THE GUT MICROBIOTA IN THE MURINE STRESS RESPONSE 

 

 

 

 

 

 

 

A Dissertation 

Presented to the Faculty of the Graduate School 

of Cornell University 

In Partial Fulfillment of the Requirements for the Degree of 

Doctor of Philosophy 

 

 

 

 

 

 

by 

Madalena Vaz Ferreira Real 

August 2025



 

 

 

 

 

 

 

 

 

 

 

 

 

© 2025 Madalena Vaz Ferreira Real



 

 

THE GUT MICROBIOTA IN THE MURINE STRESS RESPONSE 

Madalena Vaz Ferreira Real, Ph. D. 

Cornell University 2025 

 

An organism’s survival depends on its ability to adequately respond to changes in 

internal and external conditions. When these changes are particularly demanding or life-

threatening, they can act as stressors. In mammals, gut-associated microbial 

communities have been shown to be important modulators of host responses to external 

chronic stressors. However, the effects of internal or acute stressors on the gut 

microbiota are less well understood. Additionally, while most work has focused on 

community-level changes, how individual microbiota members evolve in response to 

stressors is yet to be investigated. In Chapter 1 of this dissertation, I explored how the 

gut microbiota of house mice (Mus musculus domesticus) responds to host internal 

stressors, specifically the metabolically demanding production of major urinary proteins 

(MUPs). I show that deletion of the Mup gene family caused sex-specific shifts in the 

taxonomic and functional composition of the mouse gut microbiota, including the 

depletion of microbes belonging to the Ruminococcaceae family, which has previously 

been shown to reduce the risk of metabolic disease. In Chapter 2, I investigated how the 

host and its gut microbiota respond to an acute external stressor (i.e., predator-odor 

exposure), comparing these responses to the effects of a well-established chronic stress 

paradigm (i.e., social isolation). I found that brief exposure to predator odor had a 

greater impact on the gut microbiota of wild-derived mice than prolonged social 



 

isolation, and that the gut microbiota was a better predictor of host behavior than was 

host gene expression. Finally, in Chapter 3, I explored how the host’s social 

environment affects the evolution of gut bacterial species. I found that social isolation 

accelerated divergent evolution in the native gut microbiota of wild-derived mice. The 

summation of this work contributes to our understanding of how gut symbionts shape 

the host’s ability to cope with diverse challenges.



 

iii 

BIOGRAPHICAL SKETCH 

Madalena Vaz Ferreira Real majored in Aquatic Sciences at the Abel Salazar Institute 

of Biomedical Sciences, at the University of Porto (Porto, Portugal). After completing 

her undergraduate studies in 2018, she pursued a post-graduation in Scientific 

Illustration, while working as a teaching assistant at her alma mater. In 2020, she 

became a Fulbright Fellow and was accepted as a PhD student in the Ecology and 

Evolutionary Biology program at Cornell University (NY, USA), under the advisership 

of Dr. Andrew Moeller. Since 2024, she has been visiting the department of Ecology 

and Evolutionary Biology at Princeton University (NJ, USA) as an IvyPlus Exchange 

Scholar. After her graduation in the Summer of 2025, she will start a postdoctoral 

position in the department of Fundamental Microbiology, at the University of Lausanne 

(Lausanne, Switzerland), under Dr. Philipp Engel.  



 

iv 

 

 

 

 

 

 

 

 

 

 

 

A TODOS os que me ajudaram nesta aventura. 



 

v 

ACKNOWLEDGEMENTS 

I want to gratefully acknowledge the financial support for this dissertation provided by 

the Fulbright U.S. Scholar Program, sponsored by the U.S. Department of State and the 

Portuguese Fulbright Commission. I want to clarify that its contents are solely my 

responsibility and do not necessarily represent the official views of the Fulbright 

Program, the Government of the United States, or the Portuguese Fulbright 

Commission. Funding was also provided by the National Institutes of Health (NIH) 

grant R35 GM138284-01, awarded to Dr. Andrew Moeller, and a pilot grant award 

under NIH Animal Models for the Social Dimensions of Health and Aging Research 

Network R24 AG065172, awarded to Prof. Michael Sheehan. The NIH had no role in 

study design, data collection and interpretation, and the contents of this dissertation are, 

once again, solely my responsibility, and do not necessarily represent the views of the 

NIH. 

I want to extend my sincere gratitude to my special committee—composed of Prof. 

Michael Sheehan, Dr. Maren Vitousek, and chaired by Dr. Andrew Moeller—without 

which this work would not have been possible. I also want to thank all members of the 

Moeller lab, past and present: my fellow graduate students, Samantha Goldman, Brian 

Dillard, and Abby Landers; PostDocs Jon Sanders, Brian Trevelline, Dan Sprockett, and 

Kate Lagerstrom; and technicians Weiwei Yan, Hannah Grazul, Nate Ennis, and 

Siddharth Uppal. I also want to thank undergraduate students Sylvia Bayrakdarian and 

Tess Reichard. Finally, I want to extend my profound gratitude to all the staff, faculty, 

and students at the Ecology and Evolutionary Biology Department, particularly Patty 

Jordan and Jennifer Holleran, for their aid, big and small, throughout my PhD program.   



 

vi 

TABLE OF CONTENTS 

BIOGRAPHICAL SKETCH ......................................................................................... iii 

ACKNOWLEDGEMENTS ........................................................................................... v 

TABLE OF CONTENTS .............................................................................................. vi 

LIST OF FIGURES ....................................................................................................... ix 

LIST OF TABLES ........................................................................................................ xi 

1. MAJOR URINARY PROTEIN (MUP) GENE FAMILY DELETION DRIVES 

SEX-SPECIFIC ALTERATIONS ON THE HOUSE MOUSE GUT MICROBIOTA . 1 

1.1. Abstract .............................................................................................................. 1 

1.2. Importance ......................................................................................................... 1 

1.3. Introduction ....................................................................................................... 2 

1.4. Results ................................................................................................................. 4 

1.4.1. Deletion of the Mup gene cluster .................................................................. 4 

1.4.2. Metagenomic sequencing of Mup WT and KO mice .................................... 6 

1.4.3. Mup deletion affected the taxonomic composition of the gut microbiota in 

males ....................................................................................................................... 6 

1.4.4. Mup deletion affected the functional composition of the gut microbiota in 

males ....................................................................................................................... 9 

1.4.5. Significant correspondence of taxonomic and functional profiles ............... 9 

1.4.6. Mup deletion reduced the gut microbial diversity in males ....................... 10 

1.4.7. Specific microbial taxa and functions were depleted in Mup-knockout 

males ..................................................................................................................... 10 

1.5. Discussion ......................................................................................................... 14 

1.6. Methods ............................................................................................................ 18 

1.6.1. Genome Editing .......................................................................................... 18 



 

vii 

1.6.2. Animals ....................................................................................................... 19 

1.6.3. Microbiota Analysis .................................................................................... 19 

1.6.4. Statistical Analysis ...................................................................................... 20 

2. THE MOUSE GUT MICROBIOTA RESPONDS TO PREDATOR ODOR AND 

PREDICTS HOST BEHAVIOR .................................................................................. 22 

2.1. Abstract ............................................................................................................ 22 

2.2. Importance ....................................................................................................... 23 

2.3. Introduction ..................................................................................................... 23 

2.4. Results ............................................................................................................... 25 

2.4.1. Predator-odor exposure and social isolation gave rise to stress-associated 

behaviors .............................................................................................................. 25 

2.4.2. Predator-odor exposure altered the transcriptome of an endocrine and 

immunological tissue ............................................................................................ 30 

2.4.3. Predator-odor exposure had a greater impact on the gut microbiota than 

did social isolation ............................................................................................... 31 

2.4.4. Predator-odor exposure drove persistent alterations in the relative 

abundances of specific SGBs ................................................................................ 33 

2.4.5. Predator odor–responsive SGBs co-varied with host anti-microbial and 

behavioral responses ............................................................................................ 37 

2.4.6. The gut microbiota was a better predictor of host behavior than was host 

VAT gene expression ............................................................................................ 40 

2.6. Methods ............................................................................................................ 46 

2.6.1. Animals ....................................................................................................... 46 

2.6.2. Social Isolation ........................................................................................... 46 

2.6.3. Predator Odor Exposure ............................................................................ 46 

2.6.4. Behavioral Assays ...................................................................................... 47 



 

viii 

2.6.5. Host Gene Expression ................................................................................ 49 

2.6.6. Gut Microbiota ........................................................................................... 50 

2.6.7. Statistical Analyses ..................................................................................... 52 

3. SOCIAL ISOLATION ACCELERATES EVOLUTION IN THE MURINE GUT 

MICROBIOTA ............................................................................................................. 54 

3.1. Abstract ............................................................................................................ 54 

3.2. Importance ....................................................................................................... 55 

3.3. Introduction ..................................................................................................... 55 

3.4. Results & Discussion ....................................................................................... 57 

3.4.1. Measurable intra-specific diversity in the native mouse gut microbiota ... 57 

3.4.2. Social isolation accelerates divergent evolution in the gut microbiota ..... 62 

3.5. Conclusion ........................................................................................................ 65 

3.6. Methods ............................................................................................................ 65 

3.6.1. Animals ....................................................................................................... 65 

3.6.2. Social Isolation ........................................................................................... 66 

3.6.3. Metagenomic Data ..................................................................................... 66 

3.6.5. Statistical Analyses ..................................................................................... 67 

APPENDIX .................................................................................................................. 68 

1.1. Supplemental Results ...................................................................................... 68 

1.2. Supplemental Figures ...................................................................................... 69 

1.3. Supplemental Tables ....................................................................................... 75 

2.1. Supplemental Figures ...................................................................................... 79 

2.2. Supplemental Tables ....................................................................................... 83 

REFERENCES ............................................................................................................. 90 

 



 

ix 

LIST OF FIGURES 

Figure 1.1. Mup deletion and experiment timeline. ....................................................... 5 

Figure 1.2. Mup deletion significantly changes the gut microbial taxonomic and 

functional composition of mature males.. ...................................................................... 8 

Figure 1.3. Mup deletion significantly shifts the abundance of various microbial taxa 

and functions.. .............................................................................................................. 12 

Figure 2.1. Brief predator-odor exposure altered host behavior, gene expression, and 

the gut microbiota. ........................................................................................................ 28 

Figure 2.2. Predator-odor exposure drove persistent alterations in the relative 

abundances of specific SGBs. ...................................................................................... 36 

Figure 2.3. Predator odor–responsive SGBs co-varied with host anti-microbial and 

behavioral responses. .................................................................................................... 39 

Figure 2.4. The gut microbiota was a better predictor of host behavior than was host 

gene expression. ........................................................................................................... 42 

Figure 3.1. Measurable intra-specific diversity in the native mouse gut microbiota. . 61 

Figure 3.2. Social isolation accelerates divergent evolution in the gut microbiota. .... 64 

Figure S1.1 Mup genotype birth ratio. ......................................................................... 69 

Figure S1.2. Mup deletion significantly changes the gut microbial taxonomic 

composition of mature males. ....................................................................................... 70 

Figure S1.3. Mup deletion significantly changes the gut microbial functional 

composition of mature males. ....................................................................................... 71 

Figure S1.4. Mup deletion significantly reduces microbial family diversity in mature 

males. ............................................................................................................................ 72 

Figure S1.5. Mup deletion significantly reduces microbial functional diversity in 

mature males. ................................................................................................................ 73 

Figure S1.6. Mup deletion significantly shifts the abundance of various microbial 

taxa. .............................................................................................................................. 74 



 

x 

Figure S2.1. Pair-housed mice exposed to predator odor displayed increased 

sociability. .................................................................................................................... 79 

Figure S2.2. Predator-odor exposure altered VAT gene expression. .......................... 80 

Figure S2.3. Predator odor caused a larger mean fold-change in gene expression than 

did social isolation ........................................................................................................ 81 

Figure S2.4. Social isolation increases SGB turnover. ................................................ 82 



 

xi 

LIST OF TABLES 

Table 1.1. Effect of Mup genotype on microbial species and COG function 

composition .................................................................................................................... 7 

Table 2.1. Effect of predator-odor exposure (PO) and social isolation (SI) on SGB 

composition .................................................................................................................. 33 

Table 2.2. Taxonomic classification of SGBs that displayed a persistent response to 

predator-odor exposure (DA at both D22 and D30). .................................................... 35 

Table S1.1. Effect of host sex on microbial taxonomic and functional composition. . 75 

Table S1.2. Effect of Mup genotype on microbial taxonomic and functional 

composition. ................................................................................................................. 75 

Table S1.3. Microbial taxonomic and functional profiles correspondence with 

Procrustes. .................................................................................................................... 75 

Table S1.4. Effect of Mup genotype on microbial taxonomic and functional diversity.

 ...................................................................................................................................... 76 

Table S1.5. Microbial taxa and functions showing differential abundance in WT vs. 

KO genotypes ............................................................................................................... 76 

Table S1.6. Functional enrichment analyses with a hypergeometric test. ................... 77 

Table S1.7. Power analysis of taxonomic and functional beta diversity results. ......... 78 

Table S1.8. Power analysis of taxonomic and functional alpha diversity results ........ 78 

Table S2.1. Results from linear mixed-effects model (LMM) showing effects of sex, 

predator odor (TMT), social isolation (Housing), or their interaction on mouse 

behavior ........................................................................................................................ 83 

Table S2.2. Results from pairwise LMM showing the differences in behavior between 

the different stressor treatment combinations. .............................................................. 84 

Table S2.3. Results from PERMANOVA and PERMDISP analyses showing the 

effect of predator odor (TMT) and social isolation (Housing) on the Euclidean 

distances of the visceral adipose tissue (VAT) transcriptome. ..................................... 85 



 

xii 

Table S2.4. Significant results (FDR p-value < 0,05) from the biological pathway 

enrichment analyses of differentially expressed genes (DEGs) in the VAT of TMT-

exposed (Pair TMT) and single-housed mice (Single H2O) when compared to 

unstressed controls (Pair H2O). .................................................................................... 86 

Table S2.5. Results from random forest models testing the predictive power of SGB 

abundances at D22 and VAT transcriptome on host phenotypes. ................................ 88 

Table S2.6. Summary results of predictive power (Gini Importance) of SGBs that 

were persistently associated with TMT exposure (DA at both D22 and D30). ........... 89 



 

1 

1. MAJOR URINARY PROTEIN (MUP) GENE FAMILY DELETION DRIVES 

SEX-SPECIFIC ALTERATIONS ON THE HOUSE MOUSE GUT MICROBIOTA 

 

1.1. Abstract 

The gut microbiota is shaped by host metabolism. In house mice (Mus musculus), major 

urinary protein (MUP) pheromone production represents a considerable energy 

investment, particularly in sexually mature males. Deletion of the Mup gene family 

shifts mouse metabolism towards an anabolic state, marked by lipogenesis, lipid 

accumulation, and body mass increases. Given the metabolic implications of MUPs, 

they may also influence the gut microbiota. Here, we investigated the effect of deletion 

of the Mup gene family on the gut microbiota of sexually mature mice. Shotgun 

metagenomics revealed distinct taxonomic and functional profiles between wildtype 

and knockout males, but not females. Deletion of the Mup gene cluster significantly 

reduced diversity in microbial families and functions in male mice. Additionally, a 

species of Ruminococcaceae and several microbial functions, such as transporters 

involved in vitamin B5 acquisition, were significantly depleted in the microbiota of 

Mup-knockout males. Altogether these results show that major urinary proteins 

significantly affect the gut microbiota of house mouse in a sex-specific manner. 

 

1.2. Importance 

The community of microorganisms that inhabits the gastrointestinal tract can have 

profound effects on host phenotypes. The gut microbiota is in turn shaped by host genes, 

including those involved with host metabolism. In adult male house mice, expression of 



 

2 

the major urinary protein (Mup) gene cluster represents a substantial energy investment, 

and deletion of the Mup gene family leads to fat accumulation and weight gain in males. 

We show that deleting Mup genes also alters the gut microbiota of male, but not female, 

mice in terms of both taxonomic and functional composition. Male mice without Mup 

genes harbored fewer gut bacterial families and reduced abundances of a species of 

Ruminococcaceae, a family that has been previously shown to reduce obesity risk. 

Studying the impact of the Mup gene cluster on the gut microbiota has the potential to 

reveal the effects of these genes on host phenotypes. 

 

1.3. Introduction 

The gut microbiota has emerged as a major modulator of host phenotypes, from 

metabolism [1-5] to behavior [6-8], motivating the investigation of the factors that shape 

the gut microbiota. Hosts can influence the taxonomic and functional composition of 

the microbiota through various genetically based physiological processes [9]. Gene 

knockouts allow us to test hypotheses concerning the effects of these processes on the 

gut microbiota. This approach has been employed to demonstrate the effects on the 

microbiota of innate and adaptive immune genes [10-14]. However, the effects on the 

microbiota of non-immune genes related to host metabolism have only recently started 

to be investigated [15-17]. 

Major urinary proteins (MUPs) are lipocalins involved in pheromonal 

communication [18-21], and their production represents a major metabolic investment 

for house mice (Mus musculus). In this social rodent, the Mup gene family has 

undergone an extensive parallel evolutionary expansion, accumulating 21 distinct 



 

3 

copies in a 2.2 Mbp gene cluster on chromosome 4 [21]. The genetic diversity and 

dynamic expression of Mup genes allow excreted MUPs to function as individual 

identifiers [22-25], conveying kinship, territory, social status, sex, reproductive state, 

age, health, and even diet [26-31]. This communication occurs mostly through urine 

markings [32, 33], with MUPs constituting up to 90% of the male urinary proteome 

[34], a 2–8 times higher protein content than females [35]. Mup genes are also the most 

highly expressed genes in the liver [35], representing up to 20% of the hepatic 

transcriptome in mature males [36]. Male MUP expression is particularly upregulated 

after puberty [37, 38], when social dominance is established [39, 40]. 

MUP production is a considerable energy investment for mice, particularly 

males. MUP expression is reduced under caloric restriction [41-43] and in obese and 

diabetic mice [41, 44]. In addition to being affected by energy availability, MUPs also 

regulate house-mouse metabolism. Genetically obese and diabetic mice inoculated with 

a recombinant MUP display improved insulin sensitivity mediated by a reduction in 

glucose and lipid anabolism [44]. Conversely, sexually mature Mup-knockout (KO) 

males exhibit increased anabolic phenotypes relative to wildtype (WT) individuals [45]. 

KO males displayed higher body weight and visceral adipose tissue than WT mice, 

despite lower food intake and equal energy expenditure. This metabolic shift also 

manifested through higher circulating levels of triglycerides, free fatty acids and leptin, 

and an upregulation of genes associated with lipid metabolism in KO versus WT males. 

These results point to the profound effect of MUPs on mouse metabolism. Given the 

known interactions between host metabolic function and the gut microbiota [11, 15-17, 

46, 47], Mup expression may indirectly impact the gut microbial community. 



 

4 

Additionally, Mup gene expression has been observed in the intestinal transcriptome of 

juvenile males [48] and in the duodenal proteome of adults [49], indicating gut 

commensals could also be in direct contact with MUPs. These possible mechanisms 

lead us to hypothesize that deletion of the Mup gene family may have major effects on 

the house-mouse microbial community, but this has yet to be tested. 

Here, we investigated how deletion of the Mup gene cluster impacts the gut 

microbiota of house mice. To answer this question, we generated a Mup-knockout line 

(KO; Mup-/-) with CRISPR/Cas9 and crossed it with wildtype mice (WT; Mup+/+) for 

multiple generations, yielding litters of mice discordant for Mup genotype. We then 

sequenced and compared the gut metagenomes of the homozygous progeny (KO vs 

WT). We hypothesized that sexually mature WT and KO mice would host distinct 

microbiotas, both taxonomically and functionally, and that the largest differences would 

be found in males. We found a sex-specific effect of the Mup knockout on the microbial 

taxonomic and functional profiles, demonstrating that this metabolically costly gene 

cluster shapes the gut microbiota of house mice. 

 

1.4. Results 

1.4.1. Deletion of the Mup gene cluster  

The Mup gene cluster was fully deleted using CRISPR/Cas9 to cleave upstream of Mup4 

and downstream of Mup21 (Figure 1.1A). An individual’s Mup status was confirmed 

by genotyping ear biopsies collected at weaning. Lack of MUP production was 

confirmed by measuring urinary MUP levels (Figure 1.1B). Heterozygous crosses 

generated offspring with two copies of the Mup gene cluster (wildtypes, WT), only one 



 

5 

(heterozygotes, HT), or none (Mup knockouts, KO) at the expected Mendelian ratio 

(Figure S1.1).  

 

 

Figure 1.1. Mup deletion and experiment timeline. (A) Representation of the 2 Mbp 

Mup gene cluster shows the CRISPR/Cas9 sgRNA target sites (dashed lines) and the 

primer binding sites used for genotyping (arrows). (B) Representative SDS-PAGE 

stained with Coomassie Brilliant Blue shows urinary MUPs in WT mice (blue band at 

20 kDa) and the absence of MUPs in KO individuals of both sexes (n = 2 

mice/sex/genotype). 

  



 

6 

1.4.2. Metagenomic sequencing of Mup WT and KO mice 

MUP production is upregulated in WT males after sexual maturity [37, 38] and thus we 

sampled the gut microbiota by collecting fecal pellets from 12-week–old mice (87 ± 3 

days old) housed in holding cages with same-sex littermates of diverse genotypes. A 

total of six litters were sampled. We sequenced the fecal metagenome of WT and KO 

mice, yielding an average of 9.63 ± 6.05 million reads/sample post-quality control and 

host-filtering. There were no significant differences in read count between WT and KO 

mice or between males and females, according to Wilcoxon tests (Holm–Bonferroni 

adjusted p-value = 0.90 and 0.19, respectively) and a linear mixed-effects model 

accounting for litter as a random effect (p-value = 0.51 and 0.59, respectively). For the 

alpha and beta diversity analyses, all samples were rarefied to the minimum observed 

value of 1 million reads/sample. Functional annotation of unrarefied reads produced 

478.1 ± 295.9 thousand COG counts/sample, which were rarefied to the minimum 

observed value of 84.0 thousand COG counts/sample before alpha and beta diversity 

analyses. 

 

1.4.3. Mup deletion affected the taxonomic composition of the gut microbiota in males 

To test the hypotheses that deleting the Mup gene cluster affected the composition of 

the gut microbiota, particularly in males, we compared the gut microbiota of sexually 

mature WT and KO mice. PERMANOVA analyses based on Jaccard and Bray-Curtis 

dissimilarities among samples’ MetaPhlAn 4 taxonomic profiles revealed that the 

microbiotas of males and females were significantly different (Table S1.1). Thus, we 

tested for significant differences between WT and KO individuals within each sex while 



 

7 

controlling for litter effects. These analyses revealed that mature WT and KO males 

display significantly different microbial taxonomic compositions at the species and 

genus (but not family) levels based on both presence-absence (Jaccard) and abundance-

weighted (Bray-Curtis) beta diversity dissimilarities, even while co-housed with same-

sex littermates of diverse Mup genotypes (Table 1.1; Table S1.2; Figure S1.2). This 

significant shift in the taxonomic makeup of the microbiota between WT and KO males 

was not driven by differences in dispersion between Mup genotypes, as no significant 

differences in taxonomic PERMDISP were observed (Table 1.1; Figure 1.2B). No 

significant differences were observed in females (Table 1.1; Table S1.2; Figure S1.2). 

Cumulatively, these results show that deletion of the Mup gene cluster caused a sex-

specific directional shift in the taxonomic composition of the gut microbiota.  

 

Table 1.1. Effect of Mup genotype on microbial species and COG function 

composition.a 

Sex Dissimilarity Species COG Function 

PERMANOVA PERMDISP PERMANOVA PERMDISP 

Male Jaccard 0.037 0.273 0.021 0.046 

Bray-Curtis 0.019 0.188 0.012 0.080 

Female Jaccard 0.154 0.739 0.116 0.633 

Bray-Curtis 0.158 0.398 0.481 0.200 
a Significant results are bolded (Holm–Bonferroni adjusted p-value < 0.05). 

 

  



 

8 

 F
ig

u
re

 1
.2

. 
M

u
p
 d

el
et

io
n

 s
ig

n
if

ic
a
n

tl
y
 c

h
a
n

g
es

 t
h

e 
g
u

t 
m

ic
ro

b
ia

l 
ta

x
o
n

o
m

ic
 a

n
d

 f
u

n
ct

io
n

a
l 

co
m

p
o
si

ti
o
n

 o
f 

m
a
tu

re
 m

a
le

s.
 

(A
) 

P
ri

n
ci

p
al

 c
o
o
rd

in
at

es
 a

n
al

y
se

s 
(P

C
o
A

) 
sh

o
w

 t
h
e 

o
rd

in
at

ed
 J

ac
ca

rd
 (

to
p

 r
o
w

) 
an

d
 B

ra
y

-C
u
rt

is
 (

b
o
tt

o
m

 r
o
w

) 
d
is

si
m

il
ar

it
ie

s 
in

 

sp
ec

ie
s 

(t
ri

an
g
le

s)
 a

n
d
 C

O
G

 f
u
n

ct
io

n
 (

sq
u
ar

es
) 

co
m

p
o
si

ti
o
n
 i
n
 t
h
e 

fe
ca

l 
m

et
ag

en
o
m

e 
o
f 

se
x
u

al
ly

 m
at

u
re

 m
ic

e,
 s

u
p
er

im
p
o
se

d
 w

it
h

 

a 
P

ro
cr

u
st

es
 a

n
al

y
si

s.
 T

h
e 

o
rd

in
at

ed
 p

o
in

ts
 a

re
 f

ac
et

ed
 b

y
 s

ex
 a

n
d
 c

o
lo

re
d

 b
y
 g

en
o
ty

p
e,

 s
h
o

w
in

g
 t

h
e 

ta
x
o
n
o
m

ic
 a

n
d
 f

u
n

ct
io

n
al

 

p
ro

fi
le

s 
o

f 
W

T
 (

b
lu

e)
 a

n
d
 K

O
 (

y
el

lo
w

) 
m

al
e 

(l
ef

t 
co

lu
m

n
) 

an
d
 f

em
al

e 
(r

ig
h
t 

co
lu

m
n
) 

m
ic

e.
 T

h
e 

p
er

ce
n
ta

g
e 

o
f 

v
ar

ia
ti

o
n
 i

n
 t

h
e 

ta
x
o
n
o
m

ic
 d

is
si

m
il

ar
it

y
 m

at
ri

x
 e

x
p
la

in
ed

 b
y
 e

ac
h
 P

C
o
 a

x
is

 i
s 

en
cl

o
se

d
 i

n
 p

ar
en

th
es

is
. 

T
h
e 

P
E

R
M

A
N

O
V

A
 a

n
al

y
se

s 
(c

en
te

r 

b
o
tt

o
m

) 
in

d
ic

at
e 

w
h
et

h
er

 t
h
er

e 
is

 a
 s

ig
n
if

ic
an

t 
d
if

fe
re

n
ce

 i
n
 t

h
e 

ce
n
tr

o
id

 a
n
d
/o

r 
d
is

p
er

si
o
n
 o

f 
th

e 
W

T
 a

n
d
 K

O
 g

ro
u
p
s 

fo
r 

ea
ch

 

d
is

si
m

il
ar

it
y
 m

ea
su

re
 a

n
d
 s

ex
 (

*
 =

 p
-v

al
u
e 

<
 0

.0
5
; 

n
s 

=
 p

-v
al

u
e 

>
 0

.0
5
).

 (
B

) 
B

o
x
 p

lo
ts

 s
h
o
w

 t
h
e 

sp
ec

ie
s 

(t
o
p
 r

o
w

) 
an

d
 C

O
G

 

fu
n
ct

io
n
 (

b
o
tt

o
m

 r
o
w

) 
o

v
er

la
p
 i

n
 t

h
e 

m
ic

ro
b
io

ta
 o

f 
m

a
tu

re
 m

al
es

 (
le

ft
 c

o
lu

m
n
) 

an
d
 f

em
al

es
 (

ri
g
h
t 

co
lu

m
n
),

 u
si

n
g

 b
o
th

 J
ac

ca
rd

 

(l
ef

t 
su

b
-c

o
lu

m
n
) 

an
d

 B
ra

y
-C

u
rt

is
 (

ri
g
h
t 

su
b

-c
o
lu

m
n
) 

si
m

il
ar

it
ie

s.
 T

h
e 

P
E

R
M

D
IS

P
 a

n
al

y
se

s 
(c

en
te

r 
to

p
) 

in
d
ic

at
e 

w
h

et
h
er

 t
h
er

e 

is
 a

 s
ig

n
if

ic
an

t 
d
if

fe
re

n
ce

 i
n
 t

h
e 

d
is

p
er

si
o
n
 o

f 
th

e 
W

T
 (

b
lu

e)
 a

n
d
 K

O
 (

y
el

lo
w

) 
g
ro

u
p
s 

(*
 =

 p
-v

al
u
e 

<
 0

.0
5
; 

n
s 

=
 p

-v
al

u
e 

>
 0

.0
5
).

 

 



 

9 

1.4.4. Mup deletion affected the functional composition of the gut microbiota in males 

We also tested if gene function profiles in the microbiota differed between Mup WT and 

KO mice. Using MG-RAST annotations based on Clusters of Orthologous Genes 

(COGs) (Table S1.2), we conducted the same PERMANOVA and PERMDISP analyses 

that were employed for the microbiota taxonomic profiles. As observed for taxonomy, 

the COG functional composition was significantly different between WT and KO males, 

but not between WT and KO females (Table 1.1; Figure S1.3). In contrast to the 

taxonomic results, PERMDISP indicated higher functional variation among KO males 

compared to WT individuals, based on Jaccard, but not Bray-Curtis, similarities (Table 

1.1; Figure 1.2B). No such differences in PERMDISP were observed in females (Table 

1.1; Figure 1.2B). These findings indicate mature KO males exhibited both directional 

shifts and increased inter-individual heterogeneity in the functional composition of the 

gut microbiota relative to WT mice. 

 

1.4.5. Significant correspondence of taxonomic and functional profiles  

Based on the results of PERMANOVA and PERMDISP analyses (Table 1.1), we next 

visualized the similarities among the microbiota of WT and KO mice at both taxonomic 

and functional levels using Procrustes. These analyses revealed significant 

correspondence between taxonomic and functional profiles recovered from individual 

mice using Bray-Curtis but not Jaccard (Table S1.3). Within sexes, only in mature males 

did the taxonomic and functional configurations significantly match (Table S1.3). 

Plotting the superimposed taxonomic and functional compositions of mature mice 



 

10 

revealed that samples clustered by Mup genotype in males but not females (Figure 

1.2A).  

 

1.4.6. Mup deletion reduced the gut microbial diversity in males 

Given the observed differences in gut taxonomical and functional profiles between WT 

and KO males, we investigated whether the Mup knockout also affected taxonomic and 

functional alpha diversity. A linear mixed-effects model accounting for litter as a 

random effect indicated that KO males had lower diversity (Shannon) and evenness 

(Pielou) of microbial families, but not species or genera, than their WT counterparts 

(Table S1.4; Figure S1.4). Females displayed no effect of Mup genotype on taxonomic 

diversity (Table S1.4; Figure S1.4). We also observed differences in COG function 

diversity (Shannon, but not Pielou) between WT and KO males (Table S1.4; Figure 

S1.5). No such differences were observed in females (Table S1.4; Figure S1.5). These 

alpha diversity results align with our findings from the beta diversity analyses, once 

again showing that the Mup gene cluster deletion affected the gut microbiota diversity 

of mature mice in a sex-specific manner. 

 

1.4.7. Specific microbial taxa and functions were depleted in Mup-knockout males 

Given that the Mup knockout affected the gut microbial taxonomic and functional 

composition in mature males (Figure 1.2), we next identified which specific taxa and/or 

COG functions differed between Mup WT and KO mice (Figure 1.3 and S1.6). 

Differential abundance analyses with ANCOM-BC2 detected a Ruminococcaceae 

species, species-level genome bin (SGB) 43260, that was present in both WT and KO 



 

11 

mice but significantly underrepresented in KO males relative to WT males (Holm–

Bonferroni adjusted p-value < 0.001; Table S1.5). No taxa, including SGB43260, were 

differentially abundant at this significance threshold between WT and KO females 

(Table S1.5). Several COG functions were also underrepresented in KO males (Table 

S1.5). The largest fold-change in abundance was observed in the sodium 

(Na+)/pantothenate symporter, involved in the transport of pantothenate or vitamin B5 

[50]. The most significant was component EscU of the type III secretory pathway, used 

by Gram-negative bacteria to inject virulence factors into host cells [51, 52]. 

L−asparaginase II, a periplasmic high-affinity enzyme that hydrolyses exogenous l-

asparagine into l-aspartate and ammonia [53], was also highly significant. A 

hypergeometric test revealed that no single COG category was particularly 

overrepresented among the various depleted functions in KO males (Table S1.6). The 

microbiota of KO males was instead significantly enriched in transcriptional regulators. 

Here too, no functions were highly differentially abundant in the microbiota of females 

(Table S1.5). These results point to the sex-specific effects of the Mup gene cluster 

deletion on the abundances of specific taxa and functions. 

 



 

12 

 

Figure 1.3. Mup deletion significantly shifts the abundance of various microbial 

taxa and functions. (A) Volcano plots show the log2-transformed fold change (LFC) 

in the abundance of species (top row) and COG functions (bottom row) in the gut 

microbiota of mature male (left column) and female (right column) mice. ANCOM-

BC2 analyses identified species and functions (points) that were significantly more 

abundant in WT (blue) or KO mice (yellow), labeled respectively with the family and 

species designation, and the COG Category and Function. Red lines mark the 

significance threshold (Holm–Bonferroni–adjusted p-value < 0.001). The y-axis 

indicates the -log10 transformation of the non-adjusted p-value. (B) Bar plots show the 

LFC in abundance of the COG functions that were significantly overrepresented in 

mature WT males. 

 



 

13 

1.4.8. Depleted COG functions were present in the SGB underrepresented in KO males 

Based on the results from the differential abundance analysis, which identified the 

Ruminococcaceae species SGB43260 and several COG functions significantly depleted 

in KO males, we investigated whether these COG functions were present in SGB43260. 

We functionally annotated the 29 metagenome-assembled genomes (MAGs) from 

SGB43260 in the MetaPhlAn4 reference database. The genomes had an average length 

of 2.4 ± 0.5 Mbp of which 80.2 ± 4.8 % were coding regions, with 2.2 ± 0.5 thousand 

protein-coding genes annotated. Some of these genes were annotated as COG functions 

depleted in KO males. The functions with the second and third largest fold-change in 

abundance, asparaginase and Cu/Zn superoxide dismutase, were found in 26/29 and 

16/29 genomes, respectively. No Na+/pantothenate symporters, the function with the 

largest fold-change in abundance, were observed in any of the genomes, although a gene 

identified as being part of the Na+/solute symporter family was found in one of the 

MAGs. Although missing pantothenate-specific symporters, 24/29 MAGs had a type III 

pantothenate kinase, which catalyzes the first step in the pathway that converts 

pantothenate into coenzyme A (CoA) [54]. The COG function with the most significant 

differential abundance result was component EscU of the type III secretory system 

(T3SS), but no genes associated with this secretion system were found in the annotated 

genomes. These results indicate that the depletion in some, but not all, COG functions 

could be explained by the reduction in abundance of the Ruminococcaceae species 

SGB43260 in KO males. 

  



 

14 

1.5. Discussion 

We found that the presence of the Mup gene cluster in house mice significantly altered 

the taxonomic and functional composition of the gut microbiota. In accordance with our 

hypothesis, deletion of the Mup gene cluster significantly affected the gut microbiota of 

mature male mice, but not female mice, through a shift in composition, reduction in 

diversity, and depletion of microbial taxa and functions. These differences were seen 

even among co-housed littermates, indicating that the effects of Mup were robust to the 

homogenizing effect of microbial dispersal among animals sharing a cage [55, 56]. The 

observation that Mup deletion did not change the microbial profiles of females aligns 

with our hypothesis that the effect of a Mup deletion would be stronger in male mice, 

given the sexually dimorphic expression pattern that this gene cluster exhibits [35, 36].  

Deletion of the Mup gene cluster led not only to a shift in the taxonomic and 

functional composition of mature males, but also an increase in inter-individual 

microbiota variation. The production and/or presence of MUPs at high levels may exert 

a parallel pressure on the microbial community of WT males, driving convergence 

toward a similar configuration. Mature males lacking the Mup gene cluster also 

exhibited decreased diversity and evenness in microbial families and gene functions. A 

reduction in microbial taxonomic and functional diversity is one of the hallmarks of 

obesity-related metabolic syndrome [3, 57, 58], and Mup KO males were previously 

shown to develop phenotypes associated with this syndrome, such as higher body 

weight, visceral adipose tissue, and circulating levels of triglycerides, free fatty acids 

and leptin [45]. Cumulatively, these results are consistent with a scenario in which 

knocking out the Mup gene cluster dysregulates mouse development and/or physiology, 



 

15 

increasing the amount of variation in functional content among individuals and reducing 

alpha diversity. 

In addition to community-level patterns, the abundance of a Ruminococcaceae 

species, SGB43260, was negatively impacted by the deletion of the Mup gene cluster in 

males, but not females. This taxon was present in several KO mice, showing that Mup 

presence is not required for colonization by this microbe. Ruminococcaceae have been 

previously associated in both mice and humans with lower body weight and reduced 

risk of developing metabolic syndrome [59-62]. Members of this microbial family are 

major producers of butyrate [63], a short-chain fatty acid (SCFA) that promotes gut 

health (with both anti-inflammatory and antitumoral properties [64]) and increases 

satiety [65]. The depletion of this taxon in Mup KO males might explain the previously 

observed weight gain in these animals [45], although further experiments (e.g., 

microbiota-transplant experiments into germ-free mice) will be required to assess this 

hypothesis.  

Several gene functions were also depleted in the gut microbiota of Mup KO 

males, particularly sodium-dependent transporters of pantothenate (vitamin B5) [50]. 

Pantothenate forms the core of CoA, an essential co-factor in cellular respiration [66] 

and energy metabolism, including the synthesis of SCFAs like butyrate. However, not 

all bacterial taxa are able to synthesize vitamin B5 de novo [67]. Several 

Ruminococcaceae species are auxotrophic for pantothenate  [68, 69], relying on sodium-

dependent transporters to acquire this essential vitamin and forming cross-feeding 

networks with producers [67, 70]. Functionally annotating the MAGs within SGB43260 

did not reveal any pantothenate-specific symporters, although the MAGs did contain 



 

16 

kinases that can utilize pantothenate in CoA biosynthesis. Other depleted COG 

functions, such as asparaginase and Cu/Zn superoxide dismutase, were found in the 

SGB43260 MAGs. These results suggest that the decrease in abundance of certain COG 

functions caused by Mup deletion could be linked to SGB43260, whereas others may 

not reflect changes in the relative abundance of specific taxa but instead taxonomy-

independent shifts in the functional profile of the microbiota. 

One potential caveat to the observed sex-dependent effects of Mup deletion on 

the house mouse microbiota is that our study may have been underpowered to detect 

small effects in females. However, subsampling the WT males by randomly removing 

one individual and reperforming all analyses indicated that significant results were still 

observed in iterations based on lower sample sizes (see Supplemental Results), 

suggesting that sampling effort alone cannot explain the disparity in results between 

males and females. Regardless, a priority for future work will be to test through 

expanded sampling whether WT and KO females also differ in the taxonomic and 

functional composition of their microbiota. Furthermore, lineages of WT and KO mice 

could be housed in separate cages over longer timescales (e.g., multiple generations), 

conditions which would be expected to amplify microbiota differences. Comparing the 

microbiota of immature WT and KO males and testing how the gut microbial 

composition changes when these animals reach puberty and start producing MUPs [37, 

38] could distinguish between the effect of the Mup gene cluster knockout and the sex-

specific differences in MUP production. In the present study, the observation that adult 

females, which have 2–8 times lower MUP content in their urine than do adult males 



 

17 

[35], did not display significant microbiota differences is consistent with effects of sex-

specific differences in MUP production. 

The observed shifts in the microbiota – through a decrease in overall diversity 

and depletion of taxa and functions associated with host metabolic health – could in turn 

impact the metabolism of KO mice. It will be interesting to disentangle which aspects 

of the anabolic phenotype observed in Mup-knockout males are directly caused by the 

absence of MUP production or by the shifts in the microbial community. Reciprocally 

transplanting the gut microbiotas of WT and KO mice to germ-free mice with different 

Mup genotypes could elucidate which metabolic phenotypes are caused by the 

microbiota differences observed here. The role of the identified Ruminococcaceae 

species could be explored by inoculating KO mice with this taxon and testing for 

changes in host metabolic state. Additionally, it is worth investigating the mechanisms 

through which deletion of the Mup gene cluster affects the gut microbiota. Inoculating 

KO mice with recombinant MUPs [44] could help differentiate between the effects of 

MUP production and the role of circulating MUPs on the house-mouse gut microbial 

community. Our results motivate investigation of how the gut microbiota mediates 

MUP expression. Both male and female germ-free mice have reduced MUP 

transcription [71], which suggests that the presence of the gut microbiota or some of its 

members are necessary for normal MUP production. Investigating the interactions 

between MUPs, the microbiota, and metabolism will reveal the role of this sexually 

dimorphic gene on house-mouse physiology. 

 



 

18 

1.6. Methods 

All procedures conformed to guidelines established by the U.S. National Institutes of 

Health and have been approved by the Cornell University Institutional Animal Care and 

Use Committee (protocol #2015-0060). 

 

1.6.1. Genome Editing 

The Mup gene cluster knockout with CRISPR/Cas9 on FVB x B6 hybrid mice was 

performed by Cornell’s Stem Cell and Transgenic Core Facility. The inbred mouse 

strain FVB/NJ (JAX #001800) is commonly used to generate transgenics, due to its 

large pronucleus and litter size [72]. B6(Cg)-Tyrc-2J/J (JAX #000058) are C57BL/6J 

albino mice [73]. Purified RNA (Cas9 + sgRNA; sgRNA1: 

GGGCCATAAGGAATGATCTTGGG; sgRNA2: 

GAGCTAAAGGAGACCCATATGGG) was injected into the pronucleus and 

cytoplasm of fertilized FVB x B6 embryos (n = 150). Embryos that advanced to 2-cell 

stage were transferred into pseudo-pregnant FVB x B6 females (20 embryos per 

recipient).  

The resulting offspring were genotyped with Transnetyx (Cordova, TN) using ear tissue 

samples collected at weaning. Real-time PCR was used to detect Mup presence, using 

primers that targeted the gene cluster (forward primer: 

ACAACCTGCCATTCTGTCTCTTAAT; reverse primer: 

GGCAATGAAACAAGGATTTGAGTTTTACATAT; final concentration: 900 nM). A 

second test confirmed Mup deletion by using primers flanking the gene cluster, as 

amplification is only possible if the 2.2 mbp region is absent (forward primer: 



 

19 

CAGTACTCAGGGCTTGGGATT; reverse primer: 

ACTGTTCTCGTGGGAATATGTATTGTGAA; final concentration: 900 nM). 

Successful amplification in both tests indicates a heterozygous (HT) genotype, with 

Mup present in one of the chromosomes and missing in the other. Wildtype (WT) 

individuals will only have amplification in the detection test, with only the deletion test 

being successful in knockouts (KO). The genotyping was phenotypically confirmed by 

measuring MUP concentration in the animals’ urine with sodium dodecyl sulfate–

polyacrylamide gel electrophoresis (SDS-PAGE). The gel was stained with Bio-Rad’s 

Bio-Safe™ Coomassie brilliant blue and we used Bio-Rad’s Precision Plus Protein™ 

Kaleidoscope™ Prestained Protein Standards as a ladder. 

 

1.6.2. Animals 

A breeding population was kept at a conventional mouse facility at Cornell University, 

Ithaca, NY. Heterozygous crosses generated litters of WT, HT, and KO individuals. We 

analyzed mice from 6 different litters (7 ± 1 pups per litter). Mice were weaned at 3 

weeks of age (24 ± 3 days) and housed with same-sex siblings of diverse genotypes (2-

4 animals in a cage). Animals were kept in a 12h light:12h dark cycle, with constant 

room temperature and humidity (21°C, 50%). Standard chow diet and water were 

available ad libitum. Only WT and KO males and females were included in the analysis 

(n = 20; 4 WT males; 3 WT females; 6 KO males; 6 KO females).  

 

1.6.3. Microbiota Analysis 

Fecal samples were collected at 12 weeks of age (87 ± 3 days). Microbial total DNA 



 

20 

was extracted using the Quick-DNA MagBead extraction kit (Zymo, Irvine, CA) and 

the OT-2 liquid handling robot (Opentrons, New York, NY). Library preparation 

followed the Hackflex protocol [74] using the same robot. Libraries were sequenced on 

an Illumina NextSeq 2000 (Cornell’s Biotechnology Resource Center, Ithaca, NY).  

The metagenomic data was quality controlled with FastQC (v0.12.1) [75], followed by 

trimming of the Illumina adapters (GATCGGAAGAGC) with Cutadapt (v4.1; setting: 

"--minimum-length 1 --nextseq-trim 20") [76], and removal of host reads with Bowtie2 

(v2.5.1; setting “--very-fast”; host reference genome: GRCm39 GCF_000001635.27) 

[77]. The remaining reads were taxonomically profiled with MetaPhlAn4 (v4.0.6) [78] 

and functionally annotated with Clusters of Orthologous Croups (COGs) using MG-

RAST (v4.0.3) [79]. The genomes included in SGB43260 from the MetaPhlAn4 

reference database were functionally annotated with Bakta (v1.7.0) [80]. 

All data analyses were conducted in R (v4.2.2). Before measuring alpha and beta 

diversity, the library size was normalized by randomly subsampling sequences to 1 

million reads/sample for the taxonomic data, and 84 thousand COG counts/sample for 

the functional data. The Shannon diversity index, observed richness and Pielou’s 

evenness metrics were measured using the microbiome R package (v1.20.0) [81]. The 

Jaccard and Bray-Curtis dissimilarities and Euclidean distances between samples were 

calculated with phyloseq (v1.42.0) [82]. All plots were created with ggplot2 (v3.4.1) 

[83]. All figure panels were assembled using Inkscape 1.2. 

 

1.6.4. Statistical Analysis 

Data are presented as mean ± standard deviation. Group means were compared with the 



 

21 

Wilcoxon signed-rank test. A linear mixed-effects model assessed the effect of genotype 

on microbiota diversity, while controlling for litter as a random effect (via lmerTest v3.1 

[84]). The factors (genotype, sex and/or litter) that explained the microbiota 

dissimilarity matrix were tested by running a PERMANOVA [85] via adonis2 and a 

PERMDISP [86] via betadisper (vegan v2.6 [87]). Differentially abundant taxa and/or 

functions were identified with ANCOM-BC 2 [88]. Enrichment/depletion in COG 

categories and/pathways within differentially abundant functions was tested with a 

hypergeometric test via phyper (stats v4.2.2). Throughout the analyses, a Holm–

Bonferroni adjusted p-value < 0.05 was considered statistically significant, except for 

the ANCOM-BC 2 analyses, where a Holm–Bonferroni adjusted p-value < 0.001 was 

used. 

 

  



 

22 

2. THE MOUSE GUT MICROBIOTA RESPONDS TO PREDATOR ODOR AND 

PREDICTS HOST BEHAVIOR 

2.1. Abstract 

In mammals, chronic stressors can alter gut microbial communities in ways that mediate 

host stress responses. However, the effects of acute stressors on the gut microbiota, and 

how these interact with host stress responses, are less well understood. Here, we show 

that acute exposure of wild-derived mice (Mus musculus domesticus) to predator odor 

altered the composition of the gut microbiota, which in turn predicted host behavior. 

We experimentally tested the individual and combined effects of brief (15-minute) 

exposure to synthetic fox fecal odor and prolonged (4-week) social isolation—a well-

established chronic stress paradigm. Leveraging behavioral assays, transcriptomics of 

visceral adipose tissue, and analyses of 3,500 metagenome-assembled genomes 

(MAGs) generated from our data, we found significant effects of predator odor on host 

behavior, gene expression, and gut microbiota. Gut microbial communities and host 

gene expression profiles responded more strongly to brief predator-odor exposure than 

to prolonged social isolation. The relative abundances of predator odor–responsive 

bacterial species—including members of the Muribaculaceae and Lachnospiraceae—

measured a week after a single predator-odor exposure were associated with host 

phenotypes assessed the following week, namely grooming and social behaviors and 

the expression of genes involved in anti-microbial defense, even after accounting for 

the effects of the stressors. Using random forest classifiers, we found that variation in 

gut-microbiota composition significantly predicted variation in behavior within 

treatment groups. These results indicate interactions between the gut microbiota and the 

responses of wild-derived mice to the threat of predation, and ecologically relevant 



 

23 

acute stressor. 

2.2. Importance 

Stress is an inherent part of animals’ lives, affecting not only physiology and behavior 

but also the community of microorganisms that reside within the gastrointestinal tract. 

The gut microbiota has been shown to influence how hosts respond to chronic stressors, 

but less is known about gut-microbiota effects on responses to acute stressors. We found 

that a 15-minute exposure to predator odor had greater impacts on the mouse gut 

microbiota than did 4 weeks of social isolation—a commonly used chronic stressor. 

Independent of stressor treatment, the gut microbiota significantly co-varied with mouse 

behavior measured days later. These results show that the acute threat of predation 

reshapes the mouse gut microbiota in persistent ways that predict host behavioral 

responses. 

2.3. Introduction 

Animals are faced with diverse stressors that vary in duration and intensity [89-91], 

giving rise to distinct physiological and behavioral responses [92-95]. In mammals, 

behavioral [96-98], endocrine [99, 100], and immune [96, 101, 102] stress responses are 

mediated in part by the community of host-associated microorganisms that reside in the 

gastrointestinal tract, i.e., the gut microbiota [103], which is shaped by changes in the 

internal and external host environment [98, 104-108]. Recent work has shown that 

chronic stressors, such as social isolation, can alter the gut microbiota in ways that 

mediate host stress responses [109-111].  In contrast, the effects of acute stressors on 

the gut microbiota, and the relationships between the gut microbiota and host responses 

to acute stress, remain less well-understood.  



 

24 

In rodents, the threat of predation represents an intense acute stressor, and 

animals have evolved to use environmental cues, such as predator-odor marks, to adjust 

their behavior and avoid predator encounters [112]. Previous work in mice has shown 

that a single brief exposure to synthetic fox fecal odor 2,5-dihydro-2,4,5-

trimethylthiazoline (TMT) can lead to long-lasting behavioral changes, such as 

increased fearfulness [113-117]. However, the effects of acute exposure to predator odor 

on the gut microbiota, and the extent to which predator odor–induced changes in the gut 

microbiota interact with host stress responses, have yet to be investigated.  

Here, we tested the effects of acute exposure to predator odor on the gut 

microbiota, gene expression, and behavior of wild-derived house mice (Mus musculus 

domesticus) and contrasted these effects with those of social isolation, a well-

established chronic stress paradigm [21, 22]. We longitudinally sampled the gut 

microbiota of males and females exposed to predator odor for 15 minutes at the 

beginning of the third and fourth weeks of the experiment or socially isolated for 4 

weeks. These wild-derived mice retain a more diverse gut microbiota than do lab mice 

and have recently been proposed as a solution to the microbiota-related issues around 

the poor reproducibility and low translational value of laboratory mouse studies [118, 

119]. We found that brief exposure to predator odor had a greater impact on the gut 

microbiota than did prolonged social isolation. Additionally, the taxonomic composition 

of the gut microbiota, but not host gene expression profiles, significantly predicted inter-

individual variation in host behavior, even after accounting for the effects of the 

predator-odor treatment. These findings support links between the gut microbiota and 

host responses to acute stressors.  



 

25 

2.4. Results 

2.4.1. Predator-odor exposure and social isolation gave rise to stress-associated 

behaviors 

We compared the effects of brief exposure to predator odor to those of prolonged 

exposure to a social isolation—a well-established chronic stress model. All experiments 

were conducted using wild-derived mice originally captured in Saratoga Springs (New 

York, USA) in 2012 and since inbred in the lab [120-122]. At the beginning of the 

experiment (D0), we separated groups of 6-week-old same-sex littermates into pair- and 

single-housed mice (Figure 1A). Two weeks later (D15), we exposed these mice to 35 

µL of either synthetic fox fecal odor (2,5-dihydro-2,4,5-trimethylthiazoline, TMT) or a 

control scent (deionized water, H2O) for 15 minutes. We repeated this exposure 

procedure the following week (D22). The full factorial design produced 9 unstressed 

control cages (10 male and 8 female pair-housed H2O-exposed mice), 6 predator-odor 

cages (6 male and 6 female pair-housed TMT-exposed mice), 8 social-isolation cages 

(3 male and 5 female single-housed H2O-exposed mice), and 9 cages that experienced 

both stressors (4 male and 5 female single-housed TMT-exposed mice).  

We assessed whether brief predator-odor exposure and prolonged social 

isolation acted as stressors in wild-derived mice by examining the effects of these 

treatments on host behavior with linear mixed-effects models (LMM) that accounted for 

sex, litter, and cage effects (Figure 1B; Table S1 and S2). We found that both stressors 

increased fearfulness in wild-derived mice, as both TMT-exposed (LMM p-value = 

0.022) and socially isolated mice (LMM p-value = 0.048) displayed significantly 

reduced occupancy of the exposed central area during the open field test [123]. Mice 



 

26 

individually exposed to TMT (pairwise LMM p-value = 0.015) or socially isolated 

(pairwise LMM p-value = 0.041) were significantly more fearful (i.e., displayed reduced 

center occupancy) than unstressed controls, while mice that experienced both stressors 

displayed a non-significant trend toward increased fearfulness (pairwise LMM p-value 

= 0.067). There were no significant differences in overall locomotion between the 

treatment groups (predator odor: LMM p-value = 0.257; social isolation: LMM p-value 

= 0.588). These results confirm that, in line with previous work in lab mouse strains [92, 

109, 124], predator-odor exposure and social isolation functioned as stressors that 

increased fear-associated behaviors in wild-derived mice.  

Predator-odor exposure has also been shown to lead to a reduction in self-

grooming [116], which is part of a suite of behaviors that are inhibited when rodents 

enter a more vigilant state after the threat of predation [125]. We assessed grooming 

behavior with the splash test [126], where individual mice were sprayed with a palatable 

sucrose solution and their grooming behavior was measured for 5 minutes. We observed 

that social isolation, but not predator-odor exposure, significantly decreased grooming 

behavior (LLM p-value = 0.035), such that both socially isolated mice (pairwise LLM 

p-value = 0.029) and mice that experienced both stressors (pairwise LLM p-value = 

0.039) spent significantly less time grooming than did unstressed controls (Figure 1B).  

Both brief predator-odor exposure [127] and prolonged social isolation [109, 

124] have been previously reported to decrease rodent sociability, here measured by the 

relative time spent interacting with a social stimulus (stranger) versus a non-social 

stimulus (empty cup) in the three-chamber test [128]. Contrary to prior work in lab mice 

[127], we found that pair-housed wild-derived mice exposed to predator odor displayed 



 

27 

increased sociability, spending significantly more time interacting with the social 

stimulus than the non-social stimulus (paired Wilcoxon test p-value = 0.002; Figure S1) 

and displaying a social preference index significantly greater than zero (one-sample t-

test p-value = 0.001; Figure 1B). Interestingly, the social preference of pair-housed mice 

exposed to TMT was significantly higher than that of their single-housed counterparts 

(pairwise LMM p-value = 0.009). We did not observe significant sex differences in any 

of the assayed behaviors (LLM p-values > 0.05). Overall, our results show that both 

brief exposure to predator odor and prolonged social isolation resulted in an increase in 

stress-related behaviors of wild-derived mice (i.e., fearfulness and, in the case of 

socially isolated mice, reduced self-grooming), and that predator-odor exposure in the 

context of pair housing increased sociability.  

 

  

 

 

  



 

28 

Figure 2.1. Brief predator-odor exposure altered host behavior, gene expression, 

and the gut microbiota. (A) The diagram shows the experiment timeline. Same-sex 

littermates of wild-derived mice (Mus musculus domesticus) were either pair or single-

housed for approximately 30 days. Mice from both housing groups were exposed to 

either water (H2O) or predator odor (TMT) for 15 minutes on days 15 and 22 of the 

experiment. Fecal samples for metagenomic analyses were collected on days 0, 15, 22, 

and 30. At the end of the experiment, mouse behavior was assayed over three 

consecutive days (OFT, open-field test; ST, splash test; 3CT, three-chamber test), after 

which, mice were sacrificed and their visceral adipose tissue (VAT) collected for 

transcriptomic analyses. The embedded table shows the number of male (triangles) and 

female (squares) replicate cages per treatment group. (B) Box plots show the amount of 

time that male (triangles) and female (squares) mice of different treatment groups spent 

occupying the central area of the open-field test (OFT), the time spent grooming after 

being sprayed with a 10% sucrose solution in the splash test (SP), and the social 

preference index measured during the three-chamber test (3CT). Differences between 

the groups were tested with a pairwise comparison of the marginal means from a linear 

mixed-effects model (LMM) accounting for sex, litter and cage effects (* = p-value < 

0.05). In the 3CT plot, we also tested whether the social preference index within each 

group significantly differed from zero with a one-sample t-test (** = p-value < 0.01). 

Dashed gray lines indicate the unstressed controls’ mean. (C) Principal coordinate 

analysis (PCoA) shows the first two axes of the ordinated Jaccard dissimilarity matrix 

of species genome bins (SGBs) composition at the end of the experiment (D30). Points 

represent the species profiles of male (triangles) and female (squares) mice, colored by 

treatment group: blue, Pair H2O; yellow, Pair TMT; pink, Single H2O; orange, Single 

TMT. Pair-housed cage-mates are connected by a gray line. The percent variation in 

SGB composition explained by each axis is enclosed in parenthesis. (D) Volcano plots 

show the log2-transformed fold change (LFC) in the expression of genes (points) in the 

visceral adipose tissue (VAT) of male and female mice in response to predator-odor 

exposure (left) or social isolation (right). Genes in the upper-outer quadrants were 

differentially expressed (DEGs; absolute LFC > 1 and FDR-adjusted p-value < 0.01), 

and their total number is presented at the top of the plot. (E) Bar plots show the absolute 

LFC of genes belonging to the anti-microbial humoral response biological pathway, 

which was found to be significantly enriched in the DEGs of mice exposed to predator 

odor (left) and social isolation (right). Genes are ordered and colored by the -log10(FDR-

adjusted p-value) resulting from the differential expression analysis. Gene Reg3b was 

not included in the social isolation plot as it was not differentially expressed. 

 

  



 

29 

  



 

30 

2.4.2. Predator-odor exposure altered the transcriptome of an endocrine and 

immunological tissue 

Given the observed effects of brief exposure to predator odor and prolonged social 

isolation on the behaviors of wild-derived mice, we next investigated the impacts of 

these stressors on gene expression profiles in the visceral adipose tissue (VAT). The 

VAT can function as an endocrine organ, secreting adipokine hormones that protect 

against the development of depressive- and anxiety-like phenotypes after chronic stress 

and regulate negative feedback on the HPA axis [129]. The VAT also has important 

immune functions, harboring several immune cells and secreting pro- and anti-

inflammatory cytokines [130, 131]. Both acute and chronic stress have been shown to 

increase gut permeability, leading to the translocation of microbes and their metabolites 

from the gut lumen to surrounding tissues [132]. The VAT can detect these microbial 

signals [133] and inhibit bacterial infection, systemic inflammation, and sepsis [134, 

135] through, for example, the secretion of anti-microbial peptides (AMPs) [136, 137]. 

Increased inflammation in the VAT has also been shown to impact mouse behavior 

[138]. To investigate the interactions among this immune-endocrine organ, mouse stress 

responses, and the gut microbiota, we used 3’ RNA-seq to sequence the VAT 

transcriptomes of mice dissected at the end of the experiment (D33). In total, 35 samples 

met our quality threshold of over 1 million transcripts/sample, averaging 6.54 million 

(SD = 3.22 million) transcripts/sample. 

We found that TMT exposure, but not social isolation, significantly altered the 

VAT gene expression profile (PERMANOVA of Euclidean distances accounting for 

sex and litter; predator odor: p-value = 0.035; R2 = 0.047; social isolation: p-value = 



 

31 

0.097; R2 = 0.037; Figure S2; Table S3). We tested for differentially expressed genes 

(DEGs) between mice exposed to individual stressors and unstressed controls. We 

found that, compared to social isolation, predator odor led to a greater number of DEGs 

(absolute log2(fold change) (LFC) > 1 and FDR-adjusted p-value < 0.01; Figure 1D) 

and a larger mean fold-change in gene expression (Wilcoxon test p-value = 0.026; 

Figure S3). Interestingly, these DEGs were significantly enriched in biological 

pathways related to anti-microbial humoral immune responses, including those 

mediated by anti-microbial peptides (AMPs; FDR-adjusted p-value = 0.007; Figure 1E; 

Table S4). These results show that brief predator-odor exposure had a greater impact on 

VAT gene expression than did prolonged social isolation. 

 

2.4.3. Predator-odor exposure had a greater impact on the gut microbiota than did 

social isolation 

We assessed the impacts of brief exposure to predator odor and prolonged social 

isolation on the gut microbiota by sequencing the metagenomes of fecal samples 

collected at the beginning (D0) of the experiment, immediately before the TMT 

exposures (D15 and D22), and the end of the experiment (D30; Figure 1A). Sequencing 

D0 and D30 samples yielded a mean of 33.6 million (SD = 5.0 million) reads/sample. 

Intermediate samples from D15 and D22 were sequenced at lower depth, yielding a 

mean of 10.0 million (SD = 2.4 million) reads/sample. We assembled and binned D0 

and D30 reads into metagenome-assembled genomes (MAGs) using a custom 

automated workflow (MAGmaker), following previously published methods [139]. The 

MAGmaker workflow yielded 6,404 MAGs, 3,542 of which were high quality (>90% 



 

32 

completeness, <5% contamination). We then created a custom database of 

representative species-level genome bins (SGBs) by dereplicating high-quality MAGs 

at 95% average nucleotide identity (ANI). SGBs were classified with the Genome 

Taxonomy Database [140, 141], and their relative abundances in all samples were 

quantified using InStrain [142]. The mean mappability of our samples to the SGBs was 

75.75% (SD = 3.04%), indicating that the custom-made SGB database was generally 

representative of the bacterial diversity in our samples. Samples were rarefied to the 

minimum library size before beta-diversity analyses (8,470,661 reads for D0 and D30 

samples, and 3,896,354 reads for D15 and D22 samples). 

We tested the effects of predator-odor exposure and social isolation on the SGB 

profile through PERMANOVA analyses of Jaccard and Bray-Curtis dissimilarities, 

accounting for sex, litter, and cage effects. In contrast with previous findings that 4 

weeks of social isolation altered the gut microbiota of male lab mice [109, 110], social 

isolation had no effect on mean SGB composition at D30 (Table 1; Figure 1C). Instead, 

social isolation only significantly affected Bray-Curtis dissimilarities at D22 (Table 1). 

We did, however, find that social isolation significantly increased the inter-host variance 

in the SGB composition at D30 (Jaccard PERMDISP p-value = 0.033), and socially 

isolated mice displayed significantly higher SGB turnover relative to pair-housed mice, 

as measured by the intra-individual Jaccard dissimilarity between the beginning (D0) 

and end (D30) of the experiment (LMM social isolation: p-value = 0.025; predator odor: 

p-value = 0.427; Figure S4). In contrast, TMT exposure did not significantly change the 

inter-individual variance in SGB profiles (Jaccard PERMDISP p-value = 0.177) but 

instead significantly shifted the mean SGB composition (Jaccard PERMANOVA p-



 

33 

value = 0.016). This difference was not observed before predator-odor exposure at D0 

and D15 (Table 1). Thus, predator-odor exposure drove a significant shift in the SGB 

profile of male and female mice sampled one week after the second TMT exposure 

(D30; Table 1; Figure 1C). Overall, these results indicate that two acute exposures to 

TMT were sufficient to shift the SGB profiles of the mouse gut microbiota, that this 

effect was maintained for at least one week, and that the impacts of predator-odor 

exposure on gut-microbiota composition were greater than those of chronic social 

isolation. 

 

Table 2.1. Effect of predator-odor exposure (PO) and social isolation (SI) on SGB 

composition.a 

Timepoint  Stressor  
Jaccard Bray-Curtis 

p-value R2 p-value R2 

D0  
PO 0.691 0.013 0.122 0.016 

SI 0.622 0.013 0.180 0.014 

D15  
PO 0.384 0.020 0.284 0.010 

SI 0.990 0.010 0.588 0.007 

D22 

 

PO 0.540 0.017 0.106 0.018 

SI 0.140 0.023 0.002 0.029 

D30 

 

PO 0.027 0.020 0.002 0.026 

SI 0.181 0.015 0.209 0.011 
a Significant results are bolded (p-value < 0.05). 

 

2.4.4. Predator-odor exposure drove persistent alterations in the relative abundances 

of specific SGBs 

Given the observed changes in gut-microbiota composition following predator-odor 

exposure, we next investigated which specific microbial taxa drove these patterns. We 

found that a single exposure to TMT was sufficient to significantly alter the relative 

abundances of several SGBs (D22 absolute LFC > 1 and FDR-adjusted p-value < 0.01; 



 

34 

Figure 2A). Seven of these SGBs remained differentially abundant (DA) one week after 

the second TMT exposure at D30 (Table 2). These results demonstrate that a subset of 

SGBs in the gut microbiota of wild-derived mice displayed persistent responses to brief 

predator-odor exposure and prolonged social isolation.  



 

35 

  

T
a
b

le
 2

.2
. 
T

ax
o
n
o
m

ic
 c

la
ss

if
ic

at
io

n
 o

f 
S

G
B

s 
th

at
 d

is
p
la

y
ed

 a
 p

er
si

st
en

t 
re

sp
o
n
se

 t
o
 p

re
d
at

o
r-

o
d
o
r 

ex
p
o
su

re
 (

D
A

 a
t 

b
o
th

 D
2
2

 

an
d
 D

3
0
).

 

S
G

B
s 

P
h

y
lu

m
 

C
la

ss
 

O
rd

er
 

F
a
m

il
y
 

S
p

ec
ie

s 

S
G

B
0
5
1

 
B

ac
te

ro
id

o
ta

 
B

ac
te

ro
id

ia
 

B
ac

te
ro

id
al

es
 

M
u
ri

b
ac

u
la

ce
ae

 
L

ep
a
g
el

la
 

sp
. 

0
0
2
3
6
1
1
5
5
 

S
G

B
0
0
2

 
B

ac
te

ro
id

o
ta

 
B

ac
te

ro
id

ia
 

B
ac

te
ro

id
al

es
 

R
ik

en
el

la
ce

ae
 

A
li

st
ip

es
 

sp
. 
9
1
0
5
8
7
6
7
5
 

S
G

B
0
7
4

 
B

ac
il

lo
ta

 A
 

C
lo

st
ri

d
ia

 
O

sc
il

lo
sp

ir
al

es
 

O
sc

il
lo

sp
ir

ac
ea

e
 

P
el

et
h

o
m

o
n

a
s 

sp
. 
0
0
9
7
7
4
0
5
5
 

S
G

B
1
0
5

 
B

ac
il

lo
ta

 A
 

C
lo

st
ri

d
ia

 
L

ac
h
n
o
sp

ir
al

es
 

L
ac

h
n
o
sp

ir
ac

ea
e
 

A
ce

ta
ti

fa
ct

o
r 

sp
. 
0
1
1
9
5
9
1
0
5
 

S
G

B
1
0
7

 
B

ac
il

lo
ta

 A
 

C
lo

st
ri

d
ia

 
C

h
ri

st
en

se
n
el

la
le

s 
B

o
rk

fa
lk

ia
ce

ae
 

G
a
ll

im
o
n
a
s 

sp
. 
9
1
0
5
8
5
5
9
5
 

S
G

B
1
0
8

 
B

ac
il

lo
ta

 A
 

C
lo

st
ri

d
ia

 
C

h
ri

st
en

se
n
el

la
le

s 
B

o
rk

fa
lk

ia
ce

ae
 

C
o
p
ro

p
la

sm
a

 

sp
. 
9
1
0
5
8
3
8
5
5
 

S
G

B
1
5
8

 
B

ac
il

lo
ta

 A
 

C
lo

st
ri

d
ia

 
L

ac
h
n
o
sp

ir
al

es
 

L
ac

h
n
o
sp

ir
ac

ea
e
 

C
A

G
-3

0
3
 

sp
. 
9
1
0
5
8
5
2
1
5
 

 



 

36 

  

F
ig

u
re

 2
.2

. 
P

r
ed

a
to

r-
o
d

o
r 

ex
p

o
su

re
 d

ro
v

e 
p

er
si

st
en

t 
a
lt

er
a

ti
o
n

s 
in

 t
h

e 
re

la
ti

v
e 

a
b

u
n

d
a
n

ce
s 

o
f 

sp
ec

if
ic

 S
G

B
s.

 (
A

) 
V

o
lc

an
o

 

p
lo

ts
 s

h
o
w

 t
h
e 

lo
g

2
-t

ra
n
sf

o
rm

ed
 f

o
ld

 c
h
an

g
e 

(L
F

C
) 

in
 t

h
e 

re
la

ti
v
e 

ab
u
n
d
an

ce
s 

o
f 

sp
ec

ie
s 

(p
o
in

ts
) 

in
 t

h
e 

g
u
t 

m
ic

ro
b
io

ta
 o

f 
m

al
e 

an
d

 

fe
m

al
e 

m
ic

e 
in

 r
es

p
o
n
se

 t
o
 p

re
d
at

o
r-

o
d
o
r 

ex
p
o
su

re
 (

le
ft

 c
o
lu

m
n
) 

o
r 

so
ci

al
 i

so
la

ti
o
n
 (

ri
g

h
t 

co
lu

m
n
) 

a 
w

ee
k
 a

ft
er

 t
h
e 

fi
rs

t 
T

M
T

 

ex
p

o
su

re
 (

D
2
2
; 

to
p

 r
o
w

) 
an

d
 a

 w
ee

k
 a

ft
er

 t
h
e 

se
co

n
d
 T

M
T

 e
x
p
o
su

re
 (

D
3

0
; 

b
o
tt

o
m

 r
o
w

).
 U

p
p

er
-o

u
te

r 
q

u
ad

ra
n

ts
 c

o
n

ta
in

 p
o

in
ts

 

co
rr

es
p
o
n
d
in

g
 t

o
 d

if
fe

re
n

ti
al

ly
 a

b
u
n
d

an
t 

S
G

B
s 

(D
E

S
eq

2
; 

ab
so

lu
te

 L
F

C
 >

 1
 a

n
d
 F

D
R

-a
d
ju

st
ed

 p
-v

al
u
e 

<
 0

.0
1

),
 w

h
o
se

 t
o
ta

l 
n

u
m

b
er

 

is
 p

re
se

n
te

d
 a

t 
th

e 
to

p
 o

f 
th

e 
p
lo

t.
 T

h
e 

th
re

e 
m

o
st

 s
ig

n
if

ic
an

tl
y
 d

if
fe

re
n
ti

al
ly

 a
b
u
n
d
an

t 
S

G
B

s 
in

 e
ac

h
 q

u
ad

ra
n
t 

ar
e 

la
b

el
le

d
. 
(B

) 
T

h
e 

p
h
y
lo

g
en

et
ic

 t
re

e 
sh

o
w

s 
th

e 
ev

o
lu

ti
o
n
ar

y
 r

el
at

io
n

sh
ip

 (
n
u
m

b
er

 o
f 

am
in

o
 a

ci
d
 s

u
b
st

it
u
ti

o
n
s)

 b
et

w
ee

n
 r

ep
re

se
n
ta

ti
v
e 

S
G

B
s 

in
 o

u
r 

cu
st

o
m

 d
at

ab
as

e,
 w

it
h
 t

h
e 

ti
p
s 

co
lo

re
d

 b
y
 b

ac
te

ri
al

 c
la

ss
. 
T

h
e 

ri
n
g
s 

sh
o
w

 d
if

fe
re

n
ti

al
ly

 a
b
u
n
d
an

t 
S

G
B

s 
at

 D
2
2
 (

o
u
te

r 
tw

o
 r

in
g
s)

 a
n
d
 

D
3
0
 (

in
n
er

 t
w

o
 r

in
g
s)

, 
co

lo
re

d
 b

y
 t

h
e 

st
re

ss
o
r 

g
ro

u
p
 i

n
 w

h
ic

h
 t

h
ey

 w
er

e 
en

ri
ch

ed
: 

H
2
O

 (
d
ar

k
 b

lu
e)

 v
s.

 T
M

T
 (

y
el

lo
w

) 
an

d
 P

ai
r 

(l
ig

h
t 

b
lu

e)
 v

s.
 S

in
g
le

 (
p
in

k
).

 S
G

B
s 

th
at

 d
is

p
la

y
ed

 a
 p

er
si

st
en

t 
re

sp
o
n
se

 t
o
 t

h
e 

st
re

ss
o
rs

 a
cr

o
ss

 b
o
th

 D
2
2
 a

n
d
 D

3
0
 a

re
 e

n
ci

rc
le

d
. 

 



 

37 

2.4.5. Predator odor–responsive SGBs co-varied with host anti-microbial and 

behavioral responses  

As predator-odor exposure up-regulated the expression of genes involved in AMP 

production, we investigated possible associations between the expression of these AMP-

associated genes and the relative abundances of microbes that responded persistently to 

TMT exposure (DA at both D22 and D30) (Figure 3). We controlled for the effects of 

the stressor treatments (Pair H2O, Pair TMT, Single H2O, Single TMT) by testing these 

associations using partial Spearman correlations, that residualized the AMPs expression 

levels and the relative abundances of SGBs against the treatment group. We found that, 

a week after the first TMT exposure (D22), the relative abundances of SGB051—

identified as Muribaculaceae species Lepagella sp. 002361155 and depleted in TMT-

exposed mice—and the relative abundances of TMT-responsive SGB105 and 

SGB107—identified as Lachnospiraceae species Acetatifactor sp. 011959105 and 

Borkfalkiaceae species Gallimonas sp. 910585595, respectively—were negatively 

associated with all genes involved in AMP production (Spearman’s Rho < 0; FDR-

adjusted p-value < 0.05). The relative abundances at D22 of all other TMT-responsive 

SGBs showed positive associations with genes involved in AMP production, although, 

after correcting for multiple testing, this correlation was only significant for SGB074—

identified as Oscillospiraceae species Pelethomonas sp. 009774055. These results show 

that, independent of the effects of stressor treatment, the relative abundances of predator 

odor–responsive SGBs were significantly associated with host AMP gene expression. 

We also tested whether the relative abundances of predator odor–responsive 

SGBs co-varied with host behavioral responses independently of the effects of stressor 



 

38 

treatment (Figure 3). We found that the D22 relative abundances of all predator odor–

responsive SGBs but one were negatively associated with mouse social preference. 

Lachnospiraceae members SGB158 and SGB105, which were enriched in TMT-

exposed mice, were also negatively associated with grooming behavior. These findings 

indicate that behavioral responses to predator-odor exposure, such as reduced grooming 

and social preference (particularly for single-housed mice exposed to TMT), could be 

partially explained by the relative abundances of TMT-responsive SGBs measured a 

week prior.   

  



 

39 

 

 

Figure 2.3. Predator odor–responsive SGBs co-varied with host anti-microbial and 

behavioral responses. (A) The correlation matrix shows associations between the D22 

relative abundances of SGBs that were persistently associated with predator-odor 

exposure (differentially abundant at both D22 and D30) and host anti-microbial (AMPs) 

and behavioral responses. The color of the circles corresponds to Spearman’s Rho from 

a partial correlation test that accounted for the effects of stressor treatments. The 

thickness of the circle border indicates the significance level of this same correlation 

test, corrected for multiple comparisons (FDR).  



 

40 

2.4.6. The gut microbiota was a better predictor of host behavior than was host VAT 

gene expression 

Motivated by the observation that host phenotypes could be partially explained by the 

relative abundances of specific predator odor–responsive SGBs, we tested the predictive 

power of the entire gut microbiota. To this end, we trained a random forest model 

residualized against stressor treatment to predict host behavior based on SGB relative 

abundances at each timepoint. We tested whether inter-individual variation in the gut 

microbiota significantly predicted variation host behavior within treatment groups (Pair 

H2O, Pair TMT, Single H2O, Single TMT). This analysis asked whether, after 

accounting for the effects of stressor treatment, gut microbiota composition predicted 

future mouse behavior. The model’s performance was assessed through a 5-fold cross-

validation strategy, and we conducted a permutation test to measure the statistical 

significance of the observed R2 values. We found that inter-individual variation in SGB 

relative abundances at D30 significantly predicted grooming behavior (Permutation test 

p-value = 0.016 and R2 = 0.021; linear regression p-value = 0.0195) and social 

preference (Permutation test p-value = 0.001 and R2 = 0.102; linear regression p-value 

= 0.0324), but not fearfulness (i.e., center occupancy; Figure 4A; Table S5). SGB 

relative abundances at D0, D15, or D22 were not predictive of behavior (Table S5). Of 

the SGBs we found to respond persistently to predator-odor exposure (DA at both D22 

and D30), SGB107 and SGB158 ranked among the top half most important features for 

predicting grooming and social behavior (Figure 4B; Table S6). SGB105 and SGB002 

were also important predictors of social preference, with the latter—identified as 

Rikenellaceae species Alistipes sp. 910587675—ranking in 20th place. Furthermore, we 



 

41 

found that the gut microbiota was a better predictor of host behavior than was VAT gene 

expression, which did not significantly predict behavioral phenotypes (Permutation test 

p-value > 0.1 and R2 < 0; Figure 4A; Table S5). These results indicate that the gut 

microbiota sampled at the end of the experiment, a week after the second TMT 

exposure, significantly predicted host behavioral phenotypes measured days later.  



 

42 

 

Figure 2.4. The gut microbiota was a better predictor of host behavior than was 

host gene expression. (A) The scatter plots show the relation between observed 

grooming (left column) and social (right column) behavior (residualized to account for 

the effect of stressor treatment) and the values predicted by a random forest model based 

on SGB relative abundances at D30 (top row) or VAT gene expression (bottom row). 

Points are colored by treatment group: blue, Pair H2O; yellow, Pair TMT; pink, Single 

H2O; orange, Single TMT. The significance of the linear regression is shown in the 

upper inner corner, with significant values (p-value < 0.05) in bold. (B) The bar plots 

show the zero-centered log10(Gini Importance) of SGBs persistently associated with 

predator-odor exposure (differentially abundant at both D22 and D30) for predicting 

grooming (left) and social (right) behaviors. The SGB importance rank order is 

annotated in bold, and the bars are colored blue if the SGB was among the top half most 

important features (the dashed line indicates the median importance value). The 

maximum x-axis value corresponds to the most important SGB, identified in the top 

right corner. 



 

43 

2.5. Discussion 

We found that exposure of wild-derived mice to an acute stressor—predator odor—

promoted stress-associated behaviors, altered gene expression patterns in the visceral 

adipose tissue through the activation of the AMP-mediated anti-microbial responses, 

and shifted the species composition of the gut microbiota. Surprisingly, we found that 

brief exposure to predator odor had a larger impact on the VAT transcriptome and the 

gut microbiota SGB profile than did prolonged social isolation—a well-established 

chronic stress model previously shown to impact the gut microbiota [109, 110]. 

Moreover, variation in gut-microbiota composition at the end of the experiment (D30) 

predicted host behaviors within treatment groups (i.e., after accounting for the stressor 

effects). These results show that predator odor alters the murine gut microbiota and 

suggest links between inter-individual variation in the gut microbiota and behavioral 

responses in the context of predation stress.  

By comparing the effects of two distinct stressors and their combination on the 

gut microbiota, host gene expression, and behavior, our study builds on prior work that 

focused on the effects of individual stressors [109, 110, 143, 144]. We observed that the 

combination of stressors often had non-additive effects, exemplifying the context 

dependence of the stress response [92]. For example, we observed that pair-housed mice 

exposed to predator odor displayed significantly higher social preference than did their 

socially isolated counterparts, demonstrating that the social environment can affect the 

response to consequent stressors. In our case, pair-housed mice might seek out 

proximity with conspecifics when under the threat of predation—a behavior described 

as huddling [145]—while single-housed mice exposed to predator odor display the 



 

44 

asocial behavior commonly observed after chronic stressor exposure [146]. 

We identified both general and stressor-specific responses in hosts and their gut 

microbiota. For example, we observed that both stressors increased the expression of 

AMPs in the VAT. Interestingly, the expression of these AMPs co-varied with the 

relative abundances of several stressor-responsive taxa, even after accounting for 

stressor effects (Figure 3). For example, SGB051, which was depleted in TMT-exposed 

mice, was negatively associated with all genes involved in AMP production. This SGB 

was identified as a Muribaculaceae, a family that has been associated with anti-

inflammatory markers [147-149] and found to be depleted in inflammatory bowel 

disease (IBD) [147, 150, 151]. Muribaculaceae produce short-chain fatty acids 

(SCFAs), which have been shown to stimulate mucus production and reduce 

inflammation [152], and members of this family also engage in cross-feeding behavior 

with known probiotic bacteria, such as Bifidobacterium and Lactobacillus [153]. The 

associations observed in our study could be due to either SGB051 growing better in the 

absence of AMPs or to a negative effect of this species on anti-microbial responses. One 

possible explanation for our results is that responses of the gut microbiota to predator 

odor were mediated by VAT inflammation triggered during the acute stress response. 

Alternatively, it is also possible that predator-odor exposure led to changes in the gut 

microbiota that triggered VAT inflammation. Given that both increased VAT 

inflammation [138] and stress-induced changes in the gut microbiota [109-111] have 

been previously shown to affect mouse behavior, the observed behavioral effects of 

stressor exposure may be mediated by signaling along the gut-brain-immune axis [154-

156]. Alternatively, the gut microbiota may interact with host behavior in a VAT-



 

45 

independent manner, as variation in gut species composition was a better predictor of 

mouse sociability and grooming behavior than was the VAT gene expression profile. 

Understanding the mechanisms underlying the associations between the predation 

stress–altered microbiota, host inflammatory responses, and behavior will require 

controlled experiments in germ-free animals designed to test the effects of stressor-

associated microbiota on mouse phenotypes that can complement previous work 

showing that host responses to acute stressors are microbiota-dependent [157-160].  

Several results contradicted expectations based on previous literature using lab 

mouse strains. For example, while 4 weeks of social isolation was sufficient to alter the 

gut microbiota of C57BL6 males [109, 110], in our study only predator-odor exposure 

significantly altered the species composition of the gut microbiota. This discrepancy 

could be due to differences in the robustness of the gut community to perturbation, as 

lab mouse strains have a depleted, and potentially more volatile, gut microbiota 

compared to wild mice [118, 119]. Previous work has shown that wild-derived mouse 

lines can retain microbiota from wild populations for over a dozen generations in the 

lab [161]. Additionally, there is evidence that wild-derived mice are less sociable than 

lab mouse strains [162-164], potentially increasing their tolerance for social isolation. 

This hypothesis is supported by our observation that unstressed controls lack a clear 

social preference, whereas commonly used lab strains tend to display high sociability 

[165]. The significant social preference displayed by pair-housed mice exposed to 

predator odor similarly contradicts previous results in the literature [127]. Brief 

predator-odor exposure is likely also a more potent stressor for wild-derived mice than 

prolonged social isolation. This is supported by previous work showing wild-derived 



 

46 

mice are more readily responsive to the threat of predation (e.g., higher freezing and 

flight) than lab mice [166]. Thus, this work adds to a growing body of evidence 

motivating the incorporation of wild-derived mice into research as a biologically 

relevant complement to the work conducted on classic lab strains [118, 119, 162, 167].  

2.6. Methods 

All procedures conformed to guidelines established by the U.S. National Institutes of 

Health and have been approved by the Cornell University Institutional Animal Care and 

Use Committee (protocol #2015-0060). 

2.6.1. Animals 

We used a wild-derived M. m. domesticus line, NY3, which was captured in Saratoga 

Springs (NY, USA) in 2012 and inbred in captivity [120-122]. These mice are related 

to the SAR/NachJ lines currently provided by the Jackson Laboratory [167]. Mice were 

kept on an inverted light cycle (light:10pm-10am) and all handling was conducted 

during the active (dark) phase using red light. Standard chow diet and water were 

available ad libitum. 

2.6.2. Social Isolation 

Mice were weaned at 3 weeks of age (M = 22 days old, SD = 2) and co-housed with 

same-sex littermates. At 6 weeks of age (D0, mice were M = 37 days old, SD = 8), we 

separated co-housed same-sex littermates into pair- and single-housed mice, in a total 

of 8 male and 7 female mouse pairs, and 7 male and 10 female mice that were socially 

isolated.  

2.6.3. Predator Odor Exposure 

Two weeks after the start of the experiment (D15; mice were M = 57 days old, SD = 5), 



 

47 

we exposed mice to synthetic fox fecal odor, 2,5-dihydro-2,4,5-trimethylthiazoline 

(TMT) (90% purity; BioSRQ, Sarasota, FL, USA). Mice were individually placed in an 

empty cage, brought to a fume hood, and a filter paper with 35 µL of either TMT or 

deionized water (H2O) was placed on their cage metal grid topper for 15 minutes. After 

the exposure period, we removed the filter paper, keeping the mice in the fume hood for 

an additional 15 minutes to ensure the TMT was removed from the work area. 

Afterwards, all mice were returned to their home cages. We exposed control mice first 

and the TMT-treated mice last, to ensure controls never encountered the predator odor. 

We repeated this exposure procedure the following week (D22; mice were M = 64 days 

old, SD = 5). In total, 10 males and 11 females were exposed to TMT, while 13 males 

and 13 females were exposed to H2O. Considering both stressors, there were 9 pair-

housed H2O-exposed cages (5 male pairs and 4 female pairs), 6 pair-housed TMT-

exposed cages (3 male pairs and 3 female pairs), 8 single-housed H2O-exposed cages (3 

males and 5 females), and 9 single-housed TMT-exposed cages (4 males and 5 females). 

2.6.4. Behavioral Assays 

We assayed mouse behavior on the fifth week of the experiment (M = 30 days, SD = 3), 

over three consecutive days. The tests were ordered from least to most disruptive, 

starting with the Open Field Test (OFT) [123], followed by the Splash Test (SP) [126], 

and ending with the Three-Chamber Test (3CT). All behavior trials were conducted 

between 1:00-6:00 pm, during the animals' active (dark) phase. The OFT and 3CT were 

conducted inside a soundproof chamber in a separate behavior room. The mice were 

habituated to the behavior room for at least 30 minutes. We alternated the testing order 

between experimental groups and wiped the behavioral apparatuses between trials with 



 

48 

70% ethanol. An infrared camera placed above the apparatuses recorded the behavior 

tests. All analyses were conducted by the same experimenter, who was blind to the 

experimental condition. 

2.6.4.1. Open Field Test 

Mice were placed in the center of a white PVC arena (48 × 46 × 44 cm) and allowed to 

explore for 10 minutes. We used ToxTrac [168] to detect mouse trajectories and 

calculate total distance walked and time spent in the center of the arena (34 × 32 cm; ½ 

of the area). After the test, each mouse was housed in a new individual home cage. 

2.6.4.2. Splash Test  

Our protocol was adapted from Nollet [169]. On the second day of behavior tests, each 

mouse was placed in an empty “spray” cage and sprayed on their dorsal coat with a 10% 

sucrose solution. Mice were then quickly returned to their home cage, and grooming 

behavior was measured for the following 5 minutes. We directly measured the duration 

of the grooming behavior by using a chronometer to time observed instances of 

nose/face grooming, head washing, and body grooming, as described by Kalueff and 

Tuohimaa [170]. 

2.6.4.3. Three-Chamber Test 

Mice were placed in the middle chamber of a transparent plexiglass three-chamber arena 

for a 5-minute habituation period. We then placed an overturned wire cup on diagonally 

opposite corners of each side chamber. One of the cups was empty (non-social 

stimulus), while the other contained a stranger mouse of the same sex as the test mouse 

(social stimulus). The position of the stimuli was counterbalanced between trials. We 

allowed the test mouse to explore the arena for 10 minutes. The recorded video footage 



 

49 

was manually scored with BORIS where we logged interactions with the social and non-

social stimuli following recommendations by Rein, Ma [128]. These were behaviors 

such as directly interacting with the stranger mouse between the cup’s wire bars and/or 

sniffing the cup or any protruding part of the stranger mouse (e.g., tail). Climbing on 

top of the cups or grooming in their proximity was not counted as an interaction. 

Additionally, we calculated the social preference index, which is the difference in 

interaction time between the social and non-social stimulus, normalized by the total time 

interacting with both stimuli [128]. Significantly higher interaction time with the social 

versus non-social stimulus was interpreted as the existence of a social preference [128].  

2.6.5. Host Gene Expression 

2.6.5.1. Tissue Collection and Gross Morphometry 

Mice were humanely euthanized with an overdose of CO2 followed by cervical 

dislocation. Immediately after, we measured body mass and length (nose-tip to tail-

base) and dissected the visceral adipose tissue (VAT), which was flash frozen and stored 

at -80°C. 

2.6.5.2. Transcriptomic Sequencing 

Total RNA was extracted from the VAT using the RNeasy Lipid Tissue Mini Kit 

(Qiagen, Germantown, MD, USA). Briefly, approximately 50 mg of VAT were placed 

in Bead Ruptor 2 mL tubes with 2.4 mm metal beads (Omni International, Kennesaw, 

GA, USA) and 1 mL of QIAzol Lysis Reagent (Qiagen). The tissues were homogenized 

with a Bead Ruptor (Omni International) at speed 5 for 30 s. We followed the 

manufacturer’s instructions for the remainder of both RNeasy protocols, including the 

optional drying step of placing the RNeasy spin column in a new 2 mL collection tube 



 

50 

and centrifuging at full speed for 1 minute. Total RNA was eluted twice in 30 µL of 

RNase-free water (Qiagen). We assessed the concentration and purity of the extracted 

RNA with a NanoDrop Microvolume Spectrophotometer (Thermo Fisher Scientific) for 

all samples and with a TapeStation (Agilent Technologies, Santa Clara, CA, USA) for 

10% of the samples. The 3’RNA-seq libraries were prepared at Cornell University’s 

Biotechnology Resource Center (BRC; Ithaca, NY, USA) and sequenced on a NextSeq 

500 (Illumina). 

2.6.5.3. Transcriptomic Data Processing 

We assessed the quality of our transcriptomic data with FastQC (v0.72.0) [75] and 

filtered out samples with less than 1 million reads. The remaining transcripts were 

aligned against the Mus musculus reference genome (GRCm39 GCF_000001635.27) 

using STAR (v2.7.10b; setting: --quantMode GeneCounts --outSAMtype BAM 

SortedByCoordinate) [171].  

2.6.6. Gut Microbiota 

2.6.6.1. Fecal Collections 

We sequenced the metagenome from fecal samples collected at the beginning of the 

first (D0), third (D15), fourth (D22), and fifth (D30) weeks of the experiment. D15 and 

D22 samples were collected immediately before the predator-odor exposure procedure. 

D30 samples were collected immediately before the first behavioral assay. All fecal 

samples were flash frozen and stored at -80°C.  

2.6.6.2. DNA Sequencing 

Total microbial DNA was extracted with the Quick-DNA MagBead extraction kit 

(Zymo, Irvine, CA, USA) using an OT-2 liquid handling robot (Opentrons, New York, 



 

51 

NY, USA). The metagenomic libraries were prepared with the TruSeq kit (Illumina, San 

Diego, CA, USA) at Cornell’s BRC. The D0 and D30 libraries were sequenced in a 

NovaSeq 6000 S4 (PE150 bp; Illumina) at the UC Davis Genome Center (Davis, CA, 

USA), while the D15 and D22 libraries were sequenced in a NovaSeq X (2 × 150 bp; 

Illumina) at Cornell’s BRC.  

2.6.6.3. Metagenomic Data Processing and Analysis 

We processed our metagenomic samples and assembled MAGs using a custom 

snakemake [172] workflow (MAGmaker, available at 

https://github.com/CUMoellerLab/sn-mg-pipeline). Briefly, sequencing adapters were 

trimmed with Cutadapt (v17.4; setting: "-a 

AGATCGGAAGAGCACACGTCTGAACTCCAGTCA -A 

AGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT -a 

GGGGGGGGGGGGGGGGGGGG -A GGGGGGGGGGGGGGGGGGGG -u 10 -U 

10 --minimum-length 1 --nextseq-trim 20") [76], host reads removed with Bowtie2 

(v2.5.1; setting: “--very-fast”; host reference genome: GRCm39 GCF_000001635.27) 

[77], and the quality of our metagenomic data was assessed with FastQC (v0.72.0) [75]. 

We assembled high-quality non-host reads from D0 and D30 into contiguous sequences 

(contigs) using Megahit (v1.2.9) [173] and assessed assembly quality with Quast 

(v5.2.0) [174]. We then calculated the distances between our samples using SourMash 

(v4.4.0) [175] and selected the 11 samples most dissimilar to all others to be used as 

“prototypes” against which the metagenomic reads from all samples were mapped with 

Minimap2 (v2.24) [176]. The resulting coverage information was input into three 

different binning algorithms: CONCOCT (v0.4.2) [177], MetaBAT2 (v2.15) [178], and 

https://github.com/CUMoellerLab/sn-mg-pipeline


 

52 

MaxBin (v2.2) [179]. We then used DASTool (v1.1.6) [180] to select the optimal set of 

bins among the three binners. The resulting MAGs were quality controlled for 

completeness and contamination with CheckM (v1.2.2) [181] and GUNC (v1.0.6) 

[182]. We created a custom genome database by dereplicating high-quality MAGs 

(>90% completeness, <5% contamination) into 95% average nucleotide identity (ANI) 

species-genome bins (SGBs) using dRep (v3.4.2) [183]. We profiled these SGBs using 

GTDB-Tk (v2.4.0) [141]. Finally, we calculated the relative abundances of 

representative SGBs in our samples by mapping all metagenomic reads against our 

custom genome database with Bowtie2 and profiling them using InStrain (v1.8.0) [142]. 

The counts from our mapped reads were normalized by the genome size of the 

representative SGB the read mapped to and the library size of the metagenomic sample 

from which the read originated. We then converted these normalized counts into a 

relative abundances percentage. The Jaccard and Bray-Curtis dissimilarities between 

samples were calculated with phyloseq (v1.42.0) [82]. 

2.6.7. Statistical Analyses 

All data analyses, except were indicated, were conducted in R (v4.2.2). Means are 

reported alongside standard deviations (SD). Linear mixed-effects models, controlling 

for litter and cage as a random effect, were implemented via lmerTest (v3.1) [84]. We 

tested which factors explained the microbiota dissimilarity matrix by running a 

PERMANOVA [85] via adonis2 and a PERMDISP [86] via betadisper, both included 

in vegan (v2.6) [87]. We calculated normalized counts and differential gene expression 

or taxa relative abundances between stressor treatments using DESeq2 (v1.42.1) [184] 

with LFC shrinkage using the "ashr" adaptive shrinkage estimator [185], after filtering 



 

53 

for genes or taxa with one or more counts in at least three samples. We identified 

biologically meaningful pathways within DEGs with a biological pathway enrichment 

analysis using clusterProfiler (v4.10.1) [186, 187]. We investigated whether species 

relative abundances significantly explained variation in behavior while controlling for 

stressor treatment effects with a random forest model implemented via Python’s scikit-

learn (v1.6) [188]. We evaluated our model’s performance using a 5-fold cross-

validation strategy, where we subset our data into 5 different “folds” and trained the 

model 5 times, using 4 of the folds as the training data and the remaining fold as the test 

data. Our performance metric (R2) was averaged across all 5 iterations. This way, each 

data point is used at least once to both train and test our model, and the R2 reflects the 

model’s performance across the entire dataset. We evaluated the statistical significance 

of the R2 using a permutation test, where we randomized the associations between 

behavior and species relative abundances and recalculated R2 for 1000 iterations, thus 

generating a distribution of R2 values expected under the null hypothesis that behavior 

is not associated with species relative abundances. We calculated a p-value by 

measuring the likelihood of observing the R2 resultant from our random forest in the 

null distribution. Lastly, we assessed the importance of individual SGBs in predicting 

behavior using Python’s scikit-learn (v1.6) [188]. Throughout the analyses, a p-value < 

0.05 was considered statistically significant, except where otherwise indicated. All plots 

were created with ggplot2 (v3.4.1) [83]. All figure panels were assembled using 

Inkscape 1.2.  



 

54 

3. SOCIAL ISOLATION ACCELERATES EVOLUTION IN THE MURINE GUT 

MICROBIOTA 

 

3.1. Abstract 

The mammalian gut microbiota is a key regulator of host health that can be shaped by 

the host’s social environment. Prior work has shown that social interactions can drive 

the transmission of gut bacteria between hosts, altering the taxonomic composition of 

the gut microbiota. However, relatively little is known about the effects of the host’s 

social environment on the evolution of gut symbionts. Here we show that social isolation 

accelerates the evolution of bacterial species in the gut microbiota of wild-derived house 

mice (Mus musculus domesticus). We split littermates into single-housed individuals 

and co-housed pairs and sequenced their gut metagenomes before and after 30 days. We 

quantified evolutionary rates—defined here as the amount of intra-specific genomic 

change—in 173 gut bacterial species through analyses of 3,500 metagenome-assembled 

genomes (MAGs) generated from experimental mice. Over the course of the 30-day 

experiment, gut bacterial species evolved significantly more in single-housed mice than 

in co-housed littermates. This also translated into greater intra-specific genomic 

divergence between socially isolated individuals and their littermates than between co-

housed pairs. These results demonstrate that host’s social environment shapes the 

evolutionary rates and trajectories of gut bacterial species. 

 

 



 

55 

3.2. Importance 

Animals are frequently faced with changes in the environment to which they must 

respond adequately to survive and thrive. The community of microbes that inhabits the 

gastrointestinal tract, known as the gut microbiota, can influence host phenotypes 

through its own responses to environmental change. Most previous work has examined 

these responses at the community level, such as shifts in species abundances, but has 

generally overlooked changes occurring within the genomic repertoire of each microbial 

species. These evolutionary processes are especially understudied in complex systems, 

such as a multi-species host-associated community like the mammalian gut microbiota. 

Here, we investigated how changing the host’s social environment, by socially isolating 

wild-derived mice for 30 days, affected the evolution of gut microbes. We found that 

social isolation accelerated divergent evolution in gut microbes, which has potential 

consequences for host fitness. 

 

3.3. Introduction 

Mammals, like all multicellular life, have evolved in a microbial world [189]. As a 

result, their physiology has become enmeshed with that of the microbes that reliably 

inhabit them [1, 189-193], coming to depend on microbial cues for phenotypic responses 

to changes in the environment [1, 103, 194-197]. The host’s social environment, for 

example, can profoundly impact the gut microbiota [198], which in turn modulates host 

phenotypes [109-111]. Not only are microbial lineages differentially transmitted 

through social contact [161, 199], but increased transmission among members of the 

social network homogenizes gut communities and increases their richness, while social 



 

56 

isolation limits microbial dispersal and increases inter-individual heterogeneity in gut 

community composition [200-203]. In social rodents, prolonged social isolation has also 

been shown to act as a stressor [109, 110, 204, 205], resulting in gut inflammation and 

an altered taxonomic composition of the gut microbiota [109, 110]. Whereas the effects 

of the host’s social environment on the taxonomic composition of the gut microbiota 

has been well-studied, relatively little is known about how the social environment of the 

host affects evolutionary dynamics within gut bacterial species.  

Bacteria in the mammalian gut microbiota have several characteristics that 

potentiate rapid adaptation to changes in the environment, such as short generation times 

[206], large population sizes [207], and high mutation [208, 209] and recombination 

rates [210, 211], which combine to generate high intra-specific genomic diversity [212-

215]. Recent work has begun investigating the evolution of gut microbial lineages in 

real time by, for example, co-culturing complex communities in vitro [216, 217] or 

tracking specific lineages in vivo [209, 218-222]. However, studies that measure 

evolutionary processes occurring simultaneously in more than a few species colonizing 

the gut are still rare [210, 223, 224].  

Here we leveraged genome-resolved metagenomic data from male and female 

wild-derived mice (Mus musculus domesticus) that were socially isolated or co-housed 

with a littermate for 30 days to investigate how changes in the host social environment 

affect the evolutionary dynamics in the gut microbiota. We hypothesized that social 

isolation would increase the amount of intra-specific genomic change in the gut 

microbiota of socially isolated mice—given the change in selective environment (e.g., 

gut inflammation) and the limited social transmission, which may reduce the effective 



 

57 

population size of a given lineage, increasing the amount of genetic drift—ultimately 

increasing genomic heterogeneity in the littermate gut metacommunity. We found that 

social isolation accelerated the evolution of gut bacterial lineages, increasing the 

genomic divergence between lineages in single-housed mice and their littermates 

relative to the same lineages in co-housed pairs. These results show that the host social 

environment impacts evolutionary dynamics in the gut microbiota. 

 

3.4. Results & Discussion 

3.4.1. Measurable intra-specific diversity in the native mouse gut microbiota 

We investigated the effects of social isolation on the evolution of bacterial lineages in 

the gut microbiota of wild-derived mice by leveraging a genome-resolved metagenomic 

dataset generated in Chapter 2. Briefly, 6-week-old littermates were co-housed or 

single-housed for 30 days. We sequenced fecal samples collected at the beginning (D0) 

and end (D30) of the experiment, yielding on average 33.6 million (SD = 5.0 million) 

reads/sample. We assembled and binned our reads into metagenome-assembled 

genomes (MAGs) using a custom workflow [139], resulting in a total of 6,404 MAGs, 

over half of which were high-quality (3,542 MAGs with >90% completeness and <5% 

contamination). We then created a custom database with 173 representative species-

level bins (SGBs), comprised of high-quality MAGs de-replicated at 95% average 

nucleotide identity (ANI) and classified with the Genome Taxonomy Database (GTDB) 

[140, 141]. Our database contained representatives of 15 distinct bacterial orders (Figure 

3.1A). We profiled our samples by mapping all sequenced reads to our custom SGB 

database, with a mean of 75.75% (SD = 3.04%) mapped reads per sample, indicating 



 

58 

that our custom database was broadly representative of the genomic diversity in the full 

dataset. We then used InStrain [142] to calculate the abundance of our representative 

SGBs in each sample and quantify their intra-specific genomic diversity. The latter was 

achieved by using InStrain Compare to calculate popANI: the average nucleotide 

distance between two samples for a genome, considering only as differences genomic 

positions where none of the population alleles were shared. For each of our SGBs, we 

created a distance matrix of popANI values between our samples and identified strains 

by creating hierarchical clusters at 99.999% ANI. These analyses allowed us to compare 

the degree of genomic similarity of dominant strains among pairs of samples. 

We found the native gut microbiota of wild-derived mice displayed measurable intra-

specific genomic diversity (Figure 3.1A), with SGBs in our dataset clustering on 

average into 22.24 (SD = 22.26) distinct strains. The number of unique strains per SGB 

varied between taxonomic groups, with Clostridia bacteria, such as SGBs in the order 

Oscillospirales and Lachnospirales, containing more strains per SGB than SGBs in the 

order Bacteroidales (Figure 3.1C; Kruskal-Wallis test p-value < 0.001; post-hoc Dunn’s 

test FDR-adjusted p-value < 0.05). We also found significant variation in the prevalence 

of strains in the metacommunity (Figure 3.1A). While most strains for each SGB were 

found in only one sample, 30 strains were found in 50 or more samples, and 7 SGBs 

contained a single dominant strain that was found in all samples (Figure 3.1B). All the 

latter SGBs belonged to the order Bacteroidales, which, along with SGBs from order 

Christensenellales, displayed significantly higher mean strain prevalence than SGBs 

from other orders (Figure 3.1C; Kruskal-Wallis test p-value < 0.001; post-hoc Dunn’s 

test FDR-adjusted p-value < 0.05). The mean prevalence of strains in each SGB was 



 

59 

positively associated with that SGB’s mean relative abundance (Figure 3.1D; linear 

regression p-value < 0.001 and R2 = 0.47), with SGB103, identified as the Bacteroidales 

Alistipes muris, displaying the highest mean relative abundance (M = 6.19%, SD = 

2.04%) and containing a single dominant strain found in all samples. Bacteroidales 

bacteria have been shown to be aerotolerant [225], which may promote dispersal and 

enable strains to proliferate to every host patch in the metacommunity, as was observed 

for 7 Bacteroidales SGBs (Figure 3.1B). In contrast, members of the class Clostridia, 

such as Oscillospirales and Lachnospirales, are obligate anaerobes [226, 227], which 

may limit their dispersal through the metacommunity and promote intra-specific 

diversity (higher number of unique strains per SGB). Additionally, the observed positive 

association between an SGB’s relative abundance and the prevalence of its strains 

(Figure 3.1D) indicates that abundant species increase their dispersal potential or, 

alternatively, that good dispersers are also better competitors in the gut of individual 

hosts. Overall, these results show high inter-specific variation in the amount of intra-

specific diversity in the native gut microbiota of house mice that is potentially associated 

with bacterial lifestyle.  

  



 

60 



 

61 

 

Figure 3.1. Measurable intra-specific 

diversity in the native mouse gut 

microbiota. (A) The phylogenetic tree 

shows the evolutionary relationship 

(amino acid substitutions) between 

representative SGBs in our custom 

database. Tips are colored by order. The 

bar plots show, from left to right, the 

mean relative abundance (%) of each 

SGB across all samples, the number of 

strains within each SGB (number of 

distinct 99.999% ANI clusters), and the 

mean strain prevalence within each 

SGB (average number of samples 

strains were observed in), colored by 

order. (B) The histogram shows the 

total strain prevalence in our samples, 

with the total number of strains (×103) 

observed in a given number of samples. 

The embedded tiles indicate SGBs 

containing strains that were observed in 

at least 50 samples (grey dashed line) or 

in all samples (red dashed line). Tiles 

are colored by order and labeled with 

the SGB the strain belonged to. (C) The 

box plots show the number of strains 

per SGB (left) and the mean strain 

prevalence per SGB (right) in orders 

with more than one representative SGB. 

Significant pairwise comparisons are 

denoted with different letters (Dunn’s 

test BH-adjusted p-value < 0.05). (D) 

The scatter plot shows the association 

between the mean SGB relative 

abundance (%) and the mean 

prevalence of strains within that SGB, 

colored by order. The linear 

regression’s p-value and R2 are 

annotated at the top. 

 



 

62 

3.4.2. Social isolation accelerates divergent evolution in the gut microbiota 

We next investigated the effects of social isolation on the evolution of gut commensals. 

We hypothesized that social isolation would accelerate the evolution of gut 

commensals, given the changes in the gut environment and the disruption of social 

transmission. We tested this hypothesis by calculating the average intra-individual 

genomic divergence (1 – popANI) in a given SGB between the beginning (D0) and end 

(D30) of the experiment. We found that SGBs in single-housed mice underwent, on 

average, a significantly greater amount of genomic change over the 30 days of the 

experiment than did their conspecifics in co-housed mice (Figure 3.2A; paired 

Wilcoxon signed-rank test p-value = 0.006), such that a majority of SGBs (57.36%) 

displayed higher mean genomic divergence in single-housed mice than in co-housed 

mice. Additionally, we identified 15 SGBs whose mean intra-individual genomic 

divergence in single-housed mice was greater than 0.001%. These likely represent strain 

replacement events. These SGBs were on average 3.5 times more genomically diverged 

in socially isolated mice than their conspecifics in co-housed mice. Interestingly, all 

these SGBs, except for the Bacteroidales SGB082, were members of the class 

Clostridia. As previously mentioned, these bacteria are obligate anaerobes [226, 227], 

making it unlikely that the observed replacement events result from the invasion of 

strains from the broader environment. Instead, it is possible that these strains were at 

very low abundance at D0 but became the dominant strain after 30 days of social 

isolation. These results show that social isolation significantly accelerated the intra-host 

evolution of bacterial species in the gut microbiota, likely through selection on standing 

strain variation, and that this effect was particularly noticeable in obligate anaerobes. 



 

63 

Given our observation that SGBs in single-housed mice undergo greater 

evolutionary change than their conspecifics in co-housed mice, we investigated the 

effects of social isolation on SGB divergence between different individuals. We 

hypothesized that social isolation reduces dispersal opportunities for bacteria in the gut 

microbiota, particularly those that rely on social transmission. This would create a 

subdivision of the gut metacommunity of littermates, with SGBs transmitted between 

co-housed littermates remaining more genomically similar but becoming increasingly 

diverged from their conspecifics in single-housed littermates. We tested this hypothesis 

by calculating the mean genomic divergence in the same SGB within co-housed 

littermates and between single-housed and co-housed littermates. We found that, by end 

of the 30-day experimental period, SGBs in single-housed mice were significantly more 

genetically diverged from conspecifics in co-housed mice than the same SGBs in co-

housed mice were to each other (Figure 3.2B; paired Wilcoxon signed-rank test p-value 

= 0.033). Crucially, there were no such differences at D0. Additionally, conspecific 

SGBs in different single-housed littermates were also significantly more diverged than 

conspecific SGBs in co-housed mice (Figure 3.2B; paired Wilcoxon signed-rank test p-

value = 0.002). Thus, social isolation caused SGBs not only to diverge from those in 

co-housed littermates, but also from those in other socially isolated mice. These results 

indicate that social isolation increased the genomic diversity without parallel evolution 

between conspecific bacteria in different hosts, while co-housing homogenized and 

stabilized the intra-specific genomic diversity of the gut microbiota. 

 



 

64 

 

Figure 3.2. Social isolation accelerates divergent evolution in the gut microbiota. 

(A) The box plot shows the mean genomic divergence (1 – popANI) within the same 

SGB between the beginning (D0) and end (D30) of the experiment in co-housed (blue) 

and single-housed mice (pink). Each point represents an SGB, colored by order. SGBs 

are connected to their conspecific in the other housing group by a line, colored pink if 

the SGB shows on higher divergence in single-housed mice, or blue if higher in co-

housed mice. The dashed grey line indicates our strain-clustering threshold (99.999% 

ANI), meaning that points above this value represent strain replacement events. The p-

value resulting from a paired Wilcoxon signed-rank test is shown at the top. (B) The 

box plots show the mean genomic divergence between conspecific SGBs in co-housed 

mice (blue), single- and co-housed mice (violet), and single-housed mice (pink), at the 

beginning (D0) and end (D30) of the experiment. SGBs (points) are colored by order 

and connected across comparisons by a grey line. The p-values resulting from paired 

Wilcoxon signed-rank tests are shown at the top. 

  



 

65 

3.5. Conclusion 

We leveraged a previously created genome-resolved metagenomic longitudinal dataset 

to investigate the effects of social isolation on the evolution of gut microbes. We found 

that the native gut microbiota of house mice hosts extensive intra-specific diversity, 

which varied between taxonomic groups with differing lifestyles and dispersal 

potentials. Additionally, we found that this intra-specific diversity was increased by 

single-housing mice, such that social isolation accelerated the genomic evolution of gut 

bacteria. Interestingly, this genomic change did not occur in parallel. Instead, social 

isolation led bacterial lineages on divergent evolutionary trajectories, further increasing 

intra-specific diversity in the metacommunity—likely impacting the host’s ability to 

cope with novel challenges.  

 

3.6. Methods 

3.6.1. Animals 

The mice used in this experiment belonged to a wild-derived M. m. domesticus line 

captured in Saratoga Springs (NY, USA) in 2012 and inbred in captivity since [167]. 

We kept mice on an inverted light cycle (light:10pm-10am), with all handling conducted 

during the active (dark) phase under red light. Standard chow diet and water were 

available ad libitum. All procedures conformed to guidelines established by the U.S. 

National Institutes of Health and have been approved by the Cornell University 

Institutional Animal Care and Use Committee (protocol #2015-0060). 

 

 



 

66 

3.6.2. Social Isolation 

We weaned mice at 3 weeks of age (M = 22, SD = 2 days) and co-housed them with 

same-sex littermates. At 6 weeks of age (M = 43, SD = 6 days), littermates were 

separated into single- and co-housed mice, in a total of 8 male and 7 female co-housed 

pairs, and 7 male and 10 female single-housed mice. They remained under these housing 

conditions for the following 30 (SD = 3) days. As part of Chapter 2, half of the mice 

were exposed to an additional stressor (predator odor), while the other half were used 

as controls (5 male and 4 female co-housed pairs, and 3 male and 5 female single-housed 

mice). We collected fecal samples at the beginning (D0) and end of the experiment 

(D30) and stored them at -80°C. 

 

3.6.3. Metagenomic Data  

We used the genome-resolved metagenomic dataset generated in Chapter 2. Briefly, 

fecal samples were sequenced and assembled into MAGs using a custom snakemake 

[172] workflow (MAGmaker, available at https://github.com/CUMoellerLab/sn-mg-

pipeline). We used the resulting high-quality MAGs (>90% completeness, <5% 

contamination) to create a custom genome database of species-genome bins (SGBs) by 

dereplicating MAGs into 95% average nucleotide identity bins using dRep (v3.4.2) 

[183]. SGBs were taxonomically profiled using GTDB-Tk (v2.4.0) [141] and we 

calculated their relative abundance in our samples by mapping all metagenomic reads 

against our SGB database with Bowtie2 and profiling them with InStrain (v1.8.0) [142]. 

We normalized the resulting mapped reads count by the respective metagenomic 

sample’s library size and the respective SGB’s genome size and converted it into a 

https://github.com/CUMoellerLab/sn-mg-pipeline
https://github.com/CUMoellerLab/sn-mg-pipeline


 

67 

percentage. We assessed the genomic divergence in the same SGB between different 

samples with InStrain Compare (settings: --breadth 0.05 --coverage 0.0025 --

group_length 10000000 --database_mode -ani 0.95). 

 

3.6.5. Statistical Analyses 

All data analyses were conducted in R (v4.2.2). Data are presented as mean (M) and 

standard deviation (SD). Throughout the analyses, a p-value < 0.05 was considered 

statistically significant. For the genomic divergence analyses we only included control 

mice (i.e., not exposed to predator odor). All plots were created with ggplot2 (v3.4.1) 

[83]. All figure panels were assembled using Inkscape 1.2. 

 

  



 

68 

APPENDIX 

1.1. Supplemental Results 

To investigate the effect of sampling bias on the detection of significant effects of Mup 

deletion on the microbiota in males but not in females, we subsampled our WT males 

by alternatively removing one male from the analyzes and testing for differences in 

microbiota beta and alpha diversity and differential abundance at both taxonomic and 

functional (COG) profiles. PERMANOVA analyses testing for differences in species 

composition in the gut microbiota of WT and KO males revealed that significance was 

maintained in 3/4 cases based on Jaccard dissimilarities, and in 1/4 cases using Bray-

Curtis dissimilarities (Table S1.7). In terms of COG-function composition, significance 

was maintained in 2/4 subsampling analyses based on Bray-Curtis dissimilarities but 

not maintained for analyses based on Jaccard dissimilarities (Table S1.7). Significant 

differences in microbial family Shannon diversity were observed for 1/4 cases (Table 

S1.8), but not for alpha diversity measures based on COG functions (Table S1.8). 

Importantly, subsampling had no qualitative effect on the detection of differentially 

abundant taxa or functions in males. The significant effects of Mup deletion in males 

observed in these subsampling analyses—particularly the male-specific differential 

taxon and COG-function results, which were completely robust to subsampling—lend 

further support to our conclusions that Mup deletion more strongly affected the 

microbiota of male mice than female mice.  

 

  



 

69 

1.2. Supplemental Figures 

 

Figure S1.1 Mup genotype birth ratio. Bar plot shows the total percentage of male 

(top bar) and female (bottom bar) wildtype (blue), heterozygote (red), and knockout 

(yellow) pups born in the six study litters.   



 

70 

 
Figure S1.2. Mup deletion significantly changes the gut microbial taxonomic 

composition of mature males. Principal coordinates analyses (PCoA) show the 

ordinated Jaccard and Bray-Curtis dissimilarities in Species, Genus, and Family 

composition in the fecal metagenome of sexually mature mice. The ordinated points are 

faceted by sex and colored by genotype, showing the taxonomic profiles of WT (blue) 

and KO (yellow) male (left column) and female (right column) mice. The percentage of 

variation in the taxonomic dissimilarity matrix explained by the first two PCo axes is 

enclosed in parenthesis.  

 

 



 

71 

 

 

Figure S1.3. Mup deletion significantly changes the gut microbial functional 

composition of mature males. Principal coordinates analyses (PCoA) show the 

ordinated Jaccard and Bray-Curtis dissimilarities in COG Functions composition in the 

fecal metagenome of sexually mature mice. The ordinated points are faceted by sex and 

colored by genotype, showing the functional profiles of WT (blue) and KO (yellow) 

male (left column) and female (right column) mice. The percentage of variation in the 

taxonomic dissimilarity matrix explained by the first two PCo axes is enclosed in 

parenthesis.  



 

72 

 
Figure S1.4. Mup deletion significantly reduces microbial family diversity in 

mature males. Box plots show the Shannon Diversity (top row), Pielou’s Evenness 

(middle row), and Observed Richness (bottom row) in microbial species (left column), 

genus (middle column), and families (right column) in the gut microbiota of mature 

males (left sub-column) and females (right sub-column). The linear mixed-effects 

model analyses (center top) indicate whether there is a significant difference in diversity 

in the gut microbiota of the WT (blue) and KO (yellow) mice (* = p-value < 0.05; ns = 

p-value > 0.05).    

 



 

73 

 

Figure S1.5. Mup deletion significantly reduces microbial functional diversity in 

mature males. Box plots show the Shannon Diversity (left column), Pielou’s Evenness 

(middle column), and Observed Richness (right column) in microbial COG Functions 

in the gut microbiota of mature males (left sub-column) and females (right sub-column). 

The linear mixed-effects model analyses (center top) indicate whether there is a 

significant difference in diversity in the gut microbiota of the WT (blue) and KO 

(yellow) mice (* = p-value < 0.05; ns = p-value > 0.05).  



 

74 

 

Figure S1.6. Mup deletion significantly shifts the abundance of various microbial 

taxa. Volcano plots show the log2-transformed fold change (LFC) in the abundance of 

genus (top row) and families (bottom row) in the gut microbiota of mature male (left 

column) and female (right column) mice. ANCOM-BC2 analyses identified taxa 

(points) that were significantly more abundant in WT (blue) or KO mice (yellow). Red 

lines mark the significance threshold (Holm–Bonferroni–adjusted p-value < 0.001). The 

y-axis indicates the -log10 transformation of the non-adjusted p-value. 

 

  



 

75 

1.3. Supplemental Tables 

Table S1.1. Effect of host sex on microbial taxonomic and functional composition (p-

value). 

  WT KO 

Rank Dissimilarity PERMANOVA PERMDISP PERMANOVA PERMDISP 

Species  
Jaccard 0.030 0.324 0.057 0.900 

Bray-Curtis 0.061 0.284 0.025 0.423 

Genus 
Jaccard 0.019 0.291 0.014 0.840 

Bray-Curtis 0.087 0.282 0.007 0.385 

Family  
Jaccard 0.062 0.675 0.031 0.557 

Bray-Curtis 0.035 0.806 0.047 0.441 

COG 

Function 

Jaccard 0.089 0.282 0.009 0.969 

Bray-Curtis 0.050 0.188 0.057 0.090 

 

Table S1.2. Effect of Mup genotype on microbial taxonomic and functional 

composition (p-value). 

  Male Female 

Rank Dissimilarity PERMANOVA PERMDISP PERMANOVA PERMDISP 

Species  
Jaccard 0.037 0.273 0.154 0.739 

Bray-Curtis 0.019 0.188 0.158 0.398 

Genus 
Jaccard 0.025 0.175 0.171 0.935 

Bray-Curtis 0.009 0.333 0.146 0.397 

Family  
Jaccard 0.197 0.174 0.191 0.947 

Bray-Curtis 0.061 0.126 0.404 0.169 

COG 

Function 

Jaccard 0.021 0.046 0.116 0.633 

Bray-Curtis 0.012 0.080 0.481 0.200 

 

Table S1.3. Microbial taxonomic and functional profiles correspondence with 

Procrustes (p-value). 

Rank Dissimilarity Male Female 

Species + COG  
Jaccard 0.002 0.953 

Bray-Curtis 0.001 0.230 

Genus + COG 
Jaccard 0.001 0.943 

Bray-Curtis 0.001 0.221 

Family + COG  
Jaccard 0.001 0.866 

Bray-Curtis 0.001 0.920 

 

 



 

76 

Table S1.4. Effect of Mup genotype on microbial taxonomic and functional diversity 

(p-value). 

  Male Female 

Rank Metric p-value Estimate p-value Estimate 

Species  

Shannon 0.208 0.130 0.980 0.007 

Richness 0.320 8.534 0.844 -4.424 

Evenness 0.330 0.016 0.609 0.018 

Genus 

Shannon 0.371 0.092 0.895 0.036 

Richness 0.356 3.495 0.933 0.745 

Evenness 0.428 0.019 0.693 0.017 

Family  

Shannon 0.041 0.242 0.537 0.164 

Richness 0.481 -1.390 0.968 -0.093 

Evenness 0.041 0.078 0.455 0.052 

COG 

Function 

Shannon 0.050 0.041 0.484 -0.032 

Richness 0.594 10.490 0.476 -41.561 

Evenness 0.314 0.002 0.642 -0.001 

 

Table S1.5. Microbial taxa and functions showing differential abundance in WT vs. KO 

genotypes* 

Rank Sex Taxa / COG Function p-value LFC Group 

Species  
Male SGB43260 1.078E-05 2.595 WT 

Female SGB41239 4.657E-03 -2.784 KO 

Genus 
Male GGB30296 6.324E-06 2.584 WT 

Female GGB28621 4.286E-03 -2.781 KO 

COG 

Function 

Male Na+/pantothenate symporter 2.874E-05 2.796 WT 

Male 
Uncharacterized protein, possibly 

involved in motility 
1.129E-03 1.667 WT 

Male 
Predicted hydrocarbon binding 

protein (contains V4R domain) 
1.042E-02 1.434 WT 

Male L-asparaginase II 1.534E-08 1.427 WT 

Male 
Trans-aconitate 

methyltransferase 
2.487E-02 1.427 WT 

Male Cu/Zn superoxide dismutase 1.102E-04 1.390 WT 

Male 
V8-like Glu-specific 

endopeptidase 
1.475E-04 1.385 WT 

Male 
Type III secretory pathway, 

component EscU 
2.440E-09 1.308 WT 

Male 
Fucose 4-O-acetylase and related 

acetyltransferases 
4.605E-06 1.201 WT 

Male Cephalosporin hydroxylase 5.297E-06 1.104 WT 

Male 
Uncharacterized protein required 

for formate dehydrogenase 
1.398E-05 1.101 WT 



 

77 

activity 

Male Sugar diacid utilization regulator 3.722E-02 1.095 WT 

Male 
Sulfite oxidase and related 

enzymes 
1.690E-02 0.989 WT 

Male Precorrin-2 methylase 3.460E-03 0.850 WT 

Male 

Distinct helicase family with a 

unique C-terminal domain 

including a metal-binding 

cysteine cluster 

1.379E-03 0.833 WT 

Male 

Dissimilatory sulfite reductase 

(desulfoviridin), alpha and beta 

subunits 

3.145E-02 0.788 WT 

Male 
Uncharacterized protein 

conserved in cyanobacteria 
3.573E-04 0.725 WT 

Male 

Predicted transcriptional 

regulators containing the 

CopG/Arc/MetJ DNA-binding 

domain 

1.039E-04 -0.819 KO 

Female 

Glycosylphosphatidylinositol 

transamidase (GPIT), subunit 

GPI8 

3.456E-03 2.142 WT 

Female Uroporphyrinogen-III synthase 7.652E-03 -1.781 KO 

Female Arginine kinase 2.754E-03 -1.915 KO 

* Highly significant results are bolded (Holm–Bonferroni adjusted p-value < 0.001). 

 

Table S1.6. Functional enrichment analyses with a hypergeometric test (p-value). 

 

 Male Female 

COG Category WT KO WT KO 

Cellular processes and 

Signaling 

0.545 1.000 0.205 1.000 

Information storage and 

Processing 

1.000 0.170 1.000 1.000 

Metabolism 0.425 1.000 1.000 0.189 

Poorly characterized 0.251 1.000 1.000 1.000 

 

  



 

78 

Table S1.7. Power analysis of taxonomic and functional beta diversity results (p-

value). 

Rank Dissimilarity w/o 219 w/o 224 w/o 256 w/o 257 

Species  
Jaccard 0.014 0.039 0.058 0.041 

Bray-Curtis 0.119 0.069 0.153 0.046 

Genus 
Jaccard 0.052 0.549 0.118 0.097 

Bray-Curtis 0.017 0.097 0.077 0.029 

Family  
Jaccard 0.320 0.138 0.100 0.431 

Bray-Curtis 0.084 0.126 0.053 0.175 

COG 

Function 

Jaccard 0.051 0.089 0.108 0.171 

Bray-Curtis 0.032 0.018 0.052 0.120 

 

Table S1.8. Power analysis of taxonomic and functional alpha diversity results (p-

value). 

Rank Metric w/o 219 w/o 224 w/o 256 w/o 257 

Species  

Shannon 0.209 0.806 0.567 0.565 

Richness 0.613 0.053 0.685 0.677 

Evenness 0.259 0.102 0.024 0.034 

Genus 

Shannon 0.331 0.567 0.859 0.868 

Richness 0.795 0.050 0.522 0.516 

Evenness 0.252 0.138 0.025 0.027 

Family  

Shannon 0.034 0.184 0.077 0.117 

Richness 0.185 0.084 0.629 0.494 

Evenness 0.145 0.229 0.095 0.133 

COG 

Function 

Shannon 0.087 0.057 0.102 0.152 

Richness 0.373 0.559 0.558 0.546 

Evenness 0.802 0.141 0.303 0.399 

 

 



 

79 

2.1. Supplemental Figures 

 

Figure S2.1. Pair-housed mice exposed to predator odor displayed increased 

sociability. Box plots show the amount of time that male (triangles) and female 

(squares) mice of different treatment groups spent interacting with a social stimulus (a 

stranger mouse) vs. a non-social stimulus (an empty cup) in the three-chamber test 

(3CT). Differences between the social and non-social interaction time within each 

treatment group were tested with a paired Wilcoxon test. Significant results (p-value < 

0.05) are shown in bold.  



 

80 

 

Figure S2.2. Predator-odor exposure altered VAT gene expression. Principal 

coordinate analysis (PCoA) shows the first two axes of the ordinated Euclidean distance 

matrix of the VAT transcriptome at the end of the experiment (D33). Points represent 

the transcriptome of male (triangles) and female (squares) mice, colored by treatment 

group: blue, Pair H2O; yellow, Pair TMT; pink, Single H2O; orange, Single TMT. A 

grey line connects pair-housed cage-mates. The percent variation explained by each axis 

is enclosed in parentheses. 

  



 

81 

 

Figure S2.3. Predator odor caused a larger mean fold-change in gene expression 

than did social isolation. Box plot shows the absolute log2(fold-change) in gene 

expression of DEGs in the VAT of pair-housed mice exposed to TMT (yellow) and 

single-housed mice exposed to H2O (pink), when compared to unstressed controls (Pair 

H2O). The difference between the groups was tested with a Wilcoxon test, and the 

resulting p-value shown on top of the plot. 

  



 

82 

 

Figure S2.4. Social isolation increases SGB turnover. Box plot shows the intra-

individual SGB Jaccard dissimilarity of male (triangles) and female (squares) mice 

between the beginning (D0) and end (D30) of the experiment. A linear mixed-effects 

model (LMM) accounting for sex, litter and cage effects showed a significant difference 

between pair and single-housed mice (p-value shown on top).   



 

83 

2.2. Supplemental Tables 

Table S2.1. Results from linear mixed-effects model (LMM) showing effects of sex, 

predator odor (TMT), social isolation (Housing), or their interaction on mouse behavior. 

Significant differences (p-value < 0.05) are in bold. 

 

Behavior Factor Estimate SE DF t-value p-value 

Center 

Occupancy 

Sex 0.008 0.027 40.682 0.298 0.767 

Housing -0.084 0.041 39.285 -2.039 0.048 

TMT -0.093 0.038 26.584 -2.426 0.022 

Housing:TMT 0.105 0.057 41.944 1.822 0.076 

Grooming 

Duration 

Sex 23.881 11.801 35.951 2.024 0.051 

Housing -44.327 20.299 41.979 -2.184 0.035 

TMT -26.043 20.615 41.058 -1.263 0.214 

Housing:TMT 30.741 26.866 40.768 1.144 0.259 

Social 

Preference 

Sex -0.105 0.126 40.393 -0.829 0.412 

Housing 0.084 0.194 39.010 0.432 0.668 

TMT 0.272 0.182 26.259 1.492 0.148 

Housing:TMT -0.598 0.270 41.945 -2.212 0.032 

  



 

84 

Table S2.2. Results from pairwise LMM showing the differences in behavior between 

the different stressor treatment combinations. Significant differences (p-value < 0.05) 

are in bold. 

Behavior Contrast Estimate SE DF p-value 

Center 

Occupancy 

Pair H2O - Pair TMT 0.093 0.038 2.426 0.015 

Pair H2O - Single H2O 0.084 0.041 2.039 0.041 

Pair H2O - Single TMT 0.072 0.039 1.831 0.067 

Pair TMT - Single H2O -0.009 0.044 -0.202 0.840 

Pair TMT - Single TMT -0.021 0.042 -0.502 0.616 

Single H2O - Single TMT -0.012 0.044 -0.274 0.784 

Grooming 

Duration 

Pair H2O - Pair TMT 26.043 20.615 1.263 0.206 

Pair H2O - Single H2O 44.327 20.299 2.184 0.029 

Pair H2O - Single TMT 39.629 19.204 2.064 0.039 

Pair TMT - Single H2O 18.284 21.045 0.869 0.385 

Pair TMT - Single TMT 13.586 19.098 0.711 0.477 

Single H2O - Single TMT -4.698 18.213 -0.258 0.796 

Social 

Preference 

Pair H2O - Pair TMT -0.272 0.182 -1.492 0.136 

Pair H2O - Single H2O -0.084 0.194 -0.432 0.666 

Pair H2O - Single TMT 0.243 0.185 1.312 0.190 

Pair TMT - Single H2O 0.188 0.210 0.896 0.370 

Pair TMT - Single TMT 0.514 0.197 2.615 0.009 

Single H2O - Single TMT 0.326 0.205 1.596 0.110 



 

85 

Table S2.3. Results from PERMANOVA and PERMDISP analyses showing the 

effect of predator odor (TMT) and social isolation (Housing) on the Euclidean 

distances of the visceral adipose tissue (VAT) transcriptome. Significant differences 

(p-value < 0.05) are in bold. 

 

Test Factor DF 
Sum of 

Squares 
R2 

F 

statistic 

p-

value 

PERMANOVA 

Sex 1 16444.484 0.036 1.666 0.099 

Litter 10 195428.974 0.430 1.980 0.002 

Housing 1 16593.089 0.037 1.681 0.097 

TMT 1 21563.061 0.047 2.185 0.035 

Residual 20 197377.341 0.434   

Total 33 454424.543 1.000   

PERMDISP 

Housing 1 145.910 145.910 0.242 0.647 

Housing:Residuals 32 19273.247 602.289   

TMT 1 264.474 264.474 0.498 0.505 

TMT:Residuals 32 16998.042 531.189   

 



 

86 

Table S2.4. Significant results (FDR p-value < 0,05) from the biological pathway 

enrichment analyses of differentially expressed genes (DEGs) in the VAT of TMT-

exposed (Pair TMT) and single-housed mice (Single H2O) when compared to 

unstressed controls (Pair H2O). The gene ontology (GO) containing genes involved in 

antimicrobial humoral immune response mediated by antimicrobial peptides (AMPs) 

is in bold. 

 

Comparison GO ID Description GeneRatio BgRatio p-value 
FDR p-

value 

Pair H2O - 

Pair TMT 

GO:0031016 
pancreas 

development 
7/127 75/17475 

1.194E-

06 
0.002 

GO:0031638 
zymogen 

activation 
6/127 66/17475 

8.321E-

06 
0.007 

GO:0061844 

antimicrobial 

humoral 

immune 

response 

mediated by 

antimicrobial 

peptide 

6/127 75/17475 
1.748E-

05 
0.007 

GO:0031018 

endocrine 

pancreas 

development 

5/127 45/17475 
1.813E-

05 
0.007 

GO:0016485 
protein 

processing 
9/127 249/17475 

8.619E-

05 
0.028 

GO:0043588 
skin 

development 
9/127 266/17475 

1.422E-

04 
0.036 

GO:0007586 digestion 6/127 113/17475 
1.752E-

04 
0.036 

GO:0035270 

endocrine 

system 

development 

6/127 113/17475 
1.752E-

04 
0.036 

GO:0019730 

antimicrobial 

humoral 

response 

6/127 116/17475 
2.022E-

04 
0.037 

Pair H2O - 

Single H2O 

GO:0007586 digestion 7/75 113/17475 
5.414E-

07 
0.001 

GO:0007631 
feeding 

behavior 
6/75 109/17475 

7.279E-

06 
0.004 

GO:0031016 
pancreas 

development 
5/75 75/17475 

1.737E-

05 
0.004 

GO:0061844 

antimicrobial 

humoral 

immune 

response 

5/75 75/17475 
1.737E-

05 
0.004 



 

87 

mediated by 

antimicrobial 

peptide 

GO:0019730 

antimicrobial 

humoral 

response 

5/75 116/17475 
1.408E-

04 
0.029 

GO:0031638 
zymogen 

activation 
4/75 66/17475 

1.844E-

04 
0.031 

 



 

88 

Table S2.5. Results from random forest models testing the predictive power of SGB 

abundances at D22 and VAT transcriptome on host phenotypes. Significant results (p-

value < 0.05) are in bold. 

 

Predictor Phenotype Variable 
Permutation 

test R2 
p-value 

Gut microbiota Behavior 
Center 

occupancy 
-0.545 0.623 

Gut microbiota Behavior 
Grooming 

duration 
0.021 0.016 

Gut microbiota Behavior 
Social 

preference 
0.102 0.001 

Gut microbiota 
VAT 

transcriptome 
PC1 0.008 0.033 

Gut microbiota 
VAT 

transcriptome 
PC2 -0.143 0.120 

Gut microbiota 
VAT 

transcriptome 
PC3 -0.671 0.653 

Gut microbiota 
VAT 

transcriptome 
PC4 -0.387 0.109 

VAT 

transcriptome 
Behavior 

Center 

occupancy 
-0.899 0.835 

VAT 

transcriptome 
Behavior 

Grooming 

duration 
-0.467 0.454 

VAT 

transcriptome 
Behavior 

Social 

preference 
-0.902 0.852 

  



 

89 

Table S2.6. Summary results of predictive power (Gini Importance) of SGBs that 

were persistently associated with TMT exposure (DA at both D22 and D30). SGBs 

that ranked in the top half most important features are in bold. 

Phenotype Variable SGB ID 
Gini 

Importance 
Rank 

Behavior 

Grooming 

Duration 

SGB107 2.74E-03 60 

SGB158 7.97E-04 83 

SGB051 6.47E-04 89 

SGB074 6.12E-04 90 

SGB108 3.58E-04 103 

SGB002 1.72E-05 148 

SGB105 0.00E+00 172 

Social Preference 

SGB002 1.80E-02 20 

SGB105 7.17E-03 39 

SGB107 3.04E-03 63 

SGB158 2.33E-03 71 

SGB074 2.77E-04 116 

SGB108 0.00E+00 167 

SGB051 0.00E+00 168 

VAT PC1 

SGB105 5.96E-03 34 

SGB158 6.60E-04 77 

SGB107 5.91E-04 79 

SGB002 1.32E-04 101 

SGB074 1.11E-04 105 

SGB108 3.32E-05 118 

SGB051 0.00E+00 169 

  



 

90 

 

REFERENCES 

1. Chevalier, C., et al., Gut Microbiota Orchestrates Energy Homeostasis during 

Cold. Cell, 2015. 163(6): p. 1360-74. 

2. Bäckhed, F., et al., The gut microbiota as an environmental factor that 

regulates fat storage. Proceedings of the national academy of sciences, 2004. 

101(44): p. 15718-15723. 

3. Turnbaugh, P.J., et al., Diet-induced obesity is linked to marked but reversible 

alterations in the mouse distal gut microbiome. Cell host & microbe, 2008. 

3(4): p. 213-223. 

4. Tremaroli, V. and F. Bäckhed, Functional interactions between the gut 

microbiota and host metabolism. Nature, 2012. 489(7415): p. 242-249. 

5. Moreno-Navarrete, J.M. and J.M. Fernandez-Real, The gut microbiota 

modulates both browning of white adipose tissue and the activity of brown 

adipose tissue. Reviews in Endocrine and Metabolic Disorders, 2019. 20: p. 

387-397. 

6. Sarkar, A., et al., The Microbiome in Psychology and Cognitive Neuroscience. 

Trends Cogn Sci, 2018. 22(7): p. 611-636. 

7. Sarkar, A., et al., The role of the microbiome in the neurobiology of social 

behaviour. Biol Rev Camb Philos Soc, 2020. 95(5): p. 1131-1166. 

8. Diaz Heijtz, R., et al., Normal gut microbiota modulates brain development 

and behavior. Proc Natl Acad Sci U S A, 2011. 108(7): p. 3047-52. 

9. Foster, K.R., et al., The evolution of the host microbiome as an ecosystem on a 

leash. Nature, 2017. 548(7665): p. 43-51. 

10. Rehman, A., et al., Nod2 is essential for temporal development of intestinal 

microbial communities. Gut, 2011. 60(10): p. 1354-1362. 

11. Vijay-Kumar, M., et al., Metabolic syndrome and altered gut microbiota in 

mice lacking Toll-like receptor 5. Science, 2010. 328(5975): p. 228-31. 

12. Kwon, O., et al., Altered Gut Microbiota Composition in Rag1-deficient Mice 

Contributes to Modulating Homeostasis of Hematopoietic Stem and Progenitor 

Cells. Immune Netw, 2015. 15(5): p. 252-9. 



 

91 

13. Galvez, E.J.C., et al., Shaping of Intestinal Microbiota in Nlrp6- and Rag2-

Deficient Mice Depends on Community Structure. Cell Rep, 2017. 21(13): p. 

3914-3926. 

14. Sprockett, D.D., et al., Home-site advantage for host species–specific gut 

microbiota. Science Advances, 2023. 9(19): p. eadf5499. 

15. He, Q., et al., Lactobacillus casei Zhang exerts anti-obesity effect to obese 

glut1 and gut-specific-glut1 knockout mice via gut microbiota modulation 

mediated different metagenomic pathways. European Journal of Nutrition, 

2022. 61(4): p. 2003-2014. 

16. Neugebauer, K.A., et al., Baat Gene Knockout Alters Post-Natal Development, 

the Gut Microbiome, and Reveals Unusual Bile Acids in Mice. Journal of Lipid 

Research, 2022. 63(12). 

17. Niska-Blakie, J., et al., Knockout of the non-essential gene SUGCT creates 

diet-linked, age-related microbiome disbalance with a diabetes-like metabolic 

syndrome phenotype. Cellular and Molecular Life Sciences, 2020. 77: p. 3423-

3439. 

18. Chamero, P., et al., Identification of protein pheromones that promote 

aggressive behaviour. Nature, 2007. 450(7171): p. 899-902. 

19. Nevison, C., et al., The ownership signature in mouse scent marks is involatile. 

Proceedings of the Royal Society of London. Series B: Biological Sciences, 

2003. 270(1527): p. 1957-1963. 

20. Timm, D.E., et al., Structural basis of pheromone binding to mouse major 

urinary protein (MUP‐I). Protein Science, 2001. 10(5): p. 997-1004. 

21. Charkoftaki, G., et al., Update on the human and mouse lipocalin (LCN) gene 

family, including evidence the mouse Mup cluster is result of an “evolutionary 

bloom”. Human genomics, 2019. 13(1): p. 1-14. 

22. Cheetham, S.A., et al., The genetic basis of individual-recognition signals in 

the mouse. Current Biology, 2007. 17(20): p. 1771-1777. 

23. Sheehan, M.J., et al., Selection on Coding and Regulatory Variation Maintains 

Individuality in Major Urinary Protein Scent Marks in Wild Mice. PLoS 

Genet, 2016. 12(3): p. e1005891. 

24. Hurst, J.L., et al., Individual recognition in mice mediated by major urinary 

proteins. Nature, 2001. 414(6864): p. 631-634. 

25. Logan, D.W., T.F. Marton, and L. Stowers, Species specificity in major 

urinary proteins by parallel evolution. PloS one, 2008. 3(9): p. e3280. 



 

92 

26. Penn, D. and W.K. Potts, Chemical signals and parasite-mediated sexual 

selection. Trends in Ecology & Evolution, 1998. 13(10): p. 391-396. 

27. Schellinck, H.M., A.M. West, and R.E. Brown, Rats can discriminate between 

the urine odors of genetically identical mice maintained on different diets. 

Physiology & behavior, 1992. 51(5): p. 1079-1082. 

28. Green, J.P., et al., The Genetic Basis of Kin Recognition in a Cooperatively 

Breeding Mammal. Curr Biol, 2015. 25(20): p. 2631-41. 

29. Hurst, J.L., Female recognition and assessment of males through scent. 

Behavioural brain research, 2009. 200(2): p. 295-303. 

30. Ramm, S.A., S.A. Cheetham, and J.L. Hurst, Encoding choosiness: female 

attraction requires prior physical contact with individual male scents in mice. 

Proceedings of the Royal Society B: Biological Sciences, 2008. 275(1644): p. 

1727-1735. 

31. Thom, M.D., et al., The direct assessment of genetic heterozygosity through 

scent in the mouse. Curr Biol, 2008. 18(8): p. 619-23. 

32. Hurst, J.L., The functions of urine marking in a free-living population of house 

mice, Mus domesticus Rutty. Animal Behaviour, 1987. 35(5): p. 1433-1442. 

33. Thonhauser, K.E., et al., Scent marking increases male reproductive success in 

wild house mice. Animal Behaviour, 2013. 86(5): p. 1013-1021. 

34. Beynon, R.J. and J.L. Hurst, Urinary proteins and the modulation of chemical 

scents in mice and rats. Peptides, 2004. 25(9): p. 1553-63. 

35. Penn, D.J., S.M. Zala, and K.C. Luzynski, Regulation of Sexually Dimorphic 

Expression of Major Urinary Proteins. Front Physiol, 2022. 13: p. 822073. 

36. Sheehan, M.J., P. Campbell, and C.H. Miller, Evolutionary patterns of major 

urinary protein scent signals in house mice and relatives. Mol Ecol, 2019. 

28(15): p. 3587-3601. 

37. Payne, C.E., et al., Heterogeneity of major urinary proteins in house mice: 

population and sex differences. Chemical Signals in Vertebrates 9, 2001: p. 

233-240. 

38. Thoss, M., et al., Major urinary protein (MUP) profiles show dynamic changes 

rather than individual “barcode” signatures. Frontiers in ecology and 

evolution, 2015. 3: p. 71. 

39. Thoß, M., et al., Regulation of volatile and non-volatile pheromone attractants 

depends upon male social status. Scientific Reports, 2019. 9(1): p. 489. 



 

93 

40. Luzynski, K.C., et al., Pheromones that correlate with reproductive success in 

competitive conditions. Scientific Reports, 2021. 11(1): p. 21970. 

41. Hui, X., et al., Major urinary protein-1 increases energy expenditure and 

improves glucose intolerance through enhancing mitochondrial function in 

skeletal muscle of diabetic mice. J Biol Chem, 2009. 284(21): p. 14050-7. 

42. Giller, K., et al., Major urinary protein 5, a scent communication protein, is 

regulated by dietary restriction and subsequent re-feeding in mice. 

Proceedings of the Royal Society B: Biological Sciences, 2013. 280(1757): p. 

20130101. 

43. Mitchell, S.E., et al., The effects of graded levels of calorie restriction: II. 

Impact of short term calorie and protein restriction on circulating hormone 

levels, glucose homeostasis and oxidative stress in male C57BL/6 mice. 

Oncotarget, 2015. 6(27): p. 23213. 

44. Zhou, Y., L. Jiang, and L. Rui, Identification of MUP1 as a regulator for 

glucose and lipid metabolism in mice. Journal of Biological Chemistry, 2009. 

284(17): p. 11152-11159. 

45. Greve, S., et al., The major urinary protein gene cluster knockout mouse as a 

novel model for translational metabolism research. Sci Rep, 2022. 12(1): p. 

13161. 

46. Kreznar, J.H., et al., Host genotype and gut microbiome modulate insulin 

secretion and diet-induced metabolic phenotypes. Cell reports, 2017. 18(7): p. 

1739-1750. 

47. Holmes, E., et al., Understanding the role of gut microbiome–host metabolic 

signal disruption in health and disease. Trends in microbiology, 2011. 19(7): 

p. 349-359. 

48. Kawaji, H., et al., The FANTOM5 collection, a data series underpinning 

mammalian transcriptome atlases in diverse cell types. Scientific data, 2017. 

4(1): p. 1-3. 

49. Geiger, T., et al., Initial quantitative proteomic map of 28 mouse tissues using 

the SILAC mouse. Molecular & Cellular Proteomics, 2013. 12(6): p. 1709-

1722. 

50. Reizer, J., A. Reizer, and M. Saier Jr, The Na+/pantothenate symporter (PanF) 

of Escherichia coli is homologous to the Na+/proline symporter (PutP) of E. 

coli and the Na+/glucose symporters of mammals. Research in microbiology, 

1990. 141(9): p. 1069-1072. 



 

94 

51. Hueck, C.J., Type III protein secretion systems in bacterial pathogens of 

animals and plants. Microbiology and molecular biology reviews, 1998. 62(2): 

p. 379-433. 

52. Zarivach, R., et al., Structural analysis of the essential self-cleaving type III 

secretion proteins EscU and SpaS. Nature, 2008. 453(7191): p. 124-127. 

53. Srikhanta, Y.N., et al., Distinct physiological roles for the two L-asparaginase 

isozymes of Escherichia coli. Biochemical and biophysical research 

communications, 2013. 436(3): p. 362-365. 

54. Yang, K., et al., Structural basis for substrate binding and the catalytic 

mechanism of type III pantothenate kinase. Biochemistry, 2008. 47(5): p. 

1369-1380. 

55. Campbell, J.H., et al., Host genetic and environmental effects on mouse 

intestinal microbiota. The ISME journal, 2012. 6(11): p. 2033-2044. 

56. Hildebrand, F., et al., Inflammation-associated enterotypes, host genotype, 

cage and inter-individual effects drive gut microbiota variation in common 

laboratory mice. Genome biology, 2013. 14(1): p. 1-15. 

57. Le Chatelier, E., et al., Richness of human gut microbiome correlates with 

metabolic markers. Nature, 2013. 500(7464): p. 541-546. 

58. Turnbaugh, P.J., et al., A core gut microbiome in obese and lean twins. nature, 

2009. 457(7228): p. 480-484. 

59. Menni, C., et al., Gut microbiome diversity and high-fibre intake are related to 

lower long-term weight gain. International journal of obesity, 2017. 41(7): p. 

1099-1105. 

60. Ziętak, M., et al., Altered microbiota contributes to reduced diet-induced 

obesity upon cold exposure. Cell metabolism, 2016. 23(6): p. 1216-1223. 

61. Feng, L., et al., Gut microbiota-mediated improvement of metabolic disorders 

by Qingzhuan tea in high fat diet-fed mice. Journal of Functional Foods, 2021. 

78: p. 104366. 

62. Guo, J., et al., Gut microbiota and short chain fatty acids partially mediate the 

beneficial effects of inulin on metabolic disorders in obese ob/ob mice. Journal 

of Food Biochemistry, 2022. 46(5): p. e14063. 

63. Darnaud, M., et al., Enteric delivery of regenerating family member 3 alpha 

alters the intestinal microbiota and controls inflammation in mice with colitis. 

Gastroenterology, 2018. 154(4): p. 1009-1023. e14. 



 

95 

64. Louis, P., G.L. Hold, and H.J. Flint, The gut microbiota, bacterial metabolites 

and colorectal cancer. Nature reviews microbiology, 2014. 12(10): p. 661-672. 

65. Hamer, H.M., et al., The role of butyrate on colonic function. Alimentary 

pharmacology & therapeutics, 2008. 27(2): p. 104-119. 

66. Webb, M.E., A.G. Smith, and C. Abell, Biosynthesis of pantothenate. Natural 

product reports, 2004. 21(6): p. 695-721. 

67. Magnúsdóttir, S., et al., Systematic genome assessment of B-vitamin 

biosynthesis suggests co-operation among gut microbes. Frontiers in genetics, 

2015. 6: p. 148. 

68. Soto-Martin, E.C., et al., Vitamin biosynthesis by human gut butyrate-

producing bacteria and cross-feeding in synthetic microbial communities. 

MBio, 2020. 11(4): p. 10.1128/mbio. 00886-20. 

69. Heinken, A., et al., Functional metabolic map of Faecalibacterium prausnitzii, 

a beneficial human gut microbe. Journal of bacteriology, 2014. 196(18): p. 

3289-3302. 

70. D'Souza, G.G., et al., Ecology and evolution of metabolic cross-feeding 

interactions in bacteria. Natural Product Reports, 2018. 35(5): p. 455-488. 

71. Weger, B.D., et al., The Mouse Microbiome Is Required for Sex-Specific 

Diurnal Rhythms of Gene Expression and Metabolism. Cell Metab, 2019. 

29(2): p. 362-382 e8. 

72. Taketo, M., et al., FVB/N: an inbred mouse strain preferable for transgenic 

analyses. Proceedings of the National Academy of Sciences, 1991. 88(6): p. 

2065-2069. 

73. Townsend, D., C.J. Witkop Jr, and J. Mattson, Tyrosinase subcellular 

distribution and kinetic parameters in wild type and c‐locus mutant C57BL/6J 

mice. Journal of Experimental Zoology, 1981. 216(1): p. 113-119. 

74. Gaio, D., et al., Hackflex: low cost Illumina Nextera Flex sequencing library 

construction. 2019. 

75. Andrews, S., FastQC: a quality control tool for high throughput sequence 

data. 2010, Babraham Bioinformatics, Babraham Institute, Cambridge, United 

Kingdom. 

76. Martin, M., Cutadapt removes adapter sequences from high-throughput 

sequencing reads. EMBnet. journal, 2011. 17(1): p. 10-12. 



 

96 

77. Langmead, B. and S.L. Salzberg, Fast gapped-read alignment with Bowtie 2. 

Nature methods, 2012. 9(4): p. 357-359. 

78. Blanco-Míguez, A., et al., Extending and improving metagenomic taxonomic 

profiling with uncharacterized species using MetaPhlAn 4. Nature 

Biotechnology, 2023: p. 1-12. 

79. Keegan, K.P., E.M. Glass, and F. Meyer, MG-RAST, a metagenomics service 

for analysis of microbial community structure and function. Microbial 

environmental genomics (MEG), 2016: p. 207-233. 

80. Schwengers, O., et al., Bakta: rapid and standardized annotation of bacterial 

genomes via alignment-free sequence identification. Microbial genomics, 

2021. 7(11): p. 000685. 

81. Lahti, L. and S. Shetty, Introduction to the microbiome R package. 2018. 

82. McMurdie, P.J. and S. Holmes, phyloseq: an R package for reproducible 

interactive analysis and graphics of microbiome census data. PloS one, 2013. 

8(4): p. e61217. 

83. Wickham, H., W. Chang, and M.H. Wickham, Package ‘ggplot2’. Create 

elegant data visualisations using the grammar of graphics. Version, 2016. 2(1): 

p. 1-189. 

84. Kuznetsova, A., P. Brockhoff, and R. Christensen, Package ‘lmertest’. Version 

3.1-0. Computer software]. Retrieved from https://cran. r-project. org, 2019. 

85. Anderson, M.J., Permutational multivariate analysis of variance. Department 

of Statistics, University of Auckland, Auckland, 2005. 26: p. 32-46. 

86. Anderson, M.J., K.E. Ellingsen, and B.H. McArdle, Multivariate dispersion as 

a measure of beta diversity. Ecology letters, 2006. 9(6): p. 683-693. 

87. Oksanen, J., et al., Package ‘vegan’. Community ecology package, version, 

2013. 2(9): p. 1-295. 

88. Lin, H. and S.D. Peddada, Analysis of compositions of microbiomes with bias 

correction. Nature communications, 2020. 11(1): p. 1-11. 

89. Romero, L.M., M.J. Dickens, and N.E. Cyr, The Reactive Scope Model - a new 

model integrating homeostasis, allostasis, and stress. Horm Behav, 2009. 

55(3): p. 375-89. 

90. Levine, S. and H. Ursin, What is stress?, in Neurobiology and 

Neuroendocrinology. 1991, New York: Marcel Dekker. 

https://cran/


 

97 

91. Del Giudice, M., et al., What Is Stress? A Systems Perspective. Integr Comp 

Biol, 2018. 58(6): p. 1019-1032. 

92. Du Preez, A., et al., The type of stress matters: repeated injection and 

permanent social isolation stress in male mice have a differential effect on 

anxiety- and depressive-like behaviours, and associated biological alterations. 

Transl Psychiatry, 2020. 10(1): p. 325. 

93. Du Preez, A., et al., Do different types of stress differentially alter behavioural 

and neurobiological outcomes associated with depression in rodent models? A 

systematic review. Front Neuroendocrinol, 2021. 61: p. 100896. 

94. Barnum, C.J., et al., Psychological stress in adolescent and adult mice 

increases neuroinflammation and attenuates the response to LPS challenge. 

Journal of Neuroinflammation, 2012. 9: p. 1-15. 

95. Liu, X., et al., Effects of group housing on stress induced emotional and 

neuroendocrine alterations. Brain Res, 2013. 1502: p. 71-80. 

96. Pearson-Leary, J., et al., The gut microbiome regulates the increases in 

depressive-type behaviors and in inflammatory processes in the ventral 

hippocampus of stress vulnerable rats. Mol Psychiatry, 2020. 25(5): p. 1068-

1079. 

97. Davis, D.J., et al., Lactobacillus plantarum attenuates anxiety-related behavior 

and protects against stress-induced dysbiosis in adult zebrafish. Sci Rep, 2016. 

6: p. 33726. 

98. Chevalier, G., et al., Effect of gut microbiota on depressive-like behaviors in 

mice is mediated by the endocannabinoid system. Nat Commun, 2020. 11(1): 

p. 6363. 

99. de Weerth, C., Do bacteria shape our development? Crosstalk between 

intestinal microbiota and HPA axis. Neurosci Biobehav Rev, 2017. 83: p. 458-

471. 

100. Mir, H.D., et al., The gut microbiota metabolite indole increases emotional 

responses and adrenal medulla activity in chronically stressed male mice. 

Psychoneuroendocrinology, 2020. 119: p. 104750. 

101. Agranyoni, O., et al., Gut microbiota determines the social behavior of mice 

and induces metabolic and inflammatory changes in their adipose tissue. NPJ 

Biofilms Microbiomes, 2021. 7(1): p. 28. 

102. Werbner, M., et al., Social-Stress-Responsive Microbiota Induces Stimulation 

of Self-Reactive Effector T Helper Cells. mSystems, 2019. 4(4). 



 

98 

103. Weiss, S., Microbiota as mediator and moderator of stress. Bioessays, 2021. 

43(1): p. e2000277. 

104. Moeller, A.H. and J.G. Sanders, Roles of the gut microbiota in the adaptive 

evolution of mammalian species. Philos Trans R Soc Lond B Biol Sci, 2020. 

375(1808): p. 20190597. 

105. Candela, M., et al., Intestinal microbiota is a plastic factor responding to 

environmental changes. Trends in microbiology, 2012. 20(8): p. 385-391. 

106. David, L.A., et al., Host lifestyle affects human microbiota on daily timescales. 

Genome biology, 2014. 15(7): p. 1-15. 

107. David, L.A., et al., Diet rapidly and reproducibly alters the human gut 

microbiome. Nature, 2014. 505(7484): p. 559-563. 

108. Turnbaugh, P.J., et al., The effect of diet on the human gut microbiome: a 

metagenomic analysis in humanized gnotobiotic mice. Science translational 

medicine, 2009. 1(6): p. 6ra14-6ra14. 

109. Huang, L., et al., Commensal microbe-derived propionic acid mediates 

juvenile social isolation-induced social deficits and anxiety-like behaviors. 

Brain Research Bulletin, 2021. 166: p. 161-171. 

110. Usui, N., H. Matsuzaki, and S. Shimada, Characterization of Early Life Stress-

Affected Gut Microbiota. Brain sciences, 2021. 11(7): p. 913. 

111. Dunphy-Doherty, F., et al., Post-weaning social isolation of rats leads to long-

term disruption of the gut microbiota-immune-brain axis. Brain, behavior, and 

immunity, 2018. 68: p. 261-273. 

112. Apfelbach, R., et al., The effects of predator odors in mammalian prey species: 

a review of field and laboratory studies. Neuroscience & Biobehavioral 

Reviews, 2005. 29(8): p. 1123-1144. 

113. Baumbach, J.L., et al., A single exposure to the predator odor 2, 4, 5-

trimethylthiazoline causes long-lasting affective behavioral changes in female 

mice: Modulation by kappa opioid receptor signaling. Pharmacology 

Biochemistry and Behavior, 2024. 242: p. 173822. 

114. Post, R.J., et al., Effects of juvenile exposure to predator odor on adolescent 

and adult anxiety and pain nociception. Physiology & behavior, 2014. 131: p. 

57-61. 

115. Hebb, A.L., et al., Brief exposure to predator odor and resultant anxiety 

enhances mesocorticolimbic activity and enkephalin expression in CD‐1 mice. 

European Journal of Neuroscience, 2004. 20(9): p. 2415-2429. 



 

99 

116. Janitzky, K., et al., Behavioral effects and pattern of brain c-fos mRNA induced 

by 2, 5-dihydro-2, 4, 5-trimethylthiazoline, a component of fox feces odor in 

GAD67-GFP knock-in C57BL/6 mice. Behavioural brain research, 2009. 

202(2): p. 218-224. 

117. Hebb, A.L., et al., Odor-induced variation in anxiety-like behavior in mice is 

associated with discrete and differential effects on mesocorticolimbic 

cholecystokinin mRNA expression. Neuropsychopharmacology, 2002. 27(5): p. 

744-755. 

118. Oh, J.H. and B. Rehermann, Natural versus laboratory world: incorporating 

wild-derived microbiota into preclinical rodent models. The Journal of 

Immunology, 2021. 207(7): p. 1703-1709. 

119. Bruno, P., T. Schüler, and S.P. Rosshart, Born to be wild: utilizing natural 

microbiota for reliable biomedical research. Trends in Immunology, 2024. 

120. Sheehan, M.J., P. Campbell, and C.H. Miller, Evolutionary patterns of major 

urinary protein scent signals in house mice and relatives. Molecular ecology, 

2019. 28(15): p. 3587-3601. 

121. Miller, C.H., et al., Dynamic changes to signal allocation rules in response to 

variable social environments in house mice. Communications Biology, 2023. 

6(1): p. 297. 

122. Miller, C.H., et al., Scent mark signal investment predicts fight dynamics in 

house mice. Proceedings of the Royal Society B, 2023. 290(1993): p. 

20222489. 

123. Gould, T.D., D.T. Dao, and C.E. Kovacsics, The open field test. Mood and 

anxiety related phenotypes in mice, 2009: p. 1-20. 

124. Hacquemand, R., et al., Comparison between low doses of TMT and cat odor 

exposure in anxiety-and fear-related behaviors in mice. Behavioural brain 

research, 2013. 238: p. 227-231. 

125. Blanchard, R.J., et al., The characterization and modelling of antipredator 

defensive behavior. Neuroscience & Biobehavioral Reviews, 1990. 14(4): p. 

463-472. 

126. Nollet, M., A.M.L. Guisquet, and C. Belzung, Models of depression: 

unpredictable chronic mild stress in mice. Current protocols in pharmacology, 

2013. 61(1): p. 5.65. 1-5.65. 17. 

127. Zangrossi Jr, H. and S.E. File, Behavioral consequences in animal tests of 

anxiety and exploration of exposure to cat odor. Brain research bulletin, 1992. 

29(3-4): p. 381-388. 



 

100 

128. Rein, B., K. Ma, and Z. Yan, A standardized social preference protocol for 

measuring social deficits in mouse models of autism. Nat Protoc, 2020. 15(10): 

p. 3464-3477. 

129. Liu, J., et al., Adiponectin is critical in determining susceptibility to depressive 

behaviors and has antidepressant-like activity. Proc Natl Acad Sci U S A, 

2012. 109(30): p. 12248-53. 

130. Grant, R.W. and V.D. Dixit, Adipose tissue as an immunological organ. 

Obesity, 2015. 23(3): p. 512-518. 

131. Wu, Y. and Y. Ma, CCL2‐CCR2 signaling axis in obesity and metabolic 

diseases. Journal of Cellular Physiology, 2024. 239(4): p. e31192. 

132. Kelly, J.R., et al., Breaking down the barriers: the gut microbiome, intestinal 

permeability and stress-related psychiatric disorders. Frontiers in cellular 

neuroscience, 2015. 9: p. 392. 

133. Bertin, B., P. Desreumaux, and L. Dubuquoy, Obesity, visceral fat and Crohn's 

disease. Current Opinion in Clinical Nutrition & Metabolic Care, 2010. 13(5): 

p. 574-580. 

134. Karrasch, T. and A. Schaeffler, Adipokines and the role of visceral adipose 

tissue in inflammatory bowel disease. Annals of Gastroenterology: Quarterly 

Publication of the Hellenic Society of Gastroenterology, 2016. 29(4): p. 424. 

135. Batra, A., et al., Mesenteric fat—control site for bacterial translocation in 

colitis? Mucosal Immunology, 2012. 5(5): p. 580-591. 

136. Zhang, L.-j., et al., Dermal adipocytes protect against invasive Staphylococcus 

aureus skin infection. Science, 2015. 347(6217): p. 67-71. 

137. Alcorn, J.F. and J.K. Kolls, Killer fat. Science, 2015. 347(6217): p. 26-27. 

138. Guo, D.-H., et al., Visceral adipose NLRP3 impairs cognition in obesity via IL-

1R1 on CX3CR1+ cells. The Journal of clinical investigation, 2020. 130(4): p. 

1961-1976. 

139. Sanders, J.G., et al., Widespread extinctions of co-diversified primate gut 

bacterial symbionts from humans. Nat Microbiol, 2023. 8(6): p. 1039-1050. 

140. Parks, D.H., et al., GTDB: an ongoing census of bacterial and archaeal 

diversity through a phylogenetically consistent, rank normalized and complete 

genome-based taxonomy. Nucleic acids research, 2022. 50(D1): p. D785-

D794. 



 

101 

141. Chaumeil, P.-A., et al., GTDB-Tk: a toolkit to classify genomes with the 

Genome Taxonomy Database. 2020, Oxford University Press. 

142. Olm, M.R., et al., inStrain profiles population microdiversity from 

metagenomic data and sensitively detects shared microbial strains. Nat 

Biotechnol, 2021. 

143. Patel, K.V., et al., Impact of acute stress on the canine gut microbiota. 

Scientific Reports, 2024. 14(1): p. 18897. 

144. Rodiño-Janeiro, B.K., et al., Acute stress triggers sex-dependent rapid 

alterations in the human small intestine microbiota composition. Frontiers in 

Microbiology, 2025. 15: p. 1441126. 

145. Bowen, M.T., et al., Aggregation in quads but not pairs of rats exposed to cat 

odor or bright light. Behavioural processes, 2012. 90(3): p. 331-336. 

146. Sandi, C. and J. Haller, Stress and the social brain: behavioural effects and 

neurobiological mechanisms. Nature reviews neuroscience, 2015. 16(5): p. 

290-304. 

147. Li, A.-l., et al., Effect of 2′-fucosyllactose supplementation on intestinal flora in 

mice with intestinal inflammatory diseases. International Dairy Journal, 2020. 

110: p. 104797. 

148. Zhao, R. and G.X. Shen, Impact of anthocyanin component and metabolite of 

Saskatoon berry on gut microbiome and relationship with fecal short chain 

fatty acids in diet-induced insulin resistant mice. The Journal of Nutritional 

Biochemistry, 2023. 111: p. 109201. 

149. Zhan, F., et al., Cucurbitacin E Alleviates Colonic Barrier Function 

Impairment and Inflammation Response and Improves Microbial Composition 

on Experimental Colitis Models. Journal of Inflammation Research, 2024: p. 

2745-2756. 

150. Lee, K.S., et al., An automated Raman-based platform for the sorting of live 

cells by functional properties. Nature microbiology, 2019. 4(6): p. 1035-1048. 

151. Volk, J.K., et al., The Nlrp6 inflammasome is not required for baseline colonic 

inner mucus layer formation or function. Journal of Experimental Medicine, 

2019. 216(11): p. 2602-2618. 

152. van der Hee, B. and J.M. Wells, Microbial regulation of host physiology by 

short-chain fatty acids. Trends in microbiology, 2021. 29(8): p. 700-712. 

153. Zhu, Y., et al., Exploration of the muribaculaceae family in the gut microbiota: 

diversity, metabolism, and function. Nutrients, 2024. 16(16): p. 2660. 



 

102 

154. Agirman, G., K.B. Yu, and E.Y. Hsiao, Signaling inflammation across the gut-

brain axis. Science, 2021. 374(6571): p. 1087-1092. 

155. Fung, T.C., The microbiota-immune axis as a central mediator of gut-brain 

communication. Neurobiology of disease, 2020. 136: p. 104714. 

156. Puricelli, C., et al., The gut-brain-immune axis in autism spectrum disorders: a 

state-of-art report. Frontiers in Psychiatry, 2022. 12: p. 755171. 

157. Crumeyrolle-Arias, M., et al., Absence of the gut microbiota enhances anxiety-

like behavior and neuroendocrine response to acute stress in rats. 

Psychoneuroendocrinology, 2014. 42: p. 207-17. 

158. Lyte, J.M., et al., Gut-brain axis serotonergic responses to acute stress 

exposure are microbiome-dependent. Neurogastroenterol Motil, 2020. 32(11): 

p. e13881. 

159. Rosell-Cardona, C., et al., Acute stress enhances synaptic plasticity in male 

mice via a microbiota-dependent mechanism. Neuropharmacology, 2025. 273: 

p. 110434. 

160. Vagnerová, K., et al., Interactions between gut microbiota and acute restraint 

stress in peripheral structures of the hypothalamic–pituitary–adrenal axis and 

the intestine of male mice. Frontiers in immunology, 2019. 10: p. 2655. 

161. Moeller, A.H., et al., Transmission modes of the mammalian gut microbiota. 

Science, 2018. 362(6413): p. 453-457. 

162. Vogt, C.C., et al., Female behavior drives the formation of distinct social 

structures in C57BL/6J versus wild-derived outbred mice in field enclosures. 

BMC biology, 2024. 22(1): p. 35. 

163. Tuttle, A.H., et al., Social propinquity in rodents as measured by tube 

cooccupancy differs between inbred and outbred genotypes. Proceedings of the 

National Academy of Sciences, 2017. 114(21): p. 5515-5520. 

164. Smith, J., J.L. Hurst, and C. Barnard, Comparing behaviour in wild and 

laboratory strains of the house mouse: levels of comparison and functional 

inference. Behavioural processes, 1994. 32(1): p. 79-86. 

165. Moy, S., et al., Sociability and preference for social novelty in five inbred 

strains: an approach to assess autistic‐like behavior in mice. Genes, Brain and 

Behavior, 2004. 3(5): p. 287-302. 

166. Blanchard, R.J., et al., Defensive behaviors in wild and laboratory (Swiss) 

mice: the mouse defense test battery. Physiology & behavior, 1998. 65(2): p. 

201-209. 



 

103 

167. Dumont, B.L., et al., Into the Wild: A novel wild-derived inbred strain 

resource expands the genomic and phenotypic diversity of laboratory mouse 

models. PLoS genetics, 2024. 20(4): p. e1011228. 

168. Rodriguez, A., et al., ToxTrac: a fast and robust software for tracking 

organisms. Methods in Ecology and Evolution, 2018. 9(3): p. 460-464. 

169. Nollet, M., Models of Depression: Unpredictable Chronic Mild Stress in Mice. 

Curr Protoc, 2021. 1(8): p. e208. 

170. Kalueff, A.V. and P. Tuohimaa, Grooming analysis algorithm for 

neurobehavioural stress research. Brain Research Protocols, 2004. 13(3): p. 

151-158. 

171. Dobin, A., et al., STAR: ultrafast universal RNA-seq aligner. Bioinformatics, 

2013. 29(1): p. 15-21. 

172. Mölder, F., et al., Sustainable data analysis with Snakemake. F1000Research, 

2021. 10. 

173. Li, D., et al., MEGAHIT: an ultra-fast single-node solution for large and 

complex metagenomics assembly via succinct de Bruijn graph. Bioinformatics, 

2015. 31(10): p. 1674-1676. 

174. Gurevich, A., et al., QUAST: quality assessment tool for genome assemblies. 

Bioinformatics, 2013. 29(8): p. 1072-1075. 

175. Brown, C.T. and L. Irber, sourmash: a library for MinHash sketching of DNA. 

Journal of open source software, 2016. 1(5): p. 27. 

176. Li, H., Minimap2: pairwise alignment for nucleotide sequences. 

Bioinformatics, 2018. 34(18): p. 3094-3100. 

177. Alneberg, J., et al., Binning metagenomic contigs by coverage and 

composition. Nat Methods, 2014. 11(11): p. 1144-6. 

178. Kang, D.D., et al., MetaBAT 2: an adaptive binning algorithm for robust and 

efficient genome reconstruction from metagenome assemblies. PeerJ, 2019. 7: 

p. e7359. 

179. Wu, Y.W., B.A. Simmons, and S.W. Singer, MaxBin 2.0: an automated 

binning algorithm to recover genomes from multiple metagenomic datasets. 

Bioinformatics, 2016. 32(4): p. 605-7. 

180. Sieber, C.M., et al., Recovery of genomes from metagenomes via a 

dereplication, aggregation and scoring strategy. Nature microbiology, 2018. 

3(7): p. 836-843. 



 

104 

181. Parks, D.H., et al., CheckM: assessing the quality of microbial genomes 

recovered from isolates, single cells, and metagenomes. Genome research, 

2015. 25(7): p. 1043-1055. 

182. Orakov, A., et al., GUNC: detection of chimerism and contamination in 

prokaryotic genomes. Genome Biol, 2021. 22(1): p. 178. 

183. Olm, M.R., et al., dRep: a tool for fast and accurate genomic comparisons that 

enables improved genome recovery from metagenomes through de-replication. 

The ISME journal, 2017. 11(12): p. 2864-2868. 

184. Love, M.I., W. Huber, and S. Anders, Moderated estimation of fold change 

and dispersion for RNA-seq data with DESeq2. Genome biology, 2014. 

15(12): p. 1-21. 

185. Stephens, M., False discovery rates: a new deal. Biostatistics, 2017. 18(2): p. 

275-294. 

186. Yu, G., et al., clusterProfiler: an R package for comparing biological themes 

among gene clusters. Omics: a journal of integrative biology, 2012. 16(5): p. 

284-287. 

187. Wu, T., et al., clusterProfiler 4.0: A universal enrichment tool for interpreting 

omics data. The Innovation, 2021. 2(3): p. 100141. 

188. Pedregosa, F., et al., Scikit-learn: Machine learning in Python. the Journal of 

machine Learning research, 2011. 12: p. 2825-2830. 

189. McFall-Ngai, M., et al., Animals in a bacterial world, a new imperative for the 

life sciences. Proc Natl Acad Sci U S A, 2013. 110(9): p. 3229-36. 

190. Behera, J., et al., The role of gut microbiota in bone homeostasis. Bone, 2020. 

135: p. 115317. 

191. Xue, J., K.M. Ajuwon, and R. Fang, Mechanistic insight into the gut 

microbiome and its interaction with host immunity and inflammation. Animal 

nutrition, 2020. 6(4): p. 421-428. 

192. Zheng, D., T. Liwinski, and E. Elinav, Interaction between microbiota and 

immunity in health and disease. Cell Res, 2020. 30(6): p. 492-506. 

193. Jiao, Y., et al., Crosstalk Between Gut Microbiota and Innate Immunity and Its 

Implication in Autoimmune Diseases. Front Immunol, 2020. 11: p. 282. 

194. Sommer, F., et al., The gut microbiota modulates energy metabolism in the 

hibernating brown bear Ursus arctos. Cell reports, 2016. 14(7): p. 1655-1661. 



 

105 

195. Flint, H.J., et al., The role of the gut microbiota in nutrition and health. Nature 

reviews Gastroenterology & hepatology, 2012. 9(10): p. 577-589. 

196. Alberdi, A., et al., Do Vertebrate Gut Metagenomes Confer Rapid Ecological 

Adaptation? Trends Ecol Evol, 2016. 31(9): p. 689-699. 

197. Kohl, K.D., A. Stengel, and M.D. Dearing, Inoculation of tannin‐degrading 

bacteria into novel hosts increases performance on tannin‐rich diets. 

Environmental Microbiology, 2016. 18(6): p. 1720-1729. 

198. Sarkar, A., et al., Microbial transmission in animal social networks and the 

social microbiome. Nat Ecol Evol, 2020. 4(8): p. 1020-1035. 

199. Raulo, A., et al., Social and environmental transmission spread different sets 

of gut microbes in wild mice. Nature Ecology & Evolution, 2024. 8(5): p. 972-

985. 

200. Moeller, A.H., et al., Social behavior shapes the chimpanzee pan-microbiome. 

Sci Adv, 2016. 2(1): p. e1500997. 

201. Raulo, A., et al., Social networks strongly predict the gut microbiota of wild 

mice. The ISME Journal, 2021. 

202. Moeller, A.H., et al., Dispersal limitation promotes the diversification of the 

mammalian gut microbiota. Proc Natl Acad Sci U S A, 2017. 114(52): p. 

13768-13773. 

203. Costello, E.K., et al., The application of ecological theory toward an 

understanding of the human microbiome. Science, 2012. 336(6086): p. 1255-

62. 

204. Krimberg, J.S., et al., Impact of social isolation on the oxytocinergic system: A 

systematic review and meta-analysis of rodent data. Neurosci Biobehav Rev, 

2022. 134: p. 104549. 

205. Mudra Rakshasa, A. and M.T. Tong, Making "Good" Choices: Social Isolation 

in Mice Exacerbates the Effects of Chronic Stress on Decision Making. Front 

Behav Neurosci, 2020. 14: p. 81. 

206. Korem, T., Growth dynamics of gut microbiota in health and disease inferred 

from 

single metagenomic samples. 2015. 

207. Sender, R., S. Fuchs, and R. Milo, Revised estimates for the number of human 

and bacteria cells in the body. PLoS biology, 2016. 14(8): p. e1002533. 



 

106 

208. Barrick, J.E. and R.E. Lenski, Genome dynamics during experimental 

evolution. Nature Reviews Genetics, 2013. 14(12): p. 827-839. 

209. Zhao, S., et al., Adaptive Evolution within Gut Microbiomes of Healthy People. 

Cell Host Microbe, 2019. 25(5): p. 656-667 e8. 

210. Garud, N.R., et al., Evolutionary dynamics of bacteria in the gut microbiome 

within and across hosts. PLoS Biol, 2019. 17(1): p. e3000102. 

211. Sakoparnig, T., C. Field, and E. van Nimwegen, Whole genome phylogenies 

reflect the distributions of recombination rates for many bacterial species. 

Elife, 2021. 10. 

212. Schloissnig, S., et al., Genomic variation landscape of the human gut 

microbiome. Nature, 2013. 493(7430): p. 45-50. 

213. Greenblum, S., R. Carr, and E. Borenstein, Extensive strain-level copy-number 

variation across human gut microbiome species. Cell, 2015. 160(4): p. 583-

594. 

214. Nayfach, S., et al., An integrated metagenomics pipeline for strain profiling 

reveals novel patterns of bacterial transmission and biogeography. Genome 

research, 2016. 26(11): p. 1612-1625. 

215. Scholz, M., et al., Strain-level microbial epidemiology and population 

genomics from shotgun metagenomics. Nat Methods, 2016. 13(5): p. 435-8. 

216. Rakoff-Nahoum, S., K.R. Foster, and L.E. Comstock, The evolution of 

cooperation within the gut microbiota. Nature, 2016. 533(7602): p. 255-9. 

217. Hsu, R.H., et al., Microbial interaction network inference in microfluidic 

droplets. Cell systems, 2019. 9(3): p. 229-242. e4. 

218. Frazão, N., et al., Two modes of evolution shape bacterial strain diversity in 

the mammalian gut for thousands of generations. Nature Communications, 

2022. 13(1). 

219. Frazao, N., et al., Horizontal gene transfer overrides mutation in Escherichia 

coli colonizing the mammalian gut. Proc Natl Acad Sci U S A, 2019. 116(36): 

p. 17906-17915. 

220. Barroso-Batista, J., et al., The first steps of adaptation of Escherichia coli to 

the gut are dominated by soft sweeps. PLoS Genet, 2014. 10(3): p. e1004182. 

221. Lescat, M., et al., Using long-term experimental evolution to uncover the 

patterns and determinants of molecular evolution of an Escherichia coli 



 

107 

natural isolate in the streptomycin-treated mouse gut. Mol Ecol, 2017. 26(7): 

p. 1802-1817. 

222. Zhang, P., et al., Colonization of gut microbiota by plasmid-carrying bacteria 

is facilitated by evolutionary adaptation to antibiotic treatment. The ISME 

journal, 2022. 16(5): p. 1284-1293. 

223. Groussin, M., et al., Elevated rates of horizontal gene transfer in the 

industrialized human microbiome. Cell, 2021. 

224. Yilmaz, B., et al., Long-term evolution and short-term adaptation of 

microbiota strains and sub-strains in mice. Cell Host Microbe, 2021. 

225. Zitomersky, N.L., M.J. Coyne, and L.E. Comstock, Longitudinal analysis of 

the prevalence, maintenance, and IgA response to species of the order 

Bacteroidales in the human gut. Infection and immunity, 2011. 79(5): p. 2012-

2020. 

226. Cruz-Morales, P., et al., Revisiting the evolution and taxonomy of Clostridia, a 

phylogenomic update. Genome Biology and Evolution, 2019. 11(7): p. 2035-

2044. 

227. Yutin, N. and M.Y. Galperin, A genomic update on clostridial phylogeny: G 

ram‐negative spore formers and other misplaced clostridia. Environmental 

microbiology, 2013. 15(10): p. 2631-2641. 

 


	BIOGRAPHICAL SKETCH
	ACKNOWLEDGEMENTS
	TABLE OF CONTENTS
	LIST OF FIGURES
	LIST OF TABLES
	1. MAJOR URINARY PROTEIN (MUP) GENE FAMILY DELETION DRIVES SEX-SPECIFIC ALTERATIONS ON THE HOUSE MOUSE GUT MICROBIOTA
	1.1. Abstract
	1.2. Importance
	1.3. Introduction
	1.4. Results
	1.4.1. Deletion of the Mup gene cluster
	1.4.2. Metagenomic sequencing of Mup WT and KO mice
	1.4.3. Mup deletion affected the taxonomic composition of the gut microbiota in males
	1.4.4. Mup deletion affected the functional composition of the gut microbiota in males
	1.4.5. Significant correspondence of taxonomic and functional profiles
	1.4.6. Mup deletion reduced the gut microbial diversity in males
	1.4.7. Specific microbial taxa and functions were depleted in Mup-knockout males

	1.5. Discussion
	1.6. Methods
	1.6.1. Genome Editing
	1.6.2. Animals
	1.6.3. Microbiota Analysis
	1.6.4. Statistical Analysis


	2. THE MOUSE GUT MICROBIOTA RESPONDS TO PREDATOR ODOR AND PREDICTS HOST BEHAVIOR
	2.1. Abstract
	2.2. Importance
	2.3. Introduction
	2.4. Results
	2.4.1. Predator-odor exposure and social isolation gave rise to stress-associated behaviors
	2.4.2. Predator-odor exposure altered the transcriptome of an endocrine and immunological tissue
	2.4.3. Predator-odor exposure had a greater impact on the gut microbiota than did social isolation
	2.4.4. Predator-odor exposure drove persistent alterations in the relative abundances of specific SGBs
	2.4.5. Predator odor–responsive SGBs co-varied with host anti-microbial and behavioral responses
	2.4.6. The gut microbiota was a better predictor of host behavior than was host VAT gene expression

	2.6. Methods
	2.6.1. Animals
	2.6.2. Social Isolation
	2.6.3. Predator Odor Exposure
	2.6.4. Behavioral Assays
	2.6.4.1. Open Field Test
	2.6.4.2. Splash Test
	2.6.4.3. Three-Chamber Test

	2.6.5. Host Gene Expression
	2.6.5.1. Tissue Collection and Gross Morphometry
	2.6.5.2. Transcriptomic Sequencing
	2.6.5.3. Transcriptomic Data Processing

	2.6.6. Gut Microbiota
	2.6.6.1. Fecal Collections
	2.6.6.2. DNA Sequencing
	2.6.6.3. Metagenomic Data Processing and Analysis

	2.6.7. Statistical Analyses


	3. SOCIAL ISOLATION ACCELERATES EVOLUTION IN THE MURINE GUT MICROBIOTA
	3.1. Abstract
	3.2. Importance
	3.3. Introduction
	3.4. Results & Discussion
	3.4.1. Measurable intra-specific diversity in the native mouse gut microbiota
	3.4.2. Social isolation accelerates divergent evolution in the gut microbiota

	3.5. Conclusion
	3.6. Methods
	3.6.1. Animals
	3.6.2. Social Isolation
	3.6.3. Metagenomic Data
	3.6.5. Statistical Analyses


	APPENDIX
	1.1. Supplemental Results
	1.2. Supplemental Figures
	1.3. Supplemental Tables
	2.1. Supplemental Figures
	2.2. Supplemental Tables

	REFERENCES