CELLULAR RESPONSE TO MINERALIZED COLLAGEN FIBRILS IN 3D CO-CULTURE SPRING GEL MODEL FOR CALCIFIC AORTIC VALVE DISEASE A Thesis Presented to the Faculty of the Graduate School of Cornell University In Partial Fulfillment of the Requirements for the Degree of Master of Science in Materials Science and Engineering by Chih-Yi Wang August 2024 © 2024 Chih-Yi Wang ABSTRACT Calcific aortic valve disease (CAVD) is an active progressive disease characterized by calcification of the aortic valve, causing tissue stiffening and eventually valve malfunction. As an age-related disease, CAVD has become a rising public concern due to improved healthcare and longevity. However, the lack of understanding of the disease's progression and mechanism remains unclear, limiting the development of pharmacologic treatments. Here, we aim to understand how the morphology of hydroxyapatite in a mineral-rich matrix can affect the behavior of valve cell populations in a model for later disease stages. By incorporating hydroxyapatite nanoparticles and mineralized collagen fibrils into the in vitro 3D hydrogel model that mimics the aortic valve microenvironment, we were able to observe the lesion formation, its chemical composition and spatial distribution with Raman mapping, and cell behavior through a histology approach. iii BIOGRAPHICAL SKETCH Chih-Yi Wang was born and raised in Hsinchu, Taiwan. Inspired by her father who came from a background of engineering, she developed strong interest in the field of science at a young age. Chih-Yi earned her Bachelor of Science in Materials Science and Engineering at National Tsing Hua University, where she did her undergraduate research in polymer-coated nanoparticle cancer drug delivery. During her undergraduate studies, she noticed her increasing appeal in biomaterials, especially applying materials in biomedical uses. At Cornell University, Chih-Yi joined the Estroff research group and started her thesis research in the field of biomineral, specifically the in vitro models for Calcific Aortic Valve Disease. Over the course of two years at Cornell, she has gained new set of technical skills, gets to collaborate with many wonderful researchers, and more importantly, her passion in biomaterial has grown stronger than ever. Chih-Yi will be continuing her academic journey at University of Florida for a PhD degree where she’s looking forward to applying everything that she has learned at Cornell. iv To my parents. v ACKNOWLEDGMENTS I have immense gratitude for everyone who offered guidance, encouragement, and feedback during my time in Cornell University. First, I want to thank my advisor Dr. Lara Estroff, without her guidance when the project got stuck, the project with mineralized collagen fibril will have been impossible. I would also like to thank Dr. Jonathan Butcher for the support on the CAVD project. Thanks to everyone in the CAVD project, who I enjoyed working with and provided me a lot of help, especially Stephan Sutter and Alex Cruz who are my mentors. Thanks to Cornell Center for Materials Research (CCMR) facility managers, Mark Pfeifer, Philip Carubia, and John Grazul for their knowledge in sample preparation and instrument techniques. Outside of research, I would like to thank the Estroff Group, the welcoming and lively atmosphere makes the office a pleasant place to work. I want to express my gratitude toward Lilly Tsaur, my roommate and friend, who I share lots of happy and stressful time with. I’m also grateful to my friends, Jeffrey Gao, Akshey Dhar, Ashutosh Garudapalli, Douglas Palumbo, Jin Hong Joo, Jonathan Palumbo, River Carson, Vicky Peng, and Yuhe Zhang, my time here has been so colorful because of you. Finally, to my parents, who has been supportive all this time, thank you. vi TABLE OF CONTENTS ABSTRACT ...................................................................................................................i BIOGRAPHICAL SKETCH......................................................................................iii ACKNOWLEDGEMENTS ........................................................................................ v TABLE OF CONTENTS .......................................................................................... vi LIST OF FIGURES...................................................................................................viii LIST OF TABLES....................................................................................................... x LIST OF ABREVIATIONS....................................................................................... xi CHAPTER 1. BACKGROUND AND INTRODUCTION....................................... 1 1.1 Introduction........................................................................................................... 1 1.2 Biomineralization in the Human Body…………….............................................. 3 1.3 Calcific Aortic Valve Disease............................................................................... 5 1.4 Minerals in the Valve............................................................................................ 8 1.5 Role of Valve Cell Populations in CAVD............................................................. 9 1.6 Three-Dimensional (3D) in vitro models of CAVD…………………................ 10 1.7 Thesis Overview ................................................................................................. 12 1.8 References........................................................................................................... 13 CHAPTER 2. CELLULAR RESPONSE TO MINERALIZED COLLAGEN FIBRILS IN 3D CO-CULTURE SPRING GEL MODEL FOR CALCIFIC AORTIC VALVE DISEASE………………............................................................. 17 2.1 Introduction......................................................................................................... 17 2.2 Experimental Design........................................................................................... 17 2.3 Materials and Methods........................................................................................ 20 2.4 Results………………......................................................................................... 30 2.4.1 Casting of MCF incorporated hydrogel……………..................................... 30 2.4.2 Hydrogel Compaction Analysis……………………………………………. 31 vii 2.4.3 Lesion Formation…………………….......................................................... 41 2.4.4 H&E Staining of Gel Cross Section….......................................................... 43 2.4.5 Raman Mapping of Lesions on Whole-mount………….............................. 45 2.4.6 Raman Mapping of Lesions on Cross Section……...................................... 50 2.5 References.......................................................................................................... 62 CHAPTER 3. CONCLUSION AND FUTURE DIRECTIONS ............................ 63 APPENDIX1: Raman mapping on HA_OGM cross section done dehydrated... 66 APPENDIX 2: Raman mapping on MCF_OGM cross section don dehydrated. 68 APPENDIX 3: Raman mapping on MCF_OGM cross section on-lesion. ………71 viii LIST OF FIGURES Figure 1.1 Examples of physiological and pathological mineralization in human body…………………………………………………………………………………… 4 Figure 1.2 Heart and aortic valve structure…………………………………….…. 6 Figure 1.3 Calcific Aortic Valve Disease………………………………………… 7 Figure 1.4. In vitro models used in CAVD relevant studies…………………..…. 12 Figure 2.1 Design of 3D in vitro spring gel model from Gee et al……………… 18 Figure 2.2 Experimental design…………………………………………………. 21 Figure 2.3 Characterization of HA nanoparticles and mineralized collagen fibrils………………………………………………………………………………… 23 Figure 2.4 Bright field image of MCF hydrogel under GM and OGM condition. 33 Figure 2.5. Hydrogel compaction of HA and MCF gels under both GM and OGM conditions……………………………………………………………………………. 34 Figure 2.6 T-test done on compaction of four condition, HA_GM, HA_OGM, MCF_GM, MCF_OGM. ……………………………………………………………. 35 Figure 2.7 Bright field image of MCF and HA nanoparticle…………………….. 38 Figure 2.8 Hydrogel compaction of HA_OGM and MCF_OGM………………... 39 Figure 2.9 Lesion formation. ……………………………………………………. 42 Figure 2.10 H&E cross section staining of HA_OGM and MCF_OGM gels…… 44 Figure 2.11 High resolution H&E staining of lesion area………………………… 45 Figure 2.12 Raman mapping on HA_OGM whole mount sample (perinodular)…. 46 Figure 2.13 Raman mapping on MCF_OGM whole mount sample (perinodular)... 48 ix Figure 2.14 Raman mapping on HA_OGM cross section on-lesion……………… 52 Figure 2.15 Raman mapping on HA_OGM cross section off-lesion………...…… 55 Figure 2.16 Raman mapping on MCF_OGM cross section on-lesion …………… 57 Figure 2.17 Raman mapping on MCF_OGM cross section off-lesion …………... 59 Figure 3.1 Immunofluorescence staining can help understand cell behavior…… 65 Figure A1.1 Raman mapping on HA_OGM cross section done dehydrated……... 66 Figure A2.1 Raman mapping on MCF_OGM cross section done dehydrated…… 68 Figure A3.1 Raman mapping on MCF_OGM cross section on-lesion…………… 71 x LIST OF TABLES Table 2.1 Mass ratio of MCF to collagen solution tested in spring gel model… 31 Table 2.2 Raman peak assignments……………………………………………. 61 xi LIST OF ABBREVIATIONS CAVD Calcific Aortic Valve Disease HA Hydroxyapatite MCF Mineralized collagen fibrils VIC Valve interstitial cell VEC Valve endothelial cell GM Basal-medium OGM Osteogenic media SEM Scanning electron microscopy TEM Transmission electron microscopy 1 CHAPTER 1 BACKGROUND AND INTRODUCTION 1.1 Introduction Calcific aortic valve disease (CAVD) is a progressive disease that happens when calcium deposits on the aortic valve, causing stiffening of the tissue and eventually leading to malfunction of the valve. Calcific aortic valve disease has been dramatically increasing in global burden: between 1990 and 2017, the incidence of CAVD increased by 124%[1]. In 2017, about 12.6 million cases of CAVD were reported, along with an estimated 102, 700 deaths [1] Currently, there are no approved pharmacologic treatments available to prevent or reduce the disease progression, leaving invasive transcatheter implantation (TAVR) or surgical valve replacement (SAVR) as the only practical options. Limited understanding of the complex mechanism of the disease initiation and progression has been restricting research in new treatments [2] As CAVD is a chronic disease that progresses through time, understanding the disease environment of different stages is critical. Previously thought to be a degenerative disease, recent work has shown that CAVD involves accerlerated matrix remodeling and that the formation of large calcific lesions is driven by cell-cell and cell-matrix interactions [3]. While characterization of ex vivo tissues has provided direct evidence of the structural disruption and mineral buildup associated with CAVD, it fails to capture early timepoints of the calcification process for two reasons. First, CAVD is most often diagnosed when symptoms arise, i.e., when the leaflets are already moderately calcified; second, valves are typically not 2 replaced until their removal is needed, making it difficult to access lightly calcified ex vivo valves. On top of this, the most common valve intervention is TAVR (which is indicated in cases of light to moderate CAVD), which just involves implanting a new valve directly into the malfunctioning valve, so ex vivo leaflets are only obtained through SAVR which is reserved for advanced cases that would be ineligible for the less invasive TAVR. Some solution to the limitation of ex vivo human tissues are mouse models, however, the anatomical structure of mouse aortic valve differs from human and pig valves, as their leaflets are usually only ~5–10 cells thick and do not exhibit segregated layers [4]. Moreover, lesions in mouse models can take several months to form, whereas lesions can form in in vitro model with porcine cells within 1 week, which is why in vitro CAVD models has been widely used in relevant studies [4]. Hydrogel-based in vitro models have commonly been used to capture many characteristics of the native cellular environment [5]. 3D collagen hydrogels in particular are capable of recapitulating the key elements and structure of diseased valves, as shown by Butcher and coworkers who demonstrated CAVD-like lesion formation driven by an osteogenic stress state and seeded mineral particles in a collagen-based co-culture model [6]. By incorporating different forms of minerals, from particle-like to intrafibrillar collagen fibrils, into this flexible model, I aim to understand how the morphology of pre-existing minerals, affects the lesion-forming behavior of aortic valve cells. 3 1.2 Biomineralization in the human body Biomineralization is the process of mineral deposition, often onto an organic matrix, by living organisms. The products consist of heterogeneous composites of organic and inorganic compounds [7]. Some common examples of biomineralization include vertebrate bones and teeth, as well as the exoskeletons of crustaceans and mollusks. Mineralization in the human body, under normal circumstances, is an important process responsible of developing hard tissues that give structure to the body [8]. Physiological mineralization is highly regulated. For example, bone formation is regulated by specific cell populations, and the mineral formed has uniform chemical composition and morphology. Bone formation is a continuous process of deposition and resorption that involves osteoblasts which are responsible for the growth of calcium phosphate crystals in collagen fibrils, osteoclasts which resorb and remodel mineral, and osteocytes which maintain bone structure and play an important signaling role [9] (Fig.1.1B). The resulting mineral is composed of carbonated hydroxyapatite that is poorly crystalline. The collagen provides a structural template in which mineralization proceeds, both intrafibrillar and interfibrillar, and the crystallographic c axes of the apatite nanocrystals aligned with the long axes of the collagen fibrils [10]. 4 However, unwanted mineralization may take place at ectopic sites (e.g., within soft tissues), through a variety of processes collectively known as pathological mineralization. This type of mineralization is usually detrimental to the function of the tissues and is often associated with disease conditions such as injury, inflammation, and aging. Examples of pathological mineralization include calcific valve disease, blood vessel calcification, kidney stones, and calcific tendinitis [9-12]. Unlike physiological mineralization, pathological minerals are usually a product of Figure 1.1: Examples of physiological and pathological mineralization in human body. (A) Physiological and pathological mineralization on a tissue level. (B) Bone structure and bone mineralization on a cellular level. Adapted with permission from Refs [10] and [37]. 5 dysregulation, are chemically and morphologically heterogeneous, and can arise through a variety of pathways according to the tissue it is forming in [10]. My work aims to explore the complex relationship between cells and their mineralized matrix in the context of calcific aortic valve disease. 1.3 Calcific Aortic Valve Disease The human heart is responsible for delivering blood containing oxygen, nutrients, and signaling molecules to organs around the body [2]. To sufficiently support the human body, the heart pumps 3-5 liters of blood every ~60 seconds, an average of more than 2.5 billion heartbeat in a 70-year lifetime[2, 14]. The flow of oxygenated blood out of the heart to the rest of the body is regulated by the aortic valve, located between the left ventricle and the aorta. The human heart aortic valve is composed of three leaflets, also known as cusps. To regulate the direction of blood flow, these leaflets open to allow blood to flow through and seal tightly together to prevent the blood from regurgitating into the heart. Mature aortic valve leaflets are made up of highly organized extracellular matrix (ECM) with resident valve interstitial cells (VIC), with a monolayer of valve endothelial cells (VEC) covering the surface [16]. The valve ECM has a trilaminar structure: the fibrosa facing the aorta, the spongiosa in the middle, and the ventricularis on the ventricular side. These three layers each play a role in providing the biomechanical properties the valve needs to sustain the long and repetitive cyclical strain [17]. The fibrosa is composed of collagen ensuring the tensile strength of the valve, whereas the ventricularis composed of elastic fibers providing elasticity during diastole. The spongiosa is composed of loose, 6 glycosaminoglycan-rich connective tissue, serving as shock absorbance during valve movement. Figure 1.2: Heart and aortic valve structure. (A) Human heart structure including the four chambers and the aortic valve. Aortic valve is composed of 3 leaflets (cusps). (B) The aortic valve leaflets are made up of trilaminar structure ECM with resident VICs and a monolayer VEC on the surface. (C) The three layer of valve ECM includes the fibrosa, spongiosa, and ventricularis. Adapted with permission from Refs [38], [39]. 7 Calcific aortic valve disease (CAVD) occurs when calcium deposits form within the fibrous aortic valve leaflet matrix causing leaflets to thicken and stiffen, and eventually obstruct blood flow [18]. It is a chronic disease which progresses through two clinical stages: aortic valve sclerosis and aortic valve stenosis. The first disease stage (sclerosis), involves thickening and mild calcification in leaflet without restricting their motion. In this stage, the function of the valve is not yet significantly affected. During the later stage of CAVD (stenosis), advanced calcification and stiffened leaflets prevent the proper opening and sealing of the valve, obstructing blood flow and leading to eventual heart failure. Severe aortic valve stenosis is accompanied by clinical symptoms such as shortness of breath, angina and syncope Figure 1.3: Calcific Aortic Valve Disease. (A) Healthy valve that seal tight and regulates blood flow. (B) Disease valve where mineral deposits on valve leaflets causing thickening and stiffen of the valve. (C) Aortic valve in Calcific aortic valve disease. Adapted with permission from Refs [2]. 8 [19]. Since symptoms arise only when the disease is already advanced, early diagnosis is challenging. Moreover, without any pharmacologic treatments available, severe cases of aortic valve stenosis are only treatable via surgical or transcatheter aortic valve replacement. Recent studies, both retrospective and prospective, have identified several factors linked to the risk of calcific aortic valve disease (CAVD), including age, male gender, diabetes mellitus, obesity, hypertension, smoking, and elevated levels of plasma lipoprotein(a) (Lp(a)) and LDL cholesterol [19 - 21]. Aortic valve sclerosis affects 20-30% of individuals over the age of 65 (rising to 48% among those over 85), while aortic valve stenosis affects 2% of those over 65 (8% among those over 85), making CAVD as the most prevalent heart valve disorder worldwide [23]. 1.4 Minerals in the valve Minerals in CAVD have shown heterogeneity in chemical composition and morphology. Calcium phosphate crystals including hydroxyapatite (HA; Ca10(PO4)6(OH)2), which is known as the predominant mineral in bone, and magnesium whitlockite (Ca18Mg2(HPO4)2(PO4)12) were mineral compositions found in CAVD [17, 23, 24]. The presence of the minerals was confirmed through x-ray diffraction done on calcified tissue. Raman imaging, a vibrational spectroscopy technique commonly used to determine the chemical composition of minerals, can be used differentiate various pathological calcium phosphates. For example, the ν1 PO43− symmetric stretching mode for apatite is typically observed at ~958-960 cm-1, for whitlockite, the ν1 PO43− bands are at ∼970−972 cm−1 [25]. Other than the difference in crystal structure, energy 9 dispersive X-ray spectroscopy (EDX) revealed higher Mg/Ca and lower Ca/P ratios in whitlockite than apatite [25]. More studies have been done on the morphology of the mineral deposit found in disease valve. Calcific particles, calcific fibers, and compact calcification were found in calcific lesions under scanning electron imaging (SEM) [24]. In this work, they report that unlike calcified particles, mineralized fibers were only found in 13% of the samples, suggesting that mineralized fibers might be a later product of the disease than the calcific particles [24]. Although it is still unclear whether calcific particles play a role in inducing further mineralization, different mineral morphologies are found in diseased valve and potentially have a different formation timeline. 1.5 Role of Valve Cell Populations in CAVD Calcific aortic valve disease is known to be an active disease process driven by resident cells of the aortic valve [25, 26]. The aortic valve leaflet contains 2 cell types: valve interstitial cells (VIC) and valve endothelial cells (VEC). VICs are located throughout the valve ECM. They are a heterogenous population comprising different cell types including fibroblasts and smooth muscle cells, which are involved in the secretion and maintenance of the extracellular matrix [28]. In response to appropriate cues, VICs can transdifferentiate into activated myofibroblastic, osteoblast-like, or chondrogenic cells, which results in thickening of the valve [29]. Osteogenic differentiation is the process by which mesenchymal stem cells (MSCs) or progenitor cells differentiate into osteoblasts, which are responsible for bone formation and mineralization. Osteogenic differentiation is primarily 10 controlled by the transcription factor Runx2, which is predominantly expressed in immature osteoblasts [30]. Myofibroblastic differentiation refers to the transformation of fibroblasts or other mesenchymal cells into myofibroblasts, which are contractile cells characterized by the presence of smooth muscle-like α-smooth muscle actin (α- SMA) stress fibers [31]. Chondrogenic differentiation is controlled by the Sox9 protein and involves the differentiation of MSCs or progenitor cells into chondrocytes, which are specialized cells responsible for cartilage formation and maintenance [32]. The VEC layer that lines the surface of the leaflets serves as a selective and protective barrier between the blood and the valve ECM, maintaining the homeostatic function of the tissue [6]. VEC that are in contact with blood flow are mechanosensitive and respond to the changes in the environment. As the valve opens and closes during the cardiac cycle, the ventricular and aortic sides of the valve leaflet experience different levels of mechanical stress. The ventricular side is exposed to laminar flow as blood leaves the ventricle during systole, while the closure of the valve during diastole subjects the aortic side of the leaflet to oscillatory shear stress. Repetitive shear gradually damages the endothelial layer, triggering inflammation and calcification on the aortic (fibrosa) side of the leaflet. 1.6 Three-Dimensional (3D) in vitro models of CAVD The limitations of CAVD treatment include not only the lack of pharmacologic treatments but also the lack of early detection, prevention, and mitigation strategies. To have a more in-depth understanding of the pathological mechanism of CAVD, multiple aspects have been studied, including the role of cells and the mechanism of 11 how they interact with the environment, the structure and remodeling of the ECM in disease stages, as well as the biomechanical factors (e.g. cyclic strain, pressure, hemodynamic forces, etc.) [33]. A common approach taken in these studies involves designing in vitro models of the disease. Recently, 3D cell culture scaffolds have been shown to provide cells with a similar physiochemical environment as the native valve ECM, compared to two-dimensional (2D) models that fail to effectively mimic the complex in vivo leaflet environment [33]. Aortic valve models were developed by recapitulating components of the valve microenvironment, e.g. incorporating VIC and VEC coculture, or hydrogel constructs mimicking valve structure and mechanical properties. Lam et al. developed a 3D VIC-cultured matrigel-collagen hydrogel scaffold that can sustain dynamic mechanical stimulation (Figure 1.4(a)) [34]. In Gee et al., VIC and VEC co-cultured 3D hydrogels subjected to equiaxial mechanical constraint and tension were used to investigate cell-cell and cell-matrix interactions and how they contribute to the development of calcific lesions. Phenotype activation of cells was also observed in the system (Figure 1.4(b)) [6]. Gretchen et al. presented a microfluidic bioreactor designed to provide physiologically relevant levels of laminar or oscillatory shear stress over the VEC layer on the surface of the ECM matrix (Figure 1.4(e)). This bioreactor enabled testing the effect of shear stresses on endothelial cell mesenchymal transformation [35]. 12 Figure 1.4: In vitro models used in CAVD relevant studies. Different in vitro model designs for CAVD. (a) 3D VIC-cultured matrigel-collagen hydrogel scaffold that can sustain dynamic mechanical stimulation. (b) VIC and VEC co-cultured 3D hydrogels for cell-cell and cell-matrix interaction studies. (c) Gelatin-based hydrogel with embedded VICs micropatterned on poly(ethylene glycol) (PEG)-based hydrogel to induce topographical changes and 3D alignment. (d) Microfluidic device for applying oscillatory and unidirectional steady flow on VECs seeded on collagen gels. (e) Gelatin based microfluidic bioreactor for testing the effect of shear stresses on endothelial cell mesenchymal transformation. (f) PEG-based trilaminar structure to mimic aortic valve and anisotropic polycaprolactone (PCL) nanofibers and gelatin- GAG-based hydrogel composite to mimic fibrosa and spongiosa of the aortic valve. Adapted with permission from Refs [33]. 1.7 Thesis Overview The driving questions behind this thesis include: How do cells respond to different mineral environments in terms of matrix remodeling and nodule formation? How does the disruption of the endothelial cell layer and subsequent lesion formation differ between nanoparticle and mineralized fibril environment? What is the spatial distribution of biochemical components within the matrix? And finally, do HA nanoparticles and mineralized collagen fibrils induce different degrees of osteogenic differentiation in valve cells? 13 REFERENCES [1] S. Yadgir et al., “Global, Regional, and National Burden of Calcific Aortic Valve and Degenerative Mitral Valve Diseases, 1990-2017,” Circulation, vol. 141, no. 21, pp. 1670–1680, May 2020, doi: 10.1161/CIRCULATIONAHA.119.043391. [2] K. Driscoll, A. D. Cruz, and J. T. Butcher, “Inflammatory and Biomechanical Drivers of Endothelial-Interstitial Interactions in Calcific Aortic Valve Disease,” Apr. 30, 2021, Lippincott Williams and Wilkins. doi: 10.1161/CIRCRESAHA.121.318011. [3] E. J. Farrar, V. Pramil, J. M. Richards, C. Z. Mosher, and J. T. Butcher, “Valve interstitial cell tensional homeostasis directs calcification and extracellular matrix remodeling processes via RhoA signaling,” Biomaterials, vol. 105, pp. 25–37, Oct. 2016, doi: 10.1016/j.biomaterials.2016.07.034. [4] K. L. Sider, M. C. Blaser, and C. A. Simmons, “Animal Models of Calcific Aortic Valve Disease,” Int J Inflam, vol. 2011, pp. 1–18, 2011, doi: 10.4061/2011/364310. [5] M. W. Tibbitt and K. S. Anseth, “Hydrogels as extracellular matrix mimics for 3D cell culture,” Jul. 01, 2009. doi: 10.1002/bit.22361. [6] T. W. Gee, J. M. Richards, A. Mahmut, and J. T. Butcher, “Valve endothelial- interstitial interactions drive emergent complex calcific lesion formation in vitro,” Biomaterials, vol. 269, p. 120669, Feb. 2021, doi: 10.1016/J.BIOMATERIALS.2021.120669. [7] H. C. W. Skinner and H. Ehrlich, “Biomineralization,” in Treatise on Geochemistry: Second Edition, vol. 10, Elsevier Inc., 2013, pp. 105–162. doi: 10.1016/B978-0-08-095975-7.00804-4. [8] E. Tsolaki and S. Bertazzo, “Pathological mineralization: The potential of mineralomics,” Oct. 01, 2019, MDPI AG. doi: 10.3390/ma12193126. [9] V. Šromová, D. Sobola, and P. Kaspar, “A Brief Review of Bone Cell Function and Importance,” Nov. 01, 2023, Multidisciplinary Digital Publishing Institute (MDPI). doi: 10.3390/cells12212576. [10] N. Vidavsky, J. A. M. R. Kunitake, and L. A. Estroff, “Multiple Pathways for Pathological Calcification in the Human Body,” Adv Healthc Mater, vol. 10, no. 4, Feb. 2021, doi: 10.1002/adhm.202001271. 14 [11] D. Ralph, K. van de Wetering, J. Uitto, and Q. Li, “Inorganic Pyrophosphate Deficiency Syndromes and Potential Treatments for Pathologic Tissue Calcification,” May 01, 2022, Elsevier Inc. doi: 10.1016/j.ajpath.2022.01.012. [12] D. Bazin, M. Daudon, C. Combes, and C. Rey, “Characterization and some physicochemical aspects of pathological microcalcifications,” Oct. 10, 2012. doi: 10.1021/cr200068d. [13] L. N. Poloni and M. D. Ward, “The materials science of pathological crystals,” Jan. 14, 2014. doi: 10.1021/cm402552v. [14] B. S. Franklin, M. S. Mangan, and E. Latz, “Crystal Formation in Inflammation,” May 20, 2016, Annual Reviews Inc. doi: 10.1146/annurev- immunol-041015-055539. [15] F. Shaffer, R. McCraty, and C. L. Zerr, “A healthy heart is not a metronome: an integrative review of the heart’s anatomy and heart rate variability,” Front Psychol, vol. 5, Sep. 2014, doi: 10.3389/fpsyg.2014.01040. [16] R. B. Hinton and K. E. Yutzey, “Heart valve structure and function in development and disease,” Annu Rev Physiol, vol. 73, pp. 29–46, 2011, doi: 10.1146/annurev-physiol-012110-142145. [17] M. S. Sacks, W. David Merryman, and D. E. Schmidt, “On the biomechanics of heart valve function,” Aug. 25, 2009. doi: 10.1016/j.jbiomech.2009.05.015. [18] J. M. Richards et al., “Crystallinity of hydroxyapatite drives myofibroblastic activation and calcification in aortic valves,” Acta Biomater, vol. 71, pp. 24–36, Apr. 2018, doi: 10.1016/j.actbio.2018.02.024. [19] A. Sai Harika Pujari and P. Agasthi Affiliations, “Aortic Stenosis Continuing Education Activity.” [Online]. Available: https://www.ncbi.nlm.nih.gov/books/NBK557628/?report=printable [20] P. Mathieu and B. J. Arsenault, “CAVD: Civilization aortic valve disease,” Jul. 21, 2017, Oxford University Press. doi: 10.1093/eurheartj/ehx219. [21] B. Fendley Stewart et al., “Clinical Factors Associated With Calcific Aortic Valve Disease.” [22] J. Ljungberg et al., “Traditional cardiovascular risk factors and their relation to future surgery for valvular heart disease or ascending aortic disease: A case- referent study,” J Am Heart Assoc, vol. 6, no. 5, May 2017, doi: 10.1161/JAHA.116.005133. 15 [23] S. Bertazzo and E. Gentleman, “Aortic valve calcification: A bone of contention,” Apr. 21, 2017, Oxford University Press. doi: 10.1093/eurheartj/ehw071. [24] J. M. Gawoski, K. Balogh, and W. J. Landist, “Aortic valvular tophus: identification by X-ray diffraction of urate and calcium phosphates,” 1985. [25] F. A. Shah et al., “Micrometer-Sized Magnesium Whitlockite Crystals in Micropetrosis of Bisphosphonate-Exposed Human Alveolar Bone,” Nano Lett, vol. 17, no. 10, pp. 6210–6216, Oct. 2017, doi: 10.1021/acs.nanolett.7b02888. [26] P. Ohukainen, H. Ruskoaho, and J. Rysa, “Cellular Mechanisms of Valvular Thickening in Early and Intermediate Calcific Aortic Valve Disease,” Curr Cardiol Rev, vol. 14, no. 4, pp. 264–271, Aug. 2018, doi: 10.2174/1573403x14666180820151325. [27] P. R. Goody et al., “Aortic Valve Stenosis: From Basic Mechanisms to Novel Therapeutic Targets,” Apr. 01, 2020, Lippincott Williams and Wilkins. doi: 10.1161/ATVBAHA.119.313067. [28] L. H. M. Moncla, M. Briend, Y. Bossé, and P. Mathieu, “Calcific aortic valve disease: mechanisms, prevention and treatment,” Aug. 01, 2023, Nature Research. doi: 10.1038/s41569-023-00845-7. [29] M. Bogdanova et al., “Interstitial cells in calcified aortic valves have reduced differentiation potential and stem cell-like properties,” Sci Rep, vol. 9, no. 1, Dec. 2019, doi: 10.1038/s41598-019-49016-0. [30] M. T. Valenti, L. D. Carbonare, and M. Mottes, “Osteogenic differentiation in healthy and pathological conditions,” Jan. 01, 2017, MDPI AG. doi: 10.3390/ijms18010041. [31] Y. Tai et al., “Myofibroblasts: Function, formation, and scope of molecular therapies for skin fibrosis,” Aug. 01, 2021, MDPI. doi: 10.3390/biom11081095. [32] M. R. Casanova, H. Osório, R. L. Reis, A. Martins, and N. M. Neves, “Chondrogenic differentiation induced by extracellular vesicles bound to a nanofibrous substrate,” NPJ Regen Med, vol. 6, no. 1, Dec. 2021, doi: 10.1038/s41536-021-00190-8. [33] I. Tandon, A. Ozkizilcik, P. Ravishankar, and K. Balachandran, “Aortic valve cell microenvironment: Considerations for developing a valve-on-chip,” Dec. 01, 2021, American Institute of Physics. doi: 10.1063/5.0063608. 16 [34] N. T. Lam, H. Lam, N. M. Sturdivant, and K. Balachandran, “Fabrication of a matrigel-collagen semi-interpenetrating scaffold for use in dynamic valve interstitial cell culture,” Biomedical Materials (Bristol), vol. 12, no. 4, Jul. 2017, doi: 10.1088/1748-605X/aa71be. [35] G. J. Mahler, C. M. Frendl, Q. Cao, and J. T. Butcher, “Effects of shear stress pattern and magnitude on mesenchymal transformation and invasion of aortic valve endothelial cells,” Biotechnol Bioeng, vol. 111, no. 11, pp. 2326–2337, Nov. 2014, doi: 10.1002/bit.25291. [36] S. Park, S. Choi, A. A. Shimpi, L. A. Estroff, C. Fischbach, and M. J. Paszek, “COLLAGEN MINERALIZATION DECREASES NK CELL-MEDIATED CYTOTOXICITY OF BREAST CANCER CELLS VIA INCREASED GLYCOCALYX THICKNESS”, doi: 10.1101/2024.01.20.576377. [37] A. Ressler, A. Žužić, I. Ivanišević, N. Kamboj, and H. Ivanković, “Ionic substituted hydroxyapatite for bone regeneration applications: A review,” Open Ceramics, vol. 6, Jun. 2021, doi: 10.1016/j.oceram.2021.100122. [38] G. A. Fishbein and M. C. Fishbein, “Pathology of the Aortic Valve: Aortic Valve Stenosis/Aortic Regurgitation,” Aug. 01, 2019, Current Medicine Group LLC 1. doi: 10.1007/s11886-019-1162-4. [39] P. Ohukainen, H. Ruskoaho, and J. Rysa, “Cellular Mechanisms of Valvular Thickening in Early and Intermediate Calcific Aortic Valve Disease,” Curr Cardiol Rev, vol. 14, no. 4, pp. 264–271, Aug. 2018, doi: 10.2174/1573403x14666180820151325. 17 CHAPTER 2 CELLULAR RESPONSE TO MINERALIZED COLLAGEN FIBRILS IN A 3D CO- CULTURE SPRING GEL MODEL FOR CALCIFIC AORTIC VALVE DISEASE 2.1 Introduction In vitro CAVD models of the different disease stages can help understand the mechanism of disease progression and can be a useful tool for therapeutic development and diagnosis. Specifically, a 3-dimensional (3D) in vitro valve model has provided an opportunity to understand the communication between valve cell populations and their environment. It is known that there is heterogeneity in mineral morphology found in calcified valves and that discrete crystalline particles exist alongside fibril or bone-like calcification. Observing how the cells respond to the different minerals in the matrix can help understand the mechanism of the different stages of the disease and, therefore, the progression of the disease. This work aims to observe the response of valve cells to hydroxyapatite nanoparticles and mineralized collagen fibrils incorporated into the extracellular matrix using an in vitro 3D co- culture aortic valve model system. 2.2 Experimental Design To observe the cell response to mineral environment, we need to start from building an in vitro model that can accurately present the aortic valve. The 3D in vitro spring gel model used in this work was adapted from Gee et al. [1]. The model is composed of a collagen hydrogel cast into a cylindrical well in a PDMS mold. A 18 circular spring anchors the hydrogel to accommodate ECM remodeling and provides an equiaxial stress state. To create a model that recapitulates the composition of the valve leaflet fibrosa, the collagen hydrogel is seeded with primary porcine VICs to form the extracellular matrix of the leaflet constructs. A monolayer of primary porcine VECs was cast on the surface of the collagen hydrogel to mimic the aortic-side endothelial layer. Compared to a 2-dimensional monoculture model, this co-cultured 3D hydrogel model allows us to study the communication and signaling between VIC and VEC populations and their response to their environment. [2] In this model, some of the important factors that we considered are cell population, the mineral incorporated, and the growth. The initiation of CAVD is believed to begin with early Figure 2.1 Design of 3D in vitro spring gel model from Gee et al. (A) Hydrogel compacts towards the center during the remodeling of the matrix. A stainless spring is to provide equiaxial tension for the hydrogel when undergo compaction. (B) 3D spring gel model is a type-1 collagen- based hydrogel. VICs were located throughout the bulk volume of the hydrogel and monolayer VECs were topped on the surface to mimic the structure of valve leaflets. Adapted with permission from Refs [1]. 19 inflammation and dysfunction of the valve endothelial cells, valve interstitial cells contribute to the disease progression through myofibroblastic differentiation and osteogenic differentiation. Both cell population play a role in CAVD, but more importantly, there is active communication between VIC and VEC. Previous work from Gee et al. [3] has found evidence of VEC actively contributing to VIC pathological remodeling and calcification. Therefore, having VIC and VEC co-culture is critical in our in vitro model when it comes to understanding the interaction between cells and the matrix. Primary porcine cells were used in our model. Porcine valves are a widely used analog for human adult valve endothelial and interstitial cells. Pigs develop atherosclerosis and valvular lesions without intervention, which is similar to humans [3]. Hydroxyapatite is known to be one of the main minerals found in calcified valves and bone [7, 8]. As discussed in Chapter 1, heterogeneity in mineral morphology is a hallmark of diseased valves. While spherical particles were thought to be the first mineral structure observed, perhaps even prior to lesion formation, bone- like mineralized collagen fibers are only observed in small fraction of calcific lesions. From here, we hypothesize that the formation of bone-like mineralized fibrils happens at a later stage than calcific nanoparticles. Therefore, in this work, we incorporate HA nanoparticles and bone-like, intrafibrillar mineralized collagen fibrils into the in vitro model to observe the cell behavior in the different matrix environment. Since CAVD is a long-term progression that can take up to decades, we need a faster way to generate calcification in out hydrogel model. Here we culture our hydrogel in osteogenic media (OGM), which is basal-growth medium (GM), a medium that 20 contain the necessities for cells to live, with supplements that induce osteogenic differentiation of cells. To answer our questions of how cell behave in the mineral environment, we monitor the degree of gel compaction and the lesion formation as clues to the remodeling of the matrix. Then by using Raman microscopy, we analyze the chemical composition of both the lesion and off-lesion area and map out the spatial distribution of matrix components. Finally, we combine different histological stains to look at cell behavior including osteogenic differentiation, myofibroblastic differentiation, cell movements, and cell proliferation. 2.3 Materials and Methods Experiments were carried out over a 1- to 1.5-week time frame by incubating co-culture spring gels with incorporated HA nanoparticles and mineralized collagen fibril, respectively. Samples were then fixed or sectioned for further characterization, including the degree of lesion formation, chemical composition and spatial distribution, histology, and lesion morphology. Figure 2.2 shows a schematic of the flow of the experimental methods. 21 Figure 2.2: Experimental Methods. (A) Cells isolated from porcine aortic valves were used in the in vitro model for their ability to form valvular lesions like human valves. (B) Porcine valve interstitial and endothelial cells were cultured in flask before used. (C) 3D in vitro co-culture spring gel is a type 1 collagen-based hydrogel with valve interstitial cells and minerals dispersed throughout the bulk volume of the gel, and then topped with a layer of valve endothelial cells on the surface. (D) Gels were cultured in osteogenic media at 37 °C and 5% CO2 for 10 days before cryoembedding or fixed for further characterization. (E) Characterization used include degree of gel compaction and lesion formation, Raman spectroscopy mapping for spatial distribution of different components, and different histology staining for cell behavior. 22 2.3.1 Hydroxyapatite (HA) nanoparticles Hydroxyapatite nanoparticles were synthesized by aqueous precipitation as described by Richards et al. [4]. Ammonium phosphate dibasic solution ((NH4)2HPO4, 10 mM) was added dropwise at a rate of 10 mL/min to a solution of calcium nitrate tetrahydrate (Ca(NO3)2 • 4H2O, 10 mM) while rapidly stirring over ice (4°C). Before combining, solutions were first balanced to pH 10 using concentrated ammonium hydroxide (NH4OH). Hydroxyapatite nanoparticles were dialyzed in NaOH (pH 11) before incorporation into the hydrogel. A Raman spectrum showed the signature of PO4 3- (960 cm-1) which is indicative of hydroxyapatite mineral. TEM showed HA nanoparticles in rod-like shape, with sizes around 20 nm on the longer side (Fig 2.3). 2.3.2 Mineralized collagen fibrils (MCF) Mineralized collagen fibrils (MCF) were synthesized by the Eli Sone group at the University of Toronto. MCFs are uncrosslinked collagen fibrils mineralized via a polymer-induced liquid-precursor (PILP) process, forming intrafibrillar hydroxyapatite in the collagen fibrils. Lyophilized MCFs were resuspended in pH 8 NaOH at the desired concentration (14.04 mg/mL) before incorporation into the spring gel system. Characterization of the MCFs is shown in Fig 2.3, with the same Raman spectrum of hydroxyapatite seen in the nanoparticles as well as amide III (~1247 cm-1) indicating the presence of collagen. From the TEM and SEM images we can see the fibril structure of the MCFs but no free HA nanoparticles around, indicating the existence of intrafibrillar mineral. 23 Figure 2.3: Characterization of HA nanoparticles and mineralized collagen fibrils (MCF). Raman spectra and TEM image of HA nanoparticles provided. by Stephan Sutter. TEM image of MCF provided by Professor Eli Sone from University of Toronto (unpublished data). n 24 2.3.3 PDMS spring construct mold preparation PDMS molds were made by mixing PDMS activator and PDMS base with a 1:10 mass ratio. The PDMS mixture was then cast over a positive mold with the same dimensions as the final collagen gel and cured at 60°C overnight. Each resulting PDMS negative (cylindrical well) mold was punched out and circular springs were assembled into the mold. PDMS spring constructs were autoclave-sterilized before use. 2.3.4 Porcine VIC/VEC culture Our protocol of primary cell isolation was adapted from Gould et al. [5] Valve endothelial cells (VEC) and valve interstitial cells (VIC) were harvested from fresh porcine aortic valves (provided by Timberline Meats, Dundee, NY). Cells were isolated by collagenase digestion, by swabbing the endothelial layer gently and dabbing the swab in cold collagenase solution (600u/mL collagenase in 1x DMEM, 1x P/S and 10% FBS) to dislodge the cells. Collagenase solutions were then centrifuged at 300g for 5 min. 3mL of VEC media (1x DMEM, 10% FBS, 1% 50u/mL heparin) were added into the VEC tube and centrifuged again at 300g for 5 min. VEC were resuspend in 5mL of VEC media and plated in collagen-coated T25 culture flasks for further culture and passaging. VIC were isolated by incubating valve leaflets in a tube containing supernatant collagenase for 12-18 hours. VIC are cultured at 37°C and 5% CO2 in DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. VEC were cultured under the same conditions, but the 25 flasks were pre-coated with 50 µm/mL rat-tail collagen I for 30 minutes before cells were introduced. VIC and VEC cultures were used between passage four and six. 2.3.5 Co-culture cellularized collagen hydrogel casting In a 24-well plate, 50 µL of PBS were added in the well to assist with mold adhesion to bottom of well. PDMS spring constructs were pushed to sit in the bottom of the well with forceps. PDMS spring constructs were then treated with 50µL 0.2% Pluronic acid solution in PBS for 30 minutes ahead as a lubricant for casting collagen hydrogel. Porcine valve interstitial cells (VIC) were suspended at 4.00E+05 cells per mL in 5x DMEM with 10% FBS, H2O, and type l rat-tail collagen diluted to a final gel concentration of 2 mg/mL. HA nanoparticles and MCFs were added at 0.25 mg/mL into hydrogel mixture, respectively. The mixture pH was then adjusted to pH 7–8 using either 0.1M NaOH or 0.1M HCl, and 100 µL were cast into spring construct molds. VIC- and mineral-bearing collagen gels were incubated (37°C and 5% CO2) for an hour to allow the gel to set. To seed the porcine valve endothelial cells (VEC), VEC were suspended at 5.00E+04 cells per mL in DMEM. 50µL of the VEC suspension were cast on top of the collagen hydrogel to form a domed droplet and were left in the incubator (37°C and 5% CO2) for 2 hours and 45 minutes. All constructs were cultured in general media (GM) (DMEM, 10% FBS, 1% penicillin- streptomycin) for the first night of incubation, after which the media was switched to osteogenic media (OGM) (DMEM, 10% FBS, 1% penicillin-streptomycin, 0.5M B- 26 glycerophosphate, 2.5mg/mL ascorbic acid, dexamethasone supplement) the next day. Media was exchanged every other day, and collagen hydrogels were cultured for up to 10 days. Optical microscope images were taken daily to track gel compaction. After 10 days of culture, collagen hydrogels are fixed with 4% PFA for 20 minutes, then washed 3x with PBS for 10 minutes each, and stored in 70% ethanol at 4°C. 2.3.6 Cryo-embedded and sectioning of samples Gels were punched out with a biopsy punch prior to fixation and cryo- embedded in Optimal Cutting Temperature compound (OCT; Tissue-Tek O.C.T. compound, Sakura), then sectioned at -20°C. Vertical cross-sections of gels were taken as serial sections alternating between thicknesses of 10 µm for histology and 30 µm for Raman spectroscopy. 2.3.7 Raman Spectroscopy and mapping of gels Raman spectroscopy is a technique that utilizes the inelastic scattering of monochromatic light to provide vibrational fingerprinting of molecules, offering insights into molecular composition and structure. With the high resolution of the Raman spectroscopy, single-point spectra provide high selectivity and are able to distinguish various components in the samples. Raman mapping can create hyperspectral maps of both biomolecules and mineral in the area of interest, revealing the spatial distribution and variation in molecular composition across the surface. A confocal Raman microscope (Alpha300R, WITec) was used to collect Raman data. Data collection of Raman spectroscopy analysis was done by using the 27 532 nm laser excitation source with a laser power of 45 mW. The diffraction gratings of the spectrometer were adjusted to 300 1/mm and the spectrum center was set to 2100 cm-1 for large area scanning to capture both the minerals and organic components. Raman mapping of the hydrogels was done in two ways: top-down view of whole-mount gels and cross-section. Whole-mount gels were hydrated with 1X PBS and a bright field overview of the gels were taken using a 5x objective (Zeiss, 0.25 NA). Then we switched to the 63x dipping objective for more higher-resolution images. Using the dipping objective can ensure us to image hydrated hydrogel with its original structure. Mappings were done using large area scanning and further processed and analyzed via WITec Project. 30 µm thick cross sections of the spring gel (as mentioned in 2.3.6) were used for Raman mapping. Sample cross sections were hydrated with 1x PBS. Overview of the sections were taken with the dipping objective and mapping was done with the 63x dipping objective. Raman maps were processed starting from the removal of cosmic rays, followed by different filters for individual maps of the chemical components of interest. Filters used in this study are n1 PO4 (960 cm-1) for hydroxyapatite mineral, Amide III (1226 – 1316 cm-1) for collagen, Amide III (1284 - 1359cm-1) for non- collagenous proteins (NCP), and -CH- stretch (2790 – 2872 cm-1) for lipid. 28 2.3.8 Histology Hematoxylin and eosin (H&E) staining was done to visualize the cellular and tissue structures around lesions. Sample sections were rehydrated by passing them through a series of graded alcohols (100%, 100%, 70%) and then rinse in water. Staining with Hematoxylin: The rehydrated tissue sections were stained with hematoxylin (3 min), rinsed in water to remove excess hematoxylin, and differentiated by immersing the slides in an acid-alcohol solution (acid alcohol) to remove non- specific background staining. Blue the sections by immersing them in 0.1% sodium bicarbonate to enhance the hematoxylin staining. Staining with Eosin: Sample sections were stained with eosin (45 sec) follow with a rinse in water to remove excess eosin. Dehydration, Clearing, and Mounting: Samples were dehydrated by passing them through a series of graded alcohols (95%, 100%, 100%) and cleared by immersing them in D-limonene. Lastly, a drop of mounting medium was applied on the stained section and covered with a glass coverslip. 2.3.9 Immunofluorescence staining Cross-section staining: Tissue sections were permeabilized with PBST (0.05% Tween-20 in 1x PBS) and blocked with 5% goat serum and 0.3M glycine, 1% bovine serum albumin (BSA) for 1 hour at room temperature. Samples were incubated at 4°C overnight with corresponding primary antibodies, in this case, rabbit RUNX2 Polyclonal Antibody (Invitrogen, Waltham, MA) and Anti-Hu/Mo SOX9 (Invitrogen, Waltham, MA). Primary antibodies were diluted in 1% bovine serum albumin (BSA) 29 and 0.3M glycine (VWR International, West Chester, PA) in PBST. Samples were rinsed with PBST 3 times for 10 min each. Species-specific secondary antibodies raised in goat conjugated to Alexa Fluor® (Invitrogen, Waltham, MA), in this case, RUNX2 at 488 or SOX9 at 568 fluorophores, were added at 4 μg/ml (1:500 dilution ratio) (Invitrogen, Grand Island, NY). Tissue sections were incubated with the secondary antibody solution for 1 hour at room temperature in the dark. After 3 10- minute PBS, sections were incubated with autofluorescence quenching solution (Vector TrueView Reagent) for 5 min at room temperature. Nuclei were stained with NucBlue DAPI dye (8 droplets/ml) for 20 min. Samples were washed in PBS 3 times each for 10 min, then blotted dry and mounted with Prolong® Gold Antifade mounting medium and covered with a glass cover slip. Sections were imaged with the Keyence BZ-X800 series. Whole-mount staining: After fixing with 4%PFA (in PBS), samples were blocked with the blocking buffer (10% v/v Goat serum in PBST (0.1%Tween-20 in 1x PBS)) for 1 hour at room temperature on an orbital shaker followed by three PBST washes, each for 5 min. Samples were then incubate in primary antibody solution (1% primary antibody in 1% BSA (in PBST)) overnight at 4°C. Before applying secondary antibody to the samples, three PBST washes, each for 30 min, were done at 4°C on an orbital shaker. Samples were incubated in secondary antibody solution (0.2% secondary antibody in 1%BSA (in PBST)) for 1 hour at room temperature on an orbital shaker. After three PBS washes, each for 30 min, were done at 4°C on an orbital shaker, samples were incubated in NucBlue DAPI dye (8 droplets/ml) in 1x PBS for 30 min at room temperature as nuclear counterstain. Finally, after three PBS 30 washes, each for 5 min, samples were equilibrated in optimal clearing solution (1:1 Glycerol: PBS solution) overnight at 4°C before ready to image. 2.3.10 Scanning electron microscopy (SEM) imaging Prior to SEM imaging, collagen hydrogels need to be fully dehydrated. The drying protocol was adapted from [36]. [6]. Spring gels were placed in a 24 well plate and first incubated in 2% glutaraldehyde in cacodylate buffer (0.05M in sodium cacodylate in DI, pH 7.4) at 4°C for 2 hours, followed by a 1% cacodylate buffer rinse at room temperature three times, each for 10 minutes. The gels were then fixed with 1% osmium tetroxide in cacodylate buffer at room temperature for 1 hour followed by 1% cacodylate buffer rinse three times each for 5 minutes. Next, samples were dehydrated with serial ethanol washes (25%, 50%, 75%, 90%, 100%), each for 10 minutes, and stored in 100% ethanol overnight. Samples were dried with a critical point dryer (missing instrument name) for 24 hours, then coated with carbon (time & current settings, sputter coater manufacturer (denton)) and imaged using a Sigma-500 field-emission scanning electron microscope (FE-SEM, Zeiss). 2.4 Results and Discussion 2.4.1 Casting of MCF incorporated hydrogel The casting of HA nanoparticles incorporated hydrogel was already established by Butcher and coworkers [1]. In brief, 14.04 mg/mL of HA nanoparticles were dialyzed into NaOH and were added into a collagen solution to form 0.25 mg/mL concentration of mineral incorporated hydrogel. To incorporate lyophilized MCFs into 31 the hydrogel system mentioned above, I tested different concentrations of MCFs in stock collagen solution and noticed that the solution would not form hydrogel when the MCF concentration are too high (Table 2.1). Collagen hydrogels form when fibrils crosslink with each other. When mineralized collagen fibrils, which have a weaker crosslinking ability, are mixed with collagen, there is a maximum ratio of MCFs to unmineralized collagen fibers at which a hydrogel can form. In this thesis, I focus on using the concentration of 0.25 mg/mL to match the protocol for HA nanoparticle hydrogels, but the versatility of the MCFs leaves more possible applications in the 3D in vitro valve model. Table 2.1: Mass ratio of MCF and stock collagen tested in acellular spring gel model. Mass ratio (Col:MCF) Successful gelation Note 100:1 Yes Concentration used in the rest of this study 100:2 Yes No significant difference compared to 100:1 ratio 1:1 No 1:100 No 2.4.2 Hydrogel compaction analysis Previous work from Gee et al. found that when HA nanoparticles are incorporated into a VIC and VEC co-cultured hydrogel, the hydrogels compacted significantly more when cultured under osteogenic media condition then cultured 32 basal-medium condition [1]. As the first step of testing the cells response to the MCFs, we started from observing the effect of osteogenic media (OGM) on MCF gels using basal-medium (GM) as the control. MCF gels were cultured in basal-medium (GM) and osteogenic media (OGM) to observe the compaction of the hydrogels under different culturing conditions. Bright field images of the hydrogels were taken daily (Fig 2.4). On day 8, hydrogels cultured in osteogenic media have significant compaction while those gels cultured in basal medium do not show much compaction, indicating that the effect of osteogenic media also exists in MCF hydrogels. To quantify the degree of compaction, Fiji was used to measure the area along the hydrogel boundary and compared to the original gel area (day 1) in percentage (Fig 2.5). The experiment was done with a sample size of 4 for each condition (n=4), the daily area ratio was average between the samples. The MCF_GM gel ended up at 0.9 area ratio on day 8 and the area of MCF_OGM gel started shrinking on day 4 and significantly dropped to 0.44 on day 8. Between HA nanoparticle gels and MCF gels, there is no significant difference in the final degree of compaction in both basal- medium and osteogenic media conditions. Another important feature to pay attention to are the cellularized lesions (indicated by red arrows in Fig 2.4) that form on the hydrogels. Visually there is more lesion formation in the gels cultured in OGM. Once we knew how MCF gels behave in GM and OGM conditions, we focused onto the osteogenic media condition of HA nanoparticle gels and MCF gels. 33 Figure 2.4: Bright field image of MCF hydrogel under GM and OGM condition. Pictures taken on day 1, day 5, and day 8 of culturing. Red dash circle shows the area where the hydrogel is still intact, and red arrows pointed to lesions. On day 8, hydrogel under OGM condition seemed to showed significant compaction compare to GM condition. 34 Figure 2.5: Time-point hydrogel compaction of HA and MCF gels under both GM and OGM conditions. GM conditions did not seem to compact a lot (0.9) compared to the OGM conditions which compacted to around 0.44 on day 8. Even though HA_OGM and MCF_OGM both ended up compacting around the same degree, it seemed to have gone through different process. 35 36 Figure 2.6: T-test done on compaction of four condition, HA_GM, HA_OGM, MCF_GM, MCF_OGM. (A) Bar chart showing compaction between day 4 and day 8 of each condition (n=5). (B) Bar chart showing compaction between HA_GM and MCF_GM on day 4 and day 8 (n=5). (C) Bar chart showing compaction between HA_OGM and MCF_OGM on day 4 and day 8 (n=5). [* p < 0.05] [ D p < 0.0001] [ns p > 0.05]. Figure 2.6 compares the different conditions using t-test. Within each condition, compaction of all four conditions (HA_GM, HA_OGM, MCF_GM, MCF_OGM) showed significant difference between day 4 and day 8. Between HA_GM and MCF_GM, differences in degree of compaction were not significant in either day 4 or day 8. However, between HA_OGM and MCF_OGM, there is significant difference in degree of compaction on day 4 despite they ended at a similar degree of compaction on day 8, showing that there might be some different mechanism happening during the culturing. 37 To make sure the hydrogels were always under the uniaxial strain provided by the spring, the culturing of the OGM gels were stopped right before the gels started to detach from the inner ring of the spring (around day 8 to day 10). While HA nanoparticle gels and MCF gels both reached 40% - 50% compaction roughly at the same time, HA nanoparticle gels seem to have started to compact earlier than MCF gels (Fig 2.7). On day 5, when MCF gels were still fully attached to the PDMS, compaction on HA gels has already left a gap between the hydrogel boundary and the PDMS mold (Fig 2.7). Similar situations were seen over different batches of experiment (Fig 2.8). The three replicates were seeded with different batches of VICs and VEC, which means the batch-to-batch difference between cells, e.g., speed of metabolism, can affect the rate of compaction, but the degree of the effect is not understood yet. Even though there are no specific time points that correlate to the start of the compaction or when the gel started to rapidly compact, we noticed that HA gels started compacting earlier than MCF gels. 38 Figure 2.7: Bright field image of MCF hydrogel and HA nanoparticle hydrogel under OGM condition. Pictures taken on day1, day5, and day8, red dash circle show the area where the hydrogel is still intact and red arrow shows where cellularized lesions were located. 39 40 Figure 2.8: Hydrogel compaction of HA_OGM and MCF_OGM gels over three replicates (n = 6). Culturing was stopped right before the hydrogel teared off the spring due to compaction. 41 2.4.3 Lesion formation Lesions are known to be an important hallmark of CAVD. In the bright field images of both HA_OGM gels and MCF_OGM gels, lesion formation was observed. To quantify the lesions that were formed, the area of lesion coverage was compared to the area of the hydrogel within the inner circle of the spring, shown as the red dashed circle in Fig 2.9 (B). Bright field images for lesion quantification were taken with the 5x objective of Raman microscopy to get higher resolution image of the lesions. Degree of lesion coverage is defined by the total surface area of lesion seen on the surface of the gel compared to the surface area of the hydrogel. There is no significant difference seen in the degree of lesion formation between HA_OGM gels (~30%) and MCF_OGM gels (~27%) Fig 2.9 (A). However, there seems to be a difference in the morphology of the lesions between HA_OGM and MCF_OGM gels. Lesions on HA_OGM gels looks more diffuse throughout the surface, where it seems like one big lesion covering the surface, whereas lesions on MCF_OGM gels looks more localized. Now that we knew there are lesions forming in both HA and MCF gels, we want to further investigate the structure and chemical composition of the lesions. 42 Figure 2.9: Lesion formation. (A) Degree of lesion formation on Day 9 in percentage. (B) Bright field images of gels showing lesion formation on the surface. Bright field images are taken with the 5x objective of Raman microscopy. Darker area with the red dash lines were considered lesions. Left: HA_OGM. Right: MCF_OGM. 43 2.4.4 H&E staining on gel cross sections H&E staining provides important information about the pattern, shape, and structure of cells in a tissue sample. Hematoxylin stains nucleic acids with a deep blue-purple color and eosin is pink and stains proteins nonspecifically. In a typical tissue, the cytoplasm and extracellular matrix stained by eosin will show varying degrees of pink color. Here, to get a general information of the distribution of nuclei and extracellular matrix, we stained the cross section (thickness: 10 µm) with H&E. Several observations were made based on the H&E staining. On the top surface, there is a thin, dark pink layer representing the endothelial cells that were seeded on top of the hydrogel. Beneath this endothelial layer, where the lesions are located, there is a notably high cell density compared to the rest of the hydrogel. Between the lesions and the off-lesion region of the gels there is a band-like structure that we suspect has collagen due to its shade of pink. Interesting differences between HA_OGM and MCF_OGM gels is the matrix morphology seen, including how the pink band-like structure seemed to stretch into the off-lesion area more in the HA_OGM gels than the MCF_OGM gels. In the off-lesion region, in the MCF_OGM gels, the matrix appears to have a more linear pattern while the matrix in HA_OGM gels does not show specific pattern. Also, the MCF_OGM gel seemed to have shrunk in height (~500 µm) according to the cross section that we get compared to the HA_OGM (~790 µm). With the general observations from the H&E staining, we want to further understand the chemical composition of the different matrix regions. We used Raman 44 microscopy to look at the chemical composition of the lesion and off-lesion region and used different histological stains to provide information of cell behavior. Figure 2.10: H&E cross section staining of HA_OGM and MCF_OGM gels. (A)- (B) Zoom-in image of HA_OGM and MCF_OGM across the on-lesion and off-lesion area, showing cell-dense lesion and extracellular matrix. H&E images taken with ScanScope (Aperio ScanScope Cs2, Leica). A B 45 2.4.5 Raman mapping of lesions on whole-mount 2D Raman mapping was done on whole-mount along with the PDMS mold. To maintain the structure of the hydrogels, we keep them hydrated with a drop of 1xPBS on top. Region of interest for Raman mapping includes areas on-lesion and off-lesion, which we determine through a bright field overview of the whole gel. From a HA_OGM gel, a map was taken at a perinodular region (half on the lesion, half off the lesion), due to the difference in y-axis between the lesion area and the off-lesion area, components were only captured clearly on the lesion side. Weak HA signals were seen (Figure 2.12), while collagen and non-collagenous protein seemed to be strongly overlapping with each other. Due to the noises, DNA signals are difficult to identify. Figure 2.11: High resolution H&E staining of lesion area. Images were taken with the Keyence BZ-X800 series at 40x. 46 47 Figure 2.12 Raman whole-mount map of HA_OGM gel (perinodular). Map taken at lesion periphery. Raman spectrum does not show very distinct difference between the different components, including weak signal of HA and very little difference between NCP and collagen. DNA signals were captured but not confident. 48 49 Figure 2.13 Raman whole-mount map of MCF_OGM gel (perinodular). Map taken at lesion periphery. No MCF or HA nanoparticles were seen in the map, and the collagen and NCP seemed to be overlapping similar as HA_OGM. Under the Raman microscope, I noticed that no MCFs are found on the surface of the gels, instead some were found around 100 µm below the surface of lesions, but the MCFs and lesions does not seem to have spatial correlation. Spectrum in Figure 2.13 also show very weak signal at around 780 cm-1, which we consider as the O-P-O backbone of DNA. During data collection, we noticed some difficulties. First, little HA particles and no MCFs were found on the lesion surface of the gels, instead we saw MCFs around 100 µm below the surface. This cause difficulty in finding the spatial 50 correlation of the lesions and the minerals since they are not on the same elevation. Secondly, the lesions are highest in the center, lower toward the edges, the uneven level of the lesions makes it easy to lose focus while mapping the gels. Lastly, relatively noisy signals were seen during the process of mapping. One approach for this problem is to do a 3D Raman mapping, which can capture components on different height. However, to do that we need a very long scanning time, and we need to deal with the low signal when mapping area deeper in the gel. This limits the amount of information we can get. Another approach is to map cross-sections of our spring gel samples to get information in-depth, results will be discussed in the following section. 2.4.6 Raman mapping of lesions on cross sections By doing Raman mapping on sample cross section, we were able to get both on lesion and off-lesion mapping on the same surface within shorter time and stronger signals. Further information can be collected by performing histology on the serial sections. Cross-sections were rehydrated with 1X PBS before imaging because we noticed in dry samples weaker signals of DNA and the difficulty to tell collagen and NCP apart (shown in Appendix 1 & 2). Areas of interest are located toward the middle of the cross-section to prevent potential possibility of folded edge due to sectioning. Figure 2.14 is a map done on HA_OGM cross-section that is mostly on the lesion and slightly off-lesion in the bottom area. Different from Raman maps done on the whole gels, we got stronger and cleaner spectrum. From the maps we saw that on the lesion area, dense cells were seen, as well as some correlation between the DNA and the 51 lipids. Other interesting finding include the collagen dense area below the lesion and how HA nanoparticles were not found in the lesion. 52 53 54 Figure 2.14: Raman cross section map of HA_OGM gel on-lesion. Map taken on a HA_OGM cross-section with an area of interest mostly on the lesion and slightly off lesion in the bottom area of the map. Individual maps of collagen and hydroxyapatite showed that they are located in the off-lesion area, and the lesion are more NCP, DNA, and lipid dense. 55 56 Figure 2.15: Raman cross section map of HA_OGM gel off-lesion. Map taken at off-lesion area, were lots of HA nanoparticles were seen. 57 58 Figure 2.16: Raman cross section map of MCF_OGM gel on-lesion. Map taken on lesion area, where no MCF or any newly formed HA nanoparticles were seen. Similar to HA_OGM gels, lesions are cell dense and there is a collagen dense region on the bottom edge of the lesion. 59 60 Figure 2.17: Raman cross section map of MCF_OGM gel off-lesion. Map taken at off-lesion area. This region of interest was taken right below Figure 2.16. Interestingly, compare to off-lesion area of HA_OGM, there is more empty space where no matrix are found, showing a looser off-lesion area that is also seen in H&E staining. 61 Raman shift (1/cm) Assigned Raman mode Interpretation 962 Phosphate v1 (symmetry P-O stretch) Hydroxyapatite 1663 Amide l Collagen, Non-collagenous protein 1247 Amide lll Collagen 1320 Amide lll Non-collagenous protein 875 Pro/Hypro Collagen 2790 - 2860 Symmetric CH Lipid (shoulder) 780 - 800 O-P-O backbone DNA Table 2.2: Raman peak assignments. The spectral range of Raman peaks used in this study to calculate peak areas (and peak width). 62 REFERENCES [1] T. W. Gee, J. M. Richards, A. Mahmut, and J. T. Butcher, “Valve endothelial- interstitial interactions drive emergent complex calcific lesion formation in vitro,” Biomaterials, vol. 269, p. 120669, Feb. 2021, doi: 10.1016/J.BIOMATERIALS.2021.120669. [2] H. C. W. Skinner and H. Ehrlich, “Biomineralization,” in Treatise on Geochemistry: Second Edition, vol. 10, Elsevier Inc., 2013, pp. 105–162. doi: 10.1016/B978-0-08- 095975-7.00804-4. [3] F. A. Shah et al., “Micrometer-Sized Magnesium Whitlockite Crystals in Micropetrosis of Bisphosphonate-Exposed Human Alveolar Bone,” Nano Lett, vol. 17, no. 10, pp. 6210–6216, Oct. 2017, doi: 10.1021/acs.nanolett.7b02888. [4] P. R. Goody et al., “Aortic Valve Stenosis: From Basic Mechanisms to Novel Therapeutic Targets,” Apr. 01, 2020, Lippincott Williams and Wilkins. doi: 10.1161/ATVBAHA.119.313067. [5] S.T. Gould, E.E. Matherly, J.N. Smith, D.D. Heistad, K.S. Anseth, The role of valvular endothelial cell paracrine signaling and matrix elasticity on valvular interstitial cell activation, Biomaterials 35 (11) (2014) 3596–3606, https://doi. org/10.1016/j.biomaterials.2014.01.005. [6] S. Park, S. Choi, A. A. Shimpi, L. A. Estroff, C. Fischbach, and M. J. Paszek, “COLLAGEN MINERALIZATION DECREASES NK CELL-MEDIATED CYTOTOXICITY OF BREAST CANCER CELLS VIA INCREASED GLYCOCALYX THICKNESS”, doi: 10.1101/2024.01.20.576377. [7]. N. Anousakis-Vlachochristou, D. Athanasiadou, K. M. M. Carneiro, and K. Toutouzas, “Focusing on the Native Matrix Proteins in Calcific Aortic Valve Stenosis,” Aug. 01, 2023, Elsevier Inc. doi: 10.1016/j.jacbts.2023.01.009. [8] M. Gao et al., “Capacitive deionization toward fluoride elimination: Selective advantage, state of the art, and future perspectives,” May 18, 2024, Elsevier B.V. doi: 10.1016/j.desal.2024.117392. 63 CHAPTER 3 CONCLUSION AND FUTURE DIRECTIONS In conclusion, our study successfully incorporated MCFs into the spring gel model, demonstrating that both HA and MCF exhibit a similar degree of compaction and lesion formation. Notably, compaction in HA gels begins earlier than in MCF gels. Mineral deposits are not seen in the lesion areas but are significantly present just beneath the lesions in both gel types. The matrix within the lesion areas is primarily composed of non-collagenous proteins, while a collagen-rich band appears below these areas, as we see in the H&E staining. Lesions were observed as a 3D region that form on the surface of the hydrogels. Despite the similar degree of lesion formation, there seem to be a difference in the morphology from a top-down view of on the hydrogel surface, HA_OGM being more diffuse and MCF_OGM being more localized. To further investigate the morphology, we want to include the depth dimension of the lesions into consideration. We can look at the cryo-sections throughout the gels under bright field to get a more accurate idea of the morphology and spatial distribution of the lesions. Other potential experiments to do include a series of time-point experiment, capturing the different status of gels throughout the 10-day culturing. The focus on this experiment includes when the start of lesion formation (around day 5-6) is and whether there is any disruption in the endothelial layer. Other things to pay attention to is how the spatial distribution of minerals changes from day 1 to day 10, since we 64 assumed that minerals were uniformly dispersed on day 1, the fact that minerals were not found in lesions is interesting. The effect of endothelial cell layer in HA incorporated gels are well known through previous studies, including actively inducing pathological remodeling of VICs [6]. To ensure that VEC perform similar effect in MCF incorporated hydrogel, an experiment with a VIC-only hydrogel incorporating MCFs can help answering the question. Lastly, more histology characterization can be done to help us understand more cellular behaviors, Von Kossa staining can map out the mineral deposits, VE-cadherin or CD-31 can visualize the endothelial cell layer, RUNX2 for osteogenic differentiation, Sox9 for chondrogenic differentiation, aSMA and f-actin for the myofibroblast activity and cell movement. As an example, from whole-mount aSMA staining (Fig 3.1) we see in HA gels the actin below lesion has a star-like structure where the nuclei align with the actin. In MCF gels, actin does not show the star-like structure but more of a linear pattern. Parameters used in the imaging is yet to be optimize, but this gave an idea of what information histology characterization can provide us. 65 Figure 3.1: Immunofluorescence staining can help understand cell behavior. DAPI (blue) for nuclei and aSMA staining (red) for smooth muscle actin. 66 APPENDIX 1 67 Appendix 1: Raman maps on HA_OGM cross-section off-lesion (dehydrated). 68 APPENDIX 2 69 70 Appendix 2: Raman maps on MCF_OGM cross-section perinodular (dehydrated). 71 APPENDIX 3 72 Appendix 3: Raman maps on MCF_OGM cross-section on-lesion (hydrated).