INTRODUCTION

Section:

ChooseTop of pageABSTRACTINTRODUCTION <<RESULTSDISCUSSIONCONCLUSIONSMETHODSSUPPLEMENTARY MATERIALChronic liver disease is a major public health concern worldwide, and its prevalence is expected to increase in coming years with a concomitant increase in incidence of liver diseases such as non-alcoholic liver disease (NAFLD) and non-alcoholic steatohepatitis (NASH).11. S. K. Asrani, H. Devarbhavi, J. Eaton, and P. S. Kamath, J. Hepatol. 70(1), 151–171 (2019). https://doi.org/10.1016/j.jhep.2018.09.014 Fibrosis is one of the hallmarks of chronic liver disease and stems from an aberrant wound healing process that results in the accumulation of extracellular matrix (ECM) proteins that alter the mechanical and biochemical properties of the liver.2,32. S. L. Friedman, J. Hepatol. 38, 38–53 (2003). https://doi.org/10.1016/S0168-8278(02)00429-43. T. Kisseleva and D. Brenner, Nat. Rev. Gastroenterol. Hepatol. 18(3), 151–166 (2021). https://doi.org/10.1038/s41575-020-00372-7 While these changes in tissue properties negatively impact the behavior of many cell types in the liver, with hepatocyte dysfunction being one of the most canonical consequences, non-parenchymal cells of the liver are also prominently involved in the onset of fibrosis, and as such have been the focus of increasing interest in recent years. In particular, hepatic stellate cells (HSCs) have been implicated as significant contributors to the progression of liver fibrosis, and recent work has demonstrated that liver sinusoidal endothelial cells (LSECs) play a pivotal role in regulating HSC phenotype and coordinating signaling in the liver sinusoid microenvironment.4,54. I. Mederacke, C. C. Hsu, J. S. Troeger, P. Huebener, X. Mu, D. H. Dapito, J.-P. Pradere, and R. F. Schwabe, Nat. Commun. 4, 2823 (2013). https://doi.org/10.1038/ncomms38235. G. Xie, X. Wang, L. Wang, L. Wang, R. D. Atkinson, G. C. Kanel, W. A. Gaarde, and L. D. Deleve, Gastroenterology 142(4), 918–927 (2012). https://doi.org/10.1053/j.gastro.2011.12.017LSECs are specialized endothelial cells found in liver sinusoids that play a critical role in maintaining normal liver homeostasis.6,76. J. Poisson, S. Lemoinne, C. Boulanger, F. Durand, R. Moreau, D. Valla, and P. E. Rautou, J. Hepatol. 66(1), 212–227 (2017). https://doi.org/10.1016/j.jhep.2016.07.0097. K. K. Sørensen, J. Simon-Santamaria, R. S. McCuskey, and B. Smedsrød, in Comprehensive Physiology ( Wiley, 2015), pp. 1751–1774. In healthy livers, they form a semi-permeable barrier between the blood and liver parenchyma and utilize their fenestrae and abilities as scavengers to act as waste filter in the sinusoids.8,98. K. Elvevold, B. Smedsrød, and I. Martinez, Am. J. Physiol. 294(2), G391–G400 (2008). https://doi.org/10.1152/ajpgi.00167.20079. K. K. Sørensen, P. McCourt, T. Berg, C. Crossley, D. L. Couteur, K. Wake, and B. Smedsrød, Am. J. Physiol. 303(12), R1217–R1230 (2012). https://doi.org/10.1152/ajpregu.00686.2011 However, the onset of liver fibrosis results in the loss of characteristic LSEC fenestrae and behavior in a process known as capillarization. Also associated with capillarization is the formation of a basement membrane underneath the sinusoidal endothelium, and the dysregulation of LSEC signaling pathways, an observation that suggests LSEC capillarization, is a key mediator of fibrosis progression.1010. V. Natarajan, E. N. Harris, and S. Kidambi, Biomed. Res. Int. 2017, 4097205. https://doi.org/10.1155/2017/4097205In addition to their important role in maintaining balance in sinusoidal microenvironments, recent studies have provided evidence of considerable phenotypic heterogeneity in LSECs in both development and in zonation of mature sinusoids.11–1411. R. Marcu, Y. J. Choi, J. Xue, C. L. Fortin, Y. Wang, R. J. Nagao, J. Xu, J. W. MacDonald, T. K. Bammler, C. E. Murry, K. Muczynski, K. R. Stevens, J. Himmelfarb, S. M. Schwartz, and Y. Zheng, iScience 4, 20–35 (2018). https://doi.org/10.1016/j.isci.2018.05.00312. J. Lotto, S. Drissler, R. Cullum, W. Wei, M. Setty, E. M. Bell, S. C. Boutet, S. Nowotschin, Y.-Y. Kuo, V. Garg, D. Pe'er, D. M. Church, A.-K. Hadjantonakis, and P. A. Hoodless, Cell 183(3), 702–716 (2020). https://doi.org/10.1016/j.cell.2020.09.01213. N. Aizarani, A. Saviano, Sagar, L. Mailly, S. Durand, J. S. Herman, P. Pessaux, T. F. Baumert, and D. Grün, Nature 572(7768), 199–204 (2019). https://doi.org/10.1038/s41586-019-1373-214. K. B. Halpern, R. Shenhav, H. Massalha, B. Toth, A. Egozi, E. E. Massasa, C. Medgalia, E. David, A. Giladi, A. E. Moor, Z. Porat, I. Amit, and S. Itzkovitz, Nat. Biotechnol. 36(10), 962–970 (2018). https://doi.org/10.1038/nbt.4231 Moreover, it is increasingly being demonstrated that this heterogeneity is not limited to healthy LSECs, but, in fact, exists in disease conditions such as cirrhosis and fibrosis.15–1715. P. Ramachandran, R. Dobie, J. R. Wilson-Kanamori, E. F. Dora, B. E. P. Henderson, N. T. Luu, J. R. Portman, K. P. Matchett, M. Brice, J. A. Marwick, R. S. Taylor, M. Efremova, R. Vento-Tormo, N. O. Carragher, T. J. Kendall, J. A. Fallowfield, E. M. Harrison, D. J. Mole, S. J. Wigmore, P. N. Newsome, C. J. Weston, J. P. Iredale, F. Tacke, J. W. Pollard, C. P. Ponting, J. C. Marioni, S. A. Teichmann, and N. C. Henderson, Nature 575(7783), 512–518 (2019). https://doi.org/10.1038/s41586-019-1631-316. T. Su, Y. Yang, S. Lai, J. Jeong, Y. Jung, M. McConnell, T. Utsumi, and Y. Iwakiri, Cell. Mol. Gastroenterol. Hepatol. 11(4), 1139–1161 (2021). https://doi.org/10.1016/j.jcmgh.2020.12.00717. X. Xiong, H. Kuang, S. Ansari, T. Liu, J. Gong, S. Wang, X.-Y. Zhao, Y. Ji, C. Li, L. Guo, L. Zhou, Z. Chen, P. Leon-Mimila, M. T. Chung, K. Kurabayashi, J. Opp, F. Campos-Pérez, H. Villamil-Ramírez, S. Canizales-Quinteros, R. Lyons, C. N. Lumeng, B. Zhou, L. Qi, A. Huertas-Vazquez, A. J. Lusis, X. Z. S. Xu, S. Li, Y. Yu, J. Z. Li, and J. D. Lin, Mol. Cell 75(3), 644–660 (2019). https://doi.org/10.1016/j.molcel.2019.07.028 While this broad body of evidence has been accumulating, there has been much debate about the qualities and characteristics of LSECs and how best to develop a robust framework for studying LSEC phenotype.6,8,186. J. Poisson, S. Lemoinne, C. Boulanger, F. Durand, R. Moreau, D. Valla, and P. E. Rautou, J. Hepatol. 66(1), 212–227 (2017). https://doi.org/10.1016/j.jhep.2016.07.0098. K. Elvevold, B. Smedsrød, and I. Martinez, Am. J. Physiol. 294(2), G391–G400 (2008). https://doi.org/10.1152/ajpgi.00167.200718. L. D. DeLeve and A. C. Maretti-Mira, Semin. Liver Dis. 37(4), 377–387 (2017). https://doi.org/10.1055/s-0037-1617455 Notably, however, discussion of tissue microenvironmental factors and their influence over LSEC phenotype has largely been left out of this debate, despite their demonstrated impact on liver parenchyma and non-parenchyma.19–2119. S. R. Caliari, M. Perepelyuk, E. M. Soulas, G. Y. Lee, R. G. Wells, and J. A. Burdick, Integr. Biol. 8(6), 720–728 (2016). https://doi.org/10.1039/C6IB00027D20. A. L. Olsen, S. A. Bloomer, E. P. Chan, M. D. A. Gaça, P. C. Georges, B. Sackey, M. Uemura, P. A. Janmey, and R. G. Wells, Am. J. Physiol. 301(1), G110–G118 (2011). https://doi.org/10.1152/ajpgi.00412.201021. C. P. Monckton, A. Brougham-Cook, K. B. Kaylan, G. H. Underhill, and S. R. Khetani, Adv. Mater. Interfaces 8(22), 2101284 (2021). https://doi.org/10.1002/admi.202101284Indeed, given the mechanical and biochemical changes that occur in the liver, extensive work has been done to investigate the role these parameters play in influencing LSEC phenotype in both healthy and disease settings. As such, it is well established that ECM composition has a powerful influence over maintaining or altering LSEC phenotype.22–2622. L. Gifre-Renom, M. Daems, A. Luttun, and E. A. V. Jones, Int. J. Mol. Sci. 23(3), 1477 (2022). https://doi.org/10.3390/ijms2303147723. S. March, E. E. Hui, G. H. Underhill, S. Khetani, and S. N. Bhatia, Hepatology 50(3), 920–928 (2009). https://doi.org/10.1002/hep.2308524. Y. Zhang, Y. He, S. Bharadwaj, N. Hammam, K. Carnagey, R. Myers, A. Atala, and M. Van Dyke, Biomaterials 30(23), 4021–4028 (2009). https://doi.org/10.1016/j.biomaterials.2009.04.00525. P. S. Amenta and D. Harrison, Microsc. Res. Tech. 39(4), 372–386 (1997). https://doi.org/10.1002/(SICI)1097-0029(19971115)39:4<372::AID-JEMT7>3.0.CO;2-J26. R. F. McGuire, D. M. Bissell, J. Boyles, and F. J. Roll, Hepatology 15(6), 989–997 (1992). https://doi.org/10.1002/hep.1840150603 Recently, it has been shown that LSEC phenotype is also influenced by microenvironments of different elastic modulus, or stiffness, and that changes to the stiffness of the liver microenvironment can precede ECM protein deposition in fibrosis.27,2827. A. J. Ford, G. Jain, and P. Rajagopalan, Acta Biomater. 24, 220–227 (2015). https://doi.org/10.1016/j.actbio.2015.06.02828. P. C. Georges, J.-J. Hui, Z. Gombos, M. E. McCormick, A. Y. Wang, M. Uemura, R. Mick, P. A. Janmey, E. E. Furth, and R. G. Wells, Am. J. Physiol. 293(6), G1147–G1154 (2007). https://doi.org/10.1152/ajpgi.00032.2007 Additionally, it has been shown that artificially overexpressing canonical LSEC transcription factors in non-specialized endothelial cells were insufficient in fully restoring the hallmarks of LSEC phenotype, implicating extrinsic cues from the liver microenvironment in maintaining LSEC phenotype.2929. W. de Haan, C. Øie, M. Benkheil, W. Dheedene, S. Vinckier, G. Coppiello, X. L. Aranguren, M. Beerens, J. Jaekers, B. Topal, C. Verfaillie, B. Smedsrød, and A. Luttun, Am. J. Physiol. 318(4), G803–G815 (2020). https://doi.org/10.1152/ajpgi.00215.2019 Despite these advances, however, current in vitro models of LSEC behavior and phenotype fail to fully recapitulate the full range of physiological microenvironmental stimuli needed to further interrogate the role of the microenvironment on LSEC phenotype. Specifically, it remains to be determined how ECM composition, stiffness, and secreted soluble factors cooperatively impact LSEC phenotype and heterogeneity. Additionally, current models and experimental methods utilize techniques that have limited investigatory bandwidth that restrict the scope and scale of potential inquiries. Cellular microarray platforms, however, have been shown to address these shortcomings by affording investigators the ability to thoroughly examine the influences of different microenvironmental components simultaneously on liver cells in a high throughput setting.30–3530. C. J. Flaim, S. Chien, and S. N. Bhatia, Nat. Methods 2(2), 119–125 (2005). https://doi.org/10.1038/nmeth73631. K. B. Kaylan, A. P. Kourouklis, and G. H. Underhill, J. Vis. Exp. 121, e55362 (2017). https://doi.org/10.3791/5536232. K. B. Kaylan, I. C. Berg, M. J. Biehl, A. Brougham-Cook, I. Jain, S. M. Jamil, L. H. Sargeant, N. J. Cornell, L. T. Raetzman, and G. H. Underhill, eLife 7, e38536 (2018). https://doi.org/10.7554/eLife.3853633. K. B. Kaylan, V. Ermilova, R. C. Yada, and G. H. Underhill, Sci. Rep. 6, 23490 (2016). https://doi.org/10.1038/srep2349034. A. P. Kourouklis, K. B. Kaylan, and G. H. Underhill, Biomaterials 99, 82–94 (2016). https://doi.org/10.1016/j.biomaterials.2016.05.01635. A. Brougham-Cook, I. Jain, D. A. Kukla, F. Masood, H. Kimmel, H. Ryoo, S. R. Khetani, and G. H. Underhill, Acta Biomater. 138, 240–253 (2022). https://doi.org/10.1016/j.actbio.2021.11.015 As such, they are ideal for studying LSEC behavior and for performing nuanced investigations into the roles that different microenvironmental stimuli have in impacting LSEC phenotype.In this work, we utilized a cellular microarray platform to elucidate LSEC phenotype responses and heterogeneity as a function of microenvironmental composition. Specifically, we systematically examined the combinatorial effects of variations in ECM composition, substrate stiffness, and soluble factor presence on the phenotype of human LSECs by measuring the relative expression of LYVE-1, a scavenger receptor and established LSEC differentiation marker, VE-cadherin, a highly expressed sinusoidal cadherin which is critical to LSEC function, and CD-31, a common marker of endothelia and of LSEC capillarization.36–3836. C. M. Carreira, S. M. Nasser, E. di Tomaso, T. P. Padera, Y. Boucher, S. I. Tomarev, and R. K. Jain, Cancer Res. 61(22), 8079–8084 (2001).37. C. Géraud, K. Evdokimov, B. K. Straub, W. K. Peitsch, A. Demory, Y. Dörflinger, K. Schledzewski, A. Schmieder, P. Schemmer, H. G. Augustin, P. Schirmacher, and S. Goerdt, PLoS One 7(4), e34206-e34206 (2012). https://doi.org/10.1371/journal.pone.003420638. L. D. DeLeve, X. Wang, L. Hu, M. K. McCuskey, and R. S. McCuskey, Am. J. Physiol. 287(4), G757–G763 (2004). https://doi.org/10.1152/ajpgi.00017.2004 We identified unique trends in LSEC phenotype marker co-expression as well as novel spatial patterning of these markers. We characterized the heterogeneity and plasticity of LSEC phenotype by classifying the observed responses into four unique phenotype clusters. Furthermore, we demonstrated a unique correlation between LSEC phenotype, cell contractility, and Notch signaling as a function of microenvironmental context. Taken together, these studies highlight the important role of the microenvironment on influencing canonical LSEC phenotype markers. Moreover, this work identifies the unique phenotypes that LSECs exhibit due to ECM composition, stiffness, and soluble factor alterations in models of healthy and fibrotic tissue, and contributes crucial information toward constructing a more robust framework for how LSEC behaviors should be understood and studied going forward.

RESULTS

Section:

ChooseTop of pageABSTRACTINTRODUCTIONRESULTS <<DISCUSSIONCONCLUSIONSMETHODSSUPPLEMENTARY MATERIAL

LSEC phenotype markers and attachment profile respond dynamically to microenvironment context

Given the considerable evidence that LSEC behavior is strongly influenced by both ECM composition and stiffness independently,23,2723. S. March, E. E. Hui, G. H. Underhill, S. Khetani, and S. N. Bhatia, Hepatology 50(3), 920–928 (2009). https://doi.org/10.1002/hep.2308527. A. J. Ford, G. Jain, and P. Rajagopalan, Acta Biomater. 24, 220–227 (2015). https://doi.org/10.1016/j.actbio.2015.06.028 we sought to better understand how these two microenvironmental parameters, in various, physiologically relevant combinations, could further affect LSEC phenotype. Using a cellular microarray platform, we selected 28 ECM combinations previously implicated in influencing hepatic cell phenotype from literature,21,23,3521. C. P. Monckton, A. Brougham-Cook, K. B. Kaylan, G. H. Underhill, and S. R. Khetani, Adv. Mater. Interfaces 8(22), 2101284 (2021). https://doi.org/10.1002/admi.20210128423. S. March, E. E. Hui, G. H. Underhill, S. Khetani, and S. N. Bhatia, Hepatology 50(3), 920–928 (2009). https://doi.org/10.1002/hep.2308535. A. Brougham-Cook, I. Jain, D. A. Kukla, F. Masood, H. Kimmel, H. Ryoo, S. R. Khetani, and G. H. Underhill, Acta Biomater. 138, 240–253 (2022). https://doi.org/10.1016/j.actbio.2021.11.015 and three different hydrogel stiffnesses that mimic the mechanical properties of different stages of liver fibrosis: 1 kPa for healthy tissue, 6 kPa for early-stage fibrosis, and 25 kPa for late-stage fibrosis.41,4241. M. Fraquelli, C. Rigamonti, G. Casazza, D. Conte, M. F. Donato, G. Ronchi, and M. Colombo, Gut 56(7), 968 (2007). https://doi.org/10.1136/gut.2006.11130242. S. Mueller and L. Sandrin, Hepatic Med. 2, 49–67 (2010). https://doi.org/10.2147/hmer.s7394 In total, we tested 84 unique combinatorial microenvironments for their impact on LSEC phenotype. Initially, we observed that LSEC attachment profiles appear to be highly ECM dependent, with conditions containing collagen type IV (C4) exhibiting higher attachment and conditions containing laminin α1 (LN) exhibiting lower attachment. We also observed much lower cell attachment on 1 kPa substrates compared to 6 and 25 kPa [Figs. 1(a), 1(b), and supplementary material Fig. 1]. Notably, we observed that LSEC phenotypic marker LYVE-1 showed robust expression as early as 24 h into culture, and that mean LYVE-1 expression was significantly increased at the 72 h culture timepoint—particularly on 1 kPa substrates—with considerable dependence on the ECM composition [Figs. 1(c) and 1(d)].

VE-cadherin and CD-31 expression influenced by microenvironmental composition and display junctional localization

Additionally, we observed that LSEC phenotypic markers VE-cadherin and CD-31 exhibit co-expression as early as 24 h into culture. After 72 h, however, VE-cadherin and CD-31 display further changes in expression through marked junctional localization along cell periphery [Fig. 2(a)]. These signal localizations, and the degree to which they were expressed, were also observed to be dependent on microenvironmental context, with VE-cadherin and CD-31 exhibiting different response profiles. Specifically, after 72 h, LSECs demonstrated a higher junctional localization of VE-cadherin on 6 and 25 kPa vs 1 kPa substrates, while the CD-31 expression was observed to decrease with increasing stiffness [Figs. 2(b) and 2(c)].To better understand our observations of these phenotype markers, their expression profiles were more deeply interrogated. Further inspection of the cell microarray phenotypic data revealed multimodal signal distribution profiles for each marker and consequently distinct sub-populations of cells, suggesting that a thresholding and classification system of analysis would enable improved population identification [Fig. 2(d)]. Upon introducing a cutoff threshold of expression to classify marker expression, a percent positive metric was developed for the analysis of LYVE-1 and CD-31 expressions, with a percent junction localized marker used for VE-cadherin (abbreviated as “VE-cadherin edge”). Using these metrics, phenotype marker solo and co-expression were analyzed as a function of stiffness and ECM composition. We observed that soft (1 kPa) substrates promote elevated LYVE-1 expression, and that LYVE-1 expression decreased in response to increased stiffness (supplementary material Fig. 2). Regression analysis revealed that ECM condition C1/HA positively impacted LYVE-1 expression, while C1/LU and C4/LU were determined to negatively impact LYVE-1 expression (supplementary material Fig. 3). Analysis of VE-cadherin junction localization revealed that stiff (25 kPa) substrates promoted elevated VE-cadherin junction localization, and that this effect was reduced with decreasing stiffness (supplementary material Fig. 4). Regression analysis revealed that ECM condition C1/LU positively impacted VE-cadherin junction localization, while C1/LN, C4/C5, and C1/HA were determined to negatively impact VE-cadherin expression (supplementary material Fig. 5). Interestingly, CD-31 was observed to exhibit a similar expression profile to LYVE-1 (supplementary material Figs. 6 and 7), indicating patterns of co-expression.Intrigued by potential co-expression trends with these markers, the expression profiles of LYVE-1 positive cells vs VE-cadherin junction localized cells and LYVE-1 positive cells vs CD-31 positive cells were analyzed. Strikingly, LYVE-1 positive cells and VE-cadherin junction localized cells were observed to be inversely related, while LYVE-1 and CD-31 positive cells were observed to have a highly proportional relationship [Figs. 2(e) and 2(f)]. Moreover, on 1 kPa substrates, extreme forms of these trends were observed, with LYVE-1 positive cells establishing consistently high levels of expression independent of ECM composition. Given the unique expression trend observed between LYVE-1 and VE-cadherin, we sought to understand how ECM composition and stiffness influence populations of LYVE-1+ and VE-cadherin+ cells. Specifically, we observed that the proportions of LSECs that are positive for LYVE-1 only (+/−), VE-cadherin only (−/+), both (+/+), or neither (−/−) changed dramatically with different microenvironmental conditions. Specifically, ECM was observed to influence +/+ and +/− cell populations more on soft (1 kPa) vs stiffer (6 and 25 kPa) substrates, while influencing −/+ and +/− more on stiff (25 kPa) than softer (1 and 6 kPa) substrates [Fig. 2(e)]. Additionally, −/− and −/+ populations were observed to be highly stiffness dependent, with much higher levels of population proportionality observed on stiffer (6 and 25 kPa) vs soft (1 kPa) substrates. This microenvironmental impact is also observed when the relative expression of CD-31 is included in the heterogeneity analysis (supplementary material Fig. 8). Overall, these data highlight the influence of ECM composition and stiffness on the expression of these markers and illuminate novel LSEC phenotypic heterogeneity.

Microenvironmental stimuli elicit spatial patterning and heterogeneity in LSEC expression of LYVE-1, VE-cadherin, and CD-31

Given the unique responses these phenotypic markers exhibited as a function of their microenvironmental composition, we sought to more precisely understand these trends by down-selecting to a subset of 16 ECM conditions that showed the highest average cell attachment for follow up investigations. LSECs were then cultured on these 16 ECM microarrays, and the expression of the phenotypic markers (LYVE-1, VE-cadherin, CD-31) was quantitatively assessed following 72 h of microarray culture. Notably, LSECs at this time point displayed noticeable spatial patterning of LYVE-1 expression across the multicellular cultures that are confined to the arrayed ECM domains [Figs. 3(a) and 3(b)]. This patterning was observed to be highly stiffness dependent, with longer pattern radii that is indicative of a larger fraction of the cell monolayer expressing LYVE-1 primarily observed on 1 kPa substrates compared to 6 or 25 kPa [Fig. 3(c)]. Broadly, ECM composition was observed to establish a considerable dynamic range of pattern lengths on all stiffnesses. Linear regression modeling identified conditions containing C4 as the most impactful on altering median LYVE-1 expression radii length independent of stiffness, with C4/FN promoting longer median expression radii length, while C4/C5 promoting the opposite (supplementary material Fig. 9). Additionally, VE-cadherin and CD-31 were observed to exhibit some degree of spatial patterning as well as LYVE-1. This collective phenotypic patterning was observed to be highly dependent on both ECM composition and stiffness [Fig. 3(d)]. For example, LYVE-1 and CD-31 exhibited similar patterning for C1/LN and C1/LU, yet on C1/FN, the expression of these markers diverged, while VE-cadherin showed opposite patterning trends on C1/LU vs C4/FN.

Relative presence of ECM proteins modulates spatial patterning and phenotypic heterogeneity in LSECs

With ECM composition observed to exert such a prominent role in influencing LSEC marker expression and heterogeneity, we sought to determine the dose-responsive influence of ECM composition on LSEC phenotype. To do so, the relative concentrations of four representative ECM conditions (C1/FN, C4/LU, C1/LU, and C4/C5) were studied at the following ratios for a total of 16 conditions across the same three stiffnesses (ECM A/ECM B): 200:50, 150:100, 100:150, and 50:200 [Fig. 4(a)]. The effect of the relative composition of ECM was apparent in the resultant LSEC adhesion profiles, particularly on 1 kPa substrates [Fig. 4(b)]. Additionally, varying the ratio of components for C4/C5 conditions significantly attenuated cell attachment independent of stiffness, in contrast with C4/LU conditions which exhibited little difference in attachment regardless of relative concentration or stiffness (supplementary material Fig. 10). Most notably, the relative concentration of ECM protein was observed to substantially impact the expression and spatial patterning profile of LYVE-1. For example, while C1/LU displayed minor changes in patterning profile across different concentrations, C1/FN showed a considerable increase in pattern profile, independent of stiffness, as the concentration of C1 increased and FN decreased [Fig. 4(c)]. This was also reflected in LYVE-1 percent positive cells, with ECM condition and relative composition observed to attenuate LYVE-1 expression differently. Interestingly, while varying concentrations of C4/LU and C4/C5 elicited consistent patterning profiles independent of concentration, cells on C4/C5 generally displayed longer pattern radii, while those on C4/LU displayed shorter pattern radii (supplementary material Fig. 11). Notably, relative ECM composition was observed to have little to no impact on VE-cadherin junction localization, while the CD-31 expression was also observed to be sensitive to relative ECM composition, similar to LYVE-1 (supplementary material Fig. 12). Moreover, population heterogeneity analysis of changes in LYVE-1, VE-cadherin, and CD-31 expressions as a function of relative ECM composition revealed that ECM concentration influences the heterogeneity of expression of these three LSEC phenotype markers most dramatically on stiff (25 kPa) substrates [Fig. 4(d)]. Specifically, conditions containing LU were observed to pointedly influence LSEC heterogeneity by either positively or negatively impacting triple negative, triple positive, and double positive (−/−/−, +/+/+, +/−/+, and +/+/−) populations depending on their relative concentration and ECM combination (order of phenotype markers: LYVE-1/VE-cadherin/CD-31). Notably, +/+/− populations are least prevalent on soft (1 kPa) substrates and highest on intermediate (6 kPa) substrates, with both maintaining similar levels of −/−/− (supplementary material Fig. 12).

Soluble factor presence influences LSEC phenotype in combination with ECM composition and stiffness

While ECM composition is an important feature of a cellular microenvironment, it is only part of the milieu of biochemical signals that interact with cells. Soluble factors such as cytokines and growth factors are also present and active in cell microenvironments, especially in the liver sinusoid. As such, we sought to understand how this facet of the cellular microenvironment impacted LSECs in combination with ECM composition and stiffness. We began by investigating the impact of soluble factors as a microenvironmental stimuli by culturing LSECs on cellular microarrays of 16 representative ECM conditions on 1, 6, and 25 kPa substrates using a base media formulation (− control, EGM, Lonza) and a cytokine supplemented media (+ control, EGM2, Lonza). Notably, we observed that the percentages of LYVE-1+ cells are significantly higher when cultured in supplemented media compared to the negative control, independent of stiffness [Fig. 5(a)]. Encouraged by this observation, we then tested eight single and two-factor combinations of four of the prominent growth factor components from the supplement media [vascular endothelial growth factor (VEGF), fibroblast growth factor (FGF), insulin-like growth factor (IGF), and Heparin] on these 16 ECM microarrays at the same three stiffness, combining for a total of 384 unique microenvironmental conditions.We observed that soluble factor presence caused no significant reductions in LSEC attachment, and that different combinations of soluble factors could promote differential levels of attachment depending on their components, with IGF + Heparin promoting higher levels of attachment yet FGF + Heparin promoting lower levels (supplementary material Fig. 13). Regression analysis revealed that ECM composition, stiffness, and soluble factor treatment significantly impacted LYVE-1 expression, with C1/HA and FGF promoting the largest increases in percent LYVE-1 positive cells, and 25 and 6 kPa substrates promoting the largest decreases [Fig. 5(b) and supplementary material Fig. 14]. LYVE-1 spatial patterning was also observed to be impacted by soluble factor treatment (supplementary material Fig. 15). Additionally, VE-cadherin junction localization, CD-31 expression, and LYVE-1 patterning were also impacted by the combination of ECM, stiffness, and soluble factors (supplementary material Figs. 16 and 17). We also observed that while maintaining their inverse relationship, the relative proportions of LYVE-1 and VE-cadherin shifted sizably upon treatment with soluble factors, highlighting both the robust relationship between the two markers and the considerable phenotypic plasticity of LSECs (supplementary material Fig. 18). More broadly, Principal Component Analysis (PCA) combined with hierarchical clustering analysis revealed that LSEC phenotype as a function of ECM composition, stiffness, and soluble factors is remarkably heterogenous and can be characterized into four distinct populations [Figs. 5(c) and 5(d)]. Notably, all three microenvironmental stimuli were determined to be significant determinants of cluster assignment (Wilks test, p-value 

Small-molecule inhibition of Notch and ROCK signaling pathways attenuate LSEC phenotype and spatial patterning profile

While the soluble factor experiments revealed an important role for growth factors in influencing LSEC phenotype, they typically function as agonists, affecting cell behavior by stimulating cell signaling pathways. Considering the unique trends in the data collected thus far, we sought to better understand mechanistically how and why LSECs respond to their microenvironments in such dramatic fashion. Given the demonstrated impact of these soluble factor agonists, we then considered whether pathway antagonism through small molecule inhibition could shed more light on the mechanisms regulating LSEC phenotype. Changes in tissue stiffness are hallmarks of liver fibrosis, and it is well documented that cellular contractility changes concomitantly in various hepatic cell types with increased tissue stiffness through mechanostransduction pathways.32,34,35,4332. K. B. Kaylan, I. C. Berg, M. J. Biehl, A. Brougham-Cook, I. Jain, S. M. Jamil, L. H. Sargeant, N. J. Cornell, L. T. Raetzman, and G. H. Underhill, eLife 7, e38536 (2018). https://doi.org/10.7554/eLife.3853634. A. P. Kourouklis, K. B. Kaylan, and G. H. Underhill, Biomaterials 99, 82–94 (2016). https://doi.org/10.1016/j.biomaterials.2016.05.01635. A. Brougham-Cook, I. Jain, D. A. Kukla, F. Masood, H. Kimmel, H. Ryoo, S. R. Khetani, and G. H. Underhill, Acta Biomater. 138, 240–253 (2022). https://doi.org/10.1016/j.actbio.2021.11.01543. R. C. Andresen Eguiluz, K. B. Kaylan, G. H. Underhill, and D. E. Leckband, Biomaterials 140, 45–57 (2017). https://doi.org/10.1016/j.biomaterials.2017.06.010 Additionally, Notch signaling has been implicated in LSEC capillarization in vivo, and activated Notch signaling has been observed in patients with cirrhosis.44,4544. H. Ma, X. Liu, M. Zhang, and J. Niu, Mol. Biol. Rep. 48(3), 2803–2815 (2021). https://doi.org/10.1007/s11033-021-06269-145. J.-L. Duan, B. Ruan, X.-C. Yan, L. Liang, P. Song, Z.-Y. Yang, Y. Liu, K.-F. Dou, H. Han, and L. Wang, Hepatology 68(2), 677–690 (2018). https://doi.org/10.1002/hep.29834 Considering the observed impact of stiffness on LSEC phenotype as well as the robust cell–cell junction formation observed from VE-cadherin and CD-31 data, we hypothesized that mechanostransduction and Notch signaling pathways were involved in regulating LSEC phenotype. To interrogate the influence of these pathways, we treated cells with gamma secretase inhibitor (GSI), which prevents the proteolytic cleavage of the Notch receptor, and thus the release of the Notch intracellular domain, and Rho-associated kinase (ROCK) inhibitor (Y-27632), which blocks myosin II activity, to better understand the role of Notch and mechanostransduction pathways in determining LSEC phenotype.We observed that LSEC LYVE-1 expression dramatically increased compared to control when treated with GSI, especially on stiffer (6 and 25 kPa) substrates, whereas Y-27632 treatment showed little impact [Fig. 6(a)]. Notably, LSECs treated with both GSI and Y-27632 exhibited lower VE-cadherin expression compared to control, with the largest decreases occurring on stiffer (6 and 25 kPa) substrates, while little to no effect was observed on CD-31 expression [Fig. 6(b) and supplementary material Fig. 19]. We also observed that GSI and Y-27632 treatments had a pronounced impact on LYVE-1 spatial patterning. Specifically, compared to control the treatment groups disrupted LYVE-1 spatial patterning on 6 and 25 kPa, while cells cultured on 1 kPa were less affected [Fig. 6(c)]. When the expression of all three phenotype markers was considered as a function of their ECM composition, stiffness, and small molecule inhibition, we observed that GSI and Y-27632 treatments induce a remarkable shift in phenotype proportion, particularly on stiff (25 kPa) substrates [Fig. 6(d) and supplementary material Fig. 19]. Overall, these data illuminate a previously undescribed LSEC phenotypic plasticity and underscore the role of Notch and mechanostransduction pathways in regulating LSEC phenotype.

DISCUSSION

Section:

ChooseTop of pageABSTRACTINTRODUCTIONRESULTSDISCUSSION <<CONCLUSIONSMETHODSSUPPLEMENTARY MATERIALIn this work, we demonstrate the impact that different microenvironmental stimuli have on LSEC phenotype and heterogeneity, particularly the effects these stimuli have in combination with one another (Fig. 7). For example, the influence of stiffness on LSEC LYVE-1 expression is modulated by ECM composition, with ECM conditions including laminin exhibiting strong stiffness dependence, while conditions containing proteins such as C4/C5 exhibiting less dependence on substrate stiffness. We also found that changes in the relative concentrations of ECM proteins within multicomponent mixtures regulated the spatial patterning of LYVE-1 expression within the circular multicellular domains of the cell microarray. Furthermore, substrate stiffness exhibited a cooperative influence on the responses of LSECs to exogenous stimuli, and the phenotypic responses to Notch signaling and contractility inhibitors were also modulated by the stiffness of the microenvironment. It is well documented that LSECs change their phenotype following acute liver injury in the early stages of fibrosis development through a process called capillarization. This change is characterized by a loss of fenestrae, reduced expression of canonical markers, and developing a basal membrane and has far-reaching impacts by influencing both neighboring cells of the sinusoid and immune cells more generally.5,18,38,46,475. G. Xie, X. Wang, L. Wang, L. Wang, R. D. Atkinson, G. C. Kanel, W. A. Gaarde, and L. D. Deleve, Gastroenterology 142(4), 918–927 (2012). https://doi.org/10.1053/j.gastro.2011.12.01718. L. D. DeLeve and A. C. Maretti-Mira, Semin. Liver Dis. 37(4), 377–387 (2017). https://doi.org/10.1055/s-0037-161745538. L. D. DeLeve, X. Wang, L. Hu, M. K. McCuskey, and R. S. McCuskey, Am. J. Physiol. 287(4), G757–G763 (2004). https://doi.org/10.1152/ajpgi.00017.200446. L. D. DeLeve, Hepatology 61(5), 1740–1746 (2015). https://doi.org/10.1002/hep.2737647. L. D. DeLeve, X. Wang, and Y. Guo, Hepatology 48(3), 920–930 (2008). https://doi.org/10.1002/hep.22351 Notably, similar changes in LSEC phenotype have been demonstrated in other liver pathologies, underscoring LSECs' pivotal role in maintaining liver homeostasis.48–5048. M. Pasarín, V. L. Mura, J. Gracia-Sancho, H. García-Calderó, A. Rodríguez-Vilarrupla, J. C. García-Pagán, J. Bosch, and J. G. Abraldes, PLoS One 7(4), e32785 (2012). https://doi.org/10.1371/journal.pone.003278549. S. Francque, W. Laleman, L. Verbeke, C. Van Steenkiste, C. Casteleyn, W. Kwanten, C. Van Dyck, M. D'Hondt, A. Ramon, W. Vermeulen, B. De Winter, E. Van Marck, V. Van Marck, P. Pelckmans, and P. Michielsen, Lab. Invest. 92(10), 1428–1439 (2012). https://doi.org/10.1038/labinvest.2012.10350. M. Miyao, H. Kotani, T. Ishida, C. Kawai, S. Manabe, H. Abiru, and K. Tamaki, Lab. Invest. 95(10), 1130–1144 (2015). https://doi.org/10.1038/labinvest.2015.95 While the local and global impact of LSEC dysregulation in the liver has been previously established, many questions about the specific roles that microenvironmental stimuli, like tissue stiffness, ECM composition, and soluble factor presence, play in this dysregulation have gone unanswered. Our data provide insight into these inquires and demonstrate the crucial role that microenvironmental signaling has on shaping LSEC phenotype.One of the hallmarks of liver fibrosis progression is the altered mechanical properties of the tissue, namely, the increasing stiffness of the tissue microenvironment.4242. S. Mueller and L. Sandrin, Hepatic Med. 2, 49–67 (2010). https://doi.org/10.2147/hmer.s7394 It has been previously shown that substrate stiffness has a direct impact on endothelial cell morphology and phenotype. For example, numerous studies have reported that endothelial cells exhibit differential cell spreading, focal adhesion expression, contractility, and cell organization in a stiffness-dependent manner.43,51–5543. R. C. Andresen Eguiluz, K. B. Kaylan, G. H. Underhill, and D. E. Leckband, Biomaterials 140, 45–57 (2017). https://doi.org/10.1016/j.biomaterials.2017.06.01051. R. Krishnan, D. D. Klumpers, C. Y. Park, K. Rajendran, X. Trepat, J. van Bezu, V. W. M. van Hinsbergh, C. V. Carman, J. D. Brain, J. J. Fredberg, J. P. Butler, and G. P. van Nieuw Amerongen, Am. J. Physiol. 300(1), C146–C154 (2011). https://doi.org/10.1152/ajpcell.00195.201052. F. Bordeleau, B. N. Mason, E. M. Lollis, M. Mazzola, M. R. Zanotelli, S. Somasegar, J. P. Califano, C. Montague, D. J. LaValley, J. Huynh, N. Mencia-Trinchant, Y. L. Negrón Abril, D. C. Hassane, L. J. Bonassar, J. T. Butcher, R. S. Weiss, and C. A. Reinhart-King, Proc. Natl. Acad. Sci. U. S. A. 114(3), 492–497 (2017). https://doi.org/10.1073/pnas.161385511453. E. E. Bastounis, Y.-T. Yeh, and J. A. Theriot, Sci. Rep. 9(1), 18209–18209 (2019). https://doi.org/10.1038/s41598-019-54336-254. M. C. Lampi, M. Guvendiren, J. A. Burdick, and C. A. Reinhart-King, ACS Biomater. Sci. Eng. 3(11), 3007–3016 (2017). https://doi.org/10.1021/acsbiomaterials.6b0063355. F. J. Byfield, R. K. Reen, T.-P. Shentu, I. Levitan, and K. J. Gooch, J. Biomech. 42(8), 1114–1119 (2009). https://doi.org/10.1016/j.jbiomech.2009.02.012 Indeed, our data illustrate the complex role stiffness plays in regulating LSEC phenotype, with LSECs cultured on healthy tissue mimics (1 kPa) for 3 days consistently exhibiting elevated expression of both canonical differentiation (LYVE-1) and c