COMMENTARY The Journal of Clinical Investigation Building discontinuous sinusoidal vessels

Courtney T. Griffin and Siqi Gao Cardiovascular Biology Research Program, Oklahoma Medical Research Foundation, Oklahoma City, Oklahoma, USA.

Liver sinusoidal development Blood vessels have a unified mission to circulate blood throughout the and embryonic hematopoiesis Sinusoids are fenestrated discontinuous body; however, they have additional diverse and specialized roles in various found in the liver, spleen, and organs. For example, in the liver, discontinuous sinusoids, which are bone marrow. These three organs all have fenestrated capillaries with intercellular gaps and a fragmented basement hematopoietic roles at various stages of membrane, facilitate delivery of macromolecules to highly metabolic development; therefore, sinusoids are . During embryonic development, discontinuous sinusoids well adapted to facilitate the movement of also allow circulating hematopoietic progenitor and stem cells to populate hematopoietic progenitor or stem cells and the liver and promote blood cell differentiation. In this issue of the JCI, mature blood cells between the circulatory Géraud et al. describe an essential role for the transcription factor GATA4 in system and these tissues. During embryon­ promoting the development of discontinuous sinusoids. In the absence of ic development, erythro-myeloid progeni­ liver sinusoidal GATA4, mouse embryos developed hepatic capillaries with tors derived from the yolk sac or definitive upregulated endothelial cell junction proteins and a continuous basement hematopoietic stem cells generated in the membrane. These features prevented hematopoietic progenitor cells from yolk sac, placenta, umbilical cord, and/or transmigrating into the developing liver, and Gata4-mutant embryos died aorta-gonad-mesonephros (AGM) region from subsequent liver hypoplasia and anemia. This study highlights the migrate through the circulatory system surprising and extensive transcriptional control GATA4 exercises over and populate the developing fetal liver (5, specialized liver vascular development and function. 6). Within the liver, hematopoietic stem cells proliferate and egress to colonize the spleen, thymus, and bone marrow, which becomes the primary site of hematopoiesis after birth (7). The fetal liver is also a key diversity uous capillaries allow for large molecules site of definitive erythropoiesis, and mature Capillaries are microvessels found through­ and even transmigratory cells to be readily red blood cells begin exiting the liver and out the body that serve as intermediate transported between the blood stream and entering the circulatory system by E12.5 to conduits for blood to travel between small tissue. For example, the tight junctions of oxygenate the rapidly growing embryo (8). arteries (arterioles) and (venules). continuous brain capillaries prevent circu­ Monocytes in the fetal liver also give rise to Capillaries can be broadly classified as lating toxins from entering the underlying a subset of tissue macrophages that require continuous or discontinuous based on cerebrospinal fluid and brain tissue, while diaphragmed fenestrations in sinusoidal morphological characteristics. Continuous discontinuous liver capillaries facilitate in order to exit the fetal liver capillaries are defined by dense intercel­ transport of large molecules in the blood and populate tissues throughout the body lular junctional proteins, which contribute to underlying hepatocytes for metabolism (9). Interestingly, sinusoidal morpholo­ to tight or adherens junctions, and by the (3). Fenestrations, or endothelial cell pores gy changes during rodent development: underlying basement membrane, which is that frequently possess a thin diaphragm by late gestation, junctional proteins are produced in part by adjacent pericytes (1). across their opening, can be found in both downregulated and fenestrations lose dia­ In contrast, discontinuous capillaries have continuous and discontinuous capillaries phragms and increase in size and number intercellular gaps due to a paucity of junc­ and serve as an additional selective con­ (10). The implications of these alterations tional proteins and lack an organized base­ duit for exchange of materials across the in sinusoidal morphology on fetal liver ment membrane (2). The distinct properties endothelial cell barrier (2). Importantly, function are poorly understood. of continuous and discontinuous capillar­ fenestrations and other features of contin­ In this issue, Géraud et al. provide ies have functional consequences for the uous and discontinuous capillaries can be important insight into the transcriptional surrounding tissue: continuous capillar­ influenced by developmental stage, phys­ regulation of discontinuous liver sinusoid ies are more refractory to the exchange of iological signals, or toxic insults, under­ morphology (11). This group had previous­ macromolecules between the blood stream scoring the plasticity of capillary morphol­ ly determined that the transcription factor and surrounding tissue, while discontin­ ogy and function (4). GATA4 is enriched in rat liver sinusoidal endothelial cells (LSECs) compared with Related Article: p. 1099 rat lung microvascular endothelial cells (LMECs) (12). Géraud and colleagues have Conflict of interest: The authors have declared that no conflict of interest exists. now employed transgenic mouse lines to Reference information: J Clin Invest. 2017;127(3):790–792. https://doi.org/10.1172/JCI92823. define the function of GATA4 in murine

790 jci.org Volume 127 Number 3 March 2017 The Journal of Clinical Investigation COMMENTARY

Figure 1. Gata4 deletion alters liver sinusoid morphology. Genetic deletion of the tran- scription factor Gata4 from liver sinusoidal endothelial cells (LSECs) causes upregulation of endothelial cell junction proteins and robust deposition of basement membrane proteins that prevent circulating hematopoietic progen- itor or stem cells from colonizing the fetal liver (11). The consequences of this transition from discontinuous sinusoidal to continuous capillary morphology are liver hypoplasia, anemia, and lethality of Gata4 mutant embryos (11).

LSECs (11). Specifically, they developed Géraud et al. also exploited prima­ defects, as these populations expanded and a line of mice harboring Cre driven by the ry rat LSECs and LMECs to establish a differentiated in transplantation experi­ stabilin 2 (Stab2) promoter (Stab2-Cre), transcriptional profile for discontinuous ments and in vitro (11). Instead, Géraud which is expressed in mature LSECs (11), and continuous capillaries, respectively and colleagues propose that the continu­ and exploited an existing Lyve1-Cre line (11). As rodent LSECs rapidly de-differ­ ous capillaries that aberrantly displace dis­ (13), which is active in embryonic LSECs entiate in culture, precluding knockdown continuous sinusoids in Gata4 mutant liv­ (14). Deletion of a floxed Gata4 allele in or overexpression studies, Géraud and ers prevent progenitor and hematopoietic either line resulted in mutant embryos with colleagues manipulated GATA4 expres­ stem cells from colonizing the mutant liver, hypoplastic , anemia, and prenatal sion in human umbilical endothelial where they would subsequently undergo lethality (11). Unlike the liver, other organs cells (HUVECs), which are transcription­ expansion and differentiation (Figure 1). were not obviously affected by Gata4 dele­ ally similar to continuous LMECs (11). tion in either the Stab2-Cre or Lyve1-Cre Upon GATA4 overexpression, HUVECs Outstanding questions line (11). GATA4-deficient livers exhibited acquired a transcriptional profile more Because the capillarization process (i.e., microvessels with features of continuous similar to discontinuous LSECs, indicating transformation from discontinuous sinu­ capillaries rather than those of discontin­ that GATA4 can promote a discontinuous soids to continuous capillaries) described uous sinusoids seen in WT livers (11). In LSEC gene program across species and in in these Gata4 mutant embryos encom­ particular, upregulation of endothelial cell different endothelial cell types (11). passes both upregulation of endothelial junction proteins (CD31 and VE-cadherin) Finally, Géraud et al. reported that the cell junction proteins and deposition of on LSECs was observed by immunostain­ emergence of continuous capillary features basement membrane, one might question ing, and a more robust basement mem­ in Gata4 mutant livers preceded liver hypo­ whether one or both of these phenotypes is brane was detected by electron microscopy plasia and anemia (11). Erythro-myeloid essential for blocking fetal liver coloniza­ and by immunostaining for extracellular progenitor cells that typically colonize tion. Murine embryos in which endothelial matrix components (11). No effect of Gata4 the fetal liver from the circulation around cell junctions are artificially strengthened deletion on LSEC fenestrae was reported, E10.5 were markedly reduced in livers but by genetically replacing VE-cadherin with but it would be interesting to know whether elevated in the blood of Gata4 mutants at a VE-cadherin–α-catenin fusion construct this transcription factor affects the number E11.25 (11). Hematopoietic stem cell popu­ have phenotypes that are strikingly similar or size of fenestrae or the presence of a dia­ lations followed a similar aberrant localiza­ to those of the Gata4 mutants described phragm on the pores. As LSECs are highly tion pattern in the liver and blood of E13.25 by Géraud et al., including hypoplastic endocytic (15), it would also be informative Gata4 mutants (11). These progenitor and livers, anemia, and failed transmigration to know whether GATA4 impacts endocyt­ hematopoietic stem cells from mutant of hematopoietic stem and progenitor ic vesicle density on embryonic LSECs. embryos appeared to have no intrinsic cells from the circulation into the liver

jci.org Volume 127 Number 3 March 2017 791 COMMENTARY The Journal of Clinical Investigation

(16). Therefore, strong and/or upregulat­ initial development of those sinusoids is an 1. Pavelka M, Roth J. Functional Ultrastructure: ed endothelial cell junctions are clearly a open question. If GATA4 does not contrib­ Atlas of Tissue Biology and Pathology. Vienna, Austria: Springer Vienna; 2010:254–255. sufficient impediment for fetal liver colo­ ute to discontinuous sinusoidal develop­ 2. Sarin H. Physiologic upper limits of pore size nization. Interestingly, embryos in which ment in these hematopoietic organs, iden­ of different blood capillary types and another Gata4 is deleted from embryonic hepatic tifying other transcription factors that act in perspective on the dual pore theory of microvas­ mesenchymal cells develop hyperactive lieu of GATA4 will clarify the organ-specific cular permeability. J Angiogenes Res. 2010;2:14. stellate cells and produce excessive extra­ derivation of sinusoidal vessels. 3. Cleaver O, Melton DA. Endothelial signaling during development. Nat Med. 2003; cellular matrix around the hepatic vascula­ Capillarization of hepatic sinusoids is 9(6):661–668. ture before dying at E13.5 with hypoplastic associated with advanced liver fibrosis (18); 4. Aird WC. Phenotypic heterogeneity of the endo­ livers and anemia (17). Although the Del­ therefore, the question of whether GATA4 thelium: I. Structure, function, and mechanisms. gado et al. study did not assess fetal liver transcriptionally maintains discontinuous Circ Res. 2007;100(2):158–173. colonization by hematopoietic stem and hepatic sinusoid morphology has import­ 5. Gomez Perdiguero E, et al. Tissue-resident macrophages originate from yolk-sac-de­ progenitor cells, the reported phenotypes ant clinical implications. Human patients rived erythro-myeloid progenitors. Nature. suggest that excessive sinusoidal base­ with cirrhosis and advanced liver fibrosis 2015;518(7540):547–551. ment membrane may also be sufficient to have diminished hepatic GATA4 protein 6. Medvinsky A, Rybtsov S, Taoudi S. Embry­ block this process. Nevertheless, it is not levels compared with patients with healthy onic origin of the adult hematopoietic sys­ clear why hematopoietic stem and progen­ livers or early-stage disease (17), although tem: advances and questions. Development. 2011;138(6):1017–1031. itor cells preferentially colonize the bone it in unclear whether GATA4 levels are 7. Mikkola HK, Orkin SH. The journey of devel­ marrow and spleen later in development, altered specifically in the LSECs of these oping hematopoietic stem cells. Development. despite an increase in open pore fenestrae, patients. Would upregulation of GATA4 in 2006;133(19):3733–3744. a downregulation of endothelial cell junc­ LSECs of cirrhotic livers reverse the capil­ 8. Palis J. Primitive and definitive erythropoiesis in mammals. Front Physiol. 2014;5:3. tional proteins, and a persistently discon­ larization process and associated fibrosis? 9. Rantakari P, et al. Fetal liver endothelium tinuous basement membrane that would Genetic rodent models could be designed regulates the seeding of tissue-resident macro­ presumably favor continued transmigra­ to address this question. In vivo vascular phages. Nature. 2016;538(7625):392–396. tion through the hepatic sinusoids (10). GATA4 overexpression would also vali­ 10. Bankston PW, Pino RM. The development of the Another interesting question is whether date and provide functional context for the sinusoids of fetal rat liver: morphology of endo­ thelial cells, Kupffer cells, and the transmural GATA4 is important for the maintenance finding by Géraud et al. that cultured endo­ migration of blood cells into the sinusoids. Am J of discontinuous liver sinusoids once they thelial cell lines with continuous character­ Anat. 1980;159(1):1–15. are established, as Géraud and colleagues istics (HUVECs and bEnd3 cells) assume 11. Géraud C, et al. GATA4-dependent organ-spe­ demonstrated that GATA4 is expressed in a transcriptional profile more consistent cific endothelial differentiation controls liver adult murine, rat, and human LSECs (11). with discontinuous LSECs when GATA4 is development and embryonic hematopoiesis. J Clin Invest. 2017;127(3):1099–1114. An inducible Stab2-Cre or Lyve1-Cre line overexpressed in vitro (11). Altogether, the 12. Géraud C, et al. Liver sinusoidal endothelium: a would help address the temporal influence new evidence provided by Géraud and col­ microenvironment-dependent differentiation of GATA4 over discontinuous hepatic sinu­ leagues about the importance of GATA4 in program in rat including the novel junctional soidal characteristics. Such lines could also establishing discontinuous hepatic sinusoi­ protein liver endothelial differentiation-associ­ clarify whether GATA4 and discontinuous dal morphology during mid-gestation rais­ ated protein-1. Hepatology. 2010;52(1):313–326. 13. Pham TH, et al. Lymphatic endothelial cell sinusoids are required for differentiated es interesting follow-up questions about sphingosine kinase activity is required for lym­ blood cells to egress from the fetal liver and the extent to which GATA4 exercises tem­ phocyte egress and lymphatic patterning. J Exp reenter the circulation. Macrophage precur­ poral and spatial influence over discontin­ Med. 2010;207(1):17–27. sors require diaphragms on LSEC fenestrae uous capillary morphology throughout the 14. Crosswhite PL, et al. CHD4-regulated plasmin for this egression process (9), but a robust lifetime of an organism. activation impacts lymphovenous hemostasis and hepatic vascular integrity. J Clin Invest. basement membrane may serve as a critical 2016;126(6):2254–2266. impediment, even in the presence of dia­ Acknowledgments 15. Braet F, Wisse E. Structural and functional phragmed fenestrae. An inducible Stab2-Cre This work was supported by NIH grants aspects of liver sinusoidal endothelial cell fenes­ line might also help clarify whether GATA4 R01HL134778 and R01HL111178 and trae: a review. Comp Hepatol. 2002;1(1):1. 16. Dartsch N, Schulte D, Hägerling R, Kiefer F, promotes discontinuous sinusoidal develop­ by American Heart Association grant Vestweber D. Fusing VE-cadherin to α-catenin ment in other organs, such as the bone mar­ 15GRNT25090015 (to CTG). impairs fetal liver hematopoiesis and lymph row and spleen, which undergo colonization but not blood vessel formation. Mol Cell Biol. by hematopoietic stem cells later in develop­ Address correspondence to: Courtney 2014;34(9):1634–1648. ment (7). Géraud and colleagues report that Griffin, 825 NE 13th Street, MS 45, Okla­ 17. Delgado I, et al. GATA4 loss in the septum trans­ versum mesenchyme promotes liver fibrosis in GATA4 is not expressed on STAB2-positive homa City, Oklahoma 73104, USA. Phone: mice. Hepatology. 2014;59(6):2358–2370. endothelial cells in adult bone marrow and 405.271.7073; E-mail: courtney-griffin@ 18. Hernandez-Gea V, Friedman SL. Pathogenesis of spleen (11), but whether it contributes to the omrf.org. liver fibrosis. Annu Rev Pathol. 2011;6:425–456.

792 jci.org Volume 127 Number 3 March 2017