Liver sinusoidal endothelium: A microenvironment-dependent differentiation program in rat including the novel junctional protein liver endothelial differentiation-associated protein-1

Authors

  • Cyrill Géraud,

    Corresponding author
    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    • Klinik für Dermatologie, Venerologie und Allergologie, Universitätsmedizin Mannheim, D-68135 Mannheim, Germany
    Search for more papers by this author
    • These authors contributed equally to this work.

    • fax: 49 (0) 621 383 3815

  • Kai Schledzewski,

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Alexandra Demory,

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author
  • Diana Klein,

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author
  • Miriam Kaus,

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author
  • Francis Peyre,

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author
  • Carsten Sticht,

    1. Center for Medical Research, Medical Faculty Mannheim, University of Heidelberg, Mannheim, Germany
    Search for more papers by this author
  • Konstantin Evdokimov,

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author
  • Shun Lu,

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author
  • Astrid Schmieder,

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author
  • Sergij Goerdt

    1. Department of Dermatology, Venereology and Allergology, University Medical Center and Medical Faculty Mannheim, University of Heidelberg, and Center of Excellence in Dermatology, Mannheim, Germany
    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

Abstract

Liver sinusoidal endothelium (LSEC) is a prime example of organ-specific microvascular differentiation and functions. Disease-associated capillarization of LSEC in vivo and dedifferentiation of LSEC in vitro indicate the importance of the hepatic microenvironment. To identify the LSEC-specific molecular differentiation program in the rat we used a two-sided gene expression profiling approach comparing LSEC freshly isolated ex vivo with both lung microvascular endothelial cells (LMEC) and with LSEC cultured for 42 hours. The LSEC signature consisted of 48 genes both down-regulated in LMEC and in LSEC upon culture (fold change >7 in at least one comparison); quantitative reverse-transcription polymerase chain reaction confirmation of these genes included numerous family members and signaling pathway-associated molecules. The LSEC differentiation program comprised distinct sets of growth (Wnt2, Fzd4, 5, 9, Wls, vascular endothelial growth factors [VEGFR] 1, 2, 3, Nrp2) and transcription factors (Gata4, Lmo3, Tcfec, Maf) as well as endocytosis-related (Stabilin-1/2, Lyve1, and Ehd3) and cytoskeleton-associated molecules (Rnd3/RhoE). Specific gene induction in cultured LSEC versus freshly isolated LSEC as well as LMEC (Esm-1, Aatf) and up-regulation of gene expression to LMEC levels (CXCR4, Apelin) confirmed true transdifferentiation of LSEC in vitro. In addition, our analysis identified a novel 26-kDa single-pass transmembrane protein, liver endothelial differentiation-associated protein (Leda)-1, that was selectively expressed in all liver endothelial cells and preferentially localized to the abluminal cell surface. Upon forced overexpression in MDCK cells, Leda-1 was sorted basolaterally to E-cadherin-positive adherens junctions, suggesting functional involvement in cell adhesion and polarity. Conclusion: Comparative microvascular analysis in rat identified a hepatic microenvironment-dependent LSEC-specific differentiation program including the novel junctional molecule Leda-1. HEPATOLOGY 2010

Endothelial cells (ECs) display marked heterogeneity in different organs and in different segments of the vascular tree. Liver sinusoidal endothelial cells (LSECs) are a prime example of uniquely differentiated microvascular EC that exert highly specialized functions as professional endocytes1 and participate in induction of hepatic immune tolerance.2 For endocytosis, LSEC express a broad range of different scavenger receptors including Stabilin-1 and Stabilin-2, two members of a novel scavenger receptor gene family selectively expressed in sinusoidal ECs that have been identified and thoroughly characterized by us.3 Stabilin-2 is the major hyaluronan scavenger receptor of the liver, whereas Stabilin-1 mediates endocytosis of acLDL (low density lipoprotein) and SPARC. Stabilin-1 and -2 use the constitutive clathrin-mediated endocytic pathway in LSEC (4); in addition, Stabilin-1 fulfills a second role as an intracellular cargo carrier.5 In LSEC, clathrin-coated vesicles and Stabilin-1/2+ early endosomes travel along microtubules organized in a very special net-like way linking LSEC morphology to endocytic function.6

Despite their highly specialized microvascular differentiation, LSECs retain a remarkable phenotypic and functional plasticity. In liver cirrhosis, for example, endothelial plasticity results in morphological transdifferentiation of LSEC, collectively termed sinusoidal “capillarization.” Unfortunately, not much is known about the mechanisms that control regular LSEC differentiation and LSEC transdifferentiation during pathogenic processes. LSEC-hepatocyte interactions have been recognized to be of special importance due to unidirectional cytokine crosstalk between LSEC and hepatocytes mediated by hepatocyte growth factor (HGF) and vice versa between hepatocytes and LSEC by way of EG-VEGF.7 Recently, we have been able to show that LSEC-derived Wnt2 acts as a cell type-specific autocrine growth factor in LSEC cross-stimulating the VEGF pathway.8 A major setback in deciphering LSEC-specific differentiation is the fact that LSECs are not amenable to long-term cultures in vitro. LSECs rapidly lose their characteristic morphology as well as some of their specialized functions in culture. Hitherto, attempts to improve LSEC culture conditions have had limited success,9, 10 indicating that a better understanding of the molecular programs underlying LSEC-specific differentiation in vivo and dedifferentiation in vitro is urgently needed.

Dedifferentiation of EC in culture is not unique to LSEC. Both blood vascular as well as lymphatic microvascular EC undergo marked transdifferentiation over time upon culture.11 High endothelial venule endothelial cells (HEVEC) from tonsil are a striking example of highly specialized ECs that lose their specific gene signature as soon as 48 hours after isolation.12 Thus, even short-term cultures of primary EC do not adequately mimic the respective differentiated EC phenotypes in situ. These results suggest that organ-specific EC differentiation and function is maintained by the respective tissue microenvironments.

For a comprehensive analysis of the molecular programs mediating LSEC-specific differentiation, we chose a similar, two-sided, comparative gene expression profiling approach. Selection of the genes that were both overexpressed in LSEC in comparison to LMEC and down-regulated in LSEC upon short-term cultivation resulted in identification of an LSEC-specific gene signature including genes in several functional categories. Among these molecules, liver endothelial differentiation-associated protein (Leda)-1 was identified as a novel homolog of adherens junction-associated protein-1 (Ajap-1/Shrew-1) involved in cell adhesion and polarity.13, 14 This LSEC-specific gene signature may comprehensively determine the special functional program of liver sinusoidal endothelium.

Abbreviations:

cDNA, complementary DNA; EC, endothelial cell; ECM, extracellular matrix; FC, fold change; HEVEC, high endothelial venule endothelial cell; HGF, hepatocyte growth factor; LMEC, lung microvascular endothelial cell; LSEC, liver sinusoidal endothelial cell; LDL, low density lipoprotein; qRT-PCR, quantitative reverse-transcription polymerase chain reaction; VEGF, vascular endothelial growth factor.

Materials and Methods

First antibodies: mouse anti-RhoE (Sigma), rabbit anti-Ehd3 (kind gift from Markus Plomann), mouse antirat SE-1 (ARP), mouse antirat-CD31 (BD Biosciences) custom-made polyclonal rabbit anti-Stabilin2, custom-made monoclonal mouse anti-Stabilin1 and anti-Stabilin2, rabbit anti-Lyve1 (Reliatech), and custom-made anti-Leda-1 polyclonal guinea pig antibody were generated by immunizing with a c-terminal synthetic peptide of Leda-1. Second antibodies: horseradish peroxidase (HRP), Cy3, Alexa488-conjugated donkey or goat antirabbit, antimouse, antiguinea pig (Dianova and BD Biosciences).

Animals.

Sprague-Dawley rats were purchased from Janvier (Le Genest-St-Isle, France) and received humane care according to the guidelines of the National Institutes of Health (NIH). Experiments were approved by the animal ethics committee in Baden-Wuerttemberg (Regierungspraesidium Karlsruhe AZ:35-9185.82A-35/07).

LSEC and LMEC Isolation and Purification.

Cells were isolated and purified as previously described; purity was confirmed by FACS with directly-labeled antibodies against CD31, Stabilin-2, and CD11b and resulted in >95% Stabilin-2+, CD31+, CD11b− cells.8 LSEC were plated on collagen-coated dishes and cultured using a mixture of EBM-2 (CC-1356, Cambrex) and Willams'E (Invitrogen) growth medium, containing EGM Single-Quots (CC-4133), 0.2% bovine serum albumin (BSA), 10 ng/mL HGF, and 1% ITS media supplement (I3146, Sigma) at 37°C in a humidified incubator (5% CO2).

Immunohistology.

Cryostat sections were air-dried and acetone-fixed. Specimens were blocked with 5% BSA in phosphate-buffered saline (PBS) and incubated with first antibody, followed by appropriate HRP-labeled secondary antibody. Pictures were taken with a DCRE microscope, camera, and software system (Leica).

Immunofluorescence.

Acetone-fixed cryostat sections and paraformaldehyde-fixed cells on coverslips were blocked with 3% BSA, incubated with first antibodies, followed by appropriate secondary antibodies. Specimens were analyzed by confocal microscopy (Leica). For actin staining Alexa488-conjugated phalloidin (Molecular Probes) was used.

Western Blot Analysis.

Whole cell lysates, protein determination, and sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) were carried out as described.8 After blotting on PVDF membranes (BioRad) membranes were incubated with first antibody overnight, followed by the appropriate HRP-conjugated secondary antibody. Signal intensity of SuperSignal West Pico ECL Substrate (Pierce) was detected with ECL-Hyperfilms (Amersham).

Primer Design.

Primers for quantitative reverse-transcription polymerase chain reaction (qRT-PCR) were designed using available GenBank sequences with Primer-Blast (http://www.ncbi.nlm.nih.gov/tools/primer-blast/). All primers for intron containing genes were intron-spanning. Specificity was validated with a Blast Search for crossreactions and by RT-PCR and agarose gel-electrophoresis. A list of all oligonucleotide sequences used for qRT-PCR is found in Supporting Information Table 1.

RT-PCR and Quantitative Real-Time PCR.

RNA extraction was carried out using the RNeasy Kit (Qiagen). After DNase I (Fermentas) pretreatment complementary DNA (cDNA) synthesis was performed with RevertAid H Minus M-MuLV Reverse Transcriptase (Fermentas). For quantitative real-time PCR 1 μL of cDNA was amplified with Sybr-Green Master-Mix (Applied Biosystems) with an MX3005P signal detection system (Stratagene). Gene expression was related to actin expression with the Stratagene Data-Analysis Software. All samples were amplified in triplicate.

Statistical Analysis.

SigmaPlot 11 software was used to perform either t test or one-way analysis of variance (ANOVA). A P-value <0.05 was considered statistically significant. Error bars show the standard error of the mean (SEM) of each experiment.

Rho-Kinase Activity Assay.

CycLex Rho-kinase Assay Kit (CY-1160, Cyclex) was used according to the manufacturer's instructions. Equal amounts of fresh protein lysates and dilutions of 1:10 and 1:100 were used and analyzed with a spectrophotometric plate reader.

cDNA Microarray Analysis.

Biotinylated antisense cRNA was prepared according to the Affymetrix standard labeling protocol. Gene expression profiling was performed using arrays of rat genome 230 2.0-type (Affymetrix, High Wycombe, UK). A custom CDF v. 11 with Entrez-based gene definitions was used to annotate the arrays. Raw fluorescence intensity values were normalized applying quantile normalization. Differential gene expression was analyzed based on loglinear mixed model ANOVA15 using a commercial software package SAS JMP7 Genomics, v. 3.2 from SAS (Cary, NC). A false-positive rate of a = 0.05 with Holm correction was taken as the level of significance. The raw and normalized data were deposited in the Gene Expression Omnibus database (http://www.ncbi.nlm.nih.gov/geo/; Accession No. GSE-20375).

Generation of Leda-1-Transfected MDCK Cells.

A recombinant Leda-1 cDNA was amplified by PCR (primers: Hs_Leda1_cDNA_FW_Spe1 5′tgacactagtaaaggctgaaaatctggg3′ and Hs_Leda1_cDNA_BW_Not1 5′gtcagcggccgcggcctcctgcttggctg3′ from human Leda-1 cDNA IRATp970C0957D (IMAGE-ID 5730150), purified on agarose gel, and subcloned after digestion with SpeI and NotI restriction enzymes into the expression vector pEF6/V5-His Topo (Invitrogen) according to standard molecular biology protocols. MDCK cells (ATCC) were transfected with human Leda-1 vector DNA using Fugene HD (Roche) transfection reagent. Several stably transfected clones were selected by way of limited dilution. As negative control empty vector-transfected MDCK clones were transfected and selected under parallel culture conditions.

Results

LSEC Isolated Ex Vivo Rapidly Lose Their Characteristic Morphology, as Well as Expression of Marker Genes upon Short-Term Culture In Vitro.

Freshly isolated LSEC adhered to collagen-coated cell culture plates for 2 hours displayed a round-to-oval cell shape with abundant cytoplasm and the typical sieve plates/fenestrations, and they showed strong expression of LSEC marker-genes such as Stabilin-1/2, Lyve1, and CD32b/SE-1 antigen. Upon cultivation LSEC underwent remarkable morphological changes after as few as 24-48 hours. The characteristic sieve plates/fenestrations were lost, the nuclear-cytoplasmic ratio was elevated, and they seemingly adopted the cobblestone-like appearance of continuous EC in vitro (Fig. 1A). After a culture period of 42-72 hours Stabilin-1/2, Lyve-1, and CD32b showed rapid down-regulation of messenger RNA (mRNA) and protein, whereas expression of the pan-endothelial marker CD31 remained unchanged (Fig. 1B,C). mRNA expression of Wnt-2 previously identified by us as an autocrine growth factor specific for LSEC cross-stimulating the VEGF pathway was also found to rapidly decline during culture (Fig. 1D). Thus, isolated LSECs undergo marked transdifferentiation in culture, indicating that normal LSEC differentiation in vivo depends on the control of the hepatic microenvironment that is not adequately reproduced in vitro.

Figure 1.

LSEC transdifferentiation in vitro. (A) Immunofluorescent staining of isolated rat LSEC cultured for 2 and 48 hours with anti-Lyve1 (green) and anti-Stabilin-2 antibodies (red). Toto3 was used for nuclear staining. Images were acquired using confocal microscopy. Expression of Stabilin-1, Stabilin-2 (B), CD32b, CD31, Lyve1 (C), and wnt2 (D) was analyzed by quantitative RT-PCR. Expression levels are shown as relative expression in relation to the reference gene (actin). Data are expressed as mean ± SEM of at least three independent experiments performed in triplicate. Bars indicate a P-value <0.05. Western blot analysis of Stabilin-2 (B) and CD32b (C) expression was performed with protein lysates of LSEC2h and LSEC72h using a polyclonal Stabilin-2 antibody and the SE-1/CD32b antibody. Equal loading was controlled by stripping and reprobing with an anti-β-actin antibody. The results are representative of at least three independent experiments.

Identification of the Microenvironment-Dependent Molecular Program Underlying LSEC Differentiation.

To identify the molecular program underlying microenvironment-dependent LSEC-differentiation, we chose a double-sided comparative approach. Total RNA was isolated from three different groups of samples: (1) freshly isolated LSEC (LSEC0h); (2) LSEC kept in culture for 42 hours (LSEC42h); and (3) freshly isolated CD31-sorted lung microvascular endothelial cells (LMEC0h). After cDNA synthesis these three groups (4-5 independent samples for each group) were subjected to Affymetrix DNA Microarray Analysis.

By comparing LSEC0h with LMEC0h, 364 genes (LSECspecific) were found to be overexpressed in LSEC0h with a fold-change (FC) >2 (Supporting Information Table 2). Among these genes were several well-known LSEC marker genes such as Stabilin-2, CD32b, and Lyve1. Vice versa, von Willebrand factor, a gene well known to be strongly expressed in LMEC but only weakly in LSEC, was strongly overexpressed in LMEC0h (FC = 51). Thus, the purity of the cell samples and the quality of the hybridization was validated by these marker genes. By comparing LSEC0h with LSEC42h, 465 genes (LSECdown) were found to be down-regulated at the 42-hour timepoint with FC >2 (Supporting Information Table 3). By analyzing LSECspecific and LSECdown for common genes (n = 106) and by including only genes with a FC >7 in at least one of the comparisons, 48 genes were identified that are LSEC-specific and depend on the hepatic microenvironment. The resultant genes (LSECspecific+down) (Fig. 2A) were grouped according to their gene ontology terms into the following clusters: (1) cytokine and growth factor signaling; (2) transcriptional regulators; (3) scavenger receptors, endocytosis, and transport; (4) cytoskeletal organization; (5) extracellular matrix and cell-matrix adhesion proteins; (6) immune system processes; and (7) others (Fig. 2B; Table 1). qRT-PCR was performed on all 48 genes using the three above-mentioned groups as well as an additional group of freshly isolated LSEC that had been adhered to a collagen-coated cell culture dish for 2 hours (LSEC2h) to differentiate adhesion-independent transdifferentiation in vitro from adhesion-associated changes. Differential expression of 39 genes was confirmed by qRT-PCR analysis in all groups with high statistical significance, whereas seven genes were only partially confirmed (Supporting Information Table 4).

Figure 2.

Comparative transcriptome analysis of LSEC0h, 42h and LMEC. (A) Venn diagram of a double-sided comparison of microarray expression data. In all, 106 intersecting genes were found that are overexpressed with a FC >2 in LSEC0h in comparison to LMEC as well as in LSEC0h in comparison to LSEC42h. Forty-eight of the 106 genes reach an FC >7 in at least one of the two comparisons. (B) By gene ontology analysis these 48 genes were grouped into seven functional clusters. (C) qRT-PCR analysis of genes up-regulated in LSEC42h versus LSEC0h including Esm1, Aatf, Cxcr4, Apeln in LSEC0h, 2h, 42h and LMEC. Expression levels are shown as relative expression in relation to the reference gene (actin). Data are expressed as mean ± SEM of at least three independent experiments performed in triplicate. Bars indicate a P-value <0.05.

Table 1. Microenvironment-Dependent LSEC-Specific Genes
Gene NameUniGeneHsIDFC LSEC > LMECFC LSEC0h > LSEC42h
  1. Genes preferentially expressed in LSEC were identified by comparing normalized, mean hybridization signals of LSEC0h and LMEC samples. Genes down-regulated in cultured LSEC were identified by comparing LSEC0h and LSEC42h. The fold change (FC) was determined by calculating the ratio of mean hybridization signals. In Table 1 only genes are included that reach an FC>2 in both comparisons and an FC>7 in at least one of the two comparisons. For each group (LSEC0h, LSEC42h, and LMEC) mean hybridization signals of 4-5 independent samples were used.

Cytokine and growth factor signaling
 Ccl24Rn.13882215.516.8
 Bmp2Rn.9093114.911.8
 Vegfr3/Flt4Rn.810438.54.4
 Wnt2Rn.2037647.87.2
 Similar to CCRL1Rn.1695986.730.4
 Il1aRn.123003.77.7
Transcriptional regulators
 Gata4Rn.2625111.54.5
 MafRn.1072610.87.3
 TfecRn.96598.92.3
Scavenger receptors, endocytosis and transport
 Stabilin-2Rn.16293193.613.6
 Ehd3Rn.777134.86.8
 CD209bRn.13884624.512.7
 Clec4fRn.988620.918.8
 Abca8aRn.2278919.611.7
 Slc39a8Rn.2832112.43.8
 Ap1b1Rn.3738312.32.7
 Pla2r1Rn.1169411.92.1
 Slc40a1Rn.1532411.34.6
 CtskRn.559810.22.6
 Cd5lRn.438048.98.4
 ArsbRn.940048.25.5
 Slc43a2Rn.2049887.86.0
Cytoskeletal organization
 Rnd3/RhoERn.251537.23.7
ECM and cell-matrix adhesion proteins
 Emilin1Rn.9934611.19.7
 TtrRn.140410.89.1
 Col13a1Rn.175668.55.8
 AlbuminRn.20296876.551.8
 AhsgRn.320838.48.3
 HpseRn.63928.08.2
 FcnaRn.200517.68.5
Immune system processes
 Mox2rRn.12443913.14.5
 Ifitm1Rn.2208711.32.9
 FcgrtRn.158458.03.8
 Dpp4Rn.913642.612.0
 C1qgRn.23937.811.9
 C1qaRn.1056478.713.2
 C1qbRn.67026.07.8
Others
 Dnase1l3Rn.29996110.03.8
 Similar to keratin 6LRn.20170231.06.9
 Cyp26A1Rn.8107215.520.3
 Ssg1Rn.219313.014.9
 Leda-1/RGD1304952Rn.134118.56.1
 Bace2Rn.3858110.16.1
 Itih3Rn.895767.87.6
 CideaRn.81717.75.0
 Gstt1Rn.111227.64.9
 Rarres1Rn.167375.28.5
 Dhrs3Rn.120112.611.8

In addition, 449 genes (LSECup) were also found to be up-regulated in culture with FC >2 when comparing LSEC0h with LSEC42h (not shown), indicating that LSEC dedifferentiation in culture is not just a process of cellular deterioration. Using qRT-PCR, several of these genes were shown to reach expression levels also seen in LMEC0h (Cxcr4, Apelin), whereas others were specifically induced in cultured LSEC (Esm1, Aatf) (Fig. 2C).

Impaired Wnt-2 and VEGF Signaling Parallels LSEC Transdifferentiation.

As Wnt-2 and Flt-4/Vegfr3 were confirmed to be overexpressed in LSEC compared to LMEC and to be considerably down-regulated in LSEC in vitro by qRT-PCR (Fig. 3A), the Wnt and VEGF signaling pathways were scrutinized for broader impairment in LSEC transdifferentiation in vitro. qRT-PCR analysis showed that VegfA was highly overexpressed in LMEC versus LSEC and did not decline in culture. Vegf B, C, D did not show any significant differences. Vegfr2 shown by us to be induced in LSEC by Wnt-2, but not Vegfr1, was overexpressed in LSEC versus LMEC; both receptors were down-regulated in LSEC in vitro. VEGF coreceptor Nrp2 (venous/lymphatic EC), but not Nrp1 (arterial EC), paralleled expression of Vegfr2/3. Tmem24 involved in Flt-4/Vegfr3 signaling was also found to be down-regulated upon culture (Fig. 3B). Previously, we have shown that Wnt2 is strongly overexpressed in LSEC and that it acts as an autocrine growth factor by way of Wnt receptors Fzd4, Fzd5, and Fzd9. Differential expression of Wnt signaling components in LSEC and LMEC was also previously demonstrated by us including Wntless homolog (Wls), an intracellular wnt transporter, and wnt-inhibitory factor (Wif-1), whereas expression of most Wnt ligands except for Wnt-7b and Wnt-10b was relatively weak in LSEC as well as in LMEC.8 Upon transdifferentiation in culture (42 hours), wnt2 receptors Fzd4, Fzd5, and Fzd9 and Wnt-10b and Wls were considerably down-regulated, whereas Wnt-7b, Wif1, and Devl did not change significantly (Fig. 3C).

Figure 3.

Differential expression of Vegf and wnt signaling components in LSEC0h, 2h, 42h and LMEC. qRT-PCR analysis of Vegfr3/Flt4, wnt2 (A), Vegf signaling related genes (B), and wnt signaling related genes (C) in LSEC0h, 2h, 42h and LMEC. Expression levels are shown as relative expression in relation to the reference gene (actin). Data are expressed as mean ± SEM of at least three independent experiments performed in triplicate. Bars indicate a P-value <0.05.

Tfec, Gata-4, Maf, and Lmo3 Constitute a Set of LSEC-Specific Transcriptional Regulators.

Organ-specific blood vascular EC differentiation is thought to be mediated by specific combinations of otherwise non-EC-specific transcription factors. In this respect, identification and confirmation by qRT-PCR of the three transcription factors Tfec, Gata-4, and Maf in the LSECspecific+down gene signature was quite intriguing (Fig. 4A). In contrast to Gata-4, expression of Gata family members Gata-2, -3, and -5 was much lower in LSEC versus LMEC and did not change significantly upon culture (Fig. 4B,C). Similarly, basic helix-loop-helix (bHLH) transcription factors Mitf, Tfe3, and Tfeb showed a much lower expression level in LSEC than Tfec and grossly remained unaltered throughout cultivation (not shown).

Figure 4.

Differential expression of transcriptional regulators and endocytosis related gene Ehd3 in LSEC0h, 2h, 42h and LMEC. qRT-PCR analysis of Gata4, Tfec, Maf (A), and Gata and Lmo gene families (B/C) in LSEC0h, 2h, 42h and LMEC. (D) Immunofluorescent staining of isolated rat LSEC with anti-Stabilin1/2 (green) and anti-Ehd-3 antibodies (red). Toto3 was used for nuclear staining. Images were acquired using confocal microscopy. (E) qRT-PCR analysis of Ehd3 and Ehd family members in LSEC0h, 2h, 42h and LMEC. Expression levels are shown as relative expression in relation to the reference gene (actin). Data are expressed as mean ± SEM of at least three independent experiments performed in triplicate. Bars indicate a P-value <0.05. (F) Western Blot analysis of Ehd3 expression was performed with protein lysates of LMEC, LSEC2h, and LSEC48h using a Ehd3-antibody. Equal loading was controlled by stripping and reprobing with an anti-β-actin antibody. The results are representative of at least three independent experiments.

Gata-family members and bHLH proteins are known to interact with LIM domain only (Lmo) proteins. Lmo3 was found among the LSECspecific genes with an FC of 6.4; by qRT-PCR Lmo3 mRNA levels were significantly higher in LSEC than in LMEC and declined significantly upon culture (Fig. 4B). Lmo1 was absent and Lmo2 and 4 were not differentially expressed in LSEC versus LMEC; Lmo4 mRNA levels, however, slightly decreased in LSEC after 42 hours in culture (Fig. 4C).

LSEC-Specific Endocytosis: A Possible Role for Ehd3 in Regulating Stabilin-1-Positive Vesicle Transport.

As part of the gene cluster scavenger receptors, endocytosis and transport, Ehd3 was identified as a possible new mediator of vesicular transport in LSEC. Its selective expression in LSEC and loss of expression upon culture was confirmed by qRT-PCR and with western blotting (Fig. 4E,F). In contrast to Ehd3, Ehd family members Ehd1, 2, and 4 were highly overexpressed in LMEC versus LSEC and their expression remained stable in LSEC in culture. Immunofluorescent double labeling of Stabilin-1 and Stabilin-2 with Ehd3 in freshly isolated LSEC showed partial coexpression of Ehd3 with Stabilin-1, but not with Stabilin-2 (Fig. 4D), suggesting a possible role for Ehd3 in regulating intracellular trafficking of Stabilin-1-positive endosomes.

Rnd3/RhoE Has a Distinct Pattern of Expression in LSEC and Its Down-Regulation Parallels Stress Fiber Formation in Cultured LSEC.

Rnd3/RhoE is a nonfunctional small GTPase that inhibits RhoA-mediated stress fiber formation by way of competitive inhibition of RhoA phosphorylation by ROCK-I; Rnd3, however, does not bind to the highly homologous kinase ROCK-II. Interestingly, Rock inhibition is able to dissolve actin stress fibers that develop in the center of LSEC in vitro and to keep fenestrations dilated.16 Rnd3 was confirmed by qRT-PCR and western blotting to be expressed in LSEC0h/2h, but not in LMEC and LSEC48h (Fig. 5A). Rnd3 homologs, Rnd1 and Rnd2, were expressed at much lower levels than Rnd3 in LSEC0h showing a transient, adhesion-dependent increase in LSEC2h (Fig. 5A). RhoA was found to be equally expressed in LSEC and LMEC and throughout culture (Fig. 5A). Rock-I and ROCK-II as well as ROCK-I downstream targets LimK1 and 2 and Cofilin1, but not Cofilin2, were expressed in LSEC0h/2h/42h and in LMEC without showing significant differences between samples (Fig. 5B).

Figure 5.

Rnd3 down-regulation in cultured LSEC is accompanied by an increase in stress fibers. qRT-PCR analysis of Rnd1, 2, 3, RhoA (A) Rnd3 target genes (B) in LSEC0h, 2h, 42h, and LMEC. (C) Immunofluorescent staining of isolated rat LSEC with phalloidin (actin) (green) and anti-Rnd3 antibody (red). Toto3 was used for nuclear staining. Images were acquired using confocal microscopy. (D) Rho-kinase-activity of fresh lysates of LSEC2h and LSEC48h was analyzed with a semiquantitative immunoassay measuring Mypt1 threonine-696 phosphorylation. Photospectrometric absorption at 450 nm reflects the relative amount of Rho-kinase activity. Equal amounts of lysates were used undiluted (1) and in dilutions of 1:10 (0.1) and 1:100 (0.01). The results are representative of two independent experiments. (E) Immunofluorescent staining of isolated rat LSEC after 2 and 72 hours in culture with phalloidin (actin) (green). Images were acquired using confocal microscopy. The results are representative of at least three independent experiments. Expression levels are shown as relative expression in relation to the reference gene (actin). Data are expressed as mean ± SEM of at least three independent experiments performed in triplicate. Bars indicate a P-value <0.05.

On immunofluorescence, LSEC displayed a homogeneous vesicular pattern of Rnd3 expression throughout the cytoplasm (Fig. 5C); upon cultivation, Rnd3 distribution within the cell became uneven mostly concentrating in the perinuclear region (not shown). Although cumulative ROCK-I and -II activity as measured by Mypt1 phosphorylation was decreased in LSEC in culture (Fig. 5D), reduced Rnd3 expression and subcellular relocalization were accompanied by increased stress fiber formation indicative of enhanced activity of the RhoA/ROCK-I axis (Fig. 5E).

Leda-1 Is a Novel Ajap-1 Homolog Selectively Expressed in All Liver ECs.

Within the LSECspecific+down gene signature, a novel uncharacterized gene (GenBank Access. No. 00101459.1) was identified (Table 1) whose 2,456 basepair (bp) cDNA codes for a putative type-1 transmembrane protein of 272 amino acids (aa) with a predicted molecular weight of 29 kDa, including a 27 aa n-terminal signal peptide. Selective expression of this novel protein in freshly isolated LSEC versus LMEC and down-regulation of its expression in LSEC upon culture was confirmed by qRT-PCR (Fig. 6A), and hence the novel protein was named liver endothelial differentiation-associated protein (Leda)-1. The putative Leda-1 protein sequences in mouse and human display striking identity. In the rat Leda-1 protein sequence, 6 aa at the c-terminus are missing by comparison with human and murine sequences. This difference may be explained by a supernumerary weak intron in Access. No. 00101459.1 (Supporting Information Fig. 1). After addition of these 6 aa, rat and mouse/human Leda1 show 99%/97% identity (Supporting Information Fig. 2A). A Blast search using rat Leda-1 as a query in the nonredundant National Center for Biotechnology Information (NCBI) database identified Leda-1 genes in the vertebratum phylum and revealed a significant homology to adherens junction-associated protein-1 (Ajap-1/Shrew1), a protein that targets adherens junctions in polarized epithelial MDCK cells and influences cell invasion (Supporting Information Fig. 2B).

Figure 6.

Identification of Leda-1, a new Ajap1 homolog that localizes to the abluminal side of sinusoidal and nonsinusoidal EC of the liver. (A) qRT-PCR analysis of Leda-1 in LSEC0h, 2h, 42h and LMEC. Expression levels are shown as relative expression in relation to the reference gene (actin). Data are expressed as mean ± SEM of at least three independent experiments performed in triplicate. Bars indicate a P-value <0.05. Western blot analysis of Leda-1 expression was performed with protein lysates of LMEC, LSEC2h, and LSEC48h using a custom-made polyclonal guinea pig anti-Leda1 antibody. Equal loading was controlled by stripping and reprobing with an anti-β-actin antibody. The results are representative of at least three independent experiments. ECL film exposure was varied as indicated. (B) Immunohistochemistry of rat liver and lung with a polyclonal Leda-1 antibody, guinea pig preimmune serum as negative control and a CD31-antibody. Bar = 50 μm. The results are representative of at least three independent experiments. (C) Immunofluorescent staining of rat liver sections with anti-Stabilin-2 (green) and anti-Leda-1 antibodies (red). Images were acquired using confocal microscopy. Arrows depict a large vessel that is Stabilin-2-negative but Leda-1-positive and arrowheads depict areas of the sinusoidal vessel wall that are Stabilin-2-negative but Leda-1-positive. (D) High-magnification immunofluorescent staining of rat liver sections with anti-Stabilin2 (green) and anti-Leda-1 antibodies (red). (E) Immunofluorescent staining of rat liver sections with anti-Leda-1 and anti-collagen4 antibodies (red). The results are representative of at least three independent experiments.

For further studies, a polyclonal guinea pig antibody was raised against a highly specific 22 aa c-terminal peptide of rat Leda-1. Specificity of this antibody was confirmed by immunocytochemistry using HEK293 cells transiently transfected with a cDNA coding for human Leda-1 (Supporting Information Fig. 3). When used in western blotting experiments, this anti-Leda-1 antibody labeled a protein with a molecular weight of ≈26 kDa in lysates of freshly isolated LSEC. Signal intensity in lysates of LSEC48h was drastically reduced and no signal was obtained from LMEC (Fig. 6A). Using this anti-Leda-1 antibody in immunohistochemistry, sinusoidal endothelial cells as well as endothelial cells of the central veins and portal vessels were nicely depicted in the liver, whereas no Leda-1-expressing cells were found in the lung (Fig. 6B).

Immunofluorescent double labeling of Leda-1 with Stabilin-2, Lyve-1, and CD31 in the liver showed colocalization of Leda-1 with Stabilin-2 (Fig. 6C,D) and Lyve-1 (not shown) in sinusoidal endothelium and colocalization of Leda-1 with CD31 in the endothelium of large vessels (not shown). Stabilin-2 did not show full coverage of the vessel wall preferentially being expressed in the central parts of the cell body and at the luminal side of LSEC (Fig. 6C,D). In contrast, Leda-1 labeled both the central parts as well as the periphery of LSEC and displayed full coverage of the vessel wall; in addition, Leda-1 was mostly located at the abluminal/basal side of LSEC (Fig. 6C,D). Preferential localization of Leda-1 at the abluminal side was also seen in relation to CD31 in endothelium of larger liver vessels (not shown). In addition, Leda-1 was found to partially colocalize with collagen IV, a major constituent of the basal lamina-like extracellular matrix of the space of Disse, further confirming its preferentially abluminal expression (Fig. 6E).

Leda-1 Localizes to Adherens Junctions in Polarized MDCK Cells.

To further analyze Leda-1 function, MDCK cells which do not express Leda-1 endogenously were stably transfected with full-length human Leda-1. Similar to Ajap-1, Leda-1 localized to the basolateral compartment of the membrane as demonstrated by its location below ZO-1, a cytoplasmic protein that targets tight junctions separating the apical and basolateral compartments in polarized cells (Fig. 7A,C). Furthermore, Leda-1 specifically targeted adherens junctions in MDCK cells, as shown by colocalization with E-cadherin (Fig. 7B,D). These data suggest a role for Leda-1 in cell polarity and adhesion.

Figure 7.

Leda-1 sorts to the basolateral compartment in MDCK cells and localizes to adherens junctions. Immunocytochemistry of Leda-1 transfected MDCK cells costained for ZO-1 (green) (A) and E-cadherin (green) (B) and Leda-1 (red) (A,B). Tight junction protein ZO-1 is seen in close apposition to Leda-1 but does not colocalize with it (A). In contrast, E-cadherin and Leda-1 colocalize (yellow) in the plasma membrane (B). (C) In vertical sections of Leda-1 transfected MDCK cells, Leda-1 (red) sorts to the basolateral compartment below ZO-1 (green). Insert: detail of one MDCK cell in vertical plane. (D) Costaining with E-cadherin on vertical sections also shows strong colocalization (yellow) of Leda-1 (red) and E-cadherin (green) in the lateral part of plasma membrane corresponding to adherens junctions.

Discussion

As LSECs are a prime example of organ-specific EC, this study sought to comprehensively analyze the molecular program underlying microenvironmentally controlled differentiation of LSEC in the liver. Multimodal microvascular gene expression profiling of freshly isolated LSEC versus LMEC and versus LSEC after short-term culture identified an LSEC-specific gene signature of 48 genes that is maintained by the hepatic microenvironment. Vice versa, induction of a specific set of genes was also demonstrated in cultured LSEC, indicating that LSEC in culture rather undergo a process of transdifferentiation than of mere deterioration. Up-regulation of Esm1 and Cxcr4 in cultured LSEC, genes known to be expressed in lung and tumor EC (TEC),17, 18 in combination with acquisition of cobblestone morphology and reduction in endocytic capacity suggests that LSEC transdifferentiate in vitro toward a continuous EC phenotype. Interestingly, these changes in culture are mirrored in vivo during sinusoidal capillarization in liver cirrhosis and in hepatocellular carcinoma (HCC), suggesting that related mechanisms could mediate LSEC transdifferentiation in vivo and in vitro.

This notion is further supported by overexpression of Ehd3 in LSEC, a member of the Ehd family of intracellular transport regulators.19 Colocalization of Ehd3 with Stabilin-1, but not Stabilin-2, implies a role for Ehd3 in trafficking of Stabilin-1-positive endosomes in LSEC. In addition, a strong decline in Ehd3 expression was found upon cultivation of LSEC. Interestingly, TEC isolated from rat HCC also showed strong down-regulation of Ehd3 protein as compared to normal LSEC,20 again indicating that the mechanisms that govern LSEC transdifferentiation in culture may also be responsible for pathogenic sinusoidal capillarization as in HCC.

As the functional and molecular repertoire of LSEC differs in vivo and in vitro, current LSEC culture models do not allow to adequately study LSEC biology in culture. Experiments to improve LSEC culture,9, 10 however, were evaluated by a very limited set of LSEC markers, i.e., fenestrations and SE antigen/CD32b expression. Our study strongly broadens the knowledge of LSEC-specific genes and thereby allows for a much more sophisticated analysis as well as for further improvement of LSEC culture models. Vegfs are probably the most important vascular growth and differentiation factors also for LSEC,21 and VEGF is widely used as a supplement in EC culture media including LSEC. Among the components of the VEGF signaling pathway, the three VEGF receptors and their coreceptor Nrp2 were shown to be strongly down-regulated in LSEC in vitro. Wnt-2, previously identified by us as a positive regulator of VegfR2, and its receptors were also drastically decreased upon culture. Thus, defective Wnt signaling may enhance and synergize with defective Vegf receptor activity in cultured LSEC in a vicious circle. Furthermore, primary Vegf receptor deficiency in cultured LSEC may explain impaired LSEC proliferation in culture despite the presence of high concentrations of Vegf in LSEC culture media.

As of now, a unique blood vascular EC-specific master regulator, as is Prox1 for lymphatic EC, has not be identified and specific gene expression in different blood vascular EC is thought to be mediated by combinations of otherwise nonspecific transcriptional regulators. The LSEC-specific transcription factors Tfec, Gata4, Maf, and Lmo3 identified here may well represent such an EC subtype-specific combination of transcriptional regulators. Gata4 has been shown to be important in development of the liver and of cardiac myocytes. Although Gata2, 3, and 6 are expressed in different EC and transcriptionally target EC-specific genes such as vWF, VCAM-1, and Tie-2,22 Gata4 is not generally expressed in blood vascular EC. Interestingly, endothelial Gata4 expression specifically induces the formation of the heart valves, a site where the sinusoidal endothelial marker proteins Stabilin-1 and -2 are also expressed.23 Thus, specific overexpression of Gata4 in LSEC versus LMEC-associated overexpression of Gata2, 3, and 5 renders Gata4 an attractive candidate for at least coregulating LSEC-specific gene transcription. Tfec, a bHLH transcription factor of the Mitf family contributes to IL-4-induced macrophage activation, suggesting a possible role in regulation of immune system processes in LSEC; interestingly, the Mitf-family member Tfeb is involved in placental vascularization.24 Although the proto-oncogene c-Maf has been shown to induce the angiogenic surface aminopeptidase N/CD13 in EC in vitro, our microarray analysis showed that CD13 expression was decreased in LSEC versus, LMEC indicating that MAF may target different genes in LSEC.25 Because Lmo family members have been shown to interact with Gata and bHLH transcription factors,26 Lmo3 could be involved in the regulation of LSEC-specific gene expression, possibly by interaction with Gata4 and/or Tfec.

In this study we furthermore show that the LSEC-specific differentiation program comprises a novel, highly conserved 278 aa type-1 transmembrane protein selectively expressed in liver endothelium that was named liver endothelial differentiation-associated protein (Leda)-1. The only known homolog of Leda-1 is Ajap-1/Shrew1, a type 1 transmembrane protein that targets adherens junctions in polarized cells and interacts with the E-cadherin/beta-catenin complex.14 By interaction with CD147, Ajap1 regulates tumor cell invasion.13 The intriguing abluminal localization of Leda-1 in LSEC and the basolateral sorting to adherens junctions in MDCK cells that is similar to Ajap-1 suggests a similar function for this novel endothelial protein in regulation of cell-cell and/or cell-matrix interactions in liver endothelium. As we and others failed to detect VE-cadherin in LSEC, it seems likely that LSECs do not possess classical adherens junctions. Nevertheless, it is likely that LSECs possess different kinds of junctional complexes that mediate adhesion to surrounding cells and matrix. Leda-1 might well be involved in this special adhesion apparatus. In contrast to all other known endothelial markers of the liver, which show preferential expression either in sinusoidal EC (Stabilin-1, Stabilin-2, Lyve-1, CD32b) or in nonsinusoidal EC (CD31), Leda-1 is an organ-specific endothelial protein similarly expressed by both sinusoidal and nonsinusoidal EC of the liver, indicating that Leda-1 is strictly dependent on the liver microenvironment. Therefore, it will be important to identify the hepatic factors that regulate its expression and to investigate its in vivo relevance in pathologic processes such as liver cirrhosis and HCC.

Ancillary