Endocytosis of collagen by hepatic stellate cells regulates extracellular matrix dynamics

Abstract

Hepatic stellate cells (HSCs) generate matrix, which in turn may also regulate HSCs function during liver fibrosis. We hypothesized that HSCs may endocytose matrix proteins to sense and respond to changes in microenvironment. Primary human HSCs, LX2, or mouse embryonic fibroblasts (MEFs) [wild-type; c-abl^−/−; or Yes, Src, and Fyn knockout mice (YSF^−/−)] were incubated with fluorescent-labeled collagen or gelatin. Fluorescence-activated cell sorting analysis and confocal microscopy were used for measuring cellular internalization of matrix proteins. Targeted PCR array and quantitative real-time PCR were used to evaluate gene expression changes. HSCs and LX2 cells endocytose collagens in a concentration- and time-dependent manner. Endocytosed collagen colocalized with Dextran 10K, a marker of macropinocytosis, and 5-ethylisopropyl amiloride, an inhibitor of macropinocytosis, reduced collagen internalization by 46%. Cytochalasin D and ML7 blocked collagen internalization by 47% and 45%, respectively, indicating that actin and myosin are critical for collagen endocytosis. Wortmannin and AKT inhibitor blocked collagen internalization by 70% and 89%, respectively, indicating that matrix macropinocytosis requires phosphoinositide-3-kinase (PI3K)/AKT signaling. Overexpression of dominant-negative dynamin-2 K44A blocked matrix internalization by 77%, indicating a role for dynamin-2 in matrix macropinocytosis. Whereas c-abl^−/− MEF showed impaired matrix endocytosis, YSF^−/− MEF surprisingly showed increased matrix endocytosis. It was also associated with complex gene regulations that related with matrix dynamics, including increased matrix metalloproteinase 9 (MMP-9) mRNA levels and zymographic activity. HSCs endocytose matrix proteins through macropinocytosis that requires a signaling network composed of PI3K/AKT, dynamin-2, and c-abl. Interaction with extracellular matrix regulates matrix dynamics through modulating multiple gene expressions including MMP-9.

hepatic stellate cells (HSCs) are pivotal for extracellular matrix homeostasis in liver. Quiescent HSCs contain numerous vitamin A lipid droplets and are located in the space of Disse. Upon liver injury, they are transformed to active myofibroblast-like cells, which are characterized by increased proliferation, α-smooth muscle actin expression, and robust collagen production (3–5, 13, 19, 26, 31). However, matrix also reciprocally regulates HSC function, including adhesion and migration through mechanical stiffness, integrin activation, and other pathways (2, 15, 20, 39, 57). However, the mechanisms by which HSCs sense their matrix surroundings are incompletely understood.

Collagen production and removal is a dynamic balance that determines tissue architecture. Disruption of this homeostasis can lead to tissue destruction or fibrosis. In liver, HSCs are important not only for collagen production that contributes to liver cirrhosis, but also for collagen degradation during cirrhosis resolution. Collagen has been reported to be internalized by macrophages and fibroblasts (1, 30) and recently by HSCs (36, 37). However, the mechanisms and relevance of this process remain undefined and were the focus of this investigation. Here we show that HSCs endocytose collagen and that this ultimately regulates production of the matrix-degrading protease matrix metalloproteinase 9 (MMP-9), implicating a feedback loop, whereby matrix sensing by HSCs regulates subsequent matrix degradation maneuvers. Furthermore, we provide mechanistic insight into the sensing process by showing that matrix endocytosis occurs through a macropinocytosis pathway that requires actin, dynamin, and c-abl. These studies add to our understanding of HSCs and their dynamic interaction with the matrix microenvironment.

MATERIALS AND METHODS

Reagents.

Oregon-green 488-conjugated gelatin, Fluorescein Conjugate DQ type I Collagen (DQ-collagen I), LysoTracker, and Dextran 10K were from Invitrogen (Grand Island, NY). Cysteine protease inhibitor E64D was from Calbiochem (San Diego, CA). Dynasore, Cytochalasin D, blebbistatin, methyl-β-cyclodextrin (MβCD), Filipin, nystatin, 5-ethylisopropyl amiloride (EIPA), PDGF-BB, and fibroblast growth factor were obtained from Sigma-Aldrich (Dorset, UK). VEGF and transforming growth factor (TGF)-β were from R&D Systems (Minneapolis, MN). Lysosomal-associated membrane protein-1 (LAMP-1) and CD63 antibodies were from Santa Cruz Biotechnology (Santa Cruz, CA). Early endosome antigen (EEA) antibody was from Cell Signaling Technology (Danvers, MA).

Cell culture.

Primary human HSCs were purchased from ScienCell Research Laboratories (Carlsbad, CA) and used at low passage (P3–5; catalog no. 5300). These cells have been well characterized and used widely in prior publications (10, 32, 46). Wild-type mouse embryonic fibroblast (MEF) cells, MEF cells isolated form c-abl and Arg double-knockout mice (c-abl^−/− MEF), and MEF cells isolated from Yes, Src, and Fyn knockout mice (YSF^−/− MEF) were kindly provided by Drs. E. Leof and M. McNiven, respectively (Mayo Clinic, Rochester, MN) (9, 58). Immortalized human-derived LX2 cells (54) are immortalized HSCs with a myofibroblast-like phenotype that were used in some experiments along with human HSCs. Cells were cultured in DMEM or specialized HSC medium, supplemented with 10% fetal bovine serum, 1 mmol/l l-glutamine, and 100 IU/ml streptomycin/penicillin.

Collagen endocytosis.

HSCs or LX2 cells were incubated with indicated concentrations of Oregon Green-labeled gelatin, DQ-collagen I, or Dextran 10K for indicated time in the absence or presence of 20 μm E64D. Cells were extensively washed and fixed with 4% paraformaldehyde and stained with LysoTracker. For subcellular localization analysis, cells were stained with EEA to mark early endosomes, LAMP-1 for late endosomes, and CD63 for multivesicular body marking. Images were taken with a Zeiss LSM 510 microscope.

For fluorescence-activated cell sorting (FACS), HSCs or LX2 cells were plated in six-well dishes, starved overnight with serum-free medium, and treated with indicated compounds for 30 min before incubation with fluorescent-labeled collagen or gelatin. Cells were extensively washed, trypsinized, and fixed with 2% formaldehyde and subjected to FACS analysis.

mRNA extraction, PCR array, real-time quantitative PCR.

To quantify the steady-state concentration of mRNAs, total RNA was extracted using RNeasy Mini kit (Qiagen, Crawley, West Sussex, UK) according to the manufacturer's instructions, quantified using a Spectrophotometer, and reverse transcribed. Real-time quantitative PCR was performed in an Applied Biosystems 7500 Real-time PCR System (Foster City, CA) to quantify the steady-state concentration of RNA using a SYBR Green PCR Kit. Primers are detailed in Table 1. The reaction contained 3.6–7.3 ng RNA and 0.5 μM primers. The relative gene expression (fold of GAPDH) of each gene was calculated by using ΔC_T method. RT² Profiler PCR Array (Venlo, Netherlands) was performed per the manufacturer's instruction.

Table 1.

Primers for RT-PCR analysis

Gene	Primer
GAPDH forward	CCAGGGCTGCTTTTAACTCT
GAPDH reverse	GGACTCCACGACGTACTCA
MMP9 forward	GACGTCTTCCAGTACCGA
MMP9 reverse	CTCAGGGCACTGCAGGAT

MMP9, matrix metalloproteinase 9.

Western blotting and zymography.

Conditioned medium was collected and centrifuged at 270 g for 3 min to remove cellular debris. Cells were lysed in RIPA buffer. Protein concentration in lysates was measured and used to normalize protein loading of gels. Cell extracts and conditioned medium were diluted fourfold in lysis buffer and reduced with 5% β-mercaptoethanol and then fractionated by PAGE and analyzed by Western blotting. Detection was performed using enhanced chemiluminescence.

MMP-9 activity in conditioned medium was evaluated by zymography as described (11). Briefly, 7.5% polyacrylamide gels containing 2 mg/ml gelatin were subjected to electrophoresis under nonreducing conditions. Following electrophoresis, SDS was removed by washing in 2.5% Triton X-100, and gels were incubated at 37°C for 18 h in 50 mM Tris·HCl, pH 8.0, 50 mM NaCl, 10 mM Ca₂Cl, and 0.05% Triton X-100. Gels were then stained in 0.2% Coomassie Brilliant Blue. Gelatinase activity was detected as clear bands on a dark background. Densitometric analysis of bands was performed.

Gene ontology analysis.

We used the software package The Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.7 (17) for gene ontology analysis. In brief, the GeneBank IDs of the genes from PCR array were inputted into the tools for gene ontology analysis of biological process and cellular components. The results were ranked based on adjusted P values using Benjamini-Hochberg method for multiple-comparison corrections. The count represents the number of genes involved in that function group, with the percentage of genes in the input genes that are involved in that function.

Statistical analysis.

Results are expressed as means ± SE. Significance was established using the Student's t-test and ANOVA when appropriate. Differences were considered significant when P < 0.05.

RESULTS

HSCs and LX2 cells internalize collagen.

To test the hypothesis that collagen internalization could regulate extracellular matrix dynamics, we initially investigated whether primary HSCs can internalize collagen. Collagen I was chosen for this study because it is pathologically increased in liver cirrhosis (38). When DQ-collagen I (1 μg/ml) was incubated with HSCs for 3 h at 37°C to allow visualization of internalized molecules, a pronounced intracellular vesicular accumulation of fluorescent signal was observed (Fig. 1A). As seen in confocal microscopy images in Fig. 1A, these collagen-containing vesicles (green) colocalized with LysoTracker (red), a marker for lysosomes. These data indicate that the collagen is internalized and is routed to lysosomes. As endocytosis is susceptible to temperature changes, we tested specificity of effect by performing studies at 4°C, a temperature broadly used to block endocytosis. As shown in Fig. 1A, the intracellular collagen-containing vesicles were significantly decreased at 4°C compared with 37°C. The results were further confirmed by using FACS to quantify collagen internalization. At 37°C, 91% of cells contained intracellular collagen; however, only 0.01% cells were positive at 4°C, suggesting that collagen internalization occurs through endocytosis (Fig. 1B). To explore temporal kinetics of collagen endocytosis, we incubated HSCs or LX2 cells with 0.5 μg/ml collagen for varying durations. As shown in Fig. 1C, HSC collagen endocytosis increased with incubation time; 53% of HSC cells were positive at 60 min, and 81% were positive at 360 min. The extracellular concentration of collagen also regulated collagen internalization; 30% of HSC cells incubated with 0.1 μg/ml gelatin for 3 h internalized collagen, which increased to 81% with 1.0 μg/ml and plateaued with subsequent increases in collagen concentration (Fig. 1D). Although LX2 cells showed slower endocytosis kinetics, they were able to internalize collagen as well (Fig. 1, C and D). Experiments using human and mouse pancreatic stellate cells showed similar results to that of LX2 and HSCs, suggesting that collagen internalization is a general phenomenon of stellate cells (data not shown). Additionally, experiments carried out using Oregon-green 488 gelatin (1 μg/ml; degradation product of full-length collagen) showed similar concentration- and temperature-dependent internalization, indicating that both collagen and its degradation products can be effectively internalized. Finally, we evaluated effects of HSC growth factor activation on collagen endocytosis. Both TGF-β and PDGF-BB significantly, albeit modestly, increased collagen internalization compared with vehicle-treated HSCs (Fig. 1E).

Fig. 1.

Human hepatic stellate cells (HSCs) internalize collagen in a time- and concentration-dependent manner. A: human HSCs were incubated with DQ collagen I (1 μg/ml, green) at indicated temperature for 3 h. LysoTracker (red) was added in the last 15 min of incubation. Pictures were taken using a LSM 520 confocal microscopy. In negative control (cells incubated with collagen without fluorescent labeling) or HSCs incubated with DQ collagen I (1 μg/ml) at 4°C for 3 h, LysoTracker (red) was shown scattered in the cytoplasm, but there was no collagen internalization as quantified by confocal microscopy. However, in HSCs incubated with DQ collagen I (1 μg/ml) at 37°C for 3 h, internalized collagen (green) colocalized with Lysotracker (red), as shown in the merged image as yellow. High-zoom colocalization image is also shown with colocalization, highlighted with arrows. B: human HSCs were incubated with DQ-collagen I (1 μg/ml) at indicated temperature for 3 h and then subjected to fluorescence-activated cell sorting (FACS) analysis. C: LX2 cells and human HSCs were incubated with DQ-collagen I (0.5 μg/ml) for indicated time and then subjected to FACS analysis (n = 3 independent experiments; P < 0.05 compared with time zero). D: human HSCs and LX2 were incubated with DQ-collagen I at indicated concentration at 37°C for 3 h and then subjected to FACS analysis (n = 3; P < 0.05 compared with time zero). E: HSC were preincubated with transforming growth factor (TGF)-β1 (5 ng/ml), PDGF-BB (100 ng/ml), FGF (10 ng/ml), or VEGF (5 ng/ml) for 30 min at 37°C before incubation with DQ collagen I (0.5 μg/ml) for 3 h at 37°C and then subjected to FACS analysis (n = 3; *P < 0.05 compared with control).

Collagen endocytosis occurs through macropinocytosis.

We next sought to discern the mechanism of collagen endocytosis by HSCs. A recent study suggested that macropinocytic endocytosis may be an important route for cell sensing (8). To test whether stellate cells use macropinocytosis as an endocytic pathway for collagen, we first performed studies with Dextran 10K, a classic marker for macropinocytosis. We treated cells with DQ-collagen I and Alexa Fluor 647 Dextran 10K together for 3 h. FACS analysis revealed that about 62% of cells cointernalized collagen and Dextran 10K (Fig. 2, A and B). We next evaluated the effect of EIPA, an ion exchange inhibitor that inhibits macropinocytosis (21), on collagen endocytosis. Cells pretreated with EIPA showed significantly reduced internalization of Dextran 10K as well as collagen (Fig. 2, A and B) as assessed by FACS. Additionally, under confocal microscopy, Dextran 10K colocalized with collagen and LysoTracker (Fig. 2C), indicating that HSCs internalize collagen through macropinocytosis. Finally, colocalization analysis showed internalized collagen colocalized with CD63, a marker of multivesicular bodies (Pearson's coefficient 0.56 ± 0.14) with lesser colocalization with EEA and LAMP-1 (Pearson's coefficient 0.15 ± 0.08), indicating that internalized collagen is eventually targeted for degradation (Fig. 2D), which is consistent with previous reports (22, 30).

Fig. 2.

Collagen endocytosis occurs through macropinocytosis. Human HSCs were pretreated with 5-ethylisopropyl amiloride (EIPA) (50 μM) for 30 min before incubation with Dextran 10K or DQ-collagen I (0.5 μg/ml) or both for 3 h, as labeled in the graph. A: representative FACS analysis is shown. B: quantification data of FACS are depicted (n = 3, *P < 0.05). C: human HSCs were incubated with DQ collagen I (1 μg/ml, green) and Dextran 10K (0.2 mg/ml) (blue) at 37°C for 3 h. Representative micrographs were taken using a LSM 520 confocal microscopy and showed internalized collagen (green) colocalized with Dextran 10K (blue) in the merged image. High-zoom colocalization image was also shown. D: LX2 cells were incubated with DQ-collagen I (1 μg/ml, green) for 3 h at 37°C. Cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% Triton-X. Early endosome antigen (EEA), lysosomal-associated membrane protein-1 (LAMP-1), and CD63 (red) antibodies were used for fluorescence analysis.

Collagen macropinocytosis occurs through an actin-regulated signaling network that requires dynamin-2 and phosphoinositide-3-kinase/AKT.

We next sought to examine the signaling mechanism of collagen macropinocytosis in greater detail. The involvement of the large GTPase- and endocytosis-regulating protein, dynamin-2, in collagen internalization was investigated using a dominant-negative dynamin-2 construct containing a well-characterized K44A point mutation (7). HSCs were transiently transfected to express control adenovirus or K44A dynamin-2. Cells expressing K44A dynamin-2 showed significantly decreased collagen internalization by 77% (Fig. 3A). The data were further confirmed by pretreatment of HSCs or LX2 cells with Dynasore (20 μM), a pharmacological inhibitor of dynamin, which diminished HSC collagen endocytosis by 51% (Fig. 3A). Because macropinocytosis is an actin-dependent endocytic process, we next tested pharmacological inhibitors of actin cytoskeleton dynamics and myosin activity on collagen internalization by HSCs. Cells were preincubated with Cytochalasin D (0.5 μM) to inhibit actin polymerization, ML7 (1 μM) to inhibit myosin light chain kinase, or blebbistatin (50 μM) to block myosin II for 30 min before cells were incubated with DQ-collagen I for 3 h. FACS analysis showed that Cytochalasin D, ML7, and blebbistatin all decreased collagen internalization by 47%, 45%, and 39%, respectively (Fig. 3B). These data suggest that actin and myosin are important for collagen endocytosis in HSCs. Because dynamin regulates both clathrin- and caveolae-dependent endocytosis, we next explored which of these pathways mediated collagen endocytosis using a panel of pharmacological inhibitors. Cells were pretreated with inhibitors of caveolae-dependent endocytosis [5 mM MβCD (49), 5 μg/ml filipin (44), 30 μM nystatin (41), or clathrin-dependent endocytosis (0.45 M sucrose) (16)] for 30 min, and then collagen endocytosis was measured. Whereas sucrose inhibited endocytosis, inhibitors of caveolae pathway did not, indicating that collagen endocytosis occurs through a clathrin-dependent pathway (Fig. 3C). Phosphoinositide-3-kinase (PI3K) and the downstream effector protein AKT are important in regulating completion of macropinocytosis (43). We next tested their roles in collagen internalization by HSCs using pharmacological inhibitors. Wortmannin and AKT inhibitor blocked collagen internalization by 70% and 90%, respectively (Fig. 3D). Under the current concentration and treatment durations, we did not observe significant effects of the pharmacological inhibitors on cell viability based on DAPI uptake (Fig. 3E). These studies indicate that collagen macropinocytosis occurs through a dynamin-, PI3K/AKT-dependent actin-regulated pathway.

Fig. 3.

Collagen internalization is regulated by Dynamin-2, phosphoinositide-3-kinase/AKT, and actin cytoskeleton. Human HSCs were infected with K44A mutant Dynamin-2 adenovirus for 24 h or pretreated with Dynasore (A), blebbistatin, ML7, or Cytochalasin D (B), methyl-β-cyclodextrin (MβCD), filipin, nystatin, or 0.5 M sucrose (C), or wortmannin or AKT inhibitor (D) for 30 min before DQ-collagen I (0.5 μg/ml) incubation for 3 h. Cells were subjected to FACS analysis to measure matrix protein endocytosis (n = 4, *P < 0.05 compared with positive control). E: DAPI exclusion assay was used to identify the percentage of viable cells, which was not significantly different between control and compounds assessed.

C-abl, but not Src is critical for macropinocytosis.

Both c-abl and Src family nonreceptor tyrosine kinases are important for matrix dynamics and have been implicated in PI3K/AKT signaling. We therefore hypothesized that c-abl and Src may be important for HSC collagen endocytosis. We used MEF cells isolated from c-abl and Arg double-knockout mice (c-abl^−/−) and MEF cells from YSF^−/−. These double- and triple-knockout cells delete multiple family members with redundant functions. Compared with wild-type MEF, c-abl^−/− MEF showed less collagen internalization at a range of collagen concentrations (Fig. 4A). Surprisingly, YSF^−/− MEF cells demonstrated significantly enhanced collagen endocytosis compared with wild-type cells (Fig. 4A). To explore this unexpected result further, we measured c-abl activity using phospho-CrkII as a readout. Western blot data showed increased phospho-CrkII in YSF^−/− MEF compared with WT MEF (Fig. 4B). These studies indicate that c-abl activity is upregulated in YSF^−/− MEF and may account for the unanticipated increase in collagen endocytosis observed in these cells.

Fig. 4.

Knockdown of c-abl decreases but knockdown of Yes, Src, and Fyn (YSF) increases collagen endocytosis in mouse embryonic fibroblasts (MEF) cells. MEF cells isolated form wild-type (WT), c-abl^−/− mice (c-abl^−/−), and Yes^−/−Src^−/−Fyn^−/− (YSF^−/−) cells were starved overnight and then incubated with indicated concentration of DQ-collagen I (0.5 μg/ml) for 3 h before the cells were subjected to FACS analysis. A: FACS analysis comparing collagen uptake among WT, c-abl^−/−, and YSF^−/− MEF cells after incubation with indicated concentration of DQ-collagen I for 3 h. B: p-CRKII activity is shown in cells, including quantitation of densitometry (n = 3, *P < 0.05 compared with WT).

Interaction of collagen with HSCs regulates genes related with extracellular matrix dynamics.

Finally, we studied the functional response of HSCs to extracellular collagen. As a first step, we performed a targeted PCR array to determine changes in mRNA transcripts in HSCs in response to collagen. Multiple targets related with collagen degradation or synthesis were changed upon HSC interaction with collagen, including a prominent increase in MMP-9 (Table 2). We next performed quantitative real-time PCR for a number of individual matrix regulatory proteins to confirm the array data. This analysis confirmed that MMP-9 was upregulated by 3.7-fold (Fig. 5A) after HSCs interact with collagen. Several other changes were observed as well (Fig. 5, B–D), but we focused on MMP-9 for further confirmation and analysis owing to the known role of this protein in matrix degradation. Gelatin zymography confirmed increased MMP-9 activity in response to collagen (Fig. 5E). In addition to effects on MMP activity, interaction with collagen significantly decreased collagen Iα mRNA levels by 22% based on RT-PCR (Fig. 5D). Moreover, mRNA levels of collagen XVIII α1 were also diminished in the microarray analysis (Table 2). These data indicate that extracellular collagen initiates a response in HSCs that regulates matrix dynamics through mechanisms that include MMP-9 as well as collagen production. Additionally, gene ontology analysis suggested that genes regulating wound-healing response, cell survival/apoptosis, and those encoding proteins residing in the extracellular space were among the highly regulated genes (Tables 3, 4, 5, and 6). This analysis supports previous observations suggesting important survival signal provided by cellular collagen interaction and also supports our data that interaction with extracellular matrix (collagen) by cells is able to provide feedback to its surrounding extracellular components. However, it should be noted that the present analysis could not distinguish whether the observed signaling effects are secondary to collagen ligand interaction with a collagen membrane receptor, such as an integrin family member vs. the internalization process itself.

Table 2.

Changes in HSC mRNA levels upon collagen interaction

Unigene	GeneBank	Symbol	Description	Fold Changes
Hs.227817	NM_004049	BCL2A1	B-cell CLL/lymphoma 2 (BCL2)-related protein A1	11.79415
Hs.591873	NM_000880	IL7	Interleukin 7	7.889862
Hs.1349	NM_000758	CSF2	Colony stimulating factor 2 (granulocyte-macrophage)	7.061624
Hs.412707	NM_000194	HPRT1	Hypoxanthine phosphoribosyltransferase 1	4.756828
Hs.126256	NM_000576	IL1B	Interleukin 1β	4.169863
Hs.303649	NM_002982	CCL2	Chemokine (C-C motif) ligand 2	3.863745
Hs.150749	NM_000633	BCL2	B-cell CLL/lymphoma 2	3.837056
Hs.141125	NM_004346	CASP3	Caspase 3, apoptosis-related cysteine peptidase	3.5801
Hs.83169	NM_002421	MMP1	Matrix metalloproteinase 1 (interstitial collagenase)	3.555371
Hs.591016	NM_003805	CRADD	Caspase 2 and RIPK1 domain containing adaptor with death domain	3.24901
Hs.297413	NM_004994	MMP9	Matrix metalloproteinase 9 (gelatinase B, 92 kDa gelatinase, 92 kDa type IV collagenase)	3.197479
Hs.522632	NM_003254	TIMP1	Tissue inhibitor of metalloproteinase 1	3.182146
Hs.534255	NM_004048	B2M	β2-microglobulin	2.969047
Hs.744837	NM_031850	AGTR1	Angiotensin II receptor, type 1	2.928171
Hs.480653	NM_001154	ANXA5	Annexin A5	2.928171
Hs.436873	NM_002210	ITGAV	Integrin-αV (vitronectin receptor, α-polypeptide, antigen CDS1)	2.907945
Hs.443914	NM_000454	SOD1	Superoxide dismutase 1, soluble	2.732081
Hs.244139	NM_000043	FAS	Fas (TNF receptor superfamily, member 6)	2.713209
Hs.523185	NM_012423	RPL13A	Ribosomal protein L13a	2.713209
Hs.74615	NM_006206	PDGFRA	Platelet-derived growth factor receptor α polypeptide	2.639016
Hs.516578	NM_006287	TFPI	Tissue factor pathway inhibitor (lipoprotein-associated coagulation inhibitor)	2.566852
Hs.77274	NM_002658	PLAU	Plasminogen activator, urokinase	2.531513
Hs.81564	NM_002619	PF4	Platelet factor 4	2.42839
Hs.713079	NM_000602	SERPINE1	Serpin peptidase inhibitor, clade E (nexin, plasminogen activator inhibitor type 1), member 1	2.411616
Hs.544577	NM_002046	GAPDH	Glyceraldehyde-3-phosphate dehydrogenase	2.394957
Hs.2490	NM_033292	CASP1	Caspase 1, apoptosis-related cysteine peptidase (interleukin 1, β convertase)	2.158456
Hs.440848	NM_000552	VWF	Von Willebrand factor	2.099433
Hs.647092	NM_000603	NOS3	Nitric oxide synthase 3 (endothelial cell)	1.972465
Hs.731912	NM_003879	CFLAR	Caspase 8 and Fas-associated death domain-like apoptosis regulator	1.958841
Hs.483635	NM_000800	FGF1	Fibroblast growth factor 1 (acidic)	1.931873
Hs.491582	NM_000930	PLAT	Plasminogen activator, tissue	1.765406
Hs.502876	NM_004040	RHOB	Ras homolog gene family, member B	1.765406
Hs.654458	NM_000600	IL6	Interleukin 6 (interferon β2)	1.693491
Hs.643447	NM_000201	ICAM1	Intercellular adhesion molecule 1	1.624505
Hs.631562	NM_003706	PLA2G4C	Phospholipase A2, group IVC (cytosolic, calcium-independent)	1.591073
Hs.203717	NM_002026	FN1	Fibronectin 1	1.547565
Hs.109225	NM_001078	VCAM1	Vascular cell adhesion molecule 1	1.536875
Hs.2030	NM_000361	THBD	Thrombomodulin	1.526259
Hs.591014	NM_003006	SELPLG	Selectin P ligand	1.515717
Hs.89499	NM_000698	ALOX5	Arachidonate 5-lipoxygenase	1.494849
Hs.694	NM_000588	IL3	Interleukin 3 (colony-stimulating factor, multiple)	1.394744
Hs.709191	NM_000625	NOS2	Nitric oxide synthase 2, inducible	1.394744
Hs.82848	NM_000450	SELE	Selectin E	1.394744
Hs.728756	NM_000655	SELL	Selectin L	1.328686
Hs.512937	NM_001872	CPB2	Carboxypeptidase B2 (plasma)	1.319508
Hs.713645	NM_001955	EDN1	Endothelin 1	1.319508
Hs.2007	NM_000639	FASLG	Fas ligand (TNF superfamily, member 6)	1.319508
Hs.404914	NM_003183	ADAM17	ADAM metallopeptidase domain 17	1.245862
Hs.592605	NM_002538	OCLN	Occludin	1.197479
Hs.219140	NM_002521	NPPB	Natriuretic peptide B	1.189207
Hs.730607	NM_001953	TYMP	Thymidine phosphorylase	1.156688
Hs.183713	NM_001957	EDNRA	Endothelin receptor type A	1.140764
Hs.298469	NM_000789	ACE	Angiotensin I converting enzyme (peptidyl-dipeptidase A) 1	1.123531
Hs.113916	NM_001716	CXCR5	Chemokine (C-X-C motif) receptor 5	1.117287
Hs.479756	NM_002253	KDR	Kinase insert domain receptor (a type III receptor tyrosine kinase)	1.117287
Hs.164226	NM_003246	THBS1	Thrombospondin 1	1.109569
Hs.467304	NM_000641	IL11	Interleukin 11	1.101905
Hs.594454	NM_002019	FLT1	Fms-related tyrosine kinase 1 (vascular endothelial growth factor/vascular permeability factor receptor)	1.071773
Hs.654616	NM_032992	CASP6	Caspase 6, apoptosis-related cysteine peptidase	1.057018
Hs.1407	NM_001956	EDN2	Endothelin 2	1.021012
Hs.143436	NM_000301	PLG	Plasminogen	1.006956
Hs.19383	NM_000029	AGT	Angiotensinogen (serpin peptidase inhibitor, clade A, member B)	−0.987509
Hs.478275	NM_003810	TNFSF10	Tumor necrosis factor (ligand) superfamily, member 10	−1
Hs.514821	NM_002985	CCL5	Chemokine (C-C motif) ligand 5	−1.05702
Hs.302085	NM_000961	PTGIS	Prostaglandin I2 (prostacyclin) synthase	−1.10191
Hs.655801	NM_003841	TNFRSF10C	Tumor necrosis factor receptor superfamily, member 10c, decoy without an intracellular domain	−1.10957
Hs.68061	NM_021972	SPHK1	Sphingosine kinase 1	−1.12506
Hs.76206	NM_001795	CDH5	Cadherin 5, type 2 (vascular endothelium)	−1.1487
Hs.479754	NM_000222	KIT	V-kit Hardy-Zuckerman 4 feline sarcoma viral oncogene homolog	−1.16473
Hs.93177	NM_002176	IFNB1	Interferon β1, fibroblast	−1.20581
Hs.218040	NM_000212	ITGB3	Integrin β3 (platelet glycoprotein IIIa, antigen CD51)	−1.23114
Hs.241570	NM_000594	TNF	Tumor necrosis factor	−1.33793
Hs.376675	NM_000442	PECAM1	Platelet/endothelial cell adhesion molecule	−1.34723
Hs.252820	NM_002632	PGF	Placental growth factor	−1.36604
Hs.211600	NM_006290	TNFAIP3	Tumor necrosis factor, α-induced protein 3	−1.43396
Hs.505654	NM_002205	ITGA5	Integrin α5 (fibronectin receptor, α polypeptide)	−1.45397
Hs.643813	NM_002211	ITGB1	Integrin β1 (fibronectin receptor, β polypeptide, antigen CD29 includes MDF2, MSK12)	−1.47427
Hs.517356	NM_030582	COL1BA1	Collagen, type XVIII, α1	−1.61328
Hs.490330	NM_000906	NPR1	Natriuretic peptide receptor A/guanylate cyclase A (atrionatriuretic peptide receptor A)	−1.67018
Hs.171995	NM_001648	KLK3	Kallikrein-related peptidase 3	−2.04202
Hs.531668	NM_002996	CX3CL1	Chemokine (C-X3-C motif) ligand 1	−2.62079
Hs.519842	NM_003804	RIPK1	Receptor (TNFRSF)-interacting serine-threonine kinase 1	−3.16017
Hs.516966	NM_138578	BCL2L1	BCL2-like 1	−3.36359
Hs.369675	NM_001146	ANGPT1	Angiopoietin 1	−3.70635
Hs.89640	NM_000459	TEK	TEK tyrosine kinase, endothelial	−3.83706
Hs.73793	NM_003376	VEGFA	Vascular endothelial growth factor A	−3.86375
Hs.624291	NM_004324	BAX	BCL2-associated X protein	−5.65685
Hs.513617	NM_004530	MMP2	Matrix metalloproteinase 2 (gelatinase A, 72 kDa gelatinase, 72 kDa type IV collagenase)	−11.7942

Hepatic stellate cells (HSCs) were treated with vehicle or collagen for 2 h before mRNA was extracted for PCR array. The data were arranged according to mRNA fold change upon collagen interaction.

Fig. 5.

Collagen and HSC interaction stimulates increase in matrix metalloproteinase (MMP)-9 mRNA levels and zymographic activity. A: human HSCs were incubated with BSA or DQ-collagen I (0.5 μg/ml) for 2 h, and RNA was extracted for real-time PCR for MMP-9, MMP-2 (B), smooth muscle actin (SMA) (C), and collagen Iα (D) (n = 3, *P < 0.05 compared with control). E: human HSCs were incubated with BSA or DQ-collagen I (0.5 μg/ml) for 48 h, and cell medium was collected for gelatin zymography. Representative data and quantification are shown from 3 independent zymographies (*P < 0.05).

Table 3.

Functional gene ontology of upregulated genes

Term	Count	%	Adjusted P Value Using Benjamini-Hochberg Method
Response to wounding	23	39	1.70e-13
Regulation of cell proliferation	25	42.4	2.30e-12
Cell migration	16	27.1	2.10e-10
Regulation of body fluid levels	13	22	2.30e-10
Wound healing	14	23.7	3.20e-10
Localization of cell	16	27.1	4.80e-10
Cell motility	16	27.1	4.80e-10
Regulation of response to external stimulus	13	22	5.50e-10
Regulation of apoptosis	22	37.3	1.10e-09
Regulation of programed cell death	22	37.3	1.20e-09
Regulation of cell death	22	37.3	1.10e-09
Cell activation	15	25.4	1.60e-09
Blood coagulation	11	18.6	1.80e-09
Coagulation	11	18.6	1.80e-09
Negative regulation of apoptosis	16	27.1	1.70e-09
Response to oxygen levels	12	20.3	1.70e-09
Negative regulation of programmed cell death	16	27.1	1.80e-09
Negative regulation of cell death	16	27.1	1.80e-09
Blood vessel development	14	23.7	2.20e-09
Hemostasis	11	18.6	2.20e-09

The GeneBank IDs of the genes from the PCR array were analyzed using DAVID v6.7 for gene ontology analysis of biological process and cellular components. The results were ranked based on adjusted P values using Benjamini-Hochberg statistical method for multiple-comparison corrections. The count represents the number of genes involved in that function group. % represents the percentage of genes in the input genes that are involved in that function.

Table 4.

Cellular component gene ontology of upregulated genes

Term	Count	%	Adjusted P Value Using Benjamini-Hochberg Method
Extracellular space	26	44.1	4.90e-16
Extracellular region part	29	49.2	3.70e-16
Extracellular region	34	57.6	1.70e-12
External side of plasma membrane	8	13.6	2.10e-04
Platelet α granule	5	8.5	2.20e-03
Secretory granule	7	11.9	2.20e-03
Plasma membrane part	22	37.3	2.30e-03
Plasma membrane	29	49.2	7.50e-03
Extracellular matrix	8	13.6	8.00e-03
Cell surface	8	13.6	7.50e-03
Platelet α granule lumen	4	6.8	7.60e-03
Cytoplasmic membrane-bound vesicle lumen	4	6.8	8.50e-03
Vesicle lumen	4	6.8	9.00e-03
Integral to plasma membrane	13	22	2.90e-02
Intrinsic to plasma membrane	13	22	3.30e-02
Cytoplasmic vesicle	9	15.3	3.90e-02
Vesicle	9	15.3	4.80e-02
Cytoplasmic membrane-bound vesicle	8	13.6	5.40e-02
Cytosol	13	22	5.30e-02
Membrane-bound vesicle	8	13.6	5.70e-02

Data are as described in Table 3.

Table 5.

Functional gene ontology of downregulated genes

Term	Count	%	Adjusted P Value Using Benjamini-Hochberg Method
Regulation of cell proliferation	12	46.2	4.80e-05
Regulation of programed cell death	12	46.2	3.30e-05
Regulation of cell death	12	46.2	2.30e-05
Regulation of apoptosis	11	42.3	1.80e-04
Regulation of locomotion	7	26.9	2.00e-04
Cell adhesion	10	38.5	3.90e-04
Biological adhesion	10	38.5	3.40e-04
Blood vessel development	7	26.9	5.30e-04
Vasculature development	7	26.9	5.40e-04
Positive regulation of cell proliferation	8	30.8	6.80e-04
Cell migration	7	26.9	7.60e-04
Programmed cell death	9	34.6	7.20e-04
Positive regulation of apoptosis	8	30.8	6.70e-04
Positive regulation of programmed cell death	8	30.8	6.50e-04
Positive regulation of cell death	8	30.8	6.30e-04
Regulation of cell migration	6	23.1	7.40e-04
Localization of cell	7	26.9	9.00e-04
Cell motility	7	26.9	9.00e-04
Regulation of cell motion	6	23.1	1.30e-03
Cell death	9	34.6	1.50e-03

Data are as described in Table 3.

Table 6.

Cellular component gene ontology of downregulated genes

Term	Count	%	Adjusted P Value Using Benjamini-Hochberg Method
Extracellular space	12	46.2	2.30e-06
Extracellular region part	12	46.2	3.60e-05
Cell surface	7	26.9	1.60e-03
Extracellular region	13	50	4.00e-03
Integrin complex	3	11.5	3.00e-02
Receptor complex	4	15.4	3.00e-02
Plasma membrane part	12	46.2	2.40e-02
External side of plasma membrane	4	15.4	5.50e-02
Secretory granule	4	15.4	5.70e-02
Plasma membrane	15	57.7	5.20e-02
Platelet α granule	3	11.5	4.90e-02
Integral to plasma membrane	8	30.8	5.40e-02
Intrinsic to plasma membrane	8	30.8	5.60e-02
Cell fraction	7	26.9	1.10e-01
Platelet α granule membrane	2	7.7	1.30e-01
Cytoplasmic membrane-bound vesicle	5	19.2	1.30e-01
Soluble fraction	4	15.4	1.30e-01
Membrane-bound vesicle	5	19.2	1.30e-01
Ruffle membrane	2	7.7	1.20e-01
Cytoplasmic vesicle	5	19.2	1.70e-01

Data are as described in Table 3.

DISCUSSION

Tissue collagen homeostasis is tightly regulated by collagen production and degradation. Imbalance of this homeostasis leads to tissue destruction or fibrosis. In the current study, we investigated whether HSCs internalize collagen as a means to sense their microenvironment and respond accordingly. We make the novel observation in this study that human HSCs internalize collagen through a macropinocytosis mechanism that is regulated by dynamin-2 and c-abl and that interaction of collagen with HSCs provides feedback on MMP-9 production.

In this study, we identify macropinocytosis, a nonselective endocytosis pathway, as a mechanism for collagen endocytosis by HSCs. This was based on colocalization studies with Dextran 10K as well as inhibition of matrix endocytosis by compounds that perturb macropinocytosis, including EIPA. Perturbation of actin cytoskeleton, PI3K/AKT signaling, and the large GTPase dynamin-2 also diminished HSC collagen internalization. Prior studies have implicated these pathways in macropinocytosis (27, 35). Innate immune cells such as macrophages continuously survey their environment in search of foreign particles and soluble antigens through macropinocytosis (24, 27). Our study suggests that HSCs also use this mechanism to examine and respond to environmental changes and highlight increasing evidence for innate immune surveillance function of HSCs, especially considering that some matrix proteins can stimulate Toll-like receptor 4 and other innate immune markers on HSCs (6, 45, 52).

The interaction between HSC and extracellular matrix is complex. Extracellular collagens can be recognized by cells through putative collagen receptors including integrins, the discoidin domain receptors (DDRs), and mannose receptor family, which upon ligand binding elicit different signaling pathways that respond to extracellular matrix (14, 25, 48, 56). Interaction of integrin with collagens provides multiple cues, including cell-extracellular matrix adhesion and cellular polarization that eventually regulate many aspects of cell behavior, including positive or negative feedback on synthesis of the extracellular proteins themselves. For example, prior observations suggest that negative feedback regulation of collagen synthesis is a major role of integrin α1β1 receptor stimulation in fibroblasts (23, 40). DDR receptors regulate cell adhesion, migration, proliferation, and remodeling of the extracellular matrix by controlling the expression and activity of MMPs (42, 55). Similarly, uPARAP/Endo180 from the mannose receptor family was found to be important for collagen internalization in activated rat HSCs (37) and other cultured cells (12, 29, 33, 37, 51). Our study provides evidence that collagen macropinocytosis is an additional communication mechanism between HSCs and extracellular matrix. Further studies will be required to elucidate the distinct or related roles of matrix-receptor binding and macropinocytosis when HSCs are exposed to collagen in the microenvironment.

Surprisingly, nonreceptor tyrosine kinase family members, c-abl and Src, differentially regulated HSC collagen uptake. MEF cells with c-abl and Arg double knockdown showed decreased collagen uptake compared with control MEF cells. C-abl tyrosine kinase likely regulates macropinocytosis by regulating Abi1 interactions with several actin polymerization regulatory complexes, including PI3K, whose involvement in actin dynamics is well established (53). In contradistinction, MEF cells with Src isoform deletion showed enhanced collagen uptake compared with control MEF. Indeed, CRK phosphorylation was increased in YSF MEF, indicating that lack of Src may lead to a compensatory upregulation of c-abl with ensuing increase in matrix macropinocytosis in these cells. Although most of our studies were performed in the absence of TGF-β1 stimulation, the signaling networks we studied are classically downstream from TGF-β1 stimulation, and this growth factor did stimulate collagen internalization in our studies and in prior studies (18).

Although recognized predominantly for their fibrogenic activity, HSCs also degrade matrix by producing proteolytic MMPs. Collagen degradation includes the cleavage of full-length extracellular collagen by collagenases (MMP-1, -2, -8, -13, and the membrane-bound MMP-14 and -15) followed by further byproduct cleavage by the gelatinases, MMP-2, or MMP-9 (28, 34, 47). Degraded collagen is then internalized by cells and targeted to lysosomes, a step that may account for the rheostatic function of HSCs, as proposed in the present study. Thus the proteolytic degradation of matrix may achieve the goal of allowing HSCs to “sample and respond” to extracellular matrix dynamics by modifying collagen production or collagen-degrading enzymes. Indeed, we did observe an increase MMP-9 in production in response to collagen endocytosis, supporting such a feedback loop. A prior study supporting this concept showed that HSCs generate MMP-2 in response to type I collagen fibril exposure (50). In addition to MMP, interaction with collagen significantly decreased HSC collagen Iα and collagen XVIII α1 mRNA transcripts, indicating a negative feedback on collagen production. In summary, the present studies identify a role for HSCs as a sensor and responder to the extracellular matrix microenvironment through their ability to endocytose extracellular matrix proteins and adjust their production of proteases in response to what they sample.

GRANTS

This work was supported by National Institutes of Health (NIH) Grants DK59615 and AA021171 (V. Shah), the Cell Biology Core of the Mayo Clinic Center for Cell Signaling in Gastroenterology (P30DK084567), and NIH K12 CA90628 (B. Ji).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: Y.B. conception and design of research; Y.B., D.M., M.D., and S.C. performed experiments; Y.B., D.M., M.D., B.J., X.L., S.C., and V.H.S. analyzed data; Y.B., D.M., M.D., S.C., and V.H.S. interpreted results of experiments; Y.B. prepared figures; Y.B. and V.H.S. drafted manuscript; Y.B. and V.H.S. edited and revised manuscript; Y.B. and V.H.S. approved final version of manuscript.