A dual reporter gene transgenic mouse demonstrates heterogeneity in hepatic fibrogenic cell populations



Activation of hepatic stellate cells (HSCs) and other resident mesenchymal cells into myofibroblasts expressing alpha smooth muscle actin (αSMA) and collagen I is a key event in liver fibrogenesis. However, the temporal expression profiles of αSMA and collagen I genes in these cells is unknown. To address this question, we studied αSMA and collagen α1(I) transcriptional patterns in primary cultures of HSCs, and additionally, in an in vivo model of secondary biliary fibrosis using transgenic mice that express the Discomsoma sp. red fluorescent protein (RFP) and the enhanced green fluorescent protein (EGFP) reporter genes under direction of the mouse αSMA and collagen α1(I) promoter/enhancers, respectively. The αSMA-RFP mice were crossed with collagen-EGFP mice to generate double transgenic mice. Reporter gene expression in cultured HSCs demonstrated that both transgenes were induced at day 3 with continued expression through day 14. Interestingly, αSMA and collagen α1(I) transgenes were not coexpressed in all cells. Flow cytometry analysis showed three different patterns of gene expression: αSMA-RFP positive cells, collagen-EGFP positive cells, and cells expressing both transgenes. αSMA-only and αSMA/collagen expressing cells showed higher expression levels of synaptophysin, reelin, MMP13, TIMP1, and ICAM-1 compared to collagen-only expressing cells, as assessed by real-time PCR. Following bile duct ligation, αSMA and collagen α1(I) transgenes were differentially expressed by peribiliary, parenchymal and vascular fibrogenic cells. Peribiliary cells preferentially expressed collagen α1(I), while parenchymal myofibroblasts expressed both αSMA and collagen α1(I). In conclusion, these data demonstrate heterogeneity of gene expression in myofibroblastic cells during active fibrogenesis. These reporter mice provide a useful tool to further characterize fibrogenic cell types and to evaluate antifibrotic drugs. (HEPATOLOGY 2004.)

Following repeated injury, the liver undergoes a tissue remodeling process similar to that observed in other organs.1 Besides hepatocyte regeneration, resident nonparenchymal cells differentiate into myofibroblasts to generate extracellular matrix (ECM), which is required for adequate tissue repair. If the hepatic injury persists, myofibroblasts accumulate and produce large amounts of extracellular matrix proteins, mainly collagen I, leading to fibrosis. Based on its pivotal role in fibrogenesis, inhibition of myofibroblast accumulation is currently considered a rational strategy for the treatment of chronic liver diseases.2

The cellular source of myofibroblasts in the injured liver is under extensive investigation. Studies in the 1980s demonstrated that hepatic stellate cells (HSC), rather than hepatocytes, secrete large amounts of collagen in the injured liver.3, 4 Following injury, HSCs transdifferentiate into myofibroblast-like cells, with increased proliferation, cytokine secretion, collagen synthesis, and cell contractility5. Prolonged culture of primary HSCs on plastic results in cell activation similar to that described in vivo.6 Studies performed in recent years indicate that other cell types such as portal fibroblasts and vascular myofibroblasts also have fibrogenic potential in the injured liver.7–10

A common feature for myofibroblastic cells in the injured liver is the expression of smooth muscle α actin (αSMA), which is associated with the acquisition of motogenic and contractile properties.6 Although the expression of αSMA mRNA has been well documented in cultures of HSCs and liver myofibroblasts, the expression of αSMA by immunohistochemistry is considered the gold-standard method to detect fibrogenic cells in human and rodent liver specimens.11–13 The quantification of the number of αSMA positive cells in the liver is commonly used as a marker of active fibrogenesis and is being used to estimate the efficacy of antifibrotic therapies in both human and experimental liver fibrosis.14, 15 However, the relationship between the expression of αSMA in myofibroblastic cells and increased collagen expression has not been clearly established.

To address these questions, we generated transgenic mice that express the Enhanced Green Fluorescent Protein (EGFP) reporter gene under the direction of the collagen α1(I) promoter/enhancers, and RFP (Red Fluorescent Protein) under direction of the promoter/enhancers of αSMA. COLL-EGFP mice contain 3.2 kb of the mouse collagen α1(I) promoter linked 3′ to the enhancing elements containing chromatin hypersensitivity sites 4 and 5.16 The COLL-EGFP transgene is appropriately expressed in hepatic myofibroblasts during culture activation and in carbon tetrachloride (CCL4) induced fibrosis.16 To characterize the expression of αSMA in collagen α1(I)-expressing cells, we generated mice that express the Discomsoma sp. RFP reporter gene driven by the αSMA promoter/enhancer and crossed these mice (αSMA-RFP) with the COLL-EGFP mice. Using these double transgenic mice (αSMA-RFP/COLL-EGFP), we investigated the temporal expression profiles of αSMA and collagen α1(I) genes in primary cultures of HSCs and spatial expression profiles in an in vivo model of secondary biliary fibrosis. We provide evidence for heterogeneity of gene expression in hepatic myofibroblasts in vitro and in vivo.


HSCS, hepatic stellate cells; αSMA, alpha smooth muscle actin; RFP, red fluorescent protein; EGFP, enhanced green fluorescent protein; PFA, paraformaldehyde, GFAP, glial fibrillary acidic protein; MMP, matrix metalloproteinase; ECM, extracellular matrix; PBS, phosphate-buffered saline; BDL, bile duct ligated; FACS, Fluorescence Activated Cell Sorting; PCR, polymerase chain reaction.

Materials and Methods

Generation of Transgenic Mice and Genotyping.

Coll-EGFP mice were generated as previously described.16 To generate the αSMA-RFP transgenic mouse, the vascular smooth muscle α-actin promoter/enhancer (from plasmid pSMP8)17 and the RFP reporter gene DsRED1-1 (Clontech, Palo Alto, CA) were digested with BamHI and blunted using a standard Klenow reaction. A 3.0-kb fragment of the αSMA promoter was blunt ligated into the DsRED1-1 reporter gene vector (at the blunted BamHI site) using a standard ligase reaction. This cloning strategy was required to maintain the proper reading frame of the DsRED1-1 cDNA. The resultant αSMA-RFP construct was expanded in DH5α cells and isolated using the endotoxin free DNA isolation kit (Qiagen, Valencia, CA). The αSMA-RFP construct was linearized with EcoRI and SspI and the ≈5.0-kb fragment of the transgene was purified on a TAE-agarose gel. Approximately 1 to 10 femtograms of linearized DNA was injected into ova from C3H/C57B6 mice.

Founder mice were crossed to C57B6 strain mice and the hair follicles of resultant progeny were assessed for RFP expression using fluorescence microscopy. Two founders (5403 and 5396) were shown to possess RFP fluorescence in hair and hair follicles. All genotyping of αSMA-RFP, COLL-EGFP or αSMA-RFP/COLL-EGFP mice was conducted by observing fluorescence in a small tail-snip (≈2-3 mm).

Characterization of αSMA-RFP Transgenic Mice.

Cellular expression of the αSMA-RFP transgene was colocalized with endogenous αSMA using fluorescence microscopy and immunohistochemistry. Briefly, mice were euthanized by cervical dislocation, the small bowel was resected and placed in 4% paraformaldehyde (PFA) for ≈16 hours. After, tissues were washed twice in phosphate buffered saline (PBS) and placed in 70% ethanol prior to paraffin embedding. To detect RFP, 5 μm paraffin sections were deparaffinized, mounted in glycerol-based medium and RFP fluorescence was captured using fluorescence microscopy. To detect endogenous αSMA, the covers slip was removed, the section was washed extensively in PBS to remove glycerol and the same section was stained with an anti-αSMA monoclonal antibody (Dako, Carpinteria, CA) as previously described.18 Images of endogenous αSMA staining were captured using visible microscopy. For coexpression studies of EGFP and DsRED, all tissues were fixed in 4% PFA, followed by incubation in 30% sucrose-PBS for 24 hours and cryosectioning.

Primary HSC Isolation.

HSCs were isolated from 12- to 16-week-old male mice.18 HSC purity, as estimated by the autofluorescence of the cells by ultraviolet-excited fluorescence microscopy, was between 98%-99%. Cells were seeded on uncoated plastic tissue culture dishes and cultured in Dulbecco′s modified eagle medium (DMEM; GibcoBRL, Grand Island, NY) supplemented with 10% fetal calf serum, 2 mM L-glutamine and standard antibiotics.

Experimental Models of Liver Injury.

Liver injury was induced either by CCL4 administration or bile duct ligation. CCL4 was administered in a single intraperitoneal dose (2.0 μL/g body weight; 1:1 dilution with corn oil). Control mice were treated with corn oil alone (2.0 μL/g body weight; diluted 1:1 in PBS). Liver samples were collected and total RNA was isolated using guanidinium isothiocyanate/CsCl method as previously described.19 To induce liver fibrosis, bile duct-ligation was performed essentially as described.20 Mice were sacrificed 14 days after surgery, liver samples collected and fixed in fixed in 4% PFA for 10-24 hours, washed twice in PBS, and transferred to vials containing 30% sucrose (made in PBS) for at least 24 hours or until cryostat sectioning. All animal procedures were approved by the Investigation and Ethics Committee and Institutional Animal Care and Use Committee of the University of North Carolina.

Microscopy Studies.

RFP and EGFP fluorescence in live primary cultures were assessed using an Olympus IX70 (Olympus, Melville, NY) fitted with EGFP-specific (Omega Optical, XF116-2, Brattleboro, VT) and RFP-specific filters (Omega Optical, XF137-2). Images were captured using a digital SPOT camera (Diagnostic Instruments Inc., McHenry, IL). Identical exposure times were used for each data point within an individual experiment. Confocal laser-scanning microscopy (Zeiss LSM-510, Carl Zeiss, Oberkochen, Germany) was used to assess RFP and EGFP fluorescence in livers of 14-day bile duct ligated (BDL) mice. Ten micron cryostat sections were mounted in glycerol based medium on glass slides and 1-3 μm optical sections were captured using RFP and EGFP-specific filters. Images represent overlays of scanned sections. Collagen was detected in the livers of bile duct ligated mice by Sirius red staining followed by counterstaining in methyl green.21

Flow Cytometric Analysis and Sorting.

To assess expression of RFP and EGFP during primary culture HSC activation, HSCs were isolated and grown on plastic tissue culture dishes for 5 and 10 days without splitting.18 At the specified times, HSCs were dislodged from the plates using trypsin/EDTA, washed once with growth medium, once with PBS, and than placed in tubes containing flow cytometric analysis solution (PBS containing 0.1% BSA and 0.01% azide). Fluorescence in live cells was measured within 30 minutes of isolation on a FACscan (Becton-Dickinson, Mountain View, CA) using the FL1 channel to detect EGFP fluorescence and FL2 channel to detect RFP. HSCs from four αSMA-RFP/COLL-EGFP mice were sorted 14 days after single plating on plastic tissue culture dishes using a MOFLO FACSorter. Cells were sorted based on RFP-only, EGFP-only, or RFP/EGFP expression profiles. Total RNA from sorted HSCs was isolated within 30 minutes post-sort using RNAqueous-Micro (Ambion, Austin, TX), and a standard reverse transcription reaction was conducted to generate cDNA.

Northern Blot and Semiquantitative RT-PCR.

To compare endogenous αSMA and transgenic αSMA-RFP expression, total RNA was isolated from whole liver following intraperitoneal treatment with CCL4 as described earlier. As a control, total RNA was also isolated from 14-day activated HSCs using Trizol (Gibco BRL).18 Northern blot was used to detect endogenous αSMA and transgenic αSMA-RFP expression. Ten micrograms of liver RNA and 2 μg of HSC RNA were loaded onto a MOPS/formaldehyde gel and transferred by capillary action in 20X SSC onto a nylon membrane. Random primed P32-labelled cDNA probes were generated using the αSMA or DsRED1-1 cDNAs. The probes were added to the prehybridized membranes at 1 × 106 counts/mL in Rapid-Hyb buffer (Amersham, Piscataway, NJ) and incubated at 65°C for 2-4 hours. The membranes were washed 4 times in 0.1X SSC/0.1% SDS at 60°C and then exposed to a phosphor imaging screen for 8-16 hours. Images were captured on a Typhoon Imager (Amersham).

Total RNA from FACS-sorted cells was isolated as already described and all samples were resuspended in a total of 25 μL of DEPC-treated water. One to 2 μL of each RNA sample were used in a standard (according to product insert) real-time polymerase chain reaction (PCR) reaction using a LightCycler real-time PCR analyzer and LightCycler cyber green reagents (Roche, Bassel, Switzerland). Each reaction contained 3 mM MgCl2 and was cycled at 97°C, 10 seconds; 60°C, 5 seconds; 72°C 30 seconds for 40 cycles. Linear amplification ranges were determined using LightCycler analysis software and reaction crossing points were applied to the formula described by Pfaffl.22 Primers sequences are as follows:
















The αSMA-RFP Transgene Recapitulates Endogenous αSMA Expression.

To generate the αSMA-RFP transgenic mouse, a 3.1-kb fragment of the vascular αSMA promoter/enhancer17 was placed 5′ of the RFP reporter gene cDNA (Fig. 1A). This fragment of the αSMA promoter contains an intron 1 CarG-box that is required for full tissue-specific expression in an αSMA-LacZ transgenic mouse.23 Ova injection of the αSMA transgene produced 16 founders of which 8 contained the RFP cDNA. A PCR strategy confirmed that of the 8 founders that possessed the RFP cDNA, only 4 contained the 5′ end of the transgene. RFP expression from tail-snips indicated that only two of these 4 founders (5396 and 5403) had visibly detectable levels of RFP expression as assessed by fluorescence microscopy. Organ-specific expression of the αSMA-RFP transgene indicated the transgene was highly expressed in the small intestine, bladder, stomach, vasculature and epididymis24, 25 where endogenous αSMA is expressed (Fig. 1B). Further analysis at the cellular level demonstrated that RFP was co-localized with endogenous αSMA in the muscularis layer of the small intestine (Fig. 1C).

Figure 1.

Transgenic DNA contstructs and validation of the αSMA-RFP transgene. (A) Schematic of DNA constructs that were used to generate the αSMA-RFP mouse (top) and the COLL-EGPF mouse (bottom). The αSMA-RFP construct contains exons 1 and part of exon 2 from the αSMA gene (hatched boxes). (B) αSMA-containing organs from the αSMA-RFP mouse were resected and immediately assessed for RFP expression using fluorescence microscopy. Organs from a transgene negative littermate was used as a control. (C) Coexpression of the αSMA-RFP transgene and endogenous αSMA was determined by the observation of RFP and αSMA in the muscalaris layer of the small intestine. (D) Northern blot demonstrates the concomitant inducibility of the αSMA-RFP transgene with the endogenous αSMA gene in CCL4 treated αSMA-RFP mice.

We next assessed the inducibility of the αSMA-RFP transgene in the liver. Endogenous αSMA expression has been shown to be restricted to a minority of cells in the vasculature of the normal liver, while acute liver injury (using CCL4) has been shown to induce high levels of αSMA protein in activated myofibroblasts at 48 hours to 72 hours posttreatment.26 In the αSMA-RFP mouse, CCL4 administration induced concomitant expression of RFP mRNA at 48 and 72 hours that closely correlated with endogenous αSMA mRNA expression (Fig. 1D).

αSMA and Collagen α1(I) Are Expressed in a Similar Temporal, But Heterogeneous Manner in Primary Cultures of HSCs.

To assess the temporal expression patterns of αSMA and collagen α 1(I), mice were generated that contained both the αSMA-RFP and COLL-EGFP transgenes (αSMA-RFP/COLL-EGFP). To determine the time course of αSMA and collagen α 1(I) expression during culture activation of HSCs, cells were isolated and purified to >99% homogeneity (as determined by retinoid autofluorescence), and allowed to activate on plastic for 14 days. HSCs are heterogenous with regard to the vitamin A content.27–31 However, the functional significance of this heterogeneity is unclear. It is possible that our isolation procedure to prepare HSC could result in the selection of cells with higher vitamin A content. The procedure is based on the fact that vitamin A-rich cells are light and can be easily separated using a discontinous gradient. An increase in the gradient density would have yielded a higher number of HSCs but with more contaminating cells. In order to obtain pure HSCs cultures, a low density gradient was used. Both the αSMA-RFP and COLL-EGFP transgenes were expressed in few individual cells upon plating (Fig. 2). Fluorescent microscopy analysis at day 0-1 showed small numbers of cells expressing both RFP and EGFP, while most cells expressed either RFP or EGFP. This observation was more evident as the cells began to proliferate from days 2-4 (Fig. 2).

Figure 2.

Temporal expression profiles of the COLL-EGFP and αSMA-RFP transgenes in highly purified cultures of HSCs. (A) HSCs were isolated from a COLL-EGFP/ αSMA-RFP mouse using a discontinuous gradient. Retinoid content confirmed that >99% of the cells were HSCs. Transgene expression was followed over a 14-day period using fluorescence microscopy. Each column represents the same field of cells. Images in each column were digitally captured using either EGFP- or RFP-specific filters. Images were overlayed (bottom row) to assess coexpression. The data are representative of at least 3 independent experiments.

To more accurately assess the expression of the RFP and EGFP reporter genes in culture-activated HSCs, FACS was used to quantify reporter gene expression in nonpassaged cells at days 5 and 10 (Fig. 3). HSCs from mice containing only the αSMA-RFP transgene, only the COLL-EGFP transgene, or both transgenes were isolated to >99% homogeneity based on retinoid content which could be observed in all cells at day 5 (Fig. 3A, blue spots). By day 10, HSCs had activated to the point that most retinoid-rich droplets had been lost (Fig. 3A). At day 5 of culture activation, 50% of the HSCs did not express either transgene, while 7% expressed αSMA-RFP, 14% expressed COLL-EGFP and 30% expressed both transgenes (Fig. 3B). Further culture until day 10 demonstrated that an additional 25% of the negative cells begin to express one or both of the transgenes. Thus, there was an increase in the number of cells expressing both the αSMA-RFP and COLL-EGFP transgenes (Fig. 3B). Interestingly, the percentage of cells expressing only the αSMA-RFP or COLL-EGFP transgenes remained almost constant (9% RFP, 14% EGFP) between days 5 and 10 (Fig. 3B).

Figure 3.

Three unique sub-populations of HSCs are identified using fluorescence microscopy and flowcytometry. (A) HSCs were isolated from COLL-EGFP, αSMA-RFP and COLL-EGFP/ αSMA-RFP mice and culture-activated for either 5 or 10 days. Expression of transgenes was assessed using fluorescence microscopy. Retinoid droplets (blue) were imaged using UV excitation wavelengths and DAPI blue emission filters. Images are 200× original. (B) The same cells depicted in 3A were dissociated and subjected to flowcytometry. HSCs from COLL-EGFP-only or αSMA-RFP-only mice were used to calibrate the flowcytometer (columns left and middle). Transgene status of cells in each quadrant is indicated by colored box. The data are representative of at least 3 independent experiments.

Populations of Culture-Activated HSCs Demonstrate Different Gene Expression Profiles.

To further characterize the molecular characteristics of each population of HSCs, expression of HSC-marker genes was assessed in the three populations of HSCs using real-time PCR analysis. HSCs were isolated from αSMA-RFP/COLL-EGFP mice, activated on plastic for 10 days, and FACS-sorted based on transgene expression profiles (non-expressing, EGFP-expressing, RFP-expressing or EGFP/RFP-expressing) (Fig. 4A). The sorted HSCs were replated on plastic and cultured for an additional 4 weeks. Changes in transgene expression profiles were documented using fluorescence microscopy at 3-4 day intervals. Twenty-four hours after replating, the αSMA-RFP-expressing HSCs still expressed only RFP, however, approximately 25% of the COLL-EGFP–expressing cells began to also express RFP (not shown). After 5 days, approximately 90% of the transgene negative HSCs demonstrated transgene expression profiles similar to the original cultures, (i.e., RFP-only, EGFP-only, RFP/EGFP; not shown). After 4-weeks in culture, the HSCs that were sorted based on αSMA-RFP expression, still exclusively expressed the αSMA-RFP transgene. By contrast, the HSCs sorted based on COLL-EGFP transgene expression demonstrated transgene expression profiles that were similar to the original cultures (i.e., RFP-only, EGFP-only, RFP/EGFP), demonstrating the plasticity of the EGFP+ cells by the expression of both transgenes.

Figure 4.

Subpopulations of HSCs demonstrate different HSC-marker gene expression profiles. HSCs were sorted by FACS based on transgene status, cDNA was generated, and HSC-marker genes were assessed by real-time PCR. (A) HSCs were isolated from COLL-EGFP/ αSMA-RFP mice, culture-activated for 10 days, and subjected to FACS. Sort-gates for each population are indicated by black boxes and transgene status is indicated by a colored square (top left histogram). (B) cDNA was made from sorted cells and subjected to real-time PCR analysis for synaptophysin, reelin, GFAP, MMP2, MMP13, TIMP1 and ICAM-1 expression. PCR analysis for each gene was done in dupilcate (P < .001 for each gene examined).

To assess differential gene expression in the RFP-only, EGFP-only, and RFP/EGFP-expressing populations of culture activated HSCs, real-time PCR was used. Neural-specific markers glial fibrillary acidic protein (GFAP), synaptophysin, reelin; matrix metalloproteinase gene expression (metalloproteinase; MMP-2 and -13); TIMP1 (tissue inhibitor of metalloproteinase), and ICAM-1 (intercellular adhesion molecule) were assessed from the freshly FACS-sorted HSCs described above (10-day culture activated). Although each population expressed all genes examined, HSCs exclusively expressing the αSMA transgene demonstrated significantly higher levels of reelin, synaptophysin, MMP-13, TIMP1 and ICAM-1 mRNA compared to HSCs exclusively expressing the collagen α1(I) transgene (Fig. 4B). Conversely, there was higher MMP-2 gene expression in HSCs exclusively expressing collagen α1(I) compared to cells exclusively expressing αSMA (Fig. 4B). There was no significant difference in the GFAP mRNA levels. The predominant, coexpressing yellow cells had characteristics of classic activated HSCs with expression of both collagen, αSMA, and the neural markers synaptophysin, reelin and GFAP. These expression studies further demonstrate the distinct molecular characteristics of each population.

Expression of Transgenes During Cholestasis Induced Fibrosis.

To extend the gene expression studies during active in vivo hepatic fibrogenesis, we performed a pilot study to determine the pattern of expression of αSMA and collagen α1(I) using the well-characterized model of bile duct ligation. We first assessed extracellular matrix (ECM) accumulation 2 weeks after surgery as demonstrated by Sirius red staining (Fig. 5A). As expected, sham-operated mice did not show ECM accumulation in the hepatic parenchyma, and a positive staining was only observed in vessel walls (not shown). By contrast, bile duct ligated mice showed marked ECM deposited around proliferating bile ducts and bridging fibrosis (Fig. 5A). At high magnification, ECM was identified surrounding peribiliary fibrogenic cells (Fig. 5A, left).

Figure 5.

Anaylysis of transgene expression during active fibrosis. To analyze transgene expression in vivo, COLL-EGFP/ αSMA-RFP mice were subjected to bile duct ligation (n = 3). Fourteen days post-surgery, livers were assessed for ECM by (A) Sirius Red staining. Left: Small arrows denote parenchymal fibrogenic cells expressing ECM. Middle: White box surrounds highly fibrotic region surrounding vessel or parenchyma which are represented as high magnification images to the left and right. Small arrows indicate proliferating bile ductules. Right: High-magnification image of peribiliary/perivenual myofibroblasts expressing collagen. (B) Transgene expression surrounding vessels and bile ductules in BDL mice. From left to right: White boxes represent region of high magnification image to the right. (C) White boxes indicate regions of high magnification images (to left and right) of parenchymal myofibroblasts.

To investigate both αSMA and collagen α1(I) gene expression in vivo, αSMA-RFP and COLL-EGFP transgene expression was assessed in COLL-EGFP/ αSMA-RFP mice 14 days after surgery using confocal microscopy. In sham-operated mice, only vessel walls expressed both αSMA-RFP and COLL-EGFP, while no transgene expression was detected in the liver parenchyma (not shown). In bile duct ligated mice, endothelial cells in central veins expressed αSMA-RFP only (Fig. 5B) consistent with the up-regulation of αSMA in porcine serum-induced models of liver fibrosis.32 Myofibroblasts accumulating around proliferating bile ducts predominantly expressed the COLL-EGFP transgene (Fig. 5B). Finally, perisinusoidal fibrogenic cells expressed either COLL-EGFP alone, or coexpressed αSMA-RFP and COLL-EGFP (Fig. 5C), consistent with activated HSCs. Transgene expression was not detected in hepatocytes, supporting the paradigm that parenchymal cells are not a source of collagen in liver fibrogenesis. These results indicate that different cell populations express αSMA and collagen α1(I) in livers undergoing active fibrogenesis.


The current study investigates the expression of two key genes in liver fibrogenesis, αSMA and collagen α1(I), in fibrogenic cell populations by using a double reporter gene transgenic mouse. While αSMA protein expression is commonly used to assess the presence of fibrogenic cells in the injured liver, collagen α1(I) is the main extracellular matrix protein accumulated in the fibrotic liver.4, 13 We provide evidence in vitro and in vivo that there is heterogeneity in HSCs/myofibroblasts with regard to gene expression. Primary HSCs were cultured on plastic, which is a widely accepted model to study activation of quiescent HSCs into activated myofibroblastic cells.6 After prolonged culture, most HSCs express both αSMA and collagen α1(I) simultaneously, while others exclusively express either collagen α1(I) or αSMA. Moreover, we have investigated the expression profile of αSMA and collagen α1(I) genes in vivo by submitting the double transgenic mice to a model of secondary biliary fibrosis. Similar to what was demonstrated in culture, both transgenes were differentially expressed in myofibroblastic cells, while no transgene expression was detected in hepatocytes.

To study the temporal profile of gene expression in hepatic fibrogenic cells in vitro, we examined primary cultures of HSCs. An important finding of our study is that there is heterogeneity in the expression profile of αSMA and collagen α1(I) transgenes in cultured HSCs. After prolonged culture, most cells coexpressed both genes, while others expressed either αSMA or collagen α1(I). This finding is relevant, since we demonstrate that HSCs can express collagen α1(I) in the absence of αSMA. To investigate whether this finding is due to HSCs heterogeneity or to the existence of contamination by liver fibroblasts, specific neural markers were assessed.30 All three transgenic cell populations expressed specific HSC markers such as synaptophysin, reelin, and GFAP, suggesting that each population was indeed HSC-derived. Interestingly, the level of expression of these markers varied depending on the αSMA/collagen α1(I) expression profile. Cells expressing αSMA-only or both αSMA/collagen α 1(I) showed higher expression levels of synaptophysin and reelin. By contrast, COLL-EGFP-only expressing cells expressed lower levels of these neural markers. Additionally, the genes MMP-2, MMP-13, TIMP1 and ICAM-1 demonstrated differential expression profiles between the three different populations of HCSs. These data suggest αSMA and collagen expression profiles of culture-activated HSCs can mark differences in transcriptional profiles of other HSC-expressed genes and perhaps can be used to define subpopulations of HSCs.

HSCs display marked heterogeneity in morphology based on their zonal location in the hepatic lobule.28, 31, 33, 34 Cells in the centrolobular zone are conspicuously dendritic with longer processes in comparison to those in the periportal zone. Vitamin A storage is well developed in zones 1 and 2, but reduces gradually toward the central region. Immunostaining of neural crest markers is stronger in the periportal zones than that in the centrolobular zones.35–38 Whether the morphological/phenotypical heterogeneity is associated with differences in cell function is unclear.

To begin to delineate the relationship between morphological/phenotypical heterogeneity and cell function, we assessed αSMA/collagen α1(I) expression in a model of cholestatic fibrosis. This bile duct ligation model reproduces the changes observed in patients with chronic cholestatic diseases.7 It is important to point out that while αSMA-RFP was highly expressed in culture-activated HSCs, its expression in vivo was low. In contrast, the COLL-EGFP transgene was markedly expressed both in culture and in vivo. Because αSMA protein is expressed in myofibroblasts in experimental liver fibrogenesis, it is conceivable that the αSMA-RFP transgene is not transactivated in vivo as efficiently as in culture. Although the integration status of the transgene could play a role in expression differences of the transgene in vitro versus in vivo, two different αSMA-RFP founder lines demonstrated identical tissue expression patterns indicating the transgene position likely does not influence transgene expression behavior. Perhaps additional cis-elements required to respond specifically to in vivo stimuli of liver injury are not included in the promoter/enhancer fragment used to generate the αSMA-RFP mouse.

Despite the limitations of transgenic model systems, several conclusions can be drawn based on our pilot in vivo study. First, collagen α1(I) expression is scarce in the normal liver, while its expression during liver fibrogenesis is very high. Second, parenchymal cells (i.e., hepatocytes) are not a source of collagen α1(I) in the fibrotic liver. Third, fibrogenic cells accumulating around small vessels and proliferating bile ducts, as well as activated HSCs, are a major source of collagen α1(I). And finally, fibrogenic cells can express αSMA in the absence of collagen α1(I) expression and vice versa.

An important conclusion of the current study is that αSMA and collagen α1(I) are not always coexpressed in fibrogenic cell types in vitro or in vivo. This finding has important implications, since the expression of αSMA by immunohistochemistry is considered the gold-standard method to detect fibrogenic cells in human and rodent liver specimens. Moreover, the quantification of the number of αSMA positive cells in the liver is commonly used as a marker of active fibrogenesis.2 Based on this data, we propose to reconsider the current paradigm that αSMA staining reflects the amount/activity of fibrogenic cells in the liver. Specific data demonstrating collagen expression in the liver tissue should also be provided in studies assessing liver fibrogenesis.