Potential conflict of interest: Nothing to report.
Hepatic stellate cells play an essential role in the liver's injury response. Although stellate cells are defined by the presence of cytoplasmic protrusions, the function of these characteristic structures has been obscure. We hypothesized that stellate cell protrusions act by coupling injury-associated stimuli to chemotaxis. To test this hypothesis, we developed an assay for directly visualizing the response of living stellate cells in early primary culture to local stimulation of the tips of protrusions with platelet-derived growth factor-BB (PDGF). Stellate cells exhibited elongate protrusions containing actin, myosin, and tubulin. PDGF, but not cytochrome C, localized at a protrusion tip induced a coordinated series of morphological events—cell spreading at the tip, movement of the cell body toward the PDGF, and retraction of trailing protrusions— that resulted in chemotaxis. Soluble PDGF and AG 1296, a receptor tyrosine kinase inhibitor, both reduced stellate cell chemotaxis. PDGF-induced chemotaxis was associated with an early and transient increase in myosin phosphorylation within the spreading lamella. We observed that blebbistatin, a myosin II inhibitor, completely and reversibly blocked protrusion-mediated lamella formation and chemotaxis. Moreover, blockade of MRLC phosphorylation with the myosin light chain kinase inhibitor, ML-7, or the rho kinase inhibitor, Y-27632, blocked lamella formation, myosin phosphorylation within the protrusion, and chemotaxis. Conclusion: These results support a model in which protrusions permit stellate cells to promptly detect PDGF distant from their cell bodies and transduce this signal into mechanical forces that propel the cell toward the site of injury. (HEPATOLOGY 2007.)
Hepatic stellate cells mediate the liver's response to injury and the development of cirrhosis.1, 2 Growing evidence supports a model in which injury-associated stimuli induce a series of stellate cell responses.3–5 An early event in this model is the chemotaxis of stellate cells to damaged areas of the liver, where they proliferate, synthesize extracellular matrix, and participate in tissue repair. If injury resolves, then stellate cells undergo apoptosis once repair has been completed. In contrast, if injury persists, stellate cells drive the development of fibrosis and subsequent cirrhosis. Although the regulation of stellate cell responses has been intensively studied, the cellular mechanisms through which injury-associated stimuli are coupled to these responses are incompletely understood.
A distinguishing characteristic of stellate cells is that they possess multiple elongate protrusions, which encircle the sinusoids.6 Indeed, stellate cells derive their name from the star-shaped arrangement of these protrusions that radiate outward from the cell body. The function of these distinctive structures has been a mystery. The leading supposition has been that stellate cell protrusions act by contracting around the sinusoid, thereby modulating sinusoidal blood flow. This idea, however, is based entirely on circumstantial evidence.7–9 We are unaware of any studies that have directly investigated the role of stellate cell protrusions.
The aim of this study was to test the hypothesis that stellate cell protrusions act by coupling injury-associated stimuli to chemotaxis. We addressed this aim by (1) characterizing stellate cell protrusions; (2) demonstrating that protrusions can detect localized concentrations of platelet-derived growth factor (PDGF); and (3) establishing that protrusions are capable of transducing this chemical signal into mechanical forces that drive chemotaxis.
Polystyrene-latex carboxylate beads (20 μm diameter, Bangs Laboratories) were coated with PDGF-BB (Biosource International) or cytochrome C type IV (Sigma) using a kit (Carbodiimide Kit for Carboxylate Beads, Polysciences Inc.).10 Quantikine PDGF-BB ELISA (R&D Systems) and a protein microassay (Bio-Rad) were used to measure PDGF and cytochrome C, respectively.
Stellate cells, isolated from the livers of Sprague-Dawley rats11, 12 were grown on coverslips at 37°C. Animals were treated per the NIH Guide for the Care and Use of Laboratory Animals and as approved by the UCSF Animal Research Committee. After 2 days in primary culture, coverslips with attached stellate cells were transferred to a chamber (37°C) containing serum-free HEPES-buffered physiological solution. PDGF or cytochrome C coated beads were immediately placed onto stellate cells. In some experiments, soluble PDGF (Calbiochem), AG 1296 (Calbiochem), Y-27632 (Calbiochem), ML-7 (Sigma), or blebbistatin (Toronto Research Chemicals) was added to the solution at the beginning of the experiment. Phase-contrast images were acquired using a 10× objective (Ach 0.25 NA, Olympus) on an inverted microscope (IX70, Olympus) every 2 minutes with a cooled-CCD camera (Sensys, Roper Scientific).
Quantification of Morphological Changes.
Distinct morphological alterations that occur during stellate cell chemotaxis were quantified from the captured time-lapse images (Metamorph v. 6.1, Universal Imaging). Protrusion spreading was measured by determining the surface area of a line around the plasma membrane edge of a protrusion in contact with a PDGF-coated bead (Metamorph and Graphire, Wacom Co.). The nucleus-bead distance was determined by measuring the length of a straight line between the center of the PDGF-coated bead and the center of the cell nucleus. Trailing protrusion retraction was determined by measuring the change in the length of a straight line between the tips of trailing protrusions and their initial location at the time the bead was added. Trailing protrusions were defined as protrusions with a length of at least 40 μm that radiated at a 120° to 210° angle away from the protrusion on which the bead was placed.
Cells were fixed with 4% paraformaldehyde in PBS for 15 minutes, permeabilized in 0.1% Triton X-100 in PBS for 5 minutes, blocked for 16 hours in PBS containing 2% BSA, and then incubated for 2 hours in PBS-BSA with one or two of the following antibodies: anti-myosin regulatory light chain (MRLC; clone MY-21, Sigma), anti-phosphorylated MRLC (see below), or anti-α-tubulin (clone DM1A, Sigma). An antibody directed against the phosphorylated serine-19 residue of MRLC (NCBI accession #s: P19105, P18666, A61034) was collected from a rabbit immunized with a synthetic phospho-peptide, NH2-Arg-Pro-Gln-Arg-Ala-Thr-SerPO4-Asn-Val-Phe-Ala-Cys-COOH (UCLA CURE Antibody/Radioimmunoassay Core). Primary antibody was fluorescently labeled with a secondary antibody (Molecular Probes). F-actin was stained with rhodamine-phalloidin (Molecular Probes). Images were acquired using a 40× objective (UApo/340 1.35 NA, Olympus) and appropriate filters (Chroma).
Measurement of MRLC Phosphorylation.
Total and phosphorylated MRLC staining was performed on cells grown on coverslips with grids (Electron Microscopy Sciences). These grids allowed identification of cells previously assayed for chemotaxis, even after PDGF-coated beads had washed off. Cells were co-labeled with the rabbit antibody directed against serine-19 phosphorylated MRLC and the mouse antibody directed against total MRLC as described. Cells were next incubated with species-specific fluorescent secondary antibodies (rabbit-Alexa Fluor 488 and mouse-Alexa Fluor 546, Molecular Probes) to distinguish phosphorylated from total MRLC. Images were acquired using identical microscope, camera, and filter settings for each antibody. Background signal was subtracted from each image. Using a graphics tablet and Metamorph, regions were created by drawing lines around the protrusion touching the bead, the cell body, and the trailing protrusions for each cell. In each region, the fraction of phosphorylated MRLC to total MRLC was determined by dividing the average fluorescence signal of phosphorylated MRLC labeling by that of total MRLC.
We employed hepatic stellate cells in early primary culture, an intensely studied model system, in which stellate cells closely resemble stellate cells in hepaticus in terms of morphology, protein expression, and behavior.3 Cells exhibited 4.2 ± 1.4 (SEM) protrusions per cell (n = 18 cells) that radiated away from the cell body in a symmetrical manner (Fig. 1). The base to tip dimension (i.e., length) of the protrusions was 62.3 ± 26.4 μm (n = 32 protrusions), whereas the midlength width of the protrusions was 6.5 ± 3.2 μm (n = 32 protrusions). Staining with anti-MRLC, anti-α-tubulin, and phalloidin revealed the presence of both actomyosin and microtubule filaments in each protrusion (Fig. 1). Thus, stellate cell protrusions represent distinct morphological structures that display a characteristic size and shape, and contain both actomyosin and microtubule cytoskeletons.
Local Stimulation of Protrusions with PDGF Triggers Receptor-Mediated Chemotaxis.
To investigate whether stellate cells can detect chemoattractants localized to the tips of their protrusions, PDGF and cytochrome C-coated beads were synthesized. PDGF-coated beads bound 20 ± 2 pg PDGF per bead. Cytochrome C-coated beads bound 48 ± 14 pg cytochrome C per bead. Control experiments showed that during 2 hours in solution more than 99.6% of the total PDGF-BB remained bound to the beads.
We tested whether local stimulation of protrusions with PDGF could be detected by stellate cells using time-lapse microscopy. Within minutes of coming into contact with a protrusion, a bead coated with PDGF stimulated cell movement toward the contact site (Fig. 2A and Supplemental Video 1). A series of morphological events were triggered by contact between a PDGF-coated bead and a protrusion. First, we observed membrane ruffling at the contact site and spreading of the plasma membrane and cytoplasm of the protrusion adjacent to the bead (Fig. 2A and Supplemental Video 1). As spreading advanced, the protrusion formed lamella-like structures flanking the bead (Fig. 2A and Supplemental Video 1). Next, the nucleus and cytoplasm of the cell body were found to move simultaneously toward the bead (Fig. 2A and Supplemental Video 1). Finally, protrusions at the trailing end of cells retracted (Fig. 2A and Supplemental Video 1). These results suggest that stellate cells can detect PDGF localized to the tip of a protrusion and initiate chemotaxis.
In every experiment (n = 27) in which a PDGF-coated bead touched a protrusion, the stellate cell moved toward the bead (Fig. 2C). When a PDGF-coated bead was within 10 μm of a protrusion, but not actually touching, the cell never moved toward the bead (n = 20; data not shown). When a cytochrome C-coated bead was placed in contact with a protrusion (n = 21), membrane ruffling never occurred at the contact site and the cell never moved toward the bead (Fig. 2B and Supplemental Video 2). We confirmed that protrusion-mediated chemotaxis occurred in specific response to PDGF by showing a dose-dependent and saturable inhibition by soluble PDGF (Fig. 2C). Soluble PDGF enhanced undirected stellate cell movement as illustrated by the dose-dependent increase in movement of cells away from the PDGF-coated beads (Fig. 2C). Similarly, soluble PDGF increased undirected cell movement in experiments in which cytochrome C-coated beads were used (Fig. 2C). We verified that chemotaxis toward PDGF-coated beads is mediated through PDGF's cognate receptor tyrosine kinase using AG 1296, a selective inhibitor of receptor tyrosine kinase autophosphorylation.13 AG 1296 inhibited chemotaxis in response to PDGF-coated beads in a dose-dependent manner (Fig. 2C). Our observations that stellate cell chemotaxis did not occur with cytochrome C-coated beads, and was inhibited by soluble PDGF and AG 1296, demonstrate that protrusions detect localized PDGF at their tips though specific ligand-receptor binding.
Distinct and Coordinated Alterations in Morphology Drive Protrusion-Mediated Chemotaxis.
To further characterize the morphological events provoked by PDGF binding to cognate receptors on the tips of protrusions, we quantified changes in protrusion surface area, nucleus-bead distance, and trailing protrusion retraction. Eight minutes after contact with a PDGF-coated bead, protrusion surface area began to increase (Fig. 3). The protrusion spread 45.5 μm2/min until 30 minutes after bead contact, reaching a maximal size of approximately 1,250 μm2. Ten minutes after protrusion-bead contact, the cell nucleus started moving 0.94 μm/min toward the bead (Fig. 3). This continued for nearly 32 minutes until the cell body touched the bead. Spreading of the protrusion and movement of the cell body toward a PDGF-coated bead resulted in the disappearance of the original elongate protrusion. Protrusions at the trailing edge of the cell began to retract toward the cell body 12 minutes after initial contact with the bead (Fig. 3). Trailing protrusions retracted 0.26 μm/minute and ceased retracting at 50 minutes. Our observations indicate that detection of PDGF by protrusions induces a coordinated sequence of specific changes in morphology that permits chemotaxis. These results also suggest the possibility that protrusions act by transducing PDGF-binding at the protrusion into mechanical forces that drive chemotaxis.
Protrusion-Mediated Chemotaxis Depends on Myosin II Activation Within the Lamella.
To test whether protrusions act as transducers that convert chemical signals into mechanical forces, we investigated the potential role of myosin II in protrusion-mediated chemotaxis. We employed the selective myosin II inhibitor, blebbistatin,14 which when added at the time of PDGF-bead placement completely abrogated the morphological changes associated with protrusion-mediated chemotaxis (Fig. 4 and Supplemental Video 3). On removal of blebbistatin, cells proceeded to move toward the PDGF-coated bead (Fig. 4 and Supplemental Video 3). This result suggested that stellate cell chemotaxis in response to PDGF detected by protrusions requires activation of myosin II.
The role and location of myosin II, which is activated when its regulatory light chain is phosphorylated at serine 19,15 were further investigated by quantifying the fraction of phosphorylated MRLC to total MRLC (MRLC-P:MRLC) at different sites (i.e., lamella, cell body, and trailing protrusions) within stellate cells undergoing chemotaxis. MRLC-P:MRLC was determined at time points corresponding to the beginning of lamella formation (7.5 minutes), the beginning of nuclear movement and trailing protrusion retraction (15 minutes), the end of new lamella formation (25 minutes), and the end of nuclear movement and trailing protrusion retraction (50 minutes). MRLC-P:MRLC in the protrusions and cell body of nonpolarized stellate cells before placement of a PDGF-coated bead was similar, at approximately 0.5 (Figs. 5 and 6). Within 7.5 minutes of PDGF-coated bead placement at the tip of a protrusion, MRLC-P:MRLC within the lamella (i.e., the protrusion contacting the bead) began to increase (Figs. 5 and 6). MRLC-P:MRLC at the newly formed lamella doubled to 1.0 by 15 minutes before decreasing to near baseline levels at 25 minutes after initial contact with the bead (Figs. 5 and 6). In contrast, MRLC-P:MRLC within the cell body remained near 0.5 throughout the entire course of chemotaxis (Figs. 5 and 6). Conversely, within the trailing protrusions MRLC-P:MRLC began to increase 7.5 minutes after placement of a PDGF-coated bead at the tip of a protrusion (Figs. 5 and 6), and remained elevated throughout the 50-minute course of chemotaxis (Figs. 5 and 6). These data suggest that discrete regulation of myosin activity, at different times and locations within the cell, direct the morphological alterations that underlie stellate cell chemotaxis.
To determine whether myosin II powers protrusion-mediated chemotaxis, we modulated the activity of the two key regulators of myosin II activity: myosin light chain kinase, which directly phosphorylates MRLC, and rho-associated kinase, which inhibits myosin phosphatase.15 ML-7, which has previously been shown to inhibit MRLC phosphorylation in rat stellate cells,16 abolished PDGF-coated bead-induced MRLC phosphorylation in all regions of the cell (Fig. 7A). ML-7 also blocked lamella formation and prevented stellate cell movement toward PDGF-coated beads as measured by the absence of an increase in surface area spreading and the absence of cell nucleus-bead distance shortening, respectively (Fig. 7B,C). The rate of retraction of trailing protrusions increased with ML-7 treatment, but this inhibitor also induced retraction of protrusions even amongst cells not touching PDGF-coated beads (data not shown). Y-27632, at a concentration that we have previously shown inhibits MRLC phosphorylation in stellate cells,17 abolished PDGF-coated bead-induced MRLC phosphorylation in all regions of the cell (Fig. 8A). Y-27632 also blocked lamella formation, prevented cell nucleus movement toward beads, and decreased retraction of trailing protrusions by 23% in response to placement of a PDGF-coated bead on a protrusion (Fig. 8B,C). These results demonstrate that protrusions are capable not only of detecting PDGF, but also of transducing this chemical signal into myosin II-powered mechanical forces that drive the morphological events comprising stellate cell chemotaxis.
These results support our hypothesis that stellate cell protrusions act by coupling injury-associated stimuli to chemotaxis. Moreover, our results provide the first direct evidence to identify a specific functional role for the characteristic protrusions exhibited by stellate cells. First, our data show that stellate cells early in primary culture extend distinct symmetrical protrusions, similar to those seen in situ, that contain structural (e.g., f-actin and α-tubulin), motor (e.g., myosin), and signaling (e.g., PDGF receptors) elements requisite for cellular movement. Second, we have demonstrated that PDGF localized at the tip of a protrusion triggered a series of morphological events—(1) cell spreading at the tip; (2) movement of the cell body toward the PDGF; and (3) retraction of the trailing protrusions—that resulted in directed migration of the cell. Third, our observations suggest that early and transient receptor-mediated activation of myosin within the leading protrusion drives PDGF-induced chemotaxis. These observations indicate that stellate cell protrusions are capable of detecting chemoattractants and subsequently transducing this chemical signal into mechanical forces that power chemotaxis.
The function that stellate cell protrusions serve is unexplained, despite the fact that these distinctive structures define this cell type. Stellate cells mediate the liver's response to injury and consequently must be able to promptly react to changes in the microenvironment. Stellate cells, however, occupy the perisinusoidal space, an unstirred fluid layer where diffusion limits the rate of molecular movement.6 Therefore, we propose that protrusions may act, in part, as antennae permitting stellate cells to rapidly detect specific chemoattractant signals remote from the cell body. Several observations support this concept. Placement of a PDGF-coated bead at the tip of a protrusion stimulated a specific sequence of changes in myosin activation and the actin cytoskeleton that resulted in migration toward the bead. Placement of a cytochrome C-coated bead, which has a mass and charge similar to PDGF, at the tip of a protrusion did not induce movement toward the bead. Protrusion-mediated chemotaxis was abolished by exposure to a selective receptor tyrosine kinase inhibitor and competitively inhibited by addition of soluble PDGF. These results indicate that stellate cells have the capacity to sense finely localized PDGF at the tips of their protrusions and respond by migrating toward that chemical signal.
Surprisingly, PDGF bead-induced chemotaxis was associated with an early and transient increase in MRLC phosphorylation within the spreading lamella. This event was temporally and spatially distinct from MRLC phosphorylation within the cell body and trailing protrusions. Myosin II functions primarily at the sides and trailing end of motile cells.18, 19 By localizing to these sites during movement, myosin is thought to create polarity by restricting lamellipodia formation to the front of the cell and permitting translocation by retracting the trailing end.20 Our findings, however, suggested that myosin activation within the leading protrusion plays an important role in chemotaxis. We tested this possibility by demonstrating that three mechanistically distinct inhibitors (blebbistatin, ML-7, and Y-27632) of myosin activation blocked protrusion-mediated lamella formation and chemotaxis. Our results suggest that protrusions not only detect PDGF but transduce this chemical signal into mechanical forces that power chemotaxis.
Few studies have examined chemotaxis using stellate cells in early primary culture when they exhibit protrusions. Several studies that investigated stellate cells very early in culture (i.e., within hours of their isolation) have reported an absence or reduced capacity to migrate.21–23 After isolation, however, 1 to 2 days in culture are needed for stellate cells to re-extend their protrusions. Furthermore, prior reports of stellate cell migration have used methods that do not permit study of protrusions.21, 24, 25 Most commonly, the Boyden chamber method has been used to determine the number of stellate cells that migrate across an insert in response to a chemical gradient.21, 26, 27 This method permits study of the response of populations of cells over the course of hours to days, but not surveillance of rapid cellular and subcellular events as required to examine the role of protrusions. To address our hypothesis, we used stellate cells in early primary culture that display characteristic protrusions similar to those observed extending from stellate cells in situ. We also developed an assay in which a chemoattractant, PDGF, was precisely targeted to the tip of a stellate cell protrusion and the real-time and subcellular responses of a single stellate cell were directly visualized.
Based on recent data indicating that stellate cells are phagocytic,28, 29 we considered the possibility that the PDGF-coated bead-induced movement could represent a phagocytotic response. We believe, however, that PDGF-coated beads triggered chemotaxis rather than phagocytosis because cytochrome C-coated beads did not cause movement, the beads used were too large to induce stellate cell phagocytosis,29 and phagocytosis is not associated with cell body translocation or trailing end retraction. These results suggest that PDGF-coated beads triggered chemotaxis, not phagocytosis.
The data presented in this study support a new pathophysiological model in which the presence of protrusions permits stellate cells to promptly detect injury-induced chemical signals remote from their cell bodies and transduce this stimuli into mechanical forces that propel the cell toward the site of injury. This is likely to be the initial event in the stellate cells' response to injury, suggesting that protrusions may be a valuable target for the development of clinical therapies for diseases of the liver. Finally, these findings have general biological significance because hepatic stellate cells belong to a ubiquitous class of star-shaped cells, termed pericytes,30 which are essential for wound healing and blood vessel formation in diverse tissues.3, 30
The authors thank the Research Center for Alcoholic Liver and Pancreatic Diseases (P50 AA11999), its Non-Parenchymal Liver Cell Core (R24 AA12885), and the NIDDK-supported Cell and Tissue Biology Core Facility of the UCSF Liver Center (R01 P30 DK26743) for providing isolated hepatic stellate cells. We thank Gianluca Gallo (Drexel College of Medicine) for his valuable advice regarding the use of protein-coated beads.