Potential conflict of interest: Nothing to report.
Liver regeneration is regulated by growth factors, cytokines, and other endocrine and metabolic factors. Calcium is important for cell division, but its role in liver regeneration is not known. The purpose of this study was to understand the effects of cytosolic calcium signals in liver growth after partial hepatectomy (PH). The gene encoding the calcium-binding protein parvalbumin (PV) targeted to the cytosol using a nuclear export sequence (NES), and using a discosoma red fluorescent protein (DsR) marker, was transfected into rat livers by injecting it, in recombinant adenovirus (Ad), into the portal vein. We performed two-thirds PH 4 days after Ad-PV-NES-DsR or Ad-DsR injection, and liver regeneration was analyzed. Calcium signals were analyzed with fura-2-acetoxymethyl ester in hepatocytes isolated from Ad-infected rats and in Ad-infected Hela cells. Also, isolated hepatocytes were infected with Ad-DsR or Ad-PV-NES-DsR and assayed for bromodeoxyuridine incorporation. Ad-PV-NES-DsR injection resulted in PV expression in the hepatocyte cytosol. Agonist-induced cytosolic calcium oscillations were attenuated in both PV-NES–expressing Hela cells and hepatocytes, as compared to DsR-expressing cells. Bromodeoxyuridine incorporation (S phase), phosphorylated histone 3 immunostaining (mitosis), and liver mass restoration after PH were all significantly delayed in PV-NES rats. Reduced cyclin expression and retinoblastoma protein phosphorylation confirmed this observation. PV-NES rats exhibited reduced c-fos induction and delayed extracellular signal-regulated kinase 1/2 phosphorylation after PH. Finally, primary PV-NES–expressing hepatocytes exhibited less proliferation and agonist-induced cyclic adenosine monophosphate responsive element binding and extracellular signal-regulated kinase 1/2 phosphorylation, as compared with control cells. Conclusion: Cytosolic calcium signals promote liver regeneration by enhancing progression of hepatocytes through the cell cycle. (HEPATOLOGY 2010;)
Liver regeneration is a well-orchestrated process regulated by cytokines, growth factors, hormones, and neurotransmitters that interact with hepatocytes to help restore liver mass and function within days after partial tissue loss.1 A combination of paracrine, autocrine, and endocrine interactions push quiescent liver cells toward the cell cycle, as well as maintain normal differentiated hepatic functions in the regenerating tissue. Among this regulatory signaling network, a number of calcium-mobilizing agonists have been identified, including noradrenaline,2 arginine vasopressin (AVP),3 adenosine triphosphate (ATP),4 epidermal growth factor (EGF), hepatocyte growth factor (HGF),5 and insulin.6 However, the impact of intracellular calcium signaling on liver regeneration after partial hepatectomy (PH) had never been directly studied.
Calcium signaling regulates many cellular processes throughout the life of an organism, including secretion, metabolism, and differentiation.7 Temporal and spatial patterning of calcium signals has been shown to be critical in particular for activation of immediate early genes,8 transcription factors9, 10 and mitogen-activated protein kinases (MAPKs),11 and for cytokine gene expression.9 As a result, intracellular calcium has been reported to regulate cell proliferation at multiple steps of the cell cycle.12 Oscillation frequency, amplitude, and time delay of cytosolic calcium oscillations are essential kinetic parameters to determine cell responses to extracellular stimuli, especially entry into the cell cycle.13 Recent work in a cultured liver cell line highlighted that nuclear calcium signals affected proliferation whereas cytosolic calcium did not.14 On the other hand, both nuclear and cytosolic calcium signals were important for proliferation of hepatic stellate cells.15 However, such studies have never been performed on primary hepatocytes or in vivo. During liver regeneration, calcium has long been known to be important, as suggested by early studies on hypocalcemic rats.16 Alterations in the calcium signaling machinery have been reported to occur during liver regeneration, although the data remain limited.17-19 We found that AVP, a potent calcium-mobilizing agonist in hepatocytes, increased hepatocyte entry into the cell cycle after PH in the rat.3 We also reported that remodeling of calcium signaling occurred in hepatocytes after PH.19 Together, these findings raise the question of whether calcium signaling is important for liver regeneration.
Previous studies have established that cytosolic and nuclear calcium can be regulated independently,20 and that, in particular, growth factors important for liver regeneration, such as HGF21 and insulin,6 can differentially affect cytosolic and nuclear calcium in hepatocytes. We also know that calcium signals in these two compartments can have different effects.22, 23 Therefore, the purpose of the present work was to examine the physiological involvement of cytosolic calcium during liver regeneration in the rat. We interfered with calcium signaling before PH by expressing parvalbumin (PV) in the liver, a calcium-binding protein expressed in muscle cells and neurons but absent from the liver,24 using adenoviruses coding for PV targeted to the cytosol to selectively buffer Ca2+ in this compartment.14, 21 We found that expression of PV efficiently buffers agonist-induced calcium oscillations in the cytosol, and inhibits primary hepatocyte proliferation in vitro as well as in vivo during liver regeneration.
Two-thirds hepatectomy was performed on adult female Wistar rats (weighing 200-250 g).19 Adenoviruses (Ad; 5 × 109 plaque forming units [pfu]/rat) in phosphate-buffered saline (300 μL) were injected in the portal vein after laparotomy under isoflurane anesthesia. At the time of hepatectomy and at various times after PH, liver fragments were removed, frozen in nitrogen-cooled isopentane, and stored at −80°C until use.
Plasma alanine aminotransferase was measured using a commercial kit, according to the manufacturer instructions (Sigma). Interleukin-6 was measured with a Luminex 100 analyzer (kit from R&D Systems).
Constructs encoding discosoma red fluorescent protein (DsR) and PV fusion protein targeted to the cytosol (PV-nuclear export sequence [NES]-DsR) were obtained as described14 (see also Supporting Methods).
Isolation of Hepatocytes and Cell Culture.
Hepatocytes were isolated and cultured as described.3, 25 Cells were infected 1 hour with Ad-DsR or Ad-PV-NES-DsR, at a multiplicity of infection of 20 pfu/cell, and further cultured. The next day, cells were stimulated with ATP 30 μM (cyclic adenosine monophosphate responsive element binding protein [CREB] experiments) or EGF 50 ng/mL (proliferation experiments). Medium was changed every day until the end of the experiment. SkHep cells21 were infected with the different Ad and stimulated with HGF (100 ng/mL for 8 minutes) for extracellular signal-regulated kinase (ERK) phosphorylation experiments.
Determination of [Ca2+]i Changes in Hepatocytes.
Freshly isolated hepatocytes from DsR and PV-NES-DsR–expressing animals, or Hela cells were loaded with fura-2-acetoxymethyl ester (3 μM, 20 minutes, 37°C) (Molecular Probes), and intracellular free calcium concentration ([Ca2+]i) imaging experiments were done as described.19
Liver fragments were homogenized and processed as described.19 Primary antibody dilutions were: PV (PARV19, Sigma Aldrich) 1/1000; ERK1/2 (Promega) 1/5000, phospho-ERK1/2 (Cell Signaling) 1/500; total retinoblastoma protein (pRb; Santa Cruz Biotechnology) 1/1000; phospho-pRb (Cell Signaling Technology) 1/1000; cyclin D1 (Santa Cruz Biotechnology) 1/500; actin (Sigma) 1/5000.
Histology and Immunohistochemistry.
Bromodeoxyuridine (BrdU) and cell death detection (Tunel-AP) were performed on frozen liver sections according to manufacturer's directions (Roche). Nuclear BrdU labeling pattern after PH was analyzed as previously reported.26 The number of BrdU-positive nuclei was related to the total number of nuclei (Hoescht staining). Oil red O and hematoxylin staining were performed according to standard procedures.
At different time points after stimulation, primary rat hepatocytes were fixed in 4% formaldehyde, permeabilized with 0.1% Triton, incubated with antibody against phospho-CREB (1/100; Cell Signaling Technology, Danvers, MA) at 4°C overnight, and incubated with Alexa Fluor 488–conjugated secondary antibody. Epifluorescence images were obtained on an Axioskop photomicroscope (Zeiss).
Quantitative Polymerase Chain Reaction.
Total RNA was isolated from rat liver on TRI Reagent (Sigma) according to manufacturer's recommendations. Complementary DNA was amplified by polymerase chain reaction (PCR) in the presence of specific primers detailed in Supporting Table 1. Expression of genes of interest was compared with the expression of glyceraldehyde 3-phosphate dehydrogenase and 18S ribosomal RNA.
We used the student t test to compare sample means with paired controls. Results are expressed as means ± standard error of the mean. P values ≤ 0.05 were considered to be statistically significant.
PV Expression After Recombinant Adenovirus Infection.
An increasing amount of PV-NES-DsR fusion protein expression was observed in the liver as a function of time, culminating at day 4 after portal Ad administration (Fig. 1A). At this time point, extrahepatic PV expression was high in spleen and lung, but not in kidney, heart, and muscle (Supporting Fig. 1). DsR fluorescence and PV immunohistochemistry revealed that adenovirus infection induced strong transgene expression throughout the liver, with periportal predominance (Fig. 1B). PV-NES-DsR was expressed only in the cytosol of hepatocytes, as shown on liver sections (Fig. 1B). Only marginal PV expression was observed in Kupffer cells (Supporting Fig. 2). PV was expressed during the 3 days following PH (Fig. 1C), but it was hardly detected 7 days after PH (not shown). Rats injected with Ad-PV-NES-DsR or Ad-DsR were comparable at the time of PH, in terms of initial liver weight and other parameters (Supporting Fig. 3).
Calcium Signaling Alteration in PV-Expressing Cells.
Because PV was highly expressed in hepatocyte cytosol in vivo after Ad injection, we checked if agonist-induced calcium signals were altered in isolated hepatocytes from infected rats. As shown in Fig. 2A, low concentration of noradrenaline induced well-shaped calcium oscillations in cells with low or no detectable PV-NES-DsR expression (“control cells”), whereas oscillations were nearly abolished (one isolated Ca2+ transient) in PV-NES-DsR–expressing hepatocytes. Slow Ca2+ oscillations were induced when a high noradrenaline concentration was used, whereas a Ca2+ plateau was observed in control cells, which suggests that cytosolic PV expression induced a decrease in hepatocyte sensitivity to Ca2+-mobilizing agonists. Basal cytosolic Ca2+ concentrations were not different between PV-expressing and control cells (PV-NES: 141 ± 6 nM [n = 9]; DsR: 159 ± 12 nM [n = 30]; noninfected cells: 151 ± 4 nM [n = 30]), an observation in line with biochemical properties of PV, which acts as a slow buffer Ca2+-binding protein.24 As a control, hepatocytes from Ad-DsR–infected rats were isolated and did not exhibit alteration in their calcium responses to noradrenaline as compared to noninfected cells (Fig. 2B).
It was previously reported in Xenopus oocytes that PV can attenuate inositol triphosphate–evoked calcium release.27 Our data are also consistent with previous reports on PV-expressing cell lines, in which amplitude of calcium responses was found to be reduced as compared with control cells.14, 21 We observed the same data in Hela cells infected in vitro with Ad PV-NES-DsR as compared with Ad DsR. As shown in Supporting Fig. 4, PV-expressing Hela cells exhibited attenuated Ca2+ responses to ATP (0.1-10 μM), as compared with nonexpressing cells. More precise analysis revealed that 21.4% of PV-NES-DsR–expressing cells exhibited a biphasic pattern of [Ca2+]i rise, with a rapid first phase and a slower second phase, whereas DsR-expressing cells did not. Also, the [Ca2+]i rise time was significantly longer in PV-NES-DsR–expressing cells (1.19 ± 0.16 minutes; n = 28) than in control cells (0.42 ± 0.04 minutes) (Supporting Fig. 4). Finally, the number of oscillating cells in response to ATP (1 and 3 μM) was dramatically reduced by PV-NES expression (14.3% versus 97.4% in control cells). These data demonstrate dramatic impairment of agonist-induced calcium signals induced by cytosolic expression of PV in Hela cells as well as in in vivo–infected hepatocytes. Calcium response kinetics suggests a decrease in sensitivity to Ca2+-mobilizing agonists in PV-NES–expressing cells.
Liver Regeneration Is Delayed in PV-Expressing Rats.
Ad-PV-NES-DsR or Ad-DsR were injected into the portal vein, and PH was performed at day 4, because transgene expression was the highest at this time point (Fig. 1C).
Cytosolic PV expression significantly reduced liver mass restoration, mainly in the first 24 (P < 0.001, n = 6) to 48 (P = 0.05, n = 6) hours after PH (Fig. 3A). The DNA synthesis peak 24 hours after PH was also significantly delayed in PV-NES–injected rats as compared with controls, as measured by BrdU incorporation in liver sections (P = 0.001, n = 4) (Fig. 3B). Hepatocyte mitosis, monitored by phosphorylated histone 3 (PH3) immunostaining, was reduced in PV-NES rats at 24 hours after PH (H24; P < 0.01, n = 6) (Fig. 3C). In line with these results, BrdU labeling patterns of DNA replication26 indicated a slower progression through S phase in PV-NES-DsR livers than in DsR controls (Fig. 4A), and significantly lower amounts of phospho-retinoblastoma (phospho-pRb) were detected in PV-NES-DsR than in DsR rats 24 hours (P < 0.05, n = 4) after PH (Fig. 4B). Quantitative PCR analysis of cyclin D1, cyclin A, and cyclin E messenger RNA expression after PH confirmed impaired cell cycle progression in PV-NES–expressing rats (Fig. 4C) (P = 0.05, n = 4). As a control, there was no increase in apoptotic cells in the PV-NES-DsR group as compared with control rats at 24, 48, or 72 hours after PH (Supporting Fig. 5). These findings suggest that calcium signals help quiescent G0 hepatocytes enter the cell cycle, thus restoring liver mass.
Early Events After PH Are Altered by Cytosolic Calcium Buffering.
Intracellular calcium movements control gene expression mostly by changes in protein phosphorylation. The activity of MAPKs, Ca2+/calmodulin-dependent protein kinases (CaMKs), and of numerous transcription factors has been shown to be regulated by intracellular Ca2+ movements.28 The Ras pathway is tightly controlled by calcium oscillations through calcium-regulated translocation to plasma membrane of Ras guanosine triphosphatase–activating proteins,29 and is critical for cell cycle progression after PH.30 Consistent with this, ERK1/2 was significantly less phosphorylated at 1 hour after PH in rats expressing PV-NES-DsR (Fig. 5). Moreover, in SkHep cells, a widely used hepatocyte cell line for studying calcium signaling,14, 20, 21 HGF-induced ERK phosphorylation was significantly inhibited by infection with Ad-PV-NES-DsR as compared with Ad-DsR (Supporting Fig. 6). We also examined CREB phosphorylation, an important calcium-dependent pathway for gene expression regulation,31 which is activated during liver regeneration.32 Isolated rat hepatocytes were infected in vitro with either Ad-DsR or Ad-PV-NES-DsR, and stimulated with ATP (30 μM) as a calcium-mobilizing agonist. Phospho-CREB was strongly induced by ATP in DsR-expressing hepatocytes, whereas it was induced far less in PV-NES–expressing hepatocytes (Fig. 6). It is noteworthy that, as shown in Fig. 4C, the immediate early gene c-fos was less induced in PV-NES-DsR than in DsR rats, at 1 hour (P = 0.05, n = 4) after PH. Given that this gene, as well as ERK and CREB, is regulated in part by Ca2+,8, 28, 29, 31 these findings are consistent with the hypothesis that altered cytosolic [Ca2+]i oscillations in hepatocytes result in alteration in their activation.
Because hepatocyte calcium signals modulate bile flow and glucose metabolism,33, 34 we measured bile acids and glucose in the plasma of DsR and PV-NES rats, but did not find any significant difference between the two groups, which suggests that calcium signaling disruption is compensated by other signaling pathways (Supporting Fig. 7).
Cytosolic PV Expression Dampens Hepatocyte Progression in the Cell Cycle In Vitro.
Given that hepatocyte in vivo cell cycle entry after PH was delayed by cytosolic calcium buffering, hepatocyte proliferation was investigated in vitro. BrdU incorporation was studied in isolated rat hepatocytes after in vitro infection with Ad-DsR or Ad-PV-NES-DsR. We observed that cytosolic PV expression in vitro, as was found in vivo after PH, inhibited hepatocyte S phase entry as compared with DsR-expressing control cells (Fig. 7). BrdU incorporation at the DNA synthesis peak was 1.6-fold ± 0.3-fold greater in control hepatocytes than in PV-NES–expressing cells (n = 3, P = 0.05). Thus, hepatocytes in primary culture, as well as in vivo in their physiological environment, appear to need unaltered cytosolic calcium signaling for optimal cell cycle progression.
Liver regeneration is regulated by growth factors and cytokines that affect cytosolic calcium signaling,1, 5 yet the impact of intracellular calcium mobilization on liver regeneration remains poorly explored.3, 19 In this work, we have directly interfered with cytosolic calcium signaling in vivo by expressing the calcium buffer PV in the cytosol of hepatocytes. Buffering agonist-induced cytosolic calcium oscillations led to slowing of cell cycle progression in hepatocytes in primary cultures as well as in vivo after PH, resulting in delayed regeneration of the organ.
Previous studies in cell lines6, 14, 15, 20-22 or tissues35 have shown that transfected PV can efficiently alter calcium signal profiles27 and cell cycle progression in culture.36 Hepatocytes typically respond to agonists with frequency-modulated calcium oscillations3, 19 that propagate across the cell and transmit from cell to cell as intracellular and intercellular calcium waves.33, 34 These signals have been linked to the regulation of physiological processes such as glucose metabolism,37 bile flow,3 and liver regeneration.3, 19 By attenuating the well-organized pattern of agonist-induced calcium signals before PH, we dampened hepatocyte progression through the cell cycle after PH, suggesting that such signals control critical parameters for optimal liver regeneration. Previous reports in cell lines and primary isolated cells suggested that calcium signals in the nuclear and cytosolic compartments play distinct roles for cell proliferation.14, 15 In line with this, the requirement for calcium in cell proliferation may be modulated by the degree of cell transformation.12 Also, primary hepatocytes and liver cell lines do not express the same repertoire of intracellular calcium channels in the different cell compartments,20 thus in turn altering calcium signaling.38 The present study demonstrates in vivo at the organ level that hepatocytes require unaltered cytosolic calcium signaling to progress through the cell cycle after PH, even though liver cell lines do not share this requirement.14 Interestingly, agonist-induced nuclear calcium signals are not altered in cells expressing PV exclusively in the cytosol,14, 22 perhaps because calcium can diffuse from the cytosol to the nucleus,39, 40 where it is not buffered. Calcium signals in both compartments—cytosol and nucleus—may thus be important in vivo, and further studies are required to specifically examine the impact of nuclear calcium signals on primary hepatocyte proliferation and liver regeneration.
The impact of PV expression on the different liver regeneration parameters revealed that cytosolic calcium signals may be important for early events after PH. Immediate early gene transcription and early phosphorylation of ERK and CREB, all have been reported as crucial for hepatocyte progression after PH,30, 32, 41 and are dependent on cytosolic and/or nuclear calcium signaling.8, 23, 28, 29 We suggest that attenuated [Ca2+]i oscillations in calcium-buffered hepatocytes resulted in impaired activation of these pathways. A potential reduction in CaMK activation, as previously reported,31 or reduced ERK1/2 activation that we observed in PV-NES–expressing hepatocytes, may have also contributed to altered CREB phosphorylation. Because CaMK II,42 as well as ERK1/229 activity, is sensitive to Ca2+ oscillation frequency, an attractive hypothesis would be that cytosolic PV expression, by attenuating agonist-generated Ca2+ signals, resulted in impaired phosphorylation of CREB.
We also observed that PV-NES-DsR rats exhibited impaired cytokine induction after PH, as compared with control rats, suggesting that intracellular calcium buffering may have occurred in a small fraction of Kupffer cells, with potential effects on cytokine gene induction (Supporting Fig. 2).
Our data define a role for cytosolic calcium signaling in the early triggering of hepatocyte cell cycle progression from G0 to G1 and S phases. This fits with the rise in concentration, in the liver and in the plasma, observed early after PH for several Ca2+-mobilizing agonists, suggesting these agonists might be involved in initiating the regeneration process. In particular, EGF and HGF elicit cytosolic Ca2+ oscillations in hepatocytes, the physiological impact of which has never been specifically addressed.5 Also, comitogenic Ca2+-mobilizing agonists like extracellular ATP,4 AVP,3 and noradrenaline,2 have been individually reported to contribute to early phases of liver regeneration. The present study provides evidence that buffering hepatocyte calcium signals, potentially generated by these agonists in the minutes after PH, results in delaying hepatocyte cell cycle progression. Further work will be needed to precisely explore the targets of cytosolic calcium signals during liver regeneration.
We thank Karim Benihoud for helping us in adenovirus production and Sylviane Boucherie and Mauricette Collado for excellent technical assistance. We also thank Lydie Humbert and Dominique Rainteau (INSERM U538, Paris) for plasma bile acid measurements.