Eli and Edythe Broad Center for Regeneration Medicine and Stem Cell Research, University of California San Francisco, San Francisco, CA
Department of Surgery, Division of Transplantation, University of California San Francisco, San Francisco, CA
Liver Center, University of California San Francisco, San Francisco, CA
Eli and Edythe Broad Center for Regeneration Medicine and Stem Cell Research, Department of Surgery, Division of Transplantation, University of California San Francisco, 513 Parnassus Avenue, S1457B, Campus Box 0525, San Francisco, CA 94143
Potential conflict of interest: Nothing to report.
H.W. received grant support from the California Institute for Regenerative Medicine, American Liver Foundation, and American Society of Transplantation. G.R. was supported by a fellowship from the NRSA Hepatology Training Grant at UCSF. R.N. was supported by a scholarship from the Agency of Science Technology and Research (A*STAR) Singapore.
MicroRNAs (miRNAs) constitute a new class of regulators of gene expression. Among other actions, miRNAs have been shown to control cell proliferation in development and cancer. However, whether miRNAs regulate hepatocyte proliferation during liver regeneration is unknown. We addressed this question by performing 2/3 partial hepatectomy (2/3 PH) on mice with hepatocyte-specific inactivation of DiGeorge syndrome critical region gene 8 (DGCR8), an essential component of the miRNA processing pathway. Hepatocytes of these mice were miRNA-deficient and exhibited a delay in cell cycle progression involving the G1 to S phase transition. Examination of livers of wildtype mice after 2/3 PH revealed differential expression of a subset of miRNAs, notably an induction of miR-21 and repression of miR-378. We further discovered that miR-21 directly inhibits Btg2, a cell cycle inhibitor that prevents activation of forkhead box M1 (FoxM1), which is essential for DNA synthesis in hepatocytes after 2/3 PH. In addition, we found that miR-378 directly inhibits ornithine decarboxylase (Odc1), which is known to promote DNA synthesis in hepatocytes after 2/3 PH. Conclusion: Our results show that miRNAs are critical regulators of hepatocyte proliferation during liver regeneration. Because these miRNAs and target gene interactions are conserved, our findings may also be relevant to human liver regeneration. (HEPATOLOGY 2010)
The adult liver is unique in its intrinsic ability to regenerate through proliferation of fully differentiated cells.1 Adult hepatocytes are quiescent and normally divide only once or twice a year in mice and even less frequently in humans.2 However, adult hepatocytes have the ability to divide numerous times in response to liver tissue injury or loss.3, 4 After 2/3 partial hepatectomy (2/3 PH) in mice, hepatocytes enter and progress through the cell cycle in a highly synchronized fashion.5 Every hepatocyte divides once, and every other hepatocyte divides once more to restore the liver mass within 7 days.1 A complex network of cytokine and growth factor signaling between hepatocytes and other liver cell types regulates the hepatocyte cell cycle to ensure that liver regeneration is rapid and robust.6
Although microRNAs (miRNAs) have been shown to posttranscriptionally regulate genes that orchestrate proliferation in development and cancer, their role in organ regeneration is largely unknown. Recent studies in zebrafish have revealed that suppression of miR-1337 or miR-2038 is required for fin regeneration. Zebrafish fin regeneration is mediated by stem cells that are recruited to or originate from dedifferentiation of cells residing in the injured area. In contrast, regeneration of the mammalian liver entails cell cycle entry and division of differentiated hepatocytes. Proliferating hepatocytes remain differentiated and continue to provide liver function.9 Knowing how this unique form of regeneration is regulated might enable its restoration in diseased hepatocytes and recapitulation in other, nonregenerative cell types. Here we describe the results of our analysis of the changes in miRNA expression during mouse liver regeneration, leading to the identification of miR-21 and miR-378 as regulators of organ regeneration.
Dgcr8del/fl, Alb-Cre+/−, and Dgcr8fl/fl mice were generated from matings of Dgcr8del/fl mice on a mixed C57Bl/6, 129S4 background10 with Alb-Cre+/− mice on a C57Bl/6 background (Jackson Laboratory).11, 12 Eight to 12-week-old male mice were used in this study. All procedures involving mice were approved by the Institutional Animal Care Committee at the University of California San Francisco.
Two-thirds of the liver was surgically removed under isoflurane anesthesia as described.5 One hundred μg/g body weight 5-bromo-2-deoxyuridine (BrdU, Roche) was injected intraperitoneally 35 hours after surgery.
Total RNA was isolated with Trizol and treated with DNase I (Ambion) to eliminate genomic DNA. One μg RNA was used for cDNA synthesis with Superscript III reverse transcription reagent (Invitrogen). PCR amplification was performed at 50°C for 2 minutes and 95°C for 10 minutes, followed by 40 cycles at 95°C for 15 seconds and 60°C for 1 minute in a 7300 real-time PCR system with SYBR green (Applied Biosystems). For each sample we analyzed β-actin, Gapdh, or 18S rRNA expression to normalize target gene expression. Primers for qRT-PCR were designed with Primer Express software (Applied Biosystems). Dgcr8 primers were designed to target the deleted exon 3. For miRNA analysis, RNA was isolated with the miRNeasy kit (Qiagen). Ten ng RNA was used for miRNA-specific cDNA synthesis with the TaqMan MicroRNA Reverse Transcription Kit and Taqman MicroRNA Assays (all Applied Biosystems). PCR amplification was performed at 95°C for 10 minutes, followed by 40 cycles at 95°C for 15 seconds and 60°C for 1 minute in a 7900 real-time PCR system (Applied Biosystems). The small RNA Sno202 was used to normalize target miRNA expression. Relative changes in gene and miRNA expression were determined using the 2-ΔΔCt method.13
Paraffin-embedded liver samples were sectioned and stained with the antibodies rabbit anti-Cyclin D (NeoMarkers), mouse anti-PCNA (Biosource), and rabbit anti-Ki67 (Lab Vision) at 1:100 dilutions. To visualize immunocomplexes for light microscopy with 3,3′-diaminobenzidine (DAB), we used biotinylated antirabbit or antimouse IgG antibodies and the ABC reagent (all Vector). Immunostainings with rat anti-A6 (gift from Dr. Valentina Faktor, NCI), rabbit anti-DGCR8 (Proteintech), and rat anti-BrdU (Abcam) antibodies were performed on sections of frozen liver samples embedded in optimum cutting temperature compound (Tissue-Tek, Sakura Finetek) at 1:250, 1:50, and 1:100 dilution, respectively. For fluorescence microscopy, the secondary antibodies goat antirat conjugated with Alexa Fluor 488, goat antirabbit conjugated with Alexa Fluor 594, and goat antirat conjugated with Alexa Fluor 594 (all Molecular Probes) were used at 1:500 dilutions. Nuclear DNA was stained with 10 μg/mL Hoechst 33342 (Molecular Probes).
Global miRNA Expression Profiling.
Eight-week-old male wildtype C57Bl/6 mice (Jackson Laboratory) were used for analysis of miRNA expression changes after 2/3 PH. Liver samples were obtained at 0, 1.5, 6, and 18 hours after surgery. Five mice were analyzed for each timepoint. Total RNA was isolated using Trizol and purified with the miRNeasy mini kit (Qiagen). miRNA expression profiling including labeling, hybridization, scanning, normalization, and data analysis was performed at Exiqon. Profiling was conducted with total RNA with RNA integrity number (RIN) values close to 8 measured with an Agilent 2100 Bioanalyzer. One μg total RNA of each sample and a common reference pool were labeled with Hy3 and Hy5 fluorescent label, respectively, using the miRCURY LNA Array power labeling kit (Exiqon). Hy3-labeled samples and Hy5-labeled common reference pool were mixed pairwise and hybridized to miRCURY LNA arrays v. 9.2 (Exiqon), which have a 61% coverage of the mouse miRNAs listed in miRBase v. 14.0. Hybridization was performed using a Tecan HS4800 hybridization station. Slides were scanned using an Agilent G2565BA Microarray Scanner System and image analysis was carried out with ImaGene 7.0 software (BioDiscovery). Background correction was performed to remove nonbiological contributions to the measured signal.14 Quantified signals were normalized using the global Lowess (locally weighted scatterplot smoothing) regression algorithm.15 Log2 transformed median Hy3 (sample)/Hy5 (common reference pool) ratios were calculated for each capture probe (present in four replicates on the array). A two-tailed Student's t test was performed between the samples obtained at 0 hours and the samples obtained at 1.5, 6, and 18 hours after 2/3 PH. Clustering was performed on miRNAs corresponding to capture probes with log2 Hy3/Hy5 ratios which passed filtering criteria of P < 0.001. The heatmap was generated using the Euclidean metric. The log2 median ratio values were standardized by subtracting the mean followed by dividing by the standard deviation.16
miRNA Manipulations for qRT-PCR and Luciferase Assays.
miRIDIAN miRNA Mimics or Hairpin Inhibitors (Dharmacon) were introduced into Hepa1,6 mouse hepatoma cells (ATCC) at a final concentration of 20, 40, or 80 nM. Five × 104 cells were plated in 24-well plates (Corning) and transfected using Lipofectamine 2000 (Invitrogen) 24 hours later. Hepa1,6 cells transfected with chemically modified double-stranded or single-stranded oligonucleotides based on the sequence of miR-67 from C. elegans (both Dharmacon) were used as controls for miRNA mimics or inhibitors, respectively. For luciferase assays, cells were cotransfected with 30 ng of the pMIR-REPORT vector (Ambion) modified to contain the B-cell translocation gene 2 (Btg2) or ornithine decarboxylase 1 (Odc1) 3′ untranslated region (UTR) and 30 ng of the pSV-β-Galactosidase Control Vector (Promega) to monitor transfection efficiencies. Twenty-four hours after transfection, luciferase and β-galactosidase activities were measured using the Luciferase Assay System and the Beta-Glo Assay System (both Promega) on a Synergy 2 Microplate Reader (BioTek Instruments). Luciferase activities were normalized to β-galactosidase activities for each well.
Significance was determined with a two-tailed Student's t test.
To determine the overall impact of miRNAs on liver regeneration, we performed 2/3 PH on mice with global miRNA deficiency specifically in hepatocytes. Mature miRNAs are the product of sequential cleavage of the primary transcript by the RNase III enzymes Drosha and Dicer. Because Dicer is also involved in processing of other small RNAs, we generated mice with hepatocyte-specific inactivation of DGCR8, which together with Drosha forms the microprocessor complex that is specifically required for canonical miRNA biogenesis.17 Mice with hepatocyte-specific miRNA deficiency were viable and developed normally into adulthood. However, whereas miRNA-deficient hepatocytes readily exited the G0 phase of the cell cycle, they failed to transition into S phase by 36 hours after 2/3 PH (Fig. 1A). Despite increased expression of Cyclin D1 (Ccnd1) before 2/3 PH, these cells failed to induce Cyclin A2 (Ccna2) and Cyclin B1 (Ccnb1) expression at 36 hours after 2/3 PH (Fig. 1B). Moreover, other genes or markers specific for DNA synthesis were not activated or detectable in miRNA-deficient hepatocytes (Supporting Information Fig. 1B,C). Interestingly, a subset of the mice showed compensatory expansion of adult liver progenitors, so-called oval cells (Fig. 1A, Supporting Information Fig. 2A). In contrast to hepatocytes, oval cells retained intact DGCR8 and hence miRNA expression, which explains their normal proliferative capabilities (Fig. 1A-D, Supporting Information Fig. 2B).
To identify miRNAs regulating hepatocyte S phase entry during liver regeneration, we analyzed global miRNA expression during the first 36 hours after 2/3 PH in wildtype mice. Pilot analyses led us to focus on miRNA expression changes during the first 18 hours after 2/3 PH (Supporting Information Table 1 and data not shown). Previous studies showed that many genes are differentially expressed after 2/3 PH.9, 18, 19 However, using a stringent cutoff of P < 0.001, we found significantly altered expression after 2/3 PH of only 7 of ≈430 mouse miRNAs analyzed (Fig. 2A). Intriguingly, miR-21, a known promoter of proliferation in cancer,20 was most significantly induced. miR-21 peaked at 18 hours after 2/3 PH, that is, after hepatocytes transitioned from G0 into G1 but before they passed the restriction point and entered S phase (Fig. 2B). Recent studies showed that miR-21 is transcriptionally regulated by activation protein 1 (AP-1)21 and signal transducer and activator of transcription 3 (STAT3),22 proteins activated early in liver regeneration.1 Because these and our findings fit well with lack of miR-21 impairing the transition of regenerating hepatocytes from G1 to S phase, we focused our analyses on miR-21.
Several target genes of miR-21 were previously reported.20 However, to potentially identify new targets of miR-21 involved in liver regeneration, we chose an unbiased approach. We first used the TargetScan algorithm to identify genes targeted by miR-21 in both mice and humans.23 Focusing on conserved miR-21 targets not only increased the probability of target gene prediction but also assured that our results could be extended to human liver regeneration. We then used the PicTar algorithm to scan the 3′UTR of the conserved miR-21 target genes and eliminate genes with a lower score or free energy (Supporting Information Table 2).24 Our findings of impaired G1 to S phase progression in miRNA-deficient hepatocytes and induction of miR-21 at a time when entry into S phase is negotiated suggested that miR-21 acts to promote liver regeneration. Therefore, among the 63 genes meeting the selection criteria, we focused on 17 genes with established negative effects on proliferation (Supporting Information Table 2). Among these genes were the previously reported miR-21 targets Timp3, Reck, and Pdcd4.20 Potential new miR-21 targets included Tgfbi and Smad7, components of the transforming growth factor β (TGFβ) signaling pathway, which is known to restrict liver regeneration.25 Most interestingly, however, our search retrieved Btg2, a gene restraining G1 to S phase transition that, paradoxically, is induced by 2/3 PH.18 Because Btg2 also had the highest score and free energy of the predicted conserved miR-21 target genes with established proliferation-inhibiting function, we investigated whether it is directly targeted by miR-21 (Supporting Information Fig. 4A, Supporting Information Table 2).
BTG2 inhibits proliferation by interfering with activating phosphorylation of FoxM1.26 FoxM1 is activated after 2/3 PH and its deficiency impairs DNA synthesis and Ccnb1 gene expression in regenerating mouse hepatocytes.26, 27Btg2 was previously reported to be immediately induced and peak at 4 hours after 2/3 PH.18 When we investigated the expression of Btg2 at later stages, we found that it returns to baseline levels between 6 and 18 hours after 2/3 PH. Thus, the expression pattern of Btg2 is the mirror opposite of that of miR-21 (Fig. 3A). Analysis of Dgcr8del/fl, Alb-Cre+/− mice lacking oval cells showed that miR-21 is mainly expressed in hepatocytes in the liver (Fig. 3B). Taken together with the similar nature of the cell cycle defect in hepatocytes with FoxM1 or global miRNA deficiency (Fig. 1A,B), our findings suggested that miR-21 antagonizes Btg2 in regenerating hepatocytes to facilitate efficient cell cycle progression. Indeed, Btg2 messenger RNA (mRNA) levels and activity of a reporter gene linked to its 3′UTR readily responded to miR-21 mimic or inhibitor transfection into well-differentiated mouse hepatoma cells (Fig. 3C). These manipulations also caused induction or suppression of the FoxM1 target gene Ccnb1, respectively (Fig. 3D). We conclude that direct inhibition of Btg2 by miR-21 antagonizes its inhibitory effect on FoxM1 during liver regeneration.
Another miRNA significantly altered in response to 2/3 PH was miR-378 (Fig. 2A). In contrast to miR-21, miR-378 expression declined after 2/3 PH, suggesting that it might inhibit a proliferation-promoting gene (Fig. 4A). Loss of miR-378 expression in livers of Dgcr8del/fl, Alb-Cre+/− mice lacking oval cells indicated that it functions predominantly in hepatocytes in the liver (Fig. 4B). Among 64 genes predicted as targets of miR-378 in both mice and humans, four genes were previously reported to have a positive effect on cell proliferation (Supporting Information Table 3). Intriguingly, one of these genes, Odc1, encodes a polyamine-synthesizing enzyme that is needed for efficient and timely DNA synthesis in liver regeneration.28 Furthermore, Odc1 had the highest score and free energy of the predicted conserved miR-378 target genes with established proliferation-promoting function (Supporting Information Fig. 4B, Supporting Information Table 3). Using the experimental strategy described above, we found that miR-378 inhibits Odc1 by direct targeting of a complementary sequence in its 3′UTR (Fig. 4C). These results suggest that efficient liver regeneration may involve release of Odc1 from repression by miR-378.
Although miRNAs are known to regulate cell proliferation, little information exists on the miRNAs and their target genes involved in regeneration of the liver or other organs. We investigated this question by generating mice with hepatocyte-specific inactivation of DGCR8, an essential component of the microprocessor complex. DGCR8 anchors the primary miRNA transcript for cleavage by Drosha. Thus, DGCR8 acts upstream of Dicer and its deficiency leads to disruption of processing of miRNAs, not other small RNAs.17 In line with recent findings of impaired proliferation in DGCR8-deficient mouse embryonic stem cells,29 we found that initiation of DNA synthesis was delayed in DGCR8-deficient hepatocytes after 2/3 PH. This finding suggested that miRNAs regulate hepatocyte G1 to S phase progression during liver regeneration.
To identify miRNAs regulating the regenerative capabilities of hepatocytes, we screened for miRNA expression changes in livers of adult wildtype mice after 2/3 PH. Because the miRNAs reported as promoters of proliferation in mouse embryonic stem cells are not expressed in hepatocytes, it was not surprising that none of these miRNAs was induced by 2/3 PH. Instead, we specifically found increased levels of miR-21, an established promoter of proliferation that is expressed at high levels in many types of cancer.20
Prompted by stringent target gene prediction we found that Btg2 is a direct target of miR-21. Moreover, our data suggest that increased miR-21 expression serves to antagonize the cell cycle inhibitor Btg2 during liver regeneration. The proliferation-promoting effect of increased miR-21 after 2/3 PH appears to be aided by a decrease in miR-378 expression and subsequent de-repression of Odc1. However, because the net result of global miRNA deficiency in hepatocytes was impaired S phase entry, loss of miR-378 appears to be insufficient to compensate loss of miR-21 during liver regeneration. Specific inhibition of miR-21 and miR-378 in vivo could be used to delineate their individual contributions to regulation of liver regeneration.30
Because a single miRNA typically targets many genes, the effects of miR-21 and miR-378 during liver regeneration are most likely not restricted to inhibition of Btg2 and Odc1. For example, in addition to Tgfbi and Smad7, the TargetScan algorithm predicts Tgfbr2, Acvr1c (activin A receptor 1C), and Acvr2a (activin A receptor 2A) as direct and conserved miR-21 targets involved in TGFβ and activin signaling. miR-21 might target these genes to limit the inhibitory effect of TGFβ and activin signaling on G1 to S phase transition of hepatocytes after 2/3 PH.25
The levels of the proliferation-promoting gene Ccnd1 were increased in hepatocytes with global miRNA deficiency before 2/3 PH. Our miRNA profiling revealed that miR-16, an miRNA known to inhibit Ccnd1 in the prostate, is expressed in the liver.31 Thus, loss of miR-16 may explain de-repression of Ccnd1 in Dgcr8del/fl, Alb-Cre+/− mice. In analogy, it is possible that loss of other miRNAs normally expressed in hepatocytes but not induced by 2/3 PH may contribute to impaired liver regeneration in Dgcr8del/fl, Alb-Cre+/− mice. This could explain the spontaneous oval cell activation in a subset of Dgcr8del/fl, Alb-Cre+/− mice. In addition, mouse miRNAs continue to be identified and we cannot rule out that miRNAs not represented on our arrays play a role in liver regeneration. However, in contrast to findings after DGCR8 inactivation in the skin,32 miR-21 was depleted in whole liver samples of mice with hepatocyte-specific DGCR8 deficiency. This shows that miR-21 is mainly expressed in hepatocytes in the liver and supports our conclusion that miR-21 directly regulates cell cycle progression in hepatocytes. Specific induction of miR-21 in G1 phase after 2/3 PH and impaired G1 to S phase transition in both hepatocytes with global miRNA deficiency and in those with FoxM1 deficiency further suggest that miR-21 plays a leading role in miRNA regulation of liver regeneration.
Our analyses focused on miRNA target genes that are conserved between mouse and human. Although little has been reported about miR-378's regulation or function, the expression of miR-21 is known to be increased in primary human liver cancer.33, 34 Moreover, miR-21 has been shown to promote proliferation of human liver cancer cell lines by inhibition of the phosphatase and tensin homolog (PTEN) tumor suppressor.33 Therefore, it is likely that miR-21 inhibits Btg2, and potentially other regulators of hepatocyte proliferation, also in human liver regeneration.
The authors thank Dr. Valentina Faktor for providing A6 antibody, Sandra Huling for immunostainings, and Dr. Montgomery Bissell for critical reading of the article.