Ring1B promotes hepatic stem/progenitor cell expansion through simultaneous suppression of Cdkn1a and Cdkn2a in mice

Authors

  • Hiroyuki Koike,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Yasuharu Ueno,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Takako Naito,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    Search for more papers by this author
  • Tomoya Shiina,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    Search for more papers by this author
  • Susumu Nakata,

    1. Division of Oncological Pathology, Aichi Cancer Center Research Institute, Nagoya, Japan
    Search for more papers by this author
  • Rie Ouchi,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    Search for more papers by this author
  • Yuta Obana,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    Search for more papers by this author
  • Keisuke Sekine,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    Search for more papers by this author
  • Yun-Wen Zheng,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    Search for more papers by this author
  • Takanori Takebe,

    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    2. Project Leader of Advanced Medical Research Center, Yokohama City University, Yokohama, Japan
    3. PRESTO, Japan Science and Technology Agency, Tokyo, Japan
    Search for more papers by this author
  • Kyo-ichi Isono,

    1. Laboratory for Developmental Genetics, RIKEN Center for Integrative Medical Sciences, Yokohama, Japan
    Search for more papers by this author
  • Haruhiko Koseki,

    1. Laboratory for Developmental Genetics, RIKEN Center for Integrative Medical Sciences, Yokohama, Japan
    Search for more papers by this author
  • Hideki Taniguchi

    Corresponding author
    1. Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, Yokohama, Japan
    2. Project Leader of Advanced Medical Research Center, Yokohama City University, Yokohama, Japan
    • Address reprint requests to: Hideki Taniguchi, M.D., Ph.D., The Department of Regenerative Medicine, Yokohama City University Graduate School of Medicine, 3-9 Fukuura, Kanazawa-ku, Yokohama, Kanagawa 236-0004, Japan. E-mail: rtanigu@med.yokohama-cu.ac.jp; fax: +81-45-787-8963.

    Search for more papers by this author

  • Potential conflict of interest: Nothing to report.

  • This work was supported by grants from the Special Coordination Funds for Promoting Science (11800122), grants from Stategic Promotion of Innovative Research and Development (S-innovation, 62890004), the Grant-in-Aid for Scientific Research on Innovative Area,s and the Japan Health Sciences Foundation of the Japan Science and Technology Agency (JST). This work was also supported by the Grants-in Aid (nos. 20591532, 21249071, 22390260, 23791490, 23791491, 24106510, and 24689052) of the Ministry of Education, Culture, Sports, Science, and Technology (MEXT) of Japan, by the Specified Research Grant from Takeda Science Foundation, by a grant from Yokohama Foundation for Advanced Medical Science, and by a grant from the Japan Insulin Dependent Diabetes Mellitus (IDDM) network.

Abstract

Polycomb-group (PcG) proteins play crucial roles in self-renewal of stem cells by suppressing a host of genes through histone modifications. Identification of the downstream genes of PcG proteins is essential for elucidation of the molecular mechanisms of stem cell self-renewal. However, little is known about the PcG target genes in tissue stem cells. We found that the PcG protein, Ring1B, which regulates expression of various genes through monoubiquitination of histone H2AK119, is essential for expansion of hepatic stem/progenitor cells. In mouse embryos with a conditional knockout of Ring1B, we found that the lack of Ring1B inhibited proliferation and differentiation of hepatic stem/progenitor cells and thereby inhibited hepatic organogenesis. These events were characterized by derepression of cyclin-dependent kinase inhibitors (CDKIs) Cdkn1a and Cdkn2a, known negative regulators of cell proliferation. We conducted clonal culture experiments with hepatic stem/progenitor cells to investigate the individual genetic functions of Ring1B, Cdkn1a, and Cdkn2a. The data showed that the cell-cycle inhibition caused by Ring1B depletion was reversed when Cdkn1a and Cdkn2a were suppressed simultaneously, but not when they were suppressed individually. Conclusion: Our results show that expansion of hepatic stem/progenitor cells requires Ring1B-mediated epigenetic silencing of Cdkn1a and Cdkn2a, demonstrating that Ring1B simultaneously regulates multiple CDKIs in tissue stem/progenitor cells. (Hepatology 2014;60:323-333)

Abbreviations
AAT

α1-antitrypsin

AFP

alpha-fetoprotein

Alb

albumin

BrdU

5-bromo-2′-deoxyuridine

CDKI

cyclin-dependent kinase inhibitor

ChIP

chromatin immunoprecipitation

Ck

cytokeratin

CKO

conditional knockout

DKO

double knockout

E

embryonic day

ES

embryonic stem

FCM

flow cytometry

Ggt

gamma-glutamyl transpeptidase gene

GO

gene ontology

Krt

keratin gene

PcG

polycomb group

qRT-PCR

quantitative reverse-transcriptase polymerase chain reaction

Rb

retinoblastoma protein

SD

standard deviation

TAM

tamoxifen

Tat

tyrosin aminotransferase gene

WT

wild type

Tissue stem cells play important roles in organ development and tissue repair.[1, 2] For example, embryonic development of the liver, the largest internal organ, starts on embryonic day 8.5 (E8.5) in mice, when hepatic stem/progenitor cells expressing alpha-fetoprotein (AFP) begin to proliferate; these cells ultimately differentiate into hepatocytes and bile duct epithelial cells.[3-5] Flow cytometric (FCM) studies of mouse embryonic liver from E9.5 to E14.5 have revealed the presence of hepatic stem/progenitor cells with a strong potential for proliferation and bidirectional differentiation.[6] These findings suggest that hepatic stem/progenitor cells actively replicate themselves to create a renewing pool of primitive cells. Self-renewal is one of the defining characteristics of stem cells. Knowledge of the mechanisms underlying stem cell self-renewal is crucial for understanding the full potential and nature of stem cells. However, much remains to be learned about the molecular basis of those mechanisms.

In order to maintain the undifferentiated state and stemness, it is important for embryonic stem (ES) cells and other pluripotent stem cells to keep their developmental genes transcriptionally silent, yet poised for activation, in the course of repeated cycles of cell division. Several researchers have postulated that histone-modification-based reversible mechanisms of gene silencing are critical for maintenance of stem cell pluripotency and plasticity.[7, 8] In stem cells, “bivalent domains” containing both suppressive (silencing) histone modifications (H3K27me3 and H2AK119ub1) and activating (positive) histone modifications (H3K4me3) are present in the promoter regions of octamer-binding transcription factor 3 and 4, SRY (sex determining region Y)-box 2, as well as other developmentally important genes known to be poised for activation.[9, 11] Bivalent domains are also present in the promoter regions of genes involved in neuronal differentiation of embryonic stem cells.[12] These findings underscore the importance of bivalent domains as molecular switches that regulate the pluripotency of embryonic stem cells. Moreover, bivalent domains have been found in hematopoietic and other tissue-specific stem cells.[13, 14] Thus, there is a growing body of evidence implicating bivalent domains in the molecular machinery mediating gene silencing and maintenance of pluripotency across different types of stem cells. Therefore, in order to fully understand the mechanisms of stem-cell self-renewal and differentiation, it is essential to identify the stem cell genes prone to both suppressive and activating histone modifications and clarify their role in stem cell regulation.

Molecular studies of bivalent domains have revealed that polycomb-group (PcG) proteins suppress transcription, whereas trithorax-group proteins activate transcription.[9, 10] PcG histone modifications often overlap with multiple transcription start sites of known genes and therefore have key developmental functions.[11] PcG proteins exist in subnuclear loci as distinct multimeric protein complexes consisting of more than 20 components. Ring1B, one of the PcG-complex proteins, is a key mediator of transcriptional suppression of PcG target genes.[13] Ring1B is a major E3 ubiquitin ligase involved in histone H2AK119 monoubiquitination. H2A monoubiquitination suppresses recruitment of RNA polymerases and induces chromatin aggregation, leading to transcriptional suppression.[14] Ring1B-null mouse embryos manifest abnormal gastrulation and cell-cycle inhibition and do not develop beyond 10-11 days in utero.[15] In addition, mouse embryos lacking the PcG protein, Bmi1, a factor that interacts with Ring1B and facilitates its E3 ligase activity, develop and survive past birth, but exhibit significantly low counts of hematopoietic and neural stem cells in adulthood.[16, 17] These observations indicate that PcG-mediated suppressive histone modifications are crucial for stem cell maintenance.

We found that forced expression of Bmi1 in hepatic stem/progenitor cells enhanced the proliferative capacity of cells and resulted in tumor formation; these results demonstrate that PcG proteins play a fundamental role in regulating stem cell self-renewal and carcinogenesis.[18, 19] Elucidating the functions of PcG protein complexes will require identification of their target genes. However, at present, little is known about their downstream target genes in tissue stem cells. For example, cyclin-dependent kinase inhibitor (CDKI) Cdkn2a is a classic downstream Bmi1 target in hematopoietic stem cells. However, deficiency of Cdkn2a does not restore the ability of Bmi1-deficient stem cells to self-renew, implying involvement of other target genes in stem cell expansion and differentiation.[20]

In this study, we investigated the role of Ring1B in hepatic stem/progenitor cells and identified downstream genes involved in proliferation and differentiation of those cells.

Materials and Methods

Mice

Ring1B conditional knockout (CKO) mice [Rosa26-CRE-ER(T2)+/− ;Ring1Bflox/flox],[21] Cdkn1a−/− mice (The Jackson Laboratory, Bar Harbor, ME), Cdkn2a−/− mice (The Jackson Laboratory), and the mice generated by crossing these mice were used. Tamoxifen (TAM; Sigma-Aldrich, St. Louis, MO) solution was administered into the abdominal cavity of pregnant mice at 1 mg/body/day for 3 days. The littermates obtained by crossing of the parent mice having the CRE-ER(T2) heteroallyl were compared. Genotypes of the mice were verified by polymerase chain reaction (PCR). The primer sequences (forward and reverse) are listed in Supporting Table 1. Mice were bred and maintained according to our institutional guidelines for the use of laboratory animals.

Fluorescence-Activated Cell Sorter Analysis and Cell Culture

For in vitro assays, single-cell suspensions of liver cells were prepared from E13.5 mouse embryos. Isolation and culture of c-Kit CD49f+/low CD29+ CD45 TER119 cells were performed as described previously.[6] Sorted cells were seeded (2,000 cells per dish) in 35-mm type IV collagen-coated dishes (Becton-Dickinson Labware, Bedford, MA). TAM was added at a final concentration of 800 nM after 24 hours of culture. After 5 or 7 days of culturing, expression analysis of differentiation markers and composition of the cell population were assessed.

Microarray Analysis

Total RNA was isolated from nonhematopoietic (CD45 TER119) cells of an E13.5 Ring1B-depleted and control mouse (treated with TAM since E8.5) liver using TRIzol Reagent (Life Technologies, Carlsbad, CA). Total RNA was labeled and hybridized to a 3D-Gene Mouse Oligo chip 24k (Toray Industries Inc., Tokyo, Japan), according to the manufacturer's instructions. Hybridization signals were scanned and processed by 75% percentile shift normalization in GeneSpring GX11. We considered only probe sets whose expression values differed 2-fold or more between Ring1B-deficient and control cells. For up-regulated genes, GO (gene ontology) term enrichment analysis was performed according to a method described by the GO Consortium (http://www.geneontology.org/index.shtml). The raw data of the microarray analysis have been deposited in the Gene Expression Omnibus database (GSE54116).

Results

Ring1B Is Essential for Early Organogenesis of the Liver

To investigate the role of Ring1B in liver organogenesis, we dissected tissue from Ring1B CKO fetuses homozygous for a floxed Ring1B allele that can be inactivated by Cre-mediated recombination. These mice ubiquitously expressed the cytoplasmic Cre recombinase/estrogen receptor fusion protein, Cre-ER(T2), whose activity is controlled by translocation to the cell nucleus induced by 4-hydroxy-TAM (Supporting Fig. 1A). Almost total deletion of the Ring1B first coding exon and Ring1B protein suppression required 2 days after TAM administration, as shown by PCR and western blotting analysis in Ring1B CKO embryos (Supporting Fig. 1B,C). The E13.5 Ring1B CKO mice that had received TAM since E8.5 had a liver whose size and weight were smaller than that of littermate control mice (Fig. 1A,C). Interestingly, in E13.5 Ring1B CKO embryos, organogenesis of the liver was markedly inhibited, compared with other organs (Supporting Figs. 1D,E). To clarify the role of Ring1B in liver organogenesis, we induced Ring1B depletion at various developmental stages. The results showed that in both E15.5 and E17.5 fetal livers, which had undergone TAM administration since E10.5 and E12.5, respectively, no difference was observed in the size and weight, compared to control livers (Figs. 1B,C). Moreover, in Ring1B-depleted livers, cytokeratin (Ck) 8/18-positive epithelial cell proliferation, assessed as 5-bromo-2′-deoxyuridine (BrdU) incorporation, was significantly lower than in control livers on E13.5 (Fig. 1D). Collectively, these results show that Ring1B plays an important role in proliferation of epithelial cells at the initial stage of liver organogenesis.

Figure 1.

Ring1B depletion significantly decreased proliferation of epithelial cells at the early stage of liver development. (A) Morphology (left panels) and hematoxylin and eosin staining (right panels) of control and Ring1B CKO mouse embryos on embryonic day 13.5 (E13.5) injected with TAM since E8.5. The Ring1B CKO liver was much smaller than that of control mice (scale bars: 2 mm). (B) Morphology of control and Ring1B CKO embryos and liver on E15.5 and E17.5 injected with TAM since E10.5 and E12.5, respectively. The Ring1B CKO liver (E15.5 and E17.5) was similar to the control. (C and D) Weight (C) and BrdU-positive rate in Ck8/18-positive epithelial cells (D) in liver from control and Ring1B CKO embryos on E13.5 (left panel) and E15.5 (right panel). Bars represent mean ± standard error of the mean (C: n = 8; D: n = 24). (E) AFP, BrdU, and Ck8/18 coimmunofluorescent images of control and Ring1B CKO liver (E13.5). Scale bars: 100 μm; arrow: AFP+ BrdU+ Ck8/18+ cells. (F) Subsequent to coimmunofluorescence staining for AFP, BrdU, and Ck8/18, the percentage of each group of cells per field was calculated. Bars represent mean ± SD (n = 3). (G) Gene expression analysis by qRT-PCR for E13.5 CD45 TER119 liver cells from control and Ring1B CKO embryos. All data were normalized to the values of a control liver. (H) Frequency of a c-Kit CD49f+/low CD29+ population in CD45 TER119 cells was measured with fluorescence-activated cell sorter analysis. Bars represent mean ± SD (n = 3-4). *P < 0.05; **P < 0.01; Mann-Whitney's U test.

Ring1B Depletion Inhibited Proliferation and Differentiation of Hepatic Stem/Progenitor Cells in Early Development of the Liver

Hepatic stem/progenitor cells that maintain an undifferentiated state and proliferate actively exist in large numbers in the liver during E8.5-E12.5, and subsequently their numbers decline. To verify this, we assessed liver cell number at the early stage of organogenesis. The result showed that the number on E9.5 and E13.5 was 5 × 102 and 1 × 107 cells, respectively, and the expansion ratio of 2 × 104 was observed in only 4 days. On the other hand, during E14.5-E17.5, the number increased only 2-fold (Supporting Fig. 2A). These results confirm that pronounced cell growth occurs during early liver organogenesis. Expression of AFP and incorporation of BrdU were analyzed for assessment of hepatic stem/progenitor cell frequency during liver organogenesis. As a result, among Ck8/18-positive cells, frequency of AFP-positive and proliferating hepatic stem/progenitor cells was high: 37.3% ± 9.1% on E11.5 and 37.6% ± 3.3% on E13.5; it declined to 2.9% ± 2.8% on E17.5 (mean ± standard deviation [SD], here and below; Supporting Fig. 2B,C). Because liver cell number increased 33.6-fold and frequency of AFP-positive cells remained unchanged during E11.5-E13.5, it appeared that hepatic stem/progenitor cells expanded the pool of AFP-positive cells during this period. To examine whether Ring1B is involved in hepatic stem/progenitor cell proliferation and differentiation, we evaluated Ring1B-depleted liver tissue on E13.5. The AFP-positive rate in epithelial cells was reduced 0.91 ± 0.07 times in Ring1B CKO mice, and frequency of AFP-positive proliferating cells decreased 0.43 ± 0.07 times (Fig. 1E,F). Furthermore, expression of undifferentiated marker genes (alpha-fetoprotein Afp, hematopoietically expressed homeobox, and delta-like 1) and frequency of c-Kit CD49f+/low Cd29+ hepatic stem/progenitor cells was reduced in CD45 TER119 nonblood cells from Ring1B CKO fetal liver on E13.5 (Fig. 1G,H). These results suggested that the reduced frequency of mutant hepatic stem/progenitor cells resulted from a decrease in proliferative rate of Ring1B-deficient cells.

In addition, expression of hepatocyte differentiation marker genes albumin (Alb), Aat, and hepatcoyte nuclear factor 1a decreased in the hepatic stem/progenitor cell population (c-Kit CD49f+/low CD29+ CD45 TER119) derived from Ring1B CKO livers at E13.5 (Supporting Fig. 3A). We also confirmed the reduction on expression of cholangiocyte marker genes keratin (Krt)19, Krt7, and gamma-glutamyl transpeptidase (Ggt; Supporting Fig. 3B). Furthermore, frequency of α1-antitrypsin (AAT)-positive hepatic differentiated cells among Ck8/18-positive cells decreased 0.86-fold in Ring1B-deficient liver at E15.5, compared with to wild type (WT; Supporting Fig. 3C,D). Furthermore, expression of hepatocyte differentiation marker genes Alb, Aat, and tyrosin aminotransferase (Tat) and cholangiocyte marker genes Krt7 and Ggt decreased in the CD45 TER119 cells obtained from Ring1B-deficient livers at E15.5 (Supporting Fig. 3E,F). These results suggested that hepatic stem/progenitor cells were derailed in the normal process of differentiation into hepatoctytes and cholangiocytes by Ring1B depletion at an early stage of liver development.

To evaluate the involvement of Ring1B on differentiation in the late stage of liver development, we analyzed Ring1B CKO embryos at E17.5 (+TAM E10.5-12.5). As a result, Ring1B CKO down-regulated differentiation markers (Supporting Fig. 3G,H), but the blockade of differentiation in mid-stage CKO liver (+TAM E10.5-12.5) was weaker than that in early-stage CKO liver (+TAM E8.5-10.5; Supporting Fig. 3E,F). Taken together, Ring1B regulates proliferation and differentiation of hepatic stem/progenitor cells in early-fetal liver development. Even though inhibition of differentiation by Ring1B depletion in mid-fetal liver development was observed, it was a little weaker than at the early stage.

Ring1B Is Essential for Clonal Expansion of Hepatic Stem/Progenitor Cells

To elucidate the role of Ring1B in hepatic stem/progenitor cell expansion, we used FCM to isolate c-Kit CD49f+/low CD29+ CD45 TER119 cells from E13.5 fetal liver, which show multipotential and self-renewing capacities under conditions of clonal analysis. The positive rate of Afp expression in the c-Kit CD49f+/low CD29+ CD45 TER119 fraction was confirmed to have a high frequency (68.5% ± 5.2%; Supporting Fig. 4A,B). Ring1B-deficient cell clones consisted of hepatic stem/progenitor cells derived from colonies whose size was significantly smaller than that of control colonies (Fig. 2A,B). To test whether the Ring1B deletion is related to differentiation of mutant hepatic stem/progenitor cells, we evaluated the potential of hepatic stem/progenitor derived colonies to generate two hepatic cell lineages. After 5 days in culture, we found a significantly smaller number of Ring1B-lacking colonies that contained more than 90 cells, including hepatocytes (Alb-immunoreactive cells) and cholangiocytes (Ck7-immunoreactive cells; Fig. 2C,D). These results indicate that Ring1B is essential for proliferation and differentiation of hepatic stem/progenitor cells.

Figure 2.

Ring1B depletion inhibited clonal expansion of hepatic stem/progenitor cells. (A) PCR analysis of control and Ring1B CKO hepatic stem/progenitor cells cultured with TAM show almost total deletion of Ring1B on day 1 after treatment under proliferation conditions. (B) Subsequent to coimmunofluorescence staining for Alb and Ck7, the count of Alb+, Ck7+, and Alb/Ck7+ colonies was determined. (C) Coimmunofluorescence staining for Alb and Ck7 was conducted on clonal colonies formed by c-Kit CD49f+/low CD29+ CD45 TER119 cells isolated from a control and Ring1B CKO liver after 5 days in culture. (D) Clonal colony formation upon 5 days of single-cell culture of c-Kit CD49f+/low CD29+ CD45 TER119 cells isolated from either control or Ring1B CKO livers. Arrow: a hepatic stem/progenitor cell on the first day; scale bar: 100 μm. (E-G) Effect of Ring1B depletion on self-renewal capacity. Among the clones derived from untreated Cre+/− Ring1Bflox/flox c-Kit CD49f+/low CD29+ CD45 TER119 cells, TAM was added in those that kept growing beyond 3 weeks. In 5 days after TAM addition, expanding subcultured cells were again clonally sorted and cultured with a single cell. (E) Clonal colony formation upon 5 days of single-cell culture. (F) Coimmunofluorescence staining for Alb and Ck7 was conducted on clonal colonies. (G) The number of total colonies and large clones consisting of more than 90 cells were counted at day 5 of subculture. Scale bar: 100 μm; red: Alb; green: CK7; blue: 4′,6-diamidino-2-phenylindole). Results are mean ± SD (n = 3). P values (asterisks) are from Mann-Whitney's U test. *P < 0.05; **P < 0.01.

Next, we examined the role of Ring1B on the self-renewal of hepatic stem/progenitor cells using replating assay. c-Kit CD49f+/low CD29+ CD45 TER119 cells were sorted from livers in control and Ring1B CKO mice and cultured for more than 3 weeks. Cells obtained from control mice kept proliferating even after replating and could constitute bipotent colonies from a single cell, whereas most of the cells from Ring1B CKO mice stopped proliferating and could only form a significantly reduced number of the large colonies (Fig. 2E-G). These results showed that Ring1B plays a role in regulation of self-renewal of hepatic stem/progenitor cells.

Cdkn1a and Cdkn2a Are Derepressed in the Ring1B-Deficient Liver Cells

Next, we searched for genes downstream of Ring1B that are involved in hepatic stem/progenitor cell proliferation and differentiation. Because Ring1B is involved in transcriptional repression, we focused on genes derepressed by Ring1B CKO. Genome-wide messenger RNA profiling (microarray analysis) of CD45 TER119 cells from a Ring1B-deficient fetal liver on E13.5 showed that 919 genes were derepressed (activated) more than 2-fold, compared to control mice. GO analysis showed a significant presence of categories related to cell proliferation (Fig. 3A; Supporting Table 2). Among these candidates (target genes of Ring1B), we hypothesize that CDKIs, which regulate progression through the G1 phase of the cell cycle, play an important role in regulation of hepatic stem/progenitor cell proliferation by Ring1B. To validate the microarray analysis data, we carried out quantitative reverse-transcriptase (qRT)-PCR on a number of selected targets. Among the CDKI genes, Cdkn1a, Cdkn2a, and Cdkn2d were up-regulated in mutant CD45 TER119 liver cells on E13.5 (Fig. 3B). Moreover, examination of expression of these genes during liver development revealed that expression of Cdkn1a and Cdkn2a was low before E13.5 and increased after E15.5, at the beginning of differentiation into two lineages (Fig. 3C and Supporting Fig. 5B). On comparing expression of CDKIs among several organs in E13.5, Ring1B CKO embryos confirmed that derepression of Cdkn1a specifically occurred in the liver (Supporting Fig. 5A). To distinguish between direct (Ring1B-bound) and indirect targets (not bound to Ring1B), we used the chromatin immunoprecipitation (ChIP)/PCR method. We found that the promoter regions of all CDKI genes derepressed in Ring1B CKO mice were bound by Ring1B (Fig. 3D). Overall, these data suggest that among the genes downstream of Ring1B, Cdkn1a and Cdkn2a play an important role in regulation of hepatic stem/progenitor cell proliferation by Ring1B.

Figure 3.

Cdkn1a and Cdkn2a expression was derepressed in Ring1B-depletedliver cells. (A) Significant enrichment of profiles of the 919 derepressed genes in Ring1B CKO liver cells. GO analyses were performed for the differentially expressed genes. P value of each GO term is shown (2.60 × 10−3, 1.44 × 10−2, 1.98 × 10−2, 3.83 × 10−2, and 8.17 × 102). (B) Gene expression analysis of CDKIs by qPCR in CD45 TER119 cells from a control and Ring1B CKO mouse liver (E13.5). All data were normalized to the values of glyceraldehyde-3-phosphate dehydrogenase. Results are mean ± SD (n = 3). P values (asterisks) are from Mann-Whitney's U test. *P < 0.05; **P < 0.01. (C) qPCR analysis showed relative expression of CDKI genes in CD45 TER119 liver cells during liver development. (Results are mean ± SD; n = 3.). (D) ChIP PCR analysis of CD45 TER119 cells from a control and Ring1B CKO mouse liver in the CDKI locus promoter region using anti-Ring1B and anti-rabbit (loading control) antibodies. P values (asterisks) are from Mann-Whitney's U test. *P < 0.05; **P < 0.01.

Suppressing Both Cdkn1a and Cdkn2a Expression by Ring1B Is Required for Proliferation of Hepatic Stem/Progenitor Cells

We performed a rescue assay to examine whether Cdkn1a and Cdkn2a, identified as Ring1B downstream players, are involved in regulation of proliferation and differentiation of hepatic stem/progenitor cells. To examine whether the phenotype in Ring1B-depleted hepatic/stem progenitor cells could be rescued, in addition to loss of Ring1B, inhibition of Cdkn1a and/or Cdkn2a expression was performed. At first, we analyzed embryos of CreER(T2)+/− Ring1Bflox/flox Cdkn2a−/−. We isolated c-Kit CD49f+/low CD29+ CD45 TER119 cells from these mice and examined the effect of Ring1B depletion on proliferation and differentiation of hepatic stem/progenitor cells in a clonal culture system. As a result, growth inhibition of hepatic stem/progenitor cells caused by Ring1B depletion was not hampered (Supporting Fig. 6C,D). Moreover, evaluation of liver organogenesis in E13.5 embryos that received TAM since E8.5 revealed that Ring1B-Cdkn2a double-deficient mice did not recover the reduced size of the liver caused by the Ring1B single knockout (Supporting Fig. 7A). Liver organogenesis of Cdkn2a single-deficient mice was similar to that of WT mice (data not shown).

Subsequent to this, we analyzed the clonal colony formation process of CreER(T2)+/−, Ring1Bflox/flox, and Cdkn1a−/− mouse cells. As a result, growth inhibition of hepatic stem/progenitor cells caused by Ring1B deletion was not rescued (Fig. 4A,B). Evaluation of liver organogenesis in E13.5 mice with induced Ring1B deletion since E8.5 revealed that the liver size reduction caused by Ring1B single knockout was not reversed when the Cdkn1a gene was also inactivated (double-knockout [DKO] mice; Supporting Fig. 7B,C). In liver of DKO mice (Ring1B null and Cdkn1a null), on E13.5, examination of the frequency of AFP+/BrdU+ cells among Ck8/18-positive cells showed that the number of AFP+/BrdU+ cells did not increase (Supporting Fig. 7D,E). In liver of DKO and single-knockout mice, on E13.5, expression analysis of AAT, a hepatocyte-specific marker, confirmed that the AAT-positive rate was not different between control and DKO mice (Supporting Fig. 7F,G). In CD45 TER119 nonblood cells from liver of the DKO (Ring1B null and Cdkn1a null) or single-knockout mice (Ring1B null), there was no significant difference in expression of hepatocyte and cholangiocyte marker genes, compared to control mice (Supporting Fig. 7H). Overall, these data showed that proliferation and differentiation of hepatic stem/progenitor cells were not caused by suppression of Cdkn1a or Cdkn2a by Ring1B.

Figure 4.

Cdkn1a and Cdkn2a cosuppression rescued clonal expansion in Ring1B-depleted hepatic stem/progenitor cells. (A-D) Colony assays of Ring1B-Cdkn1a DKO hepatic stem/progenitor cells. (A) Bright-field images of colonies derived from control cells, Ring1B-Cdkn1a DKO cells and DKO cells transfected with Cdkn2a microRNA (miRNA) on day 7 of culturing. Venus protein expression of transfected cells is shown in insets. (B) Inhibition of expression of Cdkn2a in Ring1B-Cdkn1a DKO cells markedly reversed the total number of large colonies containing more than 90 cells on day 7 of culturing (results are mean ± SD; n = 3-8). (C) Immunocytochemical analyses of colonies derived from control cells, Ring1B-Cdkn1a DKO cells, and DKO cells transfected with Cdkn2a miRNA on day 7 of culturing. Alb (red), Ck7 (green), and DAPI (blue) expression was merged. (D) The absolute number of Alb+ cells in colonies derived from control cells, DKO cells, and DKO cells transfected with Cdkn2a miRNA on day 7 of culturing (results are mean ± SD; n = 94-179). (E) Gene expression analysis of clonal colonies after 7 days of cell culture, derived from DKO cells transfected with Cdkn2a miRNA. “Control” means Cre-negative Cdkn1a null mutant mouse cells introduced with the negative control mRNA. Scale bar: 100 μm (results are mean ± SD; n = 3). P values (asterisks) are from Mann-Whitney's U test. *P < 0.05; **P < 0.01. DAPI, 4′,6-diamidino-2-phenylindole.

To investigate the proliferative and differentiative properties of hepatic stem/progenitor cells, we isolated c-Kit CD49f+/low CD29+ CD45 TER119 cells from Ring1B-Cdkn1a DKO mice and induced Ring1B deletion and Cdkn2a suppression in the clonal culture system (Supporting Fig. 6A,B). As a result, we confirmed that simultaneous inhibition of Ring1B, Cdkn1a, and Cdkn2a in the clonal culture system recovers the colony-forming ability of hepatic stem/progenitor cells possessing the Ring1B single deletion (Fig. 4A,B and Supporting Fig. 6C,D). Next, we also analyzed the role of Cdkn1a and Cdkn2a in differentiation of hepatic stem/progenitor cells. The frequency of Alb-expressing cells in colonies on culture day 7 decreased significantly in the Ring1B single-knockout group. Subsequent to this, we checked Alb-positive cells in colonies that lacked expression of Ring1B/Cdkn1a/Cdkn2a. Immunostaining revealed that recovery of the Alb-positive differentiated cell number was not observed in triple-inhibition cells (Ring1B/Cdkn1a/Cdkn2a), suggesting that the inhibited hepatocyte differentiation was not restored to normal levels (Fig. 4C,D and Supporting Fig. 6E,F). Next, we attempted to reveal more details of hepatic maturation level in formed colonies with qRT-PCR. Analysis for expression of hepatocyte differentiation-related genes reported on previously[22] revealed that expression of the genes, Alb, Aat, Tat, and acyl-coenzyme A oxidase 1, was partially restored by triple inhibition (depletion of Ring1B/Cdkn1a and inhibition of Cdkn2a; Fig. 4E). Deletion of Ring1B and a knockdown of Cdkn1a in C-Kit CD49f+/low CD29+ CD45 TER119 cells isolated from a CreER(T2)+/− Ring1Bflox/flox Cdkn2a−/− mouse were also performed for gene expression analysis of differentiation markers in colonies at day 7 of culture. As a result, expression of the differentiation markers in the colonies was either the same or slightly lower, compared to colonies from Ring1B-Cdkn2a-null cells (Supporting Fig. 6G). These results suggested that Cdkn1a deletion, but not incomplete inhibition, from an earlier step of liver development would be crucial for differentiation, rather than Cdkn2a.

Overall, Ring1B is essential for growth of hepatic stem/progenitor cells through inhibition of expression of both Cdkn1a and Cdkn2a. Downstream genes other than Cdkn1a and Cdkn2a may be necessary for progressing hepatic stem/progenitor cell differentiation steps.

Discussion

In this study, we demonstrate that Ring1B plays an essential role in proliferation of hepatic stem/progenitor cells during their expansion phase by suppressing both Cdkn1a and Cdkn2a, and this notion indicates that stem cell proliferation is regulated by a combination of several CDKI proteins (Fig. 4 and Supporting Fig. 6). Active cyclin-dependent kinases phosphorylate and inactivate members of the retinoblastoma protein (Rb) family, which are negative regulators of progression through the G1 and S phases, leading to induction of genes regulated by E2F and cell proliferation. Two families of CDKIs control those events. Cdkn1a is the member of the Cip/Kip family of CDKIs. Cdkn2a is the member of the INK4 proteins and specifically targets the cyclin D-dependent kinases.[23-25] Two CDKI target molecules, Cdk2 and Cdk4/6, can independently inactivate Rb/E2F and promote G1/S transition.[23, 24] Therefore, it would be logical that suppression of both Cdkn1a and Cdkn2a is necessary for maintenance of the high proliferative potential during stem/progenitor cell expansion. Gene expression analysis revealed that Cdkn1a and Cdkn2a expression was low before E13.5 and increased after E15.5 (Fig. 3C). Analysis of BrdU incorporation revealed that Cdkn1a and Cdkn2a expression shows an inverse correlation with the frequency of AFP/BrdU double-positive cells (Fig. 3C and Supporting Fig. 2B,C). When Ring1B was deleted at the early stage of liver development, impairment of liver development was observed (Fig. 1A,C,D), and, at that time, expression of Cdkn1a and Cdkn2a was strongly derepressed (Fig. 3B). In vitro colony assay revealed that simultaneous suppression of Cdkn1a and Cdkn2a is necessary for regulation of hepatic stem/progenitor cell proliferation by Ring1B (Fig. 4 and Supporting Fig. 6). Therefore, it is considered that Ring1B regulates expansion of hepatic stem/progenitor cells through repression of both Cdkn1a and Cdkn2a at the early stage of liver development before E13.5. Upon treatment with TAM from E8.5, hepatic stem/progenitor expansion was inhibited by the start of Cdkn1a and Cdkn2a derepression. Therefore, significant inhibition of early liver development was observed in Ring1B CKO mice by injection of TAM from E8.5.

The limited effectiveness of Ring1B/Cdkn1a/Cdkn2a triple inhibition against the hepatic stem/progenitor cell differentiation processes suggests that Ring1B target genes, except Cdkn1a and Cdkn2a, are required for differentiation. Gene expression analysis of CD45 TER119 cells from an E13.5 Ring1B-deficient mouse liver revealed that 919 genes were derepressed by Ring1B CKO. These genes should include some candidates controlling hepatic stem/progenitor cell differentiation (Fig. 3A). In the future, genes regulating hepatic/stem progenitor cell differentiation will probably be identified by microarray analysis to identify functionally relevant direct targets among the potential 919 target genes. Progress in understanding the epigenetic properties of hepatic stem/progenitor cells, as well as some other stem cell populations, will be crucial for their practical application—for example, to ensure that differentiated cells derived from stem cells are fully committed to their lineage and have effectively shut down alternative options. Further characterization of the epigenetic states of cells between multipotent and terminal differentiation stages will therefore be important for understanding and refining the identity of the stem cell.

PcG proteins play an important role in cancer cells as well as in normal stem cells.[26-29] Stem cells rely on PcG proteins to reversibly suppress genes encoding transcription factors required for proliferation and differentiation. We and others hypothesize that acquisition of epigenetic changes in the promoter region of these suppressed genes could disrupt stem cell phenotypes and initiate abnormal clonal expansion and thereby predispose tissue to cancer.[30-32] Therefore, Ohm et al. proposed that epigenetic lesions might be frequent in the stem cell niche owing to, for example, chronic inflammation, injury, and poor nutrition. Disruption of stem cell signals from the niche microenvironment involves an initiating mutation within the subpopulation of epigenetically disrupted progenitor cells at the earliest stages of a neoplasm.[30] Malignant transformation of hepatocytes is believed to occur, regardless of the etiological agent, through a pathway of increased liver cell turnover, induced by chronic liver injury and regeneration—for instance, in the context of hepatitis C virus infection and inflammation. Therefore, those epigenetic events and a loss of a tumor suppressor gene(s) during the first step of hepatocarcinogenesis may directly lead to a malignant phenotype of liver tissue.[33, 34] In a previous study, we demonstrated that self-renewal disrupted by Bmi1 drives malignant transformation in hepatic stem/progenitor cells; tumors originating from those cells can be diagnosed as combined hepatocellular carcinoma and cholangiocarcinoma.[18, 19] Bmi1 stimulates Ring1B's E3-ubiquitin ligase activity. Thus, these observations could explain how Ring1B can contribute to carcinogenesis in the liver. In the present study, we show that Ring1B is required for hepatic stem/progenitor cell expansion on E10.5-E13.5, but loss of Ring1B's function has no effect on cell proliferation of adult hepatocytes. If a further study identifies a specific inhibitor for either Ring1B or H2A119 ubiquitination or both in the Cdkn1a- and Cdkn2a-signaling pathway, we may be able to develop more effective cancer treatments that are less toxic for hepatocytes by suppression of disrupted stem/progenitor cell expansion. In conclusion, characterization of the Ring1B network in tissue stem cells is important for understanding both cancer progression and normal identity of stem cells.

Acknowledgment

The authors thank Mr. K. Matsunaga, Ms. A. Makita, K. Asada, and N. Tsuchida for their technical assistance, Mr. A. Tanaka for his technical help with flow cytometry, and Dr. H. Miyoshi for the lentiviral vectors. The authors also thank Ms. N. Hijikata and N. Sasaki for their laboratory assistance, as well as all the lab members in our lab for their critical comments and valuable discussions.

Ancillary