Adult Stem Cells Exhibit Global Suppression of RNA Polymerase II Serine-2 Phosphorylation§


  • Rasmus Freter,

    Corresponding author
    1. Ludwig Institute for Cancer Research, Nuffield Department of Clinical Medicine, University of Oxford, United Kingdom
    • Ludwig Institute for Cancer Research, Nuffield Department of Clinical Medicine, University of Oxford, United Kingdom
    Search for more papers by this author
  • Masatake Osawa,

    1. Cutaneous Biology Research Center/Center for Regenerative Medicine, Massachusetts General Hospital/Harvard Medical School, Massachusetts, USA
    Search for more papers by this author
  • Shin-Ichi Nishikawa

    1. Ludwig Institute for Cancer Research, Nuffield Department of Clinical Medicine, University of Oxford, United Kingdom
    Search for more papers by this author

  • Author contributions: R.F.: conception and design, collection and/or assembly of data, data analysis and interpretation, manuscript writing; M.O.: conception and design, final approval of manuscript; S.-I.N.: conception and design, data analysis and interpretation, manuscript writing, final approval of manuscript.

  • Disclosure of potential conflicts of interest is found at the end of this article.

  • §

    First published online in STEM CELLS EXPRESS July 16, 2010.


Adult stem cells, which are characterized by their capacity for self-renewal and differentiation, participate in tissue homeostasis and response to injury. They are thought to enter a state of relative quiescence, known as reversible cell cycle arrest, but the underlying molecular mechanisms remain poorly characterized. Previous data from our laboratory has shown that housekeeping gene expression is downregulated in melanocyte stem cells (MelSCs), suggesting a global suppression of mRNA transcription. We now show, using antibodies against specific phosphorylated forms of RNA polymerase II (RNApII), that adult MelSCs do not undergo productive mRNA transcription elongation, while RNApII is activated and initialized, ready to synthesize mRNA upon stimulation, and that the RNApII kinase CDK9 is absent in adult MelSCs. Interestingly, other adult stem cells also, including keratinocyte, muscle, spermatogonia, and hematopoietic stem cells, showed a similar absence of RNApII phosphorylation. Although it is difficult to show the functional significance of this observation in vivo, CDK9 inhibition resulted in enhanced survival of cells that are deprived from survival factors. We conclude that the absence of productive mRNA transcription is an early, specific, and conserved characteristic of adult stem cells. Downregulation of mRNA transcription may lead to decreased rates of metabolism, and protection from cellular and genetic damage. Screening heterogeneous tissues, including tumors, for transcriptionally quiescent cells may result in the identification of cells with stem cell-like phenotypes. STEM CELLS 2010; 28:1571–1580.


Adult stem cells have the unique ability to undergo sustained self-renewal and differentiation, which is essential for tissue homeostasis and response to injury. These cells are resistant to cytotoxic stress but retain the capacity for activation upon stimulation. Although active induction and maintenance of quiescence are thought to be key mechanisms underlying these features of the stem cell system, their molecular basis is unresolved.

Quiescence is commonly defined as a reversible exit from the cell cycle. Quiescent stem cells have been analyzed in terms of their cell cycle regulation [1], control of cellular metabolism [2] and interaction with their special microenvironment, the niche [3], but analysis of their general transcriptional machinery has been scarce. In other quiescent cell systems, such as those induced by serum starvation [4, 5], resting lymphocytes [6, 7] and yeast cells in the stationary phase [8, 9], the global suppression of mRNA synthesis has been implicated as a factor in quiescence and the control of cell cycle and metabolism.

The mRNA transcription cycle of transcription initiation and elongation, and the subsequent release of RNA polymerase II (RNApII) from the DNA is tightly regulated by phosphorylation of the C-terminal domain (CTD) of RNApII [10, 11]. The CTD of mammalian RNApII is composed of 52 repeats of the consensus sequence YS2PTS5PS. Transcription initiation requires phosphorylation of Serine five (Ser5) of the CTD by TFIIH, consisting of CDK7 and CyclinH. Phosphorylation of Serine two (Ser2) of the CTD by p-TEFb, a heterodimer of CDK9 and Cyclin T1, T2, or K, triggers productive transcription elongation, mRNA processing, and the release of the mature mRNA [12, 13]. Ser2 phosphorylation and transcription elongation is the critical target for eukaryotic gene expression [14, 15]. Inhibition of CDK9 function by 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole (DRB) or flavopiridol results in degradation of most mRNA [15] and induces apoptosis [16, 17]. Similarly, knockdown of CDK9 results in complete absence of mRNA synthesis and embryonic lethality [18, 19].

Most cells, including terminally differentiated and senescent cells, actively synthesize mRNA. In these cells, RNApII is phosphorylated on both Ser2 and Ser5, independent of the cell cycle [20, 21]. Quiescent cells, such as primary T lymphocytes, are characterized by an almost complete absence of RNApII phosphorylation [7, 21]. RNApII in yeast cells in the stationary phase [8, 9] or on Drosophila species heat shock genes [12, 22] is phosphorylated on Ser5, but not on Ser2. In this situation, the rapid activation of gene transcription is possible upon stimulation, such as the addition of nutrients or heat shock. Taken together, analysis of the specific phosphorylated sites in RNApII reveals phases of productive mRNA elongation or paused mRNA transcription initiation.

Melanocyte stem cells (MelSCs) are defined as Dct-positive and c-Kit signaling-independent cells in the lower permanent portion of the hair follicle [23, 24]. Previous data from our laboratory showed that several housekeeping genes, including GapDH, ActB, ActG, and Aldoa, are expressed at a lower level in MelSCs [24]. Additionally, MelSCs show repression of the CAG housekeeping gene promoter, as well as smaller cell size [25]. These findings prompted us to test if mRNA transcription is globally downregulated in MelSCs.

In this study, we show that RNApII Ser2 phosphorylation, reflecting productive mRNA transcription elongation, is absent in adult MelSCs in the lower permanent portion of the hair follicle, while mRNA transcription initiation is activated. In line with this, CDK9 protein and mRNA were downregulated in adult MelSCs. Interestingly, other adult stem cell systems, including hematopoietic, keratinocyte, spermatogonia, and muscle (satellite) stem cells, showed similar absence of RNApII Ser2 phosphorylation, suggesting a conserved mechanism of transcriptional quiescence in various stem cells.


Tissue Preparation and Antibody Staining

Mice of the indicated age were sacrificed by CO2 asphyxiation, shaved, and their back skin (for staining of melanocytes and keratinocytes), testis, or muscle tissue dissected by scissors. Tissue was fixed in 4% PFA in PBS at 4°C overnight, washed twice with PBS, and dehydrated in 20% sucrose at 4°C overnight. Tissue was mounted in frozen section medium (Richard-Allen Neg-50) and snap-frozen in liquid nitrogen. Sections were cut at 12 μm with a Leica CM3050S cryostat, air-dried, and blocked for 30 minutes with 2% Skim Milk (BD Difco) in PBS containing 0.1% Triton X-100 (PBS-T). Primary antibodies were diluted in blocking solution and added onto the slides at 4°C overnight. Antibodies used were rabbit anti-RNA polymerase II CTD repeat YSPTSPS (phospho S2) (Abcam ab5095) at 1/500 dilution, rabbit anti-RNA polymerase II CTD repeat YSPTSPS (phospho S5) (Abcam ab5131) at 1/50 dilution, goat anti-Trp2 (D-18, sc-10451, Santa Cruz) at 1/250 dilution, rabbit anti-CDK9 (H-169, sc-8338, Santa Cruz) at 1/50, rabbit anti-Ccnt1 (H-245, sc-10750, Santa Cruz) at 1/250 dilution, rat anti-CD9 (KMC8, BD Pharmingen) at 1/500, rat anti-c-Kit (Ack4, purified in our laboratory) at 1/250, rat anti-GFP (GF090R, Nacalai Tesque) at 1/500, rat anti-CD34 (RAM34, Ebiochem) at 1/250, rat anti-CD45 (30-F11, BD Pharmingen) at 1/1,000, rat anti-NCAM (H28.123, Millipore) at 1/50, rat anti-CD71 (R17217, eBioscience) at 1/1,000. Three times the slides were washed with PBS-T for 10 minutes, and appropriate secondary antibodies conjugated with Alexa 488 or Alexa 546 (Invitrogen, 1/500 dilution) as well as TO-PRO3 (Invitrogen, 1/1,000 dilution) were applied to the slides, stained for 1 hour at room temperature, three times the slides were washed with PBS-T for 10 minutes and mounted using ProLong Gold Antifade (Invitrogen).

Antibody Specificity

Blocking of anti-CTD-Ser2 phosphorylation (Ser2-P) and Ser5-P was performed by incubation of primary antibodies with 1 μg/ml YSPTSPS peptide phosphorylated at Ser2 or Ser5 (Abcam ab12793 and ab18488, respectively) at 37°C for 30 minutes with shaking. Preincubation of anti-Ser5-P antibody with a synthetic CTD-Ser2-P peptide resulted in a strong staining, which was completely blocked by preincubation with a CTD-Ser5-P peptide (Supporting Information, Fig.S1A, C). Similarly, the signal from anti-CTD-Ser2-P antibody could be blocked by preincubation with CTD-Ser2-P peptide, but not with CTD-Ser5-P peptide (Supporting Information, Fig. S1B, S1D). Moreover, serum starvation of NIH 3T3 cells greatly diminished signal from anti-Ser2-P antibody, which was regained after serum restimulation (Supporting Information, Fig. S2A–S2C). This increase in signal could be blocked by treatment with the CDK9 inhibitor DRB (100 μM), both in cell culture and western blot (Supporting Information, Fig.S2D, S3F). We conclude that these antibodies enable the specific and sensitive detection of CTD-phosphorylation.

Cell Culture

Isolation and culture of primary melanoblasts was performed as described previously [26]. Briefly, skin from E15.5 CAG-CAT-eGFP x Dct-Cre mouse embryo was dissected and trypsinized. GFP positive melanoblasts were sorted and cocultured on mitomycin C-treated XB2 feeder cells in the presence of bFGF and SCF. CDK9 inhibition and c-Kit starvation was performed by washing the cells and adding DRB (10 μM final concentration) 30 minutes prior to addition of Ack2-5 μg/ml final concentration overnight. Percentage of surviving cells was calculated as number of GFP positive cells in DRB-treated versus DMSO-treated condition.

3T3 fibroblasts were maintained in Dulbecco's modified Eagle's medium (DMEM) +10% FCS. Cells were cotransfected with Venus reporter plasmid and expression plasmids (pCMV) containing dominant negative CDK9 (D167N [27]) or wild-type CDK9 and Ccnt1 separated by a 2A peptide (pTEFb). Forty-eight hours after transfection, serum starvation was performed by washing and replacing medium with DMEM +0.1% FCS for 4 hours. mRNA was isolated using RNeasy columns (Qiagen). Alexa 647-conjugated anti-Annexin V staining was performed according to instructions, and the number of Annexin V positive cells was counted by FACS Canto.

FACS Sorting and Quantitative Reverse Transcription Polymerase Chain Reaction Analysis

Quantitative polymerase chain reaction (qPCR) of FACS-sorted MelSC and differentiated melanocytes cDNA libraries was performed after reverse transcription of mRNA isolated from GFPhigh SSChigh (differentiated melanocytes) and GFPlow SSClow (MelSC) [24] cells using oligo-dT30 primer. One-step quantitative reverse transcription (qRT)-PCR with gene-specific primers was conducted according to manufacturer's instructions (Qiagen) on GFP positive cells at p10 after or without Ack2 injection (MelSC and differentiated melanocytes respectively). Primer sequences are available upon request.

FACS sorting and staining of CD34 KSL cells were performed after red blood cell lysis (PharmLyse, BD Biosciences) and lineage (CD4, CD8, CD11b, Ter119, GR-1, B220) depletion using biotin-conjugated primary antibodies (BD Pharmingen) and streptavidin-conjugated magnetic beads (Biomag, Qiagen) according to manufacturer's instructions. FACS sorting was performed on a BD FACS Aria, using lineage-biotin primary and streptavidin-PE-Cy7 secondary antibodies, PE-conjugated anti-Sca-1, APC-conjugated anti-c-Kit, FITC-conjugated anti-CD34 (all BD Pharmingen), and PI for dead cell discrimination. CD34 c-Kit+ Sca-1+ Lin PI and CD34+ c-Kit+ Sca-1+ Lin PI cells were isolated and sorted separately onto a cooled Collagen type 1-coated two-well cell culture slide (BD BioCoat). The slides were incubated at 37°C for 1 hour in a moisturized hybridization chamber, fixed by adding 4% PFA at room temperature for 5 minutes, washed (three times) with PBS 10 minutes, blocked by 2% Skim Milk in PBS-T for 30 minutes, and incubated with primary antibodies at 1/1,000 dilution at 4°C overnight. Further washes, secondary antibodies, and mounting were performed as aforementioned. Images were captured with a LSM510 scanning module equipped Zeiss Axiovert 200M microscope and analyzed with ImageJ and Excel.

Transgenic Mouse

Full-length CDK9 mRNA was obtained from MGC (IMAGE:3601310), PCR amplified with the forward primer 5′ GGCGCGCCACCATGGATTACAAGGATGACGACGATA AGATGGCCAAGCAGTACGACTC 3′ and reverse 5′ GGCGCGCCTCAGAAGACAC GTTCAAATTCC 3′, resulting in addition of Asc1 sites at both ends, a Kozak sequence and N-terminal FLAG tag. The purified PCR product was digested and ligated into modified pROSA26-1 including a neo-stop cassette and IRES-eGFP [28]. TT2 embryonic stem cells were electroporated with linearized vector and subjected to neomycin selection. Individual resistant colonies were subjected to Southern blotting with external and internal probes. Correctly targeted colonies were transiently transfected with a pPGK-Cre plasmid, and expression of CDK9 mRNA fused to ROSA26 mRNA was confirmed by sequencing of the RT-PCR product. Chimeric mice raised by morula aggregation were crossed to C57BL/6, backcrossed to Bl/6 for three generations, before crossing to Dct-Cre mouse [23]. K14-SCF [29] and CAG-CAT-eGFP [30] mouse have been described previously. All animal experiments were performed in accordance with the guidelines of the RIKEN Center for Developmental Biology for animal and recombinant DNA experiments.


Active Downregulation of CTD-Ser2-P in Developing MelSCs

We previously reported a dramatic reduction in mRNA transcription in MelSC [24]. In this study, we investigated whether or not this reduction of gene transcription involves the alteration of RNApII CTD phosphorylation.

Staining of postnatal day (p) 28 back skin sections for CTD-Ser2-P showed a complete absence of productive mRNA transcription elongation in Trp2-positve MelSC compared with surrounding cells (Fig. 1A, arrowheads), suggesting a specific and active induction of MelSC quiescence rather than growth factor deprival, which would affect surrounding cells as well. However, MelSC were strongly positive for CTD-Ser5-P (Fig. 1B, arrowheads), indicating a paused state with a rapid reactivation of mRNA transcription possible. Differentiated melanocytes in the bulb region were positive for both CTD-Ser2-P and Ser5-P (Fig. 1C and 1D).

Figure 1.

Absence of mRNA transcription elongation in adult melanocyte stem cells. Staining of postnatal day (p) 28 mouse backskin for RNA polymerase II C-terminal domain (CTD) Ser2-P (green) and the melanocyte marker Trp2 (red) shows complete absence of mRNA transcription elongation in melanocyte stem cells (MelSCs) ([A], arrowheads) compared with surrounding cells. Asterisks denote dead cells in the hair follicle. Note that MelSCs show strong signal for mRNA transcription initiation (Ser5-P in green, arrowheads in [B]). Differentiated melanocytes in the hair bulb are positive for both mRNA transcription elongation and initiation (C,D). Migrating melanoblasts at embryonic day (E) 14.5 (E) as well as melanoblasts in developing hair follicles at E16.5 (F) are positive for CTD-Ser2-P (arrowheads). In E18.5 guard hair follicles of ICR mouse, some Trp2 positive cells at the lower permanent portion of the hair follicle start to downregulate mRNA elongation (filled arrowhead in [G]), while some are still positive (open arrowhead, [G]). At p0, all MelSCs in guard hair follicle of ICR mouse are negative for CTD-Ser2-P (arrowheads, [H]). CTD phosphorylation in green, Trp2 in red. Scale bar = 20 μm (A–F), Scale bar = 50 μm (G, H). Abbreviations: Ser2-P, serine 2 phosphorylation; Trp2, tyrosinase related protein 2.

One criteria of adult MelSC is their survival after injection of antagonistic c-Kit antibody at p0 [23], indicating that MelSC development is completed at early postnatal days. We analyzed the level of CTD-Ser2-P during embryogenesis, and found that migrating melanoblasts at embryonic day 14.5 (E14.5, Fig. 1E) and melanocytes homing to hair follicles (E16.5, Fig. 1F) stained positive for CTD-Ser2-P. However, downregulation of mRNA synthesis begins in some MelSCs in guard hair follicle of ICR mouse as early as E18.5 (Fig. 1G, filled arrowhead). At postnatal day 0 (p0) all MelSCs in guard hairs of ICR mouse are negative for CTD-Ser2-P (Fig. 1H). Of note, MelSC in Bl/6 downregulate CTD-Ser2 later (p2, see Fig. 2D), possibly due to slower development, as reflected by smaller size at birth [31].

Figure 2.

Biphasic induction of melanocyte stem cell (MelSC) quiescence. Transgenic mouse over expressing SCF from the K14 promoter in skin display an increased number of MelSCs (Trp2 in red, [A]). However, all MelSCs are negative for mRNA transcription elongation (green, arrowheads, [A]). During development, both K14-SCF and C57Bl/6 wild-type littermates show positive signal for C-terminal domain (CTD)-Ser2-P (green) at the day of birth (p0, arrowheads in Bi, Ci) and downregulation at p2 (Bii, Cii). At p4, however, MelSC in K14-SCF mouse displays a strong positive staining for mRNA transcription elongation (Ciii, arrowheads), while MelSCs in wild-type mouse remain quiescent (Biii). Subsequently, mRNA transcription is downregulated in MelSCs of wild-type and mutant mouse (p6 in B/Civ, p8 in B/Cv). Scale bar = 20 μm in (A), 5 μm in (B, C). Quantification of data from (B, C) is shown in (D). A total of 20 hair follicles of three mutant and three wild-type mouse of each age were measured for CTD-Ser2-P in MelSC. To avoid out-of-focus effects, quantification was performed by normalizing CTD-Ser2-P signal to nuclear signal and plotting the signal ratio from MelSCs as percent of control surrounding cells set to 100%. **, p < 10−4; ***, p < 10−5. n denotes the total number of MelSCs counted. Abbreviations: K14-SCF, SCF expression from Keratin 14 promoter; Ser2-P, serine 2 phosphorylation; Trp2, tyrosinase related protein 2.

Suppression of CTD-Ser2-P Is Independent of SCF Signaling

For many cell types, the first event leading to the induction of quiescence is deprivation of cell survival factors. In embryonic melanoblasts, SCF/c-Kit signaling is essential for migration and survival of embryonic melanoblasts. A previous study from our laboratory demonstrated that downregulation of SCF from neonatal keratinocytes coincides with induction of MelSC [32]. To investigate whether or not downregulation of SCF is required for the repression of CTD-Ser2-P, we took advantage of K14-SCF transgenic mouse as a model system. Overexpression of SCF under control of the keratin14 promoter (K14-SCF) results in a massive increase in the number of adult melanocytes [29]. However, all adult MelSCs in K14-SCF mouse are negative for CTD-Ser2-P phosphorylation (Fig. 2A, arrowheads), suggesting that neither induction nor reactivation of quiescent MelSCs depends on SCF/c-Kit signaling, as this is constitutively expressed in K14-SCF mouse. We analyzed the time-course of MelSC quiescence induction in K14-SCF mouse. At the day of birth, all MelSCs were positive for CTD-Ser2-P (Fig. 2Bi and 2Ci). At p2, CTD-Ser2-P is downregulated in both wild-type and K14-SCF mouse (Fig. 2Bii and 2Cii). Unexpectedly, we observed a clear increase in CTD-Ser2-P at p4 in K14-SCF mouse only, whereas MelSCs in wild-type littermates remained quiescent (Fig. 2Biii and 2Ciii). At p6 and p8, MelSCs in both wild-type and mutant mouse were quiescent again (Fig. 2Biv, 2Bv, 2Civ, 2Cv, quantification in 2D). Of note, expression of endogenous SCF in keratinocytes is downregulated at p4 in wild-type [32]. This suggests that a short-lived signal has a beneficial effect on the induction of MelSC quiescence at p2, which is overridden by sustained SCF signaling at p4 in K14-SCF mouse. However, this activated state is not sustained by an active signal inducing MelSC quiescence at p6, suggesting that induction and maintenance of MelSC quiescence are regulated by different pathways at different times during development.

Downregulation of CDK9 in MelSCs

To gain an insight into the mechanism underlying low CTD-Ser2-P in MelSC, we analyzed expression of the CTD-Ser2 kinase p-TEFb, consisting of a heterodimer of CDK9 and a Cyclin partner (T1, T2, or K). Endogenous inhibitors of CDK9 activity, or the absence of CDK9 itself, may cause dephosphorylation of CTD-Ser2. To test these possibilities, we analyzed expression of CDK9 during MelSC development. As shown in Figure 3A and 3B, we observed numerous MelSCs that were already low in Ser2-P (Fig. 1H), but which expressed a normal level of CDK9 at p0 in ICR mouse.

Figure 3.

CDK9 downregulation in adult MelSC. p-TEFb, consisting of a heterodimer of Ccnt1 and CDK9, is the major kinase responsible for C-terminal domain (CTD)-Serine 2 phosphorylation (Ser2-P). CDK9 is expressed in p0 ICR hair follicles ([A], magnification in [B]). Although Ccnt1 protein was detected in p28 adult MelSC (Ccnt1 in green, arrowheads in [C]), CDK9 protein was absent in MelSCs, but expressed in all surrounding cells (CDK9 in green, arrowheads in [D]). Quantitative polymerase chain reaction showed a threefold lower expression of CDK9 mRNA in MelSC compared with differentiated melanocytes, while the level of Ccnt1 was similar between the two populations ([E], *, p < .05, n = 6 experiments). Serum starvation induces Pol2S2 dephosphorylation in 3T3 fibroblast cells ([F], −FCS) which is regained after restimulation with FCS (−/+FCS). This rephosphorylation can be inhibited by DRB in a dose-dependent manner (10 μM and 100 μM DRB). Note that treatment with 10 μM DRB does not significantly inhibit CTD-Ser2-P in 3T3 cells. Lamin B1 serves as loading control. Preincubation of mouse primary embryonic melanoblasts with 10 μM DRB improves survival of anti-c-Kit (Ack2) treatment in vitro. Representative FACS plot and quantification of Ack2 treatment in G (n = 7 experiments, *, p < .05). Trp2 positive Melanoblasts surviving Ack2 treatment without addition of DRB show decreased staining for C-terminal domain (CTD)-Serine 2 phosphorylation (Ser2-P) (in green, Trp2 in red, arrowheads in [H]) compared with cocultured XB2 feeder cells (surrounding cells) in vitro. Data shown as mean ± SEM. Scale bar = 10 μm in (A), 20 μm in (B, C, D, H). Abbreviations: DMC, differentiated melanocytes; DMSO, dimethyl sulfoxide; DRB, 5,6-dichloro-1-β-D-ribofuranosylbenzimidazole; FCS, fetal calf serum; GFP, green fluorescent protein; Mbs, melanoblasts; MelSC, melanocyte stem cell; SSC, side scatter; Trp2, tyrosinase related protein 2.

When we observed the expression of Cyclin T1 protein (Fig. 3C) and mRNA (Fig. 3E) in adult MelSCs, the level of the CDK9 protein was reduced in adult MelSC compared with surrounding cells (Fig. 3D). In line with this, CDK9 mRNA was specifically downregulated in MelSC compared with differentiated cells (Fig. 3E). To our knowledge, this is the first report of the absence of CDK9 protein in any cell type. This data suggests strongly that global transcriptional suppression occurs before CDK9 downregulation, which is an outcome rather than cause of global transcriptional suppression.

Embryonic melanoblasts as well as transit amplifying and terminally differentiated melanocytes depend on SCF/c-Kit signaling for migration, proliferation and survival. Injection of an antagonistic c-Kit antibody (Ack2) into newborn mice results in complete whitening of the first post natal hair, while hair color is restored during later hair cycles [23]. This indicates that MelSC are independent of c-Kit signaling and can reconstitute the whole melanocyte system. To gain functional insights into CTD-Ser2 dephosphorylation in MelSC, we treated cultured embryonic GFP+ melanoblasts [26] with DRB, a well-known inhibitor of CDk9 function. We found an improved survival of embryonic melanoblasts upon Ack2 treatment in vitro (Fig. 3G), indicating that low inhibition of CDK9 has a beneficial effect on survival of cells upon deprivation of survival factors. Under this experimental setting, however, we cannot exclude that cocultured XB2 cells are affected by DRB treatment as well, and that the observed effect is indirect. On the other hand, cultured melanoblasts surviving Ack2 treatment in vitro exhibit a decreased CTD-Ser2 phosphorylation (Fig. 3H), suggesting that CTD-Ser2 dephosphorylation is necessary for c-Kit independence. Similarly, overexpression of dominant negative CDK9 in 3T3 fibroblasts protected the cells from apoptosis induced by serum starvation (Supporting Information, Fig.S3B), suggesting that transcriptional quiescence is beneficial for cell survival in stress conditions.

CDK9 Expression Driven by the Rosa26 Promoter Is Insufficient to Restore CTD-Ser2-P in MelSCs

To elucidate the function of CDK9 downregulation in MelSCs, we created Cre-inducible CDK9 overexpressing mice. The Rosa26 locus was successfully targeted with a FLAG-tagged CDK9 and IRES-eGFP construct (Fig. 4A and 4B). Expression of GFP and CDK9 mRNA in-frame with Rosa26 mRNA was confirmed in targeted ES transfected with a Cre-expression plasmid (Fig. 4C and 4D). However, the coat color of the third hair cycle of mice double-positive for Dct-Cre and R26-CDK9 was indistinguishable from Dct-Cre single-positive mice (Fig. 4E), suggesting that overexpression of CDK9 in melanocytes does not impair MelSC function. GFP expression was weak but detectable in MelSCs (Fig. 4G, arrowheads). Surprisingly, we did not detect phosphorylation of CTD-Ser2 in MelSC of double knock-in mouse (Fig. 4H), implicating that expression of CDK9 from the R26 locus is not sufficient to induce phosphorylation.

Figure 4.

Overexpression of CDK9 in vivo. CDK9 IRES-eGFP was knocked-in into the Rosa26 locus and expression in melanocytes induced by crossing with a Dct-Cre mouse. (A): Scheme of targeting strategy, with primer P1 on the Rosa26 mRNA and reverse primer P2 on CDK9. Correct targeting and single integration was confirmed by southern blot. Band marked “a” represents wild-type, band “b” targeted allele using external probe, band “c” shows single integration in ES clones 68, 86, and 89 using an internal probe ([B], asterisk marks unspecific background). GFP positive cells were observed only after Cre-mediated recombination in clone 68 (C). Reverse transcription polymerase chain reaction (RT-PCR) and sequencing (not shown) confirms expression of CDK9 in-frame with Rosa26 mRNA in GFP positive recombined cells using primer P1 and P2 ([D], all CDK9 as internal control). Hair color of the third hair cycle of double transgenic R26-CDK9 x Dct-Cre mouse was indistinguishable from Dct-Cre control mouse ([E], hair cycle induced by epilation, mouse from ES clone 68). Total number of GFP positive melanocytes and MelSC was equal between double and triple mutant (F, n = 6–8 mouse). GFP is expressed in MelSCs (yellow indicates merge of green GFP and red Trp2 signal, arrowheads, [G]). Absence of C-terminal domain (CTD)-Ser2 phosphorylation in MelSCs of double transgenic mouse (Trp2 in red, To-Pro3 in blue, arrowheads, [H]). Cre recombination efficiency (ratio of GFP+/Trp2+ cells) was calculated to be more than 60% (I). One-step quantitative RT-PCR shows that total CDK9 is repressed in MelSC of Dct-Cre x R26-CDK9 mouse ([J], **, p < .005). R26-CDK9 represents around 10% of all CDK9 and is expressed at lower levels in MelSC compared with differentiated melanocytes (*, p < .05). Data shown as mean ± SEM. Scale bar = 10 μm (G, H, I). Abbreviations: eGFP, enhanced green fluorescent protein; FLAG, FLAG tag; IRES, internal ribosome entry site; MC, melanocyte; MelSC, melanocyte stem cell; Ser2-P, Serine 2 phosphorylation; SSC, side scatter; Trp2, tyrosinase related protein 2.

Because the level of expression of GFP from Dct-Cre x R26-CDK9 IRES-eGFP melanocytes was very low, we used the Dct-Cre x CAG-eGFP background to count the total number of melanocytes and remaining MelSCs after Ack2 treatment. Total melanocyte and MelSC numbers in p10 Dct-Cre x CAG-eGFP were the same as those in triple Dct-Cre x CAG-eGFP x R26-CDK9 mouse (Fig. 4F), further suggesting that overexpression of CDK9 has no effect on melanocyte development or maintenance. Of note, this assay detects melanocytes that underwent Cre-recombination to express eGFP, suggesting these cells also express CDK9. We confirmed Cre recombination to be above 60% in Dct-Cre x R26-CDK9 mouse (Fig. 4I).

The total level of CDK9 was significantly lower in MelSCs in Dct-Cre x R26-CDK9 mouse compared with differentiated melanocytes (Fig. 4J). Furthermore, expression of R26-CDK9 was diminished in MelSC, suggesting that, as with other housekeeping gene promoters, the Rosa26 locus is repressed in MelSC.

CTD-Ser2-P in Other Stem Cell Types

During the imaging of MelSCs for CTD-Ser2 staining, we observed a distinct zone of CTD-Ser2-low cells in the bulge region of adult hair follicles. Staining with the keratinocyte stem cell (KSC) marker CD34 [33] (Fig. 5A, bracket) and K15 [34] (not shown) proved that these CTD-Ser2 low cells are KSCs. To determine whether this phenomenon in MelSCs and KSCs is active in other quiescent stem cells, we tested other adult stem cell systems, including muscle stem cells (satellite cells) positive for NCAM [35] (Fig. 5B, arrowhead), and M-Cadherin [36] (not shown), spermatogonia stem cells positive for CD9 [37] (Fig. 5C) and EpCAM [38] (not shown) and hematopoietic stem cells, defined as CD34-KSL cells [39] (Fig. 5F). We observed that some of these stem cell marker positive cells showed significant absence of CTD-Ser2-P (quantification in Fig. 5H). Some CD9 positive spermatogonia stem cells showed very low levels of CTD-Ser2-P (Fig. 5C, arrowheads), while c-Kit positive spermatogonia, actively cycling cells, were detected as CTD-Ser2-P positive cells (Fig. 5D).

Figure 5.

C-terminal domain (CTD)-Ser2-P is commonly suppressed in adult stem cells. Staining with the keratinocyte stem cell (KSC) marker CD34 showed KSC are also low for mRNA transcription elongation (bracket, [A]). Muscle satellite cells, positive for NCAM ([B], red), showed a similar decrease in CTD-Ser2-P. CD9 positive spermatogonia cells attached to the basal lamina were negative for CTD-Ser2-P ([C], filled arrowheads), while CD9 positive cells detaching from the lamina started to upregulate CTD-Ser2-P ([C], open arrowheads). c-Kit positive transit amplifying spermatogonia were positive for CTD-Ser2-P (D). CD34+ KSL cells, characterized as short-term repopulating HSCs, are homogeneously high positive for CTD-Ser2-P (E). Around three-fourth of long-term repopulating CD34-KSL cells are CTD-Ser2-P high positive ([F], open arrowheads), while one-fourth is negative for mRNA elongation ([F], filled arrowheads). Quantification of HSC is shown in (G). The ratio of CTD-Ser2-P versus nuclear signal was plotted as a histogram. CD34- KSL cells (black bars) show a shift toward lower CTD-Ser2-P values. Data shown is cumulative of three experiments. *, p < .05, total cell number ∼ 700 cells of each population, error bars mean ±SEM. (H): Quantification of CTD-Ser2-P staining for MelSCs, KSCs (CD34 and K15), satellite cells (NCAM and Mcad), spermatogonia stem cells (CD9 and EpCAM) and HSCs. CTD-Ser2-P/nuclear signal of equal numbers of surrounding cells was set to 100%, black bars represent CTD-Ser2-P/nuclear signal of stem cells positive for the respective marker. E values at the top denote p-value of student's t-test, n = number of cells of each population. Scale bars = 50 μm (A), 20 μm (B–D) and 10 μm (F, E). Abbreviations: EpCAM, epithelial cell adhesion molecule; HSC, hematopoietic stem cell; KLS, c-Kit- Sca1- lin- cells; MelSC, melanocyte stem cell; NCAM, neural cell adhesion molecule; Ser2-P, Serine 2 phosphorylation.

To further assess that low CTD-Ser2 is observed only in the SC compartment, we took advantage of the hematopoietic system, in which detailed staging is possible. Sorting and staining of CD34+ KSL cells, defined as short-term repopulating hematopoietic stem cell (HSC), showed a very homogenous distribution of CTD-Ser2-P (Fig. 5E and 5G, white bars). However, CD34- KSL cells clearly showed two distinct populations, one with CTD-Ser2-P levels as high as CD34+ KSL cells (Fig. 5F, open arrowheads) and another population with extremely faint CTD-Ser2-P staining (Fig. 5F, filled arrowheads). Accordingly, a histogram of CTD-Ser2-P normalized to nuclear staining showed a significant shift to lower CTD-Ser2-P signal in 27% of all CD34- KSL cells (Fig. 5G, black bars).

Taken together, we have demonstrated that a significant proportion of cells in at least five adult stem cell systems show significant global repression of productive mRNA synthesis (Fig. 5H), raising the possibility that this conserved mechanism of adult stem cell quiescence may play an important role in the maintenance of adult stem cells. Furthermore, analysis of CTD-Ser2-P may help to identify cells with stem cell properties in various organs and cancer tissues.


Adult stem cells have the unique ability to self-renew and differentiate, thereby maintaining tissue integrity and response to injury throughout life. These cells must be protected from genetic and cellular damage due to their life-long functionality. Maintaining the quiescent state of the most immature compartment is a common strategy found in most adult stem cell systems.

One prominent common feature of adult stem cells is their label-retaining capacity [41], suggesting that cell cycle quiescence is characteristic for adult stem cells. Indeed, activation of the cell cycle leads to adult stem cell depletion [42–45]. However, how the quiescence of the stem cell compartment is induced and maintained is largely unclear. It has been long known that hematopoietic stem cells can be isolated by their low retention of Pyronin Y, an RNA-binding dye [46], suggesting that the global suppression of transcriptional activity is a hallmark of quiescent stem cells. This observation, however, has not been investigated in terms of the global transcription activity of RNApII, which is responsible for all mRNA transcription. It is generally accepted that phosphorylation of the CTD repeat of RNApII is tightly coupled to the different stages of mRNA transcription [11]. Phosphorylation of CTD-Ser5 and CTD-Ser2 is predominantly found during the initiation and elongation phases of mRNA synthesis, respectively. It is now becoming clear that most mRNA transcription is regulated at the elongation phase. Signaling events result in the immediate transcription of response genes, such as heat shock genes, stationary phase exit genes, differentiation induced genes, and primary response genes [8, 22, 47-50]. RNApII at such genes is already present at the promoter and phosphorylated at CTD-Ser5, but remains stalled until signaling induces phosphorylation of CTD-Ser2 by p-TEFb. Notably, this promoter proximal pause is detected only on certain genes, while for example, housekeeping genes are actively transcribed.

The major aim of this study is to assess the phosphorylation status of RNApII in stem cell compartments. We found a significant reduction in housekeeping gene expression in MelSCs and showed that productive mRNA transcription elongation is globally downregulated in MelSCs. However, RNApII is present in a CTD-Ser5 phosphorylated form, suggesting that the transcription machinery is poised for immediate transcription upon appropriate signaling. Absence of CTD-Ser2-P specifically marked MelSCs, while surrounding keratinocytes and fibroblasts were positive. This indicates an active repression of mRNA transcription elongation, distinct from the quiescence induced by nutrient starvation. MelSCs do however express some mRNA [24]. This suggests the presence of a transcription factor that directs the remaining p-TEFb to specific genes to activate local mRNA transcription, as previously shown [13].

CDK9, the kinase subunit of p-TEFb, has been detected in all cell types and tissues examined to date [51, 52], including quiescent primary T lymphocytes [7, 21, 53]. In line with the essential role of CDK9, its knockdown or inhibition leads to the complete abolishment of global mRNA synthesis in vitro and in vivo, and subsequent apoptosis [15, 18, 19, 27]. Surprisingly, we found that CDK9 protein and mRNA expression are downregulated in adult MelSCs in vivo, while these cells maintain Ccnt1 expression. To date, there is no known negative regulator of the expression of CDK9 [54], which behaves as a typical housekeeping gene. Indeed, the CDK9 promoter shows characteristics of a constitutive housekeeping gene promoter [55]. Thus, low CDK9 expression may be primarily an outcome of the global suppression of housekeeping genes in MelSCs [24]. Nonetheless, it is possible that low CDK9 would strengthen this state of global suppression. Which factor(s) induce downregulation of CDK9 transcript in MelSCs remains for future study.

To gain insight into the significance of CDK9 downregulation, we inhibited CDK9 by chemical inhibitors and dominant negative CDK9. In both cases, inhibition resulted in improved survival in conditions of growth factor starvation, suggesting that transcriptional quiescence is beneficial in stressful conditions. To elucidate the in vivo function of CDK9, we attempted to overexpress CDK9 in melanocytes. However, our strategy failed to induce CTD-Ser2 phosphorylation in MelSCs, and did not have any effect on adult MelSCs, as CDK9 expression from the endogenous Rosa26 promoter was also repressed in these cells. Consistent with this is our observation that nascent MelSCs that are already low in Ser2-P are still CDK9 positive. This strongly suggests that global transcriptional suppression starts earlier than the downregulation of CDK9, probably through the regulatory circuit of p-TEFb activity [13]. Thus, the signal that initiates the global transcriptional suppression in MelSC should be delivered around p2.

In most cells, quiescence is first triggered by the deprivation of growth factors. To investigate whether or not this is also the case for MelSCs, we analyzed CTD-Ser2-P in K14-SCF transgenic mice, which maintain SCF expression in epidermis, while endogenous SCF expression is downregulated at around day 2 after birth [32]. Indeed, our previous studies on this strain showed that activated melanoblasts are maintained in postnatal epidermis. Our findings in the present study show that Ser2-P negative cells are present in K14-SCF mouse, suggesting that the mechanism inducing global suppression of transcription and Ser2-P downregulation does not require deprivation of growth signaling. Surprisingly, however, we also found a clear difference between wild-type and K14-SCF transgenic mice. We observed a second peak of CTD-Ser2-P during postnatal development in K14-SCF mouse. Hence, it is likely that deprivation of the SCF signal from epidermis may facilitate the induction of quiescent MelSCs under normal circumstances. Nonetheless, all these results suggest the presence of a dominant signal that induces global transcriptional suppression, even in the presence of SCF, suggesting that the induction and maintenance of transcriptional quiescence are of great importance in the MelSC system.

To investigate whether or not our observation in MelSCs is also found in other stem cells, we assessed CTD-Ser2 phosphorylation in a number of stem cell systems. Interestingly, we observed lower CTD-Ser2 phosphorylation in various other adult stem cell systems, including keratinocyte, muscle, spermatogonia, and hematopoietic stem cells, which suggests that global suppression of mRNA transcription elongation is a conserved feature of adult stem cells. Heterogeneity in the hematopoietic stem cell system has recently been addressed, revealing a fraction of 15%–25% of cells with in vivo stem cell function [40, 56, 57], in agreement with our observation of 27% of CTD-Ser2-P negative cells in the CD34-KSL population. However, whether these CTD-Ser2-P negative cells exhibit increased HSC activity remains to be shown.

Taken together, we have shown that global repression of mRNA transcription elongation is a specific and early marker of MelSCs. MelSCs display instead initiated but paused RNApII, ready to start productive mRNA transcription upon appropriate stimulation. Absence of CTD-Ser2 phosphorylation is a conserved feature of adult stem cells, reflecting their lower metabolic status and possibly protecting the cells from cytotoxic and genetic damage. We suggest that the screening of tissues, including cancer, using antibodies against CTD-Ser2-P may facilitate the detection of subsets of cells with stem cell-like phenotypes.


We thank Dr. Igor Samokhvalov for support with the generation of knock-in mice, Dr. Yoshiteru Sasaki for R26 IRES-eGFP targeting vector and support with western blotting, Dr. Lars Martin Jakt for helpful discussion, Douglas Sipp for critical reading of the manuscript, and the Laboratory for Animal Resources and Genetic Engineering at the RIKEN CDB for injection of ES cells and maintenance of mice. R.F. is supported by a Monbukagakusho scholarship.


The authors indicate no potential conflicts of interest.