|Home | About | Journals | Submit | Contact Us | Français|
Hematopoietic stem cells (HSCs) are dormant in the bone marrow and can be activated in response to diverse stresses to replenish all blood cell types. Here we identify the ubiquitin ligase Huwe1 as a crucial regulator of HSC functions via its post-translational control of N-myc. We found Huwe1 to be essential for HSC self-renewal, quiescence and lymphoid fate specification. Using a novel fluorescent fusion allele (MycnM), we observed that N-myc expression was restricted to the most immature, multipotent stem and progenitor populations. N-myc was upregulated in response to stress or upon loss of Huwe1, leading to increased proliferation and stem cell exhaustion. Mycn depletion reversed most of these phenotypes in vivo, suggesting that the attenuation of N-myc by Huwe1 is essential to reestablish homeostasis following stress.
HSCs have the unique capacity to self-renew and differentiate into all mature blood lineages. Although HSCs are a predominantly quiescent population, they can respond dynamically to extrinsic cues, such as injury and infection1. HSC exhaustion is the phenomenon whereby this typically quiescent population is driven to proliferate abnormally and the balance between asymmetric, self-renewing divisions is skewed toward rapid symmetric divisions, eventually leading to a premature depletion of the stem cell pool2. Chronic inflammation, inhibition of cyclin-dependent kinase inhibitors (CKI) or loss of particular angiocrine factors in the bone marrow niche can all lead to stem cell exhaustion3. Thus, to prevent stem cell exhaustion, mechanisms must exist in HSCs to attenuate proliferative signals.
The transition from the resting (G0) to activated (G1) cell state is primarily driven by the accumulation of biomass, manifested by increased transcriptional and translational activity along with greater uptake and anabolism of biosynthetic precursors4. The Myc family of proto-oncogenes (c-Myc, N-myc and L-myc) are particularly relevant in this context, as these have been shown to influence transcription of a wide range of genes that are required for cell growth and DNA replication5, 6. Myc and Mycn mRNAs are expressed at roughly equivalent levels in HSCs7. Conditional deletion of Myc has been shown to inhibit differentiation of HSCs whereas deletion of both Myc and Mycn genes leads to rapid depletion of the stem cell pool7, 8. As proteins, c-Myc and N-myc exhibit extensive post-translational modifications and are remarkably unstable9–11. Turnover of Myc proteins is determined by a cascade of phosphorylation and ubiquitylation events, which target them for ubiquitin-mediated proteolysis12, 13.
We and others have previously demonstrated the critical role of the SCFFbxw7 ubiquitin ligase as a regulator of HSC quiescence14–16 and shown that stem cell exhaustion observed in Fbxw7 conditional knockout mice is dependent on abundance of c-Myc protein14. The HECT family ubiquitin ligase Huwe1 (also referenced as Mule or ARF-BP1) has been shown to ubiquitylate many of the same substrates as Fbxw7, including Mcl1, c-Myc and N-myc17–19. Furthermore, Huwe1 has been previously implicated as a determinant of neural stem cell self-renewal and differentiation20. Therefore, we hypothesized that the two ligases might act in a similar or concerted fashion in HSCs.
Here we report that conditional knockout of Huwe1 in the hematopoietic system led to a loss of HSC self-renewal and impaired lymphoid specification at the earliest stages of differentiation. Using novel fluorescent fusion knock-in alleles, we observe at the single-cell level that loss of Huwe1 leads to stabilization of its substrate N-myc. Attenuation of N-myc by Huwe1 was essential to maintain quiescence of adult HSCs, as we demonstrate that depletion of Mycn in Huwe1-deficient HSC led to restoration of the stem cell pool. This work further illustrates the complex regulation of Myc transcription factors by the ubiquitin proteasome system in immune development.
Analysis of RNA sequencing data from sorted populations of hematopoietic cells revealed that HECT, UBA and WWE domain containing 1 (Huwe1) was among the most highly expressed genes related to ubiquitin conjugation in HSCs (Supplementary Fig. 1a)21. Huwe1 expression decreased during early stages of differentiation, but was abundantly expressed in mature lymphoid populations (B, T and NK cells) (Supplementary Fig. 1b). To study whether Huwe1 has a role in hematopoiesis, Huwe1 conditional knockout (“floxed”) mice were crossed to the pI:pC-inducible Mx1-Cre transgenic line to induce deletion of Huwe1 in HSCs (and their progeny) in adult mice. At early timepoints post-pI:pC administration (4–6 weeks), a slight, but significant, increase in phenotypic HSCs (Lineage-negative (Lin−) Kit+Sca1+CD150+CD48−) was observed in Huwe1-deficient bone marrow (Supplementary Fig. 1c). However, at later time points of analysis (4–6 months post-deletion), HSCs were severely reduced, in both frequency and absolute number, in the bone marrow of Huwe1F/YMx1-Cre+ mice (Fig. 1a–c), suggesting that an expansion and subsequent contraction of the self-renewing HSC population had occurred following loss of Huwe1.
To test the consequences of Huwe1 loss on HSC function in vivo, bone marrow from Huwe1F/YMx1-Cre+ or Huwe1+/YMx1-Cre+ donors was transplanted competitively with an equal number of wild-type bone marrow cells into lethally irradiated congenic recipients (Fig. 1d). pI:pC was administered to recipients 3 weeks post-transplant. Contribution from the Huwe1-deficient donors to peripheral blood chimerism declined significantly and was virtually undetectable 12 weeks after pI:pC treatment. These data suggested that the stem cell defect in Huwe1-deficient mice is cell-intrinsic and cannot be attributed to failure to home in bone marrow, as we observed these effects with deletion following engraftment. Similarly, the in vitro colony-forming ability of isolated Huwe1-deficient hematopoietic stem and progenitor cells (HSPCs: Lin−Kit+Sca1+) was also impaired upon serial passaging (Supplementary Fig. 1d). Analysis of the bone marrow of transplant recipients revealed a significant reduction in total donor-derived cells and a profound loss of donor-derived HSCs in the Huwe1-deficient cohort (Supplementary Fig. 1e,f).
Since the onset of hematopoietic defects in Huwe1 conditional knockout mice was more rapid upon transplantation, we further investigated how Huwe1-deficient HSCs would respond to acute stress. Treatment with the chemotherapeutic agent 5-fluorouracil (5-FU) specifically ablates rapidly dividing hematopoietic progenitor cells and subsequently induces cycling of otherwise quiescent HSC to replenish these progenitor pools22. Therefore, we tested how Huwe1-deficient mice would respond to serial 5-FU challenge. In this context, Huwe1-deficient mice exhibited markedly decreased survival when compared to wild-type mice (Fig. 1e). We hypothesized that this increased susceptibility to repetitive 5-FU challenge was due an inability of Huwe1-deficient HSCs to reenter quiescence. Accordingly, in wild-type bone marrow, ~70% of HSC were in resting phase (G0) 14d after 5-FU treatment, whereas only half (36%) of Huwe1-deficient HSCs were in G0 at the same time point (Supplementary Fig. 1g). We also found a significant decrease in quiescence in Huwe1-deficient HSCs under steady state conditions (Fig. 1f), although to a lesser extent than following 5-FU.
To avoid any potential complications due to activation of interferon response in the Mx1-Cre model23, we also generated Huwe1 conditional knockouts using the Vav1-Cre transgene. Vav1-Cre is pan-hematopoietic and labels the vast majority of fetal HSCs in lineage-tracing experiments by E13.524. Fetal livers from developing Huwe1-deficient or wild-type embryos (E18.5) exhibited no difference in HSC number (Supplementary Fig. 2a,b), but gave rise to significantly fewer colonies in vitro (Supplementary Fig. 2c). Conversely, adult Huwe1F/YVav1-Cre+ mice, had significantly fewer bone marrow HSCs compared to their wild-type littermates and the latency of this phenotype was notably shorter than in the inducible Mx1-Cre model (Fig. 2a). Taken together, these findings suggested that Huwe1 is essential for quiescence and self-renewal of adult HSC both in steady-state and under conditions of stress.
Interestingly, a profound reduction in early lymphoid progenitors and developing B cells was observed in the bone marrow of Huwe1-deficient mice (Fig. 2a–c). Erythroid and myeloid populations in these mice were not reduced (Fig. 2d), suggesting this reduction was not merely a consequence of declining HSC function. Early T cell commitment was also diminished in these mice, resulting in thymii that were significantly smaller in size and decreased in cellularity (Fig. 2e and Supplementary Fig. 3a). Thus, we concluded that Huwe1 also has a crucial role in early fate decisions in HSCs, demonstrated by the loss of the earliest lymphoid-biased or restricted progenitors (Flt3+ MPPs and CLPs) in the bone marrow25. This effect was cell intrinsic, as sorted Huwe1-deficient HSPCs were also defective or delayed in their ability to give rise to B and T cells in OP9 stromal co-cultures (Supplementary Fig. 3b,c).
In contrast, approximately half of Huwe1F/YVav1-Cre+ mice that were aged beyond four months exhibited signs of myelo-proliferation, accompanied by reduced hemoglobin and red blood cell counts (Supplementary Fig. 4a,b) and slightly elevated white blood cell (WBC) counts (Supplementary Fig. 4c). The majority of nucleated cells in the peripheral blood were granulocytes or monocytes (Supplementary Fig. 4d,e). The spleens of these aged mice also had a profoundly disrupted architecture characterized by loss of distinguishable boundary between white and red pulp and a 2-fold or greater expansion in CD11b+ cells (Supplementary Fig. 4f). These observations implied that Huwe1 loss alters differentiation balance leading to the significant loss of lymphoid progenitors and, in a subset of mice, indicators of myelo-proliferative neoplasms.
We hypothesized that the hematopoietic defects exhibited in Huwe1-deficient mice are a consequence of changes in the activity of Myc proteins. Myc and Mycn are the two Myc family genes that are predominantly expressed in hematopoietic progenitors7. Since the c-Myc–GFP fusion allele (MycG) has proven useful for determining post-transcriptional regulation of c-Myc in heterogeneous populations of cells14, 26, 27, we sought to generate a similar fusion allele for N-myc. A targeting construct was generated where mCherry cDNA was cloned in frame with the first translated ATG within the Mycn gene. (Supplementary Fig. 5a). N-myc immunoblot analysis of normal and targeted ES cells confirmed that an immunoreactive protein product of approximately 95 kDa was expressed exclusively in the properly targeted ESCs (Fig. 3a). Consequently, a significant shift in mCherry fluorescence was observed in ESCs that expressed the N-myc fusion product (Fig. 3b). Two independent ESC clones were used to establish Mycn–mCherry (MycnM) mouse lines where the fusion protein would be expressed in all tissues. Intercrossing of MycnM/+ mice resulted in homozygous MycnM/M pups at the expected Mendelian ratio and these mice developed to adulthood without any discernable defects.
Although robust expression of N-myc–mCherry was observed in embryonic stem cells, mCherry fluorescence was virtually undetectable in the majority of hematopoietic cells from MycnM/M adult mice. In contrast to c-Myc–GFP, which is expressed at the highest levels in DN3–4 thymocytes and Lin−Kit+Sca1− bone marrow progenitors, N-myc–mCherry expression was restricted to the most primitive stem and progenitor compartments, such as HSCs and early thymic precursors (CD4−CD8−CD25−CD44hiKit+ ETPs), and tapered off rapidly in more restricted progenitors (Fig. 3c). To assess whether the mCherry fusion could be utilized to faithfully identify cells with differential N-myc protein abundance, HSCs from the bone marrow of MycnM/M mice were sorted on the basis of mCherry fluorescence (N-mychi and N-myclo) and subjected to microarray gene expression profiling. Accordingly, genes upregulated in tumors where N-myc is amplified were significantly enriched in N-mychi HSCs (Supplementary Fig. 5b,c). In contrast to what we had previously observed in c-Mychi HSPCs14, gene signatures associated with adult stem cell populations were enriched in the N-mychi HSCs (Supplementary Fig. 5d).
Next, we investigated whether HSCs that expressed differential levels of c-Myc or N-myc protein were functionally distinct. To this end, HSC from the bone marrow of MycnM/MMycG/G mice were sorted on the intensity of GFP and mCherry fluorescence into four populations (Fig. 3d) and transplanted into lethally irradiated recipients. Recipients transplanted with c-Mychi HSC (III and IV) had significantly reduced frequency of donor-derived cells in their peripheral blood over time compared to recipients transplanted with c-Myclo HSC (I and II) (Fig. 3e). This finding agrees with our previous observation that HSPCs expressing higher levels of c-Myc have reduced self-renewal capability14. N-myc abundance, on the other hand, was not a clear determinant of stemness, as recipients transplanted with either c-Myclo N-mychi (II) or c-Myclo N-myclo (I) HSC had equivalent levels of donor chimerism over time. These data suggest that self-renewing HSC preferentially express N-myc and demonstrate that high c-Myc expression can serve as an early indicator of HSC differentiation.
Under stress conditions where HSCs are stimulated to proliferate, c-Myc been found to increase accordingly28. To test whether N-myc abundance can respond dynamically to hematopoietic stress, we injected MycnM/MMycG/G mice with 5-FU or pI:pC and measured mCherry or GFP fluorescence in HSC 4d later. Since HSCs downregulate c-kit in response to 5-FU treatment, we utilized the SLAM family markers CD48 and CD150 to identify HSCs in 5-FU–treated bone marrow (Fig. 4a). Similarly to c-Myc, we observed a significant accumulation of N-myc in HSCs from mice treated with either 5-FU or pI:pC (Fig. 4b,c), indicating that N-myc is also upregulated in HSCs as they proliferate in response to stress. This observation led us to hypothesize that depletion of N-myc might impair HSC proliferation in response to 5-FU. Surprisingly, following 5-FU treatment, MycnF/FMx1-Cre+ mice had a 2-fold increase in HSC number compared to their wild-type littermates (Fig. 4d). By comparison, loss of Huwe1 had the opposite effect, severely compromising recovery of the HSC pool after treatment.
To reliably address the impact of Huwe1 depletion on N-myc stability, we first relied on a well-characterized HSPC-like cell line, HPC-729. Similarly to HSCs, HPC-7 cells can undergo multilineage differentiation, depend on stem cell factor (SCF) for their expansion and express both N-myc and c-Myc (Fig. 5a). HPC-7 cells were transduced with retroviral shRNA constructs targeting either Huwe1 or Renilla luciferase as a negative control. Upon Huwe1 knockdown, a significant accumulation of N-myc protein, but not c-Myc, was observed (Fig. 5a). We tested several other known Huwe1 substrates, including p53, Miz1 and Mcl1 17,30,31, but no increase in these proteins was observed. Since Huwe1 has also been implicated in MAP kinase and Wnt signaling32,33, we also measured phosphorylated Erk1/2 and beta-catenin, yet neither were affected by Huwe1 depletion. To address whether the accumulation of N-myc was proteasome-dependent, the Huwe1 knockdown study was repeated in the presence of a proteasome inhibitor. Addition of MG-132 significantly stabilized N-myc in cells expressing the control shRNA, while this effect was greatly diminished in cells where Huwe1 was depleted (Fig. 5b). However, p53 was still strongly stabilized in both conditions34. Proteins that are polyubiquitylated via lysine 48 (K48) linkage are spared from degradation with MG-132 treatment, implying that there is less ubiquitylated N-myc following Huwe1 knockdown.
Next, to investigate whether loss of Huwe1 also results in N-myc accumulation in primary HSCs, we crossed the Huwe1 conditional knockout mice to the N-myc–mCherry c-Myc–GFP fusion line to generate Huwe1F/YMx1-Cre+MycnM/MMycG/G mice. Three weeks after pI:pC treatment, Huwe1-deficient HSCs had approximately 3-fold more N-myc protein than wild-type HSCs as determined by mCherrry fluorescence (Fig. 5c,d). We also observed a significant increase in N-myc in Huwe1-deficient common myeloid progenitors (CMPs), but not in more lineage-restricted progenitors such as granulocyte-monocyte progenitors (GMPs) (Fig. 5c). In contrast, c-Myc abundance was unchanged in Huwe1-deficient HSCs or myeloid progenitors. Taken together with our observations in HPC-7 cells, we can conclude with confidence that Huwe1 targets N-myc for proteasomal degradation in HSCs.
Since Myc family transcription factors have been shown to bind and influence transcription of a wide distribution of targets, we hypothesized that the stabilization of N-myc occurring in Huwe1-deficient HSCs would have profound consequences on gene expression. Thus, we performed high throughput RNA-sequencing on sorted HSCs from wild-type, Huwe1F/YMx1-Cre+ and Huwe1F/YVav1-Cre+ mice and analyzed transcripts that were differentially expressed (Fig. 6a). 53 genes were significantly upregulated and 224 genes were significantly downregulated in Huwe1-deficient HSC (Supplementary Fig. 6a). Upregulated transcripts included targets of the transcription factor E2F and genes involved in cell cycle progression (Fig. 6b), in agreement with our observation that Huwe1-deficient HSCs are more proliferative. Levels of Stat1 mRNA and interferon-responsive genes were also elevated in Huwe1-deficient HSCs (Fig. 6c). Conversely, one of the gene sets most significantly downregulated in Huwe1-deficient HSCs pertained to genes that are highly enriched in long-term self-renewing HSCs (Fig. 6b)35. These downregulated genes included many critical mediators of HSC quiescence, such as Mpl, Tek (Tie-2), Cdkn1c (p57), Runx1t1 (Eto) and Ndn (necdin)36–38 (Fig. 6c). Many key effectors of early lymphoid specification, such as Notch1, Gfi1 and Satb139–41 were also significantly downregulated in the Huwe1-deficient HSC.
To address whether these changes in gene expression were due to increased N-myc, we ectopically expressed N-myc in wild-type HSCs and measured several of the differentially expressed transcripts by qRT-PCR. Accordingly, Mpl, Ndn, Satb1 and Tek expression were decreased and Stat1 increased in HSCs upon enforced N-myc expression (Fig. 6d). To determine which of these genes were direct N-myc targets, we performed chromatin immunoprecipitation followed by deep sequencing (ChIP-seq) to identify regions bound by N-myc in mobilized human CD34+ stem and progenitor cells. Across replicates, we identified 4,466 high-confidence genomic peaks where N-myc was significantly enriched. The largest proportion of peaks were found in gene promoter regions (Supplementary Fig. 6b) and correlated with enrichment of active histone marks, such as H3K27 acetylation and H3K4 trimethylation (Supplementary Fig. 6c), a pattern that has previously been observed for loci bound by c-Myc. Importantly, many of the genes differentially expressed in the Huwe1-deficient HSCs were bound by N-myc, such as STAT1, CDK6 and SATB1 (Fig. 6e). Thus, we conclude that the transcriptional changes observed upon loss of Huwe1 are due, at least in part, to stabilization of N-myc.
To determine whether the phenotypes of Huwe1-deficient HSCs were dependent on N-myc accumulation, Huwe1F/YMx1-Cre mice were crossed to a Mycn conditional knockout line (MycnF/F) to generate Huwe1/Mycn double knockout animals (dKO). Unlike their Huwe1 cKO littermates, aged dKO mice retained a normal population of HSCs in their bone marrow (Fig. 7a,b). Depletion of Mycn in Huwe1-deficient mice also restored HSC quiescence (Fig. 7c) and serial colony forming ability (Supplementary Fig. 7a). Accordingly, Cdkn1c, Cdk6 and Stat1 gene expression were restored to normal levels in dKO HSCs (Fig. 7d). Deletion of a single allele of Mycn led to a slight, but statistically insignificant increase in phenotypic HSCs in the bone marrow of Huwe1–deficient mice (Supplementary Fig. 7b). However, early lymphoid progenitor populations were only minimally rescued in dKO mice (Fig. 7e), suggesting that the loss of lymphoid priming in Huwe1-deficient HSCs is largely N-myc independent. Upon transplantation, dKO bone marrow provided long-term myeloid reconstitution in recipient mice to a level comparable to wild-type donors (Fig. 7f), yet lymphoid reconstitution declined over time from both the dKO and Huwe1 KO donors (data not shown). Unexpectedly, aged mice deficient in Mycn alone had nearly a 2-fold increase in HSC compared to their wild-type littermates (Fig. 7b). We did not attribute this effect to compensation by c-Myc as neither Myc mRNA nor protein were increased in Mycn-deficient HSCs (Supplementary Fig. 7c,d). To assess whether other known substrates might also account for hematopoietic defects observed in Huwe1-deficient mice, Huwe1F/Y HSPCs were simultaneously infected with Cre and shRNA retroviruses and the ability of the doubly-transduced cells to form colonies was measured in vitro. shRNA hairpins targeting a panel of previously identified Huwe1 substrates including topoisomerase II binding protein 1 (Topbp1), Rev-erbα (Nr1d1), p53 (Trp53) and Mycn 42, 43, were tested in this assay, as well as a control hairpin targeting Renilla luciferase. Out of all the constructs tested, only the shRNA targeting Mycn could rescue serial colony forming ability in the Huwe1-depleted HSPCs (Supplementary Fig. 7e). Altogether, these data propose that Huwe1 is regulates HSC self-renewal by maintaining homeostatic amounts of N-myc as HSC exit and reenter quiescence.
Our findings have uncovered a novel role for the Huwe1 E3 ligase in the maintenance of HSC homeostasis, adding another example of how ubiquitylation and relative abundance of Myc proteins can have a critical role in developmental processes. The observation that Huwe1 can regulate N-myc stability, but not c-Myc, is intriguing given the unspecified function of N-myc in HSCs. Unlike what has been observed in Myc conditional knockouts, loss of Mycn does not appear to impair adult HSC differentiation7, but instead leads to an accumulation of HSC in aged mice. However, it remains an open question whether N-myc plays a unique role in fetal or stress hematopoiesis, where we and others have observed a significant enrichment in N-myc expression44.
An additional provocative discovery described here is the contrast of N-myc and c-Myc protein abundance in HSCs and multipotent progenitors. While others have reported Myc and Mycn transcripts are co-expressed in the most immature populations of HSCs, this was quantified in a heterogenous population of cells7. Utilizing our fluorescent fusions of c-Myc and N-myc, we are now for the first time able to quantify abundance of both proteins at the single-cell level. Instead of uniform co-expression of both proteins in HSC, we observe there are many cells where expression of N-myc and c-Myc was mutually exclusive. Whether or not these cells are functionally unique remains an intriguing possibility. Clearly, from our gene expression analysis of N-mychi and N-myclo HSCs there are some distinct molecular differences between these populations, but we found that these populations perform similarly in transplantation assays. Still, the evident switch from N-myc expression in self-renewing, quiescent stem cells to higher c-Myc expression in transit amplifying progenitors has exciting implications for unique functions of Myc family members at the molecular level. Recently, overexpression of Mycn was shown to enhance reprogramming of mature blood cells to induced hematopoietic stem cells (iHSCs)45, suggesting that N-myc, in particular, helps to activate a stem-like identity in hematopoietic cells.
Although there are many similarities between the hematopoietic phenotypes of Huwe1 and Fbxw7 conditionally deficient mice, there are some notable differences that might shed light on the role of these ubiquitin ligases in HSCs. Based on our observations in Huwe1 conditionally deficient mice, we propose that this gradual decline in the HSC pool indicates a failure to reenter quiescence, as opposed to compulsory cell cycle entry and differentiation, where we would expect a more pronounced phenotype. Hence, in stress scenarios, such as with transplantation or myeloablation, HSC exhausion in Huwe1-deficient mice was much more apparent. In steady state, the onset of HSC depletion was considerably slower in Huwe1-deficient mice than in Fbxw7-deficient mice, and depletion of Huwe1 in adult mice did not lead to fatal anemia. There are several potential mechanistic explanations why this could be the case. For one, we have observed that both c-Myc and N-myc accumulate in Fbxw7-deficient HSC (data not shown), while only N-myc appears to be stabilized upon loss of Huwe1. Therefore, the net sum of Myc proteins would be higher in Fbxw7-deficient HSCs and this could result in further amplification of c-Myc and N-myc target gene expression46. Alternatively, it is possible that N-myc could be a weaker transcriptional activator than c-Myc. This was hypothesized decades ago47, but was questioned by the fact that mice harboring homozygous knock-in of Mycn into the Myc locus develop normally48. However, to our knowledge, the effect of equivalently overexpressing either c-Myc or N-myc in HSCs has not been directly compared. Previous studies have also shown that K63-ubiquitylation of c-Myc by Huwe1 can affect its affinity for transcriptional co-activators or Miz-1, a component of a ternary complex with Max found at gene loci that are repressed by c-Myc18. If it were also to regulate N-myc in a similar manner, loss of Huwe1 would not only lead to stabilization of N-myc, it would also alter its activity. In support of this notion, we observed more genes are down-regulated in HSCs upon Huwe1 loss than up-regulated, contrary to what we observe in Fbxw7-deficient HSCs.
Finally, if Huwe1 and Fbxw7 are both ubiquitylating the same substrate does this mean that there is a convergence of these two E3 complexes in the same pathway? It is possible that N-myc could be ubiquitylated sequentially or in a concerted fashion by the two E3 complexes. Successive elongation of polyubiquitin chains on protein substrates has been described for RING finger and HECT domain E3 ligases in yeast49. It is also possible that the two ligases would act downstream of different signals. Substrate recognition by Fbxw7 depends heavily on the activity of priming kinases (e.g. GSK3β), whereas the events that lead to substrate ubiquitylation by Huwe1 are still poorly defined. it would be intriguing to test whether Huwe1 could still ubiquitylate N-myc with missense substitutions in its phosphodegron or upon inhibition of GSK3β, both of which would block its binding to Fbxw7. In addition to Huwe1 and Fbxw7, there are several other E3 ligases that have been shown to regulate ubiquitylation of Myc proteins50 and it is only through these sort of genetic and biochemical studies that we can begin to unravel the complex interplay of these proteins in HSCs.
Huwe1 and Mycn conditional knockout (floxed) mice, as well as the MycGFP knock-in allele have been described previously19, 26, 51. To delete gene loci specifically in the hematopoietic system, conditional knockout lines were crossed to Mx1-Cre and Vav1-Cre transgenic lines24, 52. Mx1-Cre was induced by three consecutive intraperitoneal injections of poly(I)·poly(C) (GE Healthcare, 10 mg/kg) administered every other day. 5-fluorouracil powder (Sigma-Aldrich) was dissolved in PBS at a concentration of 10 mg/ml and administered by intraperitoneal injection at a dose of 150 mg/kg. For transplant studies, congenic CD45.1+ recipients (B6.SJL-Ptprca) were exposed to 12 Gy of gamma irradiation (administered in two doses, 4 h apart) and cells were transferred intravenously by retroorbital injection. For transplant of sorted populations, 2 × 105 unfractionated CD45.1+ bone marrow cells were added for hemogenic support. Peripheral blood was sampled periodically from experimental and control mice and blood counts were determined by Hemavet analysis (Drew Scientific).
To generate the Mycn mCherry fusion (MycnM) allele, an 8.88 kb genomic DNA fragment was subcloned from a positively identified 129SvEv BAC clone (RP22: 127D5). An mCherry coding sequence was PCR amplified and subcloned into a recombineered SalI site within targeting vector just after the ATG start codon within Mycn exon 2. A pGK-gb2-loxP-neor cassette was inserted 550 bp downstream of exon 2. The long homology arm extended 5.6 kb 5′ of the mCherry sequence and the short homology arm extended 2 kb downstream of the neor cassette. The resulting targeting vector was linearized with NotI and electroporated into W4 129S6/SVEvTac ES cells. Neomycin-resistant clones were screened by short arm PCR and injected into C57BL/6 EIIA-Cre blastocysts to remove the neor cassette from the germline53. Genotyping of pups following backcross to wild-type C57BL/6 mice was performed by PCR using the following primers: Mycn_F: 5′- TTGGCTCATTCTCTCTTGGTTTGC-3′, Mycn_R: 5′- AGCATCTCCGTAGCCCAATTCG-3′, mCherry_R: 5′-GCCGTCCTCGAAGTTCATCA-3′. All animal experiments were done in accordance to the guidelines of the NYU School of Medicine or Institutional Animal Care and Use Committee.
Single-cell suspensions were derived from mechanical disruption of mouse bone marrow, spleen, thymus and fetal liver in PBS supplemented with 2% fetal calf serum (FCS, Sigma). For FACS analysis of spleen, fetal liver and peripheral blood, red blood cells were lysed with ACK buffer. Nonspecific antibody binding was blocked by incubation with 20 µg/ml Rat IgG (Sigma) for 15 min. Cells were incubated with primary antibodies for 45 min and streptavidin conjugates, when applicable, for 15 min on ice. The antibodies used in this study are listed in Supplementary Table 1. Lineage-negative cells were defined by lack of expression of Gr-1, TER-119, CD4, CD8, B220 and CD11b (except for fetal liver samples). For intracellular Ki67 and DAPI staining in HSC, following staining of surface antigens, bone marrow cells were fixed with 4% paraformaldehyde and permeabilized with 0.1% saponin prior to addition of anti-Ki67 (BD Biosciences) and 2 µg/ml DAPI. For sorting of rare HSC populations, bone marrow was enriched for stem and progenitor cells via magnetic selection with anti-CD117 microbeads (Miltenyi Biotec) prior to antibody staining and sorting. Stained cells were quantified using a BD Fortessa analyzer or isolated with a BD ARIA II. FlowJo software (Treestar) was used to generate flow cytometry plots, histograms and calculate mean fluorescence intensities.
Whole cell extracts were prepared by resuspending cells in RIPA buffer (50 mM Tris-HCl pH 8, 150 mM NaCl, 1% NP-40 substitute, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with complete protease inhibitor cocktail (Roche). Cleared lysates were boiled in Laemmli sample buffer with 1% beta-mercaptoethanol and proteins were separated by SDS-PAGE using precast 4–12% Bis-Tris gels (Life Technologies). Proteins were transferred to PVDF membrane (Millipore) at 30 V, 4 °C for 16 h. Blots were blocked for 30 min with 5% milk in TBS-T, incubated with primary antibodies for >3h and horseradish peroxidase conjugated anti-mouse IgG or anti-rabbit IgG (GE Healthcare) for 1 h, washing 4× for 10 min with excess TBS-T in between and prior to developing. Blots were developed with SuperSignal West Pico ECL substrate (Thermo Scientific) and exposed to film. The following antibodies were used in this study: Huwe1 (Bethyl, A300–486A, 1:1000), N-myc (Santa Cruz, B8.4.B, 1:500), c-Myc (Cell Signaling, 9402, 1:500), Mcl-1 (Santa Cruz, S-19, 1:200), Miz-1 (Santa Cruz, H-190, 1:200), beta catenin (Cell Signaling, 8480, 1:1000), Erk1/2 (Cell Signaling, 4695, 1:1000), phospo-Erk1/2 (Cell Signaling, D13.14.4E, 1:1000), p53 (Leica, CM5, 1:1000), vinculin (Sigma, hVIN-1, 1:5000) and actin (EMD Millipore, C4, 1:5000).
Murine ESCs were grown as previously described14. HPC-7 cells were grown in IMDM (Gibco) supplemented with 5% FCS, 100 ng/ml SCF (Peprotech), 100 U/ml penicillin-streptomycin (Cellgro), and 150 µM 1-thioglycerol (Sigma). Primary sorted HSCs were cultured in Opti-MEM with 10% FCS, 55 µM beta-mercaptoethanol (Gibco), 100 U/ml penicillin-streptomycin, 100 ng/ml SCF and 20 ng/ml Tpo (Peprotech) in 96-well round bottom plates at a density of 0.5–1 × 104/well. Differentiation of HSPCs to B and T lymphocytes on OP9 stroma was performed as described54. Retrovirus was produced by transfecting Plat-E cells with MSCV vectors and collecting supernatant 48 h and 72 h later. Viral supernatant was passed through a 0.45 µm syringe filter (Corning) and concentrated with Amicon 100 kDa centrifugal filters (Millipore). HPC-7 cells and primary HSC were infected by addition of concentrated virus in the presence of 4 µg/ml polybrene and centrifuged at 800 x g for 90 min at 30 °C. Antibiotic-resistance selection of cells was achieved by addition of 2 µg/ml puromycin or 500 µg/ml G418. MG-132 (Z-Leu-Leu-Leu-H aldehyde, Peptides International) was dissolved in DMSO and added to cultueres at a final concentration of 10 µM. For colony forming assays, 200 sorted HSPCs were plated in triplicate in 3-cm dishes with 1 ml M3434 media (StemCell Technologies). After 7 d, colonies were scored, harvested and serially replated at 5 × 103 cells/dish.
Cell populations were sorted directly into Trizol reagent (Life Technologies). Nucleic acids were extracted according to manufacturer instructions and precipitated in isopropanol with glycogen or linear acrylamide (Ambion) used as a carrier. After 30 min of DNase I treatment (Life Technologies), RNA was purified with RNA clean and concentrator columns (Zymo). RNA integrity was verified on a Agilent 2100 Bioanalyzer Pico chip. For small-scale RNA sequencing, full-length cDNA libraries were prepared using the SMARTer Ultra Low Input RNA kit (Clontech). Amplified cDNA libraries were sheared to 200–500 bp fragments using a Covaris LE220. End repair, A-tailing and Illumina adaptor ligation reactions were carried out using a NEBNext Library Prep Kit (New England BioLabs). Libraries were sequenced on an Illumina HiSeq2500 v4 using 50-bp paired-end reads. Fastq files were aligned to mm9 using TopHat allowing 2 mismatches. Aligned features were sorted with samtools, counted with htseq-count and differential expression was determined using the ‘edgeR’ package in Bioconductor, as described55. Microarray gene expression profiling of sorted N-mychi and N-myclo HSC and GSEA was carried out as previously described27. Heat map representations of row mean centered expression values were generated using the ‘pheatmap’ package in Bioconductor.
For the analysis of N-Myc binding, 3 × 107 human CD34+ cells were fixed in 1% formaldehyde for 10 min at 37°C, then lysed by the addition of lysis buffer 1 (5 mM PIPES pH 8, 85 mM KCl, 0,5% NP40) and incubated for 20 min at 4°C. The cells were then pelleted and resuspended in lysis buffer 2 (10 mM Tris/HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 1% NP-40, 1% deoxycholic acid sodium salt, 0.1% SDS) and incubated for 10 min at 4 °C. Nuclei were sonicated using a Bioruptor (Diagenode). 3 µg of N-myc antibody (B8.4.B, Santa Cruz) was coupled to 30 µl of Dynabeads (A/G-mixture) overnight and the complex was added to chromatin followed by a 6 h incubation at 4 °C with rotation. Complexes bound to the beads were washed three times with wash buffer A (20 mM Tris-HCl pH 8.1, 150 mM NaCl, 2 mM EDTA, 0.1% SDS, 1% Triton X-100), three times with wash buffer B (20 mM Tris-HCl pH 8.1, 500 mM NaCl, 2 mM EDTA, 0.1% SDS, 1% Triton X-100) and three times with wash buffer C (10 mM Tris-HCl pH 8.1, 250 mM LiCl, 1 mM EDTA, 1% NP-40 and 1% sodium deoxycholate). Beads were eluted twice with elutuion buffer (1% SDS, 0.1 M NaHCO3) incubating for 15 min each time followed by treatment with RNAseA (20 µg/ml) and proteinase K (200 µg/ml). DNA was purified by phenol-chloroform extraction and resuspended in TE buffer. ChIP-seq libraries were generated: end repair, A-tailing, adaptor ligation (Illumina TruSeq system) and PCR amplification. AMPure XP beads (Beckman Coulter, A63880) were used for DNA cleaning in each step of the process. Raw images generated by Illumina HiSeq2000 using the default set of parameters were processed by CASAVA to remove the first and last bases and then they were used to generate sequence reads in fastq format.
Reads were aligned to hg19 human genome using Bowtie256 with the standard parameters (except for –m 1 in order to report only unique alignments). MACS version 2.0.1057 was used to perform peak calling for both broad and narrow peaks. The files with the aligned reads were converted to wig format using GenomicTools58 and then to bigwig format using the corresponding UCSC tool (wigToBigWig, UCSC Browser binary utilities) in order to visualize the tracks. Snapshots were prepared using the Integrative Genomics Viewer (IGV). GEO Datasets GSM537652 and GSM772885 were used to determine histone modifications occupancy in human CD34+ cells. DeepTools59 was used to generate heatmap visualizations.
RNA was extracted from sorted populations as described for RNA sequencing. cDNA was prepared using the High-Capacity RNA-to-DNA Kit (Applied Biosystems). Real time PCR reactions were carried out using SYBR Green Master Mix (Roche) and run on a Lightcycler 480 II (Roche). Relative mRNA abundance between samples was calculated with the ΔΔCT method using Gapdh as an internal control. The specific primer sequences used for qRT-PCR in this study are listed in Supplementary Table 2.
Tissues were dissected from mice and fixed overnight in 10% buffered formalin (Fisher Scientific). Spleens were dehydrated in ethanol and embedded in paraffin for sectioning. Bones were decalcified in 0.5 M EDTA for 48 h prior to dehydration and embedding. 5µm paraffin sections were stained with hematoxylin and eosin for brightfield microscopy. Peripheral blood smears were briefly fixed in methanol, stained with Wright-Giemsa solution (Protocol, Fisher Scientific) and mounted in Cytoseal 60 (Thermo). Images were taken using a Zeiss Axio Observer microscope.
Statistical analyses were performed using GraphPad Prism or R. We calculated whether data points were normally distributed using the Kolmogorov-Smirnov test. When two groups were compared, a two-tailed t-test with Welch’s correction or Mann-Whitney test was used. When several groups were compared, a one-way ANOVA test followed by Dunnet’s test for multiple comparisons was used. P ≤ 0.05 was considered significant.
We would like to thank the members of the Aifantis laboratory for helpful discussions. Also, we thank A. Heguy and members of the NYU Genome Technology Center, for expert assistance RNA sequencing, the NYU Flow Cytometry facility for expert cell sorting and the NYU Histology Core. We would also like to thank G. Inghirami from Weill Cornell Medicine for assistance with histopathological evaluations. We thank S. Heimfeld at Fred Hutchinson Cancer Research Center for providing human CD34+ cells (Core Center of Excellence NIDDK Grant # DK56465). The Aifantis Lab is supported by National Institutes of Health (1R01CA169784, 1R01CA133379, 1R01CA105129, 1R01CA149655, 5R01CA173636), the William Lawrence and Blanche Hughes Foundation, The Leukemia & Lymphoma Society (TRP#6340-11, LLS#6373-13), The Chemotherapy Foundation, The V Foundation for Cancer Research, the Alex’s Lemonade Stand Foundation for Childhood Cancer and the St. Baldrick’s Cancer Research Foundation. B.K. is the Berger Foundation Fellow of the Damon Runyon Cancer Research Foundation (DRG-2234-15). This work was also supported by a grant from the Deutsche Forschungsgemeinschaft to E.W. (Emmy Noether Research Group, WO 2108/1-1) and by an American-Italian Cancer Foundation Post-Doctoral Research Fellowship to F. Boccalatte.
RNA-sequencing, ChIP-sequencing and microarray data were deposited into the Gene Expression Omnibus (GEO) with the following accession codes: GSE85488
AUTHOR CONTRIBUTIONSB.K. and I.A. designed the study and prepared the manuscript. B.K. performed most of the experiments described in the manuscript. F.B. completed experiments and focused on N-myc genomic and transcriptomic studies. K.M-C. initiated the Huwe1 cKO in vivo analysis. E.W. performed the N-myc ChIP-Seq. F.B. and B.A-O. analyzed the MYCN ChIP-Seq data. J.W. and C.K. were responsible for animal husbandry. C.L. provided bioinformatics analysis and guidance. X.Y. designed the Mycn mCherry targeting vector. A.L. provided Huwe1–floxed mice and helped with data analysis.
COMPETING FINANCIAL INTERESTS
The authors declare no competing financial interests.