Members of the NODE (Nanog and Oct4-associated deacetylase) complex and SOX-2 promote the initiation of a natural cellular reprogramming event in vivo

Abstract

Differentiated cells can be forced to change identity, either to directly adopt another differentiated identity or to revert to a pluripotent state. Direct reprogramming events can also occur naturally. We recently characterized such an event in Caenorhabditis elegans, in which a rectal cell switches to a neuronal cell. Here we have used this single-cell paradigm to investigate the molecular requirements of direct cell-type conversion, with a focus on the early steps. Our genetic analyses revealed the requirement of sem-4/Sall, egl-27/Mta, and ceh-6/Oct, members of the NODE complex recently identified in embryonic stem (ES) cells, and of the OCT4 partner sox-2, for the initiation of this natural direct reprogramming event. These four factors have been shown to individually impact on ES cell pluripotency; however, whether they act together to control cellular potential during development remained an open question. We further found that, in addition to acting at the same time, these factors physically associate, suggesting that they could act together as a NODE-like complex during this in vivo process. Finally, we have elucidated the functional domains in EGL-27/MTA that mediate its reprogramming activity in this system and have found that modulation of the posterior HOX protein EGL-5 is a downstream event to allow the initiation of Y identity change. Our data reveal unique in vivo functions in a natural direct reprogramming event for these genes that impact on ES cells pluripotency and suggest that conserved nuclear events could be shared between different cell plasticity phenomena across phyla.

How differentiated cells can switch their identity is a fascinating question that has attracted much attention in the last decade. A number of studies have shown how strikingly easily a differentiated cell can be experimentally reprogrammed not only into an embryonic stem cell-like state (1) but also into another, different, differentiated identity (2). Remarkably, this process, called direct cell-type conversion or transdifferentiation, also occurs naturally (2).

The molecular mechanisms underlying these events are still unclear and it remains to be determined whether key elements are shared between natural and induced reprogramming events. Factors used to reprogram differentiated cells to a stem cell-like state are important for embryonic stem (ES) cell self-renewal (1). Several studies have shed light on the molecular networks that maintain ES cell pluripotency (3). Key factors have been identified, such as Nanog (3) or SOX2 and OCT4 that are required for ES cell pluripotency and that, together with additional factors, can force fibroblasts into embryonic-like stem cells (1, 3). Besides these pluripotency factors, transcriptional repression complexes have been found necessary for ES cell self-renewal, but whether and how these complexes act together to control cellular potential during development remain to be determined. Among them, the NODE (Nanog and Oct4-associated deacetylase) complex has been recently identified in ES cells through the purification of factors associated with Nanog and has been proposed to repress the expression of Nanog and OCT4 target genes and to maintain the stem cell identity of ES cells via the inhibition of their differentiation (4). The NODE complex notably contains OCT4 and SALL4, known interaction partners of Nanog (3), as well as MTA1/2 (4). SALL4, a zinc finger transcription factor, has been shown to stabilize ES cell self-renewal and to enhance iPS reprogramming (4–9). MTA1 is a transcriptional modulator and appears to have an important role in cellular identity as its expression has been correlated with invasive cancers and metastasis (10) and as loss of MTA1 in ES cells impairs their pluripotency (4). Although SALL4 and MTA1 have also been found to associate with the nucleosome remodelling and histone deacetylase (NuRD) complex (11–13), most NuRD components did not appear to associate with the NODE complex (4).

To investigate the molecular requirements of natural direct reprogramming, we turned to a traceable single-cell direct reprogramming model. We have characterized a direct cell-type conversion event in the worm Caenorhabditis elegans, in which a rectal epithelial cell, called “Y”, changes its cellular identity to become a motor neuron named “PDA” in the normal course of development (14). This direct conversion follows a stereotyped sequence of events, allowing the identification of genes that mediate different steps in this process. In this study, we show that EGL-27/MTA, SEM-4/SALL, CEH-6/OCT, and SOX-2 are essential for the Y-to-PDA conversion and may act as a NODE-like complex, whereas the NuRD complex is dispensable for this process. Our study establishes a role for these factors together in a natural reprogramming event. Our data further point to a conserved plasticity module and suggest an important role of a NODE-like complex in regulating cell identity in vivo.

Results

Transcriptional Repressor EGL-27/MTA1 Controls Y-to-PDA Identity Changes.

Recent studies have shown that chromatin remodeling activities and members of repressor complexes such as the Polycomb Group proteins (15) and members of the NuRD (16) or the NODE (4) complexes, play important roles in the regulation of cell identity. We thus sought to test whether such repressor complexes impact on a natural reprogramming event, the transdifferentiation of the Y cell (14), in C. elegans. The Y cell is born during the early embryonic development of the worm when it becomes one of the six cells forming the rectum (Fig. 1A). Later on during larval development, the Y cell rescinds from the rectal tube, migrates away, and becomes a motor neuron called PDA, with a characteristic axon (Fig. 1B). Importantly, we have found that this process, which does not require a cell division, involves transition through a seemingly dedifferentiated, albeit not multipotent, intermediate (14, 17). To identify the molecular networks that endow the Y cell with the ability to change identity and dedifferentiate, we designed an RNAi screen targeting the early steps of Y-to-PDA. We built a RNAi sublibrary out of the Ahringer RNAi library (18), by identifying known or putative chromatin remodeling activities and their associated cofactors (Table S1). A primary RNAi screen allowed the identification of dsRNA that led to worms with a persistent Y-cell phenotype as observed by Nomarski optics (Fig. 1 B and C). The ability of these potential dsRNA candidates to impair the Y-to-PDA transition was then confirmed in a secondary screen, using a marker of the PDA neuron (14, 17) that clearly highlights PDA axon and cell body (Fig. 1B). Our RNAi screens identified egl-27 as required for Y direct reprogramming (Fig. 1D).

Fig. 1.

Initiation of the Y-to-PDA cell identity switch is defective in egl-27 mutants. (A) The rectal tube of the worm C. elegans is formed by three rings of two epithelial cells that are named Y, B, U, F, K, and K′. (B) The rectal area, before and after Y transdifferentiation. This process can be observed by DIC optics and PDA can be observed using a cog-1::GFP fluorescent marker (14, 17). (C) RNAi screening strategy used to discover candidates involved in the initiation of the Y-to-PDA cell identity switch. A primary screen was performed under DIC optics to select L3 stage worms with a persistent Y cell. A secondary screen confirmed the absence of PDA, using the cog-1::GFP marker. (D) RNAi knockdown of the egl-27 gene led to a defect in Y-to-PDA transdifferentiation in both our primary (second column) and secondary (third column) RNAi screens. % 3 epi. c, percentage of animals exhibiting a three epithelial cells in the anterior rectum mutant phenotype under Nomarski optics; % no PDA, percentage of animals where no PDA was formed, as detected using the PDA marker cog-1::gfp; n, total number of animals scored. Positive control, RNAi against the egl-5 gene (36); negative control, bacteria with an empty feeding vector. (E) The persistent Y cell remains a rectal epithelial cell in the egl-27(ok1670) loss-of-function mutant. The expression of epithelial (che-14::GFP) or rectal (egl-5p::GFP, egl-26p::GFP) markers was assessed in the persistent Y cell, which was found at its original position, in L3 and older larvae in egl-27(ok1670) mutants. Epi, epithelial; n, total number of animals scored.

We have previously shown that the Y-to-PDA transition is a multistep process that involves retraction from the rectum, followed by migration, a dispensable step (14, 17). An additional cell found in the anterior rectum of animals treated with egl-27 dsRNA (Fig. 1C) suggests that the Y cell remained in the rectum and that an early step of this process is affected. To establish whether Y reprogramming has been initiated in these animals, we analyzed the Y-cell characteristics (Fig. S1) in a deletion mutant for egl-27. Examination of the expression of the Y rectal cell markers egl-5 and egl-26 and of the epithelial marker che-14 (14, 17) showed that in egl-27 mutants a persistent Y cell is found at its original location in the rectum and remained a rectal epithelial cell (Fig. 1E). Thus, in egl-27 mutants the rectal epithelial identity of the Y cell has been correctly specified but Y-to-PDA reprogramming has not been initiated. We conclude that egl-27 activity is essential at a very early step and that in its absence the Y cell lacks the ability to change its identity.

SANT and ZnF Domains of EGL-27/MTA1 Are Required for Y-to-PDA.

egl-27 is one of the two C. elegans genes encoding a protein similar to the mammalian metastasis-associated protein family that includes MTA1 (19, 20). The nuclear factor MTA1 is a member of nuclear complexes with chromatin remodeling activities (10, 11) or transcriptional repression (4) and has been associated with ES self-renewal ability (4). However, which domains in MTA1 mediate its activity on cellular identity is not known. The longest EGL-27 protein contains bromo-adjacent homology (BAH), EGL-27 and MTA1 homology 2 (ELM2), SWI3/ADA2/N-CoR/TFIII-B (SANT), and Zinc Finger (ZnF) domains that are all located in the N terminus, followed by a long C terminus devoid of any recognizable domains apart from a coiled-coil motif (Fig. S1). To test which part of the EGL-27 protein is necessary and sufficient to allow the Y-to-PDA identity change, we determined the ability of different EGL-27/MTA1 protein fragments to rescue the Y-to-PDA defect of egl-27 mutants. Different portions of egl-27/mta1 cDNA were expressed under the control of a 6.8-kb egl-5/Hox promoter, which directs restricted expression in the rectal cells, from after Y birth to adulthood and allowed us to bypass the toxicity associated with a wider expression of the EGL-27 protein. We found that the full-length protein efficiently rescued the Y-to-PDA defect of the egl-27 mutant (Fig. 2A and Fig. S2), as did the N terminus (EGL-27^1–512) alone. These results are supported by our analysis of several deletion alleles in egl-27/mta1, suggesting that the EGL-27/MTA1 N terminus plays an important role during Y-to-PDA direct reprogramming (Fig. S1). Furthermore, the sole C-terminal part of the protein (EGL-27^517–1,129) was not able to rescue the PDA defect of egl-27 mutants (Fig. 2A and Fig. S2). These data suggest that the C terminus of EGL-27/MTA1 does not function by itself during the Y-to-PDA direct reprogramming. However, it may potentiate the activity of the EGL-27 N terminus, as alleles that affect the C-terminal region exhibit a low-penetrance Y-to-PDA mutant phenotype (Fig. S1). Finally, a form that lacks both the BAH and ELM2 domains (EGL-27^286–1,129) showed strong rescuing abilities (Fig. 2A and Fig. S2). Thus, within the N-terminal region, our data point to a crucial role of the SANT and ZnF domains for EGL-27 activity during natural reprogramming. SANT domains are found in several conserved transcriptional modulators such as the CoREST/SPR-1, MTA, SMRT, RERE, and NcoR proteins (21) and have been associated with a function in the nucleus, via interaction either with histone tails (21, 22) or with histone tail modifiers (for example, ref. 23). Given the EGL-27 similarity to MTA1, a component of nuclear complexes known to associate with the chromatin and impact on transcription, the importance of the SANT domain suggests that EGL-27 acts at the chromatin level during Y-to-PDA transdifferentiation. Additionally, these results suggest that egl-27 activity is required in one or more rectal cells to allow Y direct reprogramming. Consistent with the potential focus of egl-27 activity in Y or surrounding rectal cells, we found that an egl-27/mta1 reporter is expressed in the Y cell in the 1.5-fold embryonic stage as well as during the L1 larval stage, which precedes the initiation of Y transdifferentiation (Fig. 2B).

Fig. 2.

Characterization of egl-27 activity during the Y-to-PDA cell identity switch. (A) Rescuing of egl-27(ok1670) loss-of-function mutant with different isoforms of EGL-27. (B) Expression of egl-27 in the Y cell. The egl-27 gene, widely expressed (20), is expressed in the Y cell after its birth during embryogenesis (embryonic 1.5× stage shown) and continues to be expressed in the Y cell (larval L1 stage shown). (Center) Fluorescent images obtained with the stIs10165 [egl-27p::his-25::mcherry; unc119(+)] strain. (Left) The corresponding DIC images; (Right) a merge of both. The rectal slit is indicated and anterior is to the left. (C) Identification of potential interactors of egl-27 during the initiation of the Y-to-PDA cell identity switch. The presence of a PDA neuron, using cog-1::GFP (14), was assessed in mutants, when available, or in RNAi–knocked-down animals. *, RNAi “target 1” and “target 2” reflect nonoverlapping RNAi clones designed to control for potential off-target effects (Fig. S6); the RNAi Ahringer library (18) was used otherwise. Note that absence of a defect after RNAi may not reflect the null phenotype. &, putative null alleles were examined; £, we have previously shown that sem-4 null mutants displayed a completely penetrant “persistent Y” phenotype similar to the phenotype observed in egl-27 mutants (14); Emb, the available mutant causes embryonic lethality, and no L3 progeny could be analyzed; §, no mutant is available for these genes. (D, Upper) Schematic view of the sem-4a and -b isoforms. (Lower) The two isoforms of the zinc fingers-containing SEM-4 protein rescue the sem-4(n1971) null Y-to-PDA mutant phenotype. The a and b splices were expressed under the control of the egl-5/Hox promoter. p, two-tailed P value calculated using Fisher’s exact test; ***P < 0.001; *P < 0.05; **P < 0.01; n.s., not statistically significant. Tg, transgenic worms; Non tg, nontransgenic siblings scored as controls. Results represent the mean of triplicate scorings of a representative transgenic line for each construct ± SD; n, total number of worms scored. More Tg line scorings are shown in Figs. S2 and S4.

SOX-2 and Members of the NODE, but Not the NuRD, Complexes Are Crucial to Allow Y-to-PDA Natural Reprogramming.

Mammalian MTA1 is a member of the NuRD and NODE complexes (4, 11), which are associated with histone modification activities, notably histone deacetylase (HDAC). To investigate whether other members of these complexes are involved in Y-to-PDA transdifferentiation, we assessed the effect of a reduction of their activity, via RNAi or by examining loss-of-function or putative null mutants (SI Methods). Surprisingly, we found that none of the NuRD complex components we tested (chd-3/Mi2a, mep-1/Mbd3, lin-53/RbAp48, rba-1/RbAp46, dcp-66/p66, spr-5/Lsd-1, and lin-40, another gene with close similarity to MTA1) exhibited a Y-to-PDA defect (Fig. S3). In addition, we could not find evidence of HDAC involvement during Y-to-PDA using hdac-2 or -3 mutant analysis, hdac-1 RNAi depletion (Fig. S3), or HDAC inhibitors trichostatin A (TSA) or valproic acid (VPA) treatments (SI Methods). However, we were unable to completely exclude a requirement for HDAC as complete loss of HDAC activity in the worm causes early developmental arrest.

Because MTA1 was reported to associate with OCT4 in the NODE complex in ES cells (4), we also investigated the potential role of NODE components (Fig. 2C). Excitingly, the NODE complex was reported to also contain SALL4, a vertebrate homolog of the worm SEM-4. We have previously shown that mutants devoid of sem-4/SALL activity in worms exhibit a fully penetrant block of Y reprogramming (14). We found that RNAi-mediated knockdown of sem-4/SALL also affected Y-to-PDA (Fig. 2C). The defects we have found in the egl-27 mutant are similar to those observed in sem-4 mutants: In those animals, no reprogramming of the Y cell is initiated, and the Y cell remains as an epithelial rectal cell at its original position (14). Two different splice variants have been reported for sem-4/Sall (Fig. 2D), lacking (splice “b”) or not (splice “a”) a conserved 12-amino acids motif found in the N terminus of SALL proteins that has been involved in the recruitment of the NuRD complex and MTA1 (12, 13). Both splice variants, under the control of the egl-5/Hox promoter, were able to rescue the “no PDA” phenotype of the sem-4/SALL null mutant (Fig. 2D and Figs. S4 and S5), demonstrating that this N-terminal domain is not strictly necessary for SEM-4 activity during Y-to-PDA. Furthermore, these results show that sem-4 activity is required in the rectal cells, as is egl-27. These results are consistent with our findings that the NuRD complex does not influence Y identity change and that SEM-4/SALL could act as a NODE-like complex component.

We then examined the possible involvement of other NODE components. No clear homolog of Nanog is found in worms. However, three factors containing a POU domain related to OCT4 exist (24). We tested all three transcription factors (named ceh-6, unc-86, and ceh-18) and found that knockdown of one of them, ceh-6/OCT, led to a PDA defect, as observed with two different specific RNAi constructs (Fig. 2C and Fig. S6). In these animals, the Y cell appears to remain in the rectum (SI Methods) as in the egl-27 and sem-4 mutants, indicating that all three genes are necessary for the initiation of the Y-cell reprogramming. Consistently, ceh-6 is expressed in the Y cell early on during embryogenesis (24). Thus, all of the members of the NODE complex we could test are necessary for the first steps of Y-to-PDA direct reprogramming.

Various members of the POU family, including OCT4/Pou5f1, have been shown to interact and bind DNA with SOX-2 and to synergistically activate target gene expression (25). We thus tested whether sox-2 and other C. elegans sox family members (sox-1, sox-3, egl-13, and sox-4; Fig. S7) could be involved in Y identity change. Three different RNAi constructs targeted against sox-2 blocked Y-to-PDA transdifferentiation early, before initiation of the reprogramming event (Fig. 2C and Fig. S6). It is thus striking that four factors with a key role in the maintenance of ES cell pluripotency and/or in iPS reprogramming are necessary in vivo for a natural direct reprogramming event to occur.

EGL-27/MTA1 Associates with SEM-4/SALL, SOX-2, and CEH-6/OCT.

Among these four conserved factors, mammalian SOX-2 and OCT4 are known to physically interact (25), and in vitro coimmunoprecipitation (CoIP) experiments and mass spectrometry analyses (26–28) suggested that SOX-2 and SALL4, and OCT4 and SALL4, can be found associated in mammals. In addition, besides our findings that all four factors are genetically involved at the same step of the Y reprogramming event, egl-27/mta1, sem-4/sall, and ceh-6/oct are expressed in the Y cell since around its birth (14, 24). To assess the potential of these proteins to work together in common complexes or associated subcomplexes, we turned to a HeLa cell culture assay, as wide expression of egl-27 was toxic to worms. We first examined the subcellular localization of EGL-27 and SEM-4 and their ability to colocalize. Both proteins were found in the nucleus when transfected (Fig. 3B, a and o), and these two factors colocalized in the nucleus when cotransfected (Fig. 3B, b). Interestingly, N- or C- terminal truncated EGL-27 proteins that do not contain amino acids 511–741, such as EGL-27^[138–511] or EGL-27^{[741–1,129]}, were exclusively or significantly found in the cytoplasm (Fig. 3B, e, g, k, and m). SEM-4 could potentiate the nuclear translocation of these two EGL-27 proteins (Fig. 3B, e–h) but not of smaller C-terminal parts (Fig. 3B, k–n). These results suggest that amino acids 553–741 in EGL-27/MTA1 contain a sequence that enables nuclear localization on its own, but that additional factors, like interaction with SEM-4/SALL, can also mediate nuclear localization.

Fig. 3.

Molecular dissection of EGL-27/MTA1 and its ability to interact with SEM-4a/b, SOX-2, and CEH-6. (A) Schematic view of EGL-27/MTA1 variant proteins used. Ability to CoIP with SEM-4a/b, assessed in HeLa cells, is summarized on the Right. We found that the full-length C. elegans EGL-27/MTA1 appeared toxic to HeLa cells, as in C. elegans, decreased dramatically the transfection efficiency and resulted in very few expressing cells, precluding any biochemical study. This limitation was eliminated by removing the N terminus of the protein up to the middle of the BAH domain (amino acid 138). C-terminal parts of EGL-27 (amino acids 899–1,046 and amino acids 1,027–1,129) did not show any ability to interact with SEM-4a/b in immunofluorescence (IF) confocal observations or in CoIP experiments; it was subsequently used as a negative control for CoIP experiments. (B) IF of HeLa cells expressing different EGL-27 variants fused to CFP (in green) and YFP-fused SEM-4a protein (in red) (b, d, f, h, j, i, n, and o). Hoechst-stained cell nuclei are in blue. Areas in which SEM-4a and EGL-27 colocalize are in yellow. (b) The orthogonal view of the area of colocalization is shown as an Inset (Upper Right). White arrows point out speckles in which SEM-4a and EGL-27 colocalize (d, f, and h). (a–f and i, j) Native EGL-27, EGL-27^[138–738], and EGL-27^{[553–1,129]} proteins are found in the nucleus and colocalize with SEM-4a/SALL; (g, h and k–n) EGL-27^[138–511], EGL-27^{[741–1,129]}, EGL-27^{[899–1,046]}, and EGL-27^{[1,027–1,129]} proteins are mainly found outside of the nucleus. Cotransfection of SEM-4/SALL triggers nuclear colocalization of EGL-27^[138–511] and EGL-27^{[741–1,129]}, but not of EGL-27^{[899–1,046]} or EGL-27^{[1,027–1,129]}. Thus, although colocalization with SEM-4 is not predictive of the ability to rescue the egl-27 mutant phenotype, all rescuing EGL-27 isoforms show colocalization with SEM-4 in the nucleus. (C) CoIP and Western-blot (WB) analysis of C. elegans Flag-tagged truncated forms of EGL-27 and HA-tagged SEM-4a or -b in transfected HeLa cells. (D) CoIP and WB analysis of C. elegans Flag-tagged EGL-27^{[741–1,129]} and HA-tagged SEM-4a, CEH-6, or SOX-2 in transfected HeLa cells. (E) CoIP and WB analysis of Flag-tagged CEH-6 and HA-tagged SEM-4a/b or SOX-2 in transfected HeLa cells.

We next evaluated the ability of several EGL-27 fragments to coprecipitate with SEM-4 (Fig. 3A). The C-terminal part of EGL-27^{[553/741–1,129]} exhibited both colocalization (Fig. 3B, e, f and i, j) and a strong interaction activity with SEM-4 (Fig. 3C). However, neither the extreme C-terminal part (amino acids 1,027–1,129) nor a fragment encompassing a coiled-coil domain (amino acids 899–1,046) in EGL-27 mediates the association with SEM-4 (Fig. 3 A, B, k–n, and C). Thus, our colocalization and coprecipitation data indicated that the region between amino acids 741 and 899 mediates a strong association between EGL-27 and SEM-4 (Fig. 3 A–C). In addition, we found that the N-terminal part of EGL-27^{[138–511/738]} (Fig. 3A) was also able to associate with SEM-4 (Fig. 3 A, B, c, d and g, h, and C). Thus, two regions within EGL-27 mediate its association with SEM-4, including the region bearing rescuing activity. Notably, the two SEM-4/SALL a and b isoforms showed the same interaction properties with EGL-27/MTA1. We next asked whether we could detect an association between EGL-27/MTA1 and SOX-2 or CEH-6/Oct. Coimmunoprecipitation experiments demonstrated that EGL-27^{[741–1,129]} exhibits significant interaction with SOX-2 and CEH-6 (Fig. 3D) and conversely that CEH-6/OCT interacts with SOX-2 and the two isoforms of SEM-4/SALL (Fig. 3E). Thus, these four transcriptional modulators can associate in cells and could exert their activity as a multiproteic complex.

EGL-5/AbdB Hox Protein Likely Acts Downstream of EGL-27 Complex.

We have previously shown that loss of the posterior hox gene egl-5 leads to a similar early defect in Y reprogramming (14). We therefore investigated the relationship between egl-27 and egl-5. As expected for two genes involved in the same process, we found that a hypomorph mutant allele of egl-5 that does not result in a Y-to-PDA defect by itself enhances the Y-to-PDA defect found in mutants animals bearing a weak egl-27 allele (Fig. 4A). As egl-27 has been shown to modulate both hox gene expression and HOX protein activity in the ventral hypodermal cells (29), we examined whether egl-5 could act downstream of egl-27 to allow initiation of Y reprogramming. We did not detect a change in expression of a reporter encompassing 13 kb of the egl-5 locus in the rectal cells (SI Methods), where we showed egl-27 activity to be necessary (Fig. 2A). However, ectopic expression of EGL-5 in the strong egl-27(ok1670) mutant led to a significant rescue of the Y-to-PDA defect (Fig. 4B). These results suggest that egl-27 positively modulates EGL-5 activity. Interestingly, inducing ectopic expression of EGL-5 as late as in early L1 animals was sufficient to rescue the Y-to-PDA defect (Fig. 4B), suggesting that EGL-5 activity is needed just before the initiation of the process rather than to endow the Y cell after its birth with the competence to change identity, a step requiring lin-12/Notch (14). In addition, we found that egl-27 and sem-4 are also required around the same time, by demonstrating the ability of an egl-27 or a sem-4 cDNA expressed under the col-34 promoter to rescue the no PDA defect of the corresponding mutant (Figs. S2 and S4, respectively).

Fig. 4.

EGL-5/Hox likely acts downstream of EGL-27. (A) Genetic interaction between egl-27 and egl-5. An egl-5/Hox weak loss-of-function allele enhances the Y-to-PDA defect found in egl-27(ok151) mutants. (B) Ectopic expression of EGL-5/Hox rescues the Y-to-PDA defect in egl-27 mutants. By contrast, ectopic expression of EGL-5 or heat shock of wild type (WT) worms did not affect the Y-to-PDA process. Tg, transgenic worms; Non tg, nontransgenic siblings scored as controls; n, total number of animals scored; ***P < 0.001; **P < 0.01. Results represent the mean of at least five scorings for each strain ± SD. (C) Working model for the interaction network of SEM-4a/b/SALL, CEH-6/Oct, and SOX-2 with EGL-27/MTA and downstream EGL-5/Hox.

Discussion

In this study, we have examined the molecular requirements for the initiation of a natural direct reprogramming event in vivo, at the single-cell level within a physiological context. In mammalian cells, a mixture of four transcription factors is enough to reprogram cells ex vivo (1). Here, we have identified four conserved nuclear factors that are all strictly required in vivo at the same early step of the Y-cell reprogramming in vivo. Lowering the activity of any of these factors results in a unique phenotype, the incapacity for a rectal cell to initiate its normally occurring transdifferentiation. Furthermore, we have found that these four factors, whose mammalian counterparts are individually known to impact on the potential of murine embryonic stem cells as complexes, associate in cells. Our results suggest a model where nuclear events involving these four factors, which together may represent a conserved plasticity module, provide a permissive context for the reprogramming of a rectal cell into a motor neuron.

We identified egl-27 through a reverse genetic screen for mutants unable to initiate Y direct reprogramming and defined the protein domains mediating EGL-27 activity during this process that may also be relevant to the MTA1 requirement for ES cell pluripotency. Furthermore, we have identified the modulation of EGL-5/Hox activity as a likely downstream event. We found that egl-27 and sem-4 activities are not required before the end of the embryonic development and slightly before the egl-5 requirement, consistent with a role upstream of EGL-5. It is thus possible that chromatin reorganization leads to activation of EGL-5, as has been proposed for egl-27 modulation of MAB-5/Hox activity in the Pn.p cells (29). The identification of the key role of egl-27 in Y-to-PDA led us to investigate potential partners. We found that a reduction of the activity of the sox-2, ceh-6/Oct, and sem-4/SALL genes led to the same early block of Y reprogramming, establishing that these four nuclear factors act together in a natural cell plasticity process. The mammalian counterparts for these factors have been individually shown to be essential for mouse development, either for inner cell mass proliferation (Sall-4) (6, 8, 30) or pluripotency (Oct-4) (31) or for epiblast pluripotency (Sox-2) (32)—the potential impact of Mta1 loss on early embryonic development has not been reported. However, the phenotypes differ and whether these factors act together in vivo during vertebrate development is still an open question. Indeed, most of what is known about the activity of these factors in a common process comes from studies using ES cells or iPS reprogramming (3). Our data suggest that these factors act together in very different biological contexts, including during natural cell-type conversion.

All of the factors that we have identified here impact on ES cells’ self-renewing abilities (3). Could they nevertheless play a very different role during Y-to-PDA transdifferentiation? For example, a requirement for Sox2 in the maintenance of neural stem cells or in the differentiation of specific neuron subtypes has been described (33). It could thus be conceivable that sox-2 could function as a neural determinant during Y-to-PDA. However, we have shown that initiation of Y direct reprogramming leads first to a loss of epithelial and rectal characteristics, an apparent dedifferentiation, before neural markers are switched on (17). Thus, the persistence of a rectal Y cell observed in sox-2 depleted animals is not consistent with a role in the subsequent adoption of a neural identity, but rather with a lack of initiation of the process. In addition, data obtained in the Hobert laboratory indicate that embryos mutant for sox-2 do not exhibit neuronal cell loss or defects, suggesting that sox-2 is not involved in C. elegans neurogenesis. It is thus likely that the role of sox-2 during Y-to-PDA does not involve neural fate promotion.

Altogether, our data suggest that these four factors act as a multiproteic C. elegans NODE-like complex required for Y competence to change identity (Fig. 4C). The NODE-like complex active in the Y cell doubtless contains more activities, and it will be of high interest to determine which and how their combinatorial activities in this cell make it uniquely able to switch its identity. By contrast with ES and iPS cells, an important aspect will be to understand why the cellular potential of the Y cell does not widely increase during the process. Our data raise the intriguing possibility that these factors may represent a conserved plasticity cassette. These factors may promote an “open state” that allows cells to adopt subsequently a distinct fate. Alternatively, in line with their role during iPS reprogramming, these factors could provide the necessary transcriptional environment and chromatin structure to allow the Y cell to repress its rectal epithelial expression program and lose its initial identity. Furthermore, our data raise the intriguing possibility that pluripotent reprogramming borrows to some extent from mechanisms at work during natural reprogramming events. In further support of this possibility, Sall4, Oct4, and Sox2 are expressed during fin regeneration in fish or Xenopus, a process that requires dedifferentiation of the tissue cells surrounding the amputation site, and Oct4 and Sox2 are necessary for regeneration to take place (34, 35). Our studies highlight a unique function and association for these factors in an in vivo single-cell reprogramming event that could be promising for fundamental developmental biology, regenerative medicine, and cancer therapies.

Methods

Strains.

Standard methods were used for C. elegans handling. The relevant genes and alleles used in this study are described in SI Methods.

RNAi by Feeding.

To increase the sensitivity of our RNAi screen, the RNAi hypersensitive mutation rrf-3(pk1426) or a rrf-3(pk1426); egl-5(n1489) sensitized background were used. RNAi experiments were performed as previously described (36), and all clones from the Ahringer library were confirmed by sequencing. Bleached eggs were put on plates and their progeny scored for Y-to-PDA phenotype.

Cell Transfection Procedures and Biochemical Analyses.

A total of 5 × 10⁵ HeLa cells were cultured in a 100 × 20-mm dish with Gibco DMEM–5% FCS. Ten micrograms of plasmids DNA added to 20 μL of jetPEI were used for transfection. α-HA-tag^HRP (Cell Signaling) and α-FLAG-tag^HRP (Sigma) antibodies (Western blot) or α-HA-tag and α-FLAG-tag coupled to agarose beads from Sigma (immunoprecipitations) were used. Cells were fixed for 15 min with 4% paraformaldehyde before observations.

Microscopy and Imaging.

DIC and epifluorescence observations were performed using a Zeiss Z1 imager microscope. In vivo worm pictures of the rectal area were acquired using a Leica Spinning Disk confocal microscope. Transfected HeLa cells were observed using a Leica SP2 MP confocal microscope.

Additional materials and methods are described in SI Methods.