Both land plants and metazoa have the capacity to reprogram differentiated cells to stem cells. Here we show that the moss Physcomitrella patens Cold-Shock Domain Protein 1 (PpCSP1) regulates reprogramming of differentiated leaf cells to chloronema apical stem cells and shares conserved domains with the induced pluripotent stem cell factor Lin28 in mammals. PpCSP1 accumulates in the reprogramming cells and is maintained throughout the reprogramming process and in the resultant stem cells. Expression of PpCSP1 is negatively regulated by its 3′-untranslated region (3′-UTR). Removal of the 3′-UTR stabilizes PpCSP1 transcripts, results in accumulation of PpCSP1 protein and enhances reprogramming. A quadruple deletion mutant of PpCSP1 and three closely related PpCSP genes exhibits attenuated reprogramming indicating that the PpCSP genes function redundantly in cellular reprogramming. Taken together, these data demonstrate a positive role of PpCSP1 in reprogramming, which is similar to the function of mammalian Lin28.
Stem cells can self-renew and produce cells to be differentiated during development1,2,3,4. On the other hand, differentiated cells can change their cell fate to stem cells under certain conditions in both land plants and metazoa3,4. In flowering plants, differentiated cells can form undifferentiated cell masses called callus. With the addition of the appropriate phytohormones they can regenerate shoot and root meristems including stem cells, as was first shown with carrot in 1958 (ref. 5). Several genes have been shown to be involved in the formation of callus or regeneration of stem cells in Arabidopsis thaliana (Arabidopsis). Overexpression of a plant-specific AP2/ERF transcription factor ENHANCER OF SHOOT REGENERATION 1 (ESR1)/DORNROESCHEN (DRN) promotes the formation of shoot meristems from callus6. Induction of another AP2/ERF transcription factor WOUND INDUCED DEDIFFERENTIATION 1 (WIND1) enhances callus formation without exogenous hormones7. In bryophytes, differentiated cells have a remarkable ability of being reprogrammed into stem cells without callus formation. In the moss Physcomitrella patens (Physcomitrella), wounding can induce the transition from differentiated leaf cells into proliferating chloronema stem cells without any exogenous phytohormones8,9. To understand the molecular mechanisms underlying this reprogramming, transcriptome analysis was performed during the reprogramming10 and several factors were identified as playing a role in the process. For instance, Cyclin-Dependent Kinase A (CDKA) activation is essential for cell cycle re-entry during reprogramming9. WUSCHEL-related homeobox 13-like (WOX13L) genes are required for the initiation of tip growth during stem cell formation11.
In mammals, the induction of four factors is sufficient to reprogram somatic cells to pluripotent stem cells. Oct4, Sox2, cMyc and Klf4 were first reported as induced pluripotent stem cell (iPSC) factors able to reprogram mouse fibroblast cells into pluripotent stem cells12. Later, the same factors were applied to human fibroblast cells to generate iPSCs13. At the same time, another set of pluripotency factors Oct4, Sox2, Nanog and Lin28 was identified, which could successfully induce pluripotent stem cells from human fibroblast cells14. So far, factors belonging to the same gene family and functioning in reprogramming from differentiated cells to stem cells have not been identified between land plants and metazoa. Therefore, it is still unknown whether plants and animals use similar mechanisms for the reprogramming from differentiated cells to stem cells.
Here, we report that P. patens Cold-Shock Domain Protein 1 (PpCSP1), which shares highest sequence similarity and domain structure with Lin28 in metazoa, enhances reprogramming in Physcomitrella. PpCSP1 accumulates in the reprogramming cells. PpCSP1 expression is negatively regulated by its 3′-untranslated region (3′-UTR). When the 3′-UTR is removed, PpCSP1 transcripts increase and the reprogramming is enhanced. Deletion of PpCSP1 and three closely related PpCSP genes causes attenuated reprogramming, demonstrating a positive and redundant function of PpCSPs in the reprogramming.
PpCSP1 shares conserved domains with Lin28
Cold-shock domain proteins (CSPs) were first identified in bacteria as proteins expressed under cold-shock conditions15, and were later implicated in the process of cold acclimation in flowering plants as CSP transcripts accumulate after cold treatment in Arabidopsis and wheat16,17,18. The cold-shock domain (CSD) is highly conserved in bacteria, land plants and metazoa. CSD possesses nucleic acid binding activity and is capable of binding to single-stranded DNA/RNA and double-stranded DNA19. To better understand the evolution of CSPs, we investigated the function of the PpCSP1 gene in Physcomitrella since no previous study had focused on CSPs in non-flowering plants20. To characterize the expression pattern of PpCSP1, we generated a PpCSP1-Citrine fusion protein line (nPpCSP1-Citrine-nosT; Supplementary Fig. 1a,b). Using live imaging, we detected predominant PpCSP1-Citrine signals in chloronema and caulonema apical stem cells, which self-renew and produce cells that differentiate into chloronema and caulonema cells, respectively (Fig. 1a,b and Supplementary Fig. 1c). The signals were also detected in chloronema and caulonema side branch initial cells, which are typically destined to become chloronema apical stem cells (Fig. 1a,b). These results suggested the possible involvement of PpCSP1 in stem cell maintenance and in the reprogramming of differentiated chloronema and caulonema cells to chloronema apical stem cells8. In addition to CSD, a search for conserved domains21 (www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) in PpCSP1 identified a provisional domain PTZ00368 (universal minicircle sequence-binding protein), which is comprised of two CCHC zinc-finger domains (Fig. 1c). Most plant CSPs and some animal CSPs also have CCHC zinc-finger domains but bacteria CSPs do not. We then performed BLASTP searches using the PpCSP1 sequence as a query to identify proteins related to PpCSP1. Lin28 proteins were the top hits when the BLAST searches were performed against the database of metazoa, including Homo sapiens, Mus musculus and Caenorhabditis elegans. The Lin28 proteins, one of which is an iPSC factor in human, share one CSD and two CCHC zinc-finger domains with PpCSP1 (ref. 22 and Fig. 1c). We subsequently inferred phylogenetic relationships of PpCSP1 and other proteins with these three domains using the maximum likelihood tree reconstruction method of RAxML23. Although the low resolution of the phylogenetic tree did not enable us to examine whether PpCSP1 is orthologous or paralogous to Lin28 (Fig. 1d), PpCSP1 and Lin28 should be homologous because of the shared domains and these results led us to investigate whether PpCSP1 plays a role similar to Lin28 in reprogramming differentiated cells to stem cells.
PpCSP1 mRNA and protein accumulate during reprogramming
To investigate the function of PpCSP1 in reprogramming, we cut gametophore leaves and cultivated them on a medium without phytohormones9. Gametophores are shoots formed in the haploid generation (Supplementary Fig. 2a). When a differentiated leaf is excised from a gametophore, leaf cells facing the cut change to chloronema apical stem cells with tip growth and divide ∼30 h after excision9 (Supplementary Fig. 2b). A chloronema apical stem cell divides to regenerate itself and form a chloronema subapical cell. Therefore, chloronema apical stem cells fulfil the definition of a stem cell: they self-renew and give rise to cells that go on to differentiate. All leaf cells with tip growth behave as chloronema apical stem cells9 and this acquisition of a new fate is the most reliable sign of the reprogramming at present. To examine the spatiotemporal expression pattern of the PpCSP1 protein in cut leaves, we removed the DNA fragment containing the nopaline synthase polyadenylation signal (nosT)9 and the neomycin phosphotransferase II (nptII)24 expression cassette from the nPpCSP1-Citrine-nosT line by transiently expressing Cre recombinase25. As a result, the native 3′-UTR was fused to the PpCSP1-Citrine coding sequence (CDS) (nPpCSP1-Citrine-3′-UTR line; Supplementary Fig. 1a). During the reprogramming process, Citrine signals specifically increased in leaf cells facing the cut just after excision (Fig. 2a,b; Supplementary Movie 1; and Supplementary Fig. 1d). The Citrine signals increased continuously until tip growth started. Even though the Citrine signal increased in most edge cells, fewer than half of the edge cells protruded. These observations suggest that other factors unevenly distributed in the edge cells are also involved in reprogramming. After the protrusion, PpCSP1-Citrine signals localized more conspicuously at the phragmoplast than other parts in the cytosol. The signals were dispersed in the cytosol after cytokinesis with remaining signals at the cell septum. The signals at the phragmoplast decreased during subsequent cell divisions of chloronema apical stem cells (Supplementary Fig. 3 and Supplementary Movie 2). These indicate that PpCSP1 protein predominantly accumulates in the leaf cells facing the cut, accumulates during reprogramming, gradually decreases after reprogramming, and is maintained in stem cells. In the growing protonemata, PpCSP1-Citrine was continuously expressed in apical stem cells during the entire cell cycle (Supplementary Movie 3). When side branch cells initiated, PpCSP1-Citrine signals increased during protrusion and localized at the phragmoplast. The signals at the phragmoplast decreased during subsequent cell divisions (Supplementary Movie 3). In addition, PpCSP1 was expressed in proliferating cells in gametophore apices where both stem cells and proliferating non-stem cells exist8 (Supplementary Fig. 1e). This is reminiscent of Lin28, which regulates cell cycles in stem cells26,27. PpCSP1-Citrine localized in the cytosol but not in the nucleus (Fig. 2c). Because of the presence of the CSD and zinc-finger domains, it is plausible that PpCSP1 functions as an RNA-binding protein to regulate mRNA maturation, stability, or translation in the cytosol in a manner similar to that reported for other CSPs18,19, including Lin28 and related proteins in metazoa.
To analyse the promoter activity of PpCSP1, we made a transcriptional fusion (PpCSP1pro:LUC), in which the coding sequence of luciferase (LUC)28 is driven by the 1.8 kb PpCSP1 promoter. This construct was integrated into the PIG1 neutral site29,30 of the nPpCSP1-Citrine-3′-UTR background line (Supplementary Fig. 2c,d). With this dual reporter construct (PpCSP1pro:LUC nPpCSP1-Citrine-3′-UTR), we are able to simultaneously monitor promoter activity and protein accumulation at a single-cell level (Fig. 2d–f and Supplementary Fig. 2e–j; and Supplementary Movie 4). Time-lapse imaging showed LUC signals from PpCSP1 promoter activity increasing after excision (Fig. 2e). In edge cells that would later protrude, the intensities maximized at ∼12 h and were maintained with some fluctuation (Fig. 2e, left). However, the rates of increase and the maxima of the intensities varied among cells. In edge cells that never protruded, LUC signals initially increased but were not maintained as they were in the protruded edge cells (Fig. 2e, right). PpCSP1-Citrine levels in edge cells that would protrude continued to increase from 24 to 36 h, until these cells divided (Fig. 2f). In edge cells that never protruded, Citrine accumulation reached a maximum in 24–36 h and then gradually declined, which is consistent with the changes in promoter activity (Fig. 2e). The smaller variation in protein levels than in promoter activity in cells that eventually protrude (Fig. 2e,f, left) suggests the potential involvement of post-transcriptional regulation or a difference in stability of the transcripts and proteins of PpCSP1.
PpCSP1 is negatively regulated through its 3′-UTR
Lin28 is negatively regulated by microRNA (miRNA) let-7 (refs 31, 32, 33, 34), which directly binds to Lin28 transcripts at the 3′-UTR leading to the degradation of Lin28 transcripts31. In the Physcomitrella genome, we could not identify a miRNA similar to let-7 (refs 35, 36, 37, 38). However, the 3′-UTR of PpCSP1 is 623 bp, which is longer than the median length (334 bp) of 3′-UTRs in the Physcomitrella v1.6 genome sequence39. This suggests that regulatory elements could be located in the 3′-UTR. To determine if the 3′-UTR of PpCSP1 is involved in regulating transcript abundance, we performed 5′-digital gene expression (5′-DGE) analysis10 in the nPpCSP1-Citrine-3′-UTR and nPpCSP1-Citrine-nosT lines, in which the 3′-UTR is separated from the PpCSP1-coding region by the nosT and the nptII expression cassette (Supplementary Fig. 1a). We compared these results to previously published 5′-DGE data of leaf cut experiments10 (Fig. 3a). In the 5′-DGE analysis, ∼25-bp cDNA fragments at the 5′-ends of polyadenylated RNAs are sequenced. The tags in the 5′-UTR or CDS represent RNA molecules that are not cut in the 3′-UTR, while tags in the 3′-UTR represent RNAs that are cut or undergoing degradation. The number of tags in the PpCSP1 5′-UTR or CDS tended to increase after the leaf cut and nPpCSP1-Citrine-nosT had a generally higher value than nPpCSP1-Citrine-3′-UTR (6.6-fold in median). In wild-type and nPpCSP1-Citrine-3′-UTR lines, more sequenced tags were mapped to the 3′-UTR than the 5′-UTR or CDS, while in the nPpCSP1-Citrine-nosT line more tags were mapped to the 5′-UTR or CDS than to the exogenous 3′-UTR of nosT (Fig. 3a). These data suggest that the 3′-UTR of PpCSP1 is a degradation target similar to that of Lin28, or has a weak polyadenylation signal.
To examine the activity of the 3′-UTR, independent of its original genomic context, we generated constructs with a constitutively active elongation factor-1α (EF1α) promoter40 -driven sGFP41, fused to either the PpCSP1 3′-UTR or nosT, introduced into the PTA1 neutral site40 (Supplementary Fig. 4a–c). sGFP intensity in the EF1αpro:sGFP-nosT line increased in all of the examined leaf cells during reprogramming after cutting (Fig. 3b,c and Supplementary Fig. 4e–h). The increase of activity was more conspicuous in edge cells than in non-edge cells (Fig. 3c). On the other hand, in the 3′-UTR-fused line, cellular signals of both edge and non-edge cells (Fig. 3d,e and Supplementary Fig. 4i–l) were ∼10 times weaker than those in the nosT-fused line (Fig. 3c,e and Supplementary Fig. 4e–l). To examine the degradation activity of the 3′-UTR under unwounded conditions, sGFP signals were compared in protonemata and gametophores between the two lines (Fig. 3f,g). In gametophores and protonemata, signals of the 3′-UTR-fused line were weaker than those in the nosT-fused line as in the reprogramming process. Reverse transcriptase-quantitative PCR (RT-qPCR) determined that transcript levels of sGFP were 67.3±1.5-fold (mean±s.d., n=3) and 57.2±3.1-fold (mean±s.d., n=3) higher in the nosT than the 3′-UTR line in gametophores and protonemata, respectively. These results indicate that the PpCSP1 3′-UTR contains negative regulatory signals that function, independently of the PpCSP1 promoter, during the reprogramming process in cut leaves, as well as during regular development.
PpCSP1 does not appear to be regulated by a microRNA
miRNAs-evolved independently in land plants and metazoa42,43,44. However, some similarities exist between these two linages, such as conserved components like Dicer/Dicer-like and Argonaute proteins42. In addition, two possible Arabidopsis miRNAs (miRNA854 and miRNA855) were identified to be shared between land plants and metazoa and had binding sites within the 3′-UTR of the target mRNA45. To test whether a similar miRNA-associated regulation as let-7 miRNA regulates Lin28, we made a deletion series of the PpCSP1 3′-UTR fusing each fragment after the stop codon of the sGFP reporter gene driven by the constitutive rice Actin 1 promoter46,47 (Fig. 3h). These constructs were transiently introduced into gametophore leaf cells by particle bombardment and co-bombarded with a fragment containing the monomeric Red Fluorescent Protein 1 (mRFP) gene48 driven by the same Actin 1 promoter for normalization (Fig. 3h,i). The linear correlation of the sGFP and mRFP signals in the transformed cells was confirmed (Supplementary Fig. 4d). In comparison to the control (no UTR), signal intensities of sGFP fused with 623-, 500-, 400-, 300- and 200-bp 3′-UTR fragments decreased to 11.8, 15.7, 29.4, 54.9 and 74.5%, respectively (Fig. 3j). This gradual reduction suggests that several different regions in the 3′-UTR serve as targets for the negative regulation. We subsequently searched candidate miRNAs using the 3′-UTR as a query in the psRNATarget website (http://plantgrn.noble.org/psRNATarget/)49 and analysed small RNAs at the PpCSP1 locus in Plant Small RNA Genes WebServer (https://plantsmallrnagenes.psu.edu/cgi-bin/Ppatens_Locus_Reporter)35. However, we could not find any miRNA-targeting sequences in the 3′-UTR. In the future, additional studies such as genome-wide mRNA-protein interaction analysis50, will be needed to fully understand the molecular mechanisms of the degradation function of the PpCSP1 3′-UTR.
Increase of PpCSP1 transcript levels enhances reprogramming
Having determined that the 3′-UTR has a degradation function, we quantified transcript levels in the nPpCSP1-Citrine-nosT line, and compared them with the nPpCSP1-Citrine-3′-UTR line and wild type. Using RT-qPCR we found that transcript levels were 6.0±2.9-fold (mean±s.d., n=3) and 9.9±2.5-fold (mean±s.d., n=3) higher in the nPpCSP1-Citrine-nosT line as compared with the nPpCSP1-Citrine-3′-UTR line and wild type at 0 h after leaf cutting, respectively. These results are in agreement with the 5′-DGE analysis as the tag counts in nPpCSP1-Citrine-3′-UTR were not drastically different when compared with wild type. Collectively, these results indicate that transcript levels of PpCSP1 increased in the nPpCSP1-Citrine-nosT line.
As the PpCSP1 transcript level is ∼10-fold higher in the nPpCSP1-Citrine-nosT line, we found that this transcript increase results in protruding non-edge cells (Fig. 4a–c). However, only edge cells protrude in wild type (Supplementary Fig. 2b). We calculated percentages of excised leaves with at least one protruding non-edge cell in wild-type, nPpCSP1-Citrine-nosT, and nPpCSP1-Citrine-3′-UTR lines (Fig. 4a,b). While the percentages of excised leaves with protruding edge cells did not differ among these lines (Fig. 4a), those with protruding non-edge cells significantly increased in nPpCSP1-Citrine-nosT (Fig. 4b). Moreover, some non-edge cells of nPpCSP1-Citrine-nosT exhibited stronger Citrine signals than surrounding cells, some of which were reprogrammed to stem cells (Fig. 4c,d; Supplementary Fig. 5a–d; and Supplementary Movie 5), while Citrine signals of nPpCSP1-Citrine-3′-UTR lines were detected in cells at the cut edge but not in non-edge cells (Fig. 2a).
To confirm the increase in protruding non-edge cells in the nPpCSP1-Citrine-nosT line, we produced a PpCSP1pro:PpCSP1-Citrine line. In this construct, the PpCSP1 promoter, PpCSP1 CDS and Citrine gene were inserted into the neutral PTA1 site (Supplementary Fig. 5e,f), which enabled us to visualize increased PpCSP1-Citrine levels. RT-qPCR analysis indicated that transcript levels of PpCSP1 were 15.5±3.7-fold (mean±s.d., n=3) higher in PpCSP1pro:PpCSP1-Citrine as compared with wild type at 0 h after leaf cutting. Spatiotemporal patterns of Citrine signals and protruding cells in the PpCSP1pro:PpCSP1-Citrine line were similar to those of the nPpCSP1-Citrine-nosT line (Fig. 4e–g). We conclude that the protruding non-edge cell phenotype resulted from the transcript increase of PpCSP1-Citrine. On the other hand, we could not find morphological and growth differences in protonemata and gametophores between wild-type and nPpCSP1-Citrine-nosT lines (Supplementary Fig. 8).
To investigate the relationship between PpCSP1 and other factors involved in the reprogramming, we analysed transcript levels of WOX13-like genes11 in nPpCSP1-Citirine-3′-UTR and nPpCSP1-Citrine-nosT lines with 5′-DGE during the reprogramming of cut leaves. However, no significant differences were observed in the transcript levels (Supplementary Fig. 6a,b). On the other hand, PpCSP1 transcript levels investigated with the 5′-DGE data in Δppwox13lab line11 were detected to be lower than those in wild type at 24 h after dissection, while PpCSP1 transcripts were similarly induced until 6 h in wild type and the mutant (Supplementary Fig. 6c). These results suggest that PpCSP1 is positively regulated by WOX13-like genes but PpCSP1 does not regulate WOX13-like genes.
PpCSP quadruple deletion attenuates reprogramming
Deletion of the PpCSP1 gene (Supplementary Fig. 7a) resulted in no detectable difference in reprogramming between wild type and the mutant (Fig. 5a,b). There are three closely related genes, PpCSP2, PpCSP3 and PpCSP4, (Fig. 1d) in the Physcomitrella genome39,51. We generated single (ppcsp2, ppcsp3 and ppcsp4), double (ppcsp1 and ppcsp2), triple (ppcsp1, ppcsp2 and ppcsp3) and quadruple (ppcsp1, ppcsp2, ppcsp3 and ppcsp4) deletion mutants (Supplementary Fig. 7a–g). The percentage of excised leaves with reprogrammed cells was similar to the wild type in all single-, double- and triple-deletion mutant lines in both edge and non-edge cells (Fig. 5a,b). However, in the quadruple deletion mutant lines, cell protrusion was delayed (Fig. 5c). The delay was more severe in non-edge cells and was significant until 72 h (Fig. 5c,d), when chloronemata covered the excised leaves and further observation was impossible. Collectively, these results indicate that the four PpCSP genes are positive regulators of the reprogramming and possess redundant functionality.
PpCSP1 was expressed in not only stem cells but also proliferating non-stem cells in gametophore apices (Supplementary Fig. 1e) and appeared to localize at the phragmoplast (Supplementary Fig. 3 and Supplementary Movie 2). These data suggest the possibility that PpCSP1 is not involved in the reprogramming but in general cell cycle progression. To examine this possibility, we analysed the phenotype of the quadruple deletion mutant and the PpCSP1 transcript-increased line in protonemata and gametophores. We could not distinguish the protonemata and gametophores of the quadruple deletion mutant and the transcript-increased line from those of wild type (Supplementary Fig. 8a–f). Moreover, the duration of cell cycles of protonemata of these lines was measured with time-lapse observation and we could not find any differences (Supplementary Fig. 8g and Supplementary Movie 6). These results suggest that PpCSP1 does not play a major role in cell cycle progression in protonemata.
When we added a DNA synthesis inhibitor, aphidicolin to cut leaves, cell cycle re-entry was arrested but leaf edge cells still protruded, indicating that cell cycle progression is not required for reprogramming9 (Supplementary Fig. 9). To examine whether PpCSP1 regulates reprogramming regardless of cell cycle, we treated with aphidicolin the quadruple deletion mutant, PpCSP1 transcript-increased line, and wild type, and compared their reprogramming phenotype. In the presence of aphidicolin, the ppcsp quadruple deletion mutant and the PpCSP1 transcript-increased line exhibited attenuated and enhanced reprogramming, respectively as in the absence of the cell cycle inhibitor (Supplementary Fig. 9). These indicate that PpCSP1 functions in reprogramming independent of cell cycle progression.
On the basis of the results of this study, we propose a model for the function of PpCSP1 in the cellular reprogramming of Physcomitrella (Fig. 5e). PpCSP1 mRNA is weakly transcribed and degraded through regulatory elements localized to the 3′-UTR in all leaf cells (Fig. 3b–g). Subsequent to excision, a wound signal induces promoter activity, which results in an increase in transcript and protein levels (Fig. 2d–f). The increase of promoter activity is strong enough for reprogramming in edge cells but not in non-edge cells. Since some edge cells are not reprogrammed (Fig. 2a,d), another unidentified factor (X) must be necessary for uniform edge cell reprogramming. Furthermore, since some reprogramming still occurs in the ppcsp quadruple deletion line (Fig. 5c), another inductive pathway occurring independent of PpCSP1 must exist (Fig. 5e). In the nPpCSP1-Citrine-nosT and PpCSP1pro-PpCSP1-Citrine lines, without repression mediated by the 3′-UTR, PpCSP1 expression increases and triggers reprogramming in non-edge cells (Fig. 4a–g).
Shared domain structures and amino acid similarities between PpCSP1 and Lin28 (Fig. 1c,d) suggest that Lin28 is the most closely related protein of PpCSP1 in the metazoan genomes. Both PpCSP1 and Lin28 are dispensable for reprogramming and function to enhance the reprogramming. Lin28 is dispensable for iPSC formation and promotes the maturation of iPSCs12,13,52, although Lin28 participates in iPSC reprogramming from human fibroblast cells14. In the ppcsp quadruple deletion line of Physcomitrella, reprogramming was attenuated in edge cells but was not completely arrested (Fig. 5c,d). Non-edge cells were effectively reprogrammed in the PpCSP1 transcript-increased lines (Fig. 4b,f). However, the molecular mechanisms underlying PpCSP1 and Lin28 regulation appear to be different. Lin28 binds to precursors of miRNA let-7 and inhibits its processing31,32,34, while let-7 leads to the degradation of Lin28 transcripts31. Therefore, this negative feedback loop functions as a bistable switch to regulate cell fate31. We found that regulation of PpCSP1 transcripts is mediated by its 3′-UTR but we could not find miRNA binding sites in this region nor let-7 homologues in the Physcomitrella genome. Furthermore, the degradation of PpCSP1 transcripts is not specific to the differentiated cells (Fig. 3b–e). The activation of the PpCSP1 promoter in the reprogramming cells results in the increase of PpCSP1 transcripts (Fig. 5e).
Multicellularity with stem cells has evolved independently in land plant and metazoan lineages and the molecular mechanisms underlying reprogramming appear to differ between these lineages1,2,3,4. Nevertheless, this study showed that closely related genes encoding CSD proteins, PpCSP1 and Lin28, are involved in reprogramming, although their orthology was not clear (Fig. 1d). Therefore, it is an open question whether PpCSP1 and Lin28 have evolved from a common gene or different genes of the last common ancestor.
CSD is highly conserved in bacteria, land plants and metazoa19,20, but the biochemical function of CSD in reprogramming is unknown. In Escherichia coli, CSPs function as RNA chaperones that destabilize secondary structures in RNA53,54 and deletion of four CSP genes results in growth defect under low temperature53,54. Wheat cold-shock domain protein 1 (WCSP1) also has nucleic acid binding activity, anti-termination activity and dsDNA melting activity18. Ectopic expression of WCSP1 in an E. coli CSP deletion mutant could complement its cold-sensitive phenotype18, suggesting that the CSP function as RNA chaperone in response to cold stress is the ancestral function of CSP between bacteria and land plants. Arabidopsis CSPs (AtCSPs) also function in the stress response and during regular development17,55,56,57,58,59,60. However, no report has shown that CSPs function in stem cell establishment/maintenance or reprograming in flowering plants. GUS reporter analysis showed that AtCSPs are expressed in shoot and root meristem harbouring stem cells17,58,59,60. These suggest that AtCSPs may play a role in stem cell regulation in Arabidopsis. It will be a future challenge to investigate the biochemical functions of CSD within PpCSPs and AtCSPs in reprogramming.
PpCSP1-Citrine signals localized at the phragmoplast when the reprogrammed leaf cells divide (Supplementary Fig. 3 and Supplementary Movie 2). The signals were maintained in the reprogrammed chloronema apical stem cells and diminished in the successive cell divisions, although the diminished signals were maintained in chloronema apical stem cells (Supplementary Fig. 3 and Supplementary Movie 2). In addition, PpCSP1 was expressed in both stem cells and proliferating non-stem cells in gametophore apices (Supplementary Fig. 1e). These results suggest that PpCSP1 is involved in cell cycle regulation during or after reprogramming, as Lin28 promotes cell cycle regulators and coordinates proliferative growth26,27. However, increasing and decreasing PpCSP1 levels in nPpCSP1-Citrine-nosT and the quadruple deletion mutant lines, respectively did not change the duration of cell cycles in protonema apical stem cells (Supplementary Fig. 8). Moreover, aphidicolin blocks cell cycle re-entry, nevertheless cells facing the cut protruded without dividing, indicating that the reprogramming does not require cell cycle progression. In the presence of aphidicolin, the PpCSPs quadruple deletion mutant and PpCSP1 transcript-increased line exhibited attenuated and enhanced reprogramming, respectively (Supplementary Fig. 9). These results indicate that PpCSP1 plays a role in reprogramming. It is a future question whether PpCSP1 functions in cell cycle regulation during the reprogramming.
In human cells, overexpression of Lin28 with a set of pluripotency-associated transcription factors Oct4, Sox2 and Nanog enhances reprogramming of fibroblast cells into iPSCs14. In addition to let-7, Lin28 binds to various mRNAs including ∼50% of the human transcripts with motifs of GGAG or GGAG-like, although it is still unclear how its global mRNA-binding ability contributes to iPSC reprogramming61,62,63. Future studies are warranted to investigate both the PpCSP1 and Lin28 regulatory networks in order to find molecular mechanisms underlying the common positive reprogramming functions between PpCSP1 and Lin28.
The Gransden 2004 strain of P. patens51 was used as the wild-type strain and cultured on BCDAT medium under continuous white light at 25°C (ref. 24). The third or fourth leaves were excised from gametophores 3 weeks after inoculation and put into liquid BCDAT medium to induce the reprogramming9. Polyethylene glycol-mediated transformation24 was performed using 10 μg of linearized plasmid as below: protoplasts were prepared from 3-day-cultured protonemata which were incubated in 25 ml of 8% mannitol solution with 0.5 g Driserase (Kyowa Hakko Kogyo Co., Ltd) at 25 °C for 30 min. After filtrating the protonemata with 50-μm nylon mesh, the protoplasts were collected by centrifugation at 180g for 2 min at room temperature, and resuspended into 40 ml of 8% (w/v) mannitol. Centrifugation and washing steps were repeated twice. Washed protoplasts were suspended in MMM solution (8.3% mannitol, 0.1% MES-KOH (pH 5.6), and 15 mM MgCl2) at 1.6 × 106 cells ml−1. Then, 300 μl of the protoplast suspension and 300 μl of PEG/T solution (28.5% polyethylene glycol 6,000 in 7.2% mannitol, 0.1 M CaCl2 and 10 mM Tris-HCl (pH 8.0)) were added into 30 μl of linearized plasmids. The protoplasts were incubated at 45 °C for 5 min, and then at 20 °C for 10 min in water baths. The transformed protoplasts were diluted to 8 ml with protoplast liquid culture medium (5 mM Ca(NO3)2, 1 mM MgSO4, 45 μM FeSO4, 0.18 mM KH2PO4 (adjusted to pH 6.5 with KOH), the alternative TES, 50 mg l−1 ammonium tartrate, 6.6% mannitol and 0.5% glucose), poured into a 6-cm Petri dish, and kept under the dark condition at 25 °C overnight. The protoplasts were collected by centrifugation at 180g for 2 min at room temperature, and suspended in 8 ml of top layer protoplast regeneration medium (BCD medium supplemented with 5 mM ammonium tartrate, 10 mM CaCl2, 0.8% agar and 8% mannitol) preheated at 45 °C. The suspended protoplasts were poured into four 9-cm dishes that contained solidified bottom layer of protoplast regeneration medium (BCD medium supplemented with 5 mM ammonium tartrate, 10 mM CaCl2, 0.8% agar and 6% mannitol) which was covered with cellophane. After 3-day incubation under continuous light, the regenerating protoplasts were transferred to BCDAT medium containing antibiotics for selection for 2 weeks. Then, the plants were transferred to BCDATG medium, incubated for 1 week, and re-inoculated onto the selection medium again. Stable transformants were further analysed by PCR and DNA gel blot analyses.
Sequence data of PpCSPs can be found from Phytozome P. patens V3.3 (DOE-JGI, http://phytozome.jgi.doe.gov/) under the following accession numbers: PpCSP1 (Pp3c5_6070); PpCSP2 (Pp3c6_23240); PpCSP3 (Pp3c5_7920); and PpCSP4 (Pp3c5_7880).
Plasmid construction for expression analysis
Primers used for plasmid construction are provided in Supplementary Table 1. To insert the CDS64 in frame with the PpCSP1 CDS, a PpCSP1 genomic DNA fragment just before the stop codon and a fragment just after the stop codon, were amplified and cloned into pCTRN-NPTII 2 (AB697058); thereby generating nPpCSP1-Citrine-nosT line (Supplementary Fig. 1a). One microgram of circular Cre recombinase25 expression plasmid (AB542060), as extracted from the E. coli DH5α strain with Wizard Plus SV Minipreps DNA Purification System kit (Promega) without any restriction enzyme digestion, was introduced into the PpCSP1-Citrine line to excise the selection marker cassette and the nopaline synthase terminator flanked by two loxP sites to generate the nPpCSP1-Citrine-3′-UTR line. The regenerated lines were screened not to grow on a medium containing 20 mg l−1 G418 and candidate lines were further confirmed by PCR.
For the promoter reporter lines, a 2.2 kb fragment containing a gateway rfcA cassette (Invitrogen) and a terminator sequence of pea (Pisum sativum) rbcS3A gene was amplified by PCR from the plasmid pT1OG (LC126301) with the primer pair shown in Supplementary Table 1 and then transferred into the XbaI-HindIII cut pPIG1b-NGGII plasmid (AB537478), resulting in the plasmid pAK101. A luciferase-coding sequence was amplified from pGL4.10 (Promega) and inserted into the StuI site of pAK101, resulting in a gateway-luciferase binary vector pAK102. A 1.8 kb PpCSP1 promoter fragment was amplified and cloned into the pENTR/D-TOPO vector (Invitrogen). The PpCSP1pro:LUC plasmid was constructed by LR reaction between the entry plasmid and pAK102. This construct was introduced to nPpCSP1-Citrine-3′-UTR line to generate the PpCSP1pro:LUC nPpCSP1-Citrine-3′-UTR line (Supplementary Fig. 2c).
Plasmid constructions of PpCSP1 3′-UTR deletion series
Primers used for plasmid construction are provided in Supplementary Table 1. sGFP and mRFP were inserted into pTKM1 vector30 to generate pTKM1-sGFP and pTKM1-mRFP vectors. Different lengths of the PpCSP1 3′-UTR were amplified with wild type genomic DNA as a template and inserted just after the sGFP coding sequence at the ApaI site (Fig. 3h).
Transient expression using particle bombardment
Sixty mg gold particles (1.6 μm diameter) were coated with equal quantities of each pair of pTKM1-mRFP/pTKM1-sGFP plasmid DNA and bombarded by PDS-1000 (Bio-rad) under 94.5 KPa vacuum condition into 5-week-old gametophores. Digital images were obtained using an Olympus DP71 camera on a fluorescence microscope (SZX16, Olympus, Japan). Fluorescence intensity of specific leaf cells was quantified by ImageJ 1.48v.
Plasmid construction for EF1αpro:sGFP-3′-UTR line
Primers used for plasmid construction are given in Supplementary Table 1. Fragments of sGFP and sGFP-3′-UTR were amplified from pTKM1-sGFP-3′-UTR plasmid and cloned into pENTR/D-TOPO (Invitrogen) and subsequently inserted into the pT1OG vector (LC126301)40 (Supplementary Fig. 4a,b).
Plasmid construction for the PpCSP1pro:PpCSP1-Citrine line
Primers used for plasmid construction are provided in Supplementary Table 1. A fragment of 2.1 kb PpCSP1 promoter and PpCSP1-coding sequence was amplified from wild-type genomic DNA and inserted into pCTRN-NPTII with XhoI and BsrGI sites. The fragment containing the PpCSP1 promoter, PpCSP1-Citrine fusion gene and nptII expression cassette was subsequently digested by SmaI and inserted into the pPTA1 vector (LC122350) (which contains the targeting sequence to PTA1 locus40) to generate the PpCSP1pro:PpCSP1-Citrine line (Supplementary Fig. 5e).
Plasmid construction for the deletion of PpCSP genes
Primers used for plasmid construction are provided in Supplementary Table 1. To delete PpCSP1, PpCSP2, PpCSP3 and PpCSP4 in wild type Physcomitrella, genomic fragments containing the 5′- and 3′-flanking regions of each gene were inserted into the 5′-end and 3′-region of the nptII expression cassette of pTN182 (AB267706), of the hygromycin resistance cassette of pTN186 (AB542059), of the BSD expression cassette of p35S-loxP-BSD (AB537973) and of the Zeocin resistance cassette of p35S-loxP-Zeo (AB540628) plasmids, respectively. The generated constructs were digested by suitable restriction enzymes for gene targeting (Supplementary Fig. 7a–d).
To generate ppcsp quadruple deletion mutants, the PpCSP1-deletion construct was introduced into wild-type Physcomitrella to generate ppcsp1 lines. The PpCSP2-deletion construct, PpCSP3-deletion construct, and subsequently the PpCSP4-deletion construct were introduced into the ppcsp1 lines to generate the ppcsp1 ppcsp2 double-deletion mutants, ppcsp1, ppcsp2 and ppcsp3 triple-deletion mutants and ppcsp1, ppcsp2, ppcsp3 and ppcsp4 quadruple deletion mutants, respectively.
DNA gel blot analysis
DNA gel blot analysis9 was performed as below: ∼3 μg of genomic DNA was digested with appropriate restriction enzyme(s) (see Supplementary Figs 2, 4 and 7), run on 0.7% (w/v) SeaKemGTG agarose (BME, Rockland, ME, USA), and transferred to a Hybond N+ nylon membrane (GE Healthcare, Chicago, IL, USA). Probe labelling, hybridization and detection were performed using the AlkPhos direct labelling and detection system with CDP-Star (GE Healthcare) according to the supplier’s instructions. Primers used for probe amplification are provided in Supplementary Table 1.
BLASTP search against a data set consisting of the nr as of Jan, 2015, Klebsormidium data set from http://www.plantmorphogenesis.bio.titech.ac.jp/~algae_genome_project/klebsormidium/kf_download.htm Pinus taeda assembly 1.01 annotation v2 http://dendrome.ucdavis.edu/ftp/Genome_Data/genome/pinerefseq/Pita/v1.01/Pita_Annotation_v2/, and P. patens v1.6 data set https://www.cosmoss.org/physcome_project/linked_stuff/Annotation/V1.6/P.patens.V6_filtered_cosmoss_proteins.fas.gz, were performed using PpCSP1 through http://moss.nibb.ac.jp/cgi-bin/blast-nr-Kfl. According to BLASTP search, we noticed that Lin28 proteins are most similar to PpCSP1 in metazoan genomes. To see whether PpCSP1 and other plant CSPs are most similar to Lin28 in land plant genomes, C. elegans Lin28 was used as a query and PpCSPs together with other plant CSPs were found. Top 700 for PpCSP1 and 600 for Lin28 hit sequences were recovered and aligned using MAFFT68 with the einsi option through http://moss.nibb.ac.jp/cgi-bin/selectNalign and a preliminary tree was drawn with http://moss.nibb.ac.jp/cgi-bin/makenjtree. From both trees, sparse samplings of terminal taxa were performed to include human and mouse Lin28 homologues and Amborella trichopoda, Arabidopsis and Physcomitrella CSP homologues. These sequences were further aligned through http://moss.nibb.ac.jp/cgi-bin/selectNalign. Sites aligned ambiguously or having gaps were marked as excluded for further analysis using MacClade ver 4 (ref. 69). After removing proteins lacking conserved zinc-finger domains and choosing proteins with one cold-shock domain and two zinc-finger domains, the nexus file was submitted from http://moss.nibb.ac.jp/cgi-bin/makemltree1000. This selects an amino acid substitution model based on the data and performs maximum likelihood analysis using RAxML version 8.1.16. Bootstrap analysis was performed with 1,000 replicates prepared with SEQBOOT in PHYLIP and consensus was calculated with CONSENSE70.
RNA preparation and RT-qPCR analysis
Total RNA was purified from protonemata and cut leaves with the RNeasy Micro Kit (Qiagen). First-strand cDNA was synthesized using the ReverTra Ace qPCR RT Master Mix (TOYOBO). RT-qPCR was performed using an ABI PRISM 7500 (Applied Biosystems) with the QuantiTect SYBR Green PCR Kit (Qiagen). The cycle conditions were: 50 °C for 2 min and 95 °C for 10 min as pre-treatments, 95 °C for 15 s and 60 °C for 1 min at 40 cycles as amplification. After amplification cycles, we carried out dissociation analyses for confirmation of target validity. The sequences of primers for RT-qPCR are listed in Supplementary Table 1. Standard curves were estimated by dilution series (1, 0.1, 0.01, 0.001 and 0.0001) of one wild-type cDNA sample. Each transcript level determined by RT-qPCR analysis was normalized with TUA19.
Digital gene expression profiling with 5′-DGE analysis
Transcriptome analyses with 5′-DGE analysis10 were performed as below (DRA accession number DRR055536-DRR055559): From 5 to 10 μg of total RNA, poly(A)+ RNA was enriched with the FastTrack Kit (Thermo Fisher Scientific, Waltham, MA, USA). Then, first-strand cDNA was synthesized using biotin-labelled dT20 primers containing an EcoP15I site (Biotin-TEG-5′-CTATCAGCAGTTTTTTTTTTTTTTTTTTTT-3′) using PrimeScript II reverse transcriptase (Takara Bio, Shiga, Japan). DNA synthesis extended after the 5′-end of the mRNA, complementary to the biotin-labelled P2 DNA-RNA chimeric oligonucleotide containing an EcoP15I site and a GGG ribonucleotide sequence (Biotin-TEG-5′-CTGCCCCGGGTTCCTCATTCTCTCAGCArGrGrG-3′). The second-strand cDNA was synthesized based on the P2 sequence. After digestion with EcoP15I, the fragments were captured with streptavidin beads and ligated with P1 adaptors which were produced by annealing P1-A (5′-CCACTACGCCTCCGCTTTCCTCTCTATGGGCAGTCGGTGAT-3′) and P1-B-NN oligonucleotides (5′-N*N*ATCACCGACTGCCCATAGAGAGGAAAGCGGAGGCGTAGTGG-3′, where asterisks indicate phosphorothioate bonds). The resulting 25-bp 5′-cDNA fragments were amplified by 12 cycles of PCR using P1 (5-CCACTACGCCTCCGCTTTCCTCTCTAT-3′) and P2 primers (5′-CTGCCCCGGGTTCCTCATTCT-3′) and were then subject to 25-bp SOLiD single-read sequencing from the P1 sites. For the comparison between nPpCSP1-Citrine-nosT and nPpCSP1-Citrine-3′-UTR lines, special reference for each line with the targeted change on the scaffold_41 where the PpCSP1 locus is present was prepared and the sequence tags were mapped on the respective reference. Expression profiles of gametophore leaves 0, 1, 3 and 6 h after excision in nPpCSP1-Citrine-nosT line, nPpCSP1-Citrine-3′-UTR lines and wild type11 were analysed. Cumulative sum of tags of PpCSP1 transcript for all time points in these lines were calculated and shown in Fig. 3a.
Microscopy and image analyses
Live-imaging analysis was performed using a fluorescence microscope (IX81, Olympus) with a cooled CCD camera (ORCA-AG, Hamamatsu Photonics) or an EM-CCD camera (ImagEM, Hamamatsu Photonics). Protonemata were cultured on glass-bottom dishes with BCDAT medium for 5–7 days before time-lapse observation of protonema growth. For luciferase bioluminescence imaging, tissues were pre-cultured for 18 h in BCDAT medium, including 500 μM beetle luciferin potassium salt (Promega), before the observation. The third or fourth leaves were excised from gametophores on a plate 3 weeks after inoculation and placed on a 35-mm glass-based dish (IWAKI) covered with 2% methylcellulose. The leaves were covered with cellophane and then with 0.8% solid BCDAT medium. The petri dish was set on the stage of an IX81 microscope. Bright-field and Citrine-fluorescence images (using the × 10 objective lens) of excised leaves were taken at 20-min intervals for 72 h after excision. A U-MNIBA3 filter (Olympus) was used for Citrine. For the bioluminescence imaging with the time-lapse observation, images were taken at 2-h intervals for 72 h after the excision. A U-MGFPHQ filter (Olympus) was used for the detection. Between imaging, the stage was moved in continuous white light conditions under control of the MetaMorph software (Molecular Devices). The area and intensity of the Citrine, LUC or sGFP signal in each cell were calculated. The average intensity at each time point was calculated as the intensity of the GFP signal divided by the area of the cell. The movie of the time-lapse images was edited with ImageJ 1.48v. Images of PpCSP1-Citrine localization (Fig. 2c) were taken by an inverted microscope (IX81, Olympus) equipped with a spinning-disk unit (CSU21, Yokogawa) with a CMOS camera (ORCA-Flash 4.0, Hamamatsu Photonics). Bandpass filters (FF01-550/88-25, Semrock) for Citrine were used in the spinning-disk unit. Gametophore apex images (Supplementary Fig. 1e) were taken by a fluorescence microscope (BX51, Olympus) equipped with a colour camera (DS-Fi1c, Nikon). Citrine-fluorescence images were taken with U-MNIBA3 filter (Olympus). Aphidicolin treatment9 was performed as excised leaves were put into BCDAT liquid medium containing aphidicolin at the concentration denoted in Supplementary Fig. 9 or mock (DMSO). The leaves 72 or 120 h after excision were stained in a solution containing 0.1% aniline blue and 0.1% K3PO4 (pH 12.5) to visualize newly synthesized cell plates. Fluorescent images were taken by a fluorescence microscope (BX51, Olympus) equipped with a colour camera (DS-Fi1c, Nikon) and with long-pass filter (U-MWU2, Olympus; Supplementary Fig. 9). Protonema and gametophore images of EF1αpro:sGFP-nosT and EF1αpro:sGFP-3′-UTR lines (Fig. 3f,g) and bombardment experiment images (Fig. 3i) were taken by a fluorescence microscope (SZX16, Olympus) equipped with a colour camera (DP71, Olympus). sGFP fluorescence was taken by GFPHQ filter (Olympus). mRFP fluorescence was taken by an RFP1 filter (Olympus) for excitation and 593/40 filter (Semrock) for emission. Fluorescence linearity of the colour camera DP71 was examined with fluorescence beads. Images showing a fluorescence intensity that fitted within the linear range were chosen for quantitative analyses.
We declare that all data supporting the findings of this study are available within the manuscript and its Supplementary Files or are available from the corresponding authors upon request.
How to cite this article: Li, C. et al. A Lin28 homologue reprograms differentiated cells to stem cells in the moss Physcomitrella patens. Nat. Commun. 8, 14242 doi: 10.1038/ncomms14242 (2017).
Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank all members in the laboratory of M.H. and the ERATO Reprogramming Evolution Project for useful discussions and technical support; and Shih-Long Tu, Chunli Chen, Xiangdong Fu, Yan Sun and Changwei Shao for helpful discussions. We also thank the Functional Genomics Facility, Model Plant Research Facility, Spectrography and Bioimaging Facility, and the Data Integration and Analysis Facility of National Institute for Basic Biology, and Japan Advanced Plant Science Network for their technical support. This work was partially funded by MEXT KAKENHI grants to Y.T. and M.H., JSPS KAKENHI grants to M.H. and JST grant to M.H., Program for Advancing Strategic International Networks to Accelerate the Circulation of Talented Researchers to M.H. and P.N.B., the Center for the Promotion of Integrated Sciences of SOKENDAI, the Howard Hughes Medical Institute and the Gordon and Betty Moore Foundation (through Grant GBMF3405) and from the NIH (R01-GM043778) to P.N.B.
PpCSP1-Citrine protein accumulation in an excised leaf of nPpCSP1-Citrine-3'UTR #1 line. Selected frames (0, 24, and 48 hours) are shown in Fig. 2a.
PpCSP1-Citrine protein enriched during cytokinesis in an excised leaf of nPpCSP1-Citrine-3'UTR #1 line. Selected frames during cytokinesis are shown in Supplementary Fig. 3.
PpCSP1-Citrine protein accumulation in growing protonemata of nPpCSP1-Citrine-3'UTR #1 line.
PpCSP1 promoter activity and PpCSP1-Citrine protein accumulation in an excised leaf of PpCSP1pro:LUC nPpCSP1-Citrine-3'UTR #2 line. Selected frames (0, 24, and 48 hours) are shown in Fig. 2d.
PpCSP1-Citrine protein accumulation in an excised leaf of nPpCSP1-Citrine-nosT #136 line. Selected frames (0, 24, and 48 hours) are shown in Fig. 4c.
Protonema growth of the wild type, nPpCSP1-Citrine-nosT #136 line, and ppcsp1 ppcsp2 ppcsp3 ppcsp4 #29 line.