Article https://doi.org/10.1038/s41467-022-34529-6 pTINCR microprotein promotes epithelial differentiation and suppresses tumor growth through CDC42 SUMOylation and activation Received: 5 August 2022 Olga Boix1, Marion Martinez1, Santiago Vidal2, Marta Giménez-Alejandre1, Lluís Palenzuela1, Laura Lorenzo-Sanz 3, Laura Quevedo 4, Olivier Moscoso5, Accepted: 27 October 2022 Jorge Ruiz-Orera 6, Pilar Ximénez-Embún 7, Nikaoly Ciriaco5, Paolo Nuciforo 8, Camille Stephan-Otto Attolini 9, M. Mar Albà10,11, Javier Muñoz 7, Tian V. Tian12, Ignacio Varela 4, Ana Vivancos13, Check for updates Santiago Ramón y Cajal 5, Purificación Muñoz3, Carmen Rivas 2,14 & 1234567890():,; 1234567890():,; María Abad 1 The human transcriptome contains thousands of small open reading frames (sORFs) that encode microproteins whose functions remain largely unexplored. Here, we show that TINCR lncRNA encodes pTINCR, an evolutionary conserved ubiquitin-like protein (UBL) expressed in many epithelia and upregulated upon differentiation and under cellular stress. By gain- and loss-of-function studies, we demonstrate that pTINCR is a key inducer of epithelial differentiation in vitro and in vivo. Interestingly, low expression of TINCR associates with worse prognosis in several epithelial cancers, and pTINCR overexpression reduces malignancy in patient-derived xenografts. At the molecular level, pTINCR binds to SUMO through its SUMO interacting motif (SIM) and to CDC42, a Rho- GTPase critical for actin cytoskeleton remodeling and epithelial differentiation. Moreover, pTINCR increases CDC42 SUMOylation and promotes its activation, triggering a pro-differentiation cascade. Our findings suggest that the micro- proteome is a source of new regulators of cell identity relevant for cancer. In the last decade, improvements in genomic computational analyses, oxygen and glucose deprivation16, viral and bacterial infection17,18 and peptidomics and ribosome profiling have revealed that our genome even tumor initiation19. contains thousands of small open reading frames (sORFs) located in In epithelial tissues, cell polarity plays a pivotal role in the acqui- previously assumed non-coding regions, whose translation produces a sition and maintenance of the differentiated status and in the specia- myriad of bioactive proteins that have been largely overlooked1,2. lized functions of epithelial cells20. The establishment of epithelial cell These small proteins, shorter than 100 amino acids, are called micro- polarity relies on several coordinated events, including the reinforce- proteins, micropeptides or SEPs (from sORF-encoded peptides). To ment of cell-to-cell adhesion21,22 and the regulation of cytoskeleton date, only a subset of them have been functionally characterized, and organization and its associated downstream signaling pathways23,24. The they have been shown to play essential functions regulating a plethora small GTPase cell division control protein 42-homolog (CDC42) is of fundamental processes such as DNA repair, RNA splicing, calcium considered an essential orchestrator of epithelium formation25,26. Given signaling, cell metabolism and tissue regeneration3–11. Importantly, its role in the structure and remodeling of the actomyosin cytoskeleton, mounting evidence suggests that microproteins have a role in stress it has been widely connected with the establishment of epithelial cell response12–15. In fact, several studies have observed an increased polarity27–29 and with the maturation of epithelial cell contacts30,31, translation of sORFs in response to different cellular stresses such as among other important processes of epithelial morphogenesis. A full list of affiliations appears at the end of the paper. e-mail: mabad@vhio.net Nature Communications | (2022)13:6840 1 Article https://doi.org/10.1038/s41467-022-34529-6 Conversely, loss of epithelial identity by dedifferentiation is con- TINCR lncRNA according to the Human Protein Atlas project. sidered a fundamental initial step in epithelial tumorigenesis32,33, Endogenous pTINCR microprotein was detected in these cell lines associated with tissue disorganization, uncontrolled growth and re- localized mainly in the nucleus and at the cell-to-cell junctions acquisition of mesenchymal and stem cell features34,35. Epithelial tissue (Fig. 1k). Altogether, we have demonstrated that TINCR lncRNA homeostasis is constantly challenged by exposure to pro-mutational actually codes for pTINCR microprotein. injuries and, thus, robust protective mechanisms against malignant transformation are crucial. In this context, compelling evidence have pTINCR promotes epithelial differentiation in vitro and in vivo shown that after genotoxic or oncogenic stress, epithelial cells activate To characterize the cellular and molecular functions of pTINCR, we a terminal differentiation program that has been proposed as a safe- cloned pTINCR sORF tagged with an HA epitope in a doxycycline- guard mechanism against oncogenic alterations36–38. Altogether, cell inducible lentiviral vector. We generated two different constructs polarity and differentiation have been positioned as key tumor sup- placing the HA-tag either in the C-terminal (pTINCR-HA) or the pressive mechanisms and their disruption is considered a hallmark in N-terminal (HA-pTINCR) part of the microprotein. To minimize the epithelial cancers33,39. Identifying players that regulate and maintain possible effect of the tag on pTINCR, we introduced a flexible linker epithelial cell identity is necessary to understand development, tissue between the HA and the pTINCR sORF (Supplementary Fig. 2A and homeostasis and cancer biology. Importantly, the relevance of Supplementary Table 2). Moreover, to uncouple the function of TINCR microproteins in regulating cell identity has not been explored. lncRNA and pTINCR microprotein, we also developed a synthetic ORF TINCR was described as a lncRNA upregulated during keratinocyte (syORF) by mutating ~20% of pTINCR-HA nucleotide sequence, sig- differentiation, and it has been shown to promote epithelial differ- nificantly changing the secondary structure of the RNA while produ- entiation as an RNA molecule40–42. In this study, we report that TINCR cing the same protein (Supplementary Fig. 2A, B, and Supplementary encodes an evolutionary conserved ubiquitin-like protein (UBL) that Table 2). We have used the syORF throughout the manuscript to vali- we have named pTINCR, which regulates epithelial differentiation by date our main findings (see figure legends). Both C- and N- terminal enhancing the SUMOylation and activation of CDC42, and which acts constructs were detected upon transient expression in U2OS (Sup- as a tumor suppressor in epithelial cancers. plementary Fig. 2C, D) and displayed similar distributions, being detected in all cellular fractions but most prominently in the nucleus Results and at cell-to-cell junctions (Supplementary Fig. 2E, F), consistent with pTINCR is a conserved microprotein encoded by TINCR the endogenous localization of the protein (Fig. 1k). lncRNA and expressed in epithelial tissues The high expression of TINCR in skin prompted us to evaluate the To identify novel microproteins with a possible role in cell identity, we role of pTINCR firstly in this context. We used the human keratinocyte evaluated the coding potential of annotated lncRNAs known to be HaCaT cell line, which allows to recapitulate keratinocytes differ- regulated during differentiation and in cancer. We used PhyloCSF, a entiation by modulating the Ca2+ concentration in the culture medium. comparative genomics algorithm that analyzes codon substitution Consistent with a previous report40, we observed that TINCR mRNA frequencies across evolution to predict coding sequences43. In skin, expression was higher in differentiated (cultured under high Ca2+ our analyses pointed to TINCR as a good candidate, based on its concentration) compared to non-differentiated HaCaT cells (cultured enriched expression44 and its high PhyloCSF score (Fig. 1a and Sup- under basal/low Ca2+ conditions) (Supplementary Fig. 2G). Our results plementary Table 1). show that pTINCR overexpression during HaCaT differentiation led to TINCR is a 3.7-kilobase lncRNA upregulated during the differ- an increased upregulation of differentiation markers compared to the entiation of keratinocytes and other epithelial cells40–42. We corrobo- control cells (Fig. 2a, differentiation medium panels), indicating that rated that TINCR is expressed in human and mouse skin, and also in pTINCR improves keratinocyte differentiation. Remarkably, pTINCR other epithelia (Fig. 1b). Conservation analysis of the TINCR transcript overexpression induced an increase in the expression of some differ- showed a sORF of 264 bp (Fig. 1c). Ribosome profiling analysis in entiation markers even in low Ca2+ conditions (Fig. 2a, basal medium mouse skin (RibORF score ≥ 0.7) revealed that this sORF is translated panels), suggesting that pTINCR is sufficient to trigger differentiation into an 87 amino acid microprotein highly conserved across the tet- events. In agreement, we observed that pTINCR overexpression per se rapoda taxa, which we named pTINCR (Fig. 1c, d). To confirm that (without changing the Ca2+ concentration) switches actin cytoskeleton pTINCR sORF is translated into a stable microprotein, we performed organization from a stress fiber disposition to a cortical pattern in vitro translation using the full-length TINCR lncRNA (ENSEMBL: (Fig. 2b, second row), resembling the keratinocyte differentiation ENST00000448587.5) in the presence of 35S-methionine (Fig. 1e, f), process (Fig. 2b, third row)23,24,46,47. Moreover, pTINCR overexpression and obtained a peptide product of ~12 kDa. Of note, mutating the start induced the establishment of adherens and tight junctions similarly to codon of pTINCR impairs the translation of any detectable product high calcium conditions, as seen by the increase in E-cadherin and β- (Fig. 1e, f). In addition, by re-analyzing mass spectrometry data from catenin (markers of adherens junctions) and ZO-1 (marker of tight human organotypic skin cultures45, we identified three different tryptic junctions) at the cell membrane (Fig. 2c–e). peptides corresponding to pTINCR, providing strong experimental To further explore the role of pTINCR as a driver of epidermal evidence of pTINCR translation in skin (Fig. 1g and Supplementary differentiation, we performed an in vivo proof-of-concept experiment Fig. 1A). Furthermore, we performed Western blot and immunohisto- using a teratoma formation assay, which allows to study pluripotent chemical analyses using a custom polyclonal antibody against pTINCR stem cell differentiation in an unbiased manner. We inoculated mouse and observed that pTINCR protein is expressed in skin and also in Embryonic Stem Cells (ESCs) modified to express pTINCR or not in the several other stratified (tongue, palate, esophagus, bladder, cervix and flanks of immunodeficient mice and observed that pTINCR expression mammary gland) and simple (lung, stomach, uterus, sebaceous and led to reduced teratoma growth (Fig. 2f), consistent with more dif- sweat glands) epithelial tissues (Fig. 1h–j and Supplementary Fig. 1B). ferentiated teratomas. Furthermore, histopathological examination of Of note, apart from the expected band (Fig. 1h, indicated by the arrow), teratoma sections revealed a significant increase in skin differentiation pTINCR antibody detects by Western blot additional bands that may within pTINCR-overexpressing teratomas, supported by the increase correspond to post-translational modifications of pTINCR, in keratin deposition (Fig. 2g and Supplementary Fig. 2H). Immunos- although we cannot discard other uncharacterized pTINCR taining experiments in control teratomas (with no pTINCR over- isoforms. Finally, to determine pTINCR subcellular localization, we expression) showed endogenous pTINCR expression in skin. Some performed immunofluorescence experiments in HaCaT and MCF7 teratomas presented transitions from single to stratified epithelia and, cells, two epithelial cell lines documented to express high levels of remarkably, pTINCR expression notably increased in the stratified Nature Communications | (2022)13:6840 2 Article https://doi.org/10.1038/s41467-022-34529-6 Fig. 1 | pTINCR is an evolutionarily conserved microprotein highly expressed in of the constructs used for in vitro translation of pTINCR microprotein. f In vitro epithelia. a Screening of the coding potential of lncRNAs enriched in skin com- translation of WT and ATGmut full-length TINCR lncRNA visualized by SDS-PAGE. pared to other organs44. PhyloCSF score was obtained from LNCipedia v.5.2. b Heat Two experiments were performed and obtained similar results. g MS/MS spectrum map of TINCR lncRNA expression in human and mouse tissues. Expression in of a unique peptide derived from pTINCR microprotein in human skin organotypic human was obtained from GTEX. Expression in mouse was assessed by RT-qPCR (at cultures. h Western blot analysis of endogenous pTINCR in a panel of mouse least 3 mice per organ). Values represent the percentage of TINCR expression in epithelial tissues. Adrenal gland and spinal cord were added as negative controls. each organ normalized to the expression in the skin. c Diagram of TINCR locus. The experiment was repeated 4 times obtaining similar results. i Representative pTINCR sORF is indicated with a big black square. In red, PhyloCSF score across immunostainings of pTINCR in mouse and human skin and liver using a pTINCR TINCR. In green, Ribo-seq analysis of Tincr transcript in mouse skin. In blue, antibody. Liver was added as a negative control. Scale bars correspond to 100 µm. expression of Tincr in mouse skin by RNAseq. d pTINCR amino acid conservation j Representative immunostainings of a panel of mouse stratified epithelia using a across the tetrapoda taxa depicted by multiple sequence alignment performed by pTINCR antibody. Scale bars correspond to 100 µm. k Representative immuno- Clustal Omega. Amino acid colors indicate their properties (pink, positive charge; fluorescence images of HaCaT and MCF7 cells using a pTINCR antibody (green) blue, negative charge; red, hydrophobic; green, hydrophilic). The symbols below and Phalloidin-TRITC (red). Scale bars correspond to 50 µm. Immunostainings the alignment represent the biochemical similarity (asterisks indicate identical were performed at least twice obtaining the same results. Source data are provided conservation, colons high similarity and periods similar conservation). e Diagram as a Source Data file. areas compared to the non-stratified ones (Fig. 2h and Supplemen- squamous cell carcinoma (cSCC) cell line (hSCC1048,49), the A549 tary Fig. 2I). lung adenocarcinoma, and the MCF7 luminal breast cancer cell Next, we wanted to address the function of pTINCR in other line. In hSCC10 and A549 cells, pTINCR overexpression was asso- epithelial contexts. We overexpressed pTINCR in a set of cancer ciated with an upregulation of several genes related to epithelial cell lines of epithelial origin, including a patient-derived cutaneous differentiation and a downregulation of basal epithelial markers Nature Communications | (2022)13:6840 3 Article https://doi.org/10.1038/s41467-022-34529-6 Fig. 2 | pTINCR overexpression promotes epidermal differentiation. (syORF) in HaCaT cells. Control cells cultured for 24 h in differentiation medium a Expression of the indicated differentiation markers in HaCaT cells upon pTINCR- are shown as positive control. Percentage of cells showing membranous staining is HA (syORF) and HA-pTINCR overexpression. mRNA expression was measured indicated. The experiment was performed 3 times (c) or 2 times (d and e) inde- 4 days after inducing pTINCR with doxycycline and changing to differentiation pendently, representative images are shown. Scale bars correspond to 50 µm. medium (or maintained in basal medium). Statistical differences between pTINCR- f Effect of pTINCR overexpression on teratoma growth. Error bars represent the overexpressing cells and control cells (in basal or differentiation medium) and mean ± SEM. N = 18 teratomas in control group; N = 20 in pTINCR-HA group; N = 11 between basal and differentiated control cells (shown in brackets) are shown. Error in HA-pTINCR group. One-way ANOVA test corrected for multiple comparison was bars represent the mean ± SD of N = 3 technical replicates from a representative performed. g Quantification of keratin deposition in differentiated teratomas using experiment performed 3 times independently obtaining similar results. A 2-way QuPath. Dots represent the percentage of area with keratin deposition relative to ANOVA test corrected for multiple comparison was performed. b Immunostaining the total teratoma area. Error bars represent the mean ± SEM. N = 18 teratomas in images of F-actin 24 h after doxycycline induction of pTINCR-HA (syORF) in control group; N = 21 teratomas in pTINCR-HA group; N = 12 teratomas in HA- HaCaT cells cultured in basal medium. Control cells cultured for 24 h in differ- pTINCR group. Non-parametric Kruskal–Wallis test corrected for multiple com- entiation medium are shown as positive control. Percentage of cells showing parison was performed. h Representative H&E staining and IHC of endogenous cortical actin is indicated. The experiment was performed 3 times independently, pTINCR in a control teratoma. Pictures show single and stratified epithelia with representative images are shown. Scale bars correspond to 50 µm. keratin deposition. Scale bars correspond to 200 µm. Source data are provided as a c–e Immunostainings showing E-cadherin (c), β-catenin (d) and ZO-1 (e) after 24 h Source Data file. (E-cadherin) or 4 days (β-catenin and ZO-1) of doxycycline induction of pTINCR-HA (Fig. 3a). In agreement, pTINCR-overexpressing MCF7 cells upre- Altogether, these results strongly support the role of gulated luminal markers and downregulated basal cell genes pTINCR in promoting epithelial differentiation in vitro and (Fig. 3a). As previously observed in HaCaT cells, pTINCR expres- in vivo. sion triggers a cortical actin disposition (Fig. 3b–d) and reinforces cell-to-cell contacts in these models, as seen by an increased pTINCR is required for in vitro epithelial differentiation accumulation of β-catenin, ZO-1 and E-cadherin at the cell mem- To investigate the impact of pTINCR deficiency on cellular brane (Fig. 3b–d and Supplementary Fig. 2J, K). differentiation, we generated pTINCR-deficient cell lines by using Nature Communications | (2022)13:6840 4 Article https://doi.org/10.1038/s41467-022-34529-6 Fig. 3 | pTINCR overexpression promotes differentiation in several epithelial b–d Immunostainings of F-actin, β-catenin, ZO-1 and E-cadherin in the indicated cell types. a Expression of the indicated differentiation and basal markers upon cell lines after 24 h (actin and E-cadherin) or 4 days (β-catenin and ZO-1) of dox- the overexpression of pTINCR-HA and HA-pTINCR in hSCC10, A549 and MCF7 ycycline induction of pTINCR-HA (hSCC10 and A549) or pTINCR-HA syORF cells. mRNA expression was measured 21 days (A549 and hSCC10) or 4 days (MCF7) (MCF7). Nuclei are counterstained with DAPI. Percentage of cells showing mem- after doxycycline induction. mRNA expression is normalized to GAPDH and rela- branous staining is indicated. Results observed upon the overexpression of HA- tive to the control in each case. In MCF7, pTINCR-HA syORF was used. Error bars pTINCR are shown in Supplementary Fig. 2H, I. The experiment was performed represent the mean ± SD of N = 3 technical replicates from a representative twice independently, representative images are shown. Scale bars correspond to experiment performed at least 4 times independently obtaining similar results. 50 µm. Source data are provided as a Source Data file. A 2-way ANOVA test corrected for multiple comparison was performed. CRISPR-Cas9 technology in two of our epithelial models, HaCaT and that pTINCR protein, independently from TINCR lncRNA, is required to MCF7. pTINCR-KO cells contain a single-nucleotide insertion that dis- achieve full epithelial differentiation in vitro. rupts the ATG start codon and abolishes the translation of pTINCR (Fig. 4a and Supplementary Fig. 3A–D), but is predicted to have no pTINCR triggers an epithelial differentiation transcriptional effect on the secondary structure of TINCR transcript (Supplementary program Fig. 3E). We have validated pTINCR deficiency by Western blot (Fig. 4a) To corroborate the pro-differentiation function of pTINCR, we ana- and immunostaining (Supplementary Fig. 3D). lyzed the transcriptional profile induced by pTINCR overexpression. We subjected both HaCaT and MCF7 cells to high calcium, a well- We performed an extensive RNA-seq analysis in hSCC10 cell line upon known inducer of differentiation in the epidermis50 but also in other the overexpression of pTINCR to assess both early transcriptomic epithelial tissues such as esophagus51 or mammary glands52,53. Of note, changes (6, 12, and 24 h) and long-term changes (4, 7, 14, and 21 days), pTINCR protein expression increases upon differentiation in WT MCF7 the latter probably reflecting changes in cell identity. First, we studied cells (Fig. 4a). Importantly, we observed that TINCR lncRNA was transcriptional dynamics driven by pTINCR overexpression using upregulated upon calcium-induced differentiation to the same extent impulseDE algorithm, a framework for longitudinal sequencing in WT and pTINCR-KO cells (Fig. 4b), indicating that the regulation of experiments that reveals differential gene expression associated with the TINCR lncRNA is not affected in engineered pTINCR-KO cells. time54. We detected 6 different groups of genes that clustered together However, pTINCR-deficient cells fail to upregulate differentiation based on their dynamic expression (Fig. 5a, b and Supplementary markers to the same extent as WT cells (Fig. 4c, f). In addition, pTINCR- Fig. 4A). We performed gene ontology enrichment analysis on each KO cells did not acquire an epithelial morphology (Fig. 4d, g) nor cluster (Fig. 5c, d and Supplementary Fig. 4B). Interestingly, two of remodeled their actin cytoskeleton towards a cortical disposition them (cluster 1 and cluster 2) were dynamically opposite clusters, both under differentiation conditions (Fig. 4e, h). These results demonstrate enriched in gene ontology terms related to cytoskeleton. We used the Nature Communications | (2022)13:6840 5 Article https://doi.org/10.1038/s41467-022-34529-6 ClueGo tool to functionally organize gene ontology networks within Next, we ran DESeq2 analysis55 in samples collected at 0 h and at each cluster (Fig. 5e, f). We observed that cluster 1 is enriched in genes 21 days and looked for long-term gene expression changes in pTINCR- connected with cell morphogenesis, cell-to-cell junctions and adhe- overexpressing cells. A rotation-based functional enrichment analysis sion (Fig. 5e), and cluster 2 is formed by cytoskeleton organization- showed that gene-sets positively correlated with pTINCR expression related genes associated with cell division and chromosome segrega- were associated with actin organization and cell polarity, such as “actin tion (Fig. 5f). filament organization”, “establishment or maintenance of cell polarity” Nature Communications | (2022)13:6840 6 Article https://doi.org/10.1038/s41467-022-34529-6 Fig. 4 | pTINCR is a critical regulator of epithelial differentiation in vitro. calcium differentiation medium for 4 days. mRNA expression is normalized to a Western blot of endogenous pTINCR using a polyclonal pTINCR antibody in WT GAPDH expression. Error bars represent the mean ± SD of N = 3 technical replicates and pTINCR-KO HaCaT and MCF7 cells, stably cultured in basal or high calcium from a representative experiment performed twice (c) or 3 times (f) independently differentiation medium, as indicated. b TINCR lncRNA expression measured by RT- with similar results. n.s: not significant. 2-way ANOVA test corrected for multiple qPCR in WT and pTINCR-KO HaCaT and MCF7 cells, under basal or differentiation comparison was performed. d, g Representative phase contrast images of WT and conditions for 4 days. mRNA expression is normalized to GAPDH expression. Error pTINCR-KO HaCaT (d) and MCF7 (g) cells cultured in basal and differentiation bars represent the mean ± SD of N = 3 technical replicates from a representative conditions for 4 days. Scale bars correspond to 50 µm. e, h Representative immu- experiment performed twice independently obtaining similar results. 2-way nofluorescence images of F-actin in WT and pTINCR-KO HaCaT (e) and MCF7 (h) ANOVA test corrected for multiple comparison was performed. c, f Expression of cells cultured in basal medium or under differentiation conditions for 24 h. Total the indicated differentiation markers measured by RT-qPCR in WT and pTINCR-KO percentage of cells displaying cortical actin is indicated. Scale bars correspond to HaCaT (c) and MCF7 (f) cells cultured in basal medium or high 50 µm. Source data are provided as a Source Data file. or “cortical actin cytoskeleton” (Fig. 5g, h). In addition, other epithelia- FILAGGRIN and INVOLUCRIN, an effect that is partially restored when specific gene sets (“Epithelial cell differentiation, “Cell–cell junction”, reexpressing exogenous pTINCR (Fig. 6e and Supplementary Fig. 5G). “Epithelial cell-cell adhesion”, “Apical surface”, “Notch signaling”) or These results suggest that, upon cellular damage, pTINCR acts down- epidermal-related (“Epidermal cell differentiation”, “Skin develop- stream the activation of p53 to induce cell differentiation. ment”, “Keratinocyte differentiation”, “Cornification”) were also enri- ched after 21 days of pTINCR overexpression (Fig. 5g, h). On the other pTINCR has tumor suppressor activity in epithelial tumors hand, the genes sets that negatively correlates with pTINCR expression TINCR lncRNA has been previously documented to be downregulated were enriched in pathways associated with cell cycle, cell metabolism in human cSCC40, where inactivating mutations in TP53 are a well- and protein processing, among others (Fig. 5g and Supplementary known prevalent risk factor and an early event during the development Fig. 4C). Interestingly, we found that pTINCR-downregulated genes of cSCC (62, http://p53.free.fr). Our own analysis of transcriptomic data were also enriched in the signature of “Myc targets” (Supplementary corroborated that TINCR expression is significantly decreased in cSCC Fig. 4C), an oncogenic pathway in cSCC closely connected with the compared to healthy skin (Fig. 7a). Moreover, we screened a panel of differentiation grade of tumors and clinical prognosis of this human patient-derived-cSCC cell lines (previously characterized by malignancy44,56,57. We validated these results by RT-qPCR analysis their epithelial or mesenchymal features48,49) for TINCR expression by (Supplementary Fig. 4D, E). In conclusion, this transcriptomic study RT-qPCR, and confirmed its downregulation when compared to the strongly supports the pro-differentiating function of pTINCR revealed immortalized primary keratinocyte cell line HaCaT (Fig. 7b). Notably, by our functional studies. epithelial-like cancer cell lines expressed higher levels of TINCR com- pared to mesenchymal-like cancer cell lines. Immunohistochemical pTINCR is upregulated upon cellular stress in a p53 dependent analyses in a wide cohort of human cSCC tumors (N = 51) revealed that manner and it is required for damage-induced differentiation pTINCR expression is decreased in 50 out of 51 samples, where 40% of p53 is widely known for its role as a tumor suppressor, being mutated tumors have a > 50% reduction below normal levels (Fig. 7c, d, Sup- in more than 50% of human cancers. In addition, mounting evidence plementary Fig. 5H), supporting a tumor suppressive role of pTINCR in has shown that p53 is also involved in embryonic development and cell cSCC. The loss of pTINCR expression did not correlate with the differentiation58,59. Our previous results suggest that the role of mutational state of p53 in these tumors (Supplementary Fig. 5I, J), pTINCR in calcium-induced differentiation does not require functional possibly meaning that tumors benefit from losing pTINCR expression p53, given that HaCaT cells are p53 mutant. Given that in independently of p53 mutation status. Consistent with these results, epithelial tissues cellular damage activates a terminal differentiation pTINCR overexpression significantly decreased tumor growth in PDXs program36,37, we wondered whether the expression of pTINCR was generated with hSCC10 cells (Fig. 7e). Interestingly, pTINCR- related to p53 activation upon damage. First, we treated different overexpressing tumors presented a significantly higher deposition of cancer cell lines with several p53 inducers or stabilizers (doxorubicin, hyaline matrix (Fig. 7f), a degenerative phenomenon associated with actinomycin-D or nutlin-3a) and analyzed TINCR expression and regression, DNA damage, autophagy, apoptosis and cell death63. The pTINCR levels. We observed that only cells with functional p53 (A549 effect of pTINCR reducing tumor cell proliferation was also observed or HCT116) upregulated pTINCR at the mRNA and protein level, in hSCC10 in vitro and, importantly, in other epithelial cell types such whereas p53-KO cells (HCT116 p53KO) or with mutant p53 (hSCC10) as A549 and MCF7 (Supplementary Fig. 5K). Finally, we studied the did not show the same response (Fig. 6a, b, Supplementary Fig. 5A–D). potential tumor suppressor activity of pTINCR in other epithelial These results corroborated that pTINCR is upregulated upon stress in a cancers and observed that high expression of TINCR significantly p53 dependent manner. We analyzed published p53 ChIP-seq correlates with increased overall survival (OS) of patients with bladder experiments60,61 and we did not observe p53 binding to pTINCR carcinoma (Log rank test p = 0.00034, HR = 0.59), PDAC (Log rank test locus, either in control cells or in cells treated with DNA-damaging p = 0.0024, HR = 0.51), stomach adenocarcinoma (Log rank test agents, suggesting that the regulation of pTINCR by p53 is indirect p = 0.022, HR = 0.63), head and neck squamous cell carcinoma (Log (Supplementary Fig. 5E, F). Next, we studied the effect of pTINCR rank test p = 0.0022, HR = 0.66) and lung adenocarcinoma (Log rank deficiency in p53-dependent DNA-damage-induced differentiation test p = 0.021, HR = 0.7) (Fig. 7g). Altogether, these results highly using the pTINCR-KO MCF7 cell line. MCF7 cells are p53 WT (Supple- support the role of pTINCR as a tumor suppressor in cSCC and mentary Fig. 5D) and, as expected, TINCR transcript was upregulated potentially in other epithelial tumors. upon damage in both pTINCR-WT and pTINCR-KO cells (Fig. 6c). Consistent with previous results, this upregulation was abolished when pTINCR is an ubiquitin-like protein that interacts with SUMO MCF7 cells were transduced with the human papillomavirus (HPV) E6 and modulates SUMOylation oncoprotein, which induces p53 degradation (Fig. 6c and Supple- To understand the molecular mechanisms behind the function of the mentary Fig. 5D), and only WT MCF7 cells upregulated pTINCR at the microprotein, we did structural modeling using I-TASSER64, which protein level (Fig. 6d). Of note, we observed that pTINCR-KO MCF7 revealed that pTINCR is predicted to be a ubiquitin-like protein (Fig. 8a). cells do not respond to damage-induced differentiation the same way Ubiquitin-like proteins (UBLs) are a family of proteins with structural as WT cells and fail to upregulate differentiation markers such as similarity to ubiquitin, despite presenting low similarity in their amino Nature Communications | (2022)13:6840 7 Article https://doi.org/10.1038/s41467-022-34529-6 acid sequences65. Type I members of the UBL family contain a di-glycine complexes67. Therefore, given the absence of di-Gly motif, pTINCR (di-Gly) motif at the C-terminus for targeted conjugation to other could arise as a new Type II UBL (Supplementary Fig. 6A). Subsequent proteins, functioning as covalent modifiers in a similar manner as analysis of the pTINCR amino acid sequence using GPS-SUMO ubiquitin65. On the other hand, Type II members of this family lack the software68,69 highlighted two well-conserved and overlapping SUMO- di-Gly motif but bear one or more domains resembling ubiquitin interacting Motifs (SIM) in the C-terminal part of the microprotein structure66,67. Type II UBLs are not conjugated covalently to other (Fig. 8a). To test the possible interaction of pTINCR with SUMO, we molecules but can bind to their interactors and form protein generated a SIM mutant version of pTINCR-HA (pTINCR-SIMmut) in Nature Communications | (2022)13:6840 8 Article https://doi.org/10.1038/s41467-022-34529-6 Fig. 5 | pTINCR triggers transcriptional changes associated with actin cytos- represents a different GO term and its size is proportional to the Bonferroni- keleton, epithelial morphogenesis and cell cycle. RNA-seq analysis was per- corrected p-value (p ≤ 0.05 were considered significant). Significant GO terms are formed in hSCC10 cells upon the overexpression of pTINCR-HA syORF to assess grouped based on their similarity (κ value) and labeled in the same color. Con- both early transcriptomic changes (6, 12, and 24 h) and long-term changes (4, 7, 14, nected nodes shared common genes. Most significant GO terms in each group are and 21 days). a, b Transcriptional dynamics were analyzed using impulseDE, and labeled. g Heatmap of selected biological pathways affected by pTINCR over- genes were classified in clusters according to their similar expression dynamics. expression, showing gene enrichment values of each experimental group at 0 h and Graphs represent cluster 1 (a) and 2 (b). c, d GO terms enrichment analysis using 21 days of doxycycline induction. BarPlot shows normalized enrichment scores ClueGO software indicating the functional term enrichments of the genes in clus- (NES) for the corresponding categories using GSEA. h Gene Set Enrichment Analysis ters 1 (c) and 2 (d). Plots show the most significantly enriched GO terms. p ≤ 0.05, of pTINCR-induced genes versus the indicated gene signatures. Source data are using two-sided hypergeometric test with Bonferroni correction. e, f. GO enrich- provided as a Source Data file. ment analysis of genes from cluster 1 (c) and 2 (d) using ClueGO 2.5.1. Each node a Mock Doxorubicin ActinomycinD Nutlin-3a b HCT116 HCT116 A549 hSCC10 HCT116 A549 hSCC10 WT p53KO p53WT p53MUT 2.5 p53WT p53MUT Relative to GAPDH (x10000) 0.3 1.0 Mock Mock Mock 2.0 Relative to GAPDH (x10000) Relative to GAPDH (x10000) 0.8 1.5 0.2 0.6 Doxorubicin Doxorubicin Doxorubicin 1.0 0.4 0.1 0.5 0.2 pTINCR pTINCR pTINCR 0.0 0.0 0.0 WT p53 KO c DMSO Doxorubicin d Non-treated UV e DMSO Doxorubicin hTINCR FILAGGRIN INVOLUCRIN WT 14 30 PP 0.00012 0.00012 Relative RNA expression Relative RNA expression P 0.000001 aCat WT P 0.00013 P 0.00013 Relative RNA expression 12 25 MCF7 p53WT 10 20 E6 P 0.000001 8 PP 0.0067 0.0067 15 6 2.0 P 0.0074 P 0.0074 1.5 ns 4 pTINCR-KO 1.0 2 PP 0.0067 0.0067 d 0.5 0 0.0 WT E6 pTINCR KO pTINCR WT pTINCR pTINCR WT pTINCR pTINCR KO KO+OE KO KO+OE Fig. 6 | pTINCR is induced by damage in a p53 dependent manner and it is independently with similar results. n.s: not significant. Multiple T-TEST corrected required for damage-induced differentiation. a TINCR lncRNA expression ana- for multiple comparison was performed. d Endogenous pTINCR expression ana- lyzed by RT-qPCR after 24 h of treatment with doxorubicin, actinomycin D or lyzed by immunofluorescence after 24 h of exposure to UV light in WT, E6- nutlin-3a in the indicated cell lines. mRNA expression is normalized to GAPDH in transduced or pTINCR-KO MCF7 cells. Representative images are shown, the each case and relative to the control. Error bars represent the mean ± SD of N = 3 experiment was performed twice independently. Scale bars correspond to 50 µm. technical replicates from a representative experiment performed twice (hSCC10 e Expression of the indicated differentiation markers after 24 h of treatment with and HCT116) or 3 times (A549) independently with similar results. b Endogenous doxorubicin in WT, E6-transduced, pTINCR-KO or pTINCR-KO re-expressing pTINCR expression analyzed by immunofluorescence in the indicated cell lines pTINCR-HA (pTINCR KO + OE) MCF7 cells. mRNA expression is normalized to after 24 h of treatment with doxorubicin. Representative images are shown, the GAPDH and relative to the control in each case. Error bars represent the mean ± SD experiment was performed twice independently. Scale bars correspond to 50 µm. of N = 3 technical replicates from a representative experiment performed 3 times c TINCR lncRNA expression levels analyzed by RT-qPCR after 24 h of doxorubicin in independently with similar results. n.s: not significant. Multiple T-TEST corrected WT, E6-transduced or pTINCR-KO MCF7 cells. mRNA expression is normalized to for multiple comparison was performed. Source data are provided as a Source GAPDH in each case and relative to the control. Error bars represent the mean ± SD Data file. of N = 3 technical replicates from a representative experiment performed 3 times which most of the amino acids constituting the SIM domain were Accordingly, pTINCR-SIMmut does not reduce cell proliferation and replaced by alanines (VLLLV > AAAAV, Supplementary Table 2). GST- does not induce the remodeling of the actin cytoskeleton (Fig. 8d, e). To pull down assays confirmed that pTINCR binds to SUMO1 and SUMO2/3 understand the meaning of pTINCR-SUMO interaction we analyzed in a non-covalent manner and that the interaction is lost in the pTINCR- global protein SUMOylation in different cell lines by Western blot. We SIMmut (Fig. 8b). Interestingly, SIM-mutant pTINCR resulted in a less observed that pTINCR overexpression modifies SUMO1 and SUMO2/3 stable protein, as shown by cycloheximide treatment experiments, conjugation patterns in a cell-dependent manner, increasing the pointing to an important role of SUMO in stabilizing pTINCR (Fig. 8c). intensity of some SUMOylated bands and decreasing the intensity of Nature Communications | (2022)13:6840 9 Article https://doi.org/10.1038/s41467-022-34529-6 a b c d hTINCR hTINCR cSCC biopsy Skin pTINCRloss pTINCR loss 2000 5 Relative to GAPDH (x10000) 100 >90% Expression (TPM) 1500 4 50-90% p<0.001 80 10-50% 3 <10% 0 % of tumors 1000 10X 60 2 Tumor area 500 1 40 0 20 0 aT 16 24 10 11 17 19 N=51 C C C C C C aC Normal cSCC C C C C C C H hS hS hS hS hS hS 0 Epithelial-like Mesenchymal-like 10X cSCC cell lines e f cSCC PDX growth cSCC PDXs 1200 Control pTINCR-HA Hyaline positive Hyaline negative Control pTINCR-HA HA-pTINCR 1000 HA-pTINCR Tumor volume (mm3) 800 P 0.0575 600 P 0.016 P 0.0048 400 P 0.025 200 0 0 5 10 15 20 25 Days g Bladder Pancreatic Ductal Stomach Head and Neck Squamous Lung Carcinoma Adenocarcinoma Adenocarcinoma Cell Carcinoma Adenocarcinoma HR=0.59 (0.44-0.79) HR=0.51 (0.33-0.8) HR=0.63 (0.42-0.94) HR=0.66 (0.51-0.86) HR=0.7 (0.52-0.95) logrank P=0.00034 logrank P=0.0024 logrank P=0.022 logrank P=0.0022 logrank P=0.021 Probability Probability Probability Probability Probability Expression Expression Expression Expression Expression Low Low Low Low Low High High High High High Time (months) Time (months) Time (months) Time (months) Time (months) Fig. 7 | pTINCR acts as a tumor suppressor in several epithelial tumors. percentage of pTINCR loss in the cSCC patient cohort (N = 51). e Effect of pTINCR a Expression of TINCR lncRNA in cSCC tumors compared to healthy skin (data expression on cSCC in vivo tumor growth. Tumor growth was monitored by subtracted from TCGA data). Statistical significance was calculated using two-sided measuring the volume at the indicated time points. Error bars represent the T-TEST. b TINCR lncRNA expression in a panel of patient-derived cSCC cell lines mean ± SEM. N = 16 tumors in control group; N = 15 tumors in pTINCR-HA group; and in HaCaT keratinocytes. Cancer cell lines are grouped according to their epi- N = 16 tumors in HA-pTINCR group. Two-sided T-TEST was performed. f Repre- thelial/mesenchymal traits. mRNA expression is normalized to GAPDH in each sentative images of H&E stainings showing areas with hyaline deposition in control case. Error bars represent the mean ± SD of N = 3 technical replicates from a and pTINCR-overexpressing cSCC PDXs. Graph shows the quantification of the representative experiment performed twice independently with similar results. number of tumors with hyaline deposits. Two-sided Fisher exact test was per- c pTINCR expression was analyzed by IHC in a cSCC patient cohort (N = 51). formed. Scale bars correspond to 200 µm. g Kaplan–Meier curves showing the Representative images show the loss of pTINCR expression from healthy epidermis correlation between TINCR expression and patient survival in the indicated tumor to cSCC (left) and a magnification of healthy skin and tumor area is shown (right). types. Source data are provided as a Source Data file. Scale bars correspond to 500 µm (left) or 200 µm (right). d Quantification of the others, an effect that seemed to be partially reverted in pTINCR-SIMmut in CDC42 (Saint probability=1; FC-B > 6, Confidence Score=35) cells (Fig. 8f, g and Supplementary Fig. 6B and C). Altogether, our data (Fig. 9a). CDC42 is a small GTPase of the Rho family, whose main indicate that pTINCR is a UBL whose interaction with SUMO is essential function is to regulate the organization of the actin cytoskeleton27. Rho for its stability and its function. GTPases reversibly transit between an active (GTP-bound) and an inactive (GDP-bound) state, acting as molecular switches to regulate pTINCR binds CDC42, enhances its SUMOylation and promotes many essential cellular processes such as cell morphology, migration, its activation invasion, cell cycle progression, endocytosis and gene transcription70. Given that pTINCR binds to SUMO and, therefore, potentially interacts Moreover, CDC42 has been shown to be essential for epithelial mor- with many other SUMOylated proteins, we wanted to uncover its phogenesis by regulating cell-to-cell contacts formation and the interactome. We performed immunoprecipitation of pTINCR-HA fol- establishment of apico-basal structures28,71. Interestingly, our RNAseq lowed by mass-spectrometry analysis and detected 35 potential analysis pointed to a statistically significant positive correlation interactors with a Saint probability >0.7 and a Confidence score >25 between pTINCR expression and the gene sets “GTPase binding” and (Supplementary Table 3). Among them, we were particularly interested “GTPase activator activity” (Fig. 9b). Thus, we wanted to explore Nature Communications | (2022)13:6840 10 Article https://doi.org/10.1038/s41467-022-34529-6 Fig. 8 | pTINCR is a UBL that interacts with SUMO and modulates SUMOylation. time point relative to the starting number of cells (day 0). Error bars represent the a Upper panel, image depicting the predicted structure of pTINCR, UBIQUITIN and mean ± SD of N = 3 technical replicates from a representative experiment per- SUMO using PyMOL. Color code indicates: α-helix (dark blue); β-sheet (green); formed 3 times independently obtaining similar results. n.s: not significant. Two- SIMs (purple). Lower panel, scheme of the predicted pTINCR domains. b Analysis sided T-TEST was performed. e Effect of pTINCR-HA and pTINCR-SIMmut over- of pTINCR and pTINCR-SIMmut non-covalent interaction with SUMO1 and SUMO2 expression on actin cytoeskeleton in A549 cells. Panels show representative by GST pull down assay. The experiment was performed 3 times independently immunofluorescence images of F-actin in control, pTINCR and pTINCR-SIMmut obtaining the same result, a representative experiment is shown. c Analysis of overexpressing cells induced with doxycycline for 4 days. Scale bars correspond to pTINCR-HA and pTINCR-SIMmut stability in U2OS. Cells were treated with CHX for 50 µm. f, g. Effect of pTINCR overexpression on SUMO1 (f) and SUMO2/3 (g) levels the indicated time points and analyzed by Western blot. The experiment was and conjugation analyzed by Western blot. In HaCaT cells, pTINCR-HA syORF was performed 3 times independently obtaining the same result. A representative used. The experiment was performed 8 times in (f) and 3 times in (g) indepen- experiment is shown. d Growth curve of A549 overexpressing pTINCR-HA, dently. A representative experiment is shown. Source data are provided as a Source pTINCR-SIMmut and control vector. Growth rate represents cell number at each Data file. whether pTINCR pro-differentiation phenotype was related to CDC42. SUMOylation did decrease the levels of GTP-CDC42 in control cells and First, we validated the interaction between pTINCR and CDC42 by partially reverted pTINCR-induced CDC42 activation (Fig. 9h). These immunoprecipitation (Fig. 9c). Of note, we were not able to co- results are consistent with a previous report showing that the precipitate CDC42 with pTINCR-SIMmut. This could suggest that the SUMOylation of Rac1, another Rho GTPase, does not modify its loca- binding of pTINCR to SUMO facilitates its interaction with CDC42, lization but is required for optimal GTP loading73. Altogether, these although we cannot discard the instability of pTINCR-SIMmut as an evidences suggest that pTINCR promotes CDC42 activation by alternative explanation. Next, we assessed the effect of pTINCR on the enhancing its SUMOylation. activation status of CDC42 by pull-down experiments and observed that pTINCR, but not pTINCR-SIMmut, increases GTP-CDC42 levels pTINCR-induced CDC42 activation does not promote (Fig. 9d). UBL proteins are characterized by their common, evolutio- pro-oncogenic events narily conserved tertiary structure, generally referred to as the β-grasp The epidermal growth factor (EGF) is a well-known CDC42 activator fold. This allows them to behave as adaptors for protein-protein that is associated with the acquisition of migration and invasion interaction to ensure enzymatic activity72. Thereby, we hypothesized capacities74. Importantly, although pTINCR induced CDC42 activation that the interaction of pTINCR with SUMO and its ability to activate similarly to EGF cell treatment (Supplementary Fig. 7H), the outcome CDC42 could be connected. First, we observed that CDC42 is of the activation was different. While pTINCR triggered an actin SUMOylated and that this modification is enhanced when we over- cytoskeleton remodeling towards a cortical disposition (Figs. 2b, expressed pTINCR (Fig. 9e, f). Worthy to mention, overexpressing 3b–d), EGF led to the formation of lamellipodia and cell protrusions pTINCR did not alter the SUMOylation status of other SUMO targets (Supplementary Fig. 7A). In agreement, pTINCR overexpression had no such as p53 and B23 (Fig. 9g), suggesting again that pTINCR effect on relevant effect on the transcriptional regulation of common epithelial- SUMOylation is not a global event but rather protein-specific. To our to-mesenchymal genes (Supplementary Fig. 7B, C), nor on cell migra- knowledge, the SUMOylation of CDC42 has not been described before. tion or invasion (Supplementary Fig. 7D–G). As seen by experiments using the SUMOylation inhibitor ML792 (Supplementary Fig. 6D, E), CDC42 SUMOylation did not significantly pTINCR promotes epithelial differentiation through the modify CDC42 localization (Supplementary Fig. 6F) or protein stabi- activation of CDC42 lity, given that we only see effects beyond 24 h of cicloheximide Next, we wanted to assess whether pTINCR pro-differentiation phe- treatment (Supplementary Fig. 6G). However, the inhibition of notype was mediated by its ability to activate CDC42. We analyzed the Nature Communications | (2022)13:6840 11 Article https://doi.org/10.1038/s41467-022-34529-6 activation of CDC42 during calcium-induced differentiation in WT, KO + OE), demonstrating that it is pTINCR microprotein and not the pTINCR-KO and pTINCR-KO HaCaT cells transduced with pTINCR lncRNA that is responsible for CDC42 activation during differentiation. syORF (pTINCR-KO + OE). As observed in Fig. 9i, CDC42 is activated in To further confirm that pTINCR promotes differentiation by inducing WT cells during calcium-induced differentiation. This result agrees the activation of CDC42, we used two chemical inhibitors of CDC42 with previous reports showing that CDC42 activity is critical for pro- activation, CASIN and ML141, which block GTP loading onto CDC42 cesses associated with epithelial morphogenesis, such as cell junction (Supplementary Fig. 7H). Of relevance, treatment with CASIN or ML141 maturation25,30. By contrast, pTINCR-KO cells failed to activate CDC42 inhibits both calcium- and pTINCR-induced differentiation, as seen by and to differentiate (measured by the expression of INVOLUCRIN) an impaired upregulation of differentiation markers (Fig. 9j and Sup- upon exposure to differentiation conditions, despite having a func- plementary Fig. 7I). Moreover, CDC42 inhibition also impaired the tional TINCR lncRNA (Fig. 4b). Remarkably, this effect was partially remodeling of actin cytoskeleton and the increase in cell-to-cell junc- reverted upon pTINCR re-expression in pTINCR-KO cells (pTINCR- tions formation induced by pTINCR (Fig. 9k) as well as by calcium- Nature Communications | (2022)13:6840 12 Article https://doi.org/10.1038/s41467-022-34529-6 Fig. 9 | pTINCR promotes differentiation through the SUMOylation and acti- KO + OE) HaCaT cells during calcium-induced differentiation. INVOLUCRIN is vation of CDC42. a Analysis of the pTINCR interactome using CRAPome algorithm. shown as a marker of cell differentiation. j Expression of the indicated differ- b GSEA analysis of pTINCR-induced genes in hSCC10 cells versus the indicated gene entiation markers after 24 h of calcium-induced-differentiation or pTINCR-HA signatures. c. Interaction of pTINCR with CDC42 analyzed by co- syORF overexpression in HaCaT cells, treated or not with the CDC42 inhibitors immunoprecipitation assay. d CDC42 activation assay in hSCC10 cells after 4 days CASIN or ML-141. Error bars represent the mean ± SD of N = 3 technical replicates of pTINCR expression. pTINCR-HA syORF was used. e Representative CDC42 from a representative experiment performed 3 times independently with similar SUMOylation assay in U2OS cells transduced with pTINCR-HA, Flag-CDC42 and results. We show statistical differences within each experimental condition and His6-SUMO2. SUMOylated CDC42 is indicated with arrowheads. f Quantification of between basal control cells and differentiated control cells or pTINCR- 4 independent CDC42 SUMOylation assays performed in U2OS cells. Values of overexpressing cells (shown in brackets). n.s: not significant. 2-way ANOVA test CDC42 SUMOylation are normalized to total CDC42 and relative to the control. with multiple comparison was performed. k–l Immunostainings of F-actin, Multiple T-TEST corrected for multiple comparison was performed. E-cadherin and β-catenin upon pTINCR- (k) or calcium-(l) induced differentiation g Representative SUMOylation assay of endogenous p53 and B23 proteins in U2OS for 24 h (actin and E-cadherin) or 4 days (β-catenin), in basal HaCaT cells treated or cells transfected with His6-SUMO2 and pTINCR-HA. SUMOylation of p53 and B23 is not with CASIN (actin and E-cadherin) or ML-141 (β-catenin). pTINCR-HA syORF was indicated with arrowheads. h CDC42 activation assay in HaCaT cells after 16 h of used. Nuclei are counterstained with DAPI. Total percentage of cells showing ML-792 treatment and/or 24 h of pTINCR syORF overexpression. i CDC42 activation membranous staining is indicated. Scale bars correspond to 50 µm. Source data are assay in WT, pTINCR-KO or pTINCR-KO overexpressing pTINCR-HA syORF (pTINCR provided as a Source Data file. induced differentiation (Fig. 9l). These results demonstrate that shown that pTINCR is associated with keratinocyte proliferation78. pTINCR microprotein promotes epithelial differentiation, at least in However, the molecular mechanisms behind pTINCR functions were part, by triggering CDC42 activation (Fig. 10). not thoroughly explored. Here, we have combined several experi- To better define the molecular mechanism behind the pro- mental approaches, including detection by specific antibodies, to differentiation function of pTINCR, we performed an antibody-based undoubtedly demonstrate the presence of pTINCR in skin but also in array to analyze the phosphorylation status of 141 proteins associated other epithelial tissues. In addition, we have provided evidence of its with actin dynamics upon pTINCR overexpression (Supplementary role as a tumor suppressor through the maintenance of epithelial Fig. 7J and Supplementary Table 4). Among the proteins with increased identity. Moreover, we have performed an in-depth mechanistic ana- phosphorylation levels, we found an enrichment in members of the lysis demonstrating that pTINCR is a UBL protein that induces epi- PKC, PLC and the phosphatidylinositol lipid (PtdIns) families, such as thelial differentiation, at least in part by enhancing the SUMOylation phospho-Ser307 PIP5K or phospho-Thr638 PKC α/β II (Supplementary and the subsequent activation of CDC42. Fig. 7K). This group of proteins and their effectors have been widely We have shown that pTINCR triggers epithelial differentiation in related to epithelial morphogenesis and CDC42 protein HaCaT keratinocytes and in a set of cancer cell lines of epithelial origin, activation25,75,76. In contrast, several members of the MAPK family, as seen by the upregulation of epithelial differentiation markers and whose activation is usually connected with pro-oncogenic and pro- the reorganization of actin cytoskeleton towards a cortical actin net- proliferative outcomes, displayed decreased phosphorylation levels work. In addition, pTINCR reinforces cell-to-cell junctions and (Supplementary Fig. 7L). It is also worthy to mention that with this decreases proliferation. Furthermore, our results in teratoma forma- approach we detected a striking increase in F-ACTIN levels in pTINCR- tion assays evidenced the pro-differentiation role of pTINCR also overexpressing cells (Supplementary Fig. 7M). Collectively, these data in vivo. Lastly, an extensive transcriptomic analysis by RNA-seq sup- demonstrate that pTINCR induces epithelial differentiation by reg- ported these findings by showing a positive correlation between ulating actin cytoskeleton dynamics through the activation of CDC42 pTINCR expression and gene signatures associated with actin organi- protein and its effectors. zation, cell-to-cell junctions, cell polarity, and epithelial differentiation. In this regard, it is worth noticing that in addition to its localization at Discussion cell-to-cell contacts and in the cytoplasm, pTINCR is also found in the The microproteome has been largely overlooked until recently and, nucleus, leaving open the possibility that it may have a direct role in although some of its functions are beginning to be uncovered, much transcriptional regulation. more research is needed to ascertain the extent of its biological rele- By in silico analysis of its structure we observed that pTINCR vance. TINCR was first described as a lncRNA and it has been shown to belongs to the ubiquitin-like protein (UBLs) family, a family of proteins be involved in epithelial differentiation40–42. Here, we demonstrate that with structural similarity to ubiquitin and most of them with protein- TINCR encodes a microprotein, pTINCR, with its own role also in epi- conjugation capability72. Interestingly, pTINCR is a Type II UBL, which thelial differentiation. This opens the possibility that some of the lacks the C-terminal GG residues that Type I UBLs typically used for functions ascribed to the lncRNA could be mediated by pTINCR. To their conjugation, as in the case of UFM, Atg12, Atg8, and UBL5. Of support that our observations are due to pTINCR microprotein and not note, pTINCR has two overlapping SIMs and we have demonstrated to TINCR lncRNA, we performed gain-of-function assays using only the that pTINCR interacts non-covalently with SUMO1 and SUMO2/3. Since pTINCR sORF and not the full-length TINCR lncRNA. Also, our key the two SIM are overlapping they cannot work as two individual SIM observations have been confirmed using a synthetic ORF (syORF) that functional domains and, consequently, pTINCR can only bind to one differs by ~20% in its nucleotide sequence from the wild-type pTINCR SUMO moiety at a given time. Importantly, we have shown that sORF (wtORF) and changes its secondary structure. On the other hand, pTINCR-SUMO interaction is necessary for pTINCR stability. Further- our pTINCR-KO cells were engineered by introducing a single more, pTINCR-SIMmut does not reduce cell proliferation and does not nucleotide change in the start codon of pTINCR ORF. This change is induce remodeling of the actin cytoskeleton. Altogether, these results predicted to have a minor effect on the folding of the TINCR lncRNA suggest that pTINCR function may depend on its non-covalent inter- and did not affect its expression or transcriptional regulation. Fur- action with SUMO. In line with these results, previous evidences thermore, some of the phenotypes observed in pTINCR-KO cells were associate changes in cellular SUMOylation with embryonic develop- rescued with the overexpression of exogenous pTINCR. Altogether, we ment and cell fate stabilization79,80, keratinocyte differentiation81–83, are confident that our observations are mediated by the microprotein and tumor suppression84–86. and not by the lncRNA. In agreement with our results and parallel to The ability to bind SUMO allows pTINCR to potentially interact our work, peptides derived from pTINCR have been detected by pro- with SUMOylated proteins. In this study, we uncovered an important teomics in the human stratum corneum77, and another study has pTINCR interactor, the small GTPase of the RHO family CDC42. Our Nature Communications | (2022)13:6840 13 Article https://doi.org/10.1038/s41467-022-34529-6 differentiation damage/ Homeostasis Cancer stimuli stress p53 TINCR pTINCR Downregulated Expressed in in epithelial epithelial tissues SIM tumors pTINCR SUMO SIM CDC42-GDP pTINCR SUMO SIM CDC42-GTP Epithelial Tumor differentiation suppression Fig. 10 | Proposed working model for pTINCR function. pTINCR is an UBL domain and modulates cell SUMOylation patterns. In particular, pTINCR increases microprotein expressed in epithelial tissues and downregulated in cSCC. pTINCR is CDC42 SUMOylation and promotes its activation, leading to epithelial differ- upregulated by pro-differentiation cues -such as calcium- and also by cellular entiation and tumor suppression. damage in a p53 dependent manner. pTINCR interacts with SUMO through its SIM data show that pTINCR binds to CDC42, increases its SUMOylation and results show that pTINCR-induced CDC42 activation does not lead to promotes its activation. Interestingly, mutating the pTINCR SIM- lamellipodia or filopodia formation, or enhancement of mesenchymal domain or treating with a SUMOylation inhibitor impair the ability of traits. Instead, it led to cytoskeleton remodeling towards a cortical pTINCR to activate CDC42, suggesting that CDC42 SUMOylation is deposition of actin resembling a more differentiated state. Moreover, required for its activation. These observations are consistent with we have shown that treatment with CDC42 inhibitors (CASIN and other small GTPases, such as RAC, that are also activated by ML141) impair epithelial differentiation mediated by pTINCR and by SUMOylation73. We propose a molecular model in which pTINCR acts calcium. Several publications describe the importance of CDC42 for as a scaffold to enhance the SUMOylation of CDC42, which in turn tissue differentiation, supporting its tumor suppressor role in certain promotes its activation. We have shown that the effect of pTINCR on contexts. During epithelial differentiation, CDC42 is involved in the SUMOylation is not a global effect but protein specific. We hypothesize establishment of an apico-basal cell polarity through the formation of that pTINCR could be promoting the SUMOylation of its interactors, tight junctions and the control of asymmetrical cell divisions28,71. In such as CDC42, acting as a specificity factor that bridges UBC9 (or fact, CDC42 knock-out mice die soon after birth with severe defects in other complexes with SUMO-ligase activity) to their targets. This the formation of the epidermal barrier, indicating the essential role of hypothesis is in line with other proteins known to regulate SUMOyla- CDC42 in epidermal differentiation88. Integrative genomic analyses tion, such as the tumor suppressor p14ARF, which induces SUMOyla- have shown that elevated RHO and RAC signalling, but not tion of a variety of p14ARF-interacting proteins85. CDC42 signalling, is a common feature in SCCs89. In addition, high CDC42 is a master regulator of cell polarity and actin dynamics27. levels of VAV2 (a guanine nucleotide exchange factor of the Rho In cancer, CDC42 activation has been commonly associated with family) in both cutaneous and head and neck SCC favors tumorigen- oncogenic traits such as increased cell migration and invasion87. Our esis through activation of RAC and RHO proteins, but not through Nature Communications | (2022)13:6840 14 Article https://doi.org/10.1038/s41467-022-34529-6 CDC4290. Altogether, we conclude that the pro-differentiation role of (Invitrogen), 15% FBS, 50 mM β-mercaptoethanol, 100X non-essential pTINCR is mediated, at least in part, by the SUMO-dependent activa- amino acids (Invitrogen), 1% P/S and 1000 U/ml LIF (ESGRO, Chemi- tion of CDC42. In agreement, pTINCR induces a substantial increase in con). Low calcium (basal) or high calcium (differentiation) media the phosphorylation status of several cytoskeleton-related proteins were prepared by supplementing calcium-free DMEM (Gibco, that have been widely associated with epithelial differentiation and #21068028) with 10% calcium-depleted FBS, 4mM L-glutamine stratification, such as Merlin23,91,92, the p85-subunit α/γ of PI3K93–96 or (Invitrogen), 1% P/S, and calcium chloride (Sigma) to a final con- several members of the Par complex97,98. These last two families of centration of 0.03 mM (low calcium, basal media) or 2.8 mM (high proteins are downstream effectors of CDC42 and are pivotal for the calcium, differentiation media). FBS was calcium-depleted by incu- establishment of the apico-basal polarity of epithelial tissues, as well as bation with BT Chelex 100 resin (BIO-RAD #143-2832) according to for the control of asymmetric cell division, spindle orientation and the manufacturer’s instructions. HaCaT and MCF7 cells were main- stem cell differentiation25,76. On the other side, we observed a sig- tained in basal media. Dedifferentiation was induced by culturing nificant decrease in the phosphorylation status of many proteins of the cells in basal medium for at least three weeks. Confluence was MAPK/ERK pathway, reported to be inhibited during keratinocyte maintained lower than 70% to ensure basal conditions. Differentia- differentiation99. The effect of many of these specific phosphorylation tion was induced by culturing with differentiation medium for changes has not been reported, but our results encourage to further the specified time (see figure legends). All cells were incubated at characterize them as potential drivers of epithelial differentiation. 37 °C, 5% CO2 in a humidified incubator and routinely tested for TINCR lncRNA has been documented to be deregulated in various mycoplasma contamination by PCR and confirmed negative. tumor types, and its role as an oncogene or as a tumor-suppressor Doxycycline (Sigma Aldrich) was added at 1μg/ml, when indi- seems to be tumor specific100,101. In esophageal SCC cells, siRNA- cated, to activate Tet-inducible constructs. To induce genotoxic stress mediated silencing of TINCR represses cell proliferation and EMT and/or p53 activation, cells were incubated with doxorubicin (1 µM), features102. However, in other studies performed in cell lines derived actinomycin-D (5 nM) or nutlin-3a (10 µM) for 24 h or irradiated from cervix, head and neck, and lung SCC, its silencing enhanced cell without medium with a UV light lamp emitting at 254 nm at a rate of growth and migration103. In cSCC, TINCR lncRNA has been consistently 50 J/m2. For protein stability assays, cells were incubated with cyclo- reported to have a tumor suppressor activity103,104. We have corrobo- heximide (100 μg/ml) for the specified time periods, 24 h after rated the downregulation of TINCR lncRNA in cSCC and, moreover, we doxycycline-induced pTINCR expression. For CDC42-activation assay have demonstrated that pTINCR protein is also lost in human cSCC or F-actin immunofluorescence, hEGF (Invitrogen) was added at compared to healthy epidermis. Indeed, its reexpression in cSCC PDXs 200μg/ml for 5 min or 24 h, respectively. For SUMOylation inhibition, reduces tumor growth in vivo. Of note, a marked deposition of hyaline ML792 (#HY-108702, MedChemExpress) was added at 10 µM for a matrix was evident in tumors expressing pTINCR. Hyaline globules minimum of 16 h. For CDC42-GTP inhibition, CASIN and ML141 (#HY- have been reported in numerous types of cancer as a rare event usually 12874 and #HY-12755 respectively, MedChemExpress) were added at observed in benign lesions63,105–107. Although there is no clear con- 5 µM for 2 h minimum prior to additional treatments. sensus about its origin, this phenomenon is generally regarded as a hallmark of degeneration and apoptosis63. Interestingly, the presence RNA-Seq analysis of TCGA data of hyaline deposits increases in tissues after injury or in tumors after Transcriptomic data were extracted from TCGA (database https:// therapy63,108,109. Further supporting the role of pTINCR in tumor sup- portal.gdc.cancer.gov/) in the form of count tables or from Das pression, we have demonstrated that pTINCR is upregulated upon Mahapatra et al. study (cSCC cases)44 in the form of fastq files. In this DNA damage in a p53 dependent manner. p53 is a well-known tumor last case, paired-end reads from RNA-Seq were aligned using Tophat suppressor and the most common mutational event in cSCC (http:// v.2.1.0 to the human genome (hg19). Predicted transcripts from p53.free.fr62,110), but also a known trigger of cell differentiation59. In Ensembl database (release 87) were analyzed. Differentially expressed fact, pTINCR-deficient cells fail to upregulate differentiation markers genes (DEG) were identified using HTSeq + DESeq2 v. 1.24.055,111. upon cellular damage. Finally, to expand the relevance of pTINCR in other cancer types, we performed correlation studies using public data Ribosome profiling analysis and observed that high TINCR expression is associated with a favorable We adapted a computational approach to identify translated sORFs112 prognosis in patients with different tumors of epithelial origin, such as by analyzing a public ribosome profiling dataset from mouse skin bladder, head & neck, pancreatic, stomach and lung adenocarcinomas. (GSE8333219). In brief, read adapters were trimmed and reads mapping Taken together, these evidences strongly support that pTINCR plays a to annotated ribosomal and transfer RNAs were filtered out. The role in tumor suppression. resulting reads were mapped to the assembled mouse genome In summary, we have revealed that pTINCR is a previously over- (mm10) using Bowtie2 v2.3.4.3 with default parameters. Next, mapped looked UBL- microprotein which regulates epithelial cell identity and reads from experimental replicates were merged and we used the shows tumor suppressor activity (Fig. 10). The discovery of pTINCR as ribORF algorithm113 to predict translated sORFs in TINCR lncRNA with a bioactive product of an assumed lncRNA nurtures the idea of an significant read uniformity and frame periodicity (score ≥ 0.7), as this additional level of regulation represented by lncRNA-encoded micro- feature is indicative of ORF active translation. proteins. Exploring the microproteome could provide new regulators of cell identity critical for the regulation of physiological and patho- In vitro transcription/translation logical processes, such as cancer. lncRNA TINCR CDS was cloned in a pcDNA-3.1 vector under the control of the T7 promoter and incubated for 1 h with rabbit reticulocyte- Methods coupled transcription/translation system (Promega) in the presence of Cell culture and treatments [35 S] methionine. After incubation, translated product was resolved by HEK293T, U2OS, MCF7 and A549 cells were cultured in DMEM with 15% SDS-polyacrylamide gel electrophoresis (PAGE) and detected by GlutaMAX supplemented with 10% of fetal bovine serum (FBS) and 1% autoradiography. of Penicillin-Streptomycin (P/S). cSCC-patient-derived hSCC10 cell line was derived as described48,49 and cultured in DMEM-F12 supple- LC-MS/MS identification of pTINCR microprotein mented with 2% B27 (Gibco) and 1% P/S. mESC V6.4 cell line was Identification of pTINCR peptides was performed using publicly cultured in DMEM GlutaMax supplemented with 1% Sodium Pyruvate available data of organotypic skin cell cultures from Elias et al45. In Nature Communications | (2022)13:6840 15 Article https://doi.org/10.1038/s41467-022-34529-6 brief, dataset with identifier PDX014088 was downloaded from PRIDE cocktail) and concentrations determined using the Pierce™ BCA Protein repository (https://www.ebi.ac.uk/pride/archive). Raw files were pro- Assay Kit (Thermo Fisher). Lysates were loaded in acrylamide gels for cessed with Proteome Discoverer 2.2 (Thermo Fisher Scientific) using electrophoresis in Tris-Glycine SDS Running Buffer. Primary antibodies the standard settings against a human protein database (UniProtKB/ were incubated overnight at 4 °C. Secondary HRP-conjugated anti- Swiss-Prot, July 2018, 20,373 sequences), supplemented with con- bodies were incubated the following day for 1 h at room temperature, taminants and the pTINCR sequences. Carbamidomethylation of and ECL Prime Western Blotting Detection Reagent (Fisher Scientific) cysteines was set as a fixed modification whereas methionine oxidation was used as a chemiluminescent reagent for protein detection. Anti- and N-term acetylation were variable protein modifications. The bodies and dilutions are listed in Supplementary Table 7. minimal peptide length was set to 6 amino acids and a maximum of two tryptic missed cleavages were allowed. The results were filtered at Subcellular fractionation 0.01 FDR (peptide and protein level). Cells were homogenized in Buffer A (HEPES pH 7.8 (10 mM), MgCl2 (1.5 mM), KCl (10 mM) and DTT). The homogenate was incubated on Histology and immunohistochemistry ice for 10 min and then 10% Triton-X was added to favor cellular dis- Immunostaining was performed on paraffin‐embedded mouse and ruption. Samples were centrifuged at 11,300 g for 1 min at 4 °C and human tissues. In brief, paraffin blocks were sliced into 5‐μM thick supernatant (Cytoplasmic extract, CE) was separated from the pellet sections, deparaffinized with xylene (Fisher Scientific, Waltham, MA, (Nuclear extract, NE). CE supernatant was washed with Buffer B (HEPES USA) and rehydrated with decreasing concentrations of ethanol in pH 7.8 (0.3 mM), MgCl2 (1.4 mM) and KCl (30 mM)), followed by cen- water. For immunohistochemistry, paraffin sections underwent anti- trifugation at 15,000 g for 15 min at 4 °C. New supernatant was used as genic exposure process into the Discovery Ultra (Ventana) system with final cytoplasmic extract and pellet was disrupted in SDS lysis buffer CC1 buffer for 64 min at 95 °C. Anti-pTINCR antibody was incubated (Tris-HCl pH 8.0 (50 mM), EDTA (1 mM), SDS (2%)) to obtained final for 1 h at room temperature (Supplementary Table 7). Next, slides were cellular membranes extract. NE pellet was washed with Buffer B and incubated with the secondary antibody Discovery UltraMap anti- disrupted using Buffer C (HEPES pH 7.8 (20 mM), MgCl2 (1.5 mM), NaCl Rabbit HRP (Ventana). H&E staining was performed on 5 μm paraffin (0.42 mM), EDTA (0.2 mM), glycerol (25%) and DTT). After 30 min of sections in a Robust carousel tissue stainer (Slee Medical) according to iced-incubation and centrifugation at 15,000 g for 15 min at 4 °C, the common method. supernatant was used as the final nuclear extract. Cloning procedures Immunofluorescence pTINCR ORF was synthesized (IDT technologies) fused with a flexible Cells were seeded in fibronectin-coated coverslips (Sigma Aldrich). linker (GGGGSGGGGSGGGGS) and an HA-Tag epitope at the C-terminal When desired, cells were fixed in 2% paraformaldehyde for 15 min and or N-terminal part of the microprotein and flanked by EcoRI enzyme permeabilized with 0.5% Triton X-100 for 10 min at room temperature. restrictions sites at both ends. After enzymatic digestion, constructs Blocking step was made in 3% Bovine Serum Albumin (BSA) for 1 h. were ligated into the pENTR1A vector. pTINCR-SIM mut plasmid was Cells were incubated overnight at 4 °C with the primary antibody purchased directly cloned into the pDONR201 plasmid, in-frame with an diluted in blocking buffer. Next day, secondary antibodies were incu- HA-tag epitope in its C-terminal (Proteogenix). Subsequently, pTINCR- bated for 1 h at room temperature in the dark. Antibodies and dilutions HA, HA-pTINCR and pTINCR-SIMmut constructs (Supplementary are listed in Supplementary Table 7. Finally, cells were mounted in Table 2) were obtained by recombining donor vectors with the lentiviral Prolong Mounting Medium with DAPI (Invitrogen) and images were inducible system pINDUCER20 (Invitrogen) using the Gateway Cloning taken in a Nikon Eclipse Ti-E inverted microscope system. For quanti- Technology, following manufacturer’s instructions. As specified, some fication, at least 200 cells per staining were evaluated using ImageJ experiments were reproduced using a pTINCR synthetic ORF (syORF), software. generated by mutating the pTINCR ORF in pTINCR-HA and pTINCR SIMmut constructs (Supplementary Table 2). Cell proliferation analysis Cells were seeded into 24-well cell culture plates at density 5 × 103 cells/ Lentiviral transduction well. Every two days, cells were trypsinised and counted using a Neu- HEK293T cells were transfected with 5μg of specific plasmid and 5μg of bauer chamber. packaging vectors (pLP-1, pLP-2, pLP-VSVG, Invitrogen) using Fugene HD (Promega) following the manufacturer’s instructions. Viral super- Generation of pTINCR-KO cells natants were collected twice a day on two consecutive days, filtered HaCaT and MCF7 cells were transfected with pSpCas9(BB)−2A-Puro through a 0.45 μm syringe filter, supplemented with of 8μg/ml of plasmid, which contains a sgRNAs (AGCCGGGCGGGCGCCATG- polybrene and used to infect HaCaT, A549, MCF7, hSCC10 or mESCs. GAGGG) design to target the first exon of TINCR gene loci. Twenty-four Successfully infected cells were established by geneticin selection. hours after transfection, puromycin was added to select for infected cells. Isolation of pTINCR-KO single colonies was assessed by serial Analysis of mRNA levels limiting dilution in 96-multiwell plate. Successfully CRISPR-CAS9 edi- Total RNA was extracted with Trizol (Invitrogen) following the manu- ted clones were screened by genotyping PCR. facturer’s protocol. Genomic DNA was cleaned up and retro- transcription performed using the iScript gDNA Clear cDNA Synthesis RNA sequencing Kit (BioRad). Gene expression was analyzed by RT-qPCR using Pow- Cells were cultured and treated with doxycycline for microprotein erUp SYBR Green Master Mix (Thermo Fisher Scientific) in the 7900HT expression induction during the specified experimental time periods. Fast Real-Time PCR System (Applied Biosystems). Gene-specific pri- Total RNA was isolated with Trizol following manufacturer’s protocol. mers are listed in Supplementary Table 5, 6. Cycle threshold (Ct) values RNA quantity and purity were measured with the Nanodrop spectro- were normalized to GAPDH. photometer (Thermo Scientific) and 1μg of total RNA was processed for sequencing analysis. RNA integrity, determined by the RNA integ- Western blot rity number (RIN), was determined with the 2100 Bioanalyzer (Agilent Cells were scrapped and homogenized in medium-salt lysis buffer Technologies Inc., Santa Clara, CA). Cytoplasmatic and mitochondrial (150 mM NaCl, 50 mM Tris pH 8, 1% NP40 and protease inhibitors ribosomal RNAs were depleted using the RiboZero Magnetic Gold Kit Nature Communications | (2022)13:6840 16 Article https://doi.org/10.1038/s41467-022-34529-6 (Illumina Inc). rRNA-depleted samples were fragmented, cDNA was the two values. Whenever a major discrepancy was observed between synthesized and converted into sequentiable libraries using the Tru- both observers, the cases were discussed using a multi-headed Seq Stranded Total RNA kit protocol (Illumina Inc.). The size and microscope. quality of the libraries were assessed with a High Sensitivity DNA For mutational profiling, FFPE tissue sections of 10 μm were Bioanalyzer assay (Agilent Technologies Inc., Santa Clara, CA). Librar- stained with H&E and examined by a pathologist to select for a ies were sequenced in a The NextSeq 500™ (Illumina Inc.), with a read minimum tumor content of 20% as a requirement for further length of 2x76bp. On average, 77 million paired-end reads were gen- processing. erated per sample. Image analysis, base calling and quality scoring of the run were processed using the manufacturer’s software Real Time Amplicon-seq analysis Analysis (RTA 1.18.64) and followed by generation of FASTQ sequence DNA was extracted from 5 × 10 μm sliced FFPE sections using the files by CASAVA. Maxwell FFPE Tissue LEV DNA Purification Kit (Promega), according to the manufacturer’s instructions. DNA quality and concentration were RNAseq data analysis determined by fluorometric quantification using Qubit Fluorimeter Paired end reads were aligned to the hg38 human genome with STAR and Qubit dsDNA BR Assay Kit (Life Technologies, Carlsbad, CA). (v2.5.2b) and default parameters. Sambamba (v0.6.7) was used to Tumor DNA was sequenced with an in-house developed amplicon- convert to bam and sort resulting sam files. All subsequent analyses sequencing panel of over 60 genes and 1330 primer pairs targeting were performed in the R programming environment (https://www.R- frequent mutations in oncogenes and several tumor suppressors. A project.org) unless otherwise stated. Count matrices were generated total of 500 ng of DNA from each tissue sample were used for library with the Rsubread package114 using the inbuilt annotation hg38. Genes preparation, according to our established protocols. An initial were annotated using biomaRt (vGRCh38.p12). Normalization and multiplex-PCR with a proof-reading polymerase was performed on all contrasts were performed using the DESeq255 R package. Time points samples. Indexed libraries were pooled and sequenced in a MiSeq were compared against 0 h within the control and pTINCR- instrument (2 × 100) at an average coverage of 1000X. Initial alignment overexpressing condition. was performed with BWA (v0.7.17) after primer sequence clipping and In order to find genes whose expression was significantly asso- variant calling was done with the GATK Unified Genotyper (v3.4.0) and ciated with time we applied the runImpulseDE2 function from the VarScan2 (v2.4.3) followed by ANNOVAR annotation (v.anno- ImpulseDE254 R package on the count matrix generated with Rsubread. var_180416). Mutations were called at a minimum 3% allele frequency. The biological replicate was included as a technical confounder. Genes SNPs were filtered out with dbSNP and 1000 genome datasets. All with adjusted p-values lower than 0.001 were selected for downstream detected variants were manually checked. Technical quality para- analyses. The matrix containing the fold changes resulting from the meters are provided in Supplementary Table 8. comparisons against 0 h was filtered for those genes with significant association with time according to ImpulseDE. The cmeans function ChipSeq analysis from the e107 R package was used to find clusters of genes with similar Data was extracted from GSE58506 and GSE100292, aligned with expression patterns along time. The centers parameter was fixed at 6 GRCh37/hg19 and visualized with the Integrative Genomics Viewer after manual inspection. The ggplot2 (https://ggplot2.tidyverse.org), (IGV) (v2.13.2). viridis, ggExtra and gridExtra R packages were used for plotting resulting clusters. Genes were classified in a given cluster if their Non-covalent SUMO binding assay membership probabilities were larger than 0.5. Functional enrichment Glutathione S-transferase (GST) pull-down experiments were per- on the resulting gene clusters was performed using a hypergeometric formed by incubating protein extracts from U2OS cells stably test against the Gene Ontology database115, KEGG pathways116 and expressing pTINCR-HA or pTINCR SIMmut treated with Doxycycline Broad Hallmarks gene sets117. for 48 h, together with GST, GST-SUMO1 or GST-SUMO2/3 bound onto Networks of overlapping gene ontology terms were computed glutathione-sepharose beads for 2 h at 4 °C. Then, beads were washed using the ClueGO (v2.5.7) module of Cytoscape (v3.9.1) with default and bound proteins were eluted, separated on 14% SDS-PAGE and parameters. detected by Western-blot. Furthermore, the interaction between treatment and time was computed for times 0 h and 21 days. Functional enrichment of the Interactome analysis and validation interaction coefficients was performed using a rotation-based meth- Cells were lysed in a buffer containing 50 mM Tris-Hcl pH 7.5-8, odology. The ROAST algorithm (v0.99.1) as implemented in the R 150 mM NaCl, 1%Triton X-100 and protease inhibitors, and homo- package limma was used to represent the null distribution. The max- genized for 30 min in a rotor wheel. Lysates (3 mg) were immuno- mean enrichment statistic proposed in118, under restandardization, was precipitated with 5μg of monoclonal HA-Tag antibody overnight. considered for competitive testing. Immunocomplexes were collected using PureProteome™ Protein A Magnetic Beads (MERCK) and eluted by competition with a synthetic Human cSCC samples collection and processing HA peptide (Sigma). Eluate was digested with trypsin and analyzed by A historical cohort of fifty-one patients diagnosed with cSCC from liquid chromatography-mass spectrometry on an LTQ Orbitrap Velos January 2009 to August 2010 was included in the present study pro- instrument (ThermoFisher). Progenesis ® QI for proteomics software vided by the Tumor Bank of the Vall d’Hebron University Hospital v3.0 (Nonlinear dynamics, UK) was used for MS data analysis using Biobank. For histological examination, H&E staining of formalin-fixed default settings. The LC-MS runs were automatically aligned to an and paraffin-embedded (FFPE) tumor sample was performed. All cases automatically selected reference sample with manual supervision Peak were evaluated independently by two pathologists (SRYC, OM). lists were generated from Progenesis and loaded to Proteome Dis- For immunohistochemistry studies of pTINCR and p53, whole coverer v2.1 (Thermo Fisher Scientific) for protein identification. Pro- slide FFPE tissue sections of 5 μm of selected samples were stained as teins were identified using Mascot v2.5 (Matrix Science, London UK) to previously described. All cases were evaluated independently by search the SwissProt database (2018_11, taxonomy restricted to human an one expert dermatopathologist and one trained Molecular proteins, 20,240 sequences). Significance threshold for the identifi- Biologist blinded for patient groups, considering the percentage of cations was set to p < 0.05, minimum ions score of 20. Statistical positive cells and intensity of the staining, which was assessed semi- analysis was performed using Progenesis software. Proteins displaying quantitatively. Final results were obtained utilizing the average of greater than 2-fold change, and p < 0.05 (T-test) between IP and Nature Communications | (2022)13:6840 17 Article https://doi.org/10.1038/s41467-022-34529-6 control groups were considered significantly differential. Interactions Mice were checked daily for general condition and human were analyzed using CRAPome algorithm119. For validation of the endpoints. results, U2OS cells were co-transfected with pTINCR-HA and CDC42- For subcutaneous teratomas and patient-derived xenografts Flag plasmids and immunoprecipitation was performed as explained (PDXs), mESCs V6.4 or hSCC10 cells lines, respectively, expressing before. SDS-PAGE was used to visualized anti-HA pTINCR and anti-Flag either pTINCR HA-tagged constructs or control vector were trypsi- CDC42 proteins. nized and 1 × 106 cells were subcutaneously injected into the flanks of 8-week-old immunocompromised male NMRI mice (RjHan:NMRI). CDC42 activation assay Tumor growth was monitored twice a week using the formula The activation of CDC42 (CDC42-GTP) was estimated using the Cdc42 height × width × width × (3,1416/6) and animals were sacrificed when activation kit (Cytoskeleton Inc.). Briefly, after 4 days of pTINCR dox- tumors reached 1.7 cm3 or before at the indicated time. Doxycycline ycycline induction cells were washed with cold PBS, quickly lysed using was administered in the drinking water at 1 mg/ml supplemented with provided lysis buffer and concentrations determined using the 7.5% of sucrose every two days throughout all the experiments. Pierce™ BCA Protein Assay Kit (Thermo Fisher). GTP-bound CDC42 was pulled-down by GST-p21 binding domain (PBD) and detected with Generation of Kaplan–Meier plots anti-CDC42 antibody (Supplementary Table 7) according to the man- Kaplan-Meier (KM) plots were generated for survival analysis using KM ufacturer’s instructions. Positive control was the result of treating plotter database (http://kmplot.com/analysis), a website database control cells with EGF for 5 min prior to the procedure. based on resources from TCGA database. The final prognostic KM plots were presented with a hazard ratio (HR), 95% confidence interval SUMO conjugation assays (CI) and log-rank P value. P value < 0.05 was considered statistically SUMO conjugation assays were performed by transfecting U2OS stably significant expressing pTINCR cell line with HIS6-SUMO2 protein and, when required, co-transfected CDC42 Flag-tagged plasmids. After 48 h of Statistics and reproducibility doxycycline treatment, cells were lysed with 6 M guanidinium-HCl, Data are expressed as the mean ± standard deviation (SD) or the mean 0.1 M Na2HPO4/NaH2PO4 buffer. Then, lysates were mixed with Ni2+- ± standard error of the mean (SEM), as specified. Differences between nitrilotriacetic acid-agarose beads and incubated for 2 h at room groups were analyzed using unpaired T-TEST, Multiple T-TEST correct temperature. After washing, beads were resuspended in Laemmli for multiple comparison, one or two-way ANOVA with multiple cor- buffer and purified proteins were subjected to SDS-PAGE electro- rection, Fisher exact test or non-parametric Kruskal-Wallis test, as phoresis as indicated above. specified. Corrections for multiple comparison were performed when necessary: one and two-way ANOVA with Dunnett method, multiple Cell migration assay comparison T-TEST with Holm-Sidak method and Kruskal–Wallis A549 cells were seeded and culture until confluence was reached. test with Dunn’s test. All statistical tests were two-sided and performed Then, a pipette tip was used to scratch the surface of cell monolayer, using GraphPad Prism (v8.4.0, GraphPad Software Inc., San Diego, forming a wound. An Olympus CellR microscope equipped with a CA, USA). The experiments shown in this manuscript were repeated Hamamatsu C9100 camera was used to follow the closure of the independently at least twice obtaining similar results, unless specified. wound up to 48 h. hSCC10 cells were seeded in a Corning cell culture insert (Boyden Ethical statement chamber). Cells were allowed to migrate through for 24 h at 37 °C. The animal studies in this work comply with the European, Spanish, After that time, culture inserts were fixed in methanol for 5 min and and Catalan Regulations for the Protection of Vertebrate Animals used stained using Crystal Violet during 20 min at room temperature. Pic- for Experimental and other Scientific Purposes (Directive 2010/63; tures were taken using an Olympus CellR microscope. Cells were Spanish BOE RD 53/2013; Catalan DOGC 214/1997). All studies were counted using ImageJ. carried out in the “Lab Animal Service Campus Vall d’Hebron (LAS- CVH)”, registered and accredited at the Departament de Medi Ambient Cell invasion assay i Habitatge by Generalitat of Catalonia government with register 2 × 104 cells were seeded in a Corning cell culture insert (Boyden number B9900062. The experiments performed for this manuscript chamber) coated with a layer of Matrigel (Corning). Cells were allowed were linked to a project approved by Vall d’Hebron Ethics Committee, to invade through the membrane for 24 h at 37 °C. After that time, and the Commission of Animal Experimentation of Generalitat of culture inserts were fixed in methanol for 5 min and stained Catalonia government. using Crystal Violet during 20 min at room temperature. Pictures were Studies involving human samples have been performed in taken using an Olympus CellR microscope. Cells were counted using accordance with the Declaration of Helsinki. All patient samples stu- ImageJ. died have the approval of the Ethics Committee for Research of the Vall d’Hebron University Hospital (PR research project PR research project Human cytoskeleton phospho-array (AG) 191/2019)). Samples recruitment was not done specifically for Phosphorylation-specific antibody microarray (Fullmoon Biosystems our study. Inc.) was used to determine the up- and down-regulated proteins in hSCC10 cells after 1 day of pTINCR induction. The array Materials availability contains 141 site-specific antibodies against phosphorylated- and All the unique reagents described in this work will be made available unphosphorylated-proteins involved in cytoskeletal pathways, each according to the guidelines of the journal. replicated six times. Actin and GAPDH were used as controls. The assay was performed following manufacturer’s protocol. Reporting summary Further information on research design is available in the Nature Animal procedures Portfolio Reporting Summary linked to this article. Mice were kept in enriched shelters in a pathogen-free facility with 12 h/12 h light/dark cycle (from 08:00am to 08:00 pm) with ad Data availability libitum access to food and water. Housing conditions were main- The RNA-seq data generated for this publication has been deposited in tained at an average temperature of 22 ± 2 °C and humidity 55 ± 10%. NCBI’s Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo) Nature Communications | (2022)13:6840 18 Article https://doi.org/10.1038/s41467-022-34529-6 and are accessible through GEO Series accession number GSE175463. 19. Sendoel, A. et al. Translation from unconventional 5’ start sites The mass spectrometry proteomics data have been deposited to the drives tumour initiation. Nature 541, 494–499 (2017). ProteomeXchange Consortium via the PRIDE partner repository and 20. Campanale, J. P., Sun, T. Y. & Montell, D. J. Development and are accessible with the dataset identifier PXD026181. Published data- dynamics of cell polarity at a glance. J. Cell Sci. 130, sets included in this study: PDX014088 (https://www.ebi.ac.uk/ 1201–1207 (2017). pride/archive/), GSE139505, GSE83332, GSE58506, GSE100292. Tran- 21. Huang, J. et al. Differential regulation of adherens junction scriptomic data was downloaded from GTex portal (https://gtexportal. dynamics during apical-basal polarization. J. Cell Sci. 124, org/home/). Figures with associated raw data are Fig. 5, Fig. 9b and 4001–4013 (2011). Supplementary Fig. 4 (GSE175463), Fig. 9a and Supplementary Table 3 22. Zihni, C. et al. Tight junctions: from simple barriers to multi- (PXD026181), Fig. 1a (GSE139505), Fig. 1c (GSE83332), Fig. 1g and functional molecular gates. Nat. Rev. Mol. Cell Biol. 17, Supplementary Fig. 1a (PDX014088) and Supplementary Fig. 5E and F 564–580 (2016). (GSE58506, GSE100292), Fig. 1b (GTex). All other data that support the 23. Simpson, C. L., Patel, D. M. & Green, K. J. Deconstructing the skin: findings of this study are available within the article, its supplementary cytoarchitectural determinants of epidermal morphogenesis. Nat. information, or Source Data file. Source data are provided with Rev. Mol. Cell Biol. 12, 565–580 (2011). this paper. 24. Vaezi, A. et al. Actin cable dynamics and Rho/Rock orchestrate a polarized cytoskeletal architecture in the early steps of References assembling a stratified epithelium. Dev. Cell 3, 367–381 1. Orr, M. W. et al. Alternative ORFs and small ORFs: shedding light (2002). on the dark proteome. Nucleic Acids Res. 48, 1029–1042 (2020). 25. Pichaud, F., Walther, R.F. & Nunes de Almeida, F. Regulation of 2. Makarewich, C. A. & Olson, E. N. Mining for micropeptides. Trends Cdc42 and its effectors in epithelial morphogenesis. J. Cell Sci. Cell Biol. 27, 685–696 (2017). 132, jcs217869 (2019). 3. Merino-Valverde, I., Greco, E. & Abad, M. The microproteome of 26. Wu, X. et al. Cdc42 is crucial for the establishment of epithelial cancer: From invisibility to relevance. Exp. Cell Res. 392, polarity during early mammalian development. Dev. Dyn. 236, 111997 (2020). 2767–2778 (2007). 4. Huang, J. Z. et al. A peptide encoded by a putative lncRNA HOXB- 27. Etienne-Manneville, S. Cdc42–the centre of polarity. J. Cell Sci. AS3 suppresses colon cancer growth. Mol. Cell 68, 117, 1291–1300 (2004). 171–184.e6 (2017). 28. Nunes de Almeida, F. et al. Cdc42 defines apical identity and 5. Nelson, B. R. et al. A peptide encoded by a transcript annotated as regulates epithelial morphogenesis by promoting apical recruit- long noncoding RNA enhances SERCA activity in muscle. Science ment of Par6-aPKC and Crumbs. Development 146, 351, 271–275 (2016). dev175497 (2019). 6. Arnoult, N. et al. Regulation of DNA repair pathway choice in S and 29. Wu, X., Quondamatteo, F. & Brakebusch, C. Cdc42 expression in G2 phases by the NHEJ inhibitor CYREN. Nature 549, keratinocytes is required for the maintenance of the basement 548–552 (2017). membrane in skin. Matrix Biol. 25, 466–474 (2006). 7. Makarewich, C. A. et al. MOXI Is a Mitochondrial Micropeptide That 30. Elbediwy, A. et al. Epithelial junction formation requires confine- Enhances Fatty Acid beta-Oxidation. Cell Rep. 23, ment of Cdc42 activity by a novel SH3BP1 complex. J. Cell Biol. 3701–3709 (2018). 198, 677–693 (2012). 8. Prensner, J. R. et al. Noncanonical open reading frames encode 31. Wallace, S. W. et al. Cdc42 regulates apical junction formation in functional proteins essential for cancer cell survival. Nat. Bio- human bronchial epithelial cells through PAK4 and Par6B. Mol. technol. 39, 697–704 (2021). Biol. Cell 21, 2996–3006 (2010). 9. Senis, E. et al. TUNAR lncRNA encodes a microprotein that reg- 32. Muthuswamy, S. K. & Xue, B. Cell polarity as a regulator of cancer ulates neural differentiation and neurite formation by modulating cell behavior plasticity. Annu. Rev. Cell Dev. Biol. 28, calcium dynamics. Front Cell Dev. Biol. 9, 747667 (2021). 599–625 (2012). 10. Makarewich, C. A. et al. The DWORF micropeptide enhances 33. Royer, C. & Lu, X. Epithelial cell polarity: a major gatekeeper contractility and prevents heart failure in a mouse model of dilated against cancer? Cell Death Differ. 18, 1470–1477 (2011). cardiomyopathy. Elife 7, e38319 (2018). 34. Yuan, S., Norgard, R. J. & Stanger, B. Z. Cellular plasticity in cancer. 11. Matsumoto, A. et al. mTORC1 and muscle regeneration are regu- Cancer Discov. 9, 837–851 (2019). lated by the LINC00961-encoded SPAR polypeptide. Nature 541, 35. Dongre, A. & Weinberg, R. A. New insights into the mechanisms of 228–232 (2017). epithelial-mesenchymal transition and implications for cancer. 12. Khitun, A., Ness, T. J. & Slavoff, S. A. Small open reading frames Nat. Rev. Mol. Cell Biol. 20, 69–84 (2019). and cellular stress responses. Mol. Omics 15, 108–116 (2019). 36. Molinuevo, R. et al. The DNA damage response links human 13. Andreev, D. E. et al. Translation of 5’ leaders is pervasive in genes squamous proliferation with differentiation. J. Cell Biol. 219, resistant to eIF2 repression. Elife 4, e03971 (2015). e202001063 (2020). 14. Moro, S. G. et al. Impact of uORFs in mediating regulation of 37. Sanz-Gomez, N. et al. Squamous differentiation requires G2/ translation in stress conditions. BMC Mol. Cell Biol. 22, 29 (2021). mitosis slippage to avoid apoptosis. Cell Death Differ. 27, 15. Starck, S. R. et al. Translation from the 5’ untranslated region 2451–2467 (2020). shapes the integrated stress response. Science 351, 38. Ying, Z., Sandoval, M. & Beronja, S. Oncogenic activation of PI3K aad3867 (2016). induces progenitor cell differentiation to suppress epidermal 16. Andreev, D. E. et al. Oxygen and glucose deprivation induces growth. Nat. Cell Biol. 20, 1256–1266 (2018). widespread alterations in mRNA translation within 20 min. Gen- 39. Lee, M. & Vasioukhin, V. Cell polarity and cancer–cell and tissue ome Biol. 16, 90 (2015). polarity as a non-canonical tumor suppressor. J. Cell Sci. 121, 17. Jackson, R. et al. The translation of non-canonical open reading 1141–1150 (2008). frames controls mucosal immunity. Nature 564, 434–438 (2018). 40. Kretz, M. et al. Control of somatic tissue differentiation by the long 18. Razooky, B. S. et al. Viral infection identifies micropeptides dif- non-coding RNA TINCR. Nature 493, 231–235 (2013). ferentially regulated in smORF-containing lncRNAs. Genes (Basel) 41. Zhuang, Z. et al. Down-regulation of long non-coding RNA 8, 206 (2017). TINCR induces cell dedifferentiation and predicts progression Nature Communications | (2022)13:6840 19 Article https://doi.org/10.1038/s41467-022-34529-6 in oral squamous cell carcinoma. Front Oncol. 10, 624752 64. Yang, J. & Zhang, Y. I-TASSER server: new development for protein (2020). structure and function predictions. Nucleic Acids Res. 43, 42. Omote, N. et al. Long noncoding RNA TINCR is a novel regulator of W174–W181 (2015). human bronchial epithelial cell differentiation state. Physiol. Rep. 65. Cappadocia, L. & Lima, C. D. Ubiquitin-like protein conjugation: 9, e14727 (2021). structures, chemistry, and mechanism. Chem. Rev. 118, 43. Lin, M. F., Jungreis, I. & Kellis, M. PhyloCSF: a comparative geno- 889–918 (2018). mics method to distinguish protein coding and non-coding 66. Grabbe, C. & Dikic, I. Functional roles of ubiquitin-like domain regions. Bioinformatics 27, i275–i282 (2011). (ULD) and ubiquitin-binding domain (UBD) containing proteins. 44. Das Mahapatra, K. et al. A comprehensive analysis of coding and Chem. Rev. 109, 1481–1494 (2009). non-coding transcriptomic changes in cutaneous squamous cell 67. Luders, J., Pyrowolakis, G. & Jentsch, S. The ubiquitin-like protein carcinoma. Sci. Rep. 10, 3637 (2020). HUB1 forms SDS-resistant complexes with cellular proteins in the 45. Elias, M. S. et al. EMSY expression affects multiple components of absence of ATP. EMBO Rep. 4, 1169–1174 (2003). the skin barrier with relevance to atopic dermatitis. J. Allergy Clin. 68. Zhao, Q. et al. GPS-SUMO: a tool for the prediction of sumoylation Immunol. 144, 470–481 (2019). sites and SUMO-interaction motifs. Nucleic Acids Res. 42, 46. Nanba, D. et al. Actin filament dynamics impacts keratinocyte W325–W330 (2014). stem cell maintenance. EMBO Mol. Med. 5, 640–653 (2013). 69. Ren, J. et al. Systematic study of protein sumoylation: develop- 47. Lewis, L. et al. The reorganization of microtubules and microfila- ment of a site-specific predictor of SUMOsp 2.0. Proteomics 9, ments in differentiating keratinocytes. Differentiation 36, 3409–3412 (2009). 228–233 (1987). 70. Hodge, R. G. & Ridley, A. J. Regulating Rho GTPases and their 48. Bernat-Peguera, A. et al. FGFR inhibition overcomes resistance to regulators. Nat. Rev. Mol. Cell Biol. 17, 496–510 (2016). EGFR-targeted therapy in epithelial-like cutaneous carcinoma. 71. Jaffe, A. B. et al. Cdc42 controls spindle orientation to position the Clin. Cancer Res. 27, 1491–1504 (2021). apical surface during epithelial morphogenesis. J. Cell Biol. 183, 49. Bernat-Peguera, A. et al. PDGFR-induced autocrine SDF-1 signaling 625–633 (2008). in cancer cells promotes metastasis in advanced skin carcinoma. 72. Hochstrasser, M. Origin and function of ubiquitin-like proteins. Oncogene 38, 5021–5037 (2019). Nature 458, 422–429 (2009). 50. Hennings, H., Holbrook, K. A. & Yuspa, S. H. Factors influencing 73. Castillo-Lluva, S. et al. SUMOylation of the GTPase Rac1 is calcium-induced terminal differentiation in cultured mouse epi- required for optimal cell migration. Nat. Cell Biol. 12, 1078–1085 dermal cells. J. Cell Physiol. 116, 265–281 (1983). (2010). 51. Chandramouleeswaran, P. M. et al. Preferential secretion of thy- 74. Tu, S. et al. Epidermal growth factor-dependent regulation of mic stromal lymphopoietin (TSLP) by terminally differentiated Cdc42 is mediated by the Src tyrosine kinase. J. Biol. Chem. 278, esophageal epithelial cells: relevance to eosinophilic esophagitis 49293–49300 (2003). (EoE). PLoS ONE 11, e0150968 (2016). 75. Mitsushima, M., Toyoshima, F. & Nishida, E. Dual role of Cdc42 in 52. McGrath, C. M. & Soule, H. D. Calcium regulation of normal human spindle orientation control of adherent cells. Mol. Cell Biol. 29, mammary epithelial cell growth in culture. Vitro 20, 2816–2827 (2009). 652–662 (1984). 76. van den Bout, I. & Divecha, N. PIP5K-driven PtdIns(4,5) 53. Vandewalle, B. et al. Intracellular calcium and breast-cancer P2 synthesis: regulation and cellular functions. J. Cell Sci. 122, cell-growth and differentiation. Int J. Oncol. 2, 613–620 3837–3850 (2009). (1993). 77. Eckhart, L. et al. TINCR is not a non-coding RNA but encodes a 54. Fischer, D. S., Theis, F. J. & Yosef, N. Impulse model-based dif- protein component of cornified epidermal keratinocytes. Exp. ferential expression analysis of time course sequencing data. Dermatol. 29, 376–379 (2020). Nucleic Acids Res. 46, e119 (2018). 78. Nita, A. et al. A ubiquitin-like protein encoded by the “noncoding” 55. Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold RNA TINCR promotes keratinocyte proliferation and wound heal- change and dispersion for RNA-seq data with DESeq2. Genome ing. PLoS Genet. 17, e1009686 (2021). Biol. 15, 550 (2014). 79. Cossec, J. C. et al. SUMO safeguards somatic and pluripotent cell 56. Zheng, Y. et al. Expressions of oncogenes c-fos and c-myc in skin identities by enforcing distinct chromatin states. Cell Stem Cell lesion of cutaneous squamous cell carcinoma. Asian Pac. J. Trop. 23, 742–757.e8 (2018). Med. 7, 761–764 (2014). 80. Theurillat, I. et al. Extensive SUMO modification of repressive 57. Zhang, Y. et al. MALAT1-KTN1-EGFR regulatory axis promotes the chromatin factors distinguishes pluripotent from somatic cells. development of cutaneous squamous cell carcinoma. Cell Death Cell Rep. 33, 108251 (2020). Differ. 26, 2061–2073 (2019). 81. Heaton, P. R. et al. Analysis of global sumoylation changes 58. Sivakumar, S. et al. TP53 promotes lineage commitment of human occurring during keratinocyte differentiation. PLoS ONE 7, embryonic stem cells through ciliogenesis and sonic hedgehog e30165 (2012). signaling. Cell Rep. 38, 110395 (2022). 82. Luis, N. M. et al. Regulation of human epidermal stem cell pro- 59. Stiewe, T. The p53 family in differentiation and tumorigenesis. Nat. liferation and senescence requires polycomb- dependent and Rev. Cancer 7, 165–168 (2007). -independent functions of Cbx4. Cell Stem Cell 9, 233–246 60. Sanchez, Y. et al. Genome-wide analysis of the human p53 tran- (2011). scriptional network unveils a lncRNA tumour suppressor sig- 83. Deyrieux, A. F. et al. Sumoylation dynamics during keratinocyte nature. Nat. Commun. 5, 5812 (2014). differentiation. J. Cell Sci. 120, 125–136 (2007). 61. Hafner, A. et al. Identification of universal and cell-type specific 84. Gonzalez-Santamaria, J. et al. Regulation of the tumor suppressor p53 DNA binding. BMC Mol. Cell Biol. 21, 5 (2020). PTEN by SUMO. Cell Death Dis. 3, e393 (2012). 62. Brash, D. E. Roles of the transcription factor p53 in keratinocyte 85. den Besten, W. et al. Ubiquitination of, and sumoylation by, the Arf carcinomas. Br. J. Dermatol 154, 8–10 (2006). tumor suppressor. Isr. Med. Assoc. J. 8, 249–251 (2006). 63. Papadimitriou, J. C. et al. “Thanatosomes”: a unifying morphoge- 86. Lopez, I. et al. An unanticipated tumor-suppressive role of the netic concept for tumor hyaline globules related to apoptosis. SUMO pathway in the intestine unveiled by Ubc9 haploinsuffi- Hum. Pathol. 31, 1455–1465 (2000). ciency. Oncogene 39, 6692–6703 (2020). Nature Communications | (2022)13:6840 20 Article https://doi.org/10.1038/s41467-022-34529-6 87. Maldonado, M. D. M. & Dharmawardhane, S. Targeting Rac and 109. Rekhi, B. et al. Raspberry bodies and hyaline globules with positive Cdc42 GTPases in cancer. Cancer Res. 78, 3101–3111 napsin A immunoexpression are useful features in diagnosing (2018). clear cell carcinoma of the female genital tract in cytology sam- 88. Zhang, M. et al. Cdc42 deficiency leads to epidermal barrier ples. Cytopathology 29, 600–602 (2018). dysfunction by regulating intercellular junctions and keratiniza- 110. Petitjean, A. et al. TP53 mutations in human cancers: functional tion of epidermal cells during mouse skin development. Ther- selection and impact on cancer prognosis and outcomes. Onco- anostics 9, 5065–5084 (2019). gene 26, 2157–2165 (2007). 89. Hoadley, K. A. et al. Multiplatform analysis of 12 cancer types 111. Anders, S. & Huber, W. Differential expression analysis for reveals molecular classification within and across tissues of origin. sequence count data. Genome Biol. 11, R106 (2010). Cell 158, 929–944 (2014). 112. Ruiz-Orera, J. et al. Translation of neutrally evolving peptides 90. Lorenzo-Martin, L. F. et al. VAV2 signaling promotes regenerative provides a basis for de novo gene evolution. Nat. Ecol. Evol. 2, proliferation in both cutaneous and head and neck squamous cell 890–896 (2018). carcinoma. Nat. Commun. 11, 4788 (2020). 113. Ji, Z. et al. Many lncRNAs, 5’UTRs, and pseudogenes are translated 91. Gladden, A. B. et al. The NF2 tumor suppressor, Merlin, regulates and some are likely to express functional proteins. Elife 4, epidermal development through the establishment of a junctional e08890 (2015). polarity complex. Dev. Cell 19, 727–739 (2010). 114. Liao, Y., Smyth, G. K. & Shi, W. The R package Rsubread is easier, 92. Xiao, G. H. et al. p21-activated kinase links Rac/Cdc42 signaling to faster, cheaper and better for alignment and quantification of RNA merlin. J. Biol. Chem. 277, 883–886 (2002). sequencing reads. Nucleic Acids Res. 47, e47 (2019). 93. Dainichi, T. et al. PDK1 is a regulator of epidermal differentiation 115. Gene Ontology, C. The gene ontology resource: enriching a GOld that activates and organizes asymmetric cell division. Cell Rep. 15, mine. Nucleic Acids Res. 49, D325–D334 (2021). 1615–1623 (2016). 116. Ogata, H. et al. KEGG: kyoto encyclopedia of genes and genomes. 94. Calautti, E. et al. Phosphoinositide 3-kinase signaling to Akt pro- Nucleic Acids Res. 27, 29–34 (1999). motes keratinocyte differentiation versus death. J. Biol. Chem. 117. Liberzon, A. et al. Molecular signatures database (MSigDB) 3.0. 280, 32856–32865 (2005). Bioinformatics 27, 1739–1740 (2011). 95. Janes, S. M. et al. PI3-kinase-dependent activation of apoptotic 118. Bradley Efron, R. T. On testing the significance of sets ofgenes. machinery occurs on commitment of epidermal keratinocytes to Ann. Appl. Stat. 1, 107–129 (2007). terminal differentiation. Cell Res. 19, 328–339 (2009). 119. Mellacheruvu, D. et al. The CRAPome: a contaminant repository 96. Pankow, S. et al. Regulation of epidermal homeostasis and repair for affinity purification-mass spectrometry data. Nat. Methods 10, by phosphoinositide 3-kinase. J. Cell Sci. 119, 4033–4046 (2006). 730–736 (2013). 97. Horikoshi, Y. et al. Interaction between PAR-3 and the aPKC-PAR-6 complex is indispensable for apical domain development of epi- thelial cells. J. Cell Sci. 122, 1595–1606 (2009). Acknowledgements 98. Hirose, T. et al. Involvement of ASIP/PAR-3 in the promotion of The authors thank VHIO Proteomics, Molecular Oncology and Genomics epithelial tight junction formation. J. Cell Sci. 115, Core Facilities for technical assistance. We are grateful to Manuel Ser- 2485–2495 (2002). rano for providing several reagents, advice and critical discussion on the 99. Yang, G. N., Kopecki, Z. & Cowin, A. J. Role of actin cytoskeleton in manuscript. We also thank Alonso García and Raquel Pérez for their help the regulation of epithelial cutaneous stem cells. Stem Cells Dev. in processing and analyzing digital images, Gemma Serra and Sandra 25, 749–759 (2016). Peiró for their assistance with subcellular fractionation and immuno- 100. Ghafouri-Fard, S. et al. TINCR: An lncRNA with dual functions in precipitation experiments, Sara Arce and Joaquín Mateo for providing the carcinogenesis process. Noncoding RNA Res. 5, several reagents during the development of critical experiments of this 109–115 (2020). manuscript, and Juan Angel Recio for his help with the cSCC cohort. We 101. Zhu, Z. J. & He, J. K. TINCR facilitates non-small cell lung cancer are immensely grateful to all the members of the Abad lab for generating progression through BRAF-activated MAPK pathway. Biochem. the know-how for the identification of novel sORFs, for the critical Biophys. Res. Commun. 497, 971–977 (2018). reading on the manuscript and in general for their constant support to 102. Xu, Y. et al. Long noncoding RNA, tissue differentiation-inducing this project. Work in the Abad lab is supported by VHIO, Fero Founda- nonprotein coding RNA is upregulated and promotes develop- tion, La Caixa Foundation, Asociación Española Contra el Cancer ment of esophageal squamous cell carcinoma. Dis. Esophagus 29, (AECC), La Mutua Foundation and by grants from the Spanish Ministry of 950–958 (2016). Science and Innovation (SAF2015-69413-R; RTI2018-102046-B-I00). 103. Hazawa, M. et al. ZNF750 is a lineage-specific tumour suppressor M.A. was recipient of a Ramón y Cajal contract from the Spanish Ministry in squamous cell carcinoma. Oncogene 36, 2243–2254 (2017). of Science and Innovation (RYC-2013-14747). O.B. is recipient of a FPI- 104. Zhou, W. et al. lncRNA TINCR participates in ALA-PDT-induced AGAUR fellowship from Generalitat de Catalunya. We also acknowledge apoptosis and autophagy in cutaneous squamous cell carcinoma. funding from grant PGC2018-094091-B-I00 from the Spanish J. Cell Biochem. 120, 13893–13902 (2019). Government. 105. Dikov, D. I. et al. Hyaline globules (thanatosomes) in gastro- intestinal epithelium: pathophysiologic correlations. Am. J. Clin. Author contributions Pathol. 127, 792–799 (2007). O.B. designed and performed most of the experiments and data analy- 106. Meriden, Z. et al. Hyaline globules in neuroendocrine and solid- sis, contributed to discussion and co-wrote the manuscript. M.M. per- pseudopapillary neoplasms of the pancreas: a clue to the diag- formed a substantial number of experiments and contributed to nosis. Am. J. Surg. Pathol. 35, 981–988 (2011). discussion and writing. S.V.F. and C.R. performed most of the experi- 107. Michal, M. et al. Ovarian fibromas with heavy deposition of hyaline ments related with SUMO. M.G. and L.P generated the pTINCR-KO cell globules: a diagnostic pitfall. Int J. Gynecol. Pathol. 28, lines and provided general technical assistance throughout the project. 356–361 (2009). L.L.S. and P.M. generated and provided the patient-derived cSCC cell 108. Aishima, S. et al. p62+ Hyaline inclusions in intrahepatic cho- lines. L.Q. and I.V. analyzed cSCC transcriptomic data and correlated langiocarcinoma associated with viral hepatitis or alcoholic liver TINCR expression with clinical data in multiple tumor types. O.M., N.C., disease. Am. J. Clin. Pathol. 134, 457–465 (2010). P.N., and S.R.Y.C. performed the histopathological analyses. J.R.O. and Nature Communications | (2022)13:6840 21 Article https://doi.org/10.1038/s41467-022-34529-6 M.M.A. analyzed the Ribo-seq. P.X.E. and J.M. analyzed the MS data on Reprints and permissions information is available at organotypic skin cultures. T.V.T. analyzed the p53 ChIP-seq experi- http://www.nature.com/reprints ments. A.V. performed generated the libraries and performed the RNA- seq on cSCC cells. C.S.O.A. analyzed the RNA-seq on cSCC cells. M.A. Publisher’s note Springer Nature remains neutral with regard to jur- designed and supervised the study, secured funding, analyzed the data, isdictional claims in published maps and institutional affiliations. and wrote the manuscript. All authors discussed the results and com- mented on the manuscript. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, Competing interests adaptation, distribution and reproduction in any medium or format, as The authors declare no competing interests. long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if Additional information changes were made. The images or other third party material in this Supplementary information The online version contains article are included in the article’s Creative Commons license, unless supplementary material available at indicated otherwise in a credit line to the material. If material is not https://doi.org/10.1038/s41467-022-34529-6. included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted Correspondence and requests for materials should be addressed to use, you will need to obtain permission directly from the copyright María Abad. holder. To view a copy of this license, visit http://creativecommons.org/ licenses/by/4.0/. Peer review information Nature Communications thanks Dimitris Xir- odimas and the other, anonymous, reviewer(s) for their contribution to © The Author(s) 2022 the peer review of this work. Peer reviewer reports are available. 1 Cellular Plasticity and Cancer Group, Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain. 2Centro de Investigación en Medicina Molecular (CIMUS), Universidad de Santiago de Compostela and Instituto de Investigaciones Sanitarias (IDIS), Santiago de Compostela, Spain. 3Oncobell Program, Bellvitge Biomedical Research Institute (IDIBELL), Barcelona, Spain. 4Instituto de Biomedicina y Biotecnología de Cantabria, Universidad de Cantabria - CSIC, Santander, Spain. 5Department of Pathology, Vall d’Hebron University Hospital, Translational Molecular Pathology, Vall d’Hebron Institute of Research (VHIR), Universitat Autònoma de Barcelona and Spanish Biomedical Research Network Centre in Oncology (CIBERONC), Barcelona, Spain. 6Cardiovascular and Metabolic Sciences, Max Delbrück Center for Molecular Medicine in the Helmholtz Association (MDC), Berlin, Germany. 7Proteomics Unit, Spanish National Cancer Research Centre (CNIO), Madrid, Spain. 8Molecular Oncology Group, Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain. 9Bioinformatics- Biostatistics Unit, Institute for Research in Biomedicine (IRB Barcelona), Barcelona Institute of Science and Technology (BIST), Barcelona, Spain. 10IMIM - Hospital del Mar Medical Research Institute, Barcelona, Spain. 11Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Spain. 12Non- colorectal Gastrointestinal Cancer Translational Research Group, Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain. 13Cancer Genomics Group, Vall d’Hebron Institute of Oncology (VHIO), Barcelona, Spain. 14Departamento de Biología Molecular y Celular, Centro Nacional de Biotecnología (CNB), CSIC, Darwin 3, 28049 Madrid, Spain. e-mail: mabad@vhio.net Nature Communications | (2022)13:6840 22