The ULK3 kinase is a determinant of keratinocyte self-renewal and tumorigenesis targeting the arginine methylome

Goruppi, Sandro; Clocchiatti, Andrea; Bottoni, Giulia; Di Cicco, Emery; Ma, Min; Tassone, Beatrice; Neel, Victor; Demehri, Shadhmer; Simon, Christian; Paolo Dotto, G.

doi:10.1038/s41467-023-36410-6

Download PDF

Article
Open access
Published: 16 February 2023

The ULK3 kinase is a determinant of keratinocyte self-renewal and tumorigenesis targeting the arginine methylome

Nature Communications volume 14, Article number: 887 (2023) Cite this article

2855 Accesses
4 Citations
2 Altmetric
Metrics details

Subjects

Squamous cell carcinoma

Abstract

Epigenetic mechanisms oversee epidermal homeostasis and oncogenesis. The identification of kinases controlling these processes has direct therapeutic implications. We show that ULK3 is a nuclear kinase with elevated expression levels in squamous cell carcinomas (SCCs) arising in multiple body sites, including skin and Head/Neck. ULK3 loss by gene silencing or deletion reduces proliferation and clonogenicity of human keratinocytes and SCC-derived cells and affects transcription impinging on stem cell-related and metabolism programs. Mechanistically, ULK3 directly binds and regulates the activity of two histone arginine methyltransferases, PRMT1 and PRMT5 (PRMT1/5), with ULK3 loss compromising PRMT1/5 chromatin association to specific genes and overall methylation of histone H4, a shared target of these enzymes. These findings are of translational significance, as downmodulating ULK3 by RNA interference or locked antisense nucleic acids (LNAs) blunts the proliferation and tumorigenic potential of SCC cells and promotes differentiation in two orthotopic models of skin cancer.

The TINCR ubiquitin-like microprotein is a tumor suppressor in squamous cell carcinoma

Article Open access 10 March 2023

miR-22 promotes stem cell traits via activating Wnt/β-catenin signaling in cutaneous squamous cell carcinoma

Article Open access 03 August 2021

miR-184 represses β-catenin and behaves as a skin tumor suppressor

Article Open access 26 February 2024

Introduction

Squamous cell carcinoma (SCCs) is among the most frequent types of human cancer and represents a significant cause of death¹. These heterogeneous tumors originate from surface epithelial cells, such as the skin, aerodigestive and urogenital tracts, in response to various noxious conditions^2,3. Two main categories of gene alterations, driver mutations and more selective mutations that impact squamous cell fate commitment and/or ensuing terminal differentiation, are commonly found in these tumors³.

NOTCH signaling is a crucial determinant of keratinocyte (KC) proliferation and differentiation³. An inverse regulatory relationship exists between NOTCH and TP63 activity, the latter promoting KC self-renewal potential and early steps of tumorigenesis by acting on other key regulators of KC proliferation, such as p21/CDKN1A and MYC^4,5,6. This interplay is similarly important in oral, esophageal⁷, bronchial⁸, and cervical⁹ epithelia.

Another critical signaling with oncogenic potential for KCs is the Hedgehog (Hh)/GLI pathway^10,11. Mammalian cells express three GLI (GLI1-3) transcription factors (TFs), with GLI2 having a role in KC oncogenic conversion, not only in basal cell carcinomas (BCCs) but also SCCs^11,12. While increased Hh signaling by patched (PTCH) and smoothened (SMO) gene mutations were causally implicated in BCCs¹³, the mechanisms of GLI2 activation in the SCCs remain to be clarified.

The Unc51-like kinase 3 (ULK3) is a poorly characterized member of the ULK1-4 kinase family proposed to fulfill in mammalian cells the same central role as the Drosophila fused kinase towards GLI2 activation^14,15. This kinase has been mainly studied for its cytoplasmic functions in breast cancer and muscle cells^14,16,17 and implicated in the endosomal sorting machinery by delaying the abscission of midbodies under stress conditions in HeLa cells¹⁷. We and others have shown that ULK3 physically associates with the GLI2 protein and enhances its transcriptional activity, as indicated by the expression of its direct target gene, GLI1, and GLI2 itself^15,18,19. In the stroma of tumors, including SCCs, ULK3 is critical for the conversion of normal fibroblasts to activated cancer-associated fibroblasts¹⁸. By contrast, the function of ULK3 in the epithelial compartment of tumors is unknown.

“Transcriptional addiction” of cancer cells can result from altered cross-talk between ubiquitous and cell type-specific TFs²⁰. From this perspective, TFs can be divided into three functional classes²⁰: determinants of cell fate (such as TP63), integrators of various developmental and cell signaling pathways (such as NOTCH/CSL and GLI) and general regulators of proliferation (such as MYC, AP1, and p53). The TFs of these classes all bind and regulate super-enhancers (SEs), open chromatin regions with a high H3K27 acetylation density²¹. Many chromatin writers, including the histone acetyltransferase and transcriptional coactivator EP300, are recruited to such regions via direct interaction with TFs such as TP63 and MYC^22,23. Dramatic changes in SE organization that enhance binding sites for a cohort of TFs, including TP63, have been reported in SCCs²⁴ and are a hallmark of cancer²⁰.

Altered metabolism is also a hallmark of cancer growth. Several TFs, such as c-MYC and FOXM1, oversee cellular glycolytic metabolism by directly regulating the expression of rate-limiting enzymes and transporters, such as PKM, LDHA, PDKs and GLUT1^25,26,27. Lung, Head/Neck, and cutaneous SCCs utilize glucose metabolism to meet their anabolic/catabolic needs for rapid growth and can adapt by using alternative amino acids such as glutamine as a source of energy to circumvent the inhibition of glycolysis^28,29,30. This highlights the need to identify druggable targets and develop metabolism-targeting therapies for cancer.

The arginine methylation of protein residues is key for many cellular processes, including epigenetic modifications, and affects the control of proliferation, stem cell potential, and carcinogenesis³¹. There are two types of protein arginine methyltransferases (PRMTs), those that catalyze asymmetric (type I) versus symmetric (type II) dimethylation. PRMT1 is the best-studied type I enzyme and mainly exhibits transcription-activating functions, while PRMT5 is the best-studied type II enzyme, and is endowed mostly with transcription-repressive functions³². Despite reported opposite roles in some contexts, PRMT1 and PRMT5 have a shared transcription regulatory function as part of the arginine methylome complex associated with DNA³³. These PRMTs regulate the epigenetic modification of histone four arginine 3 (H4R3), which can impact the recruitment of several writers, including EP300, to chromatin^34,35. PRMT1 and PRMT5 are downregulated in differentiating keratinocytes, and sustained PRMT1 expression was found to suppress the differentiation of these cells³⁶.

We report here that ULK3 is a determinant of the self-renewal and oncogenic potential of human primary keratinocytes (HKCs) and SCC cells, a nuclear kinase that impinges on TP63, GLI, and metabolic transcription regulatory pathways. ULK3 binds directly and controls the function of PRMT1 and PRMT5 in chromatin modification and gene expression. The targeting of this nuclear kinase can represent a strategy for treating cancers, specifically SCCs.

Results

Clinical SCC samples and SCC-derived cell lines have elevated ULK3 levels

A dynamic KC-specific network of epigenetic modifications and TFs are involved in squamous cell fate determination and oncogenesis^2,3. Identifying and targeting nuclear kinases in this network constitute a novel strategy for cancer therapy³⁷. Analysis of large-scale transcriptomic datasets showed consistently higher levels of ULK3 gene expression in SCCs arising in multiple body sites (Fig. 1a). These findings were extended by immunofluorescence analysis of tissue arrays of skin, head and neck and cervical cancer SCCs, which showed consistent and statistically significant increases in ULK3 levels in tumor samples versus normal or nonaffected adjacent tissues (Fig. 1b). A more detailed immunofluorescence analysis of cutaneous SCCs showed prominent ULK3 nuclear expression in a large fraction of skin SCC cells with little expression in keratinocytes of normal skin (Fig. 1c and Supplementary Fig. 1a). ULK3 upregulation in SCCs was coincidental with the expression of the Ki67 or TP63 proliferation markers, as determined by double immunofluorescence with the corresponding antibodies (Fig. 1d). Actinic keratoses (AK) are premalignant keratinocyte lesions associated with an altered squamous cell differentiation program and expansion of cells of the proliferative compartment³⁸. As in SCCs, immunofluorescence analysis of these lesions showed an accumulation of nuclear ULK3 levels (Fig. 1e and Supplementary Fig. 1b), with co-localization with TP63 (Fig. 1f and Supplementary Fig. 1c).

**Fig. 1: ULK3 is upregulated in clinical SCC samples and SCC cells.**

We further evaluated ULK3 expression in a panel of SCC cell lines derived from oral and skin lesions and compared them to those in several strains of primary HKCs. Among the four ULK family members, ULK3 was the only one that was consistently upregulated (Fig. 1g and Supplementary Fig. 1d). ULK3 protein levels were also significantly upregulated in SCC cells versus primary HKCs, as assessed by immunoblot assays with two different anti ULK3 antibodies (Fig. 1h).

Thus, elevated ULK3 expression is a consistent hallmark of SCCs arising at various body sites, which is also observed in premalignant AK lesions and maintained in cultured SCC cells relative to normal HKCs.

ULK3 downregulation or inhibition suppresses the proliferative potential of HKCs and SCC cells

To assess the role of ULK3 in HKCs and SCC cells, we started by infecting these cells with two ULK3-silencing lentiviruses versus an empty control vector. Effective downregulation of ULK3 expression in both HKCs and SCC cells (Supplementary Fig. 2a, b) resulted in significant suppression of cell proliferation (Fig. 2a) with increased fractions of apoptotic and necrotic cells (Fig. 2b). The colony-forming capability, as a well-established stem/self-renewal assay³⁹, of HKCs, of skin and Head/Neck SCC cells and cervical cancer HeLa cells was also significantly reduced by shRNA-mediated ULK3 silencing (Fig. 2c and Supplementary Fig. 2c).

**Fig. 2: The proliferative potential of HKCs and SCC cells is suppressed by ULK3 downmodulation.**

As an alternative approach, we resorted to ULK3 gene deletion by CRISPR/Cas9 technology⁴⁰. To eliminate the issue of individual clone variability and possible compensatory mutations resulting from ULK3 loss, we analyzed pooled polyclonal cell populations shortly selected after infection with lentiviruses co-expressing Cas9 together with ULK3-targeting guide RNAs (gRNAs). Infection of SCC13 cells with lentiviruses expressing three different ULK3 gRNAs resulted in the effective deletion of the gene, as assessed by Surveyor assays⁴¹, with PCR amplification of the ULK3 gene-targeted regions followed by nuclease digestion of mismatched heteroduplexes (Fig. 2d). Deletion of the ULK3 gene was accompanied by loss of ULK3 protein expression (Fig. 2e) and, as shown in Fig. 2f, a marked decrease of SCC cells colony-forming capability. A similar reduction of clonogenicity and proliferative activity was observed after treatment of SCC cells with SU6668, an inhibitor of ULK3 kinase activity¹⁹ (Supplementary Fig. 2d–f).

Molecularly, ULK3 gene silencing resulted in the strong downregulation of TP63, a positive determinant of HKC proliferation and self-renewal⁴² (Fig. 3a, b), with a similar reduction of TP63 levels occurring in SCC13 cells treated with the SU6668 inhibitor⁴³ (Supplementary Fig. 3). In agreement with its reported function in other cellular systems^16,18, ULK3 loss also suppressed expression of the GLI1/2 transcription factors, which play a positive role not only in BCCs but also SCCs¹² (Fig. 3c). Conversely, ULK3 silencing resulted in the upregulation of keratin 1 and 10 (KRT1, KRT10), involucrin (IVL) and filaggrin (FLG) differentiation markers (Fig. 3d, e) and induction of CDKN1A/p21, a key mediator of differentiation-associated cell cycle withdrawal⁴⁴ (Fig. 3f, g).

**Fig. 3: ULK3 silencing downmodulates proliferation/stem markers while upregulating differentiation genes and cell cycle inhibitor p21.**

To address the consequences of increased ULK3 expression, we infected SCC cells with an ULK3 overexpressing lentivirus versus empty vector control. Increased ULK3 expression significantly enhanced the colony-forming capability of SCC cells with a concomitant enhancement of sphere formation as an alternative assay of cell expansion potential³⁹ (Fig. 3h, i), which were paralleled by up-regulated expression of pro-proliferation genes such as c-MYC and GLI2 (Fig. 3j).

Thus, ULK3 is a key determinant of HKC and SCC cell proliferative potential that affects other critical regulators of the growth/differentiation of these cells.

ULK3 controls gene expression and cellular metabolism

For further mechanistic insights, we performed a global transcriptomic analysis of three HKC strains in which the ULK3 gene was silenced by infection with two different lentiviruses versus empty vector control. As shown in Fig. 4a, several hundred genes were up or downmodulated by ULK3 silencing (GSE183084). Gene set enrichment analysis (GSEA)⁴⁵ showed a highly significant inverse correlation between the changes in gene expression elicited by ULK3 silencing and gene signatures related to poorly versus highly differentiated Head/Neck SCCs, HKCs with silencing of the ESR2 gene, a positive determinant of keratinocyte differentiation⁴⁶, loss of TP53, and upregulation of the pro-tumorigenic TP63, Hh/GLI and MYC signaling pathways (Fig. 4b). There was also a significant negative association between ULK3 expression and metabolism-related gene signatures, specifically those related to cellular glycolysis and specific amino acid and nucleotide metabolism (Fig. 4b).

A similar transcriptomic analysis was performed with SCC13 cells plus/minus ULK3 gene silencing by infection with two shRNA lentiviral vectors (GSE1830850). Gene ontology analysis showed that the genes whose expression was most significantly affected by ULK3 silencing in both HKCs and SCC cells are involved in processes related to the cell cycle, protein translation, and metabolism (Fig. 4c). More specifically, ULK3 silencing caused the downregulation in all cells of rate-limiting enzymes in glycolysis and glucose transport, such as PKM, LDHA, and SLC2A1, as well as a key TF involved in the control of metabolic genes, FOXM1^26,47 and the upregulation of PDK2, PDK3 and TIGAR, key repressors of the glycolytic process^48,49 (Fig. 4d, e). Compromised glycolysis in cancer cells is often compensated by the increased use of amino acids as an energy source, by enhanced glutaminolysis²⁹. Consistent with the suppression of glycolytic enzymes, we found that ULK3 silencing in both HKCs and SCC13 cells resulted in the upregulation of genes coding for a glutamine-to-glutamate converting enzyme (glutaminase; GLS1) and glutamine-related transporters (SLC1A5 and SLC38A6) (Fig. 4d). Such modulation of metabolism-related genes upon ULK3 silencing was independently confirmed by RT-qPCR assays and immunoblot analysis of SCC cells (Fig. 4e, f).

To directly assess whether ULK3 downregulation affects cellular metabolism, we performed a steady-state metabolomic analysis of SCC13 cells infected with two ULK3-silencing lentiviruses versus empty vector control. We found that ULK3 downregulation modified the levels of several metabolites involved in key cellular processes, including glycolysis, and the metabolism of lipids, nucleotides, and several amino acids, with increased levels of glutamate and glutathione, two key products of glutaminolysis⁵⁰ (Fig. 4g and Supplementary Fig. 4). Metabolite set enrichment analysis (MSEA) allows the identification of enriched metabolic pathways associated with a specific set of metabolites⁵¹. MSEA of SCC13 cells plus/minus ULK3 silencing showed the most significant metabolic changes in pathways related to the metabolism of several amino acids, such as arginine/proline, homocysteine degradation and glycine/serine, methionine, and, consistently with our analysis above, glutathione and glutamate metabolism (Fig. 4h).

Thus, ULK3 downregulation impacts the expression of genes relevant for HKC self-renewal, proliferation, and metabolism, with direct metabolic measurements consistent with decreased glycolysis and increased glutamine consumption.

ULK3 controls the chromatin configuration of key cell regulatory genes

The prevalent nuclear localization of ULK3 in SCC cells (as shown in Fig. 1c, d and Supplementary Fig. 2b) suggested that it may associate with chromatin. To assess this possibility, we performed chromatin immunoprecipitation combined with DNA sequencing (ChIP-seq) analysis of SCC13 cells with anti-ULK3 antibodies. Global distribution of ULK3 bound peaks, determined using MACS2⁵² and PeakAnalyzer⁵³, relative to the transcription start site (TSS) of genes in SCC13 cells showed that ULK3 associates within ±500 bp from the transcriptional start site (TSSs) of a restricted number of genes of functional significance (334) (Fig. 5a, Supplementary Fig. 5a, and GSE183933). For further mechanistic insights, we performed a ChIP-seq analysis of SCC13 cells plus/minus ULK3 silencing with antibodies against histone 3 lysine 27 acetylation (H3K27ac), a marker of active chromatin^20,54(GSE 183933). H3K27ac peak intensity was markedly reduced in cells with silenced ULK3, in areas at various distances (±1500 bp) from the TSS of genes, including most of those bound by ULK3 (Fig. 5b, c; Supplementary Fig. 5b).

**Fig. 5: ULK3 associates to chromatin regulatory regions of genes of functional significance.**

Using Integrative Genomics viewer (IGV, Broad Institute) we aligned the ULK3 and H3K27ac binding peaks of two genes with key metabolic regulatory functions, pyruvate kinase M (PKM) and lactate dehydrogenase A (LDHA)⁵⁵, and a key transcriptional regulator of this process, the forkhead box M1 TF (FOXM1)^26,27,47. As shown in Fig. 5d–f, upper panels, there was a broad coincidence of ULK3 and H3K27ac binding peaks at the TSS region of all three genes, with H3K27ac binding peaks being lost in cells with ULK3 gene silencing. A similar pattern of ULK3 and H3K27ac binding was found at the TSS of the EP300 gene, a known regulator of active chromatin and metabolism in HKCs and SCC cells (⁵⁶) (Supplementary Fig. 5c).

The results were validated by independent chromatin immune precipitation (ChIP) analysis of SCC13 cells plus/minus ULK3 silencing with two shRNA vectors. As expected, ULK3 binding to two different regions of these genes, the same as determined by ChIP-seq analysis, was markedly reduced in cells with ULK3 silencing, with a parallel decrease of H3K27ac levels (Fig. 5d–f, middle and lower panels). The results were extended by ChIP with antibodies against histone 3 lysine 9 acetylation (H3K9ac) and RNA polymerase (Pol II), as markers of active promoters^54,57, which were also reduced in cells with silenced ULK3 (Fig. 5d–f, lower panels). The decrease of H3K27ac, H3K9ac, and Pol II was specific for genes downmodulated by ULK3 silencing, as it was observed with a housekeeping TUBULIN A1A gene, which is not bound by ULK3 nor regulated upon ULK3 silencing in SCC13 cells (Supplementary Fig. 5d)

Thus, ULK3 oversees the chromatin configuration (open versus closed) of the majority of genes that it regulates, directly associating to the regulatory regions of genes with a critical role in metabolisms and transcriptional control.

ULK3 associates with the arginine methyltransferases PRMT1 and PRMT5

The above transcriptomic and ChIP-seq analysis indicated that only some genes are under direct ULK3 control. In contrast, the transcription and chromatin configuration of others are likely affected by ULK3 silencing with a different mechanism. The histone arginine methylase PRMT1 plays a key role in the balance between keratinocyte self-renewal and differentiation³⁶. By GSEA, the gene signature resulting from the silencing of the PRMT1 gene in HKCs³⁶ was among the most positively correlated with the transcriptomic profile elicited by ULK3 silencing in the same cells, with individual genes of functional importance similarly modulated by ULK3 and PRMT1 downmodulation (Fig. 6a and Supplementary Fig. 6a, b).

**Fig. 6: ULK3 associates with the arginine methylases PRMT1 and PRMT5.**

An attractive possibility is that the overlapping transcriptional consequences of ULK3 and PRMT1 gene silencing reflect a physical association between the two proteins, which may extend to PRMT5, another PRMT family member that catalyzes symmetric rather than asymmetric arginine methylation with a functional interplay with PRMT1^58,59. Like PRMT1, PRMT5 has also been reported to play a role in keratinocyte differentiation^36,60. Proximity ligation assays (PLA) detect endogenous intramolecular interactions, which are visualized by fluorescence puncta resulting from the juxtaposition of antibodies against associated proteins^18,61,62. In two different SCC cell lines, Cal27 and SCC13 cells, PLA with anti ULK3 and either anti PRMT1 or anti PRMT5 antibodies generated a large number of nuclear puncta (Fig. 6b). The ULK3 -PRMT1 and -PRMT5 association was independently confirmed by coimmunoprecipitation and immunoblotting with antibodies against the same proteins (Fig. 6c). We further investigated the interaction by coimmunoprecipitation assays in HEK293 cells transfected with expression vectors for wild-type ULK3, ULK3 with two different inactivating point mutations, (K44H), mapping within the ATP-binding site of the kinase domain¹⁷, and (K139R), mapping in the donor/acceptor region of the kinase transferase activity^16,17, and an ULK3 truncation lacking the C-terminal region (∆C). As shown in Fig. 6d, binding of endogenous PRMT1 and PRMT5 was equally detected to wild-type and catalytically inactive ULK3, but not to ULK3 with the C-terminal deletion, showing that the interaction is independent of ULK3 kinase activity and maps outside the catalytic domain. Both wild-type and kinase-dead ULK3 efficiently coimmunoprecipitated endogenous PRMT1/5 also in HeLa cells, further validating the results (Fig. 6e).

To assess whether endogenous PRMT1 or PRMT5 phosphorylation levels depend on ULK3, we performed PLA with antibodies recognizing PRMT1 or PRMT5 and against phospho-serine/threonine, an approach previously used to assess changes in PRMT1 phosphorylation³⁶. As shown in Fig. 6f, positive PLA puncta were found in control cells for both PRMT1 and PRMT5, which were significantly reduced upon ULK3 silencing. As a complementary approach, we took advantage of a chromatography resin that preferentially binds phosphoproteins and is used for microscale phosphoprotein purification⁶³. Expression vectors for tagged PRMT1 or PRMT5 were transfected alone or in combination with one for ULK3, and the cellular lysates were bound to the resin. The eluted fractions were separated by SDS-PAGE and analyzed by western blotting using FLAG antibodies. Indicative of enhanced phosphorylation, the amounts of PRMT1 and PRMT5 in the resin-bound fractions were significantly increased in cells with ULK3 overexpression (Fig. 6g).

Thus, ULK3 associates with the PRMT1 and PRMT5 arginine methyltransferases and can determine their phosphorylation state.

ULK3 directly binds PRMT1 and PRMT5 and increases their methylase activity

An attractive possibility is that ULK3 can interact and phosphorylate PRMT1/5 directly. To assess this possibility, purified recombinant His-tagged ULK3 was admixed with FLAG-tagged PRMT1 or PRMT5 proteins, followed by immunoprecipitation with the respective anti tag antibodies. Both PRMT1 and PRMT5 were effectively recovered after the pulldown of ULK3 and the latter was recovered after the PRMT1 and PRMT5 pulldown (Fig. 7a, b). Notably, the ULK3 - PRMT1/5 interaction required a native protein conformation as it was lost when the PRMTs were heat denatured before performing the binding assay (Fig. 7c).

**Fig. 7: ULK3 binds directly to PRMT1 and PRMT5 and controls their activity.**

To assess whether PRMT1/5 can be a direct substrate of ULK3 activity, we incubated the recombinant proteins in the presence of ATP followed by Phos-TAG SDS-polyacrylamide gel electrophoresis, which detects phosphorylated proteins as retarded band(s)⁶⁴. A large fraction of PRMT1 and PRMT5 was found in slower migrating forms upon incubation with ULK3, which were not detected when the in vitro incubation assay was performed in the presence of the ULK3 inhibitor SU6668⁴³ (Fig. 7d and Supplementary Fig. 7a). The gel retardation form was abolished when PRMT1 was heat-denatured before the kinase reaction, pointing to the importance of a native protein conformation (Supplementary Fig. 7b). No gel delayed forms were detected in parallel assays with recombinant GFP protein, indicating the specificity of the reaction (Fig. 7e).

An important question was whether ULK3 could directly affect the activity of PRMT1 and/or PRMT5. Accordingly, we preincubated ULK3 with increasing amounts of PRMT1 or PRMT5 proteins, followed by the analysis of their enzymatic activity towards histone H4. Arginine 3 of histone 4 (H4R3) is a substrate for both PRMT1-driven asymmetric (H4R3DA) and PRMT5-driven symmetric (H4R3DS) dimethylation activity³². Dot blot analysis using specific anti -H4R3DS and -H4R3DA antibodies showed a significant increase of the methylated products at the lower concentrations of the two enzymes when pre-incubated with ULK3, indicative of increased activity (Fig. 7f, g). Immunoblot analysis of ULK3 immunoprecipitated after in vitro preincubation assays with antibodies specific for asymmetric and symmetric dimethylated arginine showed that the kinase is in turn methylated by the PRMT1 but not PRMT5 enzyme (Supplementary Fig. 7c), with a selectivity of modification that was also observed after co-expression assays in HEK293 cells (Supplementary Fig. 7d), with functional implications that will have to be separately pursued.

Thus, ULK3 can bind and phosphorylate PRMT1 and PRMT5 directly, affecting their activity.

ULK3 is required for PRMT1 and PRMT5 recruitment to chromatin and H4 arginine 3 dimethylation

PRMT1³⁶ and PRMT5⁶⁵ are key for the self-renewal of HKCs and SCC cells^66,67. As assessed by immunoblotting, total protein levels of PRMT1 and PRMT5 in SCC13 cells were not significantly affected by ULK3 gene silencing (Fig. 8a). Cell fractionation followed by immunoblotting confirmed that the ULK3 protein is mainly nuclear in SCC13 cells and showed that the nuclear localization of PRMT1 and PRMT5 was not affected by ULK3 silencing (Fig. 8b), with results confirmed by immunofluorescence analysis (Fig. 8c).

**Fig. 8: ULK3 loss impairs PRMT1/5 chromatin association and function.**

To assess whether the association of PRMTs with chromatin is affected by ULK3 loss, SCC13 cells plus/minus ULK3 silencing were processed for ChIP with anti -PRMT1 and -PRMT5 antibodies, followed by qPCR amplification of different regions of the PKM and LDHA metabolic genes, which are bound by ULK3. We found two main regions of PRMT1 and PRMT5 binding to the LDHA and PKM genes, which significantly reduced in cells with silenced ULK3 (Fig. 8d).

To assess whether the results reflected a more generalized impact of ULK3 loss, total chromatin was immunoprecipitated with anti-histone H3 antibodies followed by immunoblotting with antibodies against PRMT1, PRMT5, and histone H3. As shown in Fig. 8e the association of both PRMTs with chromatin was strongly reduced in ULK3-silenced cells versus controls.

Histone H4 arginine 3 (H4R3) is dimethylated by PRMT1 and PRMT5³². The decreased chromatin association of PRMTs resulting from ULK3 gene silencing was associated with the downregulation of both symmetric and asymmetric forms of H4R3 dimethylation, as assessed by western blotting and immunofluorescence analysis with corresponding specific antibodies of SCC and HeLa cells with silenced ULK3 versus control. (Fig. 8f, g and Supplementary Fig. 8a–d). Consistent with the knockdown results, ULK3 loss by CRISPR/Cas9 deletion resulted in a similar decrease in both forms of H4R3 dimethylation (Fig. 8h and Supplementary Fig. 8e).

Thus, ULK3 is required for the association to chromatin of PRMT1/5 and for dimethylation of their histone H4 substrate.

ULK3 is a target of translational significance to suppress SCC cells proliferation and oncogenic potential

An important question was whether targeting ULK3 could effectively suppress SCC formation. In the first set of studies, we utilized orthotopic mouse ear injection assays of fluorescently labeled cancer cells, which allow in vivo imaging of lesions^18,68. Two different SCC cell lines, the mildly transformed SCC13 cell line⁶⁹ and the more aggressive SCCO28 cell line⁷⁰, were stably transduced with a GFP-expressing lentivirus, followed by infection with two ULK3-silencing lentiviruses versus empty vector control virus before ear injections. While control SCC cells formed rapidly expanding tumors, cells in which ULK3 was downregulated gave rise to tumors of smaller size and greatly reduced number (Fig. 9a, b).

**Fig. 9: ULK3 downregulation suppresses SCC growth in vivo.**

As an alternative approach, we employed an intradermal injection assay of SCC cells in Matrigel, which allows the formation of easily retrievable localized nodules⁶⁸. Even in this case, tumors formed by SCC cells with ULK3 silencing were significantly smaller than those derived from control cells, with tumor growth being suppressed entirely in some animals (Fig. 9c). The decreased tumorigenicity was mirrored by a reduction in the Ki67 proliferative index and increased differentiation, as indicated by the large increase of cancer cells with the KRT10 marker expression (Fig. 9d, e). Consistent with the findings with cultured cells, tumors formed by SCC13 cells with silenced ULK3 had significantly reduced levels of TP63 expression and H4R3 dimethylation relative to control lesions (Fig. 9f, g).

An exciting tool with translational potential to suppress the expression of target genes is provided by locked nucleic acids (LNAs), oligonucleotide analogs of high stability amenable to in vivo delivery^71,72. SCC13 cells were transfected with three LNAs designed to target different regions of the ULK3 mRNAs (#1–3) in parallel with scrambled controls (SCR). RT-PCR and immunoblot analysis showed effective downmodulation of ULK3 expression by all three LNAs, with only one of them (#1) also causing a slight reduction of the other ULK family members (Fig. 10a, b). Consistent with the gene silencing and CRISPR-deletion results, the colony formation of SCC cells was significantly decreased by administration of these LNAs (Fig. 10c), in parallel with a significant reduction of their sphere-forming capability, another well-established assay of cancer stem cell potential (Fig. 10d).

**Fig. 10: ULK3 targeting with locked nucleic acid (LNAs) suppresses SCC self-renewal and tumorigenicity.**

ULK3 has been reported to delay cytokinesis of cells under stress conditions at the G2/M phase of the cell cycle, by phosphorylating a component of the ESCRT-III cytosolic complex¹⁷. We assessed the impact of ULK3 loss on cell cycle distribution by analyzing SCC cells expressing a dual color Fluorescence Ubiquitin Cell Cycle Indicator system (FUCCI)⁷³. Transfection of these cells with the anti-ULK3 LNAs led to an accumulation of cells in the G1/S phases of the cell cycle (expressing the RFP reporter), with little or no effect on G2/M (expressing the GFP reporter) (Fig. 10e). At the same time, ULK3 silencing did not result in detectable nuclear alterations and/or multinucleated cells as would be expected in case of a mitotic block (Fig. 10f).

LNA administration also recapitulated the in vivo consequences of stable ULK3 gene silencing. SCC13 cells transfected with two ULK3-targeting LNAs (#2 or #3) 24 h before intradermal injection into mice formed much smaller tumors than controls (Fig. 10g) with a reduction of Ki67 proliferative index and TP63-positive cells and enhancement of differentiation (KRT10-positive areas) (Fig. 10h, i).

Therefore, targeting ULK3 is of translational value to suppress the oncogenicity of SCC cells impacting the balance between growth and differentiation.

Discussion

The proliferation and differentiation of squamous epithelial cells are tightly regulated processes, with genetic and epigenetic alterations in regulatory networks often leading to a transformed phenotype^2,3. We show here that ULK3, a kinase we previously reported to be key for the activation of cancer-associated fibroblasts¹⁸, plays an equally important role in the epithelial compartment, as a transcription regulator of self-renewal and oncogenic potential. In skin SCCs, ULK3 is preeminently expressed in the nuclear compartment, overlapping with epidermal areas of sustained cell proliferation and TP63 expression. Pronounced nuclear accumulation of ULK3 was also found in actinic keratosis lesions, a pathological condition of major clinical significance that can evolve into SCC^3,74, pointing to a role for this kinase in the early events of the keratinocyte transformation process. Large clinical data sets analysis showed that ULK3 expression levels are also elevated in lung, head/neck, and cervical SCCs. The findings are likely of functional significance, as ULK3 targeting decreased HKC and SCC proliferative potential, promoting differentiation and limiting tumor growth.

Our findings establish ULK3 as a key regulator of the transcription process underlying the balance between growth and differentiation of both HKCs and SCC cells with a concomitant impact on metabolism. ULK3 gene silencing induced the cell cycle inhibitor CDKN1A, a mediator of differentiation-associated cell cycle withdrawal⁴⁴, and several key markers of keratinocyte terminal differentiation process, KRT1, KRT10, IVL, and FLG^6,46,75. Gene set enrichment analysis (GSEA) of transcriptional profiles of multiple HKC strains plus/minus ULK3 silencing showed a strong negative correlation between ULK3 expression and gene signatures involved in KC proliferation and oncogenesis, such as increased ∆Np63, Hh/GLI and MYC signaling, as well as gene signatures related to SCC aggressiveness. The transcription factor TP63 plays a key role in the transition from simple to stratified epithelia and in maintaining skin stem cell populations²². Like TP63, GLI2 encodes for a transcription factor with a positive role in the oncogenic conversion of keratinocytes for both BCCs and SCCs^12,76. ULK3 has a central role in GLI1/2 activation by SHH or TGFβ in human adipose stem cells⁴³ and the elevated ULK3 levels we found in SCCs might be the reason for GLI activation in SCCs.

Downregulation of these key programs can explain the negative consequences of loss of ULK3 on self-renewal potential of HKCs and SCCs, as assessed by the well-established clonogenicity and sphere-forming assays⁷⁷. Consistent results were obtained by shRNA-mediated ULK3 gene silencing, CRISPR-deletion, and treatment with a pharmacological inhibitor of ULK3 activity. Abrogation of ULK3 expression by LNAs, a complementary approach of translational significance⁷², exerted a similar impact on SCC proliferative potential, with an accumulation of cells in the G1 phase of the cell cycle, and suppressing tumorigenesis, with a parallel suppression of TP63 expression and induction of differentiation.

Treatment of SCC cells with mTOR or PI3K inhibitors - as already tested in clinical trials - causes suppression of glycolysis and a concomitant increase of glutamine utilization as an alternative source of energy²⁹. A similar compensatory mechanism occurs in SCC cells upon ULK3 gene silencing, which results in the suppression of genes with a key role in glycolysis (PKM, PDK2/3, TIGAR, and LDHA) and the up-regulation of genes involved in glutamine uptake and utilization (GLS1 and SLC1A5). This gene regulation was consistent with metabolomics analysis showing, upon ULK3 silencing, an enrichment of several amino acid metabolic pathways, including glutamate, glutathione, and glycine/serine metabolism. Thus, as with inhibitors already considered for SCC treatment²⁹, targeting ULK3 activity/levels may benefit from a combinatory use of glutaminolysis inhibitors.

While most previous studies focused on ULK3 cytosolic function^14,16,17,18, we have found that ULK3 in HKCs and SCC cells is mostly localized to nuclei and, by combined ChIP-seq and ChIP assays, binds to a specific set of genes of functional significance, including several with a key role in metabolism and transcription. Notably, most of the ULK3-bound and/or regulated genes exhibited a striking loss of H3K27 acetylation upon ULK3 silencing pointing to its role in the maintenance of open chromatin configuration. Other chromatin regulatory mechanisms are likely to be also affected by ULK3 silencing, as indicated by the fact that H3K27 acetylation is not compromised by the ULK3 loss at all ULK3-bound genes.

Considering the impact on the larger number of genes modulated upon ULK3 silencing compared with the genes directly bound, it poised us to identify additional mechanisms by which ULK3 impinges on transcription. The recruitment of epigenetic modifiers to chromatin regulates chromatin activation states (activated or repressed)^20,78. We identified the histone arginine methylases PRMT 1 and 5 (PRMT1/5), which are the most represented in HKCs and contribute to transformation, proliferation, and survival in numerous cancers⁷⁹. PRMT1 is predominantly a transcriptional coactivator^32,80, while PRMT5 frequently acts as a corepressor^32,81. However, opposite functions have been reported for PRMT1⁸² and PRMT5⁸³. Thus, PRMT1/5 function is likely directed by their interacting partners and is, to some extent, cell-type specific^36,78.

Both PRMTs converge to regulate transcription as part of a methylome complex³³ and a functional redundancy exists among PRMT1/5, as PRMT1 loss or pharmacologic inhibition causes substrate “scavenging” by PRMT5^84,85. In HKCs, PRMT1 interacts with CSNK1A to sustain cell proliferation and repress premature differentiation³⁶ while PRMT5 suppresses PKCδ/p38-induced involucrin expression⁶⁰.

By multiple approaches, we showed that PRMT1 and PRMT5 directly associate with the Ct-terminal region of ULK3, that such binding requires PRMTs native conformation and is independent of active ULK3 kinase activity since it also occurs with kinase-defective point mutants. Functionally, we demonstrated that the enzymatic activity of PRMT1/5 is enhanced upon phosphorylation in vitro by ULK3. The findings with our GSEA which revealed that the loss of ULK3 in HKCs phenocopied the changes in gene expression induced by PRMT1 silencing, further supported the functional convergence.

The exact consequences of mutually exclusive PRMT5-driven symmetric or PRMT1-driven asymmetric arginine dimethylation of the same arginine residues are poorly understood³². Both PRMTs function in the epigenetic modification of the arginine 3 on histone 4 (H4R3), which can impact the recruitment of several writers, including EP300, to the chromatin^34,35,86. ULK3 downregulation, or gene deletion, blunted such epigenetic modifications in skin and oral SCCs, HeLa cells, and HEK293 cells, suggesting a specific and conserved mechanism. Mechanistically, we found that ULK3 loss decreases the recruitment of PRMT1/5 to the promoter areas of specific genes and, more globally, to chromatin.

It should be noted that so far, neither ULK3 nor PRMT1/5 have been identified in major large-scale functional genomic screening assays for fitness or dependency genes in SCC cells. Although these screens are instrumental in identifying targets, they critically depend on the depth and efficiency of the targeting gene libraries and, as a consequence, other well-known genes involved in the control of SCC development, such as NOTCH1 and CDK6, have often been missed^87,88.

Overall, the present work identifies a control mechanism involving the serine/threonine kinase ULK3 and the PRMT chromatin complex, which impinges on key transcriptional programs and the metabolism of keratinocyte and SCC cells. LNAs can effectively downmodulate ULK3 expression and exert similar effects on self-renewal and tumorigenicity of SCC cells as ULK3 gene silencing. This kinase represents an attractive target for the prevention and treatment of SCC lesions that would allow the simultaneous suppression of cancer cells and adjacent cancer-associated fibroblasts^89,90.

Methods

Mouse, cells, and human samples

All the animal procedures were performed as approved by the Institutional Animal Care and Use Committee of Massachusetts General Hospital (MGH), Boston, MA, USA (MGH # 2004N000170).

Normal primary human keratinocytes (HKCs) were prepared from discarded human samples from abdominoplasty at the Cutaneous Biology Research Center (MGH, Boston, MA, USA), as approved by Institutional Review Board protocol (IRB # 2018P003156) or were previously obtained with approval in refs. ^46,68. All HKC strains were identified by a progressive number or two letters identifying the operators. Normal human keratinocytes and patient-derived squamous cell carcinoma (SCC) cells were cultured as previously described in refs. ^46,68.

Human skin samples of actinic keratoses (AK), skin SCCs and normal matched controls of nearby non-affected areas from Mohs surgeries of skin cancers (excess of areas accumulating at the end of the incision after the wound closure) were obtained at the Department of Dermatology (MGH, Boston, MA, USA) as de-identified discarded parts not needed for diagnosis as approved by the IRB protocol (# 2018P003156). A progressive number identified all samples. Both male and female skins were used, as available.

Dr. Rocco, (MGH, Boston, Massachusetts, USA) provided oral SCCO28 cells. Dr. Tolstonog (Centre Hospitalo-Universitaire Vaudois, Lausanne, Switzerland) provided Cal27, Cal33, and FaDu oral SCC cells⁶⁸. Dr. Rheinwald (Brigham and Women’s Hospital, Boston, Massachusetts, USA) provided skin SCC cells (SCC12 and SCC13). All cells were routinely tested for the absence of mycoplasma.

Tissue arrays

Commercially available tissue arrays with normal or adjacent matched tissues as controls of skin squamous cell carcinoma (SK483), head and neck tumors (HN801c) and cervical squamous carcinomas (CR484) were from US Biomax. SK483 contains skin squamous cell carcinoma tissues with normal tissues as control, 48 cases-48 cores. HN801c contains head and neck squamous cell carcinoma tissues, 40 cases-80 cores. CR484 contains cervix squamous cell carcinoma tissues with matched adjacent cervix tissue, 16 cases-48 cores. Slides were processed with routine histology procedures as in refs. ^18,68 using rabbit anti ULK3 ATLAS antibody (AB_2677003, 1:100 dilution).

For signal quantification, acquired images for each color channel were imported into ImageJ software and converted into binary images. The intensity was measured using “measurement” in areas of interest. The signal was normalized to the number of cells, as detected by using a fluorescent DNA stain (4′,6-diamidino-2-phenylindole) to label nuclei. Images were prepared using Adobe Photoshop 2021, and statistical analysis was made using Prism 9 software.

Cell manipulations and assays

Conditions for culturing HKCs, SCCs, and other cell lines, RT-qPCR analysis, lentivirus production in 293 T cells and infection of cells were as in refs. ^46,62,68.

For the apoptosis assays, SCC13 and Cal27 cells infected with two separate lentiviruses for ULK3 shRNA or a control virus were seeded 2 × 10³/3 cm dish in triplicate and, after two days, incubated for 1 h with apoptosis/necrosis detection kit reagents (Abcam, cat # Ab176749). The apoptosis-detecting sensor (PS) has green fluorescence (Ex/Em = 490/525 nm) upon binding to the membrane. Loss of plasma membrane integrity after necrosis allows a membrane-impermeable 7-AAD (Ex/Em = 546/647 nm) dye to label the nucleus. Live cells were labeled with CytoCalcein Violet 450 (Ex/Em = 405/450 nm). Images of live cell cultures were obtained using an inverted fluorescent microscope Leica and quantified using ImageJ.

For clonogenicity assays, HKC, HeLa, or the indicated SCC cells (infected either with two separate ULK3 or a control shRNA-expressing virus and selected with puromycin for three days) were plated (10³/3 cm dish) in triplicate or quadruplicate/condition and cultured for six days before staining the colonies with 1% crystal violet. Images of the stained dishes were quantified using ImageJ.

For proliferation assays, we seeded an equal number of cells (10³/3 cm dish) in triplicate. We assessed the cell number of HKCs and SCC cells (plus/minus infection with shULK3- or shCTR- virus and selection for three days) on days 3 and 5 using a hemocytometer.

For CRISPR/Cas9-mediated gene deletion, all-in-one lentiviral vectors encoding both the RNA guide (gRNA) and the Cas9 protein for ULK3 gene deletion were from ABM GOOD. We infected SCC13 and HEK293 cells with three separate gRNA (#1, #2, #3) (or the combinations indicated) or a scrambled control guide vector (#SCR). After three days, we selected the cells for three additional days with 1 µg puromycin (Gibco) and then pooled the colonies before the analysis of ULK3 disruption, colony assays and western blot analysis.

The Surveyor Assay⁴¹ of SCC13 cells was made one week after the infection with lentiviruses co-expressing the Cas9 enzyme and one of the three RNA guides targeting ULK3 (gRNA ULK3 #1-3) or a scrambled control gRNA (SCR gRNA). We amplified by PCR the genomic regions of each targeted sequence (using separate oligo pairs #1-3) and tested 1 µg of amplified genomic DNA for the presence of a DNA-mismatch by incubating with a single-strand DNAse using a commercial kit Thermo Fisher (#A24372). The reactions of each targeted sequence were analyzed with a 2% agarose gel in parallel with a similar reaction made with SCR gRNA-infected cells, as a specificity control.

For ULK3 overexpression assays, Cal27 and SCC13 cells were infected with a lentiviral wild-type ULK3-HA expressing vector (ABM Good), or an empty vector, overnight then selected with 1 µg/ml of puromycin (Gibco) for three days. After the expansion of the cultures for one week 10³/3 cm dish cells were plated in triplicate or quadruplicate cultures and analyzed for colony assay formation after a week. Crystal violet-stained images of cell cultures were analyzed by ImageJ and Prism software for statistical evaluation.

For the spheroid assay, SCC13 and Cal27 cells infected with a lentiviral GFP-ULK3 wt vector, or an empty control vector (Genecopeia) were plated cells (10³/well) onto 8-well chamber slides coated with 100 μl Matrigel (BD Biosciences) in normal culture medium plus 2% Matrigel. After one week, the colonies were quantified using ImageJ. Cultures of control and ULK3-overexpressing SCC cells were seeded in parallel dishes, and the gene expression was analyzed after two days by RT-qPCR.

For locked antisense RNA (ASO LNA) transfection (IDT), we incubated overnight SCC13, SCC13 FUCCI, Cal27 or SCCO28 cells with complexes formed by mixing each LNA (10 nM final concentration) with 10 µl lipofectamine 2000 (Invitrogen) for 30 min before adding to the cell medium. We used three LNAs targeting ULK3 open reading frame and scrambled control (SCR). The following day cells were trypsinized, counted, and used for biochemical studies, sphere assays or in vivo tumor growth.

For spheroid assays of SCC13 cells, 10² cells 24 h after transfection with LNA or control oligo were seeded in quadruplicate in an 8-well culture slide coated with Matrigel and allowed to grow for seven days. We quantified the number of spheres obtained with ImageJ.

SCC13 FUCCI is a cell line stably expressing fluorescent ubiquitination-based cell cycle indicator (FUCCI) reporter, which allows the detection of G1/S (magenta fluorescence) and G2/M (green fluorescence) phases of the cell cycle⁷³. For each time point (24 h, 48 h, and 72 h) of experiment 10³ cells/3 cm dishes were seeded in triplicate. Cells were fixed with 4% paraformaldehyde in PBS and stained with DAPI to evaluate the total cell number. The fluorescence signals were acquired using a digital slide scanner (Nanozoomer S60, Hamamatsu) and quantified by ImageJ.

To determine the requirement of ULK3 kinase activity, the clonogenicity and cell proliferation assays were similarly carried out by adding to the culture medium, the day after the seeding, 10 µM of SU6668 (5-[1,2-Dihydro-2-oxo-3H-indol-3-ylidene)methyl]−2,4-dimethyl-1H-pyrrole-3-propanoic acid) an inhibitor of ULK3 kinase activity¹⁹ (Biotechne cat # 3335), or DMSO carrier (Sigma) as a negative control.

The shRNA sequences used are provided in Supplementary TABLE 1. The sequences of the oligonucleotides for ChIP and qPCR are in Supplementary TABLE 2. The sequences of siRNA and LNAs are in Supplementary TABLE 3. The Vectors and Recombinant proteins are in Supplementary TABLE 4. The detailed list of antibodies used is in Supplementary TABLE 5.

Immune detection

Western blots and immunofluorescence analysis were performed as in^46,62. Immunohistochemistry (IHC) of tumors and tissue sections was performed as in^18,46 and the images were acquired using a Leica or a Zeiss confocal microscope. For immunofluorescence, IHC and western blot analysis we used anti-ULK3 from Santa Cruz (# 137897, 1:100 dilution) and Abcam (# EPR4888, 1:100 dilution), anti-symmetrical dimethylation of arginine 3 on histone 4 from EpiGentek (# 10019-4R3DS,1:100 dilution), anti-asymmetrical dimethylation of arginine 3 on histone 4 from EpiGentek (# 10019-4R3DA, 1:100 dilution), anti-TP63 from Abcam (# 735, 1:100 dilution), anti-Ki67 from Abcam (# 15580, 1:100 dilution), anti-KRT10 from Covance (# 19054, 1:100 dilution). For western blot and immunofluorescence analysis the antibodies for PRMT1 and PRMT5 were from Cell Signaling (# 2449, 1:1000 dilution) and (# 79998, 1:1000 dilution) respectively. For western blot analysis we used Cell Signaling anti -GLS1 (# 88964, 1:1000 dilution), -PKM1 (# 7067, 1:1000 dilution) and -p21 (# 2947, 1:1000 dilution), and Abcam anti-INVOLUCRIN (#227530, 1:1000 dilution) antibodies. For Immunofluorescence analysis anti-VIMENTIN antibodies (Abcam # 20346, IF 1:100 dilution).

For the nuclear/cytosol separation, total lysates from SCC13 cells stably infected with two different shRNA targeting ULK3, or a control virus, were fractionated using the NE-PER Nuclear and Cytoplasmic Extraction Kit (Thermo Scientific #78833). As determined by a separate stained Coomassie gel, equal amounts of nuclear and cytosolic protein extracts were blotted and the membranes sequentially incubated with antibodies against ULK3, PRMT1, PRMT5, LAMIN B1 and ß-ACTIN (all 1:1000 dilution).

Protein interactions and immunoprecipitations

Proximity ligation assays⁶¹ for detecting endogenous interactions were performed using a Duo-link PLA kit (Sigma, # DUO92101) according to the manufacturer’s protocol and as in refs. ^18,62. After incubation with PLA blocking solution, the SCC cells were incubated with primary anti -ULK3 (MyBioSource # MBS9200567, 1:100 dilution) and -PRMT1 (Cell Signaling # 2449, 1:100 dilution) or -PRMT5 (Cell Signaling # 79998, 1:100 dilution) antibodies. The cells were incubated with the PLA probes, anti-rabbit PLUS, anti-mouse MINUS, washed, ligated, and amplified by rolling circle amplification. Images were obtained with a Nikon Eclipse Ti confocal microscope.

PLA analysis of PRMT1 and PRMT5 phosphorylation in HeLa cells plus/minus ULK3 silencing was similarly carried out by using anti-phospho Ser/Thr antibodies (Abcam #17464, 1:100 dilution) in combination with anti -PRMT1 Santa Cruz (# 166963, 1:100 dilution), -PRMT5 (Santa Cruz # 424245, 1:100 dilution), or nonimmune control antibodies (1:100 dilution), with the addition of phosphatase inhibitors (Thermo Fisher) in all the buffers.

We performed the immunoprecipitations of endogenous or transiently transfected proteins in SCC, HeLa, and HEK293 cells with anti -ULK3 (Santa Cruz # 517373, 1:100 dilution), -PRMT1(Cell Signaling # 2449, 1:100 dilution), -PRMT5 (Cell Signaling # 79998, 1:100 dilution), -FLAG (Sigma #F104, 1:100 dilution) and -V5 (Thermo Fisher E10, 1:100 dilution) antibodies using the same conditions as reported in refs. ^18,62.

For ChIP followed by western blot experiments, SCC13 cells infected with two ULK3 silencing lentivirus, or a control virus, were crosslinked with 1% formaldehyde and processed for ChIP using the conditions indicated in SimpleChiP assay kit (Cell Signaling # 56383, 1:100 dilution). The crosslinked chromatin from 16 × 10⁶ cells per condition was sonicated and equal amounts of total chromatin (50 µg), as assessed by spectroscopy analysis of purified DNA, were immunoprecipitated using anti-histone H3 (Cell Signaling # 4620, 1:100 dilution) or nonimmune control (Cell Signaling, 1:100 dilution) antibodies. After the crosslink reversal for 2 h at 56°C the immunocomplexes were analyzed by western with anti -PRMT1, -PRMT5 and -H3 antibodies (1:1000 dilution).

The in vitro protein interactions using the baculovirus-expressed proteins were performed as follows. FLAG-tagged PRMT1 (100 ng) (Active Motif #31411) or PRMT5 (100 ng) (Active Motif #31393) were mixed with His-tagged ULK3 (100 ng) (Abcam #101548) in binding buffer (150 mM NaCl, 100 mM Tris ph8, 0.5% NP40, 10% Glycerol) for 4 h at 4°C on a rotating platform in the presence of 20 µl protein G magnetic beads, anti FLAG (Sigma #F104, 1:100 dilution) and Dynabeads (Thermo Fisher) or HisPur Ni-NTA magnetic beads (Thermo Fisher). After this time, the complexes were washed three times in binding buffer w/o glycerol, once in binding buffer high salts (300 mM NaCl) and resuspended in 20 µl of SDS-PAGE loading buffer for electrophoresis. We carried out the western blots using anti -ULK3, -FLAG, -PRTM1 and -PRMT5 antibodies (1:1000 dilution). To determine the requirement of a native configuration in some experiments the recombinant PRMTs were heated for 20 min in a thermal block, while the same amount of PRMTs protein was left on ice as control, before mixing with the ULK3 protein.

For the kinase assays 1 µg of recombinant ULK3 was mixed with 1 µg of recombinant PRMT1 or PRMT5 in kinase buffer (25 mM Tris ph 7.5, 5 mM ß-glycerophosphate, 2 mM DTT, 0.1 µM NaVO₄, 10 mM MgCl₂) containing 200 µM ATP and incubated for 30 min at 30°C. The reaction was stopped by adding 1/2 volume of SDS PAGE loading buffer. Parallel reactions were carried out in the absence of ULK3 kinase as negative, untreated controls. Some kinase assays were performed in the presence of 1 or 10 µM SU6668 ULK3 kinase inhibitor (Biotechne cat # 3335), or DMSO carrier (Sigma) as a negative control. All the samples from the kinase assays were separated on 12.5% SuperSep™ pre-cast gel copolymerized with SuperSep^TM Phos-Tag^TM (50 µmol/l), 7.5% acrylamide (Wako-FUJI, #192–18001). After the separation, gels were washed for 30 min in 10 mM EDTA in transfer buffer before the blotting. The western blots were carried out using anti -ULK3, -GFP, -FLAG, -PRTM1 and -PRMT5 antibodies.

In vitro Methylation assays were performed by adding 500 ng of ULK3 to 500 ng of PRMT1 or PRMT5 in methylation buffer (50mMTris pH 8.6, 0.02% Triton X100, 2 mM MgCl₂, 1 mM TCEP) containing 50 µM SAM for 3 h at room temperature. Reactions were stopped with 1/2 reaction volume of SDS PAGE loading buffer. The western blots were carried out by using antibodies recognizing di-methylated symmetric (Cell Signaling #13222, 1:1000 dilution) or asymmetric arginine (Cell Signaling #13522, 1:1000 dilution), anti -FLAG and -ULK3 (1:1000 dilution).

The PRMTs activity was measured as follows: PRMT1 (1 µg) or PRMT5 (1 µg) were incubated with ULK3 (1 µg) (or without ULK3) in kinase buffer for 30 min at 30 C. After this time, 0, 50, 100, 200, 400 ng of PRMT1 or PRMT5 were diluted in methylation buffer containing 500 ng/assay of recombinant histone H4 (Active Motif #31493). Reactions were performed in 25 µl and stopped after 3 h by adding 5 µl of SDS PAGE sample buffer 6x. For each reaction, equal volumes (3 µl) were spotted on a nitrocellulose membrane in triplicate and incubated overnight with antibody anti H4R3 symmetric (EpiGentek# 10019-4R3DS, 1:1000 dilution) or H4R3 asymmetric (EpiGentek# 10019-4R3DA, 1:1000 dilution) arginine di methylation. The dot blots were developed with ECL reagents (Pierce), the autoradiographic films quantified using ImageJ, and the values (in arbitrary units) normalized to the highest enzymatic activity obtained in the presence of ULK3.

Tumorigenesis experiments

Ear injections

Mouse-ear injections with the SCC13 and SCCO28 SCC cells were carried out in 8 to 10-week-old female NOD/SCID mice (CB17sc-m, Taconic), as in¹⁸. Both cell lines were infected with a high-titer EGFP-expressing lentivirus¹⁸ and then with either two different ULK3 silencing shRNAs or a control vector as in¹⁸. A total of 10⁵ ULK3- or control-silenced EGFP-expressing SCC cells were injected 5 µl per site using a 33-gauge micro syringe (Hamilton) in the contralateral ear of the same animal. Two experiments with 10 mice each were injected for the SCC13 cells and the SCC028 cells. The day after injection all the mice ears were imaged using a fluorescent stereomicroscope (Leica MZ-FLIII) to confirm the efficient injection and the images were used to normalize the fluorescence intensity at the end of the experiment (21 days). At that time images of the ears were taken using a bright field and fluorescence stereomicroscope. For SCC13 cells tumor sizes were measured using a digital caliper using the following formula: V = (length × width²) × 0.5, and for SCCO28 cells the volume was calculated by determining the fluorescence intensity as in¹⁸ using ImageJ software.

Back injections

Mouse back intradermal tumorigenicity injection assays were carried out in 10-week-old NOD/SCID female mice (with IL-2 receptor γ-chain null mutation, NOD.Cg-Prkdcscid Il2rγtm1Wjl/SzJ, The Jackson Laboratory), as in ref. ⁶⁸. Assays were conducted with skin SCC13 or oral Cal27 SCC cells infected with ULK3 silencing shRNA #2 or with a control shRNA virus, selected and amplified in 1 µg/ml of puromycin. In each injection site we introduced 100 µl of HBSS containing 10⁶ cells diluted 50% with reduced growth factor Matrigel (BD biosciences, cat # 356234) using an insulin syringe. Cells, reagents and syringes were kept on ice before the injection. Each animal was injected in two sites on the same side of the back skin per combination. Three weeks after injection the tumors were removed, imaged, measured, and further processed for immunochemistry.

Intradermal injection with LNA-treated SCC13 was carried out as follows. SCC13 cells were transfected separately with two ULK3 targeting LNAs (#2 and #3) or with scrambled control (#SCR) (10 nM final concentration). After 24 h, 8-week-old NOD/SCID female mice (IL-2 receptor γ-chain null mutation, NOD.Cg-Prkdcscid Il2rγtm1Wjl/SzJ, The Jackson Laboratory) were injected with 100 µl of HBSS containing 10⁶ cells diluted 50% with reduced growth factor Matrigel (BD biosciences, cat # 356234) using an insulin syringe. Animals were injected in two sites of the back skin per combination. Ten days after the injection the tumors were removed, imaged, measured using a digital caliper using the following formula: V = (length × width²) × 0.5 and frozen in optimal cutting temperature compound (OCT, Scigen Tissue-Plus, Thermofisher) before processing for immunochemistry.

The maximal tumor size permitted by the approved IACUC protocol was 20 mm in any direction, which was not exceeded in any experiment. All mice were euthanized by carbon dioxide inhalation. No animal died or was excluded from the experiments. Mice were housed in four per cage on a 12 h dark/light cycle, with a constant ambient temperature of 65–70 F and 40% humidity. All the mice were randomly allocated to the experimental groups and group-housed.

Transcriptomic, Clariom D assays and GSEA

Transcriptomic profiles from Head and Neck (H&N), Lung, and Cervical SCCs from the TCGA Research Network updated to 2018 were plotted as Violin plots generated using individual sample values (z scores, calculated according to z = (expression in tumor sample - mean expression in reference sample) / standard deviation of expression in reference sample)). Head and neck: n(normal) = 44, n(SCC) = 526; lung SCCs: n(normal) = 45, n(SCC) = 509; cervical SCCs: n(normal) = 13, n(SCC) = 374.

TCGA Research Network datasets updated to 2016 as in² of primary tumors versus normal tissues from Bladder, Cervical, Esophageal and Lung SCCs. The analysis of these datasets was made using GEPIA software (http://gepia.cancer-pku.cn/index.html) and shows box-and-whiskers plots, with indicated the median, box (25–75%) and whiskers (5–95%), log(transcript per million+1), value cutoff P < 0.01.

BLCA: n(normal) = 19, n(SCC) = 404; CESC: n(normal) = 3, n(SCC) = 306; ESCA: n(normal) = 13, n(SCC) = 182; LUSC: n(normal) = 50, n(SCC) = 484.

Transcriptomic studies were performed using Clariom D arrays (Applied Biosystem). Total RNA was extracted by using Directzol RNA Miniprep assay Kit (Zymo Research #R2071) one week after ULK3 shRNA silencing for HKCs (three separate HKC strains silenced each with two ULK3-targeting shRNAs, or a control plasmid) and two weeks after ULK3 silencing for SCC13 cells (silenced with two separate ULK3-targeting shRNAs, or a control vector). Library probe preparation was done with 1 μg of total RNA. Clariom D arrays were used on the GeneChip™ 3000 instrument system and the results were processed with Transcriptome Analysis Console (TAC) software to analyze and visualize global expression patterns of genes, pathways, and possible alternative splicing events.

Gene set enrichment analysis (GSEA) for Clariom D array expression profiles was performed by using GSAA-SeqSP software (gene set association analysis for array expression data with sample permutation) from the Gene Set Association Analysis (GSAA) platform (version GSAA_2.0, http://gsaa.unc.edu/). Curated gene sets were obtained from the Molecular Signatures Database (MSigDB v5.2; http://www.broadinstitute.org/gsea/msigdb/).

DAVID software (https://david.ncifcrf.gov/) was used for gene ontology (GO) analysis on lists of genes modulated by ULK3 silencing in HKCs and SCC13 cells.

Metabolomics studies

Analysis of steady-state metabolites in SCC13 cells infected with two separate lentiviruses for ULK3 shRNA, or a silencing control virus, was performed at Beth Israel mass spectrometry core facility. In triplicate, one 10 cm culture dish per condition was extracted three times with 500 µl 80% cold methanol (−80°C), (each extraction was made on ice for 10 min, followed by centrifugation and resuspension of the pellet in 80% methanol, the three extractions pooled before next step) then the extracted metabolites were centrifuged and dried using a speed vacuum concentrator for 30 min with heat. Dried pelleted samples were resuspended in 100 µl of water before analysis at the Beth Israel Deaconess Mass Spectrometry Core facility directed by Dr. J. Asara.

Metabolites were evaluated individually for their significance or by Metabolite Set Enrichment Analysis to identify biological patterns significantly enriched using MetaboAnalist software (http://www.metaboanalyst.ca/).

ChIP-seq and ChIP analysis

For ChIP-seq, chromatin prepared from 32 × 10⁶ SCC13 cells as in ref. ¹⁸ by using a simple ChIP assay kit (Cell Signaling # 56383), was immunoprecipitated using 10 µg anti ULK3 (Santa Cruz # 517373, 1:100 dilution) or nonimmune control antibodies (Santa Cruz, 1:100 dilution), followed by quantification of the DNA by fluorometry on a Qubit system (Invitrogen) and quality/size determination using the Bioanalyzer assay (Agilent). Chromatin from SCC13 plus/minus ULK3 silencing was similarly immunoprecipitated using 5 µg of anti-H3K27ac antibodies (Abcam # 4729, 1:100 dilution), or non-immune control antibodies (Abcam, 1:100 dilution) and analyzed as above.

According to the manufacturer’s recommendations, a total of 10 ng DNA was used for library preparation using the NEBNext ChIP-Seq Library Prep Kit (Illumina).

Raw data files from ChIP-seq assays were aligned to the GRCh38 genome (https://www.ncbi.nlm.nih.gov/assembly/GCF_000001405.26/) with Bowtie2 Version 2.3.0 (http://bowtie-bio.sourceforge.net/bowtie2). Duplicates were removed with Picard (https://broadinstitute.github.io/picard/) and, for peak detection, MACS2 software (http://liulab.dfci.harvard.edu/MACS) was used with a p-value cutoff of 1.00e − 04. Peaks were annotated with HOMER (http://homer.ucsd.edu/homer/index.html). The Integrative Genomics Viewer (http:// software.broadinstitute.org/software/igv/) was used to illustrate ChIP-seq peaks.

DAVID software (https://david.ncifcrf.gov/) was used for gene ontology (GO) analysis.

Data sets generated for this study were deposited in the NCBI Gene Expression Omnibus.

ChIP experiments

SCC13 cells were infected with two viruses targeting ULK3 (shULK3#1 and shULK3#2) or a control virus (shCTR), selected and amplified in 1 µg/ml of puromycin. Cell cultures were crosslinked with 1% formaldehyde, and total cellular chromatin was obtained using ChIP sonication kit solutions (Cell signaling# 56383). The efficiency of the sonication process was monitored by using 1% agarose gels. For validation of ULK3 binding to promoters and of active chromatin regulation by ULK3 silencing, 30 µg of crosslinked sonicated chromatin was immunoprecipitated overnight with anti ULK3 (Santa Cruz # 517373, 1:100 dilution), H3K9ac (Upstate #06-942, 1:100 dilution), H3K27ac (Abcam # 4729, 1:100 dilution), Pol II (Upstate #05-623, 1:100 dilution) or nonimmune antibodies (Cell Signaling, 1:100 dilution), as control, in the presence of 40 µl of magnetic beads protein G (Cell Signaling). After three washes in low salts and one wash in high salt in nuclear lysis chromatin preparation buffers (Cell Signaling), the chromatin-bound to antibodies was eluted for 30 min at 65 °C and de-crosslinked for 2 h at 65 °C in the presence of 2 µl Proteinase K (10 mg/ml). PCR was carried out by using for each analyzed gene (PKM, LDHA, FOXM1) two separate oligonucleotide pairs amplifying the regions of the promoter/TSS.

To determine the consequences of ULK3 silencing on PRMT1 and PRMT5 association to specific promoters, 50 µg of crosslinked chromatin prepared as described above from SCC13 cells plus/minus ULK3 silencing with one shRNA (shULK3 #2), were immunoprecipitated overnight with 5 µg of rabbit anti PRMT1 (Cell Signaling # 2449), anti PRMT5 (Cell Signaling # 79998) or nonimmune IgGs (Cell Signaling). For each ULK3-bound metabolic gene analyzed, (PKM and LDHA), the immunoprecipitated chromatin was amplified with oligonucleotide pairs designed to amplify four different regions within the ULK3 binding peak, as determined by ChIP seq analysis. ChIP experiments were carried out in parallel duplicates. Fold enrichment was calculated by normalizing to IgG values using the 2^^-∆Ct formula.

Statistics and reproducibility

Data are presented as mean ± SEM, mean ± SD, or ratios among treated and controls, with three or more separate HKC strains and SCC cells, or patient-derived samples, used in independent experiments, as indicated in the Figure legends. All western blots were performed at least twice as independent biological replicates. When a representative image is shown, the number of samples and conditions analyzed are provided in the figure legend. For gene expression and all functional testing assays, the statistical significance of differences between experimental groups and controls was assessed by two-tailed unpaired or paired t-test, one-way ANOVA, and by comparing to random permutations (for GSEA), as indicated in the legends. In all cases, P values < 0.05 were considered statistically significant. Statistical significance analysis and plotting of the data were performed using Microsoft Excel 365 v16.67 and Prism 9 Graphpad software.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The transcriptomic (HKCs and SCC13 cells) and ChIP-seq data generated in this study have been deposited in GEO under the accession codes GSE183084, GSE183085, and GSE183933, respectively. The metabolomics data have been deposited in MetaboLights as MTBLS5155. Sequences for shRNA, siRNA, LNAs, CRISPR/cas9, PCR oligos, and antibodies used are in Supplementary Tables 1–5. Source data underlying Figs. 1a–h, 2a–f, 3a-j, 4b–g, 5c–f, 6b–g, 7a–g, 8a, b, d–h, 9a–g, 10a-i and Supplementary Figs. S1d, S2a, S2d, S2e, S2f, S4, S5d, S6e, S7a–c, S8s, S8d are provided with this paper. Source data are provided with this paper.

References

Thompson, A. K., Kelley, B. F., Prokop, L. J., Murad, M. H. & Baum, C. L. Risk factors for cutaneous squamous cell carcinoma recurrence, metastasis, and disease-specific death: a systematic review and meta-analysis. JAMA Dermatol 152, 419–428 (2016).
Article PubMed PubMed Central Google Scholar
Campbell, J. D. et al. Genomic, pathway network, and immunologic features distinguishing squamous carcinomas. Cell Rep. 23, 194–212.e196 (2018).
Article CAS PubMed PubMed Central Google Scholar
Dotto, G. P. & Rustgi, A. K. Squamous cell cancers: a unified perspective on biology and genetics. Cancer Cell 29, 622–637 (2016).
Article CAS PubMed PubMed Central Google Scholar
Dotto, G. P. Crosstalk of Notch with p53 and p63 in cancer growth control. Nat. Rev. Cancer 9, 587–595 (2009).
Article CAS PubMed PubMed Central Google Scholar
Nguyen, B. C. et al. Cross-regulation between Notch and p63 in keratinocyte commitment to differentiation. Genes Dev. 20, 1028–1042 (2006).
Article CAS PubMed PubMed Central Google Scholar
Truong, A. B., Kretz, M., Ridky, T. W., Kimmel, R. & Khavari, P. A. p63 regulates proliferation and differentiation of developmentally mature keratinocytes. Genes Dev. 20, 3185–3197 (2006).
Article CAS PubMed PubMed Central Google Scholar
Ohashi, S. et al. NOTCH1 and NOTCH3 coordinate esophageal squamous differentiation through a CSL-dependent transcriptional network. Gastroenterology 139, 2113–2123 (2010).
Article CAS PubMed Google Scholar
Rock, J. R. et al. Notch-dependent differentiation of adult airway basal stem cells. Cell Stem Cell 8, 639–648 (2011).
Article CAS PubMed PubMed Central Google Scholar
Talora, C., Sgroi, D. C., Crum, C. P. & Dotto, G. P. Specific down-modulation of Notch1 signaling in cervical cancer cells is required for sustained HPV-E6/E7 expression and late steps of malignant transformation. Genes Dev. 16, 2252–2263 (2002).
Article CAS PubMed PubMed Central Google Scholar
Rubin, L. L. & de Sauvage, F. J. Targeting the Hedgehog pathway in cancer. Nat. Rev. Drug Disco. 5, 1026–1033 (2006).
Article CAS Google Scholar
Grachtchouk, M. et al. Basal cell carcinomas in mice overexpressing Gli2 in skin. Nat. Genet 24, 216–217 (2000).
Article CAS PubMed Google Scholar
Li, C., Chi, S. & Xie, J. Hedgehog signaling in skin cancers. Cell Signal 23, 1235–1243 (2011).
Article CAS PubMed PubMed Central Google Scholar
Scales, S. J. & de Sauvage, F. J. Mechanisms of Hedgehog pathway activation in cancer and implications for therapy. Trends Pharmacol. Sci. 30, 303–312 (2009).
Article CAS PubMed Google Scholar
Maloverjan, A., Piirsoo, M., Michelson, P., Kogerman, P. & Osterlund, T. Identification of a novel serine/threonine kinase ULK3 as a positive regulator of Hedgehog pathway. Exp. cell Res. 316, 627–637 (2010).
Article CAS PubMed Google Scholar
Han, Y. et al. Phosphorylation of Ci/Gli by fused family kinases promotes hedgehog signaling. Developmental Cell 50, 610–626.e614 (2019).
Article CAS PubMed PubMed Central Google Scholar
Maloverjan, A. et al. Dual function of UNC-51-like kinase 3 (Ulk3) in the Sonic hedgehog signaling pathway. J. Biol. Chem. 285, 30079–30090 (2010).
Article PubMed PubMed Central Google Scholar
Caballe, A. et al. ULK3 regulates cytokinetic abscission by phosphorylating ESCRT-III proteins. Elife 4, e06547 (2015).
Article PubMed PubMed Central Google Scholar
Goruppi, S. et al. The ULK3 kinase is critical for convergent control of cancer-associated fibroblast activation by CSL and GLI. Cell Rep. 20, 2468–2479 (2017).
Article CAS PubMed PubMed Central Google Scholar
Kasak, L. et al. Characterization of protein kinase ULK3 regulation by phosphorylation and inhibition by small molecule SU6668. Biochemistry 57, 5456–5465 (2018).
Article CAS PubMed Google Scholar
Bradner, J. E., Hnisz, D. & Young, R. A. Transcriptional addiction in cancer. Cell 168, 629–643 (2017).
Article CAS PubMed PubMed Central Google Scholar
Hnisz, D. et al. Super-enhancers in the control of cell identity and disease. Cell 155, 934–947 (2013).
Article CAS PubMed Google Scholar
Yi, M. et al. TP63 links chromatin remodeling and enhancer reprogramming to epidermal differentiation and squamous cell carcinoma development. Cell. Mol. Life Sci. 77, 4325–4346 (2020).
Article CAS PubMed PubMed Central Google Scholar
Muhar, M. et al. SLAM-seq defines direct gene-regulatory functions of the BRD4-MYC axis. Science 360, 800–805 (2018).
Article CAS PubMed PubMed Central Google Scholar
Yang, H. et al. ETS family transcriptional regulators drive chromatin dynamics and malignancy in squamous cell carcinomas. Elife 4, e10870 (2015).
Article PubMed PubMed Central Google Scholar
Kim, J. W. et al. Evaluation of myc E-box phylogenetic footprints in glycolytic genes by chromatin immunoprecipitation assays. Mol. Cell Biol. 24, 5923–5936 (2004).
Article CAS PubMed PubMed Central Google Scholar
Cui, J. et al. FOXM1 promotes the warburg effect and pancreatic cancer progression via transactivation of LDHA expression. Clin. Cancer Res. 20, 2595–2606 (2014).
Article CAS PubMed PubMed Central Google Scholar
Shang, R., Pu, M., Li, Y. & Wang, D. FOXM1 regulates glycolysis in hepatocellular carcinoma by transactivating glucose transporter 1 expression. Oncol. Rep. 37, 2261–2269 (2017).
Article CAS PubMed Google Scholar
Zhou, L. et al. HOXA9 inhibits HIF-1α-mediated glycolysis through interacting with CRIP2 to repress cutaneous squamous cell carcinoma development. Nat. Commun. 9, 1480 (2018).
Article ADS PubMed PubMed Central Google Scholar
Momcilovic, M. et al. The GSK3 signaling axis regulates adaptive glutamine metabolism in lung squamous cell carcinoma. Cancer Cell 33, 905–921.e905 (2018).
Article CAS PubMed PubMed Central Google Scholar
Yang, J. et al. Targeting cellular metabolism to reduce head and neck cancer growth. Sci. Rep. 9, 4995 (2019).
Article ADS PubMed PubMed Central Google Scholar
Biggar, K. K. & Li, S. S. Non-histone protein methylation as a regulator of cellular signalling and function. Nat. Rev. Mol. Cell Biol. 16, 5–17 (2015).
Article CAS PubMed Google Scholar
Blanc, R. S. & Richard, S. Arginine methylation: the coming of age. Mol. Cell 65, 8–24 (2017).
Article CAS PubMed Google Scholar
Takai, H. et al. 5-Hydroxymethylcytosine plays a critical role in glioblastomagenesis by recruiting the CHTOP-methylosome complex. Cell Rep. 9, 48–60 (2014).
Article CAS PubMed Google Scholar
Tikhanovich, I. et al. Arginine methylation regulates c-Myc-dependent transcription by altering promoter recruitment of the acetyltransferase p300. J. Biol. Chem. 292, 13333–13344 (2017).
Article CAS PubMed PubMed Central Google Scholar
Li, X. et al. H4R3 methylation facilitates beta-globin transcription by regulating histone acetyltransferase binding and H3 acetylation. Blood 115, 2028–2037 (2010).
Article CAS PubMed PubMed Central Google Scholar
Bao, X. et al. CSNK1a1 regulates PRMT1 to maintain the progenitor state in self-renewing somatic tissue. Dev. Cell 43, 227–239 e225 (2017).
Article CAS PubMed PubMed Central Google Scholar
Ferguson, F. M. & Gray, N. S. Kinase inhibitors: the road ahead. Nat. Rev. Drug Disco. 17, 353–377 (2018).
Article CAS Google Scholar
Reinehr, C. P. H. & Bakos, R. M. Actinic keratoses: review of clinical, dermoscopic, and therapeutic aspects. Bras. Dermatol 94, 637–657 (2019).
Article Google Scholar
Pastrana, E., Silva-Vargas, V. & Doetsch, F. Eyes wide open: a critical review of sphere-formation as an assay for stem cells. Cell Stem Cell 8, 486–498 (2011).
Article CAS PubMed PubMed Central Google Scholar
Doudna, J. A. & Charpentier, E. The new frontier of genome engineering with CRISPR-Cas9. Science 346, 1258096 (2014).
Article PubMed Google Scholar
Qiu, P. et al. Mutation detection using surveyor™ nuclease. BioTechniques 36, 702–707 (2004).
Article CAS PubMed Google Scholar
Blanpain, C., Lowry, W. E., Geoghegan, A., Polak, L. & Fuchs, E. Self-renewal, multipotency, and the existence of two cell populations within an epithelial stem cell niche. Cell 118, 635–648 (2004).
Article CAS PubMed Google Scholar
Piirsoo, A. et al. Protein kinase inhibitor SU6668 attenuates positive regulation of Gli proteins in cancer and multipotent progenitor cells. Biochimica et. Biophysica Acta 1843, 703–714 (2014).
Article CAS PubMed PubMed Central Google Scholar
Topley, G. I., Okuyama, R., Gonzales, J. G., Conti, C. & Dotto, G. P. p21(WAF1/Cip1) functions as a suppressor of malignant skin tumor formation and a determinant of keratinocyte stem-cell potential. Proc. Natl Acad. Sci. USA 96, 9089–9094 (1999).
Article ADS CAS PubMed PubMed Central Google Scholar
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. Proc. Natl Acad. Sci. USA 102, 15545–15550 (2005).
Article ADS CAS PubMed PubMed Central Google Scholar
Brooks, Y. S. et al. Multifactorial ERbeta and NOTCH1 control of squamous differentiation and cancer. J. Clin. Invest 124, 2260–2276 (2014).
Article CAS PubMed PubMed Central Google Scholar
Wang, Y. et al. FOXM1 promotes reprogramming of glucose metabolism in epithelial ovarian cancer cells via activation of GLUT1 and HK2 transcription. Oncotarget 7, 47985–47997 (2016).
Article PubMed PubMed Central Google Scholar
Woolbright, B. L., Rajendran, G., Harris, R. A. & Taylor, J. A. 3rd Metabolic flexibility in cancer: targeting the pyruvate dehydrogenase kinase:pyruvate dehydrogenase axis. Mol. Cancer Ther. 18, 1673–1681 (2019).
Article CAS PubMed Google Scholar
Bensaad, K. et al. TIGAR, a p53-inducible regulator of glycolysis and apoptosis. Cell 126, 107–120 (2006).
Article CAS PubMed Google Scholar
Yang, L., Venneti, S. & Nagrath, D. Glutaminolysis: a hallmark of cancer metabolism. Annu. Rev. Biomed. Eng. 19, 163–194 (2017).
Article CAS PubMed Google Scholar
Chong, J., Wishart, D. S. & Xia, J. Using metaboAnalyst 4.0 for comprehensive and integrative metabolomics data analysis. Curr. Protoc. Bioinforma. 68, e86 (2019).
Article Google Scholar
Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS. Genome Biol. 9, R137 (2008).
Salmon-Divon, M., Dvinge, H., Tammoja, K. & Bertone, P. Peakanalyzer: genome-wide annotation of chromatin binding and modification loci. BMC Bioinforma. 11, 415 (2010).
Article Google Scholar
Belotserkovskaya, R., Saunders, A., Lis, J. T. & Reinberg, D. Transcription through chromatin: understanding a complex FACT. Biochim Biophys. Acta 1677, 87–99 (2004).
Article CAS PubMed Google Scholar
Akram, M. Mini-review on glycolysis and cancer. J. Cancer Educ. 28, 454–457 (2013).
Article CAS PubMed Google Scholar
Bosic, M. M. et al. Expression of p300 and p300/CBP associated factor (PCAF) in actinic keratosis and squamous cell carcinoma of the skin. Exp. Mol. Pathol. 100, 378–385 (2016).
Article CAS PubMed Google Scholar
Guenther, M. G., Levine, S. S., Boyer, L. A., Jaenisch, R. & Young, R. A. A chromatin landmark and transcription initiation at most promoters in human cells. Cell 130, 77–88 (2007).
Article CAS PubMed PubMed Central Google Scholar
Gao, G. et al. PRMT1 loss sensitizes cells to PRMT5 inhibition. Nucleic Acids Res. 47, 5038–5048 (2019).
Article CAS PubMed PubMed Central Google Scholar
Srour, N., Mersaoui, S. Y. & Richard, S. M-TAP dance: targeting PRMT1 and PRMT5 family members to push cancer cells over the edge. Cancer Cell 36, 3–5 (2019).
Article CAS PubMed Google Scholar
Kanade, S. R. & Eckert, R. L. Protein arginine methyltransferase 5 (PRMT5) signaling suppresses protein kinase Cδ- and p38δ-dependent signaling and keratinocyte differentiation. J. Biol. Chem. 287, 7313–7323 (2012).
Article CAS PubMed Google Scholar
Soderberg, O. et al. Direct observation of individual endogenous protein complexes in situ by proximity ligation. Nat. Methods 3, 995–1000 (2006).
Article PubMed Google Scholar
Goruppi, S. et al. Autophagy controls CSL/RBPJkappa stability through a p62/SQSTM1-dependent mechanism. Cell Rep. 24, 3108–3114.e3104 (2018).
Article CAS PubMed PubMed Central Google Scholar
Chen, R.-Q. et al. CDC25B mediates rapamycin-induced oncogenic responses in cancer cells. Cancer Res. 69, 2663–2668 (2009).
Article CAS PubMed PubMed Central Google Scholar
O’Donoghue, L. & Smolenski, A. Analysis of protein phosphorylation using Phos-tag gels. J. Proteom. 259, 104558 (2022).
Article Google Scholar
Saha, K., Fisher, M. L., Adhikary, G., Grun, D. & Eckert, R. L. Sulforaphane suppresses PRMT5/MEP50 function in epidermal squamous cell carcinoma leading to reduced tumor formation. Carcinogenesis 38, 827–836 (2017).
Article CAS PubMed PubMed Central Google Scholar
Zhao, Y. et al. PRMT1 regulates the tumour-initiating properties of esophageal squamous cell carcinoma through histone H4 arginine methylation coupled with transcriptional activation. Cell Death Dis. 10, 359 (2019).
Article PubMed PubMed Central Google Scholar
Chuang, C. Y. et al. PRMT1 expression is elevated in head and neck cancer and inhibition of protein arginine methylation by adenosine dialdehyde or PRMT1 knockdown downregulates proliferation and migration of oral cancer cells. Oncol. Rep. 38, 1115–1123 (2017).
Article CAS PubMed Google Scholar
Al Labban, D. et al. Notch-effector CSL promotes squamous cell carcinoma by repressing histone demethylase KDM6B. J. Clin. Invest 128, 2581–2599 (2018).
Article PubMed PubMed Central Google Scholar
Rheinwald, J. G. & Beckett, M. A. Tumorigenic keratinocyte lines requiring anchorage and fibroblast support cultured from human squamous cell carcinomas. Cancer Res 41, 1657–1663 (1981).
CAS PubMed Google Scholar
Lin, C. J. et al. Head and neck squamous cell carcinoma cell lines: established models and rationale for selection. Head. Neck 29, 163–188 (2007).
Article PubMed Google Scholar
Crooke, S. T., Liang, X. H., Baker, B. F. & Crooke, R. M. Antisense technology: A review. J. Biol. Chem. 296, 100416 (2021).
Article CAS PubMed PubMed Central Google Scholar
Roberts, T. C., Langer, R. & Wood, M. J. A. Advances in oligonucleotide drug delivery. Nat. Rev. Drug Discov. 19, 673–694 (2020).
Article CAS PubMed PubMed Central Google Scholar
Sakaue-Sawano, A. et al. Visualizing spatiotemporal dynamics of multicellular cell-cycle progression. Cell 132, 487–498 (2008).
Article CAS PubMed Google Scholar
Dotto, G. P. Multifocal epithelial tumors and field cancerization: stroma as a primary determinant. J. Clin. Invest 124, 1446–1453 (2014).
Article CAS PubMed PubMed Central Google Scholar
Eckert, R. L. et al. Regulation of involucrin gene expression. J. Invest. Dermatol. 123, 13–22 (2004).
Article CAS PubMed Google Scholar
Yue, D. et al. Hedgehog/Gli promotes epithelial-mesenchymal transition in lung squamous cell carcinomas. J. Exp. Clin. Cancer Res. 33, 34 (2014).
Article PubMed PubMed Central Google Scholar
Chen, S.-F. et al. Nonadhesive culture system as a model of rapid sphere formation with cancer stem cell properties. PLOS ONE 7, e31864 (2012).
Article ADS CAS PubMed PubMed Central Google Scholar
Beacon, T. H., Xu, W. & Davie, J. R. Genomic landscape of transcriptionally active histone arginine methylation marks, H3R2me2s and H4R3me2a, relative to nucleosome depleted regions. Gene 742, 144593 (2020).
Article CAS PubMed Google Scholar
Yang, Y. & Bedford, M. T. Protein arginine methyltransferases and cancer. Nat. Rev. Cancer 13, 37–50 (2013).
Article CAS PubMed Google Scholar
Barrero, M. J. & Malik, S. Two functional modes of a nuclear receptor-recruited arginine methyltransferase in transcriptional activation. Mol. Cell 24, 233–243 (2006).
Article CAS PubMed PubMed Central Google Scholar
Zhao, Q. et al. PRMT5-mediated methylation of histone H4R3 recruits DNMT3A, coupling histone and DNA methylation in gene silencing. Nat. Struct. Mol. Biol. 16, 304–311 (2009).
Article CAS PubMed PubMed Central Google Scholar
Lafleur, V. N., Richard, S. & Richard, D. E. Transcriptional repression of hypoxia-inducible factor-1 (HIF-1) by the protein arginine methyltransferase PRMT1. Mol. Biol. Cell 25, 925–935 (2014).
Article PubMed PubMed Central Google Scholar
LeBlanc, S. E. et al. Protein arginine methyltransferase 5 (Prmt5) promotes gene expression of peroxisome proliferator-activated receptor γ2 (PPARγ2) and its target genes during adipogenesis. Mol. Endocrinol. 26, 583–597 (2012).
Article CAS PubMed PubMed Central Google Scholar
Fedoriw, A. et al. Anti-tumor activity of the type i prmt inhibitor, gsk3368715, synergizes with prmt5 inhibition through mtap loss. Cancer Cell, https://doi.org/10.1016/j.ccell.2019.05.014 (2019).
Dhar, S. et al. Loss of the major Type I arginine methyltransferase PRMT1 causes substrate scavenging by other PRMTs. Sci. Rep. 3, 1311 (2013).
Article ADS PubMed PubMed Central Google Scholar
An, W., Kim, J. & Roeder, R. G. Ordered cooperative functions of PRMT1, p300, and CARM1 in transcriptional activation by p53. Cell 117, 735–748 (2004).
Article CAS PubMed Google Scholar
Chai, A. W. Y. et al. Genome-wide CRISPR screens of oral squamous cell carcinoma reveal fitness genes in the Hippo pathway. eLife 9, e57761 (2020).
Article CAS PubMed PubMed Central Google Scholar
Behan, F. M. et al. Prioritization of cancer therapeutic targets using CRISPR–Cas9 screens. Nature 568, 511–516 (2019).
Article ADS CAS PubMed Google Scholar
Eram, M. S. et al. A potent, selective, and cell-active inhibitor of human type i protein arginine methyltransferases. ACS Chem. Biol. 11, 772–781 (2016).
Article CAS PubMed Google Scholar
Chan-Penebre, E. et al. A selective inhibitor of PRMT5 with in vivo and in vitro potency in MCL models. Nat. Chem. Biol. 11, 432–437 (2015).
Article CAS PubMed Google Scholar

Download references

Acknowledgements

We are grateful to Dania Al Labban and Esther Revai for help with the experiments, and to Enzo Calautti, Caterina Missero for critical reading of the manuscript. The head and neck, lung, bladder, esophageal and cervical data shown in Fig. 1a are based upon data generated by the TCGA Research Network: www.cancer.gov/tcga. This work was supported by grants from the National Institutes of Health (R01AR078374; R01AR039190; content not necessarily representing the official views of NIH). GPD is a member of the SKINTEGRITY.CH collaborative research program and students in his group are supported by funding from the European Union’s Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement No 859860.

Emery Di Cicco is a recipient of a “Fellowship for Abroad 2020” from Fondazione AIRC.

Author information

Authors and Affiliations

Cutaneous Biology Research Center, Massachusetts General Hospital, Charlestown, 02129, MA, USA
Sandro Goruppi, Andrea Clocchiatti, Giulia Bottoni, Emery Di Cicco, Shadhmer Demehri & G. Paolo Dotto
Department of Dermatology, Massachusetts General Hospital and Harvard Medical School, Boston, 02114, MA, USA
Sandro Goruppi, Andrea Clocchiatti, Giulia Bottoni, Emery Di Cicco, Victor Neel, Shadhmer Demehri & G. Paolo Dotto
Personalized Cancer Prevention Research Unit and Head and Neck Surgery Division, Centre Hospitalier Universitaire Vaudois, Lausanne, 1011, Switzerland
Min Ma, Beatrice Tassone, Christian Simon & G. Paolo Dotto
Department of Immunobiology, University of Lausanne, Epalinges, 1066, Switzerland
Min Ma, Beatrice Tassone, Christian Simon & G. Paolo Dotto
International Cancer Prevention Institute, Epalinges, 1066, Switzerland
Christian Simon & G. Paolo Dotto

Authors

Sandro Goruppi
View author publications
You can also search for this author in PubMed Google Scholar
Andrea Clocchiatti
View author publications
You can also search for this author in PubMed Google Scholar
Giulia Bottoni
View author publications
You can also search for this author in PubMed Google Scholar
Emery Di Cicco
View author publications
You can also search for this author in PubMed Google Scholar
Min Ma
View author publications
You can also search for this author in PubMed Google Scholar
Beatrice Tassone
View author publications
You can also search for this author in PubMed Google Scholar
Victor Neel
View author publications
You can also search for this author in PubMed Google Scholar
Shadhmer Demehri
View author publications
You can also search for this author in PubMed Google Scholar
Christian Simon
View author publications
You can also search for this author in PubMed Google Scholar
G. Paolo Dotto
View author publications
You can also search for this author in PubMed Google Scholar

Contributions

S.G., A.C., G.B., B.T., E.D. and M.M. performed experiments and analyzed the results with G.P.D. A.C., B.T. and S.G. performed the bioinformatic analysis of transcriptomic with G.P.D. S.G., A.C., M.M., C.S. and Novogene, UK, conducted the bioinformatics analysis of Chip-seq experiments with G.P.D. S.D. and V.N. provided the clinical samples. S.G. and G.P.D. designed the study and wrote the manuscript.

Corresponding authors

Correspondence to Sandro Goruppi or G. Paolo Dotto.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Communications thanks Sebastian Mathea and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Peer Review File

Reporting Summary

Source data

Source Data

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as 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 changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Goruppi, S., Clocchiatti, A., Bottoni, G. et al. The ULK3 kinase is a determinant of keratinocyte self-renewal and tumorigenesis targeting the arginine methylome. Nat Commun 14, 887 (2023). https://doi.org/10.1038/s41467-023-36410-6

Download citation

Received: 16 December 2021
Accepted: 26 January 2023
Published: 16 February 2023
DOI: https://doi.org/10.1038/s41467-023-36410-6

Comments

By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.