DNPH1 exerts context-dependent tumor-promoting and -repressing functions during melanoma development

Abstract

Metastatic cutaneous melanoma is a highly lethal and multifarious disease that is driven by the interplay between UV-induced DNA damage and oncogenic lesions, yet enzymes coupling these processes remain incompletely defined. We uncovered the deoxynucleotide hydrolase DNPH1 as an overexpressed melanoma enzyme associated with metastasis and poorer patient survival. DNPH1 overexpression or downregulation in human melanoma cells enhanced or lowered, respectively, their in vitro growth rate. In vivo, Dnph1 loss delayed melanoma initiation and reduced distant metastasis in UV light-exposed, BRAF-V600E-expressing mice. Similarly, Dnph1 knockout in conjunction with oncogenic BRAF-V600E slowed melanoma onset upon ablation of the tumor suppressor TP53. Surprisingly, Dnph1 knockout facilitated growth of tumors after their establishment, indicating stage-specific functions during melanoma development. Mechanistically, UV light triggered the translocation of DNPH1 into the cell nucleus, where it may alter DNA repair pathway choice by suppressing nucleotide excision repair while enhancing homologous recombination and non-homologous end joining. DNPH1 also promoted NF-κB signaling, which may involve cooperation with two interaction partners, ANXA2 and FHL2, in stimulating NF-κB-dependent transcription. Altogether, these findings have identified DNPH1 as a previously unrecognized regulator in the development of cutaneous melanomas where it functions in a context-dependent manner as a tumor promoter or attenuator.

Graphical Abstract

INTRODUCTION

This year, in the US alone, about 112,000 persons are anticipated to be diagnosed with melanoma and more than 8,500 patients are predicted to succumb to this disease (Siegel et al, 2026). Cutaneous melanoma is an aggressive form of skin cancer typically driven by the interplay between ultraviolet (UV)-induced DNA damage and cooperating oncogenic events, most commonly involving BRAF activation. Once metastasized, melanomas are highly lethal as current therapies fail to provide durable benefit for many patients (Flaherty, 2023; Fletcher and Johnson, 2024). These facts emphasize the urgent need for additional avenues of therapeutic intervention.

2’-Deoxynucleoside 5’-phosphate N-hydrolase 1 (DNPH1; also known as C6orf108 or RCL) is an understudied enzyme that reportedly cleaves dNMPs into 2-deoxyribose 5-phosphate and free bases (Ghiorghi et al, 2007). It was speculated, but never validated, that DNPH1 activity might augment base salvage, thereby increasing dNTP levels and consequently promoting proliferation (Ghiorghi et al, 2007). But the limited ability of DNPH1 to hydrolyze canonical dNMPs (dAMP, dCMP, dGMP, dTMP, dUMP) does not align with this notion (Fugger et al, 2021; Wagner et al, 2023) and the nucleotide salvage pathway would just re-create dNMPs, rendering this process futile. Moreover, 2-deoxyribose 5-phosphate produced upon DNPH1 catalytic activity can be readily dephosphorylated into 2-deoxyribose that may act as an angiogenic factor (Akiyama et al, 2004; Azam et al, 2019; Brown and Bicknell, 1998; Dikici et al, 2020; Haraguchi et al, 1994; Sengupta et al, 2003; Uchimiya et al, 2002), yet no experimental evidence has yet supported such a mechanism of action for DNPH1. Thus, how DNPH1 could influence normal or malignant cellular physiology has remained largely unresolved. A recent study, however, demonstrated that DNPH1 hydrolyzes cytotoxic, noncanonical deoxynucleotides such as 5-hydroxymethyl-2’-deoxyuridine 5’-monophosphate, and this sanitary action could boost cancer cell survival (Fugger et al, 2021), hinting at a pro-tumorigenic function for DNPH1.

We had identified the DNPH1 gene in a screen (Shin et al, 2008) for targets of the oncogenic transcription factor ETV1 (Oh et al, 2012; Qi et al, 2020). Phosphorylation of ETV1 through mitogen-activated protein (MAP) kinases and MAP kinase-activated protein (MAPKAP) kinases stimulates its transcriptional activity, which places ETV1 downstream of prominent oncoproteins such as HER2/Neu, RAS or RAF that trigger MAP/MAPKAP kinase signaling (Bosc et al, 2001; Dowdy et al, 2003; Goel and Janknecht, 2004; Janknecht, 2003; Wu and Janknecht, 2002). Furthermore, the oncogenic V600E mutant of the BRAF kinase, which is found in about half of all human melanomas (Cancer Genome Atlas, 2015), can cooperate with ETV1 in transforming cells, and ETV1 is overexpressed in many melanomas (Birner et al, 2014; Jane-Valbuena et al, 2010; Mehra et al, 2013). This suggested that DNPH1 might also become overexpressed in melanomas and potentially facilitate their development. Hence, this has been studied in this report.

RESULTS

Overexpression of DNPH1 in melanomas

To examine a potential role of DNPH1 in melanomas, we first studied its expression in human skin. In normal skin, DNPH1 expression was noted in melanocytes (identifiable through MART-1 staining) as well as non-melanocytes, and DNPH1 was also robustly expressed in the transformed melanocytes within human cutaneous melanomas (Figure 1a). As expected from previous studies with human breast cancer cells (Shin et al, 2008), DNPH1 was primarily localized within the cytoplasm of melanocytes (Figure 1a). Bioinformatics revealed that DNPH1 mRNA was significantly upregulated in melanomas compared to normal skin (Figure 1b), and DNPH1 protein levels were higher within metastases than in primary tumors (Figure 1c). The latter suggested that DNPH1 is associated with aggressive disease, and accordingly high DNPH1 levels in melanomas correlated with reduced patient survival (Figure 1d). Overall, these data suggested that DNPH1 expression correlates with melanoma progression and may contribute to disease aggressiveness.

Figure 1. Expression of DNPH1 in human melanomas.

(a) Immunostaining of normal human skin and a primary cutaneous melanoma tumor with DNPH1 and MART-1 antibodies; DAPI staining highlights cell nuclei (scale bar = 100 μm). (b) DNPH1 transcript levels in normal skin and cutaneous melanomas. Data were derived from The Cancer Genome Atlas and the Genotype-Tissue Expression project. Means with standard deviation are indicated (unpaired, two-tailed t test). (c) DNPH1 protein expression in primary site and metastatic site melanoma tumors. Examples of staining are on the left (scale bar = 200 μm), while quantification is on the right (unpaired, two-tailed t test; means with standard deviation). (d) High DNPH1 mRNA levels predict reduced disease-free survival for melanoma patients. Shown is an analysis done with GEPIA (Gene Expression Profiling Interactive Analysis) on The Cancer Genome Atlas data.

Role of DNPH1 in melanoma cells

Next, we assessed DNPH1 protein levels in several melanoma cell lines. Compared to immortalized, non-cancerous melanocytes, DNPH1 was overexpressed in six out of eight tested human melanoma cell lines (Figure 2a). Overexpression of DNPH1 in non-cancerous melanocytes had no impact on in vitro cell growth (Figure 2b), and the same held true for a catalytically compromised mutant of DNPH1, S128A (Doddapaneni et al, 2011; Dupouy et al, 2010). In contrast, when overexpressing DNPH1 or its S128A mutant in melanoma cells with low endogenous DNPH1 levels (SK-MEL-24 or RPMI-7951), we observed significantly increased in vitro cell growth (Figure 2c and d). Interestingly, in SK-MEL-24 cells, the S128A mutant was less effective in stimulating cell growth than wild-type DNPH1, whereas there was no difference between the two in RPMI-7951 cells. Similarly, in melanoma cells highly expressing DNPH1, wild-type and catalytically inactive ectopic DNPH1 exhibited equal growth stimulatory effects in G-361 cells, while the catalytically inactive mutant was slightly less efficient in SK-MEL-2 cells (Supplementary Figure S1a and b). Together, these findings imply that DNPH1 can stimulate melanoma cell growth either independent of its catalytic activity (RPMI-7951 and G-361) or through mechanisms that are partially dependent on its catalytic function (SK-MEL-24 and SK-MEL-2).

Figure 2. Modulation of DNPH1 levels affects melanoma cell growth.

(a) Western blots showing DNPH1 expression in immortalized, non-malignant human dermal melanocytes and eight different cutaneous melanoma cell lines. (b) Wild-type or catalytically impaired S128A DNPH1 was overexpressed in immortalized, non-malignant human dermal melanocytes. Levels of DNPH1 protein were monitored by Western blotting (left panels) and cell growth (right panel) was assessed with two-way ANOVA (Tukey’s multiple comparisons test; n=3). (c) and (d) Analogous for human SK-MEL-24 and RPMI-7951 melanoma cells, respectively. (e) DNPH1 downregulation in human G-361 melanoma cells with four different shRNAs. Left, Western blotting. Middle, cell growth assay (n=3; two-way ANOVA with Dunnett’s multiple comparisons test). Right, clonogenic assay (n=3; one-way ANOVA with Dunnett’s multiple comparisons test). (f) Similar, DNPH1 downregulation in SK-MEL-2 cells (n=5 for cell growth assays and n=3 for clonogenic assays). Shown are means with standard deviation; ns, not significant; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

We also downregulated DNPH1 with four different shRNAs in three melanoma cell lines displaying relatively high levels of endogenous DNPH1: G-361, SK-MEL-2 and A-375 (Figure 2e and f and Supplementary Figure S1c). DNPH1 knockdown significantly reduced both cell growth and clonogenic activity. Altogether, these results indicate that DNPH1 can exert a pro-growth function in established melanoma cell lines in vitro.

Role of DNPH1 in UV light-induced melanomagenesis

Cutaneous melanoma is often induced by UV radiation and half of these tumors display the oncogenic V600E mutation in the BRAF kinase (Cancer Genome Atlas, 2015; Oba and Woodman, 2021). Hence, we elected to test how DNPH1 ablation would affect melanoma formation in the context of BRAF^V600E expression plus UV light exposure that also induces melanomas in mice (Trucco et al, 2019; Viros et al, 2014). For this, we used the Braf^CA allele, in which expression of the constitutively active (CA) V600E allele of Braf is normally precluded by a Lox-Stop-Lox cassette. Crossing with a skin-specific, tyrosinase (Tyr) promoter-based, tamoxifen-inducible Cre driver (Tyr::CreER^T2) allows the induction of Braf^CA expression in the skin.

For this, we applied 4-hydroxytamoxifen twice on the back at 9 and 10 weeks of age and then exposed these mice to UV-B light once per week over the following 22 weeks in a way mimicking minor sunburns. The humane endpoint was set at 250 days after the last 4-hydroxytamoxifen application. Tumor initiation was defined as positive once tumors could be reliably observed with the naked eye, which was when a tumor diameter of 3 mm had been reached. Excitingly, both male and female Tyr::CreER^T2; Braf^CA/CA; Dnph1^−/− mice were significantly less prone to melanoma initiation upon UV-B light exposure than respective control Tyr::CreER^T2; Braf^CA/CA mice (Figure 3a and Supplementary Figure S2a). Interestingly, male Dnph1^−/− mice were more protected from melanoma formation than female Dnph1^−/− mice, while there was no sex difference for Dnph1^+/+ mice (Figure 3a). In addition, heterozygous Dnph1^+/− male mice displayed a significantly lower median tumor-free survival than homozygous Dnph1^−/− male mice, whereas female heterozygous and homozygous Dnph1 knockout mice were indistinguishable (Figure 3a). Importantly, homozygous Dnph1^−/− knockout also reduced distant metastasis to the liver or lung, although there was only a trend when separating the mice by sex (Figure 3b and Supplementary Figure S2b). These data demonstrate that DNPH1 is needed for efficient melanoma initiation as well as metastasis, furthering the notion that DNPH1 is a pro-tumorigenic protein.

Figure 3. Analysis of UV light-treated Tyr::CreER^T2; Braf^CA/CA; Dnph1 compound mice.

(a) Kaplan-Meier survival curves (with log-rank tests; ns, not significant) showing the impact of heterozygous or homozygous Dnph1 knockout. (b) Incidence of metastasis to the lung and liver. Only macroscopically visible metastases were scored; two-tailed Fisher’s exact test. (c) Time needed for tumors to grow from 3 mm to 12 mm in diameter; shown are means with standard deviation (unpaired, two-tailed t test). (d) Percentage of cells staining positive for Ki67 in dissected primary-site melanomas; shown are means with standard deviation (unpaired, two-tailed t test).

We also determined the growth rate of the melanoma tumors after they became noticeable with the naked eye. This surprisingly indicated that Dnph1 knockout accelerated tumor growth in both male and female mice (Figure 3c). Consistently, we observed that compared to Tyr::CreER^T2; Braf^CA/CA tumors, Tyr::CreER^T2; Braf^CA/CA; Dnph1^−/− melanomas had more cells with positive staining of Ki67, a proliferation marker (Figure 3d and Supplementary Figure S2c). In contrast, there was no difference in CD31 staining, a marker of angiogenesis (Supplementary Figure S2d and e). These data suggest that DNPH1 stifles melanoma growth once these tumors have arisen, revealing a tumor-suppressive facet of DNPH1 action.

UV exposure without BRAF^V600E expression does not induce melanomas in mice (Trucco et al, 2019; Viros et al, 2014). We assessed if this would change upon Dnph1 knockout. For this, we followed the same treatment strategy with 4-hydroxytamoxifen application and UV-B light exposure as above, but the number of exposures was increased from 22 to 50. Even at 1.5 years of age, none of these mice developed any melanomas. However, likely due to the harm that UV light had caused to lymphocytes in the exposed, shaved skin, we noted the generation of huge lymph nodes. And interestingly, both male and female Dnph1^−/− mice had larger lymph nodes than respective wild-type mice (Supplementary Figure S3), which would again point at a tumor-suppressive function of DNPH1.

Cancer-critical roles of DNPH1 upon ablation of TP53

Infrequent excessive sun light exposure – possibly even just one sunburn – also causes melanoma formation as long as driver mutations such as BRAF-V600E are established. This scenario has been simulated with Tyr::CreER^T2; Braf^CA/CA; Trp53^flox/flox mice and 4-hydroxytamoxifen application without UV light treatment (Goel et al, 2009). Notably, this model closely resembles human melanoma, as the TP53 tumor suppressor is mutated in ~15% of all cases, often in conjunction with the presence of the BRAF^V600E mutation (Cancer Genome Atlas, 2015). So, we applied 4-hydroxytamoxifen once onto the dorsal skin of Tyr::CreER^T2; Braf^CA/CA; Trp53^flox/flox; Dnph1 compound mice at 5 weeks of age and then monitored tumor formation for 250 days thereafter.

Overall, we noted that melanoma initiation occurred faster in female than in male mice, but this was independent of the Dnph1 status (Figure 4a and Supplementary Figure S4a). Importantly, loss of one or two Dnph1 alleles caused a delay of tumor initiation in both sexes, with no difference between heterozygous and homozygous Dnph1 knockout mice (Figure 4a). No significant difference in metastasis to the liver and lung was noted (Figure 4b). However, the time primary tumors needed from detection at 3 mm diameter to growing to a diameter of 7 mm was reduced upon Dnph1 knockout (Figure 4c) and this was paralleled by greater Ki67 positivity (Figure 4d), while CD31 positivity was not changed (Supplementary Figure S4b). Collectively, these data indicate that DNPH1 promotes melanoma initiation, but delays subsequent tumor growth in the Tyr::CreER^T2; Braf^CA/CA; Trp53^flox/flox model, which is similar to what we observed in UV light-treated Tyr::CreER^T2; Braf^CA/CA mice.

Figure 4. Analysis of Tyr::CreER^T2; Braf^CA/CA; Trp53^flox/flox; Dnph1 compound mice.

(a) Kaplan-Meier survival curves (with log-rank tests; ns, not significant) showing the impact of heterozygous or homozygous Dnph1 knockout. (b) Incidence of metastasis to the lung and liver. Only macroscopically visible metastases were scored; two-tailed Fisher’s exact test. (c) Time needed for tumors to grow from 3 mm to 7 mm in diameter; shown are means with standard deviation (unpaired, two-tailed t test). (d) Percentage of cells staining positive for Ki67 in dissected primary-site melanomas; shown are means with standard deviation (unpaired, two-tailed t test).

Potential modes of DNPH1 action

To probe for DNPH1’s mechanism of action in an unbiased way, we either overexpressed or downregulated DNPH1 in G-361 melanoma cells. Resulting changes in various signaling pathways were then assessed. This was done with Phospho Explorer antibody arrays encompassing 1,318 different site- and phospho-specific antibodies. From this screen, we obtained 86 DNPH1-modulated hits that were components of multiple pathways (Supplementary Figure S5). Notably, “Melanocyte Development and Pigmentation Signaling” and “Melanoma Signaling” were stimulated by DNPH1, which provides further indication that DNPH1 may promote melanomagenesis. Furthermore, UV-B- and UV-C-induced MAP kinase signaling appeared to be activated by DNPH1 (Supplementary Figure S5), suggesting that DNPH1 might be involved in DNA damage repair after UV light exposure. This was further supported by RNA-sequencing of melanomas isolated from UV light-treated Tyr::CreER^T2; Braf^CA/CA; Dnph1 compound mice, pointing again at stimulation of UV-B- and UV-C-induced MAP kinase signaling by DNPH1 (Supplementary Figure S6). All this prompted us to test if DNPH1 would be capable of affecting DNA damage repair.

For this, we exposed G-361 melanoma cells to UV light and noticed subsequent nuclear translocation of a large proportion of DNPH1, which was normally predominantly localized within the cytoplasm (Figure 5a and b). This further suggested a potential role of DNPH1 in nuclear DNA damage responses. Consistently, modulation of DNPH1 levels revealed that it functionally impacts on DNA repair pathways (Figure 5c and d): it repressed nucleotide excision repair (NER), but stimulated repair through non-homologous end joining (NHEJ) and homologous recombination (HR).

Figure 5. DNPH1 and DNA repair.

(a) Examples of HA-DNPH1-overexpressing G-361 cells before and after treatment with UV light (scale bar = 20 μm). Left, staining with Hoechst 33342 dye. Right, immunofluorescent staining with anti-HA antibodies. (b) Respective quantification. Influence of DNPH1 downregulation (c) or overexpression (d) on DNA repair (n=4; two-way ANOVA with Dunnett’s multiple comparisons test). Bottom panels show Western blots for DNPH1 and GAPDH. Shown are means with standard deviation; ns, not significant; *, P<0.05; **, P<0.01; ****, P<0.0001.

Results from the Phospho Explorer antibody arrays and the RNA-sequencing experiments additionally indicated that DNPH1 positively regulates NF-κB signaling (Supplementary Figures S5 and S6). Furthermore, we found that Dnph1 knockout reduced expression of the NF-κB pathway activating kinases, IKKα and IKKβ, and accordingly less phosphorylation was observable on their substrate, IκBα (Figure 6a), indicating attenuation of the NF-κB pathway. Moreover, we identified ANXA2 and FHL2 as interactants of DNPH1 (Figure 6b) and both ANXA2 and FHL2 are known to potentiate NF-κB signaling (Dahan et al, 2013; Sarkar et al, 2011; Stilo et al, 2002; Zhao et al, 2023). Hence, we tested if DNPH1 would promote NF-κB-mediated transcription. Indeed, DNPH1 on its own slightly activated NF-κB-mediated transcription and this was significantly enhanced by ANXA2 or FHL2 (Figure 6c). These data suggest that DNPH1 may affect melanoma development through stimulation of NF-κB signaling. Please note that results from the Phospho Explorer antibody arrays and of the RNA-sequencing data (see Supplementary Figures S5 and S6) revealed many other potential mechanisms by which DNPH1 might influence melanomagenesis, including activation of JAK/STAT or mTOR signaling or repression of PTEN signaling.

Figure 6. DNPH1 and the NF-κB pathway.

(a) Western blots with extracts from melanomas of UV light-treated Tyr::CreER^T2; Braf^CA/CA (+/+) and Tyr::CreER^T2; Braf^CA/CA; Dnph1^−/− (−/−) mice. On the right, quantification (n=6; unpaired, two-tailed t test). (b) Interaction of DNPH1 with ANXA2 or FHL2. 6Myc-tagged DNPH1 was coexpressed with Flag-tagged ANXA2 or FHL2. After anti-Flag immunoprecipitation (IP), coprecipitated DNPH1 was revealed by anti-Myc Western blotting. Bottom two panels show input levels of the tagged proteins. (c) Activation of a NF-κB luciferase reporter gene by ANXA2, FHL2 and/or DNPH1; one-way ANOVA with Tukey’s multiple comparisons test (n=3). Shown are means with standard deviation; *, P<0.05; **, P<0.01; ***, P<0.001; ****, P<0.0001.

DISCUSSION

The role of DNPH1 in skin cancer had remained unknown. Here, we have provided evidence that DNPH1 exerts context-dependent functions in cutaneous melanoma, promoting tumor initiation and metastasis while restraining the growth of established tumors in vivo. This differs from breast, prostate and lung cancers, where DNPH1 has thus far been characterized to solely promote tumorigenesis (Oh et al, 2026; Wang et al, 2025). Consistent with a role as a tumor promoter, Dnph1 knockout delayed the onset of tumorigenesis in both utilized mouse melanoma models and additionally reduced metastasis in the UV light-induced model. On the other hand, we found that DNPH1 inhibited growth of established tumors, which was opposite to our results showing DNPH1 to enhance in vitro growth of melanoma cell lines. We hypothesize that this discrepancy may reflect fundamental differences between in vitro tissue culture systems and in vivo tumor biology, particularly the contributions of the tumor microenvironment and the immune system that are absent in vitro. Other explanations may apply, including that established cell lines are probably further along the evolutionary pathway to malignant transformation compared to cells in tumors being in the process of initial growth and different activities of DNPH1 might be impactful in such different circumstances; for instance, the ability of DNPH1 to stimulate oncogenic NF-κB signaling might be its dominant activity in established cell lines in vitro, thereby acting as a growth promoter, while suppression of NER may be DNPH1’s predominant activity in developing melanomas in vivo, thereby potentially compromising their growth.

Speculative mechanism 1: influencing NF-κB signaling

The NF-κB signaling pathway is upregulated in melanoma tumors, forms part of their UV response and is causally linked to melanoma development (Cooper and Bowden, 2007; Janostiak et al, 2017; McNulty et al, 2004; Yang and Richmond, 2001; Yang et al, 2010). We found that DNPH1 can interact with ANXA2 and FHL2, which are modulators of the NF-κB pathway (Dahan et al, 2013; Qiao et al, 2009; Sarkar et al, 2011; Stilo et al, 2002; Wang et al, 2020; Zhao et al, 2023), and DNPH1 cooperated with them to stimulate NF-κB-mediated transcription. This underpins the hypothesis that activation of the NF-κB pathway may contribute to the tumor-promoting activities of DNPH1, and consistently we found that Dnph1 knockout diminished levels of the NF-κB upstream kinases IKKα and IKKβ and also reduced phosphorylation of their substrate, IκBα. Notably, as we observed for DNPH1, UV light can induce the nuclear translocation of ANXA2 and FHL2 (Madureira et al, 2012; Morlon and Sassone-Corsi, 2003), suggesting that these interactants could also mediate a role of DNPH1 in the cellular response to DNA damage. Altogether, by enhancing NF-κB signaling within melanocytic tumor cells, DNPH1 may foster melanoma initiation, growth and metastasis.

While this could explain our findings that DNPH1 promotes melanoma initiation and metastasis, it does not readily explain why DNPH1 stifles the growth of established tumors in vivo. However, as we utilized a whole-body Dnph1 knockout for our mouse melanoma models, DNPH1 expression was abolished not only in melanocytes, but also in immune cells. Further, we observed an increase in lymph node size in UV light-irradiated Dnph1^−/− compared to wild-type mice, implying that DNPH1 can exert anti-tumor functions in immune cells. Consequently, we hypothesize that Dnph1 knockout could reduce the ability of immune cells to suppress tumor growth. If this would be a dominant effect, it may account for the accelerated growth of established melanomas observed upon DNPH1 ablation. Notably, NF-κB signaling can also reprogram immune cells leading to tumor suppression at later disease stages (Mao et al, 2025). As such, DNPH1 ablation in immune cells may subdue their NF-κB signaling and thereby diminish tumor immune surveillance. In this model, DNPH1 would promote early melanomagenesis in a tumor cell-intrinsic manner while supporting host-mediated immune control of established tumors. However, future studies are needed to substantiate this model through rigorous experimental data. For instance, we would expect that allografted syngeneic melanoma cells should grow faster in our whole-body Dnph1^−/− mice than in wild-type mice if DNPH1 within immune cells would exert tumor-suppressive activities. Also, we predict that infiltration of tumor-suppressing immune cells into melanomas would be diminished in Dnph1^−/− mice compared to wild-type mice.

Speculative mechanism 2: impact on DNA repair

UV light-induced DNA damage, including (6–4)-pyrimidine-pyrimidone adducts and cyclobutane pyrimidine dimers leading to DNA crosslinks, is predominantly repaired through the NER pathway (Sun et al, 2020). Our data suggest that DNPH1 may suppress NER, which could plausibly increase mutation acquisition after UV light exposure and thereby facilitate melanoma initiation; consistently, we observed that DNPH1 positively affected tumor initiation in UV light-treated Tyr::CreER^T2; Braf^CA/CA mice. However, suppression of NER after initial transformation of melanocytes may also impede their survival and thereby reduce tumor growth. Moreover, DNPH1 might also stimulate HR and NHEJ, which may partially compensate for the suppression of NER. Please note that DNPH1 appears to also facilitate DNA repair in prostate cancer cells, yet it was not explored which DNA repair pathway(s) was affected by DNPH1 (Wang et al, 2025). Interestingly, we observed that UV light induced the translocation of DNPH1 into cell nuclei, where DNA repair primarily occurs. However, more studies are needed to provide definite evidence that modulation of DNA damage repair is a mechanism by which DNPH1 affects melanomagenesis. It should be noted that UV light induces various signaling pathways, which could potentially lead to posttranslational modification of DNPH1. It is conceivable that those modifications induce the observed nuclear translocation of DNPH1, and likewise posttranslational modifications, triggered by oncogenic changes within melanocytes or imposed by the tumor microenvironment and immune cells, may affect the ability of DNPH1 to modulate NF-κB signaling.

DNPH1 enzymatic activity: to be or not to be needed?

An unanswered question is whether DNPH1 enzymatic activity is required for its in vivo functions. In RPMI-7951 and G-361 melanoma cells, wild-type and catalytically inactive DNPH1 were equally capable of stimulating cell growth, indicating that DNPH1 catalytic activity is dispensable in this regard. In contrast, in SK-MEL-24 and SK-MEL-2 cells, catalytically inactive DNPH1 was less active than wild-type DNPH1, but still slightly enhanced cell growth, indicating that DNPH1 stimulates cell growth in both an enzymatic activity-dependent and -independent manner in these two cell lines. Clearly, there are cell type-specific differences in the requirement of catalytic activity for DNPH1’s ability to promote melanoma cell growth. However, in light of the divergent growth-modulating roles of DNPH1 in melanoma cell cultures versus tumors, we caution against directly extrapolating conclusions from catalytically inactive DNPH1 mutants obtained in cell culture to the physiological setting in melanomas. While our study did not detail how DNPH1’s enzymatic activity shapes its growth-promoting activity, one speculative scenario might be that DNPH1’s ability to hydrolyze the cytotoxic 5-hydroxymethyl-2’-deoxyuridine 5’-monophosphate (Fugger et al, 2021) may be relevant in SK-MEL-24 and SK-MEL-2, but not in RPMI-7951 and G-361 cells. On the other hand, interaction of DNPH1 with ANXA2 and FHL2, which may stimulate the oncogenic NF-κB pathway, is unlikely to be dependent on catalytic activity of DNPH1, which could be why both wild-type and catalytically impaired DNPH1 stimulated growth in all melanoma cell lines tested.

Conclusions and limitations of this study

This study has revealed context-dependent cancer-critical roles for DNPH1 in cutaneous melanoma and pointed out hypothetical mechanisms of action of this highly understudied enzyme. Our results emphasize that DNPH1 promotes tumor initiation and metastasis, potentially through modulation of DNA repair pathway choice and NF-κB signaling, while restraining the growth of established tumors by possibly enhancing anti-tumor immune surveillance. These findings suggest that any therapeutic targeting of DNPH1 must be carefully stratified based on the stage of the disease. However, while our data suggest possible mechanisms of action, they do not prove that DNPH1 affects melanoma formation and progression through these processes, which is a major limitation of this study. Another limitation of our study is the reliance on oncogenic BRAF^V600E, and thus we excluded other cutaneous melanoma subtypes, including the mutant RAS, the mutant NF1 and the triple wild-type ones (Cancer Genome Atlas, 2015; Oba and Woodman, 2021). However, DNPH1 depletion also impaired growth of NRAS-mutated SK-MEL-2 melanoma cells, indicating that DNPH1 function may extend beyond BRAF-driven melanomas.

MATERIALS AND METHODS

Tissue staining

Paraffin-embedded human melanoma tissue microarray ME1004h was purchased from US Biomax. Metastatic samples in this tissue microarray were from lymph nodes (n=20), the parotid gland (n=1) and sole (n=1). Staining was done with rabbit polyclonal DNPH1 antibodies (HPA029675; Sigma-Aldrich) employing a Leica BOND-III machine (Kim et al, 2016). Other antibodies used for tissue staining were: MART-1 mouse monoclonal antibody M2–7C10 (NBP2–15197, Novus Biologicals); CD31 rabbit polyclonal antibody (ab28364, Abcam); and Ki67 rabbit monoclonal antibody SP6 (ab16667, Abcam). Immunohistochemical or immunofluorescence staining was done by the Stephenson Cancer Center Tissue Pathology Core. For quantitation, respective slides were handled with the Axioscan slide scanner (Zeiss) and its software employed for analysis (Oh et al, 2019).

Cell culture and virus production

Human melanoma cell lines A-375 (CRL-1619), CHL-1 (CRL-3619), G-361 (CRL-1424), RPMI-7951 (HTB-66), SH-4 (CRL-7724), SK-MEL-2 (HTB-68), SK-MEL-3 (HTB-69) and SK-MEL-24 (HTB-71) were obtained from the American Type Culture Collection and grown in DMEM media supplemented with 10% fetal bovine serum, 100 units/ml penicillin and 100 μg/ml streptomycin at 37°C (Shin et al, 2009). For growth of hTERT-immortalized dermal melanocytes (CRL-4059), Dermal Cell Basal Medium (PCS-200–030, American Type Culture Collection) supplemented with the Adult Melanocyte Growth Kit (PCS-200–042, American Type Culture Collection) was utilized. Lenti-/retrovirus for cDNA or shRNA expression was produced according to standard protocols (Kim et al, 2014) and concentrated by PEG8000 precipitation (Berry et al, 2014); the sequences targeted by the DNPH1 shRNAs, which were embedded into the miR30 gene (Shin et al, 2006), have been listed in Supplementary Table S1. Cells were infected once with virus and cultivated for two days and then selected for three days with 200 μg/ml hygromycin B or 1 μg/ml puromycin (Li et al, 2018). These transduced cells were then employed for experimental procedures described below. In order to prevent genetic drift, the transduced cell lines were always made fresh and no stocks were generated.

Assays on melanoma cells

Cells were detached from plates with trypsin/EDTA and counted. Equal numbers of cells were seeded into 96-well plates for growth assays or into 6-well plates for clonogenic assays (Sui et al, 2020). Cell growth was measured with the PrestoBlue kit (ThermoFisher Scientific) at indicated time points after seeding the cells (Sui et al, 2023). For clonogenic assays, cells were cultured until colonies became readily visible after about two weeks, with change of media every three days, and colonies were stained with crystal violet for visualization (Gu et al, 2023b).

Western blotting

Cells were lysed by boiling in Laemmli sample buffer (Kim et al, 2023). Tumors were homogenized with a VWR 200 Homogenizer Unit at 35,000 rpm speed, punching them up and down for 10–20 s on ice to completely suspend the tumor tissue into lysis buffer (50 mM Tris-HCl pH 7.4, 150 mM NaCl, 50 mM NaF, 0.5% NP-40, 10 μg/ml leupeptin, 1 μg/ml pepstatin A, 2 μg/ml aprotinin, 1 mM PMSF, 100 μM Na₃VO₄, 100 μM DTT). The homogenate was incubated on ice for 30–60 min and debris removed by centrifugation, with the supernatant being defined as the tumor lysate. Proteins in those lysates were resolved by sodium dodecyl sulfate polyacrylamide gel electrophoresis (Goueli and Janknecht, 2003) and then transferred onto PVDF membrane (Papoutsopoulou and Janknecht, 2000). After blocking, membranes were incubated with primary antibody followed by incubation with horseradish peroxidase-coupled secondary antibodies (Gu et al, 2025). Detection was done with enhanced chemiluminescence (Mooney et al, 2010a). IKKα/β rabbit polyclonal antibodies were from Santa Cruz Biotechnology (sc-7607), phospho-IκBα (Ser32) rabbit monoclonal antibodies were from Cell Signaling Technology (#2859), and rabbit actin antibodies were from Sigma-Aldrich (A2066).

Mouse work

Experimental mice were generated from intercrossing whole body Dnph1 knockout mice (Oh et al, 2026), which had been backcrossed for more than eight generations onto a C57BL/6 background, with Tyr::CreER^T2; Braf^CA/+; Pten^loxP/loxP mice (Jackson Laboratory #013590; originally mixed background, but then backcrossed onto C57BL/6) and with Trp53^loxP/loxP mice (Jackson Laboratory #008462; originally mixed background, but then backcrossed onto C57BL/6). To extract DNA for genotyping, about 2 mm of mouse tail was incubated in 200 μl of 0.05 M NaOH at 95°C for 45 min. After neutralization with 40 μl of 1 M Tris-HCl pH 8, 1 μl of this lysate was directly used for a 20 μl PCR genotyping reaction (Oh and Janknecht, 2012) with primers listed in Supplementary Table S2. PCR products were resolved by agarose gel electrophoresis and stained with ethidium bromide (Goel and Janknecht, 2003).

Isoflurane-anesthetized mice were first shaved on the back and then Veet hair removing cream was applied with a fine bristle brush. After 3–4 min, the cream and hair stubbles were removed with a fine bristle brush and wet gauze. The exposed skin was then dried with Kimwipes. To induce Cre recombinase activity, 10 μl of 50 mg/ml 4-hydroxytamoxifen in DMSO was absorbed onto a size one Filbert tip paint brush (after each paint, the brush was dried with Kimwipes) and painted onto the exposed skin. One day before UV-B light treatment, mice were shaved and hair removed as described above. For exposure, a UV-B (302 nm) light lamp was located 6 cm above the back of the mouse, while the head and tail of the mouse were protected by a light shield box. The dose of UV-B light was 1.8 kJ/m²; as the radiation intensity of the lamp was 30 J/(s*m²), the exposure time was 1 min.

Immunoprecipitations

Plasmids encoding tagged proteins were transiently transfected into 293T cells, which were then cultivated for about two days (Oh et al, 2013). Cells were lysed with 50 mM Tris-HCl pH 7.4, 150 mM NaCl, 50 mM NaF, 0.5% NP-40, 10 μg/ml leupeptin, 1 μg/ml pepstatin A, 2 μg/ml aprotinin, 1 mM PMSF, 20 μM Na₃VO₄, 0.1 mM DTT. After removal of debris by centrifugation, immunoprecipitations were done with Flag mouse monoclonal antibodies (M2, Sigma-Aldrich, F1804) as described (Mooney et al, 2010b). Coprecipitated proteins were detected through Western blotting with Myc rabbit polyclonal antibodies (A-14, Santa Cruz Biotechnology, sc-789) as described (Kim et al, 2012).

Antibody array analysis

Protein samples were isolated from G-361 melanoma cells and labeled with biotin with the Antibody Array Assay Kit (Full Moon Biosystems, KAS02) according to the manufacturer’s recommendations. These processed protein samples were then used to challenge Phospho Explorer antibody arrays (Full Moon Biosystems, PEX100). For detection, Alexa Fluor 555 streptavidin (Invitrogen, S-32355) was employed and the arrays were analyzed with a microarray scanner (Molecular Devices, GenePix 4100A). Fluorescence intensities were measured and normalized, with the ratio of each target protein to GAPDH being used for statistical analysis.

DNA damage assays

G-361 cells overexpressing HA-tagged DNPH1 were cultivated until they reached 70% confluency and then treated with 100 J/m² of UV-C light in a Stratalinker 2400 (Stratagene) without removal of media. After 2 h, cells were processed for staining as described (Knebel et al, 2006). To assay for DNA damage repair, human 293T cells were infected with lentivirus expressing DNPH1 cDNA or shRNA. After antibiotics selection, cells were additionally transiently transfected with DNA vectors allowing quantification of DNA repair. For assaying HR and NHEJ, pCVL SFFV-EF1s HA.NLS.Sce(opt) (Addgene #31479), pLCN DSB Repair Reporter (DRR) (Addgene #98895) and pCAGGS DRR mCherry Donor EF1a BFP (Addgene #98896) were used (Arnoult et al, 2017; Certo et al, 2011); repair of DNA double-strand breakage induced by the I-SceI endonuclease could thereby be assessed through the GFP/BFP signal ratio (for NHEJ) and the mCherry/BFP signal ratio (for HR). For assaying NER, an EGFP expression vector was irradiated with UV-C light (800 J/m²) in a Stratalinker 2400, which led to extensive DNA crosslinking and thereby blockade of any EGFP gene transcription after initial cellular uptake. This damaged plasmid was cotransfected with a mCherry expression vector into 293T cells, wherein the crosslinks within the EGFP gene became slowly removed due to cellular NER activity; hence, the EGFP/mCherry signal ratio was reflective of NER activity (Nagel et al, 2014). Cells were trypsinized 24 h after transfection and fixed for 30 min with 3.7% formaldehyde in PBS + 2% sucrose at 4°C under continuous tumbling. After three washes with PBS, cells were resuspended in PBS and fluorophore-positive cells detected and counted with the Stratedigm S1400Exi flow cytometer platform.

RNA-sequencing

Primary melanoma tumors were isolated from UV light-treated Tyr::CreER^T2; Braf^CA/CA and Tyr::CreER^T2; Braf^CA/CA; Dnph1^−/− mice. Total RNA was similarly isolated with Trizol (DiTacchio et al, 2012) and further purified with the RNeasy Mini kit (Qiagen). RNA-sequencing was done by Novogene (Sacramento, California, USA) and analysis performed as described (Gu et al, 2023a).

Luciferase assays

Human 293T cells were transiently transfected utilizing polyethylenimine in 6-well plates (Oh et al, 2021). The following amounts of DNA were transfected: 50 ng NF-κB-luciferase reporter plasmid, 1.7 μg KS⁺ as a carrier, 0.2 ng RelA expression vector, 20 ng pQCXIH or pQCXIH-Flag-ANXA2 or pQCXIH-Flag-FHL2, and 40 ng pQCXIP or pQCXIP-DNPH1. Two days after transfection, cells were lysed in 300 μl of 25 mM Tris pH 7.8, 2 mM EDTA, 1% Triton X-100, 10% glycerol, 2 mM DTT (De Haro and Janknecht, 2002). Luciferase activity in these lysates was then determined with a luminometer (Oh et al, 2020).

Statistics

Statistical analyses were performed with R-studio or GraphPad Prism 6.0h. Statistical tests used have been indicated in the figure legends. If a P value of less than 0.05 was achieved, statistical significance was assumed.

Abbreviations:

CA

constitutively active

DNPH1

2’-Deoxynucleoside 5’-phosphate N-hydrolase 1

HR

homologous recombination

MAP

mitogen-activated protein

MAPKAP

MAP kinase-activated protein

NER

nucleotide excision repair

NHEJ

non-homologous end joining

UV

ultraviolet

Tyr

tyrosinase

DATA AVAILABILITY STATEMENT

RNA-sequencing data (BioSample: SAMN31648295) are freely accessible at the National Center for Biotechnology Information as BioProject PRJNA880320 (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA880320/). All other data that support the findings of this study are available from the corresponding author upon reasonable request.