USP35 is overexpressed in CRC patients

To explore the role of USP35 in CRC, we first checked its expression in CRC patients. According to the TCGA and GEO (GSE37182, GSE21815, and GSE71187) databases, USP35 mRNA levels were increased in CRC patients (Figs. 1A, B and S1). We further examined USP35 protein expression in our CRC patient samples through immunohistochemistry (IHC) staining and western blotting (WB). Consistently, the USP35 expression was higher by approximately two folds in the cancerous tissues compared to the non-cancerous tissues (Fig. 1C, D). Moreover, the TCGA data also suggested that high expression of USP35 was associated with increased cancer mutational in CRC patients and higher recurrence rate in CRC patients receiving postoperative chemotherapy (Fig. 1E, F). These data suggest that USP35 is elevated in CRC, and overexpression of USP35 is associated with CRC progression and recurrence, potentiating USP35 as a cancer target in CRC.

Fig. 1: USP35 is overexpressed in CRC patients.

A, B USP35 mRNA levels were higher in CRC patients according to the TCGA (A) and GEO (GSE37182) databases (B). C The representative immunohistochemistry staining of USP35 in CRC tissues and adjacent tissues (n = 30). Quantitative analysis was shown in the graphs. D The expression levels of USP35 in CRC tissues and adjacent tissues were detected by western blotting (n = 6). Quantitative analysis was shown in the graphs. E High expression of USP35 was associated with increased cancer stages in CRC patients based on TCGA database. F The expression of USP35 was higher in the patients with recurrent (recur) CRC compared the patients with nonrecurrent (nonrecur) CRC after receiving chemotherapy. The data were extracted from the TCGA database. All data are presented as mean ± SD. *p < 0.05, ***p < 0.001 based on the Student’s t test.

USP35 promotes CRC cell proliferation and chemo-resistance

To understand the impact of USP35 on CRC cell proliferation, we overexpressed USP35 in LoVo and HT29 cells, and depleted USP35 in DLD-1 and HCT116 cells (Figs. 2A and S2A). The cell proliferation was monitored by Cell Counting Kit-8 (CCK-8) and clonogenic assays. We found that augmented expression of USP35 promoted cell viability as well as clonogenic activity in LoVo and HT29 cells (Fig. 2B, C). USP35 deficiency, however, dampened the proliferation in DLD-1 and HCT116 cell lines, as indicated in Fig. S2B, C. These results show that USP35 is required for CRC cell proliferation.

Fig. 2: USP35 promotes CRC cell proliferation and chemo-resistance.

A Establishment of USP35-overexpressed LoVo and HT29 cell lines. Quantitative analyses were shown in the graphs. B The cell viability of USP35-overexpressed LoVo and HT29 cells. Cells were implanted in 96-well plates for 48 h (n = 6) and the cell viability was assessed using CCK8 assay. C Representative images of clonogenic assay (n = 3). USP35 overexpression promoted cell proliferation. Quantitative analyses were shown in the graphs. The scale bars in represented 50 μm. D The USP35-overexpressed HT29 cells and the control cells were treated with different concentrations of oxaliplatin (OXA), 5-fluorouracil (5-FU), or combination of OXA and 5-FU for 48 h (n = 6). The cell viability was assessed using CCK8 assay. E The USP35-overexpressed HT29 cells and the control cells were treated with OXA (10 μM), 5-FU (10 μM), or combination of OXA (10 μM) and 5-FU (10 μM) for 24, 48, 72, or 96 h (n = 6). The cell viability was assessed using CCK8 assay. F, G The USP35-overexpressed HT29 cells and the control cells were treated with DMSO, OXA (10 μM), 5-FU (10 μM) or combination of OXA (10 μM) and 5-FU (10 μM) for 48 h (n = 3). Representative images indicated the apoptotic cells detected by flow cytometry analysis (F) and TUNEL staining (G). Data are presented as mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ###p < 0.001 based on the Student’s t test.

Since USP35 expression was comparatively higher in CRC recurrent cases (Fig. 1F), we next investigated whether USP35 contributed to chemo-resistance in CRC cell lines. Oxaliplatin (OXA) and 5-fluorouracil (5-FU) are classical drugs in the CRC chemotherapy regimens, either as single or combination therapy [3]. We therefore examined whether USP35 affected CRC cell response to OXA, 5-FU, or the combined treatment of OXA and 5-FU. We first tested whether increased USP35 expression conferred resistance to OXA, 5-FU, or combination of OXA and 5-FU in HT29 cell lines. We treated the HT29 cells using single or double agents with different concentration or at different time points, and examined the cell viability through CCK-8 assay. We found that USP35 overexpression rendered HT29 cells more resistant to the cytotoxic agents, either on the condition of single or combined treatments (Fig. 2D, E). We then tested whether USP35 depletion sensitized cells to the chemotherapies using the similar methods, and found that USP35 ablation enhanced cytotoxic effects of those drugs on the DLD-1 cells (Fig. S2D, E). The impact of USP35 on cell apoptosis in response to OXA, 5-FU, or combination of OXA and 5-FU was further detected by Annexin V/Propidium Iodide (PI) staining flow cytometry analysis and terminal dUTP nick-end labeling (TUNEL) assay. Strikingly, USP35 overexpression markedly reduced cytotoxic agents-induced cell apoptosis (Fig. 2F, G), while USP35 knockdown led to increased drug-induced cell apoptosis (Fig. S3C, D). The quantified data of apoptotic cells were presented in Fig. S3A–D. In parallel, western blotting analysis showed that OXA, 5-FU, or combination of OXA and 5-FU treatment led to significant increases in the levels of the apoptosis markers, cleaved PARP1 and cleaved Caspase-3, while forced USP35 expression in HT29 cells reduced cytotoxic agents-induced increase of these proteins (Fig. S4A). On the contrary, USP35 silencing in DLD-1 cells dramatically boosted the increase of these apoptotic markers induced by these chemotherapeutic drugs, further supporting the chemo-resistant role of USP35 in CRC (Fig. S4B).

USP35 deubiquitinates and stabilizes FUCA1

To investigate the possible mechanism for USP35-mediated cell proliferation and drug resistance, we transfected Flag-tagged USP35 expression plasmid into the HCT116 cell line, immunoprecipitated the USP35-associated proteins with anti-Flag antibody, separated the proteins by SDS-PAGE, and analyzed the immunoprecipitated proteins by mass spectrometry (Fig. 3A). Mass spectrometry identified seven candidates (PRMT5, PPM1B, FUCA1, DNAJA1, IGF2BP1, DDX5, XRCC6) which may be associated with USP35 (Fig. S5A). Among these potential targets of USP35, we would like to specifically focus on α-L-fucosidase 1 (FUCA1) in our current study (Figs. 3B and S5B, C), considering the strong relevance of altered glycosylation in cancer development and therapeutics [24].

Fig. 3: USP35 interacts with FUCA1.

A Silver staining of the SDS-PAGE gel containing anti-Flag antibody immunoprecipitated proteins in Flag-USP35-overexpressed HCT116 cells and the control cells. B Representative map of mass spectrometry peaks of FUCA1 interacting with Flag-USP35. C, D Exogenous interaction between USP35 and FUCA1. HEK293T cells were co-transfected with Flag-FUCA1 and/or Myc-USP35 plasmids. Cell lysates were immunoprecipitated with indicated antibodies, followed by immunoblotting with anti-Flag antibody (C) or anti-Myc antibody (D). E, F Endogenous interaction between USP35 and FUCA1. Cell lysates from LoVo (E) and DLD-1 (F) were immunoprecipitated with anti-USP35 antibody or IgG antibody, followed by immunoblotting with anti-FUCA1 antibody. G, H Proximity ligation assay (PLA) showing USP35 and FUCA1 interaction in LoVo (G) and DLD-1 (H) cells. The results were the representative of three independent experiments.

To validate our screened results, we first examined whether the physical interaction existed between USP35 and FUCA1. Co-immunoprecipitation (co-IP) assay was used for this purpose. HEK293T cells were transfected with Myc-tagged USP35 and/or Flag-tagged FUCA1, the USP35- or FUCA1-associated proteins were immunoprecipitated with anti-Myc or anti-Flag antibodies, and the lysates were subject to western blotting analysis. As illustrated in Fig. 3C, D, USP35 interacted with FUCA1 in a reciprocal fashion. The interaction between endogenous USP35 and FUCA1 was also detected in LoVo and DLD-1 cells (Fig. 3E, F). Moreover, we checked the interaction of endogenous USP35 and FUCA1 in CRC cells using proximity ligation assay (PLA). The specific positive signals (shown in red) were observed in the cells, further indicating the physical association between USP35 and FUCA1 (Fig. 3G, H). We also examined the localization of USP35 and FUCA1 by immunofluorescence (IF) staining followed by confocal microscopy, and found that USP35 and FUCA1 were expressed in both cytosol and nucleus (Fig. S5D, E).

Considering USP35 is a deubiquitinase, we then investigated whether FUCA1 was deubiquitinated by USP35. We transiently transfected HEK293T cells with Flag-tagged FUCA1, HA-tagged wild-type USP35 (USP35 WT), or HA-tagged catalytically inactive USP35 (USP35 C450A) [34] plasmids, immunoprecipitated FUCA1 associated proteins with anti-Flag antibody, and probed for ubiquitin (Ub) by western blotting. The proteasome inhibitor MG132 was used to enrich the USP35 and FUCA1 proteins in the cells. We found that only the catalytically active USP35 (USP35 WT), not the USP35 mutant (USP35 C450A), deubiquitinated FUCA1 (Fig. 4A). This result indicated that USP35 removed ubiquitins from FUCA1 and its enzymatic activity was required for this function. In addition, we silenced the USP35-overexpressed cells with different dose of shRNAs, and found that the ubiquitination of FUCA1 was increased with USP35 depletion in a dose dependent manner (Fig. 4B). This result further confirmed that FUCA1 was a deubiquitination target of USP35.

Fig. 4: USP35 deubiquitinates and stabilizes FUCA1.

A, B USP35 deubiquitinates FUCA1. A LoVo cells were transfected with Flag-FUCA1 alone or along with HA-tagged wild type USP35 (HA-USP35 WT) or the catalytically inactive USP35 (HA-USP35 C450A). Cell lysates were immunoprecipitated with anti-Flag antibody, followed by immunoblotting with anti-Ub antibody. MG132 was used to enrich the USP35 and FUCA1 proteins in the cells. B DLD-1 cells were transfected with Flag-FUCA1 alone or along with HA-USP35 WT and/or shUSP35-2 (0, 2 µg, or 4 µg). Cell lysates were immunoprecipitated with anti-Flag antibody, followed by immunoblotting with anti-Ub antibody. C, D USP35 increases the stability of FUCA1. LoVo cells expressing empty vector, HA-USP35 WT, or HA-USP35 C450A (C), and DLD-1 cells expressing USP35 specific scramble RNA, shUSP35-1, or shUSP35-2 (D) were treated with 50 μg/ml cycloheximide (CHX) for indicated time points. Then, the expression of FUCA1 was detected by western blotting. Quantitative analyses of CHX chase data are shown in the graphs. E, F USP35-mediated stabilization of FUCA1 is dependent on ubiquitin-proteasome system. USP35-overexpressed LoVo cells (USP35) and the control cells (EV) (E), and USP35-depleted DLD-1 cells (shUSP35-1 and shUSP35-2) and the control cells (Scramble) (F) were treated with or without 10 µM MG132 for 6 h. The expression of FUCA1 was detected by western blotting. Quantitative analyses were shown in the graphs. Data are presented as mean ± SD of three independent experiments. ns: not significant. *p < 0.05, **p < 0.01, ***p < 0.001 based on the Student’s t test.

Given the most widespread outcome of protein ubiquitination is the alteration of protein stability [35], we next investigated whether USP35 affected FUCA1 stability. Cycloheximide (CHX) chase assay was used to measure the stability of FUCA1 with different USP35 expression levels. We found that overexpression of wild-type USP35 (USP35 WT), not the catalytically inactive mutant USP35 (USP35 C450A), prevented FUCA1 from degradation in LoVo cells, whereas knockdown of USP35 accelerated FUCA1 degradation in DLD-1 cells (Fig. 4C, D). Resultantly, enhanced USP35 expression increased FUCA1 at protein levels in CRC cells and HEK293T in a dose dependent manner (Fig. S6A, B), while dampened USP35 expression clearly decreased the protein levels of FUCA1 in DLD-1 and HCT116 cells (Fig. S6C). As expected, USP35 overexpression or silencing did not affect the mRNA levels of FUCA1 (Fig. S6D, E). Since proteasome is required in the ubiquitin-proteasome pathway for protein degradation, we further tested how altered expression of USP35 would affect the levels of FUCA1 in the presence of MG132. As a result, forced expression of USP35 in LoVo and HT29 cells failed to “wax” FUCA1, and depletion of USP35 in DLD-1 and HCT116 cells failed to “wane” FUCA1 as it usually did with addition of MG132 (Fig. 4E, F), indicating USP35 protected FUCA1 from proteasome-mediated degradation. Overall, these results demonstrate that USP35 deubiquitinates FUCA1 and increases FUCA1 stability by preventing ubiquitin-proteasome pathway-mediated degradation of FUCA1.

FUCA1 mediates the effect of USP35 on cell proliferation and chemo-resistance in vitro

Considering the previously reported role of FUCA1 in cancer growth and chemo-resistance [27,28,29,30,31], we here investigated whether the influence of USP35 on cell proliferation and drug resistance was mediated or partially mediated by FUCA1 in our study.

We first knocked down FUCA1 in USP35-overexpressed LoVo and HT29 cells, using FUCA1-specific shRNA-4 with higher knockdown efficiency (Fig. S7A, B). As shown in Fig. 5A, FUCA1 expression was reduced to approximately the basal levels by FUCA1-specific shRNA-4 in the USP35-overexpressed LoVo and HT29 cells. Interestingly, FUCA1 ablation almost completely reversed proliferation-promoted effect of USP35 on LoVo and HT29 cells, as indicated by clonogenic and CCK-8 assays (Fig. 5B, C). Then, we stably overexpressed FUCA1 in USP35-depleted DLD-1 and HCT116 cell lines to restore FUCA1 expression to the near-basal levels (Fig. S8A). Consequently, augmented FUCA1 substantially ameliorated the growth-inhibitory effect of USP35 deficiency on these cells (Fig. S8B, C).

Fig. 5: FUCA1 mediates the effect of USP35 on cell proliferation and chemo-resistance in vitro.

A FUCA1-specific shRNA (shFUCA1-4) or control shRNA was introduced into the USP35-overexpressed HT29 and LoVo cells. The expression of FUCA1 and USP35 was detected by western blotting. Quantitative analysis was shown in the graphs. B, C CCK8 assay (B) and clonogenic assay (C) were used to detect the effect of FUCA1 depletion on USP35-induced cell proliferation. The scale bars in C represented 50 μM. D FUCA1 specific shRNA (shFUCA1-4) or control shRNA was introduced into the USP35-overexpressed HT29 cells. The cells were treated with different concentrations of OXA, 5-FU, or combination of OXA and 5-FU for 48 h (n = 6). The cell viability was assessed using CCK8 assay. E, F FUCA1-specific shRNA (shFUCA1-4) or control shRNA was introduced into the USP35-overexpressed HT29 cells. The cells were treated with DMSO, OXA (10 μM), 5-FU (10 μM), or combination of OXA (10 μM) and 5-FU (10 μM) for 48 h (n = 3). Representative images indicated the apoptotic cells detected by flow cytometry analysis (E) and TUNEL staining (F). Data are presented as mean ± SD. ns: not significant. *p < 0.05, **p < 0.01, ***p < 0.001 based on the Student’s t test.

Next, we investigated whether USP35-triggered OXA and 5-FU resistance was associated with FUCA1 expression. Given previously, FUCA1 was identified as a direct transcriptional target of p53 and was required for cisplatin sensitivity [27], we determined to understand whether FUCA1 expression/function was dependent on p53 status/activation. Unlike wild-type p53 which is a classic tumor suppressor, p53 mutant has oncogenic potentials [36], therefore we examined whether p53-FUCA1 axis exists in wild-type p53 (HCT116 and LoVo) and p53 mutant (DLD-1 and HT29) CRC cell lines. We first examined whether FUCA1 expression was increased by drug-induced p53 activation. We found that in HCT116 and LoVo (wild-type p53) cells, FUCA1 and p53 expression were concurrently triggered by OXA or 5-FU treatments (Fig. S9C, D). However, upon chemotherapeutic insult, FUCA1 expression was not induced in parallel with p53 expression in p53 mutant CRC cells (Fig. S9A, B). This finding indicated that mutant p53 failed to regulate FUCA1 expression in CRC cells. We next investigated whether the function of FUCA1 was affected by p53 status. We found FUCA1 depletion alone did not affect cell apoptosis in either p53 wild-type or mutant CRC cells (Fig. S10). However, upon drug treatments (single OXA, single 5-FU, or combined OXA and 5-FU), FUCA1 deficiency led to different results dependently on the p53. While FUCA1 knockdown made cell much more resistant to the drug treatments in LoVo (wild-type p53) cells, FUCA1 depletion moderately sensitized the HT29 cells (mutated p53) to the similar treatment (Fig. S10C, D). This observation suggested that the effect of altered FUCA1 expression on chemotherapeutic response was dependent on p53 status. As p53 mutation was found in approximately 60% of the CRC patients [36], we specifically focused on exploring the role of USP35-FUCA1 axis in p53 mutant DLD-1 and HT29 cells.

Cells were subject to OXA, 5-FU, or combination treatments using similar conditions as in Fig. 3, and the drug toxicity was measured by CCK-8 assay. As a result, FUCA1 depletion fully re-sensitized the USP35-overexpressed HT29 cells to different chemotherapeutic treatments (Figs. 5D and S11B). However, boosted FUCA1 expression largely de-sensitized USP35-depleted cells to the similar treatments in DLD-1 cells (Fig. S11A, C). Moreover, drug-induced apoptosis in these cells was illustrated by Annexin V/PI staining flow cytometry analysis and TUNEL assay. We found that depleting FUCA1 in USP35-overexpressed cells led to an increase of apoptotic cells to a rate as the control (EV + shNC) cells (Figs. 5E, F and S11D, E), while re-expressing FUCA1 in USP35-deficient cell decreased apoptotic cell population to the similar level as control (Scramble + Vector) cells in response to cytotoxic insults (Fig. S12A, B). Together, these data suggest that FUCA1 is an essential player that mediates USP35-associated cell proliferation and OXA/5-FU resistance in CRC cells.

FUCA1 mediates the impact of USP35 on tumor growth and chemo-resistance in vivo

To explore the role of USP35-FUCA1 axis in vivo, we subcutaneously inoculated the control (EV + shNC), USP35-overexpressed (USP35 + shNC), and USP35 overexpressed FUCA1-depleted (USP35 + shFUCA1-4) HT29 cells in the athymic nude mice, intraperitoneally injected water (H2O), single drug (OXA as a representative), or combined drugs (OXA + 5-FU) weekly into the mice, and routinely monitored the tumor growth. We observed that USP35 overexpression markedly promoted tumor growth and drug resistance, but depleting FUCA1 in USP35-overexpressed cells reversed the tumor growth to the basal levels, and re-sensitized the tumors to the chemo-treatments, as indicated by representative images, tumor weight and growth curve (Fig. 6A–C). Moreover, we performed immunohistochemical (IHC) staining of cleaved Caspase-3 and TUNEL assay in the xenograft samples. In parallel with the in vitro data, USP35 overexpression weakened drug-induced cell apoptosis while knockdown of FUCA1 in USP35-overexpressed cells restored drug-induced cell apoptosis, suggesting that USP35 contributes to OXA/5-FU resistance through stabilizing FUCA1 in vivo (Fig. 6D and S13A–C).

Fig. 6: FUCA1 mediates the impact of USP35 on tumor growth and drug resistance in vivo.

B Tumor growth curve of HT29 cells with or without altered expression of USP35 or FUCA1. Cells were subcutaneously injected into the nude mice (n = 5). A week after the inoculation, water, OXA (10 mg/kg), or OXA (5 mg/kg) plus 5-FU (50 mg/kg), was intraperitoneally injected in the mice weekly. Tumors were measured at the indicated time points. A Representative images of the dissected tumors were shown. A ruler is used to demonstrate the size of the tumors. C Quantification of tumor weights at the end point. D Cell apoptosis was detected using TUNEL assay of the tumor sections. E Representative immunohistochemistry (IHC) stainings of USP35 and FUCA1 in the human CRC tissues. F Box plot showing the relative FUCA1 levels in USP35-low and USP35-high patients (n = 30). Median USP35 or FUCA1 expression was defined as the cut-off point. G The correlation of the expression of USP35 and FUCA1 in the human CRC tissues by IHC staining (n = 30). H The expression levels of USP35 and FUCA1 in the human CRC tissues and adjacent tissues were detected by western blotting (n = 8). The data are shown as the mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, #p < 0.05, ##p < 0.01, ###p < 0.001 based on the Student’s t test.

Next, we examined the USP35-FUCA1 axis in the human CRC tissues. A total of 30 human CRC samples were included for IHC staining and analysis. As shown in the representative images, USP35 and FUCA1 shared the similar expression pattern (Fig. 6E). For example, in some samples, USP35 and FUCA1 were both intensely expressed in the CRC epithelium and lightly stained in the CRC stroma (Fig. 6E, sample 1). In the samples with low USP35 expression, FUCA1 expression was usually barely detectable (Fig. 6E, sample 2). According to the assessment by the pathologists, among 25 USP35-highly expressed CRC samples, 20 samples expressed elevated FUCA1, and among 5 USP35-lowly expressed CRC samples, 4 samples exhibited reduced FUCA1 expression (Fig. 6F). The strong correlation between USP35 and FUCA1 expression levels in the CRC tissues was further verified by Spearman’s correlation analysis (p < 0.001 and r = 0.665, Fig. 6G). We also examined the expression of USP35 and FUCA1 in human CRC tissues as well as adjacent normal tissues by western blotting, and found that FUCA1 expression was largely in accordant with USP35 expression (Fig. 6H). The quantified western blotting data and correlation analysis were shown in Fig. S13D, E. Therefore, these results suggest that USP35 expression correlates well with FUCA1 expression in CRC patient samples, and USP35-FUCA1 axis contributes to CRC tumor growth and chemo-resistance in vivo.

USP35-FUCA1 axis up-regulates nucleotide excision repair in CRC

Two studies demonstrated that USP35 deficiency sensitized cancer cells to cisplatin (the first platinum anti-cancer drug) treatment via de-stabilizing anti-apoptotic factor BIRC3 or activating cGAS-STING-TBK1-mediated expression of type I interferons [20, 23]. However, another study showed that FUCA1, as a direct transcriptional target of p53, was required for cisplatin-induced cell apoptosis [27]. Given that the latter study was in discrepant with our findings, which might be due to the different experimental settings, we here investigated the possible mechanism of USP35-FUCA1-mediated oxaliplatin (the third-generation platinum drug) resistance in CRC.

The Pt-based chemotherapies work by introducing DNA-platinum adducts, which is mainly corrected by nucleotide excision repair (NER) [37, 38]. Increased NER proficiency is one of the reasons accounting for on-target resistance of Pt-based therapy [37, 39]. We therefore examined the expression levels of some major components (e.g., XPC, XPA, ERCC1) of the NER system [40]. We found that in USP35-overexpressed LoVo and HT29 cells, the expression levels of XPC, XPA and ERCC1 were markedly increased (Figs. S14A, B and S15A, B). Accordingly, the XPC, XPA and ERCC1 levels were decreased by USP35 knockdown in a dose-dependent manner in DLD-1 and HCT116 cells (Figs. S14C, D and S15C, D). Moreover, ablating FUCA1 in HT29 cells dramatically reduced the expression of XPC, XPA and ERCC1, whereas enhancement of FUCA1 largely restored the level of these NER components (Figs. S14E, F and S15, F). In addition, to investigate the involvement of the NER pathway in USP35-mediated chemo-resistance, we silenced XPA in USP35-depleted FUCA1-overexpressed DLD1 cells using XPA-specific siRNA and examined cell apoptosis in response to chemotherapeutics by flow cytometry analysis. Strikingly, we found that XPA silencing partially rescue the phenotype, reinforcing the possibility that USP35-FUCA1 axis controls chemo-resistance by regulating the expression of NER components (Fig. S16A, B). However, the detailed mechanism how USP35-FUCA1 regulates the NER warrants further investigation.

Hence, our current results indicate that USP35-FUCA1 axis up-regulates NER in CRC cells, which may be a possible mechanism for USP35-FUCA1-mediated oxaliplatin resistance in CRC.