Rate-limiting enzymes of ketone metabolism affect gemcitabine sensitivity. Metabolic reprogramming is a hallmark of cancer due in part to the high energetic demands required to sustain tumor expansion. Specifically, ketone metabolites like acetoacetate and βHB are associated with increased staging (21). With the clinical significance of ketone metabolites in mind, we explored the expression of crucial genes for ketogenesis (HMGCS2) and ketolysis (OXCT1). A Kaplan-Meier plot demonstrated that elevated OXCT1 expression was significantly associated with poor survival in the TCGA urothelial BCa dataset (n = 408), and the Mattias Höglund BCa dataset (n = 308; Supplemental Figure 1, A and C; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.177840DS1) (22). Conversely, HMGCS2 expression was positively associated with survival in the same datasets (Supplemental Figure 1, B and D). In addition, the Markus Riester high-risk cohort BCa dataset (n = 93) revealed that OXCT1 was significantly expressed in invasive disease (pT4) while HMGCS2 was downregulated at higher disease stages, showing a correlation with muscle invasive disease (Figure 1, A and B) (23). To assess a relationship with chemoresistance, we focused on patients with MIBC who underwent neoadjuvant chemotherapy in the Peter C. Black dataset (n = 91), which showed that OXCT1 and HMGCS2 have opposing correlations with survival (Figure 1, C and D) (24). Altogether, OXCT1 and HMGCS2 consistently relate to the overall survival, staging, and, importantly, the outcome of chemotherapy treatment of patients with BCa.

OXCT1 and HMGCS2 expression in clinical sample datasets. (A and B) High-risk BCa Riester dataset (n = 93) was used to identify OXCT1 (A) and HMGCS2 (B) expression by disease stage (23). Statistical analysis done with 1-way ANOVA at 95 % CI. *P < 0.05, **P < 0.01, ***P < 0.001. (C) The Peter C. Black chemotherapy treated cohort (n = 91) was used to identify OXCT1 expression correlates with poor outcome within (P = 0.0012). (D) HMGCS2 expression correlates with survival after chemotherapy (P = 0.0020) (24).

DNA synthesis inhibitors like gemcitabine serve as a tool to examine the crosstalk between gene regulation and metabolism. Initially, we evaluated the expression of enzymes responsible for ketone body metabolism in a diverse panel of BCa cell lines and their effect on gemcitabine sensitivity. A transitional papillary line, RT4, was the most sensitive to gemcitabine with an IC50 of < 0.1 nM. A transitional carcinoma line, UM-UC3, was the most resistant with an IC50 of 600 nM, whereas the 5637 cell line (muscle invasive) and T24 (transitional carcinoma) lines had a very similar gemcitabine sensitivity with IC50 of 4 nM (Figure 2A). In this cell line panel, only RT4 cells had detectable HMGCS2 expression, while UM-UC3, the 5637 cell line, and T24 only expressed OXCT1 by Western blotting (Figure 2B). The cell line data support HMGCS2 expression in less aggressive gemcitabine-sensitive BCa and suggested that OXCT1 may play a role in more aggressive disease and gemcitabine resistance. Since the UM-UC3 had elevated OXCT1 expression and showed an intrinsic resistance to gemcitabine, we used CRISPR/CAS9 to generate a stable KO of OXCT1 (UM-UC3-OXCT1KO) cell line (Supplemental Figure 2A). The loss of OXCT1 expression induced gemcitabine sensitivity with a significantly lower IC50 of 2 nM (Figure 2C). Parental UM-UC3 and UM-UC3-OXCT1KO gene expression specific to drug resistance pathways was investigated with a PCR array. Data show differences in peroxisome proliferator activated receptor γ (PPARG); peroxisome proliferator activated receptor α (PPARA); fos-proto-oncogene, AP-1 TF subunit (FOS); hypoxia-inducible factor 1 subunit α (HIF1A); Erb-B2 receptor tyrosine kinase 2 (ERBB2); fibroblast growth factor 2 (FGF2); andNF-κB subunit 1 (NFKB1) (Figure 2D). These results suggest a possible metabolic shift in the UM-UC3-OXCT1KO. HIF1A is a metabolic regulator known to reduce oxidative processes and to inhibit PPARG in favor of an anaerobic metabolic shift (25). Using Western blotting, we confirmed that OXCT1 KO resulted in a decrease in PPARG expression as well as downstream CPT1, responsible for fatty acid entry into the mitochondria (26). Furthermore, HMGCS2 expression remained low in the UM-UC3-OXCT1KO cell line, suggesting that both ketogenesis and ketolysis were inhibited (Figure 2, E and F). Among other genes significantly altered by the OXCT1 KO, a member of the epidermal growth factor receptor (EGFR) family, ERBB2, was downregulated. This oncogene is known to be amplified in BCa. Interestingly, gemcitabine treatment is reported to elevate ERBB2 expression, making it a candidate target in breast, pancreatic, and biliary tract cancer treatment (27). The alterations in OXCT1 expression in the UM-UC3 line suggest its role in gemcitabine sensitivity.

OXCT1 Expression affects gemcitabine response. (A) BCa cell line panel and their response to gemcitabine. All cell lines were treated with gemcitabine in various logarithmic concentrations to determine their IC50 at 72 hours using an MTT assay. (B) OXCT1 and HMGCS2 expression was measured by Western blot in naive cells after 72 hours of culture. (C) Knocking out OXCT1 in UM-UC3 cell lines causes a significant shift in gemcitabine IC50. (D) Drug resistance PCR array shows significant changes in PPAR signaling pathway. (E) Western blot of parental UM-UC3 (P) and isogenic OXCT1KO (KO) have decreased expression of HMGCS2, PPARγ, and downstream CPT1. Calculations shown are normalized to β-actin and relative to the parental line. (F) PCR array and Western blot results suggest a metabolic shift that could decrease FAO in OXCT1-KO cells. (G) Cell viability of the isogenic 5637 cells were measured following incubation with gemcitabine at indicated concentrations.

Next, we developed an acquired resistance model with the 5637 cell line. The stable gemcitabine-resistant cell line (5637GR) was created through continuous gemcitabine exposure to achieve an IC50 > 100 μM. Expression analysis showed that 5637GR had significantly elevated OXCT1 expression when compared its isogenic parental line (P = 0.033; Supplemental Figure 2B). The subsequent KO of OXCT1 of the 5637GR cells using CRISPR/Cas9 (GR-OXCT1KO) resulted in a dramatic restoration of gemcitabine sensitivity, near to that of the 5637P line (5637P; IC50 = 0.001 μM; Figure 2G). Together, our data support a robust relationship between OXCT1 expression and gemcitabine resistance in both acquired and intrinsic models.

Metabolic changes induced by OXCT1 orchestrate gemcitabine response. Based on the role of OXCT1 in ketone catabolism, we examined the differential metabolic effect in the 5637 cell line acquired resistance model. For these studies, Seahorse XF measurements of oxygen consumption rate (OCR) were performed while serially manipulating the different mitochondrial respiratory chain complexes, as previously described (28). Basal, maximal, and spare respiratory capacity of 5637GR was significantly greater than that of the 5637P or the GR-OXCT1KO lines (Figure 3A). Interestingly, acute gemcitabine (100 nM) treatment caused a significant increase in basal, maximal, and spare respiratory capacity in the 5637GR cells when compared with vehicle, not observed in the 5637P or GR-OXCT1KO cells (Figure 3B). Calculated ATP-linked respiration was also greater in 5637GR than either parental or GR-OXCT1KO lines. This was further enhanced by gemcitabine treatment. The GR-OXCT1KO cells had a significantly lower spare respiratory capacity (the difference between basal ATP production and its maximal activity) compared with the parental line. Moreover, basal respiration and maximal respiration remained unchanged by the addition of gemcitabine, and spare respiratory capacity remained low. Thus, OXCT1 KO caused a metabolic shift toward anaerobic metabolism, yielding lower ATP production; this suggests that OXCT1 expression is necessary for an elevated respiration and survival under gemcitabine treatment.

Gemcitabine resistance is affected by cell respiration. (A) Oxygen consumption rate (OCR) under acute gemcitabine was measured using the Seahorse XFe24. (B) Isogenic 5637 lines, parental, 5637GR (GR), and 5637GR-KO (GR-KO) cells had differential metabolic activity. Acute addition of gemcitabine affected basal respiration, ATP production, maximal respiration, and spare respiratory capacity.

Next, the metabolic changes induced by 72 hours of gemcitabine treatment were evaluated using mass spectrometry (MS). Overall, 147 metabolites were identified by targeted analysis, of which 77 demonstrated significant differential enrichment by 1-way ANOVA post hoc analysis. Results highlight metabolites involved in pyrimidine synthesis (orotidine, dihydroorotate, and cytidine), TCA cycle (malate, fumarate, and succinate), and the pentose phosphate pathway (ribose-5-phosphate, gluconic acid, and sedoheptulose-7-phosphate; Supplemental Table 1). A heatmap of hierarchical clustering of the top 30 differentially enriched metabolites was generated relative to the untreated 5637P cells (Figure 4A). Notably, glucose was enriched in GR-OXCT1KO upon gemcitabine treatment but not in 5637GR cells. This result shows that gemcitabine treatment itself affected metabolism, and it also confirms that GR-OXCT1KO cells have a unique metabolism compared with resistant cells. On the other hand, cytosine, orotidine, NADPH, and phosphoribosyl diphosphate (PRPP), involved in nucleotide biosynthesis, were uniquely elevated in 5637GR cells regardless of gemcitabine treatment. The GR-OXCT1KO cells restored these same metabolites to parental levels, suggesting that OXCT1 expression has an effect on nucleotide biosynthesis pathways. Pathway enrichment analysis between untreated parental and 5637GR showed pyrimidine metabolism as among one of the significantly altered pathways in the development of gemcitabine resistance (Figure 4B and Supplemental Table 2). Amino acid metabolism (phenylalanine, cysteine, and alanine) as well as the ascorbate and aldarate metabolism were differentially regulated in 5637GR when compared with the 5637P cells (Figure 4B). The latter is known for being a source of oxidative stress protection, also suggesting that resistant cells have an increase in aerobic metabolism (29). The known functions of OXCT1 involving the TCA cycle as well as CoA synthesis were also identified (Supplemental Table 2). Upon gemcitabine treatment, these pathways were more significantly enriched in the resistant cells, including ketone body metabolism (Supplemental Figure 3A and Supplemental Table 3). To identify which metabolic signature precedes gemcitabine resistance, the gemcitabine-sensitive isogenic lines GR-OXCT1KO and 5637P cells were compared after 72 hours of gemcitabine treatment. Our results establish that knocking out OXCT1 caused an accumulation of palmitate and reduced carnitine, suggesting that fatty acid oxidation (FAO) is decreased in favor of anabolic processes (Supplemental Figure 3B). Interestingly, PRPP was significantly decreased in GR-OXCT1KO, highlighting the increase observed in 5637GR. Thus, the enhancement in nucleotide biosynthesis associated with gemcitabine resistance seems to be OXCT1 expression dependent. Furthermore, the metabolic pathway enrichment analysis between the sensitive lines showed that oxidative processes were inhibited in GR-OXCT1KO compared with the parental cell line. In contrast, fatty acid biosynthesis was enriched in GR-OXCT1KO cells (Supplemental Figure 3C).

Metabolomic analysis of isogenic bladder cancer lines revealed role of βHB in gemcitabine sensitivity. (A) Heatmap of the top 30 metabolites relative to the untreated 5637P cells. Gemcitabine-resistant cells have an increase in nucleotide synthesis regardless of treatment. OXCT1-KO cells show a decrease in TCA cycle intermediates while accumulating glucose, suggesting a metabolic shift. (B) Metabolomic pathway analysis between parental and gemcitabine-resistant cells. (C) Cell viability assay was performed on the indicated cell lines with the metabolically inactive βHB enantiomer (S-βHB). (D) Volcano Plot showing all the top upregulated and downregulated genes in acquired resistance. (E) Multi-omics joint-pathway analysis showing the crosstalk between metabolomics and gene expression; significantly altered pathways between parental and gemcitabine resistance are highlighted. (F) Graphic depiction of the utilization of βHB in the mitochondria. Metabolites from the TCA cycle can be used for nucleotide synthesis.

Elevated expression of OXCT1 in gemcitabine-resistant cells can deplete βHB, while targeting OXCT1 would cause its accumulation. Therefore, to simulate the metabolic effect of targeting OXCT1, we exploited the chiral property of βHB by treating the cells with the metabolically inactive enantiomer S-βHB (5 mM) at a dose range of natural ketosis without causing severe ketoacidosis (30, 31). Chiral molecules have nonsuperimposable mirror structures, where in this case, R-βHB is metabolically active and S-βHB is not. Treating the 5637P cells with S-βHB for 72 hours supported gemcitabine sensitivity with a significant reduction in IC50 (P < 0.001; Figure 4C). While a partial restoration of gemcitabine sensitivity was observed when the 5637GR cells were treated with S-βHB (P < 0.001), the extent of sensitization did not approach that of the parental line. This result reinforces that OXCT1 may have a role outside of metabolism in support gemcitabine sensitivity.

To further investigate the role of OXCT1 in gemcitabine resistance, we used an integrated multi-omics approach. Gene expression changes were evaluated in 3 independent clones of 5637P and 5637GR cells. These independently derived single-cell clones consistently demonstrated elevated OXCT1 expression in the 5637GR lines (Supplemental Figure 4A). Upregulated genes included T-box TF 2 (TBX2), cadherin like and PC-esterase domain containing 1 (CPED1), and peroxidasin (PXDN) associated with poor outcomes (32–34). The downregulated genes included annexin A6 (ANXA6, involved in autophagy), troponin T3 (TNNT3, involved in skeletal muscle development), and peptidyl arginine deiminase, type II (PADI2, involved in protein citrullination) (Figure 4D) (35–37). The top 10 differentially regulated genes were validated by quantitative PCR (qPCR) (Supplemental Table 4). Afterward, all the differentially expressed genes (Supplemental Table 5) and all the differentially enriched metabolites of 5637GR and 5637P (Supplemental Table 2) were integrated in a joint-pathway multi-omic analysis. Here, we equally considered the statistical contribution of metabolomics and transcriptomics to enrich metabolic and signaling pathways (38). Our results indicate gemcitabine resistance to be associated with pyrimidine metabolism and the pentose phosphate pathway, while also highlighting axon guidance and focal adhesion (Figure 4E). These findings suggest gemcitabine resistance involved OXCT1-mediated changes in metabolism and cell differentiation. Here, gemcitabine-induced OXCT1 expression mediated ATP generation through fatty acid oxidation (FAO) energy required for dihydroorotate and UMP synthesis via orotidine monophosphate (Figure 4F). One mechanism for gemcitabine resistance could involve the generation of cytosine to help overcome gemcitabine integration into the newly synthesized DNA in a competitive manner (39, 40).

OXCT1 serves as a master regulator for stem cell reprogramming. Master regulator analysis of the RNA-Seq results helped us identify TFs orchestrating differentially expressed target genes (41, 42). Master regulator analysis revealed 3 key TFs, TBX2, SOX15, and OVOL1, in support of gemcitabine resistance (Figure 5A). These transcription regulators and their corresponding downstream targets were evaluated in the Peter C. Black dataset that included a gemcitabine/cisplatin-treated cohort (24). There was a strong correlation with the expression levels of OVOL1 downstream targets (more than 800 targets) and poor prognosis, not observed for TBX2 or SOX15 (Figure 5B; Supplemental Figure 4, B and C; and Supplemental Table 6). Of note, OVOL2 downstream targets, some overlapping with OVOL1, were also identified. Interestingly, PPARG is among the unique OVOL1 targets. Within the differentially expressed genes in our dataset, 150 were OVOL1/OVOL2 pathway targets, with 73 being unique to OVOL1 and 36 unique to OVOL2 (P < 0.05; Figure 5C). These data suggest that OVOL1 and OVOL2 regulation of differentiation might be involved in gemcitabine resistance (43). Both regulators are known as transcriptional repressors of EMT in favor of epithelial differentiation (20, 44). Importantly, the indicated overexpression of their targets would suggest reduced repressor activity. To test this assumption, the top OVOL1 and OVOL2 downstream targets were measured in 5637P, 5637GR, and 5637GR-OXCT1KO lines by qPCR (Figure 5D). Notably, several targets upregulated in 5637GR were found to be restored to near parental levels when OXCT1 was knocked out (GR-OXCT1KO lines), supporting a regulatory relationship between OXCT1 expression and OVOL1 activity. These findings were further tested in UM-UC3 and T24 bladder lines. As with the isogenic 5637 lines, the KO of OXCT1 in the inherently resistant UM-UC3 cells (UM-UC3-OXCT1KO) demonstrated elevated OVOL1 target gene expression when compared with its parental line. In contrast to the UM-UC3 line, T24 cells were exposed to IC50 concentrations of gemcitabine for a minimum of 4 passages. This generated a cell line with resistance to gemcitabine but more sensitive than 5637GR (T24E; Supplemental Figure 5A). These cells demonstrated elevated OXCT1 upon gemcitabine treatment (Supplemental Figure 5B). Accordingly, OVOL1 and OVOL1/OVOL2 targets were upregulated in the exposed cell line; however, unique OVOL2 target genes seemed consistently downregulated even when exposed to gemcitabine (Figure 5D). Together, the multiple examples led us to reason that OXCT1 expression regulates OVOL1 to mediate gemcitabine resistance. To test this hypothesis, we evaluated OVOL1 subcellular localization in the parental and resistant cell lines. We found that OVOL1 cytoplasmic localization was greatest in the resistant isogenic lines including 5637GR, UM-UC3 parental, and T24 exposed (Figure 5E and Supplemental Figure 5, C and D). Of note, OVOL2 remained localized to the cytoplasm. These findings suggest that gemcitabine treatment reduced nuclear availability of OVOL1 through the upregulation of OXCT1.

Genomic analysis of isogenic bladder cancer lines revealed OVOL1 as a master regulator. (A) Master regulator analysis highlighting the significant transcription factors SOX15, OVOL1, and TBX2. (B) Survival curve associating OVOL1 targets in pretreatment and gemcitabine/cisplatin-treated samples from the Peter C. Black dataset (HR = 2.35, P = 0.0119) (24). (C) Venn diagram highlighting the proportion of OVOL1 and OVOL2 targets in the differentially expressed genes. More details in Supplemental Tables 5 and 6. (D) qPCR confirmation of the top OVOL1 and OVOL2 targets — 5637: P (parental), GR (gemcitabine resistant), KO (GR-OXCT1 KO); T24: PC (parental, untreated), PG (parental, gemcitabine-treated), EC (gemcitabine-exposed, untreated), EG (gemcitabine-exposed, gemcitabine treated); and UM-UC3: PC (parental, untreated) PG (parental, gemcitabine-treated) KC (OXCT1-KO, untreated), KG (OXCT1-KO, gemcitabine treated) — all normalized to their respective control. Resistant or exposed cells show a higher expression of OVOL1 targets suggesting that OVOL1 is not repressing its targets. Heatmap represents fold change to each respective control. (E) Representative Western blot images for OXCT1, OVOL1, and OVOL2 using β-actin, lamin B1, and Tom20 as cytoplasmic, nuclear, and mitochondrial markers, respectively. Representative densitometry values were normalized to their respective loading control and relative to the parental vehicle sample. Western blots were repeated 4 times. (F) Cell surface CD44 and CD36 were measured by flow cytometry of indicated lines following 72 hours gemcitabine treatment at IC50 doses. (G) Schematic representation of the roles of OXCT1 and OVOL1 in gemcitabine sensitive and gemcitabine resistant cells. Under lower OXCT1, OVOL1 is free to translocate to the nucleus and repress target genes including PPARG, promoting a more differentiated state and reduced FAO. Under gemcitabine-resistant conditions, there is higher OXCT1, and cytoplasmic OVOL1 promotes PPARG, FAO, and a more dedifferentiated state.

As OVOL1 plays a pivotal role in regulating epithelial differentiation (45), we reasoned that the loss of such signals would result in dedifferentiation and, potentially, the expression of stem features often associated with therapy resistance. Therefore, cell surface stem markers CD44 and CD36 were evaluated in cells exposed to 10 nM gemcitabine for 72 hours using FACS. The 5637GR cells had double the CD44 positivity (77%) compared with its isogenic parental line (38%; Figure 5F). The deletion of OXCT1 in the gemcitabine-resistant line reverted CD44 expression to near parental line levels (49%). Similarly, CD36 expression was significantly higher in 5637GR cells (70%) compared with the 5637P cells (30%), and the GR-OXCT1KO cells had reduced CD36 expression (53%). Subsequent hanging drop sphere-forming assays performed with the isogenic lines supported the 5637GR cells generating large spheres as having stem features compared with the parental and GR-OXCT1KO cells under control conditions (Supplemental Figure 6A). Only 5637GR cells were able to maintain sphere integrity in the presence of gemcitabine. The elevated expression of the additional stem marker, SOX9, and migration by 5637GR cells was found to be a OXCT1-dependent response, as both were limited in the GR-OXCT1KO cells, like the parental line (Supplemental Figure 6, B and C). These data support gemcitabine-induced OXCT1 as a mediator of dedifferentiation through negative regulation of OVOL1 activity (Figure 5G).

OXCT1 affects tumor growth and gemcitabine sensitivity in mouse models. To determine whether OXCT1 expression would affect tumor growth, the isogenic lines were orthotopically grafted in NOD SCID-γ mice. By 8 weeks, the 5637GR tumors were significantly larger than the parental counterparts (P < 0.0001; Figure 6, A and B). Knocking out OXCT1 in the 5637GR line produced tumors more variable in size than the parental line. Immunostaining was performed to localize OXCT1 expression. The tumor mitotic index and vascularity was determined by localizing phosphorylated histone H3 and CD31, respectively. Knocking out OXCT1 significantly diminished the mitotic index compared with 5637P (P < 0.001; Figure 6B). The greater tumor weight of the resistant line could be attributable to the significant increase in vascular infiltration, as confirmed by the elevation in CD31 expression (P < 0.05). The GR-OXCT1KO and parental tumors were similar in terms of associated vasculature. Finally, we tested if targeting OXCT1 would sensitize tumors to gemcitabine. Orthotopic grafts of 5637P, 5637GR, and GR-OXCT1KO lines were monitored for tumor growth by ultrasound and μCT imaging (Figure 7, A and B). Ultrasound measurements were found to be comparable with the final gross measurement of the extracted tumors (Supplemental Figure 7). Tumor volume measurements calculated from ultrasound imaging suggested that 5637GR tumors thrived under gemcitabine treatment while the parental and GR-OXCT1KO tumors had limited growth (Figure 7C). The 5637GR tumors were confirmed to have greater OXCT1 expression compared with the parental tumors, with negligible detection in the GR-OXCT1KO tumors by IHC (Figure 8, A and B). The parental and GR-OXCT1KO tumors had significantly fewer mitotic cells compared with the 5637GR tumors in response to gemcitabine (P < 0.0001). We observed no significant differences in CD31+ vasculature and TUNEL positivity in response to gemcitabine among the isogenic lines (Supplemental Figure 8). However, GR-OXCT1KO tumors had significantly lower stem-associated SOX2 and SOX9 expression compared with either 5637P or 5637GR tumors (P < 0.001). Although available antibodies did not allow for OVOL1 localization in the tissues, OVOL1, OVOL2, and their respective downstream targets, inclusive of those associated with stem features, were reliably upregulated in the 5637GR tumors compared with the parental and GR-OXCT1KO counterparts by qPCR (Figure 8C). Together, OXCT1 was identified to orchestrate metabolic and transcriptomic changes downstream of OVOL1 that contributed to the development of gemcitabine resistance.

OXCT1 expression in bladder orthotopic models. (A) Isogenic parental (5637P), gemcitabine resistant (5637GR), and resistant OXCT1-KO (GR-KO) were injected intramurally and allowed to grow. Inset scale bar: 5 mm. The tumor histology and immunostaining were performed to evaluate OXCT1 expression, mitotic capacity (phospho-Histone H3), and vascularity (CD31). Total original magnification, 100×. Scale bar: 100 μm. Zoom: 100 μm × 100 μm. (B) Statistical analysis of the staining quantification was done with 1-way ANOVA at 95 % CI. *P < 0.05, **P < 0.01, ***P < 0.001 (n = 8). Arrowheads point to positively stained nuclei.

Imaging of gemcitabine treatment of orthotopic bladder cancer models. (A) Representative ultrasound images of 5637P, gemcitabine-resistant, and GR-OXCT1KO orthotopic bladder tumors before and after gemcitabine treatment. An outline of the tumor area is depicted in each panel. Scale bar: 1.6 mm. (B) Representative μCT images with iopamidol contrast after gemcitabine treatment. Quantum GX2 microCT (Perkin Elmer) X-Ray kV: 90kV, X-Ray µA 88 µA, FOV: 72 mm, Pixel size: 144 µm. (C) Normalized quantification of tumor volume before and after treatment as measured with ultrasound, 2-way ANOVA at 95 % CI. ****P < 0.0001 (n = 8 per group).

OVOL1 and stem features downstream of OXCT1 determine bladder cancer gemcitabine sensitivity. (A) Representative staining images of OXCT1, phospho-Histone H3, CD31, SOX2, and SOX9 in tumor tissue sections. Total original magnification, 200×. Scale bar: 50 μm. (B) Positive staining quantification. Statistical comparison made with 1-way ANOVA at 95 % CI. **P < 0.01, ***P < 0.001, ****P < 0.0001 (n = 8). (C) qPCR confirmation of downstream OVOL1, overlapping and OVOL2 targets in bladder tumor samples are depicted in a heatmap. Fold change scale is relative to the mean of the parental samples per row.