RRM1 and RRM2 are essential genes in FA-HNSCC

As a starting point of this study, we analyzed the viability data of all cell lines from the previously executed 319 tumor-lethal siRNA re-screen [14], focusing specifically on the FA-HNSCC cell lines. A viability score of ≤−0.5 relative to the positive and negative controls was considered lethal [14] (Fig. 1a). We noted that the knockdown of both RRM1 and RRM2 resulted in a highly significant reduction of viability, which was observed in all FA-derived and the majority of non-FA-derived cell lines tested.

Fig. 1: RRM1 and RRM2 genes are essential in FA-HNSCC cell lines.

A Heatmap of Log2 normalized viability scores obtained by a siRNA library rescreen in a panel of Fanconi anemia-derived HNSCC, HPV-negative and HPV-positive HNSCC, premalignant oral cells, and primary oral fibroblasts (P.O.F.). Note that fibroblasts VU-SCC-1678 are primary fibroblasts that emerged when VU-SCC-1678 was cultured, and these should be considered as cancer-associated fibroblasts. All experiments were performed in triplicate. NT non-targeting siRNA control, UBB viability score obtained after knockdown of Ubiquitin B (UBB) as positive transfection control. B Viability after mRNA knockdown of RRM1 (turquoise) and RRM2 (purple) by the pooled siRNA and the subsequent individual siRNAs (#1–#4) in primary oral fibroblasts, premalignant oral cells (VU-preSCC-M3) and HPV-negative cell line VU-SCC-120. Viability was normalized to siNT and corrected for the transfection efficiency determined by siUBB. All experiments were performed in triplicate. C Assessment of protein expression of RRM1 and RRM2, 48 h post-transfection with pooled siRNAs, indicates (near) complete knockdown. Increased protein levels of RRM2 are observed when RRM1 is knocked-down and vice versa. D Summary of the viability after RRM1 and RRM2 knockdown for all cell lines tested. All tested FA-HNSCC cell lines showed vulnerability to RRM1 and RRM2 knockdown, whereas for the non-FA HNSCC cell lines 7/14 showed sensitivity and 1/3 of the premalignant cells. The tested primary oral fibroblast was resistant to RNR complex knockdown. ****p < 0.0001 with t-test statistic. E The effects on viability upon knockdown of either RNR complex gene, highly correlates for all cell lines tested (Pearson’s correlation R = 0.7211, p = 0.0002).

Ribonucleotide reductase catalytic subunits M1 and M2, RRM1 and RRM2, respectively, are key components of the ribonucleotide reductase (RNR) complex. RRM1 encodes the ribonucleoside-diphosphate reductase large subunit that forms the α-subunit of the holoenzyme ribonucleotide reductase (RNR) complex. Two proteins may function as β-subunit in the complex. Primarily RRM2 forms a functional heterodimeric tetramer with RRM1, but its paralogue ribonucleotide-diphosphate reductase subunit M2 B (RRM2B, formerly known as p53R2) can complex with RRM1 as well to form a functional RNR enzyme [15]. The RNR complex catalyzes de novo production of deoxyribonucleoside diphosphates and triphosphates (dN(D)TPs) from ribonucleoside di- and triphosphates (N(D)TPs) to provide proliferating cells with the required deoxynucleotides for DNA replication in S-phase. In quiescent cells, dNTPs are also generated, but particularly for DNA repair [15].

To investigate the susceptibility of nonmalignant cells, we analyzed the viability score after RRM1 and RRM2 knockdown in primary non-transformed oral (PO) fibroblasts from two healthy donors (donors #50 and #54) and from an FA patient (VU-1678). The normal fibroblasts did not reach the lethality threshold, indicating a possible therapeutic index to target HNSCC cells also of FA patients.

Next, to evaluate the effects on cell viability and to exclude off-target effects, the four individual siRNAs that are present in the SMARTpools, and used as such in the initial screen and re-screens, were individually transfected into PO fibroblasts and several cell lines. All four individual siRNAs targeting RRM1 or RRM2 caused a reduced viability in the tumor cell lines but not in the normal PO fibroblasts (Figs. 1b and S1a).

Protein expression of RRM1 and RRM2 was assessed 48 h post-transfection with the SMARTpools and indicated a large reduction of protein levels. In addition, the knockdown of one counterpart caused upregulation of the other one, consistent with previous notions [16] (Fig. 1c).

A larger panel of cell lines was analyzed for RRM1 and RRM2 knockdown, including FA-HNSCC cell lines CCH-FAHNSCC-2, VU-SCC-1604, VU-SCC-1131, and VU-SCC-1365. All four tested FA lines were below a relative viability of 0.3; an arbitrary cut-off that separates sensitive from resistant cell lines. (Fig. 1d and S1b–e, Table S1). The observed effect on viability to RRM1 knockdown correlated significantly to effects obtained with RRM2 knockdown (Pearson’s correlation test, R = 0.67, p = 0.0008) (Fig. 1e). Nonetheless, variations were observed; some non-FA HNSCC cell lines displayed almost identical sensitivity to RRM1 and RRM2 knockdown (e.g., UM-SCC-22A), while others (e.g., UM-SCC-6) displayed a considerable difference (Fig. 1d).

To investigate whether the cell lines could be further sensitized to RNR complex interference, we combined the knockdown of RRM1 and RRM2 in a dual transfection (Fig. S1b–e). The combination did not enhance the lethal effects, neither in the more sensitive nor the more resistant tumor cell lines.

Response of FA-HNSCC cell lines to RNR inhibitor gemcitabine

Gemcitabine or 2′, 2′-difluoro-2-deoxycytidine (dFdC) is a deoxycytidine analog and a known inhibitor of the RNR complex. After uptake, dFdC is phosphorylated to monophosphate dFdCMP [17]. Further phosphorylation steps lead to the formation of dFdCDP and dFdCTP, of which the first is a potent inhibitor of the RNR complex and the latter is known to be incorporated into the DNA [18, 19]. Gemcitabine is a clinically approved drug and established as therapy for e.g., pancreatic cancer and non-small-cell lung cancer, but not for HNSCC [18, 20, 21].

Therefore, we proceeded with testing gemcitabine sensitivity for FA-HNSCC (Fig. 2a–e and S2). Data were compared by AUC analysis on the dose-response curves as more resistant cell lines did not show a major shift in the curve, but just did not reach a complete viability reduction at higher concentrations of gemcitabine, hampering a reliable EC50 concentration calculation. A panel of FA- and non-FA-HNSCC cell lines and PO fibroblasts from two FA patients and one healthy donor were treated with a dose range of gemcitabine. Cell lines such as VU-SCC-OE that were more resistant to siRNA knockdown (Fig. 1), were also more resistant to gemcitabine treatment. The FA-HNSCC cell lines were all sensitive to gemcitabine treatment (Fig. 2c), in line with the RNA interference data of RRM1 and RRM2. Notably, both the FA- and non-FA fibroblasts were relatively resistant to gemcitabine (Fig. 2d), excluding a synthetic lethal interaction between RNR inhibition and a defective FA/BRCA pathway.

Fig. 2: Ribonuclease reductase (RNR) complex inhibition through gemcitabine shows tumor-specific vulnerability in Fanconi anemia-derived FA-HNSCC cell lines.

A Viability was obtained after 72 h of treatment with a dose titration of gemcitabine in susceptible sporadic non-FA HNSCC (spHNSCC) and B resistant spHNSCC cell lines. The dashed line indicates an arbitrary threshold of 0.01 µM gemcitabine as a visual reference. All experiments in A–D were performed three times in triplicate. C Viability of three independent Fanconi anemia HNSCC cell lines after gemcitabine treatment. All FA-HNSCC cell lines showed vulnerability to gemcitabine treatment. D Viability of primary fibroblasts from three independent donors, of which two from Fanconi anemia patients, were all resistant to gemcitabine treatment. E Summary of all viability data after gemcitabine exposure. The area under the curve (AUC) is shown. With an AUC threshold of 0.08, all cell lines are equally sensitive to gemcitabine as to RNR knockdown. F Gemcitabine susceptibility significantly correlates with susceptibility to RNR complex knockdown (Pearson’s correlation gemcitabine (AUC) to siRRM1 R = 0.76, p = 0.001; Pearson’s correlation gemcitabine (AUC) to siRRM2 R = 0.78, p = 0.0007). G Colony formation assay assessing the synergy between gemcitabine and radiation, indicated an dose–response significant response. Representative pictures of colonies are shown. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, **** ≤ 0.0001 with Student’s t-test.

Sensitivity to gemcitabine highly correlated with RRM1 and RRM2 knockdown, highlighting the specific targeting of gemcitabine as an RNR complex inhibitor (Pearson’s correlation test, RRM1-AUC, R = 0.75, p = 0.001; RRM2-AUC R = 0.89, p < 0.0001) (Fig. 2f).

As we expect that gemcitabine will be combined with radiotherapy in clinical setting, we tested the additive effect of both treatments in FA-HNSCC cell line VU-SCC-1131. A synergistic dose-responsive anti-tumor effect was observed in a clonogenic assay when cells were treated with 10 nM gemcitabine and exposed to 0, 1, or 2 Gray. (Fig. 2g).

To analyze gemcitabine and radiation sensitivity in 3D organoid cultures, both VU-SCC-1131 and VU-SCC-1604 were exposed to gemcitabine 24 h and 5 days after seeding both in 2D and 3D. Macroscopic and microscopic pictures of organoid cultures are indicated in Suppl Fig. S3a, b. Gemcitabine was added after either 24 h (2Ds, 3Ds) or after 5 days (2De, 3De). Whether cells were cultured in 2D or 3D did not differ with respect to gemcitabine response. However, longer preincubation culturing periods and formation of tumor-like structures caused an increase of the EC50 for both 2D and 3D with approximately 10× (VU-SCC-1604) or 20× (VU-SCC-1131). Also for irradiation, 2D or 3D culturing did not impact radiation sensitivity.

Hematopoietic analysis in Fancg−/− mice exposed to gemcitabine

The main concern when treating FA patients with chemotherapeutic agents is toxicity. Both toxicity in general and, most notably bone marrow failure due to the hypersensitivity of hematopoietic stem and progenitor cells (HSPCs) to genotoxic agents [22, 23]. To study the impact of gemcitabine in vivo, wild-type (WT) and Fancg null (Fancg−/−) mice were either treated with PBS, 120 mg/kg gemcitabine [24], or 0.8 mg/kg cisplatin (positive control). Bone marrow (BM) was harvested two days later for downstream analysis (Fig. 3a). Using an established panel of hematopoietic stem and progenitor cell (HSPC) markers [25], multiplex flow cytometry was performed to identify and analyze different HSPC populations (Fig. 3b). In mice, HSPCs reside within the LSK compartment of bone marrow cells, defined as Lineage (Lin)-negative cells expressing high levels of the Stem cell antigen-1 (Sca-1) and c-Kit receptor. For cisplatin, in comparison to WT mice, the LSK population of Fancg−/− mice was found hypersensitive resulting in a near ablation of HSPCs (Fig. 3c, Figure S4a). In contrast, gemcitabine-treated HSPCs displayed no reduction in both WT and Fancg−/− mice. Apart from stem and multipotent progenitors (MPPs), myeloid-committed cells such as common myeloid progenitors (CMP) and megakaryocyte–erythroid progenitors (MEP), were also strongly affected in cisplatin-treated Fancg−/− mice (Fig. 3c, Figure S4a). This illustrates that a relatively low dose of cisplatin leads to severe toxicity in the entire hematopoietic network of Fancg−/− mice, strongly affecting the survival and renewal of short-term and long-term cells that warrant blood homeostasis. In contrast to cisplatin, CMPs and MEPs in Fancg−/− mice were also hardly affected by gemcitabine. In summary, a clinically relevant dose of gemcitabine renders no toxicity to the HSPCs in FA mice, whereas a relatively low dose of cisplatin had a major impact as expected. The data in these mouse models suggest that gemcitabine may be utilized as a potential therapeutic agent in FA patients without causing additional toxicity by the FA defect.

Fig. 3: Cisplatin but not gemcitabine exposure affects hematopoietic subsets in Fancg−/− mice.

A Schematic outline of the toxicity study. Figure prepared using BioRender. B Flow cytometry-based gating strategy to identify different hematopoietic cell subsets. Bone marrows were flushed, erylyzed, and stained with a cocktail of surface antigen markers, and live single cells were used to distinguish the different subsets. C Relative numbers of different subsets as defined in (B), analyzed on day 2 post treatment. The graphs indicate the fold change of cells relative to PBS controls. In each graph, the bar represents mean ± sd. WT gemcitabine (n = 5), Fancg−/− gemcitabine (n = 6), WT cisplatin (n = 3) and Fancg−/− cisplatin (n = 3). p Values were calculated using two-way ANOVA. *p = 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Histopathological analysis of gemcitabine-treated Fancg−/− mice

To identify if gemcitabine exposure induces any other tissue toxicity in Fancg−/− mice, a whole-body pathology examination was performed on mice from all genetic backgrounds and treatment groups. Histopathological analysis was performed blinded on blood smears, spleen, thymus, heart, lung, reproductive organs, BM and small intestine (SI), to examen systemic toxicities. Giemsa-Wright staining of blood smears and H&E staining of the spleen (Fig. 4a), thymus (Figure S5a), heart, lung, and reproductive organs (not shown) did not reveal any gross pathological alterations upon gemcitabine- or cisplatin-exposure of Fancg−/− mice in comparison to PBS controls. In contrast to these tissues, the SI of Fancg−/− mice was not impacted by gemcitabine treatment but considerably impacted by cisplatin exposure, as indicated by the disrupted tissue and absence of defined crypts (Fig. 4b). The absence of a functional FA/BRCA pathway and the rapid cell turnover makes the SI also hypersensitive to cisplatin in the mice, but not to gemcitabine exposure in line with the bone marrow data (Fig. 3).

Fig. 4: Gemcitabine is well tolerated by Fancg−/− mice with a transient effect on bone marrow cellularity.

A–C Representative H&E sections of the spleen (A), small intestine (SI) (B), and bone marrow (BM) (sternum) (C) isolated from Fancg−/− mice treated with PBS, 120 mg/kg gemcitabine or low dose 0.8 mg/kg cisplatin and harvested on day 2 post-treatment. Red arrows indicate apoptotic cells. D Representative H&E sections of BM (sternum) isolated from Fancg−/− mice treated with PBS, 120 mg/kg gemcitabine, or 0.8 mg/kg cisplatin and harvested on day 7 post-treatment.

Although hematopoiesis was not affected in Fancg−/− mice 48 h post-gemcitabine treatment (Fig. 3c), mild toxicity in cellularity was observed in the H&E slides of the bone marrow (Fig. 4c). This was categorized as a mild effect as gemcitabine treatment decreased the absolute cell numbers, but the proportion of each hematopoietic subset was not affected. On the contrary, the cisplatin-treated Fancg−/− mice experienced severe toxicity, as both total cellularity (H&E slide) (Fig. 4c) as well as proportion of each subset was reduced substantially as determined by flow cytometry (Fig. 3c).

Gemcitabine is generally administered on a weekly basis in human clinical trials [26]. Therefore, we questioned whether the subtle reduction in BM cellularity on day two in Fancg−/− mice was restored by day seven. To test this hypothesis, mice experiments were repeated as described above, and the histopathology and flow cytometry analysis was performed on day seven post-treatment. In line with our hypothesis, H&E sections of the BM from gemcitabine-treated Fancg−/− mice showed a complete recovery (Fig. 4d). Interestingly, cisplatin-treated SI and BM analysis unexpectedly suggested a complete tissue recovery, based on the presence of nucleated cells (Figure S5b). However, flow cytometry data revealed that the hematopoietic subsets in cisplatin-treated Fancg−/− mice remained significantly reduced on day seven (Fig. S5c), as expected in FA mice and, by extension in FA patients. After treatment with gemcitabine, cell populations were restored at day 7, which again confirms that contrary to cisplatin, a deficient FA/BRCA pathway is not synthetically lethal with gemcitabine treatment.

In summary, our data strongly indicates that treatment with gemcitabine is well tolerated by FA/BRCA deficient mice and may provide an alternative treatment modality for chemoradiotherapy of HNSCC in FA patients.