Cytochrome P450 isoforms 1A1, 1B1 AND 2W1 as targets for therapeutic intervention in head and neck cancer

Epidemiological studies have shown that head and neck cancer (HNC) is a complex multistage process that in part involves exposure to a combination of carcinogens and the capacity of certain drug-metabolising enzymes including cytochrome P450 (CYP) to detoxify or activate such carcinogens. In this study, CYP1A1, CYP1B1 and CYP2W1 expression in HNC was correlated with potential as target for duocarmycin prodrug activation and selective therapy. In the HNC cell lines, elevated expression was shown at the gene level for CYP1A1 and CYP1B1 whereas CYP2W1 was hardly detected. However, CYP2W1 was expressed in FaDu and Detroit-562 xenografts and in a cohort of human HNC samples. Functional activity was measured in Fadu and Detroit-562 cells using P450-Glo™ assay. Antiproliferative results of duocarmycin prodrugs ICT2700 and ICT2706 revealed FaDu and Detroit-562 as the most sensitive HNC cell lines. Administration of ICT2700 in vivo using a single dose of ICT2700 (150 mg/kg) showed preferential inhibition of small tumour growth (mean size of 60 mm3) in mice bearing FaDu xenografts. Significantly, our findings suggest a potential targeted therapeutic approach to manage HNCs by exploiting intratumoural CYP expression for metabolic activation of duocarmycin-based prodrugs such as ICT2700.

Real time polymerase chain reaction (RT-PCR). The expression of CYP mRNA was quantified using RT-PCR assay in an Abi 7500 Real time PCR System (Applied Biosystems, Foster City, USA). The experiments were performed using with PrecisionPLUS MasterMix-Taqman-style, CYP1A1 (CAA GGT GTT AAG TGA GAA GGTG) and CYP1B1 (CAC TGG AAA CCG CAC CTC ) primers from Primerdesign® and CYP2W1 (GCA TCC AGC CAG AGA CAG G) primer from Applied Biosystems. Human (Primerdesign®) and hamster (TaqMan®) β-actin were used as endogenous controls. The gene expression was primarily determined relative to β-actin by applying the comparative ΔCt Method 32 . All samples were analysed in triplicate and by independent experiments. P values were determined using the two-tailed student t-test in Microsoft Excel 2013®; and the results were considered significant when P ≤ 0.05(*), P ≤ 0.01(**), P ≤ 0.001 (***).
Immunohistochemistry. Immunohistochemistry (IHC) was performed on tissue sections from xenografts and frozen human samples. The latter were obtained from Ethical Tissue (http:// www. bradf ord. ac. uk/ busin ess/ ethic al-tissue/), including participant consent for research purposes and approved by the University of Bradford's ethical committee (application 15-067) to detect CYP1A1, 1B1 and 2W1 proteins. Paraffin-embedded tissues from human tumour xenografts were initially deparaffinised and dehydrated followed by antigen retrieval for 20 min which was performed by boiling the tissue sections in 10 mM citrate buffer (pH 6.0). After blocking, the sections were incubated in a humidified chamber overnight at 4 °C with the CYP1A1 (1:10), CYP1B1 (1:5) and CYP2W1 (1:1000) primary antibodies (see Table S1 for detailed information) diluted in the respective blocking reagents followed by incubation with MOM biotinylated anti-mouse IgG reagent. The slides were washed and incubated for another 30 min at RT with Avidin/Biotin Complex (VECTASTAIN ABC HRP Kit) and developed using 3, 3-diaminobenzidine (DAB) detection kit. The counterstaining was carried out using Harris' haematoxylin. The tissue microarray (TMA) slides used to analyse the expression of CYP1A1, 1B1 and 2W1 were from , between the ages of 7 and 12 weeks were used. The mice were housed in cages in isolation cabinets with regular alternating cycles of light and darkness and received Teklad 2018 (Envigo) diet and water ad libitum. For the subcutaneous xenograft efficacy studies, mice were briefly given general anaesthesia and 2-3 mm 3 fragments of FaDu tumour from donor xenografts were transplanted subcutaneously in the abdominal flanks of the efficacy study mice. Once all tumours were palpable, mice were randomised into treatment groups. Treated animals received a single 150 mg/kg dose of ICT2700, administered intraperitoneally on day 0, with control animals receiving solvent (10% DMSO: 90% arachis oil). Tumour volume, using callipers, and animal body weight were recorded throughout the experiments and normalised to the respective volume on the initial day of treatment (day 0). A Mann-Whitney U-test was used for analysis of significance in the first study, whilst a t-test was used to compare small (mean size of 60 mm 3 ) and large (mean size of 240 mm 3 ) tumours in the second study.

Results
Immunohistochemical expression in HNC frozen tissues and TMA. The expression of CYP1A1, 1B1 and 2W1 was measured in primary tumours from snap-frozen human HNC tissues and tissue microarray (TMA). In TMA, the expression of CYP1A1 was weak in normal tissues and moderate in primary and metastatic tumours (Fig. 1A). With CYP1B1, moderate expression was identified in 21% of the normal tissues, 34% of the primary tumours and 30% of the metastatic samples, with just 3% of tumours showing high expression of CYP1B1 (Fig. 1B). It is important to note that most of the normal samples from the frozen tissues were negative while the primary tumours in TMA exhibited moderate expression. Expression of CYP2W1 in the normal tissues samples was found to be variable between TMA and frozen sections, barely detectable in the latter. However, the majority of tumour samples expressed CYP2W1 at a moderate (23%) to high (65%) level (Fig. 1C). In metastatic samples, the CYP2W1 expression was between 43 and 52%, revealing no significant difference from primary tumour tissue. Likewise, in the frozen tissue samples the majority of normal tissue (80%) displayed no CYP2W1 expression while the primary tumours from the same source consistently showed high expression (Fig. 1C). In general, the difference in CYP2W1 expression between normal tissue and primary tumours was highly significant (P ≤ 0.0002). Taken together the data indicated that CYP1A1, 1B1 and 2W1 isoforms were, in general, moderately or highly expressed in HNC compared with normal tissues. This supports the potential for CYP-activated prodrugs for therapeutic intervention in HNC following patient profiling for tumour CYP1A1, 1B1 or 2W1 expression.  CYPs in clinical samples we next went on to profile a panel of HNC cell lines and xenografts. CYP1A1, 1B1 and 2W1 at transcriptional and translational level was performed to explore their prevalence in HNCs. Accordingly, RT-PCR was carried out in HNC cell lines (n = 9) (A-253, Detroit-562, FaDu, OSC19, SCC4, SCC5, SCC10, SCC14 and SCC16A) using specific primers to determine the relative gene level expression of the CYPs. Isogenic cell line pairs CHO/CHO1A1 37 , SW480/SW480-2W1 26 and the glioma cell line U87 38 were used as controls for the expression of CYP1A1, 2W1 and 1B1, respectively. Evaluation of the HNC cell lines revealed varying levels of CYP1A1 expression in cells ( Fig. 2A). FaDu, SCC5 and SCC10 were shown to have higher CYP1A1 expression although significant statistical difference (P ≤ 0.01) with the CHO1A1 cell line was observed only with SCC10. Contrarily, cell lines OSC19, SCC4, SCC14 and SCC16A had significantly lower expression (P ≤ 0.05) compared to CHO1A1. As opposed to CYP1A1, expression of CYP1B1 and CYP2W1 in HNC cell lines was found to be significantly low compared to the control cells. Reduced expression of CYP1B1 was observed in all cell lines except for the moderate expression in FaDu and SCC10 cell lines (Fig. 2B) while the expression of CYP2W1 in the majority of the cell lines examined was either low or absent (Fig. 2C). The results indicate that CYP1A1 and CYP1B1 have elevated mRNA level expression in HNC cell lines when compared to the CYP2W1 isoform.
To corroborate these findings, we performed an in silico analysis of the RNA data from Cancer Cell Line Encyclopaedia (CCLE) dataset for HNC cell lines using the public cBioPortal website 39 and found that the isoforms CYP1B1 and CYP2W1 were amplified in Detroit-562 and SCC4 respectively (Fig. S1).

Immunodetection of CYP1A1, 1B1 and 2W1 in HNC.
To understand if the mRNA expression levels correlated with protein, immunodetection assays were performed on cell lines, cell lined-derived xenografts (CDXs) and frozen tumour tissues. A summary of CYP1A1, 1B1 and 2W1 gene and protein expression can be found in Table 1.
Immunofluorescence of CYP1A1, CYP1B1 and CYP2W1 in HNC cell lines. CYP1A1 was detected in A-253, Detroit-562, FaDu and CHO1A1 (positive control) by immunofluorescence but not CHO and OSC19 cell lines (Fig. S2A). CYP1B1 expression was detected in Detroit-562 and FaDu cells along with the U87 glioma cells (positive control 38 ), but absent in A-253 and OSC19 (Fig. S2B). It is notable that the expression of CYP1B1 in the cytoplasmic regions of Detroit-562 and FaDu were at a much lower level in comparison to U87. On the other hand, CYP2W1 displayed variable levels of expression in A253, Detroit-562, FaDu and OSC19 cells and CYP2W1-transfected SW480 colon cancer cells used as a positive control (Fig. S2C).

Immunohistochemical expression in HNC cell line derived xenografts (CDXs).
Expression of CYP1A1, 1B1 and 2W1 were further evaluated in a panel of selected xenograft tissues consisting of A-253, Detroit-562, FaDu and OSC19 cells as well as in primary human HNC sections by immunohistochemistry. Highest expression of CYP1A1 was observed in Detroit-562 and FaDu with expression being extra-nuclear (Fig. 3A). In comparison to CHO, FaDu (P ≤ 0.001), CHO1A1 (P ≤ 0.001) and Detroit-562 (P ≤ 0.01) showed significantly higher expression of CYP1A1 (Fig. S3A). CYP1B1 expression was observed to be weak in A-253, Detroit-562, FaDu and OSC19 (Fig. 3B). The expression of CYP1B1 in all examined cell lines were mainly identified in the cytoplasm. In general, the expression was significant between cell lines (P ≤ 0.001) and also in comparison with CHO, U87 and OSC19 was significant (P ≤ 0.001, Fig. S3B). The expression of CYP2W1 was shown to be high in Detroit-562 and FaDu and SW480-2W1, and moderate in A-253, SW480 and OSC19 cells. The expression of CYP2W1 was predominantly cytoplasmic, with the intensity being proportional to the presence of CYP2W1 protein (Fig. 3C). Compared to SW480 (Fig. S3C), CYP2W1 expression was significantly higher in SW480-2W1 (P ≤ 0.001), Detroit-562 (P ≤ 0.001), FaDu (P ≤ 0.001) and A253 (P ≤ 0.01).

Growth inhibition of duocarmycin compounds in HNC cell lines. Next, we investigated three
duocarmycin-based compounds ICT2700, ICT2706 and ICT2726 previously explored for CYP-activation; the former two duocarmycins are bioactivated by CYP1A1 and 2W1 to potent cytotoxins [25][26][27] while the latter analogue is inactivate 37 and hence here employed as a negative control when evaluated against a panel of HNC cell lines. Detroit-562 cells were the most sensitive to ICT2700 (IC 50 = 280 nM) followed by Fadu, SCC5, SCC10 and SCC16A cell lines (IC 50 circa 1 µM). ICT2706 was not as potent as ICT2700, which is consistent with previous data in CYP1A1-expressing cells 25,26 . Antiproliferative activity was not observed for ICT2726 under the conditions investigation (IC 50 ≥ 10 µM, Table 2) consistent with previous observations, hence the inclusion here as a negative control compound 26 . Antiproliferative activity of ICT2700 and ICT2706 was confirmed using isogenic cell lines expressing target CYPs (CHO/CHO1A1 and SW480/SW480-2W1) as previously reported 26,27 .
The anti-proliferative activity of the compounds observed in Fadu and Detroit-562 cells linked to functional CYP1A1 activity using a P450-Glo™ assay, which measures the luminescence produced via CYP1A1 mediated conversion of the substrate proluciferin to D-luciferin. An increase in CYP1A1 functional activity was observed in both FaDu and Detroit-562 cells although this was not as high as in CHO1A1-transfected cells (Fig. 4A); activity was reduced in the latter by treating cells with α-naphthoflavone, further indicating CYP1A1 activity (Fig. 4B). ICT2700 was observed to be most potent compound in HNC cell lines and hence was investigated in vivo.    3 . Tumours were measured after 48 h and the relative tumour volume at day 2 were compared to the initial tumour volume. A significant (P = 0.0004) difference was seen in response to ICT2700 in tumours of a relatively small (mean size of 60 mm 3 ) size at the commencement of treatment compared with the larger (mean size of 240 mm 3 ) tumour cohort (Fig. 5A). Increased DNA damage was observed after 6 h compared with both 1 and 24 h time points (Fig. 5B-C) indicating generation of duocarmycin chemotoxin metabolite previously identified 27 and responsible for cell death and reduction in tumour size. Next, ICT2700 was administered to mice bearing palpable tumours as a single dose on day 0 (150 mg/kg) and the impact on tumour growth delay and systemic toxicity (weight loss) was measured. Although an initial treatment effect was observed with negligible weight loss (Fig. S5), no statistically significant growth delay was observed after eight days (Fig. 5D).

Effect of ICT2700 on tumour growth in
Effect of CAF derived conditioned media on the activity of ICT2700. The tumour microenvironment (TME) is a complex multicellular system hosting a variety of immune and non-immune cell types and include cancer-associated fibroblasts (CAFs) as one of the most dominant components in the TME 40 . The TME contributes to tumourigenesis through a multifactorial process and some emerging evidence suggests differential expression of CYP1 enzymes in fibroblasts which might alter treatment sensitivity 41 . In accordance with this and to inform the in vivo efficacy of ICT2700, the effect of this prodrug in a co-culture of FaDu cells and HNC CAFs was investigated. Cells grown in the presence of conditioned media from cancer-associated fibroblast (CM-CAF) obtained from two HNC patients (CAF002 and CAF003) showed no change in the potency of ICT2700 (IC 50 : 1.03 µM) when compared with control Fadu cells grown in RPMI media alone (IC 50 : 1.03 µM). Moreover, FaDu cells co-cultured with CAF003 showed a time-dependent decrease in sensitivity to ICT2700. Specifically, the inhibition of cell growth by ICT2700 for FaDu/CAF003 co-culture (at 24 h) was IC 50 of 1.59 µM in contrast to CAF002 (IC 50 : 10.07 µM) (Fig. 6).

Discussion
Despite the multi-modality therapies used for the clinical management of HNCs involving surgery, radiotherapy, chemotherapy and targeted therapy the survival rates have not been improved significantly over the past few decades. Although progress has been made in biologic targeted therapies, notably cetuximab, no new targeted small molecules have been approved for clinical use. CYP superfamily isoforms responsible for drug oxidations are attractive as targets to selectively activate prodrugs in tumours 16,[42][43][44][45][46][47][48] . The differential expression of selected Table 1. Summary of CYP1A1, 1B1 and 2W1 gene and protein expression levels. a RT-PCR expression in Fig. 2A-C, b expression by immunofluorescence in Fig. S2A   www.nature.com/scientificreports/ CYPs in tumour versus surrounding normal tissue offers an excellent opportunity to develop targeted therapies with significantly reduced side effects. In the present study we have profiled CYP1A1, 1B1 and 2W1 expression in HNC and explored if these can be exploited for therapeutic gain. The mRNA and protein expression profiles of the three CYP isoforms were investigated in a panel of HNC cell lines, xenografts and primary tissues. Our results showed that the expression of CYP1A1/1B1 and 2W1 were poorly expressed in cell lines compared to the clinical and xenografts tissues, indicating in vivo microenvironment is important for CYP expression. This is further supported by an earlier study that was focused on the NCI 60 cancer cell line screen, and which revealed low CYP expression and functional activity in these immortalised cell lines 49 . The FaDu cell line which expressed CYP1A1 and low levels of 2W1 as well as exhibiting functional CYP1A1 activity was used to investigate ICT2700 in vivo. ICT2700 delayed tumour growth but only in small sized tumours. γH2AX phosphorylation indicative of DNA damage was increased 6 h after treatment in both small and large tumours and correlates with bioprecursor bioactivation as previously shown 26,27 . All animals displayed normal appearance and behaviour with no significant loss in body weight throughout the course of these studies. The data supports that duocarmycin prodrugs typified by ICT2700 can be administered safely with good tolerability.
Emerging preclinical studies reveal the intriguing role played by several physical and biological factors of the TME in conferring intrinsic drug resistance to tumour cells 50 . Among the biological components, CAFs remain the most studied and predominant cell type of the TME which have been shown to enhance tumour growth and confer chemotherapy (e.g. cisplatin) resistance through an irreversible CAF-activation process and subsequent secretion of several growth factors and chemokines 51 . However, our in vitro analyses suggested that the co-culture of HNC tumour cells with conditioned medium obtained from culturing human CAFs from HNC patients did not impair the efficacy of ICT2700. This finding, while preliminary, suggests minimal influence of the TME on the potency of the CYP-targeted agents. In contrast, cisplatin, a drug used clinically to treat SSC, exhibited impaired cytotoxic activity in cells cultured with CM-CAFs (Fig. 6), which is in accordance with other studies that have shown CAFs to enhance drug resistance 52 .
CYP1A1 is a controversial target given that smokers induce this enzyme in the lung, but careful patient recruitment of only non-smokers could mean cohorts of HNC patients could benefit from CYP1A1-targeted www.nature.com/scientificreports/ therapeutics. CYP1B1 is highly expressed in many cancer types and has been known as a potential target for a long time 53,54 . Recent evidence suggests that never smokers make up an increasing proportion of the head and neck cancer population and now represent 24% of new cancer diagnoses (CRUK LIHNCS trial, McCaul et al. unpublished data). However very little progress has been made in exploiting this isoform for prodrug activation although there is much support from the development of CYP1B1 inhibitors 36 . In conclusion, this study indicates the potential of CYP expression as targets for therapeutic intervention in HNC and warrants further investigation of CYP function in HNC tumours and the TME. In addition, the dosing schedule of ICT2700 should be optimised to confirm the potential of this duocarmycin prodrug approach for HNC treatment.