Targeted Therapy of HPV Positive and Negative Tonsillar Squamous Cell Carcinoma Cell Lines Reveals Synergy between CDK4/6, PI3K and Sometimes FGFR Inhibitors, but Rarely between PARP and WEE1 Inhibitors

Human papillomavirus positive (HPV+) tonsillar and base of tongue squamous cell carcinoma (TSCC/BOTSCC) have a favorable outcome, but upon relapse, survival is poor and new therapeutical options are needed. Recently, we found synergistic effects by combining the food and drug administration approved (FDA) phosphoinositide 3-kinase (PI3K) and fibroblast-growth-factor-receptor (FGFR) inhibitors BYL719 and JNJ-42756493 on TSCC cell lines. Here this approach was extended and Cyclin-Dependent-Kinase-4/6 (CDK4/6) and Poly-ADP-ribose-polymerase (PARP) and WEE1 inhibitors PD-0332991, and MK-1775 respectively were also examined. HPV+ CU-OP-2, -3, -20, and HPV− CU-OP-17 TSCC cell lines were treated with either BYL719 and JNJ-42756493, PD-0332991 BMN-673 and MK-1775 alone or in different combinations. Viability, proliferation, and cytotoxicity were followed by WST-1 assays and the IncuCyte S3 Live® Cell Analysis System. All inhibitors presented dose-dependent inhibitory effects on tested TSCC lines. Synergy was frequently obtained when combining CDK4/6 with PI3K inhibitors, but only sometimes or rarely when combining CDK4/6 with FGFR inhibitors or PARP with WEE1 inhibitors. To conclude, using CDK4/6 with PI3K or FGFR inhibitors, especially PD-0332991 with BYL719 presented synergy and enhanced the decrease of viability considerably, while although dose dependent responses were obtained with PARP and WEE1 inhibitors (BMN-673 and MK-1775 resp.), synergy was rarely disclosed.

In breast cancer with PIK3CA mutations and urothelial cancer with FGFR3 translocations or mutations, Food and Drug Administration (FDA) approved phosphoinositide 3-kinases (PI3K) and fibroblast growth factor receptor (FGFR) inhibitors resp. are used clinically [32,33]. For this reason, we recently investigated the effects of FDA-approved PI3K and FGFR inhibitors (alpelisib (BYL719) and erdafitinib (JNJ-42756493) resp.) in TSCC/BOTSCC cell lines with/without PIK3CA and FGFR3 mutations and found both dose-dependent and synergistic effects [34]. In addition, by administering these inhibitors together with cisplatin and docetaxel, used clinically for TSCC/BOTSCC therapy, various combinatory effects were disclosed [34]. Due to the synergistic effects of PI3K and FGFR inhibitors on the viability and proliferation of HPV + TSCC/BOTSCC lines, we were eager to explore whether these inhibitors could also be joined with others and reveal additional synergistic effects. PI3K and Cyclin-Dependent-Kinase-4/6 (CDK4/6) inhibitor combinations have namely also been shown to be of benefit in some types of breast cancer [35].
Additionally, other reports have shown that Poly-ADP-ribose-polymerase (PARP) and WEE1 inhibitors can have synergistic effects in other tumor types such as, e.g., in triple negative breast cancer with Cyclin E or BRCA1 alterations or in biliary tract cancer [36,37]. Moreover, we too have previously shown favorable effects of PARP inhibitors on some, but not all HPV + TSCC/BOTSCC cell lines [38].
Here, therefore in a number of TSCC cell lines, the effects of the CDK4/6 inhibitor (PD-0332991) combined either with the PI3K or the FGFR inhibitors (BYL719 orJNJ-42 756493 resp) as well as the effects of PARP and WEE1 inhibitors, BMN-673 and MK-1775 resp. alone or combined were explored.

WST-1 Viability Assay
Viability was analyzed by a WST-1 assay (Roche Diagnostics, Mannheim, Germany) and followed for 72 h (h) after the treatment according to the instructions of the manufacturer and repeated three times and presented as described in more detail before [34].

Cell Proliferation and Cytotoxicity Assays
Cells in 96-well plates were placed into the IncuCyte S3 Live ® Cell Analysis System using the Incucyte™ Cytotox Red Reagent (Essen Bioscience, Welwyn Garden City, UK), and followed for proliferation cytotoxicity, taking images every 2 h [34]. Of three repeated experiments a representative experiment was presented.

Statistical Analysis
To verify efficacy of single inhibitors or their combinations compared to the negative control, a multiple t-test accompanied by correction for multiple comparison of the means conferring to the Holm Sidak method was done [34,39,40]. The combined effects were evaluated applying the effect-based approach 'Highest Single Agent' [39,40], where details have been presented previously [34,39,40]. To test the effects of CDK4/6, FGFR, PI3K, PARP and WEE1 inhibitors, the effects of PD0332991, BYL719, JNJ-42756493, BMN-673 and MK-1775 resp. on the HPV + CU-OP-2, CU-OP-3 and CU-OP-20 and HPV − CUOP-17 cell lines, were examined by WST-1 viability assays. Cell viability was measured after 24, 48 and 72 h with single treatments with the indicated doses below. For BYL719 and JNJ-42756493 only two low doses were included, since higher doses have shown strong inhibition and have been published previously together with the IC50 values of both of the inhibitors [34].

PD-0332991
All CU-OP lines showed dose-dependent responses to 5-40 µM of PD-0332991, with CU-OP-20 being most sensitive and CU-OP-2 most resistant ( Figure 1A-D). The highest dose i.e., 40 µM decreased viability completely 24-72 h after treatment in all cell lines (at least p < 0.001) and during the same period a >50% decrease in viability was also found in all cell lines except CU-OP-2 with 20 µM PD-0332991 (in all relevant cell lines at least p < 0.01). With 10 µM PD-0332991 only CU-OP-20 presented a significant 50% decrease of viability at all time points (at least p < 0.05), while CU-OP-3 and CU-OP-17 only had minor decreases in viability after 72 h (at least p < 0.05). The 5µM PD-0332991 dose had no effect on any of the four cell lines.

BYL719
HPV + CU-OP-2, CU-OP-3, and CU-OP-20, and HPV − CU-OP-17 have previously been shown to present dose-dependent responses to 0.5-10 µM BYL719 [34]. Here only two very low BYL719 doses were utilized (0.5 µM and 1 µM) showing only marginal decreases in viability ( Figure 1E-H). All HPV + cell lines were sensitive to 1 µM BYL719 and had a decrease in viability, except CU-OP-17 (for all indicated, at least p < 0.05 after 48 h), while with 0.5 µM BYL719, CU-OP-3 was the only cell line that showed a statistically significant decrease in viability (with at least p < 0.001 at 48 h and p < 0.05 at 72 h). In contrast, HPV − CU-OP-17 tended here to be resistant to 0.5 µM and 1 µM BYL719.

BYL719
HPV + CU-OP-2, CU-OP-3, and CU-OP-20, and HPV − CU-OP-17 have previously been shown to present dose-dependent responses to 0.5-10 μM BYL719 [34]. Here only two very low BYL719 doses were utilized (0.5 μM and 1 μM) showing only marginal decreases in viability ( Figure 1E-H). All HPV + cell lines were sensitive to 1 μM BYL719 and had a decrease in viability, except CU-OP-17 (for all indicated, at least p < 0.05 after 48 h), while with 0.5 μM BYL719, CU-OP-3 was the only cell line that showed a statistically significant decrease in viability (with at least p < 0.001 at 48 h and p < 0.05 at 72 h). In contrast, HPV − CU-OP-17 tended here to be resistant to 0.5 μM and 1 μM BYL719.

IC50 (µM) b Drugs
Cell lines    More specifically, for PD-0332991 the IC50 values ranged between 9.8 µM and 35.9 µM for the CU-OP cell lines (Table 1). CU-OP-2 consistently had the highest IC50s, while for CU-OP-20 lower concentrations were consistently much more effective, in line with the data shown in Figure 2.
For BMN-673, the IC50 values were between 4.8 µM and extrapolated to 3219 with HPV − CU-OP-17 being the most sensitive and CU-OP 2 and 3 being the most resistant cell lines (Table 1). For MK-1775 the corresponding ranges were 0.2 µM to 9.6 µM, with CU-OP-20 being initially most sensitive, while the others showed relatively similar values.
To summarize these three new inhibitors, CU-OP-20 was generally the most sensitive cell line, with the only exception of BMN-673, where the HPV − CU-OP-17 cell line was more   The highest dose of PD-0332991 (40 µM) combined with 0.5 or 1 µM of BYL719 resulted in a retained decreased cell viability at all time points for all cell lines analogous to using the highest dose (40 µM) of PD-0332991 alone (for all at least p < 0.001) ( Figure 1A-H).
Notably, CU-OP-3 showed a statistically significant decrease for all remaining combinations of PD-0332991 with BYL719 0.5 µM, as was also the case for CU-OP-20, except with   PD-0332991 and BYL719. Combinations of all doses of the CDK4/6 inhibitor PD-0332991 with the lowest concentration of BYL719 0.5 µM resulted in synergistic, or mainly positive or neutral combinational effects (CI < 1) after 48 h, for CU-OP-3, CU-OP-20, and CU-OP-17, while this was not as much the case for CU-OP-2, that also showed a marginal negative effect ( Figure 4A). With the BYL719 1µM dose, all combinations with PD-0332991 mainly resulted in positive or neutral effects (CI < 1 or CI = 1) for all cell lines after 48 h. (Figure 4A).
To summarize, combining the CDK4/6 inhibitor with the PI3K inhibitor, generally resulted in a positive or neutral combinational effect in most cell lines except occasionally for CU-OP-2.
PD To summarize, combining the CDK4/6 inhibitor with the FGFR inhibitor, resulted in mainly neutral, but also some synergistic and negative combinational effects that also tended to be cell line specific. At 24 h after treatment, most combinations except with the lowest dose of MK-1775 (0.1 µM) also showed a tendency for a decrease in viability, but these did not reach statistical significance (p = ns).
To summarize, combining the PARP and WEE1 inhibitors did generally not show superior effects to using one of the drugs alone, with the exception of CU-OP-2, where an increased sensitivity was obtained.

Combinational Indexes with the "Highest Single Agent" Approach
The combinational indexes (CIs) of BMN-673 with MK-1775 were calculated according to the "highest single agent "approach for all cell lines 48 and 72 h after treatment. The CIs after 48 h are shown for specific doses as indicated below for the CU-OP cell lines ( Figure 6). The CIs after 72 h generally showed similar trends (data not shown).
For CU-OP-2 all the CIs < 1, indicated positive effects, and this tended to be also the case for the combinations including the 0.1 µM MK-1775 dose for CU-OP-20 and CU-OP-17, but not for CU-OP-3, where the tendency was mainly towards the negative side.
To summarize, for CU-OP-2 positive and synergistic effects were disclosed when combining BMN-673 with MK-1775, while for CU-OP-20 and CU-OP-17 the effects were relatively neutral, while for CU-OP-3 the effects were in the near neutral-negative side.  At 24 h after treatment, most combinations except with the lowest dose of MK-1775 (0.1 μM) also showed a tendency for a decrease in viability, but these did not reach statistical significance (p = ns).
To summarize, combining the PARP and WEE1 inhibitors did generally not show superior effects to using one of the drugs alone, with the exception of CU-OP-2, where an increased sensitivity was obtained.

Combinational Indexes with the "Highest Single Agent" Approach
The combinational indexes (CIs) of BMN-673 with MK-1775 were calculated according to the "highest single agent "approach for all cell lines 48 and 72 h after treatment. The CIs after 48 h are shown for specific doses as indicated below for the CU-OP cell lines ( Figure 6). The CIs after 72 h generally showed similar trends (data not shown). For CU-OP-2 all the CIs < 1, indicated positive effects, and this tended to be also the case for the combinations including the 0.1 μM MK-1775 dose for CU-OP-20 and CU-OP-17, but not for CU-OP-3, where the tendency was mainly towards the negative side.

Proliferation
PD-0332991. The two lowest doses 5 µM and 10 µM of PD-0332991 used above in the viability tests were shown to inhibit proliferation in all CU-OP cell lines to some extent and in some cases already 24 h after the treatment depending on the cell line ( Figure 7A-D). These doses were later used to be tested in combination with the two lowest doses of BYL719 and JNJ-42756493.
BYL719. Single treatment with BYL719 at the low dose concentrations (0.5 and 1.0 µM) induced a slight or no decrease in cell proliferation depending on the specific cell line in analogy to the viability data shown earlier ( Figure 7E-H).
JNJ-42756493. Single treatments with JNJ-42756493 at the low dose concentrations (0.1 and 1 µM) used here did not induce a decrease in cell proliferation in analogy to the viability data ( Figure 7I-L).

Cytotoxicity
In addition to the proliferation data above, the cytotoxic effects of the inhibitors with the same doses as used for proliferation were analysed using the Cytotox Red Reagent and the data for all four cell lines are shown in Supplementary Figure S1.  Figure 8A-C). For CU-OP-17 inhibition of proliferation was not quite as efficient analogous to the viability data, however, the combination of 10 µM PD-0332991 with 1 µM BYL719 gave the best effect ( Figure 8D).
PD-0332991 and JNJ-42756493. Combinations of PD-0332991 (5 and 10 µM) with JNJ (0.1 and 1 µM) inhibited proliferation to a higher or lower extent depending on the doses used and the cell line ( Figure 8E-H). The most pronounced inhibition of proliferation was observed for the CU-OP-3 and 20 cell lines ( Figure 8F,G) in analogy to that observed with the CI data ( Figure 4B).  To summarize, upon combining CDK 4/6 inhibitor PD-0332991 with PI3K inhibitor BYL719, an inhibition of proliferation was observed especially for all HPV + CU-OP lines, while the PD-0332991 and FGFR inhibitor JNJ-42756493 combination resulted in inhibition of mainly CU-OP-3 and 20 analogous to the CI data. For the PARP inhibitor, BMN-673 and the WEE-1 inhibitor MK-1775 combination inhibition of proliferation was observed mainly for CU-OP-2 also analogous to the CI data. (PD-0332991, BYL719, JNJ-42756493) and of PARP and WEE1 Inhibitors (BMN-673 and MK-1775)

of HPV + and HPV − TSCC Cell Lines
The effects of combined treatments with PD-0332991, and BYL719 or JNJ-42756493, or BMN-673 and MK-1775 on the CU-OP cell lines were examined further with regard to proliferation and cytotoxicity, all with PBS as positive control, and the data are depicted below ( Figure 8) and data not shown.
All TSCC cell lines showed dose-dependent decreases in viability and proliferation upon single treatments with PD-332991, BYL719, JNJ-42756493, BMN-673, and MK-1775 as shown here or previously [34], but the sensitivity of the cell lines varied depending on the inhibitor and the doses used. None of the above inhibitors showed any marked cytotoxic effects upon single or combined use with the lower doses included here against any of the Upon combining the CDK4/6 inhibitor PD-332991 with the PI3K inhibitor BYL719, mainly positive or synergistic effects with enhanced inhibition of viability and proliferation were observed in all cell lines, especially for HPV + CU-OP-3, CU-OP-20, and HPV − CU-OP-17. The CDK4/6 inhibitor PD-332991 and JNJ-42756493 combination also gave decreased viability and proliferation for all cell lines, but the effects were more variable. With the latter combination, synergistic effects were less common, with the exception of in CU-OP-3, while for CU-OP-2 and CU-OP-17 mainly antagonistic effects were observed. Finally, when the PARP inhibitor BMN-673 was combined with the WEE1 inhibitor MK-1775, although decreases in viability and proliferation were obtained, the only synergistic effect observed was in the CU-OP-2 cell line.
The fact that dose-dependent responses to the inhibitors were noted when using high doses of the drugs was not unexpected. Commenting first with regard to PD-332991, the effect of the lower dose (5 µM) of the inhibitor on the CU-OP cell lines would characterize these cells as rather resistant to this drug if one compares it with another report [41]. In that study, the HPV − Fadu head neck cancer cell line was shown as relatively sensitive, while two other HPV + head neck cancer cell lines UM-SCC-47 and UP-SCC-154 were similar to the CU-OP cell lines here (and also in an earlier study by us) less sensitive and defined as more resistant [38,41].
Notably, however, by combining 5 µM of PD-332991 with low doses of 0.5 and 1 µM of BYL719, gave for CU-OP-3 a remarkable decrease in the inhibition of viability and proliferation and could in fact be clinically useful, as for example as already initiated and shown for breast cancer [35]. However, despite the promises of using these drugs in combination, the safety of using these drugs alone or combined or in combination with endocrine therapies must also be accounted for and scrutinized on a molecular basis [42][43][44].
Some positive effects were also disclosed when combining PD-332991 with JNJ-42756493. Moreover, for both these latter two combinations, the best effects were observed for CU-OP-3, and although synergistic effects were also observed for all cell lines with the PD-332991 and BYL719 combinations, this was less so with the PD-332991 and JNJ-4256493 mix. For the latter on the other hand, there is to our knowledge not much information in the scientific literature.
With regard to the cytotoxic effects of the above drugs (PD-332991, BYL719, JNJ-4256493, BMN-673, or MK-1775), it was obvious that low doses of all the drugs used here had no major effects on any of the HPV + CU-OP-2, CU-OP-3 or CU-OP-20 cell lines. However, low doses of PD-332991 had a cytotoxic effect on the HPV − CU-OP-17 cell line. Nevertheless, when combining the PD-332991 with BYL719 or JNJ-42756493, or BMN-673 and MK-1775 with the doses used here, a major increase in cytotoxicity was not observed for this cell line or the other ones.
Of note, in this study only, very low doses of BYL719 and JNJ-42756493 were included, since our previous data showed that most of our cells were relatively sensitive to the effects of BYL719 and almost always so also for JNJ-42756493 [34]. Furthermore, when using high doses of these drugs, or corresponding other PI3K or FGFR inhibitors, their efficacy was so good when used alone, that the synergistic effects of combining them could be masked [34,45]. In this study, however, the BYL719 batch that was used was marginally less effective compared to that in one of our previous report [34].
Slightly different from the above inhibitors the sensitivity to BMN-673 showed another pattern, with CU-OP-2 initially being more sensitive than CU-OP-3 in line with a previous study by us where the PARP1 inhibitor AZD2281 reduced colony formation more readily in CU-OP-2 than in CU-OP-3 [38]. Furthermore, in that report, the response to AZD2281 was not correlated to the HPV status of the cell lines, or the initial PARP1 expression of the cell lines, nor was it correlated to the ability of the cell to conduct double-stranded DNA repair (DSB) [38]. For MK-1775, relatively low doses of the drug could be used against all cell lines. This was in accordance with one previous report [46]. There the effects of AZD-1775 (the Astra Zeneca equivalent to the Merk MK-1775) on the HPV + cell line UM-SCC-47 were investigated also in combination with cisplatin and it was shown that even low doses of AZD-1775 could sensitize this cell line to cisplatin through apoptosis [43]. In a more recent study, AZD-1775 was combined with different Chk1 inhibitors and the effects on radio sensitization were examined [44]. In that study, synergistic effects were observed when combining Chk1 inhibitors with the WEE1 inhibitor AZD-1775, and despite using lower doses of the two drugs prominent effects on radio sensitization were disclosed [47].
When we then continued with our investigation we could show that by combining the PARP inhibitor BMN-673 and the WEE1 inhibitor MK-1775 synergy was mainly disclosed for CU-OP-2 and not for CU-OP-3 as presented above for the other combinations. In addition, some positive effects were obtained for CU-OP-20 and CU-OP-17. Here, it is not possible to compare our data to corresponding data from others in head and neck cancer. So far namely, to our knowledge using PARP and WEE1 inhibitors in combination for oropharyngeal or head and neck cancer has still not been pursued, despite positive effects for other cancers, as shown and reviewed by others [36,37,[48][49][50].
Analyzing possible mechanisms for the above, obtaining an efficient decrease in cell viability and proliferation in the CU-OP cell lines after treatment with the CDK4/6 inhibitor PD-33291 and combining it with PI3K and FGFR inhibitors, BYL719 and JNJ-42756493 can have several explanations. A strong response after using these combinations could be that both CDK4/6 and FGFR inhibitors affect the PI3K pathway, which could naturally enhance the effect of the inhibitors [51]. Furthermore, the CDK4/6 protein acts in the G1-S point of the cell cycle so the inhibitor might act at this cell cycle checkpoint [48]. More specifically, Cyclin D activates CDK4/6 which phosphorylates the retinoblastoma tumor suppressor protein (pRB) and thereby inhibits its effect on E2F factors. Consequently, pRB phosphorylation allows for activation of E2F factors allowing DNA replication and the entrance of the cell again into the cell cycle [52]. CDK4/6 inhibitors have therefore been suggested to be useful for therapeutic strategy, and by inhibiting CDK4/6, pRB could function and block E2F and interrupt cell cycle progression [53]. In breast cancer, the CDK4/6 inhibitor Palbociclib, has already been successfully used for the treatment of oestrogen receptor-positive metastatic breast cancer and other breast cancer subtypes [54,55]. Furthermore, the potential positive clinical role of Palbociclib is suggested in combination with EGFR inhibitor Cetuximab in metastatic and advanced HPV unrelated head and neck cancer [56]. The latter has also been supported experimentally in HPV − cell lines and also discussed in a relatively recent review [57,58].
The fact that there were synergistic effects using the PARP inhibitor BMN-673 together with the WEE1 inhibitor MK-1775 was not completely unexpected. PARP can namely stabilize reversed forks during DNA replication and prevents fork collapse and the inhibition of PARP enhances stress on the replication forks and may be useful also on cells with high replicative stress in general [36]. Cells that then undergo replicative stress depend on WEE1 kinase, a key regulator of the G2 to M checkpoint. By inhibiting also WEE1 G2 to M is abrogated directing the cells with a deregulated G1 into premature mitosis resulting in mitotic catastrophe or apoptosis and cell death [36,37].
In this study, we found that the CU-OP-3 cell line was more sensitive to the synergistic effects of the CDK4/6 and PI3K and FGFR inhibitors, while the CU-OP-2 cell line was more sensitive to the latter PARP and WEE1 inhibitor combination. Why this is the case we do not presently know. The CU-OP-2 line harbors both a PIK3CA and FGFR3 mutation and should therefore theoretically be more sensitive to the PI3K and FGFR inhibitors, as actually is the case when compared to CU-OP-3 with the low doses used in this study. It is also notable that CU-OP-20 with a PIK3CA mutation but not an FGFR3 mutation is also somewhat sensitive to the PI3K inhibitor, but not to the FGFR inhibitor at the low doses used herein. We, therefore, suggest that is possible that a higher sensitivity to a single dose of an inhibitor may mask the synergy obtained as sometimes noted previously [42]. However, the fact that by combining different inhibitors one can obtain synergy against relatively resistant cell lines such as CU-OP-2 and CU-OP-3, that also are both radioresistant [59] is encouraging for the pursuit of personalized medicine. Nonetheless, more information on potential confounding by different genetic backgrounds, or additional mutations, would be of value to pursue in future studies.
There are limitations to this study. Despite the many cell lines used, one could suggest that even more cell lines should be tested. Furthermore, presently we have not conducted in vivo experiments on all cell lines tested so far, nor have we gone into detail with regard to molecular mechanisms or signaling pathways affected in the different cell lines. For the future one would be interested in being able to upfront have the possibility to predict which tumors would be eligible for a specific inhibitor combination. However, we are presently not there yet, but obviously, this will be one of our future goals.
To conclude, combinations of CDK 4/6 with PI3K or FGFR inhibitors, especially the former two disclose mainly positive and synergistic effects, but also some negative effects on the decrease of cell viability and the inhibition of cell proliferation of HPV +/− TSCC cell lines. In addition, in some cases, PARP1 and WEE1 inhibitors combined can present synergistic effects on specific HPV + TSCC cell lines. Combining these different inhibitors could be a beneficial therapeutic opportunity in specific cases and warrants further investigations to better stratify which approaches are best for specific tumors with specific molecular profiles.

Conclusions
All tested CDK4/6, PI3K, FGFR, PARP, and WEE1 inhibitors presented dose-dependent inhibitory effects on all tested TSCC lines. Synergy was mainly obtained when combining CDK4/6 with PI3K inhibitors, but less frequently obtained when combining CDK4/6 with FGFR inhibitors or PARP with WEE1 inhibitors and varied also depending on the individual cell lines.