Although CHS-828 has been established as an inhibitor of NAD synthesis with a mechanism of action similar to APO866, it had not been definitively identified as an inhibitor of NAMPT. Furthermore, previous attempts to identify resistance mechanisms for CHS-828 have yielded no definite results [24, 34] and resistance to APO866 has not been studied. In this study we introduce a novel potent analogue of CHS-828, TP201565, that shows >10 fold increased activity in sensitive cancer cell lines compared to APO866 and CHS-828. Using computer modelling, we find that CHS-828 and TP201565 most likely function as inhibitors of NAMPT by binding to the same site as APO866 and nicotinamide. This is further supported by the fact that a sepharose linked TP201565 analogue can precipitate NAMPT from cellular lysate and that this binding can be abrogated by co-incubation with either APO866 or TP201565. We also show how mutations in NAMPT can block the binding of APO866, CHS-828 and TP201565. Together, these data conclusively determine the pyridyl cyanoguanidines, CHS-828 and TP201565, as competitive inhibitors of NAMPT. This is in agreement with very recently reported data, which through the use of metabolic screens, in vitro NAMPT assays and computer modelling reach a similar conclusion for CHS-828 .
We developed four novel cell lines with acquired drug resistance towards NAMPT inhibitors and including NYH/CHS cells we observed mutations in all except PC-3/TP201565. Interestingly, the H191R mutation was detected in two different resistant cell lines derived from HCT-116. Furthermore, the distinct deletion D93del was located in both NYH/CHS and NYH/APO866. This suggests that these mutations exist in a small subpopulation of the parental cell line and that in these cases resistance has been obtained from selection rather than de novo mutations. The fact that we never obtained resistance from treating NYH cells with TP201565 despite prolonged growth with drug may be due to the fact that cells containing the NAMPTD93del were not present in the original flask used for inducing resistance to TP201565. We found that the observed mutations in NAMPT could induce resistance similar to or above that seen in the original resistant cell line which strongly suggests the mutations as the primary mechanistic cause of resistance. Furthermore, we demonstrated that whereas H191R abrogated NAMPT inhibitor binding in computer modelling it only reduced the enzymatic NAMPT activity in vitro by 50% - similar to what we found for the K342R mutation. The change in kcat may be due to altered substrate affinity resulting in an increased KM value. Also, we showed that the H191R mutation strongly increased the IC50 value for APO866 inhibition in vitro. The relatively high IC50 value of NAMPTwt obtained in vitro correlated with the previous results of others . D93del on the other hand reduced kcat
in vitro. Although this may reduce overall NAMPT activity compared to wild type NYH cells, it is likely that enough NAD is produced to keep up with the cellular demands and thus no growth disadvantage is inferred. NYH/CHS cells have previously been shown to retain the D93del mutation and full resistance after 30 passages without drug present in the media  (unpublished results, Olesen, UH). Similarly we have observed retained resistance for the remaining resistant cell lines after culture for more that 15 weeks without drug. We speculate that the induction of resistance by D93del and Q388R may be due to interference with or abrogation of dimerisation of NAMPT as both are located in the dimer interface. The NAMPT monomer may potentially retain lower enzymatic activity while not binding the known NAMPT inhibitors. Alternatively, the resistant cell lines may have adapted through reduced NAD consumption or up-regulation of other sources of NAD such as de novo synthesis.
Besides the data on the H191R mutation in figure 6B, the observed difference in kinetics and inhibitory constants between the different mutants cannot reliably be predicted on the basis of the computer docking study performed in this study. To achieve such correlations molecular dynamics calculations of the entire enzyme (including specific solvation) for tens of nanoseconds would be necessary. However, this is far beyond what is feasible with our current computational resources.
Notably, CHS-828 seems to show an upper level in LD50 values around 5-10 μM which is significantly lower than what is seen for APO866. We speculate that this may be due to a secondary mechanism of cytotoxicity occurring at above 1 μM. This has also previously been suggested [36, 37]. Although this currently unknown mechanism may be less important in cellular and molecular assays, it may well be of interest clinically for CHS-828, specifically since serum concentrations of the drug have been reported to reach 11 μM in clinical trials . However, it is unlikely that it is related to the previously proposed action of CHS-828 as an inhibitor of the I-κB kinase (IKK) complex . CHS-828 showed no inhibition of IKKα, IKKβ or TBK1 enzymes in vitro (unpublished results, Olesen, UH). APO866 displayed a surprisingly specific induction of cell death in tumour cells ranging five orders of magnitude from close to 1 nM to more than 30 μM in vitro and a similar specificity was seen for TP201565. Finally, the specific nature required of mutations in NAMPT to confer resistance combined with the lack of resistance induced even at prolonged drug treatment in a number of combinations of cell lines and NAMPT inhibitors may indicate that resistance is not induced frequently unless a suitable mutation in NAMPT is already present in a population of tumour cells. However, we also found that the resistance is not easily reversible. Also, no significant increase in tumour doubling times occurred in vivo for the resistant cell lines. Rather, we find that HCT-116/APO866 xenografts displayed reduced tumour doubling times compared to HCT-116. As we did not observe a similar trend for PC-3/TP201565 it seems unlikely to be due to increased NAMPT levels. Rather, the induction of resistance may have resulted in the development of an unidentified growth advantage in HCT-116/APO866. Further, the in vitro resistance also conferred insensitivity to APO866 treatment in xenograft models shown for HCT-116/APO866.
The PC-3/TP201565 cell line is highly resistant towards NAMPT inhibitors while displaying up-regulation but no mutations of NAMPT. The up-regulation could be due to increased gene copy number. We found that the basal NAMPT activity in the resistant cell line and also in transfected NAMPT over-expressing HEK293T/WT was much higher than in wild type PC-3 and HEK293T cells. The fact that PC-3/TP201565 cells displayed higher total activity than HEK293T/WT despite the latter having higher expression of NAMPT agrees with previous findings that endogenous NAMPT has higher activity compared to recombinant NAMPT . However, the IC50 was not significantly changed and although the absolute NAMPT activity remained high relative to wild type PC-3 and HEK293T at concentrations of APO866 up to 1-10 μM, it would not seem to fully explain the >10 μM LD50 value observed for APO866 in PC-3/TP201565. Further, the resistance appeared to be specific to NAMPT inhibitors as cross-resistance to other chemotherapeutics was not observed and it was not related over-expression of MDR1 and 2. Also, CHS-828 has displayed only low to moderate sensitivity to mechanisms of MDR based on over-expression of Pgp170 and MRP, increased levels of gluthation and tubulin-associated MDR [23, 40]. Thus, we believe that PC-3/TP201565 cells are likely to possess a further, yet unknown, specific mechanism of resistance. Still, the over-expression explains part of the resistance and, as also seen for HEK293T/WT cells, high levels of NAMPT may be sufficient to induce resistance which would be significant in a clinical setting. This is similar to the situation with thymidylate synthase where high expression levels predict resistance to 5-fluorouracil and a poor treatment outcome [41, 42].
The crystal structure of NAMPT in complex with APO866 has previously led to suggestions on improving the potency of APO866 based on its interaction with the NAMPT binding site [14, 43]. So far, published APO866 analogues have shown less activity in cellular assays [44, 45]. We found that the mutations in NAMPT identified in the resistant cell lines show a tendency to locate either in the binding site or dimer interface. These mutations may directly or indirectly adjust the shape of the binding site so that it is unsuitable for the relatively large NAMPT inhibitors while still accessible for nicotinamide and NMN. Also, others have recently published a single resistant cell line showing a point mutation, G217R, which is close to the active site of NAMPT and is responsible for high-grade resistance towards CHS-828 [14, 35]. We hope that the identification of a new class of NAMPT inhibitors and the characterization of resistance inducing mutations in NAMPT may be useful in developing second-generation NAMPT inhibitors with higher potency and potentially less affected by acquired resistance.