Synergistic enhancement of the potency and selectivity of small molecule transcriptional inhibitors

Abstract

In spite of their considerable therapeutic potential, the development of highly potent and selective transcriptional inhibitors has proven elusive. We demonstrate that combinations of transcriptional inhibitors of erbB2 expression and existing therapeutic agents that target erbB2 activity and lifetime lead to a synergistic increase in activity, with dose reductions as high 30 fold compared to individual agents. The synergy is selective for erbB2 overexpressing cancer cells. These results highlight the potential of a generalizable approach that will improve the utility of transcriptional inhibitors as both biochemical tools and potential therapeutics.

The aberrant function of a growing number of transcriptional activators is associated with the development and progression of human diseases such as cancer.^1–3 Molecules that interfere with the ability of transcriptional activators to control expression of their target genes thus have great promise as biochemical tools and therapeutics.^4–6 Activators regulate transcription through a complex network of interactions with transcriptional machinery proteins;^7,8 blocking these interactions inhibits transcription. However, the affinities of activators for their transcriptional machinery binding partners are modest (high nM - low µM K_Ds) and, correspondingly, small- and large molecule inhibitors of these interactions typically require micromolar concentrations to exert their effects, ultimately limiting their utility.^{6, examples: 9–11} Additionally, the selectivity of molecules that interact with shared coactivators is a recurring cause for concern.⁵ We previously^12–15 described the development of a new class of small molecule transcriptional inhibitors that mimic transcriptional activation domains; this includes i1 (Figure 1), an isoxazolidine that mimics the activation domain of the transcription factor ESX, and interferes with transcription of the ESX-regulated oncogene erbB2 at micromolar concentrations. Here we present a strategy that mitigates the potency and selectivity concerns of transcriptional inhibitors through a multi-pronged intervention against the ErbB2 regulatory pathway. The use of i1 in tandem with other agents that target the activity and lifetime of the erbB2 oncoprotein leads to simultaneous dose reductions of greater than 15-fold for both agents, and increased selectivity for erbB2+ cells up to 30-fold. This strategy should be readily applicable to other agents that disrupt the protein-protein interactions responsible for oncogene transcription.

Figure 1.

Schematic of erbB2 pathway and points of small molecule intervention.^27,28,31 Combinations that inhibit both the transcription of erbB2 and the lifetime or activity of the mature protein have a synergistic increase in activity against erbB2 driven cancer cells. Drug combinations to synergistically target disorders ranging from cancer to inflammation have been identified in recent years. These combinations often have the benefit of reduced toxicity in addition to increased selectivity relative to the single agents. However, applications to inhibitors of protein-protein interactions and, in particular, the transient and poorly characterized protein-protein interactions that regulate transcription have not reported.^24–26,29

We chose erbB2 for this investigation because of its clinical relevance and because the complexities of its regulatory network make effective treatment with individual agents a challenge. The erbB2 protein is a transmembrane tyrosine kinase that is over-expressed in approximately one quarter of breast cancers,¹⁶ where it has been shown to drive an aggressive phenotype marked by more rapid metastasis and shorter life expectancy than breast cancers that do not over-express erbB2.^17,18 Furthermore, erbB2 over-expressing (erbB2+) cancer cells are known to undergo growth arrest and cell death if erbB2 expression is suppressed.¹⁹ The clinical significance of erbB2 over-expression can be seen in the variety of existing treatments designed to suppress erbB2 signaling, including antibodies that target the protein's extracellular domain²⁰ and tyrosine kinase inhibitors, which target the protein's ability trans-phosphorylate other members of the erbB family and initiate cell survival and proliferation programs (Figure 1).²¹ These approaches have met with difficulty in clinical practice, but an increasing body of evidence suggests that although erbB2 driven cancers are adept at compensating for partial inhibition of erbB2 activity, they are still vulnerable to interventions that reduce erbB2 levels.^22,23

The discovery of multicomponent therapeutics has emerged in recent years as an effective strategy for increasing efficacy and decreasing off-target effects relative to single agents in a number of cases.^24–26 We reasoned that it might be possible to obtain similar benefits by combining i1 with other agents that target erbB2. One point of intervention along the erbB2 pathway is Hsp90, part of a chaperone complex that maintains erbB2 stability and assists in membrane localization.^27,28 The natural product geldanamycin reduces cellular erbB2 levels by binding to Hsp90 and inhibiting its function (Figure 1) but its toxicity prevent its use as a therapeutic agent.^27,28 Thus our initial efforts tested the hypothesis that dual targeting of the erbB2 pathway with the two PPI inhibitors geldanamycin and i1 would synergistically increase potency and specificity relative to the individual agents. As shown in Figure 2a, a 50:1 combination of i1:geldanamycin resulted in an IC₅₀ in SkBr3 (erbB2+) cells that is >10-fold lower than i1 alone. To test if the potency increase is truly synergistic, both the isobologram and multiplicative additivity (Bliss) models were employed. For the former, the IC₅₀s of fixed ratios of i1:geldanamycin were measured and compared to a hypothetical case representing additivity, in which both components act as though they are the same agent (Figure 2b).^24,25,29 IC₅₀ ratios (combination:single agent) that fall below the additivity line are indicative of positive synergy and by this measure, the combinations of the two PPI inhibitors exhibit an impressive degree of synergy. The 5:1 i1:geldanamycin combination is synergistic as defined by the multiplicative additivity or Bliss model (Figure 2c).^24,25,29 This degree of synergy increased over longer growth times, indicating robust inhibition of proliferation from combination treatment (Supporting Figure S1b–c). In addition, the combination of geldanamycin and i1 concomitantly produced an 85% drop in erbB2 levels (Supporting Figure S1a).

Figure 2.

a) The dose effect curves for i1 as a single agent and a 50:1 combination of i1:geldanamcyin after 3 days of dosing. b) The IC₅₀s of fixed dose ratios of i1 and geldanamycin were measured in SKBR3 cells after 3 days of dosing and plotted on an isobologram. c) The %effect (100 - %growth) for the indicated doses for the 3 day dosing period of a growth timecourse (Supplementary Figures S1c–d). Predicted additivity was calculated as indicated in the SI. d) The IC₅₀s from (b) were compared to IC50s for the same combinations in IMR90 cells (Supplementary Figure S2a–c) and the resulting ratios were plotted as shown, normalized to the effects of i1 and geldanamycin as single agents. The difference between the ratios for 5:1 and 50:1 i1:geldanamcyin is not statistically significant (p = 0.056)._Error bars indicate error compounded from one standard deviation of experiments performed in triplicate. For all other experiments, error bars indicate one standard deviation from experiments performed in triplicate unless noted otherwise. P values: *<0.05; **<0.01, ***<0.001

As outlined earlier, i1 displays modest selectivity for erbB2+ cancer cell lines and geldanamycin is broadly toxic. However, combinations of i1 and geldanamycin show increased selectivity for erbB2+ cancer cells when compared to non-tumorigenic IMR90 cells, cells whose growth is not driven by erbB2 (Figure 2d). This is most notable in comparison with geldanamycin alone, where the combinations produce a 20- to 35-fold selectivity improvement. These results indicate that the synergy is erbB2-dependent and not a result of general toxicity. These data further suggest that transcriptional inhibitors can be used in combinations with agents that have broad activity to selectively effect specific shared targets.

We next examined the potential for synergy between i1 and lapatinib, a reversible erbB2/erbB1 kinase inhibitor that is used clinically in the treatment of erbB2+ cancers (Figure 1).³⁰ An initial trial of a 500:1 ratio of i1:lapatanib produced a >10-fold decrease in the IC₅₀ relative to i1 alone in SkBr3 (Figure 3a). That this decrease was due to synergy was tested as before with via both the isobologram and multiplicative additivity (Bliss) methods in SkBr3 cells (Figure 3b). The IC₅₀ ratios of the i1:lapatinib combinations fell significantly below the additivity line, demonstrating a synergistic effect. Consistent with the impact on viability, i1 and lapatinib had moderate effects on erbB2 and phosphorylated erbB2 levels as single agents. However the i1:lapatinib combination was significantly (p < 0.05) more effective at reducing the total amount of active (phosphorylated) erbB2 than equivalent amounts of either i1 or lapatinib (Supporting Figure S3a).

Figure 3.

a) The dose effect curves for i1 as a single agent, and a 500:1 combination of i1:lapatinib after 2 days of dosing. b) The IC₅₀'s of fixed dose ratios of i1 and lapatinib were measured in SKBR3 cells after 2 days of dosing and plotted on an isobologram. c) The %effect (100 - %growth) for the indicated doses for the 3 day dosing period of a growth timecourse (Supplementary Figures S3f–g). Predicted additivity was calculated as indicated in the SI. d) The IC₅₀'s from (b) were compared to IC50's for the same combinations in IMR90 cells (Supplementary Figure S3e) and the resulting ratios were plotted as shown, normalized to the effects of i1 and lapatinib as single agents. Error bars indicate error compounded from one standard deviation of experiments performed in triplicate. For all other experiments, error bars indicate one standard deviation from experiments performed in triplicate unless noted otherwise. P values: *<0.05; **<0.01, ***<0.001

In addition to the increased potency, combinations of i1 and lapatinib are less toxic to erbB2-negative, non-tumorigenic IMR90 cells, leading to greater selectivity than the use of i1 in isolation (Figure 3d). As an additional readout for synergy, we dosed colonies of SkBr3 cells with compound (5 µM i1, 10 nM lapatinib or a combination of the two) for 9 days (Supporting Figure S3f). The combination treatment was much more effective than either i1 or lapatinib in isolation or the multiplicative sum of the individual effects, (Supporting Figure S3g). In contrast, combinations of i1 and the erbB1 selective kinase inhibitor erlotinib³¹ did not display significant synergy (Supporting Figure S4). This result is consistent with models that implicate the erbB2/erbB3 dimer as the primary driver of oncogenesis.^22,23

In conclusion, by using a small-molecule combination that simultaneously curbs expression of the genetic driving force behind a diseased state while also targeting related cellular processes, we achieve a synergistic increase in effect that is specific to the target cell population. Although synergistic combinations of well-established drugs have recently emerged for improving efficacy,^26,29 this is the first time that synergistic interactions between two molecules, which target protein-protein interactions have been used to overcome the activity and selectivity issues common to this class of molecules. In doing so, these data support a generalizable approach that will improve the utility of transcriptional inhibitors as both biochemical tools and potential therapeutics. As outlined earlier, i1 displays modest selectivity for erbB2+ cancer cell lines and geldanamycin is broadly toxic.²⁸ However, combinations of i1 and geldanamycin show increased selectivity for erbB2+ cancer cells when compared to non-tumorigenic IMR90 cells, cells whose growth is not driven by erbB2 (Figure 2d). This is most notable in comparison with geldanamycin alone, where the combinations produce a 20- to 35-fold selectivity improvement. These results indicate that the synergy is erbB2-dependent and not a result of general toxicity. These data further suggest that transcriptional inhibitors can be used in combinations with agents that have broad activity to selectively effect specific shared targets.