Identification of Hsp90 inhibitors as potential drugs for the treatment of TSC1/TSC2 deficient cancer

Inactivating mutations in either TSC1 or TSC2 cause Tuberous Sclerosis Complex, an autosomal dominant disorder, characterized by multi-system tumor and hamartoma development. Mutation and loss of function of TSC1 and/or TSC2 also occur in a variety of sporadic cancers, and rapamycin and related drugs show highly variable treatment benefit in patients with such cancers. The TSC1 and TSC2 proteins function in a complex that inhibits mTORC1, a key regulator of cell growth, which acts to enhance anabolic biosynthetic pathways. In this study, we identified and validated five cancer cell lines with TSC1 or TSC2 mutations and performed a kinase inhibitor drug screen with 197 compounds. The five cell lines were sensitive to several mTOR inhibitors, and cell cycle kinase and HSP90 kinase inhibitors. The IC50 for Torin1 and INK128, both mTOR kinase inhibitors, was significantly increased in three TSC2 null cell lines in which TSC2 expression was restored. Rapamycin was significantly more effective than either INK128 or ganetespib (an HSP90 inhibitor) in reducing the growth of TSC2 null SNU-398 cells in a xenograft model. Combination ganetespib-rapamycin showed no significant enhancement of growth suppression over rapamycin. Hence, although HSP90 inhibitors show strong inhibition of TSC1/TSC2 null cell line growth in vitro, ganetespib showed little benefit at standard dosage in vivo. In contrast, rapamycin which showed very modest growth inhibition in vitro was the best agent for in vivo treatment, but did not cause tumor regression, only growth delay.

147 Kinase inhibitor library screen and IC50 determination 148 The  As further confirmation of mutational status for these cell lines, we performed 251 immunoblotting to assess expression of TSC1 and TSC2, and mTORC1 signaling, which 252 should be activated in the absence of either TSC1 or TSC2 (Fig 1). There was no expression   (Fig 2).

308
309 Kinase inhibitor library screen and IC 50 determination

310
The cell lines SNU-878, SNU-886, SNU-398, CAL-72, and PEER were screened for 311 sensitivity to kinase inhibition using a kinase inhibitor-focused library (LINCS), which 312 contained 197 selective kinase inhibitors (S1 Table). This library was composed of 313 commercially available and self-developed pharmacophore-diverse ATP competitive kinase . In this initial screen, a single dose of inhibitor was used, 600nM. All 325 compounds that showed > 50% reduction in growth for one or more cell lines in this assay 326 were considered initial positive hits and were subject to further study on all cell lines.
327 A wide variety of inhibitors showed a positive signal in this initial assay, including multiple 328 mTOR, CDK, PLK, CHK, Aurora, and HSP90 inhibitors ( Table 2).

418
419  510 Tumor size of treated mice is significantly smaller on day 22 compared to mice, which 511 received vehicle (n=8) (b).

513
We then examined the potential benefit of combination treatment, with ganetespib and 514 rapamycin, or with ganetespib and INK 128, in this xenograft model system. Combination 515 ganetespib-rapamycin had similar effects on growth as rapamycin alone (Fig 8d). In contrast, 516 ganetespib-INK 128 showed apparent synergy with a greater effect on growth than either drug 517 alone although this did not achieve statistical significance (Fig 8 c, S22e and  530 the tumors treated with ganetespib or vehicle (Fig 9).   (Fig 10). Similarly S6 and 4EB-P1 levels were similar under all treatments, 545 with the exception of the ganetespib and INK 128 combination, which reduced both 546 considerably (Fig 10). TSC2 expression was low in all xenografts, as expected, with some 547 expression seen likely due to ingrowing vessels and connective tissue. pAKT (Ser473) 548 expression was also universally low, as expected (Fig 10). Both rapamycin alone and in 549 combination with ganetespib abolished pS6K (Thr389) and pS6 (Ser240/244) expression; 550 while INK 128 alone or in combination reduced pS6 (Ser240/244) and eliminated pS6K 551 (Thr389) (Fig 10). None of the treatments affect 4E-BP1 isoform expression, while the 552 ganetespib and INK 128 combination reduced overall 4E-BP1 expression. Livers from the 553 treated mice showed similar effects as the xenograft nodules, from each drug, except that pS6 554 (Ser240/244) was relatively highly expressed in the INK 128 treated liver (Fig 10).