The heat shock protein 90 inhibitor, AT13387, displays a long duration of action in vitro and in vivo in non‐small cell lung cancer

Abstract

A ubiquitously expressed chaperone, heat shock protein 90 (HSP90) is of considerable interest as an oncology target because tumor cells and oncogenic proteins are acutely dependent on its activity. AT13387 (2,4‐dihydroxy‐5‐isopropyl‐phenyl)‐[5‐(4‐methyl‐piperazin‐1‐ylmethyl)‐1,3‐dihydro‐isoindol‐2‐yl] methanone, l‐lactic acid salt) a novel, high‐affinity HSP90 inhibitor, which is currently being clinically tested, has shown activity against a wide array of tumor cell lines, including lung cancer cell lines. This inhibitor has induced the degradation of specific HSP90 client proteins for up to 7 days in tumor cell lines in vitro. The primary driver of cell growth (mutant epidermal growth factor receptors) was particularly sensitive to HSP90 inhibition. The long duration of client protein knockdown and suppression of phospho‐signaling seen in vitro after treatment with AT13387 was also apparent in vivo, with client proteins and phospho‐signaling suppressed for up to 72 h in xenograft tumors after treatment with a single dose of AT13387. Pharmacokinetic analyses indicated that while AT13387 was rapidly cleared from blood, its retention in tumor xenografts was markedly extended, and it was efficacious in a range of xenograft models. AT13387's long duration of action enabled, in particular, its efficacious once weekly administration in human lung carcinoma xenografts. The use of longer‐acting HSP90 inhibitors, such as AT13387, on less frequent dosing regimens has the potential to maintain antitumor efficacy as well as minimize systemic exposure and unwanted effects on normal tissues. (Cancer Sci, 2012; 103: 522–527)

The super‐chaperone system is involved in the folding and maturation of newly synthesized proteins.1, 2 HSP90 in particular aids in the folding and maturation of a distinct subset of proteins, which includes kinases, cell surface receptors and transcription factors.3, 4 The N‐terminal domain ATPase activity of HSP90 is essential for this function.5 Inhibition of this domain induces remodeling of the HSP90 chaperone complex, resulting in the recruitment of ubiquitin ligases, polyubiquitination and subsequent proteasomal degradation of HSP90 client proteins.6, 7 Through this mechanism, the inhibition of a single target enzyme can have a wide effect on the stability and, hence, the function of a large set of client proteins. As many oncogenic proteins are HSP90 clients, HSP90 inhibition has been found to have broad antitumor effects.8, 9, 10 In contrast to more recent targeted therapies, where the appearance of new driver mutations or resistance mutations result in a loss of efficacy, client protein mutation increases dependence on HSP90 chaperoning activity as these mutations tend to render the proteins less stable.11, 12, 13 Previous studies have also demonstrated that the constitutively activated mutant forms of EGFR are particularly dependent on HSP90 both in vitro and in vivo,14, 15, 16 indicating an HSP90 inhibitor may be particularly efficacious in mutant EGFR tumors.

After HSP90 inhibition, it has been observed that client protein depletion is transient and that levels return to normal upon drug withdrawal.17, 18 Extending the duration of target inhibition in a tumor would be expected to extend the antitumor effect. Inhibitor accumulation in tumors, coupled with rapid clearance from blood and normal tissue, has been observed for multiple classes of HSP90 inhibitors and may be due in part to the kinetics of the inhibitor binding to tumor‐derived HSP90 complexes.19, 20, 21, 22, 23 The potential advantage of compounds binding to HSP90 for a longer duration has been illustrated previously by comparing the monomeric ansamycin inhibitor 17‐AAG with a dimeric variant that exhibited slow off‐rate kinetics.18, 24

Several structurally distinct HSP90 inhibitors are progressing through clinical development with early indications of clinical responses in published phase I and II data.25, 26, 27 While there are many similarities in the pharmacology of these different agents,28, 29 one prominent variable is the duration of the pharmacodynamic effect reported.20, 30, 31

We have used fragment‐based drug discovery to identify the high‐affinity, long‐acting HSP90 inhibitor, AT13387, which is currently being evaluated in clinical trials. In the studies described here, we have used this inhibitor to demonstrate that maximizing the duration of effect in the target tissue enabled sustained client protein depletion and extended tumor growth inhibition with less frequent dosing of AT13387. This was especially manifested for the non‐small lung cancer cell line NCI‐H1975, in which mutant EGFR is a particularly sensitive HSP90 client. The use of longer‐acting HSP90 inhibitors in tumors could lead to less frequent dosing regimens, limit exposure of non‐target tissues to active drugs and reduce potential safety concerns in the clinic.

Materials and Methods

Materials.

AT13387 was synthesized at Astex Pharmaceuticals (Cambridge, UK) and stored as a lyophilized powder (Table 1). Synthesis of AT13387 is as described by Woodhead et al.32 All other reagents were purchased from Sigma (Poole, UK) unless otherwise stated.

Table 1.

Anti‐proliferative effect of AT13387 on a panel of human tumor cell lines

Origin	Cell line	AT13387 GI50 (nM)
Normal prostate	PNT2	480
Colon carcinoma	HCT116	48
	HT‐29	78
	SW620	210
Lung carcinoma	A549	22
	NCI‐H1975	27
	NCI‐H1993	63
	NCI‐H1650	13
Breast carcinoma	MCF‐7	53
	MDA‐MB‐231	260
	MDA‐MB‐468	25
	SK‐BR3	63
	T47D	29
	BT474	13
Multiple myeloma	U266	58
	RPMI 8226	70
Pancreatic cells	PANC1	55
Hepatoma	Huh‐7	22
Prostate carcinoma	DU145	94
	PC3	120
	LNCaP	77
	22Rv1	46
Uterine sarcoma	MES‐SA	53
	MES‐SA/Dx5	42
Ovarian	SKOV3	44
Leukemia	HL60	22
	K562	47
	MV4‐11	13
Melanoma	A375	18
	SkMel 28	44
Cholangiocarcinoma	TFK‐1	19

ATT13387, (2,4‐dihydroxy‐5‐isopropyl‐phenyl)‐[5‐(4‐methyl‐piperazin‐1‐ylmethyl)‐1,3‐dihydro‐isoindol‐2‐yl] methanone, l‐lactic acid salt.

Protein production.

The ATPase domain of HSP90α was expressed as a His₆‐tagged fusion and purified using Ni‐NTA metal‐affinity chromatography and Superdex 75 gel‐filtration chromatography. Proteins were concentrated in 20 mM Tris (pH 7.4) containing 150 mM NaCl and 1.0 mM DTT.

HSP90 competition isothermal calorimetry.

K _d values for AT13387 and 17‐AAG binding to HSP90 were determined with a competition Isothermal Calorimetry (ITC) format. ITC experiments were performed on a MicroCal VP‐ITC (Northhampton, MA, USA) at 25°C in a buffer comprising 25 mM Tris, 100 mM NaCl, 1 mM MgCl₂ and 1 mM Tris(2‐carboxyethyl)phosphine at pH 7.4 in order to maintain the higher affinity, reduced form of 17‐AAG.32, 33

Cell culture and reagents.

The human cell lines A375, 22RV1, BT474, DU145, LNCaP, MCF‐7, MDA‐MB‐468, MV4‐11, NCI‐H1650, NCI‐H1975, NCI‐H1993, PANC1, PC‐3, SKBr3, U266, and SK‐MEL‐28 cells were obtained from the American Type Culture Collection (Teddington, UK). A549, HCT‐116, HL60, HT‐29, K562, MCF‐7, MDA‐MB‐231, MES‐SA, MES‐SA/DX5, PNT2, SW620 and T47D cells were obtained from the European Collection of Cell Cultures (Porton Down, UK). RPMI‐8226 and TFK1 cells were obtained from Deutsche Sammlung von Mikroorganismen und Zellkulturen (Braunschwieg, Germany). Cells were grown in their recommended culture medium, supplemented with 10–20% fetal bovine serum and maintained at 37°C in an atmosphere of 5% CO₂.

Proliferation assays.

Cells were seeded into 96‐well plates before the addition of compound in 0.1% (v/v) DMSO. GI₅₀ were determined using a 10‐point dose response curve for three cell doubling times. After compound incubation 10% (v/v), Alamar blue (Biosource International, Camarillo, CA, USA) was added, and cells were incubated for a further 4 h. Fluorescence was read at λ_ex = 535 nm and λ_em = 590 nm.

Western blot samples and analysis.

Cells were incubated overnight at 37°C, followed by treatment with AT13387 or 17‐AAG for the indicated time. Floating and adherent cells were collected and treated as described previously.34 Samples were resolved by SDS‐PAGE and immunoblotted with antibodies specific for Akt, pAkt, androgen receptor, cleaved PARP, c‐Met, EGFR, phospho‐EGFR, Erk, phospho‐Erk, S6, phospho‐S6, Stat3, phospho‐Stat3 (Cell Signaling Technologies, Cambridge, UK), B‐Raf, CDK4, HER2 Raf‐1 (Santa Cruz Antibodies, Santa Cruz, CA, USA), actin (Abcam, Cambridge, UK), HSP70 (StressGen Biotechnologies, Victoria, BC, Canada), or GAPDH (Chemicon International, Temecula, CA, USA). This was followed by infrared dye labeled anti‐rabbit or anti‐mouse antibodies (Licor Bioscience, Lincoln, NE, USA). Blots were scanned to detect infrared fluorescence on the Odyssey Infrared Imaging System (Licor Bioscience).

Activation of signaling pathways was accomplished by serum‐starving BT474 or U266 cells for 18 h, followed by 15 min stimulation with either insulin‐like growth factor‐1 or IL‐6, before cell lysates were harvested.

Xenograft models.

Male athymic BALB/c mice (nu/nu) were obtained from Harlan UK (Bicester, UK) and were given food and water ad libitum. The care and treatment of experimental animals were in accordance with the United Kingdom Coordinating Committee for Cancer Research guidelines and with United Kingdom Animals (Scientific Procedures) Act 1986.35, 36

Mice were injected subcutaneously with tumor cells, NCI‐H1975 at 2 × 10⁶ cells in 100 μL serum‐free medium. Animals were randomized into groups of seven or eight, and treatment was started when tumors were approximately 100 mm³ in mean diameter. AT13387 was dissolved in 17.5% (w/v) hydroxypropyl‐β‐cyclodextrin and administered intraperitoneally at various dose levels in a dose volume of 10 mL/kg. Tumor burden was estimated from caliper measurements. Tumor growth delay was analyzed with the log–rank test on the survival curves (Graphpad Software, GraphPad Prism v3.02, La Jolla, CA, USA). Tolerability was estimated by monitoring body weight loss and survival over the course of the study.

Pharmacokinetic analysis.

Pharmacokinetic parameters were determined after intraperitoneal (80 mg/kg) administration of AT13387 to athymic BALB/c mice bearing subcutaneous NCI‐H1975 xenografts. Blood and tumor samples were collected at indicated times following dosing. Blood was drawn from the heart into a heparinized syringe. Tumors were dissected and all samples were stored at −20°C until analysis. Samples were prepared for analysis as described previously.34 Quantification was by comparison with a standard calibration line. The limit of detection for AT13387 was 150 nM in blood and 120 nM in tumor.

Analysis of tumor sample pharmacodynamic markers.

Analysis of tumor sample pharmacodynamic markers was carried out as described previously using the indicated antibodies.17, 34

Results

Binding of AT13387 to HSP90 in vitro.

AT13387 is a novel small molecule inhibitor of HSP90 that was discovered with a fragment‐based drug discovery approach that has been described in detail elsewhere.32

X‐ray crystallographic analysis of AT13387 in complex with the N‐terminal domain of HSP90 showed that it bound within the N‐terminal ATPase catalytic site and overlapped with the binding site for ATP.32 Direct binding of AT13387 to HSP90 was investigated with isothermal calorimetry. The K _d for AT13387 binding was 0.7 nM. This compares to a K _d of 6.7 nM for the binding of the ansamycin 17‐AAG to the same site; the mean stoichiometry of binding for AT13387 was 1.03.

The inhibition of a number of isolated kinases by AT13387 was also investigated including CDK1, CDK2, CDK4, FGFR3, PKB‐β, JAK2, VEGFR2, PDGFR‐β and Aurora B. None of the tested kinases were significantly inhibited at concentrations below 30 μM (data not shown).

Antitumor activity of AT13387.

AT13387 was a potent inhibitor of the proliferation and survival of many different cell lines from a variety of different tumor types in vitro. Across a panel of 30 tumor cell lines, AT13387 potently inhibited cell proliferation with GI₅₀ (n = 2) values in the range 13–260 nM (Table 1). The similarity in the GI₅₀ values for AT13387 in the MES‐SA cell line and the P‐glycoprotein–expressing variant, MES‐SA/Dx5, (53 vs 42 nM) indicated that it is not a substrate for the P‐glycoprotein transporter. AT13387 also inhibited proliferation of the non‐tumorigenic human prostate epithelial cell line PNT2 with a GI₅₀ (n = 2) value of 480 nM.

Among the cell lines most sensitive to AT13387 were several known to be dependent on receptor tyrosine kinases, such as EGFR, HER2, c‐Met, and FLT3. These included non‐small cell lung cancer cell lines with mutant EGFR (NCI‐H1975; GI₅₀, 27 nM) or amplified c‐Met (NCI‐H1993; GI₅₀, 63 nM). Other sensitive cell lines included a cholangiocarcinoma line (TFK1; GI₅₀, 19 nM) and IL‐6–responsive tumor cell lines, notably the multiple myeloma lines RPMI‐8226 (GI₅₀, 70 nM) and U266 (GI₅₀, 58 nM). We are unaware of any other reports demonstrating HSP90 inhibitor sensitivity in a cholangiocarcinoma cell line. It is noteworthy that IL‐6 has been shown to contribute to tumorigenesis and cell survival in cholangiocarcinomas.37, 38 The benefit of current chemotherapeutic options for cholangiocarcinomas is controversial, with some reports claiming a survival benefit while others report no impact on disease or survival benefit.40, 41, 42, 43 The sensitivity of IL‐6 receptor‐signaling and TFK1 cells to AT13387 suggests that cholangiocarcinomas are attractive candidates for further investigation in preclinical and clinical studies.

Cell‐based mechanism of action studies.

The molecular fingerprint of HSP90 inhibition is characterized by an increase in HSP72 and corresponding depletion in HSP90 client proteins (e.g. CDK4, Raf‐1 and Akt).44, 45 The time‐ and concentration‐dependence of these effects for AT13387 were demonstrated in the non‐small cell lung cancer cell line NCI‐H1975, which harbors two activating mutations (L858R and T790M) in EGFR (Fig. 1A,B). At concentrations of 30 nM and above, AT13387 induced depletion of a range of HSP90 client proteins in NCI‐H1975 cells within 6 h of exposure. In further studies, client proteins believed to be the primary drivers of cell growth were investigated. In addition to loss of mutant EGFR in NCI‐H1975 cells, AT13387 also caused depletion of BRAF in the melanoma cell line A375, the androgen receptor in the prostate cancer cell line 22Rv1, Her2 in the breast carcinoma cell line BT474 and c‐Met in the lung carcinoma line NCI‐H1993 (Fig. 1C). Inhibition of signaling via growth and survival pathways (e.g. Erk1/2 and PI3K/Akt) is an important consequence of HSP90 inhibition, and suppression of phosphorylation in these signaling pathways is often the earliest event observed.28 We have found IL‐6–dependent cells to be very sensitive to AT13387, which correlates with the inhibition of the downstream signaling pathway. IL‐6–induced phospho‐signaling through both the Stat3 and MAPK pathways was ablated in the presence of AT13387 in the multiple myeloma cell line U266 and breast carcinoma cell line BT474 (Fig. 1D).

Figure 1.

Effects of AT13387 treatment on HSP90 client protein levels. (A) NCI‐H1975 cells treated with the indicated doses of AT13387 for 18 h before lysates were harvested and equivalent amounts of protein from each lysate were resolved by SDS‐PAGE and immunoblotting with the indicated antibodies. Control samples are vehicle only treated cells. (B) NCI‐H1975 cells treated with the indicated doses of AT13387 for 1, 3, 6, 12, 24 or 48 h before lysates were harvested and immunoblotted with anti‐EGFR antibody. (C) 22Rv1, NCI‐H1975, BT474, A375 and NCI‐H1993 cells were untreated or treated for 18 h with 1 μM AT13387 before they were harvested for immunoblot analyses. (D) Serum‐starved U266 and BT474 cells were treated with 1 μM AT13387 for 18 h before they were induced with IL‐6 or IGF‐1 and harvested for immunoblot analyses. AT13387, (2,4‐dihydroxy‐5‐isopropyl‐phenyl)‐[5‐(4‐methyl‐piperazin‐1‐ylmethyl)‐1,3‐dihydro‐isoindol‐2‐yl] methanone, l‐lactic acid salt; CTL, control; EGFR, epidermal growth factor receptor; HSP, heat shock protein; pAkt, phospho‐Akt.

AT13387 inhibition of HSP90 results in prolonged client protein depletion.

Following the observations that AT13387 treatment depleted client proteins and ablated phospho‐signaling, we investigated the duration of these biochemical effects in vitro. Melanoma (A375), breast carcinoma (BT474), and lung carcinoma cell lines (NCI‐H1993 and NCI‐H1975) were treated with 1 μM of AT13387 to ensure complete inhibition of HSP90. The duration of this inhibition was then studied at various times after the inhibitor had been removed (Fig. 2A–D). At doses above the IC₅₀ value, the results demonstrated that the extent and duration of client protein depletion was generally dependent on the length of exposure to compound and that 24‐h exposure maximally suppressed levels of client proteins and phospho‐signaling in these cells (Fig. 1A,B). After removal of AT13387 from the medium, A375 melanoma cells showed significant suppression of all tested HSP90 client proteins for at least 48 h (Fig. 2A). Additionally, signaling through the MAPK (phospho‐Erk1/2) and PI3K (phospho‐S6) pathways was suppressed for at least 48 h in AT13387‐treated A375 cells. Suppression by AT13387 of the primary drivers of growth in BT474 (HER2), NCI‐H1993 (c‐Met) and NCI‐H1975 (EGFR) cell lines was found to last in excess of 7 days after removal of the AT13387 (Fig. 2B–D). Mutant EGFR found in NCI‐H1975 cells was particularly sensitive to depletion following AT13387 treatment, with this prolonged client depletion requiring only a 7‐h exposure to AT13387 (Fig. 2D). Suppression of phospho‐signaling was also maintained over this extended period. An expected consequence of this long duration of action is reduced survival of these cells. Indeed, the corresponding level of cleaved PARP in cells treated with AT13387 correlates with a significant decrease in PI3K signaling.

Figure 2.

AT13387 has a long duration of inhibition in vitro. (A) A375, (B) BT474 and (C) NCI‐H1993 cells were treated with 1 μM of AT13387 for 24 h, washed with PBS and re‐incubated with fresh culture medium. (D) NCI‐H1975 cells were treated with 1 μM AT13387 for 7 h, washed with PBS and re‐incubated with fresh culture medium. Samples were taken at the indicated times after PBS wash for immunoblot analyses. AT13387, (2,4‐dihydroxy‐5‐isopropyl‐phenyl)‐[5‐(4‐methyl‐piperazin‐1‐ylmethyl)‐1,3‐dihydro‐isoindol‐2‐yl] methanone, l‐lactic acid salt; EGFR, epidermal growth factor receptor; HSP, heat shock protein; pAkt, phospho‐Akt.

Pharmacodynamic profile of AT13387.

Given the sensitivity of EGFR to AT13387 in vitro, we wanted to determine the duration of response to AT13387 elicited in vivo. The extended duration of action of HSP90 inhibition following a single maximum dose of AT13387 (80 mg/kg) was investigated in vivo in mice bearing NCI‐H1975 xenograft tumors. Pharmacodynamic analyses revealed that the HSP90 client protein EGFR was suppressed for up to 72 h following treatment with AT13387. Indicative of the strength of growth and survival signals in the tumors, Phospho‐S6 and pAkt were also suppressed for up to 72 h (Fig. 3). Similar results for Phospho‐S6 and pAkt were obtained from studies using mice bearing A375 xenografts (data not shown).

Figure 3.

AT13387 in vivo duration of action. Athymic mice bearing NCI‐H1975 xenografts were given intraperitoneal injections of 80 mg/kg AT13387. Control animals were untreated. Samples were taken at the indicated times for immunoblot analyses. AT13387, (2,4‐dihydroxy‐5‐isopropyl‐phenyl)‐[5‐(4‐methyl‐piperazin‐1‐ylmethyl)‐1,3‐dihydro‐isoindol‐2‐yl] methanone, l‐lactic acid salt; CTL, control; EGFR, epidermal growth factor receptor; HSP, heat shock protein; pAkt, phospho‐Akt.

Pharmacokinetic profile of AT13387.

To determine whether the distribution of AT13387 to the xenograft tumor may in part account for its extended duration of action, we investigated both the blood and tumor pharmacokinetic profiles in the same mice used to establish the pharmacodynamic profiles. After intraperitoneal administration of 80 mg/kg AT13387 to NCI‐H1975 xenograft‐bearing mice, AT13387 was rapidly cleared from the blood and could not be detected beyond 24 h. In contrast, high levels of AT13387 were measured in the tumors, and the compound was detectable for at least 240 h after dosing (Fig. 4). In summary, this study demonstrated that, in mice, AT13387 was cleared rapidly from blood and distributed readily to tumors. Tumor retention of AT13387 was markedly extended, which may contribute to extended target inhibition in this target tissue. The extended tumor retention of AT13387 may also mean that less frequent dosing is required in order to maintain efficacy in xenograft models.

Figure 4.

Pharmacokinetic analyses of AT13387. Mean exposure parameters for AT13387 in male athymic Balb/c mice bearing NCI‐H1975 tumors after intraperitoneal administration of 80 mg/kg AT13387. (●) AT13387 concentrations in tumors; (■) AT13387 concentrations in blood. The dashed line represents the limit of detection. AT13387, (2,4‐dihydroxy‐5‐isopropyl‐phenyl)‐[5‐(4‐methyl‐piperazin‐1‐ylmethyl)‐1,3‐dihydro‐isoindol‐2‐yl] methanone, l‐lactic acid salt.

Antitumor efficacy of AT13387 in vivo.

The dose and schedule dependence of the antitumor effects of AT13387 were investigated in NCI‐H1975 xenografts in athymic mice (Fig. 5). When given on an intermittent basis, AT13387 could be tolerated at doses of up to 70 mg/kg twice weekly or 90 mg/kg once weekly. Bodyweight loss in mice did not exceed 20% before recovering in all cases except one, and loss was highest following the second dose. Tumor growth inhibition was similar in NCI‐H1975 for both dosing regimens. The maintenance of antitumor effects with such a prolonged off‐treatment period is consistent with the extended pharmacodynamic action of AT13387 observed for mutant EGFR and other biomarkers in vitro and in vivo and the extended retention of AT13387 in tumors.

Figure 5.

AT13387 is efficacious in multiple xenograft models on a once a week dosing schedule. (A) Athymic mice bearing NCI‐H1975 xenografts were treated for 16 days with intraperitoneal injection of (♦) 17.5% cyclodextrin or AT13387: (■) 70 mg/kg on days 1, 4, 8, 12 and 16; (▲) 55 mg/kg on days 1, 4, 8, 12 and 16; or (♢) 90 mg/kg on days 1, 8 and 16. Eight mice were included in each group. Results shown indicate mean tumor volume (mm³) with error bars representing ± SE. (B) Dosing schedules and body weight changes corresponding to (A). AT13387, (2,4‐dihydroxy‐5‐isopropyl‐phenyl)‐[5‐(4‐methyl‐piperazin‐1‐ylmethyl)‐1,3‐dihydro‐isoindol‐2‐yl] methanone, l‐lactic acid salt.

Discussion

In the course of characterizing AT13387, we observed it to have a long duration of action in tumor cells. In several cell lines, 24‐h treatment in vitro with AT13387 suppressed some client proteins for longer than 7 days. One of the more sensitive cell lines tested was the NCI‐H1975 lung carcinoma cell line, which suppressed mutant EGFR after only being exposed to AT13387 for 7 h. By 7 days post‐exposure, the viability of the remaining cells was very low and the effect could be considered irreversible. In all lines tested to date, the duration of action of AT13387 persisted longer than has been reported for a range of other small molecule HSP90 inhibitors. For example, SNX‐2112 was reported to suppress HER2 for only 24–48 h in BT474 cells in vitro.18 Prolonged pharmacodynamic effects were also noted for AT13387 in vivo when client proteins and phospho‐signaling were monitored in NCI‐H1975 xenograft tumors in mice following a single dose. In these xenograft tumors, the suppression of specific sensitive client proteins lasted up to 72 h. HSP90 client proteins were suppressed for 24–48 h with NVP‐AUY922,30 for only 24 h following treatment with SNX‐5422 (the pro‐drug of SNX‐2112),20 and <24 h following treatment with BIIB021.31 An overall review of the data for different HSP90 inhibitors in clinical development suggests a correlation between the potency of the inhibitors and their duration of action in tumors. The estimated K _d of AT13387 for HSP90 was 0.7 nM, making it the most potent inhibitor at the N‐terminal ATP site reported to date compared with other inhibitors (e.g. SNX‐2112, 30 nM; NVP‐AUY922, 1.7 nM; BIIB021, 1.7 nM; 17‐AAG, 6.7 nM).20, 21, 31 The high potency and extended action of AT13387 correlated with a long tumor retention in xenograft models.

Previous reports have suggested that high‐affinity binding, specifically to tumor‐derived HSP90 protein complexes, might account for tumor‐specific distribution of HSP90 inhibitors.23 However, another study suggested there was no difference in the affinity of 17‐AAG to HSP90 alone or to HSP90 in a multiprotein complex.46 That study found that high‐affinity binding is related to a slow dissociation rate for certain inhibitors and suggested that the observed tumor‐specific accumulation of these compounds may simply be related to increased levels of HSP90 found in tumor cells and the law of mass action.

We explored a number of intermittent dosing schedules in NCI‐H1975 xenografts in mice to investigate the relevance to the antitumor effect of the prolonged pharmacodynamic action associated with AT13387. Based on the observation that the tested client proteins were suppressed for up to 72 h in vivo, we explored twice weekly dosing and found it to be effective in a broad range of different tumor types. We also found AT13387 to be efficacious with once weekly dosing in a lung carcinoma xenograft model. Published data on other agents suggests that only compounds that have pharmacodynamic actions lasting for 48 h or more are able to maintain efficacy when dosed once a week.30

One advantage of exploiting the long tumor retention and extended pharmacodynamic effect of HSP90 inhibitors is less frequent dosing, such that systemic exposure to the agent is reduced. This is particularly apparent for AT13387, where these effects are combined with rapid clearance from the blood in mice. Notable toxicities reported preclinically and clinically for HSP90 inhibitors include liver and gastrointestinal toxicity.17, 27 Thus, the therapeutic index of HSP90 inhibitors in the clinic might be improved by exploiting longer acting agents such as AT13387.

Disclosure Statement

Brent Graham, Jayne Curry, Tomoko Smyth, Lynsey Fazal, Joe Coyle, John Lyons, Nicola G. Wallis and Neil T. Thompson are employees of Astex Pharmaceuticals.

Abbreviations

17‐AAG

17‐N‐allylamino‐17‐demethoxygeldanamycin

ATT13387

(2,4‐dihydroxy‐5‐isopropyl‐phenyl)‐[5‐(4‐methyl‐piperazin‐1‐ylmethyl)‐1,3‐dihydro‐isoindol‐2‐yl] methanone, l‐lactic acid salt

EGFR

epidermal growth factor receptor

HSP

heat shock protein

pAkt

phospho‐Akt