LY573636-sodium (tasisulam) is a small molecule antitumor agent with a novel mechanism of action currently being investigated in a variety of human cancers. In vitro, tasisulam induced apoptosis via the intrinsic pathway, resulting in cytochrome c release and caspase-dependent cell death. Using high content cellular imaging and subpopulation analysis of a wide range of in vitro and in vivo cancer models, tasisulam increased the proportion of cells with 4N DNA content and phospho-histone H3 expression, leading to G2–M accumulation and subsequent apoptosis. Tasisulam also blocked VEGF, epidermal growth factor, and fibroblast growth factor–induced endothelial cell cord formation but did not block acute growth factor receptor signaling (unlike sunitinib, which blocks VEGF-driven angiogenesis at the receptor kinase level) or induce apoptosis in primary endothelial cells. Importantly, in vivo phenocopying of in vitro effects were observed in multiple human tumor xenografts. Tasisulam was as effective as sunitinib at inhibiting neovascularization in a Matrigel plug angiogenesis assay in vivo and also caused reversible, non G2–M–dependent growth arrest in primary endothelial cells. Tasisulam also induced vascular normalization in vivo. Interestingly, the combination of tasisulam and sunitinib significantly delayed growth of the Caki-1 renal cell carcinoma model, whereas neither agent was active alone. These data show that tasisulam has a unique, dual-faceted mechanism of action involving mitotic catastrophe and antiangiogenesis, a phenotype distinct from conventional chemotherapies and published anticancer agents. Mol Cancer Ther; 10(11); 2168–78. ©2011 AACR.
Despite the massive resources expended over the past 20 years and a steadily declining death rate since the 1990s, cancer remains the second leading cause of death in the United States (1, 2). The human genome project has provided hope for a new era of cancer therapeutics by identifying novel molecular targets for drug discovery (3–5). Despite the promise of genomics-centered drug discovery, it is likely that phenotypic-based approaches will continue to play a valuable, complementary role through interrogation of the cell as a target (6–8). Indeed, many of today's standard of care anticancer agents were identified using phenotypic approaches (9).
LY573636-sodium, hereafter referred to as tasisulam, is an acyl-sulfonamide small molecule that is the product of a phenotypic drug screen of more than 14,000 compounds. A requirement of the screen was to identify compounds that had approximately a 100-fold margin of activity between cancerous and nontransformed normal cells. In vitro, tasisulam is a potent antiproliferative compound (free drug EC50 typically in the low μmol/L range) and displays preferential cancer cell death relative to normal, untransformed human cell lines (10, 11). Furthermore, NCI COMPARE analysis indicates that tasisulam possesses a unique mechanism of action relative to other agents (10, 12). In humans, tasisulam has shown activity in several different tumor subtypes, and clinical studies are ongoing or completed in soft tissue sarcoma, ovarian cancer, acute myeloblastic leukemia, metastatic breast cancer, non–small cell lung cancer, and renal cancers (13–15). In hematologic malignancy cell lines, tasisulam induced apoptosis through induction of the caspase-9–dependent intrinsic apoptosis pathway, possibly resulting from a direct effect on mitochondrial processes leading to upregulation of reactive oxygen species (ROS; ref. 16). However, the cellular mechanism(s) underlying these effects were not well understood.
We have expanded the preclinical investigation of tasisulam into a wide variety of solid tumor cell lines, with a particular focus on elucidating the cellular processes underlying the anticancer activity of this compound. Surprisingly, tasisulam disrupted 2 seemingly distinct processes in cancer cell lines, neither of which seems to cause a primary induction of ROS. The first involved induction of the intrinsic apoptosis pathway in a cell-cycle–dependent fashion. Using high content imaging (HCI)-based subpopulation analysis (17, 18), we showed that G2–M accumulation leads to caspase-3 activity and subsequent cell death. The second involved tasisulam-mediated effects on tumor vasculature. Tasisulam displayed both in vitro and in vivo antiangiogenic effects, without inhibiting the proximal growth factor/receptor signaling or inducing endothelial cell death. In addition, tasisulam promoted vascular normalization by stabilizing the existing vasculature, reducing vessel tortuosity and reducing hypoxia in vivo (19). Although identification of the precise cellular target(s) will require additional study, these results suggest that the broad activity of tasisulam in cancer preclinical models is mediated by both cytotoxic and antiangiogenic effects.
Materials and Methods
In vitro cancer cell experimentation
Cells (non-NCI COMPARE panel) were acquired from American Type Culture Collection (ATCC) and cultured according to ATCC guidelines and plated in growth medium for 24 hours before treatment. Cell authenticity was verified by short tandem repeat genotyping at RADIL, no more than 6 months before drug testing (20). Cells were plated at the rate of 500,000 cells/10 mL into 10-cm dishes for flow cytometry, which was done with a Beckman Coulter FC5, and cell cycle was determined using MOD fit.
For Western blotting, nitrocellulose membranes were blocked in TBS, 0.1% Tween-20, and 5% nonfat dry milk, probed for cleaved caspases-3 or -9, PARP/cleaved PARP (Cell Signaling Technology) and then exposed to secondary antibody for 1 hour at 25°C. Proteins were visualized with Super Signal West Femto enhanced chemiluminescence detection (ThermoFisher) and imaged with a Bio-Rad ChemiDoc XRS. For HCI, all procedures are previously described (18). For cell viability, cellular ATP production was measured by Cell Titer Glo Assay (Promega). For assessment of in vitro biochemical proangiogenic receptor tyrosine kinase inhibitory activity, tasisulam was tested at Cerep and Millipore, per industry accepted and validated guidelines.
In vitro endothelial cell experimentation
For assessment of human umbilical vein endothelial cell (HUVEC) migration, HUVECs (45,000 per well) were plated in black 96-well collagen I plates (BD Biosciences) and incubated overnight at 37°C in 5%CO2. Cells were labeled with 3.6 μmol/L CMFDA (Invitrogen) for 30 minutes in serum-free medium, whereupon complete growth medium was replenished and incubated for 30 minutes. A scratch was made in each well using a pintool affixed to a MultiMek robot 96-well head. Cells were then washed using 200 μL normal PBS. Hundred microliter serum-free medium was then added to each well. Cell migration was detected using the Acumen Explorer for day 0 (baseline) data collection. An additional 100 μL 10% serum-containing medium with 2× compound was added into each well after imaging. The plate was cultured overnight and imaged to collect endpoint data. Migration was measured using the difference between endpoint and baseline readings for each well.
Adipose-derived stem cells (ADSC; Lonza) were grown in EGM MV Microvascular Endothelial Cell Growth Medium (Lonza), and endothelial colony forming cells (ECFC; EndGenitor Technologies) were grown in the same medium with an additional 5% FBS (Gibco). Cells were maintained in an incubator at 37°C with 5% CO2.
For HCI, 5,000 ECFCs per well were plated in 200 μL starvation medium (EGM+5% FBS) and cultured overnight. Fresh starvation medium was replenished ± tasisulam for 48 hours. Cells were fixed to the plate with 50 μL concentrated Prefer (Anatech) for 30 minutes at room temperature, washed 1× with PBS, treated for 5 minutes with 100 μL 0.1% SDS and then 100 μL 0.1% Triton X-100 in PBS. Cells were stained with Hoechst 33324 to visualize nuclei, anti-Ki67 (Neomarkers) and visualized with Alexa Fluor-555 conjugated anti-rabbit (Invitrogen) or stained for nuclear fragmentation (TUNEL; In Situ Cell Death Detection Kit TMR Red; Roche) according to manufacturer's protocol.
To assess cord formation, ADSCs were plated in basal medium (MCDB-131 media supplemented with insulin, dexamethasone, ascorbic acid, transferrin, and tobramycin). On day 2, ECFCs were seeded on the ADSCs in basal medium. Following growth factors (VEGF, 20 ng/mL; bFGF, 50 ng/mL; or EGF, 50 ng/mL) and tasisulam addition, cultures were grown for 3 days, fixed to the plate for 30 minutes with cold 70% ethanol, immunostained with anti-human CD31 (R&D Systems)/Alex Fluor-488 donkey anti-sheep IgG secondary (Invitrogen), anti-smooth muscle actin-Cy3 (Sigma) and Hoechst 33342.
Assessment of endothelial cell receptor signaling required a label-free technology (Cellular Dielectric Spectroscopy; CellKey), which is capable of measuring complex impedance changes in cell monolayers. Impedance (Z) is related to the ratio of voltage/current (Z = V/I; Ohm's law). Cells respond to receptor stimulation with rearrangements in the actin cytoskeleton causing shape, adherence, and cell interaction changes, measured as changes in impedance. Recording the response of a cell to ligand stimulation allows specific evaluation of ability of a therapeutic to block the proximal signaling pathways involved in processes such as angiogenesis. A total of 45,000 cells per well are seeded in 150 μL of growth medium onto a microplate containing electrodes at the bottom of each well. After overnight incubation at 37°C in 5% CO2 incubator, the plate is loaded into the instrument where the cells are washed and medium replaced with serum-free medium. A 30-minute equilibration to establish a baseline is followed by compound addition, with impedance monitoring for 30 minutes to record effects. The baseline is reset as ligand is added at 50 ng/mL and the impedance of each well was monitored for 15 additional minutes to record ligand/receptor-induced cellular responses. Maximum change in impedance over time is reported.
In vivo Matrigel plug angiogenesis assay
ECFC and ADSC cells were mixed at the ratio of 4:1 (2 × 106/0.5 × 106), centrifuged, resuspended in Matrigel (BD Biosciences) on ice and subcutaneously injected (0.2 mL/implant) into female athymic nude mice (Harlan), 2 implants per mouse (one/flank). Six days posttreatment mice were sacrificed and implants collected. The right implant was placed into zinc-tris fixative (Pharmingen) for immunohistochemical analysis. The left implant was flash frozen for hemoglobin assay.
For hemoglobin assay, plugs were weighed and placed into homogenization tubes. XY buffer (1% Triton X-100, 25 mmol/L Tris pH 7.5, 150 mmol/L NaCl, 1 mmol/L EDTA, 1 mmol/L EGTA, HALT protease inhibitor (Pierce) added just before use) was added at 4× weight of the plug. Plugs were homogenized for 10 seconds at setting #4 on Fastprep FP120 machine. Samples were centrifuged 5 minutes at 12,000 rpm in 4°C refrigerated centrifuge. Hemoglobin concentration was assessed using QuantiChrom Hemoglobin assay kit (BioAssay Systems) following the manufacturer's protocol.
In vivo pharmacology
Tumor cells were mixed 1:1 with Matrigel and implanted subcutaneously in the right rear flank of athymic nude female mice at 5.0 × 106 cells per injection. Xenografts were grown to an average tumor volume of 200 mm3, and the mice were randomized at baseline according to tumor volume and body weight (n = 8). Physiologic saline and tasisulam doses were administered daily by intravenous injection for 5 days followed by 2 days of rest. This cycle was repeated and tumor growth inhibition was measured.
Tumor volume was estimated using the formula: v = l x w2 × 0.536, in which l = larger of measured diameter and w = smaller of perpendicular diameter. Antitumor activity was calculated as a percent reduction of treated (T) tumor volume relative to untreated control (C) tumor volume [1-(T/C)] × 100, and the data were log transformed to equalize variance across time and treatment groups. The data were then analyzed with a 2-way repeated measures analysis of variance by time and treatment using the MIXED procedures in SAS software (version 8.2). The correlation model for the repeated measures is spatial power. The MIXED procedure is also used separately for each treatment group to calculate adjusted means and SEs. Both analyses account for the autocorrelation within each animal and the loss of data that occurs when animals with large tumors are removed from the study early. The adjusted means and SEs are plotted for each treatment group versus time.
Tumor tissue immunofluorescence and imaging
Five days posttreatment, xenografts were excised and placed into zinc-tris fixative. After 24 hours, tumors were trimmed, routinely processed, embedded in paraffin blocks, and 4 micron sections were made. Slides were baked at 60°F for 1 hour and then deparaffinized in xylene (4 × 10 minutes); rehydrated with ethanol/water immersions with final washes in TBST; blocked with Protein Block (Dako) for 30 minutes; stained with a combination of Hoechst 33324, rat anti-human CD31 (Pharmingen)/anti-rat Alexa Fluor-488 (Invitrogen), rabbit anti-GLUT1 (Dako)/anti-rabbit Alexa Fluor- 647 (Invitrogen), and mouse anti-Smooth Muscle Actin/Cy3 (Sigma); imaged using an iCys Laser Scanning Cytometer (CompuCyte) and a Marianas Digital Imaging Workstation configured with a Zeiss Axiovert 200M inverted fluorescence microscope (Intelligent Imaging Innovations). Mean vessel density (MVD) was calculated as the percentage of total tissue area (Hoechst positive) that is also CD31 positive. Percent pericyte coverage of vessels was calculated as the percentage of total CD31 area that was also colocalized with SMA staining. Percent hypoxia area was calculated as the percentage of total tissue area that is also GLUT1 positve. Mean tortuosity index was calculated as the percent of large vessels/percent of small vessels. Vascular normalization index was calculated as (pericyte coverage of vessels)/(MVD*% hypoxia area*mean tortuosity index) relying on both direct and surrogate markers for these parameters as explained under Results. Quantitative data comparisons of treatment groups were done using the Dunnett's analysis in JMP statistics software (SAS).
Tumor histopathology and immunohistochemistry
Xenografts were formalin fixed, trimmed and embedded in paraffin. Three micron sections were immunohistochemically labeled using an autostainer for phosphohistone H3-serine 10 (pHH3) with a rabbit polyclonal primary antibody (Upstate) at a concentration of 1 μg/mL without antigen retrieval and with 3,3′-diaminobenzidine as chromagen. Sections were counterstained with hematoxylin. Additional sections were stained with hematoxylin and eosin (H&E). Digital images of the immunolabeled slides were captured with the Scan Scope XT (Aperio) and evaluated using the Spectrum software (Aperio) positive pixel count algorithm.
Tasisulam induced apoptosis in a broad range of in vitro cancer cell models
Tasisulam inhibited the growth of a wide range of tumor histologies, with more than 70% of the 120 cell lines tested displaying an antiproliferation EC50 of less than 50 μmol/L (Supplementary Table S1). Cell lines with an EC50 > 50 μmol/L were deemed resistant. Because tasisulam is highly protein bound (≥99.7% bound to serum albumin), the relative free concentration of 50 μmol/L tasisulam in cell culture media supplemented with 10% FBS is approximately 5 μmol/L, a clinically relevant drug concentration. Overall, tasisulam induced an antiproliferative response across a wide range of tumor histologies, suggesting that the compound likely targeted a process fundamental to cell growth. Representative in vitro antiproliferation curves in the Calu-6 non-small cell lung carcinoma (EC50 = 10 μmol/L) and A-375 melanoma models (EC50 = 25 μmol/L) are shown (Fig. 1A).
Tasisulam affected cell cycle
A recent report in hematologic malignancy cell lines observed that induction of ROS occurred 72 hours after tasisulam treatment (16). However, in the Calu-6 cell line, glutathione levels did not change and the addition of an exogenous reducing agent, N-acetylcysteine, did not prevent cell death (Supplementary Fig. S1A/B), suggesting that ROS generation may have been secondary to apoptosis-related events. Fluorescence-activated cell sorting (FACS) analysis of Calu-6 and A-375 cell lines showed that tasisulam induced a concentration-dependent increase in 4N DNA (G2–M accumulation; Fig. 1B, top and middle panels). Calu-6 cells showed striking G2–M accumulation, likely because of asynchronous cells maintaining a relatively high basal 4N DNA population (28%). Western blotting of cleaved PARP and activated caspase-9 confirmed an intrinsic apoptotic response (Fig. 1B, inset), whereas cytosolic cytochrome c levels increased dramatically following treatment (data not shown). HeLa cells, which were largely resistant to the effects of tasisulam (EC50 ∼150 μmol/L), displayed only a modest G2–M accumulation and only slight activation of apoptosis. Interestingly, double thymidine-blocked, synchronized Calu-6 cells treated with tasisulam were significantly delayed in progressing through G2 when drug was introduced immediately upon release from cell-cycle block or during S-phase. If cells were treated after DNA replication, they were able to complete a normal mitosis (Supplementary Fig. S2).
Tasisulam treatment caused apoptosis in cells with increased DNA content
To determine whether tasisulam-induced G2–M accumulation was directly linked to apoptosis, a previously described cell-cycle HCI assay that focuses on cellular subpopulations was used. The HCI assay associates multiple phenotypic readouts within the same cell, as opposed to population-based readouts that cannot associate concomitant biological events at the cellular level (18). Boundaries are established for different cell-cycle states on the basis of nuclear parameters, with an emphasis on DNA content. These boundaries loosely define the classical cell-cycle stages, but given the cell-level resolution of HCI in conjunction with the complexity of cell cycle biology, cell-cycle stages are defined via multiple interacting parameters, rather than demarcation via a single feature. Fluorescent images of tasisulam treated Calu-6 or A-375 cells probed for DNA content, tubulin, and pHH3 showed increased DNA condensation and 3- to 4-fold increase in pHH3 (Fig. 2A), consistent with G2–M accumulation. Similar results were observed in other sensitive, but not resistant, cancer cell lines (data not shown).
HCI subpopulation analysis revealed that only those cells with increased DNA content displayed activated caspase-3, with the extent of induction proportional to the increase in DNA content, nuclear area, and pHH3 (Fig. 2B). Quantitation of G2–M accumulation corroborated FACS data whereby effects of tasisulam were maximal after 48 hours (2 cell doublings), further strengthening the hypothesis that tasisulam induces cancer cell apoptosis through a cell-cycle–related mechanism.
Tasisulam inhibited endothelial cell proliferation in vitro
In the original phenotypic screen, compounds were required to have a significant margin to apoptosis in malignant versus normal cells. Tasisulam caused nonapoptotic growth arrest in HUVECs, prompting further study in other endothelial cell types. ECFCs treated with tasisulam and analyzed in a HCI proliferation and apoptosis assay showed dramatic decreases in the proliferation marker Ki67 at concentrations above 2 μmol/L total tasisulam (∼0.6 μmol/L free drug), similar to the effect observed with the multitargeted receptor tyrosine kinase inhibitor sunitinib (Fig. 3A). However, neither tasisulam nor sunitinib induced endothelial apoptosis, as measured by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL). This was in sharp contrast to the microtubule targeting agent, nocodazole, which was both antiproliferative and proapoptotic (Fig. 3A). The antiproliferative effect of tasisulam was not limited to ECFCs and was also observed with endothelial cells of different origins and primary human fibroblasts (data not shown). Furthermore, in HUVECs, there were no significant changes in cell-cycle distribution as determined by FACS (Supplementary Table S2), indicating the antiproliferative effects were non-cell cycle dependent, suggesting that tasisulam preferentially exerts its cell-cycle/proapoptotic effects on transformed cells.
Tasisulam inhibited in vitro endothelial cord formation and in vivo angiogenesis
Because endothelial proliferation is a process critical for angiogenesis, tasisulam was next assessed for inhibition of in vitro endothelial cord formation, a surrogate assay that models key morphogenic features of blood vessel formation (21). Tasisulam inhibited VEGF-, fibroblast growth factor (FGF)- and epidermal growth factor (EGF)-induced cords in a concentration-dependent manner, with EC50 values of 47, 103, and 34 nmol/L, respectively (Fig. 3B and C). To determine whether tasisulam interfered with acute proximal growth factor receptor–related signaling, a cell impedance assay was used. Changes in impedance as a result of growth factor–induced morphologic changes are a quantitative measure of receptor tyrosine kinase pathway activity (22). VEGF causes a relaxation of the junctions between cells and hence less resistance to electrical conductivity, expressed here as a decrease in impedance. EGF and FGF cause a slight increase in cell-to-cell pressure and hence a slight increase in impedance. Sunitinib, an approved angiogenesis inhibitor, completely blocked the response to VEGF, but not EGF or FGF. Tasisulam did not inhibit VEGF-, FGF-, or EGF-induced impedance changes, indicating that it did not interfere with acute proximal growth factor/receptor signaling (Table 1), consistent with the fact that tasisulam does not inhibit these kinases in vitro (data not shown). In addition, tasisulam is not inhibitory to HUVEC migration in vitro, suggesting that effects on cord formation involve an as yet unexplained mechanism (data not shown).
|.||VEGF .||EGF .||FGF .|
|.||VEGF .||EGF .||FGF .|
NOTE: Endothelial cells were treated with 50 μg/mL ligand and dZiec (extracellular impedance of the monolayer) was recorded every 2 seconds for 15 minutes. Maximum change (ohms) results are shown. VEGF causes a marked relaxation of the junctions between cells and hence less resistance to electrical conductivity, expressed as a decrease in impedance. EGF and FGF cause a slight increase in the cell to cell pressure and hence a slight increase in impedance. The untreated control cells tend to relax slightly during this incubation, resulting in a slightly decreased impedance.
To determine whether tasisulam promotes an antiangiogenic effect in vivo, a Matrigel plug model of neoangiogenesis (23) was done. After 6 days, the coimplanted cells in the Matrigel formed extensive networks of human blood vessels with functional anastomoses to the mouse circulatory system (Fig. 4A). Compared with vehicle-treated control animals, tasisulam at both 25 and 50 mg/kg (corresponding to clinically relevant doses) caused a significant reduction in hemoglobin content (Fig. 4B) and mean blood vessel density by immunohistochemical labeling of Matrigel plugs (Fig. 4C). A similar effect was seen in animals treated with a therapeutically relevant dose of sunitinib (Fig. 4C), indicating that tasisulam impaired in vivo neoangiogenesis.
Tasisulam displayed antitumor efficacy in vivo
The demonstration of in vitro cell-cycle/apoptosis effects on cancer cells and in vitro and in vivo antiangiogenic activities by tasisulam prompted us to explore its activity in tumor xenograft models. Animals received tasisulam at either 25 or 50 mg/kg per day by intravenous injection, 5 days on/2 days off, for 2 weeks, or saline control. Tasisulam displayed dose-dependent Calu-6 xenograft antitumor efficacy, with a maximal reduction in tumor volume relative to control animals of 77% (Fig. 5A; P < 0.001 beginning with day 27), without significant toxicity [determined using standard animal use in scientific research guidelines, including periodic assessment of animal weight and clinical status (for example, no major impact on activity, grooming, no clinical signs of distress that would warrant euthanasia, etc); formal bone marrow assessment in nontumor bearing animals indicated minimal effects at 200 mg/kg]. To confirm cell-cycle effects in vivo, immunohistochemical analysis was done for pHH3 (Fig. 5B). Consistent with the results observed in vitro (Figs. 1B and Fig. 2), tasisulam treatment resulted in G2–M accumulation by pHH3 immunostain (>5-fold), and an increase in mitotic figures in vivo (Fig. 5B). As mentioned above, the strong G2–M accumulation is likely due to the increased number of 4N cells within an untreated, asynchronous Calu-6 cell population. H&E staining of tumor sections also showed increased DNA fragmentation and other apoptosis-associated cellular changes in tasisulam-treated tumors compared with controls (Fig. 5B, bottom panels, arrows). Tasisulam has shown similar antitumor efficacy across a range of in vivo xenografts, including colorectal (HCT-116), melanoma (A-375), gastric (NUGC-3), leukemia (MV-4-11), pancreatic (QGP-1; Supplementary Figs. S3 and S4), and multiple tumor grafts of widely varying histology (data not shown; ref. 24).
To determine whether there was a link between antitumor activity, indicators of G2–M accumulation/apoptosis, and antiangiogenic activity in vivo, multiplexed immunofluorescence tissue imaging of tasisulam-, and control-treated xenografts was done, using a panel of cellular markers of tumor-associated vascularization. These, in turn, were used to assess the degree of vascular normalization, characterized by reduced vessel density, increased pericyte coverage of vessels, decreased vessel tortuosity, and decreased tumor hypoxia (19). Each feature was individually quantitated using image processing algorithms to generate a normalization index (see Materials and Methods). Tumors were labeled with Hoechst, anti-CD31 to visualize endothelial cells/blood vessels, anti-smooth muscle actin to visualize myofibroblasts/pericytes, and anti-GLUT1 as a surrogate marker for tissue hypoxia (25, 26). Although GLUT-1 does not directly measure hypoxia, in many xenograft models (including the Calu-6 model), it is an accurate surrogate of hypoxia. Examination of control- and tasisulam-treated Calu-6 tumors revealed that tasisulam induced changes in several vascular morphologic parameters, consistent with a vascular normalization phenotype. These changes included dose-dependent decreases in vessel density and vessel tortuosity, and trends for an increase in pericyte coverage of vessels, decrease in hypoxia, and an overall increased normalization index (Fig. 5C). Similar phenotypic effects were observed in the A-375 melanoma (data not shown).
Because tasisulam and sunitinib modulated the vasculature through apparently distinct mechanisms (Table 1), the activity of these agents, alone or together, was explored in the human renal cell carcinoma mouse xenograft model Caki-1. An amount of 10 mg/kg sunitinib displayed no tumor growth delay, whereas both 25 mg/kg tasisulam and the tasisulam/sunitinib combination showed a modest effect out to approximately day 50. Beyond day 50, the combination displayed a durable, statistically significant tumor growth delay (minimal P < 0.01 from day 26 onward; Fig. 5D) that persisted well beyond treatment cessation, with reinitiation of tumor growth occurring nearly 70 days postdosing.
Relatively few cancer therapeutics have antitumor activity mediated by both cytotoxic and antiangiogenic effects. Tasisulam induced tumor cell apoptosis as a result of G2–M accumulation, while simultaneously normalizing tumor vasculature. The G2–M accumulation occurred in a wide range of cancer cell types, but not in primary untransformed cells. A more detailed look at the HCI profiles in Calu-6 and A-375 cells suggested that caspase-3 induction begins in S-phase, before the completion of DNA replication (Fig. 2B). Tasisulam is not a direct inhibitor of protein kinases in vitro, and its cell-cycle effects are phenotypically distinct from those produced by cell-cycle inhibitors, such as microtubule poisons (e.g., paclitaxel), G2–M kinase inhibitors (e.g., Aurora or Cdk1), or topoisomerase inhibitors (doxorubicin; ref. 6). Because tasisulam had no effect on DNA synthesis (Fig. 2B and data not shown), it is possible mitotic effects are exerted through interaction with a novel target protein(s) before the G2 checkpoint, a hypothesis supported by the delay in cell-cycle progression noted in double thymidine blocked Calu-6 cells treated with tasisulam immediately after synchronization (Supplementary Fig. S2).
Angiogenesis is a biological process retained by tumors that is critical for tumor growth. Whereas the field of antiangiogenic therapy was established largely on the basis of Folkman's hypothesis to starve tumors of their blood supply (27, 28), the emerging concept of vascular normalization has gained momentum (19). Normalization is a process by which a dense, highly tortuous, and poorly delivering tumor vasculature is converted to one that is more sparse, yet very stable, mature (highly invested in pericytes), organized, and functional (19). Antiangiogenic agents such as sunitinib induce a transient normalization window lasting less than 10 days following the initial dose (29), with a net effect of an improvement in tumor blood flow, resulting in increased delivery of oxygen, nutrients, and theoretically, therapeutics. In the Calu-6 xenograft model, tasisulam induced morphologic features of vascular normalization, including increased pericyte coverage (indicative of vessel maturation) and decreased hypoxia (indicative of improved vessel functionality). Additional studies are required to confirm that tasisulam is capable of maintaining a stable vascular normalization phenotype.
It is unclear whether the proapoptotic and antiangiogenic effects of tasisulam are mediated by a singular target. The broad proapoptotic activity of tasisulam in vitro and the finding in human clinical studies that the dose-limiting toxicity is bone marrow suppression (14) suggest a cellular target shared by hematopoietic stem cells and transformed cell lines. However, preclinical in vivo studies have also shown that an endothelial cell-dependent antiangiogenic response plays an important role in the overall antitumor activity of tasisulam. Furthermore, tasisulam HCI fingerprints were clearly distinct from previously published HCI fingerprints of antimitotic agents that exhibit G2 phase-specific characteristics such as significantly increased nuclear area in conjunction with 4N DNA content and only slightly changed variation in Hoechst DNA staining (6, 7). Altogether, these findings result in a unique phenotypic MOA for tasisulam, characterized by cell-cycle–dependent apoptosis in a broad range of cancer cell lines and antiangiogenic effects in the absence of endothelial cell apoptosis or G2–M accumulation.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
The authors thank Drs. Neote, Shu, and Chen for angiogenesis assay development, Wong for informatics, Starling for scientific guidance, and J. Dempsey for flow cytometry.