WO2007067516A2 - Multiple myeloma - Google Patents

Abstract

The present invention relates, in general, to multiple myeloma and, in particular, to a method of treating multiple myeloma and compounds and compositions suitable for use in such a method. The invention further relates to methods of identifying compounds suitable for use in treating multiple myeloma and to methods of predicting a patient's responsiveness to the instant treatment methods.

Description

MULTIPLE MYELOMA

This application claims priority from U.S. Provisional Application No. 60/742,521, filed December 6, 2005, the entire content of which is incorporated herein by reference. TECHNICAL FIELD

The present invention relates, in general, to multiple myeloma and, in particular, to a method of treating multiple myeloma and compounds and compositions suitable for use in such a method. The invention further relates to methods of identifying compounds suitable for use in treating multiple myeloma and to methods of predicting a patient's responsiveness to the instant treatment methods.

BACKGROUND

Multiple myeloma arises from unchecked division of the plasma B lymphocytes, resulting in a large monoclonal population of plasma cells that typically hyper-secrete monoclonal antibodies or immunoglobin light chains (Sirohi and Powles, Lancet 363:875-887 (2004)). Within the bone marrow, stromal cells (BMSCs) contribute to the survival of these cells through the secretion of vascular endothelial growth factor (VEGF) and interleukin 6 (IL-6) (Hayashi et al, Br. J. Haematol. 120:10-17 (2003)). Autocrine production of cytokines and growth factors also contribute to the overall survival of multiple myeloma cells both within and outside the bone marrow and not surprisingly. overproduction of cytokines, or their cognate receptors, is a key determinant of the sensitivity of cells to chemotherapeutic agents (Catlett-Falcone et al,

Immunity 10(1):105-115 (1999), Novak et al, Blood 103:689-694 (2004)). In addition to playing a role in survival and progression of the disease, production of IL-6 and receptor activator of NF-κB ligand (RANK-L) by myeloma cells is responsible for the increased production of osteoclasts and associated osteolysis observed in patients with advanced disease (Chauhan and Anderson, Apoptosis 8(4):337-343 (2003)). Consequently, targeting of cytokine production and/or disruption of the signaling pathways in which they operate has become the focus of drug discovery efforts in this field.

Glucocorticoids are frontline therapies used for the treatment of all stages of multiple myeloma. Although the precise mechanism(s) by which these steroids manifest their inhibitory activities in vivo is not known, it is likely, based on responses observed in vitro, that they block the production of required survival factors leading to an increase in apoptosis (Schmidt et al, Cell Death Differ.

11:S45-S55 (2004)). Acquired resistance of myeloma cells to glucocorticoid induced apoptosis remains a significant clinical problem and one of the major impediments to long term survival in these patients. Although probably occurring in many different ways, it has been shown that resistance can arise as a consequence of reduced expression of the glucocorticoid receptor (GR), or expression of inactive or dominant negative receptor variants (Krett et al, Cancer Research 55:2727-2729 (1995), Chauhan et al, Oncogene 21:1346-1358 (2002)). However, it is now considered that the more common mechanism of resistance consists of either the hyperexpression of growth factor receptors, such as the IL-6 receptor (IL-6R) a, or dysregulation of the Bcl-2 "rheostat", with altered levels of Bcl-2 or BcI-XL (Schmidt et al, Cell Death Differ. 11:S45-S55 (2004), Chauhan et al, Oncogene 21:1346-1358 (2002)). The mechanisms underlying these epigenetic changes in glucocorticoid treated myeloma cells are poorly understood and thus the search for agents that circumvent resistance has largely been empirical. It is in the framework of this clinical problem that histone deacetylase inhibitors (HDACIs) have emerged as potential therapies for multiple myeloma. It was initially considered that HDACIs would function primarily as facilitators of cellular differentiation. However, their ability to induce apoptosis in a variety of transformed cells, both in vitro and in vivo, has suggested that they may have a broad range of targets and may not function in an equivalent manner in all cells (Duan et al, MoI. Cell. Biol. 25(5): 1608-1619 (2005), Richon et al, J. Cell Biol. 95:3003-3007 (1998), Roy et al, Cell Death Differ. 12(5):482-491 (2005), Takai et al, Cancer 101 :2760-2770 (2004)). In general, HDACIs function, at least in part, by inhibiting class I histone deacetylases, resulting in a

hyperacetylation of histones H3 and H4 with a subsequent local decondensation of chromatin and an enhancement in the expression of specific subsets of genes (Eyal et al, Epilepsia 45(7):737-744 (2004)). However, given the increasing number of proteins whose activities have been shown to be regulated by acetylation, it is likely that whereas histone acetylation may play an important role in some circumstances, it is equally likely that it serves as an indicator for other activities of these enzyme inhibitors in cells (Quivy et al, Biochem.

Pharmacol. 68:1221-1229 (2004), Xu et al, EMBO J. 22(4):893-904 (2003), Yuan et al, Science 307:269-273 (2005)). A further complication in understanding of the mechanism of action of HDACIs emerged from recent work indicating that, whereas this class of drugs wase capable of enhancing nuclear receptor (NR)- mediated transcription (10-fold), most of this activity could be blocked using MAPK inhibitors (Jansen et al, Proc. Natl. Acad. Sci. USA 101(18):7199-7204 (2004)). It was subsequently demonstrated that HDACIs are in fact capable of activating p42/44MAPK, although this activity had no effect on the ability of these compounds to increase acetylation of H3 and H4. (See also D'Anna et al, Biochemistry 19:2656-2671 (1980), Piekarz and Bates, Current Pharmaceutical Design 10:2289-2298 (2004), Acharya et al, Molecular Pharmacology 68(4):917- 932 (2005), Garcia-Manero and Issa, Cancer Investigation 23:635-642 (2005), Bruserud et al, Expert Opinions in Therapeutic Targets 10(l):51-68 (2006), Dokmanovic and Marks, Journal of Cellular Biochemistry 96:293-304), Richon et al, Proc. Natl. Acad. Sci. USA 97(18):10014-10019 (2000), Gui et al, Proc. Natl. Acad. Sci. USA 101(5):1241-1246 (2004), Takai et al, Cancer 101 :2760-2770 (2004), Subramanian et al, Proc. Natl. Acad. Sci. USA 102(13):4842-4847 (2005), Bannister et al, Current Biology 10(8):467-470 (2000), Wang et al,

Journal of Biological Chemistry 279(46):48376-48388 (2004), Blagosklonny et al, Molecular Cancer Therapeutics 1(11):937-941 (2002), Aoyagi and Archer, Trends in Cell Biology 15:565-567 (2005), (Jaansen et al, Proc. Natl. Acad. Sci. USA 101(18):7199-7204 (2004).)

The above-described findings led to the conclusion that MAPK activation and the ability to inhibit histone deacetylases were distinct pharmacological activities of the currently available HDACIs. Given these complex activities, it appeared that currently available HDACIs may not be functionally equivalent in all systems and that optimizing for activity using histone acetylation as an endpoint may or may not yield compounds with the best activity in a given disease.

The present invention results, at least in part, from the realization that it is more appropriate to identify the earliest responses in cells that track with or are responsible for the desired phenotype and to optimize compound selection and clinical trial design using this as a benchmark. Studies described herein focus on defining the mechanism of action of HDACIs in inducing apoptosis in multiple myeloma cells.

In opposing the activity of HDACs, HDACIs encourage the activity of histone acetyltransferase (HAT) enzymes that utilize acetyl CoA to modify lysines of target proteins with an acetyl group. However, acetyl CoA is not solely, or even mainly, a substrate for HATs and is utilized as a substrate for several cellular processes, including entry of acetyl groups into the TCA cycle and formation of long chain fatty acids. Cellular pools of acetyl CoA are tightly maintained, with surplus being shunted into formation of long chain fatty acids and a reduction being rapidly replenished through glycolysis or fatty acid oxidation. HDACI treatment has been shown to result in 2-7 fold increase in acetylation of histones, depending on the cell type analyzed, generally resulting in penta-acetylated H3 (Waterborg, Journal of Biological Chemistry 273(42):27602- 27609 (1998)). Considering the extent to which histone acetylation increases, and that histones are not the sole targets of protein acetylation, HDACI blockage of recycling of acetyl groups may significantly increase the amount of acetyl CoA removed from the cellular pool through protein acetylation. Because acetylation requires use of acetyl CoA as a substrate, studies described herein were undertaken to determine whether HDACI treatment would result in a measurable decrease in cellular acetyl CoA and whether that would contribute to the apoptotic activities of HDACIs.

SUMMARY OF THE INVENTION The present invention relates generally to multiple myeloma. More specifically, the invention relates to a method of treating multiple myeloma and compounds and compositions suitable for use in such a method. The invention further provides a method of determining potential responsiveness of a patient to the treatment strategy described herein. Objects and advantages of the present invention will be clear from the description that follows.

Objects and advantages of the present invention will be clear from the description that follows.

BRIEF DESCRIPTION OF THE DRAWINGS

Figures 1A-1D: Apoptogenic activity of short chain fatty acids in myeloma cells correlates with HDACI activity. Fig. IA) Comparison of molecular structures of compounds used in Figs. IB and 1C (created with

ChemDraw, CambridgeSoft, Cambridge, MA). Fig. IB) RPMI 8226, U266, or OPM2 cells were treated for 96 hrs with MAA (5 mM), VPA (2 mM), VPD (2 mM), sodium butyrate (1 mM), fumarate (5 mM), or TSA (50 nM). After 96 hrs, cells were harvested, stained with 7-AAD and Annexin V, and quantitated using FACS analysis. Data represent mean ±SEM of at least three independent experiments. Fig. 1 C) RPMI 8226 (left panel) or U266 (right panel) cells were treated for 96 hrs with vehicle, MAA (5 mM), or VPA (2 mM) alone or together with cell signaling pathway inhibitors including pertussis toxin (PTX—

200ng/ml), LY294 (25μM), or U0126 (5μM). Data are indicative of three independent experiments. Fig. ID) 0PM2 cells were treated for 24 hrs with the same concentrations of the indicated compounds used in Fig. IB. Whole cell extracts (WCE) were analyzed by Western blot for expression of acetylated histone 3 (upper panel) or GAPDH (lower panel).^• In Fig. IB and Fig. 1C and most other figures, normalized cell survival is calculated by defining the percentage of vehicle-treated cells still viable after the 96-hour treatment

(unstained by either Annexin V or 7-AAD) as equal to 100% and calculating surviving cells in treated samples as percentages relative to vehicle control (normalization more specifically analyzes apoptosis due to drug treatment by omitting the small percentage of apoptotic cells observed in the untreated control).

Figures 2A-2C: Dex and MAA induce apoptosis in multiple myeloma cells independently. Fig. 2A) RPMI 8226 cells were treated with Dex (0.5-50 nM), MAA (1-5 mM), or VPA (0.5-2 mM). After 96 hrs, cell survival was analyzed as in Fig.l . Fig. 2B) RPMI 8226 cells were treated for 96 hrs with Dex (50 nM) and MAA (5 mM) alone or together in the presence or absence of GR antagonist RU486 (500 nM). Cell survival was analyzed as in Fig. 2A. Fig. 2C) RPMI 8226, U266, OPM2, MMl. S, or MMl. R cells were treated for 96 hrs with Dex (50 nM), MAA (5 mM), or VPA (2 mM) prior to harvest and analyzed as in Fig.l. Data represent the mean ±SEM of three or more independent experiments.

Figures 3 A and 3B: HDACI treatment of myeloma cells results in accumulation of cells in Gj/Go prior to induction of apoptosis. Fig. 3A) OPM2 and U266 cells were treated with vehicle, VPA (2 mM), or MAA (5 mM), and cell survival at 24, 48, and 96 hours was analyzed by 7-AAD and Annexin V staining. The percentage of live (unstained) OPM2 (left panel) and U266 (right panel) cells present in the analyzed population is indicated for each treatment and time point. Fig. 3B) OPM2 and U266 cells plated and treated with VPA or MAA in parallel with those analyzed in Fig. 3A were analyzed for cell cycle progression using propidium iodide staining for DNA content. Percentage of live (sum of cells in Gi /G₀, S, or G₂/M phases) OPM2 (left panel) and U266 (right panel) cells present in Gi or Go phases of the cell cycle, as indicated by 2N DNA content, are indicated for each treatment and time point. Results in Fig. 3 A and Fig. 3B are representative of three independent experiments.

Figures 4A-4D: Down-regulation of IL-6 receptor mKNA is observed prior to TRAIL induction. Fig. 4A) OPM2 and U266 cells were treated with VPA (2 mM) for 0, 4, 8, 16, 24, 48, or 96 hours prior to cell lysis. WCE were analyzed by ELISA assay for expression of TRAIL. Graphs indicate the mean ±SEM of three independent experiments. Fig. 4B) WCE harvested in A were analyzed by Western blot for expression of IL-όRα, acetylation of histone 3, and inactive (uncleaved) caspase 3. Extracts were reprobed for GAPDH to analyze equivalent loading. Results are representative of at least three experiments. Fig. 4C) OPM2 and U266 cells were treated with VPA over time as in Fig. 4A. cDNA from treated cells was analyzed by real time qPCR for the relative abundance of the IL- 6Rα mRNA. Relative abundance of each mRNA was calculated using the ΔΔC_T method, in which mRNA levels of each receptor were normalized to 36B4 mRNA levels detected in the same samples, and expression of the IL-6Ro: receptor in untreated cells for each cell line was set equal to one. Relative abundance of the IL-6Rα mRNA in treated cells was then calculated as a ratio relative to 1.

Fig. 4D) U266 cells were treated with 2 niM VPA in the presence or absence of cyclohexamide (CHX - 2 μg/ml). Relative abundance of IL-6Rα mRNA was determined by analyzing cDNA made from treated cells by real time qPCR and was normalized to the abundance of 36B4 mRNA using the ΔΔCT method.

Results illustrated in Fig. 4C and Fig. 4D represent the mean of triplicate wells ±SD and are indicative of three independent experiments.

Figures 5A-5C: Repression of FGFR3 contributes to induction of apoptosis by VPA in OPM2 cells. Fig. 5A) WCE from Fig. 4B were analyzed by Western blot for expression of FGFR3 (upper panel). cDNA from Fig. 4C were re-analyzed by real time qPCR for abundance of FGFR3 mRNA. Results were normalized to abundance of 36B4 mRNA assayed from the same samples, and graphed data represent the mean of triplicate wells ±SD (lower panel). Fig. 5B) OPM2 cells were treated with 2 mM VPA in the presence or absence of cyclohexamide (CHX— 2 μg/ml). Relative abundance of IL-6Rα mRNA was determined by analyzing cDNA made from treated cells by real time qPCR and was normalized to the abundance of 36B4 mRNA using the ΔΔC_T method.

Fig. 5C) OPM2 cells were treated for 24 hrs with VPA (2 mM), VPD (2 mM), MAA (5 mM) or butyrate (1 mM). Relative abundance of mRNAs for FGFR3, IL-6Rα, and BcI-XL were determined by analyzing cDNA made from treated cells by real time qPCR and were normalized to the abundance of 36B4 mRNA detected in the same samples. Results in Fig. 5B and Fig. 5C represent the mean of triplicate wells ±SD. Data in Figs. 5A-5C is indicative of at least three independent experiments. Figures 6A-6C: VPA specifically influences expression of growth factor receptors. Fig. 6A) 0PM2, RPMI₅ and U266 cells were treated for 24 or 48 hrs in the presence or absence of VPA (2 mM). WCE were analyzed for relative expression of BCMA protein (left panel) and mRNA (right panel) by Western blotting or real time qPCR, respectively. Relative abundance of BCMA mRNA was calculated as in Fig. 4. Fig. 6B) OPM2, RPMI, and U266 cells were treated for 24 hrs in the presence or absence of VPA (2 mM). ' Cells were divided for protein (Fig. 6B) or mRNA (Fig. 6C) analysis. Fig. 6B) WCE made from vehicle- or VPA-treated cells was analyzed by Western blot for expression of death receptor 4 (DR4 - TRAIL receptor), FGFR3, vascular endothelial growth factor receptor 3 (VEGFR3), gpl30 (IL-6R/3), IL-6Rα, or acetylated histone 3.

Expression of GAPDH was also analyzed to ensure equivalent loading. Fig. 6C) cDNA made from samples reserved in B for analysis of gene expression at the mRNA. level was analyzed using real time qPCR to determine abundance of mRNAs for granulocyte maturation colony stimulating factor receptor alpha (GMCSFRQ:) or insulin like growth factor 1 receptor alpha (IGF-I Ra). Results represent the mean of triplicate wells ±SD and are representative of three independent experiments.

Figures 7A-7E: VPA induces apoptosis in myeloma cell lines and patient samples with efficacy comparable to that achieved with other alternative myeloma treatments, and synergizes with them to further -increase apoptosis observed in t(4;14) positive cells. Fig. 7A) RPMI 8226, U266, or OPM2 cells were treated with VPA (0.5, 1, or 2 mM), arsenic trioxide (As₂O₃ - 1, 2, or 5 μM), or SAHA (1, 2, or 5 μM). Cell survival was determined by 7-AAD and Annexin V staining and calculated as in Fig. 1. Fig. 7B) RPMI 8226 cells were treated with VPA (0.25 mM) alone or together with As₂O₃ (1 μM), SAHA (1 μM), or Dex (5 nM), and cell survival was determined and calculated as in Fig. 1. Cell treatments are indicated by (+) or (-) below corresponding bars. Striped bars appearing between samples treated with As₂O₃, SAHA, and Dex alone or together with VPA represent the theoretical sum of the independent effects of each treatment, or the anticipated induction of apoptosis if the activities of the co-treated compounds were independent of each other. Fig. 7C) OPM2 cells were treated with VPA (0.25 mM) alone or together with As₂O₃ (1 μM), SAHA (1 μM), Dex (5 nM), MAA (5 mM), or Velcade (5 nM). Samples were analyzed and graphed as in Fig. 7B. Decreased cell survival observed with co-treatment of cells with VPA and other single agents as compared to the theoretical sum of the effects of each treatment alone indicates synergy between the drugs used for co-treatment, and the significance of each synergy was determined using ANOVA analysis followed by Tukey's multiple comparison test. (**) = P < 0.001, (*) = P < 0.05 Fig. 7D) Partially purified plasma cells isolated from a patient sample through tandem Ficoll gradient purifications were incubated 96 hrs in the presence or absence of VPA (2 mM). Cell survival was analyzed by 7-AAD and Annexin V staining followed by FACS analysis, and data graphed indicate the % ±SEM of live cells detected as compared to similarly treated U266 and OPM2 cells. Fig. 7E) Partially purified plasma cells isolated from a second patient sample were treated for 96 hrs with VPA (2 mM), Dex (50 nM), or As₂O₃ (5 μM). Cell survival was analyzed and graphed as in Fig. 7D, and indicates the mean ±SEM of triplicate samples.

Figures 8A-8E: Treatment with HDAC inhibitors results in a detectable decrease in overall cellular levels of acetyl CoA. Fig. 8A) OPM2 cells were incubated with ³H-acetate prior to addition of VPA (2mM) or SAHA (5μM). Following 24 or 48 hours incubation, cells were lysed and samples were deproteinated and counts were measured in soluble and insoluble fractions. Fig. 8B) OPM2 cells were treated 48 hours with the indicated concentrations of VPA, butyrate (NaB), SAHA, or Dexamethasone (Dex). Following analysis of protein concentration, lysates were deproteinated and acetyl carnitines were measured by mass spec-mass spec (MS-MS) analysis. Acetyl carnitine levels were determined by normalization to lysate protein concentration prior to deproteination. OPM2 cells were treated 24 (Fig. 8C) or 48 (Fig. 8D) hours with VPA (0.5, 1, or 2mM) or SAHA (1, 2.5 or 5μM). Acetyl carnitine levels were determined as in Fig. 8A. Fig. 8E) Nuclear extracts from untreated OPM2 cells were incubated with VPA or SAHA at the indicated concentrations prior to analysis of HDAC activity. Results are representative of at least three

independent experiments and graphed values represent calculated mean +/- standard deviation of triplicate samples.

Figures 9A-9G: HDACIs inhibit glucose uptake and reduce GLUTl expression prior to induction of apoptosis. 0PM2 (Fig. 9A) or H929 (Fig. 9B) cells were treated 24 or 48 hours with VPA (ImM), SAHA (1.5μM), or

Doxorubicin (10OnM). 2xlO⁵ live cells (determined by trypan blue staining) were incubated 10 minutes with ³H-2-deoxy-glucose (2-DOG) prior to washing and lysis of the cells. Retained radioactivity was detected by addition of lysates to scintillation fluid prior to analysis. Figs. 9C and 9D) Prior to glucose uptake analysis, a sample of cells from Figs. 9A and 9B were stained with Annexin-V-PE and 7-AAD and analyzed by FACS. Fig. 9E) RNA isolated from OPM2 cells treatd 0-24 hours with 2mM VPA was reverse transcribed prior to analysis by real time quantitative PCR (RTqPCR); detected levels of GLUTl were normalized to similarly detected levels of housekeeping gene 36B4 using the ΔΔC_T method. Fold induction over control was determined by setting GLUTl levels in untreated cells equal to 1. Fig. 9F) Lysates from OPM2 cells treated 24 or 48 hours with ImM VPA were analyzed by Western blotting for expression of GLUTl and loading control GAPDH. Results are representative of at least three independent experiments and graphed values represent calculated mean +/- standard deviation of triplicate samples. Fig. 9G) OPM2 cells were treated 24 hours with ImM VPA, 1.5μM SAHA, or 5OnM PDl 73074 FGFR inhibitor prior to analysis of glucose uptake rate. Treated cells were incubated with 3H-2-DOG for 0-30 minutes before stopping glucose uptake with phloretin. The inhibition of FGFR3 reduces glucose uptake with an efficiency similar to HDACI Retained radioactivity was detected as in Fig. 9A.

Figures 10A-10I: HDACIs influence glucose uptake at the level of GLUTl expression and hexokinase activity. Fig. 10A) OPM2 cells were stably infected with empty retrovirus (OPM2-NGFR) or retrovirus expression FLAG- GLUT 1 (OPM2-GLUT1). Parent and infected OPM2 cells were treated 0, 24, or 44 hours with ImM VPA prior to analysis of glucose uptake as in Fig. 9 A.

Fig. 10B) Lysates of treated cells in Fig. 1 OA were analyzed by Western blotting for GLUTl . Parent OPM2 (Fig. 1 OC) or 0PM2-GLUT1 (Fig. 1 OD) cells were treated 48 hours with ImM VPA or VPD prior to analysis of glucose uptake rate. Treated cells were incubated with ³H-2-DOG for 0-30 minutes before stopping glucose uptake with phloretin. Retained radioactivity was detected as in Fig. 9 A. Figs. 1OE and 10F) OPM2 cells were treated 24 or 44 hours with ImM VPA prior to analysis of glucose uptake rate as in Figs. 1OC and 1OD using either H-2-DOG or ³H-O-m ethyl glucose. Samples were incubated and analyzed as in Figs. 1OC and 10D. Fig. 10G) Hexokinase activity present in lysates of OPM2 cells treated 24 or 44 hours with ImM VPA or 2.5uM SAHA was analyzed and normalized to mg protein input. Fig. 10H) OPM2 cells were treated 0 (control), 24, or 44 hours with VPA (ImM) prior to isolation and reverse transcription of RNA. cDNA was analyzed by RTqPCR and detected levels of HXKl were normalized and calculated as in Fig. 9E. Fig. 101) Hexokinase 1 and GAPDH expression in lysates of OPM2 cells treated 24 or 44 hours with ImM VPA or 2.5μM SAHA was analyzed by Western blotting. Results are representative of at least three independent experiments and graphed values represent calculated mean +/- standard deviation of triplicate samples. Figures 1 IA-I IH: Reduction of glucose uptake by HDACI causes cells to utilize long chain fatty acids. OPM2 cells were treated 24 (Figs. 1 IA, 1 IC₅ 1 IE and 1 IG) or 48 hours (Figs. HB, HD, 1 IF and 1 IH) with VPA (0.5, 1, or 2mM) or SAHA (1, 2.5 or 5μM). Following analysis of protein concentration, lysates were deproteinated and acyl carnitines were measured by MS-MS analysis. Acyl carnitine levels were determined by normalization to original lysate protein concentration.

Figures 12A-12H: HDACI treatment results in metabolism of amino acids. 0PM2 cells were treated 24 (Figs. 12A, 12C and 12E) or 48 (Figs. 12B, 12D or 12F) hours with VPA (0.5, 1, or 2mM), SAHA (1, 2.5 or 5μM) or

Doxorubicin (25, 50, or 10OnM). Amino acid levels present in whole cell extracts were determined using MS-MS analysis of deproteinated lysates and normalized to protein concentrations in the original lysates. Figs. 12G and 12H) Prior to cell lysis and analysis of amino acids, a sample of cells from Figs. 12 A, 12C and 12E (shown in Fig. 12G) or Figs. 12B, 12D, and 12F (shown in Fig. 12H) was stained with Annexin-V-PE and 7AAD and analyzed by FACS to determine cell viability.

Figures 13A-13H: Metabolism of amino acids, and disposal of released amino groups, contributes to HDACI induction of apoptosis. Fig. 13A) 0PM2 cells were treated 96 hours with vehicle or VPA (0.75mM) in the presence of absence of supplemental non-essential amino acids (2mM). Cells were stained with Annexin-PE and 7-AAD and apoptosis was analyzed by FACS analysis. Indicated percentages represent the average remaining live (unstained) cells 47- SD. Fig. 13B) OPM2 cells were treated 0 (control), 24, or 48 hours with VPA (0.5, 1, or 2mM) prior to isolation and reverse transcription of RNA. cDNA was analyzed by RTqPCR and detected levels of CPSl were normalized and calculated as in Fig. 9E. Ornithine (Fig. 13C), citrulline (Fig. 13E), and arginine (Fig. 13G) levels were determined in OPM2 cells treated 48 hours with VPA (0.5, 1, or 2mM) or SAHA (1, 2.5 or 5μM) and analyzed as described for amino acids in Fig. 12A. Fig. 13D) Spent media from OPM2 cells treated 44 hours with VPA (ImM) or SAHA (2.5μM), as well as a fresh media control, was analyzed using a urea detection kit. Urea present in samples was calculated using a standard curve. (Fig. 13F) 0PM2 cells were treated 48 hours with VPA (0.5, 1, or 2mM) prior to lysis and analysis of protein concentration. Lysates were deproteinated and alpha- ketoglutarate content was analyzed through MS-MS and compared to internal standards. Fig. 13H) OPM2 cells were pre-treated 5 hours with DFMO (5mM) or vehicle prior to treatment for 96 hours with vehicle or VPA (0.75mM) in the presence or absence of DFMO (5mM). Cell survival was analyzed as in Fig. 13 A. Results are representative of at least three independent experiments and graphed values represent calculated mean +/- standard deviation of triplicate samples. Fig 131) 0PM2 cells were treated 48 hours with vehicle (control), 2mM VPA, or 5uM SAHA. After the initial 24 hours incubation, cells were isolated, washed, and saturated with ROS sensitive dye CM-H2DCFDA. Cells were washed following dye loading and retreated in the original treatment media for an additional 24 hours. ROS production was measured through increased fluorescence as detected by FACS. DETAILED DESCRIPTION OF THE INVENTION

The present invention relates to a method of treating multiple myeloma. The method comprises administering to a mammal (human or non-human) in need of such therapy a histone deacetylase (HDAC) inhibitor in an amount sufficient to effect the therapy. The invention includes methods of treating multiple myeloma in mammals (e.g., humans) who has become refractory to other forms of treatment (including corticosteroid therapy). The HDAC inhibitors of the invention (e.g., short chain fatty acids) can also be used as a first-line therapy alone or in combination with a corticosteroid-based treatment regimen. Short chain fatty acid HDAC inhibitors can also be used alone or in combination with, for example, AS₂O₃, VeI cade, thalidomide, CPTl inhibitors, IL-ό receptor antibodies, FGF receptor tyrosine kinase inhibitors or SAHA, in treating myelomas refractory to corticosteroid therapy.

HDAC inhibitors appropriate for use in the invention include short chain fatty acids, as well as chemically distinct compounds such as SAHA or tricostatin A (TSA). Short chain fatty acids suitable for use include C₃-C12 fatty acids, preferably C₃-C₁₀, more preferably C₃-C₈, for example, methoxyacetic acid (MAA), butyric acid (BA), valproic acid (VPA), propionic acid, 3- methoxypropionic acid and ethoxyacetic acid, or pharmaceutically acceptable salts thereof. Also suitable for use are precursors of short chain fatty acids

(including those described above), or pharmaceutically acceptable salts thereof, ethylene glycol monomethyl ether being one example of a suitable precursor. Combinations of short chain fatty acids (or precursors or salts thereof) can be used, e.g., to lower the required doses. Preferred combinations comprise MAA, VPA and MAA being an example of such a combination.

The HDAC inhibitors of the invention, including the short chain fatty acids, can be administered alone or in combination with other chemotherapeutic agents suitable for use in treating multiple myeloma. For example, HDAC inhibitors (e.g., short chain fatty acids) can be used before, during or after the administration of chemotherapeutic agents including but not limited to arsenic compounds, such as arsenic trioxide or melarsoprol or arsenic sulfides (see, for example, U.S. Appln. 20040146583 and USP 6,733,792) and ATRA. In a specific embodiment, a short chain fatty acid, e.g., VPA, and Velcade can be administered in combination.

HDACIs of the invention can also be administered in combination with one or more inhibitors of polyamine synthesis (e.g., DFMO, methylglyoxal. bis(cyclopentylamidinohydrazone) (MGBCP), SAM486A (CGP48664), methylglyoxal-bis(guanylhydrazone) (methyl GAG), and polyamine analogues (e.g., BE4-4-4-4 and BEPUT)), promoters of reactive oxygen species (e.g., DFMO), inhibitors of glycolysis (e.g., 2-deoxyglucose, mannose, and hexokinase inhibitors (e.g., gluocsamine, 3-bromopyruvate and sorbose-1 -phosphate), urea cycle inhibitors, including arginase inhibitors (e.g., nor-arginine, 2(S)-amino-6- boronohexanoic acid (ABH), N(omega)-hydroxy-nor-]-arginine (nor-NOHA), and l-norvaline5), inhibitors of IL-6 signaling (e.g., antibodies to IL-6 receptor or to IL-6), fat absorption uptake inhibitors (e.g., orlistat, etc.), anti -retroviral protease inhibitors (e.g., nelfinavir, ritonavir, amprenavir, lopinavir) as they can inhibit fatty acid uptake by cells, and/or inhibitors of carbamoylphosphate synthetase (e.g., 5'_/?fiuorosulfonylbenzoyladenosine (FSBA)).

Preferred combinations of agents suitable for use in the present invention comprise MAA and VPA, MAA and SAHA, VPA and DFMO and MAA and DFMO.

It will be appreciated from the data presented in Example 2 that HDACI treatment of a multiple myeloma patient can be enhanced by placing the patient on a diet high in protein and low in sugar (Sinha et al, Neurologist 11:161-170 (2005)). Any suitable mode of administration can be used in accordance with the present invention including but not limited to parenteral administration, such as intravenous, subcutaneous, intramuscular and intrathecal administration, oral, and intranasal administration, and inhalation. The mode of administration can vary, for example, with the condition of the patient.

The invention includes pharmaceutical compositions comprising one or more HDAC inhibitor (e.g., short chain fatty acid) and a carrier. The

compositions can be, for example, in the form of a sterile aqueous or organic solution or a colloidal suspension. The composition can also be in dosage unit form, for example, as a tablet or capsule. The compositions can comprise additional active agents, such as a corticosteroid (e.g., dexamethasone) or a chemotherapeutic agent, or otherwise, as noted above.

The invention also relates to kits suitable for use in practicing the method of the invention. Such kits can comprise in one or more container means therapeutically effective amounts of one or more HDAC inhibitor (e.g., short chain fatty acid) in pharmaceutically acceptable form. The kit can also comprise an additional chemotherapeutic agent, or other active agent described above, in pharmaceutically acceptable form. The kit can further comprise a needle and/or syringe.

The optimal therapeutic dose of an HDAC inhibitor (e.g., short chain fatty acid) can vary, for example, with the HDAC inhibitor, the patient and the effect sought and can be readily determined by one skilled in the art. For example, a daily dose of short chain fatty acid can be from about 0.1 to about 150 mg per kg body weight per day (e.g., parenterally or orally). A preferred daily dose can be from about 1 to about 100 mg/kg body weight of short chain fatty acid, more preferably, from about 10 to about 20 mg/kg/day. Again, any suitable route of administration can be employed for providing the mammal with an effective dosage of the HDAC inhibitor. For example, oral, transdermal, iontophoretic, parenteral (e.g., subcutaneous, intramuscular, and intrathecal) can be employed. Dosage unit forms include tablets, troches, cachet, dispersions, suspensions, solutions, capsules and patches. (See, for example, Remington's Pharmaceutical Sciences.)

The present invention also includes methods of predicting a patient's responsiveness to the instant treatment methods. A t(4;14) chromosomal translocation is found in 10-20% of myelomas (Rasmussen et al, Br. J. Haematol. 117:626-628 (2002)). This translocation leads to aberrant expression of a constitutively active form of FGFR3. As described in the Examples that follow, myelomas cells possessing this translocation are particularly sensitive to HDAC inhibitors (e.g., short chain fatty acids such as VPA). Accordingly, the invention includes a method comprising obtaining a blood or bone marrow sample from a patient (e.g., a patient known to have multiple myeloma or a patient suspected of having multiple myeloma) and assaying DNA present in that sample for the presence of the t(4;14) chromosomal translocation (e.g., using art-recognized techniques). Presence of the translocation indicates that the patient is more likely than not to be responsive to treatment comprising administration of the HDAC inhibitors (e.g., short chain fatty acids such as VPA) described above.

In addition to the above, the invention also includes a method of determining a therapeutically effective dose of HDAC inhibitor. In accordance with this method, a patient known to have multiple myeloma or a patient suspected of having multiple myeloma is treated with a range of doses of HDAC inhibitor(s) (e.g., 1-100 mg/kg) and blood samples from that patent are analyzed for the level of the abberant form of FGFR3 at each dose (e.g., by analyzing for the protein or the mRNA using, for example, art-recognized techniques). A therapeutically effective dose is a dose that is found to effect down-regulation of the aberrant form of FGFR3 (that is, reduces production of the aberrant form of FGFR3 relative to a control). Compounds (e.g., short chain fatty acids) suitable for use in treating multiple myeloma can be identified by assaying candidate compounds for their the ability to inhibit HDAC and, more specifically, to desensitize signaling systems required for cell proliferation and survival. Candidate compounds can be screened for their ability to regulate (e.g., inhibit) expression of IL-6Rα, FGFR3 (in the context of the t(4; 14) translocation) and/or BCMA. Appropriate screening methods include those described in the Example that follows.

Certain aspects of the invention are described in greater detail in the non- limiting Examples that follows. EXAMPLE 1

EXPERIMENTAL DETAILS

Cell lines. RPMI 8226 and U266 cells were purchased from ATCC (Monassas, VA). OPM2 cells were generously provided by E. Brad Thompson, Baylor College of Medicine, Houston, TX. MMl .S and MMl .R were kind gifts from Steven T. Rosen, Northwestern University, Chicago, IL. Cells were maintained in modified RPMI 1640 (ATCC) supplemented with 8% (RPMI 8226, MMl. S and MMl. R) or 15% (U266 and OPM2) FBS. Analysis of patient isolates were done with bone marrow aspirates twice subjected to Ficoll gradient separation to isolate a reasonably pure population of plasma cells. All cells were grown in a humidified incubator maintained at 37°C and 5% CO2.

Reagents. Compounds utilized included valproate (VPA), methoxy-acetic acid (MAA), butyrate, trichostatin A (TSA), and fumarate, all of these ordered from Sigma Aldrich (St. Louis, MO) as sodium salts and dissolved in water.

Valpromide (VPD - Lancaster Synthesis, Pelham, NH) was dissolved in 100% ethanol. Other compounds utilized included SAHA (suberoylanilide hydroxamic acid - Merck, Whitehouse Station, NJ), arsenic trioxide (As₂O₃ - Sigma), and dexamethasone (Dex - Sigma). Velcade (Millenium Pharmaceuticals,

Cambridge, MA— formerly PS-341) was obtained as a generous donation for research purposes from the manufacturer. Cyclohexamide (Sigma) was provided as a lOOmg/ml stock solution. All compounds were diluted in culture media immediately prior to use.

Cell viability and apoptosis. IxIO⁵ cells were treated for 24-96 hrs with the indicated compounds in ImI total volume of RPMI 1640 media supplemented with 12% FBS. Cells were harvested by centrifugation (500xg for 5 min), washed twice in PBS₅ and stained with PE-conjugated Annexin V and 7-AAD per manufacturer's instructions (Pharmingen, San Diego, CA) prior to FACS analysis. CeZ/ cycle progression. 1.25x10⁶ cells were incubated for 0, 24, or 48 hrs in 5mls RPMI media containing indicated treatments. Cells were harvested by centrifugation, washed in PBS, and fixed in 70% cold ethanoL Cell samples were incubated lhr at 4°C and then stored at -20⁰C until staining. Fixed cells were pelleted and resuspended at 2.5x10⁶ cells/ml in Pl staining solution [20 μg/ml PI (Sigma), 50 μg/ml RNase A diluted in PBS]. After a 30 min 37°C incubation, samples were stored at 4°C until FACS analysis. Cells were staged by calculating a relative percentage of cells in Gi _/o, S, or G₂ by defining the sum of cells in those three stages as 100% of the viable cell population. Western blotting. 1-3x10⁶ cells were incubated in RPMI media containing the indicated treatments, harvested by centrifugation, washed twice in PBS supplemented with 3% FBS, and resuspended in Lysis Buffer [50 mM Tris (pH 8), 100 mM NaCl₅ 1.5 mM MgCl₂, 1% Triton X-100, 1 mM EGTA₅ 10% glycerol, 50 mM NaF, 2 rnM Na₃VO₄, IX protease inhibitor cocktail

(Calbiochem, La Jolla, CA)]. Following 30 min incubation at 4°C, WCE were cleared by centrifugation (14,000xg for 5 min) and analyzed by Bradford assay. Equal amounts of WCE were resolved by SDS-PAGE; following transfer, membranes were blocked with BSA or milk, and proteins were detected by Western blotting per antibody manufacturers' instructions using enhanced chemiluminescence detection (Amersham, Piscataway, NJ). Antibodies to IL- 6Rα (sc-661), FGFR3 (sc-13121), DR4 (sc-7863), VEGFR3 (Flt4, sc-321), GAPDH (sc-20357), and gpl30 (sc-656) were obtained from Santa Cruz

Biotechnologies (Santa Cruz, CA). Antibodies to caspase 3, acetylated histone 3 and BCMA were obtained from Cell Signaling Technology (Beverly, MA), Upstate (Lake Placid, NY) and Abeam (Cambridge, MA), respectively.

ELISA assay. TRAIL expression was detected by ELlSA assay per manufacturer's instructions (Biomol, Plymouth Meeting, PA). Briefly, known dilutions of a purified TRAIL standard or 25 μ.g of WCE (lysis procedure detailed above) were incubated in prepared wells pre-coated with an antibody to TRAIL. Following washing and addition of biotinylated antibody to TRAIL and streptavidin-HRP (reagents provided), chromogen solution was added for quantitative detection as analyzed by spectrophotometry. TRAIL expression per mg WCE was calculated from the linear regression of the standards.

Real Time quantitative PCR. To analyze the effect of compounds on mRNA expression, 1-2x10⁶ cells were plated at 0.3-0.5x10⁵ cells/ml in RPMI media containing indicated treatments. Following incubation for 0-24 hrs, cells were harvested by centrifugation and washed in PBS prior to lysis. RNA isolation (RNeasy - Qiagen, Valencia, CA) and reverse transcription (iScript— Biorad, Hercules, CA) were performed per kit manufacturer's instructions. qPCR of cDNA was done using iQ SYBR Green supermix (Bio-Rad) per kit instructions, and amplification was performed using the iCycler optical system with associated software (Bio-Rad). mRNA abundance was calculated using the ΔΔC_T method as previously described (Livak and Schmittgen, Methods 25:402-408 (2001)).

5 Primers included: IL-όRα, forward (F) - GCTCCTCTGCATTGCCATTG,

reverse (R) - CATCTGGTCGGTTGTGGCT; FGFR3, F - .

GCCTGGTCATGGAAAGCGT, R- CGGATGCTGCCAAACTTGTT (Soverini et al, Haematologia (Budap) 87:1036-1040 (2002)); GMCSFR, F - TGCTCTGTGAGTTACCACACC, R- GGCAGTCCCAGCTTAAATTCAT; i o BCMA, F - TTTCTTTGGCAGTTTTCGTG, R - GATGCAGTCTTCACAGGTGC; IGF-lRα, F - AGGATATTGGGCTTTACAACCTG, R- GGCTTATTCCCCACAATGTAGTT; and 36B4, F - GGACATGTTGCTGGCCAATAA, R - GGGCCCGAGACCAGTGTT.

15 Detection of reactive oxygen species. 3x105 0PM2 cells were treated with the indicated compounds in a ImI volume in phenol red free (PRF) RPMI media supplemented with 12.5% serum. Following 24 hours of treatment, the cells were isolated through centrifugation (treatment media was saved), washed in PRF serum free (SF) RPMI media, and incubated 30 minutes in PRF SF RPMI

2 o containing 5uM CM-H2DCFDA (5,6-chloromethyl-2\7'- dichlorodihydrofluorescein diacetate, acetyl ester - Invitrogen). Cells were washed in PRF SF RPMI and replated in the original treatment media. Following an additional 24 hours incubation, cells were collected and washed in PBS+1% BSA prior to analysis by FACS.

25 RESULTS

Evaluation of the activity of different chemical classes of HDACIs in- cellular models of Multiple Myeloma.

For the initial characterization of HDACI activities in multiple myeloma,

RPMI 8226, U266, and OPM2 cells were treated with the short chain fatty acid derived HDACIs methoxy acetic acid (MAA), valproic acid (VPA), or butyrate, as well as with the chemically distinct HDACI tricostatin A (TSA). Valpromide (VPD), a derivative of VPA lacking HDACI activity, was also included for comparative purposes (Fig. IA). Fumarate was used as a comparison because it lacks both HDACI activity and the extended carbon chains of these other compounds. These particular cellular models were chosen as they represent different stages of the disease and/or have different underlying abnormalities. RPMI 8226 cells undergo apoptosis in response to treatment with corticosteroids and thus are representative of the initial glucocorticoid-naive phase of the disease ^" (Genty et al, Leuk. Res. 28:307-313 (2004)). U266 cells are glucocorticoid resistant likely as a consequence of their ability to overexpress both the IL-6Rα receptor and soluble IL-6 (Schwab et al, Blood 77:587-593 (1991)). OPM2 were chosen as they are partially responsive to glucocorticoids and harbor a t(4;14) chromosomal translocation that leads to the ectopic expression of fibroblast growth factor receptor (FGFR) 3 (Ronchetti et al, Oncogene 20:3553-3562 (2001)). This translocation is found in 10-20% of myelomas and thus OPM2 cells model a significant subtype of this disease (Rasmussen et al, Br. J. Haematol. 117:626-628 (2002)). The doses of compounds used for this analysis were chosen based upon either the IC₅₀ determined in vitro using purified HDACs and hi stones as substrates (not shown) or, for those compounds that have been used in humans, clinically relevant doses were chosen. As demonstrated in Fig. IB, MAA (5 mM), VPA (2 mM), and butyrate (1 mM) induced apoptosis to varying degrees in all three cell lines with activity greater than or comparable to that of TSA (25 nM). Cell survival was assessed using Annexin V and 7-AAD staining and quantitated using FACS analysis. VPD (2 mM) had no significant effect on myeloma cell survival, indicating that the acidic nature of VPA is required for its apoptogenic activity. Fumarate (5 mM), acidic in nature but without the extended carbon chains of VPA, also did not significantly affect myeloma cell survival. In all of these cell lines, at pharmacologically relevant doses, VPA proved to induce apoptosis as effectively as the benchmark HDACI butyrate and better than TSA.

An evaluation was next made as to whether the ability of the chosen HDACIs to induce apoptosis in myeloma cells was due to their ability to inhibit histone deacetylation or to their activities on MAPK or other signaling pathways. It was shown in RPMI 8226 cells that whereas inhibition of PKC or MAPK signaling reduced the viability of these cells, it did not impact MAA induced apoptosis (Fig. 1C). Similarly, in U266 cells, these cell signaling inhibitors did not influence the apoptogenic actions of either MAA or VPA. It appears more likely that the cytotoxic activity of these compounds correlates with their activity as HDACIs. The effect of these HDACIs on the acetylation of histone 3 (H3) in OPM2 cells, analyzed by Western blotting, is presented for illustrative purposes. As shown in Fig. ID, treatment with MAA, VPA, butyrate, or TSA resulted in a significant enhancement of acetylated H3 (upper panel) over vehicle treated cells, while Western blotting of GAPDH indicated that comparable amounts of protein from each sample were analyzed (lower panel). As with the apoptosis assays, VPA proved to be more effective than MAA and comparable with butyrate in this acetylation assay. Just as neither VPD nor fumarate significantly influenced myeloma cell survival, neither of these compounds had an appreciable effect on the acetylation of H3. These data suggest that the cytotoxic effects of MAA, VPA, butyrate and TSA correlate with and are mediated, at least in part, through inhibition of HDAC activity. Because VPA is already approved for the treatment of seizures and anxiety, has good bioavailability in the human body and a long half-life in serum (6-8 hours), as opposed to butyrate (~6 minutes), the majority of the remaining experiments focused on the activities of this HDACI. VPA and MAA induce apoptosis in myeloma cells independently of their sensitivity to glucocorticoids.

Because of their sensitivity to glucocorticoids, RPMI 8226 cells were chosen to compare the apoptogenic activities of the synthetic glucocorticoid receptor (GR) agonist dexamethasone (Dex), and the HDACIs MAA and VPA. For these studies, RPMI 8226 cells were treated with Dex (0.5 - 50 nM), MAA (1 - 5 mM), or VPA (0.5 - 2 mM), and cell survival was measured as above. As shown in Fig. 2A, both VPA and MAA effectively induced apoptosis in a manner that was comparable to or slightly better than Dex. Since the SCFA-derived HDACIs have been demonstrated to significantly enhance nuclear receptor transcriptional activity, it was important to determine if their apoptogenic activity in RPMI cells was in any way influenced by GR status. To address this issue, cells were treated with Dex (50 nM) or MAA (5 mM) alone or together in the presence or absence of the GR antagonist RU486 (500 nM). As observed in Fig. 2B, RU486 had no significant effect in and of itself on cell survival, but completely reversed the apoptogenic activity of Dex. In contrast, co-treatment of cells with MAA and RU486, irrespective of the presence or absence of Dex, did not reverse MAA-induced apoptosis, indicating that the apoptogenic effects of MAA are independent of GR. . .

Given that the apoptogenic actions of MAA were independent of GR, an analysis was next made of the ability of MAA or VPA to induce cell death in a panel of myeloma cell lines with varying sensitivity to Dex. In Fig. 2C, GC sensitive (RPMI 8226, OPM2, and MMLS) and resistant (U266 and MMl. R) cells were treated with Dex (50 nM), MAA (5 mM), or VPA (2 mM) and analyzed for apoptotic response. Despite their disparate sensitivity to Dex

(ranging from 0-98% apoptosis), all of the myeloma cells were determined to be sensitive to MAA and VPA resulting in differing rates of increased cell death with no relationship between the degrees of sensitivity of each cell line to Dex or HDACI. These data demonstrate that the mechanism(s) that allow myeloma cells to escape the cytotoxic effects of GCs do not influence myeloma cell sensitivity to SCFA-derived HDACIs.

VPA and MAA elicit a biphasic response in myeloma cells with initial cell cycle arrest followed by apoptosis.

HDACIs have been shown to arrest cell cycle progression as well as to induce apoptosis in hematopoietic tumor cells (Richon et al, Proc. Natl. Acad. Sci. USA 93:5705-5708 (1996), Sakajiri et al, Exp. Hematol. 33:53-61 (2005)). Thus, the question raised was whether the less characterized HDACIs MAA and VPA would similarly arrest cell proliferation. For this analysis, OPM2 and U266 cells were treated with VPA (2 mM) or MAA (5 mM) for 24, 48 or 96 hours, and cells were harvested at each time point and analyzed for cell survival and staged. As demonstrated in Fig. 3 A, neither 0PM2 (left panel) nor U266 (right panel) cells treated with MAA or VPA display a significant difference from the untreated control after 24 hours treatment with respect to the percentage of live (unstained) cells present (as detected by 7- AAD and Annexin V staining), indicating that significant apoptosis has not yet occurred. Similarly, VPA and MAA treated 0PM2 (Fig. 3B₃ left panel) and U266 (right panel) cells do not show a significant change in the percentage of live cells (DNA content between 2N and 4N) present in Gi/Go (2N) phase of the cell cycle at 24 hours. At 48 hours, a 25% decrease in live cells (Fig. 3 A, left panel) is detected for OPM2 cells treated with VPA, while the corresponding MAA treated sample does not differ significantly from the untreated control. However, cell cycle analysis indicates a 15-20% increase in OPM2 cells present in the Gi /Go phase of the cell cycle for both VPA and MAA treated cells (Fig. 3B, left panel). Similarly, at 48 hours of treatment, a 20-30% decrease in live U266 cells was observed following treatment with VPA or MAA, as compared to the untreated control (Fig. 3A, right panel). Likewise a ~ 20% increase in the percentage of live U266 cells in the Gi /Go phase of the cell cycle was observed (Fig. 3B₃ right panel). These data indicate that of the reduced live cell population detected at 48 hrs in Fig. 3A, a greater percentage of the cells are present in the Gi/Go phase of the cell cycle (Fig. 3B), demonstrating that MAA and VPA can induce cell cycle arrest.

VPA down-regulates expression ofIL-6 receptor prior to induction of apoptosis.

The mechanism by which HDACIs induce apoptosis has not yet been clearly defined although it has been reported that some of these compounds can induce the expression of p21, explaining possibly their antiproliferative activities (Lavelle et al, Am. J. Hematol. 68:170-178 (2001), Mitsiades et al, Proc. Natl. Acad. Sci. USA 101(2):540-545 (2004)). Recently, Nebbioso et al. described that HDACIs also induce the expression of the death receptor ligand TRAIL in acute myeloid leukemia (AML) cells, resulting in an autocrine signaling loop through the death receptor 4 (DR4) that culminates in apoptosis (Nebbioso et al, Nature Medicine 1 l(l):77-84 (2005)). Thus, the potential contribution of these pathways to HDACI action was next evaluated in multiple myeloma cells. Firstly, the timing of apoptosis, growth arrest, and induction of TRAIL in myeloma cells in response to VPA were evaluated. To this end, U266 and 0PM2 cell lines were treated with VPA (2 mM) (or MAA₃ not shown) over an extended time course (4- 96 hrs) and analyzed the expression of TRAIL was analyzed by ELISA assay. Notably, although quantitative apoptosis is observed in these cells by 48 hours, it is significant that induction of TRAIL was not detectable until 48 hours following treatment in the OPM2 cells or 96 hours in the U266 cells (Fig. 4A). This suggests that although TRAIL may contribute to the apoptogenic activities of MAA and VPA, its expression appears to be too late to be the primary regulator of cell growth arrest and apoptosis.

The ability of HDACIs to inhibit the growth of myeloma cells (Fig. 3) led to the speculatation that these compounds may have a negative influence on one of the survival pathways shown previously to be important for the survival of these cells. Upregulation of the IL-6 receptor a (IL-6R.O! - gp80), has been described as one mechanism by which myeloma cells can escape the apoptogenic effects of Dex (Chauhan and Anderson, Apoptosis 8(4):337-343 (2003)). Since both U266 and OPM2 cells overexpress IL-6Rα, the impact of HDACIs on the expression of this receptor was evaluated. Specifically, OPM2 and U266 cells were treated for 4-96 hrs with VPA (2 mM), and the expression of the IL-όRα, acetylation of H3, or cleavage and activation of caspase 3, a hallmark of apoptosis, were assessed by Western immunoblot of cell extracts. Increased acetylation of histone 3 was detected as early as 4 hours following VPA addition (Fig. 4B). Similarly, the expression of IL-όRo: was visibly reduced following just 8 hours of treatment with VPA reaching a minimum at 16 hours (Fig. 4B). In contrast, activation of caspase 3 correlated temporally with the induction of TRAIL expression observed in Fig. 4A, at 48 hours in OPM2 cells and 96 hours in U266 cells, as detected by cleavage of the apo-caspase 3 to the activated fragment (Fig. 4B). These data demonstrate that treatment with VPA results in a blockade of survival factor IL-6 signaling prior to induction of TRAIL expression and apoptosis.

To determine whether VPA mediated down-regulation of IL-6R0; occurred at the niRNA or protein level, IL-6R0; message levels were analyzed over time following VPA treatment. For these experiments, U266 and OPM2 cells were treated with VPA (2 mM) for 4-24 hours, and RNA was harvested and analyzed by real time qPCR using primers to the IL-6Rα mRNA. For normalization purposes, the abundance of the 36B4 mRNA level was also measured and the relative abundance of the IL-6Rα-mRNA was calculated using the ΔΔC_t method (Livak and Schmittgen, Methods 25:402-408 (2001)). As illustrated in Fig. 4C, VPA treatment resulted in a reduction of IL-6R0: mRNA with kinetics similar to that observed for the reduction in the IL-6Rα protein expression. Thus, it was concluded that IL-6Ra is rapidly down-regulated at the mRNA level, and that this effect of VPA is observed prior to induction of TRAIL.

Because the effect of VPA on the abundance of IL-6RαmRNAs was so rapid, it was anticipated that this effect was also independent of protein synthesis. U266 cells were treated for 24 hours with or without VPA in the presence or absence of cyclohexamide (CHX)₅ and harvested RNA was analyzed for abundance of IL-βRαmRNA. As illustrated in Fig. 4D, treatment of U266 cells with CHX alone does not affect basal expression of the IL-6Rce mRNAs

(comparing solid bars), and similarly co-treatment of cells with CHX and VPA does not reverse repression of IL-όRαmRNA (striped bars). Therefore, the influence of VPA on transcription of the IL-δRα gene does not appear to require new protein synthesis. VPA down-regulates aberrantly expressed FGFR3 in OPM2 cells.

Recently OPM2 cells were shown to possess a translocation between chromosomes 4 and 14— t(4;14) - which leads to aberrant expression of a constitutively active form of FGFR3 (Ronchetti et al, Oncogene 20:3553-3562 (2001)). Pharmacological blockade of this receptor's activity culminated in halted cell cycle progression and induction of apoptosis, while similarly treated U266 cells, lacking expression of FGFR3, were unaffected (Grand et al,

Leukemia 18:962-966 (2004)). Because exquisite sensitivity of OPM2 cells to VPA was observed, as was rapid regulation of the IL-6RαmRNA, the question presented was whether VPA could similarly be regulating the expression of FGFR3 in 0PM2 cells.

As described previously (Grand et al, Leukemia 18:962-966 (2004)), FGFR3 expression was not detected in U266 cells (Fig. 5A). However, robust expression of FGFR3 was detected in WCE of untreated OPM2 cells (Fig. 5A - upper panel), and the receptor's expression was rapidly reduced by VPA treatment, reaching an undetectable level by 16 hours (similar to the regulation of IL-6Rα). This regulation also occurred at the mRNA level, as real time qPCR analysis of mRNA from similarly treated OPM2 cells demonstrated significantly reduced abundance of the FGFR3 mRNA after just 4 hours of VPA treatment, reflecting what was observed at the protein level (Fig. 5 A - lower panel).

It was next asked whether VPA mediated down-regulation of FGFR3 occurred independently of new protein synthesis as was demonstrated for IL-6Rα in the same cell line. To address this issue, OPM2 cells were treated for 24 hours with or without VPA in the presence or absence of CHX and FGFR3 mRNA expression was assessed. As illustrated in Fig. 5B, CHX affected neither the basal expression of the FGFR3 mRNAs (solid bars) nor the repression of FGFR3 mRNA by VPA (striped bars). Thus, down-regulation of both FGFR3 and IL-6R0C (above) is a primary response to treatment with HDACIs.

As illustrated in Fig. 1 , the SCFAs MAA, VPA, and butyrate exhibited different efficacies with respect to their apoptogenic activities in OPM2 cells despite their comparable HDACI activity. Thus, to determine whether these observed differences correlated with the ability of these HDACIs to regulate FGFR3 mRNA expression, 0PM2 cells were treated for 24 hours with VPA (2 mM), VPD (2 mM), MAA (5 mM), or butyrate (1 mM), and RNA was harvested and examined by real time qPCR for abundance of FGFR3 mRNA. As illustrated in Fig. 5 C, treatment with VPA or butyrate resulted in a 10-fold reduction of FGFR3 mRNA, while MAA repressed FGFR3 mRNA approximately 2-fold and VPD had no significant effect. Comparing the efficacy of these drugs on the respective endpoints described in Fig. IB and Fig. 5C, and observed in several independent experiments, it appears that the apoptogenic activity of each of these compounds in OPM2 cells mirrors their effect on FGFR3 expression. Analysis of the same samples for abundance of the IL-βRαmRNA indicates that VPA and butyrate similarly affect IL-6Rα message levels, but the effects of MAA more closely resemble those of VPD, with a less profound effect on IL-όRαmRNA expression. The disparate effects of MAA and VPD on OPM2 cell survival further indicate that for OPM2 cells, regulation of FGFR3 more directly affects cell survival than does regulation of IL-6Rα!. Finally, mRNA levels for the Bel- XL inhibitor of apoptosis are unchanged in OPM2 cells by any of the treatments, ruling out the possibility that these compounds repress transcription in a non specific manner. VPA down-regulates BCMA in myeloma cells.

It has been demonstrated above that VPA deprives myeloma cells of required survival and proliferative signals that are mediated through up-regulated or aberrant expression of IL-6R0; or FGFR3 on U266 and OPM2 cell lines, respectively. However, FGFR3 is not expressed on RPMI 8226 cells (see Fig. 6B), and these cells do not overexpress IL-6R<x Furthermore, IL-6R0: expression level is not modulated by VPA (Fig. 6B), and it was not possible to detect a significant induction of TRAIL expression in these cells following VPA treatment (data not shown). Because VPA-induced apoptosis proved to be independent of GR signaling in these cells (Fig. 2), it was hypothesized that, similar to the observations made in U266 and OPM2 cells, VPA was modulating the expression of a growth factor receptor necessary for RPMI 8226 cell proliferation and survival. Thus, the examination of growth factor receptors implicated in myeloma cell survival continued with a view to identifying a response that tracked with the apoptogenic action of HDACIs in RPMI cells.

One obvious change in gene expression detected by microarray analysis of patient samples representing normal plasma cells, MGUS (multiple gammopathy of undetermined significance— an expansion of the plasma B cells that can progress to multiple myeloma), and multiple myeloma is the significant up- regulation of the B cell maturation antigen (BCMA) receptor (Claudio et al, Blood 100:2175-2186 (2002), Davies et al, Blood 102:4504-4511 (2003)). This receptor is a member of the TNF superfamily of receptors (called TNFRSFl 7), and together with B cell activating factor receptor (BAFF-R) and transmembrane activator and CAML interactor (TACI), BCMA has recently been shown to mediate the B-lymphocyte survival and proliferation signals associated with TNF family members B cell activating factor (BAFF), B lymphocyte stimulator (BLyS), and a proliferation inducing ligand (APRIL) (Schneider, Curr. Opin. Immunol. 17:282-289 (2005)). BCMA is of particular interest as it has been implicated previously as a plasma cell survival factor (Schneider, Curr. Opin. Immunol. 17:282-289 (2005)), and BCMA expression has been detected on a majority of myeloma cell lines, with several of these lines being subsequently shown to express BAFF, BIyS, and/or APRIL as well (Novak et al, Blood 103:689-694 (2004), Moreaux et al, Blood 103(8):3148-3157 (2004)). Finally, it has been demonstrated that BAFF and APRIL were comparable to L-6 in their ability to rescue RPMI 8226 cells from Dex-induced apoptosis (Moreaux et al, Blood 103(8):3148-3157 (2004)). All of these data together led to the

examination of whether VPA modulated expression of BCMA in OPM2, RPMI 8226, or U266 cells. As illustrated in Fig. 6 A, BCMA expression was detected in all three cell lines. Importantly, in each cell it was observed that VPA treatment reduced the relative levels of both BCMA protein and mRNA (left and right panels, respectively). The most striking down-regulation, a ~ 20-fold down- regulation of the BCMA mRNA, was observed in RPMI 8226 cells. Down- regulation of BCMA message following 24 hrs of VPA treatment correlated with a similar fold down-regulation of BCMA receptor protein in all three cell lines. TACI mRNA was not detected in OPM2 cells and was present only at low levels in U266 cells (data not shown). In RPMI 8226 cells, where TACI was appreciably expressed, its level was not significantly changed by treatment with VPA (data not shown), indicating that this effect of VPA on BCMA expression is specific even within this family of receptors. VPA specifically regulates expression of a subset of membrane expressed receptors.

Having observed regulation of several membrane expressed growth factor receptors by VPA, it was important to be certain that this HDACI was not functioning in a non-specific manner by targeting the expression of all membrane expressed receptors. Therefore, the expression of multiple membrane receptors was examined utilizing both Western analysis and real time qPCR. Several growth factor receptors have been implicated in contributing to myeloma cell survival and proliferation, including vascular endothelial growth factor receptor (VEGFR) 3, granulocyte maturation and colony stimulating factor receptor (GMCSFR) CK, and insulin-like growth factor receptor (IGF-R) a. In addition to analyzing these receptors, expression of the death receptor (DR) 4 was also examined. Because the DR4 ligand TRAIL has been described to participate in apoptosis induced by HDACIs (Nebbioso et al, Nature Medicine 1 l(l):77-84 (2005)), it would be expected that its expression level would not be repressed by VPA. Western (Fig. 6B) and real time qPCR (Fig. 6C) analysis of the expression of DR4, VEGFR3, and IGF-lRαin OPM2, RPMI 8226, or U266 cells (indicated above) indicates that the expression of these receptors were not significantly influenced by VPA treatment. In addition to IL-6Rα and FGFR3, VPA also reduced expression of the IL-6Rcx partner receptor gp 130 in both RPMI and U266 cells, although its expression was unaffected in OPM2 cells (Fig. 6B). In contrast, VPA induced expression of GMCSFRor in both RPMI (5-fold) and U266 (15-fold) cells, while GMCSFRαmRNA was undetectable in OPM2 cells regardless of VPA treatment (Fig. 6C), demonstrating that VPA treatment does not result only in repression of membrane receptors. Finally, Western blot detection of acetylated histone 3 and GAPDH indicate that VPA treatment inhibited HDAC activity in each cell line and that WCE inputs are relatively equivalent (Fig. 6B).

VPA cooperates with other myeloma therapeutics to maximize induction of apoptosis.

SAHA, arsenic trioxide (As₂Os), and Velcade have been utilized with some success to treat Dex -refractory myeloma (Berenson et al, Clinical

Lymphoma 5:130-134 (2004), Richardson et al, N. Engl. J. Med. 352:2487-2498 (2005), Kelly et al, J. Clin. Oncol. 23:3923-3931 (2005)). For a preliminary idea of how VPA would perform in a similar setting, a comparison was made of its efficacy as an apoptogenic agent in the presence or absence of SAHA, AS2O3. or Velcade in myeloma cells. As illustrated in Fig. 7A, treatment of RPMI 8226, OPM2, and U266 cells with increasing doses of VPA (1-5 mM), SAHA (l-5μM), or AS₂O₃ (0.5-5μM) resulted in significant induction of apoptosis in each cell line. The apoptogenic effects of VPA in each cell line were comparable to those of the other therapeutics, indicating that across a spectrum of sensitivities of myeloma cell lines to cytotoxic drugs, VPA performed as a chemotherapeutic agent equivalent to those already described for treatment of refractory myeloma. To determine whether SAHA₅ As₂O₃, Dex, MAA or velcade would cooperate with VPA to increase the apoptotic response in myeloma cells, RPMI 8226 (Fig. 7B) or OPM2 (Fig. 7C) cells were co-treated with sub-maximal doses of Dex, SAΗA, As₂O₃, MAA₅ or velcade alone or together with a sub-optimal dose of VPA, as indicated. To facilitate comparison, the theoretical sum of the percentages of induction of apoptosis for each of the single agents alone, representing the expected result of co-treatment if the compounds are working independently, is presented. Co-treatment of VPA with each of these compounds resulted in a statistically significant increase in apoptosis as compared to treatment with either compound alone, suggesting that use of these agents in combination in a clinical setting may improve patient response. Interestingly, in the Dex-sensitive RPMI 8226 cells, additive effects of VPA with As₂O₃ and SAHA were similarly .

observed as was cooperation of VPA with Dex, although the observed rate of apoptosis of VPA in combination with Dex was less than the sum of either drug alone (Fig. 7B).

Similar to the RPMI 8226 cells, the apoptogenic effects of VPA also appear to be additive to those of SAHA and As₂O₃ in the OPM2 cells (Fig. 7C), where little significant difference was observed between the percentage of apoptotic cells following co-treatment with these compounds and the projected theoretical sum calculated from the percentages of apoptotic cells observed for the two drugs independently. As in Fig. 7B₅ the theoretical sum of the apoptogenic effects of each of the single agents is presented to facilitate comparison. In the OPM2 cells, however, which are less sensitive than the RPMI cells to Dex and display loss of IL-6Rα and FGFR3 in response to VPA, it was observed that Dex synergized with VPA to induce apoptosis, as cell death observed with co- treatment is significantly greater (p<0.001) than the sum of the independent effects of each agent alone (Fig. 7C). These results indicate that in a cell line where VPA perturbs the acquired resistance to Dex by down-regulating the IL- 6Ra and FGFR3, Dex regains some of its apoptogenic potential, supporting the hypothesis that addition of VPA to a corticosteroid-based treatment regimen can enhance the efficacy of Dex even when myeloma cells have acquired some mechanisms to facilitate Dex resistance. In contrast, the effects of VPA were approximately additive to those observed for SAHA (Fig. 7C), suggesting that these drugs are functioning through the same pathway (inhibition of HDACIs). In 0PM2 cells, VPA also synergized with MAA to induce apoptosis, resulting in significantly greater (p<0.05) cell death than can be explained as the sum of the apopto genie effects of VPA and MAA alone (Fig. 7C). Just as the myeloma cell lines display differential sensitivity to MAA and VPA, these data further suggest that these two SCFAs may be functioning through different mechanisms or pathways in these circumstances. Finally, the proteasome inhibitor Velcade has previously been shown to synergize with the HDACIs butyrate and SAHA to induce apoptosis in myeloma cells (Mitsiades et al, Proc. Natl. Acad. Sci. USA 101(2):540-545 (2004), Pei et al, Clin. Cancer Res. 10(11):3839-3852 (2004)). Similarly, synergy between VPA and Velcade was observed, as these drugs in combination resulted in significantly greater (p<0.001) apoptosis than the sum of the effects of the drugs used independently. Together, the results^' of Fig. 7 demonstrate the potential of combining VPA treatment with other regimens to achieve greater apoptotic rates in myeloma cells and support combination therapy utilizing VPA clinically. VPA. induces apoptosis in myeloma patient cell isolates.

The above data indicate that VPA efficiently induces apoptosis in myeloma cell lines at doses that previous studies indicate can be sustained physiologically in human patients with acceptable side effects. As a measure of the clinical potential of VPA to treat myeloma, the effect of VPA treatment on cells isolated from the bone marrow of two myeloma patients was compared to the effects observed on myeloma cell lines. Isolates from patient 1 were cultured for 96 hours in the presence or absence of VPA (Fig. 7D), and were analyzed by 7- AAD and Annexin V staining to determine the effect of VPA on cell survival. As illustrated in Fig. 7D, while 90% of the cultured patient cells survived the incubation time in culture without treatment, an approximate 2-fold increase in apoptosis was observed in samples treated with VPA, comparable to that observed with the VPA-treated myeloma cell lines. In Fig. 7E, plasma cell isolates from patient 2 were treated with VPA (2 mM), Dex (50 nM), or As₂O₃ (5μM) to compare the apoptogenic response to each of these agents. Consistent with the use of corticosteroid therapy as initial treatment for myeloma, of the agents examined here, the cells from patient 2 were least responsive to Dex with only a 10% increase in apoptosis. The response to VPA was comparable to that observed for the experimental drug AS₂O₃ with an approximate 35% and 45% decrease in living cells observed for each treatment, respectively, further supporting the clinical potential of VPA to treat myeloma.

Summarizing, HDACIs have been shown to induce apoptosis and to increase the expression of p21 and the death ligand TRAIL in a variety of hematopoietic tumor cells (Lavelle et al, Am J Hematol 68:170-826 (2001), Nebbioso et al, Nature Medicine 1 l(l):77-84 (2005)). In the study described above, it is demonstrated that HDACIs also induce apoptosis and enhance TRAIL expression, albeit after extended periods of treatment, in several multiple myeloma cell lines. Of importance, however, was the observation that HDACIs can also effectively down-regulate growth factor receptor expression within 8-16 hours, an activity that appears to correlate with the accumulation of cells in the G1/G0 phase of the cell cycle. These results suggest that a primary activity of HDACIs in multiple myeloma is the desensitization of signaling systems required for both cell proliferation and survival.

The importance of the IL-6 pathway in myeloma cell survival has been well documented, and overexpression of both the IL-δRα and its ligand has been implicated in de novo and acquired resistance to glucocorticoids. It has been shown, for instance, in several different myeloma cell lines that IL-6 protects these cells from the antiproliferative and apoptogenic effects of Dex (Juge- Morineau et al, Br. J. Haematol. 90(3):707-710 (1995)). While these observations were made using cells in culture, there is abundant data to suggest that this pathway is also relevant in clinical disease. Notably, the serum levels of the soluble form of the IL-6 receptor (sIL-6R) is frequently elevated in myeloma patients, and high levels of sIL-6R correlate with poor prognosis and can be used as an indicator of treatment response and disease progression (Papadaki et al, Acta Haematol. 97(4):191-195 (1997), Kyle, Stem Cells 13(Suppl 2):56-63 (1995)). Microarray analysis of circulating plasma cells isolated from patients have indicated that the IL-6RαrnRNA is upregulated in these cells approximately 4-fold as compared to healthy individuals (Chauhan et al, Oncogene 21:1346- 1358 (2002)). Furthermore, within the bone marrow microenvironment, the interaction of myeloma cells with BMSCs stimulates the release of IL-6, generating a feedback loop that stimulates myeloma cell survival and proliferation and reduces the efficacy of chemotherapy regimens (Hayashi et al, Br. J.

Haematol. 120:10-17 (2003)). These observations validate the IL-6 pathway as a target for anti-myeloma therapy. Indeed, studies are ongoing in the clinic to evaluate whether chemoresistance can be circumvented using a specific antibody to reduce the circulating levels of IL-6 (Rossi et al, Bone Marrow Transplant 22:Epub (2005), Brochier et al, Int. J. Immunopharmacol. 17:41-48 (1995),

Legouffe et al, Clinical Experimental Immunology 98:323-329 (1994)). Because dysregulation and constitutive activation of several growth factor receptors has been demonstrated to occur in myeloma, such an approach is of inherently limited scope. Based on the present data, however, it is believed that HDACIs are an attractive alternative as they target multiple different growth factor signaling pathways.

The OPM2 cell line is a model of a specific subset of myelomas and thus the observation that HDACIs can lead to apoptosis in this glucocorticoid resistant cell line has a very immediate clinical implication. Specifically, the translocation t(4;14) that gives rise to aberrant expression of the FGFR3 in the OPM2 cell line is also found in 10-20% of myeloma patients (Rasmussen et al, Br. J. Haematol. 117:626-628 (2002)). Importantly, blockade of FGFR3 activity, by either specific receptor tyrosine kinase antagonists or with targeted shRNAs, results in apoptosis of OPM2 cells regardless of co-treatment with IL-6 (Grand et al, Leukemia 18:962-966 (2004), Zhu et al, Molecular Cancer Therapeutics 4(5):787-798 (2005)). It is interesting to note that the HDACI VPA was more effective than MAA at inducing apoptosis in OPM2 cells and that this correlates well with the differences in their ability to down-regulate the FGFR3 mRNA and receptor protein. Because t(4;14) and FGFR3 expression are generally predictive of a particularly aggressive disease and poor patient outcome (Rasmussen et al, Br. J. Haematol. 1 17:626-628 (2002)), the specific effect of VPA on FGFR3 expression suggests that this drug may be of particular value in treating patients with this translocation. Interestingly, FGFR3 expression has also been detected in bladder and cervical carcinoma cell lines (Jebar et al, Oncogene epub:l-8 (2005)).

However, it was not possible to detect a change in FGFR3 mRNA abundance in J82 urothelial carcinoma cells following VPA treatment indicating that the actions of this drug may be directed towards the abnormal FGFR3 gene in myeloma cells.

One of the most important findings of the present studies was the observation that VPA strongly inhibits the expression of B-cell maturation antigen (BCMA), a receptor for a signaling pathway that has been recently implicated, in the progression of myeloma. The upregulation of BCMA in myeloma cells as compared to normal plasma cells (Claudio et al, Blood 100:2175-2186 (2002), Davies et al, Blood 102:4504-4511 (2003)) and the significantly increased detection of BCMA ligands in myeloma patient serum (Moreaux et al, Blood 103(8):3148-3157 (2004)) highlight the clinical potential of VPA regulation of this pathway. HDACI dependent induction of apoptosis in multiple cancer cells, including multiple myeloma, has been well established, but the mechanism is not yet clearly defined. Although histone acetylation is the indicator used to characterize and compare HDACIs it is clear from the findings described above that there is something in common in the pathways that regulate the expression of the IL-όRα, the FGFR3 (in the context of the t4: 14 translocation), and the BCMA antigen that serves as the target for HDACIs. Future studies are described toward identifying the target(s) that confer sensitivity of these growth factor receptors to HDACIs. Possible mechanisms include (a) a reduction in the activity and/or expression of a positive acting transcription factor (b) an increase in the activity of a negative acting transcription factor or (c) an alteration in the activity of a protein that governs the stability or processing of these growth factor mRNAs. There is precedent for the regulation of transcription factor function by acetylation. For instance, acetylation of NFKB enhances its activity by increasing its affinity for DNA and reducing its interaction with the IKB repressor (Quivy et al, Biochem. Pharmacol. 68:1221-1229 (2004)). Conversely, acetylation of C/EBPα reduces its DNA binding activity and its ability to regulate target gene transcription (Legace and Nachtigal, J. Biol. Chem. 279(18):18851-18860 (2004)). Finally, modulation of the acetylation state of Sp3 by treatment of cells with TSA has been shown to convert this protein from a repressor to an activator (Ammanamanchi et al, J. Biol. Chem. 278:35775-35780 (2003)). In the context of this discussion, it is worth noting that NFKB, C/EBP and SpI and 3 have all been implicated in transcriptional control of growth factors and their receptors, making these attractive targets through which HDACIs may be mediating their effects in myeloma cells. However, an analysis of the properties of these specific transcription factors in myeloma cells following HDACI treatment did not reveal obvious changes in their expression level, acetylation state, or cellular localization. Therefore, an alternative set of transcription factors may be important.

A second attractive hypothesis is that acetylation decreases the stability of a transcription factor or a coregulatory protein required for growth factor receptor expression. However, a formal test of this hypothesis awaits a clearer

understanding of the factors involved in the expression of the growth factors that have determined to be important for myeloma cell survival. However, there is precedent for such a regulatory activity in other systems. The stability of the E2F1 protein is increased subsequent to its acetylation following the treatment of cells with DNA damaging agents (Ianari et al, J. Biol. Chem. 279:30830-30835 (2004)). While estrogen receptor (ER) recruitment of the coactivator SRC- 3/ACTR is required for maximal ER transactivation of several promoters, SRC-3 also has the secondary activity of acetylating ER, resulting in the agonist- dependent turnover of this receptor (Shao et al, Proc. Natl. Acad. Sci. USA 101 :11599-11604 (2004)).

An alternative, more provocative hypothesis for the mechanism of HDACIs influence of growth factor receptor mRNA abundance is that, rather than transactivation of promoters, HDACIs may specifically affect the processing or stability of these mRNAs. The specificity of the response to HDACI treatment is intriguing in that three growth factor mRNAs, each encoding factors important to survival of myeloma cells, were affected, suggesting that rather than specific promoters, HDACIs seem to be targeting a class of mRNAs. Previous studies have determined that mRNAs of like function or that encode components of a molecular pathway or structural entity are generally processed by the same RNA binding proteins (Penalva et al, Methods MoI. Biol.257:125-134 (2004), Keene and Lager, Chromosome Res. 13:327-337 (2005)). Based on recent elegant work by the Keene, it possible that the expression of an mRNA binding protein, required for the processing of these receptor mRNAs, is inhibited either in its expression or its activity by increased acetylation. This possibility is being investigated currently by surveying the expression of key RNA binding proteins in myeloma cells treated with different HDACIs.

Although the main objective of the present study was to define the primary responses to HDACIs, the demonstration that VPA was particularly effective at inducing apoptosis in all myeloma models studied has near term clinical implications. This drug is widely used for the treatment of epilepsy, bipolar disease and migraine and has a well established safety record and well defined pharmaceutical properties (Perucca, CNS Drugs 16:695-714 (2002)). When used for these conditions the serum levels of the drug achieved would be expected to yield maximum HDACI activity. Thus, a phase II trial could be undertaken to evaluate the efficacy of this drug, alone or in combination with other drugs, in multiple myeloma patients who have failed a primary intervention with glucocorticoids. This approach would be supported by the encouraging results of ongoing clinical trials of VPA as a single agent or in combination therapy to treat a wide range of solid tumors or myelodysplastic conditions (Blaheta et al, Med. Res. Rev. 25(4):383-397 (2005)). The observation that VPA sensitizes the glucocorticoid resistant cell line 0PM2 to Dex is particularly important as it suggests that this drug, or another HDACI, could have utility in frontline steroid based therapies by (a) reducing the dose of the glucocorticoid and (b) delaying the onset of resistance. Based on these findings there is reason to believe that VPA (or another HDACI) in combination with drugs like velcade and/or arsenic trioxide will have utility in the treatment of multiple myeloma.

EXAMPLE 2 EXPERIMENTAL DETAILS

Cell lines. OPM2 cell line was provided by E. Brad Thompson, Baylor College of Medicine, Houston, TX. H929 cells were obtained from ATCC (Monassas, VA). Cells were maintained in modified RPMI 1640 (ATCC) supplemented with 12% FBS in a humidified incubator maintained at 37°C and 5% CO₂. During treatments, cells were plated at densities of IxIO⁵ or 3xlO⁵ cells/ml in media alone or including indicated treatments.

Reagents. Valproate (VPA), sodium butyrate (NaB), and were ordered from Sigma Aldrich (St. Louis, MO) as sodium salts and dissolved in water. Suberoylanilide hydroxamic acid (SAHA - Merck, Whitehouse Station, NJ), doxorubicin (Dox - Sigma), and DL-α-Difiuoromethyl ornithine hydrochoride (DFMO - Sigma) were dissolved in DMSO. Where applicable, cells were pretreated with DFMO for 5 hours prior to co-treatment with VPA. Valpromide (VPD— Lancaster Synthesis, Pelham, NH) and dexamethasone (D ex - Sigma) were dissolved in 100% ethanol. Amino acid supplement (1OmM) was obtained from Invitrogen (Carlsbad, CA). The above compounds were diluted in culture media immediately prior to use.

³H-acetate distribution. 1x10⁶ cells were incubated one hour in media containing 50μM (25μCi/ml) ³H-acctatc and then treated as indicated for 24 or 48 hours prior to washing twice in PBS+1% BSA. Washed cells were resuspended in 1 OOμ sterile water and snap frozen. Samples were thawed and pipeted repeatedly to shear DNA prior to a second freeze/thaw cycle. Protein was precipitated by addition of 800μl MeOH and cleared by centrifugation.

Supematent containing soluble counts was removed, and pellets were solubilized in 0.5ml sterile water. Counts present in both fractions were detected using a Beckman LS6000SC scintillation counter.

Amino acid and acyl carnitine analysis. 3x10 cells were treated as indicated for 24 or 48 hours prior to washing twice in PBS+1% BSA. Washed cells were sonicated in 300μl sterile water and lysates were cleared by

centrifugation. Protein concentration was measured by Bradford assay, and 1 OOμl lysate was deproteinated through addition of 800μl methanol and cleared through by prior to MS-MS analysis as previously described (Rashed et al, Pediatric Research 38:324-331 (1995), An et al, Nature Medicine 10:268-274 (2004), Jensen et al, Journal of Biological Chemistry 281 :22342-22351 (2006)). Detected levels of amino acids and acyl carnitines were normalized to mgs sample protein input prior to averaging triplicate samples.

HDAC inhibition. Nuclear extracts prepared from untreated 0PM2 cells were incubated with indicated concentrations of HDACI and HDAC activity was analyzed using kit AK-500 (BIOMOL International, Plymouth Meeting, PA) per manufacturer's instructions.

Glucose uptake. Treated cells were washed in PBS+1 % BSA and resuspended at 2x10⁵ intact cell/ml (determined by trypan blue staining prior to washing) in warmed KRH buffer (2OmM Hepes, pH 7.4, 1.25mM MgSO₄, 1.25m CaCl₂, 14OmM NaCl, 5mM KCl, 2% BSA). lOOμl (2x10⁵) cells were combined with 150μl KRH containing 2/xCi of 2-deoxy-D-[2,6-³H]-glucose (2-DOG - GE Healthcare, Piscataway, NJ) or 3-O-methyl-D-[l-³H]-glucose (3-OMG - GE Healthcare). Following 0-30 minutes incubation at 37°C, glucose transport was stopped by addition of 200μl ice cold KRH buffer containing 200μM phloretin (Calbiochem, San Diego, CA). Stopped samples were incubated on ice prior to the cells being washed twice in PBS+1% BSA. Samples were lysed by addition of 200μl IN NaOH. Retained radioactivity was detected by scintillation counting.

Cell viability and apoptosis. 1x10^s cells were treated for 24-96 hrs with the indicated compounds in ImI total volume of media. Cells were harvested by 5 centrifugation (500xg for 5 min), washed twice in PBS, and stained with PE- conjugated Annexin V and 7-AAD per manufacturer's instructions (Pharmingen, San Diego, CA) prior to FACS analysis.

Real Time quantitative PCR. 0.5-2x10⁶ cells treated as indicated were harvested by centrifugation and washed in PBS prior to lysis. RNA isolation io (BioRad - Hercules, CA) and reverse transcription (iScript— Biorad) were

performed per kit manufacturer's instructions. Real time qPCR of cDNA was done using iQ SYBR Green supermix (Bio-Rad) per kit instructions, and amplification was performed using the iCycler optical system with associated software (Bio-Rad). mRNA abundance was calculated using the ΔΔCj method as

15 previously described (Livak and Schmittgen, Methods 25:402-408 (2001)).

Primer sequences included: 36B4 forward (F):

GGACATGTTGCTGGCCAATAA. reverse (R): GGGCCCGAGACCAGTGTT; HXKl (F): GGACTGGACCGCTGAATGT, (R):

ACAGTTCCTTCACCGTCTGG; GLUTl (F):

^•2 o CTTTTCTGTTGGGGGCATGATTG, (R) : CCGCAGTAC ACACCGATGAT; and CPSl (F): GCTGGCTACCAAGAGTTTAGG, (R):

ACAGGCTGACCTTGAAATTCAAT.

Western blot analysis. SDS PAGE and Western blot analysis was performed per instructions by Bio-Rad using antibodies to GLUTl (Abeam, 25 Cambridge, MA) or HXKl- (Santa Cruz Biotechnology, Santa Cruz, CA). Retrovirus construction. Rat GLUTl cDNA was cloned into pMIGR- NGFR retroviral expression vector, which allows co-cystronic IRES mediated expression of the extracellular domain of nerve growth factor receptor (NGFR). A FLAG epitope tag was inserted into an extracellular loop of GLUTl.

Amphitrophic retrovirus was produced by co-transfection of pMIGR-GLUTl with pVSVg (Clontech, Mountain View, CA) into GP2 293 packaging cells

(Clontech). OPM2 cells were infected with GLUTl retrovirus by combining filtered spent media from the GP2 293 infection with OPM2 cells in RPMI media supplemented with 4μg/ml polybrene (Sigma). Following 48 hours incubation, cells were stained for NGFR expression with a PE-conjugatcd antibody to NGFR (Pharmingen) and sorted using FACS analysis. Sorted cells were maintained as a polyclonal population, and expression of FLAG-GLUTl was monitored weekly by immunostaining for FLAG (Sigma) and FACS analysis.

Hexokinase activity. Hexokinase activity was analyzed, as previously described (Bauer et al, FASEB Journal 18(11): 1303-1305 (2004)) and V_max was normalized to mgs protein input as determined by Bradford assay.

Alpha-ketoglutarate detection. 8x10⁷ OPM2 cells were treated as indicated, harvested by centrifugation, and lysed in 0.1M HCl containing internal standards. Samples were deproteinated, prepared, and analyzed by GC/MS as previously described (Jensen et al, Journal of Biological Chemistry 281:22342- 22351 (2006)).

Urea analysis. Urea levels in 50μl of retained media were measured using kit DIUR-500 (Bioassay systems) per manufacturer's instructions. RESULTS

Previous studies have analyzed the fold change in histone acetylation in response to HDACIs. There was interest in determining a measure of the fold change in protein acetylation on a cellular scale as influenced by HDACIs. In short, the question presented was whether HDACIs would induce a significant shift in labeled acetyl groups from a soluble pool (ie. acetyl CoA) to a protein- associated fraction through acetylation. To this end, OPM2 cells were incubated with ³H-acetate prior to treatment with HDACI in order to label the cellular pool of acetyl CoA. Following 24 or 48 hours treatment and cell lysis, samples were deproteinated using methanol to precipitate protein, and the proportion of soluble versus insoluble (protein associated) counts was determined for each treatment. As illustrated in Fig. 9A, HDACI treatment did not influence the distribution of acetate between soluble and protein associated pools at 24 hours, but at 48 hours, ^' the proportion of labeled acetate that precipitates in the protein fraction increased 1.5 fold in valproic acid (VPA) treated cells and 1.25 fold in suberoylanilide hydroxamic acid (SAHA) treated cells.

The above results indicate that HDACIs induce such a profound change in the acetylation state of chromatin and other cellular proteins that they lead to a detectible shift of labeled acetate from a soluble pool to precipitable protein. The next question asked was whether this consumption of acetyl CoA would result in a detectible change in the intracellular levels of acetyl CoA. Acetyl CoA is held in equilibrium with acetyl carnitine by the enzyme carnitine acetyl transferase. Thus, measurement of acetyl carnitine reflects the levels of acetyl CoA in the cell. OPM2 human multiple myeloma cells were treated for 48 hours with VPA, sodium butyrate (NaB), SAHA, or Dexamethasone (Dex - 10OnM). All of these agents have been shown to induce significant (60-90%) and comparable apoptosis in this cell line as well as several other myeloma cell lines within 96 hours, although no more than 10-20% apoptosis is observed at this time point of 48 hours (data not shown). Using MS-MS analysis to measure acetyl carnitine present in cell lysates, an approximate 30% decrease was observed in acetyl carnitine in cells treated with HDACIs (Fig. 8B), but no change in cells treated with Dex, indicating that treatment with HDACI is able to shift equilibrium of acetyl CoA in the cell. This effect not likely related to induction of apoptosis because these doses of Dex and HDACI induce apoptosis in OPM2 cells with similar efficiency during longer treatments. Real time quantitative PCR analysis revealed only slight change in the mRNA expression of carnitine acetyl transferase (data not shown), suggesting that altered enzyme expression is not likely responsible for the detected change in acetyl carnitine. These data suggest that HDACIs are instead reducing the overall size of the available acetyl CoA pool.

The dose response and timing of HDACI modulation of acetyl CoA were next examined, analyzing acetyl carnitine levels in OPM2 cells similarly treated with VPA (0.5-2mM) or SAHA (1-5 μM) over 24 or 48 hours. Neither HDACI affected acetyl carnitine levels at 24 hours (Fig. 8C), but at 48 hours a modest and reproducible increase was observed in acetyl carnitine at low doses of VPA and a decrease in acetyl carnitine with higher doses of VPA (Fig. 8D). SAHA resulted in a dose dependent decrease in acetyl carnitine.

An effort was made to correlate the differing outcomes of these compounds with their efficacy of HDAC inhibition. Nuclear extracts were prepared from untreated OPM2 cells and were incubated with increasing concentrations of VPA or SAHA. HDAC activity was determined through addition of a fluorescently tagged substrate that can be detected following deacetylation. A disconnect was observed between the doses of VPA and SAHA that achieved a significant change in acetyl CoA and those that most efficiently inhibited nuclear HDAC activity. While SAHA achieved 80+% HDAC inhibition at very low doses (0.5μM and higher), higher doses (>1 μM) were required to reduce acetyl CoA (Fig. 8E). In contrast, doses of VPA observed to alter acetyl CoA levels (0.5-2mM) achieved only 40-60% HDAC inhibition. VPA has been shown to most efficiently inhibit the nuclear class I HDACs (IC50 = 0.5mM) with less efficient inhibition (IC₅₀ = 2.5mM) of the class II HDACs that shuttle between cytoplasm and nucleus (Gottlicher et al, EMBO Journal 20(24):6969- 6978 (2001)). In contrast, SAHA inhibits both class I and II HDACs with similar efficiency (IC₅₀ = l.OμM) (Hildemann et al, Journal of Biotechnology 124:258- 270 (2006)). Because the doses of VPA used in Figs. SB, 8C and 8D are below the IC₅₀ for class II HDACs, it is surmised that the effect of HDACIs on acetyl carnitine levels is most likely mediated through class I HDACs.

Acetyl CoA is tightly regulated within the cell, with excess being quickly shunted into formation of long chain acyl carnitines. Likewise a decrease in acetyl CoA is rapidly amended, primarily through increased glycolysis, but also though metabolism of acyl carnitines or amino acids. Therefore the question asked was whether HDACI treatment would result in increased glycolysis as indicated initially by an increase in glucose uptake. OPM2 cells were treated 24 or 48 hours with VPA, SAHA, or Doxorubicin (Dox) prior to analysis of cellular uptake of ³H-labeled 2-deoxy-glucose. The doses selected of each drug result in comparable rates of apoptosis in OPM2 cells following 96 hours of treatment (data not shown). It was observed that, rather than increased glucose uptake, HDACIs caused a time dependent decrease in glucose uptake (5-10 fold at 48 hours - Fig. 9A). Dox treatment resulted in no decrease, and perhaps a slight increase, in glucose uptake, indicating that HDACI activity, rather than cytotoxic activity, contributes to reduced glucose transport. This effect of HDACIs was consistent between myeloma cell lines, as H929 cell responded likewise to HDACI treatment with a similar fold decrease in glucose uptake (Fig. 9B).

Reduced glucose uptake was not a result of apoptosis, as analysis of these cells showed a no more than 10% decrease in live cells as determined by annexin V and 7-AAD staining (Figs. 9C and 9D), and this increase in apoptosis was observed only at 48 hours while a significant reduction in glucose uptake was observed following just 24 hours of treatment. Furthermore, the cells were counted with trypan blue staining prior to glucose uptake analysis to ensure that the same number of intact cells was included in the assay for each treatment.

The profound decrease in glucose uptake that was observed led to the hypothesis that HDACIs were influencing expression or activity of glucose transporter proteins. Real time quantitative PCR (RTqPCR) analysis of RNA isolated from untreated OPM2 cells indicated that glucose transporters (GLUT) 1 and 8 are the members of the glucose transporter family most abundantly expressed in this cell line (data not shown). GLUT8 is primarily associated with transport of fructose, while the almost ubiquitously expressed GLUTl is known to be a primary glucose transporter utilized by hematopoietic cells, as well as though to be responsible for basal glucose uptake in most cell types, and its expression has been associated with more aggressive forms of solid or hematological neoplasms. Because the glucose uptake assays were done in the absence of insulin or IGF stimulation, it was speculated that HDACIs were influencing GLUTl. RTqPCR analysis of RNA isolated from OPM2 cells treated 0-24 hours with VPA demonstrated a rapid decline in GLUTl mRNA following HDACI treatment, suggesting that HDACIs are regulating expression of GLUTl at the level of transcription (Fig. 9E). mRNA levels of other GLUTs were either unchanged or slightly increased, thus correlating GLUTl mRNA expression with the observed decrease in glucose uptake (data not shown). Western blot analysis of OPM2 cells treated VPA for 24 or 48 hours indicated a slight reduction in GLUTl protein at 24 hours, with a more pronounced decrease at 48 hours, thus correlating with glucose uptake (Fig. 9F).

Glucose homeostasis is known to be regulated, and in some cell types maintained, by extracellular signals that are conducted through growth factor receptors and lead to activation of signaling pathways including Akt, a master regulator of glucose uptake (Barthal et al, J. Biol. Chem. 274:20281-20286 (1999)). The data in Fig. 4B, 5A and 6A demonstrate that HDAC inhibitors downregulate expression of several growth factor receptors, leading to the question of whether the loss of growth factor receptor signaling could contribute to the HDACI-induced reduction of glucose uptake observed. Because FGFR3 proved to be one of the most robustly regulated growth factor receptors, the question raised was whether pharmacological inhibition of FGFR3 would resemble the inhibition of glucose uptake observed following HDACI treatment. OPM2 cells were treated 24 hours with FGFR3 inhibitor PD 173074 or HDACI

VPA or SAHA prior to analysis of glucose uptake. Treatment with either HDACI or PDl 73074 resulted in similar fold inhibition of glucose uptake, suggesting that the immediate loss of growth factor receptor expression caused by HDACI may contribute to the reduction of glucose uptake associated with HDACI treatment (Fig. 9G).

The next goal was to determine what contribution decreased glucose uptake made to induction of apoptosis in OPM2 cells. To this end, a retroviral vector was constructed that incorporated a rat GLUTl cDNA with a FLAG tag inserted into an exofacial loop, making the surface expression of the

overexpressed FLAG-GLUTl detectible by FACS. Stably infected OPM2 cells

(OPM2-GLUT1) were sorted for expression of a co-expressed co-cistronic truncated human nerve growth factor receptor (NGFR) (data not shown). These cells stably maintained surface expression of both FLAG-GLUTl and NGFR indefinitely. Glucose uptake assays comparing OPM2 parent cells with OPM2- GLUTl cells or OPM2 cells infected with an empty retroviral vector control expressing only NGFR indicated that truncated NGFR did not affect glucose uptake while overexpression of GLUTl resulted in a 2-3 fold increase in, glucose uptake (Fig. 1 OA), demonstrating that the overexpressed GLUTl is active. Treatment of these OPM2 derivative cell lines with VPA for either 24 or 48 hours resulted in a decrease in glucose uptake comparable in magnitude to that observed in the parent OPM2 cells (Fig. 10A). Parallel Annexin-V and 7-AAD analysis of these samples indicated that the reduced glucose uptake observed in VPA-treated OPM2-GLUT1 cells could not be attributed to increased apoptosis, as similar rates of apoptosis were observed in parent and derived cell lines (data not shown). As demonstrated by Western blot analysis, VPA also did not significantly affect expression of FLAG-GLUTl (Fig. 10B), and FACS analysis confirmed that FLAG-GLLfTl remained detectable on the cell surface in the presence of VPA (data not shown), demonstrating that overexpression of GLUTl was not sufficient to restore the glucose uptake following HDACI treatment. The profound effect on glucose uptake via either endogenous or overexpressed GLUTl, without a corresponding decrease in protein expression of FLAG-GLUTl suggests that HDACIs are directly influencing the rate of glucose uptake. Therefore, glucose uptake was analyzed in OPM2 (Fig. 10C) or OPM2-GLUT1 (Fig. 1 OD) cells over 0-30 min. The curves obtained without VPA treatment indicated that glucose uptake in these two cell lines had similar apparent K_ms but disparate V_max, consistant with overexpression of active GLUTl. At no point in the curves was the glucose uptake similar in the presence or absence of VPA, demonstrating that VPA is not merely slowing the rate of glucose uptake, but is in fact reducing the overall capacity of the cell to transport glucose. VPA reduced the V_max of glucose uptake in both cell lines, while Valpromide (VPD), an analog of VPA lacking its HDAC inhibitor activity (data not shown), had no such effect, further

demonstrating that HDAC inhibition contributes to altered glucose uptake.

The inability of GLUTl overexpression to reconstitute glucose uptake led to the hypothesis that HDACIs were either directly influencing GLUTl activity or affecting the rate other steps of glycolysis. 2-deoxy-glucose (2-DOG) does not proceed in glycolysis past uptake by glucose transporters and phosphorylation by hexokinase to produce 2-DOG-6-phosphate. Without phorphorylation, glucose (or 2-DOG) can freely pass out of the cells, leaving a possibility that the reduced uptake and retention of 2-DOG in the presence or absence of GLUTl

overexpression as observed in Figs. 1 OA, 1OC and 1OD was mediated by HDACI decreasing activity or expression of hexokinases. To more directly analyze the activity of GLUTl, a comparison was made of glucose uptake in OPM2 cells treated 24 or 44 hours with VPA using either 2-DOG as previously or an alternative glucose analog 3-O-methyl glucose (3-OMG) that cannot be phosphorylated by HXK. Analysis with 2-DOG indicated a time dependent decrease in the rate of glucose uptake, with approximate 2- and 5-fold decreases observed at 24 and 44 hours, respectively (Fig. 10E). A parallel analysis of the same treated samples using 3-OMG revealed only a slight decrease in uptake at 24 hours, but a two fold decrease in uptake after 44 hours of VPA treatment (Fig. 10F). These data indicate that VPA is influencing the cumulative glucose uptake observed with 2-DOG at 24 hours without affecting the activity of glucose transporters, leading to the conclusion that HDACIs are influencing the activity of hexokinases at this time point. The decrease observed at 44 hours using 3-OMG is most likely due to reduced expression of GLUTl (Fig. 9F). The greater fold decrease after 44 hours of treatment that was observed using 2-DOG most likely represents the sum of effects of HDACIs on both hexokinase activity and GLUTl expression. Similar results were observed in OPM2 cells treated 24 or 44 hours with SAHA (data not shown).

An analysis of HXK activity in lysates of OPM2 cells following 24 or 44 hours of treatment with VPA or SAHA revealed 50% decrease in the V_max of HXK activity detected at 24 hours and maintained through 44 hours (Fig. 1 OG). RTqPCR analysis indicated that HXKl was the most abundantly expressed HXK in OPM2 cells, with a barely detectable level of HXK3 also observed (data not shown). No discernable expression of HXK2 or 4 was observed. Further qPCR analysis of RNA from OPM2 cells treated 24 or 44 hours with VPA revealed an 8-10 fold induction of HXKl mRNA expression by VPA (Fig. 10H), suggesting that decreased glucose uptake and HXK activity is provoking an attempted adaptive response of HXK induction. The observed increase in HXKl mRNA translated into increased HXKl protein expression as shown by Western blot analysis of lysates of OPM2 cells treated 24 or 44 hours with VPA or SAHA (Fig. 101). These data indicate that HDACIs are reducing HXK activity despite an induction of HXK expression, and together they suggest that HDACIs are reducing glucose uptake at the level of GLUTl expression as well as HXK activity.

The reduced glucose uptake that was observed following HDACI treatment most likely results in the decline in acetyl CoA that was originally noted. However, it would be expected that the cells would then utilize an alternative resource to replenish acetyl CoA, one likely source being long chain fatty acids or acyl carnitines. Like acetyl CoA, long chain fatty acids are most easily quantitated by analysis of their acyl carnitine counterparts. Thus it was asked whether acyl carnitines were being metabolized in the presence of HDACIs. Through MS-MS analysis of extracts of OPM2 cells treated with increasing doses of VPA or SAHA for 24 or 48 hours, a dose and time dependent decrease in staple stores for the cells was observed, including myristate, palmitate, stearate, and oleate (Fig. 11). As with acetyl CoA₃ low doses of HDACIs induced a slight increase in fatty acids, whereas higher doses resulted in up to 10 fold reduction of these fatty acids. These data suggest that the crippling of glucose uptake caused by HDACIs was resulting in a switch in metabolic fuel for the cell and a corresponding depletion of fatty acid stores. Similar results were obtained in a comparable analysis using the H929 myeloma cell line (data not shown), indicating this response to HDACIs is not limited to 0PM2 cells. However, this cellular response is apparently still insufficient to restore acetyl CoA equilibrium. Deamination of amino acids allows metabolism of their carbon chain backbone, with select amino acids being converted to pyruvate as well as TCA cycle components. Because the cells were exhibiting signs of starvation and a broad metabolic shift, it was asked whether the cells had also initiated amino acid metabolism. MS-MS analysis of WCE of OPM2 cells treated with HDACI or Dox for 24 or 48 hours revealed little change in amino acid levels at 24 hours (Figs. 12A, 12C and 12E), but a significant decrease in levels of several amino acids at 48 hours (Figs. 12B, 12D and 12F), particularly glutamate/glutamine (not discernible by these assays), aspartate/asparagine, and proline. These amino acids are gluconeogenic, meaning that in their metabolism they feed into the TCA cycle. Similar results were obtained in a comparable analysis of H929 cells. These changes occur at a time point in which maximal reduction in glucose uptake is observed, but prior to wholesale induction of apoptosis as demonstrated by annexin V and 7AAD staining of these cells (Fig. 12G and 12H).

The data so far suggested that HDACIs place the cell in a position of metabolic stress and that replenishment of a metabolic resource might reduce or delay the apoptotic effects of HDACIs. Because HDACIs reduce glucose uptake and induce metabolism of amino acids, addition of supplemental glucose predictably did not rescue the cells from apoptosis (data not shown), and thus it was elected instead to supplement the cells with amino acids. OPM2 cells were treated with a low dose of VPA that achieves approximately 50% apoptosis at 96 hours (Fig. 13A). Supplementation of these cells with 0.2mM non-essential amino acids had no effect on cell survival in the absence of HDACI, but potentiated the apoptotic response to VPA. These data suggest that the amino acid metabolism induced in the presence of HDACIs may contribute to the induction of apoptosis. Metabolism of amino acids requires that they initially be deaminated and the ammonia group released either processed into urea or transferred to an acceptor such as alpha-ketoglutarate (cc-KG). To begin to determine whether the cells were stressed through ammonia buildup, RTqPCR was used to analyze mRNA expression of carbamoyl phosphate synthase (CPS) 1, the enzyme responsible of processing free ammonia groups prior to their entry into the urea cycle. In OPM2 cells treated 24 or 48 hours with 0.5-2mM VPA, CPSl mRNA was induced 4-16 fold in a time and dose dependent manner (Fig. 13B), suggesting that the cells may be attempting to adapt to increased ammonia production as a result of HDACI treatment.

Ammonia produced through amino acid metabolism is cleared under physiological circumstances through the urea cycle or through transamination. The urea cycle rarely exists in its entirety outside of the liver and kidney, and RTqPCR analysis of 0PM2 cells treated with VPA did not reveal increased mRNA expression of other enzymes of the urea cycle (data not- shown). Indeed, mRNA expression for ornithine transcarbamylase (OTC), which incorporates carbamoyl phosphate produced by CPSl into the urea cycle, was not detectible (data not shown), indicating that the urea cycle is not functional in these cells. Likewise, no increase was observed in urea cycle intermediates ornithine, citrulline, and arginine; rather, ornithine and citrulline were decreased in OPM2 cells treated 48 hours with increasing doses of HDACI (Figs. 13C, 13E and 13G). Additionally, it was not possible to detect significant urea production in the spent media of cells treated with VPA beyond that originally present in the media (Fig. 13D), and an arginase inhibitor did not cooperate with HDACI to increase apoptosis (data not shown), as would be expected if the cells were utilizing urea synthesis to eliminate amino groups.

If these cells were alternatively using transamination to transfer the amino groups to α-KG to form glutamate or glutamine, observation of an increase, rather than decrease, in levels of glutamate and glutamine would be expected. However, the inability to differentiate glutamate and glutamine levels could mask a significant change in levels of either amino acid, and similarly the cells could excrete these amino acids to the media. To more directly determine whether the cells were utilizing transamination, MS-MS analysis was used to examine the levels of organic acids in 0PM2 cells following treatment with VPA. Little change was observed in the levels of oc-KG following 48 hours of treatment with VPA, further suggesting that transamination is not being extensively employed in these cells (Fig. 13F).

Because ornithine content was reduced and excess amine groups can be incorporated into polyamines, it was hypothesized that the myeloma cells might be relying on polyamiπe production to clear the amino groups released from amino acid metabolism and it was asked whether inhibition of polyamine synthesis would affect the apoptotic response to HDACIs. OPM2 cells were treated 96 hours with or without a low dose of VPA (0.75mM) in the presence or absence of difluoromethyl ornithine (DFMO), a specific inhibitor for ornithine decarboxylase (ODC) 1, the rate limiting enzyme for polyamine synthesis (Fig. 13H). Inhibition of ODCl potentiated the apoptotic effects of VPA but had no effect on cell survival in the absence of HDACI, suggesting that the polyamine pathway is important for mitigating the effects of HDACI, possibly through incorporation of released ammonia groups.

One function of the polyamine synthesis pathway is to mitigate the deleterious effects of reactive oxygen species (ROS), which are produced during oxidative metabolism of fatty acids. The synergy of polyamine synthesis inhibition with HDACI inhibition to induce apoptosis, together with the metabolic shift of the cells to use of fatty acids, raised the possibility that HDACI treatment may be inducing ROS production, which in the absence of DFMO would be ameliorated by polyamine synthesis. To test the viability of this hypothesis,

OPM2 cells were treated 48 hours with VPA or SAHA. During the final 24 hours of incubation, cells were saturated with CM-H2DCFDA, a dye sensitive to ROS production. As compared to the untreated control, HDACI induced an

approximate three fold increase in ROS detection.

In summary, despite an efficient increase in histone acetylation, microarray analysis of patient samples treated with HDACI suggest that only approximately 5% of genes show significant change in expression (Van Lint et al, Gene Expression 5:245-253 (1996), Butler et al, Proc. Natl. Acad. Sci. USA 22:11700-11705 (2002), Glaser et al, Molecular Cancer Therapeutics 2:151-163 (2003)), and comparison of these gene sets indicates only 20% of these genes (about 1% of the genome) are in common. Therefore, it is proposed that that central to their ability to induce apoptosis must be a profound impact on basic cellular processes such as metabolism.

The studies described above indicate that HDACIs specifically inhibit glucose uptake in myeloma cells through both acute and chronic mechanisms. While HDACIs do modulate expression of glycolysis pathway members, notably GLUTl, they also appear to specifically affect the activity of hexokinases independent of their expression. Indeed, the inability of GLUTl overexpression to recover glucose uptake suggests that HDACI inhibition of HXK activity is responsible for the acute effect on glucose uptake and is a key mechanism by which HDACIs influence metabolism, and this inhibition is further compounded secondarily by downregulation of GLUTl expression to result in a chronic inhibition of glucose metabolism. The resulting metabolic void invoked by a profound decrease in glucose uptake is initially filled by oxidation of fatty acids, and then secondarily by catabolism of amino acids. These metabolic effects occur rapidly, prior to significant induction of apoptosis. Blockade of glycolysis at the level of glucose uptake through siRNA knockdown of GLUTl was sufficient in and of itself to induce expression of p21 (Noguchi et al, Cancer Lett. 154:175-182 (2000)), raising the question of how much of the activity displayed by HDACIs toward myeloma cells can be accounted for by a direct effect on transcription of p21 as has been most widely reported. Thus, the contribution of these drugs in harnessing the runaway metabolism of cancer must be considered when describing a mechanism of action for HDACIs in the induction of apoptosis.

The foregoing studies have most extensively utilized SAHA, a newly described HDACI that has been shown to induce apoptosis in a number of types of tumor cells and that has enjoyed some success clinically, as well as valproate, which has been used clinically for decades to reduce seizures, moderate bipolar disorder and alleviate migraine headaches. Despite its successful clinical history, VPA has only recently been described as an HDACI, begging the question of what proportion of its established clinical activities that are not clearly understood mechanistically may be related to HDAC inhibition.

The glycolytic nature of cancer has been well known for several years, and recent studies utilizing transformed cells from several distinct tissues indicate that, despite the inefficiency, the majority of pyruvate produced from glucose metabolized by cancer cells is directed ultimately to synthesis of lactate rather than entering the TCA cycle. Corresponding with their reliance on glucose availability and glycolysis for energy, overexpression of several members of the GLUT family has been described for cancer cells arising from a variety of tissues, notably including lung cancer overexpression of GLUTl . Favorable clinical response has been observed in patients with non-small cell lung cancer following treatment with HDACI, suggesting that GLUTl overexpression may correlate with sensitivity to HDACIs in, a clinical setting. GLUT expression is most likely upregulated as an adaptive response to the ischemic conditions that exist within the solid growing tumor mass. GLUT overexpression and increased glycolysis at the expense of oxidative phosphorylation may in fact provide a protective advantage to cancer cells, as radiation therapy requires the presence of oxygen to induce cytotoxicity. That elevation of glycolysis is a nearly universal property of cancer cells, however, makes it an attractive chemotherapeutic target. It is described here that HDACIs specifically inhibit the activity of GLUTl , the most abundant of the hexose transporters in this myeloma model and also the GLUT family member most widely reported to be overexpressed in cancers arising from of a variety of tissues. While other cancers have been shown to overexpress GLUTl and/or other GLUT family members, the motifs present in GLUTl are highly conserved within some members of the GLUT family and ongoing experiments will determine whether HDACIs may similarly influence glucose uptake in cancer cells from other tissues.

HDACs are subdivided primarily into three classes— zinc dependent class I and II and NAD dependent class III, more commonly called the sirtuins. Class I HDACs are localized to the nucleus, while class II and III HDACs are present in both nucleus and cytoplasm. The sirtuins have previously been linked to metabolism: their activity can be influenced by Akt activity, and Sirtl was shown to deacetylase and increase the activity of ATP synthase. It was initially observed that influence of acetyl CoA levels and later of glucose uptake correlated with influence of class I HDACs, as efficient inhibition of Class I HDACs is a characteristic shared by both VPA and SAHA at these doses. Although class I HDACIs have been observed to influence expression of sirtuins, no effect on glucose uptake or cell survival following treatment with splitomicin, a specific inhibitor of sirtuins (data not shown), suggesting that class III HDAC activity is unlikely to be involved in HDACI inhibition of glucose transport.

One unique property displayed by HDACIs is their relative selectivity for induction of apoptosis in cancer cells while displaying little effect in normal cells. Given the distortion of glucose utilization in cancer versus normal cells, the effect of HDACIs on glucose uptake documented herein may contribute to that selectivity. Because normal cells have a much lower requirement for glycolytic rate, and more efficiently utilize the glucose they metabolize, if HDACIs influenced glucose uptake in normal cells as well as cancer cells, then normal cells may be better able to adapt and survive the "famine". Alternatively,

HDACIs may more efficiently target overexpressed glucose transporters, which would mean that normal cells would effectively fall below the threshold of inhibition. Future experiments will determine whether sensitivity to HDACIs can be correlated with GLUT expression and glucose uptake in transformed and normal cells from both hematological and solid tissues.

Although the mechanisms by which HDACIs induce apoptosis in myeloma cells are not clearly understood, the in vitro observations appear to translate into clinical success. Combination therapies using HDACIs with proteasome inhibitors or with other chemotherapeutics have shown some success. With the interest in combination therapies involving HDACIs, there was a desire to understand the contribution of HDACI-induced metabolic changes in the activities of these chemotherapeutics and thus studies were undertaken to determine whether metabolic pathway supplements or regulators would impinge upon the activities of HDACIs to either increase or decrease their ability to induce apoptosis. The initiation of amino acid metabolism observed when myeloma cells were treated with HDACI led to the examination of pathways that may contribute to clearance of the amino groups liberated during amino acid breakdown. During these studies, it was determined that inhibition of polyamine synthesis had little effect on myeloma cells survival in and of itself (Fig. 13D). However, inhibition of the polyamine pathway during treatment with HDACI resulted in elevated apoptosis as compared to HDACI alone. Treatment of myeloma cells with a polyamine supplement consisting of several of the products of the pathway had no effect on cell survival in the presence or absence of HDACI even at a 10-fold greater polyamine concentration than recommended for cell treatment. Thus, the products of the polyamine pathway, in the event that it is activated during HDACI treatment, are unlikely to be toxic to these cells, and these data suggest that the activity of enzymes (ODC and subsequent) are required, rather than the polyamine products themselves.

The polyamine synthesis pathway has elicited some interest as a chemotherapeutic target itself. The activity of enzymes involved in polyamine synthesis, but not the polyamine products themselves, assist in neutralizing and mitigating the effects of reactive oxygen species (ROS). Because the success of radiation treatment as well as some chemotherapies is influenced by ROS, treatment with DFMO can increase sensitivity of cells to these agents. The synergistic effect on cell survival that was observed with HDAC and polyamine synthesis inhibitors, as well as the documented increase in ROS following HDACI treatment (Fig. 131), suggests that co-administration of HDACI with polyamine synthesis inhibitors (e.g., DFMO— difluoromethyl ornithine and derivatives thereof as well as other agents that inhibit the enzyme ornithine decarboxylase 1 (S-adenosylmethionine decarboxylase (SAMDC)) may increase the efficacy of therapy. Additionally, supplementation of the cells with amino acids during HDACI treatment resulted in increased apoptosis as well. It is suggested that the shift in metabolism from glycolysis to consumption of either acyl carnitines or amino acids results in increased production of ROS, which the polyamine synthesis pathway assists in mitigating. When this pathway is blocked, the metabolic shifts results in buildup of ROS and increased apoptosis. These results raise the possibility that the efficacy of HDACI treatment may be maximized by a high protein, low sugar diet. Intriguingly, historic treatment of seizure disorders now treated with valproate included just such a diet.

One of the challenges complicating development of cancer therapeutics is selection of an in vitro response upon which to optimize further development of derivatives of a promising compound. In the case of HDACIs, the accumulation of acetylated histones has been an obvious benchmark that is easily monitored in any cell line. Results presented here suggest that such an approach may not effectively monitor the cellular responses to HDACIs that are actually producing the desired endpoint (apoptosis). The disparate sensitivities of transformed cell lines in vitro, as well as cancers treated clinically, may instead be influenced by the impact of HDACIs on cellular metabolism. On going work analyzing the contribution of altered glucose metabolism to the overall HDACI sensitivity of cancer cell lines arising from differing origins will further elucidate whether cellular metabolism and glucose uptake can provide a more effective prognostic marker as to the ability of HDACIs to induce apoptosis in a given cancer cell type.

All documents and other information sources cited above are hereby incorporated in their entirety by reference.

Claims

WHAT IS CLAIMED IS:

1. A method of treating multiple myeloma comprising administering to a patient in need thereof an amount of a histone deacetylase inhibitor (HDACI) sufficient to effect said treatment.

2. The method according to claim 1 wherein said patient is a human.

3. The method according to claim 1 wherein said patient is refractory to corticosteroid therapy.

4. The method according to claim 1 wherein said HDACI is a short chain fatty acid, or pharmaceutically acceptable salt thereof.

5. The method according to claim 1 wherein said short chain fatty acid is a C3-C12 fatty acid.

6. The method according to claim 5 wherein said short chain fatty acid is a C₃-Cs fatty acid.

7. The method according to claim 6 wherein said fatty acid is methoxyacetic acid (MAA), butyric acid (BA), valproic acid (VPA), propionic acid, 3-methoxypropionic acid or ethoxyacetic acid.

8. The method according to claim 4 wherein at least two short chain fatty acids, or pharmaceutically acceptable salts thereof, are administered.

9. The method according to claim 8 wherein at least MAA and VPA are administered.

10. The method according to claim 1 wherein said method further comprises administering an arsenic compound.

1 1. The method according to claim 1 wherein said method further comprises administering to said patient an inhibitor of polyamine synthesis.

12. The method according to claim 1 wherein said method further comprises administering to said patient a promoter of reactive oxygen species.

13. The method according to claim 1 wherein said method further comprises administering to said patient an inhibitor of glycolysis.

14. The method according to claim 1 wherein said method further comprises administering to said patient a urea cycle inhibitor.

15. The method according to claim 1 wherein said method further comprises administering to said patient an inhibitor of carbamoylphosphate synthetase.

16. The method according to claim 1, wherein said method further comprises administering to said patient a diet high in protein and low in sugar.

17. A composition comprising a corticosteroid and at least one HDACI or pharmaceutically acceptable salt thereof.

18. A method of predicting the response of a patient to administration of a HDACI comprising obtaining a biological sample from said patient, assaying DNA present in said sample for the presence of a t(4; 14) chromosomal translocation, -wherein the presence of said translocation indicates that the likelihood of said patient responding to treatment for multiple myeloma, which treatment comprises administering at least one HDACI, is at least fifty percent.

19. A method of determining a therapeutically effective dose of HDACI comprising administering to a multiple myeloma patient or a patient suspected of having multiple myeloma a multiplicity of doses of an HDACI, obtaining a biological sample from said patient, and determining, at each of said doses, the level of an aberrant form of FGFR3 present in said sample, wherein said aberrant form of FGFR3 results from a t(4; 14) chromosomal

translocationwherein a therapeutically effective dose is one that down regulates said aberrant form of FGFR3.

20. A method of screening a test compound for suitability as a candidate drug for use in treating multiple myeloma comprising assaying said test compound for the ability to inhibit histone deacetylase (HDAC) activity, . wherein a test compound that inhibits HDAC activity is a candidate drug for use in treating multiple myeloma.