An activated JAK/STAT3 pathway and CD45 expression are associated with sensitivity to Hsp90 inhibitors in multiple myeloma

Huiqiong Lin1,2, Iryna Kolosenko2, Ann-Charlotte Bjo¨rklund, Darya Protsyuk, Anders O¨sterborg, Dan Grander, Katja Pokrovskaja Tammn

Abstract

The molecular chaperone Hsp90 is required to maintain the activity of many signaling proteins, including members of the JAK/STAT and the PI3K pathways. Inhibitors of Hsp90 (Hsp90-Is) demonstrated varying activity against multiple myeloma (MM) in clinical trials. We aimed to determine which signaling pathways that account for the differential sensitivity to the Hsp90-I 17DMAG on a panel of MM cell lines and freshly obtained MM cells. Three CD45þ cell lines with an activated JAK/STAT3 pathway were sensitive to 17DMAG and underwent prominent apoptosis upon treatment, while the majority of CD45 cell lines, that were dependent on the activated PI3K pathway, were more resistant to the drug. Culturing the most resistant cell line, LP1, in the presence of IL-6 resulted in up-regulation of CD45 and pSTAT3, and sensitized to 17DMAGinduced apoptosis, primarily in the induced CD45þ sub-population of cells. The high CD45 expressers among primary myeloma cells also expressed significantly higher levels of pSTAT3, as compared to the low CD45 expressers. Ex vivo treatment of primary myeloma cells with 17DMAG resulted in a stronger caspase3 activation in tumor samples with the prevalence of high CD45 expressers. STAT3 activity was efficiently inhibited by Hsp90-Is in both cell lines and primary cells suggesting an importance of STAT3 inactivation for the pro-apoptotic effects of HSP90-Is. Indeed, over-expression of STAT3C, a variant with an increased DNA binding activity, in U266 cells protected them from 17DMAG-induced cell death. The down-regulation of the STAT3 target gene Mcl-1 at both the mRNA and protein levels following 17DMAG treatment was significantly attenuated in STAT3C-expressing cells, and transient over-expression of Mcl-1 protected U266 cells from 17DMAG-induced cell death. The finding that CD45þ MM cells with an IL-6-activated JAK/STAT3 pathway are particularly sensitive to Hsp90-Is as compared to the low CD45 expressers may provide a rational basis for selection of MM patients amenable to Hsp90-I treatment.

Keywords:
Multiple myeloma
Apoptosis
STAT3
Akt
17DMAG
HSP90

Introduction

Multiple myeloma (MM) is an incurable malignant tumor of B-cell origin. The malignant clone expands in the bone marrow, causing lytic bone lesions, defects in hematopoiesis, hypercalcemia and renal failure. MM is characterized by a heterogeneous genetic background caused by chromosomal translocations and point mutations, as well as by epigenetic modifications. In addition, the bone marrow microenvironment plays an important role in supporting growth, survival and migration of MM cells [1]. MM adhesion to the bone marrow stromal cells triggers the secretion of cytokines, such as IL-6. This factor in turn activates the JAK/STAT signaling that plays a fundamental role in proliferation and survival of MM cells. IL-6 is one of the major survival factors for MM as malignant cells are commonly dependent on a paracrine or autocrine production of this growth factor [2]. Furthermore, IL-6 signaling also confers resistance to druginduced apoptosis of MM cells [3]. These survival effects can be attributed, at least in part, to the IL-6 responsive gene STAT3, since abrogation of STAT3 leads to death of MM cells [4].
Another major pathway activated in MM is that of phosphoinositide 3-kinase (PI3K), leading, in turn, to the phosphorylation and activation of the Akt kinase, which promotes survival signaling. The PI3K pathway is activated through the growth factor IGF-1 as well as through the inactivation of a negative regulator of the PI3K pathway, the PTEN phosphatase. A number of studies have demonstrated that inhibition of the PI3K/Akt pathway leads to cell death in MM, indicating that it might represent a valid therapeutic target in this disease [5,6].
An increased expression of Hsp90 (90 kD heat shock protein) also contributes to MM tumor cell survival and is associated with poor prognosis [7,8]. Hsp90 is an abundant, highly conserved cellular chaperone that acts as a key component of a multiprotein chaperone complex that regulates folding, maturation and stabilization of a large group of ‘‘client’’ proteins [7]. These include oncogenic kinases and growth factors/cytokine receptors, such as IL-6 associated gp130, IGF-1R, Akt and JAK kinases. Therefore inhibition of Hsp90 affects all of these pathways. Moreover, Hsp90 inhibition can abrogate the protective effects of bone marrow stroma, which normally triggers these survival/ proliferative pathways [7]. Hsp90 inhibitors (Hsp90-Is) have, to date, demonstrated a variable antitumor response, either alone or in combination with other drugs in pre-clinical and clinical studies in MM. It remains unclear on what basis one should select myeloma patients for Hsp90-I-based therapies.
One reason for the differential response to Hsp90-Is may be that different oncogenic pathways are dysregulated in the individual malignant clones. Interestingly, MM cell lines and primary myeloma are found to fall into two major biological groups: either dependent or independent on the activity of the Akt kinase for their survival [9]. Further studies have shown that the activity of the PI3K pathway is almost uniformly associated with a CD45negative (CD45) MM phenotype [10]. CD45 is a trans-membrane phosphatase required for lymphocyte development and its expression is induced by IL-6 [11]. Treatment of a CD45 MM cell line with IL-6 resulted in the induction of CD45 expression and inhibition of Akt phosphorylation, demonstrating that CD45 inhibits the PI3K pathway [10]. Furthermore CD45 MMs were found to be particularly sensitive to the inhibition of the IGF-1 signaling pathway [12] while CD45þ MM—to the inhibition of the JAK pathway [13]. Thus, taken together, there are two defined subgroups within MM tumors: CD45þ with an activated IL-6/JAK/ STAT pathway and, CD45 with an activated PI3K pathway and an inactive JAK/STAT pathway.
We aimed to determine whether cells dependent on any of these pathways are particularly sensitive to the inhibition of Hsp90 in order to identify predictive biomarkers of a favourable response to these drugs in MM. Our results on cell lines and primary cells showed that CD45þ MM cells with an activated JAK/STAT3 pathway are particularly sensitive to 17DMAG and that inactivation of STAT3 plays an important role in the Hsp90I-mediated cytotoxic effects. This suggests that the IL-6-induced signalling represents an ‘‘Achilles heel’’ in the treatment strategy of MM with Hsp90-Is.

Materials and methods

Cell lines, culture conditions and drug treatment

Human MM cell line U266 [14] and U266 transfected with pCMV-neo or with STAT3C [15], RPMI8226 and L363 (from Dr. P. Marchetti, Faculte de M edecine, Universit e de Lille, France), LP-1, AMO-1, OPM-2 (from the DSMZ, Braunschweig), KMS12-PE and INA-6 [16](from Prof. M. Gramatzki, Division of stem cell transplantation, 2nd department of medicine, Kiel, Germany) were used in this study. All cell lines were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS), L-glutamine and streptomycin/penicillin in a 37 1C incubator with 5% CO2. For INA-6,500 U/ml IL-6 (PeproTech) were added into the medium according to the instructions from the supplier. All the cell lines were validated using genotyping with 9 chromosomal markers (Uppsala genotyping service). The cell lines were maintained in concentration between 2 105 and 1 106 cell/ml. The drugs 17DMAG, Ly294002 and Pyr6 [17] (Calbiochem) were diluted, stored and used according to the instructions. 10 mM of Ly294002 and 1 mM of Pyr6 were used as previously described [17,18]. The efficiency of these drugs was monitored by Western blotting with anti-phospho-Akt and antiphopsho-STAT3 antibodies, respectively.

Patients and primary myeloma cells

Bone marrow samples from patients with myeloma were obtained from the Karolinska University Hospital Solna, Stockholm, Sweden. The clinical characteristics are shown in Table 3. All patients had a confirmed diagnosis of MM and all but 3 were previously untreated and the clinical staging was done according to the Myeloma International Staging System (ISS [19]). The study was approved by the regional ethics committee, and all patients gave their informed consent in accordance with the Declaration of Helsinki. Bone marrow mononuclear cells were purified using LymphoprepTM (Axis-Shield PoC AS, Oslo, Norway).10-5 106 mononuclear cells were immediately fixed while 10–30 106 cells were further purified using anti-CD138Ab-coated magnetic micro-beads (MiltenyiBiotec). After purification, cells were resuspended at 106/ml in the RPMI medium supplemented as for the cell lines, cultured without or with 500 nM 17DMAG for 24 h and fixed. Fixation and stainings were performed as described in [20] using following antibodies, all from Becton Dickinson: CD45-V450 (clone 2D1), CD138-PerCP Cy5,5 (clone MI 15), pSTAT3-Y705Alexa 647 (clone 4/P-STAT3), Active Caspase3-FITC (clone 51-68654X) and pAkt-Ser473-Alexa 488 (clone M89-61). The following panel was used for cells that were fixed directly after mononuclear cell preparation: CD45-V450, CD138-PerCP Cy5,5, pSTAT3-Alexa 647 and pAkt-Alexa 488. Another panel was used for CD138þ selected cells and cultured ex vivo for 24 h: CD45V450, pSTAT3-Alexa 647, Active Caspase3-FITC with or without CD138-PerCP Cy5,5. The gates were set up using Fluorescence Minus One (FMO) method as shown in Supplementary Fig. 2. The data were obtained and analyzed on a LSRII Flow Cytometer and a FACSDiva software, version 6.1.2 (Becton Dickinson). Student ttest, two-sample unequal variance and two-or one-tailed distribution, as indicated, were used for statistical analysis.

Transient transfection

Amaxa Nucleofector Kit C and X-05 program were used for transfection according the manufacturer’s instructions. 2 106 cells and 3 mg of DNA (1 mg of EGFP and either 2 mg of plasmids Mcl-1 or pSG5) were used. 24 h after transfection cells were diluted into 2 105 cells/ml and treated with 200 nM 17DMAG for another 24 h.

Assays for cell viability and apoptosis

The effects of the indicated inhibitors monitored using WST-1 viability assay (see results in Supplementary Fig. S1C and D) are presented here as percent of viable cells after 24 h (Pyr6) or 72 h (Ly294002) of treatment. Summary of the expression levels of pSTAT3 and pAkt from Supplementary Fig. S1A, B and Fig. 2C–E.
Cell viability was determined by WST-1 assay (Roche) according to the manufacture’s instruction. Briefly, cells were diluted into 2 105 C/ml in 96-well plates and 24 h later treated with 200 nM of 17DMAG (Calbiochem) for 24, 48 or 72 h. The OD values were measured in a microplate reader (BIO-RAD Model 550) at the wavelength of 450 nm. The Caspase 3/7 assay (Promega) was performed according to the manufacture’s instruction and luminescence was measured in Centro LB 960 luminometer, BERTHOLD Technologies. FITC-Annexin V/PI stainings (Roche) were performed according the manufacture’s instruction and analyzed by a FACS Calibur (Becton Dickinson) and Cell Quest Pro software. For the double CD45/Annexin V staining on LP1 cells cultured with or without IL-6, cells were first incubated with CD45-APC (Becton Dickinson), washed once and then preceded to the FITC-Annexin V staining. Live cell population was gated for the CD45þ and CD45 and percent of Annexin V-positive cells was recorded in these gates. Data were obtained and analyzed on a LSRII Flow Cytometer and a FACSDiva software, version 6.1.2 (Becton Dickinson). Student t-test (two-tailed, two-sample unequal variance) was used for calculations of statistical significance.

Protein extraction, western blotting and antibodies

Total proteins were prepared and Western blotting was performed as described[21].The antibodies used were against: STAT3, p-Y705-STAT3, Akt, p-S473-Akt, JAK1 and Mcl-1, all from Cell Signaling; HSP90 and HSP70 from Santa Cruz; b-actin from Sigma, Bcl-2 from Dakopatts AB, Bcl-xL from Becton Dickinson and anti-flag from Stratagene. Secondary antibodies were: HRP-conjugated anti-mouse and anti-rabbit from Cell Signaling Technology. The proteins were detected using ECL solution (PerkinElmer).

Electrophoretic mobility shift assay (EMSA)

EMSA was performed using nuclear extracts and hSIE probe (50-GTCGACATTTCCCGTAAATC-30) as previously described [15]. For super shifts, 1 mg of anti-flag antibody (Stratagene) was added after 10 min of incubation of extracts with the labeled probe, and incubated for additional 20 min at RT before separation on PAGE.

Quantitative real time PCR

Total RNA was extracted from 2 106 cells using TRI Reagent (Ambion) following the manufacturer’s instructions. cDNA was synthesized from 1 mg of total RNA using SuperScript II Reverse Transcriptase (Invitrogen). Real-time PCR was run on a 7500 Real-Time PCR system using SYBR GREEN PCR Master Mix (Applied Biosystems). b-actin was used as control. The Mcl-1 primers were: Mcl1-F 50-GTGGCTAAACACTTGAAGACC-30, Mcl1-R 50-GAAGAACTCCACAAACCCATCCC-30. Each sample was tested in triplicate. The expression changes of Mcl-1 were analyzed using standard curve method.

Results

Characterization of myeloma cell lines

As described in the Introduction, the importance of a number of signaling pathways has been recognized in multiple myeloma (MM). These include the JAK/STAT3, activated by the cytokine IL-6, and the PI3K/Akt pathway, being mainly activated through the IGF-1R. We characterized eight MM cell lines with regard to the expression of Hsp90, the myeloma cell marker CD138 [1], the trans-membrane phosphatase CD45, and the activity as well as the dependence of the cells on the JAK/STAT3- and the PI3K/Akt pathways (Table 1 and Supplementary Fig. S1). The levels of Hsp90 were similar in all eight cell lines (Supplementary Fig. S1B), and all the cell lines were positive for CD138. Three of them were CD45þ while the other five were CD45 (Table 1). Notably, the majority of CD45þ cell lines exhibited high or detectable levels of pSTAT3 and were sensitive to the JAK2 inhibitor Pyr6 (U266, AMO1 and INA6), but were significantly more resistant to the inactivation of the PI3K/Akt signaling. Vice versa, cell lines highly sensitive to the PI3K inhibitor Ly294002 (LP1, OPM2, RPMI8226) were resistant to the inactivation of the JAK/STAT signaling (Table 1; Supplementary Fig. S1; Fig. 2C and E). These data are in line with previously published studies that described two distinct myeloma subgroups, Akt-dependent and Aktindependent [9] and the mutually exclusive activity of the JAK/STAT and the PI3K/Akt pathways in MM due to the action of the transmembrane phosphatase CD45 [10]. The L363 and KMS12PE cell lines, although CD45 and pSTAT3-negative, were found to be dependent on the both signaling pathways (Supplementary Fig. S1C and D).

Differential sensitivity of multiple myeloma cell lines to 17DMAG

Annexin V positivity and caspase 3 activation as a measurement of apoptotic cell death after 17DMAG treatment: summary of the data from Fig. 2(A and B) presented as fold change compared to control. We next determined the sensitivity (IC50) of the eight MM lines to the Hsp90-I, 17DMAG, using the WST-1 viability assay (Table 2). Four cell lines, three CD45þ and one CD45 L363, all dependent on the JAK/STAT3 pathway, had an IC50 between 140 and 750 nM of 17DMAG. The four CD45 lines, dependent on the PI3K pathway, had an IC50 of more than 2000 nM of the drug (Table 2). LP1 cells were highly resistant to treatments up to 5 mM during 24 h, and showed only a 25% reduction in viability
after 48 h of treatment with this dose (data not shown). We next determined whether loss of cell viability was due to induction of cell death.17DMAGinduced prominent cell death in all the CD45þ cell lines and L363, also dependent on the JAK/STAT3 pathway, as monitored by Annexin V staining (Fig. 1A) and using a caspase 3 assay (Fig. 1B). In contrast, both assays detected only a weak induction of cell death in the other four CD45 lines after 24 h of treatment with 200 nM of 17DMAG. Caspase 3 activation was significantly different between the two groups, either dependent (all three CD45þ lines, L363 and KMS12PE) or independent (LP1, OPM2, RPMI8226) on the JAK/STAT pathway (p¼0.03). The difference between the CD45þ (U266, INA6, AMO1) and CD45 (L363, KMS12PE, LP1, OPM2, RPMI8226) sub-groups was also statistically significant (Figs. 1B, np¼0.05), further suggesting that CD45þ cell lines and/or dependent on the JAK/STAT3 pathway are more sensitive to 17DMAG-induced cell death as compared to the CD45 cell lines.
In order to monitor the effects of 17DMAG we analyzed the expression of pSTAT3,pAkt and the total levels of these proteins in accordance with JAK and Akt being known Hsp90 clients [8,22]. Hsp70 protein levels were also monitored as an indication of the activity of the Hsp90 inhibitor, and were induced to a similar extent in all cell lines following 17DMAG treatment ([8,23], data not shown). The levels of p-Y705-STAT3 correlated well with the presence of IL-6: high in INA-6 that requires addition of IL-6 in the medium [16] and in U266 with an autocrine IL-6 production [15]. pSTAT3 levels, however, were low in AMO-1 and L363 cells, and not detectable in KMS12PE cells that were nevertheless dependent on the JAK/STAT pathway activity (Supplementary Fig. S1 and Fig. 1). 17DMAG potently inhibited pSTAT3 in all cell lines with detectable pSTAT3 (Fig. 1C–E), and also down-regulated the levels of pAkt or Akt protein as previously described [8,24]. This occurred in both sensitive and resistant lines. These data confirmed that Hsp90 inhibition compromises the JAK/STAT3 and the Akt pathways. However, no direct correlation with the inhibition of these protein’s expression and 17DMAG-induced cell death could be established.

Activation of JAK/STAT3 by IL-6 induced CD45 and sensitized to 17DMAG-induced cell death

Since we found that CD45 expression and dependence on the A representative of two independent experiments is shown. (B). MM cell lines were cultured in the absence or presence of 17DMAG as in A. and apoptosis was assessed by measuring caspase 3 activity. Data are presented as fold change compared to control. np¼0.05 for CD45þ vs. CD45 (Table 1). (C)–(E) The indicated MM cell lines were cultured in the absence or presence of 200 nM of 17DMAG for the indicated time and the levels of phosphorylated forms and of total proteins of STAT3 and Akt were analyzed by Western blotting. b-actin was used as a loading control. observation that prolonged IL-6 stimulation of the CD45 LP1 cell line resulted in the induction of CD45 expression [10]. Indeed, culturing LP1 cells (the most resistant to 17DMAG among the MM cell lines analyzed) in medium with the addition of 20 ng/ml of IL-6 for 1–3 months led to the up regulation of CD45 in 15–30% of the cells (Fig. 2A, ‘‘LP1þIL6’’, lower panel. CD45þ U266 and CD45 LP1 cells cultured in normal medium were used for comparison (Fig. 2A, upper and middle panel). Treatment of LP1þIL-6 cells with 17DMAG for 24 and 48 h resulted in a significantly increased cell death (Fig. 2B). Interestingly, it was primarily CD45þ cells that contributed to the increased apoptotic cell population (Fig. 2C and D). STAT3 phosphorylation was, as expected, induced by IL-6 and strongly inhibited by 17DMAG (Fig. 2E). Although pAkt levels did not differ between LP1 and LP1þIL-6, the inhibition of total Akt levels by 24 h of 17DMAG treatment was more complete in the LP1þIL-6 as compared to the parental LP1. This experiment confirmed that CD45 positivity and JAK/STAT3 activity are associated with sensitivity to Hsp90-I-induced cell death of MM cells. Importantly, it also showed that resistance of LP1 cells to these drugs is reversible. Finally, it further substantiated the notion that IL-6-induced survival pathways are vulnerable to the inhibition of Hsp90.

CD45-high myeloma primary cells express pSTAT3 and are sensitive to 17DMAG

There are two defined sub-populations within MM tumors: cells with high or low CD45 expression [25]. CD45 is induced by IL6 and plays a role in determining signaling and proliferation in human myeloma in response to growth factors [26] and is a marker of both the early stages of the disease and a better treatment outcome [27,28]. It was also shown that the JAK inhibitor preferentially kills CD45þ primary myeloma cells [13] in line with the notion that these cells are dependent on the JAK/STAT pathway.
First, we analyzed the levels of pSTAT3 and pAkt in the mononuclear cells from 10 myeloma patients (Table 3) using Phosphoflow and gating (see M&M and Supplementary Fig. S2). Cells were gated according to their CD138 expression and, when plotted against CD45, revealed two distinct populations: CD45-high and CD45-low (Supplementary Fig. S2 and Fig. 3A). The percent of pSTAT3-positive cells was significantly higher in the CD45-high cells (Fig. 3B left chart). Vice versa, the percent of pAkt-positive cells was higher in the CD45-low expressers; however, the difference was not statistically significant (Fig. 3B right chart). These stainings also revealed the predominance of either CD45-high or CD45-low cell compartment in each of the patient samples (Table 3) and, thus, allowed to divide the samples into two sub-groups. We next tested the sensitivity of primary MM cells of these sub-groups to 17DMAG. The CD138þ cells selected using magnetic beads were cultured ex vivo with or without 17DMAG for 24 h. pSTAT3 and active caspase 3 were then analyzed as markers of 17DMAG pro-apoptotic effects. pSTAT3, which diminished somewhat upon in vitro culturing per se, was nevertheless potently down-regulated upon 17DMAG treatment independently of CD45 status (Fig. 3C left chart). Active caspase 3 was induced to a higher degree in the samples with the predominance of CD45-high myeloma cells as compared to the samples containing mostly CD45low myeloma cells (Fig. 3C right chart). These data, although on a limited number of samples, thus indicated that also primary MM cells with a high expression of CD45 and an activated JAK/STAT pathway are more sensitive to the Hsp90-I 17DMAG, as compared to CD45-low MM cells.

Constitutively active STAT3C protects U266 cells from 17DMAG-induced cell death

STAT3 phosphorylation was potently inhibited in both cell lines and primary cells following treatment with17DMAG. In order to investigate the role of inhibition of STAT3 phosphorylation/ activity in the induction of cell death by Hsp90 inhibition, we used clones of U266 cells that stably express STAT3C, a mutant form of STAT3 that possesses stronger DNA binding activity and acts as a constitutively active STAT3 protein [15]. Both parental U266 cells and the stably transfected clones uniformly expressed CD45 (Fig. 2A and data not shown). Parental U266 cells, two neomycin-resistant clones (control group) and three STAT3C expressing clones, 1:16, 2:7 and 2:3 (test group) were used for these experiments (Fig. 4A). The protein levels of the JAK1 kinase, a known Hsp90 client protein and of pSTAT3 dropped upon 16 h of 17DMAG treatment in both the control and the STAT3Cexpressing clones (Fig. 4B). Both the caspase 3 assay (Fig. 4D; p¼0.009) and Annexin V stainings (Fig. 4C; p¼0.03) showed that the STAT3C expressing clones were more resistant to 17DMAG, as compared to the control clones (the p-values are calculated for the difference in response between the two groups). Thus, STAT3C protects U266 myeloma cells from Hsp90-I-induced cell death.

Down-regulation of Mcl-1by 17DMAG is attenuated in the presence of STAT3C

We then aimed to determine the molecular mechanisms of STAT3C-mediated protection from 17DMAG-induced cell death in U266 cells. To this end, we tested the kinetics of changes in protein levels of several STAT3-responsive genes in the control and STAT3C transfected clones upon 17DMAG treatment. Induction of apoptosis monitored in this experiment as the cleavage of caspase 3 was pronounced in the neo1 control clone but not in the STAT3C-expressing clone 2:3 (Fig. 5A), in line with the data in Fig. 4D for the 24 h time point. The levels of Bcl-XL, Bcl-2 and survivin were not significantly affected by the treatment (data not shown). However, the levels of Mcl-1, another STAT3 target, were prominently down regulated already after 6 h of 17DMAG treatment in the neo1 control clone (Fig. 5A). 17DMAG treatment caused a decrease in Mcl-1protein levels also in the STAT3Cexpressing cells; however, the remaining Mcl-1 levels were higher in these cells as compared to the neo1 control (Fig. 5A). Mcl-1 was down regulated by 17DMAG at the mRNA level as early as 6 h of 17DMAG treatment as shown by qRT-PCR (Fig. 5B).
The relative decrease of Mcl-1 mRNA was about 50% lower in the STAT3-expressing 2:3 clone as compared to the neo1 control (Fig. 5B). Thus, STAT3C might counteract Mcl-1 down-regulation at the transcriptional level [29].
We have previously shown that STAT3C elicits stronger and prolonged DNA binding as compared to the wt STAT3 protein [15]. Although the levels of pSTAT3 were equally inhibited in the control and the STAT3C-expressing clones (Fig. 4B), the DNA binding of STAT3C might be prolonged and thereby support Mcl-1 gene transcription. In support of such a hypothesis, EMSA showed that although DNA binding was inhibited in both clones upon 17DMAG treatment, the remaining levels of DNA-bound STAT3 were higher in the STAT3C-expressing clone as compared to the neo1control (Fig. 5C; compare lanes 3 and 7). A supershift using anti-FLAG antibodies showed that it is indeed the Flagtagged STAT3C that is the remaining STAT3 bound to DNA after 17DMAG treatment (Fig. 5C, compare lanes 7 and 8. Note the complete disappearance of the band in lane 8).
The Mcl-1 protein has been strongly implicated in the survival of MM cells and drug resistance [30]. Indeed, also in case of 17DMAGinduced cell death, transient Mcl-1 over-expression rendered U266 cells more resistant as compared to the vector control transfected cells (Fig. 5D). We have previously noted that STAT3C overexpression also resulted in higher basal levels of another antiapoptotic protein, Bcl-2 in the 2:3 and 2:7 clones (Fig. 5A and [15]. We therefore asked whether Bcl-2 over expression could protect U266 cells from 17DMAG-induced cell death. Subclones of U266 cells stably transfected with Bcl-2 [31] were used for this purpose. Indeed, Bcl-2 over-expression (clones 7 and 8) largely protected from 17DMAG-induced cell death to the same extent as STAT3C over-expression (clone 2:3, Fig. 5E). Thus the protective effect of STAT3C on 17DMAG-induced cell death is likely mediated through a concerted action of STAT3 downstream targets such as the antiapoptotic proteins Mcl-1 and Bcl-2.

Discussion

The identification of subgroups of patients that show greater or lesser susceptibility to particular anti-cancer therapies is a critical factor in the development of an individually targeted treatment strategy. In MM the activity of the PI3K pathway is associated with a CD45 phenotype [10,12], whilst IL-6 upregulates CD45 expression [11] and, thus, the activity of the JAK/ STAT3 pathway is associated with the CD45þ phenotype. Intrigued by the existence of these two distinct subgroups of MM, we asked whether they might be differentially sensitive to the Hsp90-I 17DMAG. A hint that this might be the case came from one previous study, where all of the MM cell lines resistant to another Hsp90-I, NPV-AUY922, had undetectable pSTAT3 levels [8]. We found that all of the MM cell lines analyzed in our study expressed very similar levels of Hsp90. However, cell lines dependent on the JAK/STAT pathway died by apoptosis while the majority of the cell lines dependent on the PI3K pathway were more resistant to 17DMAG. The finding is particularly interesting given that the JAK and Akt kinases are both Hsp90 clients and are both clearly affected by Hsp90 inhibition [8,22].
It is well established that IL-6 signaling that activates STAT3driven transcription is an important survival factor for MM in the bone marrow. IL-6 signaling can also protect MM cells from cell death induced by glucocorticoids, important drugs used in the treatment of this disease [3]. Therefore the finding that IL-6 cannot protect from Hsp90-I-induced cell death [23] and, furthermore, that MM cells dependent on the JAK/STAT signaling are particularly sensitive to Hsp90-Is is very important. It has been reported that inactivation of the JAK/STAT pathway by Hsp90-Is is associated with apoptotic cell death [8,22]. In this study we further show that inhibition of STAT3 is a key event in the pro-apoptotic activity of Hsp90-Is, since over-expression of STAT3C, a genetically manipulated form of STAT3 with higher affinity to DNA [15,32], protected from 17DMAG-induced apoptosis. STAT3 normally drives expression of anti-apoptotic Bcl-2 family members, including Mcl-1 [4]. Indeed, Mcl-1 was downregulated already at the mRNA level following treatment with the Hsp90-I, and this effect was attenuated by STAT3C overexpression. In line with these data, over-expression of Mcl-1 could partially rescue cells from 17DMAG-induced cell death. Interestingly, another anti-apoptotic protein, Bcl-2 was not down regulated by 17DMAG in U266 cells; nevertheless its overexpression also rendered resistance to this drug. These data could be explained by the overlapping and cooperative functions of the anti-apoptotic proteins of the Bcl-2 family, which are not exclusively regulated by STAT3 but also by other transcription factors.
Although the Akt kinase, a crucial player in the PI3K pathway and a known Hsp90 client, was down-regulated in the 17DMAGtreated MM cells, this did not result in loss of cell viability or cell death. One prominent example is the most resistant cell line, LP1, where both Akt phosphorylation and Akt protein levels were inhibited by the drug treatment. Inhibition of PI3K by Ly294002, however, resulted in a loss of cell viability in this cell line (mainly through induction of growth arrest), being in line with other studies [9],[33][5,6][10]. One could speculate that although efficient (especially at 24 h, Fig. 2E), inhibition of Akt by 17DMAG might be reversible, while Ly294002 that inhibits upstream signaling from PI3K has a more substantial and prolonged effect on this pathway. It remains to be clarified whether involvement of a signaling cross-talk might underlie the discordance of these results. Nevertheless, cell lines dependent on the PI3K pathway were more resistant to the inhibition of Hsp90, and, thus, activation of this pathway might represent a potential cause for the resistance of MM to Hsp90-Is.
The final maturation of plasma cells (PCs) takes place in the bone marrow, which plays a crucial role in the differentiation process characterized by a gradual decrease in proliferation and CD45 expression, and the up-regulation of Bcl-2. Just like normal bone marrow PCs, MM cells consist as both CD45þ proliferating and CD45 (and low) cells, which are less or non-proliferating. In addition, there is also a distinct compartment of MM cells that are CD45 (or low) and highly proliferative. These cells might be the most resistant to treatment in general, since patients with an elevated CD45 compartment have a poorer outcome[25,27,28]. In line with this, we also detected both CD45-high and CD45-low compartments among CD138þ primary MM cells. The presence of pSTAT3 dominated in the CD45-high compartment in line with both CD45 and STAT3 being activated by IL6 signaling. There was also a clear predominance of either of these two compartments in each patient sample, which allowed us to divide them into two distinct groups. Ex vivo treatment with 17DMAG showed that more cell death was observed in the group of samples with a predominance of the CD45-high compartment. This was in line with our data on cell lines with one interesting exception, the CD45 cell line L363, that was dependent on both PI3K and JAK/STAT3 signaling and was also highly sensitive to 17DMAG. These cells might represent an intermediate phenotype, and we are currently investigating other signaling molecules involved in proliferation and survival of these cells. This finding should be followed up in the future by clinical studies comparing the response to Hsp90-Is in the groups of MM patients with differential predominance of CD45-high vs. CD45-low compartments, respectively. Interestingly, we also found that the resistant phenotype is reversible: culturing of CD45 LP1 cells in medium supplemented with IL-6 (LP1þIL-6) led to the up regulation of CD45 and clearly sensitized these resistant cells to 17DMAG-induced cell death. In the present study, we did not address the exact mechanism of this sensitization. It has been demonstrated that the CD45 phosphatase (induced by IL-6) inhibits Akt phosphorylation activated by IGF-1 stimulation [10]. Although we did not observe any significant reduction in the basal phospho- or total Akt levels in LP1þIL-6 cells, its down regulation upon 17DMAG treatment was more complete as compared to the regular LP1 cells. Thus, IL-6 stimulation might facilitate inactivation of pAkt by Hsp90-Is. It was also shown that CD45þ MM cells are more sensitive to various apoptotic stimuli such as oxidative- and ER-stress, than CD45 MM [34]. This phenomenon was explained by the increased expression of VDAC1 in cells that express the CD45 phosphatase. Interestingly, one of the multiple effects of Hsp90-Is in MM cells is the induction of ERstress due to accumulation of misfolded proteins [7], and, therefore, the correlation between CD45 expression and sensitivity to the Hsp90-Is could be explained by this alternative mechanism.
In conclusion, by analyzing a panel of MM cell lines and primary cells from 10 MM patients, we found that cells dependent on IL-6-activated signaling (predominantly CD45þ), are particularly sensitive to Hsp90-Is, and that resistant CD45 MM cells can be sensitized to these drugs by prolonged IL-6 stimulation. We further showed that the sensitivity of the IL-6 induced pathway to 17DMAG is invariably translated into inhibition of STAT3 phosphorylation in primary cells and that inactivation of STAT3 and its down-stream targets is a critical event for Hsp90Is-induced cell death. In contrast, an activated PI3K pathway might represent a cause for resistance to Hsp90-Is in this disease. These findings are of clinical interest since they may help in stratifying MM patients for future individualized and targeted treatments.

Role of the funding source

This study was supported with grants from Karolinska Institutet, the Cancer Society of Stockholm, the Swedish Research Council and the Swedish Cancer Foundation. The funding sources were not involved in study design, collection, analysis and interpretation of data, in the writing of the report or in the decision to submit the paper for publication.

References

[1] T. Hideshima, C. Mitsiades, G. Tonon, P.G. Richardson, K.C. Anderson, Understanding multiple myeloma pathogenesis in the bone marrow to identify new therapeutic targets, Nat. Rev. Cancer 7 (2007) 585–598.
[2] M. Kawano, T. Hirano, T. Matsuda, T. Taga, Y. Horii, K. Iwato, H. Asaoku, B. Tang, O. Tanabe, H. Tanaka, A. Kuramoto, T. Kishimoto, Autocrine generation and requirement of BSF-2/IL-6 for human multiple myelomas, Nature 332 (1988) 83–85.
[3] M.A. Frassanito, A. Cusmai, G. Iodice, F. Dammacco, Autocrine interleukin-6 production and highly malignant multiple myeloma: relation with resistance to drug-induced apoptosis, Blood 97 (2001) 483–489.
[4] R. Catlett-Falcone, T.H. Landowski, M.M. Oshiro, J. Turkson, A. Levitzki, R. Savino, G. Ciliberto, L. Moscinski, J.L. FernandezLuna, G. Nunez, W.S. Dalton, R. Jove, Constitutive activation of Stat3 signaling confers resistance to apoptosis in human U266 myeloma cells, Immunity 10 (1999) 105–115.
[5] T. Hideshima, L. Catley, N. Raje, D. Chauhan, K. Podar, C. Mitsiades, Y.T. Tai, S. Vallet, T. Kiziltepe, E. Ocio, H. Ikeda, Y. Okawa, H. Hideshima, N.C. Munshi, H. Yasui, P.G. Richardson, K.C. Anderson, Inhibition of Akt induces significant downregulation of survivin and cytotoxicity in human multiple myeloma cells, Br. J. Haematol. 138 (2007) 783–791.
[6] J. Zhang, Y. Choi, B. Mavromatis, A. Lichtenstein, W. Li, Preferential killing of PTEN-null myelomas by PI3K inhibitors through Akt pathway, Oncogene 22 (2003) 6289–6295.
[7] P.G. Richardson, C.S. Mitsiades, J.P. Laubach, S. Lonial, A.A. Chanan-Khan, K.C. Anderson, Inhibition of heat shock protein 90 (HSP90) as a therapeutic strategy for the treatment of myeloma and other cancers, Br. J. Haematol. 152 (2011) 367–379.
[8] T. Stuhmer, A. Zollinger, D. Siegmund, M. Chatterjee, E. Grella, S. Knop, M. Kortum, C. Unzicker, M.R. Jensen, C. Quadt, P. Chene, J. Schoepfer, C. Garcia-Echeverria, H. Einsele, H. Wajant, R.C. Bargou, Signalling profile and antitumour activity of the novel Hsp90 inhibitor NVP-AUY922 in multiple myeloma, Leukemia 22 (2008) 1604–1612.
[9] A. Zollinger, T. Stuhmer, M. Chatterjee, S. Gattenlohner, E. Haralambieva, H.K. Muller-Hermelink, M. Andrulis, A. Greiner, C. Wesemeier, J.C. Rath, H. Einsele, R.C. Bargou, Combined functional and molecular analysis of tumor cell signaling defines 2 distinct myeloma subgroups: Akt-dependent and Aktindependent multiple myeloma, Blood 112 (2008) 3403–3411.
[10] G. Descamps, C. Pellat-Deceunynck, Y. Szpak, R. Bataille, N. Robillard, M. Amiot, The magnitude of Akt/phosphatidylinositol 30-kinase proliferating signaling is related to CD45 expression in human myeloma cells, J. Immunol. 173 (2004) 4953–4959.
[11] H. Ishikawa, N. Tsuyama, S. Abroun, S. Liu, F.J. Li, O. Taniguchi, M.M. Kawano, Requirements of src family kinase activity associated 17-DMAG with CD45 for myeloma cell proliferation by interleukin-6, Blood 99 (2002) 2172–2178.
[12] G. Descamps, S. Wuilleme-Toumi, V. Trichet, C. Venot, L. Debussche, T. Hercend, M. Collette, N. Robillard, R. Bataille, M. Amiot, CD45neg but not CD45pos human myeloma cells are sensitive to the inhibition of IGF-1 signaling by a murine antiIGF-1R monoclonal antibody, mAVE1642, J. Immunol. 177 (2006) 4218–4223.
[13] V. Ramakrishnan, T. Kimlinger, J. Haug, M. Timm, L. Wellik, T. Halling, A. Pardanani, A. Tefferi, S.V. Rajkumar, S. Kumar, TG101209, a novel JAK2 inhibitor, has significant in vitro activity in multiple myeloma and displays preferential cytotoxicity for CD45þ myeloma cells, Am. J. Hematol. 85 (2010) 675–686.
[14] K. Nilsson, H. Bennich, S.G. Johansson, J. Ponten, Established immunoglobulin producing myeloma (IgE) and lymphoblastoid (IgG) cell lines from an IgE myeloma patient, Clin. Exp. Immunol. 7 (1970) 477–489.
[15] L. Thyrell, V. Arulampalam, L. Hjortsberg, M. Farnebo, D. Grander, K. Pokrovskaja Tamm, Interferon alpha induces cell death through interference with interleukin 6 signaling and inhibition of STAT3 activity, Exp. Cell Res. 313 (2007) 4015–4024.
[16] R. Burger, A. Guenther, F. Bakker, M. Schmalzing, S. Bernand,W. Baum, B. Duerr, G.M. Hocke, H. Steininger, E. Gebhart, M. Gramatzki, Gp130 and ras mediated signaling in human plasma cell line INA-6: a cytokine-regulated tumor model for plasmacytoma, Hematol. J. 2 (2001) 42–53.
[17] L. Pedranzini, T. Dechow, M. Berishaj, R. Comenzo, P. Zhou, J. Azare, W. Bornmann, J. Bromberg, Pyridone 6, a pan-Janusactivated kinase inhibitor, induces growth inhibition of multiple myeloma cells, Cancer Res. 66 (2006) 9714–9721.
[18] L. Thyrell, L. Hjortsberg, V. Arulampalam, T. Panaretakis, S. Uhles, M. Dagnell, B. Zhivotovsky, I. Leibiger, D. Grander, K. Pokrovskaja, Interferon alpha-induced apoptosis in tumor cells is mediated through the phosphoinositide 3-kinase/mammalian target of rapamycin signaling pathway, J. Biol. Chem. 279 (2004) 24152–24162.
[19] P.R. Greipp, J. San Miguel, B.G. Durie, J.J. Crowley, B. Barlogie, J. Blade, M. Boccadoro, J.A. Child, H. Avet-Loiseau, R.A. Kyle, J.J. Lahuerta, H. Ludwig, G. Morgan, R. Powles, K. Shimizu, C. Shustik, P. Sonneveld, P. Tosi, I. Turesson, J. Westin, International staging system for multiple myeloma, J. Clin. Oncol. 23 (2005) 3412–3420.
[20] J.M. Irish, J.H. Myklebust, A.A. Alizadeh, R. Houot, J.P. Sharman, D.K. Czerwinski, G.P. Nolan, R. Levy, B-cell signaling networks reveal a negative prognostic human lymphoma cell subset that emerges during tumor progression, Proc. Nat. Acad. Sci. U.S.A. 107 (2010) 12747–12754.
[21] V. Arulampalam, I. Kolosenko, L. Hjortsberg, A.C. Bjorklund, D. Grander, K.P. Tamm, Activation of STAT1 is required for interferon-alpha-mediated cell death, Exp. Cell. Res. 317 (2011) 9–19.
[22] J. Duus, H.I. Bahar, G. Venkataraman, F. Ozpuyan, K.F. Izban, H. Al-Masri, T. Maududi, A. Toor, S. Alkan, Analysis of expression of heat shock protein-90 (HSP90) and the effects of HSP90 inhibitor (17-AAG) in multiple myeloma, Leuk. Lymphoma 47 (2006) 1369–1378.
[23] C.S. Mitsiades, N.S. Mitsiades, C.J. McMullan, V. Poulaki, A.L. Kung, F.E. Davies, G. Morgan, M. Akiyama, R. Shringarpure, N.C. Munshi, P.G. Richardson, T. Hideshima, D. Chauhan, X. Gu, C. Bailey, M. Joseph, T.A. Libermann, N.S. Rosen, K.C. Anderson, Antimyeloma activity of heat shock protein-90 inhibition, Blood 107 (2006) 1092–1100.
[24] M. Chatterjee, S. Jain, T. Stuhmer, M. Andrulis, U. Ungethum, R.J. Kuban, H. Lorentz, K. Bommert, M. Topp, D. Kramer, H.K. Muller-Hermelink, H. Einsele, A. Greiner, R.C. Bargou, STAT3 and MAPK signaling maintain overexpression of heat shock proteins 90alpha and beta in multiple myeloma cells, which critically contribute to tumor-cell survival, Blood 109 (2007) 720–728.
[25] R. Bataille, N. Robillard, C. Pellat-Deceunynck, M. Amiot, A cellular model for myeloma cell growth and maturation based on an intraclonal CD45 hierarchy, Immunol. Rev. 194 (2003) 105–111.
[26] M. Collette, G. Descamps, C. Pellat-Deceunynck, R. Bataille, M. Amiot, Crucial role of phosphatase CD45 in determining signaling and proliferation of human myeloma cells, Eur. Cytokine Netw. 18 (2007) 120–126.
[27] C. Pellat-Deceunynck, R. Bataille, Normal and malignant human plasma cells: proliferation, differentiation, and expansions in relation to CD45 expression, Blood Cells Mol. Dis. 32 (2004) 293–301.
[28] S. Kumar, S.V. Rajkumar, T. Kimlinger, P.R. Greipp, T.E. Witzig, CD45 expression by bone marrow plasma cells in multiple myeloma: clinical and biological correlations, Leukemia 19 (2005) 1466–1470.
[29] H. Isomoto, S. Kobayashi, N.W. Werneburg, S.F. Bronk, M.E. Guicciardi, D.A. Frank, G.J. Gores, Interleukin 6 upregulates myeloid cell leukemia-1 expression through a STAT3 pathway in cholangiocarcinoma cells, Hepatology 42 (2005) 1329–1338.
[30] M. Jourdan, J.L. Veyrune, J. De Vos, N. Redal, G. Couderc, B. Klein, A major role for Mcl-1 antiapoptotic protein in the IL-6-induced survival of human myeloma cells, Oncogene 22 (2003) 2950–2959.
[31] T. Panaretakis, K. Pokrovskaja, M.C. Shoshan, D. Grander, Interferon-alpha-induced apoptosis in U266 cells is associated with activation of the proapoptotic Bcl-2 family members Bak and Bax, Oncogene 22 (2003) 4543–4556.
[32] J.F. Bromberg, M.H. Wrzeszczynska, G. Devgan, Y. Zhao, R.G. Pestell, C. Albanese, J.E. Darnell Jr., Stat3 as an oncogene, Cell 98 (1999) 295–303.
[33] F. Pene, Y.E. Claessens, O. Muller, F. Viguie, P. Mayeux, F. Dreyfus, C. Lacombe, D. Bouscary, Role of the phosphatidylinositol 3-kinase/Akt and mTOR/P70S6-kinase pathways in the proliferation and apoptosis in multiple myeloma, Oncogene 21 (2002) 6587–6597.
[34] S. Liu, H. Ishikawa, N. Tsuyama, F.J. Li, S. Abroun, K.I. Otsuyama, X. Zheng, Z. Ma, Y. Maki, M.S. Iqbal, M. Obata, M.M. Kawano, Increased susceptibility to apoptosis in CD45(þ) myeloma cells accompanied by the increased expression of VDAC1, Oncogene 25 (2006) 419–429.