Redox modulation of adjacent thiols in VLA-4 by AS101 converts myeloid leukemia cells from a drug-resistant to drug-sensitive state

: Interactions between the integrin VLA-4 on acute myelogenous leukemia (AML) cells with stromal fibronectin is a decisive factor in chemotherapeutic resistance. In this study, we provide a rationale for a drug repositioning strategy to blunt integrin activation in AML cells and restore their sensitivity to chemotherapy. Specifically, we demonstrate that the non-toxic tellurrium compound AS101, currently being evaluated in clinical trials, can abrogate the acquired resistance of AML. Mechanistic investigations revealed that AS101 caused redox inactivation of adjacent thiols in the exofacial domain of VLA-4 after its ligation to stromal fibronectin. This effect triggered cytoskeletal conformational changes that decreased PI3K/Akt/Bcl2 signaling, an obligatory step in chemosensitization by AS101. In a mouse xenograft of AML derived from patient leukemic cells with high VLA-4 expression and activity, we demonstrated that AS101 abrogated drug resistance and prolonged survival in mice receiving chemotherapy. Decreased integrin activity was confirmed on AML cells in vivo. The chemosensitizing activity of AS101 persisted in hosts with defective adaptive and innate immunity, consistent with evidence that integrin deactivation was not mediated by heightening immune attack. Our findings provide a mechanistic rationale to reposition the experimental clinical agent AS101 to degrade VLA-4-mediated chemoresistance and improve clinical responses in AML patients. chemotherapy-sensitive human myeloid leukemic cells both in vitro and in vivo Inside-out signaling may then play a larger role in persistent control of activation, a process that may benefit from filters provided by the interconnections among intracellular signaling paths. provide redox sensitive sites for regulation of other integrins. Therefore, this property of AS101 may afford specificity to the compound. B. The small tellurium molecule, AS101, has been shown to have an excellent safety profile in patients treated with the compound for several years. This suggests that the compound is probably relatively specific. 17 element in the human body. Our integrated results show that eradication of residual leukemic cells in a xenograft model of AML can be achieved by a unique alternative approach to existing strategies of controlling the VLA-4 integrin activation using AS101, currently being tested in clinical trials. In light of these results clinical studies involving AML patients refractory to chemotherapy treated with AS101 and chemotherapy have been approved and will be soon initiated. Such treatment might be particularly beneficial for patients with high functional VLA-4 expression.


Abstract:
Interactions between the integrin VLA-4 on acute myelogenous leukemia (AML) cells with stromal fibronectin is a decisive factor in chemotherapeutic resistance. In this study, we provide a rationale for a drug repositioning strategy to blunt integrin activation in AML cells and restore their sensitivity to chemotherapy. Specifically, we demonstrate that the non-toxic tellurrium compound AS101, currently being evaluated in clinical trials, can abrogate the acquired resistance of AML. Mechanistic investigations revealed that AS101 caused redox inactivation of adjacent thiols in the exofacial domain of VLA-4 after its ligation to stromal fibronectin. This effect triggered cytoskeletal conformational changes that decreased PI3K/Akt/Bcl2 signaling, an obligatory step in chemosensitization by AS101. In a mouse xenograft of AML derived from patient leukemic cells with high VLA-4 expression and activity, we demonstrated that AS101 abrogated drug resistance and prolonged survival in mice receiving chemotherapy. Decreased integrin activity was confirmed on AML cells in vivo. The chemosensitizing activity of AS101 persisted in hosts with defective adaptive and innate immunity, consistent with evidence that integrin deactivation was not mediated by heightening immune attack. Our findings provide a mechanistic rationale to reposition the experimental clinical agent AS101 to degrade VLA-4-mediated chemoresistance and improve clinical responses in AML patients.

Introduction
A major problem in the treatment of acute myeloid leukemia (AML) remains the recurrence of the pathology following chemotherapy. This common and severe complication is due to resistant leukemic cells which are localized in the bone marrow. In the hematopoietic niche, specific components of the microenvironment play a crucial role in the pathogenesis of AML by promoting tumor cell growth and survival as well as drug resistance (1). Evidence supporting this concept includes the finding that AML cells bind to bone marrow stromal cells through combined β  and β 2 integrin mechanisms (2). Because relapses following chemotherapy are a major contributor to poor survival in acute myeloid leukemia (4,5), elucidation of the VLA-4 activation/deactivation process, and eventually control and manipulation of its activation is important.
Although a wealth of evidence shows that inside-out signaling can control integrin activation (6,7), it has been postulated that at least specific integrin function may be directly affected by redox rearrangements within the cysteine-rich domain of the extracellular integrin regions (8)(9)(10)(11). Thus, a disulfide bond reshuffling mechanism is proposed in which resting and active integrins differ in the number and position of unpaired cysteine residues (12). Nevertheless, the physiological significance of this mode of integrin activation is not known. 4 AS101, an organotellurium(IV) compound previously (13,14) used in Phase II clinical trials in cancer patients, is a potent immunomodulator with a variety of potential therapeutic applications (15)(16)(17). The compound has been shown to have beneficial effects in diverse preclinical and clinical studies (13,14,18).
Accumulated evidence suggests that much of the biological activity of AS101 is directly related to its specific chemical interactions with cysteine thiol residues. The Te IV -thiol chemical bond may lead to conformational change or disulfide bond formation in a specific protein, possibly resulting in the loss of its biological activity, if the thiol residue is essential for that function (19,20). Indeed we demonstrated that the specific redox-modulating activities of AS101 result in a variety of beneficial biological effects: inhibition of IL-10(20) resulting in tumor sensitization (18); neuroprotection in both Parkinson's disease models (20) and ischemic stroke (21), all mediated by the Te(IV) redox chemistry of the compound.
Likewise, the protective mechanism of AS101 against homocysteine toxicity was shown to be directly mediated by its chemical reactivity, whereby AS101 reacted with homocysteine to form homocystine, the less toxic disulfide form of homocysteine (22).
Vicinal thiols are closely spaced protein thiols in native proteins that undergo reversible conversions with disulfide bonds. Vicinal thiols do not necessarily have to be in close proximity to the primary amino acid sequence of a protein to be brought into close opposition by protein folding (23). Vicinal thiolcontaining proteins regulate a variety of receptors and transcription regulatory proteins (24,25). These thiols are also found on cell surfaces (26). Vicinal thiols in VLA-4 and, in particular on VLA on AML cells have not been yet described.
Based on the thiol-Te interaction of AS101, the present study reveals a unique approach in regulating cellular VLA-4 activity, via redox modulation of vicinal thiols on VLA-4 by AS101, and explores its role in the conversion of resistant to chemotherapy-sensitive human myeloid leukemic cells both in vitro and in vivo. 6 monoclonal anti-biotin antibody. Immunoprecipitation and Western Blot analysis was performed as described (20).
Attachment assay: 96 wells plates were coated with 80 μl of FN or VCAM-1 or BSA. Cells with or without AS101 were incubated in the wells for one hour, Thereafter, cells were washed three times. The attached cells were tested by the XTT assay at 450nm.

Enrichment of human leukemic CD45 + cells.
Human CD45 + cells were enriched from total spleen cells using the MACS magnetic cell sorter (Miltenyi Biotec, Germany) according to the manifacturer's instructions, obtaining purity of about 95%. Purified cells were used freshly for attachment assays.

Actin shape index.
A quantitative estimate of the actin shape changes was obtained as previously described (28). Briefly, actin shape index was calculated as P 2 /4πS, where P and S are the perimeter and the surface of the cell, respectively. These values were obtained by thresholding images of phalloidin staining to outline the polymerized actin. A perfectly circular shape of the polymerized actin yields a shape index of 1, and departure from a circle yields a shape index larger than 1. Actin reorganization was determined following 30 min of activation, which is the best time for observing this process in these cells. FRET analysis. Fluorescence resonance energy transfer (FRET) was measured by the donor-sensitized acceptor fluorescence technique as previously described (29). Three sets of filters were used: one Statistics. Results are expressed as mean ± S.E, Differences in cells viability between groups in response to chemotherapy were analyzed using two-way ANOVA. Differences in survival curves between groups were calculated by the Kaplan-Meier method. p<0.05 was considered statistically significant.

AS101 sensitizes leukemic cells to chemotherapy-induced death via inhibition of VLA-4 activity.
In the present study, we show that AS101 increases the sensitivity of human AML cells to chemotherapy and prevents the minimal remaining of resistant cells in a xenograft mouse model. Furthermore, the mechanism of action of this phenomenon is defined. 9 neutralizing anti VLA-5 antibodies did not affect cells sensitivity to chemotherapy (Fig. 2c). These results suggest that U937 cells acquire drug-resistance via α 4 β 1 , but not α 5 β 1 , and VLA-4 is an important target for AS101's sensitization of leukemic cells to chemotherapy-induced death. As further proof for integrin deactivation by AS101, Fig. 2d shows that AS101 prevents attachment of AML cells to both VCAM-1(a VLA-4 ligand) and Fibronectin (ligand to VLA-4 and VLA-5) ( Author manuscripts have been peer reviewed and accepted for publication but have not yet been edited.
Furthermore, biotin-BMCC labeling was found on both α 4 and β 1 peptides when the cells were prereduced with dithiothreitol (DTT). AS101 significantly lowered this thiol labeling on both chains (Fig.S3a). Importantly, the thiol residues on the α4 chain were diminuted in AML cells treated with the Interestingly, no exposed sulfhydryl groups were found on the immunoprecipitated α 5 chain of U937 cells cultured in the presence of fibronectin (Fig, 3c).  Fig. S4a and b). These data imply that the sensitivity of AML leukemic cells expressing high VLA-4 to chemotherapy may be enhanced by redox modulation of the VLA-4 integrin. In order to validate AS101's target on AML cells, VLA-4 from AML patient leukemic cells expressing high VLA-4, or VLA from mouse AML cells, was knocked out by shRNA. As seen in Fig. S5, knocking-out VLA-4 rendered these cells sensitive to chemotherapy while AS101 did not further potentiate this activity.

The resistance of AML patients leukemic cells expressing high
Signaling alterations involved in enhanced drug-sensitivity induced by AS101. Inhibition of VLA-4 activity by AS101 was associated with decreased PI3K/Akt/BCl-2 signalling and this effect was related to the enhanced drug-sensitivity induced by AS101. Fig. 4a shows that AS101decreases the expression of pAkt in U937 cells plated on fibronectin in a dose-dependent manner. Pharmacological inhibition of PI3K with LY294002 significantly increased drug sensitivity of leukemic cells when cells were plated on fibronectin, while AS101 did not further enhance this sensitivity (Fig. 4b). Moreover, overexpression of Akt in leukemic cells (Fig. 4c) abolished the sensitizing effect of AS101 to drug-induced death when leukemic cells were plated on fibronectin as reflected by both cell viability (Fig.4d) and subG1 accumulation assays (FigS6). These data collectively suggest that inhibition of VLA-4 activity by AS101 decreases the expression of pAkt and this property, at least in part, mediates AS101's sensitizing activity.
Similarly, Bcl-2, a downstream substrate of Akt, was also shown to mediate AS101's sensitizing activity. Overexpression of Bcl-2 abolished AS101's sensitizing effect in the presence of fibronectin ( Fig. 5a and   b). Moreover transient transfection of Bcl-2 antisense oligonucleotides (Fig. 5c) partially but significantly restored leukemic cells sensitivity to chemotherapy, while addition of AS101 slightly but insignificanltly further enhanced this sensitivity (Fig. 5d), suggesting Bcl-2 as a mediator of AS101's activity.

Conversion of resistant to chemotherapy-sensitive human myeloid leukemic cells in vivo by AS101
We then asked if the VLA-4 inactivation by AS101 translates into enhanced therapeutic effects in vivo.
For this purpose we inoculated leukemic cells from either U937 cell lines or from AML patients i.v. to SCID mice, and monitored their organ distribution by PCR using the human α satellite sequence. 8 days after U937 inoculation (Fig. S7a) or 4 days after inoculation of leukemic cells from an AML patient expressing high VLA-4 ( Fig. 6a), leukemic cells resided only within the BM. While they persisted in the BM of both AS101 or ARA-C-treated mice, they were totally eliminated from the BM of mice subjected to combined treatment with AS101+ARA-C ( Fig. S7a and Fig. 6a). This therapeutic effect persisted at 60 days after U937 cells implantation (Fig. S7a). At 14 days after implantation, leukemic cells from both origins were detected in all organs examined ( Fig. S7a and Fig. 6b). Nevertheless, no detectable human AML cells were found in organs of AS101+ARA-C-treated mice (Fig. 6b). Importantly, the improved sensitivity of leukemic cells to chemotherapy following treatment with AS101 was reflected by the increased survival of mice implanted with U937 cells (Fig. S7b). Moreover, the combined treatment of mice inoculated with leukemic cells from an AML patient expressing high VLA-4 significantly increased mice survival (Fig. 6c) as opposed to the relative insensitivity to ARA-C alone, while mice inoculated with leukemic cells from an AML patient expressing low VLA-4, responded well to chemotherapy alone, yielding 80% survival (Fig. 6d). Co-treatment with AS101+ARA-C further but insignificantly increased mice survival. In order to preclude the possibility that AS101 may potentially alter the innate immune response to the leukemia cells, which can be integrin α4β1 driven, we used the NSG (NOD SCID GAMMA) mice with profound immunological multidysfunction in both adaptive and innate immunologic function.. Fig. 7a shows that att 14 days after implantation, leukemic cells were detected in all organs 13 examined. Nevertheless, no detectable human AML cells were found in organs of AS101+ARA-C-treated mice (Fig. 7a). Furthermore, AS101 is shown to inhibit integrin activity in these mice in vivo. Fig. 7b shows that treatment with AS101 inhibits attachment of leukemic cells, enriched from mice spleens, to both VCAM-1 and Fibronectin. These data collectively suggest that high VLA-4 expression and activity in AML leukemic cells confers resistance to chemotherapy and VLA-4 is a target for AS101 in sensitization of AML cells in vivo. Although a wealth of evidence shows that inside-out signaling via factors in the cytoplasm can control integrin activation state (30)(31)(32), an alternative concept now suggests that at least some integrin's activation could be controlled directly by a redox site in the extracellular domain, independent of factors in the cytoplasm (33)(34)(35)(36). Thus, integrin disulfide exchange may be involved in aspects of integrin activation, altering integrin conformation and increasing the ligand binding affinity of the integrin (10,37,38). The two pathways to integrin activation, inside-out and redox modulation, may serve different purposes. The redox switch could regulate rapid and transient changes in activation state that require no "filtering" through the cytoplasm. Inside-out signaling may then play a larger role in persistent control of activation, a process that may benefit from filters provided by the interconnections among intracellular signaling paths. Using a proteomic approach, Laragione et al identified the α 4 integrin as a molecular target susceptible to redox regulation (39). Furthermore, Liu et al (40) found that ligand binding of VLA-4 induced exposure of sulfhydryl groups on the α 4 peptide.
The Te(IV)-thiol chemical bond formed between AS101 and the α 4 chain leads to conformational change in the VLA-4 integrin as seen in this study. Our study suggests that the inactivation of VLA-4 is due to binding of AS101 to the thiol groups of vicinal cysteines on the α 4 chain. At least some tellurium derivatives are known to interact with specific vicinal thiols within certain proteins (41). We thus assume that these dithiols reversibly interconvert to a disulfide bond under physiologic conditions, providing sites where changes in the redox environment can regulate protein function.
We cannot exclude the possibility that AS101 also modifies sulfhydryls on other membrane cell proteins (including other integrins) that could be involved in chemoresistance. Nevertheless, the consistency between the results obtained with α 4 β 1 expressed on cell surface leukemic cultures and on patients' leukemic cells, with regulation of chemosensitivity, and the lack of free thiols on the exofolial region of α5 strongly suggests a mechanistic link between the decrease in free cysteine residues in α 4 β 1 and the decrease in chemotherapy-induced cell resistance following treatment with AS101. Yet, a differential reactivity of various cysteines toward AS101 may exist, possibly reflecting their variable access to AS101. Indeed, α IIb β 3 is subject to S-nitrosylation at specific critical cysteines residues located within specific motifs resulting in the integrin loss of function (42). This agent does not however deactivate all other types of integrins. In general some indications suggest that AS101 is relatively specific: A. As shown, the inactivation of VLA-4 is due to binding of AS101 to the thiol groups of vicinal cysteines on the α 4 chain. These thiols do not necessarily provide redox sensitive sites for regulation of other integrins. Therefore, this property of AS101 may afford specificity to the compound. B. The small tellurium molecule, AS101, has been shown to have an excellent safety profile in patients treated with the compound for several years. This suggests that the compound is probably relatively specific. VLA-4 has emerged as a promising therapeutic target in AML. Still, its role as a clinical marker to define remission induction, disease recurrence, or prognosis remains controversial. Initially, anti-VLA-4 antibodies were shown to chemosensitize human AML cells and to eradicate MRD in experimental mice when combined with chemotherapy, implicating VLA-4 in acquired chemotherapy resistance and MRD(3). These findings suggested that high VLA-4 expression might reduce chemosensitivity, resulting in poor remission induction, MRD, disease recurrence, and short survival(3). Another study supported this prediction. Tavernier-Tardy et al (43) showed that high VLA-4 expression was associated with shorter survival of AML patients. Furthermore, although nonsignificant (p=0.058), there was a trend for lower VLA-4 expression in samples from patients in remission as compared to refractory patients. Moreover, they found a high significant correlation between the expression of CXCR4 and VLA-4. Over expression of CXCR4 on AML cells has been previously described to predict adverse overall survival (44)(45)(46). Our results are in line with those of Matsunaga et al. Nevertheless, a recent study (47) failed to confirm the prognostic role of VLA in AML as suggested by Matsunaga, and reported that functional VLA-4, but not merely VLA-4 expression, is associated with longer overall survival of adult AML patients. This apparently controversy may be hypothetically explained by the fact that high VLA-4 expression and function may merely be a surrogate for certain favorable factors that affect prognosis, and they do not contradict the principle that inhibition of VLA-4 activity on AML leukemic cells is advantageous with respect to increased chemosensitivity and improved outcome.
The ability of AS101 to sensitize leukemic cells from patients expressing high VLA-4 to chemotherapy-induced death was associated with the inhibition of Focal adhesion kinase (FAK) phosphorylation at Tyr-397, a residue that is critical for its function. FAK is a downstream signal that follows α 4 β 1 integrin engagement. Recently Recher et al (48) have shown that FAK is frequently expressed and activated in AML cells, and that FAK expression correlates with enhanced migratory properties, drug resistance, high leukocytosis, and reduced survival.