Cytotoxic Activity of Gemtuzumab Ozogamicin (Mylotarg) in Acute Myeloid Leukemia Correlates with the Expression of Protein Kinase Syk

Cytotoxic activity of gemtuzumab ozogamicin (Mylotarg) in acute myeloid leukemia correlates with the expression of protein kinase Syk

L Balaian and ED Ball

Blood and Marrow Transplantation Division, Department of Medicine and Moores UCSD Cancer Center, University of California, San Diego, La Jolla, CA, USA

Acute myeloid leukemia (AML) cells express the cell surface component in the therapy of AML.11 GO induces remissions in antigen CD33 that, upon ligation with a monoclonal antibody about 30% of patients with relapsed AML. GO, a humanized (mAb), is a downregulator of cell growth in a Syk-dependent manner. An anti-CD33 mAb coupled to a toxin, gemtuzumab immunoglobulin G (IgG) 4 mAb, contains human sequences ozogamicin (GO), is used for the treatment of AML (Mylotarg). whereas the complementarity-determining regions are derived Therefore, we investigated whether the response of AML cells from a murine antibody that binds CD33. The antibody is linked to GO treatment also depends on Syk expression. Forty primary to N-acetyl-gamma calicheamycin via a bifunctional linker and AML samples (25 Syk-positive and 15 Syk-negative) were tested is 50% loaded with 4–6 mol of calicheamycin per mole of for their response to the anti-proliferative effects of GO and antibody. Notably, the remaining 50% of the antibody is not unmodified anti-CD33 mAb. A correlation between Syk expression and the response of leukemia cells to GO and anti-CD33 linked to calicheamycin. Therefore, we examined whether the mAb was found. ‘Blocking’ of Syk by small interfering RNA effects of GO depend on CD33-coupled molecules such as Syk. resulted in unresponsiveness of AML cells to both GO and anti- As GO is effective in only a minority of patients, the signaling CD33 mAb-mediated cytotoxicity. Syk upregulation by the de- activity of CD33 may be relevant to the responses achieved, in methylating agent 5-azacytidine (5-aza) induced re-expression addition to the effect of the toxin calicheamycin. of Syk in some cases, resulting in enhanced GO and anti-CD33- The protein kinase Syk is an essential element in many antigen mediated inhibition of leukemia cell growth. Thus, the cytotoxicity of both GO and anti-CD33 in primary AML samples was asso- receptor (B-cell receptor, T-cell receptor, Fc receptors) down- ciated with Syk expression. 5-Aza restored Syk and increased stream signaling cascades, resulting in cell responses such as 12–17 the sensitivity of originally Syk-negative, non-responsive cells adhesion, phagocytosis, proliferation and differentiation. to CD33 ligation to levels of Syk-positive cells. These data During early stages of antigen ligation, Syk binds to the recep- have clinical significance for predicting response to GO and tor, becomes activated and phosphorylated and then phosphory- designing clinical trials. lates specific substrates such as phospholipase C-gamma Leukemia (2006) 20, 2093–2101. doi:10.1038/sj.leu.2404437; 18 published online 19 October 2006 with consecutive calcium influx. Late signaling events such Keywords: myeloid leukemia; Syk; anti-CD33 mAb; gemtuzumab as the activation of transcriptional factor nuclear factor- ozogamicin; 5-azacytidine. kappa B couple Syk activation with cell proliferation and differentiation.19 In myeloid cells, Syk involvement in proximal signaling mediated by activated Fc receptor family members containing the tyrosine-based activation motif (ITAM) is well documen- Introduction ted.19–23 However, recent discoveries revealed that Syk is also involved in signaling of the ITIM-bearing CD22 receptor in B CD33 is a cell surface glycoprotein specifically expressed on cells.24–27 We extended these findings by demonstrating that myeloid cells including myeloid leukemia cells.1 Monoclonal Syk (or ZAP-70) play an important role in CD33 signaling.28,29 antibodies (mAbs) against CD33 have been used in the diagnosis Upon CD33 ligation, Syk becomes phosphorylated and creates and therapy of acute myeloid leukemia (AML) for many years.2 complexes with phosphorylated forms of the CD33 molecule CD33, a member of the siglec family, is engaged in sialic acid- itself and protein phosphatase Src homology phosphatase-1 dependent cell interactions and adhesion of myeloid cells.3 The (SHP-1). Moreover, we showed that the anti-proliferative cytoplasmic tail of CD33 contains two immune tyrosine-based response of AML cells to CD33 ligation correlates with the inhibitory motifs (ITIMs) and therefore, may serve as a potential level of Syk expression. Thirty percent of primary AML samples inhibitory receptor.4–6 Engagement of CD33 induced apoptosis demonstrate no detectable Syk expression. Significantly greater and inhibition of proliferation in leukemia cells from AML and numbers of Syk/Zap-70-positive samples respond to anti-CD33 chronic myeloid leukemia patients.7–10 However, little is known mAb treatment.29 about the molecular mechanisms of the CD33 signaling events Recent discoveries established Syk as a tumor suppressor and leading to inhibition of cell growth and apoptosis. linked deficient Syk expression to a variety of human The toxin (calicheamycin)-conjugated anti-CD33 mAb gem- hematopoietic30–33 and solid tumors.33–35 In breast cancer, tuzumab ozogamicin (GO) is now established as a useful Syk kinase is a potent modulator of malignant growth and a potential tumor suppressor, presumably by controlling cell Correspondence: Dr ED Ball, Blood and Marrow Transplantation division.33 Moreover, it was demonstrated that loss of Syk Division, Department of Medicine and Moores UCSD Cancer Center, expression in breast cancer and T-cell acute lymphoblastic University of California, San Diego, 3855 Health Sciences Drive, La leukemia cells occurs at the transcriptional level, and is a result Jolla, CA 92093-0960, USA. 31,36,37 E-mail: [email protected] of DNA hypermethylation. Treatment of these cells with Received 25 April 2006; revised 11 August 2006; accepted 23 August 5-azacytidine (5-aza), a methylation inhibitor, restored Syk 2006; published online 19 October 2006 expression and function.36,37 Cytotoxicity of anti-CD33 mAb and GO correlate with Syk expression L Balaian and ED Ball 2094 5-Aza is a nonspecific Syk DNA methyltransferase inhibitor was determined by flow cytometry. CD2, CD3, CD7, CD10, that is used for the treatment of myelodysplasia38 and possibly CD13, CD14, CD15, CD16, CD19, CD20, CD33, CD34, CD64, AML.39,40 However, the effects of 5-aza on leukemia cell growth CD95, human leukocyte antigen-DR, TdT expression were are not well defined. As expression levels of Syk are likely to analyzed. All antibodies were obtained from BD Pharmingen play an important role for the competency of the immune system (San Jose, CA, USA). and may also play a role in oncogenesis, we examined a panel of 40 primary AML samples for possible correlations between Syk expression and their response to the anti-proliferative effects Cell transfection of unmodified or immunotoxin-bound anti-CD33 mAb. SMARTpool small interfering RNAs (siRNAs) (collection of at least four individual siRNA, catalog No. 60–047 ) that target human Syk (Gen Bank Accession No. NM_003177) as well as Materials and methods control RNA (nonspecific pool) were purchased at Upstate Biotechnology Inc. (Lake Placid, NY, USA). Leukemia cells were Antibodies transfected according to the manufacturer’s instructions. Briefly, The anti-CD33 mAb was obtained from Medarex Inc. (Prince- cells (3–4 Â 106) were transfected by 5 mg of Syk siRNA or ton, NJ, USA). Anti-Syk, anti-SHP-1, anti-CD33 rabbit or goat control naked siRNA mix using 6 ml Lipofectamine 2000 polyclonal antibodies were purchased from Santa Cruz Bio- (Invitrogen, Carlsbad, CA, USA). Seventy-two hours after technology Inc. (Santa Cruz, CA, USA). The anti-CD13 mAb was transfection, the cells were used for proliferation assays and purchased from Cell Sciences Inc. (Norwood, MA, USA). A horse- Western blot analysis. radish peroxidase-conjugated anti-phosphotyrosine mAb, 4G10, was obtained from Upstate Biotechnology Inc. (Lake Placid, NY, USA). Cell activation and immunoprecipitation Mononuclear cells from AML patients or AML cell lines (5– 6 Â 106) were activated by 10 mg/ml anti-CD33 or anti-CD13 Materials mAb of similar isotype for 20 min at room temperature, followed All chemicals were purchased from Sigma Chemical Co. (St by addition of polyclonal anti-mouse IgG at 20 mg/ml for Louis, MO, USA). variable times (1–60 min). The reaction was stopped by adding ice-cold phosphate-buffered saline (PBS). After three washes in ice-cold PBS, the cells were lysed in lysis buffer, containing 1% Cells (v/v) Triton X-100, 0.15 M NaCl, 50 mM Tris-HCl (pH ¼ 7.2), The human AML cell line Hodgkin’s lymphoma (HL)-60 was 0,1% sodium dodecyl sulfate (SDS), 1 mM Na-orthovanadate, obtained from ATCC (Manassas, VA, USA). Cells from AML 1mM phenylmethylsulfonyl fluoride, 1% (v/v) ethylenediamine- patients were collected from peripheral blood after informed tetraacetic acid, 1% (v/v) Aprotinin and 0.03 mM Leupeptin. consent under the auspices of the University of California San After 30 min on ice, the nuclear debris was removed by Diego Institutional Review Board. Mononuclear cells were centrifugation for 15 min at 13 000 Â g. Lysates were equalized isolated on a Ficoll-Hypaque gradient and cultured under with respect to the amount of protein as assessed by optical standard conditions in Rosewell Park Memorial Institute density at 280 nM. Specific immunoprecipitation was performed medium (RPMI)-1640 containing 10% fetal calf serum (FCS) for 2 h to overnight in the presence of 30% (vol./vol.) ‘Protein and 100 ng/ml of granulocyte-monocyte colony-stimulating A/G’, conjugated with agarose (Santa Cruz Biotech., Santa Cruz, factor. CA, USA). Immunoprecipitates were washed three times in lysis buffer and then suspended in equal volume of Laemli sample buffer for SDS-polyacrylamide gel electrophoresis (PAGE). Proliferation assay AML cells were cultured in triplicate wells in 96-well round bottom plates, with or without various amounts of anti-CD33 or SDS-PAGE and Western Blotting control anti-CD13 mAb in a final volume of 200 ml of RPMI-10 Total cell lysates or immunoprecipitates were added to separate 5 at 371C in humidified 5% CO2 in air, at 10 cells per well. The wells (8 mg/well) of SDS-PAGE (7.5–10% acrylamide) gel, cultures were pulsed with 1 microCi [3H]-thymidine per well electrophoretically size-separated under reducing conditions, during the last 3 h of the 48 h culture for cell lines. Primary AML and then transferred onto nitrocellulose for immunoblotting. The cells were pulsed during the last 16 h of culture. The amount of filters were first incubated for 1 h in 5% non-fat dry milk in PBS- 3H-thymidine incorporated into acid-precipitable DNA was T (PBS plus 0.01% Tween 20), and then incubated with the assessed via liquid scintillation counting. primary antibody for 2 h. After washing in PBS-T, the filters were incubated for 1 h in horseradish peroxidase-conjugated matching secondary antibodies. The filters were then washed in PBS-T, Flow cytometry analysis incubated with the enhanced chemoluminescence detection Leukemia cells were washed and then suspended in staining reagents (Pierce, Rockford, IL, USA), and exposed to X-ray film media (SM), containing RPMI-1640, 3% FCS, 0.01% NaN3 and (Fuji Film, Fischer, Tustin, CA, USA). In all figures, samples were 1 mg/ml propidium iodide (Calbiochem, La Jolla, CA, USA), plus run in parallel gels. saturating amounts of fluorescein isothiocyanate-conjugated anti-CD33, anti-CD64 mAbs, or an isotype-matched control mAb of irrelevant specificity. After 30 min at 41C, the cells were Statistical analyses washed in SM twice and then analyzed on a FACScan (Becton S.e.m. calculation, graph production and statistical evaluation Dickinson, San Jose, CA, USA). Dead cells and debris were were performed using Sigma-Plot 8.0 version (Systat Software excluded from analysis by characteristic forward and side- Inc., Point Richmond, CA, USA) and Microsoft Excel (Microsoft, scatter profiles and propidium iodide staining. The immuno- Seattle, WA, USA). Statistical significance of the difference phenotype of primary AML peripheral blood mononuclear cells between sample groups was calculated by using the Student’s

Leukemia Cytotoxicity of anti-CD33 mAb and GO correlate with Syk expression L Balaian and ED Ball 2095 t-test and was defined as a P-value less than or equal to 0.05. manner, at concentrations greater than 0.1 mg/ml.28,29 At The significance of observed differences in proportions was optimal concentrations (0.1 mg/ml), anti-CD33 mAb inhibited tested using the w2 test and was defined as a Pp0.05. colony formation of HL-60 cells by 450% and DNA synthesis in primary AML cells up to 40%. We previously found that CD33 ligation induced tyrosine phosphorylation of the tyrosine 28,29 Results kinase Syk, but not of src-family tyrosine kinases (Fyn, Lyn). As these results suggested an important role of Syk in CD33 The growth inhibitory effects of anti-CD33 mAb: signaling, we tested a panel of 40 primary AML samples from correlation with Syk expression different French-American-British (FAB) types for Syk expression The cell surface CD33 receptor on myeloid leukemia cells (Table 1). All samples had high blast counts (data not shown), functions as a negative regulator of cell growth. Anti-CD33 and the majority (36 of 40) contained high percentages (460%) mAb-mediated growth arrest occurs in a dose-dependent of surface CD33. By Western blotting (data not shown), Syk expression was undetectable in about 37% (15 of 40) of primary AML samples. We did not detect any correlation between Syk expression and FAB type or CD33 expression. Table 1 Syk expression on primary AML samples We previously reported that anti-CD33 mAb induced dose- dependent growth inhibition more effectively in primary AML FAB Total no. CD33 No. of Syk- No. of Syk- 29 of expression positive negative cells that expressed Syk. Here, in order to amplify the samples (median) (%) samples samples representative groups, a panel of 40 primary AML samples was tested for their response to the optimal concentration MO 1 92 1 0 (0.1 mg/ml) of anti-CD33 mAb (Figure 1a). The level of inhibition M1 5 81 2 3 in Syk-positive samples was considerably higher (mean value M2 7 85 5 2 435%) than in Syk-negative samples (mean value o20%). This M3 1 73 1 0 M4 10 86 5 5 difference was statistically significant (Po0.05; Student’s t-test). M5 9 94 5 4 Moreover, among Syk-positive samples (n ¼ 25), 68% demon- M6 2 85 2 0 strated significant growth inhibition in response to CD33 M7 1 92 1 0 ligation (responders, as defined by 425% inhibition of 3H- Unknown 4 83 3 1 thymidine uptake by anti-CD33 mAb) whereas 32% had no Abbreviations: AML, acute myeloid leukemia; FAB, French-American- significant response (non-responders, defined by o25% growth British. inhibition) (Figure 1b). In contrast, only 20% of the Syk-negative

a a-CD13 mAb 50 a-CD33 mAb 40

Inhibition of proliferation 0 + - (% of control media basal level) (n=25) (n=15) Syk Syk

b Syk + samples Syk - samples (25 of 40) (15 of 40)

R non-R 20% R 32% 68%

P =0.0033 non-R Chi-square test 80%

Figure 1 Correlation between Syk expression and the anti-proliferative effect of CD33 ligation in primary AML cells. (a) Proliferation assay. Primary AML cells were cultured for 48 h in the presence or absence of 0.1 mg/ml of anti-CD33 mAb or control anti-CD13 mAb. 3H-thymidine incorporation was measured during the last 18 h of culture. Basal proliferation in the absence of antibody (c.p.m.45500 c.p.m.) was considered 0, and results are expressed as % change for each condition. Samples were grouped according to the expression of Syk/Zap70 and their response to CD33 ligation. Responders (R) are deﬁned as samples with 425% inhibition of 3H-thymidine uptake in the presence of anti-CD33 mAb, whereas non-responders (NR) are deﬁned as samples with o25% inhibition. Error bars indicate the s.e.m. (b) Correlation between Syk expression and the response of AML primary cells to CD33 ligation. Summary of primary AML proliferation.

Leukemia Cytotoxicity of anti-CD33 mAb and GO correlate with Syk expression L Balaian and ED Ball 2096 samples (n ¼ 15) responded to anti-CD33 mAb. These results tion of Syk expression largely prevented the anti-proliferative suggest a correlation between Syk expression and responsive- effects of the naked anti-CD33 mAb (Figure 3a) as well as the ness of AML cells to CD33 ligation (P ¼ 0.0033 by w2 test). immunotoxin-conjugated mAb GO (Figure 3b). Thus, the Syk-specific siRNA converted initially Syk-positive and CD33-responsive AML cells into Syk-negative non-responsive The growth inhibitory effects of GO (immunotoxin- cells. These results show that Syk plays a central role in CD33 conjugated anti-CD33): correlation with Syk expression signaling. As Syk þ AML cells were more likely to respond to CD33 ligation with growth inhibition compared to SykÀ AML samples, we asked whether the anti-proliferative effect of GO might also depend on the presence of Syk. GO inhibited DNA synthesis in 5-Aza treatment of Syk-negative primary AML cells a dose-dependent manner in both Syk-positive and Syk-negative increases their response to GO and anti-CD33 mAb samples, but the level of inhibition was significantly higher in Syk is an important tyrosine kinase, which appears to function as the Syk þ compared to Syk samples at all concentrations tested a tumor suppressor that is silenced by hypermethylation in many (Po0.003, Student’s t-test) (Figure 2a). The difference was more cancer cells. As we found a correlation between Syk expression prominent at low doses of GO and diminished at higher doses, and the response of primary AML samples to GO and anti- probably owing to free toxin activity. CD33 m Ab, the DNA methylase inhibitor 5-aza was used in a Response to GO was defined as 450% inhibition of 3H- panel of 40 primary AML samples alone (Figure 4) and in thymidine uptake at 10 ng/ml, whereas non-responders demon- combination with GO or anti-CD33 mAb (Figure 5). strated o50% inhibition at this concentration. In Syk-positive First, we analyzed dose- and time-dependent anti-prolifera- AML samples, the number of responsive samples was signifi- tive effects of 5-aza on human leukemia cell line HL-60 cantly higher (73%) compared to Syk-negative samples (17%) (Figure 4a). Treatment conditions: dose of 100 nM for 48 h was (Figure 2b). These data show a correlation between Syk considered as suboptimal, and was used in consequent expression in primary AML cells and the inhibitory activity of experiments with primary AML cells. Based on the level of GO (P ¼ 0.02; w2 test). inhibition mediated by 5-aza, we distinguished three groups of primary AML samples (Figure 4b). The majority (21 of 40 samples) demonstrated low response (inhibition of proliferation siRNA-mediated silencing of Syk expression largely o25%). In nine of 40 samples, 5-aza treatment induced prevents the anti-proliferative response of AML cells medium (25–50% inhibition) and in 10 of 40 samples this treat- to anti-CD33 mAb and GO ment mediated high (450% inhibition) responses. Response to To determine whether Syk expression was required for CD33- 5-aza was defined as 425% inhibition of 3H-thymidine uptake mediated inhibition of DNA synthesis, we used a siRNA strategy at 100 nM after 48 h treatment, whereas non-responders to downregulate Syk expression. We transfected HL-60 cells demonstrated o25% inhibition at these treatment conditions. with Syk-specific siRNA or control siRNA using Lipofectamine In Syk-positive AML samples, the number of responsive samples 2000 (Figure 3). Seventy-two hours post-transfection, Syk was was significantly higher (60%) compared to Syk-negative undetectable by Western blotting in cells treated with the Syk- samples (26%) (Figure 4c). These data suggest a correlation siRNA, the level of SHP-1 was unchanged in these cells, between Syk expression in primary AML cells and the inhibitory demonstrating specificity of the effect (Figure 3c). Downregula- activity of 5-aza (P ¼ 0.04; w2 test).

a 120 100 Syk + (n =25) AML samples 80 Syk - (n=15) AML samples 60 40 20

0 Proliferation (% of control) Proliferation none 1 5 10 50 100 500 1000

GO concentration (ng/ml)

b Syk + samples (25 of 40) Syk- samples (15 of 40) R 17% P = 0.02; non-R 36% Chi-square test R non-R 64% 83%

Figure 2 The effect of GO on growth of AML cells is dependent on Syk expression. (a) Dose-dependent inhibition of proliferation induced by GO treatment. Proliferation assays were performed with primary AML cells as described in Figure 1a. GO was added for 48 h at the indicated concentration. Basal cell proliferation of untreated cells (410 000 c.p.m. of 3H-thymidine incorporation) was considered to be 100%. (b) Correlation between Syk expression and the response of primary AML cells to GO treatment. Summarized data (mean) for 25 Syk-positive and 15 Syk-negative samples are presented. Error bars indicate the s.e.m.

Leukemia Cytotoxicity of anti-CD33 mAb and GO correlate with Syk expression L Balaian and ED Ball 2097 ab 120 * 100 * 120 100 80 80 60 60 * 40 40 20 20 0 Proliferation (% of control) Proliferation 0 none 1 10 100 1000 cntrl RNA Syk si-RNA Proliferation (% of control) GO concentration (ng/ml) media untreated cells α-CD13 mAb α control si-RNA treated -CD33 mAb Syk si-RNA treated

c I IP IgG Syk probed with 1234 1234 α-Syk Ab Syk

II IP IgG SHP-1 probed with 1234 1234 α-SHP-1 Ab SHP-1 1 – untreated cells 2 – control RNA treated cells 3 – Syk siRNA treated cells 4 – positive control (I-Raji or II-U-937 cells)

Figure 3 siRNA silencing of Syk expression decreases the anti-proliferative response of primary AML cells to anti-CD33 mAb including GO. (a and b) Proliferation assay. HL-60 cells were transfected with Syk siRNA or control siRNA. Seventy-two hours post-transfection, cells were treated with either naked anti-CD33 mAb or control anti-CD13 mAb (a) or they received the indicated concentrations of GO (b). Proliferation assays were performed as described in Figure 1a. The basal 3H-thymidine incorporation of untreated cells was 410 000 c.p.m. (considered as 100%). Data represent the mean7s.e.m. of three independent experiments. The asterisks indicate statistically signiﬁcant difference between groups of data. (c) Western blots of HL-60 cells transfected with Syk siRNA or control siRNA. After 72 h of culture, cell lysates were subjected to immunoprecipitation and Western blotting with the indicated mAb. Shown results represent one of three independent experiments.

a b

120 100

100 5 aza 100 nM 80 5 aza 1000 nM 80 60

60 40

40 20

20 Inhibition of proliferation 0 Proliferation (% of control) n=9) 10) 0 (% of untreated cell proliferation) (n=21) ( w gh (n= none 1 2 4 5 lo edium hi Days in culture m Group response to 5-aza treatment

c Syk+ (n=25) samples Syk+ (n=15) samples n-R n-R n=11, R n=10, 73% n=4, 40% 27% R n=15, 60% P=0.04 Chi-square test

Figure 4 The effect of 5-aza on growth of AML cells is dependent on Syk expression. (a) Proliferation assay. HL-60 cells were treated with indicated concentrations of 5-aza for 1–5 days. After two washes, proliferation assays were performed as described in Figure 1a. The basal 3H-thymidine incorporation of untreated cells was 410 000 c.p.m. (considered as 100%). Data represent the mean7s.e.m. of three independent experiments. (b) Proliferation assay. Primary AML cells were cultured for 48 h in the presence of 100 nM of 5-aza and then after two washes, proliferation assays were performed as described in Figure 1b. (c) Correlation between Syk expression and the response of primary AML cells to 5-aza treatment. Summary of primary AML cell proliferation.

Leukemia Cytotoxicity of anti-CD33 mAb and GO correlate with Syk expression L Balaian and ED Ball 2098 a Syk+ samples (n=25) b Syk- samples (n=15) 120 120 100 100 80 80 60 60 40 40 20 20 0 0 none 0.1 1 none 0.1 1 α-CD33 mAb concentration (µg/ml)

c Syk+ samples (n=25) d Syk- samples (n=15) 120 120 100 100 Proliferation (% of conrol) Proliferation 80 80 60 60 40 40 20 20 0 0 none 1 10 100 none 1 10 100 GO concentration (ng/ml)

no treatment 5-aza (100nM ) 48h treatment

Figure 5 5-Aza treatment of primary AML cells increases response to anti-CD33 mAb and GO. Primary AML cells were pretreated with 5-aza (100 nM) for 48 h. After two washes, proliferation assays were performed as described in Figure 1a. Summarized data (mean) for 25 Syk-positive (a and c) and 15 Syk-negative (b and d) samples are presented. Error bars indicate the s.e.m.

Next, we analyzed the effects of the combined treatment of presence of 5-aza plus GO or anti-CD33 mAb reached 60–70% AML primary cells with 5-aza and then with either anti-CD33 and was similar in Syk-positive and Syk-negative cells. mAb or GO (Figure 5). In Syk-positive samples, pretreatment with 5-aza enhanced (about 20%) the anti-proliferative effects of antibody (Figure 5a) or GO (Figure 5c). Similar experiments Discussion with control isotype-matching anti-CD13 mAb did not result in increased growth inhibition (data not shown). Meanwhile, We demonstrate for the first time that the anti-proliferative in Syk-negative primary samples, pretreatment with 5-aza led to effects of GO in AML is associated with the expression of Syk statistically significant (Po0.05, Student’s t-test) enhancement and that 5-aza treatment of Syk-negative and, therefore, poorly of growth inhibition (up to 40% increase) induced by GO or responsive, AML cells restored sensitivity to CD33 ligation- anti-CD33 mAb (Figure 5b and d). induced cytotoxicity mediated by both GO and unconjugated We examined whether treatment with 5-aza may restore Syk anti-CD33 mAb. expression in Syk-negative AML cells through induction of Recent studies showed that ligation of CD33 molecules generalized DNA hypomethylation. In 15 Syk-negative primary mediated AML cell growth inhibition,7–10 activation of protein AML samples, five (33%) showed restored Syk expression after tyrosine phosphatases SHP-1 and SHP-2,3–5 and a recruitment of 5-aza treatment (Figure 6a and b). In Syk-positive samples, the the Syk protein tyrosine kinase.28,29 These events led to level of Syk expression was unchanged. Of note, in two of five apoptosis and inhibition of cell growth. Our previous experi- (40%) SHP-1-negative primary AML samples, 5-aza treatment ments also proved that the inhibitory effect of anti-CD33 mAb also restored SHP-1 expression (data not shown). was caused by ligation of the CD33 molecule itself rather than Treating Syk-positive AML cells with single-agent 5-aza at through Fc receptor crosslinking.41 We also demonstrated that 100 nM inhibited DNA synthesis by almost 40%, whereas the Syk expression was associated with the response of leukemia same drug concentration caused o20% inhibition in Syk- cells to the anti-proliferative effect of unconjugated anti-CD33 negative cells (Figure 6c). 5-Aza in combination with GO or mAb.29 In this study, we expanded these observations in a larger anti-CD33 caused less than additive inhibition of DNA synthesis panel of primary AML samples (n ¼ 40). Western blots detected in Syk-positive cells (Figure 6c, left panel). 25 (63%) Syk-positive and 15 (37%) Syk-negative samples. In Syk-negative cells where 5-aza did not induce re- These data confirm our previous findings that Syk was expression of Syk, the addition of GO or anti-CD33 mAb undetectable in about 30% primary AML samples and that the enhanced the anti-proliferative effect of 5-aza (Figure 6c, middle expression of Syk correlated with response to CD33 ligation.21 panel). In contrast, Syk-negative AML cells, where 5-aza Significant differences in CD33 signaling between responsive induced re-expression of Syk, were only marginally inhibited and non-responsive AML cells that expressed Syk were found by 5-aza or anti-CD33 mAb alone, whereas the combination (29 and data not shown). In CD33-responsive AML samples, produced a more than additive effect in AML cells (Figure 6c, CD33 ligation induced tyrosine phosphorylation of Syk and right panel). Even the effect of GO was almost doubled by 5-aza Syk/CD33 and Syk/SHP-1 de novo complex formation. How- in these cells. Thus, total inhibition of DNA synthesis in the ever, CD33 signaling in CD33-non-responsive AML samples

Leukemia Cytotoxicity of anti-CD33 mAb and GO correlate with Syk expression L Balaian and ED Ball 2099 b a Syk-/-, n=10, 25% 48h)

no treatment5-aza (100nM AML #13 – (1 of 25 Syk+) + Syk , -/+ Syk , AML #9 – (1 of 5 Syk-/restored) n=25, 62% n=5, 13% AML #11 – (1 of 10 Syk-/non-restored)

c I.Syk + (n=25;63% of total) II.Syk -/- (n=10;25% of total) III.Syk -/+(n=5; 13% of total) * 80 80 80 * * * 60 60 60

40 40 40

(% of control) 20 20 20

Inhibition of proliferation 0 0 0 none a-CD13 a-CD33 GO none a-CD13 a-CD33 GO none a-CD13 a-CD33 GO no treatment 5-aza (100 nM 48 h)

Figure 6 5-Aza treatment of Syk-negative primary AML cells increases response to anti-CD33 mAb and GO. (a and b) Western blot data. At 48 h, cells were harvested and analyzed by Western blotting for Syk expression. SykÀ/À is an abbreviation for Syk-negative samples where Syk was not restored by 5-aza. SykÀ/ þ is an abbreviation for Syk-negative samples where Syk was restored by 5-aza treatment. (c) Proliferation assay. Primary AML cells were treated with 5-aza for 48 h, washed and cultured in the presence or absence of anti-CD33 or CD13 mAb (0.1 mg/ml) or GO (0.01 mg/ml) for 48 h and the proliferation assay performed as described in Figure 1. Bars marked with asterisk have a statistically signiﬁcant difference between groups (Po0.05, Student’s t-test).

was different from responsive samples and from each other. In therefore, increase the efficacy of GO or unconjugated anti- AML sample No. 18, CD33 ligation did not induce tyrosine CD33 mAb. phosphorylation or recruitment of Syk or Syk/SHP-1 complex In order to augment the anti-proliferative effects of GO and formation. In AML sample No. 19, CD33 ligation induced anti-CD33 mAb, we considered methods of upregulating Syk tyrosine phosphorylation of Syk. However, no Syk/SHP-1 or Syk/ expression and/or increasing (restoring) its activity. Absence of CD33 complex formation was detected. Syk expression, mediated by hypermethylation, is linked to cell These differences in CD33 signaling confirmed our previous unresponsiveness to a variety of treatments in many tumors.33 In findings29 and revealed the importance of not only the presence many studies, 5-aza re-established Syk expression and activ- of Syk, but its functional activity. These data altogether prove the ity,36,37 leading to restoration of tumor cell responsiveness to significance of Syk in CD33 signaling. treatments. That prompted us to investigate whether or not the response of Although 5-aza effects are not specific for Syk, it may be AML cells to GO is also dependent on Syk expression. It is even important as many other critical genes in addition to Syk are more intriguing as 50% of GO is not conjugated to calichea- silenced by methylation and combined loss of these genes mycin (package insert), and hence some of its activity is in fact promotes tumor progression.42 Moreover, although 5-aza was due to anti-CD33 mAb signaling. Our results demonstrated a recently approved for the treatment of myelodysplastic syn- strong correlation between the response of AML cells to low drome and has been previously studied as an anti-AML therapy, doses of GO and Syk expression. Higher doses of GO there is not much known about its effects on proliferation of diminished differences between Syk-positive and Syk-negative AML at low doses in vitro. samples, possibly owing to the effects of free toxin. Moreover, in After trying various doses and time periods for treatment, we Syk-positive samples, the number of responders was signifi- have chosen ‘the optimal’ condition for our experiments. 5-Aza cantly higher compared to Syk-negative samples. These data (100 nM) for 48 h resulted in sub-maximal inhibition of suggested a correlation between Syk expression and the proliferation both in cell lines and primary AML cells. Moreover, response of AML cells to Mylotarg. To confirm Syk engagement it led to the restoration of Syk expression in five of 15 (33%) Syk- in CD33 signaling in anti-proliferative effects of both unconju- negative samples. In many studies, 5-aza was used in much gated anti-CD33 mAb and Mylotarg, we performed ‘blocking’ more higher concentrations for complete de-methylation, but experiments with Syk siRNA. Seventy-two hours after transfec- the purpose of our investigation was to use the minimal dose tion, Syk was undetectable by Western blot and HL-60 leukemia (in order to exclude the toxicity) which still demonstrated cells (originally responsive to anti-CD33 mAb cytotoxic activity) anti-proliferative effects on AML cells. became unresponsive. The anti-proliferative effect of GO in Based on response of primary AML cells to 5-aza, we were transfected cells also diminished. These data demonstrate the able to divide them into two groups: responders (inhibition importance of Syk in CD33- or GO-mediated cytotoxicity. 425%) and non-responders (inhibition o25%). To our surprise, As GO is widely used in the treatment of AML,10 but effective there was a correlation between Syk expression and the only in one-third of patients, understanding the mechanisms of response of AML cells to 5-aza alone. Syk-positive cells were its cytotoxicity is important. We hypothesize that determining more sensitive to inhibitory effects of 5-aza. The mechanisms Syk expression before treatment can serve as a biomarker and, underlying this correlation remain to be determined.

Leukemia Cytotoxicity of anti-CD33 mAb and GO correlate with Syk expression L Balaian and ED Ball 2100 Consecutive treatment of Syk-positive cells with 5-aza and 12 Qin S, Yamamura H. Up-regulation of Syk activity during HL60 then GO or anti-CD33 mAb resulted in 20% enhancement of cell differentiation into granulocyte but not into monocyte/ inhibition. Similar treatment of Syk-negative cells led to 40% macrophage-lineage. Biochem Biophys Res Commun 1997; 236: augmentation of GO- and anti-CD33 mAb-mediated anti- 697–701. 13 Tsubokawa M, Tohyama Y, Tohyama K, Asahi M, Inazu T, proliferative activity. Restoration of Syk expression in some Nakamura H et al. Interleukin-3 activates Syk in a human originally Syk-negative samples allowed us to hypothesize that myeloblastic leukemia cell line, AML-193. Eur J Biochem 1997; the mechanism of this phenomenon was caused by de- 249: 792–796. methylation of Syk, SHP-1 (data not shown), and possibly other 14 Raeder EM, Mansfield PJ, Hinkovska-Galcheva V, Shayman JA, molecules. Importantly, the total inhibition of proliferation in Boxer LA. Syk activation initiates downstream signaling Syk-negative (less sensitive) samples after combined treatment events during human polymorphonuclear leukocyte phagocytosis. J Immunol 1999; 163: 6785–6793. was comparable to results of inhibition in Syk-positive samples. 15 Cambien B, Pomeranz M, Millet MA, Rossi B, Schmid-Alliana A. That is, non-responsive Syk-negative samples after 5-aza treat- Signal transduction involved in MCP-1-mediated monocytic ment became responsive to anti-proliferative effects of anti- trans-endothelian migration. Blood 2001; 97: 359–366. CD33 mAb and Mylotarg. 16 Nakashima K, Kokubo T, Shichijo M, Li YF, Yura T, Yamamoto N. We suggest that for more effective treatment of AML by GO, A novel Syk kinase-selective inhibitor blocks antigen presentation other anti-CD33 mAb and 5-aza: (1) pretreatment levels of Syk of immune complexes in dendritic cells. Eur J Pharmacol 2004; expression can be used as a prognostic marker for response, and 505: 223–228. 17 Shen L, Lang ML, Wade WF. The ins and outs of getting (2) in Syk-negative patients, the level of Syk expression can be in: structures and signals that enhance BCR or Fc receptor- restored by 5-aza leading to improved therapeutic response to mediated antigen presentation. Immunopharmacology 2000; 49: anti-CD33-mediated therapy. A clinical trial testing these 227–240. hypotheses is underway at our institution. 18 Dustin LB, Plas DR, Wong J, Hu YT, Soto C, Chan AC et al. Expression of dominant-negative src-homology domain 2-containing protein tyrosine phosphatase –1 results in increased Syk Acknowledgements tyrosine kinase activity and B cell activation. J Immunol 1999; 162: 2717–2724. 19 Kraft S, Katoh N, Novak N, Koch S, Bieber T. Unexpected Larisa Balain designed and performed all of the experiments and functions of FcepsilonRI on antigen-presenting cells. Int Arch prepared the paper. Edward D Ball co-designed the experiments Allergy Immunol 2001; 124: 35–37. and co-wrote the paper. 20 Hamawy MM, Fischler C, Zhang J, Siraganian RP. Fc epsilon RI aggregation induces tyrosine phosphorylation of a novel 72 kDa protein downstream of Syk. Biochem Biophys Res Commun 1997; References 239: 670–675. 21 Xu R, Seger R, Pecht I. Cutting edge: extracellular signal-regulated 1 Griffin JD, Linch D, Sabbath K, Larcom P, Schlossman SF. A kinase activates syk: a new potential feedback regulation of Fc monoclonal antibody reactive with normal and leukemic human epsilon receptor signaling. J Immunol 1999; 163: 1110–1114. myeloid progenitor cells. Leukemia Res 1984; 8: 521–527. 22 Ma H, Yankee TM, Hu J, Asai DJ, Harrison ML, Geahlen RL. 2 Sievers EL, Appelbaum FR, Spielberger RT. Selective ablation of Visualization of Syk-antigen receptor interactions using green acute myeloid leukemia using antibody-targeted chemotherapy: a fluorescent protein: differential roles for Syk and Lyn in the phase I study of an anti-CD33 calicheamycin immunoconjugate. regulation of receptor capping and internalization. J Immunol Blood 1999; 93: 3678–3684. 2001; 166: 1507–1516. 3 Freeman SD, Kelm S, Barber EK, Crocker PR. Characterization of 23 Ebel C, Schmidt RE, Hundt M. Signal transduction via both human CD33 as a new member of the sialoadhesin family of cellular low-affinity IgG Fc receptors, Fc gamma RIIa and Fc gamma RIIIb, interaction molecules. Blood 1995; 85: 2005–2012. depends on the activity of different families of intracellular kinases. 4 Ulyanova T, Blasioli J, Woodford-Thomas TA, Thomas ML. The Immunobiology 2001; 203: 616–628. sialoadhesin CD33 is a myeloid-specific inhibitory receptor. Eur J 24 Law CL, Sidorenko SP, Chandran KA, Zhao Z, Shen SH, Fischer EH Immunol 1999; 29: 3440–3449. et al. CD22 associates with protein tyrosine phosphatase 1C, Syk 5 Taylor VC, Buckley CD, Douglas M, Cody AJ, Simmons DL, and phospholipase C-gamma 1 upon B cell activation. J Exp Med Freeman SD. The myeloid-specific scialic acid-binding receptor, 1996; 183: 547–560. CD33, associates with the protein tyrosine phosphatases, SHP-1 25 Tuscano JM, Engel P, Tedder TF, Agarval A, Kehrl JH. Involvement and SHP-2. J Biol Chem 1999; 274: 11505–11512. of p72syk kinase, p53/56lyn kinase and phosphatidyl inositol-3 6 Paul SP, Taylor LS, Sansbury EK, McVicar DW. Myeloid specific kinase in signal transduction via the human B lymphocyte antigen human CD33 is an inhibitory receptor with differential ITIM CD22. Eur J Immunol 1996; 26: 1246–1252. function in recruiting the phosphatases SHP-1 and SHP-2. Blood 26 Yohannan J, Wienands J, Coggeshall KM, Justement LB. Analysis of 2000; 96: 483–490. tyrosine phosphorylation-dependent interactions between stimu- 7 Vitale C, Romagnani C, Falco M, Ponte M, Mingari MC. latory effector proteins and the B cell co-receptor CD22. J Biol Engagement of p75/AIRM1 or CD33 inhibits the proliferation of Chem 1999; 274: 18769–18776. normal or leukemic myeloid cells. Proc Natl Acad Sci USA 1999; 27 Otipoby KL, Draves KE, Clark EA. CD22 regulates B cell receptor- 96: 15091–15096. mediated signals via two domains that independently recruit Grb2 8 Vitale C, Romagnani C, Puccetti A, Olive D, Mingari MC. Surface and SHP-1. J Biol Chem 2001; 276: 44315–44322. expression and function of p75/AIRM-1 or CD33 in acute myeloid 28 Balaian L, Ball ED. Direct effect of bispecific anti-CD33 Â anti- leukemias: engagement of CD33 induces apoptosis of leukemic CD64 antibody on proliferation and signaling in myeloid cells. cells. Proc Natl Acad Sci USA 2001; 98: 5764–5769. Leukemia Res 2001; 25: 1115–1125. 9 Mingari MC, Vitale C, Romagnani C, Falco M, Moretta L. p75/ 29 Balaian L, Zhong RK, Ball ED. The inhibitory effect of anti-CD33 AIRM1 and CD33, two sialoadhesin receptors that regulate the monoclonal antibodies on AML cells growth correlates with Syk proliferation or the survival of normal and leukemic myeloid cells. and/or ZAP-70 expression. Exp Hematol 2003; 5: 363–371. Immunol Rev 2001; 181: 260–268. 30 Goodman PA, Wood CM, Vassilev A, Mao C, Uckun FM. Spleen 10 Bae J, Martinson JA, Klingemann HG. Heteroclitic CD33 peptide tyrosine kinase (Syk) deficiency in childhood pro-B cell acute with enhanced anti-acute myeloid leukemic immunogenicity. Clin lymphoblastic leukemia. Oncogene 2001; 20: 3969–3978. Cancer Res 2004; 10: 7043–7052. 31 Goodman PA, Burkhardt N, Juran B, Tibbles HE, Uckun FM. 11 Sievers EL. Targeted therapy of acute myeloid leukemia with mono- Hypermethylation of the spleen tyrosine kinase promoter in clonal antibodies and immunoconjugates. Cancer Chemother T-lineage acute lymphoblastic leukemia. Oncogene 2003; 22: Pharmacol 2000; 46: 18–22. 2504–2514.

Leukemia Cytotoxicity of anti-CD33 mAb and GO correlate with Syk expression L Balaian and ED Ball 2101 32 Ushmorov A, Leithauser F, Sakk O, Weinhausel A, Popov S, 37 Yuan Y, Liu H, Sahin A, Le Dai J. Reactivation of syk expression by Moller P et al. Epigenetic process play a major role in B-cell- inhibition of DNA methylation suppresses breast cancer cell speciﬁc gene silencing in classical Hodgkin lymphoma. Blood invasiveness. Int J cancer 2005; 113: 654–659. 2006; 107: 2493–2500. 38 Jabbour EJ, Giles EJ. New agents in myelodysplastic syndromes. 33 Coopman PJ, Mueller SC. The Syk tyrosine kinase: a new negative Curr Hematol Rep 2005; 4: 191–199. regulator in tumor growth and progression. Cancer Lett 2006; 241: 39 Fenaux P. Inhibitors of DNA methylation: beyond myelodysplastic 159–173. syndromes. Natl Clin Pract Oncol 2005; 2: 836–844. 34 Stewart ZA, Pieterpol JA. Syk: a new player in the ﬁeld of breast 40 Claus L, Almstedt M, Lubbert M. Epigenetic treatment of cancer. Breast Cancer Res 2001; 3: 5–7. hematopoietic malignancies: in vivo targets of demethylating 35 Coopman PJ, Do MT, Barth M, Bowden ET, Hayes AJ, agents. Semin Oncol 2005; 32: 511–520. Basyuk E et al. The Syk tyrosine kinase suppresses malignant 41 Balaian L, Ball ED. Anti-CD33 monoclonal antibodies enhance the growth of human breast cancer cells. Nature 2000; 406: cytotoxic effects of cytosine arabinoside and idarubicin on acute 742–747. myeloid leukemia (AML) cells through similarities in their signaling 36 Yuan Y, Mendez R, Sahin A, Dai JL. Hypermethylation leads to pathways. Exp Hematol 2005; 33: 199–211. silencing of the SYK gene in human breast cancer. Cancer Res 42 Esteller M. DNA methylation and cancer therapy: new develop- 2001; 61: 5558–5561. ments and expectations. Curr Opin Oncol 2005; 17: 55–60.

Leukemia