Reduction of phosphoinositide-phospholipase C beta1 methylation predicts the responsiveness to azacitidine in high-risk MDS

Abstract

Lipid signaling pathways are involved in cell growth, differentiation, and apoptosis, and could have a role in the progression of myelodysplastic syndromes (MDS) into acute myeloid leukemia (AML). Indeed, recent studies showed that phosphoinositide-phospholipase (PI-PL)Cbeta1 mono-allelic deletion correlates with a higher risk of AML evolution. Also, a single patient treated with azacitidine, a DNA methyltransferase inhibitor currently used in MDS, displayed a direct correlation between PI-PLCbeta1 gene expression and drug responsiveness. Consequently, we hypothesized that PI-PLCbeta1 could be a target for demethylating therapy. First, we analyzed the structure of PI-PLCbeta1 gene promoter, then quantified the degree of PI-PLCbeta1 promoter methylation and gene expression in MDS patients at baseline and during azacitidine administration. Indeed, PI-PLCbeta1 mRNA increased in responder patients, along with a reduction of PI-PLCbeta1 promoter methylation. Also, the molecular response correlated to and anticipated the clinical outcome, thus suggesting that PI-PLCbeta1 gene reactivation could predict azacitidine responsiveness. Our results demonstrate not only that PI-PLCbeta1 promoter is hypermethylated in high-risk MDS patients, but also that the amount of PI-PLCbeta1 mRNA could predict the clinical response to azacitidine, therefore indicating a promising new therapeutic approach.

Promoter hypermethylation of CpG nucleotides in DNA is believed to be a key contributor to the molecular pathophysiology of myelodysplastic syndromes (MDS) by inactivating genes involved in the control of normal cell growth, differentiation, and apoptosis (1–3). Currently, two demethylating agents, azacitidine (Vidaza, Celgene Corp) and decitabine (Dacogen, Eisai Inc), have been approved for the MDS treatment. Both of these drugs are cytosine nucleoside analogues capable to induce a hematologic response, as compared with best supportive care treatments alone, delay the evolution into acute myeloid leukemia (AML) and prolong the overall survival (4–6).

The molecular mechanisms underlying the effect of the epigenetic therapies are not completely understood, even though it is clear that the DNA methyltransferase inhibitors can induce the reexpression of methylated silenced gene products (7). After incorporation of demethylating agents into DNA, the methyltransferases are inhibited, but complete demethylation occurs only after several cycles of replication, thus accounting for time to response to these drugs (8). In MDS, demethylating treatments have been proven to induce the reactivation of a hypermethylated p15/INK4B gene (9), as well as of other genes, such as p21WAF/Cip1 and p73 (10). However, almost all cancer-related signaling pathways may be affected by hypermethylation, and a growing number of silenced methylated genes involved in each major type of cancer is coming to light (11). Therefore, it is increasingly important to find out other molecular targets for azacitidine, because some of them could become new biomarkers to monitor the clinical response to demethylating therapies.

Lipid signaling pathways are implicated in many cellular processes, in normal and pathological conditions (12–14). Indeed, lipid signaling in disease is an emerging field of investigation, several pathways being involved in the molecular mechanisms leading to cancer, such as leukemogenesis (15, 16). Nuclear inositide signaling is essential in the control of cell growth and differentiation, and nuclear phosphoinositide-phospholipase (PI-PL)Cbeta1 appears as one of the main players of these signal transduction processes. Up to now, our group has shown that PI-PLCbeta1 nuclear localization is crucial for its function (12, 17, 18). In fact, a role for nuclear PI-PLCbeta1 in cell cycle control has been established, in the G1 progression (19), as well as in the G2/M transition (20). Also, it has been demonstrated that PI-PLCbeta1 targets either cyclin D3/cdk4 (21) or CD24 (22), which have a role in both stem cell proliferation and the early stages of the hematopoietic differentiation (23, 24).

Recent findings led us to hypothesize a role for PI-PLCbeta1 signal transduction pathways in the MDS progression towards AML (25, 26). In fact, we showed not only that PI-PLCbeta1 monoallelic deletion is associated with a higher risk of AML evolution (27), but also that MDS patients, bearing the deletion or not, usually have an altered PI-PLCbeta1 gene expression (28). Also, in a single high-risk MDS patient, we observed that azacitidine induced an increase in PI-PLCbeta1 mRNAs, along with a down-regulation of activated Akt (29), whose phosphorylation has been linked to a higher risk of evolution into AML (30, 31), and we showed that the PI-PLCbeta1 gene expression was directly related to the clinical response to treatment.

In the present study, we focused on the role of PI-PLCbeta1 methylation in MDS patients. First, we examined whether PI-PLCbeta1 promoter region is hypermethylated in MDS cells; then, we quantified the level of PI-PLCbeta1 promoter methylation and gene expression at baseline and during azacitidine therapy; and last, we investigated the connection between PI-PLCbeta1 promoter methylation and gene expression with the clinical response to azacitidine.

Results

Patient Characteristics.

Bone marrow mononuclear cells (BMMNCs) or peripheral blood (PB)MNCs from 24 patients affected by MDS (18 treated with azacitidine, and 6 with best supportive care only) were examined. Median age was 70 years (range, 50 to 84 years). MDS was diagnosed following both French-American-British (FAB) and World Health Organization (WHO) classifications (32, 33), whereas, according to the International Prognostic Scoring System (IPSS) (34), patients were divided into two risk subgroups: intermediate 2 risk (n = 16) and high risk (n = 8). However, throughout the text, all of the MDS patients are named as high-risk MDS. Patient demographics and disease characteristics are summarized in Table 1.

Table 1.

Clinical, hematologic, and cytogenetic characteristics of the MDS patients

	Diagnosis		IPSS	Screening	Age	Karyotype	PI-PLCbeta1 methylation	PI-PLCbeta1 gene expression	Clinical outcome	Death	Treatment	Response
	FAB	WHO
1*	RAEB	RAEB 2	INT 2	03/2006	84	+8	−45.79	+42.03	Major HI, relapse, death	Cachexia	AZA	Y
2	RAEB	RAEB 2	INT 2	08/2004	60	46, XY	−22.30	+15.20	CR	—	AZA	Y
3*	RAEB	RAEB 2	High	06/2006	80	Complex	−11.51	+71.62	PR, relapse, AML, death	Cachexia	AZA	Y
4*	RAEB	RAEB 2	High	01/2005	62	Complex	−40.25	+33.98	Major HI, relapse, AML, death	Infection	AZA	Y
5*	RAEB	RAEB 2	INT 2	02/2002	62	46, XY	+2.84	−1.61	Major HI, relapse, death	Infection	AZA	Y
6*	RAEB	RAEB 2	INT 2	06/2004	63	t(2;3)	+89.45	−38.98	Stable disease, AML, death	Hemorrage	AZA	N
7*	RAEB T	AML	INT 2	06/2006	79	46, XY	+40.36	−18.02	Stable disease, disease progression, death	Cachexia	AZA	N
8*	RAEB	RAEB 2	High	03/2005	68	+8	+32.23	−13.60	Stable disease, AML, death	Cachexia	AZA	N
9	RAEB	RAEB 2	INT 2	01/2007	80	+8	−32.73	+11.08	Major HI	—	AZA	Y
10*	RA	RCMD	INT 2	09/2004	50	−7	−13.30	+17.49	Major HI, disease progression, death	Septic shock	AZA	Y
11*	RAEB T	AML	INT 2	11/2005	78	46, XY	+20.89	−26.88	Stable disease, death	Infection	AZA	N
12	RAEB	RAEB 1	INT 2	10/2005	78	+21	+84.21	−41.74	Stable disease	—	AZA	N
13*	RAEB	RAEB 2	INT 2	01/2007	75	46, XY	−13.60	+13.44	HI, death	Hemorrage	AZA	Y
14	RAEB	RAEB 2	INT 2	11/2006	65	46, XY	+54.40	−13.18	Stable disease, AML	—	AZA	N
15	RAEB	RAEB 2	High	09/2007	77	46, XY	+39.57	−15.09	Stable disease	—	AZA	N
16	RAEB	RAEB 2	High	04/2007	68	46, XY	+36.62	−23.61	Stable disease	—	AZA	N
17	RAEB	RAEB 2	High	04/2007	69	46, XX	−40.75	+47.11	HI	—	AZA	Y
18	RAEB	RAEB 2	High	04/2007	69	46, XX	−38.47	+37.99	HI	—	AZA	Y
19*	RA	RCMD	INT 2	09/2004	50	−7	+3.80	−2.01	Disease progression, death	Infection	BSC	—
20*	RAEB	RAEB 2	INT 2	11/2005	78	46, XY	−2.91	+1.10	Stable disease, death	Infection	BSC	—
21	RAEB	RAEB 1	INT 2	10/2005	78	+21	+3.22	−3.48	Stable disease	—	BSC	—
22	CMML	CMML	INT 2	02/2005	64	+8	−1.91	+3.17	AML	—	BSC	—
23	RAEB T	AML	High	06/2005	75	+8	−3.75	+2.38	AML	—	BSC	—
24	RAEB	RAEB 1	INT 2	04/2002	71	46, XY	−2.55	+3.05	AML	—	BSC	—

RA, refractory anemia; RARS, RA with ringed sideroblasts; RAEB, RA with excess of blasts; RAEBT, RAEB in transformation; RCMD, refractory cytopenia with multilineage dysplasia; CMML, chronic myelomonocytic leukemia; INT, intermediate; HI, hematologic improvement; CR, complete remission; PR, partial remission; AZA, azacitidine; BSC, best supportive care.

*Patients deceased during follow-up.

Study Group and Clinical Response.

Between February 2002 and December 2008, 18 high-risk MDS patients were treated with azacitidine, receiving a median number of 17.5 (range, 10 to 49) courses. According to the revised International Working Group (IWG) criteria (35), 10 patients (55%) showed a favorable response to treatment: complete remission (one patient), partial remission (one patient), and major hematologic improvements (eight patients). The remaining eight patients did not respond to azacitidine.

PI-PLCbeta1 Putative Promoter Region Analysis.

By using the Methyl Primer Express Software (v.1.0; Applied Biosystems), two regions were identified as putative CpG islands (Fig. 1A). The first one (named as island 1) was localized between −18301/−13320, whereas the second one (named as island 2) was found in the region −2340/−1, considering +1 as the translation start site. Genomic DNA was examined in these two regions, which were first bisulfite-modified and subsequently analyzed by real-time methylation-specific PCR (MSP), to investigate the status of methylation in MDS patients, as well as in healthy subjects. Only the island 2 was differentially methylated; therefore, our subsequent analyses focused on this region (Fig. 1B). For samples with enough cells to analyze, PI-PLCbeta1 island 2 methylation status was also studied in CD34+ cells, but the levels of PI-PLCbeta1 promoter methylation were not significantly different from the results obtained from the analysis of the total MNCs (Fig. 1C).

Fig. 1.

PI-PLCbeta1 promoter methylation. (A) Structure of the PI-PLCbeta1 promoter region, showing the two putative CpG islands and the regions analyzed by real-time MSP. (B) Degree of methylation of the two putative CpG islands in healthy subjects and MDS patients. Only the island 2 is highly methylated in MDS cases, as compared with healthy subjects. (C) Quantification of the PI-PLCbeta1 island 2 methylation status in CD34+ cells and total MNCs from healthy donors and MDS patients. There is no significant difference between PI-PLCbeta1 promoter methylation levels in CD34+ cells or total MNCs within each sample, as compared with the universally methylated DNA. (D) Quantification of the PI-PLCbeta1 island 2 methylation status in cell lines, as compared with healthy subjects and the universally methylated DNA. HL60 acute promyelocytic leukemia cell line retains the highest degree of methylation, whereas U937 monocytic cell line shows the lowest degree of methylation. Results are the mean of three independent experiments ± SD. **, P < 0.01 vs. methylated DNA.

PI-PLCbeta1 Methylation Status in HL60 and U937 Cell Lines.

Several cell lines were bisulfite-modified and the status of PI-PLCbeta1 methylation was examined. Methylation was calculated as a fraction of the total PI-PLCbeta1 (methylated and unmethylated); therefore, the universally methylated DNA (Chemicon), after bisulfite conversion, was used as a positive control and considered as 100% for the methylated allele and 0% for the unmethylated allele. HL60 and U937 cell lines showed the higher and the lower level of methylation in the island 2; thus, becoming our positive and negative control in all of the subsequent experiments (Fig. 1D).

PI-PLCbeta1 Gene and Protein Expression in HL60 and U937 Cell Lines.

The amount of basal PI-PLCbeta1 promoter methylation was inversely correlated with both gene and protein expression. In fact, semiquantitative real-time PCR experiments showed that HL60 cells had the highest amount of PI-PLCbeta1 promoter methylation and low levels of mRNAs, whereas U937 cells had the lowest promoter methylation and higher PI-PLCbeta1 mRNA levels (Fig. S1A). Immunocytochemical experiments and Western blot analyses confirmed the high amount of PI-PLCbeta1 protein in U937 cells, and barely detectable protein levels in HL60 cells (Fig. S1 B and C).

Effect of Azacitidine on Cell Viability and Apoptosis in Cell Lines.

To determine whether azacitidine treatment induced either diminished survival or increased apoptosis, HL60 and U937 cells were exposed to increasing concentrations of the drug for 24 h, and viability was evaluated by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) assays. Azacitidine induced a dose-dependent decrease in viability in both cell lines: at 24 h, the IC₅₀, calculated by the Calcusyn Software (v.2.0; Biosoft), was 8.55 μM for HL60 cells, whereas it was 21.82 μM for U937 cells (Fig. S2A).

It was also investigated whether the decreased survival was related to apoptosis. After treatment with 5 μM azacitidine for 24 h, ≈45% of HL60 cells were apoptotic/necrotic. As for U937 cells, after incubation with 20 μM azacitidine for 24 h, ≈34% of the cells were apoptotic/necrotic, whereas 5 μM azacitidine for 24 h induced lower levels of apoptosis, because only 10% of the cells were apoptotic/necrotic (Fig. S2B).

Therefore, because HL60 cells were affected by azacitidine concentrations of 5 μM for 24 h, we used these conditions for all subsequent experiments, thus considering HL60 cells as a positive control and U937 as a negative control, not being affected by azacitidine at those concentrations.

Azacitidine Induces PI-PLCbeta1 DNA Hypomethylation, Gene, and Protein Expression in Cell Lines.

HL60 and U937 cells were incubated for 24 h with 5 μM azacitidine, and the methylation levels were measured by real-time MSP: azacitidine-induced PI-PLCbeta1 DNA hypomethylation was significant in HL60 cells, but not in U937 cells, as expected at these concentrations; therefore, these two cell lines were considered as the positive and negative control for all of the subsequent methylation experiments (Fig. S3A). As for PI-PLCbeta1 gene and protein expression, semiquantitative real-time PCR, immunocytochemical experiments and Western blot analyses confirmed that azacitidine-induced PI-PLCbeta1 reactivation was significant in HL60 cells, but not in U937 cells, as expected at these concentrations (Fig. S3). In particular, real-time PCR analyses were initially carried out on both PI-PLCbeta1a and PI-PLCbeta1b. However, because azacitidine specifically induced a significant increase in PI-PLCbeta1b mRNA (Fig. S3C), our subsequent analyses focused on the nuclear splicing variant.

Azacitidine Induces In Vitro Methylated PI-PLCbeta1 Reactivation.

To assess whether the PI-PLCbeta1 island 2 is a specific target for demethylating agents, the activity of the WT (unmethylated) PI-PLCbeta1 promoter/secreted alkaline phosphatase (SEAP) reporter plasmid and the in vitro methylated PI-PLCbeta1 promoter/SEAP reporter plasmid were investigated (Fig. S4). HL60 cells were transiently transfected with 1 μg/well of the WT (unmethylated) PI-PLCbeta1 promoter/SEAP reporter plasmid, in vitro methylated PI-PLCbeta1 promoter/SEAP reporter plasmid, mock pSEAP2-Basic vector, and pmaxGFP control plasmid. This latter was useful to evaluate the transfection efficiency, which was ≈80% for all of the experiments. The activity of the empty pSEAP2-Basic vector was ≈10%. In basal conditions, after transfection with the WT (unmethylated) PI-PLCbeta1/SEAP plasmid, the promoter activity increased to ≈48%, whereas the in vitro methylated PI-PLCbeta1/SEAP vector displayed lower levels, comparable with the mock pSEAP2-Basic vector. After azacitidine treatment, the promoter activity increased in both reactions, reaching a statistically significant level of ≈80% for the in vitro methylated PI-PLCbeta1/SEAP vector (Fig. S4).

Quantification of PI-PLCbeta1 Promoter Methylation in MDS Patients During Azacitidine Treatment.

Genomic DNA from healthy subjects and MDS samples was extracted and bisulfite-modified, so as to analyze the status of promoter methylation before and during azacitidine administration. As compared with the pre-treatment amount, responder patients demonstrated a marked decrease in PI-PLCbeta1 promoter methylation, whereas patients who were refractory to azacitidine displayed increasing levels of promoter methylation (Fig. 2A, Tables 1 and 2). The differences between PI-PLCbeta1 promoter methylation and baseline, calculated as a percentage rate on the pre-treatment level of the same patient, were statistically significant (Student's t test, P < 0.05 vs. baseline, 95% CI −13.65 to +29.42). As for patients treated with best supportive care only, they did not show any significant change in PI-PLCbeta1 promoter methylation (Student's t test, P > 0.1, 95% CI −4.15 to +2.79). The differences between baseline and the median levels of PI-PLCbeta1 promoter methylation during follow-up are reported in Table 1; positive values indicate an increase, whereas negative values indicate a reduction.

Fig. 2.

PI-PLCbeta1 promoter methylation and gene expression correlate with response to azacitidine. (A) Responder patients show a strong reduction in PI-PLCbeta1 methylation levels, as well as an induction of PI-PLCbeta1b gene expression, as compared with the pretreatment period. Nonresponder patients display a constant increase in PI-PLCbeta1 methylation, corresponding to a reduction in PI-PLCbeta1b levels during azacitidine treatment. Results are the mean of three independent experiments ± SD. (B) Immunofluorescence analysis showing the induction of PI-PLCbeta1 expression in responder patients during azacitidine therapy; on the contrary, subjects refractory to azacitidine display a decrease in PI-PLCbeta1 protein expression.

Table 2.

PI-PLCbeta1 promoter methylation and gene expression correlate with the clinical response to azacitidine

	No. cases
Responders
Decreased PI-PLCbeta1 gene expression and increased PI-PLCbeta1 methylation	1
Increased PI-PLCbeta1 gene expression and decreased PI-PLCbeta1 methylation	9
Nonresponders
Decreased PI-PLCbeta1 gene expression and increased PI-PLCbeta1 methylation	8
Increased PI-PLCbeta1 gene expression and decreased PI-PLCbeta1 methylation	0

PI-PLCbeta1 Gene and Protein Expression in MDS Patients During Azacitidine Treatment.

To determine whether PI-PLCbeta1 reactivation could be induced in vivo by azacitidine therapy, we analyzed the MDS samples at baseline and before the beginning of each single course of azacitidine. Overall, the differences between PI-PLCbeta1 gene expression and baseline, calculated as a percentage rate on the pretreatment level of the same patient, were statistically significant (Student's t test, P < 0.05 vs. baseline, 95% CI −10.55 to +21.35).

Notably, nine patients showed a significant increment of PI-PLCbeta1 gene and protein expression, with a statistically significant difference between pre and post-treatment. In contrast, nonresponders displayed a constant increase, as compared with baseline (Fig. 2, Table 2). Also, there was a statistically significant inverse correlation between PI-PLCbeta1 gene expression and promoter methylation during azacitidine administration within individual patients (paired Student's t test), whereas patients treated only with best supportive care did not show any significant alteration in PI-PLCbeta1 levels (Table 1). Together, these data indicate that PI-PLCbeta1b mRNA induction is associated with the clinical response to azacitidine.

As for PI-PLCbeta1 protein expression, because in most cases the number of MDS cells available for each sample was limited, we performed only immunocytochemical investigations. The analyses, carried out on samples before and during azacitidine therapy, confirmed the results obtained by real-time PCR experiments, i.e., during the treatment in responder patients PI-PLCbeta1 increased, whereas it lowered in nonresponders (Fig. 2B).

PI-PLCbeta1 Promoter Methylation and Gene Expression Correlate with Response.

Considering for each MDS patient the pretreatment level as the reference for both PI-PLCbeta1 promoter methylation and gene expression, we quantified the amount of these parameters before and during azacitidine therapy (Fig. 2, Tables 1 and 2). PI-PLCbeta1 promoter methylation or gene expression were calculated as a percentage rate on the basal level within each patient. The methylation status of responders was significantly decreased after azacitidine treatment, reaching a median level of ≈67% (Student's t test, P < 0.001 vs. baseline, 95% CI −39.21 to −18.28), whereas nonresponders had a significant increase of ≈40% (Student's t test, P < 0.001 vs. baseline, 95% CI +23.04 to +65.98). As for PI-PLCbeta1 gene expression, responder patients had a significant increase of ≈34% in PI-PLCbeta1b splicing variant (Student's t test, P < 0.001 vs. baseline, 95% CI +16.81 to +47.62), whereas nonresponders had an average reduction of ≈20% (Student's t test, P < 0.001 vs. baseline, 95% CI −31.32 to −11.51). Patients treated only with best supportive care showed no significant changes in both PI-PLCbeta1 methylation and gene expression.

PI-PLCbeta1 Promoter Methylation and Gene Expression as Predictive Factors of Response to Azacitidine.

Because PI-PLCbeta1 promoter methylation and gene expression showed a statistically significant inverse correlation, we hypothesized that these parameters could be useful to monitor the effect of azacitidine. Interestingly, the analysis of p15, an established target gene for demethylating agents (9), demonstrated not only that PI-PLCbeta1 was indeed demethylated during the therapy, and its gene expression was inversely correlated to the promoter methylation, but it also led us to hypothesize for PI-PLCbeta1 a predictive value in the responsiveness to azacitidine. In fact, although both p15 and PI-PLCbeta1 methylation levels coincided with the major variations of the clinical status (Fig. 3), our analyses demonstrated that changes in PI-PLCbeta1, but not p15, may anticipate the hematological response to the demethylating therapy (Fig. 4). After azactidine administration, the patient clinically showed a rapid worsening cytopenia (visible as a decrease in hemoglobin levels), followed by a favorable response (HI); at a molecular level, PI-PLCbeta1b expression initially lowered, and subsequently showed a marked increase. Indeed, the molecular response, detectable as an increase in PI-PLCbeta1b mRNA, was perceptible two months before the increase in hemoglobin.

Fig. 3.

PI-PLCbeta1 and p15 promoter methylation correlate with response to azacitidine. Responder patients show a reduction in PI-PLCbeta1 and p15 methylation, as compared with the pre-treatment period. Nonresponder patients display an increase in PI-PLCbeta1 and p15 methylation, and a worsening of the clinical outcome. Results are the mean of three independent experiments ± SD.

Fig. 4.

PI-PLCbeta1 promoter methylation and gene expression predict the responsiveness to azacitidine. PI-PLCbeta1 methylation is inversely associated with PI-PLCbeta1b mRNA level (square line vs. round line). Also, the molecular changes anticipate the clinical response (round line vs. star line), because the changes in hemoglobin levels are shifted as compared with the PI-PLCbeta1 mRNA amount. Results are the mean of three independent experiments ± SD.

Discussion

Although loss of gene function associated with gene hypermethylation is a relatively common feature in hematologic cancers, the exact biological significance of this abnormality has not been extensively studied.

Recently, our group demonstrated that the lipid signaling pathways can be involved in the progression of MDS into AML (26). Indeed, we disclosed that, in high-risk MDS patients, not only takes place an activation of the Akt/mTOR axis (30), which is also linked with a reduction of PI-PLCbeta1 expression (29), but also that PI-PLCbeta1 undergoes a cryptic monoallelic deletion, which has been associated with a worse clinical outcome, because patients bearing this cytogenetic aberration showed a higher probability of evolution into AML (27). Interestingly, in the same study, we also demonstrated that the PI-PLCbeta1 monoallelic deletion is correlated to a specific reduction of PI-PLCbeta1b splicing variant, the nuclear one, which is known to be involved in cell cycle progression mechanisms.

Azacitidine is a DNA methyltransferase inhibitor that has been proven to be effective in MDS, because it can induce hematologic responses in 50–60% of cases (4–6), delay the AML evolution, and prolong the overall survival. Nevertheless, the molecular mechanisms underlying this drug, as well as the cellular targets or the pathways involved, are still not fully understood.

Our group recently showed a significant correlation between PI-PLCbeta1 gene expression and the hematologic response to azacitidine in a single high-risk MDS patient (29), thus hinting at a role for PI-PLCbeta1 in methylation. Therefore, we hypothesized that PI-PLCbeta1 promoter could be hypermethylated and that azacitidine could affect the expression of this gene. A specific software designated two putative CpG island: real-time MSP analyses showed that the island 1 was methylated both in healthy subjects and MDS cases. In contrast, the island 2, located next to the ATG, was differentially methylated, in that both CD34+ cells and MNCs extracted from MDS patients showed higher levels of methylation, as compared with healthy subjects. Therefore, our subsequent experiments focused on the island 2, using HL60 and U937 cells as a positive and negative control for PI-PLCbeta1 methylation, respectively. Azacitidine induces both demethylation and cytotoxicity: MTT experiments, as well as apoptotic assays, showed that HL60 cells exposed to 5 μM azacitidine for 24 h were indeed demethylated with a percentage of apoptosis, which was <50%. Also, immunocytochemical experiments and Western blot analyses demonstrated that azacitidine induced PI-PLCbeta1 gene and protein expression only in HL60 cells, thus indicating that PI-PLCbeta1 could be involved in epigenetic processes. Therefore, to assess whether PI-PLCbeta1 is a specific target for azacitidine, the sequence of the putative island 2 has been cloned into a reporter system, containing the SEAP gene. In basal conditions, the activity of the WT (unmethylated) island 2 was higher than the in vitro methylated vector; after azacitidine treatment, the in vitro methylated promoter was demethylated, leading to an increase in the expression of the SEAP gene.

Starting from the hypothesis that azacitidine targets PI-PLCbeta1 promoter gene and induces a gene reactivation, we subsequently analyzed the status of PI-PLCbeta1 promoter methylation, gene and protein expression in 24 MDS patients, 18 under therapy with azacitidine and 6 treated with best supportive care only. Our data confirmed that the amount of PI-PLCbeta1 during azacitidine treatment is inversely associated with the degree of promoter methylation and correlates with the clinical outcome, in that subjects showing a demethylation of PI-PLCbeta1 promoter had a reactivation of PI-PLCbeta1b mRNA and a favorable clinical response. Also, we demonstrated not only that PI-PLCbeta1, as well as p15, could be used to monitor the effect of azacitidine, but also that the PI-PLCbeta1 molecular response may anticipate the clinical outcome, because the changes in PI-PLCbeta1b may be detected some cycles before the hematological response.

Here, PI-PLCbeta1 has been demonstrated to be hypermethylated, and therefore, might be considered as a target for demethylating therapies. Also, we showed not only that the induction of PI-PLCbeta1, and particularly of its nuclear splicing variant PI-PLCbeta1b, directly correlates with a favorable clinical outcome, but also that the evaluation of PI-PLCbeta1 could predict the hematological response. Therefore, these data open up a number of previously undescribed lines of investigation relevant for understanding the molecular pharmacology of azacitidine and possibly of other demethylating agents in MDS.

Materials and Methods

Patient Characteristics.

BM and PB samples came from 24 MDS patients and 10 healthy normal volunteers who had given informed consent according to the Declaration of Helsinki. All of the samples came from the Department of Hematology and Medical Oncology “L. e A. Seràgnoli” of the Policlinico S. Orsola-Malpighi, Bologna, Italy. Patients were observed between 2002 and 2008. The median follow-up was 33 months (range, 13 to 52 months). In all of the subjects participating in this study, MDS diagnosis was defined according to both FAB (32) and WHO classifications (33), whereas the IPSS was used to divide the patients into intermediate 2- and high-risk group (34). However, throughout the text, all of the MDS patients are named as high-risk MDS. As initially, FAB classification (32) was used, three patients with refractory anemia with excess of blasts in transformation were included, subsequently diagnosed as AML, after WHO classification (33). Eighteen of the MDS patients underwent s.c. azacitidine treatment (75 mg/m2/day for 7 days in each 28-day cycle), whereas six patients received best supportive care only.

Evaluation of Response.

The clinical response or treatment failure were defined in accordance with the standardized IWG response criteria for MDS (35). Hematologic response was defined as complete remission, partial remission, or hematologic improvement. Other outcomes included stable disease, failure, relapse, disease progression, or disease transformation into AML.

Isolation of Mononuclear Cells and CD34+ Cells from BM or PB Samples.

For in vitro experiments, BMMNCs or PBMNCs were isolated by Ficoll-Paque (GE Healthcare) density-gradient centrifugation, according to the manufacturer's protocol. All analyses were performed on samples from patients at baseline, and subsequently before each cycle of azacitidine administration. CD34+ cells came from total MNCs after immunomagnetic separation using the CD34 Micro Beads kit (Miltenyi Biotec) following the manufacturer's instructions.

For antibodies and reagents, tissue cell cultures, cell viability analysis, annexin V–FITC/PI staining, immunocytochemistry, and Western blot analyses, see SI Methods.

Nucleic Acids Extraction.

Genomic DNA and total RNA were isolated from cell lines, CD34+, and total MNCs from MDS patients and healthy subjects by using the QIAamp DNA Blood Mini Kit and the RNeasy Mini Kit (both from Qiagen), according to the manufacturer's protocol, and RNA was retro-transcribed as previously described (27).

Bisulfite DNA Modification and Real-Time MSP.

DNA was bisulfite-converted by the MOD50 kit (Sigma–Aldrich), according to the manufacturer's protocol. PI-PLCbeta1 promoter methylation was quantified by a standard SYBR-Green real-time MSP method with the GAPDH as a reference gene. MSP primers were designed by the Methyl Primer Express Software (v.1.0) to distinguish between methylated and unmethylated PI-PLCbeta1 DNA sequences in a specific region of the island 2 (−705 to −586), chosen by the software as the best to be analyzed, as well as methylated and unmethylated p15 (Table S1). Universally methylated DNA (Chemicon), subjected to bisulfite conversion, was used as a positive control for methylated alleles.

PI-PLCbeta1 Gene Expression Analysis.

As described elsewhere (28), PI-PLCbeta1 gene expression was quantified by a TaqMan real-time PCR method, with the GAPDH gene as the internal reference. Data were statistically analyzed by the GraphPad Prism software (v.3.0), using healthy subjects as the calibrator.

For plasmid constructs, in vitro methylation, cell transfection, and reporter gene assays, see SI Methods.