Protein Tyrosine Phosphatase Receptor Type γ Is a Functional Tumor Suppressor Gene Specifically Downregulated in Chronic Myeloid Leukemia

, Abstract Chronic myelogenous leukemia (CML) is the most common myeloproliferative disease. Protein tyrosine phosphatase receptor type γ (PTPRG) is a tumor suppressor gene and a myeloid cell marker expressed by CD34 + cells. Downregulation of PTPRG increases colony formation in the PTPRG-positive megakaryocytic cell lines MEG-01 and LAMA-84 but has no effect in the PTPRG-negative cell lines K562 and KYO-1. Its overexpression has an oncosuppressive effect in all these cell lines and is associated with myeloid differentiation and inhibition of BCR/ABL-dependent signaling. The intracellular domain of PTPRG directly interacts with BCR/ABL and CRKL, but not with signal transducers and activators of transcription 5. PTPRG is downregulated at the mRNA and protein levels in leukocytes of CML patients in both peripheral blood and bone marrow, including CD34 + cells, and is reexpressed following molecular remission of disease. Reexpression was associated with a loss of methylation of a CpG island of PTPRG promoter occurring in 55% of the patients analyzed. In K562 cell line, the DNA hypomethylating agent 5-aza-2 ′ -deoxycytidine induced PTPRG expression and caused an inhibition of colony formation, partially reverted by downregulation of PTPRG expression. These findings establish, for the first time, PTPRG as a tumor suppressor gene involved in the pathogenesis of CML, suggesting its use as a potential diagnostic and therapeutic target. Cancer Res; 8896


Introduction
Chronic myelogenous leukemia (CML), also known as chronic myeloid or chronic myelocytic leukemia, is a malignant cancer of the bone marrow myeloid lineage. It accounts for ∼15% to 20% of all cases of leukemia (1-1.5 cases per 100,000 population per year; ref. 1) and originates from a pluripotential stem cell in which a 9:22 translocation results in the production of BCR/ABL fusion protein. This has a constitutive tyrosine kinase activity and deregulates signal transduction pathways, leading to leukemia (2). Phosphorylation of key residues is required for the full transforming activity of BCR/ABL (3).
For this reason, much attention has been focused on naturally occurring negative regulators of tyrosine kinase signaling: the protein tyrosine phosphatase (PTP) family of enzymes.
The human PTP family contains 107 members, 38 of which belong to the phosphotyrosine-specific ("classic") PTP subfamily (subdivided in receptor-and nonreceptor-like types) and 61 belong to the so-called "dual-specific phosphatases." To date, only two tyrosine phosphatases, PTP1B and SHP-1, are known to dephosphorylate and partially inhibit the transformation potential of BCR/ABL (4,5). Serine-threonine phosphatase PP2A is inhibited in blast crisis CML (6). These PTPs belong to the nonreceptor class of enzymes. Recently, PTPROt, a receptor-like PTP, has been found to interfere with BCR/ABL signaling in K562 cells (7).
Here, we show that PTPRG acts as a functional tumor suppressor gene in CML, interacting with BCR/ABL and inhibiting downstream signaling events. PTPRG is specifically downregulated in peripheral blood and bone marrow leukocytes of CML patients, at least in part by a mechanism involving hypermethylation of a PTPRG CpG island located in the 5′ untranslated region. These findings imply that PTPRG might represent a potential diagnostic and therapeutic target in CML.

Tissue samples
Patients were recruited at the hematology departments of Bologna, Nuoro, and Verona, Italy among newly diagnosed CML patients during the first chronic phase. Presence of the Philadelphia chromosome and p210 BCR-ABL rearrangement was a prerequisite for enrollment. CML samples were taken at diagnosis and after the initiation of therapy with imatinib mesylate (IM). Age-and sex-matched samples from individuals diagnosed as not affected by malignant disease were used as a reference group. Analysis of the CD34 + population for the group of patients in molecular remission was not possible, as these samples were not available at the time of evaluation. Table 1 reports clinical data. Written informed consent was obtained in accordance with the Declaration of Helsinki. Cytogenetic response was classified as complete, and the patients were considered to be in complete cytogenetic remission (CCR) based on the BIOMED 2 standardized protocols.

Cells
K562 (22) and MEG-01 (23) were from American Type Culture Collection; KYO-1 (24) and LAMA-84 (25) were from DSMZ. HEK293 was purchased from Invitrogen. K562 transfectants were selected, adding 0.50 mg/mL G418 (Invitrogen) to the culture medium. All were characterized by cytogenetic analysis and antigen expression. Human samples were derived from Ficoll-purified cells (mRNA analysis) or whole blood samples (flow cytometry analysis).

Cell transfection and selection of CML cell lines
Full-length (FL) human PTPRG (26) and antisense (AS) cDNA and cell lines were described (10,27). The D1028A PTPRG cDNA was obtained by site-directed mutagenesis, wherein the Asp 1028 codon was replaced with an Ala codon (NP_002832.3) and verified by sequencing.
To avoid the source of error associated with clonal variation within cell lines, we selected a K562 clone (named B4) that maintains the capability to differentiate when treated with the well-known inducers sodium butyrate and hemin (28,29) and performed all the transfection and selection processes starting from this well-characterized clone.

Immunoprecipitation with Protein G-Sepharose 4 Fast Flow
Total protein content was assessed using Bradford assay (Sigma). Cell lysates (400 µg of total protein for each sample) were incubated with 3 µg of specific antibodies for 3 hours at 4°C. Twenty microliters of protein G-Sepharose (Sigma) for each sample were added, and the mixture was incubated for 1 hour at 4°C with gentle rocking, washed, and subjected to SDS-PAGE.
Expression and purification of PTPRG intracellular domain and pull-down assay PTPRG intracellular domain (ICD) and D1028A mutant (amino acid residues 797-1145 of NP_002832 sequence) were cloned in the T7-based HisG-tagged vector expression pRSET A (Invitrogen) using the unique BamHI and EcoRI sites, sequenced, and expressed in BL21 (DE3) pLysS Escherichia coli, and the recombinant proteins were purified. Approximately 2 μg of PTPRG and 10 μg of enhanced green fluorescent protein (EGFP) affinity purified protein were added to 500 μg of K562 lysed in LB for 3 hours at 4°C. The beads were collected, washed three times with LB, and then subjected to SDS-PAGE and Western blotting with specific antibodies as described.
When indicated, K562 cells were exposed to 2 μmol/L 5-aza-2′-deoxycytidine (DAC) for 24 hours, washed, and then plated in a drug-free medium. The same experiment was reproduced in mock-transfected K562, and two independent clones were stably transfected with PTPRG-AS cDNA that was harvested and plated in a 96-well plate (5 × 10 3 cell/ 100 μL/well) or in MethoCult as described previously. At the indicated time points, 10 μL of 5 mg/mL MTT (Sigma-Aldrich) resuspended in sterile PBS were added, and cells were lysed according to the manufacturer's instructions. Absorbance at 570 nm was read in an ELX808iu Ultra Microplate Reader (Bio-Tek Instruments, Inc.).

Xenografting in nude mice
In vivo studies using 4-week-old nu/nu Swiss mice weighing 18 to 22 g (Charles River) for each experimental condition were performed exactly as described (31).
To evaluate the change of PTPRG and BCR/ABL mRNA levels in CML patients during follow-up, we applied the formula (T − U)/(T + U) adapted from Mauri and colleagues (32). For each individual, the difference (T − U) between the mRNA levels of both PTPRG and BCR/ABL on (T) and before treatment (U) with IM was divided by the sum of the same values (T + U). When T equals U, the ratio is zero, which corresponds to no change in the expression level between the two conditions. This ratio equals −1.0 when no expression is detectable in the treated patient (T = 0). When U = 0, the ratio will be +1.0. Intermediate ratio values between −1.0 and +1.0 correspond to different expression levels.

Methylation-specific PCR
Genomic DNA (1 μg) was subjected to bisulfite modification using the CpGenome DNA modification kit (Chemicon) according to the manufacturer's protocol. PCR conditions were 95°C for 8 minutes followed by 45 cycles at 95°C for 30 seconds and an annealing temperature of 60°C (methylation-specific PCR) or 61°C (unmethylation-specific PCR) for 1 minute followed by a final extension at 72°C for 7 minutes. PCR products (168 bp for unmethylated PTPRG and 166 bp for methylated PTPRG) amplifying the region from −576 to −410 relative to the transcription start were resolved in a 2.5% agarose gel.

Flow cytometry of purified cells and whole blood
Cell lines and whole blood samples were stained with the anti-PTPRG antibody chPTPRG IgY or preimmune IgY and with monoclonal anti-CD34 PE (clone AC136, Miltenyi Biotech) as described (10). Flow cytometry was performed on a Becton Dickinson FACScan flow cytometer. Data analysis was performed with FCS Express V3 software (De Novo Software).

Statistical analysis
Data analysis was performed using unpaired and onesample t test (GraphPad Instat software). A P < 0.05 was considered statistically significant.

Results
PTPRG expression specifically correlates with decreased clonogenic capability and growth in CML cell lines K562, KYO-1, LAMA-84, and MEG-01 were plated in Metho-Cult. Each well was scored for the presence and number of colonies after 8 days. Fig. 1A shows the number of colonies grown (top) and their volumes (bottom). Only PTPRG expression correlates with a reduced clonogenic capability and volume of the colonies (Fig. 1B). The choice of the PTP panel was linked either to a receptor-like structure and known expression in hematopoietic lineages (PTPRC-CD45, PTPRJ-CD148, PTPRE-PTPε, and PTPRU-PTPμ) or to their role in CML [PTPN1-PTP1B (4), PTPN6-SHP-1 (5), and PPP2R4-PP2A (6)].
PTPRG affects the clonogenic capability of both PTPRGpositive and -negative cell lines. As shown in Fig. 1C, PTPRG expression reduced the clonogenicity in all cell lines from 31% to 40% relative to controls. Down-modulation of PTPRG expression by transfection of an AS PTPRG construct (27) resulted in an increase of colony number only in PTPRG-positive LAMA-84 and MEG-01 cell lines. Interestingly, these cell lines are of megakaryocytic origin, and it is known that PTPRG is expressed in human megakaryocytes (16). The construct is capable to downregulate PTPRG as shown in Supplementary Table S2.  Table S2). Protein expression was comparable but slightly lower than those recorded in human monocytes and was associated with the increase of the myeloid differentiation marker CD13 (Fig. 2B). Clonogenic capability of the cells was decreased in FL compared with mock-and D1028A-transfected cells. In in vivo proliferation assay performed in xenografted nude mice, K562 FL clone formed very small tumors compared with mock and D1028A clones (Fig. 2C).
Wild-type (WT) PTPRG induces a reduction of total and p210 BCR-ABL -specific tyrosine phosphorylation of its direct substrate CRKL (33)(34)(35) and of STAT5, a transcription factor that represents a downstream target of BCR/ABL (36,37), and is not occurring in mock-transfected and phosphatasedead (D1028A)-transfected clones (Fig. 2D). These results indicate that the tumor suppressor effect of PTPRG in K562 cells is mediated by interference with BCR/ABLdependent signaling.

BCR/ABL is a substrate of PTPRG
As BCR/ABL dephosphorylation occurred in PTPRGtransfected K562 (Fig. 2D), we hypothesized that BCR/ABL might represent a direct substrate for PTPRG. Histidinetagged EGFP protein (mock), WT, and D1028A ICD bound to affinity resin were then incubated with K562 cell lysates. WT and D1028A ICD precipitated a complex containing BCR/ABL, ABL, and its direct substrate CRKL (Fig. 3A). The specificity of the interaction was supported by the lack of coprecipitation of STAT5 with the complex as well as by the lack of any signal in the lane containing larger amounts of EGFP protein (mock; Fig. 3A and C). BCR/ABL precipitated from the WT PTPRG ICD is specifically dephosphorylated as the same protein bound to the catalytic inactive enzyme maintains its tyrosine phosphorylated status, thus ruling out the presence of other tyrosine phosphatase activities within the complex (Fig. 3B). The interaction also occurs when both proteins are coexpressed within HEK293 (Fig. 3D).

PTPRG expression in CML patients
PTPRG downregulation also occurred in CML patients: transcript level was 12.2-fold in Ficoll-purified bone marrow, whereas it was 4.68-fold lower in peripheral blood mononuclear cells (PBMC) than in samples derived from healthy donors (Fig. 4A, left and middle). Thirteen CML patients were quantitatively analyzed for the expression of PTPRG and BCR/ABL at diagnosis and during molecular remission, showing an inverse correlation between the two genes (Fig. 4A, right).
Taken together, these results show that the lack of PTPRG expression was associated with the occurrence of the disease and rule out intrinsic defects of the healthy hemopoietic cells of the subjects to express PTPRG. Figure 3. PTPRG pull-down assay and PTPRG-specific BCR/ABL dephosphorylation. Purified recombinant His-tagged PTPRG ICDs of native or phosphatase-inactive (D1028A) PTPRG were bound to nichel beads and reacted with K562 cell line lysates. A, Western blotting with anti-ABL, CRKL, and STAT5 antibodies. B, the blot was reacted with antiphosphotyrosine antibodies PY20 and 4G10 (pY). Dephosphorylation of BCR/ABL specifically occurred only when BCR/ABL was incubated with ICD. B, the presence of the protein in the D1028A-coated beads was confirmed by the reaction of the stripped blot with anti-ABL antibody. C, Ponceau staining of the same membrane showed the extent of protein purification achieved by affinity purification of the baits and show that the amount of control protein (EGFP) present in the control beads is higher that the specific baits. One representative of two independent experiments. D, the results of single and combined cDNA transfection with the indicated cDNA (top). HEK293 cells were transfected with p210 BCR-ABL (p210), PTPRG (FL), PTPRG D1028A (D1028A), and a combination of p210 + FL and p210 + D1028A cDNA. Equal amounts of HEK293 lysate were probed with anti-ABL and anti-PTPRG-P4 antibodies. Bottom, the result of coprecipitation of HEK293-transfected cells. The lysates were immunoprecipitated with anti-PTPRG-P4 antibody and immunoblotted with anti-ABL antibody and vice versa, showing that PTPRG-p210 interaction occurs between endogenously expressed proteins.

Demethylating agents induce PTPRG expression
PTPRG downregulation by epigenetic modification has been reported in various cancer types (18)(19)(20)(21). The hypomethylating agent (DAC) induced reexpression of PTPRG (Fig. 5A) followed by a marked inhibition of colony formation and cell proliferation/survival in semisolid and liquid media partially reversed by the concomitant inhibition of PTPRG expression through transfection with a PTPRG AS carrying plasmid in both transiently (not shown) as well as stably transfected K562 cell line clones (Fig. 5B and C). The presence of AS construct was effective in inhibiting DAC-induced PTPRG expression as evaluated after 3 and 6 days from the removal of the drug (Fig. 5C). Flow cytometry confirmed the efficacy of the construct: the percentage of PTPRG-positive cells at 3 and 6 days after DAC treatment in control K562 clone was 13.2% and 16.2%, respectively, which dropped to 6.5% and 1.1% at 3 and 6 days, respectively, on DAC treatment of AS-transfected clones.

Reduced promoter methylation and recovery of PTPRG expression in a subset of patients
CpG methylation analysis in the same patients at diagnosis and after successful treatment (all characterized by upregulation of PTPRG expression) indicated that the recovery of PTPRG expression is associated with reduced methylation of a region of its promoter in a substantial fraction of patients (Fig. 5D).

Discussion
In this study, we first observed a specific association between loss of PTPRG expression and increased clonogenic  +1). B, left, expression of PTPRG antigen in samples from healthy donors (control) and CML patients was evaluated by flow cytometry of whole blood (gate on the left) with chPTPRG antibody; an example of the analysis is shown. Center, geometric median fluorescent intensity (MFI) ratio values of the samples analyzed, including controls and CML at diagnosis and at CCR. Right, geometric MFI ratio values of PTPRG antigen in CD34 + cells from peripheral blood of healthy donors (control) and CML patients. Data are average geometric MFI ratios between the chPTPRG-stained sample and the preimmune chicken IgY. In the absence of PTPRG antigen expression, the MFI ratio equals one. capability in CML cell lines that did not occur with seven other phosphatases analyzed. The observation that the two CML cell lines that express PTPRG were still capable of growing in methylcellulose, although at a much reduced extent, prompted us to speculate that the level of PTPRG expression/activity could determine the growth capability and differentiation of these cells. This hypothesis was confirmed by the results of the PTPRG Figure 5. PTPRG promoter hypermethylation in K562 cell lines and primary cells. Involvement of PTPRG in the oncosuppressive effect of DAC. A, expression of PTPRG in K562 after 3 and 6 days of treatment with DAC; cells were exposed to 2 μmol/L DAC for 24 h and then replated in drug-free medium. RNA was extracted on the third and sixth days after treatment to verify PTPRG gene expression by PCR (ATCB, 22 cycles; PTPRG, 40 cycles). One representative experiment of three experiments. B, cells were transfected with mock and PTPRG AS cDNA vector and selected for G418 resistance. The cells were then exposed to 2 μmol/L DAC. At 24 h after drug addition, cells were washed with drug-free medium and then replated at 3 × 10 3 /mL; after 8 d, each well was scored for colony number and expressed as a percentage relative to untreated transfected cells. Top, a detail of the colonies grown (magnification, 10×); bottom, MTT-stained wells (n = 3). The same experiment was reproduced in transiently transfected K562 clones (mock and AS). C, same cells of B cultured in liquid medium. The percentage of live cells after DAC treatment was calculated, setting the value of the mock and AS K562 cell lines treated with vehicle (DMSO) to 100%. Top, MTT staining was performed at 3 and 6 d. Bottom, at the same time points, an aliquot of cells for each experimental condition was analyzed for PTPRG expression by flow cytometry (reported in the text) and mRNA levels by QPCR. n = 2. D, PTPRG methylation-specific PCR in CML patients. All patient (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11) samples at diagnosis were hypermethylated for PTPRG. Examples of PCR amplification data are shown: M, methylated amplicon; UM, unmethylated amplicon. Patients 1 to 5 were hypermethylated at the follow-up after treatment, whereas patients 6 to 11 were unmethylated. No alterations in the methylation pattern of PTPRG were observed in non-CML disease (n = 5, data not shown) under the same experimental conditions. modulation experiments, wherein overexpression determined a reduction of clonogenic capability whereas downregulation was associated with an increase in clonogenic capability only in the two PTPRG-positive cells lines MEG-01 and LAMA-84.
K562 cell line showed activity-dependent increase of the myeloid differentiation marker CD13 and decrease in clonogenic capability both in vitro and in vivo following PTPRG cDNA transfection, associated with a reduced total, BCR/ABL-, CRKL-, and STAT5-specific tyrosine phosphorylation. No detectable PTPRG degradation was found at variance with other studies reporting that the association of BCR/ABL with the serine-threonine phosphatase PP2A leads to SHP-1 activation followed by BCR/ABL dephosphorylation and proteasome-dependent proteolysis (6). PTPRG likely acts through a different pathway activating myeloid differentiation and a different pattern of dephosphorylation and/or protein association in comparison with PP2A/SHP-1 whose mRNA, at variance with PTPRG, is expressed in K562 (Fig. 1B).
CRKL dephosphorylation directly links PTPRG to the inhibition of BCR/ABL (35). Pull-down assay confirms a direct interaction between the ICD domain of PTPRG and BCR/ABL along with its direct substrate CRKL. BCR/ABL acts as a bona fide substrate for PTPRG, as it is dephosphorylated only by precipitation with the catalytic active enzyme. This interaction occurs most likely in the ABL portion of the aberrant kinase, as suggested by the ability of the bait to coprecipitate ABL, and also occurs when both targets are coexpressed in HEK293 cells. The absence of signal using higher amounts of an unrelated protein (EGFP) and the absence of STAT5 within the complex testify to the specificity of the interaction. This last observation is in agreement with the fact that, although known to be dependent on phosphorylation and activation by BCR/ ABL, STAT5 was never reported to coprecipitate with it (36,38,39).
A near-complete lack of expression was associated with the occurrence of CML in primary patient samples, whereas we observed a recovery to values close to those of healthy donors for both mRNA and protein levels following molecular remission. This shows that the lack of expression was not due to other factors independent of the disease status. The inverse correlation with BCR/ ABL mRNA levels before and after treatment suggests that the loss of the oncogenic clone is followed by the recovery of a nonneoplastic hemopoiesis characterized by PTPRG expression. Along these lines, we show that, similar to the well-established molecular analysis for identification of the t(9;22)(q34.1;q11.21) chromosomal translocation (40), the measurement of the PTPRG transcript might offer a new diagnostic tool in association with the former. This potential application needs, of course, a more extensive validation by multicenter studies. Moreover, we provide for the first time the proof of principle that flow cytometric analysis of PTPRG expression might represent a potentially useful and unique additional biomarker for CML, with the advantages typical of this application, such as ease of use, enhanced reproducibility, and cost-effectiveness.
Members of the tyrosine phosphatase family are involved in the pathogenesis of CML (4-6, 41, 42), and a therapeutic strategy based on PTP activation has been proposed (43). Having identified a molecular target, ligandstimulated activation of the residual PTPRG molecules expressed on the surface of Ph + myeloid blasts and the development of specific chemical activators can be readily proposed.
Downregulation of PTPRG expression associated with methylation of specific promoter regions has been recently described in various cancers (18)(19)(20)(21). This result, along with the protective effect of the AS construct on DAC treatment, is in accordance with these findings and adds CML to the list of neoplastic diseases where methylationdependent PTPRG downregulation occurs. This observation might explain the molecular mechanism of epigenetic drugs found to be active in CML (44). Even if we observed this association in 55% of the patients analyzed, still not all of them show the effect. However, an association between PTPRG downregulation and promoter hypermethylation within this region was reported, despite this phenomenon occurring in 27% of the patients affected by cutaneus T-cell lymphoma (20). It is likely that CpG islands within other PTPRG promoter regions might be involved, and this possibility needs to be thoroughly analyzed in future studies using high-throughput approaches.
The loss of PTPRG expression in CD34 + cells indicates that this is an early event in the pathogenesis of CML, although it is still unclear if the maintenance of critical levels of PTPRG expression can act as a "gatekeeper" in the molecular events leading to the clinical manifestation of CML.
Overall, our findings provide the first compelling evidence of the tumor-suppressive effect of PTPRG in CML, indicating that downregulation, but not necessarily a complete loss of PTPRG expression, is associated with the development of CML and that restoration of expression levels similar to, but even lower than those present in normal myeloid cells seems to exert a strong oncosuppressive effect in Ph + cells.
These results point for the first time to PTPRG as a relevant new player in the pathogenesis of CML and suggest it as a potential target for diagnostic and therapeutic applications.

Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
its critical support during the earliest phase of the study and throughout the following developments.

Grant Support
Support for this study was provided by the Consorzio Studi Universitari in Verona and by the Associazione Italiana per la Ricerca sul Cancro (grant IG 4667).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.