Expression of the protein tyrosine phosphatase beta2 gene in mouse erythroleukemia cells induces terminal erythroid differentiation.

We have cloned cDNA for protein tyrosine phosphatase β2, which had been implicated in erythroid differentiation of mouse erythroleukemia cells. Expression of cDNA constructs, in which β2 cDNA is placed under the control of mouse metallothionein-I promoter, by ZnCl2 converted a significant portion (20 to 38%) of the cells to erythroid-like cells, which is 25-50% of the erythroid differentiation efficiency observed by conventional erythroid-inducing agents. Furthermore, introduction and expression of altered protein tyrosine phosphatase β2 cDNA constructs designed to produce the enzyme lacking the phosphatase activity inhibited erythroid differentiation by 100-20%, depending upon the concentration of erythroid-inducing agents employed. These results strongly suggest that protein tyrosine phosphatase β2 is involved in triggering erythroid differentiation in mouse erythroleukemia cells.

We have cloned cDNA for protein tyrosine phosphatase ␤2, which had been implicated in erythroid differentiation of mouse erythroleukemia cells. Expression of cDNA constructs, in which ␤2 cDNA is placed under the control of mouse metallothionein-I promoter, by ZnCl 2 converted a significant portion (20 to 38%) of the cells to erythroid-like cells, which is 25-50% of the erythroid differentiation efficiency observed by conventional erythroid-inducing agents. Furthermore, introduction and expression of altered protein tyrosine phosphatase ␤2 cDNA constructs designed to produce the enzyme lacking the phosphatase activity inhibited erythroid differentiation by 100 -20%, depending upon the concentration of erythroid-inducing agents employed. These results strongly suggest that protein tyrosine phosphatase ␤2 is involved in triggering erythroid differentiation in mouse erythroleukemia cells.
Erythroleukemia cells have been the subject of extensive studies as one of the best models for terminal differentiation (1)(2)(3)(4)(5)(6). Upon exposure to inducing agents, mouse erythroleukemia (MEL) 1 cells are converted to cells that exhibit all the characteristics of erythroid cells, including the loss of growth proficiency. Previously, we presented experimental results suggesting that dephosphorylation of phosphorylated tyrosine moieties of specific cellular proteins is involved in the early stage of MEL cell differentiation, which includes induction of the differentiation by protein tyrosine kinase inhibitors (7-10), inhibition of differentiation by a specific inhibitor (Na 3 VO 4 ) of protein tyrosine phosphatases (PTPases) (11), and lack of tyrosine dephosphorylation in differentiation-defective mutant cells (11). Furthermore, two intracellular protein factors implicated in MEL cell differentiation had characteristics of a tyrosine-phosphorylated protein and a PTPase, respectively (12). In more recent studies, we found that transcripts of two new PTPases, termed RIP and PTP␤2, were induced at a very early stage of differentiation (13). Subsequently, we have cloned full-length cDNA for RIP (14) and PTP␤2. Here we report cloning of a full-length cDNA for PTP␤2 and show that introduction and subsequent expression of the full-length PTP␤2 cDNA in MEL cells, but not RIP cDNA, convert a significant portion (20 -38%) of the cells to erythroid-like cells. Dominantnegative experiments using altered PTP␤2 cDNA constructs further suggested that PTP␤2 is involved in triggering MEL cell erythroid differentiation.

MATERIALS AND METHODS
Cell Culture-MEL cells (745A) were cultured at 37°C in a CO 2 (5%) incubator in minimal essential medium with 12% (v/v) fetal calf serum. For colony formation, exponentially grown cells were incubated for 2 days under various conditions and plated on the medium containing methylcellulose (1.5%). Colonies were counted after 9 days of incubation as described previously (9).
Cloning of PTP␤2 cDNA-A cDNA library was constructed from MEL cells that had been incubated with 280 mM DMSO for 24 h (14), and clones were screened using a 377-base pair PTPase catalytic domain (BET) probe (15) radiolabeled with [ 32 P]dCTP (111 TBq/mmol; ICN) (14). Of 1.5 ϫ 10 6 plaques, 30 plaques hybridized with the probe, and a clone-(37-1) that contained the longest cDNA insert (ϳ8 kbp) was subcloned. The nucleotide sequence was determined by the dideoxy chain-termination method by using Sequenase sequencing kit (U. S. Biochemical Corp.) and Taq polymerase on the Perkin-Elmer 373A automated DNA sequencer. We also isolated a clone with ϳ6 kbp of cDNA-(45-10), and the cDNA was sequenced.
Plasmid Construction-The SacI/ApaI fragments (6.0 and 7.6 kbp), which represent full-length PTP␤2 cDNA, were excised from clone-(45-10) and clone-(37-1), respectively. The DNA was blunt-ended and inserted into the blunt-ended XhoI site of pBMGneo vector (16). The resulting plasmids, termed pMG␤2-6 K (for 6-kbp cDNA) and pMG␤2-8 K (for 7.6-kbp cDNA) contain either the full-length 6-or 7.6-kbp PTP␤2 cDNA between the second intron of rabbit ␤-globin gene and its poly(A) signal, and their expressions are under the control of mouse metallothionein-I (mMT-I) promoter. The plasmids also carry neomycin resistance gene (neo r ). The blunt-ended XhoI fragment for full-length RIP cDNA (8 kbp) (14) was also inserted into the blunt-ended XhoI site of this vector, and the resulting plasmid was termed pMGRIP. For pGK␤2Cys/Ser, a point mutation that should lead to the substitution of cysteine 1140 to serine was introduced into the full-length PTP␤2 cDNA in pBlueScript SK by Chameleon mutagenesis kit (Stratagene). The DNA was also appended at proline 52 with 10 amino acids of the human c-myc epitope (EQKLISEEDL). The SacI/NcoI fragment (4.5 kbp) of this mutant PTP␤2 cDNA (␤2Cys/Ser) was blunt-ended and inserted into the blunt-ended PstI site of pGK vector, including mouse phosphoglycerate kinase promoter and its poly(A) signal (17).
For pGK␤2⌬P, the NotI/blunt-ended BstXI fragment (3.0 kbp) of PTP␤2 cDNA was subcloned into the NotI/blunt-ended XhoI site of pBlueScript SK. This deleted fragment of PTP␤2 cDNA was appended with an artificial stop codon at threonine 1040. The SacI/SalI fragment (3.0 kbp) derived from this plasmid was blunt-ended and inserted into the blunt-ended PstI site of pGK vector.
Transfection-The plasmids (pMG␤2-6 K, pMG␤2-8 K, pMGRIP, pGK␤2Cys/Ser, and pGK␤2⌬P) were introduced into MEL cells using Lipofectin (Life Technologies, Inc.). In the case of pGK␤2Cys/Ser and pGK␤2⌬P, pGKneo vector was cotransfected with them. After 2 days, * This research was supported in part by a Grant-in-aid from the Ministry of Education, Science, and Culture of Japan and the Ryoichi Naito Foundation for Medical Research. 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.
The amino acid reported in this paper has been submitted in the Protein Identification Resource with accession number JT0626.
For Western blotting of hemoglobin, total proteins (10 g) were extracted from the cells (2 ϫ 10 6 ), electrophoresed, and immunoblotted using rabbit anti-mouse antibody (Organon Teknico-Cappel Products) as described (9). The specificity of the antibody was confirmed before use.

RESULTS
Cloning and Characterization of PTP␤2 Gene-Following exposure to erythroid-inducing agents such as DMSO or hexamethylenebisacetamide (HMBA), levels of two novel PTPase (PTP␤2 and RIP) transcripts in MEL cells are sharply increased at a very early stage of differentiation (up to 24 h), and level off thereafter (13,14). We first isolated the full-length cDNA for RIP (14) and subsequently the full-length cDNA for PTP␤2. As shown in Fig. 1, the PTP␤2 gene (7,603 base pairs) codes for a putative 1,238 amino acid protein with estimated molecular mass of 136 kDa (PIR accession number JT0626). There exist a typical catalytic PTPase domain located close to the C terminus (amino acids 961-1,197) and a transmembrane domain (amino acids 878 -897), suggesting that PTP␤2 is membrane-associated PTPase. We identified two forms of PTP␤2 mRNA in MEL, a major species with approximately 7.6 kbp and another with approximately 6 kbp, which is probably a spliced product of the 7.6-kbp mRNA. Sequencing of the cDNA for the 6-and 7.6-kbp RNAs indicated the difference between the two to be at the 3Ј-UTR. The PTPase catalytic domain of mouse PTP␤2 has considerable homology with recently reported human DEP-1 (HPTP) (18,19), human HPTP␤ (20), rabbit GLEPP1 (21), and Drosophila DPTP10D (22). More recently, cDNA (Byp) that is apparently identical to PTP␤2 cDNA was reported (23). Byp/PTP␤2 is ubiquitously expressed among mouse tissues, and the gene is located on chromosome 2 (23, 24).
Induction of PTP␤2 transcripts among transfectants by ZnCl 2 -In order to investigate the relationship between PTP␤2 (and RIP) expression and MEL cell terminal differentiation, we constructed composite DNA in which 6-or 7.6-kbp PTP␤2 cDNA was placed under the control of the mMT-I promoter. MEL cells (745A) were transfected with the two constructs, and among G418-resistant clones (34 clones for the 6-kbp cDNA and 23 clones for the 7.6-kbp cDNA), three clones (2 clones for the 6-kbp cDNA and 1 clone for the 7.6-kbp cDNA) exhibited induction of the 6-or 7.6-kbp transcripts after addition of ZnCl 2 . Fig. 2 shows the patterns of induction of the PTP␤2 transcripts in these clones ( Fig. 2A), along with those induced in the parental MEL cell by DMSO (Fig. 2B). While 6-kbp transcripts were barely detectable and not induced by ZnCl 2 in parental (745A) MEL cells, clones ␤2-37 and ␤2-49, in which the 6-kbp PTP␤2 cDNA construct was introduced, exhibited a significant increase in the 6-kbp transcript level after exposure to ZnCl 2 although we detected small amounts of the 6-kbp transcripts without ZnCl 2 , which were likely to be derived from leaked transcription through the mMT-I promoter ( Fig. 2A,  upper panel). In ␤2-57, a clone in which the 7.6-kbp PTP␤2 C with an intensity screen; bottom panel, overnight exposure at room temperature without the screen) are presented. Glyceraldehyde-3-phosphate-dehydrogenase cDNA was used as a control probe. B, MEL (745A) cells were exposed to DMSO (280 mM) and poly (A) ϩ RNA was prepared at 0, 12, and 24 h and subjected to Northern blot analysis using a 32 P-labeled probe (PTP␤2 catalytic domain sequences) as in A. C, RIP transfectant (RIP13-35) was cultured in the presence (or absence) of ZnCl 2 (160 M) for 6 h. Cells (4 ϫ 10 6 ) were collected, and total RNA (10 g) prepared using TRIzol Reagent (Life Technologies, Inc.) were subjected to Northern blot analysis using 32 P-labeled PTPase catalytic domain of RIP cDNA (MTC) (15) as a probe. The gel was stained with ethidium bromide. cDNA construct was introduced, 7.6-as well as 6-kbp transcripts (probably the spliced product of the 7.6-kbp transcripts) were induced ( Fig. 2A, lower panel). For unknown reasons, those transcripts were expressed at a considerably high level in ␤2-57 even without ZnCl 2 . We also examined a clone (␤2-20) in which PTP␤2cDNA (6 kbp) was present in the cells, but no PTP␤2 transcripts were detected ( Fig. 2A). We also constructed a composite DNA in which a full-length cDNA for RIP (14) was placed under the control of the mMT-I promoter, and over 1 dozen transfectants in which RIP transcripts are induced by ZnCl 2 were subsequently obtained. A typical induction pattern, exhibited by one of the clones (RIP13-35), is shown in Fig. 2C.
Hemoglobin Accumulation and Induction of Other Erythroid Differentiation Markers by the Expression of PTP␤2 cDNA-We next examined whether induction of PTP␤2 (or RIP) transcripts by ZnCl 2 affects erythroid differentiation. As seen in Table I, a substantial portion of ␤2-37 and ␤2-49 cells, which exhibited induction of 6-kbp PTP␤2 transcripts by ZnCl 2 , became positive to benzidine staining, a specific staining for hemoglobin (25). Essentially the same results were obtained with ␤2-57 cells where 7.6-as well as 6-kbp PTP␤2 transcripts were also induced (Table I) (Table I). The induction of hemoglobin in the PTP␤2 transfectants was confirmed by Western blotting for globin (Fig. 3A) although we detected small amounts of hemoglobin in ␤2-37, ␤2-49, and ␤2-57 cells without ZnCl 2 , which was probably derived from the leaked transcription products as shown in Fig. 2A.
Although the efficiency of erythroid differentiation (20 -38%) induced by ZnCl 2 was less than that induced (70 -80%) by conventional inducing agents, these results strongly suggest that the PTP␤2 transcripts induced by ZnCl 2 were responsible for the accumulation of hemoglobin. Since intrinsic 7.6-kbp transcripts remained at a level comparable with those of ZnCl 2induced 6-kbp transcripts regardless of the presence or absence of ZnCl 2 (see Fig. 2A), it seems that PTP␤2 translated from 6-kbp transcripts is responsible for the accumulation of hemoglobin, and there may be a specific mechanism to modulate the splicing of the 7.6-kbp transcripts in MEL cell differentiation. The reason for the low level of hemoglobin accumulation in ␤2-57 cells without ZnCl 2 , despite the significant expression of PTP␤2 transcripts, is not clear. Either more sophisticated mechanisms exist to produce active PTP␤2 protein from the 7.6-kbp transcripts in addition to the splicing described above or ␤2-57 cells had been altered to be clonable. Otherwise, the cells with such a high level of the 7.6-kbp transcripts would have been differentiated and would have become non-clonable even without ZnCl 2 .
We also investigated whether transcripts for ␦-aminolevulinate synthase (ALAS-E), another marker for MEL cell differentiation (26), are induced following ZnCl 2 treatment. Fig. 3B shows that the ALAS-E transcripts were also increased in these cells (␤2-49 and ␤2-57) by ZnCl 2 . A similar pattern of induction was also obtained with the transcripts of another MEL cell differentiation marker, glycophorin (data not shown).
Irreversibility of Erythroid Differentiation by the Expression of PTP␤2 cDNA and Effects of Specific Differentiation Inhibitors-We also examined whether hemoglobin accumulation in the PTP␤2 cDNA transfected cells (␤2-37, ␤2-49, and ␤2-57) induced by ZnCl 2 is irreversible as observed in MEL cell differentiation. For this, the cells were incubated with ZnCl 2 for 2 days, ZnCl 2 was removed, cell incubation was continued for another 3 days, and accumulation of hemoglobin was assayed. As seen in Table II, incubation with ZnCl 2 for 2 days was sufficient to induce hemoglobin accumulation, suggesting that were collected, and total RNA (20 g) were subjected to Northern blot analysis using 32 P-labeled ALAS-E (26) and glyceraldehyde-3-phosphatedehydrogenase (GAPDH) cDNA.

TABLE I Differentiation of PTP␤2 and RIP transfectants by ZnCl 2
The cells were incubated in the presence (or absence) of ZnCl 2 (160 M) for 5 days and stained with benzidine to score the benzidine-positive cells as described previously (25). at 2 days (when PTP␤2 transcripts were fully induced), the cells had already entered into an irreversible stage in expressing a series of erythroid specific genes. This is essentially the same as so-called "commitment" in which exposure to erythroid inducing agents for one or two days is sufficient for MEL cells to enter the irreversible cascade of terminal differentiation (27,28). Thus, it seems that the hemoglobin accumulation and other phenotypic changes observed following ZnCl 2 treatment are not results of transient expression of genes, but closely associated with cellular commitment to terminal differentiation.
We also examined whether specific inhibitors for MEL cell terminal differentiation such as phorbol 12-myristate 13-acetate, dexamethasone, or sodium vanadate (Na 3 VO 4 ) inhibit hemoglobin accumulation induced by ZnCl 2 . Table II shows that these inhibitors blocked the induction of hemoglobin accumulation by ZnCl 2 as they do in erythroid differentiation induced by erythroid-inducing agents.
Thus, by all the criteria so far examined, the changes observed upon induction of PTP␤2 gene expression by ZnCl 2 were, at least qualitatively, indistinguishable from those induced by erythroid-inducing agents.
Effect of Altered PTP␤2 cDNA Expression on Erythroid Differentiation-To investigate whether PTP␤2 is actually involved in erythroid differentiation induced by erythroid-inducing agents, we constructed altered PTP␤2 cDNA and examined whether expression of the altered PTP␤2 cDNA affects erythroid differentiation. In one of the constructs (pGK␤2Cys/Ser), a cysteine residue in the PTP␤2 catalytic domain (position 1140) was replaced by serine, so that the PTPase should lose its catalytic activity while maintaining the total integrity of the protein (23). Another construct (pGK␤2⌬P) in which 595 base pairs of the 3Ј-PTP␤2 ORF cDNA was deleted should code for PTP␤2 lacking two-thirds of the C-terminal intracellular PTPase catalytic domain. These altered PTP␤2 cDNA constructs were placed under the control of a strong constitutive promoter, phosphoglycerate kinase promoter (17).
Among 33 transfectants, which were obtained after transfection with pGK␤2Cys/Ser, three clones (clones ␤2Cys/Ser-4, -8, and -20) exhibited significant levels of PTP␤2 transcripts that were apparently derived from the introduced altered cDNA constructs (Fig. 5A). Employing these cells, we examined whether erythroid differentiation induced by DMSO or HMBA is affected as a result of the constitutive expression of the altered PTP␤2 cDNA. Fig. 5B shows quite clearly that all the transfectants expressing the altered PTP␤2 cDNA (pGK␤2Cys/ Ser) transcripts exhibited considerably lower levels of erythroid differentiation by DMSO (left panel) as well as HMBA (right panel) than those of the parental and control cells, which included transfectants without the altered cDNA expression (clones ␤2Cys/Ser-1 and -9). As often observed in such dominant-negative experiments, the inhibitory effect was most prominently seen at suboptimal concentrations of the inducers (the optimum concentrations for erythroid differentiation are 280 mM for DMSO and 5 mM for HMBA). The effect was also confirmed by Western blotting for hemoglobin accumulated in the cells (Fig. 5C). We obtained essentially the same results with all the transfectants (clones ␤2⌬P-2, -12, and -40) expressing PTP␤2⌬P transcripts that should be translated into protein molecules lacking a considerable portion of the cytoplasmic PTPase domain of the enzyme (data not shown). These results indicate that expression of altered PTP␤2 with no catalytic PTPase activity, either by amino acid substitution or deletion of the catalytic domain, inhibits erythroid differentiation induced by DMSO or HMBA, and the effect may be best explained as a result of competitive inhibition by altered PTP␤2 of normal PTP␤2 activity for a tyrosine phosphorylated substrates whose dephosphorylation is essential for erythroid differentiation. DISCUSSION We have demonstrated that expression of a specific PTPase, PTP␤2, induced biochemical and physiological changes that are quite similar to those associated with in vitro and in vivo erythroid differentiation. Those include: 1) induction of hemoglobin and ALAS-E and glycophorin transcripts, 2) loss of growth proficiency, 3) cellular commitment to erythroid cells, and 4) sensitivity to specific inhibitors for MEL cell erythroid differentiation. The dominant-negative experiments using transfectants expressing altered PTP␤2 transcripts further suggest that PTP␤2 is involved in in vitro erythroid differentiation induced by conventional erythroid-inducing agents.
The results presented above, however, are still premature to conclude that PTP␤2 is actually responsible for triggering erythroid differentiation. For example, although all the characteristics induced by the expression of PTP␤2 so far examined were essentially the same as that observed in erythroid differentiation, the maximum efficiency of differentiation (20 -38%) by the expression of PTP␤2 was approximately half of those FIG. 4. Colony-forming ability of PTP␤2 transfectants after ZnCl 2 treatment. The cells were cultured with DMSO (280 mM) or ZnCl 2 (160 M) for 2 days and plated on a semisolid medium (9). Colonies were counted after 9 days of incubation as described previously (9). Bars depict colony-forming ability (mean Ϯ S.E., n ϭ 8).

TABLE II Cellular commitment and effect of inhibitors
The cells were incubated with DMSO or ZnCl 2 in the presence (or absence) of the inhibitors for 2 days and further incubated without them for 3 days. The cells were then stained with benzidine to score the benzidine-positive cells (25). (70ϳ80%) obtained by conventional erythroid-inducing agents. Several possibilities can be considered to explain the difference. Expression, stability, or intracellular location of the PTP␤2 induced through the artificial cDNA constructs may be different from that induced by erythroid-inducing agents. Alterna-tively, another cascade of molecular events may exist besides the PTP␤2 dephosphorylation cascade, and activation of that or both cascades leads to the full induction. There have been reports concerning expression of ␤-globin and other erythroid-specific genes by transcription factors (3,29). Expression of transcription factors implicated for erythroid-specific gene expression, such as SCL and MafK, substantially increased the level of expression of ␤-globin (3,29) and ALAS-E (3) transcripts in MEL cells as well as the number of benzidine-positive cells, which reached 10 -20% (29) or 24 -54% (3) of the total population. It is not clear, however, from these reports whether the expression of these transcription factors resulted in erythroid differentiation that accompanied growth arrest and cellular commitment, features specific to erythroid differentiation. As we reported previously, induction of PTP␤2 transcripts starts at a very early stage of differentiation, as early as 6 h after addition of inducing agents (13), whereas induction of these transcription factors starts at a later stage (24 h) of differentiation for SCL (29) or no induction at all occurs during differentiation for MafK (T. Yamamoto, personal communication). Along with the pleiotropic induction of erythroid-specific genes and phenotypes by the expression of PTP␤2, this involvement of PTP␤2 in erythroid differentiation, if any, should be in a step quite upstream of the differentiation cascade, well before it differentiates into more specific cascades including ␤-globin gene expression that occurs at later stages. If this is the case, one could speculate that PTP␤2, with a transmembrane domain in the molecule (Fig. 1), is a membrane-associated mediator of a signal from the cell surface, which is generated by erythroid-inducing agents, to nuclei where a series of transcription factors specific to erythroid genes are subsequently activated.
The molecular mechanism of how expression of a single specific PTPase (PTP␤2) converts MEL cells to the ones whose characteristics are very similar to erythroid cells is, of course, a matter of speculation at this stage of the research. Induction of PTP␤2 may dephosphorylate a specific phosphotyrosine-containing protein whose dephosphorylation triggers a molecular cascade for terminal differentiation. Alternatively, induced PTP␤2 may dephosphorylate a broad range of cellular proteins rather nonspecifically and that among the dephosphorylated proteins, there may be one (or more) specific protein that is responsible for erythroid differentiation. Western blot analysis of phosphotyrosine-containing cellular proteins indicated no significant changes in the patterns of major tyrosine-phosphorylated proteins following ZnCl 2 treatment (data not shown), suggesting that induction of PTP␤2 dephosphorylates only a limited number of specific proteins. Experiments to identify the target proteins using more sensitive detection procedures are currently underway.