Protein tyrosine phosphatases PTP-1B and TC-PTP play nonredundant roles in macrophage development and IFN-γ signaling

Abstract

The control of tyrosine phosphorylation depends on the fine balance between kinase and phosphatase activities. Protein tyrosine phosphatase 1B (PTP-1B) and T cell protein tyrosine phosphatase (TC-PTP) are 2 closely related phosphatases known to control cytokine signaling. We studied the functional redundancy of PTP-1B and TC-PTP by deleting 1 or both copies of these genes by interbreeding TC-PTP and PTP-1B parental lines. Our results indicate that the double mutant (tcptp^−/−ptp1b^−/−) is lethal at day E9.5–10.5 of embryonic development with constitutive phosphorylation of Stat1. Mice heterozygous for TC-PTP on a PTP-1B–deficient background (tcptp^+/−ptp1b^−/−) developed signs of inflammation. Macrophages from these animals were highly sensitive to IFN-γ, as demonstrated by increased Stat1 phosphorylation and nitric oxide production. In addition, splenic T cells demonstrated increased IFN-γ secretion capacity. Mice with deletions of single copies of TC-PTP and PTP-1B (tcptp^+/−ptp1b^+/−) exhibited normal development, confirming that these genes are not interchangeable. Together, these data indicate a nonredundant role for PTP-1B and TC-PTP in the regulation of IFN signaling.

T cell protein tyrosine phosphatase (TC-PTP; also known as PTPN2) and protein tyrosine phosphatase 1B (PTP-1B; also known as PTPN1) are intracellular phosphatases with a high degree of sequence and structural homology within the catalytic domain (1, 2). PTP-1B is known to dephosphorylate the insulin receptor (3, 4), and ptp1b^−/− mice are protected from diet-induced insulin-resistance and obesity (3, 5, 6). This phenotype has incited much interest in PTP-1B inhibitors in treating type II diabetes (7, 8). However, because of the structural homology with TC-PTP, none of the small-molecule inhibitors developed to date exhibits a high degree of selectivity against PTP-1B, and only a few are currently in clinical trials.

Both TC-PTP and PTP-1B have been shown to modulate cytokine receptor signaling (reviewed in ref. 9), including IFN-γ signaling (10, 11). IFN-γ is a major inflammatory cytokine, and ligation of the IFN-γ receptor results in the activation of cytoplasmic Jak1 and Jak2 kinases as well as the transcription factor Stat1 (12). Jak1 and Stat1 are known substrates of TC-PTP (11, 13), whereas Jak2 is known to be dephosphorylated by PTP-1B (10). Consequently, we hypothesized that the deletion of both TC-PTP and PTP-1B would exacerbate the inflammatory phenotype seen in tcptp^−/− mice (14, 15).

The data presented here indicate that at least 1 copy of tcptp or ptp1b is required for normal embryonic development. Our results show that the sequential ablation of TC-PTP and PTP-1B reveals a thus-far undetected gene dosage effect. This suggests that small-molecule inhibitors against PTP-1B that also inhibit TC-PTP may have multilevel effects that are proportional to the net reduction in both TC-PTP and PTP-1B activity.

Results

The tcptp^−/−ptp1b>^−/− Double Mutant Is Lethal During Embryonic Development.

To study the redundancy of TC-PTP and PTP-1B function in vivo, we obtained tcptp^−/−ptp1b^−/− double-deficient mice. Using tcptp^+/−ptp1b^+/− mice as breeders, we obtained no double mutants at 3–5 days of age (Table 1), suggesting that the loss of both TC-PTP and PTP-1B resulted in embryonic death.

Table 1.

Genotype distribution in the progeny from heterozygous matings

Genotype(tc-ptp ptp-1b)	Obtained ratio	Expected ratio
+/+ +/+	0.053	0.063
+/+ +/−	0.149	0.125
+/+ −/−	0.062	0.063
+/- +/+	0.166	0.125
+/- +/-	0.280	0.250
+/- −/−	0.135	0.125
−/− +/+	0.068	0.063
−/− +/−	0.087	0.125
−/− −/−	0	0.063

Pups were identified and a piece of tail was obtained for genotyping at 3–5 days after birth. The observed frequencies are on the left and the expected Mendelian frequencies on the right. n = 483. χ² < 1 × 10⁻⁶.

Examination of embryos revealed no obvious embryonic, placental, or yolk sac abnormalities and a normal Mendelian ratio in E8.5–9.0 live embryos generated from tcptp^+/−ptp1b^+/− intercrosses (Fig. 1A). At E9.5–10.5, the tcptp^−/−ptp1b^−/− embryos showed signs of growth retardation, including a smaller head and underdeveloped nasal processes and branchial arches. The heart was still beating (Fig. 1B). At E11.5, most of the tcptp^−/−ptp1b^−/− embryos were dead, with collapsed vasculature and tissue necrosis. No tcptp^−/−ptp1b^−/− individuals were detected past E12 (Fig. 1A). We assessed the relative proportion of progenitor cells (CD117⁺Ter119⁻), CD117⁻Ter119⁺ erythroid cells, CD117^loTer119^lo erythroid precursor cells, CD309⁺ (Flk-1 or VEGFR2) endothelial cells, and CD11b⁺ macrophages by flow cytometry in E10–10.5 yolk sacs. As shown in supporting information (SI) Fig. S1, the absence of tcptp and ptp1b in yolk sacs did not affect a specific cell lineage, but rather impeded normal embryonic development.

Fig. 1.

Combined deficiency in TC-PTP and PTP-1B is lethal in early embryonic development. (A) Frequencies and expected frequencies of live tcptp^−/− ptp1b^−/− embryos obtained from tcptp^+/− ptp1b^+/−× tcptp^+/− ptp1b^+/− breeders at E9.5–13.5 postcoitum. (B) An E10.5 tcptp^−/− ptp1b^−/− embryo together with a tcptp^+/− ptp1b^+/− littermate.

To further characterize the early lethality of tcptp^−/−ptp1b^−/− embryos, we analyzed whole-protein lysates prepared from E9.5–10.5 embryos for differences in protein expression. Stat1 is a substrate of TC-PTP (16) that can lead to apoptosis or attenuate cytokines essential for embryonic development (17, 18). Strong Stat1 phosphorylation levels were detected in tcptp^−/− ptp1b^−/− embryos at E9.5–10.5; however, no Stat1 phosphorylation was evident in any other genotype (Fig. 2; tcptp^+/− ptp1b^−/− not shown). In addition, a small increase in Stat1 protein level also was seen in the tcptp^−/− ptp1b^−/− embryos compared with their littermates (Fig. 2). These results suggest a complementary role of TC-PTP and PTP-1B in Stat1 regulation. In contrast, Stat3 and Stat5 phosphorylation varied between individuals, possibly due to differential regulation during development, and no conclusive hyperphosphorylation was detected in the tcptp^−/− ptp1b^−/− embryos compared with their littermates (Fig. S2).

Fig. 2.

Constitutive Stat1 phosphorylation in tcptp^−/− ptp1b^−/− embryos. Two representative Western blots, showing hyperphosphorylation of Stat1 in E10.5 tcptp^−/− ptp1b^−/− embryos and no phosphorylation in either tcptp^−/− ptp1b^+/+ or tcptp^+/+ ptp1b^−/− littermates. Similar results were obtained from 4 different tcptp^−/− ptp1b^−/− embryos.

Haploinsufficiency of tc-ptp Becomes Apparent on ptp1b^−/− Background.

Both tcptp^+/− and ptp1b^−/− mice have essentially normal life spans and are fertile; however, when put on a ptp1b^−/− background, more than half of tcptp^+/− ptp1b^−/− mice die by age 3 months (Fig. 3A), and the remaining mice do not survive past 9 months (data not shown). Conversely, tcptp^+/− ptp1b^+/− mice are fertile, present no evident abnormalities, and survive past 12 months.

Fig. 3.

Haploinsufficiency of tc-ptp is apparent on the ptp1b^−/− background. (A) Survival curve of tcptp^+/− ptp1b^−/− mice compared with tcptp^−/− ptp1b^+/+ and tcptp^+/+ ptp1b^+/+, tcptp^+/− ptp1b^+/−, tcptp^+/+ ptp1b^−/− mice. (B) Percentage and absolute number of Ter119⁺ erythroid cells in the bone marrow of young (5- to 6-week-old) mice. The y-axes of the flow cytometry plots have been adjusted to show equal numbers of events. n = 6–9 per genotype. *, P < .05; **, P < .005.

Young tcptp^+/− ptp1b^−/− mice (age 5–6 weeks) were smaller than their littermates and, on examination of the major hematopoietic cell types in the bone marrow, displayed decreased numbers of Ter119^hi mature and immature erythroid cells, suggesting that they may have been anemic (Fig. 3B), similar to tcptp^−/− mice (15). In comparison, the numbers of CD11b⁺ myeloid cells and B220⁺ B lymphocytes were normal (Fig. S3), and there was no disproportionate increase in peripheral B220⁺IgM^−/loCD24^−/lo cells (Fig. S4) that would characterize chronically activated follicular type B cells. Older tcptp^+/− ptp1b^−/− mice had lower body weight and exhibited significant splenomegaly, lymphadenopathy, and accelerated thymic atrophy, as well as massive mononuclear infiltrates to the lacrimal and salivary glands and the gastric mucosa (Table S1 and data not shown), whereas age-matched heterozygous tcptp^+/− ptp1b^+/− and tcptp^+/+ ptp1b^−/− controls were mostly normal. It is also noteworthy that tcptp^−/− ptp1b^+/− mice had a significantly shorter lifespan than tcptp^−/− ptp1b^+/+ mice (mean survival, 20.4 ± 1.7 days vs. 25.3 ± 3.5 days; P < .0001) and that the incidence of live tcptp^−/− ptp1b^+/− births was somewhat lower than expected (Table 1).

TC-PTP and PTP-1B Collaborate in Macrophage Development.

Given the appearance of the tcptp^+/− ptp1b^−/− mice, which resembled that of tcptp^−/− mice to some extent (14–16), as well as the current evidence of a role for both TC-PTP and PTP-1B in the control of CSF-1 (19, 20) and IFN-γ signaling (10, 11), we hypothesized that tcptp^+/− ptp1b^−/− mice might exhibit defective macrophage development and activation.

Tcptp^+/+ ptp1b^−/− and tcptp^+/− ptp1b^−/− mice both had a larger percentage of splenic CD11b⁺Ly6G^−/lo macrophages compared with tcptp^+/+ ptp1b^+/+ controls (Fig. 4A, Left). Taking into account the significant splenomegaly in the tcptp^+/− ptp1b^−/− mice (256 × 10⁶ ± 28 × 10⁶ cells/spleen vs. 171 × 10⁶ ± 14.2 × 10⁶ for tcptp^+/+ ptp1b^−/− and 103 × 10⁶ ± 10.6 × 10⁶ for tcptp^+/+ ptp1b^+/+; P < .02 for both), the total number of macrophages in tcptp^+/− ptp1b^−/− spleen was approximately double that in tcptp^+/+ ptp1b^−/− spleen and more than triple that in normal mice spleen (Fig. 4A, Right).

Fig. 4.

Collaborative regulation of monocyte/macrophage development and the macrophage IFN-γ response. (A) Representative FACS data from the spleens of young (5-to 6-week-old) mice, showing the percentages of CD11b⁺Ly6G^−/lo monocytes and CD11b⁺Ly6G^hi granulocytes. n = 6–10 per genotype. *, P < .05; **, P < .01 compared with all other genotypes unless specified otherwise. (B) Myeloid colony formation in methylcellulose from total bone marrow from young (5- to 6-week-old) mice in response to GM-CSF and CSF-1. n = 4–6 per genotype. *, P < .05; **, P < .01. (C) Representative Western blot showing phosphorylation of Stat1 in spleen-derived macrophages after 15 min of stimulation with 100 U/mL of IFN-γ. Similar results were obtained from 3 separate experiments.

To explain this increase in splenic macrophages, we found no significant increase in the total number of myeloid precursors from tcptp^+/− ptp1b^−/− bone marrow in a mixed-colony assay (Fig. S3). In accordance with previous reports (20), tcptp^+/+ ptp1b^−/− bone marrow produced more colonies in the presence of CSF-1 than control bone marrow; however, tcptp^+/− ptp1b^−/− mice had even more CSF-1–responsive cells than tcptp^+/+ ptp1b^−/− mice (Fig. 4B). This finding was specific to CSF-1; stimulation with GM-CSF did not produce any differences in the number of colonies independent of genotype (Fig. 4B). Our results indicate that the ability to produce CSF-1−responsive colonies does not depend solely on the frequency of monocyte-committed progenitors and demonstrate that TC-PTP and PTP-1B are not redundant in CSF-1 signaling (19, 20), but play a cooperative role in macrophage development.

Given that both TC-PTP and PTP1B are known to regulate the IFN-γ signaling pathway (10, 11, 13), we examined the IFN-γ responsiveness of tcptp^+/− ptp1b^−/− spleen-derived macrophages. Macrophages derived from tcptp^+/− ptp1b^−/− spleens produced on average 3 times more NO than those derived from tcptp^+/+ ptp1b^+/+ control spleens and approximately twice as much as those derived from either tcptp^+/− ptp1b^+/− or tcptp^+/+ ptp1b^−/− spleens (Fig. S5). Furthermore, compared with tcptp^+/− ptp1b^+/− and tcptp^+/+ ptp1b^−/− mice, Stat1 phosphorylation was significantly enhanced in macrophages derived from tcptp^+/− ptp1b^−/− mice (Fig. 4C), indicating that TC-PTP and PTP-1B collaborate in the regulation of the IFN-γ response directly downstream of the receptor.

Thymic Defect in the tcptp^+/−ptp1b^−/− Mice.

Previous data have shown that tcptp^−/− mice have normal numbers of peripheral T cells, which do not proliferate in response to mitogenic stimuli (15) but produce large amounts of IFN-γ (14). There is no reported T-cell phenotype in ptp1b^−/− mice (21). Given the accelerated thymic atrophy in tcptp^+/− ptp1b^−/− mice, we studied their T-cell development and function in greater detail.

We found an age-dependent decrease in CD4⁺CD8⁺ double-positive thymocytes in the tcptp^+/− ptp1b^−/− mice compared with viable littermate controls (Fig. 5A). These results suggest an early delay in thymocyte development, and an accumulation of cells was detected at the CD44⁺CD25⁻ population (DN1 stage) in the tcptp^+/− ptp1b^−/− mice compared with viable littermate controls (Fig. 5B). Investigation of the early thymic progenitor (ETP) population and the CD117^−/lo DN1 fractions (DN1c, d, e) (22) revealed a relative increase in the proportion of the canonical CD117^hi ETP (Lin⁻CD117^hiCD25⁻; Fig. 5C), as well as a more pronounced increase in noncanonical CD117^−/lo CD127⁺ fractions with low proliferation capacity (DN1c, d/e; Fig. 5C). In addition, we detected a significant proportion of B220⁺ cells in the tcptp^+/− ptp1b^−/− thymuses, a population that was absent from both normal (data not shown) and tcptp^+/+ ptp1b^−/− thymuses (Fig. S6). Our results reflect the possibility that the tcptp^+/− ptp1b^−/− ETP and DN3 do not receive sufficient signals for expansion and differentiation, leading to the thymic defect and the development of B220⁺ cells from noncanonical precursors.

Fig. 5.

Age-dependent defect in thymic output and enhanced IFN-γ secretion by peripheral T cells from tcptp^+/− ptp1b^−/− mice. (A) Representative flow cytometry plots showing a relative increase in mature single-positive thymocytes and a decrease in double-positive thymocytes in tcptp^+/− ptp1b^−/− mice. n = 5–12 per genotype. (B) Representative flow cytometry plots showing a relative increase in the DN1 subpopulation in tcptp^+/− ptp1b^−/− thymuses. n = 5–12 per genotype. (C) Representative FACS data showing CD117 and CD127 staining on lineage-negative cells in thymuses from 13- to 15-week-old mice. n = 6–10 per genotype. (D) IFN-γ production by splenic CD90⁺ cells in the supernatant after 48 h of culture on CD3/CD28. Histograms represent mean ± SEM from 3 separate experiments. *, P < .05, tcptp^+/− ptp1b^−/− compared with all other genotypes.

The tcptp^−/− ptp1b^+/+ mice exhibited elevated levels of circulating IFN-γ shortly before death (14). In addition, their splenic T cells produced more IFN-γ on a per-cell basis after mitogenic stimulation. We did not detect any serum IFN-γ in unstimulated tcptp^+/− ptp1b^−/− mice, but we did note the capacity of tcptp^+/− ptp1b^−/− T cells to produce IFN-γ in culture. IFN-γ production by control splenocytes (tcptp^+/+ ptp1b^+/+, tcptp^+/− ptp1b^+/−, tcptp^+/+ ptp1b^−/−) was essentially equal, whereas tcptp^+/− ptp1b^−/− T cells produced almost twice as much IFN-γ when normalized to DNA synthesis (Fig. 5D). Our results indicate that T cells from tcptp^+/− ptp1b^−/− mice have an increased capacity to produce IFN-γ and thus likely contribute to the decreased survival of these animals.

Discussion

The results presented here demonstrate that TC-PTP and PTP-1B play cooperative and nonredundant roles during embryonic development as well as in postnatal macrophage and T lymphocyte development and activation. The removal of 1 copy of tcptp on a ptp1b^−/− background revealed a gene dosage effect that had previously gone undetected in tcptp^+/− mice. In these tcptp^+/− ptp1b^−/− mice, we were able to observe the collaborative effect of the 2 enzymes in macrophage development both in vivo and in culture with CSF-1. This occurred at different levels; TC-PTP appears to control only dephosphorylation of the receptor (19), whereas PTP-1B may play a role in receptor recycling as well (20). Such a role for PTP-1B in receptor maturation has been proposed in the context of Flt3, another member of the receptor tyrosine kinase class III family (23). Our data also demonstrate that IFN-γ signaling is potentiated in tcptp^+/− ptp1b^−/− macrophages, and that peripheral T cells from these mice can secrete more IFN-γ compared with cells from either tcptp^+/+ ptp1b^−/− or tcptp^+/− ptp1b^+/− mice. The normal phenotype seen in tcptp^+/− ptp1b^+/− mice further suggests that TC-PTP and PTP-1B are not interchangeable enzymes and can act either on different pathways or different areas of the same pathway.

IFN-γ signaling requires Jak1 and Jak2 and Stat1 (12, 24, 25). Stat1 and Jak1 are substrates for TC-PTP (11, 13), whereas Jak2 is dephosphorylated by PTP-1B (10). Our results demonstrate that TC-PTP and PTP-1B also are not redundant or interchangeable in the control of IFN-γ signaling and Stat1 phosphorylation. Macrophages from tcptp^+/− ptp1b^−/− spleens were more sensitive to IFN-γ than those from either tcptp^+/− ptp1b^+/− or tcptp^+/+ ptp1b^−/− control spleens; furthermore, unlike their littermates, tcptp^−/− ptp1b^−/− embryos exhibited constitutive Stat1 phosphorylation. The ablation of other negative regulators of IFN-γ signaling, SOCS1 and PIAS1, also results in decreased survival (26–28). In parallel, increasing IFN-γ leads to autoimmune inflammation (29–32). Although the precise cause of embryonic lethality in tcptp^−/− ptp1b^−/− mice remains unknown, Stat1 activation can lead to apoptosis or influence the secretion of cytokines important for the migration of cells and maintenance of the embryo (17, 18). Interestingly, Stat5 and Stat3 were not significantly hyperphosphorylated in tcptp^−/− ptp1b^−/− embryos, further supporting the likely biological significance of Stat1 activation. Furthermore, TC-PTP and PTP-1B have multiple substrates with the ability to affect the entire organism, including CSF1R, which likely contribute to the phenotype observed in tcptp^+/− ptp1b^−/− adult mice.

We also detected decreased numbers of mature and immature Ter119⁺ erythrocytes in young tcptp^+/− ptp1b^−/− animals, which were present in close to normal numbers in older tcptp^+/− ptp1b^−/− mice that survived past 10 weeks. Because excessive IFN and Stat1 signaling can result in anemia (33, 34), we propose that this discrepancy may have been due to the mixed genetic background, and that the mice that survived longer could have produced less IFN-γ or had a slightly different response to IFN-γ. (Signaling studies were conducted on cells from young animals.)

Interferons are currently used for treatment of various cancers, as well as hepatitis and multiple sclerosis (35). In the case of chronic myelogenous leukemia, a negative correlation between responsiveness to IFN-α and TC-PTP activity has been shown (36, 37). Furthermore, Stat1 is a known tumor suppressor (38), and increasing Stat1 activity could reduce tumor load and metastasis. Because the development of TC-PTP or PTP-1B–specific inhibitors is difficult because of their significant homology, blocking agents with overlapping capabilities to inhibit both enzymes may provide a useful advantage. Moreover, because tcptp^+/− ptp1b^+/− mice display no overt abnormalities, and because ptp1b^+/− mice also demonstrate improved glycemia in type II diabetes (3, 5, 6), reduced obesity (3, 5, 6), and delayed breast tumor development in ErbB2 transgenic models (39), we believe that the use of nonspecific PTP inhibitors against TC-PTP/PTP1B will still have a beneficial effect in these 3 diseases without major side effects, as long as sufficient amounts of both of these enzymes remain active.

Materials and Methods

Mice.

ptp1b^−/− mice were bred with tcptp^+/− mice to obtain tcptp^+/−ptp1b^+/− mice, which were then interbred to obtain all of the different genotypes studied. Genotyping was done as described previously (3, 15). All mice were kept in specific pathogen-free housing in the animal care facility. Protocols were approved by the McGill Animal Care Ethics Committee.

Statistical Analysis.

All histograms represent mean ± SEM. P values were determined using a 2-tailed unpaired Student t test.

More details are provided in SI Materials and Methods.