A novel phosphatidylinositol-3,4,5-trisphosphate 5-phosphatase associates with the interleukin-3 receptor.

To gain insight into the intracellular signaling cascades that are activated by the binding of interleukin-3 (IL-3) to its target cells, we have embarked on the identification of proteins that are associated with the IL-3 receptor (IL-3R). In a previous study we reported that a 110-kDa serine/threonine protein kinase is constitutively associated with the IL-3R and activated following IL-3 stimulation. We now report that a phosphatidylinositol-3,4,5-trisphosphate (PtdIns-3,4,5-P3) 5-phosphatase (5-ptase) is also constitutively associated with the IL-3R. This 5-ptase is magnesium-dependent and removes the 5-position phosphate from PtdIns-3,4,5-P3 but does not metabolize PtdIns-4,5-P2, inositol (Ins)-1,3,4,5-P4, or Ins-1,4,5-P3. This substrate specificity distinguishes it from any previously characterized 5-ptase. Interestingly, it may be bound indirectly via phosphatidylinositol 3-kinase (PI 3-kinase), another enzyme that is constitutively bound to the IL-3R. However, unlike PI 3-kinase which becomes activated following IL-3 stimulation, this receptor-associated 5-ptase activity does not increase following IL-3 stimulation, and its primary function may be to keep the principal in vivo product of PI 3-kinase, PtdIns-3,4,5-P3, at low levels in unstimulated cells, to terminate the PI 3-kinase signal following IL-3 stimulation or to metabolize PtdIns-3,4,5-P3 to a metabolically active second messenger, i.e. PtdIns-3,4-P2.

To gain insight into the intracellular signaling cascades that are activated by the binding of interleukin-3 (IL-3) to its target cells, we have embarked on the identification of proteins that are associated with the IL-3 receptor (IL-3R). In a previous study we reported that a 110-kDa serine/threonine protein kinase is constitutively associated with the IL-3R and activated following IL-3 stimulation. We now report that a phosphatidylinositol-3,4,5-trisphosphate (PtdIns-3,4,5-P 3 ) 5-phosphatase (5-ptase) is also constitutively associated with the IL-3R. This 5-ptase is magnesium-dependent and removes the 5-position phosphate from PtdIns-3,4,5-P 3 but does not metabolize PtdIns-4,5-P 2 , inositol (Ins)-1,3,4,5-P 4 , or Ins-1,4,5-P 3 . This substrate specificity distinguishes it from any previously characterized 5-ptase. Interestingly, it may be bound indirectly via phosphatidylinositol 3-kinase (PI 3-kinase), another enzyme that is constitutively bound to the IL-3R. However, unlike PI 3-kinase which becomes activated following IL-3 stimulation, this receptor-associated 5-ptase activity does not increase following IL-3 stimulation, and its primary function may be to keep the principal in vivo product of PI 3-kinase, PtdIns-3,4,5-P 3 , at low levels in unstimulated cells, to terminate the PI 3-kinase signal following IL-3 stimulation or to metabolize PtdIns-3,4,5-P 3 to a metabolically active second messenger, i.e.

PtdIns-3,4-P 2 .
Interleukin-3 (IL-3) 1 is a potent hemopoietic cytokine that is produced primarily by activated T lymphocytes and stimulates the proliferation and differentiation of pluripotent stem cells and committed myeloid and lymphoid progenitors. It exerts its action by binding to specific cell surface receptors on its target cells (1,2). These cell surface receptors are members of the hemopoietic receptor superfamily (2) and consist of two subunits, designated ␣ and ␤ (2). The 70-kDa ␣ subunit is specific for IL-3 (3), while the 140-kDa ␤ subunit (␤ c ) is shared by IL-3, GM-CSF, and IL-5 (4). In the mouse, there is a second ␤ subunit called ␤ IL-3 which shares 91% amino acid identity with ␤ c and is specific for IL-3 (5). Although neither the ␣ nor the ␤ subunits possess intrinsic tyrosine kinase activity, the IL-3R ␤ subunits, as well as several other intracellular proteins, rapidly become phosphorylated on tyrosine residues following IL-3 stimulation (6 -14). This phosphorylation is mediated, at least in part, by a constitutively associated tyrosine kinase, Jak2 (15), that becomes activated following IL-3 binding (16). To gain further insight into the signaling pathways activated, and inactivated, by the binding of IL-3, we and others have recently set out to identify proteins that are intimately associated with the IL-3R itself. To date these studies have shown that the tyrosine-specific phosphatase, hematopoietic cell phosphatase, also known as PTP1C, SHP1, or SHPTP1, associates with the IL-3R ␤ subunit following IL-3 stimulation (17,18). This phosphatase is most likely responsible for the dephosphorylation of Jak2 and the IL-3R and for the subsequent down-regulation of the response of hemopoietic cells to IL-3 (17,18). In addition, a novel 110-kDa Ser/Thr kinase has been found to constitutively associate with the IL-3R and become tyrosine-phosphorylated in response to . This kinase may be involved in the extensive serine/threonine phosphorylation of the IL-3R that occurs following IL-3 stimulation (10).
Recently, we discovered that a phosphatidylinositol-3,4,5trisphosphate (PtdIns-3,4,5-P 3 ) 5-phosphatase (5-ptase), which we have designated SHIP, for SH2-containing inositol phosphatase, becomes associated with Shc following IL-3 stimulation (20). Since Shc has been shown to bind to many cell surface receptors following growth factor stimulation (21)(22)(23)(24)(25)(26), we investigated whether this 5-ptase could also be associating with the IL-3R following IL-3 stimulation. Unexpectedly, we found a 5-ptase was indeed associated with the IL-3R but it was not SHIP. Rather, it was a previously uncharacterized PtdIns-3,4,5-P 3 -specific 5-ptase that associates constitutively with the IL-3R and is most likely bound to this receptor indirectly, via PI 3-kinase. This is the first demonstration of a 5-ptase being associated with a cell surface receptor, and, given its specificity for the major in vivo product of PI 3-kinase, it most likely plays an important role in regulating PI 3-kinase-mediated events.

Reagents-
The production and purification of COS cell-derived murine IL-3 and granulocyte-macrophage colony-stimulating factor (GM-CSF) were as described previously (27). Rabbit antiserum to the human IL-3R ␤ c subunit was generated by immunizing animals with a gluta-* This work was supported by grants from the Medical Research Council and the National Cancer Institute of Canada with core support from the British Columbia Cancer Foundation and the British Columbia Cancer Agency. 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.
¶ Supported by National Institutes of Health Grants HL55672 and HL16634.

RESULTS
Recently, we cloned the inositol polyphosphate 5-ptase, SHIP, from a hemopoietic, IL-3-dependent murine cell line, B6SUtA 1 (20). Of the 5-ptases cloned to date (31,32), SHIP was unique in that it possessed an SH2 domain, became phosphorylated on tyrosine residues, and associated with Shc following IL-3 stimulation (20). These properties suggested that it might play an important role in cytokine-mediated signaling. To determine at what site(s) within the cell that SHIP exerted its effects, we reasoned that since Shc associates with many cell surface receptors following growth factor stimulation (21)(22)(23)(24)(25)(26), it might bring SHIP to the IL-3R. To test this, B6SUtA 1 cells treated with and without IL-3 were lysed, and PtdIns-3,4,5-Pspecific 5-ptase assays were carried out with anti-Shc and anti-IL-3R immunoprecipitates. As can be seen in Fig. 1, normal rabbit serum immunoprecipitates did not contain any 5-ptase activity while anti-Shc immunoprecipitates did, but, as shown previously (20), only after IL-3 stimulation. Anti-IL-3R immunoprecipitates, on the other hand, possessed 5-ptase activity constitutively, and the level of this activity did not change with IL-3 stimulation.
These results suggested a number of possibilities. A small amount of Shc could be associating with the IL-3R constitu- FIG. 1. A PtdIns-3,4,5-P 3 5-ptase is constitutively associated with the IL-3R. 4 ϫ 10 7 B6SUtA 1 cells, treated with (IL-3) or without (C) IL-3, were lysed and immunoprecipitated with either normal rabbit serum (NRS) or anti-Shc or anti-IL-3R antibodies. One-tenth of the immunoprecipitates (as well as recombinant 5-ptase II (PtII) and blank (BL) samples as positive and negative controls, respectively) were incubated with PtdIns-3,4,5-[ 32 P]P 3 under conditions where product formation was linear with time and the reaction products were separated by TLC. tively and in this context be capable of binding SHIP. Alternatively, SHIP could be binding independent of Shc to the IL-3R or a 5-ptase distinct from SHIP could be binding to the IL-3R. To test the first possibility, B6SUtA 1 cells treated with and without IL-3 were lysed, and the lysates were immunoprecipitated with either anti-Shc or anti-IL-3R antibodies. The immunoprecipitates were then subjected to Western analysis with anti-Tyr(P) antibodies (4G10). As can be seen in Fig. 2A, anti-Shc antibodies co-precipitated the tyrosine-phosphorylated form of the 145-kDa SHIP protein following IL-3 stimulation, in keeping with our previous results (20,34). It was difficult to determine, however, whether anti-IL-3R antibodies co-precipitated the tyrosine-phosphorylated form of SHIP because of the presence of the tyrosine-phosphorylated 140-kDa IL-3R ␤ subunit (34). Nevertheless, reprobing this blot with anti-Shc antibodies revealed that Shc associated with the IL-3R only following IL-3 stimulation. This suggested that the first possibility was unlikely.
To discriminate between the last two possibilites, i.e. whether SHIP or a different 5-ptase was associated with the IL-3R, we took advantage of our previous finding that SHIP also hydrolyzed the 5-position phosphate from Ins-1,3,4,5-P 4 (20) and asked if the IL-3R-associated 5-ptase shared this property. As shown in Fig. 2B, it did not. Thus, a 5-ptase distinct from SHIP associated with the IL-3R. Moreover, the fact that IL-3 did not increase the level of Ins-1,3,4,5-P 4 (or PtdIns-3,4,5-P 3 ) hydrolyzing activity in IL-3R immunoprecipitates indicated that the Shc associating with the IL-3R in response to IL-3 either does not have SHIP attached or, more likely, is present in too low amounts to detect its associated SHIP under the assay conditions used. Consistent with this result, we could not detect SHIP in anti-SHIP immunoblots of anti-IL-3R immunoprecipitates (data not shown).
At this point in our studies, Jackson et al. (35) reported the partial purification of a novel 5-ptase that removed the 5-position phosphate from PtdIns-3,4,5-P 3 but not from Ins-1,3,4,5-P 4 and that associated with PI 3-kinase in human platelets. As a first test to determine if a similar PI 3-kinase-associated 5-ptase might be responsible for the IL-3R-associated 5-ptase activity observed in B6SUtA 1 cells, we investigated whether PI 3-kinase was constitutively associated with the IL-3R by carrying out Western analysis, using antibodies to the p85 subunit of PI 3-kinase, with anti-p85 and anti-IL-3R immunoprecipitates from B6SUtA 1 cells. As can be seen in Fig. 3A, both the ␣ and ␤ isoforms of p85 were constitutively associated with the IL-3R in these cells. Since this was somewhat of a surprise, given that we and others have observed PI 3-kinase binding to other growth factor receptors only after growth factor stimulation (33, 36 -40), we sought to confirm this with PI 3-kinase assays. As shown in Fig. 3B, anti-IL-3R immunoprecipitates clearly possessed PI 3-kinase activity in the absence of IL-3 stimulation, although this activity increased following addition of IL-3. This increase was consistent with previous reports showing that IL-3 increases total cellular PI 3-kinase activity in hemopoietic cell lines (41). We then tested whether PI 3-kinase actually possessed an associated 5-ptase activity in our murine myeloid cell line by determining whether anti-p85 im-  3) or without (C) IL-3, were subjected to immunoprecipitation with anti-IL-3R antibodies, and the immunoprecipitated protein A beads were washed, PI 3-kinase assays were performed, and TLC was carried out as described previously under "Materials and Methods." Lysates from B6SUtA 1 cells, treated with IL-3, were immunoprecipitated with anti-p85 antibodies and treated as above to serve as a positive control. C, 4 ϫ 10 7 B6SUtA 1 cells, treated with (IL-3) or without (C) IL-3, were lysed and immunoprecipitated with either anti-p85 or anti-IL-3R antiserum. One-tenth of the immunoprecipitates were incubated with PtdIns-3,4,5-[ 32 P]P 3 under conditions where product formation was linear with time and the reaction mixture was fractionated by TLC. D, 1 ϫ 10 8 B6SUtA 1 cells, treated with IL-3 for 5 min at 37°C, were lysed and immunoprecipitated with either anti-p85, anti-IL-3R, or anti-SHIP (as a positive control) antiserum. The immunoprecipitates were incubated with PtdIns-3,4,5-[ 32 P]P 3 for 30 min at 37°C and then purified 4-ptase was added, as indicated, and incubation continued for an additional 30 min at 37°C before the products were separated by TLC. munoprecipitates were capable of hydrolyzing PtdIns-3,4,5-P 3 . As can be seen in Fig. 3C, this was indeed the case and 5-ptase activity did not change with IL-3 stimulation, consistent with it being the IL-3R-associated 5-ptase. Moreover, the 5-ptase activity associated with the anti-p85 immunoprecipitates was substantially higher, on a per cell basis, than that associated with the IL-3R (Fig. 3C). This too was consistent with the possibility that the IL-3R-associated 5-ptase was binding indirectly to the receptor via PI 3-kinase since only a minor proportion of the total cellular PI 3-kinase is associated with the IL-3R (i.e. Fig. 3A).
To confirm that the ptases associated with the IL-3R and p85 were indeed 5-ptases and not 4-ptases, PtdIns-3,4,5-[ 32 P]P 3 5-ptase assays were carried with anti-p85, anti-IL-3R, and anti-SHIP immunoprecipitates from B6SUtA 1 cells treated with IL-3, as usual, and the 32 P-labeled PtdInsP 2 product was incubated with or without purified inositol polyphosphate 4-ptase before separation by TLC (29). As can be seen in Fig.  3D, 4-ptase treatment of all three immunoprecipitates resulted in the appearance of a new 32 P-labeled spot that comigrated with PtdInsP. Since the initial PtdInsP 3 substrate was labeled in the 3-position, this confirmed that the IL-3R-and p85associated enzyme was indeed a 5-ptase.
As a further test to see if the PI 3-kinase-associated 5-ptase was identical to the IL-3R-associated 5-ptase, substrate specificity comparisons were undertaken, and, as is shown in Table  I, both 5-ptases only recognized PtdIns-3,4,5-P 3 as a substrate and both required Mg 2ϩ as a cofactor. This last property distinguishes this 5-ptase from that reported in human platelets (35) and, taken together with its substrate specificity, suggests that the 5-ptase associated with PI 3-kinase is identical to that associated with the IL-3R, that it associates with the IL-3R via its constitutive interaction with PI 3-kinase, and that it is distinct from other 5-ptases reported to date.

DISCUSSION
The identification of receptor-associated proteins is an essential first step in elucidating downstream signals initiated by the binding of growth factors to their receptors. In the case of IL-3, our current results, coupled with previous data, suggest that Jak2, PI 3-kinase, a 110-kDa serine/threonine kinase, and a novel 5-ptase associate constitutively with the IL-3R while Shc and hematopoietic cell phosphatase bind only following IL-3 stimulation. Although our finding that PI 3-kinase associates constitutively with the IL-3R was somewhat of a surprise, it is consistent with a recent report by Rao and Mufson (42) in which they demonstrated that a GST fusion protein containing only the membrane proximal 67 amino acids of the IL-3R ␤ c subunit cytoplasmic domain, and that was not detectably tyrosine-phosphorylated, was capable of binding PI 3-kinase in vitro (42). As to exactly how PI 3-kinase is binding to the IL-3R (which does not contain a consensus sequence for the SH2 domains of p85 on either its IL-3R ␣ or ␤ subunits), Jucker and Feldman (43) recently reported that a 76 -85-kDa tyrosinephosphorylated protein may link PI 3-kinase to the IL-3R ␤ subunit in human TF-1 cells. This is of interest given that we recently identified a 72-kDa protein that was both constitutively associated with the IL-3R and constitutively tyrosinephosphorylated in B6SUtA 1 cells (19).
The specific 5-ptase we have identified appears to be constitutively associated with PI 3-kinase (i.e. its activity does not change in anti-p85 immunoprecipitates with IL-3 stimulation), and, since the major in vivo product of PI 3-kinase is the substrate for this 5-ptase, it is likely that one of its functions is to maintain low levels of PtdIns-3,4,5-P 3 in unstimulated cells. Following IL-3 stimulation, when PI 3-kinase activity is increased, its function may be to return PtdIns-3,4,5-P 3 levels back to resting levels after the various PI 3-kinase-activated cascades have been triggered and/or to generate sufficient PtdIns-3,4-P 2 to activate similar or as yet uncharacterized pathways.  I  The IL-3R-and p85-associated 5-ptases have identical substrate specificities Assays were carried out as described previously (20, 28 -32).