Murine SHP-1 Splice Variants with Altered Src Homology 2 (SH2) Domains

SHP-1 is a protein-tyrosine phosphatase with two Src homology 2 (SH2) domains. These SH2 domains determine which proteins SHP-1 associates with, but they also autoregulate the activity of the catalytic domain. In this report, we find that the murineSHP-1 transcript is processed to yield a series of alternatively spliced in-frame transcripts, the majority of which exclude exons encoding one or the other SH2 domain. We have examined the corresponding protein isoforms in several ways. First, our measurements of V max and K m under different conditions indicate that the SH2 variants have elevated activity because of lessened autoregulation. Second, to ascertain whether regulation by the SH2 domains reflects intra- or intermolecular effects, we analyzed the state of SHP-1 by high performance liquid chromatography and sucrose density gradient centrifugation. Our results showed that SHP-1 is a monomer and, thus, is regulated in an intramolecular manner. Third, our analyses detected shape differences between SHP-1 and the active splice variant protein deleted of the amino-terminal SH2 domain; i.e. SHP-1 was globular and resistant to proteolytic digestion, while the splice variant protein was “rod-shaped” and more susceptible to proteolytic digestion.

Protein-tyrosine phosphatases (PTPs) 1 have major effects on the activity of numerous proteins that function in cell physiology and ontogeny. One such PTP, SHP-1 (previously called PTP1C, HCP, and SHPTP1) is predominantly expressed in hematopoietic cells and has gained much attention with the finding that motheaten and motheaten viable mice have mutations in the SHP-1 gene (1,2). Mice homozygous for these mutations have a puzzling phenotype, with characteristics of both immunodeficiency and autoimmunity. Affected animals die prematurely, usually of pneumonitis (3).
SHP-1 (4 -7), SHP-2 (8 -11), and csw in Drosophila melanogaster (12) form a class of protein-tyrosine phosphatases that have two tandem SH2 domains amino-terminal to the phosphatase domain. SH2 domains, which are present in many signaling molecules, bind phosphorylated tyrosine residue motifs in a sequence-specific manner (13) and hence function to bring two signaling molecules together to help propagate a signaling cascade. However, in the case of SHP-1 and SHP-2, their SH2 domains have a second function, namely to negatively regulate the activity of the catalytic domain (14 -16). This inhibitory effect is relieved when the amino-terminal SH2 domain binds to phosphorylated peptides (17,18). Biphosphopeptides that bind to both SH2 domains of SHP-2 activate SHP-2 at lower concentrations than do single phosphopeptides, arguing that both SH2 domains participate in this inhibitory effect (19). This notion is also supported by the crystal structure for SHP-2 that shows an intramolecular interaction between both SH2 domains and the phosphatase domain (20). Despite the sequence similarity between SHP-1 and SHP-2 (ϳ50% amino acid identity; Ref. 8), other comparisons have suggested that the SH2-mediated inhibition of SHP-1 and SHP-2 might be different. Thus, SHP-1 molecules that lack the carboxyl-terminal SH2 domain (SH2C) are still subjected to negative regulation by the amino-terminal SH2 domain (SH2N; Refs. 18 and 21). However, the polypeptide segment corresponding to SH2C is necessary in SHP-2 as a spacer to allow the SH2N to block the catalytic cleft (20). To accommodate the stereochemical restrictions, it is possible that autoregulation of SHP-1 might be due to intermolecular interactions between two or more SHP-1 molecules, as has been suggested before (21).
Two forms of SHP-1 transcripts have been detected in humans (22), denoted as (I)SHP-1 and (II)SHP-1. The cDNA sequences of (I)-and (II)SHP-1 are identical except in the 5Јuntranslated region and in the first few coding nucleotides (MLSRG are the first 5 amino acids of (I)SHP-1, and MVR are the first 3 amino acids of (II)SHP-1). Because the published mouse SHP-1 amino-terminal coding sequences (MVR) are identical (5,6), probably corresponding to the human (II)SHP-1 isoform, it is unclear whether mouse has more than one species of SHP-1 protein. Here, we report findings that indicate that the mouse also produces a second isoform, similar to the human (I)SHP-1. In addition, we also found that these transcripts are each spliced in multiple ways, thus generating in-frame RNAs. In some cases, these SHP-1 RNAs lack exons that encode either one of the SH2 domains. As the SH2 domains of SHP-1 possess autoregulatory functions (see above), proteins encoded by these * This work was funded by the National Cancer Institute of Canada. 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 nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  RNAs are expected to have novel regulatory properties. Our results confirmed this expectation; each of these SHP-1 variant proteins had elevated phosphatase activity. Furthermore, our analysis showed that autoregulation of SHP-1 is an intramolecular process.
Construction, Expression, and Purification of SHP-1 and Isoforms-Total RNA from murine bone marrow macrophages was prepared by the Trizol (Life Technologies, Inc.) method. The primer pairs (II)SHP-1-74-5Ј (5Ј-GTGCCTGCCCAGACAAACTG-3Ј) and SHP-1-1859-3Ј (5Ј-CCACAGGTCTCAGTCTATCGGGT-3Ј); (I)SHP-1-90-5Ј (5Ј-AACAGCT-GTGCCACTCGATTG-3Ј) and SHP-1-1859-3Ј were used for RT-PCR to obtain cDNAs corresponding to (II)-and (I)SHP-1, respectively. Amplified cDNAs were subcloned into the S1 nuclease-blunted BamHI site of pRSETb. Ligated DNAs were transfected into DH5␣, and clones were sequenced to check for errors. Constructs with the correct sequences were transfected into BL21(DE3), and single colonies were induced to express the fusion protein in liquid cultures with isopropyl-1-thio-␤-Dgalactopyranoside as described previously (23). These vectors were designed to produce a polyhistidine-tagged fusion protein of SHP-1 and its isoforms. In addition, an enterokinase cleavage site present within the fusion peptide allowed for removal of the peptide after purification. Purification of the fusion proteins from the soluble fraction was carried out using Probond nickel-resin columns (Invitrogen) as described previously (23).
Phosphatase Assays Using Purified Bacterial Protein-Phosphatase assays were carried out essentially as described by Pei et al. (17). Each fusion protein was assayed at a concentration of ϳ150 nM in 50 l of either buffer A (50 mM NaAc, pH 5.0, 2 mM EDTA, 2 mM dithiothreitol, and different concentrations of NaCl) or buffer B (50 mM Pipes, pH 6.0 or 7.0, 2 mM EDTA, 2 mM dithiothreitol, and indicated concentrations of NaCl) with the substrate para-nitrophenyl phosphate (pNPP; Sigma) for up to 2 min at 25°C (generation of product was linear for all SHP-1 isoforms for up to 10 min). Reactions were stopped with 450 l of 0.2 N NaOH, and the para-nitrophenylate anion was measured by absorption at 410 nm in a Beckman Du ® 640B spectrophotometer. Absorbance values were recalculated to moles using the extinction coefficient of 16,600 M Ϫ1 cm Ϫ1 (410 nm) for the product. In all experimental assays, less than 1% of the substrate was converted to product. Initial velocities (obtained by dividing moles of product generated by duration of the assay in minutes) were then plotted against varying pNPP concentration. The data were analyzed with SigmaPlot version 2.0 to obtain V max and K m values from nonlinear least squares fits of the rectangular hyperbolic form of the data.
Enterokinase Reactions-Digestions with enterokinase (EKmax, Invitrogen) were performed as suggested by the manufacturer. 2 g of (I)SHP-1 mj and S6 or SHP-1(C453S) mj and S6 were incubated overnight at 4°C in 0, 0.01, and 0.001 units of EKmax with the manufacturer's buffer. An aliquot was then analyzed by SDS-PAGE and immunoblotting.
Size Exclusion Chromatography-Size exclusion chromatography was performed on a Superdex 200 column (Amersham Pharmacia Biotech) with solvent delivery by a Beckman System Gold HPLC instrument. The column buffer was 50 mM Pipes (pH 7.0), 250 mM NaCl, 2 mM EDTA, and the flow rate was 0.5 ml/min. 20 g of purified (I)SHP-1 mj and S6 protein (two different purified preparations of each were assayed) were applied in 400 l of column buffer (i.e. initial enzyme concentration at approximately 750 nM), and elution volumes were measured by absorbance at 280 nm, as well as by phosphatase activity and Western blotting. For the purpose of apparent molecular mass calibration, the elution positions of 20 g each of BSA (monomer and dimer), ovalbumin, and carbonic anhydrase were measured. The elution volume (V e ) for each protein was divided by 7.1 ml (the void volume (V o ) of the column as measured by blue dextran) and expressed as V e /V o . A plot of log molecular weight of the standard globular proteins versus V e /V o yielded a straight line.
Sucrose Density Gradients-11-ml sucrose gradients (linear from 10 to 25%) were made in the following buffer: 50 mM Pipes, pH 7.0, 250 mM NaCl, 2 mM EDTA. 20 g each of glutathione S-transferase (GST), ovalbumin, BSA, and human IgG in 200 l of the same buffer were layered onto separate gradients and centrifuged in a SW41 rotor for 17 h at 40,000 rpm at 20°C. 20 g each of (I)SHP-1 mj and S6 (two different purified preparations of each were assayed) in 200 l of buffer without sucrose (i.e. initial enzyme concentrations at ϳ1500 nM) were also subjected to the same experimental conditions. 200-l fractions were collected. 50 l of each fraction was assayed for protein with Coomassie Protein Assay Reagent (Pierce), and 20 l was assayed for phosphatase activity in 50 l of Pipes, pH 6.0, 250 mM NaCl, 2 mM EDTA, 2 mM dithiothreitol, and 20 mM pNPP for 5 min at 25°C.
Phosphopeptide-The erythropoietin receptor peptide pY429 with the sequence DPPHLKpYLYLVVSDSK (where pY represents phosphotyrosine), previously described by Pei et al. (18), was synthesized and purified by reverse phase HPLC on a C8 column.

RESULTS
Characterization of Murine SHP-1 Isoforms-As noted above, the two published sequences for mouse SHP-1 correspond more closely to the human type II isoform than to the type I isoform. It has also been reported that human (I)SHP-1 is exclusively expressed in epithelial cells, while (II)SHP-1 is only found in hematopoietic cells (22). To assess whether mice also produce a type I isoform of SHP-1, we sequenced 5Ј-RACE products from SHP-1 transcripts from a mouse adrenal epithelial tumor cell line, Y1. With this strategy, described under "Experimental Procedures," we obtained two products, fragment X (ϳ360 bp) and fragment Y (ϳ250 bp) (Fig. 1). Sequencing fragment X (Fig. 1A) showed that it encoded the aminoterminal amino acids (i.e. MVR) similar to that for human type II SHP-1. Fragment Y (Fig. 1B) contained sequences that encoded the amino-terminal amino acids (i.e. MLSRG) similar to that of human type I SHP-1 (4). These results indicate that mouse also produces two distinct SHP-1 isoforms differing at their 5Ј-ends.
Analysis of the Murine Chromosomal SHP-1 Gene-In humans, the two SHP-1 isoforms are generated from the same gene via the use of alternative 5Ј exons (22). To assess whether each transcript isoform of mouse SHP-1 is also produced from two alternatively used 5Ј exons, we characterized the entire mouse SHP-1 locus. Assuming that the human and murine genetic structure was similar, we designed primers to amplify the complete mouse SHP-1 gene. Six genomic DNA fragments ( Fig. 2A) encompassing the complete SHP-1 gene were amplified by PCR from normal mouse DNA. Fig. 2A shows the locations of these PCR fragments and the primers used to obtain the intron-exon boundary sequences that are shown in Fig. 2B. Genomic DNA sequences were compared with the SHP-1 cDNA sequences in order to deduce the intron-exon boundaries. We obtained complete sequences for all introns except for the indicated four cases in which the introns were larger than 2 kb (GenBank TM numbers U65951-U65955). Similar to the human chromosomal SHP-1 gene, mouse SHP-1 has 17 exons ( Fig. 2A). (I)exon 1, which contains 5Ј-untranslated sequences and nucleotides encoding the first 5 amino acids (i.e. MLSRG) of (I)SHP-1, is located ϳ5.8 kb upstream of (II)exon 1, which contains 5Ј-untranslated sequences and nucleotides encoding the first 3 amino acids (i.e. MVR) of (II)SHP-1. Thus, both mouse and human (I)SHP-1 and (II)SHP-1 transcripts are generated by alternative 5Ј exon usage.
Identification of Murine SHP-1 Splice Variants in Bone Marrow Macrophages-(I)SHP-1 transcript variants lacking either one (exon 2) or both exons (exons 2 and 3) encoding the SH2N domain exist in human cells (22). To determine whether we could identify such transcripts in murine cells, we assessed the SHP-1 transcripts in murine bone marrow macrophages by Northern blots (Fig. 3A). These blots, using the phosphatase domain of SHP-1 as a probe, revealed the presence of multiple bands larger and smaller than the expected full-length transcript (Fig. 3A). To examine the sequence of these bands, we used RT-PCR to amplify the different RNA molecules. Expression of (I)SHP-1 transcripts was assessed using the primers (I)SHP-1-90-5Ј and SHP-1-1859-3Ј. We found not only the expected 1944-bp (I)SHP-1 mj product, but also a larger product (ϳ2000 bp) and two smaller products (ϳ1800 and ϳ1600 bp; Fig. 3B). The sequencing analysis of these products is summarized in Fig. 3C. As depicted in Fig. 3, the different RNA isoforms arise by alternative splicing. We denote these different isoforms with specific postscripts. Thus, the major forms indicated in Fig. 3C are denoted as mj. The spliced variants are denoted as S1-S6. Two SHP-1 structures are present within the ϳ2000-bp product; (I)SHP-1 S1 (2105 bp) retains the 171-bp intronic sequence between exons 15 and 16, and (I)SHP-1 S2 (2051-bp) retains the 117-bp intronic sequence between exons 4 and 5 (Fig. 3C). Two SHP-1 structures are present within the ϳ1800-bp product; (I)SHP-1 S3 (1811 bp) excludes sequences from exon 2, and (I)SHP-1 S4 (1820 bp) excludes a 114-bp segment from exon 4 (Fig. 3C). The ϳ1600-bp product (I)SHP-1 S6 (1626 bp) excludes sequences from exons 2 and 3 (Fig. 3C). This structure is similar to the previously reported human SHP-1 variant (22). All of these transcript variants of SHP-1 retain the normal reading frame.
Expression of the (II)SHP-1 transcripts was assayed by RT-PCR using the (II)SHP-1-74-5Ј and SHP-1-1859-3Ј primer pair. The 1934-bp product represented (II)SHP-1 mj (Fig. 3B). Two structures similar to S1 and S2 of (I)SHP-1 were identified in the ϳ2000-bp product (Fig. 3C). The ϳ1800-bp product contained three structures. Two were similar to S3 and S4 of (I)SHP-1, while the third, (II)SHP-1 S5 (1817 bp), excluded sequences from exon 5 (Fig. 3C). As with (I)SHP-1 transcript variants, all of these transcripts retain an open reading frame, and as a consequence, they are capable of encoding active phosphatases. Thus, most of these splice variants of SHP-1 would express proteins with altered amino-terminal or carboxyl-terminal SH2 domains. However, we found that all of these alternatively spliced transcripts of SHP-1 were present at low levels and that the corresponding protein products were not detected (see "Discussion"). 2 Expression and Purification of Murine SHP-1 in a Bacterial System-The SH2 domains of human SHP-1 inhibit the phosphatase domain by decreasing its affinity for its substrate (17). The naturally occurring mouse SHP-1 transcript variants encode SHP-1 proteins with deletions in their SH2 domains. These splice variant SHP-1 proteins can be used to gain information on the mechanism of autoinhibition. By comparing their kinetic properties relative to SHP-1 mj, we can assess whether they have elevated activities as predicted and thus map the motifs responsible for the inhibitory effects.
The SHP-1 cDNAs that were used in this study all encode proteins that have deletions affecting one or the other SH2 domain (Fig. 4A). In this study of autoregulation, we have compared these SHP-1 splice variant proteins with two other proteins: SHP-1 mj, which contains the complete SH2N and SH2C domains, and an engineered mutant protein of SHP-1, which consisted only of the phosphatase domain (PTPase domain; Fig. 4A). These cDNAs were cloned into the BamHI site of the pRSETb vector and transfected into the bacterial strain BL21(DE3) for protein expression. Cloning into the BamHI site produces fusion proteins that contain an N-terminal polyhistidine region that allows for purification by binding to nickel  1 and 2) and thymus (lanes 3 and 4) was electrophoresed in 1.5% agarose, transferred to nitrocellulose, and probed with a 32 P-labeled DNA segment corresponding to the SHP-1 phosphatase domain. B, RNA was prepared from normal mouse bone marrow macrophages and subjected to RT-PCR. Positions of the primers are shown above a schematic diagram of (I) and (II) SHP-1 mj RNA. The RT-PCR products were electrophoresed in 0.7% agarose, transferred to nitrocellulose, and probed with 32 P-labeled sequences of SHP-1 phosphatase domain. C, schematic diagram of SHP-1 RNA isoforms and the splice variant transcripts that were found in normal mouse bone marrow macrophages.
columns (see Fig. 5A for schematic of the fusion peptide). To allow for removal of the fusion peptide, an enterokinase cleavage site was placed immediately amino-terminal to the initiator methionine of SHP-1 and its isoforms. Bacterial lysates were prepared, and the fusion proteins were partially purified and analyzed by staining SDS-PAGE gels with Coomassie Blue (Fig. 4B). As (I)SHP-1 S4 was expressed at low levels and thus contained many bacterial proteins even after purification, Western blots using a carboxyl-terminal specific antibody were used to measure the SHP-1 concentration in each preparation. Using this information, all subsequent assays were ensured to have an equimolar amount of each protein (Fig. 4C).
The Fusion Peptide Does Not Affect the Kinetic Properties of SHP-1-As noted above, all fusion proteins included an enterokinase site, which was included to allow cleavage of the oligo-His fusion peptide (Fig. 5A). Cleavage at this site would remove a T7 epitope from the protein and, consequently, result in the loss of reactivity to an anti-T7 antibody by Western blot. As shown on Fig. 5B, (I)SHP-1 mj could not be cleaved at any of four different concentrations of enterokinase (Fig. 5B, lanes  6 -9). This site, however, was cleaved by enterokinase on (I)SHP-1 S6 (Fig. 5B, lanes 2 and 3), as exemplified by the absence of reactivity of the anti-T7 antibody to the smaller immunoreactive band observed with the anti-SHP-1 antibody. We also found that enterokinase cleaves at other presumably nonspecific sites on (I)SHP-1 S6 more than on (I)SHP-1 mj. Relative to the densitometric values obtained for the undigested controls for (I)SHP-1 S6 and (I)SHP-1 mj (lanes 1 and 5,  respectively), the band intensities for lanes 2 and 3 are 0.3 and 0.49, respectively, while the band intensities for lanes 6 and 7 are 0.22 and 0.5, respectively. Thus, ϳ5-fold more enterokinase is required to obtain a comparable level of proteolytic degradation for (I)SHP-1 mj than for (I)SHP-1 S6. These data suggest a structural difference between (I)SHP-1 mj and (I)SHP-1 S6 (see below for more discussion on this point).
To test whether the presence of the oligo-His fusion peptide affected the kinetic behavior of the proteins in question, we expressed in bacteria both (I)-and (II)SHP-1 mj and selected SHP-1 splice variant proteins without the oligo-His fusion peptide (for simplicity, hereinafter referred to as "native") and compared their kinetic parameters to those of their fusion protein counterparts. This analysis was applied to (I)-and (II)SHP-1 mj, S3, and S6. Fig. 6A depicts the bacterial expression of the (I)SHP-1 mj, S3, and S6, both as native protein and as fusion products. We then measured the K m of unpurified native and unpurified fusion proteins and compared these values to those obtained with purified fusion proteins, as shown in Fig. 6B. In all cases, the enzyme activity curves were similar for unpurified native and unpurified fusion proteins, while bacterial lysates alone could not hydrolyze pNPP at any concentration (Fig. 6B). Furthermore, the K m values for unpurified fusion protein were not significantly different than the values for the purified fusion proteins (Table I). We thus conclude that the fusion peptide had little effects on the kinetic properties of the isoforms and that we could therefore carry out enzyme kinetic analyses using the fusion proteins. As both (I)-and (II)SHP-1 mj had comparable results in all of the experiments performed in this study, data are only presented for one or the other.
Both SH2 Domains Are Required for Maximal Inhibition of the Phosphatase Domain-In these studies, we analyzed purified spliced variant proteins together with an engineered mutant that consisted only of the phosphatase domain (PTPase). Since the effects on the K m of deleting either one of the two SH2 domains were not known, the PTPase protein served to define the kinetic properties of the catalytic domain in its uninhibited state. Kinetic analyses at physiological conditions (i.e. pH 7.0, 150 mM NaCl) revealed that SHP-1 mj had very little activity, and thus it was not possible to calculate a K m value from the data (Fig. 6C). However, the K m for each of the other isoforms was readily measured. In some cases, the values approached that of the PTPase (Fig. 6C, Table II). Specifically, deletion of a portion or the entire SH2N domain of SHP-1 (i.e. S3 and S6, respectively) resulted in the complete activation of the catalytic domain. (Fig. 6C, Table II); i.e. the K m values of S3 and S6 are similar to that of the PTPase, which was ϳ6 mM pNPP (Fig. 6C, Table II), similar to the value of 12.6 mM obtained previously for the catalytic domain of human SHP-1 (17). Internal deletions of the SH2C domain (i.e. S4 and S5) resulted in partial activation of the catalytic domain, as evidenced by its K m value of ϳ20 mM pNPP, which is about 3-fold higher than the K m of the PTPase.
Although the K m of S4 and S5 (both missing portions of the SH2C domain) ranged from ϳ10 to 27 mM when assayed at pH 7.0 and 50 -150 mM NaCl, their K m were Ͼ100 mM when as-sayed at pH 7.0 and 250 mM NaCl (Table II). By contrast, the K m for the PTPase was at 6 -8 mM under all assay conditions. This suggests that at high NaCl concentrations and in the absence of an intact SH2C domain, the interaction between the SH2N domain and the catalytic domain was stabilized. However, at pH 6.0 and 250 mM NaCl, the K m of S4 and S5 were ϳ17-44 mM, while the K m for (I)SHP-1 mj remained at Ͼ100 mM (Table II). This suggests that a more stable interaction exists between the SH2 domains and the catalytic domain of (I)SHP-1 mj compared with that of S4 or S5. These data suggest that the SH2N domain is principally responsible for the inhibitory effects but requires an intact SH2C domain to exert its full effects.
Activation of Murine SHP-1 mj-Human SHP-1 is inactive under physiological conditions, but this inhibitory effect is alleviated at pH 5.0 (24), in 80% glycerol (25), or in the presence of tyrosine-phosphorylated peptides that bind to the SH2N (17,18). To test whether murine SHP-1 can be activated by similar means, we performed enzyme kinetic assays under these various conditions. We first assayed SHP-1 mj and splice variant proteins in 80% glycerol, the concentration that was previously shown to have maximum stimulatory effects for human SHP-1 (25). Although there was variability in the results due to the difficulty in carrying out reactions at this high glycerol concentration, all SHP-1 isoforms were found to be active in 80% glycerol at pH 7.0 (Table II).
Acidification of the reaction conditions to pH 5.0 activated (II)SHP-1 mj with a corresponding decrease in K m to ϳ20 mM pNPP (Table II). It is possible that at this low pH the electro-FIG. 6. Kinetic analyses of SHP-1 mj and isoforms. A, expression of SHP-1 isoforms as "native" or fusion proteins. Lanes 1, 3, and 5 of this Coomassie Bluestained SDS-PAGE represent the fusion proteins for (I)SHP-1 mj, S3, and S6, while lanes 2, 4, and 6 represent the native proteins for the same three proteins, respectively. B, initial velocities at varying substrate (pNPP) concentrations at pH 7.0, 150 mM NaCl for 50 ng of unpurified native (open symbols) or unpurified fusion proteins (closed symbols) of (I)SHP-1 mj (OE), S3 (q), S6 (f), and untransfected bacterial lysate (ࡗ). All assays were performed for 1 min at different substrate concentrations and stopped with 0.2 N sodium hydroxide, and absorbance at 410 nm was determined as a measure of product generated. C, initial velocities of partially purified SHP-1 and isoforms at varying pNPP substrate concentrations at pH 7.0, 150 mM NaCl. A 0.15 M concentration of each protein was assayed for 1 min. Values for K m and V max are presented in Table II. The solid lines represent the best fit of the data to the Michaelis-Menton equation, with the values of V max and K m in each case being presented in Table II. Because of the low level of activity for (II)SHP-1 mj, a satisfactory fit of the data with respect to the magnitude of the errors associated with V max and K m was not obtained.

TABLE I
The K m (millimolar concentrations of pNPP) of selected SHP-1 isoforms either "native" or as fusion proteins at the indicated pH Bacterial lysates alone were found not to hydrolyze pNPP at any concentration. a These values were adopted from Table II for comparison. b NaCl concentration at 50 mM for this assay and at 150 mM for all others.
static repulsive effects of amino acid side chains that have gained positive charges uncouple the SH2 domain(s) from the inhibitory site. In support of this premise, increasing the NaCl concentration to 250 mM reversed the activating effects of an acidified environment by increasing the K m (Table II), although it is possible that NaCl might directly affect the catalytic domain. Some SHP-1 splice variant proteins were inhibited at pH 5.0. The K m for S6 and the PTPase were inexplicably above 100 mM pNPP (Table II). It thus appears that the manner by which glycerol and an acidic environment activates SHP-1 and its isoforms is different.
Inhibition of (I)SHP-1 mj Is Intramolecular-Our data so far indicated that inhibition requires two intact SH2 domains but did not reveal whether the inhibitory SH2 domains act intra-or intermolecularly. To examine this issue, we subjected (I)SHP-1 mj to HPLC size exclusion analysis to assess whether the enzyme was a monomer or dimer. (I)SHP-1 mj was applied onto a Superdex 200 size exclusion column that was equilibrated to pH 7.0 (50 mM Pipes pH 7.0, 250 mM NaCl, 2 mM EDTA). Using the standard curve, (I)SHP-1 mj (calculated molecular mass of 72 kDa) was determined to migrate on the size exclusion column with an apparent molecular mass of 72 kDa. (Fig. 7B). No portion of SHP-1 mj was detected as a dimer (Fig. 7A). This suggests that (I)SHP-1 mj is a globular monomer and excludes the possibility that it is a stable dimer. Interestingly, (I)SHP-1 S6 (calculated molecular mass of 59 kDa) migrated through the column with an apparent molecular mass of 82 kDa (Fig. 7B).
Proteins with extended structures elute earlier on size exclusion columns (i.e. higher apparent molecular mass) than do globular proteins of equal molecular mass. The anomalous elution volume for (I)SHP-1 S6 on the HPLC column suggested that it existed either as a dimer or had an extended rather than globular conformation. To determine whether (I)SHP-1 S6 was a dimer or had an extended structure, sucrose density gradient sedimentation analyses of (I)SHP-1 mj and S6 were carried out. This assay can resolve this issue because a protein with an extended conformation sediments slower than a globular protein with an equivalent molecular mass, while a protein of larger mass sediments faster; i.e. if (I)SHP-1 S6 had an extended conformation, it would sediment slower than a 59-kDa globular protein, while if it was a dimer, it would sediment faster. In fact, as shown in Fig. 7C, S6 sedimented with an apparent molecular mass of about 45 kDa, indicating that S6 was most likely a monomer with an extended rather than globular conformation. Again, (I)SHP-1 mj sedimented as a globular monomer, confirming the results obtained by size exclusion chromatography and suggesting that inhibition arises predominantly through intramolecular interactions (Fig. 7C).
We were concerned that autoregulation might involve only a transient intermolecular interaction that was so short lived that it did not affect the exclusion and sedimentation properties. To test this possibility, the activity of (I)SHP-1 mj was assessed over a ϳ100-fold range of enzyme concentration (17-1400 nM). We reasoned that if the inhibition was due to an intermolecular interaction, there should be an inverse concentration dependence to the specific activity observed. However, as can be seen in Fig. 7D, the specific activity of (I)SHP-1 mj was invariant over the concentration range tested. In addition, Fig. 7D also shows that the specific activity of (I)SHP-1 S6 did not vary significantly over the same range tested, indicating that it does not require dimerization to be catalytically active. The concentration independence of enzyme specific activity together with the hydrodynamic data indicating that (I)SHP-1 mj is a monomer at 17 nanomolar to 1.4 micromolar concentrations both argue that autoregulation of (I)SHP-1 mj is intramolecular.
Can the Activation Status of the Enzyme Be Attributed to the Structural Difference between (I)SHP-1 mj and (I)SHP-1 S6?-As presented above, the difference in enterokinase cleavage between (I)SHP-1 mj and S6 suggested a structural difference between the two enzymes. Structural differences between (I)SHP-1 mj and S6 were in fact detected by size exclusion chromatography and by sucrose density gradient analyses. These structural differences might be attributed to the activation status of (I)SHP-1 mj. To determine whether activation of SHP-1 mj rendered the enterokinase recognition site more accessible, (I)SHP-1 mj was incubated with 60 M of the phosphorylated erythropoietin receptor peptide pY429 (EpoR pY429). At this concentration, this peptide activates (I)SHP-1 mj maximally by binding to its SH2 domains (18). 3 Nevertheless, enterokinase could not cleave at the enterokinase recognition site of (I)SHP-1 mj, but it still cleaved the same site on (I)SHP-1 S6 (Fig. 5B, lanes 4 and 10 -13). To ensure that the EpoR pY429 was not being dephosphorylated during the overnight enterokinase reaction, the experiments and the same results were duplicated with the catalytically inactive C453S point mutant of (I)SHP-1 mj and S6. 3 However, we found that SHP-1 mj aggregated when activated with a pH 5.0 buffer or with 60 M EpoR pY429 at pH 7.0. 3 Therefore, although the enterokinase experiments did not show an increase in proteolytic cleavage of (I)SHP-1 mj in the presence of EpoR pY429, the fact that this peptide induced aggregation and that aggregated complexes are usually resistant to proteolysis makes these experiments difficult to interpret.

DISCUSSION
Conservation of Gene Structure-In this study, we showed that murine macrophages express SHP-1 transcripts with two different amino-terminal encoding sequences ((I)SHP-1 and (II)SHP-1) and numerous splice variants. Two observations argue that these isoforms and variants are unlikely to be PCR artifacts. First, SHP-1 transcripts of corresponding lengths can be detected in Northern blots of total RNA from mouse tissue.
Second, all variations involve the use of alternative splice junctions. For the generation of (I)-and (II)SHP-1 S4, a cryptic splice site was used within exon 4. This alternative splice site sequence (CTTgtgcgt) is similar to a common version of the consensus 5Ј splice site sequence (CCGgtgagt) (26). However, as further discussed below, these alternative transcripts are present at low levels, and as a consequence, we believe that they are not physiologically relevant.
Our finding that mice, like humans, produce both type I and type II forms of SHP-1 argues that this conserved feature serves an important function. It is not evident whether the different forms have arisen because of a need to regulate SHP-1 transcription using distinct promoters or to produce two SHP-1 isoforms with different functions or properties. However, considering that we detected, at most, minor differences between (I)SHP-1 mj and (II)SHP-1 mj proteins and that we found most of the same splice variants in both I and II forms, we favor the former view.
It is unclear at present whether the naturally occurring mRNA isoforms are translated into stable proteins at physiologically significant levels. The motheaten mutation, a single base pair deletion in exon 3 of the SHP-1 gene, creates a novel splice site that is subsequently utilized to splice to exon 4 (1, 2). However, this splicing is out of frame, and thus SHP-1 mj is not produced in motheaten mice (2). We found that the SHP-1 gene in motheaten macrophages is also produced as multiple spliced variant transcripts that resemble those identified in normal macrophages, except for the altered splicing caused by the motheaten mutation. 2 Interestingly, we identified the (I)SHP-1 S6 transcript in motheaten macrophages, and since this transcript does not contain exon 3 and thus does not harbor the motheaten mutation, it might produce a SHP-1 variant protein. However, we failed to detect this protein in motheaten macrophages, suggesting that it is most likely not present at physiological levels in normal macrophages. 2 Furthermore, the finding that the transcript variants in normal macrophages represent at most minor species (i.e. S3, S4, S5, and S6 are collectively at ϳ5% the level of (II)SHP-1 mj) 2 suggests that the corresponding protein products are also not expressed at physiologically relevant levels. This is not surprising, because we found in this study that these splice variants encode disregulated phosphatases. That the catalytic activities of SHP-1 and SHP-2 are negatively regulated suggests that a constitutively active SHP-1 or SHP-2 is toxic. Thus, these active splice variant proteins might be toxic to cells if present at high levels.
Mechanism of Autoregulation-The structural similarity between SHP-1 and SHP-2 (ϳ50% amino acid identity; Ref. 8) suggests that autoregulation of these two phosphatases is accomplished by similar mechanisms. The finding that human SHP-1 chromatographs as a monomer suggests a role for intramolecular interactions between the SH2 and the catalytic domains (24). Indeed, the crystal structure for SHP-2 (20) shows that the SH2 domains fold back onto the catalytic domain. In the SHP-2 structure, there are extensive interactions between the SH2N and the catalytic domain. Asp 61 , located within the SH2N domain of SHP-2 (Asp 59 for SHP-1), blocks the catalytic pocket and hydrogen bonds to the catalytic cysteine in the phosphatase domain (20). However, other data suggest that SHP-1 and SHP-2 might be autoregulated in different ways. First, the SH2 domains and catalytic domain of SHP-2, when expressed as separate polypeptide chains, can associate and inhibit the catalytic domain in vitro (16), a feature that was not reproduced with SHP-1 (17). Second, agents that activate SHP-1, such as glycerol and other 1,2-diols, inhibit the activity of SHP-2 (25). Finally, complete deletion of the SH2C domain from human SHP-1 did not impair autoreg- Elution volumes at peak absorption values (shown at the top of each peak) were used to construct the plot shown in B. Peaks were confirmed to contain (I)SHP-1 mj and S6 by Western blotting using the carboxylterminal specific anti-SHP-1 antibody (data not shown). B, HPLC elution positions for (I)SHP-1 mj and S6 relative to standard curve. Data are presented as elution volume (V e ) divided by the void volume (V o ) of the Superdex column measured using blue dextran. The molecular masses of (I)SHP-1 mj and S6 employed for calculating the y coordinate represent the theoretical value of the monomer species in each case. Data are representative of three experiments. C, sucrose density gradient sedimentation analysis. Data are presented as peak elution fraction measured by both protein determination and phosphatase activity assays. Data plotted as molecular weight to the 2 ⁄3 power (m). The molecular masses of (I)SHP-1 mj and S6 employed for calculating the y coordinate represent the theoretical value of the monomer species in each case. Data are representative of three experiments. D, activities over a concentration range. At a substrate concentration of 35 mM pNPP, (I)SHP-1 mj and S6 were assayed in 50 mM Pipes, pH 7.0, 250 mM NaCl, 2 mM EDTA, 2 mM dithiothreitol, and 0.1 mg/ml BSA for up to 5 min. (I)SHP-1 mj and S6 were assayed at approximate concentrations of 17, 50, 150, 450, and 1400 nM. Data points are averages of three experiments. Error bars are omitted for simplicity (range is less than 30% of the average). ulation (18,21). This is surprising because, based on the crystal structure for SHP-2, the SH2N domain would not be expected to reach the catalytic pocket without the SH2C domain acting as a spacer (20). Pregel et al. (21) suggest that inhibition of the human SH2C deletion mutant, as well as normal SHP-1, might be mediated by the intermolecular association of two molecules in a head to tail manner.
Our analyses of the full-length protein showed that SHP-1 mj is regulated intramolecularly; i.e. we showed using size exclusion chromatography and sucrose density gradient sedimentation analyses that murine SHP-1 mj is a globular monomer and that the specific activity of (I)SHP-1 mj was invariant over an ϳ100-fold concentration range. These results indicated that SHP-1 does not form stable dimers, and consequently, this concentration independence of specific enzyme activity indicates that SHP-1 mj was not inhibited by the SH2 domains from another SHP-1 molecule, thus ruling out transient intermolecular interactions. Our finding that partial deletions of the SH2C of murine SHP-1 (i.e. S4 and S5) activate the catalytic domain is also consistent with SHP-1 mj being inhibited intramolecularly. As argued above, such deletions would be expected to impede the SH2N from reaching and blocking the substrate binding pocket, based on the structure of SHP-2. It is not clear why our data differ from previous work (18,21). As there is a high degree of amino acid identity between human and mouse SHP-1 sequences (ϳ95%; Ref. 6), it appears unlikely that species specificity is the reason for the discrepancy. A possible reason for the difference is that our splice variants (i.e. S4 and S5) only deleted portions of the SH2C, which contrasts with the mutants from the previous report that delete the entire SH2C (18,21). Perhaps the portions of the SH2C domain in our splice variants prevented the proposed intermolecular interactions (21). Nevertheless, we conclude that SHP-1 is autoregulated through intramolecular interactions between the SH2 domains and the catalytic domain, just as with SHP-2.
Evidence for Conformational Change Associated with Activation-Many laboratories have proposed that an inactive SHP-1 exists in a closed state and undergoes a conformational change upon activation (open state). This is exemplified by the number of models presented in publications illustrating how SHP-1 "opens up" when it engages a substrate with its SH2 domains (e.g. Ref. 17). The open state for SHP-1 has been only hypothetical, since there are no data to suggest that such a state exists for an active SHP-1. Our data, which correlate the structural differences between (I)SHP-1 mj and S6 to the catalytic state of the phosphatase, support models in which SHP-1 "opens up" when activated. Both our HPLC size exclusion and sucrose density gradient sedimentation analyses detected hydrody-namic differences between SHP-1 mj and S6. These assays were consistent in showing that inactive SHP-1 mj was a globular protein (closed state) and that active S6 had an extended conformation (open state). The conformational differences between SHP-1 mj and S6 were also detected with proteolytic sensitivity assays (Fig. 5). These results indicate that our HPLC gel filtration and sucrose density gradient sedimentation assays were sensitive enough to detect such structural differences. Unfortunately, activating SHP-1 mj with either pH 5.0 buffers or activating concentrations of EpoR pY429 induced aggregation, 3 so we were not able to test directly if SHP-1 mj undergoes a conformational change upon activation.