Modulation of the ERG K+ Current by the Tyrosine Phosphatase, SHP-1*

We reported previously (Cayabyab, F. S., and Schlichter, L. C. (2002) J. Biol. Chem.277, 13673–13681) a functional interaction between the ERG-1 K+ channel and Src tyrosine kinase, which increased the current. We now show that the tyrosine phosphatase, SHP-1, which is present in microglia, is increased after brain damage, and is activated by colony-stimulating factor-1, associates with ERG-1 and regulates the current. Patch clamp recordings from the MLS-9 microglia cells were made with pipette solutions containing a recombinant SHP-1 protein: wild type (SHP-1 wild type (wt)), catalytically active (SHP-1 S6), or the substrate-trapping mutant (SHP-1 Cys → Ser). SHP-1 wt and SHP-1 S6 proteins decreased the current, an effect that was reversed by the phosphatase inhibitor, pervanadate, whereas SHP-1 Cys → Ser increased the current. Moreover, transient transfection with cDNA for SHP-1 wt or SHP-1 S6 decreased the ERG current without decreasing the protein level. Tyrosine phosphorylation of ERG-1 was decreased by transfection with SHP-1 wt and increased by SHP-1 Cys → Ser. The decrease in current by active SHP-1 was partly attributed to changes in the voltage dependence of activation and steady-state conductance, whereas inactivation kinetics and voltage dependence were not affected. Our results show that ERG-1 is a SHP-1 substrate constituting the first report that an ion current is regulated by SHP-1.

be heart-specific, where its natural mutations underlie one type of life-threatening arrhythmia (2)(3)(4). However, HERG and ether-à -go-go K ϩ channels are expressed in certain cancers, leading to intense interest in their contributions to proliferation and cell survival (5,6). We recently provided the first direct evidence of a functional role for HERG in cancer cells (7). HERG was selectively up-regulated in primary leukemias and several hematopoietic cell lines, and the HERG-channel blocker, E-4031, reduced proliferation in some of the cell lines. HERG and the three known rat ERG isoforms are expressed in the nervous system, and although ERG-2 was thought to be restricted to a small subpopulation of neurons (8), we found ERG-2 mRNA in rat microglia (1). ERG channels in neurons can modulate the resting potential (9), spike frequency (10), hormonal secretion (11,12), and neuritogenesis (9,13), although some roles have only been studied in cell lines. We previously identified ERG currents and ERG-1 protein in the brain microglia cell line (MLS-9) that we developed (1,14,15). Because MLS-9 cells lack the Kv1.3 and classical inward-rectifier currents that are prevalent in primary cultured microglia from which they are derived, ERG is expected to perform most roles of K ϩ channels in these cells.
Microglia and other ERG-expressing cells have protein tyrosine kinase-dependent signaling processes. We presented previously (1) the first report that tyrosine phosphorylation of the ERG protein modulates the ERG current. In MLS-9 cells, ERG-1 protein associates with, and is phosphorylated by, the cytosolic protein tyrosine kinase, Src. Phosphorylation is decreased by protein tyrosine kinase inhibitors, and several means of inhibiting endogenous Src activity reduce the current. Conversely, activating endogenous Src or transfecting constitutively active v-Src increases the current and alters its voltage dependence and kinetics, producing much more ERG current at negative potentials. The ERG current is then poised to promote microglial functions that require a negative membrane potential. Thus, we wanted to identify protein tyrosine phosphatase(s) that counter-modulate the ERG current. For several reasons, we considered SHP-1, an SH2 (Src homology 2)containing protein tyrosine phosphatase, as a likely candidate. (i) SHP-1 is expressed predominantly in hematopoietic cells (16), including microglia, and is up-regulated in microglia after traumatic brain injury (17). (ii) SHP-1 is activated by tyrosine phosphorylation after the mitogen, colony-stimulating factor-1 (CSF-1) binds to its receptor (18,19). Microglia proliferation is strongly stimulated by CSF-1 (20), a process that was used to produce the cell line, MLS-9 (15). (iii) An immunoreceptor tyrosine-based inhibitory motif (ITIM) consensus (Leu-Thr-Tyr(P) 829 -Cys-Asp-Leu) (21)(22)(23) is present in the cyclic nucle-otide-binding domain in the carboxyl terminus of ERG-1. ITIM is an SH2-binding motif that is present in a number of cellular receptors (NK cell inhibitory receptors, antigen receptors, CD22, and interleukin-3 receptors) that are associated with inhibitory signaling within the immune system. (iv) Brains of homozygous motheaten mice, which are SHP-1 null, have decreased numbers of all subtypes of glia (24), suggesting that SHP-1 plays an important role in the normal differentiation and distribution of astrocytes, microglia, and oligodendrocytes in the central nervous system.
Despite this wealth of information, nothing is known about the actions of SHP-1 on any ion channel. We have now exploited several SHP-1 variants to delineate the role of SHP-1 in modulating the ERG current. Inactive SHP-1 apparently exists in a "closed" state and then undergoes a conformational change to an "open" state on activation (25). By using high pressure liquid chromatography size-exclusion and sucrose density gradient sedimentation analyses, we showed previously that the inactive wild type SHP-1 (SHP-1 wt) is a globular protein (closed state) and the active SHP-1 S6 (lacking the aminoterminal SH2 domain) has an extended conformation (open state) (26). Mutation of the catalytic cysteine residue (SHP-1 Cys 3 Ser) from the signature motif eliminates enzymatic activity, while allowing normal substrate binding. It can protect the target protein from dephosphorylation by trapping the substrate (27) and thus the SHP-1 Cys 3 Ser mutant acts in a dominant-negative manner (28). In the present study, we found that SHP-1 protein interacts with and dephosphorylates the native ERG protein in MLS-9 cells. We used the three types of SHP-1 protein (SHP-1 wt, SHP-1 S6, and SHP-1 Cys 3 Ser) to show that SHP-1 is a negative regulator of ERG current and to identify ERG-1 as bona fide substrate for SHP-1. This is the first demonstration that SHP-1 regulates any ion channel.

EXPERIMENTAL PROCEDURES
Culturing the Microglia Cell Line, MLS-9 -The MLS-9 cell line was derived by treating pure cultures of rat microglia with colony-stimulating factor-1 (CSF-1) for several weeks and then harvesting microglia colonies (15,29). Like cultured rat microglia, MLS-9 cells stain with isolectin B4 and the antibodies OX-42 and ED-1. They are not labeled with antibodies against the astrocyte marker, glial fibrillary acidic protein, or the fibroblast protein, fibronectin. MLS-9 cultures were grown to 75-80% confluency in endotoxin-free minimal essential medium (MEM) containing 5% horse serum, 5% fetal bovine serum, and 50 g/ml gentamycin. For harvesting, they were washed twice with sterile phosphate-buffered saline (PBS) and then released from the flask by incubating (10 min, 37°C) with sodium citrate solution (130 mM NaCl, 15 mM sodium citrate, 10 mM HEPES, 10 mM D-glucose, pH 7.4). After adding an equal volume of MEM to the cell suspension, the cells were centrifuged at 700 rpm for 10 min and then resuspended in MEM. All cell culture reagents were from Invitrogen.
Patch Clamp Recordings-Whole-cell patch clamp recordings of ERG currents were made as described previously (1). An Axopatch 200A amplifier (Axon Instruments, Foster City, CA) was used with on-line compensation for series resistance and capacitance. The signals were filtered at 5 kHz and analyzed using pCLAMP 6.0 software (Axon Instruments). Pipettes with resistances of 3-5 M⍀ were made from thin walled borosilicate glass capillaries (WPI, Sarasota, FL). Recordings were from isolated bipolar MLS-9 cells with relatively small series resistances (4 -15 M⍀), which were compensated to Ͻ5 M⍀, and because the currents were Ͻ1000 pA, the maximal voltage error was 5 mV. Only cells exhibiting adequate voltage control were included, as judged by mono-exponential decays for the capacitive current and deactivation at very negative potentials. Curve fitting for kinetics and voltage dependence used non-linear least squares routines in Microcal Origin Version 5.0 (Microcal Software, Northampton, MA).
Chemicals-The HERG blocker, E-4031, was prepared as a 10 mM stock solution in distilled water, stored at Ϫ20°C, and then diluted in bath solution to the final concentration. Previously, we used the specific Src-inhibiting peptide, src40 -58, and we showed that the scrambled peptide (src40 -58s) was an inactive control (1). Thus, in the present study, we used cells containing the scrambled peptide, src40 -58s, as a negative time-matched control for experiments in which pipette solutions were supplemented with the recombinant proteins, SHP-1 wt, SHP-1 S6, and SHP-1 Cys 3 Ser. All peptides (synthesized at the Hospital for Sick Children, Toronto, Canada) were prepared as concentrated aqueous stock solutions in 0.1% bovine serum albumin, stored at Ϫ80°C, and then thawed and diluted in pipette solution just before use. The final pipette concentration of each protein was 0.1 mg/ml. Pervanadate was prepared as a 200 mM stock solution in distilled water, stored at Ϫ20°C, and then diluted in bath solution to the final concentration.
Transfecting MLS-9 Cells-The properties of the recombinant SHP-1 proteins with differing catalytic activity and the mammalian expression plasmids, pABA-neo (containing SHP-1 wt or SHP-1 S6), p4AD (containing SHP-1 wt), p4AE (containing SHP-1 Cys 3 Ser), and pRSV␤gal have been described (26,30). Each dish of MLS-9 cells was transiently transfected with 2 g/ml of a vector alone (pABA-neo or pRSV␤gal) or with SHP-1 wt, SHP-1 S6, or SHP-1 Cys 3 Ser, along with 1 g/ml pEGFP cDNA, using LipofectAMINE (Invitrogen) as described previously (31). Patch clamp recordings were made from green fluorescent cells 24 -48 h after transfection. For biochemical analyses, cells were harvested 24 h after transfection by scraping and then lysed in ice-cold modified RIPA buffer (see below), and the protein concentration of each cell lysate was determined using the Bio-Rad DC protein assay (Bio-Rad).
Immunoprecipitation and Western Blot Analyses-To monitor the tyrosine phosphorylation of native ERG-1, the proteins were immunoprecipitated using an anti-phosphotyrosine antibody, and then Western blots were probed for ERG-1, as follows. MLS-9 cells were lysed in a solubilization buffer (20 min, 4°C) that contained 1% Triton X-100, 25 mM Tris-Cl, pH 7.5, 150 mM NaCl, 100 mM NaF, 5 mM EDTA, 1 mM Na 3 VO 4 , and the protease inhibitors leupeptin (2 g/ml), aprotinin (2 g/ml), and phenylmethylsulfonyl fluoride (1 mM). The lysates were centrifuged at 15,000 ϫ g (15 min, 4°C) to remove cellular debris. The supernatant was cleared by incubation with protein A/G-agarose (3 mg/ml, 1 h) (Calbiochem) and centrifuged to remove the agarose. Tyrosine-phosphorylated proteins were immunoprecipitated from 100 g of total protein by incubating overnight at 4°C with anti-phosphotyrosine antibody (PY20, Medicorp, Montreal, Quebec, Canada) and then incubating for 3 h in protein A/G-agarose, followed by centrifugation. The immunoprecipitates were washed three times with ice-cold solubilization buffer containing 0.1% Triton X-100 and then eluted in 50 l of gel-loading buffer containing 120 mM Tris-HCl, pH 6.8, 2% SDS, 2% ␤-mercaptoethanol, 25% glycerol, 0.01% bromphenol blue, and 1 mM Na 3 VO 4 . For Western analysis, protein concentrations were measured as above, and then proteins were run on a 6.5% polyacrylamide gel, electrotransferred to nitrocellulose, and blocked with 5% nonfat milk in PBS containing 0.1% Tween 20 (PBST). The membrane was incubated overnight at 4°C with a polyclonal anti-HERG (human homologue of ERG-1) antibody (1:160; Alomone Labs, Jerusalem, Israel). After four washes with PBST, the membranes were incubated (1 h, room temperature) with horseradish peroxidase-conjugated secondary antibody (1: 3000; Cedarlane Labs, Hornby, Ontario, Canada). Following another four washes with PBST, labeled proteins were visualized using enhanced chemiluminescence (ECL, Amersham Biosciences) on XAR-2 film (Eastman Kodak Co.), and the signals were quantified by densitometry (Bio-Rad model GS-670). In parallel, Western blots were prepared using 50 g of total protein and probed with anti-cyclin D 1 /D 2 (1:500; Upstate Biotechnology, Inc.), and the signals were analyzed by densitometry and used to normalize amounts of tyrosine-phosphorylated ERG-1. In addition, Western analysis was used to analyze the relative phosphotyrosine levels of ERG-1 protein using an anti-phosphotyrosine antibody (4G10, Upstate Biotechnology, Inc., Lake Placid, NY) or anti-HERG antibody on total cell lysates (20 -40 g) from vector control, SHP-1 wt, and SHP-1 Cys 3 Ser transfectants.
Statistical Analysis-Data are expressed as mean Ϯ S.E. When appropriate, we used the two-tailed Student's paired t test or ordinary analysis of variance tests with the Bonferroni corrections multiplecomparison post-test, performed with INSTAT2 software (GraphPad Instat Software, version 2.04, Sunnyvale, CA). In either case, p Ͻ 0.05 was considered statistically significant.

SHP-1 Proteins
Modulate the ERG Current-There are two main mechanisms whereby SHP-1 might modulate ERG functions. (i) An ITIM consensus sequence is present in the cytoplasmic carboxyl terminus of the ERG-1 protein (see Introduction). If SHP-1 interacts with ERG-1 via this ITIM motif, we would expect it to down-regulate the ERG current. (ii) We showed previously that in MLS-9 cells, activated c-Src, and transfected, constitutively active v-Src increase the ERG current (1). It has been reported that SHP-1 relieves the autoinhibition of c-Src, thereby activating it (32). Thus, it is possible that SHP-1 would increase ERG current by activating c-Src. To distinguish between these possibilities, we used two approaches. To assess the functional role of SHP-1 in regulating the ERG current in MLS-9 cells, we used whole-cell recordings with purified SHP-1 wt or mutant proteins in the pipette or transient transfections with SHP-1 wt, mutant constructs.
Recombinant His-tagged SHP-1 wt and mutant SHP-1 S6 and SHP-1 Cys 3 Ser proteins were purified using Talon® (Clontech) columns (26). We assessed acute effects of each SHP-1 protein on ERG current by including it in the recording pipette solution. Scrambled src40 -58 (src40 -58s) peptide, which did not affect the ERG current ( Fig. 1), was used as a time-matched negative control (1). The peak currents at 30 min were 374.5 Ϯ 48.5 pA versus 403.6 Ϯ 39 pA with and without src40 -58s, respectively (n ϭ 7, p Ͼ 0.05). Spontaneous rundown was Ͻ10% at 30 min compared with the initial current during the first 5 min of recording. However, with either SHP-1 wt or SHP-1 S6 in the pipette solution, the peak ERG currents decreased significantly and to similar plateau levels (Fig. 1B). With SHP-1 wt, the average decrease was 53%, i.e. 432.4 Ϯ 50.6 pA for the first 5 min versus 203 Ϯ 53 pA at 30 min (n ϭ 7, p Ͻ 0.01). With SHP-1 S6, the current decreased by 47%, 551.6 Ϯ 68.3 pA for the first 5 min versus 290.8 Ϯ 51 pA at 30 min (n ϭ 6, p ϭ 0.01). Conversely, SHP-1 Cys 3 Ser, which acts in a dominant-negative manner, increased the ERG current by 97% from 443.6 Ϯ 27.4 pA in the first 5 min to 872.1 Ϯ 101 pA at 30 min (n ϭ 4, p Ͻ 0.01). Thus, active SHP-1 phosphatase reduces the native ERG current in MLS-9 cells. Pervanadate, an inhibitor of protein tyrosine phosphatases, prevented the action of SHP-1 (Fig. 2). The current was decreased by 52% (from 576 Ϯ 86.8 pA in the first 5 min to 277.1 Ϯ 50.3 pA after 30 min of treatment) by SHP-1 wt or SHP-1 S6 in the pipette (n ϭ 4, p Ͻ 0.05). It was restored to the control level after 500 M pervana-date was added to the bath (455.8 Ϯ 97 pA, n ϭ 4, p Ͼ 0.05 versus control cells).

FIG. 1. Recombinant SHP-1 tyrosine phosphatase modulates ERG current in MLS-9 cells.
A, ERG channels were opened by holding the membrane potential at ϩ20 mV (see text for further explanation), and then inward ERG currents were evoked by stepping to a test potential of Ϫ120 mV. Test pulses were delivered every 60 s to monitor time-dependent changes in ERG currents produced by the scrambled peptide (src40 -58s) or one of the SHP-1 recombinant proteins in the pipette (100 g/ml): wild type (SHP-1 wt), catalytically active (SHP-1 S6), and substrate-trapping mutant (SHP-1 Cys 3 Ser). n ϭ 12) was reduced by 35-40% (p Ͻ 0.01 versus vector control). The current in SHP-1 Cys 3 Ser transfectants was unchanged (29.0 Ϯ 2.4 pA/pF, n ϭ 6). Moreover, the changes in current density were not caused by decreases in channel or Src proteins. Rather, ERG-1 protein increased by 2.1-fold in SHP-1 wt and 1.8-fold in SHP-1 Cys 3 Ser transfectants (Fig. 3, p Ͻ 0.05 versus vector controls), and Src levels were not decreased (data not shown).
Effects of SHP-1 Transfectants on ERG Current Kinetics-In principle, an SHP-1-induced decrease in ERG current could result from changes in voltage dependence or from faster deactivation or inactivation. We found previously (1) that ERG channel closing was slower when MLS-9 cells were transfected with an active v-Src tyrosine kinase, and this contributed to the observed increase in current. Therefore, we used transiently transfected MLS-9 cells to assess whether SHP-1 caused changes in biophysical properties of the currents.
By using a voltage protocol (Fig. 5A) previously used to study HERG current deactivation (channel closing) (11), we found that the average time constants (Fig. 5C) were the same for vector control, SHP-1 wt, and SHP-1 S6 transfectants at all potentials tested. For all transfectants, deactivation was well fitted by a mono-exponential function. At Ϫ120 mV, the deactivation time constants were 98.8 Ϯ 25.8 ms for the vector control (n ϭ 8), 98.1 Ϯ 12.3 ms for SHP-1 wt (n ϭ 15), and 87.0 Ϯ 6.0 ms for SHP-1 S6 (n ϭ 17) (p Ͼ 0.05 (Student's t test with Bonferroni post hoc test; p Ͼ 0.05). The time course of current activation was not assessed because there was no outward activating current under the conditions of our study, i.e. normal internal and high external K ϩ . Current inactivation was examined using a triple-pulse protocol (1,8,33,34). From Ϫ80 to ϩ40 mV (Fig. 5B), inactivation dominated the current relaxation, which was well fitted by a mono-exponential function. Below these voltages, inactivation was not assessed because deactivation was significant (see Fig. 5A). The time constants were indistinguishable between control and SHP-1 transfectants over this voltage range (Fig. 5D). Together, these data suggest that SHP-1 does not exert its effects by changing the kinetics of channel closing or inactivation.
SHP-1 Transfection Alters Specific Voltage-dependent Properties of the ERG Current-We observed previously (1) that when Src tyrosine kinase was activated in MLS-9 cells, a neg-ative shift in the ERG activation versus voltage relation increased the tonically activated "window" current. Thus, we asked whether SHP-1 decreases the ERG current by an opposite shift in voltage dependence.
The Voltage Dependence of Inactivation Is Not Changed-Steady-state inactivation was monitored using a well established protocol (3,8) wherein channels were activated (and inactivated) by a 1-s-long pre-pulse to ϩ20 mV, and the peak inward current was measured at various test potentials to assess relief from inactivation (Fig. 6A). For each cell, a current versus voltage (I-V) relation was constructed (not shown), and the maximal slope conductance (G max ) was calculated from a linear fit to the I-V plot between Ϫ140 and Ϫ80 mV. We then determined the voltage dependence of the channel rectification factor, which reflects the deviation of the current from the value expected if there is no inactivation. The resulting averaged data (Fig. 6B) were fitted with Boltzmann equations, and no differences in steady-state inactivation (rectification) were  . 1st lane). The reason for two bands of such similar size is not known, but one possibility is that they represent differential phosphorylation. B and C, Western blots like those in A were analyzed by densitometry, and the relative amounts of SHP-1 and ERG-1 proteins were normalized to cyclin D 1 /D 2 . Values are mean Ϯ S.E. for the number of cells indicated. *, p Ͻ 0.05; **, p Ͻ 0.01, compared with vector-transfected cells.
Differences following SHP-1 transfections are relevant to the physiological functioning of this current. Considering that microglia do not produce over-shooting action potentials, the steady-state channel activity (proportion of tonically activated K ϩ channels) is expected to be a major factor in determining their contribution to cell function. To calculate the steady-state fraction of open channels (window current) (Fig. 6E), the fraction of channels inactivated at each voltage was multiplied by the fraction of channels activated at the same voltage. The steady-state conductance (window current) of ERG in MLS-9 cells (Fig. 6E) was very similar in shape and amplitude to heterologously expressed ERG-1 channels (8) with a peak of about 11% of the maximal conductance. For all SHP-1 transfectants, the steady-state conductance decreased to 7-9%, and the voltage at which the peak occurred was shifted by ϩ6 to ϩ10 mV compared with the vector controls. Together, these changes predict that SHP-1-transfected cells will have less tonically active current near the membrane potentials reported for microglia cells.
Pervanadate Changes the Voltage Dependence of ERG Current-Because the decrease in peak ERG current by SHP-1 was reversed by pervanadate (Fig. 2), we asked whether pervanadate also affects the voltage dependence, with or without SHP-1 transfection. In MLS-9 cells transfected with vector alone (Fig. 7A), pervanadate shifted the conductance versus voltage (G-V) curve by Ϫ16 mV, without changing the slope factor. Specifically, the V1 ⁄2,act shifted from Ϫ23.5 Ϯ 0.9 to Ϫ39.6 Ϯ 1.2 mV after pervanadate treatment (p Ͻ 0.001, n ϭ 3), whereas k act was 13.6 Ϯ 0.7 mV before and 16.1 Ϯ 1.0 mV after pervanadate (p Ͼ 0.05, n ϭ 3). In contrast, the rectification factor (R) versus voltage curve was not affected by pervanadate: V1 ⁄2,inact was Ϫ55.0 Ϯ 3.7 versus Ϫ60.3 Ϯ 4.2 mV (p Ͼ 0.05, n ϭ 4), and k inact was 24.5 Ϯ 4.4 mV before and 22.6 Ϯ 4.6 mV (p Ͼ 0.05, n ϭ 4) after pervanadate. Thus, the significant hyperpolarizing shift in activation was responsible for increased tonic channel activity (from 8 to 12% of the maximal conductance, see Fig. 7A, inset) and the Ϫ16.2 mV shift in the voltage at which the current was maximal (from Ϫ22.4 to Ϫ38.6 mV). As a result, the current at the end of long test pulses to Ϫ120 mV (holding potential, ϩ20 mV) increased by ϳ1.5-fold (data not shown). These results suggest that endogenous tyrosine phosphatase(s), such as SHP-1, counteract the

FIG. 5. SHP-1 transfections do not alter ERG current deactivation or inactivation kinetics.
A, channels were first activated (and inactivated) during a 300-ms-long pulse to ϩ80 mV (holding potential, Ϫ80 mV), and then inactivation was removed by pulses to Ϫ100, Ϫ120, Ϫ140, and Ϫ160 mV (see inset for voltage protocol). Scale bars apply to all traces. The deactivation (closing) time course was monitored by fitting the current relaxations to a mono-exponential function: I t ϭ A s exp(Ϫt/ s ), where I t is the tail current at time t; A s is the initial current amplitude, and s is the time constant of deactivation. B, to monitor inactivation, the membrane potential was first depolarized to ϩ20 mV for 1 s to fully activate (and inactivate) the channels. Then, a brief (20 ms) hyperpolarizing step to Ϫ160 mV was applied to allow rapid recovery from inactivation, followed by depolarizing steps to ϩ40, 0, Ϫ40, Ϫ80, and Ϫ120 mV (see inset for voltage protocol). Scale bars apply to all traces. The current relaxation at each voltage was fitted to a monoexponential, as above. effects on ERG current of endogenous tyrosine kinases.
In cells transfected with SHP-1 wt (Fig. 7B), pervanadate shifted the midpoint of activation (V1 ⁄2,act ) by Ϫ15 mV from Ϫ21.4 Ϯ 0.8 to Ϫ36.8 Ϯ 3.2 mV (p Ͻ 0.01, n ϭ 3). Pervanadate greatly increased the slope factor for activation, k act , i.e. it reduced the apparent voltage sensitivity from 8.0 Ϯ 0.8 to 23.7 Ϯ 2.5 mV (p Ͻ 0.005, n ϭ 3). The mid-point for inactivation (V1 ⁄2,inact ) was not significantly changed, Ϫ49.4 Ϯ 5.3 before versus Ϫ57.7 Ϯ 7.4 mV after pervanadate (p Ͼ 0.05, n ϭ 6), nor was the slope factor for inactivation changed; k inact was 29.2 Ϯ 6.6 mV before and 30.7 Ϯ 9.8 mV after pervanadate (p Ͼ 0.05, n ϭ 6). Together, these results show that pervanadate increased the voltage range for the window current (Fig. 7B, inset) mainly by changing the voltage dependence of activation. Pervanadate increased the peak of the steady-state conductance from 9 to 12% of the maximal conductance and increased by ϳ4-fold the current measured at the end of a 750-ms test pulse to Ϫ120 mV (from a holding potential of ϩ20 mV) (not shown). Pervanadate was more effective in SHP-1-transfected cells, producing a Ϫ27 mV shift in the voltage for peak channel activity, compared with a Ϫ16 mV shift in vector control cells.
ERG-1 Protein Constitutively Interacts with SHP-1 and Src in MLS-9 Cells-We first confirmed key earlier findings (1) for the MLS-9 cell batches used in the present study, i.e. that the anti-HERG antibody immunoprecipitated ERG-1 protein (Fig.  8A) and that ERG-1 constitutively interacts with Src in these cells (Fig. 8B). Because an ITIM (Leu-Thr-Tyr(P) 829 -Cys-Asp-Leu) motif is present in the cyclic nucleotide-binding domain of ERG-1, and the current was down-regulated by SHP-1, we next assessed whether SHP-1 associates with ERG-1 protein. MLS-9 lysates were immunoprecipitated with an anti-HERG antibody, and then the immunoprecipitates were run on SDS-PAGE, transferred to nitrocellulose, and probed with an anti-SHP-1 antibody. SHP-1 was co-immunoprecipitated with ERG-1 in MLS-9 cells (Fig. 8C). This is the first demonstration of interaction between SHP-1 and an ion channel. Together with our functional studies showing that pervanadate reverses the inhibitory effects of SHP-1 wt on ERG currents, these results suggest that ERG-1 is a substrate for SHP-1. We showed previously (1) that ERG-1 is also a substrate for Src, but it remains to be established whether all three proteins interact in a single complex.
Elevated Tyrosine Phosphorylation in SHP-1 Cys 3 Ser Transfectants-As shown above, SHP-1 wt or SHP-1 S6 transfection significantly reduced the ERG current density, which was unchanged in SHP-1 Cys 3 Ser transfectants (Fig. 4). Thus, we asked whether SHP-1 wt or SHP-1 Cys 3 Ser transfection modified the tyrosine phosphorylation status of the ERG-1 protein. More phosphorylation is expected if the substrate-trapping mutant, SHP-1 Cys 3 Ser, protects phosphotyrosines in the target protein from dephosphorylation by endogenous phosphatases. The relative phosphotyrosine levels FIG. 6. SHP-1 transfections alter the voltage dependence of the ERG current. A, steady-state inactivation was examined by stepping to ϩ20 mV for 1 s and then stepping to various test voltages, at which the tail currents were measured. Representative currents are shown for vector control and SHP-1 Cys 3 Ser transfectants before (upper panel) and 10 min after 3 M E-4031 was added to the bath (lower panel). B, the rectification factor (R) was calculated by first constructing the fully activated I-V relationship from data like those in A. The maximal slope conductance (G slope ) was calculated from a linear fit to the I-V relation between Ϫ140 and Ϫ80 mV (not shown), and R was calculated from R ϭ I/(G slope (V m Ϫ E K )), where V m is the test potential and E K is the K ϩ reversal potential (Ϫ6.7 mV). Values shown are mean Ϯ S.E., and the points were fitted with the Boltzmann equation (see below). C, activation versus voltage was measured as before (8,15). The membrane potential was held for 20 s (conditioning pre-pulse) at voltages between ϩ40 and Ϫ100 mV and then stepped to the test potential (Ϫ120 mV) to relieve inactivation. The peak amplitude of each inward tail current reflects channel activation during the conditioning pulse. Representative currents are shown for vector control and SHP-1 Cys 3 Ser transfectants in the absence (upper panel) or presence of 3 M E-4031 (lower panel). D, peak conductance versus voltage curves, calculated from data like those in C. Values are mean Ϯ S.E., fitted with a Boltzmann equation: G/G max ϭ 1/(1 ϩ exp((V Ϫ V1 ⁄2,act )/k act )). E, steady-state conductance versus voltage (window current) was calculated by multiplying each fitted G-V curve (from D) by its respective rectification factor (R) curve (from B).
were first compared between vector control, SHP-1 wt, and SHP-1 Cys 3 Ser transfectants by probing Western blots of MLS-9 lysates from these transfectants with anti-phosphotyrosine or anti-HERG antibody. There was a characteristic doublet at about 130 and 145 kDa with more tyrosine-phosphorylated protein in the SHP-1 Cys 3 Ser transfectants (Fig. 9A,  lanes 3 and 6). Next, tyrosine-phosphorylated proteins were immunoprecipitated with an anti-phosphotyrosine antibody, run on SDS-PAGE, and transferred to nitrocellulose, and the Western blot was probed with an anti-HERG antibody (Fig.  9B). For comparison, the levels of tyrosine-phosphorylated ERG-1 were standardized to the cyclin D 1 /D 2 levels in the lysates used for immunoprecipitation. Although cyclin D 1 /D 2 is commonly used for standardizing Western blots, its expression is linked to the cell cycle (35), likely explaining its variable expression after SHP-1 wt transfection (compare Fig. 3A with Fig. 9B). However, as another control, we found no change in levels of the signaling molecules, phosphatidylinositol 3-kinase or SHP-2 phosphatase (data not shown). When compared with vector controls, the levels of phosphorylated ERG-1 protein increased about 2-fold in SHP-1 Cys 3 Ser and decreased about 4-fold in SHP-1 wt transfectants. Thus, ERG-1 protein appears to be a substrate for SHP-1.

DISCUSSION
Physiological Relevance of SHP-1 Effects on the ERG Current-We observed previously (1) up-regulation of ERG current by Src, and this corresponded with increased tyrosine phosphorylation of the channel protein. The outcome of SHP-1 treatments cannot be easily predicted because SHP-1 can up-regulate Src kinase activity, which is expected to have the opposite effect from direct de-phosphorylation of the ERG protein. Our results are consistent with an outcome dominated by direct effects of SHP-1 on the channel protein. Namely, active forms of SHP-1 (SHP-1 wt, SHP-1 S6) are expected to reduce tyrosine phosphorylation of the channel protein and decrease the ERG current, and these effects should be antagonized by the phosphatase inhibitor, pervanadate. Our results are entirely consistent with this mechanism, i.e. SHP-1 wt transfection decreased the channel phosphorylation, active SHP-1 forms decreased the maximal current (acute application or transfection), and pervanadate restored the current. Conversely, the substrate-trapping mutant, SHP-1 Cys 3 Ser, is expected to protect the channel protein and increase its tyrosine phosphorylation level, as we observed after transfection. This should increase the current, as was observed with SHP-1 Cys 3 Ser in the pipette. The lack of effect after SHP-1 Cys 3 Ser transfection would be expected if the levels of SHP-1 Cys 3 Ser protein expressed were insufficient to counteract endogenous active SHP-1 proteins. After SHP-1 transfection there was more ERG protein, thus an increase in current was anticipated. The opposite was observed, which suggests that SHP-1-dependent channel de-phosphorylation dominates under these conditions. Although it has been reported that Src can be activated by SHP-1 (32), this would have produced the opposite outcome to that observed. Specifically, SHP-1 wt or SHP-1 S6 should have increased the channel tyrosine phosphorylation, thereby increasing the ERG current. The decrease in current in SHP-1 wt transfectants was not due to a decrease in ERG-1 or Src protein. After transient transfection with SHP-1 wt or SHP-1 Cys 3 Ser, both transfectants had increased levels of SHP-1 and ERG-1 immunoreactive proteins, and Src was not decreased. Our results strongly suggest that SHP-1 acts on the ERG-1 protein with functional consequences, and thus, ERG-1 is a novel SHP-1 substrate.
Disruption of SHP-1 tyrosine phosphatase function affects a wide range of hematopoietic cell functions, usually increasing cell proliferation, cell adhesion, and oxidative burst and decreasing programmed cell death (for reviews see Refs. 36 and 37). Moreover, the capacity of SHP-1 to down-regulate mitogenic signaling cascades is reflected in the SHP-1-deficient "motheaten" phenotype of mice, whose granulocytes and macrophages show marked proliferation in response to growth factor stimulation (19,38,39). CSF-1 is a potent mitogen for microglia, and after brain injury, CSF-1 and its receptor, c-Fms, are up-regulated in microglia (for review see Ref. 40). Traumatic injury to the central nervous system activates microglia and strongly increases expression of SHP-1, which physically associates with and is activated by c-Fms in these cells (17). Surprisingly, given the effects of SHP-1 deficiency in the peripheral immune system of "motheaten-viable" mice (38), decreased microglia proliferation was observed (17).
Our data are the first to show modulation of an ion channel by SHP-1, and the regulation we observed may have broader implications. SHP-1 is an important protein tyrosine phosphatase, which negatively regulates many cell-surface receptors, mainly for growth factors. HERG plays an important role in heart functions (2-4), and we have evidence it is involved in hematopoietic cancers (7). Thus, it is crucial to understand how HERG functions are modulated and, in particular, the role of tyrosine phosphorylation and dephosphorylation in HERG functions.
Biophysical Mechanisms Accounting for Changes in ERG Current-In principle, down-regulation of ERG currents by SHP-1 could result from any of several biophysical mechanisms as follows: (i) decreased membrane area with the same channel density, (ii) decreased single-channel conductance or number of channels, or (iii) altered voltage dependence or kinetics of the channels. The first is ruled out because there was no significant difference in membrane capacitance among the transfection groups, and the second was not investigated within the scope of the present study. We found that the third mechanism contributes significantly.
A decrease in ERG current could result from faster deactivation, slower activation, a negative shift of the voltage dependence of inactivation, or a positive shift in the voltage dependence of activation (41)(42)(43). Previously, we showed that Src-linked tyrosine phosphorylation of ERG-1 increased the current, partly by shifting the voltage dependence of activation to more negative potentials and slowing channel closing (1). Kinetic changes do not appear to account for the present results because active SHP-1 proteins in the pipette slowed deactivation, which is expected to increase, not decrease, the current, and SHP-1 transfections did not change either deactivation or inactivation rates. Although slower channel activation could reduce the current, this is unlikely because the activating pre-pulses (1 or Ͼ30 s) were much longer than needed to fully activate the channels at ϩ20 mV (15). Instead, active SHP-1 enzymes acutely added to the pipette or overexpressed through transfection altered the voltage dependence of ERG activation and decreased steady-state conductance. Active SHP-1 produced a positive shift in the conductance versus voltage curve when the channels were activated at physiologically relevant voltages (Ϫ10 mV and below), making it harder to open the channels. Interestingly, when the current was activated at positive potentials, no differences in the conductance versus voltage curve were seen; thus, a decrease in the number of active channels might also occur in this voltage range. SHP-1 transfections did not change the slope factors for activation; therefore, the voltage sensitivity of the activation gate was not apparently affected. Furthermore, the voltage dependence of inactivation was not significantly affected by SHP-1 transfections or pervanadate. Thus, changes in the steady-state conductance (calculated from the voltage dependence of activation and inactivation) can account for the reduced ERG current in SHP-1 wt and SHP-1 S6 transfectants and after acute addition of these enzymes to the pipette. The increased current in SHP-1 Cys 3 Ser transfectants cannot be explained by the same process, because the voltage dependence was not changed in the opposite direction.
Changes in ERG current activation induced by SHP-1 transfectants were reversed by the phosphatase inhibitor, pervanadate. This is consistent with a mechanism wherein reducing Ser transfectants, and hypo-phosphorylated in SHP-1 wt transfectants. MLS-9 cell lysates (100 g of total protein) were immunoprecipitated with an anti-phosphotyrosine antibody (PY20), and Western blots were probed with the anti-HERG antibody. To standardize the levels, a Western blot of the MLS-9 lysates (50 g of total protein) was probed with anti-cyclin D 1 /D 2 antibody (lower panel), and the bands were analyzed by densitometry. C, summary of relative phosphotyrosine levels of ERG-1 protein. Values are mean Ϯ S.E. for the number of independent determinations indicated on each bar. #, p Ͻ 0.05 versus vector control; **, p Ͻ 0.01 versus SHP-1 Cys 3 Ser. channel de-phosphorylation by SHP-1 makes the ERG channels more active, just as we observed previously (1) when channel phosphorylation was increased by Src. Moreover, pervanadate expanded the steady-state activity over a voltage range that is relevant to cells like microglia that lack classical action potentials.
Known structure-function relations of the HERG protein can help in interpreting the effects of SHP-1 on the current in MLS-9 cells. Activation of many voltage-gated K ϩ channels involves the S4-S5 linker and the carboxyl-terminal half of the S6 domain. It is thought that the voltage-sensing S4 domain changes conformation (44), and the S4-S5 loop couples S4 movements to the activation gate (45). In HERG, mutations in the S4-S5 loop shift the voltage dependence of channel activation to more positive potentials (43,46). HERG activation apparently differs in that the amino terminus might bind to the S4-S5 loop to stabilize the open channel (47). Consequently, deletions or mutations in the amino terminus of HERG, including the Per-Arnt-Sim domain, shift the voltage dependence of channel activation to positive potentials (34,41,(47)(48)(49). HERG inactivation is mediated by a region in or near the outer mouth of the channel (33, 41, 50 -52). The lack of effect of SHP-1 on ERG inactivation is not surprising because potential targets of SHP-1 should be on the intracellular side of the channel.
ERG-1 and SHP-1 as Part of a Signaling Complex-Although we do not know which site(s) on the ERG-1 channel are dephosphorylated by SHP-1, the site may involve components of the activation gate because SHP-1 was selective in shifting the voltage dependence of channel activation. Our results are consistent with tyrosine phosphorylation increasing the coupling between the amino terminus and the activation gate, making the channel easier to open. The lack of effect on the voltage dependence of inactivation supports earlier work showing that in HERG, activation and inactivation mechanisms are not directly linked (46,52).
We propose the following model: SHP-1 reduces the ERG current by selectively disrupting interactions between the amino terminus and the S4-S5 linker through phosphorylation of specific amino-terminal tyrosine residues. Sequence analysis of ERG-1 reveals one amino-terminal consensus site for tyrosine phosphorylation (RKVEIAFY) and three sites ( 54 YSRA, 329 YRTI, and 405 YSPF) corresponding to the YXX hydrophobic motif responsible for tyrosine phosphorylation or direct binding of SH2 domains by Src family members (53). The 54 YSRA sequence is particularly interesting because it is within the Per-Arnt-Sim domain (first 135 amino acids), which is thought to mediate protein-protein interactions in a wide range of proteins (48,49). In the future, it will be important to determine whether this site is a target for endogenous SHP-1. Co-immunoprecipitation analysis revealed physical associations of SHP-1 with ERG-1 and SHP-1 with Src proteins, and we previously demonstrated interaction between ERG-1 and Src (1). Our results are consistent with a multimolecular complex that regulates ERG channel function, but it remains to be determined whether ERG-1, Src, and SHP-1 form a multimeric complex, and where the interactions occur.