Activity-based probes for protein tyrosine phosphatases Sanjai Kumar, Bo Zhou, Fubo Liang, Wei-Qing Wang, Zhonghui Huang, and Zhong-Yin Zhang* Department of Molecular Pharmacology, Albert Einstein College of Medicine, Bronx, NY 10461 Communicated by Jack E. Dixon, University of California at San Diego School of Medicine, La Jolla, CA, April 12, 2004 (received for review January 23, 2004)
P
rotein tyrosine phosphatases (PTPs) constitute a large family of signaling enzymes (⬎100 in humans) that are important for the regulation of cell proliferation, differentiation, metabolism, migration, and survival (1, 2). Dysfunction in PTPs results in aberrant Tyr phosphorylation, which has been linked to the etiology of several human diseases, including cancer and diabetes (3, 4). Unlike protein kinases, of which Tyr- and Ser兾Thr-specific kinases share sequence identity, the PTPs show no sequence similarity with Ser兾Thr phosphatases or the broad-specificity phosphatases, such as acid or alkaline phosphatases. The hallmark that defines the PTP superfamily is the active site amino acid sequence C(X)5R, also called the PTP signature motif, in the catalytic domain. The PTPs can be broadly divided into two groups based on active site substrate specificity: the Tyr-specific and the dual-specificity phosphatases, which hydrolyze pSer兾Thr as well as pTyr. Despite variations in primary structure and differences in substrate specificity, key structural features in the active site and the mechanism of catalysis are conserved among all members of the PTP superfamily (5, 6). Although PTPs share a common catalytic mechanism, they have distinct (and often unique) biological functions in vivo. One of the major challenges in the field is to rapidly establish functional roles for PTPs, in both normal physiology and pathogenic conditions. Gene knockout analysis is useful in assessing the role of a number of PTPs in cellular signaling. However, this process is often tedious, and gene ablation in animals often results in compensatory changes through other mechanisms during embryonic development. In addition, the one-gene-at-a-time approach is clearly inadequate to deal with the dynamics and complexity in the complement of proteins within a proteome. One attractive strategy for efficient analysis of PTP function is to characterize these enzymes collectively, rather than individually. In this regard, DNA microarray methods provide significant insights on changes in the abundance of transcripts. However, the measured mRNA levels do not always www.pnas.org兾cgi兾doi兾10.1073兾pnas.0402323101
Fig. 1.
Structures of BBP and PTP activity probes I and II.
correlate with protein expression. Proteomic approaches address some of the gaps in genomic technologies by profiling and measuring bulk changes in protein levels. Unfortunately, current methodologies are merely adequate for abundant proteins. Furthermore, the amount of protein is not always proportional to biological activity, which may be subject to posttranslational regulation. Thus, standard proteomics techniques are not optimal for tracking variations in protein activity. Because the function of a PTP depends on its phosphatase activity, the development of technologies for directly measuring the dynamics in PTP activity on a global scale is of tremendous interest. Recently, a chemical approach has emerged that allows the consolidated detection and identification of collections of enzyme activities in complex proteomes (7, 8). This approach employs specific chemical probes that are directed to an enzyme active site for covalent modification in an activity-dependent fashion. Activitybased probes have been used for proteomic analysis of the Cys and Ser hydrolases, providing insights on these two families of proteases in cell biology and in diseases (9, 10). However, little success has been achieved in targeting specifically to the PTP family with activity-based probes. Here, we describe the chemical synthesis and biochemical characterization of two activity-based probes that enable the interrogation of the state of PTP activity in samples of high complexity and in the whole proteome. It is anticipated that effective application of the activity-based PTP probes will accelerate the functional characterization of PTPs, thereby facilitating our understanding of the roles of PTPs in health and diseases. Experimental Procedures Synthesis of ␣-Bromobenzylphosphonate (BBP)-Based Probes I and II.
The multistep chemical synthesis of probes I and II (Fig. 1) is provided in Supporting Text, which is published as supporting information on the PNAS web site. Detailed information on PTPs and non-PTP proteins also can be found in Supporting Text.
Kinetic Characterization of PTP Inactivation by the BBP-Based Probes.
PTP inactivation by the BBP-based probes was studied at 25°C in a pH 6.0 buffer containing 50 mM sodium succinate, 1 mM EDTA, Abbreviations: PTP, protein tyrosine phosphatase; BBP, ␣-bromobenzylphosphonate; pNPP, p-nitrophenyl phosphate; HRP, horseradish peroxidase; rt, room temperature. *To whom correspondence should be addressed. E-mail:
[email protected]. © 2004 by The National Academy of Sciences of the USA
PNAS 兩 May 25, 2004 兩 vol. 101 兩 no. 21 兩 7943–7948
BIOCHEMISTRY
Protein tyrosine phosphatases (PTPs) are involved in the regulation of many aspects of cellular activity including proliferation, differentiation, metabolism, migration, and survival. Given the large number and complexity of PTPs in cell signaling, new strategies are needed for the integrated analysis of PTPs in the whole proteome. Unfortunately, the activities of many PTPs are tightly regulated by posttranslational mechanisms, limiting the utility of standard genomics and proteomics methods for functional characterization of these enzymes. To facilitate the global analysis of PTPs, we designed and synthesized two activity-based probes that consist of ␣-bromobenzylphosphonate as a PTP-specific trapping device and a linker that connects the trapping device with a biotin tag for visualization and purification. We showed that these probes are active site-directed irreversible inactivators of PTPs and form a covalent adduct with PTPs involving the active site Cys residue. Additionally, we demonstrated that the probes are extremely specific toward PTPs while remaining inert to other proteins, including the whole proteome from Escherichia coli. Consequently, these activity-based PTP probes can be used to profile PTP activity in complex proteomes. The ability to interrogate the entire PTP family on the basis of changes in their activity should greatly accelerate both the assignment of PTP function and the identification of potential therapeutic targets.
and 1 mM DTT, adjusted to the ionic strength of 150 mM with NaCl. The inactivation reaction was initiated by the addition of a 5-l aliquot of PTP stock to a 45-l solution containing appropriately diluted probe I (final concentration of DMSO, 5%). At appropriate time intervals, aliquots of 2 l were removed from the reaction and added to a 200-l solution containing 20 mM pnitrophenyl phosphate (pNPP) in pH 6.0 buffer at 30°C (11). The kinetic parameters of the inactivation reaction were obtained by fitting the data to the following equations:
冉
冊
A0 ⫺ A⬁ ⫺kobs䡠t A t A⬁ ⫽ ⫺ e A0 A0 A0
[1]
and ki ⫻ 关I兴 . k obs ⫽ KI ⫹ 关I兴
[2]
Covalent Labeling with the PTP Probes. In a typical labeling exper-
iment, the reaction was initiated by adding the PTP probe (final concentration, 1 mM) to the preincubated enzyme (final concentration, 25 M) in the appropriate buffer at 25°C for 1 h, unless otherwise stated. All labeling reactions involving PTPs were carried out in 50 mM sodium succinate buffer (pH 6.0) that contained 1 mM EDTA and 1 mM DTT, adjusted to the ionic strength of 150 mM with NaCl. Labeling reactions involving non-PTP enzymes were carried out in their respective optimum activity conditions, and, thus, the following buffers, adjusted to the ionic strength of 150 mM, were used: alkaline phosphatase, 50 mM Tris, pH 9.0兾1 mM MgCl2; potato and prostatic acid phosphatases, 100 mM sodium acetate, pH 5.0兾1 mM EDTA; protein phosphatase 1 (PP1) and phosphatase, 50 mM 3,3-dimethylglutarate, pH 7.0兾2 mM MnCl2; papain, lysozyme, Grb2-Src homology 2 (SH2), Shc-phosphotyrosine-binding (PTB), and SNT1-PTB, 50 mM 3,3-dimethylglutarate, pH 7.0兾1 mM EDTA; calpain, 100 mM imidazole, pH 7.3兾10 mM CaCl2兾1 mM DTT; glyceraldehyde-3-phosphate dehydrogenase, 15 mM sodium pyrophosphate, pH 8.5兾7.5 mM NAD兾1 mM DTT; chymotrypsin, 50 mM Tris, pH 7.8兾50 mM CaCl2; trypsin, 100 mM Tris, pH 8.5兾1 mM EDTA; BSA, 50 mM sodium succinate buffer, pH 6.0兾1 mM EDTA兾1 mM DTT; protein phosphatase 2B, 50 mM 3,3-dimethylglutarate, pH 7.0兾0.3 M calmodulin兾2 mM MnCl2; Src kinase, 100 mM Tris, pH 7.0兾1 mM EDTA; thermolysin, 50 mM Hepes, pH 7.0兾5 mM CaCl2; and GST, 50 mM 3,3-dimethylglutarate, pH 7.0兾1 mM EDTA兾1 mM DTT. Western Blot Analyses of the Labeling Reaction. The labeling reactions were quenched by the addition of 1 volume of 2⫻ SDS loading buffer (reducing) at 75°C for 5 min. Each sample was divided into halves, which were separated on a SDS兾12.5% PAGE gel (3 g of protein per lane). The proteins from one gel were transferred overnight to a nitrocellulose membrane at 4°C. The membrane was then blocked with 5% nonfat dry milk in Tris-buffered saline (TBS) with 1% Tween 20 (TBS-T) for 1 h at 25°C. After two quick washes with TBS-T, the membrane blot was treated for 2 h with antibiotin-horseradish peroxidase (HRP) conjugate (Cell Signaling Technology, Beverly, MA; 1:1,000 dilution) in TBS-T containing 5% nonfat dry milk at 25°C. The anti-biotin-HRP-treated blot was then washed three times (for 30 min each time) with TBS-T and subsequently treated with HRP substrate for 1 min before exposing the chemiluminescent membrane blot to the film. The second gel was Coomassie blue-stained for protein visualization. In the labeling experiment involving YopH and its mutants (Fig. 6), the blot itself was stripped with stripping buffer (100 mM 2-mercaptoethanol and 2% SDS in 62.5 mM Tris䡠HCl, pH 6.7) for 30 min at 50°C to remove the anti-biotin-HRP and HRP substrate. The blot was washed quickly a few times with water and exposed to Coomassie blue for protein staining. 7944 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0402323101
Matrix-Assisted Laser Desorption Ionization Time-of-Flight (MALDITOF) MS. To prepare the modified YopH for the MALDI-TOF
experiment, 1 M YopH was mixed with 10 mM PTP probe I. YopH alone with pH 6.0 succinate buffer was used as a control. After a 3-h incubation at 25°C, a 10-l sample was desalted by a ZipTip C4 column (Millipore) and eluted in 4 l of 50% acetonitrile兾0.1% trifluoroacetic acid. After mixing 1 l of the elute with 1 l of saturated sinapinic acid solution (50% acetonitrile兾0.1% trifluoroacetic acid), a 1-l aliquot was transferred to a 100-well MALDI sample plate (Applied Biosystems). Mass spectra were recorded on a Voyager-DE STR biospectrometry workstation (Applied Biosystems) in the positive polarity mode. All spectra are summations of 100 individual laser shots. Results and Discussion Given the large number and complexity of PTPs in cell signaling, new strategies are needed for the integrated analysis of PTPs in the whole proteome. The ability to profile the entire PTP family on the basis of changes in the activity of its members should greatly accelerate both the assignment of PTP function and the identification of potential therapeutic targets. However, until now, classselective probes for PTPs were not available. The suicide substrate, 4-fluromethylaryl phosphate, was recently explored as a PTP probe (12), because hydrolysis of 4-fluoromethylaryl phosphate generates a highly reactive quinone methide intermediate, which can alkylate nucleophilic side chains at, or near, the phosphatase active site (13). Unfortunately, 4-fluromethylaryl phosphate is not specific for PTPs; it also forms a covalent adduct with other classes of phosphatases such as nonspecific prostatic acid phosphatase and protein Ser兾Thr phosphatase calcineurin (13–15). An additional drawback to this type of chemistry is that the diffusible, unmasked quinone methide electrophile could alkylate other proteins in the vicinity that carry nucleophilic residues on their surface. Consequently, chemical probes based on 4-fluromethylaryl phosphate lack the specificity required for global analysis of the PTP superfamily. Below, we describe the development of two specific activity-based PTP probes based on BBP (Fig. 1), a quiescent affinity inactivator of the Yersinia PTP YopH (16). Design and Synthesis of Biotinylated BBPs as Activity-Based PTP Probes. Ideally, an activity-based probe for the PTPs would consist
of a PTP-specific trapping device for covalent attachment to the enzyme active site and a linker that connects the trapping device with a reporter兾affinity tag for visualization and purification. For this reason, our first-generation PTP probes I and II (Fig. 1) consisted of the potential PTP-specific affinity agent BBP, a linker, and a biotin tag. Probe II differs from I in that an additional hexanoic acid spacer was inserted between BBP and biotin. This design was intended to evaluate whether the linker space between BBP and the biotin tag in probe I is sufficient to prevent steric hindrance that could block accessibility of BBP for PTP labeling or of the tag for detection and purification. One crucial aspect of the design involved separating the amine functionality at the 4 position of the phenyl ring by a methylene linker for subsequent condensation of (⫹)biotinamido-N-hydroxysuccinimide ester or (⫹)biotinamidohexanoic acid N-hydroxysuccinimide ester, to avoid the undesirable in situ elimination of the bromide leaving group during the synthesis. The chemical synthesis of PTP probes I and II involved synthesizing a common intermediate, compound 10, in a seven-step procedure (Scheme 1) and its subsequent coupling with the (⫹)biotinamido-N-hydroxysuccinimide ester analogues, followed by the deprotection of the phosphonate diethyl ester (Scheme 2). Thus, 4-cyanobenzyl bromide 3 was hydrolyzed in the presence of barium carbonate to produce 4-cyanobenzyl alcohol 4, which was subsequently reduced with lithium aluminum hydride under strict argon atmosphere to yield 4-hydroxymethyl-benzyl-ammonium chloride 5, upon acidic aqueous extraction (17). The amine functionality of Kumar et al.
Scheme 1. Synthetic routes applied to the syntheses of common intermediate 10 are shown. (i) BaCO3兾H2O; reflux, 2–2.5 h. (ii) LiAlH4兾ether, at room temperature (rt), 45 min; reflux, 3 h. (iii) Di-tert-butyl dicarbonate, KOH兾THF兾 H2O at rt, 1.25 h. (iv) PCC兾CH3COONa, CH2Cl2 at rt in the dark, 17 h. (v) HPO(OEt)2兾triethylamine, benzene; reflux, 50 h. (vi) PPh3Br2, CH3CN兾pyridine at 0°C and rt, 1.5 and 2 h, respectively. (vii) Trifluoroacetic acid neat, 15–20 min.
diesters 11 and 12 were deprotected with bromotrimethylsilane in DMF to yield probes I and II (21). Probes I and II Are Active Site-Directed and Irreversible Inactivators of PTPs. To characterize probes I and II as activity-based PTP
probes, we first examined for their effect on PTP activity using pNPP as a substrate. As expected, probe I inactivated the Yersinia PTP YopH in a time- and concentration-dependent first-order process (data not shown). Similar results were obtained with several PTPs, including PTP1B, HePTP, SHP2, LAR, PTP␣, PTPH1, VHR, and Cdc14. In addition, probes I and II were equally effective at inhibiting PTP activity, indicating that linker size does not affect probe reactivity. Thus, unless indicated otherwise, all subsequent experiments were performed with probe I because of its better solubility in aqueous solution. Inactivation of PTPs with probe I appeared to be irreversible, because extensive dialysis and兾or buffer exchange of the reaction mixture failed to recover enzyme activity. Analysis of the pseudo-first-order rate constant as a function of reagent concentration showed that probe I-mediated YopH inactivation displayed saturation kinetics (Fig. 2), yielding values for the equilibrium binding constant KI and the inactivation rate constant ki of 0.74 mM and 0.17 min⫺1, respectively. Similar saturation
BIOCHEMISTRY
4-hydroxymethyl-benzyl-ammonium chloride 5 was then first masked by tert-butyl carbamate to render (4-hydroxymethylbenzyl)carbamic acid tert-butyl ester 6 (18). Corey oxidation (19) with pyridinium chlorochromate兾sodium acetate gave the reactive aldehyde 7. The condensation of aldehyde 7 with diethyl phosphite in the presence of triethylamine produced {[4-(tert-butoxycarbonylamino-methyl)-phenyl]-hydroxymethyl}-phosphonic acid diethyl ester 8 as a racemic mixture. This racemic mixture of 8 was used as such in the following step of the synthesis. Bromination of the secondary alcohol, 8, to yield compound 9 was achieved by using dibromotriphenyl phosphorane in pyridine (20). Finally, deprotection of tert-butyloxycarbonyl with neat trifluoroacetic acid gave [(4-aminomethylphenyl)-bromomethyl]-phosphonic acid diethyl ester 10. The coupling of 10 with (⫹)biotin N-hydroxysuccinimide ester (for probe I) and (⫹)biotinamidohexanoic acid Nhydroxysuccinimide ester (for probe II) in N,N-dimethylformamide (DMF) gave compounds 11 and 12, respectively. The phosphonate
Fig. 2. Concentration dependence of the pseudo-first-order rate constant kobs for probe I-mediated YopH inactivation. The curve was generated by fitting the data to Eq. 2.
Scheme 2. Synthesis of PTP probes I and II from compound 10. (viii) (⫹)Biotinamido-N-hydroxysuccinimide ester兾N,N-dimethylformamide (DMF) at rt, 70 h. (ix and xi) TMSBr兾DMF at rt, 18 h. (x) (⫹)Biotinamidohexanoic acid N-hydroxysuccinimide ester兾DMF at rt, 70 h.
Kumar et al.
Fig. 3. Specific labeling of PTPs by probe I. PTPs (3 g) were mixed with 1 mM of probe I in pH 6.0 buffer containing 50 mM succinate and 1 mM EDTA and with an ionic strength of 0.15 M at 25°C for 1 h. The reaction mixtures were separated by SDS兾12.5% PAGE and subjected to Western blot analysis by using anti-biotin-HRP conjugate. Bound anti-biotin antibodies were visualized by chemiluminescence using an ECL kit (Amersham Pharmacia). (Upper) ECLdeveloped gels. (Lower) Coomassie blue-stained gels. LMW, Low molecular weight. PNAS 兩 May 25, 2004 兩 vol. 101 兩 no. 21 兩 7945
zylphosphonate, a nonhydrolyzable pTyr mimetic (11). Further evidence in support of the inactivation’s being directed to the active site included the fact that arsenate, a competitive PTP inhibitor, was able to protect PTP from BBP-mediated inactivation. Probe I was quite stable in pH 6.0 buffers, although significant solvolysis could be detected at pH values ⬎7.0. To minimize any loss of the probe due to hydrolysis, subsequent labeling experiments were conducted at pH 6.0. It is, however, worth mentioning that the probe displayed similar reactivity toward the PTPs at both pH 6.0 and 7.0. Interestingly, prolonged incubation (48 h) of probe I with 100 mM DTT, 50 mM azide, or 50 mM Cys in either pH 6.0 or 7.0 buffers did not produce the expected substitution products as evidenced by NMR and MS. Moreover, these scavenging nucleophiles did not affect the rate of probe I-mediated PTP inactivation. These results demonstrate that the BBP-based probes are inert to nucleophilic agents, and their propensity for rapid inactivation of PTPs suggests that a latent affinity group unleashes its reactivity only toward a suitably disposed nucleophilic residue in the active site.
Fig. 4. Matrix-assisted laser desorption ionization time-of-flight (MALDITOF) mass spectra of YopH (A) and YopH labeled with probe I (B). The high-intensity peaks show the mass of the singly charged proteins, and the low-intensity peaks are the doubly charged proteins.
kinetics was also observed for YopH inactivation by BBP (KI ⫽ 4.1 mM and ki ⫽ 0.11 min⫺1). These results suggest that the BBP-based probes are active site-directed affinity agents whose mode of action likely involves at least two steps: binding to the PTP active site followed by covalent modification of active site residue(s). The ability of BBP-based probes to target the PTP active site was expected because BBP mimics pTyr, which is known to occupy the PTP active site. Indeed, the KI values for probe I and BBP are similar to competitive inhibition constants measured for ben-
Fig. 5.
Probes I and II Form a Covalent Adduct with PTPs. To demonstrate that probes I and II inactivate PTPs by forming a covalent adduct with the enzymes, we incubated the PTP with either biotin or the probe at a 1 mM concentration and pH 6.0 for 1 h. The sample was divided into halves and separated by SDS兾PAGE. As shown in Fig. 3, covalent labeling of the PTP by the biotin-tagged probe was visualized with anti-biotin antibody-conjugated peroxidase chemiluminescence, whereas the amount of loaded protein was determined by Coomassie blue staining. As expected, biotin alone did not label the PTPs, whereas the biotin-tagged probe I was covalently incorporated into all PTPs tested, including cytosolic PTPs YopH, PTP1B, HePTP, SHP2, and FAP-1, receptor-like PTPs PTP␣ and DEP-1, dual-specificity phosphatases VHR, Cdc14, and PRL-3, and the low-molecular-weight PTP. Thus, probe I can react with a broad range of enzymes from the PTP superfamily. Furthermore, the extent of labeling correlated with PTP activity. Thus, it is possible that the probe can be used to analyze the activation status of PTPs under different physiological conditions. Again, similar results were obtained with probe II, indicating that the biotin tag has similar accessibility to the anti-biotin antibodies. To provide more direct evidence that the PTP probes can covalently label PTPs, we also analyzed the probe I-treated YopH using matrix-assisted laser desorption ionization (MALDI) MS (Fig. 4). The measured (M ⫹ H)⫹ value of 33,515 for YopH agreed with the theoretical value. The measured (M ⫹ H)⫹ value of 33,941 for modified YopH indicated a mass shift of 426, which corre-
Potential mechanisms for PTP inactivation by the BBP-based probes are shown.
7946 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0402323101
Kumar et al.
Fig. 6.
Covalent labeling of wild-type and mutant YopH with PTP probe I.
sponded exactly to the expected mass shift resulting from covalent attachment of probe I (mass, 506) to YopH concomitant with a loss of bromide (mass, 80) (Fig. 5). The absence of a peak at m兾z 33,515 indicated the high efficiency of the modification reaction (Fig. 4B).
Probe I Is Highly Specific for PTPs. Because probe I was intended for
PTP activity profiling at the whole-proteome level, we wanted to determine whether the probe exhibits cross-reactivity with other protein classes. To this end, we examined the reactivity of probe I toward a panel of non-PTP enzymes兾proteins, including alkaline phosphatase, potato and prostatic acid phosphatases, Ser兾Thr protein phosphatases PP1, PP2B, and phosphatase, the SH2 domain of Grb2, the PTB domains of Shc and SNT1, the Src kinase, Ser proteases trypsin and chymotrypsin, metalloprotease thermolysin, Cys proteases papain and calpain, glyceraldehyde-3phosphate dehydrogenase and GST (both of which possess reactive low pKa Cys), BSA, and lysozyme. As shown in Fig. 7A, none of the proteins tested were reactive to probe I. The lack of reactivity of
BIOCHEMISTRY
Mechanism of PTP Inactivation by Probe I. The PTPs share a conserved active site that recognizes aryl phosphates and a common catalytic mechanism that use a highly reactive nucleophilic Cys residue (6). This Cys (Cys-403 in YopH) displays an unusually low pKa of ⬇5 (22) and is situated at the bottom of the pTyr-binding pocket such that its S␥ atom is poised 3 Å from the phosphorus atom of pTyr (11). In the catalytic mechanism, the active site Cys initiates a nucleophilic attack on the phosphorus, leading to the formation of a thiophosphoryl enzyme intermediate. This process is assisted by the general acid (Asp-356 in YopH), which is in close proximity to the scissile phenolic oxygen of pTyr. The active site Arg (Arg-409 in YopH) is involved in initial binding of the pTyr substrate and stabilizes the transition state. If the BBP moiety binds the PTP active site in a manner similar to pTyr, it could covalently modify the PTP by means of two potential mechanisms (Fig. 5). In one mechanism, the active site Cys initiates nucleophilic attack, as in normal substrate reaction, directly on the phosphonate group to give a transient phosphoranelike intermediate (species b in Fig. 5), followed by closure to a three-membered ring species c with the expulsion of the bromide. Subsequent ring opening produces ␣-hydroxylbenzylphosphonate covalently attached to the active site Cys (species e in Fig. 5). Alternatively, the strategically positioned Asp general acid could
displace the bromide directly from the benzylic position to yield species f (Fig. 5). Both mechanisms will result in a covalent adduct with the net loss of a bromide. To distinguish between these possibilities, we carried out labeling experiments with several site-directed mutants of YopH. The C403S mutant is catalytically inactive but retains substrate-binding ability (23). As seen in Fig. 6, C403S did not show any reactivity toward probe I. In addition, pretreatment of YopH with iodoacetate, which is known to specifically alkylate Cys-403 (22), also blocked the incorporation of probe I into YopH. In contrast, incubation of the general acid-deficient mutants D356A and D356N with probe I led to a complete loss of phosphatase activity (data not shown) with an extent of labeling comparative with that of the wild-type YopH. Together, the results exclude the involvement of Asp-356 in displacing the bromide in a direct nucleophilic fashion and strongly suggest that the thiol group of the active site Cys-403 is the site of covalent attachment for the BBP-based probes. Consistent with the importance of the active site Arg in binding of the phosphoryl group in the substrate, dramatic decrease in covalent labeling was observed for R409A and R409K. Moreover, more probe I was incorporated into R409K than R409A in accord with R409K’s higher affinity for substrate than R409A’s (24). Collectively, the results are consistent with a mechanism by which the activity-based probe is targeted to the PTP active site for covalent adduct formation that involves the nucleophilic Cys. Although our results do not formally exclude the possibility that the active site Cys may directly displace the bromide at the benzylic position, it is highly unlikely, because in the crystal structures of PTP bound to pTyr or its nonhydrolyzable analogues, the benzylic position is out of the reach of the Cys nucleophile.
Fig. 7. Specificity of PTP probe I. (A) Activity of probe I toward non-PTP enzymes兾proteins. (B) Activity of probe I toward the total cell lysates from E. coli. The samples were treated with 1 mM probe I for 1 h. The reaction mixtures were separated by SDS兾12.5% PAGE and subjected to Western blot analysis by using anti-biotin-HRP conjugate. Ptase, phosphatase; G-3P dehydrogenase, glyceraldehyde-3-phosphate dehydrogenase.
Kumar et al.
PNAS 兩 May 25, 2004 兩 vol. 101 兩 no. 21 兩 7947
probe I toward nonphosphatases is easy to understand because these enzymes do not recognize aryl phosphates as substrates. Although SH2 and PTB domains bind pTyr, they lack reactive groups in the binding site and are inert to probe I. What is remarkable is that although both alkaline and acid phosphatases hydrolyze aryl phosphates and possess an active site nucleophile (Ser and His, respectively), neither could be labeled by the probe. The source of this specificity may relate to the highly reactive active site Cys in PTPs and the special affinity of the nucleophilic sulfur for phosphorus. To further evaluate the specificity of probe I for PTPs, we also incubated probe I with the total cell lysates from Escherichia coli transformed with or without YopH. Western blot analysis of the reaction mixture indicated that, with the exception of YopH, no E. coli proteins were labeled by the probe (there are no PTPs in E. coli) (Fig. 7B). This level of selectivity indicates that probe I is suitable for global analysis of PTP activity in a cell. As initial proof of principle, we tested the hypotheses that the activity of PTPs can be significantly altered in cancer cells and that the activity-based probes are capable of profiling PTP activity at the whole-proteome level. We performed a preliminary experiment with a hTERT immortalized human mammary epithelial cell line, and a human breast cancer cell line, MCF-7. As shown in Fig. 8, there are at least four PTPs (indicated by arrowheads) displaying significantly different activities between these two cell lines. Two PTPs (230 and 100 kDa) showed decreased activities in MCF-7, compared with hTERT, cells. In contrast, the activities of two PTPs migrating at 60 and 75 kDa were dramatically higher in MCF-7 cells than those in hTERT cells. In fact, the 60-kDa band may represent a PTP whose expression is induced in MCF-7 cells. Experiments are needed to further optimize the reaction conditions with mammalian cell lysates and to identify the PTPs with altered activity in MCF-7 cells. This might be accomplished by affinity capture of the biotin-tagged PTPs with avidin columns followed by liquid chromatography (LC)兾MS兾MS analysis. We were able to identify PTP1B from the protein pool eluted with biotin from the avidin column by using PTP1B antibody, and we estimated that under our conditions the probe could detect PTP1B in the subnanogram range, quantities sufficient for MS experiments. We note that there are a couple of nonspecific bands in the untreated cell lysates (Fig. 8), which may be caused by endogenous biotinylated proteins. If their presence represents a serious problem for PTP identification, the cell lysate can be precleared with anti-biotin antibodies before treatment with the PTP probe. Identification of PTPs whose activities are abnormally regulated in cancer cells may establish links between members of the PTP family and cancer. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13.
Hunter, T. (2000) Cell 100, 113–127. Tonks, N. K. & Neel, B. G. (2001) Curr. Opin. Cell Biol. 13, 182–195. Li, L. & Dixon, J. E. (2000) Semin. Immunol. 12, 75–84. Zhang, Z.-Y. (2001) Curr. Opin. Chem. Biol. 5, 416–423. Barford, D., Das, A. K. & Egloff, M. P. (1998) Annu. Rev. Biophys. Biomol. Struct. 27, 133–164. Zhang, Z.-Y. (2003) Prog. Nucleic Acid Res. Mol. Biol. 73, 171–220. Cravatt, B. F. & Sorensen, E. J. (2000) Curr. Opin. Chem. Biol. 4, 663–668. Jeffer y, D. A. & Bog yo, M. (2003) Cur r. Opin. Biotechnol. 14, 87–95. Liu, Y., Patricelli, M. P. & Cravatt, B. F. (1999) Proc. Natl. Acad. Sci. USA 96, 14694–14699. Greenbaum, D., Medzihradszky, K. F., Burlingame, A. & Bogyo, M. (2000) Chem. Biol. 7, 569–581. Sun, J.-P., Wu, L., Fedorov, A. A., Almo, S. C. & Zhang, Z.-Y. (2003) J. Biol. Chem. 278, 33392–33399. Lo, L.-C., Pang, T.-L., Kuo, C.-H., Chiang, Y.-L., Wang, H.-Y. & Lin, J.-J. (2002) J. Proteome Res. 1, 35–40. Myers, J. K. & Widlanski, T. S. (1993) Science 262, 1451–1453.
7948 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0402323101
Fig. 8. Global analysis of PTP activity with probe I in normal and cancer breast cells. MCF-7 or hTERT immortalized human mammary epithelial cells were lysed in 50 mM Mes (pH 6.0), 150 mM NaCl, 1% Triton X-100, 10% glycerol, 5 mM EDTA, 1 g兾ml aprotinin, 1 g兾ml leupeptin, and 1 mM PMSF and incubated with 1 mM probe I for 1 h at rt. Fifty micrograms of protein was loaded and separated by SDS兾10% PAGE. Biotinylated proteins were detected by immunoblotting using anti-biotin antibody.
In summary, we described the design and synthesis of two activity-based PTP probes that consist of a PTP-specific trapping device, BBP, for covalent attachment to the PTPs and a linker that connects the trapping device with a biotin tag for visualization and purification. We showed that the probes inactivate a broad range of PTPs in a time- and concentration-dependent fashion. We established that this inactivation is active site-directed and irreversible. We provided evidence that these probes form a covalent adduct with PTPs, most likely involving the active site Cys residue. More importantly, we demonstrated that the probes exhibit extremely high specificity toward PTPs while remaining inert to other enzymes兾proteins, including the whole proteome from E. coli. These properties indicate that the activity-based PTP probes can be used to profile PTP activities in both normal and pathological conditions, enabling direct isolation and identification of PTP activity in distinct physiological states. This work was supported in part by the G. Harold and Leila Y. Mathers Charitable Foundation and National Institutes of Health Grant 1U54 AI057158. 14. Wang, Q., Dechert, U., Jirik, F. & Withers, S. G. (1994) Biochem. Biophys. Res. Commun. 200, 577–583. 15. Born, T. L., Myers, J. K., Widlanski, T. S. & Rusnak, F. (1995) J. Biol. Chem. 270, 25651–25655. 16. Taylor, W. P., Zhang, Z. Y. & Widlanski, T. S. (1996) Bioorg. Med. Chem. 4, 1515–1520. 17. Gavin, J. A., Garcia, M. E., Benesi, A. J. & Mallouk, T. E. (1998) J. Org. Chem. 63, 7663–7669. 18. Far, A. R., Young, L. C., Rang, A., Rudkevich, D. M. & Rebek, J., Jr. (2002) Tetrahedron 58, 741–755. 19. Corey, E. J. & Suggs, W. (1975) Tetrahedron Lett. 16, 2647–2650. 20. Gajda, T. (1990) Phosphorus Sulfur and Silicon 53, 327–331. 21. McKenna, C. E., Higa, M. T., Cheung, N. H. & McKenna, M.-C. (1977) Tetrahedron Lett. 2, 155–158. 22. Zhang, Z.-Y. & Dixon, J. E. (1993) Biochemistry 32, 9340–9345. 23. Zhang, Y.-L., Yao, Z.-J., Sarmiento, M., Wu, L., Burke, T. R., Jr., & Zhang, Z.-Y. (2000) J. Biol. Chem. 275, 34205–34212. 24. Zhang, Z.-Y., Wang, Y., Wu, L., Fauman, E., Stuckey, J. A., Schubert, H. L., Saper, M. A. & Dixon, J. E. (1994) Biochemistry 33, 15266–15270.
Kumar et al.
