Mechanism and Kinetics of Metalloenzyme Phosphomannose

natural substrates and two inhibitors, erythrose 4-phos- phate and mannitol 1-phosphate, were investigated. Un- der nondenaturing conditions, the inta...
0 downloads 0 Views 157KB Size
Download PDF
Anal. Chem. 2005, 77, 5596-5603

Mechanism and Kinetics of Metalloenzyme Phosphomannose Isomerase: Measurement of Dissociation Constants and Effect of Zinc Binding Using ESI-FTICR Mass Spectrometry Hong Gao,†,‡ Yonghao Yu,†,‡ and Julie A. Leary*,†

Department of Chemistry and Molecular Cell Biology, Genome Center, University of California, Davis, California 95616, and Department of Chemistry, University of California, Berkeley, California 94720

Electrospray ionization Fourier transform ion cyclotron resonance (ESI-FTICR) mass spectrometry was used to study the noncovalent complexation of a metalloenzyme, phosphomannose isomerase (PMI), which catalyzes the interconversion of mannose 6-phosphate and fructose 6-phosphate. The zinc cofactor binding effect and the noncovalent interactions of the holoenzyme with its two natural substrates and two inhibitors, erythrose 4-phosphate and mannitol 1-phosphate, were investigated. Under nondenaturing conditions, the intact zinc-containing monomeric protein ions were reproducibly observed with no dissociation. Molecular ions corresponding to apo-PMI monomer were obtained by dialyzing the holoenzyme against EDTA. The binding/release of the metal ion did not alter the charge-state distributions of the protein to any significant extent, but changed the binding affinity of the substrates by at least 5-fold. Using ESI-FTICR mass spectrometry, the binding stoichiometry and specificity of the enzyme-substrate and enzyme-inhibitor complexes were directly determined. The first time report of the apparent dissociation constant for the isomeric substrates of PMI was measured to be 88.8 µM. The relative dissociation constant of the two inhibitors derived from gas-phase noncovalent complexation was very similar to the relative inhibition constant derived from solution phase kinetics. Phosphomannose isomerase (PMI) is a monomeric enzyme, of molecular mass ∼40 kDa, that catalyzes the reversible isomerization of mannose 6-phosphate (Man6P) and frutose 6-phosphate (Fru6P). In all eukaryotes and prokaryotes, this reaction is the initial committed step for the supply of guanosine diphosphate D-mannose (GDP-D-mannose), the activated mannose donor for the biosynthesis of many mannosylated structures, including glycoproteins, glycolipids, exopolysaccharides, and cell wall components in microorganisms.1-5 In the formation of GDP-D* Corresponding author. E-mail: [email protected]. Tel: 530-754-4987. Fax: 530-754-9658. † University of California, Davis. ‡ University of California, Berkeley. (1) Smith, D. J.; Proudfoot, A.; Friedli, L.; Klig, L. S.; Paravicini, G.; Payton, M. A. Mol. Cell. Biol. 1992, 12, 2924-2930.

5596 Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

Scheme 1. Biosynthetic Pathway for GDP-D-Mannose

mannose, Fru6P is first isomerized to Man6P by PMI, which is further converted to mannose 1-phosphate (Man1P) by phosphomannomutase. GDP-mannose pyrophosphorylase then catalyzes the incorporation of GTP on Man1P to produce the final product GDP-D-mannose (Scheme 1). Since the activity of PMI is essential for survival and pathogenesis of several microbes, it is considered a likely target for drugs against various microorganisms, including fungi, which can lead to serious illness and death in immunosuppressed individuals.5,6 For instance, studies have shown that over 25% of HIV-infected patients have severe mycosal infections.7,8 PMI is also required for the development of a lung infection pathogen Pseudomonas aeruginosa in cystic fibrosis patients.5 PMI activity is important in mammalian systems in terms of glycolysis and glycosylation.9,10 The level of PMI activity in leukocytes of chronic lymphatic leukemia patients is lower than normal, which makes mannose toxic, as the cells are not able to metabolize the Man6P they produce.11,12 Recent research indicates that a deficiency of PMI leads to a human disease, congenital (2) Miles, J. S.; Guest, J. R. Gene 1984, 32, 41-48. (3) Orlean, P. Mol. Cell. Biol. 1990, 10, 5796-5805. (4) Payton, M. A.; Rheinnecker, M.; Klig, L. S.; Detiani, M.; Bowden, E. J. Bacteriol. 1991, 173, 2006-2010. (5) Roux, C.; Lee, J. H.; Jeffery, C. J.; Salmon, L. Biochemistry 2004, 43, 29262934. (6) Bhandari, A.; Jones, D. G.; Schullek, J. R.; Vo, K.; Schunk, C. A.; Tamanaha, L. L.; Chen, D.; Yuan, Z. Y.; Needels, M. C.; Gallop, M. A. Bioorg. Med. Chem. Lett. 1998, 8, 2303-2308. (7) Manfredi, R.; Vezzadini, P.; Costigliola, P.; Ricchi, E.; Fanti, M. P.; Chiodo, F. Aids 1993, 7, 223-226. (8) Manfredi, R.; Vezzadini, P.; Fanti, M. P.; Chiodo, F. Sand. J. Infect. Dis. 1994, 26, 55-57. (9) Cleasby, A.; Wonacott, A.; Skarzynski, T.; Hubbard, R. E.; Davies, G. J.; Proudfoot, A. E. I.; Bernard, A. R.; Payton, M. A.; Wells, T. N. C. Nat. Struct. Biol. 1996, 3, 470-479. (10) Collins, L. V.; Attridge, S.; Hackett, J. Infect. Immun. 1991, 59, 1079-1085. (11) Delafuente, M.; Penas, P. F.; Sols, A. Biochem. Biophys. Res. Commun. 1986, 140, 51-55. (12) Delafuente, M.; Hernanz, A. Br. J. Cancer 1988, 58, 567-569. 10.1021/ac050549m CCC: $30.25

© 2005 American Chemical Society Published on Web 07/13/2005

disorder of glycosylation 1b (CDG-1b).13,14 CDG-1b results from hypoglycosylation of serum and other glycoproteins and is presented with congenital hepatic fibrosis and protein-losing enteropathy. Structural and mechanistic information about PMIs from different organisms will help in the design and optimization of compounds that selectively inhibit the fungal and other pathogenic enzymes while maintaining healthy mammalian cells. Proudfoot et al.15 defined three types of PMIs based on sequence alignment and physiochemical and kinetic properties. The enzymes from mammals and most bacteria belong to type 1 PMIs, which are homologous monofunctional proteins catalyzing the single isomerization reaction.16,17 The 1.7-Å crystal structure of type I PMI from Candida albicans9 is the only three-dimensional structure of this enzyme reported to date. This high-resolution structure of the intact enzyme shows the existence of an active site on the surface of the central domain with a single zinc atom bound. Kinetic studies have demonstrated the necessity of this metal cofactor for PMI activity.18 The active site forms a deep, open cavity that has suitable dimensions to bind its substrate Man6P or Fru6P. However, without the crystal structure of a PMI-substrate complex available, the role of the zinc cofactor in catalysis is still not clear, and the actual role and positioning of the nucleophilic residue has not yet been determined. Since the elucidation of the catalytic and regulatory mechanisms of PMI is essential for rational drug design, various other techniques such as inhibition studies,5,17,19,20 site-specific labelings,21-23 and site-directed mutations20,22 have been employed to generate complementary information. Electrospray ionization (ESI)-MS has been increasingly recognized as a powerful technique to investigate various biological events, such as protein folding,24 posttranslational modification,25 and more recently, noncovalent protein-ligand or protein-protein interactions.26-34 The soft ionization feature of electrospray allows (13) Jaeken, J.; Matthijs, G.; Saudubray, J. M.; Dionisi-Vici, C.; Bertini, E.; de Lonlay, P.; Henri, H.; Carchon, H.; Schollen, E.; Van Schaftingen, E. Am. J. Hum. Genet. 1998, 62, 1535-1539. (14) Niehues, R.; Hasilik, M.; Alton, G.; Korner, C.; Schiebe-Sukumar, M.; Koch, H. G.; Zimmer, K. P.; Wu, R. R.; Harms, E.; Reiter, K.; von Figura, K.; Freeze, H. H.; Harms, H. K.; Marquardt, T. J. Clin. Invest. 1998, 101, 1414-1420. (15) Proudfoot, A. E. I.; Turcatti, G.; Wells, T. N. C.; Payton, M. A.; Smith, D. J. Eur. J. Biochem. 1994, 219, 415-423. (16) Proudfoot, A. E. I.; Payton, M. A.; Wells, T. N. C. J. Protein Chem. 1994, 13, 619-627. (17) Wells, T. N. C.; Coulin, F.; Payton, M. A.; Proudfoot, A. E. I. Biochemistry 1993, 32, 1294-1301. (18) Gracy, R. W.; Noltman, E. A. J. Biol. Chem 1968, 243, 4109-4116. (19) Wells, T. N. C.; Payton, M. A.; Proudfoot, A. E. I. Biochemistry 1994, 33, 7641-7646. (20) Wells, T. N. C.; Scully, P.; Paravicini, G.; Proudfoot, A. E. I.; Payton, M. A. Biochemistry 1995, 34, 7896-7903. (21) Coulin, F. Biochemistry 1993, 32, 14139-14144. (22) Wells, T. N. C.; Scully, P.; Magnenat, E. Biochemistry 1994, 33, 57775782. (23) Papoutsopoulou, S. V.; Kyriakidis, D. A. Mol. Cell. Biochem. 1997, 177, 183-191. (24) Deng, Y. Z.; Smith, D. L. J. Mol. Biol. 1999, 294, 247-258. (25) Oda, Y.; Nagasu, T.; Chait, B. T. Nat. Biotechnol. 2001, 19, 379-382. (26) Carroll, K.; Gao, H.; Chen, H.; Bertozzi, C.; Leary, J. A. PLOS, submitted. (27) Ganem, B.; Li, Y. T.; Henion, J. D. J. Am. Chem. Soc. 1991, 113, 3, 62946296. (28) Loo, J. A. Mass Spectrom. Rev. 1997, 16, 1-23. (29) Loo, J. A. Int. J. Mass Spectrom. 2000, 200, 175-186. (30) Pi, N.; Yu, Y.; Mougous, J. D.; Leary, J. A. Protein Sci. 2004, 13, 903-912. (31) Pinkse, M. W. H.; Heck, A. J. R. J. Am. Soc. Mass Spectrom., 2004, 15, 1392-1399.

the weakly bound noncovalent complexes formed in solution to be transferred into the gas phase and observed by mass spectrometry, through which important information about these complexes, such as molecular weight, binding stoichiometry, and relative binding strength, can be garnered. Using titration or competition experiments, dissociation constants for a variety of biological noncovalent complexes have been measured.35-39 Composition determination of metalloproteins and characterization of the metal ion binding/release effects using the ESI-MS approach have also been well documented.28,40-43 Compared to other techniques for noncovalent binding investigations, ESI-MS offers high assay speed and the ability to obtain stoichiometric information directly. Previously, our laboratory reported the kinetic measurements of PMI in both directions using a novel tandem mass spectrometry-based strategy.44 In this present study, we use ESI-FTICR mass spectrometry to probe the noncovalent complexes of PMI with its ligands, including the zinc cofactor, the two natural substrates Man6P and Fru6P, and two known inhibitors erythrose 4-phosphate and mannitol 1-phosphate. These results, together with previous kinetic and inhibition studies, provide more comprehensive characterization of this enzyme and its associated ligands. EXPERIMENTAL SECTION Materials. Escherichia coli PMI (as a suspension in (NH4)2SO4) was obtained from Sigma Chemical Co. (St. Louis, MO) and was used without further purification. Glucose 6-phosphate (Glc6P), Man6P, Fru6P, erythrose 4-phosphate (Ery4P), and mannitol 1-phosphate (Mal1P) were also purchased from Sigma. All other salts are of analytical grade, and all solvents are of HPLC grade. Instrumentation. The protein characterization and noncovalent complex experiments were performed on a Bruker FT-ICR MS equipped with an actively shielded 7-T superconducting magnet. Solutions were infused into an Apollo electrospray source (Bruker, Billerica MA) at a rate of 2 µL/min using a syringe pump. The N2 nebulizing and drying gas pressure was maintained at 50 and 25 psi, respectively. Protein samples were analyzed either in 80:20 acetonitrile/water solution containing 1% formic acid or in 50 mM NH4OAc (pH 7.5). The bias on the glass capillary was kept at 4600 V, and 170 °C drying gas was used to assist in the (32) Sobott, F.; Robinson, C. V. Curr. Opin. Struct. Biol. 2002, 12, 729-734. (33) van den Heuvel, R. H. H.; Heck, A. J. R. Curr. Opin. Struct. Biol. 2004, 8, 519-526. (34) Yu, Y.; Kirkup, C. E.; Pi, N.; Leary, J. A. J. Am. Soc. Mass Spectrom. 2004, 15, 1400-1407. (35) Lim, H. K.; Hsieh, Y. L.; Ganem, B.; Henion, J. J. Mass Spectrom. 1995, 30, 708-714. (36) Sannes-Lowery, K. A.; Griffey, R. H.; Hofstadler, S. A. Anal. Biochem. 2000, 280, 264-271. (37) Daniel, J. M.; Friess, S. D.; Rajagopalan, S.; Wendt, S.; Zenobi, R. Int. J. Mass Spectrom. 2002, 216, 1-27. (38) Zhang, S.; Van Pelt, C. K.; Wilson, D. B. Anal. Chem. 2003, 75, 30103018. (39) Tjernberg, A.; Carno, S.; Oliv, F.; Benkestock, K.; Edlund, P.-O.; Griffiths, W. J.; Hallen, D. Anal. Chem. 2004, 76, 4325-4331. (40) Lim, J.; Vachet, R. W. Anal. Chem. 2003, 75, 1164-1172. (41) Lim, J.; Vachet, R. W. Anal. Chem. 2004, 76, 3498-3504. (42) van den Bremer, E. T. J.; Jiskoot, W.; James, R.; Moore, G. R.; Kleanthous, C.; Heck, A. J. R.; Maier, C. S. Protein Sci. 2002, 11, 17138-11752. (43) Veenstra, T. D.; Johnson, K. L.; Tomlinson, A. J.; Craig, T. A.; Kumar, R.; Naylor, S. J. Am. Soc. Mass Spectrom. 1998, 9, 8-14. (44) Gao, H.; Chen, Y.; Leary, J. A. Int. J. Mass Spectrom. 2005, 240, 291-299.

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5597

desolvation process. Further desolvation was achieved by collisions of the ions with neutral buffer gas at the nozzle-skimmer region using ∼180 V (in positive mode) capillary exit voltage. A throttle valve was installed at the nozzle-skimmer region to adjust the pressure to ∼1 × 10-6 mbar. Ions were externally accumulated in a radio frequency-only hexapole for 1 s before transfer into the ICR cell. Excessive kinetic energy was removed by colliding the ions with argon pulsed into the cell to a pressure of ∼1 × 10-8 mbar. Usually, for one transient, four loops of argon pulse were applied with a series of pump downs applied to lower the pressure in the cell to ∼1 × 10-10 mbar before ion detection. All ions were collected using gated trapping and detected using chirp excitation and broadband data acquisition using an average of 16-64 time domain transients containing 32K data points. The original time domain free induction decay spectra were zero filled, Gaussian multiplied, and Fourier transformed. All the data were acquired and processed using Bruker Xmass version 6.0.0 software. The parameters of the ESI source, ion optics, and cell were tuned for the best signal-to-noise ratio and were kept the same for a system of experiments unless mentioned otherwise. The samples of solution-phase kinetics were analyzed on a Thermo Finnigan (San Jose, CA) LCQ classic quadrupole ion trap mass spectrometer equipped with an ESI source operated in the negative ion mode. All instrumental details regarding the kinetic analysis have been described previously.44 General Procedures. To buffer exchange PMI from the original (NH4)2SO4 suspension into a mass spectrometry-friendly buffer, the stock solution was diluted 40-fold with 50 mM NH4OAc, pH 7.5, and centrifuged at 6000g for 40 min using Millipore Amicon centrifugal filters equipped with a 10 000 Da molecular weight cutoff membrane (Billerica, MA). The protein was buffer exchanged four times to completely remove (NH4)2SO4 and further diluted using 50 mM NH4OAc for MS analysis. The apoenzyme was prepared by overnight dialysis of the PMI stock solution against 50 mM NH4OAc containing 1 mM EDTA using 10 000 Da cutoff Slide-a-lyzer filters from Pierce (Rockford, IL). After removal of Zn2+, the PMI solution was buffer exchanged into 50 mM NH4OAc two to three times as described above to remove extra EDTA. The PMI concentration after buffer exchange was determined by Bradford assay, and PMI activity was assayed as described below. All ligands were prepared as 1 mM stock solution in 50 mM NH4OAc, and appropriate volumes of ligand, buffer, and enzyme were mixed and incubated for 15 min to generate noncovalent complexes. The mixture was chilled on ice and infused into the ESI-FTICR mass spectrometer for analysis. The syringe pump and spray chamber was wrapped with ice bags to prevent the protein sample from precipitating out of the solution. The solution kinetic experiments were performed in 25 mM NH4OAc, pH 7.5, at room temperature. For activity assay, PMI solution before or after buffer exchange was first diluted and then added to 5 mM Man6P solution to make a final enzyme concentration of 0.8 µg/mL. The reaction was stopped after 10 min by quenching with 4-fold MeOH. For inhibition studies, two sets of samples were prepared. One set contained Ery4P with the concentration varying from 0 to 2 mM, and the other set contained Mal1P at the same concentration range. The concentration of the substrate Man6P was kept constant at 1.25 mM in all solutions. 5598

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

Each reaction was initiated by adding an aliquot of PMI solution to a final concentration of 0.4 µg/mL and quenched with 4-fold MeOH. All quenched mixtures were further diluted by equal volumes of 1:1 MeOH/H2O solution and then delivered to the LCQ mass spectrometer to determine their compositions. The quantification of products and calculation of reaction velocities are the same as previously described for the kinetic measurements of PMI.44 Since Man6P and Fru6P have the same m/z value, MS2 spectra were used to differentiate these isomers based on the relative abundances of two diagnostic product ions (m/z 169 and 199). The values measured for two pure isomers and any unknown mixtures were then applied to a two-equation, two-unknown system as shown below:

aI169,F + bI169,M ) I169

(1)

aI199,F + bI199,M ) I199

(2)

Here, the constants I169,F and I169,M refer to the percent total ion current measured of the m/z 169 ion for Fru6P and Man6P standards, respectively. Similarly, I199,F and I199,M refer to the corresponding values measured of the m/z 199 ion for the two standards. I169 and I199 are the values measured for an unknown mixture of the two isomers. Through this system of equations, the raw percentage of Fru6P and Man6P, a and b, can be calculated. To determine the final composition of a mixture, two additional normalizations were introduced. The first normalization takes into account the bias effect of the instrumental response toward each isomer, which may come from a slight difference in ionization efficiency. To correct for this bias effect, a 1:1 mixture is analyzed, and the calculated values of a and b for this mixture are divided by the actual percentage (0.5) in order to obtain the normalization factors for each isomer, i.e., N1F ) a(1:1)/0.5; N1M ) b(1:1)/0.5. The calculated a and b for other mixtures are then divided by N1F and N1M to obtain a′ and b′; i.e., a′) a/N1F and b′) b/N1M. The second normalization ensures that. %Fru6P + %Man6P ) 100%; therefore, the final percentages a′′ and b′′ can be expressed as follows: a′′) a′/( a′+ b′); b′′) b′/( a′+ b′). The product concentration (P) is calculated by multiplying the known concentration of the initial substrate (S0) and the percentage of the product isomer measured using the above composition analysis method. The initial reaction velocity (v) is then calculated by dividing the product concentration by reaction time. The determination of the relative inhibition constant based on initial velocities will be discussed in the following sections. RESULTS AND DISCUSSION Noncovalent Complexation of PMI and Zinc Cofactor. The ESI mass spectra of E. coli PMI, in what is believed to be the denatured and native states, are shown in Figure 1. The spectrum of PMI sprayed from a solution of 80:20 acetonitrile/water containing 1% formic acid shown in Figure 1a displays a broad distribution of charge states (20+ to 51+) with a maximum at 33+, consistent with the expectation of ions originating from unfolded, loosely packed conformational states in denatured conditions during the electrospray process. The deconvoluted mass of 43 915.0 ( 1.0 Da, agrees well with the molecular mass

Figure 1. ESI mass spectra of 10 µM PMI in (a) acetonitile/H2O (80:20) with 1% formic acid, (b) 50 mM NH4OAc, pH 7.5 after direct buffer exchange, with the expansion of the 12+ charge state shown in (c).

Table 1. Measured and Theoretical Molecular Masses for PMI and Its Noncovalent Complexes PMI

measd massa (Da)

theor massb (Da)

mass error (%)

apo, denatured apo, native apo + Zn2+(holo) apo + 2Zn2+ holo + Man6P/Fru6P holo + Ery4P holo + Mal1P

43 915.0 ( 1.0 43 914.4 ( 2.8 43 977.6 ( 2.2 44 040.2 ( 1.6 44 238.2 ( 0.6 44 178.4 ( 1.2 44 242.9 ( 2.0

43 914.64 43 915.49 43 978.02 44 041.40 44 238.16 44 178.11 44 240.17

0.001 0.003 0.003 0.005 0.002 0.001 0.006

a Calculated from three charge states and at least three replicates. Calculated based on the most abundant isotopic mass using the Bruker Xmass 6.0.0 software.

b

of apo-PMI calculated based on its amino acid sequence (43 914.64 Da). Figure 1b represents the spectrum acquired at low resolution for PMI sprayed from 50 mM NH4OAc, pH 7.5, in which a narrow distribution of three major charge states (11+, 12+, and 13+) was obtained. These ions originate primarily from a more compact and folded conformation formed under physiological solution conditions. When the 12+ charge state was expanded as shown in Figure 1c, four ions were observed. The lowest m/z ion corresponds to the holo-enzyme containing one zinc ion (PMI + Zn2+); the next ion, with an approximate +40 Da mass shift from the first one, is identified as its solvent adduct coming from either Na+‚H2O or Na+‚NH3. The third ion corresponds to the holoenzyme containing two zinc ions (PMI + 2Zn2+), and the last ion, with a calculated mass of ∼40 Da higher than the third, is assigned as the solvent adduct of the third ion. All measured masses are listed in Table 1 along with their corresponding theoretical values.

Type I PMIs from various sources have been characterized as zinc-dependent metalloenzymes with one zinc atom per protein molecule,18,23 and in the crystal structure of C. albicans PMI, only one zinc cofactor was observed.9 Wells et al.17 once reported that yeast PMI has a second zinc binding site, an inhibitory site, with a dissociation constant of 6.4 µM, which is nearly 6 orders of magnitude weaker than the binding site of the primary zinc cofactor (low picomolar). Based on the high sequence homology between the E. coli enzyme and other type I PMIs, it is proposed that the second zinc ion observed in the present study originates either from an inhibitory metal ion binding site or from nonspecific associations. To differentiate between the two protein-zinc interactions, the influence of the desolvation energy (designated here by the capillary exit voltage) on the stability of the two corresponding noncovalent complexes was analyzed. As shown in Figure 2, for both the 13+ and 12+ charge states, the relative abundances of the PMI + 2Zn2+ ions versus those of the PMI + Zn2+ ions decrease gradually as the capillary exit voltage is increased. At capillary exit voltages above 182 V, binding of the second zinc is relatively negligible. In contrast, the noncovalent complex of PMI + Zn2+ is very stable and no dissociation to apoPMI was observed under the highest capillary exit voltage tested. Given the large difference between the dissociation constant of the zinc cofactor and the zinc inhibitor/nonspecific zinc clustering, it is not surprising that the less tightly bound Zn2+ is lost under higher capillary exit voltages. Since it is believed that the tightly bound zinc ion is the only essential cofactor, the capillary exit voltage was set above 182 V for the following experiments to minimize the interference of the second zinc ion and thus simplify the spectra. Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5599

Figure 2. (a) ESI mass spectra obtained for 10 µM PMI in 50 mM MH4OAc, pH 7.5 at different capillary exit voltages. The ions corresponding to PMI + Zn2+ and PMI + 2Zn2+are labeled with open circle and open triangle, respectively.

To investigate the role of the zinc cofactor in the structure and catalytic mechanism of PMI, the apo-enzyme was prepared by overnight dialysis of the PMI stock solution against 50 mM NH4OAc containing 1 mM EDTA. After buffer exchange with NH4OAc to remove EDTA, the solution was directly sprayed and the mass spectrum was acquired under the same conditions as described above. As shown in Figure 3, the apo-enzyme displays the same narrow charge-state distribution (11+, 12+, and 13+) as the holoenzyme, but at lower m/z values (calculated mass ∼64 Da lower; see Table 1). The activity assay of the apo-protein using solution kinetics as described in the Experimental Section showed undetectable product formation as expected. Reconstitution of the holo-enzyme was achieved by a stepwise titration of apo-PMI with Zn2+. Figure 3 indicates that adding Zn2+ stepwise to a 12.5 µM apo-enzyme solution results in a gradual increase of the relative abundance of the ion corresponding to the holo-enzyme. In the case of 1 equiv of Zn2+, binding appears to be completely saturated, agreeing with the low-picomolar dissociation constant of the zinc cofactor. The activity assay of this reconstituted holo-PMI solution

also showed full recovery of enzyme activity when compared to native holo-PMI (156.5 versus 157.4 units/mg). In comparing the spectra of the native holoenzyme, the apo-enzyme and the reconstituted holo-enzyme (Figure 1b and insets of Figure 3), it is clear that the charge-state distribution does not vary. This phenomenon is different from those observed for the DNA-binding domain of the vitamin D receptor43 and the colicin E9 endonuclease,42 in which metal ion binding induced an obvious variation in the charge-state distribution of the protein, suggesting conformational changes. The result obtained for PMI seems to indicate that binding of the essential zinc cofactor does not change the conformation of PMI to any appreciable degree. Data from crystal structure studies also imply that zinc plays more of a catalytic role rather than a structural role.9 Noncovalent Complexation of PMI and Isomeric Substrates. To investigate PMI in complexation with its two natural substrates, two reaction mixtures containing 10 µM holo-PMI and 50 µM concentration of either Fru6P or Man6P were analyzed. In the spectra shown in Figure 4a and b, a new series of ions with a mass shift of +260 Da were observed, which represent the noncovalent adduction of the substrate. As a control, the interaction between holo-PMI and a nonsubstrate stereoisomer, Glc6P, was also investigated. The monomer Glc6P has been reported to be a poor inhibitor of PMI with an inhibition constant at high-millimolar concentration.45 The spectrum (Figure 4c) of holo-PMI incubated with Glc6P, with a ligand-to-protein ratio of 5:1 and acquired under the same instrumental conditions as used in Figure 4a and b, shows no ligand binding. These data suggest that the complexes observed between the enzyme and its natural substrates are actual representations of the solution binding effect and not a sole artifact of gas-phase clustering. Since the PMIcatalyzed reaction is a one-substrate reversible system, incubation of either substrate with micromolar concentrations of enzyme should lead to equilibrium and the composition of the two final mixtures would be identical. Therefore, it is expected that the ratio between bound and unbound protein obtained for the two

Figure 3. (a) Expansion of the 12+ charge state of the ESI mass spectra for the titration experiment between 12.5 µM apo-PMI and Zn(OAc)2. The ions corresponding to apoprotein and PMI + Zn2+ are labeled with asterisks and open circles, respectively. The apoenzyme was obtained after overnight dialysis with EDTA. Insets show the corresponding overall charge-state distribution of each spectrum. 5600 Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

Scheme 2. Proposed Mechanism of PMI (Redrawn from Gracy and Noltmann46)

have been measured.34,37,38,47 However, in the case of PMI, the spectra obtained represent the interaction between the protein and two ligands that have the same molecular weight but potentially different dissociation constants. Therefore, only the apparent Kd measurements can be ascertained. Our method for determining this value is explained below. The actual reaction mechanism for a one-substrate reversible system can be expressed as k+1

k+2

k+3

-1

-2

-3

E + S {\ } ES {\ } EP {\ }E+P k k k Figure 4. ESI mass spectra of (a) holo-PMI with Man6P, (b) holoPMI with Fru6P, (c) holo-PMI with Glc6P, (d) apo-PMI with Man6P, (e) apo-PMI with Fru6P, and (f) apo-PMI with Glc6P. In all experiments, the concentrations of ligand and enzyme are 50 and 10 µM, respectively.

substrates should be the same, which is in fact observed (Figures 4a and b). Traces d-f in Figure 4 are the spectra of apo-PMI after incubation with the above three ligands and obtained under the same conditions as those used for acquiring Figure 4a-c. Using Glc6P as a control, the two natural substrates bind to the apoenzyme as well. This result is not surprising. As discussed above, zinc binding/release does not change the overall conformation of the enzyme to any significant extent; therefore, the substratebinding pocket should be at least partially preserved in the absence of the metal cofactor. However, in comparing Figure 4a and d or Figure 4b and e, it is obvious that the zinc cofactor does markably induce more substrate binding. More than a 5-fold increase in binding affinity can be approximated for holoenzyme compared to apoenzyme based on the relative ratio of the intact protein and protein-ligand complex ions. Although the function of the zinc cofactor in the isomerization has not yet been completely elucidated, it is proposed that the zinc cofactor coordinates with the carbonyl oxygen and the C2 hydroxyl oxygen of the hydroxyl aldehyde form of the substrate Man6P and thus facilitates the removal of the C2 hydrogen9,45,46 (Scheme 2). The observed enzyme-substrate nonconvalent complexation in the absence and presence of zinc cofactor shown here is consistent with this proposed mechanism. Apparent Dissociation Constant Measurement for the Noncovalent Complex of PMI and Isomeric Substrates. Based on the binding stoichiometry and the relative abundance of free protein and protein-ligand complex obtained using the MS approach, actual dissociation constants (Kd) for various systems (45) Noltmann, E. A. In The Enzymes, 3rd ed.; Boyer, P. D., Ed.; Academic Press: New York, 1972; Vol. 6, pp 271-313. (46) Gracy, R. W.; Noltman, E. A. J. Biol. Chem 1968, 243, 5410-5419.

(1)

In this equation, k with different subscripts represents the rate constant for each individual step. The dissociation constant of the substrate (KdS) and that of the product (KdP) can be expressed as KdS ) k-1/k+1) [E][S]/[ES], and KdP ) k+3/k-3 ) [E][P]/[EP]. As ES cannot be differentiated from EP in the noncovalent complex assays, it is impossible to measure the individual dissociation constant of each ligand. What can be measured, however, based on the relative ratio of bound and unbound protein, is the apparent dissociation constant as expressed in eq 2:

Kdapp ) [E]([S] + [P])/([ES] + [EP]) ) ([E][S] + [E][P])/([ES] + [EP]) (2) By substituting [ES] and [EP] with the expressions of their individual dissociation constant, eq 2 can be written as

Kdapp ) ([E][S] + [E][P])/([E][S]/KdS + [E][P]/KdP) ) (1 + K)/(1/KdS + K/KdP) (3) In eq 3, K )[P]/[S] and refers to the concentration ratio between product and substrate. When the reaction reaches equilibrium, which the system studied in this case does, K equals the equilibrium constant of the reaction Keq. According to previous kinetic studies 45 and the one we recently reported,44 Keq of this enzymatic system is 1.1. Equation 3 is thus simplified as

Kdapp ) 2.1KdSKdP/(1.1KdS + KdP)

(4)

Therefore, the value of Kdapp measured by this method can be considered as a weighted average between KdS and KdP. Although it is not the actual dissociation constant for a protein-ligand binary (47) Ayed, A.; Krutchinsky, A. N.; Ens, W.; Standing, K. G.; Duckworth, H. W. Rapid Commun. Mass Spectrom. 1998, 12, 339-344.

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5601

and [EL] is the enzyme-ligand complex concentration ([EL])[ES] + [EP]). The number of ligands bound per protein molecule, expressed as the ratio between complex concentration [EL] and total enzyme concentration [E]0, can be written as

[EL] [EL] 1 ) ) ) [E]0 [E] + [EL] [E] +1 [EL] 1 Kd

[L]

Figure 5. ESI mass spectra showing the titration of 10 µM holoPMI with Fru6P (a) and the corresponding saturation plot (b). The ions corresponding to free protein and protein-ligand complex are labeled with open circle and open triangle, respectively. The 13+ and 11+ charge state ions are multiplied 4-fold for clarity.

complex, Kdapp can represent the binding affinity of the two isomeric substrates toward the enzyme. To measure the value of Kdapp between PMI and its isomeric substrates using noncovalent complexation assay, a titration experiment between 10 µM holoenzyme and Fru6P was performed. The resulting spectra are shown in Figure 5a. Clearly, with the increase in Fru6P concentration, more protein-ligand complex is formed, as expected for typical saturation behaviors. However, increasing the ligand concentration above 100 µM substantially supresses the signal-to-noise ratio and also shows nonspecific ligand binding to the enzyme (data not shown). To avoid the complication caused by nonspecific binding in the analysis of the titration experiments, the Fru6P concentration was kept below 80 µM where binding of more than one ligand is relatively negligible. For the determination of Kdapp based on the titration experiment, a general equilibrium expression for a single binding site system is used as shown in eq 5.

Kdapp ) [E][L]/[EL]

(5)

In this equation, [E] refers to the free enzyme concentration, [L] is the free ligand concentration (for this system, [L])[S] + [P]), 5602

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

[L]

)

app

+1

Kd

app

+ [L]

(6)

Assuming that the ionization efficiency of the free protein and protein-ligand complex is comparable, the value of [EL] can be calculated based on the relative ion abundances, IE and IEL, respectively, i.e., [EL]) [E]0IEL/(IE + IEL). Figure 5a shows that the relative ratio of IE and IEL at different charge states is different, with lower charge state showing more complexation. This observation is consistent with the assumption that lower charge states represent more folded conformations of the protein, and thus induce more ligand binding, as has been reported previously.48 Therefore, the sum of ion abundances at all charge states was used to calculate the concentration of the complex. The value of [L] can be determined as [L] ) [L]0 - [EL], where [L]0 is the initial ligand concentration. By plotting the value of [EL]/[E]0 as a function of [L], the apparent dissociation constant can be determined by nonlinear fitting. The saturation plot obtained from the titration experiment is shown in Figure 5b, in which the experimental data were fitted by the Grafit software according to eq 6. The apparent Kd for the PMI-substrate complex was determined to be 88.8 ( 2.7 µM based on three measurements. This is the first time report of the apparent Kd value for this complicated system. As studying interactions between a protein and isomeric ligands is difficult for solution binding assays, gasphase noncovalent complexation can provide useful options for such systems. Noncovalent Complexation of PMI and Inhibitors. The noncovalent complexation between PMI and possible inhibitors was also investigated. Since the inhibition constant Ki of an inhibitor represents the dissociation constant of the enzymeinhibitor complex regardless of the reaction mechanism, the Ki ratio of two inhibitors measured through solution kinetics should be equal to their Kd ratio measured through gas-phase noncovalent complexation, if the complexes remain undisturbed during the ESI process and if gas-phase nonspecific interactions between the protein and excess ligand are negligible or controlled.34 Ery4P and Mal1P are two known competitive inhibitors of yeast PMI, and both Ki values have been reported to be within the range of 0.1-0.8 mM.16,45 The spectrum of 10 µM holo-PMI mixed with 150 µM Ery4P and 150 µM Mal1P is shown in Figure 6a. Both PMI-Ery4P and PMI-Mal1P noncovalent complexes were observed, which is consistent with the competitive feature of the two inhibitors. By using one concentration of both inhibitors in a single experiment, the relative dissociation constant of the two (48) Wang, W.; Kitova, E. N.; Klassen, J. S. Anal. Chem. 2003, 75, 4945-4955.

where I refers to the inhibitor concentration and Ki is the inhibition constant. This equation can be further rearranged to eq 9:

1 KM + S KM/Ki ) + I v VmaxS VmaxS

Figure 6. (a) ESI mass spectrum of 10 µM holo-PMI with 150 µM Ery4P and 150 µM Mal1P. (b) 1/v versus I plots for Ery4P (filled triangle) and Mal1P (filled circle).

(9)

Therefore, with the same substrate concentration S and kinetic parameters KM and Vmax, the relative Ki of the two inhibitors could be determined based on the ratio of the two slopes of the corresponding 1/v versus I plots, i.e., Ki,Ery4P/Ki,Mal1P ) slopeMal1P/ slopeEry4P. The slopes of the 1/v versus I plots for Ery4P and Mal1P, shown in Figure 6b, are 16.3 and 11.4, respectively. The value of Ki,Ery4P/Ki,Mal1P was thus measured to be 0.70, which correlates well with the value of Kd,Ery4P/Kd,Mal1P obtained from the gas-phase data. This result suggests that, for PMI, the binding domain is preserved from solution phase to gas phase during the ESI process and nonspecific association does not exist to any significant extent. CONCLUSION

inhibitors can be determined according to eq 7:

Kd,Ery4P [PMI][Ery4P] [PMI - Mal1P] ) ) Kd,Mal1P [PMI - Ery4P] [PMI][Mal1P] [PMI - Mal1P][Ery4P] (7) [PMI - Ery4P][Mal1P]

Again assuming that the ionization efficiencies of the proteinligand noncovalent complexes and the free protein are similar, the concentrations of different PMI-associated species can be calculated based on their corresponding ion abundances. The sum of ion abundances of PMI-Ery4P, PMI-Mal1P, and free PMI at all three charge states were measured to be 0.5, 0.4, and 1.0, respectively. Using an initial PMI concentration of 10 µM, concentrations of PMI-Ery4P and PMI-Mal1P were determined to be 2.7 and 2.1 µM, respectively. The concentrations of free Ery4P and Mal1P were calculated to be 147.3 and 147.9 µM, respectively, based on their initial concentration of 150 µM. Therefore, the ratio of Kd,Ery4P versus Kd,Mal1P (eq 7) is 0.77 using the noncovalent complexation assay. Using the methodology reported previously for solution kinetic measurements of PMI based on tandem mass spectrometry,44 the relative Ki values of the two inhibitors were determined. For this experiment, two series of solutions containing 1.25 mM Man6P and 0-2 mM of either inhibitor were prepared, and their corresponding reaction velocities were determined. For competitive inhibitions, the rate equation can be expressed as

v)

VmaxS KM(1 + I/Ki) + S

The noncovalent interaction of the metalloenzyme PMI has been studied by ESI-FTICR mass spectrometry. The complexes of apo-PMI with the zinc cofactor and the holo-enzyme with two isomeric substrates and two inhibitors have been generated and their binding properties have been investigated. The binding/ release study of the zinc ion on the enzyme itself and on the enzyme-substrate complexes supports the catalytic function of the cofactor. The binding studies for the two natural substrates and a nonsubstrate stereoisomer suggest that the observed gasphase noncovalent complexes represent the actual solution interactions. Using this assay, the apparent dissociation constant for the complex between PMI and the isomeric substrates was determined for the first time. The relative dissociation constant measured for two inhibitors correlates well with their relative inhibition constants, suggesting that the extent of solution binding is preserved in transferring the complexes from solution into the gas phase. In summary, the present study demonstrates that gasphase noncovalent complexation can generate useful information for complicated binding systems such as the one between a protein and isomeric ligands, which are not easily achieved by traditional solution binding assays. ACKNOWLEDGMENT The authors gratefully acknowledge the NIH Grant GM 63581for funding this work.

Received for review March 31, 2005. Accepted June 14, 2005.

(8) AC050549M

Analytical Chemistry, Vol. 77, No. 17, September 1, 2005

5603