γ-Glutamyl Hydrolase: Kinetic Characterization of Isopeptide Hydrolysis Using Fluorogenic Substrates

Abstract

γ-Glutamyl hydrolase, a cysteine peptidase, catalyzes the hydrolysis of poly-γ-glutamate derivatives of folate co-factors and many antifolate drugs. We have used internally quenched fluorogenic derivatives of glutamyl-γ-glutamate and (4,4-difluoro)glutamyl-γ-glutamate to examine the effect of fluorine substitution adjacent to the scissile isopeptide bond. Using a newly developed continuous fluorescence assay, the hydrolysis of both substrates could be described by Michaelis-Menten kinetics. Fluorine substitution resulted in a significant decrease in observed rates of hydrolysis under steady-state conditions due primarily to a ~ 15-fold increase in K_m. Using stopped-flow techniques, hydrolysis of the non-fluorinated isopeptide was characterized by a burst phase followed by a steady-state rate, indicating that formation of the acyl enzyme is not rate-limiting for hydrolysis of this isopeptide. This conclusion was confirmed by analysis of the progress curves over a wide range of substrate concentration, which demonstrated that the acylation rate (k₂) is ~ 10-fold higher than the deacylation rate (k₃). The increased value of K_m associated with the difluoro derivative limited the ability to obtain comparable pre-steady-state kinetics data at saturating concentration of substrate due to inner filter effects. However, even under non-saturating conditions, a modest burst was observed for the difluoro derivative. These data indicate that either deacylation or rearrangement of the enzyme-product complex is rate-limiting in this isopeptide hydrolysis reaction.

γ-Glutamyl hydrolase (GH, EC 3.4.19.9), a lysosomal cysteine protease, plays an important role in maintaining folate homeostasis (1). The folates are key cofactors in one-carbon metabolism leading to such essential biosynthetic products as glycine, methionine, thymidylate, and purine nucleotides (2, 3). The folylmonoglutamates are elongated to folylpolyglutamates by the enzyme folylpoly-γ-glutamate synthetase (FPGS, EC 6.3.2.17) in an ATP-dependent ligation process, effectively trapping the cofactor in the cell (4, 5). GH catalyzes the hydrolysis of the Glu-γ-Glu bonds to form folylmonoglutamates that can then be exported from the cell. Thus, these two enzymes are primarily responsible for the regulation of folate levels in the cell (Figure 1).

Figure 1.

Folylpolyglutamate synthesis and hydrolysis. The folylmonoglutamate is elongated in an ATP-dependent reaction catalyzed by FPGS. GH catalyzes the hydrolysis of the γ-glutamyl bonds.

Consistent with this notion, high GH activity has also been shown to decrease the efficacy of several polyglutamylated antifolate chemotherapeutic drugs such as methotrexate (AMPte-Glu) (1). The polyglutamylated drug is hydrolyzed by GH, generating “free” drug that is retained poorly and leads to reduced cytotoxicity. A number of single nucleotide polymorphisms (SNPs) have been identified in the human GH gene both in the promoter region and the mature enzyme (6). One of these SNPs has been found in acute lymphoblastic leukemia patients with low GH activity. This SNP has been shown to reduce GH activity on long-chain methotrexate polyglutamates leading to the accumulation of intracellular methotrexate polyglutamates in leukemia cells (7). These reports place GH as an integral component in the regulation of the intracellular level of both folates and multiple antifolate drugs.

We have developed fluoroglutamate-containing γ-glutamyl peptides as mechanistic probes for GH (8). In earlier studies with GH from hog kidney, it was observed that a methotrexate derivative, 2-amino-10-methylpteroyl (2′RS,4′RS)-4-fluoroglutamyl-γ-glutamate (AMPte-(4-F)Glu-γ-Glu) was hydrolyzed significantly more slowly than AMPte-Glu-γ-Glu (9). These findings were corroborated with recombinant rat and human enzymes where it was found that substrates containing either (2S,4S)-4-fluoroglutamate (10) or (2RS)-4,4-fluoroglutamate (11) were hydrolyzed at least 25-fold more slowly than their non-fluorinated analogs (12). Fluorine substitution for hydrogen is sterically conservative but imparts large electronic effects that can change rates of reaction (13). It was of interest to determine the mechanistic basis for the attenuated rate of GH-catalyzed hydrolysis of the fluoroglutamate-containing substrates. In the studies reported here, recombinantly expressed rat GH was used, as it not only has demonstrated greater activity on shorter peptides than human GH, but also exhibits an apparent endopeptidase mode of cleavage that results in products that are not further substrates for the enzyme (14, 15), thereby simplifying the kinetics analysis.

Despite the interest in GH, protocols available for measuring its activity have been limited to endpoint assays of cleavage products using high-performance liquid chromatography (16–20) or capillary electrophoresis (21, 22). We have reported the first continuous assay for GH based on an internally quenched fluorogenic glutamyl-γ-glutamyl peptide with an N-terminal N-methyl-p-aminobenzoyl (N-Me-pAB) moiety as the fluorophore (23). Its fluorescence is quenched in the intact substrate by fluorescence resonance energy transfer (FRET) (24) to a C-terminal 3-nitrotyrosine (Tyr(NO₂)). Cleavage of the Glu-γ-Glu bond results in release of the Tyr(NO₂) moiety, so that fluorescence of the N-terminal hydrolysis product is observed. However, this substrate could only be used at concentrations up to about 10 μM because the strong absorbance of the N-Me-pAB moiety results in a decrease in the observed fluorescence at higher substrate concentrations, a phenomenon commonly referred to as the inner filter effect (IFE) (24). A second-generation fluorogenic substrate (1, Figure 2), in which the N-Me-pAB fluorophore is replaced with an o-aminobenzoyl (Abz) group, extends the useful concentration range before significant inner filter effects are observed and provides better sensitivity with a larger fluorescent signal. In order to study directly the influence of fluorine on GH-catalyzed hydrolysis of γ-glutamyl peptides, 4,4-difluoroglutamate has also been incorporated into this framework (2, Figure 2) (25). These compounds have been used in a continuous fluorescence assay to provide a detailed kinetic comparison between the GH-catalyzed hydrolysis of glutamyl and fluoroglutamyl peptide substrates. The studies described herein define the steady-state kinetic parameters for GH-catalyzed hydrolysis of 1 and 2. Transient kinetic parameters were determined using stopped-flow fluorescence spectroscopy. These experiments provide the first definitive evidence for formation of an intermediate in the reaction catalyzed by GH. Based on the steady-state and transient kinetics data, possible mechanisms for fluorine-mediated modulation of γ-glutamyl peptide hydrolysis are discussed.

Figure 2.

GH-catalyzed hydrolysis of internally quenched fluorescent substrates 1 and 2.

EXPERIMENTAL PROCEDURES

Materials, general synthetic methods, and details on the solution phase synthesis of 1 (Scheme 1) are described in the Supporting Information. The syntheses of N-Abz-Glu·TFA (3a) and Glu-γ-Tyr(NO₂)·Et₃N (4), the two expected products resulting from GH-catalyzed hydrolysis of 1 (Figure 2), are also described in the Supporting Information. For access to larger amounts of 1 required for transient kinetics experiments, isopeptide 1 was also synthesized by manual solid phase peptide synthesis on Wang resin (1.00 g, 1.2 mmol, suspended in 10 mL dry DMF) using an Fmoc protection scheme and HBTU activation (Scheme 2, See Supporting Information for details). Abz-4,4-F₂Glu-γ-Glu-γ-Tyr(NO₂) (2) was synthesized by Dr. David Konas (25).

Scheme 1.

Scheme 2.

Analytical HPLC was controlled and monitored using either two Altex pumps, a Hewlett-Packard diode array UV/Vis detector and an HP workstation, or a Waters system with a 6000a pump and a 510 pump with a Waters 996 photodiode array detector running under Millenium³² software (version 3.2). A Vydac 218TP54 C₁₈ analytical column was eluted at 1 mL/min under one of the following conditions. Solvents: A, acetonitrile with 0.1% AcOH and 0.02% TFA (v/v); B, H₂O with 0.1% AcOH and 0.02% TFA (v/v). Gradient 1: a separation beginning with 15% solvent A and 85% solvent B, which increased to 20% A over 25 min, then ramped to 95% solvent A in 15 min. Gradient 2: a separation beginning with 3% solvent A and 97% solvent B which was linearly increased to 50% A over 43 min, then ramped to 95% solvent A in 2 min.

LC-MS analysis was performed on a system including a Finnigan LCQ mass spectrometer interfaced with either a Hitachi HPLC and Hitachi detector (UV or fluorescence), or ThermoFinnigan Surveyor HPLC components with a fluorescence detector incorporated into the system using a splitter between the fluorescence detector and mass spectrometer. The HPLC separation was conducted on a Vydac C₁₈ column with a gradient elution of solvent A and solvent B with either (Gradient 3) a linear increase from 3% to 50% solvent A over 22 min at 1 mL/min or (Gradient 4) a linear increase from 3% solvent A to 50% solvent B over 43 min at 0.8 mL/min. UV detection was set at 230 and 278 nm, and fluorescence detection was set at λ_ex = 320 nm, λ_em = 420 nm for Abz-containing compounds.

Initial Biochemical Characterization of FRET Peptides

Abz-Glu-γ-Glu-γ-Tyr(NO₂) Enzymatic Hydrolysis: LC-MS Analysis

Abz-Glu-γ-Glu-γ-Tyr(NO₂) (1, 50 μM) was incubated with GH (0.72 nM) for 85 min. The resulting mixture was analyzed by LC-MS (Hitachi HPLC components, Gradient 3); t_R = 8.0 min, (M+H)⁺ m/z 266.9 (3a); t_R = 9.0 min, (M+H)⁺ m/z 356.1 (4).

Extinction Coefficient of Abz-Glu-γ-Glu-γ-Tyr(NO₂) (1)

Insufficient quantities of 1 and 2 were available by solution phase synthesis for accurate determination of an extinction coefficient for these substrates. Therefore, the extinction coefficients of the N-terminal chromophore, N-Abz-glutamic acid, 3a, and the C-terminal chromophore, N-Ac-3-nitrotyrosine, S3, were obtained to provide an accurate extinction coefficient for the intact peptides. N-acetyl-3-nitrotyrosine (S3, Supporting Information, 268.2 mg, 1.0 mmol), previously dried over P₂O₅ at 25 °C for 12 h, was dissolved in 2 L of Milli-Q water to provide a 500 μM stock solution. Similarly, Abz-Glu (3a, Supporting Information, 132.9 mg, 0.5 mmol), previously dried over P₂O₅ at 25 °C for 12 h, was weighed under N₂. This was diluted to 1 L in H₂O to provide a 500 μM stock solution. These stock solutions were diluted in either 0.1 M HCl or 0.1 M NaOH to provide 1 mL volumes with final concentrations of 2, 5, 10, 25, 50, 75, and 100 μM. All dilutions were made in triplicate. Spectra were recorded between 220–600 nm at a scan rate of 600 nm/min at each concentration in each solvent. Plots of the average absorbance at 278.5 nm (0.1 M HCl) or 285 nm (0.1 M NaOH) versus concentration were then fit to Beer’s Law and the extinction coefficients of each chromophore summed to give values of ε = 6690 M⁻¹cm⁻¹ (λ = 278.5 nm, 0.1 M HCl) and ε = 5080 M⁻¹ cm⁻¹ (λ = 285 nm, 0.1 M NaOH). ¹ Following successful synthesis of 1 on a larger scale by solid phase synthesis, direct determination of the extinction coefficient was possible and led to a value of ε = 5110 M⁻¹ cm⁻¹ (λ = 285 nm, 0.1 M NaOH), in good agreement with that determined for 1 as described above. Considering the identical chromophores in the donor and acceptor moieties of 1 and 2, the extinction coefficient of 2 was assumed to be the same as that determined for 1. ²

Enzyme Assays

Protein concentration

For routine determination of protein concentration, the method of Bradford (26) was used with reagents purchased from either BioRad or Sigma. The manufacturer’s instructions for the microassay were followed and bovine serum albumin was used as the standard. Initially, for steady-state experiments, the GH concentration was determined by the Bradford assay adjusted for enzyme purity (ca 75%) based on quantitative SDS-PAGE gel analysis. A more reliable determination of active GH concentration using more highly purified GH was obtained by a further adjustment based on a fit to the pre-steady-state kinetics data (eq 1–3); vide infra. Concentrations of GH determined in this manner are indicated in all experimental protocols and were used in the calculation of rate constants (Table 1).

Table 1.

Burst and steady -state kinetics of GH and Abz-Glu-γ-Glu-γ-Tyr(NO₂).

Exponential phase					Linear phase
Entry	KS (μM)	Km (μM)	k2 (s−1)	k3 (s−1)	Km (μM)	kcat (s−1)
1a	67	2.6	70	83	------	------
2b	67	2.6	130	23	------	------
3c	67	3.9	75	78	2.6	1.1
4d	67	3.9	139	14	2.6	4.0

^aFit to the data for the exponential phase (eq. 1–3) using Kaleidagraph with the following parameters: [E] = 3 μM (Bradford assay), K_m = 2.60 μM (Figure 5D, Michaelis -Menten equation), k₂+k₃ = 153 s⁻¹ (Figure 5B, eq. 2).

^bFit to the data for the exponential phase (eq. 1–3) using Kaleidagraph with the following parameters: [E] = 0.86 μM based on the fraction of active enzyme (28.6%) determined from the ratio of the burst amplitude, A to [E] (Figure 4); other parameters as in footnote ‘a’.

^cGlobal fit to the data (eq. 1–3) using Dynafit (61) with the following parameter: [E] = 3 μM.

^dGlobal fit to the data (eq. 1–3) using Dynafit (61) with the following parameter: [E] = 0.86 μM.

Continuous Steady-State Kinetic Analysis

Concentrations of stock solutions of Abz-Glu-γ-Glu-γ-Tyr(NO₂) were determined based on the UV absorbance at 285 nm (ε = 5080 M⁻¹cm⁻¹, vide supra) in 0.1 M NaOH. All solutions except those of the enzyme were passed through a 0.2 μm syringe filter or a spin filter. In a typical assay, 90 μL of assay buffer pH 6.0, containing 50 mM β-mercaptoethanol, 50 mM sodium acetate, and 1 or 2 (concentration varied) was equilibrated at 25 °C in an amber microcentrifuge tube. The stock solution of GH was centrifuged at 10,000 rpm (8,000 × g) for 2 min to sediment any dust particles, and the supernatant was then carefully removed and diluted 10-fold in storage buffer consisting of 50 mM sodium acetate pH 5.5, 50 mM BME, 1 mM octyl-β-glucopyranoside, 1 mM EDTA, 1 M NaCl, pH 5.5. Preincubation of GH (15 μL) at 25 °C for 2 h was necessary to obtain maximal GH activity that appeared to be stable for at least the next 6 h. (14) To initiate the reaction, an aliquot of activated GH (10 μL) was added to the assay buffer and substrate in a microcuvette positioned in a jacketed cell holder. Final concentrations were typically 50 mM sodium acetate, pH 6.0, 50 mM BME, 0.25–75 μM substrate (1 or 2), and 0.72 nM GH (1) or 6.4 nM GH (2) in a final volume of 100 μL.

Reaction progress was monitored by time-based fluorescence acquisition (Spex Fluoromax-2) with the following parameters: time interval 1 s, integration 1.0 s, slit width 5 nm (band pass), λ_ex = 320 nm, λ_em = 420 nm. The cuvette was capped tightly and the fluorescence was monitored for 2–10 min. Samples were transferred from the microcuvette to amber microcentrifuge tubes and a 1 μL aliquot of concentrated GH (typically 2.9 μM final concentration) was added to hydrolyze completely the remaining substrate. The controls at each concentration of substrate were treated as above except that GH storage buffer was added in place of GH. Both sample and control tubes were incubated at 25 °C to allow complete hydrolysis of the samples and the final fluorescence was measured for 25 s with an interval of 5 s and integration time of 1 s to provide the fluorescence endpoints (F_∞).

The corrected fluorescence was obtained by subtracting the fluorescence of the control from that of the corresponding sample. Samples with substrate concentrations >20 μM were corrected for IFE (vide infra). A calibration curve to convert fluorescence units to micromolar product was created by plotting IFE-corrected F_∞ vs. substrate concentration. The inverse of the slope of this line (μM/fluorescence unit) multiplied by the initial rate in fluorescence units/s provides the initial rate in units of μM/s. Initial rates were then fit to the Michaelis-Menten equation. To determine k_cat, V_max was divided by the total enzyme concentration. Data fitting was completed in Kaleidagraph version 3.5 (Synergy software, Reading, PA).

Inner Filter Effect Corrections

Solutions containing varying concentrations (0.25–75 μM) of equimolar Abz-Glu and Glu-γ-Tyr(NO₂) were used to determine a correction factor for each concentration of substrate used in the assay. The measured fluorescence divided by the correction factor provided the corrected fluorescence value. This IFE correction was applied to initial rates measured at the higher substrate concentrations of 1 or 2 (35, 50, and 75 μM), whereby the initial rate was divided by an appropriate correction factor (0.91, 0.90, or 0.62 respectively). Final fluorescence values obtained by hydrolysis of 1 or 2 (10, 20, 35, 50, and 75 μM) were similarly corrected (factors of 0.97, 0.92, 0.83, 0.77, and 0.55 respectively).

Transient Kinetics

Transient kinetics experiments were carried out on Hi-Tech SF-61 or Hi-Tech SF-61DX2 stopped-flow instruments with a xenon lamp as the light source. The excitation wavelength was 320 nm (unless otherwise noted), and a 389 nm cutoff filter was used to select the emission wavelengths. Experiments were initiated by mixing equal volumes of enzyme and substrate at 25 °C or 4 °C. The substrate solutions were prepared in 50 mM sodium acetate, pH 6.0, 50 mM BME. Substrate concentrations are reported as final concentration after mixing with GH. Unless otherwise noted, solutions of GH were prepared in GH storage buffer, pH 5.5 with [NaCl] = 0.2 M to obtain a final concentration 0.1 M [NaCl] after mixing with substrate. Data were collected to at least 10 s using the program KISS (Kinetic Instruments, Inc., Ann Arbor, MI) on the Hi-Tech single-mixing stopped-flow instrument or KinetAsyst on the SF-61DX2. The dead time of the instrument is 1.5–2 ms. Data were collected at least in triplicate unless otherwise noted and were corrected by subtracting the fluorescence of the intact substrate. Standard curves were generated by measuring the fluorescence of solutions of various concentrations of the fluorescent hydrolysis product, Abz-Glu, varying concentrations of a 1:1 mixture of the reaction products, Abz-Glu and Glu-Tyr(NO₂) or N-Acetyl-Tyr(NO₂), or by internal calibration based on end points of completely hydrolyzed samples.

For samples exhibiting a burst phase, the average progress curves out to 0.5 s were fit to the burst equation (eq 1) where P represents the product formed in terms of

$$\text{P}=\text{A}(1-{\text{e}}^{(-\text{kt})})+\text{vt}+\text{C}$$ (1)

fluorescence, A is the amplitude of the burst, k is the burst rate (k_burst), t is the time in seconds after mixing, v is the slope of the steady-state linear part of the curve (v_ss), and C is the intercept extrapolated from the linear portion (27, 28). In experiments varying [S]₀ at constant [GH], the progress curves were fit to obtain k_burst. A plot of k_burst versus substrate concentration was fit to eq 2 (27, 28). A plot of the burst amplitude (A in eq 1) versus substrate concentration was fit to eq 3 (27, 28). [E] is known from the stoichiometry of the burst, and K_m is known from fitting the Michaelis-Menten equation, using v_ss (from eq 1) as the steady-state rate. Thus, a fit of the data to eq 2 provides a value of (k₂ + k₃) as well as K_s. Using the value of (k₂ + k₃) thus obtained, a fit of eq 3 provides a value of k₂. Subtraction of k₂ from (k₂ + k₃) provides the value for k₃.

$${\text{k}}_{\text{burst}}=({\text{k}}_2+{\text{k}}_3)[\text{S}]/({\text{K}}_{\text{S}}+[\text{S}])$$ (2)

$$\text{A}=[\text{E}]{({\text{k}}_2/({\text{k}}_2+{\text{k}}_3))}^2/(1+{\text{K}}_{\text{m}}/[\text{S}]){)}^2$$ (3)

RESULTS

Synthesis of FRET Peptide Substrate for GH

A γ-glutamyl peptide with the o-aminobenzoyl fluorophore at the N-terminus and the 3-nitrotyrosine quenching moiety at the C-terminus was first synthesized as shown in Scheme 1 (See Supporting Information for details). Compound 5, a protected diglutamyl peptide (23)³, was deprotected at the N-terminus and coupled to Boc-protected o-aminobenzoic acid (6) using standard EDC-mediated peptide coupling. This was followed by deprotection of the C-terminus, coupling of the 3-nitrotyrosine quenching moiety (8), and two deprotection steps to provide the desired product, 1. To study directly the effect of fluorine on GH-catalyzed hydrolysis of γ-glutamyl peptides, 4,4-difluoroglutamate was also incorporated using the same synthetic strategy to provide the related fluoropeptide, 2 (25). The two putative products of GH-catalyzed hydrolysis of 1, Abz-Glu (3a) and Glu-Tyr(NO₂) (4) (Figure 2), were also synthesized for use as HPLC standards, LC-MS standards, and for fluorescence standard curves (See Supporting Information for details).

For transient kinetic studies, an additional supply of 1 was obtained using solid phase peptide synthesis (SPPS) to provide ample quantities of 1 in a rapid, efficient synthesis (Scheme 2). As described in the Supporting Information, SPPS is the preferred method of preparing 1, due to the commercial availability of suitably protected natural amino acids. However, SPPS is currently not yet feasible for synthesis of 2 or other peptides containing non-natural amino acids due to the large excess of Fmoc-protected amino acid (e.g., (2S)-4,4-difluoroglutamic acid) required for coupling to the resin-based C-terminus of the growing peptide.

Enzyme Assay Development for GH

To facilitate detailed kinetic studies of GH-catalyzed hydrolysis of both non-fluorinated and fluorinated substrates, our recently reported FRET assay for GH has been further refined. The original assay was based on an internally quenched fluorogenic glutamyl-γ-glutamyl peptide with an N-terminal N-methyl-p-aminobenzoyl (N-Me-pAB) moiety as the fluorophore and 3-nitrotyrosine as the C-terminal quenching moiety (23). However, the N-Me-pAB moiety has a significant absorbance (λ_max = 289 nm, ε = 15,050 M⁻¹cm⁻¹, 0.1 N NaOH (29)) near the excitation wavelength (λ_ex = 290)_, leading to a large IFE that limits the use of this substrate to concentrations below 10 μM. In the second-generation fluorogenic substrate (1), the N-Me-p-AB fluorophore has been replaced with an o-aminobenzoyl (Abz) group λ_max = 322 nm, ε = 1,585 M⁻¹cm⁻¹, ethanol (30)) so that absorbance is much less at the excitation wavelength (λ_ex = 320 nm). The use of 1 extends the concentration range (0 – 35 μM) before significant inner filter effects are observed as well as provides better sensitivity with a higher fluorescent signal. The concentration range of 1 and 2 in assays could be further extended to 75 μM by correcting for the IFE as described in Experimental Procedures. A more rigorous method of calibrating the assay to convert the fluorescence units to micromolar product also improved the reproducibility of the assay. Instead of using a single calibration point as is frequently reported for FRET assays, a concentrated aliquot of GH (1 μL, 2.9 μM) was added to each assay (varying concentrations of 1 or 2) after the initial velocity measurement, and the fluorescence was measured after complete hydrolysis. A linear plot of fluorescence versus product concentration (See Supporting Information, Figure S1) provided a reliable calibration factor as the inverse slope of the line (μM/fluorescence units). These improvements resulted in a robust, sensitive, and convenient continuous assay for GH that could now be applied to the detailed kinetic analysis of 1 and 2.

Steady-State Kinetic Comparison of GH-Catalyzed Hydrolysis of Substrates 1 and 2

The steady-state kinetic parameters for GH reactions with 1 and 2 were determined at 25 °C using the fluorescence assay (See Supporting Information, Figure S2). Since the rate of hydrolysis of 2 is significantly slower than that of 1 (vide infra), a higher concentration of GH (~ 9-fold) was used to monitor the reaction of 2 (see Experimental Section). The kinetics data were fit to the Michaelis-Menton equation to provide values of K_m and V_max for 1 and 2, which are summarized in Table S1. Because it was necessary to use a higher concentration of GH in order to obtain reasonable rates of hydrolysis of 2,a comparison of k_cat and k_cat/K_m for the two substrates is more appropriate than V_max and K_m. However, given the observation of an attenuated burst amplitude in the GH-catalyzed hydrolysis of 1 under pre-steady-state conditions ([S] ≈ [GH], vide infra), determination of k_cat under steady-state conditions by this method should be done with caution in this case. With these caveats, the substrate specificity of GH, as indicated by the relative values of k_cat/K_m, showed a 25-fold preference for 1 over 2. These data are consistent with preliminary data obtained with several isopeptide derivatives of methotrexate, 2-amino-10-methylpteroyl (2RS,4RS)-4-fluoroglutamyl-γ-glutamate (AMPte-(4-F)Glu-γ-Glu), (9), and the corresponding derivatives containing either (2S,4S)-4-fluoroglutamate or (2RS)-4,4-fluoroglutamate (12). This difference appears to be mediated mainly by a much higher value of K_m (75 μM) for the fluorinated analog, 2, over that of 1 (K_m = 5.4 μM). The high value of K_m for 2 made it difficult to saturate GH before inner filter effects became too severe, leading to slightly higher errors in the determination of the kinetic constants for 2 compared to 1.

The proposed kinetic scheme for GH, that of a typical protease, is shown in Figure 3. In this scheme, rapid equilibrium association of the enzyme and substrate forms the Michaelis complex (ES), followed by formation of the acyl enzyme (ES′) with concomitant release of the first product P₁ (Glu-Tyr(NO₂) (4)). Separation of the quenching moiety (Tyr(NO₂)) from the fluorophore (Abz) in this step results in increased fluorescence because FRET is distance-dependent. The last step of the kinetic scheme is the hydrolysis of the acyl-enzyme intermediate to release the second product, P₂ (Abz-Glu (3a) or Abz-F₂Glu (3b)). Both the formation of the acyl enzyme and the hydrolysis of the acyl enzyme (to reform free enzyme) proceed through similar tetrahedral intermediates (31). The slower hydrolysis of the fluorinated analog, 2, could be due either to steric effects that result in poor binding to GH or to an electronic effect, perhaps involving the stabilization of one or both tetrahedral intermediates by adjacent fluorine substituents (32).

Figure 3.

Proposed kinetic scheme for GH. ES is the Michaelis complex and ES′ is the acyl enzyme.

Steady-State vs. Pre-Steady-State Kinetics Experiments

From the above steady-state kinetics data alone it is difficult to draw conclusions about the mechanistic differences in the hydrolysis of 1 compared to 2. The values of K_m for 1 and 2 as determined under steady-state conditions differ by almost 15-fold. K_m is defined by the individual rate constants shown in eq 4, and is not equivalent to the dissociation constant (K_S = k_-1/k₁) unless acylation is rate-limiting; i.e., k₂ ≪ k₃, and

$${\text{K}}_{\text{m}}=({\text{k}}_{-1}/{\text{k}}_1)[{\text{k}}_3/({\text{k}}_2+{\text{k}}_3)]$$ (4)

k₋₁ ≫ k₂. Pre-steady-state kinetics experiments can often allow for deconvolution of K_m to provide individual rate constants. These experiments should therefore allow for independent determination of k_cat under conditions where [GH] is identical for experiments involving isopeptides, 1 and 2. In the mechanism shown in Figure 3, k_cat is defined by eq 5. The value of k_cat provides a lower limit for the first-order rate constant

$${\text{k}}_{\text{cat}}={\text{k}}_2{\text{k}}_3/({\text{k}}_2+{\text{k}}_3)$$ (5)

governing all steps, including chemical and conformational changes, in converting substrate to product once the substrate is bound to the enzyme surface (28). It is assumed that k_-2 approaches zero due to the low concentration of P₁ as it diffuses away from the protein, leading to decreased FRET and increased fluorescence. Stopped-flow kinetic experiments were conducted to determine if k₃ might be the rate-limiting step.

Pre-Steady-State Kinetics Experiments for GH and 1

Varying concentrations of GH were mixed with 50 μM 1 (~10 × K_m) in the stopped-flow instrument and the increase in fluorescence was monitored continuously. The results at 25 °C suggested the presence of a burst; however, lowering the temperature to 4 °C allowed the observation of a significant burst (Figure 4A). The burst suggests formation of an intermediate on the reaction pathway, the breakdown or release of which is slow relative to the rate of production of the first observed product. The observed burst curve indicates that k₃, representing hydrolysis of the acyl enzyme intermediate and release of Abz-Glu, is the rate-limiting step of the reaction.⁴ Fitting progress curves to the burst equation (eq 1) provided a rate constant for the initial phase of 72 s⁻¹, independent of [GH] (Figure 4A, [1] = 50 μM, ca 10 K_m). The amplitude of the burst and the steady-state rate, v_ss, at each concentration of enzyme (Figure 4A) were determined and shown to be proportional to the enzyme concentration (Figure 4B). Assuming a full burst, the burst amplitude provides an estimate of the viable active site concentration. However, this was determined to be only about 30% of the GH concentration indicated by Bradford assay (data not shown). Several possible factors that could result in attenuation of the burst amplitude are discussed below.

Figure 4.

(A) Burst observed for rapid mixing of Abz-Glu-γ-Glu-γ-Tyr(NO₂) (1, 50 μM) in assay buffer, pH 6.0, and varying concentrations of GH, in enzyme storage buffer, pH 5.5, 4 °C. Final concentrations were 50 μM 1 and 0.29 – 1.2 μM GH as indicated. Each progress curve is the average of five experimental traces. Only 10% of the data points are shown for clarity and each line represents a fit to the burst equation (eq 1). (B) Dependence of the burst amplitude (○) and steady state velocity (■) on enzyme concentration.

At a fixed concentration of GH (0.86 μM), the kinetics of hydrolysis of 1 (2.5– 50 μM) were monitored by fluorescence and fit to eq 1 (Figure 5A) to obtain the burst rate constants (k) and amplitudes (A), as well as the steady-state velocities (v_ss). At concentrations of 1 less than K_m (e.g., 2.5 μM), a burst was not observed at the concentration of GH used, so this progress curve was only used for fitting the steady-state velocity. This dependence of the burst rate constant on substrate concentration was fit to eq 2 (Figure 5B), which provides values for K_S and (k₂ + k₃), and the dependence of the burst amplitude on substrate concentration was fit to eq 3 (Figure 5C) to provide k₂ and, k₃, the latter by subtraction of k₂ from (k₂ + k₃). The slower, nearly linear phase of the progress curves (Figure 5A) represents the steady-state rate, v_ss, the substrate concentration dependence of which could be fit to the Michaelis-Menten equation (Figure 5D) to provide the steady-state kinetic parameters, K_m and k_cat.

Figure 5.

Transient kinetics with varying Abz-Glu-γ-Glu-γ-Tyr(NO₂), 1, in assay buffer, pH 6.0, mixed with GH, (0.86 μM final concentration after mixing) in enzyme storage buffer, pH 5.5, 4 °C. (A) Burst kinetics observed by fluorescence at > 389 nm upon varying concentration of 1 (indicated on figure). Each progress curve is the average of at least three experimental traces and only 10% of the data are shown for clarity. Data for [1] > 2.5 μM were fit to eq. 1 (solid lines). Data for 1 > 2.5 μM were fit to eq 1. (B) Dependence of burst rate constant on substrate concentration with fit to eq 2. (C) Dependence of burst amplitude on substrate concentration with fit to eq 3. (D) Dependence of steady-state rate on substrate concentration with fit to the Michaelis-Menten equation.

These kinetics data are summarized in Table 1. Fitting the data using a concentration of active GH derived from the burst analysis (0.86 μM) is consistent with our observations, whereas using the GH concentration derived from the Bradford assay (3.0 μM) is not. Only when the lower GH concentration was used was k₂ found to be greater than k₃ in fitting eq 3 (Table 1, entry 1 vs. 2); the presence of a burst predicts that k₂>k₃. However, when the enzyme concentration was set at 3 μM, as indicated by the Bradford assay, the calculated k₂ is less than k₃, which is inconsistent with the observation of the burst that is clearly evident in both Figures 4A and 5A. Although several possible explanations for the attenuated burst amplitude are further elaborated in the Discussion, a definitive basis for the observed low values of the burst amplitude is not clear from these data. However, highly reproducible data have been obtained (Figures 4 and 5), and by varying substrate concentration at fixed [GH] and fitting equations describing how the burst amplitude and burst rate constant vary with substrate concentration (eq 1–3), it was possible to determine the individual rate constants, k₂ and k₃, as well as the dissociation constant, K_S and, in so doing, learn much about the mechanism of the reaction catalyzed by GH.

Pre-Steady-State Kinetics Experiments for GH and 2

To determine if the presence of geminal fluorine atoms in 2 changes the rate-limiting step, transient kinetic studies were carried out by rapidly mixing 2 (final concentration 50 μM) with varying concentrations of GH (0.29 – 1.2 μM). No burst was observed under these conditions (Supporting Information, Figure S3). However, since the substrate concentration was less than K_m (75 μM) and thus non-saturating, it may be difficult to observe a burst. The experiment was repeated with a higher concentration of 2, 150 μM.⁵ At 150 μM 2, a modest burst was observed (Figure 6, [GH] = 0.86 μM). The high value of K_m (75 μM) and limited quantity of 2 available precluded the use of [2] >150 μM for the collection of data to fit eq 1–3, as was described for substrate 1. However, fitting the limited data of Figure 6 to the burst equation (eq 1) provided an amplitude of 0.78 μM, or 26% active GH, which is very similar to the value of 29% active GH determined from the kinetic analysis of GH-catalyzed hydrolysis of 1. At this concentration of substrate, [2] = 150 μM, ~ 2 K_m, the burst rate constant, k_burst, was determined to be 13 s⁻¹.

Figure 6.

Kinetics of Abz-4,4-F₂Glu-γ-Glu-γ-Tyr(NO₂) (2) hydrolysis. Equal volumes of GH in storage buffer (pH 5.5) and 2 in assay buffer (pH 6) were rapidly mixed at 4 °C and the fluorescence monitored. Concentrations after mixing were 0.86 μM GH and 150 μM 2, average traces (n=5) shown.

DISCUSSION

The new continuous fluorescence assay described above provides detailed kinetics data for the hydrolytic reaction catalyzed by γ-glutamyl hydrolase (GH). Previous research to elucidate the kinetic properties of this enzyme has relied on end-point assays (14, 20), primarily via chromatographic separation of the products using high performance liquid chromatography (HPLC) or capillary electrophoresis (CE) (33) (vide supra). In order to obtain more detailed steady-state and transient kinetics data, a continuous spectral assay was desired. GH is specific for cleavage of a Glu-γ-Glu bond, releasing glutamic acid which is neither a fluorophore nor a chromophore. Acylation of the N-terminal Glu of the Glu-γ-Glu isopeptide with a fluorescent group together with addition of an internal quenching moiety at the C-terminus leads to internal quenching of fluorescence due to FRET (24). GH-catalyzed hydrolysis of these substrates releases the quenching moiety, leading to a fluorescent signal.

The first FRET peptide designed for use as a GH substrate contained an N-terminal p-(N-Me)aminobenzoyl (N-Me-p-AB) fluorophore with 3-nitrotyrosine (Tyr(NO₂) as the quenching moiety at the C-terminus (23). Although an effective substrate, the hydrolysis of which could be monitored continuously by fluorescence, its utility is limited due to attenuated fluorescence of the reaction products. This is a result of the inner filter effect (IFE) (vide supra). Replacement of the N-Me-p-AB moiety with o-aminobenzoyl (Abz) at the N-terminus and retention of Tyr(NO₂) at the C-terminus minimized the IFE and provided two excellent GH substrates, 1 and its 4,4-difluoroglutamate analog, 2, and these have been used to determine the reaction kinetics of GH-catalyzed isopeptide hydrolysis.

While the use of peptide substrates containing fluorophores that are quenched due to intramolecular FRET is common for the high-throughput screening of compound libraries (34, 35), complete evaluation of protease reaction kinetics using such substrates is not frequently found in the biochemical literature (36). This is very likely due to problems caused by quenching of fluorescence (IFE). Recent studies have demonstrated that use of FRET peptides to determine standard enzyme kinetics data (K_m, k_cat, k_cat/K_m) can lead to incorrect values and erroneous conclusions (37). In such cases, IFE-based errors can be minimized by the use of alternate methods such as fast-flow HPLC (37) or dilution of the fluorogenic peptide substrate with a non-fluorescent analog (38). In the current research, a proper analysis of the steady-state and kinetics was possible with isopeptide 1 without resorting to these methods (Supporting Information, Figure S2A). However, because the K_m for the fluorinated substrate, 2, is so large, it was not possible to saturate the enzyme (Supporting Information, Figure S2B) due to the IFE at high concentrations of 2. The use of fast-flow HPLC was not pursued and the scarcity of (2S)4,4-difluoroglutamic acid (25) precluded its incorporation into a non-fluorogenic peptide substrate for use in combination with 2.

A fit of the data shown in Figure S2 (Supporting Information) to the Michaelis-Menten equation provided steady-state kinetic constants for γ-glutamyl peptides 1 and 2 (Table S1). The dramatic decrease in the relative values of k_cat/K_m observed rates for GH-catalyzed hydrolysis of 2 vs. 1 (k_cat/K_m )_rel 2:1 = ca. 0.04) is in contrast to the known inductive effects of fluorine on reactions at adjacent carbon. In the case of regioselective nucleophilic attack at the γ-ester of 4,4-difluoroglutamic acid esters (25, 39), the electron-withdrawing properties of fluorine leads to a large increase in the rate of nucleophilic attack at the adjacent carbon. Interestingly, incorporation of fluorine adjacent to the scissile isopeptide bond results in a significant (~ 15-fold) increase in K_m. In order to probe more deeply into the basis for the increase in K_m, investigation of GH-catalyzed hydrolysis under non-steady-state conditions was carried out using stopped-flow techniques

As noted above (eq. 4), K_m is comprised of a dissociation constant, K_s = k_-1/k₁, modified by the rates of formation (k₂) and breakdown (k₃) of an intermediate, in this case the acyl enzyme. Experiments to determine the values of K_s, k₂, and k₃ were carried out using the stopped-flow instrument. At a fixed concentration of 1 ([1] = ~ 10 K_m), GH-catalyzed hydrolysis of the isopeptide displays burst kinetics (Figure 4A), and it can be shown that both the burst amplitude and the steady-state rate are directly proportional to the enzyme concentration (Figure 4B). At a fixed concentration of [GH], hydrolysis of 1 displays burst kinetics (Figure 5A) with the burst rate, k_burst (Figure 5B), burst amplitude (Figure 5C) and v_ss (Figure 5D) all dependent on the concentration of 1 as predicted. Fitting the data of these figures to eq. 2 (Figure 5B), eq. 3 (Figure 5C), and the Michaelis-Menten equation (Figure 5D) yielded values of K_s, k₂, k₃, K_m, and k_cat (Table 1). The mechanism shown in Figure 3 indicates the reaction steps associated with K_s (k_-1/k₁), k₂ and k₃. Intramolecular quenching of fluorescence in 1 is no longer possible on separation of the quenching moiety, Glu-γ-Tyr(NO₂) from the fluorescent donor, Abz-Glu, following GH-catalyzed hydrolysis. Since the kinetics of product formation as measured by fluorescence exhibits a rapid burst phase followed by a slower steady-state phase, it follows that the step in which Glu-γ-Tyr(NO₂) is released, k₂, must be faster than some subsequent step. As shown in Table 1, the data of Figures 5A–D were fitted in two ways, one of which used the concentration of GH determined by the Bradford assay (3 μM) and a second which used the concentration of GH indicated by the burst amplitude (0.86 μM). Only when [GH] approached 0.86 μM (entries 2 (exponential phase) and 4 (global fit)) were values of k₂ and k₃ obtained that are consistent with the observed burst kinetics; i.e., k₂ > k₃. The observation of a burst phase in the enzyme-catalyzed hydrolysis of a peptide bond is unusual. Although the first burst kinetics were reported for a protease (40), the substrate used in that classic paper was a nitrophenyl ester. In contrast, substrates 1 and 2 incorporate an isopeptide, Glu-γ-Glu, and, to the best of our knowledge, a pre-steady-state kinetics investigation of enzyme-catalyzed hydrolysis of an isopeptide bond with this structural motif has not been reported (vide infra).

It is interesting to note that a more significant burst was observed at 4 °C than at 25 °C (vide supra), the temperature at which the steady-state kinetics data (Table S1, Figure S2) were obtained. Since an increase in fluorescence is associated with release of P₁ (Figure 3, k₂), the observed burst at 4 °C suggests, at first glance, that k₂ increases with a decrease in temperature; however, this would be incompatible with the Arrhenius equation, and is therefore unlikely. A more plausible explanation is that k₂ only decreases slightly at the lower temperature while k₃ decreases significantly, so that k₂ ≫ k₃, leading to the observed burst of fluorescence. The rate constant, k₃, is likely more complex than indicated in Figure 3⁴ and could include a product release step and/or protein rearrangement to properly position the active site Cys110 for attack on the adjacent scissile bond if the enzyme acts processively. Large-scale domain motion of GH to accommodate repositioning of the adjacent scissile bond and/or active site residues, included in k₃, would be expected to be much more sensitive to temperature effects than the simple acylation step represented by k₂.

The discrepancy between the apparent concentration of GH as determined by the Bradford assay and that derived from the kinetics data should be addressed. Both the human and rat GH isozymes, expressed and purified as described in the Supporting Information, have been shown to be homogeneous by analytical ultracentrifugation (41) and, in the case of human GH, has been crystallized to provide a structure of the substrate-free enzyme (42). The simplest explanation of the kinetics data is that only ca. 30% of the enzyme is active.⁶ However, there are other possibilities. The x-ray crystal structure of hGH indicated that the enzyme has a homodimer structure with a β-mercaptoethanol molecule covalently linked to Cys110 at only one of the two active sites in the dimer (43). In addition, it has been shown recently by analytical ultracentrifugation that GH is a non-dissociating homodimer in solution (20, 41). Tyr36 is part of the monomer-monomer interface and has shown to be required for catalysis (T. J. Ryan, unpublished results). Thus, it is possible that in hGH, and by extension rGH, only one of the active sites is involved in the burst phase of catalysis; i.e., half-of-the-sites reactivity. The second site burst will be masked by the slower subsequent steps of catalysis. If true, the concentration of GH involved in the burst is predicted to be 50% of the total protein, a figure closer agreement to the value of ca. 30% determined from the burst amplitude data, but nevertheless suggesting some instability of the protein during handling. It is, of course, also possible that oligomerization to form larger inactive GH aggregates could lead to decreased catalytic activity. However, we have no evidence for the formation of aggregates under the conditions used in the experiments reported herein.

The pre-steady-state kinetics of GH-catalyzed hydrolysis of the 4,4-difluoroglutamate-containing isopeptide, 2, was also investigated using stopped-flow techniques. Due to the IFE, the maximum concentration of 2 used in the acquisition of kinetics data was 150 μM, a concentration of substrate only ~ 2 K_m. Therefore, a thorough pre-steady-state analysis was not possible. Only when [2] = 150 μM (~ 2 K_m) was a modest burst observed (Figure 6). The limited data obtained from Figure 6 indicate a >5-fold decrease in k_burst for 2 vs. 1, even at subsaturating concentration (~2 Km) of the fluorinated substrate. Although modest in magnitude, the effect of placing two fluorine atoms adjacent to the scissile bond in a Glu-γ-Glu isopeptide is to retard the burst phase, presumably associated with formation of the acyl enzyme.⁷

Product analysis by LC-MS demonstrated that the reaction proceeds with cleavage of the expected Glu-γ-Glu isopeptide bond, leading to Glu-γ-Tyr(NO₂) and Abz-Glu or Abz-(4,4-F₂)Glu from 1 and 2, respectively (Figure 2). The internally quenched derivatives of Glu-γ-Glu, 1 and 2, were chosen for study to preclude the possibility of multiple sites of cleavage, as is observed for longer chain poly-γ-glutamates(20, 44). Thus, the kinetics data reported in this paper are for GH-catalyzed hydrolysis of a single bond in a Glu-γ-Glu isopeptide.

Further analysis of the data presented in Table 1 involving a global fit to the transient kinetics data (Figure 5) allows for evaluation of the internal consistency of these data. Thus, a value of k_cat can be calculated from k₂ and k₃ (eq. 4), leading to a value of 12.7 s⁻¹, and a value of K_m can be calculated from K_s, k₂, and k₃ (eq. 5), leading to a value of 6.1 μM. These calculated values of k_cat and K_m for GH-catalyzed hydrolysis of 1 are in reasonable agreement with the steady-state component of the transient kinetics data (Table 1, entry 4).

As noted above, the observation of a burst phase in the GH-catalyzed hydrolysis of 1, a peptide, albeit an isopeptide, Glu-γ-Glu, is unusual. In terms of substrate structure and site of hydrolysis, the synthetase and hydrolase involved in the reversible post-translational modification of tubulin (45, 46) are most closely related to GH. Although the ATP-dependent ligase, tubulin polyglutamylase, that catalyzes the addition of multiple glutamates has been partially purified and its catalytic properties described (47, 48), the putative hydrolase activity by which the glutamates are removed (49) has not been characterized. Other enzymes that catalyze nucleophilic attack at the γ-position of substituted glutamine include γ-glutamyl transpeptidase (EC 2.3.2.2) (50) and transglutaminase (EC 2.3.2.13) (51). In the former reaction, acylation occurs via attack of an active site Cys or Thr on the γ-carboxamido group of a γ-glutamyl peptide followed by reaction of the acyl enzyme with an amino acid to release the peptide and form a new γ-glutamyl dipeptide. In the latter case, acylation involves attack of an active site cysteine on the γ-carboxamido group of a protein-bound glutamine followed by reaction of the acyl enzyme with an alkyl amine (e.g., protein-bound lysine, polyamines) to release NH₃ and form a protein-bound γ-substituted glutamine. The catalytic mechanisms of both have been studied in detail but with non-peptide, chromogenic leaving groups (52) or non-physiological amine acceptors (53, 54), respectively. Thus, the burst kinetics observed in several cases by these investigators involve substrates with better leaving groups (e.g., p-nitroaniline, aminocoumarins) than an amino acid or peptide such as Glu-γ-Tyr(NO₂) released in the GH-catalyzed reaction studied in this work (Figure 5).

Ubiquitin C-terminal hydrolases (UCH) (55, 56) are cysteine peptidases, family C12, that act on an isopeptide in which a C-terminal glycine is linked to the ε-amino group of an internal lysine residue of ubiquitin (57), In extensive mechanistic studies on a ubiquitin C-terminal hydrolase, UCH-L1, Stein and co-workers suggest that the enzyme exhibits a “mechanistic plasticity” where slightly different mechanisms are involved in substrate hydrolysis vs. enzyme inactivation by a peptidyl aldehyde (56). This suggestion is supported by a crystal structure of apo-UCH-L1 (58). It is proposed that binding of the peptide substrate involves displacement of a bound water to allow for proper alignment of the active site His-Cys diad to form the S⁻/ImH⁺ ion pair involved in formation of the acyl enzyme. Similarly, in GH, positioning of Cys110 and His220 to form the requisite ion pair presumably required for GH-catalyzed hydrolysis may involve a significant protein rearrangement, resulting in placing some or all of the substrate in a hydrophobic cleft. Unfortunately, data available from the published crystal structure of hGH (42) do not address this possibility because the structure obtained is of the substrate-free enzyme. However, it is possible that GH-catalyzed hydrolysis of oligo-γ-glutamates involves general-acid catalysis by the α-CO₂H of a Glu residue on the carbonyl of the adjacent scissile γ-glutamyl peptide bond. This substrate-assisted catalysis could thereby facilitate attack by Cys110 to form the acyl enzyme intermediate (Figure 7). Substrate-assisted catalysis involving the α-CO₂H could explain the observed decrease in k_burst for 2 vs. 1; i.e., electron withdrawal by the CF₂ adjacent to the carbonyl group would render that group less susceptible to general-acid catalysis

Finally, it should be noted that GH catalyzes the hydrolysis of a polymeric isopeptide and as such may operate in a processive mode as has been demonstrated in glycohydrolases such as amylase (“degree of multiple attack.”) (59). Although the research reported herein involved a diglutamate substrate containing only a single cleavage site, GH can hydrolyze multiple peptide bonds in longer chain poly-γ-glutamates. (20) Rearrangement of the initially formed E-P complex to position the next scissile bond for attack by Cys110, as required for a processive mechanism, could be the rate limiting step. If a rearrangement of E-P slows product release, it could lead to the observed burst kinetics. As noted above, the temperature dependence of k₂ vs. k₃, suggested by the observation of a burst of fluorescence at 4 °C but not at 25 °C, is consistent with slow protein rearrangement associated with a processive hydrolytic reaction at sequential peptide bonds along the poly-γ-glutamate chain.

In summary, the hydrolysis of two internally quenched derivatives of the isopeptide, Glu-γ-Glu, by recombinant rat GH has been investigated via a new continuous fluorescence assay. Comparison of the steady-state kinetics data obtained for a substrate with two fluorines adjacent to the scissile bond, 2, vs. the non-fluorinated derivative, 1, revealed that the surprisingly attenuated rate observed for 2 is due primarily to a 15-fold increase in K_m. Pre-steady-state kinetics data were obtained by stopped-flow fluorescence methods, and a discreet burst phase was observed for the GH-catalyzed hydrolysis of 1, leading to the conclusion that formation of the putative acyl enzyme occurs with a rate ~ 10-fold faster than the rate of deacylation and any subsequent step(s). Although the transient kinetics data obtained for 2 were more limited due to the combined effects of a high K_m and the IFE, a modest burst was observed at sub-saturating concentrations of the substrate, thus indicating that the two adjacent fluorine atoms do not affect the relative rates of formation or breakdown of the acyl enzyme intermediate. This detailed kinetic analysis of GH-catalyzed hydrolysis of 1 has documented an interesting and perhaps unique example of burst kinetics in isopeptide hydrolysis. It has also suggested possible half-the-sites reactivity and a processive mechanism for GH-catalyzed hydrolysis of oligomeric γ-glutamyl substrates, both of which are subjects for future research.

Supplementary Material

1File003. SUPPORTING INFORMATION AVAILABLE.

Experimental details on the synthesis of the GH substrate, N-Abz-Glu-γ-Glu-γ-Tyr(NO₂) (1) by both solution and solid phase methods, two putative products, N-Abz-Glu (3a) and Glu-γ–Tyr(NO₂) (4), and a model compound, N-Ac-Tyr(NO₂) (S3), for use in determination of the extinction coefficient of the intact peptide, 1 Figures showing the linear dependence of product fluorescence on the concentration of the quenched substrate, 1 (Figure S1), steady-state kinetics of the GH-catalyzed hydrolysis of 1 (Figure S2A) and 2 (Figure S2B), and transient kinetics of the GH-catalyzed hydrolysis of 2 at [2] < K_m (Figure S3). This material is available free of charge via the Internet at http://pubs.acs.org.

Figure 7.

Abbreviations and Textual Footnotes

Abz

o-aminobenzoyl

AMPteGlu

2-deoxy-2-amino-10-methyl-pteroyl

(2′RS

4′RS), indicates racemic carbon at the 2′- and 4′-position

AMPte-(4-FGlu)

2-deoxy-2-amino-10-methyl-pteroyl(4-fluoro)glutamic acid

BME

β-mercaptoethanol

DIPEA

diisopropylethylamine

DMAP

dimethylaminopyridine

DMF

dimethylformamide

EDC

N-(3-dimethylaminopropyl)-N′-ethylcarbodiimide

EDTA

ethylenediaminetetraacetic acid

EtN₃

triethylamine

Fmoc

fluorenylmethyloxycarbonyl

FRET

fluorescence resonance energy transfer

GH

γ-glutamyl hydrolase

HETCOR

heteronuclear correlation spectroscopy

HBTU

N,N,N′,N′-Tetramethyl-O-(1H-benzotriazol-1-yl)uronium hexafluorophosphate

HOBt

1-hydroxybenzotriazole

IFE

inner filter effect

N-Me-pAB

N-methyl-p-aminobenzoyl

SPPS

solid phase peptide synthesis

TCEP

Tris(2-carboxyethyl)phosphine hydrochloride

TEAB

triethylammonium bicarbonate

TFA

trifluoroacetic acid

Tyr(NO₂)

3-nitrotyrosine