Contribution of Lys276 to the conformational flexibility of the active site of glutamate decarboxylase from Escherichia coli Angela Tramonti 1 , Robert A. John 2 , Francesco Bossa 1 and Daniela De Biase 1 1 Dipartimento di Scienze Biochimiche ‘A. Rossi Fanelli’ and Centro di Studio sulla Biologia Molecolare del CNR, Rome, Italy; 2 Cardiff School of Biosciences, Cardiff, UK Glutamate decarboxylase is a pyridoxal 5¢-phosphate- dependent enzyme responsible for the irreversible a-decar- boxylation of glutamate to yield 4-aminobutyrate. In Escherichia coli, as well as in other pathogenic and non- pathogenic enteric bacteria, this enzyme is a structural component of the glutamate-based acid resistance system responsible for cell survival in extremely acidic conditions (pH < 2.5). The contribution of the active-site lysine residue (Lys276) to the catalytic mechanism of E. coli glutamate decarboxylase has been determined. Mutation of Lys276 into alanine or histidine causes alterations in the conformational properties of the protein, which becomes less flexible and more stable. The purified mutants contain very little (K276A) or no (K276H) cofactor at all. However, apoenzyme preparations can be reconstituted with a full complement of coenzyme, which binds tightly but slowly. The observed spectral changes suggest that the cofactor is present at the active site in its hydrated form. Binding of glutamate, as detected by external aldimine formation, occurs at a very slow rate, 400-fold less than that of the reaction between glutamate and pyridoxal 5¢-phosphate in solution. Both Lys276 mutants are unable to decarboxylate the substrate, thus preventing detailed investigation of the role of this residue on the catalytic mechanism. Several lines of evidence show that mutation of Lys276 makes the protein less flexible and its active site less accessible to substrate and cofactor. Keywords: glutamate decarboxylase; pyridoxal 5¢-phos- phate; active-site lysine; site-directed mutagenesis; Escherichia coli. In all pyridoxal 5¢ phosphate (PLP)-dependent enzymes studied so far, the e-amino group of a conserved lysine residue at the active site [1] binds the cofactor as a Schiff’s base. It has been suggested that the formation of an internal aldimine between the coenzyme and a primordial apoenzyme occurred early in the evolution of PLP-enzymes because, before becoming catalytically advantageous, it prevented rapid loss of PLP which was precious because of its ability to catalyse a number of reactions on a wide variety of biosubstrates by itself [2]. Site-directed mutagenesis supports the proposal that participation of this lysine residue in an internal aldimine with the cofactor also accelerates formation of the external aldimine between the PLP 4¢-aldehyde and substrate amino groups because transaldimination is more rapid than de novo Schiff’s base formation [3–6]. It also facilitates the proton transfers essential to many B6-dependent reactions [3–10]. In the amino acid decar- boxylases so far investigated, the corresponding lysine appears not to be involved in reprotonation after decarboxylation, but mainly to play a role in the initial transaldimination, in proper positioning of the a-carb- oxylate for decarboxylation and in product release [11,12]. Bacterial glutamate decarboxylase (Gad, E.C. 4.1.1.15) is one of the structural components of the glutamate-based acid resistance system, responsible for acid survival of enteric pathogens, such as Escherichia coli, Shigella flexneri and Listeria monocytogenes [13–15], and of other nonpatho- genic bacteria [16,17]. E. coli synthesizes two Gad isoforms, GadA and GadB, 98% identical in amino acid sequence and biochemically indistinguishable [18,19]. Gad catalyses the irreversible a-decarboxylation of L -glutamate to yield 4- aminobutyrate and CO 2 . It has been suggested that in this enzyme the active-site lysine is involved in the protonation of the quinonoid intermediate at C-4¢ during the abortive decarboxylation–transamination reaction, while a histidine has been proposed as the residue responsible for the protonation at C-a which occurs during the physiological decarboxylation reaction [20]. Site-directed mutagenesis established that His167 and His275, likely candidates as proton donors, are not responsible for the reprotonation after CO 2 elimination [21]. The present work has been undertaken with the aim of determining the contribution of the active-site lysine residue (Lys276) to coenzyme binding and to stages in the reaction catalysed by E. coli Gad. Correspondence to D. De Biase, Dipartimento di Scienze Biochimiche ÔA. Rossi FanelliÕ, Piazzale Aldo Moro, 5-00185 Roma, Italy. Fax: + 39 06 49917566, Tel.: + 39 06 49917692, E-mail: debiase@caspur.it Abbreviations: Gad, glutamate decarboxylase; PLP, pyridoxal 5¢-phosphate. Enzymes: glutamate decarboxylase (P28302) (E.C.4.1.1.15). Note: a web site is available at http://w3.uniroma1.it/bio_chem/ homein.html (Received 21 May 2002, revised 24 July 2002, accepted 26 July 2002) Eur. J. Biochem. 269, 4913–4920 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03149.x MATERIALS AND METHODS Analytical reagents Vent polymerase was from New England Biolabs. Restric- tion enzymes, T4 DNA ligase and the agarose gel DNA extraction kit were from Roche. The T7 sequencing kit, DEAE-SepharoseÒ and Sephadex G-25Ò were from Phar- macia. [a- 35 S]dATP (1000 CiÆmmol )1 )wasfromNew England Nuclear. Ingredients for bacterial growth were from Difco. Oligonucleotides were from Genenco. Gua- nidineÆHCl, 2,2,2-trifluoroethylamine, aminoacetonitrile bisulfate and Gabase were from Sigma. Other chemicals were from Merck. Site-directed mutagenesis Site-directed mutagenesis was performed by overlap exten- sion polymerase chain reactions [22], following the proce- dure described in Tramonti et al.[21].Mutagenicprimers were 5¢-GGCCATGCATTCGGTCTG-3¢,fortheGadB- K276A mutant, 5¢-GGCCATCACTTCGGTCTG-3¢,for the GadB-K276H mutant, and their complementary sequences. Fragments EcoRV/HindIII, generated by digest- ing the amplicons from the second polymerase chain reaction, were subcloned into pQgadB [19]. The newly inserted fragments of plasmid pQgadBK276A and pQgadBK276H were sequenced on both strands. Purification of mutant forms of Gad Expression and purification of mutant enzymes were as described for wild-type enzyme [19]. The E. coli strain JM109(pREP4), known to produce low levels of endo- genous GadA/B was used as host [19]. Preparations of the mutant enzymes were treated with NaBH 3 CN to inactivate any wild-type enzyme present. Calorimetric and spectroscopic analyses Thermal unfolding of GadB-K276A and GadB-K276H (1.5–2.0 mgÆmL )1 ) was analyzed under nitrogen pressure on a MicroCal MC-2D differential scanning calorimeter (MicroCal, Inc., Northampton, MA, USA). Results were corrected for instrumental baseline and normalized for protein concentration. Absorption spectra were measured on a Hewlett-Packard model 8452 diode-array spectrophotometer. CD spectra were recorded as the average of three scans on a Jasco 710 spectropolarimeter equipped with a DP520 processor at 25 °C. Fluorescence spectroscopy was performed with a LS50B fluorimeter (Perkin Elmer) at the excitation wave- length of 295 nm. Curve fitting and statistical analysis were carried out using the data manipulation software SCIENTIST (Micromath,SaltLakeCity,UT).ThePLPcontentofall the enzyme preparations was determined by treating the protein with 0.1 M NaOH and measuring absorbance at 388 nm (e 388 ¼ 6550 LÆmol )1 Æcm )1 [23]). The pH-depend- ent absorbance variation of wild-type and mutant enzymes was analyzed using the following equation: Abs H n E À Abs E Abs À Abs E À 1 ¼ 10 ÀnpK 10 ÀnpH ð1Þ where Abs HnE and Abs E are the absorbances of completely protonated and unprotonated forms of enzyme, K is the intrinsic dissociation constant and n is the number of protons involved in titration. The change in absorbance observed in the reaction of GadB-K276A and GadB-K276H mutants with glutamate was analyzed with Eqn (2) which describes two consecutive irreversible reactions of the type A ! k 1 B ! k 2 C. b ¼ a 0 Â k 1 k 2 À k 1 Âðe Àk 1 Ât À e Àk 2 Ât Þð2Þ In the above equation, b is concentration of the 412-nm absorbing species (B) and a 0 is the concentration of species Aattime0. Gad activity assay Enzyme activity was assayed by quantitating the reaction product, 4-aminobutyrate, by HPLC [24] or using Gabase, a commercial preparation containing 4-aminobutyrate aminotransferase and succinic semialdehyde dehydrogen- ase, as previously described [19]. On some occasions, enzyme activity was assayed using the pH indicator bromocresol green (0.02% w/v) as described by De Biase et al. [25]. RESULTS Physical properties of mutant enzymes Yields of the mutant enzymes, GadB-K276A and GadB- K276H, after the standard purification, i.e. in absence of added PLP, were 50 mgÆL )1 of bacterial culture, as for wild-type GadB [19]. The mutant forms were stable for several months at 4 °C. As judged by CD spectroscopy in the far-UV region, mutations did not affect the overall protein conformation. The transition temperature of reconstituted GadB-K276A was 62.3 °C, suggesting that this mutant enzyme adopts a significantly more stable conformation than wild-type GadB (51 °C). The transi- tion temperature of GadB-K276H (55.3 °C) was only slightly higher than that of the wild-type enzyme (Fig. 1). In support of the above observation, limited proteolysis by trypsin showed that GadB-K276A is more resistant to proteolytic degradation than the wild-type enzyme (data not shown). Spectral properties of GadB-K276A and -K276H mutants Absorption spectra of purified GadB-K276A showed maxima at 280 and 328 nm (Fig. 2A). However, the specific absorbance at 328 nm due to the cofactor was significantly lower than that of the wild-type enzyme and, correspond- ingly, the amount of PLP released by NaOH treatment was only 10% of that expected for a fully saturated holoenzyme. Nevertheless, the small amount of cofactor present was not displaced by either gel filtration in 0.5 M KH 2 PO 4 ,or incubation with cysteine or dialysis in 2 M guanidineÆHCl. Complete removal of the cofactor was achieved by overnight dialysis against 1 M KH 2 PO 4 ,pH4.2,withthe resulting apoenzyme having the spectrum shown in Fig. 2A, inset. Purification of GadB-K276H following the standard 4914 A. Tramonti et al. (Eur. J. Biochem. 269) Ó FEBS 2002 protocol yielded 100% apoenzyme. Unlike the wild-type apoenzyme, which precipitates instantaneously upon coen- zyme removal, both GadB-K276A and GadB-K276H apoenzymes remained stable for many weeks in 0.1 M sodium acetate, pH 4.6. For both mutant forms, the holoenzyme was regenerated by treating the apoenzyme (150 l M ) overnight with a fivefold molar excess of PLP (750 l M ). Unbound cofactor was removed by extensive dialysis against 0.1 M sodium acetate, pH 4.6. The recon- stituted mutant enzyme (Fig. 2A) contained one molecule of PLP per monomer, as judged by NaOH treatment. The absorption spectrum above 320 nm fitted well to the sum of two log normal curves having k max values of 330 nm and 388 nm (Fig. 2B). The great majority of the coenzyme was present as 330 nm-absorbing chromophore, the corres- ponding peak being broader, but less intense, in the GadB- K276H mutant enzyme (Fig. 2B). Continuous monitoring of the absorbance changes associated with reconstitution of GadB-K276A indicated that the absorbance decrease at 388 nm (free PLP) and the increase at 330 nm were biphasic (data not shown) and the curve fitted well to the sum of two exponentials (k 1 ¼ 0.58 ± 0.001 min )1 ; k 2 ¼ 0.059 ± 0.002 min )1 ) with the more rapid phase accounting for 70% of the reaction. Treatment of both reconstituted mutants with NaCNBH 3 did not affect the spectra, demonstrating that PLP is bound as the free aldehyde. Figure 2(C) shows CD spectra of wild-type and GadB mutants in the 300–500 nm range, where the chromophore of all enzymic forms absorbs maximally. Notably, the GadB-K276A mutant produced a much smaller CD signal than the GadB-K276H mutant, despite the lower absorb- ance of the latter (Fig. 2B). E. coli Gad undergoes well-established changes in activity and in the absorption spectrum of the cofactor depending on pH (Fig. 3A) [26]. At pH values higher than 5.3, the enzyme absorbs maximally at 340 nm and is inactive, whereas at lower pH values, the enzyme absorbs maximally at 420 nm. The change in activity parallels the absorbance Fig. 2. Absorption and CD spectra of GadB Lys276 mutants. (A) Absorption spectra of GadB-K276A mutant. The absorption spectra of GadB-K276A (20 l M ) as it is purified under standard conditions (dotted line), in the apoenzymatic form (solid line) and after its reconstitution with PLP (dashed line) were determined in 0.1 M sodium acetate, pH 4.6, containing 0.1 m M dithiothreitol. (B) Analysis of absorption spectra of GadB-K276A and GadB-K276H mutants. The solid lines are those of best fit to the sum of two log normal curves [32] having k max values of 330 nm and 388 nm. Only one in three of the data points collect for the GadB-K276A (d) and GadB-K276H (j) absorption spectra is shown. (C) CD spectra of wild-type and mutant enzymes. The CD spectra of wild-type GadB (solid line), and of GadB- K276A (dotted line) and GadB-K276H (dashed line) mutants, each at a concentration of 184 l M , were determined in 50 m M sodium acetate, pH 4.6, containing 0.1 m M dithiothreitol. Fig. 1. Differential scanning calorimetry of wild-type GadB and active- site lysine mutants. Thermal denaturation profiles of GadB wild-type (solid line), of GadB-K276A mutant (dashed line) and GadB-K276H mutant (dotted line). Protein samples (1.5–2.0 mgÆmL )1 )werein0.1 M sodium acetate, pH 3.6, containing 0.1 m M dithiothreitol. Ó FEBS 2002 Role of Lys276 in E. coli glutamate decarboxylase (Eur. J. Biochem. 269) 4915 change and it is clear from the pH profile that multiple protons are involved in the transition (Fig. 3A, inset) [26]. In both active-site lysine mutants, changing the pH from 3.6 to 6.4 induced an increase of the species absorbing at 388 nm and a decrease of that absorbing at 328 nm (Fig. 3B). The systematic changes in the spectrum of the mutant enzymes showed the proportion of the 388 nm chromophore increasing towards a limiting value. Again, as in the wild- type enzyme (Fig. 3A), the abrupt changes in absorbance with increasing pH did not fit well to a single ionization event and the data from both active site-lysine mutants fitted adequately to a model that required the simultaneous loss of 4–6 protons (Fig. 3B, inset). Experiments in absorption and fluorescence spectroscopy, in the presence of increasing concentrations of guani- dineÆHCl (0–6 M ), were conducted with both wild-type and the mutant enzymes. In both mutants, guanidineÆHCl in the range 0–2 M induced an absorbance change characterized by an increase at 388 nm and a decrease at 328 nm (Fig. 3C). The same behavior was also observed when sodium chloride was added (data not shown). In the same concentration range of guanidineÆHCl and sodium chloride, the absorbance spectrum of the wild-type enzyme remained unaffected. When excited at 295 nm in the absence of guanidineÆHCl, the reconstituted mutant enzymes exhibited two fluores- cence emission maxima (Fig. 4A). The first, at 332 nm, also present in the wild-type (Fig. 4B), is due to the intrinsic fluorescence of the protein. It is likely that the second, at 380 nm, is due to energy transfer to the 330-nm absorbing form of PLP at the active site. At 2 M guanidineÆHCl the emission spectrum of both mutants showed exclusively a peak at 332 nm, because the PLP in the active site has been converted into a form mainly absorbing at 388 nm. In the range 2–5 M guanidineÆHCl the change in fluorescence emission spectra indicated that the unfolding profiles of wild-type and mutant enzymes are superimposable, with the transition point (50% unfolding) centered at 3.4 M guani- dineÆHCl. Upon unfolding, a blue-shifted emission maxi- mum at 360 nm in both wild-type and mutant enzymes was observed (Fig. 4). Reaction with glutamate Addition of 20 m M glutamate to GadB-K276A produced an increase in absorbance at 412 nm and a decrease at 328 nm each with the same half-time of approximately 2 h (Fig. 5A). The change was characterized by an isosbestic point at 342 nm. The 412 nm contribution was completely abolished by adding NaCNBH 3 , a reagent known to reduce exclusively the protonated Schiff bases. After 7 h, an additional slow spectral change occurred which was com- plete within 30 h. This spectral change is characterized by a decrease at 412 nm and an increase at 340 nm, with an isosbestic point at 375 nm (Fig. 5B). The change observed at 412 nm conformed to an equation describing two consecutive irreversible reactions (see Materials and meth- ods, Eqn 2). Increasing the glutamate concentration produced a linear increase in the value observed for k 1 whereas k 2 did not change significantly (0.07 ± 0.01 h )1 ). No 4-aminobutyrate was detected at the end of the reaction although the method used was sensitive enough to detect this compound at 10% of the enzyme concentration. Fig. 3. Effect of pH and guanidineÆHCl on the absorbance spectra. (A) Absorption spectra of wild-type GadB (11.4 l M ) were determined in 0.1 M sodium acetate in the pH range 3.5–6.2. Only relevant spectra are shown. In the inset, the pH variation at 420 nm is represented. The solid line is that of best fit to Eqn (1) (Materials and methods), with pK ¼ 5.292 ± 0.007, Abs E ¼ 0.0069 ± 0.0008, Abs HnE ¼ 0.1058 ± 0.0007 and n ¼ 5.1 ± 0.4. (B) Absorption spectra of GadB-K276H (10.3 l M ) were determined as above. In the inset, the pH variation at 388nmisrepresented.ThesolidlineisthatofbestfittoEqn(1) (Materials and methods), with pK ¼ 5.60 ± 0.01, Abs E ¼ 0.032 ± 0.001, Abs HnE ¼ 0.0060 ± 0.0006 and n ¼ 8.4 ± 4.6. (C) Absorption spectra of GadB-K276H mutant (19 l M ) measured in the presence of 0, 0.1, 0.2, 0.4, 0.6, 1 and 2 M guanidineÆHCl in 0.1 M sodium acetate, pH 4.6. The pH- and guanidine-dependent absorbance changes in the GadB-K276A mutant were identical with those in GadB-K276H mutant, and therefore the data are omitted. 4916 A. Tramonti et al. (Eur. J. Biochem. 269) Ó FEBS 2002 Moreover, no pH increase was detected during the reaction with glutamate when using the pH indicator bromocresol green in an unbuffered solution, thus indicating that there was no consumption of protons. Treatment of the reaction mixture with 0.2 M NaOH at the times 0, 5 and 20 h released the full complement of cofactor as PLP (detected and measured by its 388 nm absorbance). Reaction of GadB-K276H with 20 m M glutamate resulted in a spectral change similar to that occurring in the GadB- K276A mutant (data not shown). The only difference observed between the two mutants was the amplitude of the change in absorbance at 412 nm, which at 20 l M protein was much smaller in GadB-K276H (total absorbance change of 0.03) than in GadB-K276A (total absorbance change of 0.12; Fig. 5A). As observed for the alanine mutant, the histidine mutant did not produce 4-aminobutyrate. In the active-site lysine mutants of aspartate aminotrans- ferase [7], tryptophan synthase [4], D -amino acid transam- inase [5] and alanine racemase [27] it was observed that exogenous amines can partially or totally substitute for the catalytic role of the active-site lysine. In order to study the effect of exogenous amines on GadB-K276A, 2,2, 2-trifluoroethylamine (pK a ¼ 5.7) [7] and aminoacetonitrile bisulfate (pK a ¼ 5.3) [7] were added to the enzyme in presence of sodium glutamate. When 1 M 2,2,2-trifluoroethylamine or 0.2 M amino- acetonitrile bisulfate were included in the reaction mixture containing 20 l M GadB-K276A and 20 m M sodium glutamate, the enzyme underwent spectral changes identical to those already described, but the increase in absorbance at 412 nm occurred approximately six times faster. The devel- opment of turbidity however, prevented analysis of the later phases of the reaction. Even in the presence of exogenous amines 4-aminobutyrate production was undetectable (data not shown). When both mutant enzymes were incubated with glutamate in the presence of a low concentration of guanidineÆHCl (0.4 M ), spectral changes identical to those previously described occurred, even though the initial spectrum was different, and at the end of reaction a species absorbing at 340 nm could be detected (data not shown). DISCUSSION Many of the alterations produced by mutating the active- site Lys276 of GadB can be attributed to changes in the conformational properties of the protein. The findings that the mutation to alanine increases the unfolding temperature Fig. 5. Reaction of GadB-K276A with glutamate. The absorbance spectra of GadB-K276A mutant (60 l M ) were recorded with 20 m M sodium glutamate in 0.1 M sodium acetate, pH 4.6, from 0 to 6 h (A) and from 6 to 30 h (B). Fig. 4. Effect of guanidine on fluorescence emission spectra. (A) Emis- sion spectra (k exc ¼ 295 nm) of GadB-K276H mutant (0.82 l M )in 0.1 M sodium acetate, pH 4.6, containing 0, 1, 2, 3 and 5 M guani- dineÆHCl. (B) Emission spectra of wild-type GadB (0.77 l M )in0.1 M sodium acetate, pH 4.6, containing 0, 2, 4 and 5 M guanidineÆHCl. Ó FEBS 2002 Role of Lys276 in E. coli glutamate decarboxylase (Eur. J. Biochem. 269) 4917 of the holoenzyme by 11 °C, makes it resistant to tryptic hydrolysis and prevents the precipitation observed with the wild-type apoenzyme demonstrates that the absence of this lysine residue, either as an aldimine with the cofactor or as a protonated primary amine in the apoprotein, makes the protein less flexible. The same reduced flexibility seems likely to account for the slow, but ultimately tight, binding of PLP to the apoenzyme in vitro and for the fact that preparations of the mutant enzymes are always largely as apoenzyme. Similarly, inflexibility of the protein would also explain the observation that PLP, bound to the mutant enzymes, forms an aldimine with glutamate much more slowly than free PLP. The observation that the mutant apoenzymes are able to form stable holoenzymes despite the absence of the active site lysine residue shows that, as in other decarboxylases [11,12], non–covalent interactions between protein and cofactor are sufficient to ensure tight binding. This is also in line with the finding that His275 contributes to cofactor binding via an ionic interaction with the phosphate group of PLP [21]. Because the mutant forms of GadB cannot form an internal aldimine, it is not surprising that the 420 nm chromophore, characteristic of the wild-type enzyme at pH 4.6, is absent. However, the spectrum of PLP bound to the Lys276 mutants is quite different from the spectrum of the same compound when it is free in solution. Absorption bands at 388 nm and 330 nm are present in the spectra of both free PLP and PLP bound to the mutant enzymes, but in PLP free in solution the 388 nm chromophore is the most abundant species, whereas it is only a minor component of the spectrum of the mutant enzymes. In free PLP, the 388 nm and 330 nm chromophores are attributed to the unsubstituted and hydrated aldehydes, respectively [28]. It seems likely that the absorbance changes observed are due to an increase in the proportion of the PLP hydrate when the cofactor binds. A similar structure has also been suggested to be formed in the active-site lysine mutant of aromatic L -amino acid decarboxylase [11]. The biphasic nature of the changes in spectrum occurring upon PLP binding, reported also with the wild-type enzyme [29], suggests that initial cofactor binding is followed by a slower confor- mational adjustment. The CD spectra (Fig. 2C) show that GadB-K276H retains much more asymmetry than GadB-K276A probably because movement of the cofac- tor within the active site is more restricted by the histidine side-chain. Both wild-type GadB and its active-site lysine mutants show abrupt pH-dependent spectral changes involving simultaneous transfer of multiple protons. Other PLP- dependent enzymes undergo similar pH-dependent changes which are related to activity and are attributed to protona- tion of the internal aldimine formed with lysine and the cofactor aldehyde. For example, aspartate aminotransferase is converted from an inactive 430 nm-absorbing protonated internal aldimine to an active unprotonated 362 nm- absorbing form with a pK of 6.2 in a process that fits well to the ionization of a single proton [30]. It has been suggested that in wild-type Gad, the ionization responsible for the absorbance change does not take place on the internal aldimine and much evidence indicates that the change in spectrum of the wild-type enzyme is due to a conformational transition in the protein induced by shifting the pH [31]. It seems very likely that the pH-dependent changes that occur in the spectrum of the mutants under investigation in the present work are due to the same pH- induced conformational transition observed in the wild-type and that the different forms of the cofactor present in wild- type and mutant enzymes are recording the same event at the active site. The observation that the pH-dependent spectral changes occur in enzyme forms without the internal aldimine demonstrates that the protonation responsible for the absorbance changes and for activation of the wild-type enzyme is not of the internal aldimine itself. To explain the pH-dependent occurrence of the 330 nm-absorbing chromophore, it has been proposed that, in the wild-type enzyme, high pH induces the formation of an aldamine between the internal aldimine and an enzyme cysteine residue and that the low pH conformation favors the unsubstituted internal aldimine [31]. However, because in the mutant enzymes the high pH favors the 388 nm-absorbing unsubstituted aldehyde, formation of a covalent bond between PLP and a cysteine side-chain can be excluded as the basis of the pH- dependent absorbance changes observed with the Lys276 mutant enzymes. An explanation that unites observations from both wild-type and mutant enzymes is that the pH- dependent conformational change induces an alteration in the polarity of the active site. In this hypothesis, the environment of the cofactor is more hydrated in the low- pH conformation. Thus, the increased polarity favors the 420 nm-ketoenamine tautomer in the wild-type enzyme and the 330 nm-absorbing hydrated form of PLP in the mutant enzymes. Conversely, the less hydrated environ- ment of the high-pH conformation favors the 340 nm- absorbing enolimine tautomer of the wild-type cofactor and the 388 nm-absorbing unhydrated aldehyde of PLP in the mutant enzymes (Fig. 6) 1 . Fig. 6. Chemical structures of the chromophore proposed to be present at the active site of GadB wild-type and GadB-K276 mutants at low and high pH, respectively. 4918 A. Tramonti et al. (Eur. J. Biochem. 269) Ó FEBS 2002 In the presence of low concentrations of guanidineÆHCl or of sodium chloride, there is a spectral change in the mutant enzymes similar to that which occurs upon changing the pH. We suggest that low concentrations of solutes, by subtracting water molecules, cause a change in polarity of the active site and favor the 388 nm-absorbing unhydrated aldehyde of PLP. The failure of all the methods used to detect enzymatic activity in GadB-K276A and GadB-K276H mutants shows that GadB mutated at the active-site lysine loses reactivity towards the substrate, though the mutants are still capable of slowly binding L -glutamate and forming an external aldimine. The increase in absorbance at 412 nm observed when GadB K276A was mixed with glutamate, together with the observation that this chromophore converted to one absorbing at 340 nm when NaCNBH 3 was added provides strong evidence that an external aldimine is formed between the cofactor and the amino acid. The linear dependence on glutamate concentration of the observed rate constant governing this phase shows that there is no detectable saturation of the mutant enzyme with substrate, even at high concentrations. The second order constant calculated from this experiment (3.6 · 10 )3 ±0.3 · 10 )3 M )1 Æs )1 ) is much lower than that calculated for the reaction between free PLP and glutamate in 0.1 M sodium acetate, pH 4.6 (1.47 ± 0.13 M )1 Æs )1 ). The 400-fold reduction in reaction rate contrasts markedly with the observation that PLP bound to the corresponding mutant of aspartate aminotransfer- ase forms an external aldimine with glutamate or aspartate at least three orders of magnitude more rapidly than does free PLP [3]. It seems likely that at least part of this major reduction in reactivity of the enzyme-bound PLP is due to the extensive hydration we propose to be responsible for the predominant 330 nm chromophore, as well as to the reduced flexibility discussed earlier. The cause of the subsequent and even slower change in spectrum from 412 nm to 340 nm is unknown, although the failure to detect 4-aminobutyrate shows that it is not due to decarboxylation of the substrate. A possible explanation could be that Lys276 is, either directly or indirectly, involved in correctly positioning the Ca-COO – bond orthogonal to the plane of the delocalized cofactor imine system. Moreover, the release of cofactor as PLP when the enzyme was treated with NaOH after comple- tion of this reaction shows that it is not due to transamination to pyridoxamine phosphate. A similar but much more rapid spectral change occurs in the reaction catalyzed by the wild-type enzyme, where it has been attributed to a reversible conformational change in the enzyme leading to a form that does not undergo further reaction [19,21]. ACKNOWLEDGMENTS This work was partially supported by grants from the Italian Ministero dell’Istruzione, dell’Universita ` e della Ricerca and from the Istituto Pasteur-Fondazione Cenci Bolognetti (to DDB). The Centro di Eccellenza di Biologia e Medicina Molecolare (BEMM), Universita ` di Roma La Sapienza, is also acknowledged. We thank Professor A. Giartosio for DSC measurements and Professor D. Barra for critical reading of the manuscript. REFERENCES 1. Alexander, F.W., Sandmeier, E., Metha, P.K. & Christen, P. (1994) Evolutionary relationships among pyridoxal-5¢-phosphate- dependent enzymes. Eur. J. Biochem. 219, 953–960. 2. Mehta, P. & Christen, P. (1998) The molecular evolution of pyridoxal-5¢-phosphate-dependent enzymes. In Advances in Enzy- mology and Related Areas of Molecular Biology (Daniel, L.Purich, eds) Vol. 74, pp. 129–184. John Wiley and Sons, New York. 3. Toney, M.D. & Kirsch, J.F. (1993) Lysine 258 in aspartate ami- notransferase. Enforcer of the Circe effect for amino acid substrate and general-base catalyst for the 1,3-prototropic shift. Biochem- istry 30, 4072–4077. 4. Lu, Z., Nagata, S., McPhie, P. & Miles, E.W. (1993) Lysine 87 in the b subunit of tryptophan synthase that forms an internal aldimine with pyridoxal phosphate serves critical roles in trans- amination, catalysis and product release. J. Biol. Chem. 268, 8727– 8734. 5. Nishimura, K., Tanizawa, K., Yoshimura, T., Esaki, N., Futaki, S., Manning, J.M. & Soda, K. (1991) Effect of substituion of a lysyl residue that binds pyridoxal phosphate in thermostable D -amino acid aminotransferase by arginine and alanine. Bio- chemistry 30, 4072–4077. 6. Rege, V.D., Kredich, N.M., Tai, C H., Karsten, W.E., Schnac- kerz, K.D. & Cook, P.F. (1996) A change in the internal aldimine lysine (K42) in O-acetylserine sulfhydrylase to alanine indicates its importance in transamination and as a general base catalyst. Biochemistry 35, 13485–13493. 7. Toney, M.D. & Kirsch, J.F. (1992) Bronsted analysis of aspartate aminotransferase via exougenous catalysis of reactions of an inactive mutant. Protein Sci. 1, 107–109. 8. Yoshimura, T., Bathia, M.B., Manning, J.M., Ringe, D. & Soda, K. (1992) Partial reaction of bacterial D -amino acid transaminase with asparagine substituted for the lysine that binds coenzyme pyridoxal 5¢-phosphate. Biochemistry 31, 11748–11754. 9.Grimm,B.,Smith,M.A.&Wettstein,D.(1992)Theroleof Lys272 in the pyridoxal 5-phosphate active site of Synechococcus glutamate-1-semialdehyde aminotransferase. Eur. J. Biochem. 206, 579–585. 10. Schirch, D., Delle Fratte, S., Iurescia, S., Angelaccio, S., Contestabile, R., Bossa, F. & Schirch, V. (1993) Function of the active-site lysine in Escherichia coli serine hydroxymethyl- transferase. J. Biol. Chem. 268, 23132–23138. 11. Nishino, J., Hayashi, H., Ishii, S. & Kagamiyama, H. (1997) An anomalous side reaction of the Lys303 mutant aromatic L -amino acid decarboxylase unravels the role of the residue in catalysis. J. Biochem. 121, 604–611. 12. Osterman, A.L., Brooks, H.B., Jackson, L., Abbott, J.J. & Phillips, A. (1999) Lysine-69 plays a key role in catalysis by ornithine decarboxylase through acceleration of the Schiff base formation, decarboxylation, and product release steps. Biochem- istry 38, 11814–11826. 13. De Biase, D., Tramonti, A., Bossa, F. & Visca, P. (1999) The response to stationary-phase stress conditions in Escherichia coli: role and regulation of the glutamic acid decarboxylase system. Mol. Microbiol. 32, 1198–1211. 14. Waterman, S.R. & Small, P.L.C. (1996) Identification of r S - dependent genes associated with the stationary-phase acid-resist- ance phenotype of Shigella flexneri. Mol. Microbiol. 21, 925–940. 15. Cotter, P.D., Gahan, C.G. & Hill, C. (2001) A glutamate dec- arboxylase system protects Listeria monocytogenes in gastric fluid. Mol. Microbiol. 40, 465–475. 16. Sanders, J.W., Leenhouts, K., Burghoorn, J., Brands, J.R., Venema, G. & Kok, J. (1998) A chloride-inducible acid resistance mechanism in Lactococcus lactis and its regulation. Mol. Micro- biol. 27, 299–310. Ó FEBS 2002 Role of Lys276 in E. coli glutamate decarboxylase (Eur. J. Biochem. 269) 4919 17. Higuchi, T., Hayashi, H. & Abe, K. (1997) Exchange of glutamate and c-aminobutyrate in a Lactobacillus strain. J. Bacteriol. 179, 3362–3364. 18. Smith, D.K., Kassam, T., Singh, S. & Elliott, J.F. (1992) Escher- ichia coli has two homologous glutamate decarboxylase genes that maptodistinct.Loci. J. Bacteriol. 174, 5820–5826. 19. De Biase, D., Tramonti, A., John, R.A. & Bossa, F. (1996) Iso- lation, overexpression, and biochemical characterization of the two isoforms of glutamic acid decarboxylase from Escherichia coli. Protein Expr. Purif. 8, 430–438. 20. Tilley,K.,Akhtar,M.&Gani,D.(1994)Thestereochemical course of decarboxylation, transamination and elimination reac- tions catalysed by Escherichia coli glutamic acid decarboxylase. J. Chem. Soc. Perkin Trans. 1, 3079–3086. 21. Tramonti,A.,DeBiase,D.,Giartosio,A.,Bossa,F.&John,R.A. (1998) The roles of His-167 and His-275 in the reaction catalyzed by glutamate decarboxylase from Escherichia coli. J. Biol. Chem. 273, 1939–1945. 22. Higuchi, R., Krummel, B. & Saiki, R.K. (1988) A general method of in vitro preparation and specific mutagenesis of DNA frag- ments: study of protein and DNA interactions. Nucleic Acids Res. 16, 7351–7367. 23. Peterson, E.A. & Sober, H.A. (1954) Preparation of crystalline phosphorilated derivatives of vitamin B6. J. Am. Chem. Soc. 76, 169–175. 24. Grant, P.L., Basford, J.M. & John, R.A. (1987) An investigation of transient intermediates in the reaction of 2-methylglutamate with glutamate decarboxylase from Escherichia coli. Biochem. J. 241, 699–704. 25. De Biase, D., Maras, B. & John, R.A. (1991) A chromophore in glutamate decarboxylase has been wrongly identified as PQQ. FEBS Lett. 278, 120–122. 26. Shukuya, R. & Schwert, G.W. (1960) Glutamic acid decarb- oxylase. II. The spectrum of the enzyme. J. Biol. Chem. 235, 1653– 1657. 27. Watababe, A., Kurosawa, Y., Yoshimura, T., Kurihara, T., Soda, K. & Esaki, N. (1999) Role of lysine 39 of alanine racemase from Bacillus stearothermophilus that binds pyridoxal 5¢-phosphate. J. Biol. Chem. 274, 4189–4194. 28. Harris, C.M., Johnson, R.J. & Metzler, D.E. (1976) Band-shape analysis and resolution of electronic spectra of pyridoxal phos- phate and others 3-hydroxypyridine-4-aldehydes. Biochim. Biophys. Acta 421, 181–194. 29. O’Leary, M.H. & Malik, J.M. (1972) Kinetics and mechanism of the binding of pyridoxal 5¢-phosphate to apoglutamate decar- boxylase. J. Biol. Chem. 247, 7097–7105. 30. Jenkins, W.T. & Taylor, R.T. (1965) Glutamic-aspartic transa- minase. VIII. Equilibrium kinetics with aspartate. J. Biol. Chem. 240, 2907–2913. 31. O’Leary, M.H. & Brummund, W. Jr (1974) pH jump studies of glutamate decarboxylase. J. Biol. Chem. 249, 3737–3745. 32. Johnson, R.J. & Metzler, D.E. (1970) Analysing spectra of vitamin B6 derivatives. Methods Enzymol 18A, 433–471. 4920 A. Tramonti et al. (Eur. J. Biochem. 269) Ó FEBS 2002 . Contribution of Lys276 to the conformational flexibility of the active site of glutamate decarboxylase from Escherichia coli Angela Tramonti 1 , Robert A. John 2 ,. polarity of the active site. In this hypothesis, the environment of the cofactor is more hydrated in the low- pH conformation. Thus, the increased polarity favors the 420 nm-ketoenamine tautomer in the. [21]. The present work has been undertaken with the aim of determining the contribution of the active- site lysine residue (Lys276) to coenzyme binding and to stages in the reaction catalysed by E.