- Email: [email protected]
Mutation of the Nucleophilic Elbow of the Lux-Specific Thioesterase from Vibrio harveyi
Biochemical and Biophysical Research Communications 275, 704 –708 (2000) doi:10.1006/bbrc.2000.3362, available online at http://www.idealibrary.com on
Mutation of the Nucleophilic Elbow of the Lux-Specific Thioesterase from Vibrio harveyi Jun Li, Bijan Ahvazi, Rose Szittner, and Edward Meighen 1 Department of Biochemistry, McGill University, 3655 Promenade Sir William Osler, Montreal, Quebec, Canada H3G1Y6
Received July 18, 2000
Myristoyl-ACP thioesterase (LuxD) from Vibrio harveyi causes the slow release of fatty acids for reduction into the aldehyde substrate required for the bacterial bioluminescence reaction. The active site Ser nucleophile (S 114) of the LuxD thioesterase is in a ␥-turn with a sequence (AXS 114XS) quite different from the standard motif of GXSXG found in almost all (thio) esterases and lipases. The presence of an Arg residue (R 118) in the first turn of the helix after the ␥-turn also distinguishes LuxD from other enzymes. Mutation of R 118 to Leu inactivated the enzyme and prevented acylation of the Ser 114 nucleophile, while even a conservative replacement with Lys resulted in over 75% loss of the same functions, suggesting that R 118 helps maintain the configuration of the active site. In contrast, replacement of S 116 with Gly but not Ala stimulated the esterase and deacylation rates by over threefold. Purification of the S116G mutant to homogeneity and analyses of its intrinsic fluorescence on acylation with myristoyl-CoA clearly demonstrated that this mutant was much more active than wild-type LuxD. The presence of S 116 rather than the expected Gly residue in the ␥-turn containing the Ser nucleophile may function so that release of fatty acids from LuxD is restricted allowing a more efficient delivery of fatty acids to the luminescent system. © 2000 Academic Press
The most notable structural feature of the ␣hydrolase superfamily of proteins is the nucleophilic elbow (1, 2) containing a Ser residue in the strained ⑀-conformation. The nucleophilic Ser is in a very sharp ␥-turn between a -strand and an ␣-helix with two Gly residues flanking the Ser in a motif of G-X-S-X-G in almost all cases (3). The members of the ␣-hydrolase superfamily differ from other esterases, lipases and proteinases containing the G-X-S-X-G motif in that only the central Ser has torsion angles that place it in an unfavourable region of the Ramachandran plot. 1
To whom correspondence should be addressed. Fax: 514-3987384. E-mail: [email protected]. 0006-291X/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
Consequently, the predominance of Gly residues in the flanking positions in the ␣-hydrolase superfamily is due to tight packing between the -strand and the ␣-helix in the ␥-turn rather than secondary structure restrictions on the side chain due to unfavourable torsion angles (1). The lux-specific myristoyl-ACP thioesterase, LuxD, from Vibrio harveyi has been recently shown to be a member of the ␣-hydrolase superfamily (4) with the nucleophilic Ser in the ⑀-conformation ( ⫽ 53°, ⫽ ⫺127°). However, the amino acids flanking the nucleophilic serine, S 114, are not Gly residues but Ala and Ser in a sequence of A-X-S 114-X-S, which is conserved in LuxDs in other genera from both marine and terrestrial bacteria (5, 6). A more extended consensus sequence flanking the nucleophilic serine in esterases and lipases from the ␣-hydrolase and other structure families (1) also shows that the LuxD thioesterases can be distinguished by the presence of a conserved arginine residue (R 118) on the first turn of the ␣-helix after the nucleophilic elbow (6). An unusual feature of the lux-specific thioesterases is their slow turnover rate compared to other thioesterases (6 –11). In luminescent bacteria, LuxD is responsible for cleavage of myristoyl groups from myristoyl-ACP so that the fatty acid can be reduced to the tetradecanal substrate required for the luminescence reaction (12). The rate limiting step is a slow deacylation (6) which may serve to let LuxD function as a carrier of fatty acids from the fatty acid synthase complex to the fatty acid reductase complex responsible for reducing fatty acids to aldehyde (5, 8). A controlled release of fatty acids for reduction to the aldehyde substrate needed for the bioluminescence reaction would be necessary to prevent depletion of the fatty acids required for lipid biosynthesis. The design of the nucleophilic elbow and the conservation of specific amino acids in this region would therefore be important to generate a thioesterase which can be acylated to a high level and cleaved at a slow controlled rate.
704
Vol. 275, No. 2, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
This paper focuses on the residues flanking the nucleophilic serine in V. harveyi LuxD that distinguish this enzyme from other esterases and lipases. Replacement of A 112 and S 116 flanking the nucleophilic Ser with Gly residues showed that a large enhancement of activity occurred on mutation of S 116 but not A 112 to Gly indicating that the presence of S 116 is important for the slow deacylation of LuxD thioesterases. In contrast, mutation of the R 118 residue caused a major reduction in the acylation and activity of LuxD indicating that this residue must be conserved to maintain the Ser nucleophile in the active configuration.
TABLE 1
Esterase Activities of Mutant LuxDs a Mutant
Activity (nmol/min)
Wt A112G A112S S116A S116G S116G/A112G R118K R118L
2.6 ⫾ 0.9 1.8 ⫾ 0.3 2.9 ⫾ 0.6 1.8 ⫾ 0.5 8.2 ⫾ 2.5 6.9 ⫾ 1.2 0.2 ⫾ 0.1 0 ⫾ 0.1
a Extracted from 1 ml of E. coli (⫹luxD) and assayed as described under Materials and Methods.
MATERIALS AND METHODS Materials. [ 3H]Myristoyl-CoA was prepared from [ 3H]myristic acid (11 Ci/mmol, Amersham) as previously described (13). Stock solutions of 5 mM p-nitrophenyl myristate (Sigma) were prepared in isopropanol. Site-directed mutagenesis. Mutagenesis was conducted on a 1.6 kbp fragment containing V. harveyi LuxD inserted in the SacI and BamHI sites of the M13 (mp19) sequencing vector according to the method of Kunkel (14, 15). Oligonucleotide primers from the Sheldon Biotechnology Centre (McGill) were used to mutate A 112, S 116 and R 118 to the appropriate residues. A double mutant, A112G/S116G, was produced by sequentially mutating A 112 and S 116. Expression. The mutated luxD genes were transferred to the pT7-5 vector and expressed in Escherichia coli K38 containing the pGP1-2 vector coding for the bacteriophage T7 polymerase according to the procedure of Tabor and Richardson (16). Cells were grown at 30°C in enriched media (2% tryptone, 1% yeast extract, 0.5% NaCl, 0.2% glycerol, 50 mM KHPO 4, pH 7.2) until OD 660 ⫽ 1.0, the temperature increased to 42°C for 15 min to induce synthesis of T7 RNA polymerase and then 200 g/ml of rifampicin added to block E. coli RNA polymerase followed by incubation at 30°C for 60 min. Cells were centrifuged and then resuspended in one-fifth of the volume of 0.05 M phosphate, pH 7, before extraction by sonication. Esterase activity. The esterase activity was determined at 23°C from the rate of cleavage of 50 M p-nitrophenyl myristate in 50 mM phosphate, pH 8.0, containing 0.05% Triton X-100, 10 mM -mercaptoethanol and 20% glycerol by the change in absorbance at 405 nm. Providing short times (1 to 2 min) and low amounts of extracted or purified LuxD were used, the production of p-nitrophenol (16,800 M ⫺1 cm ⫺1 at pH 8) was linear with time. The activities of mutant LuxDs in extracts were calculated from the average of the values obtained for four to six samples grown, expressed and extracted independently. Activities are given in nmol/ min for 1 ml of cell culture (200 l of extract) based on the change of OD 405 with time for 50 l of extract. Protein acylation and deacylation. Samples of the 35S-labelled wild-type and mutant LuxDs were acylated by incubation in 50 mM phosphate, 0.1 mM -mercaptoethanol, pH 7.5 with 6 M [ 3H]myristoyl-CoA for 30 sec followed by mixing with an equal volume of SDS sample buffer (24 mM Tris-Cl, 10% glycerol, 0.8% SDS, 10 mM -mercaptoethanol, 0.04% bromophenol blue). Deacylation was conducted under the same conditions except unlabelled LuxD was reacted with 6 M [ 3H]myristoyl-CoA for 30 sec and then diluted 1:3.5 into cold 20 M myristoyl CoA for various times before quenching the reaction by dilution 1:1 into SDS sample buffer. Gel electrophoresis and radioactivity measurements. SDS-PAGE was performed by the method of Laemmli (17). The gels were stained with Coomassie brilliant blue R-250, destained, soaked in En3Hance, dried under vacuum and then exposed to Kodak X-OMAT
film at ⫺80°C for up to one week. To determine the level of radioactivity ( 3H and/or 35S), the LuxD band was excised and dissolved in hyamine hydroxide overnight at 37°C before counting in Cytoscint. Counts minus background for each isotope for the mutant and wildtype LuxDs were determined from the cpms measured in channels set for 3H and 35S after correction for spillover from the other isotope (0.4% of the 3H cpm into the 35S channel and 42% of the 35S cpm into the 3H channel). The isotope ratios [ 3H cpm]/[ 35S cpm] were divided by 10 to partially account for the higher level of 3H than 35S. For the wild-type LuxD, 35S cpm were corrected for an extra 35S band of approximately half the intensity being excised with LuxD.
RESULTS Measurements of the esterase activities of mutants with substitutions of the residue two amino acids prior to the active site Ser in the nucleophilic elbow (S 114) showed that replacement of A 112 by Ser of Gly had little effect on function (Table 1). In contrast, replacement of S 116 by Gly but not Ala stimulated activity by about 3-fold with the double mutant (S116G/A112G) being activated to the same extent as the S116G mutant. The results suggest that small amino acid acids (Gly, Ala, Ser) can be located at positions 112 and 116 without disruption of the nucleophilic character of the Ser, but higher activity is obtained if Gly rather than an amino acid with a small side chain (Ala, Ser) is located in the last position of the standard G-X-S-X-G motif. In contrast, activity appears to be independent of the amino acid at the first position in this motif providing that it is small (Ala, Gly, Ser). Mutation of R 116 resulted in loss of function with no detectable activity for the R118L mutant (Table 1). Substitution of R 116 by the amino acid most closely related in structure (i.e., Lys) appears to allow retention of a low level of activity indicating that the positively charged ⑀-amino group on Lys can partially substitute for the guanidinium group on the Arg residue. As the activity of crude extracts containing LuxD can vary due to differences in expression and/or extraction of the enzyme, the mutant LuxDs were specifically labelled with [ 35S]methionine during expression with T7 RNA polymerase (16). Under these conditions, ri-
705
Vol. 275, No. 2, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 1. SDS gel electrophoresis and isotope ratio of mutant 35Slabelled V. harveyi LuxD thioesterases reacted with [ 3H]myristoyl CoA. The autoradiogram shows the SDS gel electrophoresis of the mutant and wild-type LuxDs labelled during expression with [ 35S]methionine and acylated in extracts with [ 3H]myristoyl-CoA. (a) A112G, (b) A112S, (c) A112G/S116G, (d) S116A, (e) S116G, (f ) wt, (g) R118K, (h) R118L. The isotope ratios are given below each mutant and were determined by elucidation of the 35S and 3H cpm after excision of the LuxD band from the gel as described under Materials and Methods. The different intensities of the LuxD bands reflect primarily the differences in incorporation of [ 3H]myristoyl groups into LuxD and not the level of [ 35S]methionine incorporated into the protein, which is relatively low. The radioactivity at the bottom of the gel reflects the free myristoyl-CoA running at the front.
fampicin is added to inhibit E. coli RNA polymerase so that only the LuxD protein under control of the T7 promoter on the plasmid is synthesized. The extracted proteins are then labelled with [ 3H]myristoyl-CoA before SDS gel electrophoresis, autoradiography and excision of the LuxD band to determine the amount of LuxD protein from the 35S counts and the level of acylation from the incorporation of 3H in the myristoyl groups on LuxD. Figure 1 shows an autoradiogram of the SDS gel for the mutant and wild type enzymes after acylation. The differences in intensity primarily reflect different levels of the acylated LuxD mutant enzyme as most of the radioactivity originates from 3H and not 35S in samples which are fully acylated (0.5 to 1.0 mole of acyl group per mole of enzyme). Any variation in the relative amounts of the LuxD mutants on the SDS gel can be compensated for by excision of the LuxD band and elucidation of the isotope ratio ( 3H/ 35S). As shown in Fig. 1 all the mutants with substitutions at positions 112 and 116 have the same isotope ratio and therefore level of acylation (i.e., 3H/ 35S ratio ⫽ 7 to 10) as the wild type enzyme. In contrast the level of acylation, as measured by the isotope ratio, clearly drops on mutation of R 116. For the R116K and R116L mutants, the isotope ratios decrease to 2 and 0.3, respectively, in reasonable agreement with the relative decreases in esterase activity (Table 1). As might be expected, the same isotope ratio is seen for the LuxD mutants with enhanced activity (S116G and A112G/S116G) as that for the wild type LuxD as the wild-type enzyme is fully acylated (0.5 to 1.0 pmol/mol of enzyme) under these conditions (6). The cleavage of the acyl group from S 114 has been previously been shown to be the rate limiting step in
the (thio) esterase mechanism of LuxD (6). Deacylation can be followed directly in crude extracts by acylation with [ 3H]myristoyl-CoA and then diluting the samples into a high concentration of cold myristoyl-CoA and following the loss of labelled acyl groups covalently incorporated on LuxD. Figure 2 shows that the rates of deacylation are at least two to three times higher with the S116G mutant and the double mutant, A112G/ S116G, than the deacylation rate with wild type LuxD accounting directly for the higher activity of these mutants. For the R 118 mutants it was not possible to follow their deacylation rates due to the low levels of acylated enzyme that were formed. Other mutants were deacylated at the same rate as wild-type LuxD (data not shown) consistent with these mutants having activities similar to that of wild-type LuxD. To demonstrate directly that activity was enhanced by mutation of S 116 to Gly, the S116G mutant was purified to homogeneity and the acylation was followed by fluorescence (Fig. 3). Reaction of LuxD with myristoyl-CoA causes an increase in the intrinsic fluorescence of the enzyme due to a change in environment of W 213 located in close proximity to the nucleophilic serine (18). Figure 3 compares the change in fluorescence for the S116G mutant and the wild type enzyme during the reaction with myristoyl-CoA. The fluorescence rises to a plateau in the first minute of the reaction and then decreases back to its original value after complete hydrolysis of the substrate. The results clearly show that with the same amount of substrate,
FIG. 2. Comparison of the deacylation of wild-type V. harveyi LuxD and mutants with S 116 replaced by Gly in cold chase experiments. The mutant and wild type LuxDs were labelled with [ 3H]myristoyl-CoA for 1 min before dilution into cold myristoyl-CoA and the reaction then quenched at various times with SDS sample buffer. Samples were resolved by SDS gel electrophoresis and the amount of radioactivity in the LuxD polypeptide band divided by the counts in the LuxD band at the start of the cold chase experiment plotted as a function of time. Details are given under Materials and Methods.
706
Vol. 275, No. 2, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
The presence of a conserved Arg residue (R 118 ) in the helix just after S 114 is unique to the LuxD thioesterases. Figure 4 shows that the Arg is involved in a hydrogen bonded network with Glu (E 122 ) in the same helix as well as with Ser (S 191 ) and Thr (T 192 ) residues in the first structural element (a helix) just after the proposed “cap” subdomain and just before
FIG. 3. Intrinsic fluorescence during acylation with myristoylCoA of the wild-type V. harveyi LuxD and the S116G mutant. Fluorescence emission at 337 nm on excitation at 296 nm was followed during the reaction of 1 M enzyme with 5 M myristoyl-CoA in 50 mM phosphate, pH 7.0, at 20°C.
the purified S116G mutant cleaves the substrate at a rate 300 to 400% faster than the wild-type enzyme accounting for the much shorter interval during the reaction in which the elevated plateau of fluorescence is maintained. DISCUSSION The structure of the nucleophilic elbow with the serine nucleophilic S 114 as part of the catalytic triad followed by the alpha helix showing the structural role of the conserved R 118 residue is illustrated in Fig. 4. Replacement of S 116 by Gly but not Ala increased the rate of deacylation and esterase activity of LuxD. This result shows that the ⑀-conformation of the nucleophilic serine (S 114) is still retained in the S116G mutant, however, the ability for water to attack and cause deacylation is enhanced with Gly rather than Ser at position 116. This could reflect small changes in the torsion angles for S 114 due to introducing Gly at position 116 or enhanced ability to form the oxyanion hole (1). The conservation of a A-X-S 114-X-S motif in the nucleophilic elbow of LuxD thioesterases from different luminescent marine (Vibrio, Photobacterium) and terrestrial (Xenorhabdus) bacteria in spite of higher activity in mutants with Gly at position 116 suggests that a LuxD thioesterase with higher activity may be disadvantageous to the bacteria and cause unwanted release of free fatty acids. It appears that this enzyme may have evolved as a carrier of acyl groups to be released slowly only under appropriate physiological conditions.
FIG. 4. Structural model of the V. harveyi thioesterase active site. Atomic coordinates are from the crystal structure determined by Lawson et al. (4). Top: The ␥-turn flanked by A 112 and S 116 and containing the nucleophilic serine, S 114, as part of a Glu-His-Ser catalytic triad is followed by the ␣-helix showing the multiple interactions of the R 118 side chain. Bottom: Replacement of R 118 by Lys leads to the generation of some of the same contacts as in the wild-type enzyme. Lys was placed in its optimum rotomer configuration.
707
Vol. 275, No. 2, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
the beta strand leading to the Asp (D 211 ) of the catalytic triad (4). Replacement of R 118 with Leu inactivated the enzyme and decreased the level of acylation of the active site Ser at least 30-fold. However, replacement of R 118 with Lys in LuxD allowed a low retention of activity with the level of acylation of the active site Ser being decreased about four-fold. Figure 4 shows that Lys at position 118 in its best rotomer configuration can form one hydrogen bond with E 122 and a hydrogen bond with T 192 explaining the partial activity with the R118K mutant. The experiments support a model in which R 118 plays a central role in linking the nucleophilic elbow to the cap structure and D 211 in the catalytic triad. The conservation in LuxD of residues A 112 , S 116 and R 118 different from those observed flanking the active site Ser nucleophilic in other (thio) esterases and lipases appears to be necessary to maintain the slow release of fatty acid for the bioluminescence reaction.
3. 4.
5. 6. 7. 8. 9. 10. 11.
ACKNOWLEDGMENT
12. 13.
This work was supported by a grant (MT4314) from the Medical Research Council of Canada.
14. 15.
REFERENCES
16.
1. Derewenda, Z. S., and Derewenda, U. (1991) Biochem. Cell Biol. 69, 842– 851. 2. Ollis, D. L., Cheah, E., Cygler, M., Dijkstra, B., Frolow, F.,
17. 18.
708
Franken, S. M., Harel, M., Remington, S. J., Silman, I., and Schrag, J. (1992) Protein Eng. 5, 197–211. Brenner, S. (1988) Nature 334, 528 –530. Lawson, D. M., Derewenda, U., Serre, L., Ferri, S., Szittner, R., Wei, Y., Meighen, E. A., and Derewenda, Z. S. (1994) Biochemistry 33, 9382–9388. Ferri, S. R., and Meighen, E. A. (1991) J. Biol. Chem. 266, 12852–12857. Li, J., Szittner, R., Derewenda, Z. S., and Meighen, E. A. (1996) Biochemistry 35, 9967–9973. Carey, L. M., Rodriguez, A., and Meighen, E. (1984) J. Biol. Chem. 259, 10216 –10221. Byers, D., and Meighen, E. (1985) J. Biol. Chem. 260, 6938 – 6944. Pazirandeh, M., Chirala, S. S., and Wakil, S. J. (1991) J. Biol. Chem. 266, 20946 –20952. Tai, M. H., Chirala, S. S., and Wakil, S. J. (1993) Proc. Natl. Acad. Sci. USA 90, 1852–1856. Witkowski, A., Witkowska, H. E., and Smith, S. (1994) J. Biol. Chem. 269, 379 –383. Meighen, E. A. (1991) Microbiol. Rev. 55, 123–142. Rodriguez, A., Riendeau, D., and Meighen, E. (1983) J. Biol. Chem. 258, 5233–5237. Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. USA 82, 488 – 492. Kunkel, T. A., Roberts, J. D., and Zakour, R. A. (1987) Methods Enzymol. 154, 367–382. Tabor, S., and Richardson, C. C. (1985) Proc. Natl. Acad. Sci. USA 82, 1074 –1078. Laemmli, U. K. (1970) Nature 227, 680 – 685. Li, J., Szittner, R., and Meighen, E. A. (1998) Biochemistry 37, 16130 –16138.
Mutation of the Nucleophilic Elbow of the Lux-Specific Thioesterase from Vibrio harveyi Jun Li, Bijan Ahvazi, Rose Szittner, and Edward Meighen 1 Department of Biochemistry, McGill University, 3655 Promenade Sir William Osler, Montreal, Quebec, Canada H3G1Y6
Received July 18, 2000
Myristoyl-ACP thioesterase (LuxD) from Vibrio harveyi causes the slow release of fatty acids for reduction into the aldehyde substrate required for the bacterial bioluminescence reaction. The active site Ser nucleophile (S 114) of the LuxD thioesterase is in a ␥-turn with a sequence (AXS 114XS) quite different from the standard motif of GXSXG found in almost all (thio) esterases and lipases. The presence of an Arg residue (R 118) in the first turn of the helix after the ␥-turn also distinguishes LuxD from other enzymes. Mutation of R 118 to Leu inactivated the enzyme and prevented acylation of the Ser 114 nucleophile, while even a conservative replacement with Lys resulted in over 75% loss of the same functions, suggesting that R 118 helps maintain the configuration of the active site. In contrast, replacement of S 116 with Gly but not Ala stimulated the esterase and deacylation rates by over threefold. Purification of the S116G mutant to homogeneity and analyses of its intrinsic fluorescence on acylation with myristoyl-CoA clearly demonstrated that this mutant was much more active than wild-type LuxD. The presence of S 116 rather than the expected Gly residue in the ␥-turn containing the Ser nucleophile may function so that release of fatty acids from LuxD is restricted allowing a more efficient delivery of fatty acids to the luminescent system. © 2000 Academic Press
The most notable structural feature of the ␣hydrolase superfamily of proteins is the nucleophilic elbow (1, 2) containing a Ser residue in the strained ⑀-conformation. The nucleophilic Ser is in a very sharp ␥-turn between a -strand and an ␣-helix with two Gly residues flanking the Ser in a motif of G-X-S-X-G in almost all cases (3). The members of the ␣-hydrolase superfamily differ from other esterases, lipases and proteinases containing the G-X-S-X-G motif in that only the central Ser has torsion angles that place it in an unfavourable region of the Ramachandran plot. 1
To whom correspondence should be addressed. Fax: 514-3987384. E-mail: [email protected]. 0006-291X/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.
Consequently, the predominance of Gly residues in the flanking positions in the ␣-hydrolase superfamily is due to tight packing between the -strand and the ␣-helix in the ␥-turn rather than secondary structure restrictions on the side chain due to unfavourable torsion angles (1). The lux-specific myristoyl-ACP thioesterase, LuxD, from Vibrio harveyi has been recently shown to be a member of the ␣-hydrolase superfamily (4) with the nucleophilic Ser in the ⑀-conformation ( ⫽ 53°, ⫽ ⫺127°). However, the amino acids flanking the nucleophilic serine, S 114, are not Gly residues but Ala and Ser in a sequence of A-X-S 114-X-S, which is conserved in LuxDs in other genera from both marine and terrestrial bacteria (5, 6). A more extended consensus sequence flanking the nucleophilic serine in esterases and lipases from the ␣-hydrolase and other structure families (1) also shows that the LuxD thioesterases can be distinguished by the presence of a conserved arginine residue (R 118) on the first turn of the ␣-helix after the nucleophilic elbow (6). An unusual feature of the lux-specific thioesterases is their slow turnover rate compared to other thioesterases (6 –11). In luminescent bacteria, LuxD is responsible for cleavage of myristoyl groups from myristoyl-ACP so that the fatty acid can be reduced to the tetradecanal substrate required for the luminescence reaction (12). The rate limiting step is a slow deacylation (6) which may serve to let LuxD function as a carrier of fatty acids from the fatty acid synthase complex to the fatty acid reductase complex responsible for reducing fatty acids to aldehyde (5, 8). A controlled release of fatty acids for reduction to the aldehyde substrate needed for the bioluminescence reaction would be necessary to prevent depletion of the fatty acids required for lipid biosynthesis. The design of the nucleophilic elbow and the conservation of specific amino acids in this region would therefore be important to generate a thioesterase which can be acylated to a high level and cleaved at a slow controlled rate.
704
Vol. 275, No. 2, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
This paper focuses on the residues flanking the nucleophilic serine in V. harveyi LuxD that distinguish this enzyme from other esterases and lipases. Replacement of A 112 and S 116 flanking the nucleophilic Ser with Gly residues showed that a large enhancement of activity occurred on mutation of S 116 but not A 112 to Gly indicating that the presence of S 116 is important for the slow deacylation of LuxD thioesterases. In contrast, mutation of the R 118 residue caused a major reduction in the acylation and activity of LuxD indicating that this residue must be conserved to maintain the Ser nucleophile in the active configuration.
TABLE 1
Esterase Activities of Mutant LuxDs a Mutant
Activity (nmol/min)
Wt A112G A112S S116A S116G S116G/A112G R118K R118L
2.6 ⫾ 0.9 1.8 ⫾ 0.3 2.9 ⫾ 0.6 1.8 ⫾ 0.5 8.2 ⫾ 2.5 6.9 ⫾ 1.2 0.2 ⫾ 0.1 0 ⫾ 0.1
a Extracted from 1 ml of E. coli (⫹luxD) and assayed as described under Materials and Methods.
MATERIALS AND METHODS Materials. [ 3H]Myristoyl-CoA was prepared from [ 3H]myristic acid (11 Ci/mmol, Amersham) as previously described (13). Stock solutions of 5 mM p-nitrophenyl myristate (Sigma) were prepared in isopropanol. Site-directed mutagenesis. Mutagenesis was conducted on a 1.6 kbp fragment containing V. harveyi LuxD inserted in the SacI and BamHI sites of the M13 (mp19) sequencing vector according to the method of Kunkel (14, 15). Oligonucleotide primers from the Sheldon Biotechnology Centre (McGill) were used to mutate A 112, S 116 and R 118 to the appropriate residues. A double mutant, A112G/S116G, was produced by sequentially mutating A 112 and S 116. Expression. The mutated luxD genes were transferred to the pT7-5 vector and expressed in Escherichia coli K38 containing the pGP1-2 vector coding for the bacteriophage T7 polymerase according to the procedure of Tabor and Richardson (16). Cells were grown at 30°C in enriched media (2% tryptone, 1% yeast extract, 0.5% NaCl, 0.2% glycerol, 50 mM KHPO 4, pH 7.2) until OD 660 ⫽ 1.0, the temperature increased to 42°C for 15 min to induce synthesis of T7 RNA polymerase and then 200 g/ml of rifampicin added to block E. coli RNA polymerase followed by incubation at 30°C for 60 min. Cells were centrifuged and then resuspended in one-fifth of the volume of 0.05 M phosphate, pH 7, before extraction by sonication. Esterase activity. The esterase activity was determined at 23°C from the rate of cleavage of 50 M p-nitrophenyl myristate in 50 mM phosphate, pH 8.0, containing 0.05% Triton X-100, 10 mM -mercaptoethanol and 20% glycerol by the change in absorbance at 405 nm. Providing short times (1 to 2 min) and low amounts of extracted or purified LuxD were used, the production of p-nitrophenol (16,800 M ⫺1 cm ⫺1 at pH 8) was linear with time. The activities of mutant LuxDs in extracts were calculated from the average of the values obtained for four to six samples grown, expressed and extracted independently. Activities are given in nmol/ min for 1 ml of cell culture (200 l of extract) based on the change of OD 405 with time for 50 l of extract. Protein acylation and deacylation. Samples of the 35S-labelled wild-type and mutant LuxDs were acylated by incubation in 50 mM phosphate, 0.1 mM -mercaptoethanol, pH 7.5 with 6 M [ 3H]myristoyl-CoA for 30 sec followed by mixing with an equal volume of SDS sample buffer (24 mM Tris-Cl, 10% glycerol, 0.8% SDS, 10 mM -mercaptoethanol, 0.04% bromophenol blue). Deacylation was conducted under the same conditions except unlabelled LuxD was reacted with 6 M [ 3H]myristoyl-CoA for 30 sec and then diluted 1:3.5 into cold 20 M myristoyl CoA for various times before quenching the reaction by dilution 1:1 into SDS sample buffer. Gel electrophoresis and radioactivity measurements. SDS-PAGE was performed by the method of Laemmli (17). The gels were stained with Coomassie brilliant blue R-250, destained, soaked in En3Hance, dried under vacuum and then exposed to Kodak X-OMAT
film at ⫺80°C for up to one week. To determine the level of radioactivity ( 3H and/or 35S), the LuxD band was excised and dissolved in hyamine hydroxide overnight at 37°C before counting in Cytoscint. Counts minus background for each isotope for the mutant and wildtype LuxDs were determined from the cpms measured in channels set for 3H and 35S after correction for spillover from the other isotope (0.4% of the 3H cpm into the 35S channel and 42% of the 35S cpm into the 3H channel). The isotope ratios [ 3H cpm]/[ 35S cpm] were divided by 10 to partially account for the higher level of 3H than 35S. For the wild-type LuxD, 35S cpm were corrected for an extra 35S band of approximately half the intensity being excised with LuxD.
RESULTS Measurements of the esterase activities of mutants with substitutions of the residue two amino acids prior to the active site Ser in the nucleophilic elbow (S 114) showed that replacement of A 112 by Ser of Gly had little effect on function (Table 1). In contrast, replacement of S 116 by Gly but not Ala stimulated activity by about 3-fold with the double mutant (S116G/A112G) being activated to the same extent as the S116G mutant. The results suggest that small amino acid acids (Gly, Ala, Ser) can be located at positions 112 and 116 without disruption of the nucleophilic character of the Ser, but higher activity is obtained if Gly rather than an amino acid with a small side chain (Ala, Ser) is located in the last position of the standard G-X-S-X-G motif. In contrast, activity appears to be independent of the amino acid at the first position in this motif providing that it is small (Ala, Gly, Ser). Mutation of R 116 resulted in loss of function with no detectable activity for the R118L mutant (Table 1). Substitution of R 116 by the amino acid most closely related in structure (i.e., Lys) appears to allow retention of a low level of activity indicating that the positively charged ⑀-amino group on Lys can partially substitute for the guanidinium group on the Arg residue. As the activity of crude extracts containing LuxD can vary due to differences in expression and/or extraction of the enzyme, the mutant LuxDs were specifically labelled with [ 35S]methionine during expression with T7 RNA polymerase (16). Under these conditions, ri-
705
Vol. 275, No. 2, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
FIG. 1. SDS gel electrophoresis and isotope ratio of mutant 35Slabelled V. harveyi LuxD thioesterases reacted with [ 3H]myristoyl CoA. The autoradiogram shows the SDS gel electrophoresis of the mutant and wild-type LuxDs labelled during expression with [ 35S]methionine and acylated in extracts with [ 3H]myristoyl-CoA. (a) A112G, (b) A112S, (c) A112G/S116G, (d) S116A, (e) S116G, (f ) wt, (g) R118K, (h) R118L. The isotope ratios are given below each mutant and were determined by elucidation of the 35S and 3H cpm after excision of the LuxD band from the gel as described under Materials and Methods. The different intensities of the LuxD bands reflect primarily the differences in incorporation of [ 3H]myristoyl groups into LuxD and not the level of [ 35S]methionine incorporated into the protein, which is relatively low. The radioactivity at the bottom of the gel reflects the free myristoyl-CoA running at the front.
fampicin is added to inhibit E. coli RNA polymerase so that only the LuxD protein under control of the T7 promoter on the plasmid is synthesized. The extracted proteins are then labelled with [ 3H]myristoyl-CoA before SDS gel electrophoresis, autoradiography and excision of the LuxD band to determine the amount of LuxD protein from the 35S counts and the level of acylation from the incorporation of 3H in the myristoyl groups on LuxD. Figure 1 shows an autoradiogram of the SDS gel for the mutant and wild type enzymes after acylation. The differences in intensity primarily reflect different levels of the acylated LuxD mutant enzyme as most of the radioactivity originates from 3H and not 35S in samples which are fully acylated (0.5 to 1.0 mole of acyl group per mole of enzyme). Any variation in the relative amounts of the LuxD mutants on the SDS gel can be compensated for by excision of the LuxD band and elucidation of the isotope ratio ( 3H/ 35S). As shown in Fig. 1 all the mutants with substitutions at positions 112 and 116 have the same isotope ratio and therefore level of acylation (i.e., 3H/ 35S ratio ⫽ 7 to 10) as the wild type enzyme. In contrast the level of acylation, as measured by the isotope ratio, clearly drops on mutation of R 116. For the R116K and R116L mutants, the isotope ratios decrease to 2 and 0.3, respectively, in reasonable agreement with the relative decreases in esterase activity (Table 1). As might be expected, the same isotope ratio is seen for the LuxD mutants with enhanced activity (S116G and A112G/S116G) as that for the wild type LuxD as the wild-type enzyme is fully acylated (0.5 to 1.0 pmol/mol of enzyme) under these conditions (6). The cleavage of the acyl group from S 114 has been previously been shown to be the rate limiting step in
the (thio) esterase mechanism of LuxD (6). Deacylation can be followed directly in crude extracts by acylation with [ 3H]myristoyl-CoA and then diluting the samples into a high concentration of cold myristoyl-CoA and following the loss of labelled acyl groups covalently incorporated on LuxD. Figure 2 shows that the rates of deacylation are at least two to three times higher with the S116G mutant and the double mutant, A112G/ S116G, than the deacylation rate with wild type LuxD accounting directly for the higher activity of these mutants. For the R 118 mutants it was not possible to follow their deacylation rates due to the low levels of acylated enzyme that were formed. Other mutants were deacylated at the same rate as wild-type LuxD (data not shown) consistent with these mutants having activities similar to that of wild-type LuxD. To demonstrate directly that activity was enhanced by mutation of S 116 to Gly, the S116G mutant was purified to homogeneity and the acylation was followed by fluorescence (Fig. 3). Reaction of LuxD with myristoyl-CoA causes an increase in the intrinsic fluorescence of the enzyme due to a change in environment of W 213 located in close proximity to the nucleophilic serine (18). Figure 3 compares the change in fluorescence for the S116G mutant and the wild type enzyme during the reaction with myristoyl-CoA. The fluorescence rises to a plateau in the first minute of the reaction and then decreases back to its original value after complete hydrolysis of the substrate. The results clearly show that with the same amount of substrate,
FIG. 2. Comparison of the deacylation of wild-type V. harveyi LuxD and mutants with S 116 replaced by Gly in cold chase experiments. The mutant and wild type LuxDs were labelled with [ 3H]myristoyl-CoA for 1 min before dilution into cold myristoyl-CoA and the reaction then quenched at various times with SDS sample buffer. Samples were resolved by SDS gel electrophoresis and the amount of radioactivity in the LuxD polypeptide band divided by the counts in the LuxD band at the start of the cold chase experiment plotted as a function of time. Details are given under Materials and Methods.
706
Vol. 275, No. 2, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
The presence of a conserved Arg residue (R 118 ) in the helix just after S 114 is unique to the LuxD thioesterases. Figure 4 shows that the Arg is involved in a hydrogen bonded network with Glu (E 122 ) in the same helix as well as with Ser (S 191 ) and Thr (T 192 ) residues in the first structural element (a helix) just after the proposed “cap” subdomain and just before
FIG. 3. Intrinsic fluorescence during acylation with myristoylCoA of the wild-type V. harveyi LuxD and the S116G mutant. Fluorescence emission at 337 nm on excitation at 296 nm was followed during the reaction of 1 M enzyme with 5 M myristoyl-CoA in 50 mM phosphate, pH 7.0, at 20°C.
the purified S116G mutant cleaves the substrate at a rate 300 to 400% faster than the wild-type enzyme accounting for the much shorter interval during the reaction in which the elevated plateau of fluorescence is maintained. DISCUSSION The structure of the nucleophilic elbow with the serine nucleophilic S 114 as part of the catalytic triad followed by the alpha helix showing the structural role of the conserved R 118 residue is illustrated in Fig. 4. Replacement of S 116 by Gly but not Ala increased the rate of deacylation and esterase activity of LuxD. This result shows that the ⑀-conformation of the nucleophilic serine (S 114) is still retained in the S116G mutant, however, the ability for water to attack and cause deacylation is enhanced with Gly rather than Ser at position 116. This could reflect small changes in the torsion angles for S 114 due to introducing Gly at position 116 or enhanced ability to form the oxyanion hole (1). The conservation of a A-X-S 114-X-S motif in the nucleophilic elbow of LuxD thioesterases from different luminescent marine (Vibrio, Photobacterium) and terrestrial (Xenorhabdus) bacteria in spite of higher activity in mutants with Gly at position 116 suggests that a LuxD thioesterase with higher activity may be disadvantageous to the bacteria and cause unwanted release of free fatty acids. It appears that this enzyme may have evolved as a carrier of acyl groups to be released slowly only under appropriate physiological conditions.
FIG. 4. Structural model of the V. harveyi thioesterase active site. Atomic coordinates are from the crystal structure determined by Lawson et al. (4). Top: The ␥-turn flanked by A 112 and S 116 and containing the nucleophilic serine, S 114, as part of a Glu-His-Ser catalytic triad is followed by the ␣-helix showing the multiple interactions of the R 118 side chain. Bottom: Replacement of R 118 by Lys leads to the generation of some of the same contacts as in the wild-type enzyme. Lys was placed in its optimum rotomer configuration.
707
Vol. 275, No. 2, 2000
BIOCHEMICAL AND BIOPHYSICAL RESEARCH COMMUNICATIONS
the beta strand leading to the Asp (D 211 ) of the catalytic triad (4). Replacement of R 118 with Leu inactivated the enzyme and decreased the level of acylation of the active site Ser at least 30-fold. However, replacement of R 118 with Lys in LuxD allowed a low retention of activity with the level of acylation of the active site Ser being decreased about four-fold. Figure 4 shows that Lys at position 118 in its best rotomer configuration can form one hydrogen bond with E 122 and a hydrogen bond with T 192 explaining the partial activity with the R118K mutant. The experiments support a model in which R 118 plays a central role in linking the nucleophilic elbow to the cap structure and D 211 in the catalytic triad. The conservation in LuxD of residues A 112 , S 116 and R 118 different from those observed flanking the active site Ser nucleophilic in other (thio) esterases and lipases appears to be necessary to maintain the slow release of fatty acid for the bioluminescence reaction.
3. 4.
5. 6. 7. 8. 9. 10. 11.
ACKNOWLEDGMENT
12. 13.
This work was supported by a grant (MT4314) from the Medical Research Council of Canada.
14. 15.
REFERENCES
16.
1. Derewenda, Z. S., and Derewenda, U. (1991) Biochem. Cell Biol. 69, 842– 851. 2. Ollis, D. L., Cheah, E., Cygler, M., Dijkstra, B., Frolow, F.,
17. 18.
708
Franken, S. M., Harel, M., Remington, S. J., Silman, I., and Schrag, J. (1992) Protein Eng. 5, 197–211. Brenner, S. (1988) Nature 334, 528 –530. Lawson, D. M., Derewenda, U., Serre, L., Ferri, S., Szittner, R., Wei, Y., Meighen, E. A., and Derewenda, Z. S. (1994) Biochemistry 33, 9382–9388. Ferri, S. R., and Meighen, E. A. (1991) J. Biol. Chem. 266, 12852–12857. Li, J., Szittner, R., Derewenda, Z. S., and Meighen, E. A. (1996) Biochemistry 35, 9967–9973. Carey, L. M., Rodriguez, A., and Meighen, E. (1984) J. Biol. Chem. 259, 10216 –10221. Byers, D., and Meighen, E. (1985) J. Biol. Chem. 260, 6938 – 6944. Pazirandeh, M., Chirala, S. S., and Wakil, S. J. (1991) J. Biol. Chem. 266, 20946 –20952. Tai, M. H., Chirala, S. S., and Wakil, S. J. (1993) Proc. Natl. Acad. Sci. USA 90, 1852–1856. Witkowski, A., Witkowska, H. E., and Smith, S. (1994) J. Biol. Chem. 269, 379 –383. Meighen, E. A. (1991) Microbiol. Rev. 55, 123–142. Rodriguez, A., Riendeau, D., and Meighen, E. (1983) J. Biol. Chem. 258, 5233–5237. Kunkel, T. A. (1985) Proc. Natl. Acad. Sci. USA 82, 488 – 492. Kunkel, T. A., Roberts, J. D., and Zakour, R. A. (1987) Methods Enzymol. 154, 367–382. Tabor, S., and Richardson, C. C. (1985) Proc. Natl. Acad. Sci. USA 82, 1074 –1078. Laemmli, U. K. (1970) Nature 227, 680 – 685. Li, J., Szittner, R., and Meighen, E. A. (1998) Biochemistry 37, 16130 –16138.