Protein Science (1997), 6:1937-1944. Cambridge University Press. Printed in the USA Copyright 0 1997 The Protein Society Determination of pK, values of the histidine side chains of phosphatidylinositol-specific phospholipase C from Bacillus cereus by NMR spectroscopy and site-directed mutagenesis TUN LIU, MARGRET RYAN, FREDERICK W. DAHLQUIST, AND 0. HAYES GRIFFITH Institute of Molecular Biology and Department of Chemistry, University of Oregon, Eugene, Oregon 97403 (RECEIVEDDecember 4, 1996: ACCEPTEDMay 19, 1997) Abstract Two active site histidine residues have been implicated in the catalysis of phosphatidylinositol-specific phospholipase C (PI-PLC). In this report, we present the first study of the pK,, values of histidines of a PI-PLC. All six histidines of Bacillus cereus PI-PLC were studied by 2D NMR spectroscopy and site-directed mutagenesis. The protein was selec- tively labeled with '3C"-histidine. A series of 'H-I3C HSQC NMR spectra were acquired over a pH range of 4.0-9.0. Five of the six histidines have been individually substituted with alanine to aid the resonance assignments in the NMR spectra. Overall, the remaining histidines in the mutants show little chemical shift changes in the 'H-"C HSQC spectra, indicating that the alanine substitution has no effect on the tertiary structure of the protein. H32A and H82A mutants are inactive enzymes, while H92A and H61A are fully active, and H81A retains about 15% of the wild-type activity. The active site histidines, His32 and His82, display pK,, values of 7.6 and 6.9, respectively. His92 and His227 exhibit pK, values of 5.4 and 6.9. His61 and His81 do not titrate over the pH range studied. These values are consistent with the crystal structure data, which shows that His92 and His227 are on the surface of the protein, whereas His61 and His81 are buried. The p$, value of 6.9 corroborates the hypothesis of His82 acting as a general acid in the catalysis. His32 is essential to enzyme activity, but its putative role as the general base is in question due to its relatively high pK,,. Keywords: "C"-histidine; 'H- I3C HSQC; histidine pK,,; NMR; phosphatidylinositol-specific phospholipase C; site-directed mutagenesis Phosphatidylinositol-specific phospholipase C (PI-PLC) is the fo- 1993). The bacterial PI-PLCs cleave phosphatidylinositol (PI) to cus of considerable research because of its role in generating sec- form the lipid-soluble diacylglycerol (DAG) and water-soluble myo- ond messengers and modulating membrane traffic in eukaryotic inositol 1,2-cyclic phosphate [I(1:2cyc)P], which can be converted cells (Berridge, 1993; Lee & Rhee, 1995; De Camilli et al., 1996). to acyclic I-phosphate as shown in Figure 1 (for a review, see Moreover, the PI-PLCs from bacteria such as Bacilluscereus, Bruzik & Tsai, 1994). The mammalian PI-PLCs hydrolyze phos- Bacillus thuringiensis, Staphylococcus aureus, Listeria monocyto- phatidylinositol4,5-bisphosphateto yield DAG and the correspond- genes, and Costridium novvi possess the activity of cleaving gly- ing polyphosphorylated inositol. With the recent completion of cosylphosphatidylinositol (GPI), the lipid anchor that attaches many crystal structures of B. cereus PI-PLC (Heinz et al., 1995, 1996), proteins to membranes (for reviews, see Turner, 1990; Englund, and the mammalian PI-PLCGI (Essen et al., 1996), more detailed studies of the molecular mechanisms of these enzymes are now possible. The catalytic domain of the mammalian PI-PLC has a Reprint requests to: 0. Hayes Griffith, Institute of Molecular Biology similar folding topology to that of B. cereus PI-PLC, Le., an ir- and Department of Chemistry, University of Oregon, Eugene, Oregon 97403; regular crlp-barrel, also named TIM barrel after the more regular e-mail: [email protected]. structure of this type first observed in triose phosphate isomerase. Ahhret,icrrions: DAG, diacylglycerol; DSS, 2,2-dimethyl-2-silapentane- 5-sulfonate, sodium salt: EDTA, ethylenediarninetetraacetate; GLN-INS, There are two highly conserved histidines at the active sites of both glucosaminyl(cu1~6)-D-m~o-inositol:GPI, glycosylphosphatidylinositol; bacterial and mammalian PI-PLCs. In the B. cereus isozyme, these H-bond.hydrogen-bond: HEPES, N-(2-hydroxyethyl)piperazine-N'-2- two histidines are His32 and His82 (Fig. 2). Based on the presence ethanesulfonic acid: HSQC, heteronuclear single-quantum coherence; I( l)P, of the stable cyclic product, I(1:2cyc)P, an in-line mechanism of m?n-inositol I-phosphate: I( 1 :2cyc)P. D-mJo-inositol I ,2-cyclic phosphate; NMR. nuclear magnetic resonance: PI,phosphatidylinositol: PI-PLC, general acid and base catalysis was postulated, analogous to that phosphatidylinositol-specific phospholipase C; Tris. tris(hydroxylmethy1)- for ribonuclease A (Lin et al., 1990; Volwerk et al., 1990; Lewis aminomethane. et al., 1993). 1937 1938 7: Liu et ai. D.4G -0-6-0 PI-PLC * fast + OH OH PI I( I:Zcyc)P KIP Fig. 1. The reactions catalyzed by R. cereus PI-PLC. The first reaction isfast and yields diacylglycerol (DAG) and Inyo-inositol I.2-cyclic phosphate [I(I:Zcyc)P]. The second reaction is much slower and convens I( I:Zcyc)P to nyo-inositol I-phosphate [[(I )PI. To understand the roles of His32 and His82 in catalysis, we have (<O. 1 mM). The other two weak peaks above and below the His92 initiated the investigation of the microenvironments of the histi- peakin the 'H dimension are data truncation artifacts from the dines byNMR spectroscopy. Thereare a total of six histidine strong His92 resonance. The relatively low intensity for the His32 residues (Fig. 2) in B. cereus PI-PLC, a 298-residue protein with resonance is dueto exchange broadening at the pH value (8.0) near a molecular weight of 35 kDa. In this report, we present the pK,, its pK,, (7.6). The 'H chemical shifts observed for the histidines determination for all six histidines by 2D NMR spectroscopy aided that were not substituted in the mutants are summarized in Table 1. by site-directed mutagenesis. This is the first study of histidine pK,, Overall, the remaining histidines in the mutants show little chem- values for a member of the PI-PLC family of enzymes. ical shift changes in the 'H-"C HSQC spectra except for His32 and His81 in the H82A mutant. The replacement of His81 has only a slight effect on the chemical shift of His82. The alanine substi- Results tution of His82 results in -0.4 ppm downfield shift for His32 and -0.2 ppm upfield shift for His81, respectively, in the 'H dimen- Resonance assignments for the histidines in the NMR spectra sion. Replacement of His32 with alanine causes only -0.08 ppm A 'H-"C correlated spectrum of '3Cf'-histidine-labeled B. cereus downfield shift of His82 in 'H and has no effect on the chemical PI-PLC is shown in Figure 3a. All six histidine residues are present shifts of the other histidines. and resolved. Individual resonances in thc spectrum were assigncd The 'H and "C chemical shifts of the C"Hs ofHis61 and from the NMR spectra of a set of mutants in which one histidine His227 are identical at pH 7.6 (Fig. 4). As the pH decreases, one at a time was replaced by alanine (Fig. 3b-f). The impurity peak resonance shifts downfield, whereas the other one remains the marked in the spectrum 3(e) is present in every sample but dom- same. To determine the continuity of the titration curves from the inant in the H81A sample resulting from poor overexpression of crossover point at pH 7.6, several 'H-"C HSQC spectra were the mutant protein and consequently low protein concentration acquired for the H61 A mutant between pH 6.8-8.2. The chemical Fig. 2. Stereo rihhon diagram of the crystal structure of R. cereus PI-PLC showing the histidine side chains together with myo-inositol (INS) at the active site. The plot was generated using the program MOLSCRIPT (Kraulis. 1991). Histidine pKu values of R. cereus PI-PLC by NMR 1939 His32 I " c H82A F H92A t 8.0 1.5 7.0 Fig. 3. The 'H-I'C HSQC NMR spectra of "C"~histidine-lahcled B. WWLL\I'I-PL<' acquit-cd at pH 8.0 and 25°C. (A) Wild ~ypc; (B)-(F) mutants with one histidine replaced hy alanine in each case. The arrowsin the spectra (A)-(C) indicate dala tt-uncatton artilact\ from the strong Hi\92 resonance. In (E), the peak marked with an asterisk (:F)is due to an impurity, which is present 111 every wnplc hut dominant in the H81A sample duc to poor overexpression 01' the mutant protein and consequently low protein concentration. shifts of the remaining resonance (His227) of the pair in these the two histidines were reversed. However, the "C" chemical shifts spectra are the same as in the wild type, so we were able to trace of His32 and His82 are very different. The pK,,s determined from the curve for His227 in the absence of His61. Thus, the resonance both 'H and ' 'C data are the samc, and this rules out the possibility that does not titrate over the range of pH 4-9 belongs to His61. of confusion in the assignments. Furthermore, when a weak com- Because the 'H" chemical shifts of His32 and His82 are similar, petitive inhibitor of the cnzymc, glucosaminyl(al76-D-my- one might raisc the question whether the resonance assignments of inositol, is added to the protein sample, only the resonance assigned 1940 Z Liu et al. Table 1. 'H Chemical shifrs (ppm) of the histidine C"Hs were not determined due to protein precipitation at lower pH and in the histidine mutants limitation of the buffer system at higher pH.