Two Conserved Histidine Residues Are Critical to the Function of the TagF-like Family of Enzymes*

The TagF protein from Bacillus subtilis 168 is the poly(glycerol phosphate) polymerase responsible for the synthesis of wall teichoic acid and is the prototype member of a poorly understood family of similar teichoic acid synthetic enzymes. Here we describe in vitro and in vivo characterization of TagF, which localizes the active site to the carboxyl terminus of the protein and identifies residues that are critical for catalysis. We also establish the first mechanistic link among TagF and similar proteins by demonstrating that the identified residues are also critical in the function of TagB, a homologous enzyme implicated as the glycerophosphotransferase responsible for priming poly(glycerol phosphate) synthesis (Bhavsar, A. P., Truant, R., and Brown, E. D. (Sept. 2, 2005) J. Biol. Chem. 36691–36700). We investigated the dependence of TagF activity on pH and showed that deprotonation of a residue with a pKa near neutral is critical for proper function. Alteration of histidine residues 474 and 612 by site-directed mutagenesis abolished TagF activity in vitro (5000-fold reduction in kcat/Km) while variants in four other conserved acidic residues showed minimal loss of activity. Complementation using H474A and H612A mutant alleles failed to suppress a lethal temperature-sensitive tagF defect in vivo despite confirmation of robust expression by Western blot. When corresponding mutations were made to the homologous tagB gene, these alleles were unable to suppress a tagB temperature-sensitive lethal phenotype. These results extend the mechanistic observations for TagF across a wider family of enzymes and provide the first biochemical evidence for the relatedness of these two enzymes.

The TagF protein from Bacillus subtilis 168 is the poly(glycerol phosphate) polymerase responsible for the synthesis of wall teichoic acid and is the prototype member of a poorly understood family of similar teichoic acid synthetic enzymes. Here we describe in vitro and in vivo characterization of TagF, which localizes the active site to the carboxyl terminus of the protein and identifies residues that are critical for catalysis. We also establish the first mechanistic link among TagF and similar proteins by demonstrating that the identified residues are also critical in the function of TagB, a homologous enzyme implicated as the glycerophosphotransferase responsible for priming poly(glycerol phosphate) synthesis ( We investigated the dependence of TagF activity on pH and showed that deprotonation of a residue with a pK a near neutral is critical for proper function. Alteration of histidine residues 474 and 612 by site-directed mutagenesis abolished TagF activity in vitro (5000fold reduction in k cat /K m ) while variants in four other conserved acidic residues showed minimal loss of activity. Complementation using H474A and H612A mutant alleles failed to suppress a lethal temperature-sensitive tagF defect in vivo despite confirmation of robust expression by Western blot. When corresponding mutations were made to the homologous tagB gene, these alleles were unable to suppress a tagB temperature-sensitive lethal phenotype. These results extend the mechanistic observations for TagF across a wider family of enzymes and provide the first biochemical evidence for the relatedness of these two enzymes.
Teichoic acids comprise a diverse group of phosphate-containing polymers found covalently attached to peptidoglycan in the cell walls of Gram-positive bacteria. In Bacillus subtilis 168, the major wall teichoic acid is a linear 1,3-linked poly(glycerol phosphate) polymer that is anchored to the wall through an N-acetylmannosamine-␤-(1-4)-Nacetylglucosamine disaccharide linkage unit attached via phosphodiester linkage to the 6-hydroxyl of N-acetylmuramic acid in peptidoglycan. The genes believed to be responsible for teichoic acid synthesis and export in this organism are organized into the operons tagAB, tagDEF, tagO, and tagGH (1)(2)(3). Despite a poor understanding of the function of these highly negatively charged polymers, a growing body of evidence suggests that they play an essential role in the growth of Gram-positive organisms. Depletion of the TagO (3), TagD (4), TagB (5), TagF (5), and TagG/H (6) proteins in B. subtilis resulted in gross morphological abnormalities and cell lysis. Although the organization and regulation of the genes responsible for teichoic acid synthesis in model poly(glycerol phosphate) and poly(ribitol phosphate) containing organisms is fairly well understood (1,(7)(8)(9), still little is known about the function of the gene products biochemically. This is particularly true of the enzymes involved in synthesis of the poly(glycerol phosphate) polymer itself, namely TagB and TagF. Of the teichoic acid synthetic enzymes in B. subtilis 168, only TagD (glycerol-3-phosphate cytidylyltransferse) has been studied in great biochemical detail (10 -14), whereas the enzymes involved in disaccharide synthesis and polymer export have strong functional predictions based on sequence similarity (3,15,16). The poly(glycerol phosphate), or teichoic acid glycerol phosphate (Tag), 2 polymerase was first studied using crude membrane preparations or partially purified protein samples (17,18) before being genetically associated with the tagF gene by Pooley (19). The tagF gene encodes a basic, 88-kDa protein 746 amino acids in length that shows no sequence similarity to any protein of known function. The enzyme is found associated with the membrane both in B. subtilis and when expressed in Escherichia coli, but can be readily extracted for in vitro study (20). We recently purified recombinant TagF and demonstrated biochemically that it is responsible for the Tag polymerase activity (20). TagB, which shows ϳ30% pairwise sequence identity with the carboxyl terminus of TagF, is the protein about which the least is known. Because of its similarity to TagF, and a plausible chemical requirement for a teichoic acid priming activity, TagB was proposed to function as a Tag primase, adding a glycerol phosphate moiety to the linkage unit in advance of polymerization by TagF (15). However, the complete lack of characterization of TagB and paucity of available data on TagF means that virtually nothing is known about how these enzymes carry out catalysis, and it remains unclear how truly related the polymerases and primases are mechanistically.
A growing family of TagF-like enzymes is currently being annotated while a significant void remains in our knowledge concerning the mechanisms and the very functions of the proteins within it. This family, which includes both TagF and TagB, remains poorly understood, in part because its members show no sequence similarity to any proteins of known function, which makes any structural or mechanistic predictions difficult. The TagF enzyme remains the prototype for proteins of this type and offers the best opportunity for gaining a further understanding about a family that includes the suspected teichoic acid primases and polymerases from a wide range of organisms expressing chemically distinct teichoic acid polymers. Here we describe mutational and physical analysis of the TagF enzyme and provide the first information about its structure and mechanism. We extend these results to TagB through in vivo complementation studies using corresponding mutants and provide a mechanistic and functional link between the two enzymes and the broader TagF-like family. In conjunction with the accompanying report (21), these observations also serve to support the contention that TagB is the Tag primase.

MATERIALS AND METHODS
General Methods-Strains used in this study are listed in TABLE ONE. Oligonucleotides and plasmids used in this work are described in supplemental TABLES IS and IIS, respectively. All cultures were grown in LB medium (22) supplemented with antibiotics or xylose where necessary. Antibiotics were used at the following concentrations: 75 g/ml ampicillin, 10 g/ml chloramphenicol, and 25 g/ml kanamycin. Cloning was performed in the E. coli strain Novablue (Novagen) according to established protocols (22). Transformations in B. subtilis were carried out according to procedures described previously (23,24). Restriction enzymes, T4 DNA ligase, and Vent polymerase were from New England Biolabs (Beverly, MA). Hotstar Taq polymerase was from Qiagen (Mississauga, Ontario, Canada). The Gateway cloning system was from Invitrogen. [U-14 C]Glycerol-3-phosphate, Ni 2ϩ chelating columns, and Superdex 200 columns were from Amersham Biosciences. High-performance liquid chromatography columns and scintillation fluid were from Waters (Mississauga, Ontario, Canada). Chromatography was carried out using either an Amersham Biosciences Ä kta fast-protein liquid chromatography system or Waters High-performance liquid chromatography system. Filters were from Millipore (Nepean, Ontario, Canada). Mouse ␣-His antibodies were from Amersham Biosciences, donkey ␣-mouse-horseradish peroxidase and donkey ␣-rabbit-horseradish peroxidase antibodies were from Bio/Can Scientific (Mississauga). Polyclonal rabbit ␣-TagF and ␣-TagB antibodies were raised for us by Cocalico Biologicals (Reamstown, PA) using purified full-length protein. Dithiothreitol, isopropyl-␤-D-thiogalactopyranoside, imidazole, MOPS, and ampicillin were from Bioshop (Burlington, Ontario, Canada). Potassium phosphate was from EM Science (Darmstadt, Germany). Protease Inhibitor Mixture Set III was from Calbiochem. CDPglycerol and CDP-[U-14 C]glycerol were synthesized using glycerol-3phosphate cytidylyltransferase from Staphylococcus aureus and previously described methods (25). All other reagents were from Sigma.
TagF Purification-The tagF gene was cloned into pDEST-14 using primers JS06 and JS34, which were designed to place a hexahistidine tag at the carboxyl terminus of the protein. The enzyme was then purified as previously described (20). The variant enzymes described in this work were all similarly purified.
Tag Polymerase Assay-Polymerase reactions were carried out as previously described (20). Briefly, 250-l reactions containing 3.33 mg of protein/ml of B. subtilis membranes and 50 nM purified TagF protein were initiated upon addition of CDP-glycerol to 150 -4000 M. Each reaction contained 0.2 Ci of CDP-[U-14 C]glycerol. Reactions were allowed to proceed at room temperature for 20 min before being quenched by the addition of urea to 4 M. Membranes were then separated from the supernatant by ultracentrifugation and washed twice in reaction buffer. Finally, all supernatants and pellets were analyzed by liquid scintillation counting.
pH Effects for TagF-Steady-state kinetic parameters were determined for the native enzyme between pH 5.5 and 8.0 using the following overlapping buffers: 25 mM MES-NaOH (pH 5.5-6.5), 25 mM MOPS-NaOH (pH 6.5-7.5), and 25 mM EPPS-NaOH (pH 7.5-8.0). The ionic strength of each buffer solution was calculated using the Henderson-Hasselbach equation and was adjusted to 20 mM through the addition of NaCl. At overlapping pH levels (i.e. pH 6.5 and pH 7.5) the enzyme was assayed in both buffer systems, and all data were included. The data were analyzed using Sigma Plot 8.0 (SPSS Scientific, Chicago, IL) and were fit using non-linear regression to the following equation for a single ionization, where represents the first order (k cat ) or second order (k cat /K m ) rate constant, and maxlim is the pH-independent value.
Site-directed Mutagenesis-The tagF mutants described were generated using the QuikChange method (Stratagene, La Jolla, CA). Primers JS35-JS41 and their reverse complements were used as primer pairs in the reactions. pDEST14-tagF was used as template. Mutant alleles were then amplified by PCR using primers JS04 and JS05 and cloned into pSWEET (26) for in vivo complementation. Each clone was confirmed by nucleotide sequencing.
To generate equivalent mutations in tagB, the gene was first amplified by PCR from the chromosome of B. subtilis 168 using primers tag-Boptrbsfor and tagBrev3 and cloned into the EcoRV site of pBluescript II SKϩ (Stratagene, La Jolla, CA). Single crossover PCR was then used to introduce the desired mutations. 27-mer primer pairs encompassing four codons flanking the desired mutagenic codon were used to amplify tagB in two fragments using the T7 and T3 primers (the latter two primers bind to plasmid sequence). Following amplification, parental plasmid was digested using DpnI. Both tagB fragments were gel-purified and used as template for a second PCR reaction using T7 and T3 primers. Full-length product was gel-purified and cloned into pRB374 (27) and pSWEET using appropriate restriction sites. All clones were confirmed by nucleotide sequencing.
Native Proteolysis-TagF protein (10 g) was incubated in proteolysis buffer (50 mM HEPES, 40 mM KCl, 10 mM MgCl 2 , pH 7.3) containing 2 g/ml trypsin or chymotrypsin for 0, 5, 15, or 30 min. Reactions were quenched by the addition of phenylmethylsulfonyl fluoride to 0.5 mM followed by the addition of boiling Laemmli buffer (28). Samples were then analyzed by discontinuous SDS-PAGE using 3% stacking and 12% separating gels visualized with Coomassie Blue stain. Protease protection by the substrate was tested in the presence of CDP-glycerol (2 mM).
Conditional Complementation-tagF and tagB mutant alleles were cloned into the pSWEET chromosomal integration vector. The plas-mids were linearized by restriction digest, and the linear DNA fragments containing the mutant alleles under the control of the P xyl promoter were transformed into strains EB247 or EB486. Successful chromosomal integrants (containing both the temperature-sensitive allele and a complementing tagF/tagB mutant allele) were grown in the presence of 2% xylose at 47°C. Cultures were grown to A 600 ϭ 1.0, normalized for cell density, and serially diluted. 2 l of each diluted sample were spotted onto LB-2% xylose agar and grown at 47°C.
Western Analysis of Complementation Strains-Cell lysates were analyzed by Western blot to test for expression of the variant proteins. Expression of TagB from chromosomally integrated pSWEET was undetectable by Western blot, so strains harboring the mutant tagB alleles on pRB374 were used to test expression of these variants. 4-ml cultures of each strain were grown at 30°C under conditions for expression from the mutant alleles. Cultures were grown to A 600 ϭ 0.5 (0.3 for tagB), pelleted, and resuspended in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0). Samples were then treated with lysozyme (1 mg/ml) and finally boiled in Laemmli buffer (28). The resulting samples were then analyzed by Western blot using ␣-TagF or ␣-TagB polyclonal antibody and ␣-FtsY polyclonal antibody as loading control. 6 Tag Improves Yield of TagF Protein-Previously we reported the purification and characterization of amino-terminally His 6 -tagged TagF protein (20). It was noted that this protein was difficult to purify to a substantial yield. Wild-type and mutant tagF alleles described in this study were cloned into pDEST-14 and engineered to be expressed with a carboxyl-terminal His 6 tag. Using this strategy, we were able to achieve an increase of ϳ10-fold in protein yield versus the amino-terminally tagged enzyme (ϳ0.25 mg/liter versus ϳ2.5 mg/liter of culture). Steady-state kinetics showed no difference between the two tagged versions of TagF (data not shown). Carboxyl-terminally tagged enzyme was stable when stored at Ϫ80°C for several months with no appreciable loss in activity.

Carboxyl-terminal His
pH Dependence of TagF Activity-Nothing about the TagF reaction mechanism was known or could be inferred a priori from similarity to any known enzyme. We assayed the enzyme at various pH levels to identify any ionizable groups that might be important for catalysis. Steady-state kinetic parameters were determined for the native enzyme between pH 5.5 and 8.0 using a series of overlapping buffer systems (see "Materials and Methods"). Fig. 1 shows the relationships between the apparent first (k cat ) and second (k cat /K m ) order rate constants for the TagF reaction versus pH. No value is reported at pH 5.5 in the plot of log(k cat ) versus pH, because at this pH the underlying velocity versus substrate concentration curve showed a linear relationship through all substrate concentrations tested. Therefore a value for k cat could not be extracted, although k cat /K m could still be determined from the slope of that line. The plots of log(k cat ) versus pH and log(k cat /K m ) versus pH both show an increase with pH with a slope of one followed by a plateau at higher pH. When fit to the equation for a single ionization (Equation 1), both relationships gave a pK a for the ionizable group near neutral (6.7 Ϯ 0.2 and 7.0 Ϯ 0.1, respectively). This suggests the presence of a group that must be deprotonated for function (i.e. activity is inhibited by high [H ϩ ]) in both the free enzyme (or the substrate) and the enzyme-substrate complex (29).
Mutational Analysis of TagF-The identification of an important ionizable group with a pK a near neutral implicated histidine as a possible critical residue in the function of TagF. Fig. 2 shows an amino acid sequence alignment of the carboxyl-terminal region of TagF against several other enzymes predicted to be involved in teichoic acid biosyn-thesis in B. subtilis strains 168 and W23 and Staphylococcus epidermidis. It is noteworthy that these strains produce different teichoic acids (poly(glycerol phosphate) for strain 168 and S. epidermidis and poly(ribitol phosphate) for strain W23) yet express enzymes with a high degree of sequence homology to the Tag polymerase in our model B. subtilis strain 168. Two absolutely conserved histidine residues were identified at positions 474 and 612 of TagF. In addition, five other conserved acidic residues were deemed interesting, because the Tag polymerase has previously been shown to require Mg 2ϩ (or Ca 2ϩ ) ions for catalysis (17,18). In a survey of known crystal structures, magnesium (and calcium) ions were found preferentially bound to hard oxygen ligands (30) which, in the absence of any structural information, justified the selection of the conserved Glu-604, Asp-630, Glu-639, Asp-645, and Asp-650 residues for further study. The seven identified residues were individually substituted to alanine by site-directed mutagenesis and subsequently expressed and purified to examine their effects on catalysis.
Each of the TagF variants purified in this study was tested for activity using the Tag polymerase assay (see "Materials and Methods"). TABLE TWO presents the results of these experiments in comparison to the wild-type enzyme. The Asp 3 Ala and Glu 3 Ala substitutions resulted in only minor changes in the second order rate constant k cat /K m . These variants displayed 4-fold or smaller reductions in k cat /K m with no individual change in k cat or K m being greater than 2-or 4-fold, respectively. These effects were likely due to small changes in local environment accompanying the non-conservative amino acid substitutions. The D650A variant did not express in E. coli, which suggests that this residue may play a significant role in maintaining TagF structure. In contrast to the acidic variants, neither histidine variant exhibited any detectable activity under standard assay conditions. To assay the intrinsically low activity of the H474A variant, the enzyme concentration was increased to 500 nM, and 0.4 Ci of CDP-[U-14 C]glycerol was used in a 60-min reaction. Under these conditions, we were able to extract the first order rate constant V max /K m from the slope of a linear velocity versus substrate concentration plot for the H474A variant, but the H612A variant still exhibited no detectable activity. The dramatic effects of the histidine substitutions on enzyme activity (ϳ5000-fold reduction in k cat /K m , TABLE TWO) suggested that these residues were important for polymerase activity.
Identification of a Protease-sensitive Site within the Carboxyl-terminal Region of TagF-TagF was incubated in the presence of trypsin and chymotrypsin in an attempt to gain some information about its domain structure and to identify a protease fingerprint associated with properly folded protein. Digestion of TagF by trypsin resulted in the generation of two fragments, which were stable for at least 30 min under reaction conditions (Fig. 3A). A nearly identical pattern was observed after digestion with chymotrypsin (data not shown), which suggested the presence of a protease-hypersensitive region within the enzyme's structure. Western analysis of the digestion products using ␣-His 6 antibody showed that the smaller 21-kDa fragment contained the carboxyl terminus of the protein (data not shown) and placed the theoretical site of cleavage within the region of high similarity among TagF-like proteins. The carboxyl-terminal fragment was transferred to polyvinylidene difluoride membrane, excised, and subjected to amino-terminal sequencing. The amino-terminal sequence of the 21-kDa fragment was identified as YKFDL. This sequence is unique within the TagF primary sequence and is identified by an arrow in Fig. 2.
CDP-glycerol Protects TagF from Protease Digestion-The effects of substrate binding on protease sensitivity of TagF were tested by subjecting the enzyme to trypsin digestion in the presence of 2 mM CDPglycerol. In the presence of its substrate, native TagF was significantly protected from trypsin digestion (Fig. 3A). Control experiments showed that the addition of CDP-glycerol had no effect on the digestion of BSA (data not shown). These results suggested that binding of CDP-glycerol to TagF either physically blocked the cleavage site or caused a conformational rearrangement that rendered the site no longer accessible. Together, the protease fingerprint of the free enzyme combined with protease protection in the presence of substrate provided us with a robust test that we could use to assess proper folding of the TagF variants. Each variant enzyme was tested in this way and found to be indistinguishable from the wild-type enzyme (Fig. 3B shows the results for the H474A variant).
Conditional Complementation of the tagF1 Defect Using Mutant tagF Alleles-The mutant tagF alleles were cloned into pSWEET, integrated into the chromosome of strain EB247 (tagF1 temperature-sensitive strain), and tested for their ability to complement the lethal defect. We have shown previously (20) that wild-type tagF is capable of full complementation of this defect (strain EB311), thus confirming that the tagF1 mutation alone is responsible for the observed phenotype and providing a manner in which to test proper in vivo function. Fig. 4A shows the results of dilution experiments performed on each of the conditionally complemented mutant strains to observe whether they were capable of suppressing the lethal tagF1 phenotype. Successive 10-fold dilutions of cells were plated on LB-agar containing 2% xylose and grown at 47°C. At this temperature, only the gene products expressed from pSWEET could be expected to carry out the essential Tag polymerase activity, because the mutant tagF1 gene product is incapable of supporting FIGURE 1. pH dependence of TagF activity. Wild-type TagF protein was assayed with CDP-glycerol as varied substrate. B. subtilis membranes were held constant at 3.33 mg of protein/ml. Steady-state kinetic parameters were determined at varying pH levels using the following buffers: 25 mM MES-NaOH (pH 5.5-6.5), 25 mM MOPS-NaOH (pH 6.5-7.5), and 25 mM EPPS-NaOH (pH 7.5-8.0). These data are reported as log(k cat ) versus pH (A) and log(k cat /K m ) versus pH (B).
growth at this temperature (31)(32)(33). It can clearly be seen from Fig. 4A that the organism could not survive when either the H474A or H612A alleles (strains EB984 and EB985, respectively) were provided to complement the tagF1 defect, whereas the wild-type allele restored robust growth (strain EB311). These results indicate that disruption of either critical histidine residue results in both catastrophic reduction of TagF activity as measured in vitro and elimination of the enzyme's ability to perform its function in vivo. Interestingly, we observed that the D650A   . Alignment of TagF with proteins predicted to be involved in teichoic acid biosynthesis. Amino acid sequence alignment of the carboxyl-terminal region of TagF from B. subtilis strain 168 (_168) against the TagB sequence from B. subtilis strain 168 and the TagF sequence from S. epidermidis (_SE), two organisms which contain poly(glycerol phosphate) teichoic acids. The TagF sequence was also compared with the TarB, TarF, TarK, and TarL sequences from B. subtilis strain W23 (_W23), a poly(ribitol phosphate) teichoic acid containing strain. Sequence alignment was performed using the ClustalW algorithm. Black highlights denote identical amino acids, whereas conservative substitutions are highlighted in gray.
The amino acids at positions marked by asterisks were individually substituted to alanine in this study in order to examine their importance for catalysis. The arrow marks the site of trypsin cleavage within TagF.
mutant allele (strain EB1279) was also incapable of complementation at 47°C. The Identified Critical Residues Are Also Important for TagB Activity-Mutations corresponding to those examined in tagF were made in the B. subtilis 168 tagB gene to determine if the residues important for catalysis were common across the family of TagF-like enzymes. The mutant tagB alleles were cloned into pSWEET and integrated into the chromosome of strain EB486 (tagB temperature-sensitive strain). Fig. 4B shows the results of dilution experiments performed as above for the tagB complementation strains. As was seen for tagF, alleles containing the two corresponding histidine mutations in tagB (H126A and H253A) were incapable of restoring growth at the non-permissive temperature(strainsEB1216andEB1217,respectively).Alsoconsistentwasthe inability of the tagF D650A equivalent (tagB D290A, strain EB1229) to complement the defect.
Failure of Histidine Mutants to Complement Is Not Due to Lack of Expression-Western blot analysis was used to test the expression of the TagF and TagB variants in vivo and relate that expression to the results observed in the conditional complementation experiments. Fig. 5 shows expression of both histidine variants of TagF and TagB that was comparable with the level of expression of wild type protein seen in control strains EB311 and EB894, respectively. This result confirms that the failure of the histidine variants to complement was due to a lack of activity and not aberrant expression. The TagF D650A, and corresponding TagB D290A, variants were also expressed, although they were incapable of suppressing their respective temperature-sensitive defects. Thus these residues were also observed to play an important role in vivo, although their impact on in vitro activity was impossible to determine.

DISCUSSION
The identification of the TagF enzyme as the Tag polymerase in the model organism B. subtilis (19,20) has prompted the annotation of many unknown proteins across a wide variety of Gram-positive organisms into an emerging family of TagF-like proteins. What is particularly interesting about these proteins is that, although they carry out reactions using prototypical nucleotide activated precursors, they share homology with no other proteins outside of their select group. Absent are any obvious nucleotide binding motifs such as the Walker motifs (34) or the HXGH motif found in other cytidylyltransferasas and tRNA synthetases (35), including the glycerol phosphate cytidylyltransferase TagD (13). Also absent is the CDP-alcohol phosphotransferase motif (36) found in the phospholipid synthase family of proteins. As a consequence, almost nothing can be inferred a priori about how the TagF-like family of enzymes carries out their functions.
By assaying the enzyme in buffers of differing pH we showed that TagF is inhibited at high [H ϩ ] and identified an important ionization with a pK a near neutral. This pattern is consistent with the presence of an active site base that must be deprotonated for activity. Alignment of the TagF primary sequence against other predicted teichoic acid synthetic enzymes identified many notable similarities. Of particular interest in light of the pH dependence of TagF activity were two conserved histidines at positions 474 and 612. Substitution of these two residues to alanine abolished activity (ϳ5000-fold reduction in k cat /K m ), whereas alteration of several other conserved acidic residues, which where also selected for their potential role in catalysis or metal binding, showed little effect in vitro. Chemical intuition would suggest that the removal of a proton from the terminal hydroxyl group of an extending poly(glycerol phosphate) chain to promote nucleophilic attack upon the pyrophosphate linkage of CDP-glycerol is a plausible and simple mechanism by which TagF might perform catalysis that is consistent with the presence of an active site base (Scheme 1). Unfortunately, the exceedingly low residual activities of the histidine variant enzymes has to this point made it technically impossible to test whether they retain the same dependence on pH as the wild-type enzyme or whether alteration of either residue abolished base dependence. These results do not therefore allow for the direct assignment of a catalytic role to any particular residue, rather they have identified several key residues for further investigation toward that end.
In vivo complementation studies using mutant tagF and tagB alleles corroborated the functional importance of the two conserved histidine residues. The histidine mutant alleles were unable to complement the lethal tagF1 temperature-sensitive defect at 47°C, whereas the wildtype allele restored robust growth. Furthermore, Western blot analysis of lysates from these same complementation strains confirmed that the variant enzymes were indeed expressed yet failed to properly function in vivo. It is of significant interest that when the same mutations were made at corresponding positions in the tagB gene, they rendered these alleles incapable of complementing a tagB temperature-sensitive defect. The in vivo experiments, in conjunction with TagF kinetic studies described above, clearly demonstrate the importance of the two conserved histidines in the activity of both the Tag polymerase and the predicted primase. These results therefore make a case for the conservation of mechanism across a family of enzymes that synthesize polymers of distinctly different chemical structures in both pathogenic and non-pathogenic Gram-positive organisms. . Complementation studies using wild type and mutant tagF and tagB alleles. The wild type and mutant tagF (A) and tagB (B) alleles were integrated into the chromosome of their respective temperature-sensitive strains at the amyE locus using the pSWEET system. Shown is the growth of wild-type B. subtilis and the temperaturesensitive tagF and tagB strains (EB274 and EB486, respectively) engineered to express the identified enzyme variants. Each strain was grown in liquid culture at 30°C to A 600 ϭ 1.0, diluted as indicated and spotted (2 l) onto the plates. The plates contained 2% xylose (w/v) and were incubated at 47°C, a temperature that is non-permissive for growth of the temperature-sensitive strains.  TagF (A) and TagB (B) variant enzymes in each of the strains used for conditional complementation experiments was confirmed by Western blot. Strains were grown at 30°C (to allow for the survival of all strains), normalized for cell density, and lysed. Samples were probed using polyclonal ␣-TagF or ␣-TagB antibody. Each sample was also probed using polyclonal ␣-FtsY antibody to ensure equal loading.
In addition to the histidine mutants, it was observed that the D650A (D290A in tagB) mutant allele was also unable to complement the temperature-sensitive defect. The TagF D650A variant did not express in E. coli and could not be purified to test in vitro activity. Consequently, it was also impossible to examine the proteolysis fingerprint of this variant or test for proteolysis protection in the presence of substrate, two experiments that were performed on all other variants to assess proper folding. However, the fact that the D650A variant and the corresponding TagB D290A variant did express in B. subtilis, yet failed to suppress their respective temperature-sensitive defects, raises the possibility that the failure to complement was due to the loss of Asp-650 specifically and not simply gross protein misfolding leading to degradation. Asp-650 sits within one of the most highly conserved regions of TagF (LITD 650 YSSV) and clearly warrants further study, along with His-474 and His-612, to determine what roles these residues play in the enzyme's activity. Work is ongoing in our laboratory toward this end.
With this work we also provide the first insight into the physical structure of the TagF enzyme. Sequence alignments suggested that the active site of the polymerase could possibly be found within the carboxyl-terminal region. We examined variant enzymes containing amino acid substitutions within this region and identified two histidine residues that were critical to function both in vitro and in vivo and one aspartate residue whose alteration affected function in vivo. We further identified a protease sensitive site within the carboxyl-terminal region of TagF that was significantly protected by the addition of substrate. The catalytically impaired histidine variants showed proteolysis protection that was indistinguishable from wild type suggesting that substrate binding was not drastically impaired in these variants and that the carboxyl-terminal region contained residues that were involved in catalysis. Together, these experiments confirm the location of the active site within the carboxyl-terminal region of TagF and may give further information about catalysis. The location of the cleavage site is interesting in the context of the multiple sequence alignment because it falls within a region that is not universally conserved and, because of its susceptibility to proteases, is likely surface-exposed. Protection from proteolysis may have come as a result of CDP-glycerol binding to TagF and physically blocking the cleavage site. This would be consistent with the presence of the active site near the carboxyl terminus of the protein. However, an intriguing possibility is that the binding of the substrate may have caused a conformational rearrangement that rendered the cleavage site no longer accessible. Many processive enzymes contain structures that partially or completely encircle their polymeric substrates/products (reviewed in Ref. 37). In fact, detailed structural studies on several cellulases have revealed that loops that enclose the substrate are a feature that distinguishes processive cellobiohydrolases (exo-cellulases) from non-processive endogluconases (endo-cellulases) (Ref. 38 and reviewed in Ref. 39). Furthermore, movement of such loops has been observed upon substrate binding, which is presumed to allow "grasping" of the polymer followed by processive degradation, thus explaining the pres-ence of both endo-and exo-cellulase activities in enzymes such as Cel6A from Humicola insolulens (40). It is therefore tempting to speculate that the TagF polymerase might also use such a strategy to grasp its substrate and that such a conformational change may have been evidenced in the loss of protease sensitivity in the presence of substrate.
The work presented here has provided the first structural and mechanistic characterization of any protein within the TagF-like family. Mutational analysis gave information on critical residues that were consistent with pH data and, in conjunction with proteolysis experiments, located the active site within the carboxyl terminus of the protein. Our experiments also demonstrated an important link between the TagF and TagB proteins, which justifies their inclusion within the same family and favors the assignment of primase activity to the last enzyme involved in poly(glycerol phosphate) teichoic acid synthesis in B. subtilis whose function remains unresolved. The TagF-like family of enzymes still presents many challenges to our understanding of the molecular mechanisms of teichoic acid biosynthesis, but the identification of the key residues described here makes it possible to move forward and focus on the chemistry of teichoic acid priming and polymerization.