Critical Aspartic Acid Residues in Pseudouridine Synthases*

The pseudouridine synthases catalyze the isomerization of uridine to pseudouridine at particular positions in certain RNA molecules. Genomic data base searches and sequence alignments using the first four identified pseudouridine synthases led Koonin (Koonin, E. V. (1996) Nucleic Acids Res. 24, 2411–2415) and, independently, Santi and co-workers (Gustafsson, C., Reid, R., Greene, P. J., and Santi, D. V. (1996)Nucleic Acids Res. 24, 3756–3762) to group this class of enzyme into four families, which display no statistically significant global sequence similarity to each other. Upon further scrutiny (Huang, H. L., Pookanjanatavip, M., Gu, X. G., and Santi, D. V. (1998)Biochemistry 37, 344–351), the Santi group discovered that a single aspartic acid residue is the only amino acid present in all of the aligned sequences; they then demonstrated that this aspartic acid residue is catalytically essential in one pseudouridine synthase. To test the functional significance of the sequence alignments in light of the global dissimilarity between the pseudouridine synthase families, we changed the aspartic acid residue in representatives of two additional families to both alanine and cysteine: the mutant enzymes are catalytically inactive but retain the ability to bind tRNA substrate. We have also verified that the mutant enzymes do not release uracil from the substrate at a rate significant relative to turnover by the wild-type pseudouridine synthases. Our results clearly show that the aligned aspartic acid residue is critical for the catalytic activity of pseudouridine synthases from two additional families of these enzymes, supporting the predictive power of the sequence alignments and suggesting that the sequence motif containing the aligned aspartic acid residue might be a prerequisite for pseudouridine synthase function.

All organisms chemically modify their RNA after transcription, and the isomerization of uridine to its C-glycoside isomer pseudouridine (⌿) 1 is the most prevalent modification, Fig. 1 (1). This isomerization is catalyzed by the pseudouridine synthases, enzymes that display specificity for U residues at particular positions in certain RNA molecules, a specificity that can range from handling a single specific site to mild promiscuity (2)(3)(4)(5)(6)(7). Physiological ramifications resulting from the lack of ⌿ at particular locations have recently become evident, mandating a fuller understanding of ⌿ generation in particular and RNA modification generally.
In Escherichia coli, severe growth inhibition results from disruption of rluD (formerly denoted sfhB or yfiI), which encodes the ⌿ synthase responsible for isomerization of U residues at positions 1911, 1915, and 1917 of 23 S rRNA (6,7). In eukaryotes, Steitz and co-workers (8) have elegantly demonstrated that the presence of ⌿ in the U2 small nuclear RNA is required for proper assembly of the spliceosome, work that relied on the inhibition of the responsible ⌿ synthase(s) by U2 transcripts containing 5-fluorouridine. Such inhibition of ⌿ synthases by RNA containing 5-fluorouracil is well precedented (9 -11), and this inhibition may account for a secondary mode of action of the anticancer drug 5-fluorouracil, which primarily acts by inhibiting thymidylate synthase (12). Consistent with 5-fluorouracil cytotoxicity resulting from ⌿ synthase inhibition is a long string of observations concerning cell lines treated with both 5-fluorouracil and thymidine (eliminating the need for thymidylate synthase). This treatment affects many RNAmediated events, including disruption of rRNA maturation (13,14), disruption of pre-mRNA splicing (15)(16)(17), and, perhaps, reducing translational accuracy (18). The link is thus established between ⌿ synthases and critical RNA-mediated cellular processes, the disruption of which can lead to dire consequences.
One such consequence is likely the X-linked human disease dyskeratosis congenita. Young men suffering from this disease have blotchy skin, poor dental health, sparse hair (including a lack of eyebrows), and evanescent nails; these men also tend to develop gastrointestinal tumors and suffer bone marrow failure (19). The gene responsible for dyskeratosis congenita has recently been identified and encodes a protein dubbed dyskerin, which contains a nuclear localization sequence and has two stretches of amino acids highly similar to the E. coli ⌿ synthase TruB (20). Although the detected similarity to TruB is in itself rather weak evidence for concluding that dyskerin is a ⌿ synthase, the sequence alignments of the known ⌿ synthases support that assessment (see below), and probing the functional significance of these alignments was the purpose of the experiments described in this communication. A brief presentation of the alignments is, therefore, imperative before recounting and discussing the experiments.
After Penhoet and co-workers (9) and Ofengand and coworkers (2-4) cloned the first four ⌿ synthase genes, both Koonin (21) and Santi and co-workers (22) undertook alignments and data base searches with these ⌿ synthases. Both studies found statistically insignificant identity between the four proteins over their entire length, but both studies also determined that three short stretches of amino acids (5-13 residues) displayed significant similarity. Interestingly, each of the four ⌿ synthases (RluA, RsuA, TruB, and TruA, which was formerly named either HisT or ⌿ synthase I) had clear homologs in the data base, which led Koonin (21) to conclude that each of the first four cloned ⌿ synthases represented a different family of these enzymes. Indeed, similarity to the known ⌿ synthases has been used to target unidentified open reading frames as ⌿ synthases, and this strategy has led to the cloning of eight more ⌿ synthases from E. coli, yeast, mouse, and human (as a partial clone) (6,7,(23)(24)(25)(26). The degree of relatedness among the four families, however, was rather ambiguous. In fact, as a result of the relatively low levels of similarity, each report of these alignments omitted at least one of the genes from the figure showing the alignments: Koonin (21) excluded TruA, and Santi and co-workers (22) excluded both TruA and TruB.
Reexamination of the sequence data by Santi and co-workers (27) led to the insight that among all of the ⌿ synthases and their identified homologs, a single residue, an aspartic acid, is found aligned in all sequences. This aspartic acid residue in TruA (Asp-60) was mutated to alanine, asparagine, glutamate, serine, and lysine, and all of the mutant TruA proteins were catalytically inactive although still able to bind tRNA with near wild-type affinity (27). The critical catalytic participation of the aligned aspartic acid residue strongly supports the functional significance of the aligned sequence motifs. Fig. 2 shows the region, which Koonin (21) named motif II, containing the aligned aspartic acid residue, including the first four cloned ⌿ synthases and dyskerin, which shares strong similarity with TruB in this region (20). Interestingly, this motif was also identified in both deoxycytidine triphosphate deaminase (catalyzes dCTP 3 dUTP) and deoxyuridine triphosphatase (catalyzes dUTP 3 dUMP). Based on this observation, Koonin hypothesized that this stretch of amino acids was involved in uridine binding (21), which appears to conflict with the ability of the TruA mutants to bind tRNA (27). This seeming contradiction between prediction and experimental results and the weak similarity between the four families of ⌿ synthases dictated further testing of the alignment generally and the "conserved" aspartic acid residue specifically. To this end, we undertook the mutation of the aligned aspartic acid residues in two other ⌿ synthases of different families, RluA and TruB.

EXPERIMENTAL PROCEDURES
General-Hen egg white lysozyme was purchased from Sigma. Activated charcoal (Norit SA-3) was purchased from Aldrich. Nucleoside triphosphates and competent JM109(DE3) E. coli cells were purchased from Promega Corp. (Madison, WI). The restriction enzyme BstNI was purchased from New England Biolabs (Beverly, MA). Isopropyl-␤-Dthiogalactopyranoside, HEPES, and Tris were purchased from Roche Molecular Biochemicals. Oligonucleotides were purchased from the Great American Gene Company (Ramona, CA). QuikChange™ sitedirected mutagenesis kits were purchased from Stratagene (La Jolla, CA). [5-3 H]UTP was purchased from Amersham Pharmacia Biotech. Prime RNase inhibitor was purchased from 5 Prime 3 3 Prime, Inc. (Boulder, CO). Ni-NTA superflow resin and RNA/DNA purification kits were purchased from Qiagen (Chatsworth, CA). Competent BLR(DE3) pLysS E. coli cells were purchased from Novagen (Madison, WI). Spectra/Por dialysis tubing (12,000 -14,000 NMWCO) and all other chemicals were purchased from Fisher or its Acros Organics division (Pittsburgh, PA). A Robocylcer Gradient 96 thermal cycler (Stratagene) was used for the polymerase chain reaction component of the site-directed mutagenesis. DNA sequencing was performed in the University of Delaware Cell Biology Core Facility using a Long Readir 4200 DNA sequencer (Li-Cor, Inc., Lincoln, NE).
Plasmids-The plasmid p67CF23, which contains the gene for E. coli tRNA Phe behind a T7 promoter, was a generous gift from O. Uhlenbeck (28). Plasmids containing the genes truB (2) and rluA (3) in pET15b vectors were generously provided by J. Ofengand. Because the reports of these plasmids do not assign them names, they will be referred to as p⌿55 (containing truB) and p⌿746 (containing rluA). The plasmid pT7-911Q for expression of bacteriophage T7 RNA polymerase gene was a gift from T. Shrader, and the overexpressed T7 RNA polymerase was isolated as described (29). All of these overexpressed proteins have a His 6 tag fused to their N terminus to simplify purification.
Site-directed Mutagenesis of TruB and RluA-Site-directed mutagenesis was performed using the QuikChange TM protocol (Stratagene) according to the manufacturer's instructions except for the ther-  (22). The D denotes the critical aspartic acid residue noted by Huang et al. (7). Note that the glycine five amino acids downstream of the D is not present in all identified homologs of these proteins. The numbering excludes His 6 tags added to the N terminus to simplify purification. mal cycling program. Because the Robocycler moves sample tubes between heating blocks maintained at different temperatures, additional time is required for the sample to achieve thermal equilibrium than with a single block thermal cycler (in which the sample temperature equilibrates as the block is adjusted to the new temperature). The final program for the mutagenesis polymerase chain reaction was as follows: 95°C for 1 min; 16 cycles of 95°C for 1 min, 55°C for 1.5 min, and 68°C for 15 min; and holding at 4°C.
The QuikChange TM protocol uses complementary primers containing a mismatch to alter a codon by polymerase chain reaction amplification of the entire plasmid. For TruB mutants, p⌿55 was used; for RluA mutants, the plasmid template was p⌿746. The primers used for each mutation are presented here with the upper primer broken into codons and the altered codon (and its complement on the other primer) denoted in boldface type. TruB  Overexpression and Purification of the ⌿ Synthases-For expression of wild-type TruB and RluA, JM109(DE3) cells were transformed with p⌿55 or p⌿746, respectively, transformed cells were used to inoculate LB medium (500 ml), and the culture in a baffled flask was shaken vigorously at 37°C. When the culture reached A 600 nm ϭ 0.6, isopropyl-␤-D-thiogalactopyranoside (100 mM) was added (final concentration, 1 mM). After 3 h, the cells were harvested at 6000 ϫ g for 20 min at 4°C, quick frozen, and stored at Ϫ80°C.
For purification of the enzyme, the cell pellet was thawed and resuspended in 50 mM sodium phosphate buffer (10 ml), pH 8.0, containing NaCl (300 mM) and imidazole (3 mM). All subsequent steps were performed at 4°C. Lysozyme (100 mg/ml; final concentration, 0.1 mg/ml) was added, and the mixture was stirred for 30 min to effect cell lysis. After sonication to reduce viscosity, the lysate was centrifuged (18,000 ϫ g for 30 min) to pellet cell debris. Ni-NTA superflow resin (2.5 ml) was added to the supernatant, and the mixture was gently stirred for 1.5 h to bind the His-tagged enzyme to the resin. After centrifugation (1100 ϫ g for 3 min), the pelleted resin was resuspended in the same volume of buffer initially used for cell resuspension and packed into a column. The column was washed (three times, 7 ml each) with 50 mM sodium phosphate buffer, pH 8.0, containing NaCl (300 mM) and imidazole (20 mM). Bound enzyme was then eluted with a step (10 ml) of 50 mM sodium phosphate buffer, pH 8.0, containing NaCl (300 mM) and imidazole (500 mM). Eluted TruB was dialyzed overnight against 50 mM HEPES buffer, pH 7.5, containing ammonium chloride (100 mM) and EDTA (0.1 mM) and stored at 4°C. Eluted RluA was dialyzed overnight against 20 mM triethanolamine⅐HCl buffer, pH 7.9, containing NaCl (0.35 M). This concentrated solution was either used within one day or was diluted to 0.24 mg/ml with buffer and then an equal volume of glycerol and stored at Ϫ20°C.
The overexpression and purification of the mutant ⌿ synthases were identical to that of the wild-type enzyme except that the expression host was BLR(DE3) pLysS, which produces small amounts of T4 lysozyme so that a separate treatment with lysozyme was not required. The isolated enzymes were judged to be Ͼ98% pure by SDS-polyacrylamide gel electrophoresis with Coomassie Blue staining, and 15-40 mg of purified protein was obtained from 500 ml of culture.
In Vitro Transcription and Purification of tRNA-Substrate tRNA Phe was synthesized by in vitro transcription catalyzed by T7 RNA polymerase with BstNI-linearized p67CF23 as template. The transcription reaction (1 ml) was carried out in 40 mM Tris⅐HCl buffer, pH 8.1, containing NTPs (4 mM each), MgCl 2 (24 mM), spermidine (1 mM), dithiothreitol (5 mM), Triton X-100 (0.01%), Prime RNase Inhibitor (120 units), GMP (16 mM), and T7 RNA polymerase (0.1 mg/ml). After 9 h at 37°C, protein was extracted into phenol:chloroform:isoamyl alcohol (25:24:1), and the aqueous phase was removed and washed with chloroform:isoamyl alcohol (24:1). RNA was precipitated from the removed aqueous phase by addition of 0.3 M sodium acetate buffer, pH 5.2 (0.1 volume) and then ethanol (95%, 3 volumes). After incubation at Ϫ20°C for at least 2 h, the precipitate was pelleted by centrifugation, and the supernatant was decanted. The pellet was dissolved in 10 mM Tris⅐HCl buffer, pH 7.5, containing EDTA (1 mM), and the tRNA transcript was purified using the Qiagen RNA/DNA system according to the manufacturer's instructions. Preparation of tRNA Phe containing [ (27). This [5-3 H]tRNA was typically diluted with unlabeled tRNA transcript to afford [5-3 H]tRNA of suitable specific activity for the ⌿ synthase assays.
⌿ Synthase Activity Assay-The assay for ⌿ synthase activity was a slight modification of the tritium release assay reported by Nurse et al. (2), which measures the liberation of tritium from C5 when labeled U in RNA is isomerized to ⌿. A typical assay mixture (500 l) was 50 mM HEPES buffer, pH 7.5, containing ammonium chloride (100 mM), dithiothreitol (5 mM After 10 min at room temperature, an aliquot (95 l) of each mixture was very slowly filtered through a 25 mm cellulose nitrate membrane filter (0.45 m; Whatman, Maidstone, United Kingdom), which were prewetted with the HEPES buffer. The filter was rinsed rapidly with 25 mM potassium phosphate buffer, pH 7.4 (two times, 1 ml each). After air-drying, the filter was put into a scintillation vial along with scintillation fluid (5 ml), shaken vigorously, and then counted in a liquid scintillation counter. Each protein concentration provided three filter binding data points, and the computer program GraphPad InPlot was used to plot the data and fit them to the binding curve e⅐tRNA ϭ (e ⌺ )(e⅐tRNA max )/(e ⌺ ϩ K d ), where e ⌺ is the total enzyme concentration, e⅐tRNA is the concentration of tRNA bound to enzyme, e⅐tRNA max is the maximum concentration of tRNA bound to enzyme, and K d is the dissociation constant for the enzyme-tRNA complex. The use of this simplified binding curve is allowed because the concentration of tRNA (10 nM) is 1% of the lowest enzyme concentration (1 M), so that the free enzyme concentration essentially equals the total enzyme concentration.
Test for De-uracilating Activity-An assay mixture was prepared as described above except that the volume was doubled. A mutant ⌿ synthase was added (final concentration, 100 nM), and after 30 min, half of the reaction mixture was quenched and processed as described above.
To the other half of the reaction mixture, wild-type ⌿ synthase was added (final concentration, 100 nM); after an additional 30 min of incubation, the reaction mixture was worked up as described above.

Characterization of the ⌿ Synthase Mutants-
The site-directed mutagenesis proceeded smoothly, affording the D48A and D48C mutant TruB and D64A and D64C mutant RluA. By virtue of an N-terminal His 6 tag, all four mutant enzymes were purified to very near homogeneity by chromatography over a column of Ni-NTA resin. No major differences were noted in the yield, isolation, or storage of the mutant enzymes versus the wild-type enzymes.
Nitrocellulose binding assays (30) were used to probe the ability of the wild-type and mutant ⌿ synthases to bind substrate. A small concentration of [ 3 H]tRNA was incubated with an increasing concentration of enzyme, and protein was adsorbed to nitrocellulose filters, which were subjected to liquid scintillation counting to quantitate bound [ 3 H]tRNA. A representative binding curve is shown in Fig. 3. Binding of [ 3 H]tRNA by enzyme plateaus at 70% of the tritium present, which cor-responds well with the release of only 70% of the theoretical maximum when this batch of substrate was incubated with wild-type TruB. Although the presence of non-substrate tritiated RNA might initially seem disadvantageous, this contamination allowed the determination of the counting efficiency for tritium bound to the filter by quantitation of the tritium in the filtrate for a sample on the binding plateau. Substitution of bovine serum albumin for ⌿ synthase resulted in no retention of [ 3 H]tRNA on the filter (data not shown).
As shown in Table I, the mutant ⌿ synthases all bind tRNA substrate with measured K d Ͻ 2 M. The higher K d of wild-type TruB versus mutants (Table I) likely reflects that the former is interacting with product (⌿ in place), whereas the latter interacts with substrate (U in place). The measured values are probably greater than the true K d values due to expected dissociation from the filter-bound enzyme at this binding affinity. Even with K d ϭ 2 M, under the standard assay conditions (500 nM [ 3 H]tRNA, 50 nM enzyme), a significant part (20%) of the mutant ⌿ synthases would be bound by substrate.
Activity of the ⌿ Synthase Mutants-The ⌿ synthase activity of each mutant was assessed using the well established tritium release assay (2). As shown in Table II and Fig. 4, the mutants did not catalyze the release of tritium from [ 3 H]tRNA significantly above the background levels of the method. To reduce the lower limit of detectable catalytic activity, the amounts of mutant ⌿ synthase included in the incubation mixture was increased past that of the wild-type enzyme in the comparator assay: the tritium released still did not vary significantly from the background of the method, even in one case in which the [ 3 H]tRNA (500 nM) was saturated with 20 M mutant enzyme (D48A TruB).
The tritium release assay utilizes charcoal to adsorb tRNA, and any free uracil produced during incubation of a mutant enzyme with tRNA transcript would also be adsorbed. We therefore tested whether or not the mutant enzymes release free uracil at a rate comparable to the rate of glycosidic bond cleavage by the wild-type ⌿ synthases. Half of an incubation of [5-3 H]tRNA with mutant ⌿ synthase was withdrawn and processed while wild-type ⌿ synthase was added to the other half, which was processed later. In all cases, the incubation with wild-type enzyme resulted in release of the theoretical maximum amount of tritium from [5-3 H]tRNA, as did a control incubation with wild-type enzyme and substrate (Fig. 5). DISCUSSION The mutant ⌿ synthases described here were generated and characterized to probe the value of the sequence alignments (21,22) of the first identified ⌿ synthases. The ready availabil-   ity of sophisticated programs makes it universally possible to probe the rapidly expanding data base for obscure relationships between proteins that do not display overt global similarity. A few hours' work on the Internet can generate hundreds of "related" proteins, but it remains imperative to verify experimentally the functional or structural significance of aligned amino acids. Only after many such studies can empirical measures and instinct develop that will allow accurate sorting of "hits" as informative or not.
To assess the role of the aligned aspartic acid residue in the highly dissimilar families of ⌿ synthases (Fig. 2), Asp-48 in TruB and Asp-64 in RluA were each mutated to alanine and cysteine. Alanine was chosen as a substitution to eliminate all side chain functionality while avoiding potential problems due to the conformational flexibility of glycine. The choice of cysteine was guided by a possible nucleophilic role for the aspartic acid residue, a role that may allow functional substitution of the carboxylate of aspartate by the nucleophilic thiol(ate) of the cysteine side chain. No other mutations were made because a critical role for the aligned aspartic acid residues in TruB and RluA would allow confident extrapolation from the four other replacements in TruA (D60N, D60E, D60S, and D60K) (27) despite the global dissimilarity among TruA, TruB, and RluA. The D48A and D48C mutants of TruB and the D64A and D64C mutants of RluA were, therefore, constructed and tested for ⌿ synthase activity.
The finding that all four mutants are inactive strongly supports the significance of the one aspartic acid residue that is absolutely aligned among all of the known ⌿ synthases and their homologs. By extension, the stretch of aligned amino acids containing the critical aspartic acid residue also plays a role in ⌿ synthase function, although it remains for future experiments to determine whether this stretch of amino acids serves a structural role, plays a more direct role in catalysis, or both. Likewise, further experiments are required to assess whether this stretch of amino acids is conserved, which implies radically divergent evolution of the ⌿ synthases other than in the three aligned motifs, or not conserved, which implies a convergent evolution to provide the three motifs properly positioned for catalysis on four different protein scaffolds. Structural information will be immensely valuable in this regard. The preliminary crystal structure of TruA (31) and crystals of a catalytically active fragment of the ⌿ synthase RluC (32) have been reported, and the determination of structures of members of the other ⌿ synthases families will reveal whether or not they share common folds despite their statistically insignificant global similarity.
The precise role of the critical aspartic acid residues also remains the domain of future experiments. The nitrocellulose binding assay data (Table I, Fig. 3) indicate that the mutant enzymes still bind tRNA efficiently, which argues against a large structural change. However, the CD spectra (data not shown) of the wild-type and mutant enzymes vary in the intensity of the features diagnostic for secondary structure, suggesting that either the folding or, perhaps, the degree of folding is somewhat perturbed by mutation. These differences may be a function of concentration or buffer composition or even arise from differences in interaction between the His 6 tag and the rest of the protein, possibilities currently under investigation. Regardless of the cause (structural, catalytic, or both), the results presented here demonstrate that the aligned aspartic acid residue is, indeed, critical for ⌿ synthase function.
This work does not directly address the possible chemical roles of the aligned aspartic acid residue, which could serve as a nucleophile, as a counter-ion, as a hydrogen-bond acceptor/ donor, or as a general acid/base catalyst. The replacement of aspartic acid with cysteine was chosen, in part, to determine whether or not the nucleophilic thiol group could substitute for the carboxylate. The lack of activity exhibited by D48C TruB and D64C RluA, however, does not speak forcefully to this question, for the shorter length of the side chain of cysteine relative to that of aspartic acid may simply preclude a close enough approach for the mutant enzymes to effect a nucleophilic attack.
The proposal of Koonin (21) that the motif containing the "conserved" aspartic acid residue constituted a uridine binding sequence suggested an alternate explanation of our observations. If the aspartic acid residue were essential for binding the uracil ring, the mutant enzymes might catalyze glycosidic bond cleavage at a rate unimpaired by mutation but then lose their grip on a putative uracil(ate) intermediate (27) and release it into solution. The free uracil would be adsorbed onto the charcoal during the assay protocol, and no free tritium would be detected (as was observed). Mutation of the aligned aspartic acid, then, would convert a ⌿ synthase into a "deuracilase." To test this hypothesis, [ 3 H]tRNA was incubated with each of the described ⌿ synthase mutants. The amount of tritium released by the wild-type enzyme was the theoretical maximum for the amount of substrate present and was identical whether or not the substrate was preincubated with mutant enzymes (Fig. 5). It can be concluded, therefore, that the mu- tant enzymes do not release uracil(ate) into solution at a rate comparable to glycosidic bond breakage in the wild-type enzymes. The possibility remains that the aligned aspartic acid residue is required for the "specific binding" of substrate tRNA necessary for precise positioning of the isomerized uridine in the active site. In this case, only "nonspecific" interaction with tRNA is being detected by the nitrocellulose binding assays. The results do, however, clearly eliminate the possibility that mutation of the aligned aspartic acid residue converts the ⌿ synthases into efficient deuracilases.
The results of this study also strengthen the assignment of ⌿ synthase function to dyskerin, the protein the absence of which leads to the human disease dyskeratosis congenita, for the similarity to TruB in the region containing the critical aspartic acid residue (Fig. 2) is striking (19,20). Heiss et al. (20) proposed that dyskerin is the human homolog of the yeast protein Cbf5p. Very recently, other investigators have independently proposed Cbf5p as a ⌿ synthase that handles small nucleolar RNA on the basis of its localization in a nucleolar protein complex and similarity to TruB in the region of the critical aspartic acid residue (33). Dyskerin (20) and Cbf5p (33) also share similarity with TruB in another stretch of amino acids identified by Santi and co-workers (22) and by Koonin (21), who named it motif I (Fig. 6). The experiments presented here support the assignment of dyskerin and Cbf5p as ⌿ synthases by demonstrating that the shared motif containing the aligned aspartic acid residue has functional significance in two ⌿ synthase families other than that represented by TruA, families that display no statistically significant global sequence similarity to each other. In fact, the presence of the amino acid motif may prove useful as a prerequisite for assigning ⌿ synthase activity to genes of unknown function.