Purification and Functional Characterization of the Glu-tRNAG1” Amidotransferase from Chlamydomonas

The formation of glutaminyl-tRNA (Gln-tRNA) in chloroplasts, and mitochondria occurs in a two-step reaction. This involves misacylation of tRNAG’” with glutamate by glutamyl-tRNA synthetase and subsequent amidation of Glu-tRNAG’” to the correctly acylated Gln-tRNAG’” by a specific amidotransferase Nature 331, Here we demonstrate the existence of this pathway in green algae and de-scribe the purification of the Glu-tRNAG’” amidotransferase from Chlamydomonas reinhardtii. The purified enzyme showed an M, of approximately 120,000 when analyzed by glycerol gradient sedimentation and gel filtration.

The formation of glutaminyl-tRNA (Gln-tRNA) in Bacilli, chloroplasts, and mitochondria occurs in a twostep reaction.
Here we demonstrate the existence of this pathway in green algae and describe the purification of the Glu-tRNAG'" amidotransferase from Chlamydomonas reinhardtii.
The purified enzyme showed an M, of approximately 120,000 when analyzed by glycerol gradient sedimentation and gel filtration.
An apparent M, of 63,000 of the denatured protein was demonstrated by electrophoresis on sodium dodecyl sulfate-polyacrylamide gels. This indicates that the enzyme possesses an (Ye structure. The substrate for the purified enzyme is Glu-tRNAG'" but not Glu-tRNAG'".
The enzyme requires ATP, Mg'+, and an amide donor for the conversion.
Acceptable amide donors are glutamine, asparagine, and ammonia. Blocking of the glutamine-dependent reaction by alkylation of the protein with 6-diazo-5-oxonorleucine did not inhibit the ammonia-dependent reaction, suggesting that the enzyme has separate glutamine and ammonia binding sites. As suggested by Wilcox (Wilcox, M. (1969) Eur. J. Biochem. 11, 405-412) the amidation reaction may involve glutamyl-phosphate formation, since ATP is cleaved to ADP when the enzyme is incubated with Glu-tRNAG'" and ATP. In common with other glutamine amidotransferases, the enzyme also possesses low glutaminase activity. The purified Glu-tRNAG'" amidotransferase forms a stable complex with Glu-tRNAG'" in the presence of ATP and Mg2+ but in the absence of the amide donor as determined by gradient centrifugation.
Also, the glutamine-dependent activity can be selectively blocked by alkylation of an active site cysteine with the glutamine analogs 6-diazo-5-oxonorleucine or 2amino-oxo-5-chloropentanate (Hartman, 1963;Tso et al., 1982). Modification by alkylation has little effect on the ammonia-dependent reaction suggesting separate binding sites for glutamine and ammonia (Pate1 et al., 1977;Messenger and Zalkin, 1979). Several glutamine amidotransferases also possess glutaminase activity (Zalkin, 1985;Zalkin and Truitt, 1977). Amino acid sequence comparison of these enzymes obtained with the results of enzymatic studies provided evidence for two distinct types of glutamine amidotransferases designated as the trpG-andpuFF-type subfamilies (Weng and Zalkin, 1987;Mei and Zalkin;. The Glu-tRNA"'" amidotransferase is unique among this class of enzymes in its use of Glu-tRNA as amide acceptor. The high tRNA specificity demonstrated by the Bacillus enzyme suggests that the protein is able to recognize tRNA structure (Wilcox, 1969). The unique features of this enzyme and its important biological role in providing correctly charged tRNA for protein biosynthesis make this enzyme a very interesting object of study. As there is little known about the structure and properties of these tRNA-dependent enzymes, we decided to purify the Glu-tRNA"'" amidotransferase from Chlamydomonas reinhardtii.
were incubated in a total volume of 50 gl of 100 mM Hepes, pH 7.5,lO mM MgC12, 10% glycerol, and 3 mM DTT with the indicated amounts of AdT in the presence of 1 mM ATP and 1 mM glutamine for 20 min at 37 "C. Reactions were stopped and the tRNAs were deacylated by the addition of 0.4 ml of 20 mM KOH.
The solution was neutralized with 81 ~1 of 100 mM HCl before the ["Clglutamine was isolated by chromatography on Dowex-1 at neutral pH and quantitated by liquid scintillation counting.
Aliquots of each fraction were aminoacylated with glutamate as described above, and the total acceptor activity per fraction was calculated. The tRNA"" was identified by analyzing all Glu acceptor activity peaks for possible AdT activity as described above.
tRNA Samples-For the preparation of partially purified tRNAG'" and tRNAG'" samples, total tRNA from B. subtilis, Synechocystis 6803, and C. reinhardtii were fractionated by RPC-5 reversed-phase chromatography (see Peterson et al., 1988). Aminoacylation and AdT experiments showed that the tRNAo'" and tRNAo'" fractions were separated from each other but still contained other tRNA species. Their glutamate acceptor activity is given in Table II. Purification of the Glu-tRNAG'" Amidotransferase-An exact description of the preparation and use of the chromatographic resins employed in this study have been given (Chen et al.,-1990a, Wald-Schmidt et al.. 1988). The purification is summarized in Table I as substrate.
As the B. subtilis tRNA was more readily available than C. reinhardtii tRNA we used the bacterial tRNA as substrate in the purification of AdT activity from C. reinhurdtii.
In the purification scheme we took care to enrich only the tRNAdependent activity by checking in parallel assays for the conversion of free glutamate (instead of Glu-tRNA) to glutamine. In addition, since it is known that glutamine synthetase can also carry out at low efficiency the conversion of Glu-tRNAG'" + Gln-tRNA"" (Strauch et al., 1988), we assayed also specifically for this enzyme (Lipmann and Tuttle, 1945) during the purification.
Purification of AdT from whole cell S-100 extracts was by four chromatographic steps (summarized in Table I). The first purification step was DEAE-cellulose chromatography to remove nucleic acids from the extract. Because of the high concentration of tRNA in the S-100 exact specific activity of AdT could not be determined.
Assuming an overall enzyme recovery of 50% we estimate a l-fold "purification" after this first step. The next purification by separation on phosphocellulose did not bind approximately 90% of the protein applied to the column. The AdT activity was recovered with elution by 250 mM KCl, yielding a 5fold purification.
At this step all glutamine synthetase activity was found in the flow-through of the column, separated from the AdT activity. After this step only tRNA-dependent Glu + Gln conversion was detected in protein-containing fractions. Further purification was achieved on two different FPLC resins. Mono Q chromatography resulted in a g-fold increase in specific activity, whereas the final step, Mono S chromatography, gave a 7fold purification.
The overall purification was at least 315fold. The active fraction contained a single protein band as shown by SDS-polyacrylamide gel electrophoresis (Fig. 2).

Amidotransferase
We determined the molecular weights of the native and denatured enzyme. Two independent methods, rate zonal Glu-tRNA"'"Amidotransferase from C. reinhardtii sedimentation in glycerol gradients and gel filtration through the FPLC matrix Superose 12, were used to estimate the molecular weight of the native enzyme. Pure AdT (5 pg of the Mono S fraction) was centrifuged through a lo-35% glycerol gradient which included 500 mM KC1 to prevent aggregation. Marker proteins were sedimented on a parallel gradient. The peak of AdT activity was located in fraction 7, which corresponds approximately to the position of /3-galactosidase (118 kDa) (Fig. 3A). A similar molecular weight was indicated by the results of gel filtration; P-galactosidase eluted at the same position (fraction 24) as the peak of AdT activity (Fig. 3B). When the active fractions were analyzed by SDS-polyacrylamide gel electrophoresis a single polypeptide of M, = 63,000 was observed (data not shown). The same result was obtained (Fig. 2) when the Mono S fraction with the highest AdT activity was checked. We conclude that the Glu-tRNA"'" amidotransferase is a homodimeric enzyme of M, -120,000 with an apparent M, -63,000 for the subunit.
Absorbance measurements at 260 and 280 nm indicated that the purified enzyme is not associated with DNA or RNA. AdT with micrococcal nuclease has no effect on enzyme activity (data not shown).
Purified Glu-tRNA"'" Amidotransferase Amidates Glu Acylated to tRNA"'" In order to demonstrate that purified AdT catalyzes the tRNA-dependent Glu ---f Gln conversion we carried out AdT reactions in the presence and absence of precharged C. reinhardtii Glu-tRNA"'". After isolation of the aminoacyl-tRNA and deacylation, we determined by thin layer chromatography the nature of the amino acid originally bound to tRNA.

Radioactive
Gln and Glu served as markers. As can be seen in Fig. 4 (lanes 1 and 2) no Glu + Gln conversion occurs in the absence of tRNA. However, lane 4 shows that the majority of ["'C]Glu is converted into ["C]Gln when unlabeled Gln is added as the amide donor. The conversion reaction requires ATP (lane 5).
Substrate Specificity of Glu-tRNA"'" Amidotransferase tRNA-It was known from earlier studies that the barley chloroplast AdT is specific for tRNA"'" (Schon et al., 1988b). As the results in Table II demonstrate, this is also the case for the C. reinhardtii enzyme.
In order to conduct these experiments, we first separated by reversed-phase RPC-5 chromatography the tRNA"'" and tRNA"'" species contained in unfractionated tRNA of C. reinhardtii, B. subtilis, and Synechocystis 6803 (for details see "Experimental Proce-  No tRNA C. reinhardtii tRNA"'" (30%) tRNA"'" (27%) R. subtilis tRNA"'" (73%) tRNA"'" (69%) Synechocystis tRNA"'" (27%) tRNAo'" dures"). When the [Y!]Glu-tRNA"'" and ['4C]Glu-tRNA"'" preparations from these organisms were incubated with the purified C. reinhardtii AdT only Glu-tRNA"" was a substrate in the conversion reaction. Amide Donor-When we examined the ability of the purified AdT to use different amide donors we found that a number of glutamine analogs substituted for glutamine (Table  III). Acetylation of the amino group of glutamine had no effect, whereas removal of the carboxyl and amino function in the valeramide analog led to a sharp decrease in activity. D-Glutamine was not accepted by the enzyme, indicating clear stereospecificity for the L-form by the enzyme. Asparagine is also an acceptable amide donor albeit with much lower efficiency. Interestingly, ammonia can also serve as a donor.
AZ'-Some glutamine amidotransferases use ATP as an energy source which in some cases is cleaved between the CX-P-phosphate and in others between the P-y-phosphate. To analyze the exact cleavage position of the ATP molecule in the AdT reaction, we performed assays under normal conditions using the ATP analogs AMP-PCP and AMP-CPP, in which the methylene substitution for oxygen prevents cleavage between phosphates connected by the carbon bridge. The fact that AMP-CPP could support the amidation reaction while AMP-PCP could not (Table III) suggested that ATP cleavage occurs between the @-and y-phosphate. This conclusion is also supported by the ATP hydrolysis experiment presented below. The activation is specific for ATP as GTP could not replace it (Table III).

Glu + Gin Conversion
Requires Activation by

Phosphorylation
Earlier experiments (Wilcox, 1969) provided evidence that ATP is required in the reaction to activate the y-carboxyl group of glutamate which is bound to tRNA. As a first step we wanted to investigate the site of cleavage of the phosphates of ATP. Therefore, we incubated AdT [a-32P]ATP either alone or in presence of tRNA"'" or Glu-tRNAG'". The reaction was analyzed by TLC for the appearance of radioactive ADP indicating ATP conversion (Fig. 5). The enzyme did not bring about any ATP cleavage when incubated with ATP alone or in the presence of uncharged tRNAG'" (Fig. 5, lanes 1 and 2). However, when charged Glu-tRNA"" was used, ATP was converted to ADP (Fig. 5, lane 3). This result is in agreement with the utilization of AMP-CPP, but not AMP-PCP, instead of ATP in the AdT reaction (see above).
ATP may be required in this reaction in order to provide a phosphorylated enzyme intermediate (for discussion see Buchanan, 1973). Therefore we performed an ATP-P, exchange experiment in the presence of large amounts of ADP and [32P]orthophosphate and checked for the formation of radioactive ATP by TLC as described above. Although several Glu-tRNAG'" Amidotransferase from C. reinhardtii concentrations of enzyme and substrate were used we were unable to detect labeled ATP. Thus, it appears that in the absence of tRNA or amide donor the enzyme cannot form a phosphorylated intermediate.

Are There Two Amido Donor Binding Sites on the Enzyme?
It is well known that some glutamine amidotransferases use glutamine and also ammonium chloride as amido donors (Zalkin, 1985). To distinguish between the two binding sites selective alkylation (of a cysteine residue) by the glutamine analog 6-diazo&oxo-L-norleucine (DON) has been used (Tso et al., 1982) which abolished the enzyme's ability to use glutamine but not ammonia. We therefore preincubated AdT (Mono S fraction) without amide donor but in the presence of all other reaction components with different amounts of DON. After starting the reactions by the addition of the amide donors (glutamine or ammonium chloride), the amidotransferase activity was measured. Parallel reactions without inhibitor inactivation served as control. Fig. 6 shows that DON only affected the glutamine-dependent reaction. This result indicates the existence of separate binding sites for glutamine and ammonia on the AdT molecule. C. reinhardtii AdT differs in this respect from the B. subtilis enzyme, where DON cannot inhibit the glutamine-dependent activity in crude extracts (Strauch et al., 1988).

Glu-tRNAG'"
Amidotransferase Has Low Glutaminase Activity Since it is known that glutamine amidotransferases possess glutaminase activity which is sometimes stimulated by the products of the amidotransferase forward reaction we checked the glutaminase activity of purified AdT. Under the conditions of the normal AdT reaction the enzyme (Mono S fraction) was able to convert ['*C]Gln to [Y]Glu.
In the absence of added components, AdT was able to convert in 120 min 2.5% of input Gln (20 mM) into Glu. However, this low glutaminase activity was stimulated (to 7.5%) by the presence of ADP, Pi, Mg2+ ions, and pre-charged Glu-tRNA"'".
Without the addition of enzyme, almost no conversion was detectable.
Purified Glu-tRNA"'" Amidotransferase Forms a Stable Complex with Glu-tRNAGL" As a first step towards the investigation of the reaction mechanism of the C. reinhardtii AdT, we wanted to examine the ability of the enzyme to form a complex with its tRNA substrate. Wilcox (1969) demonstrated complex formation of the B. subtilis AdT with its substrate by binding of precharged Glu-tRNAG'" to a protein out of a crude column fraction and isolation of the complex by gel filtration.
Since we had previously achieved separation of charged tRNA from different tRNA binding enzymes under mild conditions with glycerol gradient sedimentation (Chen et al., 1990a), we applied this method to the analysis of the initial process of the enzyme reaction, the complex formation with the mischarged tRNA.
We assumed that stable complexes may be obtained in the absence of the amide donor so the enzymatic amide transfer by AdT can not proceed. As seen in Fig. 70, AdT does form a stable complex with Glu-tRNA"", which is not present in the control gradients with Glu-tRNAG'" (Fig. 7B) and the enzyme alone (Fig. 7A). The complex did not form with uncharged tRNAG'", tRNA"'", and charged Glu-tRNA"", indicating a distinct substrate specificity for mischarged Glu-tRNA"'" (data not shown). Moreover, the absence of any complex without ATP in the assay demonstrated the dependence of complex formation on the presence of ATP (Fig. 7C).

DISCUSSION
The formation of Gln-tRNA"'" in Gram-positive eubacteria, archaebacteria, and in organelles (Fig. 1) requires the presence