Diadenosine 5’,5’”-P1,P4-tetraphosphate and Related Adenylylated Nucleotides in Salmonella typhimurium”

Salmonella typhimurium LT2 rapidly accumulates high levels of a family of five adenylylated nucleotides following exposure to a bacteriostatic quinone, 6-amino-7-chloro-5,8-dioxoquinoline. These compounds have been analyzed using our recently described two-dimensional thin layer chromatographic method. The five dinucleotides, which cannot be detected in exponentially growing cells, have been identified as diaden-&e ApppGpp AppppG ApppG 5’-guano-sine-5‘-(P’,PS-triphosphate)), and ApppA (diadenosine AppppA has been previ- ously detected in vitro as an enzymatic product of aminoacyl-tRNA synthetases and in vivo at submi-cromolar levels in eucaryotic cells. The induced intra- cellular concentration of AppppA and the other adenylylated nucleotides in S. typhimurium is approxi- mately 100-fold higher than that found in eucaryotic cells. We propose that these dinucleotides are alar- mones, in of M NH4HC03 (pH 8) for 30 min at room temperature. The polyethyl- eneimine cellulose was pelleted by centrifugation in a microfuge for 10 min at 4 "C, the supernatant was removed, and the elution process was repeated. The two supernatant fractions were pooled, concentrated in a spinning lyophilizer overnight, and resuspended in 100 pl of 0.25 M NH,HC03 (pH 8). This procedure gave a high yield of purified compound (approximately 90% recovered). Chemical and Enzymatic Tests-Tests for the characterization of compounds were performed as previously described (17).

Thus far, AppppA has been detected in vivo in lower eucaryotic cells (e.g. Tetrahymena pyriformis (9) and Physarum polycephalum (10)) and in a variety of mammalian cells at very low concentrations of 10-1200 nM (3,4,11). Rapaport and Zamecnik have reported a correlation between increased levels of AppppA and cellular proliferative activity and have proposed a function for this dinucleotide as a possible pleiotypic activator of proliferation (11). This hypothesis is supported by findings that AppppA triggers the initiation of in uitro DNA replication in quiescent mammalian cells (12), binds to HeLa cell DNA polymerase a (13,14), and acts as a primer for DNA synthesis i n uitro (15, 16).
We now find that a family of adenylylated nucleotides dramatically accumulate to high levels in a bacterial cell inhibited by a cytostatic quinone, ACDQ. Using our newly developed technology to resolve cellular nucleotides, we have analyzed the novel phosphorylated compounds that are produced, their induced levels, and the kinetics of their accumulation under these conditions. A preliminary account of part of this work has been presented.' AppppA and Related Adenylylated Nucleotides in S. typhimurium systems) on polyethyleneimine cellulose plates and developed as described (17). In co-chromatography experiments, 2 4 aliquots of 5 mM standard solutions were mixed with 10-pl radioactive samples before application to the TLC plate. Locations of standards were determined either by UV absorbance (17) or the phosphate detection method (19). "P-Labeled compounds were visualized by autoradiography (17). When required, the corresponding spots were cut out from the chromatogram and placed in a vial containing 4.0 ml of Betamax (WestChem Products) scintillation fluid. Radioactivity was determined by scintillation counting in a Packard Tri-Carb 300 scintillation counter.
Purification of Quinone-induced Spots and Other Radiolabeled Nudeotides-A 10-ml culture of S. typhimurium was grown and labeled as described above. Thirty min after the addition of ACDQ, the cells were extracted as above and separated into 1-ml portions. The debris was pelleted by centrifugation in a Beckman microfuge. The supernatant was filtered through an Acrodisc disposable filter assembly (0.2 pm) from Gelman and concentrated in a spinning lyophilizer (Savant Instruments, Inc.) overnight. The concentrate was resuspended in 70 pl of distilled water, spotted on 12 TLC plates, and run in the optimized two-dimensional TLC system. Autoradiograms were used as guides to locate the relevant compounds. After the spots were cut out from the chromatograms, the polyethyleneimine cellulose containing each compound was scraped from the plastic backing. Each phosphorylated compound was eluted in 0.5 ml of 0.25 M NH4HC03 (pH 8) for 30 min a t room temperature. The polyethyleneimine cellulose was pelleted by centrifugation in a microfuge for 10 min a t 4 "C, the supernatant was removed, and the elution process was repeated. The two supernatant fractions were pooled, concentrated in a spinning lyophilizer overnight, and resuspended in 100 pl of 0.25 M NH,HC03 (pH 8). This procedure gave a high yield of purified compound (approximately 90% recovered). Chemical and Enzymatic Tests-Tests for the characterization of compounds were performed as previously described (17).

RESULTS
S. typhimurium LT2 Rapidly Accumulates Six Novel Phosphorylated Compounds after Inhibition by ACDQ-ACDQ is a cytostatic quinone known to inhibit the growth of E. coli (20, 21). Its primary mode of action, a t low concentrations, appears to be the inhibition of charging of tRNA"" by reacting with essential sulfhydryl groups of leucyl-tRNA synthetase, resulting in an active site-directed modification of the enzyme (21-23). Cells exposed to this quinone accumulate ppGpp and pppGpp (at a ratio of 10:l) (20, 22). Nucleotide pool changes induced by ACDQ were investigated using our newly developed technology (17). Fig. 2a shows a schematic diagram of our standard two-dimensional TLC system (17). Fig. 26 is an autoradiogram of a normal R2P-labeled crude extract of S. typhimurium, while an autoradiogram of an extract taken after a 30-min exposure to ACDQ (5 pglml) is depicted in Fig.  2c. It is apparent that a number of spots (as indicated by arrows) rapidly appear, or increase in intensity, in the lower portion of the autoradiogram after the addition of ACDQ. Their chromatographic locations indicate that they are compounds with high negative charge. A large number of other metabolic stresses did not cause the appearance of these spots (Refs. 17 and 24 and data not shown).
T o see whether other changes occur following exposure to Standard two-dimensional TLC separation of cellular nucleotides. 0.9 M guanidine HCI (pH 6.5) was used for the first dimension solvent, and 74 g of (NH,)PSOI, 0.4 g of (NH,)HSO,, 4 g of disodium EDTA, 100 ml of H20 (pH 3.5) was used for the second dimension solvent. a, identities of metabolites resolved with this two-dimensional system, including 3'-phosphoadenosine 5'-phosphosulfate (PAPS), acetyl coenzyme A (AcCoA), succinyl coenzyme A (SucCoA), malonyl coenzyme A (MalCoA), and uridine 5'-diphospho-N-acetylmuramyl pentapeptide (UDP-AcMur-p,). Also shown are the chromatographic locations of ADP-Rib, adenosine 5"phosphosulfate (APS), coenzyme A-gluthathione disulfide (CoASSG), 3'-dephospho coenzyme A (-3'P-coA), orotidine 5'-monophosphate (OMP), p5'A"'p, and p5'A2 p. S6-13 represent phosphorylated compounds present in normal cells that have not been identified. The darkened spots represent compounds that do not adsorb to charcoal. Autoradiograms exposed from this two-dimensional separation of a 32P-labeled extract of S. typhimuriurn before (b) and 30 min after (c) addition of ACDQ (5 pg/ml) are shown. The arrows indicate the appearance of induced phosphorylated metabolites. 0 ACDQ, the acid extracts of S. typhimuriurn before and after addition of ACDQ were separated in a two-dimensional TLC system which gives excellent resolution of nucleotides with low negative charge (data not shown) (17). Aside from minor pool fluctuations, there were no other changes in the twodimensional profiles other than those occurring in the high negative charge region.
Better resolution of compounds in the lower left corner of the chromatogram was obtained by increasing the salt concentration in the first dimension solvent and lowering the salt concentration in the second dimension solvent. A first dimension solvent of 1.75 M morpholine, 0.1 M boric acid, 1.4 M HCl (pH 8.7) gives excellent resolution of ribo-and deoxyribonucleoside tri-and polyphosphates. A second dimension solvent of 3 M (NH4)&304, 2% disodium EDTA (pH 5.5) gives improved resolution of hydrophobic nucleotides with a minimal amount of smearing (17). A schematic diagram of this system is shown in Fig. 3a.
With this optimized two-dimensional system, we can clearly visualize the nucleotide pool changes which occur following exposure to ACDQ. Fig. 3c depicts an autoradiogram of a crude extract 50 min after addition of ACDQ using the optimized TLC system. Compared to an autoradiogram of the normal acid extract (Fig. 36), there are 11 spots which increase in intensity. Six major spots correspond to phosphorylated compounds which accumulate dramatically and are not de--2nd tectable (i.e. concentrations less than 1 p M ) in exponentially growing cells. Since they have not been detected previously in S. typhimurium, we designated these as quinone-induced spots (QS) 1-6. According to the "logic" of the two-dimensional TLC system (17), the relative migration of these compounds in the first dimension indicates that they probably contain 3-5 phosphates. Similarly, their location in the hydrophobic (or left) portion of the chromatogram suggests that they are either dinucleotides or CoA derivatives. Another spot (indicated by the arrow) appears to be a result of smearing during chromatography and was not seen in several similar chromatographic runs. In addition to these new spots, five other spots increase, but they have been seen previously. Two of the spots correspond to the "magic spots" (ppGpp and pppGpp) discovered by Cashel and Gallant (25). From previous work (20,22) these were known to accumulate in ACDQinhibited cells. Three other spots correspond to ppG>p (guanosine 5'-diphosphate-2',3'-cyclic monophosphate, two spots) and pppG>p (guanosine 5'-triphosphate-2',3'-cyclic monophosphate) (see Fig. 3a). ppG>p and pppG>p levels increase on chromatograms simultaneously with the "magic spots" and appear to be minor artifactual degradation products which are produced when ppGpp and pppGpp, respectively, break down before and during chromatography." Several known B. Bochner, unpublished data.   (17)) were observed to decrease 30 min after the addition of ACDQ (Fig. 2c).
Characterization of the Quinone-induced Spots-We have developed a battery of chemical and enzymatic tests which can be used to characterize and analyze nucleotides and nucleotide derivatives (17). Only one chromatogram is required to test the susceptibility of several compounds. In addition, the sensitivity of known compounds to a given test verifies the validity of that test. These tests have elucidated the structure of each of these novel metabolites.
The susceptibilities of the quinone-induced spots to the seven tests are summarized in Table I. It is evident that QS 1, 3, 4, and 5 have identical susceptibilities to the battery of tests. This, in addition to the fact that they all migrate in the same region of the two-dimensional TLC system, indicates that they may be structurally related. Their resistance to bacterial alkaline phosphatase (see Fig. 3d), 5'-nucleotidase, and nuclease P1 indicates that these compounds do not contain external phosphate groups. Therefore, they are likely to be dinucleotides rather than CoA derivatives. Furthermore, their resistance to ribonuclease T2 argues against internal 3'phosphate linkages. The sensitivity of these dinucleotides to oxidation by periodate and digestion by venom phosphodiesterase, however, suggests these' phosphorylated compounds are ribose-containing nucleotides with internal 5"phosphate linkages.
QS2 differs from these four compounds in being sensitive to digestion by alkaline phosphatase (see Fig. 3d) indicating that this nucleotide contains one or more external phosphates. Its susceptibility to venom phosphodiesterase and ribonuclease T2 suggests that this compound contains both 5'-and 3'-polyphosphates, at least one of which is probably cleaved by alkaline phosphatase. As with QS 1, 3, 4, and 5 , QS2 is oxidized by periodate. This compound resembles ppGpp and pppGpp in its susceptibility to the tests (17).
QS6 is apparently structurally unrelated to the other quinone-induced spots. While QS 1-5 adsorb to charcoal, confirming that they are nucleotides, QSS does not. We have found that some nucleotides have a decreased affinity for charcoal; for example, uracil nucleotides adsorb less readily than adenine or guanine nucleotide^.^ In order to determine whether the nonbinding of QS6 to charcoal was due to decreased affinity, we exposed this phosphorylated compound to 2-and 4-fold increased amounts of charcoal (1.3 mg/20 p1 and 2.6 mg/20 pl, respectively). In each of these instances, QS6 did not appreciably adsorb to charcoal. This characteristic is confounded by the susceptibility of QS6 to digestion by 5'-nucleotidase, indicating that this compound contains a nucleoside linked to a 5'-monophosphate (17). Based on the specificity of this enzyme, it is likely that QS6 is a nucleotide that for some reason is hindered in adsorbing to charcoal rather than a non-nucleotide that is hydrolyzed by the en-zyme. However, it is also possible that QS6 was hydrolyzed by a contaminating enzyme. The susceptibility of QS6 to digestion by alkaline phosphatase (Fig. 3d) further indicates that this compound contains at least one external phosphate. Its resistance to venom phosphodiesterase, ribonuclease T2, and nuclease P1 suggests that QS6 is not a nucleoside linked to 5'-or 3'-polyphosphates nor to a 3'-monophosphate. Its migration in the first dimension indicates that this novel compound has a net negative charge similar to that of a nucleotide containing four external phosphates. Because QS6 is present at a very low intracellular concentration, we have been unable to further clarify its structure. Its characteristics, as established by our battery of tests, do not correspond to any normal cellular metabolite. However, it is unlikely that this phosphorylated compound is structurally related to the family of dinucleotides (QS 1-5).
Identification of QSl as AppppA-QS1 migrates with ATP in the first dimension indicating that they possess similar net negative charge. This evidence suggests that QSl is a dinucleotide containing four internal phosphates. Based on the earlier findings that ACDQ inhibits leucyl-tRNA synthetase (21)(22)(23) and that AppppA has been found to be produced in vitro by aminoacyl-tRNA synthetases (1-S), we suspected that QSl might be AppppA. In fact, we found that QSl comigrated precisely with a standard of AppppA.
To further verify its identity, QSl was purified and digested by venom phosphodiesterase. The digestion was followed by chromatographic analysis with two one-dimensional TLC systems. The two systems resolve nucleotides based upon different principles of separation (i.e. negative charge of their phosphate groups and their content of nucleobases, respectively) (17), in order to avoid the possible problem of two compounds migrating together in one system. Autoradiograms depicting this digestion are shown in Fig. 4, a and b. QSl before digestion was run in lane 4. 5'-ATP and 5'-AMP were the products of a partial venom phosphodiesterase hydrolysis of QSl (lane 5). As the digestion was followed to completion (lane 6 ) , the ATP generated from the first cleavage was further dephosphorylated to 5'-AMP and PP, in a final 32P ratio of 1:0.85, respectively. The venom phosphodiesterase digestion products of QSl were verified by comparing their migration with that of a parallel digestion of purified 32P-labeled ATP. Upon addition of venom phosphodiesterase, ATP was cleaved to 5'-AMP and PPi (Fig. 4, a and b, lanes 2 and 3) in a final 32P ratio of 1:2.0, respectively. Our experiments show that the cleavage of AppppA by venom phosphodiesterase to give ATP and AMP is relatively fast compared to the cleavage of ATP. This phenomenon has been reported previously by Randerath et al., who proposed that AppppA may be a better substrate than ATP for venom phosphodiesterase (2).
It is unlikely that QSl has an undetected modification on either adenosine. The complete venom phosphodiesterase digestion products of QSl were resolved in a two-dimensional  ( l a n e 2), and 16 h ( l a n e 3) after addition of snake venom phosphodiesterase. Likewise, 32P-labeled QSl was run before ( l a n e 4 ) , 40 min ( l a n e 5), and 16 h ( l a n e  6) after addition of venom phosphodiesterase. The digestions were performed in 0.25 M NH4HC03 (pH 8) at 37 "C.
TLC system which optimally resolves modified mononucleotides (17). The AMP and PPi generated from the digestion comigrated with standards of 5'-AMP and PPi, respectively.
Any modification of the adenosine moiety or a different linkage of the phosphate to the nucleoside would have resulted in altered migration in this system. Thus, the structure of QSl has been established as AppppA.
Identification of QS2 as ApppGpp-ApppGpp and AppppGpp can be made in vitro from ppGpp and pppGpp, respectively (6). Therefore, we hypothesized that QS2 might be one of these based on its chromatographic location and susceptibility to ribonuclease T2. QS2 was purified and analyzed, analogous to the method used for QSl. Venom phosphodiesterase cleaves the a,@-pyrophosphate linkage of nucleoside 5'-polyphosphates, and at pH 8.4, it requires a free 3'-hydroxyl group (26): When QS2 was hydrolyzed by venom phosphodiesterase at pH 8.4, 5'-AMP and ppGpp (at a 32P ratio of 1:3.95, respectively) were the only digestion products (Fig. 5, a and b, lane 2). The identity of the AMP band was verified by comigration with a standard. In Fig. 5, a and 6, lanes 3 and 4 depict the migration of radiolabeled standards of ppGpp and pppGpp, respectively. As in the optimized twodimensional system, ppG>p is present with ppGpp. The lower band of Fig. 5a, lane 2, clearly migrates with the standard of ppGpp and not pppGpp in the solvent system which separates on the basis of negative charge. Consequently, the identity of QS2 was established as an adenylylated form pf ppGpq; howe y r , t t i s $nucleotide could be linked A5 ppps G3 pp or As ppp3 G5 pp, since both structures yield AMP and ppGpp as venom phosphodiesterase digestion products and show identical susceptibilities to the battery of tests.
T o determine the phosphate linkage of this adenylylated nucleotide, we subjected purified QS2 to a mixed enzymatic digestion with ribonuclease T2 and nuclease P1 at pH 6.5. Digestion by this enzyme mixture will distinguish the two phosphate linkages. comigration with standards) to give ApppG and Pi in an approximate molar ratio of 1 2 , respectively. Thus, the structure is A"'ppp"G:"pp. The linkage of this dinucleotide is consistent with the finding that ApppGpp is synthesized from ppGpp and aminoacyl-AMP in vitro (6).
The identity of QS4 as ApppG is supported by alkaline phosphatase digestion of the acid extract of ACDQ-treated cells (Fig. 3d), in which QS2 (ApppGpp) disappeared with a concomitant increased appearance of QS4 (ApppG). Commercial standards of ApppG and ApppA were available, and they comigrated exactly with QS 4 and 5, respectively. A preliminary purification of QS3 has indicated that AMP, ATP, GMP, and GTP are the partial venom phosphodiesterase digestion products of this dinucleotide, supporting its identity as AppppG; however, no commercial standard of AppppG was available for comparison. Alternative structures for QS 4 and 5 are possible; for example, AppppdG and AppppdA would show similar susceptibilities to the battery of tests and could be expected to run in the same region as QS 4 and 5, respectively, since they might have a decreased affinity for borate. However, it is unlikely that these dinucleotides would comigrate exactly with standards of ApppG and ApppA, respectively. Commercial standards of AppppdG and AppppdA were not available.
It is interesting to note that QS 1-5 preserve a relative order of migration in accordance with the empirical logic of the two-dimensional TLC system (Fig. 3c). For example, ATP, GTP, and ppGpp migrate along a diagonal line in this system: in a similar manner, AppppA (QSl), AppppG (QS3), and ApppGpp (QSZ) also form a line parallel to, but shifted to the left of, the ATP-ppGpp line. In addition, the relative positions of ApppA (QS5) and ApppG (QS4) to AppppA (QSl) and AppppG (QS3) are parallel to the positions of the unadenylylated forms, i.e. ADP and GDP to ATP and GTP, in like fashion. Consequently, the identities of QS 1-5 are consistent with their chromatographic locations.
We have located where standards of AppA, ApppppA, and AppppppA migrate in the optimized two-dimensional system (data not shown). There are no spots present in crude extracts of ACDQ-treated cells which comigrate with these dinucleotides. It is also known that adenylylated pyrimidine nucleoside 5'-polyphosphates can be synthesized in the in vitro back reaction of lysyl-tRNA synthetase (3,4,27). These would be expected to run near ATP and could be obscured. In Fig. 3d, we do see one alkaline phosphatase-resistant spot near ATP which could correspond to one of these. Quantitation of Quinone-induced Spots and Related Nucleotides- Fig. 6a depicts the kinetics of ATP, ppGpp, pppGpp, and other nucleotide pools in cells exposed to ACDQ (5 pg/ ml). These changes are comparable to those reported previously (20). Whereas ATP increases slightly from 3.0 mM (the basal level in normal cells (17)) to 3.6 mM, the ppGpp concentration increases dramatically shortly after addition of the quinone. Using the initial cellular concentration of ATP to normalize the concentrations of other phosphorylated compounds, we found that ppGpp increases 79-fold to a level of 2.2 mM, while pppGpp reaches a significantly smaller peak concentration of 164 p~. The other nucleotides undergo less dramatic changes.
The induction of QS 1-5 following exposure to ACDQ is shown in Fig. 66, which can be compared to the levels of the unadenylylated forms in Fig. 6a. The two major quinoneinduced spots, AppppA and ApppGpp, peak 30 and 20 min, respectively, after the addition of ACDQ. AppppA increases from undetectable levels ( 4 p~) to a peak concentration of 102 p~ while ApppGpp accumulates from undetectable levels to 86 p~, resulting in at least an 80-fold increase in the levels of these dinucleotides. The other minor quinone-induced spots increase less dramatically and attain levels of 8-20 pM.
The phosphate concentration of QS6 increases to a maximum concentration of 119 p~ 30 min after the cell is exposed to ACDQ. Since the identity of QS6 has not been determined, we were unable to convert these values to the concentrations of this compound. If we assume that this novel phosphorylated compound contains 5 phosphates (based on its chramatographic location), we can estimate an approximate induced intracellular concentration of 24 pM.

DISCUSSION
Using our recently developed method for monitoring the i n vivo metabolism of cellular nucleotides (17), we have been attempting to systematically identify new nucleotide signal molecules (24). As part of this study, we now find that the bacterium Salmonella typhimurium accumulates a family of adenylylated nucleotides at intracellular concentrations of up to 100 p~ following exposure to ACDQ but not under a variety of other metabolic stresses (Refs. 17 and 24 and data not shown). Characterization of these compounds by investigating their susceptibility to our battery of chemical and enzymatic tests and further studies on the purified compounds have established the structures of the two major nucleotides as AppppA and ApppGpp. Preliminary evidence indicates that the other compounds are AppppG, ApppG, and ApppA.
Adenylylated nucleotides can be synthesized in the back reaction of aminoacyl-tRNA synthetases i n vitro (2)(3)(4)(5)(6), and it is assumed that they are produced by the same mechanism i n uiuo (8,11). This reversible reaction has been found to nonspecifically adenylylate any nucleoside 5'-polyphosphate in vitro (2)(3)(4)(5)(6)27). AppppA, ApppGpp, AppppG, ApppG, and ApppA could, therefore, be produced from ATP, ppGpp, GTP, GDP, and ADP, respectively (see Fig. 1). AppppGpp has been synthesized i n vitro in a similar manner from pppGpp (6), and it is likely that cells do have the capability to synthesize this dinucleotide. As can be seen from Fig. 6a, however, cells exposed to ACDQ have "magic spot" pools similar to those of spoT mutants (22,28), that is, the level of ppGpp is much higher than that of pppGpp. With the concentration of pppGpp attaining a level of only 160 p~ it is possible that there is not a high enough concentration of this nucleotide from which the adenylylated form (AppppGpp) could be derived. This explanation, however, is confounded by the observation that ApppG is produced i n vivo even though the level of GDP, the unadenylylated form, never exceeds 100 p~ under these conditions. The specificity of nucleotide adenylylation in vivo requires further elucidation.
The finding that AppppA and other adenylylated nucleotides accumulate to high levels in a bacterial cell raises the question of their function. In eucaryotic cells, AppppA has been postulated to be a signal for cell proliferation (11). Our results suggest a different function in bacteria because these dinucleotides accumulate under a particular condition where the growth of cells is severely inhibited. We propose, instead, that AppppA and the related adenylylated nucleotides are alarmones (24,29,30), signal molecules, such as ppGpp (29,31,32) and CAMP (30), which alert the cell to the onset of particular metabolic stresses.
One possible hypothesis for the function of AppppA and the related adenylylated nucleotides is that they signal a shortage of tRNA. This is supported by the evidence that ACDQ inhibits the forward reaction of leucyl-tRNA synthetase (21)(22)(23). Fig. 1 depicts how these dinucleotides could be made in uiuo. The synthesis of these compounds from the aminoacyl-adenylate competes with tRNA charging; that is, if tRNA is present, the aminoacylation reaction is drawn forward resulting in a depletion of the aminoacyl-adenylate. This is supported by the finding that the enzymatic synthesis of AppppA by E. coli phenylalanyl-tRNA synthetase is decreased upon the addition of phenylalanine-specific tRNA ( 7 ) . However, if the cell is experiencing a shortage of tRNA, this would tend to increase the aminoacyl-adenylate pool and could promote the pyrophosphate exchange back reaction resulting in production of AppppA and the other adenylylated nucleotides.
Another possible hypothesis is that these adenylylated nucleotides signal oxidative damage to the cell. Quinones, such as ACDQ, are strong oxidizing agents (33) as they cause redox cycling (34). Leucyl-tRNA synthetase, an enzyme hypersensitive to oxidation (23,35,36), is the most susceptible target of ACDQ (21)(22)(23). In fact, it has been demonstrated that ACDQ not only inhibits the tRNA charging reaction but also decreases the binding affinity of leucyl-tRNA synthetase for PP, (22) as it reacts with a pair of sulfhydryl groups located near the active site of the enzyme (23). Consequently, this enzyme could have a higher affinity for nucleoside 5'-polyphosphates than for PPi if the back reaction were to occur. The accumulation of these dinucleotides in S. typhimurium and E. coli under a variety of other oxidative stresses, but not under other metabolic stresses, supports this latter possibil-it^.^