A new nucleotide involved in the stringent response in Escherichia coli. Guanosine 5'-diphosphate-3'-monophosphate.

A novel nucleotide has been detected in Escherichia coli subjected to the stringent response. However, this nucleotide does not accumulate in relA+ cells subjected to heat shock, in which guanosine 5'-diphosphate-3'-diphosphate does accumulate but stable RNA synthesis is not restricted. The intracellular level of this new nucleotide thus correlates well with control of stable RNA synthesis. Chemical and enzymatic analysis shows that the new nucleotide is guanosine 5'-diphosphate-3'-monophosphate. It is suggested that this nucleotide may play a role in stringent control of stable RNA synthesis.

A novel nucleotide has been detected in Escherichia coli subjected to the stringent response. However, this nucleotide does not accumulate in relA+ cells subjected to heat shock, in which guanosine 5'-diphosphate-3'diphosphate does accumulate but stable RNA synthesis is not restricted.
The intracellular level of this new nucleotide thus correlates well with control of stable RNA synthesis. Chemical and enzymatic analysis shows that the new nucleotide is guanosine 5'-diphosphate-3'-monophosphate.
It is suggested that this nucleotide may play a role in stringent control of stable RNA synthesis.
When the aminoacylation of any tRNA species is restricted in Escherichia co& there ensues a major readjustment of the pattern of transcriptive and metabolic activity, including a restriction of stable RNA synthesis, termed stringent control (reviewed in Refs. 1 to 3). This adjustment requires a functional relA gene product, which has been identified as an enzyme which produces the two regulatory nucleotides, pppGpp ' and ppGpp (reviewed in Ref. 4). Stringent control is unimpaired in spoT mutants, which accumulate only ppGpp (5,6). It follows that the ability to generate a high level of ppGpp is necessary for stringent control.
The nucleotide has proven to be a pleiotypic effector of a variety of metabolic steps and transcriptional processes in vitro (reviewed in Refs. 1 to 3). Nonetheless, its reported effects on rRNA and tRNA synthesis in vitro and in permeabilized cells have not, on the whole, seemed sufficient to account in full for stringent control of the formation of these RNA species in real life (reviewed in Ref. 7).
Under various growth regimes, high levels of ppGpp are generally correlated with restricted synthesis of rRNA and tRNA. But special cases have been observed where this correlation breaks down (7)(8)(9). In particular, we recently found that certain strains accumulate a high level of ppGpp upon heat shock but do not at the same time restrict rRNA and tRNA formation (7). (The interpretation of these results has recently been challenged by Challoner-Larson and Yamazaki (10). However, their objections are obviated by the fact that * This study was supported by Grant 5-It01 GM13626 (to J. G.) from the National Institutes of Health and Grants BC-170 (to J. G.) and PF-1231 (to C. P.) from the American Cancer Society, Inc. the original results depended on the comparison of the behavior of isogenic relA' and relA-cells following temperature upshift. Challoner-Larson and Yamazaki (10) reported no such comparison in their work, and we feel that such a comparison is essential.) It follows that a high level of ppGpp may be necessary, but it is not sufficient, for stringent control of stable RNA synthesis.
To explain this puzzle, we speculated that some derivative of ppGpp, formed during the stringent response but not upon heat shock, might be an essential element of the stringent control system (7). In this article, we describe the identification and characterization of a new member of the MS nucleotide family which might fill the bill.  1 shows radioautograms produced by a certain group of "'P-labeled nucleotides, resolved by a ' We use this abbreviation by analogy to MS1 and MS2, the original designations of ppGpp and pppGpp-.before these compounds were characterized.
We report the chemical characterization of MS3 below, and will call it by its proper name once we have completed the description of the structural analysis. development at right angles in the acidic solvent will not comigrate with ppGp in the two-dimensional pattern. Thus, we would expect material co-migrating with ppGp in both dimensions to contain artifactually generated ppGp amounting to a maximum of 0.5% the quantity of ppGpp. On the contrary, the labeled MS3 detected by this procedure (Fig. 2) amounts to nearly 3% the quantity of ppGpp. The bulk of this material must therefore be of authentic biological origin.
We should warn the reader that the method illustrated in Fig. 1 does risk considerable contamination of MS3 by spurious ppGp generated by hydrolysis of ppGpp during the first, acidic dimension. The quantity will vary with the length of time the fist dimension takes to run, which varies unpredictably from one batch of PEI-cellulose plates to the next. Indeed, with unusually slow running plates we have seen clear evidence of this chromatographic artifact, in the form of a trail of radioactivity running down from MS3 through the GTP spot to the level of ppGpp in the fist dimension. Thus, the reversed two-dimensional procedure illustrated in Fig new method. In Panel A, the nucleotides were those found in a relA+ strain growing exponentially; in Panel B, those found in the same strain after induction of the stringent response through serine hydroxamate inhibition; in Panel C, those found in an isogenic relA mutant also subjected to serine hvdroxamate inhibition.
It can be seen that an increased level In order to obtain sufficient MS3 for chemical characterization, we have employed the following three-step procedure. First, the nucleotide fraction was purified and concentrated by adsorption to activated charcoal, followed by desorption in 50% ethanol containing 2.8% NH,OH. This material was then submitted to column chromatography on DEAE-Sephadex in a neutral pH NaCl gradient containing urea by the method shown in Fig. 3. The small peak (Fraction 26) following marker GTP proved to consist mainly of MS3 (Fig. 3A), which was subsequently isolated by preparative thin layer chromatography (Fig. 3B). Here again, we have assessed the extent of artifactual hydrolysis of ppGpp to ppGp during work-up, in particular the brief exposure to 2.8% NH,OH during desorption from Norit. When authentic ppGpp was carried through the latter procedure, we again found less than 0.5% hydrolysis to ppGp. Thus, MS3 isolated in this manner should also consist mainly of authentic biological material.
of the regulatory nucleotides ppGpp and pppGpp is specific to the conditions of Panel B. Another compound, just above GTP in Panel B, shares this specificity as well. We call this compound MS3. The nucleotide predicted by our hypothesis should accumulate during the stringent response but not upon heat shock. Panel D, in which the cells were the reZA' strain subjected to heat shock, shows that the level of MS3 behaves Chemical Characterization of MS3-MS3 is absorbed by activated charcoal and labeled by ['?]guanosine (see below), indicating that it is a purine nucleotide. It is insensitive to periodate oxidation and fails to complex borate ion, indicating as predicted.
MS3 has not been detected before because it co-migrates with GTP in most commonly employed separation systems.
In the system of Fig. 1, MS3 is just resolved from GTP and co-migrates with its two isomers, ppGp and pGpp. At this point, a serious possibility of artifact needs to be considered. The fist dimension solvent of the separation system is strongly acidic, and the 3'-pyrophosphate group of ppGpp is somewhat acid-labile. Thus, some labeled ppGp could be generated from labeled ppGpp during chromatography in the first dimension, We have therefore taken some pains to determine how much, if any, of the material we term MS3 is of authentic biological origin.
First, we have simply reversed the order in which the two dimensions of the separation are developed. The second dimension solvent, 1.5 M KH2P04 at pH 3.4, is only weakly acidic; using authentic ppGpp, we find 0.5% or less hydrolysis to ppGp during chromatography in this solvent. Since the & of ppGpp is much lower than that of ppGp in the KH2P04 solvent, any ppGp generated from ppGpp during subsequent ployed resolves guanosine from all known purine nucleosides, including, in particular, 2'-and 3'-modified guanosine derivatives as shown. We conclude that the nucleoside residue of MS3 is unmodified guanosine. Since alkaline phosphatase hydrolysis converts MS3 to guanosine, it follows that the substituent on position 2' or 3' is one or more of the phosphate groups.
After equilibrium labeling with ['4C]guanosine and [""PI-PO+ the ratio of 14C to "'P found in MS3 was 97% that found in GTP and 71% that found in ppGpp, indicating that the compound contains three phosphate groups/guanosine residue. Accordingly, the chromatographic behavior of MS3 was compared with that of authentic ppGp and pGpp. Both compounds co-migrated with MS3 in the two-dimensional separation of Fig. 1 and in several other systems. Moreover, nitrous acid treatment of MS3 converted it to a product which comigrated in a two-dimensional separation with the nitrous acid product of ppGp, presumably ppXp (data not shown). All of the foregoing results indicate that MS3 is an isomer of GTP particularly slowly on nucleotides bearing a 3' substituent (17-19). Slow hydrolysis of MS3 and authentic ppGp by this enzyme proceeded with identical kinetics (Fig. 7).
bearing one or more phosphates on position 2' or 3'.
Location of the Phosphate Groups in MS3-The 3'-diphosphate groups of such nucleotides as pGpp and ppGpp are sensitive to alkaline hydrolysis, whereas the 5'-diphosphate residue of ppGp is not. MS3 is not sensitive to alkaline hydrolysis, indicating that it does not have a diphosphate group on position 3' or 2' (12). A complementary test is provided by the 14), which phosphorylates guanosine nucleotides bearing a diphosphate group in position 5'. Both MS3 and ppGp serve as acceptors in this reaction, whereas pGpp does not (Fig. 5). We conclude that MS3 is a guanosine 5'-diphosphate with another phosphate on either position 2' or 3'.
To determine the location of the third phosphate group, we hydrolyzed MS3 with snake venom phosphodiesterase and submitted the diphosphate product to two tests. First, the product co-migrated with authentic pGp(3') in a separation system which resolves this compound from pGp(2') ( Fig. 6, Panel A). Second, the diphosphate product was hydrolyzed virtually completely by ryegrass 3'nucleotidase (15, 16) (Fig.  6, Panel B), which was shown in parallel incubations to be wholly inactive against pGp (2'). These tests indicate that the third phosphate group in MS3 is esterified to position 3', and complete the structural proof that MS3 is identical to ppGp.
We have performed one further enzymatic test of this identity. Snake venom phosphodiesterase cleaves the 5' o-/3 bond of different nucleotides at characteristic rates and works FIG. 6. Enzyme digestion of MS3. '"P-labeled MS3 was purified from preparative thin layer chromatograms and subjected to digestion by snake venom phosphodiesterase. The incubation conditions were as follows: lo-p1 reaction mixtures contained Tris-HCl buffer at pH 8.9 (0.8 ymol), MgC12 (0.2 pmol), and snake venom phosphodiesterase (Sigma) at 0.2 unit. The 6-h incubation was followed by two-dimensional thin layer chromatography on a cellulose sheet together with marker nucleotides. The solvent used is isobutyric acid:concentrate NHrOH:water (66:1.7:33) for the first dimension and saturated ammonium sulfate:8.2% sodium acetate:isopropyl alcohol (80:18:2) for the second dimension (Panel A). The radioactive compound which co-migrated with marker pGp(3') was then purified and incubated with rye grass 3'-nucleotidase, 0.7 unit/ml, in 20 mM Tris-HCl buffer (pH 7.4) at 37'C for 7 h. The reaction mixture was then separated on thin layer of PEI-cellulose with stepwise sodium formate, pH 3.4, at 0.5, 2.0, and 4.0 M as solvent (Panel B). consistent with the expected precursor-product relationship between GTP (or GDP) and ppGpp. A similar lag intervenes between the labeling of ppGpp and ppGp. This lag, by the same logic, is consistent with the production of ppGp from ppGpp by hydrolysis of the 3'-P-phosphate.
A search for such an activity in vitro is in progress.

DISCUSSION
Our results show that MS3 is ppGp, a new member of the family of nucleotides involved in the stringent response. We have discussed earlier control experiments which demonstrate that artifactual hydrolysis of the 3'-pyrophosphate group of ppGpp during workup is insufficient to account for more than a minor fraction of the material we have isolated and gone on to characterize.
We should emphasize that our data contain two further sources of evidence that ppGp is of biological origin. One is the fact that the bulk of the material we isolate is phosphorylated in position 3' and not position 2' (Fig. 6). Acid or alkaline hydrolysis of ppGpp would generate a racemic mixture of 2' and 3' isomers, just as in the case of RNA hydrolysis. Second, ['%]guanosine enters ppGpp and ppGp with markedly different kinetics (Fig. 8), which constitutes further proof that the latter compound is made in the cells and not during extraction and resolution of the nucleotides.
We were led to look for a compound such as ppGp because of cases where a high intracellular level of ppGpp did not correlate with restricted stable RNA synthesis. We will report elsewhere on experiments which directly suggest that ppGp plays a role in regulating the formation of rRNA and tRNA." The existence of ppGp underlines the proposition that regulatory effects of ppGpp alone may not provide a complete account of the various facets of stringent control, or of the ot,her control systems (e.g. growth rate control) in which compounds of this group seem to participate.
It seems to us likely, indeed almost a foregone conclusion, that each of the MS nucleotides has specific regulatory functions of its own. This report brings the number of these nucleotides identified in E. coli to three. It is by no means impossible that other such compounds remain to be discovered in E. coli, especially in view of the existence of related nucleotides in Actinomycetes (22,23) and Bacillus subtilis (24).
This increasingly numerous tribe of 3'-phosphorylated nucleotides, and their putative regulatory functions, presents itself as an intracellular, prokaryotic counterpart to the families of chemically related hormones of higher organisms, each with particular although sometimes overlapping functions in a network of specificities. (We are indebted for this analogy to the late Gordon Tomkins (25).) We believe that the burgeoning literature on control effects of the most abundant of these compounds, ppGpp, will prove to be only the first installment of a much more elaborate story. Achnowledyment-We are grateful to Linda Palmer for her excellent technical assistance in some experiments.
Note Added in Proof- Lagosky and Chang (26) have recently reported in detail on the production of ppGp by acid hydrolysis of ppGpp. They have also developed a lysozyme-deoxycholate method for extracting nucleotides from cells without this artifact; using this method, we are happy to note, they find a biological level of ppGp virtually identical to the one we have observed.