Cysteine starvation, isoleucyl-tRNAIle, and the regulation of the ilvGEDA operon of Escherichia coli.

The involvement of undermodified tRNA in the regulation of the ilvGEDA operon has been investigated using Escherichia coli C6, a relA-, Cys-, Met- mutant. This strain accumulates thionucleotide-deficient or methyl-deficient tRNA when starved for cysteine or methionine, respectively. The levels of threonine deaminase, the ilvA gene product, and transaminase B, the ilvE gene product, were both lower in cysteine-starved cells, as compared with either growing or methionine-starved cultures. When cysteine was added to cysteine-starved cells, growth ensued promptly and both enzyme activities returned to control levels. Treatment of recovering cultures with valine limited growth by isoleucine limitation, but did not cause a derepression of the ilvGEDA operon. Valine treatment of nonstarved or methionine-starved cells led to the expected increase in threonine deaminase and transaminase B activities. Cysteine-starved cells slowly regained the ability to derepress the operon after 3 h of recovery in complete medium. In contrast, the induction of the lac operon was normal in cysteine-starved cultures, even in the presence of valine. The loss of derepressibility of the ilvGEDA operon was correlated with the presence of a kinetically and chromatographically altered tRNAIle in cysteine-starved cells. No changes in tRNAIle were observed after methionine starvation. Using the periodate method, we found that the charging of tRNAIle increased from the normal level of 60 to 80% or greater after starvation for cysteine. Under conditions where the ilvGEDA operon was fully derepressed in nonstarved cells, the charging of tRNAIle fell to 27%. Unexpectedly, nearly identical results were obtained with cysteine-starved cells after an identical derepression test. These results suggest that factors other than the aminoacylation state of tRNAIle may be important in the regulation of this operon. In particular, modifications to tRNA which involve cysteine may be necessary for controlling the expression of the ilvGEDA operon in E. coli.

The involvement of undermodified tRNA in the regulation of the ilvGEDA operon has been investigated using Escherichia coli C6, a relA-, Cys-, Met-mutant. This strain accumulates thionucleotide-deficient or methyl-deficient tRNA when starved for cysteine or methionine, respectively. The levels of threonine deaminase, the ilvA gene product, and transaminase B, the ilvE gene product, were both lower in cysteinestarved cells, as compared with either growing or methionine-starved cultures. When cysteine was added to cysteine-starved cells, growth ensued promptly and both enzyme activities returned to control levels. Treatment of recovering cultures with valine limited growth by isoleucine limitation, but did not cause a derepression of the ilvGEDA operon. Valine treatment of nonstarved or methionine-starved cells led to the expected increase in threonine deaminase and transaminase B activities. Cysteine-starved cells slowly regained the ability to derepress the operon after 3 h of recovery in complete medium. In contrast, the induction of the lac operon was normal in cysteine-starved cultures, even in the presence of valine. The loss of derepressibility of the ilvGEDA operon was correlated with the presence of a kinetically and chromatographically altered tRNA"" in cysteine-starved cells. No changes in tRNA"' were observed after methionine starvation. Using the periodate method, we found that the charging of tRNA'Ie increased from the normal level of 60 to 80% or greater after starvation for cysteine. Under conditions where the ilvGEDA operon was fully derepressed in nonstarved cells, the charging of tRNA'" fell to 27%. Unexpectedly, nearly identical results were obtained with cysteine-starved cells after an identical derepression test. These results suggest that factors other than the aminoacylation state of tRNA1Ie may be important in the regulation of this operon. In particular, modifications to tRNA which involve cysteine may be necessary for controlling the expression of the ilvGEDA operon in E. coli.
Transfer RNA has been implicated in the regulation of several amino acid biosynthetic enzyme operons in Escherichia coli (1, 2). The ilvGEDA operon of E. coli K-12 is multivalently controlled by the levels of aminoacylated tRNA"', tRNAhu, and tRNAVa1 (3,4), presumably through an attenuation mechanism (5, 6). The regulatory region of this operon has been sequenced, and found to code for a leader RNA, within which is a message for, a peptide rich in isoleu-cine, leucine, and valine residues (5, 6). According to a recent model, when the level of any or all of the aminoacylated regulatory tRNAs falls, translation of the leader RNA in the peptide region will be impeded and RNA polymerase will then transcribe the DNA containing the structural genes of the operon (7). When abundant levels of aminoacyl-tRNA are present, the leader peptide is translated, signalling RNA polymerase to terminate transcription at the attenuator site prior to the structural genes. Hence, any factor which alters the state of aminoacylation or function of the regulatory tRNAs may have an effect on the expression of the operon.
We have been investigating the role of thionucleotides in tRNA using E. coli C6, a re& mutant requiring both cysteine and methionine (8). Under conditions of cysteine starvation, this mutant produces tRNA deficient in thionucleotides (8-10). We observed a change in the aminoacylation behavior of tRNA"' from cysteine-starved cells (11), as compared to that found in cells grown on fully supplemented medium. Sulfurdeficient tRNA"' had a 4-fold increased rate of aminoacylation and an altered elution position on MAK' columns, relative to normal tRNA'". We wanted to see if the presence of this undermodified tRNA"' in cysteine-starved cells would alter the regulation of the iloGEDA operon. To accomplish this, threonine deaminase (ilvA gene product) and transaminase B (ilvE gene product) levels were measured in cysteinestarved and normal E. coli C6 cells. Our results show that cysteine-starved cells are unable to derepress the ilvGEDA operon, the effect being reversible by addition of the sulfur amino acid. Furthermore, we found a correlation between the regulation of this operon and the presence of the undermodified, kinetically altered tRNA'Ie present in cysteine-starved cells.

Bacterial Strains and Cell
Growth-E. coli HfrC, relA Met-Cys-, X(C6 mutant) was grown in M9 medium (12) in the presence of 0.16 mM cysteine and 0.13 mM methionine, as previously described (8, 11). Cells were maintained at 37 "C in either a gyratory shaking water bath or New Brunswick Microferm fermentor, growth being monitored at 650 nm. Cysteine starvation was carried out in 10-to 14-liter fermentor cultures in M9 medium supplemented with 0.016 mM cysteine and 0.13 mM methionine (11). Methionine starvation was carried out in a similar fashion but with 0.16 mM cysteine and 0.013 mM methionine.
Enzymatic Assays-Cells were harvested by centrifugation, washed Once in cold M9 salts, and stored at -90 "C until needed. Cell extracts were prepared by sonication as previously described (13 Transfer RNA and Aminnacylation-Transfer RNA was prepared as described previously (8) using isopropanol fractionation as the final step in purification. An aminoacyl-tRNA synthetase mixture was prepared from E. coli B cells (Grain Processing Corp., Muscatine, IA) by a previously reported procedure (17). The aminoacylation assay (8, 11) contained the following components in a total volume of 0.5 ml: 20 mM Tris-HC1, pH 7.3, 10.0 mM magnesium acetate, 2.0 mM ATP, 0.5 to 1.0 pCi of L-[14C]isoleucine or 2.5 to 5.0 pCi of L- [3H]isoleucine (352 to 360 mCi/mmol), 1.0 A~M ) unit of tRNA, and an appropriate amount of synthetase mixture. All assays were carried out at 37 "C, 5O-pl samples being removed at various times, applied to 2.3-cm Whatman paper discs, and the discs prepared for scintillation counting as previously described (8).
Aminoacyl-tRNA was isolated from reaction mixtures by extraction with an equal volume of redistilled phenol saturated with 0.05 M sodium acetate buffer, pH 5.0. The resulting emulsion was centrifuged at 15,000 X g for 20 min, and the aqueous layer removed. The phenol layer was extracted with 1 h volume of 0.05 M sodium acetate buffer, pH 5.0, and the second aqueous layer obtained as above. Aminoacyl-tRNA was precipitated from the combined aqueous layers by adding 0.1 volume of 20% potassium acetate, pH 5.0, and 2 volumes of 95% ethanol, and was stored at -20 'C.
Determination of in Vivo Aminoacylation hue/--The fraction of tRNA"' which is aminoacylated in vivo was determined by the method of Folk and Berg (18) with the following modifications. Cultures were rapidly treated with 5% trichloroacetic acid and sodium dodecyl sulfate at 0.1% prior to collection of the cell pellet. The pellet was extracted with phenol at pH 5 by sonication as described (18), but we found it necessary to include a 30-min shaking step to ensure quantitative extraction of RNA. All other steps were performed as described (18), the final dialysate being passed through a 0.2 mM Gelman Acrodisc filter. This step was needed to remove undissolved material which interfered with the subsequent assays. The per cent charging of tRNA"" was measured with the standard aminoacylation system, using 0.5-1.0 A260 unit of periodate-treated or untreated RNAs (see legend to Table V). Per cent charging is defined as the incorporation of labeled isoleucine into periodate-treated RNA X 100 divided by the incorporation of isoleucine into untreated RNA.
RPC-5 Chromatography of tRNA-[3H]-or ['4CC]isoleucyl-tRNA was obtained as above and dissolved in 0.5 ml of 0.4 M NaCl, 0.01 M sodium acetate, pH 4.5, 0.01 hf MgCl,, and 0.002 M 2-mercaptoethanol. The sample was applied to an RPC-5 column (0.6 X 21 cm) (19) and washed with 100 ml of the same buffer at a flow rate of 1.5 ml/ min at 150 p.s.i. (using a Milton Roy Model 396 minipump). tRNA was eluted using a linear gradient of 0.45 to 0.6 M NaCl in 0.01 M sodium acetate, pH 4.5, 0.01 M MgC12, and 0.002 M 2-mercaptoethanol. Fractions of 1.0 ml were collected, 3 ml of ACS scintillation fluid (Amersham) was added, and radioactivity was measured by the scintillation method.

RESULTS
Amino Acid Starvation and ilvGEDA Regulation-The importance of base modification to the regulatory functions of tRNA was first appreciated due to studies of the hisT mutant of Salmonella typhimurium (20). This mutant, unable to convert certain uridine residues to pseudouridine in tRNA, showed a derepression of both the histidine and ilvGEDA operons (21). We wanted to see if the regulation of the ilvGEDA operon was altered in the presence of other undermodified forms of tRNA. To accomplish this, E. coli strain C6 was cultured in the presence of reduced levels of cysteine or methionine, conditions which support less growth than when both amino acids are present in optimal amounts (Fig.  1). During the starvation period which follows, the cells accumulate either sulfur-deficient (8) or methyl-deficient tRNA (22) as a consequence of the relA mutation. To determine if the accumulation of undermodified tRNA altered the regulation of the ilvGEDA operon (5, 6), we measured the activity of threonine deaminase and transaminase B in fully supplemented (control), cysteine-starved, and methionine-starved E. coli C6 cells. In each case, the starvations were carried out for 6 h after growth had ceased due to depletion of the amino acid. The data of Table I show that cysteine starvation resulted in a lower specific activity of both threonine deaminase and transaminase B. This drop in the activity of both enzymes was consistently observed with cysteine-starved cells (see Table I11 also), but was not seen with E. coli strain C6 starved for methionine. It is clear that deprivation of cysteine or methionine did not lead to a derepression of the ilvGEDA operon.
The previous experiments showed that cysteine-or methionine-starved cells retained their ability to repress the ilu-GEDA operon. However, a clear interpretation of these results is difficult, since deprivation of amino acids would also limit protein synthesis with the starved cultures (see Table IV, below). This prompted us to study the derepression of the ilvGEDA operon during recovery from cysteine or methionine starvation. The derepression condition involved treatment with valine, an inhibitor of acetohydroxyacid synthetase (23). coli C6. E. coli was grown in 10-liter fermentor cultures in medium containing 0.16 mM cysteine and 0.13 mM methionine (control), 0.016 mM cysteine and 0.13 mM methionine (cysteine-starved), or 0.16 mM cysteine and 0.013 mM methionine (methionine-starved). Each culture was inoculated in the evening using 5 ml of a freshly grown culture of strain C6 to yield a calculated initial A m of 0.001. The first readings were taken about 10 h later, as shown. At the end of the growth study, cells were harvested by centrifugation, washed with M9 salts, and frozen at -80 "C.

TABLE I
Effect of amino acid staroation on isoleucine and valine biosynthetic enzymes Con.tro1 cultures were grown in complete medium and cells were harvested at As50 = 0.7. Starved cultures were incubated for 6 h after growth had ceased due to depletion of either cysteine or methionine (see Fig. 1). Cell extracts were prepared as described under "Materials and Methods." Enzyme assays were carried out in duplicate; the data are averages of two separate experiments. This treatment inhibits growth by causing an isoleucine limitation. Fig. 2 shows that the growth of E. coli C6 in complete medium was inhibited by the addition of 25 mM valine, and this inhibition was overcome by adding 12.5 mM isoleucine and 12.5 mM leucine along with valine (top). The high concentration of these amino acids was necessary due to an apparent defect in the uptake of amino acids by this mutant (13). The lower two panels of Fig. 2 show the results of experiments where starved cells were allowed to recover after the addition of either 0.16 mM cysteine or 0.13 mM methionine. Growth begins shortly after this addition in both cases and is limited by the presence of 25 mM valine. Cysteinestarved cells show a more pronounced inhibition of growth by valine, as discussed below. The effect of valine is overcome by also adding isoleucine and leucine as shown. At the conclusion of the growth experiments, the cells were harvested and washed, and extracts were assayed for threonine deaminase and transaminase B activities. Table I1 shows that treatment of control cultures with valine resulted in increased levels of both enzymes, indicating derepression of the operon. The effect is not coordinate as threonine deaminase levels were increased nearly 7-fold while transaminase B was elevated by 2-fold. Derepression was not observed with extracts from cells to which isoleucine and leucine were added along with valine. Cysteine-starved cells grew when cysteine was added (Fig. 2) and the levels of threonine deaminase and transaminase B returned to control levels (compare with  Table I). However, although valine had a clear effect on growth, no derepression of the operon was observed. This is in contrast to the results with methionine-starved cells during recovery, in which valine caused derepression, and this was prevented by isoleucine and leucine. These results suggest that cysteine-starved cells are incapable of derepression of the ilvGEDA operon. The effect appears to be specific for cysteine as methionine-starved cells were similar to normal

I1
Regulation of the ilvCEDA operon ufter amino acid starvation E. coli C6 cells were grown as described in the legend to Fig. 2, and suspended in complete medium with or without the addition of valine or valine, isoleucine, and leucine, as noted. Frozen cells were thawed and cell extracts prepared by sonication as described above. Tbreonine deaminase and transaminase B assays were carried out promptly as described under "Materials and Methods." Except where noted, all data are averages from two separate experiments. Within each experiment, each assay is carried out in duplicate. The data are averages of duplicate enzyme assays from a single experiment.

TABLE 111
Derepression of the ilvGEDA operon during cysteine starvation and recovery Cysteine-starved and control cells of E. coli C6 were grown in a 12liter fermentor culture as described in Fig. 1. At the indicated times before and during starvation, cells were removed and tested for derepression in complete medium as described in Table 11. At the same time, cells were removed for tRNA isolation. After 6 h of starvation, cysteine was added at 0.16 mM and incubation wae continued. Cells were taken at 1,2, and 3 h after the addition of cysteine and tested for derepression and also saved for tRNA isolation. For the derepression assays, cells were incubated at 37 "C in complete medium or in complete medium containing 25 mM valine. After 3 h, the cells were harvested and washed and the enzyme assays were carried out as above. cells in their ability to derepress the operon.
We determined the time course of the loss of derepressibility of the ilvGEDA operon during cysteine starvation. Table  I11 gives the results of an experiment where cells were removed at various times during cysteine starvation and tested for derepression. The derepression test involves a 3-h incubation in complete medium, with or without 25 mM valine.  Here, the effect of valine on growth was greater, increasing the doubling time to 280 min. No further changes in specific activities or derepression occurred as starvation continued to 6 h. Although the cells became more sensitive to growth inhibition by valine as starvation proceeded, no derepression of the ilvGEDA operon was observed.
The above experiments showed that the inability to derepress the operon occurred as early as 3 h after the onset of cysteine starvation. To see if this effect was reversible, cysteine at 0.16 mM was added to a culture after 6 h of starvation. A portion of this culture was removed at 1, 2, and 3 h of growth recovery and tested for the ability to derepress the ilvGEDA operon as above. With the control cultures, removed from the main culture and suspended in complete medium for an additional 3 h, the activities of threonine deaminase and transaminase B gradually increased from the values characteristic of cysteine-starved cells to those of cells grown in fully supplemented medium. Cells removed at 1, 2, and 3 h of recovery all showed some inhibition of growth when resuspended in complete medium with 25 mM valine, but the magnitude of this effect decreased markedly during recovery. However, after 1 or 2 h of recovery, valine elicited only a 2fold increase in threonine deaminase activity with little change in the level of transaminase B. By the third hour of recovery, the ability to derepress the ilvA gene is nearly normal, while transaminase B was only slightly elevated in response to valine. The latter effect may be the result of a direct effect of cysteine on either the enzyme activity or expression of the ilvE gene (24). In summary, cysteine-starved cells recover their ability to derepress the ilvGEDA operon, but only after a considerable incubation period in the presence of cysteine.
The failure to derepress the ilvGEDA operon could be the result of restricted protein synthesis in cells containing tRNAs accumulated during cysteine starvation. To test this

TABLE IV
Protein synthesis during recovery from cysteine starvation Exponential and cysteine-starved cultures were grown in complete or low cysteine medium as in Fig. 1. After 6 h of starvation, 0.16 mM cysteine was added and incubation was continued to allow growth recovery. One ml of cells were removed from exponential or recovering cultures and protein synthesis was determined after addition of 1 pCi of a mixture of 15 [''C]aminoacids (New England Nuclear). The initial rate of protein synthesis was measured by removing 5O-pl samples at 1-min intervals up to 5 min from each assay tube, application to 2.4-cm paper discs, and processing as described (25). Amino acid incorporation was linear for 5 min under each experimental condition. The data are corrected for differences in cell density as shown and are a composite from several different experiments. possibility, we measured the rate of protein synthesis (25) at various stages of recovery from cysteine starvation, the results being given in Table IV. The data show a dramatic drop in the overall rate of protein synthesis for cysteine-starved cells resuspended in complete medium. During recovery, the rate of protein synthesis increases gradually, but did not reach the exponential rate. The addition of 25 mM valine at various times during recovery dramatically lowered the rate of protein synthesis, but in no instance was total inhibition observed. These results show that cysteine-starved cells promptly resume protein synthesis when cysteine is added, although at lower than optimal rates.
To determine if cysteine starvation alters gene expression, we determined the induction of the lac operon in response to the gratuitous inducer, isopropyl-B-D-thiogalactoside (16). Fig. 3 shows that cysteine-starved cells of E. coli C6 can express the lac operon, as nearly a 100-fold increase in 8galactosidase activity was observed during 1 h of incubation in complete medium. The presence of 25 mM valine slowed the synthesis of this enzyme slightly, but a full induction was observed within 2 h of growth recovery. Cyclic AMP was required for the expression of this operon when valine was present, presumably to overcome catabolite repression. However, the effect of valine on the lac operon is complex, as exponential cells can induce the operon without the addition of CAMP. Finally, the presence of CAMP had no effect on the expression of the ilvGEDA operon during the first 3 h of recovery from cysteine starvation (data not shown). Hence, expression of the lac operon was observed under conditions where derepression of the ilvGEDA operon could not be demonstrated.
Properties of tRNA during Starvation and Recovery-Since aminoacyl-tRNA is involved in the regulation of the ilvGEDA operon, we determined the aminoacylation behavior of the relevant tRNAs in exponential and amino acid starved cultures of E. coli C6. Table V shows that cysteine-starved cells contain a tRNArIe with a faster initial rate of aminoacylation than seen with either growing or methionine-starved cells. With tRNAL"", both the rate and extent of acylation were lower in cysteine-starved cells, and a similar drop in the extent of acylation was observed with methionine-starved cells as well. RPC-5 chromatography of tRNAL"" from cysteinestarved cells (data not shown) revealed isoacceptor profiles similar to those observed for amino acid starved E. coli (26).
These effects were not further investigated since the alterations in tRNAL"" were not specific for cysteine-starved cells. Finally, little change in the rate or extent of tRNAVa1 acylation TABLE V Aminoacylation properties of tRNA from exponential and amino acid starved cultures Transfer RNA was isolated as described under "Materials and Methods" from cultures of E. coli C6 grown in complete medium or with limited amounts of either cysteine or methionine. Starved cultures were maintained at 37 "C for 6 h after growth had ceased due to depletion of the limited amino acid. The initial rates of aminoacylation were determined using the standard aminoacylation assay, utilizing 4 pg of crude synthetase protein. Extents of aminoacylation were determined with the same system but with 0.8 mg of synthetase. The data are averages of duplicate determinations carried out with two separate tRNA preparations, except for the initial velocities of leucyl-and valyl-tRNA formation. In these cases, a single kinetic experiment was carried out. Transfer RNA was isolated and aminoacylated as described under "Materials and Methods." Normal tRNA was acylated in the presence of ['4C]isoleucine while sulfur-deficient tRNA was aminoacylated using [3H]isoleucine. The labeled tRNA5 were isolated from reaction mixtures and co-chromatographed as described above. The position of the major tRNA'" for methyl-deficient tRNA was determined by co-chromatography with normal tRNA (data not shown). The symbols are: 0, normal tRNA; 0, sulfur-deficient tRNA.
was observed with tRNA from cysteine-starved cells. Hence, there is a correlation between the inability to derepress the iLvGEDA operon and a change in the aminoacylation properties of tRNA"' of cysteine-starved cells.
We then sought to determine whether the change in acylation behavior was specific to one isoaccepting tRNA"'. Normal, sulfur-deficient, and methyl-deficient isoleucyl-tRNAs were prepared in vitro and chromatographed on RPC-5 columns (19). Fig. 4 shows that the elution pattern for normal and methyl-deficient tRNAs were similar, while sulfur-deficient tRNA"" eluted at a higher salt concentration. The increased rate of aminoacylation observed for the latter tRNA is associated with an alteration in its chromatographic behavior, and these changes are both specific to tRNA from cysteine-starved cultures.
The kinetic and chromatographic properties of tRNA"' in cysteine-starved cells were investigated in more detail. We isolated tRNA at various times during starvation and recovery, and determined rates of aminoacylation and RPC-5 elution profiles (as in Table V and Fig. 4). The initial velocities of isoleucyl-tRNA formation for tRNA from prestarved, 3-h  starved and 6-h starved E. coli C6 cells were 10.5, 38.0, and 40.4 pmol of Ile/Azso/min, respectively. Corresponding values for tRNA isolated after 1,2, and 3 h of recovery from cysteine starvation were 33.5, 19.0, and 11.5 pmol of Ile/A2m/min. These results show a fairly rapid increase in the rate of isoleucyl-tRNA formation during starvation, which is nearly complete 3 h after growth ceases. After cysteine addition to starved cells, there is a gradual lowering of the rate of isoleucylation to normal values. Fig. 5A shows the isoacceptor pattern on RPC-5 for tRNA"" from pre-starved cells, demonstrating its normal behavior (also see Fig. 4). By 3 h of starvation, the amount of the major tRNA"" is much lower and is replaced by a new tRNA1Ie which elutes at higher salt concentrations. At 6 h, the normal tRNA"" is nearly absent and the altered tRNA1Ie predominates. This altered tRNA is responsible for the increased rate of aminoacylation observed for tRNA isolated during cysteine starvation (11). One hour after cysteine addition, there is a partial loss of the altered tRNA and an increase in normal tRNA1Ie. As recovery continues, the isoacceptor pattern returns to normal, as does the rate of aminoacylation. Hence, there is a correlation between by guest on March 24, 2020 http://www.jbc.org/ Downloaded from the presence of undermodified tRNA"' in cysteine-starved cells, the kinetic properties of this tRNA, and the inability to derepress the ilvGEDA operon.
It is possible that the presence of the undermodified tRNA"' in cysteine-starved cells, with its faster rate of aminoacylation, would increase the cellular level of isoleucyl-tRNA. This could explain the lower specific activities of threonine deaminase and transaminase B in cysteine-starved cells, and the failure to derepress the operon in response to valine. TO test this possibility, we first measured the per cent charging of tRNA"' in vivo, using the periodate method following pH 5 phenol extraction of the RNA (18). The data of Table VI show that, in E. coli strain C6 grown in complete medium, the charging of tRNA"" was 62%. When cells are treated with 25 mM valine, one sees a dramatic drop in the charging of tRNA'" to 27%. This is consistent with the inhibition of isoleucine synthesis by valine, and represents the amount of isoleucyl-tRNA present when the ilvGEDA operon is fully derepressed. Cysteine-starved cultures of E. coli strain C6 show a 20% increase in isoleucyl-tRNA levels, as compared to that of fully supplemented cells. In some experiments, the charging of tRNA in cysteine-starved cells approached 100% (see Fig. 6). Hence, the accumulation of the kinetically altered tRNA"" during cysteine starvation does lead to an increase in the cellular level of isoleucyl-tRNA*'". The effect of valine on cysteine-starved cells during growth recovery was investigated. The per cent charging of tRNA"" remains high even 3 h after the addition of cysteine to starved cells. However, treatment of starved cells with valine during recovery results in a decrease in the isoleucyl-tRNA level to 32%, a value similar to that observed with control cultures treated with valine. These results are quite unexpected and indicate that, with cysteine-starved cells, valine treatment causes a drop in the cellular level of charged tRNA"", but this does not lead to a derepression of the ilvGEDA operon.
The derepression anaIysis used above measures enzyme levels after a 3-h treatment with valine. In contrast, the periodate method determines the % charging of tRNA at the time when the culture is harvested, i.e., is an instantaneous value. For this reason we determined the per cent charging of tRNA"' in control and cysteine-starved cells at various times after suspension in complete medium with 25 mM valine. Fig.  6 shows that before addition of valine the per cent charging of tRNA"" was higher in cysteine-starved cells than in control

TABLE VI
In vivo aminoacyletion levels of tRNA" Fully supplemented cultures were grown in complete medium with or without 25 mM valine as described in the legend to Fig. 2. Cysteineand methionine-starved cultures were grown as in Fig. 1 and incubated for 6 h after depletion of the amino acid. The recovery experiments utilized cysteine-starved cells to which cysteine at 0.16 mM ( 5 valine) was added after 6 h of starvation. Incubation was continued for 3-4 h after addition of the sulfur amino acid. In all cases, RNA was prepared and per cent charging of tRNA1'" determined by the periodate method (19). Aminoacylation assays were performed as described under "Materials and Methods," except that a mixture of 15 unlabeled amino acids (no isoleucine) was included to prevent misacylation (24). The data are given as the mean 2 S.E., the number of separate RNA preparations being given in parenthesis. Significantly lower than control (no valine) at p < 0.01.
Significantly higher than control (no valine) at p = 0.07. cells. Thirty minutes after the addition of valine, the aminoacylation ratio fell to below 10% for both cultures. These ratios remained low for the next hour and then gradually rose in both cases as the growth effects of valine were overcome (data not shown). No significant differences in the isoleucyl-tRNA levels of control or starved cells were observed throughout the period of valine treatment. Therefore, the reason why cysteine starvation Ieads to an inability to derepress the ilvGEDA operon must be some other alteration in the regulatory tRNA(s) other than its aminoacylation state.

DISCUSSION
Cysteine starvation of E. coli strain C6 relA, Met-, Cyswas accompanied by a dramatic loss of the ability to derepress the ilvCEDA operon. This effect was reversed slowly as cysteine-starved cells recovered in complete medium. The loss of derepressibility was specific to cysteine starvation, as both growing and methionine-starved cells of the same strain were able to derepress the operon in response to an isoleucine limitation. Moreover, cysteine-starved cells were capable of a full induction of the lac operon under the same conditions where the iEuGEDA operon could not be expressed. This effect was correlated with the appearance of an altered tRNA"' during cysteine starvation, which had a faster rate of aminoacylation and an altered elution position on RPG5 chromatography. No changes in the activity or RPC-5 pattern were observed for tRNA"" from methionine-starved cells of E. coli C6. During recovery from cysteine starvation, the altered tRNA"" was gradually converted back to the normal isoacceptor, both in terms of aminoacylation and chromatographic behavior. Therefore, the loss of derepressibility of the ilu-GEDA operon was correlated with an alteration of one of the tRNAs known to be involved in the regulation of this operon (3,4).
The most likely explanation for the above findings would be an increase in the level of isoleucyl-tRNA in cysteinestarved cells of E. coli strain C6. Direct measurement of the per cent charging of tRNA"' confirmed this hypothesis, as a 20% increase in isoleucyl-tRNA was observed after cysteine starvation. This may explain why threonine deaminase and transaminase B levels are somewhat lower after 6 h of cysteine starvation. Valine treatment was used to limit isoleucine synthesis, which in turn generates a derepression signal, This treatment reduced the per cent charging of tRNA'" from 62 to 27% in growing cells and from 80 to 32% in cysteinestarved cells during recovery. These values were determined 3 h after the addition of valine and measure the instantaneous per cent of aminoacylated tRNA"' at the end of the derepression period. We also found no differences in the per cent charging of tRNA'" under these two conditions throughout the 3-h treatment with valine (Fig. 6). Taken together, these results show that valine causes a drop in isoleucyl-tRNA levels, presumably the consequence of its inhibition of isoleucine synthesis at the level of acetohydroxy acid synthetase (23). On the other hand, the data do not allow us to conclude why the ilvGEDA operon is not derepressed by this addition of valine after cysteine starvation of E. coli strain C6.
In terms of the accepted model for the regulation of this operon (5, 6 ) , failure to derepress could arise in two ways. First, the operon could be blocked at the level of transcription initiation, although little is known about the regulatory elements acting at this site. It is unlikely that ppGpp levels are too low in cysteine-starved cells to allow transcription of the operon since regulation is normal in methionine-starved cultures. Further, it was recently shown that, as opposed to relA+ strains, ppGpp was low in amino acid starved cultures of relA mutants of E. coli but rose rapidly when the missing amino acid was added (27). The derepression test used here included cysteine in the medium along with valine. The second regulation of the ilvGEDA operon occurs at the several attenuators prior to structural genes (28, 29). The drop in the isoleucyl-tRNA level observed with cysteine-starved cells in the derepression test appears to be sufficient to cause a decrease in attenuation, but does not. The other major regulatory tRNAs are not involved, as we have shown that valine specifically lowers the per cent charging of tRNA'" with no effect on the per cent charging of tRNAL"" or tRNAVe' (24). Perhaps the tRNA"' present in starved cells is more efficient in translation such that the leader peptide is translated even with less than 10% of the tRNA"' in aminoacylated form. Alternatively, a thionucleotide-containing tRNA, which fails to translate some codon earlier in the leader, may cause a block in translation leading to transcription termination at the first attenuator (5, 6). Finally, nutritional deprivation of r e" strains of E. coli leads to mistranslation of mRNA (30)(31)(32). Hence, unmodified tRNA may mistranslate the leader preceding the attenuator site(s) of the ilvGEDA operon and signal RNA polymerase to terminate transcription. Further investigation will be necessary to resolve these issues.
Several lines of evidence suggest that the modification and aminoacylation of tRNA'" are important for the regulation of the ilvGEDA operon. The hisT mutants of S. typhimurium (21) and E. coli (33), defective in pseudouridine synthesis, were found to have derepressed levels of the isoleucine-valine biosynthetic enzymes, Recent findings suggest that tRNA"' in the hisT mutant may be defective in regulating the operon at the site of transcription initiation (34). The trp operon of E.
coli was derepressed in a mutant defective in the synthesis of 6-(3-methyl-2-butenylamino)-2-methylthio-9-~-ribofuran0sylpurine, further showing the importance of base modification to the regulatory roles of tRNA (35). The results with the C6 mutant reported here differ from these studies in that no derepression of the ilvGEDA operon occurs in the presence of the undermodified tRNA", and preliminary results suggest that sulfur-deficient tRNA"' is normal in its pseudouridine content (36). It is of interest that the regulation of the expression of the gene for threonine deaminase was found to be more sensitive to changes in isoleucyl-tRNA than other enzymes of this operon (29,37). Our results agree with this, as the derepression of the ilvA gene was the first to be observed as E. coli C6 recovered from cysteine starvation. Further, whatever alteration of tRNA"" is responsible for the results given here, it is clear that the regulation of early and late genes is affected.
The changes in sulfur-deficient tRNA"', which give rise to its altered aminoacylation and chromatographic behavior, are not completely understood at present. Normally, growing E. coli C6 cells contain three species of tRNA: tRNA:' ' , presently uncharacterized but comprising less than 5% of the total; tRNA;'", a slowly acylated 4-thiouridine-containing species comprising 45% of the total; and tRNA:' , a species with a fast rate of aminoacylation which is 50% of the total (38). The latter tRNA'" is undermodified with respect to tRNA;", but it is not yet clear if these species are products of different genes or are related through base modification. Cysteinestarved cells contain one major tRNA"' whose rate of acylation is fast and is also similar to normal tRNAp in base composition (36). However, these two tRNAs are different, as they chromatograph distinctly on MAK and RPC-5 columns (Ref. 11 and Figs. 4 and 5). We report here that the tRNA"' of cysteine-starved cells is converted back to the normal pattern after 3 h of recovery on complete medium. This had also been observed with sulfur-deficient tRNA treated with sulfurtransferase in vitro, namely, that tRNAp was reformed and the rate of aminoacylation was reduced to normal levels (8). While the latter experiments strongly suggest that sulfurdeficient tRNA'Ie is an undermodified version of normal tRNA'", a careful structural analysis of these tRNAs must precede any definite conclusions. These experiments are currently in progress.