GENETICSMcappinig of the rel Loclus.Selectioln oj'Mlutanits altd TransdullctalitsPhlylogenly of the Originlacl rel Mutanit........Dominailce .....................................
ROLE OF TRANSFER RNA IN THE REGULATION OF RNA SYNTHESISOTHER PROPOSALS CONCERNING THE MECHANISM OF ACTION OF THE m-el LOCUS.PolyaniinesSuibstrate Regiulationl of RNA Synithesis ......
Regulation by Free Ribosome.s ..................
COORDINATE REGULATION OF RNA SYNTHESISDISCUSSION .. .. .LITERATURE CITED
.. 206
..206207207
..207
. 207208
I. 208208
.. 208
..209. 209210210210210
..211212212214214215217218221223
INTRODUCTIONAmiiino Acid Control of Ribonlucleic Acid
(RNA) SynthesisIn 1952, Sands and Roberts (87) reported the
results of a study on the relation between thesynthesis of proteins and of nucleic acids. Theyshowed that when a culture of a strain of Escher-ichia coli that required tryptophan and histidinewas deprived of either amino acid, not only didprotein synthesis stop, but there was also a reduc-tion in the rate of RNA synthesis. They inferredthat, since neither tryptophan nor histidine is ametabolic precursor of nucleic acid, the reductionin RNA synthesis was a secondary effect of theprimary blockage in protein synthesis. The impor-tant conclusion drawn was that there is a regula-tory connection between the synthesis of proteinsand of nucleic acids. These observations wereconfirmed and extended by other workers (36, 42,
'Present address: Department of Molecular Biology,University of Edinburgh, Edinburgh 9, Scotland.
43, 74). The original hypothesis was that thisregulation was effected by transfer RNA (tRNA)(48, 94), and it was then proposed (40) that un-charged tRNA inhibited the RNA polymerase,and that such inhibition resulted in the reducedsynthesis of all classes of RNA.One exception to the rule that starvation of a
bacterium for a required amino acid stops notonly protein synthesis but also RNA synthesis wasthe finding (7, 8) that, in the methionine-requiringstrain W6 of E. coli K-12, RNA synthesis con-tinues in the absence of methionine (see Fig. 1).This strain was referred to by Stent and Brenner(94) as having a "relaxed" control of RNA syn-thesis, in contrast to the "stringent" control thatnormally obtains. It was shown that the propertyof stringent or relaxed control of RNA synthesismaps as a single genetic locus, the RNA controlor "RC" locus.The object of this review is to relate the various
observations on the nature of the regulatory con-nection between protein synthesis and RNA syn-
Relative increase in Optical Densityof fully supplemented aliquot
FIG. 1. Effect of methionine starvation on the syn-thesis of RNA. The incorporation of radioactive uracilinto the acid-insoluble fraction oftwo different methio-nine auxotrophs, strains J-53 and W6, was followed inthe presence and in the absence of methionine. StrainJ-53 has a stringent control of RNA synthesis, whereasin strain W6 this control is relaxed. Redrawnfrom Stentand Brenner (94).
thesis, and to evaluate the theories on themechanism of this regulation that have so farbeen advanced.
Nomenclature
Various symbols have been used to representstrains with stringent and with relaxed control ofRNA synthesis. In this review, we shall use thesymbol rel to represent the genetic locus deter-mining this function. This symbol, unlike "RC,"conforms to the rules advocated by Demerec etal. for representing genotypes (19) and has alreadybeen used for this purpose (98). The mutationpresent in strain W6, being the one defining thislocus, is designated rel-1. In the absence of exten-sive complementation data, it is as yet not possibleto decide if all such mutations fall within thesame cistron.
In the system of Demerec et al. (19), the pheno-type symbol is clearly distinct from the genotypesymbol. The symbols RCStr and RCrel, for strin-
gent and relaxed control of RNA synthesis,respectively, can legitimately be used to describephenotype. Since they are unambiguous andwidely used, we urge their continued usage indescribing the phenotypes engendered by thewild-type and mutant alleles of the rel gene.
PHYSIOLOGICAL STUDIES
Amino Acid Starvation and RNA SynthesisAmino acid starvation. The principal physio-
logical difference between RCstr and RCrelstrains is manifested under conditions of effectiveamino acid starvation, when RCrel but not RCstrstrains continue to accumulate RNA. In aminoacid-starved RCstr strains, the rate of residualRNA synthesis is reduced to a few per cent of thatfound in the presence of the required amino acid,whereas amino acid-starved RCrel strains con-tinue to produce RNA at normal or slightly higherthan normal rates (94). Within these limits, con-siderable variation can occur in the rate of re-sidual RNA synthesis, as will be discussed pres-ently.To demonstrate the difference in the rate of
residual RNA synthesis between RCstr and RCrelstrains, bacteria can be amino acid-starved inseveral different ways. Perhaps the most fre-quently used technique involves the use of auxo-trophs which require an exogenously suppliedamino acid for growth. Removal of the aminoacid from the medium causes an immediate cessa-tion of protein synthesis in RCstr and RCrelstrains. Another technique involves the use ofamino acid analogues, such as f-2-thienylalanine,2-thiazolealanine, and 5-methyltryptophan (81).These analogues interfere with the synthesis of thecorresponding natural amino acid and thereforecreate a situation of amino acid starvation. Cer-tain strains of E. coli-for instance strain K-12but not strain B-can be starved for isoleucine byaddition of valine, which interferes with isoleucinebiosynthesis by feedback inhibition of the firstenzyme in the isoleucine biosynthetic pathway(78). Amino acid starvation resulting from theaddition of such inhibitory compounds to bac-terial cultures is useful for distinguishing betweenthe RCstr and RCrel phenotypes, since it makespossible amino acid starvation of bacteria thathave no amino acid requirements and because itis convenient for handling large numbers of testcultures.A third technique which creates a condition of
amino acid starvation involves a "shift-down,"or transfer of a bacterial culture, from a mediumcontaining a complete complement of aminoacids to a minimal medium containing only
nutrients required for growth. Since the repressionof the enzymes of amino acid biosynthesis in thecomplete medium results in a state of "physio-logical auxotrophy," the shiftdown creates atransient condition of amino acid starvation.RNA synthesis. The rate of residual RNA
synthesis under these conditions of amino acidstarvation can be measured by chemical estima-tion of the total amount of RNA. This methodmeasures net rather than total synthesis of RNA,since a situation in which RNA is synthesized andthen broken down is not detected. RNA synthesiscan also be measured by following the incorpora-tion of a radioactive precursor of RNA intotrichloroacetic acid-precipitable material. Suchincorporation may be detected even in the absenceof net RNA synthesis, again because of turnoverof unstable RNA. It must also be kept in mindthat failure to observe incorporation can beascribed to causes other than cessation of RNAsynthesis, for instance to the failure to take upthe precursor from the medium or to convert itto a substrate suitable for incorporation intoRNA. Thus it has been demonstrated that, duringamino acid starvation, RCstr bacteria assimilateand phosphorylate exogenously supplied pyrimi-dines much less readily than do RCrel bacteria(21). In this particular case, however, it is likelythat there is a connection between reduction inthe rate of RNA synthesis and in uptake of pyri-midines from the medium, although the nature ofthis connection is not clear.To date, various auxotrophic strains have been
starved for most of the 20 standard amino acids(32, 94), with the general result that, upon starva-tion, RNA synthesis ceases in RCstr strains andcontinues in RCrel strains. Fiil and Friesen (32)showed that in each of a series of independentlyisolated rel mutants, all derived from the samerel+ parent strain, the amount of radioactiveuracil incorporated during starvation for differentamino acids (threonine, leucine, arginine, andhistidine) was essentially the same. The onlyanomalous datum (3) so far has been the observa-tion that RNA synthesis ceases upon starvation ofa serine/gycine- rel-i strain. This anomaly doesnot simply result from the role of glycine as a pre-cursor of purine biosynthesis, since the presenceof adenine in the medium during starvation doesnot restore RNA synthesis.Two instances of variation in the rate of radio-
active uracil incorporation on amino acid starva-tion have been described. The first is in the seriesof strains isolated by Fiil and Friesen (32). Theydemonstrated that the rate of incorporationduring amino acid starvation varies widely be-tween strains and is characteristic for eachmutant. They concluded that these strains differ
in the amount of RNA that is synthesized duringamino acid starvation. In the other case, consider-able variation in the amount of uracil incorpora-tion was observed (3) among Arg- recombinantsof a cross between an Hfr strain carrying the rel-Imutation and a rel+ F- strain. Despite the widerange in the amount of uracil incorporation,recombinants could be clearly distinguished asRCstr or RCrel. It must be concluded, therefore,that genes other than the rel locus may have aneffect on this operational test of the residual rateof RNA synthesis in amino acid-starved bacteria.
Types of regulation of RNA synthesis. A cleardistinction should be made between the control ofRNA synthesis shown by RCOtr strains underamino acid starvation and the regulation of therate ofRNA synthesis during normal growth. Ingrowing cultures, the number of ribosomes perbacterium increases with increasing growth rate,as does tRNA (54). An exception to this rule isthat, at low growth rates, tRNA is synthesized inrelative excess (85). That is, bacteria are able toadjust the synthesis of these two classes ofRNA toa rate commensurate with the growth rate.The regulatory aberration of RCrel strains
manifests itself only during amino acid starvation,for, as demonstrated by Neidhardt (68), RCrelstrains are able to regulate RNA synthesis whensubjected to shifts in carbon or nitrogen sourcewhich either enhance or depress the rate of overallgrowth. The ability of RCrel strains to adjusttheir rate of RNA synthesis with growth ratehas been confirmed with an isogenic pair of RCstrand RCrel strains (G. Edlin, unpublished data).These results emphasize the fact that the rellocus affects RNA synthesis only during aminoacid starvation and that the general regulation ofRNA synthesis during exponential growth inRCrel strains is indistinguishable from that ofRCstr wild-type strains.
Consequences of RCrel ExpressionChloramphenicol particles and relaxed particles.
When bacteria are treated with chloramphenicol,protein synthesis is inhibited but RNA synthesiscontinues. The continuation of RNA synthesisin the absence of protein synthesis produces anaccumulation of "chloramphenicol particles"which have been presumed to be ribosomal pre-cursors, since the RNA encapsulated in theseparticles can be matured into complete, functionalribosomes when protein synthesis is restored byremoval of the chloramphenicol. An analogoussituation obtains during amino acid starvation ofan RCrel strain. Here, too, RNA synthesis con-tinues in the absence of protein synthesis, result-ing in an accumulation of "relaxed particles."Because of the information these particles may
provide in understanding the many steps in ribo-somal biosynthesis, their synthesis and composi-tion have been extensively studied. A consider-able literature exists on the properties of theseparticles (16, 49, 58, 67, 97), but, since thesestudies do not bear directly on the nature of therel mutation, we shall not discuss them further inthis review.
Permeability effects. There exist mutant strainsof Bacillus and closely related genera and of E.coli that excrete substantial amounts of glutamate.In general, these mutant strains have a require-ment for biotin, and the optimum condition forglutamate excretion is that of biotin-limitedgrowth (18). Two biotin-requiring strains of E.coli grown under conditions of biotin limitationhave been reported to have a lipid deficiency inthe membraneous components of the cell (15, 38).It has been proposed that this lipid deficiencyresults in a permeability defect which allowsglutamate to escape from the cell. In the absence ofan effective feedback control, the production andexcretion of glutamate continues.On amino acid starvation of an RCrel strain
but not an RCstr strain, glutamate is also excreted(P. Broda, in preparation). The rate of glutamateexcretion gradually increases after the onset ofamino acid starvation, reaching a maximum afterabout 30 min at 37 C. It is therefore proposedthat, during the relaxed response, there is an alter-ation in cellular permeability which becomes fullyexpressed at that time, allowing continued syn-thesis and excretion of glutamate. This hypothesisis consistent with the finding that RCrel strains ofE. coli become somewhat permeable to actino-mycin D during amino acid starvation (H. Mat-zura and P. Broda, in preparation).Amino acid sensitivities. In an amino acid shift-
down, a situation of temporary amino acid starva-tion is created and persists until the bacteriahave produced the amino acid biosyntheticenzymes. RCrel cells take much longer to recoverfrom such a shift-down than RCstr cells (3, 68),and this response of RCrel cells to a shift-downdepends upon their stage of growth (3). Appro-priate dilutions from a broth-grown culture of anRCrel methionine-requiring strain in variousstages of growth were plated onto tryptone-brothplates and onto minimal agar supplemented withmethionine. Whereas the colony counts on thebroth-agar were a measure of the culture'sgrowth, the minimal agar colony counts, althoughcoincident with the broth-agar colony counts inthe stationary phase of growth, were as low as20% of the broth values for cells at the beginningof the exponential phase of growth. It appears,therefore, that RCrel cells are especially sensitiveto a shift-down at this stage. This shift-down sensi-
TABLE 1. Amino acid sensitivities of recombinantsfrom a cross of the type Hfr RCrel X F- RCutr a
a Hfr strain Cavalli (Met- RCrel leucine-sensi-tive methionine-insensitive phenylalanine-insensi-tive) was crossed with an F- strain PA309 (Leu-His- Arg- RCstr leucine-insensitive methionine-insensitive phenylalanine-insensitive). Leu+ Met+His+ Arg- recombinants were tested for the RCphenotype and for their amino acid sensitivities(data from reference 3).
tivity is manifested in a dramatic fashion by theaddition of 500 ,ug/ml of leucine to the minimalagar. The percentage of survivors from an RCculture at the beginning of the exponential phaseof growth is now less than 0.1, and cells at allstages of growth, including those from the finalnongrowing culture, are sensitive to some degree.Recombinants from a conjugational cross
between an RCrel leucine-sensitive Hfr strain anda leucine-insensitive RCstr recipient were analyzed(3) for the linkage between the leucine-sensitivityproperty and the RC property (Table 1). Therewas a complete correlation, in that all the RCstrrecombinants were insensitive and all the RCrelrecombinants were sensitive.
Other sensitivities which at least partially cor-related with the RCrel phenotype were observedon the addition of phenylalanine and of methio-nine. Both parent strains used in the conjugationexperiment described above were insensitive tothese amino acids; however, all the RCrel recom-binant clones, but none of the RCstr clones, weresensitive to methionine (Table 1). Also, therewere phenylalanine-sensitive clones among theRCrel recombinants but not among the RC"trones. The conclusion was that RCrel bacteriatend to be more sensitive than RCstr cells to someamino acids in the shift-down condition. Thesensitivity to leucine and phenylalanine can berelieved by the addition of mixtures of leucine,isoleucine, and valine, suggesting that here, as invaline sensitivity, inhibition is caused by a de-rangement of the system which regulates thesynthesis of leucine, valine, isoleucine, and threo-nine (1). Sensitivity to a high external concentra-tion of an amino acid, leading to such a derange-ment, could occur in an RCrel strain but not in anRCstr strain if, on amino acid starvation, the
RCrel strain were to become more permeablethan the RCstr strain.
RCrel phenotype and conjugation. The RCrelphenotype might be a complicating factor in theanalysis of conjugation experiments. The yield ofrecombinants obtained in a genetic cross dependsboth upon the medium in which the mating isperformed and the constitution of the agar plateson which the recombinants are selected. Thelowest yields have been obtained with matingsin broth or Casamino Acids medium followed byplating on minimal agar plates (44, 82); these are,in fact, shift-down situations. The presence of theRCrel property might then influence both theyield and the types of recombinants isolated incrosses, especially when the recipient strain isRCrel. It happens that both Hfr strains, Cavalliand Hayes, were isolated from an F+ RCrel strain,so that many genetic analyses have been done onrecombinants of crosses of the type Hfr RCrel XF- RCtr.
GENETICS
Mapping of the rel LocusBacterial conjugation experiments showed that
the rel locus is located between the strA and glyAloci on the E. coli chromosome (2). More refinedtransductional mapping with phage P1 has indi-cated that the rel locus lies between the argA andcysC loci, or between minutes 53 and 54 of theTaylor-Trotter map (98). An analogous mutationin Salmonella typhimurium is reported to map in asimilar position in that species (60).A large number of RCrel strains of independent
origin have been isolated from RCstr strains byFiil and Friesen (32) and by Lavalle (50). Themutations carried by most of these mutant strainsmap in the same region as the rel locus. However,in three of Lavalle's isolates, the mutation doesnot map in the rel region (50). These threeisolates, although of RCrel phenotype, in thesense that RNA accumulates in the absence of arequired amino acid, have some physiologicalproperties which distinguish them from strainswith the classical RCrel phenotype.
Selection of Mutants and TransductantsEarly comparative investigations of the proper-
ties of RCrel and RCstr strains were done withnonisogenic strains. However, four procedureshave been developed to obtain closely relatedpairs of RCBtr and RCrel strains. The first of theseprocedures is that of penicillin selection (32); itdepends upon the observation that glucose-grownRCrel bacteria, which have been extensivelystarved for a required amino acid in the presenceof glucose and subsequently shifted to fresh
medium containing lactose as the carbon sourcetogether with the required amino acid, exhibit amuch greater growth lag than a similarly treatedRCstr strain (3). If penicillin is present in thislactose medium, the RCstr bacteria will be selec-tively killed before the RCrel bacteria beginmultiplying. If the penicillin is removed prior tothe onset of growth of the RCrel bacteria, aculture enriched in RCrel bacteria is obtained. Itwas found that mutagenesis of a bacterial culture,followed by five cycles of penicillin selection, wassufficient to allow the isolation of RCrel bacteria.It should also be possible to isolate RCstr deriva-tives of an RCrel strain by making use of theearlier resumption of growth by RCstr strainsafter such a glucose-lactose shift. The advantageof the penicillin method lies in the opportunityof selecting independently arising mutants from agiven parental strain, so that different mutantscan be compared. The major defect in this pro-cedure is that it is not known how many mutationsmay have occurred in the formation of the pheno-typically "relaxed" strain that is isolated.A second procedure, developed by Lavalle
(quoted in 32), involves transduction. It is imprac-tical to select directly for RCStr or RCrel trans-ductants; therefore, the desired rel allele is intro-duced as a joint transductant with the closelylinked thymine marker thyA. This method relieson the prior selection of a thyA mutation in therecipient strain into which the desired rel alleleis to be introduced by incubation of the Thy+wild strain in the presence of aminopterin ortrimethoprim (91). Since the cotransduction fre-quency of thyA with rel is only a few per cent, aconvenient refinement employs a lysate of phageP1 or phage 363 prepared from a donor strain ofthe genotype thyA+argA and carrying the relallele to be transferred (32,). This phage lysateis then used to infect the thyAargA+ recipient,and Thy+ transductants are then selected by plat-ing on thymine-free, arginine-supplemented agar.The Thy+ transductants are examined by replicaplating for clones that have acquired the donorargA allele. These thyA+argA transductants arethen tested for the presence of the RCrel or RCstrcharacter; the cotransduction frequency of argAand rel is of the order of 25% (32). An advantageof this method is that it is possible to use thearginine requirement for the test of the RC pheno-type, so that RNA synthesis in different strainscan be compared under the standard condition ofarginine starvation.The observation that RCrel or RCstr strains can
be prepared by transduction supports the con-clusion that a small region of the E. coli chromo-some is relevant to the expression of this pheno-type (2). Fiil and Friesen have shown, further-
more, that the mutational event involved in theorigin of the penicillin-selected RCrel mutants isalso confined to this small region of the genome.Lysates of the transducing phage 363 were pre-pared from a series of RCrel strains isolated by thepenicillin technique. These lysates were used toinfect an argA RCstr strain, and argA+ trans-ductants were selected. It was found that in eachcase a proportion of the argA+ transductants wereof RCrel phenotype and that, on amino acidstarvation, the transductants exhibited the samekinetics of RNA synthesis as the original donorstrains. It was concluded that the "degree ofrelaxedness" reflects the particular rel mutation.The third procedure was developed by R. E.
MacDonald (personal communication); it exploitsa morphological difference between colonies ofamino acid-starved clones of RCstr and RCrelbacteria. Several thousand RCstr amino acidauxotrophs are spread on minimal agar whichcontains only about 1 Ag/ml of the requiredamino acid, and incubated for 24 to 48 hr.Microcolonies appear, but their continued growthis restricted because of the amino acid limitation.The colonies of RCstr bacteria cease RNA syn-thesis at the time of amino acid exhaustion.Colonies of RCrel bacteria, however, continue tosynthesize RNA for a period after amino acidexhaustion. This difference in residual RNAsynthesis is reflected in the colonial morphology,so that RCstr and RCrel colonies can bedistinguished by phase contrast microscopy.Figure 2 illustrates this difference.An analogous procedure was employed by
Martin (60) to isolate an RCrel derivative of ahistidine-requiring strain of S. typhimurium. Aculture of this strain was mutagenized and por-tions containing 500 to 1,000 cells were placed onsterile Millipore filters (Millipore Corp., Bedford,Mass.) on histidine-supplemented minimal agarplates. When microcolonies had appeared, thefilter was transferred to another plate lackinghistidine but containing radioactive uridine.Autoradiography was used to locate colonieswhich had incorporated a large amount of uridine;in this way, an RCrel strain was isolated. Themutation was cotransducible with the argBmarker, the analogue of the argA marker of E.coli (60). These techniques for isolating RCrelstrains were used by Martin (60) and Edlin (23)to demonstrate that the RCrel phenotype, whichallows RNA synthesis to continue during histi-dine or tryptophan starvation, was not able toabolish the effects of a strong polar mutation ineither the histidine or tryptophan operons.An unexplained finding is that, despite ef-
forts in various laboratories, it has not beenpossible to isolate RCrel derivatives of the RCstr
FIG. 2. Morphological difference between RCstr andRCrel strains. A mixture of the RCOtO strain CP78 andits RCrel derivative CP79 (31) was plated on minimalagar medium containinig limiting amounts of a requiredamino acid. Colonies are observed with a low-powermicroscope with intense oblique lighting at an angle ofabout 70° from below. Colonies of the RCrel strain havea mottled appearance, whereas colonies of the RCstrstrain diffuse light more evenily and are somewhat moreopaque.
E. coli strain B, neither by transduction (N. Fiil,J. Gallant, personal communication; P. Broda,unpublished experiments) nor by penicillin selec-tion (P. Donini, personal communication). Revert-ants of RCrel bacteria to the RCstr phenotypehave been reported (R. Lavalle, in 54). However,no RCstr revertants of the E. coli strain 58-161,carrying the classical rel-i mutation, have beenobserved.
Phylogeny of the Original rel MutantStrain W6 (RCrel), in which Borek et al. (8)
first observed the phenomenon ofRNA accumula-tion during amino acid starvation, is a biotin-independent (Bio+) revertant of the Met-Bio-RCstr strain 58-161. It is not clear whether theloss of the biotin requirement and the acquisitionof the RCrel phenotype are related. Other Bio+revertants of the original 58-161 strain (2; P.Broda, unpublished results) have retained theirRCstr character. Hence, there does not seem to bea necessary connection between the Bio -- Bio+and the RCstr - RCrel mutations. One unex-plained coincidence does exist, however. It has
been observed (P. Broda, unpublished data) thatboth strain 58-161 (RCstr) and strain W6 (RCrel),excrete glutamate on methionine starvationwhereas, in other strains of E. coli, only RCrelbacteria excrete glutamate; the correspondingRCstr strains do not. In other bacterial species(18), the property of glutamate excretion isalmost invariably associated with the requirementfor biotin, so that it is possible that there is somerelation between these properties.
DominanceIt has been shown (N. Fiil, personal communica-
tion) that the rel+ allele is dominant over themutant rel allele. This was demonstrated by con-structing partial diploids by using an F-primefactor carrying the rel locus. Fiil has shown thatthe phenotype of partial diploids of the genotypesrel/Frel+ and rel+/Frel is RCstr; RNA synthesisis as stringently controlled as in the normalhaploid rel+ strain. Experiments to measure thekinetics of expression of the RCstr and RCrelphenotypes are in progress. Clearly this type ofinformation will be of great value for the evalua-tion of hypotheses on the nature of the functionof the rel gene.This system has also been used (N. Fiil, personal
communication) to test for complementationbetween different RCrel mutants. In the few casesthat have so far beentested, such complementationhas failed to produce the RCstr phenotype.
ROLE OF TRANSFER RNA IN THE REGULATIONOF RNA SYNTHESIS
Stent and Brenner (94) postulated the inter-vention of amino acids in the regulation of RNAsynthesis based on genetic and physiologicalstudies of RCstr and RCrel bacterial strains. Inessence, they suggested that any of the completeset of tRNA molecules could function as inhibi-tors of RNA synthesis when uncharged withamino acids, but that on amino acylation thetRNA molecule was no longer inhibitory. Thus, asufficiency of amino acids results in completecharging of the transfer RNA, whereas a defi-ciency would cause one or more tRNA species tobecome uncharged. This uncharged tRNA couldthen repress RNA synthesis by inhibition of theRNA polymerase. This model implicitly containsthe concept of coordinate regulation of RNAsynthesis, since it envisages a general reductionin the synthesis of all classes of RNA.Kurland and Maal0e (48) presented a similar
model based primarily on the effect of chloram-phenicol on RNA synthesis. Owing to the impor-tance of chloramphenicol in the development ofthe ideas concerning the regulation of RNA
synthesis, some of the results obtained with thisand other antibiotics will be discussed. At rela-tively low concentrations, chloramphenicol per-mits some residual protein synthesis; with suchtreatment, RNA is maximally stimulated in auxo-trophic strains if catalytic amounts of requiredamino acids are present in the medium (74). Incontrast, at higher concentrations of chloram-phenicol which completely inhibit proteinsynthesis, the presence of exogenous amino acidsis no longer required for maximal stimulation ofRNA synthesis (4, 5). At first glance it seemscontradictory that stopping protein synthesis byamino acid starvation represses RNA synthesisin RCStr strains, whereas stopping protein syn-thesis by addition of chloramphenicol does notdepress and may actually stimulate RNA syn-thesis. The most straightforward explanation forthis paradox appears to be that RNA synthesisis stimulated, not by any direct action of chloram-phenicol on the synthesis of RNA, but because itpermits the intracellular build-up of amino acidpools and of aminoacylated tRNA. It is presumedthat the breakdown of protein during amino acidstarvation contributes to this intracellular accu-mulation of amino acids when de novo proteinsynthesis is prevented by chloramphenicol orother antibiotics. In fact, it has been demon-strated that tRNA becomes fully charged withamino acids when chloramphenicol is added to anamino acid-starved culture (29, 63). Morris andDeMoss (63) observed that, although both chlor-amphenicol and puromycin stimulated RNAsynthesis during leucine starvation of an auxo-troph, the leucyl-tRNA became fully chargedonly on addition of chloramphenicol and not onaddition of puromycin. This observation led themto propose that puromycin can stimulate RNAsynthesis without increasing the level of chargedtRNA. However, Ezekiel and Elkins (in prep-aration) have tested a variety of antibiotics,including puromycin, for their ability to stimu-late RNA synthesis and have concluded that allof the compounds tested stimulateRNA synthesisby sparing amino acids and increasing the amountof aminoacyl-tRNA in the cell. (For a moredetailed analysis of these experiments, see"Regulation by Free Ribosomes," below.)The important conclusion that amino acid
regulation of RNA synthesis depends on a com-plete complement of aminoacyl-tRNA and notmerely on free amino acids was established withbacterial mutants. Fangman and Neidhardt (30,31) isolated a p-fluorophenylalanine-resistantmutant of E. coli that was shown to have analtered phenylalanyl-tRNA synthetase. This strainwas unable to activate p-fluorophenylalanine andhence to incorporate the analogue into protein.
Upon phenylalanine-starvation of this mutantstrain, addition of the analogue not only did notrestore protein synthesis but also did not restoreRNA synthesis (27, 31). This result suggestedthat not only the presence of amino acids, buttheir activation and attachment to tRNA, isrequired for the promotion of RNA synthesis.This conclusion was confirmed by the isolation oftemperature-sensitive bacterial mutants (6, 25,69, 100), one of which was shown to possess analtered valyl-tRNA synthetase which is active atlow temperature (30 C) but inactive at high tem-perature (42 C). At the high temperature, anRCstr strain carrying this mutation was unable tosynthesize RNA, although all required aminoacids were present, whereas an RCrel straincarrying the same mutation was able to synthe-size RNA at the high temperature. Hence itfollows that an essential step in the regulation ofRNA synthesis by RCstr strains is the attach-ment of amino acids to tRNA. If one constructsan RCrel strain that carries the same tempera-ture-sensitive synthetase mutation, RNA syn-thesis continues at the normal rate after proteinsynthesis is stopped by shift of the culture to thenonpermissive temperature (42 C). This resultsuggests that the rel mutation allows RNAsynthesis to proceed in the absence of proteinsynthesis without requiring the amino acylationof tRNA. This conclusion is strengthened bythe failure of chloramphenicol to appreciablystimulate RNA synthesis in a temperature-sensitive synthetase RCstr strain at the hightemperature (Ezekiel and Elkins, in preparation).This observation constitutes the only clear differ-ence between the mechanism of stimulation ofRNA synthesis by chloramphenicol and the relgene in the absence of protein synthesis. It wasalso possible to demonstrate with these mu-tants that repression of the enzymes of thevaline biosynthetic pathway depends not on freevaline but on the supply of valyl-tRNA (24).Thus, it appears that aminoacylated tRNA maybe important in the regulation of diverse cellularprocesses.With these results in mind, the discussion of the
original hypotheses of Stent and Brenner (94)and of Kurland and Maaloe (48) can be con-tinued. The model gained support from the experi-ments of Tissieres et al. (99), who demonstratedthat in vitro, uncharged tRNA was a considerablybetter inhibitor of the RNA polymerase thancharged (aminoacylated) tRNA. These authorsconcluded that regulation of RNA synthesismight occur by inhibition of theRNA polymeraseby uncharged tRNA (40, 99). Subsequent studies,however, cast doubt on the correctness of thismodel. Bremer et al. (9) made a more detailed
examination of the inhibition of the RNA polym-erase by charged and uncharged tRNA in vitro.They found that prior reaction of the RNA poly-merase with deoxyribonucleic acid (DNA) pro-tects it against subsequent inhibition by tRNAand that tRNA molecules were heterogeneouswith respect to RNA polymerase inhibition.Moreover, the difference in inhibition betweencharged and uncharged tRNA was so small thatit seemed unlikely that it could account for theobserved in vivo regulation of RNA synthesis.
Ezekiel and Valulis (29) were able to increasethe internal cellular tRNA concentration four- tofivefold by incubating amino acid-starved bac-teria with chloramphenicol, which allows tRNAand ribosomal RNA (rRNA) to accumulate.After this treatment, the rate of RNA synthesisin these bacteria and in a control culture, whichhad no prior chloramphenicol treatment, wasvaried by starvation for a required amino acidin the presence of different amounts of chlor-amphenicol. For any given rate ofRNA synthesis,the concentration of uncharged tRNA was four-to fivefold higher in the chloramphenicol-pre-treated culture than in the control. This suggestedthat uncharged tRNA does not regulate RNAsynthesis by inhibition of the RNA polymerase.
Differences in the intracellular condition oftRNA from RCstr and RCrel strains have beenexamined in some detail. Initial experiments weredesigned to test the specificity of the attachmentof amino acids to tRNA in extracts preparedfrom RCstr and RCrel strains (59). From theseextracts, purified tRNA and activating enzymeswere prepared and used in different combinationsto detect any break-down in the specificity of thetransfer reaction. No breakdown in specificitywas detected; it was therefore concluded that therel gene does not affect this reaction. Furtherexperiments explored the intracellular conditionof tRNA with respect to the degree of esterifica-tion in vivo of its terminal adenosyl 3'(2')hydroxyl with amino acids. These experimentsshowed that about 70% of the tRNA moleculesare esterified in vivo; no significant differenceswere found between RCStr and RCre I strains (101).
Differences among each of the componentsnecessary for RNA synthesis in RCstr and RCrelstrains have also been sought by Olenick andHahn (73) by in vitro studies. They preparedextracts of DNA, tRNA, and RNA polymerasefrom RCstr and RCrel bacteria and used theseextracts to stimulate RNA synthesis in an in vitrosystem. Different combinations of the extracts(for instance DNA and tRNA from RCStr bac-teria and RNA polymerase from RCrel bacteria)were added to an in vitro system. RNA synthesiswas stimulated to the same extent by all of the
combinations tested, so it was concluded that therewere no gross differences among these com-ponents in RCstr and RCrel bacteria.
Finally, an important observation regarding therole of aminoacylated tRNA in RNA synthesishas been made by inhibiting protein synthesiswith trimethoprim. Trimethoprim has been shownto inhibit the enzyme dihydrofolate reductase(11); this results in the cell's inability to synthe-size N-formylmethionyl-tRNA. Since this par-ticular species of tRNA is required for peptide-chain initiation, trimethoprim prevents proteinsynthesis in bacteria (26). More importantly forthis discussion, trimethoprim has been shown tostop RNA synthesis in RCstr strains but not inRCrel strains (90). This observation is significantbecause it implies that RNA synthesis dependsnot only upon the aminoacylation of all species oftRNA, but on some reaction subsequent to theattachment of amino acids to tRNA, since theformylation of the methionyl-tRNA occurs afterattachment of methionine to its cognate tRNA.The main conclusion to be drawn from theseexperiments is that RNA synthesis in amino acid-starved RCstr strains depends either directly onN-formylmethionyl-tRNA or, as seems morelikely, on one of the later steps in protein syn-thesis, which cannot occur if peptide-chain initia-tion is prevented.
OTHER PROPOSALS CONCERNING THEMECHANISM OF ACTION OF THE
rel Locus
Polyamines
A series of studies (13, 14, 76, 77) have led S. S.Cohen and his co-workers to propose that thepolyamines putrescine and spermidine have acausal role in the control of RNA synthesis. Theirinitial studies were made with the RCstr 15TAU-strain of E. coli, which requires thymine, arginine,and uracil for growth (47). When this strain wasstarved for arginine (or glucose), thereby haltingRNA synthesis, there was a cellular accumula-tion of putrescine but not of spermidine (76).In contrast, in the presence of chloramphenicol,which permits continued RNA synthesis in theabsence of arginine, a situation analogous toarginine starvation of an RCrel strain, spermidineaccumulates but putrescines does not. It was alsoobserved that the addition of spermidine to themedium can stimulate RNA synthesis in the argi-nine-starved RCstr strain, and that this stimula-tion could be prevented by the concomitant ad-dition of putrescine.
In a subsequent report (13), polyamine syn-thesis was studied in an RCrel derivative ofstrain 15TAU, which was compared to the parent
RCstr strain. It was reported that on argininestarvation of the RCrel strain, but not of theRCstr strain, spermidine accumulated, and thatthis accumulation inhibits putrescine biosyn-thesis. Furthermore, during normal growth, thelevel of spermidine was found to be somewhathigher in the RCrel strain than in its RCstrparent, although the amount of RNA was alsofound to be higher in the RCrel strain than in theRCstr parent strain. A further observationwas that on uracil starvation of the RCrel strain,but not the RCRtr strain, spermidine also ac-cumulated. This result was interpreted to meanthat spermidine accumulation does not neces-sarily depend on continued RNA synthesis,although the intracellular accumulation of freespermidine is correlated with RNA accumulation.As a result of these studies, it was suggested thata relaxed strain may be one in which putrescinebiosynthesis is inhibited by spermidine.The comparison between the accumulation of
polyamines and RNA synthesis was extendedin a later report (77). For these studies, the TAURCstr strain was compared to a K-12 RCrelderivative. A parallelism was demonstrated be-tween the intracellular accumulation of spermi-dine and RNA under a wide variety of growthconditions. Measurements were made underconditions of normal growth, methionine stimu-lation ofRNA synthesis, starvation for an aminoacid or thymine, and inhibition by chlorampheni-col or streptomycin. The general conclusion fromthese studies was that RNA synthesis is regulatedby appropriate concentrations of spermidineand putrescine.The stimulation of RNA synthesis by addition
of spermidine to an arginine-starved RCstrstrain appears similar to that already describedfor a variety of antibiotics such as chlor-amphenicol, streptomycin, tetracycline, etc. How-ever, no measurements of the effect of polyamine-addition on protein synthesis were reported inthe initial study (76). It has been found (Edlinand Stent, unpublished data) that addition ofspermidine to a growing bacterial culture causeda marked reduction in the rate of protein synthesis.Similar observations were made by Ezekiel andBrockman, who examined the effects of spermidineon RNA synthesis in some detail (28). Theymeasured the level of arginine-charged tRNAafter the addition of spermidine to arginine-starved RCstr bacteria in amounts sufficient tostimulate RNA synthesis. They found that thelevel of arginine-charged tRNA increased asthe stimulation of RNA synthesis increased. Theyalso demonstrated that addition of spermidine toamino acid-starved bacteria enhances proteinturnover, thereby increasing the intracellular avail-
ability of amino acids. Their findings support theidea that spermidine stimulates RNA synthesisin amino acid-starved RCstr strains by the spar-ing of amino acids and not by a more directinvolvement in the regulatory process.
Against this view are the data of Raina, Jansen,and Cohen (77), who have shown that whenspermidine is added to an arginine-starved cul-ture of strain TAU RCstr to a concentration of40 mm, RNA synthesis is stimulated, as measuredby radioactive uracil incorporation, whereasprotein synthesis also increases slightly, asmeasured by radioactive leucine incorporation.They infer from this result that the stimulationof RNA synthesis can occur without a corre-sponding intracellular buildup of the amino acidpool. Difficulties arise in comparing experimentsthat involve the use of spermidine because ofthe strong dependence on concentration and asharp dependence on the pH of the mediumbeing used.Raina and Cohen (76) also observed that radio-
active uracil incorporation into RNA continuedfor a considerably longer period when RNA syn-thesis was stimulated by addition of spermidinerather than by chloramphenicol. This findingmight be explained if spermidine is able to sta-bilize RNA and thereby retard its breakdown.Furthermore, Lazzarini and Santangelo (52)have measnred RNA synthesis and accumulationin B. subtilis during chloramphenicol inhibition ofprotein synthesis and have concluded that thefailure to accumulate RNA for more than 30 to40 min after addition of chloramphenicol resultsnot from the cessation ofRNA synthesis but fromthe rapid breakdown of ribosomal RNA in thepresence of chloramphenicol. It is possible that asimilar situation obtains during chloramphenicoltreatment of E. coli strains (71).As discussed earlier, glutamate is excreted into
the growth medium during arginine starvation ofthe RCrel 15TAU- strain but not in the RCstrstrain (P. Broda, in preparation). Since spermidineis also excreted by the RCrel strain, it is possiblethat the excretion of the two substances is related.However, during uracil starvation, spermidineappears to accumulate only in the RCrel strain,whereas glutamate accumulates in both the RCBtrand RCrel 15TAU- strains (P. Broda, in prep-aration). This result makes it difficult to draw anyconclusions concerning the relation between glu-tamate and spermidine accumulation and theregulation of RNA synthesis in RCrel strains.
In some respects, the choice of the 15TAU-strain for these experiments is unfortunate, be-cause of the biosynthetic interconnections ofarginine, uracil, glutamate, and polyamines. Asshown by Morris and Pardee (66), putrescine,
which is a precursor for spermidine, can be syn-thesized from either ornithine or arginine. Sincethe 15TAU- strain requires arginine, it is possiblethat polyamine synthesis in this strain is affectedby the presence of arginine during growth and itssubsequent removal. Moreover, glutamate is con-verted to glutamic semialdehyde and then toornithine, so that glutamate accumulation in thesestrains may also contribute to spermidine accumu-lation. The manner in which the synthesis of thesecompounds is regulated and interconnected inRCstr and RCrel strains of bacteria grown underdifferent physiological conditions remains to beclarified before the correlation between their ac-cumulation and the regulation of RNA synthesiscan be understood.
Substrate Regulation ofRNA SynthesisIt has been pointed out by several authors (12,
21, 33, 37) that RNA synthesis might be regulatedby the supply of its nucleotide precursors. Thedetailed investigation of this hypothesis has be-come possible largely as a result of two develop-ments. The first was the elucidation of the mech-anism of RNA synthesis by RNA polymerasefrom a DNA template. These in vitro studiesestablished that the ribonucleoside triphosphateswere the substrates for RNA synthesis; the discus-sion that follows assumes that they are the im-mediate precursors for RNA polymerization.The second advance has been in the develop-
ment of ion exchange thin-layer chromatography(79, 80) as a technique for the rapid and quantita-tive separation of nucleotides. This method hasbeen extensively used to investigate changes inbacterial nucleotide pools (12, 21). Operationally,nucleotide pools have been defined as acid-solublematerial (i.e., small molecules) that diffuse freelyfrom a bacterium when its membrane is madepermeable by either acid or detergent treatment.It is possible that there is compartmentalizationof nucleotides into "private pools" (10); if all theacid-soluble pools are indeed not available forRNA synthesis, then the significance of anychanges in the nucleotide precursor pools becomesdifficult to interpret. However, it has also beenargued that it is unnecessary to postulate com-partmentalization (72, 86); in this discussion, weshall assume that all nucleotides are available fornucleic acid synthesis.
Goldstein et al. (39) measured changes innucleotide pools in a strain of E. coli and foundthat, in general, nucleotide pools tended to in-crease during leucine starvation. These analyseswere made by the rather cumbersome techniquesof column chromatography and electrophoresis.By using thin-layer chromotography, Edlin andNeuhard (21) measured changes in the nucleoside
triphosphate pools in an isogenic pair of RCstrand RC rel strains. Theyreporteda gradual declinein all triphosphate pools to approximately halfthe initial levels after 1 hr ofamino acid starvationin the RCstr strain, but in the RCrel strain therewas a gradual rise in all the triphosphate poollevels, which approximately doubled after a simi-lar period of starvation. Since the rate of RNAsynthesis fell to a few per cent of the exponentialrate within about 1 min at 37 C after removal of arequired amino acid in an RCstr strain, and no
change in nucleoside triphosphate pool levels wasobserved for at least 10 min after removal of theamino acid, it was concluded that it was unlikelythat the dramatic reduction in the rate of RNAsynthesis could be attributed to a lack of sub-strates necessary for RNA synthesis.An important difference between RCstr and
RCrel strains was noted. During amino acidstarvation, RCBtr strains become refractory to theassimilation and phosphorylation of pyrimidines,and, to a lesser extent, of purines from the me-
dium. RCrei strains, which continue to synthesizeRNA in the absence of the required amino acid,are able to incorporate purines and pyrimidinesfrom the medium in normal fashion. No evidencewas presented to indicate a causal relation be-tween the failure to take up and phosphorylateuracil from the medium and the inability to syn-
thesize RNA.Gallant and Cashel (37) re-examined the effect
of amino acid starvation on nucleotide pools byusing bacteria which were rendered permeable tophosphorylated compounds by plasmolysis inhypertonic sucrose (41). When these sucrose-
shocked bacteria were incubated with nucleosidetriphosphates, radioactive uridine triphosphate(UTP) was incorporated into RNA in RCStrstrains in a manner that was only weakly depend-ent on the presence of amino acids. The incor-poration of radioactive uracil or uridine mono-
phosphate (UMP), however, was more stronglyamino acid dependent in RCstr strains than inRCrel strains. These authors proposed that theregulation ofRNA synthesis is effected by regula-tion of UTP synthesis.
These experiments are difficult to interpret in a
conclusive manner for several reasons. The bac-terial population is probably heterogeneous withrespect to the degree to which plasmolysis hasaffected the metabolism of different individualsin the population. Indeed, the rate of RNA syn-thesis in these bacteria after plasmolysis is onlyabout 5% of the rate in exponentially growingbacteria, which is characteristic of the rate ofRNA synthesis usually obtained in vitro. Oneprediction of the model proposed by Gallant andCashel was that, since synthesis of UTP was in-
ferred to be amino acid dependent in RCstr butnot in RCrel strains, it might be expected that theUMP or uridine diphosphate (UDP) kinase re-action would be amino acid dependent in extractsfrom RCrel strains. Such an amino acid depend-ence was looked for in vitro; but no significantdifferences were detected (J. Gallant, personalcommunication). However, it is possible thatnucleoside triphosphate synthesis could be af-fected by mechanisms other than by an alterationof the nucleotide kinase reaction.
Subsequently, Cashel and Gallant examinedthe amino acid dependence of nucleoside tri-phosphate formation in RCStr and RCrel strains(12) and, from the observed differences in the twostrains, concluded, in contrast to the earlier con-clusion of Edlin and Neuhard (21), that RNAsynthesis was regulated by nucleoside triphos-phate synthesis. They reported that synthesis ofnucleoside triphosphates is reduced in RCstrstrains in the absence of amino acids and is unaf-fected in RCrel strains. In addition, they showedthat the levels of UTP and cytidine triphosphate(CTP) fall rapidly to about half their normallevels in an RCstr histidine-requiring strain atabout the same time that histidine was exhaustedfrom the medium. They have also tried to estab-lish a causal relationship between RNA synthesisand triphosphate synthesis by measuring thecapacity of RCstr strains to take up and phos-phorylate uracil under conditions where RNAsynthesis has been blocked by actinomycin orproflavine. These experiments demonstrate thatRCstr strains maintain normal triphosphate syn-thesis if RNA synthesis is prevented by meansother than amino acid starvation. They againconcluded from this series of experiments thatRC"lr strains respond to amino acid deficiency bylimiting triphosphate synthesis, and that this isthe cause of the reduction ofRNA synthesis. Theyhave suggested (12) that it is phosphorylation perse which is affected by the state of the rel gene.The discrepancy between the abrupt fall in
UTP and CTP pool levels in an RCBtr histidineauxotroph upon exhaustion of histidine from themedium, reported by Cashel and Gallant (12),and the slight changes reported by Edlin andNeuhard (21) could result from differences inexperimental procedure. In one case, amino acidstarvation was produced by filtration and resus-pension of the culture (21), and in the other caseitwas achieved by growth in limiting histidine (12).In addition, starvation was for histidine in oneinstance and for arginine or leucine in the other.As mentioned earlier, RNA synthesis can also
be prevented in RCstr strains by shifting a bac-terial strain which has a temperature-sensitiveactivating enzyme from growth at low tempera-
a A culture of each strain was grown for severalgenerations in minimal glucose medium plus therequired supplements and radioactive phosphate.The cultures were transferred from growth at30 to 42 C, and portions of the culture were re-moved for triphosphate pool analysis at the timesindicated. All portions were collected and an-alyzed as described in Edlin and Neuhard (21).All values are given as millimicromoles per 5 X 108bacteria. We are grateful to F. C. Neidhardt forsupplying strain NP29 and to S. Kaplan for strain1OB6.
b Time after shift to 42 C.
ture (30 C) to a nonpermissive temperature (42C). The situation at the high temperature isanalogous to that created by amino acid starva-tion but is more specific in that RNA synthesis isblocked by alteration of a single enzymatic func-tion, rather than by the metabolically moredrastic procedure of amino acid withdrawal.Table 2 illustrates the changes that occur in thetriphosphate pools in two distinct E. coli strains,each with a temperature-sensitive valine-activat-ing enzyme, after a shift from 30 to 42 C. In onestrain, the pyrimidine triphosphate pools riseappreciably; in the other strain, the triphosphatepools appear to remain constant or fall slightly.If a culture of the parent wild-type strain withoutthe temperature-sensitive lesion is subjected to thesame shift from 30 to 42 C, all the nucleoside tri-phosphate pools immediately rise slightly, pre-sumably in response to the increased growth rateat the higher temperature.These results make it seem unlikely that RNA
synthesis stops during amino acid starvation dueto substrate limitation. However, the only tri-phosphate pool which fails to show an increasein either of the experiments presented is guanosinetriphosphate (GTP) and, in fact, the GTP pooldeclines in each instance. On the basis of the datapresented here, we would suggest that RNA syn-thesis is not regulated by substrate availability.However, these data leave open the possibilitythat GTP is involved in the amino acid regulation
of RNA synthesis. The changes in the GTP poollevel are not so great as to make it likely that itcould block RNA synthesis by substrate limita-tion. However, it is conceivable that a slight de-crease in the intracellular GTP concentrationcould prevent RNA synthesis through its role inprotein synthesis. It has been reported, for ex-ample, that GTP is required for binding of amino-acyl-tRNA to ribosomes (53). If, as suggestedmore recently by Cashel and Gallant (12), theprimary effect of amino acid-starvation of RCstrstrains is to hinder general phosphorylation,then it may be possible to formulate a theory inwhich the pools of nucleoside triphosphateswould remain at normal levels under conditionsin which stable RNA synthesis was specificallyinhibited.
Regulation by Free Ribosomes
The regulation of RNA synthesis by the intra-cellular concentration of free ribosomes has beensuggested by several investigators (62, 64, 84).In their models, it is envisaged that amino acidstarvation of an RCstr strain results in the break-down of polysomes or that it prevents their for-mation, thereby raising the level of free ribosomes,and that these free ribosomes repress RNA syn-thesis in an unspecified manner. According tothis model, the RCrel allele would allow polysomeformation in the absence of amino acids, thuspreventing the accumulation of free ribosomes,so that RNA synthesis can continue.
In certain strains of E. coli, addition of methio-nine to the growth medium causes a stimulationof RNA synthesis, while leaving the rate of pro-tein synthesis unchanged (62, 83). Matchett hasalso shown (62) that methionine stimulates thematuration of "chloramphenicol particles" intomature ribosomes after the removal of chloram-phenicol from the culture. He suggests that thisenhanced maturation of ribosomes permits theirincorporation into polysomes and results in adecrease in free ribosomes; because of a feedbackmechanism of the type outlined above, this de-crease in free ribosomes would then allow RNAsynthesis to proceed. Methionine stimulation ofRNA synthesis also results in a significant in-crease in the accumulation of cellular spermidine,as shown by Raina, Jansen, and Cohen (77).However, since methionine is also a polyamineprecursor, it is difficult to distinguish cause andeffect in these experiments.
It is not clear that there is any connection be-tween the small stimulation of RNA synthesis bymethionine observed in these studies and the effectof the rel gene on RNA synthesis. However,
similar conclusions on the regulation of RNAsynthesis by free ribosomes have been drawn fromstudies on polysome formation in amino acid-starved RCstr and RCrel strains (64, 84). It wasshown that, during amino acid starvation of anRCstr strain, there is conversion of polysomes tomonosomes (64). These monosomes are rapidlyreformed into polysomes if either the requiredamino acid or chloramphenicol is added to theculture. In contrast, polysome formation in theRCrel strain is unaffected by amino acid starva-tion and remains at the same level as in exponen-tially growing bacteria. It was concluded fromthese observations that a high level of free ribo-somes (that is, ribosomes not engaged in proteinsynthesis), present in amino acid-starved RCstrbacteria but not in amino acid starved RCrelbacteria, inhibits RNA synthesis. Morris andDeMoss also noted (64) that polysomes ap-parently reformed in amino acid-starved RCstrstrains on addition of chloramphenicol morerapidly than could be accounted for by de novosynthesis of messenger RNA (mRNA). They con-cluded that mRNA was already present in thesecells during starvation, and therefore that regula-tion of RNA synthesis is not co-ordinate. Thisaspect of RNA regulation will be discussed in alater section.More recently, however, Friesen (35) re-ex-
amined polysome formation in several strains ofE. coli in exponential growth and on amino acidstarvation. He found that the extent of breakdownof polysomes in amino acid-starved RC9tr strainsdepends both on the bacterial strain used and theamino acid withheld, so that the nature of theconnection between the inhibition of RNA syn-thesis and disappearance of polysomes is notclear.
Hartwell and McLaughlin (45) have isolatedtemperature-sensitive isoleucyl-tRNA synthetasemutants of yeast. As in E. coli, a shift to hightemperature results in cessation of RNA andprotein synthesis with no appreciable change inpolysome content (Hartwell and McLaughlin,personal communication). This observation alsocasts some doubt on the existence of a necessaryconnection between the cessation of RNA syn-thesis and disappearance of polysomes.The idea that the regulation of RNA synthesis
is mediated by polysomes or by free ribosomes,rather than by tRNA, was tested by Ezekiel andBlumenthal (in preparation). They postulated that,if starvation for different amino acids results inthe formation of different specific inhibitors (forinstance, in the form of charged or unchargedtRNA molecules), these different inhibitors mightinhibit RNA synthesis with different efficiencies.They reasoned that partial starvation for different
amino acids would then produce different rela-tionships between the rates of synthesis of RNAand of protein. They therefore compared therates of RNA synthesis and protein synthesisunder conditions in which protein synthesis waspartially blocked by the addition of amino acidanalogues, each causing starvation for a differentamino acid. When the rate of RNA synthesis wasreduced to about 35% of the exponential growthrate, protein synthesis was in each case reducedto about 70%. This finding has been interpretedas rendering less likely the possibility that RNAsynthesis is regulated by 20 different inhibitors,and as being consistent with the hypothesis of aribosomal regulation of RNA synthesis.However, Ezekiel and Elkins (in preparation)
interpret other evidence as being incompatiblewith such a ribosomal model. They examined thecapacity of a wide variety of antibiotics (chloram-phenicol, chlortetracycline, puromycin, linco-mycin, mikamycin A, erythromycin, pactamycin,and spectinomycin) to stimulate RNA synthesis inbacteria which were tryptophan-starved by addi-tion of the amino acid analogue 5-methyltrypto-phan, and in bacteria with a temperature-sensitivevalyl tRNA synthetase during incubation of thesebacteria at the nonpermissive temperature, whenthe valyl tRNA was uncharged. All of these anti-biotics stimulated RNA synthesis when aminoacid starvation was effected by addition of 5-methyltryptophan, but none of them was able tostimulate RNA synthesis significantly in the tem-perature-sensitive mutants incubated at the non-permissive temperature. They concluded thatamino acids must be attached to tRNA to stimu-late RNA synthesis, and that each of these in-hibitors is able to stimulate RNA synthesis bysparing amino acids.
These authors point out that, of these anti-biotics, chlortetracycline, lincomycin, and mika-mycin A have been reported to inhibit the bindingof tRNA to ribosomes in vitro. If, in vivo, theseantibiotics are effective in promoting the accumu-lation of charged tRNA, and also in the preven-tion of binding of the tRNA to ribosomes, itwould then be necessary to conclude that chargedtRNA can relieve the stringent control of RNAsynthesis prior to the attachment to ribosomes.However, no evidence exists on the degree towhich these antibiotics are effective in vivo inpreventing the binding of tRNA to ribosomes, sothat these experiments do not rule out the possi-bility that the regulation of RNA synthesis ismediated by ribosomes.
COORDINATE REGULATION OF RNA SYNTHESIS
So far we have discussed regulation of RNAsynthesis as if all classes of RNA molecules re-
spond to amino acid starvation in a coordinatemanner. By coordinate regulation, we mean thatthe synthesis of mRNA, rRNA, and tRNA inRCstr strains is inhibited to the same extent byremoval of amino acids. This assumption wasimplicit in the models originally proposed (40,48, 94), in which it was tacitly assumed that asingle RNA polymerase was responsible for thesynthesis of all bacterial RNA and that unchargedtRNA molecules inhibited this polymerase activ-ity, thereby reducing RNA synthesis. Only asingle E. coli RNA polymerase activity has beendemonstrated in vitro, but it is still possible thatmore than one species of RNA polymerase func-tions in vivo.There is general agreement that during amino
acid starvation the synthesis of rRNA and tRNAis strongly inhibited in RCstr strains and continuesin RCrel strains (17, 70). The rRNA and tRNAformed in the absence of protein synthesis inRCrel strains appears to be normal, except thatRNA synthesized during methionine starvationhas been shown to be submethylated (56, 57). The16S and 23S rRNA which accumulates in RCrelbacteria during amino acid starvation can bematured into ribosomes when protein synthesis isrestored, and the tRNA synthesized under theseconditions is almost fully functional in its aminoacid acceptor capacity (75).The evidence in regard to the regulation of
unstable or mRNA synthesis during amino acidstarvation is considerably less clear, and, in somerespects, contradictory. Initial experimentsseemed to support the idea that mRNA produc-tion was generally depressed during amino acidstarvation of RCstr strains to the same extent asthat of rRNA and tRNA. Gros et al. (40) char-acterized the synthesis of mRNA by the loss ofradioactivity from pulse-labeled RNA in thepresence of dinitrophenol (DNP), an inhibitorof bacterial RNA synthesis. They reported amarked reduction of "DNP labile" material inamino acid-starved RCstr strains, as comparedwith the accumulation of "DNP labile" materialin amino acid-starved RCrel strains. Moreover,the RNA extracted from amino acid-starvedRCrel strains possessed a high activity for stimu-lating protein synthesis in vitro, which was notthe case for RNA extracted from amino acid-starved RCstr strains. From a variety of experi-ments, it was concluded (40) that a single RNApolymerase was responsible for the synthesis ofRNA and that synthesis of all classes was in-hibited to the same extent in RCstr strains duringamino acid starvation, that is, that the regulationof RNA synthesis was coordinate. The generaldifficulty with experiments of this nature is thatthe amount of mRNA is inferred from the degree
of functional activity, and that this activity isassumed to be a function of the actual level ofmRNA.The question of coordinate RNA regulation
was re-examined by Friesen (34). In these exper-iments, mRNA was assayed directly by RNA-DNA hybridization. Radioactive RNA was ex-tracted from cultures of RCstr and RCrel bac-teria which had been pulse labeled either duringexponential growth or during amino acid starva-tion. The results indicated that, under the twoconditions, the fraction of the labeled RNA whichcould be characterized as mRNA by hybrid for-mation with DNA was essentially the same; i.e.,20 to 25% of the labeled RNA was hybridizable.Since the fraction of the labeled RNA that wascharacterized as mRNA seemed to be constantunder all conditions, these results also supportthe hypothesis that the regulation of RNA syn-thesis is coordinate.However, the interpretation of these hybridiza-
tion experiments is not without difficulties. Thereare several factors that could result in a significantunderestimate of the amount of mRNA synthesisduring amino acid starvation as compared to thesynthesis of stable RNA during starvation. Ahalf-life of mRNA that is short compared withthe length of the radioactive pulse-label couldlead to an underestimate of the rate of synthesisof mRNA during amino acid starvation, asmeasured by the hybridization procedure de-scribed above. The marked reduction in the up-take of radioactivity (21) by amino acid starvedRCstr bacteria could also lead to an under-estimate. Moreover, abundant species of mRNAmay not be fully hybridized, since the amount ofDNA used in the hybridization reaction is keptsmall so as to avoid significant hybridization ofrRNA and tRNA, which are present in largeexcess over mnRNA. As the following experimentsindicate, there is reason to believe that in factdirect hybridization experiments of pulse-labeledRNA do not reflect the true rate of mRNA syn-thesis in amino acid starved RCstr bacteria.
In an attempt to circumvent these difficulties,several investigators have measured the produc-tion of a single mRNA species, namely the mRNAproduced by the tryptophan biosynthetic operonof E. coli. This system has the advantage that it ispossible to hybridize pulse-labeled RNA to de-rivatives of phage k80 (480 pt) in which the bac-terial tryptophan operon has been incorporatedinto the phage genome, so that tryptophanmRNA can be specifically assayed (45). By use oftryptophan auxotrophs of RCstr and RCrelstrains, synthesis of tryptophan operon mRNAwas induced by tryptophan starvation, and thismRNA was assayed by hybridization to 480pt
phage DNA (22). These studies showed that ontryptophan starvation, tryptophan operon mRNAconstitutes a much larger fraction of the labeledRNA in the RCstr strain than in the RCrel strain.These results were interpreted as indicating a non-coordinate regulation of RNA synthesis in RCBtrstrains during tryptophan-starvation, that is, apreferential reduction in the synthesis of rRNAand tRNA. On the basis of similar experiments,the same conclusions have been drawn by Lavelle(51).Stubbs and Hall (92) studied the synthesis of
-tryptophan operon mRNA under somewhat dif-ferent conditions. Instead of inducing formationof tryptophan operon mRNA by tryptophanstarvation, they constructed a pair of RCstr andRCrel arginine auxotrophs that produce trypto-phan operon mRNA constitutively, and investi-gated the effect of arginine starvation on its pro-duction. To analyze the amounts of tryptophanoperon mRNA produced, they used the techniqueof competitive hybridization, which permits abetter quantitation than that obtained through theuse of direct hybridization. Their results showedthat arginine-starvation sharply reduced the in-tracellular level of tryptophan operon mRNA inboth the RCBtr and the RCrel strain as comparedto the amounts present in exponentially growingcultures. The explanation that Stubbs and Halladvance for their findings is that there may besome general regulation of the operons of aminoacid biosynthesis which overrides the regulatoryrole of the rel gene, and that these results do notbear directly on the question of coordinancy. Thisresult also raises the possibility that the nonco-ordinancy observed by Edlin et al. and by Lavalle(see above) for the tryptophan operon might bepeculiar to that operon.A number of other experiments seem to support
the idea of noncoordinate RNA regulation. Fromexperiments described in an earlier section, Mor-ris and deMoss (64) concluded that mRNA syn-thesis continues in RCstr strains during aminoacid starvation. Morris and Kjeldgaard (65)demonstrated induction of 3-galactosidasemRNA in amino acid-starved RCstr bacteria;this also is consistent with the idea of noncoordi-nate regulation.Forchhammer and Kjeldgaard (in preparation)
measured mRNA production in amino acid-starved bacteria by determining the activity ofRNA extracts in stimulating protein synthesis inan in vitro system. They first demonstrated thatRNA extracted from uracil-starved bacteria hasonly a few per cent of the stimulating activity ofthe RNA from an unstarved culture. After a pre-liminary uracil starvation to reduce the mRNAcontent of the bacteria, they restored uracil, but
deprived a portion of the culture of amino acids.The in vitro stimulating activity of the RNAextracted from the amino acid-starved culture in-creased in an identical fashion to the nonstarvedculture during the initial phase of recoveryand eventually leveled off at half the valueattained by the nonstarved control culture. Fromthese results, they concluded that mRNA syn-thesis continues in the absence of amino acids,although it has not been shown which classes ofRNA contribute to this stimulating activity.In an extension of earlier studies of Stern et al.(95), Sarkar and Moldave (88) have examined theRNA synthesized in amino acid-starved culturesof an RCstr strain. By analysis of this RNA withrespect to sedimentation velocity in a sucrosegradient, the distribution of polysomes, the decayin the presence of actinomycin, and its base com-position, they conclude that the RNA synthesizedduring amino acid starvation is predominantlymRNA. They concluded, therefore, that the regu-lation of mRNA synthesis is noncoordinate.
Nierlich (in preparation) pulse-labeled RNAof bacteria growing in an amino acid-supple-mented medium and in bacteria which wereamino acid starved. Previous difficulties withthis type of measurement arose from the fail-ure of amino acid-starved bacteria to take upand phosphorylate precursors from the mediumwith the same kinetics as nonstarved bacteria (21).Nierlich circumvented this difficulty by measuringthe specific activity of the radioactive ribonucleo-side triphosphate pool at intervals of a few sec-onds during the period following the addition ofradioactive material. Given the specific activity ofthe precursor pool, it is possible to convert theradioactivity incorporated into RNA into actualquantities of RNA; the rate of RNA synthesiscan then be determined. The finding was that therate of RNA synthesis in the amino acid-starvedculture is about half of the rate in the supple-mented culture, although accumulation of RNAin the starved culture is only a few per cent of thatin the control culture. Similar results were ob-tained when RNA synthesis was stopped by shift-ing bacteria with a temperature-sensitive valine-activating enzyme from low to high temperature.At the high temperature, the rate of RNA syn-thesis was again about one-half of the rate in thecontrol culture, although accumulation of RNAhad stopped. These observations are consistentwith the idea of noncoordinate regulation ofRNA, but it should be noted that there is nocharacterization of the classes of RNA which arebeing synthesized in these experiments.The conclusion that mRNA synthesis can con-
tinue during amino acid starvation was also ar-rived at from measurements of flagellar protein
formation in B. subtilis and S. typhimurium dur-ing tryptophan starvation (20). Flagellar proteincontains no tryptophan residues, so that it ispossible to study its formation in cells that aretryptophan starved. As in RCstr strains of E. coli,tryptophan starvation reduced RNA synthesis by80 to 90%; however, tryptophan starvation per-mits appreciable flagellar protein formation. Thissuggests that mRNA synthesis for these proteinsis relatively unaffected, although certain reserva-tions concerning this conclusion are in order,since evidence has been presented (55, 61) thatthe mRNA which codes for the synthesis offlagellar protein is stable.
Finally, Goodman and Manor (in preparation)have measured the relative amounts of ribosomaland soluble protein synthesized in RCstr andRCrel strains during amino acid starvation.These measurements are made as follows: aradioactive amino acid is added to RCstr orRCrel bacteria that have been amino acid starvedfor about 15 min. The bacteria are exposed toradioactivity for 30 min (protein synthesis duringstarvation is only a few per cent of the exponen-tial rate), followed by a 15-min chase with excessunlabeled amino acid. The bacteria are then cen-trifuged, washed, and suspended in fresh growthmedium. After one generation of growth, they arecollected and analyzed for radioactivity present inribosomal protein and total protein, as describedby Schleif (89). These experiments indicate that,during amino acid starvation, RCrel strains havea value for a (rate of ribosomal protein synthesis/rate of total protein synthesis) which is more thantwice the a value which is measured during ex-ponential growth. In contrast, the a value duringamino acid starvation of an RCStr strain is lessthan half the value measured during exponentialgrowth and less than one-fourth of the a valuemeasured for an amino acid-starved RCrel strain.These experiments suggest that the synthesis ofthe mRNA that codes for ribosomal protein isreduced to a greater degree during amino acidstarvation of an RCstr strain than of an RCrelstrain. If the synthesis of ribosomal proteinmRNA is regulated in the same manner as thesynthesis of of rRNA, these experiments provideadditional evidence for the noncoordinate regu-lation of RNA synthesis by the rel gene.What can be inferred from all these experiments
concerning the regulation of mRNA synthesis inRCstr and RCrel strains? These experiments seemto support the idea that amino acid regulation ofRNA synthesis by the rel gene is not coordinate,but pertains to the regulation of stable RNA(rRNA and tRNA). This does not imply thatmRNA synthesis in RCstr strains is not affectedby amino acid starvation. We believe that mRNA
synthesis is reduced in both RCStr and RCrelstrains by amino acid starvation, but that thisresults from the specific repression of manyoperons and not because of the state of the relallele. Stopping protein synthesis causes manybiosynthetic intermediates to accumulate; someof these certainly act to repress expression ofnumerous operons. Therefore. some reduction inthe overall rate of mRNA is to be expected, andthe experiments cited seem consistent with thisview. Ideally, the problem could be resolved bymeasuring mRNA synthesis for a number ofspecific operons. This has been done for thetryptophan operon, but for technical reasons, atpresent, there are not many operons which can betested in this manner.One can advance a teleological argument to
suggest why noncoordinate regulation couldconfer a selective advantage. When a bacterium issuddenly deprived of an essential nutrient, suchas a required amino acid, it is advantageous forit to overcome the deficiency as rapidly as pos-sible. The bacterium with a uniformly reduced ca-pacity for RNA synthesis would be unable to re-cover as effectively as a bacterium that maintainsthe capacity for normal mRNA production. Thisis the situation which obtains, for example, duringtryptophan starvation, when the tryptophan bio-synthetic enzymes are produced at a high rate.
DIscussIoNBefore discussing the theories that have been
advanced to explain the nature of the rel genefunction, we wish to re-emphasize two conclusionswhich seem adequately supported by experimentand which must be accommodated by any theory.The first is that the rel function pertains only tothe regulation of RNA synthesis by amino acids;the manner in which RNA synthesis responds tochanges in growth rate resulting from changes incarbon or nitrogen sources is the same in RCstrand RCrel strains. The second conclusion is thatonly the synthesis of rRNA and of tRNA, andnot that of mRNA, is regulated by the rel gene.It is therefore concluded that the function of therel gene is to adjust the rate of rRNA and tRNAsynthesis to the rate at which amino acids enterproteins. This regulation serves to adjust the levelof the protein-synthesizing machinery of the cellto a level commensurate with the availability ofamino acids for protein synthesis.The theories that have been proposed to explain
the amino acid regulation of RNA synthesis canbe summarized as follows. (i) RNA synthesis isregulated by the level of charged or unchargedtRNA (48, 63, 94). (ii) RNA synthesis is regu-lated by polyamines (13, 14,76,77). (iii) RNA syn-thesis is regulated by the availability of nucleotide
substrates (12, 37). (iv) RNA synthesis is regu-lated through an obligatory coupling betweenprotein synthesis and RNA synthesis (92, 93). (v)RNA synthesis is regulated by the supply of freeribosomes or ribosomal subunits, or by some stepin protein synthesis subsequent to the attachmentof amino acids to tRNA (62, 64, 84, 90).There is as yet no compelling evidence that sup-
ports one of the above theories to the exclusionof the others; however, at least the first of thesecan be eliminated with reasonable certainty. Theoriginal proposal that uncharged tRNA but notcharged tRNA is an effective inhibitor of theRNA polymerase enzyme is clearly not the ex-planation of amino acid regulation of RNA syn-thesis.
This assertion is based primarily on the resultsof experiments cited in "Role of Transfer RNAin the Regulation of RNA Synthesis," above.The degree of inhibition by tRNA ofRNA polym-erase activity in vitro does not seem to dependsignificantly on whether the tRNA is charged oruncharged. Moreover, the inhibition that is ob-tained does not seem great enough to account forthe marked inhibition ofRNA synthesis observedin vivo. Also, the amount of tRNA per cell, aswell as the extent to which it is charged or un-charged, can be varied over a considerable rangein vivo without affecting the normal pattern ofRNA regulation by amino acids.
In "Coordinate Regulation ofRNA Synthesis,"above, it was concluded that the function of therel gene pertains only to the regulation of the syn-thesis of rRNA and of tRNA. If this conclusionis correct, the hypothesis that the state of tRNAregulates the synthesis of RNA would have to bemodified to explain how, on amino acid starva-tion, mRNA is synthesized while synthesis ofrRNA and tRNA is inhibited. Two possiblitiesare that there is more than one species of RNApolymerase or that, during amino acid starvation,there is a selective inhibition of the RNA polym-erase such that mRNA can be transcribed whilerRNA and tRNA cannot be transcribed.An important argument against the tRNA
theory derives from the observation (90) thattrimethoprim inhibition of protein synthesis re-sults in the cessation of RNA synthesis in RCstrstrains but not in RCrel strains. Trimethopriminterferes with the formylation of the species ofmethionyl-tRNA necessary for polypeptide chaininitiation; this formylation occurs after the attach-ment of methionine to methionyl-tRNA. There-fore, it is inferred that trimethoprim blocks pro-tein synthesis and RNA synthesis in RCstr strainswithout reducing the levels of amino acid-chargedtRNA. This would suggest that an RCstr straincan undergo the stringent response even in a
situation in which all species of tRNA are fullycharged.The sum of these experiments seems to rule out
amino acid regulation of RNA synthesis bycharged or uncharged tRNA, at least by its inter-action with the RNA polymerase.The theory that RNA synthesis is regulated by
polyamines is not supported by convincing data.The stimulation of RNA synthesis in amino acid-starved RCstr strains by the addition of spermi-dine to the medium has been shown to result froma sparing of amino acids rather than from a morespecific and direct effect on RNA synthesis (28).Measurements of polyamine levels in RCstr andRCrel strains were initially made in bacterialstrains with metabolic interrelationships whichmake it difficult to correlate differences in poly-amine content with the state of the rel gene. Atpresent, it seems likely that any difference inpolyamine synthesis in RCstr and RCrel strainsis a consequence of the difference in RNA regula-tion rather than its cause.The analysis of precursor regulation of RNA
synthesis is complicated by the possibility ofnucleotide compartmentalization, but, even withthis proviso, it seems unlikely that RNA synthesisis regulated by the supply of available substrates.In general, the changes that can be observed inthe nucleoside triphosphate pools on amino acidstarvation are not of great magnitude (see Table2) in either RCstr or RCrel strains. The one pos-sible exception to this rule is the pool of GTP,which falls quite abruptly under certain conditionsof amino acid starvation of an RCStr strain. Ifthis fall in the level of GTP is the cause of thestringent response to amino acid starvation, it isunlikely that it is because of its requirement as asubstrate for RNA synthesis. An alternative isthat it might be involved indirectly through itsrequirement in protein synthesis; for example, ithas been reported that GTP is required to effectbinding of tRNA to ribosomes (53). It is difficult,however, to suggest any satisfying mechanismwhereby the synthesis ofGTP could be coupled toamino acid deprivation. Moreover, this theoryin its simple form does not explain how a fall inavailability of substrate could limit RNA syn-thesis in a noncoordinate manner.The theory of coupled synthesis ofRNA and of
protein has been thoroughly discussed by G. S.Stent in two recent articles (92, 93). It should bepointed out here that noncoordinate regulationof RNA synthesis would imply that it was thesynthesis of stable RNA that was coupled toprotein synthesis.The theories proposing that the regulation of
RNA synthesis is effected by free ribosomes arealso, in effect, suggesting that the synthesis of
stable RNA is coupled to protein synthesis. How-ever, here the emphasis is on the role of free ribo-somes as inhibitors of stable RNA synthesis. Onthis theory, stable RNA is synthesized only whenthe stable RNA already present in the cell is in-volved in protein synthesis.The initial report that polysomes disappeared
during amino acid starvation of RCstr strains butnot of RCrel strains lent credence to the idea ofregulation by free ribosomes (64). However, themore recent experiments of Friesen (35), whichshow that this result varies with the bacterialstrain used and the particular amino acid with-held, throw some doubt on the significance ofpolysome disappearance during amino acid star-vation of RCstr strains.The inhibition of protein and RNA synthesis by
trimethoprim in RCstr strains argues that theregulation occurs at some step after the attach-ment of amino acids to tRNA. Another observa-tion, consistent with this idea, is that starvationfor any of several different amino acids results inessentially the same reduction in the rate ofRNAsynthesis. This suggests that regulation is noteffected by an array of specific inhibitors corre-sponding to the different amino acids, but by amore general mechanism which becomes activatedby deprivation for any amino acid. The obvioussite of action for such a regulatory mechanism isthe polysome.A possible basis for the difference between
RCstr and RCrel strains, consistent with thetheory of regulation by the concentration of freeribosomes, would be that in RCrel strains theoperon for rRNA synthesis is insensitive to theaccumulation of free ribosomes, in contrast toRCStr strains, which are sensitive. Such a modelappears to be ruled out by the important observa-tion that the rel+ allele is dominant over the relallele. An alternative explanation is that, in aminoacid-starved RCrel strains, a sufficiently highproportion of the cell's ribosomes are in poly-somes to allow rRNA synthesis to continue,whereas polysomes break down on amino acidstarvation in RCstr strains. Since it has been re-ported (35) that polysomes do not necessarilybreak down on amino acid starvation of RCstrstrains, the validity of this model is questionable.The entire model of ribosomal involvement inthe regulation of stable RNA synthesis will begreatly clarified when it has been determined howrapidly the dominance of the rel+ allele is ex-pressed in partially diploid strains. If the RCstrphenotype is established very rapidly after theformation of the rel+/rel diploid, it would arguethat the rel gene product is some type of repressorof ribosome synthesis. If it is established that theRCstr phenotype is established more slowly, it
could be interpreted to mean that the rel geneproduct is a structural component of individualribosomes. It is conceivable that one "'RCOtre'ribosome attached to a polysome could be suf-ficient to allow breakdown of that polysome onamino acid starvation, thereby giving rise to thestringent response.
In summary, none of the models proposed todate is entirely satisfactory in explaining theamino acid regulation ofRNA synthesis. Nothingis known concerning the specific biochemicalalteration that permits RNA synthesis to continuein RCrel strains. Studies are now in progress todetermine if the rel allele causes altered GTP syn-thesis, phosphorylation, or ribosomal proteinsynthesis.
ACKNOWLEDGMENTS
This investigation was supported by Public HealthService research grant CA02129 and training grantCA05028 from the National Cancer Institute.We are grateful to G. S. Stent for numerous fruit-
ful discussions and to him and to F. C. Neidhardt fora critical reading of the manuscript. We are also grate-ful to F. C. Neidhardt for supplying strain NP29, toS. Kaplan for strain 10B6, and to R. MacDonald forproviding the photograph (Fig. 2). We are also in-debted to our many stringent and relaxed friends whosupplied information and data prior to publication.
ADDENDUM IN PROOF
This review was to have appeared in the Juneissue, but it was delayed because of objections raisedby S. S. Cohen to the discussion of his work presentedhere. Since the authors decided not to make anychanges of substance in response to these objections,readers are referred to a summary (14) of Dr. Cohen'sown views regarding the regulation of RNA synthesis.
LITERATURE CITED
1. Alfoldi, L., and E. Kerekes. 1964. Neutralizationof the amino acid sensitivity of RCreI Escherichiacoli. Biochim. Biophys. Acta 91:155-157.
2. Alfoldi, L., G. S. Stent, and R. C. Clowes. 1962.The chromosomal site of the RNA control(RC) locus in E. coli. J. Mol. Biol. 5:348-355.
3. Alfoldi, L., G. S. Stent, M. Hoogs, and R. Hill.1963. Physiological effects of the RNA control(RC) gene in E. coli. Z. Vererbungslehre 94:285-302.
4. Aronson, A. I., and S. Spiegelman. 1961. Proteinand ribonucleic acid synthesis in a chloram-phenicol-inhibited system. Biochim. Biophys.Acta 53:70-84.
5. Aronson, A. I., and S. Spiegelman. 1961. On thenature of the ribonucleic acid synthesized in thepresence of chloramphenicol. Biochim. Bio-phys. Acta 53:84-95.
6. Bock, A., L. E. Faiman, and F. C. Neidhardt.1966. Biochemical and genetic characterizationof a mutant of Escherichia coli with a tempera-
7. Borek, E.. J. Rockenbach, and A. Ryan. 1956.Studies on a mutant of Escherichia coli with un-balanced ribonucleic acid synthesis. J. Bac-teriol. 71:318-323.
8. Borek, E., A. Ryan, and J. Rockenbach. 1955.Nucleic acid metabolism in relation to thelysogenic phenomenon. J. Bacteriol. 69:460-467.
9. Bremer, H., C. Yegian, and M. Konrad. 1966. In-activation of purified E. coli RNA polymeraseby transfer RNA. J. Mol. Biol. 16:94-103.
10. Britten, R. J., and B. J. McCarthy. 1962. The syn-thesis of ribosomes in E. coli. II. Analysis ofthe kinetics of tracer incorporation in growingcells. Biophys. J. 2:49-55.
11. Burchall, J. J., and G. H. Hitchings. 1966. In-hibitor binding analysis of dihydrofolate re-ductases from various species. J. Mol. Pharma-col. 1:126-136.
12. Cashel, M., and J. Gallant. 1968. The control ofRNA synthesis in Escherichia coli. I. Aminoacid dependence of the synthesis of the sub-strates of RNA polymerase. J. Mol. Biol. 34:317-330.
13. COHEN, S. S., N. Hoffner, M. Jansen, M. Moore,and A. Raina. 1967. Polyamines, RNA syn-thesis, and streptomycin lethality in a relaxedmutant of E. coli strain 15TAU. Proc. NatlAcad. Sci. U.S. 57:721-728.
14. Cohen, S. S., and A. Raina. 1967. Some interrela-tions of natural polyamines and nucleic acids ingrowing and virus-infected bacteria, p. 157-182.In H. J. Vogel, J. 0. Lampen, and V. Bryson(ed.), Organizational biosynthesis. AcademicPress, Inc., New York.
15. Croom, J. A., J. J. McNeill, and S. B. Tove. 1964.Biotin deficiency and the fatty acids of certainbiotin-requiring bacteria. J. Bacteriol. 88:389-394.
16. Dagley, S., C. Turnock, and D. G. Wild. 1963.The accumulation of ribonucleic acid by a mu-tant of Escherichia coli. Biochem. J. 88:555-566.
17. Dagley, S., A. White, D. G. Wild, and J. Sykes.1962. Synthesis of protein and ribosomes bybacteria. Nature 194:25-27.
18. Demain, A. L., and J. Birnbaum. 1968. Alterationof permeability for the release of metabolitesfrom the microbial cell. In Current topics inmicrobiology and immunology. Springer-Verlag, New York.
19. Demerec, M., E. A. Adelberg, A. J. Clark, and P.E. Hartman. 1966. A proposal for a uniformnomenclature in bacterial genetics. Genetics54:61-76.
20. Dimmitt, K., S. Bradford, and M. Simon. 1968.Synthesis of bacterial flagella. I. Requirementfor protein and ribonucleic acid synthesis duringflagella regeneration in Bacillus subtilis. J. Bac-teriol. 95:801-810.
21. Edlin, G., and J. Neuhard. 1967. Regulation of
D BRODA BACrERIOL. REV.
nucleoside triphosphate pools in Escherichiacoli. J. Mol. Biol. 24:225-230.
22. Edlin, G., G. S. Stent, R. F. Baker, and C.Yanofsky. 1968. Synthesis of a specific messen-ger RNA during amino acid starvation of E.coli. J. Mol. Biol., in press.
23. Edlin, G. 1968. Operon polarity in relaxed RNAcontrol strains of E. coli. J. Mol. Biol. 31:135-137.
24. Eidlic, L., and F. C. Neidhardt. 1965. Role ofvalyl-sRNA synthetase in enzyme repression.Proc. Natl. Acad. Sci. U.S. 53:539-543.
25. Eidlic, L., and F. C. Neidhardt. 1965. Protein andnucleic acid synthesis in two mutants of Esch-erichia coli with temperature-sensitive amino-acyl ribonucleic acid synthetases. J. Bacteriol.89:706-711.
26. Eisenstadt, J., and P. Lengyel. 1966. Formyl-methionyl-tRNA dependence of amino acidincorporation in extracts of trimethoprim-treated Escherichia coli. Science 154:524-526.
27. Ezekiel, D. H. 1965. Requirement for p-fluoro-phenylalanine activation in control of ribo-nucleic acid synthesis. Biochim. Biophys. Acta95:48-54.
28. Ezekiel, D. H., and H. Brockman. 1968. Effect ofspermidine treatment on amino acid availabilityin amino acid-starved Escherichia coli. J. Mol.Biol. 31:541-552.
29. Ezekiel, D. H., and B. Valulis. 1966. Control ofribonucleic acid synthesis in Escherichia colicells with altered transfer ribonucleic acid con-centration. Biochim. Biophys. Acta 129:123-139.
30. Fangman, W. L., and F. C. Neidhardt. 1964.Demonstration of an altered aminoacyl ribo-nucleic acid synthetase in a mutant of Escher-ichia coli. J. Biol. Chem. 239:1839-1843.
31. Fangman, W. L., and F. C. Neidhardt. 1964. Pro-tein and ribonucleic acid synthesis in a mutantof Escherichia coli with an altered aminoacylribonucleic acid synthetase. J. Biol. Chem. 239:1844-1847.
32. Fiil, N., and J. D. Freisen. 1968. Isolation of "re-laxed" mutants of Escherichia coli. J. Bacteriol.95:729-731.
33. Fraenkel, D. G., and F. C. Neidhardt. 1961. Useof chloramphenicol to study control of RNAsynthesis in bacteria. Biochim. Biophys. Acta53:96-110.
34. Friesen, J. D. 1965. Control of bacteriophageRNA synthesis in Escherichia coli. J. Mol. Biol.13:220-233.
35. Friesen, J. D. 1968. A study of the relationship be-tween polyribosomes and messenger RNA inEscherichia coli. J. Mol. Biol. 32:183-200.
36. Gale, E. F., and J. P. Folkes. 1953. The assimila-tion of amino acids by bacteria. Biochem. J.53:493-498.
37. Gallant, J., and M. Cashel. 1967. On the mecha-nism of amino acid control of ribonucleic acidbiosynthesis. J. Mol. Biol. 25:545-553.
38. Gavin, J. J., and W. W. Umbreit. 1965. Effect of
biotin on fatty acid distribution in Escherichiacoli. J. Bacteriol. 89:437-443.
39. Goldstein, D. B., B. J. Brown, and A. Goldstein.1960. Amino acid starvation in an Escherichiacoli auxotroph. II. Acid-soluble purine andpyrimidine derivatives in normal and starvedcells. Biochim. Biophys. Acta 43:55-61.
40. Gros, F., J. M. Dubert, A. Tissieres, S. Bourgeois,M. Michelson, R. Soffer, and L. Legault. 1963.Regulation of metabolic breakdown and syn-thesis of messenger RNA in bacteria. ColdSpring Harbor Symp. Quant. Biol. 28:299-313.
41. Gros, F., J. Gallant, R. Weisberg, and M. Cashel.1967. Decryptification of RNA polymerase inwhole cells of Escherichia coli. J. Mol. Biol. 25:555-557.
42. Gros, F., and F. Gros. 1956. Role des aminoacidesdans la synthese des acides nucleiques chezEscherichia coli. Biochim. Biophys. Acta 22:200-201.
43. Gros, F., and F. Gros. 1958. Role des acidesamines dans la synthese des acides nucleiqueschez Escherichia coli. Exptl. Cell Res. 14:104-131.
44. Gross, J. D. 1963. The effect of unbalanced growthon recombinant formation in E. coli. Genet.Res. Cambridge 4:457-462.
45. Hartwell, L. H., and C. S. McLaughlin. 1968. Mu-tants of yeast with temperature-sensitive iso-leucyl-tRNA synthetases. Proc. Natl. Acad.Sci. U.S. 59:422-428.
46. Imamoto, F., N. Morikawa, K. Sato, S. Mishima,and T. Nishimura. 1965. On the transcriptionof the tryptophan operon in Escherichia coli.II. Production of the specific messenger RNA.J. Mol. Biol. 13:157-168.
47. Kanazir, D., H. D. Barner, J. G. Flaks, and S. S.Cohen. 1959. Some physiological and geneticproperties of a strain of Escherichia coli re-quiring thymine, arginine and uracil. Biochim.Biophys. Acta 34:341-353.
48. Kurland, C. G., and 0. Maal0e. 1962. Regulationof ribosomal and transfer RNA synthesis. J.Mol. Biol. 4:193-210.
49. Kurland, C. G., M. Nomura, and J. D. Watson.1962. The physical properties of the chloro-mycetin particles. J. Mol. Biol. 4:388-394.
50. Lavalle, R. 1965. Nouveaux mutants de regula-tion de la synthese de I'ARN. Bull. Soc. Chim.Biol. 47:1567-1570.
51. Lavalhi, R., and G. De Hauwer. 1968. Messengersynthesis during amino acid starvation in E.coli. J. Mol. Biol., in press.
52. Lazzarini, R. A., and E. Santangelo. 1968. Theeffect of chloramphenicol on the synthesis andstability of ribonucleic acid in Bacillus subtilis.J. Bacteriol. 95:1212-1220.
53. Lucas-Lenard, J., and A-L. Haenni. 1968. Re-quirement of guanosine 5'-triphosphate forribosomal binding of aminoacyl s-RNA. Proc.Natl. Acad. Sci. U.S. 59:554-560.
54. Maal0e, O., and N. 0. Kjeldgaard. 1966. Control
of macromolecular synthesis. W. A. Benjamin,Inc., New York.
55. McClatchy, J. K., and H. V. Rickenberg. 1967.Heterogeneity of the stability of messengerribonucleic acid in Salmonella typhimurium. J.Bacteriol. 93:115-121.
56. Mandel, L. R., and E. Borek. 1962. Ribonucleicacid synthesis under relaxed control. Biochem.Biophys. Res. Commun. 9:11-17.
57. Mandel, L. R., and E. Borek. 1963. The natureof the RNA synthesized during conditions ofunbalanced growth in E. coli K12 W-6. Bio-chemistry 2:560-566.
58. Manor, H., and R. Haselkorn. 1967. Properties ofribonucleic acid components in ribonucleopro-tein particle preparations obtained from anEscherichia coli RCrel strain. J. Mol. Biol. 24:269-288.
59. Martin, E. M., C. Yegian, and G. S. Stent. 1963.Specificity of the amino acid transfer reaction inE. coli strains carrying different alleles of theRNA control (RC) gene. Z. Vererbungslehre94:303-315.
60. Martin, R. G. 1968. Polarity in relaxed strains ofSalmonella typhimurium. J. Mol. Biol. 31:127-134.
61. Martinez, R. J. 1966. The formation of bacterialflagella. II. The relative stability of messengerRNA for flagellin biosynthesis. J. Mol. Biol.17:10-17.
62. Matchett, W. H. 1967. Methionine and the regula-tion of ribonucleic acid synthesis in Escherichiacoli. J. Bacteriol. 93:90-97.
63. Morris, D. W., and J. A. DeMoss. 1965. Role ofaminoacyl-transfer ribonucleic acid in the regu-lation of ribonucleic acid synthesis in Escher-ichia coli. J. Bacteriol. 90:1624-1631.
64. Morris, D. W., andJ. A. DeMoss. 1966. Polysometransitions and the regulation of ribonucleicacid synthesis in Escherichia coli. Proc. Natl.Acad. Sci. U.S. 56:262-268.
65. Morris, D. W., and N. 0. Kjeldgaard. 1968. Evi-dence for the non-co-ordinate regulation ofribonucleic acid synthesis in stringent strains ofEscherichia coli. J. Mol. Biol. 31:145-148.
66. Morris, D. R., and A. B. Pardee. 1966. Multiplepathways of putrescine biosynthesis in Escher-ichia coli. J. Biol. Chem. 241:3129-3135.
67. Nakada, D., I. A. C. Anderson, and B.Magasaniky 1964. Fate of the ribosomal RNAproduced b. a "relaxed" mutant of Escherichiacoli. J. Mol Biol. 9:472-488.
68. Neidhardt, F. C. 1963. Properties of a bacterialmutant lacking amino acid control ofRNA syn-thesis. Biochim. Biophys. Acta 68:365-379.
69. Neidhardt, F. C. 1966. Roles of amino acid acti-vating enzymes in cellular physiology. Bac-teriol. Rev. 30:701-719.
70. Neidhardt, F. C., and L. Eidlic. 1963. Charac-terization of the RNA formed under conditionsof relaxed amino acid control in Escherichiacoli. Biochim. Biophys. Acta 68:380-388.
71. Neidhardt, F. C., and F. Gros. 1957. Metabolic
instability of the ribonucleic acid synthesizedby Escherichia coli in the presence of chloro-mycetin. Biochim. Biophys. Acta 25:513-520.
72. Nierlich, D. P., and W. Vielmetter. 1968. Kineticstudies on the relationship of ribonucleotideprecursor pools and ribonucleic acid synthesis.J. Mol. Biol. 32:135-147.
73. Olenick, J. G., and F. E. Hahn. 1966. A study ofthe RNA polymerase reaction catalyzed by en-zymes from relaxed and stringent bacteria.Biochim. Biophys. Acta 129:429-431.
74. Pardee, A. B., and L. S. Prestidge. 1956. The de-pendence of nucleic acid synthesis on the pres-ence of amino acids in Escherichia coli. J. Bac-teriol. 71:677-683.
75. Peterkofsky, A., C. Jesensky, A. Bank, and A. H.Mehler. 1964. Studies on the role of methylatedbases in the biological activity of soluble ribo-nucleic acid. J. Biol. Chem. 239:2918-2926.
76. Raina, A., and S. S. Cohen. 1966. Polyaminesand RNA synthesis in a polyauxotrophic strainof E. coli. Proc. Natl. Acad. Sci. U.S. 55:1587-1593.
77. Raina, A., M. Jansen, and S. S. Cohen. 1967.Polymaines and the accumulation of ribonuc-leic acid in some polyauxotrophic strains ofEscherichia coli. J. Bacteriol. 94:1684-1696.
78. Ramakrishnan, T., and E. A. Adelberg. 1965.Regulatory mechanisms in the biosynthesis ofisoleucine and valine. II. Identification of twooperator genes. J. Bacteriol. 89:654-660.
79. Randerath, K., and E. Randerath. 1964. Ion-ex-change chromatography of nucleotides onpoly(ethyleneimine)-cellulose thin layers. J.Chromatog. 16:111-125.
80. Randerath, K., and E. Randerath. 1964. Anion-exchange thin layer chromatography of com-plex mixtures of nucleotides. J. Chromatog.16:126-129.
81. Richmond, M. H. 1962. The effect of amino acidanalogues on growth and protein synthesis inmicroorganisms. Bacteriol. Rev. 26:398-420.
82. Riley, M., and A. B. Pardee. 1962. Nutritionaleffects on frequencies of bacterial recombina-tion. J. Bacteriol. 83:1332-1335.
83. Ron, E. Z., and B. D. Davis. 1966. Specific stimu-lation of RNA synthesis by methionine inseveral strains of Escherichia coli. J. Mol. Biol.21:13-27.
84. Ron, E. Z., R. E. Kohler, and B. D. Davis. 1966.Increased stability of polysomes in an Escher-ichia coli mutant with relaxed control of RNAsynthesis. Proc. Natl. Acad. Sci. U.S. 56:471-475.
85. Rosset, R., J. Julien, and R. Monier. 1966. Ribo-
IiD BRODA BACTERIOL. REV.
nucleic acid composition of bacteria as a func-tion of growth rate. J. Mol. Biol. 18:308-320.
86. Salser, W., J. Janin, and C. Levinthal. 1968.Measurement of the unstable RNA in ex-ponentially growing cultures of Bacillus sub-tilis and Escherichia coli. J. Mol. Biol. 31:237-266.
87. Sands, M. K., and R. B. Roberts. 1952. The effectsof a tryptophan-histidine deficiency in a mutantof Escherichia coli. J. Bacteriol. 63:505-511.
88. Sarkar, S., and K. Moldave. 1968. Characteriza-tion of the ribonucleic acid synthesized duringamino acid-deprivation of a stringent auxotrophof Escherichia coli. J. Mol. Biol. 33:213-224.
89. Schleif, R. 1967. Control of production of ribo-somal protein. J. Mol. Biol. 27:41-55.
90. Shih, An-Ya, J. Eisenstadt, and P. Lengyel. 1966.On the relation between ribonucleic acid syn-thesis and peptide chain initiation in E. coli.Proc. Natl. Acad. Sci. U.S. 56:1599-1605.
91. Stacey, K. A., and E. Simson. 1965. Improvedmethod for the isolation of thymine-requiringmutants of Escherichia coli. J. Bacteriol. 90:554-555.
92. Stent, G. S. 1964. The operon: on its third anni-versary. Science 144:816-820.
93. Stent, G. S. 1966. Genetic transcription. Proc.Roy. Soc. (London) Ser. B, 164:181-197.
94. Stent, G. S., and S. Brenner. 1961. A genetic locusfor the regulation of ribonucleic acid synthesis.Proc. Natl. Acad. Sci. U.S. 47:2005-2014.
95. Stern, J. L., M. Sekiguchi, H. D. Barner, and S.S. Cohen. 1964. The synthesis of messengerRNA without protein synthesis. J. Mol. Biol.8:629-637.
96. Stubbs, J., and B. D. Hall. 1968. Effects of aminoacid starvation upon constitutive tryptophanm-RNA synthesis. J. Mol. Biol., in press.
97. Sypherd, P. S. 1965. Accumulation of ribonucleo-protein particles in a relaxed mutant of Escher-ichia coli. J. Bacteriol. 90:403-410.
98. Taylor, A. L., and C. D. Trotter. 1967. Revisedlinkage map of Escherichia coli. Bacteriol. Rev.31:332-353.
99. Tissieres, A., S. Bourgeois, and F. Gros. 1963.Inhibition of RNA polymerase by RNA. J.Mol. Biol. 7:100-103.
100. Yaniv, M., F. Jacob, and F. Gros. 1965. Muta-tions thermosensibles des systems activant lavaline chez E. coli. Bull. Soc. Chim. Biol. 47:1609-1656.
101. Yegian, C. D., G. S. Stent, and E. M. Martin.1966. Intracellular condition of Escherichia colitransfer RNA. Proc. Nati. Acad. Sci. U.S. 55:839-846.
