An increased content of protease La, the lon gene product, increases protein degradation and blocks growth in Escherichia coli.

The lon gene product in Escherichia coli is an ATP-dependent protease (La) that plays an important role in the breakdown of abnormal proteins and certain normal polypeptides. Since transcription of the lon gene rises as part of the heat-shock response, we studied the physiological effects of increased levels of protease La. In cells carrying additional copies of the lon gene under the control of the lac or tac promoter, induction of the protease resulted in a rapid cessation of cell growth and in a loss of viability at stationary phase. Similarly, cells carrying a multicopy plasmid encoding the lon gene contained 2-5-fold more protease La and grew much more slowly than did control cells. In such cells, insertion sequences appeared spontaneously in the lon gene on the plasmid and prevented the excess protease production and allowed more rapid growth. The cells with increased content of protease La (due to the lon plasmid or induction of the lon gene) exhibited severalfold higher rates of degradation of abnormal proteins containing amino acid analogs and of incomplete polypeptides containing puromycin. Also, a beta-galactosidase fusion protein with enzymatic activity was relatively stable in control cells but unstable in the cells with high protease La content. In these cells, the overall degradation of normal proteins increased 2-fold, and certain cellular polypeptides appeared particularly sensitive to proteolysis. Thus, rates of protein degradation in vivo are limited in part by the cellular content of the ATP-dependent protease, and increases in transcription of the lon gene enhance proteolysis and can be deleterious to the cell.


"
The lon gene product in Escherichia coli is an ATPdependent protease (La) that plays an important role in the breakdown of abnormal proteins and certain normal polypeptides. Since transcription of the lon gene rises as part of the heat-shock response, we studied the physiological effects of increased levels of protease La. In cells carrying additional copies of the lon gene under the control of the lac or tac promoter, induction of the protease resulted in a rapid cessation of cell growth and in a loss of viability at stationary phase. Similarly, cells carrying a multicopy plasmid encoding the lon gene contained 2-5-fold more protease La and grew much more slowly than did control cells. In such cells, insertion sequences appeared spontaneously in the lon gene on the plasmid and prevented the excess protease production and allowed more rapid growth.
The cells with increased content of protease La (due to the lon plasmid or induction of the Ion gene) exhibited severalfold higher rates of degradation of abnormal proteins containing amino acid analogs and of incomplete polypeptides containing puromycin. Also, a &galactosidase fusion protein with enzymatic activity was relatively stable in control cells but unstable in the cells with high protease La content. In these cells, the overall degradation of normal proteins increased 2-fold, and certain cellular polypeptides appeared particularly sensitive to proteolysis. Thus, rates of protein degradation in vivo are limited in part by the cellular content of the ATP-dependent protease, and increases in transcription of the lon gene enhance proteolysis and can be deleterious to the cell.

~~~~~~~~~~~~~~ ~ ~~ ~~
In the past several years, considerable progress has been made in our knowledge of the mechanisms and regulation of protein degradation in Escherichia coli (1,2). However, it is still unclear what factors determine the rates of degradation of a protein in vivo. One major determinant of a protein's stability is its conformation. For example, polypeptides encoded by nonsense or certain missense alleles (1)(2)(3)(4)(5)(6)(7) have been shown to be rapidly hydrolyzed to amino acids. Similarly, proteins with highly abnormal structures, such as those containing amino acid analogs or incomplete polypeptides containing puromycin (8,9) are also rapidly degraded in E. coli. Rates of intracellular protein degradation also must depend on the cells' content of proteolytic enzymes. Mutants with a reduced capacity to degrade various abnormal proteins (originally called deg) were isolated by Bukhari and Zipser (10) and subsequently were shown to map in the lon locus (11). lon mutants have a decreased rate of degradation of highly abnormal proteins (10-13) as well as certain native E. coli proteins (14, 15). These mutants show a variety of other phenotypic alterations (e.g. mucoidy or sensitivity to DNAdamaging agents) which seem to result from their decreased ability to degrade certain short-lived proteins (14, 15). The breakdown of abnormal proteins is also greatly reduced (16, 17) in htpR strains, in which lon gene expression is reduced (16).
The lon gene was shown by Chung and Goldberg (12) and Charette et al. (18) to code for an ATP-dependent protease previously designated protease La (19). The degradation of abnormal proteins in E. coli requires metabolic energy (9), and the ATP dependence of protease La appears to account, in large part, for this energy requirement. This enzyme is a new type of endoprotease, which hydrolyzes proteins and ATP in a coupled fashion (12,18,20). A variety of biochemical and genetic findings suggest that this protease catalyzes an initial rate-limiting step in the degradative pathway (1,12,13).
Although the consequences of reduced protease La function have been extensively studied, the physiological effects of high levels of this protein have not been defined previously.
A rise in lon transcription does occur as part of the heatshock response (16, 21,22). This response is induced when cells produce large amounts of an abnormal protein, and it has been postulated that this adaptation may enhance the cell's capacity to degrade such aberrant and potentially harmful polypeptides (16, 21,22). This argument makes the important assumption that the content of protease La in normal cells is rate-limiting for protein breakdown.
The present studies were undertaken to clarify whether an increase in the level of this protease enhances the rate of degradation of abnormal or normal polypeptides in vivo. To determine whether an increased content of protease La enhances proteolysis or influences cell viability, we have studied cells carrying a multicopy lon plasmid (23) and cells carrying a plasmid with the lon gene under control of an inducible promoter.
MATERIALS AND METHODS Strains, Bacteriophage, and Plasmid Construction-Strains of E. coli and bacteriophage used in this study are listed in Table I. The derivatives of previously published strains were constructed by Plvir bacteriophage-mediated transduction, by lysogenization with X bacteriophage derivatives (X), by deletion of the transposon TnlO as previously described (26), or by CaCl2-mediated transformation with plasmid DNA (27). The lon-containing plasmid pJMC40 (Fig. 1A) was kindly provided by Dr. A. Markovitz (University of Chicago); it was derived from pBR322 as previously described (23, 28). The tac 4508 base pairs in the amino-terminal lon coding sequence. B, In an attempt to subclone the lon gene away from both flanking chromosomal sequences and the plasmid pHA105, pJMC40 was digested with EcoRI and re-ligated with T4 DNA ligase at low DNA concentrations to promote circularization. The resulting plasmid mixture was transformed into competent tetracycline-sensitive cells (SY327) and plated on rich broth agar plates with 10 pg/ml tetracycline. The resulting plasmids were analyzed by restriction digestion followed by 1% agarose gel electrophoresis, staining with ethidium bromide, and visualization with short-wave ultraviolet light. They were also tested for their ability to complement a chromosomal lon mutant. No lon+ plasmids were generated by this procedure. C, Shows the structure of pSG5 and pSGl1. lon was placed under control of the loc promoter (pSG5) by inserting a promoterless lon restriction fragment into pUCl8 or tac promoter (pSGll), by insertion of this lon fragment into the tac promoter vector pKK223. promoter plasmid, pKK223, was provided by J. Brosius (Columbia University).
ard recombinant DNA techniques. Plasmid and phage DNA were Plasmids and bacteriophage were constructed in vitro using standisolated according to methods previously described (29). Restriction enzymes were purchased from either New England Biolabs or Bethesda Research Laboratories.
Enzyme Assays-&Galactosidase produced from the lon-lncZ protein fusion was assayed according to the method of Miller (30). Cells grown in rich media were centrifuged at 8OOO X g for 1 min and resuspended in minimal media prior to assay. At each point, 0.1 ml of the culture was assayed in 0.9 ml of Z buffer. mids were grown at 37 "C. After sonication, equal amounts of cell To measure protease La, cells with control and experimental plasprotein were applied to phosphocellulose columns at 4 'C, and the bound material was eluted with a salt step. The enzyme was dialyzed against assay buffer (10 mM Tris, pH 7.5,O.l mM NaEDTA, 1 mM 8mercaptoethanol, 20% glycerol). Protease La was assayed with the specific fluorometric substrate, glutaryl-alanyl-alanyl-phenylalaninemethoxynaphthylamine (Enzyme Systems Products, Livermore, CA) in the presence or absence of ATP (1 mM), as described by Waxman and Goldberg (31). Degradation of Abnormal Proteins-Amino-terminal polypeptide fragments were generated by exposure of cells growing on minimal medium to the antibiotic puromycin (100-400 pg/ml) for 20 or 30 min. During the final 5 min, the cells were exposed to [%]methionine. Prior to assay of protein degradation, the cells were washed free of the antibiotic by filtration and were resuspended in media lacking puromycin (9) and containing large amounts of nonradioactive methionine to prevent reincorporation of [36S]methionine released from protein.
To cause the incorporation of the arginine analog canavanine into cell proteins, an arginine auxotroph was exposed to media lacking arginine but containing canavanine (100 pg/ml) and [36S]rnethionine leucine auxotroph was e%osed to azoleucine (100 pg/ml) for 15 min (5 pCi/ml) for 5 min. To produce azoleucine-containing proteins, a prior to the addition of [ Slmethionine (5 pCi/ml) for 5 min. Prior to assay of protein degradation, the cells were filtered free of analogcontaining media and resuspended in media supplemented with the tive methionine. The rates of protein degradation were estimated by normal amino acid (arginine or leucine) plus an excess of nonradioacmeasuring the release of radioactive amino acids from previously labeled proteins (9).
Generation of Promoter Deletions-A series of deletions 5' to the lon structural gene was generated by Ba131 exonuclease treatment of linear DNA fragments. The extent of each deletion was verified by DNA sequencing, using the dideoxy method (32). To confirm the removal of the lon promoter, we assayed production of 8-galactosidase from a lon-locZ operon fusion carried on a pBR322-derived plasmid. The plasmids generated by this technique were used to construct a promoterless lon restriction fragment and plasmids carrying lon under control of an inducible promoter.

Cells Carrying pJMC40 Have Increased Levels of Protease
La-The plasmid pJMC40 containing the lon gene and the derivative plasmid pJMC40IS described below contain approximately 7 kilobase pairs of E. coli  ATP-dependent protease La. Extracts of cells carrying this multicopy plasmid were reported to have 2-10-fold more protease La than extracts of cells without this plasmid (12). T o confirm the elevation in the levels of this ATP-dependent protease in cells carrying pJMC40, crude extracts of these cells were subjected to phosphocellulose chromatography, and the bound fraction was eluted and assayed with the fluorometric substrate glutaryl-Ala-Ala-Phe-methoxynaphthylamine (31). Extracts of cultures with an active episomal lon gene exhibit 2-4-fold higher rates of hydrolysis of the fluorometric substrate than those of control cells carrying either no plasmid or a plasmid with an inactive lon allele (Table 11). In addition, polyacrylamide gel electrophoresis of either crude extracts or phosphocellulose eluates in the presence of sodium dodecyl sulfate revealed 3-4-fold higher amounts of the 87-kDa polypeptide', corresponding to the protease subunit (Fig. 2).
Inhibition of Growth by pJMC40-E. coli transformed with the multicopy plasmid pJMC40 containing the lon gene ( Fig.  2) were found to grow slowly and were difficult to maintain in a viable condition either in liquid or on solid growth media. Such cells frequently failed to form colonies following entrance into stationary phase or after storage on LB agar plates a t 4 "C for only a few days. Occasionally, the plasmid-carrying cells gave rise to colonies which grew faster and showed greater viability than the original pJMC40 transformants. Such colonies, however, remained resistant to tetracycline, and therefore still carried a pBR322-derived plasmid.
In an attempt to understand the basis for the faster growth and enhanced viability of these colonies, plasmids were isolated from these variants and from the original pJMC40 transformants. These plasmids were analyzed by restriction enzyme cleavage and gel electrophoresis using a variety of enzymes that recognize sequences within or flanking the lon gene (EcoRI, PstI, BumHI, and SulI). In 10 such cases analyzed, the rapidly growing variants carried an episome in which a restriction fragment from the lon gene contained an additional 800 base pairs. Further analysis positioned this inserted sequence within the EcoRI to PstI restriction fragment which contains the promoter and amino-terminal coding sequence for the lon gene (Fig. 1A). When these plasmids, designated pJMC40IS, were transformed into lon mutants, they failed to give a wild-type phenotype, unlike the original plasmid pJMC40 with an intact lon gene. When transformed into wild-type cells, the plasmid bearing an insert allowed approximately twice as much rapid growth as did the original pJMC40 plasmid (data not shown).
In subsequent studies on the effects of increased expression of the lon gene, we used the pJMC40 plasmid and as a control

TABLE I1
Protease La activity measured in phosphocellulose elwtes Cells carrying pJMC40 and pJMC40IS (N5116) were grown in Luria broth at 37 "C to an ODeoo of 0.5 and centrifuged at 400 X g for 10 min. Cell pellets were resuspended in 50 mM Tris, 10 mM MgSO., pH 8.5, at 0 "C and sonicated 4 X 30 s at 0 "C and centrifuged at 100,ooO X g for 90 min at 4 "C. Cells were dialyzed against assay buffer as described under "Materials and Methods," and fractionated on phosphocellulose as described by Waxman and Goldberg (31). Protease La activity was measured using cleavage of the specific fluorometric substrate glutaryl-Ala-Ala-Phe-methoxynaphthylamine.

Plasmid present
Relative cleavage of substrate  used the plasmid containing the insertion element within the lon gene (pJMC40IS). This approach controls for the presence of the several kilobase pairs of chromosomal DNA in addition to the lon gene on the plasmid pJMC40. When plasmids were constructed carrying either a 5' or 3' truncated end of the lon-coding sequence and were transformed into wild-type cells, we observed no effect on cell growth or viability (data not shown). Therefore, the cause of the poor growth and viability of cells carrying pJMC40 is probably the increase in amounts of the lon gene product, rather than of any controlling DNA sequences present on the lon plasmid.
Attempts to Subclone lon-In an attempt to construct a loncontaining plasmid with a minimal number of additional genes, pJMC40 was digested with the restriction enzyme EcoRI and re-ligated a t low concentrations (Fig. 1B). This procedure should yield a much smaller plasmid carrying tetracycline resistance and the lon gene with only a few hundred base pairs of chromosomal DNA flanking lon. Although this relatively simple procedure was attempted several times, very low transformation frequencies of wild-type cells were obtained following re-ligation of the cut material. None of the resulting transformants were capable of complementing a chromosomal lon mutation when transformed into a lon-cell and assayed for the mucoid phenotype on minimal media agar plates (i.e. all colonies remained mucoid). A few tetracyclineresistant transformed colonies were obtained, and restriction analysis of the plasmids present in these colonies revealed the presence of an insertion within the lon gene indistinguishable from that described for pJMC40IS above.
Constructing lon under lac Promoter Control-Since cells carrying pJMC40 are difficult to use for either physiological

Effects of Increased Protease
La in E.
coli 4511 or genetic studies and often give rise to phenotypic revertants, it was desirable to have lon under the control of an inducible promoter. Such a construct was achieved by a series of steps involving removal of the original lon promoter and replacement of this promoter with either a lac promoter or a trp-lac (tac) hybrid promoter. The lon promoter was deleted in vitro using the double-stranded DNA exonuclease Ba131, as described under "Materials and Methods," and EcoRI restriction endonuclease sites were placed flanking the promoterless gene with the aid of synthetic oligonucleotide linkers (33). T o place the lon gene under lac promoter control, the promoterless coding sequence was subcloned downstream from the lac promoter in pUC18 or into the tac promoter vector pKK223 (Fig. IC) (33). Cells overproducing the lac repressor (the laci gene product) carried on a compatible plasmid (pSG10) were transformed with either of these plasmids. The resulting transformants appear to retain viability and grow well provided lon transcription is repressed, in contrast to the cells transformed with the original lon plasmid pJMC40. However, growth of these transformants is very sensitive to induction of the lon gene. For example, when the lac inducer, isopropyl-1-thio-@-D-galactopyranoside (IPTG)? and cells carrying this plasmid were cross-streaked on agar plates, these cells failed to grow in the vicinity of the IPTG.
As expected, induction of lon under lac or tac promoter control with IPTG results in preferential synthesis of a 94kilodalton polypeptide as determined by pulse-labeling of induced cells with [35S]methionine, separation of proteins with polyacrylamide gel electrophoresis in the presence of SDS, and autoradiography of the dried gel (Fig. 3).
Induction of Protease La Inhibits Growth-To verify that overproduction of protease La is the cause of the decreased growth of cells containing pJMC40, cells with lon under lac promoter control (plasmid pSG5) or under tac promoter control (plasmid pSG11) were grown in rich media a t 37 "C to mid-log phase. One half of the culture served as a control, whereas in the other, protease La synthesis was induced by the addition of IPTG. Within 15-30 min after the addition of the inducer, the cells decreased their rate of growth, and within 1-2 generations, no further increase in optical density was observed (Fig. 4). As discussed below, these effects are probably a consequence of the excessive degradation of certain critical cell proteins. These findings thus support the earlier conclusion from studies with pJMC40 that high rates of expression of protease La are detrimental to growth.
To test whether overproduction of protease La reduces the viability of nongrowing cells, cells carrying lon under tac promoter control were grown in rich or minimal media to stationary phase. In half the culture, protease La was induced by the addition of IPTG to the medium. After different periods of incubation a t 37 "C, colony-forming units were determined by dilution and plating onto rich-media agar plates without IPTG. Although no change was seen for several hours, after 17 or 24 h in the presence of inducer, the plating efficiency of the cells with high levels of protease La was reduced 90-99% relative to the uninduced controls (Table 111) or to their initial cell number. These findings emphasize the importance of precise regulation of protease La content, especially in nongrowing cells.
Effects of High Protease Levels on Protein Degradatwn-To test whether an increased content of the ATP-dependent protease leads to more rapid degradation of abnormal proteins, a leucine auxotroph carrying pJMC40 or pJMC40IS was allowed to incorporate the amino acid analog azaleucine 'The abbreviations used are : IPTG, isopropyl-1-thio-8-D-galactopyranoside; SDS, sodium dodecyl sulfate.

Control Plasmid
Tac -Ion and SG841) carrying a control plasmid (pKK223) or the fon structural gene under tac promoter control (pSG11) were grown in M9 minimal media at 30 "C to an OD, of 0.3. Protein was labeled with 50 pCi of ["S]methionine for 30 min. Cells were disrupted by exposure to 15% trichloroacetic acid on ice for 30 min, and proteins were pelleted by centrifugation. Protein pellets were washed with 5% trichloroacetic acid and 1:l ethanol/ether and dried. Pellets were then resuspended in SDS sample buffer by boiling for 4 min. Equal counts were added to a 9-15% linear gradient polyacrylamide gel and electrophoresed in the presence of SDS. The gel was fixed in 10% acetic acid, dried, exposed to x-ray film, and then autoradiograph photographed. in place of leucine. The cultures were also exposed to [35S] methionine in order to follow the turnover of these proteins. In both cultures, proteins containing the analog were consistently degraded severalfold faster than ones containing leucine Loss of viability of ceUs overproducing protease La in stationary phase Cells (SG825) carrying pSGl1 were grown in Luria broth at 37 "C to saturation (time 0). The culture was divided in half, and IPTG was added to 500 p~ in one half, whereas the other half served as a control. The cells were incubated for an additional 5, 17, or 24 h and then serial diluted and plated for colony-forming units on Luria broth dates. ( Fig. 5), in accord with previous findings (8,9). The cells with an inactivated lon allele on the multicopy plasmid (pJMC40IS) degraded approximately 10-20% of the azaleucine-containingproteins per hour, whereas cells with an active lon gene on the episome (pJMC40) exhibited 3-5-fold more rapid degradation of these abnormal polypeptides (Fig. 5).
Analogous results were observed in cells with lon under control of the tac promoter. An arginine auxotroph carrying the plasmid pSGl1 was grown to mid-log phase in minimal media at 37 "C and washed free of arginine. These cells were resuspended in media containing canavanine, and the proteins synthesized with this analog were pulse-labeled with [35S] methionine. Following removal of the analog and the radioactive amino acid, the culture was split in two and resuspended in the original medium. One half served as a control, whereas in the other half, protease La was induced with IPTG. The induced cells consistently degraded the labeled abnormal proteins 30-50% more rapidly than did uninduced cells (Fig.  6). This effect was clearly evident within Vi h after addition of the inducer. In addition, the total amount of abnormal protein degraded was greater in the induced cells, presumably as a consequence of the higher levels of protease La. (This enhancement of the degradation of analog-containing proteins in the induced cells appeared smaller than that in cells carrying pJMC40. However, different amino acid analogs were proteins. Cells (SG840) carrying a plasmid with lon under tac promoter control were grown in minimal media at 30 "C, and canavaninecontaining proteins were generated according to "Materials and Methods." The culture was then split and half was induced with 500 pg/ml IPTG, whereas the other half served as an uninduced control. Aliquots were taken over time, and the release of label from protein was used to estimate the percentage of labeled protein degraded. Cells carrying the tac promoter plasmid without an inserted lon gene showed no difference with or without added inducer. Similar results were obtained in three independent experiments.

TABLE IV
Effects of increased copies of lon gene on degradation of puromycil polypeptides Cells (N5116) carrying either pJMC40IS or pJMC40 were grown in Luria broth at 37 "C to an ODW of 0.3, and puromycin was added to 100 pg/ml for 15 minutes. Amino-terminal protein fragments were labeled with [3H]phenylalanine for 5 min, and the cells were filtered and resuspended in fresh media lacking puromycin or radioactivity. Protein degradation was determined over time by the accumulation of trichloroacetic acid-soluble radioactivity. Similar results were obtained in independent experiments. used in different strain backgrounds, and the amount of protease La generated under these conditions may have also differed.) Incomplete polypeptides are also degraded more rapidly in cells overproducing protease La. When cultures carrying the plasmid pJMC40IS were exposed to puromycin for 15 min at 37 "C, the resulting labeled polypeptides were degraded at a rate of approximately 10% per hour. The rate of proteolysis in cells with the intact lon gene on the plasmid pJMC40 was between 20 and 25% per hour (Table IV). Similarly, when the lon gene under tac promoter control was induced following the synthesis of puromycyl polypeptides, both a greater rate and a greater amount of degradation of these protein fragments occurred than in uninduced controls (data not shown).
A Protein Fusion Is Rapidly Lost in Cells with Increased Protease La-To test whether the degradation of a particular abnormal protein is enhanced in cells carrying high levels of protease La, the stability of a @-galactosidase fusion protein was compared in cells carrying either pJMC40 or the control plasmid. Cells carrying a lon-lac2 protein fusion on an integrated bacteriophage were transformed with either pJMC40 and a protease La-@-galactosidase protein fusion produced from an integrated bacteriophage X derivative were used. Aliquots were taken over time as cells were incubated at 37 "C, and j3-galactosidase activity from the fusion protein was measured according to the method of Miller (30). Similar results were obtained in three independent experiments.
(lon') or pJMC40IS. To determine the stability of the fusion protein, chloramphenicol was added to prevent protein synthesis, and p-galactosidase activity was assayed at various times thereafter. At 37 "C, a rapid loss of this activity was observed in cells carrying pJMC40, but not in cells carrying either pJMC40IS (Fig. 7) or no plasmid (data not shown).
These data suggest that protease La, when present at high levels, will degrade a specific protein which would otherwise remain quite stable. In fact, in the control cells, this protein fusion decreased by 50% in approximately 4 h whereas in lon mutants no such loss of activity was seen. Furthermore, at 30 "C, this fusion appeared quite stable even in cells carrying pJMC40 (data not shown). Thus, the rate of degradation of this fusion protein appears to depend on both the growth temperature and the protease content of the cells.
Effects of Increased Amounts of Protease La on Degradation of Normal Cell Proteins-Unlike polypeptides containing amino acid analogs or puromycin, most proteins synthesized in growing cells are quite stable (1,2,33,34), and their rate of breakdown appears similar in lon and wild-type strains? To test the effects of high levels of the lon gene product on the degradation of these "normal" proteins, cells carrying the lon+ plasmid pJMC40 or the control plasmid pJMC40IS were grown in minimal media at 37 "C, and the proteins were labeled for 30 min with [36S]methionine. After washing, the cells were resuspended in media without radioactivity and with excess methionine. The cells carrying pJMC40 degraded these normally stable proteins approximately twice as fast (6%/h) as cells carrying the inactivated lon allele on the plasmid pJMC40IS (3%/h) (Fig. 8A). Similar results were obtained when cell proteins were labeled for 90 min (data not shown).
Analogous experiments were performed with a strain containing the lon gene under tac promoter control. Cells grown without the inducer were exposed to [35S]methionine for 30 min; the culture was then divided in two halves, one of which received IPTG to induce synthesis of protease La. The in-  (SG840) carrying a plasmid with lon under toc promoter control (pSG11) were grown at 37 "C in minimal media and pulse-labeled for approximately 1 generation (30 min) with [93]methionine. Cells were filter-washed free of radioactive amino acid and resuspended in media with excess (300 pg/ml) methionine. The culture was then split in half, with half receiving 500 pg/ml IPTG, whereas the other half served as an uninduced control. Aliquots were taken over time as cells were incubated at 37 "C. Trichloroacetic acid-soluble counts were used to estimate the rate of protein breakdown. Similar results were observed in three independent experiments. duced cultures consistently exhibited a 2-4-fold greater rate of hydrolysis of the prelabeled normal proteins (Fig. 9). The rate of degradation of these normal proteins did not rise further with time, even though more protease La was synthesized.
Although it is clear that protease La can degrade proteins present in normal cells, these experiments do not distinguish whether the enhanced proteolysis is due to slightly more rapid C. H. Chung and A. L. Goldberg, manuscript in preparation.
Effects of Increased Protease La in E. coli degradation of most cell proteins or whether certain polypeptides are selectively degraded under these conditions. To decide between these alternatives, extracts of cells carrying the lon gene under tac control were subjected to SDS-gel electrophoresis a t different times (up to 3 h) after addition of IPTG to induce protease La. As protease content increased (i.e. the 87-kDa band' in Fig. 10) and growth ceased, most polypeptide bands did not change significantly. However, several minor bands were found to decrease reproducibly in different experiments. It is interesting that the clearest examples of such polypeptides were ones associated with the fraction sedimentingat 100,000 X g (Fig. 10). In these extracts, there also was a clear accumulation of small polypeptides (i.e. less than 15 kDa), especially in the 100,000 x g pellet. Presumably, these bands represent protein fragments generated by endoproteolytic attack. At the same time, there was no obvious broadening of protein bands, as might occur upon exposure of extracts to exoproteases. Thus, the small rise in overall protein breakdown (Fig. 8A) seems to reflect the differential loss of certain polypeptides, whose disappearance may be related to the cessation of growth under these conditions.
The rate of degradation of these relatively stable polypeptides increases 2-4-fold upon starvation of E. coli for a variety of essential nutrients or ions (2,(34)(35)(36). It is unclear what proteolytic activities are responsible for this accelerated degradation of otherwise stable proteins. This process is prevented by various inhibitors of ATP production (34, 36). Although this finding suggests that an ATP-dependent system is involved, lon mutants display enhanced protein degradation upon starvation for nitrogen or a carbon source like wild-type cells (2,37). In an attempt to determine whether an increase in protease La enhances protein degradation during starvation, cells carrying either pJMC40 or pJMC40IS

43-
were deprived of nitrogen, and the rates of degradation of prelabeled cell proteins were determined (Fig. 8B). Although starved cells containing pJMC40 exhibit a higher rate of protein degradation than starved cells carrying the control plasmid, the magnitude of this difference was indistinguishable from that seen in the growing cultures. In other words, the increment in overall protein breakdown during starvation was not affected by the increase in protease La, and presumably therefore involves a distinct proteolytic system, as suggested previously (2,37,38).

DISCUSSION
These studies indicate that a severalfold increase in the intracellular level of the ATP-dependent protease La can be highly deleterious. Cellular growth is reduced (Fig. 4), and in stationary phase such cells lose viability (Table 111). Furthermore, there is a strong selection against cells with excess protease content and for cells in which the cloned gene is inactivated. Since no such selection was observed when fragments of the lon gene were carried on a multicopy plasmid, these detrimental effects must be due to the increased content of protease La and to the resulting enhancement of protein breakdown.
Interestingly, growth ceased even when the breakdown of average cell proteins had increased only by 1-2%/h (Figs. 8A and 10). In fact, the overall rate of protein degradation when protease La was induced was at most 4-6%/h, which represents only a small fraction of the rate of new protein synthesis in the growing cells. Therefore, it seems likely that the sharp decrease in cell growth and the loss of viability in stationary phase occur because certain critical cell proteins are being degraded selectively under these conditions. Following protease induction, several polypeptides, in particular membraneassociated or particulate components that sedimented in the 100,000 x g pellet, seem to be especially sensitive to proteolysis (Fig. 9). The identity of these proteins is unknown. It is unclear whether they are proteins that contain silent mutations or enzymes that normally turn over in growing E. coli.
It is interesting that in mammalian cells, many polypeptides necessary for cell growth (i.e. rate-limiting enzymes in DNA and RNA synthesis) have short half-lives (2, 39). Perhaps in E. coli there are polypeptides with analogous functions which are particularly susceptible to protease La. Isolation of phenotypic revertants which grow despite the presence of high levels of protease La should provide valuable information on the natural substrates of the lon-encoded protease and on the mechanism of the reduced viability.
These experiments provide direct evidence that protein stability in vivo depends on protease content. The presence of both the lon plasmid pJMC40 and induction of lon from an inducible promoter results in an increased degradation of various abnormal polypeptides (Figs. 5 and 6 and Table IV). Since both the initial rate of degradation and the total amount of protein hydrolyzed were greater in these cells than in the controls, protease La content seems critical in determining whether a protein undergoes degradation. Thus, in normal cells, some proteins containing amino acid analogs or puromycin or protein fusions can escape being digested, apparently because the level of this protease is insufficient for their recognition and degradation. For example, the /%galactosidase