Chromosomal Amplification of the Escherichia coli lipB Region Confers High-Level Resistance to Selenolipoic Acid

Abstract

One of the mutants (slr7 mutant) of a wild-type Escherichia coli strain resistant to selenolipoic acid reported previously (K. E. Reed, T. W. Morris, and J. E. Cronan, Jr., Proc. Natl. Acad. Sci. USA 91:3720-3724, 1994) unexpectedly grew on minimal medium following transductional introduction of a lipA null mutation. We report that the slr7 strain carries a duplication of the lip chromosomal region that causes the phenotype of the mutant strain.

Lipoic acid is a coenzyme required for the function of pathways of central metabolism present in most organisms (29). For biological function, lipoic acid must be covalently bound to its cognate enzymes and when thus bound, it serves as a long-ranging carrier of substrates between multiple active sites of large multienzyme complexes. The reactive sulfur moieties of lipoic acid reside at the distal ends of long extended protein structures, allowing lipoic acid to channel bound substrates to remarkably distant enzyme active sites (29). Although the general role of lipoic acid as a coenzyme has been known for decades, the mechanisms by which lipoic acid is synthesized and becomes attached to protein continue to be elucidated.

Escherichia coli contains three enzyme complexes that require lipoic acid as a cofactor. The three enzymes are pyruvate dehydrogenase, 2-oxoglutarate dehydrogenase, and glycine cleavage enzyme (29, 37). The most-studied lipoic acid-dependent enzyme is the pyruvate dehydrogenase complex (29) (Fig. 1). This very large protein complex (molecular weight of ca. 5 × 10⁶) is required for the synthesis of the acetyl coenzyme A (acetyl-CoA) required for entry of carbon into the citric acid cycle and fatty acids plus a variety of other molecules (7). The closely related 2-oxoglutarate dehydrogenase complex is a key enzyme of the citric acid cycle, and its product, succinyl-CoA, is a precursor in the synthesis of lysine and methionine (the pyruvate and 2-oxoglutarate dehydrogenases are often collectively called the 2-oxo acid dehydrogenases). The required covalent attachment of lipoic acid is to specific lysine residues of these enzymes that are found on protein domains of highly conserved sequences called lipoyl domains. E. coli has two distinct enzyme systems that modify lipoyl domains (26) (Fig. 2A and B). The best-understood lipoylation enzyme is lipoate protein ligase (LplA), which utilizes exogenously supplied lipoic acid to modify the specific lysine of the lipoyl domain (25). In a reaction analogous to that of biotin ligase (which modifies a structurally similar protein domain), LplA utilizes ATP to activate lipoic acid to lipoyl-AMP (Fig. 1A). The lipoyl-AMP mixed anhydride is then attacked by the ε amino group of the target lysine residue, resulting in an amide linkage (13) (Fig. 1A). LplA has been shown to be required for E. coli to utilize lipoic acid from the environment. However, lplA null mutants can be constructed without engendering a deficiency of lipoylated enzymes (25) or a nutritional requirement on minimal medium containing glucose due to the presence of a second, LplA-independent, pathway of lipoylation. This second pathway is dependent on lipB (26), which encodes a novel enzyme, lipoyl-[acyl carrier protein]-protein N-lipoyltransferase (20) (Fig. 2B) and lipA, which encodes lipoic acid synthase (24).

FIG. 1.

Pyruvate dehydrogenase reaction. The 2-oxoglutarate dehydrogenase reaction proceeds by the same mechanism as that for the pyruvate dehydrogenase reaction, except that 2-oxoglutarate replaces pyruvate and the product is succinyl-CoA. The E3 subunit is common to the two enzymes (7). The —SH HS— of E3 denotes the active-site cysteine residues. Abbreviations: ThDP, thiamine diphosphate; lip, lipoic acid; FAD, flavin adenine dinucleotide.

FIG. 2.

The two pathways of lipoic acid attachment in E. coli. The systematic names for lipoic acid and Se-lip are 1,2-dithiolane-3-pentanoic acid and 1,2-diselenolane-3-pentanoic acid, respectively. ACP, the acyl carrier protein of fatty acid biosynthesis.

In one approach taken by this laboratory to study lipoic acid attachment pathways, a lipoic acid analog, selenolipoic acid (Se-lip) (Fig. 2D), was synthesized, and E. coli mutants resistant to the analog were isolated (32). Se-lip has the same carbon backbone as lipoic acid with selenium atoms substituted for the two sulfur atoms (Fig. 2C and D). Se-lip is taken up by wild-type E. coli and used to modify lipoyl domains (32). This aberrant modification of lipoyl domains leads to inactive 2-oxo acid dehydrogenase complexes, since Se-lip cannot be reduced in vivo (32). The toxic effects of Se-lip can be overcome by addition of high levels of lipoic acid to the media or by addition of acetate plus succinate, demonstrating that toxicity is dependent upon modification of the lipoyl domains with Se-lip and concomitant inactivation of the 2-oxo acid dehydrogenases (32). One of these mutations (first called slr1 then lipA1) was shown to be an lplA allele that encoded a lipoate protein ligase of altered specificity (26, 32). The lplA1-encoded ligase discriminated against incorporation of Se-lip and thereby gave a larger proportion of lipoyl domains that were correctly modified with lipoic acid (32).

Another Se-lip-resistant mutant, a slr7 mutant, had a more complex and puzzling phenotype. Upon transductional introduction of lipA null mutation into the slr7 strain, the resulting strain remained a lipoic acid prototroph (32), although introduction of a lipA null mutation into wild-type strains results in lipoic acid auxotrophy (37). Therefore, it was proposed that the Se-lip resistance of the slr7 strain might be due to an activation of some unidentified cryptic lipoic acid biosynthetic pathway that synthesized sufficient lipoic acid to effectively compete with the selenium analog for incorporation into the 2-oxo acid dehydrogenase complexes (32). We report that this proposal is incorrect. The slr7 mutation is a duplication of the chromosomal segment that contains both lipA and lipB.

Cloning of the “lipA-complementing element.”

Prior work showed that a lipA null allele (lipA::Tn1000dkan) could be transduced into the slr7 strain without engendering the expected lipoate auxotrophy. This result led to the hypothesis that the Se-lip resistance phenotype of the slr7 mutant was due to a mutation activating some unidentified cryptic lipoic acid synthetic gene (32). We attempted to identify the putative gene by cloning a lipA-complementing fragment from the slr7 lipA::Tn1000dkan strain, strain KER279 (see Table 1 for detailed descriptions of bacterial strains and plasmids). We constructed a genomic library by digesting strain KER279 genomic DNA with EcoRI and ligating the resulting fragments into plasmid pSU19 (23), a vector of medium copy number. We then selected complementing clones after transforming this library into strain KER176 (lipA::Tn1000dkan) and selecting for growth on minimal medium containing glucose and chloramphenicol (media and genetic methods were as described previously [25, 26, 31, 32, 37]). On minimal medium, a lipA strain requires supplementation with either lipoic acid or acetate plus succinate (these supplements bypass the lipoic acid-dependent enzymes essential for aerobic growth on minimal medium containing glucose); thus, plasmid clones able to suppress lipA permitted growth on minimal medium containing glucose but lacking lipoate. A complementing plasmid (pSJ10) that contained a 4-kb chromosomal segment was obtained. Upon isolation and transformation, the plasmid was found to complement the lipA null mutation of strain KER176, but not the lipB null mutation of strain KER184. Plasmid pSJ10 failed to confer Se-lip resistance (see below). The segment of pSJ10 DNA responsible for lipA complementation was sequenced and found to be identical to the lip region of E. coli. The complementing DNA segment contained lipA but lacked the N-terminal portion of lipB. To determine the sequences responsible for complementation, we performed Tn1000 mutagenesis (14) of plasmid pSJ10. Plasmid pSJ10 was transformed into the F plasmid strain JA200 (the F plasmid carries Tn1000), this strain was mated with strain TVB127, and exconjugants resistant to chloramphenicol, kanamycin, and tetracycline were selected on Luria-Bertani (LB) medium. The exconjugants were screened for lipoic acid auxotrophy by streaking bacteria onto minimal E medium containing glucose with or without 5 ng of R-lipoic acid per ml. Twenty-four Tn1000-mutagenized isolates of pSJ10 were screened, of which 14 had lost the ability to complement lipA. All of the Tn1000 insertions lacking lipA complementation activity mapped to lipA, whereas all of the isolates that retained complementation activity had insertions outside lipA. These results demonstrated that the ability of a slr7 lipA::Tn1000dkan strain, KER279, to grow on minimal medium lacking lipoic acid was not due to a mutation of some unknown lipoic acid metabolic gene. Instead, the lipoic acid prototrophy was due to the presence of at least one functional copy of lipA on the chromosome. The intact copy was present despite the fact that a lipA::Tn1000dkan insertion was also present in the chromosome. These data strongly suggested that a chromosomal duplication of the lip region was present in strain KER279. This conclusion was strengthened by use of a second lipA null allele. A lipA::lacZ-CAT insertion constructed by T. Morris of this laboratory was transduced into strain KER270 (slr7), and the resulting strain SWJ9 (slr7 lipA::lacZ-CAT) did not require lipoic acid for growth on minimal medium containing glucose. Therefore, the result obtained was independent of the disruption allele used.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Relevant genotype or characteristic(s)	Lipoate phenotype	Construction methoda	Source or reference(s)
Strains
JK1	rpsL (W3110)	Prototroph (wild type)		37
JA200	F+			8
XL1-Blue	F′ (proAB lacIq ΔM15) ΔlacZ recA1 endA1			Stratagene
KL718	F′152 plasmid			22, 27
TVB127	rpsL lipA::Tn1000dkan recA::tet			37
KER176	rpsL lipA::Tn1000dkan	Auxotroph		37
KER184	rpsL lipB::Tn1000dkan	Auxotroph		37
KER270	rpsL slr7	Prototroph, Se-lip resistant		32
KER279	rpsL slr7 lipA::Tn1000	Prototroph		32
TM257	rpsL lipA::lacZ-CAT	Prototroph		This work
SWJ1	KER176/F′152	Diploid for lip conjugation region, prototroph		This work
SWJ9	rpsL slr7 lipA::lacZ-CAT	Prototroph	P1 (TM257) × KER270	This work
SWJ34	rpsL slr7 lipA::lacZ-CAT lipA::Tn1000dkan	Auxotroph	P1 (KER176) × SWJ9	This work
SWJ35	rpsL slr7 lipA::lacZ-CAT lipA::Tn1000dkan	Auxotroph	P1 (TM257) × KER279	This work
SWJ36	rpsL slr7 lipA::lacZ-CAT lipA::Tn1000dkan	Prototroph	P1 (KER176) × SWJ9	This work
SWJ37	rpsL slr7 lipA::lacZ-CAT lipA::Tn1000dkan	Prototroph	P1 (TM257) × KER279	This work
SWJ412	JK1/pKR111			31
SWJ413	JK1/pKR109			31
SWJ19	KER176/pSJ10			This work
SWJ414	SW3/pCRK101			This work
SWJ44	SW3/pTM61-4			This work
TVB127	lipA::Tn1000dkan rpsL ΔrecA	Auxotroph		32
Plasmids
pSU19	Chloramphenicol-resistant cloning vector			23
pCKR101	Ampicillin-resistant cloning vector			21
pKR109	Minimal lipA gene inserted into pCKR101			31
pKR111	Minimal lipB gene inserted into pCKR101			31
pTM61-4	Minimal lplA gene inserted into pCKR101			25
pSJ10	4-kb lip region EcoRI fragment of SWJ9 inserted into the EcoRI site of pSU19			This work

^aIn transductional crosses, the donor strain is shown in parentheses.

Southern blot analyses of the slr7 chromosome.

In order to obtain physical evidence for the presence of a chromosomal duplication of the lip region in strain KER279, we performed Southern blot analyses in which the chromosomal DNAs from strains JK1 (wild type) and KER270 (slr7) were compared with those from strains carrying one of the two lipA null alleles (Fig. 3A). The rationale was that the inserted DNA of the null alleles would allow detection of the multiple lipA alleles. The lipA null mutant strains were KER176 (lipA::Tn1000dkan), KER279 (slr7 lipA::Tn1000dkan), SWJ9 (slr7 lipA::lacZ-CAT), and TM257 (lipA::lacZ-CAT). Genomic DNA from each strain was digested with EcoRI, and the resulting digests were fractionated on 0.7% agarose gels and transferred to nitrocellulose. The blot was then probed with a digoxigenin-labeled probe made using a 4-kb EcoRI fragment of plasmid pSJ10 carrying lipA and all but the 24 amino-terminal residues of lipB. Southern blotting and hybridization were performed by previously described protocols (34). Detection of lip region DNA fragments utilized the Lumi-Phos system with a probe generated using the Genius system (Boehringer-Mannheim). The wild-type strain gave a single 4-kb band, as did the slr7 strains. The null mutant strains lacked the 4-kb band and instead showed other bands characteristic of the lipA insertion allele used. In contrast, the slr7 strains that also carried a lipA insertion allele showed both the 4-kb EcoRI band and the additional bands characteristic of the lipA null allele. These results clearly demonstrate the presence of a chromosomal duplication of the lip region. In order to estimate the size of the chromosomal duplication, we digested chromosomal DNAs from strains JK1 and KER270 with a variety of enzymes and compared the results with those expected from the genome sequence (Fig. 3B). Since no junction fragments were seen in these digests, the amplified region of chromosome must exceed 22 kb. However, since two lipA alleles can be efficiently cotransduced by phage P1 (see below), each duplicated region must be smaller than about 50 kb.

FIG. 3.

Southern blots demonstrate duplication of the lip region of the chromosome. (A) Chromosomal DNAs from strains were prepared (strains given above the lanes). Strain JK1 is the wild-type strain, and strain KER270 is the original slr7 strain. Strains KER176 and TM257 carry the lipA::Tn1000dkan and lipA::lacZ-CAT null alleles, respectively, and are lipoate auxotrophs. Strains KER279 and SWJ9 are slr7 strains that carry a second lipA allele, either lipA:: Tn1000dkan or lipA::lacZ-CAT, respectively. These strains are lipoate prototrophs. The DNAs were digested with EcoRI. The Ladder lane contains molecular size standards. The sizes of the chromosomal fragments (in kilobases) are given to the right of the figure. (B) Southern blots demonstrate that the slr chromosomal duplication is at least 24 kb in length. The products obtained by digestion of the chromosomal DNAs of strains JK1 (wild type) and KER270 (slr7) with several different restriction enzymes are shown. Most of the band in the strain KER270 PstI lane was lost upon transfer. The sizes of the chromosomal fragments (in kilobases) are given to the left of the figure. A map of the chromosomal region of E. coli K-12 is shown at the bottom of panel B. Restriction site abbreviations: B, BamHI; P, PstI; V, EcoRV; H, HindIII; E, EcoRI.

Some slr7 cells carry at least three copies of the lip region.

A second method was used to estimate the number of copies of the lip region present in the slr7 strains. We transduced the two differently marked lipA insertional mutations into an slr7 strain. If the strain contained only two copies of lipA, introduction of both null mutations should disrupt all copies of lipA and result in lipoic acid auxotrophy. Moreover, Southern analysis should show loss of the band representing the wild-type allele. If more than two lipA copies were present, the strain should remain a lipoic acid prototroph and Southern analysis should indicate the presence of the wild-type allele together with the bands characteristic of the two lipA null mutations (26, 32).

In this test, we analyzed the progeny of two transductional crosses. In the first cross, we used a phage P1vir lysate grown on strain TM257 (lipA::lacZ-CAT) to transduce strain KER279 (slr7 lipA:Tn1000dkan). In the second cross, we transduced kanamycin resistance into strain SWJ9 (lipA::lacZ-CAT slr7) by use of phage P1vir lysate from strain KER176 (lipA:Tn1000dkan). Transductants resistant to both chloramphenicol and kanamycin were selected in the presence of lipoic acid and then scored for lipoic acid auxotrophy. In both crosses, approximately half of the transductants required lipoic acid for growth, whereas the remaining half grew without lipoic acid. Two prototrophic strains (SWJ36 and SWJ37) and two auxotrophic strains (SWJ34 and SWJ35) were chosen at random and analyzed by Southern blot analysis (Fig. 4). This analysis clearly showed that the auxotrophic strains contained two lipA alleles (lipA::Tn1000dkan and lipA::lacZ-CAT), whereas the prototrophic strains contained three alleles (the two insertion null alleles plus the wild-type allele). Note that it is possible that transduction could remove a copy (or copies) of the amplified region through recombination of the ends of the transducing DNA fragment with the left end of one copy of the chromosomal segment and the right end of another copy. However, in transductions in which the lipA::Tn1000dkan or lipA::lacZ-CAT allele was transduced into the slr7 strain, all of the recombinants were prototrophic. If copies of the amplified region were efficiently removed by transduction, then lipoic acid auxotrophic recombinants should have resulted. Moreover, prior transductional analyses of chromosome duplications (4, 19) and triplications (4) also failed to detect such events. Finally, the branch migration step of current models of homologous recombination (35) would argue against this possibility. Therefore, it seems likely that in slr7 strains the lip region is in flux and some cells may contain two copies of the lip region while other cells contain at least three copies.

FIG. 4.

Variable degrees of lip region amplification in the cells of an slr7 culture. The figure is a Southern blot (EcoRI digestion done as described in the legend to Fig. 3A) analysis of the recombinants from transductional crosses in which a second tagged lipA null mutation was transduced into an slr7 strain that carried a tagged lipA null mutation (see text). Recombinants that grew (Y for yes) or failed to grow (N for no) on glucose-containing minimal medium were analyzed together with the parental strains (all strains grew when supplemented with 5 ng of lipoate per ml). The bands diagnostic for the Tn1000dkan allele, the lipA::lacZ-CAT allele, and the wild-type alleles are the 6.0-, 4.5-, and 4.0-kb bands, respectively. The presence (+) or absence (−) of the Tn1000dkan and lipA::lacZ-CAT alleles are shown under the blot. The smaller lipA null allele bands are not shown because they were indistinct in many lanes due to the high background and their low intensities. The sizes of the chromosomal fragments (in kilobases) are given to the right of the figure.

Se-lip resistance is conferred by plasmid-borne multiple copies of lipB.

Although we had shown that the slr7 strains contained a chromosomal duplication of the lip region, we had not shown that the presence of multiple copies of the lipB gene was responsible for conferring resistance to Se-lip. In order to test if the lipB duplication itself was responsible for conferring resistance to Se-lip, we tested the ability of plasmids that contained either a minimal lipA gene or a minimal lipB gene to confer Se-lip acid resistance to a wild-type strain. Introduction of the lipB plasmid pKR111 into the wild-type strain resulted in Se-lip resistance, whereas introduction of the lipA plasmid (pKR109) gave no increase in Se-lip resistance (Table 2). To more closely approximate the lipB gene copy numbers in the slr7 strain, we introduced one or two copies of the lip region into the wild-type strain JK1 by use of an F′ plasmid (22). F′ plasmid F152, which contains a functional lip region (20), was introduced into strain JK1 by conjugation followed by selection for lipA function on glucose-containing minimal medium supplemented with kanamycin, and the resulting strain (SWJ11) was tested for Se-lip resistance. The presence of this F′ plasmid resulted in resistance to 5,000 ng of Se-lip per ml (Table 2).

TABLE 2.

Abilities of various plasmids to confer resistance to Se-lip to two strains^a

Strain	Plasmidb	Plasmid insert	Growthc in the presence of the following concn (ng/ml) of Se-lip:
			0	50	500	5,000
JK1	pSU19	None (vector control)	++++	−	−	−
	pSJ10	lipA	++++	−	−	−
	pSJ17	lipA::Tn1000	++++	−	−	−
	pCTV616	lipA and lipB	++++	++++	+++	+++
	pCTV634	lipA Tn1000 in lipB	++++	−	−	−
	pCTV628	lipA and lipB::Tn1000	++++	++++	+++	+++
	pCRK101	None (vector control)	++++	−	−	−
	pKR112	lipB in pCRK101	++++	++++	+++	+++
	F′152	F′ plasmid (min 12.1-17.66 chromosomal segment)	++++	++++	+++	+++
SWJ3	pCRK101	None (vector control)	++++	++++	+++	+++
	pTM61-4	lplA in pCRK101	++++	−	−	−

^aThe abilities of various plasmids to confer resistance to Se-lip were tested by transforming (or mating) the plasmids into either the wild-type strain JK1 or the slr7 strain SWJ3 and plating the recombinant strains on minimal E medium containing glucose supplemented with 0 to 5,000 ng of Se-lip per ml.

^bThe pSJ and pCTV plasmids are all derivatives of the p15a plasmid pSU19, and the lip genes were expressed from the native promoters. Plasmid pCKR101 is a tac promoter expression plasmid (21).

^cSymbols: ++++, growth similar to that of the wild-type strain growing in the absence of Se-lip; −, no growth after 1 day; +++, intermediate rate of growth near the wild-type level.

Multiple copies of LplA restore Se-lip sensitivity to slr7 strains.

Our model (see below) explaining the ability of extra copies of lipB to confer increased resistance to Se-lip is that an increase in attachment of endogenously synthesized lipoic acid to the 2-oxo acid dehydrogenases effectively competes with the attachment of Se-lip by LplA. This competition model implies that an increase in the LplA activity of the slr7 strain should cancel the effect of increased LipB activity and result in loss of Se-lip resistance. To test this hypothesis, we transformed the lplA plasmid pTM61-4 (25) into strain KER270 and tested the resultant strain (strain SWJ44) for Se-lip resistance. The presence of plasmid pTM61-4 restored sensitivity to Se-lip in the slr7 strain (Table 2).

Conclusions.

Our data clearly demonstrate that the presence of multiple copies of lipB confers resistance to Se-lip and explains the slr7 phenotype. Prior work from this laboratory showed that lipB was required for posttranslational modification of lipoyl domains using de novo-synthesized lipoate (26). The other characterized Se-lip resistance allele (slr1) is an allele of the lplA (26, 32) gene, which encodes the enzyme responsible for modifying lipoyl domains utilizing exogenously supplied lipoic acid (25, 26). The lipB gene encodes a lipoyl-[acyl carrier protein]-protein N-lipoyltransferase and is essential for the utilization of de novo-synthesized lipoic acid (unpublished data). The ability of extra copies of lipB to confer high-level resistance to Se-lip indicates that increased activity of the lipB gene product increases the level of lipoylation by endogenously synthesized lipoic acid. Upon amplification of this pathway, LipB must effectively compete with the utilization of exogenous Se-lip acid via the LplA-dependent pathway. This competition was demonstrated by the loss of the slr7 Se-lip resistance upon overproduction of LplA (Table 2). Moreover, it seems probable that the capacity of the lipoate synthetic pathway exceeds that of the lipB-mediated attachment to protein, since the presence of additional copies of lipA had no effect on Se-lip resistance (Table 2).

The F′, transduction, and Southern blot data show that a very modest (two- to threefold) increase in the dosage of lipB resulted in resistance to levels of Se-lip that completely inhibited growth of the wild-type strain. The expected and more-conventional result was a linear relationship between gene dosage and the level of resistance, such as that seen with E. coli ampC. Progressive stepwise selection of ampicillin-resistant isolates results in strains carrying repetitions of ampC and the neighboring chromosomal markers that give a strict correspondence between the number of ampC copies and the level of ampicillin resistance (11, 12). In the case of Se-lip resistance, no such correspondence was seen. How can a modest increase in gene dosage result in a large increase in resistance to Se-lip? In the case of the slr7 mutation, we believe the answer lies in the unusual structural properties of the 2-oxo acid dehydrogenases. These enzymes are much more complicated than a simple monomeric hydrolyase such as AmpC. The 2-oxo acid dehydrogenases (Fig. 1) are very large protein complexes that catalyze several partial reactions involving three active sites present on three different subunits (28, 29). We believe that the behavior of the slr7 mutant results from the properties of these enzymes plus the fact that wild-type E. coli strains contain levels of the 2-oxo acid dehydrogenases that are in functional excess.

The structures and enzyme mechanism of the 2-oxo acid dehydrogenases are known to combine to give a highly nonlinear relationship between the degree of protein lipoylation and the in vitro activity of the 2-oxoacid dehydrogenase complex (1, 5, 6, 10, 15, 30, 36). The rate-limiting step in the reaction mechanism is decarboxylation of the 2-oxo acid (Fig. 1); thus, the overall rate of a 2-oxo acid dehydrogenase complex is unaffected by significant decreases in the levels of lipoic acid bound to the complex (the cofactor is not required until the subsequent partial reaction) (10). Moreover, a single lipoyl domain is able to productively access many active sites within these very large protein complexes. That is, the lipoyl domain of one E2 subunit can be acetylated by E1 subunits bound to different E2 subunits; therefore, there is only a weak dependence of the overall activity of the complex on the level of protein-bound lipoic acid (1, 5, 6, 10, 15, 30, 36). These conclusions result from a variety of experimental approaches in several laboratories that removed or inactivated major fractions of the lipoyl domains of a 2-oxo acid dehydrogenase complex. Until complete loss of functional lipoyl domains was approached, these manipulations had little or no effect on either the overall enzymatic activity of the complex or on the growth of E. coli strains that required function of the complex (1, 5, 6, 10, 15, 30, 36). The functional redundancy of the lipoyl domains plus the fact that lipoic acid is not involved in the rate-controlling step of the 2-oxo acid dehydrogenase reaction mechanism results in a strikingly nonlinear dependence of dehydrogenase activity on lipoic acid content. For example, a pyruvate dehydrogenase complex having only 2 to 5% of the wild-type level of lipoic acid attachment has been reported to retain about 20% of the activity of the enzyme complex (36). These data indicate that in the presence of Se-lip, virtually all of the lipoyl domains must be modified by the analog in order to inactivate the 2-oxo acid dehydrogenase complexes. Therefore, small increases in the level of lipoic acid attachment resulting from modest overexpression of LipB would be expected to give disproportionate increases in enzyme activity.

The second important factor in lipB-mediated Se-lip resistance is that E. coli does not require full activity of the 2-oxo acid dehydrogenases for growth on minimal medium containing glucose. Thus, the small increases in 2-oxo acid dehydrogenase enzyme activities resulting from lipB duplication would result in large increases in growth. Studies in which nonsense mutations in the genes that encode subunits of either of the dehydrogenase complexes were suppressed by tRNA suppressors of various efficiencies indicate that 10% of the wild-type 2-oxo acid dehydrogenase complex activity levels permits growth on glucose-containing minimal medium (9, 16, 17). Therefore, when the enzymatic and in vivo data are considered in context, they indicate that attachment of lipoic acid in place of Se-lip to a few percent of the 2-oxo acid dehydrogenase complexes would be sufficient for growth. This picture is not only consistent with the behavior of the slr7 mutant but argues strongly that the nonlinear mechanism proposed for the 2-oxo acid dehydrogenases (1, 5, 6, 10, 15, 30, 36) operates in vivo. In order to further test this argument, it would be advantageous to measure the extent of lipoic acid attachment in the slr7 mutant grown with Se-lip. However, lipoic acid is determined by a bioassay using a lipA strain of E. coli (18) and the lipoic acid released by hydrolysis of the 2-oxoacid dehydrogenase complexes would be contaminated with a large excess of Se-lip. Since Se-lip would inhibit growth of the assay strain, use of the bioassay is precluded.

It seems clear that the acquisition of Se-lip resistance in the slr7 strain arose by a recognized pathway. Anderson and Roth (2-4) and Roth et al. (33) argue that 10% of the cells of a culture carry a spontaneously arising duplication of some region of the chromosome. Therefore, within the population of cells spread on Se-lip-supplemented plates, a few cells would have an existing duplication of the lipB region and would survive incorporation of the inhibitor due to their increased ability to attach lipoic acid. We have not determined the maximal number of copies of the lip region in the slr7 mutant strain, but since a derivative carrying three copies can be constructed, the number is at least three. Moreover, the finding that introduction of a F′ plasmid (F′ plasmids have copy numbers of one or two per cell) carrying the lip chromosomal region results in Se-lip resistance indicates that a two- to threefold increase in lipB is sufficient to give the resistance phenotype.