Identification and Characterization of aarF, a Locus Required for Production of Ubiquinone in Providencia stuartii and Escherichia coli and for Expression of 2′-N-Acetyltransferase in P. stuartii

Abstract

Providencia stuartii contains a chromosomal 2′-N-acetyltransferase [AAC(2′)-Ia] involved in the O acetylation of peptidoglycan. The AAC(2′)-Ia enzyme is also capable of acetylating and inactivating certain aminoglycosides and confers high-level resistance to these antibiotics when overexpressed. We report the identification of a locus in P. stuartii, designated aarF, that is required for the expression of AAC(2′)-Ia. Northern (RNA) analysis demonstrated that aac(2′)-Ia mRNA levels were dramatically decreased in a P. stuartii strain carrying an aarF::Cm disruption. The aarF::Cm disruption also resulted in a deficiency in the respiratory cofactor ubiquinone. The aarF locus encoded a protein that had a predicted molecular mass of 62,559 Da and that exhibited extensive amino acid similarity to the products of two adjacent open reading frames of unknown function (YigQ and YigR), located at 86 min on the Escherichia coli chromosome. An E. coli yigR::Kan mutant was also deficient in ubiquinone content. Complementation studies demonstrated that the aarF and the E. coli yigQR loci were functionally equivalent. The aarF or yigQR genes were unable to complement ubiD and ubiE mutations that are also present at 86 min on the E. coli chromosome. This result indicates that aarF (yigQR) represents a novel locus for ubiquinone production and reveals a previously unreported connection between ubiquinone biosynthesis and the regulation of gene expression.

The gram-negative bacterium Providencia stuartii is a member of the Proteeae, which includes the genera Proteus, Morganella, and Providencia. Members of the Proteeae possess peptidoglycan that is O acetylated at the C-6 hydroxyl position of N-acetylmuramyl residues (8). This modification confers resistance to muramidases such as lysozyme and has been speculated to modulate the activity of endogenous peptidoglycan-specific hydrolases, termed autolysins (8, 9, 16). P. stuartii contains a chromosomal 2′-N-acetyltransferase, encoded by the aac(2′)-Ia locus, that has been implicated in this process (7, 29, 34, 38). This enzyme is also capable of acetylating and inactivating certain aminoglycoside antibiotics and was originally identified in clinical strains of P. stuartii overexpressing the enzyme (7, 38).

The aac(2′)-Ia gene is expressed at low levels in wild-type P. stuartii (34). The expression of aac(2′)-Ia is controlled in part by a small transcriptional activator, AarP, that is related to members of the XylS-AraC family of positive activators (18, 24). Recessive mutations that result in increased aac(2′)-Ia mRNA accumulation have also been identified in five loci (aarA, aarB, aarC, aarD, and aarG) (25, 32–35). The expression of aarP has been shown to be increased in the aarB, aarC, and aarG mutant backgrounds. These results suggest that aarP may play a central role in the activation of aac(2′)-Ia expression (32, 35).

In this study, we report the identification of the aarF gene of P. stuartii and demonstrate that aarF function is required for the expression of aac(2′)-Ia. We also present evidence suggesting that aarF is functionally equivalent to Escherichia coli yigQR and that both aarF and yigQR represent novel loci required for the production of ubiquinone.

MATERIALS AND METHODS

Bacterial strains and plasmids.

All bacteria, bacteriophages, and plasmids used in this study are described in Table 1.

TABLE 1.

Bacterial strains and plasmids

Strain or plasmid	Genotype and relevant markers	Source
Strains
E. coli
XL1-Blue	recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 Δlac-pro (F′ proAB lacIqlacZΔM15 Tn10)	Stratagene
DH5α	recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 Δ(lacZYA-argF)U169 φ80dlacZΔM15	Gibco/BRL
DH5α λpir	DH5α lysogenized with λpir	24
SM10 λpir	thi thr leu tonA lacY supE recA RP4-2-Tc::Mu Kmr λpir	28
BL21(DE3)	(B strain; rB− mB−) F−ompT hsdSB; λ prophage carries T7 RNA polymerase gene	Novagen
MC4100	F−araD139 Δ(argF-lac)U169 rpsL150(Strr) relA1 flbB5301 deoC1 ptsF25 rbsR	36
DM111	MC4100 lamB	This study
DM113	MC4100 lamB Rifr	This study
DM115	MC4100 lamB RifryigR::Kan	This study
RM1734	(MG1655) λ− F−rph-1	R. Maurer (20)
DM123	RM1734 yigR::Kan [P1 (DM115) × RM1734]	This study
AN66	thr-1 leuB6 lacZ4 glnV44 rpsL8(Strr) ubiD410	12
AN70	metB ubiE401	39
P. stuartii
PR50	Wild type	34
PR50.AFM12	aarF1	This study
PR54	aarF::Cm	This study
Plasmids
pUC4.KIXX	Cloning vector, Ap, Kan	Pharmacia
pACYC184	Medium copy vector, Cm, Tc	6
pACYC184.lacIq	pACYC184::1.3-kb XbaI-ClaI fragment containing lacIq	This study
pAFM12	pACYC184::3.6-kb Sau3AI containing aarF	This study
pMJR1560	Cloning vector, lacIq	Amersham
pBluescript SK(−)	High-copy-number vector, Ap	Stratagene
pSK.aarF	pBluescript SK(−)::1.9-kb SphI fragment from pAFM12	This study
pBluescript KS(−)	High-copy-number vector, Ap	Stratagene
pKS.aarF	pBluescript KS(−)::1.9-kb SphI fragment from pAFM12	This study
pKS.NheI	Derivative of pSK.aarF with a frameshift mutation inserted at a unique NheI site internal to the aarF coding region	This study
pSK.aarF::Cm	pSK.aarF containing a 3.6-kb chloramphenicol resistance cassette inserted within a unique NruI site in aarF	This study
pKNG101	R6K-derived suicide plasmid containing Str and sacB	22
pKNG101.aarF::Cm	pKNG101::6-kb BamHI-ApaI fragment from pSK.aarF::Cm	This study
pET21a	T7 expression plasmid, Ap	Novagen
pEF1	pET21a::3.5-kb Sau3AI fragment containing E. coli yigOPQR genes	This study
pSK-2.6	pBluescript SK(−)::2.6-kb SalI fragment from pEF1 (yigPQR)	This study
pSK.yigQR	Derivative of pSK-2.6 with yigP deleted	This study
pSK.yigR	Derivative of pSK-2.6 with yigP and initiating codon of yigQ deleted	This study
pSK.yigQRΔDraI	Derivative of pSK.yigQR with C-terminal yigR deletion	This study
pSK.yigR::Kan	pSK-2.6 containing a 1.3-kb kanamycin resistance cassette inserted within a unique BsmI site in yigR	This study
pKNG101.yigR::Kan	pKNG101::3.8-kb SalI fragment from pSK.yigR::Kan	This study

Media and bacterial growth.

Bacteria were routinely grown in Luria-Bertani (LB) broth at 37°C. To test for the aerobic utilization of nonfermentable carbon sources, M9 minimal agar plates (26) containing either 0.2% glucose or 0.5% succinate were used. For the growth of E. coli AN66 ubiD and AN70 ubiE, M9 plates were supplemented with l-leucine, l-threonine, and l-methionine each at a final concentration of 0.2 mM and with thiamine at a final concentration of 0.02 μM.

Gentamicin resistance determinations.

MICs for gentamicin were determined by an agar dilution method with twofold increasing concentrations of gentamicin. The MIC was defined as the lowest concentration of gentamicin that prevented the formation of single colonies.

Plasmid constructions.

A genomic library of P. stuartii DNA was constructed by ligation of partial Sau3AI fragments into BamHI-digested and dephosphorylated pACYC184 and was described previously (6, 24). Plasmid pAFM12 is a pACYC184 recombinant with a 3.6-kb Sau3AI fragment of P. stuartii DNA containing the aarF gene. Plasmid pSK.aarF was constructed by inserting a 1.9-kb SphI fragment from pAFM12 into pBluescript SK(−) linearized with SmaI. A genomic library of E. coli partial Sau3AI fragments constructed in pET21a was kindly provided by P. deBoer, Case Western Reserve University. Plasmid pEF1 is a pET21a recombinant and contains a 3.5-kb insert. Plasmid pSK-2.6 contains a 2.6-kb SalI fragment from pEF1 ligated into the SalI site of pBluescript SK(−). Plasmid pSK.yigQR was constructed by digesting pSK-2.6 with EcoRV to release a 457-bp fragment and religating the linearized plasmid. Plasmid pSK.yigQRΔDraI was constructed by linearizing pSK.yigQR with HincII followed by partial digestion with DraI to release a 467-bp fragment and religation. Plasmid pSK.yigR was constructed by digesting pSK-2.6 with ClaI and NarI to release a 701-bp fragment and religating the linearized plasmid.

Identification of AarF.

To identify the aarF gene product, a XhoI-XbaI fragment containing the aarF gene was excised from pSK.aarF and ligated into pBluescript KS(−) to create pKS.aarF. In pKS.aarF, the aarF gene is downstream from and in the same orientation as the T7 promoter. To create a negative control plasmid, pKS.aarF was linearized with NheI, which cuts internal to the aarF coding region, end filled with the Klenow fragment and deoxynucleoside triphosphates (dNTPs), and religated. The resulting plasmid, pKS.NheI, carries a frameshift mutation that truncates the predicted AarF protein after amino acid 99. To ensure that the aarF gene would not be expressed in the absence of isopropyl-β-d-thiogalactopyranoside (IPTG), the lacI^q gene was introduced into plasmid pACYC184 (6) as follows. A 1.3-kb fragment containing the lacI^q gene was released from plasmid pMJR1560 (Amersham) by digestion with EcoRI and HindIII and was cloned into pBluescript SK(−) that had been digested with the same enzymes to create pSK.lacI^q. The lacI^q gene was then excised from pSK.lacI^q as a 1.3-kb XbaI-ClaI fragment and was cloned into pACYC184 that had been digested with the same enzymes to create pACYC184.lacI^q. Plasmid pACYC184.lacI^q and each of the aarF derivative plasmids were cointroduced into E. coli BL21(DE3) (Novagen). Cultures were shaken in LB broth at 37°C to an optical density at 600 nm (OD₆₀₀) of 0.6 and induced with 1 mM IPTG. After 30 min, rifampin was added to a final concentration of 100 μg/ml, and cultures were shaken for an additional 2.5 h. Cells were harvested, and 15-μl aliquots were dissolved in sodium dodecyl sulfate (SDS) loading dye, boiled, and run on SDS–10% polyacrylamide gels. Total cellular protein was visualized after Coomassie blue staining.

Construction of chromosomal aarF and yigR disruptions.

To construct an aarF null allele in P. stuartii, plasmid pSK.aarF was linearized at a unique NruI site present midway in the aarF coding region at position 956. A chloramphenicol resistance cassette from pUT::mini-Tn5Cm (15), present as a 3.6-kb HindIII fragment, was end filled with the Klenow fragment of DNA polymerase I and dNTPs and ligated into NruI-linearized pSK.aarF to produce pSK.aarF::Cm. To recombine the aarF::Cm disruption into the P. stuartii chromosome, a 6-kb BamHI-ApaI fragment was excised from pSK.aarF::Cm and ligated into suicide vector pKNG101 (22) that had been digested with the same enzymes. The resulting plasmid, designated pKNG101.aarF::Cm, was integrated into the chromosome of strain PR50 by conjugal mating as described previously (24). The merodiploid was resolved by selection on 5% sucrose, and strains containing the disrupted aarF locus were identified on the basis of chloramphenicol resistance. Southern analysis confirmed that the chromosomal aarF locus had been disrupted by the chloramphenicol resistance cassette.

To construct a yigR null allele in E. coli, plasmid pSK-2.6 was linearized at a unique BsmI site internal to the yigR open reading frame and treated with T4 DNA polymerase and dNTPs to produce blunt ends. A 1.3-kb SmaI fragment containing a kanamycin resistance cassette was excised from pUC4::KIXX (Pharmacia) and ligated into linearized pSK-2.6 to produce pSK.yigR::Kan. The 3.8-kb yigR::Kan disruption was then excised from pSK.yigR::Kan with SalI and ligated into the unique SalI site of suicide vector pKNG101. The resulting plasmid, designated pKNG101.yigR::Kan, was introduced into the chromosome of strain DM113 by conjugal mating essentially as described previously (24), with the exception that rifampin was used at 100 μg/ml to counterselect against the donor strain. The merodiploid was resolved by selection on 5% sucrose, and strains containing the disrupted yigR locus were identified on the basis of kanamycin resistance. Southern analysis confirmed that the chromosomal yigR locus had been disrupted by the kanamycin resistance cassette. The chromosomal yigR::Kan disruption was then introduced into wild-type E. coli RM1734 via a P1 lysate derived from DM115. Transductants were obtained on LB agar plates containing 50 μg of kanamycin per ml, and the yigR::Kan disruption was confirmed by Southern analysis. A representative strain was designated DM123.

β-Galactosidase assays.

Plasmid pR401 containing an aac(2′)-lacZ transcriptional fusion was described previously (33). β-Galactosidase assays were performed in triplicate with cell samples harvested at the early log phase, and activity was expressed in Miller units (26). Reported values represent the average for triplicate samples.

RNA analysis.

To examine aac(2′)-Ia mRNA levels in P. stuartii strains, cultures were grown in LB broth at 37°C to an A₆₀₀ of 0.2, and RNA was prepared with TRIazol reagent (Gibco/BRL). RNA was loaded in duplicate, fractionated on a 1% agarose gel containing 2.2 M formaldehyde, and transferred to a nylon membrane by capillary transfer. Filters were probed with a digoxigenin-labeled 602-bp TaqI-SspI fragment containing the aac(2′)-Ia coding sequence. As an internal control for loading, probes were “spiked” with a labeled fragment internal to the E. coli 23S rRNA coding sequence. Filters were developed with Lumi-Phos 530 (Boehringer Mannheim Biochemicals) and exposed to autoradiography film.

Ubiquinone analysis.

Cells were first grown in LB medium supplemented with 0.5% glucose in 2-liter flasks. The cultures were shaken overnight as starter cultures of 50 ml in 250-ml flasks. Cells were then inoculated into 500 ml of the same medium to an OD₆₀₀ of 0.05 and shaken at 37°C. Cells were harvested at an OD₆₀₀ of 2.0. Typically, 3 liters of culture was used for analysis. Cells were harvested, and pellets were washed twice in 50 mM potassium phosphate buffer and stored at −20°C. Quinone extraction was performed as described by Collins (10). Thawed cells, 5 g (wet weight), in 10 ml of phosphate buffer were broken by sonication at 1-min intervals, with 1 min of cooling in between the intervals, for 5 min. Lysis was confirmed by microscopic examination. Lysed cells were resuspended in 100 ml of acetone and left to digest for 12 h at 4°C with stirring. Cell debris was removed by filtration through Whatman no. 1 filter paper. The filtrate was then evaporated to 1 ml in a rotary evaporator at 40°C. The sample was freeze-dried, and the residue was dissolved in 2 ml of acetone. Samples (100 μl) were applied to a Silica Gel F₂₅₄ plastic-backed thin-layer chromatography plate (Merck item 5735), which was developed in hexane-diethyl ether (85:15, wt/vol). Coenzyme Q8 was used as a standard. The coenzyme Q8 spots were visualized by UV illumination. The spots were cut out, and quinones were eluted with 100% ethanol. The silica gel powder was removed by centrifugation, and the spectra of the clear supernatants were recorded with a Variant DMS-90 spectrophotometer.

Nucleotide sequence accession number.

The nucleotide sequence of aarF has been deposited in the EMBL/GenBank/DDBJ Nucleotide Sequence Data Library under accession no. AF002165.

RESULTS

Identification of the aarF locus.

PR50.AFM12 is a spontaneous gentamicin-resistant derivative of wild-type P. stuartii PR50. Gentamicin resistance in PR50.AFM12 was increased 256-fold (1,024 μg/ml) over that observed for wild-type PR50 (4 μg/ml). To determine whether aac(2′)-Ia expression was increased in PR50.AFM12, plasmid pR401, containing an aac(2′)-lacZ transcriptional fusion, was introduced. PR50.AFM12/pR401 formed dark blue colonies when grown on LB agar plates containing 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside. In contrast, isogenic strain PR50/pR401 formed white colonies when grown on the same plates. The mutant allele in PR50.AFM12 was therefore designated aarF1. The regulatory effects of aarF1 on aac(2′)-Ia expression were examined in further detail (see below).

PR50.AFM12 demonstrated a reduced growth rate compared to wild-type PR50 and formed significantly smaller colonies on LB agar plates. Because PR50.AFM12 was selected spontaneously, it seemed likely that a single mutation was responsible for both the increased gentamicin resistance and the reduced growth rate observed in this strain. Therefore, to complement the aarF1 mutation, a library of PR50 genomic DNA was constructed in pACYC184 and introduced into PR50.AFM12 (6, 24). Transformants that exhibited a wild-type growth rate were easily visible in the background of microcolonies. Plasmid DNA was purified from several large colonies and retransformed into PR50.AFM12, resulting in 100% of the transformants exhibiting a wild-type growth rate. Transformants forming large colonies also exhibited gentamicin resistance that was indistinguishable from that of wild-type P. stuartii (data not shown). Analysis of a complementing plasmid, pAFM12, indicated the presence of a 3.6-kb insert.

DNA sequence analysis.

A 1.9-kb SphI fragment from pAFM12 was subcloned into pBluescript SK(−), resulting in plasmid pSK.aarF. The introduction of pSK.aarF into pAFM12 also resulted in transformants exhibiting a wild-type growth rate (Fig. 1B). The nucleotide sequence of the 1,877-bp fragment in pSK.aarF was determined on both strands. A single open reading frame of 1,632 bp, predicted to encode a 544-amino-acid polypeptide (Fig. 1A), was identified. To determine whether this open reading frame encoded aarF, a chloramphenicol resistance cassette from pUT::mini-Tn5Cm (15) was inserted into a unique NruI site within this open reading frame. The resulting plasmid, pSK.aarF::Cm, was unable to restore a wild-type growth rate when introduced into PR50.AFM12 (Fig. 1B). Based on this result and the data presented below, this open reading frame has been designated aarF.

FIG. 1.

Identification of the aarF coding region. (A) Open reading frame map of the 1,883-bp fragment of P. stuartii DNA in pSK.aarF among all six possible reading frames. Lines extending halfway through the reading frame represent potential start codons, and lines extending completely through the reading frame represent stop codons. (B) Complementation of the slow growth phenotype of PR50.AFM12 by various constructs derived from pAFM12. Shaded regions represent the extent of the aarF coding region in each construct. +, restoration of wild-type growth rate; −, failure to restore wild-type growth rate.

Identification of the aarF gene product.

To determine whether the aarF locus encoded a polypeptide of the predicted size, the aarF gene was excised from pSK.aarF and subcloned into pBluescript KS(−) to enable transcription to be driven from the T7 promoter. The resulting plasmid, pKS.aarF, was transformed into the expression strain E. coli BL21(DE3)/pACYC184.lacI^q. A 67-kDa polypeptide observed upon induction with IPTG was not observed in an uninduced control culture (Fig. 2). No induction was observed in a control strain that was transformed with pBluescript KS(−). The size of the observed polypeptide correlated with the predicted size of 62.5 kDa. To confirm that the observed polypeptide was encoded by the aarF gene, a small insertion was introduced at a unique NheI site (Fig. 1B). This insertion resulted in a frameshift leading to a severe truncation of the putative AarF protein. When the resulting plasmid, designated pKS.NheI, was introduced into the E. coli expression strain, no induced polypeptide was observed upon the addition of IPTG (Fig. 2). The introduction of plasmid pKS.NheI into PR50.AFM12 also failed to restore a wild-type growth rate (Fig. 1B).

FIG. 2.

Identification of the aarF gene product. The expression strain BL21(DE3)/pACYC184.lacI^q contained one of the following plasmids: pBS.KS (control vector), pKS.aarF (intact aarF gene transcribed by the T7 promoter), or pKS. NheI (frameshift in aarF coding sequence). Strains were grown in LB broth and induced with IPTG as indicated. Total cellular protein was visualized by SDS-polyacrylamide gel electrophoresis followed by Coomassie blue staining. The size of the AarF polypeptide (67 kDa) was estimated by the relative mobility with respect to prestained low molecular-mass markers (Bio-Rad).

Cloning and analysis of the E. coli yigQR locus.

The predicted amino acid sequence of the aarF open reading frame exhibited a high degree of homology to two putative adjacent open reading frames of unknown function, present at 86 min on the E. coli chromosome, and designated yigQ (75% identity) and yigR (77% identity) (14). The amino acid alignment of AarF with YigQ and YigR is illustrated in Fig. 3. Because of the high degree of homology between AarF and the putative YigQ and YigR proteins, it was hypothesized that the yigQR locus may be able to functionally substitute for aarF in P. stuartii. To test this hypothesis, a partial Sau3AI library prepared from a wild-type E. coli strain was introduced into PR50.AFM12, and transformants forming wild-type-size colonies were selected. Plasmids from 11 individual large colonies were analyzed by restriction mapping, and all were shown to contain inserts with a common region of DNA (data not shown). One plasmid, designated pEF1, contained a 3.5-kb insert and was chosen for further study. A 2.6-kb SalI fragment from pEF1 was subcloned into pBluescript SK(−), creating pSK-2.6. The introduction of pSK-2.6 into PR50.AFM12 also restored a wild-type growth rate. To determine the identity of the cloned E. coli DNA in pSK-2.6, sequence data were obtained from both ends and was compared to the GenBank sequence databases. The cloned E. coli fragment in pSK-2.6 extends from the SalI site at nucleotide 72625 to the Sau3AI site at nucleotide 75101, as reported by Daniels et al. (14). This fragment contains the yigQ and yigR open reading frames as well as an upstream open reading frame, designated yigP (14). To determine which open reading frame(s) was required for the complementation of PR50.AFM12, a series of derivatives of pSK-2.6 were created as described in Materials and Methods. These derivatives were introduced into PR50.AFM12 and scored for complementation by the ability of the insert to restore a wild-type growth rate (Fig. 4). Plasmid pSK.yigQR, which has a deletion of yigP, retained the ability to complement the aarF1 mutation. In contrast, a deletion removing both the yigP open reading frame and the initiating ATG codon of the yigQ open reading frame (plasmid pSK.yigR) resulted in partial complementation of the growth defect in PR50.AFM12. Similarly, a deletion removing the C-terminal 96 amino acids of the yigR open reading frame (plasmid pSK.yigQRΔDraI) also resulted in partial complementation of PR50.AFM12. Finally, insertion of a kanamycin resistance cassette into a unique BsmI site internal to the yigR open reading frame (plasmid pSK.yigR::Kan) abolished the complementation of PR50.AFM12. These data indicate that yigP is not required for the complementation of PR50.AFM12, whereas both the yigQ and yigR open reading frames are required for the complementation of PR50.AFM12.

FIG. 3.

Homology between AarF and the YigQ and YigR proteins. Proteins were aligned with the Clustal V program (21). Identical amino acids are indicated by vertical bars; similar amino acids are indicated by colons. X, ambiguity in the reported amino acid sequence of YigQ (14).

FIG. 4.

Determination of sequences required for the complementation of PR50.AFM12. Complementation of the slow growth phenotype of PR50.AFM12 by various constructs derived from pSK-2.6 is shown. The positions of various restriction sites used to create the constructs are indicated. Shaded regions represent the extents of the yigP, yigQ, and yigR open reading frames present in each construct. +, restoration of wild-type growth rate; −, failure to restore wild-type growth rate; +/−, intermediate growth rate.

Analysis of chromosomal aarF and yigR null mutants.

A chromosomal aarF::Cm disruption in wild-type PR50 was constructed by allelic replacement as described in Materials and Methods, resulting in strain PR54. PR54 exhibited a slow growth phenotype similar to that of PR50.AFM12 (aarF1) and demonstrated resistance to gentamicin (1,024 μg/ml) that was equal to that of PR50.AFM12. The introduction of pSK.aarF into PR54 restored a wild-type growth rate, whereas the introduction of control plasmid pBluescript SK(−) did not affect either growth rate or gentamicin resistance (data not shown). Plasmid pSK-2.6, containing the wild-type yigPQR locus from E. coli, also restored a wild-type growth rate when introduced into PR54 (data not shown). The ability of the E. coli yigQR locus to substitute for aarF in PR54 suggested that the two loci are functionally equivalent. To examine the role of the yigQR locus in E. coli, the yigR::Kan disruption from pSK.yigR::Kan was introduced into the chromosome of wild-type E. coli RM1734 as described in Materials and Methods to produce DM123. DM123 exhibited a slow growth phenotype similar to that of the P. stuartii aarF::Cm mutant PR54. The introduction of either pSK-2.6 or pSK.aarF restored wild-type growth to DM123, whereas the introduction of the cloning vector pBluescript SK(−) did not affect the growth of DM123 (data not shown).

Effects of aarF::Cm on aac(2′)-Ia expression.

Preliminary data suggested that the aarF1 allele increased aac(2′)-Ia expression. To determine the phenotype of an aarF::Cm disruption, plasmid pR401 [aac(2′)-lacZ] was introduced into PR54. PR54/pR401 formed dark blue colonies when grown on LB agar plates containing 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside, whereas PR50/pR401 formed white colonies when grown on the same plates. In liquid assays, the accumulation of β-galactosidase in PR54/pR401 was measured at 19.9 ± 0.74 U and represented a 17-fold increase over the value for PR50/pR401 (1.14 ± 0.05 U). However, a control construct containing a lac_p-lacZ transcriptional fusion resulted in a 102-fold increase in β-galactosidase levels in PR54 relative to those observed in PR50 (data not shown). This result indicated that the aarF::Cm allele increased β-galactosidase expression or activity independently of the aac(2′)-Ia promoter. Preliminary data indicated that the copy number of pR401 is significantly higher in the aarF::Cm background (32). In addition, these data also suggested that aarF::Cm may actually decrease aac(2′)-Ia promoter activity. To confirm these results at the level of aac(2′)-Ia mRNA accumulation, RNA was prepared from PR54 and the parental strain, PR50, and was analyzed by Northern analysis with a probe specific to aac(2′)-Ia. The results shown in Fig. 5 demonstrated that aac(2′)-Ia mRNA levels were significantly lower in PR54 than in PR50.

FIG. 5.

Effects of aarF::Cm on aac(2′)-Ia expression. The accumulation of aac(2′)-Ia mRNA was determined by Northern analysis. Lanes: 1, 30 μg of RNA prepared from PR50 (wild type); 2, 30 μg of RNA prepared from PR54 (aarF::Cm). The filter was probed with the aac(2′)-Ia coding sequence (bottom panel). As an internal control for loading, the probe was spiked with a labeled fragment derived from the E. coli 23S rRNA gene (top panel). Arrows in the top panel denote the 23S rRNA. The arrow in the bottom panel denotes the aac(2′)-Ia message.

aarF and yigR mutants are deficient in ubiquinone.

Because aac(2′)-Ia mRNA levels in the aarF mutant background are not increased above wild-type levels, an alternative mechanism must be responsible for the high-level gentamicin resistance. Mutations that disrupt the aerobic respiratory electron transport chain have been reported to result in increased aminoglycoside resistance (2–5, 25, 27). One such class of mutants are those that are defective in ubiquinone biosynthesis (3, 4). Mutants that are defective in ubiquinone biosynthesis are unable to grow aerobically on nonfermentable carbon sources, such as malate or succinate (11). To determine whether PR54 was defective in ubiquinone production, the growth of PR50 and PR54 was compared aerobically on minimal media containing either glucose or succinate as the sole carbon source. Wild-type PR50 was able to utilize either glucose or succinate as a sole carbon source (Table 2). In contrast, PR54 was unable to utilize succinate as a sole carbon source aerobically. Similarly, E. coli RM1734 was able to utilize either glucose or succinate as a sole carbon source, whereas DM123 was unable to utilize succinate as a sole carbon source. The introduction of the complementating plasmids pSK.aarF and pSK-2.6 into PR54 and DM123, respectively, restored the ability to utilize succinate (Table 2).

TABLE 2.

Phenotypes resulting from aarF and yigR disruptions

Strain	Growtha on:
	Glucose	Succinate
P. stuartii
PR50 (wild type)	+	+
PR54 (aarF::Cm)	+	−
PR54/pBluescript SK(−)	+	−
PR54/pSK.aarF	+	+
PR54/pSK-2.6	+	+
E. coli
RM1734 (wild type)	+	+
DM123 (yigR::Kan)	+	−
DM123/pBluescript SK(−)	+	−
DM123/pSK.aarF	+	+
DM123/pSK-2.6	+	+
AN66 (ubiD)	+	±
AN66/pBluescript SK(−)	+	±
AN66/pSK.aarF	+	±
AN66/pSK-2.6	+	±
AN70 (ubiE)	+	−
AN70/pBluescript SK(−)	+	−
AN70/pSK.aarF	+	−
AN70/pSK-2.6	+	−
AN70/pEF1	+	+

^aStrains were grown on M9 minimal agar plates containing either 0.2% glucose or 0.5% succinate. E. coli AN66 and AN70 were supplemented with l-leucine, l-threonine, and l-methionine each at a final concentration of 0.2 mM and with thiamine at a final concentration of 0.02 μM. +, wild-type growth rate; ±, poor growth; −, no growth. 

Cell extracts were directly examined for ubiquinone content by thin-layer chromatography as described in Materials and Methods. Extracts from P. stuartii PR50 contained high levels of ubiquinone that comigrated with the coenzyme Q8 standard (R_f, 0.2). Extracts from PR54 contained no detectable ubiquinone but contained significant amounts of a precursor (R_f, 0.117). Analysis of the E. coli yigR::Kan mutant DM123 and the isogenic parental strain RM1734 yielded similar results. Thus, aarF in P. stuartii and yigQR in E. coli are required for the production of ubiquinone.

Three E. coli genes involved in ubiquinone biosynthesis (ubiB, ubiD, and ubiE) have been mapped to the same region on the chromosome as yigQR (min 86) (12, 14, 23, 39). The ubiB gene has been tentatively identified and lies approximately 4.7 kb downstream of yigR (14). Recently, the ubiE gene was demonstrated to be equivalent to an open reading frame, designated yigO, that lies immediately upstream of the yigP open reading frame (14, 23). E. coli AN66 ubiD and AN70 ubiE were transformed with pSK.aarF and pSK-2.6 (yigPQR) (12, 39). Neither E. coli strain was complemented by pSK.aarF or pSK-2.6 (yigPQR), as scored by the restoration of growth on minimal succinate plates (Table 2). In contrast, transformation with plasmid pEF1 restored the ability of AN70 to utilize succinate as a sole carbon source. Subsequent sequence analysis demonstrated that plasmid pEF1 contained the intact yigO open reading frame in addition to yigPQR (data not shown). Therefore, the aarF gene of P. stuartii and the yigQR genes of E. coli are distinct from ubiD and ubiE.

DISCUSSION

In a search for regulators of the aac(2′)-Ia gene of P. stuartii, we identified aarF, a gene required for the production of the respiratory cofactor ubiquinone (coenzyme Q). The aarF1 and aarF::Cm mutations resulted in a 256-fold increase in gentamicin resistance above wild-type levels and caused a severe defect in aerobic growth on rich media. Initial observations obtained through the use of an aac(2′)-lacZ transcriptional fusion suggested that aac(2′)-Ia expression was increased in the aarF mutant background. However, this expression appeared to be an artifact due to increased plasmid copy number in the aarF mutant background. Direct examination by Northern analysis revealed that aac(2′)-Ia mRNA levels in PR54 (aarF::Cm) were dramatically lower than those in wild-type PR50. To our knowledge, ubiquinone has never been implicated in gene regulation, so this finding is a novel one.

In light of the above data, it seems unlikely that ubiquinone is directly involved in the regulation of aac(2′)-Ia. We recently identified a locus in P. stuartii, designated aarE, that is also required for the expression of aac(2′)-Ia. The aarE gene was found to be the P. stuartii homolog of ubiA (32, 37). In contrast, the aarD locus (25), representing the P. stuartii homolog of cydD (30, 31), is required for the function of the cytochrome d terminal oxidase and is a negative regulator of aac(2′)-Ia expression. We propose a model for the regulation of aac(2′)-Ia expression by a regulatory cascade in which ubiquinone acts as an effector molecule (Fig. 6). In this model, the reduced form of ubiquinone (ubiquinol) serves as a signal to activate aac(2′)-Ia expression through an uncharacterized pathway. It is important to note that this pathway appears to be independent of the previously identified activator AarP (24). According to this model, in a cytochrome d-deficient strain, ubiquinol is predicted to accumulate and to result in the activation of aac(2′)-Ia expression. In ubiquinone-deficient aarE and aarF mutant strains, the regulatory cascade would be disrupted, resulting in decreased aac(2′)-Ia expression. Further studies with inhibitors of electron transport may provide evidence to support this model.

FIG. 6.

Model for the regulation of aac(2′)-Ia expression. A schematic of the aerobic respiratory chain is depicted. A primary dehydrogenase couples the oxidation of a substrate (SH₂) and the reduction of ubiquinone (Q). A terminal oxidase complex oxidizes ubiquinol (QH₂) and reduces molecular oxygen to water, coupling the reaction to the generation of a proton gradient. The black box represents the putative sensor component of a regulatory cascade that responds to the levels of ubiquinol and increases aac(2′)-Ia expression.

Because aac(2′)-Ia mRNA levels are decreased in PR50.AFM12 (aarF1) and PR54 (aarF::Cm), an alternative mechanism must be responsible for the large increase in gentamicin resistance observed in the aarF mutant background. Aminoglycoside uptake requires a sufficiently negative membrane potential as well as active electron transport (3, 13, 17). High-level, nonenzymatic resistance to the aminoglycosides usually arises from mutations in components of the aerobic respiratory chain (1, 3, 4, 13, 19). PR54 was unable to utilize succinate as a sole carbon source aerobically, and extracts from this strain were found to be devoid of ubiquinone. Ubiquinone-deficient E. coli mutants were previously shown to exhibit increased gentamicin resistance and were found to accumulate gentamicin poorly (3, 4). Therefore, the high-level gentamicin resistance observed in PR54 is likely associated with decreased accumulation of the drug resulting from the absence of aerobic electron transport.

The aarF locus was found to encode a single 544-amino-acid protein. The AarF polypeptide was identified with a T7 expression system and exhibited an apparent molecular mass of 67 kDa, in agreement with the predicted size of 62.5 kDa. The predicted AarF protein exhibited extensive amino acid identity with the products of two putative adjacent open reading frames, yigQ and yigR, present at 86.6 min on the E. coli chromosome (14). This region of the chromosome has been sequenced as part of the E. coli sequencing project, and frameshifts that could merge yigQ and yigR into one contiguous open reading frame are possible. It should be noted that we have no evidence suggesting that yigQ and yigR are contiguous.

An E. coli yigR::Kan mutant was found to be defective in ubiquinone biosynthesis. Three ubiquinone biosynthesis genes, ubiB, ubiD, and ubiE, map near yigQ and yigR at min 86 on the E. coli chromosome (12, 14, 23, 39). Complementation studies showed that the yigQR genes did not complement ubiD and ubiE mutations. In addition, the ubiB gene lies upstream of yigQR (14). Therefore, aarF (yigQR) represents a novel gene in the ubiquinone biosynthetic pathway. Extracts from both P. stuartii aarF and E. coli yigR mutants contained significant amounts of a ubiquinone precursor. Future studies to determine the identity of this precursor will be required to assign a function to the aarF (yigQR) locus.