Acinetobacter baylyi Starvation-Induced Genes Identified through Incubation in Long-Term Stationary Phase

Abstract

Acinetobacter species encounter cycles of feast and famine in nature. We show that populations of Acinetobacter baylyi strain ADP1 remain dynamic for 6 weeks in batch culture. We created a library of lacZ reporters inserted into SalI sites in the genome and then isolated 30 genes with lacZ insertions whose expression was induced by starvation during long-term stationary phase compared with their expression during exponential growth. The genes encode metabolic, gene expression, DNA maintenance, envelope, and conserved hypothetical proteins.

Acinetobacter species are ubiquitous soil organisms. Starvation during long-term stationary phase (LTSP) can serve as a laboratory model for natural competitive conditions such as those found in soils (4). This model has been used to study Escherichia coli, and here, we have applied it to Acinetobacter baylyi strain ADP1 (8).

During long-term batch culture, an initially clonal population of Escherichia coli experiences five growth stages: lag, exponential, and stationary phases and then death phase and LTSP (4). Prior to LTSP, most of the cells die and serve as nutrition for starving survivors (6, 13). In LTSP, the cell population remains almost steady, declining slowly over years (reviewed in reference 4): for each newly dead cell, slightly less than one new cell is “born.”

Much of what is known about starvation physiology during LTSP has been determined through study of the growth advantage in stationary phase (GASP) phenotype. The phenotype arises from genetic changes that occur when cells experience LTSP. During LTSP, the population may have a mutation frequency approaching 1 in 600 base pairs per genome (5).

Some physiological changes that take place during LTSP have been described, as have some genes necessary for the development of GASP (13, reviewed in reference 12). Some mutant strains that exhibit GASP have mutations that enhance catabolic efficiency for processing amino acids (14-16). Another nutrient consumed is DNA, which requires genes homologous to strain ADP1's competence genes (6). Additionally, mutations that knock out SOS polymerases interfere with the formation of GASP mutants (11).

Growth of Acinetobacter baylyi.

One of our aims was to ascertain whether the phenomenon of LTSP occurs in A. baylyi. We cultured A. baylyi in aerated minimal succinate (0.01 M) broth (7) for 6 weeks at 37°C and measured the CFU ml⁻¹ over that period (Fig. 1). Death phase began after about 24 h postinoculation, while LTSP began at 48 h postinoculation and persisted for at least five more weeks.

FIG. 1.

Growth of Acinetobacter baylyi ADP1 in minimal medium with 0.01 M succinate (7) over 6 weeks. Data are from a sample representative of three trials.

Identification of starvation-induced genes.

Our second aim was to identify ADP1 genes induced by starvation during LTSP. To do this, we followed the method of Chakravorty et al. (2), digesting ADP1 chromosomal DNA with SalI and ligating the DNA to a lacZ-kan cassette from plasmid pKOK6, resulting in circular DNA that could enter the ADP1 genome by a single-crossover event following natural transformation (see the supplemental material). We used a scaled-down version of the Miller method (9) in 96-well plates to screen 3,569 kanamycin-resistant colonies of ADP1 from six different ligation pools for mutants with higher levels of β-galactosidase activity during LTSP (48 h after inoculation) than during exponential phase (6 h after inoculation). There are 305 SalI sites in each genome (1), and most of the lacZ insertions were isolated from more than one independent pool, so the screen is near saturation.

One hundred twenty-three isolates were subjected to secondary screening and exhibited greater induction of β-galactosidase activity at 7 days postinoculation than during exponential phase. Examination of the strains using inverse PCR based on the sequence of the kanamycin cassette revealed 30 unique starvation-induced (STI) lacZ insertion sites (see the supplemental material).

The procedure used to introduce the lacZ-kan cassette into each strain should have resulted in merodiploid DNA caused by a single recombination event between the chromosome and a closed circle of DNA created by ligating SalI chromosomal fragments to SalI-digested cassette DNA. Using PCR, we detected an intact target open reading frame (ORF) in every strain. The insertion of the cassette may have resulted in chromosomal rearrangements, so we also investigated the chromosome structure surrounding the cassette. We used PCR to attempt to amplify the sequence between the DNA adjacent to the 5′ end of the lacZ-kan cassette and the lacZ gene and the sequence between the DNA adjacent to the 3′ end of the lacZ-kan cassette and the kan gene. Twenty-seven strains were merodiploids without any obvious nearby rearrangements. Three strains (AGCV5, AGCV7, and AGCV11) did not produce the expected PCR products, suggesting that they may have chromosomal rearrangements.

Expression of long-term stationary-phase genes.

All 30 STI genes were induced at 7 days postinoculation compared with their expression during exponential growth (Fig. 2B; also see Table 2). When we moved the kanamycin-linked lacZ reporters into a fresh ADP1 strain through natural transformation using boiled cell extracts as DNA donors and selection for kanamycin resistance, the reporter genes of all 30 new reporter strains were still induced in LTSP (data not shown).

FIG. 2.

(A) Growth of four representative lacZ reporter strains in minimal medium with 0.01 M succinate (7). Squares, ostA::lacZ strain; diamonds, pilU::lacZ strain; triangles, argS::lacZ strain; circles, tolB::lacZ strain. (B) Induction of β-galactosidase in four representative lacZ reporter strains expressed as Miller units (9) (filled symbols, solid lines, left y axis) or nMol o-nitrophenol (ONP) per 30 min of incubation per 10⁶ CFU (open symbols, dotted lines, right y axis). Squares, ostA::lacZ strain; diamonds, pilU::lacZ strain; triangles, argS::lacZ strain; circles, tolB::lacZ strain. Old populations of cells are turbidometrically dominated by “dead” cells that are not transcriptionally active (detected using LIVE/DEAD BacLight bacterial viability kit from Molecular Probes [Invitrogen, Carlsbad, CA], which detects membrane potential; data not shown), prompting the use of nMol ONP/30 min/10⁶ CFU as an alternative measure of β-galactosidase activity.

TABLE 2.

Expression of STI gene-lacZ fusions after 7 days of incubation in batch culture

Strain	Functional group	Gene interrupted by lacZ fusion and/or its product or function	LacZ activitya (SD)
AGCV19	Envelope	ostA	2,389 (890)
AGCV23	Gene expression	trxAb	1,534 (222)
AGCV12	DNA metabolism	pilU	1,025 (388)
AGCV1	Metabolism	argS	1,014 (63)
AGCV28	Anabolism	gshA	808 (218)
AGCV7	Anabolism	leuA	718 (87)
AGCV18	Anabolismc	kdsB	682 (54)
AGCV25	Unknown	ACIAD3229, CHP	652 (126)
AGCV13	Metabolism	purA	608 (271)
AGCV10	Unknown	ACIAD0865, CHP	556 (346)
AGCV26	Unknown	ACIAD3343, CHP	427 (54)
AGCV8	Unknown	ACIAD0615, CHP	417 (15)
AGCV21	Unknown	ACIAD2746 but reversed,d CHP	390 (93)
AGCV2	Unknown	ACIAD0167, CHP	367 (48)
AGCV4	Gene expression	infB	359 (54)
AGCV14	DNA metabolism	Putative methylase	282 (131)
AGCV11	Gene expression	fusA	255 (132)
AGCV15	Catabolism	mdcA	233 (131)
AGCV5	Envelope	phaAB	225 (29)
AGCV20	Envelope	tolB	224 (15)
AGCV29	Envelope	Putative outer membrane protein	224 (15)
AGCV22	Gene expression	Putative RNA-binding protein	207 (62)
AGCV27	Anabolism	aroK	173 (25)
AGCV9	DNA metabolism	Putative mutT	157 (25)
AGCV3	Catabolism	atpA	133 (34)
AGCV17	Unknownb	ACIAD1960, CHP	92 (21)
AGCV24	Catabolism	almA	56 (3)
AGCV16	Catabolism	Putative enoyl-coenzyme A hydratase	50 (14)
AGCV6	DNA metabolism	recD	39 (14)
AGCV30	Gene expression	23S rRNA	11 (3)

^anmol o-nitrophenol/30 min/10⁶ CFU; see the supplemental material.

^blacZ fusion is in the same orientation as trxA but is inserted after the stop codon, before the next gene (rho).

^cSurrounding genes that could be part of an operon with kdsB or ACIAD1960 suggest a role in transport.

^dlacZ gene is inserted opposite the direction of the open reading frame.

To further increase confidence in the identification of STI genes, we measured the population growth of null mutants (3) lacking one of six STI genes (ACIAD0167, ACIAD0615, ACIAD1960, ACIAD3229, or ACIAD3343, encoding conserved hypothetical proteins [CHPs], or pilU) or a control gene also encoding a CHP (ACIAD2370) but adjacent to and carried on the opposite strand relative to one of the STI genes. All of the strains grew normally in minimal broth Davis (BD, Franklin Lakes, NJ) supplemented with 0.01 M succinate for the first 48 h of incubation at 37°C with aeration (P > 0.1; unpaired, two-tailed Student's t test). All six strains except the ACIAD2370::kan strain, the negative control, showed a significant decrease in survival at 7 days (P < 0.05; data not shown).

Bioinformatic analysis of STI genes.

Twenty-eight of the STI lacZ reporters were inserted in an open reading frame (Table 1). Seven of these genes (23%) encode CHPs (3). Twenty other STI genes have annotations that rely on limited similarity, conserved amino acid motifs, or the function of homologous genes studied in a distant relative. The only STI gene in Acinetobacter with an experimentally determined phenotype is almA, which is required for the degradation of long-chain N-alkanes (10).

TABLE 1.

Bacterial strains isolated in this study

Strain	Gene interrupted by lacZ fusion and/or its product or function	Locus tag (prefix, ACIAD)	Start codona	SalIb
AGCV1	argS (arginyl-tRNA synthetasec	0164	164048	163045
AGCV2	CHPd	0167	166880	168135
AGCV3	atpA; ATP synthase F1 α-subunitc	0185	181338	181779
AGCV4	infB; translation initiation factor IF-2c	0369	360876	362625
AGCV5	phaABe; pH adaption K efflux systemc	0376	367423	368026
AGCV6	recD; helicase involved in DNA repairc	0399	397904	398700
AGCV7	leuAe; 2-isopropylmalate synthasec	0530	516254	515403
AGCV8	CHP	0615	606821	606613
AGCV9	Putative MutT/Nudix proteinf	0673	668885	668708
AGCV10	CHP	0865	851697	852072
AGCV11	fusAe; protein chain elongation factorc	0884	866071	867823
AGCV12	PilUc	0911	893365	892531
AGCV13	furA; adenylosuccinate synthetasec	1258	1256743	1256803
AGCV14	Putative N6 adenine methylasef	1268	1270305	1268926
AGCV15	mdcA; malonate decarboxylasec	1753	1759599	1759642
AGCV16	Putative enoyl-coenzyme A hydratasef	1829	1836465	1837011
AGCV17	CHP	1960	1947934	1947270
AGCV18	kdsB; lipopolysaccharide synthesisc	2362	2321280	2320904
AGCV19	ostA; organic solvent tolerance outer membrane proteinc	2371	2328576	2328693
AGCV20	TolB; outer membrane proteinc	2622	2575393	2576592
AGCV21	CHP but reversedg	2746	2688538	2687944
AGCV22	Putative RNA-binding proteinf	2855	2792708	2793034
AGCV23	trxAc,h	3037-3038	2967985	2968468
AGCV24	almA; monooxygenasei	3192	3121361	3121180
AGCV25	CHP	3229	3142675	3142777
AGCV26	CHP	3343	3248853	3248995
AGCV27	AroKc	3354	3262302	3262221
AGCV28	gshA; γ-glutamate-cysteine ligasec	3549	3475214	3474012
AGCV29	Putative outer membrane proteinf	3675	3588360	3587641
AGCV30	23S rRNAj	Unknown	Unknown	Unknown

^aFirst nucleotide of annotated start codon (2).

^bFirst nucleotide of SalI site in which lacZ fusion is inserted.

^cFunction of homologous gene experimentally demonstrated in another organism.

^dCHP, conserved hypothetical protein. Homolog of previously reported genes of unknown function.

^eMay have a chromosomal rearrangement.

^fFunction proposed based on presence of conserved amino acid motif, structural feature, or limited homology.

^glacZ gene is inserted opposite the direction of the open reading frame.

^hlacZ gene is inserted in the intergenic region between the stop codon of trxA (ACIAD3037) and the start codon of rho (ACIAD3038).

ⁱFunction experimentally demonstrated in the genus studied (10).

^jThere are seven copies of this operon in strain ADP1.

The expression of STI genes suggests that ADP1 cells in LTSP are metabolically active. We found specific translation-associated genes, such as fusA and infB, to be starvation induced (Tables 1 and 2). Other STI genes encode anabolic proteins, such as those involved in amino acid synthesis (leuA and gshA), cell wall synthesis (kdsB), or catabolism (mdcA and almA). Three of the STI genes are annotated as encoding DNA maintenance, recombination, or competence proteins (Tables 1 and 2), suggesting that, as in E. coli, control of DNA is important during LTSP (6, 11). A fourth group of A. baylyi STI genes encodes proteins that are associated with the bacterial envelope (Table 2).

Here, we have reported that strain ADP1 populations persist in LTSP. We have identified 30 genes that are induced in LTSP compared with their expression during exponential phase. The induced genes encode proteins needed for catabolism, anabolism (including protein synthesis), and control of DNA. Twenty-three percent of the induced genes encode CHPs, suggesting that these proteins should no longer be considered hypothetical. We conclude that ADP1 cells, like E. coli cells, are active during LTSP. Further scrutiny of ADP1's physiology during this model of feast-famine could help reveal how Acinetobacter species survive in highly competitive conditions such as those found in soils.