Abstract

Objectives

The whole genomes of two Acinetobacter baumannii isolates recovered from a single patient were sequenced to gain insight into the nature and extent of genomic plasticity in this important nosocomial pathogen over the course of a short infection. The first, AB210, was recovered before tigecycline therapy and was susceptible to this agent; the second, AB211, was recovered after therapy and was resistant.

Methods

DNA from AB210 was sequenced by 454 GS FLX pyrosequencing according to the standard protocol for whole-genome shotgun sequencing, producing ∼250 bp fragment reads. AB211 was shotgun sequenced using the Illumina Genetic Analyzer to produce fragment reads of exactly 36 bp. Single nucleotide polymorphisms (SNPs) and large deletions detected in AB211 in relation to AB210 were confirmed by PCR and DNA sequencing.

Results

Automated gene prediction detected 3850 putative coding sequences (CDSs). Sequence analysis demonstrated the presence of plasmids pAB0057 and pACICU2 in both isolates. Eighteen putative SNPs were detected between the pre- and post-therapy isolates, AB210 and AB211. Three contigs in AB210 were not covered by reads in AB211, representing three deletions of ∼15, 44 and 17 kb.

Conclusions

This study demonstrates that significant differences were detectable between two bacterial isolates recovered 1 week apart from the same patient, and reveals the potential of whole-genome sequencing as a tool for elucidating the processes responsible for changes in antibiotic susceptibility profiles.

Introduction

Acinetobacter baumannii is an important nosocomial pathogen, with multidrug-resistant (MDR) and even pan-drug-resistant strains reported worldwide.1 In the UK, carbapenem-resistant clonal lineages limit available treatment options. One successful lineage, designated OXA-23 clone 1, belonging to European clone II, has been recovered from more than 60 hospitals, clustered mainly in London and South-East England.2 Representative isolates of this clone are usually susceptible to colistin and tigecycline only. We previously reported the emergence of tigecycline resistance during antibiotic therapy in the OXA-23 clone 1 epidemic lineage, and showed that up-regulation of the resistance–nodulation–division (RND) efflux system, AdeABC was responsible for the resistance phenotype.3

The recent availability of rapid and inexpensive whole-genome sequencing permits detailed investigation of genetic differences between pairs of bacterial isolates. In A. baumannii, whole-genome studies have thus far focused either on comparing distinct antibiotic-susceptible and MDR strains4,5 or related isolates from different patients.6 The results of these and other similar studies7 point to a high degree of genome plasticity, the rapid emergence of antibiotic resistance and significant genetic differences between closely related isolates.

Tigecycline is used as a treatment of last resort for MDR A. baumannii infection, despite a lack of formal trial data, and the emergence of resistance is a major concern. We sequenced the genomes of two A. baumannii isolates from a single patient, the first recovered before tigecycline therapy and susceptible to this agent, the second after 1 week of therapy for an intra-abdominal infection and resistant. The study aimed to gain insight into the nature and extent of genomic plasticity over the course of a short infection.

Materials and methods

Bacterial isolates

Clinical isolates AB210 and AB211 have been described previously.3 As OXA-23 clone 1 representatives, they belong to the globally successful European clone II group, and were assigned to group 1 by the multiplex PCR method described by Turton et al.8 They were typed by PFGE of ApaI-digested genomic DNA (Figure 1), as described previously,2 and the presence of blaOXA-23-like was confirmed by multiplex PCR.9

Figure 1.

PFGE profiles of AB210 (lane 2) and AB211 (lane 3).

Figure 1.

PFGE profiles of AB210 (lane 2) and AB211 (lane 3).

Antimicrobial susceptibility testing and DNA manipulations

MICs were determined by BSAC agar dilution or Etest (AB bioMérieux, Solna, Sweden) on Iso-Sensitest agar (Oxoid, Basingstoke, UK) with the results interpreted according to BSAC guidelines.9 Genomic DNA was extracted with the Wizard Genomic DNA Purification Kit (Promega, Southampton, UK) and was used as the template for DNA sequencing. Plasmids were isolated from AB210 and AB211 using the PureYield Plasmid Miniprep System (Promega) and analysed by agarose gel electrophoresis.

Whole-genome DNA sequencing and data analysis

DNA from AB210 was sequenced by 454 GS FLX pyrosequencing (Roche, Branford, CT, USA) according to the standard protocol for whole-genome shotgun sequencing, producing ∼250 bp fragment reads. AB211 was shotgun sequenced using the Illumina Genetic Analyzer (Illumina, Saffron Walden, UK) to produce fragment reads of exactly 36 bp. All sequencing was performed at GATC Biotech Ltd (Constance, Germany). A draft genome assembly for AB210 was produced from flowgram data using Newbler 2.5 (Roche). The Newbler command-line option ‘-rip’ was used to ensure reads were aligned to single contigs only. The resulting contigs were annotated by reference to the related strain A. baumannii ACICU10 (also belonging to European clone II) using the automated annotation pipeline on the xBASE server.11

Illumina reads for isolate AB211 were mapped against the draft AB210 assembly using Bowtie 0.12.0.12 For the purposes of single nucleotide polymorphism (SNP) detection, Bowtie was run with parameter ‘-m 0’ to suppress alignments that map equally to multiple locations in the genome. To detect deletions, this setting was not used. A consensus pileup was produced using SAMtools,13 and putative SNPs were called using Varscan 2.214 with the following parameters: minimum coverage (10), min-reads2 (2), min-avg-qual (15), min-var-freq (0.9). To detect microindels (insertion or deletion events) less than 3 bases long, AB211 reads were additionally mapped using Novoalign 2.5.15 Whole-genome alignments were visualized; SNPs and deletions were manually inspected using the output files from the above steps using BAMview.16

Confirmation of SNPs and chromosomal deletions

SNPs and deletions detected in AB211 in relation to AB210 were confirmed by PCR and DNA sequencing using the primers listed in Table S1 (available as Supplementary data at JAC Online). Nucleotide sequences of the resulting amplicons were determined with an ABI 3730xl DNA analyser (Applied Biosystems, Warrington, UK).

Results and discussion

Antibiotic susceptibilities

MICs of tigecycline, tobramycin, amikacin, gentamicin and azithromycin for the pre-therapy isolate AB210 were 0.5, >32, >64, >32 and >256 mg/L, respectively, while MICs for the post-therapy isolate AB211 were 16, 2, 4, 8 and >256 mg/L, respectively.

Sequencing results

Sequencing produced >128 million and >156 million nucleotide bases for AB210 and AB211, respectively. The assembly of AB210 resulted in 91 contigs larger than 500 bp, comprising 4.06 megabases of sequence and representing a median 29-fold coverage. Automated gene prediction detected 3850 putative coding sequences (CDSs), of which 3504 were homologous (defined as BLASTP e-value ≤1e-05) to a sequence in the reference genome of A. baumannii ACICU. The vast majority (96.6%) of the AB211 reads mapped to a region on the AB210 genome. The AB210 draft assembly has been deposited in GenBank (accession number AEOX00000000), and raw sequence reads for AB210 and AB211 have been submitted to NCBI's Sequence Read Archive under study accession number SRP004860.

Plasmid profile

Plasmid profiles of AB210 and AB211 were identical and showed the presence of two plasmids in each isolate (data not shown). Sequence analysis demonstrated the presence of a 9 kb contig in AB210 that displayed 99.98% identity to the previously characterized pAB0057 plasmid.5 This was seen at high sequence read coverage in both AB210 and AB211, suggesting it was present as multiple copies. Three other contigs, totalling 65 kb, were seen at below-average coverage; taken together these were a full match in length and nucleotide identity to the complete pACICU2 plasmid.10

AB210 virulence genes and resistance islands (RIs)

RIs have been detected in all sequenced A. baumannii genomes containing multiple resistance determinants. They are composite transposons that are complex in nature and have been designated AbaR (A. baumanniiresistance).4 They share a common insertion site (comM) but vary considerably among isolates in terms of the exact genetic composition, with that from ACICU, a representative of European clone II, being considerably reduced in size compared with those found in representatives of European clone I.10,17 Clinical isolates AB210 and AB211 were found to contain an AbaR-type RI. In the former isolate (GenBank accession number HQ700358) this was shown to contain sequence corresponding to nucleotides 587330–599047 of strain AB0057 (GenBank accession number CP001182), with a 2.85 kb section absent; this is an AbaR4-type island and contains blaOXA-23.

SNPs between AB210 and AB211

Eighteen putative SNPs were detected between the pre- and post-therapy isolates. Only one of these was located outside of coding regions at −35 bp upstream of ureJ, which encodes a hydrogenase/urease accessory protein (AB210 locus tag: AB210-1_2203). The location of this SNP suggests the possibility of regulatory significance, although ureJ appears to be part of a urease gene cluster that is co-transcribed as an operon in other species.18 Of the remaining 17, 8 were synonymous mutations, whereas 9 were non-synonymous, including one missense mutation (Table 1). Seventeen (94%) of the SNPs were transitions. Eight of the 9 non-synonymous SNPs could be confirmed by PCR and sequencing, while one was not validated (Table 1 and Table S1). Several of these were located within genes predicted to be involved in core biological functions, including translation (dusB), nucleic acid biosynthesis, α-ketoglutarate and arabinose transport, environmental sensing (the signal transduction histidine kinase gene adeS, which had previously been identified through a candidate gene approach3) and signalling. The mutation in adeS is believed to be responsible for up-regulation of the AdeABC efflux system, and hence tigecycline resistance. Two SNPs were located within a gene coding for a GGDEF domain-containing protein, one of which was a non-synonymous mutation while the other introduced an internal stop codon, thus giving rise to a truncated product (Table 1). These proteins are enzymes that catalyse the synthesis of cyclic-di-GMP, which has been recognized recently as an important second messenger in bacteria and is implicated in adhesin and extrapolysaccharide biosynthesis.19

Table 1.

Confirmed SNPs identified in clinical isolate AB211 resulting in amino acid substitution or termination

    Amino acid identity
 
 
SNP Position in AB210 assembly Locus tag in AB210 assembly Protein product AB210 AB211 Orthologue 
159509 AB210-1_0138 tRNA-dihydrouridine synthase, DusB ABAYE0965 
639321 AB210-1_0587 nucleoside-diphosphate-sugar epimerase ACICU_01645 
755474 AB210-1_0703 major facilitator superfamily permease ACICU_01760 
1469178 AB210-1_1405 hypothetical protein ACICU_02205 
2548057 AB210-1_2423 major facilitator superfamily permease ACICU_01217 
2852737 AB210-1_2721 signal transduction histidine kinase, AdeS ACICU_01827 
3362158 AB210-1_3207 GGDEF domain-containing protein ACICU_03492 
3362175 AB210-1_3207 GGDEF domain-containing protein ACICU_03492 
    Amino acid identity
 
 
SNP Position in AB210 assembly Locus tag in AB210 assembly Protein product AB210 AB211 Orthologue 
159509 AB210-1_0138 tRNA-dihydrouridine synthase, DusB ABAYE0965 
639321 AB210-1_0587 nucleoside-diphosphate-sugar epimerase ACICU_01645 
755474 AB210-1_0703 major facilitator superfamily permease ACICU_01760 
1469178 AB210-1_1405 hypothetical protein ACICU_02205 
2548057 AB210-1_2423 major facilitator superfamily permease ACICU_01217 
2852737 AB210-1_2721 signal transduction histidine kinase, AdeS ACICU_01827 
3362158 AB210-1_3207 GGDEF domain-containing protein ACICU_03492 
3362175 AB210-1_3207 GGDEF domain-containing protein ACICU_03492 

The asterisk indicates a stop codon.

Large structural changes in the genomes of AB210 and AB211

Three contigs in AB210 were not covered by reads in AB211, these putative deletions were designated ROD1, 2 and 3. The first, ROD1, was ∼15 kb in length. This deletion disrupted the coding sequence of the DNA mismatch repair gene mutS (AB210-1_2445) by eliminating the N-terminal mutS-I domain. Aside from encoding this mismatch recognition enzyme, ROD1 also encoded a drug/metabolite transporter (DMT) superfamily permease (AB210-1_2447) and a major facilitator superfamily (MFS) permease (AB210-1_2451), transcriptional regulators (AB210-1_2450; AB210-1_2453) and an EAL domain-containing protein (AB210-1_2448) responsible for the degradation of cyclic-di-GMP.19 At ∼44 kb, ROD2 was the largest deleted region and comprises genes encoding for transcriptional regulators (AB210-1_3253; AB210-1_3262; AB210-1_3269; AB210-1_3273), ion channels and transporters [AB210-1_3254; AB210-1_3259; (AB210-1_3275; AB210-1_3276; AB210-1_3277)], a class A β-lactamase enzyme (AB210-1_3248) and components of a type VI secretion system (AB210-1_3280; AB210-1_3281).20 Interestingly, part of the type VI secretion locus was missing even in AB210, suggesting that this was a degenerate system in both isolates. ROD1 and ROD2 are contiguous in A. baumannii ACICU, suggesting this may be a single deletion, but this could not be confirmed experimentally for AB210 by PCR (data not shown). ROD3, ∼17 kb in length, included a class 1 integron containing antibiotic resistance genes including macrolide resistance determinants [AB210-1_3691 (phosphotransferase); AB210-1_3692 (an efflux protein)] and several genes encoding aminoglycoside resistance determinants, namely aac(6′)-Ib (AB210-1_3701), two copies of aadA (AB210-1_3699; AB210-1_3700) and armA (AB210-1_3695), which encodes a 16S rRNA methylase.

Implications for Acinetobacter evolution

The extent of genomic changes detected here are consistent with the marked changes in phenotype, particularly the loss of aminoglycoside resistance in AB211. However, we were unable to determine whether these changes were the result of rapid evolution during the course of infection and treatment, or whether the patient initially had a mixed infection (or re-infection) involving different variants of the same defined clone, with subsequent selection for tigecycline resistance.

The disruption of mutS, an important DNA mismatch repair gene, is significant and suggests the possibility of a hypermutator phenotype, which may have contributed to the relatively large number of SNPs. Previous work in Acinetobacter sp. ADP1 has shown that mutS preferentially recognizes and repairs transitions,21 so its disruption in AB211 is consistent with our observation that 94% of the SNPs belonged to this class.

The absence of ROD3 is consistent with the change in aminoglycoside resistance between AB210 and AB211, with MICs of tobramycin, amikacin and gentamicin reduced at least 8-fold in AB211. The high azithromycin MICs reflect the intrinsic non-susceptibility of A. baumannii to this agent. It is notable that the development of tigecycline resistance was accompanied by increased susceptibility to other antibiotics through a large genomic deletion.

GGDEF- and EAL-containing proteins have been implicated in sessile to planktonic shifts. Taken together, the termination in a GGDEF domain-containing protein as well as the loss of an EAL domain-containing protein in ROD1 may be advantageous during the process of infection, although this remains to be experimentally determined.

In this study, whole-genome sequencing gave insight into the nature of genetic changes between isolates under selection pressure through antibiotic therapy and a hostile host environment. This study has demonstrated significant differences between two A. baumannii isolates belonging to the same epidemic lineage collected 1 week apart from the same patient. Such studies are able to shed light on the relative importance of SNPs and transposon mutagenesis on the evolution of A. baumannii and can generate hypotheses into the nature of antibiotic resistance and virulence. Although further studies are needed to assess the extent of genetic diversity among populations of A. baumannii in a single patient, we clearly demonstrated the potential of whole-genome sequencing as an important tool for helping elucidate the evolutionary processes responsible for the rapid development of antibiotic resistance in this important nosocomial pathogen.

Funding

This work was supported by an educational grant from Wyeth, now taken over by Pfizer. The xBASE facility and N. L.'s position are funded by BBSRC grant BBE0111791.

Transparency declarations

D. M. L. has received research grants from Wyeth and Pfizer, spoken at meetings organized by Wyeth and Pfizer, and received sponsorship to travel to congresses from Wyeth and Pfizer, as well as from numerous other pharmaceutical and diagnostic companies. He holds shares in GlaxoSmithKline, Merck, AstraZeneca, Dechra and Pfizer; he also acts as an executor, managing further holdings in GlaxoSmithKline and EcoAnimal Health. N. W. has received research grants from Wyeth. M. J. E., J. F. T., A. U., T. G., M. D., D. M. L. and N. W. are employees of the HPA and are influenced by its views on antibiotic use and prescribing. M. H., D. W. W. and C. P. T. have received sponsorship to attend conferences from Wyeth. N. L. and M. J. P.: none to declare.

Supplementary data

Table S1 is available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/).

Acknowledgements

We wish to thank Anthony Haines (University of Birmingham) for advice on bioinformatic analysis.

References

1
Towner
KJ
Acinetobacter: an old friend, but a new enemy
J Hosp Infect
 , 
2009
, vol. 
73
 (pg. 
355
-
63
)
2
Coelho
JM
Turton
JF
Kaufmann
ME
, et al.  . 
Occurrence of carbapenem-resistant Acinetobacter baumannii clones at multiple hospitals in London and southeast England
J Clin Microbiol
 , 
2006
, vol. 
44
 (pg. 
3623
-
7
)
3
Hornsey
M
Ellington
MJ
Doumith
M
, et al.  . 
AdeABC-mediated efflux and tigecycline MICs for epidemic clones of Acinetobacter baumannii
J Antimicrob Chemother
 , 
2010
, vol. 
65
 (pg. 
1589
-
93
)
4
Fournier
PE
Vallenet
D
Barbe
V
, et al.  . 
Comparative genomics of multidrug resistance in Acinetobacter baumannii
PLoS Genet
 , 
2006
, vol. 
2
 pg. 
e7
 
5
Adams
MD
Goglin
K
Molyneaux
N
, et al.  . 
Comparative genome sequence analysis of multidrug-resistant Acinetobacter baumannii
J Bacteriol
 , 
2008
, vol. 
190
 (pg. 
8053
-
64
)
6
Adams
MD
Chan
ER
Molyneaux
ND
, et al.  . 
Genomewide analysis of divergence of antibiotic resistance determinants in closely related isolates of Acinetobacter baumannii
Antimicrob Agents Chemother
 , 
2010
, vol. 
54
 (pg. 
3569
-
77
)
7
Adams
MD
Nickel
GC
Bajaksouzian
S
, et al.  . 
Resistance to colistin in Acinetobacter baumannii associated with mutations in the PmrAB two-component system
Antimicrob Agents Chemother
 , 
2009
, vol. 
53
 (pg. 
3628
-
34
)
8
Turton
JF
Gabriel
SN
Valderrey
C
, et al.  . 
Use of sequence-based typing and multiplex PCR to identify clonal lineages of outbreak strains of Acinetobacter baumannii
Clin Microbiol Infect
 , 
2007
, vol. 
13
 (pg. 
807
-
15
)
9
Andrews
JM
BSAC Methods for Antimicrobial Susceptibility Testing
  
Version 9.1 March 2010 http://www.bsac.org.uk/Resources/BSAC/Version_9.1_March_2010_final.pdf (22 March 2011, date last accessed)
10
Iacono
M
Villa
L
Fortini
D
, et al.  . 
Whole-genome pyrosequencing of an epidemic multidrug-resistant Acinetobacter baumannii strain belonging to the European clone II group
Antimicrob Agents Chemother
 , 
2008
, vol. 
52
 (pg. 
2616
-
25
)
11
Chaudhuri
RR
Loman
NJ
Snyder
LA
, et al.  . 
xBASE2: a comprehensive resource for comparative bacterial genomics
Nucleic Acids Res
 , 
2008
, vol. 
36
 (pg. 
D543
-
6
)
12
Langmead
B
Trapnell
C
Pop
M
, et al.  . 
Ultrafast and memory-efficient alignment of short DNA sequences to the human genome
Genome Biol
 , 
2009
, vol. 
10
 pg. 
R25
 
13
Li
H
Handsaker
B
Wysoker
A
, et al.  . 
The Sequence Alignment/Map format and SAMtools
Bioinformatics
 , 
2009
, vol. 
25
 (pg. 
2078
-
9
)
14
Koboldt
DC
Chen
K
Wylie
T
, et al.  . 
VarScan: variant detection in massively parallel sequencing of individual and pooled samples
Bioinformatics
 , 
2009
, vol. 
25
 (pg. 
2283
-
5
)
15
Krawitz
P
Rodelsperger
C
Jager
M
, et al.  . 
Microindel detection in short-read sequence data
Bioinformatics
 , 
2010
, vol. 
26
 (pg. 
722
-
9
)
16
Carver
T
Bohme
U
Otto
TD
, et al.  . 
BamView: viewing mapped read alignment data in the context of the reference sequence
Bioinformatics
 , 
2010
, vol. 
26
 (pg. 
676
-
7
)
17
Post
V
White
PA
Hall
RM
Evolution of AbaR-type genomic resistance islands in multiply antibiotic-resistant Acinetobacter baumannii
J Antimicrob Chemother
 , 
2010
, vol. 
65
 (pg. 
1162
-
70
)
18
McMillan
DJ
Mau
M
Walker
MJ
Characterisation of the urease gene cluster in Bordetella bronchiseptica
Gene
 , 
1998
, vol. 
208
 (pg. 
243
-
51
)
19
Hengge
R
Principles of c-di-GMP signalling in bacteria
Nat Rev Microbiol
 , 
2009
, vol. 
7
 (pg. 
263
-
73
)
20
Bingle
LE
Bailey
CM
Pallen
MJ
Type VI secretion: a beginner's guide
Curr Opin Microbiol
 , 
2008
, vol. 
11
 (pg. 
3
-
8
)
21
Young
DM
Ornston
LN
Functions of the mismatch repair gene mutS from Acinetobacter sp. strain ADP1
J Bacteriol
 , 
2001
, vol. 
183
 (pg. 
6822
-
31
)

Author notes

These authors contributed equally to this work.

Supplementary data