Insights into the genetic contexts of sulfonamide resistance among early clinical isolates of Acinetobacter baumannii

Since the late 1930s, resistance to sulfonamides has been accumulating across bacterial species including Acinetobacter baumannii , an opportunistic pathogen increasingly implicated the spread of antimicrobial resistance worldwide. Our study aimed to explore events involved in the acquisition of sulfonamide resistance genes, particularly sul2 , among the earliest available isolates of A. baumannii . The study utilized the genomic data of 19 strains of A. baumannii isolated before 1985. The whole genomes of 5 clinical isolates obtained from the Culture Collection University of G ¨ oteborg (CCUG), Sweden, were sequenced using the Illumina MiSeq system. Acquired resistance genes, insertion sequence elements and plasmids were detected using ResFinder, ISfinder and Plas-midseeker, respectively, while sequence types (STs) were assigned using the PubMLST Pasteur scheme. BLASTn was used to verify the occurrence of sul genes and to map their genetic surroundings. The sul1 and sul2 genes were detected in 4 and 9 isolates, respectively. Interestingly, sul2 appeared thirty years earlier than sul1 . The sul2 gene was first located in the genomic island GI sul2 located on a plasmid, hereafter called NCTC7364p. With the emergence of international clone 1, the genetic context of sul2 evolved toward transposon Tn 6172 , which was also plasmid-mediated. Sulfonamide resistance in A. baumannii was efficiently acquired and transferred vertically, e.g., among the ST52 and ST1 isolates, as well as horizontally among non-related strains by means of a few efficient transposons and plasmids. Timely acquisition of the sul genes has probably contributed to the survival skill of A. baumannii under the high antimicrobial stress of hospital settings.


Introduction
Antimicrobial resistance (AMR) is an evolutionary response of microbes to withstand the onslaught of antimicrobial agents introduced into their environment.This response has been stocking up in human bacterial pathogens since the beginning of the modern antimicrobial era, marked in 1910 by the introduction of arsphenamine (Salvarsan) into clinical use to treat syphilis cases in humans (Netherton, 1937).Twenty years later, the discovery of sulfonamido-chrysoidine (Prontosil) and development of sulfonamides (1932)(1933)(1934)(1935)(1936)(1937)(1938) was a turning point in the war against bacterial infections (Bickel, 1988).As the only available broad-spectrum antibiotic, sulfonamides were heavily used in the years before penicillin, and this continued into the early years of World War II (Mitchell et al., 2004).Nowadays, the most commonly used sulfonamide for treatment of a variety of human infections is sulfamethoxazole (SMX), mainly in combination with trimethoprim (Nunes et al., 2020).However, several sulfonamide drugs, such as sulfamethazine and sulfadiazine, are extensively used for treatment and prophylaxis of infections in livestock and aquatic animals, as well as for growth promotion purposes in husbandry (Ovung and Bhattacharyya, 2021).
Sulfonamides are chemically synthesized bacteriostatic drugs that function as competitive antagonists of para-aminobenzoic acid, which is an essential compound for the synthesis of folic acid, and thus for bacterial growth (Ovung and Bhattacharyya, 2021).They reversibly inhibit the synthesis of folic acid through binding to dihydropteroate synthase (DHPS), a catalytic enzyme in the folic acid biosynthesis pathway.Sulfonamide resistance was first reported around 1938, where particular strains of Streptococcus did not respond to the bactericidal effects of Abbreviations: AMR, Antimicrobial resistance; SMX, Sulfamethoxazole; DHPS, Dihydropteroate synthase; CS, Conserved segment; IS, Insertion Sequence; CCUG, Culture Collection University of Göteborg; ST, Sequence Type; BIGSdb, Bacterial Isolate Genome Sequence Database; MLST, Multilocus sequence typing; NCTC, National Collection of Type Cultures; GI, Genomic island; orf, Open reading frame; bp, Base pair; AbaR, A. baumannii resistance island; MARR, Multiple antibiotic resistance region; AbGRI, A. baumannii genomic resistance island.
sulfonamide (Buttle, 1939).Vertically acquired resistance to sulfonamides is conferred by mutations in the chromosomal DHPS gene (folP), enabling bacteria to produce altered forms of the DHPS enzyme that are resistant to inhibition by the drug (Skold, 2000).However, a majority of the clinically occurring sulfonamides resistance in Gram-negative bacteria is associated with the horizontal acquisition of transferable genes encoding alternative drug-resistant types of the DHPS enzyme, as first reported in 1975 (Wise Jr. and Abou-Donia, 1975).
To date, four different bacterial genes encoding resistance to sulfonamides have been identified and denoted as sul1, 2, 3, and 4 (Perreten and Boerlin, 2003;Rådström and Swedberg, 1988;Razavi et al., 2017).The sul1 gene, for which the nucleotide sequence was first determined in 1981 (Swift et al., 1981), is usually a key component of the 3 ′ conserved segment (3 ′ -CS) of class I integron (Tn21 type) and has frequently been plasmid-mediated (Rådström and Swedberg, 1988).The sul2 gene, first sequenced in 1988, is also commonly carried on plasmids but has not been found to be part of class I integrons (Rådström and Swedberg, 1988).Notably, the Sul1 enzyme (279 amino acids) was heat-labile while Sul2 (271 amino acids) was heat-stable, and their encoding genes shared only 57% nucleotide identity to each other (Rådström and Swedberg, 1988).In 2003, Perreten and Boerlin detected sul3, encoded by the 54-kb conjugative plasmid pVP440, in E. coli isolated from swine in Switzerland (Perreten and Boerlin, 2003).Sul3 (263 amino acids) showed 40.9% and 40.6% identity to Sul1 and Sul2, respectively.In the pVP440 plasmid, sul3 was not part of class I integron but was rather located in a 3.6 kb segment flanked by two copies of the insertion sequence (IS) 15Δ/26.Interestingly, ancient mutations in the chromosomal folP genes of Rhodobiaceae and Leptospiraceae were found to confer resistance to sulfonamides and were proposed as the likely origins of the sul1-3 genes based on the results of a Bayesian phylogenetic analysis of multiple alignments of their FolP/Sul sequences (Sanchez-Osuna et al., 2018).The fourth sulfonamide resistance gene, sul4, was reported in 2017 in association with class I integrons and with 31-33% nucleotide identity to the three other sul genes (Razavi et al., 2017).The sul4 gene was discovered during systematic analysis of DNA obtained from polluted Indian river sediments.In silico screening of 6489 publicly available metagenomic datasets revealed the occurrence of sul4 in different types of environmental samples obtained from seven countries across Asia and Europe (Razavi et al., 2017).
Over the last 3 decades, Acinetobacter baumannii has gained lots of interest because of the frequent occurrence of nosocomial outbreaks caused by strains exhibiting exacerbated resistance to several classes of antimicrobials such as carbapenems, extended-spectrum cephalosporins, fluoroquinolones or even to last-resort antimicrobial agents including colistin or tigecycline (Havenga et al., 2022).This bacterium is currently one of the most successful opportunistic pathogens worldwide, causing a broad range of severe infections especially among hospitalized patients, including ventilator-associated pneumonia, bloodstream infections, skin and soft tissue infections, wound infections, catheterassociated urinary tract infections and secondary meningitis.Apart from its innate resistance traits, the genomic plasticity of A. baumannii is well-suited for the acquisition of foreign resistance genes by every means of horizontal gene transfer (Sarshar et al., 2021).In addition, a few but extremely relevant virulence attributes have been evolutionarily conserved in the pan-genome of A. baumannii providing it an adaptive advantage to thrive in the hospital environment (Sarshar et al., 2021).
High rates of sulfonamide resistance have frequently been reported among clinical isolates of A. baumannii.For example, a study from Taiwan showed that 71% (286/403) of the A. baumannii clinical isolates collected between 2014 and 2015 were resistant to sulfonamides (Chen et al., 2017).Similarly, non-susceptibility to trimethoprim/sulfamethoxazole was detected in 94.5% (69/73) of non-duplicate A. baumannii urinary isolates collected over a period of 12 months from patients with severe urinary tract infection in India (Girija et al., 2019).sul1 and sul2 were present in 52.1% (38/73) and 45.2% (33/73) of these isolates, respectively.All the clinical isolates (52/52) that were reported in a recent study from Pakistan were resistant to trimethoprim/sulfamethoxazole (Karah et al., 2020).The sul1 and sul2 sulfonamide resistance genes were detected in 11/25 and 14/25 of the investigated isolates, respectively.
The relatively recent appearance of A. baumannii as a pathogen of increasing clinical relevance in human health care has raised much interest regarding its adaptability, and how its genome has successfully been able to acquire and retain new traits for AMR.As resistance to sulfonamides was one of the first appearing resistance traits in bacteria, our present study was focused on elucidating genetic events involved in the initial acquisition of the sul genes, particularly sul2, by the earliest available isolates of A. baumannii.

A. baumannii strains and growth conditions
The study involved genomic data of 19 isolates of A. baumannii collected before 1985 (Table 1), including the whole genome sequences of 5 clinical strains obtained from the Culture Collection University of Göteborg (CCUG), Sweden.The CCUG isolates were revived on Luria agar media (Fluka, Millipore Sigma, Sweden) at 37 • C. 20% glycerol was used for storage at − 80 • C.

Whole genome sequencing
Genomic DNA of the CCUG isolates was extracted using the DNeasy blood and tissue kit (Qiagen, Germany) and quantified with Qubit 2.0 fluorometer using the Qubit™ dsDNA HS Assay kit (Invitrogen, USA).Using the Nextera XT DNA Library Preparation Kit and Nextera™ DNA CD Indexes (Illumina, San Diego, USA), and the Agencourt AMPure XP system (Beckman Coulter, Life sciences, USA), an indexed paired end library was prepared for each isolate.Followed by manual normalization and pooling, the library was sequenced using the MiSeq Reagent Kit v3 (600-cycle) on Illumina MiSeq system (Illumina, San Diego, USA).All steps of DNA preparation, library construction, clean up, and genome sequencing were done according to the suppliers' instructions.The SPAdes version 3.15.4genome assembly toolkit was applied for de novo assembly of the sequence reads.

Genetic context analysis of sul genes
The occurrence of sul genes was verified and their genetic surroundings were manually annotated based on the nucleotide similarities detected by the BLASTn suite search tool (https://blast.ncbi.nlm.nih.gov/).Searches were made against the "Nucleotide collection (nr/nt)", "Whole-genome shotgun contigs (wgs)", and/or "RefSeq Genomes Database (refseq_genomes)" databases.Schematic maps of genetic contexts of sul genes were drawn using Snapgene® 6.0.5 (https://www.snapgene.com/)and modified in Inkscape version 1.0.2.

Results and discussion
In order to identify elements associated with early acquisitions of the sul genes by A. baumannii, our study included all isolates (n = 19) that were obtained from ≥1944 up to 1985 and for which we had available whole genome or sul-related sequence records.Overall, sul2 and sul1 were detected in 9 (47%) and 4 (21%) isolates, respectively.Interestingly, the earliest appearance of sul1 in A. baumannii was in 1977, while sul2 first appeared in ≥1947 (thirty years earlier than sul1).
A dominant variant of sul2 was detected in 8/9 of the sul2-positive isolates.This variant was identical to hundreds of the records available in GenBank including some of the earliest sequences of sul2 (e.g.Gen-Bank: M36657.1,CP077211.1,AE014073.1,and AY823412.1).A synonymous single base pair substitution (C309A) was detected in the sul2 gene of only one isolate (ATCC 17961), resulting in a new variant designated as sul2 ′ (GenBank: CP065432.1,locus_tag: I5593_17975).However, the nucleotide sequence of sul2 ′ was not identical to any of the other records available in GenBank (as of 22nd July 2022).Both variants encoded the same 271-long amino acid sequence (Fig. S1a), corresponding to the Sul2 sulfonamide-resistant dihydropteroate synthase (RefSeq: WP_001043260.1).
Similarly, two variants of sul1 were identified, designated as sul1 and sul1 ′ (Fig. S1b).sul1 ′ had a single base pair substitution (T115G) compared to sul1, leading to the occurrence of a single amino acid change (Ser39Ala) in the encoded protein.Searching for nucleotide similarity to sul1 in the GenBank databases yielded hundreds of identical As expected, the sul-2-positive CCUG isolates (CCUG 33549, CCUG 2488, and CCUG 890) were all resistant to sulfamethoxazole, both alone and in combination with trimethoprim, while those that were lacking sul1 or sul2 (CCUG 6644 and CCUG 26383) were susceptible to sulfamethoxazole (Table S1).Intermediate resistance to ciprofloxacin was noted for all the isolates.One isolate (CCUG 2488) was also intermediate resistant to tetracycline.Intermediate or full resistance to cefotaxime was detected in 4/5 of the isolates.Otherwise, the isolates were all susceptible to ampicillin/sulbactam, piperacillin/tazobactam, ceftazidime, imipenem, meropenem gentamicin, amikacin, tobramycin, levofloxacin, and doxycycline.ATCC 19606 and ATCC 17978 A. baumannii ATCC 19606, isolated in the USA from a urine sample ≤ 1948 (Schaub and Hauber, 1948), and ATCC 17978, collected from a cerebrospinal fluid sample in France ≤1951 (Smith et al., 2007) were both resistant to sulfonamides, while remained susceptible to a range of other antibiotics (Hamidian et al., 2020;Smith et al., 2007).sul2 was the only acquired antimicrobial resistance gene detected in the genomic records of these two isolates.The sul2 gene was located in a genomic resistance island, called GIsul2 (15,460 bp), in ATCC 19606 (Hamidian and Hall, 2017a), while a similar version of GIsul2 interrupted by the insertion sequence (IS) element ISAba1, which will hereafter be called ISAba1::GIsul2 (16,649 bp), was detected in ATCC 17978 (Nigro and Hall, 2011).

Revisiting the genetic contexts of sul2 in
Interestingly, GIsul2 was also found in the chromosomes of the type strain of Enterobacter cloacae subspecies cloacae ATCC 13047 T and of Shigella flexneri ATCC 700930 (also called serotype 2a strain 2457 T), collected in 1890 and 1954, respectively (Nigro and Hall, 2011).The annotations of GIsul2 and ISAba1::GIsul2 have been reported in detail by a number of previous studies (Hamidian and Hall, 2017a;Harmer et al., 2017;Nigro and Hall, 2011).As previously stated, the GIsul2 in ATCC 13047 T was interrupted by a mercuric ion resistance transposon related to Tn5393 (GenBank: CP001918.1).Notably, we detected direct repeats of 14 bp on the boundaries of GIsul2 in ATCC 13047 T and ATCC 700930, of which 5 bp could be a result of target site duplication during the insertion of GIsul2.We also detected GIsul2 in the genomic record of Pseudomonas aeruginosa NCTC11445 (GenBank: LR134308.1),isolated in 1957.However, the segment carrying sul2 together with ΔglmM and ISCR2 was missing in P. aeruginosa NCTC11445 (Data not shown).
The GIsul2 island in ATCC 19606 was part of a larger genomic region, called GI19606 (36,157 bp), which was inserted in the chromosome between open reading frames orf ribonuclease-PH and orf phospholipase-C (Hamidian et al., 2020).GI19606 included 28 orfs, most of which encoded for hypothetical proteins with no putative conserved domains.It had ISAba11 on one end and was flanked by a TSD of 5-bp.The origin of GI19606 was most likely acquired, with the help of ISAba11, from a plasmid related to Acinetobacter johnsonii pXBB1-9 or Acinetobacter pittii p2014N21-145-1 (GenBank: CP010351.1 and CP033569.1,respectively), as proposed previously (Hamidian and Hall, 2017a).
Among the isolates that were sequenced in this study, sul2 was detected in the same genetic context (GI19606) in A. baumannii CCUG 33549, isolated ≤1948 (Ferguson and Roberts, 1950).Interestingly, both ATCC 19606 and CCUG 33549 belonged to ST52 (3,2,2,7,9,1,5).Searching for nucleotide similarity in the GenBank databases enabled us to detect five other isolates belonging to ST52 and carrying GI19606, namely MSP4-16 (GenBank: AODW00000000.1)isolated from a mangrove soil in India in 2010 (Singh et al., 2013), ab736 (GenBank: CP015121.1)obtained from a clinical sample in USA in 2015 (Krishnamurthy et al., 2019), BA20352 and BA20475 (GenBank: JAAOQN000000000.1 and JAAOQM000000000.1,respectively) obtained from clinical blood samples in India in 2019, and A.bA112 (GenBank: JAAQWA000000000.1)obtained in Hong Kong.GI19606 was most likely inserted in the chromosome of these relatively recent ST52 isolates, as shown for the old ST52 ones, although this could not be confirmed since the sequences were located on more than one contig.Deep analysis of the relevant contigs in isolate MSP4-16 enabled us to detect two identical AT-rich direct repeats of 268 bp in GI19606 in close proximity (around 20 kb) to each other (Fig. S2), raising the possibility that the genetic architecture of GI19606 could be subject to rearrangement by direct-repeat-mediated intramolecular recombinational events.
The earliest occurrence of sul2 was found in the genome of A. baumannii NCTC 7364, isolated ≤1947, representing one of the oldest A. baumannii isolates preserved until today (Baumann et al., 1968).The type of sample and country of isolation for NCTC 7364 were unknown.NCTC 7364 was resistant to sulfonamide, trimethoprim, cephalexin, penicillin, and ampicillin (Alexander et al., 1984).Similarly to ATCC 17978, the sul2 gene was located in ISAba1::GIsul2, which was inserted in the tniBb gene of Tn6174.However, the original version of Tn6021, called Tn6022 (13,104 bp including ISAba18), was present in this isolate (Hamidian and Hall, 2011).The whole Tn6174-zone-1-Tn6022 assembly was carried on a plasmid of 148,956 bp, hereafter called NCTC7364p (Fig. S3a).Particularly, the only difference between Tn6021 and Tn6022 is related to a region of 531 bp of Tn6022 been replaced by an analogous region from Tn6174 (Fig. S3b; Hamidian and Hall, 2011).
The sul2 gene in A. baumannii ATCC 17945, isolated ≤1949 (Stuart et al., 1949), had the same genetic environment as of NCTC 7364, except that the tniE gene of Tn6022 was interrupted by a second copy of ISAba1 in ATCC 17945 (Fig. S4).The whole assembly was most likely carried on a plasmid of 151,457 bp showing >99% nucleotide identity to the NCTC7364p.However, the exact composition of the genetic context of sul2 in ATCC 17945 and its occurrence on plasmid were tentatively proposed since the corresponding sequences were distributed on 6 contigs of the genomic record of ATCC 17945.Similarly, the whole genome sequences we obtained for A. baumannii CCUG 2488, isolated ≤1959 (https://www.ccug.se/strain?id=2488), and CCUG 890, isolated ≤1962 from a clinical urine sample in Germany (Stenzel and Mannheim, 1963) revealed the occurrence of sul2 on the same Tn6174-zone-1-Tn6022 genetic context.The type of sample and country of isolation for ATCC 17945 and CCUG 2488 were unknown.

Revisiting the genetic context of sul2 in RUH 875
In A. baumannii RUH 875 (also known as NIPH 527 or A297), isolated from urine sample in the Netherlands in 1984 (Dijkshoorn et al., 1987), sul2 was part of the configuration ISAba1-sul2-ΔglmM-ISCR2-strB-strA (Hamidian et al., 2016).This configuration (6567 bp) arose by the merge of ISAba1-sul2-ΔglmM-ISCR2 (4811 bp), coming from GIsul2, with the strB-strA operon (1756 bp), derived from Tn5393.They were all part of a transposon called Tn6172 (Hamidian et al., 2016), which was carried on plasmid pA297-3 (220,063 bp; GenBank: KU744946.1).Tn6172 (11,719 bp) had two regions with 100% nucleotide identity to the corresponding regions of Tn6174 (Fig. S7).The site of insertion of Tn6172 in pA297-3 was similar to that of the whole Tn6174-zone1-Tn6022 assembly in NCTC7364p.Furthermore, large parts of pA297-3 and NCTC7364p shared >99% nucleotide identity to each other, as shown in Fig. S8.Among the differences between the two plasmids, we noted that Tn6022, part of zone 1, and part of orf97 were deleted in pA297-3 compared to NCTC7364p.This isolate, RUH 875, belonged to ST1 (1,1,1,1,5,1,1), representing one of the earliest isolates of International Clone 1 with an available genomic record.Altogether, our analysis indicated that this new genetic context of sul2 originated from the Tn6174-zone1-Tn6022 assembly by means of few genomic evolutionary events (deletions, insertions, and single nucleotide mutations) combined with inter-strain horizontal transfer of the hosting plasmid.

Summary of the genetic contexts of sul1
The oldest A. baumannii isolate with an identified sul1 gene was sulfamethoxazole-resistant strain HK302, recovered in 1977 from the respiratory tract of a patient in Zurich University Hospital, Switzerland during an outbreak of nosocomial infections (Devaud et al., 1982) and had the allelic profile of ST1 (International Clone 1).A putative structure of the HK302 resistance island was identified based on polymerase chain reaction-restriction fragment length polymorphism mapping and partial sequencing which revealed the presence of a 63.4-kb AbaR3-like resistance island carrying multiple antibiotic resistance genes, including two copies of sul1 (GenBank: HM357806.1,Krizova and Nemec, 2010).Although the whole sequence of this AbaR is currently not available, the described genetic structure corresponded to AbaR0 (GenBank: KF483599.2,Hamidian and Hall, 2018).Similarly, two copies of the sul1 gene were detected in the chromosomes of isolates A1, isolated in the United Kingdom in 1982 (Holt et al., 2015), and RUH 875 which were also members of international clone 1 (ST1).As previously described, the two copies of sul1 were part of the 3 ′ -conservative segments, either truncated or complete, of a variety of class I integrons, which were integrated, along with other transposon units, into the AbaR3-like resistance islands AbaR24 (GenBank: JN968482.3)and AbaR21 (GenBank: KM921776.1)(Nigro et al., 2011).One isolate, RUH 134 (also known as A320 and NIPH 528), collected in 1982 in the Netherlands and assigned to ST2 (corresponding to international clone 2), carried only one copy of the sul1 gene, which was located in the 3 ′ -CS of a complete In4-type class I integron of 7994 bp representing one of the main components of the IS26-bounded genomic resistance island AbGRI2-0b (GenBank: JN247441.4and CP032055.1)(Nigro and Hall, 2012).

Conclusion
Since their first introduction into clinical use, synthetic sulfonamides have been largely handicapped by the rapid appearance of sulfonamide resistance.Searching for sul genes in the genomes of ancient isolates of A. baumannii revealed a time gap of 30 years between the first appearance of sul2 ≥ 1947 and sul1 in 1977.The dynamic ability of sul2 to change contexts was notable and has most likely contributed to its prolonged existence.Importantly, the occurrence of horizontally acquired sulfonamide resistance elements in A. baumannii International Clones 1 and 2 has probably paved the way for the acquisition of additional genes conferring resistance to other classes of antibiotics.In addition, our study added an important clue on the role of plasmids and transposons in the global spread of antimicrobial resistance.One of the limitations of this study was related to the shortfall of available sulfonamide-resistant isolates in the selected timeline (− 1985).Accordingly, the identified genetic environments provided only a glimpse into the primitive arrangement of the sul1 and sul2 genes in A. baumannii.It is appealing to extend the timeline and do further research using a larger sample size, including more recent isolates, although high complexity is anticipated due to the remarkable genomic plasticity of A. baumannii.

A
.Bala et al.

Fig. 1 .
Fig. 1.Genetic context of sul2 in Acinetobacter baumannii ATCC17961.sul2 was found in GIsul2, as part of the genetic configuration sul2-ΔglmM-ISCR2-orf hypothetical protein -resG.GIsul2 was located in Tn7523 which was inserted in the chromosome of ATCC17961.GIsul2 was shown as labeled gray box.Genes and open reading frames (orf) were shown by labeled white (for the interrupted comM gene), gray (for GIsul2) or plum (for the rest of Tn7523) arrows, with the arrowhead indicating the direction of transcription.The insertion sequence element ISAcsp3-like was shown as labeled green box.ISCR2 was shown as labeled orange box.The inverted repeats (IR) of Tn7523 were shown as labeled yellow vertical bars.Target site duplications (TSD) were highlighted and shown as labeled red vertical bars.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)