Early transmissible ampicillin resistance in zoonotic Salmonella enterica serotype Typhimurium in the late 1950s: a retrospective, whole-genome sequencing study.

BACKGROUND
Ampicillin, the first semi-synthetic penicillin active against Enterobacteriaceae, was released onto the market in 1961. The first outbreaks of disease caused by ampicillin-resistant strains of Salmonella enterica serotype Typhimurium were identified in the UK in 1962 and 1964. We aimed to date the emergence of this resistance in historical isolates of S enterica serotype Typhimurium.


METHODS
In this retrospective, whole-genome sequencing study, we analysed 288 S enterica serotype Typhimurium isolates collected between 1911 and 1969 from 31 countries on four continents and from various sources including human beings, animals, feed, and food. All isolates were tested for antimicrobial drug susceptibility with the disc diffusion method, and isolates shown to be resistant to ampicillin underwent resistance-transfer experiments. To provide insights into population structure and mechanisms of ampicillin resistance, we did whole-genome sequencing on a subset of 225 isolates, selected to maximise source, spatiotemporal, and genetic diversity.


FINDINGS
11 (4%) of 288 isolates were resistant to ampicillin because of acquisition of various β lactamase genes, including blaTEM-1, carried by various plasmids, including the virulence plasmid of S enterica serotype Typhimurium. These 11 isolates were from three phylogenomic groups. One isolate producing TEM-1 β lactamase was isolated in France in 1959 and two isolates producing TEM-1 β lactamase were isolated in Tunisia in 1960, before ampicillin went on sale. The vectors for ampicillin resistance were different from those reported in the strains responsible for the outbreaks in the UK in the 1960s.


INTERPRETATION
The association between antibiotic use and selection of resistance determinants is not as direct as often presumed. Our results suggest that the non-clinical use of narrow-spectrum penicillins (eg, benzylpenicillin) might have favoured the diffusion of plasmids carrying the blaTEM-1 gene in S enterica serotype Typhimurium in the late 1950s.


FUNDING
Institut Pasteur, Santé publique France, the French Government's Investissement d'Avenir programme, the Fondation Le Roch-Les Mousquetaires.


SUMMARY Background
Ampicillin, the first semisynthetic penicillin active against Enterobacteriaceae, was released onto the market in 1961. The first outbreaks of disease caused by ampicillin-resistant Salmonella enterica serotype Typhimurium strains were identified in the United Kingdom in 1962 and 1964. This study was performed to date the emergence of this resistance in historical isolates of S. enterica serotype Typhimurium.

Methods
We studied a collection of 288 isolates collected between 1911 and 1969 from 31 countries on four continents. Antimicrobial drug susceptibility was tested for all isolates and wholegenome sequencing was carried out on a subset of 225 isolates. The mechanisms of resistance to ampicillin were identified.

Findings
Eleven isolates (3.8%, 11/288) from three phylogenomic groups were resistant to ampicillin due to the acquisition of various beta-lactamase genes carried by different plasmids, including the virulence plasmid of S. enterica serotype Typhimurium. Three isolates producing TEM-1 were isolated in France (n = 1) and Tunisia (n = 2), in 1959 and 1960, respectively, before ampicillin went on sale. The vectors for ampicillin resistance were different from those reported in the strains responsible for the outbreaks in the United Kingdom in the 1960s.

Interpretation
The relationship between antibiotic use and the selection of resistance determinants is not as direct as often presumed. Our results suggest that the non-clinical use of narrow-spectrum 3 penicillins (e.g., benzylpenicillin) may have favored the diffusion of bla TEM-1 gene-carrying plasmids in the late 1950s in this prominent zoonotic bacterium.

Panel: Research in context
Evidence before this study We searched PubMed for articles published between January 1, 1900 and December 31, 1965, with the following search terms: "Salmonella and ampicillin and resistance", "Salmonella and penicillin and resistance", "Salmonella and penicillinase", and "Salmonella and betalactamase". We also searched for articles published between January 1, 1900 and August 31, 2017, with the following search terms "Salmonella and TEM", and "Salmonella and OXA".
No language restrictions were used. We found no reports describing transmissible ampicillin resistance or beta-lactamase in Salmonella spp. isolates collected before 1962.

Added value of this study
We identified TEM-1 beta-lactamase-producing S. enterica serotype Typhimurium isolates, dating from 1959-1960, several years before the release of ampicillin onto the market. The genetic diversity of these ampicillin-resistant isolates, their resistance mechanisms and their geographic distribution suggest that ampicillin resistance had already diffused in this zoonotic pathogen at that time. We also found relative susceptibility of wild-type S. enterica serotype Typhimurium isolates to penicillin G (8 mg/L), a narrow-spectrum penicillin widely used as a 4 feed additive (subtherapeutic doses of antimicrobial agents over long periods of time) in livestock from the 1950s to the end of the 1960s.

Implications of all the available evidence
Our data shed light on the origin of resistance to a major antimicrobial agent in a prominent zoonotic bacterium. They show that environments in which penicillin G residues are abundant, such as farming environments in the late 1950s, can foster transmissible ampicillin resistance in S. enterica serotype Typhimurium. These findings highlight the need for a "onehealth" approach to tackling the issue of antibiotic resistance.

INTRODUCTION
In May 2015, the 68th World Health Assembly endorsed a global action plan to tackle antimicrobial resistance, 1 a global public health issue for the 21th century. Eighty years ago, the discovery and commercialisation of "sulfa drugs" (sulfonamidochrysoidine and sulfanilamide) opened up a new era in the fight against bacterial infections, that of antibiotics. 2 These synthetic bacteriostatic antibiotics were hailed as "magic bullets" or "wonder drugs", but were rapidly eclipsed by penicillin, a natural bactericidal substance from Penicillium rubrum that had been discovered earlier (1928) but did not go into mass production until World War 2 (WW2). 2 In his princeps paper, Alexander Fleming noted that, unlike Gram-positive bacteria, the bacteria of the "coli-typhoid" group were not sensitive to penicillin. 3 The first penicillin drug active against enterobacteria (broad-spectrum penicillin) was ampicillin, a semisynthetic derivative of 6-aminopenicillanic acid, 4 commercialised by Beecham Research Laboratories under the name "Penbritin" in 1961. 5 The first report describing beta-lactamases, encoded by "R factors" (plasmids), in Enterobacteriaceae strains with transmissible ampicillin-resistance, was published in 1965. 6 In this report, two of the three Enterobacteriaceae strains studied were found to produce the TEM penicillinase. The E.
coli isolate from Greece obtained in 1963 (the strain and its beta-lactamase were named TEM after the patient's name, Temoneria) 7 and the S. enterica serotype Paratyphi B isolate obtained in the United Kingdom in 1963 were subsequently found to contain the bla TEM-1A gene carried by a ~ 40 kb IncX2 plasmid and a ~ 100 kb IncF plasmid, respectively (Table 1). 8,9 The third strain, a S. enterica serotype Typhimurium phage-type 1a (later reclassified as definitive phage-type [DT] 2), isolated in 1962 in Brighton, United Kingdom, produced a beta-lactamase with a substrate profile different from that of TEM (Table 1). This S. enterica serotype Typhimurium strain was isolated during the first reported outbreak caused by an 6 ampicillin-resistant S. enterica serotype Typhimurium strain. It occurred in the second half of 1962, in the United Kingdom, at a time at which dozens of S. enterica serotype Typhimurium DT2 isolates resistant to ampicillin, streptomycin, sulfonamides and tetracyclines were found in humans and pigs. 10 The plasmid of a representative strain, R 1818 , later renamed R46, 9 was sequenced (GenBank accession no. AY046276) and found to be a ~ 50 kb IncN plasmid carrying the bla OXA-2 gene.
In 1963 and 1964, epidemic isolates of Salmonella enterica with transmissible ampicillinresistance were identified several times in Europe (Table 1). [11][12][13][14] In particular, S. enterica serotype Typhimurium DT29 emerged following the adoption of intensive farming methods for the rearing of calves in the United Kingdom. 12 In 1965, 1,297 and 576 isolates of this pathogen were obtained from cattle and humans, respectively. More than 96% (1,294/1,297 and 555/576) of these isolates were multidrug-resistant. NTP1, the ampicillin resistance plasmid of representative strain RT1, 13 was found to be a small (8.3 kb), multicopy, nonconjugative plasmid carrying the bla TEM-1 gene. 14 This timeline suggests that the use of ampicillin may have been the primary cause of the selection of resistant isolates of this major foodborne pathogen.
With a view to improving our understanding of the emergence of resistance to ampicillin, one of the most widely used antibiotics to date, we studied a large collection of historical isolates of S. enterica serotype Typhimurium, a zoonotic agent that has remained one of the two predominant serotypes causing human disease in Europe since WW2. 15,16 These 288 isolates were collected between 1911 and 1969, and 107 were obtained before ampicillin went on sale.

Bacterial isolates
The 288 S. enterica serotype Typhimurium isolates, analysed here were obtained from various sources (including humans, n = 149; animals, food and feed, n = 113) and geographic areas (including Europe, n = 161; Africa, n = 90; Asia, n = 19; America, n = 15), between 1911 and 1969 (table S1). They originated from the collections of the French National Reference Centre for E. coli, Shigella and Salmonella (FNRC-ESS) (n = 277), the Collection de l' Institut Pasteur (CIP) (n = 9), and the World Health Organization Collaborative Centre for Reference and Research on Salmonella (n = 2), Institut Pasteur, Paris, France. All isolates from the FNRC-ESS were maintained as stab cultures at room temperature. They have not been manipulated since being placed in storage (generally one to two weeks after their isolation from biological samples). All other isolates were stored as freeze-dried cultures.

Antimicrobial drug susceptibility testing
Antimicrobial drug susceptibility was determined by the disk diffusion method on Mueller-Hinton agar, with a panel of 32 antimicrobial drugs (Bio-Rad, Marnes-la-Coquette, France), as previously described. 17 The minimal inhibitory concentrations (MICs) of ampicillin and penicillin G (also known as benzylpenicillin) of 67 isolates were determined with Etest strips (bioMerieux, Marcy L'Etoile, France) (table S1).

Resistance transfer determination
A resistance transfer experiment was carried out with the 11 ampicillin-resistant S. enterica serotype Typhimurium isolates on solid medium. E. coli K-12 J5 Azi R and S. enterica serotype Typhimurium STM 36-57 Nal R , a nalidixic acid-resistant mutant generated in the 8 laboratory from the wild-type isolate STM 36-57 (IncF-), were used as recipient strains for plasmid conjugative transfer screening, and for estimating transfer efficiency, respectively.
Strains were mated on a filter placed on either Trypto-Casein-Soy (TCS) (when using S. enterica serotype Typhimurium STM 36-57 Nal R recipient strain) or Drigalski (when using E. coli K-12 J5 Azi R recipient strain) agar. Transconjugants were selected on the same type of agar medium as used for mating, but supplemented with either ampicillin (75 mg/L) or penicillin G (15 mg/L) and nalidixic acid (64 mg/L) or sodium azide (500 mg/L). For isolates yielding no transconjugants on TCS agar, another mating experiment was performed on M9 minimal medium rather than TCS. 18 Three transconjugants were selected at random for each experiment. Plasmid classification was based on PCR-based replicon typing. 17 For matings between S. enterica serotype Typhimurium strains, CRISPOL typing (see below) was performed to distinguish between recipient and donor strains. Each conjugation experiment was carried out in triplicate. Conjugation frequencies were calculated by dividing the number of transconjugants (counted on agar medium containing nalidixic acid and ampicillin or penicillin G) by the total number of recipient cells (counted on agar medium containing only nalidixic acid). The values shown are means from triplicate experiments.

DNA extraction
Total DNA was extracted with the Instagene kit (Biorad) or the Maxwell 16 cell DNA purification kit (Promega, Madison WI) for CRISPOL typing and whole-genome sequencing, respectively, in accordance with the manufacturer's recommendations.

9
The CRISPOL (CRISpr POLymorphisms) high-throughput method for subtyping S. enterica serotype Typhimurium and its monophasic variant was carried out on the Luminex platform, as previously described. 19 Whole-genome sequencing Whole-genome sequencing was performed on a selection of 225 of the 288 S. enterica serotype Typhimurium isolates. The selection was designed to obtain wide coverage in terms of source, spatiotemporal and genetic diversity. Fifty-three (84.1%, 53/63) isolates were not sequenced because they had a same CRISPOL type (CT) and were isolated in the same country within five years of those sequenced. The ten remaining non-sequenced isolates (15.9%, 10/63) were isolated in the same country, within two years of those sequenced, but were mostly from unknown sources (n = 2) or sources already overrepresented (pets, rodents and birds, n = 6

Read alignment and SNP detection
Illumina-generated paired-end reads and simulated paired-end reads from publicly available assembled genomes were mapped onto the reference genome of S. enterica serotype Typhimurium strain LT2, including the chromosome (GenBank accession no. AE006468) and plasmid pSLT (GenBank accession no. AE006471), with Bowtie 2 version 2.1.0. Singlenucleotide polymorphisms (SNPs) were identified with SAMtools version 0.1.19, as previously described. 22 SNPs present in insertion sequences identified by ISfinder (https://www-is.biotoul.fr/) or recombinogenic regions identified by Gubbins were removed, as previously described. 22

Phylogenetic analyses
The maximum-likelihood (ML) phylogenetic tree was built from a 32,741-chromosomal SNP alignment of all S. enterica genomes. RAxML version 7.8.6 (GTRGAMMA substitution model) was used to construct the tree. We performed 100 bootstrap pseudoreplicate analyses to assess support for the ML phylogeny. The tree was rooted on S. enterica serotype Weltevreden 840K. The final tree was visualised with FigTree version 1.4.2.

De novo assembly
The reads for each strain were assembled de novo with SPAdes version 3.6.0. 23

Genetic analyses
In silico MLST and serotyping were performed with the "Achtman 7 Gene" and "Serotype Prediction (SISTR)" tools of Enterobase, respectively. The presence and type of antimicrobial resistance genes (ARGs) or ARG-containing structures were determined with ResFinder

ROLE OF THE FUNDING SOURCE
The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

Characterisation of the ampicillin-resistant isolates
The 11 ampicillin-resistant S. enterica serotype Typhimurium isolates detected by the disk diffusion method and shown in Table 2 had penicillin MICs (for ampicillin or penicillin G) ≥ 256 mg/L (table S1, figure 2). The genomic analysis identified a beta-lactamase gene, either bla TEM (n = 9) or bla OXA (n = 2) in all these isolates. In a selection of 56 S. enterica serotype Typhimurium isolates devoid of beta-lactamase genes and belonging to the different genomic clades of the collection (see below and Appendix, page 1), the MIC of ampicillin ranged from 0.06 to 8 mg/L (median 1 mg/L), whereas the MIC of penicillin G ranged from 0.125 to 16 mg/L (median 8 mg/L) (table S1, figure 2).
The most frequent beta-lactamase gene, bla TEM-1B , carried by transposon Tn2, 26  The three S. enterica serotype Typhimurium isolates collected in France and Tunisia before the release of ampicillin contained the Tn2 transposon, in the IncX1 (n = 1) and pSLT-derived IncF (n = 2) plasmids. Tn2 was found inserted at two different sites within IncF.

Genetic relationships between the ampicillin -resistant isolates
A whole-genome phylogeny based on 32,741 SNPs evenly distributed over the non-repetitive non-recombinant core genome showed the clustering of the 11 ampicillin-resistant S. enterica serotype Typhimurium isolates into three groups (figure 4, Appendix, page 3).
The main group comprised seven ampicillin-resistant isolates collected in France between 1959 and 1969, with four types of beta-lactamase genes (bla TEM-1A , bla TEM-1B , bla OXA-1 , and bla OXA-2 ). These seven ampicillin-resistant isolates belonged to a clade consisting of 13 West and South European isolates (plus one North African isolate) collected from humans, food and animals (chicks) and belonging to ST19 and CT46 or variants of CT46 (table S1).
The second group of ampicillin-resistant isolates comprised two human ST19-CT896 isolates collected in Tunisia in 1960. They belonged to a clade consisting of 14 ST19-CT117 (or CT117-derived) isolates mostly obtained from humans (with the exception of one isolate from a sausage) in France, Belgium and Tunisia between 1953 and 1963.

16
The third group comprised the two ampicillin-resistant ST313-CT810 isolates collected in France and Tunisia in 1968. They belonged to a clade consisting of six isolates belonging to either ST302 (n = 3) or ST313 (n = 3). The four non ampicillin-resistant isolates originated from Turkey (two ST302 and one ST313) and Tunisia (one ST302) and were collected between 1956 and 1965.

DISCUSSION
We show here that transmissible ampicillin resistance actually emerged several years earlier than previously thought in S. enterica serotype Typhimurium, as the beta-lactamase bla TEM-1B gene was found on two different plasmids (IncX1 and IncF) in three isolates from Europe (France) and Africa (Tunisia) collected in 1959-1960. However, we did not aim here to determine the true extent of the global distribution of ampicillin resistance in S. enterica serotype Typhimurium, due to an obvious sampling bias (i.e., the available historical isolates studied here were predominantly from France and former French colonies in Africa and Asia).
Another limitation of our study was the mode of storage of the collection strains. Almost all the isolates were stored as stab cultures and the absence of antibiotic selection pressure may not have favored the maintenance of acquired plasmids in the long term in these strains, as previously reported for Bacillus anthracis. 26 The proportion of ampicillin-resistant isolates in our selection of Salmonella isolates may therefore be underestimated.
Nevertheless, the use of ampicillin, the first broad-spectrum penicillin indicated for the treatment of infections due to Enterobacteriaceae, which began in 1961, 5  Our results provide no proof of a causal link between the use of penicillin G and the emergence of transmissible ampicillin-resistance in livestock, but they suggest that an environment in which penicillin G residues were abundant, such as the farming environment described by Anderson 22 29,30 This difference may be due to changes in the food trade, which was much less globalised in the late 1950s-early 1960s than it has been since in the 1980s.
In conclusion, our study sheds light on the early steps in the emergence of resistance to ampicillin, the first penicillin active against Enterobacteriaceae, still in widespread use. We found that the relationship between antibiotic use and the selection of resistance determinants was not as direct as often presumed. Since the early days of antibiotic use, agriculture has played a role in the selection of resistant pathogens alongside medicine, particularly during periods of intensification. These findings highlight the need for a "one-health" approach to tackling the issue of antibiotic resistance.   The number of S. enterica serotype Typhimurium isolates studied is shown in gray, that of isolates resistant to ampicillin is shown in red, and that of isolates producing a beta-lactamase is shown in blue.