Phage-mediated transmission of colistin resistance in Acinetobacter baumannii

doi:10.21203/rs.3.rs-3295616/v1

Download PDF

Research Article

Phage-mediated transmission of colistin resistance in Acinetobacter baumannii

https://doi.org/10.21203/rs.3.rs-3295616/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 27 Jan, 2024

Read the published version in Drug Resistance Updates →

Version 1

posted

You are reading this latest preprint version

Antimicrobial resistance is a global threat to human health, and Acinetobacter baumannii is a paradigmatic example of how rapidly bacteria become resistant to clinically relevant antimicrobials. The emergence of multidrug-resistant A. baumannii strains has forced the revival of colistin as a “last-resort” drug, suddenly leading to the emergence of colistin resistance. Here, we demonstrate that the A. baumannii temperate bacteriophage Φ19606 acts as a vehicle of colistin resistance through the horizontal transfer of the eptA1 gene, encoding a lipid A phosphoethanolamine transferase which prevents colistin from binding to the outer membrane. eptA1 is coregulated with its chromosomal homolog pmrC via the PmrAB two-component system, and the expression of both genes is induced by low calcium and magnesium levels, resulting in colistin resistance. Moreover, eptA1 facilitates the emergence of spontaneous colistin-resistant mutants. Overall, Φ19606 is an unprecedented example of a self-transmissible phage vector implicated in the dissemination of colistin resistance.

General Microbiology

Acinetobacter baumanni

colistin resistance

bacteriophage

phage-mediated trasmission

Antimicrobial resistance has been listed by the World Health Organization (WHO) as one of the top ten threats to global health¹. Among multidrug-resistant pathogens, Acinetobacter baumannii is of foremost concern in hospital-acquired infections², leading both WHO and the Centers for Disease Control and Prevention (CDC) to prioritize carbapenem-resistant A. baumannii (CRAB) as a critical pathogen requiring the development of new antimicrobial strategies^1,3. CRAB has spread worldwide, with outbreaks propelled by the recent COVID-19 pandemic^4,5. CRAB infections are worrisome due to the associated mortality, healthcare and community burden, prevalence and trend of resistance, transmissibility, and poor treatability⁶. While optimal treatment for CRAB infections is still undetermined^7–9, guidelines support empirical combination therapy^8,10, including the administration of colistin (polymyxin E; hereafter Col), a “last-resort” drug that phased out due to dose-related nephrotoxicity^11,12. Concomitant with Col revival, Col resistance emerged in A. baumannii¹³, although at a relatively low frequency^14,15.

Since Col is a positively charged membrane-disrupting peptide, Col resistance can originate from lipopolysaccharide (LPS) modifications that reduce the negative charge of LPS and, therefore, the electrostatic affinity of Col for the outer membrane (OM)¹⁶. LPS modifications include the addition of 4-amino-4-deoxy-L-arabinose (AraN) or phosphoethanolamine (PetN) to the lipid A moiety^17,18. In A. baumannii Col resistance also occurs through loss-of-function mutations in the LPS biosynthetic genes lpxA, lpxC, and lpxD, resulting in LPS-deficient mutants¹⁹. However, impairment in LPS biosynthesis has a dramatic fitness cost and is rare in A. baumannii clinical isolates^20–22. Different from other Gram-negative pathogens, the AraN lipid A modification pathway is absent in A. baumannii¹⁷, making overexpression of PetN transferase(s) the principal Col-resistance mechanism^23,24. In A. baumannii, the most frequent mutations conferring Col resistance map in the chromosomal pmrCAB operon, encoding a PetN transferase (PmrC), and a two-component regulatory system (PmrAB) controlling pmrCAB transcription^23,25. Gain-of-function mutations in pmrAB increase pmrC expression and modification of lipid A with PetN^16,23,26,27. Recently, three pmrC homologs, namely eptA1, eptA2, and eptA3, have been identified in the chromosome of some clinical A. baumannii isolates^23,28, all conferring Col resistance when expressed downstream of an exogenous promoter²³.

Transferable Col resistance has extensively been documented in Enterobacteriaceae carrying the plasmid-borne mobile colistin-resistance (MCR) PetN transferases^29–31, which have seldom been detected in A. baumannii^32–36. Bacteriophages can also transfer antimicrobial resistance genes (ARGs) by either transduction or integration into the host chromosome of ARG-containing prophage genomes. Phage-encoded ARGs conferring resistance to all the main classes of antibiotics, except polymyxins, have been described in several species^37–40 including A. baumannii^41–42. Here, we report horizontal transmission of eptA1-mediated Col resistance via the A. baumannii bacteriophage Φ19606⁴³. Φ19606 provides the first example of phage-driven Col resistance in bacteria, raising further concern about the dissemination of antibiotic-resistance genes by phage-mediated horizontal gene transfer⁴⁴.

The Φ19606 bacteriophage promotes colistin resistance in A. baumannii

Comparative genomics studies of A. baumannii ATCC 19606^T sublines highlighted the uneven presence of a 52-kb prophage, named Φ19606, integrated between the gene for a hypothetical protein (HTZ92_2409) and the ssrS gene (Fig. 1a,b)⁴³. Both integrated and episomal forms of Φ19606 were found to coexist in the original A. baumannii ATCC 19606^T isolate (subline D), whereas Φ19606 was absent in a laboratory-adapted variant (subline A)⁴³.

Virions released from mitomycin C (MMC)-induced ATCC 19606(D) were used to infect ATCC 19606(A) cells, and a single turbid plaque was taken for transmission electron microscopy (TEM) imaging. Φ19606 appeared ~ 350 nm in length, with an icosahedral capsid and a segmented, noncontractile tail (Fig. 1c,d), consistent with its former assignment to the Siphoviridae family⁴³. Infectivity was tested on a panel of 14 Φ19606-free strains belonging to the most frequent sequence types (ST), including ATCC 19606(D) as control (Table S1, Extended Data 1). Spot test and PCR assays revealed that Φ19606 could be transferred and maintained in six strains, hereafter named ATCC 19606(A)Φ, 139LΦ, 141MΦ, 147NΦ, 4026Φ, and C13-373Φ, by integrating into the same chromosomal site, while being also detectable in the episomal form (Fig. 1e; Fig. S1, Extended Data 2).

Interestingly, the Φ19606 genome contains a single copy of the eptA1 gene, coding for a putative lipid A PetN transferase, mapping between the genes for anti-CRISPR (ACR) protein and a TonB-dependent receptor (Fig. 1a). Among PetN transferases, PmrC is the closest EptA1 homolog (93% identity), whereas the plasmid-encoded MCR-4.3³³ is distantly related to EptA1 (36% identity).

Ф19606 was detected in ca. 41% of 523 complete A. baumannii genomes retrieved from the NCBI database (see Methods); in ca. 29% of them it showed the canonical structure (Fig. 1a), whereas in ca. 12% it carried major mutations and/or rearrangements (Fig. 1f; Tables S2, Extended Data 1). In all Ф19606-positive strains, eptA1 mapped in the same position of the phage genome, except for two strains (DT-Ab007 and 15A45) in which eptA1 mapped outside of Ф19606 likely due to mobilization by insertions sequences (IS) (Fig. S2, Extended Data 2). Intriguingly, in strain DT-Ab007 the eptA1 gene is replaced by a class 1 integron, carrying multiple ARG cassettes (Fig. S2, Extended Data 2). The EptA1 protein sequence is highly conserved, and premature eptA1 stop codons were detected in only 10 out of 213 Ф19606-positive strains in our dataset (Fig. S3, Extended Data 2). Notably, Ф19606 was carried by phylogenetically unrelated A. baumannii strains belonging to different STs, denoting a role of Ф19606 in eptA1 horizontal transmission (Fig. S4, Extended Data 2).

Reduced Col susceptibility was observed for Ф19606-infected strains relative to the Ф19606-negative counterparts, as demonstrated by time-kill assays (Fig. 1g). However, no differences in the minimal inhibitory concentration (MIC) of Col in CAMHB were observed between Ф19606-infected and uninfected A. baumannii strains (Table S3, Extended Data 1), raising the possibility that the EptA1 enzyme is poorly functional and/or expressed at low levels under the MIC assay conditions.

To verify the PetN transferase activity of EptA1, the eptA1 ORF from A. baumannii ATCC 19606(D) was cloned under an arabinose-inducible promoter in the expression vector pVRL2⁴⁵ and introduced in the Ф19606-negative ATCC 19606(A) strain. Likewise, the pmrC ORF from ATCC 19606(A) was cloned in pVRL2 and introduced in ATCC 19606(A) as a positive control. Since the pmrC start codon has not yet been experimentally determined in A. baumannii⁴⁶, two different DNA fragments of 1,650-bp and 1,602-bp, referred to as pmrC_long (549 residues) and pmrC_short (533 residues) respectively, were cloned into pVRL2. Arabinose-induced expression of eptA1 resulted in a dramatic increase in Col MIC (64-fold), compared to the moderate increase observed for pmrC_long (4-fold) (Table 1). Conversely, pmrC_short expression did not affect the Col MIC (Table 1), denoting that the 16 N-terminal residues of PmrC_long have a prominent role in the enzyme activity. Coherently, optimal alignment between EptA1 and PmrC requires the inclusion of 16 N-terminal amino acids of PmrC (Fig. S5, Extended Data 2). The PmrC_long and EptA1 3D structural models are highly superimposable, and no differences were detected in the amino acid residues constituting the catalytic and the metal-binding sites, as inferred from the Neisseria gonorrhoeae EptA crystal structure (Fig. S5, Extended Data 2)⁴⁷.

To provide biochemical evidence of PetN transferase activity, PetN addition to lipid A was estimated by MALDIxin analysis⁴⁸. Upon arabinose induction, the level of lipid A phosphoethanolamination in ATCC 19606(A)(pVRL2eptA1) was > 2-fold higher than that in ATCC 19606(A)(pVRL2pmrC_long) while phosphoethanolaminated lipid A was not detected in ATCC 19606(A)(pVRL2pmrC_short) (Table 1; Fig. S6, Extended Data 2), providing evidence that EptA1 is a functional PetN transferase. Altogether, MIC data and MALDIxin results suggest that EptA1 is more efficient than PmrC in lipid A modification.

Expression of both pmrC and eptA1 is activated in response to Ca²⁺ and Mg²⁺ deficiency

EptA1 causes efficient PetN transfer to lipid A and determines Col resistance when ectopically expressed. However, in Ф19606-infected A. baumannii strains lipid A modification was undetectable and Col susceptibility was barely decreased, raising the possibility that Ф19606-encoded eptA1 was poorly expressed under the assay conditions. To investigate the regulation of eptA1 and pmrC, a double-reporter plasmid (pCLV2) enabling simultaneous monitoring of the expression of both genes at the single-cell level was generated by cloning the promoter regions of eptA1 and pmrC (hereafter designated PeptA1 and PpmrC) upstream of the gfp and mCherry reporter genes, respectively (Fig. 2a).

Since divalent cations Mg²⁺ and Ca²⁺ are crucial for OM integrity due to the stabilization of the negatively charged LPS layer⁴⁹, we hypothesized that PetN-mediated modification of lipid A could be induced under conditions of Mg²⁺ and Ca²⁺ deficiency, to compensate for the negative charge imbalance of the LPS layer, thereby increasing OM stability. To test this hypothesis, ATCC 19606(A)Φ harboring the pCLV2 vector was grown in a divalent cation-depleted minimal medium (DCDM9) supplemented with increasing MgSO₄ and CaCl₂ concentrations, and GFP (green) and mCherry (red) emissions were recorded. The addition of either MgSO₄ or CaCl₂ to DCDM9 significantly lowered the green and red fluorescence emissions at both the population (Fig. 2b) and single-cell levels (Fig. 2c; Fig. S7, Extended Data 2), demonstrating that both cations downregulate, to different extents, the activity of PeptA1 and PpmrC promoters. Iron and zinc, known for influencing pmrC/eptA gene expression in Enterobacteriaceae and Pseudomonadaceae⁵⁰, did not affect pmrC and eptA1 expression in A. baumannii (Fig. S8a,b, Extended Data 2). Strong induction of both PeptA1 and PpmrC was also observed upon supplementation of cation-adjusted Mueller Hinton broth (CAMHB) with ≥250 µM ethylenediaminetetraacetic acid (EDTA), to chelate divalent cations (Fig. 2d,e). Conversely, sub-MIC Col did not activate either pmrC or eptA1 expression (Fig. S8c, Extended Data 2). Notably, the activity of PeptA1::gfp and PpmrC::mCherry fusions was low in heat-inactivated human serum, saliva, and urine, compared with M9, probably due to the high content of free Mg²⁺ and Ca²⁺ (> 700 µM) in these biological fluids (Fig. S8d, Extended Data 2)⁵¹.

Given the repressive effect of Mg²⁺ and Ca²⁺ ions on both PeptA1 and PpmrC activities, Col susceptibility was tested on a panel of Φ19606-positive (n = 17) and -negative (n = 12) strains in the standard CAMHB medium (containing 160 µM Mg²⁺ and 76 µM Ca²)⁵², in divalent cation-depleted CAMHB (DCDMHB), and in DCDMHB repleted with 160 µM Mg²⁺ and 76 µM Ca²⁺ as control. For all but two [ATCC 17978 and ATCC 19606(A)] of the tested strains the Col MIC was 2 to 3-fold higher in DCDMHB than in CAMHB and Mg²⁺- and Ca²⁺-repleted DCDMHB (Table S3, Extended Data 1), as also represented by MIC₅₀ and MIC₉₀ values (Fig. 2f). Furthermore, strains harboring Φ19606 showed an overall higher Col MIC in DCDMHB than the phage-free strains (Table S3, Extended Data 1), suggesting that induction of both eptA1 and pmrC genes under conditions of Mg²⁺ and Ca²⁺ depletion has an additive effect on the Col resistance level. Even if Col resistance is increased above the breakpoint (2 mg/L) under conditions of Mg²⁺ and Ca²⁺ depletion, the clinical relevance of this observation is so far unpredictable. While elevated Mg²⁺ and Ca²⁺ concentrations in biological fluids should repress the expression of both eptA1 and pmrC (Fig. S8d, Extended Data 2), local concentrations of these cations can vary depending on the tissue, patient’s status, and therapeutic regimen^53,54.

The PmrAB two-component regulatory system controls both eptA1 and pmrC expression

To identify the transcriptional regulator(s) responsible for Mg²⁺- and Ca²⁺-repressible expression of eptA1 and pmrC, 12 Col-resistant spontaneous mutants of ATCC 19606(A)Φ carrying pCLV2 were selected on CAMHB-agar plates supplemented with 4×MIC Col, all showing higher GFP and mCherry fluorescence emission compared to the parental strain (named AΦCR1 to AΦCR12 in Fig. 3a). One mutant showing no increase in fluorescence emission was also included (AΦCR13, Fig. 3a). The estimated frequency of Col-resistant mutants was 1.36 ± 0.26×10^− 6. Whole-genome sequencing revealed that all the resistant mutants but AΦCR13 carried various missense mutations in the pmrB gene, encoding the histidine kinase sensor component of the pmrAB system (Table S4, Extended Data 1); three of them cause amino acid substitutions (17S→R, 232T→I, and 263R→C) that were previously proposed to affect PmrB signal transduction, resulting in constitutive activation of the PmrA response regulator^24,26,28. Moreover, most of the amino acid substitutions deriving from the missense mutations were localized in the PmrB cytoplasmic domain, likely affecting kinase activity and/or PmrA binding (Fig. 3b). Notably, a PmrA box, matching the previously identified consensus sequence (5′-HTTAADN₅HTTAAD-3′)²⁷, was found in the promoter region of both pmrC and eptA1 (Fig. 3c). Different from all the other mutants, AΦCR13 carried a missense mutation in lpxA, resulting in the 183Y→N substitution in the LpxA enzyme (Table S4, Extended Data 1), essential for LPS biogenesis¹⁹.

To confirm the dual role of PmrA as the transcriptional activator of both eptA1 and pmrC, the pME6032pmrAB plasmid, directing IPTG-inducible expression of pmrAB, was introduced in ATCC 19606(A)Φ(pCLV2), and PeptA1 and PpmrC activities were determined under conditions of divalent cation sufficiency and deficiency. Induction of pmrAB transactivated both pmrC and eptA1 genes only under conditions of divalent cation depletion indicating that the activity of the PmrAB two component system is inhibited by Mg²⁺ and Ca²⁺ (Fig. S9, Extended Data 2). Conversely, expression of both pmrC and eptA1 was insensitive to Mg²⁺ and Ca²⁺ repression in transactivation assays conducted with all the PmrB mutant proteins expressed from the IPTG-inducible promoter of pME6032 (Fig. 3d). This suggests that Col-resistance mutations in pmrB convert PmrAB to a permanently active state.

Trading-off and trading-up of PetN addition to lipid A in A. baumannii

Col resistance has been associated with loss of fitness and virulence attenuation in both clinical and laboratory A. baumannii strains^16,46. To investigate the impact of lipid A phosphoethanolamination on A. baumannii fitness, the in vitro growth rate of ATCC 19606(A) transformants carrying alternatively pVRL2, pVRL2eptA1, pVRL2pmrC_long, or pVRL2pmrC_short was monitored during time. Arabinose induction moderately reduced the growth rate of ATCC 19606(A) transformed with pVRL2eptA1, whereas the growth of the other transformants was unaffected (Fig. S10, Extended Data 2). This result was reinforced by in vitro competition experiments using binary combinations of ATCC 19606(A)(pVRL2eptA1), ATCC 19606(A)(pVRL2pmrC_long), and ATCC 19606(A)(pVRL2pmrC_short) under both inducing and non-inducing conditions. In all co-culture experiments, ATCC 19606(A)(pVRL2eptA1) and ATCC 19606(A)(pVRL2pmrC_short) invariably showed the lowest and highest competitive index (C.I.), respectively (Fig. 4a), whereas no disequilibrium between strains was observed in non-induced cultures (Fig. S11, Extended Data 2). These results suggest that the metabolic burden and/or the extent of cell envelope modifications negatively impact A. baumannii fitness in vitro. Lipid A phosphoethanolamination does not seem to affect A. baumannii membrane integrity, as documented by similar permeability to propidium iodide of uninduced and induced ATCC 19606(A) transformants carrying alternatively pVRL2, pVRL2eptA1, pVRL2pmrC_long, or pVRL2pmrC_short (Fig. S12, Extended Data 2). However, inducible expression of eptA1 caused a significant reduction of the cellular metabolism and ATP content, whereas expression of both pmrC_long and pmrC_short had negligible effects (Fig. S13, Extended Data 2; Fig. 4b). Therefore, loss of fitness of A. baumannii expressing the eptA1 PetN transferase does not depend on altered membrane integrity, while it can be ascribed to the metabolic burden required for lipid A phosphoethanolamination by EptA1.

Given the importance of biofilm formation in A. baumannii ecology and virulence^56,57, the effect of eptA1 or pmrC expression on biofilm formation was tested in arabinose-induced cultures of ATCC 19606(A) transformants carrying alternatively pVRL2, pVRL2eptA1, pVRL2pmrC_long, or pVRL2pmrC_short. Only eptA1 expression caused significant biofilm reduction (Fig. 4c-e), in line with the elevated energetic cost of biofilm formation and former evidence that colistin-resistant A. baumannii clinical strains are characterized by poor biofilm formation⁵⁸.

Two proteins (GO593_11845 and GO593_11855) are predicted to constitute the Φ19606 tail, and their positive charge at neutral pH (isoelectric point = 8.18 and 8.99, respectively) could mediate the adsorption of Φ19606 onto A. baumannii cells via electrostatic interaction(s) with the negatively charged OM, as it is for many phages⁵⁹. Since the addition of PetN causes a reduction of the lipid A negative charge, a spot assay was employed to evaluate the ability of Φ19606 to infect A. baumannii cells overexpressing either eptA1 or both pmrC variants. Upon arabinose induction, no plaques were observed for ATCC 19606(A)(pVRL2eptA1) and ATCC 19606(A)(pVRL2pmrC_long), while turbid plaques were observed in the absence of the inducer, as well as in ATCC 19606(A)(pVRL2pmrC_short) irrespective of induction, denoting a productive phage infection (Fig. 4f). These results indicate that both EptA1- and PmrC-mediated PetN addition to lipid A prevents Φ19606 adsorption.

Abrogation of phage adsorption was previously correlated with loss-of-function mutations in genes of the K locus, responsible for the biosynthesis of capsular polysaccharides⁶⁰. K-locus assignment for a selected panel of 514 A. baumannii strains retrieved from NCBI revealed no association between capsular type and carriage of Ф19606 (Fig. S14, Extended Data 2; Table S5, Extended Data 1), indicating that the capsule neither prevents Ф19606 infection nor constitutes the primary Ф19606 receptor. Accordingly, the Ф19606 genome lacks capsule depolymerases genes, which are extremely rare in Siphoviridae⁶¹.

Φ19606 increases the emergence of Col-resistant mutants

Although Φ19606 moderately reduces the level of Col susceptibility (Fig. 1g), we wondered if it could increase the frequency of spontaneous mutation to Col resistance. Interestingly, Col resistance (MIC > 4 mg/L) emerged at a significantly higher frequency in Φ19606-infected strains than in the uninfected progenitors (Fig. 5a). Likewise, Col resistance emerged more frequently in natural Φ19606-positive strains than in Φ19606-negative strains (Fig. 5b; Table S3, Extended Data 1). In both experiments, no differences were observed in the frequency of gentamicin-resistant mutants between the infected and uninfected groups (data not shown). Col-resistant mutants obtained from ATCC 19606(A)Φ showed higher MIC₅₀ and MIC₉₀ values than mutants obtained from ATCC 19606(A) (Fig. 5c; Table S3, Extended Data 1), indicating that among spontaneous Col-resistant mutants, those carrying Φ19606 show an overall higher level of Col resistance. Knock-out of pmrC in ATCC 19606(A)Φ and ATCC 19606(A) decreased the frequency of mutation to Col resistance by a factor ∼2 and ∼14, respectively. In the pmrC-defective ATCC 19606(A) background the mutation(s) to Col resistance cannot be attributed to lipid A PetN transferases. Remarkably, the frequency of Col resistance was ∼5-fold higher in the pmrC-deficient ATCC 19606(A)Φ than in the pmrC-proficient phage-free (eptA1-deficient) strain ATCC 19606(A) (Fig. 5d; Table S3, Extended Data 1), confirming that the Φ19606-encoded eptA1 gene provides a significant contribution to the emergence of Col resistance in A. baumannii.

Since the most frequent mutations causing clinically significant Col-resistance in A. baumannii involve pmrAB^23,24,27, we hypothesized that mutated PmrB variants directing constitutive expression of both pmrC and eptA1 could increase the Col resistance levels in strains carrying Φ19606, compared with Φ19606-free strains. As a proof of concept, Col susceptibility was compared between ATCC 19606(A) and ATCC 19606(A)Φ, both carrying the pME6032pmrAB_T187P plasmid, which directs IPTG-inducible expression of the PmrB variant carrying the 187T→P substitution (Table S4, Extended Data 1). The IPTG-modulated induction of pmrB_187T→P paralleled the growth level of both strains in the presence of increasing Col concentrations (Fig. 5e). However, in the presence of similar IPTG concentrations, ATCC 19606(A)Φ was more resistant than the phage-free control (Fig. 5e), implying that concomitant activation of both pmrC and eptA1 expression has an additive effect on Col resistance.

Mobile genetic elements, like plasmids and bacteriophages, are well-known vehicles of genes implicated in virulence and resistance to toxic agents and have driven bacterial evolution toward pathogenicity and resistance to antimicrobials. Plasmid-mediated transmission of Col resistance has extensively been documented in recent years^21,29–36, whereas Col resistance has never been associated with self-transmissible bacteriophages. Here, we demonstrated that the A. baumannii temperate phage Φ19606 can mobilize the eptA1 gene, which encodes a PetN transferase capable of conferring clinically meaningful levels of Col resistance when expressed at high levels in A. baumannii. We found that expression of the phage-encoded eptA1 gene is induced by the depletion of OM-stabilizing divalent cations Ca²⁺ and Mg²⁺, likely reflecting a compensatory mechanism to counterbalance the OM net charge by the addition of positively charged PetN to lipid A. We also demonstrate that the PmrAB two-component system serves as the master regulator of PetN transferases genes in A. baumannii, controlling the transcription of both chromosomal pmrC and Φ19606-encoded eptA1 genes, probably as the result of Φ19606 co-evolution with its host. Indeed, mutated variants of pmrAB direct constitutive expression of both pmrC and eptA1, with additive effects on Col resistance. Irrespective of whether pmrC- or eptA1-mediated, the PetN transfer to lipid A prevents subsequent infection by Φ19606. While lipid A phosphoethanolamination entails a fitness cost for A. baumannii, as inferred from the loss of competitiveness of strains overexpressing either eptA1 or pmrC, it also provides an effective mechanism to protect from Φ19606 superinfection. Temperate phages like Φ19606 gain advantages from expressing genes that increase the fitness of their host. Indeed, surface modification is a key strategy for phage superinfection exclusion⁶², and EptA1-mediated PetN addition to lipid A would combine protection from the lethal effect of a membrane-active peptide (Col) with resistance to superinfection by phages that share an entry pathway with Φ19606. Since Col resistance in A. baumannii has most often been associated with pmrC overexpression^16,23–27, the actual contribution of Φ19606-encoded eptA1 to clinically meaningful Col resistance could be underestimated, given the broad diffusion of Φ19606 in the global A. baumannii population. Even when expressed at low levels, eptA1 alleviates Col-mediated A. baumannii killing and facilitates the emergence of mutants capable of resisting clinically significant Col concentrations. In conclusion, this study provides the first evidence of phage-mediated dissemination of a Col-resistance gene in A. baumannii, unveiling a new strategy towards the evolution of antimicrobial resistance in a critical pathogen.

Bacterial strains and culture media. Bacteria were routinely grown in CAMHB (BBL Mueller Hinton II broth, Becton Dickinson) or on CAMHA plates at 37°C. The Clinical & Laboratory Standards Institute (CLSI) protocol for the preparation of iron-depleted Mueller Hinton II broth (ID-MHB)⁶³ was adapted to DCDMHB preparation. Briefly, CAMHB was treated for 16 h at 4°C with 100 g/L of the metal-chelating Chelex 100 resin (Bio-Rad) under moderate stirring, then filtered through Whatman no. 1 filter paper, and the pH was adjusted to 7.3. After autoclaving, DCDMHB was replenished with 3 µM FeCl₃⁶⁴ and 10 µM ZnCl₂⁶³. The M9 minimal medium and the Luria-Bertani (LB) medium were prepared according to Sambrook et al., 1989⁶⁵. The final addition of 1 mM MgSO₄ and 0.2 mM CaCl₂ to M9 was omitted for the preparation of DCDM9. The gentamicin (Gm) concentrations used for E. coli DH5α and A. baumannii were 10 mg/L and 100 mg/L, respectively. The Zeocin (Zeo) concentrations used for E. coli DH5α and A. baumannii were 25 mg/L and 250 mg/L, respectively. Selection for Zeo resistance was obtained on low-salt Luria-Bertani agar (LSLA, 10 g/L tryptone, 0.5 g/L NaCl, and 5 g/L yeast extract) to prevent inhibition of Zeo activity due to high ionic strength⁶⁶. The tetracycline (Tc) concentrations used for E. coli DH5α and A. baumannii selection were 12.5 mg/L and 50 mg/L, respectively. The kanamycin (Km) concentrations used for E. coli DH5α and A. baumannii selection were 25 mg/L and 50 mg/L, respectively.

Preparation of A. baumannii and E. coli competent cells. E. coli competent cells were prepared by the rubidium-calcium chloride method and transformed according to the heat shock protocol⁶⁵. Plasmid DNA was introduced in electrocompetent A. baumannii by electroporation as previously described⁶⁷.

Plasmid constructs. The eptA1 ORF was amplified from the A. baumannii ATCC 19606(D) genomic DNA with primers listed in Table S6 (Extended Data 1). The 1,839-bp amplicon was digested using the XhoI and HindIII and ligated to the corresponding sites of pVRL2⁴⁵, yielding the pVRL2eptA1 plasmid. The promoterless gene pmrC was amplified from nt 719,910 to nt 721,621 (pmrC_short) or from nt 719,861 to nt 721,621 (pmrC_long) using as the template the A. baumannii ATCC 19606(D) genomic DNA (available at https://genomes.atcc.org/genomes/1577c3a70f334038) and the primer pairs listed in Table S6 (Extended Data 1). The resulting amplicons of 1,727 bp and 1,775 bp were digested with XhoI and HindIII and ligated to the corresponding sites of pVRL2, yielding pVRL2pmrC_short and pVRL2pmrC_long, respectively. The pCLV promoter-probe E. coli-Acinetobacter spp. shuttle vector carrying the reporter genes for GFP, mCherry, and cyan fluorescent protein (CFP), and its derivatives pCLV1 and pCLV2, were constructed as illustrated in Figure S15. Briefly, the pRGC⁶⁸ and pLPV3Z⁶⁷ vectors were digested with KpnI and SacI (Fig. S15a, Extended Data 2) yielding five fragments: i) a 3,920 bp-pLPV3Z derived fragment encompassing the ColE1-like origin for replication in E. coli, the origin for replication in Acinetobacter spp. (oriAb), the parE2-paaA2 toxin-antitoxin systems ensuring plasmid stability in the absence of antibiotic selection⁴⁵, and the aacC1 gene for Gm resistance; ii) the 637 bp-fragment from pLPV3Z carrying the Zeo resistance cassette; iii) the 966 bp-fragment including the multiple cloning site (MCS) and the gfp reporter gene of pLPV3Z; iv) a 2,466 bp region encompassing the MCS and the gfp, mCherry, and cfp reporter genes from pRGC; v) a 3,910 bp-fragment representing the remaining portion of pRGC. The fragments of 3,910 bp, 637 bp, and 2,466 bp were ligated and introduced into E. coli DH5α by transformation for selection of clones containing pCLV on LSLA supplemented with 25 mg/L of Zeo (Fig. S15b, Extended Data 2). The promoter region of eptA1 (PeptA1) was amplified by PCR with the PeptA1 FW and PeptA1 RV primers (Table S6, Extended Data 1) using A. baumannii ATCC 19606(D) genomic DNA as the template, then directionally cloned into the SmaI/BamHI restriction sites of pCLV, yielding pCLV1 (Fig. S15c, Extended Data 2). The promoter region of pmrC (PpmrC) was amplified by PCR with PpmrC FW and PpmrC RV primers (Table S6, Extended Data 1) using A. baumannii ATCC 19606(D) genomic DNA as the template, then directionally cloned into the XhoI/ApaI unique restriction sites of pCLV1, yielding pCLV2 (Fig. S15d, Extended Data 2). The pmrAB genes and their mutated variants were amplified by PCR with the primers pmrAB FW and pmrAB RV (Table S6, Extended Data 1) from ATCC 19606(A)Φ(pCLV2) or its isogenic Col-resistant spontaneous mutants (AΦCR1- AΦCR12), respectively. The resulting amplicons were individually cloned into the EcoRI/SacI unique restriction sites of the pME6032 plasmid⁶⁹. All plasmids used in this work are listed in Table S7 (Extended Data 1).

Disruption of the pmrC gene. A 503-bp internal pmrC region was generated by PCR with pmrCTOPO FW and pmrCTOPO RV primers (Table S6, Extended Data 1), then cloned into pCR-BluntII-TOPO⁷⁰, yielding pTOPOpmrC (Table S7, Extended Data 1). pTOPOpmrC was used as an integrative suicide vector in A. baumannii ATCC 19606(A) and ATCC 19606(A)Φ, and pmrC knock-outs were selected on Km-containing plates. Disruption of pmrC was verified by PCR using the M13 FW and pmrCTOPO RV primers (Table S6, Extended Data 1) and amplicon sequencing. The pmrC knock-outs, namely ATCC 19606(A)pmrC::TOPO and ATCC 19606(A)ΦpmrC::TOPO, were routinely grown in the presence of Km to stabilize the pmrC mutation⁷⁰.

Induction of Φ19606. An overnight culture of A. baumannii ATCC 19606(D) was diluted 1:100 in Luria-Bertani broth (LB) supplemented with 1 mg/L MMC and incubated at 37°C and 180 rpm for prophage induction. When the culture reached an optical density at 600 nm (OD₆₀₀) = 0.5, the culture was centrifuged at 3,000×g for 15 min, and the supernatant containing the phage particles was filtered through a 0.22-µm pore size filter. The phage progeny was concentrated by polyethylene glycol (PEG) precipitation, as previously described⁷¹.

Generation of Φ19606-infected strains. The strain specificity of Φ19606 was preliminarily determined by a spot test⁷² on A. baumannii strains belonging to different STs (Table S1, Extended Data 1). Briefly, overnight A. baumannii cultures in LB medium were diluted 1:100 in fresh LB and incubated at 37°C with shaking (180 rpm) until they reached OD₆₀₀ = 0.5. CAMHA soft medium was previously prepared and maintained at 50°C. Four mL of pre-warmed (50°C) CAMHA soft [0.4% (w/v)] agar were mixed with 100 µL of the A. baumannii culture and poured on the top of a CAMHA plate. Ten-µL aliquots of the phage suspension were spotted on the plate and incubated at 37°C for 12 h before checking for the appearance of turbid plaques, denoting bacterial lysis.

Φ19606-infected A. baumannii strains were isolated from turbid plaques by streaking on CAMHA, and single bacterial colonies for each strain were inoculated in CAMHB, incubated at 37°C for 16 h, and tested for the presence of Φ19606 by PCR on purified genomic DNA using the primers listed in Table S6 (Extended Data 1).

TEM imaging. The turbid plaques derived from the spot test on A. baumannii ATCC 19606(A) were cut using a sterile scalpel, immersed in 500 µl of TM buffer (Tris HCl 10 mM; MgSO₄ 10 mM; pH = 7), and phage particles were suspended by repeating pipetting. To visualize Φ19606 virions adsorbed onto A. baumannii ATCC 19606(A) cells, 15 µl of the PEG-precipitated phage suspension from ATCC 19606(D) were mixed with an ATCC 19606(A) culture at OD₆₀₀ = 0.5 in LB supplemented with 10 mM MgSO₄ at 37°C for 15 min. After fixation with 15% (v/v) formaldehyde, 7 µL of the phage/cell suspension was deposited onto Pelco carbon and formvar coated 400 mesh copper grids (Ted Pella, Redding, California, US) and left to adsorb for 5 min. Grids were rinsed several times with water and negatively stained with aqueous 0.5% (w/v) uranyl acetate. Observations and photographs were made using a Philips CM 10 transmission electron microscope (Eindhoven, The Netherlands), operating at 60 kV. Micrograph films were developed and digitally acquired at high resolution with a D800 Nikon camera. Images were trimmed and adjusted for brightness and contrast using ImageJ v.1.53c⁷³.

CLSM imaging and automatic bacterial cell counting. Bacterial cells were deposited on a microscope glass slide overlaid with 0.5% (w/v) agarose and imaged with a Nikon A1R CLSM equipped with an Apo TIRF 100× oil immersion objective (NA 1.49). The 488 and 561 nm laser lines were employed for the GFP and mCherry excitation, respectively. Emission bandwidths at 500–550 nm and 600–720 nm were used for GFP and mCherry detection, respectively. Images were acquired at a sampling dimension of 1024×1024 pixels, then deconvoluted using the NIS-Elements software (Nikon) with default settings.

The number of fluorescent cells was determined using the ImageJ software v.1.53c, according to a previously established pipeline⁷⁴. The number of non-fluorescent cells was visually determined on differential interference contrast (DIC) images using the multi-point tool of ImageJ⁷³. The average number (± SD) of fluorescent- and non-fluorescent cells was determined for fifteen CLSM images for each condition.

A. baumannii biofilms were stained with acridine orange (AO) and visualized using Nikon A1R CLSM equipped with a CFI Plan Fluor 20× objective (NA 0.75). 488 nm and 500–550 nm were employed as AO excitation and emission bandwidth, respectively. The images were acquired at a sampling dimension of 512×512 pixels and the number of stacks was adjusted according to the biofilm thickness.

Biofilm formation and structural characterization. Bacteria were grown in CAMHB for 16 h at 37°C or, when needed, in CAMHB supplemented with 100 mg/L Gm and 4% (w/v) arabinose. Cells were diluted to OD₆₀₀ = 0.01 (corresponding to ca 5×10⁶ CFU/mL) in CAMHB or, when needed, CAMHB supplemented with 4% (w/v) arabinose, then incubated at 37°C for 48 h. Biofilm formation was measured using the standard crystal violet assay in 96-well polystyrene plates, as described previously⁷⁵. In addition, biofilms were grown on glass chamber slides (4-well on lumox detachable, Sarstedt), stained with 0.1% (w/v) AO, and observed with a CLSM as previously reported⁷⁵. Biofilm spatial characteristics were determined for at least five image stacks using COMSTAT v.2.1⁷⁶ with default image processing parameters.

Search for Ф19606 in A. baumannii genomes, Multilocus Sequence Typing (MLST), and K-locus typing. A complete list of A. baumannii genomes was retrieved from the NCBI database (accessed on 14th July 2023). Complete genomes with RefSeq annotation were selected. Redundant genomes were trimmed from the dataset, and only genomes with the highest sequence quality and coverage were retained. The ST was assigned using the MLST software available at: https://github.com/tseemann/mlst, using the Institut Pasteur scheme for A. baumannii. Genomes with undefined ST were removed from the dataset. The resulting dataset includes 523 whole-genome sequences (Table S8, Extended Data 1). The 52,671-bp sequence of Ф19606 from A. baumannii ATCC 19606(D)⁴³ was used as a query for dataset interrogation. A canonical Ф19606 was present for E-value < e^− 10 and > 95% identity across at least 95% of the total sequence length (Tables S9, Extended Data 1). Since these criteria could underestimate Ф19606 carriage due to major mutations in phage sequence, an additional BLASTn search was performed using as a query a 6,509-bp sequence from ATCC 19606(D), spanning from the gene encoding the TonB-dependent receptor (CAHBGAOC_02668) to the gene coding for the site-specific Ф19606 integrase (CAHBGAOC_02664), located upstream and downstream eptA1, respectively. This region was assumed to be present for E-value < e^− 10 and > 95% identity across at least 95% of the total sequence length (Tables S10, Extended Data 1). Genomes carrying: i) the 6509-bp region encompassing eptA1 but lacking the entire 52,671-bp Ф19606 region, and ii) the entire 52,671-bp Ф19606 sequence but presenting an IS in the eptA1 flanking regions were inspected using Snapgene software and classified as carrying Ф19606 mutant variants.

To generate the EptA1 protein consensus, EptA1 sequences were retrieved from all strains carrying Ф19606 mutant variants, except those presenting premature stop codons in the EptA1 sequence. A total of 52 sequences were aligned using Clustal Omega⁷⁷, including the reference sequence of EptA1 from ATCC 19606(D), and a consensus for EptA1 was generated using WebLogo (http://weblogo.threeplusone.com/).

Kaptive v. 2.0.1 was employed to assign a K-type to those strains of the dataset displaying a “good” or superior K-locus score⁷⁸.

Protein modeling and structural comparisons. Ab initio 3D models of PmrB, PmrC, and EptA1 protein structures were obtained using the Robetta software⁷⁹. Match marker analyses and superimposition of proteins were performed using the UCSF Chimera software⁸⁰. Amino acid sequences were aligned with Clustal Omega⁸¹, and the alignment was drawn with ESPript⁸².

Col time-kill assay. A. baumannii strains infected or not with Φ19606 were grown in CAMHB for 16 h at 37°C. The bacterial suspension was adjusted at OD₆₀₀ = 0.01 in CAMHB supplemented with 1.0 and 2.0 mg/L Col and incubated at 37°C. Aliquots (1-mL each) were taken at 0, 3, 6, and 9 h, centrifuged, and plated on CAMHA for viable (CFU) counts.

Col susceptibility testing and checkerboard assays. Col susceptibility testing was performed using the broth microdilution method according to the EUCAST guidelines (www.eucast.org) in 96-well polystyrene microplates (Sarstedt). Col was tested in the range of 0.25–512 mg/L, and each well was inoculated with 5×10⁵ CFU/mL. The Col MIC was visually determined after 18-h incubation at 37°C. Checkerboard assays were performed in CAMHB as previously reported⁸³ with at least three biological replicates. The MIC values reported in this study are the median of at least three biological replicates.

Lipid A analysis by mass spectrometry. The MALDIxin procedure for semi-quantitative assessment of PetN addition to lipid A was performed as previously described, with minor modifications⁴⁸. Briefly, bacteria were grown in CAMHB supplemented with 100 mg/L Gm with or without 4% (w/v) arabinose, washed with saline, suspended in CAMHB, and heat-inactivated at 80°C for 1 h. Two hundred µL of the suspensions were washed twice with double-distilled water, and the bacterial pellet was subjected to mild-acid hydrolysis in 400 µL of 1% (v/v) acetic acid for 60 min at 98°C. Hydrolyzed cells were centrifuged at 15,000×g for 5 min, washed twice and suspended in double-distilled water at McFarland = 20. The suspension (0.4 µL) was loaded onto the MALDI target plate and overlaid with a Norharmane matrix (1.2 µL; Sigma-Aldrich) solubilized in chloroform/methanol 90:10 (v/v) to a final concentration of 10 g/L. Samples were loaded onto MSP 96 polished steel BC targets (Bruker). Spectra were recorded in the linear negative-ion mode (laser intensity 95%, ion source 1 = 10.00 kV, ion source 2 = 8.98 kV, lens = 3.00 kV, detector voltage = 2652 V, pulsed ion extraction = 150 ns), and processed with the FlexAnalysis v.3.4 software (Bruker Daltonik), using default parameters. All mass spectra were analyzed for three independent bacterial cultures. The polymixin resistance ratio (PRR) was determined as previously reported⁴⁸.

In vitro competition assay. Competition experiments between pairs of A. baumannii strains were performed by inoculating 5×10⁵ CFU/mL of ATCC 19606(A) carrying alternatively pVRL2eptA1, pVRL2pmrC_long, or pVRL2pmrC_short. Co-cultures were made in CAMHB or in CAMHB, both supplemented with 4% (w/v) arabinose. A. baumannii ATCC 19606(A)(pVRL2pmrC_short) was used as the control to ensure a similar metabolic burden due to protein overexpression. Co-cultures were grown at 37°C for up to 12 h, then serially diluted in saline and plated onto CAMHA containing 4% (w/v) arabinose and 0, 2, or 16 mg/L Col, to assess the number of total cells, and the number of ATCC 19606(A) cells harboring either pVRL2pmrC_long or pVRL2eptA1, respectively. The C.I. was calculated as described elsewhere⁸⁴.

Membrane integrity, ATP measurements, and resazurin-based assay of metabolic activity. The A. baumannii ATCC 19606(A) carrying alternatively pVRL2eptA1, pVRL2pmrC_long, and pVRL2pmrC_short, and pVRL2 were grown in CAMHB supplemented with 100 mg/L Gm for 16 h at 37°C. Cultures were washed twice and diluted to OD₆₀₀ = 0.1 in PBS supplemented with 10 µM SYTO 9 and 60 µM propidium iodide (PI, ThermoFisher Scientific), with or without 4% (w/v) arabinose. The fluorescent emission of SYTO 9- and PI-stained samples was recorded during incubation at 37°C using a Spark 10M (Tecan) microplate reader at the excitation/emission wavelengths of 480/500 nm and 515/635 nm, respectively. Membrane integrity was expressed as the ratio between green/red fluorescent emissions⁸⁵. To estimate the ATP content, the same strains were grown for 4 h at 37°C in CAMHB supplemented with 100 mg/L Gm, and 50 µl of bacterial suspensions were diluted with an equal volume of the luciferase-containing solution of BacTiter-Glo Microbial Cell Viability Assay (Promega) and incubated at room temperature in the dark for 5 min. The OD₆₀₀ and bioluminescent emission of the bacterial suspension were measured in a Spark 10M (Tecan) microplate reader. The ATP content was expressed as the ratio between arbitrary luminescence units (A.L.U) and the OD₆₀₀ of the bacterial suspension. To estimate the metabolic activity, bacterial suspensions were diluted to OD₆₀₀ = 0.2 in CAMHB containing 10% (v/v) of the resazurin reagent PrestoBlue (ThermoFisher), and supplemented or not with 4% (w/v) arabinose and Col 4 mg/L. Resazurin fluorescence (excitation and emission at 560 and 600 nm, respectively) was monitored every 15 min for up to 4 h using the Spark 10M (Tecan) microplate reader. Fluorescence emission of sterile medium was subtracted from each sample. The RapidResa Polymyxin Acinetobacter NP test was performed according to the manufacturer’s instructions.

Analysis of the P eptA1 and PpmrC promoter activities. A. baumannii ATCC 19606(A)Φ carrying alternatively pCLV2 or the control pCLV vector were grown for 16 h at 37°C in CAMHB supplemented with 100 mg/L Gm. Cells were washed with saline and diluted to OD₆₀₀ = 0.1 in DCDM9 or M9 to minimize the fluorescent background of the medium⁶⁷. PeptA1 and PpmrC promoter activities were also tested in biological fluids collected as reported elsewhere^75,86. Cultures were incubated at 37°C and the OD₆₀₀ (bacterial growth), GFP (PeptA1 activity), and mCherry (PpmrC activity) emissions were measured using the Spark 10M microtiter plate reader (Tecan) at different time points. PeptA1 and PpmrC promoter activities were expressed as relative fluorescence units (R.F.U.), which were calculated by the ratios GFP emission/OD₆₀₀ and mCherry emission/OD₆₀₀, respectively. The R.F.U. calculated for medium autofluorescence and GFP and mCherry emission of ATCC 19606(A)Φ carrying the promoterless pCLV control vector were subtracted.

Selection of Col-resistant spontaneous mutants. Spontaneous Col-resistant mutants of A. baumannii were selected as previously described⁸⁷. Briefly, 100 µL of a bacterial suspension at OD₆₀₀ = 2 (corresponding to ca 1×10⁹ CFU/mL) were seeded on CAMHA plates containing 4×MIC Col (1 to 4 mg/L, depending on the strain). CFU counts were performed on CAMHA plates without Col to determine the number of input CFU. The frequency of Col-resistant spontaneous mutants was expressed as the ratio between the CFU obtained in Col-containing plates after 48-h incubation at 37°C and the input CFU.

Identification of pmrC and eptA1 transcriptional regulators. Spontaneous Col-resistant mutants of A. baumannii ATCC 19606(A)Φ(pCLV2) were selected on CAMHA plates supplemented with 4 mg/L Col. Colonies obtained after 48-h incubation at 37°C were suspended in 100 µl of saline, individually dispensed in a black, clear-bottom 96-well microplate (Greiner), and the OD₆₀₀, GFP, and mCherry emission values were measured using a Spark 10M microtiter plate reader (Tecan). Col-resistant strains presenting with GFP and mCherry relative fluorescence emission values > 1.5-fold the Col-sensitive parental strain ATCC 19606(A)Φ(pCLV2) were subjected to whole-genome sequencing.

Whole-genome sequencing of the parental strain ATCC 19606(A)Φ(pCLV2) was performed by GenProbio srl (Parma, Italy) using a NextSeq platform (Illumina, San Diego, CA, USA) and a MinION (Oxford Nanopore, UK) according to the supplier’s protocols. MinION long reads obtained from genome sequencing runs were used as input for a de novo genome assembly using Canu 2.2 with the estimated parameter ‘genomeSize’ of 4.0 m⁸⁸, generating a single complete genome sequence. Then, fastq files of Illumina paired-end reads (150 bp) and MinION long reads (ranging from 1,000 to 100,065 bp) were used as input for a second genome assembly through the MEGAnnotator pipeline⁸⁹. The SPAdes program v.3.15.0⁹⁰ was used for the hybrid assembly of the genome sequence, as previously reported⁴³. The resulting genomic sequence had a 323-fold coverage (114-fold coverage based on short-reads and 209 on long-reads) with genome completeness and contamination of 99.63 and 0.00, respectively.

The genome sequence of ATCC 19606(A)Φ(pCLV2) Col-resistant mutants (designated as AΦCR1-15) was obtained using the NextSeq platform. To detect single nucleotide polymorphisms (SNPs) and insertions or deletions (indels), reads derived from the Col-resistant strains were mapped against the ATCC 19606(A)Φ(pCLV2) reference genome with BWA mem v.0.7.17⁹¹, using default parameters. A consensus pileup was produced using SAMtools v.1.10⁹² and SNPs and indels were determined using VarScan.2.3.6⁹³ with the following parameters: minimum coverage (8), min-reads2 (2), min-avg-qual (15), min-var-freq (0.5), P value (99e^− 02). SNPs and indels were inspected using Artemis⁹⁴.

Statistics and reproducibility. Statistical analysis was performed with the GraphPad Instat software (GraphPad Software, Inc., La Jolla, CA). Data were analyzed using a two-tailed unpaired student’s t-test. Differences having a P ≤ 0.05 were considered statistically significant. All experiments were repeated at least three times. All replicates shown are biological replicates unless specified in the text.

Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability

The data supporting the findings in this study are provided within the Article and its Supplementary Information files. Source data are provided in this paper. The sequence data of the ATCC 19606(A)Φ(pCLV2) and the derivative Col-resistant mutants used in this study are freely available from the NCBI BioProject database under accession number PRJNA837445. The full-length sequence of the pCLV plasmid has been deposited in the GenBank database under accession number ON924783.

World Health Organization. Ten threats to global health in 2019. https://www.who.int/emergencies/ten-threatsto-global-health-in-2019 2019 (2019).
Karageorgopoulos, D. E. & Falagas, M. E. Current control and treatment of multidrug-resistant Acinetobacter baumannii infections. Lancet Infect Dis 8, 751–762 (2008).
Centers for Disease Control and Prevention. Antibiotic resistance threats in the United States, https://www.cdc.gov/drugresistance/biggest-threats.html (2019).
Clancy, C. J., Schwartz, I. S., Kula, B. & Nguyen, M. H. Bacterial superinfections among persons with coronavirus disease 2019: a comprehensive review of data from postmortem studies. Open Forum Infect Dis 8, ofab065 (2021).
Yang, S. et al. Bacterial and fungal co-infections among COVID-19 patients in intensive care unit. Microbes Infect 23, 104806 (2021).
Tacconelli, E. et al. Discovery, research, and development of new antibiotics: the WHO priority list of antibiotic-resistant bacteria and tuberculosis. Lancet Infect Dis 18, 318–327 (2018).
European Society of Clinical Microbiology and Infectious Diseases (ESCMID) guidelines for the treatment of infections caused by multidrug-resistant Gram-negative bacilli (endorsed by European Society of Intensive Care Medicine). Clin Microbiol Infect 28, 521–547 (2022).
Tamma, P. D. et al. Infectious Diseases Society of America guidance on the treatment of AmpC β-Lactamase-producing Enterobacterales, carbapenem-resistant Acinetobacter baumannii, and Stenotrophomonas maltophilia Infections. Clin Infect Dis 74, 2089–2114 (2022).
Kaye, K. S. et al. Efficacy and safety of sulbactam–durlobactam versus colistin for the treatment of patients with serious infections caused by Acinetobacter baumannii–calcoaceticus complex: a multicentre, randomised, active-controlled, phase 3, non-inferiority clinical trial (ATTACK). Lancet Infect Dis (2023) doi:10.1016/S1473-3099(23)00184-6.
Shields, R. K., Paterson, D. L. & Tamma, P. D. Navigating available treatment options for carbapenem-resistant Acinetobacter baumannii-calcoaceticus complex infections. Clin Microbiol Infect 76, S179–S193 (2023).
Durante-Mangoni, E. et al. Acute kidney injury during colistin therapy: a prospective study in patients with extensively-drug resistant Acinetobacter baumannii infections. Clin Microbiol Infect 22, 984–989 (2016).
Nelson, R. E., Hyun, D., Jezek, A. & Samore, M. H. Mortality, length of stay, and healthcare costs associated with multidrug-resistant bacterial infections among elderly hospitalized patients in the United States. Clin Infect Dis 74, 1070–1080 (2022).
Qureshi, Z. A. et al. Colistin-resistant Acinetobacter baumannii: beyond carbapenem resistance. Clin Infect Dis 60, 1295–1303 (2015).
Pormohammad, A. et al. Global prevalence of colistin resistance in clinical isolates of Acinetobacter baumannii: a systematic review and meta-analysis. Microb Pathog 139, 103887 (2020).
Humphries, R. M., Janssen, H., Hey-Hadavi, J. H., Hackel, M. & Sahm, D. Multidrug-resistant Gram-negative bacilli recovered from respiratory and blood specimens from adults: the ATLAS surveillance program in European hospitals, 2018–2020. Int J Antimicrob Agents 61, 106724 (2023).
Needham, B. D. & Trent, M. S. Fortifying the barrier: the impact of lipid A remodelling on bacterial pathogenesis. Nat Rev Microbiol 11, 467–481 (2013).
Olaitan, A. O., Morand, S. & Rolain, J.-M. Mechanisms of polymyxin resistance: acquired and intrinsic resistance in bacteria. Front Microbiol 5, 643 (2014).
Chin, C.-Y., Gregg, K. A., Napier, B. A., Ernst, R. K. & Weiss, D. S. A PmrB-regulated deacetylase required for lipid A modification and polymyxin resistance in Acinetobacter baumannii. Antimicrob Agents Chemother 59, 7911–7914 (2015).
Mustapha, M. M. et al. Phylogenomics of colistin-susceptible and resistant XDR Acinetobacter baumannii. J Antimicrob Chemother 73, 2952–2959 (2018).
Da Silva, G. J. & Domingues, S. Interplay between colistin resistance, virulence and fitness in Acinetobacter baumannii. Antibiotics (Basel) 6, E28 (2017).
Lima, W. G., Alves, M. C., Cruz, W. S. & Paiva, M. C. Chromosomally encoded and plasmid-mediated polymyxins resistance in Acinetobacter baumannii: a huge public health threat. Eur J Clin Microbiol Infect Dis 37, 1009–1019 (2018).
Kamoshida, G. et al. Lipopolysaccharide-deficient Acinetobacter baumannii due to colistin resistance is killed by neutrophil-produced lysozyme. Front Microbiol 11, 573 (2020).
Trebosc, V. et al. Dissecting colistin resistance mechanisms in extensively drug-resistant Acinetobacter baumannii clinical isolates. mBio 10, e01083-19 (2019).
Gerson, S. et al. Investigation of novel pmrB and eptA mutations in isogenic Acinetobacter baumannii isolates associated with colistin resistance and increased virulence in vivo. Antimicrob Agents Chemother 63, e01586-18 (2019).
Adams, M. D. et al. Resistance to colistin in Acinetobacter baumannii associated with mutations in the PmrAB two-component system. Antimicrob Agents Chemother 53, 3628–3634 (2009).
Arroyo, L. A. et al. The pmrCAB operon mediates polymyxin resistance in Acinetobacter baumannii ATCC 17978 and clinical isolates through phosphoethanolamine modification of lipid A. Antimicrob Agents Chemother 55, 3743–3751 (2011).
Palethorpe, S., Milton, M. E., Pesci, E. C. & Cavanagh, J. Structure of the Acinetobacter baumannii PmrA receiver domain and insights into clinical mutants affecting DNA binding and promoting colistin resistance. J Biochem 170, 787–800 (2022).
Lesho, E. et al. Emergence of colistin-resistance in extremely drug-resistant Acinetobacter baumannii containing a novel pmrCAB operon during colistin therapy of wound infections. J Infect Dis 208, 1142–1151 (2013).
Liu, Y.-Y. et al. Emergence of plasmid-mediated colistin resistance mechanism MCR-1 in animals and human beings in China: a microbiological and molecular biological study. Lancet Infect Dis 16, 161–168 (2016).
Rolain, J.-M. et al. Plasmid-mediated mcr-1 gene in colistin-resistant clinical isolates of Klebsiella pneumoniae in France and Laos. Antimicrob Agents Chemother 60, 6994–6995 (2016).
Mmatli, M., Mbelle, N. M. & Osei Sekyere, J. Global epidemiology, genetic environment, risk factors and therapeutic prospects of mcr genes: a current and emerging update. Front Cell Infect Microbiol 12, 941358 (2022).
Hameed, F. et al. Plasmid-mediated mcr-1 gene in Acinetobacter baumannii and Pseudomonas aeruginosa: first report from Pakistan. Rev Soc Bras Med Trop 52, e20190237 (2019).
Ma, F. et al. Identification of a novel plasmid carrying mcr-4.3 in an Acinetobacter baumannii strain in China. Antimicrob Agents Chemother 63, e00133-19 (2019).
Al-Kadmy, I. M. S. et al. Prevalence of genes involved in colistin resistance in Acinetobacter baumannii: first report from Iraq. Microb Drug Resist 26, 616–622 (2020).
Martins-Sorenson, N. et al. A novel plasmid-encoded mcr-4.3 gene in a colistin-resistant Acinetobacter baumannii clinical strain. J Antimicrob Chemother 75, 60–64 (2020).
Kalová, A., Gelbíčová, T., Overballe-Petersen, S., Litrup, E. & Karpíšková, R. Characterisation of colistin-resistant Enterobacterales and Acinetobacter strains carrying mcr genes from Asian aquaculture products. Antibiotics (Basel) 10, 838 (2021).
Colavecchio, A., Cadieux, B., Lo, A. & Goodridge, L. D. Bacteriophages contribute to the spread of antibiotic resistance genes among foodborne pathogens of the Enterobacteriaceae family - A Review. Front Microbiol 8, 1108 (2017).
Enault, F. et al. Phages rarely encode antibiotic resistance genes: a cautionary tale for virome analyses. ISME J 11, 237–247 (2017).
Zhang, Y., Guo, Y., Qiu, T., Gao, M. & Wang, X. Bacteriophages: underestimated vehicles of antibiotic resistance genes in the soil. Front Microbiol 13, 936267 (2022).
Blanco-Picazo, P. et al. Dominance of phage particles carrying antibiotic resistance genes in the viromes of retail food sources. ISME J 17, 195–203 (2023).
Wachino, J., Jin, W., Kimura, K. & Arakawa, Y. Intercellular transfer of chromosomal antimicrobial resistance genes between Acinetobacter baumannii strains mediated by prophages. Antimicrob Agents Chemother 63, e00334-19 (2019).
López-Leal, G., Santamaria, R. I., Cevallos, M. Á., Gonzalez, V. & Castillo-Ramírez, S. Letter to the Editor: prophages encode antibiotic resistance genes in Acinetobacter baumannii. Microb Drug Resist 26, 1275–1277 (2020).
Artuso, I. et al. Genome diversity of domesticated Acinetobacter baumannii ATCC 19606^T strains. Microb Genom 8, (2022).
Burmeister, A. R. & Turner, P. E. Trading-off and trading-up in the world of bacteria-phage evolution. Curr Biol 30, R1120–R1124 (2020).
Lucidi, M. et al. New shuttle vectors for gene cloning and expression in multidrug-resistant Acinetobacter species. Antimicrob Agents Chemother 62, e02480-17 (2018).
Beceiro, A. et al. Phosphoethanolamine modification of lipid A in colistin-resistant variants of Acinetobacter baumannii mediated by the pmrAB two-component regulatory system. Antimicrob Agents Chemother 55, 3370–3379 (2011).
Anandan, A. et al. Structure of a lipid A phosphoethanolamine transferase suggests how conformational changes govern substrate binding. Proc Natl Acad Sci U S A 114, 2218–2223 (2017).
Dortet, L. et al. Rapid detection of colistin resistance in Acinetobacter baumannii using MALDI-TOF-based lipidomics on intact bacteria. Sci Rep 8, 16910 (2018).
Clifton, L. A. et al. Effect of divalent cation removal on the structure of Gram-Negative bacterial outer membrane models. Langmuir 31, 404–412 (2015).
Gogry, F. A., Siddiqui, M. T., Sultan, I. & Haq, Q. Mohd. R. Current update on intrinsic and acquired colistin resistance mechanisms in bacteria. Front Med 8, (2021).
Jahnen-Dechent, W. & Ketteler, M. Magnesium basics. Clin Kidney J 5, i3–i14 (2012).
Girardello, R., Bispo, P. J. M., Yamanaka, T. M. & Gales, A. C. Cation concentration variability of four distinct mueller-hinton agar brands influences polymyxin b susceptibility results. J Clin Microbiol 50, 2414–2418 (2012).
Dominguez, L. J., Veronese, N., Guerrero-Romero, F. & Barbagallo, M. Magnesium in infectious diseases in older people. Nutrients 13, 180 (2021).
Chen, X. et al. Serum calcium levels and in-hospital infection risk in patients with acute ischemic stroke. NDT 18, 943–950 (2022).
Beceiro, A. et al. Biological cost of different mechanisms of colistin resistance and their impact on virulence in Acinetobacter baumannii. Antimicrob Agents Chemother 58, 518–526 (2014).
Antunes, L. C. S., Visca, P. & Towner, K. J. Acinetobacter baumannii: evolution of a global pathogen. Pathog Dis 71, 292–301 (2014).
Harding, C. M., Hennon, S. W. & Feldman, M. F. Uncovering the mechanisms of Acinetobacter baumannii virulence. Nat Rev Microbiol 16, 91–102 (2018).
Dafopoulou, K. et al. Colistin-resistant Acinetobacter baumannii clinical strains with deficient biofilm formation. Antimicrob Agents Chemother 60, 1892–1895 (2015).
Chaturongakul, S. & Ounjai, P. Phage–host interplay: examples from tailed phages and Gram-negative bacterial pathogens. Front Microbiol 5, (2014).
Gordillo Altamirano, F. et al. Bacteriophage-resistant Acinetobacter baumannii are resensitized to antimicrobials. Nat Microbiol 6, 157–161 (2021).
Knecht, L. E., Veljkovic, M. & Fieseler, L. Diversity and function of phage-encoded depolymerases. Front Microbiol 10, (2020).
Bondy-Denomy, J. et al. Prophages mediate defense against phage infection through diverse mechanisms. ISME J 10, 2854–2866 (2016).
Hackel, M. A. et al. In vitro activity of the siderophore cephalosporin, cefiderocol, against carbapenem-nonsusceptible and multidrug-resistant isolates of gram-negative bacilli collected worldwide in 2014 to 2016. Antimicrob Agents Chemother 62, e01968-17 (2018).
Hijazi, S. et al. Antimicrobial activity of gallium compounds on ESKAPE pathogens. Front Cell Infect Microbiol 8, (2018).
Sambrook, J., Fritsch, E.F., Maniatis, T. Molecular cloning: a laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y. (1989).
Luna, B. M. et al. Selectable markers for use in genetic manipulation of extensively drug-resistant (XDR) Acinetobacter baumannii HUMC1. mSphere 2, e00140-17 (2017).
Lucidi, M. et al. New shuttle vectors for real-time gene expression analysis in multidrug-resistant Acinetobacter species: in vitro and in vivo responses to environmental stressors. Appl Environ Microbiol 85, e01334-19 (2019).
Mellini, M. et al. Generation of genetic tools for gauging multiple-gene expression at the single-cell level. Appl Environ Microbiol 87, e02956-20 (2021).
Heeb, S. et al. Small, stable shuttle vectors based on the minimal pVS1 replicon for use in gram-negative, plant-associated bacteria. Mol Plant Microbe Interact 13, 232–237 (2000).
Aranda, J. et al. A rapid and simple method for constructing stable mutants of Acinetobacter baumannii. BMC Microbiol 10, 279 (2010).
Hockett, K. L. & Baltrus, D. A. Use of the soft-agar overlay technique to screen for bacterially produced inhibitory compounds. J Vis Exp 55064 (2017) doi:10.3791/55064.
Kropinski, A. M., Mazzocco, A., Waddell, T. E., Lingohr, E. & Johnson, R. P. Enumeration of bacteriophages by double agar overlay plaque assay. in Bacteriophages: Methods and Protocols, Volume 1: Isolation, Characterization, and Interactions (eds. Clokie, M. R. J. & Kropinski, A. M.) 69–76 (Humana Press, 2009). doi:10.1007/978-1-60327-164-6_7.
Schneider, C. A., Rasband, W. S. & Eliceiri, K. W. NIH Image to ImageJ: 25 years of image analysis. Nat Methods 9, 671–675 (2012).
Nichele, L., Persichetti, V., Lucidi, M. & Cincotti, G. Quantitative evaluation of ImageJ thresholding algorithms for microbial cell counting. OSA Continuum, OSAC 3, 1417–1427 (2020).
Runci, F., Bonchi, C., Frangipani, E., Visaggio, D. & Visca, P. Acinetobacter baumannii biofilm formation in human serum and disruption by gallium. Antimicrob Agents Chemother 61, e01563-16 (2016).
Heydorn, A. et al. Quantification of biofilm structures by the novel computer program COMSTAT. Microbiology (Reading) 146, 2395–2407 (2000).
Madeira, F. et al. Search and sequence analysis tools services from EMBL-EBI in 2022. Nucleic Acids Res 50, W276–W279 (2022).
Cahill, S. M., Hall, R. M. & Kenyon, J. J. An update to the database for Acinetobacter baumannii capsular polysaccharide locus typing extends the extensive and diverse repertoire of genes found at and outside the K locus. Microb Genom 8, mgen000878 (2022).
Kim, D. E., Chivian, D. & Baker, D. Protein structure prediction and analysis using the Robetta server. Nucleic Acids Res 32, W526-531 (2004).
Pettersen, E. F. et al. UCSF Chimera—A visualization system for exploratory research and analysis. J Comput Chem 25, 1605–1612 (2004).
Sievers, F. et al. Fast, scalable generation of high-quality protein multiple sequence alignments using Clustal Omega. Mol Syst Biol 7, 539 (2011).
Robert, X. & Gouet, P. Deciphering key features in protein structures with the new ENDscript server. Nucleic Acids Res 42, W320-324 (2014).
Geisinger, E. et al. Antibiotic susceptibility signatures identify potential antimicrobial targets in the Acinetobacter baumannii cell envelope. Nat Commun 11, 4522 (2020).
Lo Sciuto, A., Cervoni, M., Stefanelli, R., Mancone, C. & Imperi, F. Effect of lipid A aminoarabinosylation on Pseudomonas aeruginosa colistin resistance and fitness. Int J Antimicrob Agents 55, 105957 (2020).
D’Agostino, I. et al. Antibacterial alkylguanidino ureas: molecular simplification approach, searching for membrane-based MoA. Eur J Med Chem 114158 (2022).
Visaggio, D. et al. A highly sensitive luminescent biosensor for the microvolumetric detection of the Pseudomonas aeruginosa siderophore pyochelin. ACS Sens. 6, 3273–3283 (2021).
Lo Sciuto, A. & Imperi, F. Aminoarabinosylation of lipid A is critical for the development of colistin resistance in Pseudomonas aeruginosa. Antimicrob Agents Chemother 62, e01820-17 (2018).
Koren, S. et al. Canu: scalable and accurate long-read assembly via adaptive k-mer weighting and repeat separation. Genome Res 27, (2017).
Lugli, G. A., Milani, C., Mancabelli, L., van Sinderen, D. & Ventura, M. MEGAnnotator: a user-friendly pipeline for microbial genomes assembly and annotation. FEMS Microbiol Lett 363, (2016).
Bankevich, A. et al. SPAdes: a new genome assembly algorithm and its applications to single-cell sequencing. J Comput Biol 19, 455–477 (2012).
Li, H. & Durbin, R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics 25, 1754–1760 (2009).
Li, H. A statistical framework for SNP calling, mutation discovery, association mapping and population genetical parameter estimation from sequencing data. Bioinformatics 27, 2987–2993 (2011).
Koboldt, D. C. et al. VarScan 2: somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Res 22, 568–576 (2012).
Carver, T., Harris, S. R., Berriman, M., Parkhill, J. & McQuillan, J. A. Artemis: an integrated platform for visualization and analysis of high-throughput sequence-based experimental data. Bioinformatics 28, 464–469 (2012).

Competing interests

All authors declare no competing interests.

Author contributions

ML, GC, IA, and CS collected the data. ML drafted the manuscript. FI, GR, LV, and PV revised the draft manuscript. ML, DV, GC, FI, and PV made substantial contributions to the conception and design of the experiments. All authors had access to all the data, approved the final version of the manuscript, and agreed to submit it for publication.

Acknowledgments

We would like to thank Prof. Jeremy J. Barr (Monash University, Clayton, Victoria, Australia) for the helpful discussion, and Dr. Gerald J Larrouy-Maumus (Imperial College, London, UK) for the MALDIxin service.

Table 1: MIC and PetN transferase activity in A baumannii ATCC 19606(A) carrying arabinose-inducible eptA1, pmrC_long, and pmrC_short genes

*A baumannii* strain	Arabinose (%)	Col MIC (mg/L)^a	PetN addition (PRR)^b
ATCC 19606(A)	0	1	0.00
ATCC 19606(A)	4	1	0.00
ATCC 19606(A)Φ	0	1	0.00
ATCC 19606(A)Φ	4	1	0.00
ATCC 19606(A)(pVRL2eptA1)	0	1	0.00
ATCC 19606(A)(pVRL2eptA1)	4	64	54.72±11.52
ATCC 19606(A)(pVRL2pmrC_long)	0	1	0.00
ATCC 19606(A)(pVRL2pmrC_long)	4	4	25.51±5.90
ATCC 19606(A)(pVRL2pmrC_short)	0	1	0.00
ATCC 19606(A)(pVRL2pmrC_short)	4	1	0.00
ATCC 19606(A)(pVRL2)	0	1	0.00
ATCC 19606(A)(pVRL2)	4	1	0.00

^aMIC values in bold indicate Col resistance (Col breakpoints ≥ 2 mg/L).

^bPetN addition to the lipid A was determined according to the MALDIxin spectra reported in Fig S6 (Extended Data 2) and expressed as Polymixin Resistance Ratio (PRR) according to Dortet et al., 2018⁴⁸. PRR is the ratio of the sum of the intensities of the peaks associated with modified lipid A (for A. baumannii m/z 2,033) to the intensity of the peak of native lipid A (for A. baumannii m/z 1,910).

Extendeddata1.xlsx
Extended Data 1
Extendeddata2.pdf
Extended Data 2

Download PDF

Journal Publication

published 27 Jan, 2024

Read the published version in Drug Resistance Updates →

Version 1

posted

You are reading this latest preprint version

Phage-mediated transmission of colistin resistance in Acinetobacter baumannii

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results and Discussion

Φ19606 increases the emergence of Col-resistant mutants

Conclusions

Methods

Data availability

References

Declarations

Competing interests

Author contributions

Acknowledgments

Tables

Supplementary Files

Status:

Journal Publication

Version 1