Genome-wide phage susceptibility analysis in Acinetobacter baumannii reveals capsule modulation strategies that determine phage infectivity

Abstract

Phage have gained renewed interest as an adjunctive treatment for life-threatening infections with the resistant nosocomial pathogen Acinetobacter baumannii. Our understanding of how A. baumannii defends against phage remains limited, although this information could lead to improved antimicrobial therapies. To address this problem, we identified genome-wide determinants of phage susceptibility in A. baumannii using Tn-seq. These studies focused on the lytic phage Loki, which targets Acinetobacter by unknown mechanisms. We identified 41 candidate loci that increase susceptibility to Loki when disrupted, and 10 that decrease susceptibility. Combined with spontaneous resistance mapping, our results support the model that Loki uses the K3 capsule as an essential receptor, and that capsule modulation provides A. baumannii with strategies to control vulnerability to phage. A key center of this control is transcriptional regulation of capsule synthesis and phage virulence by the global regulator BfmRS. Mutations hyperactivating BfmRS simultaneously increase capsule levels, Loki adsorption, Loki replication, and host killing, while BfmRS-inactivating mutations have the opposite effect, reducing capsule and blocking Loki infection. We identified novel BfmRS-activating mutations, including knockouts of a T2 RNase protein and the disulfide formation enzyme DsbA, that hypersensitize bacteria to phage challenge. We further found that mutation of a glycosyltransferase known to alter capsule structure and bacterial virulence can also cause complete phage resistance. Finally, additional factors including lipooligosaccharide and Lon protease act independently of capsule modulation to interfere with Loki infection. This work demonstrates that regulatory and structural modulation of capsule, known to alter A. baumannii virulence, is also a major determinant of susceptibility to phage.

Fig 1. Identification of candidate determinants of susceptibility to Loki using Tn-seq.

(A) Loki infection of A. baumannii 17978 causes partial inhibition of bulk culture growth. Bacteria were seeded in microplates with or without Loki at the indicated MOI and growth was monitored by OD readings (liquid challenge assay). Data points show geometric mean ± s.d. (shaded bands) from n = 3 independent cultures. Where not visible, s.d. is within the confines of the symbol. Inset shows small, turbid plaques resulting from Loki infection of the same strain on solid medium. Plaques imaged with white light transillumination. Scale bar, 1 mm. (B) Tn-seq analysis of Loki susceptibility determinants in A. baumannii 17978. Volcano plots show change in gene-level fitness resulting from phage treatment at the indicated MOI compared to untreated control, plotted against inverse p value from parallel t tests. Blue data points are genes that pass significance criteria (Materials and Methods) and represent candidate Loki susceptibility determinants. Loci described further in the text are labelled. Grey data points are genes not passing significance criteria.

Fig 2. Validation of candidate phage hypersusceptibility loci.

(A) Tracks show Tn-seq fitness values (vertical bars) of individual Mariner transposon mutants at candidate Loki hypersusceptibility loci (shaded yellow) across all tested mutant banks using the indicated MOI. (B-C) Liquid challenge assays with defined mutants. Growth of the indicated A. baumannii 17978 strains with Loki (MOI 1) or no phage was measured as in Fig 1A (n = 3). “Chr” indicates the chromosomal mutation; “Plasmid” denotes the gene reintroduced under P(IPTG) control via vector pYDE152;—indicates control plasmid with no reintroduced gene; bfmS* refers to bfmS-null point mutant EGA127; IPTG “0.1” refers to mM units. (D-E) Plaque formation assays. ∼10⁸ of the indicated bacteria were overlaid on bottom agar, followed by spotting 10-fold serial dilutions of Loki. Bottom agar in D contained 1mM IPTG. Plaques were imaged with white-light transillumination after 16 h incubation at 30°C.

Fig 3. Loki resistance mutations map to capsule biosynthesis locus.

(A) Tn-seq fitness of transposon mutants in the capsule synthesis (KL3) and associated loci in the Loki challenge screen. Plotted are the average gene-level Tn-seq fitness values (W) with phage infection (Loki MOI 10–15) vs uninfected control, analyzed by unpaired t-tests. X indicates essential gene without a fitness value due to paucity of transposon insertions. Dotted line indicates the genome-wide average W during Loki challenge. (B-C, E) Plaque formation assays showing dependence on the K3 capsule for Loki infection. Loki was spotted on the following hosts: (B) acapsular itrA and wzc mutants of 17978; (C) a range of A. baumannii clinical isolates harboring the indicated K and OC loci (K loci producing the K3 capsule type are shaded blue); or (E) strains with gtr6 mutations (S1 Table). Strains in E harbored the indicated pYDE152-based plasmid and bottom agar contained 1mM IPTG. (D) Diagram of the KL3/KL22 cluster showing spontaneous Loki resistance mutations (arrowheads). The hotspot within gtr6 is expanded. KL3 and KL22 share all ORFs except pgt1 (only present in KL22).

Fig 4. Results of phage adsorption assays measuring A. baumannii binding to Loki.

Loki adsorption assays were performed with WT 17978 or the indicated mutant (S1 Table). Data points show geometric mean ± s.d. (n = 3), analyzed by unpaired t test (mutant vs WT control). P values: *, ≤0.05; **, ≤0.01; ***, <0.001; ns, not significant.

Fig 5. Susceptibility to Loki depends on BfmRS activity and correlates with control of capsule levels.

(A) bfmS and bfmR transposon mutations have opposing effects on Tn-seq fitness during Loki challenge. Data points show average gene-level Tn-seq fitness values (W) with Loki infection at indicated MOI vs uninfected control, analyzed by unpaired t-tests. Dotted line indicates the genome-wide average W during Loki challenge (MOI 10–15). (B) Capsule production by different bfmRS mutants. Bars show mean capsule level ± s.d. (n = 3), analyzed by one-way ANOVA with Dunnett’s multiple comparisons test (mutant vs WT). A representative gel is shown above the graph. (C) Liquid challenge and (D) plaque formation assays with ∆bfmRS and bfmR receiver domain point mutants. Liquid challenge data are presented as in Fig 1A (n = 3). (E) BfmR phosphorylation level was analyzed in cell lysates of receiver domain mutants vs the parent control (bfmS*) by Phos-tag western blotting. BfmR∼P as a fraction of total BfmR was quantified from n = 3 samples. Bars show mean ± s.d., analyzed by one-way ANOVA with Tukey’s multiple comparisons test. A representative blot is shown above the graph.

Fig 6. RnaA and DsbA modulate capsule production and Loki susceptibility through BfmRS and in different clinical isolates.

(A) Capsule levels measured in 17978 derivatives with ∆rnaA and ∆dsbA mutation singly and in combination with ∆bfmRS. A representative gel (top) and quantification from n = 3 samples (bottom; means ± s.d. analyzed by one-way ANOVA with Šídák’s multiple comparisons test) are shown as in Fig 5B. (B) Liquid challenge assays examining susceptibility of strains in A to Loki attack. Data presented as in Fig 1A (n = 6). (C-D) ∆rnaA and ∆dsbA derivatives of carbapenem-resistant A. baumannii strain BAA-1790 show increased capsule and Loki sensitivity. Capsule levels (C) were analyzed as in A. Bacterial growth during liquid challenge with a virulent Loki derivative (Loki*) (D) was analyzed as in B (n = 3). (E-F) rnaA overexpression via high-copy plasmid decreases capsule production and blocks Loki infection in a manner dependent on the BfmS signal receptor. WT or bfmS* 17978 bacteria harbored pJE127 (IPTG-dependent rnaA) and were grown in the presence or absence of IPTG. Representative capsule gel (E, top), capsule quantification from n = 3 samples (E, bottom), and Loki plaque assays (F) are shown. Mean capsule levels ± s.d. were analyzed by unpaired t-tests. P values: *, ≤0.05; **, ≤0.01; ***, <0.001. ns, not significant.

Fig 7. RnaA/DsbA deficiency activates BfmRS.

(A) Transcription levels of capsule biosynthesis genes wza (black) and gnaA (gray) measured by qRT-PCR in ∆rnaA and ∆dsbA mutants. Bars show mean fold change vs WT ± s.d. (n = 3); analyzed by unpaired t-tests comparing mutant vs WT. (B-C) Fluorescence reporter assays. Strains harbored a GFP transcriptional fusion to the noted promoter region and reporter fluorescence was measured as fluorescence units/A₆₀₀. (B) ∆rnaA and ∆dsbA activation of a BfmRS-dependent (18040p, black), but not BfmRS-independent (adcp, gray) promoter. (C) ∆rnaA/∆dsbA activation of 18040p requires BfmRS. Bars show mean ± s.d. (n = 3). Means were analyzed by unpaired t test comparing mutant vs WT (B) or by one-way ANOVA with Šídák’s multiple comparisons test (C). (D) Phos-tag analysis of BfmR phosphorylation. ∆rnaA and ∆dsbA increase BfmR∼P to a level matching that caused by the hyperactive ∆bfmS allele. Presented as in Fig 5E (n = 3). Asterisks show results of one-way ANOVA with Tukey’s multiple comparisons test (mutant vs WT); comparisons between mutants were not significant. (E) Heat map shows the Pearson correlation coefficients (r) analyzing relatedness of Tn-seq phenotypic signatures [41] of bfmS, rnaA, and dsb genes. (F) Tn-seq fitness of rnaA and dsb transposon mutants in the phage challenge screen. Data presented as in Fig 5A. (G) ∆dsbA but not ∆rnaA shows a synthetic growth defect with ∆bfmRS. Bacteria were incubated in LB at 37°C and growth measured as in Fig 1A (n = 3). P values: *, ≤0.05; **, ≤0.01; ***, <0.001. ns, not significant.

Fig 8. The full-length LOS outer core impedes attack by Loki.

(A) ∆gtrOC3 and ∆gtrOC4 cause enrichment of a truncated LOS intermediate. A representative LOS gel (top) and quantification from triplicate samples (bottom) are shown. Bars indicate mean normalized LOS levels ± s.d.; one-way ANOVA showed no significant difference in total LOS levels among strains (P = 0.34). (B) Tn-seq fitness of LOS outer core synthesis locus (OCL2) mutants in the phage challenge screen. Data presented as in Fig 5A. (C-D) Enhancement of Loki susceptibility in gtrOC mutants depends on capsule. (C) Plaque formation assay with ∆gtrOC3, ∆itrA, ∆gtr6, and double mutants. (D) Liquid challenge assay with strains in C (n = 3). P values: *, ≤0.05; **, ≤0.01; ***, <0.001. ns, not significant.

Fig 9. Model for modulation of phage infection by multiple susceptibility factors in A. baumannii.

Vulnerability factors that act to enhance Loki infection are shown in blue; defense factors that antagonize Loki infection are shown in red. Additional candidate defense factors, including the methyltransferase AamA or transcription factor DksA, were identified through screening but require confirmation and are not shown. (A) Loki infection of WT bacteria. Loki adsorbs in two steps: (1) interaction with the primary adsorption receptor, capsular polysaccharide of type K3, and (2) interaction with an LOS-associated secondary site. This site may be the conserved LOS inner core, or LOS-proximal capsule residues, and access is antagonized by the LOS outer core. BfmRS controls Loki infection by positively regulating both the capsule receptor (via transcriptional activation of capsule biosynthesis) and intracellular phage replication (by unknown mechanisms). Lon protease provides defense by negatively regulating intracellular phage replication, by unknown mechanisms. (B) Loki resistance arises from capsule receptor down-regulation (BfmRS deficiency), structural alteration (via mutation of hotspot gene gtr6), or complete loss (mutation of a core biosynthesis gene, e.g. itrA or wzx), blocking phage adsorption. BfmRS deficiency may also reduce phage intracellular replication. (C) By contrast, Loki hypersusceptibility arises from hyperactivity of vulnerability factors or loss of defense factors. Hyperactivation of BfmRS, defined by increased BfmR∼P (yellow halo), results from bfmS, dsbA, or rnaA mutations and may also be triggered by external stresses related to oxidative protein folding. BfmRS hyperactivity increases capsule receptor levels, enhancing phage adsorption, while also stimulating phage intracellular multiplication. Loss of LOS OC sugars increases binding of Loki to surface sites without altering its intracellular multiplication. Loss of Lon increases phage multiplication without increasing adsorption.