Comparative genomics of Acinetobacter baumannii and therapeutic bacteriophages from a patient undergoing phage therapy – Nature.com

The clinical course of the A. baumannii infection and phage treatment, known as the Patterson Case, has been described previously11. Briefly, phage treatment was initiated with two-phage cocktails, each containing four phages: cocktail PC was administered into abdominal abscess cavities through existing percutaneous drains, and cocktail IV was administered intravenously. Near the end of patient treatment, a ninth phage, AbTP3phi1, was isolated to target the phage-resistant A. baumannii strain TP3 that arose during treatment. Phage AbTP3phi1 was administered intravenously in a two-phage cocktail (IVB) in combination with one phage from cocktail IV11. As a follow up study to this phage intervention case, we determined the genomes of the phages and also of the bacterial strains that were isolated during phage treatment.

All nine phages used in the treatment cocktails were sequenced to completion and their genomes are summarized in Table1. Genome sequences of phages C2P12, C2P21 and C2P24 described as part of cocktail PC11 were determined to be identical, so phage C2P24 was renamed as phage Maestro and is used as a representative of this group. The phages described here can be categorized into two broad groups: phages Maestro, AC4, AB-Navy1, AB-Navy4, AB-Navy71 and AB-Navy97 are large (165169kb) T4-like myophages, and phage AbTP3phi1 is a 42kb Fri1-like podophage.

The six myophages can be subdivided into two clusters, with Maestro and AB-Navy71 sharing 91.6% identity and phages AC4, AB-Navy1, AB-Navy4, AB-Navy97 sharing from 93.2% to 95.5 identity (Fig.1A). Based on their sequence relationships to other A. baumannii phages and the criteria of the International Committee on Taxonomy of Viruses (ICTV)17, Maestro forms a new species in the genus Hadassahvirus, and phages AC4 and AB-Navy4 each form new species in the genus Lazarusvirus (Table1). Phages AB-Navy1, AB-Navy71 and AB-Navy97 can be assigned to species in the genus Hadassahvirus or Lazarusvirus (Table1). All of these phages are members of the Subfamily Twarogvirinae within the Family Straboviridae, which also contains the broadly-defined T4-like myophages, including the coliphage T4 itself. The 42kb podophage AbTP3phi1 is classified as a new species within the genus Friunavirus of the Subfamily Beijernickvirinae (Table1). It shares 8289% overall DNA identity, as well as genome synteny, with previously described Acinetobacter podophages, including IME200 (NC_028987), vB_AbaP_AS11 (NC_041915), Fri1 (KR149290)18 and Aci08 (NC_048081).

A DNA sequence relatedness of six T4-like myophages, showing pairwise percent DNA sequence identities as determined by ProgressiveMauve (upper section) and DNA dotplots visually representing DNA sequence alignments between phages (lower section). B Protein sequence-based relationships of 2834 Caudoviricetes phages representing all species in the ICTV taxonomy, plus the seven treatment phages and two prophages identified in strains TP1, TP2, and TP3. Unclustered singletons (17 in total) are removed from the visualization. Distinct colors are assigned at the Subfamily level; if no Subfamily was assigned, color is assigned at the Family level. Circled clusters are enlarged in CE as labeled. C Enlarged cluster representing the Autographiviridae. Nodes are colored based on their Subfamily membership, with the legend identifying prominent clades; the node representing phage AbTP3phi1 is outlined in black, and nodes representing clade-founding phages T7, phiKMV and Fri1 are labeled. D Enlarged cluster containing the T4-like subfamilies, including the Twarogvirinae; this large cluster is also linked to the T5-like Markadamsvirinae and clusters of diverse myophages including the V5-like Vequintaviridae and FelixO1-like Ounavirinae. Nodes are colored based on their Subfamily membership with the legend identifying prominent clades. Nodes representing the six treatment myophages are outlined in black, and nodes representing clade-founding phages T4, T5, FelixO1 and V5 are labeled. E Enlarged cluster containing the two prophage elements identified in strains TP1, TP2 and TP3. These prophages are not closely related to other classified phages, with the 52kb prophage 1 distantly linked to the Guernseyvirinae, and the 42kb prophage 2 related to two other unclassified siphophages.

Phage taxonomy is rapidly evolving, with multiple major recent and proposed revisions to the organization of phage taxa based on genomic relationships19. A recent global analysis of 134 Acinetobacter phage genomes placed these into eight major clusters and 38 sub-clusters, which include five proposed new subfamilies and 30 new genera20. The most abundant groups in this analysis are members of the Twarogvirinae and the Beijernickvirinae, with the vast majority of the latter (45/49 phages) falling into a single genus, the Friunavirus. A comparison of the seven treatment phages to a database constructed from species representatives of the Caudoviricetes (tailed dsDNA phages) in the ICTV taxonomy (Fig.1B) shows higher-order relationships and diversity of these phages. This analysis produced a major grouping representing the Autographiviridae, which contains AbTP3phi1 (Fig.1C) and a grouping containing the Twarogvirinae and Tevenvirinae, containing the six treatment myophages (Fig.1D). This analysis placed the five current ICTV Twarogvirinae genera into three sequence-based viral clusters, with the treatment phages placed in a viral cluster with the other members of the genera Hadassahvirus and Lazarusvirus. Likewise, all members of the genus Friunavirus were placed into a single viral cluster with AbTP3phi1.

The genome of Maestro is presented as a representative for this group of Acinetobacter myophages (Supplementary Fig.1). Maestro has a complete genome size of 169,176bp and a GC content of 36.6%. Seven tRNA genes were identified, including one that appears to specify an amber codon. Genes encoding phage integrases or proteins associated with bacterial virulence were not detected. A conserved core of 95 genes encoding proteins with direct identity to coliphage T4 (BLASTp, E<105) were identified, clustered in several regions of the genome. These include genes encoding structural proteins and proteins involved in DNA nucleotide metabolism and replication. Proteins involved in transcriptional regulation in phage T4 were found to have homologs in Maestro, which suggests Maestro follows a T4-like program of gene expression, with positive control of early, middle and late transcripts21. The holin and endolysin lysis genes in Maestro are similarly located as in T4 and have high primary structure similarity, indicating that the first two steps in lysis, the permeabilization of the inner membrane and the degradation of the cell wall are effected the same way22. The third step, disruption of the outer membrane, is accomplished in most dsDNA phages by spanin proteins23. No candidate spanins were detected in the Maestro genome, indicating that Maestro, like some other Acinetobacter phages, uses a different mechanism for OM disruption23,24. Homologs of the phage T4 RI and RIII antiholins were identified in the Maestro genome, indicating this phage has the ability to undergo T4-like lysis inhibition25. The effects of lysis inhibition in therapeutic interventions are not known, but superinfection-induced lysis inhibition delays lysis time and increases burst size in vitro and could affect in vivo phage proliferation at the site of therapeutic application. An analysis of 16 Twarogvirinae species representatives including Maestro, AB-Navy4 and AB-Navy97 by CoreGenes26 showed that 129 genes are conserved in this group, which includes the major DNA metabolism and structural functions, and a set of 41 hypothetical proteins with no identified function (Supplementary Fig.1).

During the infection process of phage T4, the long tail fibers (LTFs) bind to the phages receptor on the cell surface. In T4, the LTF is comprised of Gp34, Gp35, Gp36 and Gp37, which form the proximal LTF, two joints, and distal LTF, respectively; the distal LTF contains the phage receptor-binding function in its C-terminal domain27,28. The distal domains of the myophage LTFs, containing the predicted receptor-binding domains, were compared by multiple sequence alignment (Supplementary Fig.2) and construction of a neighbor-joining tree to determine their relationships (Fig.2). This analysis showed the myophages used in the cocktails had two different types of tail fibers, with Maestro, AC4, and Navy71 belonging to one group, and Navy1, Navy4, and Navy97 belong to the other cluster (Fig.2). This finding correlates with the phage resistance patterns observed in A. baumannii strains isolated from the patient before and during phage treatment (Table2). Strains resistant to phage AC4 were also resistant to phage Maestro and AB-Navy71, but the same strains were still partially sensitive to AB-Navy1, AB-Navy4, and AB-Navy97. Six days after the start of treatment, resistance to AB-Navy1, AB-Navy4 and AB-Navy97 was observed simultaneously. The closer relationship of the AC4 tail fiber to Maestro and AB-Navy71 likely represents a horizontal gene transfer event, as AC4 is overall more closely related to phages AB-Navy1, AB-Navy4 and AB-Navy97 (Fig.1), and this points out a limitation of using whole-genome comparisons to predict the behavior of individual phages.

The tail fibers of phages Maestro, AC4, and AB-Navy71 form one clade and the fibers of AB-Navy1, AB-Navy4 and AB-Navy97 form another.

The podophage AbTP3phi1 shows conserved protein content and synteny with other members of the Friunavirus genus, which is part of the larger group Autographiviridae that also includes the well-studied E. coli podophage T7. The genome map of AbTP3phi1 is shown in Supplementary Fig.3. As a conserved feature of this group of phages, a terminal repeat region of 396bp was identified in AbTP31 genome by the PhageTerm tool29. Like T7, these phages possess relatively small genomes of ~40kb and encode all proteins on one strand. Unlike T7, the gene encoding the RNA polymerase is located near the center of the genome, just upstream of the gene encoding the head-tail connector, an arrangement that is similar to that of phage phiKMV30. A CoreGenes analysis of 16 Friunavirus species representatives including AbTP3phi1 indicated that 29 out of 56 AbTP3phi1 protein-coding genes were conserved, which includes the DNA primase, helicase, ligase, polymerase, exo- and endonuclease, capsid, internal virion proteins, lysis proteins, the small and large terminase, and nine hypothetical proteins (Supplementary Fig.3). Not conserved are a number of hypothetical proteins (mostly located near the left end of the genome) and the C-terminal portion of the tailspike, which contains capsular depolymerase activity31. Like other known A. baumannii Fri1-like podophages, tail spike protein of AbTP3phi1 contains a pectate lyase fold (PF12708) and thus uses the bacterial capsule as its receptor, degrading the bacterial exopolysaccharide as part of its infection process20. Strains TP1, TP2 and TP3 all encode KL116 capsule loci as determined by Kaptive32, thus the AbTP3phi1 depolymerase is presumed to be active against this capsule type. As with the myophages reported in this study, spanin proteins were not found in the genome of AbTP3phi1 nor in any other A. baumannii podophage genomes24, suggesting the presence of a novel strategy for disruption of the outer membrane in these phages.

During phage treatment, A. baumannii isolates were collected from the patient via various drains or bronchial washes. These strains were tested for their phage sensitivity via plaque assays. These showed that as early as 2 days after phage administration, the efficiency of all the phages in the first two cocktails (PC and IV) was reduced when tested against the bacterial strains isolated during treatment, evident by the decreased titers on those strains compared to the initial titers observed with TP1 (Table2). In some cases, only a zone of clearing (but no individual plaques) was observed on the plates at high phage concentrations. Consistent with the myophage tail fiber protein sequence alignment (Fig.2), host resistance to phages appeared earlier with Maestro, AC4, and Navy71 as a group, and later with phages Navy1, Navy4, and Navy97 as a group. In comparison, resistance to phage AbTP3phi1 was not observed in bacterial isolates collected throughout 2 months of phage treatment, although plating efficiencies of AbTP3phi1 varied by up to three orders of magnitude on strains collected during treatment (Table2). The emergence of phage resistance early in phage treatment illustrates the potential benefits of well-characterized and rationally designed phage cocktails in treatment, which could be designed to mitigate the emergence of resistance. It also raises questions on the benefits of continued phage treatment beyond the first ~9 days, since all isolates collected after this time are fully resistant to the phage. While it is possible that the prolonged period of phage administration (over 60 days) was not required to produce the observed clinical outcome, other studies have shown that phage-insensitive mutants of A. baumannii exhibit reduced virulence33,34,35, a phenotype that has also been observed in other systems including Staphylococcus aureus36, Klebsiella pneumoniae37 and P. aeruginosa38. Thus, maintaining selection pressure for the phage-resistant phenotype may provide a benefit to continued treatment even after the pathogen has developed resistance to the treatment phage.

Some strains isolated throughout phage treatment were also tested for their antibiotic resistance profiles by traditional microtiter MIC (Supplementary Table2). At the time TP1 was isolated from the patient, they were receiving a combination of fluconazole, azithromycin, colistin, and rifampin. Shortly after TP1 isolation, meropenem was added to treatment, and shortly after the beginning of phage administration the azithromycin, colistin, and rifampin were discontinued and minocycline was initiated. The meropenem, minocycline and fluconazole treatment was continued through the end of phage treatment11. In general, the antibiotic resistance profiles of all strains isolated during the course of phage therapy remained consistent, indicating that phage therapy did not have a major impact on antibiotic resistance of the pathogen in this case. Although an initial report indicated resistance to colistin and tigecycline prior to the start of phage therapy11, sensitivity to colistin and tigecycline (in the range of 28ug/ml) was observed in strains isolated ~7 weeks after the start of phage therapy (collected on May 9, 2016). While sensitive to colistin and tigecycline, these strains were resistant to minocycline. We previously reported on a potential synergistic in vitro activity between phage cocktail and minocycline (used at sub-inhibitory concentrations of 0.25ug/ml) in inhibiting bacterial growth11. However, such results were obtained using strain TP3, and TP3 was not tested for its sensitivity to minocycline, colistin, or tigecycline in these MIC assays. Increased antibiotic sensitivity has been associated with phage resistance in organisms such as A. baumannii34,35 and P. aeruginosa39. However, some studies have observed increased antibiotic resistance in phage-resistant mutants40,41, indicating increased sensitivity to antibiotics is not a universal outcome from phage resistance and is probably dependent on the host, drug, phage, and nature of the resistance mutation. While fitness costs can be associated with phage resistance, the effects of resistance mutations can be pleiotropic with phenotypes that are not always easily predictable42.

To more fully delineate the phenotypic differences between TP1 and TP3, BioLog phenotypic microarray (PM) profiling was conducted using PM 120 (Fig.3 and Supplementary Data1). As expected given the clonal nature of the isolates, the PM demonstrated very consistent phenotypes in terms of carbon, nitrogen, phosphorus and sulfur utilization; biosynthetic pathways and nutrient stimulation; osmotic/ionic response; and pH response; as well as very consistent phenotypes in the chemical sensitivity assays (Fig.3). The phenotypic profiling results show that growth of both isolates TP1 and TP3 could be inhibited by colistin or minocycline at higher concentrations (Fig.3, yellow box and light blue box, respectively); tigecycline sensitivity is not included in the phenotype microarray (PM) panel. Isolate TP3 was found to be completely resistant to nafcillin in this assay, whereas TP1 was sensitive (Fig.3, purple box).

Each row represents a bacterial isolate (TP1 or TP3) in one phenotype panel (PM01PM20), and each column represents a specific condition (wells A01-H12) within each panel, as shown in Supplementary Data1. Effects on bacterial metabolic activity are indicated by color, with red representing growth inhibition, black representing intermediate growth and green representing growth promotion. Yellow box: colistin. Light blue box: minocycline. Purple box: nafcillin. Results are calculated using the area under the curve for 48h of growth and are presented as the average of three replicates per strain.

Three A. baumannii isolates, TP1, TP2, and TP3, were sequenced to closure using a combination of short-read (Illumina) and long-read (Nanopore) sequencing to investigate pathogen evolution during the course of phage treatment. Sequencing to closure allows for tracking of the number and position of mobile DNA elements that are often not assembled into larger contigs if the genomes are only sequenced with short-read sequencing. Strain TP1 was isolated prior to the start of phage treatment and was the clinical isolate used to determine phage sensitivity and conduct environmental phage hunts for assembly of therapeutic phage cocktails11. Strains TP2 and TP3 were isolated 6 days and 8 days after the start of phage treatment. All three strains were found to contain a single 3.9Mb chromosome and a single 8.7kb plasmid (Table3). Some variation was observed in bacterial chromosome length between strains but the plasmids contained in each strain were identical, and it is clear that these three isolates represent the evolution of strains from a common ancestor over time rather than a succession invasion by different strains. Analysis of the genomes in pubMLST43 identified all three isolates as sequence type 570 (Pasteur) and analysis in Kaptive44 identified a 20.5kb region (base position 3,774,0313,794,556 in the TP1 genome) containing 17 genes encoding a predicted capsule type of K116 (Supplementary Table3). Consistent with the broad antibiotic resistance observed in these isolates, 32 (TP1) and 35 (TP2, TP3) antibiotic resistance genes (ARGs) were identified based on searches against the CARD 2021 database45 (Supplementary Data2). The 8.7kb plasmid contained in TP1, TP2 and TP3 does not encode any identifiable AMR genes, and is identical to plasmids carried in many other A. baumannii strains deposited in NCBI. Few SNPs and indels were observed between these isolates, including 23 large (>1kb) insertions or deletions associated with the movement of mobile DNA elements. Summaries of the genomes and changes observed in strains TP2 and TP3 (relative to TP1) are summarized in Table3, and the locations of AMR genes, transposases, prophages, capsule locus in TP1 genome, and large insertion and deletions (>1kb) in TP2 and TP3, in reference to TP1, are illustrated in Fig.4. In reference to TP1, detailed sequence changes, associated coordinates and genes affected in TP2 and TP3 are listed in Supplementary Tables4 and 5, respectively.

Each gray band represents a bacterial chromosome as labeled at the replication origin. The scale on the outer ring represents DNA coordinates in Mb. The locations of AMR genes, transposases, prophages and the capsule locus are indicated for TP1 only. Major insertions or deletions (>1kb) are indicated in the TP2 and TP3 chromosomes. Of note, TP2 and TP3 contain a novel 6.7kb insertion element at the ~0.11Mb position that is not present in TP1, indicating horizontal acquisition during infection.

The most notable change in TP2 and TP3 is the acquisition of a novel 6673bp insertion sequence, inserted in a position adjacent to an existing IS3-like transposase at position 111,357 of the TP1 genome (Fig.4 and Supplementary Tables4 and 5). This acquired 6.7kb element is not native to TP1 and represents an acquisition of new DNA by horizontal gene transfer that occurred during the course of infection, and is most likely the result of DNA acquisition mechanisms unrelated to phage treatment. A. baumannii is known for its ability to rapidly acquire mobile DNA elements in the environment via conjugation and natural competence46,47, and to vary surface molecules through horizontal gene transfer48. T4-like phages like those used in treatment are generally poor transducers. In phage T4, multiple defects in ndd, denB, 42 and alc are required for transduction to occur49, and these genes are all conserved in the cocktail myophages reported in this study. In addition, transduction requires the phage to be able to productively infect the donor of the acquired DNA, which was likely to have been a different bacterial species and thus insensitive to the phages used. BLASTn searches of this sequence identified identical or nearly identical sequences in other Gram-negative bacterial genomes or plasmids, including A. baumannii (CP038644), Klebsiella pneumoniae (LR697132), E. coli (CP020524), and Citrobacter freundii (KP770032). This inserted sequence encodes a number of significant additional antibiotic resistance determinants, including a predicted aminoglycoside O-phosphotransferase (IPR002575), an NDM-1-like metallo-beta-lactamase (CD16300, IPR001279), and a CutA-like protein that may be involved in metal tolerance (IPR004323). The inserted aminoglycoside O-phosphotransferase (CARD ARO:3003687) is relatively rare in A. baumannii, found in 1.43% of A. baumannii chromosomes and 0.47% of A. baumannii plasmids, as reported by the CARD Resistance Gene Identifier. The prevalence of the inserted NDM-1-like metallo-beta-lactamase (CARD ARO:3000589) is 5.94% of A. baumannii genomes and 0.6% of A. baumannii plasmids.

Other than the 6.7kb insertion described above, all other major variations in the TP2 and TP3 genomes can be attributed to deletion or transposition of elements present in the TP1 genome (Supplementary Tables4 and 5). Another 1886bp insertion sequence was identified in TP2 and TP3 which introduces a second copy of the IS6-like transposase and an additional copy of an aminoglycoside O-phosphotransferase (IPR002575) which is also present in TP1 (locus HWQ22_16890). In this case, this insertion is a duplication of an existing AMR gene rather than the acquisition of foreign DNA. The presence of the new 6.7kb element and the duplicated 1.9kb element resulted in three extra ARGs in TP2 and TP3 (35 total predicted AMR genes) compared to TP1 (32 total predicted AMR genes) (Table3 and Supplementary Data2). This highlights the fact that bacterial pathogens do not exist as strictly clonal populations even in a single patient over time.

Prophage analysis revealed two apparently complete prophage regions (52,563bp and 42,762bp in length, respectively) in TP1, TP2, and TP3 genomes that are likely to encode active prophages (Fig.4). Phage att sites and conserved phage proteins (tail and tail tape measure protein, major head subunit and head morphogenesis protein, terminase large subunit, endolysin) were identified in both prophage regions; the coordinates of the prophages in the TP1 genome are provided in Supplementary Table3. These two prophage regions are conserved in TP1, TP2, and TP3 and no sequence change was observed among the three strains. The 52kb prophage 1 is highly conserved (with up to 100% nucleotide identity by BLASTn) in many other A. baumannii genomes, including that of ATCC 19606, which is one of the earliest available clinical isolates of A. baumannii dating to the 1940s50. This prophage region shares limited similarity to cultured phages, with its closest relative being Acinetobacter phage Ab105-3phi (KT588073), with which it shares 49.4% nucleotide identity and 22 similar proteins. The 43kb prophage region was found to be less conserved in other A. baumannii genomes, with the most closely related prophage element sharing only 69% overall sequence identity. This prophage region is ~46% related to A. baumannii phage 5W (MT349887), which also appears to be a temperate phage due to the presence of an integrase and LexA-like repressor. Other than 5W, this element is not closely related to any other cultured phages in the NCBI database, sharing no more than 10% nucleotide identity and no more than 8 proteins with other phages. Protein-based clustering of these elements (Fig.1E) showed that they are only distantly related to other cultured phages, with the closest neighbors in the Guernseyvirinae. A recent analysis of prophage carriage in A. baumannii genomes suggests that intact prophages are relatively uncommon in this species (less than one per genome) and also highly diverse, indicating a large amount of unexplored diversity in temperate phage elements51.

Five phages selected from the phage cocktails (AC4, Maestro, AB-Navy1, AB-Navy97, AbTP3phi1) were used to select for phage-insensitive mutants in vitro using A. baumannii strain TP1 as host. Three independent mutants against phages AC4, Maestro, AB-Navy97, AbTP3phi1 were isolated, and two independent mutants against phage AB-Navy1 were isolated. After resequencing and mapping mutant reads to the reference TP1 genome, changes detected with quality scores greater than 100 were examined (Table4). The majority of identified mutations were located in the bacterial KL116 capsule locus. The K116 capsule is comprised of a five-sugar repeating unit with a three-sugar backbone composed of Gal and GalNAc and a two-sugar side chain composed of Glc and GalNAc52. In all the mutants resistant to the myophages Maestro, AC4, AB-Navy97, a common 6-bp deletion was observed in a predicted capsular glycosyltransferase protein identified as Gtr76 by Kaptive (HWQ22_04225) (Fig.5). Notably, these 6-bp deletions were also observed in isolates TP2 and TP3, which evolved in vivo during phage treatment and were insensitive or showed reduced sensitivity to all myophages tested (Table2 and Supplementary Tables4 and 5). These 6-bp deletions occurred in a region containing four copies of a tandem repeat sequence TAAATT (Fig.5B), which probably is prone to mutation by strand slippage events during replication. These mutations result in the deletion of residue L243 and N244, resulting in the reduction of a predicted flexible linker between two -helices in the C-terminus of the glycosyltransferase protein. This protein is predicted to participate in capsule synthesis by forming the -D-GalNAc-(14)-D-Gal linkage of the side chain disaccharide to the trisaccharide backbone52, suggesting that this side chain plays a role in host recognition by these phages.

A Diagram of the KL116 capsule locus identified in strains TP1, TP2, and TP3 as predicted by Kaptive. Genes are represented by arrows oriented in the direction of transcription. Orange arrows represent genes involved in capsule export, yellow genes are involved in repeat unit processing, blue genes are involved in simple sugar biosynthesis, green genes encode glycotransferases and the red gene codes for the initiating transferase. All genes had 100% coverage and ranged from 90100% identity to the KL116 type in the Kaptive database. Defective capsule locus genes identified inin vitro-generated phage-insensitive mutants of TP1 are indicated by black arrows; numbers in parentheses after each phage name indicate what proportion of phage-insensitive mutants contained a mutation in that gene. B Nucleotide alignment of the sequences showing the six nucleotide deletion in one of the glycosyltransferases (gtr76) found in multiple TP1 mutants resistant to the cocktail myophages.

In mutants selected for insensitivity to phage AB-Navy1, one mutant contained the same conserved 6bp deletion identified in the other mutants, and one lacked this mutation but instead had a nonsense mutation (W183am) in carO (HWQ22_09280) (Table4). CarO is a 29kDa outer membrane transporter, loss of which has been associated with increased antibiotic resistance53,54. The role of CarO in phage sensitivity is not clear, but its truncation may lead to other cell wall defects that reduce sensitivity to this phage; truncations in CarO have been associated with reduced adherence and invasion in tissue culture and with reduced virulence in vivo55. This finding illustrates that defects in the capsule locus are not the only means by which TP1 may gain phage insensitivity. Notably, similar CarO defects were not observed in TP2 or TP3, which attained phage resistance in vivo.

In addition to the common 6bp deletion in the Gtr76 glycosyltransferase and CarO mutation, the other mutations observed in the myophage-insensitive mutants are SNPs or small indels in non-coding regions or that result in missense or silent mutations in a predicted capsular glucose-6-phosphate isomerase Gpi (HWQ22_04190) and an ABC transporter, respectively (Table4). However, these SNPs are not conserved in the in vitro mutants against myophages and were also not detected in in vivo isolates TP2 and TP3, suggesting that the defect observed in the Gtr76 glycosyltransferase is sufficient to confer insensitivity to the cocktail myophages in this strain.

Strain TP1 mutants resistant to the podophage AbTP3phi1 were also found to contain mutations in the capsule locus, but these mutations were confined to the genes encoding the glucose-6-phosphate isomerase Gpi and polysaccharide biosynthesis tyrosine autokinase Wzc (HWQ22_04255) (Table4 and Fig.5A). Loss of function in these genes is expected to result in loss of L-fructose-6-phosphate required for downstream production of capsule monomers and defects in capsule export, respectively32. This suggests that these mutants may exhibit more severe defects in K116 capsule expression, and that AbTP3phi1 requires the presence of the capsule backbone for successful infection.

Our results are consistent with the recently published work by Altamirano et al.34, where a frameshift in the glycosyltransferase and glucose-6-phosphate isomerase within the K locus were detected in two independent phage-resistant A. baumannii mutants. The consistency between our work and that study confirms the A. baumannii capsule locus being important for phage sensitivity. Both Gpi and glycosyltransferases are involved the biosynthesis of capsule K units, which are tightly packed repeating subunits consisting of 4 to 6 sugars56. The reason why one group of phages (our myophages, and the myophage FG02 in ref. 34) selected primarily for defects in the Gtr glycosyltransferase but the other phages (our podophage AbTP3phi1 and myophage CO01 in Altamirano et al.) selected for defects in Gpi is not entirely clear. These phages likely recognize different moieties of the bacterial capsule as their receptors, but it should be noted that many of the mutations associated with insensitivity observed in our study are not necessarily inactivating to the protein: the most common mutation in the capsule locus is a two-residue in-frame deletion in gtr76 (Fig.5B), and the other mutations are single-residue changes or nonsense/frameshift mutations relatively late in the reading frame. These mutations may modulate protein function rather than being inactivating.

Capsule is a known common requirement for A. baumannii phages, and defects in capsule synthesis have been shown to be responsible for phage resistance34,57. The presence of the same 6bp deletion in the capsular glycosyltransferase gene gtr76 of both the in vitro- and the in vivo-selected A. baumannii strains indicates that the same route to phage insensitivity may be followed by strain TP1 in both systems. Importantly, this demonstrates that laboratory in vitro investigations of bacterial selection and phage insensitivity can produce results that are relevant and predictive for the in vivo milieu of clinical treatment.

See the original post here:
Comparative genomics of Acinetobacter baumannii and therapeutic bacteriophages from a patient undergoing phage therapy - Nature.com