Region-specific phage compositions

Abstract

Use of phylogeographical analysis and molecular typing of strains of interest (SOIs) of isolates of an antibiotic-resistant nosocomial bacterial species in a method of preparing a region-specific phage composition with relevance for a predetermined geographical region; and a method of preparing a region-specific phage composition effectively targeting an antibiotic-resistant nosocomial bacterial species and, comprising: a) performing molecular typing of isolates of the bacterial species to identify a type of interest (TOI) that is prevalent in a predetermined geographical region, wherein said molecular typing comprises DNA sequencing and clustering the genomes into genotypes; b) performing phylogeographical analysis of strains of said TOI to identify target strains, which are phylogenetically most divergent strains of said TOI in said predetermined geographical region; c) selecting a repertoire of bacteriophages from a bacteriophage source, which repertoire exhibits lytic activity against said target strains; and d) preparing a pharmaceutical phage composition comprising said repertoire.

Description

[0001] REGION-SPECIFIC PHAGE COMPOSITIONS

[0002] FIELD OF THE INVENTION

[0003] The invention refers to a method of preparing a region-specific phage composition effectively targeting an antibiotic-resistant nosocomial bacterial species, and use of phylogeographical analysis and molecular typing of strains of interest (SOIs) of isolates of an antibiotic-resistant nosocomial bacterial species in a method of preparing a regionspecific phage composition.

[0004] BACKGROUND OF THE INVENTION

[0005] The uncontrollable rise of antibiotic resistance is one of the leading causes of human mortality. Of particular concern is the widespread presence of antibiotic-resistant bacteria in healthcare settings. Due to the limited number of effective antibiotic options, these infections often produce severe clinical outcomes. Moreover, as the conventional antibiotic development pipeline is drained, this situation will most likely not improve in the near future. Therefore, it is imperative to find alternative therapeutic approaches, with bacteriophage therapy emerging as a prominent candidate.

[0006] Bacteriophages, also known as phages, are viruses that can specifically attack and kill their host bacteria. While there have been numerous compassionate use cases where phages successfully tackled infections that resisted all conventional antibiotic treatments, the translation of this potential into positive patient outcomes on a broader scale has been progressing slowly. This limitation stems from the narrow host range of bacteriophages, which complicates their broad therapeutic applications. Specifically, many resistant bacterial species consist of hundreds of strain types, each characterised by distinct cell surface structures and defence mechanisms that restrict the host range of bacteriophages to a subset of bacterial strain types.

[0007] In response to this challenge, two primary approaches have emerged (Verbeken et al. 2022). The first one combines multiple phages into a fixed composition for broader bacterial coverage, followed by the initiation of clinical trials aimed at obtaining marketing authorisation. While several such attempts are currently in progress, as of now, no registered medicinal phage products have reached the market.

[0008] The second strategy is personalised treatment, where the patient’s bacterial sample is screened to identify the most effective phages from collections of bacteriophages (Verbeken et al. 2022; Hatfull et al. 2022). While this approach is considered promising, it comes with its own set of challenges, including limited throughput due to time-consuming phage production and regulatory approval hurdles that often follow the diagnosis of the infection. Overall, irrespective of the chosen methodology, further development of the field will largely depend on systematic procedures that identify patients requiring the same bacteriophages on a large scale (Mutalik et al. 2022).

[0009] In recent years, genomic surveillance has matured into an efficient tool to track pathogens at an unprecedented scale. Advances in sequencing technologies have revealed the genomes of hundreds of thousands of bacterial isolates. This wealth of data now presents a unique opportunity to gain insights into the distribution and transmission patterns of antibiotic-resistant bacterial pathogens. In particular, genomic surveillance has established that infections caused by top-priority antibiotic-resistant pathogens on the World Health Organisation’s (WHO) list, commonly originate within healthcare facilities. The nosocomial spread of these pathogens raises the prospect of forecasting the potential causative agents of forthcoming infections and pre-emptively matching them with suitable phages. This holds immense potential for facilitating study recruitment to achieve clinical validation and cost-effectiveness for phage therapy. Moreover, by allowing the pre-emptive preparation with suitable phages, the time for administering an effective phage treatment post-diagnosis could be significantly reduced. Despite this immense potential, genomic surveillance remains largely untapped in phage therapy.

[0010] Phage therapy is gaining increasing interest in the fight against critically resistant nosocomial pathogens. However, the narrow host range of bacteriophages hampers the development of broadly effective phage therapeutics and demands precision approaches.

[0011] Carbapenem-resistant A. baumannii (CRAB) is a bacterial species that is recognized by WHO as a top-priority nosocomial pathogen, for which innovative therapies are urgently needed, with an alarming mortality rate of 24,7% in the USA and more than 100,000 deaths globally in 2019. The phage host range of this bacterial species is especially narrow. In fact, a recent estimate suggests that it may require a phage library of more than 300 phages to effectively cover the majority of the clinical isolates (Strathdee et al. .2023).

[0012] Regelmumbal et al. (2016) disclose a personalized therapeutic five-member cocktail of wild environmental phages to rescue mice from A. baumannii wound infections. Four phages did not kill the parent strain, and one delays bacterial growth. The one constituent phage targets capsulated A. baumannii bacteria and selects for loss of receptor, shifting the population to an uncapsulated state that is then sensitized to the remaining four phages in the cocktail.

[0013] WO2021 / 138218A1 discloses a method of selecting a phage formulation using a spatio-temporal infection database.

[0014] WO2021 / 222257A1 discloses a bacteriophage composition capable of infecting and killing Pseudomonas. Bacteriophages were isolated from their original environment and tested against a panel of P. aeruginosa.

[0015] Bourdin et al. (Microbial Biotechnology 2014, 7(2): 165-176) disclose coverage of diarrhoea-associated Escherichia coli isolates from different origins with two types of phage cocktails. T4-like phages from a phage collection were tested against four collections of childhood diarrhoea-associated Escherichia coli isolates representing different geographical origins (Mexico versus Bangladesh), serotypes (69 O, 27 H serotypes), pathotypes (ETEC, EPEC, EIEC, EAEC, VTEC, Shigella), epidemiological settings (community and hospitalized diarrhoea) and years of isolation.

[0016] Sarker et al. (Virology 2012, 434(2): 222-232) discloses oral T4-like phage cocktail application to healthy adult volunteers from Bangladesh.

[0017] Warenth et al. (Pathogens 2021 , 10:690) disclose that molecular characterization of German Acinetobacter baumannii isolates and multilocus sequence typing (MLST) analysis based on whole genome sequencing reveals novel sequence types.

[0018] Hamidian et al. (Microbial Genomics 2019, 5(10): 1-40) disclose emergence, molecular mechanisms and global spread of carbapenem-resistant Acinetobacter baumannii.

[0019] Koncz et al. (bioRxiv preprint June 15, 2024, DOI: 10.1101 / 2024.06.15.599013) discloses pathogen genomic surveillance as a scalable framework for precision phage therapy.

[0020] There is a need for new strategies against antibiotic-resistant nosocomial bacteria using phages.

[0021] SUMMARY OF THE INVENTION

[0022] It is the objective of the invention to provide phages which can be used against antibiotic-resistant nosocomial bacteria, and to prepare suitable phage compositions for therapy.

[0023] The objective is solved by the subject matter as claimed and as further described herein. The present invention is based on the finding that genomic surveillance can lay the groundwork for a scalable precision phage therapy. Large-scale phylogeographic analyses in combination with experimental phage discovery and high-throughput phage typing surprisingly demonstrated that 90% of the infections can be attributed to a limited number of strain types in each world region, which can be effectively addressed by a small set of bacteriophages. Furthermore, it was found that causative CRAB strains are advantageously predictable within a six-year timeframe for individual countries, and countries with identical CRAB strain types can be rapidly identified. These combined capabilities enable proactive measures, including the development of region-specific phage collections and the formulation of phage cocktails designed to target specific strain types in a highly precise manner.

[0024] A method is provided herein which defines the most relevant target (nosocomial e.g., CRAB or CRKP) bacterial strain set of a given geographic region to optimise pharmaceutical phage products consisting of a relatively small number of phages that effectively target the highest number of relevant target strains in that region. The method can maximise the number of patients who can be treated with a phage composition (cocktail) while minimising the number of distinct phage components in the product. The pharmaceutical phage product can be optimised for a specific geographical region rather than being personalised or globally applicable.

[0025] The invention provides for a method of preparing a region-specific phage composition effectively targeting an antibiotic-resistant nosocomial bacterial species and, comprising: a) performing molecular typing of isolates of the bacterial species to identify a type of interest (TOI) that is prevalent in a predetermined geographical region, wherein said molecular typing comprises DNA sequencing and clustering the genomes into genotypes; b) performing phylogeographical analysis of strains of said TOI to identify target strains, which are phylogenetically most divergent strains of said TOI in said predetermined geographical region; c) selecting a repertoire of bacteriophages from a bacteriophage source, which repertoire exhibits lytic activity against said target strains; and d) preparing a pharmaceutical phage composition comprising said repertoire. Specifically, one, two or more TOIs are identified, and a pharmaceutical phage composition is prepared which comprises one, two or more repertoires to target all target strains of said TOIs.

[0026] Specifically, for each of said one, two, or more TOIs, a repertoire of phages is selected and used to prepare a pharmaceutical phage composition.

[0027] According to a specific aspect, a method is provided which comprises: a) performing molecular typing of isolates of the bacterial species to identify one, two or more TOIs that are prevalent in said predetermined geographical region; b) performing phylogeographical analysis of strains of each of said TOIs to identify target strains for each of said TOIs, which are phylogenetically most divergent in said predetermined geographical region; c) selecting one, two or more repertoires of bacteriophages from a bacteriophage source, wherein said repertoires exhibit lytic activity against target strains; and d) preparing a pharmaceutical phage composition comprising said one, two or more repertoires.

[0028] According to a specific aspect, molecular typing comprises DNA sequencing (such as sequencing of the entire bacterial genome, or selected nucleic acid regions or genes) and clustering the genomes into genotypes.

[0029] Typically, said molecular typing comprises genome typing which clusters pathogen genomes based on the presence or absence of specific genomic features (e.g., genes, mutations), preferably wherein these features are associated with phenotypic variability that impacts barriers to phage infection. For example, cell surface typing reflects variability in cell surface composition.

[0030] Specifically, said molecular typing comprises any one or more or all of cell surface typing, antibiotic-resistance typing, whole genome sequence typing, or multilocus variable-number tandem-repeat analysis (MLVA) typing.

[0031] Specifically, said molecular typing comprises a) evolutionary distance typing, such as multilocus sequence typing (MLST) or whole genome multilocus sequence typing (cgMLST); and / or b) surface structure typing, such as surface capsular polysaccharide typing or O antigen typing.

[0032] Preferably, said molecular typing comprises a) evolutionary distance typing, such as multilocus sequence typing (MLST) or whole genome multilocus sequence typing (cgMLST); and b) surface structure typing, such as surface capsular polysaccharide typing or O antigen typing.

[0033] Preferably, said molecular comprises combined multilocus sequence typing (MLST) and cell surface capsular polysaccharide (CPS) genotyping.

[0034] Specifically, a) said MLST clusters evolutionary-related genomes into sequence types based on the sequence of at least 6 different housekeeping genes; and b) said CPS comprises identifying polysaccharide capsule and outer lipopolysaccharide loci and classifies assemblies into CPS-types.

[0035] Specifically, molecular typing is by a high-throughput method for molecular typing of the bacteria.

[0036] Specifically, said MLST-CPS genotyping comprises a high-throughput method for molecular typing of the bacteria.

[0037] Specifically, molecular typing comprises a combination of different typing methods, such as e.g., a combination of genetic typing methods such as to capture complementary aspects of phage susceptibility.

[0038] Specifically, molecular typing comprises grouping bacterial isolates into antibioticresistant (e.g., CRAB or CRKP) types based on antibiotic resistance gene content.

[0039] Specifically, MLST is combined with CPS typing. MLST clusters evolutionary- related genomes into sequence types (TOI) based on the sequence of at least 6 or 7 different housekeeping genes. CPS typing identifies specific genomic determinants responsible for synthesizing structurally different CPSs. CPS can function as receptors for phages. Specifically, CPS typing comprises identifying polysaccharide capsule and outer lipopolysaccharide loci and classifies assemblies into CPS-types.

[0040] Specifically, the distribution of circulating antibiotic-resistant bacterial types in a predetermined geographical region can be determined by genotyping such as e.g., the combined MLST-CPS typing. Preferably, said genotyping clusters the majority of antibiotic-resistant bacterial diversity e.g., more than 50%, 60%, 70%, 80% or 90% into a limited set of genome clusters.

[0041] Specifically, said MLST clusters evolutionary-related genomes into sequence types (ST) based on the sequence of housekeeping genes.

[0042] Specifically, MLST clusters evolutionary-related genomes into sequence types (TOI) based on the sequence of at least 6 or 7 different specific housekeeping genes. For each bacterial species there are specific housekeeping genes well-known in the art. According to a specific example, for A. baumannii the following housekeeping genes can be used: gltA, gyrB, gdhB, recA, cpn60, gpi or rpoD.

[0043] According to another specific example, for Klebsielle pneumoniae, the following housekeeping genes can be used: gapA; infB; mdh; pgi; phoE; rpoB; tonB.

[0044] Further housekeeping genes can be used as listed in the public databases for molecular typing and microbial genome diversity (University of Oxford).

[0045] Specifically, prevalence of a certain bacterial serotype (e.g., a MLST-CPS serotype) is determined if the serotype accounts for at least 5% of the isolates in a predetermined geographical region. Specifically, a serotype can be classified as global if (i) it is prevalent as described herein, and (ii) it accounts for at least 2% of the isolates in at least 3 world regions spanning at least 2 continents, preferably wherein the world regions are defined according to the World Bank Development Indicators database (The World Bank, Washington, U.S.A.).

[0046] Specifically, the predetermined geographical region is any one or more of Western Europe, Eastern Europe, Southern Europe, North-America, South-America; West-Asia, East-Asia, or South-East-Asia.

[0047] According to a specific aspect, the phylogeographical analysis comprises: a) determining a number of single nucleotide polymorphisms (SNPs) in the core genome of the target strains; and b) phylogenetic analysis using a time-scaled phylogenetic tree to visualize the chronological genomic development in the core genome; wherein genetic distance is determined by the phylogenetic differences among the target strains.

[0048] Specifically, the core genome is a fraction of the whole genome shared by all of said isolates which are of the identified TOI, which core genome may differ in a number of SNPs.

[0049] Specifically, the number of SNPs is compared to a reference genome.

[0050] Specifically, the reference genome is the genome of the wild-type species. The reference genome can be selected from the genome set of the TOI based on the criterion of maximal genome length among available assemblies.

[0051] According to a specific aspect, phylogeographical analysis of strains is made for determining historical processes that may be responsible for the past to present and potentially future geographic distributions of genealogical lineages, by considering the geographic distribution of individuals in light of genetics, particularly population genetics. Specifically, phylogeographical analysis can be made to extrapolate the geographic distribution of nosocomial antibiotic resistant bacterial species for the future such as within for a certain time period up to e.g., 10, 9, 8, 7, 6, 5, 4, 3, 2, or 1 year.

[0052] Specifically, phylogeographical analysis comprises the determination of spatiotemporal dynamics of antibiotic-resistant (e.g., CRAB or CRKP) types of bacterial species or strains.

[0053] According to a specific aspect, the phylogeographical analysis is by phylogenetic analysis of a bacterial sample, wherein the sample is obtained from said predetermined geographical region.

[0054] A time-scaled phylogenetic tree can be produced using selected genomic sequences. A phylogenetic tree for selected bacterial strains typically visualizes the chronological genomic development of one or more certain genes (such as antibioticresistance genes) within a certain time-scale e.g., identifying phylogenetically diversity of said bacterial strains dominant genes. The genetic distance is typically determined by the number of mutation / evolutionary events between different bacterial species since their divergence. Phylogenetic diversity is typically determined by the sum of the branch lengths of a phylogenetic tree connecting all species, taking into account phylogenetic differences among the species.

[0055] The most distant phylogenetically diverse bacteria of a TOI are typically those which comprise genetic distance in their core genomes. The core genome refers to the DNA sequence that is present in all individuals of a particular species or strain within a population, which may or may not differ by one or more genomic mutations. Typically, the core genome is a fraction of the whole genome shared by all isolates of a TOI and used to determine the number of SNPs. The number of SNPs within the core genome can be used as a measure of genetic distance.

[0056] Specifically, bacterial core genomes are used for phylogenomic analysis.

[0057] Specifically, phylogeographical analysis is of at least one dominant sequence type (ST or TOI) clade.

[0058] Phylogeographical analysis can be carried out using established methods.

[0059] Specifically, phylogeographical analysis comprises determining phylogenetically diversity of one or more antibiotic resistance genes in particular determining for each of said one or more antibiotic resistance genes the genomic differences over time.

[0060] According to a specific aspect, the nosocomial bacterial species is one of “ESKAPE” pathogens which include e.g., Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Escherichia coli and Enterobacter cloaceae.

[0061] Specifically, the bacterial species is selected from the group consisting of: Acinetobacter baumannii, Klebsiella pneumoniae, Staphylococcus aureus, Escherichia coli, Pseudomonas aeruginosa, Enterococcus faecium, Enterococcus faecalis, Mycobacterium tuberculosis, Neisseria gonorrhoeae, Salmonella enterica, Clostridioides difficile, Helicobacter pylori, Streptococcus pneumoniae, Campylobacter jejuni, Haemophilus influenzae, Shigella spp., Enterobacter spp., and Proteus mirabilis. These bacterial species present significant challenges in clinical settings due to their resistance to antibiotics and their impact on human health.

[0062] Antibiotic-resistance genes relevant for a bacterial species are either well-known in the art, or can be identified based on sequence similarity to literature-curated resistance genes.

[0063] Typically, antibiotic resistance genes are understood as those genomic sequences which are relevant for susceptibility or resistance of a bacteria to certain antibiotics. For example, beta-lactam antibiotic-resistance genes of Acinetobacter baumannii are considered relevant if corresponding to phenotypic resistance of the bacteria to e.g., carbapenem or non-carbapenem beta-lactams.

[0064] A specific A. baumannii bacterial species of interest is a CRAB, which is particularly understood to comprise in its genome resistance determinants against carbapenems as well as against aminoglycosides and fluoroquinolones.

[0065] For example, one or more of the following antibiotic-resistance genes of CRAB bacteria can be used for phylogenetic analysis and / or phylogeographical analysis: i) carbapenem resistance genes: blaOXA-23-like, blaOXA-24 / 40-like, blaOXA- 58-like, blaOXA-143-like, blaOXA-235-like, any blaNDM, blaVIM, blaIMP, blaKPC, blaGES. ii) non-carbapenem beta-lactam resistance genes: carO, ISAba, though such genes were occasionally associated with carbapenem resistance.

[0066] According to another example, one or more of the following antibiotic-resistance genes of Klebsiella pneumoniae can be used for phylogenetic analysis and / or phylogeographical analysis: blaOXA-48-like, blaNDM, blaVIM, blaIMP, blaKPC, blaGES, blaSHV, blaTEM, blaCTX-M.

[0067] Specifically, the bacterial species (such as e.g., Acinetobacter baumannii or Klebsiella pneumoniae) is resistant against an antibiotic selected from the group consisting of: beta-lactams (such as carbapenem or non-carbapenem beta-lactams), aminoglycosides, fluoroquinolones sulfonamides, tetracyclines, macrolides, glycopeptides, and polymyxins, or any other new antibiotic class that is applied in the clinic.

[0068] Resistance can be against one antibiotic or antibiotic class, or can be multiple resistance, such as e.g., against antibiotics of two or more different antibiotic classes.

[0069] Specifically, the isolates of bacterial species are carbapenem-resistant Acinetobacter baumannii (CRAB) strains or carbapenem-resistant Klebsiella pneumoniae (CRKP) strains.

[0070] Specifically, the CRAB is resistant to at least one carbapenem beta-lactam antibiotic, and optionally at least one aminoglycoside antibiotic, and / or at least one fluoroquinolone antibiotic. For example, a CRAB strain can be resistant to at least one carbapenem beta-lactam antibiotic, at least one aminoglycoside antibiotic, and at least one fluoroquinolone antibiotic.

[0071] Specifically, the pharmaceutical phage composition comprises one or more repertoires of bacteriophages which target (in particular which have lytic activity against a) at least any one of 60, 70, 80, or 90% of isolates belonging to one or more TOIs; and / or b) at least any one of 60, 70, 80, or 90% of representative isolates prevalent in said predetermined geographical region.

[0072] As used herein “target” or “targeting” is understood as a target structure or organism (e.g., a bacterium) that is recognized by as bacteriophage. Typically, a bacterium, in particular a bacterial species, strain or isolate, can be targeted by a bacteriophage such that the bacteriophage lyses the bacterium e.g., as determined in a lysis test. A bacterium, in particular a bacterial species, strain or isolate, is particularly understood as a target bacterium, if it is lysed by a respective bacteriophage.

[0073] According to a specific aspect, the pharmaceutical phage composition comprises a repertoire of bacteriophages which lyses isolates of said TOI in a lysis test, wherein at least any one of 60, 70, 80, or 90% of the isolates are prevalent in said predetermined geographical region.

[0074] According to a specific aspect, said one or more repertoires of bacteriophages targets one or more bacterial species such as e.g. one or more nosocomial bacterial species or strains. Specifically, said one or more repertoires of bacteriophages targets species of one or more of ESKAPE pathogens, such as of the genera Acinetobacter, Klebsiella, Enterococcus, Staphylococcus, Pseudomonas, Escherichia, and Enterobacter, e.g., Acinetobacter baumannii, Klebsiella pneumoniae, Enterococcus faecium, Staphylococcus aureus, Pseudomonas aeruginosa, Escherichia coll and Enterobacter cloaceae.

[0075] Specifically, the bacteriophage source is originating from an environmental sample of the predetermined geographical region, preferably wherein the environmental sample is of waste water, soil, sewage, aquatic habitats, agricultural runoff, hospital environments, food processing facilities, animal feces, or a plant source.

[0076] Specifically, the bacteriophage source is a phage library, preferably wherein the phage library comprises at least 10, 15, 20, 30, 40, 50, 100, or 200 different bacteriophages.

[0077] According to a specific example, the bacteriophage source is a combination of sources, such as a combination of environmental samples and a phage library, which combination is used to isolate and / or select suitably phages.

[0078] Specifically, a repertoire of bacteriophages comprises or consists of 3-15 different bacteriophages.

[0079] Specifically, any one or more or all of the selected repertoires of bacteriophages comprises or consists of 3-15 different bacteriophages, preferably at least 3, 4, 5, 6, 7, 8, 9, or at least 10, preferably less than 15, 14, 13, 12, or less than 11 .

[0080] A repertoire of bacteriophages can be tested for lytic activity against selected bacterial strain. The test for lytic activity can e.g., be carried out by testing the lytic activity of individual phages against the bacterial strains, or by pool testing such as by testing the lytic activity of the repertoire in a mixture.

[0081] Lytic activity can be tested before or after or during selection of the repertoire from the bacteriophage source, such as by screening methods.

[0082] Suitable assays for determining the lytic activity are e.g., plaque assay or bacterial growth suppression measurement. Any of the assays described in the present examples can be used for determining the lytic activity. For example, plaque assays can be used to screen for the presence of lytic phage activity. An alternative method to quantify lytic activity is to measure the ability of the phage to suppress bacterial growth over time compared to the growth curve of a bacterial culture without phages. Improved antibacterial activity results in significantly slower growth or complete eradication of the investigated strain. The growth of the bacteria can e.g., be quantified by measuring the OD600 values of the bacterial culture at every 20 min for 24 or 72 hours.

[0083] Specifically, the lytic activity of an individual phage or of or repertoire of phages can be tested against multiple bacteria e.g., in parallel, such as by a spot assay.

[0084] Specifically, the lytic activity is at least 106pfu, 107pfu, 108, 109, or 101°, up to e.g., 1011pfu per mL against the target strains such as e.g., determined in a plaque assay.

[0085] Specifically, for characterizing and / or selecting a repertoire of bacteriophages or of individual bacteriophages of the repertoire, the lytic activity is at least 106pfu, 107pfu, 108, or 109, up to e.g., 1011pfu per mL against the target strains such as e.g., determined in a plaque assay.

[0086] Specifically, the lytic activity of a repertoire of bacteriophages or of individual bacteriophages of any one or more of said repertoires is at least 106pfu, 107 pfu, 107pfu, 108, or 109, up to e.g., 1011pfu per mL against the target strains such as e.g., in a plaque assay.

[0087] Specifically, the predetermined geographical region is Eastern and Southern Europe, and the repertoire of bacteriophages exhibits lytic activity against CRAB strains of the ST2 lineage, preferably the ST2-KL3 strain type, in particular the ST2-KL3 MLST- CPS type.

[0088] Specifically, the predetermined geographical region is Eastern and Southern Europe, and the repertoire of bacteriophages exhibits lytic activity against CRKP strains of the ST101-KL17 MLST-CPS type.

[0089] Specifically, the pharmaceutical phage composition comprises a repertoire of bacteriophages in a pharmaceutical formulation, or is provided as a kit of parts comprising one or more bacteriophages of said repertoire in separate containments.

[0090] Specifically, the pharmaceutical phage composition comprises one or more repertoires of bacteriophages in a pharmaceutical formulation, or is provided as a kit of parts comprising one or more bacteriophages of said one or more repertoires in separate containments.

[0091] Specifically, a pharmaceutical phage composition is produced which comprises one or more repertoires of bacteriophages in a pharmaceutical formulation.

[0092] Specifically, the pharmaceutical phage composition is provided as a kit of parts comprising one or more bacteriophages of said one or more repertoires in separate containments. Specifically, the pharmaceutical phage composition comprises a repertoire of bacteriophages, or said one or more repertoires, and a pharmaceutically acceptable carrier.

[0093] Specifically, the kit of parts comprises a) a repertoire of bacteriophages in one or more different containments; and b) a pharmaceutically acceptable carrier.

[0094] Specifically, the kit of parts comprises a) one or more repertoires of bacteriophages in one or more different containments; and b) a pharmaceutically acceptable carrier.

[0095] Specifically, the pharmaceutical phage composition is an emulsion (e.g., oil-in- water), solution (e.g., aqueous or non-aqueous), solid or semi-solid (e.g., dried, lyophilized, particulate, powder), ointment (e.g., a lotion or cream), paste, or aerosol.

[0096] Specifically, pharmaceutical phage composition comprises a storage-stable formulation.

[0097] Specifically, the formulation is for oral, topical, aural, nasal, ophthalmic application, or for intraperitoneal, intravenous, or inhalant administration.

[0098] Specifically, the formulation is for administration once a day, or more than once a day, e.g., twice a day, or once a week, or more than once a week e.g., twice a week, once a month, or more than once a month e.g., twice a month. The formulation can be provided in a suitable container or delivery device. Specifically, the formulation is comprised in kit.

[0099] Specifically, the formulation is for administration to a subject such as a human or non-human mammal.

[0100] Specifically, the pharmaceutically acceptable carrier comprises an excipient.

[0101] Specifically, the pharmaceutical phage composition comprises the repertoire of bacteriophages in combination with a drug such as e.g., an antibiotic, or a probiotic.

[0102] Specifically, the pharmaceutical phage composition is provided for treating and / or preventing an antibiotic-resistant nosocomial bacterial infection in a subject, and in particular increasing the susceptibility of an antibiotic-resistant nosocomial bacterial strain in a subject to conventional antibiotic treatment. Specifically, the phage compositions described herein can be used for reducing the virulence of an antibioticresistant nosocomial bacterial strain, in vivo or in vitro. According to a specific aspect, the methods and uses herein provide a regionspecific phage composition comprising a repertoire of bacteriophages that effectively targets at least 90%, 80%, 70%, or 60% of antibiotic-resistant strains that are prevalent in said geographical region over a time period of at least 1 , 2, 3, 4, 5, or 6 years.

[0103] According to a specific aspect, a method is provided which predicts antibioticresistant nosocomial bacterial (e.g., CRAB or CRKP) strains which are causative for a predetermined geographical region within a time period of at least 1 , 2, 3, 4, 5, or 6 years following a certain timepoint (e.g., the timepoint of selecting the repertoire of phages), by phylogeographical analysis of strains of a TOI, to identify relevant strains as a phage target which are distributed in the predetermined geographical region at the certain timepoint.

[0104] The invention further provides for the use of phylogeographical analysis and molecular typing of strains of interest (SOIs) of isolates of an antibiotic-resistant nosocomial bacterial species in a method of preparing a region-specific phage composition with relevance for a predetermined geographical region.

[0105] Specifically, the method of preparing a region-specific phage composition with relevance for a predetermined geographical region is characterized by one or more of the features of the method described herein.

[0106] Specifically, the phylogeographical analysis comprises: a) determining a number of single nucleotide polymorphisms (SNPs) in the core genome of the SOIs; and b) phylogenetic analysis using a time-scaled phylogenetic tree to visualize the chronological genomic development in the core genome.

[0107] Specifically, the core genome is a fraction of the whole genome shared by all of said SOIs, which core genome may differ in a number of SNPs.

[0108] Specifically, the number of SNPs is compared to a reference genome.

[0109] Specifically, the reference genome can be one genome of the specific TOI. The reference genome can be selected from the genome set of the TOI, such as based on the criterion of maximal genome length among available assemblies.

[0110] Specifically, methods and uses of the invention are characterized by one or more of the features as described herein.

[0111] FIGURES

[0112] Fig. 1. Relative prevalences of CRAB MLST-CPS types across the nine world regions. The least prevalent 10% of strain types are pooled in each region (other). Shades of grey indicate the relative prevalence ranges in percentage. MLST- CPS types were classified as global and prevalent. Global MLST-CPS types: ST2-KL2, ST2-KL3, ST2-KL49, ST2-KL9, ST2-KL12, ST2-KL125, ST1-KL17. Prevalent MLST- CPS types: ST2-KL160, ST2-KL7, ST2-KL77, ST2-KL235, ST492-KL104, ST636-KL40, ST2-KL152, ST2-KL22, ST499-KL18, ST1-KL18, ST15-KL22, ST25-KL139, ST25- KL22, ST78-KL3, ST79-KL49, ST79-KL9, ST16-KL24, ST2-KL10, ST2-KL6, ST2-KL13, ST2-KL234, ST2-KL81 , ST1579-KL6, ST1579-KL9.

[0113] Fig. 2. Phage sensitivity profiles of 92 CRAB isolates (columns) representing the most abundant 11 MLST-CPS types in the studied Eastern and Southern European countries. MLST-CPS types are presented in descending order of their prevalence from left to right. Grey color intensity indicates the average log 10 phage titer (PFU / mL) of three replicates for each of the 15 isolated phages. The isolates’ country of origin is indicated by the first letter of the country’s name, displayed below each column.

[0114] Fig. 3. A. baumannii phages display distinct infectivity profiles across phylogenetically diverse isolates. Phage Fishpie is active against phylogenetically divergent isolates belonging to the ST2-KL2 MLST-CPS type (A), unlike phage Porter in the case of isolates belonging to the ST2-KL12 MLST-CPS type (B). The x-axis shows the phylogenetic divergence between an isolate and a reference isolate that was sensitive to the phage. The y-axis shows the sensitivity of the isolates to the tested phage. YES / NO values indicate whether the phage can (YES) or cannot (NO) infect the isolate, i.e. clear plaques form on the lawn of the tested bacterial isolate or the bacterial growth is inhibited in the presence of the phage in liquid culture. “Same region” and “Different regions” indicate whether a certain isolate and the reference isolate originate from the same or different geographic regions. In the case of Fishpie phage the isolates originate from three geographic regions: Eastern Europe, Western Europe and Northern America, while the reference isolate originates from Eastern Europe. In the case of Porter phage all isolates originate from Eastern Europe.

[0115] Fig. 4. Constructing a precision phage cocktail against the dominant ST2- KL3 strain in Europe. A, Growth curves (ODeoo) measured for 24 hours for 41 randomly selected ST2-KL3 European isolates either in the absence (untreated) or in the presence of the Highwayman (H) and Silvergun (S) phages alone, or in combination (HS). Plotted values represent the mean ± Cl 95%. B, The effect of different phages alone or in combinations on the growth of ST2-KL3 CRAB isolates (n = 41 distinct isolates, except for the 3-phage combinations, where n = 27). Each circle represents the mean area under the bacterial growth curve value of one ST2-KL3 isolate (each measurement was carried out in three technical replicates, a.u - arbitrary unit). ** p < 0.05, *** p = 0.0001 , **** p < 0.0001 from two-sided Kruskal-Wallis test. Abbreviations: H - Highwayman, S - Silvergun, F - Fanak, Po - Porter, N - Navy-v2, Ph - PhT2-v2. C, Growth curves (ODeoo) were measured for 72 hours for 25 randomly selected ST2-KL3 isolates in the absence (untreated) or in the presence of the phage cocktail HSFPh. Plotted values represent the mean ± Cl 95%. D, Schematic representation of the KL3 CPS biosynthetic gene cluster. Functional categories of the gene products are represented with different pattern fills. The vertical arrows point to genes which harbour a loss-of-function deletion / insertion (framed with dashed line) or a point mutation (framed with dotted line) in different phage-resistant isolates listed within the frames. Text styles represent phage resistance categories as follows: Text styles represent phage resistance categories as follows: shadow, underline, bold, italic+bold and italic styles corresponding to H, S, HS, HSF and HSFPh resistance, respectively. In four cases mutations were detected in the Igt gene which is encoded trans of the KL3 gene cluster. E, Phage adsorption assay with the phage H (diamond), S (triangle), F (cube), and Ph (circle) for a wild-type ST2-KL3 isolate (Aci 110) and for its three phage-resistant derivatives: Aci 110-1 , Aci 110-2, and Aci 110-G1 which are resistant to H, HS, and HSFPh, respectively. No significant change in free phage titer indicates phage resistance is the result of cell surface structure alterations. Results show the logw reduction in free phage titres at a maximum adsorption time point in comparison to the to time point (dashed line) after mixing phages and host bacteria. Data are mean ± standard deviation, n = 3 technical replicates. F, Principle coordinate analysis plot derived from Fourier-transformed infrared measurements differentiates phage-resistant ST2-KL3 isolates that harbour loss-of- function mutations within the CPS biosynthetic pathway from those that do not ( / .e. wt and resistant lines harbouring mutation exclusively in the Igt2 gene). Symbols represent phage resistance categories as follows: cube, triangle, diamond and circle corresponding to H, HS, HSF and HSFPh resistance, respectively. Grey circle represents the wild type.

[0116] Fig. 5 Selecting the most effective phage candidate by comparing the infectivity of two phages - k146_1 (A) and k146_2 (B) - on a set of 36 Klebsiella pneumoniae clinical isolates chosen by maximizing phylogenetic diversity. All isolates belong to ST101-KL17 MLST-CPS type (TOI). The x-axis shows the phylogenetic divergence between an isolate and a reference isolate that was sensitive to the phage. The y-axis shows the sensitivity of the isolates to the tested phage, which was k146_1 (A) or k146_2 (B). YES / NO values indicate whether the phage can (YES) or cannot (NO) infect the isolate, i.e. clear plaques form on the lawn of the tested bacterial isolate or the bacterial growth is inhibited in the presence of the phage in liquid culture. “Same region” and “Different regions” indicate whether a certain isolate and the reference isolate originate from the same or different geographic regions. Isolates originate from three geographic regions: Eastern Europe, Southern Europe and Western Asia. In the case of both phages, the reference isolates originate from Southern Europe.

[0117] DETAILED DESCRIPTION

[0118] Unless indicated or defined otherwise, all terms used herein have their usual meaning in the art, which will be clear to the skilled person. Reference is for example made to the standard handbooks, such as Sambrook et al, "Molecular Cloning: A Laboratory Manual" (4th Ed.), Vols. 1 -3, Cold Spring Harbor Laboratory Press (2012); Krebs et al., "Lewin's Genes XI", Jones & Bartlett Learning, (2017), and Murphy & Weaver, "Janeway's Immunobiology" (9th Ed., or more recent editions), Taylor & Francis Inc, 2017.

[0119] The subject matter of the claims specifically refers to artificial products or methods employing or producing such artificial products, which may be variants of native (wildtype) products. Though there can be a certain degree of sequence identity to the native structure, it is well understood that the materials, methods and uses described herein, e.g., specifically referring to isolated nucleic acid sequences, amino acid sequences, fusion constructs, expression constructs, transformed host cells and modified proteins, are “man-made” or synthetic, and are therefore not considered as a result of “laws of nature”.

[0120] The terms “comprise”, “contain”, “have” and “include” as used herein can be used synonymously and shall be understood as an open definition, allowing further members or parts or elements. “Consisting” is considered as a closest definition without further elements of the consisting definition feature. Thus “comprising” is broader and contains the “consisting” definition.

[0121] The term “about” as used herein refers to the same value or a value differing by + / -5 % of the given value.

[0122] As used herein and in the claims, the singular form, for example “a”, “an” and “the” includes the plural, unless the context clearly dictates otherwise. Specific terms as used throughout the specification have the following meaning.

[0123] The term “antibiotic resistance”, as used herein shall refer to a type of drug resistance where cells of a microorganism (e.g., a bacterium or bacterial species of interest such as any bacteria involved in nosocomial infections or any mixture of such bacteria, for example Acinetobacter baumannii) have developed the ability to survive exposure to an antibiotic. The concentration of the antibiotic is one that is known to eliminate cells that lack the antibiotic resistance gene and allows for cells with the antibiotic resistance gene to survive. Typically, cells with antibiotic resistance will maintain antibiotic resistance without continued selection. However, spontaneous mutations may result in a loss of resistance, in which case, additional selection or exposure to the antibiotic may be required to eliminate cells that have lost resistance. Evolutionary stress such as exposure to antibiotics selects for the antibiotic-resistant trait. A bacterium may carry several resistance genes. Many antibiotic resistance genes reside on plasmids, whose function is to facilitate the transfer of antibiotic resistance genes from one bacterium to another. “Antibiotic resistance” has an opposite meaning as compared to “antibiotic susceptibility”, that is, a high antibiotic resistance means a low antibiotic susceptibility and vice versa.

[0124] “Antibiotic resistance” can be determined by a respective bacterial susceptibility or growth test in the presence of certain amounts of the antibiotic compound. Specifically, a bacterial species of interest is understood to be resistant against an antibiotic compound, if not affected or at least not substantially affected in the presence of a concentration of the antibiotic compound that indicates resistance.

[0125] Specific antibiotic resistance tests employ antibiotic susceptibility testing methods based on the phenotypic detection of antibiotic resistance by measuring bacterial growth in the presence of the antibiotic being tested, or different methods employing molecular techniques, microarrays, immunochromatographic methods, colorimetric methods, image methods, nephelometry, MALDI-TOF mass spectrometry, flow cytometry, chemiluminescence and bioluminescence, microfluids and methods based on cell disruption.

[0126] Results of antibacterial susceptibility testing show if bacteria are “susceptible” (can be treated with the drug), “intermediate” (may be treatable with the drug, but may require adjusted dosage), or “resistant” (cannot be treated with drug). The term “antibiotic resistant bacteria” as referred to herein shall particularly comprise those bacterial species (e.g., isolates or strains) which are classified as “resistant” or “intermediate” in the respective antibacterial susceptibility test.

[0127] The term "antibiotic resistance genes" as used herein refers to genes that confer resistance to antibiotics, for example by coding for enzymes which destroy said antibiotic compound, by coding for surface proteins which prevent the entrance of an antibiotic compound to the microorganism, actively exports it, or by being a mutated form of the antibiotic's target thereby preventing its antibiotic function.

[0128] In some aspects, the resistance gene confers resistance to a narrow-spectrum beta-lactam antibiotic of the penicillin class of antibiotics. In other aspects, the resistance gene confers resistance to methicillin (e.g., methicillin or oxacillin), or flucioxacillin, or dicloxacillin, or some or all of these antibiotics.

[0129] Specific antibiotic resistant genes include but are not limited to fosfomycin resistance gene fosB, tetracycline resistance gene tetM, kanamycin nucleotidyltransferase aadD, bifunctional aminoglycoside modifying enzyme genes aacA-aphD, chloramphenicol acetyltransferase cat, mupirocin-resistance gene ileS2, vancomycin resistance genes vanX, vanR, vanH, vraE, vraD, methicillin resistance factor femA, fmtA, mec1 , streptomycin adenylyltransferase spc1 , spc2, anti, ant2, pectinomycin adenyltransferase spd, ant9, aadA2, and any other resistance gene.

[0130] In some aspects, the pathogenic or undesired gene may be a gene encoding any gene conferring resistance to any P-lactam ( / .e., beta-lactam) antibiotic compound. In more specific embodiments, such gene may encode at least one P-lactamase. As used herein, the term “P-lactamase” denotes a protein capable of catalyzing cleavage of a 0- lactamase substrate such as a P-lactam containing molecule (such as a P-lactam antibiotic) or derivative thereof.

[0131] The term "P-lactam" or "|3 lactam antibiotics" as used herein refers to any antibiotic agent which contains a beta-lactam ring in its molecular structure, P-lactam antibiotics are a broad group of antibiotics that include different classes such as carbapenems, clavulanic acid, natural and semi-synthetic penicillins, penicillin derivatives (penams), cephalosporins (cephems), cephamycins and monobactams, that is, any antibiotic agent that contains a P-lactam ring in its molecular structure. They are the most widely-used group of antibiotics. While not true antibiotics, the P-lactamase inhibitors are also included in the group of antibiotics, P-lactam antibiotics are analogues of D-alanyl-D- alanine the terminal amino acid residues on the precursor NAM / NAG-peptide subunits of the nascent peptidoglycan layer. The structural similarity between P-lactam antibiotics and D-alanyl-D-alanine prevents the final crosslinking (transpeptidation) of the nascent peptidoglycan layer, disrupting cell wall synthesis. Under normal circumstances peptidoglycan precursors signal a reorganisation of the bacterial cell wall and, as a consequence, trigger the activation of autolytic cell wall hydrolases. Inhibition of crosslinkage by P-lactams causes a buildup of peptidoglycan precursors, which triggers the digestion of existing peptidoglycan by autolytic hydrolases without the production of new peptidoglycan. As a result, the bactericidal action of beta-lactam antibiotics is further enhanced. Generally, P-lactams are classified and grouped according to their core ring structures, where each group may be divided into different categories.

[0132] The term “carbapenem”, as used herein refers to a class of beta-lactam containing antibiotics including imipenem, meropenem, ertapenem, doripenem, panipenem, betamipron, biapenem, and tebipenem.

[0133] Beta-lactams containing pyrrolidine rings are named carbapenams. A carbapenam is a P-lactam compound that is a saturated carbapenem. They exist primarily as biosynthetic intermediates on the way to the carbapenem antibiotics. Carbapenems have a structure that renders them highly resistant to P-lactamases and therefore are considered as the broadest spectrum of P-lactam antibiotics. The carbapenems are structurally very similar to the penicillins, but the sulfur atom in position 1 of the structure has been replaced with a carbon atom, and hence the name of the group, the carbapenems. Carbapenem antibiotics were originally developed from thienamycin, a naturally-derived product of Streptomyces cattleya. The carbapenems group includes: biapenem, doripenem, ertapenem, imipenem, meropenem, panipenem and PZ-601 . The term "penam" is used to describe the core skeleton of a member of a penicillin antibiotic, i.e. a P-lactam containing a thiazolidine ring. Penicillins contain a 0- lactam ring fused to a 5-membered ring, where one of the atoms in the ring is sulfur and the ring is fully saturated. Penicillins may include narrow-spectrum penicillins, such as benzathine penicillin, benzylpenicillin (penicillin G), phenoxymethylpenicillin (penicillin V), procaine penicillin and oxacillin. Narrow spectrum penicillinase-resistant penicillins include methicillin, dicloxacillin and flucioxacillin. The narrow spectrum P-lactamase- resistant penicillins may include temocillin. The moderate-spectrum penicillins include for example, amoxicillin and ampicillin. The broad-spectrum penicillins include the co- amoxiclav (amoxicillin+clavulanic acid). Finally, the penicillin group also includes the extended spectrum penicillins, for example, azlocillin, carben icillin , ticarcillin, mezlocillin and piperacillin. Other members of this class include pivampicillin, hetacillin, bacampicillin, metampicillin, talampicillin, epicillin, carbenicillin, carindacillin, ticarcillin, azlocillin, piperacillin, mezlocillin, mecillinam, pivmecillinam, sulbenicillin, clometocillin, procaine benzylpenicillin, azidocillin, penamecillin, propicillin, pheneticillin, cioxacillin and nafcillin. P-lactams containing 2, 3-dihydrothiazole rings are named penems. Penems are similar in structure to carbapenems. However, where penems have a sulfur, carbapenems have another carbon. There are no naturally occurring penems; all of them are synthetically made. An example for penems is faropenem. 0- lactams containing 3, 6-dihydro-2H-l, 3-thiazine rings are named cephems. Cephems are a subgroup of b- lactam antibiotics and include cephalosporins and cephamycins. The cephalosporins are broad-spectrum, semisynthetic antibiotics, which share a nucleus of 7- aminocephalosporanic acid. First generation cephalosporins, also considered as the moderate spectrum includes cephalexin, cephalothin and cefazolin. Second generation cephalosporins that are considered as having moderate spectrum with anb- Haemophilus activity may include cefaclor, cefuroxime and cefamandole. Second generation cephamycins that exhibit moderate spectrum with anti-anaerobic activity include cefotetan and cefoxitin. Third generation cephalosporins considered as having broad spectrum of activity includes cefotaxime and cefpodoxime.

[0134] The fourth generation cephalosporins considered as broad spectrum with enhanced activity against Gram positive bacteria and P-lactamase stability include the cefepime and cefpirome. The cephalosporin class may further include: cefadroxil, cefixime, cefprozil, cephalexin, cephalothin, cefuroxime, cefamandole, cefepime and cefpirome.

[0135] Cephamycins are very similar to cephalosporins and are sometimes classified as cephalosporins. Like cephalosporins, cephamycins are based upon the cephem nucleus. Cephamycins were originally produced by Streptomyces, but synthetic ones have been produced as well. Cephamycins possess a methoxy group at the 7-alpha position and include: cefoxitin, cefotetan, cefmetazole and flomoxef.

[0136] Beta-lactams containing 1 , 2, 3, 4-tetrahydropyridine rings are named carbacephems. Carbacephems are synthetically made antibiotics, based on the structure of cephalosporin, a cephem. Carbacephems are similar to cephems but with a carbon substituted for the sulfur. An example of carbacephems is loracarbef.

[0137] Monobactams are b-lactam compounds wherein the P-lactam ring is alone and not fused to another ring (in contrast to most other P-lactams, which have two rings). They work only against Gram negative bacteria. Other examples of monobactams are tigemonam, nocardicin A and tabtoxin.

[0138] P-lactams containing 3, 6-dihydro-2H-l, 3-oxazine rings are named oxacephems or clavams. Oxacephems are molecules similar to cephems, but with oxygen substituting for the sulfur. Thus, they are also known as oxapenams. An example for oxapenams is clavulanic acid. They are synthetically made compounds and have not been discovered in nature. Other examples of oxacephems include moxalactam and flomoxef.

[0139] Another group of P-lactam antibiotics is the P-lactamase inhibitors, for example, clavulanic acid. Although they exhibit negligible antimicrobial activity, they contain the 0- lactam ring. Their sole purpose is to prevent the inactivation of P-lactam antibiotics by binding the P-lactamases, and, as such, they are co-administered with P-lactam antibiotics, P-lactamase inhibitors in clinical use include clavulanic acid and its potassium salt (usually combined with amoxicillin or ticarcillin), sulbactam and tazobactam.

[0140] The term "nosocomial Infections" refers to hospital-acquired infections, namely, an infection whose development is favored by a hospital environment, such as surfaces and / or medical personnel, and is acquired by a patient during hospitalization. Nosocomial infections are infections that are potentially caused by organisms resistant to antibiotics. Nosocomial infections have an impact on morbidity and mortality, and pose a significant economic burden. In view of the rising levels of antibiotic resistance and the increasing severity of illness of hospital in-patients, this problem needs an urgent solution.

[0141] Specific nosocomial infections are caused by bacterial species referred to as “nosocomial bacterial species”.

[0142] As used herein, the term “antibiotic-resistant nosocomial bacterial species” also understood as “nosocomial antibiotic-resistant bacteria”, which are herein understood as “target cells” (in particular target cells of phage therapy) or bacterial species of interest, shall refer to those bacterial species of strains, which are resistant to the antibiotic drug and may cause or are causative agents to nosocomial infections.

[0143] Common nosocomial bacteria include bacteria of the genus Acinetobacter, Klebsiella, Enterococcus, Staphylococcus, Pseudomonas, Escherichia, and Enterobacter.

[0144] Specific common examples of nosocomial bacteria include Acinetobacter baumannii, Klebsiella pneumoniae, Clostridium difficile, methicillin-resistant Staphylococcus aureus, coagulase-negative Staphylococci, vancomycin-resistant Enteroccocci, resistant Enterobacteriaceae, Pseudomonas aeruginosa, and Stenotrophomonas maltophilia.

[0145] The nosocomial-infection pathogens could be subdivided into Gram-positive bacteria (Staphylococcus aureus, Coagulase-negative staphylococci), Gram-positive cocci (Enterococcus faecalis and Enterococcus faecium), Gram-negative rod-shaped organisms (Klebsiella pneumonia, Klebsiella oxytoca, Escherichia coll, Proteus aeruginosa, Serratia spp.), Gram-negative bacilli (Enterobacter aerogenes, Enterobacter cloacae), aerobic Gram-negative coccobacilli (Acinetobacter baumannii, Stenotrophomonas maltophilia) and Gram-negative aerobic bacillus (Stenotrophomonas maltophilia, previously known as Pseudomonas maltophilia). Among many others Pseudomonas aeruginosa is an important nosocomial Gramnegative aerobic rod pathogen.

[0146] In particular and by non-limiting embodiments, a target cell may be an antibioticresistant target cell, or any mixture or population comprising said cells. Of particular interest is any of the “ESKAPE” pathogens. As indicated herein, these pathogens include but are not limited to Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter.

[0147] Thus, the target cell may be bacteria of any strain of at least one of E. coli, Pseudomonas aeruginosa, Staphylococcus aureus, Streptococcus pyogenes, Clostidium difficile, Enterococcus faecium, Klebsiella pneumonia, Acinetobacter baumanni and Enterobacter species (specifically, ESKAPE bacteria).

[0148] In further embodiments, the bacterial species of interest may include Yersinia enterocolitica, Yersinia pseudotuberculosis, Salmonella typhi, Pseudomonas aeruginosa, Vibrio cholerae, Shigella sonnei, Bordetella Pertussis, Plasmodium falciparum, Chlamydia trachomatis, Bacillus anthracis, Helicobacter pylori and Listeria monocytogens.

[0149] In other specific embodiments, the bacterial species of interest may be any E.coli strain, specifically, any one of 0157:1-17, enteroaggregative (EAEC), enterohemorrhagic (EHEC), enteroinvasive (EIEC), enteropathogenic (EPEC), enterotoxigenic (ETEC) and diffuse adherent (DAEC) E. coli.

[0150] The term “bacterial species” as used herein shall include bacteria or a certain genus or species, and particularly includes bacterial strains or isolates. The term “strain”, as used herein refers to a bacterial species exhibiting phenotypic and / or genotypic traits belonging to the same lineage, distinct from those of other strains of the same species.

[0151] The term “bacterial isolate”, as used herein refers to a bacterial culture separated from its natural origin, for example, an isolate obtained by culturing a single bacterial colony, such as derived from a heterogeneous population of bacteria.

[0152] The term “library” as used herein shall refer to a collection (in particular, a combination) of library members which are a collection of phages, or respective nucleic acid molecules (e.g., a library of genomes, fragmented genomes, or polynucleotides of phages). The library members share common features but differ in at least one mutation and / or phenotype. A library typically contains library members which are diverse, besides those that have common features.

[0153] Specifically, the present disclosure refers to a library of phages comprising a variety of different phages, such as from one or more phage sources.

[0154] In some aspects, a phage source may be a biological source material e.g., an environmental source, preferably a water, soil, sewage, aquatic habitats, agricultural runoff, hospital environments, food processing facilities, animal feces, or plant source, or a biological sample of subject or host organism, such as e.g., human or non-human animals, in particular subjects or hosts which are infected with microorganisms that are targeted by phages.

[0155] Specific phage sources may comprise a collection of phages which are wild-type phages (such as naturally-occurring in biological samples) such as obtained from an isolated biological material.

[0156] Specific biological materials used for isolating wild-type phages are e.g., samples of waste water, soil, sewage, aquatic habitats, agricultural runoff, hospital environments, food processing facilities, or animal feces or a plant source, or biological samples of subjects of hosts that are infected with a bacterium of interest e.g., clinical samples, or samples comprising a microbiome.

[0157] A "microbiome", as used herein, refers to the ecological community of commensal, symbiotic, or pathogenic microorganisms in a sample. Examples of microbiomes can include but are not limited to skin microbiome, umbilical microbiome, vaginal microbiome, conjunctival microbiome, intestinal microbiome, stomach microbiome, gut microbiome and oral microbiome, nasal microbiome, gastrointestinal tract microbiome, and urogenital tract microbiome. In some aspects, a phage source may comprise phages that comprise artificial mutants of wild-type phages, such as phage variants obtained by mutagenesis of the phage genome, which phage variants may comprise a different phenotype such as a different host specificity or a different (in particular extended) range of host specificity.

[0158] In some aspects, a phage source may comprise phages that are deposited in publicly available phage collections, such as the German Collection of Microorganisms and Cell Cultures GmbH DSMZ phage library.

[0159] The phage libraries described herein are specifically characterized by a size (which is understood as the number of diverse library members, i.e., the library diversity) which is at least 10, 102, 103, 104, 105,, library members, which are each characterized by different nucleic acid sequences. Specific phage libraries comprise the library members in separate containments, or in a mixture.

[0160] Phage libraries are advantageously used to select specific library members for one or more phenotypic activities such as to study the interaction of a phage library member with a bacterial species of interest e.g., by determining the lytic activity against the bacterium of interest. Phage library members can be selected using respective screening methods.

[0161] The screening may thus comprise one or more assays of phage activities, in particular comprising an assay to determine lytic activity, and / or selection according to the prevalence of the phages such as using geographical and / or historical data.

[0162] As a result of a screening process, a repertoire of phages can be obtained.

[0163] The term “repertoire” as used herein shall refer to a collection of species, such as phages with a variety of host specificities. The variety will specifically reflect the diversity of host specificities or ranges of host specificity e.g., to cover at least a certain percentage of all target bacterial species prevalent in a geographical region.

[0164] Specifically, the repertoire is provided in the form of a cocktail, in particular a mixture, or a kit of parts comprising the repertoire in two or more separate containments.

[0165] Specifically, the repertoire comprises a collection of isolated phages.

[0166] As used herein, the term “isolated” with reference to a bacteriophage, refers to a bacteriophage which is removed from its natural environment (e.g., removed from bacteria which it typically infects), or removed from cellular material and / or other elements that naturally exist in a biological source material. Specifically, an isolated bacteriophage is understood to be partially or completely separated from components with which it is normally associated. Methods of isolating and characterizing a bacteriophage are known in the art and include, e.g., enrichment from an environmental sample, plaque testing, culture lysis, and / or routine test dilution. Characterization procedures used to show that a phage may have utility in phage therapy include, but are not limited to, e.g., plating efficiency, phage morphology by microscopy, whole genome sequencing, one-step growth curves, and / or pulse- field gel electrophoresis.

[0167] Specifically, the repertoire comprises a collection of lytic phages, in particular phages with a certain lytic activity.

[0168] As used herein, the term “lytic” with reference to a bacteriophage, refers to a bacteriophage that infects a bacterial host and causes that host to lyse without incorporating the phage nucleic acids into the host genome. A lytic bacteriophage is typically not capable of reproducing using the lysogenic cycle.

[0169] Specifically, the repertoire comprises a collection of phages which covers a certain host specificity, or a range of host specificities.

[0170] The term “host specificity” as used herein is understood as follows. A bacteriophage comprises a certain recognition element ( / .e., "host-recognition element"), which is a phage component associated with phage-host recognition, in particular an element mediating the interaction between the phage and the host, such as an element of the phage, which participates, facilitates, improves or enables at least one of host recognition, attachment to the host, penetration, injection of the nucleic acid molecules (or any other transduced material), and / or stability of the injected material within the host (e.g., resistance to the host restriction enzymes, and the like).

[0171] Specifically, the host recognition element of a phage is localized at the tail-end of the bacteriophage.

[0172] Specifically, a repertoire of phages with relevance in a geographical region can be provided in a pharmaceutical composition designed to treat subjects at risk of being infected with at least one of the target bacterial species in said geographical region.

[0173] In some aspects, the repertoire of phages can be provided as a mixture of phages or kit of parts, wherein the repertoire is provided in one or more separate containments. The repertoire of phages can also be provided as nucleic acid molecules encoding the phage repertoire, or a repertoire of (e.g., non-pathogenic) host cells comprising the phage repertoire.

[0174] The term “pharmaceutical composition”, as used herein refers to a composition comprising one or more active compounds in a formulation comprising a pharmaceutically acceptable carrier, such as including but not limited to any excipients, stabilizers, diluents, dispersing agents, suspending agents, and / or thickening agents. A pharmaceutical composition may facilitate administration of said active compound(s) to an organism for treatment. Specifically described herein is a pharmaceutical composition comprising one or more, in particular a repertoire of bacteriophages, as active ingredients.

[0175] In some aspects, the pharmaceutically acceptable carrier is non-toxic, biocompatible, and is selected so as not to detrimentally affect the biological activity of the phage. The phages may be formulated into preparations for local delivery ( / .e., to a specific location of the body) or systemic delivery, in solid, semi-solid, gel, liquid, or gaseous forms such as tablets, capsules, powders, granules, ointments, solutions, depositories, inhalants, and injections allowing for oral, parenteral, or surgical administration. In some aspects, local administration of the pharmaceutical composition is contemplated by coating a medical device.

[0176] Suitable carriers for parenteral delivery e.g., via injectable, infusion, or irrigation and topical delivery include distilled water, physiological phosphate-buffered saline, normal or lactated Ringer’s solutions, dextrose solution, Hank’s solution, propanediol, a biocompatible oil such as including synthetic mono- or diglycerides, or fatty acids such as oleic acid. The carrier and phage may be compounded as a liquid, suspension, polymerizable or non-polymerizable gel, paste or salve. The carrier may also comprise a delivery vehicle to sustain ( / .e., extend, delay or regulate) the delivery of the phage or to enhance the delivery, uptake, stability, or pharmacokinetics of the phage. Such a delivery vehicle may include e.g, microparticles, microspheres, nanospheres, or nanoparticles composed of proteins, liposomes, carbohydrates, synthetic organic compounds, inorganic compounds, polymeric or copolymeric hydrogels, and polymeric micelles. Solutions of pharmaceutical compositions can be prepared in water suitably mixed with a surfactant, such as hydroxypropylcellulose. Dispersions can be prepared in glycerol, liquid polyethylene glycols, mixtures thereof, and in oils. Under ordinary conditions of storage and use, these preparations contain a preservative to prevent the growth of microorganisms.

[0177] In specific aspects, the actual dosage amount of a composition administered to a subject can be determined by physical and physiological factors such as body weight, severity of condition, the type of disease being treated, previous or concurrent therapeutic interventions, idiopathy of the subject, and the route of administration. In specific aspects, the pharmaceutical composition is advantageously administered in the form of an injectable composition either as liquid solutions or suspensions; solid forms suitable or solution in, or suspension in, liquid prior to injection can also be prepared. These preparations also may be emulsified.

[0178] Specific pharmaceutically acceptable carriers comprise aqueous solutions, including e.g., non-toxic excipients such as salts, preservatives, buffers and the like. Specific non-aqueous solvents are propylene glycol, polyethylene glycol, vegetable oil, and injectable organic esters such as ethyloleate.

[0179] Specific oral formulations include excipients such as mannitol, lactose, starch, magnesium stearate, sodium saccharine, cellulose, magnesium carbonate, and the like. Exemplary oral compositions take the form of solutions, suspensions, tablets, pills, capsules, sustained release formulations, or powders.

[0180] The pharmaceutical compositions is e.g., formulated for oral, nasal, buccal, rectal, vaginal, topical, orthotopic, intradermal, subcutaneous, intramuscular, intraperitoneal or intravenous use. For treatment of conditions of the lungs, aerosol delivery can be used.

[0181] In specific aspects, the pharmaceutical composition is provided in combination with other agents such as e.g., antimicrobial agents, in particular an antibiotic compound, or a probiotic. Various antibiotics can be used such as any one or more of carbapenems (e.g., imipenem / cilastatin, meropenem, ertapenem), cephalosporins (e.g., cefotaxime, ceftriaxone), aminoglycosides (e.g., gentamicin, amikacin), beta-lactams (e.g., ceftolozane / tazobactam, ceftazidime / avibactam, vaborbactam), quinolones (e.g., ciprofloxacin, levofloxacin, norfloxacin, moxifloxacin), ampicillin, avibactam, aztreonam, rifampin, sulbactam, piperacillin, tazobactam, ticarcillin, clavulanate, ceftazidime, cefepime, colistin, tigecycline, trimethoprim, sulfamethoxazole, or metronidazole.

[0182] In specific aspects, the pharmaceutical composition is provided on or within a device such as a medical device.

[0183] In particular, the repertoire of phages or pharmaceutical preparation as described herein is provided for medical use to treat a subject or patient in need of prophylaxis or treatment of a disease condition.

[0184] The term “subject” as used herein shall refer to a warm-blooded mammalian, particularly a human being or a non-human animal. Thus, the term “subject” may also particularly refer to animals including dogs, cats, rabbits, horses, cattle, pigs and poultry. The term “patient” includes human and other mammalian subjects that receive either prophylactic or therapeutic treatment. The term “treatment” is thus meant to include both prophylactic and therapeutic treatment.

[0185] Therefore, the present invention provides for a new strategy to combat the escalating crisis of antibiotic resistance, especially in the context of hospital-acquired multi-drug resistant infections, and provides for a new precision phage therapy guided by genomic surveillance. It was demonstrated how genomic surveillance can guide phage therapy for carbapenem-resistant A. baumannii (CRAB), offering a scalable and cost-effective strategy. Instead of engaging in the time-consuming process of isolating individual phages for each patient's bacterial strain, genomic surveillance was employed to assemble region-specific phage compositions that target the majority of infections within a particular geographical area.

[0186] Several significant novel aspects were applied. First, a global analysis of over ten thousand CRAB genomes was performed, identifying 29 prevalent CRAB types responsible for the majority of infections worldwide. Despite inter-regional variation, these prevalent types exhibit homogeneity within specific world regions, suggesting the potential effectiveness of region-specific phage collections for CRAB targeting. Second, phylogeographical analysis revealed a surprising temporal stability of CRAB populations within countries over a six-year period, providing a crucial time frame for these preemptively prepared region-specific phage collections. Third, in Eastern Europe, a region with high CRAB infection rates, a phylogeny-guided phage hunt was performed, matching the 11 most prevalent CRAB types with 15 new phages with therapeutic potential. Additionally, the host range of these phages was characterized using 199 clinical isolates, allowing to assess the geographic scope of their applicability. Finally, as a proof-of-concept, a precision phage cocktail was formulated that effectively inhibited a broad range of clinical isolates of the globally most prevalent CRAB type, demonstrating minimal emergence of resistant variants and efficacy in in vivo animal infection models.

[0187] The described phage therapy framework aims to design phage cocktails that provide benefits to the highest number of patients across geographical scales. This will facilitate study recruitment, crucially speeding-up clinical trials and leading to clinical validation. Moreover, a scale-up will lead to enhanced cost-effectiveness for personalized phage therapy, which will ensure the precision and practicality of phage therapy in clinical settings. Furthermore, by creating targeted collections of off-the-shelf medicinal phage products against nosocomial pathogens, the present invention provides for phage therapy as a viable adjunctive treatment for managing acute infections that require immediate intervention.

[0188] Overall, the present invention pioneers the application of bacterial phylogeography to scaling up phage therapy, not only with CRAB, but also other antibiotic-resistant nosocomial bacteria. Because of the rapidly increasing number of available genome sequences, the genomic surveillance-based framework provided herein is not limited to CRAB and can be readily applied to target other nosocomial pathogens. This broadens the potential impact of the present examples, making its findings applicable beyond CRAB and contributing to the development of effective strategies for combating a range of infectious diseases in healthcare settings.

[0189] According to specific examples, clinical isolates of antibiotic-resistant isolates (e.g., CRAB or CRKP isolates) are prioritized in order to select those strain types that are relevant from the point of drug development. Relevant means they are prevalent in the given geographic region or likely become prevalent in the future and are resistant to conventional antibiotics.

[0190] An exemplary protocol for the definition of the relevant CRAB strains (CRAB strain selections) is as follows:

[0191] - Acquiring at least 10,000 A. baumannii whole genome sequences from public repositories and / or private collections with metadata about the collection date and geographical location for each isolate.

[0192] - Selecting CRAB genomes based on genomic resistance determinants to find isolates that are likely resistant to the clinically relevant antibiotic classes: including but not limited to carbapenem, fluoroquinolone, aminoglycoside. Genomic resistance determinants mean presence / absence of genomic features, e.g. genes, or point mutations within resistance genes that relevant databases contain. The list of determinants is based on public scientific knowledge and may change over time. Alternatively, if the bacterial sample is available, bacterial susceptibility to antibiotics can be measured using standard methods based on EUCAST guidelines.

[0193] - Assigning CRAB genomes to groups using genome-based typing methods. Genomes that have the same “Type” are assigned to the same group. These groups are called strain types. It is assumed that a single phage are effective within a group regardless of other genome content. A preferred typing scheme contains MLST and CPS typing. Specifically, MLST clusters evolutionary- related genomes into sequence types (ST) based on the sequence of multiple housekeeping genes. In contrast, CPS typing identifies specific genomic determinants responsible for synthesizing structurally different CPSs, which can function as receptors for different Acinetobacter phages.

[0194] - Optionally, employing a genome downsampling strategy to remove e.g. superfluous samples from the same outbreak which would otherwise distort downstream analysis which requires a balanced data set. For downsampling, with each strain type, select one isolate from each city and each month where the city is available, otherwise select one isolate from each country and each week.

[0195] - Building phylogenetic trees for the most common (in particular, most prevalent) sequence types. Most common sequence types are those that cumulatively account for at least 50% of the available genomes. The phylogenetic trees is made separate for each sequence type (TOI). The trees then are used to determine a time frame within which local (country or city) transmissions dominate.

[0196] - Selecting strain types (previously defined by the MLST and CPS typing schemes) that have a prevalence of at least 2% or 3% or 4% or 5% in the investigated geographic region in the most recent time frame when local transmissions dominate (time frame comes from the previous step). These strains can be labelled as “prevalent”.

[0197] - Optionally adding to the strain type list those strain types that are not yet prevalent in the region, but exhibit increasing prevalence trends on the continent or in the country (depending on the target region. Increasing prevalence trends are determined by statistical modelling: temporal trends are analysed both globally and at the level of individual continents. To account for potential biases caused by geographical differences within the studied regions, continent can be included as a covariate when inferring temporal trends on a global scale. In a similar vein, world regions, e.g. Eastern Europe can be included as a covariate when inferring temporal trends for continents. False Discovery Rate (FDR) correction can be applied to account for multiple comparisons. At the end of the analysis, those “Types” that have corrected p- value below 0.05 and a regression coefficient above 0 either globally or in a continent can be classified as having “increasing prevalence trends”.

[0198] - Optionally, adding to the strain type list those strain types that were found in the region within the time frame that had been defined for local transmissions, and are both A) prevalent, and B) have increasing prevalence in another continent.

[0199] - Optionally, removing strain types from the selection that have a decreasing prevalence and are likely to disappear in the near future in the focal geographic region.

[0200] An exemplary protocol for the phage selection targeting selected CRAB strains is as follows: the aim of phage selection is to isolate phages that target the selected target strain types. It may comprise of the following steps:

[0201] - Selecting a small set of phylogenetically diverse isolates (between 1 -5) of each selected CRAB strain type to perform a phage selection experiment. Phylogenetically diverse means the isolates are not clustered on specific branches of the tree. Phage selection may mean isolating a new phage from natural sources e.g. wastewater or selecting a phage from an existing phage collection. The location of natural waste water sources is typically selected within or near to the region where the strains types are prevalent.

[0202] - Selecting phages that exhibit potent lytic activity (>106pfu) against the target clinical isolate. PFU (plack forming units) are e.g., defined by plack assay using the double-layer agar method.

[0203] An exemplary protocol for phage prioritization is as follows: the aim of phage prioritization is to retain the phages with the broadest phage host range and highest lytic activity thereby limit the number of phages in the phage composition. It may comprise the following steps:

[0204] - Selecting a larger set of phylogenetically diverse isolates in the region (at least 5 and up to 20) of each CRAB strain type to perform a phage sensitivity experiment. Phage sensitivity means testing if the phage has a lytic activity higher than 106pfu or higher than 105, or higher than 103against the target clinical isolate e.g., by the double layer agar method. An alternative method to quantify strain sensitivity to a phage is to measure the ability of the phage to suppress bacterial growth over time compared to the growth curve of a bacterial culture devoid of phages. Improved antibacterial activity results in significantly slower growth or complete eradication of the investigated strain. The growth of the bacteria can be quantified by measuring the OD600 values of the bacterial culture at every 20 min for 24 or 72 hours.

[0205] - Selecting an even larger set of phylogenetically diverse isolates from outside of the region (at least 5 and up to 20) of each CRAB strain type to perform a phage sensitivity experiment.

[0206] - Prioritizing phages (for inclusion in the phage composition) against the strain of interest (SOI) based on the results of the phage sensitivity experiments above. Typically, those phages are prioritized that have broader host range within its target strain type. Broader host range means having potent lytic activity against a larger set of tested isolates belonging to the TOI. As isolates are selected to represent large genetic distances within or even outside the region of interest, broad host range phages are more likely applicable in a larger part of the region of interest. In this context, genetic distance is defined as the phylogenetic distance of a pair of isolates. This is determined from the phylogenetic tree that had been constructed under “prioritizing target strains”.

[0207] The invention is further described by one or more of the following items.

[0208] 1 . A method of preparing a region-specific phage composition effectively targeting an antibiotic-resistant nosocomial bacterial species, comprising: a) performing molecular typing of isolates of the bacterial species to identify a type of interest (TOI) that is prevalent in a predetermined geographical region; b) performing phylogeographical analysis of strains of said TOI to identify target strains within said TOI, which are the most phylogenetically diverse in said predetermined geographical region; c) selecting a repertoire of bacteriophages from a bacteriophage source, which repertoire exhibits lytic activity against said target strains; and d) preparing a pharmaceutical phage composition comprising said repertoire.

[0209] 2. The method of item 1 , which comprises: a) performing molecular typing of isolates of the bacterial species to identify one, two or more TOIs that are prevalent in said predetermined geographical region; b) performing phylogeographical analysis of strains of each of said TOIs to identify target strains within each of said TOIs, which are the most phylogenetically diverse in said predetermined geographical region; c) selecting one, two or more repertoires of bacteriophages from a bacteriophage source, wherein said repertoires exhibit lytic activity against target strains of said TOIs; and d) preparing a pharmaceutical phage composition comprising said repertoires.

[0210] 3. The method of item 1 or 2, wherein molecular typing comprises DNA sequencing and clustering the genomes into genotypes, preferably comprising any one or more or all of multilocus sequence typing (MLST), cell surface capsular polysaccharide (CPS) typing, antibiotic-resistance typing, whole genome sequence typing, or multilocus variable-number tandem-repeat analysis (MLVA) typing.

[0211] 4. The method of any one of items 1 to 3, wherein the bacterial species is selected from the group consisting of: Acetinobacter baumannii, Klebsiella pneumoniae, Enterococcus faecium, Staphylococcus aureus, Pseudomonas aeruginosa, Escherichia coll and Enterobacter cloaceae, preferably wherein the bacterial species is resistant against an antibiotic selected from the group consisting of: beta-lactams, such as carbapenem or non-carbapenem beta-lactams, aminoglycosides, fluoroquinolones sulfonamides, tetracyclines, macrolides, glycopeptides, and polymyxins.

[0212] 5. The method of any one of items 1 to 4, wherein the isolates of bacterial species are carbapenem-resistant Acetinobacter baumannii (CRAB) strains.

[0213] 6. The method of any one of items 1 to 5, wherein the predetermined geographical region is any one or more of Western Europe, Eastern Europe, Southern Europe, North- America, South-America; West-Asia, East-Asia, or South-East-Asia.

[0214] 7. The method of item 5 or 6, wherein the predetermined geographical region is Eastern and Southern Europe, and the repertoire of bacteriophages exhibits lytic activity against CRAB strains of the ST2 lineage, preferably the ST2-KL3 strain type.

[0215] 8. The method of item 7, wherein the pharmaceutical phage composition comprises one or more repertoires of bacteriophages that target a) at least any one of 60, 70, 80, or 90% of isolates belonging to one or more TOIs; and / or b) at least any one of 60, 70, 80, or 90% of representative isolates prevalent in said predetermined geographical region.

[0216] 9. The method of any one of items 1 to 8, wherein the lytic activity is at least 106pfu (in particular, at least 106pfu per mL) against the target strains.

[0217] 10. The method of any one of items 1 to 9, wherein the bacteriophage source is originating from an environmental sample of the predetermined geographical region, preferably wherein the environmental sample is of waste water, soil, sewage, aquatic habitats, agricultural runoff, hospital environments, food processing facilities, animal feces or a plant source.

[0218] 11 . The method of any one of items 1 to 10, wherein the bacteriophage source is a phage library, preferably wherein the phage library comprises at least 10, 15, 20, 30, 40, 50, 100, or 200 different bacteriophages.

[0219] 12. The method of any one of items 1 to 11 , wherein a repertoire of bacteriophages comprises or consists of 3-15 different bacteriophages.

[0220] 13. The method of any one of items 1 to 12, wherein said MLST clusters evolutionary-related genomes into sequence types (ST) based on the sequence of housekeeping genes.

[0221] 14. The method of any one of items 1 to 13, wherein the pharmaceutical phage composition comprises one or more repertoires of bacteriophages in a pharmaceutical formulation, or is provided as a kit of parts comprising one or more bacteriophages of said one or more repertoires in separate containments.

[0222] 15. Use of phylogeographical analysis and molecular typing of strains of interest (SOIs) of isolates of an antibiotic-resistant nosocomial bacterial species in a method of preparing a region-specific phage composition with relevance for a predetermined geographical region.

[0223] The foregoing description will be more fully understood with reference to the following examples. Such examples are, however, merely representative of methods of practicing one or more embodiments of the present invention and should not be read as limiting the scope of invention.

[0224] EXAMPLES

[0225] The Examples illustrate the following process, which involves: (1) Collecting isolates of the target pathogen from surveillance programs and laboratory networks, using information on genomic sequence, and space and time of isolation; (2) Identifying dominant pathogen strain types requiring distinct bacteriophages based on pathogen genomics and phage sensitivity profiling; (3) Utilising the spatio-temporal distribution of the identified dominant pathogen types to design region-specific phage therapy interventions. These interventions maximise the therapeutic potential for the largest number of patients across geographical scales. Example 1: Methods a) Establishing a collection of Acinetobacter baumannii isolates

[0226] Due to the underrepresentation of genomic samples from the Eastern and Southern European region, 419 Carbapenem-resistant Acinetobacter baumannii CRAB clinical isolates from 5 countries from these regions were collected and sequenced (Hungary (n=253), Romania (n=120), Serbia (n=28), Montenegro (n=9), and Bosnia and Herzegovina (n=9)) between 2017 and 2022 from more than 40 healthcare facilities in 34 cities. These isolates were categorised as CRAB strains after performing the standard antimicrobial susceptibility tests according to the methods of the European Committee on Antimicrobial Susceptibility Testing (EUCAST). Nine antimicrobial agents were tested regularly including amikacin, gentamycin, tobramycin, ciprofloxacin, levofloxacin, imipenem, meropenem, trimethoprim-sulfamethoxazole, and colistin. The minimum inhibitory concentrations (MICs) were determined by E-test for carbapenems and by broth microdilution in the case of colistin. The susceptibility to the remaining antibiotics was determined by the disc diffusion method. The susceptibility test results were interpreted based on the EUCAST breakpoints (http: / / www.eucast.org). CRAB isolates were defined with both imipenem (MIC > 4 mg / L) and meropenem resistance (MIC > 8 mg / L). To study the host range of the isolated bacteriophages on an extended set of CRAB isolates, 16 additional CRAB isolates were obtained from BEI Resources (https: / / www.niaid.nih.gov / research / bei-resources-repository) and 47 from the Belgian Acinetobase collection (Valcek, A., et al. 2022). b) Genome sequencing

[0227] For the isolation of the genomic DNA, all Acinetobacter strains were routinely cultured on LB broth (LB, 5 g of tryptone, 2.5 g of yeast extract, 5 g of NaCI, and 500 ml of distilled water) or LB agar (LB plus 1.5% Bacto agar, w / v). For all plaque assays, a 0.5% LB agar overlay (LB and 0.5% Bacto agar w / v) was inoculated with 0.1 ml of a fresh overnight LB culture of the host and poured over LB agar plates.

[0228] All strains were grown at 37°C. Genomic DNA of 410 A. baumannii isolates was extracted using the GenElute™ Bacterial Genomic DNA Kit (Sigma-Aldrich) and sequenced by Illumina sequencing.

[0229] Sequencing libraries were prepared by the NexteraXT library preparation kit from Illumina according to the manufacturer protocol.

[0230] For the NexteraXT library preparation kit, genomic DNA has been fragmented to approx. 300 nt fragments and Illumina sequencing adaptors have been added by using tagmentation. Then sample-specific indexes have been added to each sample by PCR. Pooled libraries have been sequenced by Illumina NextSeq 500 using 2x150 PE sequencing chemistry in multiple sequencing runs.

[0231] Sequencing ready libraries were quality control checked by BioAnalyzer2100 instrument using High Sensitivity DNA Chip (Agilent Technologies USA, Cat. No. 5067- 4626). Sequencing was carried out on NextSeq 500 sequencing system with NextSeq 500 / 550 Mid Output Kit v2.5 (300 Cycles) chemistry (Illumina, Inc. USA, Cat. No. 20024905). c) Bioinformatic analysis of the A. baumannii genomes a. Genomes from public databases

[0232] All publicly available 15,403 A. baumannii genomes were downloaded from the NCBI database as of 09 / 2022. b. Genome assembly

[0233] The raw reads were trimmed using the cutadapt 4.3 program (Martin, M. 2011) (min quality 33, max N 0.5, min read length 30). An optimal reference genome for the reference-guided de novo assembly was chosen from the 5,271 NCBI genomes that were available on 19 / 04 / 2021 . The reference genome was selected by comparing k-mer counts using the kmc 3.2.1 software (Kokot et al. 2017; Deorowicz et al. 2013. ). We determined the number of shared k-mers between the raw reads and the 5,271 available genomes was determined, and the one with the highest shared k-mer value was selected as a reference genome for the reference-guided de novo assembly.

[0234] The assemblies were performed with the SPAdes genome assembler 3.15.5 (Bankevich et al. 2012) selecting only those contigs with lengths of at least 1 ,000 and coverage of at least 10x. Assembly fasta files that were at least an order of magnitude larger or smaller than the median file size were dropped from further analysis. c. In silico molecular typing

[0235] Whole genome assemblies were subsequently used for molecular typing. Multi Locus Sequence Typing (MLST) was performed according to the Pasteur scheme (Diancourt et al. 2010), using the mist v2.23.0 software (Seemann et al. 2023) which incorporates components of the PubMLST database (Jolley et al. 2018). In addition, Kaptive v2.0.3 software (Lam et al. 2022) was used for identifying polysaccharide capsule and outer lipopolysaccharide loci and for classifying assemblies into CPS and O-types, using the k and OC locus primary reference databases within the software, respectively. d. Detection of antibiotic resistance determinants

[0236] Antibiotic resistance determinants in all A. baumannii whole-genome sequences were detected as follows. First, open reading frames (ORFs) were predicted using the Prodigal v2.6.3 software (Hyatt et al. 2010), and then resistance genes were identified based on sequence similarity to literature-curated resistance genes compiled in the ResFinder database v2.0.0 (Zankari et al. 2012). Specifically, protein sequence similarity searches were performed with DIAMOND v2.0.15 (Buchfink et al. 2021) using 10'5E-value threshold, 80% identity and coverage thresholds, and keeping the best hit for each ORF. Beta-lactamase genes were assigned to gene families using two relevant publications (Evans et al. 2014).

[0237] Resistance determinants that were outside the scope of ResFinder were examined separately. These included the presence of the insertion sequence ISAbal , which can be required by some genes to confer resistance against carbapenems (Li et al. 2019; Turton et al. 2006; Segal et al. 2007; Corvee et al. 2007). The absence of the carO gene which encodes a porin channel (Novovic et al. 2015; Li et al. 2019), and three point mutations indicative of fluoroquinolone resistance, gyrAS81 L, gyrBA414T and parCS84L. For ISAbal (Geisinger et al. 2018; Liu et al. (2022), a list of ISAbal sequences was compiled using the European Nucleotide Archive (Leinonen et al. 2011) and a number of publications (Leinonen et al. (2011); Hawkey et al. 2015). Then, this list was manually filtered selecting only the relevant sequences based on their descriptions. Most genomes contained both the insertion sequence and one or more genes or gene fragments, but LC136852 and LC136853 seemed to only contain ISAbal . LC136852 was used to trim the rest of the sequences and acquire variations for the sequence of ISAbal . Finally, a list of 53 non-redundant ISAbal sequences was used to search ISAbal in whole genomes using NCBI BLAST v.2.13.0 (Camacho et al. 2009). with a 10-5E-value threshold. For the rest of the resistance determinants, a list of sequences was assembled including carO (Novovic et al. 2015), gyrA, gyrB and parC and a search for these was made using NCBI BLAST v.2.13.0 with 10'5E-value threshold. For point mutations, a custom script written in R (R: The R Project for Statistical Computing https: / / www.r-project.org / .) and R packages Biostrings v2.66.0 (Pages et al. 2023), dplyr v1 .1 .3 (Wickham et al. 2023) and seqinr v4.2-23 (Charif et al. 2007) was then used for identifying genomes which carry the required amino acid substitutions. e. Identifying CRAB isolates

[0238] For the purposes of this study, an A. baumannii isolate is considered CRAB if its genome contains resistance determinants against carbapenems as well as against aminoglycosides and fluoroquinolones. The present goal with this strict definition was to identify high-risk isolates that may be resistant towards a wide range of antibiotics in addition to carbapenems.

[0239] Aminoglycoside resistance was determined directly from ResFinder results. If at least one aminoglycoside resistance gene was found, the isolate was considered resistant. While ResFinder contains beta-lactam resistance genes, it does not distinguish between carbapenems and non-carbapenem beta-lactams, therefore a custom approach was applied to predict carbapenem resistance. Specifically, all available A. baumannii antibiotic resistance test results were downloaded from the Bacterial and Viral Bioinformatics Resource Center (Olson et al. 2023) (accessed 2023- 02-16) and checked whether any of the beta-lactam resistance genes, ISAbal , or carO could be associated with observed carbapenem resistance. 1 ,283 observations were found where test results were available for either imipenem, meropenem, ertapenem or doripenem, and the results could be linked to genomic sequences in our data set. It was then tested for statistical association between the experimentally determined phenotypes and the resistance determinants using contingency tables and standard statistics. An earlier study suggested that beta-lactam resistance genes blaoxA-23-iike, blaoXA-24 / 40-like, bla0XA-58-like, bla0XA-143-like, bla0XA-235-like, any blaNDM, blaviM, blaiMP, blaKPC, blaGEs were all associated with carbapenem resistance (Li et al. 2019). The present analysis was consistent with these findings; therefore, the presence of these resistance genes formed the basis of classification. Furthermore, the analysis additionally suggested that blaoxA-312 may also be associated with carbapenem resistance (16 true positives and 0 false positives, Fisher test p = 0.0005016) therefore blaoxA-312 was also added to the list of resistance determinants. However, the presence of ISAbal alone or the absence of carO proved to be poor predictors of carbapenem resistance and were not added to the list. Finally, an isolate fluoroquinolone- was considered resistant if it carried at least one of the three point mutations defined above. d) Analysis of population structure and phylogeography a. Quality filtering of genomes

[0240] For all subsequent analysis, genome assemblies were only kept where a) assembly coverage was at least 25x, b) MLST could be predicted; c) CPS type could be predicted with at least "Good" confidence ( / .e., “The locus was found in a single piece or with > 95% coverage, with < 3 truncated / missing genes and < 1 extra gene compared to the reference”); genomes were eliminated where d) GC content was extremely high or extremely low; e) number of contigs was extremely high; f) length of the longest contig, g) N50 (length of the shortest contig which, together with all longer contigs, represent 50% of the nucleotides), or h) N95 (same for 95% of the nucleotides) was extremely low, i) number of ambiguous nucleotides was extremely high, j) BUSCO v5.4.4 (Simao et al. 2015) complete score was lower than 95, k) taxonomic classification using sequence data with Kraken 2 v2.1.2 (Wood et al. 2019). suggested a species other than A. baumannii or I) the Kraken2 taxon frequency for the species was extremely low (based on Inter-Quartile Ranges, extreme outliers) (Beyer 1981). Genomes were only retained with metadata indicating m) the country of origin for the isolate and n) at least the collection year of the isolate. Biosample metadata from NCBI was extensively curated to o) eliminate samples that were not human-associated or p) to eliminate known duplicates. Following filtering, only genome assemblies of 11 ,129 A. baumannii isolates were analysed further from the 15,829 that were included in the study. b. Controlling for sampling bias by downsampling CRAB isolates

[0241] The set of 11 , 129 isolates with genome sequences is likely to be inherently biased due to overrepresentation or underrepresentation of certain regions and time periods compared to others. Such biases could reflect regional and temporal differences in reporting, rather than genuine epidemiological differences, and could therefore distort analyses of population structure (e.g., assessment of serotype diversity or regional dynamics of serotype frequencies). To address this, the 11 ,129 isolates were stratified by geographical region and time, a subsampling strategy suited for large-scale phylogeographic analyses. Specifically, we applied the following subsampling procedure: First, we only retained isolates that were classified as CRAB (8,915 isolates, 6,461 since 2016). Then, for isolates with known city information, a maximum of one sample per city per week was kept and isolates with only country information were limited to a maximum of two samples per country per week. This stratification resulted in 4,559 isolates (3,184 ST2 isolates), out of which 4,134 were collected after 2009, and 2,536 were collected since 2016. Furthermore, to mitigate potential biases introduced by uneven sampling intensity across countries, the data was further downsampled after stratification. Each country was limited to a maximum of one sample per million inhabitants. This further downsampling resulted in 2,577 isolates, out of which 2,304 were collected between 2009 and 2020, 969 between 2009 and 2015, and 1 ,455 since 2016. c. Defining prevalent and global MLST-CPS types

[0242] An MLST-CPS serotype was defined as prevalent if it accounted for at least 5% of the isolates in at least one of the 23 world regions. The 23 world regions were defined according to the World Bank Development Indicators database (legacy), accessed through the R package countrycode v1.3.0 (Arel-Bundock et al. 2023). Furthermore, a serotype was classified as global if (i) it was prevalent according to the above definition and (ii) it accounted for at least 2% of the isolates in at least 3 world regions spanning at least 2 continents. Prevalent or global serotypes were identified using the 5,225 downsampled CRAB isolates and were identified separately for isolates collected between 2016-2022 (current period) and for isolates collected between 2009-2015 (preceding period) to address temporal dynamics. d. Temporal dynamics of the relative prevalence of serotypes

[0243] The temporal dynamics of all MLST-CPS types that had been classified as prevalent was investigated (see above). For this analysis, the dataset with 4,860 downsampled CRAB isolates since 2009 was used. For each MLST-CPS type, (i) the year was identified when the first isolate was documented since 2009, (ii) the dataset was filtered to include observations since that date, (iii) each isolate was classified as belonging to the particular MLST-CPS type or not, and (iv) logistic regression analysis was applied to study trends in relative prevalence over time. Temporal trends were studied both globally and at the level of individual continents. To account for potential biases caused by geographical differences within the studied regions, the continent was included as a covariate when inferring temporal trends on a global scale. In a similar vein, world regions were included as a covariate when inferring temporal trends for continents. Note that it was only focused on MLST-CPS types with at least 50 isolates available in a given region, e.g. a particular MLST-CPS type on a particular continent was only considered if at least 50 isolates of that MLST-CPS type were collected from that continent since 2009. FDR correction was applied (through the R package stats (R: The R Project for Statistical Computing https: / / www.r-project.org / .) to account for multiple comparisons. Finally, the 15 statistically significant MLST-CPS trends calculated based on individual continents was plotted. e. Selecting regions with sufficient sample sizes For analyses that require sufficient representation of local MLST-CPS diversity, rarefaction analyses were carried out. Specifically, world regions or countries were manually included based on rarefaction analysis indicating that sufficient numbers of isolates are available to represent the local diversity of MLST-CPS types. For the rarefaction analysis the 3,275 downsampled CRAB isolates since 2016 were used and the number of isolates belonging to each MLST-CPS type in each world region or country were counted, and the rarefaction curves from these count matrices were calculated using the R package vegan v2.6-4 (Oksanen et al. 2022). After plotting the rarefaction curves, 9 out of 16 regions were manually selected to study the global distribution of CRAB strains. Similarly, 7 European and 7 non-European countries were selected for comparisons between countries. The European countries included France, Germany, Greece, Hungary, Italy, Romania, and Serbia, while the non-European countries included Brazil, China, Israel, Saudi Arabia, South Africa, Thailand, USA. f. Building time-calibrated phylogenetic trees

[0244] The 11 ,129 filtered genomes were used to build 4 separate time-calibrated phylogenetic trees for 4 sequence types: ST1 , ST2, ST492, ST636. These STs cover the most prevalent serotypes worldwide and the most prevalent serotypes in the present focal countries. For each sequence type, first, a reference genome was selected. For ST2, the reference genome was selected manually (GCF_003288775.1 , a Complete Genome from the NCBI RefSeq database, sequenced using PacBio, 120x coverage, assembled into two contigs, one of which is labelled as chromosome), for the rest of the sequence types the genome with the longest contig was selected. In the case of ties, the earliest genome was selected. Then, each genome belonging to a particular sequence type was mapped to its reference genome using snippy v4.6.0 (Seemann 2023) to produce pseudo-whole genomes. For isolates that were sequenced in the present study, the reads were mapped while for the rest of the isolates the assembled genomes were mapped. The longest contigs were kept, any duplicates (either genuinely duplicate sequences or sequences that only became duplicates after pseudo-whole genome reconstruction) were removed, and the phylogenetic tree was constructed using gubbins v3.3.0, (Croucher et al. 2015). (model fitter: raxmlng, tree builder: fasttree, maximum number of iterations: 10) to account for recombinations. Abnormally long branches were subsequently removed with TreeShrink v1.3.9 (Mai et al. 2018).. The resulting tree was dated in two steps: the tree was first rooted using root-to-tip regression with the lowest sum of the squared residuals and then the rooted tree was dated with treedater (Volz et al. 2017) without rerooting, using a strict molecular clock. After dating, tips that had been removed due to sequence duplication were added back to each tree with 0 branch lengths. g. ST2 global and regional transmission dynamics

[0245] To study the phylogeographic patterns of the ST2 sequence type it was built on a previously published method (Moura et al. 2021) and analysed genetic similarity between pairs of isolates across different temporal and spatial scales. The general idea behind the analysis is that pairs of isolates may be geographically close or distant and also genetically close or distant, and the distribution of isolate pairs across geographic- genetic distance categories characterises the transmission dynamics of the bacterium. For example, pairs that are both genetically and geographically close indicate local spread, while pairs that are genetically close but geographically distant indicate cross- border transmissions. Therefore, the number of pairs in these categories yields insights into the relative importance of local versus cross-border transmissions.

[0246] For the analysis, the time-calibrated ST2 tree reconstructed above was used. However, CRAB isolates were only included, and it was focused on the downsampled dataset to address sampling bias. This procedure resulted in a smaller ST2 tree with 3,537 tips. In the case of global transmission dynamics, pairs of isolates on the tree were considered that were collected at most 2 years apart and these pairs were categorised by geographic location (same country, different country, etc.) and genetic distance (based on the most recent common ancestor, MRCA). Then the number of pairs was counted in each combination of spatial and temporal categories and calculated relative risks separately for each temporal category. Specifically, the relative likelihood of occurrence was calculated in a given spatial category as compared with the likelihood of occurrence in the control spatial category. For example, it was found that when focusing on pairs with MRCA below 6 years, the relative risk is approximately 10 for pairs that originate from the same country as compared to those that originate from different countries on the same continent (reference category).

[0247] To account for uncertainties associated with this analysis (e.g. sampling of isolates and tree building), multiple subsamples were taken from the set of 3,537 isolates constituting the ST2 phylogenetic tree above (100 subsamples with 500 isolates each). For each subsample, a separate phylogenetic tree was reconstructed and analysed as described above. To reconstruct such a subsampled tree, the respective recombination- free polymorphic sites were selected from the Gubbins output, the subsampled sites filtered using SNP-sites v2.5.1 (Page et al. 2016) a phylogenetic tree using FastTree v2.1 .11 SSE3 (Price et al. 2009) was built and rooting and dating was performed the same way as for the ST 1 , ST2, ST492 and ST636 trees.

[0248] Then, relative risks were calculated for each tree and finally, using the 100 subsamples the mean values and 90% confidence intervals were calculated for these risks. Using 90% instead of 95% reflects that the relationships that is sought to infer require one-sided tests.

[0249] A similar procedure was followed for the phylogeographic analysis of CRAB in Europe, and also for regional transmission dynamics. For the regional analysis, pairs that were collected at most 0.5 years apart and drew 500 subsamples with 100 isolates each were included. h. Capsule visualisation of A. baumannii ST2 KL3 isolates

[0250] To visualise the capsule of different A. baumannii clinical isolates and induced phage-resistant isolates, Transmission Electron Microscopy (TEM) imaging was performed by following a previously described protocol. (Reynolds 1963). Briefly, 0.5 ml of overnight culture from each bacterial isolate (n = 4) was centrifuged for 2 min at 13000 rpm in a benchtop centrifuge and then fixed overnight at 4 °C with Karnovsky fixative solution (Karnovsky 1965) (pH 7.4). After fixation samples were briefly rinsed in distilled water (pH 7.4) for 15 min and fixed furthermore in 1 % osmium tetroxide in ddH2O (Sigma-Aldrich, St. Louis, MO, USA) for 1 h. After fixation, samples were briefly rinsed in distilled water for 10 min, then dehydrated gradually in a series of ethanol. Afterwards, bacteria were polymerized through propylene oxide (Molar Chemicals) and then embedded in an epoxy-based resin (Durcupan ACM; Sigma-Aldrich) for 48 h at 56°C. From the resin blocks 50 nm thick ultrathin sections were cut using an Ultracut UCT ultramicrotome (Leica; Wetzlar, Germany) and were mounted on a single-hole, formvar- coated copper grid (Electron Microscopy Sciences; Hatfield, PA, USA). The contrast of the samples was enhanced by staining with 2% uranyl acetate in 50% ethanol (Molar Chemicals, Electron Microscopy Sciences) and 2% lead citrate in distilled water (Electron Microscopy Sciences). i. Isolation of A. baumannii bacteriophages

[0251] The phage hunt was performed on A. baumannii wild-type isolates belonging to different MLST-CPS types and in addition to these, on phage-resistant variants of the aforementioned isolates. In all cases phages were isolated from raw sewage water harvested from waste-water treatment plants from five Hungarian cities (Szeged, Budapest, Hodmezdvasarhely, Debrecen, Bekescsaba), using the enrichment procedure described previously (Popova et al. 2021) with some modifications. In brief, the sewage water was cleared by centrifugation (4500 rpm for 10 min), followed by filtration through a 0.45-pm- membrane filter and aliquots of wastewater were mixed with an equal volume of 2* LB medium and 0.1 % of overnight bacterial culture. Samples were incubated overnight at 37°C with shaking at 140 rpm followed by centrifugation for 10 min at 4500 rpm, and then the supernatant was filtered through a 0.45 pm membrane filter twice to remove the residual bacteria and debris. Plaque assay was performed to screen for the presence of lytic phage activity using the double-layer agar method. (Anderson et al. 2011). The plates were incubated overnight at 37°C and examined for zones of lysis or plaque formation. Single plaques formed on the lawns of A. baumannii strains were picked up and this process was repeated three times, in order to obtain pure phage stock. j. Phage propagation and purification

[0252] Phage propagation was carried out using a liquid culture of the corresponding A. baumannii clinical isolate (ODeoo=0.6) at a multiplicity of infection (MOI) of 0.1. 25 ml of phage lysate was concentrated by the addition of 10% polyethylene glycol (PEG) M.W. 8000 and 1 M NaCI and incubated overnight at 4 °C. After centrifugation at 4600 rpm for 20 min, the pellet was suspended in 1 ml SM buffer (50 mM Tris-HCI [pH 7.7], 100 mM NaCI, 80 mM MgSO4). The phage titer was determined by the double-layer agar method and the titer was reported as a plaque-forming unit (PFU / ml).

[0253] For in-vivo experiments, phage stocks were prepared using the Phage-on-Tap protocol (Bonilla et al. 2016) with some modifications. The corresponding A. baumannii culture was grown in 100 ml LB at 37°C to an ODeoo of 0.6-0.7 and infected with the phage. The phage lysate was concentrated (10X) using centrifugal filters (Amicon Ultra- 15, Sigma 100 MDa cut off) and the concentrate was washed with 1X PBS at least 5 times. The phage concentrates were mixed with 0.7 volume of 1 -octanol for additional endotoxin removal, (Hietala et al. 2019) shaken for one hour at room temperature following an incubation at 4°C for 2 hours. The phage-containing phase (bottom phase) was collected with a syringe after centrifugation at 4,000 * g and 4°C for 10 min. The removal of octanol was enhanced by repeating the centrifugation step two additional times. This was followed by a few extra washing steps with PBS using Amicon filter (100 MWCO) devices. To ensure that the endotoxin concentration is low, an additional step was included using Pierce high-capacity endotoxin removal spin 1 ml column (Thermo Scientific) following the manufacturer’s instructions. e) Laboratory characterisation of the isolated phages a. Phage morphology

[0254] In order to obtain transmission electron microscopy images, six pl of the phage lysate was mounted on a TEM copper grid (300-mesh, Electron Microscopy Sciences) with carbon-coated ultrathin formvar film. (Reynolds et al. 1963). The samples were dried using filter paper after 1 minute. To obtain the negatively stained samples, the samples were contrasted with 6 pl uranyl acetate (2% w / v) in 50% ethanol for 2 min (this process was repeated 3 times). After the removal of the excessive staining solution, samples were dried under a Petri dish for 2 h before the electron microscopic evaluation. Negatively stained samples were systematically screened and the grids were examined under a transmission electron microscope (JEM-1400 Flash transmission electron microscope) at 80kV with a 16 MP Matataki Flash scientific complementary metal- oxide-semiconductor (sCMOS) camera (JEOL). b. Phage adsorption assay

[0255] 10 ml bacterial culture of wild-type isolates or induced phage-resistant isolates was pre-grown until exponential phase (OD6oo=0.6-0.7) and mixed with the phage at MOI 0,1. The cells were incubated at 37°C with aeration and at each time point 100 pL samples were taken every 5 or 10 min for 30 min. Samples were mixed with 850 pl of LB supplemented with 50 pl of chloroform, vortexed, and then centrifuged for 1 min at 13000 rpm in a benchtop centrifuge. The supernatant was diluted in LB to determine the unadsorbed phages by the plaque assay using the double-layer method. c. Determination of phage host specificity

[0256] The lytic activity of each phage was screened against multiple A. baumannii clinical isolates with the standard spot assay as previously described. (Kutter, E. (2009).

[0257] For Silvergun phage, the efficiency of infection was compared on different ST2- KL3 and ST2-KL2 isolates respectively, by determining the efficiency of plating (EOP) (for details see AciPhage report). Briefly, serial dilution of phage lysate was plated on soft agar overlays of different bacterial isolates. After incubating the plates overnight at 37°C, the individual plaques were counted and the average of 3 technical replicates was calculated and compared to the number of plaques on the original host of isolation. EOP was calculated as follows: average PFU on target bacteria I average PFU on host bacteria. f) Bioinformatic analysis of the isolated phages a. Whole-genome sequencing

[0258] Phage genomic DNA was extracted using a commercial phage DNA isolation kit (Norgen Biotek Corp.; 46850) following the manufacturer’s instructions with the sole exception of applying sterile, distilled H2O during the elution step instead of the provided elution buffer. Afterwards, samples were sequenced by Illumina sequencing. 1 ng of DNA was used as the input amount recommended in the Nextera XT Sample Preparation Guide. The rest of the protocol as written was followed. Sequencing-ready libraries were quality control checked by BioAnalyzer2100 instrument using High Sensitivity DNA Chip (Agilent Technologies USA, Cat. No. 5067-4626). NGS was carried out on the NextSeq 500 sequencing system with NextSeq 500 / 550 Mid Output Kit v2.5 (300 Cycles) chemistry (Illumina, Inc. USA, Cat. No. 20024905) Applied read length was in the 150-300 bp range. Processing of the sequencing data was started with the quality control and trimming of reads using Trimmomatic v0.36 (Bolger et al. 2014) with the following command:

[0259] TrimmomaticPE -threads X [reads_Fw].fastq.gz [reads_R].fastq.gz [Fw]_paired_output.fastq.gz [Rev]_unpaired_output.fastq.gz [Fw].fastq.gz [R]_unpaired_output.fastq.gz ILLUMINACLIP:TruSeq3-PE-2.fa:2:30:10 LEADINGS TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:32. Assembly of the thus obtained, filtered reads into contigs was carried out by either MEGAHIT (Li et al. 2015) or MetaSPAdes v.3.13.0 (Nurk et al. 2017) or meg on a per-sample basis. Contigs of 10 kb < length were then investigated to identify phages in the samples by BLAST+ v2.12.0 blastn alignment to NCBI RefSeq database, (Camacho, C., et al. 2009) using the following command: blastn -query [query], fasta -db [database] -outfmt 5 - max target seqs 1 -qcov hsp perc 5 -out[output].xml -num threads X. b. Bioinformatic analysis

[0260] Bioinformatic prediction of ORFs and annotation of the phage contigs was done by a simple customary python script. In brief, the code translated the whole input nucleotide sequence into amino acid sequence in all six reading frames. Then, regions with a length longer than 50 from start to stop codon, were collected in a list along with the positional and start-end values and the assigned identifier of the candidate gene. The collected amino acid sequences were matched to the NCBI nr database then the name and taxonomic information of the best hits were added to the identifier of the candidate subject (if any). The above data was then collectively processed to generate GenBank files for the individual phage contigs. Further, genes of interest i.e. genes related to therapeutically important functions were listed in separate files. Genes of interest were extracted from the prediction data with regular expressions together with the positional information. The therapeutic suitability and the possible life cycle of isolated phages were predicted with PhageAl software using default settings (Tynecki et al. 2020) and the Phageleads web service. (Yukgehnaish et al. 2022). Endonucleases were identified by blastp alignment. The presence of antibiotic or virulence-conferring genes were checked by utilising the RESFINDER v3.0 and VIRULENCEFINDER v1.5 database, respectively, corroborated with PhageLeads results. Insertion sequences were screened using ISfinder. (Siguier et al. 2006). Depolymerase prediction was done by Depolymerase Predictor, a tail fiber or spike protein was deemed as carrier of a depolymerase region if the value of probability was 0,85<. (Magill et al. 2023). The completeness of assembled phage genomes was elucidated via CheckV software (Nayfach et al. 2021). c. Determination of phage host specificity

[0261] The lytic activity of each phage was screened against multiple A. baumannii clinical isolates with the standard spot assay as previously described. (Kutter, E. (2009).

[0262] For Silvergun phage, the efficiency of infection was compared on different ST2- KL3 and ST2-KL2 isolates respectively, by determining the efficiency of plating (EOP). Briefly, serial dilution of phage lysate was plated on soft agar overlays of different bacterial isolates. After incubating the plates overnight at 37°C, the individual plaques were counted and the average of 3 technical replicates was calculated and compared to the number of plaques on the original host of isolation. EOP was calculated as follows: average PFU on target bacteria I average PFU on host bacteria. d. Phage pangenomic network construction

[0263] To visualize the location of the phages described herein in the broad landscape of viruses the classical approach of constructing an orthogroup based network of viruses was used, where proteins of all participant phages are clustered, and the phages’ level of connection is determined by the amount of proteins they have in common viral clusters (VCs).The tool that was found to suit this task the most vas vConTACT2 (Bin Jang et al. 2019), which implements the above methods and outputs results largely adhering to the classifications of the ICTV database. To acquire the data for the network, all phage genomes assigned to the Caudovirecetes order from the International Committee on Taxonomy of Viruses (ICTV) database were retrieved (Lefkowitz et al. 2018) (version August 22, 2022).

[0264] Literature mining to identify Acinetobacter phages that have been used as therapeutic agents was also conducted. In addition, Acinetobacter phages from the collection described herein was included, which were obtained through an extensive phage hunt program described herein.

[0265] Taxonomy information for the accession IDs originating from the ICTV database was obtained, while for the phages described herein, a blast search was conducted to identify their tail fibers and spikes. Taxonomy information was obtained based on the closest relative’s taxonomic position.

[0266] Prodigal (Hyatt et al. 2010) was used to predict ORFs for all phages, and these ORFs were used as input for vConTACT2's gene2genome module to generate a network file in Cytoscape format. A custom R script was then used to merge additional metadata information for the nodes visible in the network.

[0267] The final network file was then created with vConTACT2’s default settings, no database, the ‘BLASTP’ method, and c1 clustering. To ensure reproducibility and scalability, the entire process was implemented in a Nextflow pipeline. The network subplots were created in Cytoscape. e. Average Nucleotide Identity (ANI) of phages

[0268] The genomes of the phages described herein were compared with those considered therapeutically important or known to have a lytic life cycle. The alignments resulting from this comparison were handled as follows: for a specific section of the genome, if multiple alignments were possible, the one with a higher identity was chosen. In cases where an alignment with lower identity covered a longer section, it was used for the entire section. In situations where one genome was longer than the other, the Average Nucleotide Identity (ANI) value was adjusted by subtracting the length difference. For instance, if one genome was 90% the length of the other, even if the genomes exhibited 100% identity, the maximum ANI value achievable was 90%. f. Phage pangenomic visualisations

[0269] The genomic synteny of phages was colinearized according to a reference point (an official lytic or therapeutic phage in each case) and then the GenBank files were supplied to Clinker. (Gilchrist et al. 2021 ). The pangenomic comparisons of phages were done with a custom R script that compared each of our phage genome’s genes against the corresponding one from the reference genes using blastn. Then from all compared tail coding sequences, depolymerase regions were predicted using the Phyre2 web platform, (Kelley et al. 2015) where the best hit were selected among those that had above 0.4 confidence and contained the word ‘hydrolase’ and ‘lyase’, corresponding mostly to the methods described earlier. (Latka et al. 2019). g. Isolation of in vitro evolved phage-resistant A. baumannii mutants Phage-resistant variants were evolved in a 96-well microtiter plate in 200 pl final volume in LB medium. The ODeoo values of the wells were measured every 20 min for 24 hrs in an EPOCH 2 (Biontech) instrument, at 37°C with 10 second shaking before each measurement. Each well, except the control wells, contained 100 pl from the bacterial culture with ODeoo = 0.7 and the phage(s) in 106PFU / well concentration. Resistance was evolved to a single phage or a combination of 2, 3, or 4 different phages. If phage combinations were used, each phage was added in 106PFU / well concentration (in 1 :1 ratio). Following a 24-hour incubation period, the wells that exhibited cell growth as indicated by ODeoo values, samples were streaked onto plates containing chromogenic agar and incubated overnight at 37°C for each strain. The resulting individual colonies were restreaked at least 3 times to avoid phage contamination. The phage-resistant phenotypes were confirmed using phage susceptibility tests including the double-layer agar plate method or in vitro growth curves. The phage-resistant genotype was determined by detecting the mutations in the assembled genomes using a custom pipeline. For mutations occurring in the K-locus, the genes of the K antigen biosynthetic pathway were identified using blastn 2.13.0+. (Altschul 1997). The blast hit was aligned to the reference gene with mafft v7.520 (Katoh et al. 2013) and with a custom script identified the nucleotide and the amino acid variants. h. Antimicrobial susceptibility testing of phage-resistant isolates

[0270] Minimum inhibitory concentrations (MICs) of five clinically relevant antibiotics: Colistin (Pharma), Meropenem (Bioscience), Imipenem (MedChemExpress), Levofloxacin (MedChemExpress), Trimetoprim: Sulphametoxazole (1 :5 ratio) (Sigma, MedChemExpress) were determined using the standard microbroth dilution protocol (Wiegand et al. 2008) and interpreted using the European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines. (The European Committee on Antimicrobial Susceptibility Testing. Breakpoint tables for interpretation of MICs and zone diameters. Version 13.1 , 2023. http: / / www.eucast.org.). Cells were grown in Mueller Hinton Broth 2 (MHB2) (Millipore) and the inoculum size was set to 5 x 105bacteria per ml. Among the tested antibiotics were two carbapenems (Meropenem and Imipenem) and the last- resort antibiotic Colistin. In order to maximise reproducibility and accuracy, a robotic liquid-handling system (Hamilton) was used to automatically prepare 7-step, two-fold serial dilutions in 384-well microtiter plates. After 18 h of incubation at 37°C, raw ODeoo nm values were measured in a Biotek Synergy microplate reader. MIC was defined by a cutoff ODeoo nm value (mean +2 s.d. of A600 nm values of bacteria-free wells containing only growth medium). MIC fold change was calculated as MICwiid-type strain I MICphage-resistant Strain. i. Efficacy of phages in Galleria mellonella larvae model

[0271] Galleria mellonella larvae was used as one animal model to evaluate the effectiveness of Highwayman and Silvergun phages alone and in combination with other phages against A. baumannii clinical isolates. For each experimental condition, 10 larvae were used and experiments were repeated independently at least 3 times. Only larvae exhibiting a uniform cream color were used for the experiments. Survival of the larvae was monitored for 48 hr, every 6 hr.

[0272] As a first step, the proper inoculum size to be used was determined in further experiments. 6 different A. baumannii isolates and 5 different inoculum sizes for each isolate (CFU = 9x106, 9x107, 9x108, 9x109, 9x1010per ml) were tested. Based on these preliminary results, the inoculum size of 9x108CFU / ml was chosen as the mortality rate of the larvae was 50% at 18 hr post-infection for half of the tested isolates at this inoculum size.

[0273] Prior to the infection of the larvae, the bacterial strain was grown overnight in MHB2 medium at 37°C with aeration. To remove the growth media, the bacterial culture was washed twice with 1X Phosphate-Buffered Saline (PBS) and the density was set to 9x108CFU / ml. From this, 10 pl was injected into the first proleg of the larva. Phage treatment was administered in the same volume in PBS, in the opposite proleg, 5 minutes after the bacterial infection. This time point was chosen based on the results of preliminary experiments where we compared the survival rate of the larvae that received the same phage treatment 5 minutes and 60 minutes post-infection and found that the survival rate of the larvae receiving the treatment 5 minutes post-infection was significantly higher. j. Isolation of in vivo evolved phage-resistant A. baumannii mutants Phage-resistant A. baumannii mutants were isolated from larvae 48 hours postinfection, which marked the endpoint of the survival experiment, or earlier if larvae died before this time point. On the larva's ventral side, an incision was made with a sterile blade, and then we streaked from the body lumen onto Acinetobacter-specific chrome agar (CHROMagarTM Acinetobacter, CheBio). The resulting colonies were re-streaked at least 3 times to obtain a pure bacterial culture. The resistance phenotype was confirmed by phage typing against the phage(s) used in the treatment of the larvae and in addition, the susceptibility of the isolate was checked against the other cocktail components, PhT2-v1 and Fanak as well. If an altered phage infection pattern was detected compared to the wild-type, the isolate was sent for sequencing. k. Efficacy of phages in intraperitoneal mouse model

[0274] A mouse intraperitoneal model was adapted to study the infectivity of Aci 15 A. baumannii isolate. Female BALB / c mice, 6-7 weeks old and weighing 17-18 g, were obtained from Envigo in the Netherlands. All animal care and handling procedures adhered to the European Federation for Laboratory Animal Science Association (FELASA) guidelines, and the Enivolnvest Co., Hungary's Animal Welfare committee approved the protocols (permit number: BA02 / 2000- 12 / 2022).

[0275] Prior to the experiment, one colony of the Aci 15 isolate from a one-day-old LB agar plate was transferred into 5 mL LB liquid medium and was cultivated for 14 hours in a shaker thermostat (120 rpm, 37 °C). On the next day, the liquid culture was centrifuged (2 minutes, 10.000 g) and washed once with PBS. The gained bacterium suspension was 10x serial diluted (C, -1 , -2, -3, -4), and from the above dilutions, 200 pl was administered intraperitoneally to each mouse. Mice were separated into groups (each containing 5-5 animals) based on the infection doses. Members of the groups received about 1 x 109, aboutl x 108; aboutl x 107; aboutl x 106; aboutl x 105and aboutl x 104bacterial cells, respectively. Weight and death rates of the infected and control animals were recorded in the subsequent 7 days. For phage rescue experiments those inoculation doses were used that were at least around LD50.

[0276] Therapeutic efficacies of single and combined purified bacteriophages (phage Highwayman alone and the HSFPh phage cocktail) against Aci 15 isolate were tested as described before, (Schneider et al. 2018) with slight modifications. For this purpose, mice (40 animals) were divided into 7 groups. Members of the first three groups (GIGS) were infected with 109CFU / mice. Members of the groups G4-G6 were infected with 108CFU / mice. In addition, mice in group G7 received only from the phage suspension mix (IP) and served as phage controls. Mice in G1 and G4 served as bacterium controls and no phage administrations were performed in these groups. In contrast, members of G2 and G5 received HWM phages in MOI 6; while members of G3 and G6 received the 4-component phage cocktail, where the ratio of the bacterial and HWM concentration was MOI 6, while the other 3 phage components were MO1 1 . Phages in the case of G2, G3, G5 and G6 were administered 10 minutes after the bacteria injections. Death rates and body weights of all animals were daily recorded for seven days after the treatments. l. Comparing the cell surface properties of wild-type and phageresistant A. baumannii isolates with Fourier-transformed infrared spectroscopy

[0277] The protocol described before (Teng et al. 2022) was adapted with minor changes. Briefly, according to the manufacturer’s instructions bacterial biomass was collected from LB agar plates. The isolates were incubated for 24 h at 37°C. After incubation, bacterial suspensions were prepared by adding a full inoculation loop (1 pl) of bacteria in 1 .5 ml Eppendorf tubes containing 50 pl of 70% ethanol and 2 mm metal beads provided by the supplier. The bacterial suspension was thoroughly vortexed after which 50pl of sterile H2O was added. Then, 15pl of bacterial suspension was pipetted on a silicon plate (provided by Burker IR Biotyper equipment) in three technical replicates per isolate. We prepared two standard suspensions (provided by the supplier) and 10pl were added as controls in two technical replicates.

[0278] The plate was then dried for 20 minutes in the laminar hood at room temperature. Results were evaluated using Opus Software V7.5.18 and IR Biotyper Software V2.1 .0.195 with the default settings (32 scans per technical replicate; spectral resolution, 6 cm-1; apodization function, Blackman-Harris 3-term; zero-filling factor 4).

[0279] Following the manufacturer’s instructions, the pre-processing steps of the acquired data were as follows: the FTIR spectra measurements of individual technical replicates that did not meet the default quality criteria (0.4 < absorption < 2, signal-to- noise-ratio < 40, technical replicate of an isolate was not clustering with the other replicates of that isolate) were manually removed from further analysis to prevent inclusion of wrongful data. The distance matrix provided by the IR Biotyper software was used for further analyses. m. Screening for lysogen activity of the isolated phages with therapeutic potential

[0280] As the absence of lysogenic activity is a requirement for phages to be used in the traditional phage therapy, (Venturini et al. 2022) some of the isolated phages were tested: Silvergun, Highwayman and PhT2-v2 with therapeutic potential for lysogenic activity by adapting the protocol described before (Altamirano et al. 2021). Briefly, after performing a spot assay with the phage in question (PFU scaling from 105to 108) bacterial cells were isolated from mesas, zones of confluent bacterial or phage-resistant cells growth in the centre of the lysis spots. After applying the phage in two different concentrations, twenty colonies were isolated and tested in patch plate screens to see if the phages were capable to lysogenize the cells. This step was repeated twice to eliminate the chance of carry-over of phage particles between the steps. To ensure the absence of the lysogen activity of the phages tested, the spontaneous phage release test and the immunity assay were performed as well. Plates were incubated overnight and phage infection was identified by lysis zones. n. Validation of the wide host range of the Rocket phage

[0281] To validate the wide host range of the Rocket phage, plaque PCR was performed to detect the presence of the phage from single plaques formed by the phage on bacterial cultures of different A. baumannii isolates belonging to different MLST-CPS strains. As a negative control cells from the same plate from the vicinity of the plaques were used.

[0282] Primers used for the plaque PCR

[0283] Example 2: Region-specific phage composition, the global distribution of CRAB types

[0284] The first objective was to assess the prevalence and geographical distribution of the CRAB strains at both regional and global scales. To accomplish this, the analysis was initiated by retrieving all publicly available A. baumannii genomes from the NCBI database (15,410 genomes as of 09 / 2022), originating from 85 countries across five continents. Given the limited representation of genomic samples from Eastern and Southern European regions, 419 A. baumannii clinical isolates collected between 2011 and 2022 from 40 healthcare facilities located in 34 cities across five neighbouring countries in these regions were additionally sequenced (Hungary (n=253), Romania (n=120), Serbia (n=28), Montenegro (n=9), and Bosnia and Herzegovina (n=9)). Subsequently, the dataset was refined, narrowing down the initial 15,829 genomes to a final set of 11 ,129 genomes. The selection criteria had considerations such as genome quality, human origin, and specified information regarding the place and time of isolation. Finally, these genomes were screened for genetic determinants that collectively characterise CRAB traits, including resistance genes to carbapenems and other commonly used antibiotic classes (Methods). The analysis revealed that 80.1 % of the 11 ,129 genomes can be considered CRAB according to these criteria. As expected, there was a noticeable global increase in the relative proportion of CRAB genomes observed between the years 2000 and 2022.

[0285] As a next step, the genomes were grouped into CRAB types by employing a dual genetic typing approach to identify one or more types of interest (TOI), capturing complementary aspects of phage susceptibility. This was achieved by combining multilocus sequence typing (MLST) and cell surface capsular polysaccharide (CPS) typing. MLST clusters evolutionary-related genomes into sequence types (ST) based on the sequence of multiple household genes. In contrast, CPS typing identifies specific genomic determinants responsible for synthesising structurally different CPSs, which can function as receptors for different Acinetobacter phages.34 It is important to note that the genetic diversity of CPS is shaped by frequent horizontal gene transfers. Therefore, this trait evolves independently from ST. Finally, the distribution of currently (2016 - 2022) circulating CRAB types as defined by the combined MLST-CPS typing of the genomes in nine world regions was analysed (West-, East- and South-Europe; North- and South-America; West-, East- and South-East-Asia) where the sampling was sufficient to capture the majority of CRAB diversity after correcting for sampling bias (Methods).

[0286] In the nine world studied regions, the contemporary global diversity of CRAB is characterised by a small number of prevalent MLST-CPS types and a large number of rare ones. Specifically, in a given world region, typically 4 or 5 MLST-CPS types collectively account for 50% of the isolates, while 17 MLST-CPS types make up 90% (Figure 1). Altogether, 31 “prevalent” MLST-CPS types that have a relative prevalence in the CRAB population of at least 5% in at least one world region were identified. Nevertheless, there is substantial variation in the composition of CRAB types across different world regions. Specifically, among these 31 prevalent MLST-CPS types, only 7 exhibit a relative prevalence exceeding 2% in three or more world regions, which span across at least two continents. This makes these 7 types not only prevalent but also globally dispersed. Finally, in large contrast to the populations of the prevalent types, the remaining 10% of the global isolates can be attributed to 95 MLST-CPS types.

[0287] Then, it was investigated whether countries in close geographical proximity tend to share a more similar composition of CRAB populations. This analysis focused on seven European, closely situated and seven non-European, geographically distant countries with sufficient genomic sampling (Example 1). The analysis reveals that the MLST-CPS type composition of CRAB populations between pairs of European countries tends to exhibit more similarity compared to pairs of countries located on different continents. This higher similarity is primarily attributed to the presence of 4 prevalent types shared by at least three of the seven European countries, collectively representing 39% of the isolates. Additionally, the set of 25 country-specific MLST-CPS types constitutes only 22% of the isolates in these countries.

[0288] To summarise, the diversity within the CRAB population in each world region is characterised by a few prevalent CRAB types that are responsible for the majority of the infections and a vast number of rare strains. Additionally, while there are substantial differences in CRAB populations across world regions, countries that are geographically close to one another tend to share a more similar CRAB type composition.

[0289] Example 3 - Local CRAB variants drive infections within a 6-year period

[0290] As a next step, the temporal dynamics of CRAB populations were investigated. Initially, temporal changes in the global frequencies of the 29 MLST-CPS types that are currently (2016 - 2022) prevalent were analysed. Statistical modelling revealed that 11 out of 29 prevalent strains showed a significant expansion over the past 12 years on at least one continent (Methods). The largest expansion was exhibited by the ST2-KL3 type (), which gradually increased from 2009 onwards to become one of the three most prevalent CRAB type in seven out of the nine studied world regions. This expansion across multiple geographical regions implies a potential fitness advantage of ST2-KL3 over other types.

[0291] Despite the temporal expansions of specific prevalent MLST-CPS types, it was also observed that about76% of the prevalent types had already been present in the preceding period (2009 - 2015) with a prevalence exceeding 2% in the majority of world regions where they are currently prevalent ( / .e. >5%). This observation suggests that CRAB population compositions exhibit local stability over this timeframe. To investigate the temporal stability in more detail, the MLST-CPS type compositions of specific countries overtime were compared. The comparison focused on three countries - China, Germany, and the USA - each of which possessed a sufficient number of sequenced isolates from two distinct periods, namely 2009 - 2015 and 2016 - 2022. In each of these countries, it was observed that the previous CRAB-type composition was more similar to its contemporary counterpart than to the contemporary composition of most other countries worldwide, which confirms the temporal stability of CRAB populations in these countries.

[0292] To gain deeper insights into the spatio-temporal dynamics of CRAB types, a phylogeographical analysis of the dominant ST2 clade using established methods was carried.35 The analysis covered about70% of the available 11 ,129 A. baumannii genomes, providing insights into this pathogen’s transmission dynamics at an unprecedented scale. Specifically, a time-scaled phylogenetic tree using the core genome sequences of 7,720 ST2 isolates collected from across the globe was reconstructed (Methods). Encouragingly, a strong temporal signal within the dataset was found, affirming the accurate capture of chronological information within the reconstructed time-scaled tree. This was further supported by the estimated root date of 1973 for the tree, which aligns with epidemiological data regarding the emergence of ST2.

[0293] Then, the time-scaled tree to assess the timeframe needed for CRAB lineages to spread across different geographical distances was used. A strong phylogeographic clustering was used, with genetically closely related isolates primarily circulating within the same country over a short period. Specifically, on average, two contemporary CRAB isolates that diverged very recently ( / .e. within the past 6 years) exhibited a 19-fold higher likelihood of originating from the same country rather than from different countries, which is in agreement with our earlier analysis on the stability of the strain compositions of specific countries. This local spatial clustering was also evident in Europe, which is a region of multiple smaller countries in contrast to otherworld regions. Nevertheless, over a longer period (6-12 years), this local spatial clustering disappears, indicating substantial transmission of isolates across countries.

[0294] The dominance of local transmissions is also evident on a smaller geographic scale, such as cities. In this analysis, the focus was on Eastern and Southern European regions covered by our CRAB collection due to the availability of fine-scale geographic information. Specifically, when considering two contemporary CRAB isolates that diverged in the past 2 years, there was a 4-fold higher likelihood that they originated from the same city rather than from different cities within the country. This pattern indicates that within a 2-year timeframe, locally circulating strains in specific cities are poised to account for the majority of the CRAB infections.

[0295] The observation that spatial clustering of isolates only disappears after several years indicates the dominance of local transmissions in the case of this nosocomial pathogen. The calculated value suggests that CRAB strains currently circulating within a specific country are likely to be responsible for the majority of infections within a 6- year period.

[0296] Example 4 - Phylogeny-guided phage hunting and phage sensitivity profiling

[0297] The next goal was to demonstrate the potential of bacterial genomic surveillance in identifying a phage collection that is effective against the CRAB types responsible for the majority of infections within a specific region. To achieve this, the focus was put on the five neighbouring Eastern and Southern European countries that the genomic sample collection covers: Hungary, Romania, Serbia, Montenegro, and Bosnia and Herzegovina. For phage hunting, eleven MLST-CPS types that are both prevalent in this region were selected, collectively accounting for 86.75% of samples between 2016 and 2022, and also regionally relevant, since ten out of eleven are among the most prevalent MLST-CPS types in entire Eastern Europe. Subsequently, a phage hunt using representative clinical isolates from each of the 11 MLST-CPS types was conducted. Phages were isolated from raw communal wastewater collected in five Hungarian cities (Example 1). This resulted in the successful isolation of a total of 15 phages, all of which exhibited potent lytic activity (>106pfu) against at least one target clinical isolate. Next, the efficacy of the isolated phages was quantified on a larger collection of 92 CRAB clinical isolates from the 11 MLST-CPS types (Figure 2). As anticipated, each MLST- CPS type had a unique phage susceptibility profile. Remarkably, for eight of these MLST-CPS types, phages that effectively targeted over 95% of the tested clinical isolates were identified. For example, phage Highwayman potently targeted 28 out of the 29 tested isolates that belong to the most prevalent type (ST2-KL3) in Eastern Europe.

[0298] To further evaluate the broader relevance of the discovered phages, they were tested against an additional set of 107 A. baumannii clinical isolates using a simplified binary phage susceptibility profiling method (see Methods. These 107 isolates included further clinical isolates of the eleven MLST-CPS types from Eastern and Southern Europe and from distant locations, such as Western Europe and North America, representing an even broader genetic diversity for the eleven MLST-CPS types and 15 additional ST backgrounds. Subsequently, a systematic comparison of these binary phage susceptibilities across the entire dataset was conducted (i.e. 92+107 A baumannii isolates). The analysis revealed that variations in phage susceptibility profiles are mainly determined by CPS type, but altered ST background and large geographic distance also played a significant, albeit weaker, role. The observed impact of geographical distance hints that even within an MLST-CPS type, genetic divergence may also shape phage susceptibility. Indeed, pairs of isolates that diverged earlier tend to show less similar phage susceptibility profiles. Nevertheless, individual phages differ in their resilience to genetic variations within MLST-CPS types, with several phages retaining a killing effect even across isolates of the same MLST-CPS type that diverged genetically a long time ago. For example, an isolated phage designated as “Fishpie” is effective against all 13 tested ST2-KL2 isolates, despite their 30+ year genome divergence (Figure 3). The ability of these phages to remain effective over larger genetic distances makes them promising candidates for therapeutic applications. The same property cannot be defined for Porter, another pahge targeting ST2-KL12, where all isolates are phylogenetically close relatives with a divergence started about 5 years ago (Figure 3). The Figure shows the effectiveness of a phage against A. baumannii isolates that belong to an MLST-CPS type over various times of divergence. Specifically, the x axis shows the divergence between an isolate and a reference isolate.

[0299] Finally, specific CRAB types, such as ST636-KL40, exhibit genetic variability that is not well captured by the combined MLST-CPS typing or by the genetic distance alone, but it still markedly influences phage susceptibility. It was hypothesised that this genetic variation could induce changes in the cell surface without affecting the predicted CPS type. Therefore, the surface properties of recently diverged isolates of the ST636-KL40 type were characterised using an infrared spectroscopic method (Example 1). Importantly, cell surface variations that correlate significantly with phage susceptibility profile differences among the isolates were detected.

[0300] In summary, a phage library that is highly effective against most clinical isolates of CRAB in Eastern Europe has been successfully constructed. While bacterial susceptibility to phages is primarily dictated by CPS type, additional genetic variations that arise during the pathogen's spread across geographical regions also play a role. This underscores the importance of tailoring phage collections to the local genetic diversity of the pathogen. Example 5 - Characterization of the phage genomes

[0301] As a next step, the genomes of the 15 isolated phages were sequenced. The aim was to determine their suitability for clinical use by defining established safety criteria, including a functionally well-annotated genome, a lytic life cycle, and the absence of harmful genes. First, their protein sequence-based relationship with 3,562 known species of tailed dsDNA phages (Caudoviricetes) was analysed as done before. This allowed defining their higher-order taxonomy and genetic relatedness to phages that had been previously characterised as lytic or had been utilised for therapeutic purposes.

[0302] Out of the 15 discovered phages, which belong to 5 different taxonomic groups, 14 exhibit extensive sequence similarity to previously used therapeutic phages or well- characterised lytic phages. For example, seven phages from the Autographoviridae family share a DNA sequence identity of 80-84% with the therapeutic phage AbTP3phi1 . Furthermore, two-two myophage morphotype phages in the Twarogvirinae subfamily show 87-95% DNA sequence identity with two other previously used treatment phages, AB-Navy-4 and AB-Navy-71 , respectively. In all of these cases, despite the high sequence similarities stemming from conserved core gene sets, notable differences exist in critical genetic determinants of host specificity. Specifically, depolymerase- carrying tail proteins responsible for CPS decomposition differ from the corresponding components of previously reported treatment or lytic phages in almost every case. These findings suggest the existence of closely related Acinetobacter phages across the world, with minor genetic adaptations to accommodate various host strains. The relatively modest genetic variations compared to previously established lytic and therapeutic phages simplify the comprehensive characterization necessary for clinical application.

[0303] In contrast to the above general patterns, Silvergun exhibits a unique genetic composition that distinguishes it from all other currently known phages. Silvergun shares very few orthogroups with any other known phages, suggesting that Silvergun may be the first characterised member of a new bacteriophage family. Importantly, Silvergun is able to target the two most prevalent globally spreading strains belonging to ST2-KL3 and ST2-KL2 group, which collectively account for 31 % of the A. baumannii genomes isolated in 2022. It is possible that the presence of two predicted depolymerase regions in the tail spike and fibre proteins of this phage explains its uniquely relevant host specificity.

[0304] None of the 15 isolated phage genomes were found to contain toxin-, antibiotic resistance-genes or temperate markers that would hinder their potential future therapeutic application. All studied phages with similarity to well-characterised lytic phages are predicted to be lytic using the PhageAl tool, with the exception of one of the isolated phages called “Highwayman” whose lifestyle could not be defined reliably (Methods). Therefore, the absence of lysogenic behaviour for Highwayman was experimentally confirmed, along with Silvergun and PhT2-v2 phages, on their corresponding hosts using a previously established method.

[0305] Example 6 - Precision phage cocktail against the ST2-KL3 strain type in Europe

[0306] The first clinical case study involving Acinetobacter revealed a common challenge of phage therapy: the emergence of bacteriophage resistance, which can lead to treatment failure. To address this issue, a common strategy is the application of phage cocktails rather than individual phages.45 Therefore, the goal was to create a phage cocktail against ST2-KL3 CRAB isolates in Europe that can mitigate the emergence of phage resistance. ST2-KL3 was chosen due to its high prevalence throughout Europe. Moreover, its phylogenetic diversity in Europe is well represented among our isolates, facilitating the development of a widely applicable phage cocktail for ST2-KL3.

[0307] To start, the focus was put on Highwayman and Silvergun, two phages that together target 98% of a representative set of 56 ST2-KL3 clinical isolates. By coincubating these phages and their combination with a set of 41 randomly selected ST2- KL3 European clinical isolates in vitro, the emergence of phage resistance typically after 10-12 hours was observed, irrespective of the application of these phages individually or in combination (Fig 4A). Next, it was examined whether Highwayman or Silvergun resistance induces sensitivity to other phages in the collection. It was found that all 22 tested Silvergun resistant-lines became sensitive to at least one of four other phages from the collection. Combining one or two of these four phages with the initial cocktail of Highwayman and Silvergun significantly suppressed the growth of the target cells when measured up to 24 hours, indicating diminished resistance (Fig 4B). In the final phase, the emergence of resistance to one of the best performing phage cocktails that were termed HSFPh. HSFPh is a mixture of four isolated phages, including Highwayman, Silvergun, Fanak and PhT2-v2 was assessed. The growth of the 56 ST2-KL3 clinical isolates for up to 72 hours was monitored. Even during such a long evaluation, only sporadic instances of bacterial resistance emerging against HSFPh were detected (Fig 4C). To elucidate the genetic basis of the emerging bacterial resistance against the tested individual four phages and the corresponding phage cocktails, the genomes of 19 phage-resistant bacterial lines derived from 7 distinct wild-type ST2-KL3 isolates were sequenced. Resistance to Highwayman was driven by loss of function mutations in a specific glycosyltransferase gene (Igt2). In contrast, mutants resistant to Silvergun, Fanak, PhT2-v2 typically contained loss-of-function mutations in the genes involved in the capsular polysaccharide (CPS) biosynthesis pathway (Fig 4D). These patterns suggest that in contrast to Highwayman, resistance to these three phages requires defective capsule production. As a consequence of these mutations, the corresponding resistant strains displayed abolished phage adsorptions to their cell surface (Fig 4E). In addition, these strains displayed altered cell surface properties as revealed by the combination of infrared spectroscopy and electron microscopy (Example 1 , Fig 4F-G). Overall, as decapsulation is a known phage-resistance mechanism among other MLST- CPS strain types of CRAB, these results demonstrate that ST2-KL3 shares this feature with other CRAB types. Importantly, however, bacterial resistance to the HSFPh phage cocktail was scarce (Fig 4D).

[0308] The properties of the mutant bacterial lines were next characterised to evaluate the pleiotropic side effects of phage resistance. It was hypothesised that the emergence of resistance against the phage cocktail may render the bacterium less virulent or more susceptible to antibiotics. To test this, the in vivo virulence of the cocktail-resistant isolates in Galleria mellonella (G. mellonella) larvae infection model was measured. As expected, CPS-deficient strains exhibited significantly decreased virulence compared to the ancestor wild-type strain. Importantly, such a decrease in virulence was not observed in the case of Highwayman-resistant lines, which have intact CPS. Next, the antibiotic susceptibility profiles of 34 phage-resistant lines on a panel of clinically relevant antibiotics were measured. In the case of colistin, 44% of the tested lines transitioned from a resistant state to a sensitive state. In addition, the two isolated four-phage resistant lines became re-sensitized to meropenem, transitioning from the resistant state to the intermediate state (Methods).

[0309] Finally, the efficacy of the phages in in vivo infection models was tested. First, G. mellonella larvae infected with one of the ST2-KL3 isolates were treated with phage(s) 5 min post-infection resulting in a significant improvement in the survival rate of the larvae in all the cases. The efficiency of the phages in G. mellonella was corroborated by an infection model in mice (see Example 1). Specifically, after injecting one of the ST2-KL3 CRAB isolates intraperitoneally into mice, both treatment with Highwayman phage alone and with the HSFPh cocktail saved 100% of the animals, as opposed to the untreated animals which all died 1-day post-infection.

[0310] In sum, resistance to HSFPh phage cocktail is relatively rare, and even if it occurs, it leads to decreased virulence and enhanced susceptibility to meropenem. Notably, as carbapenem resistance is a predictor of mortality in patients infected with Acinetobacter, this resensitization by phage resistance can have clinical relevance and justifies the combinatorial use of the cocktail with meropenem to suppress the emergence of phageresistant variants. In line with this notion, the eradication of the target pathogen during phage therapy is more likely when phages are administered alongside antibiotics. More generally, these results indicate that the combined administration of phages with different modes of action can mitigate the emergence of resistance by forcing the evolutionary path into an avirulent and antibiotic-sensitive state.

[0311] Example 7: A phylogenetically diverse set of bacterial strains for testing phages is superior to a set of bacterial strains without phylogenetic information.

[0312] The present example focuses on one of the clinically relevant Klebsiella pneumoniae MLST-CPS type in Europe. Infectivity of two phages (k146_1 and k146_2 from wastewater was tested on a set of 36 isolates belonging to this TOL The isolate set was compiled to capture well the phylogenetic diversity of the TOL The infection rate for the two phages, i.e. , the proportion of isolates that can be infected by a particular phage (y-axis value = YES in Figure 5 A and B, was identical (82.85%). Thus, based solely on the infection rate, the two phages would appear therapeutically equivalent. However, a critical distinction emerged when considering phylogenetic information. Phage k146_1 uniquely maintains infectivity against the phylogenetically most divergent isolate, a property not shared by k146_2 (see circled points indicating the most distant isolates in Figure 5).

[0313] This example demonstrates a critical shortcoming in prior art phage selection methods, which rely solely on prevalence. Selecting phages based on a set of test isolates assembled based on prevalence alone fails to fully account for the evolutionary diversity within a TOL The present approach, utilizing a phylogenetically diverse test set, increases the likelihood of identifying phages with broad applicability across geographic regions and extended timeframes. Without the phylogenetic context, assessing the host range of phages becomes less accurate, diminishing the probability of selecting the most effective phage candidates. Consequently, the present methods ensure enhanced generalizability, particularly when representing a given TOI with a limited number of isolates, a common practice in the field.

[0314] Furthermore, this example demonstrates that merely sampling isolates from multiple geographic regions, as evidenced in the prior art of e.g., Bourdin et al. (Microbial Biotechnology 2014, 7(2): 165-176), fails to ensure adequate phylogenetic coverage. Notably, it was demonstrated that sampling multiple geographical regions alone does not guarantee the representation of the full evolutionary diversity within a TOI.

[0315] REFERENCES

[0316] Altschul, S. (1997). Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25, 3389-3402.

[0317] 10.1093 / nar / 25.17.3389.

[0318] Anderson, B., Rashid, M.H., Carter, C., Pasternack, G., Rajanna, C., Revazishvili, T., Dean, T., Senecal, A., and Sulakvelidze, A. (2011). Enumeration of bacteriophage particles. Bacteriophage 1 , 86-93. 10.4161 / bact.1.2.15456.

[0319] Arel-Bundock, V., Yetman, C.J., Enevoldsen, N., and Meichtry, S. (2023). countrycode: Convert Country Names and Country Codes. Version 1.5.0.

[0320] Bankevich, A., Nurk, S., Antipov, D., Gurevich, A.A., Dvorkin, M., Kulikov, A.S., Lesin, V.M., Nikolenko, S.L, Pham, S., Prjibelski, A.D., et al. (2012). SPAdes: A New Genome Assembly Algorithm and Its Applications to Single-Cell Sequencing. J. Comput. Biol. 19, 455-477. 10.1089 / cmb.2012.0021.

[0321] Beyer, H. (1981). Tukey, John W.: Exploratory Data Analysis. Addison-Wesley Publishing Company Reading, Mass. — Menlo Park, Cal., London, Amsterdam, Don Mills, Ontario, Sydney 1977, XVI, 688 S. Blom. J. 23, 413-414. 10.1002 / bimj.4710230408.

[0322] Bin Jang, H., Bolduc, B., Zablocki, O., Kuhn, J.H., Roux, S., Adriaenssens, E.M., Brister, J.R., Kropinski, A.M., Krupovic, M., Lavigne, R., et al. (2019). Taxonomic assignment of uncultivated prokaryotic virus genomes is enabled by gene-sharing networks. Nat. Biotechnol. 37, 632-639. 10.1038 / s41587-019-0100-8.

[0323] Bolger, A.M., Lohse, M., and LJsadel, B. (2014). Trimmomatic: a flexible trimmer for Illumina sequence data. Bioinformatics 30, 2114-2120.

[0324] 10.1093 / bioinformatics / btu170.

[0325] Bonilla, N., Rojas, M.L, Netto Flores Cruz, G., Hung, S.-H., Rohwer, F., and Barr, J. J. (2016). Phage on tap-a quick and efficient protocol for the preparation of bacteriophage laboratory stocks. PeerJ 4, e2261. 10.7717 / peerj.2261.

[0326] Buchfink, B., Reuter, K., and Drost, H.-G. (2021). Sensitive protein alignments at tree-of-life scale using DIAMOND. Nat. Methods 18, 366-368. 10.1038 / s41592-021 - 01101-x.

[0327] Camacho, C., Coulouris, G., Avagyan, V., Ma, N., Papadopoulos, J., Bealer, K., and Madden, T.L. (2009). BLAST+: architecture and applications. BMC Bioinformatics 10, 421. 10.1186 / 1471-2105-10-421.

[0328] Charif, D., and Lobry, J.R. (2007). SeqinR 1.0-2: A Contributed Package to the R Project for Statistical Computing Devoted to Biological Sequences Retrieval and Analysis. In Structural Approaches to Sequence Evolution: Molecules, Networks, Populations Biological and Medical Physics, Biomedical Engineering., LJ. Bastolla, M. Porto, H. E. Roman, and M. Vendruscolo, eds. (Springer), pp. 207-232. 10.1007 / 978-3- 540-35306-5 10.

[0329] Corvee, S., Poirel, L., Naas, T., Drugeon, H., and Nordmann, P. (2007). Genetics and Expression of the Carbapenem-Hydrolyzing Oxacillinase Gene blaOXA-23 in Acinetobacter baumannii. Antimicrob. Agents Chemother. 51 , 1530-1533. 10.1128 / aac.01132-06.

[0330] Croucher, N.J., Page, A.J., Connor, T.R., Delaney, A.J., Keane, J.A., Bentley, S.D., Parkhill, J., and Harris, S.R. (2015). Rapid phylogenetic analysis of large samples of recombinant bacterial whole genome sequences using Gubbins. Nucleic Acids Res. 43, e15. 10.1093 / nar / gku1196.

[0331] Deorowicz, S., Debudaj-Grabysz, A., and Grabowski, S. (2013). Disk-based k- mer counting on a PC. BMC Bioinformatics 14, 160. 10.1186 / 1471-2105-14-160.

[0332] Evans, B.A., and Amyes, S.G.B. (2014). OXA -Lactamases. Clin. Microbiol. Rev. 27, 241-263. 10.1128 / cmr.00117-13.

[0333] Geisinger, E., Vargas-Cuebas, G., Mortman, N.J., Syal, S., Wainwright, E.L., Lazinski, D., Wood, S., Zhu, Z., Anthony, J., Opijnen, T.V., et al. (2018). The landscape of intrinsic and evolved fluoroquinolone resistance in Acinetobacter baumannii includes suppression of drug-induced prophage replication. 10.1101 / 442681.

[0334] Gilchrist, C.L.M., and Chooi, Y.-H. (2021). clinker & clustermap.js: automatic generation of gene cluster comparison figures. Bioinformatics 37, 2473-2475. 10.1093 / bioinformatics / btab007.

[0335] Hatfull, G.F., Dedrick, R.M., and Schooley, R.T. (2022). Phage Therapy for Antibiotic-Resistant Bacterial Infections. Annu. Rev. Med. 73, 197-211. 10.1146 / annurev-med-080219-122208.

[0336] Hawkey, J., Hamidian, M., Wick, R.R., Edwards, D.J., Billman-Jacobe, H., Hall, R.M., and Holt, K.E. (2015). ISMapper: identifying transposase insertion sites in bacterial genomes from short read sequence data. BMC Genomics 16, 667. 10.1186 / s12864- 015-1860-2.

[0337] Hietala, V., Horsma-Heikkinen, J., Carron, A., Skurnik, M., and Kiljunen, S. (2019). The Removal of Endo- and Enterotoxins From Bacteriophage Preparations. Front. Microbiol. 10.

[0338] Hyatt, D., Chen, G.-L., LoCascio, P.F., Land, M.L., Larimer, F.W., and Hauser, L.J. (2010). Prodigal: prokaryotic gene recognition and translation initiation site identification. BMC Bioinformatics 11 , 119. 10.1186 / 1471-2105-11-119.

[0339] Jolley, K. A., Bray, J. E., & Maiden, M. C. (2018). Open-access bacterial population genomics: BIGSdb software, the PubMLST. org website and their applications. Wellcome open research, 3.

[0340] Katoh, K., and Standley, D.M. (2013). MAFFT Multiple Sequence Alignment Software Version 7: Improvements in Performance and Usability. Mol. Biol. Evol. 30, 772-780. 10.1093 / molbev / mst010.

[0341] Kelley, L.A., Mezulis, S., Yates, C.M., Wass, M.N., and Sternberg, M.J.E. (2015). The Phyre2 web portal for protein modeling, prediction and analysis. Nat. Protoc. 10, 845-858. 10.1038 / nprot.2015.053.

[0342] Kokot et al. - 2017 - KMC 3 counting and manipulating k -mer sta.pdf.

[0343] Kokot, M., Dlugosz, M., and Deorowicz, S. (2017). KMC 3: counting and manipulating k -mer statistics. Bioinformatics 33, 2759-2761.

[0344] 10.1093 / bioinformatics / btx304.

[0345] Kutter, E. (2009). Phage host range and efficiency of plating. Methods Mol. Biol. Clifton NJ 501 , 141-149. 10.1007 / 978-1 -60327-164-6_14.

[0346] Lam, M.M.C., Wick, R.R., Judd, L.M., Holt, K.E., and Wyres, K.L. (2022). Kaptive 2.0: updated capsule and lipopolysaccharide locus typing for the Klebsiella pneumoniae species complex. Microb. Genomics 8, 000800. 10.1099 / mgen.0.000800.

[0347] Latka, A., Leiman, P.G., Drulis-Kawa, Z., and Briers, Y. (2019). Modeling the Architecture of Depolymerase-Containing Receptor Binding Proteins in Klebsiella Phages. Front. Microbiol. 10, 2649. 10.3389 / fmicb.2019.02649.

[0348] Leinonen, R., Akhtar, R., Birney, E., Bower, L., Cerdeno-Tarraga, A., Cheng, Y., Cleland, I., Faruque, N., Goodgame, N., Gibson, R., et al. (2011). The European Nucleotide Archive. Nucleic Acids Res. 39, D28-D31. 10.1093 / nar / gkq967.

[0349] Li, D., Liu, C.-M., Luo, R., Sadakane, K., and Lam, T.-W. (2015). MEGAHIT: an ultra-fast single-node solution for large and complex metagenomics assembly via succinct de Bruijn graph. Bioinforma. Oxf. Engl. 31 , 1674-1676.

[0350] 10.1093 / bioinformatics / btv033.

[0351] Li, S., Duan, X., Peng, Y., and Rui, Y. (2019). Molecular characteristics of carbapenem-resistant Acinetobacter spp. from clinical infection samples and fecal survey samples in Southern China. BMC Infect. Dis. 19, 900. 10.1186 / s12879-019- 4423-3.

[0352] Liu, C., Chen, K., Wu, Y., Huang, L., Fang, Y., Lu, J., Zeng, Y., Xie, M., Chan, E.W.C., Chen, S., et al. (2022). Epidemiological and genetic characteristics of clinical carbapenem-resistant Acinetobacter baumannii strains collected countrywide from hospital intensive care units (ICLJs) in China. Emerg. Microbes Infect. 11 , 1730-1741. 10.1080 / 22221751 .2022.2093134.

[0353] Magill, D.J., and Skvortsov, T.A. (2023). DePolymerase Predictor (DePP): a machine learning tool for the targeted identification of phage depolymerases. BMC Bioinformatics 24, 208. 10.1186 / s12859-023-05341 -w.

[0354] Mai, U., and Mirarab, S. (2018). TreeShrink: fast and accurate detection of outlier long branches in collections of phylogenetic trees. BMC Genomics 19, 272. 10.1186 / s12864-018-4620-2.

[0355] Martin, M. (2011). Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet.journal 17, 10-12. 10.14806 / ej.17.1 .200.

[0356] Moura et al. (2021). Emergence and global spread of Listeria monocytogenes main clinical clonal complex | Science Advances Science Advances Vol 7, Issue 49. https: / / www.science.Org / doi / 10.1126 / sciadv.abj9805.

[0357] Mutalik, V.K., and Arkin, A.P. (2022). A Phage Foundry Framework to Systematically Develop Viral Countermeasures to Combat Antibiotic-Resistant Bacterial Pathogens. IScience 25, 104121. 10.1016 / j.isci.2O22.104121 .

[0358] Nayfach et al. (2021) CheckV assesses the quality and completeness of metagenome-assembled viral genomes | Nature Biotechnology Nature Biotechnology volume 39, pages 578-585; https: / / www.nature.com / articles / s41587-020-00774-7.

[0359] Novovic, K., Mihajlovic, S., Vasiljevic, Z., Filipic, B., Begovic, J., and Jovcic, B. (2015). Carbapenem-Resistant Acinetobacter baumannii from Serbia: Revision of CarO Classification. PLOS ONE 10, e0122793. 10.1371 / journal.pone.0122793.

[0360] Nurk, S., Meleshko, D., Korobeynikov, A., and Pevzner, P.A. (2017). metaSPAdes: a new versatile metagenomic assembler. Genome Res. 27, 824-834. 10.1101 / gr.213959.116.

[0361] Oksanen, J., Simpson, G.L., Blanchet, F.G., Kindt, R., Legendre, P., Minchin, P.R., O’Hara, R.B., Solymos, P., Stevens, M.H.H., Szoecs, E., et al. (2022). vegan: Community Ecology Package. Version 2.6-4.

[0362] Olson, R.D., Assaf, R., Brettin, T., Conrad, N., Cucinell, C., Davis, J. J., Dempsey, D.M., Dickerman, A., Dietrich, E.M., Kenyon, R.W., et al. (2023). Introducing the Bacterial and Viral Bioinformatics Resource Center (BV-BRC): a resource combining PATRIC, IRD and ViPR. Nucleic Acids Res. 51 , D678-D689. 10.1093 / nar / gkac1003.

[0363] Page, A.J., Taylor, B., Delaney, A.J., Soares, J., Seemann, T., Keane, J.A., and Harris, S.R. (2016). SNP-sites: rapid efficient extraction of SNPs from multi-FASTA alignments. Microb. Genomics 2, e000056. 10.1099 / mgen.0.000056.

[0364] Pages, H., Aboyoun, P., Gentleman, R., DebRoy, S., Carey, V., Delhomme, N., Ernst, F., Khan, H., Lakshman, A., O’Neill, K., et al. (2023). Biostrings: Efficient manipulation of biological strings. Version 2.70.1 (Bioconductor version: Release (3.18)). 10.18129 / B9. bloc. Biostrings 10.18129 / B9. bloc. Biostrings.

[0365] Popova, A.V., Shneider, M.M., Arbatsky, N.P., Kasimova, A.A., Senchenkova, S.N., Shashkov, A.S., Dmitrenok, A.S., Chizhov, A.O., Mikhailova, Y.V., Shagin, D.A., et al. (2021). Specific Interaction of Novel Friunavirus Phages Encoding Tailspike Depolymerases with Corresponding Acinetobacter baumannii Capsular Types. J. Virol. 95, 10.1128 / jvi.O1714-20. 10.1128 / jvi.O1714-20.

[0366] Price, M.N., Dehal, P.S., and Arkin, A.P. (2009). FastTree: Computing Large Minimum Evolution Trees with Profiles instead of a Distance Matrix. Mol. Biol. Evol. 26, 1641-1650. 10.1093 / molbev / msp077.

[0367] R: The R Project for Statistical Computing https: / / www.r-project.org / .

[0368] Regelmumbal et al. (2016) Antimicrobial Agents and Chemotherapy, Volume 60 Number 10, pp5806-5816.

[0369] Reynolds, E.S. (1963). THE USE OF LEAD CITRATE AT HIGH pH AS AN ELECTRON-OPAQUE STAIN IN ELECTRON MICROSCOPY. J. Cell Biol. 17, 208-212. 10.1083 / jcb.17.1.208.

[0370] Schneider, G., Szentes, N., Horvath, M., Dorn, A., Cox, A., Nagy, G., Doffkay, Z., Maroti, G., Rakhely, G., and Kovacs, T. (2018). Kinetics of Targeted Phage Rescue in a Mouse Model of Systemic Escherichia coll K1. BioMed Res. Int. 2018, e7569645. 10.1155 / 2018 / 7569645.

[0371] Seemann, T. (2023). https: / / github.com / tseemann / mlst.

[0372] Seemann, T. (2023). https: / / github.com / tseemann / snippy.

[0373] Segal, H., Jacobson, R.K., Garny, S., Bamford, C.M., and Elisha, B.G. (2007). Extended -10 Promoter in ISAba-1 Upstream of blaOXA-23 from Acinetobacter baumannii. Antimicrob. Agents Chemother. 51 , 3040-3041. 10.1128 / aac.00594-07.

[0374] Siguier, P., Perochon, J., Lestrade, L., Mahillon, J., and Chandler, M. (2006). ISfinder: the reference centre for bacterial insertion sequences. Nucleic Acids Res. 34, D32-36. 10.1093 / nar / gkj014.

[0375] Simao, F.A., Waterhouse, R.M., loannidis, P., Kriventseva, E.V., and Zdobnov, E.M. (2015). BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics 31 , 3210-3212. 10.1093 / bioinformatics / btv351.

[0376] Strathdee, S.A., Hatfull, G.F., Mutalik, V.K., and Schooley, R.T. (2023). Phage therapy: From biological mechanisms to future directions. Cell 186, 17-31. 10.1016 / j. cell.2022.11 .017.

[0377] Teng, A.S.J., Habermehl, P.E., van Houdt, R., de Jong, M.D., van Mansfeld, R., Matamoros, S.P.F., Spijkerman, I.J.B., van Meer, M.P.A., and Visser, C.E. (2022). Comparison of fast Fourier transform infrared spectroscopy biotyping with whole genome sequencing-based genotyping in common nosocomial pathogens. Anal. Bioanal. Chem. 414, 7179-7189. 10.1007 / s00216-022-04270-6.

[0378] The European Committee on Antimicrobial Susceptibility Testing. Breakpoint tables for interpretation of MICs and zone diameters. Version 13.1 , 2023. http: / / www.eucast.org.

[0379] Turton, J.F., Ward, M.E., Woodford, N., Kaufmann, M.E., Pike, R., Livermore, D.M., and Pitt, T.L. (2006). The role of ISAbal in expression of OXA carbapenemase genes in Acinetobacter baumannii. FEMS Microbiol. Lett. 258, 72-77. 10.1111 / j.1574- 6968.2006.00195.x.

[0380] Tynecki, P., Guzinski, A., Kazimierczak, J., Jadczuk, M., Dastych, J., and Onisko, A. (2020). PhageAl - Bacteriophage Life Cycle Recognition with Machine Learning and Natural Language Processing. Preprint at bioRxiv, 10.1101 / 2020.07.11.198606 10.1101 / 2020.07.11.198606.

[0381] Valcek, A., Collier, J., Botzki, A., and Van Der Henst, C. (2022). Acinetobase: the comprehensive database and repository of Acinetobacter strains. Database 2022, baac099. 10.1093 / database / baac099.

[0382] Venturini, C., Petrovic Fabijan, A., Fajardo Lubian, A., Barbirz, S., and Iredell, J. (2022). Biological foundations of successful bacteriophage therapy. EMBO Mol. Med. 14, e12435. 10.15252 / emmm.202012435.

[0383] Verbeken, G., and Pirnay, J.P. (2022). European regulatory aspects of phage therapy: magistral phage preparations. Curr. Opin. Virol. 52, 24-29.

[0384] 10.1016 / j.coviro.2021.11.005.

[0385] Volz, E.M., and Frost, S.D.W. (2017). Scalable relaxed clock phylogenetic dating. Virus Evol. 3, vex025. 10.1093 / ve / vex025.

[0386] Wickham, H., Francois, R., Henry, L., Muller, K., Vaughan, D., Software, P., and PBC (2023). dplyr: A Grammar of Data Manipulation. Version 1.1.4.

[0387] Wiegand, !., Hilpert, K., and Hancock, R.E.W. (2008). Agar and broth dilution methods to determine the minimal inhibitory concentration (MIC) of antimicrobial substances. Nat. Protoc. 3, 163-175. 10.1038 / nprot.2007.521.

[0388] Wood, D.E., Lu, J., and Langmead, B. (2019). Improved metagenomic analysis with Kraken 2. Genome Biol. 20, 257. 10.1186 / s13059-019-1891-0.

[0389] Zander, E., Nemec, A., Seifert, H., and Higgins, P.G. (2012). Association between P-Lactamase-Encoding bla OXA-51 Variants and DiversiLab Rep-PCR-Based Typing of Acinetobacter baumannii Isolates. J. Clin. Microbiol. 50, 1900-1904.

[0390] 10.1128 / jcm.06462-11.

[0391] Zankari, E., Hasman, H., Cosentino, S., Vestergaard, M., Rasmussen, S., Lund, O., Aarestrup, F.M., and Larsen, M.V. (2012). Identification of acquired antimicrobial resistance genes. J. Antimicrob. Chemother. 67, 2640-2644. 10.1093 / jac / dks261.

Claims

CLAIMS1 . A method of preparing a region-specific phage composition effectively targeting an antibiotic-resistant nosocomial bacterial species and, comprising: a) performing molecular typing of isolates of the bacterial species to identify a type of interest (TOI) that is prevalent in a predetermined geographical region, wherein said molecular typing comprises DNA sequencing and clustering the genomes into genotypes; b) performing phylogeographical analysis of strains of said TOI to identify target strains, which are phylogenetically most divergent strains of said TOI in said predetermined geographical region; c) selecting a repertoire of bacteriophages from a bacteriophage source, which repertoire exhibits lytic activity against said target strains; and d) preparing a pharmaceutical phage composition comprising said repertoire.

2. The method of claim 1 , which comprises: a) performing molecular typing of isolates of the bacterial species to identify one, two or more TOIs that are prevalent in said predetermined geographical region; b) performing phylogeographical analysis of strains of each of said TOIs to identify target strains for each of said TOIs, which are phylogenetically most divergent in said predetermined geographical region; c) selecting one, two or more repertoires of bacteriophages from a bacteriophage source, wherein said repertoires exhibit lytic activity against target strains; and d) preparing a pharmaceutical phage composition comprising said one, two or more repertoires.

3. The method of claim 1 or 2, wherein said molecular typing comprises any one or more or all of cell surface typing, antibiotic-resistance typing, whole genome sequence typing, or multilocus variable-number tandem-repeat analysis (MLVA) typing.

4. The method of any one of claims 1 to 3, wherein said molecular typing comprises a) evolutionary distance typing, such as multilocus sequence typing (MLST) or whole genome multilocus sequence typing (cgMLST); and / orb) surface structure typing, such as surface capsular polysaccharide typing or O antigen typing.

5. The method of any one of claims 1 to 4, wherein said molecular typing comprises combined multilocus sequence typing (MLST) and cell surface capsular polysaccharide (CPS) genotyping, preferably wherein a) said MLST clusters evolutionary-related genomes into sequence types based on the sequence of at least 6 different housekeeping genes; and b) said CPS comprises identifying polysaccharide capsule and outer lipopolysaccharide loci and classifies assemblies into CPS-types.

6. The method of any one of claims 1 to 5, wherein the phylogeographical analysis comprises: a) determining a number of single nucleotide polymorphisms (SNPs) in the core genome of the target strains; and b) phylogenetic analysis using a time-scaled phylogenetic tree to visualize the chronological genomic development in the core genome; wherein genetic distance is determined by the phylogenetic differences among the target strains.

7. The method of claim 6, wherein the core genome is a fraction of the whole genome shared by all of said isolates which are of the identified TOI, which core genome may differ in a number of SNPs.

8. The method of any one of claims 1 to 7, wherein the bacterial species is selected from the group consisting of: Acinetobacter baumannii, Klebsiella pneumoniae, Enterococcus faecium, Staphylococcus aureus, Pseudomonas aeruginosa, Escherichia coli and Enterobacter cloaceae, preferably wherein the bacterial species is resistant against to an antibiotic selected from the group consisting of: beta-lactams, such as carbapenem or non-carbapenem beta-lactams, aminoglycosides, fluoroquinolones sulfonamides, tetracyclines, macrolides, glycopeptides, and polymyxins.

9. The method of any one of claims 1 to 8, wherein the isolates of bacterial species are carbapenem-resistant Acinetobacter baumannii (CRAB) strains or carbapenem- resistant Klebsiella pneumoniae (CRKP) strains.

10. The method of any one of claims 1 to 9, wherein the predetermined geographical region is any one or more of Western Europe, Eastern Europe, Southern Europe, North-America, South-America; West-Asia, East-Asia, or South-East-Asia.

11. The method of any one of claims 1 to 10, wherein the predetermined geographical region is Eastern and Southern Europe, and the repertoire of bacteriophages exhibits lytic activity against CRAB strains of the ST2 lineage, preferably the ST2-KL3 MLST-CPS type.

12. The method of any one of claims 1 to 10, wherein the predetermined geographical region is Eastern and Southern Europe, and the repertoire of bacteriophages exhibits lytic activity against CRKP strains of the ST101-KL17 MLST- CPS type.

13. The method of any one of claims 1 to 12, wherein the pharmaceutical phage composition comprises a repertoire of bacteriophages which lyses isolates of said TOI in a lysis test, wherein at least 60% of the isolates are prevalent in said predetermined geographical region.

14. The method of any one of claims 1 to 13, wherein the lytic activity is at least 106pfu per mL.

15. The method of any one of claims 1 to 14, wherein the bacteriophage source is originating from an environmental sample of the predetermined geographical region, preferably wherein the environmental sample is of waste water, soil, sewage, aquatic habitats, agricultural runoff, hospital environments, food processing facilities, animal feces or a plant source.

16. The method of any one of claims 1 to 15, wherein the bacteriophage source is a phage library, preferably wherein the phage library comprises at least 10, 15, 20, 30, 40, 50, 100, or 200 different bacteriophages.

17. The method of any one of claims 1 to 16, wherein a repertoire of bacteriophages comprises or consists of 3-15 different bacteriophages.

18. The method of any one of claims 1 to 17, wherein a pharmaceutical phage composition is produced which comprises one or more repertoires of bacteriophages in a pharmaceutical formulation.

19. The method of claim 18, wherein the pharmaceutical phage composition is provided as a kit of parts comprising one or more bacteriophages of said one or more repertoires in separate containments.

20. Use of phylogeographical analysis and molecular typing of strains of interest (SOIs) of isolates of an antibiotic-resistant nosocomial bacterial species in a method of preparing a region-specific phage composition with relevance for a predetermined geographical region.

21. The use of claim 20, wherein the phylogeographical analysis comprises: a) determining a number of single nucleotide polymorphisms (SNPs) in the core genome of the SOIs; and b) phylogenetic analysis using a time-scaled phylogenetic tree to visualize the chronological genomic development in the core genome.

22. The use of claim 21 , wherein the core genome is a fraction of the whole genome shared by all of said SOIs, which core genome may differ in a number of SNPs.

Images