Acinetobacter baumannii is a Gram-negative nosocomial opportunistic pathogen. It is associated with a wide spectrum of hospital-acquired infections, such as pneumonia, bloodstream infections, urinary tract infections, and wound infections [1]. The World Health Organization published a global priority list of multidrug-resistant (MDR) bacteria in 2017 in which carbapenem-resistant A. baumannii appeared in the first priority group [2]. A. baumannii is also a part of the ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, A. baumannii, Pseudomonas aeruginosa, and Enterobacter species) pathogen group with exceptionally high resistance towards antibiotics [3]. A. baumannii has numerous antibiotic resistance mechanisms, which is reflected by the fact that the proportion of MDR strains reached 64% in 2014 [4]. Although A. baumannii causes a minor proportion of all infections associated with Gram-negative pathogens, mortality rates as high as 40–50% have been associated with these infections in intensive care units [4].

One promising method to treat MDR A. baumannii infections is phage therapy (PT), which utilizes bacteriophages (phages), which are viruses that infect bacteria. Phages are highly specific and only infect their target bacteria, and they do not affect the patient’s normal microbiota or human cells. Other advantages of phage therapy are, for example, their self-dosing capability, effective biofilm clearance, and low environmental impact. One of the disadvantages of PT is its narrow host range, which makes it difficult to find a suitable phage [5]. In addition, every phage used for phage therapy has to be characterized and found to be free of genes that encode bacterial toxins or are associated with antibiotic resistance, transposable elements, or a lysogenic life cycle [6].

Phage therapy has been shown to successfully cure A. baumannii infections when antibiotic medications have failed [4, 7]. Phages infecting A. baumannii have previously been isolated from various sources, most typically from hospital wastewater outside of Western countries [8,9,10,11]. Since antibiotic resistance is a major problem worldwide, effective therapeutic phages are needed globally.

Here, we describe the isolation and characterization of three phages from Beninese hospital wastewater. All bacterial strains used in this study were clinical Acinetobacter isolates (Table 1), most of which were collected from Finnish patients, but with unknown geographical origins. Samples for phage isolation included 34 wastewater samples that were collected during November and December 2019 in Benin from hospitals located in the adjacent cities Cotonou and Abomey-Calavi [12]. None of the hospitals were connected to a sewer system, and the samples were obtained from septic tanks or sumps. In most cases, the toilet water was not directed into these sumps. To remove debris and bacteria, the samples were filtered through a 0.2-µm polycarbonate filter (Whatman™, GE Healthcare Life Sciences), and the filtrates were transported to Finland.

Table 1 Acinetobacter strains used in the study
Full size table

Bacterial and phage incubations were conducted in lysogeny broth (LB) [13] unless otherwise stated. For solid and semisolid media, LB was supplemented with 1.5% and 0.4% agar, respectively. All bacterial and phage cultures were maintained at 37oC using standard methods [13]. A total of 23 A. baumannii strains (Table 1) were used for phage isolation by a previously described method [14]. Wastewater samples were pooled into four cocktails, and each cocktail was tested against individual strains. Each phage that was obtained was further purified by three rounds of plaque purification [13]. To determine its host range, each phage was tested against 57 Acinetobacter strains (Table 1) by the liquid culture method, using a Bioscreen C plate reader [15].

Phage DNA was isolated from a raw high-titer lysate (1010 plaque-forming units PFU/ml) using phenol-chloroform extraction and ethanol precipitation [13]. DNA samples were sequenced by the 150-bp paired-end protocol on an Illumina HiSeq platform at Novogene (https://en.novogene.com/). For de novo assembly of the phage genome sequences, a 50,000-read subset was made from both forward and reverse sequence files, using Chipster v4 [16]. Genome sequences were assembled using the A5-miseq integrated pipeline for the de novo assembly of microbial genomes [17]. BLASTn 2.10.1+ [18] was used to identify phage genome sequences among the resulting contigs. PhageTerm [19] was used to predict the physical ends of the phage genomes. The genome sequences were trimmed and organized manually based on the PhageTerm results. Assemblies were confirmed using Geneious Prime 2020.1.2 (https://www.geneious.com/) by mapping all original reads (7,913,416, 7,097,864, and 5,793,626 reads for fBenAci001, fBenAci002, and fBenAci003, respectively) back to the de novo assemblies. Preliminary annotations were made using RAST 2.0 [20,21,22], after which the annotations were checked and edited manually using Artemis 18.1.0 [23], BLASTp 2.10.1 [24], and HHpred [25]. The sequences were examined for the presence of tRNA genes using Aragorn v1.2.38 [26], VirulenceFinder 2.0 [27] was used to screen for genes coding for bacterial toxins, and ResFinder 4.0 [28] and CARD 3.0.7 [29] were used to screen for antibiotic resistance genes. All of the bioinformatic tools were applied using standard parameters.

Genome sequences were aligned using the Geneious Prime alignment tool with standard settings. Based on the alignment, the least similar protein that the phages had in common (tailspike), and one conserved structural protein (capsid protein) were selected for phylogenetic analysis. The closest relatives of the phages at the whole-genome level were identified using BLASTn, and matches with over 90% query coverage (Supplementary Table S1) were used for the construction of phylogenetic trees and a genome similarity heat map. The predicted amino acid sequences of the tailspike and capsid proteins were identified in the selected genome sequences and re-annotated, if needed.

A genome-wide phylogenetic tree was constructed using VICTOR (Virus Classification and Tree Building Online Resource) [30] using Genome-BLAST Distance Phylogeny [31] with the standard settings recommended for the prokaryotic viruses [30]. The resulting intergenomic distances were used to infer a balanced minimum evolution tree with branch support via FASTME including SPR postprocessing [32]. Branch support was inferred from 100 pseudo-bootstrap replicates. A tree was rooted at the midpoint [33] and visualized with ggtree [34]. Taxon boundaries at the species, genus, and family level were estimated using the OPTSIL program, the recommended clustering thresholds [30], and an F value (fraction of links required for cluster fusion) of 0.5 [35]. Phylogenetic trees for protein-level connections were built using the GGDC web server [31], which uses the DSMZ phylogenomics pipeline [35] adapted to single genes. A genome similarity heat map was generated using the VirClust WEB server at Virus Intergenomic Distance Calculator VIRIDIC [36].

Phages fBenAci001, fBenAci002, and fBenAci003 were isolated using A. baumannii strains #5542, #5707, and #5910, respectively. All three phages originated from the same wastewater pool, and each of them produced clear plaques with slightly different sizes, with fBenAci001 and fBenAci002 producing a halo (Supplementary Fig. S1). Under optimal conditions, all three phages gave a high titer (109-1010 PFU/ml) in their isolation hosts. All three phages were found to have a very narrow host range. Phage fBenAci001 infected only its original isolation host, and phages fBenAci002 and fBenAci003 were able to infect one additional A. baumannii strain each (Table 1).

An overview of the phage genomes is presented in Table 2. The genome sequences of phages fBenAci001, fBenAci002, and fBenAci003 are available in the GenBank database under the accession numbers MW056501, MW056502, and MW056503, respectively. All phages had approximately 41-kbp genomes and a GC content of 39.2. All predicted coding sequences (CDSs) were in the forward orientation. Direct terminal repeats with lengths of 368–394 bp were identified using PhageTerm. No genes encoding tRNAs or bacterial toxins or genes associated with antibiotic resistance, transposable elements, or a temperate life cycle were identified in the genomes.

Table 2 Overview of the phages
Full size table

A comparison of the phage genomes using VIRDIC showed that the sequence identity of all three phages to their closest relatives was between 80% and 91% (Supplementary Fig. S2). A Genome-BLAST Distance Phylogeny tree using the distance formula D0 is presented in Fig. 1A. The OPTSIL clustering yielded 21 species clusters. The tree revealed that phages fBenAci001, fBenAci002, and fBenAci003 are closely related to each other and resemble phages of the genus Friunavirus, family Autographiviridae. Each phage represents a novel, separate species. The deduced amino acid sequences of the predicted tail spike proteins (Fig. 1B) and capsid proteins (Fig. 1C) were used to build protein-level phylogenetic trees, using the GGDC web server. When comparing the protein-level phylogenetic trees with the tree based on whole genome sequences, a mosaic-like pattern was seen instead of a clear correlation.

Fig. 1
figure 1

Phylogenetic trees of the phages fBenAci001 (yellow), fBenAci002 (blue), and fBenAci003 (green) and their 21 closest relatives. (A) Genome-BLAST Distance Phylogeny tree using the distance formula D0, constructed using VICTOR. The colors of the symbols in the “species” column indicate whether phages belong to the same or different species within the same family and genus. (B) Phylogenetic tree based on the tailspike protein, calculated using the GGDC web server. (C) Phylogenetic tree based on the capsid protein, calculated using the GGDC web server. CorelDRAW was used to create the panel

Full size image

To conclude, three phages infecting clinical A. baumannii strains from Finnish patients were isolated from Beninese water samples. Phages fBenAci001, fBenAci002, and fBenAci003 were identified as three novel phages related to members of the genus Friunavirus of the family Autographiviridae. The phages were found to have a very narrow host range, which is a typical feature of A. baumannii phages [37,38,39]. Based on previous studies, the Friunavirus phages have varying capability to infect A. baumannii strains, infecting 3–50% of the strains tested. In most of the studies that showed a broad (over 30%) host range, all of the strains were isolated from the same hospital [11, 40] or from an otherwise limited area [41]. This might suggest that the bacterial strains tested in those studies had at least a partially clonal origin. When phage infectivity was tested against strains from diverse origins, the phages typically infected only a few strains [39] or just the isolation host [8].

As there were no known genes associated with a temperate life cycle, transposable elements, antibiotic resistance, or bacterial virulence in their genomes, phages fBenAci001, fBenAci002, and fBenAci003 meet the current safety requirements for therapeutic phages. However, it is worth noting that the functions of around half of the predicted gene products were unidentified. Other phages of the genus Friunavirus have been used successfully in phage therapy trials [41, 42], supporting the suitability of the phages characterized in this study for phage therapy. However, their narrow host range may limit the usefulness of these phages in therapeutic applications.

The phages characterized in this study were isolated from hospital wastewater from Benin, confirming an earlier finding by Essoh et al. [9] that wastewater from Africa is a good source for A. baumannii phages. In our experience, isolating A. baumannii phages from Finnish wastewater is very challenging, and water samples from other locations are therefore important resources for these phages. In Finland, Acinetobacter infections are often linked to travel. A Finnish study analyzing the origin of MDR Acinetobacter colonization in hospitalized patients transferred to Finland from other countries showed that most MDR Acinetobacter originated from Asia, North Africa, or the Middle East [43]. Notably, there was no MDR Acinetobacter transferred from sub-Saharan Africa between 2010 and 2019, presumably due to the low numbers of travelers to the region. However, in a report from 2019, the all-age death rate attributable to antimicrobial resistance was the highest in western sub-Saharan Africa [44]. For this reason, it is important to isolate new phages from this area to help fight against resistant infections.