Bacteriophages, phi9102 (PP898111), phi4197 (PQ432283), phi5013-M1 (PQ432284), phi5013-M2 (PQ432285), phi5038-11536 (PQ432286), phi11547 (PQ432287), phi5038-11551 (PQ432288), all mitomycin-C-inducible prophages were obtained from the clinical MDR A. baumannii strains collected from the Instituto Nacional de Cancerología (Mexico’s National Institute of Oncology), a tertiary hospital located in Mexico City. Namely, GCA_004299615.1 (phi5013-M1 and phi5013-M2), GCA_003522845.1 (phi9102), GCA_004321575.1 (phi4197), GCA_004794205.1 (phi5038-11536, phi5038-11551), and GCA_001922695.1 (phi11547). In brief, the strains were treated with mitomycin C at a final concentration of 1 μg/ml in LB broth at 37°C for an overnight incubation period. Subsequently, the supernatants were collected via centrifugation (12,000 × g for 10 min) and utilized for host-range assays. We used the previously reported strains to propagate the phages (López-Leal et al., 2021). Specifically, 10 μl of each supernatant fraction was applied to overlay plates containing 3 ml of soft agar, followed by the addition of 300 μl of bacterial cells (López-Leal et al., 2021; Kropinski et al., 2009). After incubating overnight at 37°C, the plates were scrutinized for lysis within the spotted region. Confirmation of plaque formation in susceptible strains was achieved using a previously described method (López-Leal et al., 2021; Hyman, 2019; Santamaria et al., 2014). To summarize, 100 μl of a bacteriophage solution (or a dilution from the stock) was added with 200 μl of previously cultured, susceptible A. baumannii cells. The mixture was incubated for 15 min at room temperature, after which it was mixed with soft agar and spread onto LB solid medium to create a bacterial lawn. The plates were incubated at 37°C, and the formation of lytic plaques was observed. Individual plaques were picked and subjected to three consecutive rounds of replating to ensure the purity of the bacteriophage stocks. Different plaque morphologies were also considered for bacteriophage isolation. To ensure reproducibility, all experiments were conducted in triplicate. The bacteriophages were then cultivated in 6-ml cultures of A. baumannii strains (GCA_001922705, GCA_001922745, and GCA_001922695) in LB medium (with an OD 620 nm of 0.1), supplemented with 100 μl of each bacteriophage (average of 105 PFUs/ml). DNA isolation procedures followed established protocols (López-Leal et al., 2021; Santamaria et al., 2014), and restriction enzyme digestion analysis was performed using HindIII, EcoRI, EcoRV, BamHI, and NdeI. Those phages that generated a unique restriction patterns were selected for sequencing. Phage genomic DNA sequencing was performed using a Miseq Illumina platform with a 300-bp paired-end configuration with a TruSeq DNA library. Raw reads were trimmed using TRIM_ GALORE¹ with a quality threshold of ≥30. The genome assembly was performed using the A5 pipeline (Coil et al., 2015) with a set of ~25,000 (average of 7,215,608 reads per phage genome) randomly selected paired-end reads. Then, CheckV (v1.0.3; Nayfach et al., 2021) was used to validate the quality of each assembly. The completeness was confirmed by Bandage (v0.8.1; Wick et al., 2015), using the .gfa file.

To put our induced prophages into a genomic context with other A. baumannii prophages, we used a previously characterized and reported database of 1,465 A. baumannii isolates (Hernandez-Gonzalez et al., 2022). The genomes in this database were determined as high-quality genomes according to CheckM (v1.0.2; Parks et al., 2015). BioSample information for all A. baumannii genomes was obtained using efetch from E-utilities (v16.2; Sayers, 2009). Metadata was collected for the following sections: source isolate and isolation site.

Prophage predictions were carried out using VirSorter2 (v2.2.4; Guo et al., 2021), and CheckV was used to determine the completeness and quality of phages and prophages sequence. Only phage sequences assigned as High-quality or Complete by checkv-quality or medium-quality were considered for downstream analysis. We followed the publicly available protocol for the validation of the first-instance prophage prediction.² Briefly, the final quality of prophages analyzed by CheckV were validated in a second screening using VirSorter2 (Guo et al., 2021). All (pro)phage genomes were annotated using pharokka (v1.6.1; Bouras et al., 2023).

In the first instance, all prophages and phages (all the Acinetobacter phages reported in the NCBI) with at least 80% nucleotide similarity in at least 80% of the genome length were binned by cd-hit (v4.8.1; Fu et al., 2012). Following the parameter proposed by the ICTV (International Committee on Taxonomy of Viruses), this first clustering allowed us to select those phages and prophages that belonged to the same genus (Simmonds et al., 2023; Turner et al., 2021). In this sense, groups (phage genera) with more than five members were selected for further analysis. Then, to construct homologous groups (HG) from the selected bacteriophage genomes, we ran roary (v3.13.0; Page et al., 2015), setting the BLAST search parameters to a length coverage of ≥70% and an amino acid sequence identity of ≥40%. Next, a second clustering was made based on the optimal number of clusters determined by the hierarchical clustering of viruses based on intergenomic distances calculated from their shared protein content (pan-matrix from roary). The optimal number of clusters was determined by the average-silhouette method in R. Subsequently, hierarchical trees were constructed for each cluster with 100 bootstrap replicates. The ape (v5.6.1) and ggtree (v 3.0.4) libraries in R were used to obtain the branch length distribution. Finally, phage species and phage genera were validated using VICTOR (Meier-Kolthoff and Goker, 2017).

The relationship between phage abundance and geographical location was evaluated. First, we calculated the uncertainty coefficient between bacteriophage genera and species against geographic locality. Second, the correlation between the abundance of phage species and phage genus in the different countries was computed with Pearson tests. The phage species and phage genus were taken as the result of the VICTOR analysis (Meier-Kolthoff and Goker, 2017).

In previous studies, we reported that several ST-758 A. baumannii isolates have a high abundance and diversity of prophages (López-Leal et al., 2020). Furthermore, some of these phages belonging to the genus Vieuvirus were found to be active, since they could cause infections and generate progeny in other strains of the same lineage (López-Leal et al., 2020; López-Leal et al., 2021). In this study, in order to expand our understanding regarding the induced prophage sequences and the prophage population that infects A. baumannii, we induced seven prophages from previously reported isolates (see Methods). The phages genomes of phi9102, phi4197, phi5013-M1, phi5013-M2, phi5038-11536, phi5038-11551, and phi11547 were assembled as single contigs of 52,239 base pairs (bp), 39,359, 42,180, 49,159, 48,056, 38,492 and 25,802 bp, respectively. The assembled genomes were assigned as complete and high-quality genomes according to CheckV. No tRNA or ARGs were identified. All induced prophages were similar to the reported Ab11510-phi phage (López-Leal et al., 2021; Supplementary material 1). Interestingly, strains Ab5013 and Ab5038 harbored two genomically similar phages of the genus Vieuvirus (Supplementary material 2).

To obtain a better overview of the seven isolated phages in the context of the A. baumannii prophage populations, we used a database with 1,501 high-quality A. baumannii genomes, according to CheckM (Parks et al., 2015; Supplementary material 3). In an initial analysis, 4,865 prophages were collected. However, only 967 (19.87%) of the prophage predictions were assigned as high-quality by CheckV (Supplementary material 4). To place the seven induced prophages into a genomic and phylogenomic context relative to the prophage population in A. baumannii, only high-quality predictions were used, as this approach was considered to provide more reliable results given that prophages are prone to degradation processes (Bobay et al., 2014).

These 928 final prophages represented an average of 1.67 prophages per genome. Additionally, we add 21 putative Vieuvirus genomes (Supplementary material 5), previously reported (Rastegar et al., 2024), in order to put the induced and isolated prophages (in this study) in context with other previously reported phages of the genus Vieuvirus and with prophage populations from A. baumannii isolates. Finally, we built an A. baumannii phage and prophage genome database with 995 genomes. Of these, we found that 30.02% of phages and prophages were singletons (see Methods), and 69.98% were grouped into 119 clusters using cd-hit. Of these, only phages that clustered with ≥5 others were selected to construct a gene content correlation matrix. This threshold was chosen to focus on the most prevalent phages, which could be grouped into potential genera, providing a clearer understanding of gene content similarity within more broadly distributed phage populations (Figure 1). Furthermore, we observed that the clustering shown in the heat map did not correlate with the groups found in the first level of clustering by cd-hit (see clusters in the top rows of Figure 1). This result suggested that mosaicism of bacteriophage genomes presents challenges to describing phage-relatedness (Gauthier and Hatfull, 2023). In other words, the clustering of gene content (homologous proteins) retrieved by the heat map did not reflect the genome-level clustering obtained by cd-hit. On the other hand, the heat map based on the gene content correlation matrix showed a considerable variation in gene content among the phages, resulting in 13 clusters (50.57% of the phage database). However, we observed some degree of relationship in gene content between phages from clusters 1, 2, 3, 6, and 9–11 (Figure 1, red rectangle). Next, we used the pangenome of the phages to perform a principal component analysis (PCA) based on the pan-matrix. Although only 45.5% of the total variation was captured by the first two components (out of 482 dimensions, Supplementary material 6), the first 10 dimensions accounted for 85.6% of the variance in our data. This suggests that the remaining variation is likely due to the presence of unique genes in each phage, which are not homologous to those in others. Two phage populations were delimited on the first principal (horizontal) axis. Namely, the phages from clusters 4, 5, 7, 8, 12, and 13 were placed within the PC1 component values >0, and the rest were located within the PC1 values <0 (Supplementary material 6). This indicates that the phages of clusters 4, 5, 7, 8, 12, and 13 do not share homologous genes with the rest of the phages or share very few homologous genes. On the contrary, phages from clusters 1, 2, 3, 6, and 9–11 (Figure 1, red rectangle) shared some or several homologous groups (see methods). To further elucidate this, we used a gene content distance matrix to construct a hierarchical clustering tree based on gene content (Figure 2) that only included phages from clusters 1, 2, 3, 6, and 9–11 and rooted the phylogeny using phages from Cluster 5 a closely related genus.

The first thing we noticed is that most of the clades defining the clusters (Figure 2) were well supported (>80 bootstrap values; blue circles). However, clustering boundaries between clusters 1 and 6 had weak support (bootstrap of 45). Interestingly, all members of cluster 6 correspond to prophage predictions, while phages located in cluster 1 correspond to prophage predictions and isolated phages previously reported in the NCBI (assigned as Vieuvirus). On this basis, we have tentatively separated these predominantly temperate A. baumannii phages into two clusters (Clusters 1 and 6). However, we note that this assignment was poorly supported by the phylogenetic analysis (Figure 2). Interestingly, shorter branches were observed in the Cluster 6 clade (Figure 2). Therefore, we analyzed the distribution of branch lengths for both clusters. Both distributions showed that short branches predominated in both clusters (Supplementary material 7). However, Cluster 6 (Supplementary material 7B) seems to have a higher concentration of extremely short lengths compared to Cluster 1 (Supplementary material 7A), suggesting that the prophages (Cluster 6) have a higher proportion of homolog groups or proteins very similar in sequence so they are even more closely related or have experienced less genetic divergence compared to those in Cluster 1. This indicates that the induced prophages have less HG in common; therefore, they may experience more divergence in their protein repertoire.

Finally, each cluster was validated using the VICTOR tool. Most of the clusters were validated by VICTOR at the genus level, except for phages AbTJ (Xu et al., 2020) and Ab105-1phi, which have been previously reported to share some genetic repertoire with phages of the Vieuvirus genus (López-Leal et al., 2021). However, according to the NCBI, these phages are unclassified Caudoviricetes. Another exception was for prophages P1576, P1697, P1703, P2118, and P2344 from Cluster 7 (Supplementary material 8), which showed a small degree of similarity with phage AM106 (MH115576), with a coverage of 9–15% and identity of ~90%. In other words, our approach, using the correlation of gene content variation, succeeded in grouping these phages appropriately at the genus level.

With a total of 482 bacteriophages analyzed (phages and prophages from the 13 clusters), we were able to identify and validate 97 phage species and 15 phage genera (Supplementary material 8). We then determined the sequence type and isolation location of the hosts of the 467 prophages (only prophages from the 13 clusters). This collection of A. baumannii isolates comprised 53 different STs (according to Pasteur’s MLST scheme) from 22 countries. Of these, 23.56% of the isolates primarily corresponded to 1806 STs, and 41.69% were from China (Supplementary material 9). Then, in order to identify how widely the bacteriophage genera and species were geographically spread, we first determined the uncertainty coefficient between bacteriophage genera and species. First, we determined the uncertainty coefficient between the genus-species relationships against geographic locality (country) and host ST to determine the relationship strength between these two categorical variables. We found that the mutual information was 1.27 and 0.67, for the species-country and genus-country relationships, respectively (Table 1). These values indicate that the variables share certain information, suggesting a dependence or relationship between them. Specifically, the relationship between species and country is stronger than the relationship between genus and country. Moreover, the residual uncertainty in species was 66.58% after knowing the country, whereas for genus, it was 35.11%. In addition, the residual uncertainty for the country after knowing the genus and species was 24.46 and 31.98%, respectively (Table 1). In other words, residual uncertainty is higher when species or genus is conditioned on country than vice versa, indicating that country has a considerable effect in reducing uncertainty about species or genus. Of these, phage species provide more information about the country and vice versa, compared to genus. With this information, we wanted to see the correlation between the co-occurrence of genera and phage species and the isolation country. A strong positive correlation was observed between the frequency of different species due to their co-occurrence in the same location. Of 73 phage species only located in China (28 species), Thailand (23 species), Mexico (10 species), South Korea (7 species), and Spain (5 species; Figure 3A). Of these, 26.38% of the species corresponded to the Vieuvirus genus (G2-Cluster 1; Figure 3B), followed by 11.11% of species for an unclassified genus (G4-Cluster 3), which are similar to the previously reported phage fLi-Aba03 (Rastegar et al., 2024). The rest of the species (with an average of 5.20% of species per genus) belonged to different genera. Interestingly, co-occurrence of phage species is mainly related to the Vieuvirus (Figure 3B).

Although these species belong to different genera, however, it is important to note that phages of genus G2 (Cluster 1; Figure 1) were found more frequently in Thailand, Mexico, Canada, and Iran, while phages of genus G7 (Cluster 6; Figure 1) were found mainly in China, Thailand, Canada, and South Korea (Supplementary material 8). Based on our results (Supplementary material 10), these two groups of phages could be two types of Vieuvirus located mainly in North America and South Asia (Cluster 2) and East Asia (Cluster 6). Additionally, the G15 genus, composed of a small number of phages (Cluster 13), was found mainly in isolates from Mexico (Supplementary material 11).