Polysaccharide lyases, including pectin and pectate lyases, are enzymes that cleave (1,4)-glycosidic bonds through a β-elimination mechanism (Sutherland, 1999). The predominant functional domain identified in the analyzed tail fibers is the pectin lyase-like domain (PLD), which is present in 30 tail fibers. This domain is exclusive to Ackermannviridae (1), Autographiviridae (22), and myoviruses (seven) (Table 2), which are associated with depolymerase activity studied in various recombinantly expressed proteins (Liu et al., 2019b; Oliveira et al., 2019; Popova et al., 2020a; Abdelkader et al., 2022).

Tail fibers containing PLD exhibit diversity, with an average pairwise identity of 19.2% and lengths ranging from 693 to 921 amino acids (AAs). Among these tail fibers, 23 phages with a PLD domain were explicitly documented to exhibit halo formation around their plaques (Table 1). Although the majority of tail fibers share a common PLD SUPERFAMILY (SSF51126), phage Acibel007 (gp46) deviates with only a Gene3D hit (2.160.20.10; Pectin_lyas_fold) (Table 2). Some variations involve Pfam, SMART, or CDD hits, with 10 tail fibers overlapping a Pfam hit, eight featuring Pectate_lyase_3 (PF12708), and two labeled as Beta_helix (PF13229) for myovirus phage AB1 (gp76) and Autographiviridae phage P2 (gp48). Additionally, four tail fibers contain a SMART PbH1 (SM00710) hit, corresponding to parallel beta-helix repeats in pectate lyases and rhamnogalacturonase A.

An alternate domain architecture features a PLD SUPERFAMILY (SSF51126) overlapping a Gene3D (2.160.20.10) and a CDD pyocin_knob hit at the CTD (cd19958) (Table 2). This layout is present in the tail fibers of two podoviruses (Abp1 gp47 and IME-AB2 gp71) and one myovirus (Petty gp39) (Table 2). Phage proteins sharing this domain range from 21.9% to 29.6% identity (Supplementary Table 2). Finally, Autographiviridae phage B1 gp45 encoded the only tail fiber with both a PLD SUPERFAMILY hit and a phage_tailspike_middle hit from the CDD database (Table 2). A phylogenetic tree illustrates that tail fiber proteins do not group based on capsule targets but mainly on viral morphology (Figure 2). Generally, the myoviruses group together except for SH-Ab 155599 and IME-AB2, which group with podoviruses.

The tail fibers of six phages feature a hit to the phage_tailspike_middle domain from the CDD (Table 3). This model characterizes the middle beta-helical domain of Acinetobacter bacteriophage tail spike proteins, encompassing a distinct N-terminal domain unrelated to the beta-helical substructure but implicated in virion binding. The C-terminal domain, highly variable, is suggested to play a role in receptor binding. Among these phages, AM24, IME285, BS46, and WCHABP12 (Tables 1, 3) exhibit halo formation around their plaques, with recombinantly expressed tail spike proteins from AM24 (gp50), BS46 (gp47), and IME285 (AYP68900.1) that are confirmed as functional depolymerases (Popova et al., 2019; Knirel et al., 2020; Wang et al., 2020).

Phages featuring tail fibers with the phage_tailspike_middle domain are confined to myoviruses and Autographiviridae lineages. The six identified proteins, ranging from 732 to 848 amino acids, share a pairwise identity of 60.1% (Supplementary Table 3). These proteins cluster by taxonomy, with a high % identity observed within myovirus-derived tail fibers (WCHABP12, gp16; IME285, AYP68900.1; BS46, gp47; and AM24, gp50) at 72.4–97.4% (Figure 3, Supplementary Table 2). Autographiviridae proteins B1 (gp45) and B5 (gp47) share 40.7% identity (Figure 3, Supplementary Table 3). Intriguingly, B5′s tail fiber exhibits greater similarity to myovirus phage tail fibers, with % identity ranging from 69.1% to 73.2%. The most diverse tail fiber in this group belongs to B1, featuring a SUPERFAMILY pectin lyase-like hit (SSF51126) overlapping the CDD domain (Figure 3, Table 3). This 761-AA-long tail fiber shares 22.5–40.7% identity with other tail fibers in this group.

The CATH superfamily 2.60.40.3940 is characterized by a predominantly beta class (2), sandwich architecture (2.60), and immunoglobulin-like topology (2.60.40) within a novel homologous superfamily (2.60.40.3940). In the Obolenskvirus clade of Acinetobacter myophages, nine phages, namely, AP22 (gp53), AB1 (gp77), YMC-13-01-C62 (gp46), YMC11/12/R2315 (gp84), YMC11/12/R1215 (gp20), WCHABP12 (gp15), WCHABP1 (gp6), AbP2 (gp18), and Abp9 (gp49), feature the G3DSA 2.60.40.3940 domain in the CTD of one of their tail fiber proteins (Table 4). These proteins, encoded in the morphogenesis region and positioned upstream of other tail fiber proteins, vary from 258 to 283 amino acids, sharing an overall pairwise identity of 83.0% (Supplementary Table 4). An analysis of the multisequence alignment reveals high sequence conservation at the N-terminal domain (NTD), indicating its involvement in virion attachment. Conversely, the CTD, associated with the 2.60.40.3940 hit, displays a breakdown in sequence identity among some proteins, suggesting a potential role in host cell recognition. This role is supported by the crystal structure of Acinetobacter phage AP22 gp53 CTD (PDB: 4MTM), resembling a homotrimeric globular lectin-like protein with a slender midsection, akin to host-cell binding proteins of Escherichia coli phages T7 gp17 (Garcia-Doval and Van Raaij, 2012) and T4 gp12 (Van Raaij et al., 2001). Furthermore, AP22 gp53C binds to ethylene glycol and glycerol molecules, which are surrogates of an oligosaccharide backbone. Overall, this data suggests that tail fibers encoding a 2.60.40.3940 domain are involved in recognizing oligosaccharide moieties of Acinetobacter baumannii.

The SGNH hydrolase superfamily comprises 16 well-studied protein families with a conserved catalytic fold and mechanism (Anderson et al., 2022). These enzymes, named after their catalytic Ser, His, Gly, and Asn residues, function as esterases and lipases, playing vital roles in biomass conversion, pathogenesis, and cell signaling (Akoh et al., 2004; Anderson et al., 2022). The SGNH hydrolase domain was identified in the tail fibers of 10 phages: SH-Ab 15497, fHyAci03, KARL-1, PhT2, fEg-Aba01, fLi-Aba02, fLi-Aba03, DMU1, PD-6A3, and AbTZA1, with a pairwise identity of 33.7%. Four of these phages exhibited halo formation (PD-6A3, fEg-Aba01, fLi-Aba02, and fLi-Aba03) (Wu et al., 2019; Badawy et al., 2020) (Table 5, Supplementary Table 5). Based on the structural organization of the domains, the tail fibers can be grouped into two architectures. One group features two fibritin domains: one at the NTD and one directly upstream of the SGNH hydrolase domain (Table 5). This domain layout is restricted to four members of the subfamily Tevenvirinae: vB_ApiM_fHyAci03 (fHyAci03), KARL-1, vB_AbaM_PhT2 (PhT2), and AbTZA1, which tend to group together (Figure 4). Fibritin belongs to a class of chaperones that catalyze specific phage-assembly processes, promoting the assembly of the long tail fibers and their attachment to the tail baseplate (Tao et al., 1997). Furthermore, fibritin also serves as a sensing device, controlling the retraction of the long tail fibers in adverse environments to prevent infection (Tao et al., 1997). These four proteins with this domain layout have similar lengths, ranging from 621 to 626 AA and share between 55% and 97.6% AA identity (Table 5 and Supplementary Table 5).

The second group comprises hypothetical proteins from six phages, featuring a lone SGNH domain at the CTD of the protein (Figure 4, Table 5). This domain structure is present in one Autographiviridae member, PD-6A3, and five siphoviruses (SH-Ab 15497, fEg-Aba01, fLi-Aba02, fLi-Aba03, DMU1), which group more closely together (Figure 4). PD-6A3 encodes an endolysin with activity against A. baumannii cells, which could potentially be responsible for the observed halo formation (Wu et al., 2019). In contrast, siphophages, fEg-Aba01, fLi-Aba02, and fLi-Aba03, encode two putative tail fiber proteins: one with an SGNH hydrolase domain and, in the case of fLi-Aba02 and fLi-Aba03, a Concanavalin A-like lectins/glucanases domain. The siphovirus proteins are very close in length, ranging from 620 to 660 AA, sharing 41 to 100% AA identity (Table 5, Supplementary Table 5). Phages with 100% AA identity are fEg-Aba01, fLi-Aba02, and fLi-Aba03. Comparatively, the podophage PD-6A3 tail fiber protein is significantly longer at 817 AA, although it shares approximately 41% identity with the five siphoviruses of the same layout (Figure 4, Table 5, Supplementary Table 5).

Since our data collection, the Acinetobacter podovirus Aristophanes was published. This phage does not produce a halo but encodes a tail spike SGNH hydrolase domain (gp41) (Timoshina et al., 2021). Functional study of this protein revealed a tail deacetylase causing O-acetylation of one of the K26 sugar residues which causes a slight decrease in turbidity of the host (Timoshina et al., 2021). This finding suggests that the other phages may also utilize this structural protein as a deacetylase. To get more information on the potential functions of the phages DMU1, PD-6A3, and SH Ab 15497, the proteins were modeled with AlphaFold Multimer as homotrimers, and the resulting models were submitted to the DALI server PDB search. The top hits are to a xyloglucan-active beta-galactosidase from Xanthomonas citri (7KMM) for all models, followed by a protein with an unknown function from Arabidopsis thaliana (2APJ) for DMU1 and PD-63A and a homo-dimer acetylxylan esterase from Clostridium acetobutylicum (1ZMB) for SH Ab 15497.

Galactose binding domains are present in several different protein families in eukaryotes and prokaryotes and bind to specific ligands, such as cell-surface-attached carbohydrates (Ito et al., 1991). The members of this domain exhibit a β-sandwich forming a jelly roll fold. In the tail fibers of six myophages (ME3, TAC1, Aci05, Aci02-2, Aci01-1, and Ab_121), a galactose-binding domain was identified (Figure 5, Table 6). Notably, literature reports indicate that five of these phages do not exhibit halo formation (Lee et al., 2011; Essoh et al., 2019; Asif et al., 2020). Tail fibers encoding this domain varied significantly in sequence length, spanning from 1,177 to 5,419 AA (Table 6), with diverse percent amino acid identity ranging from 12.9 to 98.8% (Supplementary Table 6). ME3, the first jumbo Acinetobacter phage identified, presented the most divergent tail fiber with an average % identity ranging from 12.4–13.2 % (Figure 5, Supplementary Table 6) (Buttimer et al., 2016). The tail fiber protein encoded by ME3 is 5,419 AA long and features multiple hits from Gene3D, SUPERFAMILY, Pfam, and PROSITE profiles (Figure 5, Table 5).

The second most divergent tail fiber belongs to TAC1, displaying an identity range of 13–42.5% (Figure 5, Supplementary Table 6), containing a myosin heavy chain Panther hit (PTHR18921) overlapping a Gene3D hit (1.10.287.1490, Phosphatidylinositol 3-kinase regulator activity). TAC1′s tail fiber also included two additional Gene3D hits corresponding to a galactose-binding domain (2.60.120.260) and a myosin heavy chain (1.20.5.340). Phages Ab_121, Aci01-1, and Aci02-2 encode a 2,211 AA tail fiber protein with an identity range of 96.8% and 98.9% (Supplementary Table 6). Aci01-1 and Aci02-2 have matching Panther (A-type inclusion protein), Gene3D (galactose binding domain), and SUPERFAMILY (galactose binding domain) hits spanning the length of the protein (Figure 5, Table 6). Phage Ab_121 tail fiber has the same Gene3D hits but differs in the location of the Panther and SUPERFAMILY hits. Phage Aci05 also has similar domain hits to phages Ab_121, Aci01-1, and Aci02-2, although it lacks a Panther and one of the Gene3D hits.

Four phages were identified to encode a lysozyme domain in their tail fibers (Table 6). Among them, three are unclassified Tevenvirinae myoviruses (ZZ1, gp162; fHyAci03, gp165; and KARL-1, gp124) while the remaining phage is an unclassified podovirus (Presley, gp80). Notably, only the characterization article of KARL-1 discussed halo formation, which was not observed previously (Jansen et al., 2018). Presley shares 12% identity to the myovirus proteins (Supplementary Table 7).

The most highly annotated proteins with the lysozyme domain belong to Tevenvirinae, reflecting the extensive research on the classical coliphage T4 (Table 7). The sequence length of these proteins varies from 593 to 600 AA long with notable conservation in percent identity ranging from 83.4% and 99.7% (Supplementary Table 7). All three proteins hit needle_T4 from the HAMAP database (Figure 6). Furthermore, each phage has three distinct SUPERFAMILY database hits corresponding to gp5 N-terminal domain-like (SSF69255), lysozyme-like (SSF53955), and phage fiber proteins (SSF69349) (Figure 6, Table 7). Three Gene3D hits are present on all proteins (2.40.50.260, 1.10.530.40, and 3.10.450.190), as well as two Pfam hits (Gp5_OB, PF06714 and Phage_lysozyme, PF00959). The lysozyme functionality of these proteins is supported by a Panther hit (T4-type lysozyme 1-related, PTHR37406), CDD hit (T4-like_lys, cd00735), and a PRINTS hit (T4 lysozyme, PR00684).

The analysis of the phage tail fibers also revealed nine unique functional domains found in three Autographiviridae, four myoviruses, and five siphovirus phages; of which, seven were reported to have halo formation: fEgAba01 (gp20), fLiAba02 (gp22), fLiAba03 (gp22), Aci07 (gp45), TaPaz (gp78), APK116 (gp43), and AS12 (gp42) (Table 8) (Popova et al., 2017, 2020a; Essoh et al., 2019; Badawy et al., 2020; Shchurova et al., 2021). Further analysis will be focused on the abovementioned phages due to their halo formation and in the case of AKP116, AS12, and TaPaz, the expression of a recombinant depolymerase enzyme. The six remaining phage tail fibers with unique domains belong to Acibel004 (gp123; 2201 AA), ME3 (gp106; 574 AA), TRS1 (gp30; 526 AA), fHyAci03 (gp247; 1259 AA), R2096 (gp66; 414 AA), and ZZ1 (gp171; 510 AA) (Table 8).

The tail fiber proteins from the podovirus AS12 (gp42, 901 AA) and the myovirus TaPaz (gp78, 878 AA) have peptidase domain hits at their CTD (Table 8). The functional domains of these tail fibers are of special interest because they have already been characterized as depolymerases (Popova et al., 2017; Shchurova et al., 2021). A PROSITE profiles hit (PS51688) was identified, which is an intramolecular chaperone auto-processing (ICA) domain that can catalyze the trimerization-dependent auto-proteolysis using two conserved serine and lysine residues. This domain has been identified in bacteriophage-encoded endosialidases and tail spike and fiber proteins (Schwarzer et al., 2007). The protein domain responsible for endosialidase activity in the ICA domain-containing tail fibers is restricted to the NTD of the proteins (Schwarzer et al., 2007). The final shared database hit overlaps the PROSITE profiles hit and is from the CDD database (cd10144) to the Peptidase S74 protein family of known phage endosialidases (Table 8). TaPaz gp78 has a Pfam Peptidase S74 hit (PF13884), while AS12 differs with an overlapping Gene3D hit (1.10.10.10), which is found with winged helix DNA-binding proteins (Table 8).

Phages APK116 (gp43, 861 AA) and PhiAB1 (gp41, 882 AA) both contain a lone pyocin knob domain at the CTD from CDD (cd19958) (Table 8). This domain layout is similar to those of Friunavirus members discussed above, except it lacks a pectin-lyase domain hit. Phage APK116 was documented to have halo formation, and gp43 was recombinantly expressed and shown to function as a depolymerase (Popova et al., 2020a). APK116 gp43 and PhiAB1 gp41 were modeled with AlphaFold and the models were submitted to the DALI server (Figure 7). The top DALI models for these two proteins are Acinetobacter phage AS12 depolymerase gp42 (6EU4) (Table 8), followed by the E. coli CAB120 depolymerase tail spike protein (6W4Q) for phiAB1 and poly(beta-d-mannuronate) c5 epimerase from Azotobacter vinelandii (5LW3) for APK116.

Two siphophages, fLiAba02 and fLiAba03 (gp22; identical proteins), were found to be the only phage tail fibers to have a Concanavalin A-like lectins/glucanases (CALG) SUPERFAMILY (SSF49899) domain hit (Table 8). Lectins are described as non-immune origin proteins possessing binding affinity toward glycoconjugates in a specific and reversible manner. As discussed above, these phages have an additional tail fiber encoding an SGNH hydrolase. To further investigate the function of the CALG domains, other characterized depolymerases were searched. Paenibacillus sp. 32352 is a soil-dwelling bacterium that produces the enzyme Pn3Pase, which degrades the capsular polysaccharide of Streptococcus pneumoniae serotype 3 (Pn3P) (Middleton et al., 2018; Wantuch et al., 2021). This protein encodes two SUPERFAMILY domain hits of interest: CALG (SSF49899; CTD) and (trans)glycosidases (SSF51445; NTD). Transglycosylases are a class of glycosyl hydrolase enzymes that can catalyze the transformation of one glycoside to another (Romero-Téllez et al., 2019). The functional domain responsible for the depolymerization of Pn3P was found to be the (trans)glycosidase domain, as knockouts of this region result in the loss of depolymerase activity compared to the loss of the CALG domain (Wantuch et al., 2021). The findings of the Pn3Pase mutation experiment can further be applied to phage Aci07 (gp45). As mentioned above, this phage has also documented halo formation. This is the only tail fiber with a hit to (trans)glycosidases from the SUPERFAMILY database (SSF51445) (Table 8). This is the same SUPERFAMILY domain as the Pn3Pase depolymerase discussed above, which suggests that gp45 of Aci07 may be a functional depolymerase.

Four phage tail fibers with no domain hits or information on halo formation will not be investigated further: ABP2 gp17, TRS1 gp29, Acibel004 gp148 and IME-AB2 gp72, and PD-AB9. Six phage tail fibers had no domain hits present, although their function as depolymerases was experimentally confirmed by researchers studying their recombinant proteins. These phages include AP22 (gp54), APK32 (gp46), APK48 (gp43), P1 (gp43), B9 (gp69), and TaPaz (gp79) (Oliveira et al., 2017, 2018; Knirel et al., 2020; Popova et al., 2020a; Shchurova et al., 2021). Furthermore, halo formation was detailed in the following phages but the protein responsible was not confirmed: AbTj (gp53) and WCHABP1 (gp5) (Zhou et al., 2018; Xu et al., 2020; Drobiazko et al., 2022). All the eight abovementioned proteins were modeled as homotrimers with AlphaFold Multimer and showed narrow midsections and larger globular regions (Figure 7).

The protein models were submitted to DALI PDB search to identify similar structural models (Figure 7, Supplementary Table 8) (Jumper et al., 2021; Holm, 2022). The top P1 DALI hit is to its own crystal structure (6E1R), followed by the putative pectin lyase gp18 (7CHU) of the Geobacillus virus E2 (Table 8). APK32 top DALI model is to Acinetobacter phage AS12 gp42 (6EU4), the characterized depolymerase discussed in the above section documenting rare domains (Table 8). The AP22 DALI hit is to its own structure (4Y9V), followed by hits to the O-specific polysaccharide lyases of Pseudomonas phages LKA1 (4RU4) and phi297 (4RU5) (Supplementary Table 8). The B9 and APK48 depolymerase models hit is to a poly(beta-d-mannuronate) c5 epimerase from Azotobacter vinelandii (5LW3 and 2PYH) and the putative pectin lyase gp18 (7CHU-A) of the Geobacillus virus E2. The TaPaz top hit is to the E. coli phage HK620 tail spike depolymerase (4XLA) and the poly(beta-d-mannuronate) c5 epimerase from Azotobacter vinelandii (5LW3). The top AbTj AlphaFold model DALI hit is to Acinetobacter phage phiAB6 tail spike depolymerase (5JSD), suggesting that this protein is responsible for the halo formation documented with the AbTj plaques. Similarly, the WCHABP1 depolymerase was modeled to a glycan biofilm modifying enzyme from Pantoea stewartii (6TGF), followed by a hit to the phiAB6 tail spike. These findings highlight a breakdown in the amino acid sequence of the proteins, but a conservation in structure, which has led to the missing functional domain hits of these proteins. The use of AlphaFold Multimer to model the proteins, and the DALI server for PDB search of the resulting models shows the power of these methods for investigating the function of tail fiber proteins lacking domain hits to investigate their potential functions.