Applied Microbiology and Biotechnology (2023) 107:3621–3636 https://doi.org/10.1007/s00253-023-12547-8 APPLIED GENETICS AND MOLECULAR BIOTECHNOLOGY CkP1 bacteriophage, a S16‑like myovirus that recognizes Citrobacter koseri lipopolysaccharide through its long tail fibers Hugo Oliveira1,2 · Sílvio Santos1,2 · Diana P. Pires1,2 · Dimitri Boeckaerts3 · Graça Pinto1,2 · Rita Domingues1,2 · Jennifer Otero1,4 · Yves Briers3 · Rob Lavigne5 · Mathias Schmelcher6 · Andreas Dötsch7 · Joana Azeredo1,2 Received: 27 December 2022 / Revised: 11 April 2023 / Accepted: 17 April 2023 / Published online: 3 May 2023 © The Author(s) 2023 Abstract Citrobacter koseri is an emerging Gram-negative bacterial pathogen, which causes urinary tract infections. We isolated and characterized a novel S16-like myovirus CKP1 (vB_CkoM_CkP1), infecting C. koseri. CkP1 has a host range covering the whole C. koseri species, i.e., all strains that were tested, but does not infect other species. Its linear 168,463-bp genome con- tains 291 coding sequences, sharing sequence similarity with the Salmonella phage S16. Based on surface plasmon resonance and recombinant green florescence protein fusions, the tail fiber (gp267) was shown to decorate C. koseri cells, binding with a nanomolar affinity, without the need of accessory proteins. Both phage and the tail fiber specifically bind to bacterial cells by the lipopolysaccharide polymer. We further demonstrate that CkP1 is highly stable towards different environmental conditions of pH and temperatures and is able to control C. koseri cells in urine samples. Altogether, CkP1 features optimal in vitro characteristics to be used both as a control and detection agent towards drug-resistant C. koseri infections. Key points • CkP1 infects all C. koseri strains tested • CkP1 recognizes C. koseri lipopolysaccharide through its long tail fiber • Both phage CkP1 and its tail fiber can be used to treat or detect C. koseri pathogens Keywords Citrobacter · Bacterial infection · Bacteriophage · Long tail fiber · Control · Diagnostics Introduction other antibiotics (e.g., ciprofloxacin, cefuroxime, aztreonam, ceftazidime, gentamicin) by chromosomal or mobile genetic Citrobacter species are Gram-negative bacilli that belong determinant elements, thus creating a need for alternative to the Enterobacteriaceae family. The genus is composed therapies (Rizvi et al. 2010; Aruna and Mobashshera 2012). of about a dozen species, but only Citrobacter freundii, Cit- Bacteriophages (phages) have emerged as a possible solu- robacter koseri (formerly named Citrobacter diversus), and tion to control drug-resistant pathogens linked to human Citrobacter amalonaticus are pathogenic to humans. In par- infectious diseases (Melo et al. 2020). Broad host range ticular, C. koseri accounts for a variable, yet large, portion phages can be more advantageous to control pathogens of Citrobacter infections (ranging from 20 to 90%), often compared to narrow-host-range phages, avoiding the need associated with urinary tract infections (Ranjan and Ranjan to identify the causative agent prior to treatment. Phages of 2013; Deveci and Coban 2014). C. koseri strains are natu- the recently defined Straboviridae family and Tevenvirinae rally resistant to ampicillin but can rapidly gain resistance to (T-even) subfamily, which have a myovirus morphotype, have been found as a good source of broad-spectrum viral candidates targeting different bacterial pathogens, at both * Hugo Oliveira species and genus levels. For instance, phage T4 is reported hugooliveira@deb.uminho.pt to have a relatively broad host range within the Escheri- * Joana Azeredo chia coli species, whereas phage S16 exhibits broad host jazeredo@deb.uminho.pt range within the Salmonella genus (Marti et al. 2013). There Extended author information available on the last page of the article are also other T-even broad-host-range reported against 13 Vol.:(0123456789) 3622 Applied Microbiology and Biotechnology (2023) 107:3621–3636 Salmonella, for instance, phage STP4-a (Li et al. 2020) and C. freundii (n = 21), C. amalonaticus (n = 1), C. youngae phage SHWT1 (Tao et al. 2021). (n = 1), and C. braakii (n = 1), as well as representatives of A common feature to all T-even phages is that their host other closely related species. For the clinical isolates of the interaction is mediated by an initial, reversible binding with Hospital de Braga, human samples were first cultured in agar the long tail fiber (LTF), followed by a second and irrevers- plates, including blood agar, chocolate blood agar, CLED ible binding with the short tail fibers. The reversible binding agar, or MacConkey agar, from which Citrobacter strains defines the phage host range. In phages T4, T2, and S16, were isolated after 18 to 24 h of incubation at 35 °C. Typing gp34 to gp38 are involved in the production of LTF proteins, was performed using MALDI-TOF MS or Gram-negative from proximal to distal segments (King and Laemmli 1971; identification cards of Vitek2 (bioMérieux) or WalkAway Riede et al. 1985; Riede 1987; Dunne et al. 2018). While in (Beckaman Coulter). Next, susceptibility profiles were per- T4, gp38 functions as a chaperone (Hashemolhosseini et al. formed using antibiogram cards for Vitek2 and WalkAway, 1996), in T2 and S16, gp38 acts as an adhesin that attaches covering imipenem, meropenem, and ertapenem, accord- to the mature LTF and modulates the receptor specificity ing to EUCAST guidelines. Beta-lactamases (AmpC) and (Riede et al. 1985, 1987; Marti et al. 2013). extended-spectrum beta-lactamase (ESBL)-producing bacte- Recently, more T-even phages have been isolated and ria were identified using an Etest (bioMérieux) with positive characterized (e.g., R3, EC14, EC35, and 8S coliphages). results for cefotetan/cefotetan + cloxacillin and for cefepime/ High-throughput techniques have confirmed that they use cefepime + clavulanic acid, respectively. All strains were lipopolysaccharide (LPS), LamB, OmpA, and OmpC as grown at 37 °C in Tryptic Soy Broth (TSB, VWR) with or surface receptors (Kortright et al. 2020). Similarly, Yersinia without agar (Merck). Different agar concentrations of 1.2% pestis myoviruses with equivalent properties, i.e., broad host or 0.6% were used to prepare Tryptic soy agar plates (TSA) range, use their long tail fiber tips for Omp and LPS recog- and soft agar overlays, respectively. nition (Chen et al. 2020; Li et al. 2020; Salem et al. 2021). To broaden the scope of phage therapy against emerg- Phage isolation ing Citrobacter infections, here we report the isolation of a new T-even C. koseri-infecting phage vB_CkoM_CkP1 Phage CkP1 was isolated from sewage water from an urban (further mentioned as CkP1). CkP1 has the typical genomic, wastewater treatment plant at Frossos (Braga, Portugal) morphologic, and infection/stability features of other T-even using the enrichment procedure, production and purifica- phages. We show that CkP1 has a host range covering all tion, exactly as described previously (Oliveira et al. 2016a, strains tested within C. koseri species and that it recognizes b). Phage stocks were stored in SM buffer (50 mM Tris–HCl the LPS using the LTF (gp267) with high nanomolar equi- pH 7.5, 100 mM NaCl, 8 mM ­MgSO4) at 4 °C. librium affinity. Furthermore, we predict protein structures of gp267 using AlphaFold-Multimer and compare to known Lytic spectra and efficiency of plating tail fiber structures of S16 and T4 (Jumper et al. 2021; Evans et al. 2022). Finally, we demonstrate that CkP1 is efficient The phage CkP1 host range and efficiency of plating (EOP) in controlling C. koseri in urine. Altogether, CkP1 displays was established against all strains listed in Table 1. Bacte- optimal traits that can be used to detect and control C. koseri rial lawns were prepared on TSA plates by adding 100 μl of infections. exponential-phase cell culture of each strain, and subsequent spotting of 10 μl of phage ­(108 PFU/ml and ten-fold dilu- tion series). After a 16-h incubation period at 37 °C, results Materials and methods were scored. The relative EOP was calculated by dividing the titer of the phage (PFU/ml) for each isolate by the titer Bacterial strains for the propagating host (CK#1) and recorded as high (≥ 0.5) or low (< 0.5). Sixty strains were used in this study, including the collection and clinical isolates from the Hospital de Braga (Braga, Por- One‑step growth curve tugal) collected over a period of 3 years, and strains kindly provided by Tomáš Kuchta lab in Slovakia (CK#6, CK#9, One-step growth curve experiments were performed as pre- CF#7, CB) or from the Salmonella Genetic Stock Centre viously described (Sillankorva et al. 2008). Briefly, 10 ml of (CK#22, CK#25, CF#1, CA#1, MM#1, SE#1), Spanish Type a mid-exponential-phase C. koseri CK#1 cells (­ OD620 = 0.5) Culture Collection (PV, EC), and American Type Culture was harvested by centrifugation (7000 × g, 5 min, 4 °C) and Collections (KP, SE#2) or standard laboratory E. coli strains resuspended in 5 ml fresh TSB medium. Next, phage CkP1 (Top10, K12) (Table 1 and associated footnote). The panel was added to the prepared host suspensions with a multi- includes several Citrobacter species, i.e., C. koseri (n = 25), plicity of infection (MOI) of 0.001 (in a final volume of 13 Applied Microbiology and Biotechnology (2023) 107:3621–3636 3623 Table 1  Lytic spectra and efficiency of plating of the Citrobacter phage CkP1 CkP1 phage Species Strain1 Origin Isolation date Patient gender Antibiotic ­resistances2 Infectivity EOP3 (dd/mm/yyyy) C. koseri CK#1 Pus 28/04/2014 Male AM + (1.0) High CK#2 Urine 05/05/2014 Male AM, AMC, CXM + (0.7) High CK#3 Skin 24/04/2014 Female AM, CXM + (0.8) High CK#4 Urine 22/04/2014 Female AM + (0.7) High CK#5 Biopsy Mat 28/04/2014 Male AM + (0.9) High CK#6 Unknown Unknown Unknown Unknown + (0.8) High CK#7 Urine 27/05/2014 Male AM, CXM + (0.9) High CK#8 Urine 28/05/2014 Male AM + (1.1) High CK#9 Unknown Unknown Unknown Unknown + (0.5) High CK#10 Urine 22/09/2014 Female AM + (0.8) High CK#11 Urine 08/11/2014 Male AM + (0.6) High CK#12 Ocular excess 27/01/2017 Male AM + (0.7) High CK#13 Urine 05/11/2014 Male AM + (0.8) High CK#14 Urine 24/10/2014 Male AM, AMC, GM, NN, TZP, SXT + (0.8) High CK#15 Bronchial aspirate 07/11/2014 Male AM + (1.0) High CK#16 Urine 06/02/2017 Female AM + (1.1) High CK#17 Urine 06/02/2017 Male AM, CXM, TZP + (0.9) High CK#18 Urine 28/08/2017 Female AM + (0.6) High CK#19 sputum 03/05/2017 Female AM + (0.9) High CK#20 skin 28/04/2017 Female AM, CXM + (0.6) High CK#21 Urine 19/04/2017 Female AM + (0.5) High CK#221 Unknown Unknown Unknown Unknown + (0.9) High CK#23 Urine 20/10/2017 Female AM + (0.7) High CK#24 Urine 29/10/2017 Male AM + (0.8) High CK#251 Unknown Unknown Unknown Unknown + (1.0) High C. freundii CF#11 Unknown Unknown Unknown Unknown − CF#2 Feces 30/04/2014 Female Unknown − CF#3 Feces 03/05/2014 Female Unknown − CF#4 Urine 15/04/2014 Male AM, CXM, SXT − CF#5 Urine 24/05/2014 Female AM, AMC, CXM, SXT, MEM, − TZP, CT, TZ, CIP, GM, NN, AN, ESBL, AmpC CF#6 Feces 30/05/2014 Male Unknown − CF#7 Unknown Unknown Unknown Unknown − CF#8 Feces 05/05/2014 Female Unknown − CF#9 Urine 12/06/2014 Female AM, AMC, CXM, SXT, MEM, − TZP, CT, TZ, CIP, GM, NN, AN, ESBL, AmpC CF#10 Urine 12/06/2014 Female AM, AMC, CT, TZ, CXM, TZP − CF#11 Urine 11/06/2014 Male AM, AMC − CF#12 Feces 12/05/2014 Female Unknown − CF#13 Urine 14/05/2014 Female AM, AMC, CXM − CF#14 Urine 13/05/2014 Female Unknown − CF#15 Peritoneal fluid 14/05/2014 Female Unknown − CF#16 Urine 07/05/2014 Male AM, AMC − CF#17 Pus 05/05/2014 Male AM, AMC − CF#18 Urine 11/05/2014 Male AM, AMC − CF#19 Urine 15/04/2014 Female AM, AMC − 13 3624 Applied Microbiology and Biotechnology (2023) 107:3621–3636 Table 1  (continued) CkP1 phage Species Strain1 Origin Isolation date Patient gender Antibiotic ­resistances2 Infectivity EOP3 (dd/mm/yyyy) CF#20 Urine 04/11/2014 Male AM, AMC − CF#21 Urine 05/11/2014 Male AM, AMC − C. amalonaticus CA#11 Unknown Unknown Unknown Unknown − CA#2 Peritoneal fluid 14/05/2014 Female AM, CXM − C. youngae CY Feces 08/05/2014 Male Unknown − C. braakii CB Unknown Unknown Unknown Unknown − M. morganii MM1 Urine Unknown Female Unknown − P. stuartii PS Urine Unknown Female AM, AMC, CXM − P. rettgeri PR Unknown Unknown Unknown − P. mirabilis PM1 Unknown Unknown Unknown Unknown − P. vulgaris PV1 Unknown Unknown Unknown Unknown − E. coli EC#11 Unknown Unknown Unknown Unknown − EC#21 Unknown Unknown Unknown Unknown − K. pneumoniae KP1 Unknown Unknown Unknown Unknown − S. enterica SE#11 Unknown Unknown Unknown − SE#21 Unknown Unknown Unknown − 1 Collection strains: CK#22-SGSC 5610; CK#25-SGSC 4696; CF#1-SGSC 5345; CA#1-SA 5615; MM#1-SGSC 5703; PV-CECT 174; EC- CECT 432; KP-ATCC 11,296, SE#1-SGSC 3029, SE#2-ATCC 13,076; reference strains: EC#1-Top10, EC#2-BL21 2 Antibiotic resistances: AM ampicillin, AMC amoxicillin/clavulanic acid, CXM cefuroxime, SXT cotrimoxazole, TZP piperacillin/tazobactam, MEM meropenem, CT cefotaxime, TZ ceftazidime, CIP ciprofloxacin, GM gentamicin, NN tobramycin, AN amikacin, ESBL extended-spectrum beta-lactamase producing bacteria, AmpC beta-lactamase producing bacteria 3 The relative efficiency of plating was calculated by dividing each phage titer (PFU/ml) by the phage titer of the propagating host (CK#1) and recorded as high (≥ 0.5) or low (< 0.5) 10 ml). Phages were allowed to adsorb for 5 min at 37 °C, Averages ± standard deviations for all experiments are indi- 120 rpm (Biosan ES-20/60). The mixtures were subse- cated for n = 3 repeats. quently centrifuged (7000 × g, 5 min, 4 °C) and the pellets were resuspended in 10 ml of fresh TSB medium. Samples Electron microscopy were taken every 5 min over a period of 30 min and then every 10 min until 1 h of infection. The phage concentration Phage particles were examined with a Jeol JEM 1400 trans- was assessed by plating 100 μl of tenfold serial dilutions, mission electron microscope (TEM), exactly as described previously mixed with 100 μl of the overnight cultured host before (Oliveira et al. 2017). and 3 ml of soft-agar TSA (0.6% agar). Averages ± standard deviations for all experiments are shown for n = 3 repeats. Phage DNA isolation, sequencing, and annotation Phage DNA was extracted from highly concentrated and Phage stability purified phage stock (≥ ­1010 PFUs/ml) using the standard procedures described elsewhere (Sambrook and Maniatis Tests were performed to evaluate the phage CkP1 stability 1989). A 2 × 150 bp paired-end DNA library (Nextera XT at either different temperature or pH values, as previously sample prep) was prepared and sequenced with an Illu- described (Oliveira et al. 2016a, b). Thermal stability was mina MiSeq platform (Illumina Inc., USA) at the Nucle- established by incubating ­108 PFU/ml of the phage at − 20, omics Core (VIB, Belgium). After processing, reads were 4, 37, 42, 50, and 60 °C for 24 h. Similarly, the effect of pH trimmed to remove adapters, contaminations, or low-quality was also evaluated for pH values of 1, 3, 5, 7, 9, 11, and 13 nucleotides and then de novo assembled with CLC Genom- using a universal pH buffer (150 mM potassium chloride, ics Workbench version 7.0 (CLC Bio, Aarhus, Denmark). 10 mM potassium dihydrogen phosphate, 10 mM sodium PCR amplifications and Sanger sequencing were performed citrate, 10 mM boric acid). In both experiments, phages were to verify regions displaying uncertain consensus sequence. diluted and plated on C. koseri CK#1 lawns for enumeration. In addition, contigs were scaffolded by amplification and 13 Applied Microbiology and Biotechnology (2023) 107:3621–3636 3625 Sanger sequencing (Eurofins Genomics) of fragments that 10 °C, 180 rpm. Protein purification was performed on an covered the remaining gaps. immobilized metal affinity chromatography column using The genome was annotated using myRAST (Aziz et al. an imidazole gradient (25 to 250 mM). Next, samples were 2008) and manually inspected. Sequence similarity detec- dialyzed against in 10 mM Tris–HCl (pH = 7) buffer and tion and secondary structure prediction was performed using quantified with the Pierce BCA Protein Assay Kit (Thermo BLASTp (Altschul et al. 1990) and HHPRED (Soding et al. Fisher) as described previously (Oliveira et al. 2016a, b). 2005), respectively. tRNAs were predicted using tRNAscan- SE (Schattner et al. 2005). The DNA homology comparisons Tail fiber binding between phage genomes were performed with BLASTN and visualized with Easyfig (Sullivan et al. 2011). The shared Fuorescence microscopy protein content was conducted with Orthovenn2 (Xu et al. 2019). The binding ability of the different gp267 LTF constructs was inferred by fluorescence microscopy observations of Tail fiber cloning, expression, and purification the C. koseri cells after incubation with each three fusion proteins. GFP alone was used as negative control. Briefly, The LTF protein (gp267) was fused to the Aequorea coerule- bacterial cells were grown in TSB medium at 37 °C to the scens green florescence protein (GFP) and produced recom- mid-log phase (­ OD620nm = 0.3–0.4) and washed with 10 mM binantly after co-expression with and without the putative Tris–HCl (pH = 7). A pellet from a 100-µl suspension was chaperone (gp268) as a bicistronic transcript in pET28a. resuspended in 20 µL of 10 µM purified gp267 LTF and Three open reading frames were made: GFP-gp267 (GFP incubated for 15 min at 20 °C. Cells were washed twice fused with the gp267 LTF); GFP-gp267/gp268 (GFP fused in 10 mM Tris–HCl to remove the unbound protein. The with the gp267 and co-expressed with the putative chaper- washed pellet was suspended in 10 µl of 10 mM Tris–HCl one gp268); and GFP-gp267trunc/gp268 (GFP fused with (pH = 7) and observed under an epifluorescence microscope a truncated variant of gp267 (starting at nucleotide 1249, equipped with U-RFL-T light source (Olympus BX51, Mag- i.e., amino acid residue 417, designed to use the predicted nitude 1000 ×) in bright field and under the FITC (470–490, C-terminal binding domain only and co-expressed with LP-516) filter. Control samples using 10 mM Tris–HCl (pH gp268). Genes 267 and 268 were amplified with Phusion™ 7) or GFP alone (instead of the fused recombinant gp267 High-Fidelity DNA Polymerase (Thermo Fisher Scientific) LTF) were prepared simultaneously. using phage CkP1 genomic DNA as template with the prim- ers Fw 5′-CCGC ​ CGG​ GAT​ CCG​ AAT​ TCA​ TGG​ CAA ​ CTA ​ TT​ Binding kinetics of tail fiber constructs to C. koseri cells AAG​CAA​ATA​CAA​TTA​AAA​AGA​AG, Fw 5′-CGC​CGG​ GATCC​GAA​TTC​ATG​TGG​GGA​ACT​GGT​GGT​TTA​AAA​ Binding kinetics and affinity of the full-length and truncated G, Rv 5′-GCC​GCC​GTC​GAC​TTA​ACC​GAT​TCT​TAC​AAT​ C-terminal version of the GFP-LTF constructs to the sur- ATA​TTT​AAA​CGC​TAC, and Rv 5′-CCG​CCG​CTC​GAG​ face of C. koseri cells were determined by surface plasmon TTA​GCC​ATG​TGA​GTG​TCT​ATG​GAA​TAG, containing resonance (SPR) analysis, using a BIAcore X system (GE the BamHI/SalI/XhoI restriction sites (underlined). Ampli- Healthcare Life Sciences) and C1 sensor chips, essentially cons were cleaned (DNA Clean & Concentrator-5 k, Zymo as previously described (Schmelcher et al. 2010). In brief, Research, USA), digested with the listed fast digest restric- the GFP-gp267trunc/gp268 protein was immobilized on the tion enzymes (Thermo Fisher Scientific) and ligated with sensor chip surfaces in both flow cells (Fc1 and Fc2) using a T4 ligase (Thermo Fisher Scientific) into the previously the amine coupling procedure (70 µl of 0.5 mg/ml at a flow constructed pET28a with GFP (Santos et al. 2019). All con- rate of 5 µl/min). Then, inactivated C. koseri cells (4% of structs contain an N-terminal 6xHis-tag followed by the gfp paraformaldehyde fixed) in HBS-T running buffer (10 mM upstream the phage gene(s). The different constructs were HEPES–NaOH, 150 mM NaCl, 3.4 mM EDTA, 0.005% used to transform E. coli TOP10 competent cells (Invitro- Tween 20, pH 7.8) were immobilized onto the protein lawn gen). Colonies were screened through colony PCR and the in Fc2 (15 µl of ~ 2 × ­109 CFU/ml at a flow rate of 3 µl/min). positive ones were used for plasmid extraction and further Finally, real-time interactions between the immobilized confirmation through Sanger sequencing (Eurofins Genom- cells in Fc2 and the proteins of interest (GFP-gp267, GFP- ics). The pET28a plasmid with GFP was used as control in gp267/gp268, and GFP-gp267trunc/gp268) were measured cloning and expression. Correct plasmids were used to trans- (30 µl at a flow rate of 10 µl/min) at different concentrations form competent E. coli Arctic Express cells (DE3) (Agilent) (1 nM to 1 µM). The Fc1 cell served as a reference in this to express the proteins when reaching a mid-exponential experiment. Association phases were recorded for 3 min phase (optical density at 620 nm of 0.6), with 1 mM isopro- and dissociation phases for 12 min. Between measurements pyl β-d-1-thiogalactopyranoside (Sigma-Aldrich), 24 h at of different concentrations of the same protein, the chip 13 3626 Applied Microbiology and Biotechnology (2023) 107:3621–3636 surface was regenerated by one to three short pulses (10 according to the supplier’s protocol. The DNA libraries were to 20 µl) of regeneration buffer (10 mM HEPES–NaOH, sequenced in the lllumina NovaSeq platform, using 150-bp 1 to 2 M NaCl, pH 7.8). Kinetic data were evaluated by paired-end sequencing reads (Stabvida). The samples gen- global fittings with the “1:1 binding with mass transfer” erated from 1122 Mbp (7,435,112 sequence reads) to 1465 and “two-state reaction” models predefined in the BIAcore Mbp (9,704,010 sequence reads) were assembled, resulting evaluation software. in a theoretical average coverage ranging from 238 × to 311x (assuming a genome size of 4.7 Mbp). Structure prediction with AlphaFold‑Multimer To detect small genomic variations (single nucleotide var- iants, SNVs, and insertions/deletions (indels)), the genomic The trimeric structures of the N-terminal (AA 1–416) and data obtained for the variants were mapped to the reference C-terminal part (AAs 417–670) of the gp267 protein were genome of the C. koseri wild type (GenBank accession no. predicted using AlphaFold-Multimer (v2.2.2) that was NC_009792) using bowtie2 (v2.3.3.1) with option --local deployed on Ghent University’s supercomputing infrastruc- (Langmead and Salzberg 2012), resulting in an overall align- ture (Jumper et al. 2021; Evans et al. 2022). Two NVIDIA ment rate of > 96% for each variant. Candidate small vari- Ampere A100 graphics processing units (GPUs) with 80 GB ations were detected using samtools (v1.9) (Li et al. 2009; of GPU memory were used. Predicted structures were Li 2011) with a custom shell script (code accessible upon inspected for their accuracy by computing and visualizing request), filtering for variants covered by at least five reads, the predicted local distance difference test (pLDDT) scores, and confirmed by visually inspecting the mapped sequence the predicted template modeling scores, and alignment data in the Integrative Genomics Viewer (IGV) (Robinson errors, using code from ColabFold (Mirdita et al. 2022). et al. 2017). Larger deletions were searched by identifica- Both predicted structures were subjected to FoldSeek, a tion of regions with low coverage (< 10 reads) using bed- search and comparison tool for protein structures, to detect tools (v2.26.0) (Quinlan 2016) but no deletions were iden- and compare our predictions to other known structures (van tified after visual inspection of these regions in IGV. As a Kempen et al. 2022). complementary approach, the variant genomes were also assembled de novo using SPAdes (v3.13.1) (Bankevich et al. Phage receptor 2012) using default parameters and the reference genome NC_009792 as “trusted contig.” The resulting contigs were Generation of phage‑resistant variants aligned with the reference genome using progressiveMauve (Darling et al. 2010) to identify potential large genomic vari- Two different approaches were used to identify the phage ations or rearrangements. receptor. First, C. koseri SGSC4696 (GenBank acces- sion no. NC_009792) strain mutants were generated using Complementation strain construction the Datsenko technique (Datsenko and Wanner 2000). Recombinogenic plasmid pMJH46 (Addgene) was trans- The rfaF (CKO_RS21630) and the gt (CKO_RS21655) ferred to electrocompetent C. koseri SGSC4696 cells. Next, genes were amplified by PCR with Phusion High-Fidelity homologous sequences were chosen to delete either ompC DNA Polymerase (Thermo Scientific) using the C. koseri (CKO_RS02455) (Fw 5′-ATG​TTA​CGC​AGC​AGC​AAC; Rv SGSC4696 (accession no. NC_009792) genomic DNA as 5′-TTAG ​ GTG​ GCG​ GTA​ CTT ​ GG)​ or ompA (CKO_RS08945) template and using the primers Fw 5′-CCA​TGG​CCA​TGA​ (Fw 5′-ATG​TTA​CGC​AGC​AGC​AAC; Rv 5′-TTA​GGT​GGC​ AAA​TAC​TGG​TGA​TTG​GTC​, Rv 5′-AAG​CTT​TCA​GGC​ GGT​ACT​TGG​) genes or both combined, using the listed TTC​CTC​TTG​TAA​TAA and Fw 5′-CCA​TGG​CCA​TGG​ primers with 50-bp recombination arms. Deletion mutants ATA ​ ACA ​ A AA​ C TA ​ ACG​ TAA​T TA ​ ATA​T TGC, and Rv were confirmed by Sanger sequencing (Eurofins Genom- 5′-AAG​CTT​TTA​ATG​AAT​CTT​GGT​TTT​GAG​GTA​CTTT, ics). Secondly, five spontaneous C. koseri SGSC4696 vari- respectively, containing the NcoI/HindIII restriction sites ants displaying resistances to the phage CkP1 after incuba- (underlined). The amplicons were digested with the Fast- tion overnight on TSA plaques were selected for genome Digest NcoI/HindIII restriction enzymes (Thermo Scien- sequencing. DNA was extracted from overnight cultures tific), ligated into pET28a with a T4 DNA ligase (Thermo using the Quick-DNA Fungal/Bacterial Miniprep Kit (Zymo Scientific) according to the manufacturer’s instructions, Research). To prepare DNA fragments (DNA libraries), 1 µg and used for transformation of E. coli TOP10 electrocom- of genomic DNA was fragmented using the Bioruptor soni- petent cells (25 μF, 200 Ω, 2.5 kV). The selection of E. coli cator (Diagenode, Denville, NJ) and then prepared using the clones was done on LB plates supplemented with 50 µg/ KAPA HyperPrep kit (KAPA Biosystems, Wilmington, MA) ml kanamycin. The final constructs were verified by Sanger 13 Applied Microbiology and Biotechnology (2023) 107:3621–3636 3627 sequencing (Eurofins Genomics). Next, electrocompetent Results C. koseri SGSC4696 cells (wild-type and phage-resistant variants) were prepared. Briefly, a culture of 100 ml of CkP1 infects all C. koseri strains tested mid-exponential C. koseri SGSC4696 cells was centrifuged and with a high killing efficiency four times (3000 × g, 15 min at 4 °C) and suspended in sub- sequent lower amounts of ice-cold 10% glycerol (40, 20, Hospital clinical isolates of C. koseri were used to enrich 10 and 1 ml). Cells were then transformed (25 μF, 200 Ω, wastewater treatment plant sewage samples, which resulted 2.5 kV) with the recombinant plasmids (pET28a::rfaF and in the isolation of the phage vB_CkoM_CkP1 (CkP1) pET28a::gt) and selected using TSA plates supplemented exhibiting uniform, clear, and small plaques (0.1 mm in with 50 µg/ml kanamycin. diameter). A comprehensive panel of 60 bacterial strains was used to determine the phage host range and its effi- Bacterial challenge test ciency of plating (Table 1). To our knowledge, CkP1 is the first described phage specific for C. koseri. Notably, CkP1 Early exponential phase (­ OD620nm = 0.2) Citrobacter cultures was able to infect all tested C. koseri strains from differ- (CK#1 and CK#2) were 10 × diluted to obtain ∼106 CFU/ ent origins, isolated from a 3-year period and which have ml and then challenged with CkP1 at MOI of 0.1 at 37 °C. distinct antibiotic resistance profiles. CkP1 also presented SM buffer was used as a control. Next, the bacterial growth a high EOP on all C. koseri strains tested. In turn, CkP1 was monitored by turbidity (­ OD620nm) every hour. Comple- was unable to lyse closely related Citrobacter (C. freundii, mentary analysis was made to assess the phage antibacterial C. amalonaticus, C. youngae, and C. braakii) and more effect at 8 h of incubation, by plating serial dilutions of the distant Enterobacteriaceae species. suspensions with and without the CkP1 in 0.9% NaCl, and To further characterize the phage infection cycle, one- counting the CFUs after an overnight incubation at 37 °C. step growth curve experiments were established under Averages ± standard deviations for all experiments are given standard growth conditions (Figure S1). CkP1 had a rela- for n = 3 repeats. tively short latent period of 25 min with an average burst size of 110 ± 38 PFUs per infected cell. Taken together, Infection assays in urine CkP1 seems to hold optimal characteristics for detection and/or treatment of C. koseri infections, including those To complement the assessment of the antimicrobial poten- caused by drug-resistant isolates. tial of the phage, CkP1 was tested in artificially contami- nated urine. Donors have signed their informed consent on collecting the urine samples. Mid-exponential growing CkP1 features high stability, able to withstand CK#1 cells were washed once with TSB and then with various pH and temperatures filtered (0.22 µm) urine collected from healthy human subjects. Finally, cells were diluted in filtered urine To further assess this phage’s potential for therapy and to ~ ­106 CFU/ml. The contaminated urine samples (190 µl) detection, stability tests were set up, exposing CkP1 to were incubated with phages at a final multiplicity of infec- different temperatures and pH values, for 24 h (Fig. 1). tion of 0.1 (10 µl) or with SM buffer (10 µl). The suspen- CkP1 remained stable between − 20 and 50 °C, maintain- sion was incubated for 4 h and for 24 h at 37 °C. After, the ing its titer. At 60 °C, the CkP1 infection was significantly cells were diluted in 0.9% NaCl, plated in triplicate, and affected, with a titer decrease of 5 log units. Under differ- CFUs were counted following an overnight incubation at ent pH conditions, no loss of CkP1’s titer was observed 37 °C. Averages and standard deviations of four repeated after 24 h within the pH range of 5 to 11. At pH 3, CkP1’s experiments are given. titer reduced approximately 1 log unit and was completely abolished at extreme pH values of 1 and 13. Furthermore, Nucleotide sequence accession numbers CkP1’s titer remained unaffected for 2 years when stored at pH 7 and 4 °C (data not shown). The complete genome sequence of the phage vB_CkoM_ CkP1 was deposited in GenBank under the accession num- ber MW239124. The sequencing reads of the C. koseri CKP1 is a new S16‑like myovirus SGSC4696 variants isolated in this study were depos- ited in the European Nucleotide Archive (ENA, https://​ Morphologically, TEM images showed that CkP1 has a www.​e bi.​a c.​u k/​e na) under the study accession number myovirus morphotype (Fig. 2), composed of an icosahe- PRJEB41322. dral head with a contractile tail (head, 98 nm; tail, 133 13 3628 Applied Microbiology and Biotechnology (2023) 107:3621–3636 Fig. 1  Phage CkP1 biophysical stability. Effect of different tempera- standard deviations of three repeated experiments are given. Signifi- ture (left) and pH (right) conditions on CkP1 phage survival after cance was determined by Student t test (*P < 0.05) 24-h exposure by enumerating the number of phages. Averages and by 18 nm) (Ackermann 1996). It was also possible to classified as a Tevenvirinae (T-even) of the Straboviridae observe the different folding positions of the CkP1 tail family. The CkP1 genome structure resembles those found fiber, stowed along the tail sheath and extended. in T-even phages. The DNA packaging and structural mod- Genomically, CkP1 has a linear DNA molecule with ule is tightly packed (gp165-211 and gp264-268), while a size of 168,463 bp, encoding 291 putative genes and the DNA replication, recombination, and modification and 11 tRNAs (Fig. 3A). The CkP1 GC content (36.9%) is the cell lysis module, composed of the endolysin (gp131), similar to other Citrobacter phages (e.g., Moon, Mijalis a T type holin (gp269), and the o- and i-spanins (gp241- with GenBank accession no. KM236240, KY654690) or 242), are scattered throughout the genome. Overall, CkP1 slightly lower (e.g., CfP1, CVT22 GenBank accession shares 38 out of 39 T4-like core-genes, i.e., the most no. KX245890, KP774835). CkP1 is related with the Sal- ancient genetic components of this group (Petrov et al. monella phage S16 (Fig. 3A), a T-even broad host range 2010). The prohead core scaffolding and protease is the phage that specifically recognizes Salmonella through only gene missing in CkP1. Still, one CkP1 orthologue is OmpC (Marti et al. 2013). CkP1 shares 36% DNA overall found in a similar genome position as of T4 genome and homology and 62% of its proteins with those of phage S16, with similar function. Fig. 2  Phage CkP1 electron micrographs. Phage particles of CkP1 negatively stained with 2% uranyl acetate observed at TEM. Long tail fibers folded along the tail sheath (white arrow) or extended (black arrow) are indicated 13 Applied Microbiology and Biotechnology (2023) 107:3621–3636 3629 Fig. 3  Phage CkP1 multiple genome alignment. A Whole-genome NC_000866), C. koseri phage CkP1, and Salmonella phage S16. All pairwise comparisons between the phage CkP1 (GenBank accession comparisons were made using tBLASTX and visualized with Easy- no. MW239124) and the Salmonella phage S16 (GenBank accession Fig. CDSs are drawn to scale and colored according to their predicted no. HQ331142). B Pairwise comparison of the long tail fiber region function between the prototype E. coli phage T4 (GenBank accession no. CkP1 gp267 is a tail fiber able to decorate C. koseri of gp267 and the tip of the long tail fiber of T4 (gp37; PDB cells entry 2XGF; Table S1, Figure S3). This fragment also con- tains eight His-X-His sequences motifs (X being any amino Based on in silico analysis, we identified the proximal to the acid), and its predicted structure forms a needle structure distal parts of CkP1 LTF formed by gp264 and gp267 pro- like T4 gp37 in which these histidine pairs coordinate iron teins, as well as their predicted accessory proteins involved ions (unlike gp37 of T2 or S16) (Bartual et al. 2010). Fur- in expression/trimerization (gp160/gp268). They are in per- thermore, neither FoldSeek nor HHPred show any meaning- fect synteny with their LTF analogs (gp34-gp37 and gp57A/ ful structural similarity for the predicted N-terminal part of gp38) found in the type phages T4 and T2 in one hand and gp267 compared to other known structures (Figure S3). Note S16 on the other hand (Marti et al. 2013; Hyman and Raaij that the N-terminal parts of gp37 of T2 and T4 have not yet 2018) (Fig. 3B). While the proximal LTF parts of CkP1, been crystallized, which could explain this finding. Overall, i.e., gp264-gp266 has relatively high similarity (> 60% aver- both predicted structures suggest that the gp267 of CkP1 age amino acid identity) to T-even proteins, the distal ends LTF is a combination of a C-terminal part that is structur- gp267 and gp268 exhibit moderate or low levels of iden- ally similar to gp37 of T4 with an additional N-terminal part tity (maximum of 54% and 36% amino acid identity to the that is not structurally similar to any other known structure. homologs in T4 and T2, respectively). In addition, the struc- To functionally characterize the CkP1 LTF, we con- tures of the N-terminal and C-terminal parts of gp267 were structed three versions of gp267 fused to a GFP (GFP-LTF), modeled using AlphaFold-Multimer. Both parts of gp267 which were co-expressed with and without the putative are well predicted by AlphaFold, reaching global pLDDT chaperone gp268 (referred to as GFP-gp267, GFP-gp267/ scores of 86.0 and 87.6 for the N-terminal and C-terminal gp268, and GFP-gp267trunc/gp268). The reasoning behind part, respectively (Figure S2). Notably, predicting the full the gp267 LTF truncation variant was to isolate the puta- gp267 structure results in a considerably lower pLDDT tive receptor-binding sequence, which is located in the score (data not shown). Both FoldSeek and HHPred results C-terminus. To assess the role of the gp267 LTF ligand, its indicate structural similarity between the C-terminal part potential binding activity was analyzed by epifluorescence 13 3630 Applied Microbiology and Biotechnology (2023) 107:3621–3636 Fig. 4  Phage CkP1 LFT binding of C. koseri cells. C. koseri cells washed twice and visualized under epifluorescence microscopy, in (CK#1) suspended in 10 mM Tris–HCl (pH = 7) were incubated with bright field or using the FITC filter GPF, GFP-gp267, GFP-gp267/gp268, or GFP-gp267trunc/gp268, microscopy. We observed that gp267 always bound to C. that this observation alone does not provide direct evidence koseri cells independent of the construct (Fig. 4). Moreover, for an actual two-state reaction, including a conformational fluorescent microscopy experiments were performed using change occurring during the binding of the CkP1 LTF to GFP-LTF proteins on all strains depicted in Table 1. All its cell wall receptor. This modeling enabled the calcula- C. koseri strains were decorated by the GFP-LTF proteins tion of association (ka1), dissociation (kd1), forward (ka2), while non-koseri strains from the Citrobacter genus or other and backward (kd2) rate constants, as well as the apparent closely related bacteria did not show fluorescent decoration equilibrium affinity constants (K), for all three variants of on the microscopy experiments. This matched the GFP-LTF the LTF (Table 2, Figure S4). The K values obtained for binding activity to the lytic spectrum of the phage CkP1, and GFP-gp267 and GFP-gp267/gp268 were 2.03 × ­108 ­M−1 and resulted in a 100% specificity and sensitivity on the panel 8.41 × ­107 ­M−1, respectively, which corresponds to affinities of strains tested. in the low nanomolar range, whereas the equilibrium affin- ity constant for GFP-gp267trunc/gp268 was approximately CKP1 LTF binds to cells with nanomolar affinity tenfold lower (8.27 × ­106 ­M−1). Similar values were obtained when a “1:1 binding with mass transfer” model was applied The equilibrium binding affinities of the different CkP1 for the fitting (data not shown). LTF versions against immobilized bacterial cells were ana- lyzed by SPR. Kinetic data was obtained at different CkP1 CkP1 recognizes C. koseri lipopolysaccharide LTF concentrations (1 nM to 1 µM) for each variant, and a through its LTF global fitting of the data overall measured concentrations was performed. The best fit was obtained when using a pre- To identify potential phage receptors, two different strategies defined two-state reaction model, as previously described were pursued. An initial approach (the Datsenko technique) for the cell wall binding domain of the Pseudomonas phage used a direct knockout approach of potential gene candidates endolysin KZ144 (Briers et al. 2009). It should be noted frequently involved in host recognition by T4-like phages. Table 2  Kinetic data and equilibrium affinity constants (K) for the binding of CkP1 GFP–LTF variants to C. koseri Ck#1 cells. For each variant, a global fitting over six different concentrations (1 nM to 1 µM) was performed, using a “two-state reaction” model LTF variant ka1 ­(M−1 ­s−1) kd1 ­(s−1) ka2 ­(s−1) kd2 ­(s−1) K ­(M−1) GFP-gp267 7.57 × ­103 9.65 × ­10−4 4.80 × ­10−4 1.93 × ­10−5 2.03 × ­108 GFP-gp267/gp268 8.89 × ­103 2.02 × ­10−3 5.27 × ­10−3 2.90 × ­10−4 8.41 × ­107 GFP-gp267trunc/gp268 1.22 × ­103 2.84 × ­10−3 9.79 × ­10−3 5.34 × ­10−4 8.27 × ­106 13 Applied Microbiology and Biotechnology (2023) 107:3621–3636 3631 As a second approach, we mapped the missing genes of Table 3  Genetic variations in strains used to assess the CKP1 potential receptors. All variants used in this study derived from C. koseri SGSC4696 (GenBank accession n. NC_009792), being generated either using Datsenko technique (V1-V3) or by isolation of spontaneous mutants challenged with CkP1 (V4-V8), as described in the material and methods. In literature, the rfaF gene NO NO V8 X – – – – – phage-resistant C. koseri variants that were sequenced and is also known as waaF. SNV single nucleotide variants. Below a summary of the spectrum of activity of both phage and its LTF against the different C. koseri variants generated in this study compared to the wild type genome (Table 3). NO NO V7 X X In the first approach, C. koseri deletion mutants of OmpC – – – – (ompC at CKO_RS02455) and OmpA (ompA at CKO_ NO NO V6 RS08945) porins were created individually and combined, X X – – – – as they are often reported as receptors of T-even phages NO NO V5 (Bertozzi Silva et al. 2016). However, none of the resulting X X – – – – mutants influenced phage infection, its EOP, or the targeted NO NO binding of GFP-LTF (Table 3). V4 X X – – – – As a second and complementary approach, five spontane- ous resistant mutants to CkP1 were isolated. The GFP-LTFs YES YES V3 X X were also not able to bind to these mutants, with the excep- – – – – tion of variant V8 for which the cells were not completely YES YES V2 decorated, but some green fluorescent dots were observed X – – – – – around the cells. The mutants were thus sequenced to iden- YES YES tify potential mutations affecting the receptor. Genomic V1 X – – – – – data demonstrated four small genetic variations as poten- tial candidates for resistance-specific mutations located in Indel: frameshift deletion at pos. 78 Indel: frameshift insertion at pos. 9 the genes encoding the 1,4-alpha-glucan branching enzyme (glgB), ADP-heptose-LPS heptosyltransferase (rfaF), and glycosyltransferase (gt) (Table 3). While the role of the glgB is unclear, the rfaF and gt are predicted to be involved in the synthesis of the lipopolysaccharide. Two variations occurred SNV: G429 to E SNV: L116 to R in the rfaF gene encoding an ADP-heptose-LPS heptosyl- Indel: deletion Indel: deletion Variant allele transferase, which is involved in lipopolysaccharide biosyn- thesis (KEGG ortholog K02843). A frameshift insertion was found at codon 9 in three mutants and another mutant car- ried a substitution of leucine to arginine at codon 116. All four mutants carrying an rfaF variant also carried a glycine ADP-heptose-LPS heptosyltransferase ADP-heptose-LPS heptosyltransferase 1,4-alpha-Glucan branching enzyme to glutamic acid substitution in another gene, glgB, which encodes an 1,4-alpha-glucan branching enzyme involved in starch and glucose metabolism (K00700). The fifth Outer membrane porin A Outer membrane porin C mutant was affected by a frameshift deletion in the gt gene. Interestingly, the mutated gene is located in close vicinity Annotation function Glycosyltransferase (~ 4,400 bp) of rfaF. No additional larger genomic varia- tions (insertions, deletions, or rearrangements) were found in the mutant strains. In an attempt to validate the phenotypic effect of these mutations, we performed in-trans comple- mentation of rfaF and gt genes, but not for glgB gene, which failed to be amplified by PCR. Nevertheless, we observed CKO_RS21655 CKO_RS21630 CKO_RS21630 CKO_RS20700 CKO_RS08945 CKO_RS02455 that the complemented mutants were again infected by phage with high EOP. Accordingly, the CkP1 GFP-LTF protein Locus tag versions could also decorate all the complemented mutants (Figure S5). CkP1 can inhibit Citrobacter growth Sensitive to phage CkP1 Decorated by GFP-LTF To complement the assessment of the phage CkP1 anti- microbial properties, we challenged two C. koseri strains (CK#1, CK#2) with the phage at MOI of 0.1 and monitored the OD over time (Figure S6). In the absence of CkP1, the ompC ompA Gene glgB rfaF rfaF cultures grew exponentially to an OD value of approximately gt 13 3632 Applied Microbiology and Biotechnology (2023) 107:3621–3636 1.0. In the presence of CkP1, both cultures were inhibited (formerly named C. diversus) is among the most common of growth until at least 8 h. At this incubation period (8 h), isolated species in human clinical specimens able to cause CkP1 reduced 4 logs of both CK#1 and CK#2 cultures, com- localized (urinary track, wounds infection, pneumonia, men- paratively with cells without phage treatments. Therefore, ingitis) and systemic life-threatening diseases (bacteremia, CkP1 proved to be efficient to reducing C. koseri in rich septicemia). Urinary tract infections are often associated media. with drug-resistant C. koseri (Ranjan and Ranjan 2013) and therefore alternative treatments are needed. C. koseri cells are efficiently controlled by CkP1 In this study, we isolated and characterized a novel T-even in urine samples phage, CKP1, specific for C. koseri. CkP1 shares 38 out of 39 T4-like core genes (Petrov et al. 2010), but contains To mimic treatments for urinary tract infections, the anti- one orthologue possibly covering the missing prohead core microbial effect of CkP1 was assessed in an in vitro setting, scaffolding and protease gene function. Whole-genome using urine samples collected from healthy donors (Fig. 5). comparisons place the broad host range Salmonella phage These samples were mixed with mid-exponential phase C. S16 as the closest CkP1 homolog. Similarly, CkP1 also koseri cells and further incubated with CkP1 at a MOI of infects all strains but limited to the C. koseri species, not 0.1. In the absence of phages, C. koseri cells were able to infecting other closely related species. The C. koseri strains grow 1 log in cell number after 24 h incubation at 37 °C. In were obtained from public collections and clinical speci- the presence of phages, CkP1 bacterial counts were reduced mens isolated from different sources, time points (spanning by proximally 2 logs and 5 logs (6 logs if we consider the three years) and which display distinct antibiotic-resistance control at the same time point) after 4 h and 24 h of incuba- patterns. tion, respectively. It can therefore be concluded that CkP1 Our initial efforts attempted to identify if common outer treatment can efficiently control C. koseri cells in spiked membrane proteins previously described in T-even phages urine under these relevant conditions. (porins OmpA and OmpC) (Bertozzi Silva et al. 2016) were also the receptors of CkP1. After single gene knockouts and double null mutants of these porins (ompC at CKO_ Discussion RS02455 and ompA at CKO_RS08945 locus), the resulting strains remained sensitive to the phage and could still be Citrobacter spp. are ubiquitous Gram-negative, faculta- infected with high efficiency. As an alternative approach, tive anaerobic bacteria that belong to the Enterobacte- we selected spontaneous bacterial variants that survived riaceae family. Currently divided into 8 members, C. koseri phage predation, and were able to pinpoint indels and SNVs Fig. 5  Control of C. koseri in urine with CkP1. Antibacte- rial effect of CkP1 on urine samples artificially contained with C. koseri (CK#1). Phage was applied at a MOI of 0.1 and the antibacterial effect measured by reduction of bacterial CFUs/ ml at 4 h and 24 h. Significance was determined by the Student t test (*P < 0.05). Averages and standard deviations of four repeated experiments are given 13 Applied Microbiology and Biotechnology (2023) 107:3621–3636 3633 in three genes (glgB, rfaF, and gt). rfaF and gt genes are Schmelcher et al. 2010; Marti et al. 2013). In this assay, the located within the C. koseri SGSC4696 rfa (also known reported apparent equilibrium affinity constants represent as waa) operon (4,705,399–4,716.689 bp) predicted to be a measure for the accumulated binding strength between involved in the synthesis and assembly of the LPS, simi- the possibly multimeric protein and the cell surface, taking larly to E. coli rfa operon (Pagnout et al. 2019). Mutations into account possible avidity effects. For instance, dupli- on rfaF (also known as waaF) have been shown to severely cation of the cell wall binding domain (CBD) of a phage truncate the lipooligosaccharide composed of Kdo and lipid endolysin resulted in an approximately 50-fold increase in A (Kanipes et al. 2006), whereas the lack of the glycotrans- equilibrium association constant (KA) as compared to the ferase affects the distribution of sugar moieties (Leipold parental single-CBD protein. This could be explained by et al. 2007). As for the branching enzyme glgB gene involved the simultaneous binding of up to two cell wall ligands by in the glycogen biosynthesis pathway, it is found outside the the dual-CBD construct (Schmelcher et al. 2011). In our rfa operon, so it is unclear whether its function is related experiments, we showed that there were no differences in to LPS biosynthesis. Four out of the five C. koseri variants the observed nanomolar range affinity equilibrium between studied contained mutations in glgB together with rfaF or gt the LTF (gp267) co-expressed with or without the additional genes. Due to unknown reasons, we were not able to amplify protein (gp268). The CkP1 LTF binding kinetics were also the glgB coding sequence by PCR and subsequently to com- equivalent to those observed in other described T-even LTF plement the mutant glgB allele. Therefore, we can currently that need accessory proteins (Marti et al. 2013). Moreover, not fully interpret the role of glgB within this context at this CkP1 LTF was able to decorate all phage-sensitive C. koseri time. The strict co-occurrence of the allele variant of glgB strains used in this study, but not the phage-resistant C. with variations in rfaF hints towards a connection between koseri mutants or non-host strains, matching its binding to these genes, but phage infection of the successful comple- the phage lytic spectrum. In-trans complementation of these mentation of rfaF variants proved that the glgB allele is not mutants again allowed cell decoration by the GFP-LTF, con- essential for resistance against CkP1. CkP1 infectivity of firming that the lack of phage infectivity of the mutants is the defective phenotypes could be fully restored by in-trans related to the phage adsorption/recognition through its LTF complementation with rfaF or gt genes from the wild-type (gp267). Of note, mutation of the gt gene did not completely host, supporting that phage CkP1 recognizes C. koseri cells abolish binding of gp267 either because the resulting modi- through the lipopolysaccharide polymer, possibly binding fication on the LPS is not substantial to hinder binding or to the inner core. gp267 binds only partially to the LPS affected by the gt Here we also describe a novel T-even LTF that functions mutation. The high affinity of this phage LTF, coupled with without the need of any accessory proteins. The T-even (e.g., the demonstrated 100% specificity and sensitivity for C. T4, T2, Bp7, and S16) LTF model is well described and koseri, shows the high potential of gp267 as a biorecogni- composed of gene products gp34 to gp37 that form the prox- tion element in the development of new diagnostic tools to imal to distal segments (King and Laemmli 1971, Dunne specifically detect this pathogenic bacterium. et al. 2018; Chen et al. 2020; Salem et al. 2021). However, in Additional tests show that the phage CkP1 has a strong the case of T4, the functional LTF (gp37) requires additional antibacterial activity against C. koseri, widening the rep- molecular chaperones (gp57A and gp38) (Hashemolhosseini ertoire of characterized Citrobacter-infecting phages pre- et al. 1996). For T2 and S16, the LTF (gp37) needs to be co- viously limited to C. freundii (e.g., CfP1, phiCFP-1, SH1- expressed with the gp38 that functions as an adhesin, attach- SH5) (Hamdi et al. 2016; Oliveira et al. 2016a, b; Zhao et al. ing to the C-terminal part of the mature LTF and modulat- 2016) and C. rodentium (CrRp3 and CrRp10) (Mizuno et al. ing the receptor specificity (Riede et al. 1985, 1987; Marti 2020) species. In comparison to other Citrobacter myovi- et al. 2013). Here we proved that the expressed proteins of ruses (CfP1 and CrRp10), CkP1 features similar infection CkP1 gp267 LTF or a truncated C-terminal part (homolog parameters but superior stability towards extreme environ- of the T-even gp37) were able to decorate C. koseri cells. mental conditions, comparable with C. freundii phage CfP1, However, the role of gp268 remains unclear (orthologue of previously isolated by our group (Oliveira et al. 2016a, b). the T-even gp38 nomenclature), which apparently does not While both withstand temperatures ranging from − 20 to need to be co-expressed with the gp267 LTF to obtain a 50 °C and pH from 3 to 11, CkP1 is also able to tolerate to functionally active LTF able to decorate cells under epif- some extent, temperatures of 60 °C after a 24-incubation luorescence microscopy. It also seems unlikely that gp267 period. Furthermore, we demonstrate that CkP1 does not contains a C-terminal chaperone, as it appears to form a nee- quickly induce resistant phenotypes and that can control C. dle structure just like the C-terminal end of gp37 of T4 LTF. koseri-driven urinary tract infections. This high stability and These results were further validated by SPR following a pro- the host range covering the whole species make CkP1 phage tocol established to measure interactions of phage-derived a promising candidate to specifically control and/or detect C. proteins with the bacterial cell walls (Loessner et al. 2002; koseri pathogens, including drug-resistant isolates. 13 3634 Applied Microbiology and Biotechnology (2023) 107:3621–3636 Supplementary Information The online version contains supplemen- Bankevich A, Nurk S, Antipov D, Gurevich AA, Dvorkin M, Kulikov tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 00253-0​ 23-1​ 2547-8. AS, Lesin VM, Nikolenko SI, Pham S, Prjibelski AD, Pyshkin AV, Sirotkin AV, Vyahhi N, Tesler G, Alekseyev MA, Pevzner PA Acknowledgements We acknowledge Tomáš Kuchta for providing (2012) SPAdes: a new genome assembly algorithm and its appli- Citrobacter strains used in our study. cations to single-cell sequencing. J Comput Biol 19(5):455–477 Bartual SG, Otero JM, Garcia-Doval C, Llamas-Saiz AL, Kahn R, Author contribution HO contributed to the design of the study, the Fox GC, van Raaij MJ (2010) Structure of the bacteriophage acquisition and interpretation of the data, and the writing of the manu- T4 long tail fiber receptor-binding tip. Proc Natl Acad Sci USA script. SS and JA contributed to the design of the study. SS, DPP, GP, 107(47):20287–20292 RD, JO, YB, RL, MS, AD, and DB acquired and interpreted the data. Bertozzi Silva J, Storms Z, Sauvageau D (2016) Host receptors for bac- All authors revised the manuscript. teriophage adsorption. FEMS microbiology letters 363(4):fnw002 Briers Y, Schmelcher M, Loessner MJ, Hendrix J, Engelborghs Y, Vol- Funding Open access funding provided by FCT|FCCN (b-on). This ckaert G, Lavigne R (2009) The high-affinity peptidoglycan bind- study was supported by the Portuguese Foundation for Science ing domain of Pseudomonas phage endolysin KZ144. Biochem and Technology (FCT) under the scope of the strategic funding of Biophys Res Commun 383(2):187–191 UIDB/04469/2020 unit. DB is supported by the Research Foundation– Chen PP, Sun HZ, Ren HY, Liu WH, Li GM, Zhang C (2020) Flanders (FWO), grant number 1S69520N. JO received a predoctoral LamB, OmpC, and the Core lipopolysaccharide of Escherichia fellowship from the UAB and a FEMS research and training grant. coli K-12 function as receptors of bacteriophage Bp7. J Virol 94(12):e00325–20 Data availability All data generated or analyzed during this study are Darling AE, Mau B, Perna NT (2010) progressiveMauve: multiple included in this published article and its supplementary information genome alignment with gene gain, loss and rearrangement. PLoS files. One 5(6):e11147 Datsenko KA, Wanner BL (2000) One-step inactivation of chromo- Declarations somal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci USA 97(12):6640–6645 Ethical approval This article does not contain any studies with human Deveci A, Coban AY (2014) Optimum management of Citrobacter participants or animals performed by any of the authors. koseri infection. Expert Rev Anti Infect Ther 12(9):1137–1142 Dunne M, Denyes JM, Arndt H, Loessner MJ, Leiman PG, Klumpp J (2018) Salmonella phage S16 tail fiber adhesin features a rare pol- Conflict of interest The authors declare no competing interests. yglycine rich domain for host recognition. Structure 26(12):1573 Evans R, Pritzel A, Antropova N, Senior A, Green T, Žídek A, Bates R, Open Access This article is licensed under a Creative Commons Attri- Blackwell S, Yim J, Ronneberger O, Bodenstein S, Zielinski M, bution 4.0 International License, which permits use, sharing, adapta- Bridgland A, Potapenko P, Cowie P, Tunyasuvunakool P, Jain P, tion, distribution and reproduction in any medium or format, as long Clancy P, Kohli P, Jumper J, Hassabis D (2022) Protein complex as you give appropriate credit to the original author(s) and the source, prediction with AlphaFold-Multimer, bioRxiv provide a link to the Creative Commons licence, and indicate if changes Hamdi S, Rousseau GM, Labrie SJ, Kourda RS, Tremblay DM, were made. The images or other third party material in this article are Moineau S, Slama KB (2016) Characterization of five podoviri- included in the article's Creative Commons licence, unless indicated dae phages infecting Citrobacter freundii. Front Microbiol 7:1023 otherwise in a credit line to the material. If material is not included in Hashemolhosseini S, Stierhof YD, Hindennach I, Henning U (1996) the article's Creative Commons licence and your intended use is not Characterization of the helper proteins for the assembly of tail fib- permitted by statutory regulation or exceeds the permitted use, you will ers of coliphages T4 and lambda. J Bacteriol 178(21):6258–6265 need to obtain permission directly from the copyright holder. To view a Hyman P, van Raaij M (2018) Bacteriophage T4 long tail fiber copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/. domains. Biophys Rev 10(2):463–471 Jumper J, Evans R, Pritzel A, Green T, Figurnov M, Ronneberger O, Tunyasuvunakool K, Bates R, Zidek A, Potapenko A, Bridgland A, Meyer C, Kohl SAA, Ballard AJ, Cowie A, Romera-Paredes B, Nikolov S, Jain R, Adler J, Back T, Petersen S, Reiman D, References Clancy E, Zielinski M, Steinegger M, Pacholska M, Berghammer T, Bodenstein S, Silver D, Vinyals O, Senior AW, Kavukcuoglu Ackermann HW (1996) Frequency of morphological phage descrip- K, Kohli P, Hassabis D (2021) Highly accurate protein structure tions in 1995. Adv Virol 141(2):209–218 prediction with AlphaFold. Nature 596(7873):583–589 Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ (1990) Basic Kanipes MI, Papp-Szabo E, Guerry P, Monteir MA (2006) Mutation local alignment search tool. J Mol Biol 215(3):403–410 of waaC, encoding heptosyltransferase I in Campylobacter jejuni Aruna K, Mobashshera T (2012) Prevalence of extended spectrum 81–176, affects the structure of both lipooligosaccharide and cap- beta-lactamase production among uropathogens in South Mumbai sular carbohydrate. J Bacteriol 188(9):3273–3279 and its antibiogram pattern. Excli J 11:363–372 King J, Laemmli UK (1971) Polypeptides of the tail fibres of bacterio- Aziz RK, Bartels D, Best AA, DeJongh M, Disz T, Edwards RA, phage T4. J Mol Biol 62(3):465–477 Formsma K, Gerdes S, Glass EM, Kubal M, Meyer F, Olsen GJ, Kortright KE, Chan BK, Turner PE (2020) High-throughput discovery Olson R, Osterman AL, Overbeek RA, McNeil LK, Paarmann D, of phage receptors using transposon insertion sequencing of bac- Paczian T, Parrello B, Pusch GD, Reich C, Stevens R, Vassieva O, teria. Proc Natl Acad Sci USA 117(31):18670–18679 Vonstein V, Wilke A, Zagnitko O (2008) The RAST server: rapid Langmead B, Salzberg SL (2012) Fast gapped-read alignment with annotations using subsystems technology. BMC Genomics 9:75 Bowtie 2. Nat Methods 9(4):357–359 13 Applied Microbiology and Biotechnology (2023) 107:3621–3636 3635 Leipold MD, Vinogradov E, Whitfield C (2007) Glycosyltransferases Riede I, Degen M, Henning U (1985) The receptor specificity of bac- involved in biosynthesis of the outer core region of Escherichia teriophages can be determined by a tail fiber modifying protein. coli lipopolysaccharides exhibit broader substrate Specificities Embo J 4(9):2343–2346 than is predicted from lipopolysaccharide structures. J Biol Chem Riede I, Drexler K, Schwarz H, Henning U (1987) T-even-type bacte- 282(37):26786–26792 riophages use an adhesin for recognition of cellular receptors. J Li H (2011) A statistical framework for SNP calling, mutation discov- Mol Biol 194(1):23–30 ery, association mapping and population genetical parameter esti- Rizvi M, Fatima N, Shukla I, Malik A (2010) Epidemiology of mation from sequencing data. Bioinformatics 27(21):2987–2993 extended spectrum -lactamases in Serratia and Citrobacter spe- Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, Marth G, cies in North India. Indian J Pathol Micr 53(1):193–193 Abecasis G, Durbin R, Genome Project Data Processing S (2009) Robinson JT, Thorvaldsdottir H, Wenger AM, Zehir A, Mesirov JP The sequence alignment/map format and SAMtools. Bioinformat- (2017) Variant review with the integrative genomics viewer. Can- ics 25(16):2078–2079 cer Res 77(21):e31–e34 Li M, Lin H, Jing Y, Wang J (2020) Broad-host-range Salmonella Salem M, Pajunen MI, Jun JW, Skurnik M (2021) T4-like bacterio- bacteriophage STP4-a and its potential application evaluation in phages isolated from pig stools infect Yersinia pseudotuberculosis poultry industry. Poult Sci 99(7):3643–3654 and Yersinia pestis using LPS and OmpF as receptors. Viruses Loessner MJ, Kramer K, Ebel F, Scherer S (2002) C-terminal domains 13(2):296 of Listeria monocytogenes bacteriophage murein hydrolases deter- Sambrook J, Fritsch ER, Maniatis T (1989) Molecular Cloning: A mine specific recognition and high-affinity binding to bacterial Laboratory Manual, 2nd edn. Cold Spring Harbor, New York cell wall carbohydrates. Mol Microbiol 44(2):335–349 Santos SB, Oliveira A, Melo LDR, Azeredo J (2019) Identification of Marti R, Zurfluh K, Hagens S, Pianezzi J, Klumpp J, Loessner MJ the first endolysin Cell Binding Domain (CBD) targeting Paeni- (2013) Long tail fibres of the novel broad-host-range T-even bac- bacillus larvae. Sci Rep 9(1):2568 teriophage S16 specifically recognize Salmonella OmpC. Mol Schattner P, Brooks AN, Lowe TM (2005) The tRNAscan-SE, snoscan Microbiol 87(4):818–834 and snoGPS web servers for the detection of tRNAs and snoR- Melo LDR, Oliveira H, Pires DP, Dabrowska K, Azeredo J (2020) NAs. Nucleic acids research 33(Web Server issue): W686–689 Phage therapy efficacy: a review of the last 10 years of preclinical Schmelcher M, Shabarova T, Eugster MR, Eichenseher F, Tchang studies. Crit Rev Microbiol 46(1):78–99 VS, Banz M, Loessner MJ (2010) Rapid multiplex detection Mirdita M, Schutze K, Moriwaki Y, Heo L, Ovchinnikov S, Steinegger and differentiation of Listeria cells by use of fluorescent phage M (2022) ColabFold: making protein folding accessible to all. Nat endolysin cell wall binding domains. Appl Environ Microb Methods 19(6):679 76(17):5745–5756 Mizuno CM, Luong T, Cederstrom R, Krupovic M, Debarbieux L, Schmelcher M, Tchang VS, Loessner MJ (2011) Domain shuf- Roach DR (2020) Isolation and characterization of bacteriophages fling and module engineering of Listeria phage endolysins for that infect Citrobacter rodentium, a model pathogen for intestinal enhanced lytic activity and binding affinity. Microb Biotechnol diseases. Viruses 12(7):737 4(5):651–662 Oliveira H, Boas DV, Mesnage S, Kluskens LD, Lavigne R, Sillankorva Sillankorva S, Neubauer P, Azeredo J (2008) Isolation and charac- S, Secundo F, Azeredo J (2016a) Structural and enzymatic char- terization of a T7-like lytic phage for Pseudomonas fluorescens. acterization of ABgp46, a novel phage endolysin with broad anti- BMC Biotechnol 8:80 gram-negative bacterial activity. Front Microbiol 7:208 Soding J, Biegert A, Lupas AN (2005) The HHpred interactive server Oliveira H, Pinto G, Oliveira A, Oliveira C, Faustino MA, Briers Y, for protein homology detection and structure prediction. Nucleic Domingues L, Azeredo J (2016b) Characterization and genome acids research 33(Web Server issue): W244–248 sequencing of a Citrobacter freundii phage CfP1 harboring a lysin Sullivan MJ, Petty NK, Beatson SA (2011) Easyfig: a genome com- active against multidrug-resistant isolates. Appl Microbiol Bio- parison visualizer. Bioinformatics 27(7):1009–1010 technol 100(24):10543–10553 Tao C, Yi Z, Zhang Y, Wang Y, Zhu H, Afayibo DJA, Li T, Tian M, Oliveira H, Pinto G, Hendrix H, Noben JP, Gawor J, Kropinski AM, Qi J, Ding C, Gao S, Wang S, Yu S (2021) Characterization of a Lobocka M, Lavigne R, Azeredo J (2017) A lytic Providencia broad-host-range lytic phage SHWT1 against multidrug-resistant rettgeri virus of potential therapeutic value is a deep-branching Salmonella and evaluation of its therapeutic efficacy in vitro and member of the T5virus genus. Appl Environ Microb 83(23): in vivo. Front Vet Sci 8:683853 e01567–17 van Kempen M, Tumescheit C, Mirdita M, Söding J, Steinegger M Pagnout C, Sohm B, Razafitianamaharavo A, Caillet C, Offroy M, (2022) Foldseek: fast and accurate protein structure search, Leduc M, Gendre H, Jomini S, Beaussart A, Bauda P, Duval JFL bioRxiv. https://​doi.​org/​10.​1101/​2022.​02.​07.​479398 (2019) Pleiotropic effects of rfa-gene mutations on Escherichia Xu L, Dong Z, Fang L, Luo Y, Wei Z, Guo H, Zhang G, Gu YQ, Cole- coli envelope properties. Sci Rep 9(1):9696 man-Derr D, Xia Q, Wang Y (2019) OrthoVenn2: a web server for Petrov VM, Ratnayaka S, Nolan JM, Miller ES, Karam JD (2010) whole-genome comparison and annotation of orthologous clusters Genomes of the T4-related bacteriophages as windows on micro- across multiple species. Nucleic Acids Res 47(W1):W52–W58 bial genome evolution. Virol J 7:292 Zhao X, Huang S, Zhao J, He X, Li E, Li H, Liu W, Zou D, Wei X, Quinlan A (2016) Bedtools version 2.26–0. https://​github.​com/​arq5x/​ Wang X, Dong D, Yang Z, Yan X, Shen Z, Yuan J (2016) Char- bedto​ols2/​relea​ses/​tag/​v2.​26.0. Accessed 04 Sept 2017 acterization of phiCFP-1, a virulent bacteriophage specific for Ranjan KP, Ranjan N (2013) Citrobacter: an emerging health care Citrobacter freundii. J Med Virol 88(5):895–905 associated urinary pathogen. Urology Annals 5(4):313–314 Riede I (1987) Receptor specificity of the short tail fibres (gp12) of Publisher's note Springer Nature remains neutral with regard to T-even type Escherichia coli phages. Mol Gen Genet 206:110–115 jurisdictional claims in published maps and institutional affiliations. Riede I, Henning U (1985) The receptor specificity of bacteriophages can be determined by a tail fiber modifying protein. Embo J 4:2343–2346 13 3636 Applied Microbiology and Biotechnology (2023) 107:3621–3636 Authors and Affiliations Hugo Oliveira1,2 · Sílvio Santos1,2 · Diana P. Pires1,2 · Dimitri Boeckaerts3 · Graça Pinto1,2 · Rita Domingues1,2 · Jennifer Otero1,4 · Yves Briers3 · Rob Lavigne5 · Mathias Schmelcher6 · Andreas Dötsch7 · Joana Azeredo1,2 Sílvio Santos Andreas Dötsch silviosantos@ceb.uminho.pt andreas.doetsch@mri.bund.de Diana P. Pires 1 Centre of Biological Engineering, University of Minho, priscilapires@deb.uminho.pt Braga, Portugal Dimitri Boeckaerts 2 LABBELS –Associate Laboratory, Braga, Guimarães, dimitri.boeckaerts@ugent.be Portugal Graça Pinto 3 Department of Biotechnology, Ghent University, Ghent, gracapinto@ceb.uminho.pt Belgium Rita Domingues 4 Departament de Genètica I de Microbiologia, Universitat ritadomingues1898@gmail.com Autònoma de Barcelona, Barcelona, Spain Jennifer Otero 5 Department of Biosystems, KU Leuven, Louvain, Belgium jenniferoterocarrera@gmail.com 6 Institute of Food, Nutrition and Health, ETH Zurich, Zurich, Yves Briers Switzerland yves.briers@ugent.be 7 Max Rubner‑Institute, Department of Physiology Rob Lavigne and Biochemistry of Nutrition, Karlsruhe, Germany rob.lavigne@kuleuven.be Mathias Schmelcher mathias.schmelcher@hest.ethz.ch 13