Due to increasing antibiotic resistance, bacterial infections have become a serious problem in hospital and community environments. A possible approach to combat such infections is phage therapy, either as an alternative to antibiotics or utilizing phage-antibiotic synergy. Experimental phage therapy proved to be successful for the treatment of bacterial infections in animal models and human patients [1
]. Interest in phage therapy has therefore increased [2
] and applications of phages are currently being investigated extensively [3
]. For the rational use of phages, it is necessary to have rapid methods of identification, particularly for new isolates as well as previously characterized phages after passaging for quality control of bacteriophage-based products.
Identification of bacteriophages is usually based on morphological characterization using electron microscopy and by genome sequencing [4
]. Multiplex PCR can be used to identify well-characterized phages with defined conserved genes typical for particular groups, for example, phages of dairy bacteria [5
] or staphylococcal phages [6
]. However, high mosaicism and modular genomic structure of tailed phages complicates their detection [7
This work focuses on the identification of staphylococcal phages that play an important role in biology, evolution and pathogenicity of staphylococci [8
]. While lytic phages shape bacterial population dynamics, temperate phages are a driving force in bacterial evolution and can benefit the host bacteria by introducing novel traits such as virulence factors as a consequence of lysogenic conversion [13
]. Bacteriophages also facilitate the horizontal transfer of bacterial DNA, including resistance genes, through transduction [14
The taxonomy of staphylococcal phages has recently been updated by the International Committee on Taxonomy of Viruses (ICTV) [17
]. All staphylococcal phages belong to the order Caudovirales,
families. Staphylococcal phages from the family Siphoviridae
are temperate and comprise 3 genera that correlate with previously described serological groups [18
]. Candidates for combating staphylococcal infections are phages belonging to Myoviridae
] as they seem unable to lysogenise host cells [11
] and recent findings suggest that lytic and polyvalent staphylococcal phages belonging to the Myoviridae
family are safe for therapeutic applications [19
Fingerprinting by means of Matrix-Assisted Laser Desorption/Ionization—Time-of-Flight Mass Spectrometry (MALDI-TOF MS) is based on rapid detection of compounds ionized directly or after minimal sample treatment. The greatest success of this method has been achieved in the field of bacterial identification [20
]. Protein profiles obtained by MALDI-TOF MS are used nowadays in thousands of clinical laboratories worldwide. Apart from bacteria, several other types of samples, including viruses, can be subjected to MALDI-TOF MS profiling analysis [21
]. However, MALDI-TOF MS profiling is not currently used for routine identification of tailed phages.
The first successful attempt at bacteriophage MALDI-TOF MS profiling was documented by Thomas et al. [22
]. This study focused on the ssRNA phage MS2 that belongs to the Leviviridae
family and infects Escherichia coli
. Treatment of MS2 phage particles, isolated by centrifugation and ultrafiltration, with 50% acetic acid was found to be necessary to obtain protein signals, as an acidic environment causes disassembly of phage protein structures. Acid treatment directly in the MALDI matrix solution was also found to be beneficial, as demonstrated by the successful generation of protein signals after preparing the phage sample in a MALDI matrix comprising 17% formic acid [23
]. To improve reproducibility and sensitivity, McAlpin et al. [25
] suggested a 10 min treatment of Yersinia pestis
podovirus φA1122 with β-mercaptoethanol to reduce inter-molecular disulphide bonds. This procedure was found to be more efficient than acid treatment.
Despite relatively long-standing knowledge that MALDI-TOF MS fingerprinting of bacteriophages was possible, most studies in the field were aimed at method optimization and involved only a limited set of strains. The largest collection of 12 strains was assessed by Bourdin et al. [26
] who used MALDI-TOF MS profiling for phage purification control after a multi-step isolation procedure. These works have demonstrated the usefulness of phage protein profiling for particular analytical purposes, however, comprehensive evaluation of sample preparation methods, MALDI-TOF mass spectral quality, reproducibility and discriminatory power have not yet been assessed critically. For this reason, we conducted a study on a relatively large set involving closely related staphylococcal phage strains, using various combinations of sample preparation techniques to establish the optimum setup and assess its applicability in identifying bacteriophages.
2. Materials and Methods
2.1. Bacterial Strains and Phages
Phages and bacterial strains used in this work are described in Table 1
. The bacteriophages from the International Typing Set for human S. aureus
and their propagation strains (PS) were obtained from Dr. P. Petráš (National Reference Laboratory for Staphylococci
, National Institute of Public Health, Prague, Czech Republic). Laboratory lysogenised strains PS 47[53+
], PS 47[77+
] were prepared previously [27
]. Phages 11, 80α and K and S. aureus
RN4220 were kindly provided by Prof. C. Wolz (Interfaculty Inst. of Microbiology and Infection Medicine Tübingen, University of Tübingen, Tübingen, Germany). Propagation strain S. aureus
for podoviruses [28
] was provided by Prof. A. Peschel (Department of Infection Biology, University of Tübingen, Tübingen, Germany). Phages Twort, 44AHJD and P68 were purchased from Félix d’Hérelle Reference Center for Bacterial Viruses (Université Laval, Québec, QC, Canada) and phage X2 from the National Collection of Type Cultures (Public Health England, Salisbury, UK). Phages B166 and B236 [29
], 131 and 812 [30
] and phage K1/420 [31
] were described previously. Phage PYO was isolated from a phage cocktail containing Staphylococcus
, E. coli
and Pseudomonas aeruginosa
phages (PYO Bacteriophagum combinierae liquidum, lot no. 000860, BioPharm, Tbilisi, Republic of Georgia). A commercial Staphylococcus
bacteriophage preparation produced by Microgen (lot no. H141, 0812, 08.14, Moscow, Russia) was purchased in a pharmacy in Samara, Russia. Bacteriophage preparation Stafal®
lot no. 140805201 was kindly provided by Bohemia Pharmaceuticals (Prague, Czech Republic). Phage preparation Duofag, staphylococcal phages SAU1 and SAU2 and P. aeruginosa
phage PAE1 were kindly provided by MB Pharma (Prague, Czech Republic). S. aureus
strains CCM 4028, CCM 4890 and CCM 8428 were obtained from the Czech Collection of Microorganisms (Masaryk University, Brno, Czech Republic).
2.2. Bacteriophage Propagation and Titration
Phages were propagated on their propagation strains (Table 1
) in a liquid medium meat-peptone broth (MPB) prepared from 13 g of nutrient broth (Oxoid, CM0001), 3 g of yeast extract (Oxoid, LP0021) and 5 g of peptone (Oxoid, LP0037) dissolved in distilled water to 1000 mL (pH 7.4). Prophage-less S. aureus
strains CCM 4890 [32
], RN4220 [33
] and CCM 8428 [34
] were used for propagation of phages with known genomic sequences to eliminate the possibility of false positive signals in MALDI-TOF MS spectra caused by induced prophages. Phages 29, 42E and 79 were propagated in a soft agar layer with 0.7% (w
) of bacteriological agar (Oxoid, LP0011) on top of solid meat-peptone agar with 1.5% (w
) technical agar (Oxoid, LP0012). Two other types of growth media were used to test the influence of medium on mass spectral quality: brain heart infusion (BHI) broth (Oxoid, CM1135; pH 7.4) and 2× yeast-tryptone broth (2YT) consisting of 16 g of tryptone (Oxoid, LP0042), 10 g of yeast extract (Oxoid, LP0021) and 5 g of NaCl dissolved in distilled water to 1000 mL (pH 7.4).
For phage titration, ten-fold serial dilutions of phage in MPB were prepared. An overnight culture of propagation strain in MPB was mixed with 1/10 volume of 0.02 M CaCl2. Bacterial suspension (0.1 mL) was added to 3 mL of 0.7% meat-peptone soft agar cooled to 45 °C and overlaid on 1.5% meat peptone agar plates. The plates were left to dry for 10 min. Phage dilutions were dropped onto the top agar and left to dry. Plates were incubated overnight at 37 °C.
2.3. Concentrating Bacteriophage to Pellets
Phage lysate, obtained after complete lysis of the bacteria, was centrifuged at 4500× g for 30 min at 4 °C and filtered through 0.45 μm pore-sized polyethersulfone syringe filters (Techno Plastic Products, Trasadingen, Switzerland) to remove bacterial debris. Phages were pelleted by centrifugation at 54,000× g for 2.5 h at 4 °C in a JA-30.50 Ti rotor (Beckman, Brea, CA, USA). For preparation of pellets from 3 mL of phage lysates, conical tubes (part no. 358119, Beckman) and adapters (part no. 358153, Beckman) with an SW 55 Ti rotor (Beckman) were used. The resulting pellet was resuspended in 350 μL of phage buffer (5 × 10−2 mol/L Tris pH 8.0, 10−2 mol/L CaCl2, 10−2 mol/L NaCl) overnight at 4 °C. For resuspending the pellets from 3 mL samples, 30 μL of phage buffer was used.
2.4. Purification of Phage Particles by CsCl Density Gradient Centrifugation
This procedure was carried out according to the description by Nováček et al. [31
]. Phage pellets resuspended in phage buffer were used as an input material for CsCl density gradient centrifugation. Soluble proteins from resuspended pellets were removed by extraction with an equal volume of chloroform. The resulting aqueous fraction (approximately 1.2 mL) was overlaid onto a preformed CsCl (Sigma-Aldrich, St. Louis, MO, USA) density gradient (1 mL of each 1.45 g/mL 1.50 g/mL, 1.70 g/mL of CsCl in phage buffer) and centrifuged at 194,000× g
for 4 h at 12 °C using a SW 55 Ti rotor (Beckman). Phage particles forming a visible zone were collected by puncturing the tube with an 0.8 mm gauge needle and syringe. Caesium chloride was removed from the phage-containing fraction by dialysis against an excess of phage buffer at 4 °C overnight using Visking dialysis tubing type 8/32”, 0.05 mm thick (part no. 1780.1, Carl Roth, Karlsruhe, Germany).
2.5. Fast Protein Liquid Chromatography (FPLC)
Bacteriophage lysates were purified using a monolithic column, CIMmultusTM
QA 1 mL (Bia separations, Ajdovščina, Slovenia) and FPLC NGC (Bio-Rad, Hercules, CA, USA). Phages were purified according to column manufacturer’s recommendation with minor modifications [35
]. After removing bacterial debris by centrifugation and filtration, 25 mL of phage lysate (108–9
PFU/mL) were mixed with 100 mM sodium phosphate buffer pH 7 at a ratio of 1:1. The column was equilibrated with 100 mM sodium phosphate buffer pH 7. The phage suspension was loaded onto the column and a linear gradient of 100 mM sodium phosphate, 2 M NaCl, pH 7 buffer was applied for elution of phages. NaCl was removed from the phage-containing fraction by dialysis as described following CsCl density gradient centrifugation.
2.6. Phage Purification by Ultrafiltration
Filtered phage lysate was used for tangential flow filtration using Pellicon XL 50 Ultrafiltration Cassettes (Millipore, Burlington, MA, USA) Biomax 300 (for Podoviridae and Myoviridae phages) and Biomax 500 (for Siphoviridae phages) according to the manufacturer’s instructions. All fluids were loaded onto the cassette with a peristaltic pump set at 40–50 mL per minute. The cassette was rinsed with 300 mL of sterile distilled water and then 300 mL of phage buffer. 350 mL of phage lysate was loaded onto the cassette and after concentrating/thickening of phages to 10–50 mL, phages were mixed with 1000 mL of phage buffer and thickened again to 10–50 mL.
2.7. Phage Precipitation by Polyethylene Glycol (PEG)
Phages were purified with PEG 8000 (Sigma-Aldrich) according to Sambrook et al. [36
] with minor modifications. NaCl and PEG 8000 were dissolved in 30 mL of filtered phage lysate to final concentrations of 0.5 M and 10 % (w
), respectively, by brief stirring. Phages were precipitated overnight at 4 °C, pelleted by centrifugation at 10,000× g
for 15 min at 4 °C and the supernatant was discarded. Pellets were dissolved in 0.5 mL of phage buffer overnight at 4 °C. Phage particles were then separated from co-precipitated bacterial debris by centrifugation at 5000× g
for 10 min at 4 °C. The residual PEG was removed by gentle extraction for 1 min with an equal volume of chloroform. The phage-containing aqueous phase was separated by centrifugation at 5000× g
for 15 min, collected and filtered through 0.22 μm pore-sized polyethersulfone syringe filters (Techno Plastic Products, Trasadingen, Switzerland).
2.8. Sample Preparation for MALDI-MS
The phage samples were mixed with a MALDI matrix solution in a 1:4 v/v ratio, the resulting mixtures were applied to three positions on the stainless steel MALDI sample plate in a volume of 0.6 μL and were allowed to dry at room temperature. Ferulic acid (FerA, 12.5 mg/mL in a water:acetonitrile:formic acid, 50:33:17, v/v mixture), or, alternatively, alpha-cyano-4-hydroxycinnamic acid (HCCA, saturated solution in water:acetonitrile:trifluoroacetic acid, 47.5:50:2.5, v/v mixture) were used as the MALDI matrix solutions.
For peptide mass fingerprinting and MS/MS analyses, the phage preparations in a volume of 5 μL were incubated with 50 ng of trypsin in 25 mM ammonium bicarbonate for two hours at 40 °C. The proteolytic mixture in a volume of 1 μL was applied to an AnchorChip MALDI sample plate, mixed with 0.6 μL of the MALDI matrix (alpha-cyano-4-hydroxycinnamic acid, 2 mg/mL in a water:acetonitrile:trifluoroacetic acid, 32:66:2, v/v mixture) and allowed to dry at room temperature.
2.9. MALDI-MS Profiling Analysis
MALDI-TOF mass spectral fingerprints were obtained using an Ultraflextreme instrument (Bruker Daltonik, Bremen, Germany) operated in the linear positive mode under FlexControl 3.4 software. External calibration of the mass spectra in the linear positive mode was performed using lysozyme (its monomer, dimer and multiply protonated ions). Laser power was set to 120% of the threshold laser power for a particular type of sample. Three independent spectra comprising 1000 laser shots each were acquired from each of the wells. Within an individual well, a minimum of 200 and a maximum of 400 shots were obtained from one position. The mass spectra were recorded within the m/z
range of 2–100 kDa. Mass spectra were processed using Flex Analysis (version 3.4; Bruker Daltonik). The MALDI-TOF mass spectra-based dendrogram was constructed using the Pearson’s product moment coefficient as a measure of similarity and the unweighted pair group average linked method (UPGMA) as a grouping method using the Biotyper 3.1 software (Bruker Daltonik). The protein fingerprints of phages with known genomic sequences were closely examined by comparing the m/z
values with the predicted molecular weight (Mw) of phage structural proteins from the NCBI Protein database or from custom RAST annotations [37
]. For matching the m/z
values to Mw of phage structural proteins, a 500 ppm tolerance was taken into account. The first amino acid of a protein was identified using TermiNator [38
] and the Mw was calculated using ExPASy ProtParam [40
2.10. MALDI-MS/MS Analysis
Analyses of digested samples were carried out using the same instrument operated in the reflectron positive mode. Seven peptide standards (Bruker Daltonik) covering the mass range of 700–3100 Da were used for external mass calibration. Peptide maps were acquired with 800 laser shots. Peaks with minimum S/N = 10 were picked out for MS/MS analysis employing the LIFT arrangement with 600 laser shots for each peptide. The MASCOT 2.2 (MatrixScience, London, UK) search engine was used for processing the MS and MS/MS data. Database searches were carried out on the NCBIprot database (non-redundant, taxonomy All Entries; downloaded from ftp://ftp.ncbi.nih.gov/blast/db/FASTA/
). For MALDI-MS data, a mass tolerance of 30 ppm was allowed for peptide mapping and 0.5 Da for MS/MS ion searches. Oxidation of methionine as an optional modification and two enzyme miscleavages were set for all searches. Peptides with statistically significant peptide scores (p
< 0.05) were considered. Manual MS/MS spectral assignment validation was carried out.
Although phage research started over 100 years ago and their therapeutic applications have been studied from the beginning [45
], necessary safety requirements for phage preparations are still being discussed [2
]. Different phage preparations are currently available on the market or under specific experimental treatment programs in Georgia, Poland, Russia, Slovakia and several clinical trials related to phage therapy have been started in the Western World. The production of bacteriophage preparations should comply with recently established requirements including phage identification as a component of quality control [2
]. MALDI-TOF MS is among the recommended methods for phage identification in master and working seed lots [46
Bacteriophage profiling by MALDI-TOF MS described in the literature has mainly been employed in bacterial identification based on phage amplification and detection of specific protein signals. Bacteriophage amplification was also found to be feasible for the identification of components in a bacterial mixture, which is difficult in direct bacterial profiling by MALDI-TOF MS. This was demonstrated by Rees and Voorhees [24
] who were able to determine both components of an E. coli
ssp. mixture by detection of proteins from phages MS2 and MPSS-1 inoculated with the bacteria. Cox et al. [47
] employed modelling of Y. pestis
phage φA1122 and E. coli
MS2 phage amplification to predict optimum growth conditions and thus to simplify the analysis workflow, which normally includes monitoring phage growth by repeated analyses at certain times. To avoid false positive identifications, undetectably low phage titres must be introduced. Therefore, Pierce et al. [48
] proposed the use of isotopically labelled 15N Siphoviridae
bacteriophage 53, whereas the presence of S. aureus
was based on detection of an unlabelled form of phage 53 capsid protein.
Over the past decade, several variants of the method based on proteomics approaches have been described: MALDI-TOF MS analysis after microwave-assisted acid digestion of MS2 phages isolated by centrifugation and ultrafiltration represents a simple and rapid method [49
]. In combination with subsequent digestion by trypsin, MALDI-TOF MS was used for detection of the amplified staphylococcal phage K, in the presence of an antibiotic, where the phage protein tryptic peptides were detected only for antibiotic resistant bacterial strains [50
]. Instead of MALDI, electrospray (ESI) has also been employed. After centrifugation, ultrafiltration and acidification, the diluted solution of MS2 protein extract yielded coat protein signals after direct injection to ESI-MS. The identity of the proteins was verified by ESI-MS/MS (Top-down) by Cargile et al. [51
] and Wick et al. [52
]. An even more sophisticated approach is represented by LC-ESI-MS/MS of phage proteins digested by trypsin [53
We have shown that pelleting of phages is time- and cost-effective and this technique can be used for sample preparation for MALDI-TOF MS even from a volume of 3 mL of phage lysate containing a sufficiently high titre (approximately 109
PFU/mL). The risk of obtaining limited analytical output when dealing with small sample volumes at lower titres should always be considered, especially when CsCl purification is not involved. The mass spectral quality should be visually monitored to avoid possible ambiguities arising from limited information gained from lower numbers of protein components detected. Similar detection limits were shown by Rees and Barr [50
]. Although pelleted phages lack the purity of phages isolated by CsCl density gradient centrifugation, the quality of the MALDI-TOF mass spectra acquired from pellets is comparable to those acquired from CsCl purified phages. At the same time, almost no differences between phages propagated in different types of media or on different propagation strains were found even when prophages were present in the genome of a propagation strain. Prophages are induced spontaneously under various stress conditions [54
]. In our previous work [57
] we showed that it was possible to detect spontaneously induced phages by PCR as they can contaminate the lysates of phages propagated on lysogenic strains. The frequency of spontaneous phage induction from lysogenic bacterial strains, as demonstrated by the appearance of infecting phage particles, has been shown to be low—around 10−8
PFU per bacterial cell [58
]. Under experimental conditions, the cell count in cultivation reaches a maximum of 108
cells per mL, therefore 103
induced phages per mL may be present. According to the experimentally set phage detection limit of MALDI-TOF MS (107
PFU/mL), induced phages cannot be detected, particularly in the presence of an excess of propagated phages. Due to ionization suppression effects, MALDI-TOF MS profiling analysis principally determinates only the majority component of the sample. Without employing other separation steps, the method cannot be used for detection of minority components, such as traces of unwanted contaminants of commercial preparations.
The results of cluster analysis of MALDI-TOF mass spectra obtained from 37 phages including 29 Siphoviridae
, 6 Myoviridae
and 2 Podoviridae
did not always correlate with their genome-based taxonomic status. Therefore, the method is not a suitable tool for classification of unknown phage strains to established phage genera. However, due to the satisfactory degree of repeatability of the mass spectra acquired from phages propagated under different conditions, the method can be used for direct phage strain identification or classification to a group of closely related ones. The possibility of reliable phage strain identification will always rely on the uniqueness of signals observed. Strain typing would be practically possible only after comprehensive examination of a wide of range of strains to affirm the specificity of the signals. It is evident from cluster analysis involving three independent cultivations of selected phages (Figure 4
), that some strains would not be distinguishable, as was demonstrated on 3A-related phages from Triavirus
genus. This is due to the fact that the variability of their mass spectra induced by the analytical method itself was greater than differences between mass spectra among different phages within a particular cluster.
Cluster analysis based on MALDI-TOF mass spectra showed that phages from the Kayvirus
genus were clustered separately from other phages. Distinguishing kayviruses from other phages is important as these phages have been successfully used in the treatment of staphylococcal infections in humans and animals [19
]. For most of the tested phages, the MALDI-TOF MS output was strain specific but kayviruses share 99% amino acid identity in most structural virion proteins, therefore they can be considered as variants of one phage strain, where individual mutant variants differ in genes encoding non-structural proteins. It is thus logical that on the basis of their MALDI-TOF MS protein profiles, these strains remain indistinguishable. MALDI-TOF mass spectra with peaks at m/z
values typical for Kayvirus
genus were obtained also from pelleted phage preparations. These signals then permitted the identification of strains using a workflow that is familiar to users of MALDI-MS systems in microbial diagnostics. Interpretation of the scoring outputs that was adopted from a routine setup used in bacterial identification represents a field for specific evaluation and probably, different score thresholds should be applied for phage analysis. Kayvirus
characteristic peaks were found even in the Duofag phage cocktail that consists of two staphylococcal phages and one P. aeruginosa
Our findings suggest that MALDI-TOF MS could be used not only for identification of laboratory cultured phages but also in the verification of phages in ready-to-use preparations. The identification success rate is dependent mostly on phage titres, while the cultivation conditions and sample purity do not play a key role. The reliability of identification of individual strains should be evaluated in comparison with other closely related strains.