Method for selecting bacteriophages

Abstract

The present invention relates to a method for selecting at least one phage capable of treating or preventing an infection or disease caused by at least one bacterium in an organism or for decolonizing an organism or decontaminating / disinfecting an environment or a surface from the at least one bacterium. Preferably, said method comprises the steps of: a) providing a plurality of genomic sequences of a plurality of phages and at least one portion of the target bacterial genome, b) determining for each phage of the plurality of phages a score obtained by evaluating at least one feature of the phage and, optionally, of the bacterium, and c) selecting from the plurality of phages at least one phage with the highest score(s) obtained in step b).

Description

[0001] Method for selecting bacteriophages

[0002] The present invention relates to a method for selecting at least one phage capable of treating or preventing an infection or disease caused by at least one bacterium in an organism or for decolonizing an organism or decontaminating / disinfecting an environment or a surface from the at least one bacterium.

[0003] BACKGROUND ART

[0004] The resistance of bacteria to antibiotics and disinfectants / decontaminants is a global public health problem. In particular, antibiotic-resistant strains of bacteria are a growing threat to animal and human health, with resistance mechanisms being identified and described for all known antimicrobials currently available for clinical use.

[0005] Alternative technologies for the biocontrol of multidrug-resistant (MDR) bacteria are being studied, but to date there are no universal strategies available for a rapid, personalized treatment of MDR infections.

[0006] Phage therapy is one of the most promising treatments for antibiotic- resistant infections. Said therapy consists in using phages to infect and kill pathogenic bacteria so as to cure the infections they cause.

[0007] Phages or bacteriophages are a wide range of viruses capable of infecting and replicating exclusively within bacterial hosts, without affecting eukaryotic cells. Bacteriophages have been shown to be safe in all tested organisms, including humans, and have been used worldwide since the 20th century.

[0008] The selectivity of phages towards a specific bacterial cell, not only at the species level but also at the strain / clone level, makes it possible to target only the pathogen responsible for the infection to be treated, without affecting commensal or environmental bacteria.

[0009] In fact, bacteriophages have a very high host specificity, both at the level of the bacterial species they infect, and sometimes, at the level of a bacterial strain within the same species; therefore, for each bacterial variant, the bacteriophage capable of attacking and killing it should be selected. For this reason, the bacteriophages should be identified and appropriately selected so as to be used as antibacterials to control / eradicate or decolonize / decontaminate / disinfect an organism or environment from a particular bacterial clone.

[0010] To date, the selection of bacteriophages has been made mostly by means of experimental tests, in particular based on in vitro antimicrobial tests.

[0011] The disadvantage of this approach is that it requires time and costs and does not consider all the interactions that the organism infected by bacteria could have with the phages.

[0012] Therefore, an effective method is needed to quickly identify ideal candidates, based on both the phage-bacteria and phage-organism interactions, for treating specific infections.

[0013] SUMMARY OF THE INVENTION

[0014] A first aspect of the present invention relates to a method for selecting at least one phage capable of treating or preventing an infection or disease caused by at least one bacterium in an organism or for decolonizing an organism or decontaminating / disinfecting an environment or a surface from at least one bacterium. In an embodiment, said method is an in silico method.

[0015] Preferably, said method comprises the steps of: a) providing a plurality of genomic sequences of a plurality of phages and at least one portion of the target bacterial genome; b) determining, for each phage of the plurality of phages, a score obtained by evaluating at least one feature of the phage genome and, optionally, of the at least one portion of the bacterial genome, and c) selecting, from the plurality of phages, at least one phage with the highest score(s) obtained in step b).

[0016] In an embodiment, the at least one feature evaluated in step b) consists of the presence of at least one molecule on the outer surface of the bacterium capable of interacting with at least one phage molecule.

[0017] In an embodiment, said step b) also comprises the evaluation of at least a second feature.

[0018] Preferably, said at least a second feature is selected from the group comprising: b2) phage virulence, where virulence means the ability of the phage to produce a lytic cycle and destroy the host bacterium; b3) at least one defense mechanism of the bacterium; b4) an immune response caused by the phage in the individual, by means of the qualitative and quantitative evaluation of the presence of gene sequences with affinity for immune defense mechanisms in the individual; and / or b5) the presence of functional genes involved in the resistance to conventional antibiotics.

[0019] Said evaluation of b5), allowing to select phages which are resistant to antibiotics, advantageously allows a combined phage and antibiotic therapy, with a synergistic effect.

[0020] Preferably, said step b) comprises the evaluation of at least two of the features b2-b5, more preferably of all the features b2-b5.

[0021] The present invention further relates to a method for preparing a phage cocktail for use in treating or preventing an infection or disease caused by at least one bacterium in an organism or for decolonizing an organism or decontaminating / disinfecting an environment or a surface from at least one bacterium.

[0022] Said method comprises providing a biobank comprising a plurality of bacteria for the expansion of the at least one phage selected according to the method of the present invention.

[0023] DEFINITIONS

[0024] In the context of the present invention, the term "organism" means an individual, i.e. , a human or animal individual, or a plant organism.

[0025] In the context of the present invention, the term "phage" or "bacteriophage" means a prokaryotic virus capable of infecting and replicating exclusively within bacterial hosts, without affecting eukaryotic cells.

[0026] In the context of the present invention, the term "lytic cycle" means a biological phage amplification cycle comprising the replication of nucleic acid (DNA or RNA) and synthesis of RNA and proteins using the organelles of the infected bacterial cell. The new virions are assembled inside the bacterial cell. At the end of the process, the bacterium undergoes lysis and the release of viruses, which then infect other bacteria.

[0027] In the context of the present invention, the term "lysogenic cycle" means a biological cycle of the phage comprising the integration of the viral DNA in the bacterial DNA.

[0028] In the context of the present invention, the term "infection" means a process characterized by penetration and multiplication of bacteria in living tissues. The concept of infection is not identified with that of infectious illness or disease since there are cases of infection without any morbid phenomenon, or asymptomatic individuals carrying the pathogen.

[0029] The term "decolonization" means the elimination of a certain bacterium which is present on the tissues of an organism even in the absence of infection.

[0030] The term "decontamination" means a process of reducing the number of pathogens from an environment or surface.

[0031] The term "disinfection" means a procedure capable of drastically reducing or better eliminating pathogens.

[0032] DETAILED DESCRIPTION OF THE INVENTION

[0033] A first aspect of the present invention relates to a method for selecting at least one phage capable of treating or preventing an infection or disease caused by at least one bacterium in an organism or for decolonizing an organism or decontaminating / disinfecting an environment or a surface from at least one bacterium. In an embodiment, said method is an in silico method.

[0034] Preferably, said method comprises the steps of: a) providing a plurality of genomic sequences of a plurality of phages and at least one portion of the target bacterial genome; b) determining, for each phage of the plurality of phages, a score obtained by evaluating at least one feature of the phage genome and, optionally, of the at least one portion of the bacterial genome; c) selecting from the plurality of phages at least one phage with the highest score(s) obtained in step b).

[0035] In an embodiment, the at least one feature evaluated in step b) consists of the presence of at least one molecule on the outer surface of the bacterium, for example on the bacterial wall, capable of interacting with at least one phage molecule.

[0036] In an embodiment, said at least one outer surface molecule of the bacterium is selected from: a polysaccharide, preferably selected from lipopolysaccharide and a membrane polysaccharide, a protein, a flagellum, and a bacterial pilus.

[0037] In an embodiment, said at least one phage molecule is a structure capable of recognizing the at least one outer surface molecule of the bacterium, receptor (receptor binding protein), and performing an interaction of the phage with the bacterium. In other words, said at least one phage molecule is an anti-receptor structure which is essential in the process of absorbing the phage on the bacterial cell.

[0038] Preferably, the plurality of phages and the at least one bacterium are analyzed separately in genomic terms and in chemical-physical terms to identify the coding regions for the at least one outer surface molecule of the bacterium and the at least one phage molecule.

[0039] Preferably, after identifying said coding regions, the interaction frequency and probability between said at least one outer surface molecule of the bacterium and said at least one phage molecule are determined. Preferably, the interaction frequency and probability between said at least one outer surface molecule of the bacterium and said at least one phage molecule are used for training a machine learning model for predicting the interaction thereof.

[0040] In an embodiment, the interaction frequency and probability between said at least one outer surface molecule of the bacterium and said at least one phage molecule are evaluated with models known to those skilled in the art. In an embodiment, the interaction frequency and probability between said at least one outer surface molecule of the bacterium and said at least one phage molecule are evaluated with a statistical predictivity model. The performance of said statistical model is preferably measured with F1 -scored or with Matthew Correlation Coefficient (MCC). Again, the ability of the model to handle new data is measured with the k-fold cross-validation technique. The k-fold cross-validation technique involves a training set being divided into k subsets, where k-1 subsets are used for training the machine learning model and the remaining subset is used for testing the machine learning model. This operation is repeated k times and the results are combined to evaluate the performance of the machine learning model. At each iteration of the model selection, the validation results are compared with previously tested configurations, and the machine learning model which produces the highest predictivity scores is stored, preferably in a computer memory.

[0041] The interaction frequency and probability between said at least one outer surface molecule of the bacterium and said at least one phage molecule are evaluated with a statistical predictivity model which preferably assigns a score between -1 and +1 to each phage of the plurality of phages based on the interaction frequency and probability.

[0042] In an embodiment, said step b) also comprises the evaluation of at least a second feature.

[0043] Preferably, said at least a second feature is selected from the group comprising: b2) phage virulence, where virulence means the ability of the phage to produce a lytic cycle and destroy the host bacterium, b3) at least one defense mechanism of the bacterium, b4) an immune response caused by the phage in the individual, by means of the qualitative and quantitative assessment of the presence of gene sequences with affinity for immune defense mechanisms in the individual, and / or b5) the resistance profile of the bacterium to conventional antibiotics given by the evaluation of the presence of functional genes involved in resistance to conventional antibiotics.

[0044] Preferably, said step b) comprises the evaluation of at least two of the features b2-b5, more preferably of all the features b2-b5.

[0045] In an embodiment, in said step b2), the virulence is evaluated in the plurality of phages, in particular, the presence in at least one phage of a gene sequence coding at least one protein or part of a protein characterizing a lytic or lysogenic biological cycle of said at least one phage. In other words, in the plurality of phages, it is evaluated which phages have a lytic biological cycle and which a lysogenic cycle. In fact, a phage with a lytic biological cycle is capable of eradicating an infection in less time.

[0046] In an embodiment, the evaluation of the presence of at least one gene sequence encoding a protein or part thereof characterizing a lytic or lysogenic biological cycle of said at least one phage, step b2), comprises the following sub-steps: b2a) analyzing the phage gene sequences; b2b) translating the phage gene sequences into all the possible amino acid sequences coded by said gene sequences; b2c) analyzing the amino acid sequences obtained in step b2b) to identify possible functional domains present in the amino acid sequences characterizing a lytic or lysogenic biological cycle of the phage, and b2d) saving the functional domains identified in step b2c. In said step b2b), the gene sequences are preferably translated into amino acid sequences using in silico techniques known to those skilled in the art. In an embodiment, said techniques for the in silico translation of amino acid sequences are for example the translation of the sequences into amino acids using all possible reading frames. In an embodiment, all the possible reading frames of the selected sequences are taken into account, discarding proteins with a length less than 40 amino acids, preferably less than 35 amino acids. In other words, all the possible amino acid sequences deriving from the analyzed gene sequences are taken into account, excluding small fragments to reduce the analysis background noise.

[0047] Therefore, in said step b2c), after having translated the phage gene sequences in all the possible amino acid sequences, the possible functional domains present in the amino acid sequences characterizing a lytic or lysogenic life cycle of the phage are preferably identified.

[0048] In an embodiment, the functional domains in the amino acid sequences are identified using databases, preferably in combination with keywords, for example, the UniProt, PFAM, Gene Onthology or NCBI databases are used. Preferably, the keywords used are integrase, excisionase and recombinase. In an embodiment, after identifying the functional domains characterizing a lytic or lysogenic life cycle of the phage, said domains are selected and encoded in features using a combination of techniques such as natural language processing (NLP) techniques, one-hot coding encodes as vectors of features and chemical-physical features, such as amino acid composition (AAC).

[0049] Preferably, said vectors of features and the chemical-physical features are used for training a machine learning model for the prediction of a lytic or lysogenic cycle of the phage.

[0050] In an embodiment, the presence in at least one phage of a gene sequence or of at least one protein characterizing a lytic or lysogenic biological cycle of said at least one phage is evaluated with a statistical predictivity model. Those skilled in the art know the statistical predictivity models applicable to this system, and how to measure the performance thereof, for example through the F1 -scored coefficients or Matthew Correlation Coefficient (MCC). The evaluation of the model is also carried out using the k-fold cross-validation technique. Said k-fold cross-validation technique is used to evaluate the best machine learning model, as described in detail above. Preferably, the results of the model evaluation are compared with previously tested configurations, and the machine learning model which produces the highest predictivity scores is stored, preferably in a computer memory.

[0051] In an embodiment, in said step b3), in the plurality of phages the presence in at least one phage of a gene sequence coding a structure or of at least one structure involved in bacterial defense mechanisms from viral infections is evaluated.

[0052] In an embodiment, the at least one structure involved in bacterial defense mechanisms from viral infections is a Clustered Regularly Interspaced Short Palindromic Repeat (CRISPR).

[0053] CRISPRs are defense systems which bacteria acquire after a first interaction with the phage to prevent recurrent interactions of the same phage, or of a phage with gene sequences shared with the phage of the first interaction. In particular, CRISPR systems are capable of “storing” small sequences of the viral genome, integrating them into the bacterial genome. Following a recurrence of the same virus, using the viral sequences integrated into the bacterial genome as a template, the CRISPR produces short RNA sequences which are complementary to the virus gene sequences and therefore cause the degradation of the phage genome. Therefore, if at least one sequence having a correlation with a phage sequence is detected in the analysis of the bacterium sequences, said bacterium has developed a mechanism to exclude the interaction of the phage, making said phage an unsuitable candidate for the treatment of the bacterial infection.

[0054] Preferably, said step b3) comprises the sub-steps of: b3a) analyzing a plurality of sequences of the bacterial genome and the phage genome; b3b) analyzing a plurality of sequences of a plurality of bacteria belonging to the same species as the bacterium of step b3a) (negative controls), b3c) analyzing the gene sequences shared between the phage and the bacterium, b3d) analyzing the gene sequences shared between the phage and the plurality of bacteria of sub-step b3b, b3e) performing an analysis between the shared sequences in step b3c) and b3d) so as to assign a correlation score, where the correlation score is -1 if there is a correlation identified in step b3c) and where the correlation score is 1 if there is a correlation identified in step b3d).

[0055] In an embodiment, said sub-step b3e) is performed by fragmenting the genome of the phage, the bacterium and the plurality of bacteria belonging to the same species of the bacterium (negative control) to obtain a library of fragments, referred to as "k-mer". Said k-mer fragments preferably have a length between 3 and 20 nucleotides, preferably between 5 and 12 nucleotides.

[0056] In an embodiment, the score is calculated using a Pearson correlation between the k-mer shared between the phage and the bacterium and between the phage and the negative controls.

[0057] In an embodiment, in said step b3), the at least one structure involved in the bacterial defense mechanisms from viral infections is a restriction enzyme, preferably an endonuclease and / or a methyltransferase. Preferably, said step b3) comprises the following sub-steps: b3a1 ) analyzing a plurality of sequences of the bacterial genome and in the phage genome; b3b1 ) analyzing the gene sequences of the bacterium coding for at least one restriction enzyme (REase), preferably for at least one endonuclease and / or a methyltransferase, b3d) analyzing the phage sequence to identify the presence of target gene sequences of the at least one restriction enzyme (RS), b3e) performing a compositional bias (CB) analysis, calculated as the ratio between the frequency observed in the sequence composition, and that expected. The expected frequency is estimated based on the observed frequencies (lowest-probability sequences - LPSs).

[0058] In an embodiment, in said step b4), the presence in at least one phage in the plurality of phages of a gene sequence coding at least one protein or of at least one protein with an affinity for immune defense mechanisms in the individual is evaluated.

[0059] Preferably, said step b4) comprises the following sub-steps: b4a) analyzing a plurality of sequences of the bacterial genome and in the phage genome; b4b) analyzing the phage gene sequences coding at least one peptide or protein potentially immunogenic for the individual, and b4c) evaluating the binding affinity of the peptide or protein to at least one system capable of recognizing non-self molecules in the body of the individual, preferably with the major histocompatibility complex type II (MHC II) or with Cytotoxic T lymphocytes (CTL). In an embodiment, the binding affinity of the peptide or protein with at least one system capable of recognizing non-self molecules in the body of the individual is assessed by fragmenting the proteins into peptides with a size between 5 and 30 amino acids, preferably between 8 and 25 amino acids.

[0060] Preferably, the binding affinity is evaluated using a prediction algorithm known to those skilled in the art, preferably using one or more algorithms selected from: NetMHCHPan. hmMHC, CTLpred, NetCTL and DeepVacPred.

[0061] In an embodiment, the method is implemented by computer for carrying out one or more of said steps b) and c).

[0062] According to the invention, an algorithm executable by said computer is arranged, comprising:

[0063] - a first software module configured to carry out step b) by calculating the score for each phage;

[0064] - a second software module configured to carry out step c) by comparing the score obtained from the different phages and determining the highest score.

[0065] The present invention further relates to a method for preparing a phage cocktail for use in treating or preventing an infection or disease caused by at least one bacterium in an organism or for decolonizing an organism or decontaminating / disinfecting an environment or a surface from at least one bacterium.

[0066] Said method comprises providing a biobank comprising a plurality of bacteria for the expansion of the at least one phage selected according to the method of the present invention.

[0067] The following examples are merely intended to disclose the invention and are not to be understood as in any way limiting it, the scope of which is defined by the claims.

[0068] EXAMPLES

[0069] Example 1 : selection of bacteriophages isolated from the environment

[0070] This example shows how the methodology has been successfully used to select bacteriophages isolated from the environment for treating a specific bacterial strain. Wastewater samples were collected and purified to be used for the isolation of bacteriophages. The purified samples were screened in vitro for the presence of bacteriophages by standard protocols (Gutierrez et al., 2010, Curr Microbiol. 61 (6):601 -8). The resulting 30 phage strains were then characterized by high-throughput sequencing methods and the resulting readings were assembled de novo with SPAdes (Philipson et al., 2018). The resulting sequences were then stored in a proprietary database for future use. To test the efficacy of the method, the 30 isolated phage strains were virtually selected against the Pseudomonas aeruginosa PAO1 strain (ATCC 10145) and then tested in vitro in a double-blind test. In particular, the ability to infect the bacterium as well as the biological cycle to predict the effectiveness of phage cocktails were evaluated. Table 1 shows the results of the analysis both in silico and in vitro and shows a strong agreement between the experimental results and the prediction.

[0071]

[0072] Example 2: exclusion of bacteria potentially resistant to selected phages.

[0073] In this example, the genome of the bacterium was analyzed to verify the presence of defense mechanisms in the bacterium which could compromise the effectiveness of the treatment with the selected phages. The entire genome of the Pseudomonas aeruginosa PAO1 strain (ATCC 10145) was analyzed for the presence of prophage genomes and three matches were found (Table 2) with high confidence scores. The in silico results were then tested with the phages in Table 1 to exclude highly similar candidates from the treatment of the infection (Table 3).

Claims

CLAIMS1. A method for selecting at least one phage capable of treating or preventing an infection or disease caused by at least one bacterium in an organism or for decolonizing an organism or decontaminating / disinfecting an environment or a surface from the at least one bacterium, said method comprising the steps of: a) providing a plurality of genomic sequences of a plurality of phages and at least one portion of the target bacterial genome; b) determining for each phage of the plurality of phages a score obtained by evaluating at least one feature of the phage genome and, optionally, of the at least one portion of the bacterial genome; c) selecting from the plurality of phages at least one phage with the highest score(s) obtained in step b), where the at least one feature evaluated in step b) is the interaction frequency and probability of at least one outer surface molecule of the bacterium and at least one phage molecule, determined by analyzing said bacterium molecule and said phage molecule.

2. A method according to claim 1 , which is an in silico method.

3. A method according to claim 1 , wherein the at least one outer surface molecule of the bacterium is selected from: a polysaccharide, preferably selected from lipopolysaccharide and a membrane polysaccharide, a protein, a flagellum and a bacterial pilus and wherein the at least one phage molecule is a receptor binding protein.

4. A method according to one of claims 1 -3, wherein the plurality of phages and the at least one bacterium are analyzed separately in genomic terms and in chemical-physical terms to identify the coding regions of said at least one outer surface molecule of the bacterium and said at least one phage molecule.

5. A method according to one of claims 1 -4, wherein said step b) also comprises evaluating at least a second feature selected from thegroup comprising: b2) phage virulence; b3) at least one defense mechanism of the bacterium; b4) an immune response caused by the phage in the individual; b5) the presence of functional genes involved in antibiotic resistance.

6. A method according to claim 5, wherein said method includes evaluating at least two of the features b2-b5, more preferably all the features b2-b5.

7. A method according to claim 5 or 6, wherein said step b2) comprises the following sub-steps: b2a) analyzing the phage gene sequences; b2b) translating the phage gene sequences into all the possible amino acid sequences coded by said gene sequences; b2c) analyzing the amino acid sequences obtained in step b2b to identify possible functional domains present in the amino acid sequences characterizing a lytic or lysogenic life cycle of the phage, b2d) saving the functional domains identified in step b2c.

8. A method according to any one of claims 1 -7, wherein said interaction frequency and probability of at least one outer surface molecule of the bacterium and at least one phage molecule, step b), and said virulence, step b2), are evaluated with a statistical predictivity model, where the assigned score varies between -1 and +1.

9. A method according to claim 5, wherein at least one of said bacterial defense mechanisms, step b3), is a Clustered Regularly Interspaced Short Palindromic Repeat (CRISPR).10.A method according to claim 9, wherein said step b3) comprises: b3a) analyzing a plurality of sequences of the bacterial genome and the phage genome; b3b) analyzing a plurality of sequences of a plurality of bacteria belonging to the same species as the bacterium of step b3a) (negative controls),b3c) analyzing the gene sequences shared between the phage and the bacterium, b3d) analyzing the gene sequences shared between the phage and the plurality of bacteria of sub-step b3b (negative controls), b3e) performing an analysis between the shared sequences in step b3c) and b3d) so as to assign a correlation score, where the correlation score is -1 if there is a correlation identified in step b3c) and where the correlation score is + 1 if there is a correlation identified in step b3d).

11. A method according to claim 5, wherein at least one of said bacterial defense mechanisms, step b3), is a restriction enzyme, preferably an endonuclease and / or a methyltransferase.

12. A method according to claim 11 , wherein said step b3) comprises: b3a1 ) analyzing a plurality of sequences of the bacterial genome and the phage genome; b3b1 ) analyzing the gene sequences of the bacterium coding for at least one restriction enzyme (REase), preferably for at least one endonuclease and / or a methyltransferase, b3d) analyzing the phage gene sequences to identify the presence of target gene sequences of the at least one restriction enzyme (RS), b3e) performing a compositional bias (CB) analysis, which is calculated as the ratio between the frequency observed in the sequence composition and the expected one, where the expected frequency is estimated based on the observed frequencies (lowest-probability sequences -LPSs).

13. A method according to any one of claims 1 -12, wherein step b4) comprises the following sub-steps: b4a) analyzing a plurality of sequences of the bacterium genome and the phage genome; b4b) analyzing the phage gene sequences coding at least one peptide or protein potentially immunogenic for the individual, and b4c) evaluating the binding affinity of the peptide or protein to at least one system capable of recognizing non-self molecules in the body of the individual, preferably with the major histocompatibility complex type II (MHCII) or with Cytotoxic T lymphocytes (CTL).

14. A method according to any one of claims 1 -13, wherein one or more of said steps b) and c) is carried out on a physical or virtual phage bank by computer.

15. A method according to claim 14, comprising the step of arranging an algorithm executable by said computer for carrying out said method, wherein said algorithm comprises:- a first software module configured to carry out said step b) by calculating said score for each phage;- a second software module configured to carry out said step c) by comparing the score obtained from the different phages and determining the highest score.

16. A method for preparing a phage cocktail for use in treating or preventing an infection or disease caused by at least one bacterium in an organism or for decolonizing an organism or decontaminating / disinfecting an environment or a surface from at least one bacterium, wherein said method comprises providing a biobank comprising a plurality of bacteria for the expansion of the at least one phage selected according to the method of one of claims 1 to 15.

Images