FR3139630A1 - METHOD FOR IDENTIFYING CONFORMATIONAL EPITOPES - Google Patents

Abstract

METHOD FOR IDENTIFYING CONFORMATIONAL EPITOPES The present invention relates to a method for identifying a conformational epitope of a protein antigen formed from at least two distinct peptides comprising sequences from the protein antigen, comprising: - a step of selecting at least a first peptide comprising a sequence from the protein antigen recognized by at least one composition comprising antibodies or B lymphocytes carrying antibodies from at least one individual immunized against the antigen protein, and - a step of selecting at least a second peptide comprising a sequence from the protein antigen located at a distance of 3.10-9 m or less from the first peptide in a three-dimensional structure of the protein antigen and recognized by at least one composition comprising antibodies or B lymphocytes carrying antibodies from at least one individual immunized against the protein antigen of the previous step, in which the at least one first peptide and the at least one second peptide form a conformational epitope of the protein antigen. (no figure)

Description

METHOD FOR IDENTIFYING CONFORMATIONAL EPITOPES Field of invention

The present invention relates to a method for identifying conformational epitopes, as well as a method for preparing antibodies directed against said conformational epitopes.

Technical background

Interest in therapeutic monoclonal antibodies continues to grow, year after year. Indeed, their highly specific affinity for antigens allows them to provide highly effective medical treatments. Thus, from 2005 to 2017 the number of monoclonal antibodies approved by the American drug agency increased from 2 to 64.

However, there remains a critical hurdle to overcome in preparing effective monoclonal antibodies.

Indeed, the selection of B lymphocytes from which monoclonal antibodies are derived is often done by screening among all antibodies recognizing all or part of the three-dimensional structure of the targeted protein antigen, and this screening with many steps remains complex as explained in the article by Saeed et al . (2017) Front. Microbiol . 8 :495. Antibodies targeting linear epitopes are also often obtained, which generally have a lower affinity than conformational antibodies that simultaneously recognize distinct parts of the three-dimensional structure of the protein antigen. These conformational antibodies correspond to epitopes derived from the three-dimensional, tertiary or quaternary structure of proteins, which most often make it possible to obtain the best affinity for the antibodies and thus to best ensure their biological functions, in particular the neutralization of their target. This neutralization by the monoclonal antibody produced can be done by direct blocking of the protein in question by the antibody or by opsonization of the microorganism targeted by this antibody. In the latter case, the constant region of the antibodies can in particular contain particular peptide sequences allowing recognition by receptors of cells of the immune system or recognition by the complement system.

Thus, enabling the obtention of monoclonal antibodies specific to conformational epitopes, thus recognizing the three-dimensional structure of the target antigens, should improve the avidity of these antibodies, and therefore their use as tools for experimental use, as markers for diagnostic use, or as immunotherapeutic treatments.

In this context, Tsumoto et al . (2019) Immunotherapy 11 :119–127 propose to prepare monoclonal antibodies directed against conformational epitopes of an antigen by immunizing mice using DNA molecules encoding the antigen, isolating B lymphocytes and then fusing them with myeloma cells expressing this antigen to obtain hybridomas producing monoclonal antibodies specific to the three-dimensional structure of the antigen.

However, although this process does indeed allow monoclonal antibodies directed against conformational epitopes to be obtained, it is relatively complex to implement.

It therefore remains to provide a simple method to implement allowing the identification of conformational epitopes and the obtaining of monoclonal antibodies directed against these conformational epitopes.

The present invention arises from the unexpected discovery by the inventors that it was possible to identify conformational epitopes of a protein antigen by combining the screening of peptides from the protein antigen against serum from individuals immunized against the protein antigen and the localization of the peptides relative to each other in the three-dimensional structure of the protein antigen.

The present invention relates to a method for identifying a conformational epitope of a protein antigen formed from at least two distinct peptides comprising sequences derived from the protein antigen, comprising:

- a step of selecting at least one first peptide comprising a sequence derived from the protein antigen recognized by at least one composition comprising antibodies or antibody-bearing B lymphocytes derived from at least one individual immunized against the protein antigen, and

- a step of selecting at least one second peptide comprising a sequence derived from the protein antigen located at a distance of 3.10 ^-9 m or less from the first peptide in a three-dimensional structure of the protein antigen and recognized by at least one composition comprising antibodies or antibody-bearing B lymphocytes derived from at least one individual immunized against the protein antigen of the previous step,

wherein the at least one first peptide and the at least one second peptide form a conformational epitope of the protein antigen.

The present invention also relates to a plurality of distinct peptides whose respective sequences are derived from the sequence of the protein antigen, each of the distinct peptides being linked to a fluorophore having a distinct fluorescence emission wavelength.

The present invention also relates to a method for selecting lymphocytes recognizing a protein antigen from a population of cells, comprising a step of identifying at least one lymphocyte binding to at least two distinct peptides comprising sequences derived from the protein antigen forming a conformational epitope of the protein antigen, in which the conformational epitope is identified by implementing the method for identifying a conformational epitope as defined above.

The present invention also relates to a method for preparing at least one antibody, or antibody fragment, directed against a protein antigen, in which the antibody, or antibody fragment, is prepared from at least one B lymphocyte obtained by implementing the lymphocyte selection method as defined above.

Detailed description of the invention

As used herein, the term “comprising” is synonymous with “including”, “containing” or “encompassing”, i.e. when an object “comprises” one or more features, other features than those mentioned may also be included in the object. Conversely, the expression “consisting of” means “made up of”, i.e. when an object “consists of” one or more features, the object cannot include features other than those mentioned.

Composition

The at least one composition comprising antibodies or antibody-bearing B lymphocytes is of any type capable of comprising B lymphocytes or antibodies.

The at least one composition may be derived from a single individual or from multiple individuals. Preferably, the at least one composition is obtained from a biological sample or swab from one or more individuals, such as a sample or swab of whole blood, serum, ascites, or bone marrow.

Preferably, the at least one composition is a population of peripheral blood mononuclear cells (PBMCs).

The at least one composition may comprise only B lymphocytes or be enriched in B lymphocytes. This selection may in particular be carried out using ligands, in particular antibodies, targeting specific membrane markers of B lymphocytes, the antibodies being able to be coupled to magnetic beads or to luminophores, in particular fluorophores, to facilitate the detection, selection, isolation or purification of the lymphocyte-antibody complexes.

By "B lymphocyte" we mean any cell of the B lineage, such as, for example, a naive B lymphocyte, an activated B lymphocyte, a memory B lymphocyte, a plasmablast or a plasma cell, especially a long-lived one.

As will be apparent to those skilled in the art, B lymphocytes recognize protein antigen through the antibodies they carry.

Antibody

As used herein, the term "antibody" includes whole antibodies as well as antibody fragments comprising at least one antigen-binding portion, such as V _L and/or V _H , Fab, F(ab') ₂ , and scFv fragments. The antibody according to the invention may be from a single species, chimeric, humanized or even human. The antibody according to the invention may be monospecific or bispecific. Furthermore, the antibody according to the invention may be monomeric or multimeric, in particular dimeric or pentameric. The antibody according to the invention may be of isotype A, D, E, G or M, preferably G.

The antibody directed against a protein antigen is preferably an antibody, recognizing a conformational epitope of the protein antigen.

The antibody directed against a protein antigen is preferably an antibody specific for a three-dimensional structure of the protein antigen.

The antibody directed against a protein antigen is preferably a monoclonal antibody.

The antibody directed against a protein antigen can be prepared from a purified B lymphocyte, by numerous techniques well known to those skilled in the art.

For example, the B lymphocyte, possibly after clonal multiplication, can be fused with a myeloma cell to give a hybridoma producing monoclonal antibodies. Also for example, the DNA sequences coding for all or part of the antibody, in particular its variable parts, can be cloned, in particular from the messenger RNA of the lymphocyte, and then expressed, recombinantly, by culture cells, possibly after insertion into a humanized antibody structure.

Individual

Preferably, the individual immunized against the protein antigen is a non-human mammal or a human.

Immunization against the protein antigen may be natural, due to an infectious agent, such as a virus, a bacterium or a eukaryote, which carries and/or expresses the protein antigen in the individual. Immunization against the protein antigen may also be induced artificially, in particular by active immunization using a vaccine comprising the protein antigen or a part thereof or inducing the production of the protein antigen in the individual, in particular a live vaccine, in particular an attenuated one, an inactivated vaccine, a subunit vaccine, a viral vector vaccine or a DNA or RNA vaccine.

As understood here a "strong responder" is an individual whose immune response to a peptide or protein antigen is strong relative to that of other individuals belonging to the same population of individuals.

Peptides

As used herein, sequences derived from the protein antigen are portions or fragments of contiguous amino acid residue sequences of the total amino acid residue sequence of the protein antigen.

Preferably, the protein antigen sequence is fragmented into at least 2, 5, 10, 15, 20, 25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100, 200, 300, 400 or 500 distinct peptide sequences.

Preferably, the protein antigen sequence is fragmented into at most 1000, 500, 400, 300, 200, 100, 90, 80, 70, 60, 50, 45, 40, 35, 30, 25, 20, 15, 10 or 5 distinct peptide sequences.

Preferably, the at least two distinct peptides are at least three, four or five distinct peptides.

Preferably, the at least two distinct peptides are at most six, five or four distinct peptides.

Preferably, the at least two distinct peptides respectively comprise from 6 to 50 amino acid residues, 6 to 40 amino acid residues, 6 to 30 amino acid residues, 6 to 25 amino acid residues, 6 to 20 amino acid residues, 6 to 15 amino acid residues, 10 to 30 amino acid residues, 10 to 20 amino acid residues, 12 to 20 amino acid residues, 10 to 18 amino acid residues, or 10 to 15 amino acid residues.

Preferably, the at least two distinct peptides respectively comprise at least 6, 7, 8, 9, 10, 11 or 12 amino acid residues.

Preferably, the at least two distinct peptides respectively comprise at most 50, 40, 30, 25, 20 or 15 amino acid residues.

Preferably, the at least two peptides comprise non-overlapping sequences from the protein antigen. As used herein, non-overlapping sequences are such that they do not overlap within the primary structure of the antigen, they are disjoint peptides.

Preferably, the at least two peptides are at a distance from each other of 3.10 ^-9 m (30 Angstrom), 2.5.10 ^-9 (25 Angstrom) 2.10 ^-9 m (20 Angstrom), 10 ^-9 m (10 Angstrom) or less in the three-dimensional structure of the protein antigen.

The three-dimensional structure of the protein antigen can be obtained by X-ray crystallography, nuclear magnetic resonance (NMR), microscopy or by computer prediction of three-dimensional structure, for example using Alphafold 2 software (Jumper et al . (2021) Nature 596 :583–589; URL: alphafold.ebi.ac.uk).

The distance between the at least two distinct peptides can be calculated in many ways well known to those skilled in the art. It can be the average of the distances between each amino acid residue of one of the peptides from each amino acid residue of the other peptide(s). Preferably, it is the minimum distance among the respective distances between each amino acid residue of one of the peptides from each amino acid residue of the other peptide(s).

As understood here, the distance between two amino acid residues is the distance between the respective alpha carbon centers of each residue.

Antigen

The protein antigen according to the invention may be a protein or a protein complex of any type. It may in particular be a monomeric or multimeric protein, in particular homomultimeric or heteromultimeric.

Preferably, the protein antigen is composed of one or more proteins comprising at most 5000, 2500, 1000 or 500 amino acid residues.

Preferably, the protein antigen is composed of a protein or proteins comprising at least 25, 50, 75, 100 or 200 amino acid residues.

Preferably, the protein antigen is composed of a protein or proteins comprising from 25 to 5000, more preferably from 50 to 2500, even more preferably from 75 to 2000 amino acid residues.

The protein antigen may consist solely of amino acid residues or comprise aprotein components in addition to the amino acid residues. The protein may be substituted by at least one polysaccharide and/or at least one lipid. Furthermore, the protein may comprise one or more prosthetic groups, such as a heme group, a cofactor, a nucleic acid or an iron-sulfur cluster.

The protein antigen can be derived from any type of organism. Preferably, the protein antigen is derived from an infectious agent or from a human or animal protein.

The protein antigen can be obtained in particular by purification from the organism producing it naturally or by recombinant means, from cell cultures of any type (eukaryotic or prokaryotic).

Acknowledgement

Preferably, the recognition of the peptides by at least one composition comprising antibodies or antibody-bearing B lymphocytes is determined using miniaturized plates (" micro-arrays " also called ELISA chips) of peptides, or by phage expression (" phage display ") of the peptides, or by ELISA against the peptides. These techniques are well known to the person skilled in the art implementing epitope mapping .

Preferably, the peptides are labeled with different fluorophores and their recognition by at least one composition comprising antibody-bearing B lymphocytes is measured by fluorescence-assisted cell sorting (FACS). In this case, a FACS will be used allowing the simultaneous detection of at least 5, 10, 15, 20 or 25 different fluorescence wavelengths.

Preferred embodiments

In a first embodiment of the method for identifying a conformational epitope as defined above, the at least one first peptide and the at least one second peptide are selected from a plurality of distinct peptides comprising a sequence derived from the protein antigen due to their recognition by at least one composition comprising antibodies or B lymphocytes carrying antibodies derived from at least two distinct individuals immunized against the protein antigen.

Preferably, according to this first embodiment, the at least two distinct individuals are strong responders to the at least one first peptide and the at least one second peptide among a plurality of distinct individuals immunized against the protein antigen.

According to this first embodiment, an epitope mapping is implemented by ELISA chip where the protein is cut into peptides, if possible overlapping, which cover it completely and the responses of an antibody composition against the peptides can be measured simultaneously. Alternatively, a phage-display approach can be used, where the expressed sequences can correspond to the targeted proteins or can be random. In the latter case, only the areas of the targeted proteins that sufficiently resemble the sequence of peptides presented by the phages and which are identified as “positive” will be taken into consideration. The three-dimensional structure of the protein antigen (possibly several if the antigen has several possible conformations) is used to calculate the distances between the peptides. This distance can correspond to the smallest distance that separates the amino acids of the two peptides, but other methods of calculating distance can be used, for example the distance between the central amino acids of the peptides. If the three-dimensional structure of the protein is not available, it can be predicted with very reliable prediction tools such as Alphafold 2. Once these data are known, we have a distance matrix between all the peptides two by two. We can also proceed to 3x3 matrices if we want to refine the method by working directly at the three-dimensional level on triplets of disjoint peptides, or even 4x4 or more, by identifying for example all the peptides at a distance of less than 3.10 ^-9 m from each other.

We then look for peptides for which the compositions show parallel or correlated responses, which means, for example, that we find the same strong responders against these peptides within a group. Since in general, there is only a fraction of the subjects who respond strongly to a given peptide (rarely more than half or even rarely more than a third), we will look for peptides for which the "strong responders of the group" are the same. We will look for example for peptides for which the subjects located in the upper percentile in terms of antibody responses (for example 25%, 30%, or even 50%) are the same to within x individuals. The lower the number of individuals in the study, the lower x will be, the higher the number of individuals studied, the higher x can be. Typically, if we study the responses of a group of 24 individuals, we can take the top 25% (6 individuals) or the top 33% (8 individuals) with a possible flexibility of a different individual in the chosen top population, resulting in 5 or 7 strong responder individuals in common on the 6 or 8 of the top percentile respectively. If we have 60 compositions, we can take for example the top 25% or the top 30%, respectively 15 or 18 common individuals, with a flexibility of 1, 2, or 3 individuals or more, respectively 13, 14 or 15 common individuals out of 15, or 15, 16, 17, or 18 individuals out of 18. To make these calculations, we can use flexible testing software that allows us to list the number of common individuals obtained according to the pairs, or triplets, of peptides and the conditions chosen. Alternatively, one can try to use correlation coefficients between responses against peptides, but this may be less accurate unless one limits oneself to groups of patients with high response, from the high antibody response percentiles.

Following the previous operations, we will obtain a table of pairs, or triplets, of peptides with the coefficients, namely numbers of strong responders in common or correlation coefficients, for each pair of peptides. From these results, number of compositions strong responders in common or strong correlation coefficient, we can thus define the best pairs of peptides that satisfy sufficiently stringent criteria in terms of numbers of strong responders in common or in terms of high correlation coefficients. If we work on antibody responses against several proteins at the same time, one way to differentiate the best correlations of pairs of peptides from the same protein from the background noise will be to look at the best values found for pairs of peptides located on different proteins. Normally these pairs do not correspond to a conformational epitope unless the proteins form a multimer, and we can therefore take as a threshold number to consider the best value found in inter-proteins or this best value minus one.

Since the distances between peptides are known, for each protein studied, we can differentiate between pairs of disjoint peptides with a distance less than 30, 25 or 20 Angstrom ( ie . 3.10 ^-9 , 2.5.10 ^-9 or 2.10 ^-9 m), corresponding to the size of a potential antibody epitope, and those with a distance greater than 20, 25 or 30 Angstrom. As we have seen, this distance threshold is important because studies have shown that an epitope recognized by an antibody, epitope B, measures on average 10 to 15 Angstrom with a maximum of 26-30 Angstrom (Cao et al . (2011) Immunome Research 7:3:1), but we can of course vary this value to take smaller distances for example 10, 15, or 20 Angstrom, or larger distances of 30 or even 35 Angstrom. It was observed by the inventors that the number of peptide pairs with a high correlation value in the peptide pair matrix is very significantly increased in proportion for pairs with a distance less than 25 Angstrom. The only possible explanation for this enrichment in close peptide pairs, for example for a maximum distance threshold of 25 Angstrom, is that these peptide pairs correspond to peptides recognized by the same antibodies in a given serum.

Once these conformational epitopes have been identified, it is easy to clone antibodies. In fact, we can identify immunized subjects who have a common response against the two or three peptides because these are the subjects who were used to select the pair or triplet of peptides forming a conformational epitope. Among all the possible pairs of peptides, we may have a preference for peptides with higher ELISA values. Among the possible patients, we may also choose those with the highest ELISA values. Once these subjects producing conformational (and potentially neutralizing) antibodies have been identified, several experimental methods are then possible to clone antibody-producing B lymphocytes from their PBMCs, thanks to knowledge of the peptides of the conformational epitope. Any method of selecting B lymphocytes recognizing the identified conformational peptides can be used; here are two possible ones:

a. For example, their PBMCs can be transformed by the EBV virus and then the antibody-producing clones can be selected by dilution/selection cycles, each time choosing the groups of cells that produce antibodies recognizing the two, or three or four peptides of the pair, depending on the number of peptides identified in the conformational epitope. This selection can be done after obtaining a sufficient number of clones. The transformation into an antibody-producing B line can also be done by fusion with transformed lines to obtain hybridomas, according to the method well known to the person skilled in the art.

b. We can try to directly select B lymphocytes recognizing the two, three, four or five peptides, depending on the number of peptides identified in the conformational epitope, by FACS from the PBMCs of individuals after having stimulated them in culture if necessary. We will then try to label the surface antibodies of the B lymphocytes of interest with the selected peptides labeled with a fluorochrome, and to select them by FACS. We can also try to select the B lymphocytes of interest by a carpet of peptides attached to a plate by adhesion/cleaning cycles to eliminate non-adherent lymphocytes.

Once the correct B lymphocytes have been cloned, monoclonal antibodies can be produced directly by traditional culture methods, or identified by sequencing, recloned by genetic engineering in cassettes which are used to reinsert the DNA corresponding to the sequenced antibody genes into the genome of lines suitable for mass production of monoclonal antibodies.

In a second embodiment of the method for identifying a conformational epitope as defined above, the at least one first peptide and the at least one second peptide are selected from a plurality of distinct peptides comprising a sequence derived from the protein antigen due to their recognition by at least one composition comprising antibodies or antibody-bearing B lymphocytes derived from the same individual.

Preferably, according to this second embodiment, the at least one first peptide and the at least one second peptide are strongly recognized among the plurality of distinct peptides comprising a sequence derived from the protein antigen.

The first embodiment above may be less easy to apply on small numbers of individuals to be analyzed. Another approach, according to the second embodiment, is to search for the peptides against which the individuals respond significantly, for example taking the peptides for which the response in an individual is greater than the median of all the peptides plus a standard deviation, and to project them directly onto the three-dimensional structure(s) of the protein antigen. If peptides thus defined as positive in antibody response in the same individual are found to be neighbors at a distance of less than 30 Angstroms ( i.e. 3.10 ^-9 m), it can be assumed that this is a three-dimensional epitope and an attempt can be made to purify the cells of this individual which simultaneously recognize the two peptides as described above. These thresholds are arbitrary and can be modified according to the number of peptide n-tuples found to have more if there are not enough signals, or on the contrary less if there are too many n-tuples (n≥2). For example, instead of taking the median plus one standard deviation, we can take the median plus two standard deviations, or the ^75th percentile, or the ^90th percentile of the antibody responses against the peptides, etc. We can also play on the size of the epitope by going from 30 to 25, or even to 20, 15, or 10 Angstrom. These approaches can be applied to reduced numbers of individuals by looking for n-tuples of neighboring peptides (n≥2) found in all or in some of the individuals tested.

The method of searching for n-tuples (n>=2) of responder and spatially close peptides from the previous paragraph also applies when there are many individuals in a group. We can thus search for peptides significantly recognized by a large number of subjects (for example, responders against a peptide located above the ^75th percentile, or the median + one standard deviation, responses against this peptide, etc.), and look at the peptides in the neighborhood in the three-dimensional structure of the protein (less than 30 Angstroms away, or 25, 20, 15, or even 10 Angstroms): either there are also some that are well recognized and they then correspond to pairs of peptides that could potentially form a conformational epitope, or there are none in appearance, but we can then systematically test all neighboring peptides to see if the peptides of a pair are not recognized simultaneously by a B lymphocyte of the individual.

Another way to identify good n-tuples (n≥2) of peptides is to choose the peptides found for example in a group of protected individuals who therefore potentially have neutralizing antibodies, and eliminate the n-tuples of peptides that are found in a group of sick individuals whose antibodies can therefore be assumed not to have been protective. This discriminatory approach can work effectively for small groups of subjects to try to select the best n-tuples (n≥2) of peptides, i.e. the best conformational epitopes.

In some cases, ELISA data may be available for only a limited number of peptides and individuals, or even a single peptide and a single individual. In this case, if a priority peptide is chosen because a response to this peptide is expected to be observed in one or more individuals, one can simply identify the peptides present within 30 Angstroms (or 25, 20, 15, or even 10 Angstroms) of this peptide in the structure of the protein antigen, whether it is a monomer or a multimer, and systematically search for whether pairs of peptides comprising the chosen priority peptide and the neighboring peptides in the structure are recognized simultaneously by B lymphocytes of the individual.

In a third embodiment, the peptides are labeled with different fluorophores and their recognition by at least one composition comprising B lymphocytes carrying antibodies is measured by FACS.

In this case, we proceed directly to the detection of the conformational epitope on the cells of an individual in 2 steps by FACS: a first step of selecting a series of peptides labeled with fluorophores having a distinct fluorescence emission wavelength whose sequences are derived from the sequence of the protein antigen which will be tested by FACS on the cells of the individual. This step will make it possible to identify peptides recognized by the cells of the individual and to choose some of them as priorities. We can then, in a second step, zoom in on each of these priority peptides, by testing by FACS this time each of these peptides labeled with a series of peptides spatially close to this peptide and labeled with fluorophores having distinct fluorescence emission wavelengths. Typically, if we start from a protein antigen, we can initially test at least 3 peptides labeled with distinct fluorophores of maximum size 60, but minimum size 10 residues, preferably at least 15 residues, in particular at least 20 residues, possibly at least 30 residues. Of course, if the protein antigen is large, it will be interesting to cover it with a maximum of peptides to identify the maximum number of peptides recognized by the individual's cells, and we can very well test a greater number of colored peptides to the extent of the possibilities of detection of distinct fluorescence wavelengths of the FACS that we have. This will make it possible to select the peptides recognized by the individual's cells according to their interest: level of response, location in the targeted protein, etc.). For each selected peptide, we can then start testing by FACS in combination with it a series of peptides close to it spatially, at less than 25 or 30 Angstroms for example, this series typically comprising at least two peptides but this can go up to several dozen depending on the protein zone located around the selected peptide and the number of possible distinct fluorophore markings, and we can thus select from the series the peptides recognized by the individual's B lymphocytes in combination with this priority peptide, and which therefore form a conformational epitope with this peptide.

If one does not wish to choose a priority peptide, one can even try to systematically test the recognition of all pairs of distinct peptides having a spatial distance of less than 30 Angstroms, or 25, 20, 15, or even 10 Angstroms, in the structure of the protein antigen by the B lymphocytes of the individual, in a manner similar to the second embodiment except that one does this for all the peptides of the protein antigen or for peptides chosen in specific areas of the protein antigen.

The invention is further explained with the aid of the following non-limiting Examples.

EXAMPLES

Example 1: Identification of antibodies targeting a conformational epitope of the Spike (S) protein of the SARS-CoV2 virus.

Example 1 below experimentally demonstrates the existence of conformational epitopes comprising at least two peptides, detected following epitope mapping against the E, M, N, and S proteins of SARS-CoV-2.

The inventors carried out epitope mapping of the E, M, N and S proteins of the SARS-CoV-2 virus from the serum of patients infected with SARS-CoV-2 and controls using consecutive peptides of size 15 amino acids overlapping by 11 amino acids and covering these 4 proteins. There were thus 16 peptides for E, 53 peptides for M, 102 peptides for N, and 316 peptides for S. The epitope mapping was carried out on the sera of 41 uninfected controls, 27 asymptomatic subjects, 23 symptomatic subjects, and 32 subjects with severe progression.

The epitope mapping ELISA data were obtained from ELISA chips from JPT Technology, for the 3 types of immunoglobulins IgA, IgG, and IgM. The inventors cleaned the data by treating the immunoglobulins separately, according to standard procedures based on the ELISA values of the negative controls (without serum), the possible presence of batch effects and over-reactivities of non-specific sera.

From the cleaned ELISA data, the inventors searched for SARS-CoV-2 specific epitopes recognized by IgG. The inventors used a statistical t-test and a Bonferroni correction (on the number of peptides tested), and with this stringent test, the inventors identified about ten SARS-CoV-2 public linear epitopes (see Table 1).

Peptide number Peptide start End of peptide p N041 161 175 1.3E-07 N056 221 235 2.3E-05 N099 393 407 1.4E-06 N100 397 411 6.3E-05 S139 553 567 2.7E-05 S140 557 571 1.0E-05 S204 813 827 1.3E-05 S287 1145 1159 7.2E-07

Peptides corresponding to linear epitopes, which are most strongly recognized by epitope mapping

Next, the inventors sought to determine the conformational epitopes by applying the procedure according to the invention by proceeding as follows:

A. Bioinformatics calculations to identify conformational epitope pairs.

1. Calculation of the number of strong responders common between two peptides.

For each pair of the 483 peptides used in the epitope mapping , the inventors identified the number of strong responders identical between the peptides of each pair, and this for each group of patients. In the context of this example, the strong responders are chosen as being subjects in the 25% having the highest response. Other values could also be taken such as the 10% having the highest response, the 30%, or the 50%, these are choices to be made according to the objectives of selection of the pairs of peptides. Other threshold criteria can also be chosen to select the strong responders, for example an ELISA response threshold, a threshold which could possibly be adjustable according to the peptides in question, or the protein in question, or the type of immunoglobulin. For the present example, we choose as strong responders against a peptide, the subjects who are simply in the 25% having the strongest ELISA response against this peptide.

2. Calculation of distances between two peptides.

For all peptide pairs from protein S, the inventors also calculated the distance between these peptides. The distances were obtained from the 3D structure of protein S available in the Protein Data Bank (Reference 6VXX), and here, the distance measured between two peptides corresponds to the minimum distance between 2 amino acids of these peptides. There were 50086 peptide pairs tested for S (for 316 peptides covering S), among them 48827 disjoint peptide pairs, and among the latter, 7287 disjoint peptide pairs had a distance less than 20 Angstrom and 41540 disjoint peptide pairs had a distance greater than 20 Angstrom.

3. Calculation of the maximum number Nmax-group of common strong responders between two peptides in the case of pairs of peptides from different proteins, to serve as a basic control.

For pairs of peptides from different proteins, the inventors were then able to count the maximum value of the number of strong responder sera (in the strongest 25%) common to the peptides of each pair. This work is carried out separately for the data from each group of subjects: 41 uninfected controls, 27 asymptomatic subjects, 23 symptomatic subjects, and 32 subjects with severe progression. For the severe group, the maximum value of common subjects in the 8 strongest responders (25% of 32) obtained between peptides from different proteins was 6.

4. For each group, selection of pairs of peptides from protein S for which the number of common strong responders is greater than or equal to Nmax-group.

In the remainder of this example, the inventors continue with the S protein because it is the largest and it allows to demonstrate the validity of the approach undertaken to identify three-dimensional epitopes. In each of the groups of subjects, the inventors identified all the pairs of disjoint peptides for which the number of common strong responder sera was greater than the maximum value Nmax-group obtained in point 3 above. In the case of the S protein, the inventors thus obtained the table of pairs of peptides with a distance lower than the chosen threshold of 20 Angstrom, having a number of common individuals greater than or equal to 6 (maximum value found in inter-proteins E-M, E-N, E-S, M-N, M-S, N-S), here with the severe group which had a good IgG response (see Table 2).

Peptide_1 Peptide_2 Number of subjects List of patients S008 S021 6 Patient_40;Patient_34;Patient_37;Patient_38;Patient_33;Patient_41 S008 S025 6 Patient_37;Patient_41;Patient_40;Patient_33;Patient_38;Patient_34 S008 S049 6 Patient_37;Patient_34;Patient_33;Patient_38;Patient_40;Patient_41 S009 S021 6 Patient_40;Patient_34;Patient_37;Patient_38;Patient_33;Patient_41 S009 S049 6 Patient_37;Patient_34;Patient_33;Patient_38;Patient_40;Patient_41 S013 S056 6 Patient_41;Patient_43;Patient_37;Patient_38;Patient_50;Patient_36 S013 S057 6 Patient_43;Patient_50;Patient_37;Patient_38;Patient_41;Patient_36 S021 S025 6 Patient_37;Patient_41;Patient_40;Patient_33;Patient_38;Patient_34 S021 S049 6 Patient_37;Patient_34;Patient_33;Patient_38;Patient_40;Patient_41 S022 S038 6 Patient_38;Patient_37;Patient_43;Patient_51;Patient_41;Patient_34 S022 S042 6 Patient_34;Patient_43;Patient_38;Patient_51;Patient_41;Patient_37 S023 S070 6 Patient_49;Patient_38;Patient_36;Patient_37;Patient_51;Patient_41 S025 S049 6 Patient_37;Patient_34;Patient_33;Patient_38;Patient_40;Patient_41 S026 S048 6 Patient_41;Patient_34;Patient_38;Patient_50;Patient_33;Patient_37 S026 S067 6 Patient_37;Patient_34;Patient_50;Patient_33;Patient_38;Patient_41 S038 S042 6 Patient_34;Patient_43;Patient_38;Patient_51;Patient_41;Patient_37 S048 S067 6 Patient_37;Patient_34;Patient_50;Patient_33;Patient_38;Patient_41 S074 S150 6 Patient_49;Patient_36;Patient_38;Patient_50;Patient_37;Patient_43 S078 S172 6 Patient_38;Patient_43;Patient_40;Patient_37;Patient_50;Patient_41 S084 S117 6 Patient_37;Patient_51;Patient_38;Patient_41;Patient_43;Patient_36 S088 S110 6 Patient_41;Patient_34;Patient_36;Patient_38;Patient_37;Patient_51 S089 S121 6 Patient_41;Patient_50;Patient_33;Patient_34;Patient_38;Patient_37 S098 S104 6 Patient_38;Patient_43;Patient_41;Patient_37;Patient_34;Patient_50 S099 S104 6 Patient_38;Patient_43;Patient_41;Patient_37;Patient_34;Patient_50 S113 S118 6 Patient_38;Patient_37;Patient_36;Patient_51;Patient_56;Patient_41 S187 S211 6 Patient_41;Patient_38;Patient_36;Patient_43;Patient_51;Patient_37 S191 S238 6 Patient_41;Patient_33;Patient_38;Patient_50;Patient_34;Patient_36 S227 S254 6 Patient_37;Patient_34;Patient_43;Patient_36;Patient_50;Patient_56 S284 S289 6 Patient_41;Patient_37;Patient_50;Patient_49;Patient_38;Patient_42

Pairs of conformational epitopes strongly recognized by the same individuals. Note that the linear epitopes in Table 1 are not found in Table 2, suggesting a different recognition mechanism.

Interestingly, it is observed that the peptides identified in the low distance pairs do not generally correspond to the very strong linear epitopes in ELISA identified in Table 1. The responses do not go up to 60,000 as can be seen for the linear epitope peptide S in Table 1 in severe patients, but we can find responders with a still high level of antibodies against the peptides identified at 5000 or even 10,000.

5. Measurement of the enrichment of close peptide pairs (distance less than 20 Angstrom) among the pairs selected in the previous point.

The inventors selected the peptide pairs with a significant number of common strong responder sera and evaluated on these peptide pairs those with a distance greater than 20 Angstroms between the peptides, and those with a distance less than 20 Angstroms (described extensively in the Table in the previous paragraph). The inventors were able to compare the distribution of the number of peptide pairs obtained with that of all the pairs of disjoint peptides from the S protein by a Fisher exact test. Under the conditions described in the example, here are the results obtained for the different groups:

Object Value Protein S Patient group Severe Distance Threshold 20 Angstrom Total number of selected pairs (with number ≥ N-max=6) 85 Number of selected pairs < 20 Angstrom (distance threshold) 29 Number of selected pairs > 20 Angstrom 56 Overall number of protein pairs < 20 Angstrom 7287 Overall number of protein pairs > 20 Angstrom 41540 Overall number of protein pairs 48827 p -value between observed number and number in protein S (Fisher) p=10-5 GOLD 3 RR 2.28

Protein S peptide pairs selected because they were recognized by the same strong responders, and measurements of enrichment in spatially close peptides compared with all protein S peptide pairs (p-value and RR).

We see that the enrichment obtained for the group of severe patients, the one that develops the most antibodies due to a prolonged infection, is very significant: There are 29 pairs of peptides with a distance of less than 20 Angstroms out of 85 pairs of peptides recognized by the same patients (i.e. 34%), while there are 7287 pairs of peptides with a distance of less than 20 Angstroms out of 48827 pairs of disjoint peptides in the protein (i.e. 15%). This enrichment in pairs of peptides with a distance of less than 20 Angstroms is extremely significant statistically with a p value = 10 ^-5 . The enrichment in pairs of spatially close peptides corresponds to a ratio of 2.28 (ratio of fractions 29/85 and 7287/48827) instead of the value 1 expected if it were linked to chance.

If epitope mapping data are available for only one protein, one can still calculate the maximum value of common subjects across all peptide pairs and take this value as the threshold for the number of strong common responders, or this value minus 1 (or even -2 or less). It is then sufficient to look at these best values to see if there is an enrichment in peptide pairs separated by an acceptable distance for an epitope (e.g. less than 10, 15, 20, or 25 Angstroms).

For N-max-group values, a less stringent search may be desired depending on the number of individuals present in the group and the chosen threshold.

The same approach can be followed this time with peptide triplets to find epitopes involving not two but three peptides at a time. In this case, we can calculate for example the distance of the peptides two by two, or even check their presence within the same sphere of diameter less than the desired size of an epitope (for example, 10, 15, 20 or 25 Angstrom).

B. Selection of PBMCs from patients producing antibodies targeting conformational epitopes

When the inventors looked for subjects with strong common responders between two peptides in the different groups, these subjects were known by name from the data which made it possible to construct the matrix of pairs of peptides having strong responders in common.

In order to find the most interesting conformational epitopes, several parameters can be taken into account. We can try to target patients with strong responses against the two (or more) peptides identified in the conformational epitope. We can also try to favor peptides that correspond to sites known not to have mutated over time (here it is the S protein of SARS-Cov-2 which is known to mutate over time). As another example of a parameter for selecting peptides, we can also favor peptides recognized by patients with particular biological or clinical characteristics. In this example, we have chosen 3 close peptides that show a fairly high level of response: peptides S008, S021, and S049 (see Table 2). There are two reasons for choosing these peptides: there are three of them which may suggest better binding for antibodies that recognize them at the same time, the level of response against these peptides is relatively strong compared to other peptides in Table 2.

We will therefore tend to prefer pairs of peptides exhibiting high ELISA response levels, but this is not an obligation of course because there may also be epitopes recognized less strongly but effective for their neutralizing effect.

Since we know the individuals who respond to the 3 peptides (see Table 2, column 4), we will choose those who generally present the strongest response against these peptides to recover their PBMCs which will be used for the selection of B clones in the next step. In this case, these were patients 40 and 41.

C. Selection of B clones from PBMCs of subjects with antibodies recognizing the conformational epitope

There are several protocols for selecting B clones recognizing peptides. In this example, a protocol based on FACS selection was chosen. The peptides were labeled with a fluorochrome, and memory B lymphocytes recognizing the 3 peptides were recognized. For this, antibodies labeling B lymphocytes (anti-CD19) are used, and the labeled peptides are added at several concentrations ranging from 10 ng/ml to 10 microgram/ml. FACS of the cells in the presence of the memory B lymphocyte marker antibody and the peptides made it possible to see on the one hand that there were B clones labeled by the 3 peptides at the optimal concentration of 1 microgram/ml, and on the other hand to purify one by one the B cells labeled by the 3 peptides in an individual well on a microplate. The next step consisted of the amplification of the cellular RNA and the cloning of the variable fragments of the heavy chain and the light chain of the IgG into a specific expression cassette for the monoclonal antibodies. These different steps of purification by FACS and cloning are well known to the person skilled in the art and in particular described in the publication of Corsiero et al . (2016) Ann Rheum Dis . 75 (10):1866-75.

Expression tests then allowed the production of small quantities of monoclonal antibodies on twelve monoclonal antibody-producing lines. Out of 12 antibodies tested, 8 of them recognized the 3 peptides S008, S021, and S049 used for their selection as well as the Spike protein by ELISA, showing that these antibodies are indeed conformational antibodies.

Example 2

Example 2 below describes a systematic method for searching for conformational epitopes comprising at least 2 discontinuous peptides, using knowledge of the 3D structure of the Spike (S) protein of SARS-CoV-2 and using flow cytometry (FACS) analysis of cells from subjects immunized against SARS-CoV-2, using a peptide known to be immunogenic by ELISA in this individual.

In this example, the inventors took cells from an individual who had manifested a prolonged form of Covid-19, for which the inventors had sufficient samples of PBMCs taken at different times. The approach used here to search for B lymphocytes recognizing a conformational epitope is different from the previous one because the inventors only knew that the individual had been infected, and that his antibodies recognized the Spike S025 peptide corresponding to residues 97 to 111. This last peptide was biotinylated using the recommendations of the supplier Thermo ScientificTM (EZ-LinkTM Sulfo-NHS-LC-Biotinylation kit) then coupled to a fluorescent conjugate, Streptavidin-BV510™.

The inventors looked at the known structural data for the Spike protein to see which amino acids in the Spike protein were spatially close to this peptide, and the inventors investigated whether peptides containing these amino acids could be recognized by FACS along with this S025 peptide, from B lymphocytes from this individual.

Peptide S025 is in the Spike NTD region which encompasses approximately the first 290 residues of Spike. When we search for all Spike residues that are not part of peptide S025 and that are located at a distance of less than 20 Angstroms from this peptide (possible size of a conformational epitope), we actually find the entire NTD domain (with the exception of a few residues). The inventors had peptides of size 15 residues and overlapping on 11 residues covering the entire NTD region (peptides S001, S002,…S070). The inventors were thus able to select the following 24 peptides, covering the NTD region well, to see if one of them also activated the patient's memory B lymphocytes in conjunction with S025: S001, S004, S007, S010, S013, S016, S019, S021, S023, S028, S031, S034, S037, S040, S043, S046, S049, S05, S055, S058, S061, S064, S067, S070.

These 24 peptides were biotinylated using the recommendations of the supplier Thermo ScientificTM (EZ-LinkTM Sulfo-NHS-LC-Biotinilation kit) and then coupled to the fluorescent conjugate Streptavidin-phycoerythrin (PE). The inventors then tested them one by one in memory B cell labeling experiments (labeled with anti-CD19-APC and anti-CD27-BV711 antibodies), to see which ones labeled the individual's B lymphocytes in conjunction with the peptide S025. The inventors thus identified 2 other peptides that labeled B lymphocytes at the same time as S025: S021 (residues 81 to 95 of Spike) and S049 (residues 193 to 207 of Spike).

The question could be asked whether residues of each of these 3 peptides did not form the same conformational epitope (recognized by the same antibody). The inventors therefore biotinylated then coupled the peptide S049 with a third fluorochrome, FITC, and searched again by FACS in the patient's cells to see if a B lymphocyte did not recognize the 3 peptides at the same time. The inventors succeeded in identifying a memory B lymphocyte of the patient presenting the 3 colors at the same time. This shows that these 3 disjointed peptides contain residues forming the same conformational epitope. It is then easy to use a cell sorter to recover the B clones recognizing these 3 peptides at the same time, and to use the tools well known to the person skilled in the art to derive a monoclonal antibody targeting this conformational epitope.

The approach developed in this example is very interesting because it easily allows the identification of conformational epitopes (here involving the amino acids of 3 disjoint peptides), even from cells of a single individual.

It is then possible to derive monoclonal antibodies recognizing a conformational epitope by purifying by cell sorting in flow cytometry (FACS) the B cells recognizing the peptides involved in the conformational epitope (all or part of the peptides). These B cells can come from the same individual, or from other individuals who are also immunized (naturally or by vaccine) against the targeted antigen. These cells can be cultured to obtain supernatants containing antibodies that can be evaluated in vitro, then the mRNAs of the variable parts of the antibodies can be for example cloned into in-line production cassettes for the production of large quantities of monoclonal antibodies. It is also possible to directly clone the mRNAs of B lymphocytes that have been sorted by FACS into a cassette system for the production of in-line monoclonal antibodies.

Example 3

Example 3 below describes a method for direct flow cytometry (FACS) conformational epitope search based on the use of series of peptides labeled with different fluorescence colors.

This example thus shows how we can directly find a conformational epitope by screening based on series of peptides marked with different fluorescence colors, from the same targeted protein.

The inventors took the cells from the individual of Example 2, who had manifested a prolonged form of Covid-19, for which the inventors had sufficient samples of PBMCs taken at different times.

The inventors produced 6 SARS-CoV2 Spike peptides of size 20 residues covering the NTD (N terminal domain) and RBD (receptor binding domain) domains of Spike:

Peptide 1 of residues 15-34 in NTD, called P1,

Peptide 2 of residues 93-112 in NTD, called P2,

Peptide 3 of residues 217-236 in NTD, called P3,

Peptide 4 of residues 438-457 in RBD, called P4,

Peptide 5 of residues 458-477 in RBD, called P5,

Peptide 6 of residues 501-520 in RBD, called P6.

These 6 peptides were biotinylated using the recommendations of the supplier Thermo ScientificTM (EZ-LinkTM Sulfo-NHS-LC-Biotinilation kit) and coupled to fluorochromes of 6 different emission colors for FACS analysis.

In a first experiment, the inventors simultaneously tested the binding of these 6 peptides on memory B lymphocytes (identified by anti-CD19-APC and anti-CD27-BV711 antibodies, fluorochromes with emission colors different from those of the peptides).

In this first experiment, the inventors obtained 4254 memory B lymphocytes out of 181,000 PBMCs tested.

This experiment showed that the individual's memory B lymphocytes recognize 2 peptides in particular:

The P2 peptide which is recognized by 23 memory B lymphocytes, and the P4 peptide recognized by 34 memory B lymphocytes, the other peptides being recognized by less than 5 memory B lymphocytes.

Peptide P2 contains peptide S025 from Example 2. Peptide P4 corresponds to the RBM (Receptor Binding Motif) region also known to often induce antibody production in infected subjects.

To see if the results of Example 2 could be reproduced by this direct FACS approach, the inventors searched for peptides located within 10 Angstroms of the P2 peptide in the Spike structure. The inventors relied on the 15 mers (overlapping by 11 residues) spanning the entire Spike protein of Example 2. Bioinformatics analysis of the 3D structure of the Spike protein showed that there were 42 of these peptides, disjoint from the P2 peptide, located within 10 Angstroms of the P2 peptide in the 3D structure of Spike. These are the peptides S5, S6, S15, S16, S19, S20, S21, S28, S29, S30, S31, S32, S33, S34, S35, S39, S40, S41, S42, S43, S44, S45, S46, S47, S48, S49, S50, S51, S52, S53, S54, S56, S57, S58, S59, S60, S61, S63, S64, S65, S66, S67.

Since many of the 42 peptides overlapped, the inventors chose 16 that represented all of the areas of the protein covered by these 42 peptides: peptides S5, S15, S19, S28, S31, S34, S39, S42, S45, S48, S51, S54, S57, S60, S63, S66.

As was done previously for the 6 peptides of the first series, these 16 peptides were biotinylated using the recommendations of the supplier Thermo ScientificTM (EZ-LinkTM Sulfo-NHS-LC-Biotinylation kit) and coupled to fluorescent conjugates of different emission colors for FACS analysis, these colors also being different from the labeling colors of the anti-CD19 and anti-CD27 antibodies used to identify memory B lymphocytes, and from the labeling color of the P2 peptide already produced. The inventors labeled the first 8 peptides out of 16 (S5 to S42) with 8 fluorochromes, and the last 8 peptides (S45 to S66) with the same fluorochromes, and carried out the combined labeling experiments between the P2 peptide and each of these two groups of peptides.

The FACS results were as follows for the experiment with the first 8 peptides S5 to S42:

Number of memory B lymphocytes = 3750 out of 152000 cells analyzed.

Number of B cells labeled by P2 peptide only: 14

Number of B cells labeled by peptide P2 and peptide S19: 4

There were 10 B cells labeled with S19 peptide only.

For the other 7 peptides tested, no labeled cells were seen.

The FACS results were as follows for the experiment with the other 8 peptides S45 to S66:

Number of memory B lymphocytes = 3794 out of 165000 cells analyzed.

Number of B cells labeled by P2 peptide only: 19

Number of B cells labeled by peptide P2 and peptide S48: 2

There were 5 B cells labeled by the S48 peptide only.

For the other 7 peptides tested, no labeled cells were seen except for peptide S66 labeling 2 B cells.

This experiment confirms that peptides S19 and S48 can each form a conformational epitope with peptide P2. Interestingly, S19 is very close to peptide S21 identified in Example 2, and S48 is very close to peptide S49 also identified in Example 2. The inventors attempted to verify by FACS whether the S19/P2 epitope and the S48/P2 epitope were recognized by the same antibody. In an experiment similar to the previous ones, but involving only the labeled peptides P2, S19, and S48 (which carried different fluorochromes), the inventors were able to identify two cells labeled by all three peptides at the same time, indicating that it is the same conformational epitope.

This example shows that the approach based on the combined analysis of series of peptides labeled with different fluorescence colors allows the identification of conformational epitopes.

As for example 2, it is then easy, thanks to the identified peptides, to purify by FACS the B lymphocytes carrying the antibodies recognizing the desired conformational epitopes, and to proceed with the production of monoclonal antibodies according to the methods well known to the person skilled in the art.

Claims (12)

Method for identifying a conformational epitope of a protein antigen formed from at least two distinct peptides comprising sequences derived from the protein antigen, comprising:
- a step of selecting at least one first peptide comprising a sequence derived from the protein antigen recognized by at least one composition comprising antibodies or antibody-bearing B lymphocytes derived from at least one individual immunized against the protein antigen, and
- a step of selecting at least one second peptide comprising a sequence from the protein antigen located at a distance of 3.10^-9m or less of the first peptide in a three-dimensional structure of the protein antigen and recognized by at least one composition comprising antibodies or antibody-bearing B lymphocytes from at least one individual immunized against the protein antigen of the previous step,
wherein the at least one first peptide and the at least one second peptide form a conformational epitope of the protein antigen. The method of claim 1, wherein the at least one first peptide and the at least one second peptide are selected from a plurality of distinct peptides comprising a sequence derived from the protein antigen due to their recognition by at least one composition comprising antibodies or antibody-bearing B lymphocytes derived from at least two distinct individuals immunized against the protein antigen. The method of claim 2, wherein the at least two distinct individuals are strong responders to the at least one first peptide and the at least one second peptide among a plurality of distinct individuals immunized against the protein antigen. Method according to claim 1, in which the at least one first peptide and the at least one second peptide are selected from a plurality of distinct peptides comprising a sequence derived from the protein antigen due to their recognition by at least one composition comprising antibodies or antibody-bearing B lymphocytes derived from the same individual. The method of claim 4, wherein the at least one first peptide and the at least one second peptide are highly recognized among the plurality of distinct peptides comprising a sequence derived from the protein antigen. A method according to any one of claims 1 to 5, wherein the peptides comprise from 6 to 30 amino acid residues respectively. A method according to any one of claims 1 to 6, wherein the peptides comprise non-overlapping sequences derived from the protein antigen. A method according to any one of claims 1 to 7, wherein the protein antigen is derived from an infectious agent or from a human or animal protein. A method according to any one of claims 1 to 8, wherein the recognition of the peptides by at least one composition comprising antibody-bearing B lymphocytes is determined by phage expression (“phage display”) or by ELISA. Method according to any one of claims 1 to 8, in which the peptides are labeled with different fluorophores and their recognition by at least one composition comprising B lymphocytes carrying antibodies is measured by FACS. A method for selecting lymphocytes recognizing a protein antigen from a population of cells, comprising a step of identifying at least one lymphocyte binding to at least two distinct peptides comprising sequences derived from the protein antigen forming a conformational epitope of the protein antigen, in which the conformational epitope is identified by implementing the method according to any one of claims 1 to 10. Process for preparing at least one antibody, or antibody fragment, directed against a protein antigen, in which the antibody, or antibody fragment, is prepared from at least one B lymphocyte obtained by implementing the lymphocyte selection process as defined in claim 11.