Identification of a human immunodominant B-cell epitope within the immunoglobulin A1 protease of Streptococcus pneumoniae

Streptococcus pneumoniae is a human pathogen causing significant morbidity and mortality worldwide. It is a transient member of the normal bacterial flora that colonizes the upper respiratory tract of the host being a major cause of various diseases such as otitis media, pneumonia, sepsis and meningitis. Despite the constant development of therapeutics, antimicrobial drugs and vaccines, pneumococcal infection still causes severe diseases in young children, elderly people and immunocompromised individuals [1, 2]. In adults, pneumococcal infection is the most common cause of community-acquired pneumonia and otitis media and, since the introduction of vaccination against Haemophilus influenzae (serotype b) and Neisseria meningitidis also the most frequent cause of meningitis.

Current immunization strategies focus on the use of S. pneumoniae polysaccharides-based vaccines, employing the 23-valent vaccine, which protects humans from two years of age, and the 7-valent toxoid-conjugated vaccine, used in children less than two years old [3, 4]. However, the available vaccines have several limitations due to the low immunogenicity of capsular polysaccharides, the high serotype variability and the genomic plasticity of this bacterium. Therefore, in the last decade there has been a great effort in developing improved vaccines to prevent pneumococcal disease.

Several surface-associated proteins, which are well conserved among the different strains of S. pneumoniae and thus represent candidates of choice for the development of novel vaccine formulations have been identified and characterized. Among this class of proteins the immunoglobulin A1 (IgA1) protease is a promising candidate since (i) it plays a major role in pathogen's resistance to the host immune response [5, 6], and (ii) it is present in all pneumococcal strains and serotypes [7, 8]. The importance of IgA protease is underlined by the fact that this host-specific enzyme is conserved in other pathogens of comparable disease and colonising similar niches [9–11].

The IgA1 protease is one of the two to four large zinc metalloproteinase present in the pneumococcal genome [7, 12]. The pneumococcal protease is a polypeptide of about 1900 amino acids associated to the bacterium via N-terminal anchoring [7, 13–15]. It is a proteolytic enzyme that specifically cleaves human IgA1 antibodies in the hinge region of the immunoglobulin heavy chain [14, 15].

Cross-inhibition experiments performed with sera from immunized rabbits have revealed considerable serological diversity of IgA1 proteases from different S. pneumoniae strains [16]. Serological analysis indicated that the sequence repeats domain of S. sanguis IgA1 protease was immunogenic in rabbits and in humans, although the antibodies recognizing this region did not inhibit enzyme activity [17]. Specific antibodies reacting with IgA1 protease have been detected in sera from patients hospitalized for pneumococcal infection [18] as well as in young children [19], highlighting the immunogenity of pneumococcal IgA1 protease in humans.

The aim of this work was to identify the immunodominant epitopes of pneumococcal IgA1 protease involved in the human antibody response against bacterial infection.

In a recent study, we isolated several antigenic regions of S. pneumoniae proteins by challenging a pneumococcal genome display library with antibodies from one patient infected by the bacterium [20]. In the present work a similar approach was employed by using "ad-hoc" patients' sera which displayed a low titer of antibodies against the pneumococcal antigens identified in the previous study (data not shown). Accordingly, sera from two patients hospitalized for pneumococcal pneumonia were used for a novel screening of the pneumococcal library. Phage pools were analyzed after two round of affinity selection and 30 out of 200 screened phage clones (15%) were isolated by immunoscreening. Among them, 9 clones (30%) contained insert encoding fragments of pneumococcal IgA1 protease and 21 clones (70%) contained insert of other bacterial proteins. At the end of the selection procedure, 3 distinct phage clones whose DNA inserts matched the sequence of the pneumococcal iga gene product were identified: clone SP1, with a DNA insert encoding for a protein fragment of 145 amino acids; clone SP2 encoding for a polypeptide of 133 residues; clone SP4 encoding for a polypeptide of 194 amino acids. As shown in Figure 1, all the protein inserts of the selected phages contain a common region of 72 amino acids corresponding to residues 420–493 of pneumococcal IgA1 protease (strain R6).

To characterize the biochemical and immunological properties of the selected protein fragments independently from the phage display context, the DNA inserts of phages SP2 and SP4 were subcloned into vector pGEX-SN [21]. This resulted in the expression of GST fusion proteins which were purified from E. coli cells under native conditions by 1-step affinity chromatography. The recombinant proteins were efficiently expressed and purified in large amounts from the cytoplasm of bacterial cells, with the yields of purified GST-SP2 and GST-SP4 being 7 and 5 mg per liter of bacterial culture, respectively. Immunoreactivity of recombinant antigens with immunoglobulins G (IgG) of human sera was examined. To this aim, an enzyme-linked immunoassay was employed (Rec-ELISA). The GST-SP2 and GST-SP4 fusion proteins were assayed with 30 sera collected from healthy adults. As a control, the IgG reactivity against wild-type GST protein was assessed for each serum. The criterion used to assign a specific reactivity against single antigens was an OD_GST-ANTIGEN greater than twice the OD_GST. Both the SP2 and SP4 polypeptides specifically reacted with more than 80% of serum samples (data not shown).

Next, the presence of immunodominant epitopes within the selected antigen fragments was investigated. To this purpose, the common region of SP2 and SP4 protein fragments, corresponding to amino acids 420–493 of pneumococcal IgA1 protease was analyzed. Figure 2 shows the alignment of this polypeptide of strain D39 to IgA proteases of other pneumococci and related species. In strain D39 such the sequence contains three 17 amino acid repeats which are part of a region of the IgA protease found to show high intraspecies conservation and a high degree of interspecies variability [22]. Pneumococcal IgA proteases fall in this region into two clusters: IgA1 proteases in genomes of strains of serotypes 1, 2, 3, 4 and 6B all have three repeats and a common downstream region while other strains share a common downstream region with S. mitis and have from two (type 23F), to seven (type 14), eight (type 19F) or even nine of these highly conserved 17 amino acid repeats. The sequences of IgA proteases of other oral streptococci such as S. sanguinis or S. oralis can not be aligned in this region since no significant homology was detected.

Two GST fusion proteins, respectively carrying residues 420–457 (epiA) and 458–493 (epiB) of the IgA1 protease from S. pneumoniae R6 strain were expressed. As shown in Figure 3, the recombinant GST-epiA and GST-epiB antigens were efficiently purified under native conditions from transformed E. coli cells. Immunoreactivity of recombinant antigens with IgG antibodies of human sera was examined and the results were compared to those obtained by use of the largest fragment of the IgA1 protease (clone SP4). The GST-epiA, GST-epiB and GST-SP4 antigens were assayed by Rec-ELISA with serum samples from the two patients hospitalized for pneumococcal pneumonia (used for the library selection) and from 46 healthy individuals. Immunoreactivity of wild-type GST was also assessed for each serum. As shown in Table 1, the recombinant GST-SP4, GST-epiA and GST-epiB proteins specifically reacted with 44 (92%), 43 (89%) and 33 (69%) of the sera, respectively, highlighting a broad antibody recognition of all antigen fragments.

Because of the broad reactivity of the epiA fragment, its recognition by antibodies present in sera from young children was also investigated. Serum samples from forty children with an age comprised between 1 and 4 years were assayed with the recombinant GST-epiA antigen by Rec-ELISA. As shown in Fig. 4, specific IgG antibodies against the epiA polypeptide were detected in sera from the majority of subjects (27 out of 40), thus demonstrating that 68% of children less than 4 years old had specific antibodies against this short polypeptide.

Next, to exclude non-specific immunoreactivity due either to altered conformation of the small polypeptides with respect to native IgA1 protease or to cross-reactivity of unrelated antibodies in human sera, a Western blot analysis was performed. To this aim, a whole cell lysate of S. pneumoniae was challenged with the affinity-purified anti-epiA antibodies. Bacterial lysates from pneumococcal iga-deficient mutant strain FP174 and from Streptococcus gordonii were assayed as controls. As shown in Figure 5, a unique protein band with an apparent molecular mass higher than 220 kDa specifically reacted with the anti-epiA antibodies in an unencapsulated strain of pneumococcus (R6). In the FP174 mutant strain, in which the iga gene was deleted, as well as in S. gordonii, such a protein band was not visible, demonstrating that the anti-epiA antibodies specifically recognized the pneumococcal enzyme.

Finally, to investigate whether the epiA polypeptide is exposed on the bacterium and is recognizable by anti-epiA antibodies, a cell-ELISA analysis was performed on whole S. pneumoniae cells. The FP174 strain was used as the negative control. The affinity-purified anti-epiA antibodies reacted with intact bacterial cells (strain R6) but not with the iga-deleted FP174 strain (data not shown), indicating that the epiA epitope within the tandem repeats of the pneumococcal enzyme is indeed exposed in the native state. Noteworthy, our results are in close agreement with a previous work where it was demonstrated that the tandem repeats of S. sanguis IgA1 protease contain immunodominant epitopes exposed in the denatured and native state [16].

Bacterial IgA1 proteases are highly specific endopeptidases that are secreted by a small number of bacteria associated with humans [23, 24]. IgA1 protease-producing bacteria include the mucosal pathogens N. meningitidis, N. gonorrhoeae, H. influenzae, S. pneumoniae, U. urealyticum and some members of pharyngeal microfloras such as Prevotella, Gemella haemolysans, S. sanguis, S. oralis and S. mitis [14, 22]. The importance of this bacterial enzyme in the interaction with the host is evidenced by the fact that most of these enzymes are specific for human IgA only, and that their impact on the host is so great as to become one of the few recognised factors of positive evolutionary selection on the human genome [25]. For the bacterial pathogens colonizing mucosal surfaces, IgA1 protease production represents a virulent factor and contributes to the pathogenesis of invasive infection by cleaving the Fc receptor from secreted immunoglobulins A1, thus preventing opsonophagocytosis [5]. IgA1 proteases have been shown to be targets of enzyme-neutralizing antibodies in serum and secretions [26], which may be induced in a state of bacterial carriage as well as during invasive infection [27].

We previously demonstrated the potential of phage-display technology in identifying B-cell epitopes recognized by the human immune response against Toxoplasma gondii and S. pneumoniae infection [20, 28–30], even though such a technique usually favors the identification of linear epitopes whereas conformational epitopes are often overlooked. Nevertheless, the present work allowed the identification of antigenic regions of the pneumococcal IgA1 protease that are broadly recognized by human antibodies. Challenging a lambda-display library of S. pneumoniae with sera from patients hospitalized for pneumococcal pneumonia enabled the selection of three distinct clones harboring a common region of the IgA1 protease. All clone sequences matched the pneumococcal enzyme encompassing amino acids 360–590. The immunoreactivity of the selected fragments, expressed as GST fusion products, was assessed; overall, IgG antibodies in serum samples from more than 90% of healthy adults reacted with at least one antigen fragment, which emphasizes the broad recognition of pneumococcal IgA1 protease by the human B cell response.

Because of the broad reactivity of the selected IgA1 protease fragments, a detailed characterization of their common region, composed by 72 amino acids, was performed. This lead to classify the sequence corresponding to residues 420–457 (epiA peptide) of the iga gene product as the most reactive polypeptide, which was recognized by antibodies present in sera from 90% of healthy adults and 68% of children with an age comprised between 1 and 4 years, thus suggesting that the human antibody response against IgA1 protease occurs during childhood. Noteworthy, our data are in agreement with a previous study showing that IgG antibodies from the majority of children less than two years old specifically reacted with pneumococcal antigens, including the IgA1 protease [19].

The evidence that affinity-purified anti-epiA antibodies also reacted with the epiB fragment (residues 458–493) indicated that a common epitope within polypeptides was recognized by human sera. Notably, a repeated 9-mer sequence VVTDKGEPE is present twice in epiA and once in epiB fragments (Fig. 1 and 2). This duplication might account for the larger reactivity of the epiA peptide with respect to the epiB peptide.

The IgA1 protease of S. pneumoniae is a major antigen expressed in all known pneumococcal strains [7]. Despite the pneumococcal enzyme has a peculiar structure among bacterial proteases, homology searches have indicated that the iga gene of S. pneumoniae is also found in other streptococci and related organisms [8, 14, 15, 22]. As already noted by others, the repeats region shows high conservation within the different pneumococcal strains and between S. pneumoniae and the most closely related S. mitis (90% identity), while homology to IgA proteases from other streptococci is significantly lower in this segment when compared to the rest of the molecule [8, 22]. The epiA peptide identified in this work is conserved in all pneumococci and in two out of three S. mitis strains [22], while it is not conserved in other oral streptococci sequenced so far. The fact that the domain covered by epiA is not shared with oral streptococci other than S. mitis suggests that this domain is probably not interchangeable, which reduces the risk of escape mutants by horizontal gene transfer. In other words, this would indicate that any reservoir of divergent IgA1 proteases in non pneumococcal species should not be imported readily by recombination into pneumococcus.

We have shown here that an immunodominant polypeptide (epiA peptide), located within the sequence repeats domain of pneumococcal IgA1 protease, is well conserved among several serotypes and strains of S. pneumoniae but not to the corresponding regions of the iga gene products from other streptococci such as S. oralis, S. sanguis and S. sanguinis. However, we also found a very high sequence homology of this protein region between S. pneumoniae and S. mitis (see Fig. 2), suggesting that the observed high rate of seropositivity against pneumococcal IgA1 protease might be due to early colonization with bacteria species other than pneumococcus.

The broad and specific recognition of the epiA polypeptide by human sera demonstrates that the pneumococcal IgA1 protease contains an immunodominant B-cell epitope. In perspective, it will be of great interest to investigate the ability of the anti-epiA antibodies in conferring protective immunity against S. pneumoniae infection in animal models. The use of phage display libraries to identify microbe or disease-specific antigens recognized by the human sera is a promising and valuable approach to epitope discovery.