Accepted Article
Received Date : 14-Jan-2014 Revised Date : 08-May-2014 Accepted Date : 10-May-2014 Article type : Original Article
Development of an enzyme-linked immunosorbent assay (ELISA) for Bartonella henselae infection detection
Fortunato Ferrara 1, Roberto Di Niro 2, Sara D’Angelo1, Marina Busetti 3, Roberto Marzari 4, Tarcisio Not 3-5, Daniele Sblattero 6*. 1. New Mexico Consortium, Los Alamos, NM, USA. 2. Department of Immunology, University of Pittsburgh, PA, USA. 3. Institute for Maternal and Child Health - IRCCS “Burlo Garofolo” Trieste, Italy. 4. Department of Life Sciences, University of Trieste, Trieste, Italy. 5. University of Trieste, Trieste Italy. 6. Department of Health Sciences and IRCAD, University of Eastern Piedmont, Novara, Italy.
* Corresponding author: Prof. Daniele Sblattero Department of Health Sciences IRCAD University of Eastern Piedmont Via Solaroli, 17 28100 Novara Italy This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an 'Accepted Article', doi: 10.1111/lam.12286 This article is protected by copyright. All rights reserved.
Accepted Article
Phone: +39 0321 660596 Fax:
+39 0321 620421
E-mail: [email protected] Running headline: Bartonella henselae antigens.
SIGNIFICANCE AND IMPACT OF THE STUDY A reliable serological assay for the diagnosis of Cat Scratch Disease (CSD) - a pathological condition caused by Bartonella henselae infection – has not yet been developed. Such an assay would be extremely useful to discriminate between CSD and other pathologies with similar symptoms but different aetiologies, e.g. lymphoma or tuberculosis. We investigate the use of two Bartonella henselae proteins – GroEL and 17-kDa – to develop a serological based ELISA, showing promising results with the potential for further development as an effective tool for the differential diagnosing of Bartonella henselae infection.
ABSTRACT Several serological diagnostics rely on enzyme-linked immunosorbent assay (ELISA) to detect bacterial infections. However for some pathogens, including Bartonella henselae, diagnosis still depends on manually intensive, time-consuming assays including microimmunofluorescence, Western blotting or indirect-immunofluorescence. For such pathogens, there is obviously still a need to identify antigens to establish a reliable, fast, and highthroughput assay (Dupon et al. 1996). We evaluated two B. henselae proteins to develop a novel serological ELISA: a well-known antigen, the 17-kDa protein, and GroEL, identified during this study by a proteomic approach.
This article is protected by copyright. All rights reserved.
Accepted Article
When serum IgG were tested, the specificity and sensitivity were 76% and 65.7% for 17-kDa, respectively, and 82% and 42.9% for GroEL, respectively. IgM were found to be more sensitive and specific for both proteins:17-kDa, specificity 86.2% and sensitivity 75%; GroEL specificity 97.7% and sensitivity 45.3%. IgM antibodies were also measured in lymphoma patients and patients with Mycobacterium tuberculosis infection to assess the usefulness of our ELISA to distinguish them from B. henselae infected patients. The resulting specificities were 89.1% and 93.5% for 17-kDa protein and GroEL, respectively. Combining the results from the two tests, we obtained a sensitivity of 82.8% and a specificity of 83.9%. Our work described and validated a proteomic approach suitable to identify immunogenic proteins useful for developing a serological test of B.henselae infection. Key words: Bartonella henselae, proteomics, GroEL, 17-kDa, antigens, cat scratch disease
INTRODUCTION B. henselae, a gram-negative bacterium of the Proteobacteria alpha 2 class (Alsmark et al. 2004), is the etiologic agent of cat scratch disease (CSD) (Dehio 2004), with an estimated incidence of 24000 people affected annually in the United States, but with the true incidence hard to calculate due to the fact that it is not a reportable disease in several states in the US (Florin et al. 2008). The infection affects the paediatric population (80% children aged 2-14 years) and young adults predominantly (Jacomo et al. 2002). The cat flea, Ctenocephalides felis, is responsible for spreading B. henselae among cats, the main reservoir for the microorganism, that can enter the host through a cat scratch or bite, potentially causing acute and chronic infections, with manifestations depending on the status of the host immune system (Chomel et al. 2003). In immunocompetent patients, CSD is
This article is protected by copyright. All rights reserved.
Accepted Article
frequently characterized by regional lymph-node enlargement and by the presence of redbrown skin papules at the site of inoculation. Patients typically remain afebrile, and uncomplicated CSD-mediated lymphadenopathy spontaneously resolves within 2-6 months. In immunocompromised patients bacillary angiomatosis and peliosis are typical features, accompanied by multi-organ infection (Massei et al. 2005). In the differential diagnosis of CSD, all possible causes of lymphadenopathy, a common paediatric problem, must be considered. Lymphadenopathy is usually due to a transitory response to an upper respiratory tract infection. In sub-acute or chronic cases, the most common culprits are gram-positive bacteria and, less commonly, Mycobacterium tuberculosis. Sometimes, however, lymphadenopathy is a clinical sign of a more serious illness, such as lymphoma (Massei et al. 2000; Ghez et al. 2001). For this reason, a non-invasive, specific diagnostic assay for B.henselae infection would be of great help in clinical practice. Rapid identification of infectious agents is critical to be able to respond promptly, to prevent disease spread, and initiate patient treatment. For many infections traditional detection methods involve time-consuming bacterial growth in culture media, isolation, biochemical identification, and, sometimes, serology. To illustrate a proteomic approach to developing a rapid diagnostic assay for a specific infectious agent, Bartonella henselae, for which there is no straightforward reliable diagnostic assay, was chosen as a model. Current methods for CDS diagnosis consist in the isolation of the bacteria from biopsy samples, an invasive and lengthy procedure (La Scola and Raoult 1999; Klotz et al. 2011). Techniques, such PCR amplification, are also biopsy-based (Anderson et al. 1994; Chung et al. 2005), and require unique expertise. Serological diagnosis, usually based on an indirect immunofluorescence assay using irradiated whole-cell antigen preparation from co-cultivated Vero cells (Zbinden 1998), is sensitive, but costly, and cross-reactivity has been reported with other species (La Scola and Raoult 1996; McGill et al. 1998). Whole-cell antigen ELISAs
This article is protected by copyright. All rights reserved.
Accepted Article
give relatively poor sensitivity and specificity (Vermeulen et al. 2007; Tsuruoka et al. 2012), suggesting that the use of highly purified bacterial proteins may provide improved sensitivity and specificity results. Follow-up of antibody titres against B. henselae could also help in assessing definitive resolution of the infection (Lesprit et al. 2003). Finding new target antigens for the serological diagnosis of B. henselae infection is therefore of considerable interest. In this study, we evaluated two suitable candidates for new ELISA assay: a well-described 17-kDa protein (Anderson et al. 1995; Sweger et al. 2000; Loa et al. 2006; Hoey et al. 2009)with homology to proteins of the virB operon, associated with the secretion of polypeptides or nucleic acids (Padmalayam et al. 2000) - and an antigen isolated by a proteomic approach and identified as the GroEL chaperon protein.
RESULTS and DISCUSSION
Antigenic protein identification Several proteins involved in the immunologic reactions against B. henselae have been recently identified (McCool et al. 2008; Eberhardt et al. 2009). We used an immunoproteomic approach for the identification of a novel B. henselae-derived immuno-dominant antigenic protein. After preparing and subjecting whole bacterial extracts to 2D-PAGE, sera from eight patients with CSD diagnosis, without hepatosplenic involvement, were pooled and used to probe blotted proteins. A comparison between silver stained gel (Fig 1 A) and WB (Fig 1 B) showed a limited number of spots labelled by the pooled IgM serum antibodies. Immunoreactive spots did not correspond to major stained polypeptides, attesting to high specificity in serum antibody recognition. A heavily stained protein with approximately a MW of 60 kDa was identified (pointed by the arrow in Fig 1B) and selected for further
This article is protected by copyright. All rights reserved.
Accepted Article
analyses. Interestingly, the immunoblot identified additional polypeptides, stained in the 1520 kDa MW range, a protein size that likely represents the well-known 17-kDa antigen.
LC-MS/MS Analysis The identified 60 kDa protein was judged as a putative immunodominant antigen, and the protein spot excised from the silver stained gel subjected to in-gel digestion. Forty-seven different peptide sequences obtained by MS/MS spectra analyzed by BLAST, showed high homology to peptides from B.henselae GroEL (accession number:O33963), covering 66% of the entire protein sequence, allowing us to positively identify the protein as GroEL (Fig 1 C). GroEL belongs to the highly conserved family of Heat Shock Proteins (HSPs). These proteins have crucial functions in cell homeostasis, protein folding (Smith et al. 1998), and play key roles in cytoprotection. HSP synthesis is extremely important during an infection, because of the challenging environmental changes in pathogen living conditions, making them an important target in the defence against infectious agents (Srivastava 2002), with immune response to HSPs observed in disease caused by bacteria, protozoa, fungi and nematode infection (Zugel and Kaufmann 1999). In particular GroEL is a chaperonin, forming an oligomeric complex with GroES (HSP10), which plays a key role in the folding and assembly of proteins (Mayhew et al. 1996; Haake et al. 1997; Marston et al. 1999; Minnick et al. 2003; Tang et al. 2006). GroEL has also been described as an immunogen in other microorganisms, including Bartonella bacilliformis (Knobloch and Schreiber 1990), and it seems that the immune response is usually directed against non-conserved regions (Lamb et al. 1989; Young and Garbe 1991). For this reason portion or epitopes of species specific GroEL, lacking homology to the human protein, have been considered as vaccine candidates (Pechine et al. 2013).
This article is protected by copyright. All rights reserved.
Accepted Article
Cloning, expression and purification of GroEL and 17-KDa proteins In order to assess the value of GroEL as a potential diagnostic, the gene was cloned and expressed as a recombinant protein in E. coli, and purified by affinity chromatography on NiNTA resin. Eluted fractions were pooled and checked for purity by SDS/PAGE followed by Coomassie-blue staining (Fig 2, lane A), showing a high degree of purity. WB analysis was performed using the His-tag at the N-terminus of the GroEL protein for detection. The result (Fig 2, lane C) confirms the recognition of a protein at the expected molecular weight (60 kDa) for the full recombinant protein. The 17-kDa antigen (Anderson et al. 1995) has also been previously proposed for ELISAbased diagnostic tests. Notwithstanding cross-reactivity that has been reported between the B. henselae 17-kDa protein and the serum of B. quintana infected patients, the positive ability of a 17-kDa-based ELISA assay to detect B. henselae-specific antibody in human samples (Loa et al. 2006) has been demonstrated. We also cloned the B. henselae 17-kDa gene to provide a positive control, a reference for the GroEL immunoreactivity test, and also to use more than one target to improve the detection of anti-B. henselae antibody in patients (Loa et al. 2006). The 17-kDa protein is scarcely soluble (Loa et al. 2006), and this was confirmed in preliminary experiments (data not shown), where the 17-kDa gene cloned into the pTrcHisB vector yielded a limited amount of recombinant protein, due to its insolubility and its deposit in inclusion bodies (Loa et al. 2006). Consequently, we used an alternative approach cloning the gene in the expression vector pMAL-c2X, to allow the expression as a fusion with the E. coli maltose binding protein (MBP). The 17-kDa gene was obtained by PCR amplification and cloned into the pMAL-c2X plasmid. Correct cloning was confirmed by DNA sequencing. The recombinant fusion protein was expressed, extracted and purified by affinity chromatography on amylose resin. The eluted fractions were pooled and checked by SDS/PAGE followed by staining with
This article is protected by copyright. All rights reserved.
Accepted Article
Coomassie-blue (Fig 2, lane B). The recombinant protein shows the expected molecular weight (~60 kDa: 17-kDa protein fused to the 42.5 kDa MBP and SV5 tag). with a very low level of degradation products. This was further confirmed by WB using a mAb recognizing the SV5 tag mapping to the C-terminus of the recombinant protein (Fig 2 lane D).
Antigenic reactivity A preliminary validation of the antigenic properties of the recombinant proteins was performed by WB analysis. A pool of eight sera from clinical cases (no patients with hepatosplenic involvement) of CSD with a positive IFA test was compared with a pool of sera from healthy donors. The results demonstrate that both GroEL and 17-kDa recombinant proteins are specifically recognized by IgM serum antibodies (Fig 2, lane E: GroEL; lane F: 17-kDa protein), whereas sera from controls (Fig 2, lanes G: GroEL; lane H: 17-kDa protein) show no reactivity. Sera were also tested against a control recombinant protein formed by MBP and the SV5 tag (MBP-SV5) to confirm the specificity of the binding to 17-kDa. No reactivity was detected, thus, demonstrating that the binding was specific for the 17-kDa target protein (data not shown). Several bacterial species – in particular within the Bartonella genus – have proteins with high homology to the two targets selected in this study (Fig 3). In particular, GroEL is one of the most conserved proteins in nature, with high sequence homology between mammals and microbes, and is frequently found to be immunogenic in a variety of bacterial and parasitic infection (Young et al. 1988; Shinnick 1991). However, the high level of shared homology between different GroEL proteins does not appear to prevent specific humoral response against species-specific HSP, as shown when similar proteins have been used as vaccination targets and species-GroEL specific antibodies were produced with no autoimmune reaction against the host GroEL (Noll and AutenriethIb 1996). For this reason we decided to use the identified GroEL protein in our ELISA assay.
This article is protected by copyright. All rights reserved.
Accepted Article
ELISA results ELISA assay has a number of advantages for serodiagnosis, including sensitivity and ease of automation. Preliminary experiments were conducted before performing our ELISA assay to establish serum dilution and the amount of antigen to be coated on the plate. When both recombinant proteins were coated at 10 μg/ml and sera (individual and pools of patients and control samples) were diluted at 1:100 we obtained a good signal-to-noise ratio (data not shown). A total of 64 patients and 87 controls sera were used. ELISA results were analysed by ROC curves. The calculated IgG response against 17-kDa protein resulted in a specificity of 76.0% and a sensitivity of 65.7% with a corresponding AUC of 0.769, confident interval (CI) 0.664÷0.853 (Fig 4 A-B). When the reactivity against GroEL protein was calculated, the specificity was 82.0% and sensitivity 42.9%, with an AUC of 0.631, CI 0.520÷0.733 (Fig 4C-D). We then focused on the IgM response, which gave a reactivity against the 17-kDa-protein with a specificity of 86.2%, sensitivity of 75.0%, and an AUC of 0.848, CI 0.781÷0.901 (Fig 4E-F), while the anti-GroEL response gave a specificity of 97.7%, sensitivity of 45.3%, and an AUC of 0.781, CI 0.707÷0.844 (Fig 4G-H). For IgM in both cases, the AUC parameters confirmed the ability of the assay to discriminate between diseased and healthy cases. The evaluation of the two antibody class responses to the two antigenic proteins showed IgM as the best choice for an ELISA-based diagnostic assay, with sensitivity and specificity values for 17-kDa almost identical to those reported in other studies (Bergmans et al. 1997). IgM are usually short lived and replaced by IgG, but the nature of IgM - as first immunoglobulin responders in infectious disease - can explain their better performance in our ELISA assay in this study where the subjects were most likely diagnosed during the acute phase of B. henselae infection. Despite these promising values, they were not deemed sufficient for an effective diagnostic test. When tested alone, the GroEL specificity was excellent (97.7%) but the sensitivity was
This article is protected by copyright. All rights reserved.
Accepted Article
disappointing (45%). For CDS diagnosis the use of ELISA based on a single antigen may be optimistic, especially since recent profiling of the feline humoral immune response (Vigil et al. 2010), demonstrated that a classifier with multiple antigens showed improved performances. Therefore combining different antigenic proteins with the well-known 17-kDa antigen (Loa et al. 2006; Hoey et al. 2009) could improve B.henselae infection diagnosis. For this reasons we combined the results of the two tests, obtaining a sensitivity of 82.8% and specificity of 83.9% in an IgM ELISA. The sera of all four patients with hepatosplenic involvement tested positive for both IgG and IgM antibodies with both bacterial antigens, pointing to the extreme usefulness of our noninvasive antibody assay as compared to biopsy-based diagnostic procedures.
Differential diagnosis Among the typical clinical manifestations of CDS the enlargement of regional lymph nodes, after a cat scratch or bite in the localized area, is a common one. In patients with lymph node enlargement it is extremely challenging to prove or invalidate the CSD diagnosis, and for this reason differential diagnosis is critical, since similar lymph node enlargements can be commonly observed both in infectious diseases (including TB), lymphoproliferative diseases, and malignancies. We investigated possibility of developing a novel serological ELISA assay for rapid and effective differential diagnosis. Sera from patients suffering from lymphoma (n=41) or TB infection (n=5) were used. The pattern of distribution of IgM antibody response is shown in Figure 5. In the case of the GroEL antigen (Fig 5 A) there were 3/46 false positive (all patients with lymphoma; 93.5% specificity), whereas in the case of the 17-kDa antigen (Fig 5 B) there were 5/46 false positive (4 lymphoma and one TB patients; 89.1% specificity). The two antigens combined in the same test gave false positive results for 7/46 patients, with a
This article is protected by copyright. All rights reserved.
Accepted Article
specificity of 84.8%. The high specificity of 89% for 17-kDa antigen and 93.5% for GroEL antigen for patients with lymphoma and TB infections will help clinicians to distinguish between CSD and other serious diseases involving lymph nodes. The presence of false positive results in our differential diagnosis experiment can perhaps be explained by the highly conservation of GroEL. For this reason to deeply investigate cross reactivity and specificity of the antigens selected for our study it could be useful to analyse sera from other bacterial infections to further assess the diagnostic power of this assay. Besides the potential cross-reactivity of the identified targets, our assay benefits from the combination of the two antigens proteins, when used for differential diagnosis, resulting in a substantial improvement of sensitivity. Our hypothesis relies on the heterogeneous humoral response to different antigens as the cause of the increased specificity of the assay (Zhang et al. 2009), in agreement with current guidelines for the diagnosis of infections involving an antibody response, thus suggesting the importance of having multiple antigens to define the antibody profile related to early diagnosis and follow-up of the specific illness (Steller et al. 2005; Sartain et al. 2006). Although further work is now needed to establish the specificity of this assay towards other members of the Bartonella genus, we believe that the 2-antigens classifier that we have defined will greatly help the differential diagnosis of B. henselae infections.
MATERIALS and METHODS
Bacteria and DNA preparation B.henselae strain Houston1 (ATCC 49882) was grown for 7 days on 5% sheep blood agar plates at 37°C in a 5% CO2, humid atmosphere. Genomic DNA extraction and purification were performed as previously described (Sweger et al. 2000).
This article is protected by copyright. All rights reserved.
Accepted Article
2D-PAGE and Western blot (WB) ZOOM-IPGRunnerSystem (Invitrogen) was used for 2D-PAGE: 50µg of cell-lysate were separated first using IPG strips (pH 3÷10) and then by second-dimension electrophoresis on 12.5% PAGE. Proteins were visualized by silver staining (Invitrogen) (Fig 1A) or transferred onto nitrocellulose membrane, blocked using 2% non-fat Milk in PBS (MPBS) for 1 h at room temperature (RT). Transferred proteins were detected by a pool of positive IFA sera diluted 1:500 and visualized by goat anti-human IgM horseradish peroxidise-conjugated (DAKO) 1:200000 and chemiluminescence (Fig 1B). SDS-PAGE on recombinant proteins was performed, with detection carried out with anti-His (Amersham) and anti-SV5 mAbs as primary antibodies, 1:5000 in MPBS, followed by goat anti-mouse IgG alkaline-phosphatase-conjugated (AP) (Jackson), 1:5000 in MPBS. Antigenicity was evaluated using sera as primary antibodies, 1:200 in MPBS, followed by anti-human IgG or IgM-AP (Jackson) 1:2000 in MPBS. Immunocomplexes were revealed by chromogenic substrate NBT-BCIP (Sigma) (Fig 2).
LC-MS/MS analysis and Protein BLAST The protein of interest was digested and analysed by LC-MS (Gu et al. 2003). The protein gel band was cut from silver stained gel, de-stained, reduced with 1x10-3 mol l-1 DTT, and alkylated with 10x10-3 mol l-1 iodoacetamide before tryptic digestion. The digested products were extracted - by bath sonication using 5% acetic acid and 5% acetic acid /50% acetonitrile – concentrated, and analysed by a μLC-nanospray-MS/MS using a QSTAR XL mass spectrometer (AppliedBiosystems) coupled with LC Packings Ultimate micro-capillary LC system (Dionex). The MS/MS data were de novo sequenced and submitted for protein BLAST search to the MS-BLAST server at EMBL (http://dove.embl-heidelberg.de/Blast2/msblast.html) using the
This article is protected by copyright. All rights reserved.
Accepted Article
Pro-BLAST 1.0 software (AppliedBiosystems). Positive hits reported by the search were further investigated by manual verification of the positive alignment of raw MS/MS spectra and sequences (Fig 1C).
Gene cloning The 17-kDa gene sequence was obtained from GenBank (U23447) and PCR amplified from B.henselae genomic DNA (primers 17-kDa-sense GCTCGGAATTCATGAAAAAATATAGCTTAGTC and 17-kDa-anti GAAGCAAGCTTCTAAAGTCGGACATCAGATTTC). The gene was cloned in the expression vector pMAL-c2X (NEB) modified to express the tag SV5 (Hanke et al. 1992) at the C-terminus of the cloned insert, with a final expected product of ~60Kda due to its fusion to MBP (Fig 2). The GroEL gene sequence was deduced from peptide analysis and subsequently searched in the genome of B.henselae in the GenBank database. The gene was PCR-amplified from genomic DNA (primers GROELsense GCTCGCTCGAGCATGGCTGCTAAAGAAGTSAAG; GROELanti TGAAGCGAATTCTTAGAAGTCCATGCCGCCCAT) and cloned in the expression vector pTrcHisB (Invitrogen). Both vectors were transformed into E.coli DH5αF’ cells.
Expression and purification of recombinant proteins Bacteria transformed with both constructs were grown in 2xYT liquid broth, with 100 µg ml-1 ampicillin, at 37 °C to OD600 0.5. Expression of the recombinant protein was induced with 0.2x10-3 mol l-1 IPTG and O/N growth at 28°C. GroEL protein was purified from the soluble cytoplasmic fraction by IMAC affinity chromatography (Qiagen), and the 17-kDa protein was affinity-purified on amylose resin (NEB), following manufacturer’s instructions. Eluted fractions were dialyzed against PBS,
This article is protected by copyright. All rights reserved.
Accepted Article
their purity was evaluated by SDS-PAGE, and protein concentration determined by Bradford assay.
Enzyme-linked Immunosorbent Assay (ELISA) Preliminary experiments were conducted before performing our ELISA assay to establish serum dilution (1:50 – 1:100 and 1.200 dilutions) and amount of antigen (2.5 μg ml-1, 5 μg ml-1 and 10 μg ml-1 ) to be coated on the plate. Results showed that when both recombinant proteins were coated at 10 μg ml-1 and sera (individual and pool of patients and controls’ samples) were diluted at 1:100 we obtained a good signal-to-noise ratio (data not shown). Therefore, ELISA-plates were coated with purified proteins, 10 μg ml-1, O/N at 4°C. Wells were blocked with MPBS for 1h at RT. Human sera (1:100 in blocking buffer) were used as primary antibodies and incubated for 1h at 37°C, followed by goat anti-human IgG or goat anti-human IgM (DAKO), AP -conjugated (1:1000 in MPBS) for 1h at 37°C; the immunocomplexes were revealed with para-Nitrophenyl-phosphate.
Human sera Sera were obtained from IRCCS Burlo Garofolo, after obtaining institutional approval and informed consent from study subjects. All 64 CDS patients were positive for IgM and IgG anti-B.henselae antibodies detected by IFA (titre > 64), 60 serum samples had CSDdependent regional lymphadenitis (25 F 35 M, median age 12y, range 6-18 y) and 4 had systemic manifestations of B.henselae infection (hepatosplenic multiple granulomatous lesions). Healthy donors (n=87; 40 F, 47 M, median age 25 y, range 18-35 y) were used as negative controls, and for differential diagnosis, sera from 5 TB patients (5 M, median age 30 y), or 41 non-Hodgkin’s lymphoma (10 F, 31 M, median age 11 y, range 3-16 y), all with negative IFA titres (
Received Date : 14-Jan-2014 Revised Date : 08-May-2014 Accepted Date : 10-May-2014 Article type : Original Article
Development of an enzyme-linked immunosorbent assay (ELISA) for Bartonella henselae infection detection
Fortunato Ferrara 1, Roberto Di Niro 2, Sara D’Angelo1, Marina Busetti 3, Roberto Marzari 4, Tarcisio Not 3-5, Daniele Sblattero 6*. 1. New Mexico Consortium, Los Alamos, NM, USA. 2. Department of Immunology, University of Pittsburgh, PA, USA. 3. Institute for Maternal and Child Health - IRCCS “Burlo Garofolo” Trieste, Italy. 4. Department of Life Sciences, University of Trieste, Trieste, Italy. 5. University of Trieste, Trieste Italy. 6. Department of Health Sciences and IRCAD, University of Eastern Piedmont, Novara, Italy.
* Corresponding author: Prof. Daniele Sblattero Department of Health Sciences IRCAD University of Eastern Piedmont Via Solaroli, 17 28100 Novara Italy This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process which may lead to differences between this version and the Version of Record. Please cite this article as an 'Accepted Article', doi: 10.1111/lam.12286 This article is protected by copyright. All rights reserved.
Accepted Article
Phone: +39 0321 660596 Fax:
+39 0321 620421
E-mail: [email protected] Running headline: Bartonella henselae antigens.
SIGNIFICANCE AND IMPACT OF THE STUDY A reliable serological assay for the diagnosis of Cat Scratch Disease (CSD) - a pathological condition caused by Bartonella henselae infection – has not yet been developed. Such an assay would be extremely useful to discriminate between CSD and other pathologies with similar symptoms but different aetiologies, e.g. lymphoma or tuberculosis. We investigate the use of two Bartonella henselae proteins – GroEL and 17-kDa – to develop a serological based ELISA, showing promising results with the potential for further development as an effective tool for the differential diagnosing of Bartonella henselae infection.
ABSTRACT Several serological diagnostics rely on enzyme-linked immunosorbent assay (ELISA) to detect bacterial infections. However for some pathogens, including Bartonella henselae, diagnosis still depends on manually intensive, time-consuming assays including microimmunofluorescence, Western blotting or indirect-immunofluorescence. For such pathogens, there is obviously still a need to identify antigens to establish a reliable, fast, and highthroughput assay (Dupon et al. 1996). We evaluated two B. henselae proteins to develop a novel serological ELISA: a well-known antigen, the 17-kDa protein, and GroEL, identified during this study by a proteomic approach.
This article is protected by copyright. All rights reserved.
Accepted Article
When serum IgG were tested, the specificity and sensitivity were 76% and 65.7% for 17-kDa, respectively, and 82% and 42.9% for GroEL, respectively. IgM were found to be more sensitive and specific for both proteins:17-kDa, specificity 86.2% and sensitivity 75%; GroEL specificity 97.7% and sensitivity 45.3%. IgM antibodies were also measured in lymphoma patients and patients with Mycobacterium tuberculosis infection to assess the usefulness of our ELISA to distinguish them from B. henselae infected patients. The resulting specificities were 89.1% and 93.5% for 17-kDa protein and GroEL, respectively. Combining the results from the two tests, we obtained a sensitivity of 82.8% and a specificity of 83.9%. Our work described and validated a proteomic approach suitable to identify immunogenic proteins useful for developing a serological test of B.henselae infection. Key words: Bartonella henselae, proteomics, GroEL, 17-kDa, antigens, cat scratch disease
INTRODUCTION B. henselae, a gram-negative bacterium of the Proteobacteria alpha 2 class (Alsmark et al. 2004), is the etiologic agent of cat scratch disease (CSD) (Dehio 2004), with an estimated incidence of 24000 people affected annually in the United States, but with the true incidence hard to calculate due to the fact that it is not a reportable disease in several states in the US (Florin et al. 2008). The infection affects the paediatric population (80% children aged 2-14 years) and young adults predominantly (Jacomo et al. 2002). The cat flea, Ctenocephalides felis, is responsible for spreading B. henselae among cats, the main reservoir for the microorganism, that can enter the host through a cat scratch or bite, potentially causing acute and chronic infections, with manifestations depending on the status of the host immune system (Chomel et al. 2003). In immunocompetent patients, CSD is
This article is protected by copyright. All rights reserved.
Accepted Article
frequently characterized by regional lymph-node enlargement and by the presence of redbrown skin papules at the site of inoculation. Patients typically remain afebrile, and uncomplicated CSD-mediated lymphadenopathy spontaneously resolves within 2-6 months. In immunocompromised patients bacillary angiomatosis and peliosis are typical features, accompanied by multi-organ infection (Massei et al. 2005). In the differential diagnosis of CSD, all possible causes of lymphadenopathy, a common paediatric problem, must be considered. Lymphadenopathy is usually due to a transitory response to an upper respiratory tract infection. In sub-acute or chronic cases, the most common culprits are gram-positive bacteria and, less commonly, Mycobacterium tuberculosis. Sometimes, however, lymphadenopathy is a clinical sign of a more serious illness, such as lymphoma (Massei et al. 2000; Ghez et al. 2001). For this reason, a non-invasive, specific diagnostic assay for B.henselae infection would be of great help in clinical practice. Rapid identification of infectious agents is critical to be able to respond promptly, to prevent disease spread, and initiate patient treatment. For many infections traditional detection methods involve time-consuming bacterial growth in culture media, isolation, biochemical identification, and, sometimes, serology. To illustrate a proteomic approach to developing a rapid diagnostic assay for a specific infectious agent, Bartonella henselae, for which there is no straightforward reliable diagnostic assay, was chosen as a model. Current methods for CDS diagnosis consist in the isolation of the bacteria from biopsy samples, an invasive and lengthy procedure (La Scola and Raoult 1999; Klotz et al. 2011). Techniques, such PCR amplification, are also biopsy-based (Anderson et al. 1994; Chung et al. 2005), and require unique expertise. Serological diagnosis, usually based on an indirect immunofluorescence assay using irradiated whole-cell antigen preparation from co-cultivated Vero cells (Zbinden 1998), is sensitive, but costly, and cross-reactivity has been reported with other species (La Scola and Raoult 1996; McGill et al. 1998). Whole-cell antigen ELISAs
This article is protected by copyright. All rights reserved.
Accepted Article
give relatively poor sensitivity and specificity (Vermeulen et al. 2007; Tsuruoka et al. 2012), suggesting that the use of highly purified bacterial proteins may provide improved sensitivity and specificity results. Follow-up of antibody titres against B. henselae could also help in assessing definitive resolution of the infection (Lesprit et al. 2003). Finding new target antigens for the serological diagnosis of B. henselae infection is therefore of considerable interest. In this study, we evaluated two suitable candidates for new ELISA assay: a well-described 17-kDa protein (Anderson et al. 1995; Sweger et al. 2000; Loa et al. 2006; Hoey et al. 2009)with homology to proteins of the virB operon, associated with the secretion of polypeptides or nucleic acids (Padmalayam et al. 2000) - and an antigen isolated by a proteomic approach and identified as the GroEL chaperon protein.
RESULTS and DISCUSSION
Antigenic protein identification Several proteins involved in the immunologic reactions against B. henselae have been recently identified (McCool et al. 2008; Eberhardt et al. 2009). We used an immunoproteomic approach for the identification of a novel B. henselae-derived immuno-dominant antigenic protein. After preparing and subjecting whole bacterial extracts to 2D-PAGE, sera from eight patients with CSD diagnosis, without hepatosplenic involvement, were pooled and used to probe blotted proteins. A comparison between silver stained gel (Fig 1 A) and WB (Fig 1 B) showed a limited number of spots labelled by the pooled IgM serum antibodies. Immunoreactive spots did not correspond to major stained polypeptides, attesting to high specificity in serum antibody recognition. A heavily stained protein with approximately a MW of 60 kDa was identified (pointed by the arrow in Fig 1B) and selected for further
This article is protected by copyright. All rights reserved.
Accepted Article
analyses. Interestingly, the immunoblot identified additional polypeptides, stained in the 1520 kDa MW range, a protein size that likely represents the well-known 17-kDa antigen.
LC-MS/MS Analysis The identified 60 kDa protein was judged as a putative immunodominant antigen, and the protein spot excised from the silver stained gel subjected to in-gel digestion. Forty-seven different peptide sequences obtained by MS/MS spectra analyzed by BLAST, showed high homology to peptides from B.henselae GroEL (accession number:O33963), covering 66% of the entire protein sequence, allowing us to positively identify the protein as GroEL (Fig 1 C). GroEL belongs to the highly conserved family of Heat Shock Proteins (HSPs). These proteins have crucial functions in cell homeostasis, protein folding (Smith et al. 1998), and play key roles in cytoprotection. HSP synthesis is extremely important during an infection, because of the challenging environmental changes in pathogen living conditions, making them an important target in the defence against infectious agents (Srivastava 2002), with immune response to HSPs observed in disease caused by bacteria, protozoa, fungi and nematode infection (Zugel and Kaufmann 1999). In particular GroEL is a chaperonin, forming an oligomeric complex with GroES (HSP10), which plays a key role in the folding and assembly of proteins (Mayhew et al. 1996; Haake et al. 1997; Marston et al. 1999; Minnick et al. 2003; Tang et al. 2006). GroEL has also been described as an immunogen in other microorganisms, including Bartonella bacilliformis (Knobloch and Schreiber 1990), and it seems that the immune response is usually directed against non-conserved regions (Lamb et al. 1989; Young and Garbe 1991). For this reason portion or epitopes of species specific GroEL, lacking homology to the human protein, have been considered as vaccine candidates (Pechine et al. 2013).
This article is protected by copyright. All rights reserved.
Accepted Article
Cloning, expression and purification of GroEL and 17-KDa proteins In order to assess the value of GroEL as a potential diagnostic, the gene was cloned and expressed as a recombinant protein in E. coli, and purified by affinity chromatography on NiNTA resin. Eluted fractions were pooled and checked for purity by SDS/PAGE followed by Coomassie-blue staining (Fig 2, lane A), showing a high degree of purity. WB analysis was performed using the His-tag at the N-terminus of the GroEL protein for detection. The result (Fig 2, lane C) confirms the recognition of a protein at the expected molecular weight (60 kDa) for the full recombinant protein. The 17-kDa antigen (Anderson et al. 1995) has also been previously proposed for ELISAbased diagnostic tests. Notwithstanding cross-reactivity that has been reported between the B. henselae 17-kDa protein and the serum of B. quintana infected patients, the positive ability of a 17-kDa-based ELISA assay to detect B. henselae-specific antibody in human samples (Loa et al. 2006) has been demonstrated. We also cloned the B. henselae 17-kDa gene to provide a positive control, a reference for the GroEL immunoreactivity test, and also to use more than one target to improve the detection of anti-B. henselae antibody in patients (Loa et al. 2006). The 17-kDa protein is scarcely soluble (Loa et al. 2006), and this was confirmed in preliminary experiments (data not shown), where the 17-kDa gene cloned into the pTrcHisB vector yielded a limited amount of recombinant protein, due to its insolubility and its deposit in inclusion bodies (Loa et al. 2006). Consequently, we used an alternative approach cloning the gene in the expression vector pMAL-c2X, to allow the expression as a fusion with the E. coli maltose binding protein (MBP). The 17-kDa gene was obtained by PCR amplification and cloned into the pMAL-c2X plasmid. Correct cloning was confirmed by DNA sequencing. The recombinant fusion protein was expressed, extracted and purified by affinity chromatography on amylose resin. The eluted fractions were pooled and checked by SDS/PAGE followed by staining with
This article is protected by copyright. All rights reserved.
Accepted Article
Coomassie-blue (Fig 2, lane B). The recombinant protein shows the expected molecular weight (~60 kDa: 17-kDa protein fused to the 42.5 kDa MBP and SV5 tag). with a very low level of degradation products. This was further confirmed by WB using a mAb recognizing the SV5 tag mapping to the C-terminus of the recombinant protein (Fig 2 lane D).
Antigenic reactivity A preliminary validation of the antigenic properties of the recombinant proteins was performed by WB analysis. A pool of eight sera from clinical cases (no patients with hepatosplenic involvement) of CSD with a positive IFA test was compared with a pool of sera from healthy donors. The results demonstrate that both GroEL and 17-kDa recombinant proteins are specifically recognized by IgM serum antibodies (Fig 2, lane E: GroEL; lane F: 17-kDa protein), whereas sera from controls (Fig 2, lanes G: GroEL; lane H: 17-kDa protein) show no reactivity. Sera were also tested against a control recombinant protein formed by MBP and the SV5 tag (MBP-SV5) to confirm the specificity of the binding to 17-kDa. No reactivity was detected, thus, demonstrating that the binding was specific for the 17-kDa target protein (data not shown). Several bacterial species – in particular within the Bartonella genus – have proteins with high homology to the two targets selected in this study (Fig 3). In particular, GroEL is one of the most conserved proteins in nature, with high sequence homology between mammals and microbes, and is frequently found to be immunogenic in a variety of bacterial and parasitic infection (Young et al. 1988; Shinnick 1991). However, the high level of shared homology between different GroEL proteins does not appear to prevent specific humoral response against species-specific HSP, as shown when similar proteins have been used as vaccination targets and species-GroEL specific antibodies were produced with no autoimmune reaction against the host GroEL (Noll and AutenriethIb 1996). For this reason we decided to use the identified GroEL protein in our ELISA assay.
This article is protected by copyright. All rights reserved.
Accepted Article
ELISA results ELISA assay has a number of advantages for serodiagnosis, including sensitivity and ease of automation. Preliminary experiments were conducted before performing our ELISA assay to establish serum dilution and the amount of antigen to be coated on the plate. When both recombinant proteins were coated at 10 μg/ml and sera (individual and pools of patients and control samples) were diluted at 1:100 we obtained a good signal-to-noise ratio (data not shown). A total of 64 patients and 87 controls sera were used. ELISA results were analysed by ROC curves. The calculated IgG response against 17-kDa protein resulted in a specificity of 76.0% and a sensitivity of 65.7% with a corresponding AUC of 0.769, confident interval (CI) 0.664÷0.853 (Fig 4 A-B). When the reactivity against GroEL protein was calculated, the specificity was 82.0% and sensitivity 42.9%, with an AUC of 0.631, CI 0.520÷0.733 (Fig 4C-D). We then focused on the IgM response, which gave a reactivity against the 17-kDa-protein with a specificity of 86.2%, sensitivity of 75.0%, and an AUC of 0.848, CI 0.781÷0.901 (Fig 4E-F), while the anti-GroEL response gave a specificity of 97.7%, sensitivity of 45.3%, and an AUC of 0.781, CI 0.707÷0.844 (Fig 4G-H). For IgM in both cases, the AUC parameters confirmed the ability of the assay to discriminate between diseased and healthy cases. The evaluation of the two antibody class responses to the two antigenic proteins showed IgM as the best choice for an ELISA-based diagnostic assay, with sensitivity and specificity values for 17-kDa almost identical to those reported in other studies (Bergmans et al. 1997). IgM are usually short lived and replaced by IgG, but the nature of IgM - as first immunoglobulin responders in infectious disease - can explain their better performance in our ELISA assay in this study where the subjects were most likely diagnosed during the acute phase of B. henselae infection. Despite these promising values, they were not deemed sufficient for an effective diagnostic test. When tested alone, the GroEL specificity was excellent (97.7%) but the sensitivity was
This article is protected by copyright. All rights reserved.
Accepted Article
disappointing (45%). For CDS diagnosis the use of ELISA based on a single antigen may be optimistic, especially since recent profiling of the feline humoral immune response (Vigil et al. 2010), demonstrated that a classifier with multiple antigens showed improved performances. Therefore combining different antigenic proteins with the well-known 17-kDa antigen (Loa et al. 2006; Hoey et al. 2009) could improve B.henselae infection diagnosis. For this reasons we combined the results of the two tests, obtaining a sensitivity of 82.8% and specificity of 83.9% in an IgM ELISA. The sera of all four patients with hepatosplenic involvement tested positive for both IgG and IgM antibodies with both bacterial antigens, pointing to the extreme usefulness of our noninvasive antibody assay as compared to biopsy-based diagnostic procedures.
Differential diagnosis Among the typical clinical manifestations of CDS the enlargement of regional lymph nodes, after a cat scratch or bite in the localized area, is a common one. In patients with lymph node enlargement it is extremely challenging to prove or invalidate the CSD diagnosis, and for this reason differential diagnosis is critical, since similar lymph node enlargements can be commonly observed both in infectious diseases (including TB), lymphoproliferative diseases, and malignancies. We investigated possibility of developing a novel serological ELISA assay for rapid and effective differential diagnosis. Sera from patients suffering from lymphoma (n=41) or TB infection (n=5) were used. The pattern of distribution of IgM antibody response is shown in Figure 5. In the case of the GroEL antigen (Fig 5 A) there were 3/46 false positive (all patients with lymphoma; 93.5% specificity), whereas in the case of the 17-kDa antigen (Fig 5 B) there were 5/46 false positive (4 lymphoma and one TB patients; 89.1% specificity). The two antigens combined in the same test gave false positive results for 7/46 patients, with a
This article is protected by copyright. All rights reserved.
Accepted Article
specificity of 84.8%. The high specificity of 89% for 17-kDa antigen and 93.5% for GroEL antigen for patients with lymphoma and TB infections will help clinicians to distinguish between CSD and other serious diseases involving lymph nodes. The presence of false positive results in our differential diagnosis experiment can perhaps be explained by the highly conservation of GroEL. For this reason to deeply investigate cross reactivity and specificity of the antigens selected for our study it could be useful to analyse sera from other bacterial infections to further assess the diagnostic power of this assay. Besides the potential cross-reactivity of the identified targets, our assay benefits from the combination of the two antigens proteins, when used for differential diagnosis, resulting in a substantial improvement of sensitivity. Our hypothesis relies on the heterogeneous humoral response to different antigens as the cause of the increased specificity of the assay (Zhang et al. 2009), in agreement with current guidelines for the diagnosis of infections involving an antibody response, thus suggesting the importance of having multiple antigens to define the antibody profile related to early diagnosis and follow-up of the specific illness (Steller et al. 2005; Sartain et al. 2006). Although further work is now needed to establish the specificity of this assay towards other members of the Bartonella genus, we believe that the 2-antigens classifier that we have defined will greatly help the differential diagnosis of B. henselae infections.
MATERIALS and METHODS
Bacteria and DNA preparation B.henselae strain Houston1 (ATCC 49882) was grown for 7 days on 5% sheep blood agar plates at 37°C in a 5% CO2, humid atmosphere. Genomic DNA extraction and purification were performed as previously described (Sweger et al. 2000).
This article is protected by copyright. All rights reserved.
Accepted Article
2D-PAGE and Western blot (WB) ZOOM-IPGRunnerSystem (Invitrogen) was used for 2D-PAGE: 50µg of cell-lysate were separated first using IPG strips (pH 3÷10) and then by second-dimension electrophoresis on 12.5% PAGE. Proteins were visualized by silver staining (Invitrogen) (Fig 1A) or transferred onto nitrocellulose membrane, blocked using 2% non-fat Milk in PBS (MPBS) for 1 h at room temperature (RT). Transferred proteins were detected by a pool of positive IFA sera diluted 1:500 and visualized by goat anti-human IgM horseradish peroxidise-conjugated (DAKO) 1:200000 and chemiluminescence (Fig 1B). SDS-PAGE on recombinant proteins was performed, with detection carried out with anti-His (Amersham) and anti-SV5 mAbs as primary antibodies, 1:5000 in MPBS, followed by goat anti-mouse IgG alkaline-phosphatase-conjugated (AP) (Jackson), 1:5000 in MPBS. Antigenicity was evaluated using sera as primary antibodies, 1:200 in MPBS, followed by anti-human IgG or IgM-AP (Jackson) 1:2000 in MPBS. Immunocomplexes were revealed by chromogenic substrate NBT-BCIP (Sigma) (Fig 2).
LC-MS/MS analysis and Protein BLAST The protein of interest was digested and analysed by LC-MS (Gu et al. 2003). The protein gel band was cut from silver stained gel, de-stained, reduced with 1x10-3 mol l-1 DTT, and alkylated with 10x10-3 mol l-1 iodoacetamide before tryptic digestion. The digested products were extracted - by bath sonication using 5% acetic acid and 5% acetic acid /50% acetonitrile – concentrated, and analysed by a μLC-nanospray-MS/MS using a QSTAR XL mass spectrometer (AppliedBiosystems) coupled with LC Packings Ultimate micro-capillary LC system (Dionex). The MS/MS data were de novo sequenced and submitted for protein BLAST search to the MS-BLAST server at EMBL (http://dove.embl-heidelberg.de/Blast2/msblast.html) using the
This article is protected by copyright. All rights reserved.
Accepted Article
Pro-BLAST 1.0 software (AppliedBiosystems). Positive hits reported by the search were further investigated by manual verification of the positive alignment of raw MS/MS spectra and sequences (Fig 1C).
Gene cloning The 17-kDa gene sequence was obtained from GenBank (U23447) and PCR amplified from B.henselae genomic DNA (primers 17-kDa-sense GCTCGGAATTCATGAAAAAATATAGCTTAGTC and 17-kDa-anti GAAGCAAGCTTCTAAAGTCGGACATCAGATTTC). The gene was cloned in the expression vector pMAL-c2X (NEB) modified to express the tag SV5 (Hanke et al. 1992) at the C-terminus of the cloned insert, with a final expected product of ~60Kda due to its fusion to MBP (Fig 2). The GroEL gene sequence was deduced from peptide analysis and subsequently searched in the genome of B.henselae in the GenBank database. The gene was PCR-amplified from genomic DNA (primers GROELsense GCTCGCTCGAGCATGGCTGCTAAAGAAGTSAAG; GROELanti TGAAGCGAATTCTTAGAAGTCCATGCCGCCCAT) and cloned in the expression vector pTrcHisB (Invitrogen). Both vectors were transformed into E.coli DH5αF’ cells.
Expression and purification of recombinant proteins Bacteria transformed with both constructs were grown in 2xYT liquid broth, with 100 µg ml-1 ampicillin, at 37 °C to OD600 0.5. Expression of the recombinant protein was induced with 0.2x10-3 mol l-1 IPTG and O/N growth at 28°C. GroEL protein was purified from the soluble cytoplasmic fraction by IMAC affinity chromatography (Qiagen), and the 17-kDa protein was affinity-purified on amylose resin (NEB), following manufacturer’s instructions. Eluted fractions were dialyzed against PBS,
This article is protected by copyright. All rights reserved.
Accepted Article
their purity was evaluated by SDS-PAGE, and protein concentration determined by Bradford assay.
Enzyme-linked Immunosorbent Assay (ELISA) Preliminary experiments were conducted before performing our ELISA assay to establish serum dilution (1:50 – 1:100 and 1.200 dilutions) and amount of antigen (2.5 μg ml-1, 5 μg ml-1 and 10 μg ml-1 ) to be coated on the plate. Results showed that when both recombinant proteins were coated at 10 μg ml-1 and sera (individual and pool of patients and controls’ samples) were diluted at 1:100 we obtained a good signal-to-noise ratio (data not shown). Therefore, ELISA-plates were coated with purified proteins, 10 μg ml-1, O/N at 4°C. Wells were blocked with MPBS for 1h at RT. Human sera (1:100 in blocking buffer) were used as primary antibodies and incubated for 1h at 37°C, followed by goat anti-human IgG or goat anti-human IgM (DAKO), AP -conjugated (1:1000 in MPBS) for 1h at 37°C; the immunocomplexes were revealed with para-Nitrophenyl-phosphate.
Human sera Sera were obtained from IRCCS Burlo Garofolo, after obtaining institutional approval and informed consent from study subjects. All 64 CDS patients were positive for IgM and IgG anti-B.henselae antibodies detected by IFA (titre > 64), 60 serum samples had CSDdependent regional lymphadenitis (25 F 35 M, median age 12y, range 6-18 y) and 4 had systemic manifestations of B.henselae infection (hepatosplenic multiple granulomatous lesions). Healthy donors (n=87; 40 F, 47 M, median age 25 y, range 18-35 y) were used as negative controls, and for differential diagnosis, sera from 5 TB patients (5 M, median age 30 y), or 41 non-Hodgkin’s lymphoma (10 F, 31 M, median age 11 y, range 3-16 y), all with negative IFA titres (
