Proof of concept for multiplex detection of antibodies against Chlamydia species in chicken serum using a bead-based suspension array with peptides as antigens

Abstract

The available differentiating tests for Chlamydia are based on detection of genetic material and only give information about the actual infection status, but reveal nothing of past infections. As the use of serological methods increases the window of detection, the goal of this study was to investigate if it is possible to develop a differentiating serological test for antibodies against Chlamydia species in chicken sera. Focus was on C. psittaci, C. gallinacea, and two closely related species, i.e. C. abortus and C. avium. To enable differentiating serology, a bead-based Luminex suspension array was constructed, using peptides as antigens, derived from known immunoreactive Chlamydia proteins. For the majority of these peptides, species-specific seroreactivity in mammalian sera has been reported in literature. The suspension array correctly identified antibodies against various Chlamydia species in sera from experimentally infected mice, and was also able to differentiate between antibodies against C. psittaci and C. gallinacea in sera from experimentally infected chickens. In field sera, signals were difficult to interpret as insufficient sera from experimentally infected chickens were available for evaluating the seroreactivity of all peptides. Nevertheless, results of the suspension array with field sera are supported by published data on the occurrence of C. gallinacea in Dutch layers, thereby demonstrating the proof of concept of multiplex serology for Chlamydial species in poultry.

Introduction

Chlamydia is an expanding genus of Gram-negative obligate intracellular bacteria, currently containing 14 different species [1], of which some were discovered only in the last decade [2], and with proposed new species pending [3]. Each species has a preferred host or host range, and many cannot only cause disease in their preferred hosts, but also have zoonotic potential [4]. The most infamous zoonotic species is Chlamydia psittaci, which has been detected in hundreds of bird species [5] including poultry [6]. In birds, C. psittaci infections can result in mild to severe disease, depending on the avian species and the Chlamydia strain, whereas an infection in humans can lead to severe pneumonia, i.e. psittacosis [7].

In the past, C. psittaci was considered the main Chlamydia species in poultry, including chickens [6,7,8]. More recent observations indicate that in chickens Chlamydia gallinacea is the most prevalent Chlamydia species [9,10,11]. Apart from a reduction in weight gain, infections with C. gallinacea are asymptomatic [9]. Conclusive evidence for zoonotic potential has not been presented yet, but cases of atypical pneumonia in slaughterhouse personnel coinciding with the occurrence of C. gallinacea have been reported [12].

Since C. gallinacea and C. psittaci can infect chickens and are likely to differ in epidemiology and zoonotic potential, it is important to have diagnostic tools available that allow detection of, and distinction between, these two species, both from a public health perspective as from an economic point of view. Current discriminatory tests rely completely on molecular methods [13] and only give information about the present infection status of animals. Discriminatory serology, however, will have added value as serology is a very easy way of monitoring (past) occurrence of any pathogen in animal production chains. At the time of writing, for Chlamydia serology in birds only a non-discriminatory ELISA is available [13, 14].

The goal of this study is to investigate if it is possible to develop a serological test that, in chicken serum, can differentiate between antibodies against different Chlamydia species, particularly C. psittaci and C. gallinacea, using peptides as antigens. Sequences of Chlamydia-derived peptides are available from the literature: a considerable number of sequences have been identified through comprehensive bio-informatics and subsequent screening with sera from infected mice [15, 16]. A subset of these peptides has successfully been used in arrays to differentiate between antibodies against various Chlamydia species in multiple mammalian hosts [17]. In the current study, we investigated whether peptides that have been used as antigens in mammalian serology also allow detection and differentiation of antibodies against Chlamydia species in chicken sera. To this end, a multiplex assay for antibody detection in chicken sera was developed, using Chlamydia-derived peptides as antigens in a Luminex suspension array, analogous to earlier work on detection of antibodies against (other) pathogens in sera from poultry and swine [18,19,20]. Here, protocols for peptide-based multiplex serology in chicken sera are given, including the use of serum-specific cutoffs for normalization, based on an internal control. Results obtained with sera from experimentally infected mice and chickens and with chicken sera from various types of farms, are discussed in detail. The proof of concept multiplex suspension array for antibodies against Chlamydial species in poultry correctly identifies Chlamydia antibodies in sera from experimentally infected animals, and observations in field samples are supported by published data on the occurrence of C. gallinacea in Dutch layers.

Materials and methods

Sera

Pooled sera from mice infected with different Chlamydia species [15, 16] were donated by Bernhard Kaltenboeck (Auburn University, Alabama, USA) and shipped by Christiane Schnee (Friedrich-Loeffler-Institut, Jena, Germany).

Sera from chickens experimentally infected with C. gallinacea or C. psittaci were available from various in-house experiments. From a published experimental infection with C. gallinacea NL_G47 [21, 22], sera were used from day 0, 14, 28, and 35 after oral inoculation of 5-week-old PCR-negative chickens (n = 8). From an unpublished experimental infection with C. psittaci 6BC, sera were available from day 0, 14, and 21 after exposure, and PCR data was available from day 4 (throat and cloaca swabs) and day 21 (airsacs, lungs and spleens) after exposure. This experiment was performed in essence as described elsewhere [23]. In short, 6-week-old SPF laying hens were placed for 1 h in an insulator in which 10 mL of a suspension with live C. psittaci 6BC (10⁶ TCID₅₀ per mL) was sprayed with a nebulizer. After infection, chickens were held in an open pen on sawdust bedding.

From a collection of ca. 2400 field sera from 2015, originating from the Dutch program for avian influenza monitoring performed by Royal GD, 120 sera from six farms were selected for testing. The collection contained 20 sera per farm, of which two farms of each of three types of farms, i.e. with conventional layers, free range layers, or broilers, were taken. The farms were selected simply by order of appearance in the collection, so the first appearing two serum sets from each farm type were picked.

Antigens

For ELISA, two commercially available antigens were used, i.e. C. abortus and Chlamydia trachomatis isolated from elementary and reticular bodies, inactivated by heat treatment and sonication (Virion/Serion, Würzburg, Germany).

For the suspension array, peptides were used as antigens. Peptide sequences derived from Chlamydia species that are relevant for poultry (C. psittaci and C. gallinacea) and from two genetically closely related species (resp. Chlamydia abortus and Chlamydia avium) were taken from literature. A selection was made based on signal strength and specificity as described [15,16,17]. In addition, for two published C. gallinacea-derived peptides, corresponding peptide sequences from two Dutch field isolates, NL_G47 and NL_F725 [24], were selected. All peptide sequences are listed in Table 1. Peptides ranging from 16 to 40 amino acids were synthesized on 50 µmol scale and pHPLC purified (Pepscan, Lelystad, The Netherlands). The peptides contained an N-terminal biotine separated from the peptide moiety by a spacer, consisting of five units of 8-amino-3,6-dioxaoctanoic acid, as used earlier [19, 20]. A compound that served as negative control (NC) was synthesized in exactly the same way, but without a peptide moiety (Table 1), and was used to determine assay background in each test for each serum.

Table 1 Peptides used as antigens

ELISA

An in-house ELISA with a mix of C. abortus and C. trachomatis antigens (Virion/Serion) was used to detect antibodies against Chlamydia species. In previous work, this ELISA was used to detect an increase in antibodies in C. gallinacea infected chickens, but a cutoff has not been established [21]. Where relevant, results are given and expressed as optical density (OD) values.

Bead sets with immobilized peptides

Biotinylated peptides and the compound that serves as negative control (NC) were immobilized as described before [19], using 5 × 10⁵ avidin coated paramagnetic beads (MagPlex-Avidin microspheres; Luminex, Den Bosch, The Netherlands). The biotinylated compounds were used at a final concentration of 4000 nM. After binding of these compounds to avidin-coated beads, the beads were washed and blocked with biotin [19].

In total, 10 spectrally distinct bead sets were used to assemble three different bead mixes. Each bead mix contained a bead set with the NC and up to 9 bead sets with peptides, as follows. Bead mix 1 contained 8 bead sets, loaded with resp. pCga001-003, pCga008-011, and NC; bead mix two contained 9 bead sets, loaded with resp. pCps001, 002, 009, 010, 013–016, and NC; bead mix three contained 10 bead sets, loaded with resp. pCab001-005, pCav001-004, and NC. The three bead mixes together constitute the Luminex suspension array when employed in parallel. By doing so, it was possible to run a cost-efficient multiplex assay for 24 peptides and internal controls with only 10 different beads.

Performing the suspension array

Suspension arrays were performed using elements of existing protocols [18,19,20, 25] as follows. Sera were diluted 1:200 in sample buffer, i.e. PBS-T with 10% PRI-blocker, a dedicated blocking reagent for bead-based immunoassays (Prime Diagnostics, Wageningen, The Netherlands). To prevent background, diluted sera were also pre-treated with neutravidin (Thermo Fisher Scientific). In parallel, of the three different bead mixes, prepared in PBS-T with 10% PRI-blocker, 50 µL was mixed with 50 µL of pre-treated serum in a 96-well plate to a final serum dilution of 1:400, with 750 beads of each bead set present.

The three bead mixes were investigated in parallel as follows. The bead/serum mixes were incubated in a 96-well plate for 30 min in the dark at room temperature on a plate shaker, after which the plate was washed three times with 100 µL PBS-T, using a magnet (LifeSep 96F, Thermo Fisher Scientific) to precipitate beads. Next, 100 µL of phycoerythrin conjugated secondary antibodies in PBS-T was added (1:1000). For detection of bound serum antibodies, Goat Anti-Chicken IgY(H+L) (Southern Biotech) was used for chicken serum, whereas AffiniPure F(ab′)₂ Fragment Goat Anti-Mouse IgG (H+L) (Jackson Immuno Research) was used for mouse serum. Beads were incubated with the secondary antibodies for 60 min in the dark at room temperature on a plate shaker. After washing, the fluorescence collected on the beads was measured with a Luminex LX200 system (Luminex), counting at least 75 beads per spectral region.

Serum-specific cutoffs and normalization

To evaluate signals acquired on peptide beads, an approach was developed to establish serum-specific cutoffs, using the corresponding assay background observed for each individual serum on the negative control (NC) beads, that are present in each of the three bead mixes. To establish serum-specific cutoffs, first the specific assay background was calculated for each serum by averaging the median fluorescent signal (MFI) acquired on NC beads (avNC), present in the three bead mixes that were tested in parallel for each serum. Next, the specific cutoff was established for each serum by taking three times the corresponding assay background, i.e. three times the average MFI observed on NC beads (3*avNC), as established empirically (not shown) by studying the effect of different multiplicities of the internal negative control NC on agreement with the known status of the experimental sera. Resulting values were taken as serum-specific cutoffs and were used to normalize data by subtraction. For normalized data this resulted in a cutoff of 0 for each serum, which allowed comparison of signals obtained with multiple sera.

Results

Sera from experimentally infected mice confirm species-specificity of peptide beads

Results of the suspension array obtained with pooled sera from mice infected with C. psittaci or C. gallinacea, or from naive mice, are presented in Figure 1. With pooled serum from mice infected with C. gallinacea, only signals on beads with peptides derived from C. gallinacea exceeded the cutoff (Figure 1, top). Not all C. gallinacea peptides were seroreactive (see below). Signals on beads with C. gallinacea peptides remained below the cutoff when tested with serum from mice infected with C. psittaci (Figure 1, middle) or with serum from naive mice (Figure 1, bottom). Likewise, peptides derived from C. psittaci were specifically recognized by the serum pool from C. psittaci infected mice (Figure 1, middle). On all beads with peptides, signals remained below the cutoff when sera from naive mice were tested (Figure 1, bottom). Results for all available mouse serum pools (n = 18; Additional file 1, with raw MFI data) showed that most peptides derived from C. gallinacea, C. psittaci, C. abortus, and C. avium were recognized in a species-specific manner in the suspension array. Before normalization, signals above the serum-specific cutoffs ranged from 377 to 12 532 MFI, with most specific signals above 1000 MFI, whereas assay backgrounds (on NC beads) remained below 200 MFI, with standard deviations ranging from 0 to 23 MFI. These results agreed with those of the in-house ELISA in which low signals (OD < 0.1) were obtained with sera from naive mice whereas high signals (OD > 0.8) were obtained with sera from infected mice, including sera from mice infected with the four Chlamydia species focused on in this study.

Figure 1

Results of the suspension array with mice sera. Results are shown for pooled sera from mice infected with C. gallinacea or C. psittaci, and naive mice. The three pools were tested with a fixed panel of beads carrying peptides or a negative control (NC) compound. Only signals on peptides are shown, expressed as normalized MFI (Y-axis) by subtraction of the serum-specific cutoffs that are based on signals on beads carrying NC. This resulted in a cutoff of 0 MFI for normalized data, represented by horizontal dashed red lines. Peptides are presented on the X-axis in the same order as listed in Table 1 and are labelled only as a group with the species names. See Additional file 1 for results (raw data) obtained with sera from mice infected with other Chlamydia species.

In the suspension array, the signals on IncA peptides derived from the strain used for infecting mice, was always higher than on the orthologous IncA peptides derived from chicken isolates (Additional file 1). Of the IncA and OmpA C. gallinacea peptides derived from Dutch chicken isolates, the IncA-derived pCga002 orthologs (pCga010 and 011) were recognized by sera from C. gallinacea infected mice, in contrast to the two OmpA-derived pCga001 orthologs (pCga008 and 009) (see Additional file 1). Sequence differences between pCga002 (IncA) and its orthologs, and between pCga001 (OmpA) and its orthologs, can be seen in Table 1.

Cross-reactions with low signals were observed for one C. avium serum on two C. psittaci peptides (< 400 MFI; pCps002 and 014). Both are different length versions derived from the same region, and do not share homology with proteins from C. avium (BLAST). Cross-reactions with a low signal were also observed with two C. pneumoniae sera on an IncA-derived peptide (< 600 MFI; pCab001) that has only limited homology to proteins from C. pneumoniae (BLAST; lowest E-value 0.52 for 45% of the query).

Taken together, results with mice sera confirmed that Chlamydia-derived peptides can be used as antigens in a bead-based suspension array for detection of serum antibodies against multiple Chlamydia species and that biotinylated peptides retain their seroreactivity when bound to avidin-coated beads.

Sera from chickens exposed to C. psittaci react with peptide beads in a species-specific manner

To investigate if the selected peptides react with sera from C. psittaci infected chickens, sera from three chickens exposed to C. psittaci were investigated. For these chickens, PCR data was available: early after the start of the experiment C. psittaci was detected in 2 out of 3 chickens (3031 and 3435); at the end of the experiment C. psittaci was detected in only one chicken (2829) (Additional file 2).

In the suspension array, the serum-specific assay background remained below 300 MFI for all sera, with standard deviations ranging from 3 to 45 MFI (Additional file 2, with raw MFI data). To visualize acquired signals on peptide beads of multiple sera in one graph, results were normalized by subtracting the calculated serum-specific cutoffs. By doing so, normalized signals were found negative on day 0 for all chickens, positive on day 14 and 21 for chicken 2829 and 3031 on one C. psittaci peptide (pCps010), or positive on day 14 and 21 for chicken 3435 on five C. psittaci peptides (incl. pCps010) (Figure 2). Three of the C. psittaci peptides were never recognized by sera from chickens exposed to C. psittaci. Of these, only pCps001 differed from the sequence of the strain used for infection (Table 2). All chicken sera from this experimental C. psittaci infection did not react with peptides derived from other Chlamydia species, apart from a cross-reaction for chicken 3031 with peptide pCab004.

Figure 2

Results of the suspension array with sera from chickens exposed to C. psittaci. Results are shown for three chickens exposed to C. psittaci 6BC. Sera from day 0, 14, and 21 were tested (resp. black, red, and green symbols); for each chicken the results of three timepoints are presented in one graph. Signals on peptides are shown and are expressed as normalized MFI. For details see the legend to Figure 1. Raw MFI values are given in Additional file 2. dpi: days post-infection.

Table 2 Seroreactivity and sequence identity of C. psittaci-derived peptides

Results with the in-house ELISA showed increased signals on day 21 (day 14 sera were not tested) compared to day 0 for all exposed chickens, which was most pronounced for chicken 3435, i.e. the chicken that responded to five peptides (Additional file 3).

Summarized, all three chickens infected with C. psittaci had an increased response in the in-house ELISA at day 21 and recognized at least 1 C. psittaci peptide in the suspension array. In total five out of the eight C. psittaci-derived peptides (that are all recognized by mice sera) were recognized by chicken sera (Table 2). Signals on four out of these five peptides were obtained only with serum from chicken 3435, i.e. the animal with the lowest Ct value in a throat swab taken 4 days after exposure. These results showed that the suspension array is able to detect a specific response against C. psittaci.

Sera from chickens infected with C. gallinacea react with peptide beads in a species-specific manner

From eight chickens infected with C. gallinacea NL_G47, sera from day 0, 14, 28, and 35 were used for this study. In the suspension array, signals on all NC beads remained below 300 MFI. Normalized signals remained below 0 on all beads on day 0, and became positive after day 0 on a subset of beads with C. gallinacea-derived peptides (Figure 3).

Figure 3

Results of the suspension array with sera from chickens infected with C. gallinacea. Results are shown for eight chickens infected with C. gallinacea NL_G47. Sera from day 0, 14, 28, and 35 were tested. For each peptide normalized signals are given for eight sera using a box-whisker plot. For details see the legend to Figure 1. Raw data are given in Additional file 4. dpi: days post-infection.

All IncA-derived peptides with sequence homology to the published peptide pCga002 and its shorter variant pCga003 were recognized by the chicken sera, including the orthologs pCga010 and 011 derived from the two Dutch field strains (Table 3). For each chicken, the highest signal (see Additional file 4 with raw MFI data) on IncA peptides was always on pCga010, derived from the NL_G47 isolate that was also used for experimental infection of these chickens. The published OmpA-derived pCga001 peptide was never recognized, but its ortholog pCga008 from the Dutch field strain NL_G47 (used for infection) was recognized by a subset of the tested sera. The ortholog from another Dutch field strain, NL_F725, was never recognized. The differences in seroreactivity are listed in Table 3, as are the peptide sequences and the corresponding sequences of the infecting strains. Weak cross-reactions were observed with peptide pCps015 with some sera (see Additional file 4 for details).

Table 3 Seroreactivity and sequence identity of C. gallinacea-derived peptides

With the in-house ELISA it had been shown that sera from day 14, 28, 35 resulted in high signals, whereas sera from day 0 had low signals, indicating that sera taken post-infection contained Chlamydia antibodies [22]. In ELISA, signals were considerably higher for sera that were seropositive for C. gallinacea in the suspension array than for seronegative sera (Additional file 5).

Summarized, of the seven C. gallinacea peptides, two were recognized by all sera from chickens infected with C. gallinacea NL_G47, three were recognized but not consistently, and two OmpA-derived peptides from strain NL_F725 and 08DC63 were never recognized. Seropositivity as determined by the suspension array correlated with signal strength of signals obtained with the in-house ELISA. These results show that a subset of the C. gallinacea-derived peptides can be used to specifically detect antibodies against C. gallinacea in infected chickens, and also suggest that this subset can be used to differentiate between antibodies against C. gallinacea and C. psittaci.

Field sera

A set of 120 field sera from layers and broilers with unknown infection status, were tested with the suspension array. The sera were from three types of chicken farms, with free range laying hens, conventional laying hens, or broilers, two farms each.

As for the experimental sera, for field sera the NC beads were tested simultaneously with the other peptide beads, resulting in information on assay background for each serum, that was used to normalize signals. In the suspension array, the background on the NC beads ranged from 115 to 2119 MFI in layers, and from 46 to 343 MFI in broilers. Sera were considered seropositive when a signal (positive after normalization) on at least one peptide was observed. This resulted in 33 seropositive sera, mostly from layers (Table 4). Of these, 20 sera were seropositive for C. gallinacea, and 19 for other Chlamydia species; 5 sera were positive for multiple Chlamydia species. The results are also presented in Figure 4 were seroreactivity of each serum is visualized (see Additional file 6 for raw MFI values).

Table 4 Summarized results of field sera tested with the suspension array

Figure 4

Results for 120 field sera tested with the suspension array. Normalized MFI values are given and categorized as indicated. Sera that are seroreactive toward peptides of more than one Chlamydia species are labelled with an asterisk below the last peptide recognized. Raw data are given in Additional file 6.

In ELISA, signals for laying hens ranged from OD 0.19 to 1.36 for free range layers and from 0.10 to 1.01 for conventional layers, signals for broilers ranged from 0.00 to 0.132. The ELISA did not reflect the serostatus of individual sera as determined by the suspension array: the range of ODs of seropositives and seronegatives largely overlapped. However, for layers as a group the mean ODs were higher for seropositives than for seronegatives (Additional file 7).

In summary, field sera recognized subsets of Chlamydia-derived peptides, but the in-house ELISA and the suspension array did not agree with each other. In the suspension array, circa two thirds of the seropositive sera recognized C. gallinacea-derived peptides.

Discussion

Current discriminatory tests for Chlamydia are based on detection of genetic material [13] and only give information about the actual infection status, but reveal nothing of past infections. In contrast, serological methods have a larger detection window. Detection of antibodies against Chlamydia has successfully been carried out with sera from various mammalian hosts [17]. The goal of the study presented here was to investigate if it is possible to develop a similar test for sera from chickens. In this study, focus was on C. psittaci and C. gallinacea, the two avian species that need to be distinguished, as one is zoonotic whereas the other currently is the prevalent species in chickens [10]. In addition, two genetically related species were included, i.e. C. abortus and C. avium, both occurring in birds [1], and the species that are most closely related to resp. C. psittaci and C. gallinacea [26]. To enable differentiating serology for the selected Chlamydia species, a bead-based Luminex suspension array was constructed, using peptides derived from immunoreactive Chlamydia proteins [15,16,17] as antigens. The suspension array correctly detected and identified antibodies against Chlamydia in sera from experimentally infected mice, showing that the peptides retained their seroreactivity in this assay format. Importantly, the test was able to differentiate between antibodies against C. psittaci and C. gallinacea in sera from experimentally infected chickens. Results with field sera confirm the high prevalence of C. gallinacea in layers. As yet, signals against peptides derived from other Chlamydia species are difficult to interpret as limited (C. psittaci) or no (C. abortus, C. avium) sera from experimentally infected chickens were available for evaluation.

Observations with infected chickens showed that upon C. psittaci infection specific antibodies were generated. Not all C. psittaci peptides were recognized, which may be an effect of the transient nature or inefficiency of the infection: PCR data showed that the C. psittaci infection of these chickens had been suboptimal and was transient. For C. gallinacea infected chickens there were no doubts concerning the infection, and signals in both ELISA and the suspension array were high upon infection.

It would be of added value to test additional sera from chickens infected with C. psittaci to investigate if the three non-reacting peptides can be recognized if chickens are not transiently infected.

An alternative explanation for the observed differences in seroreactivity of a subset of the C. psittaci peptides may be the dissimilarity of immune systems of mammals and birds, including differences in antibody development and antigen processing [27,28,29]; mice and birds may recognize different epitopes. This speculation is fed by two observations. Of the C. psittaci-derived peptides that are only seroreactive in mice, two have the same sequence as the 6BC isolate used for infecting chickens, yet do not result in a signal in the suspension array. Further, sera of C. gallinacea G47 infected chickens recognize (but inconsistently) the autologous G47 OmpA-derived peptide pCga008, whereas pooled sera from mice infected with C. gallinacea do recognize the autologous peptide, pCga003 that is based on the same region, resulting in a high signal.

With sera from infected mice some cross-reactions were observed, that could not be traced back to sequence homology and were not observed before in the peptide array [17], but one of these cross-reactions has been observed before in ELISA (pCab1, Cab_S26/3_OmpA_89-104) [16]. Since the signals of cross-reactions are low in the suspension array and low or absent in other platforms, it is likely that differences in designated serostatus of sera with low signals, are caused by differences in the selected cutoffs that are used. Also with sera from experimentally infected chickens some cross-reactions were observed. One C. psittaci infected chicken reacted with peptide pCab004, that has 71% identity with its ortholog in C. psittaci 6BC (BLAST). It’s sequence overlaps with a published peptide which is known to cross-react in ELISA with sera from mice infected with C. psittaci [16]. A subset of sera from the C. gallinacea infected chickens also cross-react with C. psittaci peptide pCps015, for which no explanation was found since the peptide has only limited homology to proteins from C. gallinacea (BLAST; lowest E-value 0.97 for 36% of the query). Regardless of the nature of these cross-reactions, in order to provide straightforward interpretation of results, it shows that panels of strong reacting peptide antigens for each species have a preference.

In this study, a generic method was developed for establishing a cutoff for each serum sample. In antibody detection, it is common practice to base cutoffs on the assay background or the average signal of a negative control group of animals, plus two or three times the standard deviation [25, 30]. Here, an approach was adopted to base a cutoff on a negative population of beads within each assay by defining a serum-specific cutoff for each individual serum sample, that is based on the individual assay background of that serum, which in singleplex platforms such as ELISA is not possible for individual samples. This serum-specific assay background was measured using internal negative control (NC) beads that were present in each bead mix. These control beads carry a compound similar to the peptides used, but without an actual peptide moiety and are expected to reflect aspecific binding of serum components in the system. Of these, in each well at least 100 beads were analysed—these are regarded as the negative population of beads within each assay and are thus used as the basis of calculating a cutoff for each serum. When analysing all results, the serum-specific assay background times three was found acceptable as cutoff for all experimental sera. This method of calculating individual serum-specific cutoffs takes away the assay background, resulting in limited aspecific reactions and/or cross-reactions. The cutoff can not take into account cross-reactions of particular peptides; by the time a fully developed assay is available, a comparison with a negative population of animals will help in both finetuning the cutoff and assembling a final peptide panel.

The approach with a universal method to establish serum-specific cutoffs was also employed for analysing field sera, but was not verified with a large set of chicken sera that are seronegative according to a reference test, as these are simply not available. Investigation of 120 Dutch field sera with the suspension array confirmed that C. gallinacea is the prevalent Chlamydia species in layers, which agrees with observations made using PCR [10], and also showed that C. gallinacea antibodies are absent in broilers, at least in the 40 sera tested. The results suggest that the current test may be used to establish if flocks are seropositive for C. gallinacea. Signals against peptides derived from other Chlamydia species than C. gallinacea were observed in mainly laying hens, but are difficult to interpret as limited (C. psittaci) or no (C. abortus, C. avium) sera from experimentally infected chickens are available to evaluate the seroreactivity of these peptide panels. So, the seroreactivity of peptides from species other than C. gallinacea was clearly demonstrated with sera from experimentally infected mice, but sera to do so unambiguously for chickens are not available. Although some information was acquired with sera from immunizations of chickens with C. psittaci and C. abortus bacterins, such information is not available for C. avium (see Additional file 8).

The observation that of these peptide panels the field sera recognized only subsets or individual peptides may indicate presence of specific antibodies, but these “sporadic” signals may also be aspecific interactions or cross-reactions with antibodies against unidentified pathogens. Some peptides cross-reacted in experimental chicken sera (e.g. pCps015) and with field sera probably have no relevance. A peptide such as pCab005 was recognized by field sera as solitary signal or coinciding with signals on C. gallinacea peptides. However, this peptide did not cross-react with experimental C. gallinacea sera, nor was it the strongest responder in the C. abortus peptide panel when tested with mice serum. A complicating issue here is that there is no data on the presence of Chlamydia species/strains (sequence variants) in the sampled chickens to verify the serological signals observed. Another complication is the aforementioned lack of experimental chicken sera for all Chlamydia species (also see Additional file 8).

Concerning possible sequence variants, for C. gallinacea IncA-derived peptides signals were always highest on the autologous versions of these peptides when testing with experimental sera. In field sera however, for each of these three peptides sera exist that are only positive on one variant. This may suggest that multiple C. gallinacea strains occur in the set used.

Regarding comparisons of the results of the ELISA and the suspension array, it is remarkable that with field sera there is no clear correlation between the two assays, whereas for sera from experimentally infected mice and chickens the results of the two assay do agree, i.e. with sera from experimental infections in both assays low signals are observed prior to infection, and high signals are observed after infection. Nevertheless, many field sera show high signals in ELISA and are not seropositive in the suspension array. An explanation may be related to the nature of the antigens. The in-house ELISA is performed with a mixture of inactivated C. abortus and C. trachomatis bacteria, in fact elementary and reticular bodies containing a multitude of antigens, which may in part be responsible for causing cross-reactions. One type of cross-reaction is within the group of Chlamydia species. In ELISA this antigen mixture enables detection in animal sera of antibodies against all other Chlamydia species, as demonstrated in this study with sera from experimentally infected mice and chickens. This is consistent with published observations where such antigens are shown to be seroreactive with antibodies against (all) other Chlamydia species [31]. The broad nature of the ELISA antigens may also be responsible for another type of cross-reactions: cross-reactions by antibodies against other bacteria that occur in chickens, Chlamydia-related or otherwise. Cross-reactions with other Chlamydia species than studied here is unlikely since C. gallinacea is the prevalent species in Dutch poultry [10], and there is no direct evidence for cross-reactivity of antigens between Chlamydia-related species (in mammalian sera) if they are not closely related [32]. The observation that in production animals ELISA signals are always low in young animals (broilers, aged between 4 and 6 weeks) but can be very high in older chickens (layers, in general 18 weeks or older) is indicative of an age-related presence of antibodies; older animals on farms undoubtedly have encountered more bacteria than young chickens or chickens used for experimental infections. Another, simple, explanation could be that the peptide panel is simply missing essential peptides, but if it is, that is only a part of the problem as the observed large number of layers seropositive for C. gallinacea match the prevalence found by PCR [10], a study in which no other Chlamydia species were detected. A complicating factor in this whole discussion is that for the ELISA cutoff has not been established [21]. For now, the mismatch between results from the in-house ELISA and suspension array remains elusive.

The value of having serum-specific cutoffs was shown when testing field sera. For two field sera, extremely high assay backgrounds were observed in the suspension array (> 1000 MFI on NC beads). For one serum, this was also reflected by high signals on other beads, whereas the standard deviation (SD) for the NC beads (measured in three parallel assays) was low (38 MFI). Despite the high serum-specific assay background, and hence the high cutoff, this serum had only three signals that exceeded the cutoff, and was designated seropositive for C. gallinacea with seroreactivity on the same peptides as seen for many other sera. For the other serum, the high serum-specific assay background was reflected by many, but not all beads, and the SD NC was also high (237 MFI). Although this serum was designated seronegative, it clearly is an aberrant serum with unusual behaviour toward the control beads. In the experimental setup with three parallel tests this was visible (NC beads are used in each parallel test), but in a complete multiplex test however this would go unnoticed as all beads would be tested in one well. So, an additional rule to filter out sera with an aberrant/irregular background signal could improve the suspension array. Summarized, these observations show the value of a serum-specific cutoff as it corrects for high assay background, but also suggest that some improvements for sera with an irregular background may be required.

The developed serological Luminex suspension array described in this work shows great promise as it is able to discern in experimentally infected chicken sera if antibodies against C. psittaci or C. gallinacea are present, and confirms that C. gallinacea is the prevalent Chlamydia species in layers. The protocol to produce beads with synthetic peptides is simple, and performing a multiplex assay with a panel of peptides is uncomplicated and not troubled by background issues, the latter in part facilitated by using negative control beads that allow individual cutoffs for each serum.

To develop this proof of concept assay into a full multiplex assay for detection and differentiation of antibodies against Chlamydia species, a panel of peptides is required that covers multiple antigenic regions of the targeted species, either conserved linear epitopes or a panel covering sequence variants. These can be obtained from the relevant publications [15,16,17], a new inventory of the ever increasing number of genome sequences, and/or by (re)screening antigenic proteins using arrays with overlapping peptides. To do this, additional chicken sera are required that are experimentally infected with the various Chlamydia species of interest. Further, for validation a set of paired samples should be available from a longitudinal study in poultry that allow detection and identification of the infecting Chlamydia species by PCR and evaluation of peptides by serology.