Development of a bead-based multiplex assay for the simultaneous detection of porcine inflammation markers using xMAP technology


Institute of Pathology, University of Bern, Murtenstrasse 31, CH-3010 Bern, Switzerland. E-mail:


Commercially available assays for the simultaneous detection of multiple inflammatory and cardiac markers in porcine blood samples are currently lacking. Therefore, this study was aimed at developing a bead-based, multiplexed flow cytometric assay to simultaneously detect porcine cytokines [interleukin (IL)-1β, IL-6, IL-10, and tumor necrosis factor alpha], chemokines (IL-8 and monocyte chemotactic protein 1), growth factors [basic fibroblast growth factor (bFGF), vascular endothelial growth factor, and platelet-derived growth factor-bb], and injury markers (cardiac troponin-I) as well as complement activation markers (C5a and sC5b-9). The method was based on the Luminex xMAP technology, resulting in the assembly of a 6- and 11-plex from the respective individual singleplex situation. The assay was evaluated for dynamic range, sensitivity, cross-reactivity, intra-assay and interassay variance, spike recovery, and correlation between multiplex and commercially available enzyme-linked immunosorbent assay as well as the respective singleplex. The limit of detection ranged from 2.5 to 30,000 pg/ml for all analytes (6- and 11-plex assays), except for soluble C5b-9 with a detection range of 2–10,000 ng/ml (11-plex). Typically, very low cross-reactivity (<3% and <1.4% by 11- and 6-plex, respectively) between analytes was found. Intra-assay variances ranged from 4.9 to 7.4% (6-plex) and 5.3 to 12.9% (11-plex). Interassay variances for cytokines were between 8.1 and 28.8% (6-plex) and 10.1 and 26.4% (11-plex). Correlation coefficients with singleplex assays for 6-plex as well as for 11-plex were high, ranging from 0.988 to 0.997 and 0.913 to 0.999, respectively. In this study, a bead-based porcine 11-plex and 6-plex assay with a good assay sensitivity, broad dynamic range, and low intra-assay variance and cross-reactivity was established. These assays therefore represent a new, useful tool for the analysis of samples generated from experiments with pigs. © 2013 International Society for Advancement of Cytometry

Pigs have been widely used as biomedical research models over the past decades [1]. Based on their comparative anatomic and physiologic characteristics [2], they represent a suitable model species for investigation of a large number of human diseases and for technical developments in surgery or anesthesia. Animal experiments using pigs have made valuable contributions not only in the field of human medicine, including research of the cardiovascular [3, 4] system, but also in the field of critical and intensive care medicine [5, 6].

For the evaluation of such experiments, analyses of blood samples for the detection of cytokine profiles or organ-specific markers are indispensable tools. Cytokine levels are traditionally measured by enzyme-linked immunosorbent assay (ELISA) allowing analysis of only a single marker at a time. These tests may be rather expensive and time consuming. In particular, the assessment of multiple markers may require a considerable volume of serum or plasma. This may be a limiting factor in which minimal sample volume is available as is the case from microdialysis studies or long experiments that require serial testing of multiple parameters. Besides, cytokines themselves may alter each other's function and regulate the production of other cytokines [7], rendering cytokine profiles or ratios more valuable than single cytokine measurements [8]. For these reasons, multiplex flow cytometry by the xMAP Technology has rapidly established itself [9, 10].

Multiplexed bead-based immunoassays for quantitation of human cytokines have been described in the past years [11-13], and several multiplexed cytokine detection assays are now commercially available to detect human, mouse, and rat cytokines based on xMAP technology. Multiplex kits for porcine cytokines are commercially available. However, no such kits for porcine markers combining cytokines, complement activation markers, and growth factors/angiogenesis factors are currently available. A major limitation for the development and use of porcine multiplex assays has been the lack of specific antibody pairs. Only relatively few reagents are currently available commercially for the detection of selected markers. In addition, microsphere-based multiplex assays for the detection of porcine cytokines have been recently reported [14-16].

In this article, we describe the development of a novel assay to detect porcine proinflammatory cytokines [interleukin (IL)-1β, IL-6, IL-8, tumor necrosis factor alpha (TNF-α), monocyte chemotactic protein (MCP)-1], the anti-inflammatory cytokine IL-10, growth factors [basic fibroblast growth factor (bFGF), vascular endothelial growth factor (VEGF), platelet-derived growth factor-bb (PDGF-bb)], complement activation markers [C5a, soluble (s)C5b-9], and a cardiac marker [cardiac troponin (cTn)-I]. We have developed an 11-plex assay including all parameters besides cTn-I (IL-1β, IL-6, IL-8, IL-10, TNF-α, C5a, sC5b-9, MCP-1, bFGF, VEGF, and PDGF-bb) and a 6-plex assay with a subset of the markers of the 11-plex assay plus cTn-I (IL-1β, IL-6, IL-8, IL-10, TNF-α, and cTn-I). As a multiplex for cytokine detection, this assay may serve as a tool to determine the “inflammation fingerprint” in a given pig experiment, and as a 6- or 11-plex may be particularly useful in, for example, the context of studying cardiac, limb, or general ischemia/reperfusion injury or models of shock.



The described assays are based on Luminex xMAP technology, a multiplexed sandwich immunoassay technique performed on the surface of 5.5-μm polystyrene beads. Details of the assay principle have been published elsewhere [17] and are available online (Bio-Plex Suspension Array System by Bio-Rad, Hercules, CA). The system was primarily set up as a single plex and assembled stepwise to a multiplex assay (see below).

Bead Coupling

Carboxylated polystyrene beads [catalog no. 171-5060(24), (25), (27), (28), (35), (36), (38), (43), (44), (46), (50), (54), (56)] and the Bio-Plex Amine Coupling Kit (catalog no. 171-406001) were purchased from Bio-Rad, and coupling procedures were performed according to the manufacturer's instructions. All traces of sodium azide were removed prior to use by dialysis (Slide-A-Lyzer MINI Dialysis Units, Pierce, Rockford, IL).

With the exception of polyclonal antibodies to detect porcine IL-6, MCP-1, bFGF, VEGF, PDGF-bb, and cTn-I, monoclonal mouse antibodies were used in the first step of the sandwich immunoassay (Table 1). The anti-human bFGF, VEGF, PDGF-bb, sC5b-9, and cTn-I antibodies were found to cross-react with porcine antigen; all other antibodies were described by the manufacturers to be specific for pig. The amount of antibody needed for a successful coupling reaction and assay performance was between 15 and 25 μg per 1.25 million beads. The efficiency of antibody coupling was validated using a biotin-conjugated antibody directed against the Fc region of the respective coupled antibody followed by streptavidin-PE by measuring the mean fluorescence intensity (MFI) on the Bio-Plex platform. MFI > 2000 was taken to indicate successful coupling against a reaction using coupled beads with streptavidin-PE only as background. An example of the original sampling is shown in a histogram of double discriminator gating and a bead map with regions for the 11-plex setting is shown in the supplementary figure.

Table 1. Overview of all the reagents and diluents used for the multiplex immunoassay to detect porcine cytokines, chemokines, growth factors, and cTn-I
 Capture antibodiesDetection antibodiesRecombinant protein
MarkerCatalog numberCloneSourceCoupling concentration (μg/ml)Catalog numberCloneSourceCatalog numberSource
  1. mAb, monoclonal antibody; pAb, polyclonal antibody; coupling concentration, antibody concentration for coupling assay.
  2. Abcam: Abcam (Cambridge, MA); CompTech: Complement Technology (Tyler, TX); Diatec: Diatec Monoclonals AS (Oslo, Norway); HyTest: HyTest (Turku, Finland); Invitrogen: Invitrogen (Carlsbad, CA); King Biotech: Kingfisher Biotech (Saint Paul, MN); MBM: MBM Sciencebridge GmbH (Göttingen, Germany); PepriTech: PeproTech EC (London, UK); R&D: R&D Systems (Minneapolis, MN).
7C5a T13/9MBM20 pAbMBM MBM
9MCP-1500-P34pAbPeproTech20PBB0089SpAbKing BiotechRP0017SKing Biotech

Multiplex Assay

All details concerning origin of reagents used for the assay are listed in Table 1. Antibody-coupled beads were diluted in assay diluent [a mixture (1:1) of PBS and 0.5% BSA pH 7.2 and Tris pH 7.6] at a concentration of 2500 beads per 50 μl. Bead suspension (50 μl per well) was added to a microplate (MultiScreen HTS 1.2-μm Durapore Membrane, Millipore Corporate, Billerica, MA). Following washing with wash buffer [a mixture (1:1) of PBS 0.1 M pH 7.2 and Tris 0.5 M pH 7.2], 50 μl of standards/samples diluted in assay buffer containing 0.5% polyvinyl alcohol (PVA) and 0.8% polyvinylpyrrolidone (PVP) were added to the wells. Incubation was performed for 30–60 min at room temperature, protected from light under continuous gentle shaking, and followed by three washing steps. The biotinylated detection antibodies were diluted in assay buffer to the concentration of 1 μg/ml and 25 μl per well. The detection antibody solution was added and incubated on the shaker for another 30 min in the dark. After three washing steps, streptavidin-R-PE (Qiagen, Hilden, Germany) was added (at 1 μg/ml). After 10 min of incubation and washing, the plate was subsequently analyzed by the Bio-Plex reader.

Dynamic Range and Limit of Detection

Standard curves with large ranges were set up to determine the dynamic range of the assay: concentrations from 2.5 to 10,000 pg/ml (PDGF-bb), 2.5–15,000 pg/ml (IL-8), 5–15,000 pg/ml (MCP-1, bFGF, and VEGF), 5–30,000 pg/ml (IL-1β, IL-6, and IL-10), 10–25,000 pg/ml (C5a), 5–24,000 pg/ml (TNF-α), 10–10,000 pg/ml (cTn-I), and 2–10,000 ng/ml (sC5b-9) (11-plex, Table 2; 6-plex, Fig. 1).

Figure 1.

Standards' recovery curves. Standard dilution curves for recombinant proteins from 11-plex (A and B) showing the range of the curves with a standards' recovery value between 70 and 130%. Data were generated with a five-parametric regression formula using the Bio-Plex Manager software and plotted in a log–log curve. A representative curve for each parameter is shown. Each presented point corresponds to an average fluorescence value of at least 100 analyzed beads for the individual parameter.

Table 2. LLOD and dynamic ranges
  1. aLowest limit of detection (LLOD) is defined as two standard deviations above the mean fluorescence intensity of 10 replicates of the zero standard in the 11- and 6-plex assays.
  2. bRange is defined as part of the standard curve where the ratio of observed to expected, known amount of standard, expressed as a percentage [(observed concentration/expected concentration) × 100] lies between 70 and 130%.
 IL-1β (pg/ml)IL-6 (pg/ml)IL-8 (pg/ml)IL-10 (pg/ml)TNF-α (pg/ml)MCP-1 (pg/ml)bFGF (pg/ml)VEGF (pg/ml)PDGF-bb (pg/ml)C5a (pg/ml)sC5b-9 (ng/ml)
LLODa (pg/ml)5555101010102.5102
Rangeb (pg/ml)5–30,0005–30,0005–15,0005–30,0005–24,0005–15,0005–15,0005–15,0005–10,00010–25,0002–10,000
 IL-1β (pg/ml)IL-6 (pg/ml)IL-8 (pg/ml)IL-10 (pg/ml)TNF-α (pg/ml)cTn-I (pg/ml) 
LLODa (pg/ml)55551010 
Rangeb (pg/ml)5–30,0002.5–30,0005–15,0005–30,0005–24,00010–10,000 

Determination of the “standard recovery” (known as back calculation of standards) is used to assess the quality of a curve fit by comparing calculated concentrations, which are based on the standard curve calculated by a five-parameter logistic equation, to expected concentrations (ratio of observed to expected known amount, expressed as a percentage) [18]. By definition, the part of the standard curve showing standard recovery percentages between 70 and 130% was considered acceptable and defined as the dynamic range (Bio-Rad Principles of Curve Fitting for Multiplex Sandwich Immunoassays, Rev B). Ten replicates of the zero standard were run to identify the average fluorescence intensity and standard deviation. The lowest limit of detection (LLOD) was defined as the lowest concentration of the standard dilution with an average fluorescence intensity of two standard deviations above the average fluorescence intensity of the zero standard. It provides a reasonable limit to differentiate between an actual minimal amount of sample and background noise and provides more valuable information concerning the minimal detectable amount of a specific protein than the dynamic range, which expresses the range of the detectable concentration in relation to the quality of curve fit. Data were obtained and analyzed using the Bio-Plex-manager Software 6.1 (Bio-Rad).


To analyze cross-reactivity, two tests were performed:

  1. Standard cross-reactivity: Single standards containing one of the recombinant proteins at a known concentration were run in the presence of all capture beads and all biotinylated reporters. Obtained fluorescence intensities were compared with the average background values of five replicates of the zero standard. In case of cross-reactivity, the concentration of a particular analyte was obtained from a standard curve, and the detection by nonrelevant antibodies was expressed as a percentage of the actual analyte concentration [(observed concentration/actual concentration) × 100].
  2. Reporter cross-reactivity: Standards containing all recombinant proteins at known concentration were run in the presence of all capture beads and single biotinylated reporter. Percentage of cross-reactivity was calculated according to the same formula as for standard cross-reactivity.


Porcine plasma samples/standard diluent-only were spiked with a known concentration of recombinant protein, aliquoted, and stored at −80°C until use. To determine intra-assay and interassay variance, four to five replicates of each spiked sample (50 μl) were either read on the same plate or on different plates. The obtained values were compared, and the respective coefficient of variation was calculated by the following formula: coefficient of variation = (standard deviation/average) × 100.

Spike Recovery

Ethylenediaminetetraacetic acid (EDTA) plasma samples or EDTA plasma prediluted in standard/sample diluent were spiked with known amounts of recombinant proteins and analyzed to determine the proximity of the expected value to the actual value measured in the spiked plasma. This method assessed variability due to assay preparation, the interference of substances present in the sample or sample matrix, and the regression analysis [19]. In addition, sample diluent-only was spiked with recombinant proteins and tested accordingly.

Correlation Between 6-plex/11-plex with Respective Singleplex

Standard/sample diluent-only was spiked with different amounts of recombinant proteins (IL-1β, IL-6, IL-8, IL-10, and TNF-α) and analyzed with 6-plex and 11-plex assays as well as singleplex assays. The results were evaluated by correlation with the respective cytokine from 6-plex with corresponding singleplex assays. Similarly, all parameters from the 11-plex assay were evaluated by correlation with the respective singleplex.

Sample Testing

To evaluate how applicable the developed multiplex assay is for the analysis of actual experimental porcine samples, for instance, from an experimental setting of cardiopulmonary bypass (CPB) or extracorporeal limb perfusion, EDTA plasma samples were analyzed at different time points [baseline, during reperfusion (120 min), and during euthanasia or off CPB]. For the 11-plex, samples were obtained from extracorporeal porcine limb perfusion experiments. This study was aimed at assessment of ischemia/reperfusion (I/R)-induced endothelial activation/injury in amputated pig limbs following prolonged preservation by the use of modified extracorporeal perfusion [20]. Briefly, porcine forelimbs were perfused with autologous blood using extracorporeal circulation (ECC) for 12 h following 6 h of cold ischemia (4°C). The limbs were subsequently replanted with a 7-day follow-up. EDTA plasma samples were analyzed for the extent of reperfusion injury. Care and use of animals in the study were in compliance with the European Convention on Animal Care and the respective Swiss National Guidelines. The study was approved by the Animal Experimentation Committee of the Canton of Bern, Switzerland.

Statistical Analysis

All data obtained from the multiplex assays were generated with a five-parametric regression formula using the Bio-Plex Manager software 6.1 (Bio-Rad), and standard curves were constructed by a five-parameter regression formula. Prism version 5.0b (Graphpad Software, La Jolla, CA) software was used for all analyses.


Setting up multiplex assays requires identifying the specific antibody pairs, the appropriate diluents as well as establishing optimized dilution factors. Different diluents were tested for their potential to reduce nonspecific binding or so-called matrix effects. In the case of experimental samples, these effects may be mediated by heterophilic antibodies, although these have previously only been described in human sera [21]. In the optimized assay, 0.8% PVP and 0.5% PVA were applied as adjuvants to the standard/sample diluent, as previously described [22]. Commercially available high-performance ELISA buffer/assay diluent was chosen as a diluent for the detection antibodies.

Lowest Limit of Detection

The assay achieved a LLOD of 5 pg/ml for IL-1β, IL-6, and IL-8, 10 pg/ml for TNF-α, cTn-I, C5a, MCP-1, bFGF, and VEGF, and 2 ng/ml for sC5b-9, respectively (6-plex and 11-plex; Table 2).


1. Standard cross-reactivity: Tests with nonrelevant cytokine standards showed no fluorescence intensity greater than two standard deviations above the average background value of five replicates of the zero standard. Recombinant cTn-I appeared to interact with all nonrelevant capture antibodies to a certain degree, showing a cross-reactivity of 0.8% with the TNF-α antibody pair and between 1.1 and 1.4% with IL-1β, IL-6, IL-8, and IL-10 antibodies (6-plex setting).

Recombinant MCP-1, bFGF, VEGF, and PDGF-bb showed interaction with all nonrelevant capture antibodies to a certain degree, with a cross-reactivity of 0.5–3% (11-plex; Table 3).

Table 3. Analysis of cross-reactivity between reagents (11-plex)
 Recombinant proteins
Beads IL-1β (pg/ml) 8554IL-6 (pg/ml) 8000IL-8 (pg/ml) 8554IL-10 (pg/ml) 4000TNF-α (pg/ml)9000MCP-1 (pg/ml) 10,000bFGF (pg/ml)10,000VEGF (pg/ml) 10,000PDGF-bb (pg/ml)10,000C5a (pg/ml) 27,200sC5b-9 (ng/ml) 3333
  1. N = below detectable range.

2. Tests for reporter cross-reactivity showed minimal detection using the cTn-I-specific antibody as a single biotinylated reporter. Cross-reactivity was 0.8% for IL-1β, 0.9% for IL-8, and 0.7% for IL-10 (6-plex). However, there was no significant recognition of recombinant proteins by nonrelevant detection antibodies (6- and 11-plex).


The coefficients of variation for intra-assay replicates lay between 5.3–12.9% (mean: 9.5%, 11-plex) and 4.9–7.4% (mean: 5.6%, 6-plex) for all samples. Interassay replicates showed coefficients of variation, which ranged from 10.1 to 26.4% (mean: 10.8%, 11-plex) and 8.1 to 28.8% (mean: 13.7%, 6-plex; Table 4).

Table 4. Intra-assay and interassay variance of the multiplex setup
  1. Intra-assay variance: eight replicates of spiked samples at different concentrations measured on the same plate. Coefficient of variation = [(standard deviation/average) × 100]. Interassay variation: the coefficient of variation for three replicates of distinct concentrations measured on different plates at different time points.
  2. aFor each assay (by 6-plex), standard dilutions for cTn-I were newly prepared, whereas cytokines were spiked and stored at −80°C.
Intra-assay (%) (n = 8)
Interassay (%) (n = 3)18.414.424.016.615.
Intra-assay (%) (n = 6) 
Interassay (%) (n = 3) 

Spike Recovery

Depending on the parameter tested, recovery values for spiked plasma samples varied from 13.1 to 42.4% and 15.8 to 114.7% for the 11- and 6-plex, respectively (Table 5 and Fig. 2). For the individual parameter, however, the percentage of recovery was fairly consistent within the different concentrations. To test for matrix effects, EDTA plasma was prediluted (1:5 or 1:10) in a dose-dependent manner with standard/sample diluent and spiked with a known concentration of the recombinant protein. The percent recovery of spiked, prediluted EDTA plasma samples showed increased recovery (% recovery between 57.8 to 83.0% and 76.3 to 102.1% for 1:5 and 1:10 predilution, respectively) when compared with undiluted plasma, suggesting a relevant matrix on spiking. Furthermore, spiking of recombinant proteins in standard/sample diluent-only without plasma showed far less variation between parameters and consistent recovery values between 90 and 118% and 75 and 115% for 11- and 6-plex, respectively.

Figure 2.

Spike recovery in plasma samples and diluent-only by 6-plex. Spike recovery to determine the proximity of the expected value to the actual value measured (6-plex): EDTA plasma samples were spiked with different amounts of recombinant protein (3200, 400, and 50 pg/ml) and analyzed to determine the proximity of the measured values to the expected ones [% recovery = (observed − neat)/expected) × 100] (A). Although spike recovery was consistent for the individual parameter measured, recovery percentages between different parameters varied for spiked plasma samples. In addition, diluent-only was spiked with the respective recombinant proteins (B). Recovery percentages were consistent for spiked diluent-only samples (between 75 and 115%). Data are expressed as average ± standard deviation.

Table 5. Spike recovery
MarkerExpected concentrationStd. diluentsEDTA plasma undilutedEDTA-P + Std. diluent (1:5)EDTA-P + Std. diluent (1:10)
Observed concentration% RecoveryObserved concentration% RecoveryObserved concentration% RecoveryObserved concentration% Recovery
  1. Std. diluent = standard/sample diluent; EDTA-P + Std. diluent (1:5/1:10) = porcine EDTA plasma prediluted in standard/sample diluent to 1:5 or 1:10.
IL-1β2000 pg/ml2006.0 ± 101.6100.3 ± 5.1695.9 ± 20.534.8 ± 1.01363.0 ± 43.768.1 ± 2.22015.0 ± 237.1100.8 ± 11.8
IL-64000 pg/ml4031.0 ± 95.1100.8 ± 2.4523.0 ± 112.113.1 ± 2.82310.0 ± 151.957.8 ± 3.83967.0 ± 10.999.2 ± 0.3
IL-85000 pg/ml5358.0 ± 808.4107.2 ± 16.21490.0 ± 56.629.8 ± 1.13691.0 ± 513.173.8 ± 10.35002.0 ± 1197.0100.0 ± 23.9
IL-102000 pg/ml1973.0 ± 118.098.6 ± 5.9268.9 ± 13.813.5 ± 0.71284.0 ± 40.364.2 ± 2.01563.0 ± 13.578.2 ± 0.7
TNF-α1300 pg/ml1320.0 ± 19.7101.6 ± 1.5551.5 ± 12.242.4 ± 0.91088.0 ± 35.583.7 ± 2.71327.0 ± 18.6102.1 ± 1.4
MCP-15000 pg/ml5530.0 ± 601.4110.6 ± 12.01263.0 ± 60.925.3 ± 1.23838.0 ± 110.576.8 ± 2.24861.0 ± 54.297.2 ± 1.1
bFGF2000 pg/ml2168.0 ± 59.9108.4 ± 3.0605.9 ± 41.930.3 ± 2.11190.0 ± 43.159.5 ± 2.21525.0 ± 185.276.3 ± 9.3
VEGF5000 pg/ml5263.0 ± 156.4105.3 ± 3.11443.0 ± 60.028.9 ± 1.22983.0 ± 125.159.6 ± 2.54171.0 ± 193.383.4 ± 3.9
PDGF-bb2000 pg/ml1999.0 ± 30.699.9 ± 1.5510.2 ± 24.725.5 ± 1.21290.0 ± 45.464.5 ± 2.31539.0 ± 122.776.9 ±6.1
C5a3400 pg/ml3653.0 ± 150.9107.4 ± 4.41153.0 ± 107.333.9 ± 3.22010.0 ± 296.559.1 ± 8.72680.0 ± 288.278.8 ± 8.5
C5b-91000 ng/ml1107.0 ± 56.4110.7 ± 5.6227.2 ± 28.822.7 ± 2.9644.8 ± 56.364.5 ± 5.6963.7 ± 136.096.4 ± 13.6

Correlation Between Multiplex Assay (6-plex or 11-plex) and Respective Singleplex

The concentration of the majority of samples measured with multiplex correlated well with the concentrations measured in the respective singleplex. Correlation coefficients (r2) were 0.9888 for TNF-α, 0.9945 for IL-1β, 0.9965 for cTn-I, 0.9967 for IL-8, 0.9968 for IL-6, and 0.9976 for IL-10 by 6-plex and singleplex (Fig. 3A). Similarly, high correlation coefficients (r2) between 11-plex and the respective singleplex assays ranging from 0.913 for sC5b-9 to 0.999 for bFGF were found (Fig. 3B).

Figure 3.

Correlation between 6-plex and singleplex as well as 11-plex and singleplex. Standard/sample diluent-only spiked with different amounts of recombinant cytokines were measured in the 6-plex assay as well as in the singleplex. Correlation between 6-plex and corresponding singleplex was good with coefficients (r2) of 0.9888 for TNF-α, 0.9945 for IL-1β, 0.9965 for cTn-I, 0.9967 for IL-8, 0.9968 for IL-6, and 0.9976 for IL-10 (A). The 11-plex assay also correlated well with corresponding singleplex assays, r2 ranging from 0.9133 (sC5b-9) to 0.9990 for bFGF (B).

Sample Testing

The course of cytokines, growth factors, and complement activation markers in EDTA plasma samples of a model of ECC limb perfusion was analyzed by 11-plex assay at different time intervals of perfusion as shown in Figure 4. Increased concentrations of cytokines (IL-1β, IL-6, IL-8, and MCP-1), growth factors (VEGF), and complement activation markers (C5a and sC5b-9) were observed during ECC. IL-10, TNF-α, bFGF, and PDGF-bb showed only a minimal increase during ECC. However, the I/R injury-induced inflammatory response and complement activation appear to be of limited duration in this model as activation markers dropped back down to baseline levels by the end of the 7-day replantation period.

Figure 4.

Plasma cytokine levels measured during extracorporeal circulation. Samples from porcine forelimb perfusion experiments in ECC models were used to validate the established 11-plex assay. EDTA plasma levels of cytokines, MCP-1, VEGF, PDGF-bb, bFGF, C5a, and sC5b-9 were reanalyzed by 11-plex at different time points as shown in the figure. During perfusion, IL-1β, IL-6, IL-8, MCP-1, and VEGF as well as complement activation marker, C5a and sC5b-9, levels were significantly higher at perfusion time points than baseline, as shown by one-way ANOVA with Bonferroni correction (*P < 0.05; **P < 0.01; ***P < 0.001). Concentrations of all measured markers dropped to normal after replantation.


In this study, an 11-plex assay for the simultaneous detection of porcine IL-1β, IL-6, IL-8, IL-10, TNF-α, C5a, sC5b-9, MCP-1, bFGF, VEGF, and PDGF-bb as well as a 6-plex assay including five cytokines and cTn-I was set up following initial optimization of all individual parameters in a singleplex setting.

Although various studies have described the setup of multiplex assays for the detection of human neutralizing monoclonal antibodies [23] or human cytokines [13, 24, 25], only few studies have been currently published describing multiplex detection of porcine cytokines [14-16]. The potential advantage of the current assay, apart from including a larger number of cytokines, is the addition of complement activation markers and growth factors as well as tissue-specific injury markers (cTn-I). The multiplex assay may be used in more general porcine models of tissue injury or, with the addition of cTn-I, in particular for the investigation of cardiac injury, for example, in myocardial infarction. However, the assay sensitivity for the complement markers (C5a and sC5b-9) as well as growth factors (VEGF and PDGF-bb) was compromised when combined with cTn-I. In addition, with the increase of background values, the LLOD of these markers was high in the presence of cTn-I antibodies. Thus, two multiplex panels were developed as an 11-plex and an overlapping 6-plex with the addition of cTn-I.

Sensitivity of the assay (for cytokines) in terms of the lowest level of detection was comparable with commercial ELISAs. In addition, the dynamic detection range for all analytes was significantly broader in our bead-based assays. For cytokines, the range was higher than in commercial ELISAs, eliminating the need for multiple dilutions of high-concentration samples. Moreover, the simultaneous detection of analytes insured an internal consistency.

Standard cross-reactivity tests showed some recognition of recombinant PDGF-bb protein by all beads in the multiplex setting. As a certain amount of cross-reactivity was observed with all beads, this phenomenon is most likely due to nonspecific binding of MCP-1, bFGF, VEGF, and PDGF-bb to the beads than due to recognition of the protein by other capture antibodies. This may also explain the minimal cross-reactivity found in reporter cross-reactivity tests when analyzing all beads with detection antibodies. As far as intra-assay and interassay variance is concerned, the presented multiplex assay showed values comparable with and within an acceptable range when compared with those observed by others [13, 26].

Although recovery values for spiked plasma samples varied to a certain extent, the percentage of recovery was stable within each tested parameter. By prediluting the EDTA plasma in standard/sample diluent, the percent recovery of each parameter following spiking was dose dependently increased to a maximum. This indicates that the low recovery values were due to absorption of the respective recombinant protein following spiking into the plasma samples. Furthermore, there was a good correlation between the results obtained in singleplex versus 6-plex or singleplex versus 11-plex. Additionally, cytokine concentrations measured by 6-plex and 11-plex assays showed correlation coefficients (r2), 0.9720 for IL-8, 0.9736 for TNF-α, 0.9873 for IL-6, 0.9904 for IL-1β, and 0.9995 for IL-10, indicating for a good sensitivity of both assays. In contrast, recovery percentages in diluent-only spiked samples were more consistent and generally higher, most likely due to significantly reduced matrix effects. Taken together, interassay variance and spike recovery values of some analytes showed deviations from expected values; however, relative concentrations of each parameter were consistent.

Cytokine concentrations in the multiplex assay correlated well with those measured using commercial ELISAs. Although in certain longitudinal studies, such as in the illustrated model, relative cytokine values may be adequate to evaluate the follow-up, some experimental setups require exact values. In general, however, absolute concentrations of antigens as measured by ELISA or similar technique have to be regarded with caution as values may differ significantly among commercially available kits [27]. In multiplex analysis, at least the issue of repeated testing (on different plates and different days) may be reduced as up to 80 samples, and in the current assay, up to 11 parameters can be measured simultaneously on one single plate.

The established 6-plex and 11-plex assays were validated using samples from in vivo porcine experiments in CPB (data not shown) and ECC models, respectively [28]. The increase in IL-1β, IL-6, IL-8, TNF-α, and cTn-I on reperfusion post-CPB fit in with data from other authors. The course of IL-10 post-CPB has, in part, been shown to be biphasic, with an initial drop in early reperfusion [29]. However, an early increase has been described by others [30]. In the limb perfusion model with ECC, plasma samples were measured using 11-plex. IL-1β, IL-6, IL-8, MCP-1, C5a, and sC5b-9 as well as VEGF levels were increased significantly from baseline during ECC, and all markers were again back to normal at the end of the replantation period. Endothelial activation due to I/R injury induces inflammatory reactions, for example, shedding of endothelial glycocalyx, activation of complement and coagulation cascades, and increased cytokine release, causing endothelial dysfunction [31-34]. In this study, we observed that prolonged preservation of amputated pig limbs using ECC has no effects on I/R-induced injury as the results fit in with the expected course of inflammatory markers within this experimental setting [20].

In conclusion, we were able to establish a bead-based multiplex assay to detect porcine cytokines (IL-1β, IL-6, IL-10, and TNF-α), chemokines (IL-8 and MCP-1), growth factors (bFGF, VEGF, and PDGF-bb), injury markers (cTn-I), and complement activation markers (C5a and sC5b-9) in porcine blood samples with a good assay sensitivity, broad dynamic range, and low intra-assay variance. The simultaneous detection of analytes allows for an internal consistency and greatly increases the amount of information obtained from a single volume-limited sample. This method should prove a valuable tool in the analysis of porcine samples obtained from experimental setups using pig as a model animal to study various states of disease, including cardiovascular disease and shock.


The authors thank the team of the Institute of Clinical Chemistry of the University Hospital of Bern for support and technical assistance.

Author Contributions

A.K.B. and J.L. performed most of the experimental work (data collection), analyzed and interpreted the data, performed statistical analyses, and wrote and critically revised the manuscript. R.R. participated in the concept and design of the study, helped to write and critically edit the manuscript, and carried a part of the overall responsibility. Y.B. participated in the concept and design of the study, helped to write and critically edit the manuscript, and carried a part of the overall responsibility. All authors read and approved the final version of the submitted manuscript.