Thanasak
Rakmanee
ab,
Irwin
Olsen
*a,
Gareth S.
Griffiths
c and
Nikolaos
Donos
a
aEastman Dental Institute for Oral Health Care Sciences, Division of Biomaterials and Tissue Engineering, UCL-Eastman Dental Institute, 256 Gray's Inn Road, London, UK WC1X 8LD. E-mail: i.olsen@eastman.ucl.ac.uk; Fax: +44(0)207975 1254; Tel: +44(0)207915 1254
bFaculty of Dentistry, Thammasat University, Patumthani, Thailand
cSchool of Clinical Dentistry, University of Sheffield, Sheffield, UK
First published on 17th November 2009
Large amounts of biological samples are usually required to measure multiple components by the enzyme-linked immunosorbant assay. However, the amounts of many tissue extracts and fluids, including gingival crevicular fluid (GCF), are generally extremely small. The aim of this study was, therefore, to develop and validate a novel multiplex bead assay (MBA) to simultaneously measure a profile of healing-related mediators in the GCF of treated periodontal wounds. An MBA was developed and validated by assessment of assay selectivity, recovery, precision and sensitivity, using eight recombinant human growth mediators as assay standards. GCF samples were collected on paper strips from healing wound (test) and healthy unaffected (control) sites of 15 patients with periodontitis, seven days post-periodontal surgery. Each GCF sample was eluted and the levels of the mediators measured using the MBA and antibody pairs specific for angiopoietin-1, vascular endothelial growth-factor, bone morphogenetic protein-2, osteoprotegerin, tissue inhibitor of metalloprotease-1 (TIMP-1), basic fibroblast growth-factor, keratinocyte growth-factor, and platelet derived growth-factor. Less than 1.8% of cross-reactivity was observed between antibodies and the eight different analytes, for which the recovery was more than 85%. Mean intra- and inter-assay precision were within the acceptance criteria of 20% and 25%, respectively. Detection of all mediators was highly sensitive (≤70 ng/L) except for TIMP-1 (215 ng/L). Angiogenic factors were the most highly secreted in the GCF seven days post-surgery. This new MBA can simultaneously measure small amounts of eight different growth mediators in the GCF of healing periodontal wounds. It might also be a valuable tool for evaluating the components of wound fluids as a prognostic indicator of the success of therapeutic intervention.
GF are pivotal in mediating periodontal wound healing and regeneration and their effects on periodontal cells in vitro and in vivo have been extensively investigated. For example, platelet derived growth-factor (PDGF) was found to promote regeneration of both hard and soft periodontal tissues,8 while keratinocyte growth-factor (KGF) has a major role in re-epithelialisation,9 including the formation of the long-junctional gingival epithelium.10 Basic fibroblast growth-factor (bFGF), vascular endothelial growth-factor (VEGF) and angiopoietin-1 (Ang-1) play important parts in vascularisation/angiogenesis during the early phase of wound healing,11–13 while bone morphogenetic protein-2 (BMP-2) has been shown to recruit undifferentiated mesenchymal cells and induce bone formation14 and cementogenesis in vivo.15 Osteoprotegerin (OPG) has a major role in bone deposition and bone remodelling,16 and tissue inhibitor of metalloprotease-1 (TIMP-1) is associated with the turnover of extracellular matrix components as well as with cell migration and angiogenesis.17–19
Studies of the molecular events which accompany the healing of damaged tissue in vivo have, however, been limited by the very small amounts of wound fluid which can generally be obtained. Thus, the majority of such studies have examined only one mediator in each situation, usually by enzyme-linked immunosorbant assay (ELISA). Although this technique is relatively specific and sensitive, it precludes the systematic examination of the profile of multiple factors in a small sample of biological material. While a ‘sequential’ ELISA could possibly overcome some of these constraints,20 it is time-consuming, impractical for routine analysis and not cost-effective, therefore there is a need for a procedure which is able to rapidly and accurately determine the profile of multiple components in small amounts of biological fluid.
Multiplex bead technology utilising flow cytometry (FCM) enables several analytes to be measured simultaneously in a small volume of biological material.21 This technique has been used for the assay of cytokines, chemokines and other inflammatory markers, and a number of multiplex panels are now commercially available. However, a multiplex bead assay (MBA) has hitherto not been applied to the measurement of multiple GF in wound fluids. Additionally, the MBA has usually utilised dual-laser FCM, which is not readily available. The present study reports, for the first time, an MBA for simultaneously measuring growth-associated mediators in wound fluids using single-laser FCM (BD FACScan®, BD Biosciences). A Tyramide Signal Amplification (TSA) procedure22 has also been utilised which, in conjunction with the MBA, significantly increases the assay sensitivity for measuring the following eight mediators in the GCF of healing periodontal wounds: Ang-1, VEGF, BMP-2, OPG, TIMP-1, bFGF, KGF, and PDGF.
For each subject, the test GCF was collected from the sites undergoing periodontal surgery (four samples) while the control GCF was collected from healthy unaffected sites (four samples), using an intracrevicular method with pre-cut filter strips as previously described.24 The volume of GCF collected was measured, the strips pooled according to the group category, transferred to plastic micro-centrifuge tubes (0.4 mL) (Alpha Laboratories) and stored at −70°C until elution using a previously described technique.25
Mixtures of the Ab-coated beads were incubated overnight with 0.1 mL of the eluted GCF samples and also with 0.1 mL of a mixture of known concentrations of the eight mediators used as standard samples for calibration (2-fold dilution series of six concentrations of the rh antigens, excluding zero, starting with 10 µg/L of TIMP-1, 4 µg/L of bFGF and 2 µg/L of each of the other mediators). After incubation, carrier beads were added to each sample and washed as above. The beads were re-suspended and incubated for 30 min in 0.5% TSA blocking buffer, centrifuged, washed and incubated for 1.5 h with a mixture of the eight (biotinylated) detection Ab (at the pre-determined optimal concentrations ranging from 45–250 µg/L). To evaluate the use of TSA in the MBA, dilutions (0, 125, 250, 500, 1000 ng/L) of a mixture of the 8 rh antigens were prepared in diluent buffer, in duplicate, and the procedure carried out with and without the addition of TSA. In the MBA without TSA, the beads were incubated for 45 min with 0.1 mL of Streptavidin-AlexaFluor488 (1:1000 dilution in diluent buffer), washed and analysed by FCM. In the MBA with the addition of TSA, the beads were resuspended in 0.1 mL of Streptavidin-HRP (1:50 dilution in the TSA blocking buffer) and incubated for 45 min, washed again as above and incubated for 10 min in 0.05 mL of TSA reagent (1:50 dilution in the amplification buffer). After washing, the beads were incubated for 45 min in 0.1 mL of Streptavidin-AlexaFluor488 (1:1000 dilution in diluent buffer). Finally, the beads were washed and resuspended in 0.1 mL of distilled water and analysed by FCM. CellQuest™ (v 3.3) and BD™ Cytometric Bead Array (CBA) software (BD Biosciences) were used for data acquisition and data analysis, respectively. The absolute amount of the analytes (ng) was determined using the standard curve obtained for each antigen, carried out at the same time. A 4-parameter logistic model of the concentration-response relationship was used for fitting the curve, as previously recommended.27 Because each GCF was eluted in a total volume of 0.1 mL, the absolute amount was calculated by dividing the outcome from the standard curve (ng/L) by 10−4 and expressed as ng/site.
Selectivity is the ability of the assay to distinguish a specific analyte in the presence of other components in the samples.28 To assess selectivity, a mixture of eight Ab-coated beads was added to 0.1 mL of a high concentration of each of the eight mediators individually (10 µg/L of TIMP-1, 8 µg/L of bFGF and 5 µg/L of each of the other mediators, in diluent buffer) and in duplicate, and the MBA carried out as above. The control was the diluent buffer alone. Selectivity of individual mediators was also evaluated by diluting each of the eight mediators at the above concentration in 10% normal human serum (NHS) (pooled from six healthy donors), which is equivalent to the final 10% concentration of the eluted GCF samples.29
Precision is the closeness of measurements when the procedure is repeated.28 To evaluate precision, the eight mediators were mixed at three concentrations (five replicates each), including the concentrations at the lower limit of quantification (LLOQ), the approximate midrange of the calibration curve and the upper limit of quantification (ULOQ). To be accepted as the lowest and highest concentrations on the standard curve, the LLOQ and ULOQ need to meet the following criteria: (1) the response (median fluorescence intensity, MFI) at the anticipated LLOQ is ≥5 times higher than the MFI of the blank (diluent buffer alone); and (2) the response should be reproducible with a mean coefficient of variation (CV) of <20%, evaluated independently using ten mixtures of the eight mediators at the anticipated LLOQ.28 The CV is defined as the ratio of the standard deviation (SD) to the mean ((SD/mean) × 100). The ULOQ is selected as the highest concentration of the calibration curve and evaluated using the same protocol as for LLOQ. The analysis of intra- and inter-assay precision was carried out three times on three separate occasions, with the acceptance criteria being a mean CV of <20% and <25% for intra-assay and inter-assay precision, respectively.27
Recovery was analysed using three solutions, each containing low, medium or high concentrations of the eight mediators (1, 3 and 10 µg/L) diluted in the 10% pooled NHS.28 Triplicate, 5 µL samples of each of the three mixtures were placed onto the paper strips separately, and the elution process carried out as above.25 The control sample was each of the three mixtures without paper strips and elution.
Assay sensitivity is defined by the limit of detection (LOD) (lowest concentration of analyte for which the response (MFI) can be reliably distinguished from the background noise),28 in which ten blank samples (diluent buffer alone) were incubated with a mixture of the eight Ab-coated beads and the MBA carried out as above. Standard curves of the eight individual mediators were generated using increasing concentrations of each mediator, as noted above. The mean and SD of the MFI of the blank for each bead type were calculated and 3 × SD was added to the mean. The resulting MFI with 3 × SD was then interpolated with the standard curve and expressed as ng/L, as previously described.27
Fig. 1 Effect of Tyramide Signal Amplification (TSA) on assay sensitivity The curves show the MFI of the MBA carried out with (••) and without (∘∘) the TSA, using increasing concentrations of the eight mediators. |
Selectivity (%) | |||||||||
---|---|---|---|---|---|---|---|---|---|
Target antigens | Concentration (ng/L) | Corresponding Ab | |||||||
Ang-1 | VEGF | BMP-2 | OPG | TIMP-1 | bFGF | KGF | PDGF | ||
a Selectivity of 100% is defined as the median fluorescence intensity (MFI) of each target antigen with its own specific corresponding Ab. | |||||||||
Ang-1 | 5000 | 100.0 | 0.3 | 0.1 | 0.0 | 0.2 | 0.2 | 0.3 | 0.0 |
VEGF | 5000 | 1.3 | 100.0 | 0.4 | 0.2 | 1.0 | 1.0 | 0.6 | 0.2 |
BMP-2 | 5000 | 1.2 | 1.8 | 100.0 | 0.3 | 1.0 | 0.8 | 0.8 | 0.2 |
OPG | 5000 | 1.7 | 1.8 | 0.5 | 100.0 | 1.4 | 1.0 | 0.8 | 0.2 |
TIMP-1 | 10000 | 0.6 | 0.3 | 0.2 | 0.1 | 100.0 | 0.4 | 0.9 | 0.1 |
bFGF | 8000 | 0.8 | 0.8 | 0.1 | 0.1 | 0.6 | 100.0 | 0.6 | 0.1 |
KGF | 5000 | 1.7 | 0.9 | 0.3 | 0.3 | 1.0 | 0.2 | 100.0 | 0.2 |
PDGF-AB | 5000 | 0.8 | 0.5 | 0.1 | 0.1 | 0.7 | 0.5 | 0.7 | 100.0 |
Analysis of the intra-assay precision showed that the mean CV of each of the mediators was <20%, thus fulfilling the precision acceptance criteria (Table 2). However, while the mean CV inter-assay of TIMP-1, bFGF and PDGF also fulfilled this criteria, Ang-1, VEGF, BMP-2, OPG and KGF were slightly elevated, but never greater than 25% (Table 2, highlighted in bold).
Sample | Ang-1 | VEGF | BMP-2 | OPG | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Concentration | CV | Concentration | CV | Concentration | CV | Concentration | CV | |||||
ng/L | Intra- | Inter- | ng/L | Intra- | Inter- | ng/L | Intra- | Inter- | ng/L | Intra- | Inter- | |
a The results show the coefficient of variation (CV) of the intra- and inter-assay for each validation sample and the mean CV of the three concentrations. CV is defined as the ratio of the standard deviation (SD) to the mean (SD/mean) × 100). | ||||||||||||
LLOQ | 125 | 9.3 | 16.8 | 200 | 9.8 | 20.1 | 62 | 19.3 | 25.5 | 62 | 15.2 | 21.7 |
Midrange | 500 | 12.2 | 19.5 | 500 | 14.7 | 22.0 | 500 | 13.5 | 26.3 | 500 | 12.5 | 25.2 |
ULOQ | 2000 | 15.4 | 27.3 | 2000 | 12.9 | 27.8 | 2000 | 14.0 | 22.2 | 2000 | 10.6 | 24.6 |
MEAN | 12.3 | 21.2 | 12.4 | 23.3 | 15.6 | 24.7 | 12.8 | 23.8 |
Sample | TIMP-1 | bFGF | KGF | PDGF | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Concentration | CV | Concentration | CV | Concentration | CV | Concentration | CV | |||||
ng/L | Intra- | Inter- | ng/L | Intra- | Inter- | ng/L | Intra- | Inter- | ng/L | Intra- | Inter- | |
LLOQ | 1250 | 7.1 | 10.0 | 400 | 8.1 | 10.5 | 200 | 11.4 | 21.3 | 100 | 9.7 | 24.0 |
Midrange | 2500 | 14.2 | 17.2 | 1000 | 13.5 | 21.8 | 625 | 18.6 | 24.2 | 500 | 12.4 | 13.4 |
ULOQ | 10000 | 14.4 | 28.1 | 4000 | 14.8 | 27.5 | 2000 | 14.0 | 20.3 | 2000 | 15.3 | 19.3 |
MEAN | 11.9 | 18.4 | 12.1 | 19.9 | 14.7 | 21.9 | 13.6 | 18.9 |
Following the elution process, at least 83.4% was recovered when using the lowest concentration (1 µg/L) and >93.5% recovered when using the highest concentration (10 µg/L). Overall, the average analyte recovery was approximately 97.2% (ranging from 78.3–116.5%), with no indication that any particular mediator was differentially affected (Table 3).
Concentration (µg/L) | Recovery (%)a | |||||||
---|---|---|---|---|---|---|---|---|
Ang-1 | VEGF | BMP-2 | OPG | TIMP-1 | bFGF | KGF | PDGF | |
a % recovery is the MFI of the sample after elution compared with that of the mediator analysed directly without elution, defined as 100%. | ||||||||
1 | 83.4 | 101.3 | 98.9 | 84.0 | 99.8 | 111.0 | 88.8 | 91.0 |
3 | 91.6 | 107.8 | 93.8 | 78.3 | 102.1 | 111.5 | 79.8 | 85.0 |
10 | 116.5 | 99.9 | 97.7 | 94.0 | 103.5 | 113.9 | 93.5 | 102.9 |
MEAN | 97.2 | 103.0 | 96.8 | 85.5 | 101.8 | 112.1 | 87.9 | 93.0 |
Measurement of the assay sensitivity showed that most of the mediators could be detected at relatively low concentrations: Ang-1, 12.8 ng/L; VEGF, 31.5 ng/L; BMP-2, 20.1 ng/L; OPG, 17.9 ng/L; bFGF, 69.1 ng/L; KGF, 6.8 ng/L, PDGF, 34.2 ng/L, although at least 214.9 ng/L of TIMP-1 was the minimum required for detection by the MBA.
Samples | Total amount of mediators in GCF (ng/site) (× 10−3) | |||||||
---|---|---|---|---|---|---|---|---|
Ang-1 | VEGF | BMP-2 | OPG | TIMP-1 | bFGF | KGF | PDGF | |
a The results show the median amounts (ng) of the analytes obtained from 15 subjects at the test and control sites with the 25th and 75th percentiles [in brackets]. The differences in amounts of the mediators between the test and control groups in each subject were averaged. b Significant differences between the test and control sites (p < 0.05). | ||||||||
Test | 338.9 [198.5, 479.0] | 344.4 [266.7, 422.0] | 25.8 [13.9, 37.8] | 33.9 [13.5, 54.4] | 2243.1 [830.4, 3655.8] | 226.9 [126.4, 327.5] | 84.2 [54.3, 114.5] | 127.2 [59.8, 194.6] |
Control | 130.8 [63.0, 198.6] | 140.5 [98.4, 182.6.0] | 20.8 [17.0, 24.6] | 12.2 [7.8, 16.6] | 162.8 [10.7, 314.8] | 164.5 [113.5, 215.1] | 12.8 [8.6, 17.0] | 57.3 [25.0, 89.7] |
Average differences | 208.0 [91.1, 324.7] | 203.8 [130.6, 277.0] | 5.0 [-5.4, 15.8] | 21.7 [2.7, 40.7] | 2080.3 [848.0, 3312.7] | 62.4 [-17.1, 141.9] | 71.4 [40.9, 101.9] | 69.9 [-10.5, 150.3] |
The TSA procedure, which was found to be rapid, simple and reproducible, markedly improved the sensitivity of the MBA. One concern of using the TSA is that it may generate non-specific signals due to diffusion of the radicalised biotin-tyramine that can react non-specifically and thus increase the assay background.22 However, in experiments not shown here, the use of an optimal concentration of the TSA was able to eliminate this difficulty. In addition, the TSA procedure also allowed low concentrations of the detection Ab to be used, compared with the ELISA.
Selectivity of the MBA for ‘specific’ target Ag is essential because biological samples are usually obtained without extraction, precipitation or further purification and are thus likely to contain multiple components which might interfere. Such endogenous analytes could complicate assay development and validation because calibration curves are generally obtained using standard solutions which do not contain other ‘contaminating’ and possibly cross-reacting factors.27 The present study showed, however, that even in the presence of relatively large amounts of endogenous Ang-1 and TIMP-1 in NHS samples, the MBA was highly selective, with only a maximum of 1.8% cross-reaction between reagents.
The FDA draft guidance acceptance criteria for assay precision is set at a CV of 20%.28 However, while this may be appropriate for validation of chromatographic assays27 which depend on the chemical and/or physicochemical properties of the molecules, it may not be fully suitable for validation of immunoassays27 which rely on the binding between the analyte (Ag) and Ab and endpoint detection system.31 It has therefore been recommended that a CV of ≤ 25% for assay precision of immunological methods is acceptable,27 and in the present study this has been shown to be met for both intra- and inter-assays.
Measurement of LOD showed that most of the mediators, except for TIMP-1, could be detected by the MBA at relatively low concentrations. Analysis of the GCF samples showed that the MBA sensitivity was sufficient to measure most of the mediators in the GCF except for KGF and PDGF, which could be detected in only half of the test samples. It is not yet known why these two mediators were not detected by the MBA in all samples, since other studies have shown them to be present in the GCF of gingival overgrowth patients32,33 and following periodontal surgery,34,35 respectively, as measured by conventional ELISA assay. The reason for this discrepancy may be due to the actual differences between the different GCF samples (e.g., timing of sample collection in relation to different stages of the wound healing process and hence different levels of growth mediators in the wound fluid). It is notable that the amounts of the angiogenic factors Ang-1, VEGF, bFGF and especially TIMP-1 were found here to be present at high levels in the test GCF, consistent with previous reports of high levels of VEGF in skin wound fluid at 7 days and of bFGF and PDGF at earlier time periods.36–38 However, such analyses require relatively large amounts of the wound fluid, in contrast to the very small volumes of GCF used here for measuring a profile of the eight growth mediators simultaneously.
In conclusion, wound healing is a complex process involving several types of cell and a number of growth mediators. The present study has developed and validated an MBA for simultaneously measuring, for the first time, the profile of mediators in a very small volume of wound fluid using single laser FCM. The TSA procedure was found to substantially increase the assay sensitivity and demonstrated that the amounts of all of the mediators were elevated in periodontal wound fluids.
Footnote |
† Electronic supplementary information (ESI) available: Growth mediators and antibodies, and schematic illustration of the multiplex bead assay. See DOI: 10.1039/b911863b |
This journal is © The Royal Society of Chemistry 2010 |