ELISA quantification was performed on specific ECM biomarkers (PIIINP, MMP-2 and -9) and GDF-15. The general ELISA methods employed by our group have been previously described in the literature.265,266 GDF-15, MMP-2 and -9 utilised the DuoSet ® ELISA assays (R&D Systems Inc, Minneapolis, Minnesota, USA) and PIIINP utilised Cusabio (Wuhan, Hubei, China). The GDF-15 assay protocol had previously been used in a published cohort study.112 All samples analysed were prepared plasma as previously outlined. The samples were all analysed on the first freeze-thaw cycle. All analyses were performed in duplicate. Analyses
152 were performed for all four assays on 96-well plates with a 6- or 7-point standard curve on each plate.
The process and principles of quantitative sandwich ELISA are outlined in Figure 3.10 Assay- specific protocols were followed for each analysis. For the R&D protocols capture antibodies were diluted to working concentrations in Phosphate-Buffered Saline (PBS). All 96 wells of the microplate were coated overnight with 100μl of capture antibody. Following incubation three automated wash cycles were performed by the Thermo Scientific Wellwash 4 Mk 2 microplate washer (Thermo Electrical Corporation, West Chester, Pennsylvannia, USA). Each well was then blocked with 300μl of Blocking Buffer (1% Bovine Serum Albumin (BSA) in PBS) and then incubated at room temperature for one hour. A further three automated wash cycles were then performed. The PIIINP Cuasbio assay provided a pre-coated and blocked microplate as part of the kit provided by the manufacturer.
Recombinant protein in known concentrations were used for each assay to prepare a six or seven point (dependent on protocol) standard curve, using a two-fold serial dilution. A blank dilution buffer sample was used for background correction. Serial dilutions were performed using several participants’ plasma samples to identify the optimum dilution for MMP-2 and PIIINP assays. For the MMP-9 and GDF-15 assays, dilutions were based upon previous reported studies. Based on these experiments, the following dilutions were used for the whole cohort: MMP-2 1:50, MMP-9 1:100 and GDF-15 1:10. PIIINP had no dilution
153 performed. Sample analysis was repeated if the results fell outside the serial dilution curve in duplicate.
154
(1) Capture antibody is bound to the 96-well microplate
(2) Sample is added, and the corresponding antigen binds to the capture antibody
(3) Detecting antibody is added, and binds a different epitope on the antigen
(4) Enzyme-linked secondary antibody is added, and binds
to the detecting antibody
(5) Substrate is added, and is converted to a detectable signal by the enzyme
Figure 3.10 The Process of a Sandwich ELISA, Microplates are coated with a specific antibody or protein (1) to bind the corresponding protein present in the samples (2). Unbound substances are washed away, and a monoclonal antibody specific to the protein being measured is added to the plate (3). A second wash removes unbound antibody and an enzyme linked antibody specific for the primary antibody is added (4), followed by an amplifier solution which develops colour in proportion to the amount of measured substance present (5). The reaction is stopped by addition of a stop solution (usually an acid); plates are then read using a microplate reader.
155 Following blocking and subsequent washing, 100μl of sample or standard were pipetted to each respective well and incubated at room temperature for 2 hours (PIIINP microplates incubated at 37oC). A further three automated wash cycles was then performed (liquid removed but wash was not performed for PIIINP analysis protocol). Detection antibody (biotinylated) was diluted in reagent diluent according to manufacturer’s instructions (MMP-9 analysis diluent included Normal Goat Serum) and then had 100μl added to each well. Incubation was for two hours at room temperature for all assays except for PIIINP which was for an hour at 37oC. A working dilution of Streptavidin conjugated to horseradish peroxidase (HRP) was prepared and 100μl was pipetted into each respective well after three automated washes. Incubation was then performed for 20 minutes at room temperature (PIIINP for one hour at 37oC) in the dark. A further three automated wash cycles were then performed. 100μl of substrate solution (1:1 mixture H2O2 and tetramethylbenzidine) were
added to each well and incubated for 20 minutes at room temperature in the dark. The PIIINP assay protocol requires the addition of 90μl of tetramethylbenzidine substrate to each well and incubated for 30 minutes at 37oC, in the dark. Finally 50μl of stop solution (2N H2SO4) were added to each well, ensuring thorough mixing. Time between adding substrate
and stop solution was dependent on observed intensity of the substrate colour. Microplate reading was performed by PHERAstar FS (BMG Labtech, Ortenberg, Germany). The microplate reader wavelength was set to 450nm. Background noise unrelated to the assay (such as optical imperfections in the plate) was corrected for by subtracting a reference measurement that was obtained at 570nm.
156 Quantification of the assay target concentration was calculated by plotting obtained optical density values to a standard curve. This standard curve was generated with the serially diluted assay standard following subtraction of the background noise from each well. Optical density values (see an example in Table 3.5) were fitted using a four-parametric logistic regression (4PL) curve-fit.267
Table 3.5 Serial Dilutions Optical Densities for GDF-15 Microplate 2
.
The 4PL model is the most suitable for fitting a standard curve for many complex biological systems and is the gold standard in ELISA analysis.267 The 4PL model is summarised by the following equation, where four parameters are calculated to fit a curve to a set of standards:
The known standard concentrations are logarithmically converted to produce a semi- logarithmic scale for fitting the 4PL model. The best-fit is calculated based upon the optical
Standard Curve GDF-15 (Microplate 2) Serial Dilution (pg/ml) Optical Densities
500 1.976 1.936 250 1.188 1.174 125 0.585 0.459 62.5 0.216 0.196 31.3 0.086 0.075 15.6 0.041 0.043 7.81 0.023 0.026
157 densities of the set of standards (for example in Table 3.5) producing an S-shaped curve. In the equation y represents the response value (i.e. optical density of the standard) and the x
represents the dose value (i.e. the known concentration of the standard), whilst a and d
represent the maximum and minimum asymptotes (horizontals) of the curve respectively (and they can be interchanged). The a and d values are the same units as the y value. The c
parameter represents the point of inflection (the midway point between a and d). The b
parameter is the hill’s slope of the curve (relates to steepness of curve at point c).
GraphPad ® PRISM 2007 version 6.0 (San Diego, California, USA) was used to calculate a 4PL curve to fit the set of standards and replicated on an online ELISA analysis tool (www.elisaanalysis.com).268-270 A 4PL curve fit was calculated for each individual microplate. The strength of each fit was tested with the strength of the correlation and was accepted if above a coefficient of >0.97. Figure 3.11 demonstrates an example of best-fit for GDF-15 microplate 2 from the COVERT-HF study (using the values generated in Table 3.5). Samples of unknown concentrations are expressed relative to the calculated standard curve. The 4 parameters calculated for this particular example is demonstrated. Reversing the 4PL curve when all 4 parameters and optical density is known allows interpolation of the unknown target concentrations. This was performed within Microsoft Excel 2010 following the 4PL calculations in GrapPad PRISM. Measured mean concentrations of duplicates underwent multiplication by the respective dilution factor (see above).
158 0 1 2 3 0 .0 0 .5 1 .0 1 .5 2 .0 2 .5 G D F - 1 5 S t a n d a r d C u r v e ( P la t e 2 ) L o g ( G D F - 1 5 ) p g / m l O p ti c a l D e n s it y
Figure 3.11 The 4PL Best-Fit Standard Curve for GDF-15 Plate 2. Values a,b,c and d are calculated for the plotting of the known set of standard concentration and best fitting a 4PL standard curve to the values. The standard curve allowed concentrations in samples to be calculated.
The precision and reliability of the ELISA assay was estimated by calculating the inter- and intra-assay coefficient of variability (CV). The inter-assay CV is an important measure of plate-to-plate consistency. This is particularly important in large cohort studies like COVERT- HF as five microplates were required to assess one biomarker assay in all samples. The highest and lowest duplicate values are utilised to calculate high and low microplate means. These high and low microplate means were combined to calculate an average mean and standard deviation of the means, which in turn allowed for the calculation of the percentage CV of means for highest and lowest values. An average of these measures allowed
159 calculation of the inter-assay CV. The intra-assay CV assesses the individual variation between duplicates upon each microplate. The results are presented as a mean of all individual intra-assay CV’s on one microplate. The mean concentration and the standard deviation of the duplicates were calculated. Subsequently the standard deviation was divided by the duplicate mean and the individual intra-assay CV percentage was calculated and the average for the plate calculated.