In this week’s Antibody Applications, we’re delighted to have our guest writer, Boby Makabi-Panzu, PhD, to tell us about Proximity Extension Assay (PEA).
Boby is a Senior Research Scientist at the Center for Molecular Medicine and Immunology, with years of experience in Virology and Immunology.
Take it away Boby!
Biomarkers are protein indicators measured to examine normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention in medical fields. In precision medicine, biomarkers are used to establish signatures from which therapies can be customized to suit individual characteristics of each patient. However, the development of such signatures requires the identification of large number of biomarkers simultaneously. Over the past decade, technological advances have enabled such “omics” approach for biomarker measurements. Particularly, a novel proteomic assay known as the Proximity Extension Assay (PEA), was shown in a recent AAAS supplement to be suitable for use in clinical settings without losing its performance characteristics and multiplexing power for establishing biomarker signatures. As a result, the development of PEA has pathophysiological importance in the context of precision medicine. In this article, I will summarize the study and the proposed PEA protocol for your reference.
What Is Proximity Extension Assay?
PEA is, by design, a throughput, good precision, and high sensitivity method merging an immunoassay with the powers of a conventional and microfluidic quantitative PCR. The assay is multiplexable and can detect up to 92 protein biomarkers at once using only a volume of 1 µl of an array of biological fluids including cerebrospinal fluid, dried blood spots, and tumor biopsies. The authors of this study demonstrated that the standard PEA format made of 92-plex panels, can be customized to a format of 19-plex panels more suitable for clinical setting uses, and still keep its performance characteristics.
How Was the Study Conducted?
To conduct their experiments, the authors proceeded, through an ongoing project, as follows: Firstly, they screened more than 400 proteins as potential markers to identify different diseases or disease stages using standard PEA standard format (92-plex panel). Secondly, among the candidate biomarkers identified with the standard format, they selected 19 to build a customized 19-plex panel format. Finally, to see whether the technology fit with the intended clinical utility in early-phase clinical trials, they compared results obtained using both assay formats (92-plex vs 19-plex).
The Standard PEA Protocol
A standard protocol was used for the occasion to assay biomarkers with both PEA formats:
- 1 µl of sample in 0.2 ml PCR tube was preincubated with 1 µl of a mixture of blocking antibodies at RT for 20 min to prevent unspecific binding of proximity probes.
- The mixture (2 µl), containing both proximity probe variants, was incubated with the previous solution (2 µl) at 37oC for 1-2 hours to permit antigen-antibody interactions.
- Samples were placed into a thermal cycler and 76 µl dilution mix was added to every 4 µl of sample and incubated for 5 min at 37oC. An addition of 20 µl of extension mix (adding up to a total of 100 µl for every sample) containing DNA polymerase at 37oC for 20 min extended oligomer to full length, creating templates for following qPCR amplification. DNA polymerase was inactivated thereafter at 85oC for 10 min.
- The extended DNA samples were quantified by qPCR. Volumes of 4 µl were extracted from each of the previous extension reactions and put into individual wells of 384 microtiter plate, 6 µl of qPCR mix was added to each well, amounting to 10 µl of qPCR reaction, leaving 96 µl of extension products from step 3. These were saved (-20o°C), to serve as backup. Concentrations of detected protein biomarkers in samples were expressed in pg/ml using a standard curve. Relative quantification, NPX (normalized protein expression) values were used to calculate correlation coefficients between analyses.
92-plex Vs 19-plex PEA: The Verdict
Using this protocol, the authors evaluated PEA performance characteristics and scalability into 19-plex panels to identify, in 80 samples, protein biomarkers including FABP4, CDH3, CPE, Dkk-4, EN-RAGE, FGF-23, FR-alpha, Gal-1, IL-10, IL-17C, IL-8, MMP-7, NTRK3, PARP-1, PRSS8, PVRL4, and SOD2 initially pre-screened in standard 92-plex panels and found these results:
- Data obtained with 92-plex screening panels correlated very well with those obtained with customized 19-plex panels. An average coefficient of correlation (R2) of 0.90 was found. The lowest correlation (R2 = 0.73) was observed with the SOD2 assay, as the samples tested were within a narrow range (<2 NPX).
- Like screening panels, custom panels displayed very good sensitivity. LOD and LLQQ of 30 fg/mL were obtained for IL-8 while assays for IL-10, FR-alpha, PARP-1, and PVRL4 were the most sensitive with LLOQ ≤ 1 pg/mL. The assay median LOD and LLOQ were 7.6 pg/mL and 11.4 pg/mL, respectively.
- A dynamic range with a median log10 range of 3.7 (2.1–5.4) was found. IL-8 had the widest dynamic range of log10 range of 5.4. Altogether, assays spanned 6.4 logs in concentration from the lowest to the highest control sample.
- The interference of the assay by plasma components such as bilirubin, lipids, and hemolysate evaluated using spiked samples was minimal and data agreed with those from screening panels.
- Cross-reactivity was not observed while testing highly related proteins, except for FR-alpha versus its highly homologous FR-beta (77% identity and 87% coverage) with a nonsignificant level (0.1%) in plasma
- Great assay linearity was found in general both at high and low ranges, with somewhat poorer accuracy observed for the MK assay, where the highest sample concentration was close to the estimated ULOQ according to minimum required dilution (MRD) established under true matrix conditions.
- Intra-assay CV varied from 4% to 8% while inter-assay CV varied between 1% and 20% for the different assays. Average intra- and inter-assay CVs were 5.8% and 13.5%, respectively, which are in parity with many commercial single ELISAs.
- Biomarker and reagent assay were stable. Samples and kits used were in general found to be insensitive (criteria: +/– 30% deviation) to both freeze-thawing and RT storage.
Based on their data demonstrating excellent correlation of values for key immunoassay parameters (sensitivity, dynamic range, specificity, linearity, precision, and stability) between the custom and standard PEA panels, the authors of this study concluded that PEA standard 92-plex panel format can be readily customized to a 19-plex panel format, which is more suitable for use in clinical settings, without losing performance.
Editor’s note: If you are studying any of the biomarkers mentioned in this article, we put together a list of antibodies with the most published figures on BenchSci for you to review.
| Target | Clonality | Host | View Published Data |
| FABP4 | Polyclonal | Rabbit | {{cta(‘2f147ea9-6065-4746-951a-86edf05ef98c’)}} |
| CDH3 | Monoclonal, 6A9 | Mouse | {{cta(‘c1695a1c-a8b1-45c8-be61-5fe937e9f545’)}} |
| CPE | Monoclonal | Mouse | {{cta(‘5c824c82-efe8-4bbb-9df6-a19d51d8855f’)}} |
| Dkk-4 | Polyclonal | Rabbit | {{cta(‘c4818963-9fa2-49d2-a35e-5a2375b3512f’)}} |
| EN-RAGE | Polyclonal | Goat | {{cta(‘e35347b1-17b1-4d93-bcca-1e036104a510’)}} |
| FGF-23 | Monoclonal | Rat | {{cta(‘b2497d8f-2431-4e89-9c6d-0448f37228e8’)}} |
| FR-alpha | Polyclonal | Goat | {{cta(‘e42cd775-e083-4743-8562-13c256922b31’)}} |
| Gal-1 | Polyclonal | Goat | {{cta(‘d8eddd29-bbc8-49e5-9cd9-9e51b15896c7’)}} |
| IL-10 | Monoclonal | Rat | {{cta(‘a0fddd8f-1183-4007-ad26-be82696175a0’)}} |
| IL-17C | Polyclonal | Goat | {{cta(‘aeeb4b39-48bb-473c-8a64-8538787eb753’)}} |
| IL-8 | Monoclonal | Mouse | {{cta(‘d8d6d613-3a63-4ecd-8762-fa41aadc714a’)}} |
| MMP-7 | Polyclonal | Rabbit | {{cta(‘e1ce2210-bdb2-463a-9a6c-66c76c96ded0’)}} |
| NTRK3 | Monoclonal | Mouse | {{cta(‘07281e26-3f77-4e65-ab0c-d1e706aae563’)}} |
| PARP-1 | Polyclonal | Rabbit | {{cta(‘f167a798-9d30-4e96-85df-df48b03d5be1’)}} |
| PRSS8 | Monoclonal | Mouse | {{cta(‘5e786f99-2980-4344-a6bb-ec48ff60217e’)}} |
| PVRL4 | Polyclonal | Goat | {{cta(‘9175a209-b5d5-41a2-b3cf-98afbc0b175e’)}} |
| SOD2 | Polyclonal | Rabbit | {{cta(‘ab32ade0-5db8-40b9-a5bd-c36efd52a90c’)}} |