New Methodologies in Cancer Biomarker Analysis

Extracellular vesicles (EVs) are small membrane-enclosed spheres secreted from all cells that can carry proteins, nucleic acids, metabolites, and other molecules. They also express proteins on their surface and can be used as translational biomarkers to aid drug development and therapeutic research.

EV-based research is promising, but studies are hampered by the sheer complexity of surface marker expression and by assay sensitivity limits. The extremely low concentration of some expressed proteins makes detection and differential analysis difficult. To solve this problem, researchers at the NIH and several other well-regarded cancer research institutes have joined forces to design a novel software analysis tool that they hope will be able to streamline complex data and create a more meaningful analysis of proteins pertinent to cancer and other pathologies^[1].

Commercially available kits for EV phenotyping do exist. Single-molecule detection kits display exquisite sensitivity and can reliably assess whether a given marker is present. The shortcoming of these kits is that they are limited to one color reporting, and therefore assume prior knowledge of which markers are being identified in a sample. Multiplex kits also exist and make use of an array of fluorophore-antibody coated beads that can detect several markers at the same time. Part of the weakness in this technique is that low-level expression can be missed in the enormous quantity of data that is generated, again making it exceedingly difficult to carry out differential analysis.

The researchers were intent on designing a methodology that could combine the strengths of both kits and provide a meaningful analysis of a high volume of data with single-molecule sensitivity. They designed MPA_PASS (multiplex analysis post-acquisition analysis Software) to be able to “stitch together” both single molecule data and high-volume multiplex data.

Cerebrospinal fluid (CSF), serum, and plasma used in this study were collected from consenting adult donors at the National Institute of Neurologic Disorders and Stroke. Plasma samples were centrifuged to remove residual platelets. To isolate EV from the human biological samples, the samples were incubated with a multiplex bead mixture consisting of bead-bound antibodies for common cell surface markers. This was followed by size exclusion to enrich for the EV fraction, and resin capture to remove impurities. 2 mL of CSF sample was concentrated to about 500 mL using a 100kDA Nanosep^® filter from Pall.

Human prostate cancer cell lines PC3 and PC3pip were also used as a source for EVs. After culturing the cells for 24-48 hours in a culture medium containing EV-depleted FBS, 50 mL aliquots of cell culture supernatants were collected and concentrated down to 5 mL using 60 mL volume 100kDA Jumbosep^TM filters. Analogous EV samples were collected from normal human neural, kidney, and colon cell lines.

Sample concentration and filtration are regularly used in sample preparation of biological fluids since it can be critical to reduce sample complexity and enrichment for target molecules.

Pall’s centrifugal filter devices are designed to prevent retentate leakage and filtrate contamination. Filtration is rapid, achieving 50X concentration with greater than 90% sample recovery. Pall centrifugal filters are available for sample volumes ranging from < 50 µL to 60 mL, in a variety of MWCO (Molecular Weight Cut Off) and membrane compositions.

Following purification of the various EV samples, the research team incubated the samples with EV-capture bead mixtures in preparation for the multiplex assays. The team used a commercially available EV kit (MACSPlex) for this part of the study. EV-capture bead mixtures were placed in a filter plate and then incubated in the presence of a chosen set of antibodies. Samples were then analyzed using flow cytometry.

The newly designed MPA_PASS software was used to conduct in-depth deconvolution of the flow cytometry dataset. One of the strengths of the method the scientists used is that multiplex analysis can capture a diverse group of EVs in terms of size and composition. Because each EV displays multiple different proteins on its surface, the signal from each bead is a function of the relative expression of a particular protein, the number of different proteins being expressed within that sample of EVs, and what other capture beads are present in the multiplex set. As mentioned earlier, this generates a huge amount of data. Both signal intensity and variation give useful information about protein abundance and protein diversity within a given sample.

The researchers used a variety of measures, including titration controls, to normalize signal intensity and ensure the reliability of their data. They found that they could obtain a high enough signal intensity from their chosen antibody markers to be able to make distinctions among expressed marker intensity in EV samples from two prostate cancer cell lines. They were able to reliably distinguish between specific and non-specific antibody binding and demonstrated that non-specific binding can be capture-bead dependent.

The scientists were able to identify unique marker combinations from different EV tissue derivations and cell lines. They were also able to use data clustering to distinguish biofluid and purification EV samples. Overall, the study results demonstrated the utility of the stitched multiplex analysis method for sample identification and for distinguishing isolation methods. The researchers believe their new software will facilitate data sharing and large data sample curation. Going forward, it is reasonable to expect MPA_PASS to be extremely useful for identifying clinical differences in samples, examining pathology-specific EV subsets, and streamlining novel EV biomarker identification.

To learn more about the centrifugal filter devices used in this study, please visit our website.