A Brighter Idea for Studying Protein Expression

One of the fundamental requirements of drug discovery and development, one that is still fundamentally difficult for researchers to attain, is to secure reliable analytical means to understand how drugs cause changes in state and function in the organism. In an ideal world, researchers would be able to see reliable indicators to distinguish real responses from spurious variation and background noise.

Many studies look at indicators of response such as expression or modification of proteins involved in the signaling pathways that the drug purportedly affects. Typical changes in expression relating to a disease state are subtle. All established technologies have shortcomings that don’t sit well with the demands of sensitivity, resolution and scalability needed by drug development to test how the drug actually affects the system in reality. Traditional methods to compare protein expression are often not sensitive enough to detect subtle changes, since background noise and technical variation are too high. Gel-based technologies like Western blotting and 2-D electrophoresis can resolve whole proteins, but they are protein-hungry techniques. They require amassing tissue and destroying cells to get enough protein for a reliable measurement, thereby sacrificing per-cell resolution (where the action might be) and valuable in situ information. Although higher sensitivity is achievable by mass spectrometry, it entails even more fragmentation, and picking up post-translational modifications by MS can be very difficult. Both 2-D electrophoresis and mass spectrometry are labor intensive and are therefore not practical when analyzing protein expression in response to drugs in typical sample sizes.

Established methods to detect expression in situ (e.g. GFP expression, FRET or BRET) require the introduction of recombinant engineered proteins, thereby precluding the possibility of monitoring expression in tissue biopsies. Screening systems to test drug targets are often based on over-expressing cell lines that bear little or no relation to what happens in typical tissues.

Current methods to study protein expression are therefore at risk of being permanently labelled as insufficiently sensitive, only capable of detecting gross changes in expression or only in artificial circumstances, that never come close to revealing the finer details of cellular regulation and response to stimuli in real life situations.

Underlining this point is the fact that, despite enormous research efforts, diagnostic tests based on protein expression are conspicuous by their absence. One of the few FDA-approved diagnostic tests is HercepTest™ from DAKO for distinguishing patients that overexpress HER2 and that may respond to the breast cancer drug Herceptin® (trastuzumab). The test is subjective and the manufacturer requires the pathologist’s experience and level of judgement to score samples.

Figure 1. The semi-quantitative HercepTest (DAKO) gives four classification scores, used for diagnosing levels of EGF receptor (HER2) overexpression in breast tumor cells. These images show ideal staining; in reality many uncertain situations arise through multiple sources of variation. Taken from the HercepTest information manual.

Illuminating Insight into Protein Expression

The demand for better resolution and objectivity for quantitation from limited tissue materials boils down to obtaining better sensitivity and specificity for the target molecule. If an assay could provide sensitivity to detect an individual protein molecule, it would mean that minimum amounts of material for analysis would be required, and tests could be performed on a per-cell basis, enabling the researcher to see the actual variation in expression among cells. If the assay were highly specific, it would make the detected signal highly reliable and would reduce background to a minimum. The method would also benefit from being able to detect proteins in any kind of material – in vitro as well as in situ in both cells and tissues.

Olink Bioscience (Uppsala, Sweden) has focused its research on making measurement tools that enable routine genomic and proteomic investigations on platforms that the individual researcher is capable of managing, yet produces the quality of information that protein expression studies demand. Olink Biosciences uses, among other things, DNA amplification technology to address the need for detection reliability and sensitivity.

In 2007, Olink Bioscience released Duolink™, a kit series in which DNA amplification is coupled to an immunoassay to bring radically new detection possibilities for monitoring protein expression. Duolink is essentially an immunofluorescence assay, with boosted sensitivity from signal amplification, and improved specificity arising from target recognition by two antibodies before a signal can be generated.

Click here to view all images (a),(b),(c) & (d) from figure 2

Figure 2. The Duolink assay utilizes two primary antibodies to recognize the target antigen (A, red and green). Each secondary antibody has a short DNA strand attached to it (A, light and dark grey). When the secondary antibodies are in close proximity by binding to the same epitope or to neighboring epitopes, the DNA strands can ligate (B) and DNA amplification can subsequently take place (C). By fluorescent labeling of the amplification product (D), the obtained detection sensitivity is up to 1,000 times higher than that of traditional immunohistochemical methods. Since the signal is generated only when both epitopes are recognized, specificity is high and non-specific background is reduced to near-zero.

The Duolink kit is based on the Proximity Ligation Assay (Söderberg et al., 2006), the most radical improvement in protein assays since the invention of the sandwich immunoassay in the 1960’s. Duolink exploits the sensitivity and specific readout of DNA amplification for studying proteins, and it can be easily performed in situ using existing antibodies. The method can be used to capture a snapshot of the levels of proteins in cells and tissues, the relative activity of membrane receptor complexes at rest and under stimuli, and the degree of post-translational modifications (PTMs) such as phosphorylation of receptors in signal transduction pathways. For detecting PTMs reliably, Duolink’s dual antibody recognition property is particularly valuable in that it eliminates the background signal that arises from the typically poor specificity of antibodies raised against phosphorylated proteins.

Click here to view all images (a),(b),(c) & (d) from figure 3

Figure 3. Detection of EGF receptors in A431 cells by (A) Duolink Q (red) and (C) a conventional immunofluorescence assay (red) shows discrete signals by Duolink Q compared to diffuse signals by immunofluorescence. Controls that were run in the absence of the anti-EGFR primary antibody show (B) negligible background in the case of Duolink Q (red) and (D) high background (red) in the case of the conventional method. Both conventional immunofluorescence and Duolink Q employed the same mouse anti-EGFR primary antibody. Immunofluorescence images employed a Texas Red labeled anti-mouse secondary antibody.

Comparisons of Duolink Q (a new kit that requires only one primary antibody) with standard immunofluorescence assays show some startling improvements. Not only is signal:noise ratio a thousand-fold higher (Figures 3A vs. 3C), the signals are present as discrete dots (Figure 3A), one per detected protein, that can be easily counted by eye or software to quantitate the level of protein in the sample.

Furthermore, the Duolink assay workflow is scalable to accommodate single cells to tissue arrays. Finally, a method is now available to meet the demands of studying protein expression with objective quantitation, high specificity, per-cell resolution and scalable to analyze many tissue samples.

Click here to view all images (a),(b),(c) & (d) from figure 4

Figure 4. Duolink detection (red) of phosphorylated epidermal growth factor receptors (EGFr) in head and neck tumors, squamous cell carcinoma (HNSCC) (A) and lip (B), and in normal skin (C) and normal placental villae (D). Quantitation of signals shows 24 signals/cell in HNSCC, 5 signals/cell in lip, 7 signals/cell in skin and 37 signals/cell in placental villae.

Duolink opens a new world of possibilities for developing quantitative assays to assess protein expression and modifications in normal and diseased tissues, as well as in response to drugs. The demand for specific assay development by pharma companies has incited Olink Biosciences to develop contract research services, which deliver optimized Duolink assays to monitor specific drug targets. Thereafter the pharma organization can bring the assay in-house based on commercially available Duolink kits.

References

HercepTest Interpretation Manual 28630. http://pri.dako.com/28630_herceptest_interpretation_manual.pdf

Direct observation of individual endogenous protein complexes in situ by proximity ligation. O. Söderberg, M. Gullberg, M. Jarvius, K. Ridderstråle, K.-J. Leuchowius, J. Jarvius, K. Wester, P. Hydbring, F. Bahram, L.-G. Larsson & U. Landegren. Nature Methods 3: 995-1000 (2006)