Previous Paper: Microfluidic Proton Sequencing

Herein is a derivation of the the recently disclosed Microfluidic Protein Sequencing. In essence it is the antibodies (or derivatives thereof) rather than the amino acid sequencing products which are affixed to the support.

Figure 1 describes the invention for the detection of only 3 amino acids. There are three important sections, which are preferably designed on a biochip.

The first is the Reaction Vessel with the protein to be sequenced affixed therein, and wherein the sequential terminal amino acid cleavages take place. Preferably the cleaving reagent is such that the cleaved amino acids are labeled, preferably fluorescently; although other detection schemes, including electronic, could be envisioned.

The next section of interest is a Valve which directs each successive amino acid derivative product to a unique and specific detection matrix. An alternative to one value, would be one value for each reaction cycle product, perhaps in a radial pattern.

These 3 detection matrixes are identical, and are quite novel in that they comprise arrays of distinct antibodies, wherein each distinct antibody is directed against a specific amino acid derivative. A novel alternative to a two dimensional array would be a linear array, wherein the distinct antibodies were sequencially placed within a single channel ("column").

Figure 2 shows what the matrixes might look like after the processes (including final wash). As shown, optimally within each matrix, only one position and thus kind of antibody has captured the amino acid derivative, displaying a detectable signal. Simply reading each successive matrix provides the target protein's sequence.

Thank you to the many experts that have written with interest, suggestions, and questions regarding the first article. Indeed, it was a question from one of these interested scientists wherewith this sequencing derivation was conceived.

1) Emerging trends in the synthesis and improvement of hapten-specific recombinant antibodies. Yau, et al Biotechnol Adv. 2003 Oct;21(7):599-637 (also see MedLine link of "Related Articles")

2) Heavy and light chain variable region sequences and antibody properties of anti-phosphotyrosine antibodies reveal both common and distinct features. Ruff-Jamison, et al J Biol Chem. 1991 Apr 5;266(10):6607-13

3) Molecular modeling and site-directed mutagenesis of an anti-phosphotyrosine antibody predicts the combining site and allows the detection of higher affinity interactions. Ruff-Jamison S et al Protein Eng. 1993 Aug;6(6):661-8

4) Requirement for both H and L chain V regions, VH and VK joining amino acids, and the unique H chain D region for the high affinity binding of an anti-phosphotyrosine antibody. Ruff-Jamison, et al J Immunol. 1993 Apr 15;150(8 Pt 1):3389-96

5) Generation of monoclonal antibodies against phosphotyrosine and their use for affinity purification of phosphotyrosine-containing proteins. Frackelton, et al Methods Enzymol. 1991;201:79-92. No abstract available.

Comments From Experts

There are many different recombinant antibody libraries available out there. By combination of a good library with the right selection strategy, you can almost isolate antibodies to all kinds of antigens. For example, there are antibodies specific to carbohydrates, which is very non-immunogenic.
---

Your protein sequencing idea is certainly interesting! My best guess is as follows; It is highly likely that one can create antibodies against Edman derivatives with medium affinity. These molecules are however so small that high affinity antibodies are very difficult to make: on the other hand, derivatisation does help quite a bit. You could see in my paper that our monoclonal against 3-methylindole had only about 11% cross-reactivity against indole so one additional methyl group really did make a difference. However, even 11% cross reactivity may be too much in your application and create false signals. So specificity could be a problem concerning for example differentiation between Ala and Gly.
---

I've only had a chance to glance at it, but that looks really cool....fast and easy as far as I can tell off the cuff. The biggest problems we've had with alternative methodologies have been in solvent choices/volubility. Your antibody detection method may be cost prohibitive, considering sequencing could require addition of all 20 antibodies to each/some wells, not to mention the antibodies would (I'm guessing) have to be 1) of pretty high quality to minimize cross-reactivity, and 2) in some way quantifiable so as to alleviate the problems caused by proline. You have a good idea, though.
---

Thank you fort the interesting link to your new concept dealing with protein sequencing. I agree with you that such a method represents a fantastic progress facilitating fast sequencing of proteins applicable to a high throughput format. Basically it should be possible to develop specific binders to any protein and small molecule using phage surface display technology as stated in the review "High-throughput applications of phage display in proteomic analyses" which will appear in one of the next issues of TARGETS. However, such a task would requires a major effort based on high throughput screening of combinatorial libraries, selection of ligand panels and validation (which represent the most important step) and would also require a major financial effort. Theoretically there are no reasons to assume that such an approach would not work.
---

I have discussed the use of antibodies with some of my colleagues within the past two weeks. It seems that some people have attempted to perform Edman's degradation on chip before, but not with the use of antibodies.
---

When we're detecting fluorescently-labeled proteins from human serum using antibody microarrays, we can get down to about 1 ng/ml. We now have amplification methods working well that get us down to 100 pg/ml or so. It also depends on the quality of the antibody.
---

We use concentrations of around 500 ug/ml for the capture antibody. Spot size in principle doesn't matter as long as your scanner resolution samples it sufficiently. (i.e. for a 100 um diameter spot, need a resolution of 10 um or so.)
---

To answer your question about generating antibodies to all 20 different amino acid residues, I think some will be easier than others. For example, Trp, and Tyr would be the easiest among all, while Gly and Ala would be the toughest. However, there would be ways to circumvent the problems. All in all, I think it is feasible but highly depends on time and resources.