Nature Methods
2, 11 - 12 (2005)
doi:10.1038/nmeth0105-11
Tadpoles by the tailGarry P NolanGarry P. Nolan is in the Department of Microbiology and Immunology, Stanford University, Stanford, California 94117, USA. gnolan@stanford.edu Intein-dependent, ultrasensitive, amplified detection of... anythingMany significant advances in biological inquiry have been dependent on the measurement of the absolute number of some molecular object. Yet deeper understanding of biological processes is often hindered by the technical limitations on how such measurements are performed. Most detection systems have at their core a specialized technique particular to the target molecule(s) under examination. Today's savviest researchers must therefore have at their tool belt a bounty of techniques ready for application. However, most of us feel some frustration when we must master yet another unfamiliar assay (often driven by a need to address some surly critique of our recently rejected paper!). How might things change if we could measure a wide range of molecular entities with ease and accuracy using a modular detection reagent that had a simple, multiplexable readout? And what if this did not require a guru's depth of knowledge, a need that, paradoxically, often causes researchers to abandon techniques either as too complicated or because they required acquisition of dyspeptic, expensive machinery? Further, would things be better if an often ignored, but crippling, variable inherent to most standard detection reagents could be dismissed as irrelevant?
Burbulis, Brent and colleagues1 take a crucial step to address all these points in an important report that promises a wide range of applications. At the heart of the approach is the supple use of intein biochemistry to chemically create a single (this is critical!) molecular hook on an expressed protein−based recognition module. This hook (described below) becomes the means to distinctively match the recognition domain to an amplifiable DNA tag for later detection. This molecular matchmaking is accomplished through the use of inteins, a curious molecular DNA parasite that insert into the coding region of genes and are expressed in frame within the proteins that host them2,
3. Soon after expression of the host protein, an intein chemically autocatalyzes its own excision-based removal and the host protein is left unscathed. Brent and colleagues applied these molecular gymnastics to create a novel detection reagent, which they quixotically term 'tadpoles' for their apparent shape and the series of metamorphosis steps they must endure until their final maturation and application.
First, the authors expressed a fusion protein made of a recognition module coupled to an intein. They purified this expressed protein and then adapted the chemistry of inteins4 to initiate excision, but in a manner that immediately couples a single DNA 'bar code' to each of the protein recognition modules from which the intein was removing itself. The DNA bar code contains a T7 RNA polymerase transcription binding and start site (for RNA-mediated detection) as well as a region for PCR amplification−mediated detection. Thus, in its final form the recognition-detection reagent is a recognition (affinity) domain, such as an antibody or a peptide aptamer, coupled directly to an amplifiable tag. These tadpole reagents were then used in a variety of proof-of-principle tests to determine their sensitivity and specificity using standard quantitative PCR as the final readout.
Astonishingly, the tadpoles were capable of detecting targets across a concentration range of up to 11 orders of magnitude. Under optimal circumstances they performed nearly 109-fold more sensitively than standard ELISA-based detection. A standard outline of a detection experiment is shown in Figure 1. Brent and co-workers applied the system first to the detection of a small organic molecule, biotin, using a streptavidin moiety fused to the intein module. This might be considered cheating, given that streptavidin has a ravishing affection for biotin—but the authors did not stop there. In a more rigorous test, they used standard-issue aptamers (short peptide sequences with modest 1- M affinity for their targets) to two important cellular proteins, E2F1 and Cdk2. These aptamers were fused to the inteins, used to create tadpole reagents and then tested. Remarkably, they performed comparably to ELISA-detected antibodies that had much higher affinities. Thus, reagents with modest affinities, such as aptamers that are quick and cheap to make, were made as useful as high-affinity (read expensive and hard to make) antibody reagents. The authors then went further with a more real-world, sensitive test of an important bacterial pathogen in whole blood sera. I can already see the reagent vendors scrambling for their phones.
 | | |
Why is this approach any different from other techniques already in play? For me, much of the importance lies in the singular addition of the DNA tag to the protein recognition module. This is aimed at roundly defeating that crippling variable I mentioned in the opening paragraph that is "as an unknowing plague" upon antibodies and other detection systems. If one bothers to read most coupling protocols one quickly learns that all available chemical reactions couple the fluorophores to the target antibody in 'an approximate range' of stoichiometries. Because antibodies are an extremely unpredictable set of proteins, and traditional protocols couple fluorophores or DNA tags 'randomly' all over the antibody, what you essentially end up with is a witch's brew of detection reagents under the guise of a homogenous agent. The system of Brent and colleagues involves a 'truth in advertising' that limits such variation, as by design it allows only a single detector per antibody or aptamer. Through an extensive series of well-designed statistical tests, the authors demonstrate that through this modest advance an extremely significant source of variation is decisively dismissed. As a consequence it allows minor variations in the concentrations of target molecules to be sensitively and quantitatively detected.
This subtle approach is cobbled together from seemingly innocuous bits of well-referenced technophile lore, but it could soon become a platform for multiplexed, ultrasensitive detection of a vast range of molecular entities. At the front end of the tadpole one can envision using nearly any encodable natural, designed or evolved protein domain that can bind numerous target classes (ranging from small organic molecules or ions, to protein fragments in the plasma of patients with early-stage cancer, protein-protein contacts, or proteins produced in early-stage viral or biological infection). At the back end of the tadpole, the readout is standard using straightforward multiplexable PCR detection. Brent and colleagues suggest obvious next steps to develop sandwich assays or even, in a nondetection application, nanoscale molecular assemblies. One can even envision applying variations of the principle to obtain far more accurate intracellular measurements of target molecules at the single-cell level (a particular interest of mine). What is important about the work is that Brent and colleagues went well beyond the norm in providing proof of concept for a detection system. The modularity of their approach, the ease with which the recognition domains can be created and simply coupled to a DNA marker for multiplexed measurements, and the extraordinary sensitivity of the approach makes this an appealing system for researchers wanting a standardized high-throughput, and accurate, detection system for... just about anything.
REFERENCES
-
Burbulis, I., Yamaguchi, K., Gordon, A., Carlson, R. & Brent, R. Nat. Methods 2, 31−37 (2005).
-
Perler, F.B. et al. Proc. Natl. Acad. Sci. USA 89, 5577−5581 (1992). | PubMed | ChemPort |
-
Amitai, G., Belenkiy, O., Dassa, B., Shainskaya, A. & Pietrokovski, S. Mol. Microbiol. 47, 61−73 (2003). | Article | PubMed | ChemPort |
-
Muir, T.W., Sondhi, D. & Cole, P.A. Proc. Natl. Acad. Sci. USA 95, 6705−6710 (1998). | Article | PubMed | ChemPort |
|
