Proximity-Dependent Initiation of Hybridization Chain Reaction (proxHCR)

(July 3rd, 2015) In this month's Nature Communications, Ola Söderberg's lab at the Uppsala University report a rather unusual way of investigating protein-protein interactions using an enzyme-free DNA technology called the Hybridisation Chain Reaction.

Hybridisation Chain Reaction (HCR) has been around for a good 10 years and was developed as an alternative to PCR. The original developers thought up the idea of using DNA hairpins as fuel packets to drive DNA-replicating machines. It works by taking advantage of the fact that energy is stored in the hairpin bends which is released to expose the hairpin to hybridisation. 

You start with a simple mixture of two hairpins of complementary sequence. This mixture is stable - the hairpins won't spontaneously unravel and so no hybridisation occurs. But if you add a linear oligonucleotide that is complementary to the stem part of the hairpin (called an "activator"), its binding to the stem destabilises the hairpin, and the release of stored chemical energy breaks the hairpin open. This in turn exposes a new segment of the hairpin which in turn serves as an activator to the next (complementary) hairpin, and so on. Result: a very long piece of DNA made without expensive enzymes. And the DNA strand is plenty long enough to be detected using a simple gel apparatus.

The next development in the story was to exploit this technology as a marker, by designing an aptamer that binds the target molecule. The inventors of HCR, Robert Dirks and Niles Pierce then of Caltech, developed an aptamer selective for ATP and showed how HCR could distinguish ATP from GTP.

Söderberg's lab took the idea one step further and showed that you can not only detect molecules, but you can also look at the interactions between molecules. How? The trick is to set the system up so that two proximity-detecting hairpins are needed to start the reaction off. To do this, you start by adding primary antibodies to the two proteins whose interaction you want to track. You then add the two proximity hairpins, each of which is fused to the secondary antibody matching the primary antibody. This means that when the two target proteins physically interact, the two hairpins - Söderberg calls them PH1 and PH2 - will hopefully be brought into close proximity by virtue of the antibody binding. 

Next bring in the activator - that binds to PH1's hairpin loop and opens it up. Classic HCR. But remember that H2 is standing right next to H1, and so H1's opened loop now opens up H2's loop. The result is a long single strand dangling in mid-air. Söderberg likens it to a fishing rod. In this case, the fish is a fluorescently-labelled hairpin (H1 and H2) complementary to the fishing rod, which hybridises and therefore gets broken open. And so the cycle continues until all the H1 and H2 hairpins are used up. 

The result is a really long strand of DNA with a fluorescent probe added for each hairpin. Longline fishing, to push Söderberg's metaphor to its limit. The resulting DNA strand, with its necklace of fluorescent attachments, is easy to detect with a fluorescent reader.

Söderberg makes strong claims for the technique. These include fast association of H1 with the activator and between H1 and H2 with no association between PH2 alone and H1 or H2, or between the PH1/activator complex alone and H1 or H2. They also claim they could pick up a fluorescence signal after 5 minutes into the reaction. 

To show how protein-protein interactions and post-translational modifications can be traced with the technique, Söderberg ran it through a battery of tests, including E-cadherin/β-catenin interactions, MEK/ERK interactions, phosphorylation of Akt and of Syk. They also showed that at least some of these interactions can be detected in formalin-fixed paraffin-embedded sections and in fresh-frozen tissue.

Obviously, time will tell whether these claims hold up for other labs that put them to the test. For one thing, we don't know whether Söderberg's optimisation of their hairpins for sensitivity and specificity will be right for all cell types. But Söderberg's in situ testing panel covered a range of cell and tissue types (HT9 cells, HG3 cells, skin tissue and colon). Another attraction of the technique is that no enzymes are needed, but at the cost of requiring not just two, but four, hairpins, increasing the preparatory work. Finally, the temporal resolution of the technique is poor, and the readout confounds two aspects of the signal: in the case of protein-protein interactions, the strength of the signal reflects the probability of the two proximity hairpins being close to each other, and that will depend not only on the number but the duration of the protein-protein interaction.

That said, Söderberg's "proximity-dependent HCR" adds a rapid and comparatively cheap probe of protein interactions.

Steven Buckingham

Picture: Chris Schlag

Last Changes: 08.14.2015