Quantum dots for Immunofluorescence

Guest post by Dave Mason

In modern cell biology and light microscopy, immunofluorescence is a workhorse experiment. The same way antibodies can recognise foreign pathogens in an animal, so the specificity of antibodies can be used to label specific targets within the cell. When antibodies are bound to a fluorophore of your choice, and in combination with light microscopy, this makes for a versatile platform for research and diagnostics.

Most small-dye based fluorophores that are used in combination with antibodies suffer from a limitation; hit them with enough light and you irreversibly damage the fluorochrome, rendering the dye ‘invisible’ or photobleached. This property is the basis of several biophysical techniques such as Fluorescence Recovery After Photobleaching (FRAP) but for routine imaging it is largely an unwanted property.

Over 20 years ago, a new class of fluorescent conjugate was introduced in the form of Quantum Dots (QDots); semiconductor nanocrystals that promised increased brightness, a broad excitation and narrow emission band (good when using multi-channel imaging) and most importantly: no photobleaching. They were hailed as a game changer: “When the methods are worked out, they’ll be used instantly” (ref). With the expectation that they would “…soon be a standard biological tool” (ref).

So what happened? Check the published literature or walk into any imaging lab today and you’ll find antibodies conjugated to all manner of small dyes from FITC and rhodamine to Cyanine and Alexa dyes. Rarely will you find QDot-conjugated antibodies used despite them being commercially available. Why would people shun a technology that seemingly provides so many advantages?

Based on some strange observations, when trying to use QDot-conjugated antibodies, Jen Francis, investigated this phenomenon more closely, systematically labelling different cellular targets with Quantum dots and traditional small molecule dyes.


Figure 3 from doi:10.3762/bjnano.8.125 shows Tubulin simultaneously labelled with small fluorescent dye (A) and QDots (B). Overlay shows Qdot in green and A488 in Magenta. See paper for more details. See UPDATE below.

The work published in the Beilstein Journal of Nanotechnology (doi: 10.3762/bjnano.8.125) demonstrates a surprising finding. Some targets in the cell such as tubulin (the ‘gold standard’ for QDot labelling) label just as well with the QDot as with the dye (see above). Others however, including nuclear and some focal adhesion targets would only label with the organic dye.


The important question of course is: why the difference in labelling when using Quantum Dots or dyes? This is discussed in more detail in the paper but one explanation the evidence supports is that it is the size of the QDots that hinder their ability to access targets in the nucleus or large protein complexes. This explanation further highlights how little we really know about the 3D structure of protein complexes in the cell and the effect of fixation and permeabilisation upon them. Why for example, can tubulin be labelled with QDots but F-actin cannot, despite them both being abundant filamentous cytosolic structures? At this point we can’t say.

So why is this study important? Publication bias (the preferential publication of ‘positive’ results) has largely hidden the complications of using QDots for immunofluorescence. We and others have spent time and money, trying to optimise and troubleshoot experiments that upon closer study, have no chance of working. We therefore hope that by undertaking and publishing this study, other researchers can be better informed and understand when (or whether) it might be appropriate to use Quantum Dots before embarking on a project.

This paper was published in the Beilstein Journal of Nanotechnology, an Open Access, peer-reviewed journal funded entirely by the Beilstein-Institut.

UPDATE [2017-06-13]: in response to a comment below, I’ve updated the overlay figure to use green/magenta instead of green/red. The original figure can be seen in the paper or here.



  1. Hi Dave
    Really nice job. Great to know this info!!

    2 little things.

    Could you have used green magenta instead of red green, so the many colour blind folks have a better chance to see what’s going on. Maybe not too late to change? Great you included scatterplots though!

    For colocalization analysis, You could have used coloc2 plugin already in Fiji. Which handles Manders’ and Pearson’s coefficients and more, and regions of interest, with scatterplots amd standardised output in PDF. No need for matlab and Jacop (perhaps not well-supported anymore)

    Liked by 2 people

    1. Hi! Thanks for the comments. You make an absolutely valid point with regard to the merged images. As the paper is already published, I don’t think we can change those versions. I’ll sort out an alternate for the post when I get a minute this week. This once again though, highlights the importance of providing the single channel data (as we have done throughout).

      Regarding the choice of plugin, all of the implementations fall back to the calculations in the original papers (Pearson”s, Manders, Costes &c) so I don’t see there’s much difference between them, I’ve always likes the Bolte / Cordelières paper (http://doi.org/10.1111/j.1365-2818.2006.01706.x) so JACoP tends to be my first choice. I certainly have nothing against coloc2.

      Not sure I agree that there’s “no need for MATLAB”: I can’t bring myself to make figures with Excel plots. Regardless, Jen plotted the data for the final figures, so the choice was hers.

      Thanks again for the comments!

      Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s