The Light Inside (literally)

Autofluorescence can be the bane of an experiment, reducing sensitivity in the detection of any emission range especially between 400-550nm. We tend to think of “autofluorescence” as if it were one singular emission source, when in reality it is the consolidation of multiple sources.

Since collagen is often imaged directly, without additional fluorescent immunodetection due to its strong autofluorescent signal (and amenable sec-ond harmonic resonance) it is a good example of how polymerization of complex macromolecules can create fluorescent signal. Although the major amino acids that comprise collagen are not aromatic, the hydrogen bonding for polymerization to align and condense into a fiber creates internal segments that become rigid and can resonate absorbed energy. This is the case for many proteins involved in cellular adhesion (collagen literally being the glue of the cell). The fact that the signal is strong enough to bother us and affect lower limit sensitivity is due to the degree of condensation of those filaments. Collagen can be dispersed and poorly aligned like in wound regeneration resulting in a dimmer signal or it can be perfectly aligned for maximum compression of signal into a small (pixel) imaging space. Thus collagen autofluorescence poses a more significant issue in imaging applications than say flow cytometry where single cells will not have such a density or detectable variation of fully polymerized collagen filaments.

Bovine Mesenteric Lymphatic Vessel

Dr. Kenton Arkill

Second Harmonic generation image (left) and 2-photon autofluorescence image (right) of a bovine mesenteric collecting lymphatic vessel, stretched with an transmural pressure head. The wavelike collagen fibers can be seen to cross over one another in a “wicker basket” type pattern. The stretched elastin fibers are clearly visible as bright straight lines, mostly in the direction of the vessel (left/right). The dimmer green background is the spillover of the collagen fluorescence.

In flow cytometry or single cell fluorescent analysis platforms, other sources of autofluorescence will pose a greater problem. For example, stimulated and actively proliferating cells, like cancer or PMA-based/ antigen stimulation or proliferation from t-cell activation will cause the activated cell to begin to become more metabolically needy. At a single cell level, vitamin and cofactor needs will dominate as sources of autofluorescence. My favorite molecule to use as an example is Riboflavin or Vitamin B2. Its a cofactor for Ox/Redox reactions and thus is heavily upregulated in cells with strong metabolic activity.

All fluorophores will also be a chromogen (hence the widely over-used term, fluorochrome), which is why autofluorescent cofactors and vitamins will often make the food we eat colored like red/orange for Vitamin A and yellow for Vitamin B2/Riboflavin. However the intensity of that color by eye does not necessarily correlate with the intensity of the fluorescent signal due to other factors (like quantum efficiency) heavily affecting fluorescent signal output.

Generally, when it comes to the function of autofluorescent molecules, its safe to say that if the tissue is filtering toxins and dense with cellular adhesion molecules (like lung), or rapidly dividing (like skin, cancer or replicating cells) or generally has higher than average cellular metabolic activity (like brain), the tissue will have higher than desirable levels of autofluorescence.

However, there is one source of autofluorescence that is not actually innate to the tissue type. I’d rather call it a source of background signal, but that isn’t exactly accurate either. It’s in be-tween the two terms, and it is the fluorescence created or exacerbated by formaldehyde- mediated amine-amine cross-linking due to cellular fixation. Formaldehyde (or paraformaldehyde) tends to be the preferred fixative for biological imaging since the cross-linking is dimeric, be-tween only two amine-containing side chains. On the other hand glutaraldehyde tends to be less preferable for routine imaging due to the higher background fluorescence created by polymeric cross-linking. This figure from Frontiers in Physiology May 2019 from Kaestner et al is one of the better side-by-side comparisons I’ve seen of not only the increase in background fluorescence resulting from freshly prepared glutaraldehyde, but also the loss of the ability to detect an antigen because of the conformational changes that resulted from exposure of the protein to cross-linking (aka loss of antigenicity with the detection reagent).

A solution to reduce the contribution of autofluorescence is in its nature designed to destroy fluorescent molecules. Any proposed solution to reduce autofluorescence would need to be applied to the sample prior to doing any immunolableing since you’ll effectively kill all fluorescence, even intended fluorescence, by these methods. Samples should be washed thoroughly before applying any additional fluorescent probe or antibody. The most common and effective way to kill a fluorophore is oxidation. If you break one of the bonds of its conjugated polyaromatic structure (every organic fluorophore), it will release the (likely planar) rigidity of the fluorescent molecule, nullifying its ability to resonate energy to an excited state. Photobleaching, the most common form of fluorophore oxidation, is produced from illuminating a labeled specimen in a water rich environment and is a condition which defines and limits the majority of biological fluorescent imaging. In this case, to proactively try and kill endogenous fluorescence, a solution of sodium borohydride in the presence of glycine is the most recommended method. Basically, try and break the fluor and then cap its ability to crosslink by substituting glycine. As you might imagine, this results in significant tissue damage. Sodium borohydride will be an indiscriminate oxidizer. Other options are to try and actively photobleach the specimens prior to labeling (again an alter-native oxidation method). I’m not the biggest fan of either of these as a reliable and consistently performing solution to this problem.

Since I can’t write a book in a single blog post (and this post is absolutely not comprehensive), I will refrain from continuing on about other solutions to autofluorescence reduction like quenching solutions. I feel like that can go into a section on methods of background reduction. I will finish by recommending that rather than fighting the battle to reduce the autofluor, accept that you can’t kill it without really hurting the morphology of your cells. Rather, amplify the fluorescence intensity of the desired signal coming from the immunodetection by enzymatic or immunological methods and use wavelengths of emission for that immunostaining that are least affected by the most intense range of autofluorescence for that tissue type. Most people will say that an emission in the very deep red visible range (600nm+) or the NIR (750nm+) will increase sensitivity of detection due to lower autofluor contributions in this range. Although this is true, there are significant caveats that contribute to overall brightness of fluorophores in this range that may not en-sure a better signal to noise ratio. But, that is indeed a whole ‘nother blog post