In situ PCR: MDx meets histology

July 22, 2015

Historically, a major activity of clinical diagnostic laboratories—and the anatomic pathology section in particular—has involved the examination of tissue sections by microscopy following preparation and sectioning. The provision of spatially defined information through the appearance of particular attributes or markers in specific cells or organ structures is a powerful tool in identifying the underlying basis of macroscopic pathology. While the basic approach of this method is one of the oldest established diagnostic lab methods, it has evolved over time to take advantage of technological advances, giving it increasingly specific and informative capabilities. Earliest iterations employed direct visual observation of tissue sections, which was rapidly augmented by the use of stains that allowed for greater contrast and differentiation of structures. (Haematoxylin staining was used as far back as 1758 on plant tissues; and after more than a century of medical use, the Haematoxylin/Eosin , or “H&E” staining pair, continues as a first choice mainstay to this day, recognizable to even many non-specialists.)

A major advance in the method came with the development of immunostaining, where labeled antibodies directed against specific antigens of interest allowed for target-specific “painting” of microscope sections. Particularly when used with fluorescent labeling strategies, immunohistochemistry (IHC) stains can be combined on a single specimen to visually indicate the spatial relationships between multiple markers at once. While not strictly quantitative, the method is more than qualitative, as intensity of staining relates to amount of antigen. Overall, the combination of classical staining and IHC provides a direct and highly valuable information source to the eye of the trained anatomic pathologist.

The story of clinical use of microscopic tissue section analysis has been one of constant evolution to take advantage of new approaches, and readers of this series may be tempted to wonder whether this adaptability has included molecular technologies as well. It has, and in the next few installments of this series we’ll examine some of these methods along with their capabilities, applications, and obstacles to use. This month, we’ll consider IHC’s options for marriage to that most central of molecular methods, polymerase chain reaction (PCR).

IHC meets PCR

The concept is a relatively simple one. Imagine being able to perform PCR (or reverse transcription [RT]-PCR) on a tissue section, somehow labeling in-situ not just protein, lipid, or carbohydrate antigens but the actual underlying genetic material, with the extreme sensitivity and specificity inherent to PCR. For genomic DNA targets the concept isn’t initially overly exciting, as given an expectation for equal genetic content of all specimen cells a dull and uninformative uniformity is to be expected. Exceptions to this occur, however, whenever the underlying assumption of equal cellular content of DNA does not hold true; identifying viral DNA in specific cells of a section, or recognizing cells in which an intrinsic DNA element has undergone significant amplification (that is, many more copies are present than in unaffected tissue) are both examples of situations where the approach can be meaningful.

In contrast, the idea of RNA-based RT-PCR approaches, with the theoretical ability to visually evaluate direct expression levels of cellular mRNAs in their physical contexts (albeit somewhat subjectively), is truly exciting. When you further consider that this approach could be made specific even to particular transcriptional splicing variants, and that specific reactions can be developed with the speed and ease of designing and ordering oligonucleotide primers as opposed to the time and cost required for development of a monoclonal antibody suitable for IHC, it all sounds too good to be true.

That’s because so far we’ve just been considering this from a theoretical perspective. As the adage goes, “In theory, there’s no difference between theory and practice; in practice, there is.” The fly in the ointment is best understood if you consider that in-situ means just that: target identification specifically restricted to a cellular or even subcellular localization. PCR (and RT-PCR) are by their very nature reactions dependent on free diffusion of reaction components (reagents, amplicons, and even to some extent, template). These two concepts are diametrically opposed, and to make “in-situ PCR” anything but an oxymoron takes some very careful balancing of conditions.

Ignoring aspects such as the impact of formalin on DNA suitability as a PCR template, conducting a successful PCR (or RT-PCR) on a fixed tissue section mounted on a glass slide will mean having to create some sort of access through the intact cell membranes and nuclear membranes, and possibly through heterogeneous connective tissues, for the PCR primers and polymerase to diffuse through to find their nucleic acid targets. (Astute readers may note that for RT-PCR targeted against mRNAs, at least the nuclear membrane should not be a problem as the targets will be cytosolic; this will be slightly more manageable.)

Solving the problem

Of course, poking gaping holes in thin sections of fixed tissues isn’t challenging; numerous digestion approaches based on proteases, lipases, detergents, or combinations of the above can readily disrupt fixed tissue and expose the nucleic acids. Doing so too aggressively, however, creates an opposite problem: the target nucleic acids may now no longer be effectively fixed in their original locations, and may diffuse about the reaction, obviating the entire purpose of in-situ. In fact, long before tissue over-digestion leads to potential template diffusion, amplicon diffusion will be an issue. If our “average” amplicon size is perhaps on the order of 500 bp, one can imagine that it doesn’t take too large of a hole in a fixed cell for concentrated amplicons to start escaping and creating a veritable diaspora of labeled amplicons all over the section.

To work, then, the tissue sample must be digested just enough to create holes big enough for polymerase and primers to get into their cellular target compartments, but not so big as to readily allow amplicons out. This narrow range of acceptable digestion, of course, won’t be the same for all tissue types either, as some will digest faster than others. Naturally, a tissue section slide may have more than one cellular type or microenvironment represented, meaning that even digestions in the Goldilocks range for one represented cell type may be too little, or too much, for other cell types present.

With all the above in mind, let’s consider how the process then works. A slide mounted, deparaffinised tissue section of interest is normally digested to what the scientist hopes is the correct degree. It is then framed with something like an adhesive chamber, which is then filled with a PCR (or RT-PCR) reaction mix containing polymerase, labeled primers (frequently biotin labeled, although other labels including direct fluorescent label is possible; alternatively, one nucleotide in the reaction mix may be labeled such as to incorporate label during primer extension). A cover is sealed over the framed tissue area, trapping the reaction mix in contact with it. The entire slide is now thermocycled (ideally, in a PCR machine with custom in-situ slide blocks, although people have been successful in only laying the slide section across a normal tube block). PCR or one-step RT-PCR (that is, a mix containing both the reverse transcriptase and the DNA polymerase functions, as there isn’t an opportunity to open the slide section post and change from a reverse transcription reagent to a polymerase reagent) occurs more or less as normal, although cycle parameters can be quite different from tube-based PCR; generally, longer dwell times at desired temperatures are required both to account for the relatively large thermal mass of the system, and also to help allow for slower reagent diffusional rates.

If all goes well, the reagents make their way into the cells, and where appropriate target is present, create localized concentrations of labeled amplicon which are slightly too large to escape from their locale. Post-PCR, a gentle washing step is used to help remove unincorporated labeled probes (or nucleotide). Again, this is something of a balancing act between not vigorous enough (labeled primer left everywhere) and too vigorous (all the valid signal washed away). Then, finally, one visualizes the remaining label, either directly if it’s a fluorophore, or via a coupled enzymatic staining immunotargeted against, for example, biotin labeled primer, and observes the slide through a microscope.

Coming soon to a lab near you?

By this point the reader is probably thinking that with all of the competing problems discussed above, that chances for seeing anything meaningful are slim to nil. Amazingly, however, with a combination of careful protocol optimization, perseverance, and a bit of luck, it is possible to get truly spectacular images showing localization of the target nucleic acids in their tissue context.

Are you about to see this arrive as a service in your local molecular diagnostic lab? Don’t hold your breath for it; the reality is, while it can work, it just doesn’t have the reliability and throughput needed to make it a good routine diagnostic method. Despite obvious potential, this approach is likely to stay relegated to the research lab, where one spectacular result every few months is enough to ensure grant renewal.

We started this month’s discussion by considering how IHC has been highly adaptable across its history; is this then the final word on marrying tissue sections to classical PCR? Thankfully, no. While it requires some highly costly and specialized equipment, use of laser capture microdissectors (LCMs) can allow direct visualization of a tissue section on a slide, and selective recovery of areas as small as a single cell or small group of cells to a reaction tube. These pre-selected cells can then be digested (fully, without fear of over-digestion) and subjected to regular (or perhaps quantitative real-time) PCR or RT-PCR. Selection of adjacent control tissue areas for comparison of molecular signal, mapped back against where the samples were taken from on the original slide, can provide similar spatially resolved information on molecular target localization and load within a multicellular image. Still other molecular methods such as FISH (an upcoming topic in this series) also act to combine molecular techniques with classical tissue microscopy, albeit not with traditional PCR. While the adaptation of molecular approaches to anatomic pathology remains challenging, these methods at the interface between the two fields have not been without success.