I’ve worked out a method for doing immunohistochemistry (IHC) on formalin-fixed paraffin embedded (FFPE) sections of bone.
Bone is very autofluorescent, so most dyes that excite in the visible range also excite substances present in bone itself, making it hard to find the specific signal generated by the dye-conjugated antibody. However, I noticed when using DAPI, an intercalating DNA stain used to light up the nuclei, that the background is almost totally black, meaning that bone isn’t autofluorescent in the UV range, where DAPI excites. So, I figured that all I would have to do is find a secondary antibody conjugated to a dye that excites in the UV range, and I should be able to do IHC with little background.
I’ve been working on a method to identify osteoclasts in FFPE mouse bone sections. In order to do this, I had to solve two problems.
- Bone is very autofluorescent.
- Osteoclasts and macrophages are essentially identical in terms of surface epitopes and enzymatic activities.
The classical stain for osteoclasts is a stain for tartrate-resistant acid phosphatase, which I modified according to this paper, but it also stains macrophages, so there’s two ways to deal with that. You can consider only TRAP+ cells on the surface of bone to be osteoclasts, or you can stain for cathepsin K, which shouldn’t stain macrophages. So that I could tell where I was in the tissue, I needed a nuclear stain, but I couldn’t use DAPI, because I wanted to be able to separate the nuclear signal and the antibody signal so I could set their exposures and gains separately. The reason a nuclear stain works well to tell where you are on a bone section is because there are many fewer cells in bone than marrow, and marrow has a distinctive spongy appearance in these sections, which I believe is partly from the shrinkage caused by dehydration during processing, but also from extraction of the fatty deposits by the hydrocarbon solvents used to dehydrate the tissue. I decided to use propidium iodide to stain the nuclei, and I found an Alexa350-conjugated secondary antibody for cathepsin K. I have been told that CD68 might work to label only macrophages, and not osteoclasts, but I haven’t checked this.
So what I have is two dyes that both excite in UV, but emit about 100 nm apart, with little overlap, and a dye that excites in the visible range(using the alexa594 filter set), but is so bright that autofluorescence can be removed by setting a high threshold. Also, PI’s just a nuclear stain, so it doesn’t matter if the image is a little saturated in that channel.
because this protocol stains using three completely distinct mechanisms, DNA intercalation, enzymatic activity, and antibody binding, I had to work on the protocol to figure out which to do first.
Generally, IHC protocols call for an antigen retrieval step in order to expose the epitopes that the antibody binds to. However, antigen retrieval involves incubating the sections at 100°C for several minutes, or treating the sections with proteolytic enzyme, or oth. This isn’t great for preserving enzymatic activity. Additionally, because PI stains all nucleic acid, not just DNA, like DAPI does, I had to add an RNase step, which meant I needed to permeabilize the cells. Permeabilizing the cells usually involves extracting away the membrane with a detergent such as Triton X100. This isn’t great for preserving membrane-associated antigens like cathepsin K.
I finally worked out the order as follows:
Deparaffinize and rehydrate the sections, stain for TRAP replacing the chromogens in the classical stain with ELF97 at 200 µM. Block the sections with a block solution containing a couple percent serum from the animal in which your secondary was raised, and add some Triton X100 at low concentration(~.1%) so that the membranes get opened, but not completely extracted. Stain with your antibodies, then wash, incubate the samples in ~50 µg/mL RNase A for 20 minutes, then in 0.5 µM PI for a couple minutes. I did 5 and it was a little brighter than necessary.
Before doing this, you should make sure your microscope has the appropriate filter setup to accomodate this. I used a Alexa594 filter set for the PI, and an Alexa 350 filter set for the antibody, and a 350em/488ex set for the ELF97.
I’ll put up some pics soon.