6 micrometer is too thin for morphological analysis of microglia. You will need at least 15 um section. You might be able to use other activation markers such as p-p38, MHC, CD45 and so on, but we never know which one is the the "definite" activation marker.

Dear Kimberly,

maybe fractal or sceleton analysis could be a possibility. There are some interesting manuscripts published, e.g.:

Karperien A, Ahammer H, Jelinek HF. Quantitating the subtleties of microglial morphology with fractal analysis. Front Cell Neurosci. 2013;7:3. doi: 10.3389/fncel.2013.00003.

Soltys Z, Orzylowska-Sliwinska O, Zaremba M, Orlowski D, Piechota M, Fiedorowicz A, Janeczko K, Oderfeld-Nowak B. Quantitative morphological study of microglial cells in the ischemic rat brain using principal component analysis.J Neurosci Methods. 2005 Jul 15;146(1):50-60.

Soltys Z, Ziaja M, Pawlínski R, Setkowicz Z, Janeczko K. Morphology of reactive microglia in the injured cerebral cortex. Fractal analysis and complementary quantitative methods. J Neurosci Res. 2001 Jan 1;63(1):90-7.

Best regards

Katrin

Dear Kimberly,

We have done a couple of studies specifically looking at changes in microglial morphology.

There are a lot of ways you can get at this information. One of simplest is to use a freeware program like imagej or Fiji. Using these packages you can convert your raw image (say a .tiff or .jpg) into a binary image and then using this new image assess basic metrics such as max diameter, length roundness etc. This is pretty good place to start. There are some other interesting packages - like fraclac which you can use to get into issues like how fractal the shape of the cells are.

If you want to go a step further then some commercial packages may be of benefit. We have extensively used Neurolucida from MBF biosciences. Using this when coupled with automated XYZ stage it is relatively straightforward to create high fidelity reconstructions of microglia at 100x. We tend to find depending on their complexity it takes 10-30minutes/cell to reconstruct. We then port the digital reconstruction's into the companion product NeuroExplorer which you can use to produce an almost limitless list of measurements. Of these I particularly like Scholl analysis undertaken on branch length, branch number and volume.

We have recently started working with Imaris and this is really nice particularly for semi-automated 3D reconstructions. In essence, however, it produces data similar to Neurolucida.

Best

Rohan.

I propose a very simple method in my recent publication. I would agree that 6um if that is what you are proposing, is less than ideal since this would not adequately encompass the soma and processes that may extend for quite some distance; this is the nature of this little beastie and you must account for it in your sampling to avoid validity problems.

Cheers

Helena

Thank you everyone for your replies.

Rohan- I have Neurolucida at the moment and under 100 X I am able to trace the soma and the branching. Problem is that I only see some of the branching and the soma that I am tracing could be cut through at a place that makes it look smaller than its largest diameter. Will this be accounted for across treatment groups? Scholl analysis sounds great and we do have the capabilities to do this, I am just worried about the thin sections for this. Would you suggest giving it the old college try?

It seems that people are saying that 6um is too thin for microglia. I know that. The tissue has been optimized for new neuron identification. That is the nature of the beast, unfortunately. I need to make use of the tissue that I have.

What measurements could I take using the 6um tissue that I have, and still be confident in my data?

Thank you in advance!

-Kimberly

Dear Kimberly,

On a good day you would set up your experiment taking into consideration all conceivable variables that could bring you unstuck later and then execute a perfect experiment. We are always dealing with tradeoffs like section thickness and fixation length in lab - its never-ending!.

So your sections are thin. Yep. I would agree with that. Will you be able to pull absolute numbers from the traces...no. The important thing that you have going for you is that all your sections across your groups have been processed the same way, so it still may be possible to see if there are any relative differences. In terms of how you should handle the metrics/measurements its hard to say - scholl may not be your best option though. Could you send me a couple of images? I can have a look and give you some more specific feedback.

Best

Rohan

Kimberley,

we have also run into this issue looking at astrocytes, microglia and endothelial cells. While 6µm sections are very thin, and less than ideal, it is just plain wrong for a reviewer to say that they not useable (believe me, my group has had this fight numerous times). Oftentimes, working in a pathology department may mean that only paraffin embedded, archived samples, are available for use. The power of "n" comes in here. If you can increase your sample size, and image LOTS of cells, then you CAN start getting very meaningful data: for example, what is the largest soma size you see? What is the average length of processes in the plane / section you have? How many primary processes are there? It will also be necessary to include the use of thin sections in your discussions.

Rohan has offered his assistance already, but please feel free to reach out to others (including Helena and Katrin, above) for support: the pdf files are interesting and worth reading.

Great, Andrew! Thank you very much. I appreciate the encouragement in the ability to make use of the tissue that I have available to me. I will absolutely reach out to others and read those pdf files tonight. Right now I am going to read your J. Neurosci. Methods on the BBB. Fascinating! We use a peripheral immune response to trigger the microglia in the hippocampus to react. This will be a big piece of what I will need to address in order to help validate my model. Thank you for the assistance!

Reposted from my duplicate deleted profile.

Hi, Kimberley. Pattern extraction for fractal analysis is a little different than for other analysis methods inasmuch as one can reasonably assume that 2d projections of 3d forms adequately convey the information of interest for fractal analysis. This is discussed in our paper noted above, which you can get free at PMC (see link). As someone else pointed out, the more images you get, the merrier.

Also, if it is relevant, you may wish to analyze large areas to assess the variation in the fractal dimension over the tissue as a whole. We've used FracLac for these types of analysis, which are built in to the software. I can look at some images and tell you more about it if that will help.

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3558688/

Quantitating the subtleties of microglial morphology with fractal analysis

Jul 18, 2013