Hi Hanie, if you process all of the samples equally then I think it is a perfectly legit way to compare the samples because you have a baseline.

If you want to highlight (no pun intended) specific structures or staining patterns then it could be a case of cherry picking.

Hi Hanie,

it really depends on what you want to compare in your control and experiments samples. I would strongly recommend NOT to use just the brightness of fluorescence intensity as an indicator of concentration of your target protein, as there are too many variables in the preparation and imaging of samples which can affect the brightness, independently from the protein's concentration (if you try to prepare a large number of control samples, you will notice that the fluorescence brightness will vary quite a lot between them.).

I would therefore suggest you compare the morphology of the sample, to see where the Alexa568 is located in the sample. To that end, a rescaling of the contrast is not only acceptable, but highly recommended, as having low contrast could "hide" in the darkness important details of your image.

Make sure you know what the contrast enhancement function does, so that you can describe the procedure to the readers and reviewers of your research (Stating "we used Imagej's automatic contrast enhancement function" sounds unprofessional :) ).

I would suggest, instead of using a automatic scaling button, to come up with your own contrast enhancement procedure (if you know a bit of computer coding, it would be useful to write a matlab script to reliably always apply the same procedure).

What i normally do is select an area i know is dark in the image, and use the maximum pixel value as my dark threshold (in imagej, under image->adjust->brightness and contrast, the value of the "minimum" slider). Then i use the brightest pixel of the image as the maximum value. If some details that are above the dark threshold are still not visible, i either reduce the maximum value, saturating some areas of the image, or i add some gamma contrast, which should be specified in the image's caption, and repeat the procedure (in imagej you can find the gamma under process->math->gamma).

Hello there Hanie

Interpretation of immunofluorescent signals via microscopy becomes particularly complicated when one is trying to stain for a novel marker. While the use of controls is a good idea, I personally feel that while setting the imaging parameters on the microscope itself one must ensure that the signal to noise is at it's most reliable. Though your positive control may stain brighter for the protein you stain with Alexa 568, if the signal is barely visible on your sample of interest, you are most likely underexposing your sample of interest. Here now, if you digitally play with the low exposure image later, you will also very likely pick up auto-fluorescent or non specific signals that may be in your sample of interest since the laser power, detector gain and other parameters were not optimized to your experimental sample.

Obviously if you are detecting a protein with really low levels of expression, you may have to digitally enhance the image, but I must agree with Paolo that you cannot use these fluorescence as quantitative. Only for representation purposes, digital correction of brightness and contrast can be done. If you intend to make any intensity measurements for comparison purpose with control, appropriate normalization with background intensities must be done. Normally the signal from A488 does not bleed into the A568 emission if you have the emission filters adjusted appropriately.

In short, I would try to optimize the acquisition settings with a positive and a negative control and then again with the experimental samples to see if all images are in the correct range (not underexposed neither overexposed). Once you find a suitable set of parameters where your positive and experimental samples are optimal for signal/noise,that would be the settings for the entire imaging protocol and later for comparison.

Hope this helps. Good Luck

Aparajita

Thank you for your time and insights everyone!

Steingrimur Stefansson

Well, I am carrying out the experiment on 3 different species and each species comes with a negative control and an experimental sample. I don't know if comparing the protein expression in each species is legit, considering that I tuned the exposure to the positive control of only one species (couldn't get the tissue (positive control) from the other two).

And… thanks to you, I think I found the title of the article I’m wrestling with ;]

Paolo Pozzi

The signal I am seeing is mostly at the peripheral of the nucleus, a bit in the cytoplasm, and in rare case on the substrate. Any of these can be true, ‘cause the protein in finally transported to the ECM. Sometimes I feel like by playing with the contrast I’m revealing more than it should be seen; rather than hiding it.

Ah! I wish ALL papers did describe how they acquired their images. To me, many of them are fake and the researcher tunes the laser to get what he wants ;)

Aparajita Lahree

Your name reminded me of a girl I used to know ~15 years ago. I don't know how she looks like now, but if you went to Carmel convent school my hunches might not be wrong :]

Anyway, the gene expression is indeed very low and I am not looking for super A568 brightness. I am not trying to quantify the fluorescence intensity as the experiment is very complicated and I do not have a uniform cell distribution and in some cases the cells are not even sorted.

True, a double control experiment (+&-) will clarify everything. However, my negative control sample is not an actual negative control (it just differs in one parameter compared to the experimental group and a HELL LOT of differences compared to the positive control).... ugh.... confusing I know ;(