First attempt on IF, I did HER2 on breast ca cells. Cytoplasmic stain (green) observed instead of membraneous as per literature. Im not sure if this cytoplasmic staining is real or am I not using the correct Z depth adjustment..any thoughts?
Membrane proteins often also found in the cytoplasm, so it is possible that your protein is also distributed in the cytoplasm. I did not see any in the nucleus, so that is also a good sign.
Sadly, this is also what off-target secondary antibody staining looks like. I would recommend trying another primary antibody with a distinct structure (tubulin, cox-iv, lamin) to make sure that your IF protocol is working properly.
To know if you're doing a good sampling you need to use the Nyquist-Shannon theorem. If you translate that theorem to a z stack, it tell you, more or less, that you need to have a lest 2 steps per resolvable element. Without going too much into details, your resolution in z is equal to 2 lambda / (NA)2 (lambda is your wavelength and NA is the Numerical Aperture of your objective) so, for at 500nm, with a 1.4NA obj, you should do a step every ~ 150 to 250nm.
Secondary Ab alono is good but not enough to probe specific staining. It is a common mistake to only use this control when testing an antibody for the first time. Usually, after developing an antibody, companies check that it recognize a protein by blocking the binding with an excess of the same protein. This shows that the antibody recognize that particular protein, but it may very well bind to other related proteins.... perhaps other family member, perhaps lower affinity... In a WB you can identify these unspecific bindinds by the MW, but in an IF to be completly sure you need a negative control like Knockout or silenced cells.
For intracellurar localisation you need to have rather good resolution - objective x60 or x100 and frame 1020*1020. It also help to make z-stacks (step around 200-250 nm) and mean scanning by 2 or 4. Then you can have better idea on the protein distribution in a cell.
First of all, you should check that there is no autofluorescence from your cells (without any antibodies at all).
Secondly you shoud do a "just secondary antibody" control. Again you should see no or very tiny signal compared to your signal that you imaged in cells incubated with primary and secondary antibodies. Using same settings in confocal microscope.
If you see intense labelling with your secondary antibody alone, that will indicate a problem. Then you should do another control with secondary alone without any permeabilization of your cells. If you still see intense labelling, especially including intense nuclear labelling, that means that there is a problem with your cells. Because I am doing electrophysiological recordings from my cells prior to IHC (see my methods papers in RG), I know that if the cells are not healthy and have low membrane potential (depolorized) the membranes become leaky and porous that will let non-specific incursion of any antibody across the plasma and nuclear membranes and get tangled in there. In brief, you get good and realiable results in immunohistochemistry if you start with good and healthy cells, provided you have good and specific primary antibodies.
The right z-step is important to get good images, but the wrong z-step does not cause a false-positive staining. So that shouldn't be your first worry. Some confocal microscopes, like the Zeiss LSM-710 that we use, are able to automatically calculate the correct z-step from the acquisition parameters that you use (objective NA, laser lambda, etc) using the Nyquist equation that is built-in the software.
Your problem could be: 1) non-specific binding of your primary antibody, 2) non-specific binding of your secondary antibody. As other people suggested, you need a control with secondary antibody alone. Another trick to use is to label your primary with two different secondary antibodies with different fluorophores - for example, if you primary antibody is rabbit, you could use two anti-rabbit secondary antibodies tagged with Alexa Fluor 488 and Alexa Fluor 555. Then if the green and the orange signals do not colocalize, you know one of your secondaries is not working right - probably labeling non-specifically. Use the secondary antibodies at low concentration so they don't saturate the primary antibody.
Dealing with non-specific binding of the primary antibody is much more difficult. The golden standard is to test it in cells transfected and not transfected with your target protein, or using KO mice in that protein as a negative control.