Seme dimers, such those formed by GPCRs, are extremely resistant to boiling for 5 min. Sometimes overnight at 37 makes the trick

Best

Thanks for your answer.

I'm studying transcription factors of 37 and 32 kDa: for the transcription factor of 37 kDa I usually see the signal at 37 kDa but also at 75 kDa; for the second transcription factor, in the last western I have done, I see a signal between 50 and 70 kDa.

Sorry, I am not familiar with transcription factors. Only one comment: Surely you checked this but are you using reducing conditions?, or comparing reducing with non-reducing conditions? 

Usually I use only reducing conditions since I add b-ME to Laemly buffer before adding it to my samples. I haven't compared reducing conditions and non-reducing conditions yet.

When i worked with b-barrel proteins (OMPs) I use a loading buffer with 8M Urea (6x LB: 200 mM Tris HCl pH 6.8; 8M Urea; 0.1 mM EDTA; 5% SDS and 0.03% Broemphenolblue). If necessary head up your samples to 42°C or 95°C for 10 minutes. But after boiling your samples you can not do MS due to carbamylation of the proteins. With 42°C i never had any trouble.

The transcription factors I have come across so far always ran at a higher MW than expected.

However, for transcription factors I would generally recommend using high salt buffer (including 500mM NaCl) which helps with the extraction an getting rid of the DNA. 

Is anything known about the dimerization interface? If so, you could try to make a mutant that doesn't dimerize anymore. If this still runs at the higher MW you might be dealing with PTMs (as suggested by Florian).