I think the rate of contents may be different in each sample.

I don't know what was knocked out.  However, if the KO protein is expressed so much, your protein band in the sample of wild type should be relatively less.

Are you talking about a specific band detected with a control serum that does not appear properly in your control group? Or did you just stain the filter (Ponceau stain) and the pattern looks differently?

Lipids can interfere with the BCA assay (http://www.ncbi.nlm.nih.gov/pubmed/1443532). If you're working with liver tissue and the high-fat fed mice have liver steatosis or simply more fat in the liver you may need to take extra steps to minimize the amount of fat in the final sample. When I process my samples for western I lyse into a home-made Laemmli-like buffer (33mM Tris-HCL@pH 6.8, 5% Glycerol, 2% SDS) via sonication, then spin the heck out of the samples (30min at the maximum setting your benchtop centrifuge can go, ~21K RCF for mine). This seems to help with getting most of the fat up to the top of the soluble fraction. RIPA should work about the same, but I've always had better luck with Laemmli's for lysis (less smearing, more consistent loading, fewer weird problem).

I agree with the previous comments about lipids. Maybe you can try to quantify with different methods and see which one works better. You could consider using Bradford and 280nm absorbance. If you are really stucked you could run different amounts of your lysate and try to understand their concentration from the blot (but I agree it could be a problem if you have low volumes)!

Brian Bower makes important points.  The other investigators also hint at the possible complexity of this problem. To me, parsimony suggests that the greatest problem is the unreliability of the standard curve used for quantification.  I suggest mixing rat brain extracts (fat imposter) with BSA in concentrations that recapitulate your experimental liver samples.  If  I am correct, you will better "normalize" your samples upon loading.

Michael

Dear all, thank you very much for your suggestions. Today I try to do again with different samples from WT HFD group.But, unfortunately I still get a same feature like my previous work. Interestingly, in my miR-122 KO mice with HFD (high fat diet) treatment, even I do so many times I can get an equal band size with WT normal chow group and miR-122 KO mice with normal chow for internal control (alpha tubulin and GAPDH). In my miR-122 KO mice with HFD I also can see severe steatosis, but I do not know why this problem only occur in WT high fat diet group. Probably like Brian comment, I should use other protocol to measure my protein samples in particular WT HFD treatment.

Picture 1 = miR-122 KO with HFD; Picture 2 = WT with HFD.

Hi Hendra. I want to make a point on the protein measurement. I use the same kit to measure my protein samples. I usually use 1mg/ml as my highest standard as according to beer-lambert law, any value greater than 1mg/ml usually will produce deviation in standard curve, hence the conc cannot be extrapolated from the standard curve when the absorbance of the samples is higher than the absorbance of the highest standard. So MAYBE your estimation could be wrong because of this. I suggest you dilute your samples, to ensure that the values are below 1mg/ml. Hope this helps!