Hi! The best way is to test several miRNA and check the most stable ones by Normfinder or Genorm software; by the way, in my plasma samples mir-103a-3p and 93a-5p, were the best choices. Important, i'm not working with cancer samples, depending on the pathology microRNA can have different expression levels and stability!Another method which is very useful with plasma and serum where a proper housekeeping is hard to find, is the global mean if you have several samples (not only 2-3). This means that you can run your Realtime without the housekeeping and normalize the ct value of each sample on the global mean of your experiment for that specific microRNA.
I add two spike.ins to the simple prior to miRNA extraction (cel-34 and cel-238) and a spike-in control to the RNA simple prior to the reverse transcription reaction (cel-54).
When adding the spike-in controls to the simple before extraction, be sure that you add them after the addition of the lysis buffer or the phenol (dependidng of the extraction method) to avoid degradation of the spike-in control sequences.
Thank you all, I will use spike-in. But wouldnt there be a difference between a protein bound native miRNA and unbound spike-in miRNAs in practice? After all protein bound miRNA should last longer and harder to isolate. btw, I plan not to use any reverse transcription, instead using isolate from serum directly.