In my opinion, it is not mandatory to use a non-coding sRNA to control for miRNA expression (though it could be perfectly the case). It should be reminded that miRNAs are transcribed by RNA pol II, much like messenger-RNAs but not for example tRNAs. The processing machinery is quite different for Pol I /II / III-processed transcripts. Besides, to date I do not know about a bona fide housekeeping microRNA.
For these reasons, I use common housekeeping genes, like EF-1alpha, bur you should select the one that in your condition really behaves like a housekeeping, for example looking at Genevestigator or reading related bibliography.
Having said that, I would suggest you to try and compare 3-4 putative housekeeping genes and select the adecuate using Bestkeeper, Normfinder or other software developed specifically to find what housekeeping gene is the most stably expressed in your samples.
In my opinion, it is not mandatory to use a non-coding sRNA to control for miRNA expression (though it could be perfectly the case). It should be reminded that miRNAs are transcribed by RNA pol II, much like messenger-RNAs but not for example tRNAs. The processing machinery is quite different for Pol I /II / III-processed transcripts. Besides, to date I do not know about a bona fide housekeeping microRNA.
For these reasons, I use common housekeeping genes, like EF-1alpha, bur you should select the one that in your condition really behaves like a housekeeping, for example looking at Genevestigator or reading related bibliography.
Having said that, I would suggest you to try and compare 3-4 putative housekeeping genes and select the adecuate using Bestkeeper, Normfinder or other software developed specifically to find what housekeeping gene is the most stably expressed in your samples.
You should experimentally validate whether U6 is a stable gene. In my experience U6 was the least stable gene. Try doing
U6, RNU1A, as well as 3-4 miRNAs that are relatively well expressed in your samples. Have a look on the literature what miRNAs are abundantly expressed in your sample of interest.
Once you have acquired the data, you will need to combine them using a stability analysis software such as GeNORM or the one said above by Carlos.
In my sample heart samples (not related to cancer), I used U6, RNU1A, miRNA-22, miRNA-26, miRNA-103 and miRNA-191.
You can use a combination of snRNAs and snoRNAs in miRNA qPCR assay (if possible) to strengthen your data or validation of true expression. Otherwise U6 snRNA can be used as ref control gene. For your info, in some instances including studies with cancer cell models, small non-coding control RNAs can even vary across cell/tissue types. So it's more acceptable when you have a set of ref genes and take the avg. expression to normalize your specific miRNA level. For more explanation please check attached publication:
Article Altered microRNA Expression Profiles and Regulation of INK4A...