Cyntia,

It looks like you are experiencing an siRNA off-target effect, which might occur due to similarity of target sequences. As you know, miRNAs can inhibit mRNA translation in two ways: perfect and imperfect binding. It may happen happen to your siRNAs as well, since they are cleaved into ssRNA after they enter the cell. The best way to check this can be through some control experiments suggested below. 

Good Luck with your research

Off-targeting (From wikipedia)

Off-targeting is another challenge to the use of siRNAs as a gene knockdown tool. Here, genes with incomplete complementarity are inadvertently downregulated by the siRNA (in effect, the siRNA acts as a miRNA), leading to problems in data interpretation and potential toxicity. This, however, can be partly addressed by designing appropriate control experiments, and siRNA design algorithms are currently being developed to produce siRNAs free from off-targeting. Genome-wide expression analysis, e.g., by microarray technology, can then be used to verify this and further refine the algorithms. A 2006 paper from the laboratory of Dr. Khvorova implicates 6- or 7-basepair-long stretches from position 2 onward in the siRNA matching with 3'UTR regions in off-targeted genes.[9]

https://en.wikipedia.org/wiki/Small_interfering_RNA#Off-targeting

What type of assay are you using to conclude that "siA only knocks out isoform B, while siB and siC knock out isoform A."?

I used qPCR with two different primer sets that recognize either isoform A, or isoform B and compared relative to a no si control.

You have to rule out that your detection method is not giving you artefact before reaching conclusion that your siRNA are not behaving as expected. How did you assess the isoform specificity of your primers? How did you reach the conclusion that isoform B is more expressed than A in HeLa?

We tested the primers in other cells systems that switch from isoform B to isoform A during differentiation, and confirmed that we could measure that change.

In our control cells (in all the si treated cells as well), the Ct values measured for isoform B were much lower (~17) than for isoform A (~25). Additionally, the literature supports that HeLa cells should primarily express the B isoform.

After treatment with the various si duplexes, we see a dramatic shift in the Ct values. For siA, only isoform B changes (relative decrease of 95% compared to control); whereas with siB and siC only siform A changes (84% and 95% decrease respectively).

I just can't understand why our si duplexes, which all target within the region in common between the two isoforms, are showing us some sort of isoform specificity.

 I agree the results are intriguing. But if you want to push this observation forward, you need stronger data to convince people and it begins by showing that your detection system is truly isoform specific. Showing that primer pair for isoform A doesnt show signal when assay on isoform B alone and vice versa (plasmid or amplicon) is a good starting point (Vandenbroucke II NAR2001, Brosseau RNA2010).  You can also assess the splicing shift by endpoint PCR using primer flanking exon 3 and output the ratio of A +B /B when you treated with your global siRNA. Good luck

RNA secondary structure or RNA binding proteins may interfere with an siRNA's efficacy.

I assume DNAse and control reactions are included (mockRT, no template, no RT primers, no PCR primers).  Amplification artifact may be a result of strand exchange during PCR.  

Run the products on an agarose gel to determine homogeneity and sequence the PCR product to determine specificity.

BLASTing your siRNA using the least stringent parameters of the algorithm may reveal possible off target annealing.

I recommend validating the knockdown of your target by Western blotting or another protein detection method, that is if your goal is to reduce the amount of protein.  If the intent is to reduce the RNA, a northern blot may help.

An additional possibility, to rule out the hypothesis of an off-targeting effect, is to test additional siRNAs. Ideally you should design new siRNAs targeting a different sequence in exons 6 or 8 (obviously, if they are big enough).

Assuming that your detection method is valid (for two isoforms of your target mRNA),  Since the siRNA needs to bind single stand region of mRNA in order to "function" (it won't bind on any "stem" structure (double-stranded region)), therefore the simple explanation on your observation might be that both isoforms transcripts have different "tertiary structure" in its native form in cells due to the difference of exon 3. So different siRNA target regions show different knockdown results in different isoforms.

Another possibility is that your cell line has SNPs that interfere with siRNA efficacy and/or detection.  Cell line SNP and other information is available from the Broad Institute's Cancer Cell Line Encyclopedia at the link below.  HeLa cells do not appear to be included in CCLE, but the pubmed link may help.

http://www.broadinstitute.org/ccle/home

http://www.ncbi.nlm.nih.gov/pubmed/23550136

first make sure your assays to isoforms are very specific. but also siRNA efficiency could be really different- depending on localization of the targets, their levels, secondary structure, bound proteins etc

My protein of interest has two isoforms. one is 75 kDa and the other is 45 kDa. Based on immunoblotting, I see siRNA knocks down only one isoform 75kDa but not the 45kDa. How can I knockdown the other isoform. Please let me know

Hi Kishore,

Seemingly, your siRNA targets a region that is coded only in the longer isoform (75kDa). It could be an alternatively spliced exon that is included only in the longer isoform (75kDa) and skipped in the shorter isoform (45kDa). In that case, you may have to design a siRNA that can target constitutively spliced exons present in the transcripts of both the isoforms.

You can design the siRNA in the region that is common to both 75 and 45Kda.