When virus RNA is produced in the cytosol, Dicer (a RNAse III Enzyme) can cut it to short nucleotides (20-25 nt) that are called siRNA (small interfering RNA).

This siRNA can be incorporated into the RNA-induced silencing complex (RISC). RISC can now knockdown mRNA that has the (complementary) sequence of the siRNA.

 

My question: Does RISC knocks down every mRNA? Does it know down the "good" cellular mRNA-transcripts or just the "bad" virus mRNA?

I mean, why does RNA interference exists at all?

The recognition is based on the sequence. If it fits, it gets diced. RNA looks the same, regardless of origin.

RNAi serves basicly for recognition of viral dsRNA. Normally, there´s no dsRNA in cells, and when it appears, it´s a sign that we have a company. So it´s more like antiviral defense, which is now used for experiments with cellular genes.

Well, partially true. It is likely that its origins were defense against transposons and viruses. However, it has also been shown to play a role in endogenous gene regulation. It has been shown that sense and antisense RNA is being produced (I think one of the first papers was in Drosophila 2000 or 2001). And it appears that this is also regulated by silencing.

RNAi is a more fine-tuned regulation system. It probably evolved as an anti-viral response, but cells later used it for their own regulation (too lazy to pull up the references). For example, it'll be easier to slow down regulation by producing a specific amount of RNAi rather than adjusting ribosomal level or etc. This fine-tuning of translation is the likely reason for the use of RNAi.