Janet A. Warrington and Mamatha Mahadevappa respond:

Carlsson, Jernås, Lindell, and Carlsson make good points. In every laboratory, people need to confirm a protocol's performance in their own hands. This is especially important when using a protocol with methods different from those used by other groups. Although we cannot comment on the Chomczynski and Sacchi protocol, we found the Trizol (Life Technologies, Rockville, MD) extraction of total RNA (a relatively routine and standard method) followed by RNeasy treatment to be more than sufficient. We do not experience a substantial loss of total yield by this method, usually less than 10%. In a comparison of total RNA yields obtained from tissue versus cultured cells, we have noted that total RNA extracted directly from tissues using RNeasy with no Trizol step may yield smaller amounts of total RNA than cultured cells treated with RNeasy alone. We can only speculate that the substantial loss of material may be due to incomplete homogenization of the tissue resulting in incomplete absorption on the RNeasy column. This finding led us to routinely use Trizol on tissues.

Finally, Carlsson and colleagues raise a concern regarding size selection or bias in the method. This was a concern of ours also, and at the time of our study, the data were analyzed for loss of information due to size selection, level of abundance, and so on. In our comparison, a very small number (less than 1%) of transcripts reproducibly detected in the total RNA samples were undetected in the poly(A) samples. The minor bias detected resulted from the level of abundance of the transcript, not the transcript length.