RNA-seq library prep problem

[rpickin]

I have made 4 attempts at generating an RNA-Seq library from a human lymphoblast cell line to no avail. I am trying to figure out where the problem is and what to try next...any and all suggestions are welcome! Thanks!

Here's what I've done so far...

I have successfully made 1 library from Poly-A RNA isolated from HeLa cells (we obtained ~15 million mappable reads from this library). I began this prep with 1ug poly-A RNA starting at the "Fragment the mRNA step". The second attempt at a biological HeLa replicate library was also started the same way and appears, upon library validation with the bioanalyzer, to look like the first library. This library is currently in the queue to be sequenced.

I am also attempting to generate biological replicate RNA-Seq libraries from lymphoblast cells in the same manner (harvest cells, isolate total RNA, select Poly-A RNA and check - through RT-PCR and Northern blots - the quality of this RNA before starting the Illumina protocol at the Fragment the mRNA step). I have attempted 4 libraries in this manner. The first two were started with ~65ng and ~45ng of poly-A selected RNA. Bioanalyzer analysis of these first two libraries resulted in fragment sizes of ~150bp or so. Both had concentrations measuring 29 and 25ng/ul each for the final library. Because the fragment sizes measured so small on the bioanalyzer, I simply stored these libraries and tried again with fresh RNA. Screening these via cloning also resulted in insert sizes of ~150bp.

The second two libraries were started 2.5ug and ~3ug poly-A isolated RNA, respectively. Again, the Illumina protocol was started at the "fragment the mRNA step" and the instructions were followed to the letter. All kits and kit components were stored and maintained correctly (including the QIAGEN kits which were purchased fresh with the first Illumina kit). The concentration of each of these libraries measured 38ng/ul and 28ng/ul, respectively. (I would expect a slightly lower concentration for the second library because there was slight loss of sample when loaded onto the gel at the "Purify the cDNA Templates" step.) The bioanalyzer results showed fragment sizes of ~150bp for G1 and a failed result for G2. I have attempted to clone these libraries as well - again, insert sizes are ~150bp.

The second set of lymphoblast cell libraries was made with a new Illumina RNA-Seq kit (lot #5332696). No other library generation attempts have been made with this kit.

The lymphoblast cell line was obtained from Coriell Institute and is a lymphoblast cell line. It is less than optimal to work with and often provides 1/2 the amount of total RNA starting material from the same number of cells as does our HeLa cell line.

I've checked with Illumina and so far the only suggestion was that I had overloaded the fragmentation step when I attempted the second set of lymphoblast libraries and to try a titration. There was no explanation of why the 1ug of HeLa RNA was successfully used for library generation...Both cell lines are human in origin and all starting RNA looked similar when checked via Ribogreen, EtBr stained gel and Northern blot probed with a mixed 28s and GAPDH probes, and RT-PCR.

Has anyone else been successful generating libraries from one cell line and then fail so miserably with a second cell line?

[scooter]

One thought I had about your ills is that it might be due to a bit excessive fragmentation. I do not agree with Illumina's comments that you have saturated the fragmentation reagent with the amount of RNA you have input. The frag buffer is simply ~10 mM Zn2+ plus some other stuff. Regardless, the fact that you are getting fragments smaller than desired does not mess with the idea that your frag buffer is saturated with RNA. I would think that would result in the accumulation of larger RNA fragments, not smaller.

Consider simply fragmenting for a shorter time or lower temp. Samples will vary somewhat on how they fragment and your cell line may produce more 'sensitive' RNA. I would think if you are starting with microgram amounts of polyA that you should be able to monitor the fragment lengths on a bioanalyzer after the purifications. This way you can be sure of what you are placing into the library prep is of the appropriate length. I would shoot for a peak size of ~200-300 bases depending on your sequencing application.