Inconsistent/low library yields using NEBNext mRNA lib prep kit?

[danaz123]

Hi all,

I don't have a ton of experience with library preps, so I'm having trouble figuring out what the issue is/what to do next in this situation. I used the NEBNext Ultra II Directional RNA Library Prep Kit for Illumina to generate cDNA libraries from RNA samples. When I piloted the protocol, my libraries came out looking good, but when I ran my actual samples recently they were super inconsistent - some ok, but most of them having really low yields (low library peak on Bioanalyzer). The library peak is expected at ~300bp. Most of them have a bump in this area, but it's small. There doesn't seem to be an issue of adapter/dimer or primer peaks.

There also doesn't seem to be a relationship between input RNA concentration and library success, and all the input RNA samples looked good on Bioanalyzer.

I'm not sure to what extent this is expected and/or if any of the bad samples are usable.
I've attached three example Bioanalyzer traces - the top one is a good library, the bottom left is a low yield (potentially usable?) library, and the bottom right is a very low yield library. Apologies - the x-axis is showing seconds instead of base pairs, but I labeled the library peaks.

If either of the low yield library examples are usable, is it still an issue that there are inconsistencies with the libraries?

And does anyone have any idea what could have caused this?

Happy to provide more info if that helps.
Thanks!


EDIT 10/2023: Just wanted to provide an update for anyone who may run into similar issues. I don't think it was a problem with the beads/ethanol carryover - I tried letting them dry a little more, but it didn't seem to have much of an impact. In the end, I was able to produce consistently usable libraries by increasing the RNA input and/or adding 1 or 2 PCR cycles at the final library amplification step. The added PCR cycles made a bigger difference for library yield. For reference, I started out using 250ng total RNA input (from mouse brain single tissue punches) + 12 final PCR amplification cycles, but, after piloting, I think a better starting point would be 300-350ng input RNA and 13 PCR cycles (of course, this may differ dramatically based on your input sample type).

[Ben3]

Hello danaz123 I think this could be the result of 3 different things. The first is fragmentation time. Did you change the time or conditions for this? I wonder if it ended up being more fragmented than previously.

The other two things are either the samples weren't amplified enough or something went wrong during the clean-ups. I've accidentally lost significant portions of libraries due to small changes in the clean-up. It's hard to know without watching you perform the library prep but those are my ideas.

[danaz123]

Ben3 thanks so much for your helpful response! Yes, it's difficult to narrow down exactly what could be going wrong. I didn't change fragmentation conditions or any other part of the protocol, but I was running a larger set of samples and I may not have been able to handle each one as carefully as I did during the pilot. I was also thinking there could be a problem at the cleanup stage - possibly ethanol carryover? I'm so paranoid about the beads drying out that I wonder if I left some of them too wet. Also considered adding an amplification cycle; might do that, and I'll think about fragmentation too. I just wish the procedure wasn't so long and arduous and the materials weren't so expensive so I could troubleshoot more easily haha!

[Ben3]

danaz123 no problem and sorry I couldn't be of more help. Another recommendation is maybe to give a test on only a sample or two and see if anything changes. Many times these kits have enough for an extra reaction or two and I would always use that to play around with the procedure because you're right, they are expensive!

[mikebenway]

Thanks for the update.
I did have one thing I keep in mind. When libraries look good but yields are low, I become suspicious of Ligation reaction setup - it really benefits from a lot of initial mixing - always more than you think, especially with automation. The symptoms are exactly as you describe. If, for some samples, extra cycles are problematic because of other metrics, like complexity or duplication, then more aggressive mixing in that step could help.

[danaz123]

mikebenway Thank you so much for this tip. I have noted it in my protocol.