Getting TruSeq to work with RNA samples

[SueFoltin]

So have isolated RNA from FFPE. Originally our isolations were performed using the Qiagen AllPrep kits. Now we are using the RNA/DNA STORM kits from Cell Data Sciences.

We have a CORE that takes our RNA samples and generates library material using Illumina's kit: RS-122-2201/2, TruSeq® StrndTL RNALT RboZroHMN/Mse/Rat. We are using ribo-depletion and have asked for them to increase the PCR cycles from 8 to 12. We are also inputting between 500-750 ng of RNA.

Some times our samples work. Some times they don't (mostly they don't).

We have tried to correlate results between input ng, ratios (260:230 and 260:280), DV200 and RIN. We don't see any correlation over hundreds of samples indicating which QC is the better value to determine which will work.

We recently submitted 144 RNA samples of which 72 were processed but only 4 were successful. (NOTE: these were all prepared by the Qiagen AllPrep kits.)

We have the option of adding more cycles of amplification but we are concerned about duplication.

Any thoughts on how we can improve our library yields? I know that the CORE has been communicating with Illumina, but if there are groups out there, working with FFPE material and have tricks they are willing to share, I would greatly appreciate it.

Thanks.

[jteeee2]

Generating RNA-Seq libraries from FFPE RNA can be quite frustrating, so I definitely understand where you are coming from.

We haven't been very pleased with the RNA we get out of the Qiagen extraction kits, so I definitely think your decision to go the RNA/DNA Storm route is a good one. We tested their kits head to head with Qiagen and others and they were the clear winners. The Ambion RecoverAll is also a decent choice.

With regard to the RNA Seq library failures, I can tell you from a ton of experience that you should probably be closer to 14 PCR cycles if your input is 500-750 ng. Yes, you will see higher duplication rates, but it's worth it if the other option is complete failure.

What kind of DV200 values are you working with? With the TruSeq Total RNA chemistry, we've generated successful libraries from material with DV200's <30%.

Are you fragmenting the RNA after ribo-depletion? If so, I would recommend skipping this step entirely and proceed to 1st strand synthesis after adding the FPE buffer.

Also, when you say that the libraries are "failing," does that mean you aren't getting any detectable yields at all? Could you attach some bioanalyzer traces of successful/unsuccessful libraries?

Another thing you may also consider is doing some qPCR analysis on your starting RNA to assess the "amplifiability" of your material. It's possible that your crosslink/modification reversal was less than stellar and your RNA is very modified, inhibiting the RT enzyme's ability to polymerize the 1st stand cDNA molecule. There are a couple of papers out there describing how to use qPCR to assess FFPE RNA quality. Here is a link to one....https://www.ncbi.nlm.nih.gov/pubmed/27077042

It seems like your success rate should be way higher than 4/72. We've never had a library failure with FFPE RNA, even with really cruddy stuff. You might also ask the core what kind of bead cleanup they are doing after the ribodepletion step. If they are using the standard bead/sample ratio, then are probably losing a lot of the RNA since it's fragmented. You should be in the 3.8-4X range for this bead cleanup (Illumina also mentions this in their protocol).

Best of luck to you

[kmcarr]

For any low quality or low quantity RNA samples our core does not use TruSeq. We have had good results with the RNA-Seq library kits from NuGEN (http://www.nugen.com/products/rna-seq). For the sample you describe we would use the NuGEN Ovation Universal RNA-Seq System with Human FFPE AnyDeplete (formerly In-DAC) target probes (part # 0340-32 or 0341-32).

[SueFoltin]

Thank you for all your suggestions. I am attaching several files. The "Final RNA run successful" excel sheet was our previous run where all the samples worked using 12 cycles of PCR, no fragmentation, bead ratio 1.8. The other excel spreadsheet contains the QC for our 144 pending samples (4 out of 72 yielding enough material). I've also included the BA files we received on those 72 samples. The products look like the correct size, but just lack yield. I have emailed Illumina for help and I got the response "The input for TruSeq Stranded RNA with ribo zero calls for good quality RNA, RINs greater than 8 and inputs of 0.1 to 1ug of total RNA. While it is really great that 48 of your FFPE derived RNAs made it through library prep, we simply can’t guarantee the kit will work for low quality samples. I am not sure how we can help with the issues you have encountered." Argh!

We are considering taking the existing samples and running another 4 cycles of PCR. However, we do not want to purchase whole kits just to do PCR.

Does anyone have recommendations for primers / taq we can use to achieve this?

I really appreciate the help.
Sue

[jteeee2]

That sounds like something Illumina would say. However, they do market these kits as being appropriate for FFPE material.

In the spreadsheets you attached, are those RNA concentrations nanodrop readings or were they measured by Qubit?

Do you have a sense of how much DNA is contaminating your preps?

[SueFoltin]

The ng/uL are based on Qubit while the ratios were determined with the Nanodrop. I don't think I've really tested to see if there is contaminating DNA as we do perform the DNase I step and perform extra wash cycles.

I'm searching now for primers / Taq. It is very frustrating that they do promote these to work with FFPE material, yet deny the quality of most isolated material. I can get those conditions for RIN, DV200, etc if I have fresh, frozen tissue to start with but aged FFPE not so much.

[jteeee2]

Glad to hear these were Qubit measurements. I was initially concerned that they were Nanodrop readings and you were actually putting far less RNA into the library prep than you thought.

A quick google search for the Illumina P5 and P7 primers will lead you to their sequences. I'd recommend the Kapa HotStart ReadyMix (KK2602). The following primer sequences will work as well.

FWD: 5' AAT GAT ACG GCG ACC ACC GAG ATC TAC AC 3'
REV: 5' CAA GCA GAA GAC GGC ATA CGA GAT 3'

[jteeee2]

Also, did the core run any other samples alongside yours? Did they include a batch control to verify the reagents were good? The fact that all of your samples performed so poorly makes me wonder if something happened during library prep.

[SueFoltin]

I don't think they run controls, unfortunately. We wondered about that too. I think we are going to request the library material be returned to us and we'll proceed with the extra 3-4 cycles of PCR. Fingers crossed that will work. If it does, I feel like I owe you a steak dinner at a minimum. LOL

[jteeee2]

I hope it works out for you. Regardless of outcome, I'd consider a using a different NGS provider . Batch controls are incredibly important and often the only way to troubleshoot sample success/failure.

[ajeffs]

Illumina have internally flagged a batch of RNA Adapter Plates because of reports of low or no library yield, lot number 20157476. It might pay to check this lot number with the Core before ruling out a possible problem with the library prep reagents.

[Olaf Blue]

Perhaps Illumina/Epicentre's ScriptSeq RNA-Seq?

http://www.epibio.com/docs/default-source/scientific-posters/2012-agbt---rapid-single-tube-workflow-for-preparing-directional-rna-seq-libraries-from-low-input-and-ffpe-rna-samples.pdf?sfvrsn=10

[SueFoltin]

Thanks for the heads up. I'll definitely forward this information to our sequencing CORE facility.

[SueFoltin]

jteeee2 - thank you for the primer and Taq information. I just spoke with the technicians at the CORE again. It seems that they hold back some of the cDNA with adaptors, so all I need are the primers / Taq to work at the next step. I know I should be more familiar with the details of the Illumina kit protocol, but are these the primers for that step? When I mentioned P5 and P7 to the techs, they weren't exactly sure what I was talking about. I have a goal this weekend to finally sit down and read end to end the detailed protocol.

[jteeee2]

For FFPE samples, none of the adaptor-ligated cDNA should be "reserved." All of it should move into the final PCR step. That's my opinion, and I'd be interested to hear what others thought. The P5 and P7 nomenclature refers to the portions of the Illumina adaptors that will end up binding to the flow cell. Primers targeted to these regions allow amplification of the entire library molecule.