DNase treatment and fragmentation of RNA

[pig_raffles]

Hi all,

I have recently started RNA extraction on different fish tissues with the aim of carrying out RNA-seq using Illumina Hi-seq.

Following a Trizol extraction protocol, I performed DNase treatment using the Ambion Turbo DNA-free kit, followed by RNeasy spin-column purification.

When the samples were run on the Bioanalyzer, RNA concentrations were quite high, yet the traces generally showed high baselines indicating sample RNA degradation (see attached image)

Whilst this may be due to a number of reasons, one of my colleagues suggested that DNase treatment itself could cause odd readings on the bioanalyzer and potentially cause RNA degradation. i was therefore wondering if anyone else had experienced similar problems with DNase treatment and total RNA quality, especially using the Ambion kit, and had any words of advice regarding its use and possible consequences

Many thanks in advance

[pmiguel]

What you probably should do is give it another shot but save an aliquot of your RNA pre-DNAse treatment to run along side your post-DNAse treated RNA. I doubt Ambion would have RNAse issues with their reagents, but who knows?

That said, most likely your RNA was already degraded prior to DNAse treatment. If not, it is also possible that you have contaminating RNAses in your RNA prep that kick in during the DNAse treatment.

--
Phillip

[NormSci]

Bypass DNase treatment if it is degrading your RNA

I have had similar problems with RNA extractions from gill tissue. Prior to DNase treatment, BioAnalyzer RIN values of my samples were typically > 9, and BioRad Experion RQI values > 8.

However, after DNase I treatment, using either the Invitrogen Amplification Grade DNase I, or the Ambion DNA-Free kit, the RNA was significantly degraded (RIN/RQI < 8, often < 6, many < 4).

Though I was never able to resolve the issue, I realized that the DNase step was not absolutely necessary, but was instead only recommended by the manufacturer (my samples were also destined for RNA-Seq on an Illumina HiSeq2000), and thus could be bypassed. Samples are typically "DNased" to aid in sample concentration normalization. However, DNA contamination within the samples is removed during the mRNA isolation step.

Assuming you want to isolate the mRNA, what you could do is DNase an aliquot of every sample and quantify it. Though the RNA will be degraded, DNA will be removed and thus the remaining concentration will represent the purified RNA. Then all you have to do is compare the difference in concentrations between the "DNased" samples and the "non-Dnased" samples (which you must also quantify). This should give a rough idea as to how much DNA is contaminating each sample, and thus how much extra "non-DNased" sample to use for mRNA library construction.

[betacamp]

Dear Norm,
Great explanation!!!
I have the same problems with my RNA sample tissue of my problematic plant....
I need all RNA preparation for the best quality without dnase treatment.... because if I use Dnase in my Rna samples all degrading...
It is a very hard work for us.
I tested many RNA kit plant prococols like Ambion,invitrogen and Qiagen..... but the best shot is home made plant protocol starting with CTAB always ....
with CTAB home made protocol I had a RIN 6 to 8 agilent traces without dnase....

[betacamp]

RNA Extraction Kits Never more...

Dear Pig_raffles,

Following...

This is my best RNA Agilent traces with homemade CTAB protocol without RNA pre-DNAse treatment.
I dont use never more RNA Extraction Kits (like Ambion or Invitrogen), because its dont work for my RNA Plant preparation samples...
With this RNA CTAB protocol homemade I get on average between 6 to 8 RNA extractions, only ONE with this best RIN result on Bioanalyser, because if I make a RNA pre-treatment DNASE, my RNA plant is completely degraded. Already many protocols tested by treatment with DNASE and any work for my sample RNA plant.

Best wishes,

Beta.

http://s1244.photobucket.com/albums/gg580/bertars/?action=view&current=RINMODELARACMYRNA-SEQ.png

[pmiguel]

Listen, you can attempt to convince yourselves that DNAse treatment is not necessary all you want, but that does change the fact that failing to do it can result in DNA sequence contamination of your RNAseq results.

Also, if your RNA is degrading during DNAse treatment, the simple explanation is that your preps still have RNases in them. That is, your RNA prep has been unsuccessful at completely removing them. So, even if you skip the DNAse step, to spare your RNA, all you accomplish is a circumvention of the QC step -- your RNA will likely degrade during a later step.

A "standard paradigm protocol" would have you isolate intact RNA free of contaminating RNases and depleted of most DNA. Then DNase treat to remove the remaining DNA, followed by some purification to remove the DNase and oligonucleotides. (I like Zymo columns for this purpose.)

No one likes to see their RNA degrade during DNase treatment, but better it happens there than at a later step where library QC may not catch it.

--
Phillip

[NormSci]

Naturally, contamination with RNases is the first thing one would suspect. To test this I ran control samples (i.e., samples where RNase free H2O was added in place of the requisite reagents required by the DNase kit) alongside DNase-treated samples.

The results showed that the control samples had almost no degradation (RIN ~ 9.6), while the DNase treated samples were highly degraded (RIN < 6). Now, correct me if I'm wrong, but if there were any residual RNases in the sample, post-extraction and pre-DNase treatment, would the control samples not also be degraded?

These results tell me that something in the DNase kit, or something that occurs during the DNase treatment (which could, I'll admit, be environmental), must be the source of degradation, and that the samples likely do not contain a significant amount of residual RNases left over from the extractions.

[betacamp]

Dear pmiguel,

I'm not trying to convince anyone do NOT use RNA DNASE treatment ...

I'm exposing the discussion of my case and my own experience with my specific type samples plant.

I´m already worked with Biomol to many kinds of samples like fungi, plants, bacteria and protozoa, and I know full well that each type has its typical extraction of RNA!!! An extraction of RNA that fits very well for a sample is not for another.

Is there a sample on top of my RNA trace Agilent to prove it, they are great my extractions and RNA-seq already made them and no problems Assembly of libraries.

Dear pig_raffles

I agree with everything you said and with the tests that you did.

All the best.

[pmiguel]

To reiterate:

(1) Any residual DNA in your RNA prep may be sequenced during RNA seq. So it is unwise to skip the DNAse step.

(2) If your RNA prep is degrading during DNAse treatment, that is a problem with a large number of possible causes. I suggested one previously. But whatever the cause, that issue needs to be dealt with so that DNAse treatment can be undertaken. Just skipping the DNAse treatment is not a legitimate solution in my opinion.

Note that having contaminating DNA sequence in your RNA seq will not necessarily interfere with assembly, so that is not an acceptable test for DNA contamination of your sequence data set.

Obviously you are entitled to your opinions on this subject. But I think I see some wishful thinking therein -- so I am pointing it out.
--
Phillip

[chadn737]

There is a lot of potential sources of contamination. Maybe its the DNase. Maybe....its the RNeasy kits you use afterwards.....especially if you contaminated the water/buffer you resuspend your RNA in.

I would question any RNA-seq result that did not DNase treat.

[NormSci]

Unless I am missing something, wouldn't the isolation of 3' tail poly-A mRNA (if that is one's goal) filter out any residual DNA contamination? If not, then somebody better tell Illumina to change their TruSeq manual.

I suppose that any DNA sequence that is complimentary to the oligo-dT primer sequence could be included in the poly-A mRNA library. But I expect that these sequences would be relatively rare compared to transcript abundance, and further, that such sequences would most likely be filtered out during pre-processing and QC steps, where rare sequences (i.e., sequencing errors) are removed. Thus, it seems to me that that the effect of putative DNA poly-A hitchhikers on any assessment of differential expression would be negligable.

Lastly, it is my understanding that treatment with DNase does not remove all contaminating DNA, but only reduces it, at best.

[chadn737]

You don't think poly A repeats are common in genomes?

[NormSci]

My apologies, perhaps I wasn't clear enough.

Poly-A repeats matching the oligo-dT primer, that have identical contiguous sequences spanning the entire read length (say, 100 bp), and that occur in the genome at a level that comes close to the abundance of expressed transcripts? No, I think such sequences would be RELATIVELY rare. Rare enough that they are more likely not to be sequenced, and for those that are, rare enough that they would be filtered out as sequencing errors. At most, such sequences would exist as repeats in the dataset only to the extent that they are repeated in the genome, and I would think that these sequences would be significantly less common than expressed transcripts.

[chadn737]

Consider that after two rounds of polyA selection, there is always a non-Insignificant number of reads from rRNA and tRNA. PolyA selection is imperfect. Primers mis-prime and you do not need 100% match to capture something. Intergenic and intronic regions are rich in A/T repeats. And since these are genomic sequences, they will not show up as sequencing errors, but instead align back to the genome. Some will be intergenic and intronic, but how do you know this is DNA contamination as opposed to alternative splicing and unannotated or noncoding transcripts or even just spurious transcription?

Of course all of this is assuming that 1) you have a reference genome and 2) you use polyA selection. What if you have to do a denovo transcriptome? What if you use a rRNA depletion method rather than polyA selection?

Its your data, but I would not waste my time or money without DNase treating.

[pmiguel]

The oligo dT step is a selection against DNA. But remember that is "oligo" dT -- you just need enough oligo dA to hybridize to the dT and you are off and running.

You are right that the DNAse treatment will not reduce the amount of DNA to zero -- it is important to use an RNA prep method that also fractionates the RNA away from the DNA. Acid phenol (TRIzol), lithium chloride precipitations, etc. are often used.

Purification schemes of this sort rely on one step depleting contaminants (in this case DNA) then another depleting them again -- hopefully with multiplicative effect.

I don't know, NormSci, you seem to know what you are doing. To tell you the truth, I am not really concerned for your experiment. I am a little concerned about the hapless grad student whose total RNA degrades the first time they do a DNase treatment, then finds your post via Google. They may feel like they now have a good reason to just skip the step that likely is just showing them they have failed to sufficiently purify their RNA.

On the other hand, that hapless grad student may have a ton of DNA contaminating their sample for whatever reason -- in which case the DNAse treatment may only be sufficient to chop it into fragments of a perfect size to be blunted and adapter-ligated. In that case maybe they are better off skipping the DNAse treatment.

Oh, and what do we really know about this DNAse anyway? How sure are we that it does only degrade RNA? How sure are we that RNA never contains deoxynucleotides? Maybe this is common in your particular organism. Face it, we are largely stumbling around in the dark. Advice we give in forums of this sort might help a few people from falling into some particularly large pits. But, as they say, your mileage may vary.


--
Phillip