various SMART cDNA methods for transcriptome sequencing

[krelmer]

Hello all,

I am planning to do some cDNA sequencing on our new Titanium. As this is our first in-house run, I will be getting the RNA synthesized to cDNA and a normalized library made by a service provider. I am looking into Evrogen, and they offer
(A): SMART Oligo II oligonucleotide 5’-AAGCAGTGGTATCAACGCAGAGTACGCrGrGrG-3’
CDS primer 5'- AAGCAGTGGTATCAACGCAGAGTTTTTGTTTTTTTCTTTTTTTTTTVN -3' .

or (B) the GsuI option:
SMART Oligo II oligonucleotide 5’-AAGCAGTGGTATCAACGCAGAGTACGCrGrGrG-3’
SmartGsu T20 5’-AAGCAGTGGTATCAACGCAGAGTA(CTGGAG)d(T)20VN-3’

These are the two methods they suggest to reduce the effect of poly-A. Does anyone have any experience to share with either of these preparations?

Evrogen also says their normalization libraries are untested with Titanium. Has anyone used them with success or failure?

For our last transcriptome run was done on a similar starting material and we also used a service provider. They used an unpublished 3' extension approach for cDNA prep & normalization. We won't use that provider again so need to try a new method. Previously the service provider also did the sequencing on GS FLX standard and we were fairly pleased with the results. Now hoping to expand the project with more data ...

Any suggestions would be very much appreciated!

Thanks,

Kathryn

[kmcarr]

Kathryn,

Our core has run many, many cDNA samples on 454 (GS20, GS FLX, and GS FLX Titanium). Almost all were prepared using Clontech SMART kits, some normalized using the Evrogen Trimmer kits which integrate with the SMART kits and a couple prepared by Evrogen. We recommend to our submitters that they use a modified CDS primer for the SMART protocol (like method A above). In our experience the results are generally good. Comparing cDNA samples to genomic we do find a higher percentage of reads rejected as mixed, and the length distribution skews shorter. We haven't had any experience with a library prepared according to method B.

[aarodriguezm]

Hi, Normalization with evrogen depends very strongly to the amount of cDNA material you are starting.
I have done a lote of normalized libraries and the big problem is do to the 3´ bias you introduce with current cDNA synthesis kits. any ideas how to resolve this question?
best regards
A.

[pmiguel]

Hi, Normalization with evrogen depends very strongly to the amount of cDNA material you are starting.
I have done a lote of normalized libraries and the big problem is do to the 3´ bias you introduce with current cDNA synthesis kits. any ideas how to resolve this question?
best regards
A.

Use random primed instead of poly-A primed cDNA.

Roche has just released a cDNA library method. (Register on 454.com/my454 to download it.) It uses ZnCl2/heat to fragment RNA so that subsequent hexamer primed reverse transcription will produce cDNA in the desired size range.

You will want to deplete your input RNA of rRNA first, most likely. Or can normalization reduce rRNA copies to be in line with all other RNAs?

--
Phillip

[aarodriguezm]

Normalization

Hi Phillip
Thanks for your rapid answer.
I think this could be suitable for modell organisms depleting with rRNA specific oligos to known rRNAs. Do to the variable efficiency of the normalization, it is very dificult to determinate how much are you able to normalize rRNA. Even if you do the poly A+ selection before the cDNA sinthesys and then normalize the library you will obtain a lot of rRNA contamination reducing the chance of sequence rare transcripts.
What do you mean? Any other ideas?
B.
Alejandro

[pmiguel]

Hi Phillip
Thanks for your rapid answer.
I think this could be suitable for modell organisms depleting with rRNA specific oligos to known rRNAs. Do to the variable efficiency of the normalization, it is very dificult to determinate how much are you able to normalize rRNA. Even if you do the poly A+ selection before the cDNA sinthesys and then normalize the library you will obtain a lot of rRNA contamination reducing the chance of sequence rare transcripts.
What do you mean? Any other ideas?
B.
Alejandro

Yes that is what I meant.

If you use a ribominus-like method to remove rRNA and a random-prime method to generate cDNAs then I would imagine that end bias would be minimal.

Other ideas for obtaining the sequence of rare transcripts:

(1) Two major reasons for a transcript to have a low relative abundance in your RNA prep.

(a)the gene has low expression in all cells of the tissue from which RNA is isolated.

(b)the gene has medium-high expression, but only in a small fraction of the cells from which the RNA is isolated.

If (b), then you could obtain sequence of rare transcripts by fractionating your cells somehow (eg FACS, microdissection, etc) prior to isolating RNA.

(2) Targeted subtraction. If you have one, or a few, genes whose expression is swamping that of other genes, you could remove them in the same way the rRNA is removed. Create biotinylated oligos that will bind the transcripts that would like to deplete your sample of. Hybridize. Pull out those transcripts with SA-beads. The transcripts remaining in solution are your depleted sample.

(3) Instead of pulling out poly-adenylated messages, pull out messages with a 5' cap using an antibody-mediated method. Whether this is suitable for your sample would depend on many factors, including the organism from which your RNA derives. Also, this would produce a 5' bias.

(4) Use a "full length" RNA kit to isolate RNA. This will, however, bias against messages that are being turned over at a high rate.

--
Phillip

[linda]

Hi
I have just run some normalized cdna library on the 454 using titanium reagents
These cdna libraries were made using pooled mRNA from the plant tissues. First strand cDNA synthesis was performed using Evrogen Mint cDNA synthesis kit (SK002) using primers

PlugOligo_1 adapter 5'_AAGCAGTGGTATCAACGCAGAGTACGGGGG_P_3'

CDS_1 adapter 5'_AAGCAGTGGTATCAACGCAGAGTAC(T)30VN _3'

(RsaI site underlined)

Subsequent PCR amplification was performed using only primer

PCR Primer M1 5'_AAGCAGTGGTATCAACGCAGAGT_3'



200ng of cDNA (qubit quantified!) was normalised using duplex-specific nuclease from Evrogen (EA001) and amplified with PCR primer M1 as above.

After the run the reads (ptp 2 regions)were quite short (250) and the distribution was quite flat (any particular peak) and the seqeuncing results was quite low ( 80MB)
I tried to titrate it using different amount of library in the emPCR but I'll get back alwys the some amount of enriched beads .

I run the cdna on an agarose gel after the normalization procedure and the smear in the range 300ng -16 kb

Do you have any idea?
Could be because I used the mint kit instead of the smart one?

[pmiguel]

Hi
I have just run some normalized cdna library on the 454 using titanium reagents
These cdna libraries were made using pooled mRNA from the plant tissues. First strand cDNA synthesis was performed using Evrogen Mint cDNA synthesis kit (SK002) using primers

PlugOligo_1 adapter 5'_AAGCAGTGGTATCAACGCAGAGTACGGGGG_P_3'

CDS_1 adapter 5'_AAGCAGTGGTATCAACGCAGAGTAC(T)30VN _3'

(RsaI site underlined)

Subsequent PCR amplification was performed using only primer

PCR Primer M1 5'_AAGCAGTGGTATCAACGCAGAGT_3'



200ng of cDNA (qubit quantified!) was normalised using duplex-specific nuclease from Evrogen (EA001) and amplified with PCR primer M1 as above.

After the run the reads (ptp 2 regions)were quite short (250) and the distribution was quite flat (any particular peak) and the seqeuncing results was quite low ( 80MB)
I tried to titrate it using different amount of library in the emPCR but I'll get back alwys the some amount of enriched beads .

I run the cdna on an agarose gel after the normalization procedure and the smear in the range 300ng -16 kb

Do you have any idea?
Could be because I used the mint kit instead of the smart one?

My guess is that your long oligo dT (30T) primer is the problem. Especially for amplicons that have the oligo dT adjacent to the PA adaptor. Those will blaze in concert, lowering the quality of all the beads near them.

If you had a way to remove that adaptor from your cDNAs prior to adding Titanium adaptors, you would probably get better results. Alternatively, if you could force the 30T side to the PB adaptor side, that would also help.

Ultimately, I think random primed cDNA libraries are the way to go for the 454. Roche now offers a protocol/kit (see 454.com/my454) for random prime cDNA library construction. I'm unaware of any normalized cDNA library protocols base on this methodology, though.

--
Phillip

[linda]

Hi Philip
Thanks for the quick reply
Because we are using the trimmer to normalize the cdna
Do you think that the SMART KIT with
Oligo II oligonucleotide 5’-AAGCAGTGGTATCAACGCAGAGTACGCrGrGrG-3’
CDS primer 5'- AAGCAGTGGTATCAACGCAGAGTTTTTGTTTTTTTCTTTTTTTTTTVN -3'
can get better results?
There is a RsaI site on the PlugOligo_1 adapter and CDS_1 adapter ( primer from the mint kit) do you think that digesting before adding the adaptor can help?
thanks

[pmiguel]

Hi Philip
Thanks for the quick reply
Because we are using the trimmer to normalize the cdna
Do you think that the SMART KIT with
Oligo II oligonucleotide 5’-AAGCAGTGGTATCAACGCAGAGTACGCrGrGrG-3’
CDS primer 5'- AAGCAGTGGTATCAACGCAGAGTTTTTGTTTTTTTCTTTTTTTTTTVN -3'
can get better results?

Possibly. But from Sanger sequencing amplicons resulting from similarly constructed cDNAs, I've seen that even the "V" anchoring of the above oligo dT does not promise that the oligo dT primer will not prime mid-polyA. So, you still may end up with long polyT tracts near the beginning of your sequence reads.

I've seen protocols that "force clone" the 3' ends of cDNAs away from the sequence priming site. If you like oligo dT methods, I would recommend one of those. (In brief, the one I've seen uses only T4 polymerase to blunt after nebulization--no T4PNK. If you add only PA adaptor to the ligation it forces ligation to the broken end because the cDNA adaptor end is not phosphorylated. The PB adaptor is added subsequently added by "step out" PCR, using PA and PB+SMART-CAP primers.)

There is a RsaI site on the PlugOligo_1 adapter and CDS_1 adapter ( primer from the mint kit) do you think that digesting before adding the adaptor can help?
thanks

No, you would need the restriction enzyme to cut internal to the polyA tract. So you would need a typeIIS or typeIII restriction site to pull that off. EcoP15I, for example. Not worth it in my opinion. Easier to do random prime.

--
Phillip