Library protocol for larger fragments (up to 900bp)

[Sandro]

Hello everybody,

For a specific application, I have to sequence fragments of different sizes up to 900bp.

As a first step, I load my DNA sample onto a gel and cut the lane in five pieces, each piece representing a different size bin (in bp about 100-250/250-400/400-550/550-750/750-900).
The DNA is purified from each gel piece separately and used for library prep. We usually use the NEB Next DNA Library Prep kit adjusted for 10 ng of starting material.
In this case I use 10 ng of DNA for the 100-250 bp bin, 20 ng for the 250-400bp bin, etc to have about the same molar amounts of DNA in each reaction. Then I do normal blunting, A-tailing, and ligation of adapters.
After 9 cycles of PCR I measure the DNA concentration by QBit.
For small inserts, we aim at a final amount of about 500 ng. This I easily achieved for the two smallest bins (although in principal as the fragments become bigger, I would like to have proportionally more DNA in ng/ul to achieve similar molar amounts). The middle bin (400-550 bp) does also reach sufficient amounts after 2 more cycles. However, the two largest bins amplify very poorly. An initial problem was the amplification of self-ligated adapters. I seemed to get rid of those by purifying the ligation product twice after the ligation step with low AMPure/DNA ratios.
Now I see some amplified libraries for the two large bins, but only a small amount and after 4 additional cycles. I don't want to amplify too much further since this apparently can induce biases. Increasing primer conc. should not make a difference, and I also don't think that the nucleotides are limiting, since I am not reaching the amounts in ng/ul of the smaller fragments.

Could it be that the enzymatic reactions (blunting, A-tailing, adapter ligation) don't work as well on large fragments (although the concentrations of DNA ends should be constant among the different bins)?
Or do I have to adjust the PCR (I use Phusion polymerase and have adjusted the elongation time to 1 min).
Or does somebody have a protocol for library preps that works also for larger inserts (and is compatible with relatively low input amounts)?

Thank you very much!
Sandro

[kerplunk412]

PCR in general is more efficient for shorter fragments (part of the reason QPCR amplicons are only ~100 bp) so that is probably part of what you are seeing. I think proofreading polymerases are slower than vanilla Taq, so you could try adding 30-60 seconds to your elongation time. You could also try using even more input material for the large sizes. This may introduce a little bit of bias, but likely less than additional PCR cycles.

Another important thing you should keep in mind is that the shorter products will also cluster more efficiently during sequencing, so you should add proportionally more of the longer products to a pool, assuming you are planning to pool prior to sequencing. I don't know exactly what the ratio should be, but probably somewhere in the range of 5-10% extra for each of your bins.

[Sandro]

Thanks for the reply!

Phusion polymerase is generally very fast and processive, 1 min should be ample time to synthesize less than a 1 kb. Nevertheless, using longer elongation times shouldn't hurt. Starting with more input material is a good idea, unfortunately I am bit limited there, especially for the large fragments.

For the clustering you are completely right. In fact, we expect the clustering efficiency to be orders of magnitude more efficient for small fragments compared to the very long ones. Therefore, we will sequence bin 1 and 2 on one lane, 2 and 4 on another, and 5 on a third one. By multiplexing different samples, we will fill the different lanes with somewhat homogeneous fragment sizes.

[nucacidhunter]

After 9 cycles of PCR I measure the DNA concentration by QBit.

I wonder if you have run PCR products in a gel or Chip or quantified by qPCR. There is possibility of higher proportion of ssDNA in large fragments amplicons which Qubit will not quantify.

[kerplunk412]

Thanks for the reply!

Phusion polymerase is generally very fast and processive, 1 min should be ample time to synthesize less than a 1 kb. Nevertheless, using longer elongation times shouldn't hurt. Starting with more input material is a good idea, unfortunately I am bit limited there, especially for the large fragments.

For the clustering you are completely right. In fact, we expect the clustering efficiency to be orders of magnitude more efficient for small fragments compared to the very long ones. Therefore, we will sequence bin 1 and 2 on one lane, 2 and 4 on another, and 5 on a third one. By multiplexing different samples, we will fill the different lanes with somewhat homogeneous fragment sizes.

I should have looked into Phusion before posting, it looks like they indeed only recommend 15-30 seconds per kb. Your plan for splitting the samples into different lanes sounds very smart to me.

One other thing you may want to look into is another polymerase, like Q5 from NEB or HiFi from Kapa. Let us know if/how you solve your problem, I am sure it will be useful to someone else somewhere down the line.