Friday, July 10, 2009

Help request - how does one block some DNA from being PCR amplified?

Dear world:

We need help.  

We would like to use PCR amplification of rRNA genes to characterize rare bacteria in a sample where there are some very very dominant bacteria.  

The problem is that we do not know what those rare bacteria are and would like to use "universal" rRNA PCR primers to amplify the rRNA genes from these organisms. Such universal primers will also amplify the rRNA genes from the dominant organisms.  

If the rare organisms are, like, really rare, almost all the PCR products will be from the dominant organisms.  We would like to obtain sequence data for the rRNA genes from the rare organisms without sequencing 1000s of the known rRNA genes from the dominant organism.  How can we do this?  

I know of attempts to block PCR amplification of specific DNAs and also attempts to digest away PCR products or bind ones to a column to get rid of them.  But I do not know if any of these methods really work.  

Anyone out there know methods that work to do this?

Added - here are the responses on FriendFeed


  1. Design PCR amp conditions to seperate Dom from rare transcripts. Qpcr rxns , Dom transcript will be majority of rxns but rare transcript will come later? In rxns play with threshold/detection limits qPCR. I hope this helps.

  2. We are doing this to discover what the rare things are- we do not know what they are in advance - so we cannot really design a qPCR assay

  3. If you know what the dominant ones are, you might be able to use something along the lines of subtractive hybridization to remove those samples prior to analysis.

  4. You can design PNA (peptide nucleic acids) that would target the dominant species. The PNA will block amplification once bound to its target.

  5. What about just adding a 10x excess of primers targeted to the middle of the amplicon for the dominant species - then most of what you amplify from them will be short and you can easily size separate it from your targets - won't be perfect, but should significantly reduce amplicons from the dominant species

  6. Can you find restriction enzymes that target a couple dominant species? If not, I think subtractive hybridization is your best best. There are kits for it but I don't know anyone who has used them.

  7. How many different dominant organisms are there likely to be and any guess on the number of rare?

  8. Jacques - Did Gary or others ever get that to work?

  9. Keith - in most of the situations where we might want to do this there is one dominant organism (e.g.., Wolbachia when we want other symbionts). As for the # of rare, that we do not know.

  10. Morgan - I think we will try the Restriction approach - seems more likely to work than subtraction.

  11. Hi there,

    I would have suggest subtractive hybridization as Morgan did. Followed by cloning is probably the more powerful technique if you don't know the sequences of your 'target' and 'noise'.

    - PCR everything
    - dilute, denature and partially re-hybridize at whatever Cot you think appropriate (depends of the expected amount of 'rare' stuff)
    - select single stranded DNA (column retention of double stranded)
    - re-PCR the eluted material and clone.

    Over 2-3 assays you should be able to go down to 10E-5 'contaminants'. Practiced that for rare cDNAs , works nicely.

    The RE approach have a drawback: molecular recombination of the resulting fragments during the PCR! Sometimes it creates more 'noise' than the initial one and 'strange' stuff.

  12. The dominant sequence should recombine much faster than the rare sequence (or could be made to with the right conditions). After each PCR stage, dissassociate and add an exonuclease that digests dsDNA. Allow recombination to occur in the presence of the exonuclease. Start each round of PCR before a significant fraction of the rare sequence has recombined (and then digested).

    This will stop working once the recombination rates are on the same order, but by this point the rare sequence should be abundant enough.

    If the sequences are similar enough, there will probably be some partial recombination between rare and dominant sequences. That should produce a signal as lots of partly digested fragments and no surviving rare sequence.

  13. Using a multiplex reaction with general primers coupled with specific primers for the dominant species has worked well for me in the past. As long as the two primer sets produce different size amplicons, you can gel extract the band produced by the more general primers. I achieved a more than 50% reduction in the dominant amplicon using this method.

  14. how about doing the PCR and then run a single-stranded conformation polymorrphism gel to separate variants and cut out the non-dominant bands (or if you have a capillary electrophoresis unit you might be able to use that). SSCP gels can be stained with silver, i believe, or you can label the PCR products. See for example
    or any number of other papers

  15. I think the SSCP or DGGE option is potentially good but I have not seen people do it when there is an enormous overabundance of one type over others ... but will look

  16. i think restriction is worth a try. You can do some controls with samples of just the dominant organism to make sure the conditions completely cut the area of interest. Using an enzyme that cuts in one place to reduce the chances that it also cuts the rare organisms. Better if there are several single cuts enzymes that can be tried individually to increase the chances of cutting the dominant but leaving the rare intact. The hybridization could work but because the search is for rare organisms it would be necessary to remove 100% of the dominant gene which may be difficult. But again, controls with pure strain can help to find the right conditions


  17. Wow, what a great display of feedback! Maybe I should start posting more questions on my blog. ;) One critique of all these methods, is how will we know if amplified/selected for "rare" organisms are truly rare without having all other sequences in the sample? Or are we assuming that the environment that we are sampling has had enough regular 16s sampling to show that a certain species is always abundant and so we can safely remove them?

  18. Why not use a cDNA normalization kit, like the one from Evrogen? Has worked well for us for transcriptome 454 sequencing when we don't want b-actin sequwnced 10,000 times, or somwthing.

  19. Hey Jonathan,

    You know about our Western English Channel time series. We have been working on several methods as described above and also using ultra deep sequencing. Would be good to speak to you about this.


    Jack Gilbert and Dawn Field

  20. pedro dimitriu7/22/2009 5:33 PM

    How about using ribosomal sequence tag primers in the manner described here?