help needed with three fragment ligation

Frank O. Fackelmayer Frank.Fackelmayer at
Thu May 11 05:09:09 EST 2000

I agree with Ned that a three fragment ligation is not a problem
usually. Of course, it depends on what you want to clone, and as always
some framgents simply don´t clone well. If all fragment ends are
non-identical, there is a good chance to get the right clone by just
putting in all three fragments, ligase buffer and ligase. It may be
necessary to check more minipreps than in a conventional two-fragment
ligation, though. We routinely do 24 to 36 preps. 
Good luck,

Ned Mantei wrote:
> In article <8fbvei$f6a$1 at>, iayork at (Ian A.
> York) wrote:
> >In article <39197CD7.21B1B328 at>,
> >Chris LaRosa  <clarosa at> wrote:
> >>Colin Herd wrote:
> >>>
> >>>     I have a puzzle with regard to a ligation I am trying to perform. I
> >>> want to ligate three pieces of DNA together, one of 7.1kb, one of 3.6kb
> >>>
> >>Can you break it down to two sets of ligations....1 to 1 , then 1 to
> >>one..?
> >
> >Agreed.  Three-fragment ligations sometimes work, but more often they
> >don't; it's not worth fooling around with it, if it doesn't work after the
> >first couple tries.
> I disagree strongly. I have many times ligated up to five fragments with
> success (success being defined as finding the right plasmid after doing
> 18 mini-preps). This even worked when one pair of ends was blunt.
> Three-way ligations should be no problem at all, unless more than one
> pair of ends is blunt. My procedure would include:
> 1) 5-10 femtomol of each fragment (a double-stranded oligo can be
> treated as a normal fragment) in a 10 ul reaction volume. Use 1 ul of
> ligase. Generally fewer fmols of larger fragments--see Nucleic Acids
> Res., 16:10301, 1988..
> 2) I generally use "classical" competent bacteria treated with only
> CaCl2 and frozen in aliquots in CaCl2--15% glycerol. I have the
> impression that some of the other chemical methods indeed give a high
> transformation frequency with picogram amounts of closed circular
> plasmid, but that there is more tendency for abnormal products to
> somehow produce plasmids. The yield of correct plasmid therfore could be
> lower. With "classical" calcium cells, you can transform 5 ul of
> ligation reaction per 125 ul bacteria.
> 3) The vector ends are dephosphorylated with calf intestine or shrimp
> alkaline phosphatase. DNA is then extracted with phenol,precipitated,
> and about 1--2 ug run on an agarose gel to remove possible undigested
> material. The gel might not be necessary in all cases. If you don't run
> the gel, make sure to remove every trace of phenol by washing the
> ethanol pellet twice with 70% ethanol (and aspirate the supernatant
> instead of just pouring it off.) I use low-gelling temperature agarose
> and phenol extraction to work up the excised bands, but take your choice.
> 4) As mentioned in this newsgroup a day or two ago, exposure of DNA to
> even medium wavelength (312 nm) UV light when excising bands from gels
> will reduce infectivity by some 99% in only a minute or two. A UV box
> emitting at 360 nm is really a good investment.
> 5) If one of the fragments is derived from another plasmid, residual
> plasmid from that fragment isolation is often a problem. The vector part
> of this plasmid can be cut with other restriction enzymes that don't cut
> in the desired fragment, so that the vector part can't recircularize. If
> one of the fragments is a PCR product derived from amplification of
> plasmid sequences, the plasmid used as template can also be carried over
> and give background problems.
> 6) Finally, there can be seemingly harmless combinations of fragments
> that just won't give a plasmid. Presumably somehow toxic for the
> bacteria, or not acceptable to the replication machinery, or... So a
> little bit of luck is still necessary.
> --
> Ned Mantei
> Department of Cell Biology, Swiss Federal Institute of Technology
> CH-8093 Zurich, Switzerland

More information about the Methods mailing list