Actual question

I have reason to believe (details see below) that in a ligation I carried out, an EcoRI sticky end (EcoRI: G'AATT_C) and an XmaI sticky end (XmaI: C'CCGG_G) were somehow ligated together, a process during which at least the EcoRI site was lost: the plasmid had a BamHI site very near the XmaI site, and a BamHI/EcoRI digest produced only one band compared to three bands in the undigested control (three conformations of circular plasmid).

How could ends as incompatible as EcoRI and XmaI be ligated?

Detailed Background or: Why I believe incompatible sites were ligated

This is chronologically further to: What are common causes of unexpected ligation products?. Briefly: I digested two plasmids, one with EcoRI and XmaI (p1), the other with EcoRI and AgeI (p2) [XmaI and AgeI produce compatible sticky ends], then carried out a ligation between a 1.4kb insert isolated from p2 (structure EcoRI-1.4kb-AgeI) into the 3.4kb backbone isolated from p1 (structure XmaI-20bp-BamHI-3.4kb-EcoRI). After transformation, I tested the ligation product by digest with EcoRI and BamHI, which should produce again 3.4kb + 1.4kb. The gel was poor quality but allowed the conclusion that there were different ligation products.

Repeating the EcoRI/BamHI digest and gel more carefully showed there were only two variants: Variant A (8 colonies) produced the two expected bands 3.4kb and 1.4kb. Variant B (4 colonies) only produced a clear 3.4kb band, nothing else visible in the lane: in other words, it only carried either an EcoRI or a BamHI site and was in total smaller than variant A (confirmed by undigested controls). Thus, I conclude that B was simply the same 3.4kb backbone, re-ligated without the 1.4kb insert. Since the BamHI site was untouched in the backbone during the preparatory EcoRI/XmaI digestion, I assume that the EcoRI site was not recovered during the ligation of variant B.

(The ligation was carried out overnight at 16 deg C, using Promega T4 DNA ligase and buffer and a molar ratio of 3:1 insert:vector at a total volume of 20uL. The E.coli for transformation were Invitrogen OneShot Stbl3 and have been routinely and successfully used in the lab for a long time. The analytical digestion used Promega EcoRI and BamHI in Buffer Multi-Core, which Promega claims offers 75-100% efficiency for both enzymes. Incubation was 1.5h at 37°C.)

  • $\begingroup$ incomplete digestion is one factor. $\endgroup$ – WYSIWYG Aug 16 '13 at 14:25
  • $\begingroup$ Incomplete digestion at what point do you mean? If the backbone (p1) was incompletely digested before the ligation, it should have produced a larger band than the one which I excised from the gel. If the analytical digest after ligation was incomplete, I would expect to see at least faint bands in addition to the one clear band corresponding to the backbone size. $\endgroup$ – Armatus Aug 19 '13 at 12:58
  • $\begingroup$ well.. if you can provide the following details then it will be easier to troubleshoot: 1. The distance between EcoRI and XmaI site, 2. How was the double digest performed $\endgroup$ – WYSIWYG Aug 21 '13 at 3:19
  • $\begingroup$ Thanks for the hint. I added those and rewrote the description a bit, hopefully that's clearer now. $\endgroup$ – Armatus Aug 21 '13 at 18:57
  • $\begingroup$ So in p1 the dropout after restriction will be 3.4 kB; A 3.4 kB difference will clearly help to differentiate between the single digest and double digests. So it seems it is not a problem of incomplete digestion. However, EcoRI and XmaI are not really compatible in a double digest experiment and sequential digest is advised. But if you anyways got the fragment of right size then it means double digest happened. How self ligation happened despite incompatible ends; perhaps due to a rare event of end joining post transformation. Just a blind guess.Anyways what was the self ligation frequency? $\endgroup$ – WYSIWYG Aug 22 '13 at 8:34

The answer from @Armatus got me looking at star activities again. EcoRI is prone to exhibit star activity in non-optimal buffers. In the case of this enzyme this tends to be a relaxation of site specificity, so HAATTC and GAATTD would be possible cut sites. The cleavage at such sites produces the standard sticky end.

So, what if there is a star site near to the XmaI site such that a digest at the true EcoRI site and also at the star site removes the XmaI site and so allows recircularisation. If the configuration was, for example:


the ligated product would have neither EcoRI nor XmnI sites.

  • $\begingroup$ Sounds like a valid explanation and is probably not even too unlikely. Unfortunately I haven't got access to the plasmid sequences anymore so I can't verify it :( $\endgroup$ – Armatus Nov 20 '13 at 19:37

My best guess to explain this phenomenon has been spontaneous degradation of the sticky ends, followed by blunt-ended ligation.

Alternatively, single-stranded exonuclease activity either from a contaminating enzyme or from EcoRI/XmaI/AgeI could be imaginable. For example, using buffer Multi-Core is discouraged for EcoRI due to potential STAR activity - maybe it developed exonuclease activity and some plasmids ended up blunt-ended before the ligation.


I want to add an extra explanation. By personal experience I can tell that incompatible sticky-ends can ligate in some conditions. The efficiency will be lower than using compatible ends but still, some molecule will ligate if the single-stranded ends can anneal just enough for the ligase to join the DNA backbone. This can especially happen at lower temperatures where the miss-annealing of the sticky ends is more stable than at higher temperatures.

The OP states:

The ligation was carried out overnight at 16 deg C.

This condition is enough to allow incompatible sticky ends to anneal often enough to get ligated. I propose the OP to run a comparative experiment to check the efficiency of incompatible-ends ligation by running one ligation at 37C for 15min and one at 16C overnight. I expect almost no background ligation to occur in the first case.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.