The use of oligos creating multiple mismatches with the genome has revealed valuable information, not only for their subsequent use, but also about replication and mismatch repair. If oligos serve as primers for replication, one might have expected them to transform only when annealed to the lagging strand of replication, which is replicated discontinuously, and not to the leading strand which is presumably replicated in a continuous fashion. However as we have found in yeast 
and others in E. coli
, although transformation is more efficient with oligos targeted to the lagging strand, it occurs with oligos targeted to the leading strand a few fold less in yeast and 30-fold less in E. coli
. Although the efficiency of transformation is greatly reduced by MMR 
, the pattern of oligo incorporation is generally very similar in the presence or absence of MMR and whether the oligo was targeted to the leading or lagging strand of replication ( and ). In all conditions examined, there was a clear asymmetry of retention of ends of the oligo, with the 5′ end being much less likely to survive than the 3′ end, as can be seen in and . These results suggest that the mechanism of incorporation is independent of MMR, is similar on the leading and lagging strands of replication, and that usually all mismatches are recognized and eliminated by MMR, or none are.
One issue that arises from these results is the degree to which the leading strand is replicated in a continuous fashion. The DNA polymerase that replicates the leading strand, Pol ε, has at least in vitro
a processivity that is not any greater than Pol δ, the polymerase that replicates the lagging strand 
. In addition, there is considerable evidence both in E. coli
and in yeast 
that replication on the leading strand can also be discontinuous. Thus it is perhaps not surprising that oligo transformation can occur on the leading strand. A large amount of evidence supports the view that lagging strand synthesis is done by Pol δ and that leading strand replication is initially carried out by Pol ε 
. A recent model proposes that synthesis after any interruption on the leading strand is completed by Pol δ 
. In that context it would be extremely interesting to know which polymerase was responsible for elongation of oligos targeted to the leading strand.
The central core of the oligo was usually incorporated, but the 5′ end was rarely incorporated and about 1/3 of the time, 10 or more nucleotides on the 3′ end were not incorporated. The loss of nucleotides from the two ends appears to occur by fundamentally different mechanisms. Phosphorothioate linkages on the 5′ end of the oligo make no difference in the pattern of loss of the 5′ end, suggesting that the nucleotides are not removed exonucleolytically, or that the enzymes involved are not affected by the altered linkages. There is considerable evidence that phosphorothioate linkages do protect against a number of exonucleases 
, and so the lack of effect suggests that the 5′ end loss is not exonucleolytic. If the oligos serve as primers for replication, then ultimately the DNA primed by the oligo would have to be joined to DNA synthesized upstream as in normal Okazaki fragment maturation 
. Thus the loss of the 5′ end sequences could be due to the formation of a flap at the 5′ end with subsequent excision of the flap by Rad27 
; the increase in 5′ end sequences in a rad27
strain indicates a role for Rad27 in flap excision of oligo sequences (). There are alternate pathways for fragment maturation not involving Rad27 
; the combination of those pathways is likely responsible for the major loss of oligo 5′ sequences, on both the leading and lagging strands. Although the leading strand of replication is generally replicated in a continuous manner, these results also indicate that new priming events on the leading strand are processed similarly to those on the lagging strand. It may be indicative of some difference between the two strands that on the leading strand the absence of Rad27 appears to effect the incorporation of some internal nucleotides in a manner different from that observed on the lagging strand (, compare incorporation of nucleotides 12 and 15 from the 5′ end in rad27 msh6
F and R strains). A surprise is that although phosphorothioate linkages on the 5′ end do not make any appreciable difference in retention of the 5′ end, they do appear to have some affect on retention of the 3′ end (for the nucleotide at position 33, 89% retention versus 79% in the R orientation, and 80% vs. 71% in the F orientation) (). If the 5′ end were lost solely due to an endonucleolytic flap cleavage 5–10 bases from the 5′ end, one would not expect phosphorothioate linkages in the 5′-most 4 nucleotides to have any effect on maintenance of 3′ end sequences.
Comparison of retention of oligo N sequences with (5′P) and without phosphorothioate linkages at the 5′ end.
Phosphorothioate linkages do offer some protection on the 3′ end in wild-type, but little if any in msh6
strains ( compared to ). What could account for the loss of 3′ end nucleotides in MMR-proficient strains that is not observed when there are phosphorothioate linkages at the 3′ end? It has been demonstrated that phosphorothioate linkages protect against at least some DNA polymerase proofreading activities 
. This suggests that some 3′ nucleotides may be lost by MMR-directed excision from the 3′ end, possibly by the 3′ proofreading exonuclease of the replicating polymerase 
. This could be an indication of a limited type of MMR function, involved in only surveillance of the 3′ end, as there is no overall difference in pattern of unmodified oligos with or without MMR (). The 3′ end of transforming oligos is not always lost, as we have been able to induce transformants using oligos in which the 3′ nucleotide has to be incorporated for reversion; this process can be quite efficient when the terminal mismatch is well tolerated, such as an 8-oxoG-A mismatch (results not shown).
There have been different conclusions on the use of phosphorothioate bonds in oligos used for transformation. The original experiments on transformation in yeast used unmodified oligos 
. Phosphorothioate linkages were later found to increase transformation in yeast by several fold in a different lab 
. In mammalian cells, the situation is complex 
. The use of oligos protected with phosphorothioate linkages at both ends induced cell cycle arrest and double-strand breaks 
. In one study it was found that oligos with phosphorothioate linkages at both ends gave greater transient correction than oligos with unmodified ends, but gave significantly fewer stable colonies 
. In the absence of MSH2
, transient correction was highest for unmodified oligos, followed in decreasing order by oligos modified at the 3′, 5′, or both 3′ and 5′ ends, but for those experiments relative viable formation of colonies was not reported 
. More recently it was found that transformation of msh2
cells was more efficient with unmodified oligos than oligos with phosphorothioate linkages at both ends and that the unmodified oligos created much less cell cycle disturbance 
. That would suggest that, at least in mammalian cells, oligos with phosphorothioate linkages at the ends can lead to double-strand breaks and cell cycle disruption and therefore fewer viable transformed colonies than the use of unmodified oligos, although the reason for the difference was not understood 
. A recent report studying oligo transformation in HeLa cells found that toxicity was correlated with increasing number of phosphorothioate bonds, possibly due to stimulation of cellular immunity, and that a few internal phosphorothioate linkages 3′ to the mismatch were most effective in creating stable transformants 
What might be the cause of cell-cycle arrest and double strand breaks observed in mammalian cells due to oligo transformation with oligos containing phosphorothioate linkages? Based on our observations, it appears that phosphorothioate linkages on the 5′ end could be problematical in replication fragment joining. It is clear that whatever process is used to join the 5′ end of the oligo into the completed replicated strand involves some sort of 5′ end processing, and the fact that a change in such processing could cause even slight differences on incorporation of 3′ end sequences, as observed in , suggests a significant change in oligo incorporation. Our experiments do not measure the incorporation of bases at the very 3′ end of the oligo, but it is clear that there is a tendency to lose bases at the 3′ end, and it may be that the usual method of primer extensive could involve a small degree of 3′ resection, which would be prevented by phosphorothioate linkages, again partially disrupting the normal incorporation.
The interesting exception to a similar pattern of transformation in wild-type and msh6
strains was provided by oligo G, where in wild-type, but not msh6
strains, nucleotides close to the center of the oligo were lost in 25% of transformants (). This loss was shown to be likely due to the occasional failure of MMR to recognize a C-C mismatch, as a well-recognized mismatch created just to the 5′ side of the C-C mismatch (oligo TG) resulted in retention of all oligo nucleotides in Trp+ revertants (). The location of the excised nucleotides relative to the retained nucleotides showed that in this case MMR-directed excision was from the 5′ end of the oligo and that excision must not have proceeded more than 4 nucleotides past the recognized mismatches or else as can be seen in , the C-C mismatch would have been removed resulting in no Trp+ revertants. Recent work analyzing single-base mispairs created by polymerase errors found that errors created by Pol α were corrected more efficiently by MMR than errors created by Pol δ, and it was hypothesized that the difference might be due to the use of the 5′ end of the Okazaki fragment as a strand discrimination signal 
. Our results are consistent with MMR-directed excision from the 5′ end of a replicating segment, and further indicate that such excision likely stops directly after the recognized mispair.
In order to make optimum use of oligo transformation, it is important to understand the parameters of oligo incorporation into the genome. As part of this work, we have shown that oligos can be used to introduce multiple changes into the genome simultaneously. With oligos that are 40 nt in length, there is a central core of greater than 15 nt that is almost always incorporated. Longer oligos would be expected to have a correspondingly longer core of nucleotide incorporation. One factor that is clearly important for the incorporation of a given part of an oligo is its distance from either end of the oligo. MMR represents a strong barrier to oligo transformation, and approaches involving transient inactivation of MMR appear most promising in circumventing oligo rejection by MMR 
. Even in the absence of MMR, however, in yeast only a small fraction of cells are transformed by a given oligo 
, for reasons that are not entirely clear. The mechanism of incorporation suggests that for a given cell, there would only be a short window for transformation in which the region of interest was being replicated and had a single-stranded region accessible for oligo annealing. Although the stability of the oligo to cellular exonucleases could be an issue in efficiency, protection of the ends with phosphorothioate linkages seems to introduce other problems, and has not led to noticeable increases in transformation efficiency in our hands. Another possibility would be that oligos were effectively being inactivated by protein binding or transport from the nucleus. Understanding the remaining causes of low transformation efficiency will be important for any potential therapeutic uses.