Array formation and integration after injection of DNA into oocyte nuclei.

We are exploring methods for gene targeting.  Such methods require the introduction of linear DNA fragments which can then be integrated into chromosomal DNA by means of homologous recombination.  However, injection of linear fragments into the gonadal syncytium does not result in homologous recombination of the gene fragment and the chromosome. Rather, such DNA is assembled into extrachromosomal arrays. These arrays are most likely formed using the nonhomologous end joining pathway. We think that it is likely that the nonhomologous end joining happens in the cytoplasm, before the DNA has access to recombine with the chromosome. Thus, this pathway is in direct competition with the homologous recombination pathway that we hope to use in gene targeting. In order to achieve gene targeting we have been working to both increase the frequency of homologous recombination and decrease the frequency of nonhomologous end joining.  In all cases, we injected a rescuing fragment from the unc-18 gene into the unc-18(e234) strain and scored for rescued animals.  DNA was injected directly into the nucleus of 5-10 oocytes per worm at a concentration of ~100ng/ul. The DNA could be incorporated in one of three ways: end joining into semi-stable arrays, integrating randomly into chromosomal DNA, or homologous recombination into the unc-18 locus. Transformed animals were tested for integration by isolating true-breeding rescued lines.  Homologous integrations were tested by outcrossing and testing for linkage.

First, it is possible that DNA injected into the gonadal syncytium does not enter the nucleus efficiently or perhaps it does not enter at all until after nuclear envelope breakdown.   By contrast, injection of circular DNA into the nucleus has resulted in frequent nonhomologous integrations and rare homologous integrations (Broverman, MacMorris, and Blumenthal PNAS 90: 4359-4363 1993).  We injected circular DNA into the nucleus but did not identify homologous integrants (see Table); however our methods were unable to generate nonhomologous integrants at high frequencies.

Second, linear DNA with 3' overhangs may inhibit nonhomologous end-joining as well act as a substrate for RAD-51 binding to initiate strand invasion with homologous sequences.   Digestion of 5' ends was performed by 15 minute treatment with lambda exonuclease at 1unit/ug DNA at 15ºC, leaving 3' overhangs that were estimated to be 500 nucleotides on each end. DNA treated in this way did not generate homologous integrations. 

            Third, it is possible that strand invasion by 3' single stranded overhanging ends can be promoted by prebinding the DNA with RecA protein. RecA is a bacterial homolog of Rad51, which is the protein that binds to 3' overhangs and catalyzes the integration of single stranded DNA into homologous double-stranded DNA. It was also thought that RecA bound to the free ends of the injected DNA would help protect the DNA from end joining. We did not identify any homologous events. The numbers of arrays and random integrants are proportionally similar with or without RecA. 

            Although none of our events is a homologous targeting event, our nonhomologous integrations suggest that it is true that a major barrier to homologous integration is access to the nucleus. It has been observed in a number of systems that the majority of integration events are nonhomologous, suggesting that homologous targeting in worms will require a method that generates a large number of integration events.

 

DNA	injected animals	transformants	integrations	homologous recombinants
Circular (vit-2 Broverman, MacMorris, and Blumenthal  1993)			63	2
Circular	102	42	0	0
Linear	322	119	4	0
Linear + RecA protein	176	78	2	0