What is the benefit of integrating an extrachromosomal array? Extrachromosomal arrays suffer from three problems that may be solved to different extents by integration into the chromosomes:
1) Extrachromosomal arrays are lost from the germ line at some frequency, so maintaining a line carrying the array involves constantly selecting animals at least every few generations. For some phenotypes this is difficult. Integrating and homozygosing the array completely eliminates this problem. Having such a stable line is convenient for a variety of experiments. For example, when developing antibodies to a protein, it is useful to have overexpressing worms to stain, and to make extracts for westerns. You can use the overexpressor to work out conditions for using the antibody, and to confirm which band on the Western is really your protein.
2) Extrachromosomal arrays are lost mitotically at some frequency, so that animals that carry them are mosaics in which one cannot necessarily determine which cells have lost the array. This pattern of mosaicism varies in different individual worms. Thus transgenes on extrachromosomal arrays may not be expressed consistently in the set of cells that one would like. This problem is only partially solved by integrating the array. Results from integrating a number of beta-galactosidase reporter constructs show that genes in integrated arrays may be artifactually activated or shut off in certain cells. Presumably due to position effect, the integrated promoters may no longer give exactly the same expression pattern as the endogenous promoter in its normal chromosomal position. In addition, the pattern of expression from these integrated constructs is not 100% reproducible from animal to animal; some cells show variable expression. Overall, however, the expression from integrated arrays can be more reliable than that from extrachromosomal arrays. Basically, you're replacing the random error in expression that comes from mitotic loss of the extrachromosomal array with a more systematic error that comes from positional effects on the integrated array. Having integrated several GFP reporter constructs, I found that about half the integrated lines gave stronger and more reproducible expression patterns than the extrachromosomal arrays from which they were derived. The other half of the integrants gave much weaker expression than the original extrachromosomal arrays. So, if you want to integrate a GFP construct it's best to set up the screen on a scale that will give you several integrants, and hopefully one or more of these will be useful. Comparing the patterns of expression in several independent integrants allows you to identify artefacts due to position effects.
3) Extrachromosomal arrays can change their properties over time. The data for this is that if an extrachromosomal array has a certain measured transmission frequency, and a number of individual worms carrying this array are used to establish new lines, these new lines may have transmission frequencies for their arrays that strongly differ from each other and from the originally measured frequency. The expression of genes from these extrachromosomal arrays therefore potentially also suffers from this kind of variation. This creates the following kind of problem: if a transgene is to be crossed into a number of genetic backgrounds and the resulting phenotypes are to be compared, how can you know that the extrachromosomal array hasn't suffered some sort of change during the strain constructions? It is assumed (though with no data I'm aware of) that integrated transgenes are stable. Therefore, when you need your transgenes to have consistent properties over many generations, it is preferable to integrate them.
How to do it:
Summary: A strain bearing an extrachromosomal array is irradiated with gamma rays or x-rays, several hundred F1 progeny carrying the array are picked to individual plates, and for each F1 several F2 carrying the array are picked to individual plates. These plates are screened for 100% transmission of the array to the F3; such a strain is homozygous for an integrated array.
Stategy: Many people pick a few hundred F1s, and then pick 4-5 F2s from each F1. The large number of F2 plates involved usually necessitates doing the screen in a few batches to avoid killing yourself setting up all the F2s at once. However, both theoretically and practically, this is not the best strategy. Statistically, you will minimize the total number of plates set up (F1 and F2 summed) per integrant recovered if you pick more F1s, and pick only 2 F2s per F1. This also has the practical benefit of spreading the work out more evenly between the two generations. My practice is to pick 200 F1s on each of three consecutive days. After a day off, I then pick 2 F2s from each F1 as the plates mature. For those rare F1 plates that seem to have thrown the array at a much higher frequency than the rest, I pick 4 F2s. After 1-2 days off, I score the F2 plates over three days. The hardest work is picking the F2's; I find that picking 400 on each of three consectutive days is tolerable. This takes me about 4 hours of relaxed on and off work per day. Averaging over several integration screens, I am fairly consistently getting about one integrant per 200 F1s (400 F2s) examined.
1. Start by using microinjection to generate extrachromosomal arrays containing the gene of interest and an appropriate marker gene. In designing the experiment, consider using a coinjection marker like unc-76, dpy-20, or lin-15, rather than the popular dominant rol-6. With rol-6, the animals carrying the array are less healthy than the animals not carrying the array; with the other markers the reverse is true. I've found lin-15 to be a great marker; the Muv animals stick out like a sore thumb making the screen for integrants a snap .
2. Selecting an extrachromosomal array to integrate:
When identifying transmitting lines, be sure to keep lines that transmit the array at a low frequency. About 20% transmission is very convenient; this is enough transmission to allow you to easily get enough animals to irradiate, but it is much lower than the 75% transmission frequency you will look for coming from animals heterozygous for an integrant. However, don't get bent out of shape looking for such a low transmitter, since this actually only makes a marginal (30%?) difference to the screen. Unfortunately, individuals from such lines tend to show huge variations in the rate at which they transmit their arrays, so that an animal that transmits the marker at a high frequency usually doesn't carry an integrated array. If all you have are high frequency transmitting lines, don't worry about it and forge ahead; you can still get the integrant by blindly picking F2s from each F1 and looking for an F2 that transmits the array to 100% of its progeny.
Another consideration is the strength the desired phenotype coming from your gene of interest in the array. You presumably should start with an extrachromosomal array that gives you a very strong phenotype. However, this is no guarantee that all the integrated lines generated from it will give a strong phenotype; position effects seem to introduce lots of variation, so that both strong and weak integrated lines will be produced.
3. You will mutagenize L4 hermaphrodites. In order to get enough L4 P0s carrying the array, you may need to set up lots of plates, especially if the array is transmitted at low frequency. For mutagenesis, use ~30 L4 animals (carrying the array) per large plate. You will need about 600 F1s (carrying the array) to be reasonably sure of getting an integrant, so depending on the transmission frequency of your array and the number of integrants desired, you should scale up accordingly.
When using lin-15 as a coinjection marker, you are faced with the annoying problem that lin-15 cannot be scored until the adult stage. Therefore, you should produce staged P0 animals to mutagenize by transferring lots of worms to a new lay plate twice a day. Then you can easily pick lots of L4 progeny to new plates to mutagenize. The day after mutagenesis, pick the non-Muv mutagenized animals to new plates. I put 30 on a small plate, and transfer them twice a day for 2 days to produce well staged F1 animals, so that non-Muv F1s can be easily picked as young adults.
4. Mutagenesis. Either x-rays or gamma rays seem to work; use 3600-4800 Rads with either type. The Horvitz lab has both an x-ray machine and a gamma source. Shai Shaham claims that either type of radiation works about the same. Yishi Jin says she gets consistent results with gamma rays, and that she got poor results the one time she used x-rays. I tried both x-rays and gamma rays on the same array and I got 2 integrants from 300 F1s with gamma rays, and 0 integrants from 200 F1s with x-rays.
5. When F1s can be scored for the presence of the array, pick the F1s to new plates (one F1 per plate). As mentioned above, you should pick about 600 F1s to be reasonably sure of getting an integrant. . Most people avoid picking the very first F1 progeny that the P0s generate: these may be derived from germ cells that had already completed meiosis at the time of mutagenesis. Theoretically, (and with some empirical evidence), you can recover a higher frequency of mutants by waiting a bit. Typically the mutagenized animals are aged 24 hrs at 20°, then transferred to new plates, and the F1s from this second set of plates are used. I continue to pass the P0s to new plates every day to make it more convenient to pick the F1s.
7. When the F2 can be scored for the presence of the array, pick F2 animals to individual plates. As discussed above, I pick 2 F2s per F1.
If you are starting with an extrachromosomal array that is transmitted at a low frequency, you can also look through the plates for potential integrants. If you started with an array with a ~20% transmission frequency, you expect most of the background F1s to have thrown few animals with the array. F1 animals carrying a (heterozygous) integrated array, in contrast, will throw 75% F2s carrying the array. If you see a plate like this, pick a lot (~8) of F2s from it. Note that you will get some false positives and false negatives here. False positives: the array may still be extrachromosomal, but may have changed to a higher transmission frequency. False negatives: due to a position effect the marker on the array may be expressed only weakly, and may now be recessive instead of dominant. Another problem is that some arrays carry genes that make the worms sick. For example, the Rol phenotype makes worms grow slowly. Thus if the plates are scored too early, the percentage of animals carrying the array may be underestimated. One thing to do in this situation is to wait a few extra days to score the plates, and then look for the plates that are the last to starve out; these probably contained a high proportion of sick F2 worms (carrying the array).
8. When the F3s are old enough, look for a plate on which 100% of the F3s carry the array. This plate should carry a homozygous integrated array. I actually don't throw the plate out unless I see 2 worms that have lost the array; a single worm lacking the array might just be due to an unrelated L1 that was inadvertantly transferred to the plate along the with adult F2.
9. Since you put your worms through mutagenesis to generate this strain, you should backcross it several times to clean up the background.