I have an ovarian cell line TOV112D which is passage 20 when I thaw it. Then I do 2 runs of Colony formation assays which give me a phenotype, but after that it just disappears. I dont know if I can trust such a phenotype - do you think I can just blame passage number and only choose the runs I like...? I was also thinking whether I may be selecting for some fast growing population that does not care anymore - but I always take care not to let them be too confluent. However it is hard to get any information in regards how many cells should be seeded for how long rather than splitting ratios which I think is always a bit subjective...? Thanks for your help!

Passaging should be done when cells are about 70-80% confluent - that's when the split ratios become less subjective.  The time/cell number thing  is very difficult as it would have to be done for each different amount of cells for each time, so won't find that sort of information readily.

 

You could be selecting for fast growing cells (or cells that detach easily) or any number of different things, but 2 passages is pretty short to be seeing this sort of change.

 

However, it is entirely possible that there is some sort of switch in growth type that is the result of the cells becoming confluent.

Thanks Bob1, this was very helpful! I must admit that with some cell lines I find it very hard to estimate the exact confluency because they grow e.g.in 'islands' and never achieve 100% confluency but obviously need splitting. I tried to figure it out by setting up growth curves but it seems like the growth rates do change over time (eg after thawing they grow slower) so this also did not solve my problem.
Anyway, I have the feeling I rather tend to split them at a lower density rather than higher because I want to avoid them becoming too confluent. Do you know whether splitting them eg at 60% could lead to selection of fast growing cells? Thanks a lot!

@ Bob1: And also, do you think such a selection would be reversible if I switched to the correct splitting ratio? Thanks!

I know what you mean about the "islands" - I have a couple of lines that grow like that too.  If you want really hard lines for this, try neuronal derived lines which have extremely long processes and stop growing when the processes touch another cell, so it looks like the cells are about 50% but in reality are fully confluent.  The way around this is to seed at a higher density (maybe 1:4 or 1:5 splits) than you might otherwise (means you will need to split more frequently), and try to ensure that you have a largely single cell suspension when seeding.  Seeding at a lower density means that you are more likely to get the islands forming, and makes it difficult again.

 

Selection for fast growing cells would be done by passaging very frequently and seeding at low density.  I don't think you will be doing too much selection so long as they don't seem to be growing faster than they used to.

 

A growth curve for any cell population should be more or less the same - some sort of logarithmic curve with a lag phase to start with and a plateau at the end.

 

I doubt the effect will be reversible, as most of these sorts of things tend not to be, but it may be, and is quite likely to go away if you freeze the cells down.