need help with methylation pattern - (Jul/01/2006 )

Hi

I'v got DNA metylation patterns of Differently Methlyation Region by using sodium bisulfite treatment. The test was to see if paternal genes were expressed or not. I have tested 4 genes (H19, IGF2R, SNRPN, MEST). 10 clones are used. H19 was completely demethylated and IGF2R was completely methylated except one clone. SNRPN and MEST really surprised me. SNRPN was perfectly methlyated, but MEST had 8 demethylated clones and 2 methylated clones.

Can anyone tell me how I should conclude the test with these results?
Any related comment is welcome!

Hi michigan
I assume you're testing human genomic DNA? These genes are all imprinted in humans with the exception of IGF2R which is imprinted in mouse but not human. From my understanding of imprinting you should see the following:

In normal individuals,
1. H19 DMR should show approximately 50% clones methylated (paternal allele) and 50% unmethylated (maternal allele). Your result shows only the maternal methylation pattern.
2. SNPRN DMR should show 50% clones unmethylated (paternal) and 50% clones methylated (maternal). Your result shows only the maternal methylation pattern.
3. MEST should show 50% clones unmethylated (paternal) and 50% clones methylated (maternal). Your result shows some evidence of differential methylation.
4. IGF2R is usually not imprinted in humans (some polymorphic imprinted cases have been reported, around 10% of cases I think) and if I recall correctly all clones would normally be unmethylated given the lack of imprinting. This is the opposite of your result.

Perhaps you need to sequence more clones to get a better representation of methylated to unmethylated clones (or have you been given abnormal DNA to test?). Or is there some bias in your PCR? All your results seem odd to me with the exception of MEST which is at least showing some evidence of differential methylation. I don't understand why IGF2R is methylated.

Also, bisulphite sequencing does not tell you whether the paternal gene is being expressed (although this might be inferred based on the methylation pattern obtained). Can you give us some more information? - i'd be interested in hearing.

Hi Karyotyper

Thank you for answering. As you guessed, I tested human cells. I was thinking these cells were benign ovarian teratoma with meiosis II error. But right now I am a little confusing.

At this time, I am trying to figure out if there are genes expressed from the paternal chromosome. That is, parthenogenetic tumor can be considered.

I don't think there are any experimental errors.

So I am collecting opinions about what kind of conclusion I should make with these results. No paternally expressed genes? Inconclusive? Need another test? I have also 96 STR markers showing a lot of homozygosity(32 of 96 markers)

Hi again michigan
I'm sorry but my comments about the methylation of IGF2R was wrong. There is a DMR within intron 2 of this gene which is maternally methylated, however the gene is definitely biallelically expressed in most humans as I said in my last post.

OK, if your cells are from a benign ovarian tetratoma I supposs you're hoping to see an exclusive maternal methylation pattern. From what I have read this is not necessarily the case and the paternal methylation pattern can sometimes still be seen depending on the gene being investigated. This might be due to the timing of imprint erasure/establishment in the oocyte. See this paper
for a very similar investigation to your own.

Isn't the SNRPN result showing you what you would expect? - ie. totally methylated (maternal) pattern.
H19 is also showing a hypomethylated (maternal) pattern and IGF2R is showing a predominately methylated (maternal) pattern as well. MEST is showing an excess of the paternal (unmethylated) pattern which is interesting.

I still think you should be talking about a maternal or paternal methylation pattern rather than whether the paternal gene is being "expressed". You're not looking at expression.
You need to investigate closely linked centromeric markers to differentiate between an M1 and MII error. Heterozygosity near the centromere suggests an MI error while homozygosity suggests an MII error. Homozygosity can be explained by either an MII error and/or recombination depending on where it is occuring on the chromosomes.

Hellow Karyotyper,

It was nice to talk to you again.

I think you almost got to my point now. Methylated SNRPN is exactly what I expected, but as you said it is interesting, MEST is my headache. I expected both of them are methylated and both of them should be methylated. It may be not parthenogenetic tumor.

Speaking of STR markers, 32 of 96 markers were homozygous, 18 of 32 homozygous markers were near the centromere and 3 heterozygous markers were near the centromere. Here again, these 3 heterozygous markers near the centromere are giving me headache. Cross-over near the centromere does not sound good.

Somehow I hate my job.

Hi again michigan
You're not dealing with an easy topic so I can see why you're frustrated. Do you have a sample from the mother? You can prove the tumour is parthenogenic by showing that there are always one or two maternal markers at all loci. There should be no non-maternal markers at any loci if the tumour is really parthenogenic.

With regard to the heterozygous markers near the centromere, it depends on how close they are. It is still possible to get recombination near the centromere. Its a lot easier to interpret this if you know the genotype of the mother. If the mother's centromeric markers are heterozygous and you're getting reduction to homozygosity in your teratoma then this strongly suggests an MII error. If you don't know the mother's genotype then you are relying on the reported heterozygosity levels of your markers to make an interpretation.

Maybe the MEST methylation pattern you're seeing is arising post zygotically in the parthenogenic cells? Alternatively maybe the MEST imprint isn't reset until very late in oogenesis - I don't know enough about the genetics of MEST to help beyond this. Try and do a PUBMED search to find more. The paper I linked in my previous post also has a lot of good information.