its nothing to do with the stain its the trypsin that can digest the CG region because its not condensed so appears light but the AT region is very condensed so the trypsin cant enter and digest this region so appears dark

Because AT rich present in heterochromatic region like in centromeric in 1, 9, 16 no. chromosome and arm on Y chromosome which show dark band because of more condensation of that region and trypsin unable to digest that protein so it take more Geimsa stain than GC rich region which is less condensed and having mostly housekeeping genes and called euchromatic region so AT rich take more geimsa stain than GC bases.    

same as iterated in answers 1 and 2 but giemsa staining of chromatin also a function of where chromosomes are in the cell cycle  and their epigenetic imprint e.g. positive heteropyncnosis-barr bodies

Yes, During digestion with Trypsin(Enzyme) , protein present in G-C rich region will be digested  whereas protein which are present in A-T rich region will not. So it take Giemsa stain. Also there is a repetitive DNA sequence present in A-T rich region so it condescended and show dark stain.

Dear Apostolos,

The Giemsa widely used in cytogenetics for visualizing chromosomes, is a basic dye that has affinity for the acidic nature of chromatin (DNA). Other dyes can also be employed for this purpose as the Feulgen reagent.

Without considering any prior treatment, chromosomes are stained uniformly by Giemsa. This changes when the chromosomes are exposed to chemical treatment denaturant (formamide, barium hydroxide, sodium hydroxide) or enzymatic (trypsin) to obtain banding as C-band or G-band. Typically, more preserved regions of chromosomes in these cases are the heterochromatin and therefore they are more strongly stained with Giemsa.

I agree with Roberto's submission. Giemsa stains chromosomes more-or-less uniformly but subsequent treatments leave different segments of the chromosome with various extent s of staining based on the A-T and G-C richness to yield the type of banding pattern obtained.

I know this is quite old but it comes up as a google result pretty high up. Others with better access to newer papers can probably give a better answer. 

As of 2011, I can't seem to find a full mechanism but here's what I've found. The giemsa stain consists of some cationic part and an anionic part (commonly azure B and Eosin Y respectively). The azure B first intercalates within the DNA then seems to form a complex with the Eosin Y. The issue arises then, why do we need the proteolytic and why does the azure B intercalate more readily within the AT rich regions? 

The second question of intercalation seems to depend on the proteolytic profile of trypsin (usually the proteolytic of choice). If the answer was simply that AT bonds naturally interact more weakly than GC , then we wouldn't need the proteolytic at all.

So all I can say now is that trypsin seems to cleave some set of dna interacting proteins that are more prevalent or accessible in heterochromatin over euchromatin. This then allows further intercalation of the azure B or accessibility to sites that were not available before. 

http://www.ucl.ac.uk/~ucapikr/projects/Ana_staining_LitRev.pdf

Article How Romanowsky stains work and why they remain valuable—Incl...

Reading these posts is obviously interesting to me! So I would suggest that the story I told in that review is a plausible explanation as to why you get the intense staining [ie its effectively a catalytic process generating an azur-eosin complex]. But the reason some chromosomal regions staining after the various treatments is anothr matter. And given the wide variety of treatments that can give rise to G bands [etc] the role of ease of access of enzyme or ease of removal of fragments does seem a plausible explanation. By the way, on p42 of that old review I did note that some chromocomes do contain 'super AT' regions. Classically this includes the malarial parasite ... but then a different complex/dye state is produced, more carmine in colour and not like the purple of G bands and as seem in standard haematology. Any comments on this guys? Richard H

I would like to know why heterochromatin is AT rich region? as discussed above, if I consider transcriptional active site which is in euchromatin region or ORIGIN of replication which is well known to be AT rich then how come it is GC rich. I will appreciate any paper suggesting this.