Subject: question about the AT/AC splice in prospero
Dear Dr. Doe,
I am an annotater for FlyBase and am reviewing the annotation of the
prospero gene in the Release 3 genome. Unfortunately, it has gone
public with errors and I'd like to clean it up, hopefully with your
help. Also, it is not apparent in the FlyBase annotation report, but
there is a large transposon inserted in the third intron in the
sequenced strain.
Currently in the FlyBase annotation report there are three mRNAs
annotated. 
One was based on GB:AF190403 which is a genomic DNA sequence and will
be deleted.
One agrees with the structure predicted by the community prospero
sequences from GenBank.
The last one is predicted by the Berkeley full length cDNA GH11848. It
has the AT/AC splice and also reads through the first intron present in
the other mRNA.
The last mRNA is the one I am wondering about. While prospero is widely
cited in the literature as having an AT/AC splice, I can't seem to find
any confirming evidence besides the one Berkeley cDNA. While the
Berkeley cDNA collection is terriffic overall, as might be expected
there is some percentage of anomalous clones with dubious structures
(incomplete clones, included introns, etc..). When prediction anything
out of the ordinary such as the AT/AC splice, we like to have more
evidence.
I have tried to track down the evidence for this unusual splice site in
the literature and so far have not been successful. One review cites
your sequence GB:Z11743 which actually has the traditional GT/AG
splice. Other reviews have pointed me to papers that I can't easily get
my hands on. I thought it would be quicker to ask you to point me to
the relevant references/evidence or perhaps provide a personal
communication. Have you isolated independent cDNAs that confirm the
unusual splice? Are they published or can you provide me with a FASTA
sequence?
Thanks for any help you can offer.
Sincerley,
Beverley Matthews
FlyBase-Harvard
Subject: Re: question about the AT/AC splice in prospero
We isolated about 25 prospero cDNAs of varying lenghts from several
different pNB40 libraries (Nick Brown's) back in the early 90's. Nearly
exactly half had the canonical splice, and the other half had the AT/AC
splice. Each of the three libraries gave both forms. The alternative
splice leads to a different length cDNA near the homeodomain, which is how we
catagorized the cDNAs.
I can go back to my archives and see what I still have for documentation.
You could also email Joan Steitz who has done some work on this
splice, as well as Roy Parker at Arizona. But let me see what I can get for you.
Best regards,
Chris
Dear Chris,
Thanks very much for your quick response. I'm glad to know that as
suspected there's a lot more evidence out there than I could see. It
would be great to get a personal communication to go with this record
so there is a better evidence trail. Either the FASTA sequence(s) or
just your written confirmation of the AT/AC splice would be great.
I would like to know if your cDNAs have the same structure as the BDGP
full length cDNA or if there are any additional forms for me to
annotate. Thanks for agreeing to look into this.  I know it can be
difficult to resurrect old results.
Thanks again,
Beverley
Subject: Re: question about the AT/AC splice in prospero
Hi Beverley,
Here is what I found in my notebook.  The exon 2/3 splice can be via a
consensus splice site or an atypical splice site. The consensus splice
generates a cDNA we refer to as prosL (L for longer). The atypical splice
generates a 87nt shorter cDNA we call prosS.  We characterized 20 cDNAs and
found the following:
4-8hr library: 10 prosL cDNAs  and 6 prosS cDNAs
8-12 hr library: 2 prosL and 2 prosS
TOTAL: 12 prosL and 8 prosS
All are different cDNAs based on length; the cDNA is long and the library
was made from oligoT priming so only one of each class was full length. We
characterized all by restriction digest and several by sequence analysis
(at least one was the full length prosS cDNA with the atypical splice).
We didn't find any other variations except length (boring) and the exon 2/3
splice, but we didn't do much sequencing back then. We have most of the
cDNAs if you want them!
Best regards,
Chris
Chris,
Thanks for the information. Just a few more questions to be completely
clear.
There is an 87 base difference in the two pros forms that I've
annotated depending on the exon 2/3 splice. In addition to having the
unusual exon 2/3 splice, the transcript based on the Berkeley full
length cDNA is different from the published form in that it includes
the intron 1 sequences. The protein appears to read right through the
intron and stays in frame.
Interestingly, there are a number of fly BLASTX hits that map
completely or largely to that intronic region. From their description
lines, they are things like SPEN protein, CREB-binding protein homolog,
homeobox protein cut, A-kinase Anchor protein, etc.. Not terribly high
scores but there are a lot of them. Must be some conserved domain in
there.
My question is whether or not your cDNAs all have intron 1 spliced out.
If you have any cDNAs that splice out intron 1 but have the atypical
splice site at the exon 2/3 junction, I'll make a new transcript form
for that.  Similarly, if you have any cDNAs that fail to splice out
intron 1 and have the canonical splice at the exon 2/3 junction, I'll
have to make a new transcript for that.
The final question is if it's OK to use your email as a personal
communication to FlyBase.
Thanks again for your help!
Beverley
Subject: Re: question about the AT/AC splice in prospero
Based on the data I have easily accessible, there is no evidence that any
of our 20 cDNAs has first intron sequence. We might have missed it for some
of the less well characterized clones...  so I can't say they all
definitely lack it... but we haven't seen it so far.
It is fine to use any of these emails as personal communications. Thanks
for checking with me.
Chris