h a l f b a k e r y
Birth of a Notion.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
or get an account
So, antibodies are in use in many many many labs, right
here in our lab, there are over 1000 different types.
They're incredibly useful. In ideal conditions, they can be
used to detect a single variant of a single protein among
1000s. Which is nice.
Now, I want to improve them. Simply genetically
mice/rats/rabbits to produce all of their antibodies with
fluorescent protein tag, such as CFP/YFP/GFP etc.
Then, simply produce antibodies in the normal way:
Inoculate, wait a bit, harvest & purify.
There are several reasons why this is great:
1. Their immune system will be fluorescent, which is
but not immediately useful to non-immunologists.
2. Antibody purification and quantification may be more
effective when assisted and monitored by fluorometry.
3. They will not require secondary antibodies for
Immunofluorescence and fluorescent Western Blot
experiments. This a: saves between 30mins and 1hr for
every experiment of this type ever performed. b:
1 layer of complexity and source of false positives.
4. The resolution of microscopy experiments will be
improved: Normally, the fluorophore is on the end of the
secondary antibody, which is attached to the primary
antibody. Each antibody is about 10nm, so, the
could be anywhere in a 20nm radius of the protein of
interest. This cuts that to 10nm. So you can get a
idea of where things are, and it's useful for super-
resolution microscopy<link>, which is becoming all easy-
5. Co-localization experiments are enhanced by the
possibility of using fluorescence resonance electron
transfer (FRET)<link>. Normally, you can use 2 separate
antibodies to probe for two independent proteins, and
to see if they're in the same place, sort of. Now, it's very
difficult to draw strong conclusions from these data, as a
microscope's resolution is a few hundred nm. Many
proteins that really don't interact at all can be found
a few hundred nm... so it's vague supportive evidence at
best. Looks good though, hence it's persistence in the
literature. Now, FRET, only works if the proteins are
REALLY close, so the CFP antibody you got from a mouse
and the YFP antibody from a Rabbit only produce a
co-localization in a FRET experiment if they're right next
to each other or directly interacting.
6. You can attach some of those whizz-bang
photoswitchable proteins, like Dronpa, to antibodies...
'cause, you know they're what the with-it kids are all
Right, got to go and wash my secondary antibody off....
[bs0u0155, Nov 14 2012]
[bs0u0155, Nov 14 2012]
[bs0u0155, Jan 17 2013]
[bs0u0155, Jan 17 2013]
||The title is somewhat misleading, since GM
animals (specifically sheep) are WKTE.
||"Flourescent tagging of GM-produced
antibodies" would be more concise.
||no takers? I thought this was a pretty good idea...
||I am sure its a great idea, but I don't actually know anything about the field, so it is difficult to form an opinion about whether this is a good idea or not.
||It's actually a pretty good idea, and it makes sense.
(Only one caveat - do two antibodies really get close
enough for FRET when they bind adjacently to a
target? And are various GFP variants good for FRET? I
thought they all had similar excitation spectra, but
I'm not at all sure.)
||This is the sort of thing you could base a startup on,
if it's not already patented.
||//And are various GFP variants good for FRET?//
||Sure are, can do lots of combinations nowadays CFP-
YFP, GFP-RFP <link>. I even suspect that triple-FRET
is possible, so you could have CFP-GFP-RFP triplets,
and prove that proteins A,B & C are adjacent, rather
than go through the tedium of doing A&B, then B&C
||In that case, I am prepared to add cinnamon to my
bun. Why isn't this already done?
||Whereabouts would you put the GFP? If you put it
on the rear end of the antibody you'd block some of
its functions. Can you insert it in a constant region,
or would this interfere with something? (I really,
really ought to know this.)
||// rear end of the antibody you'd block some of its
||I should also know this, but as long as the
antibodies get made, I don't really care if the
critters are immuno compromised. In fact it could
be useful, if they were like nude mice, and had to
be kept as such, then the protein you exposed
them to would be the only pseudopathogenic
protein they'd EVER been exposed to, so your
antibodies wouldn't have the "noise" of
||Actually, could you simply mess about with a
immune stem-cell line, give a nude mouse a bone
marrow transplant and you're done? it wouldn't be heritable, but you could freeze down the
appropriate stem cell line for each antibody, then
start/stop production in a bank of multi-purpose
nude mice at any time. You'd only need ~100,000
or so to cover the proteome, which is about 260
384 well plates. So, one freezer's worth, with a bit
of room left for super-fast emergency wine
||must find a tame immunologist.
||// You'd only need ~100,000 or so to cover the
||You'd actually want to create a line of mice with
GFP inserted in the antibody C-region genes, then
use those mice to raise antibodies against the
target protein you're interested in, shirley?
||In fact, you'd want a few lines with different GFP
variants in their C-genes so that you could (for
example) raise a green antibody against protein A
and a red antibody against protein B. Or did I miss
||Yes that was the original point exactly. I was just
thinking of an alternative which might allow a little
more commercial responsiveness. You could
synthesize peptides corresponding to a bunch of
known proteins, inoculate a whole lot of mice,
wait for a response then store the corresponding
bone marrow in the freezer, then when a request
comes in, you can scale up super quick in nude
mice. Thinking a little harder, it's actually easier
to go recombinant at that point.
||Ah, right, sorry. Problem is, you'll be making a lot of
lines, most of which will never be ordered.
Moreoveralso, how many of your peptides would
evoke responses that worked against the full
||yup, which is why the original idea's best.