Hi Bei,

First of all, I think you meant N-linked glycans, and the following discussion 
are based on this assumption. I am not aware of any direct glycosylation on 
lysines except for the O-glycosylation of hydroxylysines, which is really rare.

 
1. If you have enough protein, you should screen both deglycosylated and 
un-deglycosylated. Some proteins do crystallize with full N-linked glycans. 

2. If you do not have enough protein to try everything, or if the un-degly'ed 
won't crystallize, deglycosylation should be tested. Two things you want to 
know: 1) how easy it will be to deglycosylate the protein to nearly homogenous; 
2) how much less soluble the protein becomes after certain glycosidase 
treatment. 


3. You need to try different glycosidases. 
PNGase F is the one that you should try first as it removes the whole N-linked 
glycan. Then come the endoglycosidases: Endo F1, F3, Endo H, which leave a 
single GlcNAc on the Asn. If these all fail, then consider exoglycosidases: 
sialidase, galactosidase, hexaminidase, mannosidase, etc.. Generally speaking, 
the smaller the resulting N-glycan residue is, the less flexibility you have 
left on the protein, hence higher chance or getting highly diffractive 
crystals. 

The condition you need for a efficient cleavage with certain enzyme can vary a 
lot. This is caused mainly by the accessibility of the cleavage site. In the 
most facile cases, treatment at 4C with trace amout(1:10000) of PNGase O/N can 
result in complete removal of the N-glycans. On the other extreme, for certain 
proteins you may need more enzyme than your substrate protein and cut at 37C, 
and still only get a partial deglycosylation.

Being the enzyme that cuts the deepest on the N-glycan, one problem with PNGase 
is that the cleavage site (the NH2-C linkage on the Asn side chain) on certain 
N-linked glycosylation site could be quite inaccessible. And this actually 
happens a lot. In many cases, you may find you have to heat the reaction to 37C 
in order to get certain level of PNGase cleavage. For highly N-glycosylated 
proteins (having more than 4 or 5 sites), PNGase treatment is almost sure to 
fail. The other thing is, often, PNGase treatment results in poorly soluble 
proteins as the removal of the N-glcan exposes some surface patch for 
aggregation, or destabilized the protein itself. 

The Endo F1, F3, H, especially Endo H, can often result in better completeness 
of N-glycan removal, and better solubility of the protein, with the expense of 
one GlcNAc left on the protein. However, these ones require your N-glycans to 
be of certain type. For example, Endo H requires your N-glycans to be 
high-mannose type, which has to be generated from certain cell lines, or by a 
series of exo-glycosidase treatment. (But for insect cell produced proteins, 
endo H is very likely to work, see below.)

4. You can often achieve some degree of seperation of the glysylated form and 
un-glycosylated on ion-exchange, even when the N-glycans are not the charged 
type(for example, high mannose). This is likely due to a solubility difference 
of the differently glycosylated species. 

In the cases you have charged N-glycans, for example, proteins made from normal 
mammalian cells, the negative charge of the sialic acid on the N-glycans could 
be so significant, that it basically mask the protein's own surface charge. 
Consequently, the protein, regardless of its real PI, will bind to Q, while the 
deglycoylated will still behave normally. This will result in the separation of 
the two or more species. Actually, highly glycosylated protein with complex 
type glycans (with terminal sialic acid) often run as multiple peaks on 
ion-exchange, due to the heterogeneity of the N-glycan. 

HIC should separate the deglycolsylated protein too. 

Lectin columns should work. However the applicability of lectins depends on 
your N-glycan's type, and this could be a quite complicated topic to discuss. 
Also I am not aware of a good lectin for high-mannose.


5. Finally, to my knowledge, proteins produced from insect cells are mainly 
high-mannose type (I never experimentally tested this though, just took this 
idea from some literature). This means: 1) the N-glycans from insect cells are 
likely susceptible to Endo H treatment, which is great; 2) the N-glycans are 
not charged; 3) the N-glcans themselves are more or less homogenous as they are 
likely to be Man3,4,5 only, and are also smaller than the complex type glycans 
made from mammalian cells. This is a bless for crystallization without 
deglycosylation.


...OK, I found a paper right on this issue:

http://www.ncbi.nlm.nih.gov/pubmed/16997012

6. The incomplete glycosylation could simply be that the N-glycosulation 
machinary missed a few molecules. Especially if the protein is not from the 
same host species and is over produced. I would say: do not worry about it.

Zhijie





From: joybeiyang 
Sent: Friday, April 08, 2011 2:06 PM
To: [email protected] 
Subject: [ccp4bb] off topic: Is deglycosylation necessary for crystallization?



Dear All,

I am currently working with a protein which is glycosylated on several N and K. 
I use insect cells for secretion expression, so the protein was supposed to be 
correctly glycosylated. However after several steps of purification, there are 
still a minor doublet under the major band, which later on was proved to be my 
target protein without glycosylation. My question is:

1. Should I neglect the doublet and just crystallize at this point? (The 
unglycosylated form compare to the glycosylated form were about 5%:95%)
2. Should I deglycosylate the protein before crystallization and if so could 
anyone share with me a protocol? (Is precipitation likely to occur during 
deglycosylation?)
3. Was the unglycosylated form arise from cell death during cell culture or was 
it arise from the overexpression of target protein and inadequacy of 
glycosylation machinery and false secretion? How could I prevent it? Will a 
insect specific signal peptide work better than the native signal peptide?
4. Is it possible to separate the glycosylated and unglycosylated form through 
either a monoQ/S or a Lectin Sepharose 4B column?

Thank you very much for your input in advance and I greatly appreciate it!

With all my best wishes,

Bei

2011-04-08 

--------------------------------------------------------------------------------

joybeiyang 

Reply via email to