*** For details on how to be removed from this list visit the *** *** CCP4 home page http://www.ccp4.ac.uk ***
SUMMARY ON DEGLYCOSYLATION (built on tips from the CCP4 community) Glycoproteins are proteins to which a carbohydrate chain is covalently attached. Proteins modified in this way in many cases present a considerable challenge to X-ray structure analysis, because they do not easily form crystals. This is because the attached carbohydrate chains are often heterogeneous and flexible and interfere with the formation of crystal contacts. A well-known strategy to tackle this problem is to deglycosylate the glycoproteins prior to crystallization (see below for practical tips). This is usually done with endoglycosidases such as PNGase F (resulting in virtually complete removal of all N-linked glycans) or Endo H, Endo F1, F2, F3 (for partial deglycosylation). Also exoglycosidases can be used, alone or in combination with endoglycosidases. Alternatively, the glycosylation sites can be mutated (substitution of S/T to Ala or N to Gln or Asp in the conserved NXS/T motif). Expression in lower organisms like E. coli or Pichia pastoris is also an option, or - if a less radical procedure is required - genetically modified expression systems can be used that do not contain certain glycosyltransferases (e.g. CHO Lec cells). Practical tips for enzymatic deglycosylation Several companies like SigmaAldrich and ProZyme have excellent theoretical and practical information about deglycosylation on their web pages, which are well worth studying. Both N- and O-glycosylation occur naturally in proteins, with N-glycosylation being the more common modification. N-glycans can generally be removed rather effectively by a single enzyme, PNGase F, whereas this is not the case for O-glycosylation, where several enzymes have to act in concert to exert the same effect. Deglycosylation can be tried both under native and denaturing conditions. While native conditions are in general preferable for crystallization purposes, deglycosylation under denaturing conditions is more effective and should for this reason always been done in parallel as a positive control. To achieve higher efficiency even under native conditions, one may experiment with using higher temperatures (up to 37¨¬C, instead of 0-4¨¬C) and/or longer reaction times (1-5 days; preferably adding some sodium azide to the mixture to prevent bacterial growth). In some cases, native deglycosylation does not work well. In these cases, one may try to deglycosylate the protein under denaturing conditions and then refold the protein. Alternatively, one may use only slightly denaturing conditions, by applying various mild detergents (like ¥â-octyl-glucoside, Chaps, Triton X-100, SDS, etc.). One can also try to incubate the reaction mixture in a sonicating water bath. Yet another option is to deglycosylate the protein only partially using exoglycosidases (neuraminidase, galactosidase, etc.) instead of endoglycosidases. In practice, one often starts out with ca. 30 ¥ìl protein solution (concentrated to 0.5 to 5 mg/ml in water or a suitable buffer) and adds the glycosidase of choice (e.g. PNGase F, Endo H, Endo F1, F2, F3, neuraminidase or enzyme kits) in ratios varying from ca. 1:15 to 1:2000 (w:w). The deglycosylation reaction is then monitored regularly by taking samples and analyzing them on SDS-PAGE or IEF gels. A final check is preferably done by mass spectrometry. With some of the glycosidases like neuraminidase, one should be very careful to fully remove the enzyme (e.g. by gelfiltration), since it crystallizes easily even in minute concentrations. Another way to ensure complete removal is to use glycosidases as fusion proteins (coupled e.g. to glutathione-S-transferase; see Grueninger-Leitch, 1996) and passing the mixture through a GST affinity column (glutathione Sepharose) after the reaction is completed. Of course, it is always well worth a try to crystallize the protein also in its fully glycosylated form. Some proteins even crystallize better in their glycosylated form, due to the involvement of the glycan chains in crystal contacts. Other proteins behave best when partially deglycosylated. Glycobiology tools If one does get crystals from the glycosylated protein, one has to deal with handling carbohydrates, so here are a few practical tips: Modeling the ligand into the binding site, with or without supporting electron density, of course requires a 3D structural model of the carbohydrate ligand. This can be obtained either from scratch (e.g. using the modeling tool 'Sweet' from the www.glycosciences.de website, which converts carbohydrate sequences into 3D-models) or from a structure database (via the Uppsala HIC-Up server at http://xray.bmc.uu.se/hicup/ or directly from the PDB (http://www.pdb.org/) or other appropriate databases such as the Cambridge Structural Database (CSD; http://www.ccdc.cam.ac.uk/). The model of the carbohydrate ligand should then be refined and checked as carefully as the protein structure itself. Currently, however, tools for analyzing carbohydrate structures are not as widely known and applied as those for protein or DNA structures (see Kleywegt, 2003). A good tip is to check out the website at http://www.glycosciences.de. A couple of tools, such as pdb-care (to check carbohydrate residues in pdb files for errors) or carp (which generates Ramachandran-like plots for carbohydrates) or even GlyProt (to identify glycosylation sites in proteins and automatically attach them in silico) make life of structural glycobiologists significantly easier. Another database, currently under development, is EuroCarbDB (http://www.eurocarbdb.org/databases). And questions concerning carbohydrate nomenclature may be resolved by consulting the web site http://www.chem.qmul.ac.uk/iupac/2carb/. by Ute Krengel, University of Oslo
