III. MATERIALES Y MÉTODOS
3.5. DISEÑO ESTADÍSTICO
The mutated cDNAs were expressed in an in vitro system in the presence of p®S]Met, and the newly synthesized proteins visualized by autoradiography. Microsomal membrane was added to samples for cotranslational processing.
5.12.1 Glycosylation controls
Pre-p-lactamase and a-mating factor cDNA samples were expressed in the presence of microsomal membrane to verify processing activity, figure 5.9. Nearly all pre-p lactamase was processed to p lactamase (lanes 1 and 2). Approximately 50% a-mating factor was fully core glycosylated (19 kDa to 32 kDa), while about 20% remained unglycosylated with the remainder forming intermediate bands, lanes 3 and 4. When a low concentration of the
glycosylation inhibitor Bz-NLT was included in the reaction mix, more
intermediate bands corresponding to partial glycosylation events were seen (lane 6 ).
5.12.2 Wild type glycosylation
In the presence of membrane, glycosylation events were detected as band shifts of approximately 3000 Da each (209). When wild type gp91^^°’' cDNA was
expressed in the presence of microsomal membrane, approximately 25% was fully core glycosylated (62 kDa), 25% remained unglycosylated (53 kDa) and two intermediate glyco-forms were seen at 56 and 59 kDa, figure 5.10, lanes 1 and 2. Addition of extra microsomal membrane did not improve processing but instead reduced the amounts of protein synthesized (result not shown). When heat denatured membrane was used (lane 3) as a control, protein was still expressed but with no processing. 200 and 30 mM Bz-NLT completely inhibited glycosylation and reduced protein synthesis.
5.12.3 Expression of mutant forms of gp91"''^°* show that three sites are glycosylated
The presence of four bands from wild type gpOl^^®’' (figure 5.11, lanes 3 or 12) implies that three sites are occupied. Mutation of Asn®^ and/or Asn"*^° had no effect on the glycosylation banding pattern seen. Mutating combinations of Asn^^^, Asn^"*® and Asn^^*° resulted in corresponding loss of bands. From this data, Asn^^^, Asn^"^^ and Asn^^*° are all occupied by glycosylations, and so must be extracytosolic.
5.12.4 gpSI^^®* is an exception to rules on glycosylation
A survey of 300 polytopic mammalian glycoproteins has established a consensus of requirements for glycosylation (198).
1. The acceptor site has to be more than 10 residues from the end of a transmembrane domain.
2. A glycosylated loop must be greater than 30 residues long (on average 62 residues).
cDNA p r e - p lactamase la n e 1 m e m b ra n e - BZ-NLT (m M) - a-factor
4
-t-
+6
+ +Figure 5.9 Controls fo r protein processing
Protein transcription/translation in the presence of [^^S]-Met, visualized by
autoradiography. 0.1 pg control DNA was used in each reaction with (+) or without (-) microsomal membrane to observe processing, pre-p-iactamase cDNA was added to samples 1 and 2. In the presence of membrane, pre-p-lactamase {31.5 kDa) was processed to p-lactamase (28.9 kDa). a-factor cDNA was added to samples 3-6. In the presence of microsomal membrane, a-factor (18.6 kDa) was glycosylated to the core glycosylated form (32.0 kDa), with an intermediate band. In lane 6, 30 mM glycosylation inhibitor BzNLT was used with a-factor cDNA to show partial inhibition of glycosylation, producing a series of bands.
lane m em brane BZ-NLT (m M ) 2 3 4 5 6 7 + + ! - + - + 200 200 30 30 F ig u re 5.10 qp91P^°^ cD N A glycosylation patterns
Transcription/translation of wild type gp91^^°’' cDNA in the presence of p^S]-Met, visualized by autoradiography. Samples were with (+) or without (-) microsomal mem brane to show glycosylation. The membrane used in lane 3 had been heat denatured. Samples in lanes 5 and 7 were mixed with the glycosylation inhibitor BzNLT.
1 A
lane mem brane + 1 2 - 3 + 4 + 5 + + 6 7+ KDa 3 glycosylations 2 glycosylations 1 glycosylation u n g lycosyla te d1 B
oCO o CM CD "4- CM CO CD Vacuolar/ext r acell ular 2 4 0
III V 2 2 2 Lipid b i l a y e r 2 0 9 Cytosolic |\| 97 FAD NADPH 4 3 0
Figure ('.((Glycosylation patterns seen for various gp91phox mutants (upper panel). The model below shows the positions of the consensus glycosylation sites. Occupied sites are indicated by a branched structure. Heme coordination is shown between helices III and V, see text for details.
3. When there are several loops containing consensus sites, only the first (or most amino-terminal) is glycosylated.
Presumably these result from steric constraints, and access of glycosyl transferase to potential sites during cotranslational transport.
Histidine residues in transmembrane helices III and V of gp91^^°’' are believed to co-ordinate heme, figure 5.11. Asn^^^ and Asn^'^®, and Asn^"^° are separated by helix V, and so must be present on different extracytosolic loops. This makes gpQIPhox gpj interesting exception to the consensus on glycosylation positions.
It has been suggested that in vitro, proteins pass into the endoplasmic reticulum more slowly than in vivo. If so, this could allow abnormal sites to be occupied (198). However, the in vitro analysis presented here has been confirmed by FAB-MS studies with purified flavocytochrome 6 5 5 3, which also found that sites
Asn®^ and Asn'*^® were not glycosylated, and that Asn^^^, and Asn^^^° were occupied (unpublished collaboration with E. Stimson).
The a-subunit of the Na"' channel is the only other exception to the rules. This protein is a tandemly duplicated sequence, with glycosylations two loops (2 1 0).
The sites are widely separated, and are on independently folding domains (198). This suggests that the glycosylation of gp91P^°’‘ is a reflection of the glycosylation patterns of the two major functional domains of this molecule; the cytochrome and the flavin. The natural division between the two domains would lie between somewhere between helix V and Asn^^*° (figure 5.11, cartoon).
An X9T CGD case has been reported (184) with an Ala^®®Thr substitution. This mutation converts into an additional potential glycosylation site, and gp9 1phox jpi patient was observed to run slightly higher than in normal
controls. Treatment with N-glycanase converted the higher running gp91P^°’‘ to the expected 55 kDa band. Asn^^^ (figure 5.11) is close to the consensus sites at Asn^^^ and Asn^"*®. It is curious why the extra glycosylation has this catastrophic effect on protein stability; perhaps occupation of Asn^^ destabilizes interactions with p2 2P'°".
5.12.5 Future work
This type of analysis could be extended by introducing glycosylation consensus sites into gp91P"^°\ and also p22^^°*. Sites could be created at various positions along the protein chain. The glycosylation patterns found after in vitro synthesis would reveal whether there are other loops, and help define the boundaries of the membrane spanning regions.
6. Conclusions; a model for flavocytochrome
b s s eA structural model of flavocytochrome 6 5 5g would give a greater understanding of
the interactions of the various components of the oxidase, and the causes of the defects seen in CGD cases. Current techniques for NMR and crystallography analysis are unlikely to provide a crystal structure in the near future (33), and require impractical quantities of flavocytochrome 6 5 5 3. Structural evidence to
build a model has to come from other sources. Data has been presented in this thesis 1) showing that flavocytochrome bggg is a heterodimer, and 2) pinpointing
certain regions of gp91”^°’' as extracellular. Other structural and topological evidence concerning p2 2^^°’' and gp91^^°’‘ is available from literature:
1. Antibody binding and proteolysis
2. Secondary structure prediction (regions of a-helix and p-sheet, hydrophobic and globular regions of protein)
3. Interaction sites between flavocytochrome bggg and the cytosolic factors 4. Analysis of the nature of the various defects found in CGD cases (loss of
critical residues, structural deformations)
5. Comparisons with functional domains seen in other proteins (cofactor binding sites, heme binding motifs)
6. Computer modelling of gp91^^°’‘ based on the crystal structure of a homologous
protein, yeast ferridoxin NADP"' reductase (FNR).
This evidence is compiled in figures 6.1 and 6.2.