Crystal structures of both isoforms of human glutamic acid decarboxylase

FIELD OF THE INVENTION

The invention relates to crystal structures of the isoforms of human glutamic acid decarboxylase, GAD65 and GAD67. This invention also relates to a crystallographic model and methods for designing and selecting ligands that bind to and around the active binding site of GAD65 and GAD67.

BACKGROUND OF THE INVENTION

In this specification where a document, act or item of knowledge is referred to or discussed, this reference or discussion is not an admission that the document, act or item of knowledge or any combination thereof was at the priority date, publicly available, known to the public, part of common general knowledge; or known to be relevant to an attempt to solve any problem with which this specification is concerned.

Incorporated herein by cross reference is the contents of a paper entitled “GABA production by glutamic acid decarboxylase is regulated by a dynamic catalytic loop” published in Nature Structural & Molecular Biology, volume 14, issue 4, pages 280-286.

Nature of GABA

Gamma-aminobutyric acid (GABA) and glutamate are the primary inhibitory and excitatory neurotransmitters in mammals. The balance between GABA and glutamate controls diverse processes such as neurogenesis, movement, circadian clocks, tissue development and blood glucose regulation. GABA is synthesized from glutamate by the 65 kDa and 67 kDa isoforms of the pyridoxal phosphate (PLP) dependant enzyme Glutamic Acid Decarboxylase (GAD65 and GAD67). Despite 81% sequence similarity, GAD65 (but not GAD67) cycles between an inactive apo-form and an active PLP-bound state; this activity represents a key control mechanism for GABA synthesis. Further, GAD65 (but not GAD67) is an important auto-antigen in diabetes and neurological disorders. The molecular basis for the catalytic and immune distinctions between the two GAD isoforms remains unknown.

Nature of GAD

GAD is a member of the pyroxidol-5-phosphate (PLP) dependant transferase superfamily. Members of this diverse superfamily play a major role in amino acid metabolism and catalyse decarboxylation as well as transamination, racemisation, aldol cleavage, and beta and gamma elimination. In GAD, PLP acts as an electrophilic catalyst while covalently bound to the glutamate, thereby stabilising the carbanionic reaction intermediate. This can be depicted diagrammatically as follows:

Vertebrates possess two closely related isoforms of GAD, GAD65 and GAD67, which are products of two independently regulated genes. Gene knockout studies together with clinical data demonstrate the necessity for this enzyme activity, and delineate distinct roles for each isoform. GAD67−/− mice exhibit substantially reduced GABA levels and die at birth of cleft palate. In contrast GAD65−/− mice have normal levels of GABA and appear normal at birth, but develop fatal seizures and anxiety phenotypes. In humans, mutation of GAD67 in humans results in spastic cerebral palsy.

In addition to its key role in neurotransmission, GAD65 is found in the human pancreas where GABA may regulate the first phase insulin response. The presence of GAD in the brain and pancreas is also of immunological importance, since autoantibodies to GAD65, but rarely to GAD67, are found in neurological conditions such as stiff person syndrome and most patients with type I diabetes.

Several mechanisms have been described in the regulation of GAD and GABA synthesis; these include post-translation modifications (including phosphorylation and palmitoylation), subcellular distributions (GAD67 is primarily cytosolic whereas GAD65 is associated with synaptic microvesicles) and transcription/translational control. Most notably, however, a side reaction in GAD65 causes dissociation of the co-factor PLP and enzyme inactivation as depicted in the reaction sequence above.

In this latter reaction free pyridoxal mono phosphate (PMP) together with succinic semialdehyde is released. Thus, it has been reported that ˜80% of GAD67 isolated from cells exists in the active PLP bound state, while ˜80% of GAD65 is in the inactive apo-form without PLP. Together with the physiological information derived from murine knockout studies, these data are consistent with a model where GAD67 is responsible for production of a basal pool of GABA and GAD65 is activated in response to a sudden requirements of extra GABA, for example in response to stress. The molecular basis for this difference remains to be elucidated.

There is therefore a need for a model of the structures of GAD that can reveal how the two closely related enzymes of GAD are able to perform strikingly different roles. Specifically there is a need for a structure model that shows how GAD65 is able to allow enzyme inactivation. Furthermore there is a need for a model structure and structural data that can facilitate the design of compounds that can perform functions such as prolonging GABA production by GAD65.

SUMMARY OF THE INVENTION
Crystal Structure

The present invention therefore provides the structure coordinates of the two isoforms of an N-terminal truncation of GAD (GAD 65 and GAD67). The complete coordinates are listed in Table A.

The present invention further provides a crystal of GAD67 consisting of a monoclinic P2₁space group with unit cell dimensions of a=84.05±2.3 Å, b=62.74±2.3 Å, c=101.35±2.3 Å and β=106.69.

The present invention further provides a crystal of GAD65 consisting of an orthorhombic C222₁space group with unit cell dimensions of a=78.25±2.3 Å, b=99.05±2.3 Å and c=120.01±2.3 Å.

The present invention also provides a machine-readable data storage medium which comprises a data storage material encoded with machine readable data defined by the structure coordinates of GAD65 according to Table A or a homologue of this isoform.

The present invention also provides a machine-readable data storage medium which comprises a data storage material encoded with machine readable data defined by the structure coordinates of GAD67 according to Table A or a homologue of this isoform.

Catalytic Loop

The structure of GAD67 reveals a catalytic loop that covers the active binding site and introduces Tyr 434 as a catalytic switch. In contrast, the catalytic loop is mobile in GAD65. Mutational analysis reveals that destabilization of the catalytic loop in GAD67 promotes enzyme inactivation. It is further shown that many key residues implicated in auto-antibody binding map to mobile regions close to the active binding site of GAD65. The structure and model of the present invention show that mammals regulate the balance between GABA and glutamate by modulating the mobility of a catalytic loop. However, a cost of this mechanism may be that increased mobility in GAD65 may enhance the antigenicity of the molecule.

Accordingly, the present invention also provides a method for determining at least a portion of the three-dimensional structure of a species, such as a molecule or molecular complex which forms a binding partner of the catalytic loop (ie the loop that contains Tyr 434) or the region surrounding the catalytic loop of GAD65 and GAD 67. The molecule or molecular complex may for example stabilise, alter the conformation of, or interact with the catalytic loop. It is preferred that these molecules or molecular complexes correspond to at least a part of the active binding site defined by structure coordinates of GAD65 or GAD67 amino acids according to Table A, or a mutant or homologue thereof.

Accordingly the present invention further provides a method for identifying a binding partner for the catalytic loop or the region surrounding the catalytic loop of an isoform of an N-terminal truncation of GAD comprising the steps of:

(i) characterising the catalytic loop from the structure coordinates of Table A;

(ii) designing or selecting a binding partner that interacts with the catalytic loop or the region surrounding the catalytic loop; and

(iii) obtaining or synthesizing said binding partner.

The present invention further provides an active binding site in GAD65 or GAD67 as well as methods for designing or selecting GAD modulators including agonists, partial agonists, antagonists, partial antagonists and/or selective GAD modulators using information about the crystal structures disclosed herein. The present invention further provides GAD modulators designed or selected according to said method.

In a preferred embodiment the methods or GAD modulators of the present invention are suitable for modulating the ability of either GAD65 or GAD67 to produce physiologically active compounds, such as GABA or succinic semialdehyde. Modulation in the production of GABA is expected to be useful of treating diseases such as movement disorders, Parkinson's disease, autism, schizophrenias, depression and other mental or physical illnesses that occur as a result of GABA deficiency, perturbations in GABA or GAD.

The structures of GAD reveal how two closely related enzymes are able to perform strikingly different roles. The work shows how mobility in the catalytic loop of GAD65 is able to allow enzyme inactivation. Indeed, our structural data may facilitate the design of compounds aimed at stabilising the catalytic loop and prolonging GABA production by GAD65.

GAD65 is highly auto-antigenic with respect to GAD67 and the structures of GAD65 and GAD67 thus provides a unique structural foundation for understanding auto-immune responses. The structures reveal that a key difference between the molecules is the flexibility of the C-terminal domain together with the catalytic loop. It has previously been reported by many others that flexible loops function in native proteins as efficient antigens.

The structural model of this invention thus also provides a high-resolution picture of how mammals regulate GABA production.

BRIEF DESCRIPTION OF THE DRAWINGS

FIG. 1 is a dimeric structure of GAD67 within the asymmetric unit of the crystal. The N-terminal, PLP-binding and C-terminal domains are shaded differently and labelled. Monomer A is a slightly lighter shade than monomer B. Within the active sites, the K405-PLP Schiff base, and the PLP-GABA adduct are clearly labelled. Bound GABA product is also labelled. The “catalytic loop” (residues 430-450) forms a “flap” over the active site of an adjacent monomer.

FIG. 2 is a view of the molecular surface of GAD67 dimer and GAD67 dimer interface shaded according to sequence conservation between GAD67 and GAD65.

FIG. 2(
a) shows structural superposition between GAD67 A chain, GAD67 B chain and GAD65. Disordered regions in GAD65 are numbered.

FIG. 2(
b) shows structural superposition of GAD67 dimer and GAD65 dimer. GABA moieties bound in active sites are shown as spheres. Catalytic loops are shaded. Both figures highlight the structural shifts in the C-terminus.

FIG. 2(
c) shows the molecular surfaces of GAD65 and FIG. 2(d) shows GAD67 flexibility, as measured by atomic temperature factors. Light shading=low (ordered) and dark shading=high (flexible). Disordered residues 436-485 in GAD65 are represented as dotted lines.

FIG. 3 is a view of the molecular surfaces of GAD65 and GAD67 shaded to show differing electrostatic potential. Location of amino acids on the structure of the dimer GAD65 are shown by mutation to reduce the binding of 13 human anti-GAD65 monoclonal antibodies. The human mAbs b78 and b96.11 are representatives of two major epitope regions on GAD65, indicated by the “typical antibody footprint”. Epitopes for mAb M2, M5, M8, M9 and b78 (cluster 1) reside in one face of the C-terminal domain, in the region of the highly flexible loop for which there is no structure. Residues required for mAb binding within this cluster form a linear arrangement between the most flexible regions on the GAD65 structure, the catalytic loop and the C-terminal loop (black and doted lines respectively), being the representative mAb for this cluster (b78) enzymatically inhibitory. For two of these mAbs (M8, M9), the identified epitope residues also lie in the interface between the N-terminal and C-terminal domains (B). The epitopes within cluster 2 (M1, M3, M7, and b96.11) mapped to the opposite face of the C-terminal domain (Y). The epitope for mAb b96.11 is located in the border between the PLP and C-terminal domains, a region commonly recognised by autoantibodies present in Type 1 diabetes. The remaining 5 mAbs (M4, M6, M10, DPC) mapped to the PLP-domain, and form two sub-clusters: M4, M6 and M10, which can cross-inhibit each other, and the epitope for DPC

FIG. 4 is a view of the GAD65 catalytic loop interactions with active sites of GAD67 and GAD65.

FIG. 4(
a) shows the GAD67 monomer A.

FIG. 4(
b) shows GAD67 monomer B.

FIG. 4(
c) shows a close-up of GAD67 monomer A.

FIG. 4(
d) shows GAD65.

FIG. 4(
e) shows the superposition of active site residues of GAD67. The catalytic loop and Y434 sidechain (sticks) of GAD67 are shown. In panels A, B, and D, the 2F_o-F_c“omit” electron density contoured at 1σ is also shown (atoms from bound K-PLP cofactor, PLP-GABA and GABA product omitted from density calculation). The K405-PLP moiety, PLP-GABA atoms, and non covalently bound GABA are all visible. Hydrogen bonds are shown as dotted lines. Water molecules appear as spheres. The Y434 sidechain from the catalytic loop of chain B is also shown in both FIGS. 4(a) and 4(b). Protonation sites C4′ and Cα are labelled; in (D) alternative conformations of bound GABA are shown.

FIG. 5 is a structural superposition of GAD67 with Pig Dopa decarboxylase (PDB entry 1JS3) and E. coli GAD (PDB entry 1PMM).

FIG. 5(
a) shows the “catalytic loop” (residues 430-450) of monomer A that forms a “flap” over the active site of monomer B in GAD67 can be seen. The K405-PLP adduct, PLP-GABA adduct and GABA product are shown as sticks.

FIG. 5(
b) represents the interactions between the catalytic loop and adjacent monomer. Hydrogen bonds are shown as dotted lines. Water molecules are drawn as spheres. Residues that are different in GAD65 are lightly shaded. The alternative conformation of Y434 in monomer B is also shown. Residues 432-442 are disordered in GAD65.

FIG. 6 are graphs relating to inactivation of GAD in the presence of glutamate as described in Example 1 (below).

FIG. 6(
a) shows a comparison of % residual activity of WT GAD65 and GAD67 before and after incubation with glutamate.

FIG. 6(
b) shows a comparison of % residual activity of GAD65 and GAD65 mutants.

FIG. 6(
c) show a comparison of % residual activity of GAD67 and GAD67 mutants.

FIG. 7 depicts the sequence alignment of GAD65 and GAD67 created using ALLSCRIPT (G. J. Barton, ALLSCRIPT: A tool to format multiple sequence alignments, Protein Eng. (1993) 6, 37-40). Residue numbering is indicated for both sequences. Shaded boxes indicate residues deleted in the constructs used in this study (GAD67_Δ1-89and GAD65_Δ1-83). Differences in the primary sequences of GAD65 and GAD67 are indicated by boxes. The GAD67 secondary structure is shown above the sequence. GAD65 secondary structure is almost identical to GAD67, however, there is a difference located in the s3C β-strand of GAD67 (residues XX-YY), which in GAD65 forms two contiguous smaller strands.

GAD65 residues that form the GAD65 monoclonal autoantibody-binding epitope. Active site residues are boxed and critical functional residues boxed in black. Residues mutated in this study are indicated by an asterisk. The catalytic loop is labelled.

FIG. 8 is the molecular surface of GAD67 dimer and GAD67 dimer interface shaded according to sequence conservation between GAD67 and GAD65. The molecular surface of the GAD67 dimer and GAD67 dimer interface are shaded according to sequence conservation between GAD67 and GAD65. Light shading=100% conserved, dark shading=50% conserved, red=non-conserved. Calculated using ESPript and displayed using CCP4MG (L. Potterton et al, Developments in the CCP4 molecular-graphics project, Acta Crystallogr D. Biol. Crystallogr. (2004) 60, 2288-94).

FIG. 9 is the molecular surfaces of GAD65 and GAD67 shaded according to electrostatic potential. Molecular surfaces of GAD65 and GAD67 are shaded according to electrostatic potential(ie positive or negative). Calculated using CCP4MG.

FIG. 10 depicts the GAD65 catalytic loop interactions. The “catalytic loop” (residues 430-450) of monomer A forms a “flap” over the active site of monomer B (green) in GAD67. The PLP moiety in the active site is shown as sticks.

FIG. 11 shows a structural superposition of GAD67 with Pig Dopa decarboxylase (PDB entry 1JS3) and E. coli GAD (PDB entry 1PMM). Structural superposition of GAD67 with Pig Dopa decarboxylase (PDB entry 1JS3) and E. coli GADB (PDB entry 1PMM). (A) cartoon showing backbone. RMSD's=2.4 Å/347 residues (GAD67/GadB), 2.2 Å/422 residues (GAD67/DDC); (B) Active site residues of GAD67 and DDC. carbiDOPA inhibitor bound to DDC, GABA (GAD67) and K405-PLP-GABA (GAD67) covalent inhibitor are each shaded differently. Tyr434 from GAD67 catalytic loop is shown in dark shaking. Both GAD67 and DDC numbering is shown. The proposed H-bond between Tyr434/332 and the quinonoid Ca in DDC is shown as a dotted line.

FIG. 12 depicts a proposed reaction mechanism of GAD.

FIG. 12(
a) shows the proposed mechanism for PLP-dependent formation of the Schiff base (Enz(PLP-Glu)) between PLP and glutamate and decarboxylation to give the quinoid (Enz(Quinoid)).

FIG. 12(
b) shows how a GAD holoenzyme (i.e. PLP bound) catalyses the decarboxylation of glutamate bound to PLP; subsequent to the decarboxylation reaction, two alternate pathways have been characterised. The majority of the bound quinoid intermediate (or external aldimine) is converted to GABA alongside regeneration of the holoenzyme. Alternatively, decarboxylation-dependent transamination has been observed, where protonation of the C4′ of PLP (instead of the Cα of the quinoid intermediate that results in GABA production) leads to the formation of succinate semialdehyde (SSA), pyridoxamine phosphate (PMP) and an inactive apoenzyme lacking PLP. Kinetic analysis reveals that the steps subsequent to product release (rather than the initial decarboxylation) are rate limiting. Further, the efficiency of decarboxylation-dependent transamination varies considerably between the two GAD isoforms, and therefore it appears to account for different physiological roles as well as different proportions of active holoenzymes in cells. It has been suggested that this difference is a result of different orientations of one or more proton-donating groups in the active site or, differences in the shielding of C4′ from solvent.

DETAILED DESCRIPTION OF THE INVENTION
Structure Determination And Analysis

The first crystal structures of GAD65 and GAD67 and their active binding domains have been determined to 2.3 Å resolution.

Protein Production And Crystallisation

The coding sequences of human GAD65 and GAD67, residues 83-585 and 89-594, respectively, were expressed in Saccharomyces cerevisiae as fusions to a C-terminal hexahistidine tag. Glutamate and PLP were added to all buffers. Recombinant proteins were purified from the cell lysate by immobilized metal affinity chromatography followed by size exclusion chromatography. Enzyme activity was measured by the CO₂trapping method using benzethonium hydroxide as the trapping agent. Data generated were analysed using Prism and Ministat.

Prior to crystallization, purified holoenzymes were concentrated to 10 mg ml⁻¹and equimolar chelidonic acid was added. The proteins were crystallized by the hanging drop method. GAD65 was crystallized in 20% (v/v) ethanol, 100 mM MES (pH6.2), 10 mM 2-mercaptoethanol and 20 mM CaCl₂, and GAD67 in 18% (w/v) PEG 8,000, 100 mM MES (ph 6.3), 10 mM 2-mercaptoethanol and 20 mM CaCl₂, at 20° C.

Mutagenesis

Amino acid substitutions were introduced into the GAD65 and GAD67 sequence using Quick-change mutagenesis kit (Stratagene). The amino acid substitutions were: X, Y and Z. All mutant proteins were prepared as described for the wild-type protein. The mutants and their forward and reverse primers are listed in Table 1.

TABLE 1

Mutants
Forward primer
Reverse primer

GAD65_F283Y
5′ CATAGTCATTATTCTCTCAAGA
5′ CTTGAGAGAATAATGACTATGTTCAGACG

AGGGAGCTG

GAD65_67loop
5′ CAACCAAATGTGTGCCGGATA
5′ CAGGTCATACTGTTTATCTGGCTGAAAGA

CCTCTTTCAG
GGTATCCGGCACACATTTGGTTGCAATTCCATC

CCAGATAAACAGTATGACCTGTCC

TATG

GAD65_Y425F
5′ GCCTCCTTCCTCTTTCAGCAAG
5′ CTGAAAGAGGAAGGAGGCATGCATTTGG

ATAAAC

GAD67_Y292F
5′ CAGAGTCACTTTTCCATAAAGA
5′ CTTTATGGAAAAGTGACTCTGTTCTGAGGTG

AAGCTGGG

GAD67_65loop
5′ CCAGATGCATGCATCCTACCTC
5′ GACATCATAATGCTTGTCTTGCTGGAAGAGG

TTCCAGCAA
TAGGATGCATGCATCTGGTTGCATCC

GACAAGCATTATGATGTCTCCTAC

G

GAD67_Y434F
5′ GCAGGATTCCTCTTCCAGCCAG
5′ GGAAGAGGAATCCTGCACACATCTGG

ACAAGC

Enzyme Activity Assay For GAD65Δ1-83

The procedure for measuring enzymatic activity of GAD65Δ1-83 was based on that described previously and was performed using an anion exchange resin AG1-X8 (Bio-Rad, Hercules, Calif.) to separate ³H-GABA from ³H-glutamate (Amersham) substrate by GAD65Δ1-83 catalysis in a reaction mix after 30 minutes at 37° C. Briefly, a stock solution (0.5M KH₂PO₄pH 7.2, 10 mM 2-mercaptoethanol, 2 mM PLP, 10 mM AET and 100 mM glutamate). The reaction started after adding a solution containing a mixture of 200,000 cpm of ³H-glutamate and 100 mM of glutamate, to an eppendorf containing stock solution and purified GAD65Δ_1-83. The reaction was stopped by the addition of 0.25M of H₂SO₄. For each tested sample, there were duplicate tubes both pre-stops (0 min) and active enzyme (30 min). After incubation, 500 μl of slurry (w/v) containing anion exchange resin in MQ water was added to each reaction. Eppendorfs were centrifuged at 2000 rpm for 1 minute at room temperature and 300 μl of supernatant was collected. 1 ml of scintilant was added to the sample and counted using a counting machine (WALLAC 1409 Liquid Scintillant Counting). Glutamate decarboxylase from rat brain was used as positive control for the enzyme assay experiments. Animal was killed by decapitation and brain was removed and homogenized with the stock buffer used in the enzyme assay. The homogeneous material was centrifuged in a TL-100 ultracentrifuge (Beckman) at 10,000 g for 10 minutes, and the clarified supernatant was collected and centrifuged again for 10 minutes and used in the assay run.

X-ray Data Collection, Structure Determination And Refinement

Data were collected at the IMCA-CAT beamline at the Advanced Photon Source, Chicago, USA. Both GAD65 and GAD67 crystals diffracted to 2.3A resolution. GAD67 crystals belong to space group P2₁, and have unit cell dimensions of a=84.05 Å, b=62.74 Å, c=101.35 Å, β=106.7°, consistent with two molecules per asymmetric unit; GAD65 crystals belong to space group C222₁, and have unit cell dimensions of a=78.25 Å, b =99.06 Å, c=120.1 Å, consistent with one molecule per asymmetric unit The data were merged and processed using MOSFLM and SCALA. (P. Evans, Scaling and assessment of data quality, Acta Crystallogr. D. Biol. Crystallogr (2006) 62, 72-82; A. Leslie, Joint CCP4+ESF-EAMCB Newsletter on Protein Crystallography (1992) 26). Subsequent crystallographic and structural analysis was performed using the CCP4i interface (E. Potterton et al, A graphical user interface to the CCP4 program suite, Acta Crystallogr. D. Biol. Crystallogr (2003) 59, 1131-7) to the CCP4 suite (The CCP4 suite: programs for protein crystallography, Acta Crystallogr. D. Biol. Crystallogr. (1994) 50, 760-3), unless stated otherwise. Five percent of the dataset was flagged for calculation of the free R factor (R_free) with neither a sigma, nor a low-resolution cut-off applied to the data. A summary of the data Collection and refinement statistics are provided in Table 2.

TABLE 2

GAD67
GAD65

Data collection

Space Group
P2_L
C222_L

Cell dimensions (Å): a, b, c
84.05, 62.74,
78.25, 99.05,

101.35, β = 106.69
120.01

Resolution (Å)
97.1-2.3
54.64-2.3

Molecules per asymmetric unit
2
1

Total number of observations
141888
78118

Number of unique observations
42284
20717

Multiplicity
3.4 (2.3)
3.8 (3.8)

Data Completeness (%)
93.7 (69.2)
98.8 (98.8)

<I/σI>
17.1 (4.4)
14.9 (4.7)

R_pim(%)^b
4.5 (19.4)
3.5 (16.3)

Structure refinement

Nonhydrogen atoms

3881

Solvent
359
93

R_free(%)^c
21.4
25.1

R_cryst(%)
17.8
19.5

Rms deviations from ideality

Bond lengths (Å)
0.008
0.009

Bond angles (°)
1.2
1.3

B factors (Å²)

Mean main chain
31.2
52.6

Mean side chain
32.8
54.3

Mean water molecule
34.7
52.4

r.m.s. deviation bonded Bs
0.7
0.7

^aValues in parentheses refer to the highest resolution shell.

^bAgreement between intensities of repeated measurements of the same reflections and can be defined as: Σ(I_{h, i}− <I_h>)/ΣI_{h, i}, where I_{h, i}are individual values and <I_h> is the mean value of the intensity of reflection h.

^cThe free R factor was calculated with the 5% of data omitted from the refinement.

The structure of GAD67 was solved using the molecular replacement method and the program PHASER (A. McCoy et al, Simple algorithm of a maximum likelihood SAD function, Acta Crystallogr. D. Biol. Crystallogr (2004) 60, 1220-8). A search model was constructed from the crystal structure of Pig Dopa Decarboxylase (DDC; PDB identifier 1JS3) (REF), the closest structural homologue identified using the FFAS server (L. Jaroszewski et al, FFAS03: a server for profile-profile sequence alignments, Nucleic Acids Res (2005) 33, W284-8) (sequence identity=20%). The structure was trimmed to remove regions of high sequence divergence, leaving predominantly residues belonging to the PLP-binding domain (representing ˜60% of the total GAD67 structure). A “mixed” model consisting of conserved sidechains (all other non alanine/glycine residues truncated at Cγ atom) was then created using the SCRWL server (A. Canutescu et al, A graph-theory algorithm of rapid protein side-chain prediction, Protein Sci (2003) 12, 2001-14). Two outstanding solutions having Z-scores of 12 and 10 were produced, and packed well within the unit cell. Together with the unbiased features in the initial electron density maps, the correctness of the molecular replacement solution was confirmed.

Structure refinement and model building proceeded using one molecule in the asymmetric unit (the other Non-Crystallographic-Symmetry (NCS)-related molecule generated using NCS operators). Maximum likelihood refinement using REFMAC (G. Murshudov et al, Refinement of macromolecular structures by the maximum-likelihood method, Acta Crystallographica (1997) D53, 240-255), incorporating translation, libration, and screw-rotation displacement (TLS) refinement was carried out, employing a bulk solvent correction (Babinet model with mask). Throughout most stages of refinement, tight NCS-restraints were imposed on all residues in the two molecules in the asymmetric unit. At the later stages of refinement. All model building and structural validation was carried out using COOT (P. Emsley et al., K. Coot: Model-building tools for molecular graphics, Acta Crystallogr. D. Biol. Crystallogr. (2004) 60, 2126-32) Water molecules were added to the model using ARP/WARP (R. Morris et al, ARP/wARP and automatic interpretation of proteins electron density maps, Methods Enzymol, (2003) 374, 229-44) when the R_freereached 30%. Solvent molecules were retained only if they had acceptable hydrogen bonding geometry contacts of 2.5-3.5 Å with protein atoms or with existing solvent, and were in good 2F_o-F_cand F_o-F_celectron density.

The structure of GAD65 was determined by molecular replacement using PHASER and the refined GAD67 model. Refinement proceeded as for GAD67.

Structural Analysis

PYMOL (W. DeLano, The PyMOL User's Manual from DeLano Scientific, San Carolos, Calif., USA (2002)) was used to produce FIGS. 2, 4 and 5. Structures were superimposed using the program MUSTANG (A. Konagurthu et al, Function, and Bioinformatics, MUSTANG: A multiple structural alignment algorithm; Proteins: Structure, Function, and Bioinformatics (2006)). Accessible surface areas were calculated using the CCP4 program AREAIMOL.

Overall Description of the X-Ray Crystal Structures of GAD65 And GAD67

In order to understand the molecular regulation of GABA production, the crystal structures were determined for a truncated form of each isoform (referred to as GAD65 and GAD67 hereafter) that lack the first 83 and 89 residues, respectively. This is depicted visually in FIG. 7 for GAD65 (Seq. 1) and for GAD67 (Seq. 2). A comparative analysis of the sequence differences between GAD65 and GAD67 is set out in Table 3.

TABLE 3

Position, N-
Exposed (E)/
Exposed (E)/
Active site (C); Impli-

Residue No.

terminal domain (N),
Buried (B)/
Buried (B)/
cated in AutoAb binding

(GAD65 number

PLP domain (PLP)
Interface (I)*;
Interface (I)*;
(Ab); Implicated in T cell

in parenthesis)
GAD67 GAD65
and C-terminal (C).
GAD67 residues.
GAD65 residues.
epitope (T); Other (O).

93(87)
Thr Val
N
S

94(88)
Asp Asn
N
S
S

95(89)
Phe Tyr
N
B
S

96(90)
Ser Ala
N
S
S

97(91)
Asn Phe
N
S
S

99(93)
Phe His
N
S
S

101(95)
Arg Thr
N
S
S

107(101)
Lys Cys
N
S
S

108(102)
Asn Asp
N
S
S

111(105)
Glu Arg
α1 N
S
S

112(106)
Gln Pro
α1 N
S
S

114(108)
Val Leu
α1 N
S, I
S

115(109)
Gln Ala
α1 N
S
S

118(112)
Leu Gln
α1 N
S
S, I

119(113)
Glu Asp
α1 N
S
S

121(115)
Val Met
α1 N
B, I
B, I

122(116)
Asp Asn
α1 N
B
S
T¹

126(120)
Asn Gln
α1 N
S
S
T¹

129(123)
Arg Val
α1 N
S, I
S
T¹

131(125)
Thr Ser
α1 N
B
B, I
T¹

139(133)
Leu Ile
N
B
B

143(137)
His Tyr
N
S, I
S, I

145(139)
His Asn
α2, N
S
S

146(140)
Gln Glu
α2, N
S
S

149(143)
Glu Gln
α2, N
S
S

150 (del.)

151(del.)

152(del.)

153(144)
Gly Glu
N
S
S

154(145)
Phe Tyr
N
B
S

156(147)
Leu Trp
N
B, I
B, I

159(150)
Ser Ala
N
S, I
S, I

161(152)
His Gln
N
S
S

163(154)
Glu Gln
N
S, I
S, I

164(155)
Ser Asn
N
S
S

167(158)
Gln Glu
α3 N
S
S

170(161)
Val Met
α3 N
S
S

171(162)
Asp His
α3 N
B
B

173(164)
Arg Gln
α3 N
S
S

174(165)
Asp Thr
α3 N
S
B

179(170)
Gly Ala
N
S
B

180(171)
Val Ile
N
B, I
B, I

181(172)
Arg Lys
N
S
S

187(178)
Phe Tyr
s1A, N
B
B, I

197(188)
Ile Met
α4 N
B, I
B, I

198(189)
Ile Val
α4 N
B, I
B

202(193)
Gly Ala
α4 N
B
B

203(194)
Glu Asp
α4 N
B, I
B, I

225(216)
Met Leu
α5 PLP
B, I
B

227(218)
Gln Tyr
α5 PLP
S
S

228(219)
Ile Val
α5 PLP
S, I
B, I

237(228)
Val Ile
α5 PLP
B
B

240(231)
Ser Pro
PLP
S
S
Ab¹, mAb DPC, structural

differences between isoforms

in this loop

241(232)
Ser Gly
PLP
S
S
structural differences

between isoforms in this loop

242(233)
Lys Gly
PLP
S
B
structural differences

between isoforms in this loop

243(234)
Asp Ser
PLP
S
S
Ab¹, mAb DPC, structural

differences between isoforms

in this loop

259(250)
Ser Ala
α6 PLP
B
B

260(251)
Ile Met
α6 PLP
B
B

262(253)
Ala Ile
α6 PLP
B
B

265(256)
Tyr Phe
α6 PLP
S
S

267(258)
Tyr Met
α6 PLP
S
S

273(264)
Thr Glu
α7 PLP
S, I
S, I
Ab¹, mAb M10 and M6

279(270)
Val Leu
PLP
B
S

281(272)
Lys Arg
PLP
S, I
S

283(274)
Val Ile
s2B, PLP
B
B
T¹

284(275)
Leu Ala
s2B, PLP
B
B
T¹

289(280)
Gln His
PLP
S
S
T¹, C(within 6 Å sphere)

292(283)
Tyr Phe
α8 PLP
S, I
S
T¹, C(within 6 Å sphere)

294(285)
Ile Leu
α8 PLP
B
B

297(288)
Ala Gly
α8 PLP
B
B

298(289)
Gly Ala
α8 PLP
B
B

303(294)
Phe Ile
PLP
B
B
Ab⁴

307(298)
Asn Ser
PLP
B
B
Ab⁴

314(305)
Asn Asp
s4B, PLP
S
S
Ab², mAb 96.11

319(310)
Ile Met
s5B, PLP
B
B

322(313)
Ala Ser
α9 PLP
S
S

324(315)
Phe Leu
α9 PLP
B
B

326(317)
Ala Arg
α9 PLP
S
S
Ab¹, Ab³, T1D sera

327(318)
Lys Arg
α9 PLP
S
S
Ab³, T1D sera

336(327)
Tyr Phe
PLP
S
S

340(331)
Tyr Leu
s6B, PLP
B
B
C¹(within 6 Å)

342(333)
Asn Ser
s6B, PLP
B
B
C(within 6 Å)

356(347)
Ile Leu
α10 PLP
B
B

357(348)
Gln Leu
α10 PLP
S
S

358(349)
Glu Ala
α10 PLP
S
S

359(350)
Ile Val
α10 PLP
B
B

364(355)
Glu Lys
α10 PLP
S
S

367(358)
Asn Lys
PLP
S
S
Ab¹, mAb M4

368(359)
Leu Ile
PLP
B
B

370(361)
Leu Met
s7B, PLP

387(378)
Arg Lys
PLP
S
S

388(379)
His Trp
PLP
S
S

391(382)
Asn Ser
PLP
S
S

393(384)
Ile Val
PLP
B
B

410(401)
Leu Pro
PLP
B, I
B

416(407)
Ile Leu
s9B, PLP
B
B
C(within 6 Å)

419(410)
Lys Arg
PLP
S
S, I

421(412)
Lys Glu
PLP Connection between
S
S
Ab¹, mAb DPC

active sites

423(414)
Ile Leu
PLP Connection between
B
B

active sites α11

424(415)
Leu Met
PLP Connection between
B
B

active sites α11

426(417)
Gly Gln
PLP Catalytic loop
B
S

431(422)
Cys His
PLP Catalytic loop
S, no bonds within
S

5 A distance

433(424)
Gly Ser
PLP Catalytic loop
S, no bonds within
No struct.

5 A distance

436(427)
Leu Phe
PLP Catalytic loop
S, no bonds within
No struct.
T¹

5 A distance

438(429)
Pro Gln
PLP Catalytic loop
S, no bonds within
No struct.
T¹

5 A distance

441(432)
Gln His
PLP Catalytic loop
S, no bonds within
No struct.
T¹

5 A distance

444(435)
Val Leu

S, no bonds within
S
T¹

5 A distance

453(444)
Ile Leu
PLP Catalytic loop
B
B

461(452)
Ile Val
α12, PLP
B
B

464(455)
Phe Leu
α12, PLP
B
B

469(460)
Lys Arg
α12
B, I
B, I

474(465)
Val Thr
α13 C
S, I
S, I

478(469)
Asn Ala
α13 C
S
S

479(470)
Gln His
α13 C
S
S
Ab¹, mAb M8, M9

480(471)
Ile Val
α13 C
B
B

481(472)
Asn Asp
α13 C
S
S

492(483)
Ala Asn
α13 C
S
S
Ab¹, T¹, mAb, M7, M3 and M1

493(484)
Lys Ile
α13 C
S
S
T¹

499(490)
Glu Gly
C
S
B
T¹

500(491)
Phe Tyr
s1C, C
B
S

505(496)
Asn Asp
C
S
S

507(498)
Glu Lys
C
S
S
Ab², T¹, mAb b78

509(500)
Glu Gln
C
S
S
Ab², mAb b78

520(511)
Gln Pro
C
S
S
T¹

524(515)
Gly Thr
C
S
S
T¹

525(516)
Val Leu
C
S
S
T¹

526(517)
Pro Glu
C
S
S
Ab¹, Ab³, T¹, mAb M2 and M5

528(519)
Ser Asn
C
S
S
T¹

529(520)
Pro Glu
α14 C
S
S
Ab¹, T¹, M2 and M5

530(521)
Gln Glu
α14 C
S
S
Ab¹, T¹, M5?

532(523)
Arg Met
α14 C
S
No struct.
T

533(524)
Glu Ser
α14 C
S
No struct.
Ab¹, Ab³, T¹, M5

534(525)
Lys Arg
α14 C
S
No struct.
Ab³, T¹

536(527)
His Ser
α14 C
S, I
S
Ab¹, M5

541(532)
Lys Val
α14 C
S
S
Ab¹, b78 and M5

545(536)
Leu Arg
α14 C
S
S
Ab¹, M8 and M9

549(540)
Ser Tyr
α14 C
S
S
Ab¹, M8 and M9

555(546)
Gly Ser
s3C, C
B
S

559(550)
Gln Leu
s3C, C
S
S

563(554)
Ala Val
s4C, C
B
S

577(568)
Gln His
C
S
S
Ab¹, M7

578(569)
Ser Gln
C
S
S

GAD65 mutations with effect on antibody recognition:

Ab¹- Schwartz et al., 1999. Glu264 is essential for mAb M10 epitope but not M4 and is replaced by threonine in GAD67. Lys358 completely abolished M4 reactivity (replaced by asparagine in GAD67). Asn483 and His568 are very important for the binding of M1, M3 and M7 antibodies. Glu517, Glu520 and Glu521 are essential for M2 and M5 epitope. Ser524 and Ser527 are critical for M5 recognition only. V532K affected M5 and b78 binding and is consistent with the inability of mAb b78 to inhibit GAD65 enzyme activity (unpublished?). Pro231 and Ser234 are important for DPC epitope and localized in a region of structural differences between GAD67 and GAD65. The location of this epitope away from the active site/catalytic loop is consistent with its inability to inhibit GAD65 activity. Arg536 and Tyr540 are critical for binding M8 and M9 only, and H470Q mutation affected these antibodies epitopes. Trp379 and Glu412 are also implicated in DPC epitope. Arg317 is critical for two GAD65-specific mAb (referred in text as unpublished) and also decrease reactivity of T1D patients by 17% (Myers, Fenalti.. et al.). Leu574 -Pro and Asn247Ser results in the loss of binding for all eleven tested antibodies in a another study (this may not be so relevant for Ab binding but could be important for protein folding or maintaining conformation, similar to residue). Leu574 -Pro affected C-terminal epitopes of M1, M3 and DPA. If this Leu is close to His362-need to check- the explanation is that it affects the folding since it is critical for folding in class II decarboxylases (Capitani et al., 2003).

Ab²- Mutation 305DER307, and 498KPQ500 decreased reactivity with b96.11 by 38% and 55% respectively. (Fenalti et al, 2006)

Ab³- O'Konnor et al., (2006). 524SRL526 mutated to AAA caused drastic reduction in reactivity with b78 (inhibts GAD65 activity). Although 572DF573 are the same in GAD67 and GAD65, mutation of these residues to A completely abolished reactivity with b78 and b96.11.

Ab⁴- Mutation 294IGTDS298 (note that Gly, Thr and Asp are in the interface of the dimer) completely abolished b96.11 and b78 antibody binding.

*3.6 Angstroms being the maximum distance between atoms used to calculate residues in the dimer interface.

T¹- Patel et al., (Identification of immunodominant T cell epitopes of human glutamic acid decarboxylase 65 by using HLA-DR (alpha1*0101, beta1*0401) transgenic mice. Proc Natl Acad Sci USA, 1997 Jul. 22; 94(15): 8082-7

C¹- Within 6 Å radius from PLP ring

Preliminary studies revealed that the full length constructs were not suitable for structural studies because the N-terminal region is most likely unstructured and is extremely sensitive to proteolytic degradation. Previous studies as well as kinetic analysis revealed that N-terminally truncated GAD has comparable activity to full-length material.

The 2.3 Å structures of GAD67 and GAD65 are shown in FIG. 1 and FIG. 2. Both molecules form obligate functional dimers that bury a surface area of 6816 Å²and 5662 Å²respectively. The monomeric unit is made up of three distinct domains (termed N-terminal, PLP-binding and C-terminal; FIG. 1). The overall quality of the density in GAD67 is excellent, with all regions resolved in electron density. In GAD65 two loops (423-433 in the PLP domain and 518-520 in the C-terminal domain) are disordered. The contribution each domain makes to the dimer interface, and descriptions of the nature of the interfaces is detailed in the Table 4 which records the physical and chemical nature of the dimer interfaces. (Note that the discrepancy between the number of interface residues and surface areas of GAD67 and GAD65 dimers can be accounted for by the disorder of residues 423-433 in the PLP domain and 518-520 in the C-terminal domain of GAD65, both of which contribute to the dimer interface.)

TABLE 4

GAD67
GAD65

Buried Surface Area
6816
5662

at interface

% Polar Atoms in
29
26

Interface

% Non-Polar Atoms in
71
74

Interface

Hydrogen Bonds
42
36

Total residues/
1004/320
(32%)
966/284
(29%)

interface residues (%)

Interface residues per domain

(% composition per domain)

N-terminal
102
(32%)
102
(36%)

PLP-binding
186
(58%)
162
(57%)

C-terminal
32
(10%)
20
(7%)

The N-terminal domain includes two parallel helices that pack against the N-terminal and PLP-binding domain of the other monomer. The PLP-binding domain adopts the type I PLP-dependant transferase-like fold and comprises nine helices surrounding a 7-stranded mainly parallel β-sheet which can be seen clearly in FIG. 1. The C-terminal domain contains the four remaining helices, together with a short 2 strand antiparallel β-sheet.

GAD65 and GAD67 are 71% identical in the region structurally characterised, adopt the same fold and superpose with an r.m.s.d of 0.8 Å²(474 residues; as shown in FIG. 2a). They do, however, differ in two important regions. Rigid body shifts of 1.5 Å are observed in the C-terminal domain of GAD65 with respect to GAD67 (see FIGS. 2a and 2b). Further, B-factor analysis reveals that this region is substantially more mobile in GAD65 and contains a region not visible in electron density (residue 518-520; FIG. 2). Importantly, residues 423-433 in the PLP domain of GAD65 are also not visible in electron density. This latter region accounts almost entirely for the difference of 1150 Å²interfacial surface area between the isoforms. In GAD67 the region corresponding to residues 423-433 forms a well-ordered loop that sits on top of the active site cleft as can be seen in FIG. 1. These differences are discussed below in relation to both GAD enzyme function and autoantigenicity.

Location of B-Cell Epitopes On GAD65

Immune tolerance to GAD65 is labile since autoantibodies to GAD65 are detectable characteristically in Type 1 diabetes, whereas tolerance to GAD67 is solid; no structural basis for this differential autoreactivity has been discerned. Hence we mapped the differences in sequence onto the structure of each enzyme. The surface-exposed differences in sequence between the two isoforms were distributed over the entire structure of the molecule, with no obvious clustering apparent in any region as is evident from FIG. 8a. The differences in surface electrostatic potential of GAD isoforms are scattered throughout the molecule and do not cluster in any region and this can be seen in FIG. 9. However when we examined the location of amino acids known to influence binding of 15 human mAbs to GAD65 on the structure of the GAD65 dimer, epitopes for specific mAb localised in two distinct regions (FIG. 9). Interestingly, 10 of 15 mAb mutations that affect antibody binding localise to the C-terminal domain, or the interface between the C-terminal domain and the PLP binding domain (FIG. 3). The C-terminal domain is more mobile in GAD65 in comparison to GAD67, undergoes a 1.5 Å rigid body shift and includes a disordered region (FIG. 2).

The Active Site of Both GAD65 And GAD67 Reveals A Product Complex With GABA

Each GAD dimer contains two active sites, located in the middle of the PLP-binding domain at the dimer interface as shown in FIG. 1.

In summary, the active site is approximately tetrahedral in shape, and is constructed from residues in the PLP and C-terminal domains of both monomers. Two sides of the active site are formed by residues from the C-terminal domain and the PLP binding domain of monomer A respectively, the third face is formed by residues from the PLP binding domain of monomer B. PLP is located at the base of the cleft and is covalently attached via a Schiff base linkage to the absolutely conserved residue Lys 405 of monomer A (GAD67 numbering; FIG. 7). The conserved catalytic His 291 is found in the active site positioned on top of the PLP moiety as shown in FIG. 4.

GAD was purified in the presence of the substrate glutamate and examination of the active sites of both GAD65 and GAD67 revealed electron density consistent with the presence of the product GABA. This can be seen in FIG. 4. Continuous electron density in active sites of both GAD65 and GAD67 is compatible with the presence of a PLP cofactor covalently bonded to the active site lysine (K405) via a Schiff-base linkage (termed the internal aldimine; FIG. 4a). Notably, however, in the active site of GAD67 (monomer A) two discretely disordered conformations of GABA are observed. In one conformation, the carboxyl group of GABA forms a salt bridge with Arg567 and a hydrogen bond with Gln190. In the other conformation, continuous density between PLP and GABA reveals the presence of a trapped intermediate (termed the ‘quinoid’; FIG. 4a, b) The second active site of GAD67 (monomer B) contains only a single GABA molecule (FIG. 4b). In both active sites of GAD65, GABA adopts two discretely disordered conformations (FIG. 4d) that form analogous interactions to those seen in GAD67. However, no density was observed that suggested the presence of a quinoid intermediate in GAD65 (FIG. 4d).

Comparison of the Active Site of GAD67 And GAD65 Reveals the Structural Basis For the Regulation of GABA Production

GAD65, but not GAD67 has been reported to readily form the inactive apo-form and release PMP and succinic semialdehyde. We therefore attempted to identify key features of GAD67 that may explain this activity. Strikingly, the structure of GAD67 revealed that each active site is substantially occluded by an extended loop (residues 430-450, termed the ‘catalytic loop’) contributed in trans from the other monomer (FIGS. 1, 2a, 2b, 5, and FIG. 10). The catalytic loop is well ordered in electron density, buries ˜1000 Å at its interface, and makes 12 hydrogen bonds, 4 water mediated hydrogen bonds, 2 salt bridges and 5 hydrophobic interactions with the body of GAD67 (FIG. 10). The interactions between catalytic loop (430-450) and the rest of the protein are listed in Table 5.

TABLE 5

Loop residue
Partner
Type of bond

Interactions with other monomer

Tyr434
Gln557
vdw

Tyr434^Oη
Tyr292^N(His291^Nε2
h-bond

in other monomer)

Gln437^Nε2
Tyr556^O
h-bond

Asp439^Oδ1
Tyr556^Oη
h-bond

Asp439^Oδ2
Asn564^Oδ1
h-bond

Asp439^O
Tyr556^Oη
h-bond

Gln441
Ala544
vdw

Tyr446
Pro144
vdw

Asp450^Oδ1
Lys296^NZ
h-bond*

Asp450^O
Lys296^NZ
h-bond

TOTAL: 3 vdw, 7 h-bonds

Interactions within loop

Met430^N
Asp450^{Oδ1, Oδ2}
2 h-bonds

Met430^O
Ala432^N
h-bond

Cys431^N
Asp450^Oδ2
h-bond

Ala432^O
Phe436^N
h-bond

Tyr434^N
Gln437^Nε2
h-bond

Leu435^N
Gln437^Oε1
h-bond

Phe436
Gly449
vdw

Pro438^O
Lys440^N
h-bond

Lys440^NZ
Asp447^Oδ2, Tyr442^Oη
2 h-bonds*

Tyr442^Oη
Asp447^{Oδ1, Oδ2}
2 h-bonds

Tyr442^O
Val444^N
h-bond

Asp443^Oδ1
Ser445^{N, Oη}, Val444^N
3 h-bonds

Asp443^Oδ2
Ser445^Oη
h-bond

Asp443^O
Ser445^N, Tyr446^N
2 h-bonds

Val444^O
Asp447^N, Tyr446^N
2 h-bonds

Ser445^Oη
Tyr446^N
h-bond

Tyr446^O
Thr448^N
h-bond

Thr448^O
Asp450^N
h-bond

TOTAL: 1 vdw, 24 h-bonds

Interactions between loop and rest of monomer

Met430^N
Asn428^O
h-bond

Lys440^NZ
Glu217^Oε2
h-bond*

Tyr442^Oη
Glu217^N
h-bond

Asp447^Oδ1
Tyr216^N
h-bond

Thr448^N
Phe214^O
h-bond

Gly449^N
Phe214^O
h-bond

Asp450^O
Gln429^N
h-bond

Thr448
Lys451
vdw

Thr448^O
Lys451^N
vdw

Gly449^O
Lys451^N, Ala452^N
2 h-bonds

TOTAL: 2 vdw, 9 h-bonds

GRAND TOTAL: 6 vdw, 40 h-bonds (3 charged)

^a*= charged interaction

Importantly, the catalytic loop brings the conserved residue Tyr 434 into the centre of the active site, in close proximity to the catalytic histidine 291 (FIG. 4a, b, d). Two conformations (termed A and B) of Tyr 434 are observed. In chain A the hydroxyl group of

Tyr 434 hydrogen-bonds to the backbone nitrogen of Tyr 292 (the A conformation), whereas in chain B Tyr 434 is flipped to an alternative conformation (B) where its sidechain hydroxyl group forms a hydrogen bond to the Nε2 of the catalytic His 291 (FIG. 4a, b, d, e). In the B conformation the sidechain hydroxyl group of Tyr 434 is 2.8 Å from Ca atom of the PLP-GABA moiety

In contrast, the structure of GAD65 reveals that the catalytic loop is flexible and is not visible in electron density. As a result the active site of GAD65 is completely exposed (FIG. 2b and FIG. 5a). A structural and sequence comparison between the two isoforms reveals that the majority of residues contacted by the catalytic loop in GAD67 are identical in sequence and conformation in GAD65 (FIGS. 5b, 7, and 8b). One important exception is Phe 283, which corresponds to Tyr 292 in GAD65 and forms contacts with Tyr 434. Further, several non-conservative substitutions map to the catalytic loop of GAD65. Most notably, GAD65 contains a GlySer substitution at position 433 and a ProIn substitution at position 438. Pro438 is centred on a sharp kink in the catalytic loop and Gly 433 in GAD67 adopts a “+/+” conformation in chain A (FIG. 5b). It is therefore suggested that substitution of both of these residues in GAD65 may destabilise the conformation of the catalytic loop observed in the structure of GAD67.

Kinetic analysis confirms the importance of the catalytic loop in GAD65 auto-inactivation

Role of the Catalytic Loop

In order to investigate the role of the catalytic loop we investigated the specific activity and inactivation of GAD65 and GAD67. Consistent with published literature, our data reveal that whereas GAD67 remains active after incubation with excess glutamate, GAD65 loses 75% of activity in 20 minutes (FIG. 6). GAD65 activity could be restored by addition of PLP (data not shown).

To test the role of the catalytic loop, we generated the mutations GAD65_Y425Fand GAD67_Y434F. Both mutants were inactive with respect to decarboxylation, confirming the essential catalytic role of the conserved tyrosine.

In order to investigate the role of the catalytic loop on the inactivation rate of holoenzymes, the catalytic loops of GAD65 and GAD67 were interchanged by mutagenesis (the mutants are named GAD65_67loopand GAD67_65loop). Whilst the inactivation rate of GAD65_67loopremains similar to that of GAD65, the inactivation rate of GAD67_65loopincreased significantly (p<0.001) (FIG. 6). We also investigated the role of Phe 283/Tyr 292 in GAD inactivation. The inactivation of GAD65_F283Yand GAD67_Y292Fdid not differ significantly from the respective wildtype enzymes. However, it is shown that the rate of inactivation of a GAD65_67loop_—_F283Ymutant is significantly slowed (FIG. 6) and loses 50% rather than 75% of activity after 20 minutes. Similarly it is shown that GAD67_65loop/Y292Fis rapidly inactivated in comparison to GAD67 or GAD67_65loop. While our mutations in GAD65 did not abolish enzyme inactivation, it is interesting to note that the C-terminal domain, against which the catalytic loop packs, is more mobile in GAD65 and has shifted 1.5 Å relative to its position in GAD67. It is therefore suggested that the mobility in the C-terminal domain may also contribute to destabilising the catalytic loop. Finally, it is notable that the position of catalytic loop is in close proximity to many of the residues important for auto-antibody binding in GAD65. Several auto-antibodies have been reported that inactivate GAD65 upon binding.

Tyr 434 Functions As A Key Catalytic Switch

Together, our mutational data show that the sequence of the catalytic loop plays a key role in inactivation of GAD67 and that Tyr 434 is plays a direct role in the catalytic machinery of the enzyme. Further, these data suggest that the stabilised conformation observed in the structure of GAD67 prevents enzyme inactivation and allows continuous GABA production. It is notable that in the A conformation, the hydroxyl group of Tyr 434 would be unable to protonate the PLP Cα atom, since the hydroxyl group is >5 Å away from the Cα position. Consistent with these data, we observe unambiguous density consistent with the quinoid GABA-PLP complex (i.e. an intermediate prior to protonation and GABA release). In the B conformation the hydroxyl group of Tyr 434 is 2.8 Å from the site of protonation (Cα) of the quinoid moiety In this conformation, we observe free GABA in the active site. It is therefore suggested that Tyr 434 is directly responsible for protonating the Cα position and that the H-bond interaction between Tyr434 and His291 in GAD67 favourably raises the pKa of the Tyr434.

Structural studies on the related enzyme, DOPA decarboxylase (DDC), have revealed that, like GAD65, the region corresponding to the catalytic loop in this enzyme is disordered. DOPA decarboxylase catalyses predominantly the decarboxylation of L-aromatic amino acids into the corresponding aromatic amines, as well as half-transaminase and oxidative deaminase side reactions. Biochemical studies on DOPA decarboxylase have revealed that mutation of the equivalent residue to Tyr 434 (Tyr332) to a Phe converts the enzyme into a decarboxylation-dependant oxidative deaminase and promotes PMP release. It has thus been suggested that in DDC Tyr332 performs the protonation of the Ca atom of the quinoid intermediate that is critical for normal enzymatic activity. The position of Tyr 434 in GAD67 as shown in FIG. 11 is consistent with this hypothesis.

Together, our data provide a plausible model for the auto-regulation of GAD. In GAD67 it is suggested that the continuous presence of Tyr434 in the active site favours protonation of the Cα atom and uninterrupted GABA production (FIG. 12). Tyr425 is proposed to play the same catalytic role in GAD65. However, in the latter enzyme, the flexible nature of the catalytic loop ensures that the catalytic tyrosine is only transiently present in the active site. By analogy with DDC, it is suggested that in the absence of the catalytic tyrosine, protonation at C4′ allows semialdehyde/PMP release and enzyme inactivation. In other PLP decarboxylaseas, it has been suggested that water or the catalytic lysine (396/405) may function as a proton donor at the C4′ position. Together the structural data provides an elegant explanation for why GAD67 predominantly acts as a decarboxylase, whereas GAD65 is able to catalyse both decarboxylation and decarboxylation-dependant transamination reactions.

EXAMPLES
Example 1
Investigation of the Inactivation of GAD In the Presence of Glutamate

Purified holoenzymes (20 μg/ml) were preincubated at 30° C. for 20 min with 5 mM glutamate in the presence of 0.1% Triton X-100, 1 mM 2-mercaptoethanol, 1 mM 2-aminoethylisothiouronium bromide, 100 mM K/NaPO4, pH 7.2; enzyme activity was determined by adding L-[1-¹⁴C]-glutamic acid and incubated at 30° C. for 30 min, ¹⁴C0₂produced was trapped with benzethonium hydroxide. % Residual activity is calculated by taking the enzyme activity determined without the glutamate preincubation as 100%. Each bar is the mean of three determinations, with SD illustrated by the error bars. Statistical comparisons were performed using non-paired, two tailed Student's tests and the results are depicted in FIG. 6. FIG. 6(a) shows a comparison of % residual activity of WT GAD65 and GAD67 before and after incubation with glutamate. FIG. 6(b) shows a comparison of % residual activity of GAD65 and GAD65 mutants. FIG. 6(c) shows a comparison of % residual activity of GAD67 and GAD67 mutants.

The word ‘comprising’ and forms of the word ‘comprising’ as used in this description and in the claims does not limit the invention claimed to exclude any variants or additions.

Modifications and improvements to the invention will be readily apparent to those skilled in the art. Such modifications and improvements are intended to be within the scope of this invention.

Lengthy table referenced here

US20100216112A1-20100826-T00001

Please refer to the end of the specification for access instructions.

LENGTHY TABLES

The patent application contains a lengthy table section. A copy of the table is available in electronic form from the USPTO web site (). An electronic copy of the table will also be available from the USPTO upon request and payment of the fee set forth in 37 CFR 1.19(b)(3).

Crystal structures of both isoforms of human glutamic acid decarboxylase

Information

Publication Number

Date Filed

Date Published

Inventors

CPC

US Classifications

International Classifications

Abstract

Description

Claims

Priority Claims (1)

PCT Information