Low-resolution structure of the soluble domain GPAA1 (yGPAA1^70–247) of the glycosylphosphatidylinositol transamidase subunit GPAA1 from Saccharomyces cerevisiae Open Access

GPI (glycosylphosphatidylinositol) lipid anchoring is an alternative post-translational modification for cellular systems to anchor proteins to the outer leaflet of the cell membrane [1,2]. Approximately 0.5% of all proteins in higher eukaryotes are capable of being GPI lipid-anchored [3], including cell surface receptors like urokinase receptor, CD14, CD16, enzymes [ALP (alkaline phosphatase) and 5′-nucleotidase], adhesion proteins and antigens [Thy-1, DAF (decay-accelerating factor) and MACIF (membrane attack complex inhibition factor)]. GPI lipid-anchoring of proteins has medical implications in a variety of conditions, including paroxysmal nocturnal haemoglobinuria, prion disease pathogenesis, AD (Alzheimer's disease) and parasitic diseases such as malaria [1,4].

The GPI lipid anchor is biosynthesized stepwise by a series of enzymes located in the ER (endoplasmic reticulum) membrane [1]. Typically, the nascent substrate proprotein is exported into the ER after ribosomal synthesis via the signal peptide pathway. The so-called transamidase complex in the ER lumen recognizes a specific GPI lipid anchor attachment signal located at the C-terminus of protein substrates [5–7]. This GPI lipid anchor attachment signal consists of a sequence motif from residue ω−11 to the very C-terminus of the protein substrate, with a region of polar residues between ω−11 and ω−1, a short stretch of small polar residues from ω−1 to ω+2, including the ω-site for cleavage and the GPI attachment, a linker region that contains moderately polar residues, followed by a hydrophobic tail from ω+9 to the C-terminus of the proprotein (Figure 1) [1]. During the modification process the C-terminal propeptide after the ω-site is removed and replaced with a pre-synthesized GPI moiety by the GPI transamidase complex [1,8]. The GPI lipid anchor is a complex organic structure made up of a lipid-modified phosphatidylinositol, a tetrasaccharide with variable elaborations and phosphoethanolamine subunits [2,9]. Although the general outline of the GPI lipid anchor biosynthesis pathway is common among eukaryotes, there are distinctive taxon-specific differences that might be useful for species-specific inhibitor design as described for suppression of unicellular parasites such as Trypanosoma [10,11].

The GPI transamidase complex is made up of at least five subunits named PIG-K, PIG-T, PIG-U, PIG-S and GPAA1 in humans [12–16]. PIG-K is the catalytic subunit in this complex and proposed to be responsible for the C-terminal proteolytic processing of the substrate proteins [17–19]. GPAA1 is predicted to consist of an N-terminal TM (transmembrane) region, a globular luminal domain and six C-terminal TM segments in humans, yeast and trypanosomes [1,20]. GPAA1 is discussed to play a role in structural function and/or in supplying the mature GPI lipid anchor to the protein substrate [1].

The molecular mechanisms of the GPI transamidase complex are the most non-understood aspect of the GPI lipid anchor biology. Therefore, biochemical and structural studies are needed to delineate the biological functions of the transamidase subunits. Structural studies of the transamidase subunits have proved difficult because of their solubility issue. Recently, the low-resolution solution structures of yPIG-K (yPIG-K^24–337) and yPIG-S (yPIG-S^38–467) from Saccharomyces cerevisiae have been solved [21]. yPIG-K^24–337 consists of an egg-like domain and a small globular segment that are linked by a short stalk, while yPIG-S^38–467 appears like an open-hand domain attached to a wrist made up by the smaller domain. Since GAA1/GPAA1 is described to interact with GPI transamidase subunits and in particular with Gpi8p/PIG-K [13,22], we wanted to extend the structural insight into the subunit ensemble of the GPI transamidase. Based on sequence analytic considerations, we generated the two constructs yGPAA1^70–247 and yGPAA1^70–339 of the luminal domain of S. cerevisiae GPAA1. In comparison with yGPAA1^70–339 the recombinant protein yGPAA1^70–247 showed a higher degree of stability as well as monodispersity and enabled us to observe the first low-resolution solution structure of this protein by SAXS (small-angle X-ray scattering).

Restriction enzymes were purchased from Fermentas. Chemicals for gel electrophoresis were received from Serva. All other chemicals were at least of analytical grade and received from BIOMOL, Merck, Roth, Sigma or Serva.

To amplify the luminal domain of yGPAA1 genes encoding residues 70–339 and 70–247 (UniProt P39012, EMBL U53880.1), three oligonucleotide primers 5′-CTTTGGATCCTCTGAATGGAACATTTTGAGGGGCTATC-3′ (forward primer), 5′-AATACTCGAGCTACGATTGGTGAAATTTTTCCAAAAGG-3′ (yGPAA1^70–339 reverse primer) and 5′-CTTTCTCGAGCTAATGTTCCGTAATGGATATAGCGATGTTG-3′ (yGPAA1^70–247 reverse primer), incorporating BamHI and XhoI restriction sites (underlined), were designed. Following digestion with BamHI and XhoI, the PCR products were ligated into the pGEX-6P-1 vector. The vectors containing yGPAA1^70–339 and yGPAA1^70–247 genes were then transformed into Escherichia coli cells [strain BL21(DE3)]. The cells were grown on 100 μg/ml ampicillin-containing LB (Luria–Bertani) liquid cultures until an attenuance (D₆₀₀) of 0.6 was reached. To induce the production of GST (glutathione transferase)-tagged yGPAA1^70–339 and yGPAA1^70–247, the cultures were supplemented with IPTG (isopropyl β-D-thiogalactopyranoside) to a final concentration of 1 mM, and incubated for 16 h at 20°C.

Cells were lysed on ice by sonication for 3×1 min in buffer A [25 mM sodium phosphate, pH 7.0, 150 mM NaCl, 2 mM DTT (dithiothreitol) and 2 mM Pefabloc^SC (BIOMOL)]. The lysate was cleared by centrifugation at 12500 g for 25 min and then the supernatant was filtered (0.45 μm; Millipore). For affinity purification of GST–yGPAA1^70–247, the filtered supernatant was incubated with the glutathione–Sepharose 4B (GE Healthcare) for 1 h, equilibrated with buffer B (25 mM sodium phosphate, pH 7.0, 150 mM NaCl and 2 mM DTT). Subsequently, the medium was washed with 5 BV (bed volume) of buffer B twice and followed by 10 BV of buffer C (50 mM Tris/HCl, pH 7.5, 150 mM NaCl, 2 mM DTT and 1 mM EDTA). The on-column cleavage of GST tag from GST–yGPAA1^70–247 was performed by incubating the medium with bound proteins together with PreScission Protease (GE Healthcare) for 16 h at 4°C. Following incubation, the cleaved yGPAA1^70–247 protein was collected and applied to gel filtration column (Superdex 75 HR 10/30 column, GE Healthcare) in buffer D (50 mM Tris/HCl, pH 7.5, and 150 mM NaCl). Fractions containing yGPAA1^70–247 were identified by SDS/PAGE [23], pooled and concentrated using Amicon Ultra-4 Centrifugal Unit [10 kDa MM (molecular mass) cut-off] (Millipore).

Steady-state CD spectra were measured in far-UV light (188–260 nm) using a CHIRASCAN spectropolarimeter (Applied Photophysics). Spectra were collected in a 60 μl quartz cell (Hellma) with a path length of 0.1 mm at 20°C and a step resolution of 1 nm. The readings were for an average of 2 s at each wavelength and the recorded ellipticity values were the average of three determinations for each sample. CD spectroscopy of yeast GPAA1^70–247 (1.1 mg/ml) was performed in buffer of 50 mM Tris/HCl, pH 7.5 and 150 mM NaCl. The spectrum for the buffer was subtracted from the spectrum of the protein. CD values were converted into mean residue molar ellipticity (Θ) in units of degree · cm²/dmol using the software Chirascan Version 1.2, Applied Photophysics. This baseline-corrected spectrum was used as input for computer methods to obtain prediction of secondary structure. The CD spectrum was analysed using the K2D3 web server [24].

The synchrotron radiation X-ray scattering data for yGPAA1^70–247 were collected following the standard procedures on the X33 SAXS camera [25,26] of EMBL Hamburg located on a bending magnet (sector D) on the storage ring DORIS III of the DESY (Deutsches Elektronen Synchrotron). A photon counting Pilatus 1 M pixel detector (67 mm×420 mm) was used at a sample–detector distance of 2.4 m covering the range of momentum transfer 0.1<s<4.5/nm [s=4p sin(q)/l, where q is the scattering angle and l=0.15 nm is the X-ray wavelength]. The s-axis was calibrated by the scattering pattern of silver-behenate salt (d-spacing 5.84 nm). The scattering from the buffer alone was measured before and after each sample measurement and the average of the scattering before and after each sample was used for background subtraction. The scattering pattern from yGPAA1^70–247 was measured at protein concentrations of 2.3 and 5.8 mg/ml respectively in a buffer composed of 50 mM Tris/HCl, pH 7.5 and 150 mM NaCl. The protein as well as the buffer samples were injected automatically using the sample-changing robot for solution scattering experiments at the SAXS station X33 [27]. All data processing steps were performed automatically using the program package PRIMUS [28]. The forward scattering I(0) and the R_g (radius of gyration) were evaluated using the Guinier approximation [29] assuming that for spherical particles at very small angles (s<1.3/R_g) the intensity is represented by I(s)=I(0) exp[−(sR_g)²/3]. These parameters were also computed from the entire scattering patterns using the indirect transform package GNOM [30], which also provide the distance distribution function ρ(r) of the particle as defined by:

The MM of both proteins was calculated by comparison with the forward scattering from the reference solution of BSA. From this procedure a relative calibration factor for the MM can be calculated using the known MM of BSA (66.4 kDa) and the concentration of the reference solution by applying

where I(0)_p, I(0)_st are the scattering intensities at zero angle of the studied and the BSA standard protein respectively, MM_p, MM_st are the corresponding MMs and c_p, c_st are the concentrations. Errors have been calculated from the upper and the lower I(0) error limit estimated by the Guinier approximation. Low-resolution models of yeast GPAA1^70–247 were built by the program DAMMIN [31] as described in [32].

The sequence architecture of yeast GPAA1 (UniProt P39012 [33]) as predicted by sequence-analytic tools [1,34,35] consists of an N-terminal TM helical segment (20–42), a large luminal supposedly globular domain (42–356), followed by six TM helical regions (357–377, 389–408, 413–435, 455–480, 537–556 and 579–600) (Figures 1B and 1C). The yeast GPAA1 has a sequence identity to the human, Mus musculus and Plasmodium falciparum proteins of 26%, 26% and 19.4% respectively. Within the predicted luminal domain located between the first TM segment and the second TM region, two conserved motifs are found at residues 126–133 (RXPRX₃TE) and 228–239 (NGX₂PNXDX₂N) (Figure 1C) which had been suggested to have a role in supplying the GPI lipid anchor to the transamidase complex [1]. These two motifs are proposed to be located at the membrane–lumen interface of the ER. Furthermore, the luminal domain of GPAA1 is essential for the assembly of GPI transamidase complex [22]. Here, two constructs were designed, in which one construct (yGPAA1^70–339) covers the predicted luminal part of yGPAA1 and one shorter luminal construct, yGPAA1^70–247, that still include the conserved motifs (Figure 1C).

The S. cerevisiae GPAA1 genes, encoding the residues 70–339 and 70–247, have been amplified and cloned, and the GST-tagged proteins yGPAA1^70–339 and yGPAA1^70–247 have been produced in significant amounts in BL21(DE3) cells (Figure 2A). However, degradation of GST–yGPAA1^70–339 was observed in the corresponding SDS/PAGE (Figure 2A, lane 2). The same phenomenon was also described previously, where the recombinant full-length protein of the luminal loop of human GPAA1 underwent proteolysis in cells [22]. Therefore, only the recombinant GST–yGPAA1^70–247 was selected for further experiments.

GST–yGPAA1^70–247 was affinity-bound to the glutathione–Sepharose 4B medium from the crude lysate, before subjected to the on-column cleavage of GST tag by PreScission protease (Figure 2B). The eluted fractions were further purified via size-exclusion chromatography (Superdex 75 HR 10/30 column) as demonstrated by SDS/PAGE, revealing the high purity of the protein (Figure 2C). A Superdex 75 gel filtration column was calibrated by determining the K_av values for a set of standard proteins of known MM. Comparison of the K_av for yGPAA1^70–247 with the standard proteins suggests a native MM of approximately 24 kDa.

The secondary structure of recombinant yGPAA1^70–247 was determined from CD spectra, measured between 188 and 260 nm (Figure 3). The average secondary structure content was 28% α-helix and 27% β-sheet. This result is consistent with secondary structure predictions based on yGPAA1^70–247 amino acid sequence using the web server PredictProtein [36,37] (36% α-helix, 25% β-sheet and 38% random coil). The molar ellipticity values at 208 nm and at 222 nm are in a ratio of 1.26.

The high protein purity allowed SAXS experiments to be performed, with the aim to determine the first low-resolution structures of yeast GPAA1^70–247 in solution. The final composite scattering curve of yGPAA1^70–247 is shown in Figure 4(A). Inspection of the Guinier plot of the protein at low angles showed good data quality and no protein aggregation (Figure 4B). The R_g of yGPAA1^70–247 is 2.72±0.025 nm and the D_max (maximum dimension) of the protein is 9.14 nm (Figure 4C). Comparison of forward scattering I(0) between yGPAA1^70–247 and reference protein, BSA, yields a MM of 28 kDa, indicating that yGPAA1^70–247 is monomeric at the concentration used. Qualitative analysis of the distance distribution function ρ(r) suggests that yGPAA1^70–247 consists of a major portion, yielding a principal maximum in the ρ(r) at around 3.0 nm (Figure 4C), whereas the separated protuberance domain giving rise to a shoulder from 6.5 to 9.14 nm.

The gross structure of yGPAA1^70–247 was restored ab initio from the scattering patterns in Figure 4(A), using the program DAMMIN. The obtained shape for yGPAA1^70–247 yields a good fit to the experimental data in the entire scattering range. The corresponding fit, shown in Figure 4(A), has a discrepancy of χ²=1.3. All 10 independent reconstructions yielded a reproducible shape (Figure 4D). The NSD (normalized spatial discrepancy), which is a measure of similarity between sets of three-dimensional points [29], was computed between all 10 reconstructions, with a range of NSD from 0.396 to 0.476. The reconstruction with least NSD was selected, which has NSD=0.396. The protein appears as a two-domain molecule with a large elliptical shape, followed by a hook-like domain, and connected via a narrowed 0.8 nm short stalk (average diameter 0.8 nm). The major domain has dimensions of approximately 7.1 nm×4.8 nm, whereby the smaller domain is approximately 0.8 nm×3.5 nm.

The eukaryotic transamidase complex is composed of five major subunits: PIG-K, PIG-S, PIG-T, PIG-U and GPAA1. The substrate protein interacts as folded protein with this complex [38]. The recently determined first solution structures of yPIG-K^24–337 and yPIG-S^38–467 from yeast provided the basis for a better understanding of the structural shape and arrangement of these subunits on the ER lumen side. The hand-wrist shaped yPIG-S^38–467 (Figure 5) with its concave surfaces allows this protein to interact with other subunits and to fulfil the function of stabilizing a substrate protein–transamidase complex. yPIG-K^24–377 has been described as an elongated particle consisting of an egg-like portion and a small globular segment linked together by an 1.9 nm long stalk. The egg-like portion involves the active site machinery (Figure 5) [21], which belongs to the C13 cysteine peptide family. The conserved residue Cys⁸⁵ of yPIG-K^24–377 is exposed to the solvent [21] and proposed to form a disulfide bridge with a cysteine residue in subunit PIG-T of the yeast transamidase, bringing both subunits in close proximity (Figure 5) [21].

PIG-K co-precipitates with GPAA1 [13]. Both subunits are described to be in close proximity. PIG-K is known to bind the ω-site of the GPI anchor attachment signal in the substrate proprotein sequence [1]. The removal of the C-terminal TM domains in GPAA1 yields a non-functional GPI transamidase, while the assembly of the whole complex is not affected, suggesting the luminal domain of GPAA1 is important in the interaction with PIG-K, PIG-S and PIG-T [22]. The elongated shape of yGPAA1^70–247 with its elliptical domain, which is connected via a short stalk to the smaller hook-like domain, has a total length of 9.14 nm and would provide sufficient space to interact with yPIG-K^24–377, having a D_max of 10.3 nm (Figure 5) [21]. yGPAA1^70–247 includes the two conserved stretches of polar residues, i.e. RXPRX₃TE between residues 126–133 and NGX₂PNXDX₂N in the amino acid sequence 228–239. These two motifs have been suggested to play a role in the binding of the free GPI lipid anchor and to supply it to PIG-K [1]. The arrangement of yGPAA1^70–247 and yPIG-K^24–377 in the model presented brings both subunits in close proximity, which allows the entrance of the GPI lipid anchor for the subsequent catalytic event. The hydrophobic tail of the substrate would then supposedly be stabilized by residues of the TM helices at the N- and C-termini of yGPAA1^70–247, predicted to be involved in GPI interaction [1].

In summary, the cloning and production of the recombinant yGPAA1^70–247 provided a pure and monodisperse protein with a proper secondary structural content. The protein has been demonstrated as an elongated protein with a major elliptical- and a smaller hook-like domain that connected via a short stalk. Together with the recently determined low-resolution structures of yPIG-K^24–337 and yPIG-S^38–467 [21], the presented low-resolution structure of yGPAA1^70–247 in solution shines new light into the concerted interaction of the transamidase complex subunits together with its substrate protein and provides the basis for a better understanding of the structural shape and arrangement of these subunits on the ER lumen side.

BV

bed volume

D_max

maximum dimension

DTT

dithiothreitol

ER

endoplasmic reticulum

GPI

glycosylphosphatidylinositol

GST

glutathione transferase

IPTG

isopropyl β-D-thiogalactopyranoside

MM

molecular mass

NSD

normalized spatial discrepancy

R_g

radius of gyration

SAXS

small-angle X-ray scattering

TM

transmembrane

Wuan Geok Saw designed and conducted experiments, and performed data analysis. Birgit Eisenhaber and Frank Eisenhaber contributed to the experimental design and discussions. Gerhard Grüber designed the study and experiments, and supervised the research. All authors participated in writing the paper.

W.G.S. thanks the authority of Nanyang Technological University for awarding a research scholarship. We acknowledge the EMBL-Outstation, Hamburg, Germany, for provision of synchrotron radiation facility, and we thank M. Roessle for his great help in collecting SAXS data. We are grateful to M. S. S. Manimekalai (SBS, NTU) for her support in SAXS data analysis.

This work was supported by the New Initiative Fund FY2010, NTU.