Recombinant human dihydroxyacetonephosphate acyl-transferase characterization as an integral monotopic membrane protein

Abstract

Although the precise functions of ether phospholipids are still poorly understood, significant alterations in their physiological levels are associated either to inherited disorders or to aggressive metastatic cancer. The essential precursor, alkyl-dihydroxyacetone phosphate (DHAP), for all ether phospholipids species is synthetized in two consecutive reactions performed by two enzymes sitting on the inner side of the peroxisomal membrane. Here, we report the characterization of the recombinant human DHAP acyl-transferase, which performs the first step in alkyl-DHAP synthesis. By exploring several expression systems and designing a number of constructs, we were able to purify the enzyme in its active form and we found that it is tightly bound to the membrane through the N-terminal residues.

Highlights

• •Human DHAPAT is associated to peroxisomal membrane through the N-terminal region.
• •Recombinant human DHAPAT expressed and purified from P. pastoris cells is active.
• •Evidence of the in vitro reconstitution of DHAPAT/ADPS enzymatic complex.

1. Introduction

Ether phospholipids are characterized by an ether bond, rather than an ester bond, on the sn-1 position of the glycerol backbone. Although the detailed physiological roles of intracellular and circulating ether linked phospholipids are not yet well understood, their peculiar physical properties impact on many aspects of cell signaling and membrane biology [1], [2]. In addition to their importance as structural components of cell membranes, they are fundamental for membrane fusion and vesicle formation, and they are involved in free radical scavenging and second messenger lipids storage [3], [4]. Their importance is further underlined by the pathological conditions associated to altered synthesis. In fact, most of the known peroxisomal disorders show deficient ether lipids synthesis [5], [6]. Additionally, elevated ether lipids synthesis and uptake are a characteristic metabolic aberration of certain type of aggressive cancers, contributing to their invasiveness and survival [7].

The key enzymes involved in the ether phospholipids biosynthetic pathway, namely dihydroxyacetone phosphate acyl-transferase (DHAPAT) and alkyl-dihydroxyacetone phosphate synthase (ADPS), catalyze the acylation of dihydroxyacetone phosphate (DHAP) to form acyl-DHAP followed by the replacement of the acyl moiety with a long fatty alcohol, yielding alkyl-DHAP [8]. Mutations in the genes encoding for DHAPAT and ADPS cause a rare inherited peroxisomal disorder, rhizomelic chondrodysplasia punctata (RCDP). RCDP patients show skeletal dysplasia and profound mental retardation, which in most of the cases lead to premature death [9].

Whereas the unusual reaction mechanism of ADPS flavin fueled the interest in the biochemical characterization of this enzyme [10], [11], [12], the difficulties associated to DHAPAT expression and purification slowed progress for decades. DHAPAT performs the first step of the ether phospholipid biosynthetic pathway, namely the acylation of DHAP to form acyl-DHAP, which is the substrate of the downstream ADPS reaction. The peroxisomal localization of DHAPAT is crucial for its enzymatic activity and stability, and it is achieved by the presence of the peroxisomal targeting signal-1 (PTS1) at the C-terminus of its amino acid sequence [13]. For long, the organization of the gene encoding for DHAPAT has remained undetermined [14], although the protein itself was isolated from human placenta [15] and guinea pig liver [16] as a 680 amino acid enzyme with a molecular mass of 77 kDa [17]. In vivo experiments suggested that DHAPAT activity was dependent on the presence of ADPS. In particular, kinetic experiments indicated that endogenously generated acyl-DHAP was used preferentially by ADPS in comparison with exogenously added substrate, suggesting a close interaction of the two enzymes within the peroxisomes [18], [19]. The formation of DHAPAT/ADPS complex was initially characterized by isolating the cross-linked enzymes from rabbit Harderian gland peroxisomes, and this remains the only evidence of the heterotrimeric complex formation so far [20].

The results presented in this study aim to develop a methodology to express and purify for the first time recombinant human DHAPAT using P. pastoris to enable the study of its interaction with ADPS.

2. Materials and methods

2.1. Full-length human DHAPAT in HEK-293E cells: cloning, expression and purification

Human DHAPAT cDNA (NM_014236.3) was supplied by Origene. N-terminal His₈-eGFP tag was fused to DHAPAT through PCR and cloned into pCDNA3 vector (Invitrogen). HEK-293E (Invitrogen) cells were grown in suspension according to Longo et al. [21]. The transfection mixture contained polyethylenimine (PEI; DNA:PEI = 1:3 w/w), OPTIMEM medium (Gibco, 1:20 v/v) and pCDNA3-His₈-eGFP-DHAPAT (1 μg of DNA/10⁶ cells). Cells were collected after 24/48 h after transfection. Frozen cell pellets were resuspended in hypotonic buffer (10 mM Tris-HCl pH 7.5, 1 mM EDTA) with protease inhibitors Complete EDTA-Free (Roche), 1 mM PMSF, and 1 mg/ml DNAse, and homogenized on ice 30 times with a tight pestle. The sample was centrifuged at 70,000g for 1 h. Cell membranes were resuspended in 30 mM HEPES pH 7.8, 500 mM NaCl, 50% glycerol, and protease inhibitors. Solubilization was performed adding 1% of detergent (Anatrace) and stirring 2 h at 4 °C followed by centrifugation at 70,000g for 1 h. Fluorescence size-exclusion chromatography (F-SEC) experiment was performed according to Goering et al., 2014 [22], using a Superdex 75 5/150 column (GE Healthcare) and a Schimadzu HPLC system equilibrated with 30 mM HEPES pH 7.8, 500 mM NaCl, 5% glycerol and 0.03% DDM.

2.2. Bioinformatic analysis of DHAPAT amino acid sequence

The secondary structure prediction resulted from the comparison of the following servers: HHpred, Quick2D [23], PredictProtein [24], Proteus 2.0 [25]. The transmembrane (TM) helices/region prediction was performed by interrogating the following servers: POLYVIEW-2D (117–138) [26], TMAP (115–132, 168–196, 370–398, 565–580) [27], TMPRED (8–25, 165–194) [28], Octopus (510–560) [29].

2.3. Human DHAPAT deletion mutants: cloning, expression and purification in E. coli cells

DHAPAT shorter constructs were obtained by PCR, either lacking of N-terminal portions (DHAPATΔ120, DHAPATΔ135, DHAPATΔ144, DHAPATΔ150, DHAPATΔ156, DHAPATΔ285, DHAPATΔ530) or single domains (DHAPAT144-530, DHAPAT156-530, DHAPAT1-163), and cloned in pET28 vectors (Novagen) containing PreScission protease cleavable N-terminal tags (His₈-SUMO, GST and His₈-eGFP). Expression trials were performed in BL21, Origami, Rosetta, DH5α, Rosetta-gami and C41 E. coli strains (Novagen). GST-DHAPATΔ135 expressed using Origami was chosen for further expression and purification experiments. The culture was grown at 37 °C and 200 rpm in TB medium (Terrific Broth) until OD₆₀₀ of 1 and induced by adding 500 μM isopropyl-β-d-1-thiogalactopyranoside, shifting the temperature to 17 °C for 20 h. Collected cells were resuspended in 50 mM Tris-HCl pH 8.0, 500 mM NaCl, 5% glycerol, protease inhibitors Complete EDTA-Free (Roche), 1 mM PMSF, and 1 mg/ml DNAse. Resuspended cells were disrupted by sonication and centrifuged at 70,000g for 40 min at 4 °C. The soluble fraction was loaded on a 5 ml GSTrap FF pre-packed column (GE Healthcare), equilibrated with lysis buffer, using the ÄKTA FPLC system (GE Healthcare), and eluted with 50 mM of reduced glutathione. The sample was incubated at 4 °C with GST-tagged PreScission protease, 1:100 v/v overnight. Tag and protease were removed in a second run with the GST-trap column. The fractions containing the protein were pooled and loaded on an ion-exchange column (SOURCE 15Q, GE Healthcare), followed by a size-exclusion chromatography step (Superdex 200 10/300, GE Healthcare).

2.4. Full-length human DHAPAT: cloning, expression and purification in P. pastoris cells

Human DHAPAT cDNA fused with a N-terminal His₈-GFPuv (PreScission protease cleavable) and codon-optimized for P. pastoris expression was purchased from DNA2.0, and sub-cloned in the pJexpress 902 (DNA2.0) vector. The resulting plasmid was linearized using SacI enzyme (Thermo Scientific). The transformation was performed according to Lin-Cereghino et al. [30], using P. pastoris KM71H MutS strain (Invitrogen). The colonies were tested for DHAPAT expression by inoculating them in 24 deep-well plates containing 2 ml of BMGY medium (100 mM HK₂PO₄ pH 6.0, 6.34% Yeast Nitrogen Base, 1% glycerol, 4 × 10⁻⁵% biotin), incubating at 30 °C, 280 rpm for 60 h. The medium was exchanged with BMM medium (100 mM HK₂PO₄ pH 6.0, 6.34% Yeast Nitrogen Base, 0.5% methanol, 4 × 10⁻⁵% biotin) to induce protein expression, and after 48 h GFPuv fluorescence was measured with a Clariostar plate reader (BMG Labtech; excitation 395 nm, emission 509 nm). Positive clones were inoculated in 15 ml of BMGY medium. The pre-inoculum was incubated for 30 min at 30 °C (300 rpm). 2.5 ml of the pre-inoculum was poured in 2 L baffled-flask containing 200 ml of BMGY medium, and incubated for 72 h at 30 °C (280 rpm). Cells were harvested at 1500 g for 10 min and resuspended in 100 ml of BMM medium. To allow the induction to continue for 72 h, 0.5% methanol was added every 24 h. The best expressing clone was grown in a 5 L Bioflo 3000 Fermentor (Brunswick) at 30 °C and 72 h of methanol induction, according to Damasceno et al., 2004 [31]. Cells were collected by centrifugation at 1,500g for 10 min at 4 °C, resuspended in 50 mM Bicine pH 9.0, 500 mM NaCl, 10% glycerol, protease inhibitors Complete EDTA-Free (Roche), 1 mM PMSF, and 1 mg/ml DNAse, and lysed using a bead beater with zirconia beads (Biospec). Cell membranes were collected by centrifugation at 70,000g for 2 h at 4 °C. F-SEC experiment was performed as described above [22]. Solubilization was performed by adding 1% of FSC-12 (Anatrace) to the resuspended membranes (50 mM Bicine pH 8.2, 150 mM NaCl, 5% glycerol and protease inhibitors), stirring overnight at 4 °C. Solubilized membranes were centrifuged at 70,000g for 1.5 h. The supernatant was incubated with 5 ml of slurry Ni-Sepharose resin (GE Healthcare), stirring at 4 °C for 2 h. The buffers used for the washing steps had decreasing amounts (till 0,05%) of FSC-12. The elution step was performed using 300 mM of imidazole in the binding buffer. Tag-cleavage was performed incubating the sample with PreScission protease (GE Healthcare) in 1:10,000 v/v ratio, overnight at 4 °C. The second affinity chromatography step and the other purification steps were carried out using the ÄKTA FPLC (GE Healthcare), loading the sample on a pre-packed nickel column (His-trap, GE-Healthcare). The collected fractions were concentrated with the Amicon Ultra concentrator (30 kDa cutoff). Size-exclusion chromatography was performed using a Superdex 200 10/300 column (GE Healthcare), equilibrated with buffer containing 50 mM bicine pH 8.2, 50 mM NaCl, 5% glycerol and 0.05% FSC-12.

2.5. Radioactivity assays

DHAPAT (50 μM) was incubated for 30 min at 25 °C with 100 μM acyl-CoA and 100 μM DHAP (Sigma-Aldrich) in assay buffer (50 mM Tris-HCl pH 8.2, 50 mM NaF, 0.05% v/v FSC-12). Purified recombinant ADPS (1 μM), prepared according to Nenci et al., 2012 [11], and the radioactive substrate (100 μM), [1–14C]hexadecyl-DHAP (Sigma-Aldrich; specific radioactivity adjusted to 13,000 dpm/nmol), were added to the protein solution after the incubation. Once ADPS was added, small aliquots (10 μl) were taken at 1.5 min intervals and deposited on a DEAE cellulose membrane (Whatman), which interact with the phosphate groups of DHAP. After 4 consecutive washings with ethanol 100% to remove the excess of labeled hexadecanol, the membranes were analyzed at the scintillation counter (Tri-Carb-2100TR, Packard) to quantify the radioactivity emitted by the immobilized ADPS product. The same protocol was used to perform the activity assay of human DHAPATΔ135. All the assays were performed in duplicate. The positive control was performed as described above, using ADPS (1 μM) and adding 100 μM palmitoyl-DHAP [11].

2.6. Pull-down experiment and native gel electrophoresis

His₈-uvGFP-DHAPAT was purified as described above, without digesting with PreScission protease, (i.e. retaining the His₈-tag). Purified ADPS was prepared according to Nenci et al., 2012 [11]. The two proteins were mixed with nickel-resin, equilibrated with buffer containing 50 mM bicine pH 8.2, 50 mM NaCl, 5% glycerol and 0.05% FSC-12. The beads were washed with the same buffer and the elution was performed with 500 mM imidazole. The samples for native-PAGE (PhastGEL Native system, GE Healthcare) were prepared in a total volume of 6 μl (final concentration 1 mg/ml) by adding purified proteins with different ratios in the buffer containing 50 mM bicine pH 8.2, 50 mM NaCl, 5% glycerol and 0.05% FSC-12.

3. Results and discussion

3.1. DHAPAT is associated to peroxisomal membranes

Recombinant human DHAPAT fused to the green fluorescent protein (GFP) was expressed in HEK293-E cells retaining the peroxisomal translocation signal, as no expression was observed without PTS1 (data not shown). Partial purification of the enzyme revealed that DHAPAT is tightly associated to the membrane fraction. It was already observed that DHAPAT localizes in the inner side of the peroxisomal membrane, although there were no evidences of transmembrane regions [20]. Indeed, detergents were necessary to isolate DHAPAT, as neither glycerol nor high salt concentration were enough to solubilize it (Fig. 1A). Using a small panel of commonly used detergents, it was possible to recover up to 80% of the total GFP fluorescence, whereas with buffer without detergents the fluorescence was still associated to the membrane fraction (Fig. 1B). The SDS-PAGE analysis showed that detergents solubilized a fluorescent band in correspondence of the expected molecular weight of DHAPAT with the GFP tag (Fig. 1C). To determine the aggregation state of solubilized DHAPAT and to identify a suitable detergent for the purification, we performed a fluorescence size-exclusion chromatography (F-SEC) experiment. The results of F-SEC revealed that DHAPAT behaves very differently according to the detergent, and that most of the solubilized protein is homogenous and not aggregated (Fig. 1D). However, the scale-up of DHAPAT expression and purification from HEK293-E cells resulted difficult, as the protein was unstable and challenging to purify due to both low solubilization yields and poor interactions with affinity resins. We assumed that this is caused by mis-translocation in peroxisomes consequent to DHAPAT over-expression. Indeed, we observed that HEK-293E cells showed the formation of large fluorescence lysosomes at post-transfection time above the 48 h, likely due to mis-folded DHAPAT accumulation (data not shown).

Fig. 1.

Recombinant human DHAPAT expression in HEK-293E cells. (A) Fluorescence of samples treated with buffers containing different combinations of chaotropic agents (A - 50 mM NaCl and 30% glycerol, B - 50 mM NaCl and 20% glycerol, C - 50 mM NaCl and 5% glycerol, D – 1 M NaCl and 30% glycerol, E – 1 M NaCl and 5% glycerol, F – 500 mM NaCl and 30% glycerol, G – 500 mM NaCl and 5% glycerol) to increase eGFP-DHAPAT solubility after cell lysis. (B) eGFP-DHAPAT fluorescence recovered after solubilization with or without detergents. (C) eGFP fluorescence detection in SDS-PAGE. The fluorescent band corresponds to the expected molecular weight of DHAPAT fused to eGFP (109 kDa, considering that membrane proteins migrate with a discrepancy of up to the 30% of the expected molecular weight, due to the interaction with detergent micelles [35]). (D) F-SEC profile superposition representing the best detergents for human DHAPAT solubilization from HEK-293E cells.

3.2. The region associated with membranes is located at the N-terminus of DHAPAT

Bioinformatics analysis of DHAPAT sequence did not provide reliable tertiary structure prediction, as there are no crystal structures available of similar enzymes. Indeed, it was not even possible to predict the domain organization of the enzyme because the only conserved region is the putative acyl-transferase domain, spanning approximately from residue 138 to 283 [32], [33]. By interrogating several types of software (see Materials and Methods section) to predict the secondary structures of DHAPAT, we identified a number of highly hydrophobic α-helices, which could either be part of the active site, expected to be partially hydrophobic to host the substrate acyl chain, or interact with the peroxisomal membrane (in green in Fig. 2A). Based on these predictions, we designed a number of shorter DHAPAT constructs with the aim of obtaining a soluble portion of the protein both to pinpoint membrane-interaction sites and favor crystallographic studies.

Fig. 2.

Characterization of recombinant human DHAPATΔ135. (A) Prediction of the secondary structure organization of DHAPAT. Based on the hydropathy plot and the hydrophobic nature of the amino acids included in predicted α-helices (115–132, 168–178, 409–425, 464–481, 565–580, 600–624), we identified putative portions interacting with the peroxisomal membrane. (B) Size-exclusion chromatography elution profile of purified DHAPATΔ135 shows a major peak at 8.6 ml, corresponding to the void volume of the column and a minor peak at 14 ml, which corresponds to approximately 60 kDa size, the expected size of DHAPATΔ135 (54 kDa). (C) The purity of DHAPAT sample after SEC is confirmed by SDS-PAGE, showing a single band corresponding to approximately 60 kDa. (D) Radioactivity assay based on ADPS coupled-reaction shows that DHAPATΔ135 is not active, as it did not produce palmitoyl-DHAP, the ADPS substrate. ADPS alone with DHAPAT substrates was used as negative control and ADPS with palmitoyl-DHAP (DHAPAT product) as positive control.

We used E. coli as expression system to screen short protein variants fused to purification tags known to help the folding and protein solubility (i.e. SUMO and GST). DHAPATΔ135, DHAPATΔ144, DHAPATΔ150, lacking of N-terminal segments of different lengths, were expressed and, in particular, we succeeded in obtaining a soluble and purifiable form of DHAPAT lacking the first 135 amino acids, fused to GST tag. Although the expression yield was good and it was possible to purify DHAPATΔ135 without the use of detergents, the protein was unstable and highly prone to aggregation (Fig. 2B–C). Indeed, the non-functional state of the protein was confirmed both by the SEC profile, which showed that most of the protein was aggregated, and by the activity assay, that revealed that DHAPATΔ135 is completely inactive (Fig. 2D). Nevertheless these experiments were insightful as they indicate that the N-terminal segment is a main element for membrane association/integration and that this portion is likely involved in the enzymatic activity and in the correct protein folding.

3.3. Recombinant human full-length DHAPAT purified from P. pastoris retains the enzymatic activity

As membrane association resulted to be essential for correct DHAPAT function, we focused on the characterization of the full-length enzyme. It was already observed that the localization of DHAPAT in peroxisomes is crucial for its correct function and activity. Therefore, we chose P. pastoris as a eukaryotic expression system, given its characteristic peroxisomes proliferation under methanol induction of heterologous proteins expression [34]. As in the case of HEK-293E cells, we retained the PTS1 signal and we used the C-terminal sequence SKL, instead of the human AKL, because it is recognized more efficiently by yeast. The expression yield in P. pastoris of human DHAPAT was much higher than in mammalian cells and the protein was still tightly associated to the organelle membranes. We used F-SEC to evaluate the quality and the yield of protein solubilized using a large number of detergents. Only foscholine-12 (FSC-12) appeared to extract good amounts of protein, whereof only approximately 50% was in a non-aggregated form (Fig. 3A). DHAPAT was then purified through two steps of affinity chromatography to remove the tag after the cleavage, resulting in a pure sample which was further confirmed by the SEC profile, showing a single monodisperse peak (Fig. 3B–C). The final yield of purified DHAPAT was approximately 1 mg from 0.15 l of cell culture (corresponding to 50 g of wet cells) grown at high cell density with the fermenter. To assess whether the purified recombinant human DHAPAT was retaining its enzymatic activity, we used its partner, ADPS. This assay established is inherently very specific, because it is based on the accumulation of palmitoyl-DHAP produced by DHAPAT, which is used by ADPS to form alkyl-DHAP. First, DHAPAT was incubated with its substrates, palmitoyl-CoA and DHAP. Then, ADPS and its radioactively labeled substrate, [1–14C]hexadecanol, were added and production of [1–14C]hexadecyl-DHAP was monitored. As control, we performed the same experiment without DHAPAT (Fig. 3D). These experiments clearly indicated that there was acyltransferase activity and confirmed that, for the first time, we could produce a recombinant pure DHAPAT able to convert acyl-CoA to acyl-DHAP.

Fig. 3.

Full-length recombinant human DHAPAT expression and purification from P. pastoris. (A) Superposition of all elution profiles of solubilized DHAPAT shows that DHAPAT from P. pastoris is mainly aggregated with most detergents. (B) Size-exclusion chromatography elution profile of purified DHAPAT in complex with FSC-12 micelles (19 kDa) shows a single monodisperse peak at 10.4 ml (void volume 8.9 ml). (C) The purity of DHAPAT sample after SEC is confirmed by SDS-PAGE, showing a single band corresponding to approximately 70 kDa. (D) Radioactivity assay based on the production of radioactive alkyl-DHAP by ADPS, using acyl-DHAP produced by DHAPAT. The negative control was performed incubating acyl-CoA and DHAP without DHAPAT. Incubation of ADPS with palmitoyl-DHAP (DHAPAT product) was the positive control.

3.4. Reconstitution of the putative DHAPAT-ADPS enzymatic complex

In order to improve the expression and the purification, we initially tried to co-express DHAPAT with ADPS in HEK-293E cells, as it is known that DHAPAT activity and stability is dependent on ADPS presence [19]. While we succeeded in the co-expression, we could not co-purify the two enzymes, possibly because ADPS is negatively affected by the presence of high concentration of detergents needed for DHAPAT solubilization (see Fig. 4A). Therefore, we exploited the results obtained using P. pastoris expression system, which allowed us to obtain pure and active recombinant human DHAPAT. However, we could not obtain clear results from classic pull-down and size-exclusion chromatography experiments, due to the presence of FSC-12, which was found to cause aggregation and degradation of ADPS making chromatographic experiments unusable (Fig. 4A). Indeed, we observed that ADPS was not stable in buffer containing FSC-12, as after short time it degrades in two distinct fragments.

Fig. 4.

Recombinant ADPS and DHAPAT interaction in vitro. (A) SDS-page of pull-down experiments between ADPS and His₈-GFPuv-DHAPAT. 1, sample before incubating with resin; 2–4, unbound fractions containing manly ADPS; 5, elution with 500 mM imidazole, containing both ADPS and DHAPAT; 6–8, ADPS in buffer containing FSC-12 after 10, 20, 30 min, respectively. After 30 min incubation in FSC-12, ADPS gives rise to two new bands of 40 kDa and 20 kDa. (B) Interaction between APDS and DHAPAT analyzed by native-PAGE. 1, ADPS + DHAPAT (1:1); 2, ADPS + DHAPAT (2:1); 3, DHAPAT; 4, ADPS; 5, DHAPAT + control protein; 6, control protein.

On the other hand, we obtained interesting results using native gel electrophoresis. To demonstrate that the interaction between ADPS and DHAPAT was not due to the experimental conditions, a control using the soluble protein 7β-hydroxysteroid dehydrogenase was performed (Fig. 4B). The native gel clearly showed that, when ADPS and DHAPAT were mixed together, a new band was detected, which was different from the bands appearing from the migration of the single proteins. Moreover, no new band was detected when DHAPAT was mixed with the control dehydrogenase protein. The results obtained from the native gel suggested that, for the first time, there was a promising evidence for the formation of the complex between ADPS and DHAPAT. However, we could not confirm the evidence of complex formation with any other technique, as also cross-linking was impaired by the presence of the detergent.

In summary, we provide the first detailed experimental protocol to produce the recombinant human DHAPAT from P. pastoris, which could be used for further biochemical and functional studies. Indeed, the purified DHAPAT shows enzymatic activity that has been confirmed by reproducing in vitro the two-steps reaction performed by DHAPAT and ADPS crucial for ether phospholipids synthesis.

Author contributions statement

AM and VP conceived, supervised and designed experiments; VP, SN and AA performed experiments; FM provided new tools and reagents; AA, AM, and VP analyzed data; VP wrote the manuscript; AM and FM made manuscript revisions.

Fundings

Work in the authors' laboratories was supported by grants from the Fondazione Telethon [GGP12007] and Associazione Italiana Ricerca sul Cancro [FI-18022].