Purification and biochemical characterization of recombinant Persicaria minor β-sesquiphellandrene synthase

Materials

Pfu DNA polymerase was purchased from Biotechrabbit (Germany). Restriction enzymes and DNA ligase were purchased from Thermo Scientific (USA). DNA gel purification kits and plasmid purification kits were purchased from iNtRON Biotechnology (Korea). Farnesyl diphosphate (FPP), inorganic diphosphatase and standard alkane solution (C₈–C₂₀) were obtained from Sigma-Aldrich (St. Louis, MO, USA). Geranyl diphosphate (GPP) and geranylgeranyl diphosphate (GGPP) were purchased from Echelon Biosciences (Salt Lake City, UT, USA). Malachite Green Phosphate Assay kit was obtained from Bioassay Systems (Hayward, CA, USA). QuikChange site-directed mutagenesis kit was obtained from Agilent Technologies (Santa Clara, CA, USA). HisTrap™ HP 5 mL, and HiLoad 16/600 Superdex 200 pg were purchased from GE Healthcare (Chicago, IL, USA).

Design of recombinant P. minor sesquiterpene synthase constructs (PmSTS-Δ24)

To remove the N-terminal disordered region and enhance protein homogeneity, a truncated PmSTS construct, namely PmSTS_Δ24 was generated from the full-length recombinant PmSTS using forward primer (5′-GCCCCTCGTCATATGGCAGGTTTCAAACCTTCC-3′) and reverse primer (5′-CCAAGCTTTCATATCAGTATGGGATCGATGTAC-3′). The NdeI and HindIII restriction endonuclease sequences are underlined in these oligonucleotides, and the stop codon UGA is indicated in bold characters. The PCR amplification was performed according to the manufacturer’s guidelines. The PCR product was analyzed by agarose gel electrophoresis and purified using DNA gel purification kit following manufacturer’s guidelines. The purified PCR product was digested with NdeI and HindIII and ligated into pET28b expression vector.

Molecular cloning

The full length PmSTS (GenBank accession no: JX025008), and truncated PmSTS_Δ24 were cloned into pET28b with the affinity tag (His₆) at its N-terminus, which yielded the resulting recombinants, pET28b_PmSTS and pET28b_PmSTS_Δ24, respectively. The plasmids were transformed into competent E. coli TOP 10 cells using heat shock at 42 °C for 60 s and the transformants were selected on LB plates containing kanamycin (50 µg mL⁻¹). Positive colonies were identified by colony PCR. Recombinant plasmid was isolated from positive transformants using a plasmid purification kit. The constructs were verified by Sanger DNA sequencing (First BASE Laboratories Sdn Bhd, Malaysia).

Protein expression of recombinant P. minor sesquiterpene synthase (PmSTS)

The recombinant plasmids of pET28b_PmSTS and pET28b_PmSTS_Δ24 were transformed into competent E. coli BL21 (DE3) cells. The effect of different temperature on the solubility of recombinant protein expression was investigated by isopropyl β-D-1-thiogalactopyranoside (IPTG) induction at 37 °C and 16 °C. Briefly, a single colony was picked and cultured overnight at 37 °C in 10 mL of sterile LB culture (50 µg mL⁻¹ of kanamycin) with agitation 200 rev min⁻¹. The cells were allowed to grow at 37 °C until OD₆₀₀ reached 0.6. The cultures grown at 37 °C and 16 °C were induced by adding IPTG to a final concentration of 0.5 mM. The culture was further incubated at 37 °C and 16 °C for 4 h and 16 h at 200 rev min⁻¹ respectively. Cells were harvested by centrifugation for 10 min at 5,500 g at 4 °C and stored frozen at −80 °C until use.

Cells that had been induced at 37 °C and 16 °C were resuspended in lysis buffer containing 20 mM Tris–HCl (pH 8.0), 500 mM NaCl, 20 mM β-mercaptoethanol (βME), and lysed with 10 min sonication composed of 5 s pulse with 10 s rest at amplitude 30% power using ultrasonicator (QSONICA) on ice. Cell lysate was then centrifuged at 13,000 g for 30 min at 4 °C. The pellet and supernatant corresponding to insoluble and soluble proteins were analyzed using SDS-PAGE.

Purification of recombinant P. minor sesquiterpene synthase (PmSTS)

For protein purification, cell pellets harvested from 2 L of LB culture were used. The cell pellet was resuspended in 30 mL of binding buffer (20 mM Tris–HCl, pH 8.0, 500 mM NaCl, 20 mM βME, 20 mM imidazole). For purification in the presence of glycerol, binding buffer G (20 mM Tris–HCl, pH 8.0, 500 mM NaCl, 20 mM βME, 20 mM imidazole, 15% (v/v) glycerol) was used instead. The cell suspensions were disrupted with a 10 min sonication composed of 15 s pulse with 30 s rest at amplitude of 30% power using an ultrasonicator (QSONICA) on ice. Cell lysate was then centrifuged at 13,000 g for 30 min at 4 °C. The supernatant fraction was filtered through a syringe filter (0.2 µm pore size) before being applied into a HisTrap™ HP 5 mL (GE Healthcare), pre-equilibrated with binding buffer. After washing of HisTrap™ with 10 column volumes (CV) of binding buffer, protein was eluted with elution buffer (20 mM Tris–HCl, pH 8.0, 500 mM NaCl, 20 mM βME, 500 mM imidazole) in 2 mL fractions by 20 CV in linear gradient. Eluted protein fractions were pooled and concentrated to 2 mL using Microsap Advanced Centrifugal Device (10 kDa MWCO; Pall, New York, NY, USA) at 4 °C and were further purified using HiLoad 16/600 Superdex 200pg (GE Healthcare) at a flow rate of 0.8 mL min⁻¹. Eluted protein fractions were pooled and concentrated using a Microsap Advanced Centrifugal Device (10 kDA MWCO; Pall, New York, NY, USA).

Enzyme assay

Enzyme assays were performed using Malachite Green Phosphate Assay Kits (BioAssay Systems) in 96-well flat bottom plates. After purification, protein concentrations were determined using the Bradford Assay (Amesco). Briefly, 0.1 µM of purified enzyme was equilibrated in reaction mixture containing 20 mM HEPES, pH 8.0, 10 mM MgCl₂, 100 mU inorganic diphosphatase for 2 min at room temperature. Reactions (240 µL) were started by the addition of FPP, GPP or GGPP, and allowed to proceed at 30 °C for 5 min. After incubation, 80 µL of reaction mixture were transferred to each well (96-well plate) and the enzyme reactions were quenched by addition of 20 µL of malachite green solution. After 20 min of incubation, reactions were read at 655 nm using an iMark plate reader (Bio Rad). Negative controls were performed without the addition of purified enzyme. Monophosphate (Pi) and diphosphate (PPi) were generated according to the instructions of the manufacturer.

Enzyme characterization

The optimal temperature was determined in a series of temperatures ranging from 25 °C to 55 °C in 20 mM HEPES buffer (pH 8.0). The optimal pH was determined at room temperature from pH 6.0 to pH 10.5 using 20 mM BIS-TRIS propane, and 20 mM glycine NaOH buffer. Kinetic parameters were determined in assays with ten different substrate concentrations (2–30 µM) at pH 8.0 and 30 °C using 0.1 µM of purified enzyme. Apparent K_m, k_cat and k_cat∕K_m values were obtained with GraphPad Prism 5 software.

Product identification using gas chromatography mass spectrometry (GC-MS)

For the product identification of PmSTS and PmSTS_Δ24, two extraction methods were used: headspace solid phase microextraction (HS-SPME) and solvent extraction.

For the HS-SPME extraction method, PmSTS or PmSTS_Δ24 (∼80 µg) was incubated with substrate (60 µM FPP) in assay buffer (500 µL) containing 20 mM HEPES (pH 8.0), 10 mM MgCl₂, and 1 mM dithiothreitol. The reaction mixture was then incubated at 30 °C for 2 h, and the reaction products were extracted by HS-SPME using 100 µm polydimethylsiloxane coated fiber (Supelco, Bellefonte, PA, USA). Headspace sampling times using SPME was 30 min at 45 °C and the products were analyzed using GC-MS. GC-MS analysis was performed as described previously (Song et al., 2012; Tan & Othman, 2012) using Perkin Elmer, Turbomass Clarus 600 equipped with Perkin Elmer Elite 5 MS (30 m length, I.D. 0.25 mm, 0.25 µm film thickness).

For the solvent extraction method, the reaction mixtures were overlaid with 200 µL of hexane to trap the reaction product. The PmSTS, PmSTS_Δ24 and PmSTS_Δ24_K266E protein were incubated overnight. After incubation, the hexane layer was extracted and subjected to GC-MS analysis. GC-MS analysis was performed as described previously (O’Maille, Chappell & Noel, 2004) on an Agilent 7890A gas chromatograph equipped with HP-5MS (30 m length, I.D. 0.25 mm, 0.25 µm film thickness) and 5975C MSD with triple-axis detector.

In both methods, products were identified based on their mass spectra and Kovats Index, calculated in relation to the retention times of a series of alkanes (C₈–C₂₀). The mass spectra were compared to those in the National Institute of Standards and Technology (NIST) Library in 2011.

Protein disordered region predictions

The following servers were used for disordered region prediction of PmSTS: DISOPRED3 (Jones & Cozzetto, 2015), DisEMBL (Linding et al., 2003), and RONN (Yang et al., 2005). In all cases, PmSTS (residue 1–562) was subjected to disorder prediction using default server parameters.

Multiple sequence alignment and homology modelling

Sequences of plant sesquiterpene synthase were obtained from the SWISS-PROT database through a text search for sesquiterpene. Protein sequences of 500–600 residues were retained and the proteins that produce sesquiterpene, as judged from available GC-MS analysis, were selected. Multiple sequence alignment (MSA) was performed using Clustal Omega webserver (Sievers et al., 2011). The alignment was then visualized and analyzed using Jalview 2 (Waterhouse et al., 2009). An homology model of PmSTS as previously reported (Ee et al., 2014) was constructed using I-Tasser (Roy, Kucukural & Zhang, 2010). Superimposition of PmSTS with other terpene synthases (PDB: 3M01, 5EAT, 4FJQ, 3G4D, 1N20, 2ONG) were performed using Pymol (The PyMOL Molecular Graphics System, Version 1.8 Schrödinger, LLC.).

Site-directed mutagenesis to generate PmSTS_Δ24^K266E mutant

To generate the K266E mutant, site-directed mutagenesis was performed using QuikChange site-directed mutagenesis kit (Agilent) according to the manufacturer’s guidelines. The PCR-based mutagenesis protocol was performed with the PmSTS_Δ24 cDNAs cloned into the expression vector pET28b using forward primer (5′-GAAATGTGCAGGTGGTGGG AAAAGGTGAATATGACTAAG-3′) and reverse primer (5′-CTTAGTCATATTCACCTTTT CCCACCACCTGCACATTTC-3′). The mutagenized construct was fully sequenced before expression. The overexpression and purification of PmSTS_Δ24^K266E mutant protein was performed the same for the recombinant PmSTS.

Protein disordered region analysis of PmSTS lead to construct design of PmSTS_Δ 24 protein

Analysis of PmSTS sequence using several disorder prediction servers suggested that the N-terminal region of PmSTS (about 1–25 amino acids residues) contains disordered regions (Fig. S1). The disordered regions are amino acid regions that lack a stable secondary structures and have high conformational dynamics and flexibility that are susceptible to aggregation (Lebendiker & Danieli, 2014). To eliminate the possibility of protein aggregation caused by the disordered region, and hence facilitate the purification of homogenous protein, the recombinant PmSTS_Δ24 was designed by truncating the N-terminal predicted disordered region.

Cloning and over expression of recombinant P. minus sesquiterpene synthase (PmSTS)

Full length recombinant sesquiterpene synthase of P. minor (PmSTS) and N-terminally truncated variant (PmSTS_Δ24) were overexpressed with pET28b vector using E. coli BL21 (DE3) strain. Both the PmSTS and the PmSTS_Δ24 recombinant proteins contain a His6-tag at its N-terminus to aid in the purification of recombinant protein using immobilized metal affinity chromatography (IMAC). To monitor the expression level and solubility properties of PmSTS proteins, E. coli harboring the PmSTS or PmSTS_Δ24 gene was expressed at two different temperatures, 16 °C and 37 °C. Recombinant cells grown at both 37 °C and 16 °C showed the production of recombinant enzyme, however soluble protein expressions of PmSTS and PmSTS_Δ24 were only observed at 16 °C (Fig. 1). Expression at 37 °C drove all the proteins into inclusion bodies (Fig. 1). Lower growth temperature is known to facilitate the production of soluble recombinant protein through slowing down the transcription and translation rates, as well as reducing the strength of hydrophobic interactions that contribute to protein misfolding (Baneyx & Mujacic, 2004). The truncated version of PmSTS (PmSTS_Δ24) did not show an enhanced protein expression solubility despite the removal of the N-terminal disordered region (Fig. 1B).

Purification of recombinant PmSTS enzyme

The recombinant protein purification was conducted using nickel affinity chromatography. The recombinant protein was eluted at 200 mM imidazole, and the eluted protein fractions were identified by SDS-PAGE. The results showed high levels of E. coli contaminants eluted together with PmSTS (Fig. 2A, lane 3). Additional protein purification using size exclusion chromatography (SEC) indicated that PmSTS may have bound together with the contaminants and formed soluble aggregates (Fig. 2A). Further protein purification buffer optimization has identified that addition of 15% (v/v) glycerol in protein purification buffer has aided in reducing the contaminants and improved the homogeneity of the PmSTS protein (Fig. 2B), although the majority of the protein still remained as soluble aggregates. The SEC profile suggested that PmSTS exists as a monomer in solution. The PmSTS_Δ24 was expressed in E. coli BL21 (DE3) and purified using identical method as for PmSTS. The results revealed significant improvement in the purification profile of homogeneous PmSTS_Δ24 compared to PmSTS (Fig. 2C). Both systematic protein purification optimization and truncated protein design suggested that addition of glycerol in the buffer and elimination of N-terminus disordered region of PmSTS are important in producing homogenous PmSTS enzyme.

Enzyme characterization of PmSTS

The purified full length PmSTS and PmSTS_Δ24 proteins were used for biochemical characterization. Some sesquiterpene synthases have been reported to have broad substrate specificity, accepting both GPP and FPP as substrates (Nieuwenhuizen, Wang & Matich, 2009; Zhuang et al., 2012; Srivastava et al., 2015). However, the enzyme activity assay showed that both PmSTS and PmSTS_Δ24 were only active towards C₁₅ substrate farnesyl diphosphate (FPP). Neither C₁₀ substrate geranyl diphosphate (GPP) nor C₂₀ substrate geranylgeranyl diphosphate (GGPP) are substrates for PmSTS and PmSTS_Δ24.

As for the determination of kinetic parameters, only the activity of PmSTS_Δ24 was assayed. The purified full length PmSTS displayed enzyme instability and losing its activity over a period of time, thus making it unsuitable for the enzyme assay. The cause of the instability of PmSTS protein is yet to be investigated.

The optimal temperature for PmSTS_Δ24 activity was found at 30 °C (Fig. 3A). At 40 °C, the enzymatic activity was less than 40% compared to that at 30 °C. The optimal pH range of the enzyme was further determined at pH 7.5 to 8.0 (Fig. 3B). The enzyme activity was found to be dramatically reduced above pH 8.5. Kinetic characterization of PmSTS_Δ24 on FPP was also performed. PmSTS_Δ24 has an apparent K_m value of 10.2 µM, a k_catvalue of 0.078 s⁻¹ and k_cat∕K_m of 7.6 × 10³ M⁻¹ s⁻¹. The optimal pH and temperature, as well as the kinetic parameters of PmSTS_Δ24 are comparable to other plant sesquiterpene synthases (Table 1). As judged by the k_cat∕K_m values, the plant sesquiterpene synthases, in general, have low catalytic activity. The low catalytic activity is common as the plant sesquiterpene synthases are plant enzymes involved in secondary metabolism, which are known to have k_cat∕K_m values around 10³ M⁻¹ s⁻¹ or lower (Bar-Even et al., 2011).

P. minor sesquiterpene synthase produce β-sesquiphellandrene

The enzyme assays of PmSTS and PmSTS_Δ24 with FPP as substrate were performed by headspace solid phase microextraction gas chromatography mass spectrometry (HS-SPME-GC-MS). The HS-SPME-GC-MS analysis of volatile sesquiterpene produced by PmSTS and PmSTS_Δ24 identified β-sesquiphellandrene as the main product (∼97%) and β-farnesene (∼3%) as a minor product (Fig. 4). The HS-SPME result was further verified by solvent extraction using hexane, which showed that the major product is indeed β-sesquiphellandrene (KI:1516).

Previous GC-MS analysis of enzymatic reaction using partially purified recombinant PmSTS from E. coli showed that PmSTS produced α-farnesene (Tan & Othman, 2012). However, further studies using partially purified recombinant PmSTS from L. lactis (Song et al., 2012) and metabolite studies of A. thaliana expressing PmSTS (Ee et al., 2014), have shown that PmSTS is a β-sesquiphellandrene synthase. In this study, the PmSTS that had been purified to homogeneity was further confirmed as a β-sesquiphellandrene synthase. In an effort to prove that the point mutation K266E, introduced during the cloning process in L. lactis (Song et al., 2012), does not interfere in product specificity of PmSTS, site-directed mutagenesis was undertaken to alter residue lysine 266 to glutamic acid. The recombinant mutant protein K266E produced using an E. coli expression system displayed a product profile resembling that of PmSTS (Fig. S5). Structural analysis revealed that residue K266E is located at the exterior surface of PmSTS (Fig. 5C), and therefore is unlikely to affect the PmSTS product specificity. The truncated PmSTS with the removal of N-terminal 24 residues (PmSTS_Δ24) was shown to synthesize β-sesquiphellandrene as a major product, identical to the full length PmSTS. Thus, the N-terminal region is also not involved directly in product specificity of PmSTS, and similar properties have also been reported for truncated γ-humulene synthase from grand fir (Little & Croteau, 2002).

β-Sesquiphellandrene is a sesquiterpene found as a constituent of ginger (Zingiber officinale) (Onyenekwe & Hashimoto, 1999), turmeric (Curcuma longa) (Tyagi et al., 2015), and Alpinia conchigera (Ibrahim et al., 2009). Previous studies have shown that β-sesquiphellandrene exhibits various biological activities such as antioxidant and anticancer activities (Zhao et al., 2010; Tyagi et al., 2015). Besides plants, β-sesquiphellandrene has been found in insect as sex pheromone (Borges et al., 2007). In P. minor, sesquiphellandrene was detected at minute amount (0.1%) in the leaves and stems (Ahmad et al., 2014).

Some plant sesquiterpene gene expressions have known to be mediated by plant developmental stages or environmental stresses (Bohlmann et al., 1998; Xu et al., 2004; Zhuang et al., 2012; Yu et al., 2015). For example, sesquiterpene synthases of rice (Os08g07100) and sorghum (SbTPS1, SbTPS2), that share 32%–35% sequence identity with PmSTS, were found to produce β-sesquiphellandrene after damage by herbivores, suggesting that the emission of β-sesquiphellandrene play a role in crop defense (Yuan et al., 2008; Zhuang et al., 2012). The low abundance of sesquiphellandrene detected in P. minor may indicate similar regulation in the expression of sesquiterpene synthase (PmSTS).

Multiple sequence alignment of plant sesquiterpene synthases and PmSTS reveals an altered second metal binding motif

BLAST analysis revealed that PmSTS has high homology with sesquiterpene synthases of angiosperms, with the highest level of similarity (45%) to drimenol synthase, a cyclic sesquiterpene synthase, from Persicaria hydropiper (GenBank Accession No: KC754968.1). PmSTS contains numerous motifs highly conserved among the plant sesquiterpene synthases, including RXR, DDXXD and NSE/DTE motifs (Ee et al., 2014). However, extensive MSA analysis in this study unexpectedly discovered that PmSTS contains a modified metal binding motif N⁴⁵⁸DXXG⁴⁶²XXXV⁴⁶⁶ on helix H. This metal binding motif usually has consensus sequence (N/D)DXX(T/S/G)XXXE (Christianson, 2006; Zhou & Peters, 2009) or DDXX(D/E) (Gennadios et al., 2009) in which boldface residues typically binds to one Mg cofactor, namely Mg²⁺_B.

In PmSTS, the glutamate residue in metal binding motif (NDXXGXXXE) is found to be replaced by valine residue, N⁴⁵⁸DXXG⁴⁶²XXXV⁴⁶⁶, indicating that PmSTS contains a modified metal binding motif (Fig. 6). This alternative form of motif has not been reported in plant sesquiterpene synthases. Previous mutational analyses on other terpene synthases have shown that changes from glutamate to glutamine or aspartate in the metal binding motif greatly reduce the catalytic activity and changes the product specificity of terpene synthases (Peters & Croteau, 2002; Felicetti & Cane, 2004). However, PmSTS was found to be fully active, despite the fact that the hydrophobic side chain of V466 is not able to form a hydrogen bond with a Mg²⁺ ion. It is likely that the side chain of N458 may still chelate the Mg²⁺_B ion, with assistance from a water molecule (Zhou & Peters, 2009) or carboxylic side chain of D459 (Fig. S7), thereby PmSTS has a catalytic efficiency (k_cat∕K_m) that is comparable to other sesquiterpene synthases (Table 1).

Multiple sequence alignment between linear and cyclic plant sesquiterpene synthase was performed to identify a potential conserved motif responsible for the cyclization in sesquiterpene synthases. However, MSA analysis did not find any motif that was able to distinguish between linear and cyclic sesquiterpene synthases. A similar result was obtained from phylogenetic analysis of various α-farnesene synthases with other terpene synthases, as the phylogenetic analysis did not cluster all α-farnesene synthase together in one group (Green et al., 2007). This is not surprising given the low sequence identity within sesquiterpene synthase family (Christianson, 2006; Aaron & Christianson, 2010), even though most of the enzymes share a similar overall structure. Furthermore, based on current knowledge of sesquiterpene biosynthesis, it is still not possible to predict the absolute sesquiterpene product of a sesquiterpene synthase based on its amino acid sequence (Zulak & Bohlmann, 2010; Dickschat, 2011).

Homology model provides structural insights into PmSTS catalytic active site

The previously generated homology model was used to gain structural insight into PmSTS catalytic mechanism (Ee et al., 2014). The overall structure of PmSTS adopts an α-helical architecture containing two domains that resemble the terpene synthase family N-terminal domain (residue 1–236 for PmSTS) and terpene synthase family C-terminal metal-binding domain (residue 237–562 for PmSTS), also known as a catalytic domain (Fig. 5A). Based on the homology model, the predicted N-terminal 24 amino acid disordered region is positioned near the entrance of the active site; however, it does not affect the enzyme product specificity. Overall, the PmSTS structure is highly similar to other plant terpene synthases (Table 2). Despite lower sequence identity, PmSTS structure shared more similarity to the monoterpene synthase structure with lower RMSD than to its sesquiterpene synthase counterpart (Table 2). Structural comparison further revealed that the J–K loop and the second metal-binding motif region (H2 helix and H2-α1 loop) are highly flexible compared to the conserved motif RXR and DDXXD located at helix D of the active site (Fig. 5B).

Superimposition of the PmSTS homology model to the substrate analog complex of N. tabacum 5-epi-aristolochene synthase (NtEAS; PDB:3M01) and Gossypium arboreum (+)-delta-cadinene synthase (GaDCS; PDB:3G4F) showed that PmSTS may adopt a substrate binding like NtEAS (Fig. 5B), but is unable to bind ligand as seen in GaDCS due to the ligand binding mode of GaDCS, which may cause steric clash with the J–K loop of PmSTS (Fig. S8).

To better elucidate the function of β-sesquiphellandrene synthase, a structural study of PmSTS in complex with an FPP analog will be important to provide insights into the active site especially at the modified second metal-binding motif that is likely to interact with the Mg²⁺_B ion.

Based on previous knowledge about the reaction mechanism of other sesquiterpene synthases (Köllner, Gershenzon & Degenhardt, 2009; McAndrew et al., 2011; Garms et al., 2012), we proposed a reaction mechanism for PmSTS such that the biosynthesis of sesquiterpene begins with metal-dependent ionization of the diphosphate moiety of FPP to form a farnesyl cation and a diphosphate group (Fig. 7). The diphosphate group will interact with and be stabilized by three Mg²⁺ ions and highly conserved positively charged residues, R277, R279, and R455. The positively charged region will direct the diphosphate away from the active site. Deprotonation of the farnesyl cation may yield the minor product β-farnesene. However, most of the farnesyl cation will undergo 1,6 cyclization via a nucleophilic attack of the C6–C7 double bond generating a bisabolyl cation. Subsequently, the 1,3-hydride shift between C1 and C7 and deprotonation at C15 will lead to the formation of β-sesquiphellandrene (Garms et al., 2012).