Structure of NADP⁺-bound 7β-hydroxysteroid dehydrogenase reveals two cofactor-binding modes

The structure of a binary complex of NADP⁺ with 7β-hydroxysteroid dehydrogenase reveals two modes of cofactor binding.

Abstract

In mammals, bile acids/salts and their glycine and taurine conjugates are effectively recycled through enterohepatic circulation. 7β-Hydroxysteroid dehydrogenases (7β-HSDHs; EC 1.1.1.201), including that from the intestinal microbe Collinsella aerofaciens, catalyse the NADPH-dependent reversible oxidation of secondary bile-acid products to avoid potential toxicity. Here, the first structure of NADP⁺ bound to dimeric 7β-HSDH is presented. In one active site, NADP⁺ adopts a conventional binding mode similar to that displayed in related enzyme structures. However, in the other active site a unique binding mode is observed in which the orientation of the nicotinamide is different. Since 7β-HSDH has become an attractive target owing to the wide and important pharmaceutical use of its product ursodeoxycholic acid, this work provides a more detailed template to support rational protein engineering to improve the enzymatic activities of this useful biocatalyst, further improving the yield of ursodeoxycholic acid and its other applications.

1. Introduction  

In mammals, bile acids/salts and their glycine and taurine conjugates are recycled through enterohepatic circulation in the body (Ridlon et al., 2006 ▸). During this process, these bile acids/salts and conjugates undergo a variety of reactions induced by the intestinal microflora. These reactions include the hydrolysis of sulfate esters and the oxidation and dehydroxylation of ring hydroxyl groups (Magee et al., 2000 ▸). The 7-dehydroxylation of bile acids is a key step in these reactions, reducing the potential toxicity of the secondary bile-acid products (Coleman et al., 1994 ▸). High concentrations of secondary bile-acid products in the blood and bile have been associated with the pathogenesis of cholesterol gallstone disease and colon cancer (McGarr et al., 2005 ▸). In this respect, 7β-hydroxysteroid dehydrogenase (7β-HSDH; EC 1.1.1.201) from the intestinal microbe Collinsella aerofaciens (Liu et al., 2011 ▸), which catalyses the NAD(P)H-dependent reversible oxidation of the hydroxyl group in the β-configuration at the C-7 position of bile acids to avoid this potential toxicity, has attracted great attention as a biocatalyst for the transformations of steroid substrates (Savino et al., 2016 ▸).

Many years ago, Carrea and coworkers used 7β-HSDH to synthesize a cholic acid derivate (Carrea et al., 1992 ▸). Since then, with the wide use of ursodeoxycholic acid, an FDA-approved cholic acid derivate drug used to treat cholesterol gallstones and as an alternative to surgical intervention, the synthetic exploitation of 7β-HSDH has attracted great attention, and many researchers from academia and industry have devoted many resources to improving the enzymatic efficiency of 7β-HSDH in producing ursodeoxycholic acid (Crosignani et al., 1996 ▸; Monti et al., 2009 ▸; Eggert et al., 2014 ▸). Recently, the first structure of 7β-HSDH in an apo form was reported (Savino et al., 2016 ▸). It provided much important structural information on the conserved NADP(H)-binding site, and a theory was proposed that some specific elements in the substrate-binding pocket were related to the distinct stereoselectivity of the enzyme. In order to expand our understanding to structural knowledge of 7β-HSDH with the cofactor NAD(P)H bound, we solved the 1.70 Å resolution structure of dimeric 7β-HSDH from C. aerofaciens complexed with NADP⁺. Surprisingly, we observed that one NADP⁺ cofactor adopts a mode of binding similar to that observed in other members of this enzyme family. However, a second cofactor adopts a mode of binding that differs in the orientation of the nicotinamide, and we hypothesize that this is an intermediate mode of binding. This new structure provides a more detailed template that may support the rational protein engineering of 7β-HSDH to improve the enzymatic activities of this useful biocatalyst, further improving the yield of ursodeoxycholic acid and its other applications.

2. Materials and methods  

2.1. Macromolecule production and isothermal titration calorimetry (ITC)  

We generated the recombinant plasmid pET-28a-7β-HSDH, encoding 7β-HSDH from C. aerofaciens ATCC25986 with 6×His and glutathione S-transferase (GST) tags. In order to facilitate the removal of the 6×His tag and GST tag, a Tobacco etch virus nuclear-inclusion-a endopeptidase (TEV protease) enzymatic site and a PreScission enzymatic site were also introduced followed by the tags. The fusion protein 6×His-TEV-GST-PreScission-7β-HSDH was expressed in Escherichia coli BL21 (DE3) cells in LB medium. To obtain purified 7β-HSDH, E. coli BL21 (DE3) cells containing the recombinant plasmid that had been induced with 0.5 mM IPTG at 288 K for about 16 h were collected by centrifugation. The pellet was resuspended in buffer consisting of 20 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM DTT and disrupted in a homogenizer (Microfluidics, USA). The supernatant was obtained by centrifuging the cell lysate at 40 000g and 277 K for 1 h. The 6×His-GST tag was removed with PreScission protease (in a 1:40 ratio) by incubation at 277 K overnight. The protein was recovered from the flowthrough of a HisTrap HP column (GE Healthcare, USA). Ion-exchange column chromatography (Mono Q and Mono S; GE Healthcare, USA) was used for further purification. The high-purity fraction was then harvested using size-exclusion chromatography (Superdex 75; GE Healthcare, USA) and concentrated for crystallization. Macromolecule-production information is summarized in Table 1 ▸. The protein concentration was measured by the NanoDrop method (Desjardins et al., 2009 ▸) using an extinction coefficient of 33 720 M ⁻¹ cm⁻¹ at 280 nm (Wilkins et al., 1999 ▸).

Table 1. Macromolecule-production information.

Source organism	C. aerofaciens
DNA source	Synthetic
Forward primer	GAAGTTCTGTTCCAGGGGCCCATGAATCTGCGTGAAAAATATG
Reverse primer	TTTGTTAGCAGCCGGATCTCAATCGCGATAAAAGCTGCCCAT
Cloning vector	pET-28a
Expression vector	pET-28a
Expression host	E. coli BL21 (DE3)
Complete amino-acid sequence of the construct produced	MGSSHHHHHHENLYFQGMSPILGYWKIKGLVQPTRLLLEYLEEKYEEHLYERDEGDKWRNKKFELGLEFPNLPYYIDGDVKLTQSMAIIRYIADKHNMLGGCPKERAEISMLEGAVLDIRYGVSRIAYSKDFETLKVDFLSKLPEMLKMFEDRLCHKTYLNGDHVTHPDFMLYDALDVVLYMDPMCLDAFPKLVCFKKRIEAIPQIDKYLKSSKYIAWPLQGWQATFGGGDHPPKSDLEVLFQGPMNLREKYGEWGLILGATEGVGKAFCEKIAAGGMNVVMVGRREEKLNVLAGEIRETYGVETKVVRADFSQPGAAETVFAATEGLDMGFMSYVACLHSFGKIQDTPWEKHEAMINVNVVTFLKCFHHYMRIFAAQDRGAVINVSSMTGISSSPWNGQYGAGKAFILKMTEAVACECEGTGVDVEVITLGTTLTPSLLSNLPGGPQGEAVMKIALTPEECVDEAFEKLGKELSVIAGQRNKDSVHDWKANHTEDEYIRYMGSFYRD

ITC was employed to obtain the thermodynamic parameters of the interaction between NADP⁺ and 7β-HSDH using a MicroCal iTC200 (Malvern, England). β-Nicotinamide adenine dinucleotide phosphate sodium salt was dissolved in the same buffer as 7β-HSDH (50 mM Tris–HCl buffer containing 150 mM NaCl pH 7.5). The sample cell contained 23 µM 7β-HSDH (300 rev min⁻¹ stirring speed with an initial delay of 60 s) and 1 mM NADP⁺ was titrated in. The cell temperature was kept at 298 K and 10 µcal s⁻¹ of reference power was applied to maintain a flat baseline. A total of 20 injections of 1 mM NADP⁺ were performed, where the initial injection was of 0.4 µl and the remaining 19 injections were 2 µl each. Each injection took place over a period of 0.8 s with an interval of 150 s and a filter time of 5 s. An equivalent control experiment to subtract the heat of dilution of the NADP⁺ in buffer was performed (i.e. the same concentration of NADP⁺ was injected into the buffer solution, keeping the same experimental parameters). The results obtained were plotted using the MicroCal Origin 6.0 software.

2.2. Crystallization  

Random microseed matrix screening (rMMS; D’Arcy et al., 2014 ▸) was performed at 293 K by the sitting-drop vapour-diffusion method. 200 nl protein solution was mixed with 180 nl reservoir solution plus 20 nl seeding solution and was equilibrated against 15 µl reservoir solution. The seeding solution was prepared by crushing a large fresh crystal of lysozyme (Hampton Research) and crystallization was set up with about 60 µl of well solution [0.6 M sodium chloride, 0.1 M sodium acetate trihydrate pH 4.5, 25%(v/v) glycerol]. Commercial crystallization kits from Hampton Research and Qiagen were used for crystal screening. Suitable crystals of the complex were observed in the condition 0.1 M sodium malon­ate pH 7.0, 12%(w/v) polyethylene glycol (PEG) 3350. The crystal was mounted for diffraction using a cryoprotection solution consisting of 0.1 M sodium malonate pH 7.0, 12%(w/v) PEG 3350, 25%(v/v) glycerol. Crystallization information is summarized in Table 2 ▸.

Table 2. Crystallization.

Method	Sitting-drop vapour diffusion
Plate type	Swissci SD-3
Temperature (K)	293
Protein concentration (mg ml−1)	19.6
Buffer composition of protein solution	20 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM DTT, 1%(v/v) DMSO, 0.1 mg ml−1 NADP+
Composition of reservoir solution	0.1 M sodium malonate pH 7.0, 12%(w/v) PEG 3350
Volume and ratio of drop	200 nl protein, 200 nl reservoir
Volume of reservoir (µl)	15

2.3. Data collection and processing  

Data were obtained using an ADSC Q315r detector (Area Detector Systems Corporation, USA) on BL17U1 at Shanghai Synchrotron Radiation Facility (SSRF). Since the crystallization condition contained cryoprotectants, the crystal was flash-cooled in liquid nitrogen and directly mounted for X-ray diffraction. The diffraction data were collected at 100 K and were processed with iMosflm (Battye et al., 2011 ▸). The crystal diffracted to 1.7 Å resolution and belonged to the tetragonal space group P4₁2₁2, as checked with POINTLESS (Evans, 2011 ▸), with unit-cell parameters a = b = 72.73, c = 171.31 Å. A total of 331 630 reflections were integrated to a resolution of 1.70 Å and were merged to obtain 51 175 unique reflections with an overall R _meas of 0.090 and a completeness of 99.04%. Data-collection statistics are summarized in Table 3 ▸.

Table 3. Data collection and processing.

Values in parentheses are for the outer shell.

Diffraction source	BL17U1, SSRF
Wavelength (Å)	0.97915
Temperature (K)	100
Detector	ADSC Q315r
Crystal-to-detector distance (mm)	249.77
Rotation range per image (°)	1
Total rotation range (°)	180
Exposure time per image (s)	1
Space group	P41212
a, b, c (Å)	72.73, 72.73, 171.31
α, β, γ (°)	90, 90, 90
Mosaicity (°)	0.74
Resolution range (Å)	49.26–1.70 (1.79–1.70)
Total No. of reflections	331630 (38606)
No. of unique reflections	51175 (7164)
Completeness (%)	99.4 (97.5)
Multiplicity	6.5 (5.4)
〈I/σ(I)〉	13.7 (4.3)
R merge (%)	8.3 (33.8)
R r.i.m. † (%)	9.0 (37.4)
Overall B factor from Wilson plot (Å2)	17.134

^† R _r.i.m., the precision-indicating merging R factor, is R _merge adjusted by a factor of [1/(N − 1)]^1/2, where N is the number of times a given reflection is observed.

2.4. Structure solution and refinement  

The contents of the unit cell were analysed using the Matthews coefficient (Matthews, 1968 ▸). The Matthews coefficient of this complex was 1.96 ų Da⁻¹, corresponding to the presence of two subunits of 7β-HSDH in the asymmetric unit. The structure was determined by molecular replacement using Phaser (McCoy et al., 2007 ▸) with the monomer from PDB entry 5fyd as a search model (Savino et al., 2016 ▸). Regions that were less defined or where the electron density was ambiguous were deleted from the model. The models were completed and refined by iterative cycles of manual building using Coot (Emsley et al., 2010 ▸). Subsequent stages of refinement were carried out with REFMAC5 (Murshudov et al., 2011 ▸) within the CCP4 suite (Winn et al., 2011 ▸; Collaborative Computational Project, Number 4, 1994 ▸) and manual improvement in Coot. During refinement, the two polypeptides in the asymmetric unit were treated independently. The quality of the structure was evaluated using MolProbity (Chen et al., 2010 ▸) and PROCHECK (Laskowski et al., 1993 ▸). Coordinates and structure factors have been deposited in the Protein Data Bank with accession code 5gt9. All structural representations were generated using PyMOL (DeLano, 2002 ▸). Structure-solution and refinement statistics are summarized in Table 4 ▸.

Table 4. Structure refinement.

Resolution range (Å)	44.92–1.70
Completeness (%)	99.21
No. of reflections, working set	48520
No. of reflections, test set	2554
Final R cryst	0.1655
Final R free	0.2012
Cruickshank DPI† (Å)	0.104
No. of non-H atoms
Protein	4049
Ligand	96
Water	203
Total	4348
R.m.s. deviations
Bonds (Å)	0.0179
Angles (°)	1.8089
Average B factors (Å2)
Protein	16.718
Ligand	17.826
Water	25.259
Ramachandran plot
Favoured regions (%)	94.38
Additionally allowed (%)	5.23
Outliers (%)	0.39

^†Diffraction-component precision indicator (Cruickshank, 1999 ▸).

3. Results and discussion  

7β-HSDH was purified to homogeneity as assessed by analytic size-exclusion chromatography, SDS–PAGE and LC-MS (Agilent 6224 TOF LC/MS, USA; Figs. 1 ▸ a, 1 ▸ b and 1 ▸ c). Size-exclusion chromatography identified that the enzyme forms a dimer in solution. ITC was used to obtain the thermodynamic parameters of the NADP⁺–7β-HSDH interaction. Using a one-site model, we calculated that NADP⁺ binds to 7β-HSDH with a dissociation constant K _d of approximately 22 µM, and the ITC titrations revealed a close to 1:1 binding stoichiometry (Fig. 1 ▸ d).

Figure 1.

Purification of the 7β-HSDH protein. (a, b) Analytic size-exclusion column chromatography of the purified 7β-HSDH protein. The absorbance of the eluate was monitored at 280 nm. The protein markers used to produce the the standard curve in (b) were purchased from GE Healthcare and were used according to the manufacturer’s protocol. (c) Coomassie-stained 12% SDS–PAGE showing the purity of 7β-HSDH. Lane 1, molecular-weight markers; lane 2, 5 µg 7β-HSDH; lane 3, 10 µg 7β-HSDH; lane 4, 20 µg 7β-HSDH. (d) LC-MS shows the calculated molecular weight of the purified 7β-HSDH protein. (e) ITC profiles for the titration of 7β-HSDH with NADP⁺.

The structure of the 7β-HSDH–NADP⁺ complex contains two monomers in each asymmetric unit, referred to here as ‘complex A’ and ‘complex B’, with each monomer binding one molecule of NADP⁺ (Fig. 2 ▸ a). Like most members of the short-chain dehydrogenases/reductases (SDR) family, seven β-strands constitute the enzyme core and 11 α-helices wrap around the core to make a typical Rossmann fold, with NADP⁺ binding in a bowl-like enzymatic centre in complex A and complex B (Bennett et al., 1997 ▸). A PDB BLAST search using the DALI server (Holm & Rosenström, 2010 ▸) shows that the whole structure of the 7β-HSDH–NADP⁺ complex resembles those of many dehydrogenases and the most similar structure is, of course, the apo form of 7β-HSDH (Savino et al., 2016 ▸). Surprisingly, structural alignment of the 7β-HSDH–NADP⁺ complex and apo 7β-HSDH by the DALI server gives a Z-score of 42.8 with a root-mean-square deviation of 1.5 Å for complex A and a Z-score of 42.1 with a root-mean-square deviation of 1.3 Å for complex B (Holm & Rosenström, 2010 ▸).

Figure 2.

Structure of the 7β-HSDH–NADP⁺ protein complex. (a) The structure of 7β-HSDH bound to NADP⁺ in cartoon representation. The unique 7β-HSDH-bound NADP⁺ is shown in olive, while the normal 7β-HSDH is in green. The unique and normal NADP⁺ molecules are shown in stick representation, with C and O atoms depicted in green and red, respectively, for the unique NADP⁺ and in yellow and red, respectively, for the normal NADP⁺. (b) OMIT map (3.0σ) demonstrating that the unique NADP⁺ binds to the 7β-HSDH cofactor-binding pocket. (c) OMIT map (3.0σ) demonstrating that the normal NADP⁺ binds to the 7β-HSDH cofactor-binding pocket. (d) The alignment of two NADP⁺ from complex A (unique) and complex B (normal). (e) Alignment of the unique 7β-HSDH–NADP⁺structure (olive) with apo 7β-HSDH (PDB entry 5fyd; wheat) shown as separate cylindrical representations. The unique 7β-HSDH–NADP⁺ has the same colour as in (a).

Although structural superposition of complex A and complex B reveals that the whole backbone of the structure is conserved, details of the interactions between 7β-HSDH and NADP⁺ show that the insertion of NADP⁺ gives rise to several differences in the cofactor-binding pocket in the two complexes. One difference is that the nicotinamide group of NADP⁺ in complex A pushes helix 8 out to expand the binding site. At the same time, the nearby helix 9 adjusts its position. The middle two phosphate groups appear to push helix 2 out. Another conformational difference appears to arise from the adenine, which also pushes helix 3 out to enlarge the cofactor-binding pocket in complex A (Figs. 2 ▸ b, 2 ▸ c, 2 ▸ d and 2 ▸ e). Helix 10 and helix 11 from complex B act as a lid to close the enzymatic centre (Fig. 2 ▸ a). The results of the ITC experiment fit a one-site model much better than a two-site mode; this might mean that the difference is difficult to detect using ITC or comes from the crystal packing.

Structural alignment of the 7β-HSDH–NADP⁺ complex with NAD⁺-bound 7α-HSDH proved interesting. The typical complex B of 7β-HSDH shows a similar structural arrangement to 7α-HSDH, with the cofactors NADP⁺ and NAD⁺ in almost the same orientation, with a root-mean-square deviation of 0.296 Å over 214 C^α atoms (Fig. 3 ▸). The Arg pair (Arg40-Arg41), Thr17, Glu18 and Phe67 of 7β-HSDH are perfectly located to make hydrogen bonds to the adenosine phosphate, as predicted by Savino et al. (2016 ▸); also, Val20, Val91, Lys160, Thr189, Thr191 and Ser193 make hydrogen bonds to the nicotinamide group of NADP⁺. In NAD⁺-bound 7α-HSDH, only Asp42 and Ile69 make hydrogen bonds to the adenosine, while Ile23, Asn95, Lys163, Ile192 and Thr194 make hydrogen bonds to the nicotinamide. In complex A of 7β-HSDH there is a large bend with about 90° torsion of the nicotinamide on comparing the molecular location of NAD⁺. In the alignment of 214 C^α atoms with a root-mean-square deviation of 0.296 Å, although there is only minor translation of the adenosine phosphate, the nicotinamide of NADP⁺ is rotated about 90° to push helix 8 of 7β-HSDH out to expand the bowl. In the interactions that stabilize NADP⁺, we found the predicted arginine pair (Arg40-Arg41), Glu18 and Phe67 remain to stabilize the adenosine phosphate, while Glu18, Val20 and Gly21 orient the nicotinamide with the help of Ser240 from another molecule of 7β-HSDH.

Figure 3.

Structural interaction details of the cofactors NADP⁺ and NAD⁺ with 7β-­HSDH and 7α-HSDH. (a) Superimposition of the unique 7β-HSDH–NADP⁺ and the normal 7β-HSDH–NADP⁺ complexes shown as line representations. (b) Superimposition of the unique 7β-HSDH–NADP⁺, the normal 7β-HSDH–NADP⁺ and the 7α-HSDH–NAD⁺ (PDB entry 1fmc; Tanaka et al., 1996 ▸) complexes shown as line representations. Hydrogen-bond interactions are indicated by black arrows. 7α-HSDH is shown in red and NAD⁺ is shown in stick representation, with C and O atoms depicted in light magenta and red, respectively.

In summary, our work presents the first structure of NADP⁺ bound to dimeric 7β-hydroxysteroid dehydrogenase. In one active site, NADP⁺ adopts a standard conformation similar to that observed in other SDR-family members. In the other site, the cofactor adopts a different mode of binding with respect to the orientation of the nicotinamide. Since 7β-HSDH has become an attractive enzyme owing to the wide and important pharmaceutical use of its product ursodeoxycholic acid, our work provides a more detailed template for rational protein engineering to improve the enzymatic activities of this useful biocatalyst, further improving the yield of ursodeoxycholic acid and its other applications.

Supplementary Material

PDB reference: 7β-hydroxysteroid dehydrogenase, 5gt9