Structural Basis of Tryptophan Reverse N-Prenylation Catalyzed by CymD

Abstract

Indole prenyltransferases catalyze the prenylation of l-tryptophan (l-Trp) and other indoles to produce a diverse set of natural products in bacteria, fungi, and plants, many of which possess useful biological properties. Among this family of enzymes, CymD from Salinispora arenicola catalyzes the reverse N1 prenylation of l-Trp, an unusual reaction given the poor nucleophilicity of the indole nitrogen. CymD utilizes dimethylallyl diphosphate (DMAPP) as the prenyl donor, catalyzing the dissociation of the diphosphate leaving group followed by nucleophilic attack of the indole nitrogen at the tertiary carbon of the dimethylallyl cation. To better understand the structural basis of selective indole N-alkylation reactions in biology, we have determined the X-ray crystal structures of CymD, the CymD-l-Trp complex, and the CymD-l-Trp-DMSPP complex (DMSPP is dimethylallyl S-thiolodiphosphate, an unreactive analogue of DMAPP). The orientation of l-Trp with respect to DMSPP reveals how the active site contour of CymD serves as a template to direct the reverse prenylation of the indole nitrogen. Comparison to PriB, a C6 bacterial indole prenyltransferase, offers further insight regarding the structural basis of regioselective indole prenylation. Isothermal titration calorimetry measurements indicate a synergistic relationship between l-Trp and DMSPP binding. Finally, activity assays demonstrate the selectivity of CymD for l-Trp and indole as prenyl acceptors. Collectively, these data establish a foundation for understanding and engineering the regioselectivity of indole prenylation by members of the prenyltransferase protein family.

Graphical Abstract

INTRODUCTION

Prenylated natural products often exhibit useful cytotoxic, antimicrobial, and antioxidant properties that are not observed in their non-prenylated precursors.^1,2 In plants, fungi, and bacteria, prenylation is catalyzed by enzymes known as prenyltransferases.³ These enzymes catalyze the transfer of a prenyl moiety from donors such as dimethylallyl diphosphate (DMAPP), geranyl diphosphate (GPP), or farnesyl diphosphate (FPP) to prenyl acceptors ranging from small molecule metabolites⁴ and peptides⁵ to proteins⁶ and nucleic acids.⁷ Prenylation can occur via the primary carbon of the allylic diphosphate donor (normal prenylation) or via the tertiary carbon (reverse prenylation).⁸

Aromatic prenyltransferases exist in both membrane-bound⁹ and soluble forms.¹⁰ Soluble aromatic prenyltransferases share a common αββα or ABBA-fold, also known as a PT-barrel, consisting of a central core of ten antiparallel β-strands surrounded by ten α-helices.^11–14 Though their tertiary structure is conserved, the soluble aromatic prenyltransferases can be further divided into two groups based on their primary structure: (1) the NphB/CloQ group, which prenylate napthalenes, quinones, phenols, and phenazines, and (2) the dimethylallyltryptophan synthase (DMATS) group, which prenylate indole derivatives, tyrosine, napthalenes, and xanthones.³ Members of the DMATS group are found in both bacteria and fungi, though their sequence identity is low. Characterizing the structural and enzymatic properties of members of the DMATS group has been an active area of investigation given their role in the biosynthesis of useful natural products.^15–17 There is currently a need for additional structural studies to better inform molecular modeling of this class of enzymes.

Nature frequently employs tryptophan as a template for the biosynthesis of natural products owing to the reactivity of its indole ring.^18–20 Indoles readily undergo electrophilic aromatic substitution reactions, including alkylation (e.g., prenylation),²¹ nitration,²² and halogenation,²³ to yield a variety of bioactive natural products including antibiotics, anti-cancer agents, herbicides, anti-inflammatories, and anti-fungal compounds.²⁴ The past decade has seen the discovery of indole prenyltransferases capable of prenylating l-Trp and indole derivatives at all seven sites of the indole moiety (Figure 1).^8,25

Figure 1.

Prenylation sites of l-tryptophan, with representative prenyltransferases from bacteria (blue) and fungi (red) noted for each site. CymD from S. arenicola;²⁶ FtmPT2 from A. fumigatus;²⁷ FtmPT1 from A. fumigatus;²⁸ KgpF from M. aeruginosa;²⁹ AnaPt from N. fischeri;³⁰ FgaPT2 from A. fumigatus;³¹ SCO7467 from S. coelicolor;³² 5-DMATS from A. clavatus;²⁵ PriB from Streptomyces sp. RM-5–8;³³ TleC from S. blastmyceticus;³⁴ 7-DMATS from A. fumigatus.³⁵ Shown is the reverse N-prenylation reaction catalyzed by CymD to produce N-DMAT.

In 2010, Schultz and coworkers reported the characterization of CymD from the marine actinobacterium Salinispora arenicola, the first bacterial tryptophan prenyltransferase to be identified.²⁶ CymD adopts the canonical ABBA-fold based on its primary structure, though it shares only ~25% sequence identity with other subsequently discovered ABBA-fold bacterial prenyltransferases. CymD catalyzes the reverse prenylation of the indole nitrogen of l-tryptophan and does so in cation-independent fashion. CymD is encoded by the cymD gene of the cym gene cluster responsible for the biosynthesis of the cyclic peptides cyclomarin A and cyclomarazine A, potent anti-inflammatory and antimicrobial compounds, respectively.³⁶ CymD exclusively utilizes DMAPP as the prenyl donor, as the use of isoprenyl, geranyl, or farnesyl disphosphate yields no detectable products.²⁶ Qian and coworkers report that the mechanism of CymD proceeds through a dissociative mechanism in which the dimethylallyl cation is formed in the first step of catalysis.³⁷ In the second step, deprotonation of the indole nitrogen by a general base is proposed to precede or coincide with nucleophilic attack by the indole nitrogen at the tertiary carbon of the dimethylallyl cation to yield N-dimethylallyl-l-tryptophan (N-DMAT).

Here, we report the 1.33 Å-resolution crystal structure of CymD complexed with l-Trp and dimethylallyl S-thiolodiphosphate (DMSPP), an unreactive analogue of DMAPP. This ternary complex reveals the orientation of the prenyl donor and acceptor during catalysis and supports the proposed mechanism by which prenylation proceeds via direct nucleophilic attack by the indole nitrogen, rather than via a normal C-3 prenylation followed by an aza-Cope rearrangement – an unlikely but plausible alternative reaction pathway.^37–39 Additional structures of CymD include the native (i.e., substrate-free) enzyme at 1.70 Å-resolution and the complex with l-Trp at 1.66 Å-resolution. Activity assays and isothermal calorimetry experiments offer additional insight into the substrate scope of CymD as well as the thermodynamics of ligand binding in the active site.

MATERIALS AND METHODS

Reagents.

Unless specified, chemicals used for protein expression and purification were purchased from Fisher, Sigma, RPI, or GoldBio and were used without further purification. Reagents used for crystallization were purchased from Hampton Research. l-Tryptophan (99%), d-tryptophan (99%), indole (99+%), indoline (98%), benzothiazole (97%), benzoxazole (99+%), benzimidazole (98%), aniline (99+%), l-glutamine (99%), and l-asparagine (99%) were purchased from Acros Organics. Dimethylallyl diphosphate (DMAPP) (>95%) and dimethylallyl S-thiolodiphosphate (DMSPP) (>95%) were purchased from Echelon Biosciences.

Gene Cloning.

The codon-optimized cymD gene from S. arenicola CNS-205 (GenBank protein accession no. ABW00334.1; UniProt accession no. A8M6W6) was purchased from GenScript. The full-length cymD gene (minus the start methionine) was amplified by PCR using the forward primer 5’-TACTTCCAATCCAATGCAACCGAGGAGCTGACCACC-3’ and reverse primer 5’-TTATCCACTTCCAATGTTATTACTCGGTGCGACCACGCGC-3’. The amplified cymD gene was inserted using ligation independent cloning (LIC) into a pET-His₆-GFP-TEV-LIC vector, a gift from Scott Gradia acquired from Addgene (plasmid no. 29663). The resulting pET-His₆-GFP-TEV-CymD plasmid was amplified in NEB-5α competent E. coli cells (New England Biolabs), purified using a Monarch® plasmid miniprep kit (New England Biolabs), and sequenced at the University of Pennsylvania DNA Sequencing Facility to verify the integrity of the cymD gene.

Protein Preparation.

The pET-His₆-GFP-TEV-CymD plasmid was transformed into BL21(DE3) competent E. coli cells (New England Biolabs) and cultured overnight at 37 °C on an LB-agar culture plate supplemented with 50 μg/mL kanamycin. Single colonies of transformed cells were used to inoculate 5-mL cultures of Terrific Broth (TB) media supplemented with 50 μg/mL kanamycin which were then incubated overnight at 37 °C with shaking at 250 rpm. The 5-mL cultures were used to inoculate 6 × 1 L of TB media supplemented with 50 μg/mL kanamycin. The 1-L cultures were incubated at 37 °C with shaking at 250 rpm until OD₆₀₀ reached ~1. Protein expression was induced by adding isopropyl β-l-1-thiogalactopyranoside (IPTG) to a final concentration of 1 mM. The induced cultures were incubated overnight at 25 °C with shaking at 250 rpm. Cells were pelleted by centrifugation and stored at −80 °C prior to purification.

Cell pellets were thawed and resuspended in 20 mM sodium phosphate (pH 7.4), 500 mM NaCl, 1 mM tris(2-carboxyethyl)phosphine hydrochloride (TCEP-HCl), 20 mM imidazole. The cell suspension was treated with 1 mM EDTA, 1 mg/mL lysozyme, 1 kU of benzonase, and protease inhibitor (Roche) and incubated for 1 h at 4 °C. Cells were then lysed by sonication followed by the addition of 2 mM MgCl₂. Cell lysate was clarified by centrifugation at 38 000 g for 30 min at 4 °C. The supernatant was loaded onto a 5-mL HisTrap™ HP column (GE Healthcare) and eluted via a 20-column volume gradient to 20 mM sodium phosphate (pH 7.4), 500 mM NaCl, 1 mM TCEP-HCl, 500 mM imidazole. The protein was treated with 3 mg of TEV protease to cleave the GFP tag and concurrently dialyzed into 20 mM sodium phosphate (pH 7.4), 500 mM NaCl, 1 mM TCEP-HCl, 20 mM imidazole at 4 °C. The protein digest was loaded onto a 5-mL HisTrap™ HP column, and flow-through fractions containing pure CymD were collected and pooled. CymD was concentrated to ~5 mL using an Amicon Ultra-15 filter (molecular weight cut-off = 30 kDa) and loaded onto a HiLoad™ 26/600 Superdex™ 200 pg size-exclusion column (GE Healthcare) pre-equilibrated with 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM TCEP-HCl. Elution fractions containing CymD were identified by SDS-PAGE (Figure S1), pooled, and concentrated to ~10 mg/mL.

Crystallization.

Crystals of CymD complexed with l-Trp and DMSPP were grown by the sitting drop vapor diffusion method at 21 °C. A 100-nL drop of protein solution [9.4 mg/mL CymD, 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM TCEP-HCl, 2 mM l-Trp, 2 mM DMSPP] was added to 100 nL of precipitant solution [1% (w/v) tryptone, 0.05 M sodium 4-(2-hydroxyethyl)piperazine-1-ethanesulfonate (HEPES-Na) (pH 7.0), 12% PEG 3350] and equilibrated against a 50-μL reservoir of precipitant solution. Crystals of native CymD were grown by the sitting drop vapor diffusion method at 4 °C. A 2-μL drop of protein solution [9.3 mg/mL CymD, 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM TCEP-HCl] was added to a 2-μL drop of precipitant solution [4% 2-propanol, 0.1 M Bis-Tris propane (pH 9.0), 20% PEG monomethyl ether 5000] and equilibrated against a 500-μL reservoir of precipitant solution. The sitting drop was streak seeded with microcrystals of CymD grown in the same precipitant. Crystals of CymD complexed with l-Trp were grown by the sitting drop vapor diffusion method at 21 °C. A 2.5-μL drop of protein solution [9.8 mg/mL CymD, 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM TCEP-HCl, 2 mM l-Trp] was added to a 2.5-μL drop of precipitant solution [(0.04 M citric acid, 0.06 M Bis-Tris propane (pH 6.4)), 20% PEG 3350] and equilibrated against a 400-μL reservoir of precipitant solution.

All CymD crystals were briefly immersed in cryoprotectant solution (mother liquor supplemented with either 25% ethylene glycol or 25% glycerol) before being flash-cooled in liquid nitrogen.

Data Collection and Structure Determination.

X-ray diffraction data from crystals of the CymD-l-Trp-DMSPP complex were collected on beamline 17-ID-2 (FMX) at the National Synchrotron Light Source II (NSLS-II), Brookhaven National Laboratory. Diffraction data from CymD crystals were collected remotely on Northeastern Collaborative Access Team (NE-CAT) beamline 24-ID-C at the Advanced Photon Source (APS). Diffraction data from CymD-l-Trp crystals were collected remotely on beamline 12–2 at the Stanford Synchrotron Radiation Laboratory (SSRL), Stanford University. Diffraction data were indexed and integrated using either iMosflm⁴⁰ or XDS⁴¹ and scaled using Aimless⁴² in the CCP4 program suite.⁴³ Although some diffraction data sets were characterized by high R_merge values, the precision-indicating-merging R factor (R_pim), which reflects the increased accuracy of highly redundant data sets, was satisfactory for all shells of data in all data sets. Diffraction data collection statistics are listed in Table 1.

Table 1.

Data Collection and Refinement Statistics

	CymD	CymD-l-Trp complex	CymD-l-Trp- DMSPP complex
Unit Cell
space group	I4	P212121	I4
a, b, c (Å)	157.7, 157.7, 56.3	85.3, 86.3, 141.7	129.4, 129.4, 49.8
α, β, γ (deg)	90, 90, 90	90, 90, 90	90, 90, 90
Data Collection
beamline	APS 24-ID-C	SSRL 12–2	NSLS-II 17-ID-2
wavelength (Å)	0.91950	0.97946	0.97932
resolution (Å)	55.8 – 1.70	86.3 – 1.66	29.1 – 1.33
total no. of reflections	1209693	681939	708407
no. of unique reflections	76243	123383	95021
Rmergea,b	0.226 (1.072)	0.104 (0.360)	0.064 (0.623)
Rpima,c	0.083 (0.394)	0.070 (0.251)	0.037 (0.375)
CC1/2a,d	0.994 (0.742)	0.986 (0.886)	0.999 (0.806)
I/σ(I)a	8.0 (2.9)	8.9 (3.5)	15.0 (2.3)
redundancya	15.9 (16.0)	5.5 (5.6)	7.5 (7.0)
completeness (%)a	99.9 (100)	99.6 (99.7)	99.8 (97.3)
Refinement
no. of reflections used in refinement/test set	76236/1996	123289/2000	95019/1993
Rworke	0.167	0.163	0.166
Rfreef	0.208	0.192	0.181
no. of protein chains	2	3	1
no. of non-hydrogen atoms
protein	5266	7951	2814
ligand	46	62	38
ions	-	-	3
solvent	649	905	411
average B factor (Å2)
protein	23	18	17
ligand	30	18	16
ion	-	-	15
solvent	32	30	31
root-mean-square deviation from ideal geometry
bonds (Å)	0.007	0.006	0.005
angles (deg)	0.8	0.8	0.9
Ramachandran plot (%)g
favored	97.5	98.4	97.5
allowed	2.5	1.6	2.5
outliers	0	0	0
MolProbity score	1.12	1.08	0.97
PDB accession code	6OS3	6OS5	6OS6

^aValues in parentheses refer to the highest-resolution shell of the data.

^bR_merge = ∑|I_h – ⟨I_h⟩|/∑⟨I_h⟩; I_h = intensity measure for reflection h; ⟨I_h⟩ = average intensity for reflection h calculated from replicate data.

^cR_pim = ∑(1/(n – 1)^1/2|I_h - ⟨I_h⟩|/∑⟨I_h⟩; n = number of observations (redundancy).

^dCC_1/2 = σ_τ²/(σ_τ² + σ_ε²), where σ_τ² is the true measurement error variance and σ_ε² is the independent measurement error variance.

^eR_work = ∑||F_o| - |F_c||/∑|F_o| for reflections contained in the working set. |F_o| and |F_c| are the observed and calculated structure factor amplitudes, respectively.

^fR_free = ∑||F_o| - |F_c||/∑|F_o| for reflections contained in the test set held aside during refinement.

^gAssessed by MolProbity.

The crystal structure of CymD-l-Trp-DMSPP was solved by molecular replacement using Phaser.⁴⁴ A search model based on the crystal structure of unliganded PriB prenyltransferase [Protein Data Bank (PDB) entry 5JXM]³³ was generated using the SWISS-MODEL server⁴⁵ and converted to a poly-ALA model using Chainsaw⁴⁶ in the CCP4 program suite. Iterative cycles of refinement and manual model building were performed using Phenix⁴⁷ and WinCoot,⁴⁸ respectively. Translation-libration-screw (TLS) refinement was performed during the later stages of refinement using TLS groups determined by Phenix. Disordered protein side chain atoms showing no electron density in the 2F_o-F_c map were deleted from the model, and electron density peaks that were not confidently interpretable were left unmodeled. Refinement proceeded until R_free converged at its lower limit. The quality of the final model was assessed using MolProbity.⁴⁹ Refinement statistics are presented in Table 1. All structure figures were generated using PyMOL.⁵⁰ Root-mean-square-deviation (rmsd) values were calculated using Superpose^51,52 in the CCP4 program suite.

The refined protein structure of the CymD-l-Trp-DMSPP complex, minus ligands and solvent, was used as the search model for phasing the initial electron density maps of native CymD and the CymD-l-Trp complex. These structures were refined and evaluated following the same procedure outlined above for the CymD-l-Trp-DMSPP complex. All refinement statistics are recorded in Table 1.

Enzyme Kinetics.

The prenyltransferase activity of CymD was measured using the EnzChek Pyrophosphate Assay Kit (Invitrogen). The reaction buffer was composed of 50 mM Tris-HCl (pH 7.5), 1 mM MgCl₂, and 0.1 mM sodium azide. The 100-μL reaction mixture contained (final concentrations): 200 μM 2-amino-6-mercapto-7-methylpurine ribonucleoside (MESG), 0.1 unit of purine nucleoside phosphorylase (PNPase), 2 units of inorganic pyrophosphatase (IPPase), 1 μM CymD, and 80 μM DMAPP. The assay was initiated by adding prenyl acceptor substrate to a final concentration of 500 μM. The reaction was monitored in triplicate in a flat-bottom transparent 96-well plate using a Tecan M1000 spectrophotometer. Reaction velocity was calculated from the increase in A₃₆₀ within the linear range.

Isothermal Titration Calorimetry.

The thermodynamics of l-Trp and DMSPP binding to CymD were measured using a MicroCal iTC 200 isothermal titration calorimeter (GE Healthcare). 300 μM of l-Trp or DMSPP was titrated into 30 μM CymD in 20 mM Tris-HCl, 200 mM NaCl, 1 mM TCEP-HCl, 10% (v/v) glycerol at 25 °C. Enthalpogram data were visualized and analyzed using Origin (OriginLab, Northampton, MA).

RESULTS AND DISCUSSION

X-ray Crystallography.

Crystal structures of CymD were solved and refined to R_work and R_free values ranging from 0.163 to 0.167 and 0.181 to 0.208, respectively. The low MolProbity scores indicate no significant errors in the protein models. The native CymD and CymD-l-Trp structures were quite similar to the structure of the CymD-l-Trp-DMSPP complex, with rmsd values of 0.79 Å for 344 C_α atoms and 0.67 Å for 354 Cα atoms, respectively. CymD is a monomer in three different crystal forms based on analysis with PISA,⁵³ consistent with size exclusion chromatography of the enzyme in solution.

The ternary complex of CymD bound to l-Trp and DMSPP crystallizes in space group I4 with one protein chain per asymmetric unit. Single molecules of l-Trp and DMSPP are bound in the enzyme active site, located in the central β-barrel of the ABBA-fold (Figure 2a). The carboxylate Oε1 atom of E64, which acts as a general base in catalysis, is positioned 2.8 Å from the indole nitrogen of l-Trp with syn orientation (Figure 2b). CymD makes only one other direct interaction with l-Trp – a 2.6 Å hydrogen bond donated by Y326 to the l-Trp carboxylate. However, there are several additional water-mediated hydrogen bonds between CymD and l-Trp. The indole ring of l-Trp forms an offset π-stacking interaction with the side chain of Y326 (Figure 2c). The side chains of Y274 and Y341 are oriented in perpendicular fashion to the indole ring of l-Trp, and F192 and Y209 are nearby. These residues form an aromatic box to accommodate and orient the indole ring of l-Trp.

Figure 2.

(a) Crystal structure of CymD complexed with l-Trp and DMSPP illustrates the ABBA fold with a central active site cavity. (b) Stereoview of the active site of CymD, with selected side chains (sticks) and water molecules (red spheres) shown. Hydrogen bonds are represented by black dashed lines. The interaction between the indole nitrogen and general base E64 is highlighted in cyan. Polder omit maps corresponding to l-Trp and DMSPP are shown as green mesh and contoured at 5σ. (c) Offset π-stacking interactions in the active site of CymD. Hydrogen bonds are shown as dashed black lines. Inter-planar distances between π systems (green dashes) are indicated (Å). The angle between the indole nitrogen and dimethylallyl plane is highlighted in pink.

The CymD-l-Trp-DMSPP crystal structure mimics the precatalytic Michaelis complex, revealing the spatial relationship between l-Trp and the dimethylallyl moiety prior to catalysis. The DMSPP molecule lies parallel to the l-Trp indole ring, with the dimethylallyl moiety positioned 3.7 Å away. The tertiary dimethylallyl carbon is optimally positioned for nucleophilic attack by the indole nitrogen: the nitrogen is positioned 3.4 Å away and oriented at an angle of 104° relative to the dimethylallyl plane, close to the Bürgi-Dunitz angle of 107°.⁵⁴ Notably, the primary dimethylallyl carbon is 5.9 Å from the indole C3 atom, effectively ruling out the possibility of an aza-Cope rearrangement mechanism and affirming the direct N-prenylation mechanism proposed by Qian and colleagues.³⁷ The dimethylallyl moiety of DMSPP is parallel to the indole ring of W148, which likely stabilizes the dimethylallyl cation intermediate via a cation-π interaction during catalysis. The thiolodiphosphate group of DMSPP lies within a positively-charged pocket in the CymD active site, with the β-phosphate forming salt bridges with the side chains of R205, K207, R337, and K339. The α-phosphate, on the other hand, forms mostly hydrogen bonds with CymD, receiving hydrogen bonds from the side chains of Q77, Y274, and Y341, and forms a single salt bridge with K146 (Figure 2b).

The crystal structures of native CymD and the CymD-l-Trp complex were solved to determine the extent of substrate-induced conformational changes in the CymD active site (Figure 3). Native CymD crystallizes with two monomers in the asymmetric unit. Both CymD monomers show a molecule of bis-tris propane bound in the active site, with an amino NH group donating a 2.8 Å hydrogen bond to E64 in chain B (this distance is 3.3 Å in chain A). Structural comparison of native CymD and the CymD-l-Trp-DMSPP complex shows that the active site side chains near l-Trp do not substantially reorient to bind substrate. Even in the absence of l-Trp, the E64 carboxylate is perfectly pre-oriented and poised to deprotonate the indole nitrogen, and the Y326 side chain is positioned for offset π-π stacking. Inspection of the diphosphate-binding channel, however, reveals that several positively-charged side chains reorient to form salt bridges upon DMSPP binding.

Figure 3.

(a) Stereoview of the active site pocket in native CymD (gray) and CymD complexed with l-Trp and DMSPP (purple). (b) Stereoview of the active site pocket in CymD complexed with l-Trp and benzoic acid (cyan) and CymD complexed with l-Trp and DMSPP (purple).

The CymD-l-Trp complex crystallizes with three monomers in the asymmetric unit. The active site of chain A contains bound l-Trp as well as what appears to be a molecule of benzoic acid stacked roughly 4.0 Å above the indole ring of l-Trp (Figure 3). Benzoic acid was not present in either the protein or precipitant solutions, so its origin is unclear. Benzoic acid accepts hydrogen bonds from Y274 (2.5 Å), Y209 (2.7 Å), and Q77 (3.0 Å), and also makes an offset π-π stacking interaction (4.1 Å) with W148. Similarly, the active site in chain B contains bound l-Trp along with what appears to be a molecule of imidazole. Like benzoic acid, the imidazole forms π-π interactions with the indole group of l-Trp (3.8 Å) and the side chain of W148 (4.1 Å). Additionally, imidazole forms a 2.7 Å hydrogen bond with a bound diphosphate molecule. Chain C shows positive electron density in the active site, though it does not fit any small molecules present in the crystallization drop (e.g. bis-tris propane, citrate) and was thus deemed uninterpretable and left unmodeled. The binding of benzoic acid and imidazole in chains A and B indicates that the offset π-π stacking network in the CymD active site is capable of binding other aromatic small molecules besides the dimethylallyl carbocation and indicates the potential of expanding the chemistry of electrophilic aromatic substitution reactions catalyzed by CymD.

CymD sequence analysis and structural relationships.

The amino acid sequence of CymD was aligned with sequences of several other bacterial indole prenyltransferases sharing the ABBA-fold to identify key conserved residues as well as highlight differences that could explain the regioselectivity of each prenyltransferase (Figure S2; Table S1). CymD shares 45% sequence identity with IlaO from Streptomyces atratus, another reverse l-Trp N-prenyltransferase located in the ila gene cluster responsible for the biosynthesis of ilamycin.⁵⁵ CymD, however, shares only 25–26% identity with the C5 prenyltransferase SCO7467,³² C6 prenyltransferases PriB³³ and IptA,⁵⁶ and C7 prenyltransferases MpnD⁵⁷ and TleC.³⁴

There are several conserved residues among the bacterial indole prenyltransferases. Notably, the general base E64 is conserved, highlighting its key role in indole activation.^33,59 The aromaticity of residue 326 is conserved (Tyr in the N, C5, C6 prenyltransferases; Phe in the C6 prenyltransferases), as well as the aromaticity of residues 209 (Tyr), 274 (Tyr or His), and 341 (Tyr) that comprise the aromatic box in the active site. Several residues lining the diphosphate binding pocket of the active site are conserved as well, notably K146, R205, R337, and Y341. These residues make direct contacts with the substrate diphosphate group in the CymD-l-Trp-DMSPP complex. This observation supports the hypothesis that indole bacterial prenyltransferases have evolved so as to fix the position of the prenyl diphosphate group, with prenyl acceptor specificity and prenylation regioselectivity dictated by subtle variations in the residues that interact with and align the dimethylallyl moiety and prenyl acceptor in the active site.^59–61

Structural comparison of the ternary CymD-l-Trp-DMSPP complex and the 1.4 Å-resolution structure of the C6 PriB-l-Trp-DMSPP complex (PDB ID 5INJ)³³ offers considerable insight regarding the relationship between active site architecture and prenylation regioselectivity. PriB catalyzes the normal prenylation of l-Trp at C6, and while it shares only 25% sequence identity with CymD, both enzymes share the same ABBA-fold (the rmsd of 319 C_α atoms is 1.8 Å). The orientation of l-Trp is largely conserved between both enzymes (Figure 4a), though PriB forms four direct contacts with l-Trp, whereas CymD forms only two. In both enzymes there is a tyrosine side chain parallel to l-Trp that stabilizes the prenyl acceptor via offset π-π stacking as well as a hydrogen bond with the α-carboxylate group of l-Trp.

Figure 4.

Comparison of l-Trp and DMSPP orientation between CymD and PriB. (a) Stereoview of l-Trp and DMSPP bound to CymD (purple) and PriB (green). (b) DMSPP orientation viewed along the z-axis. (c) Connolly surfaces of DMSPP and nearby side chains shows the repositioning of the dimethylallyl moiety resulting from alternative reaction templates.

The dimethylallyl moieties of bound DMSPP, however, are positioned differently in CymD and PriB relative to l-Trp (Figure 4b). In CymD the tertiary dimethylallyl carbon lies 3.4 Å from N1, whereas in PriB the primary dimethylallyl carbon lies 3.6 Å away from C6. This shift in dimethylallyl position results from differences in the side chains lining the dimethylallyl pocket (Figure 4c). A79 in CymD is substituted with L110 in PriB, with the longer side chain effectively pushing the dimethylallyl group closer to the C6 carbon of l-Trp. Likewise, on the opposite side, the L193 side chain pushes the dimethylallyl moiety in CymD towards the N1 position of l-Trp, while the perpendicular F220 in PriB allows the dimethylallyl to shift closer to the C6 position. M132 in CymD and W165 in PriB also make favorable contacts with the two methyl carbons of DMSPP, further fixing its position over l-Trp. Finally, the aromatic capping group stabilizing the carbocation during catalysis is different between the enzymes. The larger tryptophan side chain in CymD (W148) allows the dimethylallyl to move closer towards the l-Trp nitrogen, whereas in PriB the smaller tyrosine keeps it closer to the C6 of l-Trp. Collectively, these subtle differences in active site architecture direct regioselectivity by sterically orienting the prenyl donor into the required position for optimal nucleophilic attack by l-Trp. Indeed, inspection of these residues in the aligned sequences shows some degree of correlation between side chain identity and prenylation sites (Table 2).

Table 2:

Side chains proximal to the prenyl donor moiety

Enzymes	Prenylation Site	Capping Group	Residue 1	Residue 2	Residue 3
CymD, IlaO	N	Trp	Ala	Met	Leu/Met
PriB, IptA	C6	Tyr	Leu	Trp	Phe
MpnD, TleC	C7	Tyr/Phe	Thr/Ser	Met/Ala	Ser/Ala
SCO7467	C5	Tyr	Leu	Trp	Phe

Specific activity assays.

Whereas it was previously determined that CymD exclusively accepts DMAPP as the prenyl donor,²⁶ its specificity for the prenyl acceptor had not been investigated. To screen potential substrates for CymD, a series of indoles and indole-like small molecules were evaluated by a coupled activity assay⁶² measuring the production of inorganic pyrophosphate (Figure 5). l-Trp, the native substrate of CymD, yielded a specific activity of 36.1 nmol product · μmol enzyme⁻¹ · s⁻¹. CymD also showed prenylation activity toward d-Trp, though at a rate roughly 20% of that measured with l-Trp. Indole, however, showed prenylation activity equal to that of l-Trp, demonstrating that the amino acid moiety of l-Trp is not critical for the proper orientation of the prenyl acceptor and that offset π-π stacking interactions alone are sufficient for catalysis. This is consistent with the mechanistic conclusions emanating from the crystallography results. Indoline and aniline showed very little activity, likely due to the increased nitrogen pK_a (30.7 for aniline in DMSO)⁶³ resulting from the reduction of aromaticity adjacent to the nitrogen.

Figure 5.

Specific activities of CymD with DMAPP and various prenyl acceptor substrates.

The compounds benzimidazole, benzoxazole, and benzothiazole – indole analogs with the 3-carbon substituted for nitrogen, oxygen, and sulfur, respectively – showed very little prenylation activity. The very low reactivity of benzothiazole can be attributed to decreased aromaticity caused by the poor overlap of the sulfur 3p orbitals and 2p orbitals of the adjacent carbons. The very low level of benzimidazole prenylation, however, is perplexing: its nitrogen (pK_a = 16.4 in DMSO) is more acidic than that of indole (pK_a = 21.0 in DMSO)⁶⁴ and solvent isotope studies performed by Qian and coworkers identified nitrogen deprotonation as the rate-limiting step in catalysis.³⁷ Thus, from a thermodynamic perspective, benzimidazole would be expected to show an increased rate of prenylation by DMAPP. On the other hand, the nucleophilicity of the deprotonated benzimidazole nitrogen is likely decreased by the presence of the additional nitrogen atom in the aromatic π-system. Similarly, the electronegative oxygen in benzoxazole likely lowers the nucleophilicity of the deprotonated N1 atom. These data highlight the unique versatility of indole as a biosynthetic precursor – its nitrogen can act as an acid while still retaining its nucleophilic character.

The l-amino acids glutamine and asparagine were also tested for prenylation activity since the pK_a of their carboxamide nitrogens is ca. 25 (in DMSO)⁶⁵, roughly comparable to the indole nitrogen. Furthermore, in the case of glutamine, the side chain NH₂ group is approximately isosteric with the indole NH group of tryptophan. Neither l-Gln and l-Asn showed any prenylation activity. Bis-tris propane was also evaluated since it occupies the active site of the native CymD crystal structure with one of its amino groups donating a hydrogen bond to E64. Bis-tris propane showed very little activity, indicating that while flexible nitrogen-containing small molecules may potentially occupy the active site, they do not act as suitable prenyl acceptors.

Isothermal titration calorimetry (ITC).

Thermodynamic measurements using ITC were performed to assess the binding of l-Trp and DMSPP to CymD (Table 3; Figure S3). The titration of l-Trp into CymD yielded a dissociation constant (K_d) of 4.4 μM, in good agreement with the Michaelis constant (K_M) of 3.8 μM previously determined in steady-state kinetic assays.³⁷ Notably, titration of l-Trp into CymD preincubated with DMSPP yielded a K_d of 1.1 μM – a four-fold increase in affinity – suggesting a degree of synergy between l-Trp and DMSPP (or DMAPP) binding to CymD. The structural basis of this synergy is evident in the crystal structure of CymD: the simultaneous binding of l-Trp and DMSPP establishes a network of favorable offset π-π stacking interactions. Moreover, whereas the titration of DMSPP into CymD produced no measurable heats of binding, DMSPP titrated into CymD preincubated with l-Trp yielded a K_d of 1.0 μM, further demonstrating a synergistic interaction between the two substrates. The typical intracellular concentrations of tryptophan and DMAPP in bacteria are 12 μM and 140 μM, respectively.^66,67 Inspection of the solvent-accessible surface of CymD reveals two channels accessing the active site cavity from two opposite sides of the protein surface (Figure 6), suggesting that each substrate can access the active site independently.

Table 3.

Thermodynamic values from ITC measurements of l-Trp and DMSPP binding to CymD

	Kd (μM)	ΔH (kcal mol−1)	ΔS (cal mol−1 deg−1)
l-Trp into CymD	4.4 ± 0.4	−10.6 ± 0.2	−11.0
l-Trp into CymD-DMSPP	1.1 ± 0.2	−16.0 ± 0.3	−26.4
DMSPP into CymD	-	-	-
DMSPP into CymD-l-Trp	1.0 ± 0.2	−5.2 ± 0.1	10.2

Figure 6.

Solvent-accessible surface of CymD. Bound l-Trp and DMSPP shown as yellow sticks. (Left) Channel accessing the l-Trp-binding pocket in the active site. (Right) Channel accessing the DMSPP-binding pocket in the active site.

CONCLUSIONS

Reverse N-prenylation of l-Trp by CymD proceeds through the alkylation of the indole nitrogen, which is not ordinarily a nucleophilic atom in the biological or nonbiological context. To better understand the structural basis for this reaction, CymD was crystallized in the presence of l-Trp and DMSPP, an unreactive analogue of substrate DMAPP. The orientations of bound l-Trp and DMSPP in the ternary complex reveals how the indole nitrogen is optimally positioned for deprotonation by general base E64, followed by nucleophilic attack at the allylic cation intermediate resulting from departure of the diphosphate leaving group of DMAPP. Substrate l-Trp and co-substrate analogue DMSPP are bound and oriented to direct exclusive N---C bond formation with the tertiary carbon of the dimethylallyl cation.

Structural comparison of CymD to another bacterial indole prenyltransferase, PriB, provides additional insight as to how each enzyme active site directs alternative regiochemistry for the electrophilic aromatic substitution reaction. While CymD and PriB both employ offset π-π interactions as the primary means of accommodating l-Trp with a common binding mode, differences in the residues proximal to the dimethylallyl moiety of DMSPP position the incipient carbocation intermediate adjacent to alternative ring atoms of the l-Trp indole group, thereby directing alternative prenylation reactions.

The selectivity and cooperativity of substrate binding to CymD were studied by activity assay and ITC, respectively. CymD showed prenylation activity towards only l-Trp, indole, and to a lesser extent, d-Trp. Surprisingly, CymD showed no prenylation activity towards benzimidazole despite its lower nitrogen pK_a, indicating that other factors must play important roles in the mechanism of electrophilic aromatic substitution. Finally, ITC measurements of l-Trp and DMSPP binding to CymD suggest a synergistic interaction between the two substrates. Furthermore, the crystal structure of CymD reveals that both substrates may access the active site independently through different channels.

l-Trp is a common precursor to myriad natural products, and the family of indole prenyltransferases constitute a promising platform for the biosynthesis of such small molecules in biotechnology applications.^1,15,16,59 As additional bacterial indole prenyltransferases are discovered and studied by structural methods, the correlation between primary structure, active site architecture, and prenylation substrate specificity will be further clarified.

ABBREVIATIONS

DMAPP

dimethylallyl pyrophosphate

DMSPP

dimethylallyl S-thiolodiphosphate

PDB

Protein Data Bank

ITC

isothermal titration calorimetry

GFP

green fluorescent protein

TEV

tobacco etch virus

Tris-HCl

tris(hydroxymethyl)aminomethane hydrochloride

TCEP-HCl

tris(2-carboxyethyl)phosphine hydrochloride

PEG

polyethylene glycol