Characterisation of geranylgeranyl diphosphate synthase from the sandfly Lutzomyia longipalpis

,


Introduction
The mevalonate (MEV) pathway is an essential metabolic pathway utilised by eukaryotes, archaea and certain bacteria, through which the 5-carbon compound isopentenyl diphosphate, also known as isopentenyl pyrophosphate, (IPP) is synthesized.IPP can then be reversibly converted to the allylic dimethylallyl diphosphate (DMAPP) through action of IPP isomerases.Sequential head to tail condensation reactions of DMAPP and IPP, mediated by isoprenyl diphosphate synthase (IDS) enzymes, yields longer isoprenoid chains such as geranyl diphosphate (GPP, C 10 ), farnesyl diphosphate (FPP, C 15 ) and geranylgeranyl diphosphate (GGPP, C 20 , Fig. 1A).These are essential for diverse biochemical processes including protein prenylation and terpene/ terpenoid biosynthesis (Buhaescu and Izzedine, 2007).IDS enzymes are divided into cis and trans superfamilies, defined by the stereochemistry of the allylic double bonds formed during catalysis (Z or E respectively) (Chen et al., 2020a).For completeness, it is noted that an alternative pathway to IPP and DMAPP, known as the methyl-D-erythritol phosphate (MEP) pathway, exists in most bacteria, the apicoplasts of some parasites and in plant plastids, but it is believed that insects, and animals in general, are incapable of performing these metabolic transformations (Banerjee and Sharkey, 2014).
The New World phlebotomine sandfly Lutzomyia longipalpis is a carrier of the Leishmania parasite, the causative pathogen of the neglected tropical disease leishmaniasis (Lainson et al., 1977).Male sandflies produce sex/aggregation pheromones, of which there are several defined geographical chemotypes in Brazil (Spiegel et al., 2016).These are variably comprised of three well-characterised pheromone components -the homosesquiterpenes 3-methyl-α-himachalene and 9-methylgermacrene-B, and the diterpene sobralene (Fig. 1A and B) (Hamilton et al., 1999a(Hamilton et al., , 1999b;;Palframan et al., 2018).Terpenes are produced by terpene synthases (TPS), which operate downstream of IDS activity.Studying the regulation of isoprenoid diphosphate production is a prerequisite to understanding subsequent secondary metabolism and terpene synthesis.The isoprenoid precursor for diterpenes is GGPP, which is produced by IDS enzymes known as geranylgeranyl diphosphate synthases (GGPPS).IDS enzymes are characterised by dual aspartate-rich motifs (generally DDxxD) which are important for metal cofactor coordination and diphosphate binding in the active site, and referred to as the first aspartate-rich motif (FARM) and second aspartate-rich motif (SARM).The amino acids at the − 4 and − 5 position from the FARM are generally considered as contributory factors in determining product chain length, and variations in this region categorise a GGPPS as type I (large residue at − 4/− 5; archaea), type II (two residue insertion in FARM; bacteria and plant) or type III (small residues at − 4 and − 5; fungi and animals) (Wang and Ohnuma, 1999).
GGPPS enzymes generally can accept DMAPP, GPP or FPP as allylic substrates (requiring three, two, and one equivalents of IPP to produce the final product respectively, see Fig. 1) but kinetic studies suggest that in the case of type III GGPPSs, FPP is the preferred substrate (Sagami et al., 1994).In plants, several gene paralogs are generally present, which can be ubiquitously expressed or restricted to various developmental stages and in specific tissues (Beck et al., 2013;Coman et al., 2014).Arabidopsis thaliana encodes 10 functional GGPPS proteins, with variable subcellular localisation ranging from cytosolic, mitochondrial and plastidial (Beck et al., 2013).However, essential plastidial GGPP production in Arabidopsis (preceding carotenoid biosynthesis) is derived from a single GGPPS, while two co-operative plastid-targeted enzymes carry out this function in tomato (Solanum lycopersicum), illustrating diversity in the regulation of isoprenoid chain synthesis in plants (Barja et al., 2021).
A relative paucity of information exists on the molecular characteristics of insect GGPPSs and their regulation.In contrast to plants, a single gene is usually present, indicated by the lethality imparted on Drosophila melanogaster following nonsense mutation, which highlights a lack of redundancy and the essential role of the enzyme (Lai et al., 1998).GGPPS expression in the spruce budworm moth Choristoneura fumiferana varied throughout development alongside other prenylation-linked genes, with increased expression directly preceding the pupal stage, followed by a reduction until adult emergence (Barbar et al., 2013).The bumblebee Bombus terrestris also showed age-dependent variation of GGPPS expression in the labial gland, where GGPP is converted to the diterpene pheromone component geranylcitronellol (Prchalová et al., 2016).GGPPS depletion indirectly influenced mono/sesquiterpene production in the aphids Aphis gossypii and Megoura viciae (Sun and Li, 2018;Song et al., 2022).Increased GGPPS expression in the green versus red morph pea aphid (Acrythosiphon pisum) correlated with higher carotenoid pigment levels in green morphs, while gene knockdown significantly reduced carotenoid levels in both without altering their chemical profile (Ding et al., 2019).
Structural analysis of human GGPPS found it to be hexameric (in the form of a trimer of dimers) with sequence alignments indicating this is also likely to be the case for insect members (Kavanagh et al., 2006;Barbar et al., 2013).This is in contrast to the dimeric form generally seen in farnesyl diphosphate synthases (FPPS) and fungal, bacterial, archaeal, and plant GGPPSs (Reed and Rilling, 1975;Chang et al., 2006;Kloer et al., 2006;Feng et al., 2020).One study also suggested that human GGPPS existed as an octamer in solution (Kuzuguchi et al., 1999).The human enzyme was found to co-crystallise with GGPP product bound in an inhibitory conformation in the active site, and evidence from  C. fumiferana suggests this is also likely to extend to insects (Kavanagh et al., 2006;Barbar et al., 2013).
The enzymes responsible for isoprenoid production and downstream terpene synthesis in L. longipalpis have not been characterised to date.In this study, we have identified a trans-GGPPS enzyme from the sandfly genome, and following heterologous expression and purification of recombinant protein, have determined its functional activity and quaternary structure.Unravelling the complex pathways governing aggregation pheromone biosynthesis in L. longipalpis could allow for targeted monitoring of sandfly populations to control the spread of leishmaniasis.

Phylogenetic analyses and protein structural modelling
Protein sequences corresponding to reported insect GGPPSs from B. terrestris (I1VX07), D. melanogaster (O61539), A. pisum (A0A8R2B6W9), A. gossypii (S5MIM6) and C. fumiferana (U3RD44) were retrieved from the UniProt database along with various other mammalian, fungal, bacterial and plant sequences and aligned with L. longipalpis (lzGGPPS, A0A7G3AJG6) using MUSCLE.The maximum likelihood tree was constructed in MEGA11 using LG + G + I method with partial deletion of gaps, and phylogeny was tested by bootstrap method (1000 replicates).Structural model prediction of monomeric lzGGPPS was carried out using I-TASSER, which was arranged into a hexamer in ChimeraX 1.4 using the structure of human GGPPS (PDB 2Q80) as a guide (Kavanagh et al., 2006;Yang and Zhang, 2015).

Protein expression and purification
The presence of protein localisation signals in lzGGPPS primary structure were predicted using TargetP-2.0and iPSORT (Armenteros et al., 2019;Bannai et al., 2002).DNA encoding full-length lzGGPPS in expression vector pET100/D-TOPO (N-terminally His-tagged) was synthesized by GeneArt (Thermo Fisher Scientific, Loughborough, UK).The plasmid was transformed into BL21(DE3)pLysS E. coli competent cells for heterologous protein expression.Starter cultures of 10 mL LB (with 100 μg/mL ampicillin and 34 μg/mL chloramphenicol) were incubated with shaking from a single colony overnight at 37 • C, and used to inoculate a 500 mL culture the following day.Upon culture turbidity reaching OD 600 ~0.5, cells were induced with 0.75 mM IPTG and harvested after a further 3 h at 37 • C.

Protein mass spectrometry
Protein (lzGGPPS -7.5 μg) was submitted to SDS-PAGE, with bands excised, destained and dehydrated with acetonitrile.Bands were reduced (10 mM DTT), alkylated (55 mM iodoacetamide) and digested with trypsin (0.5 μg, Promega) at 37 • C for 18 h.Formic acid (1 μL) was added to protein digests to quench protease activity.The supernatant was removed from gel pieces, centrifuged at 3000×g for 2 min and retained for nanoLC-MS/MS analysis.Samples (2.5 μL) were submitted to LC-MS/MS using an Ultimate 3000 RSLC nano HPLC system equipped with a Thermo PepMap 300 C18 column for separation (150 mm × 75 μm, 300 Å, 5 μm particle size) coupled to a Thermo LTQ FT Ultra Mass Spectrometer containing a nano-ESI source (Thermo Fisher, Loughborough, UK).The mobile phase consisted of a binary gradient of 95:5:0.1 water:acetonitrile:formic acid (A) and 5:95:0.1 water:acetonitrile:formic acid (B) with a flow rate of 400 μL/min (Table S1A).Protein sequencing was performed in data dependent acquisition (DDA) mode using collision induced dissociation (CID) fragmentation of the 3 most abundant ions in the precursor spectra.Thermo files were converted to mz5 format using MS Convert, part of the ProteoWizard package (Chambers et al., 2012).Database searching was performed using SearchGUI.X! Tandem was the search engine used, and post-processing was performed with Peptide Shaker (Barsnes and Vaudel, 2018); Peptide length 4-40, max missed cleavages 1, precursor tolerance 10 ppm with carbamidomethylation of C considered a fixed modification and oxidation of M and deamidation of N as variable modifications.
Products were identified by reverse phase liquid chromatographymass spectrometry (LC-MS) using a Waters Acquity Ultra Performance Liquid Chromatography (UPLC) system equipped with a BEH C18 column (1.7 μM, 2.1 mm × 50 mm; Waters Corporation, Wilmslow, UK) coupled to a Waters Synapt G1 Quadrupole Time-of-Flight MS operated in negative electrospray ionisation mode.Sample injection volume was 5 μL with a flow rate of 0.3 mL per minute and a column temperature of 20 • C, with the mobile phase consisting of a binary gradient of aqueous ammonium bicarbonate (5 mM, solvent A) and analytical-grade acetonitrile (solvent B; Table S1B/C give gradient parameters for short and long separation methods).The capillary voltage was set at − 2.3 kV, with desolvation gas flow rate and temperature at 600 L/h and 500 • C respectively.Products (and standards) were identified from their deprotonated ions (EIC m/z 313.1 for GPP, m/z 381.2 for FPP, m/z 395.2 for homo-FPP (HFPP),m/z 449.2 for GGPP, m/z 463.2 for homo-GGPP (HGGPP) and m/z 477.2 for bishomo-GGPP (HHGGPP)).For further confirmation of product by MS/MS, a fixed collision energy of 15 V was used to fragment the parent ion and release diphosphate (m/z 158.9) and phosphate ions (m/z 79.0).For confirmation of double bond stereochemistry, after incubation at 30 • C IDS assays were treated with 20 U shrimp alkaline phosphatase (New England Biolabs, M0371S, Ipswich, USA) for 4 h at 37 • C, with resulting alcohol products extracted using cyclohexane and submitted to GC-MS (see 2.7).

Terpene synthase assays
TPS activity assays were carried out by combining 5 μM lzGGPPS or lzIDS11 protein (or ~2 μM psTPS4) with 50 μM (E)-GPP, (E,E or Z,E)-FPP or (E,E,E)-GGPP in TPS buffer (10 mM Tris-HCl pH 7.0, 10% v/v glycerol, 10 mM MgCl 2 , 1 mM DTT, 200 μL total volume) overlaid with 200 μL cyclohexane.Assays were incubated at 30 • C for 90 min, before the aqueous and organic layers were mixed with a glass pipette, and separated by centrifugation at 3000 rpm for 2 min.The aqueous layer was discarded, and the organic layer evaporated to ~50 μL under a stream of nitrogen.
Products were identified by gas chromatography-mass spectrometry (GC-MS) using a Jeol AccuTOF GCx (Welwyn Garden City, UK) system equipped with a DB-5ms column (Agilent, Cheadle, UK).Sample injection volume was 5 μL, with helium carrier gas set at a constant flow rate of 1 mL/min and the inlet temperature set at 200 • C (Splitless mode).The oven started with a 4-min hold at 35 • C, followed by an increase of 10 • C/min to 300 • C and a 6 min hold.MS spectra were obtained in electron ionisation (EI+) mode within an m/z range of 40-500, following an 8-min solvent delay.Product identities were confirmed by comparing mass spectra and retention times to known standards/positive control assay products.

Analytical size-exclusion chromatography (SEC)
The native molecular weight (MW) in solution of lzGGPPS was assessed by analytical SEC using an ÄKTA pure fast protein liquid chromatography (FPLC) system (Cytiva, Little Chalfont, UK).Protein (0.25 mg) was applied using a 500 μL loading loop to a pre-equilibrated Superdex 200 Increase 10/300 GL column (Cytiva, Little Chalfont, UK) with a flow rate of 0.5 mL/min.Blue dextran 2000 (2000 kDa) was used to determine the column void volume (V o ), while two sets of protein standards were run separately for calibration curve construction [Thyroglobulin (669 kDa), Ferritin (440 kDa), Aldolase (158 kDa) and Ovalbumin (44 kDa)]; [Conalbumin (75 kDa), Carbonic anhydrase (29 kDa) and Aprotinin (6.5k Da)].The partition coefficient (K av ) of lzGGPPS and protein standards was calculated using the equation where V e = elution volume and V t = total column volume.Protein standard K av values were plotted against logMW using GraphPad Prism software and used to interpolate lzGGPPS apparent MW.

The L. longipalpis genome contains a GGPPS homologue
We carried out blastp and tblastn searches to identify GGPPS homologues from the recently published L. longipalpis genome (Jacobina, NCBI accession PRJNA20279) and a transcriptomic study of the midgut of lab-raised female sandflies (Jacobina, NCBI accession PRJNA649559), looking for amino acid sequence similarities and particularly presence of the conserved aspartate rich motifs (Coutinho-Abreu et al., 2020;Labbé et al., 2023).Two strong candidates, which we named lzGGPPS1 (GITU01003271.1/A0A7G3AJG6) and lzGGPPS2 (A0A1B0CWK5) were identified (Yang and Wu, 2019).The former was obtained from both transcriptomic and genomic data and the latter from genomic data only.This study focuses on lzGGPPS1 (referred to from now on as lzGGPPS), as lzGGPPS2 was identical in sequence other than being truncated at the C-terminus (Fig. S1).Construction of a phylogenetic tree comprising various reported insect GGPPS proteins, in addition to examples from other eukaryotic organisms and bacteria, revealed that lzGGPPS clustered most closely to that of D. melanogaster (Fig. 2A).Comparative amino acid alignments of insect members showed high conservation, with the majority of variation present at the N and C-termini (Fig. 2B).Notably, the SARM was identical across insect proteins, with the final aspartate residue replaced with asparagine, as were the − 4 and − 5 positions upstream of the FARM (SS).Assessment of lzGGPPS for the presence of IPP binding motifs (IBMs) showed high conservation of residues expected to be important for IPP interaction and IDS activity (13/15 -Fig.S2) (Rebholz et al., 2023b).The N-terminus was interrogated for potential signalling/mitochondrial targeting peptides with none identified.We concluded that lzGGPPS was therefore likely to represent a trans-GGPPS enzyme from L. longipalpis.

lzGGPPS is an active enzyme capable of accepting various substrates
In order to study lzGGPPS functionality, recombinant protein was expressed in bacteria and purification was carried out using immobilised metal affinity chromatography (Fig. 3).Protein identity was confirmed with LC-MS/MS peptide mass fingerprinting and database searching (Table S3).Following incubation of lzGGPPS with various allylic substrates and IPP, assessment of product profile by LC-MS showed that the enzyme was active and was capable of synthesizing GGPP (Fig. 4A, S3).In agreement with other type III GGPPS enzymes, (E,E)-FPP appeared to be the preferred substrate, although de novo activity from DMAPP was also detected with an apparent (E,E,E) product configuration (Fig. 4B,  S3E) (Sagami et al., 1994).To confirm product identity, MS/MS was utilised to fragment GGPP and release diphosphate and phosphate ions (Fig. S4).We saw no evidence of extension beyond GGPP to geranylfarnesyl diphosphate (data not shown).The stereochemistry of the GGPP produced from (E,E)-FPP and IPP was confirmed as (E,E,E), following conversion of the product to geranylgeraniol through phosphatase treatment and assessment of retention time by GC compared to a dephosphorylated (E,E,E)-GGPP standard (Fig. S5).Enzyme activity was also assessed through an orthogonal plate assay screening for the release of inorganic diphosphate (PPi) following condensation of isoprenoid units (see 2.6).Pleasingly, the relative activity across allylic substrate precursors was broadly consistent with LC-MS data (Fig. 4C).Of note, we found IPP trilithium salt (Supelco 39784) to give high background in PPi readings while IPP triammonium Fig. 2. Sequence analysis of geranylgeranyl diphosphate synthases (GGPPSs).A. Phylogenetic tree of selected insect, mammalian, fungi, bacterial and plant GGPPS enzymes inferred by maximum-likelihood analysis (unrooted).B. MUSCLE alignment of insect GGPPS enzymes.Conserved trans-IDS motifs are underlined (I-VI), with active site aspartate residues in the FARM (II) and SARM (V) highlighted in red (Ding et al., 2019).Residues upstream of the FARM (− 4 and − 5) important for determining product chain length are highlighted turquoise.salt (Echelon I-0050) did not, meaning the latter was chosen for use in subsequent assays.
Interestingly, addition of EDTA (20 mM) to inactivate lzGGPPS resulted in release of a small amount of GGPP from the enzyme, as seen by LC-MS, which was also the case with incubation of active enzyme with allylic substrate only (DMAPP), but not IPP only (data not shown).In previous work, human GGPPS was shown to co-crystallise with sequestered GGPP bound in the active site, which may explain this observation.GGPP of bacterial origin co-purifies with lzGGPPS and is released following enzyme inactivation or competition for binding with allylic substrate precursors (Kavanagh et al., 2006).Levels of GGPP released in this way were always significantly lower (by LC-MS) than those seen with productive substrates -a result supported by the PPi kinetic assays.To rule out any IDS activity from contaminating E. coli-derived proteins, we carried out LC-MS assays with another IDS-like bacterially expressed protein preparation from Lutzomyia that we have found to be inactive.This enzyme, which we have named lzIDS11 (GITU01006826/A0A1B0CVV3), gave no diphosphate products with any combination of allylic substrate and IPP (data not shown) confirming the absence of activity carrying over from the expression system.
In addition to the more familiar (E,E)-isomer, (Z,E)-FPP was also tolerated as a substrate of lzGGPPS, albeit with significantly reduced output.This experiment yielded two distinct GGPP peaks (Fig. S3D).The early-eluting GGPP peak is likely to be (E,Z,E)-GGPP from (Z,E)-FPP extension, whereas the later peak is displaced (E,E,E)-GGPP from E. coli as described above.The homosesquiterpenes produced in L. longipalpis contain an additional methyl group, probably due to incorporation of homo-IPP (HIPP) into homo-FPP by FPPS enzymes, before cyclisation into the final terpene products (Hamilton et al., 1999a(Hamilton et al., , 1999b)).The diterpene, sobralene, isby contrasta regular diterpene.Despite this, we introduced HIPP into our IDS assays to assess whether lzGGPPS is capable of producing homo-isoprenoid chains when provided with the appropriate substrate.With an equimolar mix of HIPP and IPP, lzGGPPS was able to extend (E,E)-FPP into both homo-GGPP (HGPP) and GGPP, although the latter was the major product by a ratio of approximately 1:3, when taking into account the sequestered GGPP of bacterial origin (Fig. S6A).Similarly, incubating (E)-GPP with HIPP and IPP gave the full range of products possible, including bishomo-GGPP (HHGGPP), although regular GGPP was by far the most prevalent (GGPP:HGGPP: HHGGPP 20:3:1), suggesting HIPP is a non-optimal substrate for lzGGPPS (Fig. S6B).
The identity and prevalence of the divalent metal cation can impact insect IDS isoprenoid product specificity and activity (Frick et al., 2013;Rivera-Perez et al., 2015).A comparative analysis of the impact of various divalent metal ions on lzGGPPS enzymatic activity with DMAPP + IPP as substrates suggested that Mg 2+ was the most effective cofactor (Fig. 5A).There was some activity with the inclusion of Zn 2+ and Mn 2+ , while Co 2+ gave no product and led to formation of an insoluble precipitate.Interestingly, some enzyme activity was retained in the absence of metal addition, possibly due to lzGGPPS co-purifying with Mg 2+ and/or GGPP from the bacterial expression system.With (E)-GPP as the allylic substrate, there was evidence of some minor FPP release, consistent with previously characterised GGPPS enzymes (Fig. 5B) (Sagami et al., 1994;Kavanagh et al., 2006;Barbar et al., 2013).However, increasing Mg 2+ concentration appeared to inhibit FPP production.Further comparison of metal ions with (E)-GPP + IPP as substrates confirmed Mg 2+ as the best lzGGPPS cofactor for promoting GGPP production (Fig. S7).

Pseudo terpenoid synthase activity in lzGGPPS
A recent study on the postman butterfly identified GGPPS-like enzymes exhibiting TPS activity for production of its anti-aphrodisiac pheromone, the monoterpene (E)-β-ocimene, (Darragh et al., 2021).We assessed whether lzGGPPS possesses similar activity by incubating the enzyme with substrate precursors commensurate with mono-(GPP), sesqui-(FPP) and diterpene (GGPP) synthesis.No mono-or diterpene activity could be detected by GC-MS analysis, which was also the case for (Z,E)-FPP (data not shown).However, when incubated with (E, E)-FPP, two product peaks were visible by GC-MS and identified as the sesquiterpene alcohols nerolidol (major peak) and farnesol (minor peak, Fig. S8A).This did not occur when the negative control (lzIDS11) was incubated with (E,E)-FPP, ruling out activity from contaminating bacterial phosphatases (Fig. S8B).Incubation of a characterised TPS from the striped flea beetle (Phyllotreta striolata, psTPS4) with (E,E)-FPP produced (E)-nerolidol (as expected, Fig. S8C) (Beran et al., 2016).The spectrum of the product from lzGGPPS matched that produced by psTPS4, and the retention time was identical, confirming that lzGGPPS produces (E)-nerolidol.Use of a dephosphorylated (E,E)-FPP standard confirmed the identity of the minor product as (E,E)-farnesol.
To estimate the relative TPS versus GGPPS activity of lzGGPPS, diphosphate release was screened following incubation of the enzyme with (E,E)-FPP in the presence or absence of IPP using a diphosphate assay kit.PPi release (reflecting lzGGPPS activity) was negligible in the absence of IPP when compared against either lzGGPPS with IPP or psTPS4 with (E,E)-FPP only (Fig. S9A).The diphosphate assay was also used to confirm that incubation of (E,E)-FPP with lzIDS11 yielded negligible background activity (Fig. S9B).We therefore conclude that lzGGPPS is predominantly an active GGPPS with some pseudo-TPS character caused by activation of the allylic substrate and quenching of the carbocation by water in the absence of IPP co-substrate to yield the isomeric sesquiterpene alcohols.

Kinetic characterisation of lzGGPPS and product inhibition
For a more detailed biochemical analysis of lzGGPPS, we carried out Michaelis-Menten assessments of the enzyme's kinetics.Activity of GGPPS enzymes has been shown to be stimulated by the addition of surfactants (Sagami et al., 1994;Kavanagh et al., 2006).In agreement with this, addition of 0.2% Tween 20 significantly increased PPi release by lzGGPPS (Fig. S9C).Under these conditions, when varying (E,E)-FPP concentration while holding IPP constant, lzGGPPS gave an apparent V max of 0.48 μM PPi/min/μg protein and a K m of 167.4 μM (Fig. 6A, Table 1).This represents a very high K m compared with other reported values from GGPPS enzymes, which are generally in the range of 2-20 μM (Kavanagh et al., 2006;Chang et al., 2006;Barbar et al., 2013;Heider et al., 2014).With IPP as the variable substrate, V max was 0.14 μM PPi/min/μg protein and K m was 7.9 μM, which more closely aligns with previous studies of GGPPSs (Fig. 6B, Table 1).Introduction of GGPP into PPi assays significantly reduced enzyme velocity (Fig. 6C).This is in agreement with an established consensus that GGPP product acts as a negative feedback inhibitor, with phosphates bound in the allylic site of the catalytic cavity of type III enzymes (Fig. S10) (Sagami et al., 1994;Kavanagh et al., 2006;Barbar et al., 2013).

lzGGPPS exhibits hexameric quaternary structure
To determine the MW of lzGGPPS in solution, we employed analytical SEC to separate proteins according to hydrodynamic radius.Comparison of column retention to a range of protein standards gave an apparent MW of 221 kDa (Fig. 7A and B).Immunoblot of collected fractions confirmed that lzGGPPS was the source of the A 280 peak (Fig. S11).Given that a monomer of recombinant lzGGPPS is ~41 kDa, this implies a hexameric native configuration is likely (~246 kDa), consistent with the arrangement of the human enzyme (Kavanagh et al., 2006)(Fig.7C).

Discussion
In eukaryotes, synthesis of GGPP is essential for geranylgeranylation of target proteins, granting them a hydrophobic C-terminus and promoting interaction with cellular membranes (Wang and Casey, 2016).Geranylgeranylation of RAS proteins in Bombyx mori was found to be the sole form of prenylation, whereas in mammals farnesylation is more   prevalent (Moriya et al., 2010).Moreover, in some insects GGPP is required for production of diterpene/terpenoid semiochemicals, and for carotenoid synthesis (Hojo et al., 2011;Prchalová et al., 2016;Palframan et al., 2018;Ding et al., 2019).We have identified and biochemically characterised a GGPPS enzyme from New World phlebotomine sandflies, which is capable of producing the precursor for protein prenylation and/or sobralene production.This represents the first detailed study of an IDS from L. longipalpis and only the fourth functional study of an insect GGPPS; the others being in C. fumiferana, A. gossypii and M. viciae (Barbar et al., 2013;Zhang and Li, 2014;Song et al., 2022).
The catalytic behaviour of lzGGPPS appears to agree generally with other studies of type III enzymes, where allylic substrates are preferred in the order (E,E)-FPP > (E)-GPP > DMAPP.While some activity remained with the non-typical allylic substrate (Z,E)-FPP, this was significantly lower than that for (E,E)-FPP, as would be expected for a trans-IDS.We also found that lzGGPPS could accept HIPP as a substrate, albeit less efficiently than conventional IPP.Substrate promiscuity in IDS enzymes has been long established, particularly in FPPSs, which can accept a range of artificial isoprenoids at both allylic and non-allylic binding sites (see Couillaud et al., 2022 for a recent review).Experimental evidence from the microorganism Sulfolobus acidocaldarius GGPPS has found it to be similarly open to atypical substrates, including HIPP (Ohnuma et al., 1998;Nagaki et al., 2004).Lepidoptera have evolved FPPS enzymes that selectively extend homo (ethyl-branched)-DMAPP for juvenile hormone synthesis, which can also act as a non-optimal substrate for spruce budworm GGPPS (Sen et al., 2007;Barbar et al., 2013).The fact that some chemotypes of L. longipalpis produce the homosesquiterpene pheromone components 9-methylgermacrene-B (7) and 3-methyl-α-himachalene (8) with highly defined regiochemistry (the additional carbon is incorporated in the second isoprene unit only) raises the question of how this is achieved mechanistically.Is it, for example, controlled by the flux/location of HIPP vs IPP, or do one or more of the enzymes in the biosynthetic pathway select for a specific HFPP isomer?If the latter, it could be envisaged that either a specialist HFPPS exists, or that the homosesquiterpene synthases favour one form of HFPP, or that these two properties work in tandem.Sobralene ( 6) is a regular diterpene, possessing only C 5 isoprene units, the precursor of which would be the principal product of the L. longipalpis GGPPS described here.Despite this, our results show that the enzyme can incorporate HIPP, albeit at a reduced rate compared to IPP, if the former is made available.Thus, the interplay between substrate-availability and the presence of IDSs and TPSs with varying specificities is likely to be the key to different terpene product profiles in this organism.
Despite appearing to be the most efficient substrate of lzGGPPS, (E, E)-FPP gave a very high apparent K m value following kinetic analysis.Currently, the only steady-state kinetic report of an insect GGPPS is for spruce budworm, with a K m of 2.1 μM with respect to FPP, suggesting lzGGPPS is a significantly less efficient enzyme with a lower allylic substrate affinity (Barbar et al., 2013).Nonetheless, for IPP K m readings are comparable between spruce budworm and sandfly, calculated at 4.0 μM and 7.9 μM respectively (Barbar et al., 2013).Despite use of different methods in assessment of enzyme velocity, V max estimates were also broadly similar (for (E,E)-FPP 2456 pmol/h/μg protein for C. fumiferana versus 5760 pmol/h/μg for L. longipalpis; for IPP 1904 pmol/h/μg versus 1680 pmol/h/μg) (Barbar et al., 2013).Moreover, k cat readings for lzGGPPS with regard to IPP (1.2 min − 1 = 0.02 s − 1 ) reflect a ~10-fold reduction compared to human enzyme (0.204 s − 1 ) while being close to values reported for S. cerevisiae (0.025 s − 1 ) (Kavanagh et al., 2006;Chang et al., 2006).Comparison to closer evolutionary relatives of L. longipalpis, such as D. melanogaster (quemao) will inform whether these kinetic constants are similar across dipteran insects (Lai et al., 1998).
In contrast to plants and bacteria, insect TPSs have begun to be characterised only relatively recently.From the examples we know to date, they appear to have evolved from IDS ancestors following gene duplication (Gilg et al., 2009;Beran et al., 2016;Lancaster et al., 2019).Genome analysis of Heliconius melponene butterflies identified several predicted GGPPSs, with expression analysis revealing a male bias in two of these, and subsequent functional characterisation revealing TPS activity in both (Darragh et al., 2021).We found that lzGGPPS is able to dephosphorylate (E,E)-FPP in the absence of IPP and release (E)-nerolidol and (E,E)-farnesol as products.(E)-Nerolidol production from FPP by heterologously expressed insect TPSs has been described on more than one occasion (Beran et al., 2016;Darragh et al., 2021).It is a  (Kavanagh et al., 2006).common constituent in plant essential oil and can stimulate a defensive response against herbivorous insects and pathogens (Chan et al., 2016;Chen et al., 2020b).(E,E)-Farnesol is an important intermediate in juvenile hormone production in insects (Noriega, 2014).More broadly, production of terpenoid alcohols in insects by IDS-like TPS enzymes has also been reported in the tea green leafhopper Empoasca onukii (geraniol) and the stinkbugs Murgantia histrionica, Halymorpha halys and Nezara viridula (all sesquipiperitol) (Zhou et al., 2019;Lancaster et al., 2018;Rebholz et al., 2023a).However, sesquiterpenoid alcohol production by lzGGPPS occurs at low levels and is likely due to quenching of the allylic carbocation intermediate by water in the absence of nucleophilic IPP, giving either nerolidol or farnesol depending on the site of attack.This reaction may be even less favourable in vivo through tight control of FPP/IPP substrate flux from metabolic pathways.Furthermore, the use of recombinant proteins and substrate at concentrations far above those found in cellulo may also yield non-biologically relevant products.
Insect GGPPS enzymes have previously been predicted to adopt a hexameric arrangement based on conservation of three hydrophobic regions important for inter-dimer interactions in the human enzyme (Barbar et al., 2013).For the first time, we have confirmed this to be the case, with evidence pointing to a complex of 5-6 monomer units and ruling out the dimeric complex seen in non-animal GGPPSs.The apparent molecular weight in solution of previously-described human GGPPS also appeared to correspond to a complex numbering 5-6 units, with the crystal structure confirming a propeller-bladed trimer of dimers (Kavanagh et al., 2006).It is hence likely that this assembly is represented across all insect GGPPS enzymes.
In summary, we have elucidated the enzymatic function of a protein responsible for isoprenoid biosynthesis in L. longipalpis, the first to be characterised for this species complex.Further work is required to identify potential TPSs associated with downstream terpene synthesis, and how this is regulated in the sandfly.A greater understanding and biochemical reconstitution of these pathways could give a route to sustainable aggregation pheromone production for use in control pathways -a potentially viable route to limit the spread of leishmaniasis.

C
.Ducker et al.

Fig. 6 .
Fig. 6.Michaelis-Menten kinetic analysis of recombinant lzGGPPS activity.A. Plot measuring the initial rate of PPi release with varying concentrations of (E,E)-FPP and a constant IPP concentration (100 μM), displayed as averages of three independent experiments (±SEM).B. As for A with varying concentrations of IPP and a constant (E,E)-FPP concentration (40 μM).C. Initial rate of PPi release from lzGGPPS activity with varying concentrations of added GGPP and constant concentrations of (E,E)-FPP (40 μM) and IPP (50 μM), displayed as averages of three independent experiments (±SEM).Statistical analyses were performed using one-way ANOVA with Dunnett's multiple comparisons test; significance is reported in figures by *P < 0.05; **P < 0.01.

Fig. 7 .
Fig. 7. Quaternary structure of recombinant lzGGPPS.A. Analytical SEC plot of A 280 against retention volume following injection of lzGGPPS on Superdex 200 Increase 10/300 GL column (Cytiva).B. Plot of the partition coefficient (K av ) against log MW for a range of protein standards, with the relative position of lzGGPPS indicated in red.C. I-TASSER structural prediction of lzGGPPS arranged into a hexamer in line with structure of human GGPPS(Kavanagh et al., 2006).

Table 1
Apparent kinetic parameters of lzGGPPS for substrates (E,E)-FPP and IPP.