Semi-synthetic cinnamodial analogues: Structural insights into the insecticidal and antifeedant activities of drimane sesquiterpenes against the mosquito Aedes aegypti

The Aedes aegypti mosquito serves as a major vector for viral diseases, such as dengue, chikungunya, and Zika, which are spreading across the globe and threatening public health. In addition to increased vector transmission, the prevalence of insecticide-resistant mosquitoes is also on the rise, thus solidifying the need for new, safe and effective insecticides to control mosquito populations. We recently discovered that cinnamodial, a unique drimane sesquiterpene dialdehyde of the Malagasy medicinal plant Cinnamosma fragrans, exhibited significant larval and adult toxicity to Ae. aegypti and was more efficacious than DEET–the gold standard for insect repellents–at repelling adult female Ae. aegypti from blood feeding. In this study several semi-synthetic analogues of cinnamodial were prepared to probe the structure-activity relationship (SAR) for larvicidal, adulticidal and antifeedant activity against Ae. aegypti. Initial efforts were focused on modification of the dialdehyde functionality to produce more stable active analogues and to understand the importance of the 1,4-dialdehyde and the α,ß-unsaturated carbonyl in the observed bioactivity of cinnamodial against mosquitoes. This study represents the first investigation into the SAR of cinnamodial as an insecticide and antifeedant against the medically important Ae. aegypti mosquito.

Introduction strains of Ae. aegypti, CDIAL was more efficacious than DEET [N,N-Diethyl-meta-toluamide]-the gold standard for insect repellents-at repelling mosquitoes from feeding on blood [21]. Moreover, we demonstrated that the mechanism of the antifeedant activity of CDIAL was through the activation of transient receptor potential A1 (TRPA1) channels [21].
The goal of the present study was to investigate the structural basis of the insecticidal and antifeedant activities of 1 against Ae. aegypti by generating a series of semi-synthetic CDIAL derivatives. Our efforts led to the discovery of 10 ((-)-6ß-Acetoxy-9α-hydroxydrim-7-ene-12-methyl-12-one-11-al) as a CDIAL derivative with superior and similar insecticidal activity against larvae and adult females, respectively, but weaker antifeedant activity against adult females. Herein, we describe the re-isolation of 1, analog generation, and biological evaluation of synthetic derivatives. Additionally, in order to understand the mechanisms of the most active compounds, the observed structure-activity relationship (SAR) and potential interaction of the active compounds with several nucleophilic residues of the mosquito TRPA1 are discussed.

General experimental procedures
Optical rotations ([α] D ) were measured with an LED light source monitoring at 589 nm in acetonitrile at 20 o C. The instrument used was an Anton Paar MCP 150 polarimeter (Anton Paar OptoTec GmbH, Seelze-Letter, Germany). Ultraviolet (UV) absorption spectra were measured in a 1 cm quartz tank using a Hitachi U-2910 UV/vis double-beam spectrophotometer (Hitachi High-Technologies America, Schaumburg, IL, USA). Infrared (IR) spectra were obtained on a Nicolet 6700 FT-IR spectrometer (Thermo Scientific, Waltham, MA, USA). The 1D ( 1 H, 13 C, selective COSY and NOESY) and 2D (COSY, HSQC, HMBC, and NOESY) NMR spectra were recorded at 300 K (26.85 o C) in CDCl 3 for all compounds except for 5 and 19 which were measured in methanol-d 4 , 7 which was measured in acetonitrile-d 3 , and 16 was measured in pyridine-d 5 on a Bruker Avance III HD 400 MHz instrument (Bruker, Billerica, MA, USA) using standard Bruker pulse sequences. 1 H chemical shifts are reported in parts per million (ppm) and are referenced to the residual CDCl 3 signal (δ 7.26 ppm), methanol-d 4 signal (δ 3.34 ppm), pyridine-d 5 signal (δ 7.22 ppm), or acetonitrile-d 3 signal (δ 1.96 ppm). 13 C chemical shifts are reported in ppm and are referenced to the residual CDCl 3 signal (δ 77.36 ppm), methanol-d 4 signal (δ 49.86 ppm), pyridine-d 5 signal (δ 123.87 ppm) or acetonitrile-d 3 signal (δ 118.77 ppm). Deuterated NMR solvents were purchased from Cambridge Isotope Laboratories (Tewksbury, MA, USA). High-resolution mass spectra (HRESIMS) were acquired on a hybrid spectrometer utilizing a linear ion trap and Orbitrap (LTQ Orbitrap, ThermoFisher Scientific Inc., Bremen, Germany) equipped with an ESI source in the positive-ion mode, with sodium iodide (NaI) being used for mass calibration. The spectrometer was equipped with an Agilent 1100 HPLC system (Agilent Technologies, California, USA) including a binary pump, UV detector, and autosampler. Data acquisition and analysis were accomplished with Xcalibur software version 2.0 (Thermo Fisher Scientific Inc., Bremen, Germany). The samples were prepared at a concentration of~10 μg/mL in ACN, and the injection volume was set at 30 μL. Flash and open-column chromatography were performed with SilicaFlash P60 (230-400 mesh; SiliCycle Inc., Quebec City, Canada) using solvent systems as described. Ratios of solvent systems used for chromatography are expressed in v/v as specified. Analytical thin-layer chromatography was performed on aluminum-backed precoated silica gel plates (0.24 mm; Dynamax Adsorbant, Inc., Darmstadt, Germany). Spots were visualized under UV light (254 & 320) and by spraying with modified Godin's reagent (vanillin/EtOH-Perchloric acid, 1:1, v/ v) and H 2 SO 4 -water (15%) followed by heating. Commercially available chemicals were used as purchased. Ice/water was used as the temperature bath to achieve 0˚C and dry ice/acetone was used to achieve −78 o C. Dry tetrahydrofuran (THF) and dichloromethane (CH 2 Cl 2 ) were obtained from an Innovative Technology PureSolv system (Inert, Massachusetts, USA). Unless otherwise noted, reactions were performed in oven-dried glassware under an atmosphere of dry argon.

Plant material
The stem bark of C. fragrans was purchased at the market of traditional medicines, Analakely/ Antananarivo, (Madagascar). The bark was identified by comparison with the authentic sample in the Herbarium of PBZT (Parc Botaniqueet Zoologique de Tsimbazaza, Antananarivo, Madagascar). A voucher specimen of the bark (LivCF2016) was deposited at the College of Pharmacy, The Ohio State University (Columbus, Ohio).

Extraction and isolation
The air-dried stem bark of C. fragrans was pulverized and the powder (400 g) was extracted with dichloromethane for 5 days at room temperature. The extract was filtered and concentrated in vacuo to yield a yellow-brown oily residue (80.68 g, 20.2% on dry plant material). The residue was divided into fractions using column chromatography over silica gel, eluting with a gradient system of hexanes-EtOAc (from 4:1 to 0:1). Cinnamodial (1) was recrystallized using hexanes-EtOAc (1:1) as colorless crystals (5.28 g, 1.32% on dry plant material).
6-O-acetyl-12α-methyl-pereniporin A (8) and 6-O-acetyl-12ß-methyl-pereniporin A (9). To a solution of 1 (50 mg, 0.16 mmol) in dry THF (1 mL), methylmagnesium bromide (36.4 mg, 0.486 mmol, 3 equiv) in THF (162 μL) was added dropwise at −78 o C. The reaction mixture was maintained at this temperature, with stirring for 30 min, after which it was allowed to reach r.t. and maintained for another 1.5 h. The reaction mixture was then quenched with NH 4 Cl (1 mL of a saturated aqueous solution). The aqueous solution was extracted with EtOAc (2 mL × 3) and the combined organic layers dried over anhydrous MgSO 4 , filtered, and concentrated. The resulting residue was purified by column chromatography with hexanes-EtOAC (from 3:1 to 2:1) as eluents to afford two diastereomeric compounds 8 and 9.

Mosquito cultures and rearing conditions
Eggs of the Liverpool (LVP) strain of Ae. aegypti were obtained through the MR4 as part of the BEI Resources Repository, NIAID, NIH (LVP-IB12, MRA-735, deposited by M.Q. Benedict). The eggs of the Ae. aegypti were reared to adults as described previously [22].

Toxicology assays
Larval and adult female toxicities were determined by established protocols [21,23,24]. In brief, for larvae, 10 μL of a derivative (10 mM dissolved in 100% acetone) was added to six wells of a 24-well tissue culture plate containing 1 mL of dH 2 O and five larvae per well, resulting in a final concentration of 100 μM (1% acetone). The screening concentration (100 μM) was chosen based on preliminary scouting experiments with CDIAL (1) that resulted in~50% efficacy. After 24 h in normal rearing conditions (28˚C, 80% relative humidity), the efficacy was assessed by counting the number of larvae per well that did not move after gentle prodding with a micropipette tip or fine insect pin. As a negative and positive control, respectively, the effects of 1% acetone and 100 μM CDIAL were tested in parallel.
For adults, 500 nL of a derivative (3 mM dissolved in 100% acetone) was applied to the thorax of 10 adult females (3-7 days post emergence) to deliver a dose of 1.5 nmol per mosquito. The screening dose (1.5 nmol/mosquito) was chosen based on preliminary scouting experiments with CDIAL (1) that resulted in~50% efficacy. After dosing, the mosquitoes were held under normal rearing conditions in small plastic cages (32 oz. containers) and provided with access to 10% sucrose. After 24 h, the efficacy was assessed by counting the number of treated mosquitoes that were dead or unable to fly. As a negative and positive control, respectively, the effects of 100% acetone and 1.5 nmol CDIAL were tested in parallel.
The toxicity of compound 13 was also tested by direct injection into the hemolymph of adult females using an approach similar to Raphemot et al. 2013 [25]. In brief, 500 nL of 13 (3 mM dissolved in phosphate buffered saline, PBS, with 3% DMSO) was injected into the hemolymph with a pulled-glass capillary attached to a Nanoject II injector (Drummond Scientific Company, Broomall, PA) to deliver a dose of 1.5 nmol per mosquito. The efficacy was assessed at 24 h as described above for adult females. As a negative and positive control, respectively, 3% DMSO (in PBS) and 1.5 nmol CDIAL (in PBS with 3% DMSO) were tested in parallel.
For all toxicity experiments, the mean efficacies of the derivatives and CDIAL were adjusted for effects of the negative control using Abbott's correction [26] and compared statistically using a one-way ANOVA (Bonferroni post-test) or unpaired t-test (P < 0.05).

Antifeedant assays
A capillary feeding (CAFE) choice assay was used to determine the antifeedant activity of the compounds [21,27,28]. In brief, groups of 5 adult female mosquitoes (3-10 days post-emergence) were placed in Drosophila vials (28.5 x 95 mm) covered with cotton plugs. Two 5-μL calibrated glass capillaries were inserted through the cotton into the vial. One capillary was designated the 'control' and filled with 5 μL of 10% sucrose containing 1% DMSO (the solvent of the compounds). The other capillary was designated the 'treatment' and filled with 5 μL of 10% sucrose containing 1 mM of CDIAL or a derivative (1% DMSO). The screening concentration (1 mM) was chosen based on preliminary scouting experiments with CDIAL (1) that resulted in an antifeedant index of~0.5. All vials were placed in normal rearing conditions (28˚C, 80% relative humidity) for 18-20 h after which the volume of sucrose consumed from each capillary was measured visually with calipers. To determine if a compound possessed significant antifeedant activity, the relative volumes consumed in the treatment vs. control capillaries were compared with paired t-tests (P < 0.05). The relative volumes consumed in each capillary were also used to calculate an antifeedant index by subtracting the volume consumed from the treatment capillary from that of the control capillary and dividing by the total volume consumed from both capillaries [21,27,28]. Each derivative was tested on 5-10 vials of 5 adult females. The mean antifeedant indices of the derivatives and CDIAL was compared statistically using a one-way ANOVA (Bonferroni post-test).

Computational modeling
The modeled structure of the AgTRPA1 monomer (amino acid residues: 537-1191) consists of the last 4 repeats of ankyrin repeat (AR) domain, a transmembrane domain, a linker region between the AR and transmembrane domains, and a C-terminal domain. Several computational approaches were integrated to build different parts of the AgTRPA1 structure, including ab initio modeling [29], homology modeling, and loop modeling [30], resulting in a tetrameric structural model of AgTRPA1. This structure was constructed by first forming an AgTRPA1 monomer with GalaxyFill [31] using the cryo-EM structure of hTRPA1 (PDB ID: 3J9P) [32] as a template. Subsequently, a tetramer was assembled with GalaxyHomomer [33] and ZDOCK [34]. Lastly, the tetramer was embedded in a 1-palmitoyl-2-oleoyl-glycero-3-phosphocholine (POPC) bilayer, and the system was minimized/relaxed with short (~10 ns) molecular dynamics (MD) simulations using CHARMM36 force fields [35] and GROMACS [36].
CDIAL was docked to four potential binding pockets near key nucleophilic cysteine and lysine residues (Cys621, Cys641, Cys665 and Lys710 in hTRPA1 and Cys684, Cys704, Lys728 and Lys777 in AgTRPA1) that have been implicated to form covalent bonds with electrophilic agonists [37][38][39]. Among those, the pocket centered around Cys684 in AgTRPA1 was particularly promising. All docking calculations were performed with the Lamarckian genetic algorithm using Autodock 4.2 [40]. A 96×68×78 grid box with a grid spacing of 0.375 Å centered around each of the four nucleophilic residues defined the region of the protein that the ligands would explore. 500 docking runs were performed for each AgTRPA1 pocket.

Isolation and identification of bioactive drimane sesquiterpenes
The powdered stem bark of C. fragrans (Canellaceae) was extracted with dichloromethane, concentrated, and subjected to silica gel column chromatography and recrystallization to afford cinnamodial (1) (Fig 1).

PLOS NEGLECTED TROPICAL DISEASES
Structural insights into the activities of drimane sesquiterpenes against Ae. aegypti [54,55] or primary amines such as the ε-amino group of lysine [56] , [57]. Our previous results showed that 1 could effectively kill mosquito larvae in an aqueous environment, penetrate the cuticle of adult female mosquitoes, reduce the feeding of mosquitoes when added to a sucrose solution, and reduce the propensity of mosquitoes to blood feed when dried onto the surface of a membrane feeder. Despite these promising results, nonspecific reactions of dialdehydes with endogenous free amines or water may reduce the bioavailability of these compounds. In this work, initial semi-synthetic modification was focused on producing more stable derivatives by replacing the two aldehyde groups while maintaining the α,ß-unsaturated system.
Cinnamodial (1) was used as starting material for derivatization and analog generation (Fig  2). Pinnick oxidation was employed to generate cinnamodiacid (5), wherein the C-11 and C-12 aldehydes were successfully oxidized to the corresponding carboxylic acids. The dimethyl ester derivative 6 was obtained by reacting 5 with (Trimethylsilyl)diazomethane. Reaction of 1 with hydrazine afforded a 2,3-dihydro-3-pyridazinol 7, which maintained the unsaturated system.
It was envisioned that the more reactive C-12 aldehyde of 1 could be selectively transformed into a methyl ketone by treatment with a methyl Grignard or methyl lithium reagent followed by oxidation of the resulting secondary alcohol. Treatment of 1 with one equivalent of methylmagnesium bromide or methyl lithium instead produced two diastereomeric lactols 8 and 9, instead of a secondary alcohol. As determined by NMR studies (see experimental), the lactols 8 and 9 differ at the C-12 position depending on whether the C-12 aldehyde of 1 was attacked from the α or ß face, respectively (Fig 3). Lactol 9 was then converted into the desired compound 10, with a C-11 aldehyde and C-12 ketone, upon incubation in deuterated chloroform. However, the isomer 8 remained stable in deuterated chloroform and even resisted baseinduced lactol-opening / tautomerization with 1,8-Diazabicyclo(5.4.0)undec-7-ene (DBU), see supporting information. To understand the role of the unsaturated system and confirm the necessity of the C-12 unsaturated aldehyde for its observed biological activity [21], 1 was exposed to one equivalent of ethylene glycol and a catalytic amount of p-toluenesulfonic acid in benzene to afford the monoacetal 11 (Fig 2).
Since there have been no studies showing the reaction of CDIAL with thiol or amine residues, the reaction of L-cysteine methyl ester with CDIAL was used as a model system to study the reaction of 1 with a biological substrate which may in principle react by either thiol or amine addition. CDIAL was treated with L-cysteine methyl ester under basic conditions (in pyridine) to afford cinnamothiazolidine (16). This adduct likely formed from the initial attack by the primary amine at the more reactive C-12 aldehyde to give the azomethine A intermediate, which was then attacked by the thiol to form the thiazolidine 16 (Fig 6). The reaction product and proposed reaction mechanism suggest that the α,ß-unsaturated 1,4-dialdehyde 1 may also activate TRPA1 by forming reactive pyrrole-type conjugates with the amino groups present in the protein, as demonstrated for warburganal 2, polygodial 3, and 1ß-acetoxy-9-deoxyisomuzigadial (17) [49]. Additionally, the attack of the azomethine A by the thiol of L-cysteine methyl ester suggests that the reactive imine can be attacked by nearby nucleophilic groups, such as the thiol of cysteine and hydroxyl groups of serine or threonine.

Toxicity to larval and adult female mosquitoes
All prepared CDIAL derivatives were screened for 24 h larvicidal activity against 1 st instar Ae. aegypti (Liverpool, LVP, strain) using a concentration of 100 μM in the rearing water [21]. At this concentration, CDIAL killed ca. 70% of the larvae within 24 h (Fig 7A). Nearly all of the CDIAL derivatives showed significantly lower efficacy than CDIAL. However, compounds 10 (86.5%) and 13 (100%) exhibited statistically greater efficacy than CDIAL (Fig 7A).
The derivatives were also screened for 24 h toxicity against adult female Ae. aegypti using a dose of 1.5 nmol applied to the thoracic cuticle of each mosquito. At this dose, CDIAL incapacitated 68% of the mosquitoes within 24 h. All of the CDIAL derivatives were significantly less efficacious than CDIAL except for 10, which was similar to CDIAL in efficacy (74%).
Although the acid 5, the ester 6, and the 2,3-dihydro-3-pyridazinol 7 derivatives maintained the α,ß-unsaturated functionality of 1, they were inactive in the toxicity assays against larvae and adult females. The lack of toxicity of 7 can be due to its conversion to the 7-hydroxy-pyridazine 18, which has been observed in deuterated methanol at room temperature, and also may occur in the carrier solvent used for the bioassays or upon exposure to the rearing water. Since lactol derivatives 8, 9, and 12 were not toxic to larvae and adult females, we conclude that the drimane skeleton with C-11 hemiacetal polar head groups is insufficient to elicit mosquito toxicity. These results are consistent with published findings that the biological activity Fig 5. Chemical structures of α,ß-unsaturated 1,4-dialdehydes. Polygodial (3), 1ß-acetoxy-9-deoxy-isomuzigadial (17), miogadial (14), (+)-isovelleral (15). https://doi.org/10.1371/journal.pntd.0008073.g005

PLOS NEGLECTED TROPICAL DISEASES
Structural insights into the activities of drimane sesquiterpenes against Ae. aegypti of drimane-type compounds are reduced or lost when the aldehyde functionalities are modified [50,71,72]. Acetal protection of the C-12 aldehyde as in compound 11 also abolished the toxicity to larvae and adult females, thus illustrating the crucial role of the conjugated C-12 aldehyde in eliciting a biological response. On the other hand, compound 10, a C-12 methyl  The efficacy values were calculated using Abbott's correction to account for control (100% acetone) mortality [26]. In A), compounds were added to the rearing water of 1 st instar larvae (100 μM) and efficacy was defined as the percentage of larvae that died within 24 h. In B), compounds were applied to the thoracic cuticle of adult females (1.5 nmol/mosquito) and efficacy was defined as the percentage of adults that were incapacitated (dead or flightless) within 24 h. Values are means ± SEM based on at least 6 replicates of 5 larvae each or 3 replicates of 10 adult females each. The specific numbers of replicates for each are indicated in parentheses below the compound number. Shading indicates statistical categorization of a derivative's efficacy relative to CDIAL as determined by a one-way ANOVA with a Bonferroni post-test: gray = similar (P > 0.05); filled = superior (P < 0.05); open = inferior (P < 0.05). https://doi.org/10.1371/journal.pntd.0008073.g007

PLOS NEGLECTED TROPICAL DISEASES
Structural insights into the activities of drimane sesquiterpenes against Ae. aegypti ketone of 1, exhibited superior and similar efficacy to larvae and adult females, respectively, relative to 1. Thus, the more stable methyl ketone can effectively replace the aldehyde moiety. Remarkably, the dihydroquinone 13 exhibited superior larvicidal activity to 1 and was 100% effective in the screening experiments (Fig 7A). Additional experiments determined the full concentration-response curves of 1 and 13 to better compare their relative activities (see S4 Fig  in S1 File). The EC 50 of 13 (37.8 μM; 95% C.I. = 34.35-41.5 μM) was~2.2-times more potent than 1 (82.5 μM; 95% C.I. = 70.7-96.25 μM). Despite the superior larvicidal activity of 13, it was nominally toxic to adult females. We hypothesized that the chemical modifications resulting in the dihydroquinone preserved the toxicity of the molecule, but may have reduced its ability to penetrate the cuticle of adult females. Thus, we directly injected 13 into the hemolymph of adult females (1.5 nmol/mosquito) to bypass the cuticular barrier. However, even when injected, 13 was weakly toxic to adult females within 24 h (9.8 ± 7.6% efficacy) compared to CDIAL (90.0 ± 5.0% efficacy) (P < 0.0001; unpaired t-test; N = 7 replicates of 10 mosquitoes each). Thus, 13 may exploit a larval specific mechanism to induce toxicity or it is more readily detoxified/excreted in adult females vs. larvae. In summary, the relative activities of the analogs observed against larval and adult mosquitoes (Fig 7) indicate the important role of the 1,4-dialdehyde functional group in insecticidal activity.

Antifeedant activity in adult female mosquitoes [26]
A capillary feeding (CAFE) choice bioassay was used to screen the antifeedant activity of the CDIAL derivatives against adult female Ae. aegypti (LVP strain) [21,27,28]. Briefly, mosquitoes were presented with two capillaries of 10% sucrose as a food source for 18-20 h; the 'control' capillary was treated with 1% DMSO (the carrier solvent for 1 and its derivatives in this assay) while the 'treatment' capillary was treated with 1 or a derivative thereof at a concentration of 1 mM. With the exception of 8 and 11, mosquitoes consumed significantly less sucrose from the capillaries treated with the derivatives vs. the control capillaries (paired t-tests; P < 0.05), indicative of antifeedant activity. However, the antifeedant efficacy (i.e., antifeedant index) of each derivative was inferior compared to CDIAL (Fig 8).
We have previously shown that the antifeedant activity, but not insecticidal activity, of CDIAL was associated with its modulation of TRPA1 channels [21]. Thus, we suspect that all of the derivatives were inferior agonists of TRPA1 compared to CDIAL. The reduced TRPA1 agonistic activity may be attributable to the reactivity of the unsaturated system of the molecules. Specifically, the unsaturated aldehyde in 1 favors 1,2-addition, while 5, 6, and 7 likely favor 1,4-addition. Thus, 1 would be more amenable to direct attack of an amino group on TRPA1 than the other conjugated derivatives. The mosquito antifeedant activities of 5, 6, and 7 showed a similar trend to previously reported antifeedant activity of polygodial analogues [42,43]. As mentioned above, the possible conversion of 7 to 18 during the assay may also lead to the reduced activity. The absence of activity in 11 illustrates the necessity of the C-12 conjugated carbonyl group.
Although the derivatives were inferior to CDIAL in the context of antifeedant activity, bioactivity was not completely lost except for 8 and 11 and they still provide some insights into the SAR. For example, among the derivatives, 10 and 12, were relatively effective antifeedants. The moderate activity of compound 10 may indicate that the C-12 aldehyde can be converted into a methyl ketone without complete loss of antifeedant efficacy. For the lactol 12 it is possible that a ring opening tautomerization affords the C-11 aldehyde and a C-12 conjugated carbonyl.
Unsaturated aldehyde-containing agonists, such as acrolein, are known to react with nucleophilic residues of TRPA1 channels, including cysteine, histidine, and lysine. Several cysteine residues in TRPA1 channels have been identified as potential sites that can form covalent adducts with these agonists [37,38]. However, further mutagenesis experiments also showed that unsaturated dialdehyde-containing sesquiterpenes might not bind to the same sites as those targeted by the small agonists [66]. To shed light on how CDIAL interacts with the mosquito TRPA1 channel and thereby modulating its antifeedant effects, we have developed a structural model of mosquito TRPA1 based on the single-particle cryo-electron microscopy (EM) structure of human TRPA1 (hTRPA1) [32], and then computationally docked CDIAL to the AgTRPA1 structural model. We used the TRPA1 channel of Anopheles gambiae (AgTRPA1) as a representative mosquito TRPA1, because it has been previously cloned and shown to be directly activated by CDIAL [21,73]. Moreover, the amino acid identity between AgTRPA1 and the predicted Ae. aegypti TRPA1 (AAEL009419) is very high (>83%). In particular, the putative CDIAL binding region we have identified (see below) is over 90% identical between the two species.
Our results suggest that CDIAL binds preferentially to a 'pocket' near Cys684 (Fig 9), a residue in TRPA1 channels that has been implicated to interact with electrophilic agonists [38]. Notably, six lysine residues are located within a radius of 10 Å from Cys684, including Lys656, Lys678, Lys681, Lys728, Lys738 and Lys744. Many of these lysine residues are surrounded by negatively charged residues (Glu or Asp), thereby making them more nucleophilic and likely Mosquitoes were allowed to feed equally on two capillaries of 10% sucrose with 0.01% trypan blue; the control capillary included 1% DMSO, and the treatment capillary included 1% DMSO and 1mM test compound. The difference in volume consumed between the capillaries was used to calculate the antifeedant activity [21]. Values are means ± SEM based on at least 8 replicates of 5 adult females each. The specific numbers of replicates for each are indicated in parentheses below the compound number. Shading indicates statistical categorization of derivative's efficacy relative to CDIAL as in Fig 7. https://doi.org/10.1371/journal.pntd.0008073.g008

PLOS NEGLECTED TROPICAL DISEASES
to interact with an electrophile like CDIAL. The dominant predicted CDIAL binding site is located between Cys684 and Lys728 (Fig 9). Analogous to Lys661 in human TRPA1, Lys728 is the closest lysine to Cys684. In these docked poses (Fig 9), the C-12 aldehyde of CDIAL is found close (4.2-5.0 Å) to the amino group of Lys728 that can initiate a nucleophilic attack on C-12 to form an initial CDIAL-AgTRPA1 conjugate, highlighting the importance of the PLOS NEGLECTED TROPICAL DISEASES reactive C-12 aldehyde group. This is consistent with the observation that acetal protection of the C-12 aldehyde in compound 11 abolished its antifeedant activity (Fig 8). Further, given the proximity (~5.0 Å) of the thiol group of Cys684, the polarized C-12 or C-7 of CDIAL may subsequently be attacked by this thiol. Overall, the docking results show a few possible ways in which CDIAL can interact with AgTRPA1, suggesting that the reactions of the α,ß-unsaturated 1,4-dialdehyde with nearby amine and thiol side chains (consistent with the formation of thiazolidine when CDIAL was treated with L-cysteine methyl ester in pyridine proposed in Fig 6) may be responsible for activating the AgTRPA1 channel. Overall, these preliminary computational results provide a molecular rationale for the importance of the dialdehyde moiety and the α,ß-unsaturated carbonyl for CDIAL's antifeedant activity. Nonetheless, due to the high flexibility of the loops shaping the CDIAL binding pocket, our computational studies do not exclude the possibility that other Lys residues such as Lys681 and Lys744 may also be involved in reacting with 1 after some local conformational rearrangements.

Conclusion
The current study has analyzed ten semisynthetic analogs (5-13, and 16) of the α,ß-unsaturated 1,4-dialdehyde cinnamodial (1) to identify the structural contributions of 1 toward its larvicidal, adulticidal and antifeedant activity against Ae. aegypti. Two analogs, 10 and 13 exhibited more efficacious toxicity against mosquito larvae than 1. Moreover, 10 was of similar toxic efficacy against adult females as 1. These results indicate that the reactive C-12 aldehyde can be substituted with relatively stable moieties, such as an unsaturated methyl ketone 10 or hydroquinone 13, without sacrificing insecticidal activity. All of the other CDIAL analogs showed that the bicyclic drimane skeleton alone was not sufficient to induce larval or adult toxicity.
Notably, all of the analogs possessed weaker antifeedant activity than 1 regardless of their insecticidal activity. These results support the notion that CDIAL's insecticidal and antifeedant mechanisms of action are independent [21]. While our previous results associated CDIAL's antifeedant activity with its modulation of TRPA1 [21], this work suggests that CDIAL may interact with TRPA1 by an initial nucleophilic attack at C-12 by a lysine residue to form a CDIAL-TRPA1 conjugate. The activated conjugate may then undergo further attack at the C-12 or C-7 of CDIAL by a neighboring thiol or nucleophilic residue [57].
Despite the thorough elucidation of the antifeedant structure-activity relationship against lepidopterans [45,51,56,74,75] and the toxicant activity against insect pests [72,74,[76][77][78] by several groups, this is the first investigation into the observed structure-activity relationship (SAR) of 1 against mosquitoes (Figs 7-9). Overall, we have identified several CDIAL derivatives that confirm the importance of the dialdehyde moiety for mosquitocidal activity and the α,ß-unsaturated carbonyl for antifeedant activity. The improved larvicidal activity observed in 10 and 13 suggest that the development of more stable and effective insecticidal CDIAL derivatives is possible. Future studies will focus on developing a comprehensive screening library of CDIAL derivatives and implementation of quantitative SAR to fully elucidate the insecticidal and antifeedant SARs of CDIAL, optimize its bioactivities, and identify its active pharmacophores. The latter may enable the synthesis of relatively simple synthetic molecules with similar or improved insecticidal activity compared to CDIAL, as has been recently accomplished for the spinosyns [79,80].