Differential transcription profiles in <em>Aedes aegypti</em> detoxification genes following temephos selection

Introduction

Dengue is the most prevalent mosquito-borne viral disease affecting humans globally, with the mosquito vector Aedes aegypti (L) found in nearly 100 tropical countries. Due to lack of vaccines or effective pharmaceutical treatments, dengue prevention is currently predicated on reduction of both larval and adult Ae. aegypti populations (Gubler, 2004). Adult control relies largely on formulations of pyrethroid insecticides. For larval control, the three most widely used compounds are Bacillus thuringiensis israelensis (Bti), methoprene, and temephos. Globally, temephos is the most used of these three due to its very low vertebrate toxicity and relatively low cost (WHO, 2009). Temephos is one of a few organophosphates (OP) registered to control Ae. aegypti larvae and it is an important management tool for mosquito abatement programs (EPA, US 2001).

Temephos was used for 30 years before initial reports of resistance appeared in 1995. Resistance ratios (RR) of two to ten were found in collections of Ae. aegypti from Venezuela (Mazzarri and Georghiou, 1995) and 17 Caribean countries (Rawlins and Wan, 1995). Since 2000, temephos resistance has been reported from Cuba and Venezuela (Rodriguez et al. 2001; Rodriguez et al. 2002), Thailand (Jirakanjanakit et al. 2007) and Brazil (Macoris et al. 2003; Braga et al. 2004; Lima et al. 2003, 2006, 2011; Beserra et al. 2007). Most recently reports have appeared from El Salvador (Lazcano et al. 2009), Martinique Island in the French West Indies (Marcombe et al. 2009), Argentina (Llinas et al. 2010; Seccacini et al. 2008), India (Tikar et al. 2009), Colombia (Ocampo et al. 2011), and Trinidad (Polson et al. 2010; 2011). Although resistance to temephos has been demonstrated in many areas of the world, it is the only remaining organophosphate larvicide with any appreciable use. As such, it is an important tool in managing resistance to the few alternative available larvicides.

Mechanisms of temephos resistance have been identified using chiefly bioassays with synergists, biochemical assays and most recently with microarrays. Two major mechanisms of OP resistance reported in mosquitoes involve target site mutations at the acetyl cholinesterase (AChE) and increased detoxification performed by three enzymatic systems: cytochrome P₄₅₀ monooxygenases (CYP), glutathione-S-transferases (GST) and carboxyl/cholinesterase esterases (CCE). A common pattern of increased esterase activity, increased mortality with the esterase inhibitor DEF (S.S.S-tributlyphosphorotrithioate) and no evidence of insensitive AChE have by now been reported in multiple studies involving Ae. aegypti (Wirth and Georghiou 1999; Macoris et al. 2003; Lazcano et al. 2009; Montella et al. 2007; Rodriguez et al. 2007; Sousa-Polezzi and Bicudo 2004; Bisset et al. 2011; Melo-Santos, et al. 2010). However, some other studies also found differences in CYP and GST systems among resistant and susceptible populations (Braga et al. 2005; Melo-Santos et al. 2010; Ocampo et al. 2011; Polson et al. 2011; Rodriguez et al. 2001).

In the present study, we analyzed the response to laboratory temephos selection in five mosquito strains from México and in a strain from Iquitos, Perú. We monitored for changes in the lethal concentrations (LC₅₀) and in the transcription profiles of the putative detoxification genes on the ‘Aedes Detox Chip’ DNA microarray v.2 (Strode et al. 2008). This microarray contains 318 70-mer probes representing 290 detoxification genes including 183 CYPs, 28 GSTs, 44 CCEs including nonspecific esterases, carboxyl/cholinesterase esterases, p- nitrophenyl acetate esterases, and acetylcholinesterases (AChE) and 35 additional enzymes potentially involved in response to oxidative stress in Ae. aegypti (RedOxs). This microarray has been widely used to follow changes in the expression of detoxification genes. Boyer et al. (2006) used the detox microarray to analyze the ability of Ae. aegypti larvae to tolerate temephos, Bti and toxic vegetable leaf litter. Both induction and selection were correlated with levels of larval detoxifying enzyme activities. Poupardin et al. (2008) tested the effect of exposure of Ae. aegypti larvae to sub-lethal doses of temephos on their subsequent tolerance to insecticides, detoxification enzyme activities and expression of detoxification genes. Overall, this study revealed the potential of xenobiotics found in polluted breeding sites to affect their tolerance to insecticides, possibly through the cross-induction of particular detoxification genes. Marcombe et al. (2009) investigated the molecular basis of insecticide resistance in Ae. aegypti collected in Martinique and found significantly elevated transcription of CYP, GST and CCE at both larval and adult stages. More recently, it was used to compare gene transcription in a temephos selected strain from Brazil and a decrease in expression of some of these genes following removal of temephos selection (Strode et al. 2012).

Recently we compared the gene expression profiles in six strains of Ae. aegypti during and after permethrin selection in adults (Saavedra-Rodriguez et al. 2012). Results indicated that many different genes respond to selection but consistency among strains was uncommon, even from geographically proximate strains. In the present study we report the response of this same collections following temephos selection. We compare gene expression at four levels with the ‘Aedes Detox Chip’ microarray. First, transcription patterns in the unselected F_S0 strains were measured relative to New Orleans to identify patterns of differential expression already present in the field. Second, transcription patterns were compared between strains following one generation of selection (F_S1) and F_S0 strains to identify genes that respond rapidly to selection. Third, transcription was compared between F_S1 and F_S5 strains to identify genes that respond to five generations of selection. Fourth, transcription following five generations of selection was compared among all strains relative to New Orleans (F_S5 versus NO) to identify genes commonly upregulated in all strains. Finally we performed cross resistance bioassays and compared temephos and permethrin expression profiles. Lack of cross resistance was evidenced by bioassays and different expression profiles for each insecticide selection experiment.

Results

Bioassays

Six Ae. aegypti lines collected from southern México and one collected from Iquitos, Perú (Table 1) were used in all experiments. Further each of these was divided into three replicate cages prior to selection and this replication was used to test for significance using the Limma analysis package in www.bioconductor.org. The New Orleans (NO) strain was used as a susceptible control.

Table 1

Collection sites and temephos resistance before selection. LC₅₀ in μg temephos/mL water. The LC₅₀ resistance ratio was calculated relative the susceptible New Orleans (NO) strain.

Country State	City Site	Coordinates (DDD.dddd)	Generation in which P0 was started	LC50 (FS0) ug ai/mL	95% confidence intervals	RR (FS0) P1/NO	RR (FS5) FS5/NO
Perú
	Iquitos		F2	0.029	(0.0239–0.0355)	5.8	129.0
México
Quintana Roo	Chetumal
	Calderitas	18.5563°–88.2562°	F3	0.077	(0.0583–0.1007)	15.3	310.9
	Lagunitas	18.5199°–88.3339°	F3	0.237	(0.1996–0.2974)	47.5	251.4
	Lázaro-Cárdenas	18.5359°–88.3016°	F3	0.161	(0.1147–0.2253)	32.2	389.9
	Solidaridad	18.5284°–88.3027°	F3	0.027	(0.0207–0.0363)	5.5	205.0
Yucatán	Mérida	20.9489°–89.6405°	F2	0.026	(0.0215–0.0302)	5.1	42.2
United States	New Orleans			0.005	(0.0033–0.0074)

Offspring from field collections that had been in the laboratory for no more than three generations were designated F_S0 (no prior selection). LC₅₀ values were measured in each of the unselected F_S0 lines (Table 1) using the CDC beaker bioassay http://www.cdc.gov/ncidod/wbt/resistance/assay/larval/step_2.htm. Moderate levels of resistance (RR₅₀ > 5 –10) were detected in Iquitos, Solidaridad, and Mérida F_S0 lines. High (RR₅₀ >10) levels of resistance were detected in the F_S0 of Calderitas, Lagunitas, and Lázaro-Cárdenas.

The general response to selection was an increase in LC₅₀ in all mosquito lines (Fig. 1). Table 2 lists the realized heritability (h^{) coefficients for LC}₅₀ during the selection process and the RR LC₅₀ after five generations of selection. Supplement 1 shows the numbers of selected larvae in each biological replicate at each generation of selection. Out of 34 experiments, six showed significant difference among biological replicates, including Iquitos F_S0 and F_S3, Lagunitas F_S4 and F_S5, Mérida F_S4 and Solidaridad F_S4. Each strain exhibited a distinct response pattern. Iquitos and Mérida responded very gradually to selection and had the smallest h ^{for LC}₅₀ suggesting that they had the least additive genetic variance in alleles conditioning resistance. The greatest response to selection was seen in Lázaro-Cárdenas and Calderitas and these had the largest h ^{for LC}₅₀ suggesting that they had the largest amount of additive genetic variance. However, the increase was non-linear in both strains. Calderitas exhibited a large response to selection between generations 3 and 4 as did Lázaro-Cárdenas between generations 4 and 5. This may reflect the presence of recessive alleles that condition resistance to temephos but which were initially too low in frequency to breed recessive homozygotes. Solidaridad and Lagunitas evolved resistance at an intermediate rate and had intermediate h ^{for LC}₅₀. The LC₅₀ did not constantly increase in all strains. There were marked declines in LC₅₀ in Calderitas and Lagunitas. These may have been attributable to lethal or deleterious recessive alleles that eventually increased sufficiently in frequency during the selection process to breed homozygotes. Using the experiment-wise error rate of α= 0.05, the h ^{for LC}₅₀ was only significant in the Solidaridad and Mérida collections. Lack of significance in the other strains was largely due to the non-linear effects and a large variance among the three replicates within each generation.

Figure 1

LC₅₀ response of mosquito lines to temephos selection over five generations. Regression analysis and significance is shown for each mosquito line.

Table 2

Realized heritability for temephos LC₅₀ in six field selected strains of Aedes aegypti.

Strain	Realized h2	p
Iquitos	0.24	0.073
Calderitas	2.11	0.059
Lagunitas	0.55	0.225
Lázaro-Cárdenas	1.02	0.064
Solidaridad	0.82	0.018
Mérida	0.76	0.031

Microarray validation

Expression ratios were calculated as M, the log₂ of mean transcription ratios, where M = log₂ (Cy5/Cy3), Cy5 is the adsorption at 649 nm and Cy3 is the adsorption at 532 nm. The expression ratios as measured by both microarray and quantitative-PCR were compared in twelve genes (GSTs1, CCae4C, COE-8, CYP325G3, CYP4H28, CYP4J13, CYP6Nae1, CYP9J22, CYP9J32, AAEL004388, AAEL004390 and AAEL010382). The same amplified RNA samples from 39 RNA collections were compared and a linear regression model explained 78.5% of the variance, had a slope significantly greater than zero (P<0.0001) that approximated one and an intercept not significantly less than zero (P=0.9851) (Fig. 2).

Figure 2

Correlation between microarray and real time expression ratios. Ratios are display in a log₂ scale.

Cross resistance

The LC₅₀ and RR with respect to NO for temephos was calculated for all F_S0 and F_S5 mosquitoes using NO as a standard susceptible control (Table 1). Our previous study (Saavedra-Rodriguez et al. 2012) calculated the LC₅₀ and RR for permethrin in the F_S0 and permethrin selected F_S5 mosquitoes. Figure 3 plots the RR of the LC₅₀ with temephos along the abscissa and the RR of the LC₅₀ with permethrin along the ordinate axis. Points in the circle close to the origin represent resistance in F_S0 mosquitoes. Points within the oval oriented along the abscissa represent resistance in the temephos selected F_S5 mosquitoes ranged over an order of magnitude from ~40 for the Mérida strain to ~400 for the Lázaro-Cárdenas strain. Points within the vertical oval oriented along the ordinate represent resistance in the permethrin selected F_S5 mosquitoes. These varied five-fold from ~20 for the Calderitas strain up to ~100 for the Lázaro-Cárdenas strain. Cross resistance would be manifested as points located between the two ovals and would appear somewhere in the center of this graph but instead all points are near the abscissa or ordinate.

Figure 3

Resistance ratio (RR) of the LC₅₀ with temephos along the abscissa and the RR of the LC₅₀ with permethrin along the ordinate axis. Points in the circle close to the origin represent resistance in F_S0 mosquitoes. Points within the oval along the abscissa represent resistance in the temephos selected F_S5 mosquitoes from (Saavedra-Rodriguez et al. 2012). Points within the vertical oval oriented along the ordinate all represent resistance in the permethrin selected F_S5 mosquitoes.

All of the genes in Table 6 were compared with all of the genes with significant differential expression between the F_S5 vs. NO adults used in the permethrin experiment (Saavedra-Rodriguez et al. 2012) to identify genes with differential expression in both permethrin and temephos experiments. Of the 85 differentially transcribed genes in the present experiment (Table 6) and the 70 differentially transcribed genes in the permethrin experiment, there were 13 genes with differential expression in both experiments but only in four of the collections. M values for these genes are plotted in Figure 4. Eight of the 13 were regulated in the same direction in both experiments. These were: 1) GSTs1-1 which was downregulated in both Mérida and Iquitos, 2) GSTd1-1 and 3) CYP9J32 which were upregulated in Iquitos 4) CYP6Z9, 5) Catalase and 6) COE-17 which were downregulated in Iquitos, 7) CYP6BB2 which was upregulated in Lázaro-Cárdenas in both experiments and 8) CYP9J22 which was upregulated in Mérida. However, in each of these 8 cases the trend is inconsistent. For example Figure 4 shows that GSTs1-1 in Lázaro-Cárdenas was upregulated by temephos selection but was downregulated with permethrin. CYP9J32 was upregulated with permethrin selection but unchanged by temephos selection in Lázaro-Cárdenas. Conversely CYP9J32 was upregulated with temephos selection but unchanged by permethrin selection in Calderitas. CYP9J22 was upregulated with permethrin selection but downregulated by temephos selection in Lázaro-Cárdenas. In general, there was little or no consistent evidence of cross selection for the majority of the differentially transcribed genes in both experiments.

Figure 4

The expression ratios (M) of genes with differential expression in both permethrin (Saavedra-Rodriguez et al. 2012) and the current temephos selection experiments. There are thirteen genes with differential expression in both experiments but only in four of the collections. Genes represent in more than one collection appear in boxes. Boxes were shaded in grade to assist in identification of the same gene.

Bioassays using inhibitors

The F₉ generation from three of the temephos selected strains (Solidaridad F₉, Calderitas F₉ and Mérida F₉) that were released from selection after F_S5 were bioassayed using inhibitors for CCE, CYP and GST systems. Releasing strains from temephos pressure resulted in an immediate reduction in temephos resistance for all tested strains. Mérida F_S5 was 45 fold more resistant than NO while Mérida F₉ was only 13 times more resistant than NO, resulting in a 32 fold resistance reduction. Calderitas F_S5 was 105 fold more resistant than NO while Calderitas F₉ was only 40 fold more resistant. Finally, Solidaridad F_S5 to F₉ resulted in a resistance reduction from 78 to 60 fold relative to NO.

We compared the LC₅₀ values for temephos alone and for temephos + inhibitor in each of the resistant strains (Table 7). Figure 5 shows the LC₅₀ obtained from Solidaridad F₉, Calderitas F₉ and Mérida F₉ strains using different inhibitor treatments. LC₅₀ decreased dramatically when using the temephos + DEF treatment in Solidaridad F₉, Calderitas F₉ and Mérida F₉ (497, 76 and 48 fold respectively, relative to NO). CCE inhibition resulted in mortality within 45 minutes of exposure in Solidaridad F₉ and Calderitas F₉. In comparison, temephos alone caused mortality only after 2 hours of exposure. With the GST inhibitor (DEM), LC₅₀ for Calderitas F₉ and Mérida F₉ decreased slightly (0.7 and 0.6 fold) and for Solidaridad F₉ the LC₅₀ was 1.14 fold higher. Applying the CYP inhibitor (PBO) LC₅₀ values increased 5.9, 1.8, and 2.1 fold in Solidaridad F₉, Calderitas F₉ and Mérida F₉, respectively.

Figure 5

Effect of inhibitors of esterases (DEF), cytochrome P₄₅₀ oxidases (PBO) and glutathione transferases (DEM) on temephos LC₅₀.

Table 7

Temephos LC₅₀ (μg/ml) among larvae treated with a dose of temephos as compared with larvae pre-treated with one of three inhibitors followed by a dose of temephos. Synergism ratio (SRLC₅₀) corresponds to temephos LC₅₀ divided by the inhibitor-treatment LC_50.

Treatment	Inhibitor	Target Enzymes		Solidaridad F9	Calderitas F9	Mérida F9	New Orleans
Temephos	None			0.1989 (0.1457–0.2715)	0.1520 (0.1214–0.1901)	0.0434 (0.0313–0.0602)	0.0033 (0.0026–0.0041)
Temephos	DEM 150 μg/ml	GST		0.2276 (0.1619–0.3199)	0.1065 (0.0791–0.1432)	0.0271 (0.0194–0.0378)
			SRLC50	0.8739	1.4272	1.601
Temephos	PBO 5 μg/ml	CYP		1.1754 (0.3769–3.664)	0.2773 (0.1965–0.3913)	0.0936 (0.0608–0.1440)
			SRLC50	0.1692	0.5481	0.4636
Temephos	DEF 1.5 μg/ml	CCE		0.0004 (0.0003–0.0007)	0.0020 (0.0014–0.0027)	0.0009 (0.0007–0.0011)
			SRLC50	497.25	76.0	48.22

Bioassays

Six Ae. aegypti lines collected from southern México and one collected from Iquitos, Perú (Table 1) were used in all experiments. Further each of these was divided into three replicate cages prior to selection and this replication was used to test for significance using the Limma analysis package in www.bioconductor.org. The New Orleans (NO) strain was used as a susceptible control.

Offspring from field collections that had been in the laboratory for no more than three generations were designated F_S0 (no prior selection). LC₅₀ values were measured in each of the unselected F_S0 lines (Table 1) using the CDC beaker bioassay http://www.cdc.gov/ncidod/wbt/resistance/assay/larval/step_2.htm. Moderate levels of resistance (RR₅₀ > 5 –10) were detected in Iquitos, Solidaridad, and Mérida F_S0 lines. High (RR₅₀ >10) levels of resistance were detected in the F_S0 of Calderitas, Lagunitas, and Lázaro-Cárdenas.

The general response to selection was an increase in LC₅₀ in all mosquito lines (Fig. 1). Table 2 lists the realized heritability (h^{) coefficients for LC}₅₀ during the selection process and the RR LC₅₀ after five generations of selection. Supplement 1 shows the numbers of selected larvae in each biological replicate at each generation of selection. Out of 34 experiments, six showed significant difference among biological replicates, including Iquitos F_S0 and F_S3, Lagunitas F_S4 and F_S5, Mérida F_S4 and Solidaridad F_S4. Each strain exhibited a distinct response pattern. Iquitos and Mérida responded very gradually to selection and had the smallest h ^{for LC}₅₀ suggesting that they had the least additive genetic variance in alleles conditioning resistance. The greatest response to selection was seen in Lázaro-Cárdenas and Calderitas and these had the largest h ^{for LC}₅₀ suggesting that they had the largest amount of additive genetic variance. However, the increase was non-linear in both strains. Calderitas exhibited a large response to selection between generations 3 and 4 as did Lázaro-Cárdenas between generations 4 and 5. This may reflect the presence of recessive alleles that condition resistance to temephos but which were initially too low in frequency to breed recessive homozygotes. Solidaridad and Lagunitas evolved resistance at an intermediate rate and had intermediate h ^{for LC}₅₀. The LC₅₀ did not constantly increase in all strains. There were marked declines in LC₅₀ in Calderitas and Lagunitas. These may have been attributable to lethal or deleterious recessive alleles that eventually increased sufficiently in frequency during the selection process to breed homozygotes. Using the experiment-wise error rate of α= 0.05, the h ^{for LC}₅₀ was only significant in the Solidaridad and Mérida collections. Lack of significance in the other strains was largely due to the non-linear effects and a large variance among the three replicates within each generation.

Figure 1

LC₅₀ response of mosquito lines to temephos selection over five generations. Regression analysis and significance is shown for each mosquito line.

Microarray validation

Expression ratios were calculated as M, the log₂ of mean transcription ratios, where M = log₂ (Cy5/Cy3), Cy5 is the adsorption at 649 nm and Cy3 is the adsorption at 532 nm. The expression ratios as measured by both microarray and quantitative-PCR were compared in twelve genes (GSTs1, CCae4C, COE-8, CYP325G3, CYP4H28, CYP4J13, CYP6Nae1, CYP9J22, CYP9J32, AAEL004388, AAEL004390 and AAEL010382). The same amplified RNA samples from 39 RNA collections were compared and a linear regression model explained 78.5% of the variance, had a slope significantly greater than zero (P<0.0001) that approximated one and an intercept not significantly less than zero (P=0.9851) (Fig. 2).

Figure 2

Correlation between microarray and real time expression ratios. Ratios are display in a log₂ scale.

Gene expression in unselected lines relative to New Orleans

Transcription patterns in five F_S0 strains (RNA preparation failed in Solidaridad F_S0) were measured relative to NO to identify any patterns of differential expression already present when they were collected in the field. A total of 39 genes were differentially expressed in two or more strains (Table 3). This included eight GSTs, fourteen CYPs, ten CCEs, and seven RedOx genes. Most notably, GSTe3 was 2–9 fold upregulated in all five field collections and GSTe7 was 2–4 fold upregulated in four collections. Among the CYPs, CYP9J26 and CYP9J32 were at least two fold upregulated in four strains. Many CCEs were differentially expressed but none had a consistent pattern of upregulation. Instead CCEs were significantly down regulated in the Iquitos strain and CCae4C was significantly down regulated in four strains. No trends were noted among RedOx genes. The numbers of genes differentially expressed varied among strains. From 15–21 genes were differentially expressed in Calderitas and Iquitos as compared with 46 in Mérida. From 30–40% of those differentially expressed genes were down regulated relative to NO in the Mexican collections as compared to 71% that were down-regulated in Iquitos.

Genes that respond to a single generation of selection

Transcription patterns were next compared between F_S1 and F_S0 strains to identify genes that responded to one generation of selection (Table 4). Fifteen genes responded in two or more strains. These included one sigma class GST, three CYPs, one each in the CYP4, CYP6, CYP9 families, three carboxyl-, four choline- and one juvenile hormone esterase. A heme- and a glutathione peroxidase and a superoxide dismutase were among the RedOx genes. From 27–29 genes were differentially expressed in Calderitas and Iquitos as compared with only two in Lagunitas and Mérida. From 56–62% of these genes in Calderitas and Iquitos were up-regulated as compared to 40% that were up-regulated in the other strains.

Genes that respond to four additional generations of selection

Transcription patterns were next compared between F_S1 and F_S5 generations to identify genes that respond to long term selection (Table 5). Sixteen genes responded in two or more strains. This included two GSTs, eight CYPs, one carboxyl-, one choline- and one juvenile hormone esterase. A heme- and a thioredoxin peroxidase and an aldehyde oxidase were among the RedOx genes. Most notably CYP4H28 was differentially expressed in all six strains but was down regulated in Iquitos, Lázaro-Cárdenas and Lagunitas and upregulated in Calderitas, Solidaridad and Mérida. Aldehyde oxidase 10382 was down regulated in four strains. The numbers of genes that responded to long term selection varied among strains. From 8–11 genes were differentially expressed in Iquitos, Calderitas, Lagunitas and Mérida. But 19 were differentially expressed in in Solidaridad and 23 in Lázaro-Cárdenas. In Iquitos, only four of the eight genes were upregulated and these were all epsilon GSTs.