Molecular Characterization of Ae. aegypti FAS genes
Seven putative AaFAS genes models were obtained via manual annotation using the AaegL5 assembly (35). Previously, five candidate FAS genes (AaFAS1-5), were identified based on the AaegL3 assembly of Nene et al., 2007 (23, 36), and of these, only AaFAS1 and 2 have undergone functional studies (12). The AaegL5 assembly enabled identification of two additional candidate FAS genes (AaFAS6 and AaFAS-like). The corresponding mRNA sequences showing predicted intron/exon structure and initiation and stop codons are shown in Supplemental File 1. The AaFAS1 gene model revealed a gene structure comprising 11 exons, while AaFAS2 had 5 exons and AaFAS3-5 had 6 exons (Fig. 1). The incomplete AaFAS-like and AaFAS6 gene models comprised 2 and 3 exons, respectively.
The gene models for AaFAS1-5 appear to be full length, with an average gene product length of 2,360 amino acids (Table 1). AaFAS1-5 possessed features associated with functional FAS, including an initiation methionine, a stop codon and the functional catalytic motifs (DTACSS, EAH and GSVKS) important for ketoacyl synthesis as described by Beedessee et al. 2015 (26). Additionally, AaFAS1-5 contained the YKELRLRGY motif conserved among the FAS genes of vertebrates and invertebrates, present in the polyketide synthase deshydratase domain (Fig. S1). AaFAS3 lacked 6 amino acid residues in the 3’ terminus of exon 6 and a total of 127 non-synonymous substitutions were identified in this model as compared to its AaegL3 counterpart.
Table 1. Summary of AaFAS gene family predicted from the Aedes aegypti AaegL5 assembly. The AaegL5 annotation is shown in comparison to the AaegL3 gene models reported by Nene et al., 2007 [23].
Name
|
NCBI Accession Number (AaegL5)
|
VectorBase Accession number (AaegL3)
|
Max number of exons
|
Chro-mo-some
|
Location
|
% Ident with human FAS
|
Length (nucleotides)
|
Length (amino acids)
|
No. Splice variants/isoforms
|
Notes on the revised annotation
|
% identity (AaegL3 & L5)
|
AaFAS1
|
LOC5568814
|
AAEL001194
|
11
|
2
|
NC_035108.1 (307544012..307601765, complement)
|
45.3
|
9732
|
2422
|
1
|
1 SNP in exon_4
|
99.9
|
AaFAS2
|
LOC5570229
|
AAEL008160
|
5
|
3
|
NC_035109.1 (9993811..10002427, complement)
|
36.7
|
8368
|
2385
|
1
|
1 SNP in exon_5
|
100
|
AaFAS3
|
LOC5573929
|
AAEL022506
|
6
|
2
|
NC_035108.1 (429256663..429264110, complement)
|
32.9
|
7144
|
2334
|
1
|
18 SNPs exon_1, 31 SNPs exon_2, 11 SNPs exon_3, 52 SNPs exon_4, 3 SNPs exon_5, and 12 SNPs exon_6. Deletion of six amino acid residues when compared with the AaegL3 orthologue
|
94.6
|
AaFAS4
|
LOC5573931
|
AAEL002237
|
6
|
2
|
NC_035108.1 (429327062..429334421, complement)
|
34.6
|
7068
|
2333
|
1
|
4 SNPs exon_2, 10 SNPs exon_4, 6 SNPs exon_6
|
99.1
|
AaFAS5
|
LOC5573927
|
AAEL002228
|
6
|
2
|
NC_035108.1 (429228612..429236010, complement)
|
32.9
|
7097
|
2324
|
1
|
5 SNPs exon_2, 2 SNPs exon_3, 3 SNPs exon_4, 1 SNP exon_6
|
99.5
|
AaFAS-like
|
LOC110675236
|
-
|
2
|
2
|
NC_035108.1 (429280401..429282870, complement)
|
36.1
|
2409
|
800
|
1
|
-
|
-
|
AaFAS6
|
LOC5573930
|
-
|
3
|
2
|
NC_035108.1 (429275401..429279876, complement)
|
36.8
|
4353
|
1386
|
1
|
-
|
-
|
The Bayesian inference supported AaFAS1-5 as paralogues, and revealed highest percent amino acid similarity between AaFAS1 and the H. sapiens FAS (human FAS) (Fig. 2). Notably, AaFAS1 clustered in a clade comprising the H. sapiens, Mus musculus, Apis mellifera FAS, the D. melanogaster FAS1 and 2, and an uncharacterized Anopheles gambie FAS (Fig. 2). Similarly, AaFAS2 clustered in a clade with another uncharacterized Anopheles gambiae FAS. In contrast, AaFAS3, 4, 5, 6 and -like clustered at the most branched portion of the tree, suggesting a recent diversification event. Phylogenetic analyses and amino acid alignment supported AaFAS1-5 as the counterparts of the AaegL3 genome assembly-derived gene models as follows: LOC5568814-AAEL001194; LOC5570229-AAEL008160; LOC5573929-AAEL022506; LOC5573931-AAEL002237 & LOC5573927-AAEL002228 (Fig. 2, Table 1). AaFAS-like and AaFAS6 (LOC110675236 & LOC5573930) were not identified in the AaegL3 assembly suggesting these models are unique to the AaegL5 assembly.
To investigate putative functional domains, AaFAS sequences were aligned to the human FAS using Clustal Omega (27). Human FAS contains seven catalytic domains and three noncatalytic domains (1). Collectively, AaFAS posssesed less than 50% amino acid identity to human FAS, and of the seven gene models, AaFAS1 had the highest amino acid identity (45.3%) (Table 1 and Table S5). Alignment of FAS domains also showed modest sequence identity between human FAS and AaFAS (23.03-63.56%) with greatest similarity for AaFAS1 domains (Table S5). The linear organization of mammalian FAS domains annotated by Maier et. al., 2008, is shown in Fig. 3 (1). Conservation in linear organization of motifs associated with known functional domains identified using Pfam 31.0 software is shown in Fig. 3B. Dotted lines between Fig. 3A and B compare mammalian FAS domains (Fig. 3A) and AaFAS domains (Fig. 3B). Pfam analysis did not show the presence of functional methyltransferase domains in AaFAS (Fig. 3B) and protein sequence alignment using Clustal Omega showed deletion within pseudo-methyltransferase (ΨME) domains of AaFAS compared to human FAS (16.20-23.03% identity; Table S5 and Fig. S2).
The gene model of AaFAS-like was 800 amino acids in length and contained all functional catalytic motifs, whereas AaFAS6 was 1,386 amino acids in length, and lacked catalytic motifs but contained the conserved 3’ motif YKELRLRGY conserved in FAS (Fig. S1). AaFAS-like contains ketoacyl synthase, ketoacyl synthase_C and ketoacyl-synthase C-terminal extension domains, the first 5’ domains of AaFAS1-5 and human FAS (Fig. 3B), while AaFAS6 contains ADH zinc, β-ketoreductase, PP binding and thioesterase domains, the last four domains located 3’ in AaFAS1-5 and human FAS (Fig. 3B). In the AaegL5 assembly, AaFAS-like and 6 are located on chromosome 2 at positions 429280401-429282870 and 429275401-429279876, respectively. It is possible that AaFAS-like and -6 reflect an error in genome assembly, or a gene duplication. However, molecular data and the inability to detect transcripts associated with either locus, suggest (Fig. S3) that AaFAS-like and 6 represent pseudogenes (Fig. 3B).
FAS expression during Ae. aegypti development
Mosquitoes undergo four developmental stages: egg, larva, pupa and adult. RT-PCR was used to explore the hypothesis that expression patterns of AaFAS genes vary among these stages. Five individual mosquitoes were collected for each of the 4th larval instar, pupa and adult stages. Expression of each of the AaFAS genes was normalized to the average expression of the β-actin gene (2−ΔCt) (Fig. 4).
Relative expression analyses revealed negligible AaFAS expression in larval and pupal stages, while the highest expression of all genes except AaFAS4 were observed in adult males (Fig. 4). AaFAS1 was the most predominant FAS expressed in adult mosquitoes. Differences in expression levels of any AaFAS were not observed between sugar-fed and 3 days post blood-fed (coinciding with the first gonotrophic cycle) females. The study also revealed negligible AaFAS4 expression in all developmental stages and sexes (Fig. 4).
Impact of blood feeding on expression of AaFAS1
The diet of the female Ae. aegypti typically involves both nectar and blood. The blood meal is rich in proteins and lipids; therefore, this diet may trigger lipolysis, instead of synthesis, to break down lipid molecules. We compared AaFAS1 expression, the predominant AaFAS in adult females, in sugar-fed females versus blood-fed females (feeding once or twice) (Fig. 5). Blood meals were provided only on specific days as shown in Fig. 5A, while mosquitoes from all groups were fed ad lib on 10% sugar diet throughout the experiment. Comparisons of AaFAS1 gene expression from mosquito samples collected on the same day showed no differences among feeding conditions (Fig. 5B). However, when profiled as ratios (Fig. 5C), we observed a slight, but not significant, reduction of AaFAS1 expression in females given a single blood-meal as compared to sugar-fed females on days 1, 3 and 4 post-blood meal (pbm) (Fig. 5B: F vs. B, G vs. C and H vs. D). This data suggests that diet may only play a minor role, if any, in the expression of AaFAS1 gene.
Transient knockdown of AaFAS1 gene causing upregulation of other AaFAS genes
We hypothesized that the redundancy of AaFAS genes may serve as a backup system for the mosquitoes. To test this hypothesis, we employed AaFAS1 loss-of-function studies to investigate the possibility of compensation by other AaFAS genes. Female mosquitoes were IT injected with dsRNA derived from AaFAS1 or GFP (KD control). On day 2 post-dsRNA injection, five mosquitoes were collected for assessment of AaFAS expression (Fig. 6). We observed an approximate 40% reduction in AaFAS1 expression (~39.3±13.9%) in AaFAS1-KD mosquitoes compared to the GFP-KD control (Fig. 6A). In AaFAS1-KD mosquitoes, expression levels of AaFAS2, 3 and 5 and were 191.7±38.6%, 161.4±21.8%, and 191.1±38.9%, respectively, in comparison to their levels in GFP-KD control, indicating possible compensation for the loss of AaFAS1 transcript. Conversely, the expression of AaFAS4 was 87.71±74.0% compared to AaFAS4 expression in GFP-KD control mosquitoes. To determine whether the upregulation observed in AaFAS2, 3, and 5 could possibly compensate for the loss of AaFAS1 in the AaFAS1-KD mosquitoes, we normalized the level of AaFAS genes to β-actin. We observed modest expression of AaFAS transcripts (5.6±1.44% for AaFAS2, 4.6±0.00% for AaFAS3, and 7.07±0.62% for AaFAS5 compared to β-actin), while these upregulation still did not match the remnent of AaFAS1 expression after the KD effect (36.1±11.6%). This data suggest that other AaFASs may not be able to serve as a backup system for AaFAS1, at least in adult female mosquitoes under transient KD condition.
Effect of RNAi-induced AaFAS1 knockdown on DENV2 replication in Ae. aegypti cells
Studies in cell culture have shown that FAS activity is required for flavivirus genome replication (13, 14, 37). Biochemical inhibition of FAS activity reduced DENV2 replication in both human and mosquito C6/36 cells (13, 15, 16). The lack of functional RNAi machinery in C6/36 cells hindered the use of transient KD strategy in mosquito cells. However, Ae. aegypti cells, Aag2, have functional RNAi machinery; therefore, we can investigate the role of AaFAS1, the most abundant transcript in female mosquitoes, in DENV2 replication using dsRNA transient KD in these cells (38). At 48 hours post-AaFAS1-KD (time zero of DENV2 infection), the expression level of AaFAS1 in Aag2 cells was 5.15 ± 6.33% as compared to AaFAS1 expression in GFP-KD control cells (Fig. 7A). At 24 hours post DENV2 infection (72 hours post-KD), we observed significant reduction (p<0.001) in DENV2 RNA replication in AaFAS1-KD cells as compared to the GFP-KD controls, comparable to replication in DENV2-KD (KD positive control) (Fig. 7B). KD was not associated with detrimental effects to the cells (Fig. 7C), suggesting that AaFAS1 is required for DENV2 replication in mosquito cells.
Transient inhibition of AaFAS1 reduced DENV2 infection in the midgut of Ae. aegypti
To investigate the role of AaFAS1 in DENV2 replication in vivo, mosquitoes were IT injected with dsRNA derived from AaFAS1 or GFP genes, and subsequently exposed to DENV2 infectious blood meal two-days post-injection (Fig. 8). On days 0, 3 and 7 pbm (corresponding to 2, 5 and 9 days post-dsRNA injection), whole mosquitoes were collected and analyzed for AaFAS1 gene expression (Fig. 8A). On the day of DENV2 infection by blood meal (2 days post-dsRNA injection), the level of AaFAS1 expression was downregulated by 53.73 ± 27.13% relative to GFP-KD group. On day 3 pbm, AaFAS1 expression recovered to 119.82 ± 49.43% and was comparable to the AaFAS1 expression level in the GFP-KD control. On day 7 pbm, AaFAS1 was upregulated to 191.69 ± 50.17%, suggesting a possible over-compensation post KD effect (Fig. 8A).
Investigation of DENV2-fed mosquitoes showed that, at day 3 pbm, we observed significant reduction in percent of DENV2 infected midguts compared to the GFP-KD control (Fig. 8B). The odds ratio for AaFAS1 in AaFAS1-KD mosquitoes compared to GFP-KD mosquitoes on day 3 pbm was 0.20 (95% confidence interval (CI): 0.06 - 0.60) and for DENV2 compared to GFP-KD was 0.03 (CI: 0.01 – 0.13) suggesting fewer mosquitoes were infected with DENV2 in the AaFAS1-KD group compared to the GFP-KD group on day 3 pbm. However, no differences in percent infection were observed between AaFAS1-KD and control mosquitoes on days 7 and 14 pbm. Infectious particles produced from midguts (virus titer) from AaFAS1-KD and DENV2-KD groups were significantly different from GFP-KD group when the uninfected samples were included in the analysis using the nonparametric Kruskal-Wallis test followed by Dunn’s test, with p-values adjusted with the Bonferroni method (p = 0.0002 and p < 0.0001, respectively). However, if the titers of uninfected midguts were excluded, differences in virus titer among different dsRNA treatments were not detected (tested by one-way ANOVA followed by Dunn’s test; virus titer AaFAS1-KD: 2.40x102, GFP-KD: 2.18x103, and DENV2-KD: 6.43x102 plaque forming unit (PFU/midgut).
The inhibitory effect of AaFAS1-KD on DENV2 infection did not persist in the midgut beyond day 3 pbm. The titer and percent infection in the AaFAS1-KD mosquitoes were comparable to the GFP-KD mosquitoes on day 7 pbm (AaFAS1-KD: 2.70x103, GFP-KD: 2.95x104 and DENV2-KD: 5.88x103 PFU/midgut; Fig. 8C). No differences in viral titer and percent infection were observed in midgut on day 7 pbm. To investigate whether transient AaFAS1-KD could disrupt virus dissemination, mosquito carcasses (whole body without midgut) were tested on day 14 pbm for virus infection (AaFAS1-KD: 6.57x104, GFP-KD: 8.47x104 and DENV2-KD: 0.00 PFU/carcass; Fig. 8D). Although we observed no statistical differences in mean titer in AaFAS1-KD as compared to GFP-KD control samples, two distinct populations of mosquitoes with viral titers in AaFAS1-KD carcasses were observed (Fig. 8D); some with viral titers comparable to GFP-KD control (5.77x104 PFU/carcass; i) and some with distinctively lower titers (2.95x101 PFU/carcass, ii). This observation suggsts that transient KD of AaFAS1 had a prolonged effect that can impact dissemination of DENV2 in mosquitoes.