Host plants and obligate endosymbionts are not the sources for biosynthesis of the aphid alarm pheromone

(E)-β-farnesene (EβF) is the major component of the alarm pheromone of many aphid species, but where EβF is synthesized in aphids is only partly understood. There are at least three most possible sources for the alarm pheromone: host plants, aphid obligate endosymbiont and aphids themselves. Here we eliminated the possibility of host plants and the obligate endosymbiont Buchnera aphidicola as the sources for EβF released by aphids. We excluded the possible effects of host plants on EβF biosynthesis by rearing aphids on non-plant diets. Both the diet-reared aphids, including the cotton aphid Aphis gossypii and the green peach aphid Myzus persicae, could still release EβF based on solid-phase micro-extraction combined with gas chromatography-mass spectrometer analysis. Meanwhile, we treated host aphids with antibiotics to fully eliminate Buchnera bacteria. Though the treatment seriously affected the development and fecundity of host aphids, the treated aphids could still release EβF, and there was no significant difference in the EβF concentration as per the aphid weight under different rearing conditions. Taken together, our experimental results suggest that host plants and obligate endosymbionts are not the sources for EβF released by aphids, indicating that it is most probably the aphid itself synthesizes the alarm pheromone.


Artemisia annua
. Some of these EβF synthase genes have been engineered into target plants, and the latter could release EβF with an implication in aphid control [31][32][33] . The plant-originated EβF synthase was successfully used in combination with aphid FPPS to produce EβF in vitro 19 , indicating that the aphid FPPS might be involved in the biosynthesis of aphid alarm pheromone. Unfortunately, up to date, no genes responsible for biosynthesis of EβF in aphids have been identified, nor EβF has been synthesized exclusively based on aphid genes in vitro.
The isoprenoid pathway has also been studied in microorganisms. Fujisaki et al. 34 found that the geranyl transferase ispA gene was responsible for the biosynthesis of FPP in Escherichia coli. The whole genome sequence of the obligate intracellular endosymbiont of aphid, Buchnera aphidicola, which provides essential amino acids and other nutrients for its hosts 35 also contains ispA gene [36][37][38] , though its function has not been characterized. The functional plasticity of terpenoid synthases was reported in plants, insects and microbes. For instances, Green et al. 39 found apple α-farnesene synthase possessing prenyltransferase activity; the isoprenyl diphosphate synthase in the bark beetle Ips pini had an amateur function of terpene synthase 40,41 , while the monooxygenase albaflavenone synthase in the soil bacterium, Streptomyces coelicolor, was shown to have a moonlighting terpene synthase activity 42 .
Since the genes encoding the enzymes responsible for the biosynthesis of EβF have not been identified in the aphid genome, we thus speculate that there are at least three possible sources for EβF biosynthesis: host plants, obligate endosymbionts, and aphids themselves. A previous work demonstrated that the aphids (M. persicae) reared on standard artificial diet could release EβF though the sizes of these diet-reared aphids became smaller, but an acetate-supplemented diet could rescue this defect 43 . In the present study, we used a modified diet formula and two aphid species, A. gossypii and M. persicae, to confirm that both an artificial diet and the obligate endosymbionts could not prevent EβF production in aphids, based on our solid-phase micro-extraction (SPME) combined with gas chromatography-mass spectrometry (GC-MS) results.

Results
Effects of artificial diets on the development and fecundity of aphids. The aphids M. persicae and A. gossypii were reared on improved diets modified from the published prescription 44 . The results for M. persicae and A. gossypii were totally different: M. persicae could not go down beyond the third generation (F 3 ) (only a few sterile offspring was seen), while A. gossypii went down to F 23 at this point. Our experiments showed that the developmental duration of the first-instar M. persicae nymphs of the first generation (F 1 ) reared on artificial diet was significantly longer than that of M. persicae reared on pepper plants (P < 0.01), while the developmental durations of the diet-reared second-, third-and fourth-instar nymphs were not significantly different from those of pepper-reared ones (P = 0.198, 0.581 and 0.771, respectively) ( Table 1). The development of dietreared M. persicae adults was obviously delayed compared with that of pepper-reared counterparts (P < 0.01). As a whole, the diet-reared M. persicae aphids lived highly significantly longer than pepper-reared ones (P < 0.01), and their fecundity was also extremely significantly decreased compared to that of pepper-reared M. persicae aphids (P < 0.01).
Unlike M. persicae, A. gossypii reared on artificial diets exhibited no significant difference from cotton-reared aphids in the developmental duration over different developmental stages from the fifth generation on (N 1 , P = 0.453; N 2 , P = 0.285; N 3 , P = 0.127; N 4 , P = 0.139; adult, P = 0.761) ( Table 2; data for F 17 ), although the fecundity of diet-reared A. gossypii was significantly decreased compared with cotton-reared aphids (P < 0.01). Considering that A. gossypii lived much better on the artificial diets than M. persicae, A. gossypii aphids (F 17 ) and their offspring reared on artificial diets, antibiotic-treated diets and plants were sampled for volatile analysis to examine the effects of host plants and endosymbionts on the biosynthesis of EβF in the present study.
Elimination of Buchnera and its effects on aphid development and fecundity. The elimination efficiency of antibiotic treatment against Buchnera bacteria was not only dependant on the antibiotic treatment  Figure S1), Buchnera bacteria in M. persicae could be effectively eliminated by reared on normal diets for 10 days in addition to treatment with tetracycline or rifampicin for 12-14 days. Rearing aphids on normal diets for 5 days after antibiotic treatment for 12-14 days could not successfully eliminate Buchnera from its host. For molecular detection of Buchnera in A. gossypii, we also cloned and sequenced its ispA gene (GenBank acc.no. KX999117), which was 847 bp in length and encoded a protein of 282 amino acids. Specific primers (C-ispA-F/R) designed based on this sequence was used to detect the existence of Buchnera in A. gossypii, and the results showed that the same antibiotic treatment protocol could also effectively eliminate Buchnera bacteria from A. gossypii. Meanwhile, the effects of antibiotic treatment on the development and fecundity were also observed, which indicated that antibiotic diets could fully erase the reproductive ability of both M. persicae and A. gossypii, whereas its effect on the longevity of M. persicae was not significant (P = 0.621), but significant on that of A. gossypii compared with antibiotic-free diet (P = 0.03) (Tables 1 and 2).
Quantitative analysis of cornicle secretions. The proportion of aphids emitting cornicle droplets was significantly lower when reared with artificial diet and under antibiotic treatment than that when reared on cotton plants (P < 0.001), but the proportion had no significant difference between the aphids treated with and without antibiotics (P = 0.123) ( Table 3). The mean weight of antibiotic-treated individual aphid was reduced by 50% compared with untreated aphids (P < 0.01), but there was no difference between cotton plant-reared and normal diet-reared aphids (P = 0.231). The EβF concentration of diet-reared aphid, measured as per milligram of aphid (ng), was not significantly different from those of the plant-reared (P = 0.407) and antibiotic-treated aphids (P = 0.024), though the mean amount of EβF per mg of antibiotic-treated aphid was reduced by approximately 12 ng compared with that of plant-reared aphids (P = 0.006) (Fig. 3A). In addition, the amount of EβF per antibiotic-treated aphid (ng) was significantly less than those of plant-reared (P < 0.01) and diet-reared aphid (P < 0.01) (Fig. 3B), reduced by approximately 50%, but there was no significant difference between plant-reared and normal diet-reared aphids (P = 0.380).

Discussion
The whole genome sequence of the pea aphid A. pisum was published in 2010, but no sequences belonging to the terpene synthase family have been identified. The question is: where and how the aphid alarm pheromone is biologically synthesized? There are at least two other possible sources for biosynthesis of the pheromone: host plants and obligate endosymbionts, since the cDNAs encoding EβF synthase have been sequenced and characterized in plants and microbes 27,42 . Here, we eliminated the possibility of host plants and the obligate endosymbiont Buchnera as the sources for EβF biosynthesis in two aphid species, A. gossypii and M. persicae, indicating that it is probably the aphids themselves responsible for biosynthesis of their pheromones. Why is the host plant a possible source for aphid alarm pheromone? It is known that many plant species synthesize and release the sesquiterpene EβF 27-30 . Since no genes responsible for the biosynthesis of EβF have been identified in the aphid even if the whole genome of aphid is available, the possibility for the host plant to be a source of aphid alarm pheromone cannot be excluded. It is not difficult to verify this hypothesis: If the host plant is the source for aphid-released EβF, the aphids reared on non-plant diets would not release EβF. A previous work examining the effect of artificial diet on the production of EβF suggested that the aphids (M. persicae) reared on standard artificial diet could release EβF though the sizes of these diet-reared aphids became smaller, which resulted in a decreased total amount of EBF below the alarm response threshold, but an acetate-supplemented diet could rescue this defect 43 . In the present study, our experimental data confirmed that the two aphid species reared on artificial diets could release EβF, which excluded the effects of host plants on the biosynthesis of aphid alarm pheromone, as the ingredients of the diet contained no intermediates or precursors of EβF (Supplementary  Table S1). The structural similarity between the major aggregation pheromone components of bark beetles and the monoterpenes found in the host trees led researchers to believe that the insect might use plant-derived precursors to produce their pheromone components 45 , but it is now recognized that most bark beetle aggregation   pheromone components are biosynthesized de novo 40,46 . Nevertheless, the labeling with radiolabeled acetate will provide direct evidence for the biochemical pathway of EBF biosynthesis. The mutualistic relationship between aphid and its obligate endosymbiont Buchnera evolved from about 80-150 million years ago 47,48 : The aphid hosts Buchnera, while Buchnera provides its host with essential amino acids and other nutrients 49,50 . In contrast, aphids are also infected with various facultative endosymbionts which are not essential for aphid development and reproduction. Here we only examined the effect of the obligate endosymbiont Buchnera on biosynthesis of EβF in aphids since it is not evolutionarily "safe" for the facultative endosymbionts to be responsible for such an important phenotype as aphid alarm behavior. Considering the functional plasticity of terpenoid synthases, the ispA gene encoding FPPS in Buchnera may be involved in the production of the sesquiterpene EβF. We thus inactivated Buchnera bacteria by antibiotic treatment, and then double checked the bacteria with universal primers and the primers specifically targeting ispA. Our results confirmed the validity of our antibiotic treatment procedure, as the subsequent GC-MS analysis verified that the aphids without Buchnera bacteria could still release EβF, obviating the possibility of Buchnera as a source for biosynthesis of EβF in aphids. The symbiotic bacteria in the New Zealand grass grub beetle were found to be related with the production of the beetle's sex pheromone 51 , but no direct evidence has been presented. Similarly, the bacterium Bacillus cereus isolated from Ips bark beetles was shown to be capable of converting α-pinene to verbenol (component of beetle aggregation pheromone) in vitro 52 , but no in vivo evidence showing the actual synthesis of verbenol from α-pinene in the beetle gut has been demonstrated.
The diet for rearing M. persicae was modified from the published prescription 44 , but M. persicae could not live well on it, whereas A. gossypii became rapidly adapted to it (we got F 23 offspring of diet-reared A. gossypii at this point). It was reported that the fecundity of A. pisum reared on artificial diets decreased sharply till sterility, which might be resulted from the decrease in the number of their symbiotic bacteria 53 . Our finding that the effects of the artificial diet on the development and fecundity were distinctly different in different aphids suggested that the nutritional requirement was quite different for different aphids. On the other hand, our results indicated that 90% of cotton plant-reared aphids emitted cornicle droplets when disturbed, while only approximately 60% of diet-reared and antibiotic-treated aphids emitted droplets. As the antibiotic-treated aphids were also reared on diet, the emission proportion of this treatment was not significantly different from that of normal diet-reared aphids. Thus, it is most likely that the lower proportion of droplet emission in diet-reared aphids was an effect of the diet, which might lack some important ingredients involved in the biosynthesis of EβF. It was previously found that the total amount of EβF in diet-reared aphids could be increased by adding acetate in the diet 43 . Here, our data showed that the weight of antibiotic-treated aphids reduced by approximately 50% compared with that of the untreated aphids, and the concentration of EβF (ng per aphid) in the antibiotic-treated aphids was also less than that in both plant-and diet-reared aphids. One possible explanation for this may be that removal of the obligate endosymbionts could negatively affect the development of the aphids, and the ispA gene of Buchnera might also play a positive role in the process of EβF biosynthesis by providing the precursor FPP for the host aphid.
EβF as the major component of the alarm pheromone of many aphid species plays important roles in mediating multitrophic interactions among aphids, host plants, and aphid natural enemies 11 , which has been used as a repellent against aphids or an attractant for natural enemies of aphids. Although plant-originated EβF synthase genes have been engineered into crop plants which emitted EβF and repelled aphids 32 , identification of the genes responsible for the biosynthetic pathway leading to production of EβF in aphids is certainly significant as these genes are at the core of the molecular mechanism underlying EβF biosynthesis. With these genes, we can not only synthesize the "real" aphid alarm pheromones in vitro, but also control aphid behaviors by genetic modification. Now that we excluded the possibility of host plants and obligate endosymbionts as the sources for biosynthesis of aphid alarm pheromone, what can we do next to identify the genes responsible for the synthesis of alarm pheromone in aphids? FPPS is a key enzyme in the isoprenoid pathway, supplying C 15 precursors for several classes of essential metabolites in many organisms, including dolichol, ubiquinone, sesquiterpene and sterol 12 . As shown in plants, FPP is the precursor for EβF synthesis 11 . The challenge now is to identify the enzyme(s) responsible for converting FPP to EβF in aphids. The plasticity of isoprenoid synthases including FPPS has been demonstrated in different organisms. It was repeatedly reported that recombinant FPPSs from aphid species also function as GPPS (with dual FPP/GPP synthase activity) [19][20][21][22][23] . Similarly, GPPS from Ips beetles could not only synthesize GPP from DMAPP and IPP, but also convert GPP to the monoterpene myrcene 41 . Now that the whole genome sequence of aphid is available and homology cloning of EβF synthase gene from aphids is unsuccessful, the emphasis for further work will be on the multifunctionality of isoprenoid synthases. The RNA interference technique and transcriptome sequencing would be useful in this line of work.

Methods
Aphids and artificial diet. A. gossypii and M. persicae have been maintained on potted cotton and sweet pepper plants (20 ± 0.5 °C, L16:D8) in China Agricultural University, Beijing since 2008. The artificial diets for rearing aphids were prepared according to the published recipe 44 , but cholesterol was omitted from the diet since we found it was not ideal for aphid development and a previous study showed that aphids could grow from larval stage to adults on a synthetic diet without sterol 54 . The basal composition for diet preparation was given in Supplementary Table S1. Various amino acid stock solutions were separately prepared and then used to mix with trace-metal miscible liquids, ascorbic acid, biotin, folic acid and vitamin B group (pH7.0, adjusted with 30% KOH). The final mixture was sterilized by passing through a filter (Acrodisc ® Syringe Filters with Supor ® Membrane, 0.2 μm, Life Sciences), then distributed into 1.5 mL sterile centrifugal tubes and stored at −20 °C for further use.
Artificial rearing of aphids. A two-way glass tube (2.5 cm × 3.0 cm) with the diet (150-200 µL) sealed between two layers of taut parafilm at one end was used as the feeding device. The parafilm was sterilized with ultraviolet light. Forty adult aphids were introduced into each tube, and then the bottom of the tube was covered with a filter paper and a rubber band to prevent escape of aphids. The tube was placed in a petri dish (Φ9.0 cm) with a layer of wet filter paper for keeping humidity. The adults were removed the next day but the newly-born nymphs were maintained in the tube. The diet was replaced every 2 days until the aphids reached maturity.
Effects of the artificial diet on the growth and development of aphids. One newly-born aphid was introduced into the feeding tube containing the artificial diet. The exuvial case of aphid was checked every 5 h during the nymphal stage, and the duration of each stage and the number of offspring per aphid on artificial diet was recorded. As a control, the developmental duration and fecundity of aphids on plants were also observed by using round-leaf method: 1% agar gel was poured into a petri dish (Φ3 cm), and when the gel became solidified, a round pepper or cotton leaf was laid on the gel for rearing the green peach aphid and the cotton aphid, respectively. One newly-born aphid was then introduced into each dish, with 20 replicates.
Elimination of Buchnera by antibiotic treatment. Fifty newly-born aphids were moved into the feeding tube with the artificial diet containing 100 μg/mL antibiotics (tetracycline and rifampicin, respectively). After fed for 2 d, 6 d, 10 d, 12 d or 14 d, the aphids were transferred to diet without antibiotics, and samples were taken separately for molecular detection of Buchnera after 5 or 10 additional days, with three replicates. The control group was consistently fed with artificial diet without antibiotics. The developmental durations and fecundity of aphids with and without antibiotic treatment were measured during different stages by using the feeding tube method as described above, with 20 replicates.

Molecular detection of Buchnera.
Five aphids from each treatment were subjected to DNA extraction for molecular detection. DNA extraction was performed by using the potassium acetate method as described 55 . Buchnera was detected in aphids according to the previously reported method 56 . The primers used were listed in Supplementary Table S2. The primers targeting the ispA gene of Buchnera were designed to specifically ensure that the ispA gene that is suspected to be involved in the biosynthesis of EβF is absent after antibiotic treatment. As the ispA sequence was unavailable for A. gossypii, the genome sequences of Buchnera aphidicola from other aphids (e.g., Acythosiphon pisum, M. persicae, Schizaphis graminum and Aphis glycines) were retrieved from NCBI and degenerate primers (Buchisp-F/R) were designed for PCR amplification of A. gossypii ispA. The PCR product was then cloned and sequenced according to conventional protocol. Based on the obtained ispA sequence, the primers (C-ispA-F/R) were thus designed to detect Buchnera in A. gossypii.

Analysis of volatile releases from aphids by SPME and GC-MS. Solid-phase micro-extraction
(SPME) combined with gas chromatography-mass spectrometer (GC-MS) was performed for analysis of the volatile releases from plant-reared, diet-reared and antibiotic-treated diet-reared aphids. A synthetic farnesene solution (provided by Group of Chemical Ecology of Forest Insects, Institute of Zoology, Chinese Academy of Sciences) was diluted 20 times with acetone and used as the standard, and 10 μL standard solution was transferred to a glass tube adapted to the SPME with pipette for GC-MS analysis. The standard solution was first maintained for 20 min at 30 ± 0.2 °C, and then the volatile ingredients were sampled for 30 min with 100-µm polydimethyl-siloxane SPME fibers (Supelco ® , Sigma, USA) and immediately analyzed by GC-MS on Perkin Elmer Clarus600. One or ten aphids were crushed with an insect needle prepared from a glass tube and the volatile releases were collected with SPME and then analyzed by GC-MS under the following conditions: The temperature of injection port was from 60 °C to 250 °C at the speed of 200 °C/min. Samples were purged with helium at 4 ml/min for 10 min to make the absorbed substances totally volatilized. The initial temperature of pillar was 100 °C, kept for 1 min, then changed to 180 °C at 6 °C/min and to 280 °C at 15 °C/min, kept for 10 min. Mass spectra were obtained in the EI mode at 80 eV (scanned mass range from 45 to 300 amu) and analyzed with TurboMass Ver5.4.2. Each treatment was repeated for 3 times.  57 and Fischer and Lognay 58 were combined for the quantitative analysis of cornicle secretions. The aphid was stimulated to produce droplet from the cornicle by pressing the dorsum gently with an insect pin under a dissecting microscope. When the cornicle fluid was emitted, the aphid with secretion was put in a 2-ml centrifuge tube (on the ice) that contained 200 μl of distilled hexane and 0.2 μl decane (purity >99.9%) as an internal standard for quantification. A stir bar was placed inside the centrifuge tube and placed onto a working magnetic stir plate for 10 min to completely crush the aphid and extract EβF. Then, 180 μl of clean supernatant was isolated from the homogenate by using a centrifuge and transferred to a chromatography vial. Samples were maintained at −20 °C until analysis. One microliter of sample was injected in Perkin Elmer Clarus 600 for GC-MS analysis by using the methods described 57,58 . If an aphid did not produce any cornicle fluid, it was placed in another 1.5-ml centrifuge tube for weighting. At least 100 aphids were collected for each tube, with 4 replicates. The numbers of aphids emitting or not emitting cornicle droplets were recorded. Statistical analyses. Developmental duration and fecundity data of plant-reared, diet-reared, and antibiotic-treated diet-reared aphids were analyzed by using least significant difference (LSD) multiple comparison method at 5% and 1% levels on SPSS 20.0 software. The EβF amounts were calculated by the peak areas of EβF and decane on TurboMass Ver.5.4.2 software. A covariance analysis of the EβF amounts per mg of aphid and per aphid reared under different conditions was also conducted.