Analysis of defensive secretion of a milkweed bug Lygaeus equestris by 1D GC-MS and GC×GC-MS: sex differences and host-plant effect

The composition of defensive secretion produced by metathoracic scent glands was analysed in males and females of the milkweed bug Lygaeus equestris (Heteroptera) using gas chromatography with mass spectrometric detection (GC-MS). The bugs were raised either on cardenolide-containing Adonis vernalis or on control sunflower seeds in order to determine whether the possibility to sequester cardenolides from their host plants would affect the composition of defensive scent-gland secretion. Profiles of the composition of defensive secretions of males and females raised on sunflower were closely similar, with predominant presence of (E)-2-octenal, (E)-2-octen-1-ol, decanal and 3-octen-1-ol acetate. The secretion of bugs raised on A. vernalis was more sexually dimorphic, and some chemicals e.g. (E,E)-2,4-hexadienyl acetate and 2-phenylethyl acetate were dominant in males, but absent in females. Compared to bugs from sunflower, the scent-gland secretion of bugs raised on A. vernalis was characterized by lower overall intensity of the peaks obtained for detected chemicals and by absence of some chemicals that have supposedly antipredatory function ((E)-2-hexenal, (E)-4-oxo-hex-2-enal, 2,4-octadienal). The results suggest that there might be a trade-off between the sequestration of defensive chemicals from host plants and their synthesis in metathoracic scent-glands.

the bug could not move anymore. The syringe tip was then closed with a rubber stopper. The compression lasted until droplets of liquid appeared ventro-laterally at the metathorax of the bug. Subsequently, the rubber stopper was removed from the syringe tip, 5 mL of air was drawn in and a SPME fiber coated with triple phase 50/30 μm divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS) located in a manual SPME holder (Supelco, Bellefonte, PA, USA) was immediately inserted into the tip of the barrel. The syringe with the bug and the SPME fiber was placed in the temperature-controlled shaker again. The extraction procedure was based on our previous study 11 and the adsorption time (30-120 min) and temperature (25-45 °C) were optimized with incubation for 90 min at 40 °C.
The SPME fiber was thoroughly cleaned and conditioned for 15 min before and after each analysis in an external syringe oven at 250 °C under vacuum. Before each analysis of a defensive secretion, we carried out control analysis of the cleaned SPME fiber (without sorption of analytes) and then with sorption in the clean vacant syringe to identify background compounds. The analyses of secretions obtained from individual bugs within each of the four groups (males and females raised either on A. vernalis or sunflower) were carried out in triplicate (n = 3), i.e. three milkweed bugs were used in total per group and each one was used for secretion collection three times (but not each collection provided release of scent-gland secretion). In parallel, a measurement using a non-irritated bug that underwent the same procedure as for secretion sampling except for being pressed with the plunger was performed as a blank.
Chemical analysis. The 1D GC-MS analysis was performed using a gas chromatograph with a quadrupole mass spectrometer GCMS-QP2010 Ultra instrument (Shimadzu, Kyoto, Japan). The instrument was equipped with a mid-polar Rtx-200 column -20 m × 0.15 mm i.d., 0.15 μm film thickness (trifluoropropylmethyl polysiloxane stationary phase, Restek, USA) or with a low-polar SLB-5ms column -30 m × 0.25 mm i.d., 0.25 μm film thickness (5% diphenyl/95% polydimethylsiloxane stationary phase, Supelco, USA). Helium (99.999%, Linde, Czech Republic) was used as a carrier gas at a constant linear flow rate of 35 cm s −1 . Splitless-mode injection (1 min) with SPME liner at 250 °C was employed for analysis of the true bugs' secretion. For measurement of the standard solutions, 1 μl of the sample was injected by an auto-sampler (AOC-20i, Shimadzu) in the split mode (split ratio, 1:50).
The oven temperature gradient was set as follows: 35 °C was held for 3 min, ramped at 5 °C min −1 to 130 °C, then ramped at 20 °C min −1 to 300 °C and maintained for 5 min (the total run time was 35.5 min).
The mass spectrometer was operated in the full scan mode (m/z 35-500). The ion source and interface temperatures were 200 and 250 °C, respectively.
A Shimadzu gas chromatograph with quadrupole mass spectrometer GCMS-QP2010 Plus (Shimadzu, Kyoto, Japan) equipped by a cryogenic modulator Zoex ZX1 (Zoex Corporation, Houston, USA) with a two-stage loop liquid nitrogen modulator was used for the GC×GC-MS analysis. The low-polar column SLB-5ms -30 m × 0.25 mm i.d., 0.25 μm film thickness (5% diphenyl/95% polydimethylsiloxane stationary phase, Supelco, USA) was used for the first dimension and polar Supelcowax 10 -1 m × 0.1 mm i.d., 0.1 μm film thickness (poly(ethylene glycol) stationary phase, Supelco, USA) for the second dimension. Another 2 m of the second dimension column was used as a loop.
The cryogenic modulation conditions were: 400 kPa inlet column pressure, modulation period 4 s, duration of hot pulse 375 ms and temperature of hot jet 350 °C. All other parameters were the same as for the 1D GC-MS measurement.
1D GC and GC×GC data were collected and evaluated by GCMS solution software version 2.7 (Shimadzu, Kyoto, Japan) and transformation and visualization of GC×GC data were carried out by GC Image software (version 2.0, Zoex Corporation, Houston, USA). Secretion components were identified by comparing the obtained spectra with those in the NIST 2008 Mass Spectra Library, and the identification was confirmed by comparing the retention times and mass spectra with those obtained from measurements of reference compounds. Identification by library reference only was used for substances for which a standard was not available, but only if the match factor in the NIST 2008 database was higher than 85%. OriginPro 2015 (Origin Lab Corporation, Northampton, MA, USA) software was used for data processing.
Samples of the reference compounds were prepared by dilution of the individual compounds in dichloromethane, except for a mixture of alkanes which were diluted in pentane. The concentration of stock solutions of each compound was 0.1 mg mL −1 but the final dilution of each standard was individually modified to obtain good reproducibility of retention times.

Statistical analysis.
For comparison of similarity among the four studied groups of bugs, we used principal component analysis (PCA). For qualitative comparison, the data matrix was based on presence or absence of individual compounds in the defensive secretion, obtained by the analyses conducted in triplicate, as described in Tab. 1. PCA was carried out using Statistica software (version 8.0, StatSoft Inc., Tulsa, OK, USA).

Results and Discussion
One-dimensional GC-MS analysis of the L. equestris' defensive secretion. For the pilot sampling of the defensive secretion of L. equestris males raised on sunflower seeds and its subsequent GC analysis, we employed the methodology described by Krajíček et al. 11 . As the sampling method by SPME fibre yielded peaks with very high response intensity, only one true bug individual was used for collection of the secretion, instead of three, as described by Krajíček et al. 11 . The sample was separated on a medium polar Rtx-200 column, which is considered to be universal. As can be seen in Fig. 1, individual components of the defensive secretion were not adequately separated on this column, compared to the defensive secretions of the true bug species analysed in a previous study 11 and the spectrum had substantially fewer peaks. This fact was most obvious in the area around 13-15 min, where there was only one peak in the chromatogram. According to the data obtained in separation of the individual standards on the Rtx-200 column under the same conditions 11 it can be assumed that substances, such as nonanal, (E)-2-octenal, or decanal, co-eluted in this area. Nevertheless, several other analytes were identified using this column (see Fig. 1).
Consequently, in the next experiment, a substantially less polar SLB-5ms column was employed, on which much better separation was obtained under the same experimental conditions, and thus all the subsequent 1D GC-MS analyses were performed on this column. The representative chromatograms obtained for males and females from both host plants are depicted in Fig. 2. Only peaks with a response greater than 20,000 a.u. (arbitrary units) and whose occurrence was confirmed by analysis of scent-gland secretions released by three individuals (one at a time) were chosen for identification.
In order to eliminate misleading results, the peak numbering in Fig. 2 corresponds to all identified components in the defensive secretions, although some compounds co-eluted (large peaks marked with numbers in a circle). These components could not be separated in the 1D GC-MS system and the identification was performed later by GC×GC-MS. The composition of the defensive secretion together with the retention times of the compounds are summarised in Tab. 1. In order to estimate the representation of individual substances in the secretion, the means of relative peak areas (n = 3), obtained by a method of the internal normalization and expressed as percentages, are also included in Tab. 1. The percentages of each component represent its relative abundance in the chromatogram. The means of relative peak areas are also expressed by the plus symbols (in parentheses); the higher the number of plus symbols, the higher the relative peak area value (approximately, one plus symbol equals the relative peak area up to 5%, two plus symbols up to 20% and three plus symbols over 20%). This simplification is also advantageous for comparison with GC×GC-MS results.

Two-dimensional GC×GC-MS analysis of L. equestris' defensive secretion.
Because of co-elution of some components of the defensive secretion in the 1D GC-MS system, analysis using GC×GC-MS in an orthogonal system of a low-polarity SLB-5ms column in the first dimension and a polar Supelcowax 10 column in the second dimension under the conditions described above was performed with the same sampling method as for 1D GC-MS. Secretions from males and females of L. equestris raised on toxic A. vernalis or sunflower as a control were measured, and the resultant contour plots are depicted in Fig. 3. The composition of defensive secretions was identified by comparing the obtained spectra with the Mass Spectra Library and obtained retention times with standard retention times, if available, as described in the section Material and Methods. The identified components are listed in Tab. 2 and only compounds whose occurrence was confirmed by analysis of scent-gland secretions released by three individuals (one at a time) were chosen for identification. The software used for the evaluation of blobs in GC×GC-MS does not allow for the application of the internal normalization method (as in 1D GC-MS) The plus symbol indicates only the presence of a given compound in the studied defensive secretions, www.nature.com/scientificreports www.nature.com/scientificreports/ and their number gives a visual estimate of the intensity of the individual blob (the yellow colour represent the highest intensity, see Fig. 3). This facilitates comparison of the results obtained in the different groups tested.
The use of the SLB-5ms column both in 1D GC-MS and in the first dimension of GC×GC-MS enabled subsequent comparison of the obtained data. The use of GC×GC-MS facilitated the identification of a larger number of analytes compared to the use of 1D GC-MS and, in particular, it was possible to identify the individual components of the peaks, which co-eluted in the 1D system, exhibited the highest intensities and probably represented the most important components of the defensive secretion. This corresponded to the peak with a retention time of 16.9 min in the 1D-GC-MS system, which could be separated into two peaks using the GC×GC-MS system that were identified as (E)-2-octenal and (E)-2-octen-1-ol at times 15.90 and 16.20 min, respectively. Unfortunately, the peak eluting at time 18.7 min (in 1D GC-MS) could not be completely separated into its individual components even using GC×GC-MS, but they could be at least identified. In this case, the substances decanal and 3-octen-1-ol acetate were identified (both at time 20.30 min in Fig. 4). The four analytes mentioned above were the substances with the greatest relative contents in the defensive secretion and they are likely to play an important role in either antipredatory defence or intraspecific communication of Heteroptera. Especially aldehydes, e.g. C 6 and C 8 (E)-2-alkenals and (E)-4-oxo-2-alkenals, are substances that are important for defence against predators in true bugs 43 .
For the peak eluting in the 1D GC-MS system at time 15.5 min, the separation in the GC×GC-MS system was not sufficiently effective; however, it allowed for identification of the individual components as 1-octanol, 2,6-dimethyl-7-octen-2-ol and (E,E)-2,4-hexadienyl acetate (see the differences between sexes). On the other hand, at retention times of 12.72 and 14.78 min (for GC×GC-MS), it was possible to completely separate the pair benzaldehyde and (E)-4-oxo-2-hexenal and the pair (R)-(+)-limonene and 2-ethyl-1-hexanol.
The intensities in responses of individual components of the defensive secretion can be compared in the contour plot obtained using GC×GC-MS. The contour plots depicted in Fig. 3 were obtained under the same conditions and equally adjusted by the evaluating GC Image software. It is thus possible to observe marked differences in the intensities of the responses of the individual peaks between males and females and especially between individuals raised on different host plants.
Because of the usage of the same sampling method, we could directly compare our results with those reported by Krajíček et al. 11 for three true bug species from the family Pyrrhocoridae. Analytes belonging to the most abundant and thus probably representing the most important components of the defensive secretion of L. equestris were also identified in at least one of the three pyrrhocorid species analysed by Krajíček et al. 11 . These analytes were (E)-2-octenal, decanal and (E)-2-octen-1-ol. Thirteen additional substances that occurred in the secretion www.nature.com/scientificreports www.nature.com/scientificreports/ of sunflower-raised L. equestris matched the compounds found in at least one of the three studied pyrrhocorid species including (E,E)-2,4-hexadienal, 3-methyl-1-butanol acetate, 2-ethyl-cyclohexanone, 2-hexen-1-ol acetate, and undec-2-enyl acetate 11 .
The composition of the metathoracic scent-gland secretion was analysed in several other species of milkweed bugs (Lygaeinae), though using different methods of secretion sampling and analysis, and therefore the results are not entirely comparable. Nevertheless, short-chained aldehydes (namely (E)-2-hexenal and (E)-2-octenal), alcohols and acetates were dominant also in secretions of metathoracic scent-glands in Lygaeus kalmii, Oncopeltus fasciatus, O. cingulifer, O. unifasciatellus, Tropidothorax cruciger, and Spilostethus rivularis 43-45 . Differences in the composition of defensive secretion between males and females. There were minimal differences between males and females of L. equestris raised on sunflower seeds. With regard to the diversity of substances, the compositions were practically identical in both sexes. Differences were found especially in ratios of the intensities of the largest peaks. In males, the predominant components of the secretion were decanal and 3-octen-1-ol acetate and, compared to the females, the peak of cyclooctanol was more intense, while in females, the predominant peaks were those corresponding to (E)-2-octen-1-ol and (E)-2-octenal. Most of the simple alcohols belong particularly to the group of insect attractants and, in various ratios, especially the above two alcohols indicate that these substances also have a pheromonal function 46 . In addition, across all measurements, about a one third larger amount of branched cyclohexanones was identified in females (these  Table 1. The compositions of the defensive secretion released by males and females of L. equestris raised either on toxic A.vernalis or sunflower (Helianthus annuus) as a non-toxic control obtained by 1D GC-MS conducted in triplicate; peak numbers (same as in Fig. 2), means of retention times with standard deviation (t ret ), means of relative peak areas expressed as percentages with standard deviation, plus symbol (+) denotes presence of the compound (number of plus symbols indicates relative peak area value -approximately one plus symbol equals relative peak area up to 5%, two plus symbols up to 20% and three plus symbols over 20%), and the identification method. *Bold -co-elution in the 1D GC-MS. The methods used for the identification: A -retention time and mass spectrum of the relevant substance was compared with the reference compound; Bthe mass spectrum of the relevant substance was compared with NIST 2008 MassSpectraLibrary. www.nature.com/scientificreports www.nature.com/scientificreports/ were allyl-cyclohexanones and, based on the spectral library, most probably 2-pentyl-cyclohexanone and 2-butyl-cyclohexanone). These substances are usually considered to be pheromones 47 .
Differences between females and males raised on seeds of A. vernalis were found not only in the intensity of the response of individual components, but we also found qualitative differences. In females, the intensity of obtained peaks was substantially lower than in males. This phenomenon could be associated with a higher amount of cardenolides sequestered from host plants by the females as described for the monarch butterfly (Danaus plexippus) 48 . The greatest difference in the composition of the defensive secretion is represented by the peak at a retention time of 15.5 min (for 1D GC-MS) found only in males (almost the largest in GC×GC-MS). Identification of the components of this peak was impossible when using 1D GC-MS and adequate separation was not achieved even with GC×GC-MS (see Fig. 5). Most probably, three substances co-eluted here and, on the basis of point spectra at the three most intense points in the blob, three analytes were identified as 1-octanol, 2,6-dimethyl-7-octen-2-ol and (E,E)-2,4-hexadienyl acetate and the last mentioned compound was the most abundant.
Another important difference between males and females raised on A. vernalis was the absence of 2-phenylethyl acetate in the females' chromatograms. This compound was the fourth largest peak in the 1D GC-MS chromatograms from males, and in the GC×GC-MS contour plots it was amongst the four most intense peaks. Both these acetates, found only in males, were also found only in secretion of males of other milkweed bugs (e.g. Neacoryphus bicrucis) as pheromones attracting both sexes 44,45,49 . The presence of these acetates in males raised on A. vernalis also supports the hypothesis that toxins sequestered from host plants suppress the defensive role of the metathoracic scent glands in milkweed bugs, and therefore the glands can serve their other function in producing pheromones 6,50 .
The data based on presence or absence of individual compounds in defensive secretions of males and females from different host plants were analysed means of PCA (Fig. 6). The first two principal components (PC1 and PC2) together explained for 92.83% of total variance. Males and females raised on sunflower seeds are closely similar, whereas males and females raised on A. vernalis considerably differ from each other.
Effect of host-plant toxicity on composition of defensive secretion. When pressed with the plunger during the secretion sampling, we observed that the bugs raised on A. vernalis showed a substantially lower willingness to release the scent-gland secretion. Instead, they frequently released defensive fluid from the dorsolateral space (probably containing cardenolides 32,51 ). This fluid was also sampled using SPME fibres in the same way as the scent-gland secretion. When the chromatogram of the dorsolateral space fluid was compared with a blank, (see Fig. 7) only very few volatile components could be detected. Based on GC×GC-MS chromatograms we only identified a small amount of octanal, nonanal, decanal and also α-pinene. Releasing the dorsolateral space fluid as a response to pressure corresponds with the presence of an elaborate system of subcuticular storage compartments 32 , which allows the quick discharge of the fluid upon predator attack.  www.nature.com/scientificreports www.nature.com/scientificreports/ More importantly, scent-gland secretion of the bugs raised on A. vernalis contained a smaller number of chemicals compared to the secretion of sunflower-raised bugs. Moreover, for most of the chemicals that were present in secretions of bugs from both host plants, the overall intensity of the peaks obtained was lower in the bugs raised on A. vernalis than in the bugs from sunflower (Tab. 2, Fig. 3). Out of 14 chemicals that were dominant in the secretion of sunflower-raised bugs of both sexes (Tab. 2), only 2 (cyclooctanol and α-pinene) were equally dominant in the secretion of bugs raised on A. vernalis. Six others were present in a much lower amount (e.g. (E)-2-octenal, decanal, 3-octen-1-ol acetate) and 6 were entirely absent (e.g. (E)-2-hexenal, (E)-4-oxo-2-hexenal, (E)-2-octen-1-ol). Contrastingly, out of the chemicals dominant in the secretion of the bugs raised on A. vernalis, only 4 (e.g. 2-phenylethyl acetate, (E,E)-2,4-hexadienyl acetate) were absent in secretion of sunflower-raised bugs, all of them being sex specific and present only in males (see above). In general, the secretion of individuals raised on A. vernalis contained mostly esters, variously branched alcohols, and some aldehydes and, amongst substances eluting later, also higher carboxylic acids, whereas the secretion of bugs raised on sunflower was dominated by aldehydes and oxo-aldehydes and also some alcohols and esters.
The absence (or low amount) of some of the aldehydes (3-hexenal, (E)-2-hexenal, (E)-2-octenal, heptanal, octanal, 2,4-octadienal, (E)-2-nonenal, (E,E)-2,4-decadienal) and oxo-aldehydes ((E)-4-oxo-2-hexenal) in the secretion of bugs raised on A. vernalis is striking, as they are considered to be important constituents of true-bug antipredatory defence, being aversive 52 or toxic for arthropod predators 6,16,53 and aversive for vertebrate predators, specifically for lizards 54 , geckos 55 Table 2. The compositions of the defensive secretion released by males and females of L. equestris raised either on toxic A. vernalis or sunflower (Helianthus annuus) as a non-toxic control obtained by GC×GC-MS conducted in triplicate; peak numbers (same as in Fig. 3), retention times in the first (t ret_1 ) and second (t ret_2 ) dimension, +/− symbols denote presence or absence of a particular compound (number of plus symbols indicates intensity of the individual blob) and the identification method. The methods used for the identification: A -retention time and mass spectrum of the relevant substance was compared with the reference compound; B -the mass spectrum of the relevant substance was compared with NIST 2008 MassSpectraLibrary. www.nature.com/scientificreports www.nature.com/scientificreports/ defence, with the bugs raised on cardenolide-producing host plant investing less to the defensive components of their scent-gland secretion. On the other hand, it seems that the metathoracic scent glands, which are somewhat reduced in Lygaeinae and are regarded as having mostly pheromonal function 6 , might represent a more flexible system, which interacts with the alternative defence mechanism based on sequestration of host-plant chemicals.

conclusions
Composition of the defensive secretion of males and females of the milkweed bug Lygaeus equestris was analysed using 1D GC-MS and GC×GC-MS techniques. The bugs were raised either on a toxic, cardenolide-producing host plant or on a control sunflower. The secretion was sampled by slightly compressing the bugs with the plunger of a syringe and subsequent SPME analysis. Separation in a 1D GC-MS system with a mid-polar Rtx-200 column and also with the low-polar SLB-5ms column was not sufficiently effective for some substances, which co-eluted. The use of GC×GC-MS with an orthogonal polarity columns enabled separation or at least identification of the formerly co-eluting substances, such as (E)-2-octen-1-ol, (E)-2-octenal, decanal, 3-octen-1-ol acetate, (E,E)-2,4-hexadienyl acetate, 2,6-dimethyl-7-octen-2-ol and 1-octanol, which play an important role in antipredatory defence or intraspecific communication. Dominant components in the secretion of bugs raised on sunflower were