Flower scent of Ceropegia stenantha: electrophysiological activity and synthesis of novel components

In specialized pollination systems, floral scents are crucial for flower–pollinator communication, but key volatiles that attract pollinators are unknown for most systems. Deceptive Ceropegia trap flowers are famous for their elaborate mechanisms to trap flies. Recent studies revealed species-specific floral chemistry suggesting highly specialized mimicry strategies. However, volatiles involved in fly attraction were until now identified in C. dolichophylla and C. sandersonii, only. We here present data on C. stenantha for which flower scent and pollinators were recently described, but volatiles involved in flower–fly communication stayed unknown. We performed electrophysiological measurements with scatopsid fly pollinators (Coboldia fuscipes) and identified 12 out of 13 biologically active floral components. Among these volatiles some were never described from any organism but C. stenantha. We synthesized these components, tested them on antennae of male and female flies, and confirmed their biological activity. Overall, our data show that half of the volatiles emitted from C. stenantha flowers are perceived by male and female fly pollinators and are potentially important for flower–fly communication in this pollination system. Further studies are needed to clarify the role of the electrophysiologically active components in the life of scatopsid fly pollinators, and to fully understand the pollination strategy of C. stenantha.


Introduction
Most angiosperms show mutualistic interactions with their animal pollinators, but 4-6% of flowering plants are pollinated by deceit (Renner 2006). The flowers/inflorescences of deceptive plants cheat their pollinators by either advertising a reward (e.g., breeding substrate, mating site, food) which they do not provide or by providing a reward different from what they advertise (Renner 2006). Thereby, pollinators are fooled by visual, olfactory, or tactile signals which generate a sensory impression of a desired object they will, however, not find in the flowers. In specialized deceptive pollination systems, olfactory signals (i.e., floral scents) are considered most important for pollinator attraction (Vereecken and McNeil 2010;Johnson and Schiestl 2016;Goodrich and Jürgens 2018;Wee et al. 2018); though additional traits, such as morphological or tactile cues, help to keep the disappointed pollinators interested, and trapping devices may even detain 1 3 them in the flowers to achieve successful pollination (see Araceae: Chartier et al. 2014; Apocynaceae-Asclepiadoideae : Vogel 1961;Heiduk et al. 2017;Aristolochiaceae: Oelschlägel et al. 2009Orchidaceae: Singer 2002).
Celebrities among such cheating plants are the highly sophisticated trap flowers of Ceropegia L. (Apocynaceae-Asclepiadoideae-Ceropegieae). This species-rich genus (> 220 species) is restricted to Old World tropical and subtropical habitats with hotspots of diversification in South-East Africa, India, Madagascar, and China (Meve and Liede-Schumann 2007;Bruyns et al. 2015). Ceropegia species are predominantly visited by small Diptera, though other insects/arthropods enter the flowers as well (Ollerton 1999). Hitherto, no other insects but flies where found to carry pollinaria, thus Ceropegia species are functionally highly specialized on Diptera as pollinators. As recently reviewed by Ollerton et al. (2017), 16 fly families are confirmed as Ceropegia pollinators plus additional nine families are known visitors of Ceropegia flowers. The currently available data (Ollerton et al. 2017) show that most Ceropegia species specifically attract flies of only one or two families and genera, respectively, though a few species are visited and pollinated by up to seven fly families (Ollerton et al. 2017). Recent studies on floral chemistry suggest that this pollinator specificity is achieved through species-specific floral scents (Heiduk et al. 2015(Heiduk et al. , 2016. Data on flower scent are available for a total of 14 Ceropegia species  and these data point towards specialized mimicry strategies to attract fly pollinators, such as mimicry of injured/dying insects as a food source for kleptoparasitic flies (Heiduk et al. 2015(Heiduk et al. , 2016, and rotting fruits . However, the compounds which are physiologically and/or behaviorally active in the respective fly pollinators are only identified for two species, i.e., C. dolichophylla (Heiduk et al. 2015) and C. sandersonii (Heiduk et al. 2016), which are both pollinated by kleptoparasitic milichiid flies. For all other Ceropegia species, the key floral volatiles for pollinator attraction remain to be identified. For this purpose, electrophysiological measurements are a powerful tool to screen for compounds that are potentially involved in attraction; however, their attractiveness finally has to be proven by behavioral assays (Heiduk et al. 2015(Heiduk et al. , 2016. Being highly selective chemical lures, Ceropegia flowers provide an ideal system to learn more about fly attractive chemicals, and the behavior as well as life history of flies that act as pollinators. The flowers of Ceropegia stenantha K. Schum. are pollinated by Scatopsidae both in native (Tanzanian habitats) and non-native ranges. In the plants' native habitat, four scatopsid fly species of the three genera Neorhegmoclemina, Rhegmoclemina, and Swammerdamella were described as pollinators . Flowers of plants cultivated in Germany were found to be pollinated by two different scatopsid species of the genera Coboldia and Swammerdamella .
Scatopsidae are scavengers or detritivores in larval stages (Freeman 1985;Haenni and Vaillant 1994;Haenni 1997) and are known to be associated with rotting organic material, such as fermenting leaf litter (Perez et al. 2013). Scatopsidae are also described as nectar and pollen-feeding flower visitors (Larson et al. 2001;García-Robledo and Mora 2007;Woodcock et al. 2014). One of the non-native pollinator species of Ceropegia stenantha, i.e., the oyster mushroom fly Coboldia fuscipes (Meigen 1830), is famous for being a pest insect of economic importance. Just as its common name suggests, the oyster mushroom fly causes severe damage to production of edible mushrooms, such as shiitake, champignon, and oyster mushroom (Yi et al. 2015, and references therein). Thus, mushroom-derived chemicals possibly attract these cosmopolitan flies; as their immature stages develop in diverse media, i.e., cow dung (Skidmore 1991), fungi (Dely-Draskovits 1974; 40 species of host fungi), various decaying vegetal (Duda 1928;Lyall 1929;Rabello and Forattini 1962) and dead animal matter (Ayre 1995), volatiles related to these substrates are most likely attractive as well. Though the antennal structure of these flies has been studied in detail (Zhang et al. 2016), nothing is known about the chemical perception in C. fuscipes.
To understand why Coboldia fuscipes is attracted to flowers of Ceropegia stenantha, we tested antennae of this pollinating scatopsid species to floral scent samples. Antennae of male and female flies consistently responded to a total of 13 components. Among them were novel chemical compounds known so far only from flowers of C. stenantha. They were synthesized and, together with other electrophysiologically active floral components tested as synthetic standards in further electrophysiological measurements to confirm their biological activity.

Plant material
Ceropegia stenantha is a root-, stem-and leaf-succulent twiner with smooth and slender, lanceolate leaves (Meve 2002). This fairly small perennial is scrambling or twining on shrubs, rarely exceeding 1 m in height; typically it flowers in mass displays (Fig. 1a). The corolla (up to 3 cm in length) is faintly greenish-white with small yellowish lobes. The corolla lobe tips, which are apically fused to form a cage-like structure (Fig. 1a), together with the adaxial extensions of the corolla lobe bases limit the access to the tube (Fig. 1b). The corolla tube is only slightly curved, exceptionally narrow, and merges into an also very narrow basal inflation with relatively little space around the gynostegium (Fig. 1c). C. stenantha is distributed throughout (sub-)tropical eastern Africa from Sudan to South Africa, where it grows in grassland areas (savanna) and open woodlands (see Masinde 2004).

Volatile collection
Two types of flower scent samples, i.e., thermal desorption (TD) and solvent acetone (SAc), were collected from Ceropegia stenantha flowers using dynamic headspace methods . For collection of TD samples, three flowers (first day of anthesis) of one plant individual were enclosed singly in polyester oven bags (Toppits®, Germany) for 10 min. Accumulated volatiles were subsequently pulled from the bag through a small adsorbent tube filled with a mixture of 1.5 mg Tenax-TA (mesh 60-80) and 1.5 mg Carbotrap B (mesh 20-40) (both Supelco, Bellefonte, PA, USA; Heiduk et al. 2015) for 5 min using a membrane pump (G12/01EB, Rietschle Thomas Inc., Puchheim, Germany) with the flow rate adjusted to 200 ml/min. An ambient air sample was taken in a similar way and used as a control to specify floral volatiles in the samples taken from flowers. For collection of the SAc sample, the same three individual flowers were sampled for 4 h each at a flow rate of 100 ml/min using bigger adsorbent tubes filled with 15 mg Tenax-TA and 15 mg Carbotrap B (Heiduk et al. 2015). These tubes were then eluted with 60 µl of acetone (SupraSolv, Merck KgaA, Germany) each, and the three elutes were pooled to obtain a single SAc for electrophysiological measurements and chemical synthesis (see below).
The TD samples were already used in a comparative study on Ceropegia flower scent ) and re-analyzed in the present study for comparison to the SAc sample. We here show semi-quantitative data obtained by re-analyzing the TD samples.

Chemical analyses (GC/MS)
To identify electrophysiologically active components (see below), 1.0 µl of the SAc sample was analyzed by gas chromatography/mass spectrometry (GC/MS) using a Shimadzu GCMS-QP2010 Ultra equipped with an AOC-20i auto injector (Shimadzu, Tokyo, Japan) and a Zebron ZB-5 fused silica column (5% phenyl polysiloxane; length: 30 m, inner diameter: 0.32 mm, film thickness: 0.25 µm, Phenomenex). One µl of the sample was injected (injection temperature: 220 °C; split ratio 1:1) and the column flow (carrier gas: helium) was set at 3 ml/min. The GC oven temperature was held at 40 °C for 1 min, then increased by 10 °C/min to 220 °C and held for 2 min. The MS interface worked at 220 °C. Mass spectra were taken at 70 eV (in electron ionization mode) from mass-to-charge ratio (m/z) 30 to 350, and data were processed using the GCMSolution package, Version 2.72 (Shimadzu Corporation 2012). Flower scent components of the GC/MS spectra were identified using the mass spectral data bases (for details see Heiduk et al. 2017).
A total of 22 runs were performed with seven male and seven female fly individuals (one antenna per insect). Five male and three female antennae were tested to both the floral SAc sample and the synthetic mixture, three female antennae and one male antenna were tested only to floral scent, and one antenna of both sexes was tested only to the synthetic mixture.
Measurements were performed with an Agilent 7890A (Santa Clara, California, USA) gas chromatograph. The GC was equipped with a flame ionization detector (FID), an EAD setup (heated transfer line, 2-channel USB acquisition controller; provided by Syntech; Kirchzarten, Germany), and a Zebron ZB-5 column (5% phenyl polysiloxane; length: 30 m, inner diameter: 0.32 mm, film thickness: 0.25 µm, Phenomenex). For each run, 1.0 µl of the sample was injected in splitless mode (injector temperature: 250 °C; oven temperature: 40 °C). The split opened 30 s after injection, and the oven heated by 10 °C/min up to 220 °C. The column was split at the end by a µFlow splitter (Gerstel, Mühlheim, Germany; nitrogen was used as make-up gas) into two deactivated capillaries, one (length: 2 m, inner diameter: 0.15 mm) leading to the FID setup, the other (length: 1 m, inner diameter: 0.2 mm) leading to the EAD setup. The outlet of the EAD was placed in a cleaned, humidified air flow directed over the fly antenna.
Flies used for measurements were anesthetized (CO 2 ) and their heads cut off. Two glass micropipette electrodes were filled with insect Ringer's solution (8.0 g/l NaCl, 0.4 g/l KCl, 4.0 g/l CaCl 2 ) and connected to silver wires. The caudal side of the head was connected to the reference electrode, and the recording electrode was placed in contact with the last flagellomere. The very tip of the flagellomere had to be cut Fig. 2 Synthesis of 3-acetyloxy-1-phenylbutan-2-one and 3-acetyloxy-4-phenylbutan-2-one because antennae were densely covered with mitrotrichiae which hampered immersion of the antennae into the Ringer's solution.
The total amount of scent emitted from flowers as well as the electrophysiological activity of floral volatiles in male and female Coboldia fuscipes (Scatopsidae) flies is also reported. KRI: Kovats retention index; EAD: electroantennographic detection; tr: trace amount < 0.05 Numbers in bold: whenever total amounts of individual compounds are > 5% S compound verified through authentic standard a compound synthesized in present work for having an authentic standard available *tautomeric form of 1-phenyl-2,3-butanedione

Identification of novel chemical components
The structures of the two compounds 3-acetyloxy-4-phenylbutan-2-one and 3-acetyloxy-1-phenylbutan-2-one, previously not reported as natural products, were deduced from their mass spectra (Fig. 3). The spectrum of the latter indicates a benzyl group by the ion m/z 91 and at least one acetyl-group (m/z 43). The molecular ion at m/z 206 loses 42 or 60 amu fragments, consistent with an acetyloxy-group. An acetyloxy-group at C-2 of the chain leads to an ion at m/z 87, and the related ion m/z 115 points to a neighboring carbonyl group. Connecting these fragments leaves 3-acetyloxy-1-phenylbutan-2-one as the sole likely structure.
The second compound showed similarities, such as m/z 91 (benzyl group) and 146 (loss of acetic acid), with the first spectrum. Although the molecular ion is missing, the loss of 43 amu indicates a methylketone fragment. We therefore proposed 3-acetyloxy-4-phenylbutan-2-one as the structure for this compound. Both compounds are acylated acetoins, which easily undergo internal redox reaction resulting in switching of the positions of the keto-and hydroxy-groups. Therefore, the co-occurrence of both compounds is very often observed in nature.
These structural assignments were confirmed by co-injection of floral samples and authentic material obtained by synthesis as described in the experimental section.

Discussion
Ceropegia stenantha flowers were found to emit a floral scent composed of 23 components of which 19 were already detected by Heiduk et al. (2017). Thereof, 12 aromatic components were identified and verified through authentic standards, for some of which the synthesis is described in the present paper. In electroantennographic measurements, 13 out of 23 floral scent components were consistently perceived by male and female Coboldia fuscipes flies ( Fig. 4; Table 1). Among these compounds were 3-acetyloxy-4-phenylbutan-2-one and 3-acetyloxy-1-phenylbutan-2-one, which are not known from another organism than C. stenantha (see also Heiduk et al. 2017).

Fly pollinators and floral scent chemistry
Ceropegia stenantha flowers are exclusively visited and pollinated by Scatopsidae, and we assume that this highly specific attraction is achieved by floral chemistry. Chemical attraction of fly pollinators is assumed for Ceropegia species in general (Vogel 1961;Heiduk et al. 2017) but was thus far only confirmed in two species (Heiduk et al. 2015(Heiduk et al. , 2016. Our electrophysiological measurements with Coboldia fuscipes (Fig. 4), which is one out of six scatopsid species identified as C. stenantha pollinators , point towards an important role of floral volatiles in pollinator attraction. As suggested for Ceropegia species in general (Vogel 1961), we assume that C. stenantha has a deceptive pollination strategy and it might either mimic a food source, oviposition site, or pheromones (see Vogel 1961;Heiduk et al. 2017).
Deduced from the chemical nature of most floral compounds identified in C. stenantha, we doubt that they chemically resemble female oviposition sites. The chemical profile of C. stenantha does not fit to what is described as typical oviposition site mimicry (Jürgens et al. 2013) and the aromatic-based floral scent is rather sweet smelling, whereas oviposition site mimics generally emit unpleasant scents. Furthermore, the high abundance of male individuals present inside the flowers (see above) is again intriguing in this regard; although it is described in other flies that males are attracted to oviposition sites for mating purpose (see Maier and Waldbauer 1979;Papaj 1994;Bonduriansky and Brooks 1999).

Mimicry strategy of Ceropegia stenantha
The occurrence and bioactivity of above mentioned components suggests that they signal a food source (e.g., nectar, pollen) to attract pollinators. Flowers of C. stenantha do not offer such rewards but may attract food-seeking Scatopsidae to its trap flowers by emitting reward-indicating aromatic components commonly found in rewarding flowers. Scatopsidae are known to visit such flowers to feed on pollen and/or nectar (Apiaceae: Haenni 1997; Asteraceae : Haenni 1990;Honěk et al. 2016;Dipsacaceae: Haenni 1984; Rubiaceae: García-Robledo and Mora 2007) and probably use chemical cues to find them. Food deception through signals that indicate a reward/rewarding species is a well-studied strategy with different facets, e.g., Batesian mimicry and generalized food deception (see Johnson and Schiestl 2016). However, generalized deception as well as mimicry of nectar and pollen mainly relies on visual cues (for review see Johnson and Schiestl 2016). Though C. stenantha flowers might contrast well against the background and thus be salient for flies (Fig. 1a, b), they do not seem to visually resemble a rewarding flower of another plant species. If C. stenantha would mimic a food source by emitting likely reward related volatiles, it is still suspicious that only Scatopsidae are found inside the trap flowers.
Scatopsidae often aggregate by hundreds of individuals (male and female) at specific sites not in relation with larval development sites (Fritz 1984;Haenni 2002;Nielsen 2009); however, mating behavior is frequent in these aggregations. The mechanisms triggering this aggregation behavior are unknown. C. stenantha flowers attract male and female Scatopsidae of different species ; resolving the chemistry behind the attractiveness of the flowers might help to understand the aggregation behavior in Scatopsidae. It has been shown, for example, in Drosophilidae that males and females aggregate using pheromone components which are attractive to both sexes and not necessarily species-specific (Bartelt et al. 1986). Likewise, sexual behavior in Drosophila flies is connected to activation of food-sensing neurons which helps males to find females (Grosjean et al. 2011). These neurons are activated by phenylacetaldehyde, a compound also found in C. stenantha flower scent. Under the assumption that Scatopsidae have a similar neuronal circuitry, we could hypothesize that C. stenantha uses a mixed strategy: emission of aromatic components that are related to a food source (see above) in combination with the novel components which might have a pheromone function. Solvent extracts (dichlormethane, hexane, pentane) as well as TD samples of male and female Coboldia fuscipes did not contain 3-acetyloxy-4-phenylbutan-2-one and 3-acetyloxy-1-phenylbutan-2-one, nor any of the other components identified in C. stenantha (Heiduk, unpublished data). However, C. fuscipes is only 1 of 13 scatopsid fly species attracted to C. stenantha flowers and not described from the plant's native habitat. To exclude sexual mimicry as pollination strategy in C. stenantha, extracts of pollinating scatopsid species, especially from the plant's native range, need to be analyzed for the presence of the components discussed here.

Outlook
To clarify which of the electrophysiologically active components presented in this paper finally attract the flies to C stenantha flowers, field bioassays with different mixtures of synthetic components are needed. As shown for C. dolichophylla and C. sandersonii, only small subsets of the electrophysiologically active components may successfully attract the flies (Heiduk et al. 2015(Heiduk et al. , 2016. As already suggested by Heiduk et al. (2017), the selective attractiveness of Scatopsidae to C. stenantha might be achieved through the novel flower compounds 3-acetyloxy-4-phenylbutan-2-one and 3-acetyloxy-1-phenylbutan-2-one. Our electroantennographic measurements with scatopsid flies now supported this assumption and also indicated the importance of other 1 3 scent components in this pollination system. Further studies will reveal which of the electrophysiologically active floral components attract the flies. Once the specifically attractive components are identified, their role in the life of scatopsid flies needs to be clarified to understand which sensation floral scent of C. stenantha creates in the scatopsid flies.