Introduction

The pale grass blue Zizeeria maha (Kollar) (Lepidoptera: Lycaenidae) is a small butterfly distributed widely throughout Asia, from Japan (except Hokkaido Island in the north) to southern Iran. Z. maha larvae feed on a limited number of plant species of the genus Oxalis in the family Oxalidaceae. Out of eight wild and naturalized species of Oxalis in Japan (Shimizu 1982), two species including three formae, namely Oxalis corniculata L., O. corniculata L. f. atropurpurea (Planch.) Van Houtte ex Hegi, O. corniculata f. erecta Makino, O. corniculata L. f. rubrifolia (Makino) H. Hara and O. dillenii Jacq., are described as host plants for Z. maha (Shirouzu 2006).

Oxalis species are common perennial plants that can be found as roadside weeds and on the margin of cultivated fields. The species are characterized by the biosynthesis and accumulation of oxalic acid in leaves (Shimizu 1982). Owing to its acidity and ability to bind dibasic cations such as calcium and magnesium, oxalic acid is toxic to mammals and causes hypocalcemia (Lynn 1972), kidney stone formation (Gershoff and Prien 1967) and chronic food poisoning of sheep (Seddon and Ross 1929). In insects, oxalic acid is an antibiotic factor that inhibits larval growth of Helicoverpa armigera (Lepidoptera: Noctuidae) (Yoshida et al. 1995) and is a sucking inhibitor of the brown planthopper, Nilaparvata lugens (Homoptera: Delphacidae) (Yoshihara et al. 1980). Therefore, oxalic acid can be regarded as a defensive chemical for both mammal and insect grazing. However, Z. maha evidently evolved a mechanism to overcome this defensive barrier.

In this study, we examined the effects of host plant extracts on feeding responses of Z. maha larvae in order to gain insights into the mechanism by which oxalic acid toxicity is avoided.

Materials and methods

Insects

Adults and larvae of Z. maha were collected on the campus of University of Tsukuba, Tsukuba, Ibaraki, Japan. The adult butterflies were fed with 7 % sucrose solution once per day and kept under laboratory conditions with the host plant and allowed to lay eggs. Newborn and field-collected larvae were reared on fresh host plants (Oxalis corniculata) until the molt for the third instar in an incubator (BioTRON, model LH120S, NK Systems, Osaka, Japan) under a 16L:8D photoperiod at 25 °C and 60–70 % relative humidity.

Plant material, extraction and fractionation

The leaves of O. corniculata were collected from full sun and partial sun areas at the campus of University of Tsukuba, Tsukuba, Ibaraki, Japan in August 2014. Larvae of Z. maha do not usually feed on petioles and peduncles; thus, the fresh trefoil leaves of O. corniculata (247.5 g) were extracted twice with 1.6 l of 70 % ethanol for 2 weeks at 25 °C. Combined extracts were evaporated to dryness under an aspirator vacuum by a rotary evaporator (water bath temperature less than 40 °C). The residue (13.32 g) was dissolved in 350 ml of 70 % methanol and stored as a stock solution at −25 °C. For fractionation, one-tenth (35 ml) of the stock methanolic solution was dried up and resuspended in 100 ml distilled water, then successively extracted with diethyl ether (100 ml, three times) and ethyl acetate (100 ml, three times) as shown in Fig. 1. Organic layers were dried over sodium sulfate, and solvents were evaporated, weighed and stored in a freezer (−25 °C). The water layer was evaporated, weighed and stored as an aqueous solution in a refrigerator (4 °C).

Fig. 1
figure 1

Procedure for separation of feeding stimulants for Z. maha larvae from the crude methanolic extract of O. corniculata leaves. Extracts and fractions in boxes were subjected to bioassays

Full size image

Removal and re-addition of oxalic acid in the water layer

Oxalic acid in solution is known to precipitate as calcium oxalate upon addition of calcium ion. Given the low solubility of oxalic acid in water (0.67 mg/100 ml, 20 °C; Hagler and Herman 1973), addition of calcium chloride solution to a solution containing oxalic acid followed by filtration and weighing the calcium oxalate precipitate is one of the methods used historically to quantify the oxalic acid content in plant materials (Baker 1952). We used this method to remove the oxalic acid that was expected to be present in the water layer. In advance of the addition of calcium chloride (Kanto Chemical Co., Inc., Tokyo, Japan), the concentration of oxalic acid in the water layer was estimated by gas chromatography after derivatization of an aliquot. To a 100 ml aliquot of the water layer containing 24.75 g leaf equivalent (g.l.e.) of extract, calcium chloride (0.69 g, 6.2 mmol) was added portion-wise while swirling manually. The solution was left overnight and filtered through 1 μm filter paper (No. 4, Kiriyama Glass Works Co., Tokyo, Japan). The obtained filtrate, with oxalic acid removed from the water layer, was divided into two portions for use in the feeding tests: one was subjected to the feeding test unmodified (OA removed); to the other portion 0.28 g (3.1 mmol) oxalic acid (anhydrous, Wako Pure Chemical Industries, Ltd., Osaka, Japan) was dissolved in the solution before the feeding test (OA re-added). Larval feeding activity with the OA removed and OA re-added water layers was compared with the artificial diets prepared as described below. In addition, pH values of the water layer before and after addition of calcium chloride were measured, and the effect of hydrogen chloride liberated by OA removal was tested against Z. maha larvae with an artificial diet containing hydrochloric acid (6.2 mmol) in the control diet as described below.

Artificial diet for feeding tests

A control diet was prepared by mixing 5.0 g of commercially available artificial diet (Insecta F-II, Nosan Corp. Life-Tech Department, Yokohama, Japan), 2.5 g agar powder (Wako Pure Chemical Industries, Ltd., Osaka, Japan) and 17.25 ml distilled water (total weight: 24.75 g). The mixture was heated to near boiling with a microwave, then placed in a refrigerator and used as a control diet.

Test diets were prepared by addition of crude methanolic extract or fractions from O. corniculata as aqueous solutions (or suspensions) to the artificial diets. For example, 24.75 g.l.e. crude methanolic extract (35 ml of the stock solution), after being dried up, was resuspended in 17.25 ml water and mixed with 5.0 g Insecta F-II and 2.5 g agar powder to prepare a test diet of the crude methanolic extract at a concentration of 1.0 g.l.e./g of diet.

In order to test feeding responses of Z. maha larvae to artificial diets containing various amounts of oxalic acid, oxalic acid was dissolved in 17.25 ml water and mixed as above to prepare the test diets in the range of 65–1040 μmol/g (corresponding to 0.25–4.0 g.l.e./g) of diet.

Feeding test

Bioassays for feeding activity of chewing insects can be set up by combinations of substrates (filter paper, artificial diet, plant tissues, etc.) and types of data collection (weight or area consumed, number of chewing marks, weight gain of the test insect, number or weight of feces pellets produced, direct and/or indirect observation, electrophysiological responses, etc.) (Hare 1998; Lewis and Van Emden 1986). Depending on the nature of the feeding habit of the test insect, not all of these experimental components can be applied. Preliminary tests revealed that Z. maha larvae were unable to chew neutral substrates such as filter papers or other cellulose-based thin discs. Therefore, direct measurements of the amount ingested by measuring the area consumed or by counting chewing marks were not applicable for our feeding assay. Trials with artificial diets containing crude host plant extract showed that neonate to second instar larvae were unable to feed on the artificial diets presumably because their mandibles were too small. On the other hand, the 3rd instar larvae immediately after molting, when the dorsal nectary organ was observed, were able to continue feeding on the artificial diet. With this success, we set up our feeding test to identify stimulants responsible for continuous feeding by using a combination of artificial diet as substrate and frass measurement as data collection in a no choice situation. One group of three third instar larvae was placed in a plastic petri dish (90 mm diameter, 15 mm height) lined with moistened filter paper. Five groups were used as replicates for a sample. Each group was offered one to four pieces of diet (30 mg per piece) and kept in an incubator as described above. Frass pellets were collected daily from each group, dried at 60 °C until constant weights were obtained and weighed to evaluate the feeding activities. In the test with OA removed and OA re-added water layer, ten third instar larvae were tested individually. All larval feeding tests were continued until pupation (20–25 days).

Statistics

Statistical analyses were performed with the EZR statistical software (Easy R, version 1.29; Kanda 2013).

Chemical analyses

Aliquots of the crude methanolic extract and fractions obtained by liquid-liquid partitioning were analyzed by gas chromatography-flame ionization detection (GC-FID) and by gas chromatography-mass spectrometry (GC-MS) as tert-butyldimethylsilyl (TBDMS) or trimethylsilyl (TMS) derivatives (Knapp 1979). Samples were placed in glass conical vials (GL Sciences, Inc., Tokyo, Japan), dried and reacted with a mixture of 50 μl N-methyl-N-tert-butyldimethylsilyltrifluoroacetamide (MTBSTFA; Sigma-Aldrich Co. LLC., St Louis, MO, USA) and 50 μl acetonitrile at 60 °C for 1 h. For TMS derivatives, N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA, Sigma-Aldrich) was used instead of MTBSTFA in the above reaction.

GC-FID was conducted on a HP6890 gas chromatograph (Hewlett-Packard, Palo Alto, CA, USA) equipped with a fused silica capillary column (HP-5MS, 30 m × 0.25 mm, 0.25 μm film thickness). Samples were injected in the splitless mode (sampling time: 1 min) at an injection port temperature of 280 °C. Helium was used as the carrier gas at 1 ml/min in the constant flow mode. The oven temperature was set at 50 °C for 1 min, then raised to 280 °C at 10 °C/min and held at the final temperature for 10 min. A flame-ionization detector (FID) was operated at 280 °C, and the chromatograms were analyzed with HP ChemStation software.

Mass spectra were obtained by GC-MS. Samples were injected into a HP6890N gas chromatograph operated in the same condition as GC-FID, except that the column (DB-5MS, 25 m × 0.25 mm, 0.25 μm film thickness) outlet was introduced at 280 °C into a MS-600H mass spectrometer (JEOL, Tokyo, Japan). The temperature of the ionization chamber was 190 °C, and the ionization was performed in the electron impact mode at 70 eV. The data were acquired in scan mode (scan range 40–600 amu, scan speed 0.29 s) and analyzed with TSS2000 software (version 2.00, Shrader Analytical and Consulting Laboratories Inc., Detroit, MI, USA).

Results

Feeding responses of Z. maha larvae to plant extracts and fractions

Larval feeding was significantly stimulated by the addition of the crude methanolic extract to the artificial diet, indicating the presence of feeding stimulant(s) in the extract. The mean dry weight of the frass obtained from larvae fed with a diet containing the crude methanolic extract was 2.2 times higher (mean ± SE: 36.8 ± 7.70 mg; N = 5, p < 0.05, t test) than that obtained from the control treatment (16.7 ± 2.80 mg; N = 5). After fractionation of the methanolic extract into three layers, significantly enhanced feeding activity was restricted to the water layer only. Namely, a significantly larger amount of frass was produced by the larvae fed with an artificial diet containing the water layer (78.7 ± 6.41 mg; N = 5, p < 0.05, Steel test against the control, two-tailed; Fig. 2). In contrast, larvae fed with the artificial diet containing the ether layer produced much less frass (5.2 ± 1.59 mg; N = 3), although the difference was marginally significant (p = 0.07, Steel test against control, two-tailed).

Fig. 2
figure 2

Feeding responses of Z. maha larvae to a control artificial diet and to artificial diets containing O. corniculata leaf extract. Each bar represents the dry frass weight (mean ± SE, mg per larva, N = 3–5). *p < 0.05 (Steel test against the control, two-tailed)

Full size image

Chemical analyses

After TBDMS derivatization, GC-MS analysis of the active water layer yielded a major peak (63.2 % of the total peak area) at a retention time of 13.46 min (Fig. 3). The mass spectrum of this compound showed characteristic ions of bis(tert-butyldimethylsilyl) oxalate at m/z 303 (relative intensity: 2.0 %, M+-15), 261 (53.6 %, M+-57), 189 (15.1 %), 147 (67.1 %) and 73 (100 %). By comparisons of retention times and mass spectra using both TBDMS and TMS derivatives of an authentic compound, the major compound in the water layer was identified as oxalic acid.

Fig. 3
figure 3

Typical total ion current chromatograms obtained by TBDMS derivatization of the blank control (MTBSTFA in acetonitrile; (a), standard oxalic acid (b) and water layer (c). Chromatograms were obtained with a HP6890N gas chromatograph (Hewlett-Packard, Palo Alto, CA, USA) equipped with a fused silica capillary column (DB-5MS, 25 m × 0.25 mm, 0.25 μm film thickness), programed from 50 °C for 1 min, then raised to 320 °C at 10 °C/min and held at the final temperature for 12 min. The arrows indicate the peaks of oxalic acid bis-TBDMS ester

Full size image

Oxalic acid in the methanolic extract was quantified as the TBDMS derivative by using GC-FID against a known concentration of bis-TBDMS oxalate prepared with authentic oxalic acid. Fifteen samples of the methanolic extract yielded oxalic acid ranging from 2.9 to 9.9 % (mean ± SE 7.2 ± 2.7) of the dry weight of O. corniculata leaves.

Effects of removal and re-addition of oxalic acid in the water layer on larval feeding

A significantly lower mean frass weight (15.0 mg/larva) was obtained from larvae fed with the OA-removed artificial diet, whereas 1.75 times higher weight of frass (26.3 mg/larva) was produced by larvae fed with the OA re-added artificial diet (p < 0.01, t test, two-tailed; Fig. 4). Upon addition of calcium chloride to the water layer, the pH value slightly decreased from 1.4 to 1.3. The mean dry frass weight obtained from the larvae fed with the artificial diet containing hydrochloric acid was 12.4 mg per larva, which was not significantly different from that of the control.

Fig. 4
figure 4

Feeding response of Z. maha larvae to the removal and re-addition of oxalic acid in the active water layer. Each bar represents the dry frass weight (mean ± SE, mg per larva, N = 8 for OA removed; N = 9 for OA re-added). *p < 0.01 (t test)

Full size image

Feeding responses of Z. maha larvae to artificial diets containing various amounts of oxalic acid

Dry frass weights were increased with increasing concentrations of oxalic acid in the artificial diets and reached the maximum (59.5 mg, p < 0.05, Steel test against the control, one-tailed) at 1.0 g.l.e. (260 μmol) of oxalic acid per gram of artificial diet. However, addition of oxalic acid at 520 μmmol/g and higher resulted in less frass production compared with that of the control (520 μmol/g: 9.60 mg, 780 μmol/g: 0.17 mg, 1040 μmol/g: 0.08 mg; Fig. 5).

Fig. 5
figure 5

Feeding response of Z. maha larvae to artificial diets containing various amounts of oxalic acid. Each bar represents the dry frass weight (mean ± SE, mg per larva, N = 2–5). *p < 0.05 (Steel test against the control, one-tailed)

Full size image

Discussion

In this study, we characterized oxalic acid for the first time as a lepidopteran larval feeding stimulant for Z. maha from its major host, O. corniculata.

Oxalic acid was one of the first acids identified and was discovered in wood sorrel (Oxalis acetosella L.) by Johann Christian Wiegleb in 1796 (Vaneker 2015). The free acid and its salts have been found in a variety of organisms, such as plants, animals including humans, fungi and microorganisms (Vaneker 2015). More than 215 plant families are known to accumulate calcium oxalate in their tissues (McNair 1932), indicating that oxalic acid is ubiquitously distributed in the plant kingdom. In Oxalis, oxalic acid comprising some 16 % of the dry weight appears to be present in vacuoles largely as the free acid (Ranson 1965). Levels of oxalates in O. corniculata are reported to be 4.1 % soluble and 7.0 % total oxalate (Libert and Franceschi 1987). The present data on the oxalic acid content of O. corniculata are consistent with the above descriptions both qualitatively and quantitatively. Oxalic acid was present in the biologically active water layer mainly as the free acid; hence, it was easily derivatized to its bis-TBDMS ester and detected in the range of 2.9–9.9 % (dry weight of leaves). Furthermore, upon addition of CaCl2 to the water layer, which is one of the well-established procedures for quantitative analysis of oxalic acid (Hodgkinson 1970), the water layer became turbid and fine precipitates of calcium oxalate were obtained. After removal of oxalic acid as calcium oxalate from the active water layer, the filtrate showed reduced stimulation of feeding activity (mean dry frass weight; 15.0 mg/larva), which was comparable to the control diet. Re-addition of oxalic acid to the filtrate significantly enhanced the feeding activity (26.3 mg/larva; Fig. 4). These results supported our hypothesis that oxalic acid elicits a feeding response by Z. maha larvae. Finally, the results of dose-response feeding bioassays with authentic oxalic acid (Fig. 5) led us to the conclusion that oxalic acid is a feeding stimulant of Z. maha larvae.

Although the outcome of feeding experiments is dependent upon their design (e.g., not all experiments offer a wide range of doses of a single stimulant), feeding responses in dose-response experiments may be classified into two types. The first type represents a typical sigmoidal dose-response curve, as in the case of the olive weevil to (−)-olivil (Kadowaki et al. 2003). The second type shows a bell-shaped dose-response curve with a peak followed by a decrease with increasing doses of a stimulant (David and Gardiner 1966; Hicks 1974; Yamamoto and Fraenkel 1960). Z. maha larvae showed the latter type of feeding response toward the increasing dose of oxalic acid in the artificial diet (Fig. 5). A declining response at a higher dose may be simply explained as a toxic effect of excessive amounts of the stimulant because the larvae stopped feeding and eventually died. On the other hand, such a decrease in feeding response may be the result of aversive post-ingestive feedback to excess stimuli (Provenza 1995). By such post-ingestive feedback, Z. maha larvae may be able to select leaves that contain a moderate, adequate concentration of oxalic acid in the field. The chemical composition of plants often changes with age, exposure to sunlight or other environmental factors (Flück 1963; Hemming and Lindroth 1999). Thus, oxalic acid may vary in concentration among individual host plants or among leaves of an individual plant. Plants grown under a high light environment tend to invest surplus carbon (excess for growth) in secondary metabolites. If the light condition is positively correlated with the concentration of oxalic acid in Oxalis leaves, the observation that larvae occur more frequently on host plants growing in light shade (personal observation by MY) may reflect a preference of Z. maha larvae for a moderate concentration of oxalic acid in the leaves. Investigation of the triadic relationships among the light condition, oxalic acid concentration in Oxalis leaves and feeding response of Z. maha larvae is ongoing in our laboratory.

One of the primary functions of oxalic acid in plants is considered to be chemical defense against herbivores. Oxalic acid has been reported as a sucking inhibitor of the brown planthopper, Nilaparvata lugens (Homoptera: Delphacidae) (Yoshihara et al. 1980). Yoshida et al. (1995) tested oxalic acid as a part of the resistance mechanism of chick pea, Cicer arietinum L., to Helicoverpa armigera (Lepidoptera: Noctuidae) and found that after 10 days the larval weight was reduced to 53 % of control when the larvae were fed a semiartificial diet containing 250 μmol/g oxalic acid. Likewise, a number of oxalate-induced impacts on animals, including humans, have been reported (Duncan et al. 1997; Lynn 1972). These include chronic poisoning of sheep (Bull 1929) and human death due to ingestion of the oxalate-containing plant, Rumex crispus (Reig et al. 1990). In mammalian species, LD50 values of oxalic acid for rats were determined to be 475 mg/kg for males and 375 mg/kg for females by single oral administration (Vernot et al. 1977). By the same method, they reported the LD50 values of allyl isothiocyanate (mustard oil, released by enzymatic action on sinigrin) and nicotine sulfate as 490 and 75 mg/kg, respectively. These data support that oxalic acid serves as a defense against herbivores. In the present feeding experiment, Z. maha larva ingested about 75 µmol (6.8 mg) oxalic acid during the test period (250–300 mg of artificial diet containing 0.25 mmol/g oxalic acid was consumed). Given that the mean body weight of third instar larvae was 5.6 mg at the beginning of the feeding test, it is surprising that a Z. maha larva was able to ingest a quantity of oxalic acid that exceeded its own body weight, which indicates that this insect is highly adapted to Oxalis as a specialist herbivore.

For specialist herbivores, so-called secondary plant metabolites, such as cyanide, cyanogenic glycosides, alkaloids, glucosinolates, terpenoids, coumarins and cardenolides, serve as positive cues (Rosenthal and Berenbaum 1991). Oxalic acid is no exception to this. Verschaffelt (1910) observed that the Polygonaceae-feeding leaf beetle Gastroidea viridula Deg. (Coleoptera: Chrysomelidae) gnawed the leaves of Lathryrus sylvestris (non-host) when the leaves had been immersed for some time in a solution of oxalic acid. Subsequently, Renner (1970) confirmed that the feeding of G. viridula was stimulated by oxalic acid. Matsuda and Matsumoto (1975) tested a variety of organic acids, including oxalic acid, against four species of Polygonaceae-feeding leaf beetles to analyze host plant specificity. The authors observed that the feeding of Gallerucida bifasciata Motschulsky (Coleoptera: Chrysomelidae) was stimulated by oxalic acid, as well as by malic, tartaric and citric acids. Studies of interactions between herbivorous insects and plant secondary metabolites have been conducted and have made noteworthy progress, especially in three families of butterflies, namely Papilionidae (Feeny 1991; Honda et al. 2011; Li et al. 2010; Murata et al. 2011; Nakayama et al. 2003; Nishida et al. 1987; Nishida and Fukami 1989a, b; Ono et al. 2000), Pieridae (Honda et al. 1997; Huang et al. 1993; Renwick and Lopez 1999; Renwick and Radke 1983) and Nymphalidae (Ackery 1988). On the other hand, the chemical ecology of the lycaenid-plant interaction remains almost unexplored (Fiedler 1996) with the remarkable exception of the cycasin-sequestering aposematic lycaenid Eumaeus atala florida (Röber) (Rothschild et al. 1986). Considering that lycaenid larvae show phytophagy and entomophagy, the family Lycaenidae, which is the second largest butterfly taxon with over 4500 species (Shields 1989) and possibly 6000 species (Robbins 1988), is a complex yet fascinating group of butterflies on which to study the chemical basis of host selection.

The fate of oxalic acid ingested by Z. maha larvae is not known. Specialist herbivores are demonstrated to metabolize, catabolize (detoxify) and/or sequester plant defensive secondary metabolites (Ali and Agrawal 2012; Malcolm and Brower 1989). Preliminary analysis of Z. maha fecal extracts for the presence of free oxalic acid resulted in detection of much smaller amounts of the acid than those ingested (unpublished data). Sequestration of intact oxalic acid in the body seems negative. These findings imply that oxalic acid may be catabolized (degraded) and excreted in feces or metabolized (transformed) as a carbon source in Z. maha larvae. The synthesis of oxalate-degrading enzymes, such as oxalate oxidase (EC 1.2.3.4; Datta et al. 1955) and oxalate decarboxylase (EC 4.1.1.2; Jakoby et al. 1956), is known from plants and bacteria, but, to the best of our knowledge, no information is available on the production of these enzymes in insects. Given that oxalotrophic bacteria (capable of utilizing oxalates as their sole or major carbon and energy source) have been identified from Oxalis (Sahin 2003, 2005), the involvement of these bacteria in the digestive tract of Z. maha larvae is possible. Ongoing investigations of the fate of oxalic acid in Z. maha larvae may provide insights into oxalate toxicosis.