Caffeine fostering of mycoparasitic fungi against phytopathogens

Abstract

Caffeine (1,3,7-trimethixanthine) is a typical purine alkaloid produced in more than 80 plant species. Its biological role is considered to strengthen plant's defense capabilities, directly as a toxicant to biotic attackers (allelopathy) and indirectly as an activator of defense system (priming). Caffeine is actively secreted into rhizosphere through primary root, and possibly affects the structure of microbe community nearby. The fungal community in coffee plant rhizosphere is enriched with particular species, including Trichoderma family, a mycoparasite that attacks and kills phytopathogens by coiling and destroying their hyphae. In the present study, the caffeine response of 8 filamentous fungi, 4 mycoparasitic Trichoderma, and 4 prey phytopathogens, was examined. Results showed that allelopathic effect of caffeine on fungal growth and development was differential, being stronger on pathogens than on Trichoderma species. Upon confronting, the prey immediately ceased the growth, whereas the predator continued to grow, indicating active mycoparasitism to have occurred. Caffeine enhanced mycoparasitism up to 1.7-fold. Caffeine thus functions in a double-track manner against fungal pathogens: first by direct suppression of growth and development, and second by assisting their natural enemy. These observations suggest that caffeine is a powerful weapon in the arms race between plants and pathogens by fostering enemy's enemy, and we propose the idea of "caffeine fostering" as the third role of caffeine.

Abbreviation

PDA

potato dextrose agar

Introduction

One of the unique features of plants among living organisms is the production of a variety of chemical compounds with relatively low molecular weight. These chemical compounds are referred to as secondary metabolites, and to date, over 200,000 compounds have been documented from the plant kingdom.¹ Secondary metabolites constitute an average of 2–4% of the fixed carbon of photosynthetic productivity, and are partly secreted from root through active process.² Their biological role in rhizosphere was proposed to function in nutrient acquisition, plant growth regulation, plant-microbe association and determination of microbial community structure.^3,4 Referring to interaction with microbes, root exudates serve as attractants and stimulants, and also as inhibitors and repellants against pathogens.⁵

The most popular secondary metabolite is perhaps caffeine (1,3,7-trimethylxanthine), which is typically produced in coffee and tea plants.⁶ Caffeine was first isolated as a chemical compound by a German chemist, F. Runge in 1819.⁷ Since then, pharmaceutical function of caffeine has intensively been studied, and showed that it strongly inhibits phosphodiesterase activity, leading to acceleration of cAMP signaling pathways.⁸ As to the physiological function of caffeine in plants, 2 roles have so far been proposed.⁹ First, caffeine directly restricts development and growth of other organisms, including bacteria, fungi, insects, mollusks and plants by its toxicity.¹⁰ This is often referred to as allelopathy.¹¹ Second, caffeine indirectly stimulates plant defense response by affecting signaling pathways.¹² This is referred to as priming.^9,13 However, as caffeine is efficiently produced and widely accumulated in and out of the caffeine-producing plants, additional function was conceivable. One clue is that root of plants as coffee actively secretes caffeine into culture medium, resulting in an average concentration of ∼0.03%.¹⁴ Tea seedlings also secrete caffeine up to 0.05% a day.¹⁵ Actual concentration of caffeine, however, was supposed to be gradient in rhizosphere, being much higher in the immediate neighbors of the root.¹⁴

Caffeine released from the primary root was thought to help young seedlings to establish and condition the rhizosphere to create an optimum microbial and chemical environment.¹⁶ The idea was partially substantiated by a finding that rhizosphere of Coffea arabica in Ethiopian highlands was rich in species of filamentous fungus, Trichoderma.¹⁷ Diverse Trichoderma species were also found among coffee endophytic fungi that colonize healthy tissues.^18,19 Trichoderma population was enriched in coffee waste containing caffeine ∼0.1% in wet weight.^20,21 These observation suggest a specific relationship between caffeine and Trichoderma.

Trichoderma genus contains up to 104 species,²² and is characterized by a fast growing, strong spore production, and a source of cell wall degrading enzymes.²³ Many of its species have been known to attack and kill other fungi by coiling around hyphae, penetrating and subsequently dissolving the host cytoplasm.^24,25 This phenomenon is referred to as mycoparasitism and has been applied to biocontrol in agriculture.^26,27 Trichoderma species are commonly found as both rhizosphere and endophytic fungi, perhaps because they are attracted by fungal prey and root-derived nutrients.²⁵ Trichoderma is also known to induce plants' priming²⁷ by secreting metabolites such as proteins and oligosaccharides, which activate the defense system prior to pathogen attack.²⁸ Interaction between plant and Trichoderma is thus mutually beneficial.²⁶

In this study, caffeine effects on filamentous fungi, both Trichoderma species and their prey pathogens, were examined. Results showed that caffeine exhibits dual functions: differential suppression of fungal growth and development, and acceleration of mycoparasitism. Considering these findings and so far published documents, we conclude that plants benefit from fostering enemy's enemy through secreting caffeine, and propose the idea of caffeine fostering in plant-pathogen arms race.

Results

Caffeine concentration

Eight fungal species were cultured on 1/5 strength of potato dextrose agar (1/5 PDA) plates in the absence or presence of varied concentrations of caffeine, 0.01% (0.5 mM), 0.1% (5 mM) and 1% (50 mM). The growth rate was estimated by measuring the radius of each colony after 4 and 6 d of inoculation. Of all 8 species, growth was apparently not restricted by 0.01% caffeine, significantly affected by 0.1% caffeine and completely inhibited by 1% caffeine level. Representative cases from Fusarium oxysprum (prey) and Trichoderma atroviride (predator) are illustrated in supplementary figure (Fig. S1). Based on this result, caffeine concentration was fixed to 0.1% throughout the following experiments.

Growth inhibition

Each species was inoculated on a 1/5 PDA plate with or without 0.1% caffeine, and growth was periodically monitored by measuring radius of the growing colony. In either case, all species linearly grew showing straight lines on plotted growth curve until they reached the edge of the plate (see Fig. S2 and Fig. 2 for growth curve). Growth velocity was then calculated, expressed in increase of the radius per day and statistically evaluated using analysis of covariance (ANCOVA) (Table 1). The growth velocity in the presence or absence of caffeine was species-specific among pathogenic fungi. Rhizoctonia solani grew 15.2 mm/day in the absence of caffeine, and 6.5 mm/day in the presence of caffeine. Growth was reduced to less than half of the control by caffeine. We refer to this as the inhibition rate and express it in percentage (57%). F. oxysporum grew 6.8 mm/day in the absence of caffeine and 3.4 mm/day in the presence of caffeine, showing the inhibition rate of 50%. Sclerotinia sclerotiorum grew 9.3 mm/day and 2.9 mm/day, respectively. The inhibition rate was 69%. Glomerella cingulata grew 4.8 mm/day and 3.1 mm/day, respectively, with the inhibition rate of 35%. All four Trichoderma species grew approximately 12 mm/day in the absence of caffeine. In the presence of caffeine, however, their growth rate varied: T. atroviride grew10.2 mm/day, showing the inhibition rate of 17%; T. virens grew 8.2 mm/day with the inhibition rate of 32%; T. harzianum grew 8.8 mm/day (inhibition rate 27%), and T. hamatum grew 7.4 mm/day (inhibition rate 39%). In summary, the growth velocity in the presence of caffeine was around 3 mm/day in pathogens (except for R. solani) and over 7 mm/day in Trichoderma species. The finding points to that inhibitory efficiency of caffeine is differential, being generally low in Trichoderma and high in pathogenic species. This results in a relative acceleration of Trichoderma growth in comparison with pathogen growth in the presence of caffeine. In other words, caffeine may favor predators when they coexist with prey fungi.

Figure 2.

Time course of phytopathogen growth in the absence or presence of caffeine and/or mycoparasites. The prey growth in the absence of mycoparasite was estimated by measuring radius (mm) of colonies, and that in the presence of mycoparasites was estimated by the distance (mm) between confronting border and colony center (agar block) (confrontation radius) at indicated time points. (A) Growth of F. oxysporum in the absence (open circle) or presence (closed circle) of caffeine. Confrontation with T. virens in the absence (open triangle) or presence (closed triangle) of caffeine. (B) Growth and confrontation of R. solani. Figure structure is the same as (A) except for the mycoparasite was T. atroviride. (C) Growth and confrontation of S. sclerotiorum. Figure structure is the same as (A) except for the mycoparasite was T. atroviride. (D) Growth and confrontation of G. cingulata. Figure structure is the same as (A) except for the mycoparasite was T. atroviride.

Table 1.

Growth rate and caffeine effects

	Growth (radius, mm/day)
Species	Control	0.1% caffeine	Inhibition rate (%)	p-value
Rhizoctania solani	15.2	6.5	57	<0.0001
Fusarium oxysporum	6.8	3.4	50	<0.0001
Sclerotinia sclerotiorum	9.3	2.9	69	<0.0001
Glomerella cingulata	4.8	3.1	35	<0.0001
Trichoderma atroviride	12.2	10.2	17	<0.0001
Trichoderma virens	12.1	8.2	32	0.0020
Trichoderma harzianum	12.0	8.8	27	<0.0001
Trichoderma hamatum	12.1	7.4	39	0.0059

Each fungus was inoculated on a 1/5 PDA plate with or without 0.1% caffeine, and growth was periodically estimated by measuring colony radius. For each fungus, growth rate was plotted and the velocity (mm/day) was calculated from the slope (see Fig 2 and S1). Inhibition rate of caffeine is shown in percentage. The p-value was determined by ANCOVA, and indicates the difference between samples without (control) and with (0.1%) caffeine to be significant (p < 0.05) for all fungi.

Confrontation analysis

Interaction between the prey and the predator was examined by dual culture assay (confrontation assay) in the presence or absence of 0.1% caffeine (experimental design is shown in Fig. 1A). All combinations between 4 preys and 4 predators were essentially examined. One example, interaction between F. oxysporum (prey) and T. virens (predator), is illustrated in Fig. 1B. In the control plate, the prey steadily grew after inoculation until its hyphae reached the edge of the plate. In the presence of caffeine, it also steadily grew but with reduced velocity. When the prey was co-cultured with the predator in the absence of caffeine, interference contact occurred at day-4. The colony radius up to the confronting border (confrontation radius) of the prey was 21 mm from the colony center. When the prey was co-cultured with the predator in the presence of caffeine, contact occurred before day-7. The confrontation radius of the prey was 13 mm from the colony center. In both cases, upon contact, growth of the prey toward the predator ceased, whereas that of the predator continued. This indicates an active mycoparasitism to have occurred, and caffeine to have enhanced it. Another example, interaction between G. cingulata (prey) and T. atroviride (predator), is illustrated in Fig. 1C. The inhibition pattern of the prey growth by caffeine and/or the predator was similar to the case of F. oxysporum and T. virens. In the absence of caffeine, the prey fungus steadily grew with active formation of melanin pigment. In the presence of caffeine, the prey still steadily grew with reduced velocity, but melanin formation was completely suppressed. When co-cultured with the predator, the confrontation radius of the prey in the absence and the presence of caffeine were 14 mm and 11 mm from the colony center, respectively. In addition to enhancing predator's activity, caffeine also inhibited melanin formation even in the presence of the predator. Interactions between other preys and predators were also examined and are partially shown in the supplementary figures (Figs. S3, S4). Combinations were S. sclerotiorum/T. atroviride; R. solani/T. atroviride; F. oxysporum/T. harzianum; and G. cingulata/T. hamatum. All exhibited similar inhibition patterns as those shown in Fig. 1. A notable feature is that mycoparasitic activities among T. virens, T. harzianum, T. atroviride and T. hamatum were essentially similar. In the presence of caffeine, however, T. atroviride showed the highest activity, perhaps because of its high tolerance to caffeine (Table 1). In all cases, caffeine additively enhanced mycoparasitic activity of the predator.

Figure 1.

Effects of caffeine and/or mycoparasites on phytopathogens. (A) Experimental design of confrontation assay. The prey (pathogen) (right side) and predator (Trichoderma) (left side) were inoculated 5 cm apart on a 9.5 cm petri dish containing 1/5 PDA, and radius between the center of the colony and confronting border (confrontation radius) of the prey was measured at appropriate intervals. (B) F. oxysporum. Plates are fungus alone (Control, first panel from the left), with 0.1% caffeine (+Cf 0.1%, second panel from the left); confrontation with T. virens inoculated 2 d after the prey inoculation (+T. virens, second panel from the right); confrontation in the presence of caffeine with T. virens inoculated 2 d after the prey inoculation (+Cf 0.1% +T. virens, first panel from the right). Photographs were taken 4 d (D-4), 7 d (D-7) and 11 d (D-11) after prey inoculation. (C) G. cingulata. Figure structure is the same as shown in (B) except that confronted mycoparasite was T. atroviride.

Caffeine fostering

Interaction among preys, predators and caffeine was quantitatively evaluated from time course plots (Fig. 2). Three features were apparent. First, caffeine was effective to suppress fungal growth. Second, preys immediately discontinued to grow upon confrontation with predators. Third, caffeine accelerated the mycoparasitism. When F. oxysporum and T. virens were co-cultured in the absence of caffeine, both fungi independently and linearly grew until they confronted with each other at day-4. The radius of F. oxysporum to the confronting border (confrontation radius) was ca 21 mm, and scarcely increased thereafter up to day-11. In the presence of caffeine, confrontation occurred at day-5. The confrontation radius of the prey was 13.5 mm, which remained up to day-11 (Fig. 2A). The other combinations of prey and predator exhibited similar results. R. solani confronted with T. atroviride at day-2 in the absence of caffeine, and at day-3 in the presence of caffeine (Fig. 2B). S. sclerotiorum confronted with T. atroviride at day-4 in the absence of caffeine and at day-6 in the presence of caffeine (Fig. 2C). G. cingulata confronted with T. atroviride at day-2 and at day-5 in the absence and presence of caffeine, respectively (Fig. 2D). In all cases, the prey ceased growth after confrontation, and the confrontation radius did not change during following incubation period. The confrontation radius of prey was smaller in the presence of caffeine than in the absence of caffeine. The observation is summarized with statistical evaluation (Table 2). The average radius of F. oxysporum between colony center and confronting border against T. virens was 22.5 mm in the absence of caffeine, and 13.5 mm in the presence of caffeine. Prey growth was suppressed to 60% of the control sample in the presence of caffeine (inhibition rate of 40%). This indicates that mycoparasitism was accelerated 1.fold7- by caffeine (acceleration rate of 1.7-fold). The confrontation radii of R. solani against T. atroviride were 15.7 mm and 12.2 mm in the absence and the presence of caffeine, respectively. Inhibition rate was 22% and mycoparasitism acceleration rate was 1.3-fold. Similarly, the confrontation radii in the absence and the presence of caffeine were respectively 20.9 mm and 12.9 mm in S. sclerotiorum/T. atroviride, and 13.5 mm and 10.5 mm in G. cingulata/T. atroviride. Inhibition rates were 38% in the former and 22% in the latter. Acceleration rates were 1.6-fold and 1.3-fold, respectively. In summary, caffeine fostered mycoparasites by conferring 1.3 to 1.7-fold increase in attacking capability.

Table 2.

Effects of mycoparasites on pathogen growth in the absence and presence of caffeine

Pathogens	F. oxysporum	R. solani	S. sclerotiorum	G. cingulata
Confrontation radius without caffeine (mm)	22.5+1.29	15.7+0.46	20.9+0.75	13.5+0.57
Confrontation radius with caffeine (mm)	13.5+0.48	12.2+1.47	12.9+0.75	10.5+1.00
Inhibition rate (%)	40	22	38	22
Acceleration rate (fold)	1.7	1.3	1.6	1.3
p-value	0.0061	<0.0001	0.0080	0.0072

The mycoparasites were T. virens for F. oxysporum and T. atroviride for the others. The confrontation radius was estimated by the distance between confronting border and colony center at the steady state phase in mm (see Fig. 1A). Mean value was calculated from all time points with standard deviations. Inhibition rate indicates suppression degree of the prey growth to the control in the presence of caffeine. Acceleration rate indicates accelerated degree of mycoparasitic activity in the presence of caffeine. The p-value indicates the difference between samples with and without caffeine to be significant (p < 0.05).

Discussion

Caffeine has long been known to affect fungal growth and development.^29,30 Its actual and global influence on fungal community, however, has not been well documented. The present work was designed and performed in order to understand a part of the ecological significance of caffeine. Within the frame of simple and in vitro experiments, we first found that caffeine sensitivity was different among fungi, ranging from hyper-sensitive to almost insensitive strains. Second, we found that differential effect of caffeine was favorable for mycoparasitic fungi over pathogenic fungi.

Caffeine and its derivatives, methylxanthines, are powerful inhibitor of cAMP-phosphodiesterases from mammals, plants, insects, bacteria and fungi.^9,31,32 In most cases, the drug treatment raises the level of endogenous cAMP by blocking its hydorolysis,³² and causes an imbalance of cAMP-signaling pathways involving G proteins and protein kinases.³³ This results in a serious alteration of physiological function.^34-36 For instance, when T. harzianum was treated with 0.06% caffeine, cAMP level increased 4-fold that of the control, and growth was repressed to 85% of the control.³⁷ In S. sclerotiorum, 0.1% caffeine increased endogenous cAMP level 3 to 10-fold higher than the control, and prevented mycelial aggregation, sclerotia initial formation and development.³⁸ In Colletotrichum (asexual stage of Glomerella), cAMP signaling pathway was shown to be critical for formation of melanins, which provide high pressures to appressoria to penetrate plant leaves.³⁹ Inhibition of phosphodiesterase by isobutylmethylxanthine or by high level of cAMP (5 mM) completely inhibited melanin formation in Ustilago hordei.⁴⁰ One of methylxanthine functions is thus established to modulate endogenous cAMP, which critically regulates fungal life, including morphogenesis, virulence, nutrient sensing and reproduction.⁴¹ Our observation is in accordance with this view, and suggests the possibility that observed caffeine effects were, at least partly, due to imbalance of cAMP metabolism. A question then arises as to why the caffeine effect varied among fungal species. Several factors are conceivable. For example, the quality and quantity of the target phosphodiesterases such as affinity (Km), structure of isoforms and copy numbers differ among fungi.^31,42,43 Properties of signaling components including protein kinases and G proteins^44,45 may be diverse. Moreover, cAMP itself may have different role and/or sensitivity in individual species as suggested in carbon utilization and conidiation.³¹

The differential impact of caffeine on individual fungus may cause a significant change of the structure of fungal community around caffeine-producing plants. Our second finding, showing caffeine fostering of mycoparasitism, is one of such case. This could be the result of relative increase of predator's activity due to differential growth velocity; slow in pathogens and fast in mycoparasites. Another conceivable cause is that mycoparasites positively adapt to caffeine. In mycoparatisism, the first step of predation is coiling and attacking prey's hyphae by predator's hyphae.^46,47 The coiling activity was found to be controlled by internal cAMP level⁴⁵ and G proteins.⁴⁸ G proteins, their receptors (G protein-coupled receptors; GPCRs) and cAMP receptor-like GPCRs constitute the down-stream components of cAMP cascade, and play a key role to sense the prey in mycoparasitism.²⁵ Coiling activity was directly shown to be enhanced nearly 3-fold by 5 mM of either cAMP or phosphodiesterase inhibitor isobutylmethylxanthine.⁴⁹ Judging from these data, predator fungi clearly benefit from caffeine for their parasitic activity.

Caffeine is thus a double-edged instrument for Trichoderma species; being an enhancer of parasitism on one side, and a suppressor of growth on the other side. Balancing between the 2 depends on species, and the majority of them appears to prefer the advantage as it may have acquired partial tolerance or insensitivity to caffeine toxicity. This could explain, at least partially, the abundance of Trichoderma species in rhizosphere of coffee plants.¹⁷ Caffeine-producing plants directly benefit from Trichoderma species to cope with pathogens. In addition, host plants also indirectly benefit from Trichoderma species through activation of plant growth potential, priming for defense reaction and antibiotic production.^25,50 Activation of priming is particularly beneficial for plants. Trichoderma species secret variable elicitors, including peptides and proteins, avr-like proteins and short oligosaccharides, all of which quickly activate the defense system before or immediately after the pathogen attack.^26,28

The present study together with so far available information can be summarized as following. First, caffeine constitutes the allelopathic defense system by directly suppressing pathogen growth and development. Second, caffeine fosters Trichoderma species, with which caffeine-producing plants ally to fight against pathogens. Third, Trichoderma species also ally with host plants to secure nutrients and environment. Fourth, accordingly plant-Trichoderma interaction is mutually beneficial and caffeine intermediates the connection. In this context, caffeine serves as a powerful tool or weapon in arms race between plants and pathogens by fostering enemy's enemy. This could be regarded as the third role of caffeine, and be called "caffeine fostering."

Materials and Methods

Chemicals and definition of concentration

Caffeine monohydrate was purchased from Nacalai Tesque Inc. (Kyoto, Japan). Dried mashed potatoes were purchased from Tsuji Anzen Shokuhin (Tokyo, Japan). Other chemicals were obtained from Nacalai Tesque Inc. or Wako Pure Chemical Industries (Osaka, Japan), unless otherwise stated. In physiological research papers, concentration of caffeine has commonly been expressed in percentage.⁶ Taking the molecular weight of caffeine to be 194.2, one percent approximately corresponds to 50 mM, or 10,000 ppm. In this study, the standard caffeine concentration was fixed to and expressed in 0.1%, which corresponds to ∼5 mM or 1,000 ppm.

Fungal strains and properties

Fungal strains were obtained from the National Institute of Agrobiological Sciences Genebank, Tsukuba, Japan. Mycoparasites were Trichoderma virens (MAFF # 425559), T. harzianum (MAFF#328304), T. hamatum (MAFF#236548) and T. atroviride (MAFF#235587). The pathogen species were Fusarium oxysporum (MAFF # 103007), Rhizoctonia solani (MAFF # 731100), Glomerella cingulata (MAFF # 238795) and Sclerotinia sclerotiorum (MAFF # 744080). All of these pathogens cause serious diseases not only in coffee plant but also in many other crop plants.^51-53 F. oxysporum causes vascular browning, leaf epinasty, stunting, progressive wilting, defoliation and plant death.⁵³ R. solani has a wide host range causing black root rot, stem canker and tuber blemishes in potato.^54,55 G. cingulata (sexual stage of Colletotrichum gleosporioides) causes anthracnose and brown blight.⁵³ S. sclerotiorum causes white mold, rot and blight.⁵⁶

Culture condition and confrontation

Fungal strains were inoculated on 1/5 PDA plates, cultured under dark at 23°C for 5–10 d and stored at 4°C until use. A disc from stock mycelia (3 mm diameter) was placed on 1/5 PDA containing indicated concentration of caffeine, and cultured under dark at 23°C for indicated time periods. Confrontation assay was performed on a 1/5 PDA plate by placing a disc of the phytopathogene at one side and that of the Trichoderma at the other side, each disc being 5 cm apart. Trichoderma was inoculated simultaneously with R. solani and S. sclerotiorum, and 2 d later with F. oxysporum and G. cingulata. The plate was incubated under dark at 23°C for indicated time periods. A close-up photograph was regularly taken for each plate at appropriate time intervals.

Growth rate determination

The growth rate was determined by measuring colony radius in millimeter (mm) throughout the experiments. In the case of single inoculation, distance between center of the agar disc and the edge of the colony hyphae was estimated. In the case of confrontation assay, a straight line was drown from the center of the agar disc of the prey to that of the predator and distance between center of the agar disc of the prey and the confronted border line was measured (Fig. 2A). This indicates the actual radius or growth of the prey in the presence of predator, and referred to as the confrontation radius. The growth velocity was estimated from the growth curve and expressed in increase of colony radius in mm per day. The growth inhibition by caffeine and/or predator was determined by the ratio of the growth velocity between the control and the treatment. The inhibition rate is expressed in percentage as 100 – [(radius value of the treated sample/radius value of control sample) × 100]. The mycoparasitism acceleration was expressed by the increase of predator activity (radius value of the control sample/radius value of the treated sample), and referred to as the acceleration rate expressed in fold.

Statistical analysis

Growth and confrontation experiments were performed at least twice and each assay was triplicated. Measured values were subjected to analysis of covariance (ANCOVA) in order to better investigate the effects of the caffeine.⁵⁷ Colony growths, adjusted for covariance (time) effect were compared between 2 groups (with or without caffeine) for each fungus. Values having p<0.05 were considered to be significant. R statistical software package was used to perform the analysis.⁵⁸ Note that, for statistical analysis, all measured values in addition to representative values illustrated in figures (Fig. 2 and S2) were used to make more precise evaluation.

Disclosure of Potential Conflicts of Interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors are grateful for technical assistance by Mss Yuko Seo and Yuko Kobayashi (Research Institute for Sustainable Humanosphere, Kyoto University).

Funding

This work was supported by grants from Research Institute for Sustainable Humanosphere, Kyoto University and from Japan Society for the Promotion of Science, Grant-in-Aid (JSPS KAKENHI) Grant Number 26660279 (AS).

Supplementary Material

Supplemental data for this article can be accessed on the publisher's website