Pathogenesis-related proteins protect extrafloral nectar from microbial infestation


  • Marcia González-Teuber,

    1. Department of General Botany – Plant Ecology, University of Duisburg-Essen, Universitätsstraße 5, 45117 Essen, Germany
    Search for more papers by this author
  • Sascha Eilmus,

    1. Department of General Botany – Plant Ecology, University of Duisburg-Essen, Universitätsstraße 5, 45117 Essen, Germany
    Search for more papers by this author
  • Alexander Muck,

    1. Research Group Mass Spectrometry, Max Planck Institute for Chemical Ecology, Hans-Knöll-Straße 8, 07745 Jena,
    Search for more papers by this author
  • Ales Svatos,

    1. Research Group Mass Spectrometry, Max Planck Institute for Chemical Ecology, Hans-Knöll-Straße 8, 07745 Jena,
    Search for more papers by this author
  • Martin Heil

    Corresponding author
    1. Department of General Botany – Plant Ecology, University of Duisburg-Essen, Universitätsstraße 5, 45117 Essen, Germany
    2. Departamento de Ingeniería Genética, CINVESTAV-Irapuato, km. 9.6 Libramiento Norte, Carretera Irapuato-León, 36821 Irapuato, Guanajuato, México
      *For correspondence (fax +52 462 623 9650; e-mail
    Search for more papers by this author

*For correspondence (fax +52 462 623 9650; e-mail


Plants in more than 300 genera produce extrafloral nectar (EFN) to attract carnivores as a means of indirect defence against herbivores. As EFN is secreted at nectaries that are not physically protected from the environment, and contains carbohydrates and amino acids, EFN must be protected from infestation by micro-organisms. We investigated the proteins and anti-microbial activity in the EFN of two Central American Acacia myrmecophytes (A. cornigera and A. hindsii) and two related non-myrmecophytes (A. farnesiana and Prosopis juliflora). Acacia myrmecophytes secrete EFN constitutively at high rates to nourish the ants inhabiting these plants as symbiotic mutualists, while non-myrmecophytes secrete EFN only in response to herbivore damage to attract non-symbiotic ants. Thus, the quality and anti-microbial protection of the EFN secreted by these two types of plants were likely to differ. Indeed, myrmecophyte EFN contained significantly more proteins than the EFN of non-myrmecophytes, and was protected effectively from microbial infestation. We found activity for three classes of pathogenesis-related (PR) enzymes: chitinase, β-1,3-glucanase and peroxidase. Chitinases and β-1,3-glucanases were significantly more active in myrmecophyte EFN, and chitinase at the concentrations found in myrmecophyte EFN significantly inhibited yeast growth. Of the 52 proteins found in A. cornigera EFN, 28 were annotated using nanoLC-MS/MS data, indicating that chitinases and glucanases contribute more than 50% of the total protein content in the EFN of this myrmecophyte. Our study demonstrates that PR enzymes play an important role in protecting EFN from microbial infestation.


Nectar is an aqueous solution of substances that mainly comprise primary metabolites such as sugars and amino acids (Baker and Baker, 1975; Baker et al., 1978). Benefits for plants resulting from nectar secretion include pollination in the case of flower nectar and protection from herbivores through attraction of carnivores in the case of extrafloral nectar (EFN) (Koptur, 1992; Heil, 2007, 2008). EFN is usually secreted outside the flowers, and – in contrast to floral nectar – is not involved in pollination (Bentley, 1977; Koptur, 1992).

Despite its enormous importance, surprisingly little is known of the chemical composition of nectar, and new substance classes continue to be detected. In fact, research during recent years has revealed that nectar comprises much more than sugars and amino acids dissolved in water. Floral nectar may release volatiles that contribute to its attractive function (Kessler and Baldwin, 2006; Kessler et al., 2008), and may contain other secondary metabolites, such as alkaloids (Adler et al., 2006) and phenolics (Ferreres et al., 1996), that apparently repel nectar thieves. Floral nectar even contains proteins (Lüttge, 1961; Baker and Baker, 1975; Carter and Thornburg, 2000, 2004a; Naqvi et al., 2005), which mainly comprise enzymes that protect the nectar from microbial infestation (Carter and Thornburg, 2004b; Nicholson and Thornburg, 2007). Particularly prominent are the so-called nectarins, enzymes found in the floral nectar of Nicotiana sp. (Carter and Thornburg, 2000, 2004b; Naqvi et al., 2005) that are involved in a redox cycle producing high levels of hydrogen peroxide to ensure that the nectar remains microbe-free (Carter and Thornburg, 2004b). This defensive function is important as the nectar composition makes it an excellent medium for microbial growth (Buban et al., 2003; Raguso, 2004).

Much less is known regarding the chemistry of EFN and the role that particular compounds play in its ecological interactions, even though EFN has been described for plants in more than 300 genera (Bentley, 1977; Koptur, 1992; see also Invertase activity has been detected in the EFN of various Acacia species, and allows detailed adjustments of its carbohydrate composition to exclude non-mutualistic ants (Heil et al., 2005), but we are not aware of any further report on enzymes or proteins in EFN. In order to improve our understanding of the various functions of EFN, we chose a comparative approach and used species characterized by various levels of specificity with regard to plant–ant mutualism. Among Mesoamerican Acacia species, obligate myrmecophytes secrete EFN constitutively and at high rates to nourish symbiotic ants (Heil et al., 2004). Although symbiotic ants live on myrmecophytes, their activity peak (Raine et al., 2002) and the time during which ants visit nectaries is very short (M.G.-T. and Silva Bueno, unpublished data), and thus secreted nectar probably requires ant-independent protection from microbes. In contrast, non-myrmecophyte Acacia species and related genera secrete EFN only in response to herbivore attack, thereby actively attracting ants from the vicinity (Heil and McKey, 2003).

We predicted that the EFN of myrmecophytes will contain anti-microbial protection that is more active than that in the EFN of non-myrmecophytes, as the EFN of Acacia myrmecophytes is secreted constitutively at high rates and is therefore more easily accessible to micro-organisms. Yeasts in particular are common and abundant in floral nectars (Brysch-Herzberg, 2004; Sandhu and Waraich, 2005), and may cause significant changes in nectar sugar composition, including nectar degradation (Herrera et al., 2008). As Acacia myrmecophytes actively regulate the sugar composition of their EFN (Heil et al., 2005), these plants must be able to prevent yeasts from infesting their EFN and interfering with EFN composition. We therefore selected yeast as a model micro-organism for biotests on the putative protection of EFN from microbial infestation.

First, we determined the amounts of proteins present in EFN. Second, the putative anti-microbial defence of EFN was investigated by determining the presence of fungi in fresh nectar collected from of plants growing in the field and by quantifying the activity of three classes of pathogenesis-related (PR) enzymes for which functions are known, i.e. chitinase, β-1,3-glucanase and peroxidase (Van Loon, 1999). PR proteins were also identified by means of nanoLC Q-TOF-MS/MS, and quantified in 2D gels to estimate the extent to which they contribute to overall EFN protein content. We accompanied this investigation by bioassays to determine whether enzyme activities as found in our study can effectively suppress yeast growth. We thereby identified the physiological basis of a yet undescribed function of EFN: its enzymatic protection from microbial infestation.


Total proteins

The total amount of proteins as determined by Bradford assays was significantly higher in myrmecophyte EFN than in non-myrmecophyte EFN (Figure 1a,b), both when expressed per total content of soluble solids (χ2 = 20.0; d.f. = 3; < 0.001; Kruskal–Wallis test) and per leaf dry mass (χ2 = 20.0; d.f. = 3; < 0.001; Kruskal–Wallis test). However, this effect was caused only by differences between the two life forms, as there were no significant differences between A. cornigera and A. hindsii, or between A. farnesiana and Prosopis juliflora (Figure 1). Similarly, SDS–PAGE analysis showed protein patterns that clearly differed between myrmecophyte and non-myrmecophyte species (Figure 2). Whereas numerous bands could be observed in the EFN of both myrmecophytes, protein bands appeared in much lower numbers and abundances in the EFN of A. farnesiana and P. juliflora. For myrmecophyte EFN, the molecular mass of the major protein bands ranged between 20 and 50 kDa.

Figure 1.

 Protein quantities in EFN.
(a) Relative protein content in EFN (μg proteins per mg sugars) and (b) amount of EFN proteins per leaf dry mass (μg proteins per gram leaf dry mass per 24 h) are shown for A. cornigera, A. hindsii, A. farnesiana and P. juliflora as means + SE. The sample size was seven individuals per species. Different letters indicate significant differences between the species.

Figure 2.

 Protein patterns in EFN.
SDS–PAGE (13%) profile of EFN proteins from myrmecophyte and non-myrmecophyte species. M indicates molecular weight markers, Myrm. indicates the myrmecophyte species A. cornigera and A. hindsii, and Non-myrm. indicates the non-myrmecophyte species A. farnesiana and P. juliflora.

Anti-fungal protection of EFN in nature

The occurrence of fungi in EFN under natural growing conditions was investigated by collecting samples from the field and cultivating them on malt agar plates to quantify the numbers of colony-forming units (CFU). No fungi were detected in the EFN of the two myrmecophytes; significantly higher numbers appeared in the EFN of non-myrmecophytes (χ2 = 7.2; d.f. = 3; < 0.05; Kruskal–Wallis test) (Figure 3). These results suggest that EFN can become infested by fungi under natural conditions, and that the EFN of myrmecophytes provides some protection from micro-organisms.

Figure 3.

 Presence of fungi in fresh EFN samples.
Fungal growth in EFN of the myrmecophytes A. cornigera and A. hindsii and in EFN of the non-myrmecophytes A. farnesiana and P. juliflora. Fungal growth was evaluated as the number of colony-forming units per 30 μl EFN after 48 h of incubation. Different letters indicate significant differences between the species.

Pathogenesis-related (PR) enzymes

The activities of the pathogenesis-related (PR) enzymes chitinase, β-1,3-glucanase and peroxidase were determined in EFN using standard colorimetric assays. Activity of all three PR enzymes was detected in the EFN of all four species investigated, but the activity differed between species. For example, chitinase activity differed significantly between the species (χ2 = 12.78; d.f. = 3; < 0.01; Kruskal–Wallis test), with myrmecophyte EFN generally having higher activities than the EFN of non-myrmecophytes (Figure 4a). Even the two non-myrmecophyte species differed significantly, with A. farnesiana having the lowest activity among all species investigated. Glucanase activity showed the same pattern as chitinase: it was higher in myrmecophyte than in non-myrmecophyte EFN (Figure 4b), and A. farnesiana showed the lowest activity among the four species investigated (χ2 = 11.80; d.f. = 3; < 0.01; Kruskal–Wallis test). In contrast, peroxidase activity did not differ significantly among the four species investigated (χ2 = 4.00; d.f. = 3; > 0.05; Kruskal–Wallis test), and was much lower than the activities of glucanase and chitinase (Figure 4c).

Figure 4.

 Enzyme activities.
Activities of three pathogenesis-related (PR) enzymes [(a) chitinase, (b) glucanase and (c) peroxidase] in EFN of A. cornigera, A. hindsii, A. farnesiana and P. juliflora are shown as means ± SE. The sample size was five individuals per species. Different letters indicate significant differences between the species.

Identification and quantification of PR proteins

Two-dimensional gel analysis of the A. cornigera EFN proteome revealed a relatively low number of different proteins (52 spots, see Figure 5). Around 75% of the proteins ranged in molecular mass between 20 and 37 kDa, consistent with the patterns seen in the 1D gels (Figure 2). Spots isolated from 2D gels were analysed by nanoLC-MS/MS, and the fragment spectral data were searched against the specific ‘planta’ sub-database of EBI (European Bioinformatics Institute, using the ProteinLynx GLOBAL SERVER™ ( The most abundant proteins in EFN of A. cornigera were most similar to chitinases and glucanases (Table 1). In order to quantify the extent to which these chitinases and glucanases contribute to the total amount of EFN proteins, we used the PD Quest 7.3.0 program, and performed a relative quantification by determining the volume of each spot as its absorbance (A) multiplied by its area (mm2). Glucanase proteins contributed approximately 40 ± 1.4% (= 3 gels) to the total proteins in the EFN of A. cornigera, while chitinase proteins contributed approximately 14 ± 1% (= 3 gels).

Figure 5.

 Two-dimensional gel electrophoresis.
Separation of proteins in EFN from A. cornigera by 2D gel (10% SDS–PAGE). C, chitinase proteins; G, glucanase proteins; H, glycoside hydrolase proteins; I, invertase protein; O, osmotin proteins; P, PR proteins; PX, peroxidase protein; T, thaumatin-like protein; U, unknown proteins.

Table 1.   Results of MS-BLAST searches using de novo peptide sequences
SpotAccessionDescriptionPlant speciesPeptide hits (MS-BLAST)MS-BLAST score
  1. PR families were identified as described by Van Loon et al. (1994, 2006).

C1AAC24807Class I chitinase, PR-3Solanum tuberosum2125
C2ABD66068Chitinase, PR-3Momordica charantia6289
C3ABD66068Chitinase, PR-3Momordica charantia8526
C4O81145Class I chitinase, PR-3Solanum tuberosum8441
C51302305AChitinase, PR-3Nicotiana sp.164
C61302305AChitinase, PR-3Nitoctiana sp.299
C71302305AChitinase, PR-3Nicotiana sp.1107
C8CAO78600Endochitinase, PR-3Parkia platycephala3145
C9CAO78600Endochitinase, PR-3Parkia platycephala3188
G1CAJ91137β-1,3-glucanase, PR-2Platanus x acerifolia3133
G2ABD85024β-1,3-glucanase, PR-2Lilium hybrid division VII, cv. Star Gazer285
G3BAE53384β-1,3-glucanase, PR-2Sesbania rostrata588
G4AAX81590β-1,3-glucanase, PR-2Fragaria x ananassa1253
G5AASO9873Endo-β-1,3-glucanase, PR-2Glycine latrobeana2121
G6AACO4712β-1,3-glucanase, PR-2Glycine max5280
H1ABP03049Glycoside hydrolaseMedicago truncatula4234
H2AAB77250Glycoside hydrolaseMedicago truncatula4234
IABB77250Cell-wall invertaseBambusa oldhamii4262
O1AAU95238Osmotin-like protein, PR-5Solanum phureja8443
O2CAC34005Osmotin-like protein, PR-5Capsicum annuum2108
P1AAO22065Nt PRp27-like protein, PR-17Solanum tuberosum5314
P2BAA81904Nt PRp27, PR-17Nicotiana tabacum5314
P3BAA81904Nt PRp27, PR-17Nicotiana tabacum10546
P4BAA81904Nt PRp27, PR-17Nicotiana tabacum3188
P5AAU94913PR protein 4AArachis hypogaea167
P6CAA50596PR-1a1Solanum lycopersicum181
P7CAA87071Pathogenesis-related protein, PR-1 typeSambucus nigra4275
P8BAE93153Pathogenesis-related protein 1Lolium perenne2145
P9ABB73064Pathogenesis-related protein PR-1Glycine max2117
P10CAA87071Pathogenesis-related protein, PR-1 typeSambucus nigra198
PXCAH59427Ascorbate peroxidase, PR-9Plantago major146
T1AAD55090Thaumatin, PR-5Vitis riparia3158
T2AAK59277Thaumatin-like protein, PR-5Sambucus nigra4332
T3CAA48278Thaumatin-like protein, PR-5Oryza sativa Japonica 184
T4CAA48278Thaumatin-like protein, PR-5Oryza sativa Japonica 196
T5AAM15877Thaumatin-like protein, PR-5Triticum aestivum2109
T6CAA48278Thaumatin-like protein, PR-5Oryza sativa Japonica 4230
T7AAM12886Thaumatin-like protein, PR-5Malus x domestica4154
T8CAA48278Thaumatin-like protein, PR-5Oryza sativa Japonica 3212

Effect of enzyme activities on yeast

Yeasts are among the species that are most likely to be present in floral nectar (Brysch-Herzberg, 2004; Sandhu and Waraich, 2005; Herrera et al., 2008), but no yeasts or fungi were found in myrmecophyte EFN (Figure 3). We therefore used two bioassays to investigate whether the activities of PR enzymes found in the EFN can suppress yeast growth. First, freshly collected EFN of all four species was mixed with a suspension of yeast (Saccharomyces cerevisiae) and then cultivated on malt agar plates to quantify the amount of developing yeasts in terms of colony-forming units. Myrmecophyte EFN inhibited the development of yeasts, with significantly fewer colony-forming units found in the EFN of A. cornigera and A. hindsii compared to a pure sugar solution (Figure 6a). In contrast, no significant reduction in the number of colony-forming units was seen for the EFN of the two non-myrmecophytes, A. farnesiana and P. juliflora. ‘Species’ was therefore a significant source of variance in the numbers of colony-forming units (= 5.20; d.f. = 4, 35; < 0.01; univariate anova).

Figure 6.

 Effect of EFN enzyme activities on yeast growth.
(a) Yeast growth (number of colony-forming units per 20 μl) in EFN of two myrmecophyte (A. cornigera and A. hindsii) and two non-myrmecophyte species (A. farnesiana and P. juliflora), and (b) yeast growth (number of colony-forming units per 100 μl) in artificial nectar solutions, with and without chitinase activity (6% GF = glucose/fructose solution at a concentration of 6%; 6% GF + CH = glucose/fructose solution at a concentration of 6% with chitinase activity as found for A. cornigera; 2% GFS = glucose/fructose/sucrose solution at a concentration of 2%; 2% GFS + CH = glucose/fructose/sucrose solution at a concentration of 2% with chitinase activity as found for A. farnesiana).

To investigate whether the chitinase activity as found in the EFN has a relevant effect in this context, we used purified chitinase from Streptomyces griseus and investigated its effect on yeast at concentrations as found in EFN of A. cornigera and A. farnesiana compared to yeast grown in water and sugar solution without chitinase. Chitinase activity as found in A. cornigera EFN significantly reduced yeast growth (= 4.49; d.f. = 2, 21; < 0.05; univariate anova), as a sugar solution without chitinase supported significantly more colony-forming units than the water control and the nectar mimic with chitinase activity. Therefore, a sugar solution with chitinase activity as seen in A. cornigera allowed as little microbial growth as a pure water solution (Figure 6b). On the other hand, a sugar solution with chitinase activity as seen in A. farnesiana did not significantly reduce yeast growth (= 0.92; d.f. = 2, 21; > 0.05; univariate anova), although a strong tendency towards a reduction in colony-forming units was apparent, similar to the pattern seen for A. cornigera (Figure 6b). Inhibition rates were calculated for each trial as inhibition rate (%) = (CFU in sugar solution - CFU in sugar solution plus chitinase)/CFU in sugar solution) × 100, and was 36.7 ± 8% for A. cornigera and 27.5 ± 18% for A. farnesiana. Apparently, the mean chitinase activity found in A. farnesiana EFN was not quite high enough to cause a significant effect.


Many plants secrete extrafloral nectar (EFN) to attract carnivorous arthropods as a means of indirect defence (Heil, 2008). However, since EFN is secreted by nectaries that are not protected from the environment by any anatomical structures, and since EFN contains generally attractive compounds such as sugars and amino acids, EFN requires protection from exploiters. Earlier studies suggested that the presence of non-proteinogenic amino acids in EFN (Baker and Baker, 1973) or the invertase-mediated absence of sucrose from EFN (Heil et al., 2005) might help to defend it from ‘thieves’, i.e. EFN-consuming arthropods that do not provide the secreting plant with a defensive service. Here, we investigated whether EFN is also protected from microbial infestation, as has been reported for floral nectar (Carter and Thornburg, 2000, 2004a; Naqvi et al., 2005; Nicholson and Thornburg, 2007).

The EFN of myrmecophyte Acacia species possessed more proteins than the EFN of related non-myrmecophytes, both in terms of overall quantity and the number of different proteins. Although EFN proteins may serve ant nutrition, our results demonstrate that at least some of them have another function: the protection of nectar from microbes. Chitinases and β-1,3-glucanases were found in EFN, and their activities were higher in myrmecophyte EFNs than in the EFN of non-myrmecophytes. We suggest that these differences are related to the differences in EFN production rates between the functional groups of plants (Heil et al., 2004). Myrmecophyte EFN, unlike the EFN of non-myrmecophytes, is constitutively secreted (Heil et al., 2005), and EFN in general is openly exposed to the environment, and hence easily accessible for microbial attack. Although high sugar concentrations might protect the nectar from microbial growth (Buban et al., 2003), our results suggest that PR enzymes secreted into the nectar contribute significantly to the protection of Acacia EFN from microbial infestation.

Does this protection function in nature? Freshly field-derived samples of myrmecophyte EFN were free of fungi under our experimental conditions, while the same conditions allowed detection of microbes in the EFNs obtained from non-myrmecophytes. No microbes were detected even in the EFN of myrmecophytes that had been deprived of their resident ants (M.G.-T., unpublished observations), suggesting that EFN itself exhibits anti-microbial defence. Bioassays demonstrated this defence at the functional level. In these bioassays, myrmecophyte EFN inhibited the development of yeasts while the EFN of the non-myrmecophytes did not cause significant suppression of yeast growth. We conclude that PR enzyme activities as found in myrmecophyte EFN are sufficient to prevent infestation by yeasts.

Chitinases, β-1,3-glucanases and peroxidases are common enzymes in plant pathogen resistance (Van Loon, 1999; Van Loon et al., 2006). Chitinases and β-1,3-glucanases exhibit an inhibitory activity against fungi and bacteria (Sela-Buurlage et al., 1993; Fung et al., 2002; Robert et al., 2002), while peroxidases normally function via the production of hydrogen peroxide, which then serves as an anti-microbial agent (Orozco-Cardenas and Ryan, 1999; Mydlarz and Harvell, 2007). In the floral nectar of tobacco plants, superoxide dismutase activity and the generation of hydrogen peroxide inhibited microbial growth (Carter and Thornburg, 2000). PR proteins form a major proportion of the total protein fraction in EFN of the myrmecophyte species A. cornigera, with glucanases, chitinases and thaumatin-like proteins being the most abundant classes. Chitinase and glucanase proteins alone made up more than 50% of the total protein fraction of this EFN, with glucanases the most abundant. Other proteins identified were related to sugar hydrolysis, e.g. invertase (Roitsch and González, 2004) and glycoside hydrolase (Zoran, 2008); this identification of invertase confirms previous results on the function of this enzyme in Acacia EFN (Heil et al., 2005). However, these enzymes made up a lower proportion of the total protein fraction, suggesting that the main functions of proteins in Acacia EFN relate to its protection from microbes.

The chemical composition of EFN was more complex than previously thought. The function of EFN components is not only restricted to ant attraction, but also involves protection from microbial infestation through activities of PR proteins. Moreover, the EFN of myrmecophytes possessed several additional proteins whose identity and physiological functions remain to be analysed. EFN is an ecologically important type of plant secretion, and much more thorough knowledge of its chemical composition is required to understand the various roles of EFN in the ecology and evolution of mutualistic relationships.

Experimental procedures

Plant material and study sites

We investigated the chemical composition of EFN of two myrmecophyte species [Acacia hindsii Benth. and A. cornigera (L.) Willendow], one non-myrmecophytic Acacia [A. farnesiana (L.) Willendow] and one non-myrmecophytic, closely related and sympatric species of another genus, butt the same sub-family, the Mimosoideae (Prosopis juliflora Swartz). EFN was collected from plants growing naturally in the coastal area of the state of Oaxaca, 5 km northwest of Puerto Escondido, Mexico (Pacific coast; approximately 15°55′ N and 97°09′ W, elevation 15 m), in March and April 2007 and 2008. Species were determined based on the descriptions published byJanzen (1974) and Seigler and Ebinger (1995) and by comparison with specimens held at the Herbario Nacional MEXU at Universidad Autónoma de México (UNAM).

Nectar collection

Branches of myrmecophytes were deprived of ants and other insects the day before nectar collection: thorns were cut off, ants were mechanically removed, and the branch was then placed in a mesh bag after isolating it from the rest of the plant by applying a ring of sticky resin (Tangletrap, Tanglefoot Corp., EFN secretion on branches of non-myrmecophyte species was induced by applying 1 mmol aqueous jasmonic acid solution, and these branches were then placed in mesh bags (Heil, 2004; Heil et al., 2004). In these plants, jasmonic acid is involved in the natural induction of EFN secretion after herbivore damage (Heil et al., 2004), and no chemical changes in jasmonic acid-induced nectar have yet been detected. The nectar production rate of all plants was measured after 24 h as the amounts of secreted soluble solids by quantifying the nectar volume using micro-capillaries and the nectar concentration using a refractometer as described previously (Heil et al., 2000, 2001, 2004). The leaves that had produced EFN were then collected and dried. EFN was collected from five individuals per species.

Quantification of proteins and SDS–PAGE

Quantification of total proteins was performed using the Bradford assay (Bradford, 1976). Protein levels were determined in fresh nectar immediately after collection in the field. Then, protein quantities were determined relative to the total amounts of secreted soluble solids (mg) and the dry weight (g) of the respective leaves. Differences in protein quantities among species were analysed using the Kruskal–Wallis test.

Before SDS–PAGE, fresh EFN of each species (10–20 μl for myrmecophyte species, 150–200 μl for non-myrmecophyte species) was precipitated using 10% v/v trichloroacetic acid at 4°C (nectar:trichloroacetic acid = 1:2). The mixture was incubated for 1.5 h at 4°C, and centrifuged at 19 300 g for 15 min at 4°C. Then the supernatant was removed and 0.5 ml of absolute ethanol was added. Samples were centrifuged at 10 400 g for 10 min at 4°C. Finally, proteins (15–20 μg per sample) were separated on a 13% SDS–PAGE Laemmli gel and stained with Coomassie brilliant blue.

Anti-fungal protection of EFN in nature

To screen for fungi, EFN was collected as described above and adjusted to a concentration of 3% of soluble solids (w/v) using a portable refractometer. This concentration was chosen as it was not possible to obtain more highly concentrated EFN from A. farnesiana (the same criterion was used to adjust the concentration of EFN in the following experiments). Then 30 μl of EFN (diluted 1:100 in 0.1 m PBS, pH 7.0) was plated on malt agar plates (20 g malt extract + 15 g agar). The dilution 1:100 was chosen after testing various dilutions (1:10, 1:100 and 1:1000) for all treatments. The same procedure was used for the yeast assay (see below). Plates were stored at room temperature for 48 h, and then colonies were counted to quantify the number of colony-forming units. Differences in fungal abundance (number of colony-forming units per 30 μl EFN) among the species were analysed using Kruskal–Wallis anova. The number of replicates was five individuals per species.

Biochemical assays of PR enzymes

Nectar samples were diluted 1:10 with pure water and adjusted to a concentration of 5% w/v. To quantify chitinase activity, assays based on the method described by Wirth and Wolf (1990) were performed in 96-well microplates. The total volume of 100 μl reaction preparation contained 10 μl nectar, 40 μl 50 mm Na-acetate buffer (pH 5.0) and 50 μl Remazol Brilliant Violet carboxymethyl (RBV) chitin (Loewe, Each preparation was replicated four times, incubated for 2.5 h at 37°C, and the reaction was stopped using 26 μl 0.05 M HCl. After 5 min incubation at −20°C, the plate was centrifuged at 4000 g at 4°C, and 100 μl of the supernatant were transferred to a new microplate and the absorbance measured at 550 nm in a spectrophotometer (Smax 190PC, Molecular Devices GmbH,

The activity of β-1,3-glucanase was assayed using Laminaria digitata laminarin (Sigma, as substrate. The assay mixture contained 5 μl nectar, 10 μl laminarin (20 mg ml−1 in 50 mm Na-acetate buffer, pH 5.0), 60 μl copper reactive, 60 μl arsenic reactive (see Somogyi, 1952 for preparation procedure), in a total volume of 135 μl. The amount of reducing sugars released was determined as described previously (Somogyi, 1952). One unit of activity was defined as the amount of enzyme that catalysed the release of reducing sugar moieties equivalent to 1 μmol of glucose per minute.

To quantify peroxidase activity, the reaction solution (total volume of 197 μl) contained 5 μl nectar, 0.83 μl H2O2 (30%), 1 μl guaiacol (99%) and 190 μl 50 mm Na-phosphate buffer, pH 6.0. The oxidation of the substrate was measured spectrophotometrically using an Smax 190PC (Molecular Devices GmbH) at 470 nm as described previously (Hammerschmidt et al., 1982).

Kruskal–Wallis anova was used to evaluate differences among species for the activities of each enzyme class. The number of replicates was five individuals per species.

Two-dimensional gel electrophoresis and mass spectrometry

Chitinase and glucanase proteins in the EFN of the myrmecophyte species A. cornigera were identified by 2D gel electrophoresis and MALDI-TOF/MS. The 2D PAGE procedure has been described previously (Giri et al., 2006). Three replicate gels were used for protein identification. The following modifications have been made to the published procedure. After water removal from the microtiter plate (MTP), the gel plugs were reduced using 20 μl 10 mm DTT in 25 mm ammonium bicarbonate for 1 h at 56°C, alkylated using 20 μl 55 mm IAA at room temperature in the dark for 45 min, and rinsed twice for 20 min each with 70 μl 50 mm ammonium bicarbonate/50% acetonitrile to remove the Coomassie stain. A further wash was performed with 70 μl 70% acetonitrile for 20 min. The gel plugs were then air-dried for 30 min and overlaid with 15 μl 50 mm ammonium bicarbonate containing 70 ng porcine trypsin (sequencing grade, Promega, The MTPs were subsequently covered with aluminium foil, and the proteins were digested overnight at 37°C. The resulting peptides were extracted from the gel plugs by adding 40 μl 50% acetonitrile in 0.1% trifluoroacetic acid for 20 min, and then an additional extraction for further 20 min with 70 μl of the same extraction buffer. The extracts were collected in an extraction MTP, and vacuum-dried to remove any remaining liquid and the volatile ammonium bicarbonate. A MALDImicro MX mass spectrometer (Waters, was used in reflectron mode to monitor protein digestion and database identification. The tryptic peptides were reconstituted in 6 μl aqueous 0.1% trifluoroacetic acid.

Peptides not identified by MALDI-TOF/MS were identified de novo using LC/MS/MS (Giri et al., 2006; Pauchet et al., 2008). Aliquots of peptides (1.5–6 μl) were injected onto a nanoAcquity nanoUPLC system (Waters). A mobile phase of 0.1% aqueous formic acid (15 μl min−1 for 1 min) was used to concentrate and desalt the samples on a 20 x 0.180 mm Symmetry C18, 5 μm particle size pre-column. The samples were eluted on a BEH nanoAcquity C18 column (100 mm long × 75 μm internal diameter, pore size 1.7 μm) using an increasing acetonitrile gradient in 0.1% aqueous formic acid. Phases A (0.1% formic acid) and B (0.1% formic acid in MeCN) were linearly mixed using a gradient program up to 5% phase B in A over 0.33 min, increasing to 10% B over 10 min, 40% B over 10 min, and finally increasing to 85% B over 10.5 min, holding at 85%B until the 11th min, and decreasing to 1%B at 11.1 min. The eluted peptides were transferred to the NanoElectroSpray source of a Synapt HDMS Q-TOF-type tandem mass spectrometer (Waters) through a Teflon capillary union and a metal-coated nanoelectrospray tip (Picotip, 50 × 0.36 mm, 10 μm internal diameter, Waters). The source temperature was set to 60°C, the cone gas flow was 20 l h−1, and the nanoelectrospray voltage was 3.2 kV. The TOF analyser was used in reflectron mode. The MS/MS spectra were collected over a 1 sec interval in the range 50–1700 m/z. A mixture of 100 fmol μl−1 human Glu-fibrinopeptide B and 80 fmol μl−1 reserpine in 0.1% formic acid/acetonitrile (1:1 v/v) was infused at a flow rate of 0.9 μl min−1 through the reference NanoLockSpray source every fifth scan to compensate for mass shifts in the MS and MS/MS fragmentation mode due to temperature fluctuations.

Data were collected using MassLynx version 4.1 software (Waters), and ProteinLynx GLOBAL SERVER™ software (Waters) was used for baseline subtraction and smoothing, de-isotoping, de novo peptide sequence identification and database searches. The peptide fragment spectra were searched against the EBI ‘planta’ specific sub-database downloaded on 22 July 2008 from The protein database identification search parameters were: peptide mass tolerance 20 ppm and minimum two peptides found, estimated calibration error 0.005 Da, one possible missed cleavage, carbamidomethylation of cysteines and possible oxidation of methionines. A 0.05 Da mass deviation was allowed with a calibration error of 0.005 Da for de novo sequencing. A BLAST search was performed internally using the MS-BLAST algorithm (Shevchenko et al., 2001) using a minimum of one peptide matching at an expect score of 100, with the no-gap-hspmax100-sort_by_totalscore-span1 advanced options and the PAM30MS search matrix.

In order to obtain a rough impression of the contribution of chitinase and glucanase to the total amount of proteins in the EFN of A. cornigera, all spots present in the EFN were quantified using the PD Quest 7.3.0 program (2D Analysis Software, Bio-Rad, as the volume for each spot (absorbance × mm2). First we determined the volume for all spots to represent the total proteins present in the sample. The total volume of glucanases and chitinases was then also determined and related to the amount of total proteins. Spots were considered for quantification only when present in all three replicates.

Yeast assay of chitinase activity

An assay with yeast was carried out in order to evaluate the potential effects of EFN enzyme activities on microbial growth. Commercial yeast SK Saccharomyces cerevisiae (Uniferm GmBH, was cultivated on malt extract medium agar (20 g malt extract + 15 g agar) to isolate a single strain. This single yeast strain was proliferated in liquid medium at 30°C for 24 h, and afterwards centrifuged (250 g for 10 min at room temperature), resuspended in PBS, and stored at 4°C.

The EFN of all four species were used to evaluate putative effects of nectar enzyme activities on yeast growth. EFNs were adjusted to a concentration of 5% w/v using a portable refractometer, and a 5% sugar solution (fructose:glucose 1:1) was used as a control. Aliquots (20 μl) of each EFN or the sugar solution were mixed with 20 μl of yeast suspension and incubated for 1 h at 30°C, and then 20 μl of a dilution 1:1000 in PBS was plated on malt agar plates (20 g malt extract + 15 g agar) for determination of the number of colony-forming units after 48 h. Differences among the species were analysed by univariate anova. A Tukey test was posterior applied. EFN from eight plants was used as replicates for each species.

Another assay was performed to evaluate the effects of chitinase activity as found in the EFN of A. cornigera and A. farnesiana on yeast growth. Various sugar solutions with and without Streptomyces griseus chitinase (Sigma) were prepared to create mimics of EFN (see below for composition). A 10 μl aliquot of yeast suspension (commercial yeast SK Saccharomyces cerevisiae) was incubated with 10 μl of mimic nectar for 1 h at 30°C. Various dilution series (1:100 and 1:1000) were used for determination of colony-forming units on malt agar plates (20 g malt extract + 15 g agar) after 48 h. Nectar mimics were prepared to simulate the EFN of one myrmecophyte species (A. cornigera) and one non-myrmecophyte species (A. farnesiana). The A. cornigera mimic comprised an aqueous solution of fructose and glucose (1:1 at a concentration of 6% w/v, the EFN concentration usually found for A. cornigera) with chitinase activity as found in the EFN of this species (0.18 units per ml of sugar solution). As controls, pure water solution and chitinase-free sugar solutions at the same concentration (6% w/v) were used. Mimics of A. farnesiana nectar were prepared using fructose, glucose and sucrose (1:1:1 at a total concentration of 2% w/v) and with chitinase activity as found for this species (0.01 units per ml of sugar solution). Eight repetitions were conducted for each species, and differences among treatments were evaluated separately for each species using a univariate anova. An LSD test was posterior applied.


We thank Juan Carlos Silva Bueno and Ralf Krüger for their help in the field, and Antje Loele, Domancar Orona and Rosa Maria Adame Alvarez for kindly helping with protein analysis and laboratory support. We are also grateful to Dr Arturo Reyes for supporting our work at Universidad del Mar, Puerto Escondido, México. Financial support from the Deutsches Forschungs-gemeinschaft (grant He 3169/4-2) and CONACyT (Consejo Nacional de Ciencia y Tecnología) is gratefully acknowledged. M.G.-T. is supported by a PhD fellowship from the German Academic Exchange Service.