1. Introduction
The Asteraceae are among the largest families of flowering plants with ca. 1700 genera and 23,000 - 26,000 species in all continents except for Antartica [1] , and many species are well known for their different and important uses [2] . In Argentina, it is the largest family with ca. 1400 species, 382 being endemics [3] . Flourensia DC. (Subfamily Asteroideae, tribe Heliantheae) is endemic from America with 32 species; twelve species live in Argentina and six of them are endemic of the central region: F. campestris, F. hirta, F. leptopoda, F. niederleinii, F. oolepis, and F. tortuosa [4] [5] .
Flourensia campestris (FC) and Flourensia oolepis (FO) are easily distinguished by the characteristic glossy aspect of their leaves due to resinous exudates of unknown composition, which represent ca. 40% and 20% of the biomass in young and adult leaves, respectively (Silva and Scopel, Personal communication, 2005). Plant resins, typically constituted by mixtures of volatile and non-volatile phenolic and/or terpenoids secondary compounds, are usually secreted in specialized structures located either internally or on the surface of the plant [6] . These resins, according to their constituents, are known to play important ecological functions, ameliorating or preventing damage from abiotic and biotic stresses in their natural environments (high irradiance, UVB, dissecation, herbivory, competition), whose properties have been used for the management of harmful diseases, insects, pests and weeds [7] - [9] .
FC has been shown to produce a phytotoxic compound identified as (-)-hamanasic acid A {(-)HAA; 7-car- boxy-8-hydroxy-1(2), 12(13)-dien-bisabolene} [10] . This bisabolanoid accumulates at exceptionally high concentrations in some FC organs (e.g., 1.6% of leaf dry weight). There is growing evidence that metabolites that accumulate on plant surfaces are largely produced and secreted by glandular trichomes [11] - [14] . This might be the case of FC, since short dips of fresh and dry leaves and flowers in water or ethyl acetate (EtOAc) showed high concentrations of (-)HAA, accounting for almost 20% - 30% of the total metabolites extracted [10] . In a recent paper, we have demonstrated that its natural occurrence is restricted only to some Flourensia species, strictly FC, FO and F. fiebrigii, being absent in F. hirta, F. riparia and F. niederleinii, and in 37 of the most representative co-occurring species of FC and FO [15] . The composition of the resins from the genus Flourensia has revealed the presence of other phytotoxic compounds, as in F. cernua [16] and in FO [17] - [19] . Other resin components in Flourensia spp. have also been reported to have insect antifeedant [20] - [22] , antifungal [23] , antibacterial [24] [25] , antialgal and antitermite properties [26] , and herbicidal [10] [15] [20] , and to act as tyrosinase inhibitors [27] . To our knowledge, the link between resins secretion and secretory structures in the genus Flourensia has never been addressed before.
In Asteraceae, two secreting systems occur: glandular trichomes at the surface and secretory ducts inside the organs [28] - [30] . Glandular trichome characteristics have been used to classify species within the Asteraceae [31] . Particularly in the tribe Heliantheae, most of the taxa have capitate glandular trichomes that apparently store sesquiterpene lactones [32] - [35] , which have been shown to present biological activities and toxic properties. A high density of glandular trichomes, together with the presence of secretory ducts, was revealed by light microscopic observations in leaves and stems of FC and FO [36] [37] , and of other Flourensia species (F. hirta, F. leptopoda, F. nierderleinii and F. tortuosa) [38] . However, data of secretory structures in flowers of Flourensia, as well as ultrastructural studies, are not available for the genus, and are still limited in the Asteraceae [29] [39] [40] [41] .
In this work, we examined the localization, morphology, density and ultrastructure of secretory structures―in vegetative and reproductive organs―of FC and FO by means of a combination of light, fluorescence, transmission and scanning electron microscopy. We also assessed the possible role of secretory structures in the production and secretion of (-)HAA in FC and FO. The presence and distribution of (-)HAA in different organs were studied by means of 2D-TLC, GC-FID and GC-MS.
2. Materials and Methods
2.1. Plant Material and Study Sites
Plant materials were collected in natural areas corresponding to the Punilla Valley, Córdoba province, Argentina. The study areas were located in shrub communities (total plant cover 70% - 90%), dominated by the evergreen shrubs Flourensia campestris Griseb. (FC) and F. oolepis S.F. Blake (FO). Specimens of FC were collected in “El Dragón” (La Falda), and FO plants were collected in “Dique El Cajón” (Capilla del Monte), in summer (February) of 2009 and 2010. The voucher specimens (BAA 26.498-FC- and BAA 26.499-FO-) are deposited at the Herbario “Gaspar Xuárez” of the Facultad de Agronomía, Universidad de Buenos Aires, Argentina.
2.2. Structural and Ultrastructural Analyses
2.2.1. Light and Fluorescence Microscopy
For anatomical and morphological studies, cross and longitudinal section samples of fully developed capitula (common receptacle, phyllaries, paleae, and ligulate and tubular flowers), young and mature leaves, and stems from each species were obtained. Three developmental stages were considered in stems: primary growth (PG)- corresponding to the early growth of stems during the growing season; secondary growth (with early phellogen activity (SGph)-corresponding to stems between 3 months to one year old, and woody stems (WS)-correspond- ing to stems of 2 + years. The three categories were processed separately. The materials for light microscopy (LM) were fixed in FAA and embedded in paraffin [42] . Sections (7 - 10 µm thick) were cut and stained with safranin combined with fast green (Figure 1A, Figure 2A and Figure 2B), with safranin (Figure 2C and Figures 3A-D),
Figure 1. Secretory structures in capitula and leaves of FC and FO (LM). A FC, longitudinal section of capitula showing a young phyllary (ph) and the corolla of a tubular flower with glandular trichomes (gtr) and secretory ducts (d); B FC, detail of leaf glandular trichome formed by three basal cells, two short peduncle cells and a secretory head composed by six to eight cells; C FO, cross section of a leaf showing the epidermis unistrata with thin cuticle (0.97 µm) and a glandular trichome (adaxial surface); the abaxial surface showed large substomatal chamber (sch) and over level stomata (st); secretory schizogenous ducts are commonly found attached to the vascular bundle (vb), and included in the palisade chlorenchyma. Scale bars: A 75 µm, B 10 µm, and C 50 µm.
Figure 2. Secretory structures observed in transversal sections of primary stems from FC and FO (LM and FM). A FO, young stem showing the epidermis with glandular trichomes (gtr) and their secretions (s), and secretory ducts (d); B FC, young stem with glandular trichomes and ducts; C FC, initial secondary growth showing conspicuous groups of fibers (stained red) at the protophloem; secretory ducts were found near to the primary xylem (pith = p) and in the cortex; D FC, young stem under UV excitation (FM) showing autofluorescence of glandular trichomes, peridermis (hy, 2 - 3 cell layers) and xylem vessels (x); secretory ducts do not autofluoresce (darker green). Scale bars: A and B 100 µm; C and D 250 µm.
or with aqueous toluidine blue (Figures 1B and Figures 1C) [42] , and observed with a Wild M20 microscope. For fluorescence microscopy (FM), similar sections of fresh, unstained materials were examined and photographed using ultraviolet and blue light excitation with a Universal Microscope Axioplan Zeiss. Images were recorded by a digital camera Canon PowerShot A650 IS.
Quantitative Analysis
Density of glandular trichomes (number of secretory structures per tissue surface unit area) in aerial organs was measured in fresh material previously submerged in EtOAc for 20 s in order to remove surface-secreted compounds. Epidermal layers were manually removed, mounted in 50% glycerine, and trichomes were counted under LM, using a digital videomicroscopic camera with the corresponding software (MotiCam 1.3 MP). Density and diameter of secretory ducts were measured in cross sections of common receptacle and stems (PG, WS) while ducts length was measured in longitudinal sections of tissues.
Quantitative density analysis of described parameters was performed in 5 different individuals, in 5 different samples of 1 mm2 from each one. Diameter and length of the ducts were measured in 5 different individuals, in 5 different secretory ducts from each one.
2.2.2. Transmission Electron Microscopy (TEM)
For TEM studies, plant materials were pre-fixed in 2.5% glutaraldehyde in phosphate buffer (pH 7.2) overnight and then post-fixed in OsO4 at 2˚C in the same buffer for 3 h. Following dehydration in a graded ethanol series,
Figure 3. Secretory structures observed in transversal sections of secondary stems of FC and FO (LM). A FO, secondary stem showing secretory ducts (d) in the cortex near the protophloem fibers (phf) (stained red); the most internal ducts found in the secondary phloem (sph) are products of the cambial activity; pith (p), secondary xylem (SX); B Secondary phloem with the presence of fibers (phf) and ducts; C secretory ducts in the pith, attached to the protoxylem with abundant fibers; yellowish resin was found closed to the epithelial cells; D Two young secretory ducts included in the secondary phloem, and close to the phloemathic rays (phr). Scale bars: A 150 µm, B 25 µm, C 75 µm, and D 35 µm.
the material was embedded in Spurr’s resin. Thin sections (1 µm) were made with a Reichert-Jung ultramicrotome and stained with toluidine blue for observation with light microscope. Ultrathin sections (750 - 900 nm) were made with the same ultramicrotome and then stained with uranyl acetate and lead citrate [43] . These sections were observed and photographed with a JEOL 1200 EX II.
2.2.3. Scanning Electron Microscopy (SEM)
Plant materials were fixed with FAA (2.2.1), dehydrated in a graded ethanol series and coated with a thin layer of gold. Observations were carried out on a Zeiss scanning electron microscope at 15 kV.
2.3. Phytochemical Analyses
Based on the results of type, distribution and density of secretory structures in the plant organs studied (capitula, leaves and stems), different extraction methods were performed in order to investigate the possible role of the two different secretory structures in (-)HAA storage and secretion.
The presence and content of (-)HAA in FC and FO was assessed in whole tissues, on the tissues surface (surface secreted resins), and in the resins collected directly from secretory ducts in pooled materials harvested from 5 different individuals.
2.3.1. Tissue Extractions
Whole Tissue: For whole organs, tissue extracts were obtained from grinded fresh material (50 mg), extracted twice using EtOAc (1.5 mL and 0.5 mL), vortexed for 2 min and centrifuged (10 min, 3000 g). The EtOAc phases were pooled, dried at 30˚C under N2 flow and weighed [10] .
Surface Secreted Resins: Fresh organs were washed with gentle agitation in cold EtOAc during 20 s (short dips) [10] [13] [14] . Special care was taken for dipping intact organs in order to avoid contamination from endogenous sources. This procedure did not damage the surface of the tissues as revealed by LM inspection.
Duct Resins: Resins produced in secretory ducts were collected from fresh stems of FC and FO, after peeling the external tissues and allowing the products to exude for ca. 10 min at ambient temperature. Resins were picked up using a micro syringe, weighed, and stored at −20˚C until analysis.
2.3.2. (-)HAA Analysis
(-)HAA was identified and measured by means of GC-FID and 2D-TLC [10] [15] , yielding comparable results. For the 2D-TLC, (-)HAA concentration was determined by digital analysis (ImageJ http://rsb.info.nih.gov/ij/index.html). The integrated area (relative density) of a spot corresponding to (-)HAA was compared to that of a standard (-)HAA (purified from plant tissue, not commercially available). The relation between the concentration of the standard (-)HAA and the integrated areas of the spots (relative density values) proved to be linear between 0.5 mg∙mL−1 and 2 mg∙mL−1 (r2 = 0.94). Inter batch spot (I2) variations were minimized by loading the same volume of sample (3 µL) against a fixed concentration of standard (-)HAA (1 mg∙mL−1), and running the 2D-TLC as already described by Silva et al. [10] . Means from triplicates were expressed as mg (-)HAA per g dry weight (mg∙g−1 DW) in the case of whole tissue extractions, and as mg (-)HAA per 100 mg of total surface secreted or duct resins (mg % w/w).
(-)HAA was also investigated in duct resins by GC-MS on a Perkin Elmer Clarus 600, and TurboMass 5.4.2 for data acquisition. The GC column was DB5 (60 m, 0.25 mm ID, 0.25 µm particle size) (Perkin Elmer) and the carrier gas used was He (49.6 psi). The initial temperature of 60˚C was gradually increased after 2 min to 300˚C by a ramp of 5˚C∙min−1 and held for 10 min at 300˚C. The injector was used in split mode (20 mL∙min−1) with the inlet temperature set to 250˚C. Samples were diluted in acetone or in EtOAc as follows: in 100 µL of solvent and 1/50 µL of solvent, the injection volume was 1 µL. Some peaks were identified by comparison of mass spectra using NIST MS Search 2.0., and using previous own data from volatiles in the same species [10] . Pure (-)HAA in this system showed a retention time of 23.5 min and m/z (%): 234 (12)[M+H2O], 219 (2), 191 (6), 164 (3), 149 (14), 123 (9), 109 (100), 105 (9), 79 (22), 69 (65), 55 (20), 43 (35), 39 (6).
2.4. Data Analysis
One way ANOVA or Mann-Whitney nonparametric test for the significance of the difference between the distributions of independent samples for ordinal data were applied.
3. Results
3.1. Structural and Ultrastructural Analyses
3.1.1. Light and Fluorescence Microscopy
Secretory structures in the capitula of both species were composed by glandular trichomes and secretory ducts (Figure 1A), similar to those observed in leaves (Figures 1B and Figures 1C), and stems (Figure 2 and Figure 3). The capitates glandular trichomes are biseriate with 6 - 8 head celled, a two celled short-stalk (with cutinized lateral walls) and three basal cells (Figure 1B). No significant differences in the density and distribution of glandular trichomes between the capitula of both species could be found (Table 1). As shown in Figure 1A, glandular trichomes are very abundant in paleae and phyllaries primordia. In fully developed structures, the density of glandular trichomes in the capitula was higher in the paleae, followed by the phyllaries, tubular and ligulate flowers (ca. 2.5× higher in paleae cf. to ligulate flowers, Table 1). In the bracteae, most of the glandular trichomes were found along the edges and veins, while in the corollas of ligulate and tubular flowers ca. 80% of the trichomes were located in the distal portion of these structures and associated to veins. For both species, the density of trichomes in capitula was significantly smaller than that found in leaves and stems (Table 1). No other type of glandular trichomes was detected in the capitula from both species.
Table 1. Density of glandular trichomes in capitula, leaves and stems of FC and FO.
Data (media ± SE, n = 25) represent the number of glandular trichomes per mm2. PG: stems with primary growth; SGph: stems with secondary growth and incipient phellogen activity; WS: woody stems; Letters (a - f) indicate significant differences among tissues; asterisks (*) show differences between species (p < 0.05); ND: not detected.
In leaves, glandular trichome density on the adaxial surfaces of FC was higher than in FO (Table 1). Stems (PG and SGph) in both species, showed significantly higher glandular trichome densities than the other organs; between 1.5 (FC) and 7.3 (FO) times higher than those found in the leaves and capitula (Table 1). Densities of glandular trichomes in PG and SGph stems were significantly higher in FO as compared to FC (Table 1). Glandular trichomes were absent in WS of both species (Table 1).
Secretory ducts were observed in all the floral pieces of both species and were similar in shape to those observed in the cross section of the common receptacle (data not shown). Density and diameter of secretory ducts present in the common receptacle were significantly different in FC as compared to FO (Table 2). Density was 4× smaller in FC than in FO, while diameter was ca. 1.7× larger in FC relative to FO (Table 2). In stems, the distribution and localization of secretory ducts in both species are shown and described in Figure 2 and Figure 3. The density of secretory ducts showed significant differences between developmental stages in both species; the smaller values corresponded to stems with primary growth (PG), and the larger to stems with secondary growth (WS) (Table 2, Figure 2 and Figure 3). Duct diameters followed a similar pattern, spanning from 17 μm to 165 μm. Ducts in stems with SG (cortex and secondary phloem) were significantly larger than those in PG stems; the highest values corresponded to FC (Table 2). In the longitudinal plane, these ducts could be seen as elongated spaces, between 225 μm to 1237 µm long -for older ducts located in the cortex-, and 50 μm to 275 µm for ducts present in the secondary phloem. Estimations based on the diameters and densities of FC ducts in SG stems at both phases indicate that at transversal sections ducts occupy between 0.025% (pith) and 2.4% (cortex) of the surface. Microscopical inspections of the ducts performed since early stages of development suggested a schizogenous origin (Figure 2B and Figure 2C), partially disintegrated cells being never observed in the periphery of the space. The presence of an epithelium surrounding the duct confirmed the schizogenous origin.
3.1.2. Transmission and Scanning Electron Microscopy
There were no ultrastructural differences between the glandular trichomes of FC and FO, with the exception of a broken cuticle commonly observed in FC. The secretory cells of the trichome head have a dense cytoplasm and small vacuoles (Figure 4). The outer tangential cell wall of these cells has two layers, the inner with a more electron-dense fibrillar structure than the external translucent structure pecto-cellulosic layer (Figure 4A and Figure 4B). This wall is coated by a detached cuticle that creates a subcuticular chamber (Figures 4A-C). The cytoplasm shows the presence of abundant mitochondria (Figure 4A), dictyosomes (Figure 4E), endoplasmic reticulum (Figure 4F), chloroplasts with numerous starch grains (Figure 4G), and leucoplasts (not shown). The lipophilic vacuoles have an elaborated system of inner membranes and numerous lipidic globules (Figure 4F). The dictyosomic vesicles are abundant; some are observed in contact with the plasmalemma (Figure 4B). In active glands, the outer tangential wall of the head cells presents cavities filled with substances of moderate to high electron-density (Figure 4C and Figure 4D). Substances with the same electron-density are also observed in the subcuticular chamber where secretions accumulate, giving the trichome head a spherical shape. Scanning electron images of the abaxial surface of mature leaves of FC (Figure 5A) and FO (Figure 5B) show glandular and non-glandular trichomes on the veins. A detailed inspection of the images in FC allowed to observe the presence of glandular trichomes with a ruptured or completely removed cuticle (Figure 5D and Figure 5E). In contrast, no pores or ruptures could be observed in FO (Figure 5C). These observations, together with the TEM images described (Figure 4), suggest that each species would differ in the way they release the products that accumulate in the head cells. In FC, the products would accumulate temporarily in the interfibrillar spaces of the cell wall, and transferred to the subcuticular chamber where they would accumulate until the cuticle fully distends
Figure 4. Ultrastructure (TEM) of glandular trichome cells in FC and FO. A FO, secretory cell, outer tangencial cell wall (cw) showing an inner layer with more electron-dense fibrillar structure (iw) and an external layer (ew) with a thick loose and detached cuticle (c) an electron-dense cytoplasm content, presence of mitochondria (m) and endoplasmic reticulum (ER); B Detail of previous micrograph, the arrows indicates a possible pathway of the secreted substances (s) across the cell wall and their accumulation in the subcuticular chamber (ch); dotted line indicates the external limit of the cuticle; C FC, secretory cell wall with interfibrillar spaces filled with secretions; secretions should be released through a ruptured cuticle; D FC, arrows indicate lipophilic substances in the interfibrillar spaces of the cell wall; E FC, secretory cell showing dense cytoplasm with dictyosomes (d); F FO, a view of the cell content with endoplasmic reticulum (ER) attached to the cell wall, and lipid globules (lg); G FC, detail of a chloroplast in the secretory cell with starch grains (sg). Scale bars: A 0.3; B, C, D and F 0.1 µm; E 0.2 µm; G 0.5 µm.
Table 2. Density and diameter of secretory ducts in capitula and stems of FC and FO.
Data are expressed as mean ± SE, n = 25. PG: stems with primary growth; WS: woody stems. Letters (a - d) indicate significant differences among tissues; asterisks (*) show differences between species (p < 0.05).
Figure 5. Scanning electron micrographs of leaf glandular trichomes in FC and FO. A-B FC and FO, respectively, abaxial epidermis showing the distribution of glandular (gtr) and e-glandular trichomes (tr) mostly on the main vein, and over-level stomata (st); C FO, head of a glandular trichome, no pores were observed; D FC, ruptured cuticle (arrows) in the head of a glandular trichome; E FC, glandular head with a totally removed cuticle (arrow). Upper left corner white vertical scale bars: A 15 µm; B 30 µm; C and D 3 µm; E 10 µm.
and ultimately breaks releasing the products (Figure 4C, Figure 4D, Figure 5D and Figure 5E). In contrast, in FO, the secretions of the accumulated products seem to be released through a loose cuticle, without showing any pores or cracks (Figure 4A, Figure 4B and Figure 5C).
In secretory ducts, the lumen is surrounded by a layer of specialized cells, the epithelium (Figure 6A). These cells have thin walls, conspicuous nucleus, vacuoles, abundant mitochondria, some plastids, rough endoplasmic reticulum and free ribosomes (Figure 6A and Figure 6B). The vacuoles have varied contents; some of them show an electron-dense material similar to lipid globules (Figure 6A). The plasmalemma is slightly invaginated and a fibrillar material is observed between it and the cell wall. This characteristic is more evident in the sub-epithelial cells, which have numerous plastids (including chloroplasts) and big vacuoles with contents that present moderate electron-density and some inclusions (Figure 6A).
3.2. Phytochemical Analyses
Whole tissue concentrations of (-)HAA in all aerial organs of FO were significantly lower than those originally reported in FC (Table 3); 14× to ca. 20× lower in capitula and leaves, respectively, and 2× lower in PG stems. (-)HAA distribution among organs differed between species (Table 3). While in FC the higher levels of (-)HAA were detected in capitula and leaves, in FO the highest concentration corresponded to stems with PG (Table 3). In whole stems (-)HAA content showed an age-dependent distribution, with highest levels being present in the young stems (PG) followed by SGph, and by negligible levels of (-)HAA detected in WS (Table 3).
Figure 6. Ultrastructure (TEM) of leaf secretory duct cells in FC and FO. A FC, epithelial cells from ducts showing vacuoles (v) and a conspicuous nucleus (n), and sub-epithelial cells showing wall ingrowths and vacuoles; B FO, duct epithelial cell showing the nucleus, abundant mitochondria (m), and granular endoplasmic reticulum (ER); secretions (resins) are released across the cell wall (cw) into the duct lumen. Lower A and upper B left corner white scale bars: A 2 µm; B 0.2 µm.
Table 3. Content of (-)HAA in whole tissues and resins of FC as compared to FO.
(-)HAA levels in whole tissues of FC and FO (mean ± SE; N = 3). Values in FC (Silva et al. 2012) are shown for comparison purposes. NA: not applicable; ND: not detected; PG: primary growth; SGph: secondary growth and incipient phellogen activity (last season); WS: woody stems (2 + yrs). Letters (a - d) indicate significant differences among tissues; asterisks (*) show differences between species (p < 0.05).
Surface secreted resins obtained from short dip extractions with EtOAc revealed the presence of (-)HAA in all aerial organs; concentrations being within the range of those found in whole tissue (Table 3). (-)HAA in whole tissues expressed as relative concentrations (mg % w/w) of the total resins extracted, tended to yield lower values than those found in surface secreted resins (data not shown), as could be observed in the 2D-TLC profiles (Figure 7A, PG stems). These differences may be accounted for a dilution effect in whole tissues, since the extractions included both, surface secreted and internal compounds (including duct resins). The presence of (-)HAA among the surface resins of the organs was also evidenced by GC-MS (Figure 7C, inset, PG stems). Chromatographic profiles of surface secreted resins differed between species, and slight differences were also observable among organs in each species (data not shown).
In duct resins collected from fresh stems of both species, (-)HAA was not detected by means of 2D-TLC or GC-MS (FC: Figure 7B and Figure 7C, respectively); confirming that (-)HAA only pertains to surface secreted resins.
The 2D-TLC chromatographic profiles of whole tissue of PG stems also revealed the presence of a fast chromatographic fraction represented by less polar compounds relative to (-)HAA, which was also detected in duct produced resins (cf. Figure 7A―right vs. Figure 7B―left). GC-MS analysis of stem duct resins of FC and FO
Figure 7. Identification of (-)HAA in FC young stems. A-B 2D-TLC, long broken and whole arrows indicate the first and second dimensional runs, respectively. A (-)HAA (small arrows) in surface resins (left) vs. whole tissue extract (right); additional compounds from the internal tissue can be observed in the latter; B Duct resins (left) vs. pure (-)HAA (right, small arrow); duct resins show no (-)HAA and the same compound profile observed in whole tissue extract; C GC of surface resins showing the presence of (-)HAA (small arrow) inset, compared to GC of duct resins showing a distinctive profile in which (-)HAA is absent.
showed distinctive profiles. From a total of 30 compounds found in each species only four (13%) were present in both; the volatiles santolinatriene and spathulenol could be identified by this analysis. Remarkably only 7 of these metabolites accounted for 75% (FC) and 90% (FO) of the total duct resins composition.
4. Discussion
Results showed that both species present two types of secretory structures: glandular trichomes-capitate wit3 h a biseriate-pluricelular head, and secretory ducts, confirming the observations of Delbón et al. [36] [37] . Biseriate glandular trichomes are typically found in Asteraceae, in different tribes and genera, such as Ambrosia trifida [44] , Helianthus annuus [45] [46] , Sigesbeckia jorullensis [41] , Grindelia pulchella/Heliantheae [29] , Centrapalus pauciflorus and Chrysolaena spp./Vernonieae [28] [47] , respectively, Helichrysum [48] and Inula/Inuleae [49] , Artemisia annua [50] and Tanacetum cinerariifolium/Anthemideae [51] .
Present observations demonstrate, for the first time in the genus Flourensia, the existence of glandular trichomes in reproductive organs. These were morphologically identical to those found in the vegetative organs of both species studied, and were also similar to those described in the vegetative organs of other species of the genus, like F. hirta, F. tortuosa, F. niederleinii and F. leptopoda [38] . Observations of Plectranthus ornatus (Lamiaceae) reproductive organs revealed the presence of an unusual kind of glandular trichome in respect to those found in vegetative organs [52] . In several species belonging to subfamily Lamioideae (Labiatae) a new type of capitate trichome, exclusive of calices and corollas, was described [53] . This are not the case for the species of Flourensia studied, and may constitute an important trait for the taxonomy in the genus. Similar studies in the reproductive organs of other species from the genus are needed to confirm these findings.
Despite the density of glandular trichomes found in capitula was lower than that found in leaves, and much lower than that found in stems, these trichomes may still play relevant ecophysiological functions in the reproductive organs. In this sense, secretions of glandular trichomes in floral structures have been shown to act as pollinator attractants and/or to be involved in the chemical defense against herbivore insects and microbes [52] [54] .
The densities of glandular trichomes in the leaves of both species were very similar to those previously reported by Delbón et al. [36] , while their densities in stems were never reported before. In contrast to reproductive organs, glandular trichomes in the vegetative organs of FC and FO exhibited a different pattern of distribution. While in FC leaves and stems with PG and SGph showed very similar densities, in FO, the higher densities were found in stems. Moreover, trichomes density in FO stems was between ca. 2× (SGph) to 4× (PG) higher than those found in FC. The observed striking difference, in addition to their differences in gland secretion compositions as observed in the present, may be involved in These unexpected results between two species that do not share populations [4] , pose the question of whether the increased number of trichomes may be involved in organ-specific functions that may be relevant for their corresponding adaptations through habitat specialization [6] . The ultrastructure of the head cells of glandular trichomes revealed components typically found in cellular structures that produce and secrete metabolites [29] [55] , although some glandular trichomes in Asteraceae only present smooth endoplasmic reticulum [28] . More specifically, the presence of abundant chloroplasts with visible starch grains in FC and FO is indicative of their functionality. Starch grains have been identified in leucoplasts and chloroplasts in the head cells of mature glandular trichomes of Stevia rebaudiana (Asteraceae) [56] [57] .
Detailed observations of SEM and TEM images suggest that each species would differ in the way they release the secretions to the plant surface. In FC, cuticle rupture was commonly observed. The products would accumulate temporarily in the interfibrillar spaces of the cell wall, and would be transferred and stored in the subcuticular chamber until the cuticle fully distends and ultimately breaks releasing the products. In FO, instead, cuticle rupture or pore formation was not observed by SEM in any of the materials inspected. In this species, the secreted products seem to be released to the plant surface through a loose cuticle.
Storage of secretions in the subcuticular space and cuticle rupture has been reported for other Asteraceae species [28] [29] [41] [48] [49] [58] . On the other hand, Fahn [55] reports that secretion by the apical head cells continue as long as the cells on the peduncle remains functional, which might be the case for FO.
A growing body of experimental evidence shows that terpene biosynthesis takes place within glandular trichomes [11] [12] [59] - [63] . As an example, transcriptome analysis of Artemisia annua glandular trichomes revealed the presence of contigs corresponding to multiple enzymes for terpenoid and flavonoid biosynthesis suggesting their important metabolic activity [64] . In the tribe Heliantheae, most of the studied species seems to accumulate sesquiterpene lactones in their glandular trichomes [65] . The observations of FC and FO under UV revealed a green-yellow autoflorescence inside the trichomes and on the surface of the cuticle where the secretions are released, typical of phenilpropanoids and terpenoids as those found in Lippias caberrima (Verbenaceae) [66] . Our results show that the sesquiterpenoid (-)HAA concentration was positively associated with the distribution of glandular trichomes in the different organs of both species. In FC, the higher densities of glandular trichomes were found in capitula and leaves, followed by stems with PG, and (-)HAA concentration closely resembled this pattern (cf. FC, Table 1 and Table 3). This was also the case for FO, in which the highest density of glandular trichomes in stems with PG corresponds to the highest levels of (-)HAA detected (cf. FO, Table 1 and Table 3). Although glandular trichomes are still very conspicuous in stems with SGph, the lower concentration of (-)HAA may result from a lower number of functional trichomes. A developmental stage-specific expression in glandular trichomes has been reported in other Asteraceae as Helianthus annuus [46] and Tenacetum parthenium [12] . In Artemisia annua, trichome collapse has been proposed for the decrease in artemisinin content during later stages of leaf development [67] . It is worthwhile to note that concentration of (-)HAA was always higher in FC than in FO, regardless of the organ and trichomes density. This might be due to a differential composition of the metabolites produced by the trichomes in each species, in which (-)HAA may not be the main compound in FO.
Differences in the chemical composition of glandular trichomes, even at within-plant level, have also been reported. In this sense, Appezzato-da-Glória et al. [28] found that glandular trichomes from underground organs (rhizophorous) of Chrysolaena platensis and C. obovata (Asteraceae) produced a higher diversity of compounds compared to those from the leaves. In the present work, slight differences in surface secreted resins profiles among the organs of FC and FO were found (data not shown). This variation in the mixture of chemical compounds may play a central role in ecological interactions of resins as already reviewed [6] .
The mild dipping extraction method of intact organs, which did not damage the epidermis or external tissues, proved that (-)HAA is only present in the surface secretions found in glandular trichomes bearing organs. The presence of small amounts of (-)HAA on the surface of WS may derive from the product being lixiviated by rain from the younger portions of the branch. The fact that (-)HAA is easily leached out by water [10] supports this notion.
In order to further investigate the possible role of secretory structures specifically related to the storage and secretion of (-)HAA, we compared the concentrations of (-)HAA found internally (secretory ducts) with those found on the surface of the plant (glandular trichomes). In organs such as leaves and capitula, in which resins from internal tissue were very difficult to obtain without surface contamination, whole tissue extractions were compared against surface resins. Results from 2D-TLC and GC-MS (as shown in Figure 7) demonstrate that (-)HAA is more concentrated in the surface resins of glandular trichome-bearing organs. The absence of (-)HAA inside organs like stems, where resins could be collected internally, demonstrate that this sesquiterpenoid is not produced by the secretory ducts. These facts indicate that (-)HAA is not translocated from the interior to the plant surface, but strongly suggest that it is synthesized, stored and secreted by glandular trichomes.
Based on the above, we propose that glandular trichomes in FC and FO are the secretory structures involved in the production of this putative phytotoxic allelochemical. These results are also consistent with those originally described by Hashidoko et al. [68] , where he found precursors of the (-)HAA isomer ((+)HAA) in the glandular trichomes of Rosa rugosa.
In respect to secretory ducts, the importance of resin production in Asteraceae has been long recognized by Carlquist [69] in an extensive revision of the wood anatomy of Asteraceae. Robinson [70] used morphological characteristics and resin composition of secretory ducts in phylogenetic studies to define the subtribes of the tribe Heliantheae, including the genus Flourensia in the subtribe Ecliptinae. The morphology and distribution of secretory ducts in vegetative tissues (leaves and stems) of FC and FO coincided with those already described by Delbón et al. [37] , and were similar to the ones found in reproductive organs. Using LM inspections, Delbón et al. [37] classified these ducts as shcizogenous. The presence of conserved epithelial cells, with no damaged membranes, and a well-defined clear lumen, as well as the observations at early stages of development strongly suggest the schizogenous origin of the ducts.
Albeit the importance and widespread existence of secretory ducts in woody shrubs of arid environments, morphometric data is rarely presented. In both Flourensia species, the highest density of secretory ducts was observed in stems with SG, specifically in the secondary phloem, while the larger diameters corresponded to ducts in the cortex of SG stems. Estimations based on the diameters and densities of ducts indicate that ducts in SG stems may represent as much as 2.6% per unit of volume (1 µm3), emphasizing the importance of this secretory system in FC and FO. In two varieties of Senecio filaginoides (Asteraceae), Feijóo et al. [71] reported higher density and diameter of secretory ducts in PG stems, relative to leaves. Densities (4 - 10 mm−2) were similar to those found in PG stems of FC and FO, while diameters were larger (47 µm to 203 µm, spring and winter, respectively) in S. filaginoides compared to those found in Flourensia. Diameter of secretory ducts is the most important trait that determines resin yield in conifers [72] , and also in Boswellia papyrifera (Burseraceae), a resinous tree native to Ethiopia [73] . In the latter, average diameter of axial resin ducts (113 µm) was similar to those found in FC and FO in SG stems; a significant trade-off between density and average diameter of secretory ducts was also found in B. papyrifera.
In FC and FO, the localization of the ducts in the secondary phloem, close to the cork, would facilitate the secretion of products to the surface of stems through the lenticels or cracks (by the new phellogen activity near the surface). In fact, in plants growing in their natural environment, stems usually present sticky depositions (bulk resins) that have been secreted to the surface as resin drops of different colors (yellowish to reddish) and viscosity, probably due to different degrees of oxidation.
The ultrastructural analysis of secretory ducts of FC and FO revealed the presence of chloroplasts with starch accumulation in the epithelial and subepithelial cells. The presence of abundant lipid globules in the dense cytoplasm strongly suggests their involvement in the synthesis of resins. Wall ingrowths, typically associated with transfer processes [74] , were observed in the epithelial and subepithelial cells of secretory ducts in both Flourensia species. According to Herrero and Dickinson [75] , the cell wall ingrowths associated to the plasmodesmata assures a great efficiency in the cellular exchange. Bartoli et al. [29] also described the presence of a special tissue associated to secretory ducts in Grindelia pulchella (Asteraceae).
Preliminary results from GC-MS analysis of resins collected directly from stem ducts of FC and FO showed very distinct profiles for each species, and no (-)HAA was detected as discussed above. Only some monoterpenic and sesquiterpenic volatile compounds already reported (FC: [10] ; FO: [22] ) could be recognized (data not shown). Despite the matching of most compounds with those from GC-MS libraries was generally higher than 75%, their complete identification was not possible and is under research. In their natural environment these compounds would serve as defense weapons against biotic stressors such as microbes and insects. The complete characterization of the copious resins present in the different organs deserves further investigation in order to link composition to biological activities in nature, and to explore potential sources for new biopesticides and other industrial applications.
5. Conclusions
Reproductive organs of FC and FO presented the same type of secretory structures (glandular trichomes and secretory ducts) described in vegetative organs. Our observations demonstrate, for the first time in the genus Flourensia, the presence of glandular trichomes in reproductive organs. These capitate glands were morphologically identical to those found in vegetative organs of all Flourensia species studied so far, and might constitute an important trait for the taxonomy of the genus.
Glandular trichome’s density was very similar in reproductive organs of both species. In contrast, densities of glandular trichomes in vegetative organs were significantly different between the two species; the highest density corresponded to FO stems.
Secretory ducts varied in density and diameter among the organs of both species, with the combination being maximal in stems with SG. TEM and SEM observations of glandular trichomes suggested that each species differed in the way in which secretory materials were released to the outside: through cracks or pores in FC, or through a loose cuticle in FO. In secretory ducts, the localization of lipophilic vacuoles and the presence of wall ingrowths in subepithelial cells suggest the existence of a granulocrine mechanism of secretion. Similar inspections performed since early stages of ducts development revealed a schizogenous origin.
Our studies are the first to report the distribution of the sesquiterpenoid (-)HAA in different organs of FO. Concentration of (-)HAA in all organs was significantly lower than in FC, as measured in whole tissues, with the highest contents being present in stems with primary growth. In both species, (-)HAA was only detected in surface secreted resins from those organs where glandular trichomes were also present (capitula, leaves and stems). The concentration of (-)HAA was closely and positively related to the density of the glandular trichomes. In addition, no (-)HAA was detected in the resins collected from secretory ducts. These findings strongly suggest that glandular trichomes are the sites of accumulation and secretion of (-)HAA. Biochemical investigations on the biosynthetic capacity of isolated glandular trichomes to produce (-)HAA are under way. Preliminary data on resins composition―from resins extracted directly from secretory ducts in stems― show distinctive profiles for FC and FO; only 13% of the compounds detected are shared by both species.
In addition to the elucidation of ultrastructural traits, distribution and density of secretory structures in reproductive and vegetative organs of FC and FO, our findings suggest a functional role for glandular trichomes in the secretion of the putative phytotoxic allelochemical (-)HAA.
Acknowledgements
This research was supported financially by grants from the Agencia Nacional de Promoción Científica y Tecnológica (ANPCyT) Grant PICT 0411, the Universidad de Buenos Aires, UBACyT 0566, and the Ministerio de Ciencia y Tecnología, Córdoba, Grant PID 2010. We are most grateful to Augusto Maillet for his dedication and technical assistance in the laboratory.
NOTES
*Corresponding authors.