Differential Metabolic Specialization of Foliar Oil Glands in Eucalyptus

Tree Physiology 38, 1451–1460 doi:10.1093/treephys/tpy077

Research paper

Diﬀerential metabolic specialization of foliar oil glands in Eucalyptus brevistylis Brooker (Myrtaceae) Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021

† † Jason Q.D. Goodger 1,4, , Samiddhi L. Senaratne1, , Dean Nicolle2 and Ian E. Woodrow3

1School of BioSciences, The University of Melbourne, Melbourne, Victoria 3010, Australia; 2Currency Creek Arboretum, PO Box 808 Melrose Park, Currency Creek, SA 5039, Australia; 3School of Ecosystem and Forest Sciences, The University of Melbourne, Melbourne, Victoria 3010, Australia; 4Corresponding author ([email protected]) orcid.org/0000- 0001-7392-3891

Received January 7, 2018; accepted June 6, 2018; published online July 19, 2018; handling Editor Ülo Niinemets

Trees and shrubs from the genus Eucalyptus are characterized by the presence of numerous foliar oil glands that generally house mono- and sesquiterpenes. In some species, glands are also known to house substantial quantities of unrelated secondary metabolites such as volatile, aromatic β-triketones. It is not known if these compounds are co-housed with terpenes or if they are produced in distinct, metabolically specialized glands. We showed that Eucalyptus brevistylis—a species with appreciable foliar quantities of both β-triketones and terpenes—contains two visually distinct gland types in leaves, one that is translucent and the other golden-brown. Gas chromatographic analyses of solvent extracts of the two gland types showed that the translucent glands contain sesquiterpene alcohol cubenols and cubebols (termed ‘sesquiterpene glands’), whereas the golden-brown glands contain predominantly the β-triketone conglomerone with lesser amounts of sesquiterpene hydrocarbon caryophyllenes (termed ‘triketone glands’). Analysis of leaves from trees of different ages, from young saplings through to advanced age trees, showed a gradual increase in the abundance of sesquiterpene glands relative to triketone glands as plants aged. Such ontogenetic regulation of foliar secondary metabolite concentration appears to be a common feature of Eucalyptus species, albeit at different temporal scales. A similar ontogenetic pattern was observed in ageing leaves, with mature leaves having a higher proportion of sesquiterpene glands than young leaf tips. It is concluded that regulation of the relative abundances of the two gland types with ontogeny likely reflects the different herbivores present at the different life stages of leaves and whole plants. In particular, leaf tips and young plants may be advantaged by deploying higher amounts of insecticidal β-triketones. The concurrent deployment of two metabolically distinct gland types in leaves is a rare phenomenon and a novel finding for myrtaceous trees.

Keywords: caryophyllene, conglomerone, cubebol, eucalypt, metabolomics, secretory cavity, sesquiterpene, terpene, terpenoid, triketone.

Introduction − between species with gland densities as high as 3400 cm 2 in Trees and shrubs from the genus Eucalyptus are characterized Eucalyptus protensa (Brooker and Nicolle 2013) and mean by the presence of numerous oil glands embedded within their gland volumes as large as 10 nl in Eucalyptus apiculata leaves. Each sub-dermal gland is a spherical, multi-cellular com- (Goodger et al. 2016). Consequently, terpenic oils can comprise plex composed of outer secretory cell layers surrounding a large over 6% leaf fresh weight in high oil-yielding species such as extra-cellular lumen (Carr and Carr 1970). The secretory cells Eucalyptus polybractea (Goodger et al. 2008). The constituent produce a range of mono- and sesquiterpene oils that are then terpenes housed in glands can also diﬀer between Eucalyptus stored within the lumen. The number and size of glands vary species (Brophy et al. 1991, Brophy and Southwell 2002), and

† These authors contributed equally to this work.

© The Author(s) 2018. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected] 1452 Goodger et al. impart distinctive aromas, some of which are valued for pharma- results were not consistent between all species, but did suggest ceutical, industrial and perfumery uses (Doran 1991). that at least in some species there is a level of metabolic special- Moreover, the different terpene profiles produced are thought to ization between the two embedded gland types. For example, serve important ecological roles, particularly as herbivore deter- the flavonoid rutin was detected only in dark glands of Hypericum rents (Harborne 1991, 2001) or as reducers of ozone toxicity maculatum and Hypericum erectum. Nonetheless, very low quan- (Li et al. 2018). tities of rutin were found in both dark and translucent glands of Individual foliar glands of Eucalyptus are known to contain Hypericum perforatum. The authors suggested the apparent lack mixtures of both monoterpenes and sesquiterpenes (King et al. of metabolic specialization in some species may have been arte- 2006), and analysis of extracts of bulk glands have also shown factual given the difficulty they had in imaging through the thick they can house substantial quantities of other chemicals such as leaf cuticles of those species (Kucharíková et al. 2016b).

volatile, aromatic β-triketones, non-volatile flavanones or non- In this study, we determine if β-triketones and terpenes are Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021 volatile monoterpene acid glucose esters (Heskes et al. 2012a, biosynthesized and stored in the same glands of a chosen Goodger et al. 2016). The monoterpene acid glucose esters Eucalyptus species or if there is metabolic specialization of gland have been better studied and are known to co-occur with terpe- types for the different compound classes. Volatile β-triketones nic oils in individual glands where they are localized to the outer have long been found in steam distillates of leaves from particu- region of lumena, with the terpenes more centrally located lar Eucalyptus species (Hellyer 1968, Brophy et al. 1991), but (Heskes et al. 2012a, 2012b). This unique extra-cellular spatial their specific localization to oil glands was only recently shown arrangement is thought to be mediated by the ampipathic nature (Goodger et al. 2016). The research by Goodger et al. (2016) of the compounds, being comprised of glucose centres and ter- used an enzymatic leaf digestion protocol to isolate glands and pene acid side groups, and may protect glandular secretory cells show the presence of the β-triketones conglomerone, agglomer- from the potentially auto-toxic terpenes stored within lumena one and iso-baeckeol methyl ether in Eucalyptus suberea glands (Goodger and Woodrow 2011). It is not known if flavanones or and conglomerone in Eucalyptus brevistylis glands. Similarly, the β-triketones are similarly housed together with oils in Eucalyptus use of Raman microscopy recently localized the β-triketone glands or if they are produced in distinct, metabolically specia- grandiflorone to the foliar glands of the myrtaceous shrub spe- lized glands. cies Leptospermum scoparium and Leptospermum morrisonii Such differential metabolic specialization of foliar oil glands (Killeen et al. 2015). Nevertheless, neither of those studies has not previously been observed in myrtaceous trees, but has showed if triketones co-occurred with oils in glands or if there been reported from a few herbaceous plants. The most definitive were two distinct types of glands responsible for the storage example comes from Salvia sclarea (clary sage; Lamiaceae), of triketones and terpenes, respectively. We chose to address which possess morphologically distinct peltate and capitate glan- this question using E. brevistylis as its glands are amenable to dular secreting trichomes (GSTs) protruding from epidermal isolation with the aforementioned enzymatic protocol and layers of above-ground organs. Gas chromatographic analysis of because extracts from bulk glands contained moderate individual GSTs collected from petal surfaces found capitate amounts of both triketones and terpenes, whereas those from glands mainly produce the monoterpene linalool together with E. suberea glands were almost exclusively comprised of trike- the diterpene sclareol, whereas peltate glands largely store the tones, with only very low amounts of sesquiterpenes detected sesquiterpene germacrene D (Schmiderer et al. 2008). Another (Goodger et al. 2016). Lamiaceae plant, Pogostemon cablin, possesses glandular trichomes on the surfaces of leaves and stems as well as internally Materials and methods embedded glands within the same tissues. Non-quantitative, histochemical characterization of the glandular constituents of Plant material both gland types similarly suggested metabolic specialization of Eucalyptus brevistylis Brooker (Rate’s tingle) is a large tree spe- gland types as staining was consistent with internal glands con- cies in the family Myrtaceae that is capable of growing up to taining lipids, flavones and terpenes and external GSTs housing 40 m tall (Brooker 1974). The species is endemic to the state of polysaccharides and terpenes (Guo et al. 2013). Western Australia (WA) where it holds the conservation status of Further evidence of metabolic specialization of gland types Priority 4: Rare, Near Threatened (Department of Parks and comes from Hypericum species (Hyperaceae) that contain two Wildlife 2015, Western Australian Herbarium 2017). Trees are types of internally embedded glands termed ‘dark’ and ‘translu- found in a very limited distribution in the Walpole area, where cent’ glands (Soelberg et al. 2007, Li et al. 2013, Kucharíková they are restricted to more fertile soils of higher clay content et al. 2016a, 2016b). In particular, a recent study used matrix- (Ladiges et al. 1987). Commonly associated plant species assisted laser desorption/ionization-high resolution mass spec- include the overstorey trees Corymbia calophylla and Eucalyptus trometry to localize constituents to the different gland types in marginata, and the dominant understorey species Taxandria par- leaves of 17 Hypericum species (Kucharíková et al. 2016b). The viceps and Acacia divergens.

Tree Physiology Volume 38, 2018 Triketone and sesquiterpene gland types in Eucalyptus 1453

Leaf samples were collected from trees growing along before being divided into halves by cutting along the mid-vein. Boronia Rd in Mount Frankland National Park (34° 48′ 48′ S, Leaf area was determined from the images using ImageJ. One 116° 53′ 14′ E, elevation 180 m) in October 2015 under per- half of each leaf was digested using pectinase following the mit numbers CE004912 and SW017061, procured from the aforementioned gland isolation protocol. Enzymatically digested Department of Parks and Wildlife, WA. The forest trees were of leaf halves were then partitioned into three diﬀerent tissues advanced age (estimated to be at least 50 years old) and leaves (glands, epidermis with attached cuticle, and vasculature tissue) were harvested using pole pruners at a height of 4–6m. as described in Goodger and Woodrow (2011). Each tissue Additional leaf samples were collected in March 2016 from trees type was collected separately, frozen in liquid nitrogen and planted at Currency Creek Arboretum, SA (35° 25′ 45′ S, 138° ground in tapered vials using a micropestle. The intact half of 45′ 46′ E). The arboretum trees had been grown from seed each leaf was also ground to a ﬁne powder under liquid nitrogen

sourced from the same Boronia Rd, Mount Frankland National in a mortar using a pestle. Ground tissues and leaf halves from Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021 Park population in 1994 and planted at the arboretum in 1995. the six leaves were extracted in hexane. The contents of each Leaves were collected by hand at a height of 2 m. All sampled extract were determined using GC–FID and the results for the leaves were immediately placed in snap-lock plastic bags and isolated tissues compared with the undigested leaf half from returned to the laboratory on dry ice where they were stored at each leaf. −80 °C until analysed. ff Seed from an individual E. brevistylis tree seed lot was pur- Determination of the number of di erent gland types chased from East Manjimup Nursery and Seed Centre, Forest Fully expanded leaves collected from two arboretum trees were Products Commission, Manjimup, WA in February 2016. The randomly selected and enzymatically digested. Ten groups of 10 seed had been collected in December 1998 from a tree growing isolated translucent glands and 10 groups of 10 isolated along Boronia Rd, Mount Frankland National Park, WA (34° 48′ golden-brown glands were collected from each tree and each 46′ S, 116° 53′ 28′ E). Seed was bulk sown into a tray contain- group was ground and extracted in hexane (35 μl) and analysed ing vermiculite, perlite and sand (1:1:1, v/v/v) in March 2016. using GC–FID as described. The consistency of the constituent Individual seedlings were transplanted to 300 ml forestry tubes profiles in both gland types was assessed using multiple groups containing commercial potting mix 1 month after germination of glands because hexane extraction of a single isolated gland and transferred to a glasshouse at the University of Melbourne was insufficient for accurate GC–FID quantification due to the (see Goodger and Woodrow (2010) for glasshouse growth small size of glands. conditions). After 5 months, seedlings were potted up into 1.65 l pots containing commercial potting mix and maintained in Analysis of population and plant ontogenetic variation the glasshouse until sampled. in gland types Fully expanded leaves were collected from seven 50-year-old Gland isolation protocol forest trees, three 12-year-old arboretum trees and 11 Foliar glands were isolated from fully expanded E. brevistylis glasshouse-grown saplings 15 months after sowing. Sapling leaves collected from the forest and arboretum trees following leaves were collected from a lateral branch at a height of 1 m. digestion with pectinase as described in Goodger et al. (2010). Leaves were ground and extracted with hexane for GC–FID After digestion and subsequent tissue disruption, glands were analyses. isolated from other leaf tissues by sieving. Two types of glands fl were readily distinguishable based on their golden-brown or The in uence of leaf expansion on gland type abundance translucent white appearance when viewed under a stereomicro- Five leaves of different ages and sizes along a single branch scope (see Figure 1). Samples of the two different gland types were collected from each of two forest trees and one arboretum were collected separately using a pipette. Each sample com- tree. Leaves ranged from recently emerged and unexpanded to prised 200 isolated glands and was ground and extracted with large and fully expanded. Leaves were imaged for area determi- hexane for gas chromatography with mass spectrometry (GC–MS) nations using ImageJ and then immediately ground, extracted in and gas chromatography with flame ionization detection (GC–FID) hexane and analysed by GC–FID. For the spatial analysis, fully analyses. Isolated glands were imaged under the stereomicro- expanded leaves collected from arboretum and forest trees were scope and images analysed for gland volume and circularity using randomly selected and 5 mm strips were dissected from the ImageJ software (version 1.50i, National Institutes of Health, apical, middle and basal regions of each leaf in an orientation Bethesda, MD, USA). perpendicular to the mid-vein. Leaf strips were then digested in pectinase for 20 min to facilitate the removal of the upper epi- ff Assessment of triketone occurrence in di erent leaf tissues dermis and attached cuticle, thereby enabling better visualization Six fully expanded leaves collected from the arboretum trees of glands and their colours (see Figure 1A). Images were taken were randomly selected and imaged using a flat-bed scanner of three zones within each strip: leaf margin, lamina and mid-vein

Tree Physiology Online at http://www.treephys.oxfordjournals.org 1454 Goodger et al.

Figure 1. Two different gland types are embedded within mature leaves of E brevistylis. (A) Image of leaf with cuticle and epidermis removed as viewed Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021 under a stereomicroscope with transmitted lighting. Translucent and golden-brown glands are clearly distinguishable. Scale bar represents 1 mm. (B) Isolated translucent ‘sesquiterpene’ glands as viewed under a stereomicroscope. Scale bar represents 150 μm. (C) Isolated golden-brown ‘triketone’ glands as viewed under a stereomicroscope. Scale bar represents 150 μm. using a stereomicroscope with attached camera. The number of on calibration series of commercial standards of each compound golden-brown and white glands in images of each zone were (Sigma-Aldrich, St Louis, MO, USA). The sesquiterpene alcohols counted. and the β-triketone conglomerone were quantified based on a standard series of the oxygenated sesquiterpene caryophyllene Gas chromatographic analyses oxide (Sigma-Aldrich), for want of commercial standards. It should be noted that the GC column used was unable to differentiate Constituents in hexane extracts were identified by GC–MS and between enantiomers and so no stereochemical information is quantified by GC–FID. The GC–MS system comprised a Gerstel provided. 2.5.2 autosampler, a 7890A gas chromatograph and a 5975C quadrupole MS (Agilent Technologies, Santa Clara, CA, USA). Statistical analyses μ Samples (1 l) were injected in splitless mode and the MS was One- and two-way ANOVA, paired t-tests, linear regressions and ’ adjusted to manufacturer s recommendations using tris-(per- parallel line analyses were performed using Sigmaplot Version fl uorobutyl)-amine (FC-43). The following MS source conditions 13 (Systat Software, San Jose, CA, USA) and Minitab Version 17 ° ° were used: injection temperature 250 C, transfer line 280 C, (MiniTab Inc., State College, PA, USA). ionsource230°C, quadrupole 150 °C, 70 eV (EI mode), −1 – – 2.66 scans s and scanning range m/z 50 600. The GC MS Results column was a 30 m VF-5MS with 0.2 μm film thickness and a ff 10 m Integra guard column (J & W, Agilent). Helium was used as Discovery of di erent types of foliar glands − the carrier gas at a flow rate of 1 ml min 1. The column tempera- We show here for the first time that leaves of a Eucalyptus spe- ture was held at 120 °C for 1 min following injection, then cies contain two visibly distinct types of glands embedded within − ramped at 7 °C min 1 to 180 °C and held at that temperature for the leaf lamina, and that these gland types house distinct groups a further 10 min. Mass spectra were evaluated using Agilent MSD of metabolites. The first type of gland was translucent in appear- ChemStation E.02.02.1431 for GC–MS and sesquiterpenes and ance (Figure 1A and B) and these glands were shown to exclu- β-triketones identified using the NIST 11 and Adams 2012 mass sively contain sesquiterpenes, namely the structurally similar spectra libraries (Adams 2012). Constituent identity was con- alcohols cubebol and 10-epi-cubebol in high abundance, with firmed by their relative retention times and by comparison to avail- lower abundance cubenol and epi-cubenol (Figure 2A). For sim- able authentic standards. Identified constituents were consistent plicity, we termed these glands ‘sesquiterpene glands’. The with a previous report on dichloromethane extracts of E. brevistylis second type of gland appeared a brownish-gold colour and their leaves that used GC–MS and a combination of homonuclear and occurrence in leaves was particularly evident once the cuticle heteronuclear 13C- and 1H-NMR techniques to structurally eluci- had been removed (Figure 1A and C). This gland type predom- date sesquiterpene constituents (Cornwell et al. 2000). The GC– inantly contained the β-triketone conglomerone, together with FID system comprised a Perkin Elmer Autosystem XC (Perkin lower amounts of the sesquiterpene hydrocarbons α- and β-car- Elmer, Melbourne, Australia) fitted with a low polarity column yophyllene (Figure 2B). We termed these glands ‘triketone Zebron ZB-5 (30 m × 0.25 mm i.d., Phenomenex, Torrance, CA, glands’. Interestingly, only trace levels of monoterpenes were USA). The injector temperature was 250 °C and detector tem- detected in either gland type. Both gland types were distributed perature was 220 °C. The carrier gas and temperature pro- throughout the leaf with triketone glands appearing to have a gramme were the same as for the GC–MS. Quantification of the slightly greater abundance towards the margins and basal sesquiterpene hydrocarbons α-andβ-caryophyllene were based regions of leaves.

Tree Physiology Volume 38, 2018 Triketone and sesquiterpene gland types in Eucalyptus 1455

tridecane (internal standard) 10-epi-cubebol

cubenol & OH epi-cubenol

cubebol Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021

O O GC-FID response (mV)

tridecane O O (internal standard)

β-caryophyllene α-caryophyllene conglomerone

Retention time (min)

Figure 2. Fractionation and identiﬁcation by GC–MS of the chemical constituents in hexane extracts of isolated E. brevistylis glands. (A) Sesquiterpene alcohol cubebols and lower abundance cubenols extracted from translucent ‘sesquiterpene glands’. (B) The aromatic β-triketone conglomerone and lower abundance sesquiterpene hydrocarbon caryophyllenes extracted from golden-brown ‘triketone glands’.

Glands could be readily isolated from fully expanded E. brevis- undigested leaf halves (including mesophyll cells) and found tylis leaves using an enzymatic digestion protocol and despite 99.96 ± 0.02% of conglomerone is housed in leaf glands with the difference in colour and constituents between gland types, only trace amounts detected in the vascular, epidermal and the overall shape and size of glands, and the arrangement of mesophyll tissues. Therefore, we are confident in stating that cells bounding them appeared superficially similar (Figure 1B conglomerone is exclusively housed within glands of E. brevisty- and C). Indeed, measurements of gland sizes (including cells lis leaves and that the trace levels detected in the other leaf tis- bounding the lumen) from a subset of randomly selected, iso- sues are likely due to contamination from glands damaged lated glands found no significant difference in mean gland vol- during leaf dissection for enzymatic digestion. ume for the two types (ANOVA F = 0.7, P > 0.05, DF = 23). In addition to the two gland types observed, the possibility The mean (±SE) gland volume was 1.1 ± 0.1 nl for the sesqui- also existed that there were different types of triketone glands terpene glands and 1.3 ± 0.1 nl for the triketone glands. responsible for conglomerone and the caryophyllenes, respect- Despite having equivalent volume, the sesquiterpene glands ively and/or different types of sesquiterpene glands responsible were slightly more spherical in shape (Figure 1B and C) with a for the cubebols and cubenols, respectively. To explore if there mean (±SE) circularity index for cross-sections of 0.99 ± 0.01 were further subdivisions of metabolite specialization in the E. compared with 0.97 ± 0.01 for the triketone glands (ANOVA brevistylis glands, 10 groups of 10 randomly selected glands of F = 5.2, P < 0.05, DF = 23). each of the two gland types were isolated from leaves of arbor- To determine if the triketone conglomerone was exclusively etum trees. Ideally individual glands of each type would have stored in glands or distributed throughout other leaf tissues as been analysed, but due to the small volume of E. brevistylis well, we again made use of the enzymatic digestion protocol. glands (<2 nl) groups of glands were necessary to accurately This enabled partitioning of fully expanded leaves into three dif- quantify all constituents on the GC–FID system used. The mean ferent tissues: glands, vasculature and epidermis with attached (±SE) conglomerone to total caryophyllenes ratio was 8.2 ± 0.7 cuticle. We compared extracts of the isolated tissues to across the 10 groups of triketone glands. The consistency of this

Tree Physiology Online at http://www.treephys.oxfordjournals.org 1456 Goodger et al. result suggests that there is only one type of triketone gland that Abundance of gland types varies with leaf expansion produces both the triketone conglomerone and the lower abun- We next examined how the abundance of the two gland types α β ± dance sesquiterpenes - and -caryophyllene. The mean ( SE) was influenced by leaf expansion in arboretum and forest trees total cubebol to total cubenol ratio in sesquiterpene glands was by analysing a range of leaf sizes from newly formed, unex- ± 4.9 0.1 across the 10 groups of sesquiterpene glands. As panded ‘tips’ to fully expanded leaves (see Figure 4B for leaf with the triketone glands, the consistency of this result suggests outlines). We observed a significant linear decrease in the per- that there is only one type of sesquiterpene gland that produces centage of triketone glands (based on constituent totals) as both cubebols and the lower abundance cubenols. leaves expanded on trees (Figure 4A). Linear regressions performed on the data from each tree was significant and gave Abundance of gland types varies with plant ontogeny equations of y = −1.9x + 73 (r2 = 0.89; F = 25, P < 0.05) for

To examine if the presence of the two different gland types was a the arboretum tree, y = −2.2x + 85 (r2 = 0.97; F = 114, P < Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021 common feature of the species and was evident at different plant 0.01) for the first forest tree and y = −1.9x + 90 (r2 = 0.89; F = ages, we extracted the glandular constituents from leaves of seven 25, P < 0.05) for the second forest tree (Figure 4A). A parallel advanced age forest trees, three mature but younger arboretum line analysis showed no significant difference between the trees, and 11 glasshouse-grown saplings (Figure 3). The forest regression line slopes for each tree (F = 0.33, P = 0.7), but a trees had higher mean (±SE) total glandular constituents of 48 ± significant difference in the intercepts (F = 29, P < 0.0001), −1 3mgg leafDW (ANOVA F = 46.6, P < 0.001, DF = 20), com- consistent with the differences in sesquiterpene gland to trike- pared with 24 ± 6 for the arboretum trees and 16 ± 2forthesap- tone gland constituents observed in Figure 3. lings. Leaf extracts from all plants contained constituents from both We further examined the influence of leaf expansion by count- triketone and sesquiterpene glands, but the relative proportions of ing the respective numbers of white and golden-brown coloured the two gland types appeared to vary with plant age. The percent- glands in fully expanded forest tree leaves divided into nine age of triketone glands in leaves relative to sesquiterpene glands zones based on three apical–basal axis regions (basal, middle (based on constituent totals) was highest in saplings (mean ± SE and apical) with each region sub-divided into three lateral axis = 91 ± 3%, range = 71–99%), intermediate for the young arbor- zones (margin, lamina and mid-vein; see Figure 4B). The etum trees, but most variable (mean ± SE = 79 ± 10%, range = golden-brown triketone glands were relatively more abundant at 60–96%), and lowest in old forest trees (mean ± SE = 49 ± 3%, the basal region of leaves and along leaf margins, ranging from a range = 34–57%; ANOVA F = 26.6, P < 0.0001). high of 36% in the basal margin region to a low of only 2% in

Figure 3. Quantiﬁcation of glandular constituents from whole leaf extracts of fully expanded leaves sampled from glasshouse-grown saplings, young arboretum trees and mature forest trees of E. brevistylis. Sap, glasshouse-grown sapling number; CC, Currency Creek arboretum tree number; WA, Mount Frankland National Park forest tree number.

Tree Physiology Volume 38, 2018 Triketone and sesquiterpene gland types in Eucalyptus 1457 Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021

Figure 4. Gland-type abundance changes in E. brevistylis during leaf expansion. (A) Linear decrease in triketone gland abundance (based on constituents) with increasing leaf area along branches of E. brevistylis trees. Circle symbols and solid regression line represent an arboretum tree. Triangle symbols fitted with a short-dashed regression line, and square symbols fitted with a long-dashed regression line represent two forest trees. (B) Leaf outlines of expanding leaves along a representative branch from (i) a young arboretum tree and (ii) an advanced age forest tree; the scale bar represents 25 mm. Arrows denote the positions leaf sections were dissected from the oldest, fully expanded leaf for the gland spatial analysis. (C) Spatial analysis of gland type abundance in leaf sections from fully expanded E. brevistylis leaves from forest trees. Data represents the mean of six leaves per region. the apical mid-vein region (Figure 4C). A two-way ANOVA brevistylis would be the only one of these to house the com- showed a significant difference in the proportion of triketone pounds in separate glands. Nevertheless, the possibility exists glands across the leaf width from the leaf margin to the leaf mid- that this trait is unique to E. brevistylis given it has been rib (F = 25.8, P < 0.001), and along the leaf length between described as a taxonomically isolated, relict species based on its apical, middle and basal portions of leaf (F = 3.3, P < 0.05), but unusual leaf morphology, particularly as a seedling (Brooker no interaction between leaf width and length (F = 0.4, P > 1974), and its unusual floral morphology, including its short 0.05). style and outer stamens that lack anthers (Nicolle 2018). Moreover, it is the only species classified in series ‘Pedaria’ MS of section Longistylus in subgenus Eucalyptus (Nicolle 2018), so Discussion it does possess unique traits. Future examination of other The results presented here definitively show that there are two triketone-containing Eucalyptus species such as E. agglomerata, different types of glands in the leaves of E. brevistylis: sesquiter- E. insularis and E. lateritica will confirm if specialized triketone pene glands that produce sesquiterpene alcohol cubenols and glands are common to the genus or are exclusive to E. brevistylis. cubebols and triketone glands that produce predominantly the triketone conglomerone and lesser amounts of sesquiterpene Sesquiterpene biosynthesis hydrocarbon caryophyllenes (Figure 2). This is the first report of Despite referring to one gland type as ‘triketone glands’ here, it such differential metabolic specialization of gland types in a tree is important to note that those glands did not exclusively pro- species. Exclusively translucent glands have been observed in duce the triketone conglomerone, but also contained the sesqui- leaves of all other Eucalyptus species examined to date terpene hydrocarbons α- and β-caryophyllene, albeit in much (Goodger et al. 2010, 2016, Heskes et al. 2012a, Brooker and lower abundance (Figure 2B). Nevertheless, these compounds Nicolle 2013). It is possible that metabolic gland specialization were not detected in the sesquiterpene alcohol-dominated ‘ses- occurs in some of these species, but that it is not immediately quiterpene glands’ (Figure 2A). This suggests that there are dif- obvious based on visual inspection of leaves. This is, however, ferent sesquiterpene synthase enzymes responsible for the unlikely given the leaves of most of these species possess exclu- sesquiterpene hydrocarbons and alcohols, respectively, and that sively monoterpenes and sesquiterpenes and these terpene each is active in a different gland type in E. brevistylis. The co- classes have been shown to co-occur in individual foliar glands occurrence of the two caryophyllenes in one gland type is not of the few species where glands have been directly sampled surprising given they have been shown to be biosynthesized (King et al. 2006, Goodger et al. 2010). together by a single, multi-product sesquiterpene synthase in Arabidopsis flowers (Chen et al. 2003, Tholl et al. 2005), but Triketones in Eucalyptus their co-occurrence with conglomerone in triketone glands is Some other species of Eucalyptus also contain triketones in their curious. The co-occurrence of cubebols and cubenols has also leaves (Bignell et al. 1997), and it seems improbable that E. been observed previously, particularly in the Myrtaceae, but is

Tree Physiology Online at http://www.treephys.oxfordjournals.org 1458 Goodger et al. likely due to post-biosynthetic modifications as a result of leaf were slightly less spherical in shape. Such elongated glands are ageing rather than due to shared biosynthesis. For example, commonly observed along leaf margins of Eucalyptus leaves both sesquiterpene alcohols were detected together in steam when the cuticle is removed using the enzymatic protocol distillates from a wide range of Myrtaceae plants including (Goodger et al. 2010, J. Goodger, unpublished observations). It Corymbia, Eucalyptus, Kunzea, Leptospermum and Melaleuca, but is noteworthy that a study on oil gland formation in Eucalyptus the abundance of cubenols decreased when solvent extractions lehmanii and Eucalyptus conferruminata found glands formed were used (Cornwell et al. 2000). Moreover, the accumulation during early leaf development arose from epidermal initials of cubenols in steam distillates could be prevented by the add- whereas those formed later in leaf expansion could arise from ition of buffering phosphate salts, thereby showing they were initials in the mesophyll layers (Carr and Carr 1980). It is not by-products created by acid-solvolysis of cubebols during distil- known if the different leaf progenitors give rise to different gland

lation (Cornwell et al. 2000). The authors claimed such acid- types in those species, but if such a developmental process is Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021 solvolysis also occurs as leaves age and concluded cubenols are occurring in E. brevistylis then it could account for the observed a product of leaf ontogenetic degradation of cubebols, rather leaf ontogenetic patterns and indeed the occurrence of the dif- than co-biosynthesis. ferent gland types in the species.

Ontogenetic variability in secondary metabolites Conclusion We observed a level of population variability in total glandular constituents within the groups of E. brevistylis plants sampled from Our analysis of E. brevistylis leaves has shown for the first time the forest, arboretum and glasshouse (Figure 3). Nonetheless, the existence of two types of foliar glands in a myrtaceous spe- the differences between ontogenetic stages were greater than cies. The gland types appear morphologically identical, but their the variability within each group and showed a constant, but discrete complements of secondary metabolites result in differ- temporally slow, increase in glandular constituents from young ent colours: golden-brown for the triketone glands and translu- saplings through to advanced age trees. This increase was cent for the sesquiterpene glands. We also showed that the largely due to a gradual increase in sesquiterpene alcohols relative abundance of the two types of glands is under ontogen- housed in sesquiterpene glands as plants aged. Such an onto- etic control at both the whole plant and leaf level. Thus, triketone genetic delay to the production of foliar secondary metabolites glands predominate in all sapling leaves and in young leaves of appears to be a common feature of Eucalyptus species, albeit at mature trees, then sesquiterpene gland production progressively different temporal scales, as it has been documented for mono- increases until the relative proportions of the two gland types is terpenes in E. polybractea (Goodger and Woodrow 2009), for comparable in fully expanded leaves of mature trees. monoterpenes, sesquiterpenes and monoterpene acid glucose These findings present opportunities for at least two areas of esters in E. froggattii (Goodger et al. 2013), and for cyanogenic future study. Firstly, given that it is now possible to isolate meta- glycosides in E. yarraensis (Goodger et al. 2007), E. polyanthe- bolically active foliar glands from eucalypts (Goodger et al. mos (Goodger et al. 2004), E. cladocalyx (Goodger et al. 2006) 2010), purification of the triketone glands based on their colour and E. camphora (Neilson et al. 2011, 2006). difference provides a relatively simple platform for mapping the Concomitant with the apparent influence of whole plant pathways of β-triketone and sesquiterpene synthesis. Both com- ontogeny on gland type abundance, and acting in a similar man- pound classes are of considerable interest due to their strong ner, was the effect of leaf ontogeny in E. brevistylis. As mature insecticidal properties (Cheng et al. 2007, Park et al. 2017), but tree leaves expanded, the relative abundance of triketone glands the biosynthesis of β-triketones remains relatively unstudied. decreased due to an increase in the production of sesquiterpene Secondly, the age-related pattern of deployment of the two gland glands (Figure 4A). Further exploration of this phenomenon at a types in E. brevistylis is an excellent system in which to study spatial level showed that triketone glands were relatively more how trees tune their chemical defences to meet different envir- abundant along leaf edges and at the basal portion of leaves onmental challenges, in particular, those presented by different (Figure 4C). This suggests that triketone glands are the first suites of herbivores. Young eucalypts, and the young leaves on formed glands in young leaves, and as leaves expand apically mature eucalypts, are particularly vulnerable to insect herbivory, and to a lesser degree, laterally, new lamina is comprised largely so it appears logical that this challenge is met by increased of sesquiterpene glands. The production of new lamina predom- deployment of glands rich in insecticidal β-triketones. We specu- inantly from the central and apical regions results in a higher pro- late that the suite of herbivores encountered by mature trees portion of original triketone glands remaining across the leaf may differ from those that prey on young saplings or immature base, and pushes triketone glands that were more apically leaves. Consequently, older E. brevistylis trees may be advan- located in unexpanded leaves towards the leaf margins. This lat- taged by deploying an increased abundance of defensive ses- ter action of leaf expansion may be the reason triketone glands quiterpenes in their leaves.

Tree Physiology Volume 38, 2018 Triketone and sesquiterpene gland types in Eucalyptus 1459

Acknowledgments Goodger JQD, Woodrow IE (2010) The influence of micropropagation on growth and coppicing ability of Eucalyptus polybractea. Tree Physiol We thank Metabolomics Australia for GC–MS analyses. 30:285–296. Goodger JQD, Woodrow IE (2011) α, β-unsaturated monoterpene acid glucose esters: structural diversity, bioactivities and functional roles. Conflict of interest Phytochemistry 72:2259–2266. Goodger JQD, Ades PK, Woodrow IE (2004) Cyanogenesis in Eucalyptus None declared. polyanthemos seedlings: heritability, ontogeny and effect of soil nitrogen. Tree Physiol 24:681–688. Goodger JQD, Gleadow RM, Woodrow IE (2006) Growth cost and onto- Funding genetic expression patterns of defence in cyanogenic Eucalyptus spp. Trees 20:757–765. This research was funded by the Australian Research Council Goodger JQD, Choo TYS, Woodrow IE (2007) Ontogenetic and temporal (LP150100798) and Eucalypt Australia (2015-24). S.L.S. was trajectories of chemical defence in a cyanogenic eucalypt. Oecologia Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021 supported by the Holsworth Wildlife Research Endowment (man- 153:799–808. Goodger JQD, Heskes AM, King DJ, Gleadow RM, Woodrow IE (2008) aged by Equity Trustees). Micropropagation of Eucalyptus polybractea selected for key essential oil traits. Funct Plant Biol 35:247–251. Goodger JQD, Heskes AM, Mitchell MC, King DJ, Neilson EH, Woodrow IE References (2010) Isolation of intact sub-dermal secretory cavities from Adams RP (2012) Identification of essential oil components by gas chro- Eucalyptus. Plant Methods 6:20. matrography/mass spectrometry, 4th edn. Allured Business Media, Goodger JQD, Heskes AM, Woodrow IE (2013) Contrasting ontogenetic Carol Stream, IL. trajectories for phenolic and terpenoid defences in Eucalyptus froggat- Bignell CM, Dunlop PJ, Brophy JJ, Fookes CJR (1997) Volatile leaf oils of tii. Ann Bot 112:651–659. some south-western and southern Australian species of the genus Goodger JQD, Senaratne SL, Nicolle D, Woodrow IE (2016) Foliar essen- Eucalyptus (series I). Part XIV. Subgenus Monocalyptus. Flavour Fragr J tial oil glands of Eucalyptus subgenus Eucalyptus (Myrtaceae) are a 12:177–183. rich source of flavonoids and related non-volatile constituents. PLoS Brooker MIH (1974) Six new species of Eucalyptus from Western One 11:e0151432. Australia. Nuytsia 1:297–314. Guo J, Yuan Y, Liu Z, Zhu J (2013) Development and structure of internal Brooker MIH, Nicolle D (2013) Atlas of leaf venation and oil gland pat- glands and external glandular trichomes in Pogostemon cablin. PLoS terns in the eucalypts. CSIRO Publishing, Collingwood, Australia, 226 One 8:e77862. pp. Harborne JB (1991) Role of secondary metabolites in chemical defense Brophy JJ, Southwell I (2002) Eucalyptus chemistry. In: Coppen JJW (ed) mechanisms in plants. In: Chadwick DJ, Marsh J (eds) Bioactive com- Eucalyptus: the genus Eucalyptus. CRC Press, Taylor & Franics Group, pounds from plants. John Wiley and Sons, Chichester, pp 126–139. London, pp 102–160. Harborne JB (2001) Twenty-five years of chemical ecology. Nat Prod Brophy JJ, House APN, Boland DJ, Lassak EV (1991) Digests of the Rep 18:361–379. essential oils of 111 species from northern and eastern Australia. In: Hellyer RO (1968) The occurrence of β-triketones in the steam-volatile oils Boland DJ, Brophy JJ, House APN (eds) Eucalyptus leaf oils. Inkata of some myrtaceous Australian plants. Aust J Chem 21:2825–2828. Press, Melbourne, pp 29–155. Heskes AM, Goodger JQD, Tsegay S, Quach T, Williams SJ, Woodrow IE Carr DJ, Carr SGM (1970) Oil glands and ducts in Eucalyptus L’Hérit. II. (2012a) Localization of oleuropeyl glucose esters and a flavanone to Development and structure of oil glands in the embryo. Aust J Bot 18: secretory cavities of Myrtaceae. PLoS One 7:e40856. 191–212. Heskes AM, Lincoln CN, Goodger JQD, Woodrow IE, Smith TA (2012b) Carr DJ, Carr SGM (1980) Correlation between sizes of stomata and of Multiphoton fluorescence lifetime imaging shows spatial segregation oil glands in two related Eucalyptus species. Aust J Bot 28:551–554. of secondary metabolites in Eucalyptus secretory cavities. J Microsc Chen F, Tholl D, D’Auria JC, Farooq A, Pichersky E, Gershenzon J (2003) 247:33–42. Biosynthesis and emission of terpenoid volatiles from Arabidopsis Killeen DP, van Klink JW, Smallfield BM, Gordon KC, Perry NB (2015) flowers. Plant Cell 15:481–494. Herbicidal β-triketones are compartmentalized in leaves of Cheng A-X, Lou Y-G, Mao Y-B, Lu S, Wang L-J, Chen X-Y (2007) Plant Leptospermum species: localization by Raman microscopy and rapid terpenoids: biosynthesis and ecological functions. J Integr Plant Biol screening. New Phytol 205:339–349. 49:179–186. King DJ, Gleadow RM, Woodrow IE (2006) Regulation of oil accumula- Cornwell C, Reddy N, Leach D, Wyllie S (2000) Origin of (+)-δ-cadinene tion in single glands of Eucalyptus polybractea. New Phytol 172: 440–451. and the cubenols in the essential oils of the Myrtaceae. Flavour Fragr J ˇ 15:352–361. Kucharíková A, Kimáková K, Janfelt C, Cellárová E (2016a) Interspecific Department of Parks and Wildlife (2015) Conservation codes for variation in localization of hypericins and phloroglucinols in the genus Western Australian Flora and Fauna. The Government of Western Hypericum as revealed by desorption electrospray ionization mass spectrometry imaging. Physiol Plant 157:2–12. Australia, Kensington, WA. https://www.dpaw.wa.gov.au/images/ ˇ documents/plants-animals/threatened-species/Listings/conservation_ Kucharíková A, Kusari S, Sezgin S, Spiteller M, Cellárová E (2016b) code_definitions.pdf Occurrence and distribution of phytochemicals in the leaves of 17 Doran JC (1991) Commercial sources, uses, formation, and biology. In: in vitro cultured Hypericum spp. adapted to outdoor conditions. Front Boland DJ, Brophy JJ, House APN (eds) Eucalyptus leaf oils, use, chem- Plant Sci 7:Article 1616. istry, distillation and marketing. Inkarta Press, Melbourne, pp 11–28. Ladiges PY, Humphries CJ, Brooker MIH (1987) Cladistic and biogeo- Goodger JQD, Woodrow IE (2009) The influence of ontogeny on essen- graphic analysis of Western Australian species of Eucalyptus L’Herit., tial oil traits when micropropagating Eucalyptus polybractea. For Ecol informal subgenus Monocalyptus Pryor & Johnson. Aust J Bot 35: Manage 258:650–656. 251–281.

Tree Physiology Online at http://www.treephys.oxfordjournals.org 1460 Goodger et al.

Li B, Hansen SH, Janfelt C (2013) Direct imaging of plant metabolites in Park CG, Jang M, Shin E, Kim J (2017) Myrtaceae plant essential oils and leaves and petals by desorption electrospray ionization mass spec- their β-triketone components as insecticides against Drosophila suzu- trometry. Int J Mass Spectrom 348:15–22. kii. Molecules 22:1050. Li S, Tosens T, Harley Peter C, Jiang Y, Kanagendran A, Grosberg M, Schmiderer C, Grassi P, Novak J, Weber M, Franz C (2008) Diversity of Jaamets K, Niinemets Ü (2018) Glandular trichomes as a barrier essential oil glands of clary sage (Salvia sclarea L., Lamiaceae). Plant against atmospheric oxidative stress: relationships with ozone uptake, Biol 10:433–440. leaf damage, and emission of LOX products across a diverse set of Soelberg J, Jørgensen LB, Jäger AK (2007) Hyperforin accumulates in species. Plant Cell Environ 41:1263–1277. the translucent glands of Hypericum perforatum. Ann Bot (Lond) 99: Neilson EH, Goodger JQD, Woodrow IE (2006) Novel aspects of cyanogenesis 1097–1100. in Eucalyptus camphora subsp. humeana. Funct Plant Biol 33:487–496. Tholl D, Chen F, Petri J, Gershenzon J, Pichersky E (2005) Two sesquiter- Neilson EH, Goodger JQD, Motawia MS, Bjarnholt N, Frisch T, Olsen CE, pene synthases are responsible for the complex mixture of sesquiter- Møller BL, Woodrow IE (2011) Phenylalanine derived cyanogenic penes emitted from Arabidopsis flowers. Plant J 42:757–771. diglucosides from Eucalyptus camphora and their abundances in rela- Western Australian Herbarium (2017) Descriptions by the Western tion to ontogeny and tissue type. Phytochemistry 72:2325–2334. Australian Herbarium, Department of Parks and Wildlife. Text used Downloaded from https://academic.oup.com/treephys/article/38/10/1451/5055867 by guest on 27 September 2021 Nicolle D (2018) Classification of the eucalypts (Angophora, Corymbia with permission. The Government of Western Australia, Kensington, and Eucalyptus). Version 3. http://www.dn.com.au/Classification-Of- WA. https://florabase.dpaw.wa.gov.au/help/copyright (22 February The-Eucalypts.pdf. 2017, date last accessed).

Tree Physiology Volume 38, 2018