Leaf Ontogeny Influences Leaf Phenolics and the Efficacy of ...

P1: JQX Journal of Chemical Ecology [joec]

pp997-joec-473743

October 17, 2003

20:28

Style file version June 28th, 2002

C 2003) Journal of Chemical Ecology, Vol. 29, No. 11, November 2003 (°

LEAF ONTOGENY INFLUENCES LEAF PHENOLICS AND THE EFFICACY OF GENETICALLY EXPRESSED Bacillus thuringiensis cry1A(a) d-ENDOTOXIN IN HYBRID POPLAR AGAINST GYPSY MOTH

KARL W. KLEINER,1,∗ DAVID D. ELLIS,2,3 BRENT H. McCOWN,2 and KENNETH F. RAFFA1 1 Department

of Entomology University of Wisconsin Madison, Wisconsin 53706, USA 2 Department

of Horticulture University of Wisconsin Madison, Wisconsin 53706, USA

(Received October 23, 2002; accepted July 11, 2003)

Abstract—We tested the hypothesis that ontogenetic variation in leaf chemistry could affect the efficacy of genetically expressed Bacillus thuringiensis cry1A(a) d-endotoxin, and thus provide spatial variation in (1) foliage protection and (2) selective pressures that could delay the resistance of folivores. Our model consisted of clonal hybrid Populus plants (NC5339). Consumption of foliage and relative growth rates of gypsy moth, Lymantria dispar (L.) increased, and phenolic glycoside concentrations decreased, as leaves from transformed plants containing the cry1A(a) d-endotoxin and nontransformed plants matured from leaf plastochron index (LPI) 1–6. Feeding and growth rates were negatively correlated with phenolic glycosides in both transformed and nontransformed foliage. The presence of the B. thuringiensis d-endotoxin was at most, additive to the effect of the phenolic glycosides. Feeding and growth rates were positively correlated with condensed tannins in transformed foliage, but there was no relationship with condensed tannins in nontransformed foliage. The results indicate that the presence of foliar allelochemicals of poplar can enhance the effectiveness of genetically expressed B. thuringiensis d-endotoxin against gypsy moth larvae. However, the spatial variation in gypsy moth performance in response to the combination of foliar allelochemicals and d-endotoxin was not greater 3 Current address: BC Research Inc., Forest Biotechnology Centre Vancouver, British Columbia Canada

V6S2L2.

∗ To whom correspondence should be addressed. Current address: Department of Biological Sciences,

York College of Pennsylvania, York, Pennsylvania 17405, USA. E-mail: [email protected]

2585 C 2003 Plenum Publishing Corporation 0098-0331/03/1100-2585/0 °


2586

pp997-joec-473743

October 17, 2003

20:28


KLEINER, ELLIS, MCCOWN, AND RAFFA than the effect of ontogenetic variation in foliar allelochemicals alone. These results suggest that for this important pest, foliage protection may be obtained without genetically engineered defenses, and instead, by relying on ontogenetic and clonal variation in allelochemicals. The benefits of combining novel resistance mechanisms with natural ones will depend upon the specific folivore’s adaptation to natural resistance mechanisms, such as allelochemicals. Moreover, some of the greatest benefits from transgenic resistance may arise from the need to protect trees from multiple pests, some of which may not be deterred by, or may even prefer, allelochemicals that confer protection from a few species. Key Words—Bacillus thuringiensis, Lymantria dispar, Populus, transgenic resistance, allelochemicals.

INTRODUCTION

The evolution of resistant insect biotypes poses the greatest threat to the deployment of plants genetically engineered with insecticidal proteins (Gould, 1988; Raffa, 1989; Brattsten, 1991; Bauer, 1995). Resistance to toxins can develop when the selection pressure on an insect population is strong and consistent through time (Brattsten et al., 1986; Brattsten, 1991; Denholm and Rowland, 1992). The discovery of the soil-born bacterium, Bacillus thuringiensis, over 40 years ago, and the more recent discovery of additional B. thuringiensis strains has provided renewed promise for the development of environmentally safe and effective insecticides. Not only is the toxicity of B. thuringiensis specific to particular taxa of invertebrates, but early deployment was accompanied by optimism that resistance to a naturally evolved pathogen that produced multiple endotoxins would pose relatively few problems (H¨ofte and Whiteley, 1989; Gill et al., 1992). However, resistance to B. thuringiensis has emerged in both laboratory (McGaughey and Beeman, 1988; Stone et al., 1989; Bauer et al., 1994) and field (Shelton et al., 1993; Tabashnik, 1994a) populations, thus raising concerns about the long-term efficacy of transgenic plants expressing B. thuringiensis toxins. The application of genetic engineering for pest control on trees has some economical and logistically attractive properties relative to traditional control tactics, but it must be managed wisely. The long-lived nature of trees combined with the constant genetic expression of insect resistance can provide conditions conducive for the development of biotype resistance in insects (Raffa, 1989). Strategies to reduce biotype evolution rely primarily on varying the pattern and predictability of any single selective force (Gould, 1988; Raffa, 1989; Brattsten, 1991). These have been summarized by Bauer (1995, 1997) to include (1) mixtures of toxins, (2) synergists to increase toxicity, (3) rotations of toxins, (4) temporal and spatial refuges, (5) sublethal toxins to enhance natural controls, (6) ultrahigh doses to kill resistant individuals, and (7) gene regulation of temporal and spatial toxin expression. Some of these same tactics resemble the naturally evolved chemical


pp997-joec-473743

LEAF PHENOLICS AND EFFICACY OF

October 17, 2003

20:28


Bacillus thuringiensis cry1A(a) d-ENDOTOXIN

2587

defense systems of long-lived trees (Feeny, 1970; Schultz, 1983a,b; Whitham, 1983; Berenbaum, 1988; Hoy et al., 1998). One way that plants can decrease damage due to insect herbivores and delay the resistance of insects to natural defense mechanisms is through temporal and spatial variation in the expression of allelochemicals (Schultz et al., 1983b; Coleman, 1986; Gould, 1988; Karban et al., 1997; Kleiner et al., 1999). In addition, the presence of more than one class of secondary compounds in a plant may confer a greater degree of resistance to herbivore damage than a single compound (Berenbaum and Neal, 1985; Berenbaum and Zangerl, 1988; Gould, 1988; Hay et al., 1994). These natural processes of reducing damage and delaying herbivore resistance suggest mechanisms that may be used to enhance and prolong the efficacy of genetically engineered insect resistance (but see discussions by Gould, 1988; Iwao and Rausher, 1997; Onstad and Gould, 1998a,b on how interactions among multiple selection pressures sometimes can increase the development rate of insect resistance to pesticides or, in natural systems, plant resistance to insects). The natural variation in the allelochemistry of Populus foliage provides a potential mechanism for introducing variation in the deployment of short-rotation plants with genetically engineered insect resistance. Populus is characterized by indeterminate growth, as leaves are produced continually from budbreak to senescence. Therefore, a gradient of leaf development stages is present at all times. Concentrations of phenolic glycosides are highest in the youngest leaves, and decrease as they expand and mature (Lindroth et al., 1987; Bingaman and Hart, 1993). Since the toxicity of B. thuringiensis to Lepidoptera is enhanced by phenolic glycosides (Arteel and Lindroth, 1992; Hwang et al., 1995), the effect of the genetically expressed B. thuringiensis d-endotoxin should be greater in immature foliage than in mature foliage. Condensed tannin concentrations increase as leaves age (Meyer and Montgomery, 1987; Kleiner et al., 1998). Because condensed tannins have an inhibitory effect on the efficacy of B. thuringiensis to Lepidoptera (Navon et al., 1993; Hwang et al., 1995), larvae feeding on older leaves should be less affected than those feeding on younger leaves. Independently, variation in each class of allelochemicals contributes to greater B. thuringiensis efficacy in young foliage than in older foliage. In general, insect damage of young tissues is more detrimental to growth and yield than damage to older tissues (Reich et al., 1993), hence protection of these tissues would be more desirable. This natural array of allelochemicals could in effect, provide some of the spatial variation necessary to (1) increase the efficacy of genetically expressed d-endotoxin in young plant tissue and (2) reduce biotype evolution from a toxic selection pressure. In this paper, we report on the efficacy of the B. thuringiensis cry1A(a) dendotoxin genetically expressed in the foliage of hybrid Populus clone NC5339 against gypsy moth larvae, Lymantria dispar (L.) with the allelochemical concentration of leaves at different ontogenetic stages.


2588

pp997-joec-473743

October 17, 2003

20:28


KLEINER, ELLIS, MCCOWN, AND RAFFA METHODS AND MATERIALS

Tree Culture. Transformed trees of Populus clone NC5339, P. alba L. × P. grandidentata Michx., “Crandon,” (a cross between the Leuce and Tachamahaca sections of the genus) were produced by electrical discharge particle acceleration gene insertion, as described in McCown et al. (1991), and are from the same genetic stock as Transclone Bt-11 (Robison et al., 1994; Kleiner et al., 1995). The plasmid DNA (pTVBTGUS) contained three chimeric genes: (1) a neomycin phosphortransferase (NPT) gene enabling kanamycin resistance selection, (2) a chimeric gene for B-glucuronidase (GUS) for histochemical detection of gene expression, and (3) the gene encoding the B. thuringiensis kurstaki insecticidal cry1A(a) d-endotoxin protein, (Barton et al., 1987; McCabe et al., 1988). The cry1A(a) gene was derived from the HD-1-Dipel strain (Abbott Laboratories, Chicago, IL) and has moderate toxicity to gypsy moth larvae when added to artificial diet (van Frankenhuyzen et al., 1991, 1992) or genetically expressed in Populus clone NC5339 (Robison et al., 1994; Kleiner et al., 1995). The amount of B. thuringiensis protein in transgenic plant tissue is extremely low (0.001–0.1% of the total protein), so insect bioassays are the most sensitive method for detecting its presence (Fischhoff et al., 1987; Delannay et al., 1989; Fuchs et al., 1990; Cheng et al., 1992). The cry1A(a) gene used in this study is expressed at too low a level for the quantification of the d-endotoxin protein using Western blot analysis. Transformed and nontransformed control plants were propagated from microcultured shoots. The top 3–4 cm of 5–8 cm shoots were excised and rooted ex vitro (Sellmer and McCown, 1989), acclimated, and transferred to 15-l pots (20 × 20 × 46 cm) containing a 2:2:1 mix of peat:sand:field soil and 114 g of Osmocote 13-13-13 fertilizer (Sierra Chemical Corp., Milpitas, CA, USA). Trees were grown under glasshouse conditions at ambient light and temperature conditions during June–September, 1994. Insect Culture. Gypsy moth larvae were reared from eggs (laboratory colony NJSS F41) provided by the USDA Animal and Plant Health Inspection Service, Otis, MA. Eggs masses were surface sterilized in an aqueous solution containing 2% sodium hypochlorite and 1% polyoxyethylene sorbitan monooleate (Tween 80) for 5 min and rinsed with water. Neonates were transferred to 50-cell plastic jelly trays (Hopple Plastics, Florence, KY) containing a low protein (no casein) wheat germ-based diet (ODell and Rollinson, 1966). All larvae were reared in an environmental chamber at 27◦ C and a photoperiod of 16 L:8 D. Insect Bioassays. A no-choice feeding assay was conducted using methods described by Kleiner et al. (1995). The ontogenetic stage of leaf development was standardized by sampling leaf plastochron index (LPI) leaves, 1–6. An apical leaf that has attained a lamina length of 2 cm is designated LPI 0 and the next oldest leaf designated as LPI 1 and so on down the stem. The use of the plastochron index adjusts plants of different sizes or developmental stages to a standardized


pp997-joec-473743


October 17, 2003

20:28



2589

morphological time scale (Larson and Isebrands, 1971). On the morning of the assay (between 0700–0800 hours), leaves were excised at the base of the petiole from each of 10 transformed and 5 nontransformed control trees, sealed in plastic sandwich bags, placed on ice and brought to the University of Wisconsin Department of Entomology Gypsy Moth Quarantine Facility. Leaves were surface sterilized with a 0.53% sodium hypochlorite solution and rinsed in two washes of water. A #8 cork borer was used to remove leaf disks (169.7 mm2 ) from the leaves. Five leaf disks and larvae were used for each leaf from LPIs 3–6. Leaves from LPIs 1 and 2 were small, and, therefore, only one and two leaf disks, respectively, were removed from these leaves for assays. The mean value of within leaf replicates was used to represent leaf level responses for comparisons among leaves. The day before the assay, late second-instars were removed from diet and allowed to molt to third-instar. On the day of the assay, individual preweighed larvae were placed into each cell of a jelly tray along with a leaf disk inserted into an agar block. Initial larval weights did not differ (P > 0.05) among foliage treatments or leaf positions After feeding on leaf disks for 3 d, larvae were weighed, transferred to jelly trays containing low protein diet, and allowed to feed an additional 5 d, and weighed again. Transferring larvae to diet allowed us to check for mortality daily. Monitoring growth while on diet allowed us to evaluate whether surviving larvae avoided feeding or ingested a sublethal dose (see Kleiner et al., 1995). Relative growth rate (RGR) for individual larvae feeding on leaf disks and on diet was calculated as (weightt2 − weightt1 /weightt1 /1t), where 1t is the time interval between measurements at t1 and t2. Total wet weight gain was calculated as the difference between the initial weight and the weight after feeding on diet. Leaf consumption was calculated for each larva on a dry weight basis. One reference leaf disk was removed from each leaf and placed in a glass vial and oven dried (45◦ C) for an estimate of initial dry leaf disk weight. Portions of leaf disk remaining from the assay were oven-dried and weighed and subtracted from the initial estimate to calculate leaf consumption. In some instances where larval feeding was negligible, the reference disk weighed less than the assay disk, resulting in negative consumption values. Negative consumption values were recorded as zero. Foliar Chemistry. Leaf material not used for bioassays was frozen on dry ice, freeze-dried, and ground to a 40 mesh with a Wiley mill. Leaf extracts for the analysis of condensed tannins were prepared by exhaustively (4x) extracting 45–55 mg of dry tissue in 1 ml ice cold 70% acetone with 10 mM ascorbic acid and sonicating for 30 min in ice water. The hydrolytic conversion (BuOH:HCL) of proanthocyanidins to anthocyanidins was used to measure condensed tannin content using a modification of the method of Porter et al. (1986). Condensed tannins from P. alba × P. grandidentata “Crandon” foliage were isolated by absorption chromatography, using a Sephadex LH-20 column and ethanol and acetone as solvents (Hagerman and Butler, 1980), and used to construct standard curves.


2590

pp997-joec-473743

October 17, 2003

20:28


KLEINER, ELLIS, MCCOWN, AND RAFFA

We also assayed the foliage for salicortin and tremulacin, the two most biologically active phenolic glycosides to Lepidoptera (Lindroth et al., 1988; Lindroth and Peterson, 1988). Salicortin and tremulacin were quantified using high performance liquid chromatography (HPLC) as described by Nichols-Orians et al. (1992). HPLC was performed on a Shimadzu LC-10AS with a SPD-10A UV–vis detector (Wood Dale, IL). We extracted 30–34 mg of leaf tissue in 2 ml of methanol with 1.5 mg/ml catechol as an internal standard. Samples were sonicated in ice water for 15 min, centrifuged, and the extracts filtered (0.2 um) for injection onto a reverse-phase C18 column (Spherisorb SS ODS-2, Supelco Inc., Bellefonte, PA) with a gradient of distilled H2 0 and methanol and a flow rate of 1.0 ml/min. The detector was set at 274 nm. Standard curves were constructed from purified salicortin and tremulacin. Statistical Analyses. To examine the effect of leaf position and foliage treatment on foliage consumption or larval growth, we used the PROC MIXED program in SAS (SAS Institute, 1999). Leaf position and foliage treatment were designated as fixed effects, and tree nested within foliage treatment was specified as the random effect. The weighting factor was the number of larvae used for each leaf position. To test for autocorrelation of the residuals among the leaf positions for each dependent variable, we conducted a repeated measures analysis with the AR(1) covariance structure and leaf position nested within tree as the subject effect. Models with and without the repeated statement were compared using the null model likelihood ratio test (SAS Institute, 1999). The difference in the −2 Res Log Likelihoods between the two models has a χ 2 distribution with q − 1 degrees of freedom. Repeated measure models for each of the four dependent variables (foliage consumption, RGR on test foliage, RGR on diet, and total weight gain) were not significantly different than models without the repeated statement (P > 0.05 and 1 df), indicating that each leaf position could be considered as an independent response. A slopes analysis was conducted to test whether changes in larval performance across leaf positions were the same for both foliage treatments. We used PROC MIXED as described above, designating both foliage treatment and leaf position as fixed effects, but considered leaf position (lp) as a continuous variable. The weighting factor was the number of larvae used for each leaf position. PROC MIXED does not provide r 2 values, therefore, weighted linear regression (PROC REG) was used to examine the relationship between the independent variable, leaf position, and the dependent variable, larval performance, for transformed and nontransformed trees separately. The weighting factor was the number of larvae used for each leaf position. Simple linear regression (PROC REG) was also used to examine putative relationships between leaf position and leaf chemistry. Weighted, multiple linear regression methods (PROC REG) were used to evaluate the relationship of individual measures of foliar chemistry from both foliage treatments on foliage


pp997-joec-473743


October 17, 2003

20:28



2591

consumption and growth rates of larvae while on foliage (Freund and Littell, 1991). For these models, foliage consumption or relative growth rates were the dependent variables and foliage treatment, foliar chemistry, and the interaction of treatment by foliar chemistry, were the independent variables. If the interaction is not significant, and the slopes of both lines are negative, then the effect of the d-endotoxin and foliar chemistry are additive. If the interaction is significant, and the slope of the line between foliar chemistry and larval performance is more negative for transformed than nontransformed foliage, then there is a synergistic effect. Simple linear regression was used to examine the relationship of leaf chemistry with larval performance on each type of foliage.

RESULTS

Foliage consumption increased from LPI 1 to LPI 6 for larvae feeding on transformed and nontransformed foliage (leaf position, F = 7.97; df = 4, 52; P < 0.001) (Figure 1A). Less transformed foliage was consumed than nontransformed foliage (treatment, F = 11.77; df = 1, 13; P < 0.005). There was no significant

FIG. 1. Simple weighted regressions of gypsy moth (A) foliage consumption, (B) relative growth rate on foliage, (C) relative growth rate on artificial diet following foliage assay, and (D) total weight gain following leaf and artificial diet assays, on leaf plastochron index for transformed cry1A(a) d-endotoxin and nontransformed foliage.


pp997-joec-473743

October 17, 2003

20:28



2592

Leaf position × Treatment interaction for the amount of foliage consumed during the bioassay (F = 1.21; df = 4, 52; P = 0.32). Growth rates on foliage during the bioassay increased from LPI 1 to LPI 6 (leaf position, F = 6.18; df = 5, 65; P < 0.001) (Figure 1B). Larvae feeding on transformed foliage had lower relative growth rates than larvae feeding on nontransformed foliage (treatment, F = 19.08; df = 1, 13; P < 0.001). There was no significant Leaf position × Treatment interaction for relative growth rates of larvae feeding on foliage (F = 1.03; df = 5, 65; P = 0.40). Growth rates on diet following the bioassay were positively related to the position of the leaf that was consumed (leaf position, F = 3.76; df = 5, 65; P ≤ 0.005) (Figure 1C). Growth rates on diet did not differ between larvae that had fed on transformed versus nontransformed foliage (treatment; F = 0.43; df = 1, 13; P = 0.52), indicating that larvae were able to recover from consuming transformed foliage. There was no significant Leaf position × Treatment interaction for relative growth rates of larvae after transfer to artificial diet (F = 0.93; df = 5, 65; P = 0.46). Although larval growth rates recovered once gypsy moths were transferred to unamended artificial diet, leaf position still affected total weight gained during the bioassay. Larvae gained more weight overall after feeding on older leaves (leaf position, F = 3.80; df = 5, 65; P < 0.005) (Figure 1D). After feeding on diet, larvae did not fully recover from the effects of feeding on transformed foliage. Total larval weight gain was, over the course of feeding on both transformed foliage and diet, lower than for control larvae (treatment, F = 12.46; df = 1, 13; P < 0.004). There was no significant Leaf position × Treatment interaction for total larval weight gain (foliage + diet) during the assay (F = 0.67; df = 5, 65; P = 0.65). One larva died after feeding on LPI 1 from a transformed plant. Larvae feeding on nontransformed leaves consumed more foliage as leaf position increased as compared to larvae feeding on transformed leaves (see Table 1 for regression slopes) (lp × Treament, F = 4.04; df = 1, 58; P ≤ 0.05). The slopes

TABLE 1. REGRESSION FROM COEFFICIENTS SIMPLE REGRESSIONS OF FOLIAGE CONSUMPTION OR LARVAL PERFORMANCE BY GYPSY MOTHS ON Populus LEAF POSITION Transformed Slope Intercept Leaf consumption (mg dry weight) RGR on foliage (mg/mg/day) RGR on diet (mg/mg/day) Total weight gain on foliage + diet (mg)

r

Nontransformed P

0.375 0.504 0.41 0.003 0.006 −0.021 0.41 0.001 0.015 0.479 0.17 0.177 0.005 0.066 0.32 0.012

Slope Intercept 0.79 0.01 0.033 0.007

0.061 0.006 0.422 0.083

r

P

0.57 0.42 0.39 0.38

0.002 0.022 0.035 0.037


pp997-joec-473743


October 17, 2003

20:28



2593

FIG. 2. Simple regressions of tremulacin, salicortin, and condensed tannins on leaf plastochron index for transformed and nontransformed foliage.

did not differ between transformed and untransformed trees for the relationships between leaf position and relative growth rate on foliage (lp × Treatment, F = 1.05; df = 1, 73; P = 0.30), leaf position and relative growth rate on diet (lp × treatment, F = 1.25, df = 1, 73; P = 0.26), and leaf position and total weight gain (foliage + diet) (lp × Treatment, F = 0.30; df = 1, 73; P = 0.58). Concentrations of the phenolic glycosides tremulacin (Figure 2A) and salicortin (Figure 2B) were greatest in leaves at LPI 1 and decreased as leaf position increased to LPI 6 (Table 2; Figure 2A and B). Tremulacin concentrations decreased by up to 5-fold and salicortin concentrations decreased up to 2.4-fold over this ontogenetic gradient. Tremulacin concentrations were negatively related to leaf position for both nontransformed and transformed plants. Salicortin concentrations were negatively related to leaf position in transformed plants only.


pp997-joec-473743

October 17, 2003

20:28



2594

TABLE 2. REGRESSION COEFFICIENTS FROM SIMPLE REGRESSIONS OF FOLIAR CHEMISTRY ON LEAF POSITION IN HYBRID POPLAR CLONE NC5339 Transformed Slope Intercept Tremulacin Salicortin Condensed tannins

−1.25 −0.51 −0.11

10.79 9.14 3.75

Nontransformed

r

P

Slope Intercept

0.92