genetic variation and relationships of constitutive and ...

P1: GXB Journal of Chemical Ecology [joec]

pp751-joec-459294

February 26, 2003

11:8

Style file version June 28th, 2002

C 2003) Journal of Chemical Ecology, Vol. 29, No. 2, February 2003 (°

GENETIC VARIATION AND RELATIONSHIPS OF CONSTITUTIVE AND HERBIVORE-INDUCED GLUCOSINOLATES, TRYPSIN INHIBITORS, AND HERBIVORE RESISTANCE IN Brassica rapa

DONALD F. CIPOLLINI,1,∗ JEREMIAH W. BUSCH,1,3 KIRK A. STOWE,2,4 ELLEN L. SIMMS,2 and JOY BERGELSON1 1 Department

of Ecology and Evolution The University of Chicago 1101 E. 57th St., Chicago, Illinois 60637 2 Department

of Integrative Biology University of California, Berkeley 3060 Valley Life Sciences Building # 3140 Berkeley, California 94720-3140 (Received June 25, 2002; accepted October 15, 2002)

Abstract—We examined genetic variation in inducibility and in constitutive and herbivore-induced levels of glucosinolates, trypsin inhibitors, and resistance to herbivory in families of Brassica rapa originating from a wild population. We also examined phenotypic and genetic correlations among absolute levels of these traits in control and induced plants. We grew seedlings of 10 half-sib families in pairs in pots, and exposed one plant per pair to folivory by Trichoplusia ni larvae. Two days later, we sampled all plants for total glucosinolate and trypsin inhibitor levels and examined the preference and consumption by T. ni larvae of previously damaged (induced) and undamaged (control) plants. There was no significant variation among sire families in the induction of glucosinolates or trypsin inhibitors by T. ni feeding. Total glucosinolate levels in either control or induced plants did not vary by family. In contrast, trypsin inhibitor levels in both control and induced plants varied significantly by family. Trichoplusia ni fed less on induced plants than on control plants in the bioassay, but neither the ∗

To whom correspondence should be addressed at current address: Wright State University, Department of Biological Sciences, 3640 Colonel Glenn Highway, Dayton, Ohio 45435. E-mail: [email protected] 3 Current address: Indiana University, Department of Biology, Jordan Hall, Bloomington, Indiana 47405. 4 Current address: University of Evansville, Department of Biology, 1800 Lincoln Avenue, Evansville, Indiana 47722.

285 C 2003 Plenum Publishing Corporation 0098-0331/03/0200-0285/0 °


pp751-joec-459294

February 26, 2003

286

11:8


CIPOLLINI ET AL. induction of resistance by prior T. ni feeding nor absolute levels of damage done to control and induced plants varied significantly by sire family. Temporal blocking strongly affected trypsin inhibitor levels and the response of some families in the bioassays. There were no significant phenotypic or genetic correlations of levels of glucosinolates or trypsin inhibitors with each other or with damage in either control or induced plants. Overall, these results suggest that in the B. rapa population that we studied, both total glucosinolate content and biological resistance to herbivory by T. ni was nonvariable and almost universally inducible by prior T. ni feeding. In contrast, control and induced levels of trypsin inhibitors varied genetically and have the capacity to respond to future selection imposed by herbivores. However, the role of these defenses in constitutive or induced resistance to T. ni in this species remains unclear. Key Words—Brassica rapa, induction, genetic correlation, glucosinolates, herbivory, phenotypic plasticity, resistance, quantitative genetics, trypsin inhibitors, variation.

INTRODUCTION

Most plant species must defend themselves against damage by insect herbivores and other natural enemies (Simms and Rausher, 1987). In the face of such enemies, selection for increased resistance to herbivory is thought to have generated a vast taxonomic diversity in proteins and secondary metabolites with defensive function (Erlich and Raven, 1964; Rosenthal and Berenbaum, 1992). Accordingly, to examine the hypothesis that putative plant defenses are adaptive in the face of herbivore attack, evolutionary ecologists have long sought to determine the costs and benefits of the expression of plant resistance. The production of chemical and physical defenses is assumed to be costly to fitness in plants, although the detection of costs has remained challenging (Brown, 1988; Simms, 1992; Bergelson and Purrington, 1996; Zangerl et al., 1997; Agrawal, 1998; Baldwin, 1998; Mauricio, 1998; Purrington, 2000; Cipollini et al., 2002). Allocation of limited resources to defense is thought to incur direct “physiological costs” by decreasing the amount of resources that could otherwise be allocated to fitness (Chew and Rodman, 1979; Zangerl et al., 1997; Mauricio, 1998; Cipollini, 2002). Allocation to defense may also incur indirect “ecological costs,” such as reduced competitive ability or tolerance to herbivory, that are seen only when plants are growing in an ecologically relevant context (e.g., Bergelson, 1994; Stowe, 1998; van Dam and Baldwin, 1998; Cipollini, 2002). If defense is indeed costly to fitness, then plant fitness would be maximized if individuals expressed costly defenses only when warranted by the environment (i.e., when spatially and temporally variable populations of insect herbivores are present) (Karban and Baldwin, 1997). Phenotypic plasticity in the production of defenses (i.e., inducibility) may serve as a cost-saving mechanism, by restricting the production of defenses to environments in which herbivores are present (Karban and Baldwin, 1997). However, some


pp751-joec-459294

GENETIC VARIATION IN

Brassica rapa

February 26, 2003

11:8


287

authors have suggested that unique benefits (rather than cost-savings) of employing inducible defenses could favor inducibility over constitutive strategies in plants (Karban et al., 1997). Induced responses allow plants to maintain higher fitness than uninduced plants when herbivores are present in the field (Agrawal, 1998, 2000; Baldwin, 1998). Nevertheless, inducibility may not continue to evolve in response to herbivores if there is not heritable genetic variation in a population for this form of phenotypic plasticity. Although several studies have demonstrated genetic variation in constitutive levels of defenses (e.g., Kollipara et al., 1994; Mauricio, 1998; Kliebenstein et al., 2001), few studies have examined genetic variation in the induction of defenses within a plant species (Zangerl and Berenbaum, 1990; van Dam and Vrieling, 1994; Agrawal et al., 2003). The first evidence of genetic variation in induction was in a study of wild parsnip (Zangerl and Berenbaum, 1990), in which heritable genetic variation in constitutive and induced levels of furanocoumarins was detected in two populations that historically experienced different selective pressures from the parsnip webworm, Depressaria pastivacella. Brassica rapa is defended from herbivores in part by glucosinolates. Glucosinolates are ubiquitous throughout the Brassicaceae (Rodman, 1991) and are known to vary quantitatively and qualitatively at the constitutive level in species such as Arabidopsis thaliana (Mauricio, 1998; Kliebenstein et al., 2001). Upon wounding, preformed glucosinolates are hydrolyzed by the enzyme myrosinase (thioglucosidase) to glucose and sulfate, along with various nitrile, isothiocyanate, and thiocyanate ions that are thought to be toxic or deterrent to generalist insect herbivores and some pathogens (Giamoustaris and Mithen, 1995; Hopkins et al., 1998a; Lambrix et al., 2001; Kliebenstein et al., 2002). Although constitutively produced, glucosinolates and myrosinase are also wound-inducible in many brassicaceous plants, including B. rapa (Broadway and Colvin, 1992; Bodnaryk, 1994; but see Agrawal, 2000), and are commonly associated with induced resistance to generalist pests (e.g., Agrawal, 1998). In contrast, glucosinolates serve as hostacceptance cues and feeding stimulants for specialist beetles and lepidopterans (e.g., Renwick and Lopez, 1999), which appear to be relatively undeterred at constitutive or wound-induced levels (Broadway and Colvin, 1992; Bodnaryk, 1994; Giamoustaris and Mithen, 1995; Kliebenstein et al., 2002; but see Agrawal, 2000). In addition to glucosinolates, brassicaceous plants may be defended by proteinase inhibitors. Serine proteinase inhibitors are found at relatively high levels in many brassicaceous species, including B. rapa (Broadway, 1989) and competitively inhibit serine proteases, such as trypsin and chymotrypsin, thereby inhibiting proteolysis in animals and microorganisms (Broadway, 1995). Serine proteinase inhibitors are constitutively produced and developmentally regulated in Brassica napus and Brassica oleraceae, but are also wound-inducible (Broadway and Missurelli, 1990; Cipollini and Bergelson, 2000, 2001). Proteinase inhibitors can lower in situ growth of many insect herbivores (Broadway and Colvin, 1992),


pp751-joec-459294

February 26, 2003

288

11:8


CIPOLLINI ET AL.

are associated with induced resistance in many plants (Thaler et al., 1996; Zhao et al., 1997), and can deter herbivory and reduce insect performance when expressed in transgenic plants (Johnson et al., 1990). As for glucosinolates, some insects appear to possess effective physiological counteradaptations to the presence of host-plant proteinase inhibitors in their gut and are less affected by proteinase inhibitors than are other herbivores (Broadway, 1995; Jongsma and Bolter, 1997). Data from a pilot study of B. rapa indicated that genetic variation existed among paternal half-sib families (individuals sharing the same paternal parent, but different maternal parents) in the induction of glucosinolates by larval Trichoplusia ni feeding but that levels were not correlated with the amount of leaf area removed by T. ni in a bioassay of the same plants (K. Stowe, J. Busch, and E. Simms, unpublished data). Here, we examined whether induction of trypsin inhibitors by T. ni feeding, in conjunction with glucosinolates, varies among families and can explain resistance to herbivory in B. rapa. Genetic variation in inducibility and in absolute levels of glucosinolates and trypsin inhibitors has never been examined in an integrated study. Relative roles in defense and trade-offs that could constrain their evolution also remain unclear. Specifically, this experiment was designed to address the following questions: (1) Does B. rapa exhibit genetic variation in the induction of glucosinolates, trypsin inhibitors, and biological resistance elicited by T. ni feeding? (2) Do constitutive and induced levels of glucosinolates, proteinase inhibitors, and biological resistance vary genetically in B. rapa? (3) Are there significant genetic or phenotypic correlations between glucosinolates, trypsin inhibitors, and biological resistance in control and induced plants?

METHODS AND MATERIALS

Study Species. Brassica rapa L. (syn. B. campestris: Brassicaceae; turnip rape) is a self-incompatible annual plant, native to Eurasia, that is represented by both crop cultivars and naturalized populations throughout North America. Seeds used in the current study were the second-generation offspring of plants collected from a wild population at the San Joaquin Fresh Water Reserve near the University of California, Irvine. Trichoplusia ni (Lepidoptera: Noctuidae; cabbage looper moth) is a generalist herbivore that uses brassicaceous plants as preferred hosts (Broadway, 1995). T. ni larvae have often been used in induction experiments and bioassays of resistance with brassicaceous plants, and they appear to vary in their sensitivity to different resistance factors (e.g., Broadway and Colvin, 1992; Broadway, 1995; Agrawal, 2000). Breeding Design. We used plants from preexisting paternal half-sib families, created for an earlier experiment, as seed sires in a quantitative genetics breeding


pp751-joec-459294


Brassica rapa

February 26, 2003

11:8


289

design to produce plants for the current study. Based on data from the preliminary experiment, we identified 10 paternal half-sib families that appeared to vary in mean glucosinolate levels. We used them as sires to increase our ability to detect genetic correlations between glucosinolates, trypsin inhibitors, and biological resistance (Falconer and Mackay, 1989). During the spring of 1999, we raised an individual from each chosen half-sib family (10 total plants) in the University of Chicago greenhouse along with 45 randomly chosen individuals from the wild population in California. Following the onset of flowering, each individual from the 10 half-sib families (sires) was crossed with three randomly chosen wild plants (dams), producing 30 full-sib seed families nested within 10 paternal halfsib seed families. To control for maternal effects on seed development (Agrawal et al., 1999), five flowers on each dam were hand-pollinated in each cross with pollen collected from several flowers from each individual sire. A separate camel hair paintbrush was used to collect and transfer pollen for each cross. Seeds were collected following maturation on maternal plants in late spring. Experimental Design. During the summer of 1999, experimental plants were grown from seed in a temperature-controlled greenhouse at the University of Chicago (20-25◦ C) under ambient light supplemented by sodium vapor lights. Within each of the two temporal blocks, six plants from each of the 30 full-sib families were planted, yielding 180 individuals per block and 360 individuals in the entire experiment. Several seeds from each family were initially planted in flats of Promix BX potting medium, and seedlings were watered twice daily. Four days following germination, selected plants were transplanted to 15-cm pots in the same medium to decrease the effect of root binding on the uptake of nutrients, which could interfere with the production of chemical defenses (Baldwin, 1988; Cipollini and Bergelson, 2001). At this time, two individuals from each full-sib family were paired by size and transplanted into each pot. Thus, there were three full-sib pairs of plants per half-sib family used in each temporal block of this experiment. One individual from each pair was subsequently induced (see below) while the other plant served as an undamaged control. Full-sib pairs were generated based upon similarity in size to control for any potential size effects on levels of glucosinolates or proteinase inhibitors (Broadway and Missurelli, 1990; Hopkins et al., 1998a). Paired plants did have the potential to communicate chemically through soil- or airborne factors (e.g., Siemens et al., 2002). The extent of communication among seedlings was unknown, but likely made our results conservative if it occurred. Moreover, paired plant designs are statistically powerful, and facilitated our bioassay that integrated both preference and consumption of control and induced plants. Plants within each pot were randomly assigned control or induced status at the four-leaf stage, 14 days following transplantation. T. ni larvae used to induce and to bioassay resistance were raised from eggs (obtained from Entopath Inc.), on artificial cabbage looper diet (Southland Products Inc.). Approximately 15 larvae were raised in each diet cup in a growth chamber set to 22◦ C with a


290

pp751-joec-459294

February 26, 2003

11:8


CIPOLLINI ET AL.

14L:10D light cycle. To ensure that similar instars were used to induce and to bioassay the plants during each temporal block, one cohort of larvae was started to be used for the induction of plant defenses, and another was started 2 days later for use in bioassays of resistance. Larvae used to induce plants were contained on leaves in clip cages. Clip cages were constructed out of the base plates of 30-mm clear plastic Petri dishes, using a metallic hair clip glued to each cage as a hinge. Additionally, a 15-cm thin wooden dowel was glued to each cage that could be inserted into the soil at the base of each plant to suspend the cage on a leaf without weighing down and damaging the leaf. On induced plants, one 9-day-old larva was placed in a cage on the terminal end of the second true leaf and was allowed to feed. An empty cage was placed on the second true leaf of each control plant. After approximately 12 hr, during which time larvae ate ∼9.6 cm2 of leaf area, all larvae and cages were removed from all plants. In the three cases where larvae did not feed during this time period, an equivalent amount of leaf area was mechanically damaged by crushing. This has been shown to induce trypsin inhibitors and glucosinolates in Brassica (Bodnaryk, 1994; Cipollini and Bergelson, 2000, 2001). Three samples of leaf tissue were taken for chemical analysis from the third true leaf from all plants 48 hr after removal of cages. Two leaf disks (∼1 cm diam.) were taken from either side of the midrib to quantify glucosinolate content and leaf disk dry weight. A leaf sample (∼4 cm2 ) was removed from the tip of the same leaf to quantify trypsin inhibitors. Two leaf samples were placed in prelabeled microfuge tubes, flash frozen in liquid nitrogen, placed in a cooler of Dry Ice, and stored in a −80◦ C freezer until analysis. One of the leaf disks was placed in a small, prelabeled coin envelope, placed in a drying oven, and subsequently weighed on a microbalance. T. ni larvae raised as described were used to bioassay resistance of each pair of plants. Although all plants were wounded during the leaf sampling protocol, only previously herbivore-induced plants had had sufficient time to respond to damage with changes in leaf chemistry (Cipollini and Bergelson, 2000). One 9-day-old larva was placed on the soil between each pair of plants and allowed to choose freely and feed on the plants. To eliminate a directional bias, larvae were placed such that the long axis of the body was perpendicular to each of the two plants. Following larval placement, pots were enclosed in spun mesh polyester bags to prevent larval escape. Twelve hours later, all larvae were removed, and damage to each leaf on each control and induced plant was estimated by using a clear acetate grid. Induction of biological resistance was measured as the difference between the amount of leaf area removed from each induced plant and the amount of leaf area removed from its paired control plant (yielding a negative value if resistance was induced in the previously damaged plant). This bioassay measure integrates both insect preference for control or induced plants and the amount of damage it caused to either plant.


pp751-joec-459294


Brassica rapa

February 26, 2003

11:8


291

Chemical Analyses. Total glucosinolate content of a leaf sample can be inferred from the amount of glucose liberated from a glucosinolate extract upon hydrolysis with the enzyme, myrosinase (Broadway and Colvin, 1992; Siemens and Mitchell-Olds, 1998). We analyzed glucosinolate content according to Siemens and Mitchell-Olds (1998). Our method of sample preparation differed from that protocol only in that leaf disk samples were heated for 5 min at 62◦ C in 0.5m MeOH to denature endogenous myrosinase. Glucosinolate concentrations were expressed as milligrams of glucose per gram of leaf dry weight. Trypsin inhibitor content was measured in soluble protein extracts by using a radial diffusion assay through a trypsin-containing agar as in Cipollini and Bergelson (2000). Protein concentration in leaf extracts was quantified by using the Bradford assay (Bradford, 1976) with Biorad protein dye reagent and bovine serum albumin as a standard. Trypsin inhibitor concentrations were expressed as milligrams of trypsin inhibitor per gram of extract protein. For each chemical defense, leaf samples obtained from control plants were used to quantify constitutive defense levels, and leaf samples obtained from damaged plants were used to quantify induced defense levels for each full-sib pair in the experiment. Statistical Analyses. Variation in Induction of Chemical Defense and Biological Resistance. We first tested for variation in induction (damaged levels minus control levels for each full-sib pair) of glucosinolates, trypsin inhibitors, and biological resistance using three separate ANOVAs. Each analysis was conducted using sire, block, and dam (nested within sire) as main effects. To determine whether half-sib families performed differently in the two blocks, a sire × block interaction term was also included. All statistical analyses were performed on SAS (Version 8, SAS Institute, Inc., Cary, North Carolina, USA). Variation in Constitutive and Induced Levels of Defenses and Biological Resistance. Subsequently, we tested for variation in constitutive and induced levels of glucosinolates and trypsin inhibitors and for variation in damage to control and induced plants using six separate ANOVAs. Each analysis was conducted using sire, block, and dam (nested within sire) as main effects, and a sire × block interaction term. Genetic Correlations among Defenses and Biological Resistance. To determine the degree to which the expression of glucosinolates, trypsin inhibitors, and the amount of damage received in the bioassay reflect the same genetic character, we calculated genetic correlations among the 10 sire half-sib family means of these traits (Falconer and Mackay, 1989). Genetic correlations were performed separately for control and induced plants. Alpha levels were Bonferroni adjusted to 0.0168 for both control and induced plants to maintain an overall alpha level of 0.05 for three tests. Phenotypic Correlations among Defenses and Biological Resistance. To examine for the presence of nonrandom associations between putative defenses and damage, phenotypic correlations among glucosinolate levels, trypsin inhibitor


pp751-joec-459294

February 26, 2003

11:8

292


CIPOLLINI ET AL.

levels, and the amount of damage received in the bioassay for each individual in the experiment were performed. Phenotypic correlations were performed separately for control and induced plants. Alpha levels were Bonferroni adjusted to 0.0168 for both control and induced plants to maintain an overall alpha level of 0.05 for three tests.

RESULTS

Variation in Induction of Glucosinolates, Trypsin Inhibitors, and Biological Resistance. No model effect explained variation in the induction of glucosinolates (Table 1), or trypsin inhibitors (Table 1) by T. ni feeding (data not shown). No model effect affected the induction of biological resistance by T. ni feeding, except the sire × block interaction (Table 1, Figure 1). These results reveal a lack of variation among sires in the induction of defense chemicals and biological resistance by prior T. ni feeding in the B. rapa population that we studied, but they indicate that unknown factors associated with temporal blocking affected the induction of biological resistance in different sires. Variation in Constitutive and Induced Levels of Glucosinolates. No model effect explained phenotypic variation in glucosinolate levels in either control plants (Table 2) or in induced plants (Table 3) (data not shown). These results indicate that there is a lack of variation among the sire families for the genes responsible for glucosinolate production, and no effect of experimental blocking. Variation in Constitutive and Induced Levels of Trypsin Inhibitors. In contrast to total glucosinolates, some model effects explained a significant amount of the phenotypic variation in trypsin inhibitor levels. Sire and dam (nested within sire) influenced the level of trypsin inhibitors in control plants (Table 2, Figure 2A), revealing the importance of additive genetic variation and maternal effects on constitutive levels. Moreover, control plants in the first block had higher trypsin inhibitor levels than control plants in the second block, revealing the importance TABLE 1. F VALUES FROM ANOVAS EXAMINING INDUCTION OF GLUCOSINOLATE, TRYPSIN INHIBITOR, AND BIOLOGICAL RESISTANCE IN B. rapa BY T. ni FEEDINGa

Source

df

Glucosinolate

Trypsin inhibitor

Biological resistance

Sire Dam (sire) Block Sire × block

9 20 1 9

1.37 (0.208) 0.93 (0.556) 0.01 (0.943) 0.74 (0.671)

0.98 (0.460) 1.35 (0.161) 0.59 (0.445) 0.67 (0.735)

0.38 (0.944) 0.94 (0.533) 0.11 (0.737) 1.98 (0.046)

aP

values are shown in parentheses. Bold type indicates significance.


pp751-joec-459294


February 26, 2003

11:8


Brassica rapa

293

FIG. 1. Induction by T. ni feeding of biological resistance to T. ni in sire half-sib families of B. rapa. Black bars represent block 1 and gray bars represent block 2. Bars represent mean differences (± SE) in leaf damage by T. ni between paired induced and control plants.

of environmental or experimental variation among blocks. Both sire and block affected the level of trypsin inhibitors in induced plants, with trypsin inhibitor levels again being much higher in the first block than in the second block (Table 3, Figure 3A). These results indicate genetic variation in induced levels of trypsin inhibitors, as well as an effect of environmental or experimental differences between temporal blocks. Variation in Damage on Control and Induced Plants. Only the sire × block interaction explained variation in damage to control plants in the bioassay (Table 2, Figure 2B), while only the blocking term explained variation in damage to induced plants in the bioassay (Table 3, Figure 3B). The results indicate a general lack of TABLE 2. F VALUES FROM ANOVAS EXAMINING ABSOLUTE LEVELS OF GLUCOSINOLATE, TRYPSIN INHIBITOR, AND LEAF DAMAGE IN B. rapa PLANTS PREVIOUSLY UNDAMAGED BY T. ni FEEDINGa

Source

df

Glucosinolate

Trypsin inhibitor

Damage

Sire Dam (sire) Block Sire × block

9 20 1 9

0.78 (0.635) 0.97 (0.498) 0.44 (0.506) 1.17 (0.322)

3.80 (