Plants possess myriad defenses against their herbivores, including constitutive and inducible chemical compounds and regrowth strategies known as tolerance. Recent studies have shown that plant tolerance and resistance are positively associated given they are co-localized in the same molecular pathway, the oxidative pentose phosphate pathway. However, given that both defensive strategies utilize carbon skeletons from a shared resource pool in the oxidative pentose phosphate pathway there are likely costs in maintaining both resistance-tolerance strategies. Here we investigate fitness costs in maintaining both strategies by utilizing a double knockout of cyp79B2 and cyp79B3, key enzymes in the biosynthetic process of indole glucosinolates, which convert tryptophan to indole-3-acetaldoxime (IAOx) and is further used to produce indole glucosinolates. These mutant plants are devoid of any indole glucosinolates thus reducing plant resistance. Results show that knocking out indole glucosinolate production and thus one of the resistance pathways leads to an approximate 46% increase in fitness compensation shifting the undercompensating wild-type Columbia-0 to an overcompensating genotype following damage. We discuss the potential mechanistic basis for the observed patterns.

Glucosinolates

Glucosinolates constitute a large and diverse group of defensive secondary metabolites characteristic of the order Brassicales, which includes A. thaliana, our organism of study (Muller et al. 2010). Glucosinolates (mustard oil glucosides) are nitrogen- and sulfur-rich natural plant secondary products that consist of a sulfonated oxime and a β-thioglucose moiety, but differ in side chain structures (Pfalz et al. 2009). There have been ~40 glucosinolates found in Arabidopsis, out of the 120 glucosinolates identified, most of which are classified into three subgroups based on the biosynthetic amino acid precursor, those subgroups being indole, aliphatic and benzenic (Sonderby et al. 2010). Indole and aliphatic glucosinolates constitute most of the diversity of glucosinolates in A. thaliana (Brown et al. 2003). Indole glucosinolates, our chemicals of interest, are composed of four individual compounds: glucobrassicin, 4-methoxy-glucobrassicin, neoglucobrassicin and 4-hydroxyglucobrassicin.

Many studies have shown that glucosinolate breakdown products deter generalist and specialist herbivores on A. thaliana (e.g., Agrawal and Kurashige 2003) and act in defense against pathogens (Schlaeppi et al. 2010). Upon herbivory, glucosinolates stored in the vacuole are mixed with the enzyme myrosinase (known as the “mustard bomb”), a β-thioglucosidase that is separated in scattered specialist cells known as myrosin cells (Ratzka et al. 2002). Myrosinase cleaves the β-glucose moiety from glucosinolates, leading to a variety of toxic breakdown products, such as bioactive nitriles, epithionitriles and isothiocyanates based on reaction conditions and protein factors such as epithiospecifier protein (Zhang et al. 2006).

In A. thaliana, indole glucosinolate synthesis involves a catalyzed conversion of tryptophan to indole-3-acetaldoxime (IAOx) carried out by two cytochrome P450s, cyp79B2 and cyp79B3 (Hull et al. 2000, Celenza et al. 2005). IAOx is further catalyzed through four subsequent reactions to form glucobrassicin, the most abundant indole glucosinolate found in A. thaliana (Bender and Celenza 2009). Glucobrassicin can then be further modified to the other three indole glucosinolates found in A. thaliana with modified indole rings (Pfalz et al. 2009).

Costs of Resistance

To assess costs between plant tolerance and resistance 100 seeds of Col-0 and 100 seeds of cyp79B2 cyp79B3 double mutant lines were planted and grown. While single cyp mutant lines show little deficiency in the ability to produce indole glucosinolates, double mutants are completely devoid of any indole glucosinolates (Glawischnig et al. 2004; Zhao et al. 2002). The cyp79B2 cyp79B3 double mutant line was kindly obtained from the laboratory of John Celenza at Boston University (Department of Biology, Boston, Massachusetts). 

The cyp79B2 cyp79B3 double mutant was uncovered from T-DNA insertion lines of cyp79B2 and cyp79B3 in the Col-0 background, identified from the Salk Institute collection of T-DNA insertion lines by PCR. Primers 79B2-5P (5’-TGGACAAGTATCATGACCCAATC ATCCACG-3’) and LB (5’-GGCAATCAGCTGTTGCCCGTCTCACTGGTG-3’) were used to identify a T-DNA insertion at 1512 bp after the ATG of the cyp79B2 gene. The insertion is in the second exon of the cyp79B2 gene. Primer 79B3-5P (5’- TGTTCTATGCATGGACTGGT GGTCAACATG-3’) and LB were used to identify a T-DNA insertion at 1425 bp after the ATG site of the cyp79B3 gene. The insertion is in the intron between the two exons of the cyp79B3 gene. The T-DNA insertion in the cyp79B3 gene is a tandem T-DNA insertion with LB flanking sequences at both ends of the T-DNA insertion. The insertions were confirmed by DNA sequencing of the PCR fragments generated with the LB primer and the gene-specific primers (Zhao et al. 2002). 

Arabidopsis lines were grown in a greenhouse on the campus of the University of Illinois, Champaign, under 12 hours of light (~100 uE/M²/sec) and dark. Plants were grown individually in 3.5-inch pots in L1 Sunshine mix. Seeds/seedlings were kept moist during germination, and plants were then watered daily to maintain soil moisture without saturating the soil. Plants were not fertilized. When inflorescences reached 6 cm, about 3.5 weeks, 50 plants of each ecotype were randomly clipped, leaving approximately 1 cm of inflorescence (comparable to natural mammalian herbivory (Scholes et al. 2016)).

At 6.5 weeks, 30 plants of Col-0 (15 clipped, 15 unclipped) were analyzed for indole glucosinolate concentration. Inflorescence material was taken from both clipped and unclipped plants. In addition, the level of indole glucosinolates were assessed from 30 plants (15 clipped and 15 unclipped) samples of the cyp79B2 cyp79B3 double mutant line to verify that there were in fact, undetectable levels of indole glucosinolates, consistent with the findings of Zhao et al. 2002. All samples were frozen in liquid nitrogen and stored at -80C prior to freeze-drying. Freeze-dried tissues were ground into a fine powder and stored at -20C prior to glucosinolate analysis. Glucosinolates were extracted from finely ground freeze-dried tissue, converted to desulphoglucosinolates with arylsulfatase and analyzed via high-pressure liquid chromatography (HPLC) as described by Brown et al. (2003). Freeze-dried powder 50 mg and 0.5 mL of 70% methanol were added to 2.5 mL tubes and placed on a heating block at 95C for 10 min, mixing frequently. Samples were cooled on ice and 0.125 mL glucosinolabin was used as an internal standard and centrifuged at 3,000xg for 10 minutes. The supernatant was saved and the pellet was re-extracted with another 0.5 mL 70% methanol at 95C for 10 minutes and the two extracts were combined. Protein was subsequently precipitated with 0.15 mL of a 1:1 mixture of 1 M barium acetate and 1 M lead acetate and centrifuged at 12,000xg for 1 min. Each sample was then loaded onto a column containing DEAE Sephadex A-25 resin for desulfation via arylsulfatase for 18 h and the remaining desulfo-GS eluted. Desulphoglucosinolates were separated on an HPLC system (Agilent 1100 HPLC system, with a G1311A bin pump, a G1322A vacuum degasser, a G1316A thermostatic column compartment, a G1315B diode array detector and an HP 1100 series G1313A autosampler) with a variable ultraviolet detector set at 229 nm wavelength. Elution of desulphoglucosinolates occurred over 45 minutes with a linear gradient of 0% to 20% acetonitrile in water with a flow rate of 1.0 mL/min. Glucosinolate concentration was established using glucosinalbin as an internal standard, a glucosinolate not found in A. thaliana. UV response factors for different glucosinolates were used as determined by Wathelet et al. (2001).  Indole glucosinolates were estimated by adding the four indole glucosinolates observed in Col-0.

Upon plant senescence (8 weeks), remaining plants of both ecotypes were analyzed for fitness (70 plants per ecotype; 35 clipped, 35 unclipped). Fitness measures included the number of siliques, seeds per plant, and the average seed weight per plant. Our previous studies have shown that seeds are a good measure of plant fitness as there are no significant differences in germination success between clipped and unclipped plants of A. thaliana (e.g., Siddappaji et al. 2013). We measured total silique number for each plant and counted total seed production in 3 randomly selected siliques from each plant. Seeds per silique were averaged per ecotype and treatment and then multiplied by total silique number for each plant to obtain seed totals per plant. Average seed weights were measured by weighing 50 seeds per plant from 10 plants per ecotype x treatment group. Each weight measurement was then divided by the number of seeds to yield average seed weight for each ecotype x treatment group. Additionally, we measured rosette diameter at time of senescence for Col-0 and cyp79B2 cyp79B3.

Potential differences in composite seed production were assessed using an analysis of variance and Type III sums of squares with two treatment factors (genotype and clipping). Rosette diameter was used as a covariate to adjust for differences in plant size. In addition, we regressed total seed production on rosette diameter of A. thaliana for both Col-0 and cyp79B2 cyp79B3 double mutant lines to justify the use of rosette diameter as an appropriate covariate in the model (see Fig. S3). Differences among treatments within genotypes were determined using Tukey pairwise comparisons. Glucosinolate production, between clipped and unclipped Col-0 plants, was assessed using a two-sample t-test. Statistical analyses were conducted in Systat (version 13; Systat, Chicago, Illinois, USA).

Word, Excel, Systat