Resistance gene copy number variation: deciphering stability and inheritance in glyphosate-resistant Amaranthus palmeri

Background/Question/Methods

Herbicide application, a plant stressor, is a selective agent promoting the evolution of resistance in agricultural weed populations. Copy number variation (CNV) of genes has been increasingly detected, and associated with traits critical to environmental adaptation. In Amaranthus palmeri, CNV of the herbicide glyphosate’s target gene, EPSPS, ranges from 1 to >160 copies. This is the primary source of genetic variation of phenotypic glyphosate resistance. In addition, copies of EPSPS occur as extrachromosomal circular DNA (eccDNA) structures. This extrachromosomal positioning of CNV raises fundamental questions about the stability and inheritance of gene copies conferring resistance. We used 750-900 F1 individuals from 30 crosses (mean=30.08 progenies/cross) to examine the following questions: (i) Is CNV stable within individual plants? Do plant tissues produced throughout development exhibit stable EPSPS copy number?; (ii) When experimentally crossed, does progeny CNV deviate from Mendelian expectations?; (iii) If deviations are observed, does the direction or magnitude of deviation depend on parental CNV? Variation in copy number of EPSPS was estimated by probe-based droplet digital PCR and estimates were used to perform controlled crosses. An additional cohort of F0 plants (n=119) were sampled 5x throughout their juvenile to reproductive stages (10-day intervals) to quantify intra-individual variability in CNV.

Results/Conclusions

Variability in CNV within individual plants is minimal for low copy number plants, but intra-individual variation of copy number estimates increases with mean plant copy number. Moreover, variability in CNV within individual plants over their full developmental period departs from Poisson mean-variance relationship expectations at high mean CNV. Standardized residuals of the Poisson expectations exhibit a significant linear relationship displaying more variation in CNV among temporal replicates than expected by the Poisson expectation alone. This suggests that high CNV may exhibit greater instability within a plant. Our results also illustrate a deviation from classical Mendelian inheritance in all F1 crosses. We observed significant differences in progeny CNV mean among crosses, and overall observed a moderate, significant positive relationship between parental and progeny CNV. Progeny from low CNV parents (15-49 copies) exhibited increased CNV relative to parental means (74.0%). In contrast, progeny from high CNV parents (50-92 copies) exhibited a much lower frequency of progeny that exhibited greater CNV than parents (38.4%). This may suggest a genomic constraint on the generation and maintenance of extreme resistance gene copy numbers. Understanding sources of gene copy number variation will inform ongoing study of the evolutionary trajectory of resistance in A. palmeri populations.