- An impediment to use of exotic and bioengineered trees in many places is their propensity for spread by pollen and/or seeds. Our laboratory has been using gene editing to induce mutations in floral genes as means to impart stable and reliable genetic containment when this is desirable from social (markets, regulation, public opinion) or ecological perspectives. Our studies focus on one of the most widely planted, productive, and sometimes invasive forest tree species, Eucalyptus grandis x Eucalyptus urophylla.
We used CRISPR-Cas9 to target two genes expected to be important to reproductive development in Eucalyptus. Based on studies in the model plant Arabidopsis, proteins encoded by the eucalypt homolog of EDA33 (EMBRYO DEVELOPMENT ARREST 33) are expected to be essential for seed dispersal, and proteins encoded by the homolog of TDF1 (TAPETAL DEVELOPMENT AND FUNCTION1) are expected to be essential for pollen development. Bioinformatic studies have shown that each has a single putatively orthologous, functional locus in the Eucalyptus genome. This study focused on determination of the nature of mutations induced, with an emphasis on identifying biallelic loss-of-function mutations, and effects of loss-of-function mutation on vegetative growth and morphology in the greenhouse. Ongoing work in our laboratory is studying the effects of mutation of these genes on floral development and fertility.
We generated two CRISPR-Cas9 constructs, each with two guide RNAs that target each gene, and one Cas9-only (no guide RNAs) control construct. We then Agrobacterium-transformed hybrid eucalypt SP7 (E. grandis x urophylla), for which an efficient transformation system was available. We generated 17 and 18 gene insertion events for the EDA33 and TDF1 construct, respectively, and ten events for the Cas9 control. Taking advantage of the natural polymorphisms in the hybrid that we discovered after resequencing near to the target region, we generated and sequenced allele-specific PCR products for each locus. We identified ten insertion events in EDA33 and six in TDF1 events for which there were clear loss-of function mutations in both alleles. Overall, CRISPR-Cas9 was highly efficient in generating knockout mutations, with a rate (per transgenic regenerated shoot) of 83% in EDA33 and 46% in TDF1. The knock-out events and five Cas9 control events were then micropropagated, acclimated, and transplanted into 942 cc3 pots for a greenhouse trial to study variation in plant growth rate and leaf morphology.
The three month greenhouse trial used a randomized block design with an average of six ramets per genotype and 14 non-transgenic controls. We measured height and diameter at the start and end of the trial, and relative chlorophyll density, leaf area and weight, and leaf oil gland density at the end of the trial. In a linear mixed effect model, we conducted an overall F-test where blocks and construct were fixed effects and events were considered random effects.
We found that blocks were highly statistically significant sources of variation (P<<0.05), which was not surprising as trees were sorted into blocks based on size at the start of the experiment. Presumably due to variation in propagation history, the wild type controls were larger at the outset of the study and continued to grow significantly differently from the transgenic trees, so comparisons between the knock-out and non-mutated events were predominantly based on the transgenic controls. At harvest, none of the traits were different at the 5% significance level among the transgenic genotype groups, though leaf area and old gland density were different at the 10% significance level. Equivalence testing of the differences in means of plant characteristics measured at harvest between the knock-out groups and transgenic controls showed that there was unlikely to be a 10% or greater reduction of any of the trait values in the knock-outs. Our results suggest that the functions of these genes are largely restricted to reproductive development in eucalypts, and imparting sterility through their loss-of-function is unlikely to affect productivity. Longer term studies, preferably in the field under normal forestry conditions, are needed to confirm these results.