CITRUS GENOME SEQUENCE RESOURCES: DEVELOPMENT AND UTILIZATION TO CREATE HLB RESISTANT CULTIVARS

Huanglongbing (HLB), also known as citrus greening, is the most destructive, devastating disease of citrus in the world. HLB affects all citrus species, and commercial citrus cultivars such as sweet oranges, grapefruit, and many mandarins, are most susceptible. The best long-term strategy, recognized and recommended by the National Academy of Sciences report on HLB, is to develop citrus scion and rootstock cultivars that can tolerate or resist HLB. Traditional breeding and the use of genome editing technologies, such as the CRISPR system, may be able to create highly tolerant or even resistant new citrus cultivars. However, successful implementation to achieve the goals will be substantially enhanced by having the highest quality genome sequences of both sensitive and tolerant types, to identify the most critical genes associated with sensitivity, tolerance, and resistance. This information can be used in traditional breeding to enable application of new breeding approaches such as genomic prediction and selection.It is important to understand that most commercially important citrus in Florida (sweet orange and grapefruit, for example) are derived through serial somatic mutations, and therefore traditional breeding approaches to incorporate HLB-tolerance cannot preserve their exact genotypes, nor exactly replicate their unique quality attributes. For this reason, gene editing is an attractive option because it can be used to change only the specific genes involved in susceptibility or resistance, while leaving all other attributes of the cultivar unaffected and intact.The research will focus first on producing genome sequence assemblies of 5 commercially significant types of citrus, three of which are sensitive (sweet orange, grapefruit, and Clementine mandarin) and two considered tolerant (Lisbon lemon and LB8-9 Sugar Belle® mandarin hybrid). As sequencing technology has made tremendous advances in recent years, it has become possible to produce chromosome scale assemblies of nearly complete sequence, thereby improving the effectiveness of gene editing experiments. Comparisons of these genomes to identify differences among them in terms of HLB responses will identify the important genes upon which editing ought to be focused. The project will also develop a catalog of these genes and define sites within the sequences to be targeted by CRISPR gene editing technology, to develop improved tolerance or perhaps even full resistance in these specific commercially important citrus types. As sequencing costs continue to decline and the quality of the output increases, these techniques will be used to develop genome sequence assemblies of other citrus species and their relatives to expand the genomic knowledge basis of citrus in general, to access additional sources of resistance to HLB and other limiting diseases, as well as providing the platform to explore genetic control of other important traits for citrus improvement, such as fruit quality. Such information can be utilized not only for gene editing for cultivar improvement, but also it will facilitate and greatly enhance the efficiency of developing valuable new cultivars through traditional breeding approaches. This is fertile ground for future and essential research, to generate and exploit new citrus genome sequence resources to maximize the likelihood of developing HLB resistant citrus trees.

1) Sequence genomesPreparation of high molecular weight genomic DNA: High molecular weight genomic DNA will be extracted and the integrity of the extracted citrus genomic DNA will be determined by performing pulsed-field electrophoresis, and its purity by OD260/OD280 ratio. PacBio HiFi sequencing: Genomes will be sequenced using Pacific Biosciences HIFi technology. Sequencing will be performed in commercial labs offering the service to a target depth of at least 40-50x coverage.Assembling the genomes: State-of-the-art methods will be used to assemble the genomes. Based on our previous experiences the following aspects will be taken into consideration to produce high quality reference genomes. a. Heterozygosity. Intraspecific heterozygosity of most citrus species is ~0.2-0.3%, and between species divergence ~2%, providing sufficient inter-haplotype variation to allow the two haplotypes of a diploid cultivar to be assembled into separate contigs. Such haplotype-specific assemblies are possible when sequencing depth is sufficient; 85x sequencing redundancy allows for 40+x coverage per haplotype and haplotype separation. b. Assembly methodology. Genomes will be assembled using custom pipelines developed by our collaborators at UC Berkeley and the US Department of Energy Joint Genome Institute c. Long-range linking data. A limitation of shotgun sequencing approaches is that contiguity achieved by assembling overlapping reads is usually limited to megabase-scale. Citrus chromosomes are ~30 Mbp. To bridge this gap, chromatin conformation capture ("HiC") has been shown to provide useful intra-chromosomal contig linkages and has been successfully applied to several plant genomes.d. Chromosome-scale assembly, curation, and polishing. Chromosome-scale assemblies will be made, using appropriate state of the art assembly techniques. These assemblies will be further curated by manual inspection and editing using the JuiceBox toolkit, to ensure the robustness of the chromatin-guided assembly. After these steps, attempts will be made to close any remaining gaps, using PBJelly. Finally, Illumina short-read sequence will be combined with the high depth PacBio sequence to "polish" the assembly, identifying residual errors (based on disagreements with aligned sequences) and correcting them. The resulting accuracy will have fewer than one error per megabase. This polishing step will also flag problematic poorly assembled regions that cannot be corrected with existing methods, for the benefit of downstream users.2) Sequence transcriptomesa. Preparation of high-quality RNAs and transcriptome sequencing: Total RNA from leaf samples and other relevant tissue types will be extracted using Trizol reagent. mRNA will be purified using the polyA Spin mRNA Isolation Kit and quantified using a Qubit fluorometer with Qubit RNA HS Assay Kit. Reverse transcription will be performed using SuperScript II with Oligo dT primer. cDNA will be amplified using the standard Phusion PCR protocol. The library preparation for amplicon sequencing will be done using the manufacturer's protocol, to provide full-length reads for nearly all transcripts captured as mRNA.b. Transcriptome assembly: Raw data will be uploaded automatically by Metrichor software for cloud-based base-calling and quality filtering. Poretools will be used to extract fasta reads from Basecalled fast5 files. The majority of cDNAs should be sequenced as full length. Error correction will be carried out using both hybrid and de novo correction methods. Hybrid error correction can be carried out by proovread using short-read (2 x 100 bp paired-end reads) sequencing data previously generated on the Illumina HiSeq platform. Nanocorrect will be used for de novo error correction. Canu will be used to assemble nanopore reads, and racon will be used for polishing canu assemblies.3) Annotate genomes and comparative genome analysisGenome annotation: The citrus genomes will be annotated using a previously developed annotation pipeline that includes methods to identify transposable elements and other repetitive elements, using a de novo repeat finding algorithm RepeatModeler v1.0.5,to produce a comprehensive library of transposable elements. Manual review will ensure that genes from large families (e.g., R-genes) are not incorrectly identified as repetitive elements. Each genome will be masked, producing a GFF-formatted file output by RepeatMasker, that can be made publicly available through JGI's Phytozome. To predict protein-coding gene structures (i.e., exon-intron organization) and alternate splice forms for each gene, the long-read transcriptome resources generated (see above), as well as a range of short-read RNA-Seq data that captures a wide variety of tissues and includes samples with and without HLB infection, will be used. Using the methods outlined by Wu et al. (2014) the long- and short-read transcriptome-data are aligned, and protein sequences from nearby species are aligned to identify regions of putative gene homology. For citrus, most genes have homology evidence, and with the new transcriptome resources generated in this project ~75% of citrus genes will be supported by expression data.Standard functional analysis with KOG, KEGG, ENZYME, Pathway and the InterPro family of protein analysis tools, as well as gene family information, will be computed and made available through Phytozome.4) Extract full-length sequences for HLB-responsive gene sequences and predict candidate sites for genome editingPrevious meta-analyses of citrus genes reported to be responsive to CLas inoculation and HLB disease development in 22 citrus gene profiling experiments have identified nearly 3500 microarray probe sets that can detect differential gene expressions before and after CLas inoculation, between HLB-susceptible and -tolerant citrus genotypes, and/or among different stages of HLB development. These RNA-seq results may be reanalyzed against the new genome assemblies, to determine whether some previously identified potential causal genes are structural variants absent in the new assemblies. As new data emerge from genome-wide association studies, genetic linkage mapping, transcriptome profiling, gene coexpression network analysis, etc. the list will be narrowed to about 100 genes that may consistently respond to CLas inoculation and HLB disease development. These genes would be top candidates for CRISPR-based knocking down or knocking out for HLB resistance.To extract full-length sequences and understand their structures, DNA and protein sequences will be extracted from citrus genomes and compared. They will be annotated with PSI-blast and hmmsearch. The sequences will be aligned using Clustal Omega, MAFFT, and phylogentic trees will be built with MEGA, PHYLIP. Gene structure and transcript isoforms will also be compared. The expression of selected responsive genes will be confirmed using qPCR.The assembled chromosome-scale genome assemblies and the list of HLB-responsive genes can be used to assess specificity and off-target potential of potential gRNA designs, and the cutting efficiency of Cas9 enzyme directed by these gRNAs. Online computational tools can predict candidate cutting sites within the responsive genes; these candidate sites can be scored with the assistance of available web tools that are pre-loaded with citrus genome sequence. Selected cutting sites can be manually checked then to ensure high specificity and no or minimal off-target cleavage potential. Priorities will be given to those sites in the CDS region and close to an ATG start codon. When choices for a specific gene are limited, we can increase the on-target effects by introducing more than one cutting site. Primers for making CRISPR guide RNA for the cutting sites will be designed according to the CRISPR plasmids currently used.