QUANTITATIVE AND DEVELOPMENTAL GENETICS OF SMALL VEINS IN ZEA MAYS

The major goal of this project is to identify genetic influencers of vein initiation and spacing in the leaves of Zea mays, with particular interest in the small veins responsible for high vein density and C4 photosynthesis.In C4 grasses such as Zea mays, leaf veins can be grouped into lateral, intermediate, small, and transverse classes. Amongst these, small veins are of particular interest because they are unique to C4 grasses, found only in the photosynthetically active leaf blade and not the sheath, and are responsible for maintaining Kranz anatomy and the two-cell vascular spacing rule that enables C4 metabolism. Most plant species perform C3 photosynthesis, but the rarer C4 photosynthetic pathway is more photosynthetically efficient under high temperatures and drought. Because C4 photosynthesis requires the specialized vascular arrangement of Kranz anatomy, the genetics of vascular anatomy is a crucial target for crop improvement and climate resilience.Since C4 photosynthesis was discovered in the late 1960s and was found to facilitate higher photosynthetic efficiency and drought tolerance in plants, there has been a long-standing goal of engineering the C4 pathway into C3 crops such as rice. However, efforts to do so have been hampered by a lack of anatomical understanding--we do not know how the Kranz anatomy required for C4 metabolism develops. Crucially, in maize, it is the small vein vascular subtype found in the leaf blade that enables the high vein density upholding the 2-cell spacing rule required for C4 photosynthesis.This project consists of three primary objectives:1. Map the genetic architecture influencing small vein quantitative traits using deep-learning facilitated high-throughput phenotyping and GWAS.To quantify vascular traits across 728 inbred varieties from the maize Wisconsin Diversity Panel (WiDiv), I have developed a computer vision model in PyTorch which uses a U-NETneural network architecture to perform semantic segmentation of different vascular subtypes on images of cleared and stained maize leaves. To associate genetic variation with my vascular subclass traits, I implemented a GWAS pipeline using the FarmCPU approach and have preliminary results from the genotypes measured so far. The subobjectives described below describe the steps required to arrive at final gene candidates for vein density.a. Digitally image and phenotype the Wisconsin Diversity panel.Completing the analysis of all available WiDiv genotypes will maximize the statistical power of my maize leaf vein GWAS, increasing its ability to detect multiple or subtle, low effect QTL. To facilitate this work, I supervise two part time undergraduate research assistants who contribute to this project by scanning leaves.b. Upgrade neural network with leaf compartment model.I will implement an additional model designed to separate the leaf image into its primary compartments (sheath, auricle, blade). This will allow me to 1) run a compartment-specific vein density GWAS to identify developmentally distinct blade and sheath specific loci, and 2) neutralize random variation introduced from sample preparation that influences sheath vs. blade ratio.c. Identify vein density candidate genes by GWAS.When phenotyping is complete, I will use the FarmCPU model from the rMVP package to perform a GWAS on a recent set of 46 million genomic SNPs. When available, I will order UniformMu insertion lines for promising candidates and check whether presence of the insertion is linked to a detectable change in the measured vascular phenotype.2. Identify genes controlling small vein development using scRNAseq on developing maize leaf primordia.As a complementary method of understanding small vein development, I have been developing single-cell sequencing (scRNAseq) to profile the transcriptomes of maize leaf primordia. Because of its ease of use, low cost, and high-quality data, we have decided to proceed using Fluent Biosciences' PIPseq technology to investigate the early stages of vascular specification.a. Obtain a transcriptomic time-course dataset across meristem and individual leaf primordia.To capture transcriptional signature of developing veins at cellular resolution, I will perform scRNAseq of individual leaf primordia. I will isolate and sequence nuclei due to their ease of extraction (manual chopping and filtration) and lack of cell-wall digestibility bias that influences protoplast approaches. I will use the Fluent Biosciences PIPseq technology to perform scRNAseq on 2-week old transgenic plants. I will prepare transcriptomes from 20k nuclei from entire meristems with leaves, and 2k nuclei from isolated leaf primordia of progressive developmental stages collected using a stereomicroscope.b. Identify vascular/pro-vascular cell clusters using known marker genes and divide them into subclusters for each vein type.I will identify cell types for which genetic markers are unknown, such as cells fated to become small veins, which appear morphologically between plastochron 4-5. I will use Seurat to perform data integration, feature selection, dimensionality reduction, and cell clustering, resulting in cluster assignments for cells in the dataset and a UMAP projective visualization. I will identify putative vascular cell clusters based on enrichment of known marker genes. I will computationally identify vascular subclusters associated with each vein type/their precursors based on developmental timing.c. Use in situ hybridizations to validate candidate genes as being localized to developing small veins.After using the cluster-specific scRNAseq differential gene expression and trajectory analysis to determine a set of 10 candidate genes likely to influence small vein development, the next step is spatial validation. In situ hybridization is a tool to validate locales of gene expression that does not require transformation. I will perform in situs in 2-week-old meristems for 10 of the most promising candidates to show whether these genes are expressed in relevant tissues and patterns prior to or during small vein development.3. Determine developmental timing and physiological consequences of bundle sheath fusions, a cellular spacing defect affecting small veins.While analyzing leaves from diverse genotypes, I found surprising "bundle sheath fusions," (BS fusions) where additional bundle sheath cells appear between two vascular bundles instead of mesophyll cells, violating the Kranz anatomy spacing rule. Although this phenotype resembles loss of function mutations in the scarecrow and shortroot radial patterning genes, it is common across many genotypes in my analysis. I trained my vision model to detect these abnormalities; we are thus able to quantify this trait and find associated SNPs. Anatomical analysis reveals that BS fusions always involve at least one small vein.a. Determine developmental timing of bundle sheath fusions using histology.Knowing when the anomalous BS fusions are first evident is the first step in eventually determining their overall ontogeny and relationship to small vein spacing. I will determine when BS fusions are first observable by collecting and fixing meristems of 2-week-old plants of the most extreme genotypes found by the computer vision tool, as a single individual meristem includes an entire sequence of developmental timepoints. I will use evidence such as cell shape, position, and staining properties to identify the early regions of ectopic bundle sheath.b. Measure photosynthetic efficiency and stomatal conductance of lines with high/low levels of bundle sheath fusion.I will determine whether high levels of this anomalous microscopic trait are negative, neutral, or positive for photosynthetic efficiency, information useful for agronomic breeding programs. I will collaborate with local physiology experts to use a LICOR LI-6800 to measure quantitative photosynthetic traits of genotypes with varying levels of BS fusions.