- Plant sexual reproduction requires a broad array of molecular mechanisms to proceed successfully. Some of these mechanisms are well-studied, but our knowledge of them as a whole is fundamentally incomplete. Pollen tube growth is a key part of this process, facilitating the delivery of the sperm cells to the female ovule. In this work we examine pollen function on several scales in the model organism Zea mays. First, we characterize two homeologous genes in the nop family that promote pollen tube growth. These genes are highly expressed in maize pollen. The proteins encoded by these genes contain predicted functional domains involved in calcium and phosphoinositide binding and membrane organization, key mechanisms in pollen tube growth. Mutations in nop genes led to a reduction in pollen fitness, demonstrated by reduced transmission of mutant alleles when heterozygous mutant plants were crossed through the male, but not through the female. Examinations of nop mutant pollen tubes revealed shorter tubes in mutants than in wild type pollen. In one nop mutant, tube lengths were particularly sensitive to chemicals that interfered with phosphoinositide signaling, suggesting a role for nop intimately connected to those pathways.
Next, we studied a larger set of genes that were highly expressed in maize pollen. To identify these genes, we sequenced the maize pollen transcriptome at four developmental stages. Highly expressed genes in some of these stages were tested for their contributions to pollen fitness. Mutations in these genes were linked to fluorescent seed markers that could be tracked using a phenotyping system that we developed. Transmission rates of these mutant alleles were quantified, leading to the identification of several mutants that had negative effects on pollen fitness, suggesting functional roles for the genes they disrupted. One of these genes was gex2, which we linked to problems at fertilization. In this experiment, we demonstrated the utility of phenotyping screens to track kernel markers, linking high expression to contributions of genes to pollen fitness.
The final section of this dissertation describes the method we developed to track kernels on a large scale. We created a novel high-throughput phenotyping system to scan maize ears. The system spins a maize ear while capturing a video, which is then processed into a flat projection of the surface of the ear. This platform creates a permanent record of the ear, which can then be measured to describe a variety of phenotypes. We designed and trained a deep-learning-based computer vision pipeline to rapidly identify kernel phenotypes. Our system identifies fluorescent and non-fluorescent kernels accurately, expanding our ability to study the contributions of many genes to pollen function.
This dissertation describes an effort to understand the function of genes, molecules that are invisible to the human eye. By building on the work of many scientists, we present several ways to "see" gene functions through genetic assays, phenotyping, and sequencing. These methods describe several scales of molecular function, from single genes to entire sets of genes. Ultimately, we hope to lay the groundwork for future studies that take advantage of multiscale phenotyping to uncover the molecular mechanisms of plant sexual reproduction.