Ashleigh Lister

Nitrate induces widespread transcriptional reprogramming in plants, but how transcription factors coordinate these responses across cell types and evolve across species with divergent root patterning remains unclear. Using single-cell transcriptomics in Arabidopsis thaliana and Solanum lycopersicum (tomato) under varying nitrate conditions, we demonstrate that many transcriptional responses are cell type and developmental stage-specific. Arabidopsis NIN-LIKE PROTEIN 7 (NLP7) directly modulates both broadly expressed and cell-type-specific targets, but is not required to establish the cell type specific nature of these patterns. Comparative analysis reveals evolutionary shifts; key nitrogen programs are localized to different cell types in tomato, with the exodermis as a major regulatory hub. Identification of tomato NLP7a and NLP7b direct targets at cellular resolution supports their role in direct modulation of cell type N-transcriptional programs. This is evidenced by NLP7-controlled exodermal cell wall remodeling in low nitrate, highlighting functional conservation despite divergent cellular topographies.

The diversity of plant inflorescence architecture is specified by gene expression patterns. In wheat (Triticum aestivum), the lanceolate-shaped inflorescence (spike) is defined by rudimentary spikelets at the base, which form as a result of delayed spikelet and floral development compared with central spikelets. While previous studies identified gene expression differences between central and basal inflorescence sections, gene expression patterns along the apical-basal axis remain poorly resolved due to bulk tissue-level techniques. Here, we optimize Multiplexed Error Robust Fluorescence In Situ Hybridization, a spatial transcriptomics technique, in wheat inflorescence tissue, enabling transcript localization for 200 genes to cellular resolution across 4 stages of development. Cell segmentation and clustering of 50,000 cells identified 18 expression domains and their enriched genes, revealing the spatio-temporal organization of spikelet and floral development, and characterizing tissue-level gene markers. Using these domain- and cell-level maps, we characterize expression patterns of genes differentially expressed across the apical-basal axis. We identify distinct, spatially coordinated expression patterns distinguishing axillary meristems and their subtending leaf ridges across the apical-basal axis before visible spikelet formation, highlighting factors patterning meristem identity and transition. To support the broader research community, all raw and processed data are publicly available, including through an interactive WebAtlas interface (www.wheat-spatial.com).

In wheat, inflorescence architecture critically determines yield potential, yet its structural complexity and asynchronous development have hindered cellular-resolution studies of spikelet and floret formation. Here, we integrate spatial transcriptomics, high-sensitivity multiplexed error-robust fluorescence in situ hybridization (MERFISH), and snRNA-seq across six developmental stages to generate a spatiotemporal atlas of the wheat inflorescence. We identified 20 cell types, spatially resolved into three categories: 1) proliferating cells within spikelet, marked by active division; 2) supporting cells along the central axis, including pith, cortex, and vasculature; and 3) developmental cells located both inside and at the base of the spikelets. The multi-omics approach enabled identification of the rare cell type ovary. Trajectory inference revealed that spikelets and florets originate from two temporally and spatially distinct sub-clusters of proliferating cells (R7), each defined by high expression of developmental regulators. These findings challenge the conventional model sequential meristem transitions (inflorescence-spikelet-floret) in wheat. Integration of time-series snATAC-seq and snRNA-seq delineated cellular transcriptional regulatory networks (cTRNs) governing spikelet formation, mediated by auxin and cytokinin signaling, and floret formation, driven by MADS-box transcription factors. Cell identity was maintained by cell type-specific accessible chromatin regions (csACRs), which are enriched for SNPs associated with spike-related traits. For instance, SNPs within csACRs of the WFZP and DUO1 promoters affect TaNAC30 binding, regulating supernumerary spikelet phenotypes. Our work provides a mechanistic framework for wheat inflorescence development and identifies csACRs and cTRN nodes as potential targets for optimizing yield-related inflorescence architecture.

The diversity of plant inflorescence architectures is specified by gene expression patterns. In wheat ( Triticum aestivum ), the lanceolate-shaped inflorescence (spike) is defined by rudimentary spikelets at the base which initiate first but subsequently lag in development compared with central spikelets. While previous studies identified gene expression differences between central and basal inflorescence sections, the spatio-temporal dynamics and gradients along the apical-basal axis remain poorly resolved due to bulk tissue-level techniques. Here, using spatial transcriptomics, we profiled 200 genes across four stages of wheat inflorescence development to cellular resolution. Cell segmentation and unsupervised clustering identified 18 expression domains and their enriched genes, revealing dynamic spatio-temporal organisation along the apical-basal axis of the inflorescence. Along this axis, we uncovered distinct and spatially coordinated gene expression gradients patterning meristems prior to the visible delay in basal spikelet development. This study demonstrates the potential for spatial transcriptomics time-series to advance plant developmental biology.

Spikelets are the fundamental building blocks of Poaceae inflorescences, and their development and branching patterns determine the various inflorescence architectures and grain yield of grasses. In wheat (Triticum aestivum), the central spikelets produce the most and largest grains, while spikelet size gradually decreases acropetally and basipetally, giving rise to the characteristic lanceolate shape of wheat spikes. The acropetal gradient corresponds with the developmental age of spikelets; however, the basal spikelets are developed first, and the cause of their small size and rudimentary development is unclear. Here, we adapted G&T-seq, a low-input transcriptomics approach, to characterize gene expression profiles within spatial sections of individual spikes before and after the establishment of the lanceolate shape. We observed larger differences in gene expression profiles between the apical, central, and basal sections of a single spike than between any section belonging to consecutive developmental time points. We found that SHORT VEGETATIVE PHASE MADS-box transcription factors, including VEGETATIVE TO REPRODUCTIVE TRANSITION 2 (VRT-A2), are expressed highest in the basal section of the wheat spike and display the opposite expression gradient to flowering E-class SEPALLATA1 genes. Based on multi-year field trials and transgenic lines, we show that higher expression of VRT-A2 in the basal sections of the spike is associated with increased numbers of rudimentary basal spikelets. Our results, supported by computational modeling, suggest that the delayed transition of basal spikelets from vegetative to floral developmental programs results in the lanceolate shape of wheat spikes. This study highlights the value of spatially resolved transcriptomics to gain insights into developmental genetics pathways of grass inflorescences.

Spikelets are the fundamental building blocks of Poaceae inflorescences and their development and branching patterns determine the various inflorescence architectures and grain yield of grasses. In wheat, the central spikelets produce the most and largest grains, while spikelet size gradually decreases acro- and basipetally, giving rise to the characteristic lanceolate shape of wheat spikes. The acropetal gradient correlates with the developmental age of spikelets, however the basal spikelets are developed first and the cause of their small size and rudimentary development is unclear. Here, we adapted G&T-seq, a low-input transcriptomics approach, to characterise gene expression profiles within spatial sections of individual spikes before and after the establishment of the lanceolate shape. We observed larger differences in gene expression profiles between the apical, central and basal sections of a single spike than between any section belonging to consecutive developmental timepoints. We found that SVP MADS-box transcription factors, including VRT-A2 , are expressed highest in the basal section of the wheat spike and display the opposite expression gradient to flowering E-class SEP1 genes. Based on multi-year field trials and transgenic lines, we show that higher expression of VRT-A2 in the basal sections of the spike is associated with increased numbers of rudimentary basal spikelets. Our results, supported by computational modelling, suggest that the delayed transition of basal spikelets from vegetative to floral developmental programmes results in the lanceolate shape of wheat spikes. This study highlights the value of spatially resolved transcriptomics to gain new insights into developmental genetics pathways of grass inflorescences. One sentence summary: Large transcriptional gradients exist within a wheat spike and are associated with rudimentary basal spikelet development, resulting in the characteristic lanceolate shape of wheat spikes.

Whilst much sequencing effort has focused on key mammalian model organisms such as mouse and human, little is known about the relationship between genome sequencing techniques for non-model mammals and genome assembly quality. This is especially relevant to non-model mammals, where the samples to be sequenced are often degraded and of low quality. A key aspect when planning a genome project is the choice of sequencing data to generate. This decision is driven by several factors, including the biological questions being asked, the quality of DNA available, and the availability of funds. Cutting-edge sequencing technologies now make it possible to achieve highly contiguous, chromosome-level genome assemblies, but rely on high-quality high molecular weight DNA. However, funding is often insufficient for many independent research groups to use these techniques. Here we use a range of different genomic technologies generated from a roadkill European polecat (Mustela putorius) to assess various assembly techniques on this low-quality sample. We evaluated different approaches for de novo assemblies and discuss their value in relation to biological analyses. Generally, assemblies containing more data types achieved better scores in our ranking system. However, when accounting for misassemblies, this was not always the case for Bionano and low-coverage 10x Genomics (for scaffolding only). We also find that the extra cost associated with combining multiple data types is not necessarily associated with better genome assemblies. The high degree of variability between each de novo assembly method (assessed from the 7 key metrics) highlights the importance of carefully devising the sequencing strategy to be able to carry out the desired analysis. Adding more data to genome assemblies does not always result in better assemblies, so it is important to understand the nuances of genomic data integration explained here, in order to obtain cost-effective value for money when sequencing genomes.

Background: Whilst much sequencing effort has focused on key mammalian model organisms such as mouse and human, little is known about the correlation between genome sequencing techniques for non-model mammals and genome assembly quality. This is especially relevant to non-model mammals, where the samples to be sequenced are often degraded and low quality. A key aspect when planning a genome project is the choice of sequencing data to generate. This decision is driven by several factors, including the biological questions being asked, the quality of DNA available, and the availability of funds. Cutting-edge sequencing technologies now make it possible to achieve highly contiguous, chromosome-level genome assemblies, but relies on good quality high-molecular-weight DNA. The funds to generate and combining these data are often only available within large consortiums and sequencing initiatives, and are often not affordable for many independent research groups. For many researchers, value-for-money is a key factor when considering the generation of genomic sequencing data. Here we use a range of different genomic technologies generated from a roadkill European Polecat ( Mustela putorius ) to assess various assembly techniques on this low-quality sample. We evaluated different approaches for de novo assemblies and discuss their value in relation to biological analyses. Results: Generally, assemblies containing more data types achieved better scores in our ranking system. However, when accounting for misassemblies, this was not always the case for Bionano and low-coverage 10x Genomics (for scaffolding only). We also find that the extra cost associated with combining multiple data types is not necessarily associated with better genome assemblies. Conclusions: The high degree of variability between each de novo assembly method (assessed from the seven key metrics) highlights the importance of carefully devising the sequencing strategy to be able to carry out the desired analysis. Adding more data to genome assemblies not always results in better assemblies so it is important to understand the nuances of genomic data integration explained here, in order to obtain cost-effective value-for-money when sequencing genomes.

Background: Thorough understanding of complex model systems requires the characterisation of processes in different cell types of an organism. This can be achieved with high-throughput spatial transcriptomics at a large scale. However, for plant model systems this is still challenging as suitable transcriptomics methods are sparsely available. Here we present Grid-seq, an easy to adopt, micro-scale spatial-transcriptomics workflow that allows to study expression profiles across small areas of plant tissue at a fraction of the cost of existing sequencing-based methods. Results: We compare the Grid-seq method with widely used library preparation methods (Illumina TruSeq). In spatial experiments we show that the Grid-seq method is sensitive enough to identify expression differences across a plant organ. We further assess the spatial transcriptome response of Arabidopsis thaliana leaves exposed to the bacterial molecule flagellin-22. Conclusion: We show that our method can be used to identify known, rapidly flagellin-22 elicited genes, plant immune response pathways to bacterial attack and spatial expression patterns of genes associated with these pathways.

A high quality genome sequence of your model organism is an essential starting point for many studies. Old clone based methods are slow and expensive, whereas faster, cheaper short read only assemblies can be incomplete and highly fragmented, which minimises their usefulness. The last few years have seen the introduction of many new technologies for genome assembly. These new technologies and new algorithms are typically benchmarked on microbial genomes or, if they scale appropriately, human. However, plant genomes can be much more repetitive and larger than human, and plant biology makes obtaining high quality DNA free from contaminants difficult. Reflecting their challenging nature we observe that plant genome assembly statistics are typically poorer than for vertebrates. Here we compare Illumina short read, PacBio long read, 10x Genomics linked reads, Dovetail Hi-C and BioNano Genomics optical maps, singly and combined, in producing high quality long range genome assemblies of the potato species S. verrucosum . We benchmark the assemblies for completeness and accuracy, as well as DNA, compute requirements and sequencing costs. We expect our results will be helpful to other genome projects, and that these datasets will be used in benchmarking by assembly algorithm developers.