Periodic Reporting for period 1 - LeafMap (Evolution of leaf shape diversity in Angiosperms using single cell approaches)
Reporting period: 2023-07-01 to 2025-07-31
The LeafMap project aimed to uncover how genetic networks influence the extraordinary diversity of leaf morphologies in flowering plants. While leaves of different species begin their development as morphologically indistinguishable primordia, they develop into distinct leaf forms through changes in developmental programs. Understanding this process at the resolution of individual cells and across evolutionary lineages represents a significant challenge in plant developmental biology.
To address this, the project employed state-of-the-art single-cell and spatial genomics approaches in both Brassicaceae (Arabidopsis thaliana, Cardamine hirsuta) and Solanaceae (Solanum lycopersicum, Capsicum annuum) lineages, which have independently evolved complex leaves. The objectives were threefold: (1) reconstructing the molecular basis of leaf development at cellular resolution in Brassicaceae, (2) identifying spatial transcriptional patterns during leaf morphogenesis, and (3) comparing the evolution of regulatory networks underlying leaf complexity across lineages.
Throughout the fellowship, substantial progress was made toward these objectives. High-quality single-cell multiome datasets were generated across four species, providing the first multiome-based comparative atlas of simple versus complex leaf development. These analyses revealed both conserved and lineage-specific cell states and regulators, as well as convergence in the redeployment of meristematic programs to support leaf complexity. In parallel, experimental pipelines for nuclei isolation and whole-mount FISH were established, ensuring robust validation capacity. Progress in spatial transcriptomics was slower than anticipated, but contingency strategies were implemented to address this.
Overall, the project has delivered novel insights into the evolution of leaf development, generated valuable datasets and methodological advances for the community, and positioned the fellowship holder at the forefront of plant single-cell biology.
1.1 Objectives
Objective 1: Reconstructing the genetic basis of leaf development and complexity at cellular resolution.
This objective offers a comprehensive understanding of leaf development from a genome-wide and cellular perspective. Under Objective 1, we successfully established a robust nuclei isolation and single-cell multi-omics pipeline for Arabidopsis thaliana and Cardamine hirsuta, generating high-quality single-cell transcriptome and chromatin accessibility datasets across various developmental stages. The generated data revealed both conserved and species-specific cell populations. Notably, regulators involved in leaf complexity, such as STM and RCO, could be traced within specific cell states, providing mechanistic insights into the developmental divergence between simple and complex leaves. Additionally, the multimodal data allows us to understand not only the gene expression of different cell populations but also their regulatory landscapes, thanks to the paired chromatin accessibility data. However, due to the large volume of data generated, analyzing and interpreting the chromatin accessibility data is still ongoing.
Objective 2: Discovery and exploration of GRNs in spatial context during leaf morphogenesis
Progress was slower than expected due to technical difficulties, especially the discontinuation of suitable commercial kits for high-resolution spatial transcriptomics in plants. This unexpected problem led us to put contingency plans into action. The host institute invested in new equipment to process frozen sections, including a CryoStar NX70 cryostat from Epredia (approx. 70,000 EUR), which enabled us to explore multiple options to achieve our goal. So far, we have successfully optimized cryosectioning of the target plant materials and performed pilot experiments to support the future success of these methods. Additionally, a multi-color whole-mount RNA in situ hybridization technique was adopted using Cardamine hirsuta shoot and leaf tissues. This pilot study aims to locate cell cycle regulators during leaf development and is currently being used to identify candidate regulators based on the single-cell data generated. This successful optimization highlights our dedication to improving our techniques for the benefit of the project.
Objective 3: Understanding leaf complexity from an evolutionary point of view
Under Objective 3, we extended the single-cell multiome framework to Solanaceae, generating a cross-lineage dataset of ~180,000 nuclei from four species. Current improvements in machine learning techniques have also enabled us to create a common embedding of the transcriptome space across the four species, which opens up new avenues for comparative single-cell studies. Our findings suggest that the convergent recruitment of meristematic programs in the evolution of complex leaves is a significant step towards understanding leaf complexity from an evolutionary perspective.
To address this, the project employed state-of-the-art single-cell and spatial genomics approaches in both Brassicaceae (Arabidopsis thaliana, Cardamine hirsuta) and Solanaceae (Solanum lycopersicum, Capsicum annuum) lineages, which have independently evolved complex leaves. The objectives were threefold: (1) reconstructing the molecular basis of leaf development at cellular resolution in Brassicaceae, (2) identifying spatial transcriptional patterns during leaf morphogenesis, and (3) comparing the evolution of regulatory networks underlying leaf complexity across lineages.
Throughout the fellowship, substantial progress was made toward these objectives. High-quality single-cell multiome datasets were generated across four species, providing the first multiome-based comparative atlas of simple versus complex leaf development. These analyses revealed both conserved and lineage-specific cell states and regulators, as well as convergence in the redeployment of meristematic programs to support leaf complexity. In parallel, experimental pipelines for nuclei isolation and whole-mount FISH were established, ensuring robust validation capacity. Progress in spatial transcriptomics was slower than anticipated, but contingency strategies were implemented to address this.
Overall, the project has delivered novel insights into the evolution of leaf development, generated valuable datasets and methodological advances for the community, and positioned the fellowship holder at the forefront of plant single-cell biology.
1.1 Objectives
Objective 1: Reconstructing the genetic basis of leaf development and complexity at cellular resolution.
This objective offers a comprehensive understanding of leaf development from a genome-wide and cellular perspective. Under Objective 1, we successfully established a robust nuclei isolation and single-cell multi-omics pipeline for Arabidopsis thaliana and Cardamine hirsuta, generating high-quality single-cell transcriptome and chromatin accessibility datasets across various developmental stages. The generated data revealed both conserved and species-specific cell populations. Notably, regulators involved in leaf complexity, such as STM and RCO, could be traced within specific cell states, providing mechanistic insights into the developmental divergence between simple and complex leaves. Additionally, the multimodal data allows us to understand not only the gene expression of different cell populations but also their regulatory landscapes, thanks to the paired chromatin accessibility data. However, due to the large volume of data generated, analyzing and interpreting the chromatin accessibility data is still ongoing.
Objective 2: Discovery and exploration of GRNs in spatial context during leaf morphogenesis
Progress was slower than expected due to technical difficulties, especially the discontinuation of suitable commercial kits for high-resolution spatial transcriptomics in plants. This unexpected problem led us to put contingency plans into action. The host institute invested in new equipment to process frozen sections, including a CryoStar NX70 cryostat from Epredia (approx. 70,000 EUR), which enabled us to explore multiple options to achieve our goal. So far, we have successfully optimized cryosectioning of the target plant materials and performed pilot experiments to support the future success of these methods. Additionally, a multi-color whole-mount RNA in situ hybridization technique was adopted using Cardamine hirsuta shoot and leaf tissues. This pilot study aims to locate cell cycle regulators during leaf development and is currently being used to identify candidate regulators based on the single-cell data generated. This successful optimization highlights our dedication to improving our techniques for the benefit of the project.
Objective 3: Understanding leaf complexity from an evolutionary point of view
Under Objective 3, we extended the single-cell multiome framework to Solanaceae, generating a cross-lineage dataset of ~180,000 nuclei from four species. Current improvements in machine learning techniques have also enabled us to create a common embedding of the transcriptome space across the four species, which opens up new avenues for comparative single-cell studies. Our findings suggest that the convergent recruitment of meristematic programs in the evolution of complex leaves is a significant step towards understanding leaf complexity from an evolutionary perspective.
Work Package 1 - Single cell multiomic comparison of leaf development
Under this Work Package, we planned to establish a nuclei-based single-cell multiome dataset (RNA + ATAC) in Arabidopsis and Cardamine to reconstruct gene regulatory networks during leaf morphogenesis. We successfully optimized nuclei isolation (M1.1) for downstream single-cell approaches from small tissue inputs. This methodology was recently presented at the Gordon Research Conference for Single Cell Approaches in Plant Biology (August 2025), and we are currently preparing a manuscript to further share our methodological improvements, which produce reproducible, high-quality nuclear suspensions. This optimization was also strongly supported by the Max Planck Genome Centre.
Using this method, we generated a time- series single- cell multimodal dataset for A. thaliana and C. hirsuta, comprising 16 samples and approximately 100, 100,000 high-quality nuclei data points (M1. 2). To compare the developmental cellular space of the two species (M1.3) I also analyzed a protoplast-based, lower-resolution dataset from a PhD project at the host institute. This project is currently being prepared for publication in an international scientific journal. Additionally, I conducted a comparative analysis of the data generated within the project. These datasets demonstrated a high degree of cell-type preservation, albeit with some nuanced regulatory differences in homologous cell types. For instance, we observed that in Arabidopsis, certain auxin biosynthetic genes are strongly co-expressed with the transcription factor encoding gene AtWOX1. In contrast, the expression of the same auxin biosynthetic genes in Cardamine is predicted by either ChWOX1 or ChCUC2 in a cell-type-specific manner. This supports previous findings on the specific role of the margin-patterning transcription factors, ChCUCs, in complex leaf development (Hu et al. PNAS. 2024 doi: 10. 1073/pnas. 2321877121).
We also identified a species-specific cell population in Cardamine with regulatory signatures linked to complex leaf development. Specifically, STM and KNAT1, two meristem-associated transcription factors, are expressed in this cluster. This finding emphasizes not only the involvement of meristem-derived regulatory programs in complex leaf development but also their influence on the cell state makeup of different leaf types. Comparing the cell type composition with data from meristematic tissue further revealed that this cell state is shared between the complex leaf of Cardamine and certain subpopulations of shoot apical meristem cells in both species.
Overall, this data uncovered the developmental similarities and differences between these two closely related species with different leaf morphologies, providing a solid foundation for our understanding of system- level evolutionary changes that lead to morphological innovations (D1.1).
To better understand the regulatory evolution of these differences between Arabidopsis and Cardamine, I am continuing this project to investigate chromatin accessibility differences and their relationships with gene expression in the generated data.
Work Package 2 - Discovering pattern generators using spatial transcriptomics
We aimed to establish spatial transcriptomic maps in Brassicaceae leaf primordia to uncover transcriptional gradients during morphogenesis. Initial work focused on optimizing cryosectioning of shoot apices (M2.1) which was successfully achieved. Under this Milestone, the host institute also invested in upgrading its cryosectioning facilities with an industry-leading cryostat (see above) with of the aim to ensuring the success of the project. We obtained reproducible tissue sections with intact morphology, suitable for downstream spatial profiling. The implementation of the STOmics spatial transcriptomics system is currently underway, and it is expected to yield the initially desired data within a reasonably short timeframe. Integration with MorphoGraphX (MGX) software has begun, with early efforts focused on reconstructing cryosections into three-dimensional tissues for future spatial integration.
Progress in genome-wide spatial transcriptomics was hindered by the discontinuation of specific commercial kits within Germany, necessitating an adaptation of our strategy (M2.2). To mitigate these challenges, we initiated a contingency plan (M2.3) focusing on multi-color whole-mount fluorescent RNA in situ hybridization in collaboration with Molecular Instruments (https://www.molecularinstruments.com/(opens in new window)). This method enables targeted high-resolution spatial mapping of selected genes. This allows us to validate candidate regulators identified from the single-cell datasets in situ.
These methodological developments will enable us to connect spatial gene expression patterns with morphogenetic processes once genome-wide data become available. Although Deliverable 2.2 (spatial distribution of genes and eGRNs) is pending, the progress in sample preparation, contingency implementation, and development of complementary validation pipelines has laid the foundation for achieving the spatial component of the project.
Work Package 3 - Functional validation of leaf developmental regulators
The initial plan was to create new CRISPR/Cas9 mutants of selected regulators and cis-regulatory elements (M3.2) based on candidate genes identified from WP1 and WP2. However, due to time constraints within the fellowship, new mutant lines could not be established. Instead, we can utilize the extensive collection of existing mutants and reporter lines available in the host laboratory to begin functional validation of candidates (M3.1).
A major achievement for this WP related to WP2 was the development of a whole-mount FISH pipeline, which now enables spatial validation of gene expression with cellular resolution in developing leaf primordia. This complements the single-cell datasets by providing in situ confirmation of regulatory patterns. For example, FISH probes for complex-leaf specific cell type regulators are being used to test predictions based on the gene regulatory networks reconstructed in WP1. Although Deliverables 3.1 and 3.2 (mutant generation and morphodynamic analysis) were not fully completed, the infrastructure established for molecular validation, along with the use of existing genetic resources, ensures that functional testing of candidate regulators is feasible and already in progress.
Work Package 4 - Evolutionary comparison of leaf shape diversity in Angiosperms
This WP expanded the single-cell multiomic framework beyond Brassicaceae to include Solanaceae species (M4.1). We successfully generated high-quality datasets from Solanum lycopersicum (tomato) and Capsicum annuum (pepper), in addition to the Arabidopsis and Cardamine datasets from WP1. In total, ~180,000 nuclei were profiled across four species, creating one of the most comprehensive comparative single-cell atlases of leaf development to date.
Cross-species integration showed that species with complex leaves (Cardamine and tomato) clustered together in cellular state space, despite belonging to different evolutionary lineages. This suggests a convergent redeployment of gene regulatory programs during the development of complex leaves. Notably, regulators of meristematic tissues, such as STM orthologs, were consistently enriched in meristematic and complex-leaf–specific cell populations. This indicates a convergent recruitment of meristematic programs in independent lineages with complex leaves. The shared cell states between meristematic tissues and complex-leaf-specific cell populations also highlight potential cellular homology between particular meristematic sub-populations and cells in complex leaves, thus advancing a a long-standing debate in the field of leaf development and complexity. These findings provide direct evidence that the evolution of leaf complexity in Brassicaceae and Solanaceae involved both conserved genetic frameworks and lineage-specific innovations (e.g. RCO in Cardamine). Computational analyses of cross-species eGRNs are ongoing (M4.2). Still, preliminary results already demonstrate the power of the generated dataset to uncover the evolutionary origins and innovations behind morphological diversity in angiosperm leaves.
Deliverables 4.1 and 4.2 (pan-angiosperm atlas and identification of conserved and specific modules) are mostly accomplished, with a manuscript being prepared to share these findings.
Under this Work Package, we planned to establish a nuclei-based single-cell multiome dataset (RNA + ATAC) in Arabidopsis and Cardamine to reconstruct gene regulatory networks during leaf morphogenesis. We successfully optimized nuclei isolation (M1.1) for downstream single-cell approaches from small tissue inputs. This methodology was recently presented at the Gordon Research Conference for Single Cell Approaches in Plant Biology (August 2025), and we are currently preparing a manuscript to further share our methodological improvements, which produce reproducible, high-quality nuclear suspensions. This optimization was also strongly supported by the Max Planck Genome Centre.
Using this method, we generated a time- series single- cell multimodal dataset for A. thaliana and C. hirsuta, comprising 16 samples and approximately 100, 100,000 high-quality nuclei data points (M1. 2). To compare the developmental cellular space of the two species (M1.3) I also analyzed a protoplast-based, lower-resolution dataset from a PhD project at the host institute. This project is currently being prepared for publication in an international scientific journal. Additionally, I conducted a comparative analysis of the data generated within the project. These datasets demonstrated a high degree of cell-type preservation, albeit with some nuanced regulatory differences in homologous cell types. For instance, we observed that in Arabidopsis, certain auxin biosynthetic genes are strongly co-expressed with the transcription factor encoding gene AtWOX1. In contrast, the expression of the same auxin biosynthetic genes in Cardamine is predicted by either ChWOX1 or ChCUC2 in a cell-type-specific manner. This supports previous findings on the specific role of the margin-patterning transcription factors, ChCUCs, in complex leaf development (Hu et al. PNAS. 2024 doi: 10. 1073/pnas. 2321877121).
We also identified a species-specific cell population in Cardamine with regulatory signatures linked to complex leaf development. Specifically, STM and KNAT1, two meristem-associated transcription factors, are expressed in this cluster. This finding emphasizes not only the involvement of meristem-derived regulatory programs in complex leaf development but also their influence on the cell state makeup of different leaf types. Comparing the cell type composition with data from meristematic tissue further revealed that this cell state is shared between the complex leaf of Cardamine and certain subpopulations of shoot apical meristem cells in both species.
Overall, this data uncovered the developmental similarities and differences between these two closely related species with different leaf morphologies, providing a solid foundation for our understanding of system- level evolutionary changes that lead to morphological innovations (D1.1).
To better understand the regulatory evolution of these differences between Arabidopsis and Cardamine, I am continuing this project to investigate chromatin accessibility differences and their relationships with gene expression in the generated data.
Work Package 2 - Discovering pattern generators using spatial transcriptomics
We aimed to establish spatial transcriptomic maps in Brassicaceae leaf primordia to uncover transcriptional gradients during morphogenesis. Initial work focused on optimizing cryosectioning of shoot apices (M2.1) which was successfully achieved. Under this Milestone, the host institute also invested in upgrading its cryosectioning facilities with an industry-leading cryostat (see above) with of the aim to ensuring the success of the project. We obtained reproducible tissue sections with intact morphology, suitable for downstream spatial profiling. The implementation of the STOmics spatial transcriptomics system is currently underway, and it is expected to yield the initially desired data within a reasonably short timeframe. Integration with MorphoGraphX (MGX) software has begun, with early efforts focused on reconstructing cryosections into three-dimensional tissues for future spatial integration.
Progress in genome-wide spatial transcriptomics was hindered by the discontinuation of specific commercial kits within Germany, necessitating an adaptation of our strategy (M2.2). To mitigate these challenges, we initiated a contingency plan (M2.3) focusing on multi-color whole-mount fluorescent RNA in situ hybridization in collaboration with Molecular Instruments (https://www.molecularinstruments.com/(opens in new window)). This method enables targeted high-resolution spatial mapping of selected genes. This allows us to validate candidate regulators identified from the single-cell datasets in situ.
These methodological developments will enable us to connect spatial gene expression patterns with morphogenetic processes once genome-wide data become available. Although Deliverable 2.2 (spatial distribution of genes and eGRNs) is pending, the progress in sample preparation, contingency implementation, and development of complementary validation pipelines has laid the foundation for achieving the spatial component of the project.
Work Package 3 - Functional validation of leaf developmental regulators
The initial plan was to create new CRISPR/Cas9 mutants of selected regulators and cis-regulatory elements (M3.2) based on candidate genes identified from WP1 and WP2. However, due to time constraints within the fellowship, new mutant lines could not be established. Instead, we can utilize the extensive collection of existing mutants and reporter lines available in the host laboratory to begin functional validation of candidates (M3.1).
A major achievement for this WP related to WP2 was the development of a whole-mount FISH pipeline, which now enables spatial validation of gene expression with cellular resolution in developing leaf primordia. This complements the single-cell datasets by providing in situ confirmation of regulatory patterns. For example, FISH probes for complex-leaf specific cell type regulators are being used to test predictions based on the gene regulatory networks reconstructed in WP1. Although Deliverables 3.1 and 3.2 (mutant generation and morphodynamic analysis) were not fully completed, the infrastructure established for molecular validation, along with the use of existing genetic resources, ensures that functional testing of candidate regulators is feasible and already in progress.
Work Package 4 - Evolutionary comparison of leaf shape diversity in Angiosperms
This WP expanded the single-cell multiomic framework beyond Brassicaceae to include Solanaceae species (M4.1). We successfully generated high-quality datasets from Solanum lycopersicum (tomato) and Capsicum annuum (pepper), in addition to the Arabidopsis and Cardamine datasets from WP1. In total, ~180,000 nuclei were profiled across four species, creating one of the most comprehensive comparative single-cell atlases of leaf development to date.
Cross-species integration showed that species with complex leaves (Cardamine and tomato) clustered together in cellular state space, despite belonging to different evolutionary lineages. This suggests a convergent redeployment of gene regulatory programs during the development of complex leaves. Notably, regulators of meristematic tissues, such as STM orthologs, were consistently enriched in meristematic and complex-leaf–specific cell populations. This indicates a convergent recruitment of meristematic programs in independent lineages with complex leaves. The shared cell states between meristematic tissues and complex-leaf-specific cell populations also highlight potential cellular homology between particular meristematic sub-populations and cells in complex leaves, thus advancing a a long-standing debate in the field of leaf development and complexity. These findings provide direct evidence that the evolution of leaf complexity in Brassicaceae and Solanaceae involved both conserved genetic frameworks and lineage-specific innovations (e.g. RCO in Cardamine). Computational analyses of cross-species eGRNs are ongoing (M4.2). Still, preliminary results already demonstrate the power of the generated dataset to uncover the evolutionary origins and innovations behind morphological diversity in angiosperm leaves.
Deliverables 4.1 and 4.2 (pan-angiosperm atlas and identification of conserved and specific modules) are mostly accomplished, with a manuscript being prepared to share these findings.
During this action, both the developed techniques and the scientific results achieved have proven impactful for the broader research community. The project produced one of the largest single-cell multiomic datasets for plants to date, including approximately 180,000 nuclei from Arabidopsis, Cardamine, tomato, and pepper. This resource represents the first comparative cellular-resolution atlas of leaf development, solving long-standing questions about how complex leaf shapes develop and offering a framework for new hypotheses on the mechanisms behind morphological innovation. The finding that complex-leaf species cluster together in cell-state space, regardless of their phylogenetic relationships, provides a striking example of evolutionary convergence. The identification of lineage-specific refinements in marginal regulators, such as RCO and LMI1, highlights the subtle ways in which conserved modules are redeployed to generate diversity. These discoveries not only deepen our understanding of leaf complexity but also give broader insights into the principles of regulatory evolution.
The cross-species comparison broadens the significance of the data beyond the initial models studied. The atlas can serve as a foundation for future research in both basic and applied biology, including crop species where leaf shape influences light capture and productivity. By linking meristematic programs to complex leaf development, the results open pathways to explore whether similar redeployments occur in other plant systems, providing a theoretical basis for studies in plant and animal developmental systems more generally. Besides conceptual progress, the fellowship contributed important methodological advances. An optimized nuclei-based, spin-free isolation protocol was developed for small tissue inputs, enabling reproducible, high-quality multiomic data across species. This method, now presented at international conferences and nearing publication, addresses a major technical bottleneck in applying single-cell approaches to photosynthetic tissues, where plastid content often hampers data quality. Additionally, a whole-mount FISH pipeline was adopted for leaf primordia, allowing in situ validation of candidate regulators predicted from single-cell datasets. Although progress in genome-wide spatial transcriptomics slowed due to the discontinuation of commercial kits, contingency plans are in place to ensure that the spatial aspect of regulatory programs can still be studied. The integration of these approaches with MorphoGraphX is in progress and will eventually connect morphodynamic analyses with genome-wide gene expression data, with potential applications far beyond plant research. Overall, the project has delivered the first high-resolution multimodal developmental map of green tissues across multiple angiosperm lineages, introduced methodological innovations that will be adaptable to a wide range of plant systems, and prompted new discussions in the community about convergence and constraints in development and evolution. These outcomes closely align with, and in some cases surpass, the expected impacts outlined in the DoA by producing datasets, insights, and new methodological pipelines that were not available at the start of the project.
The cross-species comparison broadens the significance of the data beyond the initial models studied. The atlas can serve as a foundation for future research in both basic and applied biology, including crop species where leaf shape influences light capture and productivity. By linking meristematic programs to complex leaf development, the results open pathways to explore whether similar redeployments occur in other plant systems, providing a theoretical basis for studies in plant and animal developmental systems more generally. Besides conceptual progress, the fellowship contributed important methodological advances. An optimized nuclei-based, spin-free isolation protocol was developed for small tissue inputs, enabling reproducible, high-quality multiomic data across species. This method, now presented at international conferences and nearing publication, addresses a major technical bottleneck in applying single-cell approaches to photosynthetic tissues, where plastid content often hampers data quality. Additionally, a whole-mount FISH pipeline was adopted for leaf primordia, allowing in situ validation of candidate regulators predicted from single-cell datasets. Although progress in genome-wide spatial transcriptomics slowed due to the discontinuation of commercial kits, contingency plans are in place to ensure that the spatial aspect of regulatory programs can still be studied. The integration of these approaches with MorphoGraphX is in progress and will eventually connect morphodynamic analyses with genome-wide gene expression data, with potential applications far beyond plant research. Overall, the project has delivered the first high-resolution multimodal developmental map of green tissues across multiple angiosperm lineages, introduced methodological innovations that will be adaptable to a wide range of plant systems, and prompted new discussions in the community about convergence and constraints in development and evolution. These outcomes closely align with, and in some cases surpass, the expected impacts outlined in the DoA by producing datasets, insights, and new methodological pipelines that were not available at the start of the project.