top of page

A spatial human thymus cell atlas mapped to a continuous tissue axis

Yayon*, Kedlian*, Boehme* et al. (BioRxiv) DOI: 10.1101/2023.10.25.562925

A spatial human thymus cell atlas mapped to a continuous tissue axis


  • Human Thymus Development

  • Spatial Transcriptomics

  • Multi-omics Data integration

  • Tissue Biology

  • Developmental Biology

Main Findings

The thymus is a central organ for the formation and education of developing T cells. Common lymphoid progenitors enter into the thymus and start to commit towards the T cell lineage. Through a sequence of well-coordinated selected processes (i.e. b-selection, positive and negative selection) the thymus ensures that T cells can recognise self-MHC molecules, prevent self-antigen recognition and promote specific lineage commitment of T cells towards a CD4 and CD8 T cell lineage. These commitment processes occur within the context of highly defined tissue environment with compartmentalised cell types ensuring the different developmental steps of thymocytes. Already at 12 weeks’ post-conception, the thymus starts developing and producing mature T cells, raising the question how the fetal thymus differences from the thymus observed in infants and children.

This preprint investigates the spatial organisation of developing thymus from fetal to paediatric stages. The authors propose a common coordinate framework (CCF) to homogenise structural differences by applying a continuous gradient across the tissue that reflects the relative localisation in different thymic areas/zones. This CCF could serve as a reference for future investigations into the spatial organisation of this organ and help to explore thymocyte maturation in a continuous space.

The authors built a spatial annotation describing the relative position of a spot based on the weighted distance to defined “boundaries” (capsule-to-cortex, and cortex-to-medulla). This way, the authors can account for the complex structural variability of the lobules in the thymus and define a relative location within a bigger area i.e. a “deeper” or a more “shallow” part of the medulla. The authors further use this continuous cortio-medullary axis (CMA) to segment the lobule into different layers to better annotate transcriptional information gained from spatial transcriptomics, predominantly 10x Visium datasets of the fetal and paediatric thymus. Moreover, the authors combine this high-resolution annotation with published and new single-cell RNA-seq, as well as CITE and VDJ-seq to infer space and time trajectories of different cell populations within the thymus microenvironments.

The authors make several interesting observations:

  1. (1)  The general trajectory of T cell development is conserved across the fetal and paediatric thymus as well as its spatial distribution across the tissue organisation with mature T cells positioned closer to the medulla vs. DN and DP thymocytes more abundant in the cortex.

  2. (2)  The spatial distribution of cytokine expression across the developmental stages showed a similar compartmentalisation, although there were cytokines with altered expression levels especially in the fetal thymus (e.g. IL-34, IL1R, CX3CL1, IL-33, SPP1).

  3. (3)  The authors found that cortical thymic epithelial cells (cTEC) while transcriptionally similar showed different sublocalisation in both the fetal and paediatric thymus. Similarly, medullary thymic epithelial cells (mTEC) are distributed along the medullary depth. Additionally, mTECs that have characteristics of “mimetic cells” have the highest abundance in deep medullary tissues. Interestingly, cells that were identified as mcTEC progenitors seem to have a different localisation in the fetal and paediatric thymus, with a more subcapusular localisation in fetal and a more cortex-medullary junction position in the paediatric thymus.

  4. (4)  The Hassall’s corpuscle associates with differentiated mTECs (“mimetic” like cells) that express a set of genes including keratin-related genes.

  5. (5)  Lastly, the authors visualise that CD8 committed T cells stay longer in the cortex compared to CD4 committed T cells which egress earlier into the medulla. CD8 committed T cells only exit the cortex at a (semi)mature stage. They attribute this to the differential dynamic use of chemokine receptors (which they validate on both transcriptomic and protein level). The authors attribute the earlier exit of CD4 committed T cells into the medulla by expression of CCR4, while CD8 committed T cells may use CXCR3 to egress from the cortex.

Overall, this is a powerful study integrating single-cell RNA-seq and spatial transcriptomics data to give an in-depth tissue localisation  of given cell types across thymic developmental stages. Overall, the data are in line with previously published information and has an important confirmatory impact on our understanding of thymic development and functions. Importantly, the authors seem to have found a powerful way to not only integrate multiple large datasets across platforms but to align interindividual spatial differences by calculating a relative position within the lobule using key macroscopic structures.


  • Perhaps the strongest limitation of the study is that it is (by nature) fairly descriptive and will require some more experimental studies to understand some of the impact that spatial differences identified in this study may have for thymocyte development.

  • Due to the latter comment, it would have been nice if the author could include a bit more information how their spatial findings could relate to thymic development in mice, especially since mouse models are the predominant model to study thymic development more mechanistically. While we don’t think this is a must have of this study, it would certainly add to the paper and enable further direct human-mouse translational comparisons.

  • While this reviewer does not dare to judge the in-depth modelling and bioinformatic analysis of the study, one did wonder if the model would allow for a similar analysis pipeline in a diseased or older thymus where the organisational structure is more perturbed.

  • Generally, it would be good if the authors would extend or precise their observations a little more and hypothesize what some of the changes could mean (e.g. why the differences in spatial cytokine expression between the fetal and paediatric thymus?, what could be the role of specifically more differentiated mTECIII populations?)

  • Finally, given the studies aim to provide a in-depth cellular and spatial atlas of the thymus, it is essential for the patient cohort data, encompassing both previously published and newly generated datasets, to include detailed information on the pathologies necessitating cardiocorrective surgery and any prior treatment regimens. This inclusion is crucial to ensure that the atlas reflects a representation of the thymus that is as close to a healthy state as possible.


The preprint advances our understanding of the spatial composition and organisation of the thymus across its fetal to paediatric development. While the biological novelty may be to somewhat restricted, the study has an important confirmatory impact. It certainly will be a valuable resource for future studies focussing on the human thymus. It will allow for a much more spatial insight into its tissue context.

The thymus could here serve perhaps also as an example of how tissues can be spatially analysed across time point and between individuals in a harmonised fashion. While the significance of our understanding of thymic development cannot be underestimated, it would have been helpful if connections to mouse models, in terms of mouse thymic development could be made using their system. Predominantly because it would enable more mechanistic studies in some of the identified cell populations and their roles within thymic development and spatial tissue contexts.


Reviewed by Dörte Symmank and Felix Clemens Richter as part of a cross-institutional journal club between the Vanderbilt University Medical Center (VUMC), the Max-Delbrück Center Berlin, the Medical University of Vienna and other life science institutes in Vienna. We thank Christoph Bock for his Input on this Review. 

The authors declare no conflict of interests in relation to their involvement in the review.

bottom of page