Functional tissue units and their primary tissue motifs in multi-scale physiology

Abstract

Background: Histology information management relies on complex knowledge derived from morphological tissue analyses. These approaches have not significantly facilitated the general integration of tissue- and molecular-level knowledge across the board in support of a systematic classification of tissue function, as well as the coherent multi-scale study of physiology. Our work aims to support directly these integrative goals.

Results: We describe, for the first time, the precise biophysical and topological characteristics of functional units of tissue. Such a unit consists of a three-dimensional block of cells centred around a capillary, such that each cell in this block is within diffusion distance from any other cell in the same block. We refer to this block as a functional tissue unit. As a means of simplifying the knowledge representation of this unit, and rendering this knowledge more amenable to automated reasoning and classification, we developed a simple descriptor of its cellular content and anatomical location, which we refer to as a primary tissue motif. In particular, a primary motif captures the set of cellular participants of diffusion-mediated interactions brokered by secreted products to create a tissue-level molecular network.

Conclusions: Multi-organ communication, therefore, may be interpreted in terms of interactions between molecular networks housed by interconnected functional tissue units. By extension, a functional picture of an organ, or its tissue components, may be rationally assembled using a collection of these functional tissue units as building blocks. In our work, we outline the biophysical rationale for a rigorous definition of a unit of functional tissue organization, and demonstrate the application of primary motifs in tissue classification. In so doing, we acknowledge (i) the fundamental role of capillaries in directing and radically informing tissue architecture, as well as (ii) the importance of taking into full account the critical influence of neighbouring cellular environments when studying complex developmental and pathological phenomena.

Figure 1

Example workflow illustrating the acquisition and processing of FTU data from a three-dimensional reconstruction of human colon tissue. Step 1: The FTU template (A) is prepared according to the biophysical constraints under consideration, such that the long axis of the resulting cylindrical block of tissue is that of the feeding capillary (CAP) on which it is metabolically dependent. This template is applied to an appropriate volumetric region in the three-dimensional histology image dataset (B). The various cells within this region (coloured boxes) are typed and their position recorded (Note: red boxes represent endothelial cells, here shown lining the feeding capillary – CAP – and the erythrocytes within its lumen). Step 2: The cellular annotations across the full extent of the FTU cylindrical boundaries (C) are stored, together with the image data and the anatomical provenance of the tissue sample. Step 3: As the resulting primary tissue motif for the above colonic FTU uses standard reference ontology terms to represent both (i) a non-redundant list of distinct cell types, as well as (ii) the anatomical region of origin for the sourced tissue material, a terse graphical depiction of the constitution of this FTU may be automatically included in the context of whole-body anatomy maps, such as the one schematized by the ApiNATOMY tool [9] in (D). In this schematic, the outer boundary of the map represents the various epithelial surface categories (each individually coloured and labelled), and the inside tiles represent vascular (red) and neural (purple) structures respectively.

Figure 2

Step-by-step example illustrating the automation of primary tissue motif comparison. [A] FTU knowledge about 5 distinct tissues (in this particular example, derived from histology textbooks [16,17]) generated lists of distinct constituent cell types for each of the corresponding derivative primary tissue motifs. [B] Each distinct cell type in [A] was mapped to the equivalent term from the CellType ontology and assigned its unique term ID. [C] An all-vs-all pairwise comparison between the primary tissue motifs (ptm) was carried out as follows: (i) for every unique combination of ptm pairs (such that a pair consists of ptm_X and ptm_Y), an all-vs-all semantic similarity score c() for each unique combination of CellType term pairs is calculated (such that one CellType term is drawn from ptm_X and another from ptm_Y); (ii) the set p{} of highest scoring exclusive pairs of CellType terms is identified – exclusivity in a pair ensures that, once a CellType term from one ptm is selected to match another CellType term from another ptm, neither of these two CellType terms are included in any other pair in p{}; (iii) the sum of c() scores in p{} are divided by the average number of cell types across ptm_X and ptm_Y to generate s(ptm_X,ptm_Y). [D] The set of ptm elements is clustered over the pairwise score s(). See also Table  1 for concrete values of c(), p{} and s() involving the 5 distinct tissues referred to in [A].

Figure 3

Mock up of whole-body treemap schematic depicting a multi-organ endocrine pathway consiting of interlinked FTU-level molecular networks. Figure  1D is extended to depict a number of primary tissue motifs representing FTUs involved in the endocrine regulation of electrolyte and blood pressure levels during exercise. Every individual motif is labeled (in blue – see below for key to label abbreviations). In this mockup of an ApiNATOMY [9] schematic, nesting of one box within another represents the part_of relation such that: (i) tiles representing motifs are nested within tiles representing the anatomical region of origin for the tissue material from which the FTU was acquired, and (ii) tiles representing the constitutent cells of the motif are nested within the corresponding motif tile. The position of nodes in the treegraph overlaid onto the treemap depicts the location of substances (i.e. molecules or charged atoms), with respect to the motif constituents, as follows: a node within the boundary of a cell tile represents an intracellular substance; on the boundary of a cell tile represent a molecule tethered to the plasma membrane of that cell; outside all cell boundaries represents a substance located in the extracellular tissue fluid of the corresponding FTU. TISSUE MOTIF LABELS: [LGIN: Large Intestine; SMIN: Small Intestine; LIVR: Liver; STMC: Stomach; ADRC: Adrenal Cortices; HART: Heart; ARTL: Arterioles; BLOD: Blood; LUNG: Lungs; KDNY: Kidneys; ADRM: Adrenal Medullae; CORD: Spinal Cord; MEDL: Medulla Oblongata; PITR: Pituitary; SKLM: Skeletal Muscles; SWET: Sweat Glands]. EDGE COLOUR: [Black: Molecular Binding; Blue: Intercompartmental Translocation].

Figure 4

Annotation of the location of cell nuclei within a vascular diffusive field. Cellular nuclei within 40 μm of the centre of the vascular lumen (light blue shading) were identified and annotated onto three broad cellular categories, namely: 'endothelial’ (red dot), 'epithelial’ (green dot) and 'connective tissue’ (orange dot). The interface shown in this screenshot can be found at: [ http://aberlour.hgu.mrc.ac.uk/eAtlasViewer_demo/application/TPRDemo/wlz/colonRecon.php].