Quantification of the Whole Lymph Node Vasculature Based on Tomography of the Vessel Corrosion Casts

Abstract

Lymph nodes (LN) are crucial for immune function, and comprise an important interface between the blood and lymphatic systems. Blood vessels (BV) in LN are highly specialized, featuring high endothelial venules across which most of the resident lymphocytes crossed. Previous measurements of overall lymph and BV flow rates demonstrated that fluid also crosses BV walls, and that this is important for immune function. However, the spatial distribution of the BV in LN has not been quantified to the degree necessary to analyse the distribution of transmural fluid movement. In this study, we seek to quantify the spatial localization of LNBV, and to predict fluid movement across BV walls. MicroCT imaging of murine popliteal LN showed that capillaries were responsible for approximately 75% of the BV wall surface area, and that this was mostly distributed around the periphery of the node. We then modelled blood flow through the BV to obtain spatially resolved hydrostatic pressures, which were then combined with Starling's law to predict transmural flow. Much of the total 10 nL/min transmural flow (under normal conditions) was concentrated in the periphery, corresponding closely with surface area distribution. These results provide important insights into the inner workings of LN, and provide a basis for further exploration of the role of LN flow patterns in normal and pathological functions.

Conflict of interest statement

The authors declare no competing interests.

Figures

Figure 1

Workflow of vascular casting, tomography and image processing. The blood in LN vasculature was flushed out and replaced with Mercox II resin to cast the lumen of the vasculature. The LN was then surgically excised and placed in a pipette tip before dissolving the tissue with potassium hydroxide (1). The freeze-dried samples were scanned with high-resolution synchrotron tomography and the radiographs were reconstructed into stack images using a phase-retrieval algorithm (2). The images were pre-processed by removing the pipette tip image using a cone crop before intensity-based segmentation and manual artefact processing (3). The binary data were then skeletonized and diameters and length of the vessels as well as the surface area density were calculated. Pressures and velocities of blood flow were estimated in each vessel based on an assumption of Poiseuille flow (4). The results were visualized with Imaris and further parameters quantified and plotted with Matlab (5).

Figure 2

High-resolution 3D images of four mouse LNs. 3D representation of four LN datasets shows the quality of vascular casts. The degree by which the capillaries are preserved varies between the cases. LN1-LN4 (A–D) are referenced in the rest of the manuscript. Scale bars: 200 µm (A–D).

Figure 3

Analysis of vessel length and diameter distribution in the vascular networks. Based on the 3D segmented images of the LN vasculature (A) the binary images were skeletonized and the length and diameter of the vessels were calculated. The vasculature was then reconstructed (B) with colors corresponding to each diameter category (C). Skeletonization was used to quantify vessel characteristics such as diameter (D) and length (E). Distributions are shown for the node shown in (B) in histograms with bins size of 1 µm for diameter (C) and 5 µm for length (D). Finally, we have performed the analysis for the other three dataset and the distribution of diameters (G) and lengths (H) show similar expected trends between the datasets. Scale bar: 200 µm (A).

Figure 4

Analysis of the surface area distribution in the vascular networks. The datasets were divided in 80 by 80 by 80 pixel boxes and surface area was calculated within each box and normalized surface area is shown for all four LNs (A–D). The heatmaps represent the mid plane for all the LNs. Vessels appeared more densely distributed in the periphery of the LN.

Figure 5

Distribution of blood pressure and velocities in the vascular network. The blood pressure in the LN vessels was skewed towards the vein pressure represented by LN being mostly blue in panel A. The distribution showed distinct sets of vessels with a peak at high pressures (~25 mmHg) that was associated with arterioles, vessels with pressures near LN vein pressure (assumed to be 10 mmHg) that was associated with venules and especially high-endothelial venules, and capillaries in between these two sets of vessels. Distribution of velocities in the LN vessels was close to a log normal distribution (C), with higher velocity in larger vessels such as arterioles and large venules. The distribution of mean vessel diameter, velocity and pressure are shown for different subsets of vessels separated into three categories of arterioles (red), venules (blue) an capillaries (green).

Figure 6

Distribution of parameters important in blood lymph fluid exchange based on the Starling equation. Distribution of the blood pressure ((A) mid section, and (B) whole LN) shows that the blood pressure in most of regions of the LN is close to the LN venous pressure (assumed to be 10 mmHg). Distribution of surface area ((C) mid section, and (D) whole LN) appears exponential. Distribution of fluid flow from LN to BV that linearly correlates with pressure multiplied by surface area ((E) mid section, and (F) whole LN) shows a non-uniform distribution with higher values in the periphery of the LN. The contours in the top panels (A, C and E) are for a single slice in the middle of the LN, and the histograms in the bottom panels (B, D, and F) are for the whole LN. The LN was divided into 100 by 100 by 100 pixel boxes (81 × 81 × 81 µm3) for these analyses.

Figure 7

Effect of LN pressure on blood lymph fluid exchange flow.Mean pressure in lymphatic channels of LN linearly affects the amount of fluid exchange from LN to BV and changes the distribution and pattern of exchange flow. At low Plymph (A) there are fluid exchange in both directions based on the location in the LN, and increasing Plymph shifts the fluid transport in favor of LN to BV (B-D). The contours in the top panels are for a single slice in the middle of the LN, and the histograms in the bottom panels are the results for the whole LN. The LN was divided into 100 by 100 by 100 pixel boxes (81 × 81 × 81 µm3) for these analyses.