Novel insights into the fenestrated scavenger endothelium of the liver sinusoid

Abstract

<p>The sinusoids (specialized small blood vessels) of the liver are covered by endothelium (blood vessel wall cells) with open transcellular pores (holes that go from one side to the other) called fenestrations. This allows for efficient bidirectional transfer of solutes between the blood and the hepatocytes (main metabolic liver cell). These fenestrations can disappear or reduce in number and size in disease states or in ageing. We therefore sought to map the literature on compounds, that affect these fenestrations, and to hypothesize how the mechanism regulating them operates. <p>The fenestrations are unevenly distributed along the sinusoid, with there being a greater fraction of the cell surfaces covered by these pores towards the end (the pericentral area) compared with the start of the vessel (the periportal area). There are also lymphatic vessels in the periportal area, in a space behind the portal vein and hepatic artery, which is often omitted from consideration in anatomical illustrations and flow models of the liver. We therefore sought to make a digital model at the single sinusoid level, including these ultrastructural details, to assess their influence on fluid flow parameters. <p>The liver endothelium is a scavenging endothelium, that is to say high capacity waste removal cells specialized in macromolecular and nanoparticle sized waste from the blood stream. Albumin is the single most abundant protein in blood, with chemically modified forms of it being found in several pathologies, especially diabetes or liver disease. It was found by a Japanese research group, Iwao et al., that when albumin is highly oxidized, it is rapidly removed from the blood stream, mainly by the liver. The properties of the liver sinusoidal endothelium as a scavenger endothelium, and the clearance kinetics led us to believe this was done by the liver sinusoidal endothelium and its stabilin receptors, because of the functions of these in respect to other modified albumins. We indeed found that this was the case. <p>The analysis of fenestrations from microscopy images is a laborious process, and contains the possibility of introducing user bias into quantifications. We assessed three different methods of image analysis for the purpose of quantifying fenestration parameters. These were manual, semi-automated/thresholding based, and fully automated/neural network based approaches. The manual classification method had little bias with regards to number, whilst showing significant user bias for diameter/size of fenestration. The semi-automated was the least biased with regard to diameter/size, but significantly biased with regards to number. The fully automated also showed considerable user bias for all parameters, however it can be used for batch processing. The methods are roughly ordered by speed (manual, semi-automated, fully automated), with regards to larger data sets.