Multiphoton imaging in human liver tissues: validation of a new tool for drug discovery.

Chronic liver disease (CLD) is one of the world's leading causes of death. Rates of mortality due to CLD have steadily increased over the past 50 years and CLD is now the 5th most prevalent killer in the UK. There are over 100 known forms of liver disease, each of which requires a tailored course of treatment. Therapies are often unavailable due to delays in diagnosis or lack of insight into disease pathogenesis, and liver transplantation is often the only curative treatment. This is creating a huge burden on healthcare services for available and suitable donors.

To avoid the need for transplantation, a greater understanding is needed on how the immune system responds throughout the course of liver disease. Herein I propose to develop and validate new methodology for testing therapeutics in authentic CLD models derived from human tissue. Experiments in mice which receive putative treatments or genetic alterations to mimic human liver disease are a valuable tool for various stages of drug discovery. Immune responses can be monitored in vivo liver injury progresses; unfortunately, although we have access to representative mouse models of liver injury induced by toxicity and we can measure the induction and resolution of fibrosis in vivo, there are no representative models for immune-mediate liver damage. Autoimmune liver diseases and chronic inflammatory liver diseases lack representative mouse models of comparable inflammation. Additionally, available models do not always display most symptoms associated with human liver disease, in part because the mouse immune system differs significantly to that of humans. As such, a broader and more relevant strategy for drug-discovery is required.

Our laboratory has identified that the major cell type of the liver, the hepatocyte, is able to phagocytose and clear dead cells effectively (apoptotic and necrotic cells are abundant in liver injury and hepatocytes make up 80% of the liver cell composition). I showed that both mouse and human hepatocytes clear dead cells in a similar manner. Moreover, I pioneered experiments whereby perfused donor human liver tissue was treated with a drug which prevented hepatocytes from clearing dead cells in the same manner as it did in mice. My experiments demonstrated that it may prove more informative to use human liver tissue for the testing of therapeutics in place of mice to study resolution of liver injury with a readout of phagocytosis and dead cell clearance.

I propose to adapt the models that I developed using static microscopy techniques, to real time imaging using our new multiphoton microscopes (most advanced of their kind in Europe). Multiphoton microscopy allows high resolution imaging deep into tissues in real time and is performed in living anaesthetised mice. This fellowship aims to refine and validate the use of this technology for monitoring immune cells in samples of mouse and human liver explants in comparison to live imaging of mice.

Our team recently observed that hepatocytes can selectively delete live immune cells which normally dampen inflammation, known as T-regulatory cells (Treg). This cell subtype has conserved properties in mice and men. We also have a compound that perturbs Treg capture by hepatocytes, and I intend to apply the same principles as for the aforementioned dead cell clearance assays developed in my PhD to adapt the human liver imaging technologies to live measurements by multiphoton microscopy. These experiments will reveal an unappreciated role for hepatocytes in immune regulation in mice and men. I have letters from translational researchers stating that successful comparison of T cell-hepatocyte interactions in perfused mouse and human tissue compared to mouse intravital experiments would encourage them to reduce their use of mice and increase the use of human liver tissue in basic biology and drug discovery research.

Chronic liver disease affects over 300 million people worldwide and transplantation is often the only available cure. Mouse models of liver disease are moderate/severe and often fail to yield clinically translational data. This is, in part, due to the variances in immune cell function between mouse and human. Imaging mice via intravital microscopy has allowed for real-time in vivo insights into liver immune cell function. However, these are severe procedures and still present difficulty with clinical application. There is a pressing need for translational models of liver disease.

During my PhD, I studied how hepatocytes engulf dead cells. This process (efferocytosis) caused hepatocytes to fail cell division in vitro and become multinucleate. I also observed that in mice with ischemia reperfusion injury, hepatocytes next to necrotic legions had more nuclei. To prove that efferocytosis caused this, we set up a model whereby donor human livers were perfused with an inhibitor of efferocytosis and then injured, which prevented multinucleation of hepatocytes near the injury. This proved that perfused human tissue could be used to replicate murine in vivo observations. Our group has also shown that regulatory T cells (Treg) were captured alive by and deleted by hepatocytes. As Tregs have shown potential for dampening liver disease, halting their deletion by hepatocytes may prove of benefit.

This proposal aims to improve the real-time imaging of liver immune cells while adhering to the 3Rs of animal testing. We aim to refine the imaging of mT/mG mouse livers by comparing in vivo intravital microscopy to ex vivo multiphoton imaging of livers for Treg-hepatocyte interactions in the presence of inhibitor C1. We shall then establish the first use of multiphoton microscopy for imaging perfused human liver and testing drug efficacy. This model will encourage others to test compounds on human liver disease tissue, yielding more clinically relevant results.

Over 3000 articles were published in 2016 involving the use of mouse models for studies in liver disease. Typically, ~5 animals were used per experimental group, which is often multiplied by the number of liver injury models tested per investigation and the time points required. Accounting for positive, negative, and vehicle controls, each experiment could warrant the use of ~25 mice on average. Current murine models of liver disease are poor representations of human pathology; mice often do not display symptoms associated with liver disease, such as weight gain or steatosis. In addition, less-well understood ailments, including autoimmune diseases such as primary biliary cholangitis (PBC) are currently unrepresented by mouse models. Our multiphoton model of imaging ex vivo human tissue will be applicable to all explant livers we receive from the QE hospital, regardless of aetiology. This will allow for the immunological study and therapeutic screening for many diseases. Industries with access to intravital microscopes will be able to use our model for their disease of interest.

Intravital microscopy, although providing an important insight to liver immunology, presents multiple risks to the wellbeing of animals. Mice must be anaesthetised throughout the entire imaging period, which is difficult to ensure and can lead to suffering. Should the results gained from ex vivo imaging of perfused mouse livers prove comparable to intravital, this would remove the need for anaesthesia for many experiments involving tissue studies that do not require recirculating lymphocytes from peripheral blood. Overall, this would lead to a refinement of live tissue imaging, which could be employed by groups interested in liver immunology.

Testing compounds or assessing the immunological interactions in human liver at early stages of an investigation would deliver powerful refinements in the planning of animal experiments. One example is to use ex vivo human tissue to select the most effective treatments. A wider variety of compounds can be tested with ex vivo systems, as ex vivo livers can be divided between multiple experimental conditions. Unsuccessful treatments can be excluded, reducing the number of required animals usually by 5 replicates each time. As such, mouse experiments can be simplified to a single control and experimental group. Groups using liver injury models have also stated that opting to use ex vivo human tissue could reduce animal usage by up to 60%. Pharmaceutical industries including Medivir and AstraZeneca, add that initial testing of compounds on human material could fast track drug discovery and save time and funds when validating drugs in vivo.

Understanding of immune cell function in human livers would allow the selection of appropriate mouse models; immune responses vary between mouse strains and inflammatory activators. This will replace the need for testing different mouse models for experimentation. Using ex vivo human tissue would also allow the selection of appropriate time points for administering drugs. Many studies often require multiple time points to monitor drug efficacy. Initial testing on human tissue would help inform experiment design in animal experiments.

Finally, using human tissue first, enables the selection of the measurable variable, which yields the largest experimental difference from the control. This allows a larger effect size to be applied to power analyses, which will reduce the sample size of animals needed to determine statistical significance. Overall, the requirement for murine models will be reduced due to the selection of effective treatments prior to animal testing and better informed experimental planning; scientists will be able to use published results achieved using our human tissue model to assist their planning of investigations. Please see attached letters of support from stakeholders on how my new methodology will reduce their use of mice in their studies.