Precision-cut tissue slices: a novel ex vivo model for fibrosis research

Pathological scar formation, i.e. fibrosis, is characterized by a disproportionate production and deposition of extracellular matrix proteins in tissues resulting in loss of organ function. Fibrotic diseases account for up to 45% of worldwide mortality, yet there are no effective antifibrotic therapies currently available. To improve and accelerate antifibrotic drug discovery, there is an urgent need for reliable and reproducible (human) in vitro methods that reflect the cellular diversity that epitomize specific organs. This thesis delineates the successful development of a novel ex vivo/in vitro model for intestinal and renal fibrosis, namely precision-cut intestinal slices (PCIS) and precision-cut kidney slices (PCKS) prepared from murine, rat and human tissue. Our results demonstrated that the slices remain viable during culture and maintain their organ-specific phenotype. Moreover, in both PCIS and PCKS we observed that fibrosis could be induced by either culture activation or treatment with a profibrotic stimulus. Based on these findings, both models were subsequently used to evaluate the efficacy of various putative antifibrotic drugs. Using PCIS, we demonstrated that pirfenidone, LY2109761 and sunitinib could mitigate fibrogenesis on a gene level, warranting further evaluation of these compounds for the treatment of intestinal fibrosis. In addition, we demonstrated that IFNγ could be used to halt renal fibrogenesis. Furthermore, studies with precision-cut liver and intestinal slices revealed that rosmarinic acid elicited organ- and species-specific effects, illustrating the pressing need for good translational models for drug discovery. Taken together, this thesis delineates that precision-cut tissue slices can be used to unravel fibrosis and evaluate the antifibrotic potential of therapeutics.


INTRODUCTION
Fibrosis is characterized by the excessive deposition of extracellular matrix (ECM) proteins, including collagen and this pathological process is the result of chronic inflammation or injury. Fibrogenesis is a physiological phenomenon serving as a part of the physiological repair processes after injury or inflammation. During the repair process, ECM is formed by myofibroblasts to temporarily replace the injured tissue. Afterwards, the inflammatory response will be inhibited, ECM will be degraded and the myofibroblasts will undergo apoptosis. During chronic injury, this repair process can be deregulated, which leads to abnormal accumulation of ECM and culminates into fibrosis (Guarino et al., 2009). Alterations of the original tissue architecture of an organ due to high collagen levels will lead to stiffness and loss of functional cells, resulting in impaired function of the organ (Ghosh and Vaughan, 2012; Wynn, 2008).
Fibrosis has been described in almost every organ, and the fibrotic process shares common mechanisms in different organs (Ghosh and Vaughan, 2012). Inflammation and injury, being essential steps in tissue fibrosis, can trigger profibrotic signaling via activation of cytokines and differentiation of resident fibroblast, endothelial cells, epithelial cells or stellate cells into myofibroblasts, which are responsible for the deposition of ECM (Ghosh and Vaughan, 2012), an indication of the typical multicellular character of the pathology. Transforming growth factor β (TGFβ), produced by several tissue-resident and blood-derived cells, is crucial in the development of fibrosis, through its influence on the production of ECM, matrix metalloproteinases (MMP) -enzymes that degrade ECM -tissue inhibitor of metalloproteinases (TIMP) and on T cell function (Hold et al., 2009). Also platelet-derived growth factor (PDGF), produced by macrophages, stellate cells and mesangial cells, is an important factor in the development of organ fibrosis, it plays a vital role in proliferation, migration and survival of myofibroblasts (Bonner, 2004;Trojanowska, 2008). Hence, the levels of PDGF are increased in fibrotic tissue (Trojanowska, 2008).
This thesis focuses on intestinal and renal fibrosis, since the pathophysiological mechanism underlying both pathologies is not fully elucidated. The various chapters delineate the development of a new ex vivo model for both intestinal and kidney fibrosis, which is a great tool in unraveling the mechanism of fibrogenesis and can be utilized in evaluating the efficacy of potential antifibrotic drugs. This chapter provides a timely overview of the current knowledge regarding intestinal and kidney fibrosis as well as the in vitro and in vivo models that are used to study fibrosis in these organs.

Intestinal fibrosis
Intestinal fibrosis (IF) occurs in patients after radiation therapy or as a reaction of intestinal tissue to chronic inflammation. Moreover, IF is a major complication in inflammatory bowel disease (IBD), represented by ulcerative colitis (UC) and Crohn's disease (CD) as the two major types of IBD. Fibrosis is an interaction between the innate immune system and multiple cellular sources of excessive ECM deposition (Fiocchi and Lund, 2011). However, in the intestine, internal factors such as gut microbiota, external and environmental factors also play important roles (Lawrance et al., 2001). Like in other organs, TGFβ is important in the early onset of fibrosis in the intestine, and is in the inflamed areas in the intestine of patients with CD (di Mola et al., 1999). Moreover, TGFβ induced a dose-dependent expression of type I collagen and heat shock protein 47 (HSP47) in intestinal fibroblasts (Kitamura et al., 2011;Mulsow et al., 2005). One of the main mysteries in IF is the source of the mesenchymal cells such as fibroblasts and myofibroblasts. A variety of hypotheses exist about the transition and activation of mesenchymal cells into fibroblasts and myofibroblasts during the onset of fibrosis (Flier et al., 2010;Powell et al., 2011;Rieder et al., 2011b). However, none of these hypotheses have been proven, mainly because of the shortage of relevant IF models.
The natural history of IBD is highly heterogeneous and frequently complicated by IF and stricture formation. This appears to be the case for both entities of IBD, UC and CD (Cosnes et al., 2002; Gumaste et al., 1992). More than 30% of CD and about 5% of UC patients develop a distinct fibrostenosing phenotype with progressive narrowing and potential for intestinal obstruction (Gumaste et al., 1992;Louis et al., 2003). Intestinal stricture formation is a frequent indication for surgery in CD (Farmer et al., 1985) and strictures frequently recur leading to repeated surgeries (Fazio et al., 1993). It appears that both entities of IBD and in particular CD exhibit a progressive nature with changes in disease behavior throughout the disease course. In CD, which is a transmural disease, chronic mucosal inflammation induces remodeling of the entire intestinal wall. This process is a cascade of events that includes epithelial cell and intestinal damage and repair, angiogenesis and lymphangiogenesis and activation of immune cells and mesenchymal cells (MCs). MCs include fibroblasts, myofibroblasts and smooth muscle cells and are the major source of ECM components (Rieder, 2013).
It is difficult to predict which patients will develop a fibrostenosing phenotype (though a majority will do so eventually) and how rapidly they will progress. No specific therapy to prevent or treat IF is known. Therefore, it is of utmost importance to develop (in vitro) human and animal models to study the mechanism of IF, which could lead to the discovery of potential targets for therapeutic compounds.

Renal fibrosis
Almost all progressive chronic kidney diseases eventually develop into renal fibrosis (Boor et al., 2010;Liu, 2011). Renal fibrogenesis is considered as an unsuccessful wound healing process, in which almost all the different cell types in the kidney and also infiltrating lymphocytes, macrophages and fibrocytes are involved (Liu, 2011). After injury, inflammatory cells infiltrate, fibroblasts are activated and ECM is produced (Liu, 2011). The myofibroblast is the key cell type in producing ECM components, however, mature fibroblasts, tubular epithelial cells, macrophages and fibrocytes are also a source of ECM in the kidney (Boor et al., 2010). In fibrosis, the tubular epithelial cells, released from the basement membrane, can undergo epithelial mesenchymal transition, forming fibroblasts (Zeisberg and Neilson, 2010). The interstitial spaces are then filled with matrix components, mostly collagen I, III and fibronectin (Zeisberg and Neilson, 2010). The two main families of proteases involved in matrix degradation in the kidney are MMPs and members of the plasmin-dependent pathway. These proteases not only fragment the ECM but also cleave non-matrix substrates, thereby producing profibrotic growth factors (Zeisberg and Neilson, 2010). Eventually all these cellular and molecular events will lead to renal failure.
To develop therapies for renal fibrosis, detailed knowledge of the disease mechanism is essential. Therefore, adequate in vivo, but also in vitro/ex vivo models, like kidney slices, are necessary.

Markers of intestinal fibrosis in IBD
To enable progress in understanding the mechanism of IF, it is essential to identify markers of IF, in order to (1) stratify patients into different levels of risk before the development of fibrosis, and (2) detect early stages of fibrosis before clinical symptoms have occurred. An optimal fibrosis marker should detect early stages of fibrosis, identify trajectory of fibrosis development, be predictive of future fibrosis, be predictive of and responsive to the effect of antifibrotic therapies and be predictive of non-responsiveness to anti-inflammatory therapies. Accomplishing these goals will open the door for targeted antifibrotic therapy, and the ability to test candidate antifibrotic therapies in clinical trials.

Currently available markers of IF
No specific and accurate predictors or diagnostic tools for IF exist and to date no marker of fibrosis is in routine clinical use. Several targets have been tested for this purpose (Table 1).
Genetic signatures are attractive as they are stable, present long before the disease onset and are not affected by alterations in the disease course. Several genes have been evaluated for their association with fibrostenosing CD. Alternations in the nucleotide-binding oligomerization domain containing 2 (NOD2) gene, the first discovered and best explored genetic variant, are weakly associated not only with CD fibrostenosis, but also with ileal disease location and fistulizing disease (Adler et al., 2011) and hence lack specificity. Other genetic variants have been described as being linked to fibrostenosis, such as those in the matrix metalloproteinase (MMP)-3 gene (Meijer et al., 2007) or in the rs1363670 locus near the interleukin (IL)12B gene . Interestingly, an increasing number of risk alleles, including NOD2, IBD5, diskslarge homolog (DLG)5, autophagy-related protein 16-1 (ATG16L1), and IL23 receptor (IL23R), confer an increasing risk for intestinal resections in CD (Weersma et al., 2009). Gene variants are promising markers, but their population frequency is low and they exhibit incomplete penetrance. The major benefit of genetic biomarkers is their stability over time and independence from environmental factors, though epigenetic modifications at the sites of fibrogenesis might prove important as well.  The most widely used criteria to predict fibrostenosing CD are clinical factors. These include the need for corticosteroids, early disease onset, perianal fistulizing disease or small bowel disease location (Beaugerie et al., 2006;Louis et al., 2003). These factors however are encompassing many different disease phenotypes. On the other hand, the Montreal classification (Silverberg et al., 2005) merely identifies fibrosis after it has become clinically apparent and can only be used as a descriptor rather than a predictor. Thus, alternative, noninvasive predictive tools are required. One predictive tool that might be of use as a biomarker is a panel of serologic markers. 15 Circulating antibodies against microbial products are found in some patients with IBD, such as anti-Saccharomyces cerevisiae (ASCA) among others. These are believed to arise from aberrant immune responses towards the luminal microbiota (Dotan et al., 2006). These antibodies are qualitatively and quantitatively associated with, and predictive of a more complicated disease phenotype, including fibrostenosis (Dubinsky et al., 2006; Rieder et al., 2011a). However, they are not specific for this phenotype, but rather predict complicated CD, including fistulizing disease and the need for surgery. ECM molecules and growth factors, such as laminin, collagens, collagen propeptides or telopeptides (Kjeldsen et al., 1995;Koutroubakis et al., 2003;Loeschke and Kaltenthaler, 1989), basement membrane components or fibronectin (Allan et al., 1989), YKL-40 (also known as human cartilage glycoprotein, a chitinase-like protein), basic fibroblast growth factor (bFGF) and others have been investigated as biomarkers of fibrosis, with inconclusive or negative results. No existing marker of fibrosis showed specific promise in IBD. Evaluating the markers in fibrosis during renal fibrosis, could also be relevant for further study in IBD

Markers of renal fibrosis
While the inciting agents in renal fibrosis are often different from the liver, lung, and skin, the fibrotic process appears to be shared. While lab markers of renal function can inform about progression of renal disease they do not necessarily reflect fibrotic burden and they can be influenced by a wide variety of factors. Unique to the kidney is the accessibility of urine as a direct, renal-specific read-out for renal fibrosis, allowing the identification of local markers of fibrosis. Urine levels of TGFβ1, CTGF and collagen IV increase with progression of chronic kidney disease (CKD) (Cheng et al., 2006;Gilbert et al., 2003;Io et al., 2004;Tsakas and Goumenos, 2006). Urine plasminogen activator inhibitor-1 (PAI-1) has also been shown to correlate with renal fibrosis in patients with diabetic nephropathy (Torii et al., 2004).
Ultrasound imaging has also entered the field of renal fibrosis. Using the doppler ultrasound technique to calculate the 'resistive index' and 'atrophic index' has shown promise in predicting time to dialysis and survival in chronic kidney disease (Sugiura and Wada, 2009).
One of the challenges of universal fibrosis markers is that they may be nonspecific, so that a patient with CD may appear to have elevated serum markers of IF when they have fibrosis in a different organ (skin, liver, lung, kidney, etc.). There may be benefit in obtaining markers of IF from the gut (biopsies) or its output (stool) to increase the specificity of markers for IF.
In summary, we can share the knowledge of fibrotic diseases from various organ e.g., kidney, liver, lung, intestine and skin. Multiple potential biologic and imaging markers are already in clinical use. While organ-specific markers, such as creatinine for renal disease or alveolar epithelial cell specific proteins in the lung, likely will not be helpful markers of IF, multiple markers with a direct link with fibrogenesis have been identified. As fibrotic mechanisms are shared across organs, these provide possible candidates for use in the intestine as well ( Table 2). These markers are indispensable for (novel) models in intestinal and renal fibrosis.

NOVEL MODELS FOR INTESTINAL AND RENAL FIBROSIS
Numerous in vitro and animal models are available to study fibrosis. Cell culture models using fibroblast or myofibroblast cell lines or freshly isolated primary cells, are used to study cell behavior during exposure to fibrotic and antifibrotic compounds, but the interplay of the different cell types involved in organ fibrosis can not be investigated, unless cocultures are used (Iredale, 2007). Furthermore, cell lines can differ from primary cells (Wilkening et al., 2003) and freshly isolated cells can react differently compared with cells in the tissue, because the interplay with other cells and matrix is absent (Snowdon and Fallowfield, 2011). In experimental animal models, fibrosis can be induced by induction of chronic injury or infection, and the detailed cellular and molecular pathways of fibrosis can be studied. Recently, precision-cut tissue slices (PCTS) have shown their use in the study of fibrosis, since the multicellular process of fibrosis can be mimicked in tissue slices and they allow the study of fibrosis in human organs by using human tissue.
In this section the use of PCTS from the kidney and intestine in fibrosis research is evaluated and in particular, the utilization of PCTS in the study of the mechanisms of fibrosis and the effect of antifibrotic drugs is discussed. Most of the techniques to induce fibrosis in PCTS are based on in vivo models or techniques that have been used in fibrosis research to induce fibrosis. Therefore, these in vivo models are discussed briefly in relation to the possibilities and utilization of PCTS in fibrosis research.

In vivo models for intestinal fibrosis
Most of the IF animal models available are based on intestinal inflammation typically induced by toxins to mimic CD and UC. In addition, different models based on transgenic animals and gene delivery systems have been developed. To induce IF, 2,4,6-trinitrobenzene sulfonic acid (TNBS) is administered intrarectally, which results in transmural inflammation and ECM deposition.

Early onset of fibrosis and PCIS
We recently established that PCIS can be used as a model to study the early onset of fibrosis in the intestine (Chapter 3). PCIS from rat and human small intestine remain viable up to 24 and 48 h, respectively. After incubation for 24 and 48 h, the gene expression of HSP47 was increased in human and rat PCIS  which is in line with the observation of Honzawa et al. who reported that serum level of HSP47 protein, which is required in the synthesis of normal collagen (Taguchi and Razzaque, 2007), is higher in CD patients than in UC patients (Honzawa et al., 2010). When rat PCIS were cultured in the presence of TGFβ1, the gene expression of procollagen 1A1 and αSMA was up-regulated compared with control slices . These results suggest that the early onset of IF can be mimicked ex vivo and that the model is a new tool to unravel the complicated mechanisms of IF.
However, more (human) studies are necessary to validate this ex vivo model of fibrosis. It would be interesting to investigate whether the TNBS or the DSS model can be applied in vitro to induce the early onset of fibrosis in vitro. In addition, the use of animal or human intestine with established IF can be used to prepare fibrotic PCIS to study the mechanism of end-stage fibrosis and to screen potential antifibrotic drugs (Chapter 4 and 5 of this thesis).

In vivo models for renal fibrosis
There are several models that can be used to study renal fibrosis characterized by increased synthesis and accumulation of ECM (Zeisberg et al., 2005). Fibrosis can be induced by exposure of animals to (toxic) compounds. The chemical N-(3,5dichlorophenyl)-succinimide can be used to provoke tubular interstitial nephritis, which can serve as a model for interstitial renal fibrosis (Barrett et al., 1983). In the renal cortex of rats exposed to this chemical, an increased activity of proline hydroxylase, elevated hydroxyproline content and increased water content was found. All these phenomena are associated with chronic renal fibrosis (Barrett et al., 1983).
Exposure of experimental animals to ochratoxin-A (OA), a mycotoxin found in grain, resulted in renal disease, comparable to endemic nephropathy. Chronic interstitial renal pathology correlated to exposure to OA in grains and animal products (Aukema et al., 2004). Four weeks of exposure to OA in piglets resulted in increased collagen formation and fibroblast proliferation, leading to renal fibrosis (Aukema et al., 2004). The genetically modified mouse strain COL4A3 − / − mouse, is used as an animal model for the Alport syndrome, which is a type IV collagen disease that leads to hematuria, proteinuria, renal fibrosis and ultimately to renal failure (Gross et al., 2010). The strengths and weaknesses of animal models used in renal fibrosis research were summarized by Meng et al. (Meng et al., 2014). Although, these models are widely used, most of them are limited to some specific species and irrelevant for the human disease. The above mentioned models for renal fibrosis might be translated into the renal slices model, either to induce fibrosis in slices or to use slices of (fibrotic) renal tissue to test antifibrotic drugs (Chapter 6).

Precision-cut kidney slices
Multiple cell types in the kidney contribute to the development of renal fibrosis. Therefore, precision-cut kidney slices may be used as an ex vivo model with all different cell types present, as described for liver (Westra et al., 2014), to study the onset and late stage of renal fibrosis.

Early onset of fibrosis and precision-cut renal slices
Prolonged culture of human kidney slices paves the way to use slices for the study of fibrosis (Vickers and Fisher, 2005). After 72 h of culture, kidney slices show signaling pathways of repair and a fibrogenic response as seen by an increased deposition of collagen IV and increased gene expression of collagens, laminins, contractile proteins and markers of proliferation (Vickers and Fisher, 2005). Therefore, this model is applicable to study mechanisms of fibrotic renal disease (Vickers and Fisher, 2005).
As a model for renal fibrosis, renal cortex slices of transgenic mouse harboring the luciferase reporter gene under the control of collagen I promoter, were used to investigate the mechanisms of collagen up-regulation in renal tissue (Tharaux et al., 2000). In kidney slices incubated with AngII, the procollagen 1a2 gene was activated, and by using specific inhibitors it could be shown that MAPK/ ERK, AP-1 and TGFβ are involved in this process (Tharaux et al., 2000). These in vitro results indicate that AngII also directly induces collagen 1 gene expression, not only indirectly by the in vivo induction of hypertensive nephropathy.
Renal slices from rats recovered for 5 weeks from an acute kidney injury 24 Chapter 1 | GENERAL INTRODUCTION (AKI) induced by ischemia-reperfusion (I/R) injury, were used to study oxidative stress in renal disease (Basile et al., 2012). In vivo experiments showed that AKI enhances the profibrotic response to AngII, which is released during oxidative stress. Dihydroethidium (DHE) incorporation as parameter for oxidative stress, was increased in renal slices recovered after AKI compared with slices from control rats (Basile et al., 2012). These data indicate that the kidney slices are an important model for the study of fibrosis in AKI induced by I/R injury. In this thesis, chapter 6, the PCKS model was demonstrated as a novel tool to test the pathophysiology of fibrosis and to screen the efficacy of antifibrotic drugs ex vivo in a multicellular and profibrotic milieu. An IFNγ conjugate targeted to the PDGFRβ significantly reduced TGFβ1-induced fibronectin, collagen I and collagen III mRNA expression (Poosti et al., 2015).

End-stage fibrosis and precision-cut kidney slices
Nagae et al. used fibrotic kidney to prepare whole kidney, medulla or cortex slices, from UUO rats to study the effect of adrenomedullin (ADM), a vasodilating peptide that reduces renal fibrosis in hypertensive animals on cyclic adenosine monophosphate (cAMP) production (Nagae et al., 2008). In the kidney slices of UUO rats, cAMP production was increased compared with slices of control kidneys and an ADM-neutralizing antibody partially blocked the enhanced cAMP production (Nagae et al., 2008). However, they used rather thick slices (1.5 mm) for a very short incubation time of maximally 45 min. Using thinner, PCKS would probably have prolonged the time period that these kidney slices could be used. Still, even in these relatively thick renal fibrotic slices pharmacological intervention studies were possible.
Flamant et al. studied the effect of discoidin domain receptor 1 (DDR1), a nonintegrin collagen receptor that displays tyrosine-kinase activity, on the development of renal fibrosis, using fibrotic kidney slices besides in vivo experiments (Flamant et al., 2006). Renal cortical slices from high-NaCl fed DDR1 −/− mice showed decreased monocyte chemoattractant protein-1 (MCP-1) secretion following stimulation with lipopolysaccharide compared with slices from high NaCl fed wildtype mice, indicating the importance of DDR1 in the process of renal inflammation and fibrosis (Flamant et al., 2006).

CONCLUSIONS AND PERSPECTIVES
PCTS from intestine and kidney represent a useful tool to investigate the early onset of fibrosis in both human and experimental animals. Utilizing the slice model can lead to new insights into the (human) mechanisms of the onset of organ fibrogenesis. In addition, in the future, treatment of PCTS with the fibrosis-inducing compounds or factors, used to induce fibrosis in vivo in the different organs, may help to further elucidate the mechanisms of human fibrosis. Furthermore, tissue slices from fibrotic intestine and kidney from both human and animals could be used to study the factors that play a role in established fibrosis. Last but not least, human (fibrotic) tissue slices will pave the way for the testing of novel therapeutic pharmacological interventions for human fibrosis, which up till now could only be performed in animal models.