Neural Cell Adhesion Molecule and Polysialic Acid in Ductular Reaction: The Puzzle Is Far From Completed, But the Picture Is Becoming More Clear

Ductular reaction (DR) is a histological lesion common to most forms of chronic liver injury (CLI). The functional role of DR in physiological and pathological liver repair was first recognized by Desmet, who suggested that DR was actually the “pacemaker of biliary fibrosis.”¹ Further studies indicated that DR is not restricted to biliary diseases and is associated with most liver diseases, along with hepatic progenitor cell (HPC) activation (formerly called type II or atypical DR).

DR appears to be a major determinant of liver scarring and progression. However, we still tend to think of liver repair in terms of a tango between myofibroblasts and macrophages and only seldom consider that liver repair is actually the result of multiple interactions between cell types of different origin (epithelial, mesenchymal, endothelial, and inflammatory), whose ultimate aim is to restore epithelial integrity. These complex interactions generate the DR, a highly dynamic “construction site” assembled where repair is needed.

The main epithelial component of DR is cholangiocyte-like cells that show a “reactive” phenotype, characterized by the expression of a variety of cytokines, chemokines, growth factors, and angiogenic factors and their cognate receptors.² Different from normal cholangiocytes, reactive ductular cells are arranged in strings, rather than in ductules, and are able to trespass on the limiting plate and “infiltrate” the lobule. The other epithelial components of DR are HPCs (i.e., small, oval cell-like cells coexpressing albumin and cytokeratin 19) localized in a periportal niche in close contact with the terminal cholangioles abutting the canals of Hering. HPCs are bipotential cells capable of differentiating toward both the biliary and hepatocellular lineages.³ Intermediate hepatobiliary cells (IHBCs; i.e., K7⁺ and/or SRY (sex determining region Y)-box 9-positive cells reminiscent of small hepatocytes) constitute the third epithelial phenotype in DR.

The classic view about the histogenesis of these epithelial phenotypes is that reactive cholangiocytes and IHBCs are the progeny of the bipotential HPCs. This is most likely the case in acute liver damage, where the outcome of liver repair is a restitutio ad integrum. However, in chronic liver diseases (CLDs) associated with progressive fibrosis, the reactive cholangiocyte component becomes preponderant, suggesting that reactive ductular cells are a by-product of HPC-driven repair and accumulate in conditions of persistent damage.^4,5 Furthermore, recent data provide new vigor to the hypothesis that DR cells can derive from multiple sources, including transdifferentiation of hepatocytes or proliferation of ductal cells, consistent with the view that the liver has the ability to use different sets of tools to repair the damage, and the particular strategy adopted depends upon the specific etiology.

Regardless of their histogenesis, DR cells undergo phenotypic and functional changes that enable them to establish paracrine communications with the multiple cell types working in the construction site (or “hepatic reparative complex”). These interactions orchestrate complex morphogenetic responses, such as branching tubulization, fibrogenesis, and angiogenesis,⁶ under the tight control of morphogenetic signaling pathways, in particular, Hedgehog, Wnt/β-catenin and Notch.^2,6

Recent research has convincingly shown that the extent of DR cells correlates with the severity of fibrosis in a variety of genetic and acquired liver diseases, from Alagille syndrome to hemochromatosis and from nonalcoholic steatohepatitis to hepatitis C virus.^4–7 Thus, several groups are investigating the factors able to trigger activation and expansion of HPC/DR cells and modulate their interactions with the extracellular matrix (ECM) components.⁵

Generation of the hepatic reparative complex requires the detachment of HPCs from their niche, followed by their spreading along the portal boundaries toward the damage area(s). The gain of migratory properties reflects phenotypically in the expression of several epithelial and mesenchymal markers.^8,9 Among them, the neural cell adhesion molecule (NCAM) is a surface glycoprotein that, in the nervous system, lung, and gut, regulates morphogenetic processes relative to cell migration and tissue differentiation and, in particular, cell-cell and cell-matrix interactions.^10,11 During liver ontogenesis, NCAM is transiently expressed by ductal plate cells, when configured in a single layer, and then it is down-regulated as the ductal plate duplicates. After birth, NCAM is not expressed by normal bile ducts (except for the canals of Hering and peribiliary glands of large intra- and extrahepatic bile ducts¹²), but its expression is then up-regulated in DR cells during CLDs. Up-regulation of NCAM and expression of neuroendocrine features, such as chromogranin A and the parathyroid hormone parathormone–related peptide, in DR cells was first reported on by Roskams et al. more than 20 years ago.¹³ Our group reported on the extensive up-regulation of NCAM expression (along with that of the antiapoptotic protein, B-cell lymphoma 2) in immune-mediated cholangiopathies, ductal plate malformations, and alcoholic liver disease.¹⁴ Ultrastructural analysis of cells coexpressing NCAM/epithelial cell adhesion molecule isolated by immunomagnetic isolation revealed an epithelial phenotype consistent with that of HPCs. These earlier findings suggested that NCAM was involved in liver repair by mediating interactions of HPCs with the ECM and with surrounding stromal cell types, similarly expressing NCAM, as is the case of portal myofibroblasts.^14,15 However, after these initial observations, the function of NCAM in DR cells remained unknown. An elegant article from Tsuchiya et al., published in this issue of Hepatology, adds another important piece to the puzzle.

To exert its biological functions, NCAM requires a posttranslational modification, such as the conjugation with polysialic acid (polysialilation). Polysialic acid (PolySia) is a highly polar structural component of ECM that plays an important role in the development of neural tissue.¹⁶ PolySia possesses a strong affinity for several adhesion molecules expressed by neural epithelial cells, including NCAM, synaptic cell adhesion molecule 1, and neuropilin 2.¹⁷ Among them, NCAM is the best characterized and the most relevant, being able to bind up to 100 PolySia residues.¹⁸ PolySia transfer to NCAM is an enzymatic process catalyzed by two polysialyltransferases belonging to the glycosyl-transferase family (ST8SiaII and ST8SiaIV) that act independently.¹⁹ Conjugation with PolySia changes the adhesive properties of NCAM to antiadhesive as a result of the high hydrophilic content of PolySia chains.²⁰ This process promotes plasticity and migration of NCAM⁺ cells during development. High expression levels of PolySia-NCAM are a defining feature of growing axons during embryonic development as well as of neural progenitor cells.²¹

The study from Tsuchiya et al. unveils the role of PolySia and its carrier, NCAM, in the generation of the DR during repair from liver injury (Fig. 1).²² Using both in vitro and in vivo mouse models of CLI, the investigators elegantly showed that PolySia-NCAM binding triggers HPC-mediated repair mechanisms. In the normal liver, expression of PolySia-NCAM was minimal and confined to a few ductular cells and fibroblasts scattered in the portal area. Subsequent to liver damage, PolySia-NCAM expression increased significantly in both HPC and DR cells and appeared localized to the aspect of DR cells in contact with other nonepithelial NCAM⁺ cells, likely hepatic stellate cells, based on their coexpression of alpha-smooth muscle actin and desmin.²³ These investigators also found that transgenic mice defective for both polysialyltransferases ST8SiaII and ST8SiaIV showed abnormal bile duct development, indicating that PolySia-NCAM interaction is involved also in liver development and tubular morphogenesis.²²

Fig. 1.

PolySia-NCAM interaction modulates activation of HPCs and DR in chronically injured liver. In severe forms of CLI, activation of the HPC compartment is associated with NCAM up-regulation. Because HPCs express the glycosyltransferase, ST8SiaIV (the main polysialyltransferase in the liver, which is also induced by liver damage), NCAM is bound to PolySia. By acting as “lube oil,” PolySia-NCAM binding is a prerequisite for the migration of HPCs to the site of damage (A). In contrast, down-regulation of the PolySia-NCAM complex may accompany the differentiation of HPCs to hepatocytes (B). PolySia cleavage from NCAM by the degrading enzyme, endoN, results in an abortive ductular reaction with DRC aggregates restricted within the portal area (C). DRC, ductular reactive cell.

To understand the functional consequences of the cooperation between NCAM and PolySia, Tsuchiya et al. cultured BMOL cells (an HPC line coexpressing NCAM and PolySia, as well as ST8SiaIV) onto laminin-coated plates to stimulate cell aggregation and then challenged them with hepatocyte growth factor (HGF) to stimulate cell migration. Cell scattering was inhibited in BMOL cells treated with endosialidase (endoN), an enzyme that cleaves PolySia from NCAM, thus suggesting that PolySia is essential for HGF-induced HPC migration by weakening cell-matrix interactions. Using the same in vitro model, the investigators showed that BMOL cell differentiation toward the hepatocyte lineage induced by Oncostatin M was associated with the down-regulation of PolySia with the PolySia cleavage from NCAM. This finding suggests that HPCs that are differentiating into a mature cell phenotype switch off mechanisms governing cell migration. Similarly, when BMOL cells were cocultured with NCAM⁺ myofibroblasts, PolySia cleavage by endoN reduced the migratory properties of BMOL. Finally, in a mouse in which a cholestatic damage was induced in vivo by 3,5-diethoxy-carbonyl-1,4-dihydrocollidine, interference with PolySia/NCAM by endoN treatment resulted in an “abortive” DR, characterized by DR cells that remained confined within the portal area, being unable to migrate into the liver parenchyma (Fig. 1). It is interesting to note that, in this model, inhibition of DR was associated with an increased biliary damage, indicating that HPC/DR-mediated repair is primarily an adaptive response.

In summary, the findings of Tsuchiya et al.²² add a new piece to the puzzle of liver repair by unraveling the biological function of NCAM expression by HPC/DR and its role in facilitating the migration of NCAM⁺ HPCs from the periportal niche to the area(s) to be repaired. Furthermore, NCAM appears to be expressed in some peripheral intrahepatic cholangiocarcinoma (CCA),²⁴ suggesting that interference with PolySia-NCAM binding may reduce the spread of the cancer and, particularly, at the level of perineural invasion, which is critical for the early dissemination of CCA.

However, further studies are needed to clarify the role of its modulation in various conditions and at different times, as well as to address the inhibitory effects of PolySia/NCAM also in conditions of established damage. This study further demonstrates the need to follow a “multidimensional” approach in addressing the mechanisms of liver repair and its physiological or pathological consequences, as well as to consider carefully the interplay between cellular effectors, matrix components, growth factors, and inflammatory signals.

Abbreviatons

CCA

cholangiocarcinoma

CLDs

chronic liver diseases

CLI

chronic liver injury

DR

ductular reaction

ECM

extracellular matrix

endoN

endosialidase

HGF

hepatocyte growth factor

HPC

hepatic progenitor cell

IHBC

intermediate hepatobiliary cell

NCAM

neural cell adhesion molecule

PolySia

polysialic acid