CC BY-NC-ND
0Human Pathology (2015) 46, 84-93
www.elsevier.com/locate/humpath
Original contribution
Elaboration of tubules with active hedgehog ®>CmssM:*k
drives parenchymal fibrogenesis in gestational alloimmune liver disease^'^
Akihiro Asai MD, PhDab*, Samyukta Malladia, Jonathan Mischa, Xiaomin Pan MDa, Padmini Malladi MSa, Anna Mae Diehl MDc, Peter F. Whitington MDa b
3Stanley Manne Children's Research Institute, Chicago, IL 60614
bDepartment of Pediatrics, Ann and Robert H Lurie Children's Hospital of Chicago, Feinberg School of Medicine of Northwestern University, Chicago, IL 60611
cDepartment of Medicine, Duke University School of Medicine, Durham, NC 27710 Received 8 July 2014; revised 17 September 2014; accepted 19 September 2014
Keywords:
Neonatal hemochromatosis; Liver progenitors; Liver development; Neoductules; Neocholangioles; Hedgehog signaling; Osteopontin
Summary Gestational alloimmune liver disease (GALD) produces severe neonatal liver disease that is notable for paucity of hepatocytes, large numbers of parenchymal tubules, and extensive fibrosis. Liver specimens from 19 GALD cases were studied in comparison with 14 infants without liver disease (normal newborn liver; NNL) to better understand the pathophysiology that would produce this characteristic histopathology. GALD liver parenchyma contained large numbers of tubules comprising epithelium expressing KRT7/19, EPCAM, and SOX9, suggesting biliary progenitor status. Quantitative morphometry demonstrated that in GALD, the area density of KRT19+ tubules was 16.4 ± 6.2 versus 2.0 ± 2.6 area% in NNL (P < .0001). Functional hepatocyte mass was markedly reduced in GALD, 16.3 ± 6.2 versus 61.9 ± 11.0 area% of CPS1+ cells in NNL (P < .0001). A strong inverse correlation was established between CPS1+ area density and KRT19+ area density (r2 = 0.66, P < .0001). Tubules showed active hedgehog signaling as determined by SHH and nuclear GLI2 expression and expressed the profibrogenic cytokine SPP1. SPP1 protein content and SPP1 expression were greater in GALD than NNL (15- and 13-fold respectively; P = .002). GALD liver contained large numbers of activated myofibroblasts and showed greater than 10-fold more fibrosis than NNL. The extent of fibrosis correlated with the area density of KRT19+ tubules (r2 = 0.387, P = .001). The data support a pathogenic model in which immune injury to fetal hepatocytes provides a stimulus for expansion of parenchymal tubules, which, by way of Hh activation, produce fibrogenic signals leading to vibrant fibrosis.
©2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
☆ Disclosures: All authors have nothing to disclose. ☆☆ Funding/Support: This work was supported by funding from the Siragusa Foundation, Chicago, IL, and the Pediatric Liver Research Fund, Lurie Children's Hospital Foundation, Chicago IL.
* Corresponding author. 240 Albert Sabin Way, S-DOC S6-334, Cincinnati OH 45229.
E-mail addresses: aki4810@gmail.com (A. Asai), sammalladi@gmail.com (S. Malladi), jmisch2008@gmail.com (J. Misch), xpan@luriechildrens.org (X. Pan), pmalladi@luriechildrens.org (P. Malladi), annamae.diehl@duke.edu (A. M. Diehl), p-whitington@northwestern.edu (P. F. Whitington).
1. Introduction
Severe fibrotic liver disease in the newborn indicates the onset of liver injury during fetal life. Such disease has been called "congenital cirrhosis" and has been associated with the neonatal hemochromatosis phenotype, wherein there are iron overload and tissue siderosis in a pattern similar to that seen in
http://dx.doi.org/10.1016/j.humpath.2014.09.010
0046-8177/© 2014 The Authors. Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/3.0/).
hereditary hemochromatosis [1]. The most frequent cause of fetal liver injury leading to congenital cirrhosis and the neonatal hemochromatosis phenotype is gestational alloimmune liver disease (GALD) [2]. GALD-related alloimmunity is specifically directed at fetal hepatocytes. No nonhepatocyte elements in the liver appear to be injured, and tissues outside the liver appear to be unaffected by the primary immune process. The mechanism of alloantibody-induced hepatocyte injury appears to involve the fetal innate immune system. The terminal complement cascade is activated by the classical pathway and results in the formation of membrane attack complex. Immunohistochemical staining for C5b-9 complex (the neoantigen created during terminal complement cascade activation and culmination with formation of membrane attack complex) shows nearly all hepatocytes in cases of GALD to have complement-mediated injury [3]. The immediate result of such injury may be liver failure in the fetus and may lead to fetal death [4]. However, in most cases, the process moves more slowly, starting in midgestation and resulting in liver failure and often cirrhosis by term [2,5]. Newborns with GALD typically show clinical liver failure usually within the first days of life.
The immune injury in GALD is specifically directed at hepatocytes and results in liver disease confined to the lobule. The liver histopathology includes severe depletion of hepatocytes relative to normal newborn liver (NNL) [5]. Remaining hepatocytes appear as giant cells and pseudoacini, which are constructs of hepatocytes surrounding a central space often containing bile. Another epithelial structure appears in abundance in the livers of GALD cases: namely, tubular forms, which consist of epithelial cells surrounding a narrow central lumen that is usually devoid of bile. They are usually narrow and elongated in contrast to the generally round pseudoacini. They are morphologically similar to the "ductular reaction" observed in humans with massive and submassive necrosis associated with acute liver failure [6,7]. It is thought that this represents a regenerative effort and that the ductules may be derived from stem cells residing in the canals of Hering and/or ductal plate remnants [7]. Further expansion of regenerative elements results in the "ductular reaction" in which neoductules (also known as neocholangioles) expressing biliary progenitor markers extend from the ductal plate/canals of Hering into the lobule. Biliary atresia and other cholestatic liver diseases of infancy are notable for "ductular reaction" along the limiting plate ofthe portal triad. In the case of biliary atresia, it appears that a repair response to biliary injury leads to a hyperactive hedgehog (Hh)-driven regenerative signal with resultant biliary dysmorphogenesis [8]. The liver histopathology in GALD and biliary atresia are in stark contrast; however, the 2 diseases are similar in that they exhibit extensive fibrosis very early in life.
In GALD, fibrosis is in the lobule, whereas in biliary atresia, it is portal. Tubules that appear throughout the lobule in GALD share several morphologic features, with reactive ductules appearing along the limiting plate in biliary atresia. In addition, GALD is a disease of the fetal liver wherein Hh and other developmental pathways are maximally active. We hypothesized that formation of parenchymal tubules drives the lobular
fibrogenesis in GALD, in analogy to the role of reactive ductules in portal fibrogenesis in biliary atresia [8,9]. Hence, we examined the relationships among tubule formation, Hh activity, osteopontin production, myofibroblast expansion and activation, and fibrosis in GALD. The results suggest that expanded formation of tubules is likely central to lobular fibrogenesis in GALD and that active Hh signaling by tubules is involved.
2. Materials and methods
2.1. Cases and reference materials
GALD cases: liver specimens from 19 GALD cases were included in the study. The subjects of this study were newborns with liver failure and the NH phenotype. All 19 liver specimens showed evidence of complement-mediated hepatocyte damage and were among cases previously reported [3]. Their postpartum ages ranged from 1 to 12 weeks. Of these, 14 specimens were collected by autopsy and 5 were hepatic explants. All 19 specimens were used for histology and immunohistochemistry studies, whereas the 5 hepatic explants were in addition snap-frozen at the time of harvest and processed for RNA and protein expression analyses. Reference cases (referred to as NNL throughout) comprised the following. Liver specimens from 8 newborns who died from perinatal asphyxia were obtained at autopsy. These cases showed no histologic liver damage and have been similarly used in other published studies [3,5,10]. Liver specimens from 6 children aged 4 months to 5 years were snap-frozen at harvest and processed for RNA and protein analysis. These cases included 4 deceased organ donors, 1 infant undergoing laparotomy, and the disease-free liver of 1 child undergoing hepatoblastoma resection. The specimens all showed normal histology and have been used for gene and protein expression analyses in comparison with neonatal and infant liver disease in other published studies [8,10]. This study was approved by the institutional review board of Lurie Children's Hospital of Chicago.
2.2. Molecular techniques
The expressions of the various genes listed in the Table were determined using real-time reverse transcription poly-merase chain reaction, as previously described [8,10]. The detailed methods and primer sequences used are presented in Supplementary Table S1. The expression of each gene was normalized to the expression of GAPDH in each sample. The results for individual genes are listed according to how they appear in the following results and will be presented as supporting evidence for the other studies performed. The expressions of genes relative to GAPDH in GALD liver and NNL are provided in Supplementary Table S2. In the Table, the normalized expression of each gene in GALD and NNL was computed as the fold-expression relative to the median expression in the NNL cohort. The statistical significance of
Table ] Expressions of genes relevant to cell lineage, Hh
pathway, and fibrogenesis determined by RT-PCR in GALD
liver expressed as fold-difference from NNL
Gene Description Fold difference Wilcoxon
symbol NNL, mean P value
(range)
KRT7 Keratin 7 9.7 (2.2-30.8) .004
KRT19 Keratin 19 16.6 (3.8-28) .002
CPS1 Carbamoyl-phosphate 0.7 (0.02-2.1) NS *
synthase 1
EPCAM Epithelial cell adhesion 50.1 (29.4-68.5) .002
molecule
SOX9 SRY (sex determining 8.9 (2.6-19.0) .002
region Y)-box 9
SHH Sonic Hh 2.1 (0.6-6.5) NS *
GLI1 GLI family zinc finger 1 3.5 (0.8-8.4) NS *
GLI2 GLI family zinc finger 2 8.1 (4.4-18.4) .002
GLI3 GLI family zinc finger 3 7.7 (1.1-23.0) .024
PTCH1 Patched 1 1.6 (0.6-4.1) NS *
HHIP Hh interacting protein 4.2 (1.9-10.9) .041
SMO Smoothened 0.6 (0.1-1.5) NS *
SPP1 Secreted 13.3 (4.2-24.4) .002
phosphoprotein 1 (also
known as osteopontin)
S100A4 Fibroblast specific 1.1 (0.3-1.6) NS *
protein 1
ACTA2 a-Smooth muscle actin 8.2 (3.2-14.0) .002
(also known as a-SMA)
GFAP Glial fibrillary acidic 0.5 (0.03-1.0) NS *
protein
DES Desmin 15.5 (6.8-26.6) .002
ELN Elastin 24.4 (7.1-57.9) .002
COL1A1 Collagen, type I, alpha 1 26.1 (5.3-75.7) .041
COL3A1 Collagen, type III, alpha 1 7.5 (2.7-14.0) .006
COL4A2 Collagen, type IV, alpha 2 13.7 (7.7-26.2) .002
COL5A1 Collagen, type V, alpha 1 10.7 (3.2-28.3) .006
Abbreviation: RT-PCR, reverse transcription polymerase chain
reaction.
* NS, Not significant, P > .05.
the expression in GALD liver versus NNL was determined by Wilcoxon test.
2.3. Histology, immunohistochemistry, and quantitative morphometry
Immunohistochemistry was used to characterize cell lineage, to localize cells that produce Hh pathway factors, and to demonstrate complement-mediated hepatocyte damage as described [3,8]. The methodology and antibodies used are provided in Supplementary Table S3. To determine area density of cell types expressing various markers, immunohis-tochemically stained liver sections were analyzed using a point-counting method as previously described [5]. To quantify tissue fibrosis, area density of fibrosis and collagenous fiber was measured in picro Sinus Red (PSR)-stained
liver sections using computerized quantitative morphometry. Details are provided in the supplementary information.
2.4. Measurement of SPP1 (also known as osteopontin) protein concentration
Liver SPP1 concentration was measured by immunoassay using electrochemiluminescence detection technology (Meso Scale Discovery, Gaithersburg, MD) according to the manufacturer's instructions. Published work shows that this immunoassay system provides precise measurement of SPP1 at low concentration in small biological samples [11]. We validated the assay characteristics with recombinant SPP1 and found the precise range of measurement to extend from 0.066 to 200 ng/well (r2 = 0.999). Measurements were made in triplicate for each sample.
3. Results
3.1. Characterization of tubular epithelial lineage and the relationship between the densities of tubules and hepatocytes
Immunohistochemical phenotyping with various lineage markers was performed to characterize tubules. KRT7 and KRT19 were strongly and evenly expressed in tubules distributed throughout the lobule (Fig. 1A). These markers of biliary epithelial lineage are expressed on biliary progenitors during development and regeneration [12] as well as mature bile ducts. Hepatocyte forms (multinucleated giant cells and pseudoacini) were negative for both markers. In contrast, in NNL, these markers were expressed in interlobular bile ducts, epithelium at the limiting plate of portal areas and canals of Hering (Fig. 1A, inset). Quantitative morphometry showed a marked increase in KRT19+ cell area density relative to NNL: GALD KRT19+ area density (n = 16) was 16.4 ± 6.2 area% versus 2.0 ± 2.6 area% in NNL (n = 8; P < .0001). KRT7 expression was 9.7-fold greater in GALD than NNL, and KRT19 expression was 16.6-fold greater (Table).
The urea cycle enzyme CPS1 is expressed in the liver by functional hepatocytes [13]. In specimens from GALD, CPS1 localized mainly to hepatocyte forms (Fig. 1B). Hepatocytes in NNL showed uniform CPS1 positivity (Fig. 1B, inset). Quantitative morphometry showed a marked reduction in functioning hepatocytes in GALD cases: GALD cases (n = 16) had 16.3 ± 6.2 area% ofCPS1+ cells versus 61.9 ± 11.0 area% in NNL (n = 8; P < .0001). An inverse correlation was established between CPS1+ area density and KRT19+ area density in newborn liver, including GALD and NNL (Fig. 1C).
Liver sections were further studied by immunohisto-chemistry for expression of epithelial cell adhesion molecule (EPCAM), which is expressed on biliary epithelial stem cells and progenitors [12,14,15] and the foregut-derived-stem cell transcription factor SOX9 [15-17]. In GALD, epithelial cells
Fig. 1 Lineage markers expressed by epithelium of lobular tubules in GALD suggests that they are biliary progenitors. A, Tubules strongly expressed KRT19. Inset, KRT19 expression in NNL is limited to ductal plate and canals of Hering. B, Only occasional epithelial cells in tubules expressed the hepatocyte marker CPS1. Inset, uniform CPS1 expression in hepatocytes of NNL. C, In this set of newborn livers, including GALD and NNL, area density of CPS1+ cells inversely correlated with that of KRT19+ cells: r2 = 0.66, P < .0001. D, Epithelial cells in tubules stained for EPCAM. Inset, biliary EPCAM expression in NNL. E, Most epithelial cells in tubules stained for SOX9. Inset, interlobular bile duct SOX9 expression in NNL. All images acquired at *200 magnification.
in tubules uniformly expressed EPCAM (Fig. 1D) and most expressed SOX9 (Fig. 1E). Hepatocyte forms were negative for EPCAM and SOX9. In NNL, ductal plate epithelial cells expressed both EPCAM and SOX9, and interlobular bile ducts expressed SOX9 (insets of Fig. 1D and E). EPCAM expression in GALD liver was 50-fold greater than that in NNL, and SOX9 expression was 8.9-fold greater (Table). These findings further suggest that the plentiful tubules spread throughout the lobule in GALD liver are potentially progenitors designated for elaboration of biliary epithelium and hepatocytes.
3.2. Hh pathway and SPP1 production in tubules
In order to determine if tubules in GALD exhibit Hh signaling, we examined liver specimens from GALD for expression of molecules related to the Hh signaling pathway. Most tubules in GALD specimens strongly expressed Hh
ligand sonic (SHH) by immunohistochemistry (Fig. 2A), whereas SHH could not be detected in hepatocyte forms or other elements of the liver. Scant SHH expression was detected in NNL, mainly in cells along the limiting plate of some portal tracts (Fig. 2A, inset). In order to determine what cells might be responding to the SHH signal, livers were examined for expression of the Hh-responsive gene product, nuclear transcription factor for Hh signal activation glioblastoma-2 (GLI2). Nuclear staining for GLI2 provides evidence for Hh response [18] and was seen prominently in the epithelial cells of tubules (Fig. 2B). Nuclear GLI2 expression could not be detected in NNL. Double staining for EPCAM and GLI2 confirmed that epithelial cells constituting tubules are Hh responsive (Fig. 2C). The results of gene expression assays (Table) show variable expressions of Hh pathway-related genes. GLI2, GLI3, and HHIP were significantly over-expressed relative to NNL, whereas SHH, GLI1, PTCH1, and SMO were not. These results are not surprising as the
Fig. 2 Tubules in GALD produce Hh ligand, show Hh responsiveness, and produce Hh response product. A, Tubules express SHH. Inset, faint SHH expression by periportal hepatocytes and epithelial cells along the limiting plate in NNL. B, Tubules show Hh responsiveness, as demonstrated by positive GLI2 staining of most epithelial nuclei. C, Double staining for GLI2 (brown) and EPCAM (blue-green) demonstrates tubules are Hh-responsive. D, Tubules are the main site of SPP1 expression in GALD liver. E, NNL shows SPP1 expression by epithelial cells along the limiting plate and interlobular bile ducts. F, When analyzed by immunoassay in snap-frozen liver, SPP1 concentration is significantly higher in GALD (n = 5) than in NNL (n = 6): Wilcoxon test, P = .002. Data are displayed as box-and-whisker plot.
GALD livers were referenced to infant liver, wherein Hh pathway is highly active in development [8]. The location of the activity as indicated by immunohistochemistry clearly separates GALD from NNL. In sum, these findings indicate that the epithelium of parenchymal tubules is Hh responsive as well as producing SHH.
SPP1 (also known as osteopontin) is a potent profibro-genic cytokine and a known target gene of the Hh pathway [19]. In GALD specimens, only tubules strongly expressed SPP1 (Fig. 2D). The location of SPP1 staining was consistent with its production by epithelial cells, and in addition, its accumulation within the lumens of tubules suggests that they also secrete SPP1. Scant SPP1 expression was seen in ductal plate cells and interlobular bile ducts in NNL (Fig. 2E). In order to determine the magnitude of SPP1 production in GALD relative to NNL, we measured liver SPP1 protein concentration. GALD liver showed 15-fold greater SPP1 protein content than NNL (Fig. 2F). Similarly, SPP1 expression was 13.3-fold greater in GALD than in NNL (Table). In summary, hepatic SPP1 production (expression) is markedly increased in GALD liver, and parenchymal tubules appear to be responsible for its hyperproduction.
3.3. Phenotype of fibrosis-related cells in GALD liver
Expansion of the mesenchymal cell compartment is an important feature of fibrotic liver. We examined GALD livers by immunohistochemistry for expression of fibroblast-specific protein 1 (FSP1; also known as S100A4), a marker for fibroblastic transformation [20]. Large numbers of S100A4+ cells with fibroblast-like morphology were found in the stroma surrounding tubules, but the epithelium of tubules did not express S100A4 (Fig. 3A). To gain greater granularity of the relationship between tubules and S100A4+ stromal cells, we performed double staining with KRT7 and S100A4. The stains did not colocalize to a single cell type: epithelial cells of tubules were KRT7+, whereas S100A4+ cells closely abutted tubules but did not overlay them (Fig. 3B). This confirmed exclusive localization of S100A4 to fibroblasts and not to tubule epithelium. In order to determine if S100A4+ fibroblasts were responsive to Hh signals, double staining for GLI2 and S100A4 was performed. Some S100A4+ fibroblasts harbored GLI2-positive nuclei, indicating their Hh responsive status (Fig. 3C).
Fig. 3 Increased fibroblasts and myofibroblasts in GALD. A, Staining for S100A4 (also known as FSP1) localized to interstitial cells, but not to tubules. B, Double staining for KRT7 (brown) and S100A4 (blue-green) demonstrates that cells showing this fibroblast marker surround and closely approximate tubule epithelium. C, Double staining for GLI2 (brown) and S100A4 (blue-green) shows that most Hh-responsive cells are epithelial, but some fibroblasts (arrows) also show nuclear GLI2. D, E, and F, Serial sections from the same specimen. D, Staining for a-SMA (also known as ACTA2) is extensive in interstitium surrounding tubules and intense in the wall of an artery in the portal area (upper right). Inset, delicate a-SMA staining of sinusoidal endothelial cells and intense staining of an artery (right) in NNL. E, DES+ myofibroblasts are scattered in the lobule, whereas the portal area (upper right) shows no positivity other than in the arterial wall. Inset, DES+ portal blood vessels in NNL. F, ELN+ myofibroblasts densely populate the interstitium surrounding tubules and portal area stroma (upper right). Inset, ELN staining of portal stroma in NNL.
To further characterize the fibrosis-related cell phenotype, we performed immunohistochemistry on GALD liver specimens for a-smooth muscle actin (a-SMA; also known as ACTA2), a marker for activated myofibroblasts [21]; glial fibrillary acidic protein (GFAP), a marker for quiescent hepatic stellate cells (HSCs) [22]; desmin (DES), a marker for activated HSCs [23]; and elastin (ELN), a marker for portal fibroblasts [21,24]. a-SMA+ cells were observed throughout lobular stroma, but not in tubules and not in the stroma of the portal triad (Fig. 3D). NNL showed delicate a-SMA staining of sinusoidal endothelial cells and intense staining in walls of arteries (Fig. 3D, inset). No GFAP+ cells were identified in GALD specimens, whereas scattered GFAP+ cells were identified in the lobule in NNL (data not shown). Scattered DES+ cells were seen in the lobule in GALD specimens, but not in the stroma of the portal triad (Fig. 3E). NNL showed no DES positivity other than in the walls of portal blood vessels (Fig. 3E, inset). The area occupied by DES+ cells was a small fraction of that staining for a-SMA. Extensive ELN staining was seen in lobular stroma and throughout the stroma of portal triads in GALD liver (Fig. 3F), whereas staining was seen only in the stroma of portal areas in NNL (Fig. 3F, inset). In support of these findings, GFAP
expression was not increased in GALD relative to NNL, whereas ACTA2, DES, and ELN were all several-fold over-expressed (Table). Altogether these findings indicate a marked increase in myofibroblasts in GALD liver. The extensive ELN staining coincided with that for a-SMA. The extent of ELN staining relative to DES suggests that the increased population of myofibroblasts arose through expansion of the pool of designated portal fibroblasts rather than from HSCs.
3.4. The amount of fibrosis in GALD and its relationship to tubule density
Computerized morphometry was performed on PSR-stained liver sections to quantitate total fibrosis and collagen area density. GALD cases showed diffuse fibrosis throughout the liver lobule (Fig. 4A), which when observed with polarized light showed yellow-green birefringence characteristic of organized collagen (Fig. 4B). NNL showed fibrosis and organized collagen only in portal areas (Fig. 4C and D). The fibrosis area density, measured by image analysis of PSR-stained liver sections photographed in bright field, was 35.6 ± 13.4 area% in GALD (n = 17) versus 2.7 ± 1.3 area% in NNL (n = 9;
GAUD PSR polarized
, ft -;
M- / .. v,
; v , ; -J,,
• ■■". - 'Ij/A ■■■ ■'w ■■ : ...
p V/v* v
■ „ - ' " ■ < - - > -
Küi ^iMi
■ . v . '
NNL PSR polarized
Fig. 4 Extensive lobular fibrosis and collagen deposition in GALD. A, PSR-stained GALD specimen shows extensive fibrosis (red) throughout the lobule, whereas the portal area (P, lower left) shows minimal fibrotic expansion. B, The same specimen shows extensive cross-linked collagen coinciding with the fibrosis (green-yellow). C and D, NNL shows fibrosis and collagen only in the portal area. E and F, Computerized morphometry performed on the liver from GALD cases (n = 19) shows an excess of fibrosis and collagen relative to NNL (n = 8). Data are displayed as box-and-whisker plot; Wilcoxon for GALD versus NNL, P < .0001 for both fibrosis and collagen.
P < .0001; Fig. 4E). The collagen area density, measured by analysis of images obtained with polarized light, was 16.1 ± 5.7 area% in GALD (n = 19)versus 1.4±0.7area%inNNL(n = 9; P < .0001; Fig. 4F). Genes related to collagen formation in human liver disease were all several-fold overexpressed in GALD liver relative to NNL: COL1A1, 26-fold; COL3A1, 7.5-fold; COL4A2, 13.7-fold; and COL5A1,10.7-fold(Table). In sum, these data show increased fibrogenesis and extensive lobular fibrosis and collagen deposition in GALD.
Among GALD subjects, some have fewer hepatocyte forms and more tubules, whereas others have more
hepatocyte forms and fewer tubules (Fig. 1C). Observation suggested that specimens with relatively more tubules had greater fibrosis than those with relatively more hepatocyte forms (Fig. 5). To evaluate this potential relationship, we performed linear correlation analysis between the area density KRT19+ cells and the area density of fibrosis and collagen. KRT19+ cell area density correlated well with the extent of both total fibrosis (Fig. 5E) and collagen (Fig. 5F). In summation of the foregoing data, lobular fibrosis and collagen deposition are markedly increased in GALD liver and correlate with the density of KRT19+ tubules.
Fig. 5 The degree of fibrosis and collagen deposition correlates with the density of tubules. A and B, Images from a representative GALD case show few CPS+ hepatocyte forms and extensive KRT19+ tubules (brown), with extensive fibrosis (red). C and D, Images from a representative GALD case show more CPS+ hepatocyte forms and fewer KRT19+ tubules, with less extensive fibrosis. E and F, Computerized morphometry performed in a sample set of newborn livers containing GALD (n = 16) and NL (n = 8) shows a positive correlation between KRT19+ (tubule) area density and fibrosis area density (r2 = 0.387, P = .001) and collagen area density (r2 = 0.343, P = .003).
4. Discussion
This study of GALD provides evidence for a novel mechanism of fibrogenesis in which hyperactive developmental pathways play a significant if not singular role in the evolution of liver lobular fibrosis. Liver injury in GALD is mediated by maternally derived IgG and fetal complement, is directed specifically at hepatocytes, and results in dramatic hepatocyte injury and loss. In order for the fetus to survive,
effort is required to replace lost and damaged hepatocytes and maintain some critical level of liver function. This effort appears to engage foregut-derived stem cells and results in extensive elaboration of tubules expressing KRT7/19, EPCAM and SOX9 throughout the lobule. These tubules exhibit active Hh signaling and produce SPP1 in excess. The data suggest that Hh signaling constitutes an important mechanism in the repair process in response to hepatocyte injury in GALD. However, an untoward effect of excessive
Hh-pathway activation results, namely, extensive interstitial fibrosis. The data suggest that Hh signaling may mediate the expansion of the mesenchymal compartment comprising myofibroblasts and the massive increase in fibrosis seen in GALD. Thus, the data suggest that in GALD, regeneration and lobular fibrosis are linked, at least partially through Hh.
Our results provide evidence for a pathway for aberrant lobular tubulogenesis in GALD that could only occur in developing liver. Expression of SOX9, EPCAM, and KRT7/19 by epithelium of tubules suggests that they comprise foregut-derived biliary progenitors [17,25]. However, their uniform distribution throughout the lobule is quite different from the periportal ductular reaction seen in massive hepatic necrosis in mature liver [6]. At midgestation in normal human embryonic development, biliary progenitors may be seen in the lobule admixed with hepatoblasts [26]. In mature liver undergoing injury, regeneration requires recruitment of stem cells from limited rests in the ductal plate, whereas in developing liver, interruption of hepatocyte maturation as seen in C/EBPalpha (Cebpa)-knockout mice results in expansion of progenitors located in the lobule [27,28]. There is no understanding of how achievement of a full complement of hepatocytes feeds back on the system to limit their further production. The inverse relationship between the density of KRT19+ tubules and that of hepatocytes in GALD liver suggests that loss of hepatocytes prompts regeneration leading to a markedly expanded progenitor compartment in the location normally occupied by hepatocytes in developing liver.
The findings in this study provide evidence that aberrant developmental signaling and extensive elaboration of tubules in GALD create an environment conducive to lobular fibrogenesis. Expansion of progenitors in Cebpa knockout mice is associated with fibrosis [27]. The strong correlation between the area density of tubules and the extent of fibrosis in GALD liver suggests a similar relationship. The connective tissue surrounding tubules in GALD liver is densely occupied by activated myofibroblasts. The origin of the myofibroblasts cannot be determined from the data. ELN staining coinciding with extensive a-SMA staining suggests lineage similar to portal fibroblasts. S100A4+ fibroblasts that closely surround tubules are Hh responsive and potentially are receiving stimulus from SHH elaborated by epithelial cells in tubules. Tubules appear to be the source of the markedly increased expression of the profibrogenic cytokine SPP1 observed in GALD. A binding element for the Hh pathway transcription factor GLI1 in its promoter induces greater SPP1 expression with activation of Hh signaling [19,29]. Our data show active Hh signaling and SPP1 expression to be colocalized to tubules. Hh-induced SPP1 expression has been shown to contribute to lobular fibrosis in experimental and human nonalcoholic liver disease [30,31] and alcoholic liver disease [32-35]. In addition, it has been suggested that SOX9 directly regulates SPP1 expression in HSCs and that Hh acts through SOX9 to promote fibrosis [36]. However, in our studies, SOX9 and SPP1 expression localized to tubules, and neither was observed in mesenchy-
mal cells. Nevertheless, the possibility exists that Hh regulates SPP1 expression in tubules both via GLI1 and SOX9 pathways. Furthermore, Hh signal may directly influence myofibroblastic transformation of HSCs [37,38], and excessive SOX9 expression in fetal hepatocytes may mediate ectopic extracellular matrix deposition characteristic of fibrosis by way of HSC transformation [39]. Thus, there appear to be multiple tubule-associated signal pathways in GALD that could contribute to fibrosis associated with expanded numbers of parenchymal tubules: Hh-driven SPP1 production, SOX9-driven SPP1 production, Hh-driven myofibroblastic transformation of HSCs, and SOX9-driven ectopic extracellular matrix deposition via HSC transformation. The relative contributions of these mechanisms cannot be determined from the data.
In summary, the results of this study provide evidence for a novel model of pathogenesis in GALD wherein injury to fetal hepatocytes leads to tubulogenesis and fibrosis. This patho-physiologic process occurs in the background of developing liver and represents chronic liver disease of the fetus. Beginning around midgestation and extending to term, immune injury to hepatocytes precludes normal lobular development, which promotes expansion of tubules and in turn promotes fibrogen-esis. The process culminates in the unique liver pathology associated with neonatal hemochromatosis.
Supplementary data
Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.humpath.2014.09.010.
References
[1] Knisely AS, Mieli-Vergani G, Whitington PF. Neonatal hemochromatosis. Gastroenterol Clin North Am 2003;32:877-89 [vi-vii].
[2] Whitington PF. Gestational alloimmune liver disease and neonatal hemochromatosis. Semin Liver Dis 2012;32:325-32.
[3] Pan X, Kelly S, Melin-Aldana H, Malladi P, Whitington PF. Novel mechanism of fetal hepatocyte injury in congenital alloimmune hepatitis involves the terminal complement cascade. Hepatology 2010;51:2061-8.
[4] Whitington PF, Pan X, Kelly S, Melin-Aldana H, Malladi P. Gestational alloimmune liver disease in cases of fetal death. J Pediatr 2011;159:612-6.
[5] Bonilla SF, Melin-Aldana H, Whitington PF. Relationship of proximal renal tubular dysgenesis and fetal liver injury in neonatal hemochromatosis. Pediatr Res 2010;67:188-93.
[6] Demetris AJ, Seaberg EC, Wennerberg A, Ionellie J, Michalopoulos G. Ductular reaction after submassive necrosis in humans. Special emphasis on analysis of ductular hepatocytes. Am J Pathol 1996;149:439-48.
[7] Theise ND, Saxena R, Portmann BC, et al. The canals of Hering and hepatic stem cells in humans. Hepatology 1999;30:1425-33.
[8] Omenetti A, Bass LM, Anders RA, et al. Hedgehog activity, epithelial-mesenchymal transitions, and biliary dysmorphogenesis in biliary atresia. Hepatology 2011;53:1246-58.
[9] Whitington PF, Malladi P, Melin-Aldana H, Azzam R, Mack CL, Sahai A. Expression of osteopontin correlates with portal biliary proliferation and fibrosis in biliary atresia. Pediatr Res 2005;57:837-44.
[10] Bonilla S, Prozialeck JD, Malladi P, et al. Neonatal iron overload and tissue siderosis due to gestational alloimmune liver disease. J Hepatol 2012;56:1351-5.
[11] Pavkovic M, Riefke B, Gutberlet K, Raschke M, Ellinger-Ziegelbauer H. Comparison of the MesoScale Discovery and Luminex multiplex platforms for measurement of urinary biomarkers in a cisplatin rat kidney injury model. J Pharmacol Toxicol Methods 2014;69:196-204.
[12] Zhang L, Theise N, Chua M, Reid LM. The stem cell niche of human livers: symmetry between development and regeneration. Hepatology 2008;48:1598-607.
[13] Butler SL, Dong H, Cardona D, et al. The antigen for Hep Par 1 antibody is the urea cycle enzyme carbamoyl phosphate synthetase 1. Lab Invest 2008;88:78-88.
[14] Cardinale V, Wang Y, Carpino G, et al. Multipotent stem/progenitor cells in human biliary tree give rise to hepatocytes, cholangiocytes, and pancreatic islets. Hepatology 2011;54:2159-72.
[15] Carpino G, Cardinale V, Onori P, et al. Biliary tree stem/progenitor cells in glands of extrahepatic and intraheptic bile ducts: an anatomical in situ study yielding evidence of maturational lineages. J Anat 2012;220:186-99.
[16] Carpentier R, Suner RE, van Hul N, et al. Embryonic ductal plate cells give rise to cholangiocytes, periportal hepatocytes, and adult liver progenitor cells. Gastroenterology 2011;141:1432-8 [38 e1-4].
[17] Furuyama K, Kawaguchi Y, Akiyama H, et al. Continuous cell supply from a Sox9-expressing progenitor zone in adult liver, exocrine pancreas and intestine. Nat Genet 2011;43:34-41.
[18] Fukaya M, Isohata N, Ohta H, et al. Hedgehog signal activation in gastric pit cell and in diffuse-type gastric cancer. Gastroenterology 2006;131:14-29.
[19] Das S, Harris LG, Metge BJ, et al. The hedgehog pathway transcription factor GLI1 promotes malignant behavior of cancer cells by up-regulating osteopontin. J Biol Chem 2009;284:22888-97.
[20] Strutz F, Okada H, Lo CW, et al. Identification and characterization of a fibroblast marker: FSP1. J Cell Biol 1995;130:393-405.
[21] Cassiman D, Libbrecht L, Desmet V, Denef C, Roskams T. Hepatic stellate cell/myofibroblast subpopulations in fibrotic human and rat livers. J Hepatol 2002;36:200-9.
[22] Buniatian G, Hamprecht B, Gebhardt R. Glial fibrillary acidic protein as a marker of perisinusoidal stellate cells that can distinguish between the normal and myofbroblast-like phenotypes. Biol Cell 1996;87:65-73.
[23] Geerts A, Eliasson C, Niki T, Wielant A, Vaeyens F, Pekny M. Formation of normal desmin intermediate filaments in mouse hepatic stellate cells requires vimentin. Hepatology 2001;33:177-88.
[24] Li Z, Dranoff JA, Chan EP, Uemura M, Sevigny J, Wells RG. Transforming growth factor-beta and substrate stiffness regulate portal fibroblast activation in culture. Hepatology 2007;46:1246-56.
[25] Kawaguchi Y. Sox9 and programming of liver and pancreatic progenitors. J Clin Invest 2013;123:1881-6.
[26] Terada T. Ductal plate in hepatoblasts in human fetal livers: I. Ductal plate-like structures with cytokeratins 7 and 19 are occasionally seen within human fetal hepatoblasts. Int J Clin Exp Pathol 2013;6:889-96.
[27] Akai Y, Oitate T, Koike T, Shiojiri N. Impaired hepatocyte maturation, abnormal expression of biliary transcription factors and liver fibrosis in C/EBPalpha(Cebpa)-knockoutmice. Histol Histopathol 2013;20:107-25.
[28] Flodby P, Barlow C, Kylefjord H, Ahrlund-Richter L, Xanthopoulos KG. Increased hepatic cell proliferation and lung abnormalities in mice deficient in CCAAT/enhancer binding protein alpha. J Biol Chem 1996;271:24753-60.
[29] Yoon JW, Kita Y, Frank DJ, et al. Gene expression profiling leads to identification of GLI1-binding elements in target genes and a role for multiple downstream pathways in GLI1-induced cell transformation. J Biol Chem 2002;277:5548-55.
[30] Syn WK, Agboola KM, Swiderska M, et al. NKT-associated hedgehog and osteopontin drive fibrogenesis in non-alcoholic fatty liver disease. Gut 2012;61:1323-9.
[31] Syn WK, Choi SS, Liaskou E, et al. Osteopontin is induced by hedgehog pathway activation and promotes fibrosis progression in nonalcoholic steatohepatitis. Hepatology 2011;53:106-15.
[32] Banerjee A, Lee JH, Ramaiah SK. Interaction of osteopontin with neutrophil alpha(4)beta(1) and alpha(9)beta(1) integrins in a rodent model of alcoholic liver disease. Toxicol Appl Pharmacol 2008;233:238-46.
[33] Morales-Ibanez O, Dominguez M, Ki SH, et al. Human and experimental evidence supporting a role for osteopontin in alcoholic hepatitis. Hepatology 2013;58:1742-56.
[34] Patouraux S, Bonnafous S, Voican CS, et al. The osteopontin level in liver, adipose tissue and serum is correlated with fibrosis in patients with alcoholic liver disease. PLoS One 2012;7:e35612.
[35] Seth D, Haber PS, Syn WK, Diehl AM, Day CP. Pathogenesis of alcohol-induced liver disease: classical concepts and recent advances. J Gastroenterol Hepatol 2011;26:1089-105.
[36] Pritchett J, Harvey E, Athwal V, et al. Osteopontin is a novel downstream target of SOX9 with diagnostic implications for progression of liver fibrosis in humans. Hepatology 2012;56:1108-16.
[37] Michelotti GA, Xie G, Swiderska M, et al. Smoothened is a master regulator of adult liver repair. J Clin Invest 2013;123:2380-94.
[38] Swiderska-Syn M, Syn WK, Xie G, et al. Myofibroblastic cells function as progenitors to regenerate murine livers after partial hepatectomy. Gut 2014;63:1333-44.
[39] Hanley KP, Oakley F, Sugden S, Wilson DI, Mann DA, Hanley NA. Ectopic SOX9 mediates extracellular matrix deposition characteristic of organ fibrosis. J Biol Chem 2008;283:14063-71.
