Galanin System in the Human Bile Duct and Perihilar Cholangiocarcinoma

Background: Perihilar cholangiocarcinoma (pCCA) is characterised by poor outcomes. Early diagnosis is essential for patient survival. The peptide galanin (GAL) and its receptors GAL1–3 are expressed in various tumours. Detailed characterisation of the GAL system in pCCA is lacking. Our study sought to characterise GAL and GAL1–3 receptor (GAL1–3–R) expression in the healthy human bile duct, in cholestasis and pCCA. Methods: Immunohistochemical staining was performed in healthy controls (n = 5) and in the peritumoural tissues (with and without cholestasis) (n = 20) and tumour tissues of pCCA patients (n = 33) using validated antibodies. The score values of GAL and GAL1–3–R expression were calculated and statistically evaluated. Results: GAL and GAL1–R were expressed in various bile duct cell types. GAL2–R was only slightly but still expressed in almost all the examined tissues, and GAL3–R specifically in cholangiocytes and capillaries. In a small pCCA patient cohort (n = 18), high GAL expression correlated with good survival, whereas high GAL3–R correlated with poor survival. Conclusions: Our in-depth characterisation of the GAL system in the healthy human biliary duct and pCCA in a small patient cohort revealed that GAL and GAL3–R expression in tumour cells of pCCA patients could potentially represent suitable biomarkers for survival.


Introduction
Cholangiocarcinoma (CCA) is a rare tumour accounting for 2% of all malignancies with increasing mortality rates [1][2][3]. CCA arises from the ductal epithelium of the biliary tree and is classified in accordance with its localisation as either intra-or extrahepatic, which is then further subdivided into perihilar and distal cholangiocarcinoma [2][3][4]. Perihilar cholangiocarcinoma (pCCA) is anatomically defined as a biliary tree tumour located between the origin of the cystic duct and the liver [2,3]. It accounts for approximately 60% of CCA [1,[5][6][7]. pCCA is often diagnosed at advanced stages, and consequently, is characterised by poor clinical outcomes with 1-and 3-year survival rates of 26.2% and 3.4%, respectively. Therefore, early diagnosis would be essential for treatment success [5,6].
It is known that in the pathogenesis of CCA, various signalling pathways are associated with the activation of tumour onset and progression [2,3,8,9]. Moreover, recent studies have revealed that the neuroendocrine system, in particular, plays an important role

Ethics
This study followed the regulations of the Austrian Gene Technology Act and the Helsinki Declaration of 1975 (revised 2013), and was conducted under approval from the Salzburg State Ethics Research Committee as a non-clinical drug trial or epidemiological investigation. It is registered under the number 1102/2020. All patients gave written informed consent, with anonymity maintained via the codification of patient material.

Sample Characteristics
To characterise the GAL system in the human bile duct, we conducted immunohistochemical stainings on human bile duct sections with extensively validated antibodies [44].
As there is a paucity of information to date regarding the GAL system in the biliary tree, we included an analysis of five bile duct samples derived from healthy individuals (Control; Supplementary Table S1), as well as thirty-three histologically evaluated tumour samples derived from patients with pCCA (Supplementary Table S2). pCCA samples were surgically obtained between 2007 and 2019 at the Department of Surgery, University Hospital, Paracelsus Medical University of Salzburg, and were provided for the experiment along with corresponding peritumoural bile duct tissues (PITs; Supplementary Table S3). PITs refer to tissue samples taken furthest from the tumour, and were used to determine whether all the tumour material had been removed by the surgery. They can also refer to the resection margin of the bile duct, and were examined by a pathologist. PITs were further subdivided into PIT with cholestasis (n= 10; PIT+C) and PIT without cholestasis (n = 10; PIT-C) to examine the impact of cholestasis on the expression of the GAL system (Supplementary Table S1 and Table 1). In Control, PIT-C and PIT+C tissue sections, the following cell types were analysed for GAL/GAL 1-3 -R expression: cholangiocytes, mucous glands, fibroblasts, nerve fibres, arterial and venous endothelium, capillary endothelium, adipocytes and smooth muscle. In the samples of pCCA, solely tumour cells were eligible for analysis.

Clinical Parameters
The following clinical parameters were evaluated: age; sex; survival time; clinical signs of cholestasis and cholangitis; TNM (tumour, node and metastasis) classification; infiltration of veins, nerves and lymphatic vessels; grading; residual tumour after surgery; and Bismuth and UICC classification (Supplementary Tables S1 and S2).

Immunohistochemical Staining (IHC)
Tissue sections were washed with xylol, followed by isopropanol and dH 2 O. After incubation with dH 2 O, slides were placed in antigen retrieval buffer (1 mM EDTA, 10 mM Tris; pH = 9) for 40 min at 95 • C. After cooling for 20 min at RT, slides were transferred to dH 2 O and washed for 3 × 3 min in PBS-T (1x PBS + 0.05% Tween ® 20 detergent) to remove residual buffer. For blocking, each slide was covered with Dako REAL™ peroxidase blocking solution (Agilent Technologies, Glostrup, Denmark) and incubated for 10 min. Afterwards, slides were again washed in PBS-T, followed by incubation with the 1st antibody (Ab) diluted in Dako antibody diluent with background reducing components ( Table 2). Primary Abs were previously tested for specificity and extensively validated by the laboratory [44]. The 2nd Ab only (negative) controls were alternatively covered with antibody diluent only. Following 40 min of incubation at 37 • C, slides were washed in PBS-T. Subsequently, each slide was covered with the 2nd Ab, using Dako EnVision + System-HRP Cells 2023, 12, 1678 4 of 24 labelled polymer anti-rabbit solution (Agilent Technologies, Glostrup, Denmark) for 30 min and washed again for 3 × 3 min in PBS-T. For staining, slides were incubated with the peroxidase substrate 3,3 -Diaminobenzidin (DAB) for 10 min to provide comparable results. Staining was terminated by immerging the slides in tap water for 2 × 2.5 min. Subsequently, slides were immersed in Mayer's haemalum solution for 5 min, briefly rinsed in tap water and dipped twice in 3% HCl-ethanol. Finally, tissue sections were rinsed 10-15 times in increasing concentrations of isopropanol (70% → 96% → 100%) and subsequently in 100% xylol. Sections were mounted with cover slides and dried at RT until hardening of the mounting medium (Histokitt No. 1025/500, Karl Hecht GmbH & Co. KG, Sondheim, Germany) occurred. Negative controls (Supplementary Figure S1) and appropriate positive control sections (Supplementary Figure S2), selected for each antigen of interest, were included in immunohistochemical staining (GAL or GAL 1-3 -R-overexpressing cell lines) [44].

Bright Field Microscopy and Slide Scanner
Bright field microscope images were acquired using the VS-120-L Olympus© slide scanner 100-W system. The settings were adjusted to 20× magnification of one z-scan layer. The scan focus points were set manually with settings adjusted to "focusing samples only" and maximum sample detection sensitivity. Scans were used for image analysis.

Image Analysis and Scoring
Analysis was performed using Olympus OlyVia (V3.8) software (Olympus Soft Imaging Solutions, Tokio, Japan). Each section was analysed for the expression of GAL and GAL 1-3 -R via evaluation of staining intensities in cholangiocytes, connective tissue layers, endothelium, nerve fibres, smooth muscle and adipocytes. This was conducted using a previously described scoring system [45] by scaling the staining intensity from 0-3 (0 = no staining, 1 = mild intensity, 2 = moderate intensity, 3 = strong intensity) and counting the percentage of positive cells with minimum values of ≤5% and maximum values of ≥95%. Final score values were derived by multiplying the intensity and percentage: score value [S] = intensity [I] x percentage [E]. Two analysts performed the evaluation and competed scoring independently. Final score values were calculated, and mean values of both evaluations were used for further statistical analysis (Supplementary Tables S3-S7).

Statistical Evaluation
Using GraphPad Prism software version 9.0.0 (GraphPad Software©, San Diego, CA, USA), the statistical description of peptide and receptor expression was performed by calculating the mean ± SEM score values from IHC for each tissue structure and antigen. Differences between the two unpaired groups were calculated to determine significance using the Mann-Whitney-test. Multiple-comparison tests were used to compare pCCAderived with healthy-and PIT-derived cholangiocytes, applying the Kruskal-Wallis-test and post hoc Dunn's multiple comparison test. p < 0.05 was considered statistically significant. Pearson correlation was applied to analyse potential associations between parameters. For analysis of survival, Kaplan-Meier curves were used.

Results
The control tissue sections showed intact structures without destruction of the bile duct epithelial layer or other histological alterations. By contrast, the peritumoural (PIT) samples showed an infiltration of immune cells, an accumulation of adipocytes and damage to the epithelial layer with galled cholangiocytes, especially in samples with cholestasis. As expected, the tumour tissue sections showed destroyed structures of the bile duct wall that formed irregular gland-like structures with mucin-producing cancer cells and disordered tissue infiltration. An investigation of intact histological structures was impossible in these sections.

Galanin in Bile Duct Tissue
In general, the GAL peptide was expressed in various cell types, including cholangiocytes, muscle cells, nerve fibres, adipocytes and endothelial cells. Therefore, GAL was localised to the plasma membrane and cytoplasm of the various cell types (GAL expression and corresponding score values are depicted in Figures 1 and 2, and Supplementary Table S3). A detailed overview of the localisation of GAL expression can be found in Table 3. In many sections, GAL revealed diffuse immunostaining with weak positive background signals. This might be due to the secretion of GAL from GAL-expressing cells [46,47]. The GAL expression results for Control, PIT+C and PIT-C, including all comparisons and associated p values, are summarised in Table 3.

Cholangiocytes
High GAL expression was found in cholangiocytes, specifically lining the bile duct lumen ( Figures 1A-C and 2A). GAL expression did not differ between healthy tissue and PIT with or without cholestasis (score values of Control = 195 ± 11; PIT-C = 209 ± 18; PIT+C = 205 ± 13).

Arteries, Veins and Capillaries
In healthy tissues, the arterial endothelial cells were positive for GAL in all layers of the connective tissue. By comparison, GAL-like immunoreactivity in PIT with or without cholestasis (PIT+C and PIT-C) was reduced by 54-85% in the arterial endothelium compared to control tissue (artery in mucosa: PIT-C: p = 0.0420; PIT+C: p = 0.0025; artery in adventitia: PIT+C: p = 0.0051) ( Figure 1D-F). In cholestasis patients in particular, GAL expression was reduced by more than 78% in mucosal arteries (PIT+C, p = 0.0025) and by 85% in arteries within the tunica adventitia (PIT+C, p = 0.0051) ( Figure 2B).
The endothelial cells of venous and capillary blood vessels showed similar results as arteries compared to the healthy control tissues with highest GAL staining intensities ( Figure 1G-I). Like in arteries, GAL expression in PIT both with and without cholestasis (PIT+C and PIT-C) was lower within mucosal veins compared to the healthy controls (PIT+C: 40%, p = 0.0303; PIT-C: 36%; p = 0.0383). In the tunica adventitia, GAL expression in the veins was, on average, 75% lower in cholestasis patients compared to the healthy controls (p = 0.0051, Figure 2B).
The IHC staining of the capillary endothelium is depicted in Figure 1J-L. GAL expression was reduced by 56% in capillaries of the tunica adventitia in cholestasis (p = 0.0152, Figures 1J-L and 2B). A trend towards lowered mean expression was also found in the mucosal capillaries of cholestasis patients.

Muscle Cells, Nerve Fibres, Adipocytes and Connective Tissue
Several cell types were positive for GAL in healthy tissues, without significant alterations in PIT+C and PIT-C ( Figure 2A). For example, biliary muscles exhibited GAL expression in healthy tissue, which, on average, tended to be lower in PIT tissue (PIT-C: 61% and PIT+C: 46%). In biliary nerve fibres, a trend towards increased GAL expression  Supplementary Tables S1 and S2, and score values in Supplementary Table S3.  Supplementary Tables S1 and S2, and  score values in Supplementary Table S3. Mann-Whitney-Test. * p < 0.05; ** p < 0.01.

Cholangiocytes
High GAL expression was found in cholangiocytes, specifically lining the bile duct lumen (Figures 1A-C and 2A). GAL expression did not differ between healthy tissue and PIT with or without cholestasis (score values of Control = 195 ± 11; PIT-C = 209 ± 18; PIT+C = 205 ± 13).

Arteries, Veins and Capillaries
In healthy tissues, the arterial endothelial cells were positive for GAL in all layers of the connective tissue. By comparison, GAL-like immunoreactivity in PIT with or without

Galanin 1 Receptor-GAL 1 -R
IHC analysis revealed that GAL 1 -R was expressed in cholangiocytes, nerve fibres, adipocytes and blood vessels (Figure 3 and Supplementary Table S4). According to its membrane-bound characteristics, receptor staining exhibited no notable background signal and was restricted to the cell membrane and cytoplasm of the various cell types. The localisation of GAL 1 -R expression, as well as the results for Control, PIT+C and PIT-C, including all the comparisons made and the p values, are summarised in Table 3.   Supplementary Tables S1 and S2, and score values in Supplementary Table S4.

Cholangiocytes
Cholangiocytes showed moderate IHC staining for GAL 1 -R in healthy tissues and PIT ( Figure 3A-C). GAL 1 -R was localised in cholangiocytes towards the apical membrane. Although mean GAL 1 -R expression levels were more than twice as high in cholestasis, the differences did not reach statistical significance in comparison to the control tissue, especially when considering the high variability of GAL 1 -R IHC staining ( Figure 4A).

Cholangiocytes
Cholangiocytes showed moderate IHC staining for GAL1-R in healthy tissues and PIT ( Figure 3A-C). GAL1-R was localised in cholangiocytes towards the apical membrane. Although mean GAL1-R expression levels were more than twice as high in cholestasis, the differences did not reach statistical significance in comparison to the control tissue, especially when considering the high variability of GAL1-R IHC staining ( Figure 4A).

Arteries, Veins and Capillaries
Arteries, veins and capillaries revealed low-to-moderate GAL 1 -R IHC staining ( Figure 3D-F). GAL 1 -R was virtually undetectable in the arteries of PIT-C (Figures 3E and 4B). In cholestasis, GAL 1 -R expression showed a trend toward reduction in the mucosal arteries compared to the healthy controls. Neither observation reached statistical significance due to the high GAL 1 -R staining variability in healthy tissues ( Figure 4B).
In veins, GAL 1 -R showed similar alterations to arteries, with nearly undetectable levels in PIT-C ( Figure 3G-I). In contrast to arteries, this effect was only observed in veins embedded in the tunica adventitia. GAL 1 -R expression in capillaries of the tunica adventitia also tended to be lower in PIT-C and PIT+C compared to the control tissues ( Figure 4B).

Muscle Cells, Nerve Fibres, Adipocytes and Connective Tissue
Myocytes, fibroblasts and connective tissues did not show any GAL 1 -R immunoreactivity ( Figure 4A). Nerve fibres embedded in small nerve fascicles expressed low levels of GAL 1 -R, with similar staining intensities between all groups ( Figure 4A). In large nerve fascicles, PIT-C exhibited a trend towards higher expression compared to the control. Of note, the immunoreactivity of GAL 1 -R accumulated in the adipocytes of PIT ( Figure 3M-O). Compared to controls, adipocytes in both PIT groups (PIT-C and PIT+C) showed 5-times higher GAL 1 -R expression, with significance for the cholestasis group (p = 0.0341) ( Figure 4A).

Galanin 2 Receptor-GAL 2 -R
GAL 2 -R showed overall weak staining intensities in the biliary tree. Its expression was found in cholangiocytes, adipocytes and only at low levels in endothelial cells, but was restricted to the cell membrane in the various cell types ( Figure 5). The score values of GAL 2 -R expression are provided in Supplementary Table S5. The localisation of GAL 2 -R expression, as well as the results for Control, PIT+C and PIT-C, including all the comparisons made and the p values, are summarised in Table 3.

Cholangiocytes
The cholangiocellular expression of GAL 3 -R was present in the control and PIT, with low immunoreactivity in all groups ( Figure 7A-C). In comparison to the control tissues, GAL 3 -R expression showed a trend towards lower levels in PIT-C and similar levels to the controls in PIT+C ( Figure 8A).

Cholangiocytes
In the healthy control tissue, GAL 2 -R exhibited low immunoreactivity in cholangiocytes, lining the bile duct lumen ( Figure 5A). Similar to GAL 1 -R, immunostaining was restricted to the cell membrane, apical towards the duct lumen. In PIT, the cholangiocellular expression of GAL 2 -R tended to be lower in cholestasis, in comparison to the control tissue ( Figures 5B,C and 6A).

Arteries, Veins and Capillaries
Arteries, veins and capillaries were otherwise negative for GAL 2 -R, with only slightly positive cells in the veins and capillaries of healthy tissues. The vessels of the peritumoural tissue GAL 2 -R were nearly undetectable in IHC ( Figures 5D-L and 6B).

Muscle Cells, Nerve Fibres, Adipocytes and Connective Tissue
GAL 2 -R was detected in muscle cells and nerve fibres in cholestasis, as well as in adipocytes, but predominantly in PIT without cholestasis (no significance; Figure 6A). In connective tissues and fibroblasts, immunoreactivity for GAL 2 -R was detected in the control tissue, as well as in PIT with and without cholestasis ( Figure 6A).   Tables S1 and S2, and  score values in Supplementary Table S5.

Galanin 3 Receptor-GAL 3 -R
GAL 3 -R showed very specific expression in the biliary tissues (Figure 7  and Supplementary Table S6) restricted to the cell membrane. The score values of GAL 2 -R expression are provided in Supplementary Table S6. The localisation of GAL 3 -R expression, as well as the results for Control, PIT+C and PIT-C, including all the comparisons made and the p values, are summarised in Table 3.

Cholangiocytes
In the healthy control tissue, GAL2-R exhibited low immunoreact cytes, lining the bile duct lumen ( Figure 5A). Similar to GAL1-R, immu stricted to the cell membrane, apical towards the duct lumen. In PIT, th expression of GAL2-R tended to be lower in cholestasis, in comparison t ( Figures 5B,C and 6A).     Supplementary Tables S1 and S2, and score  values in Supplementary Table S6.

Cholangiocytes
The cholangiocellular expression of GAL3-R was present in the control and PIT, with low immunoreactivity in all groups ( Figure 7A-C). In comparison to the control tissues, GAL3-R expression showed a trend towards lower levels in PIT-C and similar levels to the controls in PIT+C ( Figure 8A).

Arteries, Veins and Capillaries
Endothelial cells lining arteries and veins exhibited no immunoreactivity ( Figures 7D-I and 8B). In capillaries, GAL 3 -R-positive endothelial cells were observed in the healthy and PIT samples, and were localised in the mucosal and muscular bile duct layers ( Figure 7J-L). In comparison to the control, GAL 3 -R expression levels were reduced, on average, by 81% (p = 0.0030) in PIT-C and by 68% (p = 0.0326) in PIT+C compared to the healthy controls ( Figure 8B). Capillaries localised in the tunica adventitia were negative for GAL 3 -R.

Muscle Cells, Nerve Fibres, Adipocytes and Connective Tissue
Myocytes, neurons, adipocytes and fibroblasts did not show any immunoreactivity for GAL 3 -R expression ( Figure 8A).

The Galanin System in pCCA
To monitor the GAL system in pCCA, sections of pCCA were analysed via IHC. For this analysis, sections of PIT, PIT-C and PIT+C were merged. Since the bile duct wall tissues were destroyed due to tumour infiltration, the evaluation of expression was performed on tumour cells only and compared to the healthy control and corresponding PITs (Figures 9 and 10 and Supplementary Table S7). Localisation of the expression of the GAL system, as well as the results for Control, PIT and pCCA, including all the comparisons made and the p values, are summarised in Table 4.

Galanin
GAL was highly expressed in tumour cells of pCCA ( Figure 9C). Therefore, peptide expression was comparable to healthy control samples with >95% GAL-positive cells (Figures 9A and 10A). Between PIT and CCA tissue, which were derived from the same patient, no distinct alterations in immunoreactivity were observed (Figures 9B and 10). GAL expression was detected in the plasma membrane and in the cytoplasm in all study groups (Table 4).

GAL1-R
Expression of GAL1-R was observed in healthy control, peritumoural and pCCA tissues ( Figure 9D-F). The expression of GAL1-R was similar between control and cancer tissues (score values: healthy control = 40 ± 11, pCCA = 37 ± 6) but doubled in PIT ( Figure  10B). In comparison to pCCA, GAL1-R expression was significantly higher (p = 0.0459) in PIT (score values: PIT = 75 ± 13). GAL1-R expression was detected in the plasma membrane and cytoplasm in all study groups (Table 4).

GAL2-R
Weak GAL2-R expression was detected in healthy control tissue, with slightly lower expression in peritumoural and cancer tissues ( Figures 9G-I and 10C). GAL2-R expression in PIT and pCCA samples also revealed no significant differences. GAL2-R expression was detected in the plasma membrane in all study groups (Table 4).

GAL3-R
GAL3-R was found only in cholangiocytes of the healthy control tissue, with weak immunoreactivity localised apical to the cell membrane ( Figure 9J). The immunoreactivity of GAL3-R was barely notable in cholangiocytes of PIT and tumour cells of pCCA tissues ( Figure 9K-L). Compared to the healthy controls, GAL3-R expression in the pCCA tissue was not only significantly reduced but almost absent in tumour cells of pCCA (p = 0.0120, Figure 10D). Where present, GAL3-R expression was detected in the plasma membrane in all study groups (Table 4).

Galanin
GAL was highly expressed in tumour cells of pCCA ( Figure 9C). Therefore, peptide expression was comparable to healthy control samples with >95% GAL-positive cells ( Figures 9A and 10A). Between PIT and CCA tissue, which were derived from the same patient, no distinct alterations in immunoreactivity were observed (Figures 9B and 10). GAL expression was detected in the plasma membrane and in the cytoplasm in all study groups (Table 4).

GAL 1 -R
Expression of GAL 1 -R was observed in healthy control, peritumoural and pCCA tissues ( Figure 9D-F). The expression of GAL 1 -R was similar between control and cancer tissues (score values: healthy control = 40 ± 11, pCCA = 37 ± 6) but doubled in PIT ( Figure 10B). In comparison to pCCA, GAL 1 -R expression was significantly higher (p = 0.0459) in PIT (score values: PIT = 75 ± 13). GAL 1 -R expression was detected in the plasma membrane and cytoplasm in all study groups (Table 4).

GAL 2 -R
Weak GAL 2 -R expression was detected in healthy control tissue, with slightly lower expression in peritumoural and cancer tissues ( Figures 9G-I and 10C). GAL 2 -R expression in PIT and pCCA samples also revealed no significant differences. GAL 2 -R expression was detected in the plasma membrane in all study groups (Table 4).

GAL 3 -R
GAL 3 -R was found only in cholangiocytes of the healthy control tissue, with weak immunoreactivity localised apical to the cell membrane ( Figure 9J). The immunoreactivity of GAL3-R was barely notable in cholangiocytes of PIT and tumour cells of pCCA tissues ( Figure 9K,L). Compared to the healthy controls, GAL 3 -R expression in the pCCA tissue was not only significantly reduced but almost absent in tumour cells of pCCA (p = 0.0120, Figure 10D). Where present, GAL 3 -R expression was detected in the plasma membrane in all study groups (Table 4).

Correlation of Galanin and Receptor Expression with Clinical Features
Cases were divided into those with high and low expression of GAL or GAL 1-3 -R tumour cells, respectively. Cases of high expression were defined as having staining intensities above the mean expression in tumour cells of pCCA. Kaplan-Meier survival analysis revealed in a small patient cohort (n = 18), where survival data were available, that high-intensity GAL staining was associated with patient survival (p = 0.0286, Figure 11A). This is supported by the finding that GAL expression tended to inversely correlate with tumour grading ( Figure 11B). In contrast, GAL 3 -R exhibited an inverse correlation with survival rates (p = 0.0182, Figure 11C). The comparison of mean GAL 3 -R expression in this small cohort among tumour grades did not show any significant difference ( Figure 11D). GAL 1 -R and GAL 2 -R expression correlated with neither patient survival nor tumour grading in the respective cohorts (Supplementary Figure S3).

Correlation of Galanin and Receptor Expression with Clinical Features
Cases were divided into those with high and low expression of GAL or GAL1-3-R in cholangiocytes, respectively. Cases of high expression were defined as having staining intensities above the mean expression in malignant cholangiocytes. Kaplan-Meier survival analysis revealed in a small patient cohort (n = 18), where survival data were available, that high-intensity GAL staining was associated with patient survival (p = 0.0286, Figure 11A). This is supported by the finding that GAL expression tended to inversely correlate with tumour grading ( Figure 11B). In contrast, GAL3-R exhibited an inverse correlation with survival rates (p = 0.0182, Figure 11C). The comparison of mean GAL3-R expression in this small cohort among tumour grades did not show any significant difference ( Figure 11D). GAL1-R and GAL2-R expression correlated with neither patient survival nor tumour grading in the respective cohorts (Supplementary Figure S3).

Discussion
GAL was previously described in rodent cholangiocytes to be associated with biliary cell proliferation [42,43]. In human cholangiocytes, we observed high levels of GAL expression. Unlike McMillin et al. [42], we did not observe an increase in cholangiocellular GAL expression in cholestasis. In the mentioned study, the proliferative effect of GAL in rodents was mediated by the activation of GAL 1 -R, which represented the near-exclusively expressed GAL receptor in cholangiocytes [42,43]. We observed that all GAL receptors were expressed by cholangiocytes in healthy human tissue, in particular, GAL 1 -R. However, the expression patterns of receptors changed in cholestatic tissue, with slight downregulation of GAL 2 -R and a concurrent increase in GAL 1 -R, making GAL 1 -R the major receptor in cholangiocytes in cholestasis. Although this upregulation of GAL 1 -R was not statistically significant, such a switch in receptor expression could lead to enhanced cholangiocellular proliferation through activation of the mitogenic ERK1/2 signalling pathway [42]. With the simultaneous production of GAL and GAL receptor expression, our results further support the assertion of GAL signalling operating via an autocrine mechanism [42,43]. Furthermore, we found GAL-producing nerve fibres in biliary nerve fascicles, as well as in smooth muscle layers embedded in the connective tissues of the bile duct wall. Similarly, GAL was previously observed in the nerve terminals of the smooth muscles and submucosal ganglia of the intestine, and was shown to regulate bowel motility [48,49]. This leads to the suggestion that GAL might be involved in the regulation of bile duct motility.
Regarding the distribution of GAL 3 -R in the biliary tree, we found low, but specific, expression in the cholangiocytes and endothelial cells of capillaries. It is assumed that GAL 3 -R abates immune responses in colitis [50]. In PIT (with or without cholestasis), we detected significantly lower expression of GAL 3 -R in endothelial cells in capillaries in comparison to healthy tissue. Remarkably, this was also observed in tumour cells of pCCA in comparison to the controls, suggesting that the suppression of GAL 3 -R may play a role in CCA development. The immunoexpression of GAL and GAL 1-3 -R in healthy controls and PIT was restricted to the plasma membrane and the cytoplasm of cholangiocytes; epithelial cells of the mucosa, muscularis, adventitia, smooth muscle cells and nerve bundles; and endothelial cells of arteries, veins and capillaries, as well as of adipocytes. Similar findings were previously reported for tumour samples of colorectal carcinoma patients [31]. In CCA, however, we only evaluated tumour cells but obtained similar results for this tissue type.
GAL and GAL-Rs play diverse roles in tumour development and proliferation [51]. For example, in squamous cell carcinoma, high GAL levels are associated with poor survival outcome and cancerogenic immune escape, as well as accelerated cancer progression [46,52]. In gastric cancer, however, studies associate GAL with anti-tumour effects, as the downregulation of GAL correlates with metastasis and cancer progression [53]. In the present study, we found a trend towards an inverse correlation of GAL expression in the cholangiocytes of pCCA with tumour grading. Moreover, it is reported that the hypermethylation of GAL promotor regions epigenetically silences GAL in gastric cancer cells, negatively correlating with tumour size and tumour-suppressive properties [54]. In healthy tissues, distal to the gastric tumour, neuronal secreted GAL was reported to enhance cell survival and protect benign cells from caspase-mediated cell death [55].
To the best of our knowledge, our study is the first to determine the GAL system in pCCA, which, histologically, is an adenocarcinoma derived from dysplastic cholangiocytes [37][38][39][40]. The tumour breaches the bile duct wall and grows into the surrounding tissue, forming irregular gland-like structures [56,57]. In our study, tumour cells of pCCA demonstrated high GAL immunoreactivity, with similar mean expression in comparison to healthy cholangiocytes pCCA patients with available survival rates (n = 18) were divided into two groups; high GAL expressors and low GAL expressors. Remarkably, survival curves revealed significant differences in patient survival in this small cohort. Moreover, GAL expression showed a trend of inverse correlation with tumour grading, with high GAL expression in differentiated Grade 1 tumours and consecutively declining GAL expression with tumour dedifferentiation.
Regarding GAL receptors, GAL 1 -R was previously shown to mediate GAL-induced cholangiocellular hyperplasia during cholestasis [42,43]. Other studies assumed that GAL 1 -R signalling inhibits cell cycle progression [51]. In our study, GAL 1 -R was not upregulated in pCCA, but rather, it was higher in benign PIT, indicating that GAL 1 -R might cause antiproliferative effects in CCA. Nevertheless, in the biliary vascular system of peritumoural tissues, we found significantly decreased GAL and GAL 3 -R expression in endothelial cells. In dermal inflammation, GAL is associated with vasoconstrictive effects, mediated by GAL 3 -R [44,58,59]. Therefore, reduced GAL and GAL 3 -R expression in biliary blood vessels might indirectly correlate with vasodilative effects and increased vascularisation in PIT. Furthermore, angiogenesis and advanced blood supply are two hallmarks of cancer [60]. Consequently, low levels of GAL in tumour tissues could lead to enhanced vascularisation, which can still be seen in neighbouring benign tissue. GAL 2 -R was also slightly reduced in tumour cells of pCCA. GAL 2 -R is associated with mitogenic and apoptotic signalling pathways [52,61]. It is suggested to initiate cell proliferation by activating ERK1/2 and AKT signalling [52]. ERK1/2 is also part of a cell signalling pathway in CCA that enhances proliferation and cell survival [2,3,8,42,62]. Moreover, a reduced GAL 2 -R mRNA level due to promoter methylation was observed in colorectal, prostate and breast cancer [35]. Therefore, the downregulation of GAL 2 -R in CCA could protect malignant cells from apoptotic signals and promote tumour development.
Low expression of GAL 3 -R was previously reported to correlate with enhanced metastasis and lymph node invasion in colorectal cancer [31]. We detected significantly lower GAL 3 -R expression levels in pCCA compared to healthy controls. Future studies are now warranted to validate this finding in larger patient cohorts. Detailed knowledge of the functions of the GAL system in CCA could open new avenues for the development of therapeutic strategies. In high GAL 3 -R expression in CCA tissues, GAL 3 -R-selective antagonists could be used as novel therapeutic agents. Another therapeutic option could be the use of radiolabelled cytotoxic agents linked to GAL 3 -R antagonists.
A limitation of this study is that the results were solely obtained via immunohistochemistry, as we intended to provide a first overview of the expression pattern of the GAL system in the healthy human biliary tract, cholestasis and CCA. A group with preneoplastic lesions was not included, but will be included in further studies comprised of larger patient cohorts.

Conclusions
To our knowledge, this is the first detailed characterisation of the expression pattern of the GAL system in the healthy human biliary duct. An additional analysis of cholangiocytes in cholestasis and pCCA revealed alterations in the expression of the GAL system that mainly affected the receptors. Remarkably, PIT and pCCA tissues primarily exhibited the downregulation of GAL and receptors compared to the controls. For PIT, the endothelial cells of blood vessels were primarily affected, and for pCCA, tissue alterations were found in cholangiocytes for GAL 3 -R ( Figure 12). Our preliminary study on a small patient cohort revealed a potential direct correlation of GAL expression and an indirect correlation of GAL 3 -R expression with patient survival, which needs confirmation in a bigger sample cohort.  Figure S3). Supplementary Tables S1-S7: Clinical parameters of study groups (healthy controls: Supplementary Table S1 and PIT Supplementary Table S2) and IHC score values of GAL1-3-R stainings for all study groups (Supplementary Tables S3-S7  and PIT. ↑ upregulation and ↓ downregulation compared to control; one arrow represents a relative difference of ≥25%, two arrows ≥ 50%, and three arrows ≥ 75%.  Figure S1) and positive controls (Supplementary Figure S2) included in IHC and Kaplan Meier curves of GAL 1 -R and GAL 2 -R (Supplementary Figure S3). Supplementary Tables S1-S7: Clinical parameters of study groups (healthy controls: Supplementary Table S1 and PIT Supplementary Table S2) and IHC score values of GAL 1-3 -R stainings for all study groups (Supplementary Tables S3-S7