Tyrosine phosphorylation of YAP‐1 in biliary epithelial cells mediates posthepatectomy liver regeneration and is affected by serotonin

Experimental data suggested activation of yes‐associated protein (YAP‐1) as a critical regulator of liver regeneration (LR). Serotonin (5‐HT) promotes LR in rodent models and has been proposed to act via YAP‐1. How 5‐HT affects LR is incompletely understood. A possible mechanism how 5‐HT affects human LR was explored. Sixty‐one patients were included. Tissue samples prior and 2 h after induction of LR were collected. Circulating levels of 5‐HT and osteopontin (OPN) were assessed. YAP‐1, its phosphorylation states, cytokeratin 19 (CK‐19) and OPN were assessed using immunofluorescence. A mouse model of biliary epithelial cells (BECs) specific deletion of YAP/TAZ was developed. YAP‐1 increased as early as 2 h after induction of LR (p = 0.025) predominantly in BECs. BEC specific deletion of YAP/TAZ reduced LR after 70% partial hepatectomy in mice (Ki67%, p < 0.001). SSRI treatment, depleting intra‐platelet 5‐HT, abolished YAP‐1 and OPN induction upon LR. Portal vein 5‐HT levels correlated with intrahepatic YAP‐1 expression upon LR (R = 0.703, p = 0.035). OPN colocalized with YAP‐1 in BECs and its circulating levels increased in the liver vein 2 h after induction of LR (p = 0.017). In the context of LR tyrosine‐phosphorylated YAP‐1 significantly increased (p = 0.042). Stimulating BECs with 5‐HT resulted in increased YAP‐1 activation via tyrosine‐phosphorylation and subsequently increased OPN expression. BECs YAP‐1 appears to be critical for LR in mice and humans. Our evidence suggests that 5‐HT, at least in part, exerts its pro‐regenerative effects via YAP‐1 tyrosine‐phosphorylation in BECs and subsequent OPN‐dependent paracrine immunomodulation.


| INTRODUCTION
Compelling evidence suggests a central role of circulating factors and paracrine signaling in postoperative liver regeneration (LR). Given the step-wise progression of liver regeneration, the process has been divided into priming, proliferation, and termination phases. Manipulations of mechanisms involved during the priming phase have been documented to significantly affect LR. Accordingly, processes involved in induction of LR are critical for progression of normal liver regeneration. In this context, platelets have been documented to elicit a positive effect on LR after partial hepatectomy in mice. [1][2][3][4][5][6][7] Interestingly, immediate platelet accumulation within the regenerating liver was documented both in rodents and during human liver regeneration, while an impairment of platelet accumulation was associated with reduced LR in mice undergoing partial hepatectomy. However, this acute accumulation of platelets appears to be transient, as it was found to be resolved by 4 h post partial hepatectomy. 8 The inhibition of platelet aggregation by clopidogrel has previously been shown to prevent hepatocyte proliferation after 70% partial hepatectomy in mice, further underlining a critical role of platelets in LR. 9 Furthermore, several clinical studies found that low circulating platelet counts were associated with poor clinical outcome after liver resection. [10][11][12] Indeed, several of the relevant growth factors for liver regeneration are stored in platelets. 13,14 In particular, serotonin (5-HT), a mediator predominantly stored in platelet dense granules, is a potent promoter of LR, and Tryptophahydroxylase-1 (TPH1) knock-out mice with low 5-HT levels suffer from delayed LR. 5 We could recently document that intra-platelet (IP) 5-HT content correlates with clinical outcome. 3,4,15 In particular, low levels of IP 5-HT were observed in patients who develop postoperative complications and postoperative liver dysfunction (LD) as a surrogate for insufficient postoperative LR.
As 5-HT packing into platelets is negatively affected by 5-HT reuptake inhibitors such as selective 5-HT or 5-HT and noradrenalin reuptake inhibitors (SSRI/SNRI), we recently evaluated the outcome of patients using this type of drugs at the time of liver surgery and noted an increased risk for postoperative LD in this cohort. 16 These data further support the relevance of 5-HT in human postoperative LR. While the positive effect of 5-HT on LR is well-established, the mechanisms responsible for 5-HT-mediated LR still need to be elucidated. The existing evidence in TPH1 −/− mice suggests that 5-HT might elicit its pro-regenerative effects via activation of yes-associated protein 1 (YAP-1) signaling. 17 In particular, impaired LR coincided with a significant reduction in YAP-1 expression after partial hepatectomy in TPH −/− mice. 17 Of note, YAP-1 activation has been shown to be centrally involved in liver disease and regeneration. 18 In fact, inhibition of MST1 and 2, known upstream regulators of YAP-1 activation, have been found to significantly affect LR in mice. 19 Further, a selective MST1/2 inhibitor (XMU-MP-1) also effectively induced liver regeneration via induction of YAP-1. 20 However, a limited number of reports failed to document a significant reduction of LR after inhibition of YAP-1 signaling in mice. [21][22][23] In this context, recent evidence demonstrated that in a carbon tetrachloride toxic liver injury model, biliary epithelial cells (BECs) as well as hepatocytes increase YAP-1/TAZ activity during LR after toxic injury. However, a conditional YAP-1/TAZ knock-out in hepatocytes did not affect liver regeneration in their model, suggesting that hepatocyte-specific YAP-1 signaling might be dispensable for LR. 23 In clear contrast, conditional YAP-1/TAZ knock-out in BECs led to a significant decrease in LR after toxic injury. 23 The authors further documented that the effect of YAP-1 activation in BECs was associated with osteopontin (OPN) release, which mediated inflammatory responses during liver injury. 23 It is important to note that YAP-1 can be activated by mechanisms independent of the classical cytosolic retention pathway of the HIPPO pathway, mediated via serine phosphorylation and subsequent cytosolic degradation. 5-HT receptors mediate their effects via G protein-coupled receptor signaling. Of interest, G protein-coupled receptors induce tyrosine YAP-1 phosphorylation followed by nuclear retention of YAP-1 and continuous activation. [24][25][26] Despite extensive explorations of these processes in rodents after partial hepatectomy, the relevance of YAP-1 signaling during human LR remains incompletely understood. Accordingly, within this project, we explored the relevance of YAP-1 signaling during the priming phase of human liver regeneration via analysis of selected tissue and blood samples from patients undergoing liver resection and further confirmed the relevance of BECs YAP-1 activation using a conditional murine knockout model. We unraveled that 1) activating tyrosine phosphorylation of YAP-1 increases rapidly after induction of LR and this occurs predominantly in BECs, 2) conditional knockout of Yap/Taz in BECs significantly decreases liver regeneration after partial hepatectomy in mice, 3) 5-HT induces tyrosine phosphorylation of YAP-1 in BECs resulting in increased OPN expression, and 4) circulating OPN rapidly increases after liver resection in humans and is associated with inflammatory cytokines. In conjunction with experimental evidence, our data suggest that 5-HT elicits its pro-regenerative effects in part via tyrosine phosphorylation of YAP-1 in BECs, leading to increased production and release of OPN which ultimately mediates inflammatory responses within the human liver, known to be critical inducers of LR.

| METHODS
A total of 61 patients from two different institutions, comprising the General Hospital of Vienna [Austria]/ Medical University of Vienna [Austria] and the Clinic Favoriten [Austria] were included in this study. Perioperatively, peripheral blood samples were assessed 1 day before surgery (PREOP), as well as one (POD1) and 5 days after liver resection (POD5). Additionally, blood samples were collected before liver resection in the portal vein and 2 h after induction of LR in the liver vein (draining the regenerating liver lobe). Baseline characteristics, perioperative laboratory parameters, surgical procedure, and baseline liver pathology were assessed and recorded (Table S1). Brisbane-2000 nomenclature was used to define major resections (≥3 segments) and only major resections were included in this analysis. 27 Furthermore, in a subset of 25 patients, intraoperative tissue samples were obtained. We collected samples at baseline (before parenchymal transection) and after induction of LR (2 h after ligation of either the left or right portal branch as standard approach for a left or right lobectomy) to evaluate alterations in the priming phase of LR. The 2 h liver tissue sample was collected from the regenerating liver lobe that remained in the patients, not the lobe that was subsequently resected. The study was conducted in adherence to the Declaration of Helsinki and was approved by the responsible Institutional Ethics Committee. Ahead of participation, informed consent was obtained from all patients (EK Nr. 16-253-0117, EK 14-122-0714).

| Optimized blood sample preparation
We have previously demonstrated that conventional plasma preparation frequently suffers from in vitro platelet activation. 28 Accordingly, an optimized plasma preparation technique was applied as previously described by us. 3,4,14,[29][30][31] Briefly, blood was drawn into prechilled tubes containing citrate, theophylline, adenosine, and dipyridamole, it was immediately placed on ice and further processed within 30 min. After an initial centrifugation step at 1000 g and 4°C for 10 min, the plasma supernatant was subjected to further centrifugation at 10 000 g and 4°C for 10 min (to remove remaining platelets). The supernatant was stored in aliquots at −80°C.
Serum samples were retrieved by blood collection without the addition of anticoagulants and by centrifugation (at 1000 g and room temperature for 10 min) 30 min after collection. The supernatant was stored in aliquots at −80°C.
2.2 | Cholangiocyte culture/5-HT stimulation/expression of downstream target assessment The normal human cholangiocyte (NHC) cell line was maintained as previously described in supplemented Dulbecco's modified Eagle's medium/F12 with 10% fetal bovine serum. NHC have grown to approximately 60% confluency, and then serum starved for 24 h before treatment with serotonin or dimethyl sulfoxide control for 6 h.
RNA was purified from NHC whole cell lysate, and cDNA synthesized with random primers. Target genes were quantified in triplicate with real-time polymerase chain reaction (Light Cycler 480 II, Roche Diagnostics) and SYBR Green (Roche Diagnostics). 18S was utilized as the housekeeping gene, and relative quantification was performed using the 2 −ΔΔCT .

| Quantification of the intraplatelet 5-HT pool
Plasma and serum samples as well as platelet extracts were analyzed by commercially available enzyme-linked immunosorbent assay (ELISA) tests for human 5-HT (IBL, 5-HT ELISA) OPN, interleukin (IL)-6, IL-8, and CTFG according to the manufacturer's instructions. To calculate the actual intra-platelet 5-HT pool, plasma 5-HT levels, reflecting the actual circulating amount of 5-HT, were subtracted from serum 5-HT levels, which contain all 5-HT released by platelet activation during blood clotting. Perioperative parameters of hepatic cell death (AST and ALT) were measured in serum samples by routine laboratory blood tests.

| Immunoblot
NHC whole cell lysates were collected in the presence of phenylmethylsulfonyl fluoride, and protease/phosphatase inhibitors sodium dodecyl sulfate-polyacrylamide gel electrophoresis was performed, and proteins were transferred to nitrocellulose membranes. Membranes were incubated in primary antibody overnight at 1:1000 dilution. Horseradish peroxidase conjugated secondary antibody was applied 1:5000 for 1 h at room temperature. Immunoblots were visualized using enhanced chemiluminescence (ECL) (GE Healthcare Life Sciences). Proteins were quantified by densitometry using ImageJ, normalized to actin, and fold change was calculated relative to the vehicle.

| Immunofluorescence staining
After formaldehyde fixation, liver specimens were embedded in paraffin and cut into 10 µm slices for further immunofluorescence staining. Ahead of the staining procedure tissue section had to be deparaffinized and rehydrated. Tissue samples were 10 min exposed to 60°C and two-times incubated in 100% Xylol immersion for 5 min. Per reverse serial dilution, stepwise rehydration was carried out using decreasing ethanol concentrations (96%, 70%, and 50%) and ddH2O. For antigen retrieval, tissue slides were heated in citrate buffer (pH 6) and before further processing, allowed to cool down for 30 min. Consequently, to improve the efficiency of antibody penetration, the slides were immersed for 10 min in 1×TBS + 0.1% Triton X-100 and afterward washed three times in a 1×TBS + 0.1% Tween®20 buffer. To ensure an efficient reagent application and wetting of the entire sample, the tissue slides were encircled with a DAKO pen upon further processing. Then blocking solution (10% FCS, 1% BSA, 0.05% PBS-Triton, and PBS) was applied to minimize background staining, followed by a 2-h incubation at room temperature in the dark. To prevent drying out slides were stored in a humidity chamber. The excess fluid was removed before the next step, the antibody application. Since an indirect staining method was used, two antibodies are required for staining, first, an unlabeled primary and, subsequently, a secondary fluorophorelabeled. Accordingly, the first step was an overnight incubation with the diluted primary antibody (as described in Table S2A), 100 µL per tissue section. Slides were stored overnight at 4°C, again in the dark, using a humidity chamber. Afterward, the excess fluid was removed and slides were three times washed in a 1×TBS + 0.1%Tween®20 buffer. The secondary antibodies were applied in a dilution according to Table S2B. For incubation, everything was stored for 1.5-2 h in the humidity chamber at room temperature and in the dark. To prevent aggregate formation the secondary antibody has to be centrifuged for 2 min at 14.1rcf antecedently. Before nucleus counterstaining, 3 washing steps in a 1×TBS + 0.1%Tween®20 buffer were carried out, each for 10 min. Subsequently, tissue samples were incubated with 5 µg/ml Hoechst 33342 for 8 min at room temperature in the dark using a humidity staining chamber (nuclei staining). Finally, three washing steps were done (1×TBS + 0.1%Tween®20 buffer) for 10 min each, before the slides were mounted with ProlongGold Antifade mounting reagent and coverslips. The edges were additionally sealed with nail polish. The mounted slides were stored at 4°C in the dark until further use to prevent the fluorescence signal's attenuation.
Microscope images were obtained via Widefield Fluorescence Nikon A1plus Ti Microscope. Suitable size scales were added afterward in ImageJ Fiji, as well as the adjustment of brightness and contrast. Data analysis and mean fluorescence intensity measurements of the previously obtained images were performed via CellProfiler 3.1.9.

| Partial hepatectomy model
Male age-matched mice were randomly assigned to control or treatment. Surgical procedures were performed in small cohorts to ensure uniform timing in relation to the day:night cycle. Mice were anesthetized by vaporized isoflurane for an average operation time of 15 min. Two-thirds partial hepatectomy was conducted as previously described. 32 Briefly, Cholecystectomy followed by sequential ligation and excision of the left median, right median, and left lateral lobes was performed. Hemostasis was achieved. The abdomen was then closed in two layers with running 4-0 Vicryl. The resected tissue was collected for further molecular analysis at baseline. Liver regeneration was assessed 40-, 72-, and 120 h posthepatectomy. All excised tissue was either frozen and stored at −80°C or fixed in 10% (v/v) buffered formalin overnight at room temperature.

| YAP/TAZ cholangiocyte deletion in vivo
Yap/Taz double floxed mice (Yap fl/fl /Taz fl/fl ) (Strain #030532) and CK19 CreERT (Strain #026925) were obtained from Jackson Laboratory. Mice were crossbred to obtain homozygous Yap fl/fl /Taz fl/fl -CK19 CreERT genetically modified mice. Cre recombinase expression was induced by administering tamoxifen (4 mg) intraperitoneally daily for 3 days, starting 7 days before partial hepatectomy. Control mice received equal amounts of tamoxifen. The study design is illustrated in Figure 1D.
All animal experiments were approved by the institutional review board (A00004631-19-R21).

| Immunohistochemistry
Paraffin sections were cut at a thickness of 5 µm. Tissues were stained for histological analysis with primary antibodies listed in Table S2C. For quantification of Ki67 positive hepatocytes, 10 high-power fields (HPF) (200× magnification) were visualized per sample and manually counted.

| Statistical analyses
IBM SPSS software v 20.0 (IBM Corp.) and GraphPad Prism 5 (GraphPad Software) were utilized for statistical data analyses, which was based on nonparametric testing for either paired or independent samples (Mann-Whitney U test, Wilcoxon test, Spearman-Rho correlation or Pearson correlation analysis). Survival curve analysis was performed using Log-rank (Mantel-Cox) test. p values < 0.05 were considered statistically significant.

| Immediately after induction of liver regeneration in humans total YAP-1 increases predominantly in BECs
Given extensive experimental evidence suggesting a central role for YAP-1 in LR, we initially assessed intrahepatic YAP-1 levels and dynamics in patients undergoing major liver resection. Baseline YAP-1 expression was primarily found in ductular structures in the resting liver, as evaluated via co-localization of YAP-1 (marker reference 33 ) and CK19 (marker reference 34 ) upon immunofluorescence imaging ( Figure 1). We then assessed tissue from the same patients obtained 2 h after induction of regeneration (study design illustrated in Figure 1A). A significant increase in YAP-1 predominantly in CK19 + BECs was observed after induction of LR (quantification is given in Figure 1B and representative pictures are illustrated in Figure 1C for baseline and regeneration). Furthermore, the BEC-specific marker CK19 colocalized, suggesting a predominant increase of YAP-1 in BECs upon the priming phase of liver regeneration in humans.
F I G U R E 1 Total YAP-1 increases predominantly in BECs immediately after induction of liver regeneration in humans. Sample collection study design for human liver regeneration tissue is given in (A), including a description of assessed parameters of the respective samples (tissue and blood). All patients underwent major hepatic resection, with either left or right hepatectomy. Details of patients characteristics can be found in Table S1. Quantification YAP dynamics of 25 paired samples is illustrated in (B). A representative sample of an immunofluorescence staining of YAP-1, CK19, and DAPI as well as merged images are given for human liver tissue at baseline as well as  To test if isolated depletion of YAP/TAZ in BECs would be sufficient to affect LR after hepatic resection, a conditional CK19 YAP/TAZ knockout was generated (YAP fl/fl ; TAZ fl/fl -CK19CreERT) on a C57BL/6 background (N = 6/7 per group) at 10 weeks of age. The experimental design is illustrated in Figure 1D. As expected YAPfl/fl; TAZfl/fl did not show a significant phenotype ( Figure 1E-G). However, YAPfl /fl ; TAZ fl/fl -CK19 CreERT did demonstrate a significant reduction in liver regeneration. In particular, we found Ki67, a marker of cell proliferation, strikingly reduced at 40 h after partial hepatectomy in YAP fl/fl ; TAZ fl/fl -CK19 CreERT (N = 4, Figure 1E, p < 0.001). This also translated into a significant reduction of liver to bodyweight ratio at 72 h after partial hepatectomy (N = 5-7, Figure 1F, p < 0.01) as well as high mortality in the YAP fl/fl ; TAZ fl/fl -CK19 CreERT group as compared to none in YAP fl/fl ; TAZ fl/fl mice (N = 6-7, Figure 1G, p = 0.06).
3.3 | IP 5-HT correlates with intrahepatic YAP-1 expression and in vitro 5-HT stimulation of human BECs results in YAP-1 downstream target transcription Previous evidence suggests a direct effect of 5-HT on YAP-1 expression, potentially mediating the proregenerative effect of 5-HT during LR. 17 To test this hypothesis in humans, we assessed intraplatelet (IP) 5-HT levels and the correlation with intrahepatic YAP-1 expression. Samples were collected both from the peripheral circulation as well as the porto-mesenteric venous system supplying the liver. We observed that preoperative IP 5-HT levels in peripheral blood correlated with baseline YAP-1 expression in the liver (r = 0.0712, p = 0.031, Figure 2A). The correlation with hepatic YAP-1 expression was even stronger when evaluating IP 5-HT levels in the portal vein (r = 0.812, p = 0.014, Figure 2B). Similarly, circulating IP 5-HT levels correlated with YAP-1 expression after induction of LR (peripheral IP 5-HT levels: r = 0.702, p = 0.035, Figure 2C; portal venous IP 5-HT levels: r = 0.703,   Figure 2D), indicating a possible association of YAP-1 expression with the biologically available 5-HT pool.
As we previously demonstrated that SSRI/SNRI treatment effectively reduces IP 5-HT levels, we subsequently assessed YAP-1 expression dynamics in patients with perioperative SSRI/SNRI use. Importantly, we observed that patients using SSRI/SNRI failed to increase YAP-1 expression after liver resection (p = 0.017, Figure 2E). To further confirm the association of 5-HT with YAP-1 activation in human BECs, we assessed YAP-1 well-established downstream transcriptional targets 35 in cultured normal human cholangiocytes (NHC) following stimulation with 5-HT. In line with previous results, we found a significant induction of YAP-1 downstream targets (connective tissue growth factor (CTGF), Cysteine-rich angiogenic inducer 61 (CYR61) and Ankyrin repeat domain-containing protein 1 (ANKRD1). 35 as reference for these targets) upon 5-HT stimulation ( Figure 2F).

| OPN is a YAP-1 downstream target in BECs and correlates with YAP-1 and inflammatory responses during in human LR
As we had observed a direct effect of 5-HT on YAP-1 downstream targets in BECs and documented that YAP-1 increased immediately after induction of LR, we assessed OPN as a YAP-1 downstream target in BECs, given its particular function in LR. We initially documented a significant increase of OPN upon induction of LR specifically in ductular structures ( Figure 3A, p = 0.028, a representative sample is illustrated in Figure S1). We observed that during regeneration YAP-1 and OPN mRNA expression levels correlated significantly ( Figure 3B, r = 0.617, p = 0.001). Even more striking, the fold induction of YAP-1 and OPN expression closely correlated with each other ( Figure 3B, r = 0.902, p < 0.001) suggesting YAP-1-dependent regulation of OPN during human LR, as previously described in experimental models. When comparing patients who did not respond with a YAP-1 increase after induction of LR to patients with YAP-1 induction, we observed a significantly lower increase in OPN levels during the priming phase of LR ( Figure 3C, p = 0.008). We observed a trend towards reduced induction of OPN upon liver regeneration in patients on SSRI treatment, however, this did not reach statistical significance ( Figure 3C, p = 0.150), which is in line with results obtained for YAP-1 in SSRI/SNRI users. OPN is secreted by BECs, where it seems to elicit its effects as a potent chemoattractant. 36 Accordingly, its circulating fraction can be measured in plasma. We, therefore, further evaluated circulating OPN levels in patients undergoing liver resection. We observed a significant increase of circulating OPN levels during this early time point of LR ( Figure 3D, p = 0.017) when comparing preoperative levels to blood from the liver vein, draining the regenerating liver lobe, 2 h after induction of LR. We further noted that OPN levels increased perioperatively with significantly higher levels on POD1 when compared to baseline ( Figure 3D, p = 0.001). However, a decline was observed on POD5 ( Figure 3D, p = 0.049). We then evaluated circulating levels of additional canonical YAP-1 targets (CTGF) and found that patients with high postoperative OPN levels (cutoff 274 ng/mL) also had significantly higher circulating CTGF levels on POD1 ( Figure 3E, p = 0.022).
OPN elicits its effects during LR predominantly through induction of inflammation. 36,37 In particular, knock-out of OPN results in reduced interleukin (IL)-6 and IL-8 levels. 38 Accordingly, we assessed if circulating OPN levels on POD1 correlated with circulating levels of IL-6 and IL-8. Indeed, patients with high postoperative OPN levels (cutoff at 274 ng/mL) also had significantly higher circulating IL-6 and IL-8 levels on POD 1 after liver resection ( Figure 3E, IL-6: p = 0.008; IL-8 p = 0.029).

| OPN primarily localizes in BECs during the early phase of LR
To further confirm interdependence of YAP-1 and OPN, we performed immunofluorescence staining on human liver tissue samples before and after induction of liver regeneration. As illustrated in Figure 4A OPN and YAP-1 intensely colocalized, OPN being primarily observed at the apical side of biliary structures. To document the biliary phenotype of OPN positive cells, we co-stained CK19 and OPN demonstrating biliary origin of OPNpositive cells. A representative example is given for the significant increase of intrahepatic OPN upon induction of LR in Figure 4B.

| DISCUSSION
While LR has been extensively studied in rodent models, there is limited data on these mechanisms in human LR. This is particularly true for YAP-1 signaling, which has been documented multiple times to be a central molecule in murine liver pathophysiology but human validation of these processes is largely missing. Within this study, we provide evidence in human samples that 1) YAP-1 activity increases immediately after induction of LR, occurs predominantly in BECs, and is predominantly tyrosine phosphorylation-dependent, 2) 5-HT is a potent inducer of tyrosine phosphorylation of YAP-1 in BECs resulting in increased OPN expression, and 3) circulating OPN rapidly increases after liver resection and is associated with inflammatory cytokines known to be inducers of hepatic regeneration. Furthermore, we document that the isolated deletion of YAP/TAZ signaling in BECs is sufficient to significantly reduce liver regeneration after partial hepatectomy in murine models. These results suggest a critical role of BEC YAP-1 tyrosine phosphorylation and concomitant OPN induction in mediating the well-defined pro-regenerative effects of 5-HT during human LR. Furthermore, these observations provide a potential explanation of the association of SSRI/SNRI with increased incidence of postoperative LD and morbidity in patients undergoing liver resection. More importantly, as therapeutic interventions on the level of 5-HT to promote LR pose several obstacles, 3 specific manipulations of LR via tyrosine phosphorylation of YAP-1 might represent an attractive new therapeutic target.
The Hippo pathway and its regulation of the downstream effector YAP-1 have been extensively studied in experimental models of LR. 36 While overall, a YAP-1mediated pro-regenerative phenotype has been documented, a significant amount of heterogeneity has been observed when manipulating this pathway. 36 Recent evidence identified a potential explanation for this heterogeneity. Using a liver injury model with selective YAP-1/TAZ knock-out in hepatocytes versus BECs, the authors were able to document that only mice with BECspecific YAP-1/TAZ knock-out suffered from impaired LR after carbon tetrachloride toxic liver injury. 36 In line with these results we were able to document that the isolated deletion of YAP/TAZ in mice undergoing 70% partial hepatectomy did indeed reduce posthepatectomy LR. Furthermore, in humans, we predominantly observed YAP-1 expression and activation in BECs.
F I G U R E 4 Osteopontin primarily localizes in BECs during the early phase of liver regeneration. A representative sample of a immunofluorescence staining of YAP-1, OPN, and Hoechst as well as merged images are given for human liver tissue upon regeneration (A). Co-staining of CK19 and OPN is given in a representative sample at baseline as well as 2 h after induction of liver regeneration in the regenerating lobe (B). BECs, biliary epithelial cells; OPN, osteopontin; YAP-1, yes-associated protein 1.
Given the immediate increase of YAP-1 in BECs and its activating tyrosine phosphorylation in our analyses, BEC-specific YAP-1 signaling might be particularly important during the priming phase of human LR.
Platelets and platelet-derived growth factors promote LR in mice and humans. [40][41][42] They represent one of the first responders after induction of hepatic regeneration as they rapidly accumulate within the liver after hepatic resection. 15 One proposed mechanism underlying platelet-mediated LR is the release of 5-HT from dense granules. [2][3][4] Indeed, approximately 90% of biologically available 5-HT is stored in platelets. While depletion of IP 5-HT is associated with delayed LR after partial hepatectomy 5 downstream effects targeted by 5-HT remain unclear. Previous experimental results suggested that 5-HT might affect YAP-1 signaling during LR. 17,43 In this context, TPH −/− mice show significantly reduced levels of YAP-1 expression during LR with concomitantly reduced regenerative capacity. In line with these results, we observed that portal venous 5-HT levels were associated with intrahepatic YAP-1 expression during LR. As we predominantly observed YAP-1 expression and immediate induction upon LR in BECs, we further assessed the effects of 5-HT on human NHCs. Here, we were able to confirm a significant upregulation of YAP-1 downstream targets in response to 5-HT stimulation. Of interest, the process of 5-HT uptake into platelets is negatively affected by 5-HT reuptake inhibitors such as SSRI and SNRI. [44][45][46] We recently documented that patients undergoing liver resection and with perioperative SSRI/SNRI intake not only display depleted IP 5-HT levels, but also suffered from a significantly increased rate of postoperative LD, suggesting a negative effect of 5-HT modulation on LR in humans. 16 Strikingly, we found that patients undergoing liver resection while on SSRI/ SNRI therapy failed to increase YAP-1 expression and downstream upregulation of OPN upon induction of LR, further supporting the hypothesis that YAP-1signalling might mediate the pro-regenerative effects of 5-HT.
Platelets have been shown to accumulate within the regenerating liver as early as 15 min after partial hepatectomy. However, the influx of platelets during liver regeneration is supposedly transient, as the described accumulation was resolved within 4 h after induction of LR. 8 Importantly, inhibition of platelet accumulation exclusively inhibited LR if performed F I G U R E 5 YAP-1 and its phosphorylation states predominantly occur in bile ducts in the portal triad. A representative sample of a immunofluorescence co-staining of serine phosphorylated YAP-1, tyrosine phosphorylated YAP, total YAP, CK19, and OPN is given zoomed in on a portal triad. 1 marks portal vein, 2 portal artery, 3 portal bile duct. OPN, osteopontin; YAP-1, yes-associated protein 1. before partial hepatectomy, while inhibition after 2 h did not show any effect on the regenerative capacity of the murine liver. In addition, Kono et al. 47 documented that the intrahepatic increase of 5-HT is closely linked to intrahepatic platelet accumulation during LR. Consequently, the intrahepatic bioavailability of plateletderived 5-HT is increased during the priming phase of LR, which further suggests an involvement of 5-HT in YAP-1 activation in BECs. Of note, 5-HT may exert its effect independent from the classical Hippo pathway. While the Hippo pathway culminates in serine phosphorylation of YAP leading to cytosolic retention, G-proteincoupled receptors, such as 5-HT receptors, have been found to mediate their effects via tyrosine phosphorylation of YAP-1 and nuclear retention. 24 Indeed, we observed an immediate increase in YAP-1 tyrosine phosphorylation, in human liver tissue as well as in the human cholangiocyte cell line NHC after 5-HT stimulation, while serine phosphorylation was not affected. This data suggest that modulation of Hippo-specific regulation of YAP-1 might not be the predominant regulatory mechanism associated with YAP-mediated regenerative effects and adds value to the accumulating evidence suggesting a critical role for tyrosine phosphorylation in regulating YAP-1 activity in multiple contexts.
Intriguingly, 5-HT seems to elicit antiproliferative effects in BECs. 48 Accordingly, the mechanisms involved in BEC-specific YAP-1 activation during early LR most likely rely on paracrine signaling. In this context, OPN has been suggested as a soluble protein promoting proregenerative inflammatory responses during the priming phase of LR. 49 In the liver BECs are the predominant source of OPN. 49 OPN has been proposed as a critical paracrine mediator in multiple pathophysiological processes of the liver, besides LR. 37,49 It represents a YAP-1 downstream target with chemoattractant functions mediating the increase of inflammatory cytokine levels such as IL-6 and IL-8, 38,50 which represent critical inducers of LR as they induce transition from the G0 to the G1 phase of the cell cycle in hepatocytes during the priming phase of liver regeneration. 51 OPN knock-out mice showed defective initiation of LR after partial hepatectomy, F I G U R E 6 YAP-1 tyrosine phosphorylation occurs during the priming phase of in human liver regeneration. A representative sample is of an immunofluorescence staining of YAP-1 and tyrosine phosphorylated YAP-1 is given in baseline as well as regenerating liver tissue (A). Quantification of serine phosphorylated YAP-1, tyrosine phosphorylated YAP and total YAP at baseline (PRE) as well as upon regeneration (REG) is given in (B) (N = 26). Western blot and quantification of cultured human BECs upon stimulation with 5-HT (compared with vehicle control) is illustrated in (C) (N = 2). (D) summarizes the proposed mechanism of IP-5HT on hepatic regeneration. Illustrated is that IP-5HT is released from platelets and stimulates YAP-1 tyrosine phosphorylation in BECs, mediating OPN secretion from BECs that mediates an indispensable inflammatory response to promote liver regeneration during the priming phase of hepatic regeneration **p < 0. 01, *p < 0.05. 5-HT, serotonin; BECs, biliary epithelial cells; YAP-1, yes-associated protein 1.
which was mediated via insufficient induction of IL-6 expression and consequent impairment of STAT3 activation. 49 Very similar to these data, we observed an immediate increase of circulating OPN in the liver vein, draining the already regenerating liver lobe, which further increased when measured peripherally on POD 1. OPN closely colocalized with YAP-1 in BECs in the regenerating liver and intrahepatic YAP-1 expression correlated with OPN induction during LR. Importantly, the increase of circulating OPN was accompanied by proinflammatory cytokines, indicating that OPN might indeed mediate its effects via modulation of an early inflammatory response during LR in humans as a critical initiator of regenerative processes during hepatic regeneration. 51 While intrahepatic inflammation during LR is considered to be vital for the induction of LR, a potential negative impact on the physiologic function of the liver in the setting of underlying liver disease has been suggested. 52 Indeed, 5-HT has not only been described as a promoting factor during LR, but also negative effects on intrahepatic pathophysiology were observed. 7 The identified mechanism of 5-HT mediated inflammatory response via YAP-1 activation and OPN induction in BECs might indeed explain some of these observations. While knock-down of 5-HT reduces an initial inflammatory response indispensable during the priming phase of LR, chronic inflammatory responses can lead to injury such as hepatic steatosis or aggravation of viral hepatitis. 52,53 Furthermore, the simple duration of IL-6 exposure was found to critically affect liver injury and repair in mice, with adverse effects of prolonged exposure. Short-term administration over 1-2 days resulted in accelerated LR, while mice exposed over a time course of 5-7 days suffered from impaired regeneration, suggesting that precise regulation of these processes is necessary to allow for adequate regenerative responses after hepatic resection. 54 Accordingly, future research is warranted to elucidate the critical balance of inflammation during liver regeneration in patients with heterogeneous types and extents of underlying liver diseases, as they might adversely affect intrahepatic immune responses. Hence, it is fundamental to further investigate how cytokine responses affect hepatic regeneration in humans in different clinical scenarios of liver disease.
We do need to point out that our study suffers from limitations. Certainly, an inherent limitation of translational research is that we can not perform genetic alternation or treatment models to ultimately prove our observations and that some assessments can only be performed on certain types of samples that have not be available in our biorepository. While we try to include a treatment model (SSRIs) in our in-human analyses to get closer to an "intervention" model in humans, we do need to stress that we could not do this for every part of our manuscript simply because of the nature of translational research. To further counteract this inherent limitation, we tried to further explore our observations on a tamoxifen-inducible YAP/TAZ knockout model as well as cell culture experiments, which helped us to confirm some of our observations. Nonetheless, our translations data, therefore, should only be interpreted on the background of available experimental evidence. Additionally, we do need state that some of our experiments do suffer from a small sample size, which needs to be taken into account when interpreting our results. This is particularly true for results presented in Figure 2A-D. However, we also want to stress that these are in human samples that are certainly very challenging to be obtained and we believe that the observations we made in these patients are certainly informative and important, particularly when evaluated on the background of existing experimental evidence, suggesting similar processes to take place during hepatic resection in rodent models.
Taken together, within this study, we were able to identify a novel mechanism how IP 5-HT, a well-defined promotor of LR, might partially mediate its effects. Generated evidence suggests that IP 5-HT is released during the priming phase of LR, resulting in YAP-1 tyrosine phosphorylation in BECs. This activation increases OPN production as well as secretion, acting as a chemoattractant to allow for an immediate inflammatory response during early LR. Our data suggest a critical relevance of 5-HT mediated YAP-1 signaling in human LR and support further exploration of YAP-1 tyrosine phosphorylation as a potential therapeutic target during the priming phase of LR.

AUTHOR CONTRIBUTIONS
All listed authors have made substantial contributions to the conception and design, and/or acquisition of data, and/or analysis and interpretation of data and participated in drafting the article or revising it critically for important intellectual content as well as given final approval of the version to be published.

ACKNOWLEDGMENTS
The study was funded by the NIH (R01DK122813), the Medical Scientific Fund of the Mayor of the City of Vienna (P-19098), and FWF (P-32064). Open Access Funding by the Austrian Science Fund (FWF) Several illustrations in the manuscript were generated with biorender (https://biorender.com).