With no lysine kinase 1 in hepatic stellate cells promotes mouse and human liver fibrosis through glycolytic triosephosphate isomerase 1

Abstract

Background and Aims:

Liver fibrosis and cirrhosis are major global health burdens with limited treatments. Activated HSCs drive fibrosis through increased proliferation, migration, collagen deposition and glycolysis. The aim of this study is to elucidate the signaling events that lead to enhanced glycolysis, subsequent HSC activation, and liver fibrosis.

Approach and Results:

Utilizing a phospho-kinase array, we identified with no lysine kinase 1 (WNK1) as a new kinase in primary human HSCs activated by the profibrotic PDGF-B. PDGF-mediated glycolysis in HSCs was attenuated by the WNK1 selective inhibitor, WNK-IN-11, as measured by live-cell metabolic assay, Glifon-300 fluorescence, Glucose-Glo assay, and glucose transporter 1 immunofluorescence. To understand how WNK1 promotes glycolysis, we employed chemical genetics where an engineered WNK1 carrying threonine 301 to glycine mutation (WNK1^T301G) mutant with a modified ATP pocket accommodated a specific ATP-γ-S analogue to thiophosphorylate its substrates. Thio-phosphorylated proteins were analyzed by mass spectrometry, which identified the glycolytic protein, triosephosphate isomerase 1 (TPI1), as a novel WNK1 direct substrate. TPI1 interacted with WNK1 in condensates, membrane-less phase separation structures. Condensate disruption and knockdown of TPI1 or WNK1 reduced HSC proliferation and migration in vitro. In vivo, pharmacological or genetic WNK1 inhibition significantly attenuated liver fibrosis in carbon tetrachloride and high-fat, choline-deficient, l-amino acid diet mouse models. Finally, WNK1 inhibition reduced fibrogenesis ex vivo in human precision-cut liver slices.

Conclusions:

These findings identify WNK1 as a previously unrecognized kinase that promotes HSC activation and liver fibrosis via TPI1-mediated glycolysis, highlighting the WNK1-TPI1 axis as a potential therapeutic target.

Graphical Abstract

INTRODUCTION

Advanced liver fibrosis or cirrhosis is a major global health concern, contributing to more than 1.4 million deaths worldwide in 2021.^[1] In the United States, chronic liver diseases and cirrhosis are 1 of the 10 leading causes of death.^[2] At present, liver transplantation remains the only treatment for cirrhosis.^[3] Activation of HSCs is a key event in the progression of liver fibrosis.^[4–6] In response to paracrine fibrogenic cues, most notably PDGF-B and TGFβ1, HSCs become highly proliferative, migratory, contractile, and secrete excessive amounts of collagen, driving pathological extracellular matrix accumulation.^[4,6] Activation of HSCs and subsequent liver fibrosis is fueled by a metabolic shift characterized by increased glucose uptake and subsequent glycolysis.^[7–9] Notably, genetic or pharmacological inhibition of key glycolytic enzymes has been shown to attenuate liver fibrosis in vivo.^[7,8,10] Nevertheless, the signaling cascade leading to this fibrometabolism in HSCs remains largely unknown.

Glycolysis is a catabolic pathway that produces pyruvate from intracellular glucose.^[11] Glycolysis can proceed through the direct conversion of fructose-1,6-bisphosphate to glyceraldehyde-3-phosphate or through the conversion of fructose-1,6-bisphosphate to dihydroxyacetone-phosphate then to glyceraldehyde-3-phosphate, facilitated by the enzyme triosephosphate isomerase 1 (TPI1).^[12,13] The role of TPI1 in HSCs is not explored. Enhanced glycolysis increases glucose uptake in the cell, which is achieved by trafficking glucose transporters (GLUTs) to the plasma membrane.^[14] GLUTs and glucose uptake have been shown to be involved during metabolic dysfunction–associated steatotic liver disease and liver cancer,^[15,16] but not during fibrosis. It has been shown in immortalized cell lines that GLUT translocation to the plasma membrane can be facilitated by the action of kinases, such as with no lysine kinase 1 (WNK1).^[17,18] WNK1 is a serine/threonine protein kinase recognized to be involved in various cancers through phosphorylating its 2 known substrates, SPS1-related proline/alanine-rich kinase (SPAK) and oxidative stress-responsive kinase 1 (OXSR1).^[19] However, the role of WNK1 in fibrosis in general, and more specifically during HSC activation and liver fibrogenesis, is yet to be examined.

In this study, we identify WNK1 as a novel regulator of HSC activation and subsequent liver fibrosis. We demonstrate that PDGF-mediated WNK1 activation promotes glycolysis in HSCs. Using a chemical-genetic strategy combined with quantitative proteomic analysis, we identified the glycolytic enzyme TPI1 as a new substrate of WNK1. WNK1 interacts with TPI1 in condensates to drive glycolytic reprogramming of HSCs and enhance their activation. Finally, in vivo genetic and pharmacological inhibition of WNK1 attenuated in vivo liver fibrosis in mice and ex vivo fibrosis in human precision-cut liver slices (PCLS). Collectively, these findings identify WNK1 as a previously unrecognized regulator of metabolic reprogramming in HSCs and liver fibrosis, thereby suggesting WNK1 as a potential therapeutic target.

METHODS

Human liver tissues

All experiments were conducted in accordance with both the Declarations of Helsinki and Istanbul. Human liver tissues were obtained through written consent from patients under the Mayo Clinic Institutional Review Board–approved protocol #17-010608. Written consent was given in writing by all subjects. Patient demographics are detailed in Supplemental Table S1, Supplemental Digital Content 1, https://links.lww.com/HEP/K527. Adjacent healthy edges of human liver tumor resections and cirrhotic liver tissues were processed within 1 hour of collection; they were used for primary human HSC isolation and PCLS.

Cell culture

Primary human HSCs were isolated from the adjacent healthy edges of human liver tumor resections or cirrhotic liver explants. Similar to mouse HSCs,^[20,21] primary human HSCs were isolated by perfusing a 45 cm³ human liver tissue sample through the main vessels using a 24-gauge catheter (Becton Dickinson, #381512). The human liver tissue was perfused with SC-I followed by solution 1 (Supplemental Tables S2, https://links.lww.com/HEP/K527 and S3, https://links.lww.com/HEP/K527). Chopped liver tissue was incubated with solution 2, passed through a 70 μm cell strainer, and HSCs were isolated using an Accudenz gradient protocol. The detailed protocol is included in the Supplemental Materials, https://links.lww.com/HEP/K527.

Precision-cut liver slices

Fresh human liver tissues, including adjacent edges of human liver tumors and cirrhotic liver explants, were transported in Williams’ E medium (GIBCO #12551032) at 4 °C and processed within 1 hour from the collection time. The detailed protocol, partially adapted from what has been described,^[22] is included in the Supplemental Materials, https://links.lww.com/HEP/K527.

In vivo experiments

All animal experiments were conducted in accordance with the ARRIVE guidelines and approved by the Mayo Clinic Institutional Animal Care and Use Committee (A0000474923). Eight-week-old C57Bl/6J mice were provided by Envigo. Wnk1^flox/flox mice^[23] were crossed with lecithin retinol acyltransferase (Lrat)^Cre (Jax #069595) to obtain heterozygous (Wnk1^Δ/+) or homozygous (Wnk1^ΔHSC) HSC-selective deletion of Wnk1.

Chemical genetics

A plasmid containing the WNK1 kinase domain and a plasmid containing the phosphorylated region of the known substrate OXSR1 were generated (Supplemental Figure S1, https://links.lww.com/HEP/K527), with nucleotide sequences codon-optimized for bacterial expression (Supplemental Table S4, https://links.lww.com/HEP/K527) (GenScript project #U968KCYLG0). Utilizing QuickChange XL Gold site-directed mutagenesis kit (Agilent Technologies #200516) and indicated primers in Supplemental Table S5, https://links.lww.com/HEP/K527, threonine 301 was mutated into alanine (WNK1^T301A) or glycine (WNK1^T301G) in the WNK1-kinase domain plasmid to enlarge the WNK1 ATP pocket (Supplemental Table S6, https://links.lww.com/HEP/K527).

Statistics

Experiments include at least 3 biological replicates. Numerical data are expressed as mean ± SEM. ANOVA or Kruskal-Wallis, parametric or nonparametric t tests were used to assess the statistical significance between groups as appropriate with GraphPad Prism 10.2.3 (GraphPad Software, Inc.). The normality of the variables was assessed by the Shapiro-Wilk test. p value < 0.05 was considered significant.

Detailed methods regarding in vitro, in vivo, and ex vivo experiments, as well as the list of the primary and secondary antibodies (Supplemental Tables S7, https://links.lww.com/HEP/K527 and S8, https://links.lww.com/HEP/K527) are included in the Supplemental Materials, https://links.lww.com/HEP/K527.

RESULTS

WNK1 promotes glycolysis in activated HSCs

Activated HSCs exhibit increased energetic requirements via enhanced glycolysis in response to PDGF, leading to liver fibrosis.^[7,9,24] However, the signaling promoting glycolysis in activated HSCs and subsequent liver fibrosis remains unknown. Thus, we first aimed to identify novel kinases that lead to glycolysis during HSC activation. Commercial primary human HSCs were treated with vehicle or PDGF for 15 minutes and subjected to a human phosphokinase array. In addition to the expected canonical and pleiotropic kinases such as extracellular signal-related kinases 1 and 2, protein kinase B (AKT), and glycogen synthase kinase 3, WNK1 was one of the top kinases activated by PDGF (Figure 1A, Supplemental Figure S2A, https://links.lww.com/HEP/K527). The activation of WNK1 by PDGF was confirmed at the protein level by immunofluorescence in primary human HSCs (Figure 1A) and by western blot in isolated primary mouse HSCs (Figure 1B). WNK1 activation was promoted by PDGF, but not by TGFβ, in commercial primary human HSCs (Supplemental Figure S2B, https://links.lww.com/HEP/K527). A public bulk RNA sequencing dataset (GSE119606)^[25] showed that HSCs expressed significantly higher levels of WNK1 compared to WNK2, WNK3, and WNK4 (Supplemental Figure S2C, https://links.lww.com/HEP/K527) and that PDGF, but not TGFβ, increased WNK1 mRNA expression in primary human HSCs (Supplemental Figure S2D, https://links.lww.com/HEP/K527). Congruently, PDGF promoted WNK1 activation in in-house isolated primary human HSCs from both cirrhotic explant and normal adjacent edges from liver tumors, which was abolished by WNK-IN-11, a WNK1-selective allosteric inhibitor (Supplemental Figures S2E, F, https://links.lww.com/HEP/K527). These results suggest that WNK1 is a PDGF-specific downstream signaling molecule in activated HSCs.

FIGURE 1.

PDGF-mediated WNK1 activation promotes glucose uptake and glycolysis in HSCs. (A) Commercial primary human HSCs treated with vehicle or 20 ng/mL PDGF for 15 minutes and analyzed by human phospho-protein kinase array and immunofluorescence staining (n = 4, Mann-Whitney). Scale bar: 100 μm. (B) In-house isolated primary mouse HSCs treated with vehicle or 20 ng/mL PDGF for 15 minutes and analyzed by WB (n = 4, t test). (C) Commercial primary human HSCs treated with vehicle or 20 ng/mL PDGF in the presence of DMSO or 10 μM WNK-IN-11 in medium containing 10 mM glucose and monitored for 48 hours using a live-cell metabolic analyzer (left and middle) or Incucyte (right) (n = 3, 2-way ANOVA). (D) Commercial primary human HSCs treated with vehicle or 20 ng/mL PDGF in the presence of DMSO or 10 μM WNK-IN-11 in medium containing 10 mM glucose and analyzed after 24- and 48 hours using glucose-Glo assay (n = 6, 1-way ANOVA for each of the time points). (E) LX2 cells transfected with GLUT1-GFP and treated with vehicle, 20 ng/mL PDGF, or 20 ng/mL PDGF + 10 μM WNK-IN-11 for 1 hour were analyzed by total internal reflection fluorescence microscopy (n = 3, t test). Scale bar: 25 μm. (F) LX2 cells transfected with Glifon-300 for 48 hours and treated with vehicle or 20 ng/mL PDGF in addition to DMSO or 10 μM WNK-IN-11 in medium containing 10 mM glucose for 1 hour before live cell imaging (n = 3, t test). Scale bar: 150 μm. *p < 0.05, **p < 0.01. Abbreviation: WNK1 carrying threonine 301 to glycine mutation.

Next, we investigated whether WNK1 promotes glycolysis in PDGF-activated HSCs. Glycolytic flux, including glucose consumption and lactate secretion, was measured in commercial primary human HSCs using a live-cell metabolic monitoring system for 48 hours. PDGF stimulated glycolysis compared to vehicle, as evidenced by increased glucose consumption and lactate secretion (Figure 1C). However, this increase was significantly attenuated by the WNK1 inhibitor, WNK-IN-11 (Figure 1C). These effects were not due to differences in cell proliferation (Figure 1C). These results were confirmed by a bioluminescence-based glucose consumption assay, where PDGF-mediated glucose consumption at 24- and 48-hour time points was reduced by WNK-IN-11 (Figure 1D). Augmented glycolytic flux demands higher glucose uptake in the cell, which is achieved through GLUT trafficking to the plasma membrane.^[14] GLUT1 is the main glucose transporter expressed by human and mouse HSCs (Supplemental Figure S3, https://links.lww.com/HEP/K527). PDGF can induce glycolysis after only 1 hour of stimulation.^[7] Therefore, LX2 cells, an HSC line, overexpressed GLUT1-green fluorescence protein, treated for 1 hour with vehicle, PDGF or PDGF + WNK-IN-11, and analyzed by total internal reflex fluorescence microscopy. PDGF increased GLUT1 presence at the plasma membrane, which was attenuated by WNK-IN-11 (Figure 1E). Glucose uptake was also measured in LX2 cells using Glifon-300, a fluorescent sensor that allows direct, real-time visualization of glucose uptake.^[26] Consistent with the previous results, PDGF increased Glifon-300 levels, indicating increased glucose uptake compared to vehicle, which was abrogated by WNK-IN-11 (Figure 1F), suggesting that WNK1 promotes glycolysis. However, WNK1 activation was not dependent on glucose (Supplemental Figure S4, Supplemental Digital Content 1, https://links.lww.com/HEP/K527). Taken together, these data indicate that PDGF activates WNK1, which promotes glycolysis in HSCs.

Glycolytic TPI1 is a novel WNK1 substrate

To understand how WNK1 promotes glycolysis, the activation of canonical PDGF and WNK1 downstream molecules was examined. PDGF treatment of commercial primary human HSCs increased WNK1 phosphorylation, which was not decreased by WNK-IN-11, as expected (Figure 2A). In addition, PDGF-mediated AKT phosphorylation was not significantly attenuated by WNK-IN-11 (Figure 2A), suggesting that AKT phosphorylation is not a major WNK1 downstream signaling molecule. Next, we tested the effect of PDGF on the main canonical WNK1 substrate, OXSR1. Despite being a well-established WNK1 substrate, OXSR1 phosphorylation was not enhanced by PDGF (Figure 2A), suggesting that the PDGF-WNK1 axis does not involve OXSR1 in HSCs.

FIGURE 2.

Chemical genetics identifies glycolytic enzyme TPI1 as a direct WNK1 substrate. (A) Commercial primary human HSCs treated with vehicle or 20 ng/mL PDGF for 30 minutes in the presence of DMSO or 10 μM WNK-IN-11 and analyzed by immunoblotting (n = 4, pAKT/tAKT: Kruskal-Wallis, the rest of the graphs: 1-way ANOVA). (B) Schema of the chemical genetics approach used to identify direct WNK1 substrates. (C–E) LX2 cells were transfected with mock, wild-type WNK1 (WNK1^WT), and mutant WNK1 (WNK1^T301G) for 48 hours, and treated with vehicle or 20 ng/mL PDGF for 15 minutes. Cells were then either analyzed by immunoblotting (C, n = 4, Kruskal-Wallis) or used for immunoprecipitation with anti-thiophosphoesther antibody (D, E). Immunoprecipitates were analyzed by mass spectrometry and represented with a Venn diagram (D, n = 1), and by WB (E, n = 4, one-way ANOVA). (F) LX2 cells were transfected with control siRNA (siControl) or TPI1 siRNA (siTPI1) for 72 hours, followed by transfection with the Glifon-300 plasmid for 48 hours. Cells were then treated with vehicle or 20 ng/mL PDGF for 1 hour and imaged by epifluorescence microscopy. Scale bar: 100 μm (n = 4, one-way ANOVA). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Abbreviations: AKT, protein kinase B; TPI1, triosephosphate isomerase 1; WNK1 carrying threonine 301 to glycine mutation.

To identify novel WNK1 direct substrates that promote glycolysis and subsequent HSC activation, we employed chemical genetics.^[27–29] Kinases contain a structurally conserved ATP-binding pocket that is essential for phosphorylating substrates. To identify direct substrates of WNK1, we engineered the ATP-binding pocket to accommodate modified ATP analogs that can be tracked after phosphorylation. Here, the ATP-binding pocket of WNK1 was expanded, enabling it to accommodate larger ATP analogs (N6-alkylated-ATP-γ-S) that have low affinity for wild-type WNK1 and contain a γ-thiophosphate to enable identification of mutant WNK1-phosphorylated proteins. Thio-phosphorylated WNK1 substrates were immunoprecipitated using an anti-thiophosphate-ester antibody and identified by mass spectrometry (Figure 2B). The AlphaFold3 predicted structure of the WNK1 kinase domain and the crystal structure of the closely related WNK3 kinase domain were used to identify the gatekeeper residue that restricts the WNK1 ATP pocket size, threonine 301 (T301) (Supplemental Figures S5A, B, https://links.lww.com/HEP/K527). T301 was therefore mutated into smaller amino acids with shorter side chains, alanine (T301A, WNK1^T301A) or glycine (T301G, WNK1^T301G), to enlarge the binding pocket (Supplemental Figures 5A, B, https://links.lww.com/HEP/K527). The respective kinase domains containing the ATP pockets of wild-type WNK1 (WNK1^WT), WNK1^T301A, and WNK1^T301G, as well as the region of the known WNK1 substrate, OXSR1, containing the phosphorylation site (Supplemental Figure S1, https://links.lww.com/HEP/K527, Supplemental Table S4, https://links.lww.com/HEP/K527), were purified and incubated with ATP or larger ATP analogues. Phosphorylation of OXSR1 as a slower migrating band was evaluated (Supplemental Figure S5C, https://links.lww.com/HEP/K527) to identify the optimal ATP analogue-WNK1 mutant pair. WNK1^WT, but not WNK1^T301G, could effectively utilize unmodified ATP to phosphorylate the substrate. In contrast, 6-(1-MeBu)-ATP was utilized by WNK1^T301G, but not WNK1^WT. WNK1^T301A showed intermediate affinity with any of the ATP analogues (Supplemental Figure S5C, https://links.lww.com/HEP/K527). Based on these results, WNK1^T301G and 6-(1-MeBu)-ATP were used to perform the chemical genetics screen.

To identify novel substrates, LX2 cells were transfected with mock or plasmids containing the full sequence of WNK1^WT or WNK1^T301G. Cells were then treated with vehicle or PDGF for 15 minutes to identify direct substrates, pelleted and incubated with 6-(1-MeBu)-ATP-γ-S. WNK1 overexpression was similar in WNK1^WT or WNK1^T301G conditions (Figure 2C). Substrates were immunoprecipitated from cell lysates using an anti-thiophosphate ester antibody and analyzed by nanoLC-tandem mass spectrometry. A total of 615 proteins were detected (Supplemental Table S9, https://links.lww.com/HEP/K527), where 29 of them were increased at least 1.5-fold in the WNK1^T301G PDGF condition compared to the other conditions (Figure 2D). Among these 29 proteins, 25 can be highly phosphorylated (Phosphosite-Plus), and 11 were intracellular and related to metabolism (Figure 2D).^[13,30–38] One of these 11 proteins, TPI1, converts dihydroxyacetone-phosphate to glyceraldehyde-3-phosphate during glycolysis,^[12,13] but it has not been previously reported to be phosphorylated by WNK1. TPI1 was confirmed by western blot to be detected in the thiophosphate ester pulldown at significantly higher levels in PDGF-treated WNK1^T301G LX2 cells compared to the other conditions (Figure 2E), suggesting that PDGF-activated WNK1 directly phosphorylates TPI1. Finally, the effect of TPI1 on glucose uptake was assessed using Glifon-300-tranfected LX2 cells and TPI1 small interference RNA (siRNA) knockdown. PDGF promoted glucose uptake as visualized by increased red Glifon-300 fluorescence compared to vehicle control. However, the signal was attenuated by siTPI1 (Figure 2F). Altogether, these results suggest that TPI1 is a direct phosphorylation substrate of WNK1 and promotes glucose uptake in PDGF-activated HSCs.

TPI1 is localized in WNK1 condensates in activated HSCs

To understand how WNK1 interacts with TPI1, we first examined their protein structures using the AlphaFold3 prediction tool. While the TPI1 structure was predicted with high confidence, the WNK1 structure contained several low-confidence disordered regions (Figure 3A). Intrinsically disordered regions of proteins are a driving force of phase separation and condensate formation.^[39] Thus, the presence of predicted disordered regions indicates that activated WNK1 might form condensates. To prove this hypothesis, commercial primary human HSCs were treated with vehicle or PDGF and analyzed by immunofluorescence for activated phospho-WNK1. Compared to the vehicle, PDGF not only increased the total levels of phospho-WNK1, consistent with previous results (Figures 1A–C), but also induced the formation of WNK1 puncta (Figure 3B). One particularity of the condensates is their size, which varies from micrometer to sub-micrometer.^[39,40] Compared to the vehicle, PDGF increased the number of puncta per cell and the size of single puncta from 2.85 to 3.99 μm (Figure 3B). In addition, PDGF increased the colocalization of TPI1 with these phospho-WNK1 puncta compared to vehicle (Figure 3C). Finally, to prove that these puncta are WNK1 condensates, commercial primary human HSCs were treated with 1,6-hexanediol to disrupt condensates. Immunofluorescence analysis demonstrated that PDGF-mediated phospho-WNK1/TPI1 puncta formation was abolished by 1,6-hexanediol (Figure 3D). In summary, these results suggest that activated phospho-WNK1 and TPI1 interact within condensates.

FIGURE 3.

TPI1 is localized in WNK1 condensates in activated HSCs. (A) Predicted protein structures of WNK1 and TPI1 generated using AlphaFold3. (B, C) Commercial primary human HSCs were treated with vehicle or 20 ng/mL PDGF for 15 minutes, then analyzed by immunofluorescence. The number of condensates per cell and condensate size (B) as well as Mander’s coefficient for pWNK1 and TPI1 colocalization (C) are quantified (n = 3). (D) Commercial primary human HSCs were treated with DMEM or 85 mM HD dissolved in DMEM, in addition to vehicle or 20 ng/mL PDGF for 15 minutes and analyzed by immunofluorescence staining (n = 3). Scale bars: 20 μm, t test, *p < 0.05, **p < 0.01. Abbreviations: HD, 1,6-hexanediol; TPI1, triosephosphate isomerase 1; WNK1 carrying threonine 301 to glycine mutation.

WNK1 promotes HSC proliferation and migration

Cells increase their polymerized, filamentous actin (F-actin) compared to monomer globular actin during migration,^[41,42] which is one of the HSC activation hallmarks.^[6] Thus, the effect of WNK1 on purified globular actin and F-actin was next investigated. Commercial primary human HSCs were treated with vehicle or PDGF, in addition to DMSO or WNK-IN-11 for 24 hours. PDGF increased the F-actin/globular actin ratio compared to vehicle, which was abolished by WNK-IN-11 (Figure 4A). Similarly, PDGF-mediated increase of F-actin/phalloidin staining in commercial primary human HSCs was decreased by WNK-IN-11 (Figure 4B). Additionally, PDGF increased HSC migration in a transwell assay, which was abrogated by WNK-IN-11 (Figure 4C, Supplemental Figure S6A, https://links.lww.com/HEP/K527). To validate that the observed result was due to WNK1 loss-of-function rather than off-target effects of the inhibitor, we knocked down WNK1 by siRNA in our in-house isolated primary human HSCs (Supplemental Figure S6B, https://links.lww.com/HEP/K527) and performed a transwell assay. Consistently, PDGF increased HSC migration, which was reduced by WNK1 siRNA (Figure 4D, Supplemental Figure S6C, https://links.lww.com/HEP/K527), suggesting that WNK1 promotes HSC migration.

FIGURE 4.

WNK1 promotes HSC proliferation and migration in vitro. (A) Commercial primary human HSCs were treated with vehicle or 20 ng/mL PDGF, in addition to DMSO or WNK-IN-11 for 24 hours. Cells were lysed and subjected to 134,000g ultracentrifugation to obtain G-actin from the supernatant and F-actin from the pellet. Both fractions were analyzed by WB (n = 5). (B) Commercial primary human HSCs were treated with vehicle or 20 ng/mL PDGF, in addition to DMSO or WNK-IN-11 for 24 hours and analyzed by phalloidin staining (n = 5). (C) Commercial primary human HSCs were pretreated with DMSO or 10 μM WNK-IN-11 for 4 hours and then treated with vehicle or 20 ng/mL PDGF for 8 additional hours in transwell inserts. Cells were stained with crystal violet and imaged with a histology microscope (n = 3). (D) In-house isolated primary human HSCs were transfected with control siRNA (siControl) or WNK1 siRNA (siWNK1) for 72 hours, treated with vehicle or 20 ng/mL PDGF for 8 additional hours in transwell inserts, stained with crystal violet and imaged with a histology microscope (n = 4 from 3 different patients). (E) Commercial primary human HSCs were pretreated with DMSO or 10 μM WNK-IN-11 for 4 hours and then treated with vehicle or 20 ng/mL PDGF in the presence of DMSO or WNK-IN-11 for 8 additional hours. Cell proliferation was analyzed by Ki67 immunostaining (n = 3). (F) In-house isolated primary human HSCs were transfected with siControl or siWNK1 for 72 hours, then treated with vehicle or 20 ng/mL PDGF for 24 additional hours. Cell proliferation was analyzed by Ki67 immunostaining (n = 4 from 3 different patients). Scale bars: 100 μm, one-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Abbreviation: WNK1 carrying threonine 301 to glycine mutation.

Another hallmark of PDGF-mediated HSC activation is proliferation.^[6] Hence, the effects of WNK1 on Ki67 expression or 5-ethynyl-2’-deoxyuridine (EdU) incorporation in HSCs were next examined. Commercial primary human HSCs were pretreated for 4 hours with DMSO or WNK-IN-11, followed by a 24-hour treatment with vehicle or PDGF. PDGF increased Ki67⁺ HSCs compared to vehicle, which was reduced by WNK-IN-11 (Figure 4E). Similarly, PDGF promoted EdU incorporation in in-house isolated primary human HSCs from a cirrhotic liver tissue, which was abrogated by WNK-IN-11 (Supplemental Figure S6D, https://links.lww.com/HEP/K527). Congruently, in-house isolated primary human HSCs were transfected with control or WNK1 siRNA for 48 hours and used for a 24-hour proliferation assay. PDGF-stimulated Ki67 levels in siControl HSCs were reversed by siWNK1 (Figure 4F). WNK-IN-11 did not affect commercial primary human HSC viability, as shown by ATP levels (Supplemental Figure S6E, https://links.lww.com/HEP/K527). Additionally, adjacent edges of human liver tumor tissues used for human HSC isolations were heterogeneous regarding fibrosis, where one of the tissues showed increased collagen deposition by Sirius red staining (Supplemental Figure S7A, https://links.lww.com/HEP/K527). Despite these differences in fibrosis, HSCs isolated from these three patient tissues responded to PDGF similarly by increasing their migration and proliferation compared to vehicle (Figures 4D–F). Finally, the isolated HSC phenotype was similar regardless of the degree of fibrosis (Supplemental Figure S7B, https://links.lww.com/HEP/K527). Altogether, these results suggest that WNK1 downstream of PDGF promotes HSC migration and proliferation.

Glycolytic TPI1 promotes HSC proliferation and migration

Given that TPI1 is a new WNK1 substrate (Figure 2) and that WNK1 promotes HSC proliferation and migration (Figure 4), we next sought to investigate TPI1’s functional role in HSCs. Commercial primary human HSCs were transfected with control or TPI1 siRNA for 48 hours and used in a transwell migration assay for 8 additional hours. TPI1 knockdown was confirmed at the protein level (Supplemental Figure S8A, https://links.lww.com/HEP/K527). PDGF-mediated HSC migration was abrogated by TPI1 siRNA (Figure 5A, Supplemental Figure S8B, https://links.lww.com/HEP/K527). Since TPI1 co-localized with WNK1 condensates (Figures 3C, D), the role of condensates on commercial primary human HSC migration was examined. PDGF-stimulated HSC migration was abolished by the condensate inhibitor 1,6-hexanediol (Figure 5B, Supplemental Figure S8C, https://links.lww.com/HEP/K527), suggesting that condensates support HSC migration. Since WNK1 and TPI1 promote glycolysis (Figure 1),^[12,13] we examined the role of glycolysis on HSC migration. As expected, PDGF-mediated commercial primary human HSC migration was reduced by 2-deoxyglucose, a glycolysis inhibitor (Figure 5C, Supplemental Figure S8D, https://links.lww.com/HEP/K527), suggesting that glycolysis fuels HSC migration.

FIGURE 5.

TPI1, condensates and glycolysis promote HSC proliferation and migration in vitro. (A) Commercial primary human HSCs were transfected with control siRNA (siControl) or TPI1 siRNA (siTPI1) for 72 hours, treated with vehicle or 20 ng/mL PDGF for 8 additional hours in transwell inserts, stained with crystal violet and imaged with a histology microscope (n = 3). (B) Commercial primary human HSCs were pretreated with DMEM control or 85 mM HD for 4 hours and then incubated with vehicle or 20 ng/mL PDGF for 8 additional hours in transwell inserts. Cells were stained with crystal violet and imaged with a histology microscope (n = 3). (C) Commercial primary human HSCs were pretreated with DMEM control or 6 mM 2DG for 4 hours and then incubated with vehicle or 20 ng/mL PDGF for 8 additional hours in transwell inserts. Cells were stained with crystal violet and imaged with a histology microscope (n = 3). (D) Commercial primary human HSCs were transfected with siControl or siTPI1 for 72 hours, then treated with vehicle or 20 ng/mL PDGF for 24 additional hours. Cell proliferation was analyzed by EdU staining (n = 3). (E) Commercial primary human HSCs were pretreated with DMSO or 85 mM HD for 4 hours. Vehicle or 20 ng/mL PDGF was added for 24 additional hours, and cell proliferation was analyzed by EdU staining (n = 3). (F) Commercial primary human HSCs were pretreated with DMEM control or 6 mM 2DG for 4 hours. Vehicle or 20 ng/mL PDGF was added for 24 additional hours, and cell proliferation was analyzed by EdU staining (n = 3). Scale bars: 100 μm, one-way ANOVA, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Abbreviations: 2DG, 2-deoxyglucose; EdU, 5-ethynyl-2’-deoxyuridine; 1,6-hexanediol; TPI1, triosephosphate isomerase 1.

Next, the role of TPI1, condensates and glycolysis in commercial primary human HSC proliferation were examined by EdU incorporation or Ki67 immunofluorescence. PDGF-stimulated HSC proliferation was reduced by TPI1 siRNA knockdown, condensate inhibitor 1,6-hexanediol and glycolysis inhibitor 2-deoxyglucose (Figures 5D–F). Taken together, these results demonstrate that WNK1 substrate TPI1, condensate formation and glycolysis promote HSC migration and proliferation.

Genetic deletion of Wnk1 selectively in HSCs attenuates liver fibrosis in vivo

Since WNK1 promoted HSC proliferation and migration (Figure 4), we next investigated the role of HSC-selective WNK1 deletion in a mouse model of liver fibrosis. Lrat^Cre mice were crossed with Wnk1^fl/fl mice to obtain heterozygous Wnk1^Δ/+ or homozygous Wnk1^ΔHSC offspring, where Wnk1 is selectively deleted in HSCs. Wnk1^fl/+ controls or Wnk1^Δ/+ mice were administered with olive oil or carbon tetrachloride (CCl₄) to induce liver fibrosis (Figure 6A). CCl₄-treated Wnk1^fl/+ littermate controls displayed increased collagen type 1 (COL1) and alpha smooth muscle actin (αSMA) protein levels compared to olive oil-treated Wnk1^fl/+ controls (Figure 6B). However, COL1 and αSMA levels were significantly reduced in Wnk1^Δ/+ mice (Figure 6B). Similarly, CCl₄ administration elevated COL1 and collagen type 3 (COL3) immunofluorescence levels compared to olive oil in Wnk1^fl/+ controls, which were significantly reduced in Wnk1^Δ/+ mice (Figures 6C, D). Finally, Sirius red staining was augmented by CCl₄ in Wnk1^fl/+ controls and abrogated in Wnk1^Δ/+ mice (Figure 6E). CCl₄-treated heterozygous Wnk1^fl/+ controls also displayed increased protein levels of GLUT1, desmin and GLUT1-desmin colocalization compared to oil-treated controls, which were markedly reduced in WNK1^Δ/+ mice (Supplemental Figure S9A, https://links.lww.com/HEP/K527). Moreover, CCl₄ treatment of Wnk1^fl/+ littermate controls increased pWNK1 levels, TPI1 levels, pWNK1-TPI1 colocalization and pWNK1 condensate formation in desmin⁺ HSCs compared to olive oil group, which was reduced in WNK1^Δ/+ mice (Supplemental Figure S9B, https://links.lww.com/HEP/K527). Similar results were obtained with homozygous mice, where the CCl₄-mediated fibrosis in Wnk1^fl/fl controls was significantly decreased Wnk1^ΔHSC mice, as shown by Sirius red, western blot, COL1 and COL3 immunofluorescence (Supplemental Figures S10A–D, https://links.lww.com/HEP/K527). In addition, Wnk1^ΔHSC mice displayed reduced CCl₄-mediated GLUT1, desmin, and GLUT1-desmin colocalization in the liver (Supplemental Figure S10E, https://links.lww.com/HEP/K527), suggesting that WNK1 in HSCs regulates glycolysis-related protein levels.

FIGURE 6.

Genetic deletion of Wnk1 selectively in HSCs attenuates CCl₄-induced liver fibrosis in mice. Wnk1^fl/+ or Wnk1^Δ/+ male and female mice were treated with either olive oil or CCl₄ for 4 weeks, as shown in the experimental strategy (A). Livers were analyzed by WB (B), collagen 1 (COL1) and collagen 3 (COL3) immunofluorescence (C and D), and Sirius red staining (E). n = 5 to 12 mice per group, scale bars: 100 μm, αSMA WB graph: Kruskal-Wallis, the rest of the graphs: 1-way ANOVA *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. Abbreviations: CCl₄, carbon tetrachloride; COL1, collagen type 1; αSMA, alpha smooth muscle actin; WB, western blot; WNK1, with no lysine kinase 1.

To further validate the role of WNK1 in liver fibrosis, we next employed the high-fat, choline-deficient, l-amino-acid (HF-CDAA) diet model, which induces steatohepatitis and progressive sinusoidal fibrosis.^[43] Wnk1^fl/fl controls or Wnk1^ΔHSC mice were fed HF-CDAA diet for 6 weeks (Figure 7A). Sirius red staining showed extensive collagen deposition in HF-CDAA-fed Wnk1^fl/fl controls, which was markedly reduced in Wnk1^ΔHSC mice (Figure 7B). Consistently, HF-CDAA diet-induced αSMA, COL1, and COL3 protein levels were reduced in Wnk1^ΔHSC mice compared to Wnk1^fl/fl controls (Figures 7C–E). Moreover, Wnk1^ΔHSC mice displayed reduced HF-CDAA diet-mediated GLUT1, desmin, and GLUT1-desmin colocalization in the liver (Supplemental Figure S11, https://links.lww.com/HEP/K527). Taken together, these results demonstrate that genetic deletion of Wnk1 selectively in HSCs attenuates liver fibrosis in vivo and highlights WNK1 as a potential therapeutic target in chronic liver injury.

FIGURE 7.

Genetic deletion of Wnk1 selectively in HSCs attenuates liver fibrosis in the HF-CDAA model in vivo. Wnk1^fl/fl or Wnk1^ΔHSC male and female mice were fed on a HF-CDAA diet for 6 weeks, as shown in the experimental strategy (A). Livers were analyzed by Sirius red staining (B), WB (C), as well as COL1 and COL3 immunofluorescence (D and E). n = 3–5 mice per group, scale bars: 100 μm, t test, *p < 0.05, ****p < 0.0001. Abbreviations: COL1, collagen type 1; COL3, collagen type 3; HF-CDAA, high-fat, choline-deficient, l-amino-acid; αSMA, alpha smooth muscle actin; WNK1, with no lysine kinase 1.

Pharmacological inhibition of WNK1 attenuates liver fibrosis in mice and in human precision-cut liver slices

To assess the therapeutic potential of WNK1 for the treatment of liver fibrosis, we tested 2 small-molecule inhibitors, WNK1-selective WNK-IN-11 or pan-WNK inhibitor WNK463, in the CCl₄-induced liver fibrosis model. C57Bl/6 mice were administered with olive oil or CCl₄ twice a week and WNK1 inhibitors 3 times a week for 4 weeks (Figure 8A). CCl₄ administration significantly increased COL1 and αSMA protein levels compared to olive oil, which were reduced by WNK463 and WNK-IN-11 (Figure 8B). Congruently, CCl₄-mediated COL1 and COL3 immunofluorescence signals, as well as Sirius red staining, were significantly attenuated by WNK463 and WNK-IN-11 (Figures 8C, D). These results suggest that in vivo WNK1 pharmacological inhibition decreases CCl₄-mediated liver fibrosis. In a separate cohort, body weight, liver enzymes, and histological analysis did not show any significant changes between the vehicle and WNK1 inhibitors, except for the WNK463, inducing white pulp distortion in the spleen (Supplemental Figure S12, https://links.lww.com/HEP/K527).

FIGURE 8.

Pharmacological inhibition of WNK1 attenuates liver fibrosis in mice and in the human PCLS model. (A–D) C57Bl/6 male and female mice were treated with olive oil or CCl₄ twice a week, as well as vehicle (10% DMSO, 90% corn oil), WNK463 or WNK-IN-11 3 times a week, as shown in the experimental strategy (A). Livers were analyzed by WB (B), collagen 1 (COL1) and collagen 3 (COL3) immunofluorescence (C), and Sirius red staining (D). (n = 3–10 mice per group, αSMA WB graph: Kruskal-Wallis, the rest of the graphs: 1-way ANOVA) (E) Adjacent normal liver tissues from tumor resections and cirrhotic liver tissues were analyzed by immunofluorescence (n = 3 patients per group, Mann-Whitney). (F) Schematic of the experimental strategy for generating precision-cut liver slices (PCLS) from patient liver resections. Healthy human adjacent edge liver tissues from tumor resections (n = 9 from 7 different patients, Kruskal-Wallis) or human cirrhotic explant tissues (n = 3 independent replicates from 1 patient, t test) were processed in the laboratory within one hour from the collection using a vibratome to obtain 250 μm thick PCLS, as shown in the schema. PCLS were treated with vehicle (deionized water) or 20 ng/mL PDGF in the presence of DMSO or 10 μM WNK-IN-11 for 5 days, with daily media and treatment refreshment, and analyzed by WB. Scale bars: 100 μm. *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001. Abbreviations: αSMA, alpha smooth muscle actin; CCl₄, carbon tetrachloride; COL1, collagen type 1; PCLS, precision-cut liver slices; WNK1, with no lysine kinase 1.

To assess translational relevance, we examined the expression of the WNK1 gene in human liver tissues using the Human Liver scRNAseq Atlas (https://shiny.igc.ed.ac.uk/Human_Liver_scRNAseq_Atlas/). Liver mesenchymal cells showed increased collagen 1 alpha (COL1A1), COL3A1, and TPI1 levels, but decreased levels of WNK1 expression, from healthy to chronic liver disease condition (Supplemental Figure S13A, https://links.lww.com/HEP/K527). Since liver mesenchymal cells include several cell types (Supplemental Figure S13B, https://links.lww.com/HEP/K527), we examined the single-cell distribution of these genes. Activated HSCs had increased expression of COL1A1 and COL3A1, as expected, but did not show WNK1 differential expression (Supplemental Figure S13C, https://links.lww.com/HEP/K527), suggesting that WNK1 mRNA expression is not increased in chronic liver diseases. Thus, the levels of activated WNK1 in normal and cirrhotic liver tissues (Supplemental Table S1, https://links.lww.com/HEP/K527) were next investigated. Compared to normal adjacent edge liver tissue, WNK1 phosphorylation was significantly higher in cirrhotic liver tissue (Supplemental Figure S14A, https://links.lww.com/HEP/K527), suggesting that WNK1 activation is associated with human chronic liver fibrosis. Moreover, the expression of phospho-WNK1, TPI1, as well as pWNK1-TPI1 colocalization were markedly increased in desmin⁺ HSCs in cirrhotic livers compared to the normal adjacent edge tissues (Figure 8E). This is in line with the activation of WNK1-TPI1 axis in activated HSCs during human liver fibrosis.

Lastly, the effect of WNK1-selective inhibitor, WNK-IN-11, was evaluated in a human ex vivo model of liver fibrosis using PCLS (Figure 8F, schema). Human liver tissues were obtained from the adjacent healthy edges of tumor resections or cirrhotic liver explants and 250-μm-thick PCLS were obtained, preserving the native 3-dimensional architecture and cell-cell interactions of the liver. The PCLS were treated with vehicle or PDGF to induce human ex vivo liver fibrosis, in addition to DMSO or WNK-IN-11 for 3 or 5 days. PDGF stimulation of adjacent edges of liver tumor tissues increased the expression of COL1 compared to the vehicle condition, both at 3 and 5 days of culture, which was significantly reduced by WNK-IN-11 (Figure 8F, Supplemental Figure S14B, https://links.lww.com/HEP/K527). In cirrhotic liver tissue, PCLS, WNK-IN-11 significantly reduced COL1 levels at 5 days of culture, but not at 3 days of culture (Figure 8F, Supplemental Figure S14C, https://links.lww.com/HEP/K527). Additionally, no significant changes in ATP levels were observed throughout the 5-day culture period, indicating that tissue viability was stable (Supplemental Figure S14D, https://links.lww.com/HEP/K527). Together, these results demonstrate that pharmacological inhibition of WNK1 in vivo and ex vivo attenuate murine and human liver fibrosis, respectively.

DISCUSSION

In this study, by using chemical genetics, proteomics, single-cell transcriptomics, transgenic models, and a human 3-dimensional liver fibrosis model, we revealed that WNK1 is a new kinase in activated HSCs during liver fibrosis and functions as a central signaling node linking growth factor stimulation to metabolic reprogramming in HSCs. More specifically, we identified a novel mechanism where WNK1 promoted glycolysis through phosphorylating a specific glycolytic enzyme, TPI1, which has not been previously linked to liver fibrosis. In addition, we observed the formation of new TPI1-WNK1 condensates which facilitated their function in inducing HSC migration and proliferation. Lastly, we provide evidence that genetic and pharmacological inhibition of WNK1 in vivo in mice and ex vivo in a human 3D model inhibits liver fibrosis.

Glycolysis is a key metabolic pathway that converts glucose into pyruvate to produce cellular energy. We and others have previously described how growth factor-mediated glycolysis in HSCs promotes liver fibrosis.^[7,9] PDGF is a key driver of HSC activation in liver fibrosis,^[6] yet the mechanisms by which it promotes glycolysis are unknown. Glycolysis has been shown to be activated by kinase signaling, such as lymphocyte-specific protein tyrosine kinase in T cells or AKT and mitogen-activated protein kinases in cancer cells.^[44,45] However, the PDGF-dependent kinase signaling that drives glycolysis in HSCs has not been delineated. Here, we reveal a new signaling pathway where PDGF-induced WNK1 phosphorylation at threonine 60 drives glucose uptake and subsequent glycolysis during HSC activation.

WNK1, a kinase recognized to be involved in cancer, acts through its canonical substrates, OXSR1 and SPAK1, which regulate ion transport and cell volume.^[19] Although the growth factor that activates WNK1 and its substrates has not been clearly identified, it has been shown that WNK1 and OXSR1 crosstalk with TGFβ pathway.^[46] Nonetheless, in our system, TGFβ failed to activate WNK1. Additionally, we found that PDGF activated WNK1 in HSCs without affecting OXSR1 phosphorylation, suggesting the existence of other unknown substrates. Using a chemical genetics approach combined with mass spectrometry, we identified TPI1, a glycolytic enzyme, as a new substrate of WNK1. The identification of the phosphorylation site on TPI1 would be a logical next step. Importantly, both WNK1 and TPI1 promoted HSC proliferation and migration. Recently, WNK1 has been reported to form condensates through its intrinsically disordered C-terminal domain.^[47] On the other hand, TPI1 has not been previously reported to localize in condensates. In HSCs, we found that TPI1 localizes within WNK1 condensates, suggesting these structures provide a platform through which WNK1 coordinates glycolysis. It is shown that intranuclear condensates are involved in fibrosis.^[48] Nonetheless, the role of condensates in HSCs is unknown. Our data demonstrates that WNK1 condensates are essential for HSC proliferation and migration.

HSC activation is hallmark of liver fibrosis. Liver fibrosis remains a major clinical challenge. Drugs like Resmetirom and Semaglutide show some promise,^[49,50] but with limited antifibrotic effects. Our data demonstrate that genetic and pharmacological inhibition of WNK1 significantly reduced liver fibrosis in mice, without any noticeable toxicity. Utilizing human PCLS ex vivo liver fibrosis model, we demonstrate that WNK1 inhibition reduced collagen deposition, confirming translational relevance and supporting WNK1 as a promising antifibrotic target. However, murine and human 3D models do not fully reflect the complexity of the human physiological environment. Interspecies differences, including a potentially faster drug metabolism in murine livers,^[51] may represent some limitations regarding the translational applicability of our findings, and require further validation.

In summary, our findings revealed WNK1 as a previously unrecognized regulator of metabolic HSC activation and liver fibrosis, opening new avenues for targeting fibrogenic signaling. Efficient and safe pharmacological inhibitors targeting WNK1 or TPI1 during liver fibrosis are currently unexplored in clinics, defining a potential area of investigation for drug development, notably through targeting WNK1 activation or TPI1 phosphorylation.

Supplementary Material

Supplemental Digital Content is available for this article. Direct URL citations are provided in the HTML and PDF versions of this article on the journal’s website, www.hepjournal.com.

Abbreviations:

αSMA

alpha smooth muscle actin

AKT

protein kinase B

CCl₄

carbon tetrachloride

COL1

collagen type 1

COL1A1

collagen 1 alpha 1

COL3

collagen type 3

COL3A1

collagen 3 alpha 1

EdU

5-ethynyl-2’-deoxyuridine

F-actin

filamentous actin

GLUT

glucose transporter

HF-CDAA

high-fat, choline-deficient, L-amino-acid

LRAT

lecithin retinol acyltransferase

OXSR1

oxidative stress-responsive kinase 1

PCLS

precision-cut liver slices

siRNA

small interference RNA

SPAK

SPS1-related proline/alanine-rich kinase

TPI1

triosephosphate isomerase 1

WNK1

with no lysine kinase 1

WNK1^WT

wild-type WNK1

WNK1^T301G

WNK1 carrying threonine 301 to glycine mutation