Abstract

   Bilirubin, an endogenous metabolite with many significant physiological
   roles, particularly anti‐inflammatory properties, shows great promise
   as a treatment for inflammatory diseases. However, the binding targets
   and downstream signaling mechanisms of bilirubin remain unclear. Here,
   by using quantitative phosphorylation proteomics and several powerful
   chemical biology techniques such as the Cellular Thermal Shift Assay
   (CETSA), molecular docking, and MicroScale Thermophoresis (MST), it is
   identified and confirmed that with‐no‐lysine (K) kinase 1 (WNK1) is the
   primary target of bilirubin at physiological concentrations. Bilirubin
   binds to the kinase domain of WNK1, activating its kinase activity and
   increasing the intracellular chloride ion concentration via the
   downstream SPAK/OSR1‐KCC2 pathway in neurons. Manipulating endogenous
   bilirubin levels by deleting Blvra, the bilirubin synthesis enzyme, and
   Ugt1a1, its metabolic enzyme, significantly promotes and inhibits the
   activation of the lipopolysaccharide (LPS)‐induced NLRP3 inflammasome,
   respectively, in mouse hippocampus. Similarly, exogenous bilirubin
   supplementation suppressed LPS‐induced NLRP3 inflammasome activation in
   mouse hippocampus in a WNK1‐dependent manner. Quantitative
   phosphoproteomic analysis of WNK1 downstream signaling elucidated the
   broad biological roles of WNK1, notably its function in suppressing
   inflammation. The findings clarify the direct targets and signaling
   mechanisms underlying the anti‐inflammatory effects of bilirubin and
   pave the way for exploring its novel functions.

   Keywords: bilirubin, neuroinflammation, NLRP3, phosphorylation, WNK
   kinase
     __________________________________________________________________

   At physiological concentrations, bilirubin binds to the kinase domain
   of WNK1, thereby augmenting its activity and facilitating the
   phosphorylation of downstream SPAK/OSR1. This phosphorylation inhibits
   KCC2 activity, leading to elevate intracellular chloride levels in
   neurons. Moreover, the interaction between bilirubin and WNK1
   suppresses the activation of LPS‐induced NLRP3 inflammasomes, thereby
   exerting anti‐inflammatory effects.

   graphic file with name ADVS-12-2407349-g002.jpg

1. Introduction

   Bilirubin, a heme metabolite produced during the hepatic processing of
   hemoglobin, was once viewed merely as a metabolic waste product or a
   cause of brain damage in neonates when present in excess.^[ [52]^1 ^]
   However, recent studies have recognized its significant physiological
   functions.^[ [53]^2 ^] Emerging clinical evidence indicates that
   bilirubin is a crucial signaling molecule involved in the
   pathophysiology of numerous immune, infectious, and metabolic diseases.
   Even minor decreases in serum bilirubin levels are associated with an
   increased risk for these conditions.^[ [54]^3 ^] Furthermore, various
   synthesized nanoparticles have been developed to elevate bilirubin
   levels for the treatment of inflammatory and autoimmune diseases in
   clinical trials.^[ [55]^4 ^] The beneficial effects of bilirubin are
   attributed to its versatile functions, including hormone mimicry,
   clearance of oxygen free radicals, and suppression of nearly all immune
   system effectors.^[ [56]^5 ^] The physiological function of bilirubin
   is presumed to result from inhibition of protein phosphorylation and
   activation of various nuclear and cytoplasmic receptors.^[ [57]^6 ^]
   However, the molecular mechanisms by which bilirubin affects cell
   signaling and contributes to its physiological functions remain poorly
   understood.

   In the bloodstream, most bilirubin is bound to albumin, which
   facilitates stable circulation. Free bilirubin, being lipophilic, can
   cross both peripheral and central physiological barriers and cell
   membranes.^[ [58]^7 ^] As a signaling molecule, bilirubin is believed
   to influence various cellular regulatory systems by binding to cell
   membranes and intracellular target proteins, thereby modulating
   subsequent cell signaling.^[ [59]^8 ^] Identifying these target
   proteins is crucial for fully understanding the molecular mechanisms
   underlying bilirubin's extensive biological effects. In the peripheral
   system, bilirubin interacts with several intracellular and nuclear
   receptors, such as peroxisome proliferator‐activated receptors (PPAR)
   and liver‐specific fatty acid binding protein (FABP1), modulating
   various metabolic pathways.^[ [60]^3 ^] At pathological high levels,
   bilirubin can bind to Mas‐related G‐protein coupled receptors (MRGPRs)
   to induce cholestatic itch.^[ [61]^9 ^] Notably, bilirubin plays a
   particularly significant role in the central nervous system, where it
   is involved in signaling pathways that counteract inflammation
   associated with chronic neurodegenerative, neuropsychiatric, and other
   neurological diseases. However, severe elevation of bilirubin can lead
   to or exacerbate neural injury. We have previously reported that
   elevated serum bilirubin levels in patients with stroke directly
   activate the transient receptor potential melastatin 2 (TRPM2) channel
   on neuron cell membranes, worsening brain damage.^[ [62]^10 ^] Despite
   these findings, the precise target proteins through which bilirubin
   exerts its cellular regulatory functions and contributes to its
   anti‐inflammatory roles remain unclear.

   Bilirubin exerts widespread inhibitory effects on protein
   phosphorylation, significantly modulating intracellular signaling
   pathways across various biological processes.^[ [63]^6 , [64]^11 ^] For
   instance, bilirubin alleviates group 2 innate lymphoid cells (ILC2)
   driven airway inflammation by downregulating the phosphorylation of
   extracellular signal‐regulated kinase (ERK)1/2.^[ [65]^12 ^] Protein
   phosphorylation is crucial for modulating protein function and
   controlling inflammatory, immune, and metabolic processes, such as
   NOD‐like receptor thermal protein domain‐associated protein 3 (NLRP3)
   inflammasome assembly and activation.^[ [66]^13 , [67]^14 , [68]^15 ^]
   These findings prompted us to investigate the physiological protein
   targets of bilirubin by analyzing its phosphorylation regulatory
   network.

   In this study, we conducted quantitative phosphorylation proteomics to
   analyze the kinase‐substrate network modulated by bilirubin in the
   cochlear nucleus. We utilized a combination of the Cellular Thermal
   Shift Assay (CETSA), molecular docking, and MicroScale Thermophoresis
   (MST) to identify the targets of bilirubin and the binding sites among
   differentially phosphorylated proteins. Using these robust techniques,
   we determined that with‐no‐lysine (K) kinase 1 (WNK1) binds directly to
   bilirubin at physiological concentrations. WNK1, a ubiquitously
   expressed serine‐threonine kinase, is part of a signaling cascade that
   is essential for chloride ion homeostasis, immune regulation,
   autophagy, and other processes.^[ [69]^16 , [70]^17 ^] Our results
   demonstrated that bilirubin has a strong direct interaction with the
   kinase domain of WNK1, promoting its kinase activity and elevating
   intracellular chloride ion levels in neurons. Furthermore, we examined
   the effects of bilirubin on WNK1 in vivo and in vitro, identifying an
   essential role for WNK1 in LPS‐induced inactivation of the NLRP3
   inflammasome through the activation of WNK1 kinase function.

2. Results

2.1. Integrative Analysis of Phosphoproteomics and CETSA Identifies WNK1 as a
Preferred Target of Bilirubin

   To elucidate the direct binding targets and downstream signaling
   pathways of bilirubin, we conducted quantitative phosphoproteomic and
   drug‐protein binding studies, followed by integrated analysis. Both
   human and animal studies have demonstrated that bilirubin
   preferentially deposits in specific brain regions, which are recognized
   as bilirubin‐sensitive, including the brainstem nuclei (e.g., cochlear
   nucleus, vestibular nuclei, superior olivary complex), hippocampus and
   cerebellum. Bilirubin‐sensitive cochlear nuclei from mice were
   incubated in artificial cerebrospinal fluid (ACSF) with or without 3 µM
   bilirubin (Figure [71] 1A).^[ [72]^18 ^] Tandem mass tag (TMT)‐based
   phosphoproteomic methods were used to assess changes in protein
   abundance and phosphorylation at the whole‐proteome level
   (Figure [73]1A).^[ [74]^19 , [75]^20 ^] Principal component analysis
   (PCA), Pearson correlation coefficients, and relative standard
   deviations (RSD) indicated significant differences between
   bilirubin‐treated and control samples (Figure [76]S1A–C, Supporting
   Information). We quantified 11116 phosphorylation sites across the 3474
   proteins (Figure [77]1B). Bilirubin treatment resulted in a notable
   reduction in significantly regulated phosphorylation sites and
   proteins, with 340 proteins showing decreased phosphorylation at 469
   sites and 91 proteins showing increased phosphorylation at 126 sites
   (Figure [78]1C). These modifications predominantly affected
   serine/threonine residues, with 75.8% involving phosphorylation and
   17.82% dephosphorylation of serine residues (Figure [79]1D). Motif
   analysis identified the significant motif xxxxPx_S_PxxxRx (Figure
   [80]S2A–C, Supporting Information). Enrichment analysis revealed that
   the phosphorylated proteins responsive to bilirubin were mainly
   associated with neurodevelopment, neurogenesis, autophagy, and synaptic
   structures (Figure [81]S1D,E, Supporting Information).

Figure 1.

   Figure 1
   [82]Open in a new tab

   Integrative analysis of phosphorylation proteomics and CETSA to
   identify bilirubin's downstream signaling and binding targets. A)
   Workflow for identifying bilirubin targets. B) Volcano plot
   illustrating phosphorylation sites identified after bilirubin
   treatment. C) Number of significantly regulated proteins and
   phosphorylation sites following bilirubin treatment. D) Distribution of
   significantly altered phosphorylation site amino acid types. E)
   Distribution of top candidate targets matching bilirubin from the
   integrated analysis of two proteomics datasets. Protein kinases (pink
   and red) are labeled and mapped onto the human kinase spectrum (bottom
   right), with WNK1, the most significantly associated kinase target,
   highlighted in red. F) CETSA‐Western blotting experiment confirms the
   interaction between bilirubin and WNK1. G) Western blotting analysis of
   WNK1 Thr60 phosphorylation in control and bilirubin‐treated mouse
   cochlear nuclei for 30 min. In panels B–D), all bilirubin treatment
   groups were compared with their respective control groups.
   Significantly regulated phosphorylation sites were defined as those
   having an absolute value of bilirubin/control >1.5 and an adjusted p <
   0.05. See also Figure [83]S1 (Supporting Information).

   Protein kinases are key regulators of cell signaling mediated by
   reversible phosphorylation that relies on protein–protein binding
   between kinases and their regulators. Protein kinase analysis
   identified several kinase families, including mitogen‐activated protein
   kinase (MAPK) and WNK, that were significantly affected by bilirubin
   (Figure [84]S1F, Supporting Information). Notably, the WNK family
   members exhibited the most significant decrease in phosphorylation
   levels following bilirubin treatment (Figure [85]S1G, Supporting
   Information).

   To further clarify the potential binding targets of bilirubin, we used
   a label‐free technique based on protein thermal stability, known as
   CETSA, combined with mass spectrometry (MS). This method is widely
   applicable for exploring protein‐molecule interactions.^[ [86]^21 ,
   [87]^22 ^] SHSY‐5Y cell lysates were incubated with bilirubin, followed
   by denaturation at 55 °C. Proteins bound to bilirubin exhibited altered
   solubility owing to changes in thermal stability (Figure [88]1A). MS
   analysis of the soluble fraction revealed an increase in the abundance
   of 458 proteins and a decrease in that of 149 proteins, indicating
   their direct or indirect association with bilirubin (Figure [89]S3,
   Supporting Information). Functional annotation and enrichment analysis
   of proteins with increased solubility highlighted an overrepresentation
   of proteins involved in the “cellular redox homeostasis” pathway,
   aligning with the predicted function of bilirubin as a redox reagent.
   Additionally, proteins related to “UDP‐glucosyltransferase activity,” a
   pathway directly linked to bilirubin metabolism, were identified.
   Conversely, proteins with decreased solubility were significantly
   enriched in the “peroxisome” pathway, consistent with studies showing
   that bilirubin modulates peroxisome proliferator‐activated receptor
   alpha (PPARα) activity.^[ [90]^23 ^] Notably, the “protein
   serine/threonine kinase activity” pathway was also enriched among
   proteins with increased solubility (Figure [91]S4A,B, Supporting
   Information).

   Integrated analysis of phosphoproteomic and binding proteomic data has
   identified 38 proteins as candidate bilirubin‐binding targets that also
   affect phosphorylation. Among these, WNK1 emerged as the most promising
   candidate, with a p‐value of 7.68 × 10^−7 (Figure [92]1E; Figure
   [93]S5A, Supporting Information). Further analysis using the Allen
   Brain Atlas and immunofluorescence staining revealed that WNK1 is
   highly expressed in bilirubin‐sensitive brain regions, such as the
   hippocampus and cochlear nucleus (Figure [94]S5B–E, Supporting
   Information). Additionally, domain enrichment analysis of downregulated
   phosphorylated proteins after bilirubin treatment highlighted the
   “Serine/threonine‐protein kinase OSR1/WNK, CCT domain” (Figure
   [95]S6A,B, Supporting Information). These results suggest that WNK1 is
   a potential target for bilirubin in the brain.

   To validate these findings, SH‐SY5Y cells were incubated in a
   serum‐free medium with bilirubin (6 µM) or an equivalent amount of DMSO
   as the control, followed by protein denaturation at different
   temperatures (Figure [96]1A). Western blotting analysis of the soluble
   fraction showed an enhanced thermal stability of WNK1 in the
   bilirubin‐treated group (Figure [97]1F). Phosphoproteomic analysis
   revealed a significant decrease in WNK1 phosphorylation at Thr58
   (corresponding to Thr60 in humans), Ser183, and Ser185 (Figure [98]1B;
   Figure [99]S7A–D, Supporting Information). Western blotting results
   further demonstrated a substantial reduction in WNK1 phosphorylation at
   Thr58 (corresponding to Thr60 in humans) in the cochlear nucleus after
   bilirubin treatment (Figure [100]1G). These results strongly suggest
   that WNK1 is a target of bilirubin.

2.2. Bilirubin Directly Binds to WNK1

   To investigate the direct interaction between bilirubin and WNK1, we
   initially obtained a three‐dimensional structure of WNK1 from
   Alphafold2.^[ [101]^24 ^] Subsequently, we performed molecular docking
   between the crystal structure of bilirubin and WNK1 using DOCK 6.9
   application (Figure [102]2A–C).^[ [103]^25 ^] The results revealed a
   high affinity between bilirubin and the 195–483 region of human WNK1
   (hWNK1, Score = −78.5), which corresponds to the catalytic domain of
   WNK1 (Figure [104]2B). The detailed analysis identified multiple
   residues within this region that involved hydrophobic interactions
   (Ile227, Lys233, Val235, Ala248, Val281, Phe283, Leu299, Thr301,
   Met304, Thr308, Thr311, and Phe356), salt bridges (Lys233), π‐π
   stacking (Phe356), and hydrogen bonds (Cys250) with bilirubin. Notably,
   several of these residues were active sites for the WNK1 kinase domain
   and ATP binding sites (Ile227, Lys233, Val235, Ala248, Cys250, Val281,
   Thr301, Met304, Thr308, and Phe356) (Figure [105]2C).

Figure 2.

   Figure 2
   [106]Open in a new tab

   Bilirubin directly binds to WNK1. A) Chemical structure and molecular
   formula of bilirubin. B) Molecular docking showing an overview of
   bilirubin‐bound hWNK1 (left panel) and a close‐up view of the
   bilirubin‐binding site (right panel). The grid score for
   bilirubin‐bound hWNK1 is shown. C) Specific recognition of
   bilirubin‐bound hWNK1. 3D diagram showing bilirubin binding to the deep
   cavity of the WNK1 protein kinase domain, and a 2D diagram illustrating
   the interactions through hydrogen bonds, salt bridges, π–π
   interactions, and hydrophobic interactions. D) MST experimental traces
   of full‐length hWNK1 with bilirubin, with vertical bars indicating
   fluorescence changes (gray cold; blue hot). A formula for calculating
   Fnorm is also presented. E) MST demonstrating a direct interaction
   between bilirubin and EGFP‐tagged WNK1 (15 min) in lysates from
   EGFP‐WNK1 expressing HEK293T cells (n = 6 per group). F) MST
   experimental traces of full‐length hWNK1 with biliverdin, with vertical
   bars indicating fluorescence changes (gray cold; blue hot). A formula
   for calculating Fnorm is also presented. G) MST demonstrating a direct
   interaction between biliverdin and EGFP‐tagged WNK1 (15 min) in lysates
   from EGFP‐WNK1 expressing HEK293T cells (n = 3 per group). In panels E)
   and G), EGFP in lysates from EGFP‐expressing HEK293T cells was
   co‐incubated with either bilirubin or biliverdin for 15 min as a
   control (n = 3/group). Error bars indicate mean ± SD.

   To validate the direct binding of bilirubin to WNK1, we performed a
   Microscale Thermophoresis (MST) analysis, which assesses the binding
   capacity of a compound by measuring the movement of proteins over a
   microscopic temperature gradient.^[ [107]^26 , [108]^27 ^] We
   overexpressed full‐length WNK1 fused with Green Fluorescent Protein
   (GFP) in HEK‐293 cells, incubated the lysates with bilirubin at varying
   dilutions, and subjected them to precise MST measurements in
   capillaries. This approach conclusively demonstrated a direct
   interaction between bilirubin and full‐length hWNK1, with a strong
   binding affinity (Kd = 67.2 nM, p < 0.0001) (Figure [109]2D,E). This
   binding concentration is significantly lower than the concentrations at
   which bilirubin exerts toxic effects in the central nervous system and
   is comparable to its physiological concentration in vivo. Furthermore,
   biliverdin, a structurally similar intermediate product generated
   before bilirubin formation, was used as a control. Biliverdin showed
   minimal binding (Kd = 29.7 µM, p = 0.3739) (Figure [110]2F,G).
   Additionally, compared to other members of the WNK kinase family, WNK1
   exhibited a significantly higher binding affinity to bilirubin (Kd =
   0.53 µM for hWNK2; Kd = 0.76 µM for hWNK3; Kd = 0.28 µM for hWNK4)
   (Figure [111]S8A, Supporting Information). Collectively, these findings
   indicated that WNK1 is a physiological target of bilirubin.

2.3. Bilirubin Directly Binds to the WNK1 Kinase Domain

   We hypothesized that the kinase domain of WNK1 is a preferred binding
   region for bilirubin. Given the availability of the crystal structure
   of the N‐terminal region (206–483) of hWNK1 (hWNK1‐N), which contains
   the kinase domain, we conducted molecular docking analyses to
   investigate the binding of bilirubin to hWNK1‐N.^[ [112]^28 ^] The
   Sitemap module predicted five potential bilirubin‐binding pockets
   within the hWNK1‐N region, including the ATP‐binding pocket centered on
   ANP, the ligand‐binding pocket centered on the ligand molecule 7AV, and
   binding pockets centered on Val318, Tyr420, and Ser469 (Figure
   [113]3A,B). The results demonstrate that bilirubin exhibited the
   highest affinity for the ATP‐ and substrate‐binding domains of hWNK1,
   with docking scores of −7.928 and −8.198, respectively
   (Figure [114]3C,D). In contrast, the scores for Sites 3–5 were
   relatively lower, at −2.729, −1.787, and −3.526, respectively (Figure
   [115]S9A–C, Supporting Information).

Figure 3.

   Figure 3
   [116]Open in a new tab

   Bilirubin binds to the WNK1 kinase domain. A) Interaction surface
   diagram of bilirubin with the hWNK1 kinase domain crystal structure
   (PDB ID 5TF9). B) Cartoon diagram illustrating the interaction of
   bilirubin with the hWNK1 kinase domain crystal structure, highlighting
   the binding pocket. C) Specific recognition of the bilirubin‐bound
   hWNK1 ATP‐binding pocket. D) Specific recognition of the
   bilirubin‐bound hWNK1 ligand‐binding pocket. E) Schematic
   representation of the hWNK1 protein domains. F) MST demonstrating a
   direct interaction between bilirubin and EGFP‐tagged WNK1 (1–190), WNK1
   (191–491), and WNK1 (491–2382) (15 min) in lysates from different
   domains of EGFP‐WNK1 expressing HEK293T cells (n = 3/group). G) MST
   demonstrating a direct interaction between bilirubin and EGFP‐tagged
   point mutations of WNK1 (1–491) (15 min) in lysates from different
   plasmids expressing HEK293T cells. In panels F) and G), EGFP in lysates
   from EGFP‐expressing HEK293T cells was co‐incubated with bilirubin for
   15 min as a control (n = 3/group). Error bars indicate mean ± SD.

   To further validate these findings, we constructed GFP‐tagged fusion
   proteins with different lengths of WNK1 truncations (1–491, 1–190,
   191–491, 491–2382) based on the WNK1 domains (Figure [117]3E) and
   analyzed their binding capacity with bilirubin using MST analysis.^[
   [118]^29 ^] The results indicated that the fragments containing the
   kinase domain of WNK1 (1–491 and 191–491) exhibited significantly
   higher binding affinities with bilirubin (Kd = 0.315 and 0.028 µM,
   respectively) than the non‐kinase segments (Kd = 1.828 µM for WNK1
   1–90; Kd = 3.189 µM for WNK1 491–2382) (Figure [119]3F).

   Molecular docking identified several crucial amino acid residues within
   binding site 1 of the kinase domain for the binding of WNK1 and
   bilirubin. These residues included Thr301 (T301), Met304 (M304), Lys351
   (K351), and Phe356 (F356) (Figure [120]3C). To investigate this
   further, we constructed WNK1‐N mutants by replacing these residues with
   alanine (A). The results showed that the T301A mutant was unable to
   bind bilirubin, whereas the M304A, K351A, and F356A mutants exhibited
   reduced binding affinity (Figure [121]3G). This finding highlights the
   critical role of these residues in bilirubin binding. Additionally,
   comparative analysis of WNK1 sequences across various mammalian species
   revealed that the kinase domain of WNK1 is highly conserved (Figure
   [122]S8B, Supporting Information). This suggests that the interaction
   between WNK1 and bilirubin is a universal and evolutionarily
   significant feature.

2.4. Bilirubin Activates WNK1 Kinase to Elevate Intracellular Chloride Ion
Levels

   Given the binding of bilirubin to the kinase domain of WNK1, we aimed
   to explore its effect on WNK1 with a particular focus on its kinase
   activity. To evaluate WNK1 kinase activity, we utilized an in vitro
   kinase enzyme assay system to quantify ADP production by the amino acid
   fragment 181–507 of human WNK1 (hWNK1) over a range of bilirubin
   concentrations. The kinase‐dead WNK1 mutant hWNK1(181–507 S382A) served
   as a negative control (Figure [123]S10, Supporting Information).^[
   [124]^30 ^] Our findings indicate that bilirubin activates the kinase
   activity of WNK1, with an EC50 value of 896 nM, while showing no effect
   on the kinase‐dead S382A WNK1 mutant (Figure [125]4A).

Figure 4.

   Figure 4
   [126]Open in a new tab

   Bilirubin activates WNK1 kinase to elevate intracellular chloride ion
   levels via the WNK1‐KCC2 Pathway. A) Activity of the purified hWNK1
   kinase domain with varying bilirubin concentrations, measured using the
   ADP‐Glo kinase assay. Kinase functional analysis of hWNK1 (181–507
   S382A) as a control. Measured luminescence values were normalized to
   reactions lacking bilirubin. Data are from triplicate measurements in
   three independent experiments. B) Western blotting analysis of OSR1
   phosphorylation (left) and SPAK phosphorylation (right) in the
   hippocampus of Wnk1^fl/fl and Wnk1^fl/fl,Thy1‐Cre mice treated with
   ACSF and bilirubin. C) Immunofluorescence staining of bilirubin‐induced
   [Cl⁻][i] increase in DIV8 primary cultured hippocampal neurons. MQAE
   staining (green) indicates [Cl⁻][i] (fluorescence intensity inversely
   proportional to intracellular chloride concentration), and hSyn
   staining (red) marks live neurons. Scale bars = 50 µm. D)
   Immunofluorescence staining of bilirubin‐induced [Cl⁻][i] increase in
   DIV8 primary cultured hippocampal neurons, with CellTracker staining
   (red) marking live cells and hSyn‐CRE (green) marking Wnk1^hSyn‐cKO
   neurons. Scale bars = 50 µm. E) Summary data showing the fold‐change in
   MQAE fluorescence intensity of neurons treated with different bilirubin
   concentrations as shown in C). F) Summary data showing the fold‐change
   in MQAE fluorescence intensity of neurons as shown in D). G) Western
   blotting analysis of KCC2 phosphorylation in the hippocampus of
   Wnk1^fl/fl and Wnk1^fl/fl,Thy1‐Cre mice treated with ACSF and
   bilirubin. H) Immunofluorescence staining of bilirubin‐induced [Cl⁻][i]
   changes in DIV8 primary cultured hippocampal neurons, with or without
   CLP257. MQAE staining (green) indicates [Cl⁻][i], and CellTracker
   staining (red) marks live cells. Scale bars = 50 µm. I) Summary data
   showing fold‐changes in MQAE fluorescence of neurons treated with
   different drugs as shown in panel H) and Figure [127]S14A (Supporting
   Information). Error bars indicate mean ± SD. *p < 0.05, **p < 0.01,
   ***p < 0.001, and ****p < 0.0001.

   To further investigate the impact of bilirubin on WNK1 kinase activity,
   we generated hippocampal neuron‐specific Wnk1 knockout mice. Given that
   conventional Wnk1 knockout mice are embryonically lethal, we employed
   the CRISPR/Cas9 method to create mice carrying the floxed Wnk1 gene.
   Then we crossed Thy1‐Cre mice with Wnk1^flox/flox mice to generate
   hippocampal neuron‐specific Wnk1 knockout mice. We examined the effects
   of bilirubin on the phosphorylation of WNK1 downstream targets,
   Ste20‐related proline alanine‐rich kinase (SPAK), and oxidative
   stress‐responsive kinase 1 (OSR1), in the hippocampus of Wnk1^flox/flox
   and Wnk1^flox/flox,Thy1‐Cre mice. Upon bilirubin treatment,
   phosphorylation of OSR1 and SPAK was elevated in the hippocampus of
   Wnk1^flox/flox mice. However, in Wnk1^flox/flox,Thy1‐Cre mice,
   bilirubin treatment had minimal impact on OSR1 and SPAK
   phosphorylation. These findings indicate that the effect of bilirubin
   on OSR1 and SPAK phosphorylation is WNK1‐dependent (Figure [128]4B). To
   rule out the potential effects of osmolarity on the phosphorylation of
   WNK1, OSR1, and SPAK, we measured the osmolarity of different bilirubin
   concentrations. We found that low concentrations of bilirubin didn't
   change the osmolarity (Figure [129]S11, Supporting Information).

   Previous studies have demonstrated that WNK1 influences the
   intracellular chloride ion concentration through SPAK/OSR1 by
   modulating chloride cotransporters, such as the Na‐K‐2Cl cotransporter
   NKCC1 and K‐Cl cotransporters (KCCs).^[ [130]^17 , [131]^31 ^]
   Therefore, we hypothesized that bilirubin activation of WNK1 kinase
   activity might affect intracellular chloride levels. To test this
   hypothesis, we treated hippocampal neurons with varying concentrations
   of bilirubin for 30 min and observed a significant increase in
   intracellular chloride concentration. This was detected using the
   sensitive molecular probe N‐(Ethoxycarbonylmethyl)‐6‐Methoxyquinolinium
   Bromide (MQAE), as indicated by a reduction in MQAE fluorescence
   intensity (Figure [132]4C,E).^[ [133]^32 ^]

   To ascertain whether bilirubin regulates chloride ions through WNK1, we
   generated Wnk1 gene knockout primary cultured neurons. We achieved
   neuron‐specific deletion of Wnk1 in hippocampal neurons by transducing
   primary neurons isolated from Wnk1^flox/flox embryonic mice with a
   lentivirus expressing hSyn‐Cre‐EGFP.^[ [134]^33 , [135]^34 ^] Our
   results demonstrate that Wnk1 deficiency significantly reduced
   intracellular chloride ion ([Cl⁻][i]) levels in neurons. Importantly,
   following Wnk1 deletion, bilirubin stimulation failed to increase
   [Cl⁻][i] levels (Figure [136]4D,F), indicating that WNK1 is required
   for bilirubin‐induced increase in [Cl⁻][i] levels. Intracellular
   chloride concentration is crucial for modulating inhibitory synaptic
   transmission and, consequently, neuronal excitability. Consistent with
   this, we found that bilirubin induces neural hyperexcitability in the
   cochlear nuclei and hippocampus neurons, and the WNK1 inhibitor
   WNK‐IN‐11 effectively suppresses bilirubin‐induced hyperexcitability
   (Figure [137]S12A–D; Figure [138]S13A–C, Supporting Information).

   Since bilirubin‐induced activation of WNK1 kinase activity increases
   [Cl⁻][i] in neurons, we sought to elucidate the downstream effector
   proteins of WNK1 that are involved in this process. Neuronal [Cl⁻][i]
   is primarily regulated by the Cl⁻‐K⁺ symporter KCC2, with additional
   modulation by factors such as the Na⁺‐K⁺‐Cl⁻ co‐transporter NKCC1,
   GABA[A] receptors, and glycine receptors (GlyR), all of which
   contribute to [Cl⁻][i] homeostasis.^[ [139]^31 ^] To investigate the
   mechanism underlying bilirubin‐induced changes in [Cl⁻][i], we
   selectively targeted these proteins using pharmacological inhibitors
   (VU0463271 for KCC2, bumetanide for NKCC1, bicuculline for GABA[A]
   receptors, and strychnine for GlyR) and activated KCC2 using CLP257.^[
   [140]^35 , [141]^36 ^] Consistent with the role of KCC2 as a key
   regulator of Cl⁻ efflux in neurons, activation of KCC2 with CLP257
   counteracted the effect of bilirubin on [Cl⁻][i], preventing its
   elevation (Figure [142]4H). Conversely, other pharmacological agents
   did not produce similar effects (Figure [143]4I; Figure [144]S14A,
   Supporting Information). Additionally, we observed a decrease in the
   relative expression of phospho‐S940 KCC2, a phosphorylation event known
   to enhance KCC2 activity, in Wnk1^fl/fl mice hippocampal tissue
   following bilirubin treatment, whereas no such effect was observed in
   Wnk1^fl/fl,Thy1‐Cre mice (Figure [145]4G).^[ [146]^37 ^] The
   bilirubin‐induced effect on [Cl⁻][i] was not observed in primary
   astrocytes (Figure [147]S15A,B, Supporting Information) or the HEK 293T
   cell line (Figure [148]S15C, Supporting Information), both of which
   lacked KCC2 expression (Figure [149]S16A,B, Supporting Information),
   further confirming that the effects of bilirubin and WNK1 on [Cl⁻][i]
   elevation are dependent on KCC2. These findings suggest that KCC2
   significantly contributes to the elevation in [Cl⁻][i] of neurons
   following bilirubin binding to WNK1.

2.5. Bilirubin‐WNK1 Binding Suppresses LPS‐Induced Neuronal NLRP3 Activation

   The physiological concentration of bilirubin exerts broad
   anti‐inflammatory effects. Previous studies have reported that WNK1
   mediates chloride ion transport to inhibit the formation of NLRP3
   inflammasome in macrophages.^[ [150]^38 ^] Consequently, we
   investigated whether bilirubin targeting WNK1 is involved in its
   anti‐inflammatory mechanism by modulating NLRP3 in neurons. We employed
   two genetically modified mouse strains, biliverdin reductase (Blvra)
   knockout Blvra^−/− and UDP glucuronosyltransferase family 1 member A1
   (Ugt1a1) knockout Ugt1a1^+/− , to mimic the low‐ and high‐bilirubin
   conditions, respectively. Blvra^−/− mice lack the gene encoding
   biliverdin reductase A, an enzyme crucial for converting biliverdin to
   bilirubin, resulting in reduced serum bilirubin levels.^[ [151]^39 ^]
   Conversely, Ugt1a1^+/− mice exhibit decreased activity of
   UDP‐glucuronosyltransferase 1A1, the enzyme responsible for bilirubin
   conjugation, leading to elevated serum bilirubin levels.^[ [152]^40 ^]
   And we confirmed that bilirubin levels in the serum and hippocampal
   tissues of Blvra^−/− and Ugt1a1^+/− mice met the expected criteria
   (Figure [153]S17, Supporting Information). Moreover, Ugt1a1^+/− mice
   exhibited reduced KCC2 phosphorylation in the hippocampus, whereas
   Blvra^−/− mice showed increased KCC2 phosphorylation (Figure [154]S18,
   Supporting Information).

   Furthermore, a murine model of infection‐induced inflammation was
   established in neonatal wild‐type, Blvra^−/− , and Ugt1a1^+/− mice by
   intraperitoneal administration of lipopolysaccharide (LPS) (Figure
   [155]5A).^[ [156]^41 , [157]^42 ^] We assessed the expression levels of
   NLRP3 in the hippocampal region and the aggregation of the adaptor
   protein ASC in hippocampal neurons to indicate the level of
   inflammasome activation.^[ [158]^43 , [159]^44 ^] We observed elevated
   NLRP3 expression (Figure [160]5B) and increased ASC aggregation in
   low‐bilirubin Blvra^−/− mice compared to wild‐type controls, whereas
   these parameters were reversed in high‐bilirubin Ugt1a1^+/− mice
   (Figure [161]5C,D). Additionally, Caspase‐1 enzyme activity decreased
   in high‐bilirubin mice (Figure [162]5E), supporting the inhibitory role
   of bilirubin in NLRP3 activation.

Figure 5.

   Figure 5
   [163]Open in a new tab

   Bilirubin‐WNK1 suppresses LPS‐induced NLRP3 activation in the mouse
   hippocampus. A) Diagram of the experimental design. LPS (0.6 µg g^−1)
   or normal saline (NS) was injected intraperitoneally (i.p.) for 5
   consecutive days starting on day 7. Hippocampal tissues were harvested
   6 h after last injection. B) Western blotting analysis of NLRP3 in
   hippocampal tissues from WT, Blvra^−/− , or Ugt1a1^+/− mice after
   LPS/saline treatment. C) Immunofluorescence staining of ASC speck
   formation in hippocampal tissues from WT, Blvra^−/− , or Ugt1a1^+/−
   mice after LPS/saline treatment. β3‐Tubulin was used to label neurons
   and DAPI to stain cell nuclei. Main image scale bars = 50 µm, inset
   scale bars = 5 µm. D) Summary data showing the percentage of cells
   containing ASC specks as shown in C) (n = 6). (E) Relative activity of
   caspase‐1 in hippocampal tissues from WT, Blvra^−/− , or Ugt1a1^+/−
   mice after LPS/saline treatment (n = 3). F) Experimental design
   diagram. LPS (0.6 µg g^−1), LPS combined with bilirubin (5 µg g^−1), or
   normal saline (NS) was injected intraperitoneally for 5 consecutive
   days starting on day 7. G) Western blotting analysis of NLRP3 in
   hippocampal tissues from Wnk1^fl/fl,Thy1‐Cre mice after treatment. H)
   Immunofluorescence staining of ASC speck formation in hippocampal
   tissues from vehicle‐treated and Wnk1^fl/fl,Thy1‐Cre mice after
   treatment. β3‐Tubulin was used to label neurons and DAPI to stain cell
   nuclei. Main image scale bars = 50 µm, inset scale bars = 5 µm.
   Corresponding images are also provided in Figure [164]S21 (Supporting
   Information). I) Relative activity of caspase‐1 in hippocampal tissues
   from Wnk1^fl/fl and Wnk1^fl/fl,Thy1‐Cre mice after treatment (n = 3).
   Error bars indicate mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, and
   ****p < 0.0001.

   To investigate the role of WNK1 in mediating the inhibitory effect of
   bilirubin on NLRP3 activation, we used Wnk1^fl/fl,Thy1‐Cre mice which
   is the hippocampal neuron‐specific Wnk1 knockout mice. Mice were then
   administered intraperitoneal injections of saline, LPS, or a
   combination of LPS and bilirubin (Figure [165]5F). After five days of
   LPS treatment, we observed increased staining areas for GFAP (an
   astrocyte marker) and IBA1 (a microglia marker). Furthermore, a portion
   of the adaptor protein ASC (encoded by PYCARD) co‐localized with IBA1
   and GFAP. However, bilirubin does not affect ASC aggregation in either
   microglia or astrocytes (Figures [166]S19 and [167]S20, Supporting
   Information). Western blotting analysis and Caspase‐1 enzyme activity
   assays revealed that under LPS stimulation, the deletion of Wnk1
   abolished the inhibitory effect of bilirubin on NLRP3 inflammasome
   assembly, Caspase‐1 enzyme activity, as well as the cleavage‐mediated
   maturation of interleukin‐1β (IL‐1β) (Figure [168]5G–I; Figures
   [169]S21 and [170]S22A, Supporting Information). In contrast, neither
   bilirubin nor Wnk1 deficiency had a significant effect on the
   phosphorylation‐mediated activation of the NF‐κB p65 subunit upon LPS
   stimulation (Figures [171]S22B, Supporting Information). Building on
   the observation that bilirubin reduces Caspase‐1 activation, IL‐1β
   cleavage, and ASC speck formation, which are key markers of NLRP3
   inflammasome activation, our results underscore that bilirubin
   primarily inhibits the activation phase of the NLRP3 inflammasome. In
   conclusion, these findings demonstrate that bilirubin binding to WNK1
   effectively suppresses NLRP3‐mediated neuroinflammation.

2.6. Phosphoproteomic Analysis Reveals Potential Downstream Signaling
Pathways of Bilirubin WNK1 Binding

   Apart from inflammation regulation, WNK1 signaling pathways are also
   involved in angiogenesis, cancer proliferation, metastasis, and other
   functions.^[ [172]^17 , [173]^33 ^] Therefore, WNK1‐mediated kinase
   signaling was further explored to gain a deeper understanding of the
   biological functions of bilirubin‐WNK1 binding. We generated Wnk1
   specific knockdown in human SHSY‐5Y cells and performed a quantitative
   phosphoproteomic analysis (Figure [174]6A). Wnk1 knockdown resulted in
   significant alterations in the phosphorylation levels of 785 sites in
   577 proteins (fold change <0.667, p < 0.05) and 235 sites in 199
   proteins (fold change >1.5, p < 0.05) out of 6916 detected
   phosphorylation sites (Figure [175]6B,C), indicating that WNK1 has
   extensive downstream phosphorylation pathways. Enrichment analysis
   demonstrated that proteins with decreased phosphorylation levels were
   predominantly associated with biological processes such as
   “microtubule‐based processes,” “RNA processing,” and “cell cycle,” as
   well as other signaling pathways such as “RNA transport” and “longevity
   regulating pathway” (Figure [176]6E). Proteins with increased
   phosphorylation levels were enriched in biological processes such as
   “regulation of gene expression” and “mRNA export” and were associated
   with signaling pathways like “DNA damage response,” “circadian rhythm
   genes,” and “IL‐2 and IL‐18 signaling pathways” (Figure [177]6D), which
   correspond to the anti‐inflammatory effects of WNK1.

Figure 6.

   Figure 6
   [178]Open in a new tab

   Phosphoproteomic analysis reveals WNK1 downstream signaling pathways.
   A) Diagram of the experimental design. SH‐SY5Y cells were subjected to
   WNK1 knockdown using shRNA. Both normal and WNK1 knockdown SH‐SY5Y
   cells were collected for quantitative phosphoproteomics analysis. B)
   Volcano plot illustrating identified phosphorylation sites after WNK1
   knockdown by shRNA in SH‐SY5Y cells. C) Number of significantly
   regulated proteins and phosphorylation sites following WNK1 knockdown.
   D) Wiki Pathways (red) and biological processes (blue) in Gene Ontology
   enrichment analysis of proteins whose phosphorylation was upregulated
   after WNK1 knockdown. E) KEGG pathway analysis (red) and biological
   process (blue) in Gene Ontology enrichment analysis of proteins whose
   phosphorylation was downregulated after WNK1 knockdown. F) Overall
   phosphorylation change (−log[10](p‐value)) of a protein complex related
   to WNK1 downstream function, estimated as the change in phosphorylation
   on member proteins. G) The phosphorylation levels of 70 proteins were
   significantly altered in both WNK1 downstream and bilirubin‐responsive
   omics. H) CSNK1E and BAG3 among 70 proteins in G) show opposite changes
   in phosphorylation levels in the two omics.

   Although the phosphorylation of classical downstream targets of WNK1,
   such as NKCC1 (SLC12A1) (Thr212, Thr203/207) and NKCC2 (SLC12A2)
   (Thr105, Ser130), was not detected in the phosphoproteomic analysis of
   Wnk1‐deficient SHSY‐5Y cells, we identified a reduction in the
   phosphorylation levels of other significant proteins.^[ [179]^31 ^]
   Specifically, the plasma membrane calcium transporter ATP2B1 (Ser1211
   and Ser1219), sodium‐coupled neutral amino acid transporter SLC38A1
   (Ser52), and vesicular excitatory amino acid transporter SLC17A5
   (Ser16) showed decreased phosphorylation upon Wnk1 knockdown
   (Figure [180]6F). This suggests that WNK1 may affect cellular
   transmembrane substance transport through modification of these
   proteins. Importantly, the phosphorylation levels of critical nodes in
   the inflammatory signaling pathway, including Erk1 (Mitogen‐activated
   protein kinase 3, MAPK3) at Tyr204, Erk2 (Mitogen‐activated protein
   kinase 1, MAPK1) at Tyr187, and AP‐1 transcription factor c‐Jun (JUN)
   at Ser58 and Ser63, were significantly upregulated following WNK1
   knockdown (Figure [181]6F). This indicates that WNK1 broadly inhibits
   the activation of the inflammatory signaling pathways.

   To further explore the contribution of the WNK1‐mediated
   phosphorylation signaling network to the bilirubin signaling pathway,
   we integrated phosphoproteomic datasets that were influenced by
   bilirubin and Wnk1 knockdown. The analysis revealed that regardless of
   the specific phosphorylation sites, 70 proteins were co‐regulated by
   bilirubin and WNK1 (Figure [182]6G). Further analysis of the specific
   phosphorylation sites of these proteins revealed that only a few sites
   are simultaneously regulated by WNK1 and bilirubin. For example,
   phosphorylation of casein kinase CSNK1E at Ser408 and Bcl2‐associated
   athanogene BAG3 at Ser282 (corresponding to Ser283 in humans) increased
   after bilirubin stimulation, but decreased after Wnk1 knockdown
   (Figure [183]6H), suggesting that they may be downstream targets of the
   bilirubin‐WNK1 signaling pathway. Overall, these results suggest that
   bilirubin‐WNK1 binding induces a variety of important cellular
   biological effects, including anti‐inflammatory effects, and reveals
   the potential downstream molecular mechanisms of WNK1.

3. Discussion

   In recent years, bilirubin has garnered increased attention because of
   its diverse physiological functions. Importantly, bilirubin exhibits
   broad anti‐inflammatory effects and has shown significant potential for
   the treatment of inflammatory diseases.^[ [184]^4 , [185]^5 ^]
   Understanding the mechanism of bilirubin's anti‐inflammatory effects is
   of great significance for advancing the development and application of
   bilirubin‐based nanomedicines. In this study, we identified WNK1 as a
   direct binding target of bilirubin, which mediates the SPAK/OSR1‐KCC2
   pathway to elevate intracellular chloride ions and inhibit NLRP3
   activation in neurons. Consistent with our results, bilirubin has been
   shown to mitigate alum‐induced peritonitis by inactivating the NLRP3
   inflammasome.^[ [186]^45 , [187]^46 ^] WNK1 has been reported to be a
   negative regulator of the NLRP3 inflammasome in macrophages.^[ [188]^38
   ^] Additionally, we found that the phosphorylation levels of critical
   nodes in the inflammatory signaling pathway, such as IL‐2 and IL‐18,
   were significantly altered following WNK1 knockdown. These results
   suggest that the binding of bilirubin to WNK1 may represent a broad
   mechanism by which bilirubin exerts its anti‐inflammatory effects.
   Given the role of sustained neuroinflammation and NLRP3 inflammasome
   hyperactivation in various neurodegenerative diseases such as
   Alzheimer's disease (AD), Parkinson's disease (PD), and multiple
   sclerosis (MS), this study underscores the therapeutic potential of
   bilirubin and its WNK1 target in these neuroinflammatory disease
   management.^[ [189]^43 , [190]^45 , [191]^47 ^] However, the specific
   roles and regulatory mechanisms of the proposed bilirubin‐WNK1
   signaling pathway in different inflammatory diseases require further
   investigation.

   The phosphoproteomic analysis showed that bilirubin reduces the
   phosphorylation levels of WNK1 at residues Thr58 (Thr60 in humans),
   Ser183, and Ser185. WNK1 Thr58 phosphorylation has been implicated in a
   positive correlation with its kinase activity; however, this
   relationship is still under debate. For instance, a study by Ke et al.
   suggests that WNK1 regulation of intracellular chloride ion
   concentration is independent of Thr58 phosphorylation. The compound AK6
   reduces phosphorylation at Thr60 while concurrently increasing
   intracellular chloride ion concentrations, indicating a potential
   negative correlation between Thr60 phosphorylation and WNK1 kinase
   activity.^[ [192]^48 ^] Additionally, reduced phosphorylation on S183
   and S185 is not known to increase WNK1 kinase activity. This raises
   other possibility of regulatory mechanisms of bilirubin on WNK1
   activity independent of these residues. As shown in Figure [193]4A,
   bilirubin treatment can directly activate the amino acid region 181–507
   of hWNK1, suggesting that the active site of WNK1 can be modulated by
   bilirubin‐WNK1 binding, independent of upstream phosphorylation
   regulation. Small molecule activation of kinase function can occur
   through various mechanisms, including direct binding to the kinase
   active site, allosteric binding leading to conformational changes, or
   promotion of kinase oligomerization.^[ [194]^49 ^] Therefore, bilirubin
   may exert its effects through direct interaction with the kinase active
   site. Further exploration of the specific mechanism by which bilirubin
   activates the WNK1 kinase function will fill the gap in WNK1 agonist
   research and advance the biological applications of bilirubin as a
   therapeutic agent. Furthermore, we hypothesize that bilirubin binding
   may introduce a steric hindrance effect, preventing intracellular
   upstream kinases from phosphorylating critical residues such as Thr58,
   Ser183, and Ser185.

   Bilirubin exerts a wide range of biological effects, including
   physiological actions such as anti‐inflammatory effects and metabolic
   regulation, as well as toxicological effects when present at excessive
   concentrations (i.e., hyperbilirubinemia), which can lead to
   neurotoxicity and pruritus. Several studies have identified multiple
   bilirubin targets with diverse binding affinities. FABP1 binds
   bilirubin at concentrations between 1–4 µm. Additionally, UnaG, a
   modified member of the FABP family derived from eel, has a high
   affinity for bilirubin (Kd = 98 pM) and is widely used as a tool for
   bilirubin detection.^[ [195]^50 ^] FABP1 and PPARα (Kd = 5.13 µM) in
   liver and adipose tissue, modulating various energy metabolic
   pathways,^[ [196]^3 , [197]^51 ^] while MRGPRs (Kd = 54.4–92.9 µM) in
   peripheral sensory neurons mediate itch sensation.^[ [198]^3 , [199]^9
   ^] Additionally, bilirubin interacts with TRPM2 channels (EC50 =
   1.9 µM), exacerbating Ca^2⁺‐dependent brain injury during stroke.^[
   [200]^10 ^] In comparison, our study found that bilirubin binds
   strongly to WNK1, with a binding affinity across a broader
   concentration range (Kd = 67.2 nM), which is relevant to both its
   anti‐inflammatory effects at physiological concentrations and its
   neurotoxic effects at higher concentrations in the central nervous
   system. These results suggest that WNK1 may play a crucial role in
   mediating bilirubin's dual effects, which depend on its concentration,
   tissue specificity, and physiological conditions. WNK1 is an
   intracellular serine/threonine kinase involved in regulating various
   cellular functions. While we did not find direct evidence linking WNK1
   to other reported peripheral targets of bilirubin, this remains an
   interesting avenue for future research.

   As we have stated above, our previous research demonstrated that acute
   ischemic stroke triggers a transient increase in bilirubin levels
   (>25 µM) in the region surrounding the injury, where bilirubin acts as
   a volume neurotransmitter to activate TRPM2 channels (EC50 = 1.9 µM),
   exacerbating Ca^2⁺‐dependent excitotoxicity in the brain. TRPM2 is a
   highly Ca^2⁺‐permeable, non‐selective cation channel expressed in
   immune cells and neurons, mediating Ca^2⁺ influx that initiates
   signaling cascades responsible for neuronal excitability and injury
   associated with hyperbilirubinemia pathology. In this study, we
   discovered that bilirubin binds to WNK1 with a high binding affinity at
   low concentrations (Kd = 67.2 nM), exerting anti‐inflammatory effects
   in neurons. WNK1, an intracellular serine/threonine protein kinase,
   regulates ion transport (Na⁺, K⁺, Cl⁻) through the downstream kinases
   SPAK/OSR1, which maintains ion homeostasis. Our results suggested that
   the active site of WNK1 can be directly modulated by bilirubin‐WNK1
   binding, independent of upstream phosphorylation regulation such as
   that mediated by TRPM2. Besides, it is important to note that bilirubin
   is not the specific ligand for TRPM2. Structural derivatives of
   bilirubin, such as biliverdin, can activate TRPM2 channels. In
   contrast, WNK1 can bind bilirubin but not biliverdin, indicating that
   WNK1 is a highly selective target of bilirubin in the central nervous
   system.

   Moreover, astrocytes can exert either neuroprotective or neurotoxic
   effects depending on the microenvironment. Astrocytes engage in complex
   crosstalk with neurons in response to bilirubin. Studies have shown
   that bilirubin can activate the NLRP3 inflammasome and MAPK pathways in
   astrocytes in vitro, and it can also induce autophagy via
   mTOR/PKC/calcium signaling.^[ [201]^52 ^] Compared to astrocytes,
   neurons are more vulnerable to bilirubin exposure. Bilirubin influences
   various stages of neural circuit development, enhances synaptic
   transmission, and activates multiple ion channels, including TRPM2 and
   acid‐sensing ion channels, thus increasing neuronal excitability. In
   this study, we demonstrated that bilirubin binds to WNK1 at
   physiological concentrations, stimulating WNK1 kinase activity and
   suppressing LPS‐induced NLRP3 inflammasome activation. While both
   neurons and astrocytes can mount NLRP3 inflammasome responses, we
   observed that bilirubin‐induced changes in intracellular chloride
   concentration occur in hippocampal neurons but not in astrocytes, which
   lack KCC2 expression. These findings suggest that the
   bilirubin‐WNK1‐KCC2 pathway may represent a neuron‐specific signaling
   mechanism mediating the anti‐inflammatory effects of bilirubin in the
   hippocampus.

   In addition to its role in inflammation modulation, bilirubin also
   plays a role in blood pressure regulation, although the exact
   mechanisms are not fully understood.^[ [202]^3 , [203]^8 ^] Previous
   studies have identified WNK1 as a risk gene for salt‐sensitive
   hypertension, as it regulates the activity of the Na^+–Cl^−
   cotransporter (NCC) responsible for NaCl reabsorption in the distal
   convoluted tubule via the WNK1‐SPAK/OSR1 signaling cascade, which may
   be critically important in the development of hypertension.^[ [204]^16
   ^] Our findings suggest that bilirubin binding to WNK1 may be a novel
   and essential mechanism underlying its role in blood pressure
   regulation. Furthermore, phosphorylation enrichment analysis suggested
   that casein kinase CSNK1E and BAG3 may be downstream targets of the
   bilirubin‐WNK1 signaling pathway. Given the roles of CSNK1E and BAG3 in
   circadian rhythm regulation and the rhythmic fluctuations in the
   expression and phosphorylation levels of WNK1,^[ [205]^53 , [206]^54 ^]
   we hypothesize that the bilirubin‐WNK1 pathway may play a regulatory
   role in rhythmic processes.

   In conclusion, this study integrates phosphoproteomics with advanced
   molecular‐protein interaction techniques to identify WNK1 as a key
   target for bilirubin at physiological concentrations. Our results
   demonstrate that bilirubin binds to the kinase domain of WNK1,
   enhancing its activity and leading to the phosphorylation of downstream
   SPAK/OSR1, which in turn inhibits KCC2 activity and elevates
   intracellular chloride levels in neurons. Additionally, we employed an
   inflammation model to show that bilirubin binding to WNK1 effectively
   suppressed LPS‐induced NLRP3 inflammasome activation, thereby exerting
   anti‐inflammatory effects in mouse neurons. While WNK1 is highly
   conserved across multiple species, further studies involving human
   tissues or more clinically relevant models will be essential to
   validate the translational relevance and broader applicability of our
   findings. Moreover, the phosphoproteomic analysis of downstream
   signaling pathways of WNK1 has provided valuable insights into its
   diverse biological roles, particularly in the regulation of
   inflammatory pathways. Our findings establish WNK1 as a direct
   intracellular target of bilirubin's anti‐inflammatory effects, offering
   new perspectives on the physiological roles of bilirubin and suggesting
   potential therapeutic applications.

4. Experimental Section

Animal and Ethical Approval

   All experimental procedures were conducted in accordance with the
   Institutional Guidelines for Animal Care and Use and approved by the
   Institutional Animal Care and Use Committee of Shanghai Sixth People's
   Hospital, affiliated with Shanghai Jiao Tong University School of
   Medicine (Approval No:2020‐0285). Efforts were made to minimize animal
   discomfort and suffering throughout the study period. Blvra^−/− ,
   Ugt1a1^+/− and Wnk1^flox/flox C57BL/6 mice containing loxP sites
   flanking exon 2 were generated by Cyagen Biosciences, Inc. (Suzhou,
   China). The Wnk1^flox/flox C57BL/6 mice were bred with Thy1Cre
   transgenic mice to generate conditional Wnk1 knockout mice. Animals
   were housed under specific‐pathogen‐free (SPF) conditions with
   controlled temperature and a 12‐hour light‐dark cycle and were provided
   with standard food and water ad libitum.

TMT‐Tag Quantitative Phosphorylation Proteomics

   C57BL/6 mice at postnatal day 10 (P10) were euthanized on ice, and
   their cochlear nuclei were dissected and incubated in artificial
   cerebrospinal fluid (ACSF), with or without 3 µM bilirubin, followed by
   rapid freezing in liquid nitrogen. The ACSF composition was 124 mM
   NaCl, 5 mM KCl, 1.2 mM KH[2]PO[4], 1.3 mM MgCl[2], 2.4 mM CaCl[2],
   24 mM NaHCO[3,] and 10 mM glucose (pH 7.4).

   Tissue samples were ground in liquid nitrogen, lysed in urea buffer
   with protease inhibitors, sonicated, and centrifuged. Protein
   concentration was determined using the BCA assay. Trypsin digestion was
   performed, followed by TMT labeling, according to the manufacturer's
   protocol. Tryptic peptides were fractionated using high‐pH
   reverse‐phase HPLC and phosphopeptide enrichment was conducted using
   IMAC microspheres. LC‐MS/MS analysis was performed using a Q Exactive
   Plus instrument, with MS/MS data searched against a specified database
   using MaxQuant, employing predefined parameters for enzyme specificity,
   mass tolerances, modifications, and false discovery rate thresholds.

Bioinformatics Analysis

   Phosphoproteomic analysis identified 13495 phosphorylation sites
   distributed across 3792 proteins, with 11116 sites on 3474 proteins
   containing quantitative information. To ensure data reliability, we
   filtered the identified sites using a localization probability
   threshold of >0.75. After filtering, 10246 phosphorylation sites on
   3540 proteins were retained, including 9404 sites on 3325 proteins with
   quantitative information. This dataset was subsequently used for
   bioinformatics analysis.

   Differential phosphorylation sites were identified based on a
   fold‐change threshold of 1.5 and a t‐test p‐value < 0.05. Using these
   criteria, we found that 126 phosphorylation sites exhibited increased
   modification levels, and 469 sites showed decreased modification levels
   in the B/C (Bilirubin Treated/Control) comparison group.

   Proteins with quantitative information were subjected to systematic
   bioinformatics analysis, including protein annotation, functional
   classification, functional enrichment analysis, and cluster analysis
   based on enrichment results.

   For the quantification of significantly differential phosphorylation,
   each phosphorylation site was analyzed in three biological replicates.
   The average quantitative value for each site across three replicates
   was calculated for both groups. The ratio of the average values between
   the two groups was used as the relative quantification of differential
   modification (Ratio). Quantitative values were log2‐transformed to
   approximate normal distribution, and statistical significance was
   evaluated using a two‐tailed t‐test. A p‐value < 0.05 combined with a
   fold change >1.5 was considered as a significant increase, whereas a
   fold change <1/1.5 was considered as a significant decrease.

   For the assessment of quantification reproducibility, the
   reproducibility was evaluated using principal component analysis (PCA),
   relative standard deviation (RSD), and Pearson's correlation
   coefficient across biological replicates.

Plasmid Construction

   All WNK1 clones were derived from the original untagged hWNK1 cDNA
   ([207]NM_018979), which was modified and subcloned into the pcDNA3.1,
   referred to as “full‐length WNK1” throughout this study. Truncated
   constructs, including 1–190aa, 191–491aa, 1–491aa, and 491–2382aa
   fragments, were generated by PCR using the full‐length template and
   subsequently ligated into linearized pcDNA3.1 vectors. WNK2 was
   obtained from Addgene Plasmid #24569, and WNK3 was obtained from
   Addgene Plasmid #81723. Both were subjected to base modifications and
   cloned into pcDNA3.1. The WNK4 clone was derived from hWNK4 cDNA
   ([208]NM_032387.5), modified, and subcloned into the pCMV3‐C‐GFPSpark
   vector.

Cell Culture and Transfection

   Human embryonic kidney cells (HEK‐293T, ATCC: CRL‐11268) and SH‐SY5Y
   cells (ATCC: CRL‐2266) were cultured in Dulbecco's Modified Eagle
   Medium (DMEM, cat. no. 52100–39, Thermo Fisher Scientific) supplemented
   with 10% fetal bovine serum (FBS, cat. no. F7524, lot no. 022M3395,
   Sigma‐Aldrich) and 1% (v/v) streptomycin/penicillin (cat. no. L0022;
   Biowest) in a humidified atmosphere containing 5% CO₂ at 37 °C. Plasmid
   transfection was performed using Lipofectamine 3000 Transfection
   Reagent (cat. no. L3000075, Thermo Fisher Scientific). After
   transfection, the culture medium was replaced with a complete medium.
   To ensure consistency, the cells were characterized after one passage
   in the same culture medium to confirm maintenance of the desired
   characteristics and functionality.

Wnk1 Knockdown Stable Cell Line Generation

   Lentiviral constructs harboring short hairpin RNAs (shRNAs) targeting
   hWnk1 were used for the knockdown. The sequences used were
   1‐GCTGCGTATTGAAGATATTAA and 2‐CGTCAGTATCAGTCCCTATAA. The pLKO.1
   lentiviral expression vector containing the puromycin resistance gene
   was used to co‐express individual shRNAs. Recombinant lentiviral
   particles were generated by the transient transfection of plasmids into
   HEK‐293T cells. Stably transfected SH‐SY5Y cells were established using
   the lentiviral constructs. Following viral infection for 24 h, the
   medium was replaced with a complete medium supplemented with 2 µg mL^−1
   puromycin. After an additional 24‐hour incubation period with fresh
   puromycin‐containing medium, surviving cells were collected, expanded,
   and assessed for knockdown efficiency.

Chemical Structures

   Chemical structures were drawn using ChemDraw software.

Western Blotting

   Cells or tissues were lysed by grinding in RIPA buffer (cat. no. PC101;
   Epizyme Biotech) supplemented with a protease inhibitor cocktail (cat.
   no. 5872S; Cell Signaling Technology) at 4 °C. The lysate was
   subsequently centrifuged at 16000 × g for 30 min at 4 °C and the
   resulting supernatant was collected. The total protein concentration
   was determined using the Pierce BCA Protein Assay Kit (cat. no. 23 225,
   Thermo Fisher Scientific). The samples were denatured in a Protein
   Sample Loading Buffer (cat. no. LT101; Epizyme Biotech) at 95 °C and
   subjected to SDS‐polyacrylamide gel electrophoresis. Following
   electrophoresis, the proteins were transferred onto a nitrocellulose
   membrane (cat. no. HATF00010; Millipore) using a transfer buffer
   composed of 25 mM Tris, 190 mM glycine, and 20% methanol. Membranes
   were blocked with blocking buffer for western blotting (cat. no. P0252;
   Beyotime Biotechnology) at room temperature for 15 min, followed by
   overnight incubation with the primary antibody at 4 °C. After primary
   antibody incubation, the membranes were washed and incubated with
   secondary antibody at room temperature for 1 h. Protein bands were
   visualized using a Bio‐Rad chemiluminescence detection system upon the
   addition of a chemiluminescent substrate (Pierce ECL Western Blotting
   Substrate, cat. no. 32 106, Thermo Fisher Scientific).

Immunofluorescence

   The mice were anesthetized with 1% pentobarbital (0.04 mL/10 g) before
   the sternum was cut to expose the heart. A needle was inserted into the
   left ventricle to perfuse the pre‐oxygenated cold ACSF with
   low‐molecular‐weight heparin. Perfusion was continued until the tail,
   limbs, and liver turned pale (while the heart was kept beating), at
   which point the solution was switched to lavage buffer containing 4%
   paraformaldehyde. The brain tissue was fully fixed by perfusion for
   ≈15 min. Nuclear staining was performed using
   4′,6‐diamidino‐2‐phenylindole (DAPI) (2 mg mL^−1, Sigma).

Molecular Docking and Virtual Screening

   The crystal structure of WNK1 (PDB ID: 5TF9) was retrieved from the
   Protein Data Bank. Molecular docking studies were conducted using the
   Schrödinger Maestro software suite (Schrödinger, 2020–3). The protein
   was prepared using the Protein Preparation Wizard to address missing
   residues, remove water molecules, and optimize hydrogen bonding and
   energy using the OPLS2005 force field with a 0.30 Å RMSD.

   The Receptor Grid Generation module was employed to define the grid
   files. Grid file 1 (site 1) was generated with ANP at the center to
   represent the ATP‐binding pocket, whereas grid file 2 (site 2) was
   centered around ligand molecule 7AV to represent the ligand‐binding
   pocket. Additionally, the Sitemap module was used to predict the
   binding pocket of hWNK1. Grid files 3–5 (sites 3–5) were generated
   using VAL318, TYR420, and SER469.

   The 2D structure of bilirubin (C33H36N4O6) was subjected to energy
   minimization using the LigPrep module (OPLS2005 force field, 0.30 Å
   RMSD) to generate the corresponding 3D structure. Molecular docking was
   then performed using the Glide module in high‐precision (XP) mode,
   facilitating the docking of receptor and ligand molecules through
   geometric and energy matching.

   Further molecular docking and virtual screening of bilirubin, which has
   a potential binding affinity to WNK1, were conducted using Guangzhou
   Yinfo Information Technology. The structural prediction of WNK1 was
   accomplished using the AlphaFold multimer in ColabFold, while the
   potential binding sites of WNK1 were predicted using the online tool
   POCASA 1.1. Molecular docking‐based virtual screening between WNK1 and
   bilirubin was performed using DOCK 6.9.

MQAE Intracellular Cl⁻ Concentration Measurements

   Intracellular chloride concentration ([Cl⁻][i]) was measured using the
   MQAE fluorescent dye (Life Technologies) following established
   protocols (Zhang, Lee et al. 2015). Cultured primary hippocampal
   neurons were loaded by incubating cells in artificial cerebrospinal
   fluid (ACSF) containing 5 mM MQAE for 30 min at 37 °C. Imaging was
   subsequently performed using Zeiss confocal microscopy.

Primary Neuronal Culture

   Hippocampal neurons were isolated from C57BL/6 embryonic day 18 (E18)
   embryos, as previously described (Lee, Abiraman et al. 2021). Briefly,
   neurons were plated at a density of 5 × 10⁶ cells per dish and cultured
   in a neurobasal medium supplemented with 2% B27, 0.6% glucose, 1%
   GlutaMAX, and 0.5% penicillin/streptomycin. Cultures were maintained at
   37 °C in a humidified atmosphere containing 5% CO₂ until further
   experimentation. The neuronal cell culture reagents were purchased from
   Invitrogen.

Microscale Thermophoresis (MST)

   In HEK293T cells, proteins tagged with green fluorescent protein (GFP)
   were expressed. The cells were lysed using the M‐PER Mammalian Protein
   Extraction Reagent (cat. no. 78505, Thermo Scientific) supplemented
   with protease inhibitors. Lysates were centrifuged at 12000 × g for
   15 min at 4 °C, and the supernatant containing proteins was diluted
   10‐fold with 1x PBST. Protein fluorescence intensity and solubility
   were assessed using capillary scanning at 60% LED power.

   For the binding assay, the proteins were mixed with ligands and
   incubated for 15 min at room temperature before being loaded into
   Monolith NT.115 Capillaries or NT.115 Premium Capillaries (NanoTemper
   Technologies, MO‐K022 and MO‐K025). MST measurements were performed on
   a Monolith NT.115 instrument (NanoTemper Technologies) at 25 °C.
   Instrument parameters included no more than 80% excitation power and a
   medium MST power. Data analysis was conducted using the MO. Affinity
   Analysis software (NanoTemper Technologies).

MS‐CETSA Sample Preparation

   Cellular Thermal Shift Assay (CETSA) was performed following
   established procedures.^[ [209]^21 , [210]^55 ^] Briefly, SH‐SY5Y cells
   were treated with a serum‐free culture medium containing 6 µM bilirubin
   or an equivalent amount of DMSO for 30 min. After three washes with
   PBS, cells were collected and subjected to a 3‐minute heat shock at
   55 °C, followed by immediate cooling on ice. The cells were lysed by
   repeated freeze‐thaw cycles in liquid nitrogen, and lysates were
   centrifuged at 20000 × g for 20 min at 4 °C to separate soluble
   proteins from denatured precipitated proteins. The supernatant
   containing the soluble proteins was collected for subsequent mass
   spectrometry analysis.

LC‐MS Analysis

   For peptide labeling, iTRAQ labeling reagent was used according to the
   number of samples. Briefly, peptide samples were dissolved in 0.5 m
   TEAB and added to the iTRAQ reagents, then left at room temperature for
   2 h. Then, peptide fractionation was performed using a Shimadzu LC‐20AB
   system with a 5 µm 4.6 × 250 mm Gemini C18 column. Dried peptide
   samples were reconstituted with mobile phase A (5% ACN, pH 9.8),
   injected, and eluted using a gradient of mobile phase B (95% ACN, pH
   9.8). Peaks were monitored at 214 nm, collected into 20 fractions, and
   freeze‐dried. The reconstituted samples were centrifuged and injected
   into a Thermo UltiMate 3000 UHPLC system. The peptides were ionized
   using a nanoESI source and analyzed on a Q‐Exactive HF X mass
   spectrometer (Thermo Fisher Scientific) in the DDA mode. MS data were
   acquired using a top 20 data‐dependent acquisition protocol at 60000
   resolutions in MS1 and 15000 resolutions in MS2.

Kinase Activity Measurements

   WNK1 kinase activity was measured using the ADP‐Glo Kinase Assay
   (Promega, Cat. V9101) following the manufacturer's instructions.
   Purified protein was incubated with bilirubin across a range of 12
   serially diluted concentrations, beginning at a maximum concentration
   of 20 µm. After the incubation period, luminescence was recorded using
   a microplate reader (Synergy H1M, BioTek)

Caspase‐1 Activity Measurements

   Caspase‐1 activity was measured using a Caspase‐1 Activity Assay Kit
   (Beyotime Biotechnology, Haimen, China). Briefly, 50 µg of protein from
   the mouse hippocampal tissue was added to a reaction buffer containing
   Ac‐YVAD‐ρNA. The absorbance was read at 405 nm using a microplate
   reader, and caspase‐1 activity was normalized to that of the control.

Measurement of Unbound Bilirubin Using UnaG Fluorescence

   To establish a standard calibration curve, reaction mixtures were
   prepared containing 50 µL of UnaG solution (1 µm) and 50 µL of
   bilirubin solutions at graded concentrations (1.4278, 0.7139, 0.3573,
   0.1785, 0.0893, and 0 µm). After a 10‐minute incubation, fluorescence
   intensity was measured with a microplate reader (Synergy H1M, BioTek)
   at excitation and emission wavelengths of 485 and 528 nm. Mouse serum
   samples were diluted and incubated directly with UnaG, while
   hippocampal samples were homogenized in native lysis buffer (Beyotime,
   Cat.No.: P0013), centrifuged, and the supernatant incubated with UnaG.
   UCB levels were extrapolated from the standard curve, and hippocampal
   UCB concentrations were normalized to protein content and compared to
   wild‐type mouse levels.

Statistical Analysis

   Generally, all continuous variables were analyzed using parametric
   tests (two‐tailed unpaired T‐test assuming equal variances when
   comparing 2 groups, or one‐way ANOVA and post‐hoc Tukey LSD tests for
   >2 group comparisons). For proteomic analyses, we used un‐adjusted p
   values + log2(FC) thresholds to prioritize differentially abundant
   proteins. Power calculations were not performed for individual
   experiments. Specific statistical tests used for individual experiments
   are detailed in the figure legends. Sample size (n) for each
   statistical analysis and data presentation (mean ± SD) are provided in
   the figure legends. All statistical analyses were performed using
   GraphPad Prism 10.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   L.M., J.L., Q.Y., and Z.L. contributed equally to this work. S.Y.,
   C.L., and F.L. conceptualized the study; G.Y., R.S., J.L., and Q.Y.
   provided conceptual guidance; methodology was developed by F.L., L.M.,
   Z.L., Q.Y., J.L., K.S., and Y.C; investigation was conducted by L.M.,
   Z.L., B.K., Y.W., and Y.C; data curation was performed by L.M. and
   F.L.; formal analysis was carried out by F.L.; visualization was
   executed by C.W., H.Y., and K.S.; resources were provided by S.Y.,
   L.M., J.L., C.L., and Q.Y.; funding acquisition was managed by C.L.,
   S.Y., and G.Y.; project administration was led by S.Y. and C.L., with
   coordination by G.Y.; supervision was provided by S.Y., C.L., and G.Y.;
   L.M., F.L., and C.L. wrote the original draft and prepared it; F.L.,
   C.L., J.L., Q.Y., and R.S wrote te original draft and edited the
   original manuscript and reviewed it.

Supporting information

   Supporting Information
   [211]ADVS-12-2407349-s001.docx^ (109.7MB, docx)

Acknowledgements