MARK4 enhances stress granule formation under oxidative stress and increases tau accumulation Authors: Sho Nakajima, Kotone Watanabe, Grigorii Sultanakhmetov, Aoi Fukuchi, Keiya Ito, Sawako Shimizu, Taro Saito, Akiko Asada, Kanae Ando Published: 2026 ABSTRACT AbstractOveractivation of Microtubule affinity regulating kinase 4 (MARK4) is believed to contribute to Alzheimer's disease pathogenesis. MARK4 promotes the accumulation of the microtubule‐binding protein tau, thereby enhancing tau‐induced neurodegeneration. However, the underlying mechanisms by which MARK4 enhances tau accumulation are not fully understood. T‐cell intracellular antigen 1 (TIA1), a critical regulator of stress granule (SG) formation, has been suggested to initiate tau abnormality. Here, we report that MARK4 and TIA1 synergistically induce stress granule (SG) formation. MARK4 is localized in SGs with TIA1 in mammalian cultured cells and primary neurons. MARK4 suppresses TIA1 dimerization and enhances SG formation under oxidative stress. Co‐expression of MARK4 and TIA1 promotes tau accumulation, and knockdown of a fly ortholog of TIA1 suppressed tau toxicity in a Drosophila model. These results identify MARK4 as a novel regulator of SG formation and suggest a mechanistic link between oxidative stress and tau pathology.Keywords: Drosophila melanogaster, microtubule affinity regulating kinase 4, neurodegeneration, phosphorylation, stress granules, tau, T‐cell intracellular antigen 1 Abbreviations RasGAP SH3 domain binding protein 1 liquid–liquid phase separation Microtubule affinity regulating kinase stress granule T‐cell intracellular antigen 1 SGs are membrane‐less organelles composed of untranslated mRNAs, RNA‐binding proteins, and various proteins gathered by liquid–liquid phase separation (LLPS) [1]. SGs are formed transiently in response to stress and protect cells by interacting with cellular homeostatic processes; however, their persistence has been associated with the pathogenesis of cancer and neurodegenerative diseases [1, 2, 3]. SG formation in cancer cells enhances tumorigenesis and promotes their adaptation to harsh microenvironments [4]. In neurodegenerative diseases that are associated with deposition of protein aggregations, SGs have been suggested to promote conformational changes and abnormal assemblies of disease‐causing proteins [5, 6, 7, 8]. SGs are heterogeneous in their components and properties [9], and SGs containing the RNA‐binding protein T‐cell intracellular antigen 1 (TIA1) have been specifically linked to the abnormality of tau protein, whose accumulation causes neuronal loss in a number of neurodegenerative diseases, including Alzheimer's disease [10, 11, 12, 13, 14]. Hyperphosphorylated tau inclusions colocalize with TIA in cultured cells and postmortem brains of tauopathy patients [10, 11]. TIA1 promotes SG formation [12] and increases toxic tau oligomers [15], and a reduction in TIA1 protects against neurodegenerative phenotypes in the PS19 tau mice [13, 14]. TIA1 activity is regulated by oxidation, and reactive oxygen species (ROS) such as H2O2 oxidize the SG‐nucleating protein TIA1 and inhibit SG assembly [3]. However, the molecular details underlying the enhancement of tau toxicity via TIA1‐containing SGs are not fully elucidated. Microtubule affinity regulating kinase (MARK) 4 belongs to the evolutionarily conserved Par‐1 family that regulates the cytoskeleton, cell polarity, and energy metabolism [16]. Aberrant expression of MARK4 is associated with several types of cancer and with Alzheimer's disease [17, 18, 19, 20, 21, 22]. MARK4 phosphorylates tau at Ser262 and Ser356 located in the microtubule‐binding repeats [23], and phosphorylation at these sites is critical for tau to gain neurotoxicity [24, 25, 26, 27, 28, 29]. Mammals express four MARK family members (MARK1–4) with similar functional domains [30, 31]. Interestingly, although all the members of the MARK family are expressed in the brain and increase tau phosphorylation in the microtubule‐binding repeats, only MARK4 has been genetically linked to AD [18, 19, 20, 21], and MARK4 enhances tau toxicity more than other members in a Drosophila model [32]. These reports suggest that upregulation of MARK4 enhances tau toxicity via a MARK4‐specific, yet‐to‐be‐identified mechanism. Here, we report that MARK4 is a component of SGs in mammalian cultured cells and primary neurons. Our results suggest that the interaction between MARK4 and TIA1 regulates SG formation and MARK4 and tau accumulation. These results suggest a novel function of MARK4 in stress responses and tau abnormality. Materials and Methods Plasmids Myc‐MARK4 and Myc‐MARK2 were generated with polymerase chain reaction (PCR)‐based methods using the plasmids described previously [33]. MARK4 deletion constructs (MARK41–709, MARK41–368, MARK4Δ386–709, MARK4386–709, MARK4Δ360–400,Δ580–586) were generated with Myc‐MARK4 by using PCR‐based methods. MARK4D199A and MARK4T214AS218A were prepared by using PCR‐based site‐directed mutagenesis. mCherry‐MARK4 was generated with Myc‐MARK4 by using the Gibson Assembly Cloning Kit (NEB, E5510S). Tau (0N4R) was a kind gift from Dr. Mike Hutton (Mayo Clinic, Jacksonville). TDP‐43 was a kind gift from Dr. Takashi Nonaka (Tokyo Metropolitan Institute of Medical Science) [34]. pEGFP‐TIA1 is a kind gift from Dr. Mutsuhiro Takekawa (The University of Tokyo) [3]. Reagent and antibodies Anti‐MYC antibody (4A6, Merck Millipore, Burlington, MA, USA, 05‐724), anti‐HA antibody (Proteintech, Rosemont, IL, USA, 51 064‐2‐AP), anti‐MARK4 antibody (Proteintech, 20 174‐1‐AP), anti‐TIA1 antibody (Proteintech, 12 133‐2‐AP), anti‐G3BP1 antibody (Proteintech, 13 057‐2‐AP), anti‐FMRP antibody (DSHB, University of Iowa, IA, USA, 7G1‐1), anti‐DCP1A antibody (SIGMA Aldrich, St. Louis, MO, USA, WH0055802M6), anti‐GFP antibody (Merck Millipore, 11 814 460 001), anti‐tau antibody (T46, Thermo Fisher Scientific, Waltham, MA, USA, 13‐6400), anti‐phosphorylated tau antibody (pS262‐tau, Abcam, Cambridge, UK, ab92627), anti‐GAPDH antibody (Novus Biologicals, Centennial, CO, USA, NB100‐56875), anti‐actin antibody (Proteintech, 66 009‐1‐lg), secondary antibodies labeled with Alexa 488 or Alexa 546 (Thermo Fisher, A‐11008, A‐11001, A‐11035, A‐11010) were purchased. Sodium arsenite (Sigma‐Aldrich, 57 654 633), H2O2 (Fujifilm Wako, Osaka, Japan, 081‐04215), and 1,6‐hexanediol (Fujifilm Wako, 087‐00432) were purchased. Cell culture and transfection HeLa cells (RRID:CVCL_0030) were purchased from the RIKEN BioResource Center, and authenticated by the supplier upon acquisition. Master stocks were established immediately after receipt and cryopreserved in liquid nitrogen. All experiments were performed using mycoplasma‐free cells thawed from these original frozen stocks within the past three years. Cells were cultured in DMEM (high glucose) supplemented with 10% deactivated FBS, penicillin, and streptomycin at 37 °C in 5% CO2. For immunoblotting and immunocytochemistry, HeLa cells were transfected with the indicated plasmids using Lipofectamine 3000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's protocol. Primary neuronal cultures were prepared from embryos obtained from pregnant mice purchased from Sankyo Labo Service Cooperation. Inc (Toyo, Japan). Pregnant dams were received and housed under standard animal facility conditions consisting of a 12 h light/12 h dark cycle with ad libitum access to food and water in standard ventilated cages. Pregnant mice were euthanized by cervical dislocation performed by appropriately trained personnel in accordance with institutional guidelines, and embryos were collected immediately thereafter. Embryonic sex was not determined. Animals used for experiments were maintained in the laboratory following protocols for ethical animal handling established by Tokyo Metropolitan University (approval number: A8‐003). All efforts were made to minimize animal suffering and to reduce the number of animals used. Primary neurons were prepared from the mouse brain cortex at embryonic day 15 (E15) and plated on poly‐l‐lysine‐coated dishes in DMEM and Ham's F‐12 (1:1) supplemented with 5% fetal bovine serum, 5% horse serum, 100 U/mL penicillin, and 0.1 mg/mL streptomycin at a density of 1.0 × 106 cells/ml. The medium was replaced with Neurobasal medium supplemented with 2% B‐27 (Invitrogen), 0.5 mm l‐glutamine, 100 U/mL penicillin, and 0.1 mg/mL streptomycin after 4 h of plating. Immunofluorescent staining HeLa cells cultured on glass cover slips were fixed with 4% paraformaldehyde in phosphate‐buffered saline (PBS) for 15 min at room temperature. For induction of SG, cells were treated with 0.5 mm sodium arsenite or 1 mm H2O2 for one hour before fixation. After blocking with PBS containing 5% skim milk, cells were incubated with each antibody in PBS containing 5% skim milk and 0.1% Triton X‐100 overnight at 4 °C followed by incubation with secondary antibodies coupled to Alexa 488 or Alexa 546 (Thermo Fisher, A‐11008, A‐11001, A‐11035, A‐11010) (1:500). Nuclei were stained with DAPI (Thermo Fisher). Cells were examined under a fluorescence microscope BZ‐700 (Keyence, Osaka, Japan). SDS/PAGE and western blotting HeLa cells were lysed in 20 mm Tris‐Cl, pH 8.0, 150 mm NaCl, 1 mm EDTA, 0.1% SDS, 1% NP‐40, 1 mm dithiothreitol, 10 μg/mL leupeptin, 0.2 mm Pefabloc SC, 0.2 mm NaF, and 0.2 mm deoxycholate. After brief centrifugation at 100 g for 5 min, the supernatants were mixed with Laemmli's SDS‐sample buffer containing 0.0625 m Tris/HCl (pH 6.8), 2% SDS, bromophenol blue, 10% glycerol, and 5% 2‐mercaptoethanol. After Laemmli's SDS/PAGE was carried out, proteins were transferred to PVDF (Millipore) membranes using a semi‐dry blotting apparatus or submarine apparatus. Membranes were probed with the primary antibodies, followed by a secondary antibody conjugated with horseradish peroxidase. The reactions were detected using a Millipore Immobilon western chemiluminescent HRP substrate (Millipore). FRAP analysis FRAP was performed with a confocal laser‐scanning LSM710 microscope (Carl Zeiss, Jena, Germany) with a 63×/1.4 NA objective lens. Images were acquired every 500 ms. Bleaching was conducted for 100% laser strength with 50 iterations. GFP intensity of the bleached area was normalized using the intensity of a neighboring cell after background subtraction. Recovery curves were generated and analyzed using imagej and GraphPad Prism 10. Fly stocks Flies were maintained at 25°C under light–dark cycles of 12:12 h, and food vials were changed every 2–3 days. The transgenic fly line carrying the human 0N4R tau, which has four tubulin‐binding domains (R) and no N‐terminal insert (N), is a kind gift from Dr. M. B. Feany (Harvard Medical School) [35]. GMR‐GAL4 and UAS‐GFP were obtained from the Bloomington Drosophila Stock Center (Indiana University). UAS‐MARK4 strain was reported previously [33]. UAS‐ROX8 RNAi (HMS00472) was obtained from NIG‐fly (National Institute of Genetics, Japan). Expression of all proteins was induced by the GMR‐GAL4 driver in the eyes. All experiments were performed using age‐matched female flies. The study was approved by the Research Ethics Committee of Tokyo Metropolitan University. Histological analysis To study the effect of MARKs on tau‐induced neurodegenerative phenotype, we analyzed neurodegeneration in the lamina structure 10 days after the explosion, as described previously (Iijima et al., 2010). Fly heads were fixated in Bouin's fixative for 48 h at room temperature, incubated for 24 h in 50 mm Tris/150 mm NaCl, and embedded in paraffin after gradual ethanol dehydration. Horizontal (from top to bottom of the head) 7‐ μm sections were prepared with standard hematoxylin and eosin staining. Brightfield images of the stained section were captured by the Keyence BZ‐X710 microscope. Neurodegeneration in the lamina region was counted as a vacuole area per total lamina size in ImageJ (NIH) using color threshold and particle measure functions. Heads from five to seven flies were analyzed for each genotype. Western blotting of Drosophila heads Western blotting of Drosophila samples was carried out as described [33]. Briefly, 15 heads of flies at 4–5 days after the eclosion per genotype were homogenized in 2× Laemmli Buffer, boiled for 5 min at 95 °C, centrifuged at 15 800 g for 5 min to remove debris, and separated by 10% acrylamide SDS/PAGE. Proteins were transferred to PVDF membranes (Millipore). Membranes were blocked in 5% skim milk in TBS‐T and probed with primary antibodies: Anti‐Myc Tag Antibody (clone 4A6, Sigma‐Aldrich), tau monoclonal antibody (T46, Invitrogen, 13–6400), anti‐pS262‐tau antibody (Abcam, ab92627), anti‐pS356‐tau antibody (Abcam, ab92682), anti‐actin antibody (Sigma‐Aldrich, A2066), anti‐pS202/pThr204‐tau antibody (AT8, Invitrogen, MN1020), anti‐pSer396‐tau antibody (Invitrogen, 44‐752G). Membranes were incubated with peroxidase‐conjugated anti‐mouse IgG (DAKO, P0447) and anti‐rabbit IgG (DAKO, P0399). Proteins were visualized using Immobilon Western Chemiluminescent HRP Substrate (Millipore) and detected by Fusion FX (Vilber). The integral intensity of bands was quantified by ImageJ (NIH). Western blot experiments were repeated at least three times with independent cohorts. qRT‐PCR Quantitative reverse transcription PCR was carried out as previously reported [32]. More than 30 flies for each genotype were collected and frozen. Heads were mechanically isolated, and total RNA was extracted using Isogen Reagent (NipponGene, Tokyo, Japan) according to the manufacturer's protocol with an additional centrifugation step (11 000 g for 5 min) to remove cuticle membranes prior to the addition of chloroform. Total RNA was reverse‐transcribed using PrimeScript Master Mix (Takara Bio, Shiga, Japan). qRT‐PCR was performed using TOYOBO THUNDERBIRD SYBR qPCR Mix (Osaka, Japan) on a Thermal Cycler Dice Real Time System (Takara Bio). The average threshold cycle value (CT) was calculated from at least three replicates per sample. Expression of genes of interest was standardized relative to rp49. Relative expression values were determined by the ΔΔCT method. Primers were designed using Primer‐Blast (NIH) or obtained from Fly Primer Bank (http://www.flyrnai.org/flyprimerbank). The following primers were used for RT‐PCR: Drosophila Rox8, forward 5′‐AGCCGAAGACAGACATCAGTT‐3′, reverse 5′‐CTCTCCGAATGGGGCGAAAG‐3′, Drosophila rp49, 5′‐GCTAAGCTGTCGCACAAATG‐3′, reverse 5′‐GTTCGATCCGTAACCGATGT‐3. Statistical analysis The images of protein domain structure were created in DOG 2.0 [36]. Data analysis was performed using Microsoft Excel and GraphPad Prism v9.4. Values bars are given as mean ± standard deviation. Data were analyzed by a one‐way ANOVA with Tukey's or Dunnett's multiple comparison test. If data did not pass the normality test, data were analyzed by Kruskal–Wallis test with Dunn's multiple comparisons. P‐values <0.05 were considered statistically significant. To compare the frequency of MARK4 localization to stress granules (SGs) between sodium arsenite‐treated and H2O2‐treated cells (Fig. 1E), cells containing TIA1‐positive SGs were identified, and the number of cells exhibiting granules positive for both TIA1 and MARK4 was counted. This value was divided by the total number of cells containing TIA1‐positive SGs to calculate the percentage of cells showing MARK4 localization to SGs. Data from three independent transfection experiments (biological replicates) were pooled for analysis, and statistical significance was assessed using the chi‐square test. MARK4 localizes to stress granules. (A) HeLa cells were transfected with myc‐tagged MARK4 and immunostained with anti‐Myc (MARK4, magenta) and antibodies against stress granule (SG) markers (green), such as anti‐ G3BP1, anti‐TIA1 (TIA1), and anti‐FMRP. Line plots of fluorescent intensity indicate colocalization of MARK4 and SG markers. Scale bars: 20 μm. (B) MARK4‐containing puncta are eliminated by 1,6‐hexanediol. Cells were treated with vehicle (top) or 5% 1,6‐hexanediol (lower panel) for 3 min and immunostained with anti‐Myc and anti‐FMRP antibodies. Scale bars: 20 μm. (C) MARK4 does not localize to the p‐body. Cells transfected with MARK4 were immunostained with anti‐MARK4 and anti‐DCP1A antibodies. Representative staining images and line plots. Scale bars: 10 μm. (D) Endogenous MARK4 localizes to SG in primary neurons. Mouse primary neurons were treated with 0.5 mm sodium arsenite for 1 h and immunostained with anti‐MARK4 (Magenta) and anti‐G3BP1 (green). Scale bars: 10 μm. (E) MARK4 localizes to stress granules induced by H2O2. HeLa cells transfected with MARK4 were treated with 0.5 mm sodium arsenite for 1 h or 1 mm H2O2 for 1 h and immunostained with anti‐Myc (MARK4, green) and anti‐TIA1 (magenta). TIA‐1 positive granules with (white squares) or those without MARK4 (yellow squares) were observed, and MARK4 was found more frequently in TIA1‐positive granules in the cells treated with H2O2. Scale bars: 20 μm. To compare the frequency of MARK4 localization to SGs in sodium arsenite‐treated cells and those in H2O2‐treated cells, the number of cells containing granules both positive for TIA1 and MARK4 was divided by the number of all the cells with TIA1‐positive granules. Sodium arsenite; n = 141, H2O2; n = 122; ****P > 0.001, chi‐square test). Results MARK4 is a component of stress granules We found that MARK4 shows punctate patterns in the cytoplasm in HeLa cells (Fig. 1A). These MARK4‐containing puncta were also positive for stress granule markers, TIA1, RasGAP SH3 domain binding protein 1 (G3BP1), or fragile X mental retardation protein (FMRP) (Fig. 1A). This punctate pattern disappeared when cells were treated with 1,6‐Hexanediol, which inhibits LLPS (Fig. 1B), suggesting that these puncta are membrane‐less organelles. Other membrane‐free organelles in the cytoplasm formed via LLPS include P‐bodies [37]. The puncta containing MARK4 were not stained with the antibody against DCP1A, a p‐body marker (Fig. 1C), suggesting that they are not likely to be p‐bodies. These results suggest that MARK4 proteins are localized to stress granules. We asked whether endogenous MARK4 in neurons localizes to stress granules. Immunostaining of primary neurons that were treated with sodium arsenite to induce stress granule formation revealed that MARK4 colocalizes with stress granule markers (Fig. 1D). MARK4 preferentially localizes to stress granules induced by oxidative stress Stress granule formation is induced by various stressors, while the molecular composition and formation mechanisms of SGs differ depending on stress conditions [9]. Sodium arsenite and hydrogen peroxide are both known to induce SGs, while hydrogen peroxide induces noncanonical types of SGs [38, 39, 40, 41]. We analyzed the localization of MARK4 to stress granules formed under different stress conditions. We found that, in the cells with sodium arsenite treatment, 50% of stress granules contained MARK4. In the cells treated with H2O2, more than 80% of stress granules contain MARK4 (Fig. 1E). These results indicate that MARK4 is a component of stress granules induced by H2O2. The spacer domain in MARK4 is required for its localization to stress granules Mammals express four MARK family members (MARK1–4), and they share the functional domains, such as the kinase domain, UBA domain, spacer domain, and KA1 domain (Fig. 2A) [42, 43, 44]. We asked whether another member of this family localizes to stress granules. In contrast to MARK4, MARK2 expressed in HeLa cells did not colocalize with G3BP1 (Fig. 2A), suggesting that sequences that differ between MARK2 and MARK4 may contribute to the stress granule localization of MARK4. Since many proteins localized to stress granules have higher LLPS propensity [45], we compared the LLPS propensity score of MARK2 and MARK4 along their sequence by using catGRANULE [46]. We found that MARK4, but not MARK2, has a high LLPS propensity in part of its spacer region (Fig. 2B). The spacer domain in MARK4 is required for its localization to SGs. (A) MARK2 does not localize to stress granules. Top: Right: HeLa cells transfected with myc‐MARK4 or myc‐MARK2 were immunostained with anti‐Myc and anti‐G3BP1 antibodies. Representative images are shown. Scale bars: 10 μm. (B) Schematic diagrams of functional domains and liquid–liquid phase separation (LLPS) propensity of MARK2 and MARK4. MARK4 sequence shows higher scores in the spacer domain, especially around 360–400 and 580–586, indicated by the red lines. These sequences are not well conserved between MARK4 and MARK2. (C) Distribution of a series of MARK4 deletion mutants. HeLa cells were transfected with myc‐tagged MARK41–709, MARK41–368, MARK4Δ386–709, MARK4386–709, MARK4Δ360–400,Δ580–586, MARK4Δ360–400, and MARK4Δ580–586. Cells were immunostained with anti‐Myc (MARK4, magenta) and anti‐G3BP1 (green). Scale bars: 20 μm. (D) MARK4 kinase activity is not required for localization to stress granules. HeLa cells transfected with MARK4 mutants without kinase activity (MARK4D199A and MARK4T214AS218A) were immunostained with G3BP1. Representative images and line plots are shown. Scale bars: 20 μm. To determine the region that mediates the localization of MARK4 to stress granules, we generated a series of deletion mutants of MARK4 and analyzed their colocalization with an SG marker. MARK41–709 colocalized with G3BP1; however, MARK41–368 did not colocalize with G3BP1. Deletion of 386–709 (MARK4Δ386–709) abolished its colocalization with G3BP1, while a fragment of MARK4 containing 386‐709 (MARK4386–709) colocalized with G3BP1 (Fig. 2C). These results suggest that MARK4 localizes to stress granules via the spacer domain. In the spacer domain, MARK4 sequences around 360‐400 and 580‐586 show peaks in LLPS scores, which are not observed in MARK2 (Fig. 2B). We found that the mutant forms of MARK4 with these regions deleted (MARK4Δ360–400,Δ580–586) formed foci that did not colocalize with G3BP1 (Fig. 2C). We also analyzed the effects of deletion of each region: MARK4Δ580–586 colocalized with G3BP1, whereas MARK4Δ360–400 formed foci that did not colocalize with G3BP1 (Fig. 2C). These results indicate that MARK4360–400 is critical for its localization to G3BP1‐positive stress granules. We also examined whether MARK4 kinase activity is required for localization to stress granules. The mutant forms of MARK4 without kinase activity, MARK4D199A [47] and MARK4T214AS218A [48, 49], were generated and examined for colocalization with G3BP1. Both mutants colocalized with G3BP1 (Fig. 2D), indicating that MARK4 kinase activity is not required for its localization to stress granules. MARK4 promotes stress granule formation We noticed that MARK4‐transfected HeLa cells form SGs more frequently than cells transfected with vectors carrying EGFP, and MARK4‐expressing cells contained more SGs compared to cells transfected with EGFP (Fig. 3A). We were also interested in whether MARK4 affects the physical properties of stress granules, such as fluidity. The mobility of TIA1 within stress granules has been widely used as a measure of stress granule dynamics and fluidity [50, 51]. Therefore, we performed FRAP analyses in HeLa cells expressing EGFP‐TIA1 with or without mCherry‐MARK4. MARK4 expression significantly reduced the fluorescence recovery of EGFP‐TIA1 within stress granules (Fig. 3B), indicating decreased stress granule fluidity. MARK4 promotes stress granule formation. (A) HeLa cells were transfected with EGFP (GFP) or myc‐tagged MARK4 (MARK4) and immunostained with anti‐G3BP1 to detect SGs. The percentage of stress granule (SG)‐containing cells (Mean ± SD, n = 4; **P < 0.01, Student's t‐test) and density of SGs (the number of stress granules/cytosol area (μm2)) (Mean ± SD, n = 12; ***P < 0.005, Student's t‐test) were higher in MARK4‐transfected cells than those in GFP‐transfected cells. (B) MARK4 reduces TIA1 diffusion in SG. Fluorescence Recovery After Photobleaching (FRAP) was performed in cells transfected with EGFP‐TIA1 or EGFP‐TIA1 and mCherry‐MARK4. Left: Representative images of a SG before the breach, 0 s after a breach, and 145 s after a breach. GFP intensity in SGs was normalized to the intensity before the breach and expressed as mean ± standard deviation. ****P < 0.0001 (comparison of fits with extra sum‐of‐squares F test), best fit values of half‐recovery time: 12.07 for TIA1, 13.68 for TIA1 + mCherry‐MARK4. Scale bars: 5 μm (left panels) and 1 μm (right panels). (C) MARK4 without SG localization does not promote SG formation. HeLa cells transfected with MARK41–368 were immunostained with anti‐G3BP1 antibodies. The percentage of cells with stress granules (Mean ± SD, n = 3; **P < 0.01, Student's t‐test) and the number of SGs per area (Mean ± SD, n = 11–12; **P < 0. 01, Student's t‐test). (D) MARK4 without kinase activity does not promote SG formation. HeLa cells transfected with MARK4T214AS218A were immunostained with anti‐G3BP1. The percentage of cells with SGs (Mean ± SD, n = 3; **P < 0.01, Student's t‐test) and the number of SGs per area (Mean ± SD, n = 12; **P < 0.01, Student's t‐test). We asked whether MARK4 localization to SGs is required for the enhancement of SG formation. Cells expressing MARK4Δ1–368, which does not localize to SGs (Fig. 2), had fewer SGs compared to those expressing full‐length MARK4 (Fig. 3C). Also, cells expressing a kinase‐dead form of MARK4 (MARK4T214AS218A) showed fewer SGs than those expressing wild‐type MARK4 (Fig. 3D). These results suggest that MARK4 localized to SGs enhances SG formation via kinase activity. MARK4 reduces TIA1 oxidation Oxidative stress triggers cellular stress responses to induce SG formation; however, H2O2 treatment induces oxidation of TIA1 and promotes its dimerization/oligomerization, which reduces its ability to nucleate functional SGs [3]. We asked whether MARK4 affects the dimerization of TIA1 induced by H2O2 treatment. Western blotting revealed that H2O2 treatment increased the levels of dimerized/oligomerized TIA1 (ox‐TIA1) in HeLa cells, as reported in U2OS cells [3]. However, cells with MARK4 co‐expression did not show an increase in ox‐TIA1 (Fig. 4). This result suggests that MARK4 suppresses TIA1 oxidation and maintains its function to promote SG formation under oxidative stress. MARK4 reduces TIA1 oxidation in HeLa cells transfected with EGFP‐TIA1 or Myc‐MARK4 and EGFP‐TIA1 were treated with 1 mm H2O2 for 1 h. Cell lysates were subjected to Western blotting with anti‐Myc (MARK4) and anti‐TIA1. GAPDH was used as a loading control. Representative blots and quantitation (Mean ± SD, n = 3; *, P < 0.05, Ordinary one‐way ANOVA tests followed by Tukey's multiple comparisons test). MARK4 and TIA1 synergistically increase tau protein levels Next, we investigated the effects of MARK4 and SGs on tau proteins. Tau expressed in HeLa cells showed a diffused pattern in the cytoplasm and partially colocalized with MARK4 in TIA1‐positive granules (Fig. 5A). Western blotting revealed that co‐expression of MARK4 and TIA1 increased total tau levels (Fig. 5B). Expression of either MARK4 or TIA1 alone did not increase tau levels in HeLa cells (Fig. 5B), indicating that the synergistic effects of MARK4 and TIA1 are required for tau accumulation. The levels of Ser262‐phosphorylated tau were increased by co‐expression of MARK4, and it was further increased by co‐expression of MARK4 in TIA1‐overexpressing cells (Fig. 5B). TIA1 overexpression increased MARK4 levels (Fig. 5B), which is likely to cause the increase in pSer262‐tau. TIA1 and MARK4 synergistically increase tau protein levels. (A) Tau is partially localized to stress granules (SGs). HeLa cells were transfected with tau and Myc‐tagged MARK4 and immunostained with anti‐tau (tau, red), anti‐Myc (MARK4, magenta), and anti‐TIA1 (green). Arrows indicate stress granules containing tau, MARK4, and TIA1. Scale bars: 10 μm. (B) MARK4 and TIA1 overexpression increased levels of total tau and Ser262‐phosphorylated tau. Western blotting of HeLa cells transfected with tau, Myc‐tagged MARK4, and EGFP‐tagged TIA1. Actin was used as a loading control. Representative blot and quantification results are shown (Mean ± SD, n = 3; *P < 0.05, **P < 0.01, ***P < 0.005, ****P < 0.0001, Ordinary one‐way ANOVA tests followed by Tukey's multiple comparisons test). (C) MARK4 and TIA1 co‐expression does not increase TDP‐43. HeLa cells cotransfected with HA‐TDP‐43 with Myc‐MARK4, EGFP‐TIA1, or Myc‐MARK4 and EGFP‐TIA1 were subjected to western blotting. Representative blots and quantitation (Mean ± SD, n = 3; *P < 0.05, **P < 0.01, Ordinary one‐way ANOVA tests followed by Tukey's multiple comparisons test). (D) MARK4 without kinase activity does not increase tau levels. HeLa cells were transfected with tau, TIA1, and MARK4T214AS218A and subjected to western blotting. Anti‐GAPDH was used as a loading control. Representative blots and quantitative analysis (Mean ± SD, n = 6; ns, P > 0.05, Ordinary one‐way ANOVA tests followed by Tukey's multiple comparisons test). (E) HeLa cells were transfected with tau with alanine substitutions at Ser262 and Ser356 (S2A) and MARK4, TIA1 or MARK4 and TIA1. Cells were subjected to western blotting with anti‐tau, anti‐Myc (MARK4), and anti‐GFP (TIA1). Representative blots and quantification (Mean ± SD, n = 3; **P < 0.01, Ordinary one‐way ANOVA tests followed by Tukey's multiple comparisons test). (F) HeLa cells were transfected with tau with glutamate substitutions at Ser262 and Ser356 (S2E) and MARK4, TIA1 or MARK4 and TIA1. Cells were subjected to western blotting with anti‐tau, anti‐Myc (MARK4), and anti‐GFP (TIA1). GAPDH was used as a loading control. Representative blots and quantification (Mean ± SD, n = 3; **P < 0.01, Ordinary one‐way ANOVA tests followed by Tukey's multiple comparisons test). (G) MARK41–368 and TIA1 co‐expression increases tau levels. HeLa cells transfected with tau and Myc‐tagged MARK41–368, EGFP‐tagged TIA1, or Myc‐MARK41–368 and EGFP‐TIA1, were subjected to Western blotting. using, anti‐tau (tau), anti‐pS262 (pSer262), anti‐Myc (MARK4), anti‐GFP (TIA1). Anti‐GAPDH was used as a loading control. Representative blots and quantification results (*P < 0.05, **P < 0.01, Ordinary one‐way ANOVA tests followed by Tukey's multiple comparisons test). (F, G) MARK4 and TIA1 co‐expression increased tau with Ser262 substitutions. We also tested whether the simultaneous expression of MARK4 and TIA1 increases other neurodegenerative disease‐associated proteins. TDP‐43, which aggregates and accumulates in ALS and FTLD, is known to localize to stress granules [52]. We found that MARK4 and TIA1 did not increase TDP‐43 levels but rather decreased them (Fig. 5C). These results suggest that MARK4 and TIA1 may affect tau via a mechanism that is not shared with TDP‐43. Roles of tau phosphorylation in tau accumulation caused by TIA1 and MARK4 Tau phosphorylation at Ser262 and Ser356, the direct targets of MARK4, stabilizes tau proteins and increases their levels in cultured cells, Drosophila, and mice [27, 28, 29, 53]. We examined whether MARK4 kinase activity is required for the accumulation of tau protein by MARK4 + TIA1. Co‐expression of a kinase‐dead form of MARK4T214AS218A and TIA1 did not increase tau protein (Fig. 5D), indicating that MARK4 activity is required for tau accumulation in TIA1‐overexpressing cells. Unexpectedly, we found that tau phosphorylation at MARK4‐target sites was dispensable for its accumulation caused by co‐expression of MARK4 and TIA1: co‐expression of MARK4 and TIA1 increased tau proteins with alanine substitutions at Ser262 and Ser356 (S2A) (Fig. 5E, compare S2Atau and S2Atau + MARK4 + TIA1) and tau with glutamate substitutions at these sites (S2E) (Fig. 5F, compare S2Etau and S2Etau + MARK4 + TIA1). However, there was a difference in the effects of TIA1 on S2Atau and S2Etau: For the S2Atau, co‐expression of MARK4 and TIA1 does not significantly increase tau accumulation compared with MARK4 alone (Fig. 5E, compare S2Atau + MARK4 and S2Atau + MARK4 + TIA1), whereas a significant increase is observed for the S2E mutant (Fig. 5F, compare S2Etau + MARK4 and S2Etau + MARK4 + TIA1). These results suggest that TIA1 may promote the accumulation of phosphorylated tau in MARK4‐expressing cells. We also found that MARK41–368, which does not colocalize to SGs, also increased tau levels when co‐expressed with TIA1 (Fig. 5G). These results suggest that the accumulation of tau proteins caused by the co‐expression of MARK4 and TIA1 is not mediated by direct interaction with tau in SG and that MARK4 and TIA1 synergistically create a cellular condition that favors tau accumulation. Knockdown of TIA1 suppresses tau‐induced neurodegeneration in Drosophila Finally, we asked how TIA1 is involved in tau‐induced neurodegeneration by using a Drosophila model. Expression of human tau in Drosophila retina induces photoreceptor degeneration. We found that knockdown of Rox8, a functional ortholog of TIA1 in Drosophila (Fig. 6A), suppressed tau‐induced retinal degeneration (Fig. 6B). Tau protein levels and tau phosphorylation at Ser262 were not significantly affected by the knockdown of Rox8 (Fig. 6C). Rox8 knockdown also suppressed tau‐induced neurodegeneration in the background of MARK4 overexpression (Fig. 6D). Knockdown of the Drosophila homolog of TIA1 suppresses tau‐induced neurodegeneration in Drosophila. (A) Rox8 RNAi expression reduced Rox8 mRNA. qRT‐PCR of head lysate expressing Rox8 RNAi driven by a pan‐retinal driver GMR‐GAL4. (B) Rox8 expression suppressed tau‐induced neurodegeneration. Representative images of paraffin sections of fly retina stained with hematoxylin and eosin. Human wild‐type tau was co‐expressed with control RNAi (tau+control RNAi) or Rox8 RNAi (tau+Rox8 RNAi) by GMR‐GAL4. Without human tau expression, Rox8 RNAi does not cause any morphological change in the retina (compare control RNAi and Rox8 RNAi). Neurodegeneration was identified by the presence of vacuoles (arrows). Scale bars, 50 μm. Vacuole areas in the lamina were quantified and expressed as the ratio (****P < 0.0001; Student's t‐test, n = 5). (C) Rox8 knockdown did not affect tau protein expression. Western blotting of head homogenate with anti‐tau (top) and anti‐pSer262 (bottom). Actin was used as a loading control. (Mean ± SE, n = 3, P > 0.05, Student's t‐test). (D) Representative images of fly retina co‐expressing tau, MARK4, and control RNAi (tau + MARK4 + control RNAi) or tau, MARK4, and Rox8 RNAi (tau + MARK4 + Rox8 RNAi). Vacuole areas in the lamina were quantified and expressed as the ratio (*P < 0.05, ns, P > 0.05, Student's t‐test, n = 9). Scale bars, 50 μm. Flies were at ten‐day posteclosion. Discussion MARK4 has been reported to play critical roles in the pathogenesis of cancer and neurodegenerative diseases [17, 18, 19, 20, 21, 22]. MARK4 upregulation in cancer cells enhances metastasis and migration [17], and in the pathogenesis of neurodegenerative diseases, MARK4 phosphorylates microtubule‐binding protein tau at the KXGS motif to increase its toxicity [27, 28, 29]. While MARK family members are highly conserved and all phosphorylate tau at the same sites, MARK4 enhances tau accumulation and toxicity more than other MARKs, suggesting an additional mechanism by which MARK4 enhances tau toxicity [32]. In this study, we report that MARK4 is a component of the SG in cultured cells and primary neurons, and a regulator of SG formation in cultured cells under oxidative stress. Since another MARK family member, MARK2, does not distribute to SGs, MARK4 interaction with SGs may explain its specific effects on tau toxicity. Aberrant SG formation has been suggested in the pathogenesis of both cancer and neurodegenerative diseases, and our findings suggest that MARK4 plays roles in disease pathogenesis beyond its role as a tau kinase. Stress granules are dynamic organelles that exhibit liquid‐like properties, and their fluidity reflects the mobility of molecules within the condensate and their exchange with the surrounding cytoplasm. Reduced fluidity is associated with a more stable or less dynamic state and may promote the persistence of protein assemblies, including oligomeric species [2]. TIA1 has been reported to promote tau oligomer formation and to exacerbate tau‐induced neurodegeneration in mouse models of tauopathy [14]. We found that MARK4 increases stress granule formation and decreases TIA1 mobility within stress granules (Fig. 3). These findings suggest that MARK4 may alter stress granule dynamics and stability, potentially creating a cellular environment that favors the formation or persistence of toxic tau species. Oxidative stress is known as an SG inducer, while its effects are complex [54]. Treatment with a high concentration of H2O2 (1 mm) induces SG formation, while a lower concentration of H2O2 attenuates arsenite/ER stress‐induced SG formation. H2O2 oxidizes TIA1 and reduces its SG assembly function, thus suppressing SG assembly by impeding the interaction between TIA‐1 and its target mRNAs [3]. Our results indicate that MARK4 protects TIA1 against H2O2‐induced dimerization, which may contribute to enhanced SG formation under oxidative stress conditions. Supporting this model, MARK4 enhanced SG formation in H2O2 ‐treated cells but not in arsenite‐treated cells (Fig. 1). This result aligns with the involvement of MARK4 in oxidative stress, such as its upregulation in response to oxidative stress and increasing reactive oxygen species in adipocytes [55], cultured cells [56], and a diabetic cardiomyopathy model [57]. We found that knockdown of the Drosophila homolog of TIA1, Rox8, suppresses tau toxicity (Fig. 6). Tau is an intrinsically disordered protein and is found in multiple assemblies, such as monomers, dimers, oligomers, protofibrils, and fibrils, in diseased brains [58]. Once formed, fibrils spread and propagate in a prion‐like manner [59]. Thus, elucidating the early stage of pathological changes of tau proteins is critical for understanding disease pathogenesis and the development of an effective therapy [60]. In Drosophila, tau proteins are phosphorylated at disease‐related sites and tau neurodegeneration in a phosphorylation‐dependent manner without mature fibrils, indicating that this model recapitulates an early stage of tau abnormality [26, 35, 61]. Suppression of tau toxicity in this model by Rox8 knockdown is in line with the reports that SGs affect oligomer formation, the early stage of tau aggregation [62]. Deletion of Mark4 and reduction of TIA1 have both been reported to reduce tau aggregates and protect against neurodegeneration in the PS19 tau mice [13, 14, 22]. Together with our findings in cultured cells and Drosophila, these observations support a functional link between MARK4 and TIA1 in tau pathogenesis. Nevertheless, additional studies will be required to determine the extent to which the MARK4–TIA1 interaction directly contributes to tau toxicity in mammalian disease models. MARK4 is a druggable target [63], raising the possibility that modulation of MARK4 activity may have therapeutic potential for cancer and tauopathy. Conflict of interest The authors declare no competing interests. Author contributions SN: Investigation, formal analysis, writing—original draft, KW: Investigation and validation, GS: Investigation and formal analysis, AF: Investigation, KI: Investigation, SS: Investigation, TS: Investigation, resources, and supervision, AA: Investigation, resources, supervision, writing—review and editing, KA: Conceptualization, funding acquisition, investigation, project administration, supervision, writing—original draft, writing—review and editing. Acknowledgement The authors thank Dr. Mel Feany (Harvard University) for fly stocks, Dr. Mutsuhiro Takekawa (The University of Tokyo) for pEGFP‐TIA1 cDNA, and Dr. Takashi Nonaka (Tokyo Metropolitan Institute of Medical Science) for TDP‐43 cDNA. This work was supported by research awards from the Takeda Science Foundation (to K.A.), TMU strategic research fund (to K.A.), and a research grant from the Japan Society for the Promotion of Science, grant number 24 K02860 (to K.A.), AMED under Grant Number JP24wm0625509 and JP25wm0625509 (to K.A.). Data accessibility The datasets used and analyzed in this study are available from the corresponding author upon request. References Associated Data Data Availability Statement The datasets used and analyzed in this study are available from the corresponding author upon request.