PI3Kγ stimulates a high molecular weight form of myosin light chain kinase to promote myeloid cell a

Myeloid cells play key roles in cancer immune suppression and tumor progression. In response to tumor derived factors, circulating monocytes and granulocytes extravasate into the tumor parenchyma where they stimulate angiogenesis, immune suppression and tumor progression. Chemokines, cytokines and interleukins stimulate PI3Kγ-mediated Rap1 activation, leading to conformational changes in integrin α4β1 that promote myeloid cell extravasation and tumor inflammation Here we show that PI3Kγ activates a high molecular weight form of myosin light chain kinase, MLCK210, that promotes myosin-dependent Rap1 GTP loading, leading to integrin α4β1 activation. Genetic or pharmacological inhibition of MLCK210 suppresses integrin α4β1 activation, as well as tumor inflammation and progression. These results demonstrate a critical role for myeloid cell MLCK210 in tumor inflammation and serve as basis for the development of alternative approaches to develop immune oncology therapeutics.

Inflammation contributes to infectious, cardiovascular, autoimmune, and neurodegenerative diseases and cancer¹. Vascular endothelial cells and myeloid cells (monocytes, macrophages, and neutrophils) play critical roles in inflammatory responses. Myeloid cells invade inflamed tissues, where they can contribute to pathology by expressing inflammatory mediators, reactive oxygen species, angiogenic factors, and immunosuppressive factors². Endothelial cells respond to inflammatory stimuli by promoting myeloid cell adhesion and extravasation, by becoming more permeable and by undergoing angiogenesis³. Sustained recruitment of inflammatory cells to tissues can lead to progressive damage induced by release of oxygen radicals, which can lead to the development of cancer⁴. Targeting the mechanisms controlling myeloid cell responses to inflammatory stimuli could lead to additional approaches to suppress chronic inflammation and cancer.

A variety of inflammatory factors released by damaged tissues activate G protein coupled receptors (GPCRs), receptor tyrosine kinases (RTKs) or Toll-like receptor/interleukin1 receptor family members (TLR/IL1Rs) to initiate myeloid cell recruitment during inflammation⁵. In previous studies, we demonstrated that myeloid cell recruitment during tumor inflammation depends on the VCAM-1 receptor integrin α4β1 but not on other integrins^6,7,8. Inactivation of or deletion of integrin α4β1 in myeloid cells prevented monocyte and granulocyte extravasation and suppressed tumor inflammation and growth, as did selective antibody and small molecule antagonists of this integrin^6,7. We found that inflammatory stimuli promote integrin α4β1 activation, cell adhesion, and myeloid cell trafficking to tumors by activating a single PI3K isoform, PI3Kγ⁷. PI3Kγ mediated integrin activation requires PLCγ, which cleaves phospholipids to produce Ca2+ and diacylglycerol (DAG), the DAG-activated RapGEFs, CalDAG-GEFI and II, Rap1a and RIAM, thereby activating integrin α4β1 in a RIAM-dependent manner^7,9. Pharmacological or genetic blockade of PI3Kγ or integrin α4β1 suppressed adhesion and recruitment of both monocytes and granulocytes into tumors and inhibited disease progression^6,7,9. Our studies and those of others have shown that integrin α4β1 conformational changes (activation) depend on Rap1 (Ras-proximate-1), a Ras-like small GTP-binding protein and its effector protein RIAM, which localizes talin as well as paxillin to the integrin cytoplasmic tails^{9,10,11,12,13}. While talin binds integrin β chain cytoplasmic tails, inducing a shift in the conformation of the extracellular domain of the integrin and resulting in an allosteric increase in ligand-binding affinity¹³, paxillin binds to the integrin α4 cytoplasmic tail, promoting α4 integrin conformation changes that promote adhesion and trafficking of lymphocytes and myeloid cells^6,9,10,11,12. These findings indicate that targeting the signaling events that lead to inflammation might provide significant benefit in the treatment in cancer. Deciphering the molecular mechanisms by which myeloid cell integrins are activated could lead to the development of approaches to treat acute and chronic inflammatory and vascular diseases.

Here we show that PI3Kγ activates a high molecular weight form of myosin light chain kinase, MLCK210, leading to myosin-dependent activation of Rap1 GTP loading and integrin α4β1 conformational changes that promote tumor inflammation and progression. These studies also identify MLCK210 inhibitors as potential immune oncology therapeutics.

MLCK210 colocalizes with integrin α4β1

We previously determined that tumor derived growth factors and chemokines, such as SDF1α (CXCL12), TGFβ, and IL1β, promote PI3Kγ-mediated integrin α4β1 conformational changes and adhesion to endothelium in circulating myeloid cells during tumor inflammation^7,9. To identify proteins that closely associate with myeloid cell integrin α4β1 that might regulate its activation, we stimulated primary bone-marrow-derived CD11b+ myeloid cells with Lewis Lung carcinoma tumor cell condition medium (TCM) or basal media, then solubilized cells and immunoprecipitated integrin α4β1. We then performed silver staining and immunoblotting to detect integrin α4β1 interacting proteins (Fig. 1a). Immunoblotting revealed that paxillin and talin co-immunoprecipitated with integrin α4β1 upon stimulation with TCM (Fig. 1a), as predicted based upon our prior published studies^9,10,11,12. Silver staining identified two major co-precipitating proteins with molecular weights of 80 kDa and 210 kDa upon stimulation with TCM but not under basal conditions (Fig. 1a; Supplementary Fig. 1a). These proteins were excised from silver-stained gels and subjected to protein sequencing by tandem mass spectrometry (Fig. 1b). Tandem mass spectrometry analysis identified the 80 kDa co-precipitating protein as Gelsolin and the 210 kDa co-precipitating protein as a high molecular weight form of myosin light chain kinase (MLCK) (Fig. 1b; Supplementary Fig. 9). A 210 kDa isoform of MLCK was previously shown to play a key role in regulating vascular permeability^{14,15,16,17,18,19}. Gelsolin is an actin filament severing and capping protein in podosomes that is inhibited by PIP3, the product of PI3Kinases²⁰. MLCK210 is an alternatively spliced MLCK isoform that contains the entire 108 kDa non-muscle MLCK protein as well as six N-terminal IgG-C2-like domains and two tandem DXR actin-binding domains (Supplemental Fig. 1b)^{14,15,16,17,18}. MLCKs contain a core catalytic domain and a calmodulin-sensitive auto-inhibitory domain that folds back on the catalytic domain to inhibit kinase activity. Upon activation by calmodulin, MLCK phosphorylates Myosin Light Chain (MLC), thereby stimulating myosin ATPase²¹. As activated myosin undergoes a conformational change that allows it to cyclically bind and release actin filaments, it promotes contractility that leads to cell shape change and migration²².

To determine whether MLCK210 and integrin α4β1 colocalize in intact plasma membranes, CD11b+ myeloid cells from wildtype (WT) and mutant mice in which the alternatively transcribed isoform of Mylk encoding MLCK210 was selectively deleted without impacting MLCK108¹⁵ (here designated Mlck210^−/− mice), were incubated with basal culture medium or culture medium containing the chemokine SDF1α. Plasma membranes were purified, solubilized, and immunoblotted to detect integrin α4 and MLCK210. MLCK210 was detected in plasma membranes only from SDF-1α-stimulated wildtype (WT) cells but not in from Mlck210^−/− cells or basal WT cells (Fig. 1c). In contrast, integrin α4 was constitutively associated with plasma membranes (Fig. 1c). These results indicate that MLCK210 associates with the membrane only upon chemokine receptor signaling while integrin α4β1 is constitutively present in plasma membranes.

To examine the role of MLCK210 in the distribution of integrin α4 in myeloid cells, myeloid cells were incubated with SDF-1α or BSA coated 0.9 µm microbeads, fixed, permeabilized, and analyzed for clustering or aggregation of integrin α4 (green) by immunofluorescence microscopy. We found that SDF-1α but not BSA induced clustering or aggregation of integrin α4 in WT but not Mlck210^−/−myeloid cells (Fig. 1d–e). Additionally, myeloid cells were transfected with control All Stars non-silencing or Mylk siRNA, then incubated with SDF-1α-coated beads. Western blotting verified MLCK210 gene knockdown (Fig. 1f). Confocal microscopy assessment of the effect of gene knockdown on integrin localization showed that SDF- 1α beads induced integrin clustering in All Stars non-silencing siRNA-transfected cells but not in Mylk siRNA-transfected cells (Fig. 1g).

To determine whether MLCK210 and integrin α4β1 colocalize in the same membrane domains upon chemokine stimulation, we transiently transfected Mlck210^−/− cells with Flag-tagged wildtype and kinase-dead MLCK210. Because antibodies directed against MLCK react with multiple MLCK isoforms, selective detection of MLCK210 by immunofluorescence in intact cells could only be accomplished by transfecting Mlck210^−/− cells with these Flag-tagged constructs. Flag-tagged wildtype and kinase-dead Mlck210 transfected myeloid cells were incubated with SDF-1α or BSA coated 0.9 µm microbeads, fixed, permeabilized, and analyzed for colocalization of MLCK210 (red) and integrin α4 (green) by immunofluorescence microscopy. We found that SDF-1α but not BSA induced colocalization of MLCK210 (red) and integrin α4 (green) in Mlck210 WT but not KD transfected cells (Fig. 1h; Supplementary Fig. 2a–c). MLCK210 and integrin α4 overlap in the merged images can be observed as regions of yellow color indicated by arrows. Additionally, we found that SDF-1α but not BSA induced colocalization of paxillin (red) and integrin α4 (green) in Mlck210 WT but not KD transfected cells (Supplementary Fig. 2d, e). Taken together, these results demonstrate that MLCK210 closely associates with integrin α4 in the plasma membrane and is required to promote integrin α4β1 clustering in response to chemokine stimulation.

MLCK210 is required for integrin α4β1-mediated cell adhesion

Our prior studies showed that integrin α4β1 mediates myeloid cell adhesion to vascular cell adhesion molecule 1 (VCAM-1) on endothelium, leading to tumor inflammation and growth^6,7,8. We found that a wide array of chemokines and cytokines stimulated adhesion to endothelial monolayers (HUVEC) or VCAM-1-coated substrates of CFDA-labeled peripheral blood CD11b+ myeloid cells from WT mice but not from mice with an integrin α4Y991 inactivating mutation (Fig. 2a). To determine whether MLCK210 is required for integrin α4-mediated adhesion, Mlck210^−/− and WT murine peripheral blood CD11b+ myeloid cells were incubated with basal medium or medium containing IL-1β, IL-6, SDF-1α, TNFα, VEGF-A, or Lewis lung carcinoma cell conditioned medium (TCM). Cells were then allowed to adhere to endothelial cell monolayers (HUVEC) or to recombinant soluble VCAM-1 coated on plastic plates. While cytokine/chemokine-stimulated WT myeloid cells adhered to HUVEC and VCAM-1, Mlck210^−/− myeloid cells did not adhere (Fig. 2b). Previously, we showed that integrin α4-mediated adhesion of Gr1^lo monocytes and Gr1^hi granulocytes to endothelium required the activity of both PI3Kγ and Rap1^7,9. As both CD11b+ Gr1^lo monocytes and CD11b+ Gr1^hi granulocytes extravasate into tumors from peripheral blood, we asked whether the adhesion of each of these two cell types was similarly regulated by PI3Kγ, Rap1, and MLCK210. We found that the adhesion (MFI) of both CD11bGr1^lo monocytes and CD11bGr1^hi granulocytes from Pik3cg^−/−(p110γ^−/−), Mlck210^−/−, or Rap1a^−/− mice to VCAM-1-coated plates was strongly inhibited compared to the adhesion of cytokine-stimulated WT cells (Fig. 2c). In support of these findings, MlcksiRNA-treated myeloid cells also exhibited poor adhesion to HUVEC and VCAM, in contrast to control All Stars siRNA-transfected and untransfected cells (Fig. 2d). These results indicate that MLCK210 plays a key role in the PI3Kγ-Rap1 pathway that promotes integrin α4-mediated adhesion.

a Adhesion to HUVEC or VCAM-1, expressed as mean fluorescence intensity (MFI), of cytokine- and tumor conditioned medium (TCM)-stimulated murine WT or integrin α4Y991A myeloid cells (n = 3). *p < 0.0001 for cytokine-stimulated WT vs. α4Y991A cells. No significant difference (ns) for adhesion to HUVEC (p > 0.9999) or VCAM-1 (p = 0.9465) of basal WT vs. α4Y991Acells. b Adhesion to HUVEC or VCAM of chemokine, cytokine- and tumor conditioned medium (TCM)-stimulated murine WT or Mlck210^−/− myeloid cells (n = 3). *p < 0.0001 for cytokine-stimulated WT vs Mlck210^−/− myeloid cells. No significant difference (ns, p > 0.9999) for basal WT vs. Mlck210^−/− cells. c Adhesion to purified VCAM-1 coated plates of murine CD11bGr1^lomonocytes and CD11bGr1^hi granulocytes from WT, p110γ^−/−, Mlck210^−/−, or Rap1a^−/− mice (n = 3). *p < 0.0001 for cytokine-stimulated WT vs p110γ^−/−, Rap1a^−/−, and Mlck210^−/− cells. No significant difference (ns, p = 0.9999) for basal WT vs p110γ^−/−, Rap1a^−/−, and Mlck210^−/−. d Adhesion to HUVEC or VCAM of basal or cytokine-stimulated All Stars or Mlck (Mylk)siRNA-transfected murine myeloid cells (n = 3). *p < 0.0001 for cytokine induced adhesion of Mlck vs All Stars siRNA-transfected cells to VCAM. *p < 0.0001 for adhesion of IL1β and IL-6 stimulated, *p = 0.0234 for SDF-1α, *p = 0.0203 for TCM, *p = 0.0133 for TNFα and *p = 0.0091 for VEGF-A stimulated Mlck vs All Stars siRNA-transfected cell adhesion to HUVEC. No significant difference (ns, p > 0.9999) for basal Mlck vs All Stars siRNA-transfected cells. e Dose response analysis of the effect of MW01-022AZ on murine myeloid cell adhesion to VCAM-1 stimulated by SDF-1α or IL1β (n = 3). f Effect of MW01-022AZ on cytokine-stimulated human myeloid cell adhesion to HUVEC or VCAM-1 (n = 3). *p < 0.0001 for vehicle vs MW01-022AZtreated, SDF-1α IL1β and VEGF-A stimulated cells. No significant difference (ns) for basal vehicle vs MW01-022AZ on HUVEC (p = 0.8530) or VCAM-1 (p = 0.9666). g Chemotaxis of Mlck or All Stars siRNA-transfected murine myeloid cells toward basal, SDF-1α or tumor cell conditioned medium (TCM) stimulated cells (n = 3). *p = 0.0212 for TCM and *p = 0.0189 for SDF-1α-stimulated All Stars vs Mlck siRNA. No significant difference (ns, p = 0.9999) for basal cells. a, d, f, g Significance determined by one-way Anova with Tukey’s multiple comparisons. All replicates are biologically independent samples. Data are presented as mean values ± SEM. Experiments were performed twice. Source data are provided with this article as a Source Data file.

Full size image

To determine whether pharmacological inhibitors of MLCK could suppress cytokine/chemokine-mediated integrin activation and adhesion, WT murine myeloid cells were incubated with medium, SDF-1α, or IL-1β in the presence or absence of either MW01-022AZ²³ or ML-7, two MLCK inhibitors. We found that MLCK inhibitors suppressed SDF-1α and IL-1β induced adhesion of murine myeloid cells to VCAM-1 in a dose-dependent manner (Fig. 2e, Supplementary Fig. 3a). In contrast, inhibition of MLCK210 did not affect adhesion of myeloid cells to other substrates (Supplementary Fig. 3b). MW01-022AZ also inhibited cytokine/chemokine-stimulated human myeloid cell adhesion to HUVEC or VCAM (Fig. 2f). As cell adhesion is a pre-requisite for cell migration and invasion, we evaluated the effect of MLCK inhibition on myeloid cell migration. MLCK knockdown by siRNA transfection as well as the MLCK inhibitor ML-7 suppressed myeloid cell migration in transwell assays, while control transfection or vehicle had no effect on cell migration on VCAM-1 coated surfaces (Fig. 2g; Supplementary Fig. 3c, d).

In contrast to integrins on cells within tissues, integrins on the surface of circulating immune cells reside in stable, inactive conformations that are unable to bind ligand or facilitate adhesion^5,12. Upon exposure to cytokines or chemokines, intracellular signal transduction events promote integrin unfolding and acquisition of ligand-binding and adhesive capacity within 0.1 s, an event that is termed integrin activation^12,24. We previously showed that recombinant VCAM-1 conjugated to the Fc portion of the immunoglobulin heavy chain could bind to integrin α4β1 on cytokine-stimulated myeloid cells but not on unstimulated myeloid cells or on myeloid cells with inactivating mutations in integrin α4, PI3Kγ, Ras, or Rap1a^7,9. To determine if MLCK210 also regulates integrin α4β1 ligand binding, we incubated human peripheral blood myeloid cells with the MLCK inhibitor MW01-022AZ or vehicle and then with recombinant VCAM-1-Fc and anti-Fc fluorochrome-conjugated antibody and quantified the mean fluorescence intensity of bound VCAM. We found that cytokine/chemokine stimulation induced VCAM-1 binding, while the MLCK inhibitor MW01-022AZ, as well as the inhibitor ML-7, suppressed VCAM-binding (Fig. 3a, b; Supplementary Fig. 3e). Mn2+, which activates integrins independently of cytokine/chemokine receptor signaling by establishing salt bridges between integrins and their ligands^25,26, stimulated VCAM-Fc binding independently of cytokine stimulation (Fig. 3a, b; Supplementary Fig. 3e). MW01-022AZ also inhibited binding of HUTS21, a monoclonal antibody that selectively binds to an epitope that is only exposed on the β1 subunit of heterodimeric integrin complexes of cytokine/chemokine-stimulated human immune cells (Fig. 3c, d). In contrast, MW01-022AZ had no effect on the binding of P4C10, an anti-β1 integrin antibody that binds independently of activation (Fig. 3c, d). MLCK inhibition also suppressed VCAM-Fc binding to murine myeloid cells, as Mlck210^−/−, Mylk siRNA-transfected and MW01-022AZ-treated myeloid cells failed to bind VCAM-Fc upon cytokine/chemokine stimulation (Fig. 3e–h). Importantly, MLCK210 depletion had no impact on integrin α4 cell surface expression levels. However, we observed a 5–10% decrease in total β1 expression levels in Mlck210^−/− cells (Supplementary Fig. 4a, b). The integrin β1 subunit can partner with over 10 distinct α subunits and is expressed well in excess of any α4 subunit; therefore, these results indicate that Mlck210 deletion does not affect integrin α4 cell surface expression and does regulated α4 integrin activation. As we previously showed that VCAM-Fc binding to myeloid cells strictly depends on integrin α4β1 expression and function^6,7, these results show that MLCK210 promotes integrin α4β1 ligand binding (activation) in murine and human myeloid cells.

a Quantification of VCAM-Fc binding to IL-1β, SDF-1α, VEGF-A, or Mn2+ stimulated vehicle or MW01-022AZ treated human CD11b+ myeloid cells (n = 3). *p < 0.0001 for IL-1β, SDF-1α, and VEGF-A stimulated, MW01-022AZ vs vehicle-treated cells. No significant difference (ns) for basal (p = 0.9965) or Mn2+ (p = 0.6015) stimulated MW01-022AZ vs vehicle-treated cells. bFlow cytometry profiles of cells from a. c Quantification of HUTS21 or P4C10 antibody binding to basal medium, SDF-1α or Mn2+ stimulated, vehicle or MW01-022AZ treated human CD11b+ cells (n = 3).*p < 0.0001 for HUTS21 binding to SDF-1α stimulated and *p = 0.0002 to Mn2+-stimulated MW01-022AZ vs vehicle-treated cells. No significant difference (ns) for basal MW01-022AZ vs vehicle cells (p > 0.9999). No significant difference (ns) for P4C10 binding to vehicle vs MW01-022AZ treated basal (p = 0.1384), SDF-1α (p > 0.9999) or Mn2+ (p = 0.9997) stimulated cells. d FACs profiles of cells from c. e Quantification of VCAM-Fc binding to SDF-1α or Mn2+ stimulated murine CD11b+ myeloid cells in MW01-022AZ and Vehicle-treated cells (n = 3). *p < 0.0001 for SDF-1α stimulated cells. No significant difference (ns) for basal (p > 0.9999) or Mn2+ (p = 0.9991)-treated cells. f FACs profiles of cells from e. g VCAM-Fc binding (MFI) to basal medium, SDF-1α IL-1β, IL-6, VEGF-A, TNFα, and Mn²⁺ stimulated murine myeloid cells from WT or Mlck210^−/− mice (n = 3). *p < 0.0001 for WT vs Mlck210^−/−SDF-1α IL-1β, IL-6, VEGF-A, TNFα-stimulated cells. No significant differences (ns) for secondary antibody alone (p > 0.9999), basal (p = 0.9997), or Mn2+ (p = 0.9928) treated cells. h VCAM-Fc binding (MFI) to basal medium, SDF-1α and Mn²⁺ stimulated murine myeloid cells from control All stars or Mlck siRNA-transfected cells (n = 3). *p < 0.0001 for SDF1α and Mn2+ stimulated Mlck siRNA vs All Stars transfected cells. No significant differences (ns) for basal Mylk siRNA vs All Stars transfected (p > 0.9999) or untransfected vs All Stars transfected cells (p = 0.9998). For panels a, c, e, g, h, significance determined by one-way Anova with Tukey’s multiple comparisons. Analysis of VCAM-Fc, HUTS, or P4C10 binding was by single color flow cytometry. All replicates are biologically independent samples. Experiments were performed twice. Data are presented as mean values ± SEM. Source data are provided with this article as a Source Data file.

Full size image

MLCK210 is required for Rap1 activation

Our previous studies showed that PI3Kγ promotes integrin activation and adhesion by activating Rap1⁷. To determine if MLCK210 plays a role in integrin activation and adhesion upstream of PI3Kγ, we analyzed Akt phosphorylation in response to cytokine/chemokine stimulation in WT and Mlck210^−/− myeloid cells. Deletion of MLCK210 had no effect on Akt phosphorylation (Fig. 4a). These results suggest that MLCK210 acts downstream of PI3Kγ. We next asked whether MLCK210 could play a role in Rap1 activation, as Rap1 activation is a late step in the signaling cascade leading to integrin activation²⁷. Indeed, we found that Rap1 activation, as measured by GTP loading, was impaired in SDF-1α-stimulated Mlck210^−/− but not WT myeloid cells (Fig. 4b). In previous studies, we observed that expression of activated Ras, PI3Kγ, and Rap1 promoted integrin activation and adhesion in the absence of cytokine stimulation⁷. Importantly, we found that expression of activated Ras (RasG12V) or PI3Kγ (p110γCAAX) constitutively activated adhesion of WT but not Mlck210^−/− myeloid cells, suggesting that MLCK210 is required for myeloid cell adhesion downstream of Ras and PI3Kγ (Fig. 4c; Supplementary Fig. 5a). In contrast, expression of activated RapG12V stimulated α4β1-mediated adhesion in WT and in Mlck210^−/− myeloid cells (Fig. 4c; Supplementary Fig. 5a), indicating that MLCK210 promotes integrin activation downstream of Ras and PI3Kγ and upstream of Rap.

a Akt phosphorylation in WT and Mlck210^−/− cytokine-stimulated murine myeloid cells. b Rap1 GTP loading in cytokine-stimulated WT and Mlck210^−/−murine myeloid cells. c Adhesion to VCAM of WT and Mlck210^−/− murine myeloid cells (n = 3) after transfection with active Ras (G12VRas), active PI3Kγ (p110γCAAX), or active Rap (RapG12V). p = 0.9791 untransfected WT vs Mlck210^−/−, *p < 0.0001 G12VRas and p110γCAAX WT vs Mlck210^−/−, p = 0.9245 RapV12 WT vs Mlck210^−/−. Significance testing by unpaired t-test. d Myosin Light Chain (MLC) S18/Thr19 phosphorylation in basal or cytokine-stimulated WT and Mlck210^−/− murine myeloid cells. e Immunoblotting to detect Flag-tagged MLCK210, total MLCK210 and HSP90 in WT and Mlck210^−/− murine myeloid cells transfected with control (pcDNA) and MLCK210 expression plasmids. f Immunoblotting to detect phosphorylated MLC and total MLC in SDF1a stimulated or unstimulated WT or Mlck210^−/− myeloid cells that had been transfected with control (pcDNA) or flag-tagged WT and Kinase-dead (KD) MLCK210 expression plasmids. Graph: Relative pS18/Thr19-MLC expression in 3 independent samples. g Cytokine-stimulated VCAM-1 adhesion of WT and Mlck210^−/− myeloid cells transfected with control (pcDNA), WT or Kinase Dead (KD) MLCK210 expression constructs (n = 3). *p = 0.0001 for cytokine-stimulated Mlck210^−/− cells and Mlck210−/− cells expressing Mlck210KD cDNA vs WT cells. No significant differences (ns) for WT vs Mlck210^−/− cells expressing Mlck210 WT cDNA, p = 0.9936 for IL1β and p > 0.9999 for SDF-1a and VEGF-A treated cells. No significant differences (ns) for basal conditions (p> 0.9999). Significance determined by unpaired t-test for 4c and by one-way Anova with Tukey’s multiple comparisons for panels e, g. All replicates are biologically independent samples. Experiments were performed twice. Data are presented as mean values ± SEM. Source data are provided with this article as a Source Data file.

Full size image

Myosin promotes integrin activation

Since the major known substrate for all MLCK isoforms is myosin light chain (MLC), we investigated a role for myosin in integrin α4β1 activation. First, we found that MLC was phosphorylated in WT cells but not in Mlck210^−/− cells (Fig. 4d). Next, we found that MLCK210 catalytic activity was required for MLC phosphorylation in chemokine-treated myeloid cells. While SDF-1 stimulated strongly MLC phosphorylation in WT but not Mlck210^−/− myeloid cells, only expression of Flag-tagged intact MLCK210 (WT) but not Flag-tagged kinase-dead MLCK210 (KD) or control DNA (pcDNA) promoted SDF-1-induced MLC phosphorylation in Mlck210^−/− cells (Fig. 4e, f). MLCK210 catalytic activity was also required for integrin α4β1-mediated adhesion to VCAM, since expression of catalytically active MLCK210 (WT), but not kinase-dead MLCK210 (KD), restored cytokine/chemokine-stimulated adhesion to VCAM-1 in Mlck210^−/− cells (Fig. 4g). Taken together, these results indicate that MLCK210 acts downstream of PI3Kγ but upstream of Rap1 and plays a key role in promoting integrin α4 activation. MLCK can be activated by the influx of calcium ions, resulting in calmodulin-dependent MLCK activation²⁰. To explore how MLCK210 becomes activated by PI3Kγ signaling, we evaluated the effects of the PKC inhibitor R0-32-0432 and the calcium/calmodulin inhibitor W7 on cytokine-mediated adhesion to VCAM. We found that adhesion was suppressed by W7 but not the PKC inhibitor (Supplementary Fig. 5b), indicating a role for calmodulin in MLCK210-dependent integrin activation.

As MLCK210 expression in myeloid cells is required for myosin light chain phosphorylation and for Rap1 activation, our results suggest myosin could play a critical role in integrin α4β1 activation. In fact, myosin has recently been suggested to serve as a scaffold for anchoring signaling proteins at the membrane²⁸. We found that primary murine myeloid cells express non-muscle myosin heavy chain MHCIIA (encoded by Myh9) but not MHCIIB (encoded by Myh10) or MHCIIC (encoded by Myh14) protein and mRNA (Fig. 5a, b). Myh9 siRNA, but not Myh10 siRNA (used here as a negative control siRNA), inhibited Myh9 expression (Fig. 5c), and prevented cytokine-stimulated integrin α4 VCAM-Fc binding in SDF-1α but not Mn2+-stimulated myeloid cells (Fig. 5d–e). Similarly, Myh9 siRNA but not Myh10 siRNA prevented cytokine-stimulated myeloid cell adhesion to VCAM-1-coated substrates (Fig. 5f), which is strictly integrin α4β1 dependent⁶. Myh9 siRNA, but not control Myh10 siRNA, also prevented Rap1 GTP loading in SDF-1α stimulated myeloid cells (Fig. 5g). Importantly, blebbistatin, a myosin-selective ATPase inhibitor²⁹, suppressed cytokine and activated PI3Kγ (p110γCAAX) mediated stimulation of myeloid cell adhesion to VCAM-1, indicating that myosin catalytic activity promotes α4β1-mediated cell adhesion (Fig. 5h). We also observed that integrin α4 (green) colocalized with myosin (red) in SDF-1α bead-stimulated myeloid cells in Mlck210^−/− myeloid cells expressing WT but not kinase-dead MLCK210, indicating that MLCK210 catalytic activity promotes myosin-integrin interactions (Fig. 5i; Supplementary Fig. 5c). Catalytically active MLCK210 was also required to promote chemokine-induced colocalization of α4β1 and Rap1 (Fig. 5i; Supplementary Fig. 5c). Together, these results indicate that MLCK210 phosphorylates and activates myosin, leading to clustering of adaptor and cytoskeletal proteins that are required to activate integrin.

a Immunoblotting of primary murine myeloid cells (CD11b+) and Lewis lung carcinoma (LLC) cells with anti-myosin heavy chain A, B, and C or HSP90 antibodies. b RT-PCR analysis of mRNA expression of Myh9, Myh10, and Myh14 in myeloid cells (n = 3, *p < 0.0001 Myh9 vs Myh10 or Myh14). c Effect of Myh9 and Myh10 siRNA on Myh9 expression (n = 3, p = 0.0568). d, eQuantification (d) and FACS profiles (e) of VCAM-Fc binding to All Stars, Myh9 and Myh10 siRNA-transfected myeloid cells (n = 3). *p < 0.0001 for SDF-1 stimulated, p > 0.9999 for EBM stimulated, and p > 0.9000 for Mn2+ stimulated Myh9 vs All Stars and vs Myh10 siRNA-transfected cells. f Cell adhesion to VCAM of Myh9 and Myh10 siRNA-transfected myeloid cells (n = 3). *p < 0.0001 for SDF-1 and IL-1 stimulated Myh9 vs All Stars and Myh10 siRNA. g Effect of Myh9 and Myh10 siRNA on Rap1 activation by SDF-1α in myeloid cells. h Cytokine and p110γCAAX stimulation of myeloid cell adhesion to VCAM-1 in the presence of vehicle or the myosin ATPase inhibitor blebbistatin (n = 3). p = 0.9812 vehicle vs blebbistatin basal medium, *p < 0.0001 vehicle vs blebbistatin for SDF-1α, IL1β, VEGF, p110γCAAX, and p = 0.9915 vehicle vs blebbistatin for pcDNA. i Immunofluorescence detection of integrin α4 (green) and myosin or Rap1 (red) in SDF1-bead-stimulated Mlck210^−/− myeloid cells that were transfected with WT or kinase-dead Mlck210 cDNA. Nuclei were detected with Dapi (blue). Scale bar = 5 µm. All replicates are biologically independent samples. Data are presented as mean values ± SEM. Significance determined by one-way Anova with Tukey’s multiple comparisons for panels b, d, f, and by unpaired t-test for panels c, h. Experiments were performed twice. Source data are provided with this article as a Source Data file.

Full size image

MLCK210 mediates tumor inflammation and progression

Since MLCK210 is required for integrin α4β1 activation in myeloid cells and myeloid cell infiltration of tumors is integrin α4β1 dependent^6,7,8,10, we speculated that Mlck210^−/− myeloid cells would exhibit defects in trafficking into sites of tumor inflammation in vivo. Mlck210^−/− myeloid cells, like p110γ^−/−, Rap1a^−/−, and integrin α4Y991A myeloid cells^6,7,8,10, exhibited defective recruitment to Lewis lung carcinoma tumors (LLC) but not control tissues, such as spleens, after adoptive transfer into animals with tumors (Fig. 6a–c). Short-term myeloid cell recruitment to LLC tumors was also suppressed when Mylk was knocked down by Mylk siRNA expression in adoptively transferred myeloid cells (Fig. 6d, e). LLC tumor progression as measured by tumor volume over time and endpoint weights were substantially impaired in Mlck210^−/− mice compared to WT mice (Fig. 6f–i), as were tumor volumes and weights in mice bearing pancreatic ductal adenocarcinomas (PDAC) (Fig. 6j–l). To evaluate the effect of Mlck210 deletion on the tumor microenvironment, we performed flow cytometry to analyze myeloid cell content in Mlck210^−/− and WT tumors (Fig. 6m, n). The accumulation of CD11b+ Ly6C+ monocyte/macrophages as well as CD11b+ Ly6G+ granulocytes in tumors was suppressed in Mlck210^−/− mice (Fig. 6m–o; see Supplementary Fig. 6a for FACS gating strategy). As monocyte and granulocyte recruitment to tumors was suppressed in Mlck210^−/− bone marrow, these results suggest that loss of myeloid cell MLCK210 leads to improved immune responses in tumors.

a Schematic of fluorescent myeloid cell adoptive transfer into tumor bearing animals and method of analysis. Quantification of transferred cells was by single color flow cytometry. bNumber of CFDA-labelled WT and Mlck210^−/− CD11b+ cells in tumors 2 h or 24 h after adoptive transfer into mice (n = 3 mice), *p = 0.0011 2 h and *p < 0.0001 24 h WT vs Mlck210^−/− by t-test. c Number of CFDA-labelled WT and Mlck210^−/− CD11b+ cells in spleens 2 h or 24 h after adoptive transfer into mice (n = 3 mice), p = 0.9952 2 h and p = 0.727 24 h WT vs Mlck210^−/− by t-test. d Number of CFDA-labelled control All Stars and Mylk siRNA-transfected CD11b+ cells in tumors 2 h after adoptive transfer into mice (n = 3 mice), *p = 0.011 All Stars vs Mylk siRNA-transfected by t-test. e Number of CFDA-labelled control All Stars and Mylk siRNA-transfected CD11b+ cells in spleens 2 h after adoptive transfer into mice (n = 3 mice). Significance testing by one-way Anova with Tukey’s posthoc testing. f Schematic of LLC lung tumor implantation and growth. g–i Mean ± SEM tumor volume over time (g), endpoint volume *p = 0.0003 (h), and weight *p < 0.0001 (i) of LLC tumors implanted in WT (n = 8 mice) and Mlck210^−/− mice (n = 8 mice). j Schematic of LMP pancreatic carcinoma lung tumor implantation and growth. k Time course of pancreatic ductal adenocarcinoma (LMP) tumor growth in WT and Mlck210^−/− tumors (n = 4 mice). l Weight of pancreatic ductal adenocarcinoma tumors from in WT and Mlck210^−/− tumors (n = 4 mice) *p = 0.0019 for WT vs Mlck210^−/− BM. m FACs profiles of myeloid cells in WT and Mlck210^−/− LLC tumors. For Facs gating scheme see Supplementary Fig. 7a. n Quantification of myeloid cells in WT and Mlck210^−/− LLC tumors (n = 4 mice). Significance testing was performed by unpaired two-sample Student’s t-test. Data are presented as mean values ± SEM. Experiments were performed twice. Source data are provided with this article as a Source Data file.

Full size image

In prior studies, we used bone marrow transplantation to demonstrate that immune cell PI3Kγ and integrin α4 are both necessary for tumor inflammation and growth^6,7. We performed similar bone marrow transplantation studies to determine whether immune cell MLCK210 promotes tumor growth in vivo (Fig. 7a). WT mice were irradiated, and bone marrow was reconstituted with either WT or Mlck210^−/− bone marrow; reconstitution was verified by assessing Mlck210 genotypes in DNA isolated from peripheral blood leukocytes (Fig. 7b). Tumor growth, as measured by gross morphology, volume, and weight, was suppressed in mice reconstituted with Mlck210^−/− but not WT bone marrow (BM) (Fig. 7c–e). Residual tumors in mice transplanted with Mlck210^−/− BM were less than 75 mm³.

a Schematic of bone marrow transplantation and tumor implantation. b Genomic PCR to detect Mlck210 gene in peripheral blood cells from BMT transplanted mice (n = 2 mice). c–e Tumor volumes (c), weights at endpoint (15 days) (*p < 0.0001 WT vs Mlck210^−/− BM) (d) and images of gross tumor morphology (e) from LLC tumors in WT mice transplanted with bone marrow from WT (n = 10 mice) or Mlck210^−/− mice (n = 10 mice). f Representative FACS profiles and g quantification of Gr1-CD11b+ (p = 0.9724) Gr1loCD11b+ (*p = 0.0280), Gr1hiCD11b+ (p = 0.8641) myeloid cells in representative WT vs Mlck210^−/− BM transplanted tumors harvested at endpoint (15 days) (n = 3 mice). For Facs gating scheme, see Supplementary Fig. 7b. hRepresentative histographs and i quantification of F4/80+ macrophages/field (n = sections from 5 WT mice and 5 Mlck210^−/−) *p = 0.0009, CD31+ blood vessels/field (n = sections from 5 WT and 3 Mlck210^−/− mice) *p = 0.001, CD8+ T cells/field (n = sections from 5 WT and 4 Mlck210^−/− mice) *p = 0.019 in tumor sections and 100X H&E-stained tumor sections to detect percent necrosis/field (n = sections from 5 mice) *p = 0.0004 from mice transplanted with WT vs Mlck210^−/− bone marrow. Data are presented as mean values ± SEM. Significance testing was performed by Student’s t-test. Experiments were performed twice. Source data are provided with this article as a Source Data file.

Full size image

To evaluate the effect of Mlck210 deletion on the tumor microenvironment, we performed flow cytometry and immunofluorescence analysis of myeloid cell, endothelial cell, and T cell content in tumors from Mlck210^−/− and WT BM transplanted tumors. Using flow cytometry, we determined that the major effect of MLCK210 loss was on recruitment of Gr1lo bone-marrow-derived monocyte/macrophages to tumors (Fig. 7f, g; see Supplementary Fig. 6b for FACS gating strategy). Immunofluorescence analysis of frozen tumors demonstrated that deletion of Mlck210 strongly reduced both macrophage and endothelial cell content in tumors, while increasing CD8+ T cell content in tumors (Fig. 7h, i). Tumors in mice bearing Mlck210^−/− bone marrow were strikingly more necrotic than WT tumors (Fig. 7i). As monocyte and macrophage recruitment to tumors was suppressed, while T cell recruitment and necrosis were increased in mice with Mlck210^−/− bone marrow, these results suggest that loss of myeloid cell MLCK210 leads to improved immune responses in tumors. To determine whether MLCK210 loss might affect T cells directly, we examine the expression of MLCK210 in T cells isolated from WT and Mlck210^−/− animals and performed western blotting to detect MLCK isoforms. We found that T cells only expressed the 108 kDa isoform of MLCK while myeloid cells expressed the 210kD isoform (Supplementary Fig. 7). These results indicate that depletion of MLCK210 in bone-marrow-derived myeloid cells promotes the recruitment of CD8+ T cells into tumors.

To examine further whether MLCK210 in myeloid cells promotes tumor growth, we evaluated the effect of clodronate and control liposomes on tumor growth in the WT and MLCK210^−/− background (Fig. 8a). Clodronate is a bisphosphonate that is ingested by macrophages, leading to apoptotic cell death. Clodronate liposomes but not control liposomes reduced tumor growth in the WT background to the same degree as did Mlck210 deletion (Fig. 8b, c); however, clodronate had no additive inhibitory effect on tumor growth in the Mlck210^−/− background (Fig. 8b, c). Macrophages were efficiently depleted from WT and Mlck210 knockout mouse livers (Fig. 8d, e) and variably reduced in tumors (Fig. 8f, g). The results support the conclusion that Mlck210 deletion, like clodronate liposomes, interferes with the accumulation of tumor-promoting macrophages.

a Schematic of clodronate and control liposome treatment of LLC tumor bearing mice. b, c LLC tumor volumes over time (b) and endpoint tumor volumes (c) for WT (n = 6 mice treated with control liposomes (con) and 8 mice treated with clodronate liposomes (clod) and for Mlck210^−/− [n = 7 mice treated with control liposomes (con) and clodronate liposomes (clod)]. One WT animal treated with clodronate liposomes died on day 14 post-tumor inoculation and was not included in the final tumor weight graph. d–g Quantification (d, f) and FACs plots of Ly6C- (e) liver and (g) tumor macrophages from WT and Mlck210^−/− mice after treatment with clodronate and control liposomes (n = 4 mice). For Facs gating scheme, see Supplementary Fig. 7a. h Schematic depicting anti-CD8 and control antibody treatment of HPV+ Meer WT and Mlck210^−/− tumor bearing mice. i Time course of tumor growth for control and anti-CD8 antibody-treated WT and Mlck210^−/− mice (n = 12 WT and n = 8 Mlck210^−/− control mice, and n = 11 WT and n = 8 Mlck210^−/− anti-CD8-treated mice). Statistical significance determined by linear regression analysis. j, k Quantification (j) and FACs plots (k) of tumor associated CD8+ and CD4+ T cells in mice from i (n = 4 mice). For Facs gating schemes, see Supplementary Fig. 7c. l, m Quantification (n = 4 mice) and FACs plots (m) of splenic CD8+ and CD4+ T cells in mice from i. For Facs gating schemes, see Supplementary Fig. 7c. Statistical significance determined by one-way Anova with Tukey’s posthoc testing. Data are presented as mean values ± SEM. Experiments were performed twice. Source data are provided with this article as a Source Data file.

Full size image

To determine whether recruitment of CD8+ T cells is required for the inhibition of tumor growth by Mlck210 deletion, we evaluated the effect of CD8+ T cell depletion on tumor growth in WT and Mlck210^−/− animals (Fig. 8h). Antibody-mediated depletion of CD8+ T cells prevented inhibition of tumor growth in Mlck210^−/− mice but had little added effect on tumor growth in WT mice (Fig. 8i). We observed that tumors implanted in the Mlck210^−/− background recruited substantially more (approximately five times more) CD8+ T cells to tumors, while CD4+ T cells content was not significantly different in tumors in Mlck210^−/− and WT mice (Fig. 8j, k; see Supplementary Fig. 6a for Facs gating strategy). CD8+ T cell antibody treatment effectively removed CD8+ T cells from both spleen and tumors (Fig. 8j–m). Based upon these studies, we concluded that Mlck210 deletion in myeloid cells promotes to CD8+ T cell recruitment to tumors and subsequent tumor suppression.

Importantly, tumor growth and inflammation were also strongly suppressed in mice treated with the MLCK inhibitor MW01-022AZ (Fig. 9a–d). Gr1hi granulocyte, as well as Gr1lo monocyte/macrophage infiltration, were suppressed in drug-treated tumors (Fig. 9e, f). Importantly, macrophage and endothelial cell content was suppressed, while T cell content was increased in drug-treated tumors (Fig. 9g, h). As we previously documented that tumor associated myeloid cells promote angiogenesis and inhibit T cell recruitment to tumors, these studies support the conclusion that myeloid cell MLCK210 plays a significant role in regulating tumor inflammation and growth in vivo by promoting myosin-mediated integrin activation and myeloid cell recruitment.

a Schematic of LLC lung tumor implantation in C57Bl6 mice and MW01-022AZ treatment. b–d Effect of MW01-022AZ on LLC b gross tumor morphology (n = tumors from 6 mice) c tumor volume (n = 9 mice and d tumor weight at endpoint (21 days) (n = 9 mice). *p < 0.0001 for vehicle vs MW01-022AZ treatment, determined by t-test. e FACS profiles and f quantification of Gr1-CD11b+ (p = 0.1146) Gr1loCD11b+ (*p = 0.0402), Gr1hiCD11b+ (*p = 0.0146) myeloid cells in representative MW01-022AZ vs vehicle-treated tumors at endpoint (21 days) (n = 3 mice), determined by t-test. For Facs gating scheme, see Supplementary Fig. 7b. g Quantification and h representative histographs of F4/80+ macrophages (n = sections from 6 mice) *p < 0.0001 for vehicle or MW01-022AZ by t-test, CD31+ blood vessels (n = sections from 7 mice) *p < 0.0001 for vehicle or MW01-022AZ by t-test, CD8+ T cells (n = sections from 9 mice) *p= 0.0244 for vehicle vs MW01-022AZ by Mann–Whitney test in tissues from mice treated with vehicle or MW01-022AZ. i Model of integrin activation demonstrating that cytokine stimulation activates a pathway whereby PI3Kγ induces MLCK210 and myosin activation and colocalization with Rap1, resulting in Rap1-mediated integrin α4 conformation changes and interaction of the integrin with the actin cytoskeleton. Data are presented as mean values ± SEM. Statistical significance determined by Student’s t-test. Experiments were performed twice. Source data are provided with this article as a Source Data file.

Full size image

Our prior studies showed that pharmacological or genetic blockade of PI3Kγ suppressed integrin α4β1-mediated myeloid cell adhesion and recruitment of monocytes and granulocytes into tumors and inhibited disease progression^6,7,30. Here we identified high molecular weight myosin light chain kinase, MLCK210, as a key regulator of integrin activation during tumor inflammation and progression. We found that MLCK210 promotes myosin-dependent, Rap1-mediated, integrin α4β1 activation, leading to myeloid cell adhesion to endothelium, tumor inflammation, and tumor progression.

Our studies showed MLCK210 colocalizes with integrin α4β1 in the plasma membrane of cytokine and chemokine-stimulated myeloid cells and that integrin α4β1 ligand binding and adhesion in response to cytokine/chemokine stimulation requires MLCK210. We found that Rap1 activation, which is required for integrin α4 activation, is impaired in MLCK210^−/− myeloid cells; expression of activated Rap1 restored adhesive functionality to Mlck210^−/− cells. We also showed that MLCK210 controls integrin activation downstream of PI3Kγ, a key regulator of tumor inflammation. These results identify a key PI3Kγ-MLCK210-myosin-Rap-integrin α4 molecular pathway that regulates tumor inflammation (Fig. 9i). As myeloid cells lacking MLCK210 exhibit defects in integrin α4β1 activity despite high expression levels of the non-muscle MLCK108 isoform, these results suggest that the unique N-terminal domain of MLCK210 plays a role in promoting integrin activation. Importantly, inflammation and tumor growth were substantially impaired in Mlck210^−/− mice and in mice transplanted with Mlck210^−/− bone marrow. These results are similar to those observed in PI3Kγ^−/− and integrin α4Y991A mutant mice^6,7,30, indicating that MLCK210 plays a significant role in integrin activation in vivo.

Our studies show that MLCK210 expression in myeloid cells is required for MLC phosphorylation and for Rap1 activation, suggesting that the MCLK substrate myosin could play a critical role in integrin activation. Indeed, knockdown of Myh9, the major myosin isoform expressed in myeloid cells, as well as the myosin ATPase inhibitor blebbistatin inhibited integrin activation and adhesion. Support for a role of myosin in integrin activation is also provided by studies showing that integrin α4-myosin interactions promote α4-mediated activation in transfected Chinese Hamster Ovary cells³⁰ and that myosin IIA-Rap1 interactions can promote Rap1-dependent integrin activation in fibroblasts^31,32.

Activation of integrin α4β1 by inside-out signaling promotes leukocyte extravasation from the vasculature^9,12. Integrin activation depends on Rap1^6,7 and its effector protein RIAM, which localizes talin to the membrane^33,34,35. Talin binds integrin β chain cytoplasmic tails, resulting in an allosteric increase in ligand binding affinity^33,34,35. Paxillin binding to the integrin α4 cytoplasmic tail promotes integrin activation, adhesion, and trafficking of lymphocytes and myeloid cells^9,10,11,12. We showed here that integrin α4β1, MLCK210, talin, and paxillin are all colocalized in cytokine-stimulated myeloid cells and that MLCK210 expression is necessary to promote Rap1 and paxillin colocalization with α4 integrin. Our data show that MLCK210 catalytic activity is required to promote integrin clustering and adhesion, as catalytically active MLCK210 but not inactive MLCK210 colocalizes with α4β1 and restores integrin clustering and adhesion to Mlck210^−/− cells. Furthermore, MLCK210 phosphorylates myosin light chain, thereby inducing myosin ATPase activity and integrin activation in myeloid cells. The unique N-terminal head group of MLCK210 has been shown to target this kinase to the membrane and to actin filament³⁶, suggesting that MLCK210 may promote integrin activation by facilitating colocalization of integrin, signaling proteins, and cytoskeletal components (Fig. 9i).

Integrin cytoplasmic tails play critical roles in integrin activation. Integrin alpha and beta subunit cytoplasmic tails interact with each other through salt bridges between the membrane proximal GFKKR motif of the alpha subunit and the membrane proximal HDRRE motif of the beta chain^{37,38,39,40,41,42}. R→A mutations in the GFFKR sequence in the cytoplasmic tails of α4, αIIb, and αL constitutively activate these integrins⁴³. Disruption of this salt bridge forces the cytoplasmic tails apart and alters extracellular domain conformations to increase ligand-binding affinity. The interaction of talin with NPxY sites in integrin β chain cytoplasmic tails also alters the packing of transmembrane domains and activates integrins⁴³. Talin binding is facilitated by Rap1 activation, which promotes Rap1 effector protein RIAM to localize talin near the membrane⁸. Paxillin binding to the integrin α4 cytoplasmic tail also enhances integrin activation, as a Y→A mutation of the Y991 paxillin-binding site to A991 reduces talin binding and inhibits adhesion and trafficking of lymphocytes and myeloid cells^6,10,11,12. Our studies here show that MLCK210 also play a key role in activating integrin α4β1 by promoting myosin-dependent Rap1 activation and localization and subsequent Talin and Paxillin binding to the integrin cytoplasmic tail, thereby contributing to cancer progression. A previous study described a role for MLCK in β2 integrin activation; those studies described a myosin-independent role for low molecular weight MLCK in beta subunit integrin activation⁴⁴. In contrast, our studies demonstrate a role for a high molecular weight variant of MLCK, MLCK210, in mediating integrin activation, inflammation, and cancer progression.