Mitofusin 2 regulates neutrophil adhesive migration and the actin cytoskeleton

Neutrophils rely on glycolysis for energy production. How mitochondria regulate neutrophil function is not fully understood. Here, we report that mitochondrial outer membrane protein Mitofusin 2 (Mfn2) regulates neutrophil homeostasis in vivo. Mfn2-deficient neutrophils are released from the hematopoietic tissue and trapped in the vasculature in zebrafish embryos. Human neutrophil-like cells deficient with MFN2 fail to arrest on activated endothelium under sheer stress or perform chemotaxis. Deletion of Mfn2 results in a significant reduction of neutrophil infiltration to the inflamed peritoneal cavity in mice. Mfn2, but not Mfn1, -null mouse embryonic fibroblast cells have altered actin structure and are impaired in wound closure. MFN2-deficient neutrophil-like cells display heightened intracellular calcium levels and Rac activation after chemokine stimulation. Mechanistically, MFN2 maintains mitochondria-ER interaction. Restoring mitochondria-ER tether rescues the chemotaxis defect and Rac activation resulted from MFN2 depletion. Finally, inhibition of Rac restores chemotaxis in MFN2-deficient neutrophils. Altogether, we identified that MFN2 regulates neutrophil migration via suppressing Rac activation and uncovered a previously unrecognized role of MFN2 in regulating the actin cytoskeleton.


Introduction
Neutrophils, the most abundant circulating leukocytes in humans, constitute the first line of host defense. Upon stimulation by either pathogen or host-derived proinflammatory mediators, neutrophils are recruited to inflamed tissue using spatially and temporally dynamic intracellular 4 signaling pathways. Activation of the surface receptors, primarily the G-protein-coupled receptors (de Oliveira et al., 2016;Gambardella and Vermeren, 2013;Mocsai et al., 2015;Pantarelli and Welch, 2018), leads to the activation of phosphatidylinositol 3-kinases (PI3K) that produces phosphatidylinositol (3,4,5)P3 and activates small GTPases such as Rac. Rac promotes actin polymerization at the leading edge and drives cell migration (Futosi et al., 2013). In parallel, G-protein-coupled receptors activate phospholipase C, which generates IP3 and promotes the Ca 2+ release from intracellular stores (Tsai et al., 2015). Although intracellular Ca 2+ is a well characterized second messenger that activates Rac and regulates cell migration in slowly migrating cells (Price et al., 2003), its role in neutrophil migration is less clear.
Cell migration requires the coordination of multiple cellular organelles, including mitochondria. Mitochondria carry out oxidative phosphorylation to produce ATP, regulate the intracellular redox status and distribution of Ca 2+ that can regulate cell migration. In addition, mitochondria morphology changes via fusion and fission (Campello and Scorrano, 2010) to adapt to changing metabolic needs under different conditions. Mitochondria fission promotes cell migration by providing mitochondria and ATP at energy demanding sites such as the protrusion or the uropod (Campello et al., 2006;Zhao et al., 2013).
In neutrophils, mitochondrial biology is distinct and contradictory. The Warburg effect is documented in neutrophils that they primarily use glycolysis for ATP generation (Borregaard and Herlin, 1982). Neutrophils have a relative low number of mitochondria, low respiration rates and low enzymatic activity of the electron transport chain (Peachman et al., 2001).
However, disrupting mitochondrial membrane potential by pharmacological inhibitors abolished chemotaxis of primary human neutrophils (Bao et al., 2015;Bao et al., 2014;Fossati et al., 2003). Although mitochondria-derived ATP possibly regulates neutrophil chemotaxis in vitro (Bao et al., 2015), removal of extracellular ATP improves neutrophil chemotaxis in vivo (Li et al., 2016). These conflicting reports prompted us to search for mechanisms delineating the role of mitochondria in neutrophil migration outside the realm of ATP or cellular energy (Bi et al., 2014;Schuler et al., 2017;Zanotelli et al., 2018).
Human neutrophils are terminally differentiated and undergo apoptosis within 24 hours in culture and thus are not genetically tractable. We have overcome this hurdle by developing a neutrophil-specific knockout platform in zebrafish (Zhou et al., 2018a). The zebrafish is a suitable model for neutrophil research because of its highly conserved innate immune system.
In our previous work, we have confirmed the requirement of mitochondrial membrane potential and electron transport chain in the migration of zebrafish neutrophils (Zhou et al., 2018a). In addition, we have visualized a highly fused tubular network of mitochondria in zebrafish neutrophils, which is consistent with a previous report investigating primary human neutrophils (Maianski et al., 2002). Here we present evidence that Mitofusin 2 (Mfn2) regulates Rac activation to coordinate neutrophil adhesion and migration. In addition, we reveal a previously unknown function of Mfn2 in regulating the actin cytoskeleton, contributing to the understanding and management of patients with Mfn2-related mitochondrial diseases.

Neutrophils depleted with mfn2 accumulate in zebrafish vasculature
To address whether a highly fused mitochondrial network benefits neutrophil migration, we generated zebrafish transgenic lines with neutrophil specific deletion of proteins that regulate mitochondrial shape. Mitofusins, Mfn1 and Mfn2, are required for mitochondrial outer membrane fusion (Chen et al., 2003) and Opa1 regulates inner membrane fusion (Song et al., 2007). In embryos with mfn2 deletion in neutrophils, Tg(lyzC:Cas9-mfn2 sgRNAs) pu23 , a majority of neutrophils circulate in the bloodstream (Fig. 1a, b and Supplementary Movie 1). This is in sharp contrast to the control or the wild-type embryos in which over 99% of neutrophils are retained in the caudal hematopoietic tissue or in the head mesenchyme (Harvie and Huttenlocher, 2015). This abnormal distribution of neutrophils was further confirmed in a second transgenic line expressing different sgRNAs targeting mfn2, Tg(lyzC:Cas9-mfn2 sgRNAs#2) pu24 (Fig. 1a, b and Supplementary Movie 2). Neutrophils were sorted from both lines and their respective loci targeted by the 4 sgRNAs were deep sequenced. The overall mutation frequency ranged from 24% to 60%. In contrast, circulating neutrophils were not observed in embryos expressing sgRNAs targeting opa1, although the velocity of neutrophil migration in the head mesenchyme was significantly reduced (Supplementary Fig. 1 and Supplementary Movie 3), indicating that decreased neutrophil retention in tissue is not simply due to defects in mitochondrial fusion.
Next, we determined whether neutrophils in the vasculature were able to respond to acute inflammation induced by a tail transection or perform chemotaxis to LTB 4 . Significant defects in both assays were observed in the line with neutrophil specific mfn2 deletion (Fig. 1c-f and   7 Supplementary Movie 4). Taken together, mfn2 regulates neutrophil tissue retention and extravasation in zebrafish.

MFN2 regulates adhesion and migration in of neutrophils in vitro and in vivo
To get to the mechanism how Mfn2 regulates neutrophil migration, we knocked down MFN2 in human neutrophil-like cells, HL-60, using shRNAs and obtained two individual lines with 80% and 50% reduction of MFN2 ( Mfn2 is required for neutrophil infiltration in mice, we bred Mfn2 flox/flox mice  with the S100A8-Cre strain (Abram et al., 2013) for neutrophil specific depletion. With 50% of Mfn2 transcript reduction in neutrophils obtained in this strain, significant reduction of 8 neutrophil infiltration into the inflamed peritoneal cavity was observed ( Fig. 2 k-m). Blood cell composition was not altered by Mfn2 depletion (28% and 32% granulocytes in the Cre + and Crelines, respectively), consistent with a previous report that Mfn2 does not regulate blood cell development under hemostatic conditions (Luchsinger et al., 2016). Therefore, MFN2 is required for neutrophil chemotaxis and infiltration in mammals.

Mfn2 regulates the actin cytoskeleton and migration of mouse embryonic fibroblasts
In addition to neutrophils, we investigated the function of Mfn2 in mouse embryonic fibroblasts  We have demonstrated a role of Mfn2 in regulating cell migration in different model systems.
Next we want to investigate the underlying molecular mechanisms. Mfn2 localizes to both the mitochondria and the ER membrane and regulates the tethering between the two organelles in MEF cells (Naon et al., 2016). In dHL-60 cells, MFN2 colocalized with both the mitochondria and the ER, with Manders' colocalization coefficiencies of 0.60±0.085 and 0.69±0.13, respectively ( Fig. 4a, b). Mitochondria also colocalized with the ER (Manders' colocalization coefficiency 0.52±0.097) and distributed throughout the cell body ( Fig. 4c, d). The morphology of the mitochondria and the ER was further visualized using electron microscopy in dHL-60 cells ( Supplementary Fig. 4). When MFN2 was inhibited, mitochondria lost their structure and interaction with ER, and formed a cluster in the middle of the cell body (Fig. 4c, d).
Interestingly, reconstitution of the MFN2 knockdown dHL-60s with an artificial tether (which bridges mitochondria and ER independent of MFN2) (de Brito and Scorrano, 2008) restored the morphology and structure of mitochondria in MFN2-deficient dHL-60 ( Fig. 5a-d). Furthermore, expression of the artificial tether rescued the chemotaxis defect in MFN2-deficient dHL-60 cells

MFN2 suppresses RAC activation in dHL-60 cells
The close proximity of the ER and the mitochondria regulates multiple cellular signaling pathways including calcium homeostasis (de Brito and Scorrano, 2008). Indeed, MFN2-deficient dHL-60 cells exhibited higher levels of Ca 2+ in the cytosol and reduced levels in the mitochondria after fMLP stimulation (Fig. 6a, b). ATP levels were not affected in the MFN2 knockdown dHL-60 cells ( Supplementary Fig. 5a), in line with the observation that mitochondria are not a major source of ATP in neutrophils (Amini et al., 2018;Borregaard and Herlin, 1982).
Mitochondrial membrane potential and the ROS level in mitochondria were also reduced when MFN2 was depleted, especially after fMLP stimulation ( Supplementary Fig. 5b-e). This may due to the altered level of Ca 2+ in mitochondria, since mitochondrial Ca 2+ activates the electron transportation chain (Glancy et al., 2013). We attempted to buffer cytosolic Ca 2+ using BAPTA to determine whether elevated cytosolic Ca 2+ is responsible for the chemotaxis defects in MFN2 knockdown cells. However, a global cytosolic Ca 2+ inhibition abrogated the ability of dHL-60 to migrate ( Supplementary Fig. 6a, b), possibly due to the requirement of precisely regulated cytosolic Ca 2+ , spatially and/or temporally, for neutrophil migration (Mandeville and Maxfield, 1997;Marks and Maxfield, 1990). The mitochondrial calcium uniporter (MCU) is one of the primary sources of mitochondrial uptake of calcium which regulates migration of many cell types including primary human neutrophils (Zheng et al., 2017). The MCU inhibitor Ru360 did not cause further reduction of chemotaxis in MFN2 knockdown dHL-60 cells (Supplementary Notably, the predominant cortical actin, nascent focal contacts and extensive membrane ruffles in Mfn2-null MEF cells (Fig. 3) resembled the classic phenotype of fibroblasts expressing the constitutively active Rac (Hall, 1998), indicating that Rac may be over-activated in Mfn2depleted cells. Rac activation follows different time courses when cells are in suspension and when adhering to substrates (He et al., 2013). To measure Rac activation, we plated the cells on substrate-coated plates and used the phosphorylation of PAK (Graziano et al., 2017) as a readout.
The phosphorylation of PAK peaked at 30 s post stimulation and returned to the baseline at 5 m post stimulation in control cells. Whereas in the MFN2 deficient cells, the phosphorylation of PAK were elevated at all the time points investigated compared with the control (Fig. 6c,  Movie 13), further confirming that MFN2 regulates neutrophil migration through suppressing RAC activation. Together, Mfn2 mediated mitochondria-ER tether regulates the mitochondrial calcium uptake and RAC signaling to orchestrate chemotaxis in dHL-60 cells.

Discussion
Here we report that Mfn2 is crucial for neutrophil adhesion and migration, providing evidence that Mfn2 regulates the actin cytoskeleton and cell migration. By maintaining the tether between the mitochondria and ER, Mfn2 orchestrates intracellular Ca 2+ signaling and regulates Rac activation. Therefore, we have identified the mechanism for how Mfn2 regulates neutrophil adhesive migration, and highlighted the importance of mitochondria and their contact with the ER in neutrophils.
Mfn1 and Mfn2 possess unique functions, although both mediate mitochondrial outer membrane fusion (Chen et al., 2003). Our data that only MFN2, but not MFN1, is required for neutrophil adhesive migration supports that mitochondrial fusion is not likely important for neutrophil migration. Our observation goes against a body of literature that mitochondrial fission promotes cell migration in other cell types (Campello et al., 2006;Zhao et al., 2013). Alternatively, we propose a model that, in neutrophils, the interaction of mitochondria with the ER is more critical than a fused network. Although MFN2 is not the only protein that can mediate the mitochondria-ER tether (Eisenberg-Bord et al., 2016), mitochondria and ER interaction was significantly reduced upon MFN2 deletion in dHL-60, suggesting that MFN2 is at least one of the critical tether proteins in neutrophils. The fused mitochondrial network in neutrophils is possibly a result of the abundant expression of the mitofusins. Mutations in human MFN2 cause Charcot-Marie-Tooth disease type 2A (CMT2A), a classical axonal peripheral sensorimotor neuropathy (Zuchner et al., 2004). MFN2 is also implicated in many other diseases such as cancer, cardiomyopathies, diabetes and Alzheimer's disease (Filadi et al., 2018). Currently, over 100 dominant mutations in the MFN2 gene have been reported in CMT2A patients though how these mutations lead to disease is still largely unknown. The challenges in MFN2 research are that MFN2 regulates mitochondrial fusion and a plethora of cellular functions such as mitochondrial dynamics, transport, mtDNA stability, lipid metabolism and survival (Chandhok et al., 2018). In addition, gain-of-function and loss-of-function mutations are reported that affect different aspects of cellular functions (Chandhok et al., 2018). Our findings provide a new direction to understand the consequences of MFN2 deficiency in disease pathology, namely the actin cytoskeleton and Rac. Our findings also imply a possibility that the defects in immune cell migration in humans may affect immunity or chronic inflammation and indirectly regulate the 13 progression of the aforementioned diseases. Future work will be required to carefully evaluate the individual mutations of MFN2 identified in human diseases in immune cell migration. It is possible that mutations disrupting mitochondria-ER tether, but not membrane fusion, result in defects in cell adhesion and the cytoskeleton regulation.
Our conclusions present a significant departure from the prevailing focus of bioenergy, in other word ATP, in cell migration. In many cell types, including neutrophils, the relevance of mitochondria-derived ATP in cell migration is emphasized (Bao et al., 2015;Bao et al., 2014). A recent report has confirmed the established literature that mitochondria do not provide ATP in neutrophils (Amini et al., 2018). Intriguingly, OPA1 deletion suppresses the production of neutrophil extracellular traps and alters the cellular ATP levels by indirectly suppressing glycolysis. In contrast, MFN2 deletion does not affect ATP levels ( Supplementary Fig. 6) or affect neutrophil extracellular trap formation (Amini et al., 2018), suggesting again distinct biological functions of OPA1 and MFN2. In vascular endothelial cells, mitochondria also serve as signaling rather than energy-producing moieties (Lugus et al., 2011). In our study, in addition to the altered Ca 2+ level, mitochondrial membrane potential and ROS, both critical for neutrophil chemotaxis and migration (Fossati et al., 2003;Zhou et al., 2018a), were reduced in stimulated MFN2-deficient dHL-60 cells. It remains to be determined whether the modest decreases in mitochondria membrane potential or ROS contribute to the defect of neutrophil migration upon MFN2 depletion.
Our result in leukocyte is consistent with previous results in murine fibroblasts (de Brito and Scorrano, 2008) that knocking out Mfn2 results in excessive cytosolic Ca 2+ and defective 14 mitochondrial calcium uptake. Intriguingly, chronic blockade of mitochondrial calcium import by depleting the mitochondrial calcium uniporter resulted in the reduction of the ER and cytosolic Ca 2+ pools and a migration defect (Prudent et al., 2016). Although the phenotype is similar with MFN2 depletion, the underlying mechanisms is possibly different. Whereas MFN2 reduces the cytosolic Ca 2+ levels after the activation of the chemokine receptor, MCU is required for maintaining the Ca 2+ store and elevates the Ca 2+ , suggesting a requirement of delicate and precise calcium signaling in orchestrating neutrophil migration. Although cytosolic Ca 2+ triggers the activation of Rac in slow moving cells (Price et al., 2003), previous work in neutrophils suggests that Rac activation was independent of cytosolic Ca 2+ (Geijsen et al., 1999). This discrepancy could be explained with the differences in assay conditions (suspension vs. adhesion) or how Ca 2+ levels are manipulated in experiments (elevation vs. reduction). Further work will be required to determine whether/how elevated calcium regulates Rac activation in neutrophils.
In summary, combining evidence from different models, we have discovered an essential role Mfn2 plays in neutrophil adhesion and migration, and determined the downstream mechanism, which provides insights and potential therapeutic strategies for inflammatory diseases and mitochondrial diseases.

15
The zebrafish and mice experiments were conducted in accordance to the internationally accepted standards. The Animal Care and Use Protocols were approved by The Purdue Animal To generate transgenic zebrafish lines, plasmids with the tol2 backbone were coinjected with Tol2 transposase mRNA into embryos of the AB strain at one-cell stage as described (Zhou et al., 2018a). Constructs for neutrophil-specific knockout in zebrafish were generated as described (Zhou et al., 2018a)  All mice used in this study were purchased from Jackson Laboratories. Conditional Mfn2 knockout mice (B6.129(Cg)-Mfn2 tm3Dcc /J) were crossed to S100A8-Cre (B6.Cg-Tg(S100A8-cre, -EGFP)1Ilw/J) transgenic mice to obtain a homozygous floxed Mfn2 alleles with or without the Cre. All mice were used at age 6-12 weeks, and both male and female were used for experiments.

Microinjection
Microinjections of fish embryos were performed as described (Deng et al., 2011). Briefly, 1 nl of mixture containing 25 ng/µl plasmid and 35 ng/µl Tol2 transposase mRNA was injected into the cytoplasm of embryos at one-cell stage.

Tailfin wounding and Sudan black staining.
Tailfin wounding and Sudan Black staining were carried out with 3 dpf embryos as described (Zhou et al., 2018b). Briefly, embryos were fixed in 4% paraformaldehyde in phosphate-buffered saline overnight at 4°C and stained with Sudan black.

Live imaging
Time-lapse images for zebrafish circulation, LTB 4 bath, flow adhesion assay were obtained with AXIO Zoom V16 microscope (Zeiss). Time-lapse fluorescence images for zebrafish neutrophil motility were acquired using a laser scanning confocal microscope (LSM 710, Zeiss) with a 1.0/20 x objective lens at 1 min interval of 30 min. Neutrophils were tracked using ImageJ with MTrackJ plugin and the velocity was plotted in Prism 6.0 (GraphPad). Time-lapse fluorescence images for dHL-60 migration were acquired using a laser scanning confocal microscope (LSM 710, Zeiss) with a 1.0/20 x objective lens at 10 sec interval for 5 min. Cells were stained with 1 µM ER-tracker (Invitrogen E34251) and 20 nM TMRM (Invitrogen T668) for 20 min, washed twice with HBSS and added to fibrinogen coated wells. After 30 min, cells were treated with 1 nM fMLP to induce chemokinesis.

Confocal imaging
For confocal imaging, images were obtained using a laser-scanning confocal microscope (LSM 800, Zeiss) with a 1.4/63x oil immersion objective lens. Images were analysis with ImageJ.
For fluorescence intensity measurement, images within an experiment were acquired using identical camera settings and background was subtracted using ImageJ with the rolling ball radius of 50. Mean fluorescence intensity of selected areas was measured by Measurement in ImageJ and plotted in Prism (GraphPad). Colocalization was using ImageJ Plugin Coloc 2.
Interaction between channels was quantified by Manders' colocalization coefficient as described (de Brito and Scorrano, 2008).

Flow adhesion
Neutrophil flow adhesion assay was performed as described (Zhou et al., 2014). Briefly, 5x10 5 HUVEC cells in 2 ml were plated onto 10 µg/ml fibrinogen-coated 35 mm plate (Corning 430165), and incubated at 37 o C. Then the HUVEC monolayer was primed by 20 ng/ml human TNF-a (Life technologies PHC3015) for 4-6 h. dHL-60 cells were harvested and resuspended at a cell density of 5x10 5 cells/ml in complete medium. dHL-60 cells were flowed on top of HUVEC monolayer at a speed of 350 µl/min using a syringe pump. Cells adhering to the monolayer were recorded using AXIO Zoom V16 microscope (Zeiss) with camera streaming for 5 min. The total number of adherent neutrophils were quantified at 5 min.

Bone Marrow Neutrophil isolation
Femurs and Tibias from mice 8-12 weeks of age were isolated and whole bone marrow was isolated, and passed through a 70 µm filter followed by RBC lysis (Qiagen 158904). Bone marrow neutrophils were isolated using a negative selection column (MACS 130-097-658).
Neutrophils viability was determined by trypan blue staining showing >99% viability.

Peritonitis model
1ml of 4% thioglycollate (Sigma B2551) was injected directly into the peritoneal cavity of mice 6-8 weeks of age. After 3 hours of incubation, peritoneal ascites were collected by introducing 8 ml of PBS into the cavity and collecting the ascites immediately afterwards. Cells were subjected to RBC lysis and viability was determined using trypan blue staining. Cells were stained with CD11b (BD 557686) and Ly6G (BD 566453) on ice for 30 minutes and washed 3 times with staining buffer. Cells profiles were collected with a BD fortessa analyzer and analyzed with Beckman kaluza software. Neutrophil population was defined as FCS/SSC high and CD11b+Ly6G high . Percentage of neutrophils in the lavage relative to total viable cells in each experiment was normalized to the sex-matched littermate control.
Immunostaining dHL-60 cells were resuspended in mHBSS and attached to fibrinogen-coated slides for 30 min.
Then cells were stimulated with 100 nM fMLP for 3 min and fixed with 3% paraformaldehyde in PBS for 15 min at 37 o C. The immunostaining of fixed cells were performed as described (Fayngerts et al., 2017). Briefly, after fixation, cells were permeabilized in PBS with 0.1% Triton X-100 and 3% BSA for 1 h at room temperature. dHL-60 cells were incubated with phalloidin -  interval of 10s and followed by fMLP injection and image recording for another 2 min with 1s interval. The fluorescence intensity of basal Ca 2+ was normalized to 0.

Mitochondrial membrane potential, ROS, and ATP measurement
Mitochondrial membrane potential was measured using TMRM.

Mutational Efficiency Quantification
The mutation efficiency of neutrophil-specific knockout in zebrafish was quantified as described (Zhou et al., 2018a). To determine the mutation efficiency in Tg(lyzC:Cas9-mfn2 sgRNAs),

Statistical analysis
Statistical analysis was performed with Prism 6 (GraphPad). Two-tailed student's t test, or ANOVA was used to determine the statistical significance of differences between groups. A P value of less than 0.05 indicated in the figures by asterisks was considered as statistically significant.