Cell motility is required for diverse processes during immunity and inflammation. Classically, leukocyte motility is defined as an amoeboid type of migration, however some leukocytes, like macrophages, also employ a more mesenchymal mode of migration. Here, we sought to characterize the mechanisms that regulate neutrophil and macrophage migration in vivo by using real-time imaging of leukocyte motility within interstitial tissues in zebrafish larvae. Neutrophils displayed a rounded morphology and rapid protease-independent motility, lacked defined paxillin puncta, and had persistent rearward polarization of stable F-actin and the microtubule network. By contrast, macrophages displayed an elongated morphology with reduced speed and increased directional persistence and formed paxillin-containing puncta but had a less-defined polarization of the microtubule and actin networks. We also observed differential effects of protease inhibition, microtubule disruption and ROCK inhibition on the efficiency of neutrophil and macrophage motility. Taken together, our findings suggest that larval zebrafish neutrophils and macrophage display distinct modes of migration within interstitial tissues in vivo.
Macrophages and neutrophils are key components of the innate immune system of vertebrates, and are among the first responders to tissue injury and infection (Wittamer et al., 2011; Tauzin et al., 2014). These cells need to migrate efficiently through complex and changing in vivo environments, both basally while patrolling and in a directional manner in response to damage cues (Lämmermann and Germain, 2014). Classically, leukocytes are thought to migrate through interstitial tissues at fast speeds in a largely adhesion-independent manner. This type of migration is termed amoeboid motility, and neutrophils provide a classic example of amoeboid cells (Lämmermann et al., 2008, 2013; Jacobelli et al., 2010). Mesenchymal migration, as characterized in cancer cells and fibroblasts, is slower, generates stronger adhesions and often requires degradation of the extracellular matrix (ECM) (Sahai and Marshall, 2003; Friedl and Wolf, 2010). Macrophages have the capacity to modify their migration mode in three-dimensional (3D) in vitro models, switching between amoeboid and mesenchymal modes of migration depending on the matrix architecture (Van Goethem et al., 2010). However, the mode that macrophages utilize during interstitial migration in vivo remains unclear. Here, we took advantage of transparent zebrafish larvae to image neutrophil and macrophage migration and identified distinct modes of migration, with macrophages exhibiting a more mesenchymal mode of migration within interstitial tissues.
RESULTS AND DISCUSSION
Macrophage and neutrophil morphology and motility are different in vivo
Previous work has suggested that mesenchymal and amoeboid cells have distinct morphologies and motility patterns (Van Goethem et al., 2010; Friedl and Wolf, 2010; Jones et al., 2015). To study myeloid cell migration in vivo, we used double-transgenic zebrafish larvae Tg(mpeg1:EGFP)×Tg(mpx:mCherry), to label macrophages and neutrophils, respectively. Macrophages patrolled throughout the body, whereas neutrophils were stationary in the caudal hematopoietic tissue, but were motile in the head region in larvae at 3 days post fertilization (dpf) (Ellett et al., 2011; Mathias et al., 2009; Deng et al., 2011). Imaging of superficial tissues behind the otic vesicle in the head region of larvae allowed for the simultaneous analysis of random neutrophil and macrophage migration within interstitial tissues (Fig. 1A; Movie 1). We observed distinct morphologies of neutrophils and macrophages during interstitial random migration (Fig. 1B). To quantify cell morphology, we measured cell perimeter and roundness as readouts of size and shape, respectively. Migrating neutrophils are compact cells, represented by a small diameter and increased roundness compared to macrophages (Fig. 1C,D). In contrast, migrating macrophages have a larger size and more variable shape and form multiple protrusions (Fig. 1C,D). Migration dynamics between these cells were also different. Macrophages had a significantly slower speed and increased directionality compared to neutrophils (Fig. 1E,F). The high dispersion of macrophage velocity suggests that distinct populations may employ different modes of migration. Taken together, and in agreement with published findings (Ellett et al., 2011), macrophages migrate in a more mesenchymal mode than neutrophils during interstitial migration in vivo.
Macrophages form paxillin-containing structures and require proteases for motility in vivo
Amoeboid cells are able to migrate in 3D in the absence of integrin-mediated adhesions or proteolysis of the ECM (Lämmermann et al., 2008; Liu et al., 2015; Van Goethem et al., 2010), while cells using a mesenchymal mode of migration generate integrin clustering in focal complexes and show protease-dependent migration (Huttenlocher and Horwitz, 2011; Friedl and Wolf, 2010). Indeed, there is evidence to suggest that macrophages require proteases for efficient migration in zebrafish (Travnickova et al., 2015; Shan et al., 2016). To determine whether macrophages require proteases for random migration in zebrafish, we employed a protease cocktail using published methods (Van Goethem et al., 2010). Our findings demonstrate that macrophages, but not neutrophils, have impaired random migration in the presence of protease inhibition (Fig. 2A). These findings suggest that macrophages display a more mesenchymal mode of migration in vivo.
To analyze adhesion structures in vivo, we transiently expressed paxillin fused to mCherry under cell-specific promoters. Paxillin is a major scaffolding protein found in focal adhesion complexes (Hu et al., 2014; Webb et al., 2004). In neutrophils, paxillin showed a homogeneous distribution throughout the cytoplasm (Fig. 2B; Movie 2). By contrast, during random migration, in macrophages we observed dynamic paxillin-containing puncta, primarily in the mid-front section of the cell body and within protrusions (Fig. 2B; Movie 2). These puncta localized on one side of the cell, as would be the case for adhesion structures (Fig. 2C; Movie 3) (Van Goethem et al., 2010; Abshire et al., 2011). Interestingly, puncta that formed during active protrusion were observed to disperse if the protrusion was retracted (Fig. 2D). These paxillin dynamics are in line with what has been reported for classic mesenchymal cells, like fibroblasts, in vitro (Rid et al., 2005).
Neutrophil and macrophage migration are dependent on Arp2/3
Motile leukocytes in 3D environments generate dynamic actin at the leading edge (Yoo et al., 2012; Vargas et al., 2015). Actin-related protein 2/3 complex (Arp2/3) is a key component at the cell front that regulates actin dynamics during pseudopod-driven migration (Weiner et al., 1999). To assess the role of Arp2/3 in migration, we used the Arp2/3 inhibitor CK-666 (Vargas et al., 2015; Lam et al., 2015; Leithner et al., 2016). In larvae treated with CK-666, neutrophil migration was impaired, with velocity decreased by ∼89%. Macrophage migration was also impaired by the presence of CK-666, with a ∼75% reduction in their mean velocity (Fig. 3A). Both cell types showed altered morphologies; macrophages generated extensions, whereas neutrophils became round and did not generate visible protrusions (Fig. 3B). To make sure that CK-666 was not having toxic effects, we washed out CK-666 and found that neutrophils and macrophages resumed migration (Fig. S2A). During migration to a wound, Arp2/3 inhibition completely blocked neutrophil-directed migration, and macrophage migration was partly affected (Fig. 3C). The observation that macrophages can still move and generate extensions in the presence of Arp2/3 inhibition suggests that other actin nucleators (most likely formins) are able to mediate macrophage motility (Suraneni et al., 2015, 2012; Wu et al., 2012), while neutrophils appear to completely rely on Arp2/3 to generate their pseudopod-based movement in vivo.
Microtubules and PIP3 localize in protrusions in macrophages
Cell polarization is necessary for efficient migration. One of the early signaling intermediates that accumulates at the leading edge is the intracellular second messenger phosphatidylinositol (3,4,5)-trisphosphate (PIP3) (Van Keymeulen et al., 2006; Servant et al., 2000). Ratiometric imaging of PHAKT–GFP, to detect PIP3, and mCherry, showed the localization of PIP3 largely at the leading edge of neutrophils in vivo (Yoo et al., 2010; Norton et al., 2016; Lam et al., 2012) (Fig. S1A; Movie 4). Expressing these same constructs under the mpeg1 promoter allowed us to investigate PIP3 dynamics in macrophages during random migration. Similar to the localization in neutrophils, PHAKT–GFP accumulated at the leading edge of macrophages and also localized in projections at the rearward side of the cell during both protrusion elongation and retraction (Fig. S1A; Movie 4). This resembles what Welf et al. described for randomly migrating fibroblasts in vitro, where PIP3 localized to the leading edge, but also showed that protrusions on other sides of the cell accumulated PIP3 (Welf et al., 2012; Weiger et al., 2009).
We next sought to determine whether microtubules showed different polarization in neutrophils and macrophages in vivo. Microtubules actively polarize during cell migration, and play a role in transporting proteins to the cell edges (Hanania et al., 2012), promoting turnover of adhesion complexes (Ezratty et al., 2005) and regulating the actin cytoskeleton (Chang et al., 2008). Microtubule organization in motile cells is variable, and depending on the cell type can be polarized to the front or rear of the cell (Etienne-Manneville, 2013; Yoo et al., 2012; Xu et al., 2005). To compare microtubule polarity in neutrophils versus macrophages in vivo, we transiently expressed different microtubule-binding fluorescent probes under cell-type specific promoters. The microtubule-associated protein Mapre1a (mpeg1:mapre1a-GFP) localized both toward the front and rear of macrophages (Fig. S1B). To improve signal resolution, we transiently expressed the Ensconsin microtubule-binding domain fluorescent probe (EMTB-3xGFP) in macrophages and neutrophils (von Dassow et al., 2009; Yoo et al., 2012; Faire et al., 1999). Neutrophil microtubules labeled with this probe showed a strong polarity toward the rear of the cell, consistent with previous reports (Yoo et al., 2012; Eddy et al., 2002) (Fig. 4A, Movie 5). The microtubule-organizing center (MTOC) localized in front of the nucleus in both neutrophils and macrophages during random migration (Fig. 4A; Fig. S1B) (Yoo et al., 2012). By contrast, in medaka leukocytes the MTOC can oscillate between the front and rear of the nucleus during directed migration (Crespo et al., 2014). Interestingly, macrophages had fewer microtubules in the rearward portion of the cell. Microtubules extended dynamically into protrusions at the front of macrophages (Fig. 4A; Movie 5). In some cases, we observed microtubule loops at the ends of protrusions when analyzing the EMTB-3xGFP probe in macrophages but not neutrophils (Fig. 4A; Movie 5). Microtubule loops were originally described in neurons (Roos et al., 2000). These findings suggest that the polarity of microtubules is different in neutrophils and macrophages in vivo, with microtubules localized to both the front and rear of macrophages.
Effects of microtubule and ROCK inhibition on neutrophil and macrophage polarization and migration
Since we observed different microtubule distribution in neutrophils and macrophages, we next tested the effects of microtubule disruption on their migration in vivo. Prior studies have shown that microtubule disruption impairs neutrophil-directed migration, but increases their random motility (Yoo et al., 2012; Niggli, 2003; Ganguly et al., 2012). In larvae treated with nocodazole, a microtubule-depolymerizing agent, neutrophils showed a significant increase in random velocity, whereas macrophage velocity was not significantly altered, although the microtubule network was disrupted in both cell types (Fig. 4B; Fig. S1C). Both neutrophils and macrophages showed rounding of the cell body (Fig. 4B; Movie 6), consistent with previous reports (Takesono et al., 2010; Xu et al., 2005; Fonseca et al., 2010). Interestingly, in response to directed cues induced by wounding, a significant percentage of macrophages, but not neutrophils, displayed an elongated morphology in the presence of nocodazole, as assessed through a blinded quantification analysis. Time-lapse microscopy showed that macrophages appeared to have a rear detachment defect upon nocodazole treatment (Fig. S1D; Movie 7, see arrow). We never observed this phenotype in neutrophils, indicating that neutrophils and macrophages likely have different rear detachment mechanisms in vivo.
During migration, neutrophils form a defined rear uropod that has high actomyosin contractility which enables rapid motility (Hind et al., 2016). The uropod can be detected in vivo using the calponin homology domain from utrophin (UtrCH–EGFP), an F-actin probe that binds to stable F-actin (Yoo et al., 2010; Burkel et al., 2007; Riedl et al., 2008). In contrast to neutrophils, which displayed a defined rearward polarity of UtrCH, macrophages had a more dispersed distribution of UtrCH that colocalized with Lifeact throughout the cell body, as quantified by fluorescence intensity measurements (Fig. 4C). Microtubule disruption had different effects on UtrCH localization in these cells. In neutrophils, nocodazole induced a hyperpolarization of stable F-actin to the rear, with concentrated UtrCH in the uropod. By contrast, nocodazole did not induce a defined rearward polarization of UtrCH in macrophages (Fig. 4C). These findings suggest that polarization of F-actin dynamics is differentially regulated in neutrophils and macrophages in vivo.
Previous studies have demonstrated that RhoA and Rho-associated kinase (ROCK) proteins are required for rear retraction during monocyte and neutrophil migration in vitro (Worthylake et al., 2001; Yoo et al., 2012; Alblas et al., 2001). To determine how ROCK inhibition affects neutrophil and macrophage migration in vivo, we inhibited ROCK with Rho Kinase Inhibitor III (Rockout). In larvae treated with Rockout, macrophage velocity was strongly reduced, by ∼75%, whereas neutrophil velocity was only partially reduced, by ∼49% (Fig. 4D; Movie 8). At higher concentrations, both neutrophil and macrophage migration were blocked, suggesting that both cell types are sensitive to ROCK inhibition in vivo (Fig. S2B). Both cell types changed their morphology, but only macrophages showed multiple long projections upon ROCK inhibition (Fig. 4D; Movie 8). Taken together, these findings suggest that both neutrophils and macrophages are dependent on ROCK signaling for migration in zebrafish larvae. This finding is distinct from what was found for human macrophages, where macrophage mesenchymal migration is increased upon ROCK inhibition in vitro (Gui et al., 2014).
Here, we show that larval neutrophils and macrophages employ distinct mechanisms that regulate their motility through interstitial tissues in zebrafish. Neutrophils employ a classic amoeboid mode of migration and display a rounded morphology and a contractile uropod, with highly polarized actin and microtubule networks. Macrophages by contrast, exhibit a more mesenchymal-like mode, with the formation of adhesion puncta, tethering at projections and dependence on proteases. The dependence on both proteases and ROCK signaling, suggests that zebrafish macrophages exhibit characteristics of both amoeboid and mesenchymal modes of migration. Indeed, our findings suggest that not all leukocytes migrate in a classic amoeboid mode in vivo, and that there is a continuum in the types of mechanisms that leukocytes use to maneuver through complex tissues. Finally, analyzing both cell types at the same time enables new ways to study and compare the molecular mechanisms that regulate leukocyte responses and motility under the same spatiotemporal conditions in vivo.
MATERIALS AND METHODS
General zebrafish procedures and drug treatment
All adult and larval zebrafish (Danio rerio) were maintained according to protocols approved by the University of Wisconsin-Madison Institutional Animal Care and Use Committee. Embryos and larvae were maintained in E3 medium with 1% Methylene Blue (Sigma-Aldrich) at 28.5°C. When needed, E3 plus 0.003% 1-phenyl-2-thiourea (PTU; Sigma-Aldrich) was used at 24 hpf to reduce pigmentation of the larvae. For wounding and live imaging assays, larvae at 3 days post fertilization (dpf) were anesthetized with E3 plus 0.2 mg/ml 3-amino benzoic acidethylester (Tricaine; Western Chemical). Tail transections were performed on 35 mm (Falcon) milk-coated dishes, with a scalpel size 10 (Integra). When indicated, larvae were pre-incubated in drug baths 1 h before the experiment. All drugs used were suspended in DMSO or methanol–acetic acid (9:1, as stock solutions), and dissolved in E3 plus 0.2 mg/ml Tricaine. Drug working solutions were: nocodazole, 1 µM (Sigma-Aldrich); CK-666, 200 µM (Sigma-Aldrich); and Rockout, 100 µM or 200 µM (Sigma-Aldrich). The DMSO (Sigma-Aldrich) concentrations in control solutions were 0.01% for nocodazole and 0.2% for CK-666, 0.2% and 0.4% for Rockout. Protease Inhibitor mix contains: (L-3-trans-carboxyoxirane-2-carbonyl)-l-leucine (3-methylbutyl) amide (E64c; Peptide International), aprotinin (Sigma-Aldrich), leupeptin (Sigma-Aldrich), pepstatin A (Sigma-Aldrich) and GM6001 (Calbiochem) (Van Goethem et al., 2010). The concentrations used were: E64c, 100 μM; GM6001, 5 μM; Aprotinin (0.04 TIU/ml), Leupeptin (6 μM), and Pepstatin A (2 μM). DMSO (0.2%) and methanol–acetic acid (9:1; 0.04%; Fisher Scientific) were used as control at the concentration of the protease inhibitors mix in all experiments. For drug washout experiments, larvae were preincubated for 1 h with CK-666, mounted in low-melting-point agarose and imaged for 45 m. Drug washout was performed by using medium with DMSO (0.2%) three times for 30 min each. Larvae were imaged again for 45 m afterwards.
Transient expression and cloning of tol2 constructs
3 nl of solution containing between 10 ng/µl and 12.5 ng/µl of DNA plasmid, 20 ng/µl of transposase mRNA in Danieau buffer [1.7 µM NaCl (Fisher Scientific), 21 mM KCl (Dot Scientific), 12 mM MgSO4·7H2O (Sigma-Aldrich), 18 mM Ca(NO3)2 (Fisher Scientific), 150 mM HEPES pH 7.6 (Fisher Scientific)] was injected into one-cell-stage embryos. Plasmids used were: tol2(mpeg1:paxillin-mCherry) at 12 ng/µl, and tol2(lyz:paxillin-mCherry) at 12 ng/µl; tol2(mpeg1:PHAKT-GFP) at 10 ng/µl, and tol2(mpx:PHAKT-GFP) at 10 ng/µl (Yoo et al., 2010); tol2(mpeg1:mapre1a-GFP) (Distel et al., 2010) at 10 ng/µl; tol2(mpeg1:EMTB-3xGFP) at 12 ng/µl, and tol2(lyz:EMTB-3xGFP) at 12.5 ng/μl (Yoo et al., 2012); and tol2(mpeg1:γtubulin-GFP) at 12.5 ng/µl. The D. rerio paxillin–mCherry construct was constructed as follows. Danio rerio paxillin was cloned from total D. rerio mRNA by reverse transcriptase (M-MLV, Promega) followed by PCR using paxillin-specific primers (forward, 5′-ATGGACGATTTAGATGCTCTTCTCG-3′; reverse, 5′-GCTGAAGAGCTTGACGAAGC-3′). Kozak consensus and linker sequences (LELKLRILQSTVPRARDPPVAT) were fused onto D. rerio paxillin coding linked to D. rerio codon-optimized mCherry by overlap recombination (In-Fusion, Takara Bio). D. rerio codon-optimized mCherry was custom synthesized as a template for the reactions (ThermoFisher Inc). Next, paxilin–mCherry was cloned into the tol2 mpeg1 promoter-containing backbone plasmids by overlap recombination (overlap regions in the primers are indicated by lowercase letters) (In-Fusion, Takara Bio) by using specific primers (forward, 5′-tggtcttgataaagggtaccGGCCACCATGGACGATTTAGATGC-3′; reverse, 5′-tgcaataaacaagttaacTCACTTGTACAGCTCGTCCATCCC-3′), and into the tol2 lyzC promoter-containing backbone plasmid by using specific primers (forward, 5′-atcagcagtgatacaggtaccGGCCACCATGGACGATTTAGATGC-3′; reverse, tgcaataaacaagttaacTCACTTGTACAGCTCGTCCATCCC-3′). Cloning of Mapre1a–GFP into the tol2 mpeg1 backbone plasmid was performed by overlap recombination (In-Fusion, Takara Bio) by using specific primers (forward, 5′-caaaaggatccggcgccACCATGGCTGTGAACGTGTTCTC-3′; reverse, 5′-ctcctcctccgaattcAAACTCTTCTGGTTCACCACCCTC-3′). Cloning of EMTB-3xGFP into the tol2 mpeg1 backbone plasmid was done also by overlap recombination (In-Fusion, Takara Bio) by using specific primers (forward, 5′-gacagcaaaaggatccCGATTCGAATTCGCCACCATGG-3′; reverse, 5′-tctagaatcagtcgacCGACTCACTATAGTTCTAGAGGCTCGAG-3′). PHAKT–GFP and γ-tubulin–GFP sequences were isolated from other plasmids by restriction enzyme digestion, and ligated into the tol2 mpeg1 backbone plasmids.
Zebrafish transgenic lines, fixed tissue and live imaging of larvae
For live imaging we used 2.5–3 dpf larvae from transgenic lines specifically expressing fluorescent proteins in either macrophages (mpeg1 promoter) or neutrophils (mpx and lyz promoters). When indicated, these lines were in-crossed to generate double-transgenic larvae, or used for transient expression experiments. The wild-type AB strain was used as a background line for all experiments. Transgenic lines used were: Tg(mpeg1:mCherry); Tg(mpeg1:EGFP) (Ellett et al., 2011); Tg(lyz:EGFP) (Ellett et al., 2011); Tg(mpx:UtrCH-GFP); Tg(mpx:mCherry) and Tg(mpx:LifeAct-mRuby) (Yoo et al., 2010); Tg(mpeg1:UtrCH-GFP) and Tg(mpeg1:Lifeact-mRuby); and Tg(mpeg1:H2B-mCherry) (Vincent et al., 2016). Larvae were mounted laterally in glass-bottom dishes or two-well u-Slide chambers (IBIDI) using 750 µl of E3 plus Tricaine medium in 1% low-melting-point agarose (Fisher Scientific). Another 750 µl of E3 plus Tricaine (including drugs, when needed) was added over the gelled agarose. All imaging procedures were performed at temperatures between 23 and 27°C. To track macrophage and neutrophil behavior, ∼60 µm depth z-series (starting from the surface) of each larvae were taken every 3 min on a spinning disc confocal microscope. The microscope used was a Zeiss Axio Observer.Z1 (Zeiss) with a 20×/0.8 NA Plan-Apochromat lens (Zeiss), and a 63×/1.3 NA water immersion Plan-Neofluar DIC 1 mm Korr lens (Zeiss) for intracellular probe experiments. The spinning disc module corresponds to a CSU-X1 (Yokogawa), coupled to an EMCCD evolve 512 camera (Photometrics). Multi time-lapse images were acquired using Zen 2 imaging software (Zeiss). Laser power, exposure time and camera gain were defined for each set of experiments. For wound recruitment experiments, macrophages and neutrophils were counted at 2 and 1 hours post wounding (hpw), respectively, in fixed larvae. Larvae were fixed [1.5% formaldehyde (Sigma-Aldrich), 0.1 M PIPES (Sigma-Aldrich), 1 mM MgSO4 (Sigma-Aldrich) and 2 mM EGTA (Sigma-Aldrich)] at 4°C overnight, and washed three times (5 min) with PBS pH 7.4 (Sigma-Aldrich) prior to imaging. Images were taken at a magnification of 112× on a Axio Zoom.V16 stereo zoom microscope (Zeiss) with a Plan-Neofluar Z 1× lens (Zeiss), coupled with an AxioCam MRm camera (Zeiss). This set up was run by Zen 2012 acquisition software (Zeiss). Fluorescence intensity and exposure time were defined for each set of experiments. The wound area was defined as all the fin tissue (excluding notochord tissue) 100 µm from the wound edge.
Multi time-lapse images were acquired and post-processed using Zen 2 imaging software (Zeiss), FIJI, and Imaris imaging analysis software (Bitplane). Morphology analysis was performed in FIJI (Schindelin et al., 2012), briefly, subtracting background of images using sliding paraboloid (1–5 pixel), then thresholding the image. The wand tool was used to define regions of interests (ROIs), followed by using the measure function on the generated ROIs. Measure functions used on FIJI were: perimeter (length of the outside boundary of the selection, calculated on software version 1.51j) and roundness (roundness=4×area/π×major axis2). For the motility measurements shown in Fig. 1E,F, time-lapse images were stacked as maximum intensity projections to generate a 2D traceable view of the 3D data set. Cell tracking was performed and analyzed manually with the MTrackJ plug-in on FIJI, using the brightest centroid feature (25×25 pixel area) aimed at the cell body (Schindelin et al., 2012). For 3D speed measurements shown in Figs 2–4 and Fig. S2, cell tracking analysis was performed with the Imaris imaging analysis software (Bitplane). Leukocytes were tracked by using the spot function, where spot size was defined for each movie within a range of 10–15 µm. Single tracks were curated after software detection. The resulting tracks were filtered in two ways, first by only considering tracks with more than six spots for macrophages and five spots for neutrophils. This filter eliminates short and false tracks. Second, filtering for total displacement length was used to reduce the number of cells without real motility or with a ‘vibrational’ movement. Only macrophages with more than 6 µm, and neutrophils with more than 14 µm of total displacement were analyzed. For the speed measurements shown in Fig. 3A and Fig. S2A, the total displacement length filter was not used. For actin fluorescent probe analysis, a directional rectangular area was created around the cell and the mean fluorescence intensity value at defined distances measured with Zen 2 software (Zeiss). All images were background subtracted. The fluorescence intensity and the length were normalized (as a percentage of the total), and data was plotted on Excel 2016 (Microsoft). A polynomial trend line (order of 6) is included in all data sets.
Statistical analyses were performed using Prism 6 (GraphPad) and R Project for Statistical Computing software (R Development Core Team, 2013). Comparisons between two groups were performed with a unpaired two-tailed t-test. Each result is representative of at least three independent experiments. For tracking analysis, data obtained from MtrackJ and Imaris, in three independent experiments, was pooled and analyzed using a least-squares means analysis on R software as previously published and in consultation with a biostatistician (Rosowski et al., 2016). Results were graphed with Prism 6, where individual data points were displayed and color-coded by replicate. Means±s.d. are included in each graph.
We want to thank Dr Taylor Starnes and Julie Rindy for technical assistance and Jens Eickhoff for advice on statistical analyses. We thank Dr Thomas Hall (Institute for Molecular Bioscience, The University of Queensland, Australia) and Dr Bill Bement (Department of Integrative Biology, University of Wisconsin-Madison, USA) for kindly providing microtubule probes. We thank Dr Emily Rosowski for her helpful suggestions and for critical reading of the manuscript. We would also like to thank Dr William Vincent, Dr Laurel Hind, and Dr Sofia de Oliveira for critical reading of the manuscript.
Conceptualization: F.B., A.H.; Methodology: F.B.; Formal analysis: F.B., A.H.; Investigation: F.B.; Resources: P.L., R.F.; Writing - original draft: F.B., A.H.; Writing - review & editing: P.L., R.F., A.H.; Supervision: A.H.; Project administration: A.H.; Funding acquisition: A.H.
This work was funded by National Institutes of Health (R35 GM1 18027 01 to A.H.). Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.