Cell culture and reagents

TaH12810 cells (line H7, generous gift from Elizabeth Glass, The Roslin Institute, Edinburgh) were established ex vivo from the peripheral blood from Holstein calves previously infected with T. annnulata Hisar sporozoites [22]. The invasive and motile behavior of the TaH12810 cells has been recently described in more detail [10]. TaH12810 and Thei cells [23,24] (generous gift from Gordon Langsley) were cultivated in RPMI 1640 (Lonza) supplemented with 10% foetal calf serum (FCS, Amimed), 10 mM Hepes pH 7.2 (Merck), 2 mM L-glutamine (Gibco), 70 µM β-mercaptoethanol (Merck), and antibiotics (Lonza). Buparvaquone was a gift of Dirk Dobbelaere (Vetsuisse Faculty, Bern). TaH12810 cells expressing EGFP-actin or lifeact-mCherry (LA-mCherry) were generated by transfection with either pEGFP-hbeta-actin (generous gift of D. Gerlich; Institute of Molecular Biotechnology, Vienna) or pLenti-LA-mCherry (generous gift of Olivier Pertz). Plasmids: pEzrin-YFP [25] (generous gift of Miguel Quintavilla), moesin-GFP [26] (generous gift of Francisco Sánchez-Madrid). Chemicals: PP2 and SU6656 (Biaffin GmbH), H-1152 (Alexis Biochemicals, #ALX-270-423), Antibodies: mouse mc anti-MAP4K4 (clone MO7, Abnova), mouse mc anti-phospho Tyrosine (p-Tyr-100, 9411 Cell Signaling Technology), rabbit pc anti-Phospho-Ezrin (Thr567)/Radixin (Thr564)/Moesin (Thr558) (ERM) Antibody (Cell Signaling Technology, #3141). Rabbit pc anti-Ezrin/Radixin/Moesin (Cell Signaling Technology, #3142) and rabbit pc anti-TaSP [27] (generous gift from Jabbar Ahmed); TRITC/ALexa488 phalloidin (Molecular Probes).

IF microscopy

Matrix-embedded cells were fixed in 4% paraformaldehyde solution in PBS for 15 min. Cells were then permeabilized by incubation in 0.5% TritonX-100 for 10 min. Blocking of non-specific epitopes was performed in blocking buffer (10% FBS in PBS, 0.02% sodium azide) for 15 minutes. Primary antibodies were applied at 1:50 to 1:200 dilutions in blocking buffer at 4°C over night. Non-confocal images were acquired in wide-field mode either on a Nikon 80i or on an inverted Nikon Eclipse TE2000-U microscope by using Openlab software. Confocal IF images were acquired on a laser-scanning microscope (Leica SP-2 and SP-5) using Leica software. Image procession and quantification were performed using Imaris, Adobe Photoshop and ImageJ software.

Time-lapse imaging

Time-lapse imaging using video microscopy was performed using a Nikon Eclipse TE2000-U or a Leica LX inverted microscope, both equipped with a temperature- and CO2-controlled chamber. Data acquisition and image processing were performed using NIS software for the Nikon Instrument and Leica application suite (LAS) and ImageJ software for the Leica microscope. DIC and fluorescence images were acquired and assembled in AVI movies by NIS and exported for web in m4v format by QuickTime player.

Matrix embedding

Cells were embedded in matrices in 15 well ibidi slides as described below: Collagen gels were prepared according to [28]. For 100 µl collagen solution, 5 µl sodium bicarbonate (7.5% stock, 50 mM final) was combined with 10 µl 10x PBS. 85 µl bovine collagen I solution (PureCol, 3 mg/ml stock, advanced BioMatrix) was added and carefully mixed on ice. 15 µl cell suspension (1-5 x 10⁵ cells/ml, 1.5-7.5 x 10³ cells/well) were mixed with collagen solution, transferred into the ibidi slides and allowed to polymerize for 30 minutes at 37°C in the incubator. This results in a final collagen concentration of 2.5%. Matrigel: 10 µl cell suspension at 1.5 x 10⁵ cells/ml were mixed with 90 µl growth factor-reduce basement membrane matrix (BD Biosciences), transferred into 15 well ibidi slides and allowed to polymerize for 30 minutes. OQel: 1 vial freeze-dried QGel matrix was resuspended in 500 µl ice-cold matrix resuspension buffer according to the manufacturers (QGel SA, Lausanne) instructions. 5 µl cell suspension at 3 x 10⁵ cells/ml were added to ready-made QGel matrix, transferred into ibidi slides and allowed to polymerize for 30 minutes at 37°C in the incubator.

Rho-GTPase pull-down

Rho, Rac, Cdc42 pull-down assay (Cell Biolabs. Inc.) according to manufacturer instructions. In brief: Cells were lysed in pull-down lysis buffer (25 mM HEPES (pH 7.5), 150 mM NaCl, 1% NP-40, 1 mM EDTA, 2% glycerol). 800 µg protein was used per pull-down. Rho was pulled down using 40 µl slurry Rhotekin RBD agarose (20 µg bound Rhotekin RBD protein), and Rac and Cdc42 were pulled down using 40 µl slurry PAK1 PBD agarose (20 µg bound PAK1 PBD) for 1 h at 4°C. Pull-downs were then separated using 15% SDS-PAGE. Detection antibodies used: mouse mc anti-RhoA (No 240302), mouse mc anti-Rac1 (No 240106), mouse mc anti-Cdc42 (No 240201)

Quantifications, repetitions and statistical analyses

Speed equals distance (d)/time (t) and the speed of single cells were determined as the average d/t_i, where t_i is the interval time. Bleb area was determined by measuring the area in µm² of the bleb using ImageJ software.

Quantitative data for speed and motility mode determination were collected from at least three independent experiments. IFA studies and Rho-GTPase pull downs were done at least twice.

T-tests (unpaired, two-tailed) for statistical analysis were performed using Prism software.

Rounded/amoeboid motility of infected macrophages in collagen and matrigel

T. annulata-infected macrophages migrate efficiently in different 3D environments and they are able to penetrate and migrate in collagen and matrigel matrices. We previously demonstrated that motility of infected cells depends on their capability to polarize [7,29] and to trigger membrane blebs at the leading edge [10]. To better understand morphodynamic characteristics of this rounded/amoeboid cell motility, we visualized and quantified motility of infected cells in collagen and matrigel matrices by time-lapse video microscopy. We first compared speed of TaH12810 cells [10] migrating either in collagen, matrigel or a synthetic matrix based on cross-linked polyethylene glycol (QGel) (Figure 1A). Cells migrated efficiently in both collagen (movie S1) and matrigel (movie S2) while they remained stationary when embedded in QGel matrix (not shown). Interestingly, the comparison of cell speeds in collagen and matrigel revealed no significant differences with an average speed of approximately 0.5 µm/min. Considering the different stiffness and pore sizes of collagen and matrigel, we asked what morphological alterations we could observe at the single cell level in matrigel. Live-cell microscopy revealed two types of motile behavior, which we named tunneling and saltatory. Cells migrating in the tunneling mode, penetrated the matrix rounded with occasionally apparent rear constrictions contrasting the marked central constrictions of cells migrating in the saltatory mode (see below). Tunneling mode cells generated a tunnel-like cavity with a diameter that corresponds to the diameter of the cell (movie S3). In this tunnel, the cells can move at high speed in both directions, simply by switching the polarity axis. Interestingly, degradation and/or engulfment of the matrix occurred mostly on both ends of the tunnel but rarely in an angle off the initial axis of migration. In contrast, cells migrating in the saltatory mode penetrated the matrix in a series of sequential steps that results in a (“saltatory”) forward migration speed that oscillates between stagnation and rapid forward movements (movie S4). These steps are described in more detail below. A significantly larger number of cells (Figure 1B) migrated in the saltatory mode (37.65% ± 5.4), indicating that this form of migration is more favorable when large distances need to be overcome. Consistently, cells migrating in the tunneling mode (13.94% ± 2.89) tended to switch to the saltatory mode (movie S2). We measured the speed (distance/time) of cells migrating in these two different modes and found that cells migrated at maximal speeds when inside a tunnel (from one end to the other) whereby they reached approximately 9 µm/minute (see movie S3). Maximal speeds in the saltatory mode were approximately 3-fold lower with maxima around 3.5 µm/min. Speed diagrams shown in Figure 1C visualize cell displacements in µm from one interval to the next over a time period of 15 hours. The frequency of the oscillation displays the frequency of speed changes over time. In the tunneling mode, the speed is close to zero µm/m (valleys), when the cells hit one end of the tunnel and maximal (peaks) when migrating trough the tunnel. These data show that TaH12810 cells can migrate efficiently in both collagen and matrigel matrices by adopting either a tunneling or a saltatory mode of migration. The cells can migrate at very high speeds in pre-formed matrigel tunnels. However, saltatory migration is more efficient because it results in higher velocity (displacement from origin/time, see dot plots in Figure 1C) and might be used by cells for directional and chemotactic movements. The saltatory mode implies cellular compliance and the capability to expand membranous protrusions into the matrix in order to extend preexisting cavities.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Modes of motility of

T. annulata-infected macrophages embed in matrigel.

(A) Motility of cells embedded in collagen, matrigel or Qgel matrices was recorded for 15 h by live-video microscopy. Vertical scatter dot plots show speed calculated as µm/min (collagen: 55 cells, matrigel: 19 cells, QGel: 10 cells). T-test, *** p < 0.0001. B) Infected cells adopt different modes of motility in matrigel. These include forward movement by tunneling, which requires matrix degradation/engulfment at the leading edge (movie S3 and S5). Alternatively, cells can move in a saltatory mode, where cells squeeze through narrow pores and forward movement oscillates between rapid movements and phases of stagnation (movie S4 and S6). Vertical scatter dot plots show percentage of cells migrating in tunneling (14%) or saltatory (38%) mode. 48% of the cells could not be allocated to one of the two modes (immobile). 147 cells total, T-test, ** p = 0.0014. C) X/Y plots show representative paths of cells migrating in the tunneling (left) or in the saltatory mode (right). Dot plots show representative track lengths of tunneling and saltatory migration in µm. Speed diagrams compare fluctuations in speed development of tunneling and saltatory movements; peak = maximum speed, valley = minimal speed or stagnation.

https://doi.org/10.1371/journal.pone.0075577.g001

Actin cytoskeleton dynamics are increased in leading edge protrusions of cells penetrating matrigel

Closer inspection of migrating cells revealed that cells extended protrusions at the leading edge in the direction of migration and generated pores with a diameter corresponding to approximately 1/4 cell diameter (Figure 2A, movies S6, S7 & S14). Transmigration of the cell body through these pores appeared to be restricted by the compressibility of the nucleus (Figure 2B), which sets the limit for circumferential compression of a cell [30]. In the tunneling mode, cells extended filopodia-like protrusions at the leading edge, while the formation of membrane blebs appeared markedly reduced (Figure S1 and movie S5). During transmigration in the saltatory mode, the cells assumed an hourglass shape, which was hallmarked by a neck zone of maximal compression (Figure 2B, arrowheads). 3D reconstruction of multiple confocal sections revealed that the zone of maximal compression was reinforced inside the cell by an F-actin-rich ring (Figure 2C, arrow). The protrusions at the leading edge were rich in F-actin as well, suggesting that de-novo actin polymerization is required for forward extension of the cell. Indeed, using cells expressing lifeact-GFP, we found that F-actin polymerization was predominately active at the leading edge of invading cells and particularly in protrusions penetrating the matrigel (Figure 2B and movie S8), while polymerization activity was less abundant in the tail. Zones particular rich in F-actin were the submembranous skeletons in the neck zone and near regions at the leading edge where membranous protrusions emerge. To determine what the function of F-actin polymerization at the leading edge might be, we visualized the initial steps of protrusion formation in cells expressing lifeact-GFP. We found that filamentous or sheet-like, actin-rich assemblies resembling lamellipodia of matrix-invading epithelial cell [31], initiated the protrusions. However, instead of assuming spindle shaped morphology like these cells, parasitized macrophages remained rounded and progressively expanded the initial protrusions until large enough to accommodate the nucleus (Figure 2D). Furthermore, we observed accumulation of F-actin at the neck zone and at the leading edge, and overall a markedly increased content of F-actin in the protrusion that clears the way for the cell. Taken together, massive F-actin polymerization at the leading edge follows initial polarization of the cells and is likely needed for invasive migration of these cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Actin polymerization at invasion front of cells migrating in saltatory mode.

A) TaH12810 cells were embedded in matrigel and migration was monitored by time-lapse imaging. Green dots indicate circumference of cavities in migration path, red lines indicate position of pores between holes. B) Time-lapse microscopy of EGFP-actin expressing TaH12810 cell invading matrigel (movies S7 and S8). Top: grey scale images, bottom: EGFP-actin fluorescence. C) 3D reconstruction of confocal microscopy sections of TaH12810 cell migrating in matrigel. Actin cytoskeleton stained with TRITC-phalloidin (red). Arrow indicates actin–rich circle in maximal constriction zone at pore after nuclear translocation (inset shows F-actin in gray scale, 2x magnification). D) As in B but focus on initial phase of matrigel invasion. EGFP-actin fluorescence intensity was measured along line indicated and plotted against line length in µm. Arrows indicate peak fluorescence at cell cortex where protrusion emerges, arrowheads F-actin-rich ring at circular constriction zone, asterisks leading edge F-actin.

https://doi.org/10.1371/journal.pone.0075577.g002

ERM protein localization to leading edge, neck and tail of the cells

We have recently described that ERM proteins accumulate in lamellipodia of infected cells migrating on 2D substrata [7]. To determine whether ERM proteins localized to leading edge structures in cells migrating inside 3D matrices as well, we visualized ERM protein distribution at different stages of matrix invasion by immunofluorescence microscopy (Figure 3). We localized the leading edge of migrating Theileria-infected macrophages by two ways. First, we localized the parasite by nuclear staining with Hoechst, which visualized both host cell and parasite nuclei. In polarized, migrating cells, the parasite is always located between the host cell nucleus and the trailing edge of the cell (Figure 3A, arrows). Parasite localization with respect to the host cell nucleus can thus be used as a directionality indicator to determine the leading edge. Second, we found that the protein cortactin (Figure S2A and B) and tyrosine phosphorylated proteins (Figure S2C) accumulated at the invasion front in migrating cells. Cortactin is a well-established organizer of the cortical actin cytoskeleton at the leading edge to promote motility and invasiveness of cancer cells [32]. Therefore, we used anti-cortactin immunostaining as leading edge indicator and asked as to whether ERM proteins co-distributed with cortactin. We found that cortactin and ERM protein detection overlapped in invading cells and that co-distribution of the two proteins was particularly evident in the neck zone, where ERM protein accumulation remained prominent throughout most of the penetration process. Once nuclear translocation trough the pore was accomplished, ERM proteins were detected at the rear of migrating cells, suggesting a function of ERM proteins for tail compression and rear retraction. To determine ERM protein distribution in living cells, we expressed YFP-fused ezrin and monitored YFP-ezrin dynamics in cells migrating through matrigel by time-lapse imaging. We detected YFP-ezrin at the plasma membrane near the leading edge and highly enriched in the tail (Figure 3B arrow). After nuclear translocation, YFP-ezrin began to accumulate at the invading leading edge again (Figure 3B arrowhead), which suggested a function of ERM proteins both at leading and trailing edges.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. ERM proteins localize sequentially to leading edge, the neck zone and the trailing edge.

A) If microscopy analysis of ERM protein localization in cells migrating in matrigel using anti-ERM antibodies (red). Leading edge of cells was determined using cortactin localization (green). The location of the trailing edge was confirmed by visualizing host and parasite nuclei (arrows) using hoechst stain. Arrowheads indicate neck zones. B) Quantification of YFP-ezrin fluorescence intensity (leading to trailing edge). Top row shows still images of YFP-fluorescence of time lapsed-image acquisition for a period of 9 min. Bottom row are the corresponding gray-scale images. Middle row shows intensity histograms of YFP-fluorescence along the white line (top row left). Arrows indicate tail, arrowhead leading edge.

https://doi.org/10.1371/journal.pone.0075577.g003

Host cell MAP4K4 and ERM proteins co-distribute to leading and trailing cell compartments

ERM protein activity and function are regulated by C-terminal phosphorylation. One C-terminal regulatory kinases is MAP4K4 (mitogen-activated protein kinase kinase kinase kinase 4), a serine/threonine kinase, which phosphorylates ERM proteins for lamellipodia formation [33]. MAP4K4 is an essential kinase for macrophage function in the context of inflammatory signaling induced by TNF-alpha [34], a parasite-induced cytokine expressed in T. annulata-infected macrophages [22]. If MAP4K4 would be involved in the regulation of ERM function, we would expect it to be localized near leading and trailing edges of invading cells. By IF microscopy, we detected MAP4K4 at the leading edge of matrigel invading cells with apparent polarization (Figure 4A); when differential interference contrast (DIC) images in grey-scale were overlaid with RGB images generated with anti-MAP4K4 staining (Figure 4A magnifications), we observed preferential accumulation of MAP4K4 inside membrane blebs. Interestingly, anti-ERM and anti-MAP4K4 staining did not accurately co-localize; instead, ERM proteins localized predominately to the neck zone and cortex of membrane blebs (Figure 4A, magnification). We confirmed the localization pattern of MAP4K4 and ERM proteins by ectopically expressing MAP4K4 (GFP-MAP4K4) and Ezrin (ezrin-YFP) as fusions to GFP and YFP, respectively. We visualized dynamic alterations in subcellular distribution of the two proteins by fluorescent live-cell imaging in TaH12810 cells migrating inside matrigel. Analogous to the IF data, we detected ectopically expressed MAP4K4 predominately at the leading edge of migrating cells (Figure 4B, movies S9 & S10). However, we also observed GFP fluorescence in the trailing end of the cells (Figure 4B, lower), which suggested that MAP4K4 could be distributed to the rear and needed there as well. We observed strong ezrin-YFP fluorescence in trailing edge membranes and somewhat weaker in protrusions at the leading edge (Figure 4B, lower, movie S13), where F-actin dynamics are particularly high (Figure 4B, lower, movie S12; see movie S11 for grey scale live-cell imaging of same cell). Furthermore, ezrin-YFP accumulated in the neck zone during matrix penetration, where it remained throughout the penetration process. Combined, these observations indicate that spatially restricted accumulation of ERM proteins and of their regulator MAP4K4 could contribute to leading edge actin and membrane dynamics in T. annulata-infected cells.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. The ERM kinase MAP4K4 localizes to leading and trailing edges of migrating TaH12810 cells.

A) IF microscopy analysis of ERM and MAP4K4 protein localization in cells migrating in matrigel using anti-ERM and anti-MAP4K4 antibodies. Magnifications are 4x of boxed area and highlight MAP4K4 accumulation in membrane blebs. Arrows indicate direction of migration. B) Time-lapse microscopy of TaH12810 cells expressing either YFP-ezrin (movies S11 & S13) or EGFP-MAP4K4 (movies S9 & S10). LA-mCherry was used to visualize leading edge actin dynamics (movie S12). Cells migrate in matrigel either in saltatory (upper) or tunneling (lower) mode (from left to right: grey scale, GFP or YFP fluorescence, overlay). Dotted line indicates cavity boundary, stars indicate leading edge of cells.

https://doi.org/10.1371/journal.pone.0075577.g004

Filopodia-like protrusions lead the way while membrane blebbing expands the matrix for invading cells

To further understand dynamic alterations at the leading edge during matrigel invasion, we monitored the invasion process at high spatial and temporal resolution. Figure 5A shows how a cell migrates in saltatory mode, thereby penetrating the matrix and translocating the nucleus in a stepwise process. Figure 5B focuses on the invasion zone at the very leading edge of the same cell and shows the dynamic morphological alterations in that zone. We observed that membrane blebs expand along existing filopodia-like protrusions, whereby expansion began at the base of the protrusion and extended outward. We quantified the expansion and retraction processes by monitoring and measuring the size of a growing bleb every 15 seconds for 3 minutes (Figure 5B and C). Unlike membrane blebs of TaH12810 cells embedded in collagen (movie S1), which showed rapid bleb expansion and slow retraction [10], expansion and retraction dynamics of leading edge blebs in cells embedded in matrigel were comparable (Figure 5C). Confocal microscopy analysis of the blebbing protrusions showed that they consisted of rounded expansions that were lined with F-actin and decorated with multiple filopodia (Figure 5D). We have previously shown that Src kinase activity is needed for the polarization of Theileria-infected macrophage and herein we detected increased protein tyrosine phosphorylation at the leading edge of invading cells (Figure S2). To determine whether Src kinase activity could be a determinant for TaH12810 morphology during invasion, we treated infected cells embedded in matrigel with the Src kinase inhibitor Su6656 (not shown) or PP2 (Figure 5E). Both inhibitors reduced asymmetric, cortical F-actin accumulation and the formation of invasion structures compared to control cells. In contrast, cell treated with the Rho kinase inhibitor H-1152 displayed polarized morphology and massive protrusions (Figure 5E). However, analogous to what we observed in collagen-embedded TaH12810 cells [10], these protrusions appeared disorganized and the nucleus failed to transmigrate. 5F schematically shows dynamic morphological alterations during matrigel invasion, bleb formation, matrix expansion and retraction of the trailing edge.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Membrane blebs evolve along filopodia-like protrusions at leading edge.

A) Still images of DIC live-cell microscopy (movie S14). Arrowheads indicate filopodia-like protrusions, arrow neck zone of maximal compression. B) Higher magnification of leading edge of cell shown in A. Arrowhead highlights blebbing protrusion at maximal expansion. C) Quantification of surface area of blebbing protrusion shown in B. D) Confocal microscopy image of TaH128101 cells invading matrigel. Green: F-actin, red: parasite surface (anti-TaSP), blue: DNA. Insets show leading F-actin and host cell nuclear DNA. E) as D but cells were incubated in the presence of either the Src kinase inhibitor PP2 (5 µM) or the Rho kinase inhibitor H-1152 (5 µM). F) model of F-actin distribution and leading edge membrane dynamics of TaH12810 cells invading matrigel.

https://doi.org/10.1371/journal.pone.0075577.g005