Ethics Statement

All animal studies were carried out in compliance with the guidelines of the Institutional Animal Care and Use Committee (IACUC) of the University of Pennsylvania and in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The animal protocol was approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Pennsylvania, Philadelphia.

Parasite culture

Wild-type (RH) strain, RHΔphil1, RHΔphil1/TgPHIL1-C5 (Comp) and RH-OVA-tdTomato T. gondii tachyzoites [22], [23] were maintained by serial passage in confluent primary human foreskin fibroblast (HFF) (ATCC CRL-1634) monolayers in Dulbecco's Modified Eagle's Medium (DMEM), supplemented with 10% (v/v) heat-inactivated fetal bovine serum (FBS) and 10 mM HEPES pH 7.0, as previously described [24]. The medium was changed to DMEM supplemented with 1% (v/v) heat-inactivated FBS and 10 mM HEPES pH 7.0 just prior to infecting the confluent monolayers with parasites.

ex vivo imaging

C57BL/6 mice (Jackson Laboratories, Bar Harbor, ME) were anaesthetized by intraperitoneal injection of Ketamine/Xylazine. Ear hair was removed with a hair removal solution (Nair, Princeton, NJ). 1–2×10⁵ freshly egressed RH-OVA-tdTomato tachyzoites in 1 µL PBS were injected intradermally using a 26-gauge Hamilton syringe. Five min post injection the mouse was euthanized in a CO₂ chamber and the ear was removed and placed in a temperature-controlled imaging chamber (Warner Instruments, Hamden, CT), and embedded in 1.2% (w/v) low-melting agarose (Sigma-Aldrich, St. Louis, MO). Tissue was perfused with RPMI complete (RPMI plus 10% FBS, 1% penicillin/streptomycin, 1% glutamine, 1% HEPES, 1% nonessential amino acids, and 0.1% beta-mercaptoethanol), perfused with 95% oxygen and 5% carbon dioxide, and maintained at 37°C. ex vivo imaging was performed using a Leica SP5 two-photon microscope equipped with a picosecond laser (Coherent Chameleon; 720 nm–1020 nm) and tunable external detectors that allow simultaneous detection of emissions of different wavelengths and second harmonic signals (SHG). tdTomato was excited using a laser light of 900 nm. The LAS-AF software (Leica) was used for image acquisition.

Pitta chamber construction and parasite preparation for 3D motility assays

To construct imaging (“Pitta”) chambers, washed 22×22 mm borosilicate glass coverslips were adhered to washed 24×50 mm borosilicate glass coverslips with strips of double-sided tape (Scotch 3M, St. Paul, MN) spaced 3 mm apart, for an internal volume of ∼10 µL.

Parasites were harvested by syringe release of infected HFF monolayers through a 27-gauge needle, and then filtered through a 3 µm Nuclepore filter (Whatman, Piscataway, NJ), centrifuged at 1,000× g for 4 min, washed and resuspended at a concentration of 1–2×10⁸ parasites/mL in 3D Motility Media (1× Minimum Essential Medium lacking sodium bicarbonate, 1% (v/v) FBS, 10 mM HEPES pH 7.0 and 10 mM GlutaMAX L-alanyl-L-glutamine dipeptide) supplemented with 0.3 mg/mL Hoechst 33342 (H33342). Matrigel (BD Biosciences, San Jose, CA) was kept on ice to prevent polymerization prior to mixing with the parasite suspension and 3D Motility Media in a ratio of 3∶1∶3 by volume, respectively. Pitta chambers were perfused with ∼10 µL of this parasite/Matrigel suspension (final H33342 concentration ∼43 µg/mL) and incubated at 27°C for 7 min on a ThermoPlate (Tokai Hit, Shizuoka-ken, Japan). The chambers were then placed in a NanoScanZ piezo Z stage insert (Prior Scientific, Rockland, MA) in a custom-built, preheated microscope enclosure (UVM Instrumentation and Model Facility, Burlington, VT), and incubated at 35±0.2°C for 2 min to allow for temperature equilibration and completion of Matrigel polymerization prior to data acquisition. The room housing the microscope was also heated to the same temperature (35±1°C) to maintain temperature equilibration and minimize thermal convection currents within the microscope enclosure. The kinetics of Matrigel polymerization varied slightly depending on the reagent lot number, but were generally consistent under these empirically optimized conditions.

For live trail deposition in Matrigel, tachyzoites were harvested as described above, and incubated at 25°C for 15 min in 3D Motility Media with a 1∶25 (v/v) dilution of anti-TgSAG1 antibody (Argene, Shirley, NY) directly conjugated to Alexa Fluor 488 (Invitrogen) as previously described [9]. Labelled parasites were washed once with 3D Motility Media, then mixed with 3D Motility Media and chilled Matrigel, perfused into Pitta chambers and processed as for H33342-labelled samples.

Heat-killed tachyzoite preparations were prepared by incubating at 56°C for 30 min as described previously [25]. Cytochalasin D was dissolved to a concentration of 1 mM in high quality DMSO and stored in the dark at −20°C. This stock solution was diluted to a final compound concentration of 0.2 µM in 3D Motility Media supplemented with H33342 for parasite treatment. The suspension was then incubated at 37°C for 15 min to inhibit parasite motility as described previously [26], [27], prior to mixing with Matrigel and imaging.

Imaging and data acquisition

Fluorescent parasite nuclei were imaged using a 20× PlanApo λ objective (NA = 0.75) on a preheated Nikon Eclipse TE300 epifluorescence microscope with neutral density (ND) 4, ND 8 (Nikon Instruments, Melville, NY) and ND OD = 1.0 filters (Thorlabs, Newton, NJ) installed in tandem. Reference slides containing images with a defined chirality were used to confirm the correct orientation of the microscope and interpretation of trajectory handedness with the analysis software.

Time-lapse stacks were captured using an iXon 885 EMCCD camera (Andor Technology, Belfast, Ireland) cooled to −70°C and driven by NIS Elements v. 3.20 software (Nikon Instruments, Melville, NY). The camera was set to frame transfer sensor mode, with a vertical pixel shift speed of 1.0 µs, vertical clock voltage amplitude of +1, readout speed of 35 MHz, conversion gain of 3.8×, EM gain setting of 3 and 2×2 binning. The z-slices were imaged with an exposure time of 16 ms; stacks consisted of 41 z-slices spaced 1 µm apart for a total x, y, z, t imaging volume of 402 µm×401 µm×40 µm for 67 stacks in 60 s. All experiments were completed within 80 min of harvesting the parasites, and control (i.e., untreated) samples were assayed both at the very beginning and end of each experiment to ensure imaging and parasite conditions remained constant. To observe live trail deposition, tachyzoites labelled with Alexa Fluor 488-conjugated anti-TgSAG1 were imaged as described above but with a single ND filter (ND 8) installed, and with a camera EM gain setting of 10.

Tracking

Datasets were exported from NIS Elements and read directly in Imaris ×64 v. 7.6.1 (Bitplane AG, Zurich, Switzerland). Using the ImarisTrack module, parasites were tracked within a region of interest (ROI) that was cropped 1 µm from each of the original x, y and z dimensions to avoid tracking artifacts associated with objects close to the border. Background object subtraction and elliptic spot detection in the z-dimension were enabled, with the estimated x, y diameters set to 4.0 µm and the estimated z diameter set to 8.0 µm. A filter was applied to the datasets for tracks that came within 0.4 µm of a ROI border, as well as a filter to exclude all tracks with durations of less than ∼2 s, to further avoid tracking artifacts as mentioned above. An autoregressive motion tracking algorithm was applied with a maximum distance of 12.0 µm and a maximum gap size of 1. Translational and rotational drift correction were employed where necessary (e.g., a dataset where stationary parasites all have the same displacement vector) and datasets were manually inspected following drift correction to ensure the algorithm had been executed appropriately. Positional coordinates were then exported in comma-separated values (.csv) format for further processing and analysis using the Bugs software suite.

Bugs software processing and analysis

Subsequent processing and analysis of the acquired trackpoints for each organism were accomplished by a custom software suite we developed specifically for this purpose. A key tenet of the software is that the user be able to graphically and quantitatively verify each step in the process. To that end, the digitized track data were converted to and maintained at all times in a format (PDB, Brookhaven Protein Data Bank format; www.pdb.org) that allows direct visualization and analysis of all the organisms' discrete trackpoints (or smoothed trajectories, described below) with molecular graphics software such as PyMOL [28], [29]. For further quantitative analysis and plotting, derived values such as velocity, curvature and torsion were written to .csv format files for use with spreadsheet or plotting software. The “Bugs” software suite is written as portable Fortran modules linked together by shell scripts, and is available by download by SFTP from http://sourceforge.net/projects/bugs/. For questions about the software, contact mrould@uvm.edu.

Parasites with a total trajectory displacement (simple distance from first to last trackpoint) of greater than 2 µm were considered moving based on analysis of a heat-killed parasite preparation (see Results); these trajectories were subject to Fourier smoothing and calculation of parameters as discussed below. Parasites with a total trajectory displacement of 2 µm or less were considered stationary and excluded from further analysis.

Fourier smoothing

The rather sparse temporal sampling of the position of each organism, coupled with positional error in those coordinates necessitated some form of smoothing and interpolation be applied to the trackpoint coordinates. A simple, modified Fourier fit to each organism's entire set of trackpoints performed the best of several approaches tested. We evaluated these approaches using an “even-odd” cross-validation criterion in which the even-numbered trackpoints were used in the fit, and the cross-validated residual discrepancy was determined between the observed and fit coordinates of the odd-numbered trackpoints (which were not used to determine the parameters of that fit). Similarly, the odd-numbered trackpoints were used in the fit and the fit discrepancy measured with the even-numbered trackpoints. The total cross-validated discrepancy was calculated as the square-root of the sum of these two squared discrepancies. This form of cross-validation mimics the effects of a data acquisition rate that is half of the true rate, and quantifies how well we would be able to predict the “missing” data in that case. The three spatial coordinates of each organism's trajectory were fit independently as a function of time (x(t), y(t), z(t)) using singular value decomposition [30]. The fit parameters were the coefficients of the standard Fourier series, and , where n runs from 0 to the order, m, of the Fourier series, supplemented by an additional linear term, time. The latter extra term allowed better fits using a lower Fourier order (and hence requiring fewer free parameters) by permitting compensation for any difference in the first, x(t₀), and last, x(t_f), values (since the Fourier series is intrinsically periodic). The target function was the sum of the squared discrepancies between the fit values x(t_i) and the observed values X_obs(t_i) of the coordinates. Optimal Fourier order, m, for the fit for each organism's trajectory was determined using even-odd cross-validation. Fourier fits were applied as long as there were a sufficient number of trackpoints (≥12) in a given trajectory to do so. The number of trackpoints was not necessarily correlated to trajectory length; consequently, a subset of trajectories from moving parasites was considered “unfittable” and excluded from subsequent analysis.

Trajectory analysis

Coordinates of the smoothed trajectory at a given time are calculated by evaluation of the Fourier series (supplemented by the linear time term) with coefficients as fit above. A trajectory sampled at regular intervals of arc length is readily calculated by computational integration of arc length over time. Analytically-derived expressions for the first, s′, second, s″, and third, s′″, time derivatives of the 3D Fourier series (including the time term) are used to calculate the instantaneous velocity (v), curvature (κ) and torsion (τ) at any given point along the trajectory via the equations [31]:

The total trajectory length for a given organism is the (computationally) integrated path length (contour length) along the smoothed trajectory from the first to last trackpoint. The total displacement is the simple distance (“as the crow flies”) from the first to last trackpoint for an organism. The last trackpoint is derived from the last z-stack acquired in the 60 s dataset, or the last z-stack acquired before the fluorescent parasite nucleus moves out of the volume being imaged.

Discrete measurements of velocity were calculated as the simple distance between two consecutive trackpoints sampled at regular (0.5 µm) intervals of pathlength along the Fourier-smoothed trajectory, divided by the time interval between those points. The maximum velocity corresponds to the highest value for velocity observed along the Fourier-smoothed trajectory.

Statistical analysis

Values for motility parameters were exported from the Bugs software suite in .csv format, imported into Microsoft Excel for data compilation, and graphed and analyzed using GraphPad Prism v. 6.01 (La Jolla, CA). Parameters calculated from motility assays and graphed as scatter plots were analyzed using paired (RH vs. Δphil1 and Δphil1 vs. Comp) and unpaired (RH vs. Comp) Student's t-test; frequency distributions were analyzed by performing a Kolmogorov-Smirnov test. Statistical significance between samples is indicated with asterisks. Results shown are the mean ± standard deviation of four to five independent experiments, each performed in either triplicate or quadruplicate.

T. gondii move with corkscrew-like trajectories in a three-dimensional environment

To investigate how T. gondii tachyzoites move in animal tissue, we used a model of T. gondii infection where live tachyzoites expressing a tandem Tomato fluorescence reporter (RH-OVA-tdTomato) are injected into the mouse earflap and imaged by two-photon laser scanning microscopy [32]. Unlike the twirling, circular gliding and helical gliding observed in 2D motility assays, tachyzoites in the earflap appeared to undergo a form of corkscrew-like motility (Figures 1A and 1A', and Video S1). This suggested that the standard 2D motility assay may not be a completely accurate reflection of parasite motility within host tissues.

Download:

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

Figure 1. T. gondii moves in a corkscrew-like manner in three dimensions.

(A) RH strain T. gondii tachyzoites expressing a tandem tomato fluorescence cassette (RH-OVA-tdTomato) were injected into a mouse earflap and imaged by two-photon laser scanning microscopy. A maximum intensity projection (MIP) shows parasites that move in a corkscrew-like fashion (e.g., dashed white box). Scale bar = 29 µm. See Video S1 for the corresponding movie. (A') Higher magnification view of dashed white box in (A); scale bar = 12 µm. (B) Assembly and dimensions of the imaging (“Pitta”) chamber. Coverslips were assembled using double-sided tape, and perfused with a 1∶3∶3 mixture of parasites treated with Hoechst 33342, 3D motility media and chilled Matrigel, respectively. (C) Pitta chambers were incubated at 27°C for 7 min, followed by 2 min equilibration in the preheated 35°C microscope enclosure. Fluorescent parasite nuclei were imaged for 60 s in a 402 µm×401 µm×40 µm volume (approximately 67 z-stacks, with 41 z-slices captured every 0.88 s) by time-lapse fluorescence videomicroscopy. Datasets were visualized during acquisition by generating MIPs, tracked using the Imaris software, and then analyzed using the Bugs software as indicated in the workflow. (D) A MIP showing that parasites also move in corkscrew-like trajectories in Matrigel. Scale bar = 50 µm. See Video S3 for the corresponding movie. (D') Higher magnification view of dashed white box in (D); scale bar = 10 µm. The colour scheme for all MIPs was inverted for better visualization of parasite trajectories.

https://doi.org/10.1371/journal.pone.0085763.g001

We sought to study this behaviour with larger numbers of parasites, and in a more defined and controlled environment. To accomplish this, an in vitro assay was developed where imaging (“Pitta”) chambers were perfused with parasites resuspended in Matrigel, a soluble basement membrane preparation that has been widely used as a matrix to study cell migratory behaviour in 3D [33]–[35]. Since the parasites were pre-labelled with Hoechst 33342, a fluorescent nucleic acid stain, the positions of the fluorescent parasite nuclei could be readily imaged in the chamber volume by time-lapse fluorescence videomicroscopy (Figures 1B and 1C). Analysis of the datasets revealed that all of the parasites deposited TgSAG1-containing trails (Video S2) as they moved in left-handed, corkscrew-like trajectories through the Matrigel (see maximum intensity projection (MIP) in Figures 1D and 1D', and Video S3). These data demonstrate that the trajectories of T. gondii tachyzoites are strikingly different in a 3D environment compared to the types of motility observed on a coated glass coverslip, and that it is possible to reconstitute this behaviour in vitro using a Matrigel-based system.

Quantification of parameters of 3D trajectories

To further characterize and quantify the parameters of this form of motility, we developed semi-automated methods for tracking the positional coordinates of the parasites and reconstructing their trajectories for detailed analysis. Two types of “stationary” parasite datasets were generated to estimate an overall error due to Brownian motion and tracking error from the analysis software. Out of several criteria tested, the total trajectory displacement (i.e., the shortest distance from the initial to the final position of the parasite) performed the best in terms of distinguishing between moving and stationary parasites (data not shown). In a heat-killed preparation, 97.6% of the parasite trajectories had a displacement of 2 µm or less (Figures S1A and S1B), with mistracking artifacts (e.g., trajectories connecting two parasites that were clearly stationary) accounting for the remaining 2.4% (data not shown). Similar results were obtained when parasites were treated with 0.2 µM cytochalasin D (Figures S1C and S1D), a concentration previously described to inhibit parasite motility in 2D assays [26], [27].

When the trajectories were smoothed by applying a modified Fourier fit (Figures 1C, 2A, 2B, and S2, S3, S4, S5), it was apparent that the parasite's velocity along the trajectory oscillated between minima and maxima, reflecting bursts of speed as the parasites moved through the Matrigel (Figures 2C and 2D). Furthermore, a complex and periodic relationship between velocity, curvature (i.e., degree to which the trajectory bends within a plane; see Figure S6A and S6B) and torsion (i.e., degree to which the trajectory twists out of the plane; see Figure S7A and S7B) was evident along the trajectory. The trajectories decreased in curvature and torsion as the parasites accelerated (i.e., as their velocity increased), but once the local maximum velocity was reached, gradual deceleration was associated with progressively increasing curvature and torsion (Figures 2C and 2D). When curvature and torsion reached their maxima, the parasite either repeated the cycle, or paused temporarily for a variable period of time marked by low velocity and rapidly fluctuating curvature and torsion. Note that the torsion values are negative to reflect the left-handed nature of the trajectories, such that increasing torsion is depicted as increasingly negative values on the plots [36]. Since a regular helical trajectory would have constant curvature and torsion values along the length of the helix, the parasite-derived trajectories can best be characterized as periodic but irregular helices.

Download:

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

Figure 2. Visualization and analysis of two representative 3D trajectories of parasites in Matrigel.

(A) and (B) Positional coordinates for two representative wild-type (RH) parasite trajectories are visualized as connected, discrete trackpoints (white), and overlaid with the trajectory after smoothing (green). Scale bar = 7 µm. (C) and (D) Plots of velocity (red), curvature (green) and torsion (blue) values along the length of the parasite trajectories shown in panels A and B, respectively. The curvature and torsion values would be constant through time for a regular helix; the variation in these measurements along the parasite trajectories shows that they move in irregular-shaped yet periodically fluctuating corkscrews.

https://doi.org/10.1371/journal.pone.0085763.g002

Δphil1 parasites are less motile in a Matrigel-based environment

To assess how a change in the parasite's crescent morphology could affect motility behaviour, we tested the ability of TgPHIL1 (Photosensitized INA-Labeled protein 1) knockout parasites to move in the 3D motility assay. TgPHIL1 is a cytoskeleton-associated protein, the absence of which results in parasites that are shorter and wider than parental wild-type (RH) parasites. Complementation of the Δphil1 parasites with a copy of TgPHIL1 driven by its endogenous promoter (“Comp”) rescues the morphological defect (Figure S8) [22].

Thousands of trajectories were analyzed (RH n = 6,467, Δphil1 n = 9,305 and Comp n = 3,743), and comparable percentages of parasites moved for all three lines (Figures 3A and 3B; summarized in Table 1). However, the Δphil1 parasites showed significantly shorter trajectory lengths than the RH parasites from which they were derived (Figures 3A, 3C and 3D; Table 1). The Δphil1 parasites also exhibited significantly reduced mean and maximum velocities compared to RH (Figures 3E and 3F; Table 1). These effects were either partially or fully restored in the Comp parasites (Figures 3E and 3F; Table 1), demonstrating that disruption of TgPHIL1, which alters parasite morphology, is associated with a significant impact on T. gondii motility in Matrigel.

Download:

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

Figure 3. 3D motility of the Δphil1 parasites.

(A) MIPs for wild-type (RH), TgPHIL1 knockout (Δphil1) and complemented (Comp) parasites. Scale bar = 50 µm. The colour scheme for all MIPs was inverted for better visualization of parasite trajectories. The percentage of total parasites moving (B) was comparable for the three parasite lines, but the cumulative frequency distribution (C) and histogram (D) of the smoothed trajectory lengths for RH (black), Δphil1 (red) and Comp parasites (grey) reveal that the Δphil1 parasites do not move as far as the RH or Comp parasites within the same timeframe (Kolmogorov-Smirnov test, D = 0.199, p<0.0001 and D = 0.114, p<0.0001, respectively). The Δphil1 parasites also exhibited significantly decreased mean velocity compared to the RH parasites (E) and significantly reduced maximum velocity compared to both RH and Comp parasites (F) (paired t-test, significance indicated by asterisks). Closed data points are the results from five independent experiments comparing RH and Δphil1 parasites; open data points are the results from four independent experiments comparing Δphil1 and Comp parasites. Each of the independent experiments (assigned a different colour in the scatter plot) was performed in either triplicate or quadruplicate. The total number of parasites analyzed was 6,467 for RH, 9,305 for Δphil1 and 3,743 for Comp. * p<0.05, ** p<0.001, ns = not significant.

https://doi.org/10.1371/journal.pone.0085763.g003

Download:

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

Table 1. Summary of 3D motility parameters.

https://doi.org/10.1371/journal.pone.0085763.t001