A circular zone of attachment to the extracellular matrix acts as a guidance system for Toxoplasma gondii 3D motility

Toxoplasma gondii is a protozoan parasite that infects 30-40% of the world’s population. Infections are typically subclinical but can be severe and, in some cases, life threatening. Central to the virulence of T. gondii is an unusual form of substrate-dependent motility that enables the parasite to invade cells of its host and to disseminate throughout the body. A hetero-oligomeric complex of proteins that functions in motility has been characterized, but how these proteins work together to drive forward motion of the parasite remains controversial. A key piece of information needed to understand the underlying mechanism(s) is the directionality of the forces that a moving parasite exerts on the external environment. The linear motor model of motility, which has dominated the field for the past two decades, predicts continuous anterior-to-posterior force generation along the length of the parasite. We show here using three-dimensional traction force mapping that the predominant forces exerted by a moving parasite are instead periodic and directed in towards the parasite at a fixed circular location within the extracellular matrix. These highly localized forces, which are generated by the parasite pulling on the matrix, create a visible constriction in the parasite’s plasma membrane. We propose that the circular ring of inward-directed force corresponds to a circumferential attachment zone between the parasite and the matrix, through which the parasite propels itself to move forward. The combined data suggest a closer connection between the mechanisms underlying parasite motility and host cell invasion than previously recognized. In parasites lacking the major surface adhesin, TgMIC2, neither the inward-directed forces nor the constriction of the parasite membrane are observed. TgMIC2-deficient parasites remain capable of moving at speeds similar


Introduction
Parasites of the phylum Apicomplexa cause pervasive human disease and are responsible for millions of deaths annually. Toxoplasma gondii is the most prevalent apicomplexan parasite, infecting 30-40% of the human population worldwide [1][2][3]. T. gondii infections are typically subclinical in immunocompetent individuals but may cause blindness [4,5]. In immunocompromised individuals and the developing fetus, the disease can be life-threatening [3,6,7].
T. gondii is a single-cell protozoan, and acute disease is caused by the highly motile tachyzoite lifecycle stage. Like other apicomplexan parasites, the T. gondii tachyzoite uses a unique form of substratedependent motility to invade into and egress from cells of its host, to migrate across biological barriers, and to disseminate throughout the infected host. Motility is, therefore, central to the pathogenesis of toxoplasmosis [8][9][10]. The parasite can move at speeds of up to 3 µm/s without the cilia, flagella, or leading edge protrusions that drive the motility of other eukaryotic cells. Within a three-dimensional (3D) extracellular matrix, the parasite moves along a complex helical trajectory characterized by regular oscillations in velocity and changes in trajectory curvature and torsion [11].
Motility is driven, at least in part, by a complex of proteins known as the glideosome, which is anchored to the inner membrane complex at the parasite periphery ( Figure 1A). In the "linear motor" model of motility, short actin filaments are bound through a bridging protein to the cytosolic tails of transmembrane adhesins, such as TgMIC2, and translocated towards the posterior end of the parasite via a class XIV myosin, TgMyoA [12][13][14][15]. If the extracellular domains of the transmembrane adhesins are attached to ligands on the substrate, the rearward translocation of actin results in forward movement of the parasite. Other apicomplexan parasites, including those that cause malaria (Plasmodium spp.) and cryptosporidiosis, express similar glideosome proteins, suggesting that the mechanism of motility is conserved [16].
Although abundant evidence supports the importance of TgMyoA and other components of the glideosome in motility [10,14,[17][18][19][20][21], several recent observations have raised questions about whether these proteins generate force as described by the linear motor model [17,[22][23][24][25][26] and, if so, whether this is the only motility mechanism available to the parasite. For example: (a) parasites conditionally depleted of TgMyoA and other glideosome components show greatly reduced motility but are not completely immotile [17,22,25], and (b) the linear motor model predicts uniformly forward directed motion, but under some conditions parasites are seen to repeatedly switch the directionality of their movement [27][28][29][30].
Because parasite motility is essential for virulence, elucidation of the underlying mechanism(s) is of both fundamental cell biological interest and potential clinical relevance. One key piece of missing information needed for a full mechanistic understanding of motility is the directionality of the forces the parasite exerts on the external environment as it moves along its 3D helical trajectory. The linear motor model predicts continuous anterior-to-posterior forces along the entire periphery of the moving parasite.
We show here using 3D traction force mapping that the predominant forces exerted by the parasite are instead organized into a highly localized circumferential ring of inward-directed force at a fixed location within the matrix. These inward-directed forces create a visible constriction of the parasite's plasma membrane, reflecting a circular zone of tight attachment of the parasite to the matrix through which the parasite passes through as it progresses forward along its helical trajectory. In the absence of this circular attachment zone, the parasite is still able to move but this movement loses much of its directionality.

Development of a quantitative 3D traction force mapping assay
As a first step towards visualizing the forces exerted by parasites moving in 3D, we used a bead displacement assay in the well-established Matrigel extracellular matrix model [11,25,31]. Fluorescent microspheres embedded within the Matrigel were pulled towards a moving parasite from nearby anterior, posterior, and lateral locations within the matrix, and these microspheres returned to their original positions after the parasite had passed ( Figure 1B, Videos 1&2). These results demonstrate that the parasite does indeed exert a detectable force on the surrounding matrix. Unexpectedly, we also noted constrictions in the body of the fluorescent parasite as it moved ( Figure 1B, Video 1). These constrictions were even more apparent by brightfield microscopy (Fig. 1C, Video 3). Each constriction formed at the apical end of the parasite and remained stationary relative to the matrix as the parasite moved through, reminiscent of the "moving junction" through which the parasite penetrates during host cell invasion (discussed further below).
To develop a more quantitative force mapping assay and explore the relationship between the constrictions and matrix deformation, we used Alexa-Fluor 647-conjugated fibrinogen [32] to generate a fluorescent 3D fibrin matrix (Suppl. Figure 1A). The use of a fluorescent matrix enabled mapping of matrix displacements at all spatial points in our imaged volume rather than at the limited number of discrete positions offered by fluorescent microspheres. Parasite motility in matrices generated using 2.25 mg/ml fibrinogen was indistinguishable from that observed in Matrigel (Suppl. Figure 1B, C). Visual comparison of successive image volumes confirmed that parasites moving through fluorescent fibrin also deform the matrix (Suppl. Figure 1D and Video 4) and undergo the striking constrictions seen in Matrigel.
To calculate and map the deformations of the fibrin matrix, we used the Fast Iterative Digital Volume Correlation (FIDVC) algorithm developed by Christian Franck's group (https://github.com/FranckLab/FIDVC) to iteratively determine the 3D displacement fields between consecutive volumetric image stacks in our time series [33]. Each image volume is divided into 16,807 subvolumes (49 x 49 x 7 subvolumes in x, y, z). The algorithm compares the voxel intensity pattern between corresponding subvolumes at two consecutive timepoints, calculating one 3D displacement vector per subvolume. The data can be displayed as a map for the entire imaging volume (e.g., Video 5) or for any 2D plane (x-y, x-z, or y-z). Most of the 2D images below show the displacement maps in the x-y plane on one of the seven z subvolume levels.
To filter out background signal and identify displacements caused by the parasite, we used timepoints when no parasites were moving to calculate a displacement detection threshold for each individual dataset (see Methods and Suppl. Figure 2A). Using this approach, the limit of detection for 3D displacement magnitudes in the system ranges from 0.042-0.046 µm; a similar 3D displacement detection threshold (0.043-0.046 µm) was observed in samples containing no parasites (Suppl. Figure 2B), validating the approach. When analyzing data in the x-y plane only, the background cutoff was calculated using the x-y displacement magnitudes (rather than the x-y-z displacement magnitudes); this lowered the displacement detection threshold to 0.028-0.031 µm.
In order to translate matrix displacements into the magnitude of forces generated by the parasite, we determined the viscoelasticity of the fibrin matrix using laser trapping (see Methods). The fibrin gel behaves as a predominantly elastic matrix (Suppl. Figure 3) with an elastic modulus of 15.6 pN/µm and a viscous modulus of 0.09 pN*s/µm. The minimal detectable force magnitude in this system is therefore 0.66-0.72 pN in 3D and 0.45-0.49 pN in the x-y plane.

Parasite constrictions are tightly linked to periodic bursts of matrix deformation and motility
With this system for 3D traction force mapping in hand, we analyzed the pattern and directionality of forces produced by wild-type parasites as they moved through the fibrin matrix. Plotting fibrin displacement as x-y maps showed that the matrix deforms in towards the parasite from all x-y directions at discrete, periodic time points along the trajectory, before relaxing back to its initial position (e.g., Figures 2A-D and Videos 6, 7). Viewing the data as projections along each axis demonstrates that the matrix is simultaneously pulled in towards the parasite in all three dimensions (Suppl. Figure 4 and Video 5). In approximately 50% of cases, inward deformation occurred between two consecutive image volumes (1.6 seconds apart) and was followed in the next image volume by relaxation (e.g., Figure 2A-C); in the remainder of cases the deformation was held through multiple time points (e.g., Figure 2D) before relaxing (Suppl. Figure 5). During periods when the matrix was not being pulled inward or relaxing outward, no deformations of the matrix above background were detectable (Suppl. Figure 2A). This observation suggests either that the parasite pulls on the matrix only during some portions of its trajectory or that the parasite produces both strong and weak deformations, and the weak deformations -which could be continuous -are below our level of detection.
When the displacement maps were overlaid onto the images of the moving parasites, it became immediately apparent that the timing of each large matrix deformation coincided with a parasite constriction and that the displacement vectors pointed primarily in towards the constriction ( Figure 2C, D; additional examples are shown in Suppl. Figure 6). The fact that the displacements were directed inwards rather than away from the parasite indicates that the pinching of the parasite membrane results from pulling forces generated by the parasite rather than the parasite squeezing through a narrow pore within the matrix, which would push the matrix away. When a constriction persisted over multiple time points, the fibrin continued to deform into the constriction and/or to hold the deformation over those same time points as the parasite passed through (e.g., Figure 2D).
The time it took a parasite to move completely through a constriction ranged from 1.6-56.2 seconds (average 6.8 ± 5.2 seconds, measured in brightfield) with the longer times due to a stall in forward progression of a constricting parasite. The time between one constriction finishing and another beginning was also variable ( Figure 3A). In 46.3% of the cases, the end of one constriction was followed immediately by the beginning of another, while in 13.8% of the cases (n=188) a second constriction would start before the first constriction finished ( Figure 3A, B and Video 8). Most importantly, all parasites that move at least one body length undergo a constriction and the total trajectory length is directly proportional to the number of constrictions observed ( Figure 3C), strongly suggesting that the parasite-generated forces that create the constriction play an important role in forward movement.
The much reduced width of the parasite during a constriction (2.56 ± 0.38 µm at its widest point before constriction vs.1.39 ± 0.24 µm during constriction; Figure 3D) led us to question whether the constriction presents a physical obstacle to the forward motion of the nucleus (normal nuclear diameter measured by DNA staining = 2.00 ± 0.26 µm). In parasites labeled with both a fluorescent DNA stain and a fluorescently-labeled antibody against the major surface protein, TgSAG1 (to visualize the constriction), the normally round parasite nucleus did indeed become thinner and more elongate as it passed through the constriction and regained its round shape once it was through ( Figure 3E). The ratio of the nuclear dimensions perpendicular and parallel to the long axis of the parasite was 0.93 ± 0.08 in a nonconstricting parasite and 0.50 ± 0.06 during a constriction. We examined whether the periodic need to squeeze the nucleus through the constriction might be responsible for the periodic changes in speed observed along the parasite's trajectory [11], but tracking analysis of constricting parasites did not reveal any consistent relationship between the instantaneous speed of the parasite and position of the nucleus relative to the constriction.
Interestingly, the distribution of the fluorescent anti-TgSAG1 antibody on the parasite surface changed as the parasite moved: much of the fluorescent signal appeared to be "swept" from anterior to posterior at the constriction as the parasite moved through (Suppl. Figure 7 and Video 9). The redistributed antibody was ultimately shed and deposited behind the moving parasite in the form of a fluorescent helical trail within the Matrigel (Suppl. Figure 7 and Video 9).

Parasites that do not generate constrictions do not detectably deform the matrix
Since TgMyoA is thought to play a central role in force production during parasite motility ( Figure   1A), we conducted force mapping experiments using a parasite line lacking TgMyoA [17,18]. As previously reported [25], these parasites are much less motile in 3D than wild-type parasites: only 18 of 769 TgMyoA knockout parasites examined (2.3%) moved one body length or more, compared to 99 of 278 wild-type parasites (35.6%; Suppl. Figure 8A). Of the 18 TgMyoA knockout parasites that moved at least one body length, none showed constrictions like those seen in wild-type parasites. The pattern of movement was also different from wild-type parasites; rather than moving along smooth helical trajectories, the TgMyoA knockout parasites moved in tightly twisting arcs or in a stair-like pattern consisting of right-angled turns associated with a sharp bend in the body of the parasite (Suppl. Figure 8B and Video 10). In stark contrast to the wild-type parasites, we also saw no evidence by traction force mapping of a ring of inward-directed forces produced by the TgMyoA knockouts. In fact, we did not detect any displacement of the matrix by these parasites (Suppl. Figures 2C and 8C-E), suggesting that any forces produced by the moving knockout parasites are below the limit of detection in our traction force assay.
We also tested whether the surface adhesin, TgMIC2, plays a role in generating the constrictionassociated inward forces on the matrix. The motility defect in parasites lacking TgMIC2 is less severe than the defect in parasites lacking TgMyoA, with 36 of 297 TgMIC2 knockout parasites (12.1%) capable of moving more than one body length (see also [31,34]). However, these parasites often moved in an erratic, stop and start pattern, bending sharply and moving in tightly twisting arcs ( Figure 4A and Video 11). Moreover, the TgMIC2 knockout parasites also failed to form constrictions and did not detectably deform the fibrin matrix (Fig. 4A, Suppl. Figures 2D & 9C-E) despite being able to move at maximum speeds similar to wild-type parasites [31]. Taken together, these data confirm that motility is significantly altered, but not completely eliminated, in parasites lacking TgMyoA or TgMIC2. Furthermore, the data demonstrate that both proteins are required to produce the constriction and the constriction-associated inward forces on the matrix observed during the motility of wild-type parasites.

TgMIC2-deficient parasites move less directionally
The larger number of TgMIC2 knockout parasites capable of long runs of motility (compared to the MyoA knockouts) enabled us to analyze their phenotype in greater detail. The nuclei of the parasites lacking TgMIC2 did not change shape as they moved in 3D ( Figure 4B and Suppl. Figure 9A), providing further evidence that these parasites do not undergo the constrictions seen in wild-type parasites. Given their apparent propensity for tight turns, we also quantified the ability of the TgMIC2 knockout parasites to move directionally, by comparing displacement distance to trajectory length. While parasites with and without TgMIC2 traveled along trajectories of similar mean length (38.47 ± 9.3 µm vs. 42.22 ± 8.0 µm, respectively), parasites lacking TgMIC2 moved approximately half as far from their starting point as wild-type parasites ( Fig. 4C and Suppl. Figure 9B). These data suggest that the circumferential zone of attachment to the matrix, which requires TgMIC2, functions as a guidance system that converts meandering motility into more linear, directed forward progression.

Discussion
The linear motor model of apicomplexan motility predicts that a parasite moves by pulling continuously on the matrix in an anterior-to-posterior direction along the length of the parasite. It was, therefore, unexpected that the predominant forces revealed by our new 3D traction force mapping assay were instead periodic and highly localized, pointing inwards from all directions towards a circumferential location on the parasite surface. This circular band of inward-directed force forms at a fixed position within the matrix and creates a visible constriction at the parasite periphery. The constriction passes from the anterior to the posterior end of the parasite as the parasite moves forward. Typically, when the constriction reaches the parasite's posterior end a new constriction forms at the apical end and the cycle repeats. These results suggest that the moving parasite periodically assembles ring-shaped zones of tight attachment to the matrix, which deform the matrix and serve as stationary platforms that the parasite propels itself through, one body length at a time.
Attachment to the substrate, force generation, and motility in 2D vs. 3D. Both malaria sporozoites and T. gondii tachyzoites are capable of moving in circles on 2D surfaces. The pioneering work of Munter, Frischknecht and colleagues combined traction force mapping with reflection interference contrast microscopy (RICM) during the 2D circular gliding of malaria sporozoites [35]. They concluded that the parasite adheres to the substrate at its apical and posterior ends; each of these adhesion sites produces stalling forces parallel to the long axis of the parasite, and disengagement of the adhesion sites drives forward motion. Traction forces perpendicular to the direction of movement were also detected at the center of the parasite but these were interpreted as nonspecific and nonproductive. The apical and posterior adhesion sites were shown to move forward with the parasite, rather than remaining stationary with respect to the substrate.
T. gondii tachyzoites demonstrate a second form of productive motility on 2D surfaces called helical gliding: the banana-shaped parasite starts with its left side in contact with the substrate, moves forward one body length while rotating along its longitudinal axis until it lies on its right side, and then flips back to its left side to repeat the cycle [36]. This form of motility is the equivalent of the fully helical trajectories seen in 3D but constrained in 2D by contact of the curved parasite with the rigid substrate. In RICM studies of 2D helical gliding [19,37] the tachyzoite was seen to first attach to the substrate at its apical end. The zone of attachment then expanded along the length of the parasite and, as the parasite moved forward and its apical end twisted up off the cover slip, the zone of attachment shrunk until it was reduced to a spot at the posterior end. In contrast to Plasmodium sporozoites, the position of the tachyzoite adhesion site remained fixed relative to the substrate and the parasite slid over this attachment site as the parasite moved forward. 2D traction force mapping revealed forces parallel, but not perpendicular, to the longitudinal axis of the tachyzoite; these traction forces, together with spring-like forces generated by the parasite's microtubule cytoskeleton, were postulated to produce forward motion [37].
When these previous studies of T. gondii 2D motility are combined with the 3D results presented here, a consensus emerges that the basic unit of motility consists of apical attachment to the substrate, followed by apical-to-posterior translocation of the adhesion site, which is fixed in its position relative to the substrate. When the adhesion site reaches the posterior end of the parasite, a new apical attachment site is formed, and the cycle begins again. The adhesion site in a 3D matrix is circular, and the pulling forces exerted by the parasite on the matrix manifest as a visible constriction in the parasite's plasma membrane. In 2D, no deformation of the parasite has been reported at the adhesion site, likely reflecting differences in attachment to the substrate via a point/patch on the parasite surface vs. circumferential attachment. We also saw no evidence in 3D for a persistent stall force at the rear of the parasite. The biggest difference between the 2D and 3D force mapping data is that, in 3D, all force vectors point towards one highly localized circumferential region on the surface of the parasite, and many of these force vectors contained a strong component perpendicular to the long axis of the parasite.

3D motility and invasion: variations on a theme. During the invasion of host cells by T. gondii,
proteins secreted from the parasite's apical organelles assemble a ring-shaped zone of tight attachment between the membranes of the two cells. This "moving junction" does not in fact move: it is anchored to the host cytoskeleton and is hypothesized to provide a fixed platform against which the parasite exerts force [38][39][40][41][42]. Parasite surface adhesins, which are engaged with the junction via their extracellular domains, are thought to be translocated by TgMyoA in an anterior-to-posterior direction, pushing the parasite through the junction and into the host cell. The body of the parasite narrows dramatically as it passes through the moving junction, bearing a striking resemblance to the constrictions seen during 3D motility.
Two central players in moving junction formation are TgAMA1, an adhesin secreted onto the parasite surface from the micronemes, and TgRON2, which is secreted by the parasites into the host cell plasma membrane and serves the ligand to which TgAMA1 binds. Neither TgAMA1 [43,44] nor TgRON2 (RVS and GEW, unpublished data) plays any detectable role in parasite motility. Instead, TgMIC2 -another microneme protein secreted onto the parasite surface -likely plays an analogous role during motility to that of TgAMA1 during invasion, with TgMIC2 binding to ligands in the extracellular matrix rather than to TgRON2 in the host cell membrane.
As the parasite moves forward during both invasion and 3D motility, it simultaneously rotates around its long axis [11,36,45]. Our results provide a possible explanation for this rotational motion. In both cases, parasite receptors bind to their cognate ligands in the form of a circular ring of attachment that is fixed in position relative to the host cell (in the case of invasion) or the extracellular matrix (in the case of motility). It has long been proposed that the TgMyoA motor driving invasion and motility is arranged helically within the parasite, along the spiraling subcortical microtubules (e.g., [37]). Alternatively, a recent modeling study suggests that the unique geometry of the parasite periphery can generate a self- As a parasite invades a host cell, antibodies are "shaved" off the parasite surface at the moving junction by an as yet unknown mechanism [46]. Remarkably, a similar phenomenon is observed during  The numerous parallels between invasion and 3D motility noted here suggest that the two processes may be more mechanistically similar than previously recognized [12,19], with implications for our understanding of each. For example, our data suggest that 3D motility can be best described as the parasite undergoing sequential invasion-like events through circular zones of attachment to the matrix; during invasion the parasite only needs to move a single body length to penetrate into the host cell, but during motility the parasite can string multiple constriction events together to move multiple body lengths ( Figure 3C). The data also raise the possibility that the characteristic constriction of the parasite seen at the moving junction during invasion does not reflect the parasite being squeezed as it pushes itself through the small diameter opening into the host cell, as is commonly thought (e.g., [47,48]), but rather the parasite attaching to and pulling on the host cell plasma membrane as it penetrates.  Figure 3), the relationship between acceleration and force is complex and we cannot calculate with any confidence how much total displacement / force would be required to achieve a given parasite velocity and relate this to the displacement detection threshold of the assay.
The marked regularity with which the parasite generates the circular zones of attachment on its surface, yielding on average one constriction per body length of motion, could be due to repetitive bursts of formin1-mediated actin polymerization at the parasite's apical tip [19] or to periodic release of adhesins such as TgMIC2 from the micronemes. Alternatively, the observation that constrictions occur back-to-back almost 50% of the time (Fig. 3A), may suggest that a new constriction forms at the apical tip before the prior constriction has ended, but this new constriction stalls as the TgMyoA (or possibly TgMyoH [19]) to which it is connected pulls against the adhesive forces of the more posterior previous constriction. Future real-time analysis of actin polymerization and microneme secretion in motile parasites and higher sensitivity force mapping will be necessary to further interrogate the periodic nature of ring assembly.
The lack of a motility-associated constriction in TgMyoA-deficient parasites could reflect an inability either to assemble ring of adhesins at the apical tip of these parasites or, more likely, to move through the incipient ring once it has formed. Alternatively, it could be that the ring-shaped structures that form during motility are contractile, and contractility requires TgMyoA. The inward deformation of the fibrin matrix towards the constriction reported here is consistent with a contractile ring, as is the physical constriction of the parasite plasma membrane. Further experiments will be required to distinguish between these possibilities.
If the coupling of actin flow to the circular zone of attachment is important for forward motion, how do the parasites lacking TgMIC2 move at normal speeds in the absence of constrictions? It is possible that both the wild-type and mutant parasites generate small forces distributed over their entire surface (in addition to larger forces generated at the constriction in wildtype) that are collectively sufficient to drive motility but individually below our level of detection. Alternatively, the loss of TgMyoA or TgMIC2 might lead to compensatory changes in the expression of other genes with overlapping or redundant function [23,49,50]. The TgMIC2 KO parasites do in fact show changes in the expression of several other microneme and motility-associated proteins [31]. Perhaps most interestingly, parasites lacking TgMyoA and/or TgMIC2 might use a motility mechanism that is entirely different from that of wild-type parasites (e.g., [18,22]), does not involve a constriction, and is not as efficient as the wild-type mechanism but is nevertheless capable of supporting motility and sustaining the parasite's lytic cycle. By analogy with observations from animal cells [51], parasites might use different mechanisms for motility in different situations such as squeezing through tight junctions [8] vs. migrating through loose interstitial tissues. Any attempt to develop new therapeutics targeting parasite motility [52,53] will need to take into account the possibility that the parasite has multiple mechanisms to draw upon. If the motility of the mutants is driven by a different mechanism than that of wild-type parasites, higher sensitivity 3D traction force mapping may reveal informative differences in their pattern of force generation.

A guidance system for motility?
The forward motion of the TgMIC2 knockout parasites, which lack the circular attachment zone, is significantly more disorganized than the motility of wild-type parasites. The mutant parasites can still generate relatively long trajectories, but the trajectories take them less far from their starting point than wild-type parasites. As discussed above, attachment to the matrix via a circular ring of adhesion likely contributes to the rotational twisting of the parasite as it moves forward and thereby to the overall helicity of the parasite's trajectory. Extensive studies of the helical swimming behavior of bacteria has shown that a curved cell shape and helical trajectory are a particularly efficient way for small cells to move through viscous media [54][55][56][57][58]. Engagement of T. gondii with the extracellular matrix through a circular band of adhesion may therefore not only create a fixed platform for the parasite to pull against and move forward; it may also function as a guidance system to help the parasite move efficiently through the various environments it encounters as it disseminates through the tissues of the unfortunate organisms it infects. Matrigel Matrix. Pitta chambers containing parasites embedded in polymerized Matrigel were prepared as previously described [11]. Briefly, parasites were harvested from infected HFFs via syringe release and passed through a 3 µm Nucleopore filter (Whatman, Piscataway NJ). Parasites were pelleted and resuspended at in Live Cell Imaging Solution (LCIS) buffer. For fluorescence imaging, the LCIS contained 0.5 mg/ml Hoechst 33342 (Thermo Scientific, Waltham, MA) and/or a 1:20 dilution of Alexa546-conjugated anti-TgSAG1 (see Table 1). Fluorescent anti-TgSAG1 (100µg) was prepared using were flash frozen and stored at -80C until use. Fibrin gels [32] containing parasites were prepared in Pitta chambers as follows. The parasite culture medium was replaced with LCIS buffer before harvesting via syringe release and filtering. 500 µl of the parasite suspension were pelleted and resuspended in LCIS buffer to achieve a higher parasite concentration. Parasites were mixed with fibrinogen (final fibrinogen concentration of 2.25, 4.5, or 9 mg/ml). LCIS buffer containing thrombin (Sigma, Burlington, MA) and FBS was added to the parasite-fibrinogen suspension for final concentrations of 1.5x10 8 parasites per ml, 1 unit/ml of thrombin, and 10% FBS. The mixture was immediately pipetted into a Pitta chamber and allowed to polymerize at room temperature for three minutes before mounting on the microscope and imaging.

Methods
Image Acquisition. A Nikon Eclipse TE300 epifluorescence microscope (Nikon Instruments, Melville, NY) equipped with a NanoScanZ piezo Z stage insert (Prior Scientific, Rockland, MA) was used to image parasites. See Table 1   Tracking Parasite Motility. Datasets were analyzed in Imaris ×64 v. 9.2.0 (Bitplane AG, Zurich, Switzerland). Fluorescently labeled parasite nuclei were tracked using the ImarisTrack module within a 1018 pixel × 380 pixel region of interest to prevent artifacts from tracking objects near the border. Spot detection used an estimated spot diameter of 3.0 × 3.0 × 6.0 µm (x, y, z). A maximum distance of 6.0 µm and a maximum gap size of 2 frames were applied to the tracking algorithm. Tracks with durations under 10 seconds or displacements of less than 2 µm were discarded to avoid tracking artifacts and parasites moving by Brownian motion, respectively [11]. Accurate tracking was confirmed by visual inspection of parasite movements superimposed on their calculated trajectories.
For the comparison between Matrigel and fibrin, we analyzed three biological replicates, each consisting of three technical replicates. Student's T-tests were used to determine statistical significance between fibrin and Matrigel samples.

Fibrin Deformation with FIDVC. The Fast Iterative Digital Volume Correlation (FIDVC) algorithm
was used to calculate 3D fibrin displacements [33] by comparing two consecutive image volumes. The image volumes were cropped to 384 × 384 pixels (x,y) × 48 slices (z) before running FIDVC. The initial interrogation window size was set to 32 × 32 × 32 and the program was run incrementally. The displacements arrows were plotted using a combination of previously described MATLAB code for 2D quiver color-coded for directionality [32] and custom MATLAB code. To filter out background signal, we calculated the mean magnitude of the 16,807 displacement vectors for each timepoint comparison in each dataset. The displacement detection threshold for that dataset was set as three standard deviations above the lowest of these mean values.       Suppl. Figure 1D.

Rheology. A Lumicks C-
Video 5: 3D displacement map surrounding the moving parasite shown in the second panel of Figure 2C and frame 2 of Video 6. Length of arrows indicating displacement magnitude are multiplied 10-fold for display. The parasite is outlined on each z plane with red circles. Single frames from this video displaying projections onto the x-y, x-z and y-z planes are shown in Suppl. Figure 4.

Video 6:
x-y displacement map overlaid on the moving parasite shown in Figure 2A-C. Playback is in real time. Arrow size corresponds to relative displacement magnitude and arrow color to displacement direction as described in Figure 2.

Video 7:
x-y displacement map overlaid on the moving parasite shown in Figure 2D. Playback is in real time. Arrow size corresponds to relative displacement magnitude and arrow color to displacement direction as described in Figure 2.