Temperature dependence of the motility of the parasitic microswimmer Trypanosoma brucei Authors: Hannes Wunderlich, Leon Schramm, Marinus Thein, Lucas Brehm, Klaus Ersfeld, Matthias Weiss Published: 2026 ABSTRACT The parasitic unicellular microswimmer Trypanosoma brucei is the causative agent for severe diseases in humans and cattle. Trypanosomes have a complex life cycle with the best-characterized stages being the bloodstream form (BSF) in the vascular system of the mammalian host and the procyclic form (PCF) in the gut of the tsetse fly vector. Morphologically, PCF and BSF trypanosomes appear very similar, being propelled by a single flagellum that is wrapped around the spindle-shaped cell body. Despite this similarity, we show here that PCF and BSF trypanosomes have significantly different motility patterns and are also very different in their sensitivity to the ambient temperature. PCF cells show a clear run-and-tumble pattern with an improved motility for increasing temperatures, most likely facilitating long-range transport in the gut of the poikilothermic tsetse fly. These features are even preserved in mutants with compromised post-translational modification of microtubules. In contrast, BSF trypanosomes show a considerably more erratic motility that hardly varies with temperature, potentially indicating that an active exploration of the host’s body fluids with a directed motility is not necessary for the parasite’s life cycle. Significance Trypanosoma brucei is a unicellular parasite that is responsible for severe diseases in humans and cattle. Trypanosomes have a complex life cycle with the best-characterized stages being the bloodstream form (BSF) in the mammalian host and the procyclic form (PCF) in the gut of the tsetse fly. Both PCF and BSF trypanosomes are microswimmers with a single flagellum that is wrapped around the spindle-shaped cell body. Despite this morphological similarity, we show that PCF and BSF trypanosomes are significantly different in their (temperature-dependent) motility patterns. We hypothesize that the PCF mode of motility supports active movement over larger distances in the vector, whereas BSF trypanosomes may rely on the host’s bloodstream for long-range transport and dissemination. Introduction Microswimmers, i.e., self-propelled organisms with a significant motility at low Reynolds number, are ubiquitous in biology, from single bacteria and sperms that push themselves forward by beating flagella (“pushers”) to algae that seem to imitate human breast strokes (“pullers”); see Ishikawa1 for a recent review. A particularly interesting microswimmer of the puller type is the single-celled parasitic organism Trypanosoma brucei, the causative agent of the sleeping sickness in humans and of the wasting disease nagana in cattle.2 Trypanosomes undergo a complex life cycle within the mammalian host and the tsetse fly vector, which includes distinct morphological and physiological adaptations.3,4 The best-characterized stages are the bloodstream form (BSF), residing in the vascular system and tissues of the mammalian host,5,6 and the procyclic form (PCF) in the gut of the vector.7 Both forms can easily be cultured in vitro, whereas attempts to cultivate other life-cycle stages have been unsuccessful. The motility of both BSF and PCF trypanosomes is caused by a microtubule-based axoneme structure in a single flagellum that emerges from the flagellar pocket, near the posterior cell pole, and extends helically alongside the cell body (see Langousis and Hill and Gluenz et al.8,9 for instructive electron micrographs), being attached via specialized proteinaceous structures and known as the flagellar attachment zone. The concerted action of molecular motors shears microtubule bundles, hence inducing a bending wave that propagates along the flagellum.8,10,11 The resulting hydrodynamic flow eventually moves the entire cell into the direction of the flagellar tip. Interestingly, the spindle-shaped cell body is stabilized by a nematically ordered and cross-linked array of subpellicular microtubules12,13,14 that supply the cell with a significant bending elasticity. The periodic bending of the helically attached flagellum therefore directly transmits to the elastic cell body, leading to a complex bending pattern of the entire cell that has been studied in some detail in elaborate model simulations.10,11,15 Microtubule-mediated motility is not only essential for the dispersal of the parasite in the host and vector but also for its virulence and survival in both organisms.16,17 Therefore, any changes in the so-called tubulin code18 (e.g., changes in the post-translational modification of tubulin dimers) can be expected to affect the motility of PCF and/or BSF trypanosomes beyond effects that are due to different gene expression profiles—even more so since flagellar and subpellicular microtubules are biochemically stable during interphase and do not show the familiar dynamic instability seen in mammalian culture cells. So far, the majority of quantitative motility studies has been focused on BSF trypanosomes due to the direct medical relevance.15,19,20 More recently, PCF trypanosome motility has also been studied quantitatively,21 mostly with an emphasis on the effects of changes in the post-translational modification of microtubules.22,23,24 However, a thorough comparison of BSF and PCF trypanosome motility patterns has been lacking so far, as has a quantitative study on the impact of temperature on these parasitic microswimmers. Given that PCF trypanosomes live in a poikilothermic environment with variable temperatures, whereas BSF cells experience constant temperature of warm-blooded mammals, one may expect differences between the two forms, at least when changing the temperature in the physiological range. However, even the BSF, although resident in homeothermic animals, needs to cope with a considerable variation in temperature, as these cells are hosted by a wide range of vertebrates that differ in their body temperatures by several degrees.25,26 Here, we have addressed these points by quantitatively studying the motility patterns of PCF and BSF trypanosomes at different temperatures. Our data show that both forms feature markedly different motility patterns and are unequally sensitive to temperature changes. We also show that PCF mutants with altered post-translational modifications of tubulin show differential effects upon temperature changes that are mostly consistent with the effects seen for the wild-type strain. Altogether, our data indicate that PCF and BSF trypanosomes are surprisingly different in their motility characteristics, despite being very similar in their general propulsion mechanism. Materials and methods Cell culture Procyclic (PCF) Trypanosoma brucei 427 cells were cultivated in SDM-79 medium (Life Technologies, UK) supplemented with 10% fetal bovine serum (Capricorn Scientific, Germany) and 7.5 mg/L hemin (Merck, Germany) at 27°C, unless stated otherwise. Knockout PCF cell lines were obtained as described before.23,24 BSF T. brucei cells were cultured in HMI-9 medium (Life Technologies, UK) supplemented with 10% fetal bovine serum (Capricorn Scientific, Germany) at 37°C and 5% CO2. Cell densities were determined using a CASY cell counter (Roche Innovatis, Germany). Motility analysis The motility of T. brucei cells was analyzed by brightfield microscopy using quasi-two-dimensional chambers, as described before.22,23,24 In brief, to restrict cell motion to a thin slab for long-term tracking, custom-made PDMS (polydimethylsiloxan) chambers with a height of 20 μm were produced by standard soft-lithography. Chambers were filled with cell suspensions for microscopy (typical density of 3–4 × 106 cells/mL) and chamber inlets were sealed to prevent evaporation-driven flows. Prior to experiments, chambers were equilibrated in Milli-Q water to prevent PDMS swelling during imaging. All temperature-dependent experiments were conducted either by equilibrating the entire microscope environment (setup 1) or by using an external temperature control system (setup 2). In setup 1, imaging was carried out on a Leica DMI6000B inverted microscope equipped with a Leica DFC360FX camera and a 20 ×/0.7NA air objective. Here, motility measurements were performed at two temperatures, 25°C and 33°C, using a custom-made incubator box surrounding the entire microscope. In setup 2, motility measurements were performed on a Zeiss inverted microscope (AXIO Observer Z1) equipped with an Andor Neo sCMOS camera and a 20 ×/0.3NA air objective. In this configuration, the temperature of the cell suspension was controlled by an ibidi temperature controller system (ibidi, Germany), consisting of a heated stage insert and an external control unit. The chamber temperature was continuously monitored via the integrated temperature sensor and recorded throughout the experiment to ensure stable thermal conditions ranging from 24°C to 40°C. Motility recordings typically consisted of at least ten 3-min videos, acquired at a rate of 10 frames per second, yielding about 50–100 trajectories per video. Image processing and tracking Image sequences were exported as TIF stacks and further processed using ImageJ.27 Cell tracking was performed using the TrackMate plugin28 with identical detection and linking parameters for all conditions. The general tracking workflow, parameter choices, and validation procedures were identical to those described before.23,24 Extracted trajectories were filtered to exclude tracks of dead or immobile cells; i.e., we discarded tracks with an end-to-end distance below 5 μm, tracks that did not exhibit at least one displacement exceeding 20 μm within a 10-s time window, as well as trajectories that were unreasonably short (insufficient statistics) or long (excessively long trajectories typically indicated non-physiological behavior, causing a prolonged residence within the field of view). The remaining trajectories were analyzed using custom-written MATLAB scripts (The MathWorks, Natick, MA, USA), including previously provided codes from an analysis toolbox for single-particle tracking data.29 In total, our in-depth analysis included n = 1,780, 2,310, 3,018, 4,037, 6,193, and 5,129 and n = 322, 467, 692, 558, 413, 1,053 trajectories at temperatures T = 24.0°, 27.0°, 30.0°, 33.5°, 37.0°, 40.0°C for PCF and BSF trypanosomes, respectively. For mutant PCF trypanosomes, we analyzed n(25.0°C) = 288 and n(33.0°C) = 490 (VASH); n(25.0°C) = 410 and n(33.0°C) = 191 (TTLL1); n(25.0°C) = 503 and n(33.0°C) = 145 (TTLL4C); n(25.0°C) = 476 and n(33.0°C) = 107 (TTLL6B) trajectories in detail. Results and discussion BSF and PCF trypanosomes have different motility patterns To obtain two-dimensional trajectories of PCF and BSF trypanosomes, we used a previously established microfluidic chamber and acquired brightfield microscopy image series with a time resolution of Δt = 100ms between successive frames, using a temperature-controlled setup (see materials and methods). Trajectories of the cells’ center of mass were subsequently obtained from image series by a well-established tracking algorithm.28 Trajectories were then subjected to a quality-control step; that is, tracks from immobile trypanosomes and too-short trajectories were discarded. In particular, trajectories were required to contain N ≥ 100 consecutive positions without blanks; i.e., r(t = iΔt) = ri = (xi, yi) with i = 1, …, N. Representative images of PCF and BSF trypanosomes, together with typical trajectories, are shown in Figures 1A and 1B, respectively. Visual inspection of the example tracks suggests that the two trypanosome forms vary in their motility pattern, albeit being very similar in their gross morphology.Figure 1Temperature-dependence of PCF and BSF motility. Differential-interference contrast images of T. brucei in the (A) procyclic (PCF) and (B) bloodstream (BSF) stage. Nucleic acid-rich regions are highlighted in pink, with the larger and smaller regions being nucleus and kinetoplast, respectively. Three colored representative trajectories (scale bars as indicated) indicate that PCF but not BSF trypanosomes seem to feature distinct mobility phases (run versus tumble). (C and D) EA-TA-MSDs of PCF and BSF trypanosomes for different temperatures (color-coded symbols). For better visibility, data for successive temperatures have been shifted upwards by a factor of 2. Dotted and dashed lines indicate ballistic and diffusive scaling for comparison. Full-colored lines, superimposed to symbols, indicate the best fit according to Equation 2 for τ < 5 s, from which the parameters D and v0 were extracted. Lines in the range τ > 10 s indicate best fits of an asymptotic diffusion law, ⟨r2(τ)⟩t,E=D∞t. (E) Both BSF and PCF show an increase of D with temperature (open symbols, error bars depicting the 95% confidence interval of the fit process). While BSF trypanosomes (red squares) only show an increase by 30%–40% from 24°C to 40°C, PCF cells almost triple their value (black circles). Still, values for BSF cells are consistently higher at all tested temperatures, reflecting their more vigorous flagellar beating. The long-time diffusion coefficient D∞ (filled symbols), extracted as described in the main text and rescaled by the indicated factors to match the axis range, shows the same trend as D. (F) Despite their higher activity, BSF trypanosomes show almost no variation in values for v0, indicating an almost constant swimming velocity on the ensemble level. In contrast, PCF trypanosomes feature an almost 3-fold increase in the swimming velocity v0 on the ensemble level. Please note that error bars (representing the 95% confidence interval of the fit process) are in all cases smaller than the symbol size. Temperature-dependence of PCF and BSF motility. Differential-interference contrast images of T. brucei in the (A) procyclic (PCF) and (B) bloodstream (BSF) stage. Nucleic acid-rich regions are highlighted in pink, with the larger and smaller regions being nucleus and kinetoplast, respectively. Three colored representative trajectories (scale bars as indicated) indicate that PCF but not BSF trypanosomes seem to feature distinct mobility phases (run versus tumble). (C and D) EA-TA-MSDs of PCF and BSF trypanosomes for different temperatures (color-coded symbols). For better visibility, data for successive temperatures have been shifted upwards by a factor of 2. Dotted and dashed lines indicate ballistic and diffusive scaling for comparison. Full-colored lines, superimposed to symbols, indicate the best fit according to Equation 2 for τ < 5 s, from which the parameters D and v0 were extracted. Lines in the range τ > 10 s indicate best fits of an asymptotic diffusion law, ⟨r2(τ)⟩t,E=D∞t. (E) Both BSF and PCF show an increase of D with temperature (open symbols, error bars depicting the 95% confidence interval of the fit process). While BSF trypanosomes (red squares) only show an increase by 30%–40% from 24°C to 40°C, PCF cells almost triple their value (black circles). Still, values for BSF cells are consistently higher at all tested temperatures, reflecting their more vigorous flagellar beating. The long-time diffusion coefficient D∞ (filled symbols), extracted as described in the main text and rescaled by the indicated factors to match the axis range, shows the same trend as D. (F) Despite their higher activity, BSF trypanosomes show almost no variation in values for v0, indicating an almost constant swimming velocity on the ensemble level. In contrast, PCF trypanosomes feature an almost 3-fold increase in the swimming velocity v0 on the ensemble level. Please note that error bars (representing the 95% confidence interval of the fit process) are in all cases smaller than the symbol size. To thoroughly analyze cell motility, we followed our previous approach29 and calculated first for each trajectory the time-averaged mean square displacement (TA-MSD).(1)⟨r2(τ)⟩t=⟨r(t+τ)−r(t)2⟩tand the ensemble average ⟨r2(τ)⟩t,E (EA-TA-MSD) of these for each temperature. The resulting EA-TA-MSDs are shown in Figures 1C and 1D with increasing temperatures being encoded from blue to red. All EA-TA-MSDs were seen to feature a non-uniform power-law increase, with an almost ballistic behavior ∼τ2 for intermediate lag times, whereas, for small and large lag times, a roughly linear scaling emerged (cf. dotted and dashed lines in Figures 1C and 1D for comparison). Please note that data for increasing temperatures have been shifted vertically for better visibility (factor 2 between successive temperatures). Despite this shift, EA-TA-MSDs of PCF trypanosomes can be seen to depend more strongly on the ambient temperature than BSF cells, discernable by the elevated spreading of curves in Figure 1C. Given that trypanosome cells are elongated objects that undergo a vibrant bending motion due to the action of the flagellum, the cell’s center of mass displays a seemingly erratic movement on short timescales τ < 10Δt. Therefore, the EA-TA-MSD can be expected to be captured by a simple diffusive scaling ⟨r2(τ)⟩t,E∼τ on short timescales. On longer timescales, the net motion of the parasitic microswimmer with a finite drift velocity becomes dominant; i.e., the EA-TA-MSD can be expected to show a crossover to a ballistic scaling ⟨r2(τ)⟩t,E∼τ2. Following this rationale, we have fitted all EA-TA-MSDs obtained for PCF and BSF trypanosomes at different temperatures with an empiric superposition of both types of motion,(2)⟨r2(τ)⟩=Dτ+(v0τ)2.As can be seen in Figures 1C and 1D, this fit captures all MSD data very well in the range τ ≤ 5 s; i.e., one can use the values of the parameters D and v0 to rate the temperature dependence of EA-TA-MSDs for PCF and BSF trypanosomes. For larger timescales, however, marked deviations from the ballistic motion can be seen and EA-TA-MSDs rather tend toward a linear scaling. These deviations from a ballistic motion are, for example, due to intermittent tumbling phases that interrupt periods of directed motion (see discussion below). As a consequence, the trajectories will adopt features of a random walk on large timescales, eventually leading again to a linear scaling with an asymptotic diffusion constant.30 To obtain an estimate on this asymptotic behavior in the available range of lag times, we have fitted EA-TA-MSDs with an offset-free linear function ⟨r2(τ)⟩t,E=D∞τ in the range τ ≥ 10 s, allowing for a comparison of D and D∞. Inspecting the temperature dependence of D and v0 confirms the visual impression that PCF and BSF trypanosomes are differently affected by the ambient temperature (Figures 1E and 1F): Both parameters increase almost 3-fold within the tested temperature range for PCF cells, whereas BSF trypanosomes feature an almost constant v0 and only a 30%–40% increase in D. Values for D∞ were considerably higher than D for PCF and BSF, as expected, yet their variation with temperature closely followed that of D (see Figure 1E). Thus, BSF cells are considerably less temperature sensitive but show consistently higher values for D and lower values for v0 than PCF cells. Higher values for D most likely reflect the elevated activity of the flagellum in BSF cells, often described as a more vigorous flagellar beating. This leads to a vivid but erratic wiggling of the cell’s center of mass on short timescales from which only a rather poor directed motility (with low v0) emerges on larger timescales. To quantify the latter observation more thoroughly, a dissection of the short-term zig-zagging of cells due to the flagellum-induced bending and the more ballistic center-of-mass motion on larger timescales is needed. Following our previous approach,23 we have therefore calculated for each temperature the characteristic value of the ensemble- and time-averaged velocity autocorrelation function (VACF),(3)C∗=⟨v(t)v(t+τ)⟩t⟨v(t)2⟩tE.Here, the instantaneous velocity is defined via the step taken within a period τ, v(t) = [r(t + τ) − r(t)]/τ, and values C∗ > 0 indicate a persistent motion while C∗ = 0 signals a Markovian random movement (see Rehfeldt and Weiss29 for a detailed discussion). As expected, C∗ increased for all cells and all temperatures for small τ, indicating that the center-of-mass velocity is successively less affected by the cell’s short-term wiggling, approaching more and more a smooth ballistic motion (Figures 2A and 2B). For τ > 1 s, however, the values of C∗ are seen to plateau or decrease again, reflecting that cells do not simply follow a straight trajectory but rather include more random segments on elevated timescales (cf. also the example trajectories in Figures 1A and 1B). Given that all data for C∗ have a common optimum in the same range of periods τ, we have fixed τc = 1 s as the critical timescale at which a time-resolved trypanosome swimming velocity along a trajectory can be estimated best without including contributions from the short-term zig-zag motion.Figure 2Persistence and local straightness of PCF and BSF motility. (A and B) Characteristic VACF values C∗, defined in Equation 3, show for PCF and BSF trypanosomes at all temperatures a peak or plateau near τ ≈ 1 s. This suggests that, in all cases, the cell’s short-term zig-zag motion is averaged out beyond a critical timescale τc = 1 s, retaining only the long-range motility pattern. (C and D) The PDF of the local trajectory straightness p(S), defined in Equation 4, features marked changes for increasing temperatures in case of PCF trypanosomes but not for BSF. PCF cells show a bimodal PDF with a strong peak in the range S ≈ 0.8 and a successive increase of the fraction of values S < 1/2 with temperature. In contrast, p(S) for BSF trypanosomes is unimodal with a strong peak at S ≈ 0.15 throughout the tested temperature range. Gray-shaded regions indicate the range S < S0 with S0 = 0.36, which is consistent with purely random motion (see main text). Persistence and local straightness of PCF and BSF motility. (A and B) Characteristic VACF values C∗, defined in Equation 3, show for PCF and BSF trypanosomes at all temperatures a peak or plateau near τ ≈ 1 s. This suggests that, in all cases, the cell’s short-term zig-zag motion is averaged out beyond a critical timescale τc = 1 s, retaining only the long-range motility pattern. (C and D) The PDF of the local trajectory straightness p(S), defined in Equation 4, features marked changes for increasing temperatures in case of PCF trypanosomes but not for BSF. PCF cells show a bimodal PDF with a strong peak in the range S ≈ 0.8 and a successive increase of the fraction of values S < 1/2 with temperature. In contrast, p(S) for BSF trypanosomes is unimodal with a strong peak at S ≈ 0.15 throughout the tested temperature range. Gray-shaded regions indicate the range S < S0 with S0 = 0.36, which is consistent with purely random motion (see main text). Having fixed the critical timescale for all subsequent analyses, we next analyzed the local straightness of trajectories, defined for any track segment in the period from (j − n)Δt to (j + n)Δt within a single trajectory as(4)S=|rj+n−rj−n|∑k=12n|rj+k−n−rj+k−1−n|.The straightness S measures the end-to-end distance versus the taken path length within a period (2n + 1)Δt and hence quantifies how directed a trajectory segment is. Random motion is associated with small straightness values (S → 0 for n → ∞), whereas a directed, ballistic path yields S → 1. In our case, we have chosen a total window of 3τc, i.e., at any position j in a trajectory, the straightness was evaluated between the preceding and successional 15 positions. For this finite value of n, a standard Brownian random walk process yields an average straightness ⟨S⟩ ≈ 0.18 (standard deviation σ ≈ 0.09), suggesting that values S < S0 with S0 = ⟨S⟩ + 2σ = 0.36 indicate a purely random motion. As a result, we observed that the probability distribution function (PDF) of straightness values, p(S), featured a bimodal form for PCF cells with strong contributions around S > 0.8 at all temperatures (Figure 2C). This indicates a considerable fraction of trajectory segments with a straight and ballistic signature, in line with the visual impression from Figure 1A. For increasing temperature, the more erratic fraction with low S values in the range S < S0 increased; i.e., PCF trypanosomes appear to move less often in a directed fashion at high temperatures. Quantifying this observation, the fraction of straightness values with S < 1/2 was seen to increase from about 35% at 24°C to almost 50% at 40°C. In contrast to this finding, BSF trypanosomes featured fairly low values for S with a single distinct peak near to S = 0.15, irrespective of the temperature (Figure 2D). In fact, the majority of straightness values was seen in the range S < S0, consistent with a purely random motion. This indicates that trajectories of BSF cells contain few directed segments and rather move in a more randomized fashion, in agreement with the visual impression of tracks (cf. Figure 1B). In line with this, the fraction of straightness values with S < 1/2 was seen to only fluctuate randomly around 67% ± 3% over the tested temperature range. The bimodal form of p(S) at all temperatures suggests that trajectories of PCF trypanosomes can be dissected into segments with high and low straightness, corresponding to run (S > 1/2) and tumble (S < 1/2) phases. With this definition, run phases of PCF trypanosomes exhibit an almost straight trajectory segment, whereas a crumbled random motion within a narrow region is observed during tumble periods (cf. example trajectories in Figure 1A). Such an automatic dissection has been successful and in meaningful agreement with the visual impression of trajectories at the standard cultivation temperature of 27°C for PCF cells,23 suggesting that it is also a useful approach at all other temperatures studied here. In contrast, the unimodal shape of p(S) for BSF trypanosomes with its distinct peak in the range 0.1 ≤ S ≤ 0.2 indicates that there are no discernable states in the trajectories, in line with the visual impression of example trajectories (cf. Figure 1B). Indeed, a formal dissection into two distinct states with S < 1/2 and S > 1/2 resulted in very short and randomly dispersed run segments within trajectories that could not be distinguished by eye from neighboring segments that were classified as being in the tumble state (cf. also example trajectories in Figure 1 that do not exhibit appreciable periods of ballistic motion). We therefore concluded that a meaningful segregation of trajectories into distinct run and tumble states is not meaningful for BSF cells. Instead, we rather focus on a deeper analysis of PCF trypanosomes in the following subsection. How can the observed differences in motility between PCF and BSF trypanosomes be rationalized? A differential regulation of the microtubule cytoskeleton, including both the subpellicular cytoskeleton and the flagellar axoneme, might come to mind immediately. Since trypanosomes do not express variant tubulin isotypes, differential regulation could be achieved by post-translational modification and/or the differential expression of cytoskeleton-associated proteins. The latter has indeed been described, and some abundant proteins are differentially expressed in PCF and BSF.31,32,33,34 However, it is unknown whether these differentially expressed proteins convey differential properties to microtubules or even alter the flagellar activity. Another rationale could invoke the very different environments that BSF and PCF trypanosomes are residing in. The development of the parasite in the tsetse fly is characterized by a complex and pre-determined migration through various internal compartments, lasting 2–3 weeks.35,36 Although specific modes of motility during various stages of development have not yet been investigated in detail, it appears advantageous that PCF trypanosomes use a directional motility to be able to advance along their life cycle development and to find the appropriate next niche in the fly at varying ambient temperatures. BSF Trypanosoma in the mammalian host may not have a comparable necessity to actively explore their environment in order to progress with differentiation, especially when being transported along the vascular system in the host’s bloodstream to numerous other habitats like the skin, adipose tissues and, responsible for the lethality of the disease, the brain.37,38,39 Still, motility of the parasite is critical for virulence (also for BSF cells). Trypanosomes with impaired flagellar function are unable to maintain an infection in mice40,41 and immune evasion in the mammalian host requires cellular motility to recycle the cell’s variant surface glycoprotein (VSG) surface coat.16 It is, however, unclear, whether facilitating an exchange of VSG on the cell surface necessitates a directed motility with a constant shear flow on the plasma membrane. Even when BSF trypanosomes move rapidly but in a non-directed fashion, it is conceivable that the swiftly changing hydrodynamic flows are sufficient to massively enhance the diffusion of VSG on the plasma membrane, hence increasing their chance to get endocytosed in the flagellar pocket. Thus, the immune evasion remained possible, even without directed motility. PCF trypanosomes feature a robust run-and-tumble motility As a first step to explore PCF motility in more detail, we have used the automatic classification of trajectories into run-and-tumble states at any point in time t to dissect the PDF p(v) of local velocities, v = |r(t + τc) − r(t)|/τc, into two complementary contributions: By assigning each velocity the straightness-derived state, we obtained PDFs for run and tumble velocities, vR and vT, whose sum is the undissected PDF of all velocities (after proper normalization with the weight of the sub-ensembles). Two examples for this dissection, at different temperatures, are shown in Figures 3A and 3B. As expected from the EA-TA-MSD data, velocities in the run state were higher than in the tumble state. The PDF for velocities in the tumble state, p(vT), featured a single peak around v = 1μm/s at all temperatures, whereas the complementing PDF in the run state, p(vR), covered much higher velocities. In fact, not only the mean but also the width of p(vR) was seen to increase with temperature (cf. Figures 3A and 3B), indicating that run phases of PCF trypanosomes feature, on average, a higher velocity when increasing the temperature but with a larger degree of variability.Figure 3Velocity distribution and Arrhenius behavior of PCF cells. (A and B) Using the local straightness, the PDF of velocities along PCF trajectories, p(v) (black stair histogram), can be dissected into complementary contributions from the run-and-tumble states (colored stair and bar histograms, representing p(vR) and p(vT)). When increasing the temperature, p(vR) extends to higher velocities and becomes markedly broader while p(vT) remains virtually unchanged. (C) Arrhenius plot of the average run velocity, ⟨vR⟩, and the standard deviation of velocities, σ(vR). The relaxation rate ν with which the characteristic VACF value C∗ grows to its maximum (cf. Figure 2A) also shows an Arrhenius-like behavior (red symbols, see main text for details). Lines are a best fit according to an Arrhenius law (see main text for details). Velocity distribution and Arrhenius behavior of PCF cells. (A and B) Using the local straightness, the PDF of velocities along PCF trajectories, p(v) (black stair histogram), can be dissected into complementary contributions from the run-and-tumble states (colored stair and bar histograms, representing p(vR) and p(vT)). When increasing the temperature, p(vR) extends to higher velocities and becomes markedly broader while p(vT) remains virtually unchanged. (C) Arrhenius plot of the average run velocity, ⟨vR⟩, and the standard deviation of velocities, σ(vR). The relaxation rate ν with which the characteristic VACF value C∗ grows to its maximum (cf. Figure 2A) also shows an Arrhenius-like behavior (red symbols, see main text for details). Lines are a best fit according to an Arrhenius law (see main text for details). Given that the PDF p(vR) is fairly broad for all temperatures, we explored to what extent this is due to cell-to-cell variations (i.e., whether PCF trypanosomes display an individualized propulsion behavior). To this end, we extracted run-state segments with a length of 200–300 positions from individual trajectories (several hundred trajectories per temperature, each contributing one segment) and asked via a two-sample Kolmogorov-Smirnov test (at the 5% level) whether the pairwise comparison of these sets of run velocities were significantly different. As a result, only 2%–3% of pairs were rated as significantly different, indicating that the velocity heterogeneity within single trajectories is comparable for all cells. The width of p(vR) is therefore not just determined by cell-to-cell variations but rather emerges from velocity variations within each trajectory. Plotting the mean velocity in the run state as a function of the inverse absolute temperature revealed an Arrhenius scaling ⟨vR⟩∼ exp{−Ea/(kBT)} with an activation energy Ea ≈ 18kBT where kBT is the thermal energy at T = 300 K (Figure 3C). Interestingly, the standard deviation of p(vR) was also seen to follow an Arrhenius scaling, σ(vR)∼exp{−Ea′/(kBT)}, with a slightly larger activation energy (Ea′≈22kBT). While the general Arrhenius scaling of σ(vR) can be understood when considering variations of the velocity to linear order via the total differential, σ(vR) ∼ dvR ∼ exp{−Ea/(kBT)}, the somewhat higher activation energy Ea′>Ea indicates additional effects that are not included in this simple linear approximation. While the observed Arrhenius scaling of ⟨vR⟩ may come as a surprise at first glance, it is actually in agreement with earlier observations on sperm motility, where a temperature-sensitive calcium channel enhances the flagellar beating frequency.42 An Arrhenius scaling was also found for the beating frequency of cilia.43,44 This suggests that the enhanced propulsion of PCF trypanosomes could be due to an increased flagellar beating frequency as well. While our spatiotemporal resolution is insufficient to quantify the actual deformation of the trypanosome flagellum (for which high-speed differential interference contrast (DIC) microscopy at the optical resolution limit revealed complex wave patterns at a frequency of about 20 Hz45), we can still obtain an indirect measure on the flagellar activity: the increase of the characteristic VACF value C∗ with the period τ (Figure 2A) basically describes how rapidly the cell’s short-term zig-zag motion (due to the beating of the flagellum) is averaged out. An elevated flagellar beating frequency will lead to a faster zig-zagging of the cell body, and hence the plateau value of C∗ should be reached for smaller τ. In line with this reasoning, the relaxation rate ν, describing the increase of the VACF value in the range τ ≤ 1 s via a simple exponential relaxation fit, C∗ = a + b[1 − exp(−ντ)], shows a clear Arrhenius-like temperature dependence (Figure 3C). This observation indicates that a faster flagellar beating is most likely responsible for the larger velocity of PCF trypanosomes at higher temperatures. Following earlier considerations,44 we speculate that the temperature-dependent beating frequency could be due to changes in the enzymatic activity of molecular motors (i.e., their ATP hydrolysis rate). However, since the flagellum is partially wrapped around the cell body, one may also need to consider changes in the effective amplitude of the beating. In fact, the velocity of a simplified polymer-like microswimmer that deforms sinusoidally with wavelength λ, featuring a beating frequency ω, and beating amplitude ϵ, is given by v ∼ϵ2ω/λ.46 Therefore, changes in λ and ϵ may also contribute to the faster swimming of PCF trypanosomes. As stated before, our data do not allow us to draw any conclusions on the (complex) bending waves that travel along the flagellum. However, the following line of reasoning provides an idea of how temperature could also alter the effective beating amplitude ϵ. The found activation energy Ea ≈ 18kBT is well in the range that has been reported for the Arrhenius scaling of the viscoelasticity of cross-linked polymer systems (30,160 kJ/mol, corresponding to 10–60 kBT).47,48 In these systems, the activation energy encodes the typical energy scale for re-arranging cross-linkers between stable polymer fibers. Given that the subpellicular cytoskeleton of trypanosomes consists of biochemically stable microtubules that are cross-linked by microtubule-associated proteins, it is tempting to assume that also in this case a typical activation energy for the re-arrangement exists. If so, an increasing temperature resulted in a softening of the subpellicular cytoskeleton, hence decreasing the cell body’s elasticity. A decreased elastic modulus of the trypanosome cell body in turn has recently been shown by model simulations to result in an increased swimming velocity11 as the softening increases the effective beating amplitude ϵ. Thus, the experimentally observed increase of ⟨vR⟩ with temperature may partially also be a consequence of a successive softening of the subpellicular cytoskeleton of PCF trypanosomes. BSF trypanosomes most likely express a different set of cross-linking microtubule-associated proteins with different properties,31,32,33,34 and, as a consequence, the elastic properties of the subpellicular cytoskeleton may be almost independent of temperature in the tested range, resulting in the observed temperature-independent motility. As a next step, we determined the periods that PCF trypanosomes remained in the run or tumble state. To this end, we only considered periods τR and τT that lasted longer than τc and discarded those periods that were open-ended at the start/end of the trajectory. The resulting PDF of periods in the run phase, p(τR), showed only minor variations for increasing temperatures (Figure 4A). Only at 40°C (i.e., at the edge of the physiological range) did marked deviations emerge. In all other cases, the PDFs were well described by a Weibull distribution with a scale parameter of 4.5 s and a shape parameter of 0.85, indicating a heavy-tailed Weibull distribution. Please note that Weibull distributions are frequently used to describe the outcome of stochastic processes with a memory and/or a non-constant rate, even though there may not be a direct link to an actual physical process. The PDF of tumble periods, p(τT), showed highly similar results (Figure 4B); i.e., PCF trypanosomes remain for similar periods in the two modes of motion, irrespective of the temperature. As a consequence, the mean time spent in the run phase, ⟨τR⟩, and the mean time in the tumble state to re-orient the run direction, ⟨τT⟩, are basically temperature independent. Assuming the re-orientation to be completely random (i.e., every turning angle in [0, 2π] is equally probable), PCF trypanosomes move on timescales t ≫⟨τT⟩ + ⟨τR⟩ like an active Brownian particle with a typical step length Δr = ⟨vR⟩×⟨τR⟩ that is taken within a typical period T = ⟨τT⟩ + ⟨τR⟩. This suggests an effective asymptotic diffusion constant ∼Δr2/T whose temperature dependence is dominated by the experimentally observed changes of ⟨vR⟩. This simplified line of reasoning provides a qualitative explanation of the experimentally observed increase of D∞ with temperature (Figure 1E). While such an analysis would also be highly interesting for BSF trypanosomes, the lack of a meaningful dissection approach for separating distinct run-and-tumble phases prevented us from doing this.Figure 4Analysis of run and tumble periods of PCF cells. PDF of periods spent in the (A) run and (B) tumble phase for PCF trypanosomes (temperature color coded). Except for T = 40°C, all PDFs are very similar and are approximately captured by a heavy-tailed Weibull distribution (shape parameter α = 0.85, scale parameter w = 4.5 s; black dashed line). Analysis of run and tumble periods of PCF cells. PDF of periods spent in the (A) run and (B) tumble phase for PCF trypanosomes (temperature color coded). Except for T = 40°C, all PDFs are very similar and are approximately captured by a heavy-tailed Weibull distribution (shape parameter α = 0.85, scale parameter w = 4.5 s; black dashed line). Finally, we performed the same analyses for mutant PCF trypanosomes in which enzymes for the post-translational modifications of tubulin were depleted. Deleting such enzymes in T. brucei can alter the acetylation, (de)tyrosination, or poly-glutamylation of tubulin,49 potentially changing key properties of microtubules in the subpellicular cytoskeleton and/or the flagellum. In fact, altering post-translational modifications of tubulin has been shown, for example, to weaken the flagellar motility in algae,50 to alter the activity of axonemal dyneins in cilia,51 and to affect trafficking in primary cilia.52 These findings suggest that key microtubule structures in trypanosomes (e.g., the flagellum) may also be sensitive to such perturbations. In particular, we used mutants in which enzymes for the detyrosination (VASH) and polyglutamylation (TTLL1, TTLL4C, TTLL6B) of tubulins were knocked out. These mutant PCF trypanosomes have been characterized on the cell-biological level before in some detail22,23,24 and partially showed significant morphological changes and growth defects. For brevity, we only have compared data obtained for 25°C and 33°C, focusing on the observables vR, τR, and τT for which we have chosen a boxplot representation for a simpler comparison (Figure 5). Here, boxplots show the median (colored bullet point), while boxes indicate the interquartile range and whiskers denote the 1.5-fold interquartile range. To obtain a meaningful rating whether a change is statistically significant and potentially biologically relevant, we have probed the data for both temperatures by a two-sample Kolmogorov-Smirnov test (p = 0.01) and also calculated Cohen’s measure d for the effect strength. When the difference was rated statistically significant and d ≥ 30% (indicating an at least medium effect strength), the data pair was rated to be sufficiently different and to be biologically relevant (marked by asterisks in Figure 5).Figure 5Results for PCF trypanosome mutants at two temperatures. Data are shown as boxplots (see main text for the definition);median values, shown as blue and red bullet points, refer to data at 25°C and 33°C, respectively. (A) Except TTLL6B mutants, all trypanosomes show a significantly increased run velocity vR at the higher temperature that is also biologically relevant (indicated by an asterisk, cf. definition in the main text). (B and C) Periods spent in the (B) run and (C) tumble phase only show a significant change for τR in the VASH mutant. Altogether, the motility pattern of the wild-type PCF is robustly maintained in the mutants despite the applied changes in the tubulin code. See main text for details. Results for PCF trypanosome mutants at two temperatures. Data are shown as boxplots (see main text for the definition); median values, shown as blue and red bullet points, refer to data at 25°C and 33°C, respectively. (A) Except TTLL6B mutants, all trypanosomes show a significantly increased run velocity vR at the higher temperature that is also biologically relevant (indicated by an asterisk, cf. definition in the main text). (B and C) Periods spent in the (B) run and (C) tumble phase only show a significant change for τR in the VASH mutant. Altogether, the motility pattern of the wild-type PCF is robustly maintained in the mutants despite the applied changes in the tubulin code. See main text for details. As a result, we found that all PCF mutants except for TTLL6B show a biologically relevant increase of velocities in the run phase, vR at higher temperature (Figure 5A). These data suggest at first glance that the polyglutamylation of beta tubulin (via TTLL6B) has a stronger impact than affecting the polyglutamylation of alpha tubulin (via TTLL1, TTLL4C). However, since the degree of polyglutamylation reduction varies with the particular gene that is knocked out, a double knockout of TTLL1 and TTLL4C might eventually result in a similar effect to the single knockout of TTLL6B. Still, our data indicate that an enhanced motility is a fairly robust feature of PCF trypanosomes, even when altering the tubulin code. Similar to the results for the wild-type strain (Figure 4), periods spent in the run-and-tumble phases were almost unchanged at higher temperature except for τR in the VASH mutant (Figure 5B and 5C). Still, even in this case, the change appears quite subtle. Altogether, these data underline the robustness of the motility pattern found in PCF trypanosomes, even when the tubulin code is altered. Our data also indicate that the pronounced temperature dependence of the parasite’s motility is not bound to the presence of the tested enzymes. However, we have only deleted single genes for enzymes catalyzing post-translational modification, while effects are likely to be cumulative since T. brucei possess eight different enzymes involved in polyglutamylation. Conclusion In summary, we have shown here that PCF and BSF trypanosomes feature distinct motility patterns and are very different in their response to temperature changes. While PCF trypanosomes show a distinct run-and-tumble motility with an increasing swimming velocity for higher temperatures, BSF trypanosomes move considerably more random without distinguishable motility phases and little changes upon temperature variation. Supposedly, these differences reflect the varying approaches with which both trypanosome forms progress along their life cycle in strongly different environments (fly gut versus bloodstream and tissue). The remarkable stability of the PCF motility pattern after deleting genes of enzymes catalyzing post-translational modification of tubulins, despite morphological changes and growth defects, further highlights the robustness of trypanosomal microswimmers against significant perturbations. It is likely that PCF and BSF trypanosomes also have very different motility features when exposing them to constraints (e.g., when exploring soft and narrow mazes such as brain or skin tissue). Data availability The data supporting the findings are available from the corresponding author upon reasonable request. Acknowledgments Financial support by the 10.13039/501100001659DFG priority program Physics of Parasitism (ER692/3-1, ER692/3-2, WE4335/5-1, and WE4335/5-2) is gratefully acknowledged. H.W., L.S., and M.W. also acknowledge financial support by the 10.13039/501100001663VolkswagenStiftung (Az. 92738). Author contributions H.W. and L.S. performed live-cell microscopy experiments and analyzed the data. M.T. and L.B. curated and provided the different cells. K.E. and M.W. designed research, complemented the data analysis, and wrote the manuscript. Declaration of interests The authors declare no competing interests.