ABSTRACT
The alphaproteobacterium Magnetospirillum gryphiswaldense has the intriguing ability to navigate within magnetic fields, a behavior named magnetotaxis, governed by the formation of magnetosomes, intracellular membrane-enveloped crystals of magnetite. Magnetosomes are aligned in chains along the cell’s motility axis by a dedicated multipart cytoskeleton (“magnetoskeleton”); however, precise estimates of its significance for magnetotaxis have not been reported. Here, we estimated the alignment of strains deficient in various magnetoskeletal constituents by live-cell motility tracking within defined magnetic fields ranging from 50 μT (reflecting the geomagnetic field) up to 400 μT. Motility tracking revealed that ΔmamY and ΔmamK strains (which assemble mispositioned and fragmented chains, respectively) are partially impaired in magnetotaxis, with approximately equal contributions of both proteins. This impairment was reflected by a required magnetic field strength of 200 μT to achieve a similar degree of alignment as for the wild-type strain in a 50-μT magnetic field. In contrast, the ΔmamJ strain, which predominantly forms clusters of magnetosomes, was only weakly aligned under any of the tested field conditions and could barely be distinguished from a nonmagnetic mutant. Most findings were corroborated by a soft agar swimming assay to analyze magnetotaxis based on the degree of distortion of swim halos formed in magnetic fields. Motility tracking further revealed that swimming speeds of M. gryphiswaldense are highest within the field strength equaling the geomagnetic field. In conclusion, magnetic properties and intracellular positioning of magnetosomes by a dedicated magnetoskeleton are required and optimized for bacterial magnetotaxis and most efficient locomotion within the geomagnetic field.
IMPORTANCE In Magnetospirillum gryphiswaldense, magnetosomes are aligned in quasi-linear chains in a helical cell by a complex cytoskeletal network, including the actin-like MamK and adapter MamJ for magnetosome chain concatenation and segregation and MamY to position magnetosome chains along the shortest cellular axis of motility. Magnetosome chain positioning is assumed to be required for efficient magnetic navigation; however, the significance and contribution of all key constituents have not been quantified within defined and weak magnetic fields reflecting the geomagnetic field. Employing two different motility-based methods to consider the flagellum-mediated propulsion of cells, we depict individual benefits of all magnetoskeletal constituents for magnetotaxis. Whereas lack of mamJ resulted almost in an inability to align cells in weak magnetic fields, an approximately 4-fold-increased magnetic field strength was required to compensate for the loss of mamK or mamY. In summary, the magnetoskeleton and optimal positioning of magnetosome chains are required for efficient magnetotaxis.
INTRODUCTION
The spirillum-shaped alphaproteobacterium Magnetospirillum gryphiswaldense and related magnetotactic bacteria (MTB) perform a very specialized form of flagellum-driven motility, called magnetotaxis. Alignment of these organisms in the Earth’s magnetic field is caused by formation of intracellular organelles, the magnetosomes, consisting of a membrane-enveloped magnetic mineral core of magnetite or greigite (1). Magnetic navigation in M. gryphiswaldense results from a combination of the purely passive and physically based alignment in magnetic fields with a complex chemosensory system that controls the active flagellum-driven propulsion of the cell body (2, 3). This combined “magneto-aerotaxis” is believed to facilitate navigation within steep redox gradients, typically found in the sediment of aquatic habitats (4). In order to serve most efficiently as a compass and to ensure a sufficient alignment of cells, it is assumed that magnetosomes have to be concatenated into longer chains to build a magnetic dipole sufficiently strong to align the cell in the rather weak geomagnetic field (25 to 65 μT [5]). In M. gryphiswaldense, magnetosome chain assembly is mediated by attachment of individual magnetosomes to the actin-like MamK filament by the adaptor protein MamJ (6, 7). Furthermore, MamK is also required for the equipartitioning of magnetosome chains during cell division (8, 9). Deletion of mamK resulted in short, fragmented, and mispartitioned chains (6), whereas lack of mamJ caused clustered and agglomerated magnetosomes (7) (the phenotypes of the respective deletion strains are depicted in Fig. 1A). Recently, a further determinant named MamY was found to be required specifically for the positioning of magnetosome chains parallel to the motility axis of the cell (i.e., the direction of movement [4]) along regions of highest positive inner cell curvature (10). Deletion of mamY resulted in magnetosome chains located opposite to the geodetic motility axis (10) (as shown in Fig. 1A). Based on the identification of MamY as a chain-positioning determinant, it was concluded that magnetosome chains in M. gryphiswaldense are localized along the cell’s geodetic motility axis by a dedicated multipart magnetosome-specific cytoskeleton (i.e., the “magnetoskeleton,” requiring the combined action of MamK, MamJ, and MamY) for most efficient magnetic navigation (10).
Motility soft agar assay for magneto-aerotaxis. (A) Phenotypes (TEM micrographs) of the wild-type (WT), ΔmamY, ΔmamK, and ΔmamJ strains. In the wild-type strain, magnetosomes are aligned along the regions of highest inner positive cell curvature (geodetic axis). In the ΔmamY strain, magnetosome chains exhibit an improper localization at the opposing regions of negative inner curvature along the longitudinal cell axis. The phenotype of the ΔmamK strain is characterized by shorter and fragmented magnetosome chains (as evident in the micrograph), as well as ectopic localization of magnetosomes (not shown). In the ΔmamJ strain, magnetosomes form clusters. Localization patterns of magnetosomes are indicated by purple dashed lines. (B) Representative swim halos formed by the indicated strains after 2 days within a homogeneous 600-μT magnetic field (direction indicated by black arrow) are shown in panel i. Length-to-width ratios (vertical/horizontal diameter) of distorted ellipsoid swim halos are shown in panel ii. Bars and error bars depict the means ± SDs of four independent replicate experiments (performed in different weeks), each based on triplicate culture samples (considered here as subsamples). Therefore, per experiment, three precultures per strain were grown in parallel but separate culture vessels for at least three passages before inoculation into motility soft agar. Each data point (colored dots) depicts the subsample average of one experiment. The outcome of the individual experiments is shown in Fig. S1B in the supplemental material. Statistical comparison of subsample averages was performed against the wild-type strain by an ordinary one-way analysis of variance with Dunnett’s multiple-comparison posttest (**, P < 0.01; ****, P < 0.0001; ns, not significant [P ≥ 0.05]). Note that the observed differences were significant only for the ΔmamY, ΔmamJ, and ΔmamABop strains due to the higher standard deviations for the wild-type and ΔmamK strains. (C) For comparison, measurements of the magnetic response (Cmag) of the wild-type, ΔmamY, ΔmamK, and ΔmamJ strains are shown. Measurements were performed on living cells using a strong permanent magnet (∼100 mT). Note, the Cmag of a nonmagnetic strain (e.g., ΔmamABop strain) is zero (data not shown). In box plots, the bar indicates the median, the box indicates the interquartile range, and the whiskers indicate the 5th and 95th percentiles. The mean is shown as a purple plus sign. Colored dots represent data points above and below the 95th and 5th percentiles. N, number of estimates from independently grown cultures. Values were measured during culture collection handling in the lab over a time period of 2.5 years. Statistical comparison was performed against the wild type by an ordinary one-way analysis of variance with Dunnett’s multiple-comparison posttest (****, P < 0.0001; ns, not significant [P ≥ 0.05]).
Despite the detailed knowledge of the molecular determinants and mechanisms involved in magnetosome chain formation and positioning, the significance of the magnetoskeleton for magnetotaxis and the individual contributions of all key constituents in environmentally relevant magnetic fields have remained unstudied in a quantitative manner. Previous estimations of magnetic alignment (6, 7, 10, 11) were mostly based on the Cmag assay, an optical method based on the change in light scattering of the bacteria depending on cellular orientation (12). Cmag measurements are a sensitive, fast, and simple method to analyze magnetic alignment. However, for maximum sensitivity, these estimations were commonly performed in nonhomogeneous magnetic fields stronger than the geomagnetic field by multiple orders of magnitude. Moreover, Cmag measurements do not consider the putative disorientation of cells from the direction of the applied magnetic field caused by their flagellum-mediated propulsion. Therefore, previous Cmag-based estimates of strains lacking single constituents of the magnetoskeleton (6, 7, 10, 11) very likely do not accurately reflect the degree of alignment during magnetotaxis in magnetic fields within the order of magnitude of the geomagnetic field. To address these and related questions in magnetotaxis, we adopted an aero/chemotaxis-based soft agar swimming assay in magnetic fields for the quantitative assessment of magnetic alignment. In addition, we constructed a specialized microscopic setup, equipped with three pairs of coils, which is suitable for the motility analysis of MTB under defined conditions at single-cell resolution. In comparison to our previous related system (2), this setup allows us to precisely apply defined homogenous magnetic fields along each axis of the microscope. Using both techniques (the soft agar swimming assays and live-cell motility tracking), we investigated how mutants displaying various defects in their magnetosome chain arrangement (i.e., ΔmamK, ΔmamY, and ΔmamJ) (Fig. 1A) perform with respect to their alignment within homogenous and defined magnetic fields. Our results indicate that although cells lacking mamK or mamY still form fragmented or mispositioned magnetosome chains (that result in a weaker and less efficient alignment), the presence of the full magnetoskeleton and optimal positioning of magnetosome chains within the cells provide a significant benefit for magnetic navigation. Furthermore, the two methods presented in our study demonstrate alternative strategies, in addition to the widely used Cmag assay, to investigate and accurately quantify the motility behavior of MTB and magnetic alignment.
RESULTS
Cells of M. gryphiswaldense form aero-/chemotactic swim halos in semisolid soft agar plates (2). Subsequent spreading from the site of inoculation occurs due to oxygen consumption of cells by respiration and to oxygen’s limited diffusion within the agar (as well as by metabolization of medium components, such as, e.g., nitrate as alternative electron acceptor for anaerobic respiration [2, 13]). In magnetic fields, distorted ellipsoid swim halos are formed when MTB swim aligned with the magnetic field lines (2, 14). The degree of distortion provides a direct visual readout and simple estimation of motility behavior and alignment. In order to assess quantitative information on the impact of individual magnetoskeletal constituents on alignment, we estimated the degree of distortion reflected by the length-to-width ratio of ellipsoid swim halos formed in a homogenous magnetic field of defined strength (600 μT) using a coil setup.
Using this assay, we found different levels of distortion of swim halos for the wild-type, ΔmamY, ΔmamK, and ΔmamJ strains (Fig. 1B, panel i). In the wild-type strain, the average length-to-width ratio (± standard deviation [SD]) was 3.96 ± 1.05. In comparison, the average length-to-width ratio of the distorted halo formed by the ΔmamY strain was significantly reduced (1.94 ± 0.21; ∼49% of the length-to-width ratio determined for the wild-type strain) (Fig. 1B, panel ii). We also observed a reduction in the length-to-width ratio for the ΔmamK strain (3.28 ± 1.03; ∼83% of the wild-type strain’s length-to-width ratio) (Fig. 1B, panel ii). The difference in comparison to the wild-type strain, however, was not significant (Fig. 1B, panel ii) due to the larger experiment-to-experiment variations in the length-to-width ratios for the wild-type and ΔmamK strains (see Fig. S1B in the supplemental material). In contrast, the ΔmamJ strain, which lacks an ordered chain-like arrangement of magnetosomes, did not show any detectable significant distortion after 2 days of incubation (length-to-width ratio of 1.07 ± 0.08). Accordingly, swim halos of the ΔmamJ strain were almost undistinguishable from those of a nonmagnetic mamAB operon deletion strain (ΔmamABop strain), which, as expected, formed circular swim halos (length-to-width ratio, 1.02 ± 0.04) due to its inability to align with the magnetic field (Fig. 1B). However, after prolonged incubation times (3 days), the very faint and outermost section of the swim halo formed by the ΔmamJ strain also became slightly distorted in the direction of the magnetic field (Fig. S1A), indicating that a small proportion of ΔmamJ cells in fact did align more strongly. This behavior was consistent with the observation of a few ΔmamJ cells by transmission electron microscopy (TEM) that exhibited very short and also distorted chain-like arrangements of magnetosomes (in addition to the commonly observed clusters) (Fig. S1C).
The outcome of the soft agar swimming assay, that the ΔmamK and ΔmamY strains are partially impaired in magnetic alignment, is also reflected by previous Cmag-based estimates based on formaldehyde-fixed nonmotile cells. Hence, previous studies have reported Cmag values corresponding to ∼75% of the wild-type strain’s Cmag for the ΔmamK strain (6) and ∼64% of the wild-type strain’s Cmag for the ΔmamY strain (10). For comparison, previous Cmag measurements also revealed a strong impairment in magnetic alignment for the ΔmamJ strain (ΔmamJ, Cmag ≈ 0.15, corresponding to ∼9% of the wild-type Cmag [11]; ΔmamABop, Cmag ≈ 0 [15]). In contrast, when magnetic alignment was reassessed in our study by measurement of Cmag using nonfixed motile cells (Fig. 1C), the Cmag assay did not reveal any significant differences between the wild-type strain and the ΔmamK or ΔmamY mutant. For comparison, the average (±SD) estimated Cmag values for the ΔmamY strain (1.12 ± 0.19) and ΔmamK strain (1.10 ± 0.21) were close to the value for the wild-type strain (1.17 ± 0.19). The ΔmamJ strain also exhibited an increased average Cmag of 0.75 ± 0.18 (∼64% of the wild-type Cmag), which is in a similar range as a previous estimate for living cells (Cmag ≈ 0.5 [11]). These results indicated that the strong permanent magnets (∼100 mT) for Cmag determination also forced a significant proportion of ΔmamJ cells to align when fixing of cells was omitted (as also observed previously [11]). In summary, the soft agar swimming assay combined with the possibility to apply a homogenous and weaker magnetic field provided a sensitive readout of magnetotactic behavior, which in comparison to the Cmag assay reveals differences in the degrees of magnetic alignment based on motile living cells.
In order to directly assess the navigational behavior on the single-cell level in homogenous magnetic fields of different strengths below the setting used for the soft agar swimming assay, we built a microscopic platform equipped with triaxial pairs of magnetic coils (Fig. 2A and B), with the objective to use this system also for future motility- and magnetotaxis-related studies. In comparison to our previous biaxial setup (2), this system allows the precise application of defined magnetic fields (by using a three-axis low-field magnetic microsensor) (Fig. 2C) not only along each axis of the focal plane but also along the optical axis. In consequence, this setup also has the capability to cancel out the geomagnetic field. Using this microscope, living cells were observed in dark-field illumination (within a zero field and in homogenous magnetic fields of 50 μT, 100 μT, 200 μT, and 400 μT), combined with the recording and subsequent analysis of their swimming trajectories (as demonstrated in Movies S1 and S2).
Microscopic setup used for motility tracking experiments. (A) In order to apply defined and homogeneous magnetic fields along each axis of the focal plane and the optical axis, the microscope was equipped with a custom-made 3D coil cage (i). Coils were controlled by three bipolar power supplies (ii), connected to a computer via a digital-to-analog interface (iii). (B) Close-up view of the coils. (C) An HMC5883L low-field magnetic microsensor used for calibration of the setup (connected to an Arduino Uno controller, iv in panel A). Coils and sample holder are outlined by purple and yellow dashed lines.
The resulting polar coordinate plots for the wild-type strain, magnetoskeleton mutants (ΔmamK, ΔmamY, and ΔmamJ strains), and a nonmagnetic ΔmamABop control strain are shown in Fig. 3. As already evident by visual inspection of the plots, the wild-type strain outperformed all other strains with respect to the degree of alignment under any of the tested field strengths, indicated by the higher number of swimming trajectories oriented close to the axis of the applied magnetic field with increasing field strengths. In order to provide a quantitative measure, we plotted the distribution of values obtained for the absolute cosines of the apparent heading angle θ of each track (Fig. 4B) (the corresponding statistical analysis is shown Table S1). The heading angle θ denotes the angle between the orientation of a cell (represented by the direction of the velocity vector of a swimming track) and the direction of the magnetic field (Fig. 4A). The population median of the absolute cos θ is ∼0.7 if the cells exhibit a random and nonbiased motion, and the value becomes equal to 1 if cells swim perfectly aligned within the magnetic field. An alternative graphing option is shown Fig. S2, by plotting the variance in sin θ (see Materials and Methods) as a function of the magnetic field strength. This value is a measure of the dispersion of cell orientations relative to the constant magnetic field (16). The variance in sin θ equals 0.5 if cells exhibit a random motion, and the value decreases to zero when swimming trajectories are parallel to the magnetic field lines.
Representative swimming trajectories of the wild-type (WT), ΔmamY, ΔmamK, and ΔmamJ strains and a nonmagnetic ΔmamABop strain. Trajectories were recorded in a zero field (canceled geomagnetic field) and in homogeneous magnetic fields of increasing strength (50 μT, 100 μT, 200 μT, and 400 μT). The direction of the applied magnetic fields corresponds to the horizontal axis of the polar coordinate plot. Angles of 0° and 180° correspond to magnetic north and south, respectively. The track length corresponds to the distance traveled (indicated below the graphs in the colors of the inner and outer rings of the polar coordinate plot).
Estimation of magnetic alignment by live-cell motility tracking. (A) The degree of alignment within the magnetic field (applied along the x axis) is denoted by the apparent heading angle θ for each track between the direction of the velocity vector (green arrow) and the axis of the magnetic field vector (blue arrow), when swimming trajectories are projected in the focal plane (represented by the x and y axes). In order to minimize movement along the optical axis into the z direction (and within the focal plane along the y axis), all geomagnetic field components were first zeroed out by using all three pairs of coils, followed by the application of different magnetic field strengths along the x axis using one coil pair. In addition, the number of focal planes was minimized by the sample preparation method (see Materials and Methods for details). Note, two different coordinate systems were used to describe cell orientations. The apparent heading angle θ, velocity vector (green arrow), and the magnetic field vector (blue arrow) refer to polar coordinates. The x, y, and z axes refer to a Cartesian coordinate system. (B) Estimated degree of alignment of the indicated strains within a zero field (canceled geomagnetic field) and in homogeneous magnetic fields of increasing strengths (50 μT, 100 μT, 200 μT, and 400 μT). In order to denote the degree of alignment within the magnetic field, the distribution of the absolute cosines of the heading angle θ obtained from individual swimming trajectories was plotted. The absolute cos θ was used to account for the fact that magnetospirilla can swim in both directions of the magnetic field. Accordingly, an alignment score of 1 indicates perfect alignment along the horizontal axis (x axis), whereas an alignment score of 0 indicates a perfect alignment along the vertical axis (y axis). The horizontal dashed line denotes a random unbiased directional movement (corresponding to a median alignment score of 0.707 and heading angles of 45°, 135°, 225°, and 315°). In box plots, the bar indicates the median, the box indicates the interquartile range, and the whiskers indicate the 5th and 95th percentiles. The mean is shown as a plus sign and depicts the population average of the absolute cos θ. Colored dots represent data points above and below the 95th and 5th percentiles. Data points were collected and pooled from ≥3 replicate experiments (performed on different days) per strain and field condition. Statistical analysis and the total number of analyzed swimming trajectories per strain and field condition are listed in Table S1 in the supplemental material.
The observed degrees of alignment for the wild-type strain (Fig. 4B) were approximately in accordance with the Langevin function (see Materials and Methods). This equation predicts that the average of cos θ approaches saturation (corresponding to a value of 1) with increasing field strength, as the disorientation influence of the randomizing thermal energy decreases (17). According to the Langevin equation, the degree of alignment should come close to the optimum within a magnetic field of ∼400 μT for the microoxically grown wild-type cells of M. gryphiswaldense used in our study (Fig. S3). This prediction is in good agreement with the experimentally determined population average for the absolute cosines of θ of 0.853 ± 0.249 (mean ± SD) and population median of 0.978 (which should provide a more robust measure) within the 400-μT magnetic field (Fig. 4B). As also evident from the data set obtained for the wild-type strain (Fig. 4B and Fig. S2), a magnetic field with a strength corresponding approximately to the geomagnetic field (50 μT) resulted in an increase of the population median for the absolute cos θ from 0.717 to 0.853 (and in a decrease of the variance in sin θ from 0.5 to 0.38), corresponding to a change in θ of ∼13° directed to the axis of the applied 50-μT magnetic field, in comparison to the angles obtained by random swimming patterns of the bacteria in a zero field.
To compare the individual strains, we focused on the percentage of swimming trajectories exhibiting angular deviations of ≤10° from the x axis (Fig. 4A), which is an arbitrary definition, specifying a range located very close to the axis of the applied magnetic field. In the case of the wild-type strain, the percentages of swimming trajectories that displayed angular deviations of ≤10° from the axis of the applied magnetic field were 19% for a field strength of 50 μT, 25% for 100 μT, 29% for 200 μT, and 43% for 400 μT, in contrast to only 11% of the trajectories for the zero field. In comparison, the ability to swim parallel to the magnetic field lines was impaired for the ΔmamK and ΔmamY strains. Consequently, the ΔmamK strain was nearly not aligned within the 50-μT magnetic field. Furthermore, the level of alignment determined for the ΔmamK and ΔmamY strains within the 200-μT magnetic field and under the highest tested field strength (400 μT) was similar to the degree of alignment observed for the wild-type strain within the 50-μT and 100-μT magnetic fields, respectively (Fig. 4B and Fig. S2). Accordingly, for the ΔmamK and ΔmamY strains, only 20% and 26% of the swimming trajectories, respectively, were located within the range of ≤10° from the axis of the applied 400-μT magnetic field. These estimates corresponded to 48% and 62% of the percentages of swimming trajectories within this range determined for the wild-type strain (43% at 400 μT), respectively. Of note, the percentages of swimming trajectories for the ΔmamK strain located within the range of ≤10° for the magnetic field strengths of 50 μT, 100 μT, and 200 μT corresponded to 13%, 14%, and 20%, respectively. Likewise, at similar field strengths, 15%, 20%, and 24% of the swimming trajectories exhibited angular deviations of ≤10° from the axis of the applied magnetic field in case of the ΔmamY strain. Consistent with the observations in the soft agar swimming assay (Fig. 1B), the ΔmamJ strain failed to align under almost all of the tested field conditions (Fig. 4B and Fig. S2). Although the population average and median of the absolute cos θ were slightly increased under a 100-μT, 200-μT, and 400-μT magnetic field for the ΔmamJ strain, according to our statistical analysis (Table S1), none of these data sets differed significantly from the data for the zero-field condition or the corresponding field conditions applied to a nonmagnetic ΔmamABop control strain. As noted above, few cells within the ΔmamJ population exhibited a quasi-chain-like arrangement of magnetosomes (as observed by TEM) (Fig. S1C), which might have resulted in the slight, but still nonsignificant, increase in the absolute cos θ compared to that of the zero field. As expected, none of the strains exhibited a significantly different level of alignment within the zero field in comparison to that of the nonmagnetic ΔmamABop control strain under all tested field strengths (Table S1). In summary, motility tracking allowed the precise quantification of the individual contributions of all magnetoskeleton constituents and confirmed that ΔmamK and ΔmamY cells are clearly impaired in their ability to swim parallel with the magnetic field, whereas ΔmamJ cells are nearly unable to align within the tested weak magnetic fields.
In addition to the analysis of alignment, we also determined swimming speeds (in a standard wet mount, reflecting uncontrolled oxic conditions) (Fig. 5). Swimming speeds were in similar ranges for the wild-type, ΔmamK, ΔmamY, and ΔmamABop strains (on average, 25 to 30 μm/s; ranging from as low as 6 μm/s up to ∼70 μm/s; summarized over all tested field strengths) and similar to previous estimates for M. gryphiswaldense (2, 3, 18, 19). Remarkably, in the case of the wild-type, ΔmamK, and ΔmamY strains, we observed that average swimming speeds were highest within the 50-μT field, reflecting the geomagnetic field condition (although data sets were not in all cases significantly different from those of the zero-field condition) (Table S1). For example, average speeds (± standard error of the mean [SEM]) for the wild-type strain increased from 27.3 ± 0.34 μm/s within the zero field to 28.9 ± 0.30 μm/s (50-μT magnetic field) and decreased again to 27.6 ± 0.30 μm/s, 25.9 ± 0.31 μm/s, and 25.6 ± 0.35 μm/s within the 100-μT, 200-μT, and 400-μT magnetic field, respectively. In comparison to the wild-type strain, the ΔmamK and ΔmamY mutants exhibited a more prominent consecutive drop in average swimming speeds when the field strength was increased from 50 μT to 100 μT. Furthermore, maximum swimming speeds in the 50-μT magnetic field were lower for the ΔmamY strain (60.8 μm/s) than for the wild-type strain (71.8 μm/s) and ΔmamK strain (68 μm/s). This trend was not observed for the nonmagnetic ΔmamABop control strain and the ΔmamJ mutant, arguing against a general metabolic fatigue of the cells (17) during subsequent video recordings (which were conducted starting with the zero field followed by increasing field strengths during each experiment within a time span of only a few minutes).
For unknown reasons, the ΔmamJ strain displayed a considerably lower average swimming speed (in the range of 14 to 16 μm/s) (Fig. 5). In contrast, the efficiency of spreading in soft agar was not obviously affected as the sizes of the swim halos formed by the ΔmamJ strain were reproducibly of diameters similar to the swim halos formed by the fast-swimming ΔmamABop strain (Fig. 1B, panel i, and Fig. S1A) and to those of the other investigated strains if plates were incubated in the absence of an external magnetic field (data not shown). This behavior was persistent over several replicate experiments even if special care was taken to maintain cells under conditions that allowed maximum motility. Additional experiments based on the wild-type strain and the motility tracking analysis of cells withdrawn directly from different locations in soft agar plates indicated that cells with elevated swimming speed and magnetic alignment can be enriched using the soft agar swimming assay (Fig. S4). Hence, individual wild-type cell populations directly derived from the dense sections (Fig. S4A) of the distorted swim halo close to the magnetic poles exhibited increased magnetic alignment (average ± SD for the absolute cosines of θ up to 0.921 ± 0.181 at 400 μT) and higher average swimming speeds (up to 43.4 ± 0.33 μm/s [± SEM] at 400 μT) compared to the estimates based on batch cultures grown in liquid medium under microoxic conditions (Fig. 4 and 5). When this method was applied to the ΔmamJ strain (Fig. S5), average swimming speeds increased (up to 31.8 ± 0.79 [± SEM] at 400 μT); however, cell populations were still mostly lacking cells reaching swimming speeds above 50 μm/s. In addition, these supplementary experiments (Fig. S5) indicated that the faint and slightly distorted outermost parts of the swim halo observed for the ΔmamJ strain after prolonged incubation times (Fig. S1A) are apparently not caused by a permanently increased magnetic dipole moment of the cells located within these regions and, therefore, are likely not the consequence of a genetic alteration or suppressor mutation that compensates for the loss of mamJ. Currently, we cannot explain if the decreased swimming speeds of the ΔmamJ strain are of genetic nature (i.e., a second site mutation in a motility related gene) or reversible. Related observations in other bacteria, e.g., have also been attributed to flagellar motor torque-related control mechanisms (20). In summary, however, the major finding of the swimming speed analysis was the observation that the magnetic properties of the M. gryphiswaldense wild-type strain appear to be optimized to achieve maximum motility within the geomagnetic field regime.
Estimation of swimming speeds by motility tracking. Box plots depict swimming speed distributions of the indicated strains within a zero field (canceled geomagnetic field) and in homogeneous magnetic fields of increasing strengths (50 μT, 100 μT, 200 μT, and 400 μT). In box plots, the bar indicates the median, the box indicates the interquartile range, and the whiskers indicate the 5th and 95th percentiles. The mean is shown as a plus sign. Colored dots represent data points above and below the 95th and 5th percentiles. Each point depicts the average swimming speed of a single swimming trajectory. Data points were collected and pooled from ≥3 replicate experiments (performed on different days) per strain and field condition. Statistical analysis and the total number of analyzed swimming trajectories per strain and field condition are shown in Table S1 in the supplemental material.
DISCUSSION
The cytoskeletal network for magnetosome chain formation (i.e., magnetoskeleton) is complex; however, its significance and the individual contributions of all key proteins for magnetotaxis had not been quantified yet in defined and homogenous magnetic fields within the order of magnitude of the geomagnetic field. Here, we employed a soft agar swimming assay and live-cell motility tracking to depict the individual contributions of all magnetoskeleton key constituents. Previous studies (6, 7, 10, 11) were mostly based on the optical Cmag assay (12), using formaldehyde-fixed dead cells and strong permanent magnets (∼100 mT), which deliver nonhomogeneous magnetic fields that have strength that is ∼2,000 times larger than the Earth’s magnetic field. Hence, although the Cmag assay provides a time-efficient and streamlined procedure to analyze magnetic alignment, this assay improperly reflects the degree of alignment when cells perform magnetotaxis within the geomagnetic field. Although some limitations of the Cmag assay can be overcome (e.g., by implementation of appropriate coil systems within spectrophotometer devices [21–23]), the two methods presented in our study are based on the active flagellum-mediated movement of the cells within defined magnetic fields and therefore are more representative to describe the real benefit of the magnetoskeleton for magnetotaxis.
The soft agar swimming assay provides a methodologically rather simple and sensitive approach to obtain a direct visual readout of motility behavior based on the magnetic alignment and aerotactic response of the cells. While offering the option for quantification by estimation of dimensionless length-to-width ratios, this assay might become time-consuming if multiple field conditions have to be tested. Moreover, swim halos can be more difficult to interpret since they are a complex phenotype, based not only on motility and chemo-/aerotaxis but also on growth (i.e., respiration and medium consumption), often resulting in the formation of several superimposed ring- or aerotactic dome-shaped structures (24). In contrast, our motility tracking setup requires a complex and experimentally demanding instrumentation but also allows for comparison of motility parameters (e.g., alignment or swimming speed) in a very quantitative and reproducible manner based on single-cell data under environmentally relevant magnetic field conditions. In addition, motility tracking offers the possibility to further adjust the setup for use in future studies to investigate the response of MTB to different (chemotactic) stimuli such as, e.g., oxygen or light. It should be noted that the cells’ observed behavior under the microscope is still likely somewhat different from that in natural aquatic habitats; e.g., environmental sediments are often complex and highly structured (25), including possible local geomagnetic fields (26). Furthermore, it should be pointed out that although the observed degrees of alignment for the wild-type strain in our study (Fig. 4; see also Fig. S2 in the supplemental material) were roughly in accordance with the Langevin prediction (Fig. S3), deviations of the cells’ observed behavior can occur since this equation is valid only for a population of independent noninteracting particles with identical magnetic properties (27). These requirements are not fully met due to the inherent biological variation of a bacterial population. In addition, the orientation of cells can be further affected by factors other than their magnetic moment and thermals, such as, e.g., collision of individual cells, flagellar movement, including conformational changes and polymorphic transitions of flagella during directional reversals (25, 28), liquid perturbations, and hydrodynamic forces close to the coverslip (29).
Our estimations based both on the soft agar swimming assay (Fig. 1B and Fig. S1A and B) and on the cells’ swimming trajectories (Fig. 4B and Fig. S2) agree in the finding that all strains lacking magnetoskeleton key determinants (ΔmamK, ΔmamY, and ΔmamJ strains) are impaired to align within the tested weak-field regime. Magnetic alignment of the ΔmamK strain based on a soft agar swimming assay has been investigated before by Sakaguchi et al. (14). In this study, it was found that the length of the wedge-shaped swim halo of the Magnetospirillum magneticum AMB-1 ΔmamK strain was approximately 70% of the halo formed by the AMB-1 wild-type strain, using a less controlled version of the assay conducted in our study (with a neodymium magnet placed next to the agar plate). However, since AMB-1 encodes a second MamK-like paralog (30, 31) and since it has a different chain configuration than M. gryphiswaldense (6), both ΔmamK strains are not fully comparable due to different phenotypes (shorter and fragmented ectopic chains in M. gryphiswaldense [6] and small groups of 2 to 3 dispersed magnetosomes in AMB-1 [32]).
Whereas the results of the soft agar swimming assay (based on a field strength of 600 μT) (Fig. 1B and Fig. S1A and B) pointed toward a somewhat more important role of MamY in M. gryphiswaldense than of MamK (with respect to magnetic alignment), according to our motility tracking data (corresponding to field strengths in the range of 50 to 400 μT) (Fig. 4B and Fig. S2), the impairment in magnetotaxis caused by deletion of mamK or mamY is reflected by the approximately equal contributions of both proteins. Accordingly, to achieve the same degree of alignment as the wild type in a 50-μT and 100-μT magnetic field, an approximately 4-fold-larger magnetic field strength (∼200 μT and ∼400 μT, respectively) was required to compensate for the loss of either mamY or mamK (Fig. 4B and Fig. S2). Decreased efficiency of magnetotaxis in the case of the ΔmamK strain can likely be attributed to the fact that the higher proportion of cells with shorter, fragmented, and ectopic magnetosome chains within the population (6) results in a lower overall magnetic dipole moment, whereas ΔmamY cells presumably continuously deviate from the axis of the applied magnetic field, due to the misalignment of the magnetosome chain with the cell body axis (10). In contrast, the ΔmamJ strain displayed an almost complete inability to align (Fig. 1B and 4B; also Fig. S1A and B and S2) due to its inability to form coherent magnetosome chains (7).
In addition to the degree of alignment, we also analyzed the dependency of swimming speed on the magnetic field strength (Fig. 5). Our observation that swimming speeds of the wild-type, ΔmamK, and ΔmamY strains (corresponding to wild-type-like, rudimentary and fragmented, or mispositioned magnetosome chains, respectively) are highest within the 50-μT field is in contrast to the classical model of magnetotaxis. This model predicts that the migration velocity along the magnetic field lines increases monotonically with increasing magnetic field strength since the disorientation influence of the randomizing thermal energy decreases (17, 27). A relation between magnetic field strength and swimming speed has been previously observed in the related magnetospirillum M. magneticum AMB-1 (33). However, the effects of magnetic fields related to the geomagnetic field on swimming speed probably had been overlooked in previous magnetospirillum-related studies since field strengths much higher than 50 μT were used (2, 18, 26, 33, 34), or the increments between tested field strengths were larger with a focus based on the relation of swimming speeds to the oxygen gradient (3, 19). Our observations indicate that the magnetic properties of M. gryphiswaldense are optimized to achieve maximum motility within the geomagnetic field. They are in agreement with results by Pan et al. (17), who observed that the field-parallel migration velocity of magnetotactic cocci decreases with increasing field strengths from 0.1 to 0.5 mT. The authors attributed these observations to the fact that in cocci the magnetosome chain is inclined with respect to the flagellar propulsion axis, which becomes more relevant in field strengths above the geomagnetic field (17). While in magnetotactic cocci chains are obviously not perfectly aligned along the motility axis of the cell, resulting in helical trajectories (17, 35), a misalignment or mispositioning of the magnetosome chain with respect to the cell body axis does also apply for the ΔmamY and ΔmamK strains (Fig. 1A). As a consequence, the altered chain configuration in both strains might explain the more prominent increase and consecutive drop in swimming speeds with increasing field strengths in comparison to those of the wild-type strain (Fig. 5). Also, in comparison to the wild-type and ΔmamK strains, the ΔmamY strain did achieve lower maximum swimming speeds within the applied magnetic fields. This observation might be attributed to the misalignment of the magnetosome chain with the cell body axis in the ΔmamY strain (10), resulting in larger deviations from the axis of the applied magnetic field. The slight decrease in average swimming speeds for the wild-type strain above 50 μT indicates that even in wild-type cells magnetosomes are not always perfectly aligned with the cell body axis, in agreement with observations of a recent study in M. magneticum AMB-1 (36).
Apart from a relation based on the merely passive alignment of cells in magnetic fields (i.e., as a compass needle), an apparent increase and drop in swimming speeds might also be explained by a putative magnetoreceptive mechanism (26, 33, 37). Such a mechanism that employs active sensing of the magnetic field strength had been suggested in M. magneticum (26, 33). The postulated model was based on a direct interaction of MamK with a methyl-accepting chemotaxis protein (MCP) sensor protein (38) to translate torque exerted on the magnetic dipole into concomitant active flagellum-mediated responses. Given the lack of experimental evidence for such an interaction in M. gryphiswaldense, the relevance of a magnetoreceptive mechanism in M. gryphiswaldense remains obscure. The observation that a subsequent shift from a zero field to a 50-μT magnetic field resulted in no significant increase in alignment but elevated swimming speeds for the ΔmamK strain (Fig. 4B and 5) is an indication for a putative mamK-independent magnetoreception in M. gryphiswaldense.
In conclusion, our results indicate not only that the presence of all key determinants of the sophisticated magnetoskeleton cytoskeletal network and optimal positioning of magnetosome chains along the longitudinal motility axis of the cell provide a clear benefit for magnetotaxis but also that the magnetic properties of M. gryphiswaldense are optimized to achieve maximum motility for navigation in the weak geomagnetic field. Moreover, the methods presented in our study represent alternative strategies to the Cmag assay to analyze and precisely quantify magnetic alignment and also motility behavior, which can be further combined to enrich and analyze distinct cell fractions with specific motility properties.
MATERIALS AND METHODS
Strains and growth conditions.All M. gryphiswaldense strains used during this study are listed in Table 1. Strains were grown in flask standard medium (FSM) (39) under microoxic conditions as described in detail below. Optical density (OD) and magnetic response (Cmag) of M. gryphiswaldense cultures were estimated photometrically at 565 nm as reported previously (12).
Strains used in the study
Soft agar swimming assay.For the preparation of FSM (39) swim agar plates, 0.2% (wt/vol) agar was used, and the concentration of potassium lactate (carbon source) was lowered to 1.5 mM. Forty milliliters of soft agar was poured into a petri dish (8.5-cm diameter). In order to achieve reproducible conditions and to minimize growth (or oxygen)-dependent influences on magnetosome formation, strains were grown in six-well plates (in FSM [39], within a culture volume of 3.5 ml) for several passages prior to inoculation into soft agar under defined atmospheric conditions (2% headspace oxygen, 28°C) in a microoxic incubator (Scholzen Microbiology Systems AG), which provides a much larger headspace-to-culture volume ratio than tubes. Using these growth conditions, we observed no obvious correlation between the culture OD at 565 nm (OD565) and the length-to-width ratio of the distorted halos formed. To use similar amounts of cells for the experiments, cultures (corresponding to an OD565 between 0.1 and 0.3) were diluted to an OD565 of 0.1, and 5 μl of diluted cell suspension was inoculated into soft agar swim plates. Swim plates were incubated at 28°C for 3 days under atmospheric conditions (21% headspace oxygen) within a homogenous magnetic field (600 μT), using a coil setup described previously (2, 40). Plates were documented after 2 and 3 days with a ChemiDoc XRS+ imager (Bio-Rad) using Image Lab, version 5.2.1, software (1.5-s exposure time without additional filter). Swim halo diameters were determined in ImageJ Fiji (41). Estimation of the length-to-width ratio was based on the outermost swim halo formed after 2 days. For the tracking analysis of cells collected directly from soft agar plates (Fig. S4 and S5), larger petri dishes with a diameter of 14 cm (filled with 120 ml of soft agar) were used, and plates were documented after 3 to 4 days of incubation.
Motility tracking.For motility tracking experiments, small inocula were used (starting OD565 of ∼0.0005 to 0.001), and cells were grown only to an OD565 of ∼0.05 in FSM (39) for several passages in six-well plates (28°C, 2% headspace oxygen) to obtain a high number of actively dividing and highly motile cells within the culture (2). Motility tracking experiments were performed in dark-field illumination with an FN1 Eclipse upright microscope (Nikon) equipped with an S Plan Fluor 20× differential inference contrast (DIC) N1 objective (numerical aperture [NA], 0.5), a dark-field condenser (NA, 0.95), and a pco.edge 4.2 sCMOS camera (PCO). The FN1 microscope is especially suited for sensitive magnetic field applications since it was developed to provide a minimum of electronic and vibrational noise, as well as sufficient space to equip the microscope with coils.
In order to apply defined magnetic fields, a three-dimensional (3D) printed coil cage consisting of three pairs of coils positioned along the optical axis (z axis) and both axes of the focal plane (x and y axis) was custom manufactured by the company Claricent (Stephan Eder). Coils were designed to provide a homogenous magnetic field with less than 5% deviation in magnetic field strength within ±1.6 cm from the center of the cage (manufacturer’s specifications). Additionally, due to the fixed-stage design of the microscope, the sample always remained centered within the magnetic field. To obtain stable imaging conditions and to avoid magnetization of the surroundings, the microscope was placed onto an active pneumatic vibration isolation stainless steel table (Supertech Kft International GmbH). Coils were voltage controlled by using two BOP 50-2 M bipolar power supplies (Kepco) and a BPA-25-2 bipolar power amplifier (Claricent). Coil pairs mounted along the x and y axes of the focal plane were computer controlled within NIS-Elements, version 5.1, software (Nikon) by connection of the power supply voltage control inputs to the analog outputs of a USB-6008 NI DAQ Digital-to-Analog convertor (National Instruments). Calibration of the coils was performed with the objective in focus position, using a factory-calibrated GM08 gauss meter equipped with a high-sensitivity transverse probe, TP002HS (Hirst Magnetic Instruments), which was zeroed within a zero gauss chamber, and with a three-axis HMC5883L low-field magnetic microsensor (Honeywell) connected to an Uno R3 microcontroller board (Arduino). Calibration and zeroing of the HMC5883L chip were performed by determining the offset readings of each axis with Magneto, version 1.2, software (https://sites.google.com/site/sailboatinstruments1/home) by rotation of the sensor in a reference magnetic field. Offset values were implemented within the unified HMC5883L Arduino library reference code (https://github.com/adafruit/Adafruit_HMC5883_Unified). Readings of the calibrated HMC5883L sensor agreed with measurements of the factory-calibrated HIRST probe within the range of observed magnetic field strengths.
Before the experiment, cells were diluted to an OD565 of 0.01 to allow simultaneous recording of sufficient numbers of swimming trajectories while minimizing collision of individual cells. In order to minimize the number of focal planes and to obtain stable imaging conditions without sample drift and evaporation, 10 μl of cell suspension was placed under a 22- by 22-mm coverslip, which was fixed to the slide by a thin square frame of vacuum grease. To reduce dust particles and other debris within the dark-field illumination, coverslips were cleaned two times by ultrasonification for 15 min in 2% (vol/vol) Hellmanex III (Sigma-Aldrich), followed by two subsequent washes in Millipore H2O (ultrasonification for 15 min each). Coverslips were dried with compressed air before use. Movies were recorded for 40 s at 24°C with a frame rate of 25 frames per s (fps). The camera region of interest (ROI) was set to 1,500 by 1,500 pixels. Tracking was performed with NIS-Elements, version 5.1 (Nikon), employing the spot detection algorithm and random motion model, with the standard deviation multiplication factor set to 2.5, maximum gap size set to 3, and maximum object speed set to 80 μm/s. Tracks with 30 or fewer frames and line speeds lower than 5 μm/s were excluded from the analysis (representing mostly nonmotile cells that moved due to Brownian motion). Cells which were affected or tracked due to liquid perturbations or cells that were colliding or became stuck were manually excluded from the analysis. Values given in the paper are defined as follows. Line speed is the length of a straight line from the track origin to the current point (line length) divided by the time elapsed. Swimming speed is the track segment length divided by the amount of time elapsed between two positions. Population averages of swimming speeds were calculated based on the average swimming speeds of individual tracks. Heading is the apparent angle θ for each track between the direction of the velocity vector and the axis of the magnetic field vector when trajectories are projected in the focal plane (16). The Langevin function (27, 42) is defined as:
Statistical tests.Statistical tests were conducted with Prism, version 7.04 (GraphPad), as described in the respective legend to each figure. Data sets were tested for normality using the D’Agostino and Pearson, Shapiro-Wilk, and Kolmogorov-Smirnov tests.
ACKNOWLEDGMENTS
This work was supported by the Deutsche Forschungsgemeinschaft (grant Schu1080/9-2 to D.S.) and the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (grant agreement no. 692637 to D.S.).
We thank Julia Schmiedel, Agata Käsbohrer, and Katharina Silbermann for technical assistance during some experiments. In addition, we are thankful to Damien Faivre and Klaas Bente (MPI Potsdam) for helpful hints regarding the construction of the motility-tracking microscope.
FOOTNOTES
- Received 27 August 2019.
- Accepted 8 November 2019.
- Accepted manuscript posted online 15 November 2019.
Supplemental material is available online only.
- Copyright © 2020 American Society for Microbiology.
