Long term observation of Magnetospirillum gryphiswaldense in a microfluidic channel

TL;DR

This study monitored individual Magnetospirillum gryphiswaldense bacteria in a microfluidic channel over four hours, revealing velocity decay, varied stopping behaviors, and highlighting the impact of illumination and channel walls on long-term observations.

Contribution

It provides the first long-term observation of magnetotactic bacteria in microfluidic channels, emphasizing environmental factors affecting bacterial behavior during extended studies.

Findings

Bacteria showed a gradual velocity decrease of about 25 nm/s^2.

Different post-stop behaviors were observed, including rotation and immobility.

High-intensity illumination and channel walls significantly influence long-term bacterial observations.

Abstract

We controlled and observed individual magnetotactic bacteria (Magnetospirillum gryphiswaldense) inside a 5 {\mu}m high microfluidic channel for over four hours. After a period of constant velocity, the duration of which varied between bacteria, all observed bacteria showed a gradual decrease in their velocity of about 25 nm/s$^2$. After coming to a full stop, different behaviour was observed, ranging from rotation around the centre of mass synchronous with the direction of the external magnetic field, to being completely immobile. Our results suggest that the influence of the high intensity illumination and the presence of the channel walls are important parameters to consider when performing observations of such long duration.

Figures

Full text

magnetotactic bacteria, microfluidic, magnetic control

I Introduction

When we observe motile bacteria in an optical microscope, we generally have the problem that the bacteria are in focus for a limited period of time. Microtechnology allows us to fabricate microfluidic channels with a channel height of a few $\mathrm{\SIUnitSymbolMicro m}$, so that the bacteria can be forced to remain in focus (within the depth of field). The bacteria might however still move out of the field of view. Magneto-tactic bacteria (MTB) Klumpp et al. (2018) offer a solution, since they can be forced to swim along a predefined pattern by an external magnetic field Pichel et al. (2018).

In this paper, we describe an experiment in which we observed individual MTB swimming in a figure-8 pattern for a duration of several hours. Our question was how the velocity of the bacteria develops over time, and what happens if the MTB stop swimming. This question has relevance for the application of MTB as carriers for targeted drug delivery Felfoul et al. (2016), in which the MTB will be travelling for thirty minutes or more towards the tumor site. The genes for magnetosomeformation are being identified Uebe et al. (2018), and it is being investigated whether they can be expressed in other types of microorganisms Kolinko et al. (2014). Novel venues can be taken for cells acting as drug delivery agents. If cells from the human microbiota can be genetically engineered to express magnetosomes, their lifetime in the human body is likely to be larger than that of cells not naturally occurring in the human body. Also the type and number of tasks that such magnetically stearable cells can perform may increase. Long term behaviour is also of importance for the application of MTB as transporters inside microfluidic systems themselves Chen et al. (2014).

As far as we know, there have been no observations of individual swimming bacteria over a period of more than a few minutes. MTB have been observed in chambers fabricated from cover slides glued to microscope slides. Reufer et al. Reufer et al. (2014) observed the trajectories of MSR-1 MTB swimming along the top or bottom surface for a few seconds. Erglis Erglis et al. (2007) used $28\text{\,}\mathrm{\SIUnitSymbolMicro m}$-thick double-sided tape to reduce the cell height, and observed single MSR-1 for a period up to $200\text{\,}\mathrm{s}$.

There are no reports on long time observations of individual MTB in microfabricated chips. Other motile baceria have been observed (E-Coli in a channel with a height of $60\text{\,}\mathrm{\SIUnitSymbolMicro m}$ Ahmed and Stocker (2008) and S. marcescens in a channel with a height of $10\text{\,}\mathrm{\SIUnitSymbolMicro m}$ Binz et al. (2010)), but observation times have been below one minute. Männik et al. Männik et al. (2009) observed the growth of a culture of E. Coli up to two days in microfluidic chips with channel heights of $5$ to $7\text{\,}\mathrm{\SIUnitSymbolMicro m}$, but did not track single bacteria for longer than a few seconds. Rather than using restriction and magnetic bacteria, one can also move the observation cell mechanically in the tracking microscope. To keep the bacteria in focus, the tracking has to be done in three dimensions Taute et al. (2015). The longest trajectories observed were $100\text{\,}\mathrm{s}$.

We observed MSR-1 inside a glass microfabricated microfluidic chip with a channel height of only $5\text{\,}\mathrm{\SIUnitSymbolMicro m}$, to ensure they stay in the depth of field during the entire experiment. In combination with magnetic control, we could observe individual MTB swimming in a figure-8 pattern. The bacteria were observed one by one, for periods up to $70\text{\,}\mathrm{m}\mathrm{i}\mathrm{n}$, for a total duration of the experiment of $260\text{\,}\mathrm{m}\mathrm{i}\mathrm{n}$.

II Experimental

II.1 Cultivation of the magnetotactic bacteria

Magnetospirillum gryphiswaldense strain MSR-1 (DSM 6361) cells were grown heterotrophically in a liquid medium (pH $7.0$) containing succinate as the energy source, as previously described by Lefevre et al. Lefèvre et al. (2014). FeSO4 was used as the iron source. The MSR-1 cells were cultivated at $26\text{\,}\mathrm{\SIUnitSymbolCelsius}$ for $3$ to $5\text{\,}\mathrm{d}\mathrm{a}\mathrm{y}\mathrm{s}$ in sterile microfuge tubes prior to microscopic analyses.

The sampling was done using a magnetic “racetrack” separation method as described in Wolfe et al. (1987). Figure 1 shows a transmission electron microscope (TEM) image of an MSR-1, in which the magnetosome chain can be clearly identified.

II.2 Transmission electron microscopy imaging

For the TEM analysis, the medium with MSR-1 MTB was applied on carbon coated copper grids (CF200-Cu) and allowed to absorb for $30\text{\,}\mathrm{s}\mathrm{e}\mathrm{c}$. Excess sample was blotted off by touching the edge of the grid with a clean piece of filter paper and stained with $2\text{\,}\mathrm{\char 37\relax}$ uranyl acetate solution for $30\text{\,}\mathrm{s}\mathrm{e}\mathrm{c}$. The morphology of the MTB was examined by a JEM-2100F TEM (JEOL, Japan) with bright field image at an accelerating voltage of $200\text{\,}\mathrm{k}\mathrm{V}$.

II.3 Microscope

Figure 2 shows the experimental setup. We used an upright reflected light microscope (Zeiss Axiotron II) with a 20$\times$ lens with a numerical aperture of $0.5$ optimized for reflected light, a working distance of $2.1\text{\,}\mathrm{m}\mathrm{m}$, and field of view of $25\text{\,}\mathrm{m}\mathrm{m}$ (Zeiss Epiplan HD DIC). Images were taken by a CCD camera (Point Grey FL3-U3-13S2M-CS) at $10\text{\,}\mathrm{f}\mathrm{p}\mathrm{s}$ with a resolution of 1328$\times$1024.

As light source, a collimated blue LED with an average wavelength of $470\text{\,}\mathrm{n}\mathrm{m}$ was used (Thorlabs M470L2-C4). The manufacturer specifies an approximate beam power of $210\text{\,}\mathrm{m}\mathrm{W}$ in a beam diameter of $37\text{\,}\mathrm{m}\mathrm{m}$. Upon 20$\times$ magnification, this leads to a theoretical power density of $78\text{\,}\mathrm{k}\mathrm{W}\mathrm{/}\mathrm{m}^{2}$. Since for imaging the aperture is nearly closed, a large fraction of the power is blocked. By measuring the intensity difference between a fully open and nearly closed aperture by means of the average intensity on the camera, we estimate a reduction by a factor of $30$, leading to an estimated power density of approximately $3\text{\,}\mathrm{k}\mathrm{W}\mathrm{/}\mathrm{m}^{2}$.

II.4 Microfluidic Chip

The MTB were observed inside a microfluidic chip with a channel height of $5\text{\,}\mathrm{\SIUnitSymbolMicro m}$, identical to our experiments in Pichel et al. (2018). An overview of the microfluidic chip and region of interest for observation can be seen in Figure 3.

II.5 Magnetic field

The magnetic field is generated by a motorised permanent magnet placed underneath the sample Pichel et al. (2018). This magnet has its magnetisation orthogonal to the axis of rotation, so that it creates an in-plane magnetic field at the location of the sample. The in-plane angle of the field can be controlled by the rotation of the motor axis. The motor was programmed to loop in a figure-8 trajectory, so that the bacteria, on average, will not change their position. The programmed trajectory can be manually overridden, to steer the bacteria back to the centre of view to correct for drift. The tracking results of a typical re-centring process can be seen in Figure 4.

II.6 Image Processing

The image sequence was processed offline to extract the coordinates of the bacteria of interest. The procedure is identical to the method published in Pichel et al. (2018). The low-contrast nature of the image required pre-processing steps. Subsequently, we performed background subtraction, lowpass filtering, thresholding, and finally selecting the resulting blobs based on size. The centre of gravity was registered as the position of the bacteria. A nearest-neighbour algorithm with maximum search radius was used to build the trajectories from the detected bacteria. The resulting trajectories were manually cleaned for drop-outs. The velocity was calculated from the trajectories. Due to noise, the centre of gravity of the blobs jitters by one or two pixels ($180\text{\,}\mathrm{n}\mathrm{m}$ per pixel). As the velocity is calculated from the frame-by-frame displacement at $10\text{\,}\mathrm{f}\mathrm{p}\mathrm{s}$, this will result in a small residual velocity of approximately $3\text{\,}\mathrm{\SIUnitSymbolMicro}\mathrm{m}\mathrm{/}\mathrm{s}$.

III Results

III.1 Long term tracking

Data was recorded for a period of $5\text{\,}\mathrm{h}$. During this period, a single MTB was tracked at a time. An example of a recorded video is available as additional material (figure8.mov). When the selected MTB stopped moving, the magnetic field was directed to be parallel to the microfluidic channel in order to find and trap a new MTB.

Figure 5 shows a composite image of one single MTB. The total trajectory with a length of $6\text{\,}\mathrm{s}$ is shown. On top of the trajectory five frames of the recorded movie are superimposed. At the start of the observation, the MTB follows big figure-8 shaped trajectories at a velocity of up to $50\text{\,}\mathrm{\SIUnitSymbolMicro}\mathrm{m}\mathrm{/}\mathrm{s}$. After $2.4\text{\,}\mathrm{s}$, the MTB reverses direction and continues to swim at a very low speed of $5\text{\,}\mathrm{\SIUnitSymbolMicro}\mathrm{m}\mathrm{/}\mathrm{s}$, resulting in a small figure-8 shaped trajectory. This behaviour is in agreement with the bimodal velocity distribution previously observed by Reufer et al. Reufer et al. (2014).

Figure 6 shows the velocity of two MTB as functions of time. The MTB initially show a constant velocity, after which the velocity gradually decreases with time. The initial velocity and the duration of the period of constant velocity vary. For MTB4, which is the same MTB as in Figure 5, several reversals can be observed between $18$ and $22\text{\,}\mathrm{m}\mathrm{i}\mathrm{n}$.

There appears to be more or less a similar rate, about $1.5\text{\,}\mathrm{\SIUnitSymbolMicro}\mathrm{m}\mathrm{/}\mathrm{s}$ per minute ($25\text{\,}\mathrm{n}\mathrm{m}\mathrm{/}\mathrm{s}^{2}$), in the decrease of velocity. Therefore we plotted the velocity of all observed MTB as a function of time, taking the point at which the bacteria stops moving forward as a reference, see Figure 7. The figure suggests that we captured the full behaviour of MTB4 and MTB5, but observed the other MTB at the end of the decay process. This is most likely because we became more skilled in capturing new MTB over the course of the $5\text{\,}\mathrm{h}$ experiment.

III.2 Modes of motile behaviour

When the MTB slow down, we can observe a rotation around the long axis of the MTB (Figure 8), which is in agreement with propulsion by a rotating flagellum Purcell (1977).

When the MTB stop swimming, we can still observe movement. There is a difference between MTB that have a magnetosome, and those that do not. The external rotating field will always exert a torque on an MTB with a magnetosome, even if it is dead. We observe these MTB rotating around an axis that appears to be very close to the centre of their body (Figure 9). We also observe MTB that are not rotating at all (Figure 10). These would be either MTB without a magnesome, or MTB that are firmly stuck to the channel wall. Since we observe small random motion, in agreement with Brownian motion, we assume they are non-magnetic.

There are other MTB that appear to be stuck to the surface of the channel. They rotate around a point that is not in the centre of their body (Figure 11). MTB with and without magnetosome displayed this behaviour. Magnetic MTB follow the rotation of the field but non-magnetic MTB rotate randomly.

IV Discussion

In each case of a long term control sequence we observe a decline in velocity. There a several hypotheses.

A possible explanation might be that the direct lighting from our observation is heating the samples we are observing to a temperature that kills the MTB. However, no noticeable increase in chip temperature was observed. Since the thermal conductivity of glass is high, it does not support a steep thermal gradient, so the entire chip would be at the same temperature. We were still able to find new motile MTB during a $5\text{\,}\mathrm{h}\mathrm{o}\mathrm{u}\mathrm{r}$ experiment. It is therefore unlikely that the observation area increases significantly in temperature.

We assume that local ion depletion is also not a problem, since in some cases when an MTB comes to a halt, other non-magnetic ones still cross the screen without a problem.

One could imagine MTB stop swimming prior to cell division. Clear size changes however are not visible, even after observing the bacteria for a long period after they stop. From growth curves, we can estimate the mean time between cell division to be on the order of $4$-$8\text{\,}\mathrm{h}\mathrm{o}\mathrm{u}\mathrm{r}\mathrm{s}$ Sun et al. (2008). The chance that all observed MTB stopped swimming because there were close to cell division is negligble (less than $0.1\text{\,}\mathrm{\char 37\relax}$).

One might suggest the reduction in velocity is simply due to fatigue of some sort. Perhaps the flagellar motor has intervals of activity over such a long period of time. The efficiency of the flagellar motors is reported to be near $100\text{\,}\mathrm{\char 37\relax}$ Kinosita et al. (2000). Moreover, we have never observed an MTB starting to move again after coming to a halt.

The most likely explanation is an overdose of light. In the microscope, we focus a very bright LED light souce of $450\text{\,}\mathrm{n}\mathrm{m}$ wavelength on the field of view. This high intensity light source might damage the bacteria under observation. It is known that magneto-tactic bacteria respond to light. For MC-1 bacteria, illumination seems to have the same effect as an increase in oxygen concentration Frankel et al. (1997). Non-magnetic AMB-1 bacteria have been observed to migrate towards the light Li et al. (2017). There are no published reports on photo-toxicity of magneto-tactic bacteria, but we could assume a similar sensitivity to light as other types of bacteria. Light illumination at $415\text{\,}\mathrm{n}\mathrm{m}$ at a dose of $750\text{\,}\mathrm{k}\mathrm{J}\mathrm{/}\mathrm{m}^{2}$ for instance strongly reduces the viability of Propionibacterium acnes cultures Ashkenazi et al. (2003). Similarly, Santos et al. Santos et al. (2013) showed that a UV-light dose of $300\text{\,}\mathrm{k}\mathrm{J}\mathrm{/}\mathrm{m}^{2}$ at $365\text{\,}\mathrm{n}\mathrm{m}$ wavelength is sufficient to reduce the viability of most of a set of nine different types of surface water bacteria. Assuming an illumination power density at the sample plane of $3\text{\,}\mathrm{k}\mathrm{W}\mathrm{/}\mathrm{m}^{2}$ (see Section II), this would be equivalent to an exposure time of less than $250\text{\,}\mathrm{s}$.

This photo-toxic explanation is also in agreement with the observation that we can always find a new motile MTB. Only the MTB that are within the illumnated area (about $1.2\text{\,}\mathrm{m}\mathrm{m}^{2}$ will suffer from the intense light source. The MTB outside this area remain unaffected until we use them for observation. It is therefore reasonable to assume that indeed an overdose of light is responsible for the decrease in MTB motility over time.

V Conclusion

We observed magneto-tactic bacteria of type MSR-1 inside a microfluidic chip for a total of $260\text{\,}\mathrm{min}$. During this time, individual bacteria were magnetically steered in figure-8 patterns for a duration of from twenty to fifty minutes.

The MTB occasionally reverse direction, which is accompanied with a sudden drop in velocity. All observed bacteria showed a gradual decrease in velocity until they came to a full stop. The time until the start of the decrease varied, with a maximum of $30\text{\,}\min$. The decay rate however was relatively constant, at about $25\text{\,}\mathrm{n}\mathrm{m}\mathrm{/}\mathrm{s}^{2}$.

When the MTB slow down but are still swimming, we can observe rotation around their long axis. After coming to a halt, we observed three different behaviors. 1) Many MTB still rotate in the field around an axis perpendicular to their long axis and close to their centre of mass. 2) Some MTB appear to be stuck to the channel wall with a flagellum and rotate around one end. They either rotate synchronously with the rotation of the field or randomly. The latter group must be non-magnetic. 3) Finally there are MTB that do not move at all, either because they are firmly stuck or non-magnetic.

As far as we know, this experiment is the first observation of individual motile bacteria for an extended period of time. The experiment was enabled by the availability of glass microfabricated chips with low channel height, so that the bacteria stay in focus. We learned from the experiment that one should consider the influence of the microscope light and the presence of the channel walls.

Acknowledgements.

The authors would like to thank Mohammed Elwi Mitwally of the German University in Cario for help with the manual control of the MTB.