Introduction

Upon an inflammatory stimulus intravascular immune cells change their motion behavior from patrolling the endothelium to faster rolling with the aim to leave vasculature via the leucocyte adhesion cascade¹. However, current imaging methods to study these alterations, like intravital microscopy, are invasive and suffer a limited field of view¹. Contrarily, by repeated T2* weighted imaging, time-lapse MRI offers a useful tool to track iron-labeled cells non-invasively with whole-brain coverage^2-4. We use contrast simulations based on assembled k-space, a rotating phantom system mimicking migrating cells, and in vivo time-lapse MRI in mouse brains to validate that single patrolling immune cells can be resolved. Further, we study patrolling monocytes and their altered motion behavior under inflammatory conditions distant from the site of inflammation.

Methods

MRI. Time-lapse MRI, both in phantoms and in vivo, was performed on a 9.4 T Biospec (Bruker Biospin, Ettlingen, Germany) with cryogenic probe using a T2* weighted gradient echo sequence with the following scan parameters: TR: 643 ms, TE: 8.0 ms, FA: 60°, averages: 4, image matrix: 180 x 256, in-plane resolution: 67 x 55 µm², 38 slices, slice thickness: 300 µm, scan time: 8 min 12 s, 20 repetitions.
Simulations. Simulations of motion-dependent contrast were done with MATLAB by stepwise changing a signal voids position in a synthetic phantom and filling artificial k-space with fractions of the Fourier transforms of the individual images (figure 1a). An overlay was created by a voxelwise multiplication with in vivo time-lapse MRI data.^2,3
Rotating phantom system. A system (figure 2) was built to rotate agar gel phantoms with embedded iron-labeled monocytes (in vitro labeling as described previously²) inside the scanner imitating moving cells. A stepper motor (Nanotec) with mounted planetary gear allows to mimic cell motion at velocities close to the estimated detection limit (< 1 µm/s ^2,3). Repeated T2* weighted images using the time-lapse sequence were acquired for a static phantom and a rotation of 4.4x10^-3 rpm resulting in cell velocities of up to 2.3 µm/s. Hypointense spots then represent single cells.
In vivo time-lapse MRI. In mice, monocytes were labeled in vivo by i.v. injection of Ferucarbotran (Resovist®, Bayer AG) 24 h before scanning. First, in vivo time-lapse MRI was performed on 5 healthy BALB/c mice. Next, to investigate inflammatory conditions, a subcutaneous inflammation was induced by injection of polyacrylamide-gel containing lipopolysaccharides into both flanks of C57BL/6 mice. Mice were scanned 24 h (n=6), 48h (n=6), 72 h (n=5), 7 d (n=5), and 30 d (n=5) post induction³. Additionally, n=8 control C57BL/6 mice were examined without pellet application.³ Single cells were again identified as hypointense spots, labeled, counted, and tracked using ImageJ and MATLAB.

Results and discussion

The superpositions of simulated cells and real MRI data (figure 1b) agree with the time-lapse images of iron-labeled monocytes inside a rotating phantom and evidently show the increasing blurring of contrast with faster cell movement. The simulations suggest a maximum detectable speed of around 1 µm/s. However, the phantom measurements demonstrate that the detectable velocity range of moving cells is also dependent on the generated contrast in a static position. Hence, high contrast cells close to the edge of the phantom become more and more blurred but remain visible, while low contrast cells disappear when the phantom is rotating (figure 3).
Moreover, in vivo time-lapse MRI enables single-cell tracking of iron-labeled immune cells (figure 4a). By classifying cell motion into different patterns, in vivo time-lapse MRI is able to distinguish between short-term patrolling (in one or two consecutive timeframes), patrolling in confined regions (long-term short-range, in three or more consecutive timeframes), and cells that patrol over extended distances (long-term long-range, in three or more consecutive timeframes with in-plane motion) (figure 4b). For the latter, in 5 healthy mice, a total of 231 cells were detected moving across several voxels with a mean velocity of 0.19 ± 0.06 µm/s (figure 4c) agreeing with the speed of patrolling monocytes as determined by intravital microscopy¹.
Furthermore, with the onset of an immune response monocytes start rolling with much higher velocities (~40 µm/s)¹ and thus cannot be resolved by time-lapse MRI anymore ^2,3. As a consequence, the number of total detected cells decreases upon a peripheral inflammatory stimulus and restores when the inflammation is healed (figure 5a). In addition, the remaining detectable cells lean towards a long-term short-range motion behavior (figure 5b) indicating, although less in total of cells, a more intense patrolling behavior³.

Conclusion

In summary, contrast simulations are in line with in vivo time-lapse MRI data and can be validated by phantom measurements imitating moving cells. All three approaches indicate that patrolling immune cells can be detected and followed dynamically. Furthermore, in vivo time-lapse MRI is able to measure the velocity of cells migrating over an extended distance, to distinguish between different motion patterns of slowly moving cells, and to detect changes in motion behavior due to inflammation. Therefore, we understand time-lapse MRI as being a useful tool for preclinical research to track intravascular patrolling single cells and to study inflammatory diseases. In the future, we aim at accelerating MR acquisition to potentially capture rolling immune cells as well.

Acknowledgements

This study was supported by the German Research Foundation (DFG; SFB 1009 TP-Z02 to CF), the Joachim Herz Foundation (Add-on Fellowship for Interdisciplinary Life Sciences to MM), the Interdisciplinary Centre for Clinical Research (IZKF, core unit PIX) and the Medical Faculty of the University of Muenster (MEDK dissertation fellowship to FF and IF).