Multimodal Imaging Techniques Show Differences in Homing Capacity Between Mesenchymal Stromal Cells and Macrophages in Mouse Renal Injury Models

The mouse mesenchymal stem/stromal cell (MSC) D1 line was obtained from the American Type Culture Collection (CRL-12424) and RAW 264.7 mouse macrophages were obtained from the European Collection of Authenticated Cell Cultures (#91062702). Both cell types were originally derived from BALB/c mice [18, 19]. Cells were transduced with a lentiviral vectors encoding a bicistronic construct of Firefly Luciferase (FLuc) and ZsGreen under control of the constitutive promoter EF1α or a previously described bioluminescence resonance energy transfer (BRET) reporter [17], also under the control of EF1α. The production and titration of viral particles was performed using established protocols [20]. The cells were transduced with a multiplicity of infection of 5 in the presence of polybrene (8 μg/ml) for the MSCs and without polybrene for the RAW macrophages. At least 90 % of the MSCs expressed the transgenes after transduction whereas macrophages, due to the poor transduction efficiency, were sorted based on ZsGreen fluorescence using an Aria fluorescence-activated cell sorter (BD Biosciences). Both cell lines were maintained in Dulbecco’s Modified Eagle’s medium with 10 % foetal calf serum at 37 °C in a humidified incubator with 5 % CO₂.

Iron oxide labelling for MR detection was carried out by co-incubation of the cells with the particles for a period of 24 h prior to administration to mice, after which cells were washed and harvested for injection. Cationic SPIONs produced in-house [21, 22] were used for labelling the MSCs (labelling concentration: 25 μg[Fe]/ml; iron content after labelling: ~ 5 pg[Fe]/cell) whereas macrophages were labelled with ferumoxytol (AMAG pharmaceuticals, labelling concentration: 20 μg[Fe]/ml, iron content after labelling: ~ 6.5 pg[Fe]/cell). The use of different SPIONs is required because each cell type responds differently to such materials, e.g., ferumoxytol alone does not effectively label MSCs [23], but do label RAW macrophages without affecting their morphology or viability [24]. In all experiments, cells were trypsinised, pelleted, resuspended in ice-cold phosphate buffered saline (PBS) and kept on ice until injection. The cell suspension (100 μl) was administered to mice intravenously via the tail vein or intracardially (IC) via the left ventricle of the heart via ultrasound guidance (Prospect system, S-Sharp, Taiwan) [9].

Adriamycin nephropathy was induced in female BALB/c SCID mice by injecting adriamycin (ADR, doxorubicin hydrochloride, Tocris, UK) IV at 6.3 mg/kg body weight (BW) in 0.9 % saline. Cells were administered 14 days post ADR, when renal impairment is observed [25]. For IRI, male mice were anaesthetised with isoflurane and a flank incision was made for unilateral clamping of the renal pedicle for 40 min, using an atraumatic vascular clamp. Subsequently, the vascular clamp was removed and restoration of renal blood flow confirmed visually prior to repair of muscle and skin layers. In this model, renal impairment is observed 24 h post-clamping [26] and cells were administered at this time point.

For bioluminescence imaging, a subcutaneous injection of luciferin (150 mg/kg body weight, Promega, UK) was administered to mice under anaesthesia, which were imaged 15 min later in a bioluminescence imager (IVIS Spectrum, Perkin Elmer, UK). Imaging data were normalised to the acquisition conditions and expressed as radiance (photons/s/cm²/sr (p/s/cm²/sr)). For ex vivo BLI, mice were culled 10 min post administration of luciferin, after which organs were harvested and immediately imaged. Bioluminescence signals of whole animals or individual organs ex vivo were quantified by drawing regions of interest (ROIs) from which the total flux (photons/second) was obtained. For simultaneous tracking of MSCs and RAWs in the same animal, MSCs were transduced with a BRET reporter and imaged in vivo and ex vivo as previously reported [17]. In brief, the imaging protocol involves the tail vein cannulation of the mice, injection of the substrates (furimazine or luciferin) IV and sequential acquisition of data. The BRET signal from the MSCs is obtained immediately after furimazine injection and because the half-life of this substrate is very short, the signal is cleared within approximately 10 min. This allows subsequent injection of luciferin and collection of Fluc signal from the RAW macrophages in the same imaging session, without crosstalk between the different reporter systems. Ex vivo imaging involves a similar method, but the firefly luciferase signal is collected before the BRET signal [17].

MR data was obtained with a Bruker Avance III console interfaced to a 9.4 T magnet system (Bruker Biospec 94/20 USR) and a 4-channel receive-only abdominal surface coil in combination with a 72-mm resonator. Mice were imaged at baseline (before injury) and at multiple time points post administration of SPION-labelled cells. A gated Fast Low-Angle Shot (FLASH) T₂* sequence was used with the following acquisition parameters: TE 5.5 ms, TR 262.5 ms, flip angle 20 °, matrix size 386 × 386 pixels, field of view 35 × 35 mm, slice thickness 0.5 mm, number of slices 20, averages 3, acquisition time 5 min, 35 s. T₂^* relaxation times were obtained from a T₂^* map by drawing ROIs around the cortex of the kidney or a region of the liver as previously described [9]. The T₂^* maps were obtained with a multigradient echo (MGE) sequence with 8 echoes starting at 4.5 ms with 4.5 ms increments and TR 900 ms, flip angle 50 °, matrix size 256 × 256 pixels, field of view 35 × 35 mm, slice thickness 0.5 mm, number of slices 20, averages 2, acquisition time 5 min, 45 s. For postmortem imaging of kidneys, the organs were fixed in formaldehyde, embedded in agarose and imaged with a FLASH T₂^* sequence with following acquisition parameters: TE 6.3 ms, TR 1300 ms, flip angle 20 °, matrix size 400 × 400 pixels, field of view 17 × 17 mm, slice thickness 0.2 mm, number of slices 70, averages 24, acquisition time 3 h, 20 min.

We applied an imaging protocol involving the double labelling of MSCs with SPIONs and FLuc to allow their imaging via MRI and BLI, respectively. In MRI, the SPION label creates local magnetic field inhomogeneities in the areas where cells are delivered to, leading to decay of transverse magnetisation, which is observed as hypointense (dark) contrast in T₂^*-weighted imaging. This is usually used to image a specific organ or body region. BLI, on the other hand allows assessment of the whole-body distribution of cells. SPION⁺FLuc⁺ MSCs were administered into the left cardiac ventricle of mice with adriamycin-induced injury or IRI. Irrespective of the injury model, MR imaging performed within a few hours of cell administration showed hypointense contrast in the renal cortices (Fig. 1a). This confirmed that administration via the arterial route leads to successful delivery of the cells to the kidneys. Hypointense contrast was additionally observed in the medulla of kidneys following IRI (Fig. 1a, blue arrows); however, this phenomenon was also observed in the absence of administered cells (see Fig. 1 in Electronic Supplementary Material (ESM)), suggesting a surgery-specific effect. The contrast in the renal cortices was progressively lost in the subsequent days, indicating that the cells were cleared from the kidneys. Quantification of the T₂^* relaxation time, the time constant that describes the decay of transverse magnetisation, revealed a statistically significant reduction on the administration day but recovery to baseline values in the following days (Fig. 1b, c). For mice that underwent IRI, we compared the relaxation times of injured kidney with that of the uninjured contralateral kidney but saw no differences between the two (see Fig. 2a in ESM). In the liver of IRI mice, the T₂^* relaxation time dropped progressively throughout the experimental period to values that were significantly different from baseline by days 1 and 2, whereas the T₂^* relaxation time was significantly lower than baseline on all days after cell administration in the ADR model (Fig. 1b, c). This suggests that SPIONs are transported from the kidneys to the liver in the days subsequent to cell administration. BLI revealed whole-body distribution of MSCs on the administration day, as anticipated from injection via the arterial route. A progressive loss of signal intensity was seen in the days following cell administration, suggesting cell death. By day 2, only a weak signal was detected in the kidneys, which is consistent with the loss of MR contrast. Taken together, these data suggest loss of MSCs from the kidneys via progressive cell death, following which, the SPION label is transported to the liver, most likely by the host’s reticuloendothelial system.

Having observed no persistence of MSCs in the kidneys, we investigated whether macrophages have a different fate. From here, our study focuses on the unilateral IRI model given the presence of an internal control kidney in each animal. A total of 5 × 10⁶ cells were administered, based on a dose-finding study that showed that these cells are well tolerated when injected via the left ventricle of the heart [9]. MR imaging revealed a similar renal distribution as observed with the MSCs, with strong negative contrast in the cortex on the administration day and progressive loss on the following days (Fig. 2a). Indeed, the changes in T₂^* relaxation times in the renal cortices and liver follow the same behaviour as we observed for the MSCs (Fig. 2b), and as before, no differences were seen when comparing the injured kidney with the control kidney (see Fig. 2b in ESM). However, when animals were analysed individually, we noticed that some appeared to display a greater negative contrast in the injured kidney on the administration day (see Fig. 3 in ESM). BLI imaging displayed a different distribution to that seen with the MSCs, with a moderate loss of signal on day 1, but an increase in signal in the spine and liver by day 2 suggesting that the cells had populated these organs (Fig. 2c). This implies that for this cell type, the reduction in T₂^* relaxation time in the liver was probably a combination of two components: (i) SPION debris from RAW macrophages that have died after administration and (ii) viable SPION⁺ FLuc⁺ RAW macrophages that home to the liver. The differences in cell fate between the two cell types are also evident when mice are imaged ventrally (see Fig. 4 in ESM), with a marked signal originating from the liver of mice that receive macrophages but not from those that receive MSCs.

To confirm the MR data obtained in vivo, we scanned the kidneys postmortem at a higher resolution. Ex vivo imaging of the kidneys of mice that received MSCs exhibited the same features as observed in vivo: (i) hypointense contrast in the renal cortex on day 0, which was lost in subsequent days; (ii) no major differences between the distribution of cells in the cortex of healthy and injured kidneys; and (iii) a darkening of the medulla of the injured kidney (Fig. 3). Imaging of the kidneys of animals that received macrophages displayed a similar pattern, but the contrast in the cortex was stronger which is likely a consequence of the higher injection dose (5 × 10⁶ RAWs vs 10⁶ MSCs), meaning that more cells are present in the kidneys on the administration day. Interestingly, a careful examination of the images revealed that more hypointense spots were present in the injured kidney, particularly on day 1, when compared with the healthy kidney (Fig. 3, red arrows). Taken together, the in vivo and ex vivo MR imaging of macrophages suggest a greater accumulation or persistence of cells in the injured kidney, but this could not be unambiguously demonstrated in a quantitative manner using this imaging modality.

To quantitatively determine whether macrophages do preferentially persist in the injured kidney as suggested by the MR data, mice were culled 24 h post cell administration for ex vivo imaging of the organs via bioluminescence. Imaging of the lungs, liver, spleen, heart and kidneys revealed that cells were present in all major organs at this time point with lungs and liver displaying the strongest signal intensity (Fig. 4). Analysis of the signal intensity in each of the kidneys showed an increase in the mean signal in the injured kidneys when mice received the MSCs IC, but this was not statistically significant (p = 0.16, Fig. 4a). Mice that received the macrophages IC, on the other hand, displayed a statistically significant difference between kidneys, with the injured having a greater bioluminescence intensity (p = 0.032, Fig. 4b). Because it is known that RAW macrophages are able to bypass the lung vasculature when administered IV [9], we assessed whether this route of administration can also lead to cells accumulating preferentially in the injured kidney. Administration of 10⁷ cells via this route confirmed their ability to extravasate the lungs, with cells also populating the liver, spleen and kidneys (Fig. 4c). A stronger bioluminescence intensity was detected in the injured kidneys (p = < 0.001) suggesting a homing effect where these cells actively migrate to the site of ischaemic injury. Of note, our experimental setup only allowed us to detect those differences when imaging the organs ex vivo. Our attempts to quantify the signal emanating from the kidneys in vivo were unsuccessful due to the presence of cells in other organs, the surgical scar overlying the injured kidney and the low spatial resolution of BLI (ESM Fig. 5).

Having determined differences in the homing and persistence of MSCs and RAW macrophages in mice with IRI, we next sought to identify whether the biodistribution of the MSCs could be influenced by the presence of exogenous macrophages. For this, we co-injected MSCs and RAW macrophages into the same mice and applied a method to track them individually, using a combination of NanoLuc-based BRET reporter and FLuc. Shortly following IC administration, both cell types showed a similar whole-body distribution, with good co-localisation of the MSC and macrophage signal on day 0 (Fig. 5a). On day 1, the signal weakened, in agreement with the data in Figs. 1 and 2, with some MSCs still present in the abdominal area, including the kidneys and brain, whereas the macrophage signal was most intense in the spine. A different scenario emerged when the cells were administered IV, with both types of cells being found exclusively in the lungs on the day of administration. A strong reduction in signal intensity was observed by day 1, with MSCs still located in the lungs, whereas macrophages were found not only in the lungs, but also in the abdomen.

Ex vivo imaging of organs on day 1 showed similar results to those obtained when the cells were administered individually. After IC administration, MSCs were found in most organs including the kidneys, with a stronger signal intensity in the injured kidney (Fig. 5b). In contrast to the data shown in Fig. 4a, we saw a statistically significant difference between the kidneys in this experimental setup. We have previously shown that MSCs expressing this BRET reporter have a light output much greater than that of cells expressing FLuc [17] and it is thus possible that these results reflect a greater sensitivity of the reporter, allowing us to detect a statistically significant difference which was not observed with FLuc. However, it is also possible that the increased numbers of RAW macrophages in the injured kidney may restrict blood flow through the glomerular capillaries, causing a transient accumulation of MSCs. Following IV administration, no MSCs were detected in any organ apart from the lungs (Fig. 5b). RAW macrophages, on the other hand, produced a stronger signal in the injured kidney that was statistically significant irrespective of the administration route, demonstrating that their homing is not affected by co-administration of MSCs.