Monitoring of diffusion properties and transverse relaxation time of mouse ischaemic muscle after administration of human mesenchymal stromal cells derived from adipose tissue

Abstract Objectives Application of non‐invasive imaging methods plays an important role in the assessment of cellular therapy effects in peripheral artery disease. The purpose of this work was to evaluate the kinetics of MRI‐derived parameters characterizing ischaemic hindlimb muscle after administration of human mesenchymal stromal cells derived from adipose tissue (hADSC) in mice. Materials and methods MRI experiments were performed on a 9.4T Bruker system. The measurement protocol included transverse relaxation time mapping and diffusion tensor imaging. The monitoring period encompassed 14 days after femoral artery ligation and subsequent cell administration. The effect of hADSC transplantation was compared with the effect of normal human dermal fibroblasts (NHDFs) and phosphate‐buffered saline injection. Results The most significant differences between the hADSC group and the remaining ones were observed around day 3 after ischaemia induction (increased transverse relaxation time in the hADSC group in comparison with the control group) and around day 7 (increased transverse relaxation time and decreased third eigenvalue of the diffusion tensor in the hADSC group in comparison with the control and NHDF groups) at the site of hADSC injection. Histologically, it was associated with increased macrophage infiltration at days 3‐7 and with the presence of small regenerating fibres in the ischaemic tissue at day 7. Conclusions Our results underscore the important role of macrophages in mediating the therapeutic effects of hADSCs and confirm the huge potential of magnetic resonance imaging in monitoring of cellular therapy effects.


| INTRODUC TI ON
From 3% to 10% of the population in the Western world suffer from peripheral arterial disease, and this represents a major health problem of ageing population.
The natural response to major artery occlusion is a complex process, which can be divided into three phases (with some overlapping): (a) cell necrosis and apoptosis phase caused by hypoxia; (b) inflammatory phase characterized by lymphocyte, monocyte and neutrophil migration into the damaged tissue for removal of dead cells; and (c) the muscle regeneration phase. 1 During this process, vasculogenesis, angiogenesis and arteriogenesis occur to establish a functional vascular network. In appropriate conditions, skeletal muscle has the ability to recover after ischaemic injury. However, postischaemic vessel growth and remodelling processes are markedly impaired in patients suffering from arterial diseases. Critical limb ischaemia-an advanced stage of this disease-is associated with a high risk of amputation and death. 2 The prevalence of this condition is approximately 1.3%. 3 Therefore, new therapeutic approaches for these patients are being searched.
Application of mesenchymal stromal cells (MSCs) in the repair of ischaemic tissues has been extensively studied. [4][5][6] MSCs are multipotent (differentiating into adipocytes, chondrocytes and osteoblasts), non-haematopoietic, fibroblast-like plastic adherent cells that can be isolated from various tissue sources (including bone marrow and adipose tissue). Their therapeutic potential is mainly related to the secretion of growth factors and interleukins that can have immunomodulatory, angiogenic, anti-inflammatory and anti-apoptotic effects (the paracrine effect). They may also be a vehicle for gene therapy and drug delivery. Although the initial enthusiasm that these cells are immune-privileged has diminished, it is accepted that they are less immunogenic than other cell types. 7 This property encourages researchers to exploit human rather than mouse MSCs in the studies of mouse models of various diseases. 8 Such studies provide useful preclinical data necessary for designing of clinical trials. Importantly, human MSCs are easier to isolate, expand in culture and less prone to undergo spontaneous transformation to tumorigenic cells than mouse MSCs. A large number of works indicate that human MSCs may exert immunosuppressive effect in immunocompetent mice. [8][9][10] On the other hand, macrophage infiltration, suggestive of transplant rejection, was also observed after administration of human MSCs to rodents. 11,12 Huge variability of the results can be partly explained by the plasticity of MSC phenotype depending on the microenvironment of the host. 13,14 The research progress in the field of cellular therapy depends substantially on the application of non-invasive imaging methods allowing in vivo monitoring of the effects of therapeutic interventions. Magnetic resonance imaging (MRI) technique offering several contrast mechanisms is well suited for a multiparametric characterization of injured tissue. The transverse relaxation time (T2) is known to increase with tissue oedema, necrosis and inflammation. 15,16 Due to fibrillar muscle structure, self-diffusion of water is restricted by membranes and is greater along the fibre orientation than in other directions. Diffusion-weighted magnetic resonance imaging (DWI or DW-MRI) and its special kind, diffusion tensor imaging (DTI), allow the mapping of the diffusion process of water. 17 Diffusion tensor is usually calculated from six or more different diffusion-weighted acquisitions, each obtained with a different orientation of the diffusion-sensitizing gradients. The first eigenvector of this tensor describes the fibre direction, while the second and third eigenvectors represent diffusion perpendicular to the long axis of the cell. The corresponding eigenvalues (λ 1 , λ 2 and λ 3 ) are the diffusion coefficients along these directions, while fractional anisotropy (FA) describes the degree of anisotropy of diffusion. Therefore, quantitative indexes obtained from diffusion tensor imaging characterize local tissue microarchitecture.
The parameters obtained from relaxation and diffusion imaging of ischaemic hindlimbs were found to change dynamically during degeneration and regeneration processes after femoral artery ligation [18][19][20][21] and provided a useful insight into the therapeutic effects of human and mouse endothelial progenitor cells in mouse models of hindlimb ischaemia. 22,23 Recently, our group has reported that M2 macrophages are involved in the repair process of ischaemic muscle tissue after transplantation of hADSC in immunocompetent mice. 24 The purpose of this work was to evaluate the changes in diffusion and relaxation properties of muscle tissue during this process.

| Ethical statement
The experiments were performed in accordance with the Declaration of Helsinki, with the approval of the Local Committee on Bioethics in Katowice. The experimental protocol was approved by the Local Ethics Commission (KB430-17/14).

| Isolation of hADSCs
Mesenchymal stromal cells were isolated from subcutaneous adipose tissues collected during surgery in Maria Skłodowska-Curie Institute -Oncology Center, Gliwice Branch (Poland). The adipose tissues were digested with collagenase using a modification of a previously described protocol (according to Rossini et al 25 ).

| Mouse model of hindlimb ischaemia
In our work, we used the commonly exploited femoral artery ligation model. 10-to 12-week-old male C57BL/6NCrl (immunocompetent) mice were anaesthetized with 2% isoflurane and underwent surgical ligation of the left femoral artery to create unilateral hindlimb ischaemia. The artery was ligated at two points using surgical sutures, according to Brenes et al 26 This step was not followed by artery excision.
We are aware that variation in surgical procedures (femoral artery ligation vs ligation + excision of all branches) leads to the variability in the level of ischaemia. Since our study was devoted to the assessment of M2 macrophage role in the repair of ischaemic tissue, we wanted to limit the surgery-related tissue damage. Although vessel excision would result in more severe ischaemia, diffusion tensor imaging and relaxometry have already been shown efficient in visualization of degeneration and regeneration processes after femoral artery ligation in mice (Heemskerk et al). 18 Therefore, we aimed to evaluate the influence of hADSC administration on these processes.
Although we expected to observe ischaemic changes across the entire calf muscle cross-section, the therapeutic cells were administered into the gastrocnemius muscle (posterior ROI; see Section 2.72.2), according to Brenes et al. 26

| Administration of hADSCs and NHDF cells
An hour after ligation, 1 × 10 6 hADSC cells and 1 × 10 6 NHDF cells in 100 μL of PBS were administered into the gastrocnemius muscles of the mice. The control mice were injected with 100 μL of PBS.
The total number of mice included in the study was 57 (19 per each group).

| Magnetic resonance imaging
Magnetic resonance imaging experiments were performed on a 9.4 T vertical 89 mm bore system (Bruker BioSpin) equipped with a Bruker Micro 2.5 gradient system and a transmit/receive birdcage radiofrequency coil with an inner diameter of 30 mm. During data acquisition, the animals were anaesthetized with 2%-3% sevoflurane. Body temperature and respiration rate were monitored using ECG/respiratory unit (SA Instruments, Inc). A field-map-based shimming (MAPSHIM, Paravision 6.0) was used to optimize B0 field homogeneity.
Effectiveness of femoral artery ligation was visually evaluated using magnetic resonance angiography. The images of vascu-

| MRI data analysis
Quantitative parametric images were obtained on pixel-by-pixel basis with Paravision 6.0 (Bruker) software.
Mean diffusivity (MD) was calculated according to the equation: where λ 1 , λ 2 and λ 3 denote the first, second and third eigenvalue of the diffusion tensor.
Fractional anisotropy was computed using the following equation: T2 maps were calculated using mono-exponential fitting: where k is the proportionality constant related to signal gain or attenuation, S o is the proton density, and TE is the echo time. (1) The slices covered most of the hindlimb volume. After review of the images from a given study, 3 to 4 consecutive slices covering the calf muscle were selected for the analysis. Three 3D regions of interest were manually drawn in these sections for both ligated and non-

| Statistical analysis
Statistical analysis was performed with statistica software (Statsoft).
The differences in the MR parameters between the ligated and non- The P values of < .05 were accepted as statistically significant, while those falling into range from 0.05 to 0.1 were considered to indicate trends.
Absolute values of the evaluated parameters are presented in ( Figure S1). Sigma-Aldrich). Analyses of the specimens were conducted using Nikon Eclipse 80i microscope (Nikon Instruments Inc).

| Magnetic resonance angiography
In order to confirm the effectiveness of surgical intervention, the magnetic resonance angiogram of a mouse hindlimb vessel system was done immediately after femoral artery ligation ( Figure 1B).

| Monitoring of diffusion and relaxation properties
The exemplary T2, MD, FA, λ 1 , λ 2 and λ 3 maps from three consecutive slices of mouse hindlimbs at 3 days after the femoral artery ligation are presented in Figure 1C-H.

| Transverse relaxation time (T2)
The results of the comparisons of the transverse relaxation times obtained for the studied groups are presented in Figure 2A.

Anterior and medial ROIs
Femoral artery ligation resulted in a statistically significant increase in T2 values or upward trends (relative to the non-ligated limb) at days In the medial ROI, a trend towards increased T2 rel is seen at day 3 in the NHDF group in comparison with the control group.

Posterior ROI
In the control group, only a relatively slight T2 increase (relative to the non-ligated limb) at day 1, followed by a subsequent normalization at later time points, is observed. However, T2 was found to be increased at days 1 and 3 in the NHDF group and during the whole evaluation period in the hADSC group (the significant changes at days 1, 3 and 7 and a trend detected at 14 days). The between-group comparison revealed a significantly higher T2 rel in the hADSC group at days 3 and 7 in comparison with the control group and higher T2 rel in the hADSC group at day 7 as compared to the NHDF group. Additionally, it was found that at day 1, T2 rel is higher in the NHDF than in the control group.

| Mean diffusivity (MD)
The results of the mean diffusivity comparisons for the studied groups are shown in Figure 2B.

Anterior and medial ROIs
MD was significantly increased or showed an upward trend in the medial ROI (relative to the intact leg) at days 1 and 3 in the NHDF and hADSC groups. The between-group analysis revealed that MD rel at day 3 tends to be higher in the hADSC group than in the control group. In the medial ROI, the statistically significant decrease in MD (below the value for the non-ligated leg) was noted at day 7 in all evaluated groups. Although it was still decreased at day 14 with respect to the other side in the hADSC and control groups, no significant MD rel differences between the groups were found using the Kruskal-Wallis test.
As regards the anterior ROI, a statistically significant increase in MD (relative to the non-ligated limb) is seen at day 3 in the hADSC group. The between-group analysis revealed that at this time point, MD rel tends to be higher in this group than in the control group. MD decreases (relative to the non-affected leg) at day 7 in the NHDF and control groups, but no statistically significant differences between the groups were found by the Kruskal-Wallis test.

Posterior ROI
In all evaluated groups, a statistically significant decrease in MD or a trend towards its reduction (relative to the non-ligated limb) was found at days 3 and 7 in this ROI. The minimum MD rel value was attained for the hADSC group at day 7. However, the between-group comparison confirms the statistically significant difference only between the hADSC and NHDF groups, whereas in case of the hADSC and control groups, the difference is statistically insignificant.
Additionally, for the NHDF group, MD rel tends to be increased at day 3 as compared to the control group.

| Fractional anisotropy
The results of the fractional anisotropy comparisons for the studied groups are shown in Figure 2C.

Anterior and medial ROIs
In

| First eigenvalue λ 1
The results of the first eigenvalue, λ 1 , comparisons for the studied groups are shown in Figure 3A.

Anterior and medial ROIs
In all studied groups, λ 1 was found to be decreased or showed a downward trend in the anterior and medial ROIs located in the ligated limb (relative to the intact one) at day 7. The decrease was also apparent at day 3 in the control group. The between-group analysis revealed a trend towards the increased λ 1,rel in the anterior ROI at day 7 in the NHDF group relative to the control group.

Posterior ROI
The lowest λ 1,rel in the posterior ROI was observed at day 7 in all studied groups. The between-group comparison revealed a statistically lower λ 1,rel in the hADSC group than in the NHDF group at this time point. A statistically significant decrease in λ 1 with respect to the non-ligated limb was additionally observed at day 3 in the hADSC group, and at days 1 and 3 in the control group.

| Second eigenvalue λ 2
The results of the second eigenvalue, λ 2 , comparisons for the studied groups are shown in Figure 3B.

Anterior and medial ROIs
In the medial ROI of all studied groups, the ischaemic limb λ 2 values were found to be significantly increased or reveal an upward trend (relative to the non-ligated side) at days 1 and 3 (with the exception of day 3 in the control group). In the same ROI, the between-group analysis exposes a trend towards higher λ 2,rel in the hADSC group than in the control group at day 3.
In the anterior ROI of the control group, λ 2 tends to be elevated above the level typical for the non-ligated limb at days 1 and 3, while in the hADSC group, this parameter increases significantly at day 3.
The statistically significant reductions of λ 2 (relative to the non-ligated side) in the anterior and medial ROIs at day 7 were observed in all studied groups (with the exception of the hADSC group in the anterior ROI). However, no significant λ 2,rel changes are seen between the groups in the anterior and medial ROIs at day 7 when using the Kruskal-Wallis test.

Posterior ROI
In the posterior ROIs of all studied groups, the femoral artery ligation induced a statistically significant increase in λ 2 or its trend towards an elevation (relative to the non-ligated side) at day 1.
However, at day 3 in the NHDF and control groups, the λ 2 values are found to fall markedly below the level typical for the non-operated limb, and at day 7, a further decrease takes place for all three groups.
The between-group comparison reveals a significantly lower λ 2,rel in the hADSC group than in the NHDF group at day 7. By day 14, λ 2 returned to the values typical for the non-operated limb only in mice injected with NHDF, but the Kruskal-Wallis test shows no differences between the groups.

| Third eigenvalue λ 3
The results of the third eigenvalue, λ 3 , comparisons for the studied groups are collected in Figure 3C.

Anterior and medial ROIs
In the anterior and medial ROIs of all evaluated groups, the statistically significant increases of λ 3 or its trends towards elevation (relative to the non-ligated limb) are observed at days 1 and 3.
Moreover, the hADSC and NHDF administration leads at day 7 to a statistically significant reduction of this parameter in the operated limb as compared to its value in the healthy leg. However, the between-groups comparisons performed for both ROIs using the Kruskal-Wallis test reveal no significant changes in λ 3,rel . While λ 3 in the anterior ROI returned to the values typical of the intact leg in all groups by day 14, these values were slightly decreased in the control and hADSC groups in the medial ROI at this time point.

Posterior ROI
In the posterior ROI, the femoral artery ligation effect-the λ 3 increase-is seen already at day 1 in all evaluated groups (relative to the non-ligated limb). This elevation is statistically significant for the NHDF and control groups, while for the hADSC group, only a trend is observed. In the operated limb (for all studied groups), λ 3 is evidently lower as compared to the other side at day 7. Of note, the between-group analysis indicates a significant reduction of λ 3,rel in the hADSC group at day 7 as compared to the other groups.
Moreover, the comparison of the hADSC group and the control one lets us notice a trend towards the λ 3,rel elevation at day 3. Although a decreasing λ 3 trend in the operated limb vs the healthy limb is seen at day 14 in these groups, the Kruskal-Wallis test does not show significant changes.

| Summary of the differences in relaxation and diffusion properties of injured muscle between hADSC, NHDF and control groups
Summarizing, the most significant differences between the hADSC, NHDF and control groups were observed around day 3 after femoral artery ligation (increased T2 rel with respect to the control group) and at day 7 (increased T2 rel and decreased λ 3,rel in comparison with the control and NHDF groups) at the site of hADSC injection (the posterior ROI). The representative T2-weighted images, T2 maps and λ 3 maps acquired at day 7 after femoral artery ligation are presented in Figure 4.

| D ISCUSS I ON
In this study, we demonstrated the effects of xenotransplantation of hADSCs on dynamics of degeneration and regeneration processes after femoral artery ligation in immunocompetent mice. We also analysed the ischaemic tissue response to human NHDF cells.
These procedures triggered a cellular and antibody response-it was monitored with T2 relaxometry and DTI and verified using immu-  Although the precise mechanisms of immunomodulatory and pro-angiogenic effects of MSCs are unknown, a growing body of data indicate that these effects are mediated by macrophages. 38 Transplanted MSCs were shown to increase the recruitment of these phagocytes and cause a switch from M1 to M2 polarization accompanied by increased angiogenesis in the injured tissue. [39][40][41] It has been speculated that this switch is related to engulfment of dead MSCs by macrophages. 31,42 Our immunohistochemical analysis confirmed the M2 phenotype of macrophages after hADSC transplantation. The results from our laboratory indicate that the crucial role in this process is played by interleukin-6 secreted by hADSCs. 24 This cytokine stimulates the M2 macrophages responsible for the repair of damaged muscle and formation of new blood vessels. 24 Although the observed T2 rel increase in the hADSC group is probably related to the presence of inflammatory cells at days 3-7, a possible interrelation between T2 relaxation and microcirculation was speculated on the basis of a mouse model of hindlimb ischaemia. 19 The areas characterized morphologically as 'early regeneration' were described by high T2 values and high density of newly formed microvessels. Since these immature vessels are likely to be hyperpermeable, the resulting T2 elevation was interpreted as being due to fluid accumulation in the interstitial space. However, as the

| CON CLUS ION
By using T2 relaxometry and DTI as well as histology, we found that injection of hADSCs into ischaemic hindlimbs in immunocompetent mice resulted in a modification of muscle degeneration and regeneration processes with respect to the control and NHDF groups.
The most significant differences were observed around day 3 after femoral artery ligation (increased T2 in the hADSC group in comparison with the control group) and around day 7 (increased T2 and

ACK N OWLED G M ENTS
This work was performed within the framework of the project no.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.