In vivo monitoring of inflammation after cardiac and cerebral ischemia by ¹⁹F magnetic resonance imaging

Abstract

Background

In this study we developed and validated a new approach for in vivo visualization of inflammatory processes by magnetic resonance imaging (MRI) using biochemically inert nanoemulsions of perfluorocarbons (PFCs).

Methods and Results

Local inflammation was provoked in two separate murine models of acute cardiac and cerebral ischemia, respectively, followed by intravenous injection of PFCs. Simultaneous acquisition of morphologically matching proton (¹H) and fluorine (¹⁹F) images enabled an exact anatomical localization of PFCs after application. Repetitive ¹H/¹⁹F MRI at 9.4 Tesla revealed a time-dependent infiltration of injected PFCs into the border zone of infarcted areas in both injury models, and histology demonstrated a colocalization of PFCs with cells of the monocyte-macrophage system. We regularly found the accumulation of PFCs in lymph nodes. Using rhodamine-labelled PFCs, circulating monocytes/macrophages were identified to be the main cell fraction taking up injected nanoparticles.

Conclusions

PFCs can serve as “positive” contrast agent for detection of inflammation by MRI, permitting a spatial resolution close to the anatomical ¹H image and an excellent degree of specificity due to lack of any ¹⁹F background. Since PFCs are non-toxic, this approach may have a broad application in the imaging and diagnosis of numerous inflammatory disease states.

INTRODUCTION

Inflammation is associated with a large number of human diseases such as atherosclerosis, glomerulonephritis, inflammatory bowel disease, transplant rejection, neurodegenerative brain diseases, brain and spinal cord trauma, myocarditis, and ischemic heart disease. Thus, the medical problem is vast while an exact diagnosis is often difficult. Accordingly, therapy is frequently limited to symptomatic treatment, and the success of the prescribed therapy is difficult to assess. Although recent advances involve various imaging modalities such as PET, CT, MRI, optical, and ultrasound,^1-3 the visualization of inflammatory processes still poses a serious challenge, since especially in the initial phase the affected tissue does not exhibit specific physical properties which can be used to create contrast between inflamed and healthy regions.

Among the different non-invasive imaging modalities capable of whole-body imaging, such as PET and SPECT, MRI provides superior resolution and the potential to generate the required contrast to non-inflamed areas by gadolinium enhancement. However, this attempt relies on the transient accumulation of intravascularly applied gadolinium contrast agent in the interstitial space due to enhanced endothelial permeability,^4,5 which is a rather nonspecific phenomenon found to be associated with a variety of diseases. A more defined approach to delineate inflammatory areas from surrounding tissue is the tagging of infiltrating, immunocompetent cells with contrast agents.^6,7 Noninvasive visualization of immigrating cells by MRI so far used predominantly superparamagnetic iron oxide particles (SPIOs) taking advantage of the high affinity of these species for the monocyte-macrophage system.^8,9 Despite its excellent sensitivity, this attempt has the disadvantage, that the particles are not detected directly. Local deposition results in regional magnetic field inhomogeneities and thereby in a depletion of the MR signal. As a consequence anatomical ¹H MR images are often difficult to interpret because it is not always clear whether dark areas are caused by these nanoparticles or by other inhomogeneities. At present there is no method for a true positive MRI identification of infiltrating cells into inflamed tissue.

In this study we demonstrate the feasibility and safety to image inflammation in mice with a “positive” contrast at high local resolution using fluorine MRI. The naturally occuring stable fluorine isotope ¹⁹F (100%) is MR active and exhibits a sensitivity close to the ¹H nucleus.^10,11 Due to the lack of any ¹⁹F background in the body, observed signals originating from injected¹⁹F-containing compounds exhibit an excellent degree of specificity. The merging of recorded ¹⁹F images with simultaneously acquired, morphologically matching ¹H images enables an exact anatomical localization of fluorinated substances as “hot spots”¹². In the present investigation we used nanoparticles containing perfluorocarbons (PFCs) - a family of compounds which is known to be biochemically inert. Some of the PFC members, such as perfluorodecalin, perfluorotripropylamine, perfluorodichloroctane, and perfluorooctyl bromide (also known as perflubron) were already employed in patients as artificial blood substitutes.¹³ However, we used perfluoro-15-crown-5 ether, a PFC in which all 20 fluorine nuclei are chemically and magnetically equivalent thereby exhibiting superior properties for ¹⁹F MRI detection.¹⁴ In contrast to previous studies utilizing ¹⁹F MRI of PFCs for tracking of injected stem/progenitor cells after ex vivo loading,^15,16 we applied emulsified PFCs systemically resulting in an efficient and selective enrichment in circulating cells of the monocyte/macrophage system. This approach enabled us to monitor the infiltration of immunocompetent cells into inflammatory areas in an acceptable acquisition time with a spatial resolution close to the anatomical ¹H image.

METHODS

An expanded Methods section can be found in the online Data Supplement.

Preparation of the PFC emulsion

Purified egg lecithin (E 80 S, 4% wt/wt, a generous gift from Lipoid (Ludwigshafen, Germany)) was dispersed in isotonic phosphate buffer (10 mM phosphate, 150 mM NaCl, pH 7.4) by magnetic stirring at room temperature for 30 minutes. When Lissamine™ rhodamine B (rhodamine dihexadecanoic phosphatidylethanolamine, (rhodamine DHPE), Molecular Probes (Leiden, The Netherlands)) was used as a fluorescent lipid marker, a lipid mixture of lecithin and rhodamine DHPE (99.5/0.5 mol/mol) was dissolved in ethanol and the solvent was subsequently removed under reduced pressure at 35 °C, followed by evaporation under high vacuum. The resulting lipid film was hydrated with buffer by gentle mixing and stirring. After adding the perfluoro-15-crown-5 ether (10 % wt/wt, Fluorochem Ltd. (Glossop, UK)), the dispersion was pretreated with a high-performance disperser (T18 basic ULTRA TURRAX, IKA Werke GmbH & CO. KG, Staufen, Germany) at 14.000 rpm for 2 minutes. The resulting crude emulsion was high pressure homogenized (70 MPa, 10 cycles, APV Gaulin Micron Lab 40, APV, Unna, Germany). The formed nanoemulsion was filtered through a 0.22 μm sterile filter unit (Millex-GS, Millipore, Ireland) and stored until application at 6 °C.

Animal experiments

Animal experiments were performed in accordance with the national guidelines on animal care and were approved by the Bezirksregierung Düsseldorf. The male mice (C57BL/6, 20-25 g body weight, 10-12 weeks of age) used in this study were bred at the Tierversuchsanlage of Heinrich-Heine-Universität, Düsseldorf, Germany. They were fed with a standard chow diet and received tap water ad libitum. A total number of 60 mice has been investigated: blood analysis and controls with PFC/saline injections, 30/10; myocardial infarction, 12; cerebral ischemia; 8. Myocardial infarction was provoked by ligation of the left anterior descending coronary artery (LAD). In a separate experimental series focal cerebral ischemia was induced by photothrombosis (see online Data Supplement for a description of both injury models at length). A detailed scheme of the experimental protocols applied to the different groups is shown in online Data Supplement Fig. 1.

PFC injections

Mice were anesthetized with isoflurane (2.0%) using a home-built nose cone. A total volume of 100 μl (for fluorescence experiments) or up to 500 μl (for MRI) of the PFC emulsion was given intravenously through the tail vein at the time indicated in the different experiments.

Magnetic resonance imaging

Data were recorded on a Bruker DRX 9.4 Tesla Wide Bore (89 mm) NMR spectrometer operating at frequencies of 400.13 MHz for ¹H and 376.46 MHz for ¹⁹F measurements. A Bruker microimaging unit (Mini 0.5) equipped with an actively shielded 57-mm gradient set was used and images were taken from a 30-mm birdcage resonator tunable to ¹H and ¹⁹F. After acquisition of the morphological ¹H images, the resonator was tuned to ¹⁹F and anatomically matching ¹⁹F images were recorded. For superimposing the images of both nuclei, the “hot iron” colour lookup table (ParaVision, Bruker) was applied to ¹⁹F images.

Mice were anesthetized with 1.5% isoflurane and were kept at 37 °C. For functional cardiac analysis ¹H images of murine hearts were acquired essentially as described¹⁷ using an ECG- and respiratory-triggered fast gradient echo cine sequence (field of view (FOV) 30×30 mm², matrix 128×128, slice thickness (ST) 1 mm). Corresponding ¹⁹F images were recorded from the same FOV using a multislice rapid acquisition with relaxation enhancement (RARE) sequence: RARE factor 64, matrix 64×64, ST 2 mm, 256 averages, 19.12 min acquisition time. For fusion with ¹⁹F images additional ¹H datasets with a ST of 2 mm were recorded. Brain images were acquired using multislice RARE sequences for both nuclei from a reduced FOV of 20×20 mm² but otherwise unaltered geometry (please refer to the online Data Supplement for a more detailed description of MRI setup, acquisition parameters, and quantification procedures).

Blood analysis

Blood was obtained from the vena cava inferior at various times after injection of the PFC emulsion as indicated in the different experiments. Determination of serum markers of liver function was performed by the Central Laboratory of the University Hospital Düsseldorf using clinical routine protocols. In separate experiments, mononuclear cells were isolated from the blood samples by centrifugation over Histopaque density gradient (2.5 ml layers of both 1083 and 1119 (Sigma), 25 min, 700 g at room temperature). Thereafter, the tube was either immediately transferred into the NMR spectrometer for MRI (see online Data Supplement for details) or the mononuclear cells were collected from the interface of the layers and analyzed by FACS (see next section).

Flow cytometry

In preceding experiments with the murine macrophage cell line RAW 246.7 loaded in vitro with rhodamine-labelled PFCs (online Data Supplement Fig. 2) we confirmed that fluorescence of rhodamine bound to the coat of the PFC particles is detectable by FACS analysis (data not shown). Freshly prepared peripheral blood mononuclear cells (PBMC) were stained for flow cytometric analysis according to standard procedures (see online Data Supplement for details). Cells were analyzed on a FACScalibur flow cytometer (Becton Dickinson) and samples were gated on live cells based on forward as well as side scattering and by exclusion of propidium iodide-positive cells. For each sample at least 10,000 live events were acquired and analyzed with the CellQuestPro software.

Immunohistochemistry

To avoid a dissociation of rhodamine label and markers of the initial PFC carrier due to downstream processes after infiltration, all organs analyzed by immunohistochemistry were excised 1 day after injection. Slides were air dried and red fluorescence images were recorded without further processing because of water solubility of rhodamine-labelled PFCs and the impossibility of adequate histologic fixing of the nanoparticles. The sections selected for photographs were related to anatomical landmarks in order to ensure retrieval of the same area post immunohistochemistry. After processing for immunofluorescence of CD11b (see online Data Supplement for a detailed description of protocols applied to heart and brain slices, respectively), cardiac and cerebral sections were again microscoped making use of the anatomical landmarks defined in the previous session. Slides were viewed with an Olympus BX50 fluorescence microscope equipped with standard filter sets and using objectives without (before immunostaining) and with (after mounting) cover glass correction. We deliberately refrained from a merging of images taken before and after immunostaining, since an exact overlay was hampered by unavoidable minute alterations of the dried histologic slices during immunohistochemic incubation steps and subsequent mounting.

The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.

RESULTS

PFC infiltration into the heart after infarction assessed by in vivo ¹⁹F MRI

Cardiac infarction was induced by ligation of the left anterior descending coronary artery (LAD), a procedure well known to be associated with an acute inflammatory response. Two hours after ligation 500 μl of 10% perfluoro-15-crown-5 ether emulsion (average size around 130 nm, zeta potential = -31.3±1.5 mV) was applied via the tail vein (for details on the PFC emulsion please refer to online Data Supplement, Methods section).

After surgery and application of the contrast agent all animals (n=6) were imaged five times within 7 days. The infarcted area was localized by acquisition of fast gradient echo ¹H cine movies via akinesis of the affected region within the left ventricle. Subsequently, anatomically matching ¹⁹F images were recorded for tracking of the injected PFCs. A typical example of consecutively recorded ¹H and ¹⁹F images obtained 4 days after ligation of the LAD is illustrated in Fig. 1A. The enddiastolic ¹H image (Fig. 1A, left) clearly shows the presence of ventricular dilatation and wall thinning within the infarcted area (I), and in the corresponding ¹⁹F image (Fig. 1A, middle) a signal pattern matched in shape of the free left ventricular wall. Merging of these images (Fig. 1A, right) confirms the localization of PFCs within the anterior, lateral, and posterior wall. In all animals studied, ¹⁹F signal was also detected in the adjacent chest tissue, where thoracotomy (T) for LAD ligation was performed. Note that there is no background signal from other tissue. Repetitive measurements from day 1 after LAD ligation revealed a time-dependent accumulation of PFCs within the infarcted region as shown in a representative example in Fig. 1B. Enddiastolic ¹H images acquired 1, 3, and 6 days after induction of myocardial infarction show the progressive left ventricular dilatation as consequence of the insult. Merging with the matching ¹⁹F images (red) demonstrates the successive infiltration of PFCs into the affected area of the heart and the region of the chest injured by surgery. Detected ¹⁹F signals were restricted to the area near the infarcted region of the heart, at no time infiltrating PFCs were observed within the septum (see online Data Supplement Table 1 for individual data of all animals studied).

Figure 1. Infiltration of PFCs after myocardial infarction as detected by in vivo ¹⁹F MRI.

(A) Anatomically corresponding ¹H and ¹⁹F images from the mouse thorax recorded 4 days after ligation of the left anterior descending coronary artery (LAD) showing an accumulation of ¹⁹F signal near the infarcted region (I) and also at the location of surgery, where the thorax was opened (T). PFCs were injected at day 0 (2 h after infarction) via the tail vein. (B) Sections of ¹H images superimposed with the matching ¹⁹F images (red) acquired 1, 3, and 6 days after surgery indicate a time-dependent infiltration of PFCs into injured areas of the heart and the adjacent region of the chest affected by thoracotomy. Note that at day 4, an additional bolus of PFCs had been injected to compensate for clearance of the particles from the bloodstream after 3 days (see text).

Although strong PFC signals were found in ex vivo ¹⁹F images of blood components (see below), in vivo signals from PFCs in the circulation were not detectable at all (e.g. no signal within ventricular chambers, Fig. 1). Even when ¹⁹F images were acquired immediately after injection, no ¹⁹F signal from the streaming blood could be observed, since the pulse sequence used for ¹⁹F MRI (RARE) results in a signal void of flowing blood particles. Therefore, detected signals can be unequivocally attributed to accumulated PFCs in the tissue without contamination from ¹⁹F signals of circulating PFCs.

Uptake and transport of PFCs by cells of the monocyte-macrophage system

To characterize the mode by which PFCs can enter the injured heart tissue, murine blood samples were investigated ex vivo by ¹⁹F MRI after intravenous application of the emulsion. ¹⁹F images acquired after density gradient centrifugation of blood collected at different points after injection revealed a time-dependent accumulation of the ¹⁹F signal within the layer of the mononuclear cells (Fig. 2). However, three days after injection the PFCs were completely cleared from the bloodstream and were no longer detectable by ¹⁹F MRI.

Figure 2. Uptake of PFCs by mononuclear cells.

Matching ¹H and ¹⁹F MR images of a 15-ml Falcon tube after centrifugation of the collected mouse blood over Histopaque density gradient show a time-dependent accumulation of ¹⁹F signal in mononuclear cells after tail vein injection of 500 μl PFC emulsion. Blood samples were taken 2 h, 1, 2, and 3 days after PFC injection

In order to further specify the cell population containing the PFCs, experiments were performed using rhodamine-labelled PFCs. This enabled us to trace the fluorescence label not only within the mononuclear blood cells by flow cytometry but also within the inflamed region by means of fluorescence microscopy of tissue sections.

After tail vein injection of fluorescently labelled PFCs and subsequent collection of blood samples, we analyzed the layer of mononuclear cells containing the PFCs as assessed by ex vivo ¹⁹F MRI (Fig. 2). As shown in Fig. 3A two hours after injection of rhodamine-labelled PFCs almost a fifth of the mononuclear cells were found to be positive for rhodamine with the large majority of the labelled cells (~80%) exhibiting the monocyte/macrophage marker CD11b (Fig. 3B top). Approximately half of this cell type was detected to be loaded with PFC particles (Fig. 3C). The remaining of the rhodamine-positive cells were observed to be B cells (B220, Fig. 3B middle) with a marginal amount of T cells (< 2%; CD3, Fig. 3B bottom). Control experiments in vitro with a murine macrophage cell line confirmed that the labelled PFCs are avidly taken up by macrophages (online Data Supplement Fig. 2).

Figure 3. Flow cytometry of murine mononuclear cells 2 h after tail vein injection of rhodamine-labelled PFCs.

(A) Peripheral blood mononuclear cells (PBMC) from a control mouse (upper panel) and a mouse treated with rhodamine-labelled PFCs (lower panel) were analyzed for rhodamine fluorescence by flow cytometry. Dot blots show rhodamine vs. FITC fluorescence, numbers in the upper left quadrants indicate the percentage of rhodamine-positive PBMC. (B+C) PBMC from both mice were stained with FITC-labeled anti-CD11b, anti-B220 and anti-CD3 monoclonal antibodies (mAb). (B) Gated on rhodamine-positive cells, histograms display staining of specific mAb (open) and isotype-matched control mAb (grey). Numbers indicate the percentage of rhodamine-positive cells expressing the specific cell marker. (C) Histograms show rhodamine fluorescence from control (grey) and treated mouse (open). Numbers indicate the percentage of rhodamine-positive cells within the cell population analyzed.

The fate of rhodamine-labelled PFCs in cardiac tissue was investigated by histology. Microscopic survey images obtained from the same mouse shown in Fig. 1A are displayed in Fig. 4A. Micrographs show a pattern of rhodamine fluorescence which is similar to the signal distribution in the corresponding ¹⁹F MR image acquired immediately before organ excision (Fig. 1A right). The main fluorescence signals were located exclusively within the injured area. No rhodamine fluorescence was observed in the septum and in necrotic areas as confirmed by staining with triphenyltetrazolium chloride (data not shown).

Figure 4. Colocalization of rhodamine-labelled PFCs and monocytes/macrophages in the heart 4 days after myocardial infarction.

(A) Overview images of the heart from frozen sections (8 μm) obtained from the same mouse shown in Fig. 1A. (B) Anatomically matching sections before (PFC) and after processing for immunofluorescence of CD11b (DAPI, CD11b). The black rectangle in the bright field image (scale bar, 500 μm) represents the section displayed in the adjoining fluorescence images. Because of water solubility of the rhodamine-labelled PFCs and the impossibility of adequate histologic fixing of the particles, rhodamine fluorescence images had to be recorded before immunohistochemistry. Therefore, the sections selected for photographs were carefully related to anatomical landmarks in order to ensure retrieval of the same area after immunohistochemistry. Rhodamine fluorescence appeared to be diffusively distributed over the cells, while green fluorescence patterns were restricted to surface structures of the infiltrated macrophages/monocytes. Although the PFC image is slightly shifted to the left as compared to the CD11b and DAPI images, a colocalization of red and green fluorescence can be unequivocally recognized. PFCs were injected at day 3 after LAD ligation via the tail vein. The scale bar represents 50 μm.

Immunostaining of tissue sections for the monocyte/macrophage marker CD11b with FITC revealed some colocalization of fluorescence patterns for cells of the monocyte-macrophage system (green) and for rhodamine-labelled PFCs (red) as shown in Fig. 4B. It should be noted, however, that technical reasons precluded a precise merge of the differently labeled sections: due to the water solubility of the rhodamine-labelled PFCs, red fluorescence images had to be taken prior to immunohistochemisty for CD11b and required the careful selection of anatomical landmarks in order to ensure retrieval of the same area.

PFC infiltration into the brain after focal cerebral ischemia

In another set of experiments focal cerebral ischemia was chosen as an additional model of acute inflammation. After inducing ischemia by photothrombosis, all animals (n=4) were imaged at regular intervals up to 4 weeks after surgery. In RARE ¹H images the ischemic region appeared initially as bright area (Fig. 5A, left top), and the corresponding ¹⁹F images clearly show infiltration of PFCs into the border zone of the infarct which was detected at the earliest at day 4 after photothrombosis. ¹⁹F signal was transiently also observed supracranially at the location of skin incision (Fig. 5A, left bottom). Characteristic ¹H and ¹⁹F images (Fig. 5A) acquired from an individual mouse 7, 9, 12, and 19 days after inducing focal cerebral ischemia definitely show a movement of the PFCs with the rim of the shrinking infarct over time (see online Data Supplement Table 2 for individual data of all animals studied).

Figure 5. Infiltration of PFCs into the brain after induction of focal cerebral ischemia by photothrombosis.

(A) Sections of brain ¹H images (top) from an individual mouse superimposed with the corresponding ¹⁹F images (red, bottom) showing a movement of the PFCs with the rim of the infarct over time. Initially, additional signal was also observed supracranial at the location of surgery. Images were obtained 7, 9, 12, and 19 days after induction of focal cerebral ischemia. PFCs were injected at day 0 (2 h after infarction) and day 6 via the tail vein. (B) In vivo and post mortem brain images acquired 7 days after photothrombosis. (Left) Section of merged ¹H and ¹⁹F images taken immediately before organ excision. (Middle) Microscopic survey of the injured hemisphere from 8-μm frozen sections (bright field; scale bar, 2 mm). The black rectangle represents the section displayed in the adjoining fluorescence image. (Right) Infarcted area immunostained for CD11b (scale bar, 500 μm).

To support the notion that PFCs were carried into the ischemic region by monocytes/macrophages, again experiments with rhodamine-labelled PFCs were conducted (n=4). Microscopic survey images after FITC immunostaining for CD11b exhibited a comparable pattern of green fluorescence as observed for the ¹⁹F signal in the preceding MR experiment (Fig. 5B). Furthermore, comparison of red and green fluorescence at large magnification indicated colocalization of PFCs and CD11b-positive cells (online Data Supplement Fig. 3).

Detection threshold and absolute quantification

The sensitivity of our present approach can be estimated from Fig. 2 by correlating the number of cells contained in the layer of the mononuclear cells with the signal-to-noise (S/N) ratio in the corresponding areas of ¹⁹F images. Two days after PFC injection, the mean S/N ratio within this layer was determined to be 24 at a voxel size of 0.44 μl (FOV 30·30 mm², matrix 64·64, ST 2 mm). The mononuclear cell layer contained 1.16·10⁶ cells distributed vertically over about 1 mm and horizontally over the inner diameter of the tube (14 mm, as derived from axial ¹H images), which results in a cell number of about 3300 per ¹⁹F MR voxel within this layer. Assuming a minimal S/N ratio of 3 as detection threshold, as little as ~400 cells are expected to be MRI visible under these conditions. Taking into account that only a fraction of the mononuclear cells is loaded with PFCs (Fig. 3), the detection limit may be even lower.

A similar conclusion was reached in a separate set of experiments, in which RAW 264.7 macrophages were incubated ex vivo with PFCs under in-vivo-like conditions and analyzed by ¹⁹F MRI after immobilization in agarose (for details see online Data Supplement, Methods section). Stepwise dilution of PFC-loaded macrophages revealed that less than 200 cells were detectable within a voxel of 0.44 μl (online Data Supplement Fig. 4). By calibration of the absolute ¹⁹F signal intensities with PFC concentration standards (R² = 0.99892; online Data Supplement Fig. 5), the average PFC-loading per cell was calculated to be 0.73±0.19 pmol (n=8). Assuming a similar uptake of PFCs in vivo, the number of PFC-containing cells within ischemic areas can be quantified by interpolation from ¹⁹F signal intensities of the affected regions (online Data Supplement Tables 3+4).

Control experiments after PFC injection

Without further intervention, at no time ¹⁹F signals were observed within the heart or the brain. However, ¹⁹F images showed a distinct signal in the spleen one day after injection of the PFC emulsion and a weaker signal in the liver which increased up to days 2-3 reaching an intensity similar to the signal from the spleen (online Data Supplement Fig. 6). Interestingly, at the same time additional signals regularly appeared in lymph nodes located in the area of the upper thorax and the head, and became clearly visible as shown in Fig. 6. The signals in the liver persisted for several months, but no adverse effects of the PFCs were observed in these animals, and serum markers of liver function were comparable to those of saline-treated animals (e.g. the ratio of glutamic oxaloacetic transaminase to glutamic pyruvic transaminase (GOT/GPT) was 2.53±1.01 (PFC, n=8) vs. 2.26±0.57 (saline, n=7)).

Figure 6. Accumulation of PFCs in lymph nodes as detected by in vivo ¹⁹F MRI.

Axial ¹H MR images (A+B top) of a mouse superimposed with the corresponding ¹⁹F MR images (red) recorded 3 days after PFC injection via the tail vein. Orientation of axial slices is indicated in the corresponding sagittal images (A+B, bottom). (A) upper thorax (FOV, 30·30 mm²), (B) head (FOV, 20·20 mm²). Abbreviation: Ln, lymph nodes.

DISCUSSION

The present study describes a novel approach to visualize local inflammatory processes by ¹⁹F MRI using in vivo tagging of circulating monocytes/macrophages with biochemically inert PFCs. Our results show that intravenous application of emulsified PFCs after provoking local inflammation by acute cardiac or cerebral ischemia results in the accumulation of ¹⁹F labelled cells within injured areas. Detection of infiltrating monocytes/macrophages by ¹⁹F MRI at a field strength of 9.4 T is feasible in the mouse at an acceptable acquisition time (20 min) with a resolution close to the anatomical ¹H image. Therefore, PFCs can serve as a “positive” contrast agent for inflammatory processes (Fig. 7) exhibiting a high degree of specificity due to the lack of any ¹⁹F background.

Figure 7.

Schematic drawing illustrating the use of PFCs for monitoring of inflammatory processes. After injection emulsified particles are taken up by the monocyte-macrophage system and vehicled to areas of inflammation. Due to the lack of any ¹⁹F background signal, the detected signals are highly specific for infiltrating immunocompetent cells loaded with PFCs (Figure designed with use of ScienceSlides software, VisiScience Corporation, USA).

Compared to previous ¹H MRI approaches to visualize the infiltration of immunocompetent cells into inflamed areas by use of SPIOs, the presented method has the advantage of a direct positive detection of the tagging agent and therefore has the potential to work also in tissues which generally appear very dark in ¹H MRI, such as the lungs. Even though techniques have been recently described to image SPIOs or other iron oxide particles with a bright contrast,¹⁸ the physical basis of detection is still the disturbance of the regional magnetic field by these particles. Therefore, it often remains difficult to unequivocally assign alterations in local contrast to accumulating SPIOs. Furthermore, iron-based contrast agents are readily metabolized, whereas the fluorinated crown ether used in this study is biologically inert and cannot easily be degraded. This is due to the very stable C-F bond and the dense electron cloud of the fluorine atom, which results in a protective sheath.¹⁹ Experimentally, this provides the unique possibility to specifically and permanently label circulating monocytes/macrophages and follow their fate within the body. It is of note, that an absolute quantification of the observed signals is feasible (online Data Supplement Figs. 4+5 and Tables 1-4) which can be translated into the number of infiltrating immunocompetent cells.

Recent ¹⁹F MRI tracking studies of cells loaded ex vivo with PFCs and subsequently injected into mice either required long acquisition times (up to 3 h)¹⁵ or were limited in spatial resolution (voxel size of 26 μl¹⁶ compared to 0.2-0.4 μl in the present work). In this latter investigation the limit of detection was reported to be about 6000 labelled cells. The substantial higher sensitivity observed in our study is most likely due to the fact that the monocyte-macrophage system in vivo more effectively takes up the injected PFCs as compared to stem/progenitor cells incubated ex vivo. Labelling of about 50% of the total monocyte/macrophage cell population (Fig. 3C) raises the question about function and integrity of the loaded cells. Previous studies revealed that perfluoro-15-crown-5 ether labelling had no significant effect on cell proliferation, function or maturation.^15,16 It seems likely that this also applies to the monocyte-macrophage system, since both the time course of accumulation and the localization of PFC-containing monocytes/macrophages within ischemic areas are in good agreement with previous data on myocardial^20,21 and cerebral infarction^6,9 suggesting unaltered infiltration kinetics and distribution of loaded cells. Furthermore, we did not observe any adverse effects on the animals after injection of PFCs, and there were no changes in the release of liver enzymes, although this organ is a major site of PFC accumulation.

An interesting observation of this study was that lymph nodes are clearly delineated in ¹⁹F images. Although the bulk of PFCs were found in CD11b-positive cells, it should be noted that about 20% of the injected particles were taken up by B cells (Fig. 3B). However, it is difficult to decide whether the labelling of lymph nodes is due to trapping of labelled B cells or results from the accumulation of PFCs in resident macrophages. Therefore, it can not be excluded that local PFC deposition may also occur via an alternative pathway: nanoparticles carried by the lymphatic flow to the sites of inflammation could have been taken up by immunocompetent cells already present at the sites of injury before PFC injection.

Perfluorocarbons such as perflubron have been evaluated clinically as artificial blood substitute. In these early studies it was already observed that perflubron is phagocytized by the reticuloendothelial system.^22,23 In principle, perflubron should thus work as well as perfluoro-15-crown-5 ether used in the present study for ¹⁹F imaging of inflammatory processes. Perflubron has the additional advantage that it is readily cleared from the body through exhalation by the lungs within one week.²⁴ Viewed from the MRI side, perflubron has a lower MRI sensitivity caused by signal splitting due to magnetically different ¹⁹F nuclei. However, this problem can be overcome by dedicated detection methods,²⁵ the incorporation of gadolinium into the PFC droplets,²⁶ or the preparation of emulsions with a higher PFC content. Furthermore, it should be noted that the voxel size in cardiac MR diagnostics at 3 T is in the range of 2-30 μl, while it was only 0.2-0.4 μl in our study at 9.4 T, which translates in a substantial sensitivity increase in the clinical setting.