TECHNICAL REPORTS

In vivo magnetic resonance imaging of acute brain inammation using microparticles of iron oxide
2007 Nature Publishing Group http://www.nature.com/naturemedicine
Martina A McAteer1,4, Nicola R Sibson2,4, Constantin von zur Muhlen1, Jurgen E Schneider1, Andrew S Lowe2, Nicholas Warrick1, Keith M Channon1, Daniel C Anthony3 & Robin P Choudhury1
Multiple sclerosis is a disease of the central nervous system that is associated with leukocyte recruitment and subsequent inammation, demyelination and axonal loss. Endothelial vascular cell adhesion molecule-1 (VCAM-1) and its ligand, a4b1 integrin, are key mediators of leukocyte recruitment, and selective inhibitors that bind to the a4 subunit of a4b1 substantially reduce clinical relapse in multiple sclerosis. Urgently needed is a molecular imaging technique to accelerate diagnosis, to quantify disease activity and to guide specic therapy. Here we report in vivo detection of VCAM-1 in acute brain inammation, by magnetic resonance imaging in a mouse model, at a time when pathology is otherwise undetectable. Antibody-conjugated microparticles carrying a large amount of iron oxide provide potent, quantiable contrast effects that delineate the architecture of activated cerebral blood vessels. Their rapid clearance from blood results in minimal background contrast. This technology is adaptable to monitor the expression of endovascular molecules in vivo in various pathologies. Multiple sclerosis is a disease of the central nervous system characterized by multifocal white matter lesions1. At present, the diagnostic criteria for multiple sclerosis, which incorporate both clinical and magnetic resonance imaging (MRI) characteristics, require the demonstration of lesion dissemination in both time and space2,3. Transverse relaxation time (T2)-weighted and gadolinium-enhanced longitudinal relaxation time (T1)-weighted MRI detects some but not all lesions, whereas advanced MRI techniques such as diffusion imaging4, magnetization transfer5 and magnetic resonance spectroscopy6 may provide additional insights. These approaches are, however, limited in two key respects: rst, they image downstream injury, reecting relatively advanced pathology; and second, although they provide an indication of severity, they cannot accurately assess disease activity7. There is a pressing need for molecular imaging techniques that can identify early pathogenesis to accelerate accurate diagnosis and to guide specic therapy. VCAM-1 and its ligand, a4b1 integrin (also called very late antigen4, VLA-4), are important mediators of mononuclear leukocyte recruitment and lesion initiation8. VCAM-1 is not constitutively expressed on cerebral vascular endothelium, but is upregulated with endothelial activation9. Selective blockade of this interaction in experimental autoimmune encephalitis, the rodent analog of multiple sclerosis, results in abolition of both lymphocyte recruitment and the paralysis that usually follows10. Similarly, selective adhesion molecule inhibitors that bind to a4b1 block its association with VCAM-1, resulting in a substantial reduction in both new or enlarging lesions on MRI and clinical relapse in multiple sclerosis11. We reasoned that the sensitivity of MRI for detecting early inammation might be enhanced by using a molecular contrast agent targeted to VCAM-1. Such an approach might potentially provide a more precise and earlier diagnosis, and insights into disease activity, prognosis and response to specic therapy. Microparticles of iron oxide (MPIO, size range 0.761.63 mm) have been used for cellular imaging and tracking12. For some molecular imaging applications, the size of these particles would preclude delivery to the target site. For imaging endovascular structures, however, MPIO possess several positive attributes. First, MPIO convey an amount of iron that is orders of magnitude greater than that conveyed by the ultrasmall particles of iron oxide (USPIO) that have been used previously for MRI contrast. Second, the effects of MPIO on local magnetic eld homogeneity, and therefore on detectable contrast, extend for a distance roughly 50 times the physical diameter of the microparticle12. Third, the size of MPIO means that these particles are less susceptible than USPIO to extravasation or nonspecic uptake by endothelial cells and therefore they retain specicity for molecular targets13. Accordingly, we have developed a VCAM-1 antibodyconjugated 1-mm MPIO that shows specic and quantitative binding to activated endothelial cells in culture. We report that, when coupled with MRI, this targeted MPIO contrast agent detects VCAM-1 expression in vivo in mouse brain inammation with high specicity and exceptional conspicuity. RESULTS VCAM-MPIO bind to TNF-a stimulated sEND-1 cells in vitro Monoclonal antibodies to VCAM-1 were conjugated to 1-mm diameter MPIO (hereafter called VCAM-MPIO). We tested the capacity of this construct for specic and quantitative binding in vitro on cells

1Department of Cardiovascular Medicine, John Radcliffe Hospital, University of Oxford, Oxford OX3 9DU, UK. 2Experimental Neuroimaging Group, Department of Physiology, Anatomy and Genetics, Sherrington Building, Parks Road, Oxford OX1 3PT, UK. 3Department of Pharmacology, Manseld Road, Oxford OX1 3QX, UK. 4These authors contributed equally to this work. Correspondence should be addressed to R.P.C. (robin.choudhury@cardiov.ox.ac.uk).
Received 24 May; accepted 13 July; published online 23 September 2007; doi:10.1038/nm1631

NATURE MEDICINE

ADVANCE ONLINE PUBLICATION
of a mouse endothelial line (sEND-1) that b 120 a were exposed to graded doses of tumor100 necrosis factor-a (TNF-a), an inammatory 80 stimulus used to provoke surface expression of VCAM-1. After extensive washing, antiR 2 = 0.body binding to MPIO was quantied under 0 differential interference contrast microscopy. ng/ml TNF- 0 ng/ml TNF- 2 ng/ml TNF- 10 ng/ml TNF- VCAM-MPIO was retained sparsely by unIgG-MPIO VCAM-MPIO VCAM-MPIO VCAM-MPIO Dose TNF- (ng/ml) stimulated sEND-1 cells, reecting low * basal expression of VCAM-1. The number c d 400 * of VCAM-MPIO bound to sEND-1 cells 300 increased in response to increasing doses of TNF-a (Fig. 1a,b). Isotype IgG-MPIO nega200 tive control constructs did not bind to 100 TNF-astimulated sEND-1 cells. To demonFc-VCAM+ Fc-ICAM+ 0 strate specic retention further, we preNo block Fc-VCAM Fc-ICAM incubated VCAM-MPIO with a chimeric protein containing the extracellular domain Figure 1 MPIO binding to cultured sEND-1 cells. (a) After stimulation with TNF-a (010 ng/ml), cells of VCAM-1 (Fc-VCAM-1). Blocking with were exposed to VCAM-MPIO or isotype IgG-MPIO. In the absence of TNF-a, there was minimal VCAMthis soluble ligand almost completely abol- MPIO retention by sEND-1 cells. Scale bar, 10 mm. (b) Dose-dependent retention of VCAM-MPIO in ished subsequent VCAM-MPIO retention by response to incremental doses of TNF-a (R2 0.94, P o 0.01). Binding persisted after extensive TNF-astimulated sEND-1 cells. By contrast, washing and was restricted to cellular areas. (c) Confocal microscopy of sEND-1 cells stimulated with TNF-a. (50 ng/ml). Green uorescence reects expression of VCAM-1 on the cell surface. Prior pre-incubation with the soluble extracellular incubation of VCAM-MPIO with Fc-ICAM-1 had no effect on VCAM-MPIO binding (autouorescent domain of intercellular cell adhesion mole- yellow-green spheres, arrows), whereas pre-incubation with Fc-VCAM-1 abolished VCAM-MPIO retention cule-1 (Fc-ICAM-1) had no effect on VCAM- by cells almost completely, despite demonstrable surface expression of VCAM-1. sEND-1 cell nuclei are MPIO retention, as assessed by confocal stained blue. Scale bar, 5 mm. (d) Retained VCAM-MPIO (mean s.d.) after TNF-a stimulation with and without pre-incubation with soluble Fc-VCAM-1 or Fc-ICAM-1 (*P o 0.0001). microscopy (Fig. 1c,d). Flow cytometry conrmed that sEND-1 cells had low basal expression of VCAM-1, but showed strong micro-injection to the left striatum. We subsequently administered upregulation of VCAM-1 with TNF-a (Fig. 2a). Pre-incubation of antibody-MPIO constructs systemically by tail vein injection. CirculaVCAM-1 antibody with soluble Fc-VCAM-1 specically inhibited tion time allowed for both specic binding in the brain and clearance VCAM-1 binding, whereas pre-incubation of VCAM-1 antibody of unbound contrast from the blood. Mice underwent MRI, under general anesthesia, 4.55.5 h after injection of IL-1b and 1.52.0 h with Fc-ICAM-1 had no effect (Fig. 2b,c). after administration of contrast agent. VCAM-MPIO caused a marked MRI contrast effect manifest as In vivo MRI detects VCAM-MPIO bound to brain endothelium To induce endothelial activation and VCAM-1 expression in vivo, intensely low signal areas that appeared to delineate blood vessels on mice were given interleukin-1b (IL-1b) by unilateral stereotaxic the IL-1binjected side of the brain. Nonspecic retention was almost absent from the non-injected hemisphere (Figs. 3 and 4 and Supplementary Video 1 online). The dynamics of leukocyte-endothelial binding are complex and depend on multiple ligand-receptor intera 100 b 100 actions. To mimic leukocyte binding more closely, dual antibody Fc-ICAM TNF-+ Fc-VCAM Basal conjugated MPIO were constructed, targeting both VCAM-1 and No block IgG P-selectin. These dual antibodyconjugated MPIO also bound specically but did not further enhance contrast effects (Fig. 3b). Control mice that underwent the same injection regime with an irrelevant isotype antibodyconjugated MPIO showed no contrast effect (Fig. 3c). Pre-treatment of mice with VCAM-1 antibody 30 min before VCAM VCAM VCAM-MPIO administration abolished retention of VCAM-MPIO despite IL-1b injection (Figs. 3d and 4b and Supplementary Video c * * online). Similarly, control mice (with no intracerebral injection

Cell number (relative) Median fluorescence (VCAM) 25 0
AM V TN CA VC F M Fc AM - + -V TN VC CA F- A M+ + Fc M -IC TN AM F+ + G F+ Ig G Ig VC

Cell number (relative)

Figure 2 Flow cytometry. (a) Low basal expression of VCAM-1 by sEND-1 cells with marked upregulation in response to TNF-a. (b) Fc-VCAM-1 potently and specically inhibited the interaction of VCAM-1 antibody with sEND-1 cells, whereas Fc-ICAM-1 had no effect. (c) Quantitative analysis conrmed the absence of nonspecic binding of IgG-MPIO both basally (IgG) and in the presence of TNF-a (IgG TNF-a+), and low VCAM-1 expression under basal conditions (VCAM) with marked upregulation after TNF-a stimulation (VCAM TNF-a+). Antibody binding to VCAM was almost completely abrogated by Fc-VCAM-1, whereas Fc-ICAM-1 had no effect (*P o 0.001).

MPIO per field

Figure 3 In vivo T2*-weighted coronal images from 3D gradient echo data sets each with B90-mm isotropic resolution. Four images are shown per brain. (a) Mouse injected intrastriatally with 1 ng of IL-1b in 1 ml of saline 3 h before intravenous injection of VCAM-MPIO (B4.5 mg iron per kg). Intense low signal areas (black) on the left side of the brain reect the specic MPIO retention on acutely activated vascular endothelium with almost absent contrast effect in the contralateral control hemisphere. (b) Similar, unilateral MPIO contrast effects in a mouse injected as in a but with VCAM+P-selectin-MPIO. (c) Absence of MPIO effects in a mouse injected as in a but with IgG-MPIO control. (d) Absence of MPIO effects in a mouse injected with IL-1b into the striatum and with VCAM-MPIO intravenously after pretreatment with VCAM-1 antibody, which effectively blocked VCAM-MPIO binding. In ad, MRI data were obtained 12 h after MPIO injection. Scale bar, 5 mm.
or with injection of normal saline vehicle only) that received VCAMMPIO systemically showed no specic contrast effects. To appreciate the extent and architecture of the contrast effect, segmented areas were rendered to create a three-dimensional (3D) volumetric map of contrast binding that clearly demonstrated delineation of vascular structures in the IL-1bstimulated hemisphere and almost total absence of binding on the contralateral, non-activated side (Fig. 4a). Pre-treatment to block VCAM-1 abolished VCAMMPIO retention (Fig. 4b). Quantitative analysis of MPIO binding As compared with brains without IL-1b injection, specic contrast was increased more than 100-fold (3,999 1,mm3 versus 106 mm3, mean s.d.; P 0.02) after administration of VCAM-MPIO. No further increase in specic contrast was observed for the dual VCAM and P-selectin antibodyconjugated MPIO (Fig. 4c). Distribution of MPIO on histology Histological examination indicated that VCAM-MPIO lined venules in the IL-1bstimulated hemisphere (per section) and

showed sparse retention in the contralateral, non-activated hemisphere (3 4, P o 0.0001; Fig. 5). MPIO were conned to the lumen of the vessel without extravasation. Isolated MPIO were the most common nding. Small clusters, similar to those seen in vitro, were present in relatively small numbers (Fig. 5b,c). Phagocytic cells with MPIO within their cytoplasm were occasionally identied. Expression of VCAM-1 was conrmed by immunohistochemistry, showing a distribution that was limited to vascular endothelium (Fig. 5b). Safety and tolerability Injection of antibody-conjugated MPIO was well tolerated in all mice. None of the 16 mice showed signs of ill effect during close observation for up to 5 h after injection.
Figure 4 Three-dimensional volumetric maps of VCAM-MPIO binding and quantitative analyses of MPIO contrast effects. VCAM-MPIO is shown in red. (a) In each mouse, 41 contiguous images were segmented by an automated analysis of signal intensity histograms. MPIO contrast effects delineated the architecture of cerebral vasculature in the IL-1bstimulated hemisphere (left half of image) with almost total absence of binding on the contralateral, non-activated side. The midlines are indicated by vertical sections. (b) Pre-administration of VCAM-1 antibody abolished VCAM-MPIO retention. (c) As compared with brains without IL-1b injection, specic contrast was increased 4100-fold after administration of VCAM-MPIO. Dual conjugated MPIO targeting both VCAM-1 and P-selectin also bound specically but did not further enhance contrast effects. Substitution of IgG-MPIO (IgG/IL-1b+), sham intracerebral injection (VCAM/NaCl), no intracerebral injection (VCAM/IL-1b) and pre-blocking (VCAM/IL-1b+ with block) were not associated with specic contrast effects. Bars indicate mean values for each group (*P 0.02).
Specific contrast ( 106 m3)
9,000 8,000 7,000 6,000 5,000 4,000 3,000 2,000 1,000 0
-1 + + IL -1 IL -1 + A IL M + -1 P + se l VC IL N -1 aC l

Counts per section

sMPIO cMPIO IL-1+

sMPIO cMPIO IL-1

Figure 5 Postmortem light micrographs of mouse brain. (a) Cresyl violet staining shows MPIO lining a venule on the injected side of the brain. MPIO were conned to the lumen of the vessel (thin arrows). Binding most often comprised isolated MPIO. (b) Occasional clusters of MPIO were seen (clear arrow), some of which appeared to reect phagocytic cells with MPIO within their cytoplasm. Presence of VCAM-1 (brown) was conrmed by immunohistochemistry, which showed expression limited to vascular endothelium (large arrows) and colocalization of VCAM-1 immunostaining and in vivo VCAM-MPIO binding. (c) Quantication of MPIO binding in the IL-1bstimulated (IL-1b+) and unstimulated hemispheres of mice receiving VCAM-MPIO demonstrated retention of MPIO in the former with a strong preponderance of single MPIO particles (sMPIO) with less frequent clusters (cMPIO; *P o 0.0001). Scale bar, 10 mm.
DISCUSSION We have reported the development and application of a molecular imaging probe that facilitates the identication of VCAM-1 expression in mouse brain in vivo by MRI. The specicity and potency of the contrast effects are marked and derive from a combination of targeted delivery of a large amount of iron oxide to sites of early inammation and rapid clearance of MPIO from the blood. The appeals of this approach are manifold. In the context of multiple sclerosis, this technique images a process that occurs early in disease pathogenesis. Unlike existing techniques, our approach does not require tissue destruction or compromise to the blood-brain barrier. Inhibiting the interaction between leukocytes and VCAM-1 has clinical benet11, and the ability to image VCAM-1 raises the possibility of targeting treatment to individuals with increased VCAM-1 expression. VCAM-1 participates in other inammatory conditions, including atherogenesis14,15, transplant rejection16 and cancer17, for which therapeutic VCAM-1 targeting may also be effective18,19. More broadly, by modifying the ligand, MPIO constructs could readily be adapted to image other endovascular targets that are differentially expressed in a broad range of pathologies. Ultrasound has shown promise for imaging endothelial molecules with targeted microbubbles20, and an enhanced technique using sensitive particle acoustic quantication has been applied to measure ICAM-1 and VCAM-1 in rat brain21. Ultrasound techniques are, however, inherently limited by the need for acoustic windows and ultrasound cannot reliably penetrate human skull. By contrast, MRI provides exquisite spatial resolution and tissue contrast, without ionizing radiation, making it the imaging modality of choice for many brain pathologies and driving the need for molecular magnetic resonance contrast.

Gadolinium-based contrast agents shorten T1, providing positive contrast on T1-weighted images22. Expression of ICAM-1 in rat brain has been imaged ex vivo by using paramagnetic liposomes23, and we have targeted E-selectin expression in rat brain in vivo by using a sialyl Lewisx moiety conjugated to Gd-DTPA24. The small quantity of gadolinium that can be delivered to an endothelial monolayer, however, limits its contrast effect. By comparison, USPIO provide greater contrast but may require sophisticated ligands that mediate internalization by endothelial cells to achieve adequate local concentrations25. USPIO have become popular owing to their long half-life in blood, which is a positive attribute for applications such as the measurement of changes in cerebral perfusion. This property is more of a hindrance in targeted contrast agents, however, because it leads to high background contrast for an extended period. A further potential drawback of USPIO is that contrast is manifest in T2*-weighted images as indistinct areas of low signal that can be difcult to distinguish from the ordinary heterogeneity of normal tissue. In addition, because USPIO can be taken up nonspecically by endothelial cells, there is potential to compromise the specicity of molecular targeting13. MPIO convey an amount of iron that is orders of magnitude greater than USPIO, and cause a local magnetic eld inhomogeneity extending for a distance roughly 50 times their physical diameter12. Although smaller than leukocytes and thus not prone to small-vessel plugging, MPIO preclude translocation across the endothelium owing to their size and incompressible nature, as conrmed by histology. The pattern of MPIO binding appeared almost identical to the patterns of lymphocyte binding in venules in the rat experimental autoimmune encephalomyelitis model of multiple sclerosis10. Trials have demonstrated clear clinical benets from inhibition of the interaction between VCAM-1 and its ligand11. The ability to image VCAM-1 expression, in conjunction with existing diagnostic approaches, may offer clinically important opportunities to enhance specicity and to accelerate diagnosis. Early diagnosis and delivery of specic guided intervention may improve outcomes and allow response to treatment to be more precisely monitored. Our approach uses commercially available reagents to provide a generic platform technology for endovascular molecular MRI and potentially allows the substitution of alternative ligands. With respect to translation to clinical use, our approach has three key factors. First, MPIO size. At the doses used here, short-term ill effects were not seen in mice, nor was there evidence of tissue infarction owing to smallvessel plugging. Indeed in vivo imaging and subsequent histological analysis of non-injected hemispheres both conrmed that nonspecic MPIO retention was minor. Second, MPIO composition. The MPIO used here are nonbiodegradable and are not suitable for use in humans. Iron oxidecontaining contrast media are, however, in clinical use, and it should be feasible to synthesize biodegradable particles26,27. Third, iron dose. The dose of iron (4.5 mg iron per kg (body weight)) reects closely the dose of 2.6 mg per kg that has been used extensively for human oncological MRI with USPIO28, and is considerably less than that used (30 mg per kg) in a study targeting USPIO to image VCAM-1 in mice25. In conclusion, this molecular imaging approach manifests exceptionally potent contrast effects in MRI. In a mouse model of acute inammation, cerebral blood vessels were delineated and VCAM-MPIO binding was quantied. Alternative ligand-MPIO constructs provide clear opportunities for diagnostic imaging using specic endothelial cell markers that are differentially expressed in various pathologies, including inammatory diseases, cancer and atherothrombosis.

METHODS

Antibody conjugation to MPIO. We used myOne tosylactivated MPIO (1-mm diameter; iron content 26%) with p-toluenesulphonyl (tosyl)-reactive surface groups (Invitrogen) for antibody conjugation. We washed MPIO with sodium borate buffer (0.1 M, pH 9.5) and added puried monoclonal rat antibodies specic to mouse VCAM-1 (clone M/K2, Cambridge Bioscience), IgG-1 (clone Lo-DNP-1, Serotec) or a combination (50/50 wt/wt) of VCAM-1 and P-selectin (clone RB40.34, Fitzgerald Industries; MPIO per 40 mg of antibody for all). We added 3 M ammonium sulfate to give a nal concentration of 1 M. We incubated the solution with constant rotation at 37 1C for 20 h. After incubation, we collected MPIO by using a Dynal magnet (Invitrogen) and discarded the supernatant containing any unbound antibody. We added PBS plus 0.5% BSA and 0.05% Tween 20 (pH 7.4) and incubated MPIO at 37 1C overnight to block the remaining active tosyl sites. We washed MPIO with PBS plus 0.1% BSA and 0.05% Tween 20 at 4 1C before storing at a concentration of 2.MPIO per ml of PBS plus 0.1% BSA and 0.05% Tween 20 at 4 1C. We calculated that the primary amine and sulfydryl groups of 1-mm tosylactivated MPIO have a capacity to bind covalently 1.IgG molecules per MPIO. Mouse protocol. After anesthesia with isouorane (2.02.5% in 70% N2O: 30% O2), adult male NMRI mice (g) were placed in a stereotaxic frame under a Wild M650 operating microscope (Leica). Using a glass pipette (tip o50 mm), we stereotaxically injected 1 ng of mouse recombinant IL-1b in 1 ml of low endotoxin saline into the left striatum (coordinates from Bregma: anterior 0.5 mm, lateral 2 mm, depth 2.5 mm), to induce endothelial activation. After 3.1 0.2 h, we injected mice through a tail vein with VCAM-MPIO, VCAM+P-selectinMPIO, or IgG-MPIO (microparticles; B4.5 mg iron per kg; n 3 per group). We had two control groups of mice (n 2 per group): in one, mice were injected intracerebrally with 1 ml of low endotoxin saline; in the other, mice were not injected intracerebrally. We subsequently administered VCAM-MPIO intravenously to both control groups. To determine selectivity of VCAM-MPIO, we injected a further group of mice (n 3) with 0.2 mg of VCAM-1 antibody per kg at 3.3 0.4 h after intracerebral IL-1b injection to block VCAM-1binding sites. We subsequently administered VCAM-MPIO 15 min later. After VCAM-MPIO injection, mice were placed in a quadrature birdcage coil with an in-built stereotaxic frame for imaging. We maintained anesthesia with 1.51.8% isouorane in 70% N2O/30% O2, monitored electrocardiography, and maintained body temperature at B37 1C with a circulating warm water system. All procedures were approved by the UK Home Ofce. In vivo MRI. We used a 7-Tesla horizontal bore magnet with an Inova spectrometer (Varian) to acquire a T2*-weighted 3D gradient echo data set with the following parameters: ip angle, 351; repetition time, 50 ms; echo time, 5 ms; eld of view, 22.5 22.5 31.6 mm3; matrix size, 360; two averages; total acquisition time, B1 h. The mid-point of acquisition was 1.7 0.5 h after MPIO injection. We serially imaged the same mouse and found maximal contrast at 12 h with diminution by 4 h (data not shown). We zero-lled the data to 360 and reconstructed it off-line, giving a nal isotropic resolution of 88 mm3. MRI analysis. For each magnetic resonance image, we manually masked the brain to exclude extracerebral structures. We segmented low-signal areas in 41 contiguous images, spanning a depth of 3.6 mm from the dorsal hippocampus ventrally. To control for minor variations in absolute signal intensity between individual scans, we calibrated low-signal areas on ten evenly spaced slices per brain. We applied the median signal intensity value to the fully automated, histogram-based batch analysis of the 41-slice sequence. We extracted data for left and right sides of the brain simultaneously with identical parameters. We summated voxel volumes and expressed them as raw volumes in cubic micrometers without surface rendering or smoothing effects. To ensure true laterality, we quantied contrast in each hemisphere 1 mm from the midline outwards. ImagePro Plus (Media Cybernetics) was used to segment and to quantify contrast volume by an operator who was blind to the origin of all data. We present the data as specic contrast, dened as left minus right contrast volume. Additional methods. Information on cell culture and histology is available in the Supplementary Methods online. Statistical analysis. Data are expressed as mean s.d. and compared, where indicated, by two-tailed t-tests. We assigned statistical signicance at P o 0.05.

Note: Supplementary information is available on the Nature Medicine website. ACKNOWLEDGMENTS We thank W. N. Haining for expertise in FACS analysis; T. Bannister for image analysis; D.R. Greaves for critical appraisal of the manuscript; and P. Townsend for overall laboratory management. This work was funded by the Wellcome Trust (R.P.C.) and the Medical Research Council (N.R.S. and D.C.A.). AUTHOR CONTRIBUTIONS R.P.C. and M.A.M. designed the contrast agent. M.A.M. manufactured the contrast agent and, in conjunction with N.W., K.M.C., C.v.z.M. and J.E.S., undertook the in vitro experiments. N.R.S., D.C.A., R.P.C. and M.A.M. designed the in vivo experiments. N.R.S., A.S.L. and D.C.A. conducted the MRI component, and C.v.z.M. and D.C.A. undertook histological analysis. R.P.C. supervised image analysis and analyzed the data. M.A.M., N.R.S. and R.P.C. contributed to the writing of the manuscript, and all authors discussed and rened the manuscript. COMPETING INTERESTS STATEMENT The authors declare competing nancial interests: details accompany the full-text HTML version of the paper at www.nature.com/naturemedicine/.
Published online at http://www.nature.com/naturemedicine Reprints and permissions information is available online at http://npg.nature.com/ reprintsandpermissions
1. Compston, A. & Coles, A. Multiple sclerosis. Lancet 359, 12211231 (2002). 2. McDonald, W.I. et al. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the Diagnosis of Multiple Sclerosis. Ann. Neurol. 50, 121127 (2001). 3. Polman, C.H. et al. Diagnostic criteria for multiple sclerosis: 2005 revisions to the McDonald Criteria. Ann. Neurol. 58, 840846 (2005). 4. Goldberg-Zimring, D., Mewes, A.U.J., Maddah, M. & Wareld, S.K. Diffusion tensor magnetic resonance imaging in multiple sclerosis. J. Neuroimaging 15, 68S81S (2005). 5. Ropele, S. et al. A comparison of magnetization transfer ratio, magnetization transfer rate, and the native relaxation time of water protons related to relapsing-remitting multiple sclerosis. AJNR Am. J. Neuroradiol. 21, 18851891 (2000). 6. Bitsch, A. et al. Inammatory CNS demyelination: histopathologic correlation with in vivo quantitative proton MR spectroscopy. AJNR Am. J. Neuroradiol. 20, 16191627 (1999). 7. Guttmann, C.R., Meier, D.S. & Holland, C.M. Can MRI reveal phenotypes of multiple sclerosis? Magn. Reson. Imaging 24, 475481 (2006). 8. Elices, M.J. et al. VCAM-1 on activated endothelium interacts with the leukocyte integrin VLA-4 at a site distinct from the VLA-4/bronectin binding site. Cell 60, 577584 (1990). 9. Carlos, T.M. et al. Vascular cell adhesion molecule-1 mediates lymphocyte adherence to cytokine-activated cultured human endothelial cells. Blood 76, 965970 (1990). 10. Yednock, T.A. et al. Prevention of experimental autoimmune encephalomyelitis by antibodies against a4b1 integrin. Nature 356, 6366 (1992). 11. Polman, C.H. et al. A randomized, placebo-controlled trial of natalizumab for relapsing multiple sclerosis. N. Engl. J. Med. 354, 899910 (2006). 12. Shapiro, E.M. et al. MRI detection of single particles for cellular imaging. Proc. Natl. Acad. Sci. USA 101, 1090110906 (2004). 13. Briley-Saebo, K. et al. Hepatic cellular distribution and degradation of iron oxide nanoparticles following single intravenous injection in rats: implications for magnetic resonance imaging. Cell Tissue Res. 316, 315323 (2004). 14. Cybulsky, M.I. et al. A major role for VCAM-1, but not ICAM-1, in early atherosclerosis. J. Clin. Invest. 107, 12551262 (2001). 15. Nakashima, Y., Raines, E.W., Plump, A.S., Breslow, J.L. & Ross, R. Upregulation of VCAM-1 and ICAM-1 at atherosclerosis-prone sites on the endothelium in the ApoE-decient mouse. Arterioscler. Thromb. Vasc. Biol. 18, 842851 (1998). 16. Orosz, C.G. et al. Role of the endothelial adhesion molecule VCAM in murine cardiac allograft rejection. Immunol. Lett. 32, 712 (1992). 17. Maurer, C.A. et al. Over-expression of ICAM-1, VCAM-1 and ELAM-1 might inuence tumor progression in colorectal cancer. Int. J. Cancer 79, 7681 (1998). 18. Dedrick, R.L., Bodary, S. & Garovoy, M.R. Adhesion molecules as therapeutic targets for autoimmune diseases and transplant rejection. Expert Opin. Biol. Ther. 3, 8595 (2003). 19. Gosk, S., Gottstein, C. & Bendas, G. Targeting of immunoliposomes to endothelial cells expressing VCAM: a future strategy in cancer therapy. Int. J. Clin. Pharmacol. Ther. 43, 581582 (2005).

20. Villanueva, F.S. et al. Microbubbles targeted to intercellular adhesion molecule-1 bind to activated coronary artery endothelial cells. Circulation 98, 15 (1998). 21. Reinhardt, M. et al. Ultrasound derived imaging and quantication of cell adhesion molecules in experimental autoimmune encephalomyelitis (EAE) by Sensitive Particle Acoustic Quantication (SPAQ). Neuroimage 27, 267278 (2005). 22. Aime, S. et al. Insights into the use of paramagnetic Gd(III) complexes in MRmolecular imaging investigations. J. Magn. Reson. Imaging 16, 394406 (2002). 23. Sipkins, D.A. et al. ICAM-1 expression in autoimmune encephalitis visualized using magnetic resonance imaging. J. Neuroimmunol. 104, 19 (2000). 24. Sibson, N.R. et al. MRI detection of early endothelial activation in brain inammation. Magn. Reson. Med. 51, 248252 (2004). 25. Nahrendorf, M. et al. Noninvasive vascular cell adhesion molecule-1 imaging identies inammatory activation of cells in atherosclerosis. Circulation 114, 15041511 (2006). 26. Sakhalkar, H.S. et al. Leukocyte-inspired biodegradable particles that selectively and avidly adhere to inamed endothelium in vitro and in vivo. Proc. Natl. Acad. Sci. USA 100, 1589515900 (2003). 27. Chen, H.H. et al. MR imaging of biodegradable polymeric microparticles: a potential method of monitoring local drug delivery. Magn. Reson. Med. 53, 614620 (2005). 28. Will, O. et al. Diagnostic precision of nanoparticle-enhanced MRI for lymph-node metastases: a meta-analysis. Lancet Oncol. 7, 5260 (2006).

Imaging cell fate by quantitative and functional MRI
Monique Bernsen Dept. of Radiology Erasmus MC Rotterdam; The Netherlands

www.encite.org

Cell imaging by MRI
Cell labeling with paramagnetic, superparamagnetic or 19F compounds
Paramagnetic Gd-chelates Superparamagnetic Iron oxide particles 19F compounds

Higuchi et al.

Sensitive visualization possible
GdHPDO3A-labelled pancreatic islets in liver in vivo Single SPIO-labelled cells in vitro

Zhang et al.

Biancone et al.
Cell quantification by MRI
Cell quantification - Signal intensity changes

0.15 mM

0.45 mM

0.80 mM

Limitations/pitfalls in using signal intensity changes for quantification
- Change in subject position - Inhomogeneity of coil sensitivity
Cell quantification - Relaxation rate (R2 or R2*) Independent of coil bias/signal homogeneity

R2* map values

0.3 y = 0.0251x + 0.0048 0.25 R2* values 1/ms R = 0.9839
6 MPIO concentration au. 12

Kotek et al.

Cell quantification - Relaxation rate

Rad et al.

Cell quantification - Labelled cell implants in cartilage defects

Cartilage

Cells seeded in defect
Control 10,000 cells Bone

Van Buul et al.

Cartilag e
2 mm Cells seeded in defect 2 mm
10,000 cells Bone Control (0%)

2 mm 2 mm

100,000 cells (10%)

1,000,000 cells (100%)

Limitations/pitfalls in using relaxation rate for quantification - Relaxation rate dependent on compartmentalization - Relaxation rate dependent on intra-voxel distribution - Relaxation rate dependent on particle clustering
Relaxation rate dependent on compartmentalization
Relaxation rate dependent on intra-voxel distribution

Mohammadi-Nejadet al.

R2* relaxation
cell dilution cell division

R2* (1/s) AU

0 0,0 0,5 1,0 1,5 2,0 2,5

Fe concentration (g/ml)

Mohammadi-Nejad et al.
Relaxation rate dependent on particle clustering
Relaxation rate variations

SPIO dilution

SPIO complex MPIO dilution cell dilution

cell division

Distinguishing life from dead cells

R2' relaxation

SPIO complex cell dilution

R2* (1/s)

R2' (1/s)

Kuhlpeter et al.

Distinguishing proliferation

R2 relaxation

80 R2 (1/s) 60

0 0,00

0,60 Fe concentration (g/ml)

Collaborators

Dept of Radiology Sandra van Tiel Dept of Surgical Oncology Gerben Koning Timo ten Hagen Lex Eggermont Dept of Orthopedics Eric Farrell Gerben van Buul Nicole Kops Koen Bos Gerjo van Osch Harrie Weinans Dept of Cell Biology and Genetics. Paula van Heijningen Jeroen Essers Dept of Neonatology Ingrid Renes Dept of Exp. Cardiology Heleen van Beusekom Dirk Duncker Dept of Cardiology Robert-Jan van Geuns Wim van der Giessen TU Delft Ulla Woronieck TU Eindhoven Gustav Strijkers Klaas Nicolay Dept of Nuclear Med. Magda Bijster Marion de Jong GE Healthcare Gavin Houston

Jamal Guenoun

Joost Haeck
Gaby Doeswijk Gerben van Buul
Leila Alic Jifke Veenland Erik Meijring Ihor Smal Esben Plenge
HenkSmit Stefan Klein Gyula Kotek
Piotr Wielopolski Wiro Niessen Gabriel Krestin

FP7-ENCITE

ENCITE - European Network for Cell Imaging and Tracking Expertise
Call: EU financial contribution: Project duration: Cooperation Health-2007-1.2-4 In vivo image-guidance for cell therapy, Large-scale integrating project (IP) 12m 1 June May 2012

Consortium partners:

European Institute for Biomedical Imaging Research (Coordinator), AT Erasmus Universitair Medisch Centrum Rotterdam, NL King's College London, UK Weizmann Institute of Science, IL Max-Planck Institut fr Neurologische Forschung, DE Tel Aviv University, IL Universit di Torino, IT Institute for Clinical and Experimental Medicine, CZ University of Freiburg, DE University of Mons-Hainaut, BE University Paris Descartes, FR Friedrich-Alexander University, DE Leiden University Medical Center, NL The University of Milano Bicocca, IT Radboud University Nijmegen Medical Centre, NL
Foundation for Applied Medical Research, ES Institut Curie, FR BioSpace, FR Medres, DE Cage Chemicals, IT University of Navaro, ES The Hebrew University of Jerusalem, IL Westflische Wilhelms-Universitt Mnster, DE Katholieke Universiteit Leuven, BE The Chancellor, Masters and Scholars of the University of Cambridge, UK Agencia Estatal Consejo Superior de Investigaciones Cientficas, ES Consorci Institut Catala de Ciencies Cardiovasculars, ES Vrije Universiteit Medisch Centrum Amsterdam, NL Universittsspital Basel, CH