In vivo imaging of transplanted hepatocytes with a 1.5-T clinical MRI system--initial experience in mice

Abstract

The feasibility of in vitro mature mouse hepatocyte labeling with a novel iron oxide particle was assessed and the ability of 1.5-T magnetic resonance imaging (MRI) to track labeled mouse hepatocytes in syngenic recipient livers following intraportal cell transplantation was tested. Mouse hepatocytes were incubated with anionic iron oxide nanoparticles at various iron concentrations. Cell viability was assessed and iron oxide particle uptake quantified. Labeled hepatocytes were intraportally injected into 20 mice, while unlabeled hepatocytes were injected into two mice. Liver T2 values, spleen-to-muscle relative signal intensity (RI_{spleen/muscle}), and liver-to-muscle relative signal intensity (RI_liver/muscle) on gradient-echo T2-weighted imaging after injection of either labeled or unlabeled hepatocytes were compared with an ANOVA test followed by Fisher’s a posteriori PLSD test. Livers, spleens and lungs were collected for histological analysis. Iron oxide particle uptake was saturable with a maximum iron content of 20 pg per cell and without viability alteration after 3 days of culture. Following labeled-cell transplantation, recipient livers showed well-defined nodular foci of low signal intensity on MRI—consistent with clusters of labeled hepatocytes on pathological analysis—combined with a significant decrease in both liver T2 values and liver-to-muscle RI_liver/muscle (P=0.01) with minimal T2 values demonstrated 8 days after transplantation. Conventional MRI can demonstrate the presence of transplanted iron-labeled mature hepatocytes in mouse liver.

Introduction

Liver transplantation is currently the only curative treatment for a variety of acute and chronic liver diseases. The chronic shortage of donor livers suitable for orthotopic transplantation has generated interest in alternative therapies, including hepatocyte transplantation, especially regarding the treatment of life-threatening inherited metabolic disorders of the liver [1–4]. Clinical trials have already shown the feasibility and safety of transplanting genetically modified autologous and allogenic hepatocytes [5–10]. These approaches could avoid surgical removal and replacement of the entire liver for the correction of a single defective protein [11, 12].

Intrasplenic and intraportal hepatocyte injection [13, 14] both lead to hepatocyte integration within the liver parenchymal plates [15]. It is, however, well established that in normal rodent liver the proportion of transplanted hepatocytes which engraft in the liver parenchyma is low, approximately 1% [16]. A significant repopulation with transplanted cells is reached only in cases of acute or chronic liver injury. This is due to the complex process of cell integration, which involves crossing of the endothelial cell barrier that lines the sinusoids, and disruption of junctions between resident hepatocytes [17]. An important issue in liver-targeted cell therapy is the detection and location of transplanted hepatocytes within the liver parenchyma. This is currently achieved by using recombinant retroviruses carrying reporter genes, such as the Escherichia coli β-galactosidase gene, which stably integrates into the genome of the recipient’s target cells [18–20]. Labeled cells are detected by monitoring β-galactosidase activity in the recipient liver; an approach necessitating repeated liver biopsy.

Cells labeled with superparamagnetic iron oxides (SPIO) can easily be detected in vivo by magnetic resonance imaging (MRI), which has high T2* susceptibility to SPIO [21]. Spontaneous fluid-phase endocytosis can lead to SPIO uptake [22]. However, optimal cell labeling with SPIO requires additional procedures, based on transfection agents or complex nanoparticle structures [23–25]. Larger iron oxide particles appear to have better labeling efficiency [26]. We recently reported the synthesis of a novel iron oxide labeling agent for cellular imaging [27]. Because of their anionic coating, these iron oxide nanoparticles are efficiently and spontaneously endocytosed after electrostatic adsorption to the cellular outer membrane [28].

The aims of this study were twofold: (1) to assess the feasibility of in vitro mature mouse hepatocyte labeling with a novel anionic iron oxide particle, and (2) to assess whether a 1.5-T human MRI system is able to track labeled mature mouse hepatocytes following cellular transplantation in syngenic recipient livers.

Materials and methods

All experiments on mice in this study complied with the French Ministry of Agriculture guidelines and our Intitutional Animal Ethics Committee requirements. A total of 64 C57BL6 mice (8–10 weeks old, weight ranging from 25 to 30 g) were included. Thirty mice were euthanized for hepatocyte isolation. Four mice were used as control animals, and did not undergo liver cell transplantation. The remaining 30 mice were used for imaging studies following hepatocyte transplantation. Figure 1 summarizes the experimental set-up of this study.

Fig. 1.

Experimental set-up. Overall, 64 animals were included in this study. Thirty mice were included for hepatocyte isolation and in-vitro experiments including cell viability assays, and iron oxide quantification. Thirty-four dictinct mice were included for imaging experiments (four control animals, two animals undergoing unlabeled cells injection and 20 undergoing labeled cell injection). Eight animals died during the experimental process

Hepatocyte isolation and labeling with anionic magnetic nanoparticles

Hepatocyte isolation

Hepatocytes were isolated with a two-step collagenase digestion technique [29]. Briefly, the livers of 30 mice were perfused with 50 ml of Ca²⁺/Mg²⁺-free N-(2-hydroxyethyl)-piperazine-N′-2-ethanesulfonic acid (HEPES) buffer (pH 7.4) then with 50 ml of HEPES buffer containing 5 mM Ca (PAA Laboratory, France) and 0.02% (w/v) collagenase B. Digested tissues were filtered through 70-μm filters to remove cell aggregates, and were centrifuged (3×5 min) at 50 g. The cells were finally suspended in RPMI culture medium. From the 30 livers used for hepatocyte isolation, a total of 14 provided the hepatocytes for in vivo cell transplantation. The remaining livers provided hepatocytes for in vitro experiments (magnetophoresis, cell viability).

Hepatocyte labeling

A colloidal suspension of anionic iron oxide nanoparticles dispersed in water was synthesized with Massart’s method [30]. The superparamagnetic particles thus obtained consist of monocrystals of maghemite with negative citrate charges on their surface and a mean diameter of 8 nm. Isolated hepatocytes retrieved from previously digested mouse livers were incubated for 15 min at 37°C in air-5% CO₂ in RPMI medium supplemented with 5 mM sodium citrate and with the nanoparticles at iron concentrations of 0.5 mM, 1 mM, 2 mM and 5 mM. After incubation, the cells were centrifuged at 1,200 rpm for 10 min, further resuspended and incubated in RPMI medium for 90 min at 37°C in air-5% CO₂ in order to permit particle internalization.

Hepatocytes retrieved after labeling procedures and control unlabeled hepatocytes retrieved immediately after liver digestion procedures, were used for cell viability procedures, magnetophoresis quantification or in vivo procedures (see paragraphs below).

Cell viability

All cell viability procedures were performed in triplicate. Samples of labeled hepatocytes (following incubation with the nanoparticles at the four distinct iron concentrations) were plated and cultured for 3 days (Fig. 1). Cell viability was then determined by the Trypan blue exclusion test.

Quantification of hepatocyte iron uptake

We used magnetophoresis to measure iron uptake by labeled hepatocytes, as described elsewhere [31]. Experiments were applied in triplicate to cell suspensions incubated with each of the four iron concentrations. Magnetophoresis is used to measure the migration speed of magnetically labeled cells in a magnetic field gradient in a medium of known viscosity [31]. Labeled hepatocytes were centrifuged at 1,200 rpm for 10 min, and subsequently resuspended in RPMI medium at a density of 10⁴ cells/ml. Then, 400 μl of this suspension was placed in a 1-mm-thick Hellma chamber. Cell displacement in steady-state conditions was recorded with a video camera assembled on the microscope and was analyzed with dedicated image software (NIH, Bethesda, USA). The diameter, displacement of 30–50 cells, and cell velocity distribution were analyzed. Iron mass per cell was deduced from the balance between the friction drag force (proportional to the cell velocity and cell diameter) and the magnetic force experienced by the cell (proportional to the cell iron load) [31].

In vivo MRI

All animals were operated on under general anesthesia, induced by intraperitoneal injection of 0.2 ml of a mixture of 0.9 ml of ketamine hydrochloride (Imalgene, Merial, Lyon, France), 0.2 ml of xylazine (Rompun, Bayer, Germany), and 3.8 ml of saline. Thirty-four syngenic C57BL6 mice—distinct from those used to retrieve hepatocytes—were used (Charles River, France). Three mice died during anesthesia and five during surgery. A total of 26 mice were finally included in the study (Fig. 1): four mice were used as control animals, while the remaining 22 animals underwent liver cell transplantation, whether with labeled or unlabeled hepatocytes.

Hepatocyte transplantation

Each set of transplantation experiments involved two to five mice infused with the same batch of hepatocytes isolated from the liver of one syngenic mouse. Thus, two to five mice were transplanted on the same day with hepatocytes retrieved from the liver of one syngenic mouse. All 22 mice were placed supine, and the abdomen was accessed via a midline and transverse laparotomy. A PVC catheter was inserted in a branch of the superior mesenteric vein and pushed towards the portal trunk. Hepatocytes used for liver cell transplantation were collected immediately after cell labeling procedures with the nanoparticles at an iron concentration of 5 mM, kept at 37°C, and injected into 20 animals. The hepatocytes (total of 8×10⁵ cells in 250 μl) were infused via an automatic pump at a flow rate of 10 μl/s. Control unlabeled hepatocytes (total of 8×10⁵ cells in 250 μl), incubated in RPMI-only medium, were injected into two animals using the same protocol as with labeled hepatocytes.

MRI

MRI was performed with a 1.5-T device (Symphony, Siemens, Erlangen, Germany) and a surface extremity coil 2, 8, 14, 21 and 32 days after cellular transplantation with labeled cells (Table 1). In order to allow systematic pathological examination following each imaging procedure, distinct animals were used for each of these time-points. Each animal was randomly assigned to the delay between labeled cell transplantation and MRI. Mice transplanted with unlabeled cells were imaged 4 days following transplantation. For all imaging procedures the animals were maintained under general anesthesia and further immobilized in tight-fitting individual plastic tubes placed prone within the magnet. Four sequences were acquired (Table 2), comprising transverse and coronal gradient-echo steady-state free precession (GE SSFP), permitting rapid anatomic depiction, transverse T1 spin-echo (SE) weighted imaging (WI), transverse and coronal T2 GE WI, and a quantitative transverse multi-section T2 multi-echo sequence (16-echo train).

Table 1.

MRI timetable: number of mice imaged at the different time-points after labeled-hepatocyte transplantation (total of 20 mice) with mean liver T2 values depicted on the multi-echo T2 sequence

Days after hepatocyte transplantation	Number of mice	Liver T2 (ms) (mean ± standard error)
Controls	4	60.7±4
2	3	62.4±2
8	5	47.4±2
14	3	53.7±5
21	4	53±4
32	5	48.2±8

The mean liver T2 values of mice transplanted with control unlabeled hepatocytes (data not shown) was 63.4±2 ms (4 days after cellular transplantation)

Table 2.

MRI parameters

Sequence no.	Sequence type	Plane	TR/TE/α	Thickness (mm)	FOV (mm2)	Matrix	NSA	Time
1	Localizer	3 planes
2	GE SSFP	Coronal/transverse	7.77/3.9/80	2.5	208×170	256×208	11	1 min 37 s
3	SE T1 WI	Coronal/transverse	380/15/90	2	98×90	512×416	8	5 min 41 s
4	GE T2 WI	Coronal/transverse	235/19/20	3	208×170	172×128	3	3 min 37 s
5	Multiecho T2 WI	Transverse	3000/12.2 (16 TE)/90	3	98×98	256×256	2	11 min 43 s

Image analysis

Relative liver-to-muscle SIs

For each animal, all regions of interest (ROIs) were positioned by one of the authors (A.L.), assisted by NIH Image software. For each liver analysed on MRI using GE T2 sequences, three distinct ROIs were manually drawn, avoiding vessels previously identified on GE SSFP sequences, and two ROIs were manually drawn along the lumbar muscles on the same transverse section as that selected for liver measurements. The T2 signal intensity (SI) of each ROI was collected and averaged to provide both the mean SI_liver and the mean SI_muscle. Similarly, one ROI positioned on the spleen of each mouse providing the spleen SI (SI_spleen). The mean sizes of the liver, muscle, and spleen ROIs were, respectively, 136, 105 and 34 pixels.

Using these SI values, the relative liver-to-muscle SIs (RI_liver/muscle) were determined with Eq. (1):

$${RI}_{\mathit{liver}/\mathit{muscle}}=\left(\frac{{SI}_{\mathit{liver}}}{{SI}_{\mathit{muscle}}}\right).$$ (1)

Similarly, the spleen-to-muscle signal intensities (RI_{spleen/muscle}) were determined with Eq. (2):

$${RI}_{\mathit{Spleen}/\mathit{muscle}}=\left(\frac{{SI}_{\mathit{Spleen}}}{{SI}_{\mathit{muscle}}}\right).$$ (2)

Liver T2 measurements

For each animal, three ROIs were positionned within the liver parenchyma on the multi-section multi-echo T2 sequence avoiding vessels. ROIs were positioned as close as possible to the location selected on GE T2 sequences. The three SI values for each ROI were collected for each TE and averaged. The resulting quantitative T2 data were analyzed with dedicated software (IDL; RSI, Boulder, Co., USA). T2 (or 1/R2) values were determined by fitting an exponential decay of the liver SI.

Visual analysis

The presence and location of signal abnormalities within transplanted livers was looked after on both T1 WI and T2 WI.

Histological analysis

Immediately after each MRI procedure, each animal was sacrificed and its liver, spleen and lungs were removed for histological analysis (M.H.). All samples were immediately fixed in 4.5% paraformaldehyde, embedded in paraffin, and stained with hematoxylin-eosin. Some sections were also stained with Prussian blue to detect and localize iron.

Statistical analysis

The liver-to-muscle RI_liver/muscle, spleen-to-muscle RI_{spleen/muscle} and liver T2 values oberved after transplantation with either unlabeled or labeled hepatocytes were compared by using analysis of variance (ANOVA) followed by Fisher’s a posteriori probable least square difference (PLSD) test. Significance was assumed at P<0.05. All statistical analyses were done with Statview 4.5 software (Abacus Concepts, Berkeley, USA) on a Macintosh personal computer (Palo Alto, Calif., USA).

Results

Cell viability

Following hepatocyte labeling and after 3 days of culture, cell viability determined by trypan blue exclusion test always exceeded 90%, regardless of the labeling iron concentration.

Quantification of hepatocyte iron uptake

Magnetophoresis showed a homogeneous velocity distribution, consistent with homogeneous labeling of the hepatocyte population. The iron load per cell increased with the iron concentration used during the 15-min labeling step and saturated at 20±4 pg per cell for an iron concentration of 5 mM (Fig. 2).

Fig. 2.

a Mean hepatocyte iron content after 15 min of incubation with iron oxide nanoparticles at various iron concentrations (0.5, 1, 2, and 5 mM): mean iron load (± standard error) determined by magnetophoresis. b Smear of magnetically labeled mature hepatocytes retrieved after cell labeling procedures

In vivo MRI

The liver-to-muscle RI_liver/muscle fell significantly after transplantation of labeled hepatocytes compared with the four control animals (which did not undergo liver cell transplantation) or with the two animals transplanted with unlabeled cells (P=0.01), as shown by GE T2 sequences (Fig. 3). The mean liver T2 values of mice transplanted with control unlabeled hepatocytes (data not shown) was 63.4±2 ms (4 days after cellular transplantation), values similar to those observed on control animals (60.7±4 ms). Multisection multi-echo SE T2 sequences showed a significant fall in liver T2 values after transplantation of labeled hepatocytes (Table 1; P=0.01). Minimal values occured between 8 and 14 days after cell injection. There were no significant variations over time of spleen-to-muscle RI_{spleen/muscle} (P=0.7) (Fig. 4), especially when comparing control animals and animals transplanted with labeled cells, whatever time-point considered.

Fig. 3.

Mean liver-to-muscle relative SI (RI_liver/muscle) calculated with Eq. (1) in control animals (without liver cell transplantation) and in animals undergoing labeled cell transplantation. A significant decrease in the liver-to-muscle RI_liver/muscle was shown by GE T2WI as early as 2 days following transplantation. This decrease persisted for up to 32 days following transplantation

Fig. 4.

Mean spleen-to-muscle relative SI (RI_{spleen/muscle}) calculated with Eq. (2) in control animals (without liver cell transplantation) and in animals following the transplantation of labeled cells. No significant variation of the spleen-to-muscle relative RI was demonstrated over time

Whatever delay between celullar transplantation and MRI procedures, all 20 recipient livers transplanted with labeled cells showed well-defined nodular foci of low SI on all imaging sequences (Fig. 5), which were not present in the two mice transplanted with unlabeled hepatocytes (imaged 4 days after cellular transplantation) or in the four control animals (Fig. 6).

Fig. 5.

a Transverse and b coronal SE T1-weighted images (TR/TE 380/15 ms; α=90°; thickness 2 mm; FOV=98×90 mm; matrix 512×416; NSA=8) of the liver in a C57Bl6 mouse 8 days after intraportal injection of 8×10⁵ labeled hepatocytes, showing mutliple foci of low SI (arrow). c Transverse GE SSFP (TR/TE 7.77/3.9 ms; α=80°; thickness 2.5 mm; FOV=208×170 mm; matrix 256×208; NSA=11) images also show low but heterogeneous hypointense foci scattered throughout the parenchyma of the right lobe (arrow). The portal branches are clearly identified as flowing hyperintense structures (arrowhead). d Hematoxylin-eosin-saffran (HES) staining, ×200 view and e close-up ×400 view of mouse liver immediately after MRI. Prussian Blue staining shows intensely labeled cells consistent with hepatocytes (arrow) scattered within the liver parenchyma but always grouped around portal tracts

Fig. 6.

a Transverse SE T1-weighted images (TR/TE 380/15 ms; α= 90°; thickness 2 mm; FOV=98×90 mm; matrix 512×416; NSA=8) and b, c transverse GE SSFP (TR/TE 7.77/3.9 ms; α=80°; thickness 2.5 mm; FOV=208×170 mm; matrix 256×208; NSA=11) images of the liver in a C57Bl6 control animal which did not undergo liver cell transplantation showing a homogeneous liver both on T1 and T2 images, without any low SI foci. The galbladder (arrowhead) and the portal branches (arrow) are well depicted on both T2-weighted images

Prussian blue staining performed in the livers of the 20 mice transplanted with labeled hepatocytes showed the presence of clusters of strongly labeled hepatocytes located within the liver parenchyma, close to the portal veins (Fig. 5d,e). All Prussian blue-stained cells had the typical morphology of hepatocytes. These clusters of hepatocytes, distributed throughout the liver parenchyma accounted for the foci of low SI detected on MRI. At all time-points, the distribution of labeled cells within the recipient livers was identical. Figure 7 illustrates the MRI and histological aspect of transplanted livers 21 days following transplantation. Labeled cells systematically formed aggregates of close to ten strongly stained individual cells, leading to patchy, heterogeneously distributed dotted appearance on MRI. Whatever time-point considered, no sign of diffuse and homogeneously distributed low grade staining was identifed. Prussian blue-stained cells were absent in livers transplanted with unlabeled hepatocytes.

Fig. 7.

a Transverse SE T1-weighted images (TR/TE 380/15 ms; α= 90°; thickness 2 mm; FOV=98×90 mm; matrix 512×416; NSA=8) of the liver in a C57Bl6 mouse 21 days after intraportal injection of 8×10⁵ labeled hepatocytes, showing mutliple foci of low SI (arrow). b HES staining, ×400 view of mouse liver immediately after MRI. Prussian blue staining shows intensely labeled cells consistent with hepatocytes (arrow) still located close to portal tracts. There is no presence of diffuse low-grade staining. Labeled cells are not identified within the portal tracts

In all animals, histological analysis showed no iron oxide particles in the recipient lung tissue. Iron stain was faint, and homogeneously distributed within recipient spleens, without clusters of labeled cells, suggesting endogeneous iron without iron oxide particle uptake.

Discussion

Pluripotent cells and cells derived from the phagocytic system are generally used for MRI tracking [21, 32–34], but lymphocytes [35] and myoblasts [36–38] have also been successfully labeled with iron oxide contrast agents, including commercially available contrast agent [39]. To our knowledge, this is the first study to examine both the in vitro labeling feasibility of mature hepatocytes with iron oxide nanoparticles, and their in vivo detection with a conventional 1.5-T MRI device after transplantation in mice.

Transfection agents [23] and electroporation are frequently used to increase cellular uptake of magnetic labels [40]. Recently, however, Smirnov et al. [35] reported that a new iron oxide-based contrast agent with an external anionic coating accumulated passively in cells after endocytosis [35]. Tumor cells [41] and splenocytes have already been successfully labeled, attaining intracellular iron contents of up to 5 pg/cell, which are sufficient for detection by conventional MRI [42]. The same contrast agent was used here, yielding a mean iron content of up to 20 pg/hepatocyte for a very short incubation time (15 min) and without viability impairment. Contrary to polymer-coated magnetic nanoparticles combined with transfection agent, it is a peculiarity of dextran-free anionic iron oxide nanoparticles to allow efficient magnetic labeling by using very short incubation time. Indeed, the endocytosis of anionic nanoparticles is triggered by a first step of nanoparticle adsorption onto the cell plasma membrane. This adsorption is due to electrostatic interaction and is nonspecific and rapid [30]. These results are in keeping with those obtained with other cell types [43].

To our knowledge, this is the first time that engraftment of intraportally injected mature hepatocytes has been visualized in vivo with a conventional 1.5-T MRI device. Shapiro et al. [44] recently visualized the engraftment of individual labeled hepatic cells in vivo by means of MRI, but these authors used a high iron load per cell (up to 50 pg) and a 7-T magnet. The use of a standard 1.5-T magnet permits the acquisition of sequences routinely used and therefore more promising for rapid transfer to human applications. Moreover, despite the drawback related to the reduced signal-to-noise ratio of 1.5-T magnets compared with high-field magnets, our study suggests the potential feasibility of MRI cell detection following liver cell transplantation not only in rodents but also in larger animals, such as pigs or primates. This in turn could allow future human applications, whether at 1.5 T or 3 T.

We detected transplanted hepatocytes with all the MRI sequences tested including morphologic T1 SE sequences. Multisection multi-echo T2 sequences allowed us to quantify the signal loss following intraportal injection of labeled cells. However, no further significant decrease in the liver T2 values was found 8 days after transplantation onwards. Thus, the precise relationship between the T2 signal loss and cell engraftment is unclear. Interestingly, a nonsignificant fluctuation of liver-to-muscle RI and T2 SI over time was present. The reasons for such a variation are unclear. The presence of few strongly labeled cells dispersed throughout the liver, or alternatively the release of part of the iron oxide caused by cell death, could partly explain this phenomenon. We also acknowledge that only three mice were imaged at day 14 following cellular transplantation, which could have lead to a selection bias. Repeated studies with electronic microscopy on pathological specimens could provide additional information on the arrangement of iron oxide particles within or around transplanted hepatocytes.

Labeled hepatocytes were mainly found close to portal tracts and along liver sinusoids, as confirmed by histological studies. This is in keeping with the results of a trafficking study of intraportally injected mesenchymal cells [43], and suggests that labeled hepatocytes remain trapped within the portal spaces. In our experimental conditions, this could be related to the relatively high iron load per cell, or to functional alterations of the transplanted hepatocytes.

Prussian blue staining was always negative in lung tissue, while the distribution and intensity of stained cells in the spleen was very different from that seen in the liver. Indeed, iron staining was homogeneous within the spleen parenchyma, reflecting the presence of endogenous iron in splenocytes. This was further confirmed by the absence of significant variation of the spleen-to-muscle RI_{spleen/muscle}. These results suggest that few if any intraportally injected hepatocytes entered the systemic circulation.

There are several limitations to our study, related in part to the cell type we selected. Stem cells, or hematopoiteic progenitor cells have been previously used for iron oxide cell labeling and cell trafficking experiments [34, 42]. The potential of using pluripotent cells probably outranges that of mature hepatocytes; however, our aim was to keep our protocol as close as possible to potential human applications, including the use of genetically modified mature hepatocytes. Furthermore, using mature hepatocytes could allow the use of self-retrieved hepatocytes, and thus avoid the burdensome harvesting process of immunologically compatible stem cells. In this initial study, we deliberately selected syngenic hepatocytes for cell transplantation. Thus, using specific immunohistological reactions enabling a distinction between host and recipient hepatocytes was not possible. However, all histological findings showed the presence of dark-blue staining, indicating the presence of iron oxide particles surrounding hepatocyte nuclei. As stated by Shapiro et al. [44], donor cell death can result in particle endocytosis by Küpffer cells in the liver. Any iron oxide particles released by transplanted hepatocytes dying within the liver would probably be taken up by Küpffer cells and adjacent hepatocytes. This would lead to weak, diffuse staining of both host hepatocytes and Küpffer cells [44], but we observed no histological signs of this phenomenon. Moreover, infusion of either labeled cells, or of free iron particles lead to different MRI patterns of contrast with increased underlying “graininess” and reduced “dotted” appearance when free iron oxide particles are injected [44]. We, however, did not perform the intraportal injection of pure anionic iron oxide contrast agent in our study as a control. Most importantly, although labeled hepatocytes were identified both by MRI and by histological analysis as late as 32 days following transplantation, their functionality in vivo remains to be demonstrated. Further immunohistochemical studies are warranted in order to demonstrate this point.

We also acknowledge that we deliberately selected intraportal infusion—a method used for cellular transplantation models [18, 19, 45, 46]—rather than intrasplenic administration for liver cell transplantation. In this latter condition, hepatocytes migrate to the liver but a certain proportion engrafts in the spleen. However, comparing intraportal and intrasplenic delivery of hepatoctyes is beyond the scope of this study.

As a conclusion, we show here that intraportally injected hepatocytes labeled with a novel iron oxide contrast agent can be detected using a standard 1.5-T MRI device. Iron oxide labeling combined with MRI could help to avoid the need for repeated liver biopsies in liver cell transplantation studies.