lable at ScienceDirect

Biomaterials 35 (2014) 1015e1024

Contents lists avai

Biomaterials

journal homepage: www.elsevier .com/locate/biomater ia ls

In vivo hepatocyte MR imaging using lactose functionalizedmagnetoliposomes

Ashwini Ketkar-Atre a, Tom Struys a,b, Tom Dresselaers a, Michael Hodenius a,c,Inge Mannaerts d, Yicheng Ni e, Ivo Lambrichts b, Leo A. Van Grunsven d,Marcel De Cuyper c, Uwe Himmelreich a,*

aBiomedical MRI/MoSAIC, Department of Imaging and Pathology, Biomedical Sciences Group, Katholieke Universiteit Leuven, Herestraat 49, B3000 Leuven,Belgiumb Lab of Histology, Biomedical Research Institute, Hasselt University, Campus Diepenbeek, Agoralaan, B3590 Diepenbeek, Belgiumc Laboratory of BioNanoColloids, Interdisciplinary Research Centre, Katholieke Universiteit Leuven, Etienne Sabbelaan 53, B8500 Kortrijk, BelgiumdDepartment of Cell Biology, Liver Cell Biology Lab, Vrije Universiteit Brussel, Laarbeeklaan 103, B1090 Brussel-Jette, Belgiume Theragnostic Laboratory, Department of Imaging and Pathology, Biomedical Sciences Group, Katholieke Universiteit Leuven, Herestraat 49, B3000 Leuven,Belgium

a r t i c l e i n f o

Article history:Received 18 September 2013Accepted 8 October 2013Available online 5 November 2013

Keywords:Asialoglycoprotein receptor (ASGPr-1)HepatocyteLiverMRILactoseMagnetoliposomes

* Corresponding author. Biomedical MRI Unit/MoLeuven, O&N1, BUS 505, Herestraat 49, B-3000 Le330925; fax: þ32 16 330901.

E-mail address: Uwe.Himmelreich@med.kuleuven

0142-9612/$ e see front matter � 2013 Elsevier Ltd.http://dx.doi.org/10.1016/j.biomaterials.2013.10.029

a b s t r a c t

The aim of this study was to assess a novel lactose functionalized magnetoliposomes (MLs) as an MRcontrast agent to target hepatocytes as well as to evaluate the targeting ability of MLs for in vivo ap-plications. In the present work, 17 nm sized iron oxide cores functionalized with anionic MLs bearinglactose moieties were used for targeting the asialoglycoprotein receptor (ASGP-r), which is highlyexpressed in hepatocytes. Non-functionalized anionic MLs were tested as negative controls. The sizedistribution of lactose and anionic MLs was determined by transmission electron microscopy (TEM) anddynamic light scattering (DLS). After intravenous administration of both MLs, contrast enhancement inthe liver was observed by magnetic resonance imaging (MRI). Label retention was monitored non-invasively by MRI and validated with Prussian blue staining and TEM for up to eight days post MLsadministration. Although the MRI signal intensity did not show significant differences between func-tionalized and non-functionalized particles, iron-specific Prussian blue staining and TEM analysisconfirmed the uptake of lactose MLs mainly in hepatocytes. In contrast, non-functionalized anionic MLswere mainly taken up by Kupffer and sinusoidal cells. Target specificity was further confirmed by high-resolution MR imaging of phantoms containing isolated hepatocytes, Kupffer cell (KCs) and hepaticstellate cells (HSCs) fractions. Hypointense signal was observed for hepatocytes isolated from animalswhich received lactose MLs but not from animals which received anionic MLs. These data demonstratethat galactose-functionalized MLs can be used as a hepatocyte targeting MR contrast agent to potentiallyaid in the diagnosis of hepatic diseases if the non-specific uptake by KCs is taken into account.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction hepatocarcinoma (HCC), fibrosis and cirrhosis) but also to monitor

Liver biopsies and serum biomarkers are the gold standard forpredicting disease stage and prognosis in a wide range of liverdiseases. However, interobserver variability, rare but possiblecomplications of the invasive biopsy collection and the lack ofquantitative information at various time points are some of theunsolvable limitations of these techniques [1,2]. In order to monitorthe dynamic changes associated with liver diseases (such as

SAIC, Katholieke Universiteituven, Belgium. Tel.: þ32 16

.be (U. Himmelreich).

All rights reserved.

the success or failure of novel therapeutic approaches (for examplestem cell therapy), the development of non-invasive imagingtechniques would be beneficial for diagnosis and staging of liverdiseases.

Many (pre-) clinical studies have already indicated MRI as thepreferred methodology for liver imaging [2e5]. Apart fromanatomical information, contrast agents have been utilized forseveral decades in clinical liverMRI in particular for the diagnosis ofliver tumors [6e10]. Clinically approved contrast agents for hepaticMRI include non-specific gadolinium chelates, which are distrib-uted in the extracellular interstitial space [11], reticulo-endothelialsystem (RES) specific contrast agents like functionalized super-paramagnetic iron oxide particles (SPIO) that are selectively taken

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e10241016

up by Kupffer cells (KCs) and endothelial cells in the liver [12,13]and as a third category, hepatocyte specific contrast agents,which target functional hepatocytes [14]. However, none of thesecan predict the functional status of the liver longitudinally [6,14]. Ithas been reported that the uptake of hepatocyte targeted contrastagents is reduced in hepatitis [15] and hereby also reflects hepaticfunction [16,17]. Therefore, determining hepatic function usinghepatocyte targeted contrast agents would be of great benefit todiagnose liver diseases at an early stage.

After intravenous injection of SPIOs most are rapidly clearedfrom the blood by KCs, which are part of the RES. These particlesthen accumulate in several organs such as the liver, spleen andlymph nodes through the actions of the mononuclear phagocyticsystem. Hepatocyte targeting can be achieved by the introductionof cell recognizing ligands on the SPIOs surface. Frequently usedreceptors for hepatocyte targeting includes the asialoglycoproteinreceptors (ASGP-r), which are abundantly present on the surface ofhepatocytes [18]. ASGP-r recognizes galactose or N-acetylga-lactosamine residues of desialylated glycoprotein and the uptake ofgalactosylated imaging probes correlates well with hepatic func-tion [15e17,19]. To achieve this, addition of lactobionic acid [20],PVLA (polyvinlybenzyl-O-b-D-galactopyranosyl-D-gluconamide)[21], lactosylated amine [22], polycaprolactone-g-dextran [23],chitosoan e linoleic acid [24,25] and lactose functionalized anionicMLs [26] have been introduced to target ASGP-r from either freshlyisolated hepatocytes or hepatocarcinoma cell lines. These studieshave also demonstrated the feasibility of galactosyl coupled imag-ing probes for hepatocyte targeting. However, none of themaddress the long-term fate and distribution of the functionalizedimaging probes in different hepatic cell types in vivo after systemicadministration.

Liposomes incorporating iron oxide cores are generally referredto asMLs and have been used frequently as MR contrast agents [27].The inner layer of the phospholipid bilayer is strongly chemisorbedonto the iron oxide core, while the outer layer is loosely adsorbedwhich allows to a limited degree exchange of these lipids withlipids from vesicles co-incubated with the MLs [28]. In this study,anionic MLs bearing 5% DMPG have been used as negative control.These anionic MLs were further functionalized by incorporation of1% DOPE-lac; where the lactose is conjugated to the lipid by itsglucose residue, leaving a terminal galactose residue accessible forASGP receptor mediated uptake by hepatocytes. Previously, it wasshown that galactosylated anionic MLs could specifically targetASGP-r present on HepG2 cells [26]. However, their specificity totarget primary hepatocytes in vitro or in vivo has not beenevaluated.

Here, we report on the target specificity and the label retentionof lactose-functionalized MLs and non-functionalized anionic MLsin vivo in healthymice usingMRI, transmission electronmicroscopy(TEM) and histological staining. The future aim of this approach isthe usage of lactose MLs as an agent for hepatocyte imaging andmagnetic separation, and ultimately for contrast enhancement inthe diagnosis of liver disease using MRI.

2. Materials and methods

2.1. MLs materials and characterization

Dimyristoylphosphatidylcholine (DMPC), dimyristoylphosphatidylglycerol(DMPG), and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-lactosyl (DOPE-lac) were purchased from Avanti Polar Lipids (Alabaster, AL, USA). The MLs weresynthesized as described earlier [26]. For the hydrodynamic diameter measure-ments, particles were diluted with TES buffer (5 mM, pH 7.0) and were measuredwith Zetasizer Nano (Malvern, UK) at 25 �C. The average value of hydrodynamicdiameter was determined by three independent measurements using threedifferent runs of different particle batches. Electrophoretic mobility was determinedwith a Zetasizer Nano instrument (Malvern,UK) at 25 �C. Respective iron andphosphate contents were measured as described previously [28].

2.2. Hepatic cell isolation

Six animals were used for parenchymal (hepatocytes) and non-parenchymal(Kupffer cells and stellate cells) hepatic cell isolations performed on day two andfour post lactose or anionic MLs intravenous administration. After intraperitonealanesthesia, the liver was perfused through the portal vein for 5 min with pre-perfusion buffer (SC1, containing 8000 mg/L NaCl, 400 mg/L KCl, 88.17 mg/LNaH2PO4, 120.45 mg/L Na2HPO4, 2380 mg/L HEPES, 350 mg/L NaHCO3, 190 mg/LEGTA, 900 mg/L glucose, pH 7.3) at 37 �C until it was completely discolored and thenfor 5 min with perfusion buffer (SC2, containing 8000 mg/L NaCl, 400 mg/L KCl88.17 mg/L NaH2PO4 H2O, 120.45 mg/L Na2HPO4, 2380 mg/L HEPES, 350 mg/LNaHCO3, CaCl2 .2H2O, pH 7.3) along with 0.25 mg/mL collagenaseP (Roche AppliedScience, Mannheim, Germany). Centrifugation for 2 min at 50 g was performed toseparate hepatocytes from the non-parenchymal cell fraction. HSCs and Kupffer cellswere collected from the non-parenchymal fraction by fluorescent activated cellsorting (FACS Aria; BectoneDickinson, Erembodegem, Belgium) using the endoge-nous UV in HSCs and an APC coupled F4/80 antibody (Life Technologies Corporation,Carlsbad, California, USA). Mouse hepatocytes were further purified by gradientcentrifugation on a 25% Percoll gradient (GE Healthcare Life Science, Diegem,Belgium).

2.3. MR imaging

All experimental protocols were approved by the Institutional Animal CareCommission and Ethical Committee of the Katholieke Universiteit Leuven and per-formed in accordance with international standards on animal welfare.

2.3.1. In vivo imaging30 male C57bl6 (weight 25e30 g) mice received lactose MLs and anionic MLs

through intravenous injection of 200 ml (at 200 mg Fe/ml) MLs diluted in PBS. Micewere scanned under 1e2% isoflurane (carrier gas O2) with a 9.4T Bruker Biospecsmall animal MR scanner (Bruker Biospin, Ettlingen, Germany; horizontal bore20 cm) equippedwith actively shielded gradients (600mTm�1). A quadrature radio-frequency resonator (transmit/receive; inner diameter 7 cm, Bruker Biospin) wasused for image acquisition. Animals were scanned on the day of the injection and onday two, four, six and eight post particle injections. The in vivoMR imaging protocolused for liver imaging consisted of 2D T2*-weighted fast low-angle shot (FLASH) andamulti-slice-multi-echo (MSME) sequence. The FLASH sequence (TE¼ 2.3ms, TR setto 203 ms for six slices with thickness of 1 mm each and an in-plane resolution of117 mm2) was used to determine the decrease in the signal intensity (SI) post in-jection. T2 values (maps) were determined from the MSME experiments and wereused for a semi-quantitative analysis. Parameters for the MSME sequences were TRat least 3000 ms, echo spacing of 7 ms, with 234 mm2 in plane resolution with sixslices of thickness 1 mm each. In order to evaluate particle distribution post intra-venous administration in other organs, mice were subjected to whole body scanwith a Rapid Acquisition with Relaxation Enhancement (RARE) sequence(TE ¼ 15.88 ms, TR ¼ 6000 ms, spatial resolution of 200 mm2, slicethickness ¼ 0.5 mm with 50 slices) was performed. Mice were monitored using amonitoring and gatingmodel (type 1030) from SA Instruments Inc. (Stony Brook, NY,USA) for controlling physiological parameters. Temperature (with rectal probe) andrespiration were monitored and maintained during the acquisition at 37 � 1 �C and60e90min�1, respectively. All in vivoMRImeasurements were respiration triggered.The study design is schematically illustrated in Fig. 1.

2.3.2. Ex vivo imagingIsolated hepatic cell fractions were scanned by MRI in order to assess the tar-

geting specificity and MRI detectability limits using different amounts of cells (from5000 to 25,000 cells/ml). Samples were prepared by suspending isolated cells in100 ml PBS followed bymixing with 1.5% agarose (Sigma) in a 1:1 ratio. Agar and cellsuspensions were quickly transferred to 500 ml microcentrifuge tube one-thirdprefilled with solidified agar. All microcentrifuge tubes containing different hepat-ic cells at different cell density were assembled in a custom made plastic containercompletely filled with agarose. Upon solidifying, the agarose gel phantoms werescanned using a 9.4T MRI scanner. Similar coil configuration was used for phantomimaging as for animal imaging. 3D high resolution, T2*-weighted MR images wereacquired using a gradient echo sequence (FLASH, TR¼ 200ms, TE¼ 15ms with fieldof view 6.0 � 6.0 � 2.25 cm resulting in an isotropic resolution of 234 mm3). T2 mapswere acquired using the MSME sequence also used for in vivo experiments (FOV andmatrix adjusted for the size of the agar phantom).

2.3.3. Image processingAll images were processed with Paravision 5.1 (Bruker Biospin, Ettlingen, Ger-

many). In vivo images were analyzed on a blind basis, irrespective of the particlesinjected. Reduction in in vivo SI was determined by selecting ROI on the liver image.White circles in Fig. 2 indicate the ROIs chosen on the liver section to calculate themean SI. The SI was normalized using two same size ROIs drawn in the muscle re-gion (SI of muscle ¼ 100%). The relative mean SI was determined as SI Liver/SImuscle*100. T2 values were determined from MSME scans. Pixel wise T2 maps werecalculated for a slice of interest using A þ c*exp(�t/T2) with A, c and T2 as variables.The whole liver was manually delineated as ROI on the T2 maps. Thus T2 values

Fig. 1. Experimental scheme: Detailed experimental scheme followed in the study.

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e1024 1017

reported here represent the T2 relaxation time of the complete liver post MLsadministration. T2 values were compared for the pre- and post-contrast injectionconditions. In short, retention of MLs in the liver was analyzed from normalized SIand comparison of T2 time with baseline T2 value. For the analysis of SI care wastaken to maintain ROIs at similar location and size while avoiding imaging artifactsand vasculature.

For phantom image analysis ROIs were drawn in the samples avoiding the edgesof microcentrifuge tubes. Here, relative SI of the cell samples was normalized for SIof agarose (SI¼ 100%). SI was expressed as mean� standard deviation (S.D) over thechosen ROI for every sample.

2.4. Histology

2.4.1. Prussian blue stainingPost intravenous administration of lactose and anionic MLs, animals were

sacrificed on the day of the particle administration and on day two, four, six andeight (as shown in Fig. 1). 14 liver tissue samples collected at different time pointswere fixed overnight in 10% neutral buffered formalin and liver lobes were dividedfor histology and for transmission electron microscopy (TEM). Formalin fixed liver

Fig. 2. In vivo MR imaging: Serial MR scans performed on the indicated time schedule clearthe liver was normalized to muscle SI for every time point. (n ¼ 6, ***p < 0.001, **p < 0.01

lobes used for identifying SPIO- containing cells using Perl’s Prussian blue stainingwere further embedded in paraffin and 5 mm thick sections were made. The paraffinsections were deparaffinised and dehydrated by passing through alcohol series. Theslide was placed in a mixture containing equal volume of 10% potassium ferrocya-nide (Sigma, Bornem, Belgium) and 20% HCl (Vel Labs, Leuven, Belgium) for 20 minat room temperature and then rinsed with tap water for 15 min. After 10 min ofnuclear fast red for counterstaining, the slides were passed through a series ofalcohol and xylene before mounting with DPX (Sigma). Images were acquired usinga Mirax Desk (Carl Zeiss, Göttingen, Germany).

2.4.2. Ultrastructural analysis with TEM14 liver tissue lobes collected at different time points (the day of the particle

administration, Day two, four, six and eight) were cut into a cube of 2 mm2 and fixedovernight in 2% glutaraldehyde and 0.05 M sodium cacodylate buffer (pH 7.3) at 4 �C.Tissue samples were post-fixed in 2% OsO4 in 0.05 M sodium cacodylate buffer (pH7.3) for 1 h and stained with 2% uranyl acetate in 10% acetone for 20 min Further,samples were dehydrated in graded concentrations of acetone and were embeddedin epoxyresin (Araldite). Semi-thin slices (500 nm) were cut, stained with toluidine-blue and used for selecting regions of interest. Ultra-thin sections were mounted on

ly showed hypointense liver post MLs administration. Percentage reduction in the SI of).

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e10241018

0.7% formvar coated grids, contrasted with uranyl acetate followed by lead citrateand examined in a Philips EM 208 transmission electron microscope operated at80 kV. Digital images were taken with the MORADA 10/12 camera (Olympus,Hamburg, Germany). TEM analysis was performed with a Philips EM 208 S electronmicroscope (Philips, Eindhoven, The Netherlands). The microscope was providedwith a Morada Soft Imaging System camera to acquire high resolution images of theevaluated samples. The images were processed digitally with the iTEM-FEI software(Olympus SIS, Münster, Germany).

2.5. Statistical analysis

Statistical analysis was performed using Graphpad Prism 5 software (Graphpad,La Jolla, USA). Significant differences between functionalized and non-functionalized particles were determined using the two way analysis of variance(ANOVA) test with a Tukey post or Bonferroni post-test. Data were plotted asmean � S.D. P-values �0.05 were considered statistically significant.

3. Results

3.1. Synthesis and characterization of particles

Hydrodynamic radii measured with dynamic light scattering(DLS) indicated that lactose and anionic MLs were of similar hy-drodynamic diameter (approx. 40 nm). Zeta potential measure-ments confirmed a negative surface charge on both the MLs. MLsproperties are indicated in Table 1.

3.2. Determining in vivo fate of MLs by MR imaging

Animals were scanned before MLs injection, on the day of theinjection and at two, four, six and eight days post injection. T2*-weighted MRI shows hypointense contrast in the liver, indicating asignificant reduction in the SI (Fig. 2). Lactose MLs resulted in areduction of 60 � 5% in SI versus muscle whereas anionic MLsshowed a similar drop of 63 � 4%. Although significant differenceswere found for day 2 and day 6, overall both the MLs showed asimilar signal recovery profile. As T2*-weighted MRI showed nodifference in label retention between the functionalized and non-functionalized ML’s, no direct information on the cellular distri-bution of MLs could be obtained based on the MR images alone.Similarly, Fig. 3 shows color codedmaps of T2 values for bothMLs atdifferent time points. T2 relaxation times were calculated consid-ering the entire liver as one ROI. They showed a significant reduc-tion in the T2 relaxation times for both MLs (lactose MLs:11.0 � 1 ms and anionic MLs: 10 � 1 ms) with respect to controllivers before ML injection (T2 ¼ 18.1 � 1.6 ms). Again, the recoveryprofile was similar for both types of MLs indicating the necessity offurther microscopic analysis in order to determine the uptakespecificity and label retention.

In addition to dedicated high-resolution MRI of the liver region(Fig. 2), whole body MRI scans and Prussian blue staining ofselected organs were performed to assess the distribution of MLs inthe animals (see Fig. 4). Kidneys did not show any change in the SIcompared to pre-contrast SI (data not shown). Due to anatomicalpositioning of the spleen and the susceptibility artifacts caused bythe air in the lungs, determination of a reduction in the SI due toMLs administration was not possible in these organs with ourscanning protocols. Therefore, ex vivo high resolution imaging(FLASH 3D T2*-weighted scans) of the isolated spleen was per-formed to quantify the reduction in the SI with respect to the spleen

Table 1Hydrodynamic diameter and surface charge of magnetoliposomes.

MLs Diametera [nm] Zeta potential [mV]

Anionic, non-functionalized MLs 39.91 � 0.6 �20.6 � 0.7Lactose MLs 39.77 � 0.7 �23.9 � 2.6

a Measurements were performed in TES buffer (pH ¼ 7.0, 5 mM).

of non-injected control animals. Using ex vivoMRI or Perl’s Prussianblue staining, no significant differences in the iron content of thespleenwere found (see also Fig. 4B), which was most likely also dueto the high intrinsic iron content. Lung tissue samples subjected toPrussian blue staining indicated no signs of any ML uptake in thelung alveoli (Fig. 4C). With our scanning protocol, MRI signal in-tensity changeswere found to be restricted to the liver (SI reductionby 60%).

3.3. Cellular distribution of MLs by histology

3.3.1. Prussian blue stainingLiver tissue samples collected for MLs uptake confirmation with

Prussian blue staining are shown in Fig. 5. Irrespective of theinjected MLs, initial tissue samples showed sparsely stained liversections and thus did not provide robust information on the targetspecificity. By day 4, lactose MLs injected livers showed distinctblue staining mainly in bi-nucleated hepatocytes (indicated byblack arrows in Fig. 5). Also some uptake was observed in the re-gions of sinusoids (red arrows, Fig. 5) where KCs and LSECs reside.However, animals which received non-functionalized anionic MLsshowed prominent blue staining in the regions of sinusoids (redarrows, Fig. 5) indicating presence of anionic MLsmainly in KCs andLSECs. For both the MLs, blue staining was observed until the latesttime point 8 days post particle administration.

3.3.2. Ultrastructural analysisUltrastructural analysis was performed to determine the cellular

distribution and fate of the MLs after systemic administration. TEMimages indicate presence or absence of MLs in hepatocytes (Fig. 6),KCs (Fig. 7) and LSECs (Fig. 8). Fig. 6 shows that hepatocytes did notcontain iron loaded endosomes when animals received non-functionalized anionic MLs (Fig. 6AeC) at any of the follow uptime point. Conversely, in animals receiving lactose MLs, wedetected iron oxide cores heterogeneously distributed in theendosomes of hepatocytes (Fig. 6D) at the initial time point. At latertime points, MLs clustering in the endosomes (by day 4, Fig. 6E) andin larger lysosomes (by day 8, Fig. 6F) were detected for animalsthat received lactose MLs. Similar lysosomes were also observed inthe hepatocytes of anionic MLs but without iron oxide particles(Fig. 6C). Figs. 7 and 8 show particle distribution in KCs and LSECs,respectively. Animals that received non-functionalized, anionicMLs always showed label retention in KCs and LSECs at all the timepoints (Fig. 7AeC and Fig. 8AeC). For both cells types, heteroge-neously distributed iron oxide MLs in multiple endosomes wasdetected. On the other hand, only very few KCs and LSECs werefound to contain lactose MLs for later time points. The ultrastructural analysis of the particle distribution in different cell typesis consistent with data obtained from Prussian blue staining (Fig. 5).

No MLs were found in hepatic stellate cells (HSCs), indicatingthat therewas no apparent toxicity caused by theMLs, whichmighthave caused activation of HSCs.

3.4. Hepatic cell isolation and MR imaging

For confirmation of the in vivo results, the different liver celltypes were isolated at defined time points from animals that havereceived either non-functionalized anionic or lactose MLs. Theisolated cells were studied by ex vivo MRI. Fig. 9 shows images oftwo phantoms containing three different types of hepatic cells(hepatocytes, KCs and HSCs) isolated on day two and four postparticle administration. Cellular uptake of MLs was quantified bycalculating reduction in the SI with respect to agarose for all threehepatic cell types on day two post MLs administration (Fig. 9B).Samples containing different cell densities (5000e10,000 cells/ml)

Fig. 3. T2 relaxation time: Color coded T2 maps were used to analyze an absolute reduction in the relaxation time and indicate possible washout of lactose/anionic MLs. The T2relaxation time of the liver was significantly lower than pre-contrasted liver for the complete follow up schedule. (n ¼ 5, ***p < 0.0001, **p < 0.01).

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e1024 1019

were used for a better comparison of the targeting specificity. Thehighest iron oxide uptake was confirmed for hepatocytes (SIreduction of 35 � 20%) and KCs (SI reduction of 32 � 24%) fromanimals that received lactose MLs indicated by the prominent

Fig. 4. Biodistribution of MLs: (A) Both the lactose-functionalized and non-functionalizedmuscle signal intensity (data not shown)). (B) Ex vivo imaging of spleen at 4 h post MLsnon-injected animals. (C) MLs could pass through lung alveoli and no Prussian blue stainin

hypointense spots, which are due to the presence of MLs in thecells. Hepatocytes from animals that had received non-functionalized anionic MLs did not show MR signal reduction (SIreduction of only 4 � 8%), indicating very low anionic MLs uptake

MLs did not show any change in the SI of the kidneys (normalized to pre-contrastedadministration indicated no difference in the SI compared to Spleen collected fromg was observed in the tissue sections at different time points. Scale bar ¼ 50 mm.

Fig. 5. Prussian blue staining: Till day two post MLs administration, tissue sections were found to be sparsely stained with Prussian blue stain. Insets at day two and four indicatethe light cytoplasmic blue staining. At later time points, lactose MLs injected livers indicated prominent blue staining mainly in the bi-nucleated hepatocytes (black arrows) whereaslivers which received anionic MLs showed blue staining predominantly in the sinusoidal space (red arrows). Scale bar: 20 mm, insets: 50 mm.

Fig. 6. MLs distribution in Hepatocytes: (AeC) show no uptake of anionic MLs by bi-nucleated hepatocytes at any of the time points. (DeF) Lactose MLs received hepatocytesshowed presence of MLs for all the time points. Samples collected immediately (<24 h) indicated presence of MLs distributed heterogeneously in the endosomes (E). By day 4,particles were seen in the aggregated form in the endosomes and later in the bigger lysosomes (L) by day 8. (N: Nucleus; E: Endosomes; L: Lysosomes); Scale bar: (A) and (CeF) ¼ 2 mm, (B) ¼ 5 mm.

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e10241020

Fig. 7. MLs distribution in KCs: (AeC) Show homogenously distributed anionic MLs present in the multiple endosomes of KCs. (DeF) Unspecific uptake of lactose was also found inKCs but not in the multiple endosomes. (N: Nucleus; E: Endosomes); Scale bar: (AeF) ¼ 2 mm.

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e1024 1021

by hepatocytes. However the MRI of KCs isolated from animalswhich received non-functionalized, anionic MLs showed SI reduc-tion of 20 � 19%. This demonstrates that lactose MLs were retainedin hepatocytes and in KCs for the initial time point (day 2), whereasanionic MLs were only taken up by KCs but not by hepatocytes.These MRI data also confirm the TEM and in vivo MRI data (seeFig. 2). Fig. 2 indicates a significant reduction in the total SI due tolactose MLs compared to the anionic control MLs at day 2. Thisreduction seems to be a cumulative effect of iron uptake by hepa-tocytes and KCs. In contrast, non-functionalized, anionic MLs werepresent only in KCs and thus, could not reduce the signal as effi-ciently as lactose MLs did at day 2 (as seen in Fig. 2). By day 4, KCsbut not hepatocytes isolated from animals that received anionicMLs showed a SI reduction. In contrast, hepatocytes from animalsthat received lactose MLs showed retention of the MLs on day 4 bypresence of hypointense spots in the respective MR images.

4. Discussion

The main objective of this study was to assess the in vivo targetspecificity and label retention of galactose functionalized MLscompared to non-functionalized control MLs in healthy mice usingMRI and histological staining. Galactosylated imaging probes havethe potential to target either freshly isolated hepatocytes or hep-atocarcinoma cells specifically [20e26]. Studies also demonstratedthe feasibility to assess the hepatic function depending on uptake oftargeted probes [16,17]. However, determining the fate of i.v.injected SPIOs is challenging, as approximately 80% of the injecteddose is taken up by KCs and 5e10% of the injected dose ends up inthe spleen [6]. Once compartmentalized within lysosomes of RES

cells, the iron oxide particles are broken down and stored in theform of ferritin and/or hemosiderin. Hence, developing hepatocytetargeting probes is challenging as hepatocyte are not the only celltype of the liver and thus size, access limitations due to themicrostructure of the liver and unspecific uptake by phagocyticcells have to be considered [29e31].

Soenen et al. synthesized galactose bound anionic MLs andsuccessfully targeted ASGP-r expressing HepG2 cells [26]. However,the cellular distribution of these functionalized MLs in vivowas stillunexplored. Generally, MLs consist of an iron oxide core(17 � 0.4 nm, measured by TEM) that is encapsulated by a phos-pholipid bilayer. The lipid bilayer of MLs allows excellent flexibilityin terms of functionalization. Biomedical functionalization of MLsinclude the addition of polyethylene glycol (PEG) to enhance bloodcirculation times, conjugation of a peptide or antibody for activetargeting, conjugation of a fluorescent dye as a bimodal contrastagent and utilization as carriers for drug delivery [27]. The hydro-dynamic diameter of lactose MLs was measured by DLS and foundto be 39.7��0.7 nm. These lactose MLs should therefore be able topass through liver sinusoids (diameter of about 100 nm) into thespace of Disse [32]. This particle size confirmed that our particlesare suitable for hepatocyte targeting and possible further applica-tion as drug carriers.

After systemic injection of non-functionalized anionic or lactoseMLs, the whole body and cellular distribution was examined withMRI and TEM. Immediately after particle administration of MLs, allthe cell types indicated heterogeneously distributed iron oxideparticles for both the MLs as indicated by TEM (Figs. 6e8). Animalsthat received lactose MLs showed uptake in hepatocytes and also toa minor extent by KCs and LSECs. From the anatomical structure of

Fig. 8. MLs distribution in LSECs: (AeC) Show heterogeneously distributed anionic MLs present in the multiple endosomes of LSECs till day 8. (DeF) Very few lactose MLs wereunspecifically taken up by LSECs for initial days. Possibly due to lysosomal degradation MLs were not present in LSECs till day 8. (N: Nucleus; E: Endosomes); Scale bar: (AeE) ¼ 2 mm, (F) ¼ 5 mm.

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e10241022

the liver, non-parenchymal cells are localized in the sinusoids whileparenchymal cells are situated behind a fenestrated endothelium.Hepatocytes account for 65e70% of the total liver cells whereasnon-parenchymal cells (like LSECs, KCs and HSCs) account for 30e35% of total liver cells. The localization of KCs and LSECs near bloodvessels and their phagocytic capacity explains the potential uptakeof lactose MLs by KCs and LSECs. On the other hand, anionic MLswere not at all taken up by hepatocytes but only by KCs and LSECs(Figs. 6e8AeC). Also later at eight days after anionic MLs injection,large lysosomes without iron oxide particles were present in thehepatocytes (Fig. 6C) in contrast to iron loaded lysosomes in he-patocytes from animals that received lactose MLs (Fig. 6F). As ly-sosomes are considered the last phase of the receptor mediatedendocytosis pathway (in this case ASGP-r) where the ligand isdegraded, this finding confirms the hepatocyte specificity of lactoseMLs [18]. In vivo MR imaging could not provide definitive infor-mation on the specificity or possible difference in the label reten-tion due to a similar SI and T2 relaxation time recovery profile of thenon-functionalized and lactose MLs. However, ex vivo MR imagingof isolated cell phantoms confirmed the specificity of lactose MLs inthe form of hypointense contrast by hepatocytes and KCs isolatedfrom animals that received lactose MLs. In contrast, only KCs indi-cated hypointense contrast for cells isolated from animals thatreceived anionic control MLs (Fig. 9). These observations confirmedthe in vivo specificity and fate of lactose MLs.

In vivo MR imaging showed a gradual increase in the SI and T2relaxation time by day six and eight (Figs. 2 and 3) post MLsadministration, whereas electron microscopy and histology stain-ing indicated clustering of MLs also at later time points (Figs. 6e8).

Roca et al. indicated that the aggregate size should be considered inorder to explain changes in relaxivity of particles in suspension[33]. It is apparent that MLs present in the cluster form at later timepoints were insufficient to quench the MR signal significantlycompared to the initial time points (day two and four). Although,the location of aggregates in vivo is not comparable to their effect insuspension, gradual particle degradation is a probable reason forinsufficient signal quenching at later time points.

Here we determined the specificity of hepatocyte targetingprobe in healthy animals but it needs further assessment in thedisease models where hepatocyte volume reduces or affectsexpression of ASGP-r. Application of such targeted probes could beused to establish the biomarkers for liver diseases with MR imag-ing. It is a well known fact that liposomes in themselves are veryversatile entities and have frequently been used as drug carriersdue to their biocompatibility and ability to undergo surface ma-nipulations by chemical means [27]. Thus, such particles could alsobe used as drug (nano) carriers. Recently, differentiating stem cellsinto hepatocyte like cells is considered as a promising source ofhepatocytes. But these differentiation cultures result into cells frommeso- and endo-derm alongwith functional hepatocytes [34]. Suchnanoprobes could assist as isolation marker for hepatocytes bymagnetic separation from the mixed population in vitro and alsocould be utilized as an imaging probe in cell tracking studies.

5. Conclusion

We have successfully validated the specificity of lactose MLs totarget hepatocytes in vivo in healthy animals based on ultra-

Fig. 9. Ex vivo imaging of hepatic cells: (A) Liver cells were isolated on day two, four and phantoms were scanned with different cell density (5000e25,000 cells/ml). Prominenthypointense signal was observed when KCs and hepatocytes isolated from lactose MLs injected animals were scanned whereas hepatocytes which received anionic MLs did notshow any signal reduction with the same number of cells, indicating no uptake. Irrespective of the particle injected, on day four KCs showed retention of MLs whereas only he-patocytes from lactose MLs showed signal reduction and not due to anionic MLs. (B) Reduction in the signal intensity on day two post MLs administration in different cells wasplotted with respect to agarose SI indicating maximum signal reduction was possible with hepatocytes and KCs isolated from animals receiving lactose MLs and very less reductionwith anionic MLs injection.

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e1024 1023

structural analysis and ex vivo MRI data. Non-functionalized, con-trol MLs showed similar signal in vivo MRI intensity profiles as thefunctionalized MLs but were mainly due to unspecific uptake byKCs and LSECs. While in vivo targeting of hepatocytes by usinglactose MLs was in principle confirmed in this study by histologyand ultra-structural properties, the in vivo application needsfurther validation in disease models that result in hepatocyte lossor therapy models that result in hepatocyte regeneration.

Conflict of interest

Authors declare no conflict of interest.

Acknowledgments

The authors would like to thank Ann van Santvoort for thetechnical help with the MRI scans. Marc Jans is gratefullyacknowledged for assistance with processing of the TEM samples.We gratefully acknowledge the financial support by the Europeancommission (project EC-FP7-NMP-2008-Large ‘ViBRANT’, 228933),the Flemish government (SBO-IWT-80017 ‘iMAGiNe’ and SBO-IWT-090066 ‘HEPSTEM’) and by the KU Leuven Program Financing

‘IMIR’. A.A. is grateful for receiving financial support for a PhDfellowship (DBOF) from the KU Leuven.

References

[1] Manning DS, Afdhal NH. Diagnosis and quantitation of fibrosis. Gastroenter-ology 2008;134:1670e81.

[2] Castera L. Noninvasive methods to assess liver disease in patients with hep-atitis B or C. Gastroenterology 2012;142:1293e302.

[3] Clasen S, Boss A, Schmidt D, Fritz J, Schraml C, Claussen CD, et al. Magneticresonance imaging for hepatic radiofrequency ablation. Eur J Radiol 2006;59:140e8.

[4] Kanematsu M, Goshima S, Watanabe H, Kondo H, Kawada H, Noda Y, et al.Detection and characterization of focal hepatic lesions with diffusion-weighted MR imaging: a pictorial review. Abdom Imaging 2013;38:297e308.

[5] Koike N, Cho A, Nasu K, Seto K, Nagaya S, Ohshima Y, et al. Role of diffusion-weighted magnetic resonance imaging in the differential diagnosis of focalhepatic lesions. World J Gastroenterol 2009;15:5805e12.

[6] Bellin MF. MR contrast agents, the old and the new. Eur J Radiol 2006;60:314e23.

[7] Gupta S, Rajvanshi P, Lee CD. Integration of transplanted hepatocytes into hostliver plates demonstrated with dipeptidyl peptidase IV-deficient rats. ProcNatl Acad Sci U S A 1995;92:5860e4.

[8] Karabulut N, Elmas N. Contrast agents used in MR imaging of the liver. DiagnInterv Radiol 2006;12:22e30.

[9] Morana G, Salviato E, Guarise A. Contrast agents for hepatic MRI. Cancer Im-aging 2007;7(A):S24e7.

A. Ketkar-Atre et al. / Biomaterials 35 (2014) 1015e10241024

[10] Puppi J, Modo M. Use of magnetic resonance imaging contrast agents to detecttransplanted liver cells. Top Magn Reson Imaging 2009;20:113e20.

[11] Bellin MF, Vasile M, Morel-Precetti S. Currently used non-specific extracellularMR contrast media. Eur Radiol 2003;13:2688e98.

[12] Seneterre E, Weissleder R, Jaramillo D, Reimer P, Lee AS, Brady TJ, et al. Bonemarrow: ultrasmall superparamagnetic iron oxide for MR imaging. Radiology1991;179:529e33.

[13] Weissleder R, Elizondo G, Wittenberg J, Lee AS, Josephson L, Brady TJ. Ul-trasmall superparamagnetic iron oxide: an intravenous contrast agent forassessing lymph nodes with MR imaging. Radiology 1990;175:494e8.

[14] Gandhi SN, Brown MA, Wong JG, Aguirre DA, Sirlin CB. MR contrast agents forliver imaging: what, when, how. Radiographics 2006;26:1621e36.

[15] Tanimoto A, Kuribayashi S. Hepatocyte-targeted MR contrast agents: contrastenhanced detection of liver cancer in diffusely damaged liver. Magn ResonMed Sci 2005;4:53e60.

[16] Kim EM, Jeong HJ, Park IK, Cho CS, Kim CG, Bom HS. Hepatocyte-targetednuclear imaging using 99mTc-galactosylated chitosan: conjugation, targeting,and biodistribution. J Nucl Med 2005;46:141e5.

[17] Kim EM, Jeong HJ, Park IK, Cho CS, Moon HB, Yu DY, et al. Asialoglycoproteinreceptor targeted gene delivery using galactosylated polyethylenimine-graft-poly(ethylene glycol): in vitro and in vivo studies. J Control Release 2005;108:557e67.

[18] Stockert RJ. The asialoglycoprotein receptor: relationships between structure,function, and expression. Physiol Rev 1995;75:591e609.

[19] Pricer Jr WE, Ashwell G. The binding of desialylated glycoproteins by plasmamembranes of rat liver. J Biol Chem 1971;246:4825e33.

[20] Kamruzzaman Selim KM, Ha YS, Kim SJ, Chang Y, Kim TJ, Ho Lee G, et al.Surface modification of magnetite nanoparticles using lactobionic acid andtheir interaction with hepatocytes. Biomaterials 2007;28:710e6.

[21] Yoo MK, Kim IY, Kim EM, Jeong HJ, Lee CM, Jeong YY, et al. Superparamagneticiron oxide nanoparticles coated with galactose-carrying polymer for hepa-tocyte targeting. J Biomed Biotechnol 2007;2007:94740.

[22] Huang G, Diakur J, Xu Z, Wiebe LI. Asialoglycoprotein receptor-targetedsuperparamagnetic iron oxide nanoparticles. Int J Pharm 2008;360:197e203.

[23] Wu DQ, Lu B, Chang C, Chen CS, Wang T, Zhang YY, et al. Galactosylatedfluorescent labeled micelles as a liver targeting drug carrier. Biomaterials2009;30:1363e71.

[24] Lee CM, Jeong HJ, Kim SL, Kim EM, Kim DW, Lim ST, et al. SPION-loadedchitosan-linoleic acid nanoparticles to target hepatocytes. Int J Pharm2009;371:163e9.

[25] Cheong SJ, Lee CM, Kim SL, Jeong HJ, Kim EM, Park EH, et al. Super-paramagnetic iron oxide nanoparticles-loaded chitosan-linoleic acid nano-particles as an effective hepatocyte-targeted gene delivery system. Int J Pharm2009;372:169e76.

[26] Soenen SJ, Brisson AR, Jonckheere E, Nuytten N, Tan S, Himmelreich U, et al.The labeling of cationic iron oxide nanoparticle-resistant hepatocellularcarcinoma cells using targeted magnetoliposomes. Biomaterials 2011;32:1748e58.

[27] Soenen S, Vande Velde G, Ketkar-Atre A, Himmelreich U, De Cuyper M.Magnetoliposomes as MRI contrast agents. WIREs Nanomed Nanobiotechnol2011;3:197e211.

[28] De Cuyper M, Joniau M. Magnetoliposomes. Formation and structural char-acterization. Eur Biophys J 1988;15:311e9.

[29] Dong L, Gao S, Diao H, Chen J, Zhang J. Galactosylated low molecular weightchitosan as a carrier delivering oligonucleotides to Kupffer cells instead ofhepatocytes in vivo. J Biomed Mater Res A 2008;84:777e84.

[30] Popielarski SR, Pun SH, Davis ME. A nanoparticle-based model delivery systemto guide the rational design of gene delivery to the liver. 1. Synthesis andcharacterization. Bioconjug Chem 2005;16:1063e70.

[31] Rensen PC, Sliedregt LA, Ferns M, Kieviet E, van Rossenberg SM, vanLeeuwen SH, et al. Determination of the upper size limit for uptake andprocessing of ligands by the asialoglycoprotein receptor on hepatocytesin vitro and in vivo. J Biol Chem 2001;276:37577e84.

[32] Rieder H, Meyer zum Buschenfelde KH, Ramadori G. Functional spectrum ofsinusoidal endothelial liver cells. Filtration, endocytosis, synthetic capacitiesand intercellular communication. J Hepatol 1992;15:237e50.

[33] Roca AG, Veintemillas-Verdaguer S, Port M, Robic C, Serna CJ, Morales MP.Effect of nanoparticle and aggregate size on the relaxometric properties of MRcontrast agents based on high quality magnetite nanoparticles. J Phys Chem B2009;113:7033e9.

[34] Sancho-Bru P, Najimi M, Caruso M, Pauwelyn K, Cantz T, Forbes S, et al. Stemand progenitor cells for liver repopulation: can we standardise the processfrom bench to bedside? Gut 2009;58:594e603.