Internalization of paramagnetic phosphatidylserine-containing liposomes by macrophages

Background Inflammation plays an important role in many pathologies, including cardiovascular diseases, neurological conditions and oncology, and is considered an important predictor for disease progression and outcome. In vivo imaging of inflammatory cells will improve diagnosis and provide a read-out for therapy efficacy. Paramagnetic phosphatidylserine (PS)-containing liposomes were developed for magnetic resonance imaging (MRI) and confocal microscopy imaging of macrophages. These nanoparticles also provide a platform to combine imaging with targeted drug delivery. Results Incorporation of PS into liposomes did not affect liposomal size and morphology up to 12 mol% of PS. Liposomes containing 6 mol% of PS showed the highest uptake by murine macrophages, while only minor uptake was observed in endothelial cells. Uptake of liposomes containing 6 mol% of PS was dependent on the presence of Ca2+ and Mg2+. Furthermore, these 6 mol% PS-containing liposomes were mainly internalized into macrophages, whereas liposomes without PS only bound to the macrophage cell membrane. Conclusions Paramagnetic liposomes containing 6 mol% of PS for MR imaging of macrophages have been developed. In vitro these liposomes showed specific internalization by macrophages. Therefore, these liposomes might be suitable for in vivo visualization of macrophage content and for (visualization of) targeted drug delivery to inflammatory cells.


Background
Inflammation plays a crucial role in many pathologies, including cardiovascular diseases, neurological disorders and oncology, and is generally considered as an important predictor for disease progression and outcome [1,2]. Therefore, modulation of the inflammatory response by dedicated therapy is of particular interest.
The efficacy of traditional therapeutic compounds of low molecular weight is often limited by short blood circulation half-lives and adverse side effects due to non-specific systemic distribution and accumulation. Additionally, it is difficult to obtain quantitative information on the amount of drug accumulating in the diseased tissue. Drug delivery via a nanocarrier system provides an attractive alternative to alleviate these drawbacks. For example, Doxil is a clinically approved nanocarrier system for cancer treatment, which consists of doxorubicin encapsulated in liposomes [3,4]. This formulation limits cardiotoxicity and prolongs the blood circulation half-life compared to free doxorubicin, which results in an enhanced time window for drug delivery and extravasation of the liposomes through the leaky tumor vasculature.
The surface composition of nanocarriers containing drugs can be tailored to tune clearance kinetics, for instance polyethylene glycol (PEG) is often incorporated to prolong the blood half-life [5]. Furthermore, the larger size of nanocarriers promotes a higher level of uptake in diseased tissues by the enhanced permeability and retention (EPR) effect [6,7]. Importantly, to address the inflammatory response in cardiovascular disease, the drug-containing nanocarriers should be delivered with high specificity to inflammatory cells in the diseased tissue. This can be achieved by introducing ligands that mediate nanocarrier recognition and internalization by the inflammatory cells.
An attractive route to target macrophages is by incorporation of the lipid phosphatidylserine (PS) in lipid-based nanoparticles, such as liposomes. In mammalian cells, PS is predominantly present in the inner leaflet of cell membranes. When a cell becomes apoptotic, PS is exposed on the outer leaflet of the cell membrane, which serves as a trigger for phagocytosis by macrophages [8,9]. The incorporation of PS in the liposomal membrane can therefore promote uptake by macrophages. Previously it was shown that incorporation of PS in liposomes indeed resulted in enhanced uptake by macrophages [10,11].
Also, magnetic resonance imaging (MRI) contrast agents can be incorporated to image drug delivery and obtain quantitative information on the local concentration of drugs at the target site [12,13]. Previously, Harel-Ader et al. developed liposomes with PS containing iron-oxides for MRI visualization of inflammatory cells in myocardial infarction [14] and Maiseyeu et al. described liposomes with PS containing Gd-DTPA-distearylamide for MR imaging of macrophages in atherosclerotic plaques [10]. However, a detailed characterization and optimization of MRI-detectable PS-containing liposomes, including the conditions under which they most effectively target macrophages and induce strongest contrast in MRI, is still lacking.
In this study, we therefore describe the design and characterization of MRI-detectable liposomes that are targeted to macrophages using PS. Liposomes containing different molar percentages of PS were prepared and liposome size and morphology were studied by dynamic light scattering (DLS) and cryogenic transmission electron microscopy (cryoTEM). Fluorescent labels, incorporated in the liposomes, enabled detailed analysis of liposome binding and internalization by macrophages using confocal laser scanning microscopy (CLSM) and fluorescence activated cell sorting (FACS). The ability of the liposomes to induce contrast changes in MR images was studied in macrophages and quantified on the basis of the measured changes in T 1 and T 2 relaxation times.

Characterization of liposomes
Liposomes containing 0, 6, 12 and 37 mol% of phosphatidylserine (1,2-distearoyl-sn-glycero-3-phospho-L-serine = DSPS or PS) were prepared. In this paper we will refer to these liposomes as PC-L, PS-6-L, PS-12-L, and PS-37-L, respectively (see also Table 1 and Methods). Thin layer chromatography (TLC) confirmed the presence of PS in the liposome formulations by the appearance of a spot corresponding to DSPS (Figure 1a). The spot became more intense with increasing mol% PS in the lipid preparation mixture, which shows that PS was successfully incorporated in increasing amounts in the final liposome preparations up to 37 mol%.
Representative DLS spectra for the different types of liposomes are presented in Figure 1b. For all formulations a single dominant peak was observed, indicating a relatively narrow range of liposome diameters. PS-6-L and PS-12-L had the same mean hydrodynamic diameter as PC-L (Table 2). However, PS-37-L had a somewhat smaller diameter (p <0.05 vs. PC-L). Incorporation of PS resulted in a significant increase of the polydispersity index (PDI, p <0.05 vs. PC-L), which was also observed as a modest broadening of the DLS peaks ( Figure 1b). We think that changes in the membrane rigidity or stability due to incorporation of PS leads to a smaller size after extrusion. Liposome morphology was investigated in more detail using cryoTEM (Figure 1c). CryoTEM images revealed predominantly single unilamellar liposomes for all formulations. For PC-L, PS-6-L and PS-12-L liposomes were spherical, whereas for PS-37-L occasionally non-spherical, deformed liposomes were observed ( Figure 1c, black arrows).
The ability of the liposomes to generate contrast in MRI is determined by their potency to change the longitudinal (T 1 ) and transversal (T 2 ) relaxation times, which is expressed by the longitudinal (r 1 ) and transversal (r 2 ) relaxivity. The r 1 and r 2 of the liposomes at 9.4 T and room temperature, normalized to Gd concentration, were 3.0-4.0 mM -1 Ás -1 and 42-60 mM -1 Ás -1 , respectively ( Table 2). Incorporation of PS did not significantly affect the longitudinal and transversal relaxivity. All liposome formulations displayed a similar relatively high r 2 /r 1 ratio.

Association of PS-containing liposomes with macrophages
In vitro experiments were performed to determine which formulation of PS-containing liposomes resulted in highest association with mouse macrophages (RAW cells). RAW cells were incubated with PC-L, PS-6-L, PS-12-L and PS-37-L and association of liposomes with the macrophages was characterized by several readouts exploiting the various components of the liposomes, including quantitative T 1 and T 2 mapping with MRI, quantitative Gd determinations by inductively coupled plasma mass spectrometry (ICP-MS), and CLSM. Average R 1 (=1/T 1 ) and R 2 (=1/T 2 ) values for the different groups are summarized in Figure 2c and 2d, respectively. Incubation with liposomes always resulted in enhanced R 1 and R 2 values (p <0.05 vs. no L). In agreement with quantitative Gd determinations, however less pronounced, both R 1 and R 2 were highest for the incubations with PS-6-L (0.942 ± 0.004 s -1 and 37.3 ± 1.2 s -1 , respectively, p <0.05 vs. all).
The relaxivities r 1 and r 2 in the cellular environment were estimated from the quantitative Gd determinations in relation to changes in R 1 and R 2 ( Table 3). For PC-L, r 1 and r 2 were 1.9 ± 0.3 mM -1 Ás -1 and 32.5 ± 3.6 mM -1 Ás -1 , respectively. For PS-6-L, r 1 and r 2 were 0.8 ± 0.4 mM -1 Ás -1 and 16.3 ± 5.8 mM -1 Ás -1 . As shown previously, PC-L and PS-6-L relaxivities in aqueous solution were similar ( Table 2). The lower cellular relaxivities for PS-6-L   compared to PC-L therefore suggested a different PS-6-L uptake mechanism in RAW cells and consequently a different cellular distribution, which was investigated in more detail as described further on. CLSM imaging of the near-infrared (NIR)-labeled lipids incorporated in the liposomal membrane revealed association of all types of liposomes with RAW cells (Figure 2e). No NIR autofluorescence signal was detected in RAW cells incubated without liposomes. In agreement with MRI, NIR fluorescence and therefore liposome association was highest for cells incubated with PS-6-L and intermediate for PC-L, while PS-12-L and PS-37-L showed similarly low levels of NIR fluorescence.
To confirm that the observed association of PS-6-L with RAW cells was mediated by their phagocytic character, endothelial H5V cells, for which no or minor phagocytosis was expected, were incubated with the different liposome formulations. MRI T 1 and T 2 maps for untreated and liposome-incubated cell pellets, shown in Figure 3a, revealed only minor differences between the various groups. Nevertheless, incubation with liposomes resulted in enhanced R 1 values (p <0.05 vs. no L) and the highest values were detected for cells incubated with PS-6-L (0.524 ± 0.003 s -1 , p <0.05 vs. all, Figure 3b). No significant differences were detected for R 2 (p >0.05 vs. all). However, R 1 and R 2 values were significantly lower compared to RAW cell incubations for all groups (p <0.05 vs. RAW cells). ΔR 1 (= R 1,PS-6-L -R 1,no L ) values were 8.5 times lower for H5V cells compared to RAW cells, and for ΔR 2 this was even 45 times. CLSM of the H5V cells  revealed no NIR fluorescence for incubations with PC-L and very few faint spots for PS-6-L ( Figure 3c).

Divalent cation dependency of liposome association with macrophages
The above-described experiments suggested that liposomes with 6 mol% DSPS (PS-6-L) were the most optimal formulation for targeting of macrophages. Therefore, PS-6-L was used in the experiments described from here. Association of PS-containing vesicles with the macrophage cell membrane depends on the presence of divalent cations such as Ca 2+ and Mg 2+ [11]. To test whether PS-6-L binding depended on the presence of divalent cations, which could be indicative for an interaction between these ions and PS resulting in membrane binding, RAW cells were incubated with PC-L and PS- With increasing Ca 2+ and Mg 2+ concentrations, the FACS fluorescence intensities of RAW cells increased for both PC-L and PS-6-L (Figure 4a and 4b, respectively). The average NIR fluorescence for cells incubated with PC-L or PS-6-L in medium lacking Ca 2+ and Mg 2+ (HBSS-) was equal (p >0.05, Figure 4c). Importantly, for incubations in medium with high Ca 2+ and Mg 2+ concentrations (HBSS+), fluorescence was significantly highest for PS-6-L (p <0.05 vs. all).
DLS showed that the diameter of both PC-L and PS-6-L increased after 2 h of incubation in HBSS+ (Figure 4d). For PC-L the average hydrodynamic diameter changed from 164.3 ± 0.9 nm in HBSS-to 209.5 ± 25.2 nm, while for PS-6-L the diameter increased from 104.6 ± 19.4 nm to 170.8 ± 33.5 nm. This size increase could additionally enhance the uptake of both types of liposomes by the cells.

Binding versus internalization
To study whether PS-6-L were internalized by macrophages, RAW cells were incubated with PC-L or PS-6-L in HBSS+ at either 4°C or 37°C. Incubation at 4°C inhibits phagocytosis and thus a comparison between 4°C or 37°C enabled a differentiation between binding to the cell membrane and internalization. FACS analysis of cells incubated at 4°C revealed no significant differences in average fluorescence intensities after incubation with PC-L and PS-6-L (Figure 5a-b). At 37°C, however, a significantly higher fluorescence intensity was observed for PS-6-L (p <0.05 vs. all, Figure 5a-b).
CLSM confirmed the FACS measurements (Figure 5c). Incubation of RAW cells with PC-L and PS-6-L at 4°C resulted in minor association of liposomes. CLSM using higher laser intensities showed that the liposomes appeared as a rim around every cell, bound to the cell membrane. No significant internalization was observed. For incubations with PC-L at 37°C CLSM images were comparable to incubations at 4°C, with minor association of liposomes, and higher laser intensities revealed that PC-L were mainly bound to the cell membrane. CLSM confirmed that incubation with PS-6-L at 37°C resulted in massive internalization of the liposomes, as shown by the high NIR signal inside RAW cells.

Discussion
Macrophages play a decisive role in several cardiovascular diseases. For example, in atherosclerosis high macrophage content is one of the hallmarks of plaque vulnerability [2]. The inflammatory response after myocardial infarction is important for cardiac remodeling and outcome [1]. Therefore, macrophages form a significant therapeutic target in cardiovascular diseases and tools for noninvasive MR imaging of macrophages are highly desired. Iron oxides have been successfully applied for the MR visualization of macrophages in cardiovascular diseases [15][16][17]. Nevertheless, targeting of iron oxides to CD11b/CD18, which is expressed on macrophages, did not improve specificity for MR imaging of macrophages in a mouse model of atherosclerosis [18]. Recently, Gd-labeled liposomes were used to visualize monocytes and/or macrophages infiltration in the mouse myocardium up to 7 days after myocardial infarction [19].
In this study, we describe the design and characterization of paramagnetic liposomes targeted to macrophages by incorporation of PS in the liposomal membrane. The liposomes contained Gd-DOTA-DSPE for MRI detection. Gd-DOTA-DSPE is a phospholipid that presents a high r 1 and the Gd-DOTA complex displays a high thermodynamic and kinetic stability [20]. As expected, at 9.4 T, the longitudinal relaxivity is not as high as at lower, clinical field strengths [20,21]. Importantly, incorporation of PS did not significantly affect liposomal r 1 and r 2 values. The r 2 /r 1 ratio of the liposome formulations at 9.4 T was relatively high, which means that the liposomes will display a significant T 2 effect as well. Nevertheless, by appropriately choosing the MRI sequence parameters, the T 1 effect of the liposomes can be effectively exploited ( Figure 2).
A distinct difference between the PS-containing liposomes used in this study and previously reported formulations for use in in vivo MRI studies is the incorporation of 5 mol% polyethylene glycol (PEG) lipids in the liposomal membrane. PEG reduces the interactions between the liposomes, reducing aggregation and ensuring a monodisperse formulation (Table 2 and Figure 1). Additionally, PEG increases the in vivo blood circulation half-life by reducing the interactions with plasma proteins, assuring a longer interaction time with macrophages [5]. According to previous studies, incorporation of 5 mol% PEG in PS-containing liposomes is not impeding the interaction of PS with macrophages, since at least 10-15 mol% PEG would be needed to completely shield the liposomes from any interactions with proteins [22][23][24]. We therefore did not expect a decrease in the uptake by shielding of the PS.
Liposomes containing 6 mol% PS resulted in the highest uptake by RAW murine macrophages (Figure 2). Maiseyeu et al. and Rimle et al. have observed optimal uptake by macrophages of liposomes without PEG when these contained 5-12 mol% PS [10,11]. Interestingly, these experimentally determined optimal concentrations are in the range of 2-10 mol% PS found in the membranes of mammalian cells [25], which suggests that macrophages are optimally equipped to recognize and phagocytose nanoparticles that express approximate physiological concentrations of PS. Association was specific for macrophages as uptake by endothelial H5V cells was significantly lower (Figure 3).
Uptake of PS-containing liposomes by macrophages was stimulated by the presence of divalent cations (Figure 4). Higher uptake was not primarily caused by divalent cation-mediated clustering of the liposomes, since incubation of liposomes in HBSS+ resulted in moderate changes in liposome size for both PC-L and PS-6-L. The HBSS+ buffer contained a physiologically relevant concentration of 1.26 mM Ca 2+ , compared to for example approximately 1.24 mM Ca 2+ in mouse blood [26]. For the PS-mediated recognition of apoptotic cells by macrophages, different engulfment receptors have been identified, such as scavenger receptors, oxidized low-density lipoproteins recognizing receptors and CD68 [27], which for the LOX-1 scavenger receptor has been proven to be Ca 2+ -dependent [28]. Which of these receptors are important for PS-mediated uptake of liposomes remains unknown.
With respect to MR imaging of liposome uptake, a relatively high association of PS-6-L with macrophages, as determined with ICP-MS, resulted only in a modest increase in R 1 (Figure 2). This is probably related to compartmentalization of PS-6-L in intracellular vesicles after phagocytosis, which limits effective access of bulk water protons to the Gd contrast agent [29,30]. T 1 shortening requires direct physical contact between Gd and water protons to be most effective. This interpretation is corroborated by the observation that the estimated cellular relaxivity of PS-6-L (r 1 = 0.8 ± 0.4 mM -1 Ás -1 ) was lower than the one of PS-6-L in aqueous solution (r 1 = 3.0 ± 0.3 mM -1 Ás -1 ). Furthermore, internalization of PS-6-L was observed by CLSM for incubations at 37°C ( Figure 5).
The next step will be to apply and study the uptake of PS-6-L in a relevant animal model of cardiovascular inflammation, for example in atherosclerosis or myocardial infarction. Christiansen et al. have shown that echocardiography of PS-containing microbubbles trapped in infarcted myocardium correlated moderately well with MPO activity, which are excreted by inflammatory cells [31].
Apart from use in imaging applications, PS-containing liposomes are a promising vehicle for targeted drug delivery. Liposomes loaded with Q10, ATP or adenosine delivered to infarct myocardium were demonstrated to reduce infarct size and salvage ischemic myocardium [32][33][34]. Also, liposomes have been used as a vehicle for delivery of glucocorticoids drugs to perform anti-inflammatory cancer therapy [12]. Targeting could enhance the specificity of drug delivery to macrophages. Alternatively, PS-liposomes themselves can be used for therapy of inflammation as well [14,[35][36][37]. As PS-liposomes mimic apoptotic cells, they inhibit pro-inflammatory cytokines release and promote secretion of anti-inflammatory cytokines. However, for therapy purposes higher PS-concentrations (up to 30 mol%) were used [14,36], which in this study did not enhance uptake by macrophages.

Conclusions
In summary, paramagnetic liposomes, containing 6 mol% of PS, showed enhanced uptake by macrophages compared to liposomes without PS, while significantly less uptake was observed for non-phagocytic cells. Association of PS-containing liposomes to macrophages was increased by the presence of divalent cations in the incubation medium and resulted mainly in internalization of liposomes, whereas only minor binding was observed. Therefore, these liposomes can be used for molecular MR imaging of macrophages and might as well be suitable for targeted drug delivery to macrophages in cardiovascular diseases.

Characterization of liposomes
Total lipid concentrations of the final liposome formulations were determined by a phosphate determination according to Rouser [38]. Hydrodynamic numberweighted size and size distribution were assessed with dynamic light scattering (DLS, ZetaSizer NanoS, Malvern Instruments, Worcestershire, UK) at 23°C.
To confirm the presence of DSPS lipids in the PScontaining liposomes, thin layer chromatography (TLC) was performed on an aluminum sheet coated with silica gel 60 F 254 (Merck BV, Schiphol-Rijk, the Netherlands) [39]. As eluent a mixture of chloroform, methanol, glacial acetic acid and water (65:25:8:4 v/v) was used. Liposomes were applied (expected concentrations of DSPS: PC-L 0 mg/mL, PS-6-L 2 mg/mL, PS-12-L 4 mg/mL and PS-37-L 6 mg/mL) and allowed to migrate for 30 min. As controls standard solutions of DSPS (0.5, 1, 2, 4 and 8 mg/mL) were used. Finally, primary and secondary amines in DSPS and Gd-DOTA-DSPE were detected with ninhydrin.
Liposomal longitudinal and transversal relaxation times (T 1 and T 2 ) were determined with a 9.4 T small animal MR scanner (Bruker Biospin GmbH, Ettlingen, Germany) equipped with a 35-mm-diameter quadrature birdcage RF coil (Rapid Biomedical, Rimpar, Germany). For T 1 measurements an inversion recovery fast low angle shot (FLASH) sequence was used, with the following parameters: overall repetition time (TR) 15 sec, TR 4 ms, echo time (TE) 2 ms, flip angle (α) 15 o , number of excitations (NEX) 4, field of view (FOV) 3x3 cm 2 , matrix 128x128, 1 mm slice thickness, 32 segments and 60 inversion times ranging from 72 to 4792 ms. T 2 relaxation times were determined using a multi-slice multi-echo sequence with the following parameters: TR 2000 ms, 32 TEs ranging from 9 to 288 ms, α 180 o , NEX 4, FOV 3x3 cm 2 , matrix 128x128 and 1 mm slice thickness. T 1 and T 2 relaxation times were calculated by mono-exponential fitting with a custom-built fitting program (Mathematica 6, Wolfram Research Europe, Oxfordshire, UK). Relaxivities r 1 and r 2 (in mM -1 Ás -1 ) were determined from R i = R i,0 + r i Á[Gd], with i = 1,2, R i = 1/T i , R i,0 the relaxation rate of a sample without liposomes and [Gd] between 0.001 and 1 mM Gd.

Association of PS-containing liposomes with RAW cells
To determine the mol% of PS present in liposomes resulting in maximal uptake by macrophages, RAW cells were incubated with PC-L, PS-6-L, PS-12-L and PS-37-L for 2 h at 37°C (1 mM total lipid). For MRI and inductively coupled plasma mass spectrometry (ICP-MS), cells were harvested by scraping and non-bound liposomes were removed by centrifugation (3x5 min, 500 g, RPMI medium at 37°C). Cells were fixed in 4% PFA (250 μL) and a loosely packed cell pellet was allowed to form by storage at 4°C (>2 days). For confocal laser scanning microscopy (CLSM), cells were cultured on coverslips. After incubation with the liposome formulations, cells were fixed with 4% PFA (20 min). Finally cells were washed with and stored in phosphate bufferd saline (PBS).

Association of PS-containing liposomes with H5V cells
To confirm that PS-containing liposomes were not taken up by endothelial cells, H5V cells were incubated with PC-L, PS-6-L, PS-12-L and PS-37-L for 2 h at 37°C (1 mM total lipid). For MRI, cells were washed with medium (37°C) and PBS (37°C). Afterwards, cells were harvested with trypsin/EDTA, fixed with 4% PFA and a loosely packed pellet was allowed to form. For CLSM, cells were cultured on gelatin-coated coverslips and handled as described above.

Divalent cation dependency of liposome association with RAW cells
The association of liposomes to RAW cells under different calcium and magnesium concentrations was studied. RAW cells were incubated with PC-L and PS-6-L (2 h, 37°C, 1 mM total lipid) in Hank's buffered salt solution (HBBS) containing 1.26 mM Ca 2+ and 0.90 mM Mg 2+ (HBSS+), HBSS without Ca 2+ and Mg 2+ (HBSS-) and RPMI medium (0.424 mM Ca 2+ and 0.407 mM Mg 2+ ). Afterwards, cells were harvested by scraping, washed in the appropriate medium (HBSS+, HBSS-or RPMI, 37°C), fixed in 4% PFA (20 min) and stored in 0.01% sodiumazide in PBS for FACS.
To investigate possible clustering of PC-L and PS-6-L under high calcium and magnesium concentrations, liposomes were incubated in HBSS+ or HBSS-(2 h, 37°C). Changes in hydrodynamic number-weighted diameter and size distribution were measured with DLS as described above at 37°C.

Binding versus internalization
To evaluate phagocytosis of PS-containing liposomes by macrophages, RAW cells were incubated with PC-L or PS-6-L at 4°C or at 37°C (1 mM total lipid in HBSS+, 2 h). Incubation at 4°C inhibits phagocytosis. For FACS, cells were incubated with liposomes in HBSS+ and harvested and washed as described above. For CLSM, cells were cultured and incubated with liposomes in microscopy chambers (Ibidi GmbH, München, Germany). Afterwards, cells were washed with HBSS+ (4°C or 37°C), fixed with 4% PFA (20 min), washed and stored in PBS.

Cellular relaxation rates and relaxivities
The cellular relaxation rates of cell pellets (R 1 and R 2 ) were determined at 9.4 T using the MRI protocol as described above. Furthermore, the cell pellet volume was determined using a 3D FLASH sequence with the following parameters: TR 25 ms, TE 3.7 ms, α 30 o , NEX 1, FOV 25.6x25.6x25.6 mm 3 and matrix 256x256x256. Cell pellets were segmented with OsiriX Imaging Software (www.osirix-viewer.com) and pellet volumes were calculated. The Gd content of cell pellets was determined with ICP-MS (DRCII, Perkin Elmer, Waltham, USA) after destruction in nitric acid and perchloric acid (1:2 v/v) at 180°C. Next, gadolinium concentrations were derived using the cell pellet volume. Cellular relaxivities were calculated from R i = R i,0 + r i Á[Gd], with i = 1,2, and R i,0 the relaxation rate of untreated cells.