Liposomal Encapsulation of Glucocorticoids Alters Their Mode of Action in the Treatment of Experimental Autoimmune Encephalomyelitis

Glucocorticoids (GCs) are in widespread clinical application because of their potent anti-inflammatory and immunomodulatory properties. The therapeutic spectrum ranges from the treatment of acute and chronic inflammation in atopic and autoimmune diseases to the induction of cell death in hematopoietic malignancies (1). Since their first application to patients with rheumatoid arthritis more than 70 y ago (2), a plethora of chemical derivatives differing in potency and side effects have been developed. For example, prednisolone represents the gold standard for the treatment of rheumatoid arthritis (3), whereas methylprednisolone (MP) is the first choice in multiple sclerosis (MS) pulse therapy (4). In contrast, the fluorinated compound triamcinolone is preferentially administered to asthma patients and used in ophthalmology (1, 5). Besides chemical modification, recent developments also include encapsulation in polyethylene-glycol-coated long-circulating liposomes (6), which results in a prolonged half-life in the circulation, slower blood clearance rates, and superior pharmacokinetics (7). Notably, liposome-encapsulated doxorubicin is already in clinical use for anti-cancer therapy (8). In the case of GC, encapsulation into liposomes has proved effective in a rat model of arthritis (9) and in the treatment of experimental autoimmune encephalomyelitis (EAE), a widely used animal model for MS (10, 11).

The mechanism and the cellular targets of GCs differ between individual diseases and remain incompletely understood. Generally, GCs induce lymphocyte apoptosis (12) and downregulate gene expression of cytokines and adhesion molecules in a variety of cell types (13, 14). In the context of neuroinflammatory diseases, modulation of T cells, macrophages, microglia, and the blood-brain barrier could all have potentially important roles (15). Traditionally, most of these effects are attributed to genomic actions (transactivation and transrepression) mediated by the cytosolic GC receptor (GR). However, some GC effects such as activation of endothelial NO synthetase or the drop in mitochondrial membrane potential occur very rapidly (16, 17). Therefore, nongenomic mechanisms of GC action were proposed, including a direct interaction of the GR with cytosolic signaling molecules and the intercalation of GC into the plasma membrane (18).

The features of different GC in the treatment of MS can be studied using its animal model EAE. In C57BL/6 mice, immunization with myelin oligodendrocyte glycoprotein peptide 35-55 (MOG_35-55) leads to a chronic disease course, characterized by a fulminant inflammatory response, demyelinating lesions, and subsequent axonal damage (19). Recently, we used this model to demonstrate that GR-dependent modulation of T cell functions is responsible for the therapeutic efficacy of free dexamethasone (Dex) in EAE (20). Dex induces apoptosis of peripheral but not CNS-infiltrating T cells. In addition, it downregulates adhesion molecules and thereby interferes with T lymphocyte recruitment into the CNS. Importantly, the presence of the GR in T cells but not in myeloid cells such as macrophages was essential for the treatment of EAE by free Dex, highlighting the target cell specificity of GC action in EAE.

In this study, we investigated a number of different GC derivatives and GC encapsulated in liposomes. We could confirm the superior therapeutic efficacy of liposomal GC compared with all tested free GC derivatives. Unexpectedly, our study revealed that liposome-encapsulated and free GC use different modes of action. Unlike the free compounds, liposome-encapsulated GCs exert only minor effects on peripheral T cells. In contrast, they strongly affect macrophage functions such as NO production, MHC class II (MHC II) surface levels, and cytokine expression. For this reason only the combined deletion of the GR in T cells and macrophages markedly impedes the beneficial effects of liposomal GC in the treatment of EAE. Thus, the formulation determines the mode of GC action and may influence their activity spectrum in MS therapy.

C57BL/6 mice were purchased from Harlan (Borchen, Germany) and used at an age of 8–12 wk. GR knockout (GRN), GR^flox, GR^lck, and GR^lysM mice on a C57BL/6 background were reported previously (20) and bred in our own animal facility under specific pathogen-free conditions. All animal experiments were approved by the responsible authorities in Lower Saxony (Niedersächsisches Landesamt für Verbraucherschutz und Lebensmittelsicherheit).

EAE was induced and analyzed as described previously (20). The mice were immunized with 50 μg MOG_35-55 peptide in CFA and treated twice with 400 ng pertussis toxin. Animals were weighed and scored daily for clinical signs of disease on a scale from 0 to 10 depending on severity. Scores were as followed: 0 = normal; 1 = reduced tone of tail; 2 = limp tail, impaired righting; 3 = absent righting; 4 = gait ataxia; 5 = mild paraparesis of hind limbs; 6 = moderate paraparesis; 7 = severe paraparesis or paraplegia; 8 = tetraparesis; 9 = moribund; 10 = death (20). Treatment with free GCs was performed with i.p. injection on three consecutive days at a dose of 100 mg/kg, starting once the mice had reached an average clinical score of 2–3. The following compounds were used: methylprednisolone-21-hydrogensuccinat (Urbason solubile, Sanofi-Aventis), prednisolone-21-hydrogensuccinat (Prednisolut, mibe, Jenapharm), triamcinolon-acetonid-21-hydrogen-phosphate (Volon A, Dermapharm, Grünwald), and dexamethasone-dihydrogen-phosphate (Dexa-ratio-pharm, Ratiopharm). Liposome-encapsulated GCs were injected once i.p. or i.v. on the first day of therapy at a dose of either 10 mg/kg or 3 mg/kg. Control mice were injected with equal volumes of PBS.

The liposomes were prepared as described previously using the film-extrusion method starting from a lipid solution containing dipalmitoyl phosphatidylcholine, polyethylene-glycol 2000-distearyl phosphatidylethanolamine, and cholesterol (11, 21). A lipid film was created, hydrated with a solution of prednisolone- or methylprednisolone-phosphate, and dispersed into particles containing approximately 4.5% prednisolone (range 3–6% for all preparations produced by the supplier) and an average of 60 μmol phospholipid. The liposome preparations were tested to be stable with no detectable leakage.

Histologic analysis was performed as described previously (20). Paraffin-embedded spinal cord sections were prepared from paraformaldehyde-perfused mice and stained with a rat anti-human CD3 Ab (1:200; Serotec, Düsseldorf, Germany) or a rat anti-mouse Mac-3 Ab (1:200; BD Biosciences, Heidelberg, Germany) followed by application of a biotinylated rabbit anti-rat Ab (1:200; Vector Laboratories, Burlingame, CA). Visualization was achieved using the peroxidase-based ABC detection system (Vector Laboratories). Amyloid precursor protein (APP) was stained with a mouse anti-APP A4 Ab (1:1000; Chemicon, Hofheim, Germany) and developed using a biotinylated goat anti-mouse/rabbit IgG (Dako, Hamburg, Germany). For quantification, 100 visual fields of the cervical, thoracic, and lumbar spinal cord were counted in a blinded fashion at a 400-fold magnification, two infiltrates per cross-section and three cross-sections per animal. A median was determined and the results were calculated as number of infiltrating cells per square millimeter (20). For CD163 staining, spinal cord sections were stained with a rabbit anti-mouse CD163 (1:1000; Santa Cruz Biotechnology, San Diego, CA) followed by biotinylated anti-rabbit IgG (1:200, Vector) and visualized with the peroxidase-based ABC detection system. Quantification was achieved by manual microscopic counting of positively staining cells from whole slices (9–12 slices per animal) from lumbar, thoracic, and cervical spinal cord sections in a blinded fashion.

All Abs and reagents were obtained from BD Biosciences unless otherwise indicated: anti-CD3ε (145-2C11), anti-CD4 (RM4-5), anti-CD8α (53-6.7), anti-CD11a/LFA-1 (2D7), anti-CD11b (M1/70), anti-CD44 (IM7), anti-MHC II (H2^b, AF6-120.1), anti-Mac-3 (M3/84), anti-GR-1 (RB6-8C5), and annexin V. The Abs were directly labeled with FITC, PE, PerCP, PE-Cy7, Cy5, allophycocyanin, or allophycocyanin-Cy7. F4/80 was purchased from Serotec. Stainings were performed as described previously (20) and analyzed using a FACSCanto II device (BD Biosciences) in combination with FlowJo software (Treestar, Ashland, OR).

The mice were injected with a 4% thioglycollate solution i.p. 4 days prior to the isolation of peritoneal exudate cells by repeated flushing with PBS as described (22, 23). Following centrifugation, the cells were seeded in 48-well plates in RPMI 1640 medium with 10% FCS and standard antibiotics and incubated for 3 h. Nonadherent cells were removed by repeated washing with PBS containing 0.1% BSA and the adhered macrophages used for experimentation. Purity of the preparations was usually ∼90% as determined by flow cytometry based on CD11b staining. The macrophages were directly used for FACS analysis or RNA isolation, or they were incubated for 48 h in the presence of 1000 U/μl IFN-γ and 20 ng/ml LPS followed by the measurement of NO production.

Lymphocytes were isolated from the spinal cord by density centrifugation following perfusion of the mice with NaCl as described previously (20). The dissected tissue was passed through a metal mash, homogenized, and resuspended after centrifugation in a Percoll gradient. Finally, the lymphocytes were harvested at the interfaces between the layers, washed with PBS and analyzed by flow cytometry or subjected to RNA preparation.

The macrophage culture supernatants were individually collected and used for the NO release assay. Supernatant (50 μl) was mixed with 50 μl of each of the two Griess reagents, 1% sulfanilamide, and 0.1% N-naphthylethylene-diamine-dihydrochloride (both diluted in 2.5% phosphoric acid) and incubated for 5 min at room temperature for the color to develop. The concentration of stable NO₂⁻ ions that are formed when the released NO reacts with H₂O was measured by spectrophotometry at 570 nm and compared with an NaNO₂ standard curve.

Microglia were isolated from primary mixed glial cell cultures as described previously (24, 25). Mixed glial cell cultures were prepared from brains of newborn mice on postnatal days 0, 3, or 5 and cultured for a period of 9–14 d (basal Eagle’s medium, 10% FCS, 50 U/ml penicillin, and 50 μg/ml streptomycin). Microglial cells were shaken off the primary mixed brain glial cell cultures and subjected to RNA preparation.

Total RNA from lymphocytes was isolated using the Quick-RNA MiniPrep (Zymo Research, Irvine, CA). Reverse transcription into cDNA was achieved with the help of the iScript kit (Bio-Rad, Munich, Germany). Quantitative real-time PCR was performed on an ABI 7500 instrument (Applied Biosystems, Darmstadt, Germany) using the SYBR Master Mix from the same company according to the manufacturer’s instruction. The results were normalized to the mRNA expression of hypoxanthine phosphoribosyltransferase and evaluated using the ΔΔCt method (26).

Microglial cells were digested with proteinase K followed by DNA extraction according to standard protocols. The nonrecombined (275 bp) and the recombined (390 bp) GR^flox alleles were amplified using the three primers GRflox-1, GRflox-4, and GRflox-8 as described previously (27). The PCR fragments were separated on a 1.5% agarose gel, and quantification of the band intensities was achieved using the GelPro Analyzer 4.5 software (Media Cybernetics, Bethesda, MD).

Analyses were performed with unpaired t test (Microsoft Excel and Graph Pad Prism Version 4). Data are depicted as mean ± SEM; p values >0.05 were considered as nonsignificant: *p < 0.05, **p < 0.01, ***p < 0.001. To determine differences referring to the disease course, the whole curves rather than individual time points were compared between experimental groups starting on the day after the first treatment.

We had reported previously that Dex efficiently ameliorates EAE in a dose-dependent manner (20). To assess the potency of other GC derivatives that are in clinical use for various indications, including MS therapy, we compared the fluorinated compounds Dex and triamcinolone with the nonhalogenated GC prednisolone and MP. EAE was induced by immunization of C57BL/6 mice with MOG_35-55, and upon disease manifestation the different GCs were administered at 100 mg/kg i.p. on three consecutive days. The potency of triamcinolone was comparable to the one of Dex, whereas the therapeutic efficacies of prednisolone and MP were lower (Fig. 1A). This finding indicates that the best results in the treatment of EAE are achieved with halogenated GCs.

Earlier studies demonstrated that liposome-encapsulated GCs are superior to free GCs in the treatment of EAE in rats, but the mechanism remained elusive (10, 11). To allow for characterization of their mode of action, we established a protocol to treat MOG-EAE in C57BL/6 mice with liposomal GCs. A single dose of 10 mg/kg liposomal prednisolone (PL) was injected i.p. into mice suffering from EAE and compared with our standard protocol of three consecutive daily treatments with 100 mg/kg free Dex (20). Importantly, the therapeutic benefit of a single administration of PL was similar to the repeated injection of the 10-fold higher dose of the more potent halogenated free Dex (Fig. 1B). Of note, PL remained significantly effective even at a concentration of 3 mg/kg, confirming that it acts in a dose-dependent manner (Supplemental Fig. 1). Furthermore, the effect of PL and liposome-encapsulated MP applied at a similar dose of 10 mg/kg was indistinguishable in this model (Fig. 1C).

Based on these results, we decided to perform further mechanistic experiments by comparing the effects of three i.p. injections of 100 mg/kg free Dex to one i.p. injection of 10 mg/kg PL. The dominant rationale to compare exactly these two treatment regimens was that their efficacies were similar (Fig. 1B). We considered this a better strategy than comparing liposomal with free prednisolone because the potencies of these two regimens strongly differ, making comparisons of the mechanisms difficult.

Next, we investigated whether the amelioration of EAE after PL therapy correlated with improved histopathologic features. In accordance with our previous studies (20), Dex administration significantly reduced the infiltration of T cells (CD3⁺) and macrophages (MAC3⁺) into the CNS on day 3 after the onset of therapy (Fig. 2). In striking contrast to its profound therapeutic efficacy, however, PL treatment had a much milder effect on leukocyte infiltration than Dex (Fig. 2). This finding was further confirmed by determining the number of infiltrating T cells and macrophages using FACS analysis (data not shown).

Besides differences in the inflammatory response, reduced early axonal loss could also explain the diminished disease severity after PL treatment. To address this issue, we stained spinal cord sections for APP (Fig. 2). Axonal loss was significantly reduced in mice 3 d after the onset of Dex therapy, but remained almost unchanged after application of PL (Fig. 2). Therefore, changes in histopathologic features do not adequately reflect the therapeutic benefit conferred by treatment of EAE by liposomal GC, which suggests that they ameliorate EAE by mechanisms distinct from those engaged by Dex.

Having the differential effects of Dex versus PL in histopathology, we next wondered how the GR is involved in suppression in inflammation by PL. First, we induced EAE in heterozygous GRN^+/− mice in which GR expression is strongly reduced (28). PL administration ameliorated EAE in GRN^+/+ control mice while mutant GRN^+/− mice were refractory to PL therapy (Fig. 3A). This finding suggests that the GR is essential for liposomal GC action.

Previous experiments revealed that Dex fails to ameliorate EAE in the absence of the GR in T cells (20). Because PL displays similar therapeutic efficacy as Dex, we wondered whether its action on EAE also depends on the presence of the GR in T cells. We induced EAE in GR^lck mice lacking the GR in the T cell lineage and treated the diseased mice with PL as in previous experiments. Importantly, GR^lck mice that are completely refractory to Dex therapy (20) could be treated effectively with a single injection of PL (Fig. 3B). In agreement with previous results (20), the disease developed earlier in GR^lck mice (data not shown) and the course was aggravated. Nevertheless, amelioration of EAE after application of PL was almost as efficient as compared with control GR^flox mice. Comparing the efficacies of PL administered i.p. versus i.v. excluded the possibility that the application route had any influence on the treatment success (Supplemental Fig. 2).

In an attempt to explain why the GR in T cells was of little importance for EAE therapy by liposomal GC, we investigated the modulation of different T cell features in healthy mice on day 1 to 3 after treatment. Dex strongly and rapidly reduced splenic T cell numbers within 3 d, whereas the effect of PL was much slower and overall weaker (Fig. 4A). The discrepancy between both treatment regimens was even more pronounced in regard to the modulation of cell adhesion molecules. As noted above, Dex significantly reduced surface levels of LFA-1 and CD44 within 3 d (Fig. 4B). In contrast, PL had no significant effect on the expression of either molecule, after neither 1 nor 3 d (Fig. 4B). Collectively, these data suggest that T cells are little affected, and that this does not make a major contribution to the therapeutic efficacy of PL in the treatment of EAE.

Because PL confers therapeutic benefit largely independently of T cell modulation, we wondered whether macrophages or microglia are targeted by PL instead. It is known that the GR is efficiently deleted in neutrophils and macrophages of GR^lysM mice (29), and we can now show that the same holds true for microglia (Supplemental Fig. 3). Our analysis of microglia isolated from newborn GR^lysM mice confirmed that recombination of the GR^flox allele on postnatal days 0–5 increased from 41 to 76% (Supplemental Fig. 3), presumably indicating that the GR is largely absent from adult microglia in GR^lysM mice. However, despite the broad deletion of the GR in macrophages, neutrophils, and microglia, EAE induced in GR^lysM mice was still significantly ameliorated after application of PL. Nonetheless, the overall efficacy of PL therapy in GR^lysM mice was partially reduced compared with GR^flox control mice (Fig. 5A). Therefore, myeloid cells play some role for PL therapy of EAE, but they are not an essential target.

Liposome-encapsulated drugs are preferentially taken up by phagocytic cells, and our results in GR^lysM mice suggest that the GR in myeloid cells plays at least some role during EAE therapy with PL. Therefore, we investigated whether macrophages from the peritoneal exudate of naive mice were modulated by PL. MHC II surface expression, a characteristic that affects the ability of macrophages to serve as APCs, was significantly reduced by Dex as well as PL for 1 (Fig. 5B) and 3 d (Supplemental Fig. 4A) after the onset of treatment. This effect was completely abolished in GR^lysM mice, confirming that it was strictly dependent on GR expression (Fig. 5C). We also analyzed NO production by IFN-γ/LPS-stimulated macrophages, a feature that is believed to contribute to the harmful effects of infiltrating macrophages and resident microglia during EAE and MS (30). Dex substantially diminished NO levels compared with controls and, although falling short of statistical significance, this effect was even consolidated after treatment with PL (Fig. 5D, Supplemental Fig 4B). Comparable results were obtained 1 and 3 d after the onset of treatment. Absence of the GR in macrophages fully abolished repression of NO production by PL (Fig. 5E).

Because the GR is a transcriptional regulator, we investigated the control of gene expression in peritoneal macrophages after Dex and PL treatment in vivo (Fig. 6). Repression of proinflammatory cytokines and chemokines by PL was either similar compared with Dex (IL-1β, IL-6) or augmented (IP-10; Fig. 6A). Furthermore, mRNA levels of some anti-inflammatory genes that are typical for alternatively activated macrophages were more efficiently upregulated by PL than by Dex (CD163, CD206, arginase 1, and Ym1; Fig. 6B). As exemplified for arginase 1 and Ym1 by using GR^lysM mice, the modulation of gene expression by PL was strictly dependent on GR expression in macrophages (Fig. 6C). In contrast, regulation of Fizz 1, another marker of alternatively activated macrophages, and of the anti-inflammatory cytokines TGF-β and IL-10 was less systematic, and the alterations in gene expression induced by Dex and PL were only minor (Fig. 6B). Thus, liposomal GCs potently inhibit proinflammatory functions of macrophages and polarize them toward the anti-inflammatory M2 phenotype.

Guided by our observation that PL had only a minor effect on leukocyte infiltration into the CNS (Fig. 2) while potently regulating gene expression of macrophages (Fig. 6), we investigated infiltrating leukocytes from diseased mice 3 d after treatment with PL or Dex. The mRNA levels of the proinflammatory mediators IFN-γ, IL-17A, RANTES, IL-6, IL-1β, and IP-10 were significantly reduced by PL and Dex without any obvious difference between the two treatment regimens (Fig. 7A). The situation was somewhat different in regard to genes associated with the M2 phenotype of alternatively activated macrophages. Annexin 1 was more strongly elevated by PL compared with Dex, the expression levels of CD163 and CD206 were equally increased by Dex and PL, and TGF-β levels were unchanged (Figs. 7B, 8A). Analysis with immunohistochemistry revealed that CD163⁺ cells were mainly found in the meninges in the spinal cord of mice suffering from EAE and confirmed that their numbers were significantly increased after PL treatment (Fig. 8B, 8C). It thus appears that, although PL only slightly diminishes leukocyte infiltration into the CNS, it impedes the proinflammatory function of infiltrating leukocytes and additionally polarizes the macrophages toward an anti-inflammatory phenotype. This in turn would provide one explanation for the therapeutic benefit gained from the treatment of EAE with liposome-encapsulated GCs.

Because the individual lack of the GR in either T cells or macrophages has only a minor effect on the efficacy of PL in the treatment of EAE, we deleted the GR simultaneously in both compartments. EAE was induced in GR^lcklysM double-knockout (dko) mice and treated with PL according to our standard protocol. The onset of the disease in GR^lcklysM dko mice was earlier than in GR^flox control mice (data not shown), and the severity of EAE was strongly aggravated in the combined absence of the GR in T cells and myeloid cells (Fig. 9). Importantly, GR^flox control mice efficiently responded to PL application, whereas in GR^lcklysM dko mice the response was clearly weaker (Fig. 9). Nevertheless, there was some residual treatment efficacy left in GR^lcklysM dko mice. Thus, PL action during the treatment of EAE is mediated by the GR in T cells and macrophages–microglia, but it must additionally target other cell types presumably of nonhematopoietic origin. We therefore conclude that liposomal encapsulation of GCs strictly alters their target cell specificity compared with free GCs.

High-dose MP pulse therapy is the measure of choice in the clinical management of MS, but systematic studies on the use of other GC derivatives and formulations such as liposomes are rare (4, 31). Our data demonstrate that GCs fall into two categories concerning their efficacy in the treatment of EAE, with the two fluorinated compounds Dex and triamcinolone being superior to the nonhalogenated GC prednisolone and MP. This finding is in contrast to a report showing that MP completely reversed EAE in the C57BL/6 model, where the dosage, delivery route, and application frequency differed significantly from our protocol (32).

Rat models have shown that liposome-encapsulated GCs are more preferable than free GCs in EAE therapy (10, 11). They have also been used successfully to treat proteolipid protein-induced EAE in SJL mice, but without further addressing their mode of action (33). Furthermore, a good therapeutic effect was achieved in a rat model of adjuvant arthritis by using a dosing regimen comparable to ours (34). To obtain further insight into the specific mechanisms of liposome-encapsulated GCs, we applied the MOG-EAE model in C57BL/6 mice because there are ubiquitous and conditional GR knockout mice available on this genetic background. We found that PL and free Dex are equally effective in ameliorating disease symptoms under the applied conditions despite injecting PL only once instead of three times, using a 10-fold lower dosage of PL, and the lower potency of prednisolone compared with the halogenated Dex. We provide surprising evidence that liposomal encapsulation alters the mode of GC action compared with free GC.

Liposome-encapsulated drugs are taken up by phagocytes such as macrophages, which leads to the hypothesis that this could be the preferential route of transport to inflamed tissues and could result in a higher local GC concentration. Accordingly, liposomes were found to accumulate in the CNS of EAE rats (10) and in the inflamed joints during collagen-induced arthritis (35). Taking this into consideration, we found that the mode of action and especially the GC target cells fundamentally differ between liposomal and free GCs. Our previous study revealed that free Dex primarily acts on peripheral T cells while modulation of macrophage function was dispensable for the efficacy of EAE therapy. Contrary to this, we found that PL has little effect on peripheral T cell numbers and that the expression levels of CD44 and LFA-1 are not diminished by PL at all. Considering these findings we conclude that, in contrast to free Dex, liposomal GCs exert only marginal effects on T cell functions. This conclusion is also in agreement with our observation that T cell-specific deletion of GR hardly corrupts the treatment efficacy of PL. Therefore, T cells are no major target of liposomal GC action in the treatment of EAE.

The concept that liposomes are preferentially taken up by phagocytic cells concurs with our observation that effects of PL on macrophages are equivalent to free GCs or even more pronounced in some aspects. Production of NO, which exerts deleterious effect in EAE and MS (36, 37), and MHC II surface expression, which is required for Ag presentation, are similarly reduced by PL and free Dex in peritoneal macrophages after in vivo treatment. Expression of proinflammatory cytokines and chemokines was equally or even more strongly repressed by PL than by Dex. Furthermore, the expression of a number of molecules associated with alternative M2 macrophages (38), which are implicated in anti-inflammatory responses and tissue remodeling and characterized by the expression of surface (CD163, annexin 1, CD206), intracellular (arginase 1), and secreted (Ym1) molecules, was preferentially upregulated by PL in the periphery and in the spinal cord. Nonetheless, deletion of the GR in macrophages and microglia reduced the treatment success of PL only to a minor degree. Only the combined deletion of the GR in T cells and myeloid cells led to a marked but still incomplete reduction of the therapeutic efficacy of PL. There are two conclusions one can draw from this observation. First, T lymphocytes and macrophages or microglia can substitute each other to some degree as target cells of liposomal GCs, and neither alone is essential as is the case of EAE therapy by Dex. Second, there is some residual treatment capacity of PL left, even when the GR is deleted in both cell types, which indicates that there are mechanisms at work beyond the modulation of hematopoietic cells. It had been suggested previously that the higher effective GC concentration owing to the engulfment of PL by phagocytes could favor nongenomic effects by other receptors capable of binding GC. However, our analysis of heterozygous GR knockout mice clearly shows that the GR is indispensable for PL therapy of EAE. As a result, we believe that the most likely explanation of our finding is that nonhematopoietic cell types are targeted by liposomal GCs and therefore contribute to their therapeutic efficacy. Possible candidates are endothelial cells or neural cells, and it will be important in the future to determine these specific targets because they may potentially open new perspectives of how to interfere with human MS.

A striking observation was that the histopathologic features were strongly affected by Dex, but only slightly by PL despite similar therapeutic efficacy of both regimens. How does PL ameliorate the disease symptoms if not by reducing CNS infiltration? In view of GCs being shown to be neurotoxic (39), we quantified APP-positive axons as a measure of early neuronal damage. However, the number of APP-positive axons was only slightly reduced in the CNS of PL-treated mice and did not explain the discrepancy between the clinical outcome and CNS infiltration. It is known that GCs restore the integrity of the blood-brain barrier (40, 41); however, if PL would preferentially target the endothelial cells of the blood-brain barrier, this should concomitantly diminish T cell and macrophage infiltration, which is not the case. An alternative explanation is provided by our observation that PL affects the phenotype of the macrophages, because we detected an upregulation of markers characteristic for alternatively activated M2 macrophages (38) by PL, both in the periphery and in the infiltrated CNS. It is known that GCs have the capacity to polarize macrophages toward the M2 phenotype (42, 43). Furthermore, M2-polarized monocytes and microglia have been associated with a reduced disease severity of EAE (44), and M2-immunomodulating myeloid cells have been described in MS lesions (45). Thus, it is tempting to speculate that liposomal-encapsulated GCs favor the conversion of macrophages to the M2 phenotype, which then exert anti-inflammatory rather than proinflammatory activity. This process could either happen in the periphery followed by the migration of the monocytes–macrophages into the CNS or directly in situ after infiltration of the spinal cord, or both. We propose that this effect in part explains the excellent therapeutic efficiency of PL despite the marginally reduced lymphocyte infiltration.

Collectively, liposomal encapsulation of GCs augments their therapeutic value and alters their target cell specificity and their mode of action. This finding clearly indicates liposomal GCs as a promising future therapeutic tool for the treatment of MS, but the altered target cell specificity has to be considered when applying this formulation in clinical trials.

We thank Alexandra Bohl, Martina Weig, Birgit Curdt, Nancy Meyer, Amina Bassibas, and Julian Koch for technical assistance and Cathy Ludwig for language corrections.

The authors have no financial conflicts of interest.