Mesenchymal stem cell application for treatment of neuroinflammation-induced cognitive impairment in mice

Neuroinflammation is an inflammatory response within the CNS. It is beneficial as a physiological response to trauma or infection. However, chronic neuroinflammation leads to cognitive impairments [1]. Many neurological disorders and age-related neurodegenerative diseases are accompanied by neuroinflammation [2–8].

The nAChRs expressed in the brain are involved in cognitive functions such as memory, thinking, comprehension, learning capacity, orientation and language [9–12]. The α4β2 nAChRs are essential for learning and memory [13], whereas α7 nAChRs are involved in the cholinergic anti-inflammatory pathway [14,15]. nAChRs are abundantly expressed within the brain in neurons, astrocytes and microglia [9,16,17]; play a substantial role in controlling neuroinflammation [18]; and directly interact with Aβ_1–42, the main pathogenic factor in Alzheimer's disease [19]. In addition, nAChRs found in the outer membrane of mitochondria control the mitochondrial pathway of apoptosis [20] and therefore support the viability of brain neurons [21].

Previously, the authors reported that intraperitoneal injections of bacterial lipopolysaccharide (LPS) resulted in neuroinflammation in mice and caused episodic memory impairment accompanied by a decrease in α7 nAChRs and accumulation of Aβ_1–42 in the brain [22]. The brain mitochondria of LPS-treated mice also contained less α7 nAChRs and more Aβ_1–42 and became less resistant to apoptogenic influence [23]. Therapeutic approaches that appear to be efficient in preventing or curing these negative effects of neuroinflammation include the α7 nAChR-specific orthosteric agonist PNU282987 [24], the membrane-stabilizing anti-inflammatory drug N-stearoylethanolamine [25] and mesenchymal stem cells (MSCs) [26].

MSCs are multipotent cells capable of differentiating into various cell types, including neurons. In addition, MSC-produced humoral factors stimulate neurogenesis and are neuroprotective [27,28]. At present, MSCs isolated from bone marrow, adipose tissue, skeletal muscle, heart, umbilical cord and placenta are used in numerous experiments and clinical trials aimed at regenerating peripheral tissues as well as the damaged brain [29].

Previously, the authors reported that MSCs isolated from human umbilical cord and intravenously injected into LPS-treated mice penetrated the brain and prevented (if injected simultaneously with LPS) or cured (if injected 3 weeks after LPS) the negative consequences of neuroinflammation, restoring episodic memory and levels of nAChRs and Aβ_1–42 in the brain and brain mitochondria. The positive effect appeared to be mediated by soluble factors produced by MSCs, as injections of MSC-conditioned medium were also efficient [26]. In the present study, the authors aimed to optimize MSC dose, source and route of injection to investigate the length and stability of the therapeutic effect of either MSCs or their conditioned medium and to explore the possibility of potentiating this with SDF-1α, which is known to contribute to cell proliferation and neurogenesis within the brain [30].

Methods

Materials

All reagents were of chemical grade and purchased from Sigma-Aldrich (MO, USA) unless specifically indicated. Antibodies against α4 (181–192), α7 (179–190), β2 (190–200) and α7 (1–208) nAChR fragments and rabbit cyt c-specific antibodies were generated using methods previously developed in the authors' lab [31–34]. The antibodies were biotinylated according to standard procedures [35]. The antibodies anti-Nestin, anti-mouse Fab Alexa 555-conjugated and anti-rabbit Fab Alexa 488-conjugated as well as the kits for IL-1β, IL-6, IL-10 and TNF-α determination (Invitrogen, MA, USA) and NeutrAvidin–peroxidase conjugate (Molecular Probes, OR, USA) were purchased from ALT Ukraine, an official representative of Thermo Fisher Scientific in Ukraine. In addition, BrdU (category number: 5002) and anti-BrdU (category number: B8434) were purchased from Sigma-Aldrich, and antibodies against Aβ_1–42 (category number: ab2539) were purchased from Abcam (Cambridge, UK).

Animals

C57BL/6J female mice aged 2–3 months were kept in the animal facility of Palladin Institute of Biochemistry, National Academy of Sciences of Ukraine, Kyiv, Ukraine, in quiet, temperature-controlled rooms and provided with water and food pellets ad lib. All procedures, including intravenous and intraperitoneal injections and behavioral experiments, as well as death by cervical dislocation, were carried out in accordance with Directive 2010/63/EU for animal experiments and were approved by the animal care and use committee of Palladin Institute of Biochemistry. Female mice were used because they are not as aggressive as male mice. The authors did not observe any specific hormonal impact in the behavioral experiments.

MSC isolation, propagation & characterization

Umbilical cord-derived MSCs (UC-MSCs) were obtained from the Wharton jelly of healthy donors (39–40 weeks of gestation) who signed an informal agreement to provide material for scientific research as previously described [26]. Briefly, the Wharton jelly was cut into 0.4- to 0.5-mm pieces, which were then cultured in humidified air with 5% CO₂ at 37°C in α-MEM complete growth medium (Biowest, Austria) supplemented with 10% fetal bovine serum (Invitrogen), penicillin 100 U/ml (Arterium, Ukraine) and streptomycin 100 μg/ml (Arterium, Ukraine). The medium was changed every 4–5 days. After 14 days of cultivation, cells reached 70–80% confluence and were passed using the standard method with trypsin–EDTA [36]. Expression of the surface marker proteins CD34, CD45, CD90, CD73 and CD105 was determined at the second passage by flow cytometry (BD FACSAria) with fluorescein isothiocyanate- and phycoerythrin-conjugated antibodies (United States Biological, MA, USA) according to minimal criteria for defining multipotent mesenchymal stromal cells [37]. Cells of the second passage were used for transplantation into LPS-treated mice. The cell-conditioned medium was obtained by culturing second passage UC-MSCs in serum-free medium for 2 days.

Adipose-derived stem cells (ASCs) were obtained as previously described [38–40] from the liposuction material of donors who provided an informal agreement to use their biological material in scientific research. Briefly, adipose tissue obtained upon liposuction was rinsed twice with phosphate-buffered saline (PBS) and centrifuged at low speed, and an equal volume of collagenase type I was added to the final concentration of 0.1%. The mixture was incubated at 37°C for 45 min with periodic shaking and then centrifuged at 800 × g. The pellet was then rinsed with PBS, and the cells were seeded in plastic flasks. Cells were propagated in a manner identical to that used for UC-MSCs and utilized in the experiment at the level of the second passage.

Obtaining rhSDF-1α

The authors obtained rhSDF-1α by expression of the protein in Escherichia coli strain BL21 (DE3) transformed by pET24-hSDF-1α according to the standard autoinduction protocol [41]. For the isolation of inclusion bodies containing rhSDF-1α, E. coli cells were disrupted by lysozyme treatment. SDF-1α refolding and chromatographic purification were performed as previously described [42]. The biological activity of the purified protein was analyzed by transwell migration assay with human lymphocytes [43].

Animal treatment & brain preparations

In the first set of experiments, mice were treated as described by Lykhmus et al. [26]. Three groups of C57Bl/6 mice, with five animals per group, were intraperitoneally injected with 2 mg kg^-1 LPS (E. coli strain 055:B5) in 0.1 ml of saline on day 0. The same day, one of these groups received a single injection of 10⁶ UC-MSCs in 0.1 ml of incubation medium in the tail vein. Another group was injected intraperitoneally with 0.3 ml of UC-MSC-conditioned medium. Injections of conditioned medium were repeated every 7 days, starting from day 0, for 7 more weeks. All groups of mice were examined in behavioral tests every week, and blood was taken from the tail vein on days 0, 14, 28 and 56. Mice were killed a week after the last conditioned medium injection (day 56), and their brains were removed. The brain of one mouse from each group was used for electron microscopy examination as described later. The brains of other mice were homogenized, fractionated into mitochondria and non-mitochondria and characterized as previously described [44,45]. The freshly isolated mitochondria were examined for cyt c release under the effect of apoptogenic factors. In addition, both mitochondrial and non-mitochondrial fractions were lysed in detergent-containing buffer for further use in ELISA experiments [23]. Protein concentration was measured using a BCA kit (Thermo Fisher Scientific, France).

In the second set of experiments, the authors used six groups of mice, with five animals in each group, injected with LPS as described earlier. 2 weeks after LPS injection (day 14), mice received a single intravenous injection of 0.1 ml of 10⁴ UC-MSCs, 10⁵ UC-MSCs, 10⁵ UC-MSCs followed by SDF-1α, 10⁶ UC-MSCs or 10⁶ ASCs. One group was injected intraperitoneally with 0.3 ml of 10⁶ UC-MSCs. SDF-1α (20 μg per mouse) was introduced intraperitoneally a week after UC-MSCs. Mice were examined in behavioral tests every week during the first 6 weeks and again on days 69 and 119. Mice were killed on day 119, and their brains were removed, homogenized in detergent-containing buffer as previously described [23] and used for determination of nAChR subunits and Aβ_1–42 by sandwich ELISA.

In the third set of experiments, mice were divided into five groups, with five animals in each group. One group remained untreated, and four groups were injected a total of three times with 2 mg kg^-1 LPS (E. coli strain 055:B5) in 0.1 ml of saline, with a 1-month interval between injections. Two weeks after the last LPS injection, one group received a single intravenous injection of 10⁶ UC-MSCs, a second group received 10⁶ UC-MSCs followed by SDF-1α and a third group was intraperitoneally injected with 10⁶ UC-MSCs. Two weeks after MSC injection, one mouse per group was injected intraperitoneally with BrdU (2.0 mg per mouse in 0.2 ml of PBS). Mice were examined using a novel object recognition (NOR) test every week and killed on day 21 after MSC injection. Their brains were homogenized in detergent-containing buffer as previously described [23] and used for determination of cytokines, nAChR subunits and Aβ_1–42 by sandwich ELISA. BrdU-injected mice were killed on day 7 after BrdU injection, and their brains were fixed with 4% paraformaldehyde and used for immunohistochemistry as described later.

Electron microscopy

Animals were deeply anesthetized with Calypsol (ketamine, Malladi Drugs & Pharmaceuticals Ltd, India) 75 mg/kg and transcardially perfused with PBS followed by 4% formaldehyde and 2.5% glutaraldehyde solution in PBS (pH 7.4). Brains were then washed in PBS. Horizontal 150-μm sections were cut using a VT1000 A vibratome (Leica, Wetzlar, Germany), postfixed in 1% osmium tetroxide in PBS, dehydrated in ascending concentrations of alcohol and embedded in EPON 812 (Fluka Chemie AG, Switzerland). The EPON-embedded samples were ultrathin-sectioned (50–70 nm thick) with a diamond knife, picked up on 150-mesh copper grids and stained with lead citrate and uranyl acetate. A 100CX electron microscope (JEOL, Tokyo, Japan) operating at 80 kV was used to examine the grids. Photos were taken at a magnification of ×10,000 and scanned with a flatbed scanner at 600 dpi. Morphology of the mitochondria was then analyzed.

Sandwich ELISA assays

Levels of nAChR subunits and Aβ_1–42 bound to α7 nAChR in the brain or mitochondrial preparations were determined as described by Lykhmus et al. [26]. Detergent lysates of brain tissue or mitochondria were applied to the wells of Nunc MaxiSorp immunoplates (Thermo Fisher Scientific, France; 1 μg of protein per 0.05 ml per well) coated with rabbit α7 (1–208)-specific antibody (20 μg/ml). The bound subunits were detected with secondary biotinylated α4 (181–192)-, α7 (179–190)- or β2 (190–200)-specific antibody, and the bound Aβ_1–42 was detected with biotinylated Aβ_1–42-specific antibody. The bound biotinylated antibodies were revealed with NeutrAvidin–peroxidase conjugate and o-phenylenediamine-containing substrate solution. The optical density was read at 490 nm using a Stat Fax 2000 ELISA reader (Awareness Technologies, CT, USA). Cytokine levels (IL-1β, IL-6, IL-10 and TNF-α) were measured by sandwich ELISA according to the recommendations of the kit manufacturer (Invitrogen).

Cyt c release from mitochondria

The freshly isolated mitochondria (120 μg of protein per 1 ml) resuspended in 10 mM HEPES, 125 mM KCl, 25 mM NaCl, 5 mM sodium succinate and 0.1 mM Pi(K) (pH 7.4) were incubated with different CaCl₂ doses for 5 min at room temperature and immediately pelleted by centrifugation (10 min at 7000 × g) at 4°C. The supernatants were tested for the presence of cyt c by sandwich ELISA assay as previously described [34,44]. Briefly, the immunoplates were coated with cyt c-specific polyclonal antibody and blocked with 1% bovine serum albumin. The mitochondrial supernatants were applied to the wells with adsorbed antibody for 2 h at 37°C. After extensive washing, the biotinylated anti-cyt c antibody was applied for an additional 1 h to be further revealed with NeutrAvidin–peroxidase conjugate and o-phenylenediamine-containing substrate solution. The authors used similar antibodies for coating and detection because, as a result of their polyclonal nature, they recognized multiple epitopes on the cyt c molecule. Such an approach has been shown to be efficient in multiple previous experiments [23,24,26,34,44,45].

Behavioral experiments

Mice were tested using the NOR behavioral test as previously described [22,46] prior to and post-treatment. Briefly, mice were encouraged to explore two identical objects for 15 min, and then, after a 15-min break, one object was replaced with a novel one of similar size but different shape or color, and the number of explorations of the two objects was registered in a subsequent 15-min session. Results of the NOR test are presented as discrimination index, calculated as the difference in number of ‘novel’ and ‘familiar’ object explorations divided by the total number of explorations of two identical objects. The lack of novel object preference, expressed as a decrease in discrimination index, was qualified as episodic memory impairment.

Immunohistochemistry

Mouse brains were fixed with 4% paraformaldehyde and washed in 0.1 M PBS. Transverse brain sections 30–40 μm thick were cut starting from the dorsal end using a VT1000 A vibratome and kept in 0.1 M PBS with 0.02% sodium azide at 4°C. Before immunohistochemistry, the sections were incubated in 2 N HCl for 30 min at 37°C, washed with PBS and processed in blocking buffer (1% bovine serum albumin and 0.5% Triton X-100, Sigma-Aldrich in PBS) for 1 h at room temperature. The slices were then incubated overnight at 4°C with a mixture of the primary antibodies monoclonal mouse anti-BrdU (1:200) and polyclonal rabbit anti-Nestin (1:100) dissolved in blocking buffer. Next, the slices were washed with 0.01 M PBS and incubated in the dark at room temperature for 2 h with the fluorescently labeled secondary antibodies donkey anti-mouse Fab Alexa 555-conjugated (1:500) and anti-rabbit Fab Alexa 488-conjugated (1:500) dissolved in 0.1 M PBS. The slices were then washed with 0.01 M PBS, mounted on glass slides in fluorescent mounting medium (Dako, CA, USA) and analyzed with a FluoView^™ FV1000-BX61WI confocal microscope (Olympus America Inc., MA, USA) at a magnification of ×400. Images of the hippocampal dentate gyrus region were captured at 1024 × 1024 resolution and had similar brightness and contrast for all samples. A total of 30 immunohistochemistry images were analyzed for each of the four groups of mice. Quantitative analysis was performed using ImageJ software (http://rsb.info.nih.gov/ij/index.html). Only cells co-stained for BrdU and Nestin were counted. Results are presented as the number of BrdU + Nestin+ cells/mm².

Statistical analysis

Both ELISA and behavioral experiments were performed in triplicate for each mouse, and mean values for individual mice were used for statistical analysis using one-way analysis of variance and Origin 9.0 software. A total of 30 immunohistochemistry images were examined from each experimental group. Statistical analysis was performed with Excel tables (Microsoft Corporation, WA, USA) and Prism 5.0 software (GraphPad Software, CA, USA) for Windows (Microsoft Corporation). One-way analysis of variance with Tukey's post hoc test was used for evaluation of group differences in variables with a normal distribution. Data are presented as mean ± standard deviation or mean ± standard error. Differences between groups were considered significant at p < 0.05.

Results

Effect of UC-MSCs or their conditioned medium when injected simultaneously with LPS

Behavioral experiments demonstrated that a single LPS injection resulted in episodic memory impairment after 1 week, and memory was not restored up to day 56 (Figure 1A). A single injection of UC-MSCs prevented memory decline during all periods of observation. Conditioned medium injections were efficient for 2 weeks, after which discrimination index, as a measure of episodic memory, went down. LPS injection provoked a stable IL-6 increase in the blood of mice (measured on days 14, 28 and 56), which was prevented by either MSCs or conditioned medium injections (Figure 1B). The blood sera of mice injected with UC-MSC-conditioned medium, but not with UC-MSCs, contained antibodies against components of the conditioned medium (Figure 1C). The antibodies appeared after the first injection and reached a plateau by day 30 (data not shown). Antibodies capable of binding total UC-MSC antigens were found in the blood of all groups of mice and were not increased upon UC-MSC injection.

LPS injection decreased levels of α4, α7 and β2 nAChR subunits in the brains of mice, and both UC-MSCs and their conditioned medium restored levels (Figure 2A). LPS injection resulted in a significant increase in α7-bound Aβ_1–42, whereas both UC-MSCs and conditioned medium prevented it (Figure 2B).

Transmission electron microscopy analysis was performed on the CA1 region of the hippocampus and specifically focused on mitochondria. Mitochondria were identified by distinctive features, such as a double membrane with the inner membrane folded into the matrix in the form of finger- or plate-like cristae and an internal matrix of varying density that sometimes contained discrete electron-dense granules. As shown in Figure 3A, mitochondria of the control group were round, with a clearly defined double membrane, regularly distributed cristae and compact structure. LPS treatment resulted in the appearance of swollen mitochondria with a vacuolated matrix, ruptured or blurred cristae and defective outer membrane. Destructive changes were less pronounced in the brain mitochondria of mice treated with MSC-conditioned medium, and UC-MSC treatment prevented these morphological changes altogether, with mitochondria having the same shape as that observed under control conditions.

Isolated brain mitochondria of LPS-treated mice contained less α4, α7 and β2 nAChR subunits compared with control mice. Both UC-MSCs and their conditioned medium prevented a decrease in nAChR subunits (Figure 3B) and decreased Ca²⁺-stimulated cyt c release from isolated mitochondria. Conditioned medium produced a weaker effect compared with UC-MSCs (Figure 3C).

Dose, source & route of injection of MSCs injected 2 weeks after LPS & effect of SDF-1α

In the second set of experiments, MSCs were injected 2 weeks after LPS, when memory decline was already observed. In this study, the authors compared the effects of different UC-MSC doses (10⁴, 10⁵ or 10⁶ per mouse) and revealed the effect of SDF-1α. The authors also compared intravenous and intraperitoneal routes of injection using MSCs obtained from both human adipose tissue (ASCs) and umbilical cord (UC-MSCs).

As shown in Figure 4A, LPS injection resulted in a significant decline in episodic memory as measured by the NOR test. Injection of 10⁶ UC-MSCs restored memory after 2 weeks, and the effect was maintained up to day 119. Doses of 10⁴ and 10⁵ UC-MSCs per mouse were less efficient; memory restoration was not complete and tended to decrease with time. However, mice that received SDF-1α injection a week after 10⁵ UC-MSCs demonstrated rapid and continuous memory improvement that was similar to that observed in mice that obtained a 10⁶ dose. The efficiency of intraperitoneally injected ASCs and UC-MSCs was similar to that noted in intravenously injected UC-MSCs (Figure 4B).

The brains of mice injected with LPS only and examined at the end of the experiment (day 119) demonstrated a significant decrease in α4 nAChR subunits and a tendency toward a decrease in α7 subunits. Injections of 10⁴ and 10⁵ UC-MSCs did not influence nAChR levels in the brain, whereas 10⁵ UC-MSCs + SDF-1α, 10⁶ UC-MSCs and 10⁶ ASCs significantly increased levels of α4 and α7 nAChR subunits. Intraperitoneally injected ASCs and UC-MSCs also augmented levels of β2 subunits, with intraperitoneally injected UC-MSCs appearing to be the most efficient (Figure 4C). LPS injection increased the level of α7-bound Aβ_1–42 (Figure 4D), but the level was decreased by 10⁶ UC-MSCs injected either intravenously or intraperitoneally as well as 10⁶ ASCs and 10⁵ UC-MSCs + SDF-1α, but not by 10⁴ or 10⁵ UC-MSCs.

Effect of UC-MSCs after regular LPS injections

In the third set of experiments, the authors aimed to reveal the effect of MSCs on brains significantly damaged by chronic inflammation. For this purpose, mice were injected with LPS a total of three times, with a 1-month interval between injections, and a single injection of UC-MSCs was given 2 weeks after the last LPS injection. In this study, the authors used a similar dose (10⁶ per mouse) of intravenously injected UC-MSCs followed by either SDF-1α or no SDF-1α and intraperitoneally injected UC-MSCs. As shown in Figure 5A, long-term LPS treatment resulted in a significant decline in episodic memory. Intravenously injected UC-MSCs restored memory after 2 weeks, whereas intraperitoneally injected UC-MSCs and UC-MSCs followed by SDF-1α began to restore memory after 1 week.

Mice were killed 3 weeks after UC-MSC injection, when their memory was completely restored. Brains were examined for levels of pro- and anti-inflammatory cytokines, nAChR subunits and α7-bound Aβ_1–42. The authors found that levels of IL-1β, IL-6 and IL-10 in the brains of LPS-treated mice were not changed compared with those in non-treated mice, whereas TNF-α level was slightly increased (Figure 5B). Intravenously injected UC-MSCs, either with or without SDF-1α, but not intraperitoneally injected UC-MSCs, significantly decreased both IL-10 and TNF-α levels.

Regular LPS injections decreased levels of α4 nAChR subunits but increased levels of β2 subunits; therefore, it seemed likely that the decrease in α4-containing nAChRs was compensated by other β2-containing nAChR subtypes. UC-MSCs augmented α4 and α7 subunits. Intraperitoneally injected UC-MSCs also decreased levels of β2 subunits (Figure 5C). Moreover, LPS treatment increased the level of α7-bound Aβ_1–42 in the brain, whereas all three types of UC-MSC injections decreased it (Figure 5D).

To understand the mechanism of the SDF-1α effect, the authors took into account that this factor is able to improve homing and proliferation of neural progenitor cells in the brain [30]. To test whether this was the case in the current study, the authors injected mice with BrdU a week after SDF-1α injection (2 weeks after UC-MSC injection) and killed them a week later. The brains were examined by immunohistochemistry using double staining for BrdU and Nestin – a marker of mouse neural progenitor cells. As shown in Figure 6, LPS treatment decreased the number of proliferating neural progenitor cells (BrdU + Nestin+) in the dentate gyrus of the mouse brain, whereas UC-MSC injection increased it. The effects of intravenous and intraperitoneal UC-MSC treatment were similar. SDF-1α significantly improved the effect of intravenously injected UC-MSCs.

Discussion

The data obtained in the first set of experiments indicated that either a single injection of UC-MSCs or regular injections of their conditioned medium begun simultaneously with LPS injection attenuated the increase of pro-inflammatory cytokine IL-6 in the blood and prevented negative effects of inflammation, such as downregulation of α4β2 and α7 nAChRs, accumulation of α7-bound Aβ_1–42 in the brains of mice and episodic memory impairment. The ultrastructural analysis revealed that LPS treatment resulted in destruction of hippocampal mitochondria, and transplantation of UC-MSCs promoted the recovery of mitochondrial structure in the hippocampal CA1 region. This indicates that MSC-based therapy begun immediately after a pathogenic influence prevents the negative consequences of neuroinflammation.

In contrast to a stable and continuous MSC effect, the effect of MSC-conditioned medium declined after 2 weeks, possibly due to an immune reaction in recipient mice. Previously, the authors reported that regular injections of MSC-conditioned medium are almost as efficient as a single injection of MSCs [26]. These data indicated that the MSC effect is mediated by soluble factors, which opened up an attractive opportunity to use MSC-conditioned medium instead of allogeneic or even syngeneic MSCs. The data presented here demonstrate that MSC-conditioned medium is inefficient for therapeutic purposes because it stimulates antibody production in the host. In the authors' experiments, the immune reaction was not directed against protein components of the culture medium (e.g., bovine serum), as MSC-conditioned preparation was obtained in serum-free medium. Therefore, the antibodies were generated against MSC-produced soluble factors injected intraperitoneally. By contrast, MSCs injected intravenously did not stimulate additional antibody production against either their soluble factors or the antigens within the cell lysate.

Previously, the authors reported that intravenously injected MSCs penetrate the brains of LPS-treated mice [26], and therefore their soluble factors are released into the brain microenvironment and do not provoke an immune reaction. Because of the low expression of MHC class I antigens, MSCs themselves do not stimulate an allogeneic response. The relatively high level of antibodies capable of binding the antigens of lysed MSCs found in the authors' experiments was not augmented upon MSC injection and did not prevent the positive MSC effect in LPS-treated mice.

In spite of the presence of specific antibodies, the effects of conditioned medium on IL-6, nAChR subunits and Aβ_1–42 were still observed at the end of the study (day 56). This indicates that the initial, short-term signal from MSC-derived soluble factors is sufficient to prevent an LPS-induced inflammatory reaction, downregulation of nAChRs and accumulation of Aβ_1–42 in the brain but is not sufficient to constantly support episodic memory. The important role in supporting the vitality of brain cells is played by the brain mitochondria. Here the authors showed that MSC-conditioned medium supported mitochondrial integrity and apoptotic resistance less strongly than MSCs and, accordingly, was less efficient in augmenting the episodic memory of LPS-injected mice.

In the second set of experiments, the authors determined the optimal UC-MSC dose (10⁶ per mouse) and showed that it could be decreased by an order (10⁵ per mouse) if MSC injection was followed by application of SDF-1α. The MSC effect was stable for at least 4 months when MSCs were injected 2 weeks after LPS. Similar to the data of the first set of experiments, the positive effect of MSCs on memory was accompanied by upregulation of α4β2 and α7 nAChRs and a decrease in α7-bound Aβ_1–42 in the brain. These data support the authors' previous observation that MSCs not only can prevent but can also cure the negative effect of inflammation in the brain [26]. They also demonstrate the important contribution of SDF-1α to the MSC-mediated effect. In this experiment, the authors showed that intraperitoneal injection of MSCs is similarly (or even more) efficient compared with intravenous injection and that MSCs derived from human adipose tissue (ASCs) are no worse than those obtained from umbilical cord (UC-MSCs). These observations are important for the potential therapeutic application of MSCs in humans because they allow for the attainment of syngeneic cells from adipose tissue and suggest an alternative route of delivery. However, it should be noted that the effect of intraperitoneally injected cells may not be identical to that of intravenously injected cells. In the authors' experiments, UC-MSCs injected intraperitoneally augmented a wider range of nAChR subunits in the brain compared with intravenously injected cells but did not influence IL-10 and TNF-α levels when injected after prolonged LPS treatment.

To further understand the therapeutic potential of MSCs, the authors applied them to mice that were injected with LPS three times over 2 months (days 0, 30 and 60) so that their brains could be expected to be significantly damaged by chronic inflammation. Indeed, these mice demonstrated impaired episodic memory, and their brains contained elevated amounts of pro-inflammatory cytokine TNF-α as well as α7-bound Aβ_1–42. Changes in the nAChR content in the brain were different compared with short-term (acute) LPS treatment. Compared with nontreated mice, α4 subunits were downregulated, whereas β2 subunits were upregulated; α4β2 nAChRs were possibly substituted with other β2-containing nAChR subtypes (e.g., α3β2). Intravenous UC-MSC injection restored episodic memory after 2 weeks, and SDF-1α application accelerated this effect. In spite of low efficiency with regard to brain IL-10 and TNF-α, intraperitoneally injected UC-MSCs were similarly efficient compared with intravenously injected UC-MSCs.

MSCs are able to differentiate into different types of cells, including neurons. However, according to the existing paradigm, the therapeutic effect of MSCs is mainly due to soluble factors (cytokines and growth factors) produced in the microenvironment. The authors' data demonstrating the efficiency of xenogeneic MSCs and their conditioned medium support this paradigm. Previously, the authors showed that intravenously injected human UC-MSCs penetrate the brains of both LPS-treated and α7-/- mice – possibly because their blood–brain barrier is weakened by inflammation – and persist there for quite a long time [26,46]. In the model of α7-/- mice, a positive effect on brain mitochondria was observed until MSCs were present in the brain [46]. Here the authors showed that human UC-MSCs stimulated proliferation of mouse neural progenitors – which can differentiate into glial and neuronal cells – in the dentate gyrus, the brain area closely related to the hippocampus and involved in episodic memory formation. This is consistent with data showing that the dentate gyrus is the main site of neurogenesis in adults [47]. Application of SDF-1α significantly improved this MSC effect. SDF-1α, or CXCL12 [48,49], is known to contribute to cell proliferation and neurogenesis within the brain, in particular by affecting neural progenitor cells [30]. Therefore, application of SDF-1α potentiates the influence of MSCs on neurogenesis and allows for a significant decrease in MSC dose for therapeutic application.

Conclusion

The data described in the present article allow us to conclude that MSC-based therapy prevents the negative consequences of neuroinflammation by decreasing levels of proinflammatory cytokines, stimulating expression of α4β2 and α7 nAChRs in the brain, preventing accumulation of Aβ_1–42 and supporting the apoptogenic resistance of brain mitochondria. The therapeutic effect of MSCs is mediated by soluble factors produced in the microenvironment. However, the use of MSC-conditioned medium instead of the cells is not advantageous because its effect is transient and stimulates an immune reaction. MSCs not only can prevent but can also cure the negative effect of inflammation in the brain when introduced up to 2 months after the advent of chronic inflammation. MSCs of different origins (umbilical cord or adipose tissue) are similarly efficient and can be injected either intravenously or intraperitoneally, although the effects seen with each method may be slightly different. MSCs stimulate proliferation of neural progenitor cells in the brain, and application of SDF-1α potentiates the effect of MSCs on the brain and memory.