Transplantation of bone marrow derived macrophages reduces markers of neuropathology in an APP/PS1 mouse model

Background We investigated early hallmarks of putative therapeutic effects following systemic transplantation of bone marrow derived macrophages (BM-M) in APP/PS1 transgenic mice. Method BM-M were transplanted into the tail vein and the animals analysed 1 month later. Results BM-M transplantation promoted the reduction of the amyloid beta [37-42] plaque number and size in the cortex and hippocampus of the treated mice, but no change in the more heavily modified pyroglutamate amyloid beta E3 plaques. The number of phenotypically ‘small’ microglia increased in the hippocampus. Astrocyte size decreased overall, indicating a reduction of activated astrocytes. Gene expression of interleukin 6 and 10, interferon-gamma, and prostaglandin E receptor 2 was significantly lower in the hippocampus, while interleukin 10 expression was elevated in the cortex of the treated mice. Conclusions BM-M systemically transplanted, promote a decrease in neuroinflammation and a limited reversion of amyloid pathology. This exploratory study may support the potential of BM-M or microglia-like cell therapy and further illuminates the mechanisms of action associated with such transplants.


Background
Microglia are a key protagonist in the central nervous system (CNS) immune system. They regulate amyloid beta (Aβ) content through phagocytosis, playing a central role in the pathology and progression of Alzheimer's Disease (AD) [1][2][3].
Depending on the surrounding stimuli, microglia may assume either a more pro-inflammatory (M1) or antiinflammatory (M2) state [4], though this might be a fluid continuum [5]. Short-term microglial activation is a natural part of neuroprotection in the brain, contributing to Aβ clearance [6], whereas chronic activation has been associated with promoting neurodegenerative disorders such as AD [7][8][9]. As AD progresses microglia chronic activation becomes detrimental and triggers a progressive damaging cycle to the brain [9].
Therefore, an adjustment of dysfunctional microglia cells via, for example, the replenishment with young functional microglia, can be an effective therapeutic strategy. A few early reports have explored this idea [10][11][12][13], although with different cell types, delivery methods and AD animal models.
The first study was in rats using stereotactic amyloid injections and using brain-derived microglia [10]. Stereotactic inject of cells does induce damage in the brain and leads to inflammation un-related to the disease. The second study used monocytes which are less differentiated compared to our BM-M and known to be phagocytic less active [11]. The third study is close to the aim of our study as bone marrow derived microglia were transplanted, however this study explored the priming with IL-4 to derive M2-like microglia. The cells were not characterized beyond the use of CD206 as an M2 marker and no cell tracking data was provided [12]. In addition, this was again using a rat model in which stereotactic injections were used. Changes observed in this model does not capture the element of aging, focusing solely on the role of amyloid alone as a model for AD.
The aim of this study was on using the most appropriate AD model, cell production that is easily scalable and translatable as well as using well characterized cells.

BM-M differentiation from bone marrow
Femur and tibiae of young C57BL/6 mice were isolated with a scalpel. Each bone was placed in an Eppendorf tube and centrifuged, 400 g for 1 min at 4°C. The isolated pellet of bone marrow of one femur and one tibia of the same mouse was pooled and resuspended with DMEM low glucose (Gibco) with 1% penicillum/streptomycin (Gibco) and 10% FBS (Hyclone). Afterwards, the cells were cultured in 100 mm petri dishes at 37°C, 5%CO 2 and 20%O 2 with medium change after 3 days. After 10 days the nonadherent bone marrow-derived stem cells were used for differentiation into microglia-like cells. In detail, the cell suspension was centrifuged, 200 g for 5 min at room temperature. The pellet was resuspended with a medium mix and placed on a 100 mm petri dish. This medium mix included one-part DMEM (low glucose + 10%FBS) with 20 ng/ml GM-CSF (Peprotec) and one part astrocyteconditioned DMEM medium, obtained from culturing astrocytes over 24 h in this medium. The differentiation took an additional 7 days without medium change by incubation at 37°C, 5%CO 2 and 20%O 2 . All adherent cells were detached using trypsin and used for transplantation.

Flow cytometry of microglia
Briefly, cells were trypsinized and filtered through a 40 μm filter and fixed with 2% (v/v) paraformaldehyde (Pierce, 16% Formaldehyde, Methanol-free) for 15 min at room temperature. Cells were centrifuged at 500 g for 5 min and washed with phosphate buffered saline (PBS). Subsequently a blocking step with 0.5% (v/v) BSA for 30 min at room temperature. Cells for CD68 staining were permeabilized with 0.2% (v/v) Tween20 before blocking. . Cells were washed again and fluorescence was measured using the BD Influx. Dot blot graphs were created using BD FACS™ Software. A gate for total cells was set on the FSC vs. SSC plot. PE was analysed using a 561 nm laser and a BP 585/29 filter. Marker expression analysis of the gated cells was done using the appropriate isotype as negative control.

Microscopy and image processing
Images from the immunostained mice brain slides were captured with a Carl Zeiss AxioScan.Z1 slide scanner microscope, using ZEN blue (2012) software and acquired under the same exposure time settings for all the brains. Scan was performed under the 20x objective, at the wavelengths 545 nm (Zeiss 43 HE filter), 620 nm (Chroma ET49006 filter) or 365 nm (Zeiss 49 HE filter) depending on the fluorophore. From each mouse, 3 brain sections were scanned and analysis was performed in the anterior, middle and posterior cortex, hippocampus and brainstem. ZEN blue (2012) slide scan software was used to visualize the sections images and to minimally adjust the image background. The number and area of Aβ positive plaques, Iba-1 positive and GFAP positive cells were further evaluated and quantified using ImageJ software (tool of the public domain, https://imagej. nih.gov/ij/). Images were initially converted to 8-bit channels, scale was set from pixels to μm, threshold was established independently for each mouse and the region of interest was manually outlined and measured. The analyze particles function on Image J (Aβ plaques size set: 100-infinity μm 2 ; microglia size set: 100-infinity μm 2 ; circularity 0.00-1.00) was used to determine number and area of plaques and microglia cells. Plaque number per mm 2 was calculated using the counted number of plaques divided by the total measured region. Plaque size was calculated by dividing the total area covered of plaques by plaque number [14]. The same number and size calculations were made for subgroups of small, medium and large plaques (100-500 μm 2 , 500-1500 μm 2 and > 1500 μm 2 , respectively) individually for total cortical, hippocampal or brainstem region. Calculations were performed in an unbiased manner by an investigator blind to the treatment conditions of the samples.

RNA isolation and qRT-PCR
RNA was extracted from hippocampus, cortex and brainstem using peqGOLDTriFastTM (PeqLab, 30-2040, Erlangen, Germany) reagent according to the manufacturer's instructions. RNA was treated with DNaseI (Life Technologies, EN0521) to remove genomic DNA contamination. This procedure was followed by cDNA synthesis using Superscript III-reverse transcriptase (Life Technologies, 18,080,085) and Oligo (dT) 18 -Primers (Thermo Scientific, SO132) at 50°C for 1 h. cDNA was used as PCR template in an 1:10 dilution and each sample run in triplicate. Quantitative PCR was performed on StepOnePlus™ Real-Time PCR System (Applied Biosystems) using Express SYBR GreenER qPCR Supermix Universal (Life technologies, 1,178, 401 K), 0.2 μM primer each on the DNA (primers as published by us before) with the following cycle conditions: primary denaturation at 95°C for 3 min at 95°C, 35 cycles with 30s at 95°C, 30s at 60°C and 30s at 72°C followed by fluorescence measurement. Absolute quantification was performed for every single gene with three technical repeats per sample. Serial dilutions of plasmid controls with known molecule concentrations were used as a positive control and to generate standard curves. Expression of target genes were normalized using 36B4 (large ribosomal protein P0, RPLP0) as reference gene.

Statistical analysis
Statistical analysis was performed using the GraphPad Prism 6 (GraphPad Software). Data were tested for normality using the D'Agostino-Pearson omnibus normality test. Values between two groups were compared by the two-tailed unpaired Student's t test for normally distributed data. P values of p < 0.05 (*), p < 0.01 (*) and p < 0.01 (***) were considered significant.

Aβ [37-42] numbers and size
Abeta [37][38][39][40][41][42] covers the bulk of amyloid in AD brains in this mouse model [16] and was used to quantify the changes after BM-M transplantations. Twenty-eight days after administration of BM-M or PBS, mice brains were evaluated for changes in Aβ deposition. The number and size of plaques were quantified in cortex, hippocampus and brainstem individually as these regions are differently loaded with amyloid plaques in this mouse model [17]. We found that transplantation of BM-M resulted in 9% (p < 0.05) reduction of plaque size in the hippocampus only (Fig. 2). Although we could not detect a change in total Aβ [37][38][39][40][41][42] plaque numbers, our data shows that transplantation resulted in a reduction of the number of larger plaques (> 1500 μm 2 ) particularly in the cortex (50%, p < 0.03) and hippocampus (70%, p < 0.02) (Fig. 3). These results suggest that there is an effect mediated by the transplanted BM-M on the Aβ [37][38][39][40][41][42] plaques and that this is more pronounced in the hippocampus and for larger plaques.

Aβ-pE3 numbers and size
To evaluate the ability of the transplanted BM-M to invade the core of amyloid plaques we also quantified one of the modified amyloid forms known to be resistant to proteolysis and often found in the center of plaques -the pyroglutamate-modified Aβ peptide (Aβ-pE3) [18]. Double staining of Aβ [37][38][39][40][41][42] and Aβ-pE3 clearly confirms this composition of amyloid plaque types in our mouse model (Fig. 2f). No differences were found regarding the number or size of Aβ-pE3 plaques, neither in total brain area nor in the different brain regions analyzed (Fig. 4), which indicates that under the current transplantation conditions, BM-M cells do not have an impact on this subgroup of Aβ plaques (Fig. 4f-h).
BM-M transplantation was associated with an increased incidence of small microglia but had no effect on microglia clusters (associated with high microglia activation) (Fig. 5a-e). Our data shows an increase compared to controls in the number of small microglia in the hippocampus only (20%, p < 0.03 - Fig. 5c), which links to the results shown above for Aβ plaque content. Observationally, when co-staining IBA-1 with Aβ [37][38][39][40][41][42] , the association of large microglia clusters with large Aβ plaques was evident, while small soma cells were generally not associated with plaque colocalization ( Fig. 5f-h).

Astrocytes
Using immunohistochemistry, the total area covered by astrocytes showed a decrease in the treated animals compared to controls (23%, p < 0.02) particularly in the brainstem (37%; p < 0.03) (Fig. 6). The decrease indicates a reduction of astrocyte activation which in turn is associated with reduced neuroinflammation [19] .

Discussion
In previous studies, we demonstrated that microglia cells can be reliably generated in vitro [20]. It should be noted that while in this study, the level of CD206+ in BM-M was comparatively low compared to other 'classic' M2 markers, Microglia are generally found to be lower expressing for CD206 compared to blood macrophages [21]. There are three studies that have some similarity [10][11][12][13]. However, these either use an exogenous injection of Aβ as an AD animal model (which is associated with brain regional limitations) or use a different cell type (monocytes) or different age ranges. In the current study setting, we used a transgenic mouse model that shows accumulation of Aβ throughout time in several regions of the brain, moreover we have used BM-M cells and not monocytes, which have a higher resemblance with the microglia cells that are present in the brain as they are more differentiated and primed.
In a previous study, we tracked enhanced green fluorescent protein (eGFP) marked microglia intravenously transplanted to the brains of aged mice 28 days after transplantation [13]. We are therefore confident that the transplanted BM-M reach the site of investigation. In the present study, we have the advantage of obtaining results with cells that are more representative of a future clinical study, i.e., non-eGFP labelled. On the other hand this brings a limitation, since it is not possible to obtain a quantification relationship between the number of cells that reached the brain and the observed effects. What cannot be determined however, is to what extent the observed effects are directly attributable to 'standard' activity by the BM-M themselves or alternatively to secondary immunological reactions associated with the (potentially transient) presence of BM-M. c Percentage of GFAP+ area covered in cortex, hippocampus and brainstem individually, shows a significant decrease in the brain stem area of BM-M treated mice. Bar graphs display the mean ± SEM (error bars) (*p < 0.05). d Percentage of GFAP+ area covered in total brain area shows a significant reduction upon BM-M transplantation

Amyloid clearance
Our data shows that systemically transplanted BM-M are associated with a decrease in Aβ plaque size, especially in the hippocampus. We assume that BM-M transplantation improved the clearance of the diffuse halo of larger plaques: Aβ dense-core plaques are usually surrounded by more soluble diffuse Aβ material [22,23] a neuropathological feature of AD consistently observable in transgenic mouse models such as APP/PS1 [17]. In contrast, Aβ pE3 is more hydrophobic, with higher aggregation propensity and stability [24] and more resistant to degradation [25][26][27] all factors that may explain the

Neuroinflammation
Neuroinflammation is one of the main hallmarks of AD [28]. Factors IL-6, IL-10, IFN-γ and PTGER-2 that are generally observed to be elevated in AD patients [29][30][31] were found to be reduced in the hippocampus of BM-M transplanted mice.
While expression of IL-10 was reduced in the hippocampus, it was elevated in cortex of the treated mice. While we cannot currently account for the observation, it is line with a general finding:
Microglia are particularly prevalent in the hippocampus [32], and regulated by their microenvironment [33]. Thus, the consistently stronger effect we tend to observe may be due to a quicker propagation of anti-inflammatory cytokines produced by the transplanted M2 BM-M.
The one departure from this trend was the reduction of reactive astrocytes following BM-M-transplants, which was significantly more pronounced in the brainstem, not in the hippocampus.

Conclusion
Our results provide further evidence how BM-M transplantation may be impacting hallmarks of AD pathology: the observed reduction in average Aβ [37][38][39][40][41][42] plaque size; and in some neuroinflammation-associated gene expression while coupled with an increase in small IBA-1 + cells numbers may point to a potential therapeutic utility in microglia-focussed regenerative medicine.
While we could mainly demonstrate the observed effects for the hippocampus, larger trials are needed to verify whether these effects are lasting and ubiquitous with respect to AD pathology.