BV-2 Microglial Cells Overexpressing C9orf72 Hexanucleotide Repeat Expansion Produce DPR Proteins and Show Normal Functionality but No RNA Foci

Hexanucleotide repeat expansion (HRE) in the chromosome 9 open-reading frame 72 (C9orf72) gene is the most common genetic cause underpinning frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS). It leads to the accumulation of toxic RNA foci and various dipeptide repeat (DPR) proteins into cells. These C9orf72 HRE-specific hallmarks are abundant in neurons. So far, the role of microglia, the immune cells of the brain, in C9orf72 HRE-associated FTLD/ALS is unclear. In this study, we overexpressed C9orf72 HRE of a pathological length in the BV-2 microglial cell line and used biochemical methods and fluorescence imaging to investigate its effects on their phenotype, viability, and functionality. We found that BV-2 cells expressing the C9orf72 HRE presented strong expression of specific DPR proteins but no sense RNA foci. Transiently increased levels of cytoplasmic TAR DNA-binding protein 43 (TDP-43), slightly altered levels of p62 and lysosome-associated membrane protein (LAMP) 2A, and reduced levels of polyubiquitinylated proteins, but no signs of cell death were detected in HRE overexpressing cells. Overexpression of the C9orf72 HRE did not affect BV-2 cell phagocytic activity or response to an inflammatory stimulus, nor did it shift their RNA profile toward disease-associated microglia. These findings suggest that DPR proteins do not affect microglial cell viability or functionality in BV-2 cells. However, additional studies in other models are required to further elucidate the role of C9orf72 HRE in microglia.


INTRODUCTION
Microglia are resident immune cells in the brain that perform vital functions during brain development, homeostasis, and aging. These include migration, phagocytosis of cell debris, pathogens, or excess or non-functional synapses, as well as sensing environmental stimuli and switching their phenotype and function accordingly (1). Defects in microglial function and chronic changes in their physiology have been associated with a variety of developmental and neurodegenerative diseases, but the exact role of microglia in the pathogenesis of frontotemporal lobar degeneration (FTLD) and amyotrophic lateral sclerosis (ALS) is not known (2)(3)(4)(5).
Here, we have investigated the effects of the C9orf72 HRE on microglial cells by introducing the C9orf72 HRE into mouse BV-2 cells and assessing the presence of the C9orf72 HRE-associated pathological hallmarks and microglial cell functionality. Our results suggest that microglial cells harboring the C9orf72 HRE present specific HRE-associated pathological hallmarks but remain functional.

Fluorescence in situ Hybridization (FISH) and Immunocytochemistry
Cells were fixed with 4% (v/v) formaldehyde (28908; Thermo Scientific) in DPBS for 20 min at RT, washed two times with DPBS, and stored at +4 • C. FISH was performed as described in Chew et al. (44) using fluorescently labeled locked nucleic acids, TYE 563-(CCCCGG) 3 , to detect the sense foci. TYE 563-(CAG) 6 was used as a negative control probe (Exiqon). For immunocytochemistry, cells were permeabilized using 0.1% (v/v) Triton X-100 in DPBS for 10 min at RT on a rocker and washed three times with DPBS. For blocking, cells were incubated in blocking solution (0.015% (v/v) goat immunoglobulin G isotype control (02-6202; Invitrogen) in DPBS) for 30-50 min at RT on a rocker. Primary and secondary antibodies were diluted in blocking solution and incubated with the cells either at RT for 1.5 h or overnight at +4 • C followed by washing three times with DPBS. Immunocytochemistry was performed using the following primary antibodies: anti-poly-GA (1:500, MABN889; EMD Millipore), anti-poly-GP (1:500, ABN455; EMD Millipore), anti-poly-GR (1:500, MABN778; EMD Millipore), anti-p62  Thermal cycler conditions were set as follows: 5 min +95 • C preincubation followed by 45 cycles of 1) 10 s +95 • C denaturation, 2) 15 s annealing, and 3) 20 s +72 • C elongation. Expression values of genes of interest were calculated using the C T method and normalized to expression levels of β-actin. For each time point and experiment, mean expression values of 2R samples were set to 100%. Expression values of 66R samples were normalized to mean expression values of 2R samples for each experiment separately. Amplification specificity was checked using melting curves (LightCycler 480 software 1.5.1.62 SP3; Roche Molecular Systems). All the primer pairs used here gave only one peak in the melting curves.

TDP-43 Translocation Analysis
BV-2 cells were co-transfected with either 2R or 66R plasmids in combination with pLVX plasmid, fixed 24 or 48 h after transfection, and double-stained with anti-poly-GA (and goat anti-mouse Alexa Fluor 568) and anti-TDP-43 (and goat antirabbit Alexa Fluor 647) antibodies as described previously. Confocal microscopy images were acquired from three (24 h) and two (48 h) independent cell culture experiments using laser-scanning microscopy (LSM800; Zeiss). An automated segmentation technique was developed in MATLAB to annotate cell nuclei and cell bodies in images. The images included four channels: AF647 (for goat anti-rabbit Alexa Fluor 647), AF568 (for goat anti-mouse Alexa Fluor 568), ZsGreen1, and DAPI. The ZsGreen1 and DAPI channels marked cell bodies and cell nuclei, respectively. An iterative 2D median filtering, 10 iterations with a window size of 7 × 7 pixels, was applied on the images of ZsGreen1 and DAPI channels to smooth cell bodies and cell nuclei for the segmentation. A preliminary segmentation of cell nuclei was acquired by thresholding the DAPI channel for intensity values above 20, image intensities ranged from 0 to 255, and subsequently, the connected component analysis was applied. Components with an area smaller than 5,000 pixels (21.125 µm 2 ) were discarded. Holes in segmented components were filled using morphological operations. The preliminary segmentation of cell nuclei contained under-segmentation errors, which were eliminated using the watershed transform. The location of markers was determined using the regional minima of the Euclidean distance transform of the preliminary segmentation of cell nuclei. The same procedure was applied to segment cell bodies from the ZsGreen1 channel. The threshold, at which the preliminary segmentation of cell bodies was acquired, was set to 1 because only the background was equal to zero in this channel. Components of an area smaller than 3,000 pixels (12.675 µm 2 ) were discarded, and holes were filled using morphological operations. The individual cell nuclei were assigned as markers for a marker-based watershed segmentation, addressing under-segmentation error in cell bodies. Cells with cell body average intensity value below 6 were eliminated from the final segmentation. This threshold was set based on the average intensity of all cell bodies. In addition, apoptotic cells were eliminated from the segmentation. Apoptotic cells were defined as cells in which the spatial distribution of the intensity values of the cell body is heterogeneous, i.e., a concentration of high-intensity values within the cell body of an area smaller than the nucleus. For each cell, the final segmentation for the area of cell body, the area of a nucleus, the sum intensities of cell body in AF647 channel, and the sum of the intensities of a nucleus in the AF647 channel were quantified. An exemplary picture of the segmentation toolbox is shown in Supplementary Figure 1. The source code is freely available at https://github.com/AndreaBehan/FMIS-software.
For TDP-43 translocation analysis, four stages of intracellular TDP-43 distribution were defined depending on the ratio of nuclear to cytoplasmic TDP-43 signal (see Figure 4): none (TDP-43 signal-if any-is only present in the nucleus, and cytoplasmic signal is below threshold of cytoplasmic background control); mild (cytoplasmic TDP-43 signal is above the threshold of the cytoplasmic background control but below nuclear TDP-43 signal, which is above the threshold of the nuclear background control); moderate (nuclear TDP-43 signal is above the threshold of nuclear background control, and the cytoplasmic TDP-43 signal, which is above the threshold of the cytoplasmic background control signal, is above the nuclear TDP-43 signal); severe (nuclear TDP-43 signal is below the threshold of nuclear background control, and the cytoplasmic TDP-43 signal is above the threshold of the cytoplasmic background control). TDP-43 signal intensities were measured as SUM intensities of AF647 channel within the nucleus or cytoplasm (nuclear signal subtracted from whole cell-signal) as described previously and normalized to the nuclear or cytoplasmic area. Nuclear and cytoplasmic background control values were obtained by calculating AF647 SUM intensity in the nucleus (DAPI positive area) and cytoplasm (ZsGreen1 positive, DAPI negative area) of ZsGreen1-positive 2R or 66R cells immunostained without primary antibody. Cells were excluded from the analysis if located on edges of the images or if cell compartments were not wellrecognized by the program.
p62 and LAMP2A Single-Cell Analysis BV-2 cells were co-transfected with either 2R or 66R plasmids in combination with pLVX plasmid, fixed 24 or 48 h after transfection, and stained with anti-p62 or anti-LAMP2A antibodies followed by goat anti-mouse Alexa Fluor 568. Images were acquired from three (LAMP2A) and two (p62) independent cell culture experiments using laser-scanning confocal microscopy (LSM800; Zeiss). Fluorescence intensities of Alexa Fluor 568 were quantified in ZsGreen1-positive cells using ImageJ (Rasband, W.S., ImageJ, U. S. National Institutes of Health, Bethesda, Maryland, USA, https://imagej.nih.gov/ij/, 1997-2018). SUM intensity values were normalized to cell areas. For each experiment, average area-normalized SUM intensity values of negative control pictures (cells stained without primary antibody) were subtracted from area-normalized SUM intensity values of samples.
Nitrite Assay and TNFα, IL-1β, and IL-6 ELISA BV-2 cells were treated 24 h post-transfection with LPS/IFNγ as described previously. After the 24-h treatment, the culture medium was collected, centrifuged at 10,000 × g for 10 min at +4 • C, and supernatants were stored at −20 • C until analysis. Nitrite levels (as an indication of nitric oxide production) of three technical replicates per each biological replicate were measured via Griess reaction (G-7921; Invitrogen). To calculate the nitrite concentration for each biological replicate, means of their corresponding technical replicates were used. Extracellular tumor necrosis factor alpha (TNFα) (88-7324-22; Invitrogen), interleukin (IL)-1β (88-7013-22; Invitrogen), and IL-6 (88-7064-22; Invitrogen) concentrations were measured using ELISA according to kit instructions. For ELISA, no technical replicates were used. Absorbance was measured using a plate reader (Infinite M200, Tecan Group Ltd.). Data are shown as concentrations obtained from the standard curve. Data were FIGURE 1 | BV-2 cells harboring the C9orf72 hexanucleotide repeat expansion do not form intranuclear sense RNA foci. Cells were transiently co-transfected with 2R or 66R in combination with a plasmid coding for ZsGreen1 (green). Nuclei were stained with DAPI (blue). Sense RNA foci were detected via FISH using TYE 563-(CCCCGG) 3 LNA probes (red) in N2a cells transfected with 66R plasmid (A) but not in N2a cells transfected with 2R plasmid (C). No fluorescent signal was detectable when using TYE 563-(CAG) 6  obtained from three independent experiments, each including three biological replicates.

Phagocytosis Assay
BV-2 cells were transfected as described previously. Six hours before starting the phagocytosis assay, 12,000 cells per well on 96-well plates were plated in BV-2 cell medium. pH-sensitive red fluorescent bioparticles (P35364; Invitrogen) were added to the cells 24 or 48 h after transfection. Cells without particles were used for background subtraction in the red channel. Untransfected cells were used for background subtraction in the green channel (ZsGreen1). Microscopy images were acquired with 10× objective from brightfield, green, and red fluorescent channels (300 ms acquisition time) every 15 min using IncuCyte S3 (Essen BioScience). Three (48 h samples) or four (24 h samples) independent experiments were conducted. Per experiment, four images per well of three technical replicates per biological replicate were acquired. For analysis, the IncuCyte software (v2019B) was used. For the green and red channels, Top-Hat segmentation (radius 100 µm, two GCU/RCU thresholds) and an area filter of 70 µm 2 for the green channel were used.
To estimate the phagocytic capacity of only transfected cells, the size of the red fluorescent area within the green fluorescent area was calculated and normalized to total green fluorescent area for each image. Absolute values were calculated for each biological replicate and pooled from three to four independent experiments.

Cell Viability Assay
To measure cell viability 24 h after transfection, BV-2 cells were scraped, centrifuged at 800 × g for 3 min, resuspended in cell culture medium, and diluted in an equal volume of 0.4% trypan blue (T10282; Invitrogen). The numbers of trypan bluepositive cells and the total number of cells were counted using a hemocytometer (DHC-B01; NanoEnTek). The average number of trypan blue-positive cells was normalized to total cell number from eight technical replicates of two biological replicates per experiment. To measure cell viability 48 h after transfection, the BV-2 cell medium was replaced 2 and 24 h after transfection. Cells were maintained for an additional 24 h at +37 • C and 5% CO 2 .
To take both attached and floating cells (which may represent either activated or dead cells) into account, BV-2 cells were scraped in their conditioned medium, diluted in additional BV-2 medium (in total 3 mL per well), and stained and counted as described previously. To assess whether microglial cells harboring the C9orf72 HRE show RNA foci or DPR proteins, BV-2 cells were transiently transfected with plasmids encoding the GGGGCC-motif 66 times (66R) or 2 times (2R; control). These plasmids have been previously used to overexpress the GGGGCC-motif in vitro in U251 glial fibrillary acidic protein-positive glioma cells (45) or in vivo in mouse brain via intraventricular adeno-associated virus (AAV)-mediated delivery (44). In the latter study, the 66R expression led to the generation of sense RNA foci and expression of different DPR proteins (poly-GA, poly-GP, and poly-GR) in the CNS, neurodegeneration in specific brain areas, as well as behavioral deficits similar to the patients carrying the C9orf72 HRE (44).

Statistical Analyses and Data Presentation
In this study, we used transfection of N2a neuroblastoma cells, a neuronal cell line, as a positive control to detect sense RNA foci. We found that overexpression of the 66R but not 2R in N2a cells led to the formation of sense RNA foci (Figures 1A,C), as detected by FISH and TYE 563-(CCCCGG) 3 probe targeted against the expanded GGGGCC repeats. Notably, FISH with the TYE 563-(CAG) 6 negative control probe showed no signal in 66R-transfected N2a cells, demonstrating the specificity of the TYE 563-(CCCCGG) 3 probe against the expanded GGGGCC repeats ( Figure 1B). In contrast to the N2a cells, however, 66Rtransfected BV-2 cells did not show sense RNA foci 24 or 48 h after transfection using the TYE 563-(CCCCGG) 3 probe (Figures 1D,E). Next, we investigated whether the 66R-overexpressing BV-2 cells display DPR proteins. Immunocytochemistry (ICC) and Western blotting (WB) showed that poly-GA and poly-GP proteins were present in BV-2 cells 24 and 48 h after transfection with the 66R plasmids (Figures 2A,B). No signal was detectable in BV-2 cells transfected with the 2R control plasmid or in 66R-expressing BV-2 cells incubated without the primary antibodies, indicating the specificity of the signal (Supplementary Figures 2, 3). Whereas, the poly-GA immunoreactivity was distributed throughout the whole cell body (cytoplasm and nuclei), the poly-GP signal was restricted to nuclei. No aggregates or inclusions of DPR proteins were detected (Figure 2A). In the WB analysis, the DPR proteinspecific bands were detected at abundant levels at the molecular weights between 25 and 35 kDa in 66R-expressing BV-2 cells ( Figure 2B). No high molecular weight bands specific to 66Rexpressing cells were observed (Supplementary Figure 4). No poly-GR immunoreactivity was detected using WB or ICC (Figures 1A,B). Similar to the BV-2 cells, we have also observed poly-GA and poly-GP, but not poly-GR, DPR proteins in N2a cells transfected with 66R by WB and ICC (unpublished data), confirming that these are the main DPR protein species produced at detectable levels from the 66R construct. These results altogether show that BV-2 cells harboring the C9orf72 repeat expansion produce soluble poly-GA and poly-GP but not poly-GR DPR proteins or sense RNA foci.

BV-2 Cells Expressing the C9orf72 HRE Show a Transient Cytoplasmic Translocation and Increased Levels of TDP-43
TDP-43 is an RNA-binding protein, which shuttles between nucleus and cytosol (46). Hyperphosphorylation of TDP-43 and accumulation of cytoplasmic TDP-43 have been observed in the CNS of patients with FTLD and ALS, including C9orf72 HRE carriers (40,(47)(48)(49). These features have been described in glial cells of C9orf72 HRE carriers to some extent as well, but without further differentiation of the subtype, i.e., astrocytes, oligodendrocytes, or microglia (40,48). Phosphorylation, total levels, and subcellular distribution of TDP-43 were examined in BV-2 cells 24 and 48 h after transfection with 2R or 66R plasmid to assess whether the expression of the C9orf72 HRE leads to TDP-43 pathology in microglial cells (Figures 3, 4). To assess TDP-43 localization in the cytoplasm and nucleus of 2R-and 66R-transfected BV-2 cells, co-transfection of 2R or 66R with a ZsGreen1-encoding plasmid was conducted. According to ICC, 24 h after transfection 94% and 48 h after transfection 88% of cells expressing ZsGreen1 were also positive for poly-GA. No crossreactivity between poly-GA and TDP-43 secondary antibodies was observed (Supplementary Figure 5). To assess the degree of TDP-43 translocation from nucleus to cytoplasm, four different TDP-43 distribution types were categorized depending on the ratio of nuclear to cytoplasmic TDP-43 levels (Figures 4A,B).
No difference in TDP-43 phosphorylation or total TDP-43 levels between 2R-and 66R-expressing BV-2 cells could be detected in either time point (Figure 3;  Supplementary Figure 6). Twenty-four hours after transfection, BV-2 cells expressing the 66R plasmid displayed a significantly decreased number of cells with TDP-43 localized in the nucleus only (no cytoplasmic translocation) and an increased number of cells with mild TDP-43 cytoplasmic translocation as compared with 2R cells (Figure 4C). Forty-eight hours after transfection, both 2R-and 66R-expressing cells showed a similar mild TDP-43 cytoplasmic translocation, and there were no differences between the cells in TDP-43 subcellular localization (Figure 4D). In addition, the total TDP-43 levels in ZsGreen1-positive cells were quantified. Twenty-four hours after transfection, 66Rexpressing cells showed higher TDP-43 levels within the whole cell body as compared with 2R-expressing cells (Figure 4E). Forty-eight hours after transfection, no difference in total TDP-43 levels between 2R-and 66R-expressing cells could be detected ( Figure 4F). Together, these data suggest that a transient cytoplasmic translocation and upregulation of TDP-43 levels takes place in BV-2 cells after introducing the C9orf72 HRE.

Expression of the C9orf72 HRE Reduces Levels of Polyubiquitinylated Proteins
Protein homeostasis is ensured at the cellular level through the function of the ubiquitin-proteasome system (UPS) and chaperone proteins, such as heat shock proteins (Hsps). Hsps facilitate the correct folding of proteins and can direct misfolded proteins to degradation through the UPS or autophagy. Inefficient clearance of misfolded proteins is associated with neurodegeneration (50). To assess whether expression of the C9orf72 HRE leads to changes in the UPS or Hsps in microglial cells, levels of polyubiquitinylated proteins Hsp25 and Hsp70 were examined using WB. Twenty-four and 48 h after transfection, 66R-expressing BV-2 cells showed lower levels of polyubiquitinylated proteins as compared with 2R-expressing cells (Figures 5A,B; Supplementary Figures 7A,B). Hsp25 was not present in the BV-2 cells at detectable levels (data not shown). No difference in levels of Hsp70 could be detected between 2R-and 66R-expressing cells neither 24 nor 48 h after transfection (Figures 5A,B; Supplementary Figures 7C,D). These results suggest that expression of the C9orf72 HRE might affect the UPS function without altering Hsp levels.

Expression of the C9orf72 HRE Does Not Influence Viability of BV-2 Cells
Caspase 3 cleavage from the full-length protein (proenzyme) into the active form and loss of plasma membrane integrity are considered common signs of cell death (51). To elucidate whether BV-2 cells expressing 66R undergo caspase 3-dependent apoptosis or necrosis, caspase 3 cleavage was assessed using WB and plasma membrane integrity loss using trypan blue staining. Twenty-four and 48 h after transfection, BV-2 cells showed no caspase 3 cleavage products ( Figure 6A). Moreover, no difference in plasma membrane integrity (Figures 6B,C) could be detected when comparing 2R-with 66R-expressing BV-2 cells. These findings indicate that BV-2 cells harboring the C9orf72 HRE do not undergo apoptosis or necrosis even though they abundantly express DPR proteins.

Expression of the C9orf72 HRE Slightly Alters p62 and LAMP2A Levels but Not Phagocytosis in BV-2 Cells
Whereas, autophagy is the process through which intracellular components (ranging from proteins to cell organelles) are degraded, phagocytosis refers to the ingestion and digestion of extracellular cell debris and pathogens. Both pathways share a common set of signaling molecules (such as reactive oxygen species, redox equivalents) and proteins. These proteins include the lysosomal lysosome-associated membrane proteins (LAMP)-1 and−2, which fuse with vesicles formed during phagocytosis/autophagy to provide an acidic environment for proteolytic degradation. Whereas, this step occurs later in the cascade, two other key proteins are involved in the induction of autophagy. During the formation of a double-membrane vesicle, also called the phagophore, microtubule-associated protein 1 light chain 3 beta (LC3B) I is lipidated and incorporated into the phagophore membrane. The lipidated form of LC3BI is termed LC3BII, and the conversion of LC3BI to LC3BII marks the initiation of autophagy. The cargo to be degraded is targeted to the phagophore via adaptor molecules (52). Degradation of one of these adaptor proteins, p62, is used as another marker for autophagy. Also, it has been proposed that the C9orf72 proteins are involved in the initial phase of autophagy (53,54).
Because C9orf72 HRE carriers show accumulation of p62 in the brain, which might indicate compromised autophagy (40), we investigated whether the levels of proteins involved in autophagy/phagocytosis are altered upon C9orf72 HRE expression. To assess autophagy in BV-2 cells, p62 levels (normalized to β-actin), LC3BI and LC3BII levels (normalized to β-actin), and LC3BI to LC3BII conversion were measured via WB. Twenty-four and 48 h after transfection, no differences in p62, LC3BI, or LC3BII levels or LC3B conversion could be detected between the 66R-and 2R-expressing BV-2 cells (Figure 7; Supplementary Figure 8). Also, no alterations in C9orf72 (55 kDa) and LAMP-2A levels were detected 24 or 48 h after transfection by WB (Figure 7; Supplementary Figure 8).  To quantify p62 and LAMP2A levels at a single-cell level, co-transfection of 2R or 66R with a ZsGreen1-encoding plasmid was conducted, and p62 and LAMP2A levels were measured in ZsGreen1-positive cells using ICC and confocal microscopy. Whereas, the 66R-transfected cells showed mildly increased levels of p62 at 24 h after transfection, the levels were lower at 48 h after transfection as compared with 2R cells (Figure 8). No difference in the LAMP2A levels between the 2R-and 66R-transfected cells could be detected 24 h after transfection. However, 48 h after transfection, 66R-transfected cells showed slightly higher levels of LAMP2A compared with 2R-transfected cells (Figure 8).
Next, we assessed whether phagocytosis, one of the key functions of microglial cells, is affected upon expression of the C9orf72 HRE. We used co-transfection of 2R or 66R and ZsGreen1 plasmids and live cell imaging to measure phagocytosis in transfected cells (see section Phagocytosis Assay; Figure 9C). Twenty-four and forty eight hour after transfection, no difference in the phagocytic activity of BV-2 cells expressing 66R could be detected as compared with 2R-expressing cells at any time point measured (Figures 9A,B). These results show that expression of the C9orf72 HRE in BV-2 microglial cells does not change their phagocytic capability.

Response to LPS/IFNγ Treatment Is Not Affected by the Expression of the C9orf72 HRE in BV-2 Cells
One of the key functions of microglia is the ability to react to environmental stimuli. In vitro experiments have shown that lipopolysaccharide (LPS) and interferon (IFN) γ cause increased production and secretion of tumor necrosis factor alpha (TNFα), interleukin (IL) 1β and 6, and nitric oxide (NO) by microglia (55)(56)(57). To assess whether the expression of the C9orf72 HRE influences the response of microglia to LPS and IFNγ, BV-2 cells were treated concomitantly with both at 24 h after transfection for 24 h. Extracellular levels of TNFα, IL-6, IL-1β, and NO (measured indirectly as nitrite) were elevated in both 2R-and 66R-expressing BV-2 cell media upon treatment. However, no difference in the levels of NO, TNFα, IL-1β, or IL-6 after the LPS/IFNγ treatment could be detected between 2R-and 66Rtransfected BV-2 cells. In addition, no differences in NO and IL-6 levels in the media of 66R cells compared with and 2R cells could be detected in vehicle-treated cells (Figures 10A-D). These results indicate that expression of the C9orf72 HRE does not lead to activation of BV-2 cells nor does it impair their ability to respond to LPS and IFNγ treatment and subsequent production of inflammatory modulators.

Expression of the C9orf72 HRE Does Not Trigger Disease-Associated Microglia RNA Signature in BV-2 Cells
Previous studies have described a subgroup of microglia termed disease-associated microglia (DAM) based on their characteristic RNA profile in the brain of, e.g., Alzheimer's disease model mice and patients. DAM derive from homeostatic microglia through two stages. The first stage is considered independent of triggering receptor expressed on myeloid cells 2 (Trem2), whereas the second stage is Trem2 dependent. Each stage is characterized by a specific mRNA profile, where homeostatic microglial genes are downregulated, but genes involved in phagocytosis, lipid metabolism, and lysosomal pathways are upregulated. Whereas, some subgroups of DAM are considered functionally compromised and eventually neurotoxic, others might be even neuroprotective (58)(59)(60)(61)(62).
To assess whether expression of the C9orf72 HRE shifts BV-2 cells toward the DAM phenotype, mRNA expression levels of four DAM markers were examined. Whereas, TYRO protein tyrosine kinase binding protein (Tyrobp) expression is known to be elevated during stage 1, expression of Trem2, C-type lectin domain family 7, member a (Clec7a), and cystatin F (leukocystatin; Cst7) is increased during stage 2 (59). We found that 24 h after transfection, no changes in the mRNA profiles of these markers between 2R-and 66R-expressing BV-2 cells could be detected (Figures 9A-D). However, 48 h after transfection, 66Rexpressing BV-2 cells showed slightly decreased expression of Cst7 (Figure 11B), but no changes in the expression of Trem2, Tyrobp, and Clec7a compared with 2R-expressing cells could be detected (Figures 11A,C,D). These data suggest that the expression of the C9orf72 HRE does not switch BV-2 cells to the DAM phenotype.

DISCUSSION
To date, most of the mechanistic studies related to the C9orf72 HRE have concentrated on neurons, but there is no information on how the C9orf72 HRE influences the phenotype or function of microglia. In this study, we aimed to address these knowledge gaps by investigating the effects of the C9orf72 HRE on mouse BV-2 microglial cells. We tested whether the microglial cells expressing 66 GGGGCC repeats (66R) (44) show C9orf72 HREderived pathological hallmarks and cellular downstream effects similarly to as previously reported in neurons, and whether the HRE affects microglial-specific functions.
In human post-mortem brains and brains of the 66Rexpressing mice, sense and/or antisense RNA foci and DPR protein inclusions have been detected in neurons (11,13,16,21,33,44). Here, we observed that BV-2 cells expressing the C9orf72 HRE did not present detectable sense RNA foci, even though the expression of the HRE in neuronal N2a cells led to abundant formation of sense RNA foci. The RNA foci have been shown to sequester different RBPs in neurons, and this has been suggested to lead to deleterious effects as the sequestered RBPs can no longer mediate their normal functions (18,23). The underlying reason for the absence of sense RNA foci in BV-2 cells harboring the C9orf72 HRE is not clear but might be related to a potentially differential expression of RBPs in neuronal and microglial cells. Alternatively, microglial cells may be more effective in clearing the aggregated RNAs containing the expanded repeats or sequestered RBPs from the RNA foci compared with neuronal cells. Future investigations will confirm whether, e.g., human microglia show a similar phenotype or whether the lack of sense RNA foci is a specific feature of BV-2 cells expressing the C9orf72 HRE.
Despite the lack of sense RNA foci, poly-GA and poly-GP DPR proteins were detectable at high levels already 24 h after introducing the C9orf72 HRE to BV-2 cells. Neither poly-GA nor poly-GP formed cytoplasmic or nuclear inclusions or high molecular weight species, suggesting that they are present in corresponding quantifications (C) from four independent cell culture experiments. Protein samples were harvested from BV-2 cells expressing the 2R or 66R plasmid for 24 and 48 h. P62, LAMP2A, LC3BI, LC3BII, and C9orf72 levels were normalized to those of β-actin and additionally LC3BII levels to those of LC3BI. 2R was set as 100% for each time point and experiment. Data are shown as median ± interquartile range (LC3BII/LC3BI, 48 h) or as mean ± SD. To compare LC3B conversion between 2R and 66R 48 h after transfection, Mann-Whitney U-test was used. Otherwise, two-tailed independent samples t-test was used; n = 11, df = 20 (24 h); n = 12, df = 22 (48 h)-for C9orf72 data, no equal variances were assumed, df = 11.355 (24 h), df = 16.021 (48 h). df, degrees of freedom; kDa, kilodalton; LAMP2A, lysosome-associated membrane protein 2 a; LC3B, microtubule-associated protein 1 light chain 3 beta.  soluble forms in BV-2 cells. Notably, the same 66R construct as used in the present study to deliver the C9orf72 HRE led to intracellular poly-GA and poly-GP inclusions as well as soluble poly-GP when expressed in mouse neurons in vivo (44). In secondary cell lines, such as human embryonic kidney (HEK)293T cells and U251 cells, the here used 66R plasmid led to DPR protein-specific signals at or below 20 kDa (45). Because the expression of individual DPR proteins in HEK293T cells results in poly-GA and poly-GP inclusions (63), secondary cell linespecific metabolic differences or disparate length of the expanded repeat expressed might underlie the absence of inclusions in our model.
Cytoplasmic poly-GA proteins have been shown to impair the translocation of TDP-43 from the cytoplasm into the nucleus (64). Hence, C9orf72 HRE-associated TDP-43 pathology might at least partially result from the presence of DPR proteins. TDP-43 pathology has been reported in vivo in neurons expressing 66R (44). Here, we detected a transient increase in the ratio of cytoplasmic to nuclear TDP-43 in BV-2 cells expressing C9orf72 HRE. This transient increase coincided with increased total TDP-43 levels on the single cell level. However, there were no differences in TDP-43 phosphorylation in BV-2 cells harboring the C9orf72 HRE compared with the control cells. Furthermore, we did not detect intracellular inclusions or high molecular weight forms of aggregated TDP-43 in BV-2 cells upon the expression of C9orf72 HRE. The lack of a strong TDP-43 translocation phenotype in this study might be due to the short expression time of the HRE. It has been shown in some C9orf72 HRE carriers and animal models that DPR protein pathology might also occur without TDP-43 pathology. This might either be due to the fact that DPR protein pathology precedes the TDP-43 pathology or that another independent trigger, such as cellular stress, is required for the TDP-43 pathology to develop (65)(66)(67). Indeed, some previous studies have suggested that the formation of phosphorylated TDP-43 inclusions is preceded by the presence of RNA foci and DPR protein inclusions (67)(68)(69). Thus, the lack of phosphorylated TDP-43 inclusions in our study might be due to the fact that the BV-2 cells did not harbor RNA Induced pluripotent stem cell-derived motor and cortical neurons of C9orf72 HRE carriers show increased caspase 3 activation and loss of plasma membrane integrity (70). However, fibroblasts from C9orf72 HRE carriers did not show caspase 3 activation or plasma membrane integrity loss when cultured in vitro (71). On the other hand, expression of poly-GA proteins (50× poly-GA) by transient transfection in HEK293T cells or virus-mediated transduction in primary mouse neurons caused caspase 3 activation as well as plasma membrane integrity loss (33). In our model, BV-2 cells did not show signs of cell death 24 or 48 h after transfection with the C9orf72 HREcontaining plasmid, even though DPR proteins (including poly-GA proteins) were already expressed at these time points. This indicates that the susceptibility to C9orf72 HRE-derived toxicity might be cell type dependent. In addition, the toxicity of the DPR proteins might be associated with their high molecular weight, aggregated species, which were not detected in the present study in BV-2 cells.
DPR proteins, such as poly-GA, have previously been shown to inhibit proteasomal activity (33,72,73). Here, lower levels of polyubiquitinylated proteins were observed in BV-2 cells expressing the C9orf72 HRE. However, further studies are warranted on the underlying mechanisms of this finding to clarify whether it results from inhibition or enhanced activation of the UPS. Also, it needs to be further elucidated whether this finding in C9orf72 HRE-expressing BV-2 cells represents a protective mechanism to prevent the TDP-43 and DPR proteins from aggregating. The function of molecular chaperones, such as Hsps, is also intimately linked to the regulation of cellular proteostasis. Members of the HSP70 family have been found to be transcriptionally upregulated in the frontal cortex and cerebellum of C9orf72 HRE-ALS and C9orf72 HRE-ALS/FTLD cases as compared with sporadic and control subjects (74,75). In our study, no changes in Hsp70 levels could be detected, suggesting that the observed upregulation in the human CNS might represent alterations taking place in the later rather than the acute stages of disease. Altogether, our results imply that expression of the C9orf72 HRE may influence cellular proteostasis through affecting the function of the UPS but not the levels of the Hsps in BV-2 microglial cells.
In previous studies, microglia of C9orf72 HRE-cases have shown enlarged lysosomes as compared with sporadic ALS cases in post-mortem spinal cord at autopsy (75,76). This suggests that the lysosomal pathway of microglia might be affected by the C9orf72 HRE. Here, analyses at single cell level indicated that expression of the C9orf72 HRE caused a slight increase in the LAMP2A levels at 48 h after the transfection. Also, transiently elevated p62 levels at 24 h were detected but these decreased at the 48-h time point. No changes in the levels of these proteins were detected in the WB analyses. These changes, together with the other present data, do not suggest major alterations in the autophagosomal and lysosomal pathways and they were not associated with functional impairment of the BV-2 microglial cells, e.g., in terms of their phagocytic capacity.
In a previous study, the expression of poly-GA in HeLa cells was reported to reduce the translocation of p65, a transcription factor and a subunit of nuclear factor kappa-light-chainenhancer of activated B cells (NFκB), from the cytoplasm to the nucleus upon TNFα stimulation. Furthermore, it prevented the shuttling of cytoplasmic TDP-43 into the nucleus, indicating that poly-GA impairs nucleocytoplasmic transport mechanisms (64). Here, expression of the C9orf72 HRE did not alter the response of the BV-2 cells to LPS/IFNγ, as demonstrated by similar levels of NO, IL-6, IL-1β, and TNFα in C9orf72 HRE-expressing and control cells, suggesting unaltered NFκB signaling and nucleocytoplasmic transport. Furthermore, these results indicate that expression of the C9orf72 HRE does not lead to impaired activation of the BV-2 cells in response to inflammatory stimuli.
Finally, we hypothesized that the expression of C9orf72 HRE might lead to a DAM phenotype in BV-2 cells. DAM derive from homeostatic microglia via two stages, the first stage being independent of and the second dependent on Trem2, and each stage is characterized by a specific mRNA profile (58)(59)(60)(61)(62). In a previous study, single-cell RNA sequencing from the spinal cords of mSOD1 (G93A) ALS model mice revealed the presence of DAM (59). However, it is not known whether DAM occur and play a role in C9orf72 HRE-associated ALS or FTLD. In our BV-2 cell model expressing C9orf72 HRE, there were no alterations in the transcript levels of specific DAM marker genes.
We only detected a mild decrease in Cst7 mRNA levels in C9orf72 HRE-expressing cells 48 h after transfection. A strong downregulation of Cst7 expression has previously been reported to lead to increased phagocytic activity of mouse microglia (77). Thus, our finding of only slightly decreased Cst7 expression is in line with the unaltered phagocytic activity in the BV-2 cells expressing the C9orf72 HRE. Taken together, acute expression of the C9orf72 HRE does not appear to lead to dramatic changes in the phenotype or functionality of BV-2 cells.
One limitation of this study is the use of transient transfection. Because of this, it has to be taken into consideration that the untransfected cells might affect the outcome on protein and RNA levels assessed from the total cell lysates. Even though changes in polyubiquitinylated proteins were detectable via WB, moderate changes in TDP-43, p62, and LAMP2A could be observed only using microscopy. This suggests that also other mild changes might be detectable only by single cell-based analyses, especially in model systems using transient transfection. However, so far, no C9orf72 HRE-expressing stable microglial cell lines are commercially available, and subcloning of the HRE into another plasmid, which encodes a suitable selection marker, has proven to be challenging. Because human microglia of post-mortem CNS of C9orf72 HRE carriers present sense and antisense RNA foci to a much lower extent compared with neurons (17), it could be considered that microglia harbor the C9orf72 HRE or present HRE-derived pathological hallmarks less frequently than neurons. Indeed, only approximately 13% of microglia showed RNA foci in post-mortem brains of C9orf72 expansion carriers with FTLD (17). Therefore, our transient in vitro model with cells exhibiting and not exhibiting the HRE resembles examination of whole brain lysates in bulk. Despite the lack of sense RNA foci, poly-GA and poly-GP DPRs were expressed at high levels in our BV-2 cell model. Interestingly, it has been demonstrated that the DPRs may be secreted and taken up by other cells, indicating cell-to-cell spreading of the DPRs (78,79). Moreover, a recent study showed that cell-to-cell transmission of poly-GA inhibited proteasome function and led to TDP-43 accumulation in neighboring neurons (73). Therefore, the overexpression of C9orf72 HRE in transiently transfected BV-2 cells might in fact influence also the adjacent untransfected cells. Another potential limitation of the experiments in the present study is that the BV-2 cells expressing the C9orf72 HRE were examined in isolation as single cell type cultures. Microglia are known to surveil their surroundings and readily respond to environmental cues. Moreover, it has been shown that microglia isolated from the brain alter their gene expression and thus may present with an altered phenotype as compared with the microglia in their in vivo brain environment (80). For example, in human post-mortem brains of C9orf72 HRE carriers, microglia show an activated phenotype and enlarged lysosomes (47,76,81). Thus, the lack of a strong functional change in our study might derive from the lack of interactions with other cell types, including neurons. Therefore, in our upcoming studies, we aim to investigate the BV-2 cells harboring the C9orf72 HRE in co-cultures with neurons and assess whether the neuronal environment alters any of the outcomes investigated in the present study. These studies will also allow investigation of potential non-cell autonomous effects of the microglia. Finally, in general, mouse microglia, especially secondary cell lines, may not fully recapitulate the features of human microglia and, therefore, future investigations in human microglia models are as well needed.
Nevertheless, to our knowledge, we are the first to introduce the C9orf72 HRE into microglia-like cells and characterize the C9orf72 HRE-derived pathological hallmarks and their effects on microglial phenotype and functionality. Our data in this acute microglial cell model of C9orf72 HRE suggest that the C9orf72 HRE-associated gain-of-toxic-function mechanisms do not affect microglial phenotype or functionality in terms of cell death, phagocytic activity, and inflammatory response despite moderate alterations in the levels of specific proteins involved in these microglial functions. Moreover, acute expression of the C9orf72 HRE in BV-2 cells does not induce their phenotypic change to DAM-like cells. In the future, induced pluripotent stem cell-or blood monocyte-derived microglial cells from carriers of the C9orf72 HRE may provide useful model systems to investigate the effects of the C9orf72 HRE on human microglial cell physiology. In these models, both C9orf72 HRE-associated loss-of-function (C9orf72 haploinsufficiency) and gain-oftoxic-function mechanisms can be studied concomitantly. As mounting evidence on altered innate immune system and potential glial cell involvement in different neurodegenerative diseases are emerging, it is important to provide insights into their contribution also in the context of C9orf72 HRE-associated ALS and FTLD pathogenesis, a topic that is currently not well-understood. These investigations may reveal potential new biomarker or therapeutic candidate targets for further translation into clinical settings in the future.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.  GenomMed. This publication is part of a project that has received funding from the European Union's Horizon 2020 research and innovation program under the Marie Skłodowska-Curie grant agreement no. 740264. We wish to thank Dr. Leonard Petrucelli (Mayo Clinic, Jacksonville, FL, USA) for generously sharing the 2R and 66R plasmids. We also thank Dr. Tania Gendron (Mayo Clinic, Jacksonville, FL, USA) for kindly providing the protocol and advice on FISH. UEF Cell and Tissue Imaging Unit is acknowledged for providing IncuCyte S3 and LSM800 training and facilities, and especially we wish to thank Dr. Janne Capra who helped with IncuCyte S3 experiments and image analysis. Finally, we wish to thank Prof. Kai Kaarniranta (Department of Ophthalmology, Institute of Clinical Medicine, University of Eastern Finland, Kuopio, Finland) for kindly providing us with the Hsp25, Hsp70, and p62 antibodies. This study is part of the research activities of the Finnish FTD Research Network (FinFTD).