Introduction
Microglial cells are the resident macrophages of the brain and spinal cord. Because the vulnerable neural tissue is largely shielded from leukocyte infiltrates and antibodies, microglial cells are the major and practically the only line of defense in the CNS, where they perform a variety of functions which are carried out by different cell types in the periphery. Microglia provide tissue surveillance within the brain, searching primarily for degenerating neurons, pathogens (Gehrmann et al.
1995), and apoptotic cells (Eyo and Dailey
2013). They are also involved in cytotoxicity to fight neuronal infection by viruses and bacteria, which can induce collateral damage to neurons (Gehrmann et al.
1995). Although microglia are poor antigen-presenting cells during their resting state, upon activation, they can upregulate MHC class I/II proteins and become efficient antigen presenters to T cells infiltrating the brain in certain inflammatory conditions (Aloisi
2001). There is also substantial evidence that microglial cells actively participate in synaptic stripping, tissue repair, and extracellular communication (Harry
2013). Due to their role in the CNS, microglial cells are extremely sensitive to small physiological and pathological changes, in part due to their specialized potassium channels (Dissing-Olesen et al.
2007; Gehrmann et al.
1995).
The activation of microglia required to combat pathogens in the CNS is a double-edged sword. Hyper-activation of microglial cells can directly cause the chronic inflammation observed during neurodegeneration, mainly due to the excessive secretion of cytokines and chemokines (Streit
2006). Many devastating neurodegenerative conditions, including Alzheimer’s disease (Mandrekar-Colucci and Landreth
2010), Parkinson’s disease (Sanchez-Guajardo et al.
2013), and HIV-associated dementia (HAD) (Watkins and Treisman
2015), are strongly correlated with exacerbated microglial activation.
It is generally agreed that HIV-1 replication in the CNS is initiated from invading perivascular macrophages, and then progresses to infection of microglial cells and, to a lesser extent, astrocytes within the brain parenchyma (Churchill et al.
2009; Conant et al.
1994; Eugenin and Berman
2007; Gorry et al.
2003; Hazleton et al.
2010; Watkins and Treisman
2015; Wiley et al.
1986). HIV-1 can infect microglial cells via
CD4 receptors and a variety of chemokine
co-receptors including CCR3, CCR5, and CXCR4, with CCR5 being the most important of the three (He et al.
1997). For example, IL-4 and IL-10 facilitate entry and replication of HIV-1 in microglia through the upregulation of CD4 and CCR5 expression, respectively. HIV-associated dementia (HAD) as well as less severe conditions known as HIV-associated neurocognitive disorders (HAND) strongly correlates with microglia activation, presumably due to combined deleterious effects of viral proteins and cytokines on neurons (Rock et al.
2004; Watkins and Treisman
2015).
Here, we describe a robust method for establishing immortalized microglial cell lines from a wide range of species, including humans. The immortalized cells have microglia-like morphology, express key microglial surface markers, demonstrate appropriate migratory and phagocytic activity, and have the capacity to mount an inflammatory response characteristic of primary microglia. Importantly, these cells can be used to generate a stable cell lines latently infected with HIV proviruses. In a related manuscript, we provide a detailed characterization of the responses of microglial cells to inflammatory activation signals (Alvarez-Carbonell et al.
2016). These extensively characterized cell lines will provide important tools to study microglial cell function and the mechanics and dynamics of HIV transcription in the CNS.
Methods
Isolation of primary microglial cells
Fresh CNS cortical tissue was obtained from adult patients undergoing brain surgery at University Hospitals of Cleveland (Human Tissue Procurement Facility, UH). Fresh mouse brain cortical tissue was obtained from newborn mice (CWRU). For effective manual dissociation of brain tissue, we used the Neural Tissue Dissociation Kit (P) (Miltenyi Biotec), following the manufacturer’s instructions. Dissociated cells (from approximately 1 g from human tissue or 400 mg from mouse tissue) were incubated with CD11b Microbeads (Miltenyi Biotec) for standard magnetic cell sorting. Isolated CD11b+ cells were then resuspended in DMEM:F12/10% FBS medium and cultured for 7 days prior to further treatment. Maturation of the primary cellular culture was monitored by phase-contrast microscopy.
Infection of primary human microglia with HIV and primary macaque microglia with SIV
Primary human microglial cells, cultured in coversli ≤ chamber slides, were infected with replication competent R5 HIV (AD8gNef-GFP). After 4 days, the virus was removed and 10 μM Raltegravir was added for 72 h. The cells were then fixed (4% para-formaldehyde) and permeabilized (0.1% Triton X-100). Cells were then stained with DAPI and Alexa Fluor® 647-conjugated phalloidin (for actin detection; ThermoFisher Scientific), and imaged by Delta vision deconvolution microscopy to detect GFP expression (HIV), nuclei, and actin.
Similarly, primary macaque microglial cells, kindly provided by Dr. Janice Clements (JHU), were thawed and plated onto coverslip chamber slides. After a week in culture, cells were infected with replication competent SIV 17E-Fr particles. In this case, since this virus lacked the GFP-reporter, SIV expression was measured by immunostaining for the SIV p27 gag protein using a SIVmac p27 monoclonal antibody (55-2F12; NIH AIDS Reagents Program) followed by anti-mouse secondary-Alexa Fluor® 488-conjugated antibody (ThermoFisher Scientific).
Immortalization of primary microglial cells
Primary microglial cells isolated from human, macaque or mouse brain, or cryopreserved human microglia (Sciencell, Cat. #1900) were infected with vesicular stomatitis virus G envelop simian virus 40 large T antigen viral particles (VSVG SV40), containing the pBABE-puro SV40 LT construct (Addgene, Plasmid #13970) by spinoculation. Transformed microglial cells were allowed to expand in the presence of 2 μg/mL of puromycin (selection antibiotic), and antibiotic-resistant colonies were selected.
A fraction of the SV40-immortalized cells, as well as CHME-5 microglial cells, (Janabi et al.
1995) were superinfected VSVG hTERT-neomycin viral particles containing the pBABE-neo-hTERT construct (Addgene, Plasmid #1774) in order to express human telomerase reverse transcriptase. Infection was also carried out by spinoculation, and antibiotic-resistant colonies were selected in the presence of 2 μg/mL puromycin and 600 μg/mL neomycin, or 600 μg/mL neomycin in the case of CHME-5-hTERT cells (hT-CHME-5).
Species confirmation using CycT1
DNA from immortalized human microglial cells (hμglia) was isolated using the DNeasy Blood and Tissue Kit (Qiagen), and the CYCT1 gene amplified using the human CYCT1-specific primers Fwd 5′-TCC AGA ACT TCC AGT GTT GC-3′ and Rvs 5′-TGC TTC TGG GAA ATA AAT GC-3′, which yields a 500-Kb product. As a control, DNA from hT-CHME-5 cells was isolated and the CYCT1 gene amplified with the rat CYCT1-specific primers Fwd 5′-ACA GGG AAA CAG TCC ACC AG-3′ and Rvs 5′-TAT GAT TTA TCT GAT AGT-3′, which yield a 400-Kb product. Similarly, CYCT1 from macaque microglial cells was amplified for purified DNA and macaque CYCT1-specific primers Fwd 5′-ACA GGG AAA CAG TCC ACC AG-3′ and Rvs 5′-TAT GAT TTA TCT GAT AGT-3′. In each case, we used the Phusion Flash High-Fidelity (ThermoFisher F548L) polymerase, and the following PCR program: initial denaturing at 98 °C for 10 s, 30 cycles of denaturing at 98 °C for 1 s, annealing at 62 °C for 5 s, and extension at 72 °C for 15 s/Kb of product, and a final extension step at 72 °C for 1 min.
Expression of microglial cell surface markers
Surface expression of microglia specific markers was detected by fluorescence microscopy. Cells were cultured on glass coverslips, fixed, permeabilized, and incubated for 1 h with biotin-anti-human CD11b (BioLegend 301304), or FITC-anti-human CD14 (BD 555397), or anti-P2RY12 (Abcam ab86195) or biotin-anti-human P2RY12 (Bioss bs-12072R) antibodies. Alexa Flour 647 mouse anti-GFAP (BD Pharmingen 560298) was used as negative control. Cells were then washed three times and incubated for 1 h with the secondary antibodies PE Streptavidin (BioLegend 405203) or Alexa Flour® 647 goat anti-rabbit (life technologies A21244) followed by exposure to DAPI-containing washing solution for nuclear staining, and fluorescence exposure for imaging.
Flow cytometry analysis (FACS) was performed using the LSR Fortessa instrument for cell sorting, the FACSDiva software (BD, NJ) for data collection, and the WinList 3D software for data analysis was used to measure surface expression of CD11b, CD14, CD68 (eBioscience 12-0689), CD16 (eBioscience 12-0167), CD32 (eBioscience 17-0329), CD64 (eBioscience 8012-0649), CD163 (eBioscience 12-1639), P2RY12, and TGFβR (Millipore ABF17). In addition, we also measured surface expression of CCR5 (BD 556889) and CD4 (BD 556615) in primary astrocytes as well as clones 1A1 and C20, and monitored the expression of these receptors across four passages (first, second, sixth, and tenth). For FACS analysis, we used 1 × 105 cells resuspended in 1 mL of cold PBS in the presence of 0.5 μg of the antibody for 1 h at 25 °C. Appropriate secondary antibodies were used in the absence of fluorophore-conjugated primary antibody. Cell-antibody complexes were centrifuged, and the pellet resuspended in 300 μL of PBS before FACS analysis.
Microglial cell migration and phagocytosis
Migration of hμglia cells was measured by monitoring cell movement on the culture plate surface during a period of 10 h in time-lapse experiments, using an automated, computer-controlled stage encoder, Leica DMI 6000B scope. Briefly, cells were plated at a density of 2.5 × 104 cells per well in a 24-well plate, and pictures were taken every 30 min on pre-selected fields (8 fields total). Time-lapse movies were produced using MetaMorph® image analysis software (Molecular Devices, Downington, PA). The traveled distances of all the cells within the 8 fields were averaged, and the numbers plotted in a distance vs. time graph.
Similar time-lapse experiments were conducted in the presence of dead neuronal cells obtained after treatment of neurons with 0.05% trypsin, followed by 1–3 min vortex, in order to evaluate the phagocytic capacity of the hμglia cells. Cell death was verified by propidium iodide staining; at least 90% of cells were positive. Brightfield images were produced to count the number of dead neuronal cells present in the field as a function of time. The numbers were plotted in a number of dead neurons vs. time graph.
Cytokine production
A representative line of hμglia cells (clone C20) was untreated or treated with TNF-α (10 pg/mL) for 16 h prior to collection of the supernatant and isolation of total RNA (see below). The supernatants were tested on Quansys Biosciences’s (Logan, UT) Q-Plex Array™ kit (human) for secretion of IL-1α, IL-1β, IL-2, IL-4, IL-5, IL-6, IL-8, IL-10, IL-12p70, IL-13, IL-15, IL-17, IL-23, IFNγ, TNF-α, TNF-β, Eotaxin, GROα, I-309, IP-10, MCP-1, MCP-2, RANTES, TARC, Ang-2, FGF, HGF, PDGF, TIMP-1, TIMP-2, VEGF, CD-163, Fractalkine, GM-CSF, and TGF-β. Absolute values of detected protein secretion were quantified in picograms/milliliter, from which the fold change expression was calculated.
RNA-seq analysis
RNA sequencing (RNA-seq) was used as a tool to profile and confirm the microglia phenotype of clone C20, an immortalized human microglia cell line, in order to further validate our method to develop models of immortalized microglial cells. Total cellular RNA was used for preparation of RNA-seq libraries using Illumina TruSeq stranded Total RNA with Ribo Zero Gold kit, which includes removal of both cytoplasmic and mitochondrial ribosomal RNAs. Sequencing was performed on an Illumina HiSeq 2500 instrument at a depth of ∼35 million or more paired-end, 100-bp-long, strand-specific reads per sample. The resulting reads were quality controlled with fastqc (
http://www.bioinformatics.babraham.ac.uk/projects/fastqc/), and low-quality reads were removed from the library using Trim Galore (
http://www.bioinformatics.babraham.ac.uk/projects/trim_galore/). The reads that passed the quality filter were pseudoaligned to the Gencode V23 human transcriptome, and the reads that mapped to each transcript were quantitated using Kallisto (Bray et al.
2016) with 100 rounds of bootstrapping. Differential expression tests were performed with Sleuth (
http://pachterlab.github.io/sleuth/
) using the Wald test (SRA accession number SRP075430). Differentially expressed protein-coding transcripts showing twofold or more change in expression were used in pathway analysis using the GSEA pathway analysis tool (Subramanian et al.
2005) on the Hallmark (50 pathways represented) and Kegg gene sets subset (186 pathways represented) of the curated gene sets (C2) collections of the mSigDB v5 databases using 1000 permutations. Pathway terms containing less than 2 or more than 500 genes were eliminated from the pathway analysis. Pathways that show enrichment with a
p value <0.05 and the stringent familywise error rate (FWER) of <0.25 were considered significant (Subramanian et al.
2005).
Single cell RNA-seq reads from the study of human brain cells by Darmanis et al. (Darmanis et al.
2015) was downloaded from NCBI SRA (accession no. GSE67835). Each file corresponding to an individual cell was checked for quality, aligned to the transcriptome and quantitated as described above. The alignment was performed both at the level of individual cells and after concatenation of the files to generate read volumes comparable to our bulk RNA-seq to improve the accuracy of analysis in direct comparison tests. RNA-seq reads from the analysis of mouse brain cells by Zhang et al. (Zhang et al.
2012) were downloaded from NCBI SRA (accession number GSE52564) and analyzed as described above. Heatmaps were generated by means of the Hierarchical Clustering Heatmap Python recipes (
http://code.activestate.com/recipes/578175-hierarchical-clustering-heatmap-python/
), using tpm (transcripts per million) values. Protein-coding transcripts that were expressed at 5 tpm or higher in at least one of the cell types under study were included in the analyses in Fig.
8. For the middle and bottom panels of Fig.
8a, genes with over twofold higher expression in human and mouse neurons versus astrocytes and microglia (middle panel) or genes enriched in astrocytes when compared to neurons and microglia (lower panel) were selected as neuron-enriched and astrocyte-enriched genes, respectively. Genes enriched in C20 cells versus human oligodendrocytes and astrocytes (middle panel) or versus human oligodendrocytes and neurons (bottom panel) were selected for comparison and their overlap with neuron- and astrocyte-specific genes were shown as Venn diagrams.
Characterization of immortalized mouse microglial cells
Representative clonal populations of immortalized microglial cells derived from mouse brain (muμglia) were characterized by detection of CD11b and Iba1 protein expression as well as by detection of CD11b, Iba1, P2RY12 and TGFβR mRNA expression, and LPS-mediated inflammatory responses. For immunohistochemistry, cells were plated onto glass coverslips in a 24-well plate at a density of 100,000 cells/well. After 24 h, cells were switched to serum-free media. Following a 10-min fixation in 4% paraformaldehyde (PFA) with calcium and magnesium, cells were permeabilized in PBS pH 7.4 with 0.2% Triton-X and 3% normal donkey serum (NDS) for 1 h. Cells were blocked in PBS pH 7.4 with 3% donkey serum, 0.5% BSA and 0.2% Triton-X. Primary antibodies—rabbit anti-Iba1 (Wako) or rat anti-CD11b (ABD Serotec)—were added at a 1:500 dilution and cells incubated overnight at 4 °C. After washing, Alexa secondary antibodies were added at a 1:500 dilution for 1 h at room temperature. Samples were coverslipped using Prolong Gold and imaged on a Leica DM5000B scope.
The constitutive expression of CD11b (Mm00434455_m1), Iba1 (Mm00479862_g1), P2RY12 (Mm01289605_m1), and TGFβR (Mm03024091_m1) transcripts was measured by qPCR using the expression of each gene from one reference cell line as control. The mRNA expression of the pro-inflammatory markers IL-1β (Mm00434228_m1), IL-6 (Mm00446190_m1), and TNF-α (Mm00443258_m1) upon stimulation with 1 μg/mL of LPS was also detected by qPCR. For this, cells were plated in 6-well plates at a density of 500,000 cells/well and switched to serum-free media after 24 h. Cells were washed with PBS, lysis buffer (Ambion) was added to each well, and cells were removed using a cell scraper. Cells were passed through a 20G syringe ten times. Equal volume of 70% ethanol was added to each sample and the Ambion® RNA purification kit used to isolate RNA. Samples were treated with an on-column Purelink DNase kit (Life Technologies) according to kit instructions. Five hundred micrograms of RNA was converted into cDNA using a High Capacity RNA-to-cDNA kit (Applied Biosystems). Samples were pre-amplified for genes of interest using the Taqman Preamplification kit and gene expression assessed using Taqman Assays on a StepOne Plus. Gene expression levels were measured relative to GAPDH and normalized to the reference cell line to assess relative transcript levels across samples.
Development of HIV latency models in microglia
HIV infection of hμglia was carried out as previously described for CHME-5 (Wires et al.
2012) to obtain clonal populations of hμglia/HIV cells. Briefly, infection by spinoculation was carried out with vesicular stomatitis virus G-(VSVG) pseudotyped HIV particles bearing a fragment of HIV-1
pNL4-3, containing
Tat,
Rev,
Env,
Vpu, and
Nef (some cell lines contain an older HIV construct carrying no
Nef (Pearson et al.
2008) cloned into the pHR′ backbone together with the reporter gene 2dE green fluorescence protein (GFP), as previously shown (Dull et al.
1998; Pearson et al.
2008); (Fig.
11a). The viral particles were produced by the triple transfection of 293 T cells using lipofectamine, and the vector titer was determined as described previously (Kim et al.
2006b). GFP
+ cells were isolated 48 h post-infection by fluorescence-activated cell sorting (FACS), further cultured, expanded, and allowed to enter into a latent state. Evaluation of HIV latency was performed by treatment with indicated doses of TNF-α, IL-1β, or LPS. To keep the levels of HIV basal expression low, cells were maintained in 1% FBS (in DMEM supplemented with 1X normocin).
Acknowledgements
We thank past and present members of the Karn laboratory: Stephanie Milne, Biswajit Das, Uri Mbonye, Kien Nguyen, Meenakshi Shukla, Paul Wille, Amy Graham, Julia Freedman, Julian Wong, Hongxia Mao, and Michael Greenberg for gifts of the materials, help, and useful discussions. We also thank Kamel Khalili (Temple), Brandon Harvey (NIDA), Kurt Hauser (VCU), and Janice Clements (JHU) for the useful discussions and materials. We thank Scott J. Howell (CWRU Visual Sciences) for the help with the time-lapse experiments, the CWRU Genomics Core for the RNA sequencing services, and the CWRU/UH Center for AIDS Research (P30 AI36219) for the flow cytometry services. This work was supported by R01 DA036171 and DP1 DA028869 (JK).