Glial cells, especially microglia, oligodendrocytes and astrocytes represent more than 50% of the brain cells and play a crucial role in tissue repair and diseases of the central nervous system. Brain tissue damage leads to acute activation of glia, which initiates protective processes to maintain the integrity of the neuronal network (Sofroniew and Vinters, 2010; Anderson et al., 2016). In disease, brain tissue displays a chronic activation profile of glia, which involves both protective and neurotoxic markers (Heneka et al., 2015; Zuchero and Barres, 2015). Glial cells therefore play a dual role in disease and the functions of activated glia may depend on the stage of the disease (Molofsky et al., 2012; Heneka et al., 2015; Pekny et al., 2016; Priller and Prinz, 2019). In addition, glial risk genes identified in GWAS on Alzheimer’s disease (AD) and Parkinson’s disease (PD) cases indicate that glial dysfunction is a driving factor in the development and progression of CNS diseases (Bohlen et al., 2019; Hammond et al., 2019; Jansen et al., 2019). Although glial cells represent a potential avenue for treatment, their exact function in brain diseases is still unclear, and more information is needed to understand when and how glial cells could be targeted for pharmacological intervention.

Microglia are the primary immune cells of the nervous system. They permanently scan the brain parenchyma for insults and react to changes in the microenvironment by adopting different activation states (Nimmerjahn et al., 2005; Tremblay et al., 2011; Li and Barres, 2018). These are traditionally characterized by their pro-inflammatory (M1) and anti-inflammatory (M2) cytokine profile, which are accompanied by morphological changes (Heneka et al., 2015). This classification is derived from in vitro situations and may insufficiently describe the broad spectrum of activation states in the tissue environment (Jurga et al., 2021). In addition, microglia interact with neurons early in development and regulate the neuronal network by synaptic pruning (Paolicelli et al., 2011; Schafer et al., 2012; Kettenmann et al., 2013). Astrocytes are responsible for a variety of homeostatic and metabolic processes in the brain. Their close connection to the neurons ensures neurotrophic support and regulates synaptic transmission and plasticity (Zuchero and Barres, 2015; Allen and Eroglu, 2017). Similar to the activation states attributed to microglia, different types of reactive astrocytes have been reported in neurodegeneration (A1) and injury (A2) (Liddelow et al., 2017). In a recent publication, the A1/A2 classification of reactive astrocytes was revised and the authors stated that the term “reactive astrocytes” describes multiple states that astrocytes can adopt in response to the tissue environment (Escartin et al., 2021). Oligodendrocytes produce the myelin sheath insulating the axon, which is fundamental for action potential propagation in the CNS. Recently, also trophic neuronal support has been added to oligodendrocytes’ function (Nave and Werner, 2014; Zuchero and Barres, 2015).

In summary, glial activation in disease is by far more complex than the traditional classification of activations states. In this regard, advances in transcriptomic and proteomic techniques have proven to be very valuable because they enable analyses of cell type-specific expression profiles. Over the last years, datasets accumulated mainly from studies using glia derived from mouse models since the source of human glial cells is limited. Recently, comparative studies revealed substantial differences in the transcriptome of rodent and human glial cells, which urges the use of human models in translational research (Zhang et al., 2016; Galatro et al., 2017; Gosselin et al., 2017; Zhou et al., 2020). Further, transcriptomic analyses of glial cells, that were isolated and cultured from post-mortem tissue showed that these cells underwent drastic changes in vitro (Gosselin et al., 2017). This might affect their disease-specific signature and therefore analysis of acutely isolated cells are advantageous for the assessment of their in vivo expression profile.

Recently, immunopanning has been established as a promising method for the prospective purification of human glial cells from fetal or surgically dissected tissue (Zhang et al., 2016). The protocol was adapted from a purification method for rodent astrocytes and involved passing the isolated cell suspension over petri dishes coated with antibodies specific for microglia, oligodendrocytes or astrocytes, respectively (Foo et al., 2011). Here, we show for the first time that glial cells can be successfully isolated from post-mortem human adult brain using an immunopanning-based protocol. Based on the isolation method published by Zhang et al. (2016) we established a protocol to selectively isolate cells from cortical gray matter and prepared samples in parallel for RNAseq and MassSpec. Our analysis confirmed the presence of glia-specific markers in the respective fraction. This method facilitates comparative studies of glial cells derived from healthy, aged and diseased post-mortem human adult brain. Ultimately, a combined transcriptomic and proteomic approach will potentially lead to the identification of new disease markers and contribute to understanding the role of glia in pathology.

To test whether we can selectively isolate glial cells from cortical post-mortem tissue, gray matter tissue from the inferior frontal gyrus dissected during autopsy was enzymatically digested with papain to generate a single cell suspension. Next, we pulled out selected glial cells using immunopanning, which involves a series of incubation steps of the single cell suspension on Petri dishes pre-coated with antibodies. Therefore, we adapted the protocol published by Zhang et al. (2016) and used an anti-CD11b antibody for microglia, an anti-O4 antibody for oligodendrocytes and an anti-HepaCAM antibody for astrocytes for sequential purification (see Graphical Abstract). We expect 2–4 million cells in total isolated from 1 gram of gray matter using a papain based protocol. By extrapolating the cell numbers counted using microscopy pictures of the panning dishes (Supplementary Figure 1), we estimated that panning with CD11b antibody and HepaCAM antibody resulted in cell numbers of at least 500,000 and 50,000 cells, respectively, panning with O4 antibody yielded at least 10,000 cells.

Immunopanned cells were collected for qPCR analysis and samples from three donors (Table 1) were assessed for microglia (IBA1, CD68)-, oligodendrocyte (MBP)- and astrocyte (GFAP, ALDH1L1)-specific markers. Our analysis showed that microglia, oligodendrocytes and astrocytes markers were enriched in the respective sample (Figure 1). In comparison to whole gray matter lysates, microglial markers are highly increased while astrocytes and oligodendrocyte markers are only slightly higher (Supplementary Figure 2).

Next, we isolated cells by immunopanning and lysed side-by-side either in RLT lysis buffer (Qiagen RNA extraction kit) for RNA or UREA buffer for protein analysis. We collected samples from three isolations (Table 1) and analyzed them by RNA sequencing and MassSpec. We found that on both RNA and protein level, the known cell type-specific markers were enriched for microglia (IBA1, CX3CR1, C1QC, ITGAM), oligodendrocytes (MBP, PLP1, BCAS1, MOG) and astrocytes (GFAP, AQP4, ALDH1L1, HepaCAM), respectively (Figures 2, 3). Detection of microglia-specific marker that are not expressed in peripheral macrophages confirmed that we isolated microglia using the CD11b antibody (TMEM119, P2RY12, SALL1, Figure 4). In summary, the analysis showed that we were able to enrich microglia, oligodendrocytes and astrocytes from post-mortem gray matter tissue by our immunopanning protocol.

Analysis of brain tissue is crucial to understand cellular and molecular processes in neurological diseases. Whole brain tissue analysis might mask changes occurring in a specific cell type and therefore the use of isolated glia cells facilitates identification of cell-specific changes in disease. In the last years, emerging number of studies have been published using nuclear single cell sequencing isolated from post-mortem human brain in order to identify differentially expressed genes in disease (Galatro et al., 2017; Habib et al., 2017; Del-Aguila et al., 2019; Mathys et al., 2019; Zhou et al., 2020). In these analyses nuclei from whole brain tissue are isolated and cell types are identified based on the nuclear presence of cell type-specific transcripts. A recent publication showed that analysis of transcripts regarding the activation state of microglia are missed due to low nuclear abundances, emphasizing the need for alternative methods to complement these studies (Thrupp et al., 2020). Additionally, the quality of the RNA obtained from frozen tissue is not as high as from cells isolated from fresh tissue. Furthermore, while single nuclei sequencing has been incredibly informative, a large amount of mRNA is located in the cell bodies as well, and immunopanning allows for the isolation of the whole cell body.

Immunopanning is a selective isolation method that has been established to purify glia cells from rodent and human tissue (Foo et al., 2011; Zhang et al., 2016). Here, we applied immunopanning on post-mortem human gray matter tissue to isolate glia cells for transcriptome and proteome analysis.

We showed that we successfully isolated microglia and astrocytes with our immunopanning protocol. Although microglial numbers are sparse and make up about 5% of the cell population in the gray matter region (Mittelbronn et al., 2001), they can be highly enriched by immunopanning (Supplementary Figure 2). Mizee et al. (2017) reported an isolation method where they obtain 145,000 cells per gram gray matter from occipital cortex using CD11b coated magnetic beads. Here, we used papain digestion and omitted the percoll gradient centrifugation and obtained at least 500,000 cells per gram gray matter from frontal cortex on CD11b coated dishes. Our data confirm the presence of disease-associated microglial markers (e.g., APOE, CD74, CST3) which were detected with single cell sequencing but not with nuclear sequencing in the study by Thrupp et al. (2020). Further, the results of the transcriptome analysis indicate that we did not reach a complete depletion of microglia before incubation with O4 and HepaCAM coated plates (Figures 2, 4) and more optimization is needed to reach higher purity levels. Astrocytes are abundant in the gray matter but the yield of isolated astrocytes is lower in comparison to microglia and this might be due to lower viability of astrocytes in the post-mortem tissue (Sofroniew and Vinters, 2010). While astrocyte cultures from post-mortem brain could be established by a few proliferating astrocytes that form colonies and grow confluent within 2–4 weeks in culture (De Groot et al., 1997, 2001), isolation of post-mitotic mature astrocytes had been challenging in the past. More recently, with the discovery of new astrocyte surface markers, immunopanning and cell sorting protocols were used to isolate astrocytes from human tissue obtained from brain surgeries, but it was still an open question whether these methods can be used for isolation of astrocytes from human post-mortem brain tissue (Sun et al., 2013; Zhang et al., 2016). Here we showed that we were able to immunopan sufficient astrocytes from post-mortem tissue using an antibody against HepaCAM to perform RNA sequencing and proteome analyses.

Our data showed that oligodendrocytes were enriched in the respective sample in comparison to the microglia enriched sample. Oligodendrocytes were also detected in the astrocyte enriched samples, based on markers including surface antigen Claudin-11, which was only observed in the RNA dataset (Supplementary Figure 3). Previously Zhang et al. (2016) showed that their protocol depleted oligodendrocytes from the cell suspension prior to astrocyte panning (Zhang et al., 2016), which was required because HepaCAM is expressed by oligodendrocytes to a moderate extent. In the current study cell numbers were low after enrichment with the O4 antibody and qPCR results showed that enrichment of oligodendrocytes in the oligodendrocyte enriched sample is low in comparison to whole brain tissue (Supplementary Figure 2). We suspect that an inefficient depletion of oligodendrocytes by the O4 antibody, which targets Claudin-11, explains the low yield in the oligodendrocyte enrichment and presence of oligodendrocyte markers in the astrocyte enrichment. We expect overall low numbers of oligodendrocytes in the cell suspension due to their low abundance in the gray matter. For future studies, a screen for new oligodendrocyte surface marker may lead to identification of new antibodies that can be used for oligodendrocyte depletion.

Although immunopanning provides a fast and selective isolation of brain-derived cells from human post-mortem tissue, there are also limitations that apply to this technique. In addition to the expression analyses, we also tested whether we could culture microglia and astrocytes for functional assays after immunopanning these cells from gray matter. We were not able to establish a microglial or astrocyte culture after trypsinization of the immunopanned cells (Zhang et al., 2016) followed by earlier-established culture conditions for human primary glial cells (De Groot et al., 2001; Supplementary Table 1). It is conceivable that cells derived from post-mortem tissue are more vulnerable in comparison to cells isolated from tissue obtained during surgeries. It would be informative to address this in future experiments to compare tissue from both sources in this experimental setting. Performing the isolation directly after autopsy may increase viability but we were not able to shorten post-mortem delay or include tissue from surgeries due to practicalities at our department. Further, we used cortical gray matter tissue for immunopanning in this study while we and others showed in earlier studies that microglial cultures from post-mortem tissue are primarily established from white matter. Mizee et al. (2017) isolated microglia from both gray and white matter using a different isolation method and for both sources they confirmed viability of acutely isolated cells by FACS analyses. In line with other studies, viability was independent of donor age and post-mortem delay (Galatro et al., 2017; Mizee et al., 2017), however, viable cultures could only be established from white matter microglia. This indicates that the use of gray matter from post-mortem tissue for in vitro cultures is limited. So far, we have used the immunopanning protocol for subcortical white matter only in a pilot study, where the inclusion of a myelin removal step was necessary for successful immunopanning. Whether viable cultures from white matter or other brain regions could be established using immunopanning, remains to be investigated. In conclusion, the present immunopanning protocol cannot be applied to establish in vitro cultures from post-mortem gray matter.

Another limitation of this study could be the potential effect of the post-mortem delay and tissue treatment on gene expression and protein levels. Since this is almost unavoidable when working with human tissue, identified markers for glial cells or diseases identified using post-mortem human brain tissue must eventually be validated using in vivo and in vitro models. In addition, in this study we successfully obtained sequencing and proteomics data from three donors. Keeping in mind the general high variation in human donor material it is not possible to perform comparative analyses between controls and AD. Also, the amount of sample obtained using immunopanning could be a limitation. Here, we prepared cell lysates from one immmunopanning plate for RNAseq or Mass Spec analysis. However, for additional validation by Western blot, qPCR or similar methods, more plates could be prepared to reach sufficient amount of RNA or protein for analysis.

In conclusion, we have successfully established an immunopanning protocol which can be used for microglia, oligodendrocyte and astrocyte enrichment for transcriptome and proteome analysis from human post-mortem brain. This method facilitates studies of glial cell-specific profiles depending on age, disease and brain regions. Combined comparative analyses of the transcriptome and proteome will lead to the identification of glia cell-specific markers that can aid in the identification of new potential targets for therapy and diagnosis in neurodegenerative diseases.

The data presented in the study are deposited in the figshare repository, accession number 10.6084/m9.figshare.17032157 and 10.6084/m9.figshare.17032202.

Ethical review and approval was not required for the study on human participants in accordance with the local legislation and institutional requirements. The patients/participants provided their written informed consent to participate in this study.

AN, LF, and JH designed the study. AN and ID performed glial cell purification. LF, CW, DC, JB, EL, SG, and AA performed the proteomics and RNA sequencing analysis. AN designed the figures and wrote the manuscript with input from all authors. All authors were involved in the interpretation of the analysis and approved the final version of the manuscript.

All Roche contributors are full time employees of F. Hoffman-La Roche, Basel, Switzerland. The authors declare that this study received funding from F.Hoffman-La Roche. The funder had the following involvement with the study: study design, data collection, data analyses, and writing of the article.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.