Quantifications of CSF Apoptotic Bodies Do Not Provide Clinical Value in Multiple Sclerosis

Multiple sclerosis (MS) is an inflammatory disease of the central nervous system (CNS) that leads to the death of neurons and oligodendrocytes, which cannot be measured in living subjects. Physiological cellular death, otherwise known as apoptosis, progresses through a series of stages which culminates in the discharge of cellular contents into vesicles known as apoptotic bodies (ABs) or apoptosomes. These ABs can be detected in bodily fluids as Annexin-V-positive vesicles of 0.5–4.0 μm in size. In addition, the origin of these ABs might be detected by staining for cell-specific surface markers. Thus, we investigated whether quantifications of the total and CNS cell-specific ABs in the cerebrospinal fluid (CSF) of patients provided any clinical value in MS. Extracellular vesicles, from CSF of 64 prospectively-acquired subjects, were collected in a blinded fashion using ultra-centrifugation. ABs were detected by flow cytometry using bead-enabled size-gating and Annexin-V-staining. The origin of these ABs was further classified by staining the vesicles for cell-specific surface markers. Upon unblinding, we evaluated the differences between diagnostic categories and correlations with clinical measures. There were no statistically significant differences in the numbers of total or any cell-specific ABs across different disease diagnostic subgroups and no significant correlations with any of the tested clinical measures of CNS tissue destruction, disability, MS activity, and severity (i.e., rates of disability accumulation). Overlap of cell surface markers suggests inability to reliably determine origin of ABs using antibody-based flow cytometry. These negative data suggest that CNS cells in MS either die by non-apoptotic mechanisms or die in frequencies indistinguishable by current assays from apoptosis of other cells, such as immune cells performing immunosurveillance in healthy conditions.


INTRODUCTION
Multiple sclerosis (MS) is a chronic immune-mediated disease of the central nervous system (CNS), leading to the demyelination of axons and neurodegeneration. Alongside traumatic brain injury, it is the most frequent cause of neurological disability in young adults (1). Findings from prior studies led to the hypothesis that MS can be largely divided into two stages, starting with the inflammatory phase in the periphery and later entering into the neurodegenerative phase (2). Although great progress has been made in understanding the inflammatory components of the disease, the neurodegenerative components are still obscure.
Currently, there are two main ways to measure neurodegenerative process in living human subjects: structural imaging and measurement of neurofilament light chain protein (NFL) (3,4). Structural (MRI) imaging identifies CNS tissue destruction as brain/spinal cord atrophy. However, MRI imaging fails to provide cellular or molecular information and therefore cannot reliably measure the loss of crucial cell types such as neurons and oligodendrocytes, especially when loss of these CNS cells may be masked by infiltration of CNS tissue by immune cells or by compensatory astrogliosis. With emergence of ultra-sensitive Single Molecule Analysis (Simoa) NFL assay (5,6) blood (serum or plasma) NFL levels can be measured both in healthy subjects and MS patients. In MS NFL levels are increased during MS activity and they can weakly predict subsequent MS progression on a group, but not individual levels (4,(7)(8)(9)(10)(11). Thus, there is still a need to develop biomarkers of neuronal and oligodendroglial injury/loss that can be applied on a patient level.
Pathology studies have shown the presence of apoptotic cells, mainly oligodendrocytes and neurons, at the site of MS lesions (2, 12,13), suggesting that CNS cell apoptosis plays an important role in the irreversible neurological disability during the progressive stage of MS. This hypothesis was further supported by animal modeling with antiapoptotic protein Bcell lymphoma-2 (Bcl-2)-overexpressing transgenic mice; in comparison with wild type (WT) these transgenic mice showed reduced experimental autoimmune encephalomyelitis (EAE)severity, despite similar inflammatory response (14).
Apoptotic cells progress through a series of stages including chromatin condensation, DNA fragmentation, membrane blebbing, and cell shrinkage, which all culminate in the discharge of cellular contents into extracellular vesicles, known as apoptotic bodies (ABs) or apoptosomes (15,16). Previous studies have tried to isolate and identify ABs from subject's body fluid and use them as markers of respective disease-related degenerative processes (17)(18)(19)(20). Thus, the goal of the current study was to identify the presence of apoptotic cells in the CNS of living subjects by measuring the total and cell-specific ABs in cerebrospinal fluid (CSF) of patients. Additionally, we asked whether densities of total or cellspecific ABs differentiate MS from healthy donors (HDs), and within MS patient cohorts correlate with clinical measures of CNS tissue destruction, disability, MS activity and severity (i.e., rates of disability accumulation).

CSF Collection and Processing
All subjects were recruited under IRB-approved protocols (Comprehensive Multimodal Analysis of Neuroimmunological Diseases of the Central Nervous System, ClinicalTrials.gov Identifier: NCT00794352; and Evaluation and Follow-up of Patients with Cryptococcosis, ClinicalTrials.gov Identifier: NCT00001352) and all patients provided written informed consent. CSF from subjects were collected per standardized operating procedures (21). CSF aliquots were prospectively labeled using alphanumeric code, stored on ice until further use and analyzed in a blinded fashion.

Isolation of Apoptotic Bodies
ABs were isolated from culture supernatants and CSF as previously described (19,(22)(23)(24). Briefly, cells were isolated and removed by pelleting at 335 g for 10 min. To remove cell-debris, cell-free supernatants were centrifuged at 1,000 g for 10 min; followed by another centrifugation at 2,000 g for 30 min to pellet ABs. Pelleted ABs were resuspended and washed with PBS.

Flow Cytometry
ABs were stained with Annexin V-FITC, and cells were stained with Annexin V-FITC and Propidium Iodide (TACS R Annexin V Kit; Trevigen Inc., Gaithersburg, MD) as per manufacturer's instructions. CSF ABs were also stained for CNS cell-specific surface markers to identify their origin: We used cell-surface markers previously employed in isolation of human CNS cells from brain specimens using immune-panning, and validated by cell-specific RNA profiles (25,26) marker; Anti-Galactocerebroside Antibody; EMD Millipore, Burlington, MA; Clone # mGalC; tagged with PerCP-Cy5.5; Novus Biologicals), CD31 (Endothelial cell surface marker (27); Human CD31/PECAM-1 PE-conjugated Antibody; R&D Systems; Clone # 9G11), and CD14 (Myeloid lineage cell surface marker (28); Alexa Fluor R 700 anti-human CD14 Antibody; BioLegend, San Diego, CA; Clone # HCD14). Briefly, after wash with PBS, pelleted ABs were resuspended in Annexin V-FITC and fluorescence-tagged antibodies against cell-specific surface markers in Annexin V-binding buffer (provided with TACS R Annexin V Kit) and incubated in dark for 15 min at room temperature. Stained ABs were washed with binding buffer and then analyzed using fluorescence-activated flow cytometer (BD LSR II Flow Cytometer, BD Biosciences, San Jose, CA). Gating on ABs included size gate [1-4 µm (29)], using Flow Cytometry Size Calibration Kit (ThermoFisher Scientific, Grand Island, NY). The vesicles in 1-4 µm size gate were further analyzed for Annexin V and cell-specific surface markers' staining.

Statistical Analyses
ABs data for subjects' CSF samples were acquired with the operator blinded to subjects' clinical diagnoses. After data acquisition for all subjects, ABs per ml of CSF were compared across disease diagnostic subgroups (HD, NIND, OIND, RR-MS, and P-MS) using one-way ANOVA; as we have acquired CSF samples from only two CIS subjects, they were not included in analyses. Also, within MS subjects, ABs per ml of CSF were compared across disease activity (active vs. non-active

In vitro Model Validation
We validated our "identification and assessment of Abs" model using human neuronal cell line (SK-N-SH) cultures. As a positive control for induction of apoptosis we used Staurosporine treatment (0.5 µM, 24 h). Apoptotic cells were identified by staining with Annexin V and PI and were analyzed using flow cytometry. According to manufacturer's (TACS R Annexin V Kit) instructions both Annexin V and PI-negative cells are live, only Annexin V-positive cells are early-apoptotic, both Annexin Vand PI-positive cells are late-apoptotic and only PI-positive cells are necrotic (Figure 1A). After Staurosporine treatment, the % of apoptotic cells was significantly elevated (Figure 1B). Quantifying the induction of apoptosis by Staurosporine in our culture conditions, we next sought to quantify ABs in cell culture supernatants in order to demonstrate that our assay could differentiate between the release of ABs from control and Staurosporine-treated cultures. To this end, size gates [1-4 µm, an average size of ABs (29)] were applied in combination with Annexin V staining. First, 1-4 µm size gates were set using 1, 4, and 6 µm beads (Figure 2A; Flow Cytometry Size Calibration Kit, ThermoFisher Scientific). Within 1-4 µm vesicles ABs were identified as Annexin V positive (Figure 2B). Total ABs were quantified (1-4 µm and Annexin V-positive events); after Staurosporine treatment the total number of ABs in cell culture supernatants were significantly elevated ( Figure 2C).

Analyses of CSF Apoptotic Bodies
Verifying the flow-cytometry-based ABs detection in cell culture supernatants, we next applied the same assay to prospectivelyacquired CSF samples. As described in the methods, extracellular vesicles were collected using the differential centrifugation approach. 1-4 µm vesicles were selected using size gate set by known sized beads ( Figure 3A). From these vesicles total ABs were detected using Annexin V staining (1-4 µm and Annexin V-positive events; Figure 3B).
While total number of ABs in the CSF may be clinically useful, we considered the possibility that apoptosis occurs in the intrathecal compartment also under physiological conditions: e.g., any activated immune cell may cross the blood brain barrier as a part of active immunosurveillance mechanism. Such immune cell (e.g., T lymphocyte), if not re-activated in the CNS compartment, may either re-cycle back to blood via lymphatic system, or undergo apoptosis as part of physiological termination of the immune response (35,36). Thus, we envisioned that the assay that could identify the cellular origin of ABs may have significantly higher clinical utility. To this end, we employed fluorescently-tagged antibodies specific for surface markers of different CNS cells. Selected surface markers/antibodies were previously validated as CNS cell-types-specific, because they were used to isolate specific CNS cells (i.e., neurons, oligodendrocytes, astrocytes, microglia, and endothelial cells) from human brain samples via immunopanning. Subsequent sequencing of thus-isolated CNS cell types validated that expected cellspecific transcripts were only expressed in appropriate CNS cell-type (25,26).
The utilized cell-specific markers were: CD90+ ABs (neurons derived ABs), HepaCAM+ ABs (astrocytes derived ABs), GalC+ ABs (oligodendrocytes derived ABs), CD31+ ABs (endothelial cells derived ABs), and CD14+ ABs (myeloid cells derived ABs) ( Figure 3C). Total number of ABs and cell-specific ABs were adjusted for CSF volume to obtain the number of ABs per ml of CSF (ABs/ml).
Upon unblinding the diagnostic categories, we observed no statistically significant differences in number of total as well as CNS cell-specific ABs across disease diagnostic subgroups or MS activity (Figures 4, 5). However, while using non-overlapping cell-surface markers (i.e., each selected cell surface marker is specific for one CNS cell type and should not be expressed on any other CNS cells), we observed substantial co-expression of these markers on individual ABs. This overlap could be quantified by how much the sum of cell-specific ABs exceeds total ABs (Figure 4 and Supplementary Data File 1). Because the sum of cell-specific ABs always exceeded total number of ABs, we conclude that ABs most likely exhibited non-specific binding of antibodies. High non-specific antibody binding is a well-known problem affecting apoptotic cells, as apoptosis-induced changes in plasma cell membrane upregulate "eat me" signals recognized by phagocytes, including enhanced, non-specific binding of antibodies (37)(38)(39).
Consequently, we observed no correlations between the numbers/concentrations of total or any cell-specific ABs with accurate clinical and imaging measures CNS tissue destruction, disability, MS severity, and disability progression ( Table 2).

DISCUSSION
MS has been studied extensively regarding the inflammatory component of disease (40). However, neurodegenerative component of MS, or immune-mediated destruction of specific CNS cells cannot be measured in living subjects. In this study, we attempted to analyze apoptosis in living subjects by assessing ABs in CSF. While there have been previous attempts to TABLE 2 | Correlation analysis (Spearman r and P-values) between adjusted total (T-ABs) and CNS cell-specific ABs (ND-ABs, neurons; AD-ABs, astrocytes; OD-ABs, oligodendrocytes; ED-ABs, endothelial cells; and MD-Abs, myeloid cells) in subjects' CSF (ABs/ml CSF) and their clinical measures of CNS tissue destruction, disability, and severity.

T-ABs
ND-ABs AD-ABs OD-ABs ED-ABs MD-ABs analyze blood/serum extracellular vesicles as markers of CNS disorders (41)(42)(43), there is no evidence for the presence of CNS ABs in blood/serum; this is likely due to their large size (0.5-4 µm) (16,24,29,44) which prevents ABs from crossing the blood brain barrier (BBB) or their rapid elimination from the blood by the splenic or hepatic reticulo-endothelial system. Moreover, blood/serum naturally has a basal level of ABs from immune system cells which arise during regular immune responses (45,46). Our in-vitro studies validated that selected flow-cytometry assay measures ABs released to culture supernatants. Additionally, using CNS cell-specific surface markers previously validated in human immunopanning isolation of specific CNS cell-types provided high expectation that enumeration of CNS cell-specific ABs may be of clinical value. Unfortunately, after breaking the diagnostic codes we observed no differences between diagnostic categories and no correlations with any clinical or imaging outcomes of disability, CNS tissue destruction or MS severity. While some may argue that our study was under-powered to detect differences between diagnostic categories, we had good representation of subjects from all four diagnostic categories and found no biologically plausible trends. We concluded that expanding our dataset using the same assay would be futile, as such test could never be applied on a patientlevel and therefore cannot outperform current tests such as NFL.

COMRIS-CTD
There are several possible interpretations of our negative results: as cell-surface proteins are often shed during apoptosis (47)(48)(49) and changes in cell membrane structure induced by apoptosis increase non-specific binding of antibodies (37)(38)(39), accurate determination of the origin of apoptotic bodies using flow cytometry may not be possible. The interference from non-specific binding is supported by the observed overlap of multiple CNS cell-type specific surface markers on the individual ABs. If non-specific antibody binding, rather than shedding of cell-surface markers from apoptotic cells was the main cause of our negative results, then attempting to use alternative reagents for detection of cell-surface molecules, such as DNA-aptamers (50) may be of use. Unfortunately, such alternative reagents are not commercially available for validated CNS cell-surface markers. Flow cytometry may also not be an ideal method for analyzing ABs, as older flow cytometers have low resolution for subcellular particles. Our employment of enhanced gating guided by size beads and validation of our assay in cell-culture supernatants mitigated this impediment.
The fate of ABs after their release from the CNS cells is unknown; while some may be secreted to the CSF via extracellular fluid, others, perhaps most, are likely phagocytosed closer to their origin (51)(52)(53). Especially, in the context of pro-inflammatory environment rich in myeloid cells such as activated microglia and infiltrating macrophages, this local capture of ABs may be much more efficient in MS and OIND controls than in healthy subject, mitigating expected differences between diagnostic categories. Thus, CSF concentrations of ABs may not reliably reflect their CNS origin.
We conclude that measuring ABs in the CSF using flow cytometry does not provide desired clinical value. We present our negative report in an effort to prevent other investigators from pursuing this path without incorporating substantial technical advancement that may mitigate problems identified in our study. Thus, a need to develop CNS cell-specific biomarkers reflective of neurodegenerative mechanisms associated with CNS diseases remains.

DATA AVAILABILITY STATEMENT
The flow cytometry data of patients' CSF ABs and their demographic and clinical information are provided in Supplementary Data File 1.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by NIH-CNS-IRB. Written informed consent to participate in the study was provided either by participant or by his/her legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
BB and RM designed the study. RM, JM, and JP performed the experiments. RM, PW, and BB analyzed the data. RM wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version.