Abstract
Objective: Recent clinical evidence indicates that persistent reservoirs of SARS-CoV-2 in human brain tissue are associated with various neurologic symptoms. While brain tumors have unique vascular abnormalities and immunosuppressive environments, it is unclear whether SARS-CoV-2 can infect brain tumors.
Methods: Brain tumor samples were collected from a cohort of 72 COVID-19 patients during the SARS-CoV-2 BA.5 wave in Guangzhou. SARS-CoV-2 infection was confirmed by quantitative reverse-transcription polymerase chain reaction (qRT-PCR) and immunohistochemical (IHC) staining. Immune cell infiltration within the tumor tissues was assessed using IHC. RNA-sequencing was performed to investigate virus-host interactions in the brain tumors.
Results: Brain tumor samples from 72 COVID-19 patients were examined and SARS-CoV-2 RNA was detected in 11% of the samples, which included samples from craniopharyngiomas, pituitary neuroendocrine tumors (PitNETs), meningiomas, and gliomas. SARS-CoV-2 infection was present in tumor and endothelial cells within these brain tumors. SARS-CoV-2-positive tumors had greater immune cell infiltration, particularly an increase in CD8+ T cells in gliomas and pituitary PitNETs, along with the activation of innate signaling pathways. The transcriptomic analysis revealed that activation of the complement cascade within tumors may drive changes in the immune microenvironment of SARS-CoV-2-positive tumors.
Conclusions: These findings provided evidence of SARS-CoV-2 infection in brain tumors and suggested a role in altering the tumor immunosuppressive microenvironment.
keywords
Introduction
SARS-CoV-2 infection is associated with a wide range of neurologic complications, including headaches, encephalitis, cerebrovascular disorders, and long COVID syndrome, highlighting an ability to affect the central nervous system (CNS)1–4. Recent studies have detected SARS-CoV-2 RNA and proteins in brain tissue5,6 with viral RNA and proteins detected in postmortem brain samples up to 7 months after initial infection5. The virus is hypothesized to access the central CNS through peripheral nerve pathways or the bloodstream, particularly in conditions in which the blood-brain barrier (BBB) is compromised, such as during brain tumor progression6,7. Brain tumors are characterized by an immunosuppressive microenvironment, which may create a favorable niche for viral persistence and replication8,9. However, SARS-CoV-2 infection in brain tumors and the effect on the tumor immune landscape remain poorly understood.
Tumors of the CNS are immunologically “cold” and exhibit limited infiltration of tumor-infiltrating lymphocytes (TILs) and various immune effector cells compared to peripheral tumors10. This immunosuppressive phenotype contributes to the poor response of brain tumors to immunotherapy, such as immune checkpoint blockade (ICB)11. Innate immune responses, particularly those involving the complement system, have dual roles in antiviral defense and immune regulation in the tumor microenvironment12. Recent studies have implicated hyperactivation of the complement cascade in the pathogenesis of COVID-1913,14, suggesting a link between viral infection and dysregulated immune responses. In addition, innate immune responses, including activation of the complement system, have immunoregulatory functions that remodel the tumor immune microenvironment (TIME), potentially influencing tumor progression15,16.
In the current study, the presence of SARS-CoV-2 was investigated in brain tumors from a cohort of 72 COVID-19 patients. Viral RNA was detected in 11% of the tumor samples, including craniopharyngiomas, pituitary neuroendocrine tumors (PitNETs), meningiomas, and gliomas. By conducting a comprehensive analysis of immune infiltration patterns and molecular profiles, SARS-CoV-2 persistence in brain tumors was shown to be associated with alterations in the immune microenvironment, particularly via the complement system, reshaping the TIME. These findings provide new insights into the interaction between viral infection and the brain TIME with potential implications for therapeutic strategies involving the immunomodulatory function of SARS-CoV-2.
Materials and methods
Patients and samples
A total of 99 tissue samples from 72 patients diagnosed with brain tumors were collected from January to March 2023 during the SARS-CoV-2 BA.5 wave in Guangzhou, all of whom were diagnosed with COVID-19. The Ethics Committee of Sun Yat-sen University Cancer Center granted approval for the collection and use of human brain samples for this study (approval number B2020-165-01). Informed consent was obtained from all participants.
RNA extraction and SARS-CoV-2 RNA quantification
Samples were gathered and processed using the RNeasy Mini kit (74106; Qiagen, Hilden, Germany). The Detection Kit for 2019 Novel Coronavirus RNA (DA0932; DAAN, Guangzhou, China) was used to identify and measure SARS-CoV-2 RNA following the manufacturer’s instructions. SARS-CoV-2 RNA was detected by RT-qPCR targeting the viral N and ORF1ab genes with no-template and positive controls included in each run to ensure assay quality. Samples were considered SARS-CoV-2-positive based on the manufacturer’s criteria if a distinct amplification curve was generated with a CT value ≤ 40 for N/ORF1ab genes.
Antibodies and reagents
The following antibodies were used in the current study: anti-SARS-CoV-2-NP (40143-MM05, 1:100; Sino Biological, Beijing, china); anti-NRP1 (60067-1-Ig, 1:200; Proteintech, Rsemont, IL, USA); rabbit anti-CD31 (77699S, 1:500; Cell Signaling Technology, Danvers, MA, USA); rabbit anti-ACE2 (10108-T24, 1:200; Sino Biological); rabbit anti-TMPRSS2 (14437-1-AP, 1:200; Proteintech); mouse anti-CD147/BSG (66443-1-Ig, 1:200; Proteintech); anti-CD3 (85061s, 1:300; CST, Danvers, MA, USA); anti-CD4 (ab133616, 1:1000; Abcam, Cambridge, United Kingdom); anti-CD8 (ab93278, 1:200; Abcam); anti-CD8 (66868-1-Ig, 1:200; Proteintech); anti-CD20 (ab9475, 1:1000; Abcam); anti-CD68 (76437s, 1:500; CST); anti-C3 (ET7106-75, 1:100; Huabio, Fort Worth, TX, USA); anti-C1q (HA721439, 1:100; Huabio); anti-IgG (ZA-0448; ZSGB-BIO, Beijing, China); anti-IgM (ZA-0450; ZSGB-BIO); anti-TTF1 (66034-1-Ig, 1:200; Proteintech); rabbit anti-GFAP (16825-1-AP, 1:200; Proteintech); anti-MUC1 (HA601142, 1:200; Huabio); anti-panCK (HA601138, 1:200; Huabio); anti-PD-1 (18106-1-AP, 1:200; Proteintech); anti-IFN-γ (AB231036, 1:200; Abcam); and anti-FOXP3 (22228-1-AP, 1:200; Proteintech).
Hematoxylin and eosin (H&E) staining
Tissue samples from patients were fixed in 4% paraformaldehyde, embedded in paraffin, and cut into 4-μm sections. The slides were heated at 65°C for 2 h, then deparaffinized using xylene, rehydrated, and washed with water. The sections were subsequently stained with hematoxylin and eosin, dehydrated, cleared with xylene, and mounted using Mounting Medium (ZLI-9516; ZSGB-BIO). The slides were scanned with a KFbio scanner (KF-PRO-020; Zhejiang, China).
Immunohistochemistry (IHC) and immunofluorescence assay (IFA)
Brain tissue sections were deparaffinized, rehydrated, and underwent antigen retrieval in a pressure cooker at 120°C for 2 min using Tris-EDTA (pH 9.0) for IHC. Endogenous peroxidase activity was inhibited with 3% hydrogen peroxide. After blocking with 20% goat serum for 30 min brain tissue sections were incubated overnight at 4°C with primary antibodies, washed, then treated with secondary antibodies.
Brain tissue sections were blocked with 10% normal goat serum [NGS] (Solarbio, Beijing, China) in phosphate-buffered saline with Tween 20 [PBST] (TargetMol, Boston, MA, USA) at room temperature for 1 h, followed by incubation with primary antibodies diluted in 1% NGS for IFA. After an overnight incubation of the primary antibodies at 4°C the sections were washed 3 times with PBST, then stained with the appropriate secondary antibodies in PBS at room temperature for 1.5 h. IFA was performed using a Branch TSA 4-color combination kit (TissueGnostics, Vienna, Austria) according to the manufacturer’s instructions as previous procedure when primary antibodies originated from the same host17. After washing the sections were stained with DAPI (Solarbio). Finally, after washing with PBST the sections were mounted using Mounting Medium (ZSGB-BIO, China).
Quantification of CD68-, CD3-, CD4-, CD8-, CD20-, C1q-, C3-, IgG-, and IgM-positive cells was performed using QuPath software (version 0.3.2; Queen’s University, Belfast, Northern Ireland). Between 5 and 8 randomly selected fields of view per sample were analyzed to maximize data yield from the limited positive samples available for each brain tumor type. Positive cells were manually counted within each field (area per field: 0.2 mm2). The results are expressed as the average number of positive cells per square millimeter (cells/mm2). All quantitative analyses were performed independently by two researchers blinded to the sample groupings. The final value for each sample represents the mean of the scores from two observers.
Bulk RNA sequencing and data analysis
Total RNA was isolated from SARS-CoV-2-positive or -negative brain tumors using TRIzol (Invitrogen, Carlsbad, CA, USA) and subjected to RNA sequencing by LC-BIO Bio-tech Ltd. (Hangzhou, China). RNA concentration and quality were assessed using a Nanodrop 2000 (Thermo Fisher Scientific, Waltham, MA, USA). An mRNA Purification Kit (Invitrogen) was utilized to remove rRNA and extract poly(A)+ RNA using oligo d(T). The extracted mRNA was subsequently fragmented and short fragments were primed with random hexamers for the synthesis of first- and second-strand cDNA. Adenine (A) was added to the 3′ ends of the blunt fragments and indices and adapters were ligated during the adapter ligation and amplification process. The libraries were sequenced on an Illumina Hiseq 4,000 platform (Illumina, San Diego, CA, USA), producing 150-bp paired-end reads.
RNA-seq reads were aligned to the human reference genome (hg19) using HISAT2 (v2.1.0; https://github.com/DaehwanKimLab/hisat2) with default parameters. Transcript-level read counts were generated with htseq-count (v0.11.2; source). For differential expression analysis, Raw counts were imported into DESeq2 (v1.28.1) in RStudio (https://posit.co/download/rstudio-desktop/). Normalization was performed internally by DESeq2 using the median-of-ratios method. Differential gene expression was defined as a P value (<0.05) and fold-change (>2). The volcano map of the differentially expressed genes was produced by the R package, ggplot2 (v3.3.0). Functional enrichment analysis of Gene Ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways was performed with the clusterProfiler R package. All services were provided by LC Biotech Corporation (Hangzhou, China).
Statistical analysis
Statistics were performed using GraphPad Prism 9 software (GraphPad Software, San Diego, CA, USA). The results were represented as the mean ± standard deviation (SD). The Mann–Whitney U test was used for all comparisons between two independent groups. An adjusted P < 0.05 was considered statistically significant.
Results
SARS-CoV-2 RNA is present in the brain tumor tissues of COVID-19 patients
Ninety-nine tissue samples were collected from 72 patients diagnosed with brain tumors from January to March 2023 during the SARS-CoV-2 BA.5 wave in Guangzhou, all of whom were also diagnosed with COVID-19 (Figure 1A). A total of 48.6% of these patients were male with a median age of 46.5 years [interquartile range (IQR): 36.5–59 years; clinical information in Table S1]. The samples included brain tumor tissues, adjacent paracarcinoma tissues, sphenoid sinus mucosae, and cerebrospinal fluids. The cohort included various brain tumors, comprised of gliomas [n = 25 (34.7%)], meningiomas [n = 19 (26.4%)], and PitNETs [n = 14 (19.4%); Figure 1B]. SARS-CoV-2 RNA specifically targeting the viral N and ORF1ab genes was examined in all 99 tissue samples by performing reverse transcription quantitative PCR (RT-qPCR). Viral N or ORF1ab RNA was identified in the sphenoid sinus mucosae [6/11 (55%)], brain tumor tissues [8/72 (11%)], and cerebrospinal fluids [1/6 (17%)] but not in paracarcinoma tissues (Figure 1C and D). Further analysis revealed SARS-CoV-2 RNA in 1 of 2 craniopharyngiomas (50%), 3 of 14 PitNETs (21%), 2 of 19 meningiomas (11%), and 2 of 25 gliomas (8%; Figure 1E and F). These findings suggested that SARS-CoV-2 infects human brain tumor tissues.
Presence of SARS-CoV-2 RNA in the brain tumor tissues of COVID-19 patients. (A) Overview of the study cohort, which included 72 brain tumor patients from Guangzhou during the Omicron BA.5 surge; details on COVID-19-positivity, age, and gender are provided. (B) Categorization of tissue samples, including various brain tumor types, such as gliomas, meningiomas, and pituitary neuroendocrine tumors (PitNETs), as well as non-brain tumor tissues, such as sphenoid sinus mucosae, paracancerous tissues, and cerebrospinal fluids. (C) Detection of SARS-CoV-2 RNA using the ORF1ab and N genes across all tissue samples. (D) Proportion of positive detection for the ORF1ab and N genes in all tissue samples. (E) Detection of SARS-CoV-2 RNA via the ORF1ab and N genes in different brain tumor subtypes. (F) Proportion of ORF1ab and N genes detected in various brain tumor subtypes.
SARS-CoV-2 protein is expressed in human brain tumor tissues
IHC was performed to detect the SARS-CoV-2 nucleocapsid (N) protein in brain tumor samples. The viral N protein was detected in SARS-CoV-2 PCR-positive sphenoid sinus mucosae and various types of brain tumors, including PitNETs, meningiomas, gliomas, and craniopharyngiomas (Figures 2A, B and S1). However, due to low viral loads, some brain tumor tissues had no detectable viral protein expression. SARS-CoV-2 entry factor expression, including the classical receptor, ACE2, and the co-receptor, TMPRSS2, was analyzed with alternative receptors (CD147 and NRP1) to determine the mechanisms underlying viral entry into brain tumor tissues. IFA analysis revealed that ACE2 and TMPRSS2 expression is absent in brain tumors but NRP1 and CD147 expression is strong (Figures 2C and S2). These findings suggested that SARS-CoV-2 may directly infect brain tumor cells.
SARS-CoV-2 protein detection in human brain tumor tissue. (A) Representative IHC images of SARS-CoV-2 nucleocapsid protein in different brain tissue types that were SARS-CoV-2 PCR-negative or -positive; scale bars: 50 μm. (B) Patient characteristics, including age (>65 years), male gender, PCR-positivity, and IHC-positivity, for different brain tumor types. (C) Representative immunofluorescence images of SARS-CoV-2-related receptors, including ACE2, TMPRSS2, CD147, and NRP1, in different brain tissue types. Nuclei are stained with DAPI.Scale bars: 200 μm.
Specific markers for malignant cells found in different brain tumors were used to determine the ability of SARS-CoV-2 to infect tumor cells: TTF1 for PitNETs; MUC1 for meningiomas; GFAP for gliomas; and CK7 for craniopharyngiomas. These markers were then co-stained with viral nucleoprotein (NP). The results revealed that the viral NP protein is expressed in tumor cells from different brain tumors (Figure 3A). In addition, brain endothelial cells were permissive to SARS-CoV-2 infection (Figure 3B). These findings suggested that SARS-CoV-2 infects both tumor cells and endothelial cells in CNS tumors.
SARS-CoV-2 infection in brain tumor tissue. (A) Immunofluorescence co-staining of tumor markers (green) and nucleocapsid protein (red) in different brain tumor samples. Nuclei are stained with DAPI (blue). Scale bars: 100 μm. (B) Immunofluorescence co-staining of CD31 (green) and nucleocapsid protein (red) in different brain tumor samples. Nuclei are stained with DAPI (blue). Scale bars: 100 μm.
Alterations in immune cell infiltration in SARS-CoV-2-positive brain tumors
Immune cell infiltration patterns were compared between SARS-CoV-2 PCR-positive and -negative brain tumors using IHC for specific immune cell subset markers to determine the effects of SARS-CoV-2 on the TIME. SARS-CoV-2-positive and -negative brain tumors exhibited significant macrophage infiltration, as indicated by CD68 immunostaining. Infiltrating macrophages were the predominant immune cell type inside the intratumor inflammatory infiltrate, especially in meningiomas (Figure 4A–H). SARS-CoV-2-positive brain tumor patients presented a significantly greater density of CD68+ macrophages than the negative counterparts (Figure 4A–H). A greater infiltration of CD3+ T cells was demonstrated in SARS-CoV-2-positive PitNET and glioma patients than SARS-CoV-2-negative patients (Figure 4A–D). This trend was not evident in craniopharyngioma or meningioma patients (Figure 4E–H). Further analysis of T-cell subtypes revealed that CD8+ T cells were more abundant than CD4+ T cells across the intratumor zone, except for craniopharyngiomas (Figure 4A–H). Specifically, CD8+ T-cell infiltration was more abundant in SARS-CoV-2-positive PitNET and glioma patients than SARS-CoV-2-negative patients (Figure 4A and C). Only a few CD20+ B cells were sparsely distributed across all types of brain tumors and SARS-CoV-2 infection did not significantly alter B-cell infiltration in any of the four types of brain tumors examined (Figure 4A–D).
Alterations in immune cell infiltration in SARS-CoV-2-positive brain tumors. (A, C, E, and G) Representative images of H&E staining and immunohistochemical analysis of immune cell markers (CD68, CD3, CD4, CD8, and CD20) in SARS-CoV-2 PCR-negative and -positive brain tumors, including (A) pituitary neuroendocrine tumors, (C) gliomas, (E) meningiomas, and (G) craniopharyngiomas. Scale bars: 100 μm. (B, D, F, and H) Quantitative analysis of the density of positive immune cells (cells/mm2) for each marker in (B) pituitary neuroendocrine tumors, (D) gliomas, (F) meningiomas, and (H) craniopharyngiomas; 5–8 fields/sample were analyzed. (I) Spearman’s test shows a correlation between viral load and the degree of immune cell infiltration. The data represent the mean ± SD; statistical significance was determined by performing the Mann–Whitney U test and the adjusted P values are reported for multiple testing.
IFA staining analysis was performed with functional markers for CD8+ (IFN-γ and PD-1) and CD4+ T cells (FOXP3) to determine the immune phenotype of T cells (effector, exhausted, or regulatory). The density of effector CD8+ T cells (IFN-γ+ CD8+ T cells) was significantly higher in SARS-CoV-2-positive brain tumors than -negative brain tumors (Figure S3A). In contrast, the density of exhausted CD8+ T cells (PD-1+ CD8+ T cells) was significantly lower in positive tumors, except craniopharyngiomas (Figure S3B). However, the density of regulatory CD4+ T cells (FOXP3+ CD4+ T cells) did not differ significantly between the two groups (Figure S3C). These findings indicated that the infiltrating CD8+ T cells in SARS-CoV-2-positive brain tumors are predominantly of an effector phenotype. Moreover, correlation analysis was performed between viral copy number and intratumoral immune cell infiltration density. The viral load was associated with CD8+ T cell infiltration but no correlation existed with CD68+ macrophages, CD3+ T cells, or CD4+ T cell infiltration (Figure 4I). These findings suggested that SARS-CoV-2 infection is associated with alterations in the immune microenvironment of brain tumors.
Activation of intratumor innate immune signaling in SARS-CoV-2-positive brain tumors
Transcriptome sequencing was performed on a diverse set of brain tumor samples to characterize the molecular alterations associated with the persistence of SARS-CoV-2 in brain tumors in an unbiased and comprehensive manner, including six PitNET, three glioma, two craniopharyngioma, and five meningioma samples. GO and KEGG enrichment analyses revealed that significant upregulation of genes involved in pituitary hormone metabolism in SARS-CoV-2-positive PitNETs, revealing the potential effect of SARS-CoV-2 infection on the physiologic functions of PitNETs (Figures 5A and S4A). SARS-CoV-2-positive gliomas exhibited upregulation of antibacterial and antiviral immune responses with suppression of multiple metabolism-related pathways (Figures 5B and S4B).
Activating innate immune signals in SARS-CoV-2-positive brain tumors. (A–C) Gene Ontology (GO) enrichment analysis comparing gene expression profiles between SARS-CoV-2 PCR-positive (Pos) and -negative (Neg) brain tumor samples. The differentially expressed genes and the corresponding GO terms for biological processes are illustrated for (A) pituitary neuroendocrine tumors, (B) gliomas, and (C) meningiomas. (D and E) Heatmaps of gene expression profiles involved in (D) innate immune signaling and (E) cancer-related pathways across SARS-CoV-2-positive and -negative brain tumor subtypes. The Z-scores represent normalized gene expression.
SARS-CoV-2-positive meningiomas exhibited upregulation in pathways related to neurodegeneration, mitochondrial OXPHOS, and calcium signaling but downregulation in viral response, cell cycle, and oncogenic pathways (Figures 5C and S4C). In addition, a significant reduction was noted in the expression of zinc finger protein-related genes, such as ZNF177, ZNF140, and ZBTB11, in SARS-CoV-2-positive meningiomas (Figure S4D). These genes are essential for protection against SARS-CoV-2 infection and the downregulation may weaken the tumor response to SARS-CoV-2 infection. Several oncogenic signaling pathways were downregulated in SARS-CoV-2-positive craniopharyngiomas, whereas multiple immune responses, including complement activation, inflammatory cytokine signaling, and IFN-γ signaling, were markedly activated (Figure S4E and F).
Multiple innate immune responses, including complement activation, phagocytosis, and antiviral responses, were upregulated in SARS-CoV-2-positive tumors based on a comprehensive analysis of SARS-CoV-2-positive tumors, whereas neural signal transmission and oncogenic pathways were reduced (Figures 5A–C and S4A–D). In addition, significant upregulation existed in complement- and antiviral response-related gene expression (e.g., CSF1R, C1R, C1S, C1QB and IFI27) in SARS-CoV-2-positive tumors (Figure 5D). Altered expression of key genes within oncogenic signaling pathways, such as WNT, JAG2, WNT7A, and HEY2, was also noted (Figure 5E). These results suggested that infection with SARS-CoV-2 in brain tumors is associated with activation of innate immune responses, while simultaneously suppressing pro-tumorigenic signaling pathways, highlighting the complex interaction between viral persistence and tumor biology.
SARS-CoV-2-positive tumors exhibit significant activation of the complement cascade
Gene set enrichment analysis (GSEA) was performed to identify the common pathways activated in SARS-CoV-2-infected brain tumors versus uninfected controls to elucidate the mechanism by which SARS-CoV-2 infection affects the TIME. The analysis revealed significant activation of the complement system in SARS-CoV-2-positive tumors across all types of brain tumors (Figures 6A, B and S5A). Specifically, the KEGG complement activation pathway showed robust upregulation of genes affected by SARS-CoV-2 encoding key components of complement components, including C1q (C1qA, C1qB, and C1qC), C1 proteases (C1R and C1S), CFB, and complement C3 (Figure 6C). Immunostaining was performed for various complement proteins to validate that the complement cascade is activated in SARS-CoV-2-positive brain tumors. A significant deposition of complement components (C1q and C3) was detected within the tumors, which was correlated with the deposition of IgG and IgM (Figure 6D). In contrast, only negligible staining was observed in the brain tumors of the SARS-CoV-2-negative control group (Figures 6D, E and S5B). These findings indicated that complement activation is a prominent intracellular response associated with SARS-CoV-2 infection in brain tumors.
SARS-CoV-2-positive tumors exhibited significant activation of the complement cascade. (A) The Venn diagram shows the overlap of activated pathways related to SARS-CoV-2 infection among glioma, craniopharyngioma, meningioma, and pituitary neuroendocrine tumor samples. (B) Gene set enrichment analysis (GSEA) of complement activation and the classical pathway in SARS-CoV-2-positive pituitary neuroendocrine tumor and glioma samples compared to SARS-CoV-2-negative samples. (C) Heatmap of the relative levels of complement-related gene expression in SARS-CoV-2-positive and -negative brain tumor samples. (D) Representative IHC images of complement components (C1q and C3) and immunoglobulins (IgG and IgM) in pituitary neuroendocrine tumor and glioma tissues that were SARS-CoV-2-negative or -positive. Scale bar: 200 μm. (E) Quantification of the IHC intensity of C1q, C3, IgG, and IgM in SARS-CoV-2-positive and -negative samples from pituitary neuroendocrine tumors, gliomas, meningiomas, and craniopharyngiomas; 5–8 fields/sample were analyzed. (F) Co-localization of C1q/C3 with CD68 in SARS-CoV-2-negative or -positive brain tumors. Tumor sections were double-stained for IF: C1q/C3 (green) and CD68 (macrophage marker, red). Scale bar: 100 μm. Five fields/sample were analyzed. Pit, pituitary neuroendocrine tumors; Men, meningiomas; Glia, gliomas; Cra, craniopharyngiomas. The data represent the mean ± SD; statistical significance was determined by performing the Mann–Whitney U test and the adjusted P values are reported for multiple testing.
Immune cell densities were quantitatively compared between regions with high and low complement deposition (C1q, C3) within the same tumor sections from SARS-CoV-2-positive cases to verify the association between complement activation and immune infiltration. This analysis revealed that areas with intense complement deposition contained a significantly higher density of CD68+ macrophages and CD8+ T cells compared to areas with low complement deposition (Figure S6A and B). Furthermore, multiplex immunofluorescence co-staining was performed for complement components (C1q and C3) together with CD68+ macrophages or CD8+ T cells. The results demonstrated greater co-localization of complement components (C1q and C3) with CD68+ macrophages in SARS-CoV-2-positive brain tumors compared to negative tumors (Figures 6F and S6C), while CD8+ T cells were notably enriched within areas of complement deposition (Figure S6D). These findings suggested that macrophages and CD8+ T cells are key cellular responders to complement activation. Moreover, multiplex immunofluorescence was performed to analyze the co-localization of complement components (C1q and C3) with immune and tumor cells. Complement signals primarily overlapped with tumor cells, followed by macrophages (~15%), with minimal CD8+ T cell co-localization (Figure S6C–E). These results suggested that SARS-CoV-2-infected tumor cells may actively contribute to local complement initiation and amplification, possibly by upregulating and secreting some components (e.g., C3) or serving as sites for deposited complement fragments. Taken together, these findings indicated that local complement production arises from complex interactions between tumor and infiltrating immune cells, particularly macrophages.
Discussion
SARS-CoV-2 infection in brain tumors and the effect on the TIME across nine types of human brain tumors during the BA.5 wave were comprehensively analyzed in the current study. SARS-CoV-2 was detected in several types of brain tumors within the cohort of COVID-19-positive brain tumor patients, including craniopharyngiomas, PitNETs, meningiomas, and gliomas. SARS-CoV-2 infection was shown to be associated with alterations in immune cell infiltration and intratumor innate immune signaling activation, especially the complement cascade.
The current study provided histopathologic evidence of SARS-CoV-2 infection in multiple brain tumor types from COVID-19 patients. Viral RNA was identified in 8 of 72 (11%) tumor samples with no detection in adjacent paracarcinoma tissues. In addition, viral RNAs and/or proteins were identified in four types of brain tumors (craniopharyngiomas, PitNETs, meningiomas, and gliomas), supporting the notion that SARS-CoV-2 has recently undergone active replication or at least completed the processes of viral entry and uncoating. These findings are similar to previous reports of SARS-CoV-2 detection in the cerebral cortex and pituitary gland of COVID-19 patients5,18. However, the susceptibility of gliomas to SARS-CoV-2 infection remains poorly characterized with current evidence limited to case reports19 and in vitro monolayer systems20,21. Disruption of the BBB in brain tumors, coupled with a immunosuppressive tumor microenvironment7,8, may facilitate viral neuro-invasion through peripheral nerves and bloodstream routes6. Immunohistochemical localization demonstrated viral infection in tumor and endothelial cells within the brain tumors. Expression of the classical SARS-CoV-2 receptor, ACE2, and the key co-factor, TMPRSS2 was absent or very low in tumor cells among the examined brain tumor tissues. However, marked high expression of the alternative receptors, neuropilin-1 (NRP1) and CD147 (BSG), was noted. SARS-CoV-2 infection of brain tumor cells may not rely on the ACE2-TMPRSS2 axis but could be mediated through mechanisms, such as endocytosis via the NRP1 and/or CD147 pathways. Unlike endothelial cells and astrocytes, which may express elevated levels of ACE222,23, the data herein indicated that brain tumor cells exhibit a distinct “ACE2−/TMPRSS2−, NRP1+/CD147+” phenotype. This finding suggests that the virus may utilize fundamentally different “molecular keys” to invade different brain cell types. In addition, the results of the molecular analysis revealed higher expression of SARS-CoV-2 entry mediators, particularly NRP1, in human brain tumor tissues than normal brain tissue17,24, which further suggests a potential tropism of SARS-CoV-2 for infecting brain tumor cells. Thus, the findings herein indicated that brain tumors might be vulnerable to SARS-CoV-2 infection, highlighting the need for further investigation into the mechanisms underlying viral tropism and the effect on tumor biology and progression.
Greater macrophage infiltration was demonstrated in SARS-CoV-2-positive brain tumors than negative controls. Macrophages are the most abundant immune cells in the brain tumor microenvironment are are critical for viral recognition and bridging innate and adaptive immunity25. CD8+ or IFN-γ+ CD8+ T cells were more abundant in SARS-CoV-2-positive brain tumors than negative controls. Moreover, CD8+ T cell infiltration exhibited a correlative trend with viral load. These CD8+ T lymphocytes are essential for the adaptive immune response against viral infections and the presence in brain tumors suggests activated antitumor immunity26. This finding strongly supports a “dose-dependent” relationship in the strength of the adaptive anti-viral immune response, suggesting that SARS-CoV-2 infection directly or indirectly drives the recruitment and/or expansion of tumor-specific or virus-specific CD8+ T cells. SARS-CoV-2 infection might disrupt the immunosuppressive environment of brain tumors, create a more immunogenic microenvironment, and thereby improve the effectiveness of immunotherapy. This finding is consistent with studies demonstrating differences in the immune microenvironments of lung cancer patients with and without SARS-CoV-2 infection27. Moreover, various case studies have indicated that SARS-CoV-2 infection may have oncolytic effects on individuals with blood cancer, metastatic colon cancer, and clear cell renal cell carcinoma28–32. If SARS-CoV-2 infection can trigger a lasting systemic anti-tumor immunity, a positive impact on delaying tumor progression or improving patient outcomes may occur. However, the overall effect of SARS-CoV-2 infection on brain tumors likely depends on multiple factors, including tumor type, genetic background, patient immune status, as well as the viral infection dose and timing.
An unbiased and comprehensive transcriptome analysis was performed to clarify the mechanisms underlying the changes in the TIME associated with SARS-CoV-2 infection. The results revealed that SARS-CoV-2 infection leads to substantial transcriptional changes in several key biological processes, including complement activation, phagocytosis, cell cycle regulation, and oncogenic pathways. Innate immune responses were significantly enriched in SARS-CoV-2-infected brain tumors. The GSEA and immunohistochemical data showed that the classical complement pathway is strongly activated in SARS-CoV-2-positive tumors with key components, such as C1q and C3, being upregulated and deposited in tumor tissues. This coordinated multi-pathway response is more consistent with an integrated host defense against SARS-CoV-2 than non-specific, dysregulated tissue inflammation. The complement cascade is an essential component of the innate immune system that orchestrates the release of inflammatory mediators, phagocytic responses, and cell lysis33,34. We consider that activation of the complement system is the core bridge connecting viral infection and T-cell responses. Immune cell densities between regions with high and low complement deposition (C1q and C3) were quantitatively compared. The results showed greater co-localization of complement components with CD68+ macrophages and CD8+ T cells. Recent studies have shown that activation of the complement system has a role in the pathogenesis of severe SARS-CoV-2 infection13,35,36. Complement activation products (C5a and C3a) are powerful chemokines that directly chemoattract T cells, including CD8+ T cells, to inflammatory sites37. In addition, these complement components, such as C1q, and subsequently deposited immunoglobulins (IgG/IgM) can form antigen-antibody complexes. The latter can serve as strong immune stimulatory signals through the bridging of antigen-presenting cells, such as macrophages, indirectly promote the activation and clonal proliferation of CD8+ T cells, thereby exerting antiviral and even potential antitumor effects.
The IFA analysis herein revealed that complement components (C1q/C3) are predominantly co-localized with tumor cells (~80%), implicating virus-infected tumor cells as a major source and site of complement deposition. Macrophages accounted for most of the remaining signal (~20%), suggesting a role as responders and potential local producers. The activation is likely initiated by pre-existing SARS-CoV-2-specific antibodies forming immune complexes with local viral antigens via the classical pathway, which is then amplified by the tumor microenvironment. This localized complement response may function as an endogenous immune adjuvant within the infected tumors. Activating the complement system locally in tumors by delivery of complement component agonists or nanomaterials could replicate the immune activation caused by viral infections, counteract the immunosuppressive environment of brain tumors, and enhance the effectiveness of current immunotherapies, such as immune checkpoint inhibitors. In addition, there is a potential risk of abnormal complement-driven immune activation in the brain among COVID-19 patients with brain tumors. Monitoring markers of complement activity could help identify those patients who might benefit from managing CNS complications.
The current study had some limitations. First, RT-qPCR primers targeted the conserved regions of N and ORF1ab; it is impossible to determine whether the viral RNA is derived from BA.5 or earlier variants based solely on this data. Second, we were unable to systematically and completely obtain the SARS-CoV-2 infection history of all 72 patients prior to the BA.5 infection event. Third, SARS-CoV-2 infection was associated with alterations in the brain TIME immune microenvironment but evidence of a causal relationship is lacking. Finally, it remains unclear whether the immune infiltration resulting from viral infection specifically targeted the virus, tumor cells, or both. Further studies are needed to investigate the tumor progression and long-term effects of SARS-CoV-2 infection.
Notably, despite a limited sample size, a uniform viral background was noted in the BA.5 wave and consistently detected SARS-CoV-2 with complement activation in multiple brain tumor types. This finding suggests that infection-triggered immune remodeling is a common feature of these cancers. The data support a model wherein SARS-CoV-2 infection profoundly remodels the TIME, potentially creating a state that is permissive for anti-tumor immunity in addition to simply recruiting virus-specific T cells, as follows: (1) The transcriptomic analysis revealed that SARS-CoV-2-positive tumors are characterized by a significant enrichment of pathways related to “interferon-gamma response,” “immune cell infiltration,” and “phagocytosis” coupled with the activation of multiple innate antiviral immune pathways. This signature indicates a broad shift from an immunosuppressive state to an immunologically active state, aligning with an effector anti-tumor phenotype. (2) The robust complement cascade activation serves as a key mechanistic link. Complement cleavage products are potent chemo-attractants and inflammatory mediators that can non-specifically recruit and activate various immune cells, including macrophages and T cells, and enhance antigen presentation. Thus, the virus-triggered complement response likely acts as an immunologic adjuvant, reshaping the immune microenvironment. These findings highlighted the role of SARS-CoV-2 in modulating the tumor immune landscape and offer insights into the complex interaction between viral infections and tumor progression.
Conclusions
Overall, the current study provides the first human-tissue evidence that SARS-CoV-2 can infect various primary brain tumors. Importantly, infection is associated with significant remodeling of the local TIME, characterized by robust complement system activation, increased infiltration of immune effector cells, such as CD8+ T cells, and downregulation of specific oncogenic pathways. These findings suggested that SARS-CoV-2 infection may transiently reverse the immunosuppressive state of brain tumors by triggering a complement-driven innate immune response. This work offers a novel perspective on virus-tumor-immune interactions and highlights potential avenues for immunomodulatory therapeutic strategies.
Supporting Information
Conflict of interest statement
No potential conflicts of interest are disclosed.
Author contributions
Conceived and designed the analysis: Weikang Chen, Xiaobing Jiang, Xiaoxing Li.
Collected the data: Haojie Bai, Maoling Tian, Lei Zhou, Wange Lu.
Contributed data or analysis tools: Yanxia Huang, Depei Li, Linyi Liu.
Performed the analysis: Weikang Chen, Haojie Bai, Yanxia Huang, Luyao Wu, Wei Li.
Wrote the paper: Weikang Chen, Haojie Bai, Xiaobing Jiang.
Data availability statement
The data generated in this study are available upon request from the corresponding author.
Acknowledgments
We gratefully acknowledge the staff at the Precision Medicine Research Institute of the First Affiliated Hospital of Sun Yat-sen University for support and facilitation of the study.
- Received August 15, 2025.
- Accepted January 6, 2026.
- Copyright: © 2026, The Authors
This work is licensed under the Creative Commons Attribution-NonCommercial 4.0 International License.
References
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.