Comprehensive BCMA Expression Profiling in Adult Normal Human Brain Suggests a Low Risk of On-target Neurotoxicity in BCMA-targeting Multiple Myeloma Therapy

Abstract

B-cell maturation antigen (BCMA) is a target for the treatment of multiple myeloma with cytolytic therapies, such as chimeric antigen receptor T-cells or T-cell redirecting antibodies. To better understand the potential for “on-target/off-tumor” toxicity caused by BCMA-targeting cytolytic therapies in the brain, we investigated normal brain BCMA expression. An immunohistochemistry (IHC) assay using the E6D7B commercial monoclonal antibody was applied to 107 formalin-fixed, paraffin-embedded brain samples (cerebrum, basal ganglia, cerebellum, brainstem; 63 unique donors). Although immunoreactivity was observed in a small number of neurons in brain regions including the striatum, thalamus, midbrain, and medulla, this immunoreactivity was considered nonspecific and not reflective of BCMA expression because it was distinct from the membranous and Golgi-like pattern seen in positive control samples, was not replicated when a different IHC antibody (D6 clone) was used, and was not corroborated by in situ hybridization data. Analysis of RNA-sequencing data from 478 donors in the GTEx and Allen BrainSpan databases demonstrated low levels of BCMA RNA expression in the striatum of young donors with levels becoming negligible beyond 30 years of age. We concluded that BCMA protein is not present in normal adult human brain, and therefore on-target toxicity in the brain is unlikely.

Introduction

B-cell maturation antigen (BCMA) is a member of the tumor necrosis receptor superfamily encoded by the TNFRSF17 gene with expression in normal tissue reportedly restricted to the following immune cells: plasmablasts, plasma cells, a small subset of memory B-cells within the bone marrow and peripheral organs (gastrointestinal tract, trachea, spleen, lymph nodes), and in activated plasmacytoid dendritic cells. ¹ This limited normal tissue expression profile, along with the enhanced expression in plasma cell tumors,^2,3 makes BCMA an ideal target for specific and potent immunotherapeutics like T-cell redirectors and chimeric antigen receptor T (CAR-T) cell therapies.

A key safety consideration regarding the use of T-cell-activating therapies for the treatment of cancer is the potential for immune-mediated cytotoxicity, also referred to as “on-target/off-tumor toxicity,”^4,5 particularly of normal cells that are terminally differentiated, for which cell loss could be significantly detrimental to organ functions. As opposed to immune cell–redirecting antibodies, CAR-T cells have been documented to cross an intact blood–brain barrier into the central nervous system (CNS),^6–8 making in-depth investigation of target expression in normal CNS cells a critical part of safety assessment of CAR-T cell targets. Although multiple BCMA-targeting CAR-T cell therapies are in active development, information regarding the expression of BCMA protein in human brain is limited. A thorough understanding of BCMA expression in the brain is needed for the differentiation of on-target/off-tumor toxicity from the well-documented target-agnostic neurotoxicity associated with CAR-T cell therapy.

Previous investigations of BCMA expression in brain have been conducted, but were limited to specific anatomic regions of cerebrum, cerebellum, and medulla oblongata. Carpenter et al. ⁹ performed immunohistochemistry (IHC) on tissue microarrays of formalin-fixed, paraffin-embedded (FFPE) human tissue samples. These authors used a goat polyclonal BCMA-targeting antibody (R&D Systems AF193), and no immunoreactivity was detected in cerebrum or cerebellum samples. Bu et al. ¹⁰ used the same polyclonal antibody as well as an additional rabbit polyclonal antibody (US Biological B0807-50G) and corroborated the Carpenter et al. data, with no immunoreactivity detected using the R&D Systems antibody. However, as the US Biological antibody labeled cerebellar climbing fibers, the authors embarked on a more comprehensive investigation performing IHC (using both antibodies) and RNA in situ hybridization (ISH) on human and cynomolgus cerebellum and medulla oblongata FFPE samples. The authors also performed immunoprecipitation with mass spectrometry on additional fresh frozen cerebellum and medulla brain samples. No mRNA was detected by ISH, and the mass spectrometry data were equivocal; therefore, the authors ultimately determined that immunoreactivity with the US Biological antibody was nonspecific cross-reactivity that did not reflect BCMA protein expression. These investigations suggest that BCMA protein expression in normal adult human brain is unlikely; however, not all regions of the brain were examined, and publicly available transcriptomic data and development neurobiology research, as described below, indicated that a more comprehensive investigation into BCMA protein expression was warranted.

Publicly available gene expression (RNA) data sets such as RNA-sequencing (RNA-seq) and microarray data sets offer a new opportunity for examining BCMA RNA expression in the brain. The largest of these data sets comes from the GTEx database (https://gtexportal.org/home/), which contains samples from 50 unique human body sites, including 13 brain regions from hundreds of donors. These data show that the expression of BCMA RNA in the brain is extremely low [0–0.21 transcripts per million (TPM)]. By comparison, tissues known to contain resident plasma cells (intestine and spleen) have greater than five TPM. However, publications documenting the role of BCMA in neural development in mice and microarray data available from the Allen Brain Institute (https://www.brainspan.org/) suggest that BCMA may be expressed in the striatum.

As part of research efforts to understand the role of immune mediators in neurodevelopment, fetal mouse studies provided additional insight into BCMA and the role it may play in brain development, particularly in the hippocampus and striatum. Osorio et al. ¹¹ detected BCMA protein by immunofluorescence staining of neurons in frozen mouse fetal hippocampus sections but did not describe what controls were used to validate antibody performance and did not examine mice beyond postnatal day 10.5. Embryonic day 18 hippocampal pyramidal neurons responded in vitro to APRIL [an endogenous ligand of BCMA, BAFF receptor (BAFFR), and TACI] by increasing axonal length, which was blocked by the addition of anti-BCMA antibodies or when a truncated form of BCMA was expressed. McWilliams et al. ¹² detected BCMA RNA via RT-qPCR in mouse striatal samples through adulthood. BCMA protein expression was not measured in mouse brain, but this study demonstrated that axonal outgrowth of midbrain dopaminergic neurons was specifically prevented when BCMA, but not TACI, blocking antibodies were added. Given the importance of the striatum to neurocognitive and motor function, and the potential for BCMA-targeting CAR-T cells to enter the neuroparenchyma, a comprehensive investigation into the baseline expression of BCMA in the human brain is an unmet need.

To address this apparent gap, new fit-for-purpose IHC and ISH assays were developed to detect BCMA expression in human normal brain tissue sections. These assays were used along with multiple orthogonal approaches, including transcriptomic analysis. Based on the totality of the data, the weight of the evidence does not support the presence of BCMA protein in the adult anatomic regions and population studied.

Materials and Methods

Immunohistochemistry Method Development

Generation of Assay Controls

Seven cell lines were selected according to endogenous BCMA RNA expression levels, based on RNA-seq data from the publicly available CCLE (https://sites.broadinstitute.org/ccle/) and Genentech (https://ega-archive.org/dacs/EGAC00001000055) data sets. The selected cell lines served as qualified reagent controls with BCMA expression at negative, low, medium, and high levels (Table 1). As BCMA is part of larger family of tumor necrosis factor receptors, including BAFFR and TACI, which share ligands with BCMA, additional work was done to confirm the lack of antibody cross-reactivity to potentially conserved epitopes. Engineered cell lines expressing TACI or BAFFR were generated by transduction of HEK293 parental cells with lentiviruses carrying the human TACI or the human BAFFR gene coupled with a Tag epitope. Cell lines were cultured to 100% confluence, non-enzymatically collected, and pelleted by centrifugation. Cell pellets were fixed in 10% neutral-buffered formalin for 24 hr, routinely processed to paraffin, and embedded to form an array in a single paraffin block.

Table 1.

Antibodies Screened.

Vendor	Cell Signaling Technology	R&D Systems	BioLegend	Novus Biologicals
Clone name	E6D7B	1004023	Clone 19F2	Vicky-1
Isotype	Rabbit monoclonal IgG	Mouse monoclonal IgG2b	Mouse monoclonal IgG2a with kappa light chain	Rat monoclonal IgG1
Immunogen	Synthetic peptide including amino acid residues around leucine 115 (cytoplasmic domain)	No information	Recombinant IgG Fc-BCMA fusion protein	Synthetic peptide: amino acid residues 2–54 (extracellular domain)
Applications (given by each provider)	IHC, WB, IP, FC	ELISA, FC, IHC, CyTOF	FC	IHC, IF, ICC, WB, ELISA, FC
Concentrations used in IHC assay screening	7.8 µg/ml (optimal IHC concentration)	10 to 2 µg/ml	5 to 1 µg/ml	10 to 2 µg/ml
Resulting IHC staining in positive and negative controls	Generated a specific and sensitive IHC staining toward BCMA expressed by the cells	Generated unacceptable sensitivity and specificity (H929 presented a low-intensity brown wash, U937 that should be negative presented positive staining)	Generated unacceptable sensitivity and specificity (H929 presented positivity on dead cells only)	Generated unacceptable sensitivity and specificity (H929 presented positivity on dead cells only, U937 and K562 that should be negative presented positive staining)

Abbreviations: BCMA, B-cell maturation antigen; IHC, immunohistochemistry; WB, Western blot; IP, immunoprecipitation; FC, flow cytometry; CyTOF, cytometry by time flight; ICC, immunocytochemistry; ELISA, enzyme-linked immunosorbent assay.

FFPE samples of colon, spleen, and lymph node were acquired from the Janssen R&D Nonclinical Safety Molecular Pathology tissue repository as endogenous tissue controls due to the presence of resident BCMA-expressing plasma cells. All tissues were quality controlled (QCed) for their anatomic location, lack of histopathology, and suitability for IHC and ISH, and only samples that met QC criteria were used in experiments. To be considered suitable for IHC, the sample had to show the expected immunolabeling pattern by synaptophysin IHC. Suitability for ISH was confirmed by abundant positive signal for the mRNA housekeeping gene, peptidylprolyl isomerase B (PPIB), and a lack of DapB (negative control probe) signal.

Screening Candidate IHC Reagents

The IHC assay was developed on the Leica Bond Rx autostainer (Leica Biosystems; Buffalo Grove, IL). Table 1 summarizes the IHC candidate reagents considered to develop the definitive BCMA IHC assays. The IHC reagents were tested on the cell pellet reagent controls to assess their specificity and sensitivity toward the BCMA protein. Of the four reagents tested, only the rabbit monoclonal anti-BCMA antibody clone (E6D7B) presented acceptable IHC binding specificity and sensitivity. Therefore, the anti-BCMA antibody clone (E6D7B) was used in a refined assay as primary antibody for the detection of endogenous expression of BCMA protein on tissue sections.

Assay Conditions

The final IHC method used rabbit monoclonal anti-BCMA antibody clone (E6D7B) (Cell Signaling Technology, cat. no. 88183; Danvers, MA). FFPE blocks were sectioned at 4 µm, and the samples were mounted on SuperFrost Plus glass slide (VWR, cat. no. 48311-703; Radnor, PA). Briefly, unbaked glass slides were loaded in the autostainer and were deparaffinized following the generic Leica deparaffinization protocol. Heat-mediated antigen retrieval with an EDTA-based solution (pH ~9.0) was performed at 100C for 20 min. Slides were then pretreated with endogenous peroxidase solution for 10 min. Dako serum-free protein block (Agilent, cat. no. X0909; Santa Clara, CA) was applied on the samples for 10 min just before a 30-min incubation with a 1/200 dilution of the rabbit monoclonal anti-BCMA antibody clone (E6D7B) [final concentration of 7.8 µg/ml in antibody diluent (Diagnostic BioSystems, cat. no. K0004; Pleasanton, CA)]. After extensive washing steps, the bonded primary antibody was detected by the chromogenic Leica refine DAB detection kit (Leica, cat. no. DS9800; Buffalo, NY) according to the manufacturer’s recommendations. The slides were mounted with glass coverslips and were examined with a bright field microscope.

RNA In Situ Hybridization Assay Development

Controls

Previously described cell pellet array and human FFPE colon samples served as reagent controls. For each ISH staining assay, the FFPE blocks were sectioned at 4 µm, and the samples were mounted on SuperFrost Plus glass slides.

Reagents

The ISH assay was developed on the Leica Bond Rx autostainer (Leica Biosystems; Wetzlar, Germany) using the following key reagents: the mRNA detection probes including human BCMA-specific probe (Hs-TNFRSF17; ACDBio, cat. no. 585791; Newark, CA), human-positive tissue PPIB control probe (Hs-PPIB; ACDBio, cat. no. 313908), and a negative control probe DapB (ACDBio, cat. no. 312038). The hybridized probes were detected using the RNAscope 2.5 LSx Reagent Kit-Red (ACDBio, cat. no. 322750).

Assay Conditions

The BCMA ISH assay was performed on the Leica Bond Rx autostainer. Briefly, glass slides were loaded in the autostainer, baked at 60C for 30 min, and deparaffinized following the generic Leica deparaffinization protocol. Heat-mediated antigen retrieval with an EDTA-based solution (pH ~9.0) was performed for 15 min at 85C for the cell pellet reagent controls or 95C for the tissue samples. Nonspecific enzymatic digestion with proteinase K (provided in the ACDBio RNAscope 2.5 LSx kit) was then applied on each histological sample for 15 min at 40C. Hybridization of the specific probes occurred at 42C for 120 min. After extensive washes, the specifically bonded probes were detected by a series of signal amplification steps. Finally, an alkaline phosphatase enzymatic activity reacted upon a chromogen, producing a red precipitate signal visible under a bright field microscope.

The ISH staining signal results in an intracellular dot-like pattern. The number of positive dots per cell generally correlates to the amount of detectable mRNA transcripts present. The size of each dot depends preferentially on the overall probe set design. Evaluation of the ISH signal was visually performed according to the manufacturer’s guidelines. ¹³ The minimum number of dots when considering a cell positive and its relationship to correlative protein expression depends on each target of interest. As there can be differences between the level of mRNA and protein, a side-by-side characterization between ISH signal level and IHC immunoreactivity intensity provides acceptable indication of the ISH signal threshold to consider relative to protein expression.

Preanalytical variables in tissue harvesting and processing, such as prolonged time in ethanol, may influence the quality of mRNA that can be detected with the ISH method. The PPIB is a housekeeping gene and its ISH signal was used to assess the overall quality of the mRNA present in the FFPE samples (i.e., QC check). Each cell pellet included in the control array presented an ISH PPIB signal well above the manufacturer’s minimum recommended threshold of three to four dots per cell.

Investigational Human Brain Tissues for IHC and ISH

FFPE human normal brain tissues were commercially sourced from various Janssen-approved vendors. Each brain tissue sample was checked for location accuracy by H&E and IHC suitability by synaptophysin staining [rabbit monoclonal anti-synaptophysin antibody 1/8000 (Abcam, cat. no. 32127; Cambridge, UK)]. A total of 107 brain samples spanning a wide range of brain loci were used in this study. The samples originated from 63 different donors. A subset of samples was further evaluated for ISH suitability using PPIB (positive control probe).

Colocalization Studies

Colocalization studies between the IHC immunoreactivity observed with E6D7B and Golgi apparatus and neurofibrillary tangle markers were performed by immunofluorescence. Assessment of the staining was completed using a Zeiss confocal microscope (Zeiss; White Plains, NY). The colocalization experiments included the following antibody probes: cis- and trans-Golgi markers, mouse monoclonal anti-TGN46 antibody 1/500 (LSBio, cat. no. LS-C133654-100; Seattle, WA), mouse polyclonal anti-GOLGA2 antibody 1/100 (Abnova, cat. no. H00002801-B01P; Taipei City, Taiwan), mouse monoclonal anti-GOLGA5 antibody 1/300 (NovusBio, cat. no. NBP2-66875; Littleton, CO), and mouse monoclonal anti-GOLM1 antibody 1/100 (Sino Biological, cat. no. 13066-MM12; Wayne, PA). Each specific antibody was detected by the following secondary binders: goat polyclonal Alexa Fluor 488 AffiniPure F(ab′)₂ Fragment Anti-Mouse IgG (H+L) (Jackson ImmunoResearch Laboratories, cat. no. 115-546-146; West Grove, PA) and goat polyclonal Alexa Fluor® 594 AffiniPure F(ab′)₂ Fragment Anti-Rabbit IgG (H+L) (Jackson ImmunoResearch Laboratories, cat. no. 111-586-144).

Bielschowsky Silver Stain

To highlight neurofibrillary tangles, key tissue samples were stained with the Bielschowsky silver stain method (Abcam ab245877). The method was executed according to the manufacturer’s recommendation. Briefly, 4-µm tissue sections were deparaffinized and hydrated before incubation in a solution of silver nitrate for 15 min at 40C, followed by 10 min in ammoniacal silver solution. The silver staining was developed in developer solution under agitation until desired coloration. The precipitated silver was fixed with 5% sodium thiosulfate for 2 min and the slides were dehydrated and mounted with Permount (Fisher Scientific, SP15-100; Hampton, NH) before digital capture. Detection of phosphorylated Tau protein was also performed using the tangle marker probe: mouse monoclonal anti-pTau 1/100 (RnD Systems, cat. no. MAB34941-100).

Transcriptomic Analysis of BCMA Expression in Normal Brain

GTEx RNA-seq Data

RNA-seq data of the GTEx Analysis Release V8 were obtained from the dbGaP database (https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs000424.v8.p2) and were processed using the Omicsoft Array Suite tool (https://www.arrayserver.com/wiki/). Human genome build GRCh38 and gene model GENCODE Release 33 (https://www.gencodegenes.org/human/) were adopted for mapping RNA-seq sequences to genome and quantifying gene expression. All 2641 samples of 13 brain regions, including 246 caudate samples and 204 putamen samples, were selected for the profiling of BCMA expression in human brain. TPM was adopted as the unit of gene expression measurement.

Allen BrainSpan RNA-seq Data

All gene expression data were downloaded from the brain Allen BrainSpan website (https://www.brainspan.org/) using the link https://www.brainspan.org/api/v2/well_known_file_download/267666525. Data were converted from reads per million kilobases to TPM to enable compatibility with the GTEx data. BCMA expression was reported with samples grouped by brain region and development stage.

Aggregated RNA-seq Data

To determine how the expression of BCMA gene in human striatum varies with stage of development, BCMA gene expression data in striatum samples from the Allen BrainSpan data set and in caudate and putamen samples from the GTEx data set were extracted. The combined data were plotted together and split by age group of the donor.

Results

IHC Assay Development

The cell pellet reagent control array staining results are summarized in Table 1 and Fig. 1. The mouse monoclonal anti-BCMA antibody clone (1004023) (RnD Systems, cat. no. MAB1931), the human monoclonal anti-BCMA antibody clone (Clone 19F2) (BioLegend, cat. no. 357502; San Diego, CA), and the rat monoclonal anti-BCMA antibody clone (Vicky-1) (Novus Bio, cat. no. NBP1-97637SS) did not generate acceptable specificity and sensitivity toward BCMA protein expressed nor by the control cell pellets (data not shown). The rabbit monoclonal anti-BCMA antibody clone (E6D7B) generated the expected pattern of BCMA-specific immunoreactivity (i.e., membrane and/or Golgi-like pattern) in cell pellet controls (Fig. 1). H929 and MM1.R cells presented marked positive-membrane staining and a marked Golgi-like staining pattern in their cytoplasm. The JEKO-1 and Raji cells displayed a low- to very-low-intensity staining, mostly observable in the Golgi-like structure present in the cells’ cytoplasm. No immunoreactivity was observed with E6D7B in the BCMA-negative cell lines (K562, U937, and HEK293), as expected. BCMA immunoreactivity generated by the clone (E6D7B) was orthogonally confirmed with an ISH assay specific for detecting BCMA mRNA. HEK293-BAFFr and HEK293-TACI cell lines overexpressing the closely related targets BAFFR and transmembrane activator and CAML interactor (TACI) did not present any BCMA immunoreactivity, further indicating the specificity of E6D7B to BCMA.

Figure 1.

IHC and ISH assay qualification: cell pellet array and control tissues. Detection of BCMA protein by immunohistochemistry and BCMA RNA by in situ hybridization (ISH) on formalin-fixed, paraffin-embedded (FFPE) cell pellets and tissue controls. (A–G) Representative images of FFPE cell pellets expressing various levels of BCMA: H929 (A), MM1R (B), Jeko-1 (C), Raji (D), K562 (E), U-937 (F), and HEK293 (G). In high-expressing cell lines, BCMA is detected on the cell membrane and in the perinuclear region (A and B). In low-expressing cell lines, BCMA is only detected in the perinuclear region (dark blue chevrons, C and D). (A′–G′) Corroborative ISH on same cell lines. Light blue chevrons indicate detection of BCMA RNA. (H) BCMA IHC on human colon FFPE samples. Immunoreactivity is membranous and perinuclear in putative resident plasma cells within the lamina propria. Inset: Higher magnification highlighting putative plasma cells. (I) ISH on human colon FFPE. Putative plasma cells are positive. Inset: Higher magnification highlighting putative plasma cells. (J and K) IHC on BCMA-negative cells transfected with BAFFR (J) or TACI (K). No immunoreactivity is detected, despite successful transduction of the cell lines as indicated by anti-Tag immunoreactivity (insets). Primary BCMA antibody: Cell Signaling Technology clone E6D7B, Chromogen DAB. ACD RNAscope BCMA probe (cat. no. 585798), Chromogen RedAP. Primary anti-Tag antibody Cell Signaling Technology clone 9A3, Chromogen RedAP. Abbreviations: BCMA, B-cell maturation antigen; IHC, immunohistochemistry. Scale bars: A–G′ = 25 µm; H = 100 µm; I = 80 µm; J–K = 50 µm.

In tissue sections, specific positive labeling was identified in cells with morphology consistent with plasma cells in human lymph node, spleen, and colon samples. Like in the BCMA-positive cell pellet reagent controls, plasma cells presented a strong membrane staining and a Golgi-like positive structure when present in the plane of section. The relative intensity of the positive cells in the lymph node and spleen was less than that observed in the colon tissue samples.

ISH Assay Development

BCMA-specific ISH staining was identified in cell lines presenting a range of endogenous expression, from low to high BCMA mRNA fragments per kilobase million (FPKM) scores (Table 1, Fig. 1). Cell lines with high BCMA FPKM scores (H929 and MM1.R) presented 15 or more dots per cell. In cells with lower FPKM scores (JEKO-1 and Raji), five dots and one dot per cell were observed, respectively. As expected, the BCMA ISH assay was not able to generate specific signal in cell lines displaying extremely low or zero FPKM scores (K562, U937, and HEK293).

The direct correlation between the BCMA ISH signal and BCMA IHC immunoreactivity in the same control cell pellets was visually estimated (Table 1, Fig. 1). Acceptable concordances were observed on the FFPE cell pellets between the intensity of the BCMA IHC immunoreactivity and the BCMA ISH signal. Not all Raji cells displayed BCMA ISH positivity in the plane of section, and the positive cells presented mainly one dot and rarely two dots. For that cell line, the overall density of positive BCMA ISH signal matched that observed with the BCMA IHC assay reaching the threshold for positivity as determined by recommended scoring by the manufacturer (https://www.indicalab.com/wp-content/uploads/2018/04/MK_51_103_RNAScope_data_analysis_guide_RevB.pdf). Although occasional ISH signal was present in negative control cell lines (K562, U-937, and HEK293) the signal was infrequent and did not reach the 1 dot per 10 cells threshold.

In human normal colon tissue sections (positive control), specific positive BCMA ISH labeling was identified in cells with morphology consistent with plasma cells. Most of those cells displayed a low level of BCMA ISH signal (one dot per cell and occasionally two to three dots per cell). Groups of plasma cells exhibiting a single BCMA ISH dot signal appeared strongly stained by the IHC assay in subsequent adjacent sections, indicating that, similar to Raji control cells, IHC-detectable BCMA protein expression in tissue could be predicted for cells presenting a sustained ISH level as low as one dot per cell.

The presence of observable positive signal generated by the BCMA ISH assay required a particular overall mRNA quality threshold, as assessed by the PPIB QC ISH assay (four dots per cell or greater). Plasma cells that presented a specific positive BCMA ISH signal generally had a minimum of four dots per cell present in the QC PPIB ISH signal. In contrast, there was no corresponding BCMA ISH signal detected in areas where plasma cells had low PPIB ISH signal, lower than four dots per cell, despite strong and specific positive signal at the protein level (IHC on immediately adjacent sections).

BCMA IHC on Human FFPE Normal Brain Samples

Immunoreactivity with rabbit monoclonal anti-BCMA antibody clone (E6D7B) (CST) was observed in a small subset of neurons in specific anatomic regions of several normal human brain samples. The immunoreactivity pattern presented as cytoplasmic fibril- or needle-like structures found in the cell body (soma) and axonal processes of certain neurons (Fig. 2) and rarely in glial cell processes. Neurons with IHC immunoreactivity were mainly observed in the dorsal striatum, thalamus, midbrain, and medulla oblongata. Among all neurons present on the tissue sections, the density of cells with immunoreactivity varied by the location examined and was generally low. A similar regional pattern of immunoreactivity was consistently reproduced across brain tissues and donors, regardless of the demographic or commercial origin of the samples. Accounting for the relative low number of samples assessed in this study, no clear parallel was found between the IHC immunoreactivity with E6D7B and available demographic data associated with the tissue samples.

Figure 2.

Side-by-side Cell Signaling and Santa Cruz D6 IHC on brain samples. (A–C) Three representative human FFPE samples—one of putamen (A) and two of medulla (B and C)—showing immunoreactivity with Cell Signaling E6D7B clone (red tags). (D–I) IHC using a different antibody clone (Santa Cruz Biotech clone D6) at two external molecular pathology providers on the same representative samples. The immunoreactivity seen in A–C did not reproduce (D–I). Primary BCMA antibodies: Cell Signaling Technology clone E6D7B, Santa Cruz Biotechnology clone D6. Chromogen: DAB. IHC assays were optimized for each clone and performed on a Leica Bond autostainer. Abbreviations: BCMA, B-cell maturation antigen; FFPE, formalin-fixed, paraffin-embedded; IHC, immunohistochemistry. Scale bar = 100 µm.

The unique fibril-/needle-like subcellular, non-membranous, non-Golgi-like pattern of immunoreactivity in human normal brain was unexpected as it greatly differed from the characteristic pattern in cells that are known to express this protein (e.g., plasma cells). The unexpected immunoreactivity warranted further characterization. To that end, orthogonal experimentations were conducted on representative human brain samples where IHC immunoreactivity was observed.

To further assess BCMA expression in the human brain samples, external molecular pathology service providers (CROs) were contracted to develop independent BCMA IHC assays, at least as sensitive as E6D7B, for application in detecting low levels of the BCMA protein. Two BCMA IHC assays developed at Reveal Biosciences (San Diego, CA) and Hematogenix (Tinley Park, IL) were based on the mouse monoclonal anti-BCMA clone D6 from Santa Cruz Biotechnology. Those two assays generated acceptable and comparable specificity and sensitivity in targeting BCMA on the same cell pellet reagent and colon tissue controls used to develop the internal BCMA IHC assay. A set of the same brain samples that demonstrated immunoreactivity using the E6D7B clone were sent to each CRO for staining in a blinded manner. In the control samples, the BCMA IHC based on the Santa Cruz anti-BCMA antibody clone D6 presented very similar staining pattern to the internal BCMA IHC (Supplementary Fig. 1). In contrast, no BCMA staining or fibril-like immunoreactivity was seen in the brain samples. Thus, a BCMA IHC assay based on the Santa Cruz mouse monoclonal anti-BCMA antibody clone D6, performed independently at two contract laboratories, did not reproduce the immunoreactivity originally observed in the brain samples with rabbit monoclonal anti-BCMA antibody clone (E6D7B).

BCMA ISH on Human FFPE Normal Brain Samples

The specific detection of mRNA transcripts from the gene of the protein of interest in the same cells and anatomic regions could increase confidence in the specificity of the observed immunoreactivity. Therefore, we sought to apply a BCMA ISH, a method that reveals gene expression with anatomic and cellular resolution, to corroborate the observed immunoreactivity when using the E6D7B clone. A total of 49 brain samples, chosen at random without regard to E6D7B-mediated immunoreactivity, were stained with the mRNA QC probe PPIB. Of these 49 samples, 25 presented an acceptable PPIB control signal (4 dots per cell or more) and were used in subsequent experiments. The BCMA-specific probe was applied on those 25 samples, and only 4 of those samples generated a very low (1–2 dots per cell) positive signal spotted in 1 or 2 cells among the thousands present in the section. A similar signal could be detected in negative control cell pellet samples (HEK293 cells), but this signal was below the threshold of what is considered positive based on the assay. The immunoreactivity produced with the E6D7B clone was seen in occasional neurons throughout the brain sections, whereas the ISH signal was exceedingly rare.

Colocalization Assays

To further characterize the immunoreactivity resulting from the BCMA IHC assay performed on normal human brain tissues, immunofluorescent colocalization experiments with Golgi apparatus markers were conducted. In plasma cells that endogenously express BCMA protein, BCMA can be found on the cell membrane or within the Golgi apparatus. ¹⁴ The Golgi apparatus of neurons differs greatly in shape, size, and distribution when compared with plasma cells; therefore, it was deemed appropriate to determine whether BCMA immunoreactivity colocalized with known Golgi proteins within the neuronal cell bodies. To cover the span of the Golgi apparatus, cis-Golgi (GOLGA2 and GOLM1, PMID: 18953438) and trans-Golgi (TGN46, PMID: 29311477) IHC markers were examined. In tissue-resident plasma cells and reagent control cell pellets, the BCMA IHC assay located the protein at the plasma membrane and invariably in a Golgi-like structure in the cell cytoplasm. As anticipated, confocal observations carried out between Golgi markers and the BCMA antibody resulted in an evident colocalization in cell pellet controls. In the neurons of the normal human brain samples, no colocalization could be observed between cis- and trans-Golgi markers and the immunoreactivity resulting from E6D7B (Fig. 3).

Figure 3.

Confocal microscopy study of colocalization with Golgi markers. Confocal microscopy was used to determine whether immunoreactivity seen with the E6D7B clone was located in the Golgi apparatus. (A–D) H929 cells served as positive control cells for BCMA localization to the Golgi apparatus. Blue: DAPI (A); green: GOLM1 (B); red: E6D7B clone labeling BCMA (C); merged images showing colocalization of GOLM1 and BCMA in the perinuclear region confirming detection of BCMA in the Golgi apparatus (arrows, D). (E–H) Medulla. Blue: DAPI (E); green: GOLM1 (F); red: E6D7B clone immunoreactivity (G); merged image (H). Aside from autofluorescence in scattered neurons (chevrons), no colocalization between GOLM1 and E6D7B immunoreactivity was detected in medulla. Immunofluorescent staining was performed with the Leica Bond Rx autostainer, and GOLM1 and E6D7B bond antibodies were detected, respectively, with Goat polyclonal Alexa Fluor 488 AffiniPure F(ab′)₂ Fragment Anti-Mouse IgG (H+L) (Jackson ImmunoResearch Laboratories, cat. no. 115-546-146) and Goat polyclonal Alexa Fluor® 594 AffiniPure F(ab′)₂ Fragment Anti-Rabbit IgG (H+L) (Jackson ImmunoResearch Laboratories, cat. no. 111-586-144). The staining was assessed by confocal microscopy (Zeiss, LSM700). Abbreviation: BCMA, B-cell maturation antigen. Scale bar: A–D = 50 µm; E–H = 25 µm.

In addition, as the immunoreactive fibrils are morphologically reminiscent of neurofibrillary tangles seen in neurodegenerative diseases, such as Alzheimer’s disease, colocalization of pTau (the most common aggregating protein seen preclinically and clinically in Alzheimer’s disease) ¹⁵ with BCMA was also examined. The colocalization was assessed by confocal microscopy. The pTau IHC staining indicated the absence of tangles in the tissue samples assessed; therefore, the immunoreactivity observed with E6D7B did not display any colocalization with that marker (Supplementary Fig. 2). The pTau colocalization study did not reveal any relationship with the immunoreactivity seen in the tissue samples. As protein aggregates can be formed by a wide variety of proteins aside from pTau, a Bielschowsky silver staining was used to identify most tangles independently of the protein(s) involved. Select neurons displayed punctate intraneuronal argyrophilic aggregates, and most of the cells presented a clear cytoplasm. In the medulla, the distribution and morphology of the argyrophilic aggregates did not resemble the pattern generated by E6D7B IHC, suggesting that the immunoreactivity would most likely not originate from aggregated proteins (Supplementary Fig. 2).

RNA-seq Analysis

GTEx

Across brain samples, BCMA RNA detection is generally negligible (0–0.21 TPM). In the caudate nucleus and putamen (components of the striatum), there are a small subset of samples with slightly higher TPM values (1–3 TPM).

Allen BrainSpan

BCMA (TNFRSF17) RNA expression (Fig. 4) is detected in the striatum in samples from fetal (average TPM = 5.2), infant (average TPM = 15.1), juvenile (average TPM = 9.5), pubertal (one donor, TPM = 2.2), and young adults (19- and 20-years old, average TPM = 4.6). BCMA RNA levels are negligible (average TPM less than 0.4) in all other brain regions where multiple development stages were examined.

Figure 4.

RNA-seq: three-panel graphic GTEx, Allen BrainSpan, and aggregated striatum data (A). GTEx bulk RNA-seq data for all brain regions. All regions exhibit a median RNA expression level below 0.2 TPM. A small number of samples in the caudate nucleus and putamen (components of the dorsal striatum) show slightly higher levels of expression (between 1 and 4 TPM). (B) Allen BrainSpan bulk RNA-seq data for all brain regions plotted by development stage. Low levels of BCMA expression are detected in younger age groups. (C) Aggregated BCMA RNA expression data for striatum from Allen BrainSpan and components of the striatum (GTEx: caudate nucleus and putamen) plotted by donor age. BCMA RNA expression in the striatum decreases with increasing age. Abbreviations: BCMA, B-cell maturation antigen; TPM, transcripts per million.

Aggregated BCMA RNA-seq Expression Data: Striatum

GTEx and Allen BrainSpan BCMA RNA expression levels for the striatum and components of striatum (caudate nucleus and putamen) samples (n=478) show a clear decline in expression with increasing age (Fig. 4). BCMA RNA expression is generally negligible in the striatum of donors over the age of 30 years.

Discussion

Given conflicting literature related to the expression of BCMA in the CNS, and the relevance of that information to BCMA-targeting immunotherapies, we pursued a comprehensive survey of the presence of BCMA protein in human normal brain. This study employed multiple complementary assays to meet this objective, including IHC, ISH, and RNA-seq data analysis.

BCMA protein expression in normal human brain was examined by IHC on qualified commercially sourced FFPE brain samples using two different monoclonal antibodies (E6D7B and D6). Both antibody clones sensitively and specifically detected BCMA in cell pellet and tissue controls and were reported by the vendors to target similar regions of the intracellular domain of the BCMA protein.^16,17 In the brain samples, immunoreactivity was only detected when the E6D7B clone was used. The immunoreactivity occurred in a very small subset of neurons of brain sections arising from the striatum (caudate nucleus, putamen), midbrain, medulla, and occasionally cerebrum. As the D6 antibody staining was negative in the same regions, these contradictory IHC results necessitated further investigation.

In a previous study, polyclonal antibodies were used to assess BCMA expression in cerebrum, cerebellum, and medulla by IHC. ¹⁰ Although they did not examine the striatum and midbrain, their detailed examination of the medulla offered the opportunity for comparison with the current study. As in this study, one of the two polyclonal antibodies assessed in the prior study produced immunoreactivity in the medulla; however, the morphology and distribution of that immunoreactivity was markedly different from what was seen with the E6D7B antibody clone in this study. Furthermore, in the prior study, the investigators could not detect BCMA expression in the medulla and cerebellum by ISH or mass spectrometry and concluded that the observed immunoreactivity was nonspecific. ¹⁰

Confocal microscopy was used to further understand the intracellular immunoreactivity seen with the E6D7B clone, to determine whether it could be due to the presence of protein within the Golgi apparatus, the subcellular locale where BCMA is reportedly expressed in plasma cells. ¹⁴ In cells known to express BCMA (multiple myeloma cell lines), BCMA immunoreactivity colocalized with that of Golgi apparatus immunomarkers. This colocalization was not seen in neurons, signifying that the immunoreactivity seen with the E6D7B clone occurred outside of the neuronal Golgi system.

To determine whether immunoreactivity seen with the E6D7B clone could be corroborated by BCMA RNA expression data, ISH was performed on a subset of FFPE brain samples that were confirmed adequate for ISH assessment by PPIB staining. BCMA ISH signal was limited to 1–2 dots in 1–2 neurons in 4 of 25 brain samples examined. This level of detection is considered within background according to RNAscope grading scheme. ¹⁸ Furthermore, this rare background level signal did not occur in the same location as where the IHC signal was seen with the E6D7B clone. Examination of two publicly available RNA-seq databases (GTEx and Allen BrainSpan) showed generally low BCMA RNA distribution in striatal tissue in donors younger than 30 years of age. RNA expression of BCMA in other regions of the brain and in striatum beyond the age of 30 years is negligible. This is consistent with an investigation linking BCMA signaling to neuronal development in the striatum of mice and shows the importance of considering the developmental stage of the donor when drawing conclusions about gene expression in normal tissue. Donor demographic data were not always available for the current study; however, most of the samples came from older adults (age range, 26–-85 years), and immunoreactivity with E6D7B was seen across age groups.

A recent review on BCMA and neurotoxicity references microarray data used to examine the expression of BCMA in neurons. ¹⁹ The work included in this article focused on RNA-seq data, because microarray data are based on relative expression and not true quantification of transcript abundance, and the use of microarray to detect low-abundance transcripts is not recommended. Wang et al. ²⁰ have compared RNA expression data generated by both microarray and RNA-seq platforms. The authors use RT-qPCR as a “gold standard” for measuring gene expression and compared it with both microarray and RNA-seq for a collection of genes. They found that for highly expressed genes, both RNA-seq and microarray show high concordance with RT-qPCR; however, for low-abundance transcripts, concordance between microarray and RT-qPCR was very low with microarray both over- and underpredicting gene expression levels. By contrast, RNA-seq showed high concordance with RT-qPCR under these conditions. Based on the RNA-seq data from GTEx and Allen BrainSpan, BCMA transcript in the brain is a very-low-abundance transcript, which makes the interpretation of microarray data for this gene ill-advised.

The rabbit monoclonal anti-BCMA antibody clone (E6D7B) generated the expected pattern of BCMA-specific immunoreactivity (i.e., membrane and/or Golgi-like pattern) in cell pellet controls. In FFPE tissue sections, specific positive labeling was identified in cells with morphology consistent with plasma cells (strong membrane staining and a Golgi-like positive structure) in human lymph node, spleen, and colon samples. However, in tissue sections of the brain, the IHC immunoreactivity lacked anticipated membranous and Golgi-like pattern seen in positive control samples. The expression was characterized by cytoplasmic fibril- or needle-like structures found in the cell body (soma) and axonal processes of a small subset of neurons and rarely in glial cell processes, did not colocalize to the Golgi, was not replicated with a different IHC antibody, and was not corroborated by ISH data. Therefore, it was determined that immunoreactivity within the brain with rabbit monoclonal anti-BCMA antibody clone (E6D7B) was nonspecific cross-reactivity that did not reflect BCMA expression.

Considering the collective evidence, including all the orthogonal studies performed, we concluded that BCMA protein was not present in normal adult human brain by IHC. This study included 107 FFPE brain samples from 63 unique donors and evaluated RNA-seq data from an additional 288 donors. Although this sample set encompasses the most in-depth look at BCMA expression in the human brain to date, to our knowledge, the chance that a specific subset of the human population could express BCMA protein in the brain beyond 30 years of age cannot be definitively ruled out. In addition, this study did not include brain samples from multiple myeloma patients or from people with neurological comorbidities. This may be relevant because previous studies have found that BCMA expression in plasmacytoid dendritic cells (normally negative for BCMA) can be induced due to toll-like receptor agonism. ²¹ As the expression of BCMA appears to be increased in multiple myeloma tumor cells as compared with normal plasma cells as part of a prosurvival pathway, ²² it is possible that factors driving BCMA expression within the tumor could change expression patterns of BCMA elsewhere in the body.

Further investigation is necessary to explore the question of BCMA expression in the brain in the setting of multiple myeloma. A proportion of multiple myeloma patients who are treated with BCMA-targeting therapies have experienced neurotoxic adverse events. ²³ Symptoms of these neurotoxicities are wide-ranging and include clustered movement and cognitive disorders in some patients. A recent case report described a patient who was treated with an anti-BCMA CAR-T therapy (ciltacabtagene autoleucel) and had neurological adverse events, including a movement disorder consistent with parkinsonism. ²⁴ The authors examined this patient and one healthy subject and reported BCMA expression in neurons and astrocytes in the caudate nucleus, using a rabbit polyclonal antibody (US Biological B0807-50). However, that article did not present details on the development of their assay, such as staining results from orthogonally qualified BCMA-positive and BCMA-negative control samples, which would be needed to rule out nonspecific immunoreactivity. ²⁴ As the mechanism underlying neurological symptoms remains undefined, evaluation of a large number of brain samples from myeloma patients with neurological symptoms is needed; however, the absence of BCMA protein detection in the present data suggests a low risk of direct on-target/off-tumor cytotoxicity in the brain.