Multiplexed selectivity screening of anti-GPCR antibodies

Abstract

G protein-coupled receptors (GPCRs) control critical cellular signaling pathways. Therapeutic agents including anti-GPCR antibodies (Abs) are being developed to modulate GPCR function. However, validating the selectivity of anti-GPCR Abs is challenging because of sequence similarities among individual receptors within GPCR subfamilies. To address this challenge, we developed a multiplexed immunoassay to test >400 anti-GPCR Abs from the Human Protein Atlas targeting a customized library of 215 expressed and solubilized GPCRs representing all GPCR subfamilies. We found that ~61% of Abs tested were selective for their intended target, ~11% bound off-target, and ~28% did not bind to any GPCR. Antigens of on-target Abs were, on average, significantly longer, more disordered, and less likely to be buried in the interior of the GPCR protein than the other Abs. These results provide important insights into the immunogenicity of GPCR epitopes and form a basis for designing therapeutic Abs and for detecting pathological auto-Abs against GPCRs.

Fig. 1.. Experimental workflow and data analysis.

(A) Abs grouped by the phylogenetic subfamily of their GPCR target were coupled to unique color-coded beads and pooled to generate subfamily-specific SBAs (1). Dual epitope-tagged GPCRs were expressed in Expi293F cells, and the membranes were solubilized in detergent, resulting in heterogeneous mixtures of solubilized membrane proteins including the GPCRs. Total protein concentration was normalized across lysates, and aliquots were incubated with the SBAs (2). PE-conjugated anti-1D4 mAb was added to detect the GPCRs captured by the bead-coupled Abs (3). A Luminex FLEXMAP 3D instrument was used to report the data (4). The data were then processed and integrated into an interactive web interface (5). The selectivity of the Abs was determined using Ab-specific thresholds (6). (B) Flowchart of data analysis. Data generated as described in (A) were subject to QC testing for SBA generation and GPCR expression. Next, the data were scaled and centered. Abs were then annotated as binding “on-target” or off-target, and the latter were further subclassified by the proposed cause of off-target binding. (C) Schematic of data visualization. Each example beeswarm plot represents a single Ab, and each dot represents a single cell-based sample. Blue dots represent samples that ectopically express the target GPCR. Gray dots represent samples that ectopically express GPCRs other than the target. The plots enable quantitative identification of four types of Ab binding behavior. Abs binding only the intended GPCR are considered validated. Abs binding intended and additional unintended GPCR targets or unintended GPCRs only are analyzed further to identify potential causes for off-target binding. Abs not binding to any target are considered not validated. Created in BioRender.com.

Fig. 2.. Quantification of overexpressed and solubilized GPCRs by phylogenetic group.

(A) Relative quantities of solubilized dual epitope–tagged GPCRs were determined by SBA immunoassay. Here, FLAG was used for capture, and 1D4 was used for detection. GPCRs are grouped by subfamily. (B) Relative quantities of solubilized dual epitope–tagged GPCRs in the frizzled subfamily. Here, HA was used for capture, and 1D4 was used for detection. (C) Representative examples of GPCR expression and solubilization reproducibility. GPCRs selected from three different subfamilies were expressed in biological triplicate and quantified in technical duplicates (N = 6). Quantification was carried out as described in (A). Significance was determined by a one-way analysis of variance (ANOVA) (with P < 0.05) followed by Dunnett’s multiple comparison tests to mock. ****P < 0.0001 and ***P < 0.001. Sample sizes and P values are listed in table S2. Horizontal bars represent the medians of each group with the 25th and 75th percentiles. All relative levels of solubilized GPCRs were derived from MFI data and plotted here as log₂-transformed values of arbitrary units (AU).

Fig. 3.. Ab selectivity threshold and summary statistics.

(A) Data-driven selection of a hit threshold for defining Ab selectivity. The plot shows the theoretical number of Abs that would be categorized as on-target (green) and Abs that exhibit cross-reactivity (binding of unintended GPCRs, orange) as a function of SDs from the population peak. The threshold was selected to align with the plateau and corresponds to 12 SDs (vertical dashed line). (B) The numbers of Abs fall into different categories based on evidence of GPCR expression, target detection, and cross-reactivity. Green bars indicate Abs that captured only the intended GPCR target, irrespective of the level of GPCR expression. (C) Histogram showing the distribution of the number of validated Abs per GPCRs.

Fig. 4.. Detection of GPCRs with paired Abs.

(Left) Beeswarm plots showing binding events for multiple Abs targeting the same GPCR. Blue dots, intended GPCR; gray dots, unintended GPCRs. Dashed lines correspond to the selectivity cutoff for each HPA Ab. Color coding of HPA Ab ID corresponds to the color coding of the antigen in the snake plot diagram. (Right) Snake plot diagrams showing the antigen sequence used to generate the Ab on the entire protein sequence. Some Abs have the same antigen sequence. Generated with Protter (37). Both columns are divided per subfamily.

Fig. 5.. Deconvolution of observed Ab cross-reactivity behavior.

(A) Deconvoluting off-target binding through expression ratios and sequence homologies. The expression ratio between each on-target GPCR (denominator) and off-target GPCR (numerator) is plotted from highest to lowest. Abs binding off-target GPCRs are listed on the x axis according to the format “on-target GPCR.off-target GPCR.” The black dashed line denotes a twofold difference in expression ratio. The dotted line denotes equal on-target and off-target GPCR expression. The size of each circle conveys the E value of the target GPCR, off-target GPCR pair. The lower the E value, the bigger the circle and the more similar the GPCRs are to each other in the primary structure. Colors indicate whether the off-target GPCR was only captured in one sample (peach) or more than one sample containing it (teal). (B) Summary of the number of cross-reacted GPCRs per group tested shown as a percentage of total Abs per GPCR subfamily and as an absolute number. (C) Comparison of the antigen length used for producing the HPA Abs binding on-target (N = 245) versus those annotated to bind co/off-target (N = 45) or no-target (N = 109). Significance between the groups was determined by a Kruskal-Wallis test (P = 2.5 × 10⁻¹⁵). Asterisks represent P values from Wilcoxon tests. ****P < 1 × 10⁻⁶; n/s, not significant. (D) The predicted local distance difference test (pLDDT) was conducted for the three Ab selectivity classes (green line, on target; red dashed line, off-target; black dot-dashed line, no target) to determine the levels of confidence in structures predicted for the respective antigens on full-length proteins. (E) The averages of relative solvent-accessible surface areas (SASA) were calculated for antigens from the three Ab selectivity classes color coded as in (D) to determine the accessibility of the antigens on the full-length proteins.

Fig. 6.. A brief overview of the web-based interface.

(A to C) The interactive web-based interface contains tabs with information about each Ab. (A) The expression of each GPCR is visualized as expression density plots, and the performance of Abs targeting the selected GPCR is visualized as beeswarm plots. (B) Cross-reactivity of Abs with phylogenetically related targets visualized as heatmaps. (C) Comparison of paired Abs. The interface was generated through the shiny R package.