Unmasking features of the auto‐epitope essential for β₁‐adrenoceptor activation by autoantibodies in chronic heart failure

Abstract

Aims

Chronic heart failure (CHF) can be caused by autoantibodies stimulating the heart via binding to first and/or second extracellular loops of cardiac β₁‐adrenoceptors. Allosteric receptor activation depends on conformational features of the autoantibody binding site. Elucidating these features will pave the way for the development of specific diagnostics and therapeutics. Our aim was (i) to fine‐map the conformational epitope within the second extracellular loop of the human β₁‐adrenoceptor (β₁EC_II) that is targeted by stimulating β₁‐receptor (auto)antibodies and (ii) to generate competitive cyclopeptide inhibitors of allosteric receptor activation, which faithfully conserve the conformational auto‐epitope.

Methods and results

Non‐conserved amino acids within the β₁EC_II loop (compared with the amino acids constituting the EC_II loop of the β₂‐adrenoceptor) were one by one replaced with alanine; potential intra‐loop disulfide bridges were probed by cysteine–serine exchanges. Effects on antibody binding and allosteric receptor activation were assessed (i) by (auto)antibody neutralization using cyclopeptides mimicking β₁EC_II ± the above replacements, and (ii) by (auto)antibody stimulation of human β₁‐adrenoceptors bearing corresponding point mutations. With the use of stimulating β₁‐receptor (auto)antibodies raised in mice, rats, or rabbits and isolated from exemplary dilated cardiomyopathy patients, our series of experiments unmasked two features of the β₁EC_II loop essential for (auto)antibody binding and allosteric receptor activation: (i) the NDPK^211–214 motif and (ii) the intra‐loop disulfide bond C²⁰⁹↔C²¹⁵. Of note, aberrant intra‐loop disulfide bond C²⁰⁹↔C²¹⁶ almost fully disrupted the functional auto‐epitope in cyclopeptides.

Conclusions

The conformational auto‐epitope targeted by cardio‐pathogenic β₁‐receptor autoantibodies is faithfully conserved in cyclopeptide homologues of the β₁EC_II loop bearing the NDPK^211–214 motif and the C²⁰⁹↔C²¹⁵ bridge while lacking cysteine C²¹⁶. Such molecules provide promising tools for novel diagnostic and therapeutic approaches in β₁‐autoantibody‐positive CHF.

Introduction

Dilated cardiomyopathy (DCM)—defined as progressive cardiac dilatation and dysfunction without coronary heart disease—mostly affects younger adults. It frequently entails severe heart failure and transplantation. Prevalence and annual incidence amount to 0.5% and 0.1%, respectively. ¹ According to the 2008 ESC cardiomyopathy classification—within the non‐familial forms—autoimmune DCM has been recognized as a proper pathogenetic entity. ² Autoimmunity arising from acute myocardial inflammation induced by microbial or viral agents is considered one key factor in the pathogenesis of autoimmune DCM. Progress to chronic immune‐cardiomyopathy and chronic heart failure (CHF) is common in patients unable to clear the infecting agent, ³ showing persistent inflammation, ⁴ and/or having a familial pre‐disposition to autoimmune reactions. ⁵ DCM autoimmunity encompasses autoantibodies directed against myocardial antigens ⁶ , ⁷ , ⁸ and cardiovascular G‐protein‐coupled receptors (for review, see Boivin‐Jahns and Jahns ⁹ ), most notably the β₁‐AR. ¹⁰ , ¹¹ , ¹² , ¹³ The latter (termed anti‐β₁‐aabs) are suspected to play a pivotal pathogenic role because anti‐β₁‐aab‐positive compared with aab‐negative DCM patients have a more severely depressed cardiac function ¹³ and a threefold increased risk for cardiovascular death. ¹⁰ The incidence of anti‐β₁‐aabs can precede DCM symptoms by many years, ¹⁴ and selective removal of anti‐β₁‐aabs can decelerate disease progression and improve outcome in β₁‐aab‐positive DCM patients. ¹⁵ , ¹⁶ Most DCM‐relevant anti‐β₁‐aabs trigger a similar conformational switch of the β₁‐AR molecule as the true agonists ¹⁷ and elicit detrimental cardiac effects similar to chronic catecholamine challenge. The pathogenic relevance of stimulating anti‐β₁‐aabs in DCM is corroborated by DCM equivalents induced in rodents by β₁‐AR immunization, isogenic antibody transfer, and corresponding rescue experiments. ¹¹ , ¹⁸ , ¹⁹ , ²⁰ At bottom line, chronic cAMP stimulation by anti‐β₁‐aabs appears to play a major role in DCM pathogenesis. Because blockade of sympathetic hormone receptors by beta‐receptor antagonists has proven beneficial in CHF, a first logical step was to explore whether these drugs could also (pharmacologically) neutralize the cellular effects of stimulating anti‐β₁‐aabs. ²¹ Although β₁‐selective (e.g. bisoprolol and metoprolol) and non‐selective beta‐blockers (e.g. alprenolol and carvedilol) were able to significantly reduce anti‐β₁‐aab‐stimulated cAMP production, in the end, all tested substances—even at saturating concentrations—blocked anti‐β₁‐aab‐mediated effects only partially [resulting in a 50% (alprenolol) up to 70% (carvedilol) reduction of the antibody‐induced fluorescence resonance energy transfer (FRET) signals]. Consequently, clinical disease management would greatly profit from diagnostic determination and direct targeted intervention against stimulating anti‐β₁‐aabs, but the prerequisite knowledge of the structural features of the auto‐epitope crucial for cAMP stimulation is yet too limited for that.

DCM‐relevant anti‐β₁‐aabs target the first (β₁‐EC_I) and/or second extracellular loops (β₁‐EC_II) of the β₁‐AR. ²² , ²³ IgG binding to immobilized linear peptides representing these domains has been used to determine the prevalence of anti‐β₁‐aabs in various diseases. However, these assays exhibit low sensitivity/specificity and, in general, fail to discriminate patients from healthy subjects in a non‐random fashion, because the disease‐relevant anti‐β₁‐aabs target specific conformations of β₁‐EC_I and/or β₁‐EC_II that are not properly represented by immobile linear peptides. To date, DCM‐relevant anti‐β₁‐aabs can only be detected by native assays or functional readouts. ¹⁹ , ²⁴ , ²⁵ However, the many patients potentially to be screened for anti‐β₁‐aabs (i) require clinical diagnostics based on conventional procedures [e.g. enzyme‐linked immunosorbent assay (ELISA)] for measuring IgG binding to synthetic representations of the very auto‐epitope conformation specifically targeted in the course of allosteric β₁‐AR activation. ¹⁷ , ²⁴ Interestingly, circular peptide representations of β₁‐EC_II (alone or as an add‐on to cardioselective β₁‐blockers) could stop progression and even revert CHF in rats subjected to β₁EC_II immunization(s), whereas cardioselective β₁‐blockers alone were only able to stabilize the dilated cardiomyopathic phenotype. ¹⁹ , ²⁰ Therefore, cyclopeptides might provide an adequate synthetic mimic of the DCM‐relevant conformational auto‐epitope. Following up on this idea, we unmask here the structural features within β₁‐EC_II that are essential and specific for allosteric receptor activation by anti‐β₁‐aabs and required in cyclopeptides for a neutralization of that effect.

Methods

Anti‐β₁EC_II antibodies

Anti‐β₁EC_II antibodies were raised in male Lewis rats ¹¹ or rabbits (Dianova, Hamburg, Germany) or mice (mouse Mab 23‐6‐7, BioGenes, Berlin, Germany ²⁶ ) by immunization with β₁‐EC_II/glutathione S‐transferase (GST) fusion proteins. In addition, a monoclonal rat antibody was generated by fusing spleen cells from anti‐β₁EC_II‐positive immunized rats with a multiple myeloma Sp2/0‐Ag14 cell‐line (rat Mab 13F6 ²⁶ ). Hybridoma cells expressing mouse Mab 23‐6‐7 and rat Mab 13F6 were submitted to DSMZ‐Deutsche Sammlung von Mikroorganismen und Zellkulturen GmbH (Braunschweig). Depositing was carried out according to the rules of the Budapest Treaty (accession numbers DSM ACC3121 and ACC3174).

Human anti‐β₁‐aabs were gained from representative CHF patients. ²¹ Because of the relatively large amounts of serum necessary to isolate human IgG for the serial functional assays, as a proof of concept in the present work, IgG was isolated from three DCM patients only {two men [56a, New York Heart Association (NYHA) IV, left ventricular ejection fraction (LVEF) 29%, and 53a, NYHA III, LVEF 31%] and one woman (29a, NYHA IV, LVEF 26%)}. In all three patients, LVEF was determined by ventriculography, and coronary heart disease was excluded by coronary angiography. At the time of invasive diagnostics, all three patients were stable under standard CHF medication including angiotensin‐converting enzyme inhibitor/AT1 receptor blockers, beta‐blockers, and aldosterone antagonists. In none of the patients, exposure to cardio‐toxic substances, myocarditis, or other systemic heart diseases was evident from clinical history. IgG was freshly isolated from the respective (human or animal) sera and dialyzed against the appropriate assay buffers. Immuno‐reactivity of rodent IgG against a linear 25 amino acid (AA) β₁EC_II peptide (residues 199–123) was determined by ELISA. ¹² , ¹³

Cyclopeptides

Cyclopeptides corresponding to AA residues 200–220 of the human β₁‐AR and constituting the entire β₁‐EC_II ²⁷ were cyclized between the C‐terminal glutamate and the free N‐terminal amino group (Peptide Speciality Laboratories, Heidelberg, Germany). Non‐conserved residues differing between the β₁‐AR and β₂‐AR were sequentially substituted by alanine. Cysteine²¹⁶ (C²¹⁶) was replaced with α‐aminobutyric acid (B) to prevent aberrant disulfide bridging to C²⁰⁹. Corresponding 22‐mer cyclopeptides of the second extracellular loop of the human β₂‐adrenergic receptor (β₂‐EC_II) served as negative controls. In 18‐mer cyclopeptides representing a minimal EC_II structure encompassing residues 204–219 of the human β₁‐AR, C²¹⁵ or C²¹⁶ was replaced with serine to selectively disrupt potential intra‐loop disulfide bridges (for further details, see Figure 2 and Table S1). Cyclopeptides were freeze‐dried, reconstituted in water, and stored at −20°C. For the functional assays, anti‐β₁‐abs/β₁‐aabs were pre‐incubated with 40 mol of cyclopeptides/mol IgG, assuming a 1:1 stoichiometry between the cyclic 22AA‐EC_II peptide variants (~3.0 kDa) and one IgG molecule (~150 kDa). In fact, the molar excess was only 20‐fold, assuming two variable (Fab) chains with two antigen‐binding regions per IgG molecule. Antigen‐binding regions are further subdivided into hypervariable (HV) and framework (FR) regions; HV regions comprise about five to eight AAs directly contacting a portion of the antigen's surface. ²⁸ IgG was freshly isolated from the respective (human or animal) sera by caprylic acid precipitation, dialyzed against the appropriate assay buffers, and allowed to interact with the Cyclopeptides (CPs) for 16 h at 4°C (cold room) in an Eppendorf cup on a gently rotating incubator. After short spin, the supernatant was used for the assays.

Figure 2.

Schematic representation of the 22‐mer C/C/B‐β₁ cyclopeptide mimicking the human β₁EC_II bearing the essential motif of the clinically relevant auto‐epitope: circles with capitals specify amino acids (single letter code). Numbers specify their position in the sequence of the human β₁‐AR. ‘B' stands for the artificial amino acid α‐aminobutyrate substituting C²¹⁶ in order to prevent a non‐physiological S‐S‐bond C²⁰⁹↔C²¹⁶. Boxed letters N and C indicate the termini of the predicted β₁EC_II loop. The full line between C²⁰⁹ and C²¹⁵ indicates the essential intra‐loop disulfide bridge; black circles depict amino acid residues essential for the neutralization of stimulating rodent anti‐β₁EC_II‐abs. Grey circles represent additional amino acid residues supposed to be relevant for the neutralization of stimulating human anti‐β₁EC_II‐aabs. Amino acid residues R²⁰⁷, E²⁰⁵, S²⁰³, and T²²⁰ were not found essential by directed mutational analysis.

cAMP measurements by a fluorescence resonance energy transfer sensor

cAMP measurements by a FRET sensor using HEK‐β₁E₁ cells stably co‐expressing the human β₁‐AR and a FRET sensor for intracellular cAMP followed published procedures. ²¹ Cells were grown on poly‐d‐lysin‐coated 96‐well plates (Ibidi, Martinsried, Germany). Endogenous β₂‐AR was blocked with ICI‐118551 (0.1 μM). Cells were imaged with a fluorescence microscope optimized for the cyan fluorescent protein (CFP)/yellow fluorescent protein (YFP) donor/acceptor pair of the cAMP sensor (iMIC 2000, TILLPhotonics, Martinsried, Germany); CFP/YFP emissions were continuously recorded for 32 min. FRET efficiency reflecting intracellular cAMP levels was calculated every 6 s using TillvisION v4.5.59 (TILLPhotonics) and OriginPro 8G (OriginLab, Northampton, USA). Antibody effects normalized to agonist maximum were derived from successive determinations of baseline (10 min) and effect of the antibody tested (20 min) and of (−)‐isoproterenol (1 μM, 2 min) performed on sets of ≈500 cells.

β₁‐AR mutants

cDNA constructs of the human β₁‐AR bearing point mutations E202A, R207A, D212N, P213A, or V219A were generated by site‐directed mutagenesis (Table S2), confirmed by sequencing, transiently expressed in HEK293 cells, and characterized by Western blotting and radioligand displacement. ²¹ To ensure reproducibility (and sufficient supply) of the test agent ‘anti‐β₁EC_II antibody' for our experiments with human β₁‐AR mutants, we used monoclonal anti‐β₁EC_II‐abs raised in mice (Mab 23‐6‐7 ²⁶ ) or rats (Mab 13F6 ²⁶ ), thereby reducing the variability due to potentially different HV regions of polyclonal rabbit, rat, or human IgGs used in the experiments before. ²⁸ HEK293 cells expressing human β₁‐AR with the indicated point mutations were exposed (60 min, 37°C) to anti‐β₁EC_II‐Mabs (0.2 μg/μl) or (−)‐isoproterenol (1 μM) in the presence of 3‐isobutyl‐1‐methylxanthine (100 μM) and ICI‐118551 (37.5 pM). Cellular cAMP was measured by RIA (Coulter‐Immunotech, Krefeld, Germany).

Statistics

After testing for normal distribution (comparable variances on the means of each group), data were analysed by one‐way analysis of variance (ANOVA) with subsequent Dunnett's post‐hoc test for multiple comparisons (comparing all means against the control/reference mean) using GraphPad PRISM 8 (version 8.3.1, GraphPad Software Inc., USA).

Ethics

Animal experiments were approved by the animal ethics committee of the Government of Lower Franconia (Vote No. 55.2‐2531.01‐52/08, Experimental Animal Use and Care Committee, Government of Lower Franconia, Bavaria, Germany). The study of patient's biomaterials complies with the Declaration of Helsinki and was approved by the Medical Ethics Committee of the Medical Faculty of the University of Würzburg (Vote No. 186/07). Informed consent of the donors was obtained.

Results

Role of intra‐loop disulfide bridges in auto‐epitope conformation

The β₁‐EC_II contains three cysteines significantly determining loop conformation: C²⁰⁹ and C²¹⁵ form an intra‐loop disulfide bridge, while C²¹⁶ forms a bridge to β₁‐EC_I. ²⁹ In cyclopeptides, a possible aberrant bridge C²⁰⁹↔C²¹⁶ can alter the physiological conformation. To address whether this plays a role for binding and allosteric receptor activation by anti‐β₁EC_II‐abs, we synthesized two minimal (18‐mer) cyclopeptides representing AA residues 204–219 of the β₁‐EC_II, which had either C²¹⁵ (termed 18 C/C/S) or C²¹⁶ (termed 18 C/S/C) replaced by serine, thereby allowing either the physiological bridge C²⁰⁹↔C²¹⁵ or the aberrant bridge C²⁰⁹↔C²¹⁶ (Table S1). The reactivity of these cyclopeptides with stimulating anti‐β₁EC_II‐abs was assessed using sera of 99 rats having developed high titres of anti‐β₁EC_II‐abs and a cardiac DCM phenotype after repeated immunizations with β₁EC_II/GST fusion proteins. ¹⁹ , ²⁰ First, we investigated pre‐adsorption of these rat anti‐β₁EC_II‐abs to either 18‐mer cyclopeptide (at 40‐fold molar excess) prior to ELISA with linear 25‐mer β₁EC_II peptides. Binding of the large majority of sera (83/99 = 84%) was neutralized by 18 C/C/S (bearing the physiological bond C²⁰⁹↔C²¹⁵), but not by 18 C/S/C (bearing the aberrant bond C²⁰⁹↔C²¹⁶). Only four sera were neutralized by 18 C/S/C, and 12 rat sera could not be fully blocked with either cyclopeptide (Table 1).

Table 1.

Effect of intra‐loop disulfide bonds on the binding of anti‐β₁‐EC_II‐abs ^a

Neutralizing cyclopeptides (40 mol/mol IgG)	Effective pre‐adsorption of anti‐β1ECII rat sera (n) (≥90% reduction in ELISA reactivity with β1‐ECII peptides)
18‐mer C/C/S (C209↔C215)	83 (84%)
18‐mer C/S/C (C209↔C216)	4 (4%)
Not fully blocked	12 (12%)

^aDetermined by enzyme‐linked immunosorbent assay (ELISA) of IgG binding to a linear 25 AA β₁‐EC_II peptide. ¹² , ¹³

Because β₁‐AR‐mediated cAMP stimulation is the readout delineating the clinically relevant cardio‐pathogenic potential of anti‐β₁EC_II‐abs, ¹⁰ we next analysed neutralization of the cAMP stimulatory effect of rabbit anti‐β₁EC_II‐abs using HEK‐β₁E₁ cells co‐expressing human β₁‐AR with a cAMP FRET reporter. ²¹ In these cells, intracellular cAMP levels can be continuously monitored via the CFP/YFP emission ratio of the FRET sensor, and cAMP stimulation by anti‐β₁EC_II‐abs can be normalized to the maximal stimulation obtained with 1 μM (−)‐isoproterenol, measured in the same batch of cells. Rabbit anti‐β₁EC_II‐abs increased cAMP levels by about 18.8 ± 1.2% of the maximum achieved with the full agonist (−)‐isoproterenol. Pre‐absorption of rabbit anti‐β₁EC_II‐abs to 18 C/C/S (the 18‐mer cyclopeptide with the physiological bond C²⁰⁹↔C²¹⁵; 40‐fold molar excess) reduced the stimulatory effect of the antibodies by more than 50% (9.4 ± 2.2%; P < 0.01). By contrast, pre‐absorption of rabbit anti‐β₁EC_II‐abs to 18 C/S/C (the cyclopeptide with the aberrant bond C²⁰⁹↔C²¹⁶; 40‐fold molar excess) failed to block the antibody‐induced increases in intracellular cAMP (18.1 ± 2.0%, n.s.; Figure 1). In conclusion, inhibition of anti‐β₁EC_II‐induced β₁‐AR activation by β₁EC_II‐mimicking cyclopeptides requires the natural intra‐loop disulfide bridge C²⁰⁹↔C²¹⁵ and is disrupted by the non‐physiological C²⁰⁹↔C²¹⁶ bond.

Figure 1.

Neutralization of the stimulatory effects of rabbit anti‐β₁EC_II‐abs by pre‐incubation with cyclopeptides containing intra‐loop bonds C²⁰⁹↔C²¹⁵ (18 C/C/S) or C²⁰⁹↔C²¹⁶ (18 C/S/C): Stimulatory effects of polyclonal rabbit anti‐β₁EC_II‐abs on the human β₁‐AR coupled to a CFP/YFP‐FRET sensor for intracellular cAMP, normalized to maximal stimulatory effects obtained by 1 μM (−)‐isoproterenol. Rabbit anti‐β₁EC_II‐abs were pre‐absorbed with the indicated cyclopeptides (40 mol CP/mol IgG). Unblocked: effect of stimulating rabbit anti‐β₁EC_II‐abs alone. Data are given as mean ± SEM (n ≥ 5 per experiment; differences between the conditions were analysed by one‐way ANOVA with subsequent Dunnett's post‐hoc test for multiple comparisons; **P < 0.01).

β₁EC_II loop residues essential for proper conformation of the auto‐epitope

Even the 18‐mer cyclopeptide with the appropriate disulfide bridge yielded only incomplete inhibition, indicating that an optimal representation of the auto‐epitope conformation might require longer constructs with a less bulky C²¹⁶ substitution. Accordingly, a 22‐mer cyclopeptide was generated, representing the entire predicted β₁EC_II loop and having cysteine C²¹⁶ replaced by α‐aminobutyric acid (B), an artificial cysteine homologue not engaging in disulfide bridges. This optimized construct (22 C/C/B‐β₁, depicted in Figure 2) inhibited anti‐β₁EC_II antibody‐induced stimulation by ≥80% (Figure 3, first black column on the left) and served as a template to map further residues within β₁EC_II essential for the proper conformation of the auto‐epitope. Nine variants of 22 C/C/B‐β₁ were synthesized, each having a different non‐conserved AA (compared with the AA constituting the EC_II loop of the β₂‐AR) sequentially replaced by alanine (E202A, S203A, D204A, R207A, R208A, N211A, D212A, P213A, and K214A). Conserved AA residues essential for adrenoceptor function per se were omitted from the scan (for details, see Table S1).

Figure 3.

Neutralization of the functional effects of polyclonal rat anti‐β₁EC_II‐IgG by 22‐mer cyclopeptides mimicking the human β₁‐EC_II with non‐conserved amino acids (compared with the amino acids constituting the EC_II loop of the β₂‐AR) sequentially replaced by alanine: sera of 20 cardiomyopathic Lewis rats (immunized with β₁EC_II/GST fusion proteins) were pre‐absorbed with the indicated cyclopeptide mutants (40 mol CP/mol IgG, 4°C, 16 h). Grey bars: IgG binding to the linear 25 AA (199–223) β₁EC_II peptide as determined by ELISA (triplicates). Black bars: β₁‐AR‐mediated cAMP stimulation in HEK293 cells expressing native β₁‐AR functionally coupled to a cAMP FRET sensor. Decreases in ELISA reactivity (grey bars) or receptor activation/cAMP stimulation (black bars) following pre‐absorption with the different cyclopeptide mutants are shown, normalized to the values obtained without blocking cyclopeptides. Columns represent mean values ± SEM of n = 20 rat sera. Differences between the non‐mutated 22 C/C/B‐β₁ cyclopeptide and CP mutations were analysed by one‐way ANOVA with subsequent Dunnett's post‐hoc test for multiple comparisons; *P < 0.05; **P < 0.01; ***P < 0.001; †P < 0.0001. Internal negative control: non‐mutated 22C/C/D‐β₂ (sequence and alignment, see Table S1).

The reactivity of the 22‐mer C/C/B‐β₁ variants with clinically relevant stimulating anti‐β₁EC_II‐abs was determined by (i) pre‐absorption of autoantibody binding to the linear 25 AA β₁EC_II peptide (determined by ELISA) and by (ii) pre‐absorption of the receptor‐stimulating capacities of anti‐β₁EC_II‐abs (determined with cAMP FRET reporter cells). Sera of β₁EC_II/GST‐immunized cardiomyopathic rats served as a source for DCM‐relevant autoantibodies. Mean results obtained with sera from 20 individual rats are summarized in Figure 3, demonstrating that receptor binding (grey columns) and receptor stimulation (black columns) were efficiently pre‐absorbed (>80%) by all cyclopeptide mutants tested, except for those having disrupted an NDPK motif encompassing AA residues 211–214. The most disruptive mutation was P213A, which almost completely abolished the scavenger effect of the respective cyclopeptide (blocking capacity 22 C/C/B‐β₁ vs. P213A: 87.9 ± 3 vs. 7.2 ± 8%, P < 0.0001). The same was observed with a stimulating monoclonal mouse anti‐β₁EC_II antibody (Figure S1).

To confirm the clinical relevance of these results, comparable neutralization experiments were performed with stimulating anti‐β₁‐aabs isolated from selected DCM patients. Figure 4A shows a representative recording of cAMP reporter signals obtained with patient‐derived anti‐β₁‐aabs, which induced an increase in β₁‐AR‐mediated cAMP production of ≈30% of the maximal signal obtained with saturating concentrations (1 μM) of the full agonist (−)‐isoproterenol. That stimulatory effect was almost fully abolished by pre‐incubation with the (non‐mutated) cyclopeptide 22 C/C/B‐β₁, whereas the mutant P213A (earlier found ineffective after pre‐absorption of rodent anti‐β₁EC_II‐abs; Figure 3 and Figure S1) also failed to block the stimulatory effect of human anti‐β₁‐aabs (Figure 4A). The blocking profile of the entire set of mutated cyclopeptides on cAMP stimulation by human anti‐β₁‐aabs isolated from selected antibody‐positive DCM patients (male/female) is shown in Figure 4B.

Figure 4.

Neutralization of human anti‐β₁‐aabs from DCM patients by 22‐mer cyclopeptides mimicking β₁EC_II with non‐conserved amino acids (compared with the amino acids constituting the EC_II loop of the β₂‐AR) sequentially replaced by alanine: human anti‐β₁‐aabs (IgG fractions) were pre‐absorbed with the indicated cyclopeptide mutants (40 mol/mol IgG, 4°C, 16 h). Anti‐β₁‐aab‐induced cAMP production was measured in HEK293 cells expressing the native β₁‐AR functionally coupled to a cAMP FRET sensor. (A) Representative recordings of FRET ratios obtained upon addition of anti‐β₁‐aabs prepared from a male DCM patient followed by the maximal signal achieved with 1.0 μM of (−)‐isoproterenol (Iso). (B) Prevention of cAMP stimulation after pre‐absorption of patient anti‐β₁‐aabs. Results are normalized to the values without pre‐absorption. Columns represent mean ± SEM from at least three to four independent experiments with IgG prepared from different exemplary DCM patients (two men and one woman). Differences between the non‐mutated 22 C/C/B‐β₁ cyclopeptide and CP mutations were analysed by one‐way ANOVA with subsequent Dunnett's post‐hoc test for multiple comparisons; *P < 0.05; **P < 0.01; ***P < 0.001. Internal negative control: non‐mutated 22C/C/D‐β₂ (sequence and alignment, see Table S1).

In summary, our data suggest that the NDPK^211–214 motif within β₁‐EC_II is of paramount relevance for β₁‐AR‐induced cAMP stimulation by potentially cardio‐pathogenic anti‐β₁‐aabs. The stimulatory effect of human DCM‐associated autoantibodies appeared even more sensitive to disruptions of the NDPK^211–214 motif than the effect of anti‐β₁EC_II‐abs raised in rodents. Moreover, cAMP stimulation by human anti‐β₁‐aabs was also affected by the adjacent mutation R208A (Figure 4B vs. Figure 3, black columns), suggesting that the auto‐epitope targeted by human anti‐β₁‐aabs is slightly larger, encompassing the motif R (CY)NDPK^208–214. This finding is not unexpected because our rodents through repeated immunizations have undergone a maturation process in their adapted immune system, entailing increases in antibody avidity and epitope narrowing, ³⁰ which human DCM patients most probably have not undergone (explaining the generally lower levels and lower avidity of human anti‐β₁‐aabs ²⁵ ).

The black labels in Figure 2 summarize the results obtained in our experiments aiming at a further refinement of cyclopeptides intended as synthetic targets or therapeutic scavengers of stimulating anti‐β₁‐aabs. Minimal requirements for such agents encompass a core domain of eight essential AA ranging from positions 208 to 215 of the human β₁‐AR sequence, which harbours the essential intra‐loop bridge C²⁰⁹↔C²¹⁵. Aberrant disulfide bridging to C²¹⁶ must be precluded. The core domain should be embedded in a protein circle encompassing the entire β₁EC_II loop to be fully efficient.

Directed mutational analysis of the presumed auto‐epitope in intact human β₁‐ARs

Our above results regarding the auto‐epitope of stimulatory anti‐β₁EC_II‐abs led to the assumption that the generated neutralizing (scavenger) cyclopeptides represent competitive inhibitors faithfully mimicking the three‐dimensional structure of the auto‐epitope as presented on native β₁‐ARs. To confirm this assumption, we then investigated whether the very same AA residues essential for antibody neutralization by cyclopeptides would also be relevant for the interaction of stimulating anti‐β₁EC_II‐abs with full‐length human β₁‐ARs. For this purpose, we generated β₁‐AR constructs bearing point mutations of AA residues either presumed to be essential components of the core auto‐epitope (D212N and P213A) or flanking the core auto‐epitope while not essentially contributing to its formation (R207A and V219A). The mutant β₁‐ARs were expressed in HEK293 cells. Expression and functionality of the β₁‐AR mutants were ascertained (i) by ligand binding and (ii) by cAMP responses compared with the wild‐type β₁‐AR expressed under same conditions (Table S3).

To ensure reproducibility of the test agent anti‐β₁EC_II antibody, we then measured the cAMP responses induced via the mutant β₁‐ARs by two different monoclonal anti‐β₁EC_II‐abs (Mab 23‐6‐7 or Mab 13F6, raised in mice or rats, respectively ²⁶ ). Whereas the cardio‐noxious potential of anti‐β₁EC_II‐Mabs generated in mice (as Mab 23‐6‐7) has been clearly shown independently by many research groups, ³⁰ , ³¹ , ³² the clinical relevance of rat anti‐β₁EC_II‐Mabs remained to be demonstrated. Thus, according to our previously described strategy ¹¹ —after purification from the cell supernatant—we (intravenously) transferred Mab 13F6 into healthy rats every 4 weeks. Compared with animals receiving a rat IgG isotype control, rats receiving Mab 13F6 within (6 to) 9 months of Mab transfer developed a dilated cardiomyopathic phenotype (Figure S2), demonstrating its disease‐inducing potential.

To rule out effects of the mutations on β₁‐AR function per se, increases in cAMP were again normalized to the maximal stimulation achieved for each β₁‐AR mutant by the full agonist (−)‐isoproterenol (1 μM). As summarized in Figure 5, β₁‐ARs bearing mutations within the presumed auto‐epitope (D212N and P213A) were significantly (P < 0.0001) less sensitive to stimulation by rat or mouse anti‐β₁EC_II‐Mabs than the wild‐type β₁‐AR. In contrast, cAMP stimulation by rat or mouse anti‐β₁EC_II‐Mabs was not significantly (P ≥ 0.01) affected by mutations in the flanking regions (R207A and V219A), previously found to be not involved in the scavenger effect of cyclopeptides (Figures 2 and 3B). In conclusion, the same AA residues essential for antibody neutralization by cyclopeptides (Figures 3 and 4) appear also essential for the anti‐β₁EC_II‐induced activation of native human β₁‐AR (Figure 5). Consequently, cyclopeptides that faithfully mimic the auto‐epitope conformation relevant for allosteric β₁‐AR activation should also efficiently block the effects of stimulating anti‐β₁EC_II‐abs.

Figure 5.

Stimulation of mutant β₁‐ARs bearing point mutations within or flanking the presumed auto‐epitope of functional anti‐β₁EC_II antibodies: to ensure reproducibility of the test agent anti‐β₁EC_II antibody, HEK293 cells expressing human β₁‐AR with the indicated point mutations were exposed to monoclonal anti‐β₁EC_II‐abs raised in mice (mouse Mab 23‐6‐7, ²⁶ white columns) or rats (rat Mab 13F6, ²⁶ grey columns). cAMP levels were determined in the cell lysates by radio‐immunoassay. For each experiment, increases in cAMP following antibody exposure were normalized to the maximal responses achieved with 1.0 μM of the full agonist (−)‐isoproterenol. Data are given as mean ± SEM of at least six independent experiments per column. Differences between the wild‐type β₁‐AR and the indicated mutants were analysed by one‐way ANOVA with subsequent Dunnett's post‐hoc test for multiple comparisons; **P < 0.01; †P < 0.0001.

Discussion

We here provide a set of direct and indirect evidence delineating the minimal structure and conformation of the β₁‐AR epitope targeted by anti‐β₁‐aabs supposed to play an important role in various cardiovascular disorders. ⁹ Allosteric β₁‐AR activation by these autoantibodies is considered a severe risk factor if not a cause of CHF. ¹⁰ , ¹¹ It is long known that such stimulating aabs mostly target the β₁‐EC_II. ²³ However, attempts at further narrowing the target epitope have yielded ambiguous results: autoantibodies involved in Chagas cardiomyopathy are thought to target residues 201–205 ³³ ; in post‐partum cardiomyopathy, the autoantibody target has been placed at residues 200–210; and for DCM, a whole series of overlapping sequences have been published (residues 183–208, 197–202, 206–212, and 213–218 ²² , ²³ , ³³ ), encompassing almost the entire β₁‐EC_II and even portions of the adjacent transmembrane domain VI. Almost all these data have been obtained with linear peptides as probes. However, the surface of a protein is essentially a continuum of potential epitopes, but the fact that functional epitopes are often found to be discontinuous shows that mapping by linear peptide scanning may not be generally useful for comprehensive mapping or analysis of functional (energetic) epitopes. ³⁴ By now, it seems quite clear that disease‐relevant anti‐β₁‐aabs are only those that stimulate the β₁‐AR ²¹ and that such aabs bind to a conformational epitope, thereby stabilizing an active conformation and prolonging the active state of the β₁‐AR. ¹⁷ Consequently, stimulating anti‐β₁‐aabs poorly cross‐react with linear peptides, ¹² , ¹³ , ²⁴ which probably explains why epitope mapping with linear peptides has not yielded a clearer picture of the relevant β₁‐AR auto‐epitope(s). However, such a clearer picture is desperately needed for the development of specific diagnostics and therapeutics addressing this disease mechanism.

One crucial and potentially clinically relevant finding of our study is the identification of a minimal peptide structure that efficiently hinders receptor stimulation by DCM‐associated human anti‐β₁‐aabs in an experimental approach using cells expressing functionally coupled human β₁‐ARs. We have previously demonstrated that cyclopeptides mimicking that structure can stop progression of CHF and even partially revert the CHF phenotype in a human‐analogous rat model of autoimmune DCM. ¹⁹ , ²⁰ This therapeutic effect is conveyed not only by scavenging (and thus neutralizing) cardio‐pathogenic anti‐β₁EC_II‐abs but possibly also by induction of B‐cell tolerance. ³⁵ In support of this hypothesis, we observed in the above therapy model not only a reduction of autoantibody titres due to in vivo scavenging of circulating anti‐β₁EC_II‐abs but also a reduction of anti‐β₁EC_II‐secreting B cells. ¹⁹ While long‐lived plasma cells express very little or no immunoglobulins on the cell surface, ³⁶ B cells do and could thus also serve as targets of β₁EC_II‐CPs. To detect the few antigen‐specific memory B cells within splenocytes of CP‐treated vs. untreated immunized rats, in our previous study, we differentiated memory B cells into short‐lived plasma blasts by boosting the rats with β₁EC_II/GST‐FPs 3 days prior to the analysis of the spleens. Whereas long‐lived plasma cells were not targeted by β₁EC_II‐CPs, preventive as well as therapeutic application of the same CPs resulted in a ~ 80% reduction of the frequencies of splenocytes secreting anti‐β₁EC_II‐abs. ¹⁹ This finding indicates that in immunized anti‐β₁EC_II‐positive rats, repeated injections of β₁EC_II‐CPs may lead to impaired B‐cell receptor (BCR)‐mediated β₁EC_II‐specific memory B‐cell expansion (and differentiation into anti‐β₁EC_II‐producing plasma cells). Thus, cyclopeptides based on the criteria outlined in the present study may complement therapeutic strategies that exclusively aim at (non‐specific) autoantibody scavenging, such as extracorporeal IgG absorption ³⁷ or the systemic application of aptamers. ³⁸

Furthermore, the very same peptide core conveying to cyclopeptides the potency to prevent allosteric β₁‐AR activation by anti‐β₁EC_II‐aabs is obviously also a crucial prerequisite of the allosteric stimulation mechanism itself, which implies that the cyclopeptide actually mimics the 3D structure of the auto‐epitope that provides the allosteric trigger. The proposed antibody‐binding region as outlined in Figure 2 comprises the essential motif NDPK^211–214 framed by the equally essential disulfide bridge between cysteines C²⁰⁹↔C²¹⁵. In the intact receptor, this structure is constrained by another adjacent disulfide bridge linking C²¹⁶ to the transmembrane region next to EC_I. Both pairs of disulfide bridges are relevant for receptor function. ³⁹ , ⁴⁰ According to the 3D structure of the turkey β₁‐AR, ⁴¹ the assumed auto‐epitope is located at the C‐terminal end of the backward‐oriented α‐helix constituting β₁‐EC_II. Several features pre‐dispose the NDPK^211–214 motif as an autoantigen and allosteric trigger: N²¹¹, D²¹², and K²¹⁴ are prone to bond with the binding surface of the antibody, while P²¹³ is needed to interfere with the H‐bond formation to neighbouring helix residues, thereby bending the local α‐helix ⁴² and creating a unique structure for antibody recognition. The allosteric activation trigger is most probably provided by D²¹², the exchange of which to asparagine is known to cause paradoxical receptor activation by antagonistic ligands. ⁴³

Limitations

The fact that human anti‐β₁‐aabs are polyclonal and that the concentration of specific anti‐β₁‐aabs in the circulation of a patient is probably much lower than in specifically immunized anti‐β₁EC_II‐positive rodents represents a major challenge for a therapeutic use of the here described cyclopeptides in the future. Immunoglobulin G after maturation by its (antigen‐binding) HV regions assumingly recognize stretches of five (to eight) AAs as a target epitope. ²⁸ , ³⁴ The EC_II loop of the β₁‐AR is thought to be composed of 30 AAs (AA 195–225 of the receptor protein), and most functionally active human anti‐β₁‐aabs were shown to be directed against this segment of the β₁‐AR membrane protein, representing a readily accessible target on the cell surface. Thus, it seems conceivable that polyclonal human anti‐β₁‐aabs may target different epitopes (stretches of five to eight AAs) even within the 30 AAs that constitute β₁EC_II, which might result in diverging functional effects and/or downstream effects. ¹² Moreover, not only the specific target epitope (at the surface of a same target protein) but also the target protein itself (which might be sarcolemmal proteins, such as myosin or troponin, or myocyte membrane proteins such as β₁‐AR, β₂‐AR, β₃‐AR, M2‐muscarinergic, and/or angiotensin II AT₁ receptors), or cross‐reactions (or molecular mimicry ⁴⁴ ) between the target protein and other functionally relevant proteins may contribute to the diversity of functional effects and/or downstream effects of human aabs. In case of cross‐activation of β₁‐AR by human anti‐myosin aabs, ⁴⁵ β₁‐derived cyclopeptides would probably fail to hinder anti‐myosin‐aab‐induced stimulation of cardiac β₁‐AR. Assessment of the effect of human aabs on downstream signalling might be even further complicated by the co‐existence of aabs against different target proteins, which could then enhance (additive effect) or inhibit (opposing effect) downstream signalling and would require a panel of different epitope‐mimicking CPs to be therapeutically used on a case‐to‐case basis (e.g. human anti‐β₁‐aabs were shown to increase cAMP production, whereas human anti‐M2‐aabs inhibit cAMP production, ⁴⁶ or myocardial damage induced by anti‐β₁‐aabs in heart failure was found to be alleviated by human anti‐β₂‐aabs ⁴⁷ ).

Next steps and clinical perspective

As a consequence (and because of the pilot character of our study, testing only a few exemplary DCM patients), it will be necessary to fine‐map the target epitopes of anti‐β₁EC_II‐aabs of a larger number of CHF patients (with different aetiologies, e.g. DCM vs. ICM vs. hypertensive heart disease) to confirm our findings. Because human anti‐β₁‐aabs are polyclonal, their target epitope(s) might encompass not only the here unmasked structural motif in β₁‐EC_II, but also structural motifs in other extracellular domains of the human β₁‐AR, for example, in β₁‐EC_I, ¹⁰ , ²² which were not addressed in the present study. Moreover, in future (prospective) CHF studies, the patients should be systematically screened also for the presence of autoantibodies against other cardiac membrane proteins. The functional effects and the specific target epitopes of each of these aabs should then be analysed more in detail (e.g. for functional tests by adapting the here described FRET approach and for epitope fine‐mapping by adapting the here described Ala‐scan approach). In the end, this could lead to the identification of functionally relevant (targetable) key epitopes also in other G‐protein‐coupled membrane receptors. Fine‐mapping of functionally relevant key epitopes would allow to tailor epitope‐specific cyclopeptides that (i) act as autoantibody scavengers in the circulation and—in the light of the data obtained by β₁EC_II‐CP treatment of anti‐β₁EC_II antibody‐positive rats ¹⁹ —might equally (ii) impair BCR‐mediated autoantibody‐specific memory B‐cell expansion. Because the general risk of non‐specific immune (e.g. T‐)cell activation increases with the size of a given (cyclo‐)peptide from >25 AAs on, ⁴⁸ the exact knowledge of a key epitope required to neutralize (or scavenge) functional human aabs would allow to design much shorter (per se less immunogenic) cyclopeptides, as we started in the present study by testing 22‐mer instead of 25‐mer β₁EC_II‐CPs (used in our previous therapy study in anti‐β₁EC_II antibody‐positive rats ¹⁹ ). Even shorter CPs could be imagined, comprising 18, 16, or even less AAs.

With the availability of a wide panel of autoantibody‐specific (preferably short) scavenger CPs, a future therapeutic strategy could—in the end—comprise a kind of personalized approach in autoantibody‐associated CHF, that is, the injection of an individualized mixture of tailored cyclopeptides, depending on the autoantibody profile and the downstream ‘net effect' of cardio‐noxious vs. cardio‐protective autoantibodies in an individual CHF patient.

Conflict of interest

The University of Würzburg has filed for patent protection of the HEKβ₁E₁ cells (DE 102009 019578.5.2009) and for the methods and substances described in the present manuscript (EP 05 00 7056.4, WO 2006/103101 A2, EP 07 01 6637.6, WO 2009/027063 A2, WO 2010/61/2006.091, EP 11 01 2535.6). R.J. and M.J.L. are stockholders of the biotech company AdvanceCor GmbH (formerly CorImmun GmbH), München‐Martinsried, Germany. The remaining authors have no conflicts of interest to declare.

Funding

The present study was funded by the German Federal Ministry of Research and Education (Bundesministerium für Bildung, Wissenschaft, Forschung und Technologie and Bundesministerium für Bildung und Forschung), Grant GoBio No. 0315363 and Grant MolDiag No. 01ES0901.

Supporting information

Table S1 ‐ Amino‐acid sequences and denomination of cyclopeptides.

Table S2 ‐ Primer for site directed mutagenesis by nested PCR

Table S3 ‐ Characteristics of β₁AR‐mutants expressed in HEK293‐cells

Figure S1 ‐ Neutralization of a monoclonal mouse anti‐β₁EC_II‐antibody by 22mer‐cyclopeptides corresponding to the human β₁‐EC_II, each having a different non‐conserved amino‐acid (compared to the amino‐acids constituting the EC_II‐loop of the β₂‐AR) sequentially replaced with alanine.

Figure S2 ‐ Time course of left ventricular diameters from rats immunized against β₁EC_II or receiving a monoclonal rat anti‐β₁EC_II and corresponding control animals, determined by echocardiography.

Wölfel, A. , Sättele, M. , Zechmeister, C. , Nikolaev, V. O. , Lohse, M. J. , Boege, F. , Jahns, R. , and Boivin‐Jahns, V. (2020) Unmasking features of the auto‐epitope essential for β₁‐adrenoceptor activation by autoantibodies in chronic heart failure. ESC Heart Failure, 7: 1830–1841. 10.1002/ehf2.12747.