β₁-adrenergic receptor N-terminal cleavage by ADAM17; the mechanism for redox-dependent downregulation of cardiomyocyte β₁-adrenergic receptors

Abstract

β₁-adrenergic receptors (β₁ARs) are the principle mediators of catecholamine action in cardiomyocytes. We previously showed that the β₁AR extracellular N-terminus is a target for post-translational modifications that impact on signaling responses. Specifically, we showed that the β₁AR N-terminus carries O-glycan modifications at Ser³⁷/Ser⁴¹, that O-glycosylation prevents β₁AR N-terminal cleavage, and that N-terminal truncation influences β₁AR signaling to downstream effectors. However, the site(s) and mechanism for β₁AR N-terminal cleavage in cells was not identified. This study shows that β₁ARs are expressed in cardiomyocytes and other cells types as both full-length and N-terminally truncated species and that the truncated β₁AR species is formed as a result of an O-glycan regulated N-terminal cleavage by ADAM17 at R³¹↓L³². We identify Ser⁴¹ as the major O-glycosylation site on the β₁AR N-terminus and show that an O-glycan modification at Ser⁴¹ prevents ADAM17-dependent cleavage of the β₁-AR N-terminus at S⁴¹↓L⁴², a second N-terminal cleavage site adjacent to this O-glycan modification (and it attenuates β₁-AR N-terminal cleavage at R³¹↓L³²). We previously reported that oxidative stress leads to a decrease in β₁AR expression and catecholamine responsiveness in cardiomyocytes. This study shows that redox-inactivation of cardiomyocyte β₁ARs is via a mechanism involving N-terminal truncation at R³¹↓L³² by ADAM17. In keeping with the previous observation that N-terminally truncated β₁ARs constitutively activate an AKT pathway that affords protection against doxorubicin-dependent apoptosis, overexpression of a cleavage resistant β₁AR mutant exacerbates doxorubicin-dependent apoptosis. These studies identify the β₁AR N-terminus as a structural determinant of β₁AR responses that can be targeted for therapeutic advantage.

Graphical Abstract

1. INTRODUCTION

Catecholamines enhance the mechanical performance of the heart by activating cardiac β-adrenergic receptors βARs). While cardiomyocytes co-express β₁AR and β₂AR, the β₁AR is the predominant subtype (constituting ~75-80% of the total βAR population) and the principle driver of catecholamine-dependent sympathetic responses in the healthy heart ¹. β₁ARs provide hemodynamic support in the setting of acute stress by activating a G_s-cyclic adenosine monophosphate (cAMP)-protein kinase A (PKA) pathway that phosphorylates substrates that enhance excitation/contraction coupling ². Although β₁ARs also couple to cardioprotective G_s-independent mechanisms such as extracellular signal-regulated kinase (ERK), chronic/persistent β₁AR activation leads to a spectrum of changes (including cardiomyocyte hypertrophy/apoptosis, interstitial fibrosis, and contractile dysfunction) that contribute to the pathogenesis of heart failure ^3,4.

Like other G protein-coupled receptor (GPCR) superfamily members, βARs are characterized by an extracellular N-terminus, seven membrane-spanning α-helical segments that are joined by three extracellular loops and three intracellular loops, and an intracellular C-terminus. While βARs have served as prototypes in structural studies designed to elucidate the molecular dynamics of GPCR activation, studies to date have focused on specific amino acid side chains in the transmembrane helices that form the ligand binding pocket and effector docking sites on the βAR’s intracellular surface ⁵. The relatively short/unstructured βAR N-terminus - a region traditionally viewed as having a negligible role in mechanisms that contribute to receptor activation and regulation - has routinely been removed for these structural studies ⁵. However, our recent studies challenge the notion that the β₁AR N-terminus plays a negligible role in receptor activation/regulation showing that the β₁AR extracellular N-terminus is a target for glycan-regulated proteolytic cleavage and that this mechanism influences β₁AR signaling responses in cardiomyocytes ⁶. We showed that the β₁AR is detected in cardiomyocytes as both full-length and N-terminally truncated species, that N-terminal cleavage is controlled by an O-glycan modification on the β₁AR N-terminus at Ser³⁷/Ser⁴¹ (residues in the vicinity of predicted β₁AR N-terminal cleavage sites) and that the β₁AR extracellular N-terminus functions as a heretofore-unrecognized structural determinant of β₁AR activation ⁶. N-terminally truncated β₁ARs are signaling competent, but their signaling properties differ from that of the full-length β₁AR they display differences in their signaling bias to cAMP/PKA vs. ERK pathways - and only the N-terminally truncated form of the β₁AR constitutively activates AKT and confers protection against doxorubicin-dependent apoptosis in cardiomyocytes ^6,7.

Efforts to date to identify the O-glycan modifying enzymes and specific proteases that participate in the maturational processing of the β₁AR N-terminus have relied on in vitro approaches. On the basis of in vitro cleavage assays (using short peptides based upon the β₁AR N-terminal sequence as substrate and purified matrix metalloproteinase [MMP] or a disintegrin and metalloproteinase [ADAM] enzymes), Goth et al. concluded that the β₁AR N-terminus contains two major cleavage sites at R³¹↓L³² and P⁵²↓L⁵³. They showed that O-glycosylated and unmodified forms of these peptides can serve as substrates for many different MMP and ADAM family enzymes and that O-glycosylation influences the efficiency of peptide cleavage at some sites by some enzymes ⁸. It is important to note that this previous study interrogated the role of O-glycosylation by subjecting peptides to in vitro O-glycosylation reactions with GalNAc-T2, a reaction that adds a single GalNAc moiety to one or more sites on the peptide. Results obtained in this type of reductionist assay system may not necessarily recapitulate the proteolytic cleavage events that accompany maturational processing of full-length β₁ARs in a more physiologically relevant context (such as in differentiated cells such as a cardiomyocyte). First, this approach does not evaluate the role of more physiologically relevant extended and sialylated O-glycan structures that decorate the β₁AR N-terminus in a cellular context (and are predicted to undergo remodeling in the setting of inflammation, metabolic disorders, and/or cardiac hypertrophy ^9,10). Second, in vitro cleavage assays are not suited to evaluate cleavage events on membrane proteins, a situation where substrate and enzyme may localize to distinct membrane subdomains or (when tethered to the same membrane) be conformationally restrained. Third, the previous study did not resolve the importance of individual O-glycosylation sites in the control of β₁AR N-terminal cleavage. Finally, the in vitro approach cannot be used to explore the role of glycan-regulated β₁AR N-terminal cleavage events in the response to pathophysiologic cardiac injury. This study addresses these issues by mapping the O-glycosylation sites that regulate β₁AR N-terminal cleavage in cardiomyocytes, identifying ADAM17 as the protease required for β₁AR N-terminal cleavage in cardiomyocytes, and implicating ADAM17-dependent β₁AR N-terminal cleavage in an oxidative stress-dependent mechanism that calibrates cardiomyocyte catecholamine responsiveness.

2. MATERIALS AND METHODS

2.1. Materials

Antibodies were from the following sources: Rabbit polyclonal anti-βAR (ab3442, raised against residues 394-408 in human β₁AR) was from Abcam (Cambridge, MA). Mouse monoclonal anti-FLAG M2 and ant-β-actin were from Sigma-Aldrich (Saint Louis, MO). Rabbit monoclonal anti-cleaved caspase-3 (Asp175) (5A1E) was from Cell Signaling Technology (Danvers, MA). IRDye 800CW or 680RD goat anti-rabbit or goat anti-mouse IgG (H+L) were from LI-COR Biosciences (Lincoln, NE). GM6001, GI254023X (ADAM-10 inhibitor), MMP-2/MMP-9 inhibitor II, MMP-9/MMP-13 inhibitor II, doxorubicin, and H₂O₂ were from Sigma-Aldrich. MMP-2/MMP-3 inhibitor I and TAPI-2 were from Santa Cruz Biotechnology (Dallas, TX). GW280264X (ADAM-10/ADAM-17 inhibitor) was from AOBIOUS Inc (Hopkinton, MA). All other chemicals were reagent grade.

2.2. Chinese hamster ovary (CHO) pro5 and CHO-IdID cell culture and transfection

CHO-pro5 cells (obtained from American Type Culture Collection, Manassas, VA) were cultured in minimum essential medium Eagle-α modification (α-MEM), supplemented with 5% fetal bovine serum (FBS), 100 units/ml penicillin-streptomycin and 100 mM glutamine. CHO-IdID cells (generous gift from Dr. Monty Krieger, Massachusetts Institute of Technology) were cultured in Ham’s F-12 medium, supplemented with 5% FBS, 100 units/ml penicillin-streptomycin, 100 mM glutamine, either with or without galactose (Gal, 20 μM) and N-acetylgalactosamine (GalNAc, 200 μM). Cell transfections were performed in 60 mm culture dishes with 1 μg cDNA of β₁AR plasmid using jetOPTIMUS DNA Transfection Reagent (Polyplus Transfection, IIIkirch, France) according to manufacturer’s instructions.

2.3. Plasmids

A plasmid that drives expression of the human S⁴⁹R³⁸⁹-β₁AR harboring an N-terminal FLAG-tag was from Addgene (Watertown, MA). The various single residue substituted β₁AR mutant constructs used in this study were generated using the QuikChange mutagenesis system (Agilent Technologies) according to manufacturer’s instructions. The integrity of all constructs was confirmed by DNA sequencing (Genewiz, South Plainfield, NJ).

2.4. Cardiomyocyte culture and adenoviral infections

Cardiomyocytes were isolated from the ventricles of 2-day-old Wistar rats by a trypsin dispersion technique using a differential attachment procedure to enrich for cardiomyocytes followed by irradiation as described previously ^11,12. Methods to infect cardiomyocytes with adenoviruses that drive expression of WT or mutant forms of human β₁AR (Ad-WT-β₁AR and Ad-β₁AR-31H32A/52A53A prepared by Welgen Inc. Worcester, MA) were published previously ¹².

2.5. Immunoblotting

Immunoblotting was performed on cell extracts according to methods described previously ^6,7 or manufacturer’s instructions. Dilutions for primary and secondary antibodies were as follows: Abcam anti-β₁AR (ab3442) at 1:3000 followed by IRDye 800CW goat anti-rabbit IgG at 1:5000. Anti-FLAG M2 at 1:700 and anti-β-actin at 1:1000 followed by IRDye 800CW or 680RD goat anti-mouse IgG at 1:5000. Anti-cleaved caspase-3 at 1:700 followed by IRDye 800CW goat anti-rabbit IgG at 1:5000. Each panel in each figure represents results from a single gel exposed for a uniform duration using LI-COR Odyssey CLx imaging system (LI-COR Biosciences Lincoln, NE) for detection and image Studio Lite Ver 5.0 software for quantification of protein expression. All results were replicated in at least three experiments on separate culture preparations.

2.6. Statistical analysis

Results are shown as mean±SEM and were analyzed by using one-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison post hoc test; p values < 0.05 were considered statistically significant.

3. RESULTS

3.1. Maturational processing of the β₁AR N-terminus; proteolytic cleavage at R³¹↓L³²

A previous study concluded that the β₁AR N-terminus contains two major cleavage sites at R³¹↓L³² and P⁵²↓L⁵³ (Fig 1A) based upon in vitro cleavage assays using short peptides corresponding to the β₁AR N-terminal sequence and purified ADAM or MMP enzymes ⁸. Since the proteolytic cleavage events that can be identified in this type of reductionist assay system may not necessarily reflect proteolytic cleavage events that regulate the maturational processing of the full-length β₁AR in cells, we used a mutagenesis strategy to map the sites that are targets for proteolytic cleavage of full-length β₁ARs in cells.

Figure 1: R³¹↓L³² is the major site for N-terminal cleavage during maturational processing of the β₁AR in cells.

Panel A, Top: The β₁AR N-terminal sequence showing a single N-linked glycosylation site in a consensus Nx(T/S) sequence at position 15 (blue), the two known O-linked glycosylation sites at positions 37 and 41 (red) and the two putative cleavage sites at positions 31-32 and 52-53 (green). Panel A, Bottom: Schematic of the N-terminally Flag-tagged β₁AR mutants harboring single residue substitutions at one or both putative cleavage sites used in our studies. Panel B: Lysates prepared from CHO-Pro5 cells or cardiomyocytes (CM) that heterologously overexpress the WT-β₁AR or β₁AR mutants designed to be cleavage resistant were subjected to immunoblot analysis with the anti-β₁AR (raised against a β₁AR C-terminal epitope that recognizes both full-length and N-terminally-truncated β₁ARs) and anti-FLAG (that recognizes the N-terminal FLAG tag on full-length, but not truncated, β₁ARs) antibodies. Positions of the full-length and N-terminally truncated β₁ARs are denoted by filled and open triangles, respectively. A representative experiment is shown on top with results from 4 (CHO-Pro5 cell) or 7 (CM) separate experiments on separate culture preparations quantified at the bottom. Data are shown as mean±SEM with % of full-length β₁AR (upper band) expressed as a percent of total (full length + truncated) β₁AR. Data analysis for CHO-Pro5 cells was by ANOVA with post hoc Tukey’s test for multiple comparisons. Data analysis for CMs was by t-test. In each case, * p<0.05 compared with WT-β₁AR.

Fig 1B shows that the WT-β₁AR heterologously overexpressed in CHO-Pro5 cells is detected as two major bands on immunoblot analysis. Our previous studies established that the upper ~69KDa band (which is recognized by both an anti-β₁AR antibody directed against a β₁AR C-terminal epitope and anti-FLAG antibody which recognizes the N-terminal FLAG tag) is the full-length β₁AR and the ~55KDa lower band (which is recognized by anti-β₁AR antibody, but not by anti-FLAG) is an N-terminally cleaved form of the β₁AR ^6,7. Of note, while β₁AR constructs used in this study carry an N-terminal FLAG tag, the electrophoretic motility patterns of β₁AR constructs without the N-terminal epitope tag is quite similar (⁶ and data not shown).

The electrophoretic mobility patterns for the P52A/L53A-β₁AR (harboring substitutions designed to prevent cleavage at the P⁵²↓L⁵³ site) and the WT-β₁AR are identical, suggesting either that P⁵²↓L⁵³ is not a physiological cleavage site or that the P52A/L53A substitutions are not sufficient to prevent a 52-53 bond cleavage. In contrast, the R31H/L32A-β₁AR (or the β₁AR harboring double R31H/L32A-P52A/L53A substitutions) is detected as a single band that co-migrates with the full length WT-β₁AR. The observation that a R31H/L32A substitution prevents the formation of the N-terminally cleaved form of the β₁AR indicates that R³¹↓L³² is the major β₁AR cleavage site during maturational processing of full-length β₁ARs in CHO-Pro5 cells. The additional observation that WT-β₁ARs are detected as full-length and N-terminally cleaved forms in cardiomyocytes, and that R31H/L32A-P52A/L53 substitutions prevent β₁AR cleavage in cardiomyocytes (Fig 1B), indicates that a similar mechanism controls β₁AR maturational processing in cardiomyocytes.

3.2. β₁AR N-terminal cleavage is prevented by O-glycosylation

We previously reported that glycan modifications at Ser³⁷ and/or Ser⁴¹ prevent β₁AR N-terminal cleavage ⁶. β₁AR constructs were heterologously expressed in CHO-IdID cells to determine whether the protection afforded by O-glycosylation is specific for the cleavage event at R³¹↓L³². CHO-IdID cells are UDP-galactose/UDP-N-acetylgalactosamine 4-epimerase-deficient and therefore cannot synthesize UDP-Gal or UDP-GalNAc, the nucleotide sugars required for the addition of galactose (Gal) and N-acetylgalactosamine (GalNAc) to N- or O-linked oligosaccharides under conventional culture conditions with glucose as the sole sugar source ¹³. The unique advantage of studies in the CHO-IdID cell line is that the 4-epimerase defect can be bypassed and oligosaccharide synthesis can be fully restored by the addition of Gal and GalNAc to the culture medium.

Figure 2 shows that in the presence of Gal/GalNAc (under conditions that permit glycosylation), WT-β₁AR and P52A/L53A-β₁AR are detected as both full-length and N-terminally truncated species, whereas β₁AR constructs harboring single residue substitutions at the R³¹↓L³² cleavage site (R31H/L32A and R31H/L32A;P52A/L53A) are recovered as a single band corresponding to the full-length fully-glycosylated receptor. These results indicate that the truncated β₁AR species identified in CHO-IdID cells grown in the presence of Gal/GalNAc is formed as a result of N-terminal cleavage at R³¹↓L³².

Figure 2. β₁AR N-terminal cleavage at R³¹↓L³² is prevented by O-glycosylation; the β₁AR N-terminus also contains another glycan-regulated cleavage site.

IdID cells that are not transfected (NT) or transfected with WT-β₁AR or β₁AR mutants harboring single residue substitutions at putative cleavage sites were cultured with or without Gal (20 μM) and GalNAc (200 μM) as indicated. Lysates were probed with the anti-β₁AR antibody that recognizes an epitope on the β₁AR C-tail (both full-length and truncated forms of the receptor) and anti-FLAG.

Figure 2 also shows the banding patterns of WT and ‘cleavage-resistant’ β₁AR constructs in CHO-IdID cells grown without Gal/GalNAc (under conditions that do not support glycosylation). Under these conditions, WT and P52A/L53A-β₁ARs are detected by the β₁AR antibody (but not by anti-FLAG) primarily as a ~45kDa, non-glycosylated, N-terminally truncated species. While there is a small amount of a somewhat larger β₁AR species generated under these conditions, we previously noted that this likely represents a partially glycosylated species that is produced as a result of a low level of sugar scavenging from serum glycoprotein ⁶. Importantly, β₁AR constructs harboring single residue substitutions at the R³¹↓L³² cleavage site (the R31H/L32A and R31H/L32A;P52A/L53A-β₁ARs mutants) are detected as both a full-length un-glycosylated β₁AR species (recognized by both anti-β₁AR and anti-FLAG antibodies) and ~45kDa un-glycosylated/N-terminally cleaved species (recognized by anti-β₁AR, but not anti-FLAG). The observation that 31H32A and 31H32A/52A53A substituted β₁ARs can be cleaved under conditions that prevent glycosylation indicates that the β₁AR N-terminus also contains an additional glycan-regulated cleavage site that is distinct from the R³¹↓L³² site.

3.3. An O-Glycan modification at Ser⁴¹ prevents β₁-AR cleavage at a site that is distinct from R³¹↓L³²

We previously implicated O-glycan modifications at Ser³⁷ and/or Ser⁴¹ as post-translational modifications that prevent β₁AR cleavage ⁶. The previous studies did not parse possible independent effects of O-glycan modifications at position 37 vs. 41. This may be relevant since the Ser⁴¹ O-glycosylation site (LVPAS³⁷PPAS⁴¹L) is the site that conforms more closely to an MMP consensus cleavage sequence (PxS↓L, the hallmark features of an MMP cleavage site consisting of a Pro in the P3 position, small amino acid such as Gly, Ala, or Ser in the P1 position, and a Leu in the P1’ position – the +1 position C-terminal to the scissile peptide bond ¹⁴) and in vitro studies show that a short peptide based residues 23-46 of the β₁AR N-terminus is cleaved at the Ser⁴¹-Leu⁴² scissile bond by several MMPs (MMP-2, -7, -8, -9, -12, and -13 ⁸). We generated β₁AR mutants harboring single residue substitutions at each position (S37A-β₁AR and S41A-β₁AR; as schematized in Fig 3A) to test the hypothesis that S⁴¹ is the major target for the O-glycan modifications that regulate β₁AR N-terminal cleavage in cells.

Figure 3. An O-Glycan modification at S⁴¹ prevents β₁-AR cleavage at an MMP/ADAM consensus motif at S⁴¹↓L⁴².

Panel A: Schematic of the β₁AR mutants harboring single residue substitutions at O-glycosylation sites (in red) or cleavage sites (in green). Note: To simplify the cumbersome nomenclature, the 31H32A/52A53A-β₁AR construct is labeled β₁AR-31/52 in this and subsequent figures. Panel B: Lysates from CHO-Pro5 cells that were not transfected (NT) or that heterologously overexpress the indicated WT or mutant β₁AR constructs were subjected to immunoblot analysis with anti-β₁AR and anti-FLAG antibodies. A representative experiment is shown on top with results from 6-14 separate experiments on separate culture preparations quantified at the bottom. Data are shown as mean±SEM, with % of full length β₁AR (upper band) expressed as a percent of total (full length + truncated) β₁AR. Data analysis was by ANOVA with post hoc Tukey’s test for multiple comparisons. * p<0.05 compared with WT-β₁AR; ¤ p<0.05 compared with β₁AR-S37A; † p<0.05 compared with β₁AR-31/52; # p<0.05 compared with β₁AR-31/52-S37A. Panels C and D: Lysates from CHO-Pro5 cells that were not transfected (NT) or that heterologously overexpress the indicated WT or mutant β₁AR constructs were subjected to immunoblot analysis with anti-β₁AR and anti-FLAG antibodies. The data is representative of results in 5-6 separate experiments. Note: Under certain experimental conditions (as in Panel D) a non-specific band that migrates slightly slower than the full-length β₁AR can be detected.

Fig 3B shows that an S37A substitution alone leads to a modest increase in β₁AR electrophoretic mobility (the effect being most evident when comparing the mobilities of the N-terminally truncated WT- vs. S37A-β₁AR species); the S37A substitution also leads to a significant (albeit modest) 21.7% decrease in the abundance of the full-length β₁AR species. These results indicate that the Ser at position 37 is a target for an O-glycan modification, but this O-glycan modification plays a relatively minor role to prevent β₁AR N-terminal cleavage. Fig 3 also shows that the S37A substitution results in a small but significant decrease in the abundance of full-length receptors when introduced into the 31H32A/52A53-β₁AR backbone (the magnitude of this decrease being ~50% of what is observed when the S37A substitution is introduced into the WT-β₁AR backbone). These results support the conclusion that an O-glycan modification at Ser³⁷ plays a significant - albeit minor - role to prevent β₁AR N-terminal cleavage at both R³¹↓L³² and some other cleavage site.

The effect of a single S41A substitution is considerably more striking. Fig 3B shows that the S41A substitution results in a marked increase in the electrophoretic mobility of the N-terminal truncated β₁AR species and it results in a profound decrease in the abundance of both full-length WT and 31H32A/52A53A-β₁ARs (56.8% and 69.7% respectively). These results identify Ser⁴¹ as a major β₁AR N-terminal O-glycosylation site and implicate the O-glycan modification at Ser⁴¹ as the dominant regulator of β₁AR N-terminal cleavage at a site that is distinct from the R³¹↓L³² cleavage site.

We generated an additional set of mutants in which the S41A substitution was introduced along with single residue substitutions designed to disrupt cleavage at a site adjacent to this residue (with PAS⁴¹↓LLP mutated to PAA⁴¹AAP) as a strategy to more precisely map this second cleavage site. Fig 3C shows that when the S⁴¹LLP sequence is replaced by A⁴¹AAP in the 31H32A/52A53A-β₁AR backbone, the β₁AR becomes cleavage resistant; in contrast, a construct with A⁴¹AA substitutions in the 52A53A-β₁AR backbone is cleaved similar to the WT-β₁AR (data not shown). These results support the conclusion that S⁴¹↓L⁴² is a second β₁AR N-terminal cleavage site and that an O-glycan modification at Ser⁴¹ prevents cleavage at this second site.

Finally, we generated a β₁AR mutant that contains alanine substitutions at both the S⁴¹ O-glycosylation site and the adjacent cleavage site - but retains the R³¹↓L³² cleavage site - to determine whether the O-glycan modification at S⁴¹ exerts a more long-range effect on β₁AR N-terminal cleavage at R³¹↓L³². Figure 3D shows that the S41A/L42A/L43-β₁AR is detected as both a full-length and N-terminally truncated species, but the percent of the S41A/L42A/L43A-β₁AR that is recovered as a full-length species is decreased by 25.6 ±3.2% when compared to the WT-β₁AR (p<0.05, n=6). Collectively, these results indicate that O-glycosylation at S⁴¹ prevents cleavage at the adjacent S⁴¹↓L cleavage site and it also exerts a long-range effect to inhibit cleavage at the more distal R³¹↓L³² site.

3.4. The β₁AR N-terminus is cleaved at R³¹↓L³² and S⁴¹↓L⁴² by ADAM17

The previous in vitro cleavage assays that identified MMP2, MMP3, and ADAM17 as enzymes that cleave a short peptide based upon residues 23-46 in the β₁AR N-terminus at R³¹↓L^{32 8} and results reported herein that identify PAS⁴¹↓LLP (a sequence that conforms to an MMP consensus cleavage motif) as a bone fide cleavage site on the β₁AR N-terminus provided the rationale to test the hypothesis that one or both of the cleavage sites on the N-terminus of the full-length β₁AR is a target for cleavage by an MMP family enzyme in cells.

Fig 4A shows that treatment with GM6001 (a pan-MMP inhibitor) leads to increased accumulation of the full length WT-β₁AR in association with a decrease in the abundance of the N-terminally truncated species; cleavage resistant 31H32A/52A53A-β₁ARs are not influenced by the GM6001 treatment. Similar results were obtained in CHO-Pro5 cells and cardiomyocytes, indicating that the R³¹↓L³² N-terminal cleavage event that accompanies maturational processing of full-length β₁ARs in both CHO-Pro5 cells and cardiomyocytes involves an MMP activity.

Figure 4. The β₁AR N-terminus is cleaved at R³¹↓L³² by ADAM17.

Lysates from CHO-Pro5 cells that heterologously overexpress WT-β₁AR or the 31H32A/52A53A-β₁AR mutant (labeled β₁AR-31/52) that were treated for 24 hrs with vehicle or 10 μM GM6001 (Panel A) or vehicle, 10 μM GM6001, 3 μM ADAM10/17 inhibitor, 400 nM ADAM10 inhibitor, or 20 μM TAPI-2 as indicated (Panel B) were subjected to immunoblot analysis with anti-β₁AR and anti-FLAG antibodies. For cardiomyocytes (Panels A and B, right), both Ad-WT-β₁AR or Ad-β₁AR-31/52 infections and drug treatments were initiated on culture day 1, with lysates prepared for immunoblot analysis on culture day 4. In each panel, a representative experiment is shown on top and data are quantified as described in Figure 3 on the bottom (* p<0.05 compared with vehicle control).

Additional experiments with a panel of compounds specific for individual MMP family enzymes identifies a specific role for ADAM17 in CHO-Pro5 cells. Specifically, the effect of GM6001 to increase the accumulation of full length WT-β₁ARs (and decrease the abundance of the N-terminally truncated species) is mimicked by an ADAM10/17 inhibitor and TAPI-2, but not a specific inhibitor of ADAM10 (Fig 4B) or MMP-2/3, -2/9, or 9/13 (data not shown). The effect of GM6001 to increase the accumulation of full length WT-β₁ARs (and decrease the abundance of the N-terminally truncated species) also is mimicked by the ADAM10/17 inhibitor in cardiomyocytes (Fig 4B).

We used a similar strategy with β₁AR constructs harboring a S41A substitution to identify the enzyme activity that cleaves the β₁AR N-terminus at PAS⁴¹↓LLP – the site protected by the Ser⁴¹ O-glycan modification. Fig 5A shows that the cleavage event facilitated by the Ser⁴¹ glycosylation defect in the 31H32A/52A53A-β₁AR backbone is blocked by GM6001 and Fig 5B shows that it is also blocked by the ADAM10/17 inhibitor and TAPI-2, but not the specific inhibitor of ADAM10; MMP-2/3, -2/9, or 9/13 inhibitors also did not prevent this cleavage (data not shown). These results indicate that the β₁AR N-terminus is a target for cleavage by a single enzyme, namely ADAM17, but the precise site on the β₁AR N-terminus that is cleaved by ADAM17 is regulated by the type and position of the O-glycan attachment on the β₁AR N-terminus (i.e., by substrate); S⁴¹↓L⁴² cleavage is prevented by an O-glycan modification at Ser⁴¹.

Figure 5. The β₁AR N-terminus is cleaved at S⁴¹↓L⁴² by ADAM17.

Lysates from CHO-Pro5 cells that heterologously overexpress 31H32A/52A53A-β₁AR or 31H32A/52A53A-S41A-β₁AR mutants (for simplicity labeled as β₁AR-31/52 or β₁AR-31/52-S41A) that were treated for 24 hr with vehicle or 10 μM GM6001 (Panel A) or vehicle, 10 μM GM6001, 3 μM ADAM10/17 inhibitor, 400 nM ADAM10 inhibitor, or 20 μM TAPI-2 as indicated (Panel B) were subjected to immunoblot analysis with anti-β₁AR and anti-FLAG antibodies.

3.5. The mechanism underlying redox-inactivation of cardiomyocyte β₁ARs; N-terminal truncation at R³¹↓L³² by ADAM17.

We recently showed that oxidative stress (resulting from treatment with either H₂O₂ or doxorubicin) leads to a decrease in β₁AR expression and catecholamine responsiveness in cardiomyocytes ⁷. The mechanism underlying this form of β₁AR regulation remains uncertain. Our previous study showed that the H₂O₂-dependent decrease in β₁AR expression can be prevented by pharmacologic inhibitors of novel protein kinase C (PKC) activity, but not by inhibitors of PKA or Src. ⁷. A nPKC-regulated mechanism that might contribute to the control cardiomyocyte β₁AR expression was not identified. The notion that an N-terminal cleavage event might contribute to redox-inactivation of β₁ARs also was not considered. On the basis of literature identifying phorbol-12-myristat-13-acetate (PMA, a pharmacologic activator of PKC) as one of the strongest activators of ADAM17 sheddase activity ¹⁵, we examined whether redox-inactivation of β₁ARs is attributable to an N-terminal cleavage event mediated by ADAM17.

Fig 6 shows that H₂O₂ decreases the abundance of full-length WT-β₁ARs, but the 31H32A/52A53A-β₁AR mutant is not affected by this treatment. The H₂O₂-dependent decrease in full-length WT-β₁AR expression is prevented by GM6001, the ADAM10/17 inhibitor, and TAPI-2. These results implicate an ADAM17-dependent β₁AR N-terminal cleavage at R³¹↓L³² in the mechanism underlying the H₂O₂-dependent downregulation of full-length β₁AR expression in cardiomyocytes.

Figure 6. The mechanism for H₂O₂-dependent downregulation of cardiomyocyte β₁ARs; N-terminal cleavage at R³¹↓L³² by ADAM17.

Cardiomyocytes were infected with adenoviruses that drive expression of WT-β₁AR or the 31H32A/52A53A-β₁AR mutant (labeled β₁AR-31/52) on culture day 1, pretreated for 48 hr (starting on culture day 3) with vehicle, ADAM10/17 inhibitor (3 μM), TAPI-2 (20 μM), or GM6001 (10 μM) as indicated, and then challenged for 1 hr with 0.1 mM H₂O₂ on culture day 5. A representation experiment is shown on top, with results for 4 (WT-β₁AR) or 3 (β₁AR-31/52) separate experiments on separate cardiomyocyte culture preparations quantified (as described in Fig 3) at the bottom. *, p<0.05 compared with β₁AR; ¤, p<0.05 compared with H₂O₂+GM6001-treated β₁AR or GM6001-treated β₁AR.

Figure 7 shows that doxorubicin treatment leads to a decrease in the abundance of the full-length WT-β₁AR. Again, this response is attributable to ADAM17-dependent cleavage at R³¹↓L³² since it is prevented by 31H32A/52A53A substitutions in the β₁AR N-terminus or treatment with GM6001, the ADAM10/17 inhibitor, and TAPI-2.

Figure 7. The mechanism for doxorubicin-dependent downregulation of cardiomyocyte β₁ARs; N-terminal cleavage at R³¹↓L³² by ADAM17.

Infections with adenoviruses that drive expression of WT-β₁AR or the 31H32A/52A53A-β₁AR mutant (labeled β₁AR-31/52) was performed on culture day 1, with treatment with doxorubicin (Dox, 10 uM) in the presence of vehicle, ADAM10/17 inhibitor (3 μM), TAPI-2 (20 μM), or GM6001 (10 μM) initiated on culture day 3 and continued for 48 hrs. Lysates were subjected to immunoblot analysis with anti-β₁AR and anti cleaved Caspase-3 antibodies (with actin serving as a loading control). Representative immunoblots are depicted at the top with results from 4 or 5 separate experiments on separate cardiomyocyte culture preparations quantified (as described in Fig 3) at the bottom. * p<0.05 compared with β₁AR; ¤ p<0.05 compared with Dox+GM6001-treated β₁AR or GM6001-treated β₁AR. Results for cleaved caspase-3 in doxorubicin-treated β₁AR-31/52 cultures are normalized to values in doxorubicin-treated WT-β₁AR cultures. n=4, # p<0.05 compared with Dox-treated WT-β₁ARs.

We previously reported that heterologous overexpression of an N-terminally truncated form of the β₁AR protects against doxorubicin-induced apoptosis in cardiomyocytes ⁷. Fig 7 shows that heterologous overexpression of 31H32A/52A53A-β₁AR (a construct that cannot be cleaved to form the cardioprotective N-terminally truncated β₁AR species) exacerbates doxorubicin-dependent apoptosis (tracked as an increase in caspase 3 cleavage). This result lends further support to the conclusion that β₁AR N-terminal truncation constitutes a mechanism that confers protection against doxorubicin-induced apoptosis.

4. DISCUSSION

Cardiomyocyte βARs are among the most intensely studied members of the GPCR superfamily primarily because they control a large number of physiologically important processes that impact on the pathogenesis of cardiovascular disease (and hence constitute clinically important targets for drug discovery). Studies to date implicate cardiomyocyte β₁ARs in the activation of cellular responses required for normal physiological control of cardiac contractility as well as responses that contribute to the pathogenesis of cardiac arrhythmias, ventricular remodeling, and the evolution of heart failure. However, our current understanding of the molecular basis for βAR signaling responses derives from literature predicated on the assumption that β₁ARs signal exclusively as full-length receptor proteins. While evidence that the β₁AR N-terminus can serve as a target for glycosylation and proteolytic cleavage was published over 10 years ago ¹⁶, the evidence that an O-glycan-regulated cleavage event leads to the generation of distinct molecular species that differ in their coupling to downstream effector responses is very recent ^6,7. We showed that β₁ARs are detected as both full-length and N-terminally truncated species in model cell lines and cardiomyocytes, that N-terminal cleavage is prevented by O-glycan modifications at Ser³⁷ and/or Ser⁴¹ (whereas the N-glycan attachment at Asp¹⁵ does not influence β₁AR N-terminal cleavage), and that the N-terminally truncated form of the β₁AR constitutively activates a Gi-AKT pathway that confers protection against doxorubicin-induced apoptosis ^6,7. This study extends the analysis by mapping the precise proteolytic cleavage sites on the β₁AR N-terminus and the protease activity responsible for β₁AR N-terminal processing in cardiomyocytes.

This study identifies R³¹↓L³² as the major cleavage site on the β₁AR N-terminus; the N-terminally truncated β₁AR species generated during the maturational processing of β₁ARs in both CHO cells and cardiomyocytes results from cleavage at R³¹↓L³². While studies in IdID cells indicate that a glycan modification prevents cleavage at this site, the nature of this control remains uncertain. The observation that β₁ARs are expressed as a heterogeneous population of both full-length and N-terminally truncated receptor species could suggest that β₁ARs (like most glycoproteins) are synthesized as glycoforms – that inherent inefficiencies in the enzymatic pathways for glycoprotein biosynthesis results in the generation of β₁ARs variants that differ either in the sites that carry glycan attachments or the composition of the glycan attachment at any individual glycosylation site. Insofar as there is likely to be some specificity in glycan-regulation of β₁AR cleavage, this heterogeneity in β₁AR O-glycosylation could lead to the generation of receptor subpopulations that differ in their susceptibility to cleavage at the R³¹↓L³² scissile bond. In this context, chamber-specific differences in glycosylation-associated gene (glycogene) expression and the cardiac glycome ¹⁰ (which would result in a single protein, identical in sequence, displaying different O-glycosylation patterns in different cells) could underlie our previous observation that atrial tissue is relatively enriched in the N-terminally truncated form of the β₁AR compared to ventricular tissue ⁶. Our studies also establish that R³¹↓L³² is not the only glycan-regulated cleavage site on the β₁AR N-terminus. The observation that the R31H/L32A-P52A/L53A mutant (that is resistant to cleavage at the R³¹↓L³² site) is cleaved in IdID cells grown in the absence of Gal/GalNAc provided the rationale for additional studies that used a mutagenesis approach to map a second cleavage site on the β₁AR N-terminus to S⁴¹↓L⁴². Of note, cleavage at this site (which is C-terminal to the serine residue identified in this study as the major β₁AR N-terminal O-glycosylation site) is completely abrogated by the O-glycan modification at S⁴¹.

In vitro studies suggest that the β₁AR N-terminus contains as many as four potential cleavage sites that are targets for cleavage by many different MMP and ADAM family enzymes - and that individual sites show differential sensitivities to cleavage by various MMP and ADAM enzymes ⁸. However, our cell-based experiments indicate that β₁AR cleavage is restricted to two sites, R³¹↓L³² and S⁴¹↓L⁴², that cleavage at both sites requires ADAM17 activity, and that the precise cleavage product generated from the full-length β₁AR is influenced by an O-glycan attachment at S⁴¹. Of note, our studies assessed the consequences of S41A substitution that completely eliminates O-glycosylation at position 41. It is tempting to speculate that metabolic derangements associated with diabetes or various disease-induced changes in glycogene expression could lead to more subtle changes in β₁AR N-terminal O-glycosylation and that this might be sufficient to enhance β₁AR N-terminal cleavage at S⁴¹↓L⁴². While there is considerable evidence that glycosylation patterns change in parallel with changes in cellular metabolism in certain cancers (with these altered glycosylation patterns on certain serum or tumor tissue proteins serving as a biomarkers for patient prognosis or response to treatment ¹⁷), the notion that the metabolic derangements associated with diabetes also lead to changes in protein O-glycosylation (either globally on membrane proteins in general or specifically on the β₁AR) has never been considered.

This study shows that cleavage at both R³¹↓L³² and S⁴¹↓L⁴² requires ADAM17 activity. The classical paradigm for GPCR signaling places MMP and ADAM family enzymes downstream from GPCR activation in the mechanism for GPCR crosstalk to epithelial growth factor receptors. According to this model, an agonist-activated GPCR stimulates an MMP activity that cleaves an EGFR ligand (such as heparin binding EGF, transforming growth factor α, or ampiregulin) from a membrane precursor, resulting in ligand-mediated autocrine and/or paracrine transactivation of EGFRs. This form of cross-talk has been described for cardiac β₁ARs, where a β₁AR-dependent EGFR transactivation pathway (involving MMP-dependent release of HB-EGF) confers cardioprotection under conditions of catecholamine toxicity ^4,18. Of note, ADAM17 is one of many MMP/ADAM family enzymes that have been implicated in various GPCR-EGFR transactivation pathways ¹⁹, but the precise MMP or ADAM family enzyme(s) that acts downstream from the agonist-activated cardiomyocyte β₁AR has never been identified. Our study identifies a relationship between the β₁AR and ADAM17, but this involves a novel upstream role for ADAM17 in the maturational processing of cardiac β₁ARs. While we did not identify a mechanism to account for the specificity in β₁AR processing by ADAM17 (and not other MMP or ADAM family enzymes), it is worth noting that the mature form of ADAM17 is reported to localize to cholesterol-enriched membrane subdomains (lipid rafts)^20-22. An ADAM17 sheddase activity sequestered in lipid rafts would be in close proximity to the cardiomyocyte β₁AR, which also localizes to this subcellular compartment ²³. Finally, this study implicates ADAM17 as the enzyme that cleaves the β₁AR N-terminus. Studies to determine whether ADAM17 plays a dual role in both the maturational processing of β₁ARs and their downstream signaling to EGFR transactivation, or whether these actions are mediated by distinct MMP/ADAM family members that can be pharmacologically discriminated, are the focus of ongoing studies.

We previously reported that oxidative stress leads to β₁AR downregulation and decreased catecholamine responsiveness in cardiomyocytes. Our previous studies implicated a nPKC activity in this mechanism, but the exact process underlying the redox-dependent loss in β₁AR responsiveness was not identified. This study shows that redox-dependent β₁AR downregulation is due to β₁AR N-terminal cleavage at R³¹↓L³² by ADAM17. Several obvious molecular mechanisms might contribute to redox-induced ADAM17-dependent β₁AR cleavage and should be considered. First, there is evidence that ADAM17 activity is tightly regulated through changes in its subcellular localization. The bulk of the mature ADAM17 enzyme localizes to an intracellular compartment under basal conditions; only a small pool of the mature enzyme is available at the cell surface ²⁴. The observation that PKC activation leads to an increase in ADAM17 expression and activity at the cell surface ²⁴ (in the context of our previous observation that the H₂O₂-dependent decrease in β₁AR expression requires nPKC activity) could suggest that a ROS-activated nPKC isoform regulates ADAM17 trafficking, enhances its cell surface expression, and thereby triggers β₁AR downregulation. Second, ADAM17 is a redox-sensitive enzyme; it contains two evolutionarily conserved cysteinyl sulfhydryl groups in its extracellular disintegrin domain that serve as redox-sensors. The observation that oxidation of these cysteines is required for maximal ADAM17 activity ²⁵ provides a rationale for future studies that focus on whether β₁AR cleavage is due to a redox-dependent increase in ADAM17 activity (as has been described in various other cells types ^26-28). Finally, while the precise structural requirements for ADAM17 substrate recognition remain uncertain, a redox-dependent change in β₁AR secondary structure (or its O-glycosylation) could in theory serve to render the β₁AR N-terminus a better substrate for ADAM17-dependent cleavage.

Finally, this study builds upon recent evidence that β₁ARs are one of a small number of GPCRs that undergo limited N-terminal proteolysis (i.e., ectodomain shedding). In the case of the β₁AR, we previously showed that N-terminal proteolysis provides a mechanism to regulate coupling to downstream signaling pathways; N-terminally truncated (but not full-length) β₁ARs constitutively couple to a cardioprotective AKT signaling pathway that protects against doxorubicin-dependent apoptosis ⁷. Results reported herein showing that heterologous overexpression of a cleavage-resistant form of the β₁AR (a maneuver that prevents the formation of the cardioprotective N-terminally truncated β₁AR species) exacerbates doxorubicin-dependent apoptosis are consistent with this formation. The observation that β₁AR ectodomain cleavage is specifically mediated by ADAM17 exposes a novel role for this enzyme in the control of cardiomyocyte β₁AR responsiveness. ADAM17 (or TACE, TNFα converting enzyme) was first identified as a membrane-anchored metalloprotease that releases bioactive TNFα from cells; it subsequently was characterized as a sheddase for a large number of membrane-anchored growth factors, adhesion molecules, cytokines and cytokine receptors that contribute to the pathogenesis of a wide range of clinical disorders, including cancers, various inflammatory states, and neurodegenerative disorders ²⁹. However, the notion that an ADAM17 inhibitor might be useful in the treatment of these disorders must be balanced by a consideration of ADAM17’s cardiac action. Early studies established that ADAM17 is critical for normal heart development and valvulogenesis and more recent studies show that ADAM17 is expressed in the adult heart, its expression is increased in cardiac tissue from patients with myocarditis or dilated cardiomyopathy ^30,31, that cardiomyocyte-specific ADAM17 expression is required for post-ischemic angiogenesis and optimal post-MI myocardial repair ³², and that ADAM17 expression affords protection in the setting of pressure overload ³³. Our studies identify yet another cardioprotective role for ADAM17, as the enzyme required for the generation of cardioprotective N-terminally truncated β₁ARs. This raises yet an additional concern that therapies that inhibit ADAM17 activity for the treatment of cancer or inflammation may have adverse effects in the heart and that off-target cardiotoxicity might limit their usefulness in the clinic.

Highlights:

• The β₁AR N-terminus functions as a structural determinant of β₁AR responses in cardiomyocytes.

• The β₁AR N-terminus is cleaved at R³¹↓L³² by ADAM17.

• β₁ARs contain a second O-glycan-regulated ADAM17-dependent N-terminal cleavage site at S⁴¹↓L⁴².

• β¹AR inactivation during oxidative stress is due to ADAM17-dependent cleavage at R³¹↓L³².

• N-terminally truncated β₁ARs afford protection against doxorubicin-dependent apoptosis