The deubiquitinase ubiquitin–specific protease 20 is a positive modulator of myocardial β1-adrenergic receptor expression and signaling

Reversible ubiquitination of G protein–coupled receptors regulates their trafficking and signaling; whether deubiquitinases regulate myocardial β1-adrenergic receptors (β1ARs) is unknown. We report that ubiquitin-specific protease 20 (USP20) deubiquitinates and attenuates lysosomal trafficking of the β1AR. β1AR-induced phosphorylation of USP20 Ser-333 by protein kinase A-α (PKAα) was required for optimal USP20-mediated regulation of β1AR lysosomal trafficking. Both phosphomimetic (S333D) and phosphorylation-impaired (S333A) USP20 possess intrinsic deubiquitinase activity equivalent to WT activity. However, unlike USP20 WT and S333D, the S333A mutant associated poorly with the β1AR and failed to deubiquitinate the β1AR. USP20–KO mice showed normal baseline systolic function but impaired β1AR-induced contractility and relaxation. Dobutamine stimulation did not increase cAMP in USP20–KO left ventricles (LVs), whereas NKH477-induced adenylyl cyclase activity was equivalent to WT. The USP20 homolog USP33, which shares redundant roles with USP20, had no effect on β1AR ubiquitination, but USP33 was up-regulated in USP20–KO hearts suggesting compensatory regulation. Myocardial β1AR expression in USP20–KO was drastically reduced, whereas β2AR expression was maintained as determined by radioligand binding in LV sarcolemmal membranes. Phospho-USP20 was significantly increased in LVs of wildtype (WT) mice after a 1-week catecholamine infusion and a 2-week chronic pressure overload induced by transverse aortic constriction (TAC). Phospho-USP20 was undetectable in β1AR KO mice subjected to TAC, suggesting a role for USP20 phosphorylation in cardiac response to pressure overload. We conclude that USP20 regulates β1AR signaling in vitro and in vivo. Additionally, β1AR-induced USP20 phosphorylation may serve as a feed-forward mechanism to stabilize β1AR expression and signaling during pathological insults to the myocardium.

␤-Adrenergic receptors (␤ARs) 3 are G protein-coupled receptors (GPCRs) that respond to signals from the sympathetic nervous system and regulate cardiovascular functions (1,2). Out of the three ␤AR subtypes expressed in the healthy heart, the ␤ 1 AR and ␤ 2 AR subtypes are the most abundant, representing 70 -80% and 20 -30% of total ␤ARs, respectively (3). These receptors couple to the heterotrimeric G protein G s upon catecholamine stimulation and activate adenylyl cyclase to promote cAMP production and protein kinase A (PKA) activation (1,2). The ␤ 1 AR is the main regulator of catecholamine-induced cardiac chronotropy and inotropy because mouse hearts lacking either ␤ 1 AR or both ␤ 1 -and ␤ 2 AR function normally at the basal state but fail to respond to ␤-adrenergic agonists; in contrast, mice lacking only ␤ 2 AR have preserved cardiac responsiveness to catecholamine stimulation (4)(5)(6)(7)(8).
Activated ␤ 1 AR and ␤ 2 AR are regulated via phosphorylation by GPCR kinases (GRKs) and binding of the multifunctional adaptor proteins called ␤-arrestins, which precede receptor endocytosis and post-endocytic sorting (1,9). ␤ 1 AR signaling is significantly attenuated in failing hearts due to receptor desensitization and down-regulation (10,11). However, ␤ 2 AR, which is mostly desensitized but not down-regulated in the failing heart, can promote anti-apoptotic signaling and is considered to be cardioprotective according to studies in mice and cardiomyocytes (12)(13)(14). ␤ 1 AR downregulation during heart failure may be compensatory to reduce cardiac metabolic demand and limit cardiotoxic signaling associated with chronic ␤AR stimulation. The downregulation of ␤ 1 AR in failing hearts is observed at both mRNA and protein levels; the latter process entails trafficking and lysosomal degradation of internalized receptors (10,(15)(16)(17). Another hallmark of failing hearts is the refractory nature of ␤ARs due to sequestration of desensitized receptor protein in late endosomes (16). Recruitment of phosphoinositide 3-kinase (PI3K) activity to the ␤ 1 AR complex pro-motes ␤ 1 AR sequestration into late endosomes, and blockade of PI3K recruitment mobilizes the trafficking of ␤ 1 AR from late endosomes to the cell surface (16,17).
GPCR lysosomal trafficking and degradation can also be modulated by ubiquitination, a post-translational modification that appends monomers or polymeric chains of the 76-amino acid protein ubiquitin to substrate proteins (18,19). Although the dynamics and mechanisms of ␤ 2 AR ubiquitination have been broadly characterized (20 -24), the mechanisms that regulate ␤ 1 AR ubiquitination and associated vesicular trafficking are largely unknown. In this study, we discovered that ubiquitin-specific protease 20 (USP20) deubiquitinates ␤ 1 AR and regulates ␤ 1 AR trafficking. We also generated and characterized a novel USP20 knockout mouse model to determine the effect of this deubiquitinase (DUB) on ␤AR expression and cardiac response to catecholamine stimulation.

USP20 functions as a cognate deubiquitinase for the ␤ 1 AR
Agonist stimulation of FLAG-tagged ␤ 1 AR expressed in HEK-293 cells with isoproterenol (Iso) resulted in a detectable increase in receptor ubiquitination, which peaked at 60 min after agonist treatment and decreased to basal levels beyond 6 h of agonist stimulation (Fig. 1, A and B). Agonist-induced lysosomal trafficking of the ␤ 1 AR was initiated after 1 h of Iso stimulation as assessed by ␤ 1 AR-LAMP1 colocalization. However, significant amounts of ␤ 1 AR trafficked to lysosomes only after 6 h of agonist treatment (Fig. 1, C and D). Agonist-induced ␤ 1 AR ubiquitination was detectable only when 10 mM N-ethylmaleimide (NEM, an inhibitor of deubiquitinating enzymes), but not when an inhibitor of 26S proteasomal activity such as MG132, was included in the lysis and immunoprecipitation buffers. This suggests that ubiquitination of ␤ 1 ARs is rapidly A, HEK-293 cells stably transfected with FLAG-␤ 1 AR were stimulated with 1 M isoproterenol (Iso) for indicated times and subjected to FLAG IP followed by serial immunoblotting (IB) with anti-ubiquitin antibody (rabbit polyclonal, Bethyl Laboratories Inc.) and polyclonal FLAG antibody. The 1st lane shows the background signal obtained from HEK-293 cells transfected with vector. B, ubiquitin smears were quantitated and normalized to cognate FLAG-␤ 1 AR bands and plotted as % maximum signal (see "Experimental procedures"). The graph includes means Ϯ S.E. from four independent experiments. *, p Ͻ 0.05 compared with 0 min, one-way ANOVA, and Bonferroni's test. C, HEK-293 cells with stable transfection of ␤ 1 AR-CFP were stimulated with 1 M Iso for the indicated times, and the distribution of ␤ 1 AR (red) and LAMP1 (green) was visualized with LSM-510 confocal microscope. Representative images are shown, and quantification of colocalization from Ն13 z-stack images from three independent experiments are included in the bar graph in D. We used Imaris software to quantify volume of colocalization from images in a Z-stack. *, p Ͻ 0.05, versus 0 h, one-way ANOVA, and Bonferroni's test. Scale bars in C, 10 m.
Deubiquitination and ␤ 1 AR trafficking reversed by deubiquitinase activity that associates with the ␤ 1 AR. We previously demonstrated that the two homologous DUBs USP20 and USP33 function redundantly to deubiquitinate and regulate post-endocytic sorting of the ␤ 2 AR to lysosomes (23). We therefore tested whether ubiquitination and trafficking of the ␤ 1 AR are regulated by deubiquitinases in a similar manner as the ␤ 2 AR. USP20 knockdown significantly increased ubiquitination of the ␤ 1 AR in both unstimulated and agonist-stimulated cells when compared with cells transfected with control siRNA (Fig.  2, A and B). In contrast, depletion of USP33 had no significant effect on ␤ 1 AR ubiquitination compared with control knockdown (Fig. 2C). For a number of GPCRs, ubiquitination serves as a sorting signal to promote their trafficking to late endosomes and lysosomes (19,25). We therefore tested whether USP20-mediated deubiquitination of the ␤ 1 AR affects lysosomal trafficking of internalized ␤ 1 ARs. After 6 h Iso stimulation of cells transfected with control siRNA, we observed a sig- Figure 2. USP20 serves as a cognate deubiquitinase for the ␤ 1 AR. A, HEK-293 cells with stable FLAG-␤ 1 AR were transiently transfected with siRNAs targeting no mRNA (CTL) or USP20 for 48 h, serum-starved for 60 min, and stimulated Ϯ1 M Iso for 60 min. IP and immunoblotting (IB) are as in Fig. 1A. Lysate blots for USP20 and ␤-actin are shown in the lower two panels. Some batches of USP20 antisera cross-react with USP33, see under "Experimental procedures." B, ␤ 1 AR ubiquitination was quantitated as in Fig. 1B. Graph represents means Ϯ S.E. from five independent experiments. *, p Ͻ 0.05, versus control-nonstimulated (NS) and USP20-Iso; **, p Ͻ 0.01 compared with all others, two-way ANOVA, Holm-Sidak's post-test. C, ␤ 1 AR ubiquitination was determined as in A, but siRNA targeting USP33 was used. Blots shown are representative of three similar experiments performed. D, distribution of ␤ 1 AR (red) and LAMP1 (green) in HEK-293 cells transfected with siRNA targeting either no mRNA (control, CTL) or USP20 Ϯ Iso were visualized with LSM-710 confocal microscope. Representative images are shown, and quantification of colocalization from Ն20 z-stack images from three independent experiments are included in the bar graph in E. We used Imaris software to quantify volume of colocalization from images in a Z-stack. #, p Ͻ 0.05 versus CTL-NS; *, p Ͻ 0.01, versus all others, two-way ANOVA, Holm-Sidak's post-test. F, lysates serially immunoblotted for USP20 and ␤-actin show knockdown of USP20. Scale bars in D, 10 m.
Deubiquitination and ␤ 1 AR trafficking nificant increase in ␤ 1 AR colocalization with the lysosomal marker protein LAMP1 (Fig. 2, D-F). Additionally, USP20 knockdown led to a statistically significant enhancement of ␤ 1 AR/LAMP1 colocalization under basal and Iso-stimulated conditions as compared with quiescent cells with intact USP20 expression (Fig. 2, D-F). Accordingly, USP20 serves as a cognate DUB for the ␤ 1 AR: USP20 deubiquitinates ␤ 1 AR and attenuates its trafficking to lysosomes.

␤ 1 AR activation promotes site-specific USP20 phosphorylation by PKA␣
We previously identified that a serine residue in the consensus PKA phosphorylation motif (Arg-Lys-Phe-Ser) within the unique insertion domain of USP20 is phosphorylated by PKA␣ upon agonist activation of the ␤ 2 AR (24). To determine whether ␤ 1 AR activation also provokes USP20 phosphorylation on this serine residue (Ser-333 in human USP20, which is conserved across different species, including mouse), we stimulated HEK-293 cells expressing ␤ 1 AR with increasing doses of Iso or dobutamine (Dob, a selective full ␤ 1 AR agonist). As shown in Fig. 3, A-D, ␤ 1 AR activation triggered robust USP20 phosphorylation within 5 min even at subsaturating doses of either agonist. Additionally, this ␤ 1 AR-induced USP20 phosphorylation is mediated by the PKA␣ isoform expressed in HEK-293 cells, because PKA␣ knockdown significantly decreased phosphorylation of USP20 induced by 5 and 20 min of stimulation with Iso or Dob (Fig. 3, E and F). However, USP20 phosphorylation triggered by these agonists prevailed with control or PKA␤ knockdown (Fig.  3, E and F). Accordingly, the PKA␣ isoform in HEK-293 cells phosphorylates USP20 Ser-333 downstream of both ␤ 1 AR (Fig. 3) and ␤ 2 AR (24) activation.

Phospho-USP20 deubiquitinates ␤ 1 AR and blocks ␤ 1 AR trafficking to lysosomes
We next ascertained the role of USP20 phosphorylation in the regulation of ␤ 1 AR trafficking by using two complementary approaches: by siRNA-mediated down-regulation of PKA␣ to stabilize de-phosphorylated USP20 in cells, and by overexpressing phosphorylation-impaired (S333A) and phosphomimetic (S333D) mutant constructs of USP20. Down-regulation of PKA␣ significantly increased the colocalization of internalized ␤ 1 AR with LAMP1 both before and after 6 h of agonist stimu- . ␤ 1 AR activation promotes phosphorylation of USP20 on serine 333 by PKA␣. A, HEK-293 cells stably expressing FLAG-␤ 1 AR were stimulated for 5 min with indicated amounts of Iso, and equivalent cell extracts were analyzed for USP20 phosphorylation (p-USP20) and expression (T-USP20). B, phospho-USP20 bands were normalized to total USP20 in each sample and plotted as % maximum signal showing means Ϯ S.E. from n ϭ 3 independent experiments. # denotes p Ͻ 0.01 compared with nonstimulated (NS) samples, one-way ANOVA, Bonferroni's post-test. C and D, experiments and analyses were as in A and B, except that the ␤ 1 AR-selective agonist dobutamine was used. #, p Ͻ 0.05; *, p Ͻ 0.01 compared with nonstimulated. E, HEK-293 cells stably expressing FLAG-␤ 1 AR were transiently transfected with siRNAs targeting no mRNA (control), PKA␣, or PKA␤ for 48 h, serum-starved for 60 min, and then stimulated with 1 M Dob or 1 M Iso for 5 or 20 min. Cell lysates were immunoblotted for p-USP20, t-USP20, PKA␣, PKA␤, and GAPDH. F, quantification of p-USP20 normalized to t-USP20. n ϭ 3, *, p Ͻ 0.01; **, p Ͻ 0.001, ANOVA, Bonferroni's post-test.
Deubiquitination and ␤ 1 AR trafficking lation as compared with quiescent cells with intact PKA␣ expression (Fig. 4, A and B). PKA␣ knockdown also preserved agonist-induced ubiquitination of the ␤ 1 AR (Fig. 4C). We overexpressed recombinant adenoviruses encoding eGFP or USP20 WT, S333A, or S333D constructs in HEK-293 cells stably transfected with FLAG-␤ 1 AR and analyzed Iso-induced ubiquitination of the receptor (Fig. 5). In the absence of exogenous USP20, Iso stimulation produced robust ␤ 1 AR ubiquitination. However, either WT USP20 or S333D overexpression led to significant deubiquitination of the receptor (Fig. 5, A and B). In contrast, S333A overexpression did not lead to receptor deubiquitination. These data suggest that USP20 Ser-333 phosphorylation by PKA␣ is required for its DUB activity toward the ␤ 1 AR.
These findings are paradoxical to our earlier findings with the ␤ 2 AR; PKA␣ knockdown augmented ␤ 2 AR deubiquitination and concomitantly blocked ␤ 2 AR lysosomal trafficking (24). Our earlier work also showed that USP20 dissociates from the agonist-activated ␤ 2 AR complex, and additionally, the phosphomimetic mutant S333D only weakly associated with the ␤ 2 AR compared with the phospho-defective S333A mutant (23,24). In contrast to the ␤ 2 AR, which dissociates from USP20 upon activation (Fig. 6) (23), agonist-activated ␤ 1 AR forms a stable complex with USP20, suggesting that phosphorylated USP20 associates robustly with the ␤ 1 AR (Fig. 6). Additionally, although overexpressed USP20 WT and S333D constructs displayed equivalent binding with the ␤ 1 AR, S333A binding was significantly reduced (Fig. 7, A and B). Despite the differences observed in the interaction with the ␤ 1 -AR versus ␤ 2 -AR, both phosphomimetic and phosphorylation-impaired forms of USP20 possessed similar intrinsic enzyme activity as assessed by active-site labeling and covalent binding of ubiquitin-vinyl methyl ester (Ub-VME) (Fig. 7, C and D). Ub-VME is an activesite-directed probe that has been widely used to assess enzyme activity of DUBs and can distinguish between active and inactive forms of DUBs (26 -28). As expected, Ub-VME did not bind a catalytically inactive USP20 in which the active-site cysteine and histidine residues have been mutated (USP20-CH) (23,29).
To determine the effects of USP20 phosphorylation on ␤ 1 AR trafficking, we overexpressed WT, S333A, S333D, and USP20-CH constructs in HEK-293 cells stably transfected with CFP-tagged ␤ 1 AR and analyzed receptor colocalization with LAMP1 at 6 h of agonist stimulation. As shown in Fig. 8, both WT USP20 and phosphomimetic S333D overexpression resulted in almost no agonist-induced increase in ␤ 1 AR colocalization with LAMP1 compared with vector. In contrast, both phospho-defective USP20 S333A and USP20-CH mutants significantly increased ␤ 1 AR trafficking to lysosomes when compared with vector-transfected cells (Fig. 8). These data suggest that USP20 activity and its phosphorylation status play a critical role in attenuating the trafficking of internalized ␤ 1 AR to lysosomes.

Deubiquitination and ␤ 1 AR trafficking Baseline characteristics and function of USP20 -KO mice hearts
Because USP20 regulates ␤ 1 AR, which is the predominant ␤AR subtype for cardiac contractility, we hypothesized that disruption of USP20 in vivo could impact ␤ 1 AR function in the myocardium. To determine the role of USP20 in vivo, we used USP20 gene trap mice (USP20 -KO) in which we confirmed the absence of USP20 protein expression by Western blotting (Fig.  9A). We next compared USP20 -KO and WT mice for overall cardiac morphology, histopathology, and function ( Fig. 9, B-E, and Table 1). USP20 -KO mice are similar to WT controls in terms of body and heart weights ( Fig. 9). Similar to WT mice, USP20 -KO mice have normal heart morphology and no fibro-sis in cross-sections assessed by Masson's trichrome staining ( Fig. 9). However, conscious, nonanesthetized USP20 -KO mice presented a faster heart rate than WT during noninvasive echocardiography, a response that can be attributed to greater excitement or stress in these animals during the procedure (Table 1). Although the LV wall dimensions of USP20 -KO mice were comparable with WT, left ventricle volumes and internal dimensions were significantly greater in the USP20 -KO mice compared with WT mice during both systole and diastole as assessed by M-mode echocardiography. Nevertheless, basal systolic function was normal in these mice and comparable with WT as determined by the percentage of LV fractional shortening (%FS, 53.5 Ϯ 7.2, n ϭ 16 for USP20 -KO versus %FS, 57.5 Ϯ 6.5, n ϭ 15 for WT). These echocardiography findings are further complemented with invasive hemodynamic studies on anesthetized WT and USP20 -KO mice, which demonstrated a similar heart rate as well as equivalent load-dependent and load-independent cardiac function but with a modest difference in diastolic function (dP/dt min ) and no significant differences in systolic function and cardiac output (Tables 2 and 3). Accordingly, USP20 gene deletion does not have a major adverse effect on cardiac development and on overall baseline cardiac performance in mice.

␤ 1 AR signaling and expression are down-regulated in USP20 -KO mouse hearts
To ascertain whether the absence of USP20 affects myocardial ␤ 1 ARs, we first repeated invasive hemodynamic measurements with Iso stimulation. Compared with WT controls, we observed a marked blunting of Iso-induced cardiac contractility (dP/dt max , Fig. 10A) and relaxation (dP/dt min, Fig. 10B) in USP20 -KO. The heart rates between the WT and USP20 -KO measured before Iso infusion were not significantly different: mean beats/min Ϯ S.D.; WT 397 Ϯ 78, n ϭ 10; KO 339 Ϯ 52, n ϭ 10. However, the increase in heart rate induced by Iso stimulation was significantly blunted in the USP20 -KO mice compared with the WT mice ( Fig. 10C). Additionally, as shown in Fig. 10D, Dob infusion did not induce cAMP increase in USP20 -KO LVs, whereas direct activation of adenylyl cyclase with the water-soluble forskolin analog NKH477 (30) resulted in equivalent cAMP production in both WT and USP20 -KO LVs (n ϭ 5-7 mice of each genotype, mean cAMP pmol/mg: To determine whether the impaired ␤AR activity was due to a reduction in receptor density, we measured the expression levels of ␤ 1 and ␤ 2 ARs in LV membranes by radioligand binding (Fig. 11, A and B). Interestingly, although the total ␤AR levels were not significantly different between WT and USP20 -KO mice, ␤ 1 AR levels were significantly decreased, and ␤ 2 AR levels were reciprocally increased in USP20 -KO LVs. These data suggest that USP20 functions as the cognate DUB to regulate trafficking and prevent degradation of the ␤ 1 AR in vivo, and as such, the ␤ 1 ARs are tonically down-regulated in USP20 -KO. Because previous studies have shown a redundant regulation of the ␤ 2 AR by USP20 and USP33 (23), we also ascertained whether the expression level of USP33 was Deubiquitination and ␤ 1 AR trafficking altered in USP20 -KO hearts. Indeed, we detected 2-fold more USP33 in USP20 -KO LV compared with WT LV extracts ( Fig.  11, C and D). These findings suggest that although the ␤ 1 AR is primarily regulated by USP20, cardiac ␤ 2 AR expression and trafficking might be regulated by USP33 in addition to USP20 in vivo.

USP20 Ser-333 phosphorylation is triggered in mouse hearts following chronic catecholamine stress and pressure overload
Excessive or chronic catecholamine stimulation of cardiac ␤ 1 ARs is deleterious to heart function and leads to adverse cardiac remodeling, perhaps by promoting pathological signaling via CaMKII signaling or via PI3K/AKT/GSK3 signaling (1,8,31,32). ␤ 1 ARs also transactivate epidermal growth factor receptors in a ␤-arrestindependent manner and mitigate apoptosis induced by chronic Iso stimulation (33). To determine whether chronic catecholamine stimulation affects cardiac USP20 and/or activity, which in turn might affect ␤ 1 AR trafficking and signaling, we subjected WT mice to chronic saline or catecholamine exposure for 1 week, and we tested the levels of USP20 and phospho-USP20 in extracts prepared from left ventricles. The experimental mice (C57BL/6, 8 -10 weeks old) had normal heart function at baseline as measured by echocardiography (Table 4). A 1-week infusion of saline did not alter cardiac function. However, chronic Iso (3 mg/kg/day for 7 days) led to a marked increase in LV chamber size and a decrease in fractional shortening (Table 4), consistent with previous studies (33). Although saline-treated mice showed weak phospho-USP20, Iso-infused mice showed a 2-4-fold increase in USP20 phosphorylation (Fig. 12). However, both saline-and Iso-treated hearts had identical levels of total USP20. The sustained phospho-USP20 signals suggest the interesting possibility of pre-served expression and signaling via the ␤ 1 AR in nonlysosomal compartments with chronic Iso stimulation.
␤ 1 AR and ␤ 2 AR also contribute to pressure-overload hypertrophy induced by transverse aortic constriction (TAC) (34,35). Previous studies have also shown that with moderate pressure gradients, only ␤ 1 ARs are responsible for left ventricular hypertrophy (35). We tested whether USP20 expression or phosphorylation are affected during hypertrophy, as this might regulate expression and desensitization of ␤ARs. Two weeks of chronic pressure overload was induced by TAC in WT, ␤ 1 AR KO, and ␤ 1 AR/␤ 2 AR double KO mice as described previously (34, 36 -39). At 2 weeks, the blood pressure gradient was measured to assess the efficacy of surgery. Pre-and post-TAC echocardiography was also performed on conscious mice. The sham-operated mice for WT, ␤ 1 AR KO, and ␤ 1 AR/␤ 2 AR double KO displayed no preversus post-surgery change in fractional shortening, and moreover, Western blottings of heart extracts showed weak to no phospho-USP20 signals ( Fig. 13 and Table 5). The WT 2-week TAC group but not the ␤ 1 AR KO nor ␤ 1 AR/␤ 2 AR double KO displayed robust phosphorylation of USP20 as well as deterioration of heart function as evaluated by ECHO (% FS, WT: pre-sham 57.8 Ϯ 4.8 and post-sham 55.8 Ϯ 5.5; pre-TAC 57.7 Ϯ 3.3 and post-TAC 43.1 Ϯ 13.6, see Table 5). The WT TAC group but neither ␤ 1 AR KO nor ␤ 1 AR/ ␤ 2 AR double KO showed significant hypertrophy (heart weight/body weight WT-sham 4.6 Ϯ 0.5 and WT-TAC 6.5 Ϯ 1.0, see Table 5). According to these data, an increase in USP20 phosphorylation induced by TAC requires expression of ␤ 1 ARs. Collectively, these in vivo data show a correlation of USP20 phosphorylation in two models of cardiac dysfunction (chronic Iso and TAC) and suggest that ␤AR expression and Figure 6. Kinetics of agonist-induced binding of USP20 with ␤ 1 AR and ␤ 2 AR are dramatically different. A, HEK-293 cells stably expressing FLAG-␤ 1 AR were stimulated with 1 M Iso for the indicated times. Receptors were immunoprecipitated (IP) with anti-FLAG affinity gel, followed by immunoblotting for USP20. HEK-293 cells without FLAG-␤ 1 AR were used as control cells (Mock). Center panel, the amount of receptors detected by FLAG antibody. Bottom panels, total cell lysate immunoblotted for USP20 and ␤-actin. B, USP20 band in each IP was normalized to the cognate ␤ 1 AR band, and the mean values (ϮS.E.) from three independent experiments were plotted as % maximum signals denoting ␤ 1 AR/USP20 association. C and D, same procedures were used as in A and B except that HEK-293 cells that stably express FLAG-␤ 2 AR were used. B and D, n ϭ 3, *, p Ͻ 0.05 compared with 0 min (Ј); one-way ANOVA, Bonferroni's post-test.
Deubiquitination and ␤ 1 AR trafficking signaling in the heart is regulated by USP20 and its phosphorylation. In addition, it appears that ␤ 1 AR regulates and preserves its own trafficking itinerary as well as signaling by triggering downstream phosphorylation of USP20. Thus ␤ 1 ARdependent USP20 Ser-333 phosphorylation is a feed-forward mechanism in vivo that might facilitate and sustain pathological cardiotoxic ␤ 1 AR signaling.

Discussion
We have identified USP20 as a novel deubiquitinase that regulates ␤ 1 AR expression, trafficking, and signaling in cultured cells and in mouse hearts (Fig. 14). Our findings reveal ␤ARinduced phosphorylation of USP20 as a critical layer of regulation with unique reciprocal effects on the lysosomal trafficking of ␤ 1 AR and ␤ 2 AR subtypes (Fig. 14). USP20 phosphorylation is induced by ␤AR stimulation and by cardiac pressure overload through the ␤ 1 AR; phospho-USP20 deubiquitinates ␤ 1 AR and blocks lysosomal trafficking of the ␤ 1 AR, whereas phospho-USP20 promotes both ubiquitination and lysosomal degradation of the ␤ 2 AR (24). Furthermore, conditions that stabilize unphosphorylated USP20 are more favorable for ␤ 1 AR lysosomal trafficking, whereas unphosphorylated USP20 favors stabilization of ␤ 2 AR expression and signaling. Thus, phospho-USP20 presents as a molecular signature that engenders trafficking bias in the lysosomal sorting of the two major ␤AR subtypes.
We previously identified USP20, and its homolog USP33, as interacting partners of the ␤ 2 AR and as mediators of ␤ 2 AR ubiquitination and recycling (23). USP20 and USP33 share 70% identity and play redundant roles in the regulation of ␤ 2 AR ubiquitination and trafficking (23). In fact, significant lysosomal degradation of the ␤ 2 AR occurs in HEK-293 cells only after down-regulation of both deubiquitinases, whereas receptor recycling is maintained when expression of either one of these DUBs is preserved (23). In contrast, we found that only USP20 and not USP33 protects ␤ 1 AR from lysosomal degradation even though ␤ 1 AR and ␤ 2 AR share 52% identity (40). Ablation of USP20 in the heart coincides with a reduction in density of the ␤ 1 AR even though cardiac USP33 is up-regulated in the USP20-deficient heart. Thus, although USP33 up-regulation may preserve ␤ 2 AR expression in the heart, it is ineffective in regulating cardiac ␤ 1 AR expression that is mainly modulated by USP20. Accordingly, the deubiquitinases USP20 and USP33 may play a prominent role in defining which ␤AR subtype signaling is sustained in normal, failing, or pressure-overloaded hearts.
Although USP20 phosphorylation is triggered by agonists, it is reversed or blocked by ␤-blockers or antagonists (24). Possibly, ␤-blockers by inhibiting cAMP production can inhibit PKA-mediated phosphorylation of USP20, thus differentially affecting the trafficking and lysosomal degradation of ␤AR subtypes in the failing heart. The ␤-blockers also show differential pmol of ubiquitin-VME in vitro at 37°C. Unmodified USP20 and USP20 that is covalently linked with ubiquitin-VME were detected by immunoblotting with rabbit polyclonal anti-HA antibody (Cell Signaling Technology). D, ratios of USP20-Ub and unmodified USP20 for each construct are plotted. No significant difference between WT, S/A, and S/D (one-way ANOVA, Bonferroni's test).

Deubiquitination and ␤ 1 AR trafficking
effects on the expression level of ␤AR subtypes in the failing heart (41)(42)(43). Although ␤-blocker therapy improves cardiac function by antagonizing catecholamine activity, ␤AR density can also be increased by some ␤-blockers (for example metoprolol), which contribute to the improved cardiac performance observed in heart failure patients (44,45). However, other ␤-blockers such as carvedilol do not up-regulate or stabilize ␤AR expression in failing hearts, but nonetheless, they decrease morbidity and mortality of patients with heart failure (43, 46 -51). Although carvedilol has been shown to block USP20 interaction with the ␤ 2 AR and inhibit ␤ 2 AR recycling (52), future work should reveal the exact effects of various ␤ 1 AR subtype-selective and nonselective blockers on ␤ 1 AR-induced USP20 phosphorylation and interaction of USP20 with the ␤ 1 AR.
DUBs exist in cells as fully processed enzymes; however, they require conformational activation induced by binding of substrates (53,54). The binding of the activity-based Ub-VME probe suggests that Ser-333 phosphorylation does not affect the intrinsic DUB activity of USP20. However, Ser-333 phosphorylation by PKA␣ dictates USP20's specific association and affinity for each ␤AR subtype, which further defines deubiquitination and effects on post-endocytic sorting to lysosomes of each ␤AR subtype. Accordingly, phospho-USP20 deubiquitinates ␤ 1 AR and blocks lysosomal trafficking of the ␤ 1 AR, whereas it promotes both ubiquitination and lysosomal degradation of the ␤ 2 AR. In contrast, conditions that stabilize unphosphorylated USP20, are more favorable for ␤ 1 AR lysosomal trafficking and for stabilization of ␤ 2 AR expression and signaling. Although pathological insults provoke USP20 phosphorylation in vivo, the effects of phospho-USP20 on the development of cardiac dysfunction and heart failure remain to be defined when USP20 S333A and S333D knockin mouse models become available.
We also identified the physiological effects of USP20 ablation in the heart of our newly characterized USP20 -KO mice. The basal systolic function in USP20 -KO and WT are comparable, but USP20 -KO manifests modest load-dependent relaxation impairment determined by invasive hemodynamic studies on anesthetized mice. Nevertheless, overall cardiac performance was normal as cardiac output in USP20 -KO mice was similar to WT. Importantly, USP20 -KO mice show markedly impaired cardiac responsiveness to ␤-agonist stimulation. This impaired cardiac contractility induced by adrenergic stimulation is ascribed to the inability of the heart to promote cAMP production after stimulation with the ␤ 1 AR-selective agonist dobutamine. However, USP20 ablation did not affect the signaling machinery downstream of ␤ 1 AR as NKH477-induced adenylyl cyclase activation was preserved. The impaired ␤-adrenergic signaling in USP20 KO mouse was associated with lower density of ␤ 1 AR. The inability of the remaining ␤ 1 AR to activate cAMP in the USP20 -KO LVs may be linked to their ubiquitination status, which promotes their compartmental-

Deubiquitination and ␤ 1 AR trafficking
ization in lysosomes or induces conformational changes that do not favor G protein activation. Under chronic or intermittent stress of pressure overload, the densities of both ␤ 1 -and ␤ 2-ARs are significantly reduced (55). Future studies should reveal the exact role of USP20 in regulating ␤ 1 AR expression and signaling in the heart subjected to pressure overload.
The lessened Iso-induced cardiac contractility in USP20-KO mice is consistent with ␤ 1 AR deficiency because mice lacking ␤ 1 AR also present normal basal cardiac function and impaired cardiac responsiveness to ␤-adrenergic stimulation (5, 7). However, the exact underlying mechanism that leads to modest reduction in relaxation in USP20 -KO mice remains to be defined. Moreover, preserved expression of ␤ 2 AR in cardiac membranes of USP20 -KO was associated with up-regulation Figure 9. Cardiac morphology, histopathology, and morphometry of WT and USP20 -KO mice. A, immunoblot of cardiac lysates from 3-to 4-month-old WT and USP20 -KO mice with USP20 and GAPDH antibodies. Blots are representative of similar results obtained from Ͼ10 mice of each genotype. B, representative whole hearts from 3-to 4-month-old WT and USP20 -KO mice (n ϭ 3, scale bar ϭ 1 mm). C, heart weight to body weight ratio from 3-to 4-month-old WT (n ϭ 12) and USP20 -KO (n ϭ 12) mice. Error bars indicate average Ϯ S.E. D, representative Masson Trichrome staining of cardiac sections from 3-to 4-month-old WT and USP20 -KO (n ϭ 5, scale bar, 0.5 mm). E, representative baseline echocardiograms via M-mode, short axis imaging, of WT and USP20 -KO mice (see Table 1 for ECHO analyses).  Deubiquitination and ␤ 1 AR trafficking of USP33. Whether USP20 regulates additional GPCRs and proteins in the heart and whether USP20 modulates alternative signaling mechanisms in pressure overloaded ventricles and heart failure also remain to be defined.
In conclusion, we have determined a novel role of USP20 in the regulation of myocardial ␤ 1 AR density and cardiac responsiveness after catecholamine stimulation. Although catecholamines stimulate ␤AR signaling that is critical for inducing powerful contractions of the heart muscle, catecholamines also trigger GRK2-mediated signal desensitization that is followed by receptor endocytosis and degradation (56 -60). These events are maladaptive in the long term as they either cause ␤ARs to be degraded or to be refractory to stimulus and thus reduce cardiac performance. In this scenario, ␤ 1 ARs also promote noncanonical signaling pathways that promote apoptosis or cardiac hypertrophy (61). Accordingly, restoring ␤AR activ-ity without enabling endocytosis or directly inhibiting the endocytic machinery has been found to be cardioprotective (16,(61)(62)(63)(64)(65)(66)(67)(68)(69). Thus, preserving ␤AR levels and activity by preventing lysosomal trafficking of ␤ARs is a valuable therapeutic strategy for heart failure; in this context, exploiting deubiquitinases and their activity might prove beneficial.

Experimental animals
Genetically engineered ␤ 1 AR-KO and ␤ 1 AR/␤ 2 AR-KO mice were described previously (5,7). The USP20 gene trap heterozygous mouse line (Usp20 tm1a(EUCOMM)Hmgu ) was purchased from the International Knockout Mouse Consortium. The USP20 gene trap heterozygous mice were mated to produce homozygous gene trap and WT mice that were used as experimental cohorts. Additionally, we generated USP20 gene trap KO with C57BL/6N ES cell lines (HEPD0524_2_C09 and HEPD0524_2_H10, EUCOMM), with the help of Duke Transgenic Core. All three USP20 gene trap KO lines (designated as USP20 -KO) were studied and were assigned into age-and gender-matched experimental cohorts for cardiac phenotyping. Genotypes of all breeding and experimental animals were confirmed by PCR-based and Western blotting assays. All in vivo measurements were made by observers blinded to specimen identity. All animals were handled according to the approved protocols and animal welfare regulations of the Duke University Institutional Animal Care and Use Committee.

Cell lines
Human embryonic kidney (HEK-293) cells were purchased from the American Type Culture Collection and grown in minimal essential medium supplemented with 10% fetal bovine  Deubiquitination and ␤ 1 AR trafficking serum and 1% penicillin/streptomycin. Plasmid transfections in these cells were performed at 40 -50% confluency in the presence of Lipofectamine 2000 TM (ThermoFisher Scientific) and following manufacturer's protocol as described before (52). Stable cell lines expressing FLAG-tagged ␤ 1 AR, CFP-tagged ␤ 1 AR, or FLAG-tagged ␤ 2 AR were selected by supplementing growth medium with 1 mg/ml G418, and upon passaging they were maintained in growth medium containing 400 g/ml G418, as described previously (71).

Cardiac morphological and histological examinations
Morphological and histological cardiac assessments were performed as described previously (72). Hearts were excised and briefly washed in PBS before photographing with a Nikon SMZ 800 stereomicroscope and a DS-Fi1 camera. Histological heart specimens were fixed in 10% formaldehyde solution, sectioned into 5-m thick slices, and stained with Masson trichrome before visualization at ϫ20 magnification under a Zeiss Axio Imager widefield fluorescence microscope. ZEN software version 2.3 (Carl Zeiss Microscopy) was used for histological image processing and stitching, and NIH ImageJ was used for quantification and analysis.

Transthoracic echocardiography
All cardiovascular phenotyping was performed at the Duke Cardiovascular Physiology Core. Echocardiography was per- Figure 11. ␤ 1 AR is down-regulated, and ␤ 2 AR and USP33 are up-regulated in USP20 -KO LVs. A, ␤AR subtypes in LV of WT and USP20 -KO mice were measured by radioligand binding with 125 I-cyanopindolol and using the ␤ 1 AR subtype-selective antagonist CGP20712A as reported before (72,74). Nonspecific binding was defined by the nonselective antagonist propranolol. Experiments were performed in triplicate. n ϭ 21 for WT and n ϭ 17 for USP20 -KO. Error bars indicate average Ϯ S.E. *, p Ͻ 0.05, ANOVA Bonferroni's post-test. B, bar graphs show relative proportion of ␤ 1 and ␤ 2 AR in USP20 -KO and WT LVs calculated from the binding values shown in A. C, immunoblot (IB) of cardiac lysates from 3-to 4-month-old WT and USP20 -KO mice probed with USP20, USP33, and GAPDH antibodies. D, band intensities of USP20 and USP33 normalized to GAPDH in each sample are shown and represented as means Ϯ S.E. *, p Ͻ 0.05, versus USP33 in WT hearts, ANOVA, Bonferroni's post-test.

Table 4 Echocardiographic measurements in WT mice after chronic Iso
Values are expressed as mean Ϯ S.D. The following abbreviations are used: LV, left ventricle; s, systole; d, diastole; IVS, intraventricular septum; LVD, left ventricle internal dimension; LVPW, left ventricular posterior wall; ET, ejection time; BPM, beats/min; FS, fractional shortening of LV; LVm, LV mass; mVcf, mean velocity of circumferential fiber shortening; mVcfc, mean velocity of circumferential fiber shortening corrected for heart rate; cir/s, circumferences/s. Statistical significance was determined by two-way ANOVA with post hoc Sidak analysis.

Deubiquitination and ␤ 1 AR trafficking
formed on 2-4-month-old awake mice using a Vevo 2100 highresolution imaging system (VisualSonics, Toronto, Ontario, Canada) as described previously (72). Left ventricle structure and dimension were assessed via parasternal short axis view in M-mode, whereas aortic ejection time was measured by pulsed-wave Doppler. All data and images were analyzed with the Vevo 2100 Imaging System by an examiner blinded to mouse identity.

Hemodynamic evaluation in intact anesthetized mice
Mice were anesthetized with a mixture of ketamine (100 mg/kg i.p.) and xylazine (2.5 mg/kg i.p.) and connected to a rodent ventilator via endotracheal intubation (8,34). Intra-ventricular pressure was measured in real-time with a 1.4F high-fidelity micromanometer catheter (Millar Instruments) inserted into the left ventricle via the right carotid artery. Isoproterenol was administered through a polyethylene catheter (PE10) inserted into the right external jugular vein. Ventricular contractility was assessed at baseline and 45 s after injection of increasing doses of isoproterenol.

Left ventricular pressure-volume analysis
After bilateral vagotomy, cardiac catheterization was performed with a 1.4F high-fidelity micromanometer catheter (Millar Instruments) on anesthetized mice as described for the hemodynamic experiments and as described previously (8,34). Left ventricle volume and intraventricular pressure during systole and diastole were measured in real-time and recorded using a P-V conductance system (PowerLab, AD Instruments). Various preload conditions were elicited by transient inferior vena cava occlusion. Parallel conductance (Vp) of the blood pool was determined by a 10-l injection of 15% saline into the right jugular vein and used for correction of P-V loop data. Data were analyzed with P-V analysis software (PVAN data analysis software version 3.3; Millar Instruments) by an observer blinded to mouse genotype.

Determination of cardiac cAMP levels
For catecholamine-induced cAMP, 3-4-month-old mice were injected with a single bolus of a ␤ 1 AR-selective agonist of dobutamine (10 g/g body weight) or an equivalent volume of vehicle (0.9% NaCl) given intraperitoneally. For direct adenylyl cyclase activation, mice were treated with 1 g/g body weight of the water-soluble forskolin analog NKH477 (30). 2 min after drug injection, the mice were rapidly euthanized, and their hearts were excised, flash-frozen in liquid nitrogen, and stored at Ϫ80°C until assays. Total cAMP levels in mouse ventricles were measured with Direct cAMP ELISA kit (catalog no. ADI-900-066; Enzo Life Sciences Inc) according to the manufacturer's instructions. Optical density associated with cAMP levels was measured at 405 nm with Synergy Neo2 Hybrid Multi-Mode Reader (Biotek).

Radioligand-binding assays
Radioligand-binding assays were performed as described previously (72)(73)(74). Briefly, membranes isolated from LV of WT and USP20 -KO mice were diluted to a concentration of 0.5 mg/ml in ice-cold binding assay buffer containing 50 mM Figure 12. ␤AR stimulation induces USP20 phosphorylation in vivo. Soluble extracts from left ventricles of WT mice exposed chronically to saline or Iso were immunoblotted for p-USP20 and total USP20. Quantification of p-USP20 normalized to USP20, saline n ϭ 6; Iso n ϭ 5, *, p ϭ 0.01, unpaired, two-tailed t test. Figure 13. Cardiac pressure overload induces USP20 phosphorylation in a ␤ 1 AR-dependent manner. A, soluble extracts from left ventricles of WT mice with sham or TAC surgery were immunoblotted for p-USP20 and total USP20. B, quantification of p-USP20 normalized to USP20, sham n ϭ 5, and TAC n ϭ 11, **, p ϭ 0.003, unpaired, two-tailed t test. The maximum p-USP20/ USP20 ratio in the experimental cohort was taken as 100%. C and D, soluble extracts from left ventricles of mice that underwent sham or TAC surgery of indicated genotypes were serially immunoblotted for p-USP20 and total USP20. E, graph shows quantification of p-USP20 normalized to USP20 plotted as in B. *, p Ͻ 0.05, ANOVA, Bonferroni's post-test.

Deubiquitination and ␤ 1 AR trafficking
Tris-Cl (pH 7.4), 2 mM EDTA (pH 8.0), 12.5 mM MgCl 2 , and 180 g/ml L-ascorbic acid. 500 pM 125 I-cyanopindolol was added to all membranes, and inhibition of specific 125 I-cyanopindolol binding was determined by adding 20 M nonselective antagonist propranolol or ␤ 1 AR-selective antagonist CGP20712A. Samples were prepared in triplicate in polypropylene 96-well Deubiquitination and ␤ 1 AR trafficking plates (Costar, Cambridge, MA), and at least three independent experiments were performed. After 90 min of incubation at room temperature, membrane-bound radioligand was harvested onto Whatman GF/B glass fiber filters using a Brandel cell harvester (Brandel, Gaithersburg, MD). Filters were washed three times with ice-cold washing buffer (50 mM Tris-Cl (pH 7.4), 2 mM EDTA (pH 8.0), 12.5 mM MgCl 2 ), and bound radioactivity were measured with a Packard Cobra gamma counter.

RNA interference
Control siRNA targeting no mRNA and siRNA targeting USP20, USP33, PKA␣, or PKA␤ were purchased from Dharmacon GE Healthcare as described previously (23,24). Transfections were performed in serum-free medium in 40 -50% confluent cells using 20 g of siRNA in the presence of Lipofectamine 2000 TM and following the manufacturer's protocol. 4 h after transfection, serum-containing medium was added to the cells and maintained for 48 h until experiment was performed. Cells in which knockdown of target protein attained Ͼ80% efficiency were used for experimental analyses.

Immunoprecipitation and immunoblotting
For cardiac samples, freshly isolated mouse ventricles were solubilized in an ice-cold lysis buffer containing 20 mM Tris-Cl (pH 7.4), 137 mM NaCl, 20% (v/v) glycerol, and 1% (v/v) IGE-PAL CA-630, supplemented with phosphatase and protease inhibitors (1 mM sodium orthovanadate, 10 mM sodium fluoride, 10 mM phenylmethylsulfonyl fluoride, 5 g/ml leupeptin, 5 g/ml aprotinin, 1 g/ml phosphatase inhibitor mixture 2 (P5726), and 1 g/ml phosphatase inhibitor mixture 3 (P0044); all were from Sigma. Cardiac homogenates were centrifuged at 35,000 ϫ g for 30 min at 4°C, and the supernatant was used for experiments. For immunoprecipitation assays, HEK-293 cells stably expressing FLAG-tagged ␤ 1 -or ␤ 2-AR were solubilized in ice-cold lysis buffer containing 50 mM HEPES (pH 7.5), 2 mM EDTA (pH 8.0), 250 mM NaCl, 10% (v/v) glycerol, and 0.5% (v/v) IGEPAL CA-630 supplemented with the aforementioned phosphatase and protease inhibitors and also with 10 mM N-ethylmaleimide to inhibit cellular deubiquitinase activities and preserve receptor ubiquitination. Lysates from these cells were centrifuged at 13,000 rpm for 20 min at 4°C, and 1-2 mg of supernatant whole-cell extracts were immunoprecipitated using anti-FLAG M2 resin. After overnight end-over-end rotation at 4°C, immune complexes were washed three times with lysis buffer and eluted in 2ϫ SDS-PAGE sample buffer. Immune complexes and 20 g of corresponding lysates were resolved on 4 -20% gradient gels and transferred onto nitrocellulose membranes. Membrane blocking and antibody incubation were in 5% (w/v) dried skim milk powder dissolved in TTBS (0.2% (v/v) Tween 20, 10 mM Tris-Cl (pH 8.0), and 150 mM NaCl), whereas in-between washes were performed in TTBS. Note that many batches of commercial anti-USP20 IgG also detect endogenous USP33, but anti-USP33 IgG does not detect endogenous USP20; optimal separation of the two proteins is achieved on 4 -20 or 4 -12% Tris glycine gradient gels. Immunoblotted proteins were detected with enhanced chemiluminescence (SuperSignal West Pico reagent, Pierce). Signals were detected and acquired with a charge-coupled device camera system (Bio-Rad Chemidoc-XRS) and analyzed with Image-Lab software (Bio-Rad).

Quantification of ␤ 1 AR ubiquitination
The FLAG-␤ 1 AR IPs were serially immunoblotted for ubiquitin and FLAG tag. Band densities detected by the ubiquitin antibody between 70 and 300 kDa were determined in all lanes, and those values from the lane that had no FLAG-␤ 1 AR was used as the negative control to subtract nonspecific signals. Similar densitometry was performed for the FLAG blot to evaluate the signals of prominent ␤ 1 AR bands between 45 and 70 kDa. The specific values obtained for ubiquitinated species in each lane were then divided by the corresponding ␤ 1 AR value, which was used in producing bar graphs. In Fig. 1B, maximal   Figure 14. Effects of USP20 and its phosphorylation on ␤ 1 AR signaling. A, activation of the ␤ 1 AR induces USP20 phosphorylation on Ser-333 catalyzed by PKA␣. Phosphorylated USP20 deubiquitinates ␤ 1 AR and attenuates lysosomal trafficking, thus facilitating ␤ 1 AR signaling. B, PKA␣ down-regulation by RNAi or overexpression of weakly interacting USP20 S333A mutant leads to increased lysosomal trafficking of ubiquitinated ␤ 1 ARs. C, in USP20 -KO mice, myocardial ␤ 1 AR levels are significantly decreased, whereas levels of ␤ 2 AR and USP33 (homolog of USP20) are increased. Accordingly, although USP20 and USP33 serve as redundant DUBs for the ␤ 2 AR, USP20 functions as the predominant DUB to regulate ␤ 1 AR expression and trafficking in the myocardium.
Deubiquitination and ␤ 1 AR trafficking ubiquitin signal in each experiment is taken as 100%. In Figs. 2B, and 5B, the signals from control ϩ agonist is fixed as 100% in each experiment.

Adenovirus-mediated transduction
HA-tagged adenovirus expressing WT, phosphomimetic, phospho-deficient, or catalytically inactive USP20 mutants were generated with the AdEasy system (Agilent Technologies) according to published protocols (75,76), and their identity was confirmed by sequencing.
For conjugation assays with Ub-VME (U-203; R&D Systems) and for co-immunoprecipitation experiments, cells were transduced with adenovirus for 2 h and harvested 12-14 h posttransduction in co-immunoprecipitation buffer deprived of NEM. For Ub-VME conjugation assays, the level of expression of each USP20 construct was estimated by densitometric comparison of known amounts of purified HA-USP20 (76) as detected by immunoblotting with polyclonal anti-HA antibody. The final volume of lysate used in Ub-VME binding was adjusted to obtain 4 pmol of USP20 in each reaction. 20 pmol of Ub-VME was added to the reaction. In these assays, the maximal shift for USP20 WT required a 30-min incubation of the lysate ϩ Ub-VME mixture at 37°C. We used this condition to compare the reactivity of S333A, S333D, and USP20-CH constructs. The reactions were terminated by addition of SDS-PAGE sample buffer, and further analysis was carried out by Western blotting with a polyclonal anti-HA antibody (Cell Signaling Technology).

Immunofluorescence staining and confocal imaging
Immunofluorescence staining and confocal imaging were performed as described previously (23,24). HEK-293 cells stably expressing exogenous ␤ 1 AR-CFP were transfected with siRNA or plasmids encoding pcDNA3-HA and HA-tagged USP20 WT or mutants. 24 h post-transfection, the cells were seeded on poly-D-lysinecoated 35-mm glass bottom plates (MatTek Corp, Ashland, MA), serum-starved for 1 h the next day, stimulated with 1 M Iso, and fixed with 5% formaldehyde diluted in calcium/magnesium-containing Dulbecco's PBS (DPBS). 15 min after fixation, cells were permeabilized for 20 -30 min with 0.1% Triton X-100, incubated with primary antibody overnight at 4°C, and with cognate secondary antibody for 1 h at room temperature. Washes after cell fixation and antibody incubations were performed with DPBS, whereas DPBS containing 2% BSA was used for permeabilizing solution and antibody dilutions. Cells were visualized with Zeiss LSM-710 or LSM-510 META confocal microscope with filter settings for respective fluorophores: excitation nm are 435 CFP, 488 (Alexa 488), 568 (Alexa 594), and 633 (Alexa 633); emission nm are 460 -500 CFP, 515-540 Alexa fluor 488, 585-615 Alexa fluor 594, and 650 (Alexa Fluor 633). Images acquired by the LSM operating software (ZEISS ZEN imaging software) were analyzed with Imaris software (Bitplane, South Windsor, CT) to quantify fluorophore colocalization. A total of 13-20 z-stack images from three independent experiments were used for analyses.

Micro-osmotic pump implantation
Micro-osmotic pump implantation for chronic isoproterenol administration was performed as described previously (77,78). Mice were anesthetized with a mixture of ketamine (100 mg/kg) and xylazine (2.5 mg/kg). Saline-or isoproterenol-filled micro-osmotic pumps (Alzet model 1007D, Durect Corp., Cupertino, CA) were subcutaneously implanted in 3-4month-old WT mice and adjusted to a dosage of 3 mg/kg/day for 7 days. After treatment, mice were rapidly euthanized, and their hearts were processed for biochemical analyses.

TAC
Transverse aortic constriction was performed in 8 -12week-old mice anesthetized with a mixture of ketamine (100 mg/kg) and xylazine (2.5 mg/kg) as described previously (34). After thoracotomy, the aortic arch was ligated with a 6.0 nylon suture to the width of a 27-gauge needle, inducing a 60 -80% constriction. The sham control mice underwent a similar procedure but without aortic ligation. After needle removal and chest/skin closure, animals were maintained for 2 weeks and then euthanized for molecular analyses of cardiac tissues.

Statistical analysis
Data from at least three independent experiments were averaged and represented as means Ϯ S.E. Two-tailed Student's t test was used for basal assessments of WT and USP20 -KO mice. Studies involving more than two groups of mice, as well as time-course and dose-response experiments were analyzed with ANOVA followed by post hoc correction for multiple comparisons. Statistical analyses were performed using GraphPad Prism 7 (GraphPad, Inc.), and significance was established for p Ͻ 0.05.