Regulated and Aberrant Glycosylation Modulate Cardiac Electrical Signaling

Regulated and aberrant glycosylation modulate cardiac electrical signaling

Marty L. Montpetita, Patrick J. Stockera, Tara A. Schwetza, Jean M. Harpera, Sarah A. Norringa, Lana Schafferb, Simon J. Northc, Jihye Jang-Leec, Timothy Gilmartinb, Steven R. Headb, Stuart M. Haslamc, Anne Dellc, Jamey D. Marthd, and Eric S. Bennetta,1

aDepartment of Molecular Pharmacology & Physiology, Programs in Cardiovascular Sciences and Neuroscience, University of South Florida College of Medicine, Tampa, FL 33612; bDNA Microarray Core, The Scripps Research Institute, La Jolla, CA 92037; and cDivision of Molecular Biosciences, Imperial College London, London SW7 2AZ, United Kingdom; and dDepartment of Cellular and Molecular Medicine, The Howard Hughes Medical Institute, University of California at San Diego, La Jolla, CA 92093

Edited by Richard W, Aldrich, University of Texas, Austin, TX, and approved July 2, 2009 (received for review May 18, 2009) Millions afflicted with Chagas disease and other disorders of aberrant phied and more susceptible to arrhythmias (3). Electrical remod- glycosylation suffer symptoms consistent with altered electrical sig- eling occurs during development and aging, among species, and naling such as arrhythmias, decreased neuronal conduction velocity, throughout the heart (4, 5). In nearly all cardiac pathologies and hyporeflexia. Cardiac, neuronal, and muscle electrical signaling is including hypertrophy, heart failure, and long QT syndrome controlled and modulated by changes in voltage-gated ion channel (LQTS), at least one type of remodeling occurs (3, 6). activity that occur through physiological and pathological processes Voltage-gated ion channels are heavily glycosylated, with such as development, epilepsy, and cardiomyopathy. Glycans at- glycan structures comprising upwards of 30% of the mature channel mass (7, 8). Previous reports indicated that the sugars tached to ion channels alter channel activity through isoform-specific ϩ mechanisms. Here we show that regulated and aberrant glycosyla- attached to cardiac voltage-gated Na channels (Nav) and Kv tion modulate cardiac ion channel activity and electrical signaling may impact channel gating (9, 10). Glycosylation of Nav and Kv through a cell-specific mechanism. Data show that nearly half of 239 subunits were shown to alter channel gating in isoform- and subunit-dependent manners (11–14). Most studies established glycosylation-associated genes (glycogenes) were significantly differ- that the sugar-dependent gating effects were imposed by the entially expressed among neonatal and adult atrial and ventricular terminal residue attached to carbohydrate structures, sialic myocytes. The N-glycan structures produced among cardiomyocyte acid. A recent work showed that variable sialic acid levels types were markedly variable. Thus, the cardiac glycome, defined as attached to a single Nav isoform, Nav1.5, were responsible for the complete set of glycan structures produced in the heart, is differences in channel gating observed among neonatal and remodeled. One glycogene, ST8sia2, a polysialyltransferase, is ex- adult atrial and ventricular cardiomyocytes (15). pressed only in the neonatal atrium. Cardiomyocyte electrical signal- There are at least two major types of disorders of aberrant ؊ ؊ ing was compared in control and ST8sia2( / ) neonatal atrial and glycosylation that result in decreased glycoprotein sialylation and ventricular myocytes. Action potential waveforms and gating of less afflict nearly 20 million people: Congenital disorders of glycosyl- sialylated voltage-gated Na؉ channels were altered consistently in ation (CDGs; Ϸ30 known diseases) and Chagas disease, caused by ST8sia2(؊/؊) atrial myocytes. ST8sia2 expression had no effect on parasitic infection. The primary target organ for Chagas disease and ventricular myocyte excitability. Thus, the regulated (between atrium for some CDGs is the heart, leading to an increased susceptibility and ventricle) and aberrant (knockout in the neonatal atrium) expres- to conduction anomalies, cardiac arrhythmias, and heart failure sion of a single glycogene was sufficient to modulate cardiomyocyte (16–19). Thus, these diseases lead to aberrant cardiac glycosylation excitability. A mechanism is described by which cardiac function is and to symptomatic changes in cardiac excitability. controlled and modulated through physiological and pathological Previous studies established that N-glycosylation, typically sialic

processes that involve regulated and aberrant glycosylation. acids, modulate voltage-gated ion channel gating through isoform- PHYSIOLOGY specific mechanisms. Individuals afflicted with certain disorders of action potentials ͉ cardiomyocyte ͉ glycomics ͉ ion channels ͉ sialic acids reduced glycosylation present with arrhythmias consistent with changes in cardiac ion channel function. However, little is known about a direct role for glycans in cardiac function. Here we show he glycome, defined as the complete set of glycan structures that electrical communication in the heart is modulated by regu- Tproduced by the body, is comprised of hundreds of thousands lated and aberrant glycosylation. Specifically, the cardiac glycome of unique structures (1). Such structural diversity is the result of the is different in the atria than in the ventricles and the glycome is activity of nearly 250 known glycosylation-associated genes such as remodeled differentially during development of each cardiac cham- glycosyltransferases, glycosidases, and nucleotide sugar synthesis ber. Further, regulated and aberrant glycosylation modulate cardi- and transporter genes (glycogenes) that are responsible collectively omyocyte excitability and Nav channel function consistently. for producing the glycans attached to lipids and proteins (2). Functional roles for glycans in cellular communication include cell Results adhesion, self-recognition, protein trafficking and clearance, and Cardiac Glycogene Expression Is Highly Regulated. To test the poten- receptor activation (2). tial role for regulated glycosylation in cardiac excitability, our Electrical signaling occurs in all cells of the body and is of primary importance to excitable cell function. Neurons, skeletal muscle, and cardiac muscle communicate through production and conduction Author contributions: M.L.M., P.J.S., T.A.S., J.M.H., S.J.N., S.R.H., S.M.H., and E.S.B. designed research; M.L.M., P.J.S., T.A.S., J.M.H., S.A.N., S.J.N., J.J.-L., T.G., S.R.H., S.M.H., and E.S.B. of orchestrated electrical signals called action potentials (AP). performed research; S.J.N., S.R.H., S.M.H., A.D., and J.D.M. contributed new reagents/ Neuronal and muscle APs are transient membrane depolarizations analytic tools; M.L.M., L.S., S.J.N., J.J.-L., T.G., S.R.H., S.M.H., and E.S.B. analyzed data; and produced by the concerted activities of many types of voltage-gated M.L.M. and E.S.B. wrote the paper. ion channels and transport proteins. Slight alterations in ion chan- The authors declare no conﬂict of interest. nel activity can lead to altered excitability observed as a change in This article is a PNAS Direct Submission. AP waveform and/or conduction. For example, the types and 1To whom correspondence should be addressed. E-mail: [email protected]. ϩ relative densities of the set of voltage-gated K channel (Kv) This article contains supporting information online at www.pnas.org/cgi/content/full/ isoforms expressed change as the healthy heart becomes hypertro- 0905414106/DCSupplemental.

www.pnas.org͞cgi͞doi͞10.1073͞pnas.0905414106 PNAS ͉ September 22, 2009 ͉ vol. 106 ͉ no. 38 ͉ 16517–16522 Downloaded by guest on September 30, 2021 A tetra-antennary structures, which are mono-, di-, tri-, and tetra- All glycogenes (110 of 239) sialylated with a mixture of two forms of sialic acids, N- acetylneuraminic acid (NeuAc) and N-glycolylneuraminic acid (NeuGc). Nucleotide sugar (15 of 35) The data indicate N-glycan structures are regulated throughout cardiomyocyte development, exemplified by the change in the relative levels of NeuAc and NeuGc attached to atrial and ventric- Glycosidases (16 of 39) ular myocyte N-glycans. Recent studies demonstrated that MALDI-MS analyses of permethylated glycans provide reliable relative quantitative information based on signal intensities, par- Glycosyltransferases (79 of 165) ticularly when comparisons are made over a small mass range in the same spectrum (20). In making such comparisons, the adult N- 0 1020304050 glycan profiles exhibited a significant increase in the ratio of NeuGc % Genes Differentially Expressed to NeuAc relative to the neonatal samples (Fig. 2, inset), indicating a change in glycan structures, particularly sialic acids, during B VN:AN VA:AA cardiomyocyte development. Overall Overall Regulated and Aberrant Sialylation Impact Cardiomyocyte Excitabil- Nucleotide Sugar Nucleotide Sugar Chamber ity. To determine whether the remodeled glycome modulates

Comparison Glycosidases Glycosidases cardiac function, we observed the impact of the regulated and aberrant expression of a single glycogene on cardiomyocyte Glycosyltransferases Glycosyltransferases electrical activity. We studied the effect of the polysialyltrans- 0 5 10 15 20 25 0 1020304050 ferase, ST8sia2 (responsible for addition of sialic acid poly- % Genes Differentially Expressed % Genes Differentially Expressed mers to N- and O-glycans), on cardiac function for several AA:AN VA:VN reasons that include: (i) Cardiac dysfunctions including ar-

Overall Overall rhythmias and cardiomyopathy that are likely caused by changes in ion channel activity are prevalent in diseases of Developmental Nucleotide Sugar Nucleotide Sugar aberrant sialylation such as Chagas disease and some CDGs comparison Glycosidases Glycosidases (16–19); (ii) all our previous data indicated that changes in ion channel sialylation contribute significantly to the modulation Glycosyltransferases Glycosyltransferases of ion channel function (11–13, 15); and (iii) cardiac ST8sia2 0510152025 0 1020304050 expression is regulated. ST8sia2 is expressed at much higher % Genes Differentially Expressed % Genes Differentially Expressed levels in the neonatal atrium than in the neonatal ventricle as Fig. 1. Cardiac glycogene expression is regulated. Relative cardiac glycogene determined through microarray (Fig. S1A and Table S1B) and expression levels were determined using microarray analysis as described. Bar qPCR (Fig. S2) analyses and is essentially absent in the adult. graphs represent the percentage of the 239 tested glycogenes and the families We questioned whether regulated and aberrant ST8sia2 ex- of glycogenes (glycosyltransferases, glycosidases, and sugar nucleotide syn- pression impacted myocyte excitability by comparing cardiomy- thesis and transporter genes) signiﬁcantly differentially expressed among ocyte AP waveform and Nav function in neonatal atrial and myocyte types. (A) Among neonatal and adult atrial and ventricular myocytes ventricular myocytes isolated from control and ST8sia2 knock- (P Ͻ 0.01). (B) Developmental and chamber-speciﬁc expression differences (Ϫ/Ϫ) Ͻ out [ST8sia2 ] mice (Fig. 3) (21). between two myocyte types. (P 0.05); NA, neonatal atrium; NV, neonatal (Ϫ/Ϫ) ventricle; AA, adult atrium; AV, adult ventricle. Several parameters of the ST8sia2 atrial myocyte AP were significantly different compared to control data (Fig. 3). The time to AP peak was significantly reduced in ST8sia2(Ϫ/Ϫ) atrial myocytes analysis began by determining whether glycogene expression while the AP duration (APD) was increased significantly. No Ϫ Ϫ levels are regulated between atria and ventricles and during significant differences were observed between the ST8sia2( / ) and cardiomyocyte development. We measured the relative expres- control ventricular myocyte AP parameters, consistent with very sion of hundreds of glycogenes using GeneChip microarray low ST8sia2 expression in the ventricle. analysis of mRNA isolated from neonatal and adult atria and ventricles (Fig. 1, Fig. S1, and Table S1). Microarray data were Regulated and Aberrant Expression of a Single Glycogene Modulate validated using real-time RT-PCR (qPCR) analysis (Fig. S2). Nav Gating Consist with Observed Changes in Cardiomyocyte Excit- The microarray and qPCR data were consistent. ability. To correlate AP waveform changes with a direct effect on The data showed that 110 of the 239 glycogenes tested were ion channel function, whole cell Nav gating was measured in significantly differentially expressed among the four myocyte types neonatal atrial and ventricular myocytes isolated from (Ϫ/Ϫ) studied (Fig. 1A). Each family of glycogene (glycosyltransferases, ST8sia2 and control animals (Fig. 4). Nav gating was not glycosidases, and sugar nucleotide synthesis/transporter genes) con- significantly different in ST8sia2(Ϫ/Ϫ) versus control ventricular tributed proportionately equally to the large number of regulated myocytes (Fig. 4, right panels). However, there was a significant glycogenes. Glycogene expression between atria and ventricles at a Ϸ7 mV depolarizing shift in the steady state conductance- (Ϫ/Ϫ) single developmental stage and throughout development of each voltage relationship measured for Nav expressed in ST8sia2 chamber was also markedly regulated (Fig. 1B). atrial myocytes compared to control data (Fig. 4A). All other voltage-dependent gating parameters measured for the (Ϫ/Ϫ) The Cardiac Glycome Is Remodeled. To determine if the regulated ST8sia2 atrial Nav were shifted along the voltage axis by expression of glycogenes translated into a remodeled cardiac gly- nearly the same magnitude and shifted toward the values mea- come, neonatal and adult atrial and ventricular myocytes were sured for ventricular myocyte Nav (Fig. 4, left panels). These subjected to glycan profiling by MALDI-TOF-MS analysis (Fig. 2 data are consistent with our previous report indicating Nav and Table S2). The resultant N-glycan mass spectra are comprised sialylation impacts channel gating in the neonatal atrium but not of high mannose-type structures and complex-type glycans. The in the neonatal ventricle (15). These data also correlate with the complex N-glycans observed comprise core fucosylated bi-, tri-, and AP data. First, myocyte capacitance and whole myocyte Naϩ

16518 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0905414106 Montpetit et al. Downloaded by guest on September 30, 2021 Fig. 2. N-glycan glycomic proﬁling indicates marked changes during cardiomyocyte development. MALDI-TOF-MS proﬁles of the permethylated N-linked glycans of myocytes isolated from neonatal and adult atria and ventricles. (A) Neonatal atrium; (B) neonatal ventricle; (C) adult atrium; (D) adult ventricle. Major peaks are labeled, with the assigned compositions listed in Table S2. (Insets) Expansions of the mass region m/z 2950–3050 (highlighted), showing the regulated change in the relative levels of NeuAc and NeuGc sialic acids during development. Note for the two atrial samples, that the ratio of di-NeuAc to NeuAc/NeuGc to di-NeuGc determined by comparing the relative intensities of these three closely related peaks changes from Ϸ2.5:1.5:1 in the neonate to 1:1:2 in the adult

sample. A similar phenomenon is observed during ventricular development and is repeated throughout the spectra. Each peak was subjected to MSMS analysis PHYSIOLOGY to clarify their structure. All molecular ions are present in sodiated form ([MϩNa]ϩ).

current densities were no different in control versus ST8sia2(Ϫ/Ϫ) Discussion myocytes. Secondly, if Nav activation requires a greater depo- Regulated and Aberrant Glycosylation Modulate Cardiomyocyte Ex- (Ϫ/Ϫ) larization (as observed in ST8sia2 atrial myocytes), then the citability. The data indicate that ST8sia2 expression in neonatal AP should rise more slowly (as observed in Fig. 3A). In addition, atria is necessary for normal AP waveforms and normal Nav the slower Nav inactivation rate may increase APD, although sialylation and gating. Because ST8sia2 expression varies between effects of ST8sia2 expression on other ion channels also respon- atria and ventricles, the data suggest that the regulated expression sible for setting APD may be involved. of a single glycogene modulates cardiac function. Together, the data provide an example in which regulated and aberrant glycosylation ST8sia2 Expression Increases Level of Nav Sialylation. Immunoblot gel modulate cardiomyocyte excitability. shift analyses were performed to question whether altered AP Current dogma predicts that excitability is modulated through (Ϫ/Ϫ) waveform and modulated Nav function in ST8sia2 myocytes regulated expression and redistribution of ion channel isoforms. were caused by a direct effect of ST8sia2 expression on Nav Here, the data showed that Ͼ45% of the Ϸ250 cardiac glycogenes sialylation (Tables 1 and 2 and Fig. S3). The data indicated that the are significantly differentially expressed throughout the developing level of Nav sialylation is greater in control atrium than in myocardium. Further, the data indicated that the regulated and ST8sia2(Ϫ/Ϫ) atrium, with no significant difference in sialylation aberrant expression of one glycogene (of Ͼ100 regulated glyco- (Ϫ/Ϫ) levels observed between ST8sia2 and control ventricular Nav genes) was sufficient to alter cardiomyocyte AP waveforms and (Table 1 and Fig. S3A). Endoneuraminidase (EndoN) treatment, modulate less sialylated Nav gating consistently. This study describes used to remove polysialic acid chains of five or more residues (22, a mechanism by which cardiac function can be modulated through 23), indicated that only the control atrial Nav was polysialylated regulated and aberrant glycosylation. (Table 1 and Fig. S3B). These data suggest that ST8sia2 expression Individuals suffering from Chagas disease experience arrhyth- and function result in increased Nav sialylation. mias and conduction anomalies more frequently than those with

Montpetit et al. PNAS ͉ September 22, 2009 ͉ vol. 106 ͉ no. 38 ͉ 16519 Downloaded by guest on September 30, 2021 A A B conductance Normalized (+/+) Atrium (+/+) Ventricle 1.00 S8sia2(+/+) ST8sia2(+/+) 1.00 A (-/-) Atrium (-/-) Ventricle (-/-) * (-/-) ST8sia2 4 * 0.75 ST8sia2 0.75 Control AtriaKO Atria 3 0.50 0.50 # 55 # 2 55 0.25 50 0.25 (-mV) ] B, C 50 * ] (-mV) a 1 a 45 [V [V 40 mV 100 ms 0.00 45 40 0.00

ime to AP ime toPeak (ms) AP 0 Time to AP Peak (ms) Peak to AP Time T Normalized conductance -80 -60 -40 -20 -80 -60 -40 -20 Membrane potential (mV) Membrane potential (mV)

B # C # 25 60 C D current Normalized 1.00 # 105 95 1.00 * # 20 * 50 100 * 90 ] (-mV) ] (-mV) a a 95 85 (ms) 40 0.75 0.75

15 [V (ms) [V KO Vent 90 Control Vent 90 80

50 30 10 20 0.50 0.50 APD APD 5 10 0.25 * 0.25 0 0 0.00 0.00 Fig. 3. Regulated and aberrant expression of ST8sia2 impact AP waveforms. Normalized current Ϫ Ϫ -120 -100 -80 -60 -120 -100 -80 -60 APs were measured from isolated control and ST8sia2( / ) neonatal atrial and Prepulse Potential (mV) Prepulse potential (mV) ventricular myocytes as described. (Top left) Typical AP waveforms measured Inactivation Fast from control (in black) and ST8sia2(Ϫ/Ϫ) (in red) atrial myocytes. (A–C) The E F (ms) 6 * 6 τ mean Ϯ SEM AP waveform parameters (relevant portions of AP marked with # ϭ (Ϫ/Ϫ) ϭ arrows). Black, control atrium (n 7); red, ST8sia2 atrium (n 9); dark 4 4 Ϫ Ϫ

( / ) 1 nA # ϭ ϭ * 1 nA blue, control ventricle (n 7); blue, ST8sia2 ventricle (n 5). (A) Time to 2 ms 2 ms # AP peak (ms). (B) AP duration at 50% repolarization (APD50, ms). (C)AP 2 * # 2 * # τ

# (ms) duration at 90% repolarization (APD90, ms). Significance tested using a two- * * Ϫ Ϫ tailed t-test and comparing ST8sia2( / ) to control parameters. *, significant 0 0 (P Ͻ 0.005); #, not significant (P Ͼ 0.1). Fast Inactivation -60 -40 -20 0 20 40 -60 -40 -20 0 20 40 Membrane Potential (mV) Membrane Potential (mV) fast inactivationfast (ms) τ 10 G 10 recovery for from non-chagasic dilated cardiomyopathies (17, 24). A recent study H measured mouse ECG as a function of time postinfection using two 9 9 strains of Trypanosoma cruzi to infect (24). The data indicated that 8 * * # 8 Ϸ60% of infected mice (compared to Ϸ5% of control) showed 7 7 some conduction abnormality. For one of the T. cruzi strains, nearly 6 6 for recovery from for recovery

τ 5 5 all postinfection conduction anomalies involved aberrant atrial fast (ms) inactivation and/or atrioventricular conduction, observed as extended P waves Fig. 4. ST8sia2 expression modulates atrial Nav gating consistent with and PQ segments. Such ECG anomalies might be predicted with (Ϫ/Ϫ) changes in AP. ST8sia2 and control myocyte Nav gating. Left panels (A, slower AP upstroke and shifted Nav voltage dependence as ob- C, E, G) Atrium, ST8sia2(Ϫ/Ϫ) (red, n ϭ 10); control (black, n ϭ 4). Right panels (Ϫ/Ϫ) served here for ST8sia2 atrial myocytes, and are consistent (B, D, F, H) Ventricle, ST8sia2(Ϫ/Ϫ) (blue, n ϭ 4); control (dark blue, n ϭ 6). (Ϫ/Ϫ) with our findings for neonatal ST8sia2 ECGs (data not shown). Data are mean Ϯ SEM values. (A and B) Steady state activation. Data are peak normalized conductance at a membrane potential. Lines are fits of How Might Regulated and Aberrant Voltage-Gated Ion Channel Gly- data to single Boltzmann relationships. (Insets) Half-activation voltage. (C cosylation Impact Excitability? We show here that voltage- and D) Steady state channel availability. Data are the fraction of channels dependent gating parameters measured for the less sialylated available at a constant, fully-activating depolarization. Lines are fits of (Ϫ/Ϫ) ST8sia2(Ϫ/Ϫ) atrial Na were shifted to more depolarized potentials data to single Boltzmann relationships. (Dashed line) ST8sia2 data v shifted along voltage axis by Ϫ7.3 mV (to mimic measured shift in half- compared to control Nav gating parameters (Fig. 4). Such a shift in activation voltage). (Insets) Half-inactivation voltage. (E and F) Fast inac- steady-state activation and inactivation curves would lead to depo- tivation time constants. Lines are nontheoretical, point-to-point. (Dashed larization of the window current, defined as the voltage range at line) The ST8sia2(Ϫ/Ϫ) data shifted along voltage axis by Ϫ7.3 mV. (Insets) which a small percentage of channels are persistently active. Pre- Typical whole cell Naϩ current traces elicited by stepping to a Ϫ40 mV test vious reports suggest that point mutations of two Nav isoforms that potential. (G and H) Time constants of recovery from fast inactivation at a cause similar shifts in window current voltage range may be Ϫ130 mV recovery potential. Significance tested using a two-tailed t-test Ϫ Ϫ responsible for such maladies as epilepsy and LQTS (25, 26). Such and comparing ST8sia2( / ) to control parameters. *, significant (P Յ 0.005); Ͼ changes in window current will cause a shift in the voltages at which #, not significant (P 0.1). a persistent Naϩ current exists relative to the threshold and resting membrane potentials, and this could increase susceptibility to altered excitability. Together, the data suggest that a role for regulated cardiac Nav In addition, we observed a marked reduction in the rate of glycosylation (sialylation) is to help control and modulate cardiac (Ϫ/Ϫ) recovery from fast inactivation for the ST8sia2 atrial Nav. excitability and conduction achieved through a shift in the voltage Previous reports suggested that faster recovery rates increase the dependence of Nav window current and/or a change in Nav recovery probability of reentrant excitation, possibly resulting in LQT-3 (27) rate. or in idiopathic ventricular fibrillation (IVF) (28). Reentrant excitation and/or conduction block arrhythmias are prevalent in individuals suffering from Chagas disease (17). Here we report that Two Mechanisms by Which Regulated Ion Channel Glycosylation aberrant sialylation has a much larger impact on recovery rate than Modulate Cardiac Ion Channel Function and Cardiomyocyte Excitabil- that previously implicated in IVF etiology (28). At a Ϫ130 mV ity. We hypothesize that two mechanisms of regulated glycosylation (Ϫ/Ϫ) recovery potential, the less sialylated ST8sia2 atrial Nav recov- are responsible for modulation of voltage-gated ion channel activ- ers from inactivation 30% faster than the more heavily sialylated ity: (i) Inherent differences in glycosylation (unique glycosylation control atrial Nav, likely increasing the susceptibility to reentrant signatures) among ion channel subunits and isoforms as described excitation (Fig. 4G). previously (11, 12, 29) (subunit-specific) and (ii) regulated and

16520 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0905414106 Montpetit et al. Downloaded by guest on September 30, 2021 Table 1. Atrial Nav are more sialylated with ST8sia2 expression Atrium (n ϭ 8) Ventricle (n ϭ 8)

(ϩ/ϩ) ϩ (Ϫ/Ϫ) ϩ (ϩ/ϩ) ϩ (Ϫ/Ϫ) ϩ Myocyte type Sialidase (ϩ/ϩ)(Ϫ/Ϫ) Sialidase Sialidase (ϩ/ϩ)(Ϫ/Ϫ) Sialidase

Predicted MW (kDa) 241* Ϯ 2.8 256 Ϯ 2.3 246* Ϯ 3.2 240# Ϯ 2.5 233* Ϯ 2.7 243 Ϯ 2.8 243# Ϯ 2.2 234* Ϯ 1.8

(Ϫ/Ϫ) Predicted mean molecular weight (MW) Ϯ SEM for control (ϩ/ϩ) and ST8sia2 (Ϫ/Ϫ) atrial and ventricular Nav Ϯ sialidase treatment. Significance tested Ϫ Ϫ ϩ ϩ using a two-tailed t test to compare the following: (i) Untreated ST8sia2( / ) with ST8sia2 ( / ) samples, and (ii) Sialidase treated with untreated samples. *, significant (P Ͻ 0.01); #, not significant (P Ͼ 0.1).

aberrant changes in the ability of the cardiomyocyte to glycosylate Regulated Glycosylation Likely Occurs Across a Spectrum of Rates. It ion channel proteins as shown here (cell-specific). might be inferred from our data that regulated glycosylation occurs With respect to the first mechanism (subunit-specific), as differ- over a developmental timeframe or at the rate of protein synthesis. ent isoforms or additional subunits with unique glycosylation However, there is evidence that both mechanisms of differential ion signatures are expressed, so changes the level/location of sialic acids channel glycosylation take place on a more dynamic time scale. For and other sugar residues that may contribute to channel gating. An example, internalized surface proteins might reenter directly into the Golgi glycosylation pathway (34). There is also evidence that example of this is the differential expression of Na-v ␣ subunit extracellular sialyltransferase and glycosidase expression is regu- isoforms through skeletal muscle development. Nav1.4 is essentially lated and their activity may rapidly change the level of glycosylation the sole Nav ␣ subunit in adult skeletal muscle, while Nav1.5 is the dominant isoform expressed in embryonic and early postnatal attached to glycoproteins, resulting in a change in protein function (35–40). An example of this is the anti-aging protein, klotho, which muscle (30–32). Reports indicated that Na 1.4 is more heavily v has extracellular glycosidase activity (36–38, 40). Klotho expression glycosylated than Na 1.5 (11, 33). We showed that the greater v is highly regulated among tissues, throughout development and numbers of sialic acids attached to Nav1.4 impact the voltage aging, and with pathologies such as hypertension and acute myo- dependence of channel gating (11). Thus, through a switch in cardial infarction [for review, see (40)]. Thus, as klotho expression isoform, muscle Nav function is altered by the varied level/location is regulated, so changes klotho glycosidase activity. Klotho activity of Nav glycosylation. was shown to regulate surface expression of two renal epithelial ion ␤ ␤ We reported previously that sialic acids attached to the 1 and 2 channels, TRPV5 and ROMK1, by removing ␣2–6 sialic acids from Nav auxiliary subunits modulate Nav1.5 gating (12, 29). Our data the channels (36–38). This desialylation leads to stabilization of ␤ ␤ show that the regulated expression of Nav1.5 with 1 and/or 2 (or channels in the plasma membrane, thereby increasing channel neither) can establish a sialic acid-dependent Ϸ15 mV range of surface expression and ionic flux. In a second example, hippocam- channel activation voltages partially dependent on the array of pal network excitability is modulated by endogenous neuramini- glycosylation signatures created as the Nav complex is remodeled dase that removes external sialic acids (41). If similar mechanisms (e.g., Ϯ␤1 and/or Ϯ␤2). are involved in cardiomyocyte function, then extracellular glyco- Here we show that regulated and aberrant glycosylation syltransferases/glycosidases would likely produce rapid adjustments modulate voltage-gated ion channel activity through changes in to surface channel glycosylation. Thus, the rate of modulation of ion cardiomyocyte ion channel glycosylation, and this leads to altered channel gating imposed through regulated glycosylation would cardiomyocyte excitability, supporting the second mechanism (cell- occur across a spectrum of rates, ranging from relatively rapid specific). The data showed that the cardiac glycome is different in changes in surface glycan structures to the slower rate of protein the atria than in the ventricles and is remodeled differentially during synthesis to the much slower time course of development.

development of each chamber. The regulated and aberrant expres- PHYSIOLOGY sion of a single glycogene (ST8sia2) was sufficient to modulate the Summary. The data shown here indicate that the cardiac glycome is remodeled between atria and ventricles and during cardiac devel- cardiomyocyte AP and gating of less sialylated Na consistently. v opment. Glycogene expression is highly regulated, with Ͼ45% of Thus, there are two mechanisms by which regulated and aberrant glycogenes significantly differentially expressed among myocyte glycosylation modulate cardiac excitability, a subunit-specific mech- types. The N-glycans produced are also highly variable. The regu- anism as described previously, and regulated changes in the cardiac lated and aberrant expression of a single glycogene, ST8sia2, is glycome as described here. Both mechanisms likely occur simulta- apparently sufficient to alter AP waveform and Nav function by neously to produce a spectrum of possible ion channel gating changing the level of Nav sialylation. Given that expression of Ͼ100 mechanisms for the ion channel subunits responsible for contrib- glycogenes (including ST8sia2) are regulated throughout the de- uting to the AP waveform. Modulation of channel function by veloping myocardium, the potential impact of a remodeled glycome regulated or aberrant glycosylation results in a greater level of on cardiac function is profound. complexity of the orchestrated activity of these ion channels, Glycosylation exists in all cells and is likely regulated throughout leading to altered AP waveforms, and ultimately to changes in the body, suggesting that this paradigm is relevant to control and cardiac excitability. modulation of electrical signaling in other excitable tissues. There

Table 2. Atrial Nav are apparently polysialylated when ST8sia2 is expressed Atrium (n ϭ 7) Ventricle (n ϭ 8)

(ϩ/ϩ) ϩ (Ϫ/Ϫ) ϩ (ϩ/ϩ) ϩ (Ϫ/Ϫ) ϩ Myocyte type EndoN (ϩ/ϩ)(Ϫ/Ϫ) EndoN EndoN (ϩ/ϩ)(Ϫ/Ϫ) EndoN

Predicted MW (kDa) 236* Ϯ 2.7 260 Ϯ 3.7 237 Ϯ 2.8 238# Ϯ 2.5 240# Ϯ 0.8 238 Ϯ 1.8 236 Ϯ 2.0 235# Ϯ 1.8

(Ϫ/Ϫ) Predicted mean molecular weight (MW) Ϯ SEM for control (ϩ/ϩ) and ST8sia2 (Ϫ/Ϫ) atrial and ventricular Nav Ϯ EndoN treatment. Significance tested using a two-tailed t test comparing EndoN-treated samples with untreated samples. *, significant (P Ͻ 0.001); #, not significant (P Ͼ 0.3).

Montpetit et al. PNAS ͉ September 22, 2009 ͉ vol. 106 ͉ no. 38 ͉ 16521 Downloaded by guest on September 30, 2021 are diseases of aberrant sialylation such as Chagas disease and Glycomics (http://www.functionalglycomics.org), as described previously (47). CDGs that afflict nearly 20 million people and lead to altered N-glycans were released from extracted glycoproteins of tissue preparations by cardiac function consistent with the data presented here. In addi- PNGase F digestion, permethylated using the sodium hydroxide procedure, and tion, if glycome remodeling occurs during the pathogenesis of such puriﬁed by Sep-Pak C18 cartridges, as previously described (47). disorders as LQTS, heart failure, or diabetes, then the resulting Derivatized glycan samples were dissolved in methanol/water 8:2 (vol/vol) and mixed in a 1:1 ratio with 10 mg/mL 2,5-dihydroxybenzoic acid in 80:20 (vol/vol) aberrant glycosylation might contribute to the relative activity of ␮ cardiac glycoproteins that impact cardiac electrical signaling and methanol/water. A total of 0.5- to 1- L aliquots were spotted onto a target plate and dried under vacuum. MS spectra were obtained using a Voyager DE STR thereby the symptoms associated with these disorders. Together, MALDI-TOF (Applied Biosystems) mass spectrometer in the reﬂectron mode with these data describe a mechanism by which cardiac function and delayed extraction. Peaks observed in the MS spectra were selected for further likely function of other excitable tissues are controlled and modu- MS/MS. MS/MS data were acquired using a 4800 MALDI TOF/TOF (Applied Bio- lated by regulated and aberrant glycosylation. systems) mass spectrometer.

Materials and Methods Electrophysiology. Neonatal cardiomyocytes were isolated as described previ- Please refer to the SI Text for detailed methods. ously (15). APs were recorded at room temperature using an Axopatch 200B amplifier (Axon Instruments) and pCLAMP software (Axon Instruments). An Microarray Analysis. Analysis of gene expression was conducted using a custom Ϸ700 pA supratheshold depolarizing current was injected for 1 ms from a Ϫ80 mV gene microarray (GLYCOv2 chip) produced by Affymetrix for the Consortium for resting membrane potential to elicit APs. Recording solutions were described Functional Glycomics (http://www.functionalglycomics.org) (42). Hybridization previously (9). Nav recordings were performed as described previously (15). and scanning of the chip were performed according to the recommended pro- tocols (43). Immunoblot. Immunoblot analysis was performed on isolated atrial and ven- Array normalization was performed using invariant set normalization avail- tricular myocytes as described previously (15). able in DNA-Chip Analyzer (dChip) (www.dchip.org) software package (44, 45). ANOVA was performed using BRB Array Tools, developed by Dr. Richard Simon ACKNOWLEDGMENTS. We thank Dr. Minoru Fukuda (Burnham Institute for and Amy Peng Lam. Heatmaps were generated with dChip program (44, 45). The Medical Research, La Jolla, CA) for generously providing the endoneuraminidase validity of the GLYCO chips was verified using qPCR (Fig. S2). Previous reports also used. This work was supported in part by grants from the National Institute for confirmed validity of the chips (46). Arthritis, Musculoskeletal, and Skin Diseases (to E.S.B.), the National Institute for General Medical Sciences (Consortium for Functional Glycomics), the American Glycomic Analysis. Myocytes isolated from neonatal and adult atria and ventri- Heart Association (to E.S.B.), and the James and Esther King Florida Biomedical cles were processed and analyzed by Core C of the Consortium for Functional Research Program (to E.S.B.).

1. Freeze HH (2006) Genetic defects in the human glycome. Nat Rev Genet 7:537–551. 24. Bustamante JM, Rivarola HW, Fretes R, Paglini-Oliva PA (2005) Weekly electrocardio- 2. Ohtsubo K, Marth JD (2006) Glycosylation in cellular mechanisms of health and disease. graphic pattern in mice infected with two different Trypanosoma cruzi strains. Int Cell 126:855–867. J Cardiol 102:211–217. 3. Marionneau C, et al. (2008) Distinct cellular and molecular mechanisms underlie 25. Spampanato J, Escayg A, Meisler MH, Goldin AL (2003) Generalized epilepsy with functional remodeling of repolarizing Kϩ currents with left ventricular hypertrophy. febrile seizures plus type 2 mutation W1204R alters voltage-dependent gating of Na(v) Circ Res 102:1406–1415. 1.1 sodium channels. Neuroscience 116:37–48. 4. Grandy SA, Trepanier-Boulay V, Fiset C (2007) Postnatal development has a marked effect 26. Abriel H, et al. (2001) Novel arrhythmogenic mechanism revealed by a long-QT syn- on ventricular repolarization in mice. Am J Physiol Heart Circ Physiol 293:H2168–H2177. drome mutation in the cardiac Na(ϩ) channel. Circ Res 88:740–745. 5. Plotnikov AN, et al. (2004) Cardiac memory evolves with age in association with 27. Nuyens D, et al. (2001) Abrupt rate accelerations or premature beats cause life- development of the transient outward current. Circulation 110:489–495. threatening arrhythmias in mice with long-QT3 syndrome. Nat Med 7:1021–1027. 6. Nattel S, Maguy A, Le BS, Yeh YH (2007) Arrhythmogenic ion-channel remodeling in 28. Chen Q, et al. (1998) Genetic basis and molecular mechanism for idiopathic ventricular the heart: Heart failure, myocardial infarction, and atrial fibrillation. Physiol Rev fibrillation. Nature 392:293–296. 87:425–456. 29. Johnson D, Montpetit ML, Stocker PJ, Bennett ES (2004) The sialic acid component of 7. Roberts RH, Barchi RL (1987) The voltage-sensitive sodium channel from rabbit skeletal the beta1 subunit modulates voltage-gated sodium channel function. J Biol Chem muscle. Chemical characterization of subunits. J Biol Chem 262:2298–2303. 279:44303–44310. 8. Schmidt JW, Catterall WA (1987) Palmitylation, sulfation, and glycosylation of the 30. Yang JS, Sladky JT, Kallen RG, Barchi RL (1991) TTX-sensitive and TTX-insensitive sodium alpha subunit of the sodium channel. Role of post-translational modifications in channel mRNA transcripts are independently regulated in adult skeletal muscle after channel assembly. J Biol Chem 262:13713–13723. denervation. Neuron 7:421–427. 9. Ufret-Vincenty CA, et al. (2001) Role of sodium channel deglycosylation in the genesis 31. Goldin AL (2001) Resurgence of sodium channel research. Annu Rev Physiol 63:871– of cardiac arrhythmias in heart failure. J Biol Chem 276:28197–28203. 894. 10. Ufret-Vincenty CA, Baro DJ, Santana LF (2001) Differential contribution of sialic acid to 32. Trimmer JS, Cooperman SS, Agnew WS, Mandel G (1990) Regulation of muscle sodium ϩ the function of repolarizing K( ) currents in ventricular myocytes. Am J Physiol Cell channel transcripts during development and in response to denervation. Dev Biol Physiol 281:C464–C474. 142:360–367. 11. Bennett ES (2002) Isoform-specific effects of sialic acid on voltage-dependent Naϩ 33. Cohen SA, Levitt LK (1993) Partial characterization of the rH1 sodium channel protein channel gating: Functional sialic acids are localized to the S5–S6 loop of domain I. from rat heart using subtype-specific antibodies. Circ Res 73:735–742. J Physiol 538:675–690. 34. Snider MD, Rogers OC (1986) Membrane traffic in animal cells: Cellular glycoproteins 12. Johnson D, Bennett ES (2006) Isoform-specific effects of the beta2 subunit on voltage- return to the site of Golgi mannosidase I. J Cell Biol 103:265–275. gated sodium channel gating. J Biol Chem 281:25875–25881. 35. Taatjes DJ, Roth J, Weinstein J, Paulson JC (1988) Post-Golgi apparatus localization and 13. Johnson D, Bennett ES (2007) Gating of the shaker potassium channel is modulated regional expression of rat intestinal sialyltransferase detected by immunoelectron differentially by N-glycosylation and sialic acids. Pflugers Arch 456:393–405. microscopy with polypeptide epitope-purified antibody. J Biol Chem 263:6302–6309. 14. Watanabe I, et al. (2007) The glycosylation state of Kv1.2 potassium channels affects trafficking, gating, and simulated action potentials. Brain Res 1144:1–18. 36. Chang Q, et al. (2005) The beta-glucuronidase klotho hydrolyzes and activates the 15. Stocker PJ, Bennett ES (2006) Differential sialylation modulates voltage-gated Naϩ TRPV5 channel. Science 310:490–493. channel gating throughout the developing myocardium. J Gen Physiol 127:253–265. 37. Cha SK, et al. (2008) Removal of sialic acid involving Klotho causes cell-surface retention 16. Libby P, Alroy J, Pereira ME (1986) A neuraminidase from Trypanosoma cruzi removes of TRPV5 channel via binding to galectin-1. Proc Natl Acad Sci USA 105:9805–9810. ϩ sialic acid from the surface of mammalian myocardial and endothelial cells. J Clin Invest 38. Cha SK, et al. (2009) Regulation of ROMK1 channel and renal K excretion by Klotho. 77:127–135. Mol Pharmacol 76:38–46. 17. Yacoub S, Mocumbi AO, Yacoub MH (2008) Neglected tropical cardiomyopathies: I. 39. Gross HJ, Merling A, Moldenhauer G, Schwartz-Albiez R (1996) Ecto-sialyltransferase of Chagas disease: Myocardial disease. Heart 94:244–248. human B lymphocytes reconstitutes differentiation markers in the presence of exog- 18. Marquardt T, Denecke J (2003) Congenital disorders of glycosylation: Review of their enous CMP-N-acetyl neuraminic acid. Blood 87:5113–5126. molecular bases, clinical presentations and specific therapies. Eur J Pediatr 162:359–379. 40. Wang Y, Sun Z (2009) Current understanding of klotho. Ageing Res Rev 8:43–51. 19. Kranz C, et al. (2007) Expanding spectrum of congenital disorder of glycosylation Ig 41. Isaev D, et al. (2007) Role of extracellular sialic acid in regulation of neuronal and (CDG-Ig): Sibs with a unique skeletal dysplasia, hypogammaglobulinemia, cardiomyopa- network excitability in the rat hippocampus. J Neurosci 27:11587–11594. thy, genital malformations, and early lethality. Am J Med Genet A 143A:1371–1378. 42. Bax M, et al. (2007) Dendritic cell maturation results in pronounced changes in glycan 20. Wada Y, et al. (2007) Comparison of the methods for profiling glycoprotein glycans— expression affecting recognition by siglecs and galectins. J Immunol 179:8216–8224. HUPO Human Disease Glycomics/Proteome Initiative multi-institutional study. Glyco- 43. Lockhart DJ, et al. (1996) Expression monitoring by hybridization to high-density biology 17:411–422. oligonucleotide arrays. Nat Biotechnol 14:1675–1680. 21. Angata K, et al. (2004) Sialyltransferase ST8Sia-II assembles a subset of polysialic acid 44. Li C, Wong WH (2001) Model-based analysis of oligonucleotide arrays: Expression index that directs hippocampal axonal targeting and promotes fear behavior. J Biol Chem computation and outlier detection. Proc Natl Acad Sci USA 98:31–36. 279:32603–32613. 45. Li C, Hung WW (2001) Model-based analysis of oligonucleotide arrays: Model valida- 22. Hallenbeck PC, Vimr ER, Yu F, Bassler B, Troy FA (1987) Purification and properties of tion, design issues and standard error application. Genome Biol 2:RESEARCH0032. a bacteriophage-induced endo-N-acetylneuraminidase specific for poly-alpha-2,8- 46. Comelli EM, et al. (2006) A focused microarray approach to functional glycomics: sialosyl carbohydrate units. J Biol Chem 262:3553–3561. Transcriptional regulation of the glycome. Glycobiology 16:117–131. 23. Troy FA, Hallenbeck PC, McCoy RD, Vimr ER (1987) Detection of polysialosyl-containing 47. Jang-Lee J, et al. (2006) Glycomic profiling of cells and tissues by mass spectrometry: glycoproteins in brain using prokaryotic-derived probes. Methods Enzymol 138:169–185. Fingerprinting and sequencing methodologies. Methods Enzymol 415:59–86.

16522 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.0905414106 Montpetit et al. Downloaded by guest on September 30, 2021