LITAF Regulates Cardiac L-type Calcium Channels by Modulating NEDD4-1 Ubiquitin Ligase.

Background - The turnover of cardiac ion channels underlying action potential duration (APD) is regulated by ubiquitination. Genome-wide association studies (GWAS) of QT interval identified several single-nucleotide polymorphisms located in or near genes involved in protein ubiquitination. A genetic variant upstream of LITAF (lipopolysaccharide-induced tumor necrosis factor) gene prompted us to determine its role in modulating cardiac excitation. Methods - Optical mapping was performed in zebrafish hearts to determine Ca2+ transients. Live cell confocal calcium imaging was performed on adult rabbit ventricular myocytes (ARbCM) to determine intracellular Ca2+handling. LTCC (L-type calcium channel) current was measured using whole cell recording. To study the effect of LITAF on Cav1.2 channel expression, surface biotinylation and westerns were performed. LITAF interactions were studied using co-immunoprecipitation and in situ proximity ligation assay (PLA). Results - LITAF knockdown in zebrafish resulted in a robust increase in calcium transients. Overexpressed LITAF in 3-week-old rabbit cardiomyocytes resulted in a decrease in ICa,L and Cavα1c abundance, whereas LITAF knockdown increased ICa,L and Cavα1c protein. LITAF-overexpressing decreases calcium transients in ARbCM, which was associated with lower Cavα1c levels. In tsA201 cells, overexpressed LITAF downregulated total and surface pools of Cavα1c via increased Cavα1c ubiquitination and its subsequent lysosomal degradation. We observed co-localization between LITAF and LTCC in tsA201 and cardiomyocytes. In tsA201, NEDD4-1, but not its catalytically inactive form NEDD4-1-C867A, increased Cavα1c ubiquitination. Cavα1c ubiquitination was further increased by co-expressed LITAF and NEDD4-1 but not NEDD4-1-C867A. NEDD4-1 knockdown abolished the negative effect of LITAF on ICa,L and Cavα1c levels in 3-week-old rabbit cardiomyocytes. Computer simulations demonstrated that a decrease of ICa,L current associated with LITAF overexpression simultaneously shortened APD and decreased calcium transients in rabbit cardiomyocytes. Conclusions - LITAF acts as an adaptor protein promoting NEDD4-1-mediated ubiquitination and subsequent degradation of LTCC, thereby controlling LTCC membrane levels and function and thus cardiac excitation.

I on channel proteins are regulated by various types of posttranslational modifications. In most cases, ubiquitination acts as a signal for endocytosis of the ion channels, which are subsequently degraded through lysosomal or proteasomal-dependent pathways. Ubiquitination also occurs at the exit of endoplasmic reticulum and the trans-Golgi-network. [1][2][3][4] Several genome-wide association studies for loci that modify the QT interval and the risk for sudden cardiac death 5,6 have identified 3 single-nucleotide polymorphisms located in or near genes involved in protein ubiquitination: (1) the RNF207 (Ring finger protein 207) ubiquitin ligase 7,8 ; (2) RFFL (Ring finger and FYVE [Fab 1, YOTB, Vac 1, and EEA1]like domain-containing E3 ubiquitin-protein ligase); 9,10 and (3) LITAF (lipopolysaccharide-induced tumor necrosis factor), a regulator of endosomal trafficking [11][12][13] and inflammatory cytokines, 14,15 and an adapter molecule for members of the NEDD4 (neural precursor cell expressed developmentally down-regulated protein 4)-like family of E3 ubiquitin ligases. 16,17 The genetic variant rs8049607 ( Figure I in the Data Supplement) was associated with a modest QT-interval-prolongation effect (1.2 ms, P=5×10 -15 ). 5,6 It is ≈41 kb upstream of the LITAF start codon 18 and located within an intergenic enhancer region. 19 Furthermore, data from expression quantitative trait loci analyses showed that rs8049607 was also associated with reduced LITAF mRNA transcript levels in the left ventricle ( Figure I in the Data Supplement). 20 LITAF is a mediator of local and systemic inflammatory responses 15 and is locally upregulated in many inflammatory diseases, such as Crohn disease and ulcerative colitis. 21 Notably, whole-body LITAF deletion diminished experimental endotoxic shock and inflammatory arthritis in mice. 15 Importantly, loss-of-function mutations in LITAF cause Charcot-Marie-Tooth disease, an inherited peripheral neuropathy. These mutations 22,23 are clustered around the hydrophobic region required for membrane localization and cause mislocalization and impaired endosome-tolysosome trafficking of membrane proteins 11,24 (Figure 1A).
Although there has been some controversy as to the activity of LITAF as a transcription factor, 14,21,25 many studies have established its functional role in endosomal trafficking and multivesicular body formation. 11,12,26 Indeed, LITAF interacts with members of the ESCRT (endosomal sorting complex required for transport), including TSG101 (tumor susceptibility gene 101) (via LITAF's tetrapeptide motif, P[S/T]AP; Figure 1A) and STAM1 (signal transducing adaptor molecule 1) (physical interaction with LITAF), recruits them to the early endosomal membrane, and controls endosome-to-lysosome trafficking and exosome formation. 11 The N-terminus of LITAF contains 2 PXY motifs ( Figure 1A), which are important for interacting with members of the NEDD4 family of HECT (homologous to the E6-AP carboxyl terminus) domain ubiquitin ligases via their WW domains. 12,27 Based on the genome-wide association studies' findings 5, 6 and LITAF's functional role in endosome-to-lysosome trafficking, we hypothesized that LITAF is a candidate for regulation of cardiac excitation, likely acting as an effector of ion channel complex trafficking or degradation through lysosomes. Therefore, we set out to investigate the possible role of LITAF in the regulation of ion channels in zebrafish heart and rabbit cardiomyocytes. In this study, we present data that support a role for LITAF in modulating membrane abundance and function of voltage-gated L-type calcium channels (LTCCs) via the ubiquitin ligase NEDD4-1 in the heart.

METHODS
We declare that all supporting data are available within the article (and its Data Supplement). All animal experiments and procedures were approved by the Rhode Island Hospital Institutional Animal Care and Use Committee. Experiments performed on zebrafish (Danio rerio) are in accordance with animal protocols approved by the Harvard Medical School Institutional Animal Care and Use Committee. An expanded Methods section is found in the Data Supplement.

Genetic Knockdown of LITAF in Larval Zebrafish Hearts Affects Calcium Transients
In an effort to delineate the effects of cardiac LITAF in vivo, we designed a morpholino oligomer to the zebrafish ortholog (LITAF; Figure 1A) targeting the ATG. In initial dose-ranging studies, there was no evidence of cardiac or systemic toxicity (data not shown). Electrophysiological studies of the ventricular myocardium revealed that a LITAF knockdown in zebrafish larvae caused a robust increase in the amplitude of [Ca 2+ ] i transients compared with control morphants at 48 hours post-fertilization ( Figure 1B through 1E). These data suggest that LITAF modulates Ca 2+ handling in zebrafish. Because the phasic [Ca 2+ ] i transient in zebrafish largely depends on Ca 2+ influx through transmembrane Ca 2+ channels, 28 we reasoned that LITAF may be critical to regulating voltage-gated LTCC in the heart.

Effect of LITAF Overexpression on Ca 2+ Cycling in Adult Rabbit Cardiomyocytes
Because our in vivo observations in zebrafish embryos showed an increase in Ca 2+ transients in LITAF morphants ( Figure 1B through 1E), we expected LITAF to also interfere with intracellular Ca 2+ handling in rabbit cardiomyocytes. Therefore, we utilized cultured adult rabbit cardiomyocytes, which in our hands and in agreement with Tian et al 29 largely preserve t-tubules in culture (data not shown), to study the effect of adenovirally expressed LITAF on cardiac Ca 2+ cycling.
We performed live-cell confocal imaging and measured the amplitude of electrically evoked Ca 2+ transients in cells overexpressing LITAF or GFP (green fluorescence protein) as control (Figure 2A and 2B). To determine the sarcoplasmic reticulum-Ca 2+ load and assess NCX (Na + /Ca 2+ exchanger) function, 20 mmol/L caffeine was applied at the end of the experiments. We observed that LITAF overexpression in adult rabbit cardiomyocytes significantly decreased the amplitude of Ca 2+ transients (Figure 2A and 2B). This decrease was paralleled by a decrease in sarcoplasmic reticulum Ca 2+ content assessed by rapid application of 20 mmol/L caffeine. There were no changes in fractional release of Ca 2+ (Figure 2B). NCX activity calculated by measuring the rate of decay of caffeine-induced Ca 2+ transients was not different between the groups ( Figure 2B). No change in SERCA2 (sarco/endoplasmic reticulum Ca 2+ -ATPase 2) activity was found as well ( Figure 2B), which was calcu-lated via a derived rate of decay, subtracting the rate of decay of caffeine-induced Ca 2+ transients from the rate of decay of pacing-induced Ca 2+ transients. This is corroborated by Western blot data showing no significant changes in total levels of SERCA2, NCX, calsequestrin 2, or serine 16-phosphorylated phospholamban upon LITAF overexpression in cardiomyocytes ( Figure II in the Data Supplement). Together, these data indicate that the reduction in Ca 2+ transients is likely due to reduced Ca 2+ influx, which is supported by a significant downregulation of the total pool of Cavα1c in LITAF-overexpressing cells ( Figure 2C and 2D).

LITAF Modulates I Ca,L and Cavα1c Protein in 3-Week-Old Cardiomyocytes
To study any effect of LITAF on Cavα1c and LTCC current (I Ca,L ) in detail, we switched to cultured 3-week-old   19 ΔF/F 0 ), fractional release and rates of Ca 2+ removal by NCX (Na + /Ca 2+ exchanger; k caff ), and SERCA2 (sarco/endoplasmic reticulum Ca 2+ -ATPase 2; k SR ). Student t test, P<0.05 (2-3 heart preparations). C, Adult rabbit cardiomyocytes lysates were probed with anti-Cavα1c, anti-HA, and anti-GAPDH to indicate Cavα1c, exogenous LITAF and GAPDH (loading control) protein levels. D, Respective change in Cavα1c abundance, normalized to GAPDH (n=5 animals, performed in triplicate; mean±SEM). Student t test, P<0.05. rabbit cardiomyocytes. We have recently developed this model system to investigate various ion channels underlying action potential duration (APD; data not shown). 30 For example, these cells exhibit stable I Ca,L current after 48 hours in culture ( Figure 3A and 3B). The cells were transduced with adenovirus encoding GFP or HA (hemagglutinin)-LITAF. LITAF overexpression caused a significant 45% decrease in maximal I Ca,L density (from −10.2±0.9 pA/pF to −5.6±0.3 pA/pF; P<0.05; Figure 3A and 3B) with no changes in voltage-dependent activation or inactivation kinetics ( Figure III in the Data Supplement). Consistent with the electrophysiological data, Cavα1c protein levels were significantly downregulated by 45% in 3wRbCM (3-week-old rabbit cardiomyocytes) overexpressing LITAF (P<0.05; Figure 3C). By contrast, suppression of LITAF levels by short hairpin RNA (shRNA) resulted in a significant 40% upregulation in maximal I Ca,L density (from −2.2±0.3 pA/pF to −3.1±0.4 pA/pF; P<0.05; Figure 3D), which correlates with a significant 64% increase in Cavα1c abundance ( Figure 3E) and a 30% downregulation of endogenous LITAF ( Figure 3E). These data corroborate our findings in zebrafish and validate usage of both systems in the studies of LITAF effects.
We also transduced neonatal rabbit cardiomyocytes with adenovirus encoding GFP (as a control) and HAtagged LITAF and determined respective Cavα1c protein levels. Similar to adult rabbit cardiomyocytes and 3wRbCM, we observed a significant downregulation of total Cavα1c protein levels by LITAF ( Figure IVA in the Data Supplement). In addition, we used adenovirally expressed shRNA to knockdown levels of endogenous LITAF in neonatal cardiomyocytes. After 48 hours, the

LITAF Has No Significant Effect on the Major Repolarizing K + Current, I Kr (delayed rectifier potassium current) in 3wRbCM
The aforementioned genome-wide association studies 5,6 implied a role for LITAF in QT interval and, therefore, APD regulation. Interestingly, we have previously shown that 2 other genes identified in these studies, the RING finger ubiquitin ligases RNF207 and RFFL, affect a major repolarizing current in larger animals, (viz I Kr 7,9 ). Therefore, we set out to look for LITAF-dependent effects on this current. To this end, 3wRbCM were transduced with adenovirus encoding GFP or HA-LITAF and cultured for 48 hours. LIT-AF overexpression had no significant effect on I Kr current (−30 mV: control: 0.72±0.13 pA/pF; LITAF: 0.77±0.14 pA/pF; P=0.87; 4 animals). Consistent with the electrophysiological data, we observed no change in the protein levels of human ether-à-go-go-related gene (HERG), which underlies I Kr ( Figure VA and VB in the Data Supplement). Furthermore, our Western blot data for other K + channels, that is, KvLQT1 (voltage-gated K channel gene causing long QT1 syndrome), Kir2.1 (inward-rectifying potassium channel 2.1), Kv4.2 (voltage-gated potassium channel 4.2), and Kv1.4 (voltage-gated potassium channel 1.4), showed no changes with LITAF overexpression (Figure VC through VH in the Data Supplement). In summary, our data suggest a quite specific effect of LITAF on LTCC rather than potassium channels in cardiomyocytes.

Physical and Functional Interaction Between LITAF and L-type Ca 2+ Channels in tsA201 Cells
To explore the mechanisms underlying LITAF inhibition of LTCC, we used a heterologous expression system, viz tsA201 cells, which are frequently used to study LTCC in vitro because they process the multisubunit complex correctly and efficiently. 31 We transfected tsA201 cells with expression plasmids for Cavα1c, Cavβ3, and Cavα2δ-1 to express functional LTCC and plasmid for HA-tagged LITAF or empty control plasmid. Cell-surface biotinylation was performed on transfected cells and protein levels of Cavα1c in total lysates and cell-surface fractions determined ( Figure 4A and 4B). Overexpression of LITAF resulted in a significant downregulation of both total and surface-membrane levels of Cavα1c.
In contrast, no changes in cell-surface levels of Cavβ3 and Cavα2δ-1 were observed ( Figure 4A and 4B). To see whether the functional interaction between LITAF and LTCC is based on a physical interaction, we performed coimmunoprecipitations with extracts from tsA201 cells cotransfected with expression plasmids for all 3 LTCC subunits and HA-tagged LITAF. Extracts were incubated with HA antiserum or isotype control, and immunoprecipitates probed with anti-Cavβ3 (Figure 5A) or anti-Cavα1c ( Figure 5B) antibody. Western blot analyses suggest that LITAF is found in a protein complex with Cavα1c, as well as Cavβ3. Furthermore, to confirm colocalization between endogenous LITAF and LTCC in rabbit cardiomyocytes, we performed in situ proximity ligation assay. Representative images are shown in Figure 5. Specificity of the assay was shown by the lack of staining using mouse anti-LITAF or rabbit anti-LTCC as negative controls ( Figure 5C). Similarly, no signals were obtained omitting primary antibodies from the assay ( Figure 5C). As a positive control for the assay, we used rabbit anti-Cavα2δ-1 and mouse anti-Cavα2δ-1 to detect endogenous Cavα2δ-1 ( Figure 5C). Using the combination of rabbit anti-LITAF and mouse anti-Cavα1c, the appearance of puncta suggests colocalization between LITAF and Cavα1c in cardiomyocytes ( Figure 5C). Similar results were obtained using mouse anti-LITAF and rabbit anti-Cavα1c ( Figure 5C).

LITAF-Dependent Regulation of Ubiquitination and Degradation of Cavα1c in tsA201 Cells
Because ubiquitination plays an important role in all 3 major protein degradation pathways, 32 we examined whether Cavα1c was ubiquitinated in a LITAF-dependent manner. We reconstituted LTCC by transfecting tsA201

. Physical interaction between LITAF (lipopolysaccharide-induced tumor necrosis factor) and L-type calcium channel (LTCC) in tsA201 cells and 3-week-old rabbit cardiomyocytes (3wRbCM).
A, Immunoprecipitation (IP) of lysates from tsA201 cells transfected with plasmids for Cavα1c (L-type calcium channel alpha-1C subunit), Cavβ3, Cavα2δ-1, GFP (green fluorescence protein), or HA (hemagglutinin)-tagged LITAF using isotype control (lane 1) or HA antibody (lanes 2 and 3). A representative immunoblot against Cavβ3 shows an interaction between LITAF and the Cavβ3 subunit (IP; the asterisk indicates the heavy chain of the IP capture antibody). Also shown is the immunoprecipitated HA-LITAF protein. Input levels of Cavβ3, HA-LITAF, and tubulin are shown below. B, IP of lysates from tsA201 cells transfected with plasmids for Cavα1c, Cavβ3, Cavα2δ-1, GFP, or HA-tagged LITAF using HA antibody. A representative immunoblot against Cavα1c shows an interaction between LITAF and the Cavα1c subunit (IP). Also shown is the immunoprecipitated HA-LITAF protein. Input levels of Cavα1c, HA-LITAF, and tubulin are depicted below. C, Duo-link in situ proximity ligation assay using rabbit anti-LITAF and mouse anti-Cavα1c antibodies (alternatively mouse anti-LITAF and rabbit anti-Cavα1c antibodies) in 3wRb-CM, which are amenable to proximity ligation assay and express detectable levels of LITAF and LTCC. Colocalization between molecules is indicated by red puncta. No puncta were detected in negative controls in which primary antibodies were omitted or only one antibody was used (rabbit anti-LITAF, mouse anti-Cavα1c, mouse anti-LITAF, or rabbit anti-Cavα1c antibodies). As positive control for the assay, a combination of rabbit polyclonal anti-Cavα2δ-1 and mouse monoclonal anti-Cavα2δ-1 was used to detect endogenous Cavα2δ-1. Nuclei were stained with DAPI (4',6-diamidino-2-phenylindole) (blue). cells with expression plasmids encoding Cavα1c, Cavβ3, Cavα2δ-1, HA-tagged ubiquitin, and Flag-tagged LITAF, or empty control plasmid. Using anti-HA antibody to pull down ubiquitinated protein from cell lysates, followed by immunodetection of Cavα1c, we observed a LITAFdependent increase in the ubiquitination level of Cavα1c ( Figure 6A, left). Although total Cavα1c abundance was significantly reduced upon coexpression of LITAF, no changes in total Cavβ3 and Cavα2δ-1 levels were seen ( Figure 6A, right). Next, we treated cells expressing functional LTCC and LITAF or control plasmid with chloroquine or MG132 (N-benzyloxycarbonyl-L-leucyl-L-leucyl-L-leucinal) to determine which protein degradation pathway is involved in LITAF-mediated Cavα1c downregulation. The lysosomal inhibitor chloroquine completely prevented the effect of LITAF on Cavα1c protein levels, whereas the proteasomal inhibitor MG132 did not impair downregulation of Cavα1c by LITAF ( Figure 6B).

Functional Interaction Between LITAF, the Ubiquitin Ligase NEDD4-1, and L-type Ca 2+ Channels in 3-Week-Old Cardiomyocytes
Because our data obtained from tsA201 cells suggest a possible functional interaction between LITAF and NEDD4-1 ubiquitin ligase in LTCC degradation, we designed adenovirus-expressing shRNA against rabbit NEDD4-1. Transducing 3wRbCM with NEDD4-1 shRNA, endog-enous NEDD4-1 protein was significantly downregulated and a concomitant increase in total Cavα1c levels observed ( Figure 7A). To assess a possible requirement for NEDD4-1 in LITAF-mediated downregulation of LTCC in 3wRbCM, we transduced cells with adenovirus expressing GFP and NEDD4-1 shRNA (control) or LITAF and NEDD4-1 shRNA. I Ca,L densities in cardiomyocytes expressing both LITAF and NEDD4-1 shRNA and cells expressing NEDD4-1 shRNA and GFP were virtually identical ( Figure 7B), implying that NEDD4-1 is crucial for LITAF-mediated I Ca,L regulation. To corroborate our findings in 3wRbCM, we also transduced cells with adenovirus expressing LITAF and scrambled RNA (control) or adenovirus-expressing LITAF and NEDD4-1 shRNA ( Figure 7C). In the presence of coexpressed NEDD4-1 shRNA, LITAF overexpression caused a significant 45% increase in peak I Ca,L density compared with the control (from −3.1±0.4 pA/pF to −4.5±0.5 pA/ pF; P<0.05), suggesting that the negative effect of LITAF on I Ca,L densities requires the presence of NEDD4-1.

Decrease of I Ca,L Is Responsible for Shortening of APD and Decreasing Ca 2+ Transient Amplitude in Myocytes With LITAF Overexpression
To study the causal relationship between LITAF overexpression and changes of electrophysiological pheno-

NEDD (neural precursor cell expressed developmentally downregulated protein) 4-1-dependent downregulation of L-type calcium channel (LTCC) by LITAF (lipopolysaccharide-induced tumor necrosis factor) in 3-wk-old rabbit cardiomyocytes.
A, Protein levels of total Cavα1c (L-type calcium channel alpha-1C subunit), NEDD4-1, and tubulin in cells expressing scrambled control RNA or shRNA against endogenous NEDD4-1 (left; the asterisk indicates an unspecific band). Respective changes in NEDD4-1 and Cavα1c abundance, normalized to tubulin (n=5 animals, performed in triplicate; mean±SEM). Student t test, P<0.05 (right). B, Current-voltage relationships of LTCC current (I Ca,L ) peak currents for baseline conditions from cells expressing GFP (green fluorescence protein) and short hairpin RNA (shRNA) against endogenous NEDD4-1 (control) or LITAF and NEDD4-1 shRNA. C, Threeweek-old rabbit cardiomyocytes were transduced with adenovirus expressing scrambled RNA and LITAF (control) or LITAF and shRNA against endogenous NEDD4-1. Current-voltage relationships of I Ca,L peak currents for baseline conditions from respective cells are depicted (cells from 5 animals; mean±SEM; Student t test, P<0.05).
type, we used a physiologically detailed computational model of rabbit ventricular myocyte with membrane voltage coupled to spatially distributed subcellular Ca 2+ dynamics. Details of the model are provided in the Data Supplement. We first determined the I Ca,L current conductances in the model that reproduces the I Ca,L peak current versus voltage curves measured in voltageclamp mode ( Figure 3B) for myocytes with GFP and LITAF. We found that reducing the number of functional sarcolemmal LTCC from 4 to 2 in each calcium release unit, where LTCCs are colocalized with RyR2 (ryanodine receptor 2) Ca 2+ release channels, reproduced the ≈50% reduction of whole-cell I Ca,L current with LITAF compared with GFP. The whole-cell current is the summation of LTCCs from ≈16 000 calcium release units spatially distributed throughout the cell. Results in Figure 8A show that, with this 50% reduction in the total number of LTCCs, the model reproduces well the quantitative voltage-clamp measurements of Figure 3B. We then paced the myocytes at a 2.5 Hz in current-clamp mode for I Ca,L conductances (total number of LTCCs) corresponding to GFP and LITAF. The results show a 50% decrease of I Ca,L conductance, resulting from LITAF overexpression, significantly reduces both local Ca 2+ release (confocal line-scan equivalent in Figure 8B) and whole-cell Ca 2+ transient amplitude ( Figure 8C). It also shortens APD from 206 ms with GFP to 182 ms with LITAF ( Figure 8D). The decrease of APD appears relatively small in view of the large decrease of I Ca,L current during the action potential plateau phase ( Figure 8E). Examination of other currents reveals that a shift of NCX current towards reverse mode during the AP plateau partly counterbalances the effect of decreased I Ca,L current on APD. This shift is associated with a decrease of steady-state intracellular Na + concentration [Na + ] i (10.8 mmol/L with GFP versus 9.4 mmol/L with LITAF). Reduced [Na + ] i with LITAF then promotes forward mode NCX current ( Figure 8F), thereby prolonging APD and partly counterbalancing the effect of I Ca,L reduction on APD shortening. We also note that, on the contrary, the reduced Ca 2+ transient amplitude with LITAF promotes reverse mode NCX, but this effect is not as significant as the shift towards forward mode caused by [Na + ] i reduction. LITAF overexpression in 3wRbCM showed a similar (insignificant) APD shortening (mean ventricular APD90±SEM: LITAF 189±19 ms [n=21] versus GFP 224±15 ms [n=25], P=0.29). Power analysis determined that we would need 122 cells from each group to achieve statistically significant results. Of note, morpholino-mediated downregulation of LITAF in zebrafish resulted in APD prolongation that did not reach statistical significance (mean ventricular APD80±SEM: morphants 283±10 ms versus wild-type 247±9 ms, P=0.11, n=7 and n=9).

DISCUSSION
LITAF regulates endosomal trafficking [11][12][13] and inflammatory cytokines 14,15 and acts an adapter molecule for members of the NEDD4-like family of E3 ubiquitin ligases. 16,17 Sequence variation in LITAF (rs8049607) is associated with QT interval prolongation 5,6 and reduced LITAF mRNA expression in the left ventricle ( Figure I in the Data Supplement). 20 The present study provides the first empirical evidence that LITAF exerts an inhibitory effect on the abundance and function of cardiac LTCC, in part, by controlling NEDD4-1 ubiquitin ligases. Knockdown of LITAF in zebrafish larvae resulted in a robust increase in cardiomyocyte calcium transients on Fura-2 imaging. Overexpressed LITAF in 3-weekold rabbit cardiomyocytes resulted in a decrease in I Ca,L and Cavα1c protein levels, whereas a LITAF knockdown increased I Ca,L and Cavα1c protein levels. We observed a decrease in calcium transients in LITAF-overexpressing adult rabbit cardiomyocytes, which was accompanied by lower Cavα1c abundance. In tsA201 cells, overexpressed LITAF downregulated the total and surface pools of Cavα1c via increased Cavα1c ubiquitination and its subsequent lysosomal degradation. Coimmunoprecipitation showed that LITAF formed a complex with LTCC. Furthermore, in situ proximity ligation assay indicated colocalization between LITAF and LTCC in cardiomyocytes. In tsA201 cells, NEDD4-1, but not its catalytically inactive form NEDD4-1-C867A, increased Cavα1c ubiquitination compared with control. Cavα1c ubiquitination was further increased by coexpressed LITAF and NEDD4-1 but not NEDD4-1-C867A. Knockdown of NEDD4-1 using shRNA abolished the negative effect of LITAF on I Ca,L and Cavα1c protein levels in 3-week-old rabbit cardiomyocytes. Computer simulations demonstrated that a decrease of I Ca,L current associated with LITAF overexpression simultaneously shortened APD and decreased Ca 2+ transient amplitude in rabbit ventricular myocytes.

LITAF: Tissue and Substrate Specificity
Currently, we cannot rule out that LITAF/NEDD4-1-mediated downregulation of LTCC exists in other tissues, such as smooth muscle, somatodendritic neurons or endocrine cells, which also express Cavα1c at significant levels, 35 as LITAF and NEDD4-1 are ubiquitously expressed proteins. 36 Thus it is conceivable the reported single nucleotide polymorphism may also result in changes of I Ca,L in noncardiac tissue. Similarly, LITAF may also negatively affect forward trafficking of other α1subunit isoforms that require a β auxiliary subunit for surface expression, for example, Cav1.3 and Cav2.3, 37,38 which are also expressed in the heart. In contrast, T-type calcium channels, which do not require a beta subunit for their function, 39 are likely not controlled by LITAF. Further studies are warranted to explore these possibilities.

NEDD4-1-Mediated Effects of LITAF on Cavα1c Protein Levels
The NEDD4 family of HECT ubiquitin ligases contains 9 members, 17 which are all expressed in the heart. Most studies have focused on NEDD4-1 and NEDD4-2, and a plethora of potential targets have been identified in vitro. 17 Not surprisingly, various ion channels have been reported to be ubiquitinated by these ubiquitin ligases. For example, voltage-gated potassium (HERG 40 ) or sodium channels (Nav1.5 41 ) are ubiquitinated by NEDD4-2, resulting in their lysosomal degradation. Interestingly, a recent report by Rougier et al 33 suggests that NEDD4-1 promotes downregulation of newly synthesized Cavα1c at the endoplasmic reticulum/Golgi level in tsA201 cells. Prompted by these findings, we looked into the possibility that LITAF-dependent downregulation requires NEDD4-1.
Here, we present data that clearly show that upon knockdown of endogenous NEDD4-1, the effect of LITAF on I Ca,L is completely abolished in cardiomyocytes ( Figure 7B), thus confirming a hypothesized requirement of a NEDD4-1-dependent polyubiquitination of Cavα1c. Prior authors did not observe any NEDD4-1-dependent increase in Cavα1c ubiquitination, in contrast to our findings of increased ubiquitination level of Cavα1c by NEDD4-1 or LITAF ( Figure 6A, 6C, and 6D). These differences likely reflect the discrete experimental conditions used. For example, to amplify the ubiquitination signal, we cotransfected tsA201 cells with an expression plasmid for HA-tagged ubiquitin. Additionally, we added 10 mmol/L N-ethylmaleimide and iodoacetamide to the lysis buffer to block reactive cysteines and thus prevent deubiquitination of proteins by deubiquitinases during sample processing. 42 Notably, a significant LITAF-dependent increase in the overall ubiquitination level of total protein isolated from neonatal rabbit cardiomyocytes was observed (data not shown). In agreement with the study by Rougier et al 33 who reported that Cavβ2 was essential for NEDD4-1-mediated regulation of Cavα1c in tsA201 cells, we observed an absolute requirement for the accessory subunit Cavβ3 (or Cavβ2; data not shown) in LITAF-mediated downregulation of Cavα1c in the same cell line (data not shown). This is corroborated by our coimmunoprecipitation findings in tsA201 cells, demonstrating that LITAF was found in a protein complex with Cavα1c as well as Cavβ3 (Figure 5A and 5B) or Cavβ2 (data not shown) and by our in situ proximity ligation assay, which demonstrates colocalization between LITAF and LTCC in cardiomyocytes ( Figure 5C).
As shown in Figure 6B, the lysosomal inhibitor chloroquine but not the proteasomal inhibitor MG132 blocked LITAF-dependent Cavα1c downregulation. Experiments in tSA201 cells ( Figure 6A) also show Cavα1c polyubiquitination, indicated by the high molecular smear of ubiquitinated Cavα1c, in the presence of LITAF. Polyubiquitination linked through Lys48 or Lys11 generally leads to proteasomal degradation. In contrast, Lys63 linkages perform nondegradative roles (eg, cell signaling) but are also required for lysosomal degradation. 43 Thus, we hypothesize that LITAF causes NEDD4-mediated Lys63 polyubiquitination of Cavα1c.
Contrary to the study by Rougier et al, 33 we did not notice a LITAF-dependent downregulation of the accesso-ry subunits Cavβ3 (Cavβ2) and Cavα2δ-1. This scenario is reminiscent of the selective degradation of the B′β subunit of PP2A (protein phosphatase 2A) mediated via KLHL15 (Kelch like family member 15), a ubiquitin ligase adapter. 44 Although KLHL15 interacted with both the core AC dimer and the B′β monomer of PP2A, only the B′β subunit was ubiquitinated and degraded. It is conceivable that structural requirements mediated through LITAF could favor an exclusive ubiquitination of Cavα1c by NEDD4. Furthermore, it has been shown that both β and α2δ-1subunits can reach the membrane in the absence of α subunits. [45][46][47][48] This could explain that no changes in respective subunits are seen on the membrane despite lower Cavα1c levels ( Figure 4B).
LTCC levels on the membrane are determined by forward trafficking from the endoplasmic reticulum/Golgi apparatus as well as by channel internalization and recycling of endocytosed channel. One could entertain the possibility that LITAF provides NEDD4-mediated LTCC quantity control on the Golgi. Cardiomyocytes must maintain a sufficient number of LTCC channels on the membrane required for proper cardiac excitation. In contrast, limiting LTCC channels on the cell surface is required to prevent excessive calcium entry into the cell, a hallmark of cardiac dysfunction. We anticipate that regulatory pathways exist that allow tight control of LTCC forward trafficking by regulating stability or expression of LITAF. Such pathways could be adversely affected by certain cardiac diseases, subsequently impacting I Ca,L .

Regulation of NEDD4-1 Ubiquitin Ligase by LITAF
LITAF is a small zinc-binding monotopic membrane protein 49 found on the Golgi apparatus and multivesicular bodies. 12 It contains 2 N-terminal PXY motifs, which are required for physical interaction with WW domains of the HECT ubiquitin ligases NEDD4-1 12 and ITCH (itchy E3 ubiquitin protein ligase). 27 We demonstrated that LITAF interacts and colocalizes with LTCC subunits (Figure 5A through 5C) and forms a complex with LTCC. Previous studies have already observed colocalization of LITAF with NEDD4-1 on the Golgi apparatus 12 and control of Cavα1c levels by NEDD4-1 at the endoplasmic reticulum/Golgi level. 33 Therefore, it is very likely that the LITAF-mediated ubiquitination of Cavα1c is happening in the trans-Golgi-network ( Figure VI in the Data Supplement). It is also conceivable that the physical adaptor LITAF promotes NEDD4-1 ubiquitination activity and recruits NEDD4-1 to its Cavα1c substrate on trans-Golgi-network membranes ( Figure VI in the Data Supplement). Such a scenario is reminiscent of the small NEDD4 family-interacting proteins, NDFIP1 (Nedd4 family interacting protein 1) and NDFIP2. 17 They contain cytoplasmic PXY motifs and are localized to the Golgi apparatus, endosomes, and multivesicular bodies. Previously, NDFIP1 was shown to recruit various NEDD4 ubiquitin ligases to membranes, promote autoubiquitination of these ubiquitin ligases by relieving their inherent autoinhibition, and induce substrate ubiquitination. 50 Similarly, a recent study by Kang et al 51 revealed that NDFIP1 recruits NEDD4-2 to the Golgi apparatus to mediate polyubiquitination-dependent degradation of HERG. It is interesting to note that both LITAF and NDFIP1 have been reported to promote exosome secretion, and both molecules could be detected in exosomes produced by cells overexpressing LITAF or NDFIP1, respectively. 26,52 Surprisingly, a LITAF double mutant harboring mutations in both PXY motifs (LIT-AF-Y23A-Y61A) behaved like wild-type LITAF regarding downregulation of I Ca,L in cardiomyocytes (data not shown). We, therefore, think that a direct physical interaction between LITAF and NEDD4 is not an absolute requirement for the observed LITAF-dependent downregulation of LTCC by NEDD4. One could envisage that both proteins are found in a larger multiprotein complex. Alternatively, despite the 2 PXY mutations, residual physical interaction between NEDD4 and LITAF could be sufficient for LTCC ubiquitination.

Transcriptional Effects of LITAF
Although LITAF has recently been shown to target intracellular membranes, 49 which would support its purported role in endosomal trafficking, 11,12 there have been numerous studies implicating LITAF, which is highly expressed in monocytes, macrophages, lymph nodes, and spleen, 53 as an important mediator in systemic and chronic inflammation. 54 Earlier studies identified LITAF as a lipopolysaccharide-induced transcription factor for TNF (tumor necrosis factor)-α. 55 Despite some controversy regarding these findings, 14,21 recent studies using highly specific chromatin immunoprecipitation, confirmed LITAF's role as a transcription factor. For example, lipopolysaccharide-induced LITAF acts as a transcriptional activator for TNF-α, which mediates a proinflammatory and profibrogenic pattern in nonalcoholic fatty liver disease. 25 We, therefore, entertained the possibility of LITAF-dependent transcriptional effects on Cavα1c expression. However, our quantitative polymerase chain reaction data clearly indicated that Cavα1c transcript levels were not changed on LITAF overexpression in neonatal cardiomyocytes ( Figure IVC

LITAF, a Potential Candidate to Alter QT Interval
Based on the effects we have observed of LITAF on calcium transients in zebrafish (Figure 1), which largely depend on Ca 2+ influx through transmembrane Ca 2+ channels, 28 and I Ca,L in rabbit cardiomyocytes (Figure 3), we propose to add genetic variants of LITAF to the growing list of genetic risk factors for sporadic or drug-induced arrhythmias. Identification and expansion of molecular risk factors could help to identify vulnerable patients and aid in the development of new strategies targeting individuals with a prolonged QT interval. The high degree of sequence conservation among known LITAF orthologs 56 implies a conservation of function during vertebrate evolution from bony fish to mammals. Here, we provide strong evidence for a role of LITAF in regulation of calcium transients in zebrafish embryos ( Figure 1) and in the control of LTCC levels on the membrane of rabbit cardiomyocytes (Figures 2 and 3). The association between LIT-AF and QT interval variation has been reported in various genome-wide association studies. 5,6 Indeed, computer simulation of rabbit ventricular myocytes predicted a modest APD shortening in the presence of overexpressed LITAF, primarily because of the compensatory response of NCX. We observed similar magnitude of shortening of the APD in 3wRbCM that did not reach statistical significance in the setting of large variance in the APD of cultured myocytes possibly due to cell-to-cell variability in remodeling of ion channels in culture. Complementing these experiments, LITAF knockdown in zebrafish resulted in prolongation APD that did not reach statistical significance. These observations may reflect a role for LITAF in damping dynamic changes in excitation similar to those reported for other regulators of membrane protein turnover. In this framework, changes in LITAF activity in either direction create vulnerabilities to other modulators of APD, whether genetic or environmental.

Conclusions
Here, we provide both in vivo and in vitro data that imply a role for LITAF in the regulation of LTCC in the heart through modulation of the activity of the HECT ubiquitin ligase NEDD4-1. The end result is ubiquitination and subsequent lysosomal degradation of Cavα1c ( Figure VI in the Data Supplement). This, in turn, results in lower LTCC levels on the cell surface and decreased I Ca,L of cardiomyocytes. We conclude that LITAF controls membrane levels and function of LTCC and is a novel regulator of cardiac excitation.