Increasing evidence demonstrates that perturbation of excitation-contraction coupling in cardiomyocytes leads to cardiac dysfunction and heart failure.¹ Critical to excitation-contraction coupling are cardiac dyads, microdomains at membrane contact sites between transverse-tubules and junctional sarcoplasmic reticulum membranes.^1,2 Junctophilin 2 (JPH2), spanning the cleft from the transverse-tubule to the sarcoplasmic reticulum, plays a central role in formation and maintenance of transverse-tubule/sarcoplasmic reticulum junctions.³ Multiple mutations in JPH2 are associated with human cardiomyopathies. Consequently, there is growing interest in increasing our understanding of proteins that interact with JPH2 to comprise the cardiac dyad proteome.

Methods that rely on affinity pulldown coupled with mass spectrometry to characterize membrane protein-protein interactions (PPIs) are subject to limitations because of the hydrophobic nature of membrane proteins that necessitates harsh extraction conditions, thus potentially disrupting relatively weak or transient PPIs. To overcome these limitations, a biotin ligase-based system, biotin identification (BioID), has been developed to assess PPIs.⁴ BioID has proven to be a useful strategy to screen for PPIs. However, this method has the limitation of artefacts that might arise because of overexpression of the transfected or transduced target-biotin ligase fusion protein. To overcome these limitations, we developed a mouse knock-in strategy using BioID2, a second generation of BioID,⁴ fusing BioID2 to the endogenous JPH2 coding sequence, and used this to investigate the cardiac dyad proteome in living cardiomyocytes.

To knock in the 3xFLAG-BioID2 cassette into the JPH2 locus, we used the CRISPR-Cas9 genome-targeting system (Figure [A]). All mouse protocols were approved by the Institutional Animal Care and Use Committee. The 3xFLAG-BioID2-JPH2 fusion protein was expressed at levels equivalent to those of wild-type JPH2 protein (Figure [B]). Protein levels of two well-known JPH2 binding partners in junctional membrane complexes (RYR2 and CACNA1C) were not altered (Figure [B]). We also found no significant differences in cardiac function between homozygous knock-in (KI) and control mice (Figure [C]). Altogether, these data confirmed that expression of the BioID2-JPH2 fusion protein had no adverse effects on JPH2 expression and function. Immunoblot and immunofluorescence analyses with streptavidin antibody demonstrated that self-biotinylation and proximity labeling saturation in homozygous KI mouse hearts was achieved after 3 days of biotin administration (Figure [D–F]). Confocal imaging of isolated cardiomyocytes from BioID2-JPH2 KI mice demonstrated striking colocalization of JPH2 antibody signal with the streptavidin signal (Figure [G]). To identify biotinylated proteins that were in proximity to JPH2 within the cardiac dyad, we performed a NeutrAvidin pulldown and a FLAG (peptide sequence DYKDDDDK) affinity purification on whole-heart lysates from KI mice and controls. As shown in Figure (H), NeutrAvidin pulldowns were more specific and resulted in retrieving greater amounts of JPH2 proximal proteins than comparable FLAG affinity pulldowns. Characterization of biotinylated proteins using label-free tandem mass spectrometry (MS/MS) enabled identification of 550 (>2 peptides for each protein) JPH2-proximal proteins in homozygous KI mice. The distribution of biotinylated proteins between control and homozygous KI mice was illustrated using a volcano plot (Figure [I]). Self-biotinylated JPH2 and known interacting partners within the cardiac dyad, such as RYR2 and CACN1AC, were highly enriched in the fraction from KI mice. Gene ontology enrichment analysis showed that the top cellular components were sarcolemma, cation channel complexes, and transverse-tubule (Figure [J]).

Figure. BioID2 knock-in mouse strategy identifies cardiac dyad proteome in vivo.A, The JPH2 wild-type allele, the 3xFLAG-BioID2 ssDNA donor designed for insertion and the final knock-in allele. The lengths of ssDNA and homology arms are shown. B, Western blot analysis of RYR2, CACNA1C, and JPH2 in heart tissues from Ctrl and homozygous KI mice at 2 months of age. GAPDH served as a loading control. C, Echocardiographic measurements for Ctrl (n=7) and homozygous KI (n=9) mice at 2 months of age. Graphs indicates no significant changes between KI and Ctrl mice in percentage fractional shortening (%FS), left ventricular internal diameter at end-diastole (LVIDd), and left ventricular internal diameter at end-systole (LVIDs). An unpaired ttest was performed with a confidence of 95% (significant P<0.05) D, Schematic overview of biotin injection. Adult (2-month-old) Ctrl and homozygous KI mice were treated with biotin by intraperitoneal injection once daily at a dose of 24 mg/kg body weight for 7 days. Mouse hearts were collected at indicated days after biotin injection. E, Western blot analysis of biotinylated proteins in mouse heart for indicated days. Biotinylated proteins were detected with HRP-conjugated streptavidin after SDS-PAGE separation. F, Immunofluorescence staining of biotinylated proteins (streptavidin, red) and α-actinin (green) in mouse heart sections by biotin injection for indicated days. G, Adult cardiac myocytes isolated from either Ctrl or homozygous KI mice after 3 days of biotin injection were fixed and costained of biotinylated proteins (streptavidin, red) and JPH2 (green) antibodies. H, Silver staining of eluted proteins of NeutrAvidin pulldown and FLAG pulldown from Ctrl and homozygous KI mice hearts. I, Volcano plot from mass spectrometry data of differentially biotinylated proteins between control (Ctrl) and homozygous KI mice (n=3). We calculated the adjusted P values using the Benjamini-Hochberg algorithm; proteins were considered significantly enriched if they had an adjusted P<0.05 and a fold change >5. The top 20 significant proteins are reported, and several known cardiac dyadic proteins are highlighted in red. J, Gene ontology (GO) enrichment analysis of differentially biotinylated proteins by proteomics. Number of genes belonging to each category is indicated. BP, biological process, CC, cell component. K, Overlapping proteins identified as likely specific interactors of JPH2 in our BioID2 knock-in (BioKI) method and in the Quick et al⁵ HA-immunoprecipitated method. L, Western blot analysis of eluted proteins from NeutrAvidin pulldown and FLAG pulldown with antibodies against RYR2, CACNA1C, FSD2, JPH2, and BAG3.

Quick and colleagues recently identified cardiac dyad binding partners/neighboring proteins of JPH2 using a transgenic mouse line overexpressing a JPH2-HA fusion protein specifically in cardiomyocytes.⁵ Out of 18 proteins suggested to be specific interactors of JPH2 by Quick and colleagues, only 2 (RYR2 and SPEG) overlap with proteins found to be proximal to JPH2 by our approach (Figure [K]). It is notable that important members of the cardiac dyad known to interact with JPH2, such as the L-type calcium channel proteins (eg, CACNA1C, CACNB2, CACNB3) and newly discovered potential dyadic proteins (eg, FSD2, LRRC10), were detected using our BioID2 knock-in strategy but were not detected by the JPH2-HA transgenic approach, attesting to the greater sensitivity and specificity of our BioID2 knock-in strategy. We next performed pulldown analyses with NeutrAvidin or FLAG antibodies, followed by Western blot analyses for known cardiac dyad proteins (RYR2, CACNA1C, and FSD2) and nondyad proteins Bag3 (negative control) (Figure [L]). Results again confirmed that levels of biotinylated proteins from NeutrAvidin pulldown were much higher than levels of the same proteins precipitated by FLAG pulldown.

In summary, our study represents proof of concept for the use of BioID2 knock-in in the study of PPIs in vivo, where the target protein fused to the biotin ligase is present at levels comparable to those of the endogenous protein, obviating potential artefacts arising from overexpressed target-fusion proteins. Our data indicated that BioID2 knock-in was more sensitive and specific than other methods and provided a robust method to identify membrane PPIs in living mice. In addition, our work identified some potential novel cardiac dyad proteins that may play essential roles in cardiac dyad formation, maintenance, and function.

Drs Chen and Evans are funded by grants from the US National Institutes of Health, the National Heart, Lung, and Blood Institute; and Dr Chen holds an American Heart Association Endowed Chair in Cardiovascular Research.

None.

*Drs Feng and Liu contributed equally.

https://www.ahajournals.org/journal/circ

The data, analytical methods, and study materials that support the findings of this study will be available to other researchers from the corresponding authors on reasonable request. The mass spectrometry raw data have been deposited in Mass Spectrometry Interactive Virtual Environment: MSV000084352.

越来越多的证据表明，心肌细胞中的激发-收缩偶联的扰动会导致心脏功能障碍和心力衰竭。¹激发-收缩耦合的关键是心脏双性体，即在横向小管和交界性肌质网膜之间的膜接触部位的微区。^1,2 Junctophilin 2（JPH2），从横向小管到肌浆网跨开裂，在横向小管/肌浆网连接处的形成和维持中起着核心作用。JPH2中的^3个多重突变与人类心肌病有关。因此，人们对提高我们对与JPH2相互作用以构成心脏二元组蛋白质组的蛋白质的了解日益增长。

依赖于亲和力下拉结合质谱表征膜蛋白相互作用的方法受到限制，因为膜蛋白的疏水性要求苛刻的提取条件，因此有可能破坏相对较弱或短暂的PPI。为了克服这些限制，已经开发了一种基于生物素连接酶的系统，即生物素识别（BioID），以评估PPI。⁴ BioID已被证明是筛选PPI的有用策略。然而，该方法具有由于转染或转导的靶标-生物素连接酶融合蛋白的过表达而可能出现的假象的局限性。为了克服这些限制，我们使用BioID2（第二代BioID ^4）开发了一种小鼠敲入策略将BioID2与内源性JPH2编码序列融合，并用于研究活心肌细胞中的心脏二联体蛋白质组。

为了将3xFLAG-BioID2盒敲入JPH2基因座，我们使用了CRISPR-Cas9基因组靶向系统（图[A]）。所有小鼠实验方案均得到了机构动物护理和使用委员会的批准。3xFLAG-BioID2-JPH2融合蛋白的表达水平与野生型JPH2蛋白相同（图[B]）。结膜复合物（RYR2和CACNA1C）中两个著名的JPH2结合伴侣的蛋白质水平未改变（图[B]）。我们还发现纯合敲入（KI）和对照小鼠之间的心脏功能无显着差异（图[C]）。总而言之，这些数据证实了BioID2-JPH2融合蛋白的表达对JPH2的表达和功能没有不利影响。用链霉亲和素抗体进行的免疫印迹和免疫荧光分析表明，使用生物素3天后，纯合KI小鼠心脏实现了自我生物素化和邻近标记的饱和（图[D–F]）。来自BioID2-JPH2 KI小鼠的离体心肌细胞的共聚焦成像显示，JPH2抗体信号与链霉亲和素信号显着共定位（图[G]）。为了鉴定心脏二元组中接近JPH2的生物素化蛋白，我们对来自KI小鼠和对照的全心脏裂解物进行了NeutrAvidin下拉和FLAG（肽序列DYKDDDDK）亲和纯化。如图（H）所示，与类似的FLAG亲和力下拉菜单相比，NeutrAvidin下拉信号更具特异性，并导致检索到更多的JPH2近端蛋白质。使用无标记串联质谱（MS / MS）对生物素化蛋白进行表征，可以鉴定纯合KI小鼠中的550个（每个蛋白> 2个肽）JPH2近端蛋白。使用火山图说明了生物素化蛋白在对照和纯合KI小鼠之间的分布（图[I]）。自体生物素化的JPH2和心脏二联体中的已知相互作用伙伴（如RYR2和CACN1AC）在KI小鼠的组分中高度富集。基因本体富集分析表明，细胞的主要成分是肌膜，阳离子通道复合物和横管（图[J]）。使用火山图说明了生物素化蛋白在对照和纯合KI小鼠之间的分布（图[I]）。自体生物素化的JPH2和心脏二分体内的已知相互作用伴侣（如RYR2和CACN1AC）在KI小鼠的组分中高度富集。基因本体富集分析表明，细胞的主要成分是肌膜，阳离子通道复合物和横管（图[J]）。使用火山图说明了生物素化蛋白在对照和纯合KI小鼠之间的分布（图[I]）。自体生物素化的JPH2和心脏二分体内的已知相互作用伴侣（如RYR2和CACN1AC）在KI小鼠的组分中高度富集。基因本体富集分析表明，细胞的主要成分是肌膜，阳离子通道复合物和横管（图[J]）。

数字。 BioID2敲入小鼠策略可在体内鉴定心脏二联体蛋白质组。A，JPH2野生型等位基因，3xFLAG-BioID2 ssDNA供体，设计用于插入和最终敲入等位基因。显示了ssDNA和同源臂的长度。B，2个月大时Ctrl和纯合KI小鼠心脏组织中RYR2，CACNA1C和JPH2的蛋白质印迹分析。GAPDH用作加载控件。C，在两个月大时对Ctrl（n = 7）和纯合KI（n = 9）小鼠的超声心动图测量。图形显示KI和Ctrl小鼠之间的百分比缩短百分比（％FS），舒张末期的左心室内径（LVIDd）和收缩末期的左心室内径（LVIDs）没有明显变化。不成对的t以95％的置信度（显着性P <0.05）D进行了生物测试，生物素注射液的示意图。成年（2个月大）Ctrl和纯合KI小鼠每天一次腹膜内注射生物素，剂量为24 mg / kg体重，治疗7天。在注射生物素后的指定天数收集小鼠心脏。E，所示天数小鼠心脏中生物素化蛋白的蛋白质印迹分析。在SDS-PAGE分离后，用结合HRP的链霉亲和素检测生物素化的蛋白质。F，通过指定天数在小鼠心脏切片中对生物素化蛋白（链霉亲和素，红色）和α-肌动蛋白（绿色）进行的免疫荧光染色。G在生物素注射3天后，从Ctrl或纯合KI小鼠中分离出的成年心肌细胞被固定，并染色了生物素化蛋白（链霉亲和素，红色）和JPH2（绿色）抗体。H，来自Ctrl和纯合KI小鼠心脏的NeutrAvidin下拉蛋白和FLAG下拉蛋白的洗脱蛋白的银染。余，从控制（CTRL）和纯合KI小鼠（n = 3）之间差异的生物素化的蛋白质的质谱数据火山图。我们使用Benjamini-Hochberg算法计算了调整后的P值；如果蛋白质的P值经过调整，则认为蛋白质显着富集<0.05，倍数变化> 5。报告了排名前20位的重要蛋白质，几种已知的心脏二元蛋白质以红色突出显示。J，蛋白质组学对差异生物素化蛋白质的基因本体论（GO）富集分析。显示了属于每个类别的基因的数目。BP，生物过程，CC，细胞成分。K重叠蛋白在我们的BioID2敲入（BioKI）方法和Quick等⁵ HA免疫沉淀方法中被确定为JPH2的可能特异性相互作用物。L，用抗RYR2，CACNA1C，FSD2，JPH2和BAG3的抗体对中性抗生物素蛋白下拉蛋白和FLAG下拉蛋白的洗脱蛋白进行蛋白质印迹分析。

Quick及其同事最近使用特异于心肌细胞中过表达JPH2-HA融合蛋白的转基因小鼠品系鉴定了JPH2的心脏二联体结合伴侣/相邻蛋白。⁵Quick和他的同事认为18种蛋白质是JPH2的特异性相互作用因子，只有2种（RYR2和SPEG）与我们的方法发现与JPH2接近的蛋白质重叠（图[K]）。值得注意的是，检测到已知与JPH2相互作用的心脏二元组的重要成员，例如L型钙通道蛋白（例如，CACNA1C，CACNB2，CACNB3）和新发现的潜在二进位蛋白（例如，FSD2，LRRC10）。使用我们的BioID2敲入策略，但未通过JPH2-HA转基因方法检测到，证明了我们的BioID2敲入策略具有更高的敏感性和特异性。接下来，我们使用NeutrAvidin或FLAG抗体进行下拉分析，然后对已知的心脏二联蛋白（RYR2，CACNA1C和FSD2）和非二联蛋白Bag3（阴性对照）进行蛋白质印迹分析（图[L]）。

总而言之，我们的研究代表了在体内PPI研究中使用BioID2敲入的概念证明，其中与生物素连接酶融合的靶蛋白的存在水平与内源蛋白的水平相当，从而避免了潜在的人工产物来自过度表达的目标融合蛋白。我们的数据表明，BioID2敲入比其他方法更为灵敏和特异，并提供了一种可靠的方法来鉴定活小鼠中的膜PPI。此外，我们的工作还确定了一些潜在的新型心脏二联体蛋白，这些蛋白可能在心脏二联体的形成，维持和功能中起重要作用。

Chen博士和Evans博士由美国国立卫生研究院，美国国立心脏，肺和血液研究所资助。Chen博士担任美国心脏协会心血管研究基金会主席。

没有。

*冯博士和刘博士的贡献相等。

https://www.ahajournals.org/journal/circ

支持该研究结果的数据，分析方法和研究材料将根据合理的要求提供给相应作者的其他研究人员。质谱仪原始数据已存储在“质谱仪交互式虚拟环境”中：MSV000084352。