Although cardiac TT remodeling and associated dysfunction of Ca²⁺ release have been extensively reported in HF (Balijepalli et al., 2003; Louch et al., 2004, 2012; Song et al., 2006), the processes underlying the formation and maintenance of TTs in general are not known, although we do know that BIN1 seems to be a critical component of TT assembly. In skeletal muscle, this protein has several unique features that make it well-suited for this role, including a natural curvature, a charge profile that allows anchoring in the negatively charged phospholipids in the membrane bilayer, and a PIP2 binding site in Exon 10 that is unique to the muscle (and TT bearing) isoform of the amphiphysin-II protein family. With the exception of Exon 10, nearly all of these properties appear to be shared by the cardiac isoforms of this protein leaving us with an incomplete understanding of these processes in heart.

In cardiomyocytes, Hong et al., 2014 have shown that one of four cBIN1 isoforms in mouse adult cardiomyocyte (BIN1+13+17) was involved in cardiomyopathy and arrhythmia by regulating ion flux and TT morphology (Hong et al., 2014). It has been shown that exon 10 of BIN1 contains a basic amino-acid sequence (RKKSKLFSRLRRKKN) that is important to TT biogenesis in skeletal muscle and the interaction of Exon10 with PIP2 appears to be crucial for localization of BIN1 to TTs in skeletal muscle (Lee et al., 2002). However, our data show that mouse cBIN1 without Exon10/11 still binds a number of phosphoinositides including PIP2. Moreover, overexpressed cBIN1 was capable of tubule formation. These data indicate that Exon 10/11 is not essential for the interaction between cBIN1 and PIP2 for cardiac TT formation. Furthermore, we found that at least one critical PIP2 binding site on cBIN1 is not located at either exon13 or exon17 but does appear to adjoin the BAR domain where its mutation prevented tubule formation. Thus, the current study suggests a novel mechanism that allows cBIN1 and PIP2 to play a crucial role in TT biogenesis. It is also important to note that recent work has also identified the presence of exon 11-containing isoform 8 in rat (Li et al., 2020) but not apparently in mouse (Hong et al., 2014) myocardium. Furthermore, Guo et al. (2021) have shown that at least two different exon 11-containing isoforms 8 and 13 (i.e., Bin1+11 and Bin1+11+17, respectively) are also present in human heart. These results indicate that there are indeed exon 11-containing isoforms in hearts other than mouse that are likely to interact with PIP2 through this motif as well as possible additional binding sites described here that do not rely on the presence of exon 11.

One important finding is the co-localization of PLCβ1, BIN1, and PIP2, at TTs in adult ventricular myocytes. This is likely to be the PLCβ1b isoform since the other minor cardiac isoform (PLCβ1a) is located in the cytoplasm unlike membrane-bound PLCβ1b (Grubb et al., 2008, 2011). TT disruption in HF abolished both the normal striated patterns of Cav1.2 and of BIN1 organization without an overall loss of BIN1 (Supplementary Figure 4). Thus, BIN1 organization is lost in HF without reduced protein levels when PIP2 is no longer available to insure BIN1 membrane insertion and its absence also reduces BIN1 interactions with neighboring proteins. Finally, this co-localization of PLCβ1, BIN1, and PIP2 at TTs also suggests the existence of a signalasome responsible for TT formation and maintenance, supporting the idea that TT remodeling may result from a loss of one or more members of a putative signaling complex.

Hypertension and other cardiovascular diseases that often develop into HF are known to increase levels of endothelin-1 (Miyauchi et al., 1989; Yasuda et al., 1990; Tsuji et al., 1991; Bohm and Pernow, 2007) and angiotensin-II (Gray et al., 1998; Strawn et al., 2000; Molkentin and Dorn II, 2001; Ayabe et al., 2006; Palaniyandi et al., 2009). These factors serve as agonists for a G protein-coupled signaling cascade through Gαq activation which in turn activates several isoforms of phospholipase C, including PLCβ1, found in the sarcolemma. In addition, mRNA and protein levels of PLCβ1 are increased in ischemia (Asemu et al., 2003) and in stroke-prone SHRs as young as 8–10 months there is increased sarcolemmal protein kinase C (PKC) activity (Kawaguchi et al., 1993), suggesting increased PLCβ1 activity. Here, we propose that it is the resulting decrease in membrane PIP2 content itself that has important consequences for TT maintenance and efficient excitation-contraction (EC) coupling.

In order to investigate the role of PIP2, we first focused on the properties of BIN1. Recently, it has been reported that BIN1 serves as an anchor for Ca_v1.2 localization to cardiac TTs, thus promoting specific microtubule-induced trafficking of new Ca_v1.2 molecules to the T-tubular membranes (Hong et al., 2010). This is an important observation because it demonstrates that BIN1 may insure the localization of proteins crucial for EC coupling to the TT and thus serve as an important candidate around which to build the structural and functional TT unit. With this in mind, we measured BIN1 protein expression in SHRs in comparison to normal WKY rats from ages 2–20 months and found no change in BIN1 levels with HF progression but rather extensive disorganization of this protein. We have previously reported that SHRs show significant abnormalities in systolic and diastolic function by 12–17 months as a sign of advancing HF (Shah et al., 2014). In the same study we showed that SHRs show TT disorganization in some myocytes as early as 9-months and that by the age of 17-months a significant reduction in TT organization is apparent in nearly all myocytes. Our current results show that PIP2 levels were significantly decreased during HF development in SHRs especially after the age of 12 months which is consistent with the onset of TT loss during HF development.

In this study, we wanted to explore the properties of BIN1 in a cell system that did not have inherent expression of this protein or TTs. This approach allowed us to examine the properties of BIN1 in tubule formation and how it would respond to PIP2 depletion. We used mouse HL1 cells which do not express BIN1 and found that the exogenous overexpression of all isoforms of BIN1 induced tubulation. We then used this model system to demonstrate that PIP2 depletion could be accomplished in any number of ways including activation of PLCβ or by inhibition of PIP2 synthesis via PI4-kinase (PI4K), both of which subsequently were found to abolish TTs in adult ventricular myocytes. The translocation of the membrane PIP2-binding reporter PLCδ1-PH-GFP from sarcolemma to cytosol following these pharmacological treatments suggested that tubule loss was a result of sarcolemma PIP2 depletion. These results suggest that not only does BIN1 have the potential to promote tubulation but it also requires the presence of membrane PIP2 to do so.

We postulated that the N-BAR domain alone, without the Exon 10/11 sequence such as in cBIN1 isoforms, can still bind PIP2. In fact, the concept of structurally related F-BAR domains capable of binding phosphoinositol was studied previously by Frost et al. (2008) Mutagenesis analysis of the F-BAR domain of FBP17 protein showed that combined Lys56Glu and Arg104Asp substitutions resulted in an absence of tubulation. Considering that Lys56 and Arg104 are residuals of the α-helices that may be directly involved in the integrity of the helical structures or in maintaining coil-to-coil interactions, the argument of these cationic residues being the anchoring sites for phospholipid interactions remains inconclusive. The N-BAR domain of BIN1 also forms oligomeric lattices with narrower tubules than those induced by F-BAR proteins (Frost et al., 2009). We noticed a triple-lysine motif (Lys164-Lys165-Lys166) in a short loop segment between two helices and subsequently demonstrated that mutations of these cationic residues abolished tubule formation. This cationic motif is located at the distal tips of an N-BAR dimer and the mutant is not expected to affect the integrity of the BAR domain structure (Figures 8A,B). When we modeled the BIN1 lattice in self-assembly during tubule induction and highlighted the pattern of tip-to-tip locations in Figure 8C, we found an alternative scenario of unified T-tubule formation through a charge-based BIN1-to-phosphoinositol interaction. These theoretical results support the idea of a novel mechanism by which PIP2 may induce BIN1 oligomerization with resulting TT formation and maintenance.

In order to extend our observations in the HL-1 cell culture system to native cardiac myocyte structure and function, we repeated the PIP2 depletion experiments in native ventricular myocytes and found a similar dramatic reduction of TT organization following pharmacological PIP2 depletion. PIP2 depletion induced by either PI4K inhibition or PLCβ1 activation had virtually identical effects to disrupt TT organization to that reported for HF myocytes. As postulated earlier, PLCβ1 activation, as a member of putative BIN1, PIP2, and PLCβ1 complex located in the TTs, may lower endogenous PIP2 membrane levels in a manner similar to what we observed in culture. Thus, PIP2 depletion via PLCβ1 hydrolysis may eventually disrupt BIN1 targeting to plasma membranes where PIP2 ordinarily helps to localize this protein causing the cBIN1/PIP2 complex to lose its ability to maintain TTs and inducing TT remodeling in HF.

HF myocytes show a reduction of both the rate and magnitude of SR Ca²⁺ release that contribute to poor contractile performance at the subcellular and cellular levels and ultimately to systolic dysfunction in the heart. Slow rise times and times to peak in myocytes after PIP2 depletion is consistent with the finding that PIP2 loss disrupts TTs thereby reducing Ca²⁺ influx through L-type Ca²⁺ channels and impairing CICR from SR. This point is important since cellular regions lacking TTs then have large distances across which remote TT Ca²⁺ influx must diffuse to reach the SR membrane (and so-called orphaned RyRs) in order to induce CICR compared to areas with normal TTs. Thus, instead of a simultaneous activation of contraction throughout the cell, individual cell regions without local TTs will do so with a delay resulting in a dyssynchronous contraction, reducing contractile efficiency. Another feature of altered Ca²⁺ cycling is slow reuptake (Beuckelmann et al., 1992; Balke and Shorofsky, 1998; O’Rourke et al., 1999; Gomez et al., 2001; Wasserstrom et al., 2009) and we observed a slower decay time of Ca²⁺ transients in these cells, affecting the synchrony of relaxation. The primary reason for this slowing is that the time to peak release is slowed and, furthermore, there is increased heterogeneity of release along the cell length. This variability in release presumably occurs where orphaned RyRs must wait for CICR from nearby release units to increase local Ca²⁺ concentration and activate release at those sites which no longer have nearby L-type Ca²⁺ current to trigger release. Because of the delay in release, there is a slowing in rise time and time-to-peak as seen both in our experiments with pharmacologically-induced PIP2 depletion and TT loss (Figure 6) as well as in HF (Louch et al., 2006, 2012; Song et al., 2006; Shah et al., 2014; Singh et al., 2017). The result is also a slowing in recovery of the Ca²⁺ transient as different regions of the cell go through delayed release and resulting removal of cytoplasmic Ca²⁺.

Sarcolemmal PIP2 may also help localize other sarcolemmal proteins including K⁺ channels and the Na-Ca exchanger, thereby affecting cytosolic Ca²⁺ homeostasis through other mechanisms (Hilgemann and Ball, 1996; Huang et al., 1998). The progressive amplification at the multicellular level of this intracellular phenomenon probably contributes to the impairment of overall myocardial function observed in HF.

It is possible that the immunohistochemistry results may be affected by some experimental artifact that could bias our interpretation of apparent co-localization between these proteins. However, we took great care to insure that Cav1.2 was used as an indicator of the location of TTs and that any subsequent localization of the other proteins was aligned with this accepted standard for the localization of TTs. The close alignment between Cav1.2 and Bin1 (Figures 1A,B) as indicated in the merged overlap image cannot be explained as resulting due to chance without invoking explanations that are clearly unrealistic. Thus, although there may be some minor experimental artifacts that might reduce the accuracy of co-localization, such as poor alignment in the z-plane due to the point spread function inherent in confocal microscopy, it is highly unlikely that our results and interpretation are inaccurate overall.

Overall, our work provides extensive evidence that (1) PIP2 binds to Bin1, which has been a highly controversial issue in the heart, (2) Bin1 isoforms are capable of inducing tubule formation, and (3) PIP2 depletion causes loss of TTs in the heart in a manner that mimics the TT remodeling that occurs in HF. On the other hand, we have not provided definitive proof that the PIP2 depletion is directly responsible for TT loss. Our current methodologies do not permit us to measure this interaction directly but our results provide critical new evidence that PIP2 depletion occurs in HF and that pharmacological PIP2 depletion is associated with the loss of TTs in cardiac myocytes. The hope is that, in the future, we will be able to observe this relationship with higher precision in order to provide definitive proof of this reliance of TT remodeling on PIP2 regulation during the development of cardiac disease.

Our data suggest that BIN1 localization at the sarcolemma is required for TT formation and that PIP2 functions as a membrane-targeting and stabilization element for the formation of TTs. These observations are crucial to our understanding of TT remodeling in HF since reduced sarcolemmal PIP2 levels would then be expected to reduce BIN1 targeting to the sarcolemma and hence tubulation. This finding suggests a novel mechanism for TT formation and maintenance, both normal and abnormal. Approximately 80% or more of all L-type Ca²⁺ channels are present on TTs, which in turn are critical for the synchronous release of Ca²⁺ in myocytes and normal EC coupling in heart. Our observation that both delayed Ca²⁺ release and reuptake are closely related to TT organization due to PIP2 depletion strongly suggests that the loss of PIP2 may be the mechanism for TT disruption during disease development and the resulting defects in Ca²⁺ cycling and progressive decline in myocardial mechanics. Our study provides important new insights into the mechanistic understanding of TT organization and its implications in Ca²⁺ dynamics in normal and diseased states.

Based on our results, it might be possible to interfere with degenerative TT remodeling by targeting pathways leading to PIP2 loss. A better mechanistic understanding of the roles of BIN1 and PIP2 in TT remodeling may be important to the development of novel therapeutic interventions aimed at the restoration of this putative phospholipid-protein interactive complex both for the prevention of and the reversal of HF once it has developed.