Study of Short Forms of P/Q-Type Voltage-Gated Calcium Channels
Qiao Feng
Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the Graduate School of Arts and Sciences
COLUMBIA UNIVERSITY
2017
© 2017 Qiao Feng All rights reserved ABSTRACT
Study of Short Forms of P/Q-Type Voltage-Gated Calcium Channels
Qiao Feng
P/Q-type voltage-gated calcium channels (CaV2.1) are expressed in both central and peripheral nervous systems, where they play a critical role in neurotransmitter release. Mutations in the pore-forming 1 subunit of CaV2.1 can cause neurological disorders such as episodic ataxia type 2, familial hemiplegic migraine type 1 and
spinocerebellar ataxia type 6. Interestingly, a 190-kDa fragment of CaV2.1 was found in mouse brain tissue and cultured mouse cortical neurons, but not in heterologous
systems expressing full-length CaV2.1. In the brain, the 190-kDa species is the
predominant form of CaV2.1, while in cultured cortical neurons the amount of the
190-kDa species is comparable to that of the full-length channel. The 190-kDa fragment contains part of the II-III loop, repeat III, repeat IV and the C-terminal tail.
A putative complementary fragment of 80-90 kDa was found along with the 190-kDa form. Moreover, preliminary data show that the abundance of the 190-kDa species and the 80-90-kDa species relative to the full-length channel is upregulated by increased intracellular Ca2+ concentration.
Truncation mutations in the P/Q-type calcium channel have been found to cause the neurological disease episodic ataxia type 2. Some of the disease-causing truncations resemble the 190-kDa truncated channel that we found. Three pairs of
truncated versions of CaV2.1 were engineered to resemble putative products of proteolytic cleavage in the three intracellular loops. Electrophysiological properties of these truncated channels were studied. The truncated channel corresponding to the
N-terminal fragment produced by cleavage in the II-III loop has a suppressive effect on full-length P/Q channel currents, resembling the effects of several truncation mutants that cause episodic ataxia type 2. The complementary pair of truncated channels created by a truncation site in the I-II loop forms a functional channel when coexpressed. These results shed light on the functional effects of proteolytic cleavage in the intracellular loops of the P/Q channel.
Table of Contents
List of Figures and Tables...... vi
Acknowledgements...... viii
Dedication...... x
Chapter 1: Introduction...... 1
1.1 Overview of VGCCs...... 2
1.2 Physiological and pharmacological properties (current types/subtypes) of VGCCs...... 4
1.3 Molecular properties of VGCCs...... 5
1.3.1 Molecular Composition and Structure of VGCCs...... 5
1.3.2 Functions of VGCC subunits...... 8
1.4 Localization and function of VGCCs...... ,,,...8
1.5 Regulation of VGCCs...... ,,....10
1.5.1 Protein phosphorylation...... 10
1.5.1.1 Regulation of CaV1 by protein phosphorylation...... 10
1.5.1.2 Regulation of CaV2 by protein phosphorylation...... 11
1.5.2 Ubiquitination...... 15
1.5.2.1 Ubiquitination of ion channels...... 15
1.5.2.2 Ubiquitination of the L-type calcium channel...... 17
1.5.2.3 Ubiquitination of the N-type calcium channel...... 21
1.5.2.4 Summary of ubiquitination of voltage-gated calcium channels...... 24
1.6 Proteolysis of ion channels...... 25
i
1.6.1 Proteolysis of L-type calcium channels...... 25
1.6.1.1 Proteolytically cleaved C-terminal fragment of L-type calcium channel 1
subunit acting as an autoinhibitory domain...... 25
1.6.1.2 Mid-channel proteolysis of the L-type calcium channel 1 subunit...... 36
1.6.2 Proteolysis of voltage-gated sodium channels...... 40
1.7 Splice variants and truncated VGCCs...... 44
1.7.1 A C-terminal fragment of 1C acting as a transcription factor...... 44
1.7.2 Splice variants and truncated forms of the P/Q-type calcium channel...... 51
1.7.2.1 A variety of fragments of 1A...... 51
1.7.2.2 A two-domain form of 1A with inhibitory function...... 56
1.7.2.3 A C-terminal fragment of 1A acting as a transcription factor...... 59
1.7.2.4 1A truncations causing episodic ataxia type 2...... 66
1.7.2.5 Splice variants of 1A...... 80
1.8 Thesis introduction...... 82
Chapter 2: Materials and Methods...... 84
2.1 Molecular biology...... 85
2.2 Cell culture...... 86
2.2.1 Mouse cortical neuron culture...... 86
2.2.2 Rat hippocampal neuron culture...... 87
2.2.3 HEK 293T cell culture...... 88
2.3 SDS-PAGE sample preparation...... 88
ii
2.3.1 Preparation of tissue/cell lysate using Pierce IP Lysis Buffer...... 88
2.3.2 Preparation of tissue/cell lysate using SDS lysis buffer...... 89
2.3.3 Purification of 1A with nickel affinity gel...... 90
2.3.4 Preparation of oocyte membrane fraction...... 91
2.4 SDS-PAGE and Western blot...... 92
2.5 Transfection...... 93
2.6 RNA synthesis...... 95
2.7 Oocyte preparation and cRNA injection...... 95
2.8 Brain slices...... 96
2.9 Surface biotinylation...... 97
2.10 Immunocytochemistry and confocal microscopy...... 99
2.10.1 Fixed-cell imaging...... 99
2.10.2 Live-cell antibody feeding...... 100
2.11 Two-electrode voltage clamp (TEVC)...... 100
2.12 Data analysis...... 101
Chapter 3: Biochemical Studies of the 190-kDa Fragment of 1A...... 102
3.1 Introduction...... 103
3.2 The 190-kDa fragment of 1A in the mouse brain...... 104
3.3 The 190-kDa fragment of 1A in cultured mouse cortical neurons...... 106
3.4 1A expressed in heterologous systems shows normal molecular weight...... 108
3.5 Evidence that the 190-kDa fragment of 1A exists in live brains and live cells...... 110
iii
3.6 The 190-kDa fragment of 1A contains the C-terminal tail but not the N-terminal
tail...... 113
3.7 Is the 190-kDa fragment of 1A on the plasma membrane?...... 116
3.8 Effect of calcium activity on the relative amount of the 190-kDa fragment of 1A...119
3.9 Possible alternative splicing of mouse1A...... 122
3.10 Discussion...... 125
3.10.1 The presence of the 190-kDa species in live brains and cultured neurons...... 125
3.10.2 Identity of the 190-kDa species...... 131
3.10.3 Is the 190-kDa species on the plasma membrane?...... 139
3.10.4 Origin of the 190-kDa species...... 145
Chapter 4: Visualization of 1C and 1A in Cell Cultures...... 151
4.1 Introduction...... 152
4.2 Fixed-cell imaging in cultured neurons under non-permeabilizing conditions...... 153
4.3 Fixed-cell imaging in HEK 293T cells...... 157
4.4 Live-cell antibody feeding in cultured neurons...... 159
4.5 Fixed-cell imaging in cultured neurons under permeabilizing conditions...... 160
4.6 Discussion...... 161
Chapter 5: Functional properties of fragment-channels...... 169
5.1 Introduction...... 170
5.2 Functional properties of fragment-channels produced by cleavage in the II-III loop of
1A...... 170
iv
5.3 Functional properties of fragment-channels produced by cleavage in the I-II loop and
the III-IV loops of 1A...... 173
5.4 Expression of fragment-channels...... 175
5.5 Discussion...... 179
5.5.1 Why is C1 not expressed?...... 179
5.5.2 A1 and A2 form a functional channel...... 182
5.5.3 Why do B1 and B2 not conduct currents when coexpressed?...... 184
5.5.4 Possible mechanisms underlying the dominant negative effect of B1...... 188
5.5.5 Other issues...... 193
Chapter 6: Summary and Future Directions...... 197
6.1 Functional importance of the 190-kDa truncated ...... 198
6.2 Origin of the 190-kDa truncated ...... 201
References...... 204
v
List of Figures and Tables
Figure 1 Schematic of the cardiac CaV1.2 autoinhibitory signaling complex and the
mechanism of cAMP/PKA-mediated upregulation 35
Figure 2 Membrane topology of 1A, showing positions of truncation mutations,
including EA2 and non-EA2 mutations that have been studied, as well as
unstudied EA2 mutations that resemble other mutations that have been
studied 67
Figure 3 Possible mechanisms for the dominant negative effect of truncated CaV2 1
subunits and the rescuing effect of the partial N-terminal tails 79
Figure 4 Seven loci of human 1A alternative splicing identified by transcript scanning 81
Figure 5 Western blot analysis of the mouse brain with the II-III loop antibody reveals
the 190-kDa fragment of 1A in the mouse brain 106
Figure 6 Western blot analysis of 1A in cultured mouse cortical neurons 108
Figure 7 Western blot analysis of 1A expressed in heterologous systems 109
Figure 8 Homogenization in SDS lysis buffer at 95°C does not reduce the amount of
the 190-kDa species 112
Figure 9 The 190-kDa fragment of 1A contains the C-terminal tail but not the
N-terminal tail 114
Figure 10 Western blot analysis of 1A in mouse cortical slices after surface
biotinylation 119
Figure 11 Effects of calcium activity on the proteolysis of -spectrin, PARP and 1A in
vi
cultured mouse cortical neurons 120
Figure 12 CV and MaxEnt scores of potential splice donor sites in exon 1 of CACNA1A 124
Figure 13 Distribution of the CV scores of 126,480 known human splice donor sites 124
Figure 14 Four EGFP/HA-tagged human 1A constructs 155
Figure 15 Fixed-cell imaging in cultured rat hippocampal neurons under
non-permeabilizing conditions 156
Figure 16 Fixed-cell imaging in HEK 293T cells under non-permeabilizing conditions 158
Figure 17 Live-cell antibody feeding in cultured rat hippocampal neurons 160
Figure 18 Fixed-cell imaging in cultured rat hippocampal neurons under permeabilizing
conditions 161
Figure 19 Functional properties of fragment-channels hPQB1 and hPQB2 172
Figure 20 Functional properties of fragment-channels produced by cleavage in the I-II
loop and the III-IV loops of 1A 174
Figure 21 Expression of fragment-channels in the cytoplasm and on the plasma
membrane of Xenopus oocytes 178
Table 1 Properties of EA2 truncation mutants and engineered truncation mutants 68
Table 2 Different conditions tried for fixed-cell imaging with cultured neurons under
non-permeabilizing conditions 156
Table 3 Possible mechanisms underlying the dominant negative effect of B1 and the
expected phenomena 193
vii
Acknowledgements
My experience at Columbia University has been a fascinating one. During the six years as a graduate student, I had the luck to meet many wonderful people, who helped me grow both professionally and personally.
First, I would like to thank my advisor Prof. Jian Yang. Jian’s passion for science and meticulous work attitude greatly inspired and motivated me in my research.
Everytime when I encountered difficulties in my experiments, Jian encouraged me to think outside the box and try different problem-solving strategies. Without his support and mentorship, I would not have been able to complete my thesis project.
I would also like to thank the members of my thesis committee, Profs. Martin
Chalfie, Henry Colecraft, Chloe Bulinski and Yong Yu. They spent a great amount of time attending my committee meetings and giving me valuable advice for my research.
Dr. Yong Yu has also been my mentor, who taught me important lab techniques when he was in the Yang lab for his postdoc. I deeply appreciate the great help and support that all of them offered me.
I also want to thank all the past and present members of the Yang lab who have helped me. Yannis Michailidis was a wonderful mentor, who spent a lot of time training me, answering my questions, and helping me troubleshoot experiments. Prof.
Zafir Buraei, Kathryn Abele Henckels, Minghui Li, Kevin Zhang and Nicole Benvin also offered me great help and support for my research. Ye Gong worked intensively on part of my project and made an important contribution to my thesis. I would also
viii like to thank visiting student Dan Zhang as well as undergraduate students Gilbert
Garbiso, Morgan Goodman, and Hasan Ismail, who enthusiastically participated in my project and added to the warm atmosphere of our lab.
I also owe a debt of gratitude to Profs. Lawrence Chasin and James Manley, who offered me helpful professional opinions. I also appreciate the help of Profs. John
Hunt and Michael Sheetz, in whose labs I did my rotations and learned important research skills.
My friends Nicole Benvin, Yuanjun Gao, and Chi Wang have been incredibly supportive of me during my time at Columbia. I am so grateful for being great listeners, sharing their wisdom with me, and offering personal advice. I also want to thank my friends Michael Lorimer, John Olson, Ming-Jui Liu, Jianyi Ren and
Benedykt Shwarek for bringing so much fun to my life in NYC.
Last, but not least, I would like to thank my parents. I could not have achieved so much without their love, support, and encouragement. I am extremely thankful that they are always there for me, and the crucial role they play in my life.
ix
To my Parents and Motherland
x
Chapter 1
Introduction
1
Voltage-gated calcium channels (VGCCs) play important roles in lots of physiological processes, including muscle contraction, neurotransmitter release, endocrine secretion, and gene expression (1). In the brain and heart, the pore-forming
1 subunits of VGCCs are present not only in the full-length forms, but also in a variety of short forms, which have important functions. C-terminal fragments of the
L-type calcium channel CaV1.1 and CaV1.2 subunits can be produced by proteolysis and act as autoinhibitory domains to regulate L-type channel activity (2). A cryptic
promoter can also produce a C-terminal fragment of CaV1.2, which serves as a transcription factor (3). CaV1.2 is also subject to “mid-channel proteolysis”, which produces two complementary fragments and serves as a negative feedback regulatory
mechanism (4). Short forms of the P/Q-type calcium channel (CaV2.1) have also been
found in the brain. A 95-kDa fragment of CaV2.1, with unknown origin, has an inhibitory effect on the full-length P/Q-type channel (5, 6). A C-terminal fragment of
CaV2.1 can be produced by an internal ribosomal entry site and acts as a transcription
factor (7). Numerous other short forms of CaV2.1 found in immunoblot analyses have not been studied (5, 8-12).
CaV2.1 is crucial for neurotransmitter release in both central and peripheral
nervous systems (13). Mutations in CaV2.1 can cause as least three types of autosomal dominant neurological disorders: episodic ataxia type 2 (EA2), familial hemiplegic migraine type 1 (FHM1) and spinocerebellar ataxia type 6 (SCA6) (14). EA2 can be caused by several types of mutations, including truncation mutations, i.e., mutations
2 that introduce premature stop codons into the gene of CaV2.1 and lead to short forms
of CaV2.1 (15). Interestingly, as stated above, a variety of short forms of CaV2.1 are
also present in healthy brains. Why do these short forms of CaV2.1 not cause any problems like the EA2 truncation mutants? Why and how do cells produce these short
forms of CaV2.1?
To begin to address these questions, I studied CaV2.1 in the mouse brain using biochemistry methods. The experiments showed that one of the major forms of
CaV2.1 in the mouse brain is a 190-kDa N-terminally truncated CaV2.1. Then I
explored the origin and function of 190-kDa truncated CaV2.1 using methods including biochemistry, molecular biology and electrophysiology.
1.1 Overview of VGCCs
Voltage-gated calcium channels (VGCCs) are important mediators of calcium influx. The calcium ion is an important intracellular messenger with many different signaling functions (16). Upon membrane depolarization, VGCCs open and allow
Ca2+ to enter the cell, converting the electrical signal on the plasma membrane to an intracellular Ca2+ transient (1). Different types of VGCCs have different molecular compositions and different physiological, pharmacological, and regulatory properties.
Activities of VGCCs are elaborately regulated by a variety of signals. Moreover, several types of VGCCs have been found to exist in truncated forms, which have different properties from the full-length channels and also influence the functions of
3 the full-length channels.
1.2 Physiological and pharmacological properties (current types/subtypes) of
VGCCs
According to the level of membrane potential at which the channels are activated,
VGCCs are divided into two main groups, high-voltaged-activated (HVA) and low-voltage-activated (LVA) calcium channels (17). HVA channels include L-, N-,
P/Q- and R-type calcium channels, while the T-type calcium channels are designated as LVA channel. L-type calcium channels are distinguished by their high voltage of activation, large single channel conductance, slow voltage-dependent inactivation and sensitivity to dihydropyridines (18-21). In contrast to the L-type calcium channels, the
T-type calcium channels are activated by weak depolarization, are inactivated quickly, and have tiny single channel conductance (22). The N-type calcium channels have varied voltage dependence and kinetics depending on the cell type (17), and are distinguished by the specific inhibition by -conotoxin GVIA (22).
P- and Q-type calcium currents were originally discovered separately in different cell types . The two distinct types of calcium currents were later found to be from the same type of CaV1 subunit (see 1.3), and their differences probably result from alternative splicing and assembly with different auxiliary subunits (23, 24). P-type calcium currents, first recorded in cerebellar Purkinje neurons (25), are distinguished by their slow inactivation and specific inhibition by -agatoxin IVA with high
4 sensitivity (26). Q-type calcium currents, first recorded in cerebellar granule neurons, inactivate more quickly and are inhibited by -agatoxin IVA with a lower sensitivity
(26). Both P- and Q-type calcium currents are potently inhibited by -conotoxin
MVIIC (26). R-type calcium channels are resistant to all the aforementioned inhibitors (26). SNX-482 was found to inhibit R-type calcium channels at a much higher selectivity than on other types of voltage-gated calcium channels (27).
However, SNX-482 was later found to also inhibit A-type potassium channels,
especially KV4.3, which was even more sensitive to SNX-482 than R-type calcium channels are (28).
1.3 Molecular properties of VGCCs
1.3.1 Molecular Composition and Structure of VGCCs
A VGCC is a heteromultimeric protein complex comprised of several subunits.
The 1 subunit is the pore-forming subunit, through which calcium ions flow when the channel is activated (1). Mammalian 1 subunits are encoded by 10 CACNA1x genes (29). The four subtypes of L-type calcium channels, CaV1.1-CaV1.4, have 1 subunits named 1S, 1C, 1D and 1F, respectively, and these 1 subunits are encoded by CACNA1S, -C, -D and -F, respectively. The 1 subunits of P/Q-, N- and
R-type calcium channels (CaV2.1-CaV2.3) are encoded by CACNA1A, -B, and -E, and are named 1A, 1B and 1E, respectively. The 1 subunits of T-type calcium channels (CaV3.1-CaV3.3) are encoded by CACNA1G, -H, and -I, and are named 1G,
5
1H and 1I, respectively.
The 1 subunit consists of 1800-2500 amino acid residues and contains four homologous repeats (repeats I-IV), each comprised of six transmembrane segments
(S1-S6). The S1-S4 segments in each repeat form the voltage-sensing domain (VSD), and the S5-S6 segments from all four repeats constitute the pore domain (30). A
“re-entrant” loop, known as the pore loop or P loop, between S5 and S6 was predicted to form part of the pore lining of the channel (13, 17). However, the structure of the rabbit CaV1.1 complex recently determined by cryo–electron microscopy (cryo-EM) shows that the re-entrant segment between S5 and S6 is organized into two pore helices (P1 and P2) and a selectivity filter in between (30).
The subunit is an intracellular subunit (30, 31) of ~60 kDa. There are four subfamilies of subunits (1-4), encoded by four genes (CACNB1-4) (29, 31), and each subfamily includes splice variants (31). The subunit has five distinct regions, the N-terminus, the SH3 domain, the HOOK region, the GK domain and the
C-terminus (31). The middle SH3-HOOK-GK module is called the core, which is critical for many key functions of the subunit. The GK domain is able to bind with high affinity to the -interaction domain (AID) in the cytoplasmic loop connecting repeats I and II (I-II loop) of the 1 subunit. Cryo-EM structure of the CaV1.1 complex shows that the subunit is located adjacent to the voltage-sensing domain of repeat II (VSDII) of 1 (30).
The 2 subunit has four subfamilies, 2-1 through -4, encoded by
6
CACNA2D1-4, and can form splice variants (29). The N-terminus of 2 has a signal sequence directing the precursor of 2 into the lumen of the ER, so that 2 eventually becomes extracellular after being trafficked to the plasma membrane (29).
The precursor of 2, containing about 1100 amino acid residues, is subjected to posttranslational proteolysis, which cleaves the protein into 2 and , which are linked by a disulfide bond (32). Both 2 and have been shown to be glycosylated
(30, 33). Several domains have been identified in 2, including a Von Willebrand
Factor-A (VWA) domain and two Cache domains (29, 30). Through these domains,
2 interacts with the extracellular loops of repeats I-III of 1 (30).
The subunit is a glycoprotein of about 30 kDa (17, 34). There are eight subfamilies of subunits (1-8), encoded by CACNG1-8 (34). Several splice variants have been found. Hydrophobicity plots predict the subunit to have four transmembrane helices (TMs) and intracellular N- and C-termini (34). This predicted topology of the subunit was confirmed using immunocytochemistry (35) and in the
Cryo-EM structure of the CaV1.1 complex, which also shows that the subunit interacts with VSDIV of via TM2 and TM3 (30). In addition, studies show that subunits are associated with L-, P/Q- and N-type calcium channels (34).
7
1.3.2 Functions of VGCC subunits
As a fundamental component of VGCCs, the 1 subunit contains the voltage sensor and the ion-conducting pore. The main voltage sensor is the S4 transmembrane segment (36, 37). The S4 segment of each repeat contains positively charged arginine and lysine residues, and thus can move outward and rotate upon membrane depolarization, inducing a conformational change that opens the pore. The S5-S6 segments from the four repeats of 1 forms the pore domain, in which there is a selectivity filter constituted by the side chains of four conservative glutamate residues in the four repeats and the carbonyl oxygen atoms of the two preceding residues in each repeat (30). The four conservative glutamate residues are critical for the high calcium selectivity of the pore (38).
Although 1 alone is able to produce currents, coexpression of 2 and especially enhances the expression level of 1, promotes trafficking of 1 to the plasma membrane, and allows for more conventional gating properties (17, 31, 39).
Coexpression of the subunit has much smaller functional effects. Most subunits cause a small reduction of currents by causing a hyperpolarizing shift of the inactivation curve (34, 40). Some subunits have also been shown to regulate the trafficking, localization, and biophysical properties of AMPA receptors (31).
1.4 Localization and function of VGCCs
The L-type channel family includes one skeletal muscle-specific isoform (CaV1.1)
8 and three neuronal isoforms, (CaV1.2, CaV1.3 and CaV1.4). CaV1.1 is responsible for
excitation-contraction coupling in skeletal muscle (1). CaV1.2 is widely distributed in
the brain, heart, smooth muscle, and adrenal medulla (1, 13). CaV1.2 is responsible for excitation-contraction coupling in cardiac and smooth muscle, and contributes to intracellular Ca2+ transients, endocrine secretion, as well as regulation of transcription
(1, 13). CaV1.3 is expressed in the brain, adrenal medulla, as well as cochlear hair cells, and contributes to intracellular Ca2+ transients, endocrine secretion, pacemaking,
and auditory transduction (1, 13). CaV1.4 is expressed in the retina and plays an essential role in visual transduction (1, 13, 17).
The P/Q-, N- and R-type channels ,(CaV2.1, CaV2.2 and CaV2.3), are widely distributed in the central and peripheral nervous systems, and are mainly responsible
2+ for neurotransmitter release as well as dendritic Ca transients (1, 13). CaV2.1 is expressed throughout the entire brain, with particularly high levels in cerebellar
Purkinje cells (41-43). In cerebellar Purkinje cells, CaV2.1 is robustly expressed in dendrites in the molecular layer (42, 43) axon terminals in deep cerebellar and Deiters’ nucleus, as well as in the soma and along the lengths of axons (42). CaV2.1 also contributes to nearly half of the VGCC currents in cerebellar granule cells (26).
CaV2.1 is also expressed at neuromuscular junctions (44). CaV2.1 contributes to two distinct types of calcium currents, P-type and Q-type, depending on the splice variants, secondary modifications, and accessory subunits (23, 45). P-type currents are found throughout the central and peripheral nervous systems, especially in cerebellar
9
Purkinje cells (46). In cerebellar granule cells, ~76% of the CaV2.1 currents are
Q-type, and the remaining are P-type (26). The CaV2.1 currents recorded at the synapse between hippocampal CA3 and CA1 neurons are also Q-type (47).
Furthermore, CaV2.1 expressed in Xenopus oocytes also conducts Q-type currents
(48). CaV2.2 is mostly distributed in dendritic shafts and presynaptic terminals (49).
CaV2.2 can be found in a variety of regions in the brain (49), dorsal root ganglia,
dorsal horn neurons, and neuroendocrine cells (50). CaV2.3 is also widely distributed in the central and peripheral nervous systems, especially in hippocampal CA3 neurons, where it contributes to 25% of the high-voltaged-activated (HVA) calcium channel
currents (46). The T-type channels, (CaV3.1, CaV3.2 and CaV3.3), are expressed throughout the body in a variety of organs, and play a major role in cardiac pacemaking and oscillations (1, 13, 50).
1.5 Regulation of VGCCs
1.5.1 Protein phosphorylation
1.5.1.1 Regulation of CaV1 by protein phosphorylation
Both CaV1.1 and CaV1.2 are modulated by PKA-mediated phosphorylation in response to activation of the -adrenergic receptor pathways (2). A major phosphorylation site in cardiac 1C is S1928 (51), which can be phosphorylated by
PKA (52) and PKC (53). Phosphorylation of S1928 has been shown to be responsible
for PKA-mediated upregulation of CaV1.2 activity (54). Additional key
10 phosphorylation sites in cardiac 1C are S1700 and T1704 (55), which are conserved in 1S as S1575 and T1579, respectively (56). S1575/S1700 is a substrate for PKA and CaMKII, while T1579/T1704 is a substrate for casein kinase II (55, 56). Basal
activity of CaV1.2 in unstimulated cells is regulated by phosphorylation of both S1700 and T1704, whereas further upregulation of channel activity through stimulation of the cAMP/PKA pathway requires only the phosphorylation of S1700 (55). These two important phosphorylation sites will be discussed in detail in 1.6.1.1.
Several isoforms of PKC have been shown to co-immunoprecipitate with CaV1.2
(57). A-kinase anchoring protein (AKAP) 150, which binds to both CaV1.2 and PKC,
helps to recruit PKC to CaV1.2 (57). The N-terminus of CaV1.2 is subject to alternative splicing involving two mutually exclusive exons, e1a (46 amino acids long), and e1b (16 amino acids long) (58, 59). PKC downregulates the activity of
CaV1.2-e1a by phosphorylating T27 and T31 in the N-terminus (57). CaV1.2-e1b, which does not have the two threonine residues, does not show inhibition by PKC
(57). An analogous inhibitory effect has been found on CaV1.3, and its activity is decreased when S81 at the N-terminus is phosphorylated by PKC (57).
In arterial smooth muscle cells, PKC increases local CaV1.2 activity and thereby creates sites of sustained Ca2+ influx, leading to increased blood pressure (60).
1.5.1.2 Regulation of CaV2 by protein phosphorylation
CaV2 can be regulated by PKA, PKC, and CaMKII. The effects of these kinases
11 on CaV2 activity vary depending on the cell type. PKA activation enhances N-type currents in rat nodose neurons (61) but reduces N-type currents in mouse dorsal root ganglion neurons (62) and rat neostriatal neurons (63). PKA activation enhances currents of the P/Q-type channel expressed in Xenopus oocytes (64, 65), but slightly reduced the non-L- and non-N-type currents, which are mostly P/Q-type currents, in rat cerebellar granule cells (26). PKC activation enhances N-type channel activity in most central and peripheral neurons studied (66, 67), but depresses N-type channel currents in acutely isolated guinea pig hippocampal neurons (68). The specific effect of a protein kinase on a calcium channel may depend on the splice variant of the 1 subunit, the types of auxiliary subunits, and other cell-specific proteins.
In Xenopus oocytes, PKA activation has been shown to potentiate the currents of
P/Q-type channels (64, 65) and, to a lesser extent, N-type channels (65). In sea slug hair cells, activation of PKA not only causes potentiation of P/Q-type currents, but also mediates voltage-dependent facilitation of P/Q-type channels (69). In
NGF-differentiated trk-PC12 cells, activation of PKA prevents the inhibition of N- and P/Q-type channels by ethanol (70). PKA activation has also been shown to
alleviate the voltage-dependent inhibitory effect of PtdIns(4,5)P2 on the P/Q-type channel, probably by phosphorylation of the 1 subunit of the channel (71). The
C-terminal tail of 1B contains a binding site for 14-3-3, a family of regulatory proteins (72). GST pull-down and coimmunoprecipitation analyses in tsA-201 cells showed that phosphorylation of S2126 ,(in the RFYS motif within the 14-3-3 binding
12 site) by PKA or CaMKII promotes the binding of 14-3-3, which counteracts the subunits to decelerate inactivation of the CaV2.2 (72). In most of the regulatory
processes described above, it is unknown whether PKA phosphorylates CaV2 itself or other regulatory proteins. Further research is necessary to understand more details of
how PKA regulates CaV2.
Regulation of CaV2 by PKC is better understood than that by PKA. In most cases
studied, PKC activation upregulates CaV2 activity. It has also been reported that PKC activation potentiates Q-type channel-mediated synaptic transmission at CA3-CA1
synapses in rodent hippocampus (47). PKC modulates CaV2 function through phosphorylation of two regions of the 1 subunit, the I-II loop and the II-III loop. In the I-II loop of 1A and 1B, there are four consensus PKC phosphorylation sites,
422T, 425S, 453S, and 456S, corresponding to positions in 1B. The first two consensus sites are located within one of the two G binding sites and result in crosstalk between G proteins and PKC (73). Early studies showed that activation of
PKC reduces G protein-dependent inhibition of Ca2+ channels and upregulates basal calcium channel currents, especially N-type currents (66, 67). Further studies suggest that phosphorylation of T422 mediates the antagonistic effect of PKC on G protein modulation by disrupting the binding of G protein to the channel, whereas phosphorylation of either T422 or S425 is sufficient to increase N-type channel activity (74, 75).
PKC also modulates CaV2 functions by phosphorylating the synaptic protein
13 interaction (synprint) site in the II-III loop. Binding of syntaxin 1A to the synprint site results in two effects on the N-type channel: a hyperpolarizing shift in the steady-state inactivation curve and a tonic inhibition of the channel by G subunits (76).
Phosphorylation of the synprint peptide sequence of 1B with PKC or CaMKII, but not PKA or PKG, strongly inhibits binding of syntaxin 1A or SNAP25 (77). Moreover, phosphorylation of the synprint site of 1B eliminates the syntaxin 1A-mediated shift in the steady-state inactivation curve but does not influence the syntaxin 1A-induced
G protein inhibition of CaV2.2 (75). Further research revealed that binding of syntaxin
1A and SNAP25 is regulated by PKC phosphorylation at S774 and S898 and by
CaMKII phosphorylation at S784, S896, and S898 (78). Interestingly, in tsA-201 cells
where syntaxin 1A and CaV2.2 are coexpressed, PKC activation opposes the effect of syntaxin 1A on the steady-state inactivation curve without affecting the binding of
syntaxin to CaV2.2 (78).
CaMKII can also regulate CaV2 activity without phosphorylating the channel.
CaMKII binds to a peptide sequence in the C-terminal tail of 1C, which is conserved as amino acid residues 1897-1912 in the C-terminal tail of CaV2.1 (79, 80). CaMKII
constitutively binds to CaV2.1, decelerating voltage-dependent inactivation, causing a positive shift in the voltage dependence of inactivation, and enhancing
2+ Ca -dependent facilitation of CaV2.1 channels (80). These effects of CaMKII are
reduced by a peptide that eliminates binding of CaMKII to CaV2.1. Therefore, the binding instead the phosphorylation is responsible for the regulatory effects (80).
14
1.5.2 Ubiquitination
1.5.2.1 Ubiquitination of ion channels
Ubiquitination plays an important role in regulating the surface density of ion channels. The process of ubiquitination can occur at the exit of the endoplasmic reticulum (ER) (81) as well as the Golgi complex (82, 83), and labels newly synthesized ion channels for degradation in the proteasome (81, 84). Ion channels on the plasma membrane can also be ubiquitinated, which would lead to internalization and subsequently degradation in the proteasome and lysosome (81, 85).
Ubiquitination is an enzyme-catalyzed process in which a ubiquitin protein is covalently attached to a substrate protein (86). Ubiquitination usually results in an isopeptide bond between the carboxyl group of the last amino acid residue (G76) of ubiquitin and the -amino group of a lysine residue on the substrate protein (86). This process requires the sequential actions of three enzymes: a ubiquitin-activating enzyme (E1), a ubiquitin-conjugating enzyme (E2) and a ubiquitin-protein ligase (E3)
(86). Ubiquitin itself can also be ubiquitinated, thereby forming polyubiquitin chains
(86). Ubiquitin has several lysine residues, and polyubiquitin chains linked through different lysine residues affects the fate of the substrate protein in different ways (86).
Ubiquitination has been found to regulate the activity and fate of a variety of ion channels (81). Several of these ion channels bear a PY motif (L/P-P-X-Y-X-X-, where L is a leucine, P is a proline, X is any amino acid, Y is a tyrosine, and is a hydrophobic amino acid) (87, 88), which has been identified as a binding site for E3
15 ubiquitin ligases of the Nedd4/Nedd4-like family (89, 90). The Nedd4/Nedd4-like E3 ubiquitin ligases, in turn, ubiquitylate the channels and downregulate their surface expression, typically by inducing internalization of the channels (81).
Nedd4/Nedd4-like E3 ubiquitin ligases have been shown to ubiquitinate and downregulate a wide range of ion channels, including the epithelial sodium channel
(ENaC) , NaV1.2, NaV1.3, NaV1.5, the ClC-5 chloride/proton exchanger (81), and the
KCNQ1 potassium channel (KV7.1) (91). All these channels bear a PY motif in their
C-terminal regions and are downregulated by Nedd4-2, the most studied member of
the Nedd4/Nedd4-like family. NaV1.2, NaV1.3, ClC-5 (81), and KCNQ1 (91) have also been shown to be regulated by other membranes of the Nedd4/Nedd4-like family,
such as Nedd4-1 (also called Nedd4) and WWP2. KV1.3 has also been shown to be
ubiquitinated and downregulated by Nedd4-2, although KV1.3 does not have a PY
motif (92, 93). A proline-rich motif (PQTPF) in the C-terminal tail of KV1.3 has been suggested to serve as the target for Nedd4-2 (93). The L-type calcium channel 1C subunit has a non-canonical PY motif (LPYVAL) at the end of the S4-S5 loop and beginning of S5 in repeat IV (4). Mutation of this motif has been shown to markedly
reduce mid-channel proteolysis of CaV1.2, a recently discovered form of regulation of
VGCCs (4). This indicates that the Nedd4/Nedd4-like family and/or ubiquitination
may play an important role in CaV1.2 mid-channel proteolysis. More about ubiquitination of the L-type calcium channel is summarized in 1.5.2.2.
Other ion channels have also been shown to be regulated by ubiquitination (81).
16
An important function of ubiquitination is “quality control” of ion channels.
Unassembled and/or misfolded ion channels in the ER can be degraded through
ER-associated degradation (ERAD), which involves the ubiquitin-proteasome system
(81, 94). Ubiquitination is reversible. Ubiquitin moieties can be removed from a substrate protein by deubiquitinating enzymes (DUBs) (95). Deubiquitination of ion channels usually stabilizes the channels on the plasma membrane. Ubiquitin-specific protease (USP) 2-45, a member of the DUB family, has been reported to counteract
Nedd4-2-mediated ubiquitination of ENaC (96, 97) and KCNQ1 (KV7.1) (98) which
subsequently increase the surface expression of the channels. In the case of CaV1.2, however, USP2-45 deubiquitinates the channel and reduces surface expression of the
channel (99). More about the regulation of CaV1.2 by USP2-45 is summarized in
1.5.2.2.
1.5.2.2 Ubiquitination of the L-type calcium channel
CaV subunits modulate the function of CaV1 and CaV2 channels in two ways.
First, they affect the biophysical properties of the channels by increasing the opening probability and hyperpolarizing the voltage dependence of activation (100, 101).
Secondly, CaV subunits increase the number of channels on the plasma membrane
(102-104). Early studies on the P/Q-type channel suggested that the subunit masks an ER retention motif in the CaV1 I-II loop, thereby promoting the trafficking of the
1 subunit to the plasma membrane (102, 105), although no specific motif was
17 identified (105). More recently, it was suggested that the 1C I-II loop actually contains an acidic ER export motif (106). Studies of fusion proteins containing a CD4 protein fused to different fragments of 1C and 1B show that both channels contain
ER retention signals in the C-terminal tails but not in the I-II loops (84).
subunits markedly enhance surface expression of 1C, and this effect of the subunit can be mimicked by the proteasome inhibitor MG132. This indicates that the checkpoint controlling the surface expression of 1C is proteasomal degradation instead of ER retention. subunits do not exert their function by masking the ER retention signal (84). Instead, they promote surface expression of 1C by protecting
1C from ER-associated degradation (ERAD) (84). Without subunits, 1C is subject to robust ubiquitination by the E3 ubiquitin ligase RFP2 and interacts with the
ERAD complex proteins derlin-1 and p97, which mediate the transport of 1C to the proteasome for degradation (84). subunits interfere with the ubiquitination of 1C and the interaction of the channel with the ERAD complex, thereby promoting the trafficking of the channel to the plasma membrane (84).
Recently, a splice variant of the CaV1.2, named CaV1.2e21+22, was found to exert a dominant negative effect on WT CaV1.2 by sequestering subunits and subsequently enhancing the ubiquitination of WT CaV1.2 (107). CaV1.2e21+22 contains both exons 21 and 22, which encode S2 as well as part of the S1-S2 loop of repeat III and are usually mutually exclusive (107). This variant is highly expressed in neonatal rat hearts, and also abundant in the hearts of mice with induced cardiac hypertrophy, as well as hearts
18 of patients with dilated cardiomyopathy (107). CaV1.2e21+22 is retained intracellularly and does not conduct currents (107). CaV has significantly higher affinity for
CaV1.2e21+22 than for WT CaV1.2, but does not increase the overall or surface expression of CaV1.2e21+22 (107). By competing with WT CaV1.2 for CaV,
CaV1.2e21+22 enhances ubiquitination of WT CaV1.2 and subsequently proteasomal
degradation of WT CaV1.2 (107).
As mentioned in 1.5.2.1, the Nedd4/Nedd4-like E3 ubiquitin ligases bind to the
PY motifs of a variety of ion channels, ubiquitinate the channels, and subsequently downregulate their surface expression. Nedd4-1 (also called Nedd4), a member of the
Nedd4/Nedd4-like family, has also been shown to regulate the functional expression
of CaV1.2 (108). Overexpression of Nedd4-1, but not Nedd4-2 or WWP2, decreases
whole-cell CaV1.2 currents (108). Nedd4-1 also downregulates surface expression and overall expression of 1C, 2, and 2-1(108). Although CaV1.2 has a non-canonical
PY motif (LPYVAL) (4), overexpression of Nedd4-1 does not increase ubiquitination of 1C, 2, or 2-1 (108). Nedd4-1 promotes the transport of CaV1.2 from the
ER-Golgi network to the proteasome and lysosome for degradation, without
ubiquitinating the channel (108). The transport of CaV1.2 to the lysosome relies on clathrin-coated vesicles (108). The regulatory effect of Nedd4-1 relies on the subunit and does not involve the internalization of surface CaV1.2 (108). The SH3 domains of all four subfamilies of subunits contain a YDVV motif (YXX motif), which is a signal for selection into clathrin-coated vesicles (108). Mutation of this
19 motif eliminates the Nedd4-1-induced lysosomal degradation of CaV1.2 and partially reversed the effect of Nedd4-1 on Cav1.2 currents, indicating that Nedd4-1 may target
CaV-interacting proteins (108). Note that these findings point out a dual role of CaV: increasing and decreasing the functional expression of Cav1.2 (102-104). The authors did not point out that the non-canonical PY motif of 1C also encompasses a YXX motif (YVAL). Thus, it would be interesting to study whether this motif also
contributes to the regulatory effect of Nedd4-1 on CaV1.2.
While there is no evidence that any member of the Nedd4/Nedd4-like family can bind to the non-canonical PY motif of 1C, the motif has been found to be important for the mid-channel proteolysis of 1C (4). Mutation of the motif significantly reduces the mid-channel proteolysis of 1C (4), suggesting the participation of some member of the Nedd4/Nedd4-like family in this process. It has also been found that
1C bears two PEST sequences, PEST1 and PEST3, in the I-II loop and the II-III loop, respectively (109). PEST sequences have been shown to serve as signals for rapid degradation of proteins by calpain and the ubiquitin-proteasome system (110).
The mid-channel proteolysis of 1C is greatly reduced by the deletion of PEST3 and is virtually abolished by the deletion of PEST1 (4). Moreover, the mid-channel proteolysis of 1C can be significantly reduced by a cocktail of ubiquitin aldehyde, a general inhibitor of ubiquitin-recycling processes, and MG-132, a proteasome inhibitor (4). These findings suggest a pivotal role of the ubiquitin-proteasome system in the mid-channel proteolysis of 1C. Additional information about the mid-channel
20 proteolysis of 1C is summarized in 1.6.1.2.
Deubiquitinating enzymes (DUBs) also regulate the functional expression of
CaV1.2. Ubiquitin-specific protease (USP) 2-45, a member of the DUB family, was
recently reported to downregulate the functional expression of CaV1.2 (99). USP2-45 deubiquitinates both 1C and 2-1, and decreases surface expression of 1C, as well as the overall expression of 1C, 2, and 2-1 (99). USP2-45 interacts with
2-1 but not 1C and 2 (99). 2-1 is necessary for the regulatory effect of
USP2-45 on CaV1.2 (99). Note that these findings point out a dual role of the 2 subunit: increasing (111, 112) and decreasing the functional expression of CaV1.2. The finding that 2-1 can be ubiquitinated and deubiquitinated indicates that part of
2-1 is exposed to the cytosol so that E3 ubiquitin ligases and USP2-45 can have access to 2-1. However, the recently discovered cryo-EM structure of CaV1.1 shows that 2-1 is an extracellular protein with no transmembrane or intracellular segment (30). This discrepancy is probably due to the different conditions or different types of channels used in the experiments.
1.5.2.3 Ubiquitination of the N-type calcium channel
In general, the effect of ubiquitination on CaV2.2 is similar to that on CaV1.2: promoting proteasomal degradation and decreasing functional expression of the channel (85, 113, 114). The effect of subunits on CaV2.2 also resembles that on
CaV1.2: suppressing proteasomal degradation and increasing functional expression of
21 the channel (115, 116). However, the detailed mechanisms of these regulatory
processes show obvious differences between CaV2.2 and CaV1.2.
Mouse CaV2.2 has a site of alternative splicing involving a pair of mutually exclusive exons, e37a and e37b, which encode the proximal C-terminus (117).
CaV2.2-e37b is the major form of CaV2.2 in the nervous system, except in some cells,
such as the nociceptors of dorsal root ganglia, where CaV2.2-e37a is enriched (117).
CaV2.2-e37b is subject to a higher level of ubiquitination than CaV2.2-e37a (114).
Since CaV2.2-e37b is more sensitive to proteasomal degradation it is associated with
lower current density than CaV2.2-e37a (114).
CaV2.2 contains plenty of intracellular lysine residues that can potentially be ubiquitinated. However, the 12 lysine residues in the I-II loop are suggested to have a critical role in the ubiquitination and proteasomal degradation of the channel, since mutation of all the 12 lysine residues prevents proteasomal degradation (115).
Palmitoylated 1B I-II loop is polyubiquitinated and degraded by the proteasome in the absence of subunits (115). Mutation of all 12 lysine residues to arginines prevents ubiquitination and proteasomal degradation (115). However, even if proteasomal degradation is prevented by the lysine-to-arginine mutations or proteasome inhibitor MG132, subunits are still needed for the expression of palmitoylated 1B I-II loop on the plasma membrane (115). Interestingly, subunits do not protect palmitoylated 1B I-II loop from polyubiquitination, but can prevent proteasomal degradation and promote surface expression of the construct (115). In
22 accordance with these findings, full-length 1B also relies on subunits to traffic to the plasma membrane, even if proteasomal degradation is prevented by mutation of the 12 lysine residues in the I-II loop (115) or by the proteasome inhibitor MG132
(116). Therefore, functional expression of 1B is controlled by two distinct checkpoints: proteasomal degradation and a proteasome-independent checkpoint controlling the transport to the plasma membrane. subunits help the channel to pass both checkpoints, while inhibition of ubiquitination only helps the former.
The mechanism by which CaV promotes trafficking of 1B to the plasma membrane is not fully understood. Similar to 1C, 1B bears ER retention signals in the C-terminal tail (84), but these signals do not effectively retain the channel in the
ER. 1B with a mutation (W391A) that prevents binding of subunits to its AID is not selectively retained in the ER, but cannot traffic to the plasma membrane even though the overall expression level can be enhanced by the proteasome inhibitor
MG132 (116). It has been shown that the AID of CaV1 exists as a random coil in the absent of CaV (118), whereas in the presence of CaV, the AID forms a continuous
-helix along with the entire linker between S6 of repeat I and the AID (118-120).
This conformational change induced by the binding of CaV might be crucial for the targeting of the channel to the plasma membrane (115).
Ubiquitination also regulates the internalization of CaV2.2. It has been reported that the microtubule-associated protein 1B light chain 1 (MAP1B-LC1) interacts simultaneously with the C-terminal tail of 1B and the ubiquitin-conjugating enzyme
23
E2 L3 (UBE2L3) (113). This interaction occurs on the two main variants of the
CaV2.2 channels, CaV2.2-e37a and CaV2.2-e37b (85), and presumably causes the ubiquitination of 1B on the plasma membrane via an unknown E3 ubiquitin ligase
(85, 113). The ubiquitination induces dynamin-dependent endocytosis (85) and
subsequently proteasomal degradation of CaV2.2 (113).
1.5.2.4 Summary of ubiquitination of voltage-gated calcium channels
Ubiquitination can downregulate the functional expression of both CaV1.2 and
CaV2.2. It has been shown that ubiquitination causes CaV1.2 to be transported from the ER to the proteasome for degradation, but there is no reported evidence that
ubiquitination induces internalization of surface CaV1.2. Surface CaV2.2, however, can be internalized after ubiquitination mediated by MAP1B-LC1 and UBE2L3.
CaV promotes surface expression of both CaV1.2 and CaV2.2. In the case of
CaV1.2, CaV protects the channel from ubiquitination and ER-associated degradation
(ERAD). In the absence of CaV, inhibition of proteasomal degradation by MG132 is sufficient to allow surface expression of CaV1.2. In the case of CaV2.2, however, CaV seems to promote surface expression of the channel without preventing ubiquitination.
Moreover, inhibiting ubiquitination or proteasomal degradation is not enough to allow surface expression of CaV2.2 in the absence of CaV. The conformational change induced by the binding of CaV seems to be necessary for CaV2.2 to be transported to the plasma membrane.
24
Ubiquitination of CaV2.1has been studied to a lesser extent. MAP1B-LC1 and
UBE2L3 have been shown to interact with not only CaV2.2 but also CaV2.1 (85),
indicating that CaV2.1 may also be regulated by ubiquitination. However, it has been reported that there is little ubiquitination of CaV2.1 in the absence of CaV (84).
Further studies are needed to reveal whether and how CaV2.1 is regulated by ubiquitination and the mechanism by which CaV promotes surface expression of
CaV2.1.
1.6 Proteolysis of ion channels
1.6.1 Proteolysis of L-type calcium channels
1.6.1.1 Proteolytically cleaved C-terminal fragment of L-type calcium channel
1 subunit acting as an autoinhibitory domain
The full-length 1S subunit of the CaV1.1 channel in rabbit skeletal muscle is
212-kDa, as predicted from the cDNA clone (121). However, immunoblot analysis of membrane preparations of rabbit skeletal muscle revealed two size forms of 1S, 190 kDa and 212 kDa, and more than 80% of the 1S subunits are in the short 190-kDa form (122, 123). Two size forms of 1S were also detected in cultured rat skeletal muscle myocytes and found to share common sites of PKA-dependent phosphorylation with the 1S in rabbit skeletal muscle (124). The 1C subunit of the
CaV1.2 channel was also found to have two size forms, a short form of 190-210 kDa, and a long form of 210-242 kDa in rat brain (125, 126) and in rabbit heart (51, 127).
25
Immunoprecipitation and immunoblot analysis indicated that the short form is a truncated form of 1 missing the C-terminus (51, 122, 123, 126, 127).
The amount of the short form of 1C was not reduced by the use of protease inhibitors and lower temperature during sample preparation or perfusion of rats with membrane-permeable protease inhibitors (126). Moreover, 1S and1C expressed in heterologous systems are only present in the full-length form (127-130). These results indicate that the short form of the channel is present in live tissue and cells, instead of due to proteolysis during sample preparation.
The two size forms of CaV1 are differentially phosphorylated. While both size forms of 1C purified from rat brain could be phosphorylated in vitro by PKC,
CaMKII and PKG, only the long form could be phosphorylated by PKA. Therefore,
PKA preferentially phosphorylates sites in the distal C-terminal region that are present only in the long form of 1C (126, 131, 132). PKA also preferentially phosphorylates sites in the distal C-terminal region of 1S purified from rabbit skeletal muscle and cultured rabbit skeletal muscle myocytes (133, 134). Immunoprecipitation and two dimensional phosphopeptide mapping revealed that the major PKA phosphorylation sites are in the distal C-terminal region of 1S are S1854 (133) and, to a lesser extent,
S1757 (134). Interestingly, phosphorylation of S1854 has an inhibitory effect on the phosphorylation of other regions of 1S (133). Similar phenomena were observed on rat and rabbit cardiac 1C subunits, of which only the long form could be phosphorylated by PKA in vitro (51, 52). Protein microsequencing and
26 phosphopeptide mapping revealed that a major phosphorylation site in cardiac 1C is
S1928 (51), which can be phosphorylated by PKA in response to activation of the
-adrenergic receptor pathways (52) and by PKC (53). Phosphorylation of S1928 by
PKA has been shown to mediate, in part, CaV1.2 activation (54).
A C-terminally truncated protein may be due to posttranslational proteolysis or alternative splicing. The hypothesis that the short form of L-type calcium channel 1 subunit arises from proteolysis has been favored since the two size forms were first discovered in rabbit skeletal muscle (122). cDNA cloning detected only one mRNA for the rabbit skeletal muscle 1S (121), which accounts for the 212-kDa full-length form, while the 190-kDa short form is much more abundant. The C-terminal tails of
1S and 1C were shown to be sensitive to calpain and chymotrypsin, which can cleave the 1 subunit in vitro into a short form of similar or identical size to the short form found in the animal tissue (135-137). The amount of the short form of 1 in comparison with the full-length form can be altered. Ca2+ influx mediated by NMDA receptors was shown to induce proteolytic conversion of the long form of 1C into the short form in rat hippocampal slices through a calpain-dependent mechanism, while other classes of 1 subunits were not affected (136). In rare cases, the putative
C-terminal fragment of 20-40 kDa produced in vivo could be detected in immunoblot analysis. One study showed that an antibody against the C-terminal tail of 1C detected a 37-kDa species in E16 and adult ventricles, but no full-length channel was detected, probably due to the low amount (138). These results support the hypothesis
27 that the short form of L-type calcium channel 1 subunit arises from proteolysis.
Although the putative C-terminal fragment of 20-40 kDa produced in vivo has rarely been detected in immunoblot analysis, numerous studies showed evidence for its existence and shed light on its function. The Hosey group found that in freshly isolated rabbit cardiac myocytes, most 1C subunits are in the 200-kDa short form instead of the 242-kDa full-length form, as detected by an antibody against an internal domain of 1C in immunoblot analysis (127). However, immunocytochemistry analysis revealed that the C-terminus is present and colocalizes with the main body of the 1C in rabbit cardiac myocytes. In addition, the amount of C-termini relative to the total amount of 1C subunits is similar to that of the transfected HEK cells, where
1C is present only in the full-length form (127). This indicates that the cleaved
C-terminal fragment may remain functionally associated with the channel.
What effect does the cleaved C-terminal fragment have on the channel? Studies have shown that removal of 147-500 amino acid residues from the C-terminus of 1C leads to a significant increase in channel currents (137, 139, 140), suggesting that the
C-terminus participates in inhibiting channel function. Additionally, the Hosey group also found that a proline-rich domain (PRD, residues 1974-2000) in the C-terminal tail of rabbit 1C is not only important for membrane association of the chymotrypsin-cleaved C-terminal fragments, but also contributes to the inhibition of channel currents (137). Engineered C-terminal fragments were then applied to different C-terminally truncated 1C subunits, which resulted in two of the
28
C-terminal fragments, CT4 (residues 1909-2171) or CT7 (residues 2030-2171), having inhibitory effects on the truncated 1C, with CT4 having a greater effect (140).
Moreover, the segment between residues 1733-1905 is not only needed for the truncated 1C to be inhibited by the C-terminal fragments, but also able to associate with the C-terminal fragments, indicating that residues 1733–1905 might act to anchor or stabilize the inhibitory domain within the C-terminal fragments (140). One model suggests that the C-terminal tail of 1C is proteolytically cleaved at approximately residue 1900 and the distal C-terminus (DCT) binds to the DCT binding region between residues 1733-1900 to serve as an inhibitory domain (140).
The exact site of proteolysis was never known until the Catterall group identified it as the carboxyl side of A1664 by digesting 1S protein purified from rabbit skeletal muscle with three proteases and analyzing the proteolytic peptides using LC/MS/MS
(130). Proteolysis at this site in theory will result in a truncated 1S of 190 kDa, consistent with previous observations (122, 123). A1664 in rabbit 1S is conserved in rabbit and rat 1C as A1800, where a cleavage in theory will result in a truncated 1C of 203 kDa, consistent with previous observations (51, 125-127).
Like the Hosey group, the Catterall group also obtained results showing that the
C-terminal fragments of L-type calcium channel 1 subunits interact with and have an inhibitory effect on the channel (130, 141). Coimmunoprecipitation experiments showed that the C-terminal fragment of rabbit 1S can associate with the truncated channel in a heterologous system (130). Yeast two-hybrid assay revealed the specific
29 interaction between a proximal C-terminal interaction domain (residues 1556–1612) and a distal C-terminal interaction domain (residues 1802–1841) (130). Note that the two domains are proximal and distal to the cleavage site (A1664), respectively.
Similar results were obtained with rabbit 1C, whose C-terminal fragment associates with the truncated channel in a heterologous system and serves as an autoinhibitory domain by decreasing the coupling between voltage sensing and channel opening and by shifting the voltage dependence of activation to more positive membrane potentials
(141). Moreover, site-directed mutagenesis, coimmunoprecipitation, and structural modeling revealed a binding interaction between a pair of arginine residues (R1696,
R1697) in the proximal C-terminal regulatory domain, (previously called proximal
C-terminal interaction domain). Additionally, three negatively charged amino acid residues (E2103, E2106, D2110) are present in the distal C-terminal regulatory domain (previously called distal C-terminal interaction domain) (141). This
interaction is critical to the autoinhibitory effect of the C-terminal fragment on CaV1.2 channel function (141). Interestingly, the autoinhibitory effect caused by the non-covalent interaction between the cleaved C-terminal fragment and the truncated
1 subunit is much more potent than when the C-terminal fragment is covalently associated with the proximal C-terminal domain in the full-length channel (141).
These results indicate that the cleaved C-terminal fragment and the truncated 1 subunit form a one-to-one autoinhibitory complex (141).
As the molecular identity and autoinhibitory function of the L-type calcium
30 channel C-terminal fragment were being revealed, progress was also being made on understanding the cAMP/PKA-dependent regulation of L-type calcium channels.
Activities of L-type calcium channels in both skeletal and cardiac muscle are regulated by the -adrenergic receptor pathway in the fight or flight response (2).
Stimulation of the -adrenergic receptor pathway activates the cAMP/PKA signaling pathway, which then increases L-type Ca2+ currents through PKA-mediated phosphorylation of the L-type calcium channels and/or associated proteins (2). In both skeletal and cardiac muscle, PKA-dependent upregulation of L-type calcium channel activity requires anchoring of PKA to the CaV1 subunit via an A-kinase anchoring protein (AKAP), typically AKAP15 (2). AKAP15 binds to 1S and 1C via a modified leucine zipper interaction (142, 143). Interestingly, the AKAP15-binding site is distal to the proteolytic cleavage site (142, 143). This means that PKA and AKAP15 are anchored to the C-terminal fragment cleaved off from the 1 subunit. Because binding of AKAP15 and PKA to the distal C-terminal domain is required for cAMP/PKA-dependent regulation of L-type calcium channels (2), these results offer another piece of evidence that cleaved C-terminal fragment can remain associated with the channel.
These observations connected the proteolysis of L-type calcium channel
C-terminal tail with cAMP/PKA-dependent upregulation of L-type Ca2+ currents and raise the possibility that cAMP/PKA-mediated phosphorylation of the 1 subunit does not inhibit the truncated 1 subunit with a bound distal C-terminal fragment. The
31
Catterall group proposed a biophysical model, in which cardiac 1C subunits are present in three forms, full-length, truncated, and truncated plus a non-covalent bound distal C-terminal fragment, in which cAMP/PKA-mediated phosphorylation of the
C-terminal relieves the autoinhibition by the distal C-terminal fragment (141). Using this model, they precisely simulated the increase in Ca2+ currents rat ventricular myocytes (141).
Further studies have made revealing discoveries on the molecular mechanism of the disinhibition of the autoinhibitory complex. The Catterall group found that
cAMP/PKA-mediated upregulation of cardiac CaV1.2 activity in a heterologous system requires co-expression of a carefully titrated amount of AKAP15 (55). This
indicates that cardiac CaV1.2 forms an autoinhibitory signaling complex comprised of the C-terminally truncated 1C, the C-terminal fragment, AKAP15, and PKA (55).
With optimized expression of the different components of the autoinhibitory signaling
complex, PKA-dependent regulation of CaV1.2 activity was able to be reconstituted in a heterologous system with a dynamic range similar to that in cardiac myocytes (55).
Moreover, mass spectrometry revealed two previously unidentified sites of in vivo phosphorylation in rabbit skeletal muscle 1S, S1575, and T1579 (56), which are conserved in 1C as S1700 and T1704, respectively (55). S1575/S1700 is a substrate for PKA and CaMKII, while T1579/T1704 is a substrate for casein kinase II (55, 56).
These two phosphorylation sites are “strategically” located in the proximal C-terminal regulatory domain and are exposed to the interface between the proximal and distal
32
C-terminal regulatory domains. At this interface, negatively charged phosphates could disrupt the ionic interactions between the two domains and relieve autoinhibition (55,
56). Site-directed mutagenesis and electrophysiology revealed that basal activity of
CaV1.2 in unstimulated cells is regulated by phosphorylation of both S1700 and
T1704, whereas further upregulation of channel activity through stimulation of the cAMP/PKA pathway requires only the phosphorylation of S1700 (55).
Transgenic mouse lines were created to test the functional role of the 1C distal
C-terminus, S1700, and T1704 (144-146). Mice expressing only 1C without the distal C-terminus (DCT) have severely impaired cardiac function and die perinatally
(144). Counterintuitively, cardiomyocytes from these mice have reduced CaV1.2 currents, even though the autoinhibitory distal C-terminus is absent since mutant channels have low cell surface expression (144). The -adrenergic up-regulation of
CaV1.2 activity is completely lost in myocytes from the DCT mice (144). In two other mouse lines, S1700 alone or together with T1704 was mutated to alanine. In both mouse lines, basal CaV1.2 activity and -adrenergic-stimulated CaV1.2 activity were both significantly reduced, and both mouse lines suffer impaired cardiac function and premature deaths (145, 146). These results indicate that the distal
C-terminus and the two phosphorylation sites, S1700 and T1704, have crucial regulatory functions and are indispensible for cardiac development and function in vivo. Note that in the two mouse lines with mutated phosphorylation sites,
-adrenergic upregulation of CaV1.2 activity is not completely lost, and the reduction
33 in cellular contractility is not as severe as the reduction in CaV1.2 current (145, 146).
These findings suggest that there are other mechanisms for the -adrenergic upregulation of CaV1.2 activity and regulation of cardiomyocyte contractility independent of S1700 and T1704.
In summary, most CaV1 subunits in skeletal and cardiac muscle are present as a truncated 1 subunit with a non-covalently bound C-terminal fragment, in which the distal C-terminal regulatory domain associates with the proximal C-terminal regulatory domain of the truncated 1 subunit via ionic interactions. PKA is anchored to the C-terminal region of the channel via AKAP, which binds to the AKAP binding site on the C-terminal fragment. The truncated 1 subunit, C-terminal fragment,
AKAP, and PKA forms an autoinhibitory signaling complex, which has basal channel activity. In the fight or flight response, stimulation of the -adrenergic receptors causes cAMP/PKA-mediated phosphorylation of S1575 (in skeletal muscle 1S) or
S1700 (in cardiac 1C). The phosphate group disrupts the ionic interactions between the proximal and distal C-terminal regulatory domains and relieves autoinhibition, causing up-regulation of channel activity. A schematic diagram of the autoinhibitory signaling complex and its regulatory mechanism is depicted in Figure 1.
34
Figure 1. Schematic of the cardiac CaV1.2 autoinhibitory signaling complex and the mechanism of cAMP/PKA-mediated upregulation. The C-terminal tail of 1C is proteolytically cleaved on the carboxyl side of A1800. The portion between residues 1682-1740 and the portion between residues 2081-2124 serve as the proximal C-terminal regulatory domain (PCRD) and the distal C-terminal regulatory domain (DCRD), respectively. Pink dots represent the pair of arginine residues, R1696 and R1697. Red dots represent the group of negatively-charged residues, E2103, E2106, and D2110. The ionic interaction between these two groups of residues is critical to the autoinhibitory effect of the C-terminal fragment on the channel. The portion between residues 1740 and 1800 enhances the binding of the distal C-terminal fragment to the proximal C-terminus, and thus is also necessary for the autoinhibitory effect. The proline-rich domain (PRD, 1974-2000) enhances the efficacy of the autoinhibition. The autoinhibitory signaling complex also includes AKAP15, which is anchored to the AKAP15-binding domain (ABD, residues 2063-2080) of the C-terminal fragment. AKAP15 anchors PKA to the C-terminal fragment. PKA is activated by cAMP and subsequently phosphorylates S1700 (represented by the yellow dot). The negatively charged phosphate group disrupts the ionic interaction between the proximal and distal C-terminal regulatory domains and relieves autoinhibition. (Modified from Gao et al., 2001 (140)).
There are still unanswered questions about the L-type calcium channel autoinhibitory signaling complex. The type of enzyme that catalyzes the proteolytic cleavage in the C-terminal tail is still not confirmed, although both calpain and chymotrypsin are known to produce truncated 1 of similar size to that found in the animal tissue. The signal that induced the proteolysis is also unknown. Moreover, it is 35 still unclear what mechanisms cause the remaining sensitivity of cardiac CaV1.2 to the
-adrenergic receptor pathway in mice with both S1700 and T1704 mutated to alanine.
Are other phosphorylation sites responsible for this? What are the functions of those previously discovered major phosphorylation sites distal to the cleavage site, i.e.,
S1854 and S1757 in 1S, and S1928 in 1C? More research is needed to answer these questions and further understand the proteolysis of the L-type calcium channel
C-terminal tail and the autoinhibitory signaling complex.
1.6.1.2 Mid-channel proteolysis of the L-type calcium channel 1 subunit
A recent study in our lab by Michailidis and colleagues provided evidence for another type of regulatory proteolysis of the L-type calcium channel—mid-channel proteolysis (4). Mid-channel proteolysis is proteolytic cleavage of the main body of the subunits, instead of the C- or N-terminal tail. A research from our group (4) showed that the mid-channel proteolysis of rat brain 1C is a calcium-dependent process, serving as negative feedback regulation of the channel.
This research (4) utilized surface biotinylation of brain tissue and cultured neurons to focus on the 1C on the plasma membrane. Western blot analysis of rat cortical slices revealed 1C subunits of several different sizes. Two most prominent forms were the 240-kDa full-length channel, and a 150-kDa fragment-channel recognized by both an antibody against the II-III loop and an antibody against the
C-terminal tail. The 150-kDa fragment-channel presumably contains the part of II-III
36 loop, repeats III and IV and the C-terminus, and results from mid-channel proteolysis in the II-III loop. A complementary fragment-channel containing the N-terminus and repeats I and II was detected by an antibody against the N-terminal tail. Additional lengths of 1C were also detected.
To address the question whether the complementary 1C fragments remain associated with each other or sometimes separate, a tagged 1C construct was used
(4). This construct at the N-terminus was tagged with GFP, and repeat III was tagged extracellularly with HA tag, which was later immunostained in red. Confocal imaging showed that in some locations of the cell surface, red signal does not colocalize with green signal, indicating that some complementary fragment-channels separate from each other on the plasma membrane. Using constructs with HA tags in different positions, this research (4) also showed that there may also be cleavage sites in the I-II loop and the III-IV loop.
The mid-channel proteolysis was revealed to be regulated by cellular calcium activity. Reagents that increase intracellular [Ca2+], L-type calcium channel activity and/or cell excitability increase the intensity ratio of the 150-kDa fragment relative to the full-length channel, and also increase the separation of red and green. Reagents that decrease L-type calcium channel activity and/or cell excitability have the opposite effect. Calpain, which is responsible for proteolysis of voltage-gated sodium channels
(1.6.2.1), was shown to contribute to the mid-channel proteolysis of 1C and the separation of the fragment-channels. Interestingly, the ubiquitin-proteasome system
37 was also shown to be involved in the mid-channel proteolysis. Confocal imaging showed that inhibitors of the ubiquitin-proteasome system, and mutations made to a putative ubiquitination motif (PY motif) in an intracellular loop of repeat IV, markedly reduced the separation of the fragment-channels on the plasma membrane. Two PEST sequences, PEST1 and PEST3, were found in 1C in the I-II loop and the II-III loop, respectively, and were shown to play important regulatory roles (109). PEST sequences have been shown to serve as signals for rapid degradation of proteins by calpain and the ubiquitin-proteasome system (110). Our group (4) showed that deletion of PEST1 or PEST3 greatly reduces the separation of the fragment-channels.
Using engineered complementary fragment-channels, our group (4) showed that the two complementary fragments do not conduct currents individually, but together can form a functional channel and conduct currents with an amplitude of about 35% of that of the full-length channel. The channel formed by complementary fragment-channels also displays altered biophysical properties, with the I-V relationship and voltage dependence of steady-state inactivation shifted ~10 mV to the depolarizing direction. Incubating neurons with reagents that enhance L-type calcium channel activity not only increases mid-channel proteolysis, but also reduces currents of L-type calcium channels, corresponding to the smaller current amplitude of channels formed by complementary fragments. Blocking the L-type calcium channel has the opposite effects. Note that the decrease in current amplitude following channel activation may also be due to internalization of full-length channels. Nevertheless,
38 these results suggest that mid-channel proteolysis serves as a negative feedback regulation of the L-type calcium channel, and most likely provides protection to the neurons against calcium-induced cell death (147). Our group (4) examined the mid-channel proteolysis across the life span, and found that mid-channel proteolysis increases steadily with age, being ~7 times more intense in 16-month old rats than in
10-day old rats. This phenomenon was seen in additional studies (148). The increase in mid-channel proteolysis associated with aging is probably a compensatory response to the aging-associated increase in L-type calcium channel currents (149-151).
Both the Western blot analyses and imaging experiments (4) provided us with evidence that mid-channel proteolysis happens at several locations in 1C. Three pairs of complementary fragment-channels, (A1 & A2; B1 & B2; C1 & C2), were constructed to mimick the products of mid-channel proteolysis in loop I-II, II-III, and
III-IV, respectively. None of the fragment-channel conducts currents on its own, but all of them can conduct currents when paired with a complementary fragment. Our group (4) also investigated the influence of these fragment-channels on the full-length
CaV1.2 channel. C2 is the only fragment that has a dominant negative effect on the currents of the full-length channel. Neither B1 nor B2 has a significant effect on the biophysical properties of the full-length channel. A1, A2, and C1 all can alter the I-V relationship and voltage dependence of steady-state inactivation of the full-length channel.
This research from our group (4) uncovered a novel form of negative feedback
39 regulation of the L-type calcium channel but also raised new questions. Where are the precise cleavage sites? Does mid-channel proteolysis take place on the plasma membrane or in the cytoplasm? What pulls the complementary fragments apart? What is the fate of the fragment-channels after they separate? Do the fragment-channels undergo conformational changes in response to membrane depolarization, like a
95-kDa truncated 1A (152), and convey information into the cell? Does mid-channel proteolysis also happen to other types of VGCCs, such as P/Q- and N-type calcium channels? As a novel and intriguing regulatory process, mid-channel proteolysis needs to be further investigated.
1.6.2 Proteolysis of voltage-gated sodium channels
Vo ltage-gated sodium channels (VGSCs or NaChs) are essential proteins in neurophysiology. VGSCs initiate action potentials (APs) in excitable tissues, and thus are critical for the regulation of the excitability of neurons and myocytes (153). The pore-forming subunit of a VGSC is the subunit, which consists of four homologous domains, each containing six transmembrane helices (153), resembling VGCCs. The four homologous domains of a VGSC are joined by intracellular linkers, and the N- and C-termini are also located intracellularly (153).
VGSCs are regulated through proteolysis under normal (154) and stress conditions (155). Pharmacological activation of VGSCs leads to downregulation of surface VGSCs by endocytosis followed by proteolysis of the subunit, which
40 produces a 90-kDa fragment, as detected using a radioactive VGSC agonist (154).
Axon stretch injury causes proteolysis of the III-IV loop and probably other regions of the VGSC subunit, resulting in a 55-kDa fragment that can be detected by an antibody against the I-II loop (155). Proteolysis of VGSCs can also be induced by anoxia or ischemia (156).
Calpain has been shown to be an important protease underlying the proteolysis of
VGSCs (157). Activation of calpain leads to proteolysis of NaV1.2 in rat brain lysate and cultured rat cortical neurons (158). The proteolysis causes the loss of the 260-kDa full-length NaV1.2 subunit and produces several fragments, including a 160-kDa and 110-kDa band recognized by an antibody against the III-IV loop and a 150-kDa,
100-kDa and 80-kDa band recognized by an antibody against the I-II loop (158). It is therefore inferred that the NaV1.2 subunit is cleaved at two sites (158). A cleavage in the I-II loop produces a 100-kDa N-terminal fragment and a 160-kDa C-terminal fragment, and a cleavage in the II-III loop produces a 150-kDa N-terminal fragment and a 110-kDa C-terminal fragment (158). The 80-kDa band recognized by the I-II loop antibody is most likely due to a cleavage in the N-terminal tail (158).
Interestingly, most of the channel fragments remain on the plasma membrane 6 hours after calpain activation (158). These fragments may interact with each other to form channels with modified biophysical properties, like the fragments of the L-type calcium channel (4).
An in vitro model of traumatic brain injury shows that stretch injury to cortical
41 neurons can cause activation of calpain and subsequently lead to proteolysis of
VGSCs, producing the 160- and 110-kDa fragments recognized by the III-IV loop antibody (158, 159). Chronic inhibition of NMDA receptors (NMDARs) and VGSCs, but not L-type calcium channels, significantly inhibits calpain activation and VGSC proteolysis following stretch injury (159). Moreover, an early transient inhibition of
NMDARs also significantly reduces VGSC proteolysis following injury (159). These results indicate that calcium influx through NMDARs is critical for the proteolysis of
VGSCs by calpain.
It has been shown that proteolysis of the VGSC subunit prevents normal inactivation of the VGSC and thus leads to persistent influx of Na+ (160-162), which may subsequently lead to Ca2+ influx and cause further pathological changes (155).
Spinal cord injury in rats has been shown to cause calpain-dependent cleavage of
VGSCs associated with an upregulation of persistent sodium current (INaP) in
motoneurons (163). INaP is an important contributing factor to spasticity following spinal cord injury (164, 165). Intraperitoneal injection of calpain inhibitor MDL28170 not only reduces production of the 120-kDa product of VGSC protealysis (recognized
by antibody against the II-III loop), but also reduces INaP in motoneurons and spasticity (163). These results reveal the important role of VGSC proteolysis in promoting the complications from spinal cord injury and offer a potential therapeutic tool for recovery.
Inhibition of calpain is a promising therapeutic approach for neuropathy and
42 nerve injury. Overexpression of the calpain inhibitor calpastatin in transgenic mice significantly reduces hippocampal cell death following excitotoxic insult (166).
Following traumatic brain injury, transgenic mice overexpressing calpastatin showed less proteolysis of VGSCs and several other calpain substrates. In addition, the transgenic mice have lower levels of behavioral deficits compared with wild-type mice, although there is no attenuation of the tissue damage (167, 168). Inhibition of calpain by MDL28170 has also been shown to attenuate biochemical, functional, and behavioral deficits in a model of diabetic neuropathy (169).
Other proteases have also been found to catalyze VGSC proteolysis. It has been shown that in heart disease, aminopeptidases catalyze the excision of the N-terminal
methionine of NaV1.5, followed by acetylation of the resulting N-terminal alanine
(170). Truncations in NaV1.7 due to nonsense mutations have been shown to cause loss of function of the channel and subsequently lead to congenital insensitivity to pain (171). Interestingly, a missense mutation (R907Q) neighboring one of the nonsense mutations (R908X) in the S5-S6 extracellular loop of repeat II also causes the disorder (172). Further research has shown that R907Q renders the Y908-H909 bond more sensitive to matrix metalloproteinase 9 (MMP-9), indicating that
aberrantly increased proteolysis may cause the loss of function of NaV1.7-R907Q and subsequently lead to congenital insensitivity to pain (173).
43
1.7 Splice variants and truncated VGCCs
1.7.1 A C-terminal fragment of 1C acting as a transcription factor
As discussed in 1.6.1.1, 1C and 1S can be proteolytically cleaved to produce a
C-terminal fragment of 20-40 kDa, which has an autoinhibitory effect on the full-length L-type calcium channel and is involved in the upregulation of channel activity in the fight or flight response. Interestingly, C-terminal fragments of 1C have also been found to serve as transcription factors.
L-type calcium channels, especially CaV1.2 and CaV1.3, have been known to regulate gene expression through Ca2+ influx (13). A decade ago, a new mechanism through which L-type calcium channels regulate gene expression was reported. A distal C-terminal fragment of 1C was found to act as a transcription factor (148).
This 75-kDa C-terminal fragment, termed calcium channel associated transcription regulator (CCAT), is enriched in the nuclei of both HEK 293T cells transfected with
CaV1.2 and rat cortical neurons, especially inhibitory neurons that produce
-aminobutyric acid (GABA) (148). CCAT transfected into neurons, cardiac myocytes, and HEK 293T cells form distinct nuclear punctae, as revealed by confocal imaging
(148). However, the exact N-terminal residue of the CCAT is unknown, and judging from the molecular weight, it comprises virtually the whole C-terminal tail.
The CCAT contains three critical domains: an N-terminal transcription activation domain (residues 1642-1811), nuclear localization domain (residues 1814-1864), and
C-terminal transcription activation domain (residues 2011-2143) (GenBank accession
44 number AAA18905) (148). The N-terminal transcription activation domain modulates transcriptional activation, while the C-terminal transcription activation domain is essential for transcriptional activation (148). The nuclear localization of the CCAT is developmentally regulated. The level of nuclear CCAT increases substantially with age, starting at embryonic day 18 through adulthood (148). The nuclear localization of the CCAT is also regulated by changes in intracellular calcium. Activation of L-type calcium channels or glutamate receptors leads to a decrease in nuclear CCAT, while chelating Ca2+ or inhibiting L-type calcium channels or NMDARs leads to an increase in nuclear CCAT (148). Microarray experiments revealed that 16 mRNAs were significantly upregulated, and 31 genes were significantly downregulated by
CCAT expressed in Neuro2A cells (148). One of the main upregulated targets of
CCAT is gap-junction protein Cx31.1 (148). Chromatin immunoprecipitation (ChIP) analysis revealed that CCAT associates with the promoter of Cx31.1 (148).
Upregulation of Cx31.1 by CCAT also takes place in cultured rat cortical neurons and thalamic neurons, and can be reduced by depolarization of the plasma membrane, in accordance with the finding that increased intracellular [Ca2+] decreases nuclear
CCAT (148).
Connexins play a critical role in forming electrical synapses and facilitating intercellular communication in developing and mature nervous tissues (174).
Upregulation of Cx31.1 by CCAT suggests an important role of CCAT in electrical tissues. Moreover, CCAT was also shown to significantly increase the length and
45 number of neurites of cerebellar granule neurons (148). Taken together, these results suggest that the C-terminal fragment of 1C, as a transcription factor/regulator, plays an important role in the development and function of excitable tissues.
C-terminal fragments of 1C, as transcription factors/regulators, also have been found to downregulate the expression of CACNA1C (138). One study reported that immunoblot analysis of mouse ventricular myocytes with an antibody against the 1C
C-terminal tail detected only a 37-kDa band, while the same analysis of tsA-201 cells expressing CaV1.2 revealed only the 240-kDa full-length 1C (138). These results are consistent with the previous findings that 1C is present in two size forms in animal tissues but in only one size form in heterologous systems (see 1.6.1.1), suggesting that the 37-kDa C-terminal fragment (CTF) might be produced by proteolysis. In mouse ventricular myocytes, the CTF (residues 1821-2171) translocates to the nuclei (138).
Chromatin immunoprecipitation (ChIP) analysis revealed that the CTF interacts with
3 of the 5 putative promoter elements in the CaV1.2 promoter (138). Overexpression
of the CTF led to a 34% decrease of CaV1.2 promoter activity, as revealed by a luciferase reporter assay (138). Interestingly, an N-terminally truncated CTF (residues
1906-2171) is more enriched in the nuclei, has stronger interaction with the CaV1.2
promoter, and, when overexpressed, causes an 80% decrease of CaV1.2 promoter activity (138).
The CTF in mouse ventricular myocytes also antagonizes cellular hypertrophy through transcriptional regulation. Serum, a known inducer of cardiac myocyte
46 cellular hypertrophy in vitro, leads to an increase in the nuclear localization of the
CTF, which then facilitates the suppression of CaV1.2 promoter activity, mRNA level, protein level, and currents (138). Moreover, CTF overexpression attenuates serum-induced cellular hypertrophy and, by suppressing atrial natriuretic factor (ANF) promoter activity, attenuates serum-induced ANF expression.
This 37-kDa CTF has a molecular weight similar to the autoinhibitory CTF in cardiac myocytes mentioned in 1.6.1.1. It is unclear whether the two CTFs are related or are even the same fragment. It is possible that a single proteolytically produced
CTF of 1C serves both as an autoinhibitory domain by interacting with CaV1.2, and as a transcription repressor of CaV1.2 by translocating into the nucleus and interacting
with the promoter of CACNA1C. The suppressive effects of the CTF(s) on CaV1.2
expression and currents suggest a negative feedback loop, in which elevated CaV1.2 protein or current level leads to increased suppressive effects of the CTF. Further studies will be needed to confirm this negative feedback mechanism.
A C-terminal fragment (CTF) of CaV1.2 has also been shown to be present in non-excitable cells and play an important role in cellular differentiation. In Western blot analysis of rat dental pulp stem cells (rDPSCs), an antibody against a middle segment of the 1C recognized a 190-kDa band, while an antibody against the
C-terminal tail recognized a 75-kDa band (175). Interestingly, unlike many other calcium channel CTFs, the 75-kDa CTF in rDPSCs was not observed in the nucleus but only in the cytoplasm (175). Knockdown of 1C abolished the potential of
47 rDPSCs to differentiate into neural cells, and expression of the CTF is sufficient to rescue the differentiation (175). The apparent absence of the CTF in the nuclei suggests that the CTF may regulate cellular differentiation through an indirect way instead of acting as a transcription factor. However, it cannot be excluded that the
CTF acts as a transcription factor, yet its amount in the nucleus is too low to be detected using the imaging method in the study.
It has long been thought that CTFs of L-type calcium channels are products of proteolysis. Several years ago, it was found that a cryptic promoter in exon 46 of
CACNA1C drives the expression of a CTF that acts as a transcription factor (3).
Western blot analysis revealed that 1C expressed in Neuro2A cells is present in two sizes, a 250-kDa full-length form and a 40-kDa CTF, as detected by an antibody against the C-terminal tail (3). A luciferase reporter assay reveals that this CTF has transcriptional activity (3). A nonsense mutation proximal to the CTF does not abolish the production or transcriptional activity of the CTF, indicating that the CTF is not a proteolytically cleaved fragment of full-length 1C (3). Mutation of residue M2011 to an isoleucine causes an 8-kDa decrease in the size of the CTF, indicating that the CTF is the product of an independent translation event that starts at M2011 (3). Deletion of the promoter upstream of the CACNA1C gene abolishes the production of full-length
1C but not the CTF (3), indicating that there is a cryptic promoter in the CACNA1C gene upstream of the region encoding the CTF. Northern blot analysis of Neuro2A cells expressing CaV1.2 revealed two transcripts of 1C, one of 7.5 kb corresponding
48 to the full-length channel, the other of 1.4 kb presumably corresponding to the CTF
(3). Similar results were obtained from Northern blot analysis of rat cerebral cortex, diencephalon, and cerebellum, and further revealed that the two transcripts are regulated reciprocally during development (3). Moreover, single-cell quantitative PCR revealed that in all cells from rat forebrain, mRNAs containing exon 47 are approximately 80-fold more abundant than mRNAs containing exons 2, 8, and 44, and some cells express even only exon 47 (3). Taken together, these results offer convincing evidence that the CTF results from an independent transcript produced by a cryptic promoter in the CACNA1C gene. Luciferase reporter assays revealed that the
238-bp region upstream of M2011 (start site of the CTF) in exon 46 is sufficient to drive the transcription. Exon 46 and exon 47 together offer an even stronger promoter activity, but exon 47 itself has no promoter activity (3).
Mutating M2011 (start site of the CTF) to an isoleucine forces the translation to start from M2073, resulting in a shorter CTF that does not have transcriptional activity (3). Scrutiny of the peptide sequence between M2011 and M2073 revealed an evolutionarily conserved leucine zipper composed of the residues IFIL (3). Mutation of IL to AA reduced transcription by 75% while mutation of all four residues eliminated transcription completely, indicating the leucine zipper is necessary for the transcriptional activity (3).
Rapid amplification of cDNA 5’ ends (5’ RACE) was used to identify transcription start sites (TSSs) in CACNA1C, and revealed multiple TSSs, which will
49 produce proteins of 240-kDa, 110-kDa, and 15-kDa, respectively (3). The presence of these proteins was confirmed by Western blot analysis of mouse brain (3). Conditional transgenic mice lacking exons 14 and 15 of CACNA1C do not produce full-length
1C but still generate a 115-kDa and 15-kDa protein, in accordance with the TSSs identified (3).
This first report of an exonic promoter that drives the transcription of part of the same gene, has greatly expanded our understanding of the L-type calcium channel and the origins and functions of fragment-channels. On the other hand, it also raises more questions. The authors did not explain the discrepancy in the different sizes of the
C-terminal fragments that they observed in CaV1.2-expressing Neuro2A cells (40-kDa) and predicted according to the TSS (15-kDa, consistent with the protein observed in the mouse brain). Is the discrepancy caused by anomalous migration or are there two different CTFs both produced by the cryptic promoter? Are all the other previous reported CTFs of 1C generated by cryptic promoters too or are some of them produced by proteolysis? A few years before the cryptic promoter in exon 46 was reported, the same group of researchers reported a 75-kDa CTF of 1C that acts as a transcription factor (148), as discussed at the beginning of 1.7.1. There is a big difference between the sizes of the 75-kDa CTF and the CTF due to the cryptic promoter (15 kDa), and the two CTFs also have different distributions in the brain and are regulated differently during development (3, 148). Therefore they may be different fragments of 1C and have different origins. Moreover, in cardiac myocytes, the CTF
50 of 1C is present in a 1:1 ratio with the complementary part of the channel and is also associated with the channel (127); this is more in accordance with proteolysis instead of a cryptic promoter. Therefore, although there is convincing evidence that a cryptic promoter is present in CACNA1A and can generate a CTF of 1C, proteolysis may be the origin of many other CTFs.
1.7.2 Splice variants and truncated forms of the P/Q-type calcium channel
1.7.2.1 A variety of fragments of 1A
The P/Q-type calcium channel 1A subunits from brain tissues of different animals have been analyzed using immunoblotting by many labs. Every result showed more than one band (5, 8-12). The sizes of the bands reported in these studies vary in the range of 60-250 kDa. Although these studies used tissues from different animals
(mice, rats, rabbits, and humans), there are some similarities between the results.
Antibodies against the peptide CNA1 (PSSPERAPGREGPYGRE) or CNA3
(SEPQQREHAPPREHV) in the II-III loop always detected two to three bands of
160-250 kDa, and the major band was always in the range of 180-210 kDa, in which no more than two bands were detected (8-10). Antibodies against epitopes near the end of the C-terminus of 1A always detected three bands: 220-250 kDa, 160-170 kDa, and 60-85 kDa (11, 12). Other detection methods or immunoblotting with other antibodies showed bands in different patterns. An antibody against the N-terminal tail and an antibody against a long segment of the II-III loop (including both CNA1 and
51
CNA3) detected four and five bands, respectively, in the range of 95-220 kDa (5, 8).
Moreover, an analysis using a radioactive antagonist of 1A detected four bands ranging between 140 and 220 kDa (176).
In most of these results, the molecular weight of the longest 1A band is 220 kDa, lower than the theoretical molecular weight of full-length 1A, which is 250 kDa in rat (UniProtKB - P54282), 270 kDa in mouse and rabbit (P97445 and P27884), and 280 kDa in human (O00555). One of the studies reported that human 1A had a band of 250 kDa, which is smaller than the theoretical molecular weight of 280kDa
(12). This discrepancy is probably due to anomalous migration, or an extremely low amount of full-length 1A in the tissue.
Where do the other bands originate from? Are they present in the brains of live animals or produced in vitro? We cannot exclude the possibility that some of those bands are due to protein degradation during sample preparation or non-specific staining by the antibodies. However, plenty of evidence is in favor of the presence of shorter forms of 1A in the brains of live animals. Some of the studies reported that the banding pattern is highly reproducible in different preparations (5, 176). Inclusion of high concentrations of several protease inhibitors and handling of samples at 0°C did not alter the relative quantities of the bands (5, 8, 9). In one of the studies, 1A was detected as multiple bands, while 1B in the same preparation was not (5). These results all argue against in vitro degradation. Moreover, immunoprecipitation analysis showed that a 190-kDa band, the major band in rat brain, can be recognized by
52 multiple antibodies against different parts of 1A (8, 9). A 60-kDa band recognized by an antibody against the 1A C-terminal region was shown to contain the histidine tract (11), an evolutionarily conserved sequence in the 1A C-terminal tail. A 95-kDa
1A fragment is present in preparations from wild-type rabbit and mice, but not in
CACNA1A-knockout mice (5, 6). Therefore, these bands are unlikely due to non-specific staining.
As mentioned in 1.6.1.1, 1S, and1C expressed in heterologous systems are only present in the full-length form, but those from animal tissues are usually present in two size forms. Interestingly, 1A expressed in heterologous systems also exhibits a much simpler banding pattern than in animal tissues. Immunoblot analysis of 1A expressed in HEK 293, HEK 293Tm and PC12 cells with antibodies against the
C-terminal region or tags on the C-terminus always showed only two bands, a long form of 220-280 kDa (presumably full-length 1A), and a short form of 75 kDa (11,
12, 177). In one case, an extra 55-kDa band is present on the Western blot but was not discussed (7). Compared with 1A in animal tissues, 1A expressed in cell lines lacks the middle-sized form of 160-170 kDa recognized by antibodies against the
C-terminal region. Moreover, when an 1A construct with an HA-tag on repeat IV was expressed in HEK 293 cells and analyzed by immunoblot with an anti-HA antibody, even the 75-kDa band was absent and only a 250-kDa band was detected
(178). The 75-kDa form was revealed to be a C-terminal fragment of 1A produced by an internal ribosome entry site (IRES) and acts as a transcription factor with
53 important regulatory functions (see 1.7.2.3). This explains why the anti-HA antibody cannot detect it. These results indicate that live animal brains have certain mechanisms, which cell lines do not have, to produce short forms of 1A. These short forms of 1A may play crucial roles in the development and function of the brain.
What mechanisms cause so many different sizes of 1A? The two potential mechanisms mentioned most often in the literature are post-translational proteolysis and alternative splicing (5, 8, 9, 176). As mentioned in 1.6.1.2, 1C was shown to undergo mid-channel proteolysis resulting in two complementary fragments of 90 kDa and 150 kDa, respectively (4). A similar proteolytic cleavage may happen to 1A and produce a proximal fragment of ~90 kDa and a distal fragment of 160~190 kDa, depending on the species. Consistent with this possibility, 1A bands of these sizes were detected by pertinent antibodies in many of the previous studies (see above).
Alternative splicing is another potential mechanism since a190-kDa short form of
1A was shown to contain the N-terminal tail, C-terminal tail and the II-III loop.
These results indicate that this short form of 1A is not a product of proteolysis, but may instead be a splice variant (8). In another study, a 95-kDa short form of 1A was found to contain the N-terminal tail, repeat I, I-II loop, repeat II and part of the II-III loop (5). The authors surmised that this 95-kDa species is a product of alternative splicing, in consideration of the presence of two transcripts in Northern blot analysis of rat brain 1A (179). Moreover, it has been suggested that the four-domain sodium and calcium channels originated during evolution through two rounds of gene
54 duplication of a single domain ancestral channel (180, 181). The expression of the two-domain isoforms still occurs through alternative splicing, especially during embryonic development (180, 182). Thus, it is tempting to speculate that the 95-kDa two-domain form of 1A is also has evolutionary origins due to alternative splicing.
However, among the numerous splice variants that have been discovered, none can account for an isoform as small as 190 kDa (see 1.7.2.5). The possibility of post-translational proteolysis cannot be ruled out. Although the origin of the 95-kDa two-domain fragment of 1A is unknown, progress has been made on revealing its physiological properties, which is summarized and discussed in 1.7.2.2.
As will be discussed 1.7.2.2 and 1.7.2.3, the physiological properties of two short forms of 1A have been studied. However, as shown in the above-mentioned immunoblot analyses, numerous other short forms of 1A are present in the brain, and their physiological properties are still unknown. It has been revealed that the different size forms of 1A are differentially phosphorylated. A 220-kDa form of 1A from the rat brain is phosphorylated preferentially by PKA in vitro, while PKC and PKG preferentially phosphorylate a 190-kDa form in vitro (8). This suggests that different sizes of 1A may be differentially regulated in vivo to fulfill the complex regulation and function of the P/Q-type calcium channels. Further research is needed to understand the identities, physiological properties, and functions of the different size forms of 1A.
55
1.7.2.2 A two-domain form of 1A with inhibitory function
Immunoblot analysis of rabbit brain membrane protein with an antibody against part of the II-III loop revealed five different size forms of 1A, among which the shortest has a molecular weight of 95 kDa (5). This 95-kDa species is enriched in the cerebellum and cerebral cortex, with lower expression in the hippocampus and thalamus (6). This short form of 1A is unlikely to result from degradation during sample preparation for a variety of reasons. Firstly, 1B, which shares significant sequence homology with 1A, is present only as a 210-kDa form in the same preparation. Secondly, two preparations purified using different methods exhibited similar banding pattern, both having the 95-kDa band as a dominant species. Thirdly, inclusion or omission of several protease inhibitors did not affect the relative quantities of the different size forms (5). A 95-kDa short form of 1A is also present in the mouse brain, but is absent in the brain of CACNA1A-knockout mouse, signifying that the 95-kDa species and full-length 1A are different products of the same gene (6).
Microsequencing of the 95-kDa species yielded 13 peptide sequences, which are all contained in the first half of 1A up to amino acid 829 (5). The authors surmised that the actual C-terminal end of this 95-kDa protein is approximately residue 969, since the epitope of the antibody that recognized this protein corresponds to residue
779-969. This would include the region corresponding to CNA1 and CNA3. However, among the several size forms of 1A recognized by antibodies against CNA1 and
56
CNA3, the 95-kDa species is not present (8-10). This finding indicates that the
95-kDa species probably ends before CNA1, which begins at residue 887 in rabbit
1A (UniProtKB - P27884). Therefore, the 95-kDa species includes about 1/3-1/2 of the synprint site in the II-III loop.
It is tempting to speculate that the 95-kDa two-domain form of 1A is an evolutionary throwback due to alternative splicing, resembling the two-domain splice variants of 1S, 1C, and the voltage-gated sodium channel SCN8A (180, 182).
However, among the numerous splice variants of 1A that have been discovered, none can account for an isoform as small as 95 kDa (see 1.7.2.5). Another possible origin of the 95-kDa form of 1A is post-translational proteolysis. The L-type calcium channel 1C subunit was shown to undergo mid-channel proteolysis in the II-III loop before the CNA1 sequence (4). The 95-kDa species may be one of the two fragments produced by proteolytic cleavage at the equivalent site in 1A. The complementary fragment, of 175 kDa in theory, may account for the 160-190-kDa species recognized by antibodies against CNA1, CNA3 and the C-terminal region in other studies (see
1.7.2.1).
Deglycosylation studies showed that the 95-kDa form of 1A from the rabbit brain is glycosylated, presumably at N283 (5). This is the first evidence that a VGCC
1 subunit is glycosylated. Like most glycoproteins, the 95-kDa form of 1A was shown to be expressed on the plasma membrane (6, 152). Co-immunoprecipitation analysis showed that the 95-kDa form of 1A associates with the subunit, and its
57 association with the subunit depends on a critical amino acid residue, Y392, in the alpha-interaction domain (AID) (6). Trafficking of the 95-kDa form to the plasma membrane relies on the association with the subunit (6). The 95-kDa form of 1A does not conduct currents, but it can generate a high density of gating currents, upon membrane depolarization (152). This indicates that this short form of 1A can undergo conformational changes in response to membrane depolarization. Because the 95-kDa form of 1A contains at least part of the synprint site, the authors (6) surmised that it may be able to interact with the synaptic exocytotic machinery and couple depolarization to neurotransmitter release in a Ca2+-independent manner (183,
184).
Moreover, the 95-kDa form of 1A has a potent inhibitory effect on the current amplitude of full-length 1A, and the inhibitory effect can be abolished by mutating
Y392 to serine in the 95-kDa form, signifying that the mechanism of inhibition is sequestration of free subunits (6). Inhibition of full-length channel currents by truncated 1 subunits is a common phenomenon among cation channels. It has been found on L-type (185, 186), N-type (187-190), P/Q-type calcium channels (see 1.7.2.4) as well as potassium channel KvLQT1 (191). Note that many of the truncated 1 subunits studied are not naturally present in normal wild-type animals. The presence of the inhibitory 95-kDa 1A fragment in normal brains suggests that this fragment may play a critical role in regulating the activity of P/Q-type calcium channels to ensure normal development and function of the nervous system. In order to further
58 reveal the functions and importance of the 95-kDa form of 1A, future research should focus on how the expression and function of the 95-kDa fragment change in response to different physiological conditions. Moreover, it would be interesting to find a way to inhibit the expression of the 95-kDa form of 1A in animals and see how the phenotype changes. Such studies may shed light on a critical regulatory mechanism of the nervous system.
1.7.2.3 A C-terminal fragment of 1A acting as a transcription factor
As mentioned in 1.7.2.1, immunoblot analysis of mouse and human brain tissues with antibodies against the C-terminal region of 1A detected a 60-85-kDa short form of 1A (11, 12), which has also been detected as a 75-kDa species in cell lines expressing 1A (7, 11, 12, 177). This short form, (hereafter assigned the molecular weight of 75 kDa), of 1A was revealed to be a C-terminal fragment (CTF) of 1A produced by an internal ribosome entry site (IRES). It acts as a transcription factor with important regulatory functions in normal cells and mediates the polyglutamine-induced cellular toxicity in spinocerebellar ataxia type 6 (SCA6).
The 75-kDa 1A CTF was originally thought to be a product of proteolysis (11,
12, 177). It can be produced even in HEK 293T cells transfected with only the
C-terminal tail of 1A (177). Thus it was reasonable to think that there is a proteolytic cleavage site in the 1A C-terminal tail. However, the latest studies have shown that this is not the case.
59
Liquid chromatography-tandem mass spectrometry (LC-MS/MS) analysis of the
75-kDa 1A CTF produced in HEK 293 cells expressing human 1A revealed that the sequence of the N-terminus of 1A CTF is MIMEYYR (7). The initial methionine residue implies that the 1A CTF might be the direct product of a translation event, but the most convincing evidence that argues against proteolysis is that stop codons inserted upstream of the start site of 1A CTF did not abolish the expression of the
CTF (7). Nevertheless, the expression of the 1A CTF still relies on part of the upstream sequence (7). To test the IRES activity of the upstream sequence, the ~1-kb sequence upstream of the start site of 1A CTF in the 1A cDNA was inserted into a bicistronic reporter vector between the two cistrons (7). As a result, the expression of the second cistron was drastically enhanced, while the mRNA level of neither cistron changed (7). In contrast, insertion of the same sequence into a promoterless reporter vector does not enhance gene expression (7). These results indicate that the upstream sequence does not contain a cryptic promoter, nor does it increase mRNA stability.
Moreover, real-time PCR showed that the ratio of the expression levels of RNA upstream and downstream of the start site of 1A CTF is approximately 1:1 in
HEK293 and PC-12 cells transfected with human 1A and in human cerebellum.
These results argue against the presence of a cryptic promoter or alternative splicing
(7), and offer convincing evidence that the 75-kDa 1A CTF is produced by an IRES.
Further experiments revealed that the 75-kDa 1A CTF serves as a transcription factor (7). This is consistent with previous findings that the CTF is present in the
60 nuclei of mouse cerebellar cells, 1A-transfected HEK 293 cells, and probably human cerebellar cells (11). However, there were also contradictory results showing that the
1A CTF is present in the nuclei in SCA6 cerebellum but not in healthy cerebellum
(12). The 1A CTF was shown to interact with the DNA sequences and enhance the expression of genes such as BTG1, PMCA2, and GRN, all of which are involved in neural development and function (7). The 1A CTF was also shown to potentiate
NGF-mediated neurite outgrowth in PC12 cells and partially rescue the defective phenotype of CACNA1A knockout mice (7).
The 75-kDa 1A CTF not only contributes to normal neural development and function, but also is related to the polyglutamine-induced cellular toxicity in
Spinocerebellar ataxia type 6 (SCA6). SCA6 is an autosomal dominant neurodegenerative disease characterized by cerebellar atrophy especially in the superior vermis, with more severe loss of Purkinje than granule cells (14). SCA6 is caused by an expansion of the trinucleotide CAG repeats in exon 47 of the CACNA1A gene, which encodes the polyglutamine (polyQ) tract near the C-terminal end of 1A
(192). SCA6 alleles usually have 21-30 CAG repeats, as opposed to generally over 40 in other polyQ diseases, such Huntington’s disease, while normal alleles have about
13 repeats (192). Although some studies in heterologous expression systems reported
that expanded polyQ can cause functional alteration of CaV2.1, two studies on
SCA6-knockin mice found no effect of expanded polyQ on the electrophysiological
properties of CaV2.1 (14). These findings suggest that the main pathogenic
61 mechanism of SCA6 is not due to alterations of biophysical properties or membrane
expression of CaV2.1.
An increasing number of studies suggest that the main pathogenic mechanism of
SCA6 is due to cellular toxicity of the SCA6 1A CTF. Many studies showed that the
75-kDa 1A CTF with an expanded polyQ tract is toxic to cultured cells (7, 11, 177,
193). In addition, transgenic mice expressing SCA6 1A CTF exhibit ataxia and cerebellar atrophy resembling the symptoms of SCA6 in humans (7). The mechanisms of the cellular toxicity have been intensively studied by not yet fully understood. Two independent studies have shown that overexpressing the polyQ tract lacking flanking sequences does not cause toxicity, while overexpressing a longer fragment including more of the flanking regions can cause significant toxicity to cultured HEK 293T,
HEK 293, and mouse cerebellar granule cells, regardless of the length of polyQ (4 to
33 glutamine residues) (11, 177). These results indicate that the protein context of the polyQ tract contributes to the pathogenic mechanism of SCA6. One of the two above-mentioned studies showed that expanded polyQ tract does not add to the toxicity of the 1A CTF, but can make the CTF more resistant to proteolysis (177). In contrast, the other study showed that 1A CTF with expanded polyQ is more toxic than those with shorter lengths of polyQ, and showed no difference in the stabilities of the 1A CTFs (11). A possible reason for this discrepancy is that auxiliary subunits were coexpressed in the second study but not in the first one. Nevertheless, these results revealed that both the expanded polyQ and the flanking regions are
62 indispensable to the toxicity of SCA6 1A CTF.
It has also been shown that expanded polyQ increases nuclear localization of the
1A CTF in both SCA6 human cerebella and in transfected PC12 cells (12), indicating that nuclear localization may be important for the toxicity of SCA6 1A
CTF. Protein sequence analysis revealed four putative nuclear localization signals
(NLSs) upstream of the polyQ tract, and confocal imaging showed that three of them were shown to be critical for the nuclear localization of 1A CTF (11). Triple mutation of the three NLSs significantly decreased the toxicity of the 1A CTF with expanded polyQ. An additional artificial nuclear export signal (NEL) decreased the toxicity even more (11), indicating that the toxicity depends on nuclear localization of the CTF. However, nuclear localization is not the only reason flanking sequences are indispensible to the toxicity, since a polyQ tract without flanking regions also exhibits strong nuclear localization, but is not toxic to HEK 293 or mouse granule cells (11).
Another possible contributing factor in the toxicity of the SCA6 1A CTF may be its low solubility. In SCA6 human cerebella, more cytoplasmic 1A CTFs are insoluble compared with normal human cerebella, but extracerebellar tissues did not show this difference (12). Immunocytochemistry revealed that CTFs with expanded polyQ form visible aggregates in the cytoplasm and nuclei of transfected PC12 cells
(12). Similar aggregates were also observed in the cell bodies and dendrites of SCA6 cerebellar Purkinje cells but not in normal brain tissues or extracerebellar tissue of
SCA6 brains (12). These results suggest that SCA6 pathogenesis may be partially due
63 to the aggregate formation, which may cause cellular toxicity by sequestration of essential proteins. It has been shown that the whole CTF is less soluble than a more truncated protein with almost only the polyQ tract (177). This indicates that the low solubility of SCA6 1A CTF and the aggregation formation probably rely on both the expand polyQ tract and the flanking regions. This offers another explanation why the flanking regions are important for the toxicity of SCA6 1A CTF.
The expanded polyQ in SCA6 has also been shown to cause disruption of normal
1A CTF function. Normal 1A CTF serves as a transcription factor to enhance the expression of a group of genes, which are part of a Purkinje cell-specific developmental program (7). It has been shown that the expanded polyQ tract abolishes the normal function of the 75-kDa 1A CTF in regulating the expression of these genes and potentiating neurite growth (7). The pathogenic mechanism of SCA6 is at least partially due to the loss of normal 1A CTF function. On the other hand, a gain of function may also contribute to this pathogenic mechanism. As mentioned above, nuclear localization of the 1A CTF is increased by the expanded polyQ (12), whose toxicity is in turn dependent on the nuclear localization of the CTF (11). These results imply a possibility that the SCA6 1A CTF may function as an aberrant transcription factor and cause cellular toxicity through activation of aberrant gene expression. Moreover, coimmunoprecipitation and pull-down analyses showed that
1A C-termini can interact directly with myosin IIB in rat cerebellum and COS-7 cells, and expanded polyQ can increase this interaction (193). The study also showed
64 that the expanded polyQ increases the expression of a CD4-linked 1A CTF on the plasma membrane. Additionally, this effect of the expanded polyQ can be reversed by a myosin II inhibitor (193). These results suggest that the enhanced interaction with myosin IIB and increase in surface expression may contribute to SCA6 pathogenesis.
Alternatively, the pathogenesis may be caused by myosin sequestration by mutant
1A subunits or 1A CTFs. However, the effects of expanded polyQ on the surface expression of 1A CTF and on the interaction with myosin II were only studied in
COS-7 cells, and the 1A CTF used in this study not only lacks most of the region upstream of the polyQ tract but also is tagged with CD4 and GFP (193). However, further investigation is needed to determine if similar effects also happen in vivo.
In summary, SCA6 has complex pathogenic mechanisms. The 1A CTF with expanded polyQ causes cellular toxicity, which relies on both the expanded polyQ and the flanking regions. The flanking regions contribute to the cellular toxicity by offering nuclear localization signals (NLSs) and by decreasing the solubility of the
1A CTF to facilitate the formation of aggregates, which may sequestrate essential proteins and thus cause cell death. The expanded polyQ increases nuclear localization and is also necessary for the formation of aggregates. Expanded polyQ also abolishes the normal function of the 1A CTF as a transcription factor, and subsequently disrupts the normal development and function of cerebellar cells. The SCA6 1A CTF may also cause cellular toxicity through gain of function, e.g., aberrant gene activation or increased binding to regulatory proteins. More studies are necessary to
65 further understand the function of the 1A CTF and the pathogenesis of SCA6.
1.7.2.4 1A truncations causing episodic ataxia type 2
Mutations in CACNA1A cause several autosomal dominant neurological disorders, including familial hemiplegic migraine type 1 (FHM1), spinocerebellar ataxia type 6 (SCA6) and episodic ataxia type 2 (EA2) (14). While FHM1 and SCA6 reported so far are caused by missense mutations and expansion of the trinucleotide
CAG repeats, respectively (14), EA2 can be caused by a variety of mutations including missense, nonsense, deletion, and insertion mutations (15). Most EA2 mutations reported so far result in premature stop codons leading to the truncation of
1A, which may arise through nonsense mutations, frameshift caused by deletion or insertion mutations, and aberrant splicing (15). A number of studies have investigated the properties of some of the EA2 truncation mutants and engineered truncation mutants (Figure 2). The results of these studies are summarized in Table 1.
As shown in Table 1, the truncation mutants whose currents were measured all
exhibited smaller currents than full-length CaV2.1 or no currents at all. Even the removal of a segment as small as the C-terminal tail can abolish the current-conducting function of the channel, as in R1824X. Most of the investigated truncation mutants also showed less surface expression than full-length 1A.
Strikingly, ten of the twelve truncation mutants were shown to have a dominant
negative effect of suppressing full-length CaV2.1 currents. Given the essential role of
66 the P/Q-type calcium channel in synaptic transmission in the central and peripheral nervous systems (1, 17), the dominant negative effect might be a major pathogenic mechanism of these EA2 truncation mutants.
a b g l m n k f q e
j r c p d o
h i
Figure 2. Membrane topology of 1A, showing positions of truncation mutations, including EA2 and non-EA2 mutations that have been studied, as well as unstudied EA2 mutations that resemble other mutations that have been studied. (a*) V327 aberrant splicing (194). (b**) 330X (195). (c**) 384X (195). (d**) 402X (195). (e*) F624fs (196, 197). (f**) 612X (195). (g) Q681fs (196, 198, 199). (h*) S943fs (196, 200). (i*) E1004fs (196). (j**) P1219X (rabbit) (6, 152). (k*) D1268fs (196, 201). (l**) P1217fs (rat) (188-190). (m) R1281X (178, 196, 199, 202-205). (n) Y1446X (195, 196, 198). (o) R1549X (196, 199, 202, 203, 206). (p*) Q1564X (207). (q) R1669X (199, 203). (r) R1824X (196, 208, 209). Unstudied EA2 mutations are indicated by *. Non-EA2 mutations are indicated by **.
67
Table 1. Properties of EA2 truncation mutants and engineered truncation mutants
Mutation Conducting currents Less surface expression Reducing full-length Altering full-length CaV2.1 activation Reducing full-length Reducing full-length Associated with Associated with
than full-length 1A CaV2.1 current and inactivation parameters4 1A protein 1A surface subunits full-length 1A
expression expression
A330X1,8 No (195) No (195) No5 (195)
Q384X1,8 No (195) No (195) Slightly5 (195) No (195) No (195)
V402X1,8 No (195) Yes (195) No5 (195) No (195) Yes (195)
K612X1,8 No (195) Yes (195) Slightly5 (195) No (195)
Q681fs Yes (199) Yes (199) No (199) Yes (199)
P1219X1,6 No (152) No (6) Yes (6) Yes (6)
P1217fs1,7 Yes (188, 190)
R1281X Yes2 (202); No (203, Yes (199) Yes (178, 199, 203); No (203) Yes (178); No (199) Yes (178, 199) No (178) Yes (178)
204) No3 (204)
Y1446X No (195) Yes (195) Slightly5 (195) No (195)
R1549X Yes2 (202); No (203) Yes (199) Yes (203) No (203) Yes (199)
R1669X No (203) Yes (199) Yes (203) No (203) Yes (199)
R1824X No (209) Yes (209); No3 (208) No (209)
1: Not an EA2 mutation. 2: The currents are much smaller than those conducted by full-length
CaV2.1. 3: The short isoform of human 1A (AF004883) was used. 4: Voltage dependence and kinetics of activation and steady-state inactivation. 5: Lacking data for voltage dependence of steady-state inactivation. 6: Construct was made from rabbit X57477. The equivalent mutation in human AF004883 would be R1213X. 7: Rat 1A. The equivalent mutation in human AF004883 would be P1269fs. 8: Constructs were made from rabbit X57476. Note: Residue numbers are from AF004884 unless otherwise indicated.
Interestingly, one of the truncation mutants, 330X, which resembles an EA2 truncation mutant caused by aberrant splicing (V327 aberrant splicing) (194), did not show a dominant negative effect, nor does it conduct currents. V327 aberrant splicing is the shortest EA2 truncation mutant reported so far. However, the loss of function of this mutant cannot explain why this mutant causes an autosomal dominant disorder, since heterozygous CACNA1A-knockout mice showed no apparent abnormality (210),
68 arguing against haploinsufficiency as the pathogenic mechanism. The V327 aberrant splicing mutant may cause EA2 by competing with wild-type channels for the limited number of presynaptic slots for P/Q-type channels (211). Alternatively, it may alter
the voltage dependence of inactivation of full-length CaV2.1, change the way
full-length CaV2.1 is regulated by signaling molecules, or associate with and alter the function of other essential proteins in the cell.
Contradictory results were obtained as to whether the EA2 truncation mutant
R1281X alters full-length CaV2.1 currents (Table 1). The discrepancy is probably
partially due to the different isoforms of CaV2.1 used in the experiments. One of the studies used the short splice isoform of 1A (GenBank accession number AF004883), which lacks the last 244 amino acid residues encoded by exon 47, and revealed no dominant negative effect of R1281X (204). Another study used the long splice isoform (GenBank accession number AF004884) which included the residues encoded by exon 47, revealing a significant dominant negative effect of R1281X (203).
Comparison of the long and short isoforms in parallel showed that R1281X has a much stronger dominant negative effect on the long isoform of 1A than on the short one (203). Interestingly, one group found that another EA2 truncation mutant,
R1824X, also displayed potent dominant negative effect in one study, which involved the long isoform of 1A (209), but not in another study, which used the short isoform
(208). These results suggest that the endmost segment of the C-terminal tail contains regulatory elements that enhance the suppressive effect of truncated 1A on the
69 full-length channel. However, the isoform difference is not the only reason for the contradictory results on the dominant negative effect of EA2 truncation mutants.
While the dominant negative effect of R1281X on the short isoform of 1A could not be detected in Xenopus oocytes (204), another study that used HEK 293 cell as the expression system detected the dominant negative effect (178). It has been shown that
R1281X, as a misfolded protein, binds to full-length 1A and induces its degradation, through a mechanism involving the endoplasmic reticulum (ER) retention machinery and the proteasome (178). It has been reported that Xenopus oocytes have a weak ER retention machinery for mammalian proteins (212), which may be another contributing factor to the lack of dominant negative effect of R1281X on the short isoform of 1A in oocytes.
While contradictory results have been reported as to whether EA2 truncation mutants decrease the protein level of full-length 1A, all the investigated truncation mutants were shown to decrease surface expression of the full-length channel (Table
1). The decrease in surface expression may be one of the main reasons for the decrease in full-length channel currents caused by the truncation mutants. However, it cannot be excluded that the some truncation mutants may also suppress full-length
CaV2.1 currents by altering the single-channel properties of full-length CaV2.1.
The molecular mechanisms for the suppressive effect of EA2 truncation mutants
on full-length CaV2.1 have been intensively studied and are still not fully understood.
It has been shown that only those 1A truncation mutants that include the
70
-interaction domain (AID) suppress full-length CaV2.1 currents (195) (Table 1), signifying that the mechanism of the suppressive effect may involve the subunit.
Coimmunoprecipitation analysis confirmed that 402X, a truncation mutant with AID, associates with the subunit, while 384X, which does not have the AID, does not associate with (195). Another 1A truncation mutant P1219X, which resembles the
EA2 truncation mutants S943fs (200) and E1004fs (196), was also shown by coimmunoprecipitation to associate with the subunit (6). Moreover, a single mutation (Y392S) in the AID of P1219X not only disrupted its interaction with , but also abolished its suppressive effect on the currents of full-length CaV2.1 (6). These results indicate that the dominant negative effects of some EA2 truncation mutants, including S943fs and E1004fs, may be partially due to competition with the full-length 1A for the subunits.
Some other EA2 truncation mutants, however, may exert their dominant negative effects through -independent mechanisms. When expressed in Xenopus oocytes, the
EA2 truncation mutant R1281X displays a dominant negative effect, which cannot be reversed when the amount of cRNA coding for and 2 subunits was increased
(203). Coimmunoprecipitation analysis revealed that R1281X does not associate with
subunits (178). Instead, it associates with full-length 1A and disables the binding of subunits to full-length 1A (178). Moreover, it has been shown that R1281X, like a lot of misfolded proteins, is retained in the ER and degraded by the proteasome
(178), through a mechanism called endoplasmic reticulum-associated degradation
71
(ERAD) (94). Additionally, by interacting with full-length 1A, R1281X also promotes the ER retention and premature proteasomal degradation of the full-length
1A, exhibiting a “destructive interaction mechanism” (DIM) (178). As expected for a mechanism involving ER retention, the dominant negative effects of R1281X and some other EA2 mutants show dependence on the experiment system and temperature.
R1281X displayed a potent dominant negative effect on the short isoform of 1A
(AF004883) in HEK 293 cells (178), but not in Xenopus oocytes (204), in which the
ER retention machinery is weak (212). Moreover, the dominant negative effects of
EA2 mutants Q681fs, R1281X ,and E1761K (a missense mutation) in HEK 293T cells all can be abolished by a lower culture temperature (199), which have been shown to overcome ER retention of misfolded proteins (212, 213). Therefore, the
“destructive interaction mechanism” (DIM) may play an important role in the pathogenesis of R1281X and many other EA2 truncation mutants.
Truncated CaV2 and CaV3 often suppress full-length CaV2 and CaV3, respectively
(187-189). CaV2.2-Dom I-II, a truncated 1B containing the segment from the
N-terminus up until the end of the II-III loop, displayed a dominant negative effect
(187). CaV2.2-Dom I-II does not conduct currents, but suppresses the protein
expression, surface expression, and currents of full-length CaV2.2, while not changing the I-V parameters, steady-state inactivation parameters and single-channel properties
of full-length CaV2.2 (187). CaV2.2-Dom I, containing the segment from the
N-terminus up until the end of the I-II loop, also showed a similar suppressive effect
72 on the expression and currents of full-length CaV2.2 (187). Similar truncated forms of
CaV3.1 and Cav3.2 1subunits suppress the currents of full-length CaV3.1 and Cav3.2, respectively (188). Interestingly, there is marked cross-suppression between different
types of channels within the CaV2 family (188). For example, CaV2.1-P1217fs, a truncated 1A resembling the EA2 truncation mutant D1268fs (201), suppresses not only currents of full-length CaV2.1, but also currents of full-length CaV2.2. In addition,
CaV2.2-Dom I-II can also suppress currents of full-length CaV2.1 (188). However,
there is no cross-suppression between CaV2 and CaV3 channels (188). These results
indicate that these truncated CaV2.1 and CaV2.2 channels probably share the same mechanism for their dominant negative effects.
CaV2.2-Dom I-II displays a suppressive effect on full-length CaV2.2 currents with or without and 2 subunits, and transfection of a higher amount of cDNA of the subunit does not reduce the dominant negative effect of CaV2.2-Dom I-II (187).
Moreover, the 1B I-II loop itself, which was shown to interact with the subunit by yeast two-hybrid analysis, does not suppress full-length CaV2.2 currents (187). These
results suggest that the dominant negative effect of CaV2.2-Dom I-II is not due to sequestration of subunits.
Coimmunoprecipitation analysis showed that CaV2.2-Dom I-II associates with full-length 1B (190). Could CaV2.2-Dom I-II exert its dominant negative through a destructive interaction mechanism (DIM), as previously discovered on the EA2 truncation mutant 1A R1281X (178)? In contrast to DIM, which involves the
73 proteasome, the dominant negative effect of truncated 1B does not seem to rely on the proteasome, since the proteasome inhibitor lactacystin has no effect on CaV2.2
protein level in the presence of CaV2.2-Dom I (188).
CaV2.2-Dom I not only reduces the protein expression and current amplitude of
full-length CaV2.2 (187), but also reduces the mRNA level of full-length CaV2.2 and the protein level of coexpressed 2 (189). Surprisingly, CaV2.2-Dom I also reduces the mRNA level of a mutant inward-rectifier potassium ion channel Kir2.1-AAA in
the presence of, but not in the absence of, full-length CaV2.2 (189). These results are consistent with the unfolded protein response (UPR), which is initiated when the ER is unable to deal with the load of newly synthesized proteins requiring folding (214).
In the UPR, the endoplasmic reticulum resident RNA-dependent protein kinase
(PERK) is activated and subsequently phosphorylates the ribosomal translation initiation factor eIF2 (214). Phosphorylation of eIF2 causes dissociation of ribosomes from the mRNA, leading to translation arrest (214) and mRNA destabilization (215). It has been shown that dominant negative PERK markedly
reduced the suppressive effect of CaV2.2-Dom I on full-length CaV2.2 currents, while
having no effect on full-length CaV2.2 currents in the absence of CaV2.2-Dom I (188).
On the other hand, overexpression of wild-type PERK or treatment with the PERK
activator thapsigargin reduces full-length CaV2.2 currents in the absence of truncated
CaV2.2 (188). Moreover, dominant negative PERK also markedly reduced the
suppressive effect of CaV2.1-P1217fs on full-length CaV2.1 currents, while having no
74 effect on full-length CaV2.1 currents in the absence of CaV2.1-P1217fs (188). These findings suggest that the UPR is also involved in the dominant negative effects of this truncated 1A and presumably some EA2 truncation mutants including D1268fs, which resembles CaV2.1-P1217fs.
The N-terminal tails of CaV2.1 and CaV2.2 contain elements critical for the dominant negative effects of truncated CaV2 1 subunits. Two arginine residues in a
RAR motif in the N-terminal tail of CaV2 1 subunits was previously shown to be essential for G protein-mediated inhibition of CaV2 channels (216, 217). Recently, they were also shown to be critical for the dominant negative effect of truncated 1A and 1B (189, 190). Mutation of the two arginine residues in CaV2.1-P1217fs (R57 and R59) and CaV2.2-Dom I-II (R52 and R54) to alanine reverses their suppressive
effects on the currents of full-length CaV2.1 and surface expression of CaV2.2, respectively (190). Mutating the two critical arginine residues in the full-length channels, however, does not weaken the suppressive effects (190).
Surprisingly, the dominant negative effects of truncated CaV2 channels can be alleviated by partial N-terminal tails of the cognate 1 subunits. In one study, a partial
CaV2.1 N-terminal tail containing residues 46-100, and a partial CaV2.2 N-terminal tail containing residues 43-95, were engineered to have a CAAX motif attached to their C-terminus, in order to promote their association with intracellular lipid bilayers
(190). These partial N-terminal constructs, CaV2.1-(46-100)-CAAX and
CaV2.2-(43-95)-CAAX, do not influence the currents or surface expression of
75 full-length CaV2 (190). However, CaV2.2-(43-95)-CAAX almost completely reverses
the suppressive effect of CaV2.2-Dom I-II on CaV2.2 surface expression, and
CaV2.1-(46-100)-CAAX partially reverses the suppressive effect CaV2.1-P1217fs on
full-length CaV2.1 currents (190). Coimmunoprecipitation analysis showed that
CaV2.2-(43-95)-CAAX disrupts the association between CaV2.2-Dom I-II and
full-length CaV2.2 (190). This is most likely the main mechanism by which the partial
N-terminal tails alleviates the suppressive effects of truncated channels. Interestingly, these partial N-terminal tails cannot reverse the suppressive effects of the truncated channels when the two critical arginine residues are mutated in the full-length channel
CaV2.1. One reason for this observation is that these mutations cause a more robust interaction between the full-length channel and the truncated channel (190). A model has been proposed to explain the dominant negative effects of truncated CaV2 1 subunits and how the partial N-terminal tails alleviates the dominant negative effects
(190) (Figure 3).
In summary, most EA2 truncation mutants of 1A display dominant negative effects of suppressing the surface expression and currents of full-length CaV2.1, which may be the main pathogenic mechanism of EA2. Different EA2 truncation mutants may exert their dominant negative effects through different molecular mechanisms.
Three molecular mechanisms have been proposed. Firstly, some EA2 truncation
mutants may suppress the expression and currents of full-length CaV2.1 by competing with full-length 1A for subunits. Secondly, some EA2 truncation mutants may
76 promote the degradation of full-length 1A though a “destructive interaction mechanism” (DIM) involving ER-associated degradation (ERAD). By interacting with full-length 1A, these EA2 truncation mutants induce ER retention and premature proteasomal degradation of the full-length 1A. Thirdly, some EA2 mutations may suppress the synthesis of full-length 1A by associating with the full-length 1A and initiating the unfolded protein response (UPR). During the UPR, the ER resident RNA-dependent protein kinase (PERK) is activated and subsequently leads to translation arrest and mRNA destabilization of full-length 1A and other proteins.
As mentioned in 1.7.2.1, many 1A subunits are present in short forms in the brain tissues of wild-type animals and healthy human beings. One of these short forms of 1A, the 95-kDa two-domain form of 1A mentioned in 1.7.2.2, is one of the major forms of 1A in rabbit and mouse brains. It is presumably a truncated 1A whose peptide chain ends in the II-III loop (5, 6), resembling EA2 truncation mutations such as 943fs (200) and 1004fs (196). Other short forms of 1A may also resemble EA2 truncation mutants. Why do these short forms of 1A not cause any disorder like the EA2 truncation mutations do? Firstly, the short forms and full-length forms of 1A may belong to different isoforms of 1A, which serves different functions and do not interfere with each other. Secondly, the slightly different truncation positions may lead to substantial difference between “normal” truncated forms of 1A and the EA2 truncation mutants. Note that in the studies of the
77 properties of the 95-kDa two-domain form of 1A, the truncated construct P1219X was used (6, 152). The endogenous 95-kDa fragment, however, is presumably much shorter (see 1.7.2.2). Therefore, it is possible that the endogenous 95-kDa fragment does not have the dominant negative effect that P1219X has (6, 152). Additionally, the
EA2 truncations in the II-III loop reported so far are all due to frameshifts, which adds an extra peptide sequence to the truncated 1A. It is possible that the extra peptide sequences contribute to the pathogenic effects of these EA2 truncations. Alternatively, normal and EA2 truncated forms of 1A may be regulated by signaling molecules in different ways, and subsequently cause different effects on the cells. Thirdly, some short forms of 1A may be produced by proteolysis, which produces two complementary fragments of 1A. While each single fragment may have a dominant negative effect, two complementary fragments together may abolish the dominant negative effects and even form a functional channel. It has been shown that two complementary 1B fragments, CaV2.2-Dom I-II and CaV2.2-Dom III-IV, both display a dominant negative effect when expressed individually(187). When coexpressed, however, they can form a functional channel (187). Future work on both non-disease and EA2 forms of truncated 1A may reveal more details of the mechanisms for EA2 pathogenesis and discover potential therapeutic tools.
78
A
docking
B
no docking
C
or
competition no competition
D
E
Figure 3. Possible mechanisms for the dominant negative effect of truncated CaV2 1 subunits and the rescuing effect of the partial N-terminal tails. (A) Left, cartoon of the full-length CaV2 1, showing the four repeats, N-terminal tail (blue line), and the two critical arginine residues (black dots). The N-terminal tail interacts with an intramolecular docking site through the two critical arginine residues. Right, full-length CaV2 1 with the two critical arginine residues mutated (asterisks). The intramolecular docking is reduced by the mutations. (B)
Two-domain truncated CaV2 1. No intramolecular docking site is available. (C) The N-terminal tail of the truncated 1 binds to the docking site on the full-length 1 subunit, initiating the unfolded protein response (UPR). The binding is reduced by mutations of the two critical arginine residues in the truncated 1. On the other hand, when the two critical arginine residues in the full-length 1 are mutated, the binding is stronger because there is no competition between the N-terminal tails of the full-length and truncated 1 subunits. As a result, the dominant negative effect becomes stronger. (D) The partial N-terminal tail. (E) The partial N-terminal tail binds to the docking site in the full-length 1, thereby weakening the binding of the truncated 1 (left). When the two critical arginine residues in the full-length 1 are mutated, binding of the truncated 1 is stronger, and makes it more difficult for the binding to be disrupted by the partial N-terminal tail (right). (Modified from Dahimene et al., 2016 (190)).
79
1.7.2.5 Splice variants of 1A
Soong et al. systematically scanned the whole cDNA of human 1A using PCR primers (each pair spanning 1-4 exons), and discovered seven loci of alternative splicing (218) (Figure 4). At locus 1, inclusion of V slows inactivation, and also alters
G-protein and PKC regulation (23). At locus 4, insertion of NP slows activation and inactivation, and decreases the affinity of the channel to -agatoxin IVA (45). It is hypothesized that the Q-type channel (low -agatoxin IVA affinity) lacks V at locus 1 and contains NP at locus 4, while the P-type channel (high -agatoxin IVA affinity) contains V at locus 1 and lack NP at locus 4. However, secondary modifications and accessory subunits may also contribute to the functional properties of the P-type and
Q-type channels (23, 45). At locus 5, the mutually exclusive exons e37a and e37b are also present in 1B and 1E (219). In the case of 1B, the two exons are associated with different ubiquitination levels and different sensitivities to proteasomal degradation (see 1.5.2.3). At locus 6, one or both of exons 43 and 44 can be deleted
(218). The more exons are deleted at this locus, the greater the current amplitude is and the stronger the calcium dependence of inactivation is (218). At locus 7, exon 47 can be included or excluded because of a frameshift (218). Additionally, there is an isoform with a shorter exon 47 due to a frameshift (220). Functionally, these three isoforms do not differ significantly when coexpressed with any of the four different beta subunits (220).
80
Figure 4. Seven loci of human 1A alternative splicing identified by transcript scanning. CI, Ca2+ inactivation region, containing the EF-hand and IQ domain, which is likely important for Ca2+ -dependent inactivation (CDI). At locus 2, deletion of exons 16 and 17 would cause a molecular weight loss of 7 kDa and would delete half the “P-loop” and the entire S6 of repeat II, probably producing a nonfunctional channel. At locus 6, deletion of exons 43 and 44 would cause a molecular weight loss of 6 kDa. At locus 7, deletion of exon 47 would cause a molecular weight loss of 16 kDa. (Adapted from Soong 2002 (218)).
Rabbit and rat share two distinct isoforms of 1A that differ in the region encoded by exon 19: a short form termed rbA and a long form term BI (221). These two isoforms differ in their sub-cellular distributions and interactions with SNARE proteins. In cerebellar Purkinje cells, the BI isoform is preferentially localized along the length of dendrites, whereas the rbA isoform is preferentially localized in the soma
(221). Presynaptic membrane proteins syntaxin and SNAP-25 both interact with the
BI isoform of 1A, while only SNAP-25 has been found to interact with the rbA isoform (222). It is surmised that both BI and rbA isoforms of 1A are also present in humans (222), although only the BI isoform has been found in humans (223). 81
Exon 19 of 1A is also subject to another type of alternative splicing, which gives rise to two splice variants that lack most of the synprint site (224). mRNAs coding for these splice variants can be detected in most brain areas and are also found in two types of neuroendocrine cells: PC12 cells (a rat pheochromocytoma cell line) and the magnocellular neurosecretory cells (MNCs) of the hypothalamus (224). Both splice variants are functional, but the variant with a larger deletion in the synprint site displays lower current density and a marked shift in the voltage dependence of inactivation to the depolarizing direction (224).
Alternative splicing loci 1 and 4 (Figure 4) have been found in both rat (23) and human (218). The BI and rbA isoforms are shared by rabbit, rat (221), and most likely in humans (222). Additionally, the mutually exclusive e37a and e37b are present in
1A, 1B, and 1E (219). These findings suggest that most, if not all, types of alternative splicing of 1A is conserved across species.
1.8 Thesis introduction
The P/Q-type calcium channel plays crucial physiological roles (1, 13).
Mutations in the pore-forming 1A subunit of the P/Q-type channel cause severe neurological disorders, including EA2, which can be caused by a variety of truncation mutations (15). Many of the EA2 truncation mutants and lots of non-disease-related truncated versions of 1A have been found to have dominant negative effects (see
1.7.2.4). Interestingly, many short forms of 1A are also present in healthy brains (5,
82
8-12). These findings raised several questions about the short forms of 1A in healthy brains. Are they products of protein degradation or are they of physiological importance? What parts of the 1A protein are they? How are they produced? What are their functions? Why do they not cause any problems like the EA2 truncation mutants do? Are they related to the pathogenesis of EA2? This thesis will attempt to address these questions in the following chapters.
Chapter 3 identifies a 190-kDa truncated form of 1A and investigates the origin of this truncated channel. In this chapter, I also explore the functional importance of the 190-kDa truncated channel from a biochemical perspective. Chapter 4 describes the attempt to visualize 1A on the plasma membrane. Chapter 5 examines the functions of truncated P/Q-type channels and their effects on the full-length P/Q-type channel. In this chapter, I also raise concern about the investigation of the dominant negative effects of truncated P/Q-type channels, including EA2 mutants. Chapter 6 summarizes the whole thesis and proposes future directions.
83
Chapter 2
Materials and Methods
84
2.1 Molecular biology
The mouse 1A construct for transfection in HEK 293T cells was obtained from
Addgene (Catalog #26573). The rat 3 (NCBI accession number NM_012828.2) construct used for transfection of HEK 293T and neurons is cloned in a pcDNA3.1(+) vector, and the one used for RNA synthesis is cloned in a pGEMHE vector. The rabbit
2- (NCBI accession number NM_001082276.1, D87 mutated to H, S620 deleted) construct used for transfection of HEK 293T and neurons is in a modified pEGFP-C1 vector without the EGFP gene, and the same 2-1 gene in pGEMHE vector was used for RNA synthesis. Human 1A (GenBank accession number U79666.1, Q2314 insertion) was used for RNA synthesis and for making all the engineered 1A constructs. In all the tagged human 1A constructs, an 8-amino acid linker
(GDPFALCR) was inserted between EGFP (as in pEGFP-C1) and human 1A. HA tags were inserted using splice-by-overlap PCR. In hPQGH1, hPQGH2, hPQGH3 and hPQGH4, the HA tag was inserted between residues C272-G273, G645-P647,
G1478-P1479, and G1718-I1719, respectively. Note that in hPQGH2, T646 was deleted and the sequence ARHYPYDVPDYAVTFDEGTA was inserted, as previously reported (225). To make hPQGH4x2, one more HA tag was inserted between
P1770-C1771 in hPQGH4. LGH3 (4) and L-SEP2-HA6 were both based on rat brain
1C (GenBank accession number M67515.1). In L-SEP2-HA6, a super-ecliptic pHluorin (SEP) (GenBank accession number AAS66682.1), with an N-terminal linker
85
TRLVPRGSGG and C-terminal linker GGLVPRGSFDEMQ, was inserted between
Q683-T684, and two HA tags were inserted between S992-A993 and G1136-P1137.
Human 1A was truncated using PCR to make the fragment channels, A1 (M1-A444),
A2 (D445-C2506 with M on the N-terminus), B1 (M1-N719), B2 (A720- C2506 with
M on the N-terminus), C1 (M1-M1521), and C2 (M1522- C2506). To make the
FLAG-tagged fragment channels, MDYKDDDDK was added to the N-terminus of A1,
B1, and C1 by PCR. To make the HA-tagged fragment channels, MAYPYDVPDYA was added to the N-terminus of A2, B2, and C2 by PCR, with a 2-amino acid linker
(ST). All the above residue numbers refer to the amino acid residues in the unmodified human 1A.
2.2 Cell culture
All cell cultures were kept 37°C in a humid atmosphere with 5% CO2 unless otherwise indicated.
2.2.1 Mouse cortical neuron culture
On postnatal day 0 (P0), C57BL/6 mouse pups were euthanized using a carbon dioxide chamber and then decapitated. Cerebral cortices were isolated in ice-cold
Hanks' Balanced Salt Solution (HBSS, Life Technologies 14185052, 1X, pH 7.4) supplemented with 1:200 penicillin and streptomycin (Sigma-Aldrich P0781), in a laminar flow hood. The isolated cortices were minced and digested at 37°C for 15 min
86 in 0.25% Trypsin and 47500 U/mL of DNase I (EMD Millipore 260913) in HBSS
supplemented with 10 mM of MgCl2 and 1 mM of CaCl2. The solution of enzymes were then removed and the digested cortices were triturated with a serological pipette in “final DMEM”, DMEM (Life Technologies 11965092) supplemented with 10% fetal bovine serum (Fisher Scientific SH3007002) and 10% F-12 (Life Technologies
21700-075). The suspension with triturated cortices was centrifuged at 200 g for 30 s.
The supernatant was collected and centrifuged at 1500 g for 3 min. The supernatant was aspirated and the precipitated cells were resuspended in final DMEM. 2-3×106 cells were plated in each poly-D-lysine-coated 60-mm dish. 4-6 h after plating, cells were washed gently with DMEM (without supplements), and Neurobasal medium
(Life Technologies 21103-049) supplemented with B-27 (1/50, Life Technologies
17504-044), GlutaMAX (1/100, Life Technologies 35050-061) and 20 g/mL of gentamicin. Every 3-4 days, half of the culture medium was replaced with the same volume of Neurobasal medium supplemented with B-27, GlutaMAX, and gentamicin.
2.2.2 Rat hippocampal neuron culture
Pregnant Sprague-Dawley rats were euthanized using a carbon dioxide chamber and decapitated. Embryos at age E18-19 were quickly moved to ice-cold Hanks'
Balanced Salt Solution, where the hippocampi were isolated. The hippocampi were digested at 37°C for 15 min in 0.125% Trypsin. The trypsin solution was then removed and the digested hippocampi were triturated using a syringe with a 20G
87 needle in final DMEM (see 2.2.1). 3.33×104 cells were plated on each poly-D-lysine-coated 12-mm glass coverslips. 16 h later and then every 3-4 days, half of the culture medium was replaced with the same volume of Neurobasal medium supplemented with B-27 and GlutaMAX.
2.2.3 HEK 293T cell culture
HEK 293T cells were maintained in DMEM supplemented with 10% fetal bovine serum (VWR 45001-108) and 1% penicillin and streptomycin (Sigma). When cells reached 85-90% confluency, they were rinsed with PBS (Life Technologies
10010-049), dissociated using 0.05% trypsin-EDTA solution (Life Technologies
25300-054), resuspended in DMEM supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin, and seeded in new dishes at an appropriate density.
2.3 SDS-PAGE sample preparation
2.3.1 Preparation of tissue/cell lysate using Pierce IP Lysis Buffer
For brain tissue lysate, a C57BL/6 mouse was euthanized using a carbon dioxide chamber and decapitated. The desired brain region was quickly isolated in ice-cold
PBS (Life Technologies) containing 1/1000-1/500 Halt Protease Inhibitor Cocktail
(Life Technologies 78430). The isolated brain tissue was homogenized using a syringe with a 22G needle in Pierce IP Lysis Buffer with 1/50 Halt Protease Inhibitor Cocktail.
The homogenate was incubated at 4°C on a rotator, vortexing every 10 min, and then
88 centrifuged for 30 min at 16,100 g at 4°C. The protein concentration of the supernatant (lysate) was measured using the Coomassie Plus (Bradford) Assay Kit
(Life Technologies 23236). One volume of lysate was mixed with 1/2 volume of Blue
Loading Buffer (New England Biolabs B7703S) and 50 mM of extra dithiothreitol
(DTT, Life Technologies 20291), and incubated at 37°C for 30 min before being analyzed using SDS-PAGE or frozen at -80°C for future use.
For whole-cell lysate of cultured neurons, the neurons were first gently rinsed twice with prewarmed Neurobasal medium without supplements and then scraped in ice-cold Pierce IP lysis buffer with 1/50 Halt Protease Inhibitor Cocktail. The neurons were homogenized by pipetting and vortexing. The homogenate was incubated at 4°C on a rotator with vortexing every 10 min, and then centrifuged for 30 min at 16,100 g at 4°C. The supernatant (lysate) was taken and further processed in the same way as the brain tissue lysate.
For whole-cell lysate of HEK 293T cells, the cells were first gently rinsed twice with prewarmed DMEM without supplements and then scraped in ice-cold Pierce IP lysis buffer with 1/50 Halt Protease Inhibitor Cocktail. The cells were homogenized using a syringe with a 22G needle. The homogenate was then processed in the same way as the homogenate of cultured neurons.
89
2.3.2 Preparation of tissue/cell lysate using SDS lysis buffer
Brain tissue was isolated as described in 2.3.1 and quickly transferred to SDS lysis buffer (100 mM NaCl, 50 mM Tris, 10 mM EDTA, 10 mM EGTA, 2% SDS, pH
8.0) in a glass homogenizer in a 95°C wash bath. The tissue was rapidly homogenized and further incubated at 95°C for 5 min. The homogenate was incubated at room temperature on a rotator for 30 min with vortexing every 10 min, and then centrifuged for 30 min at 16,100 g at room temperature. The supernatant (lysate) was taken and the protein concentration was measured using the Coomassie Plus (Bradford) Assay
Kit. One volume of lysate was mixed with 1/3 volume of 30% (v/v) glycerol in PBS,
1/6 volume of Blue Loading Buffer (New England Biolabs) as well as 50 mM of extra
DTT (Life Technologies), and incubated at 37°C for 30 min before being analyzed using SDS-PAGE or frozen at -80°C for future use.
For cultured neuron lysates, the neurons were first gently rinsed twice with prewarmed Neurobasal medium without supplements, and then scraped in SDS lysis buffer with 50 mM DTT at 95°C. Then neurons were homogenized by pipetting, and incubated at 95°C for 5 min. The homogenate was then processed in the same way as the homogenate from brain tissues.
2.3.3 Purification of 1A with nickel affinity gel
The following solutions were needed: nickel lysis buffer (300 mM NaCl, 50 mM
NaH2PO4, 1% (v/v) Triton X-100, pH 8.0), nickel wash buffer (300 mM NaCl, 50 mM
90
NaH2PO4, 20 mM imidazole, pH 8.0) and nickel elution buffer (300 mM NaCl, 50 mM NaH2PO4, 250 mM imidazole, pH 8.0).
Lysate of mouse brain tissue or cultured neurons was prepared by homogenization and centrifugation, as described in 2.3.1, with nickel lysis buffer substituted for Pierce IP Lysis Buffer. 50 L of lysate was set aside as the whole-cell lysate sample.
Nickel affinity gel (Ni Sepharose High Performance, GE Healthcare 17-5268-01) was washed with nickel lysis buffer, mixed with the lysate and 10 mM imidazole in a spin column (Life Technologies 69725), and incubated at 4°C on a rotator for 2 h. The flow-through was discarded and the nickel affinity gel was washed with nickel wash buffer. The protein bound to the nickel affinity gel was eluted using nickel elution buffer. The concentrations of the whole-cell lysate and the eluate were measured before they were mixed with 1/2 volume of Blue Loading Buffer (New England
Biolabs) and 50 mM of extra DTT, and subsequently incubated at 37°C for 30 min.
The samples were being analyzed using SDS-PAGE or frozen at -80°C for future use.
2.3.4 Preparation of oocyte membrane fraction
Xenopus oocytes were washed with oocyte lysis buffer (mM: Tris 50, EDTA 1,
EGTA 1, pH 8.0) and then homogenized with buffer containing 1/50 Halt Protease
Inhibitor Cocktail using a syringe with a 25G needle. The homogenate was centrifuged at 1,500 g for 15 min at 4°C. The supernatant was then centrifuged at
91
100,000 g for 40 min at 4°C. The surface of the pellet was washed with oocytes lysis buffer with 1/500 Halt Protease Inhibitor Cocktail. The pellet (membrane fraction) was then resuspended in Pierce IP lysis buffer with 1/50 Halt Protease Inhibitor
Cocktail. The protein concentration was measured before the sample was mixed with
1/2 volume of Blue Loading Buffer (New England Biolabs) and 50 mM of extra DTT, and incubated at 37°C for 30 min. The sample was being analyzed using SDS-PAGE or frozen at -80°C for future use.
2.4 SDS-PAGE and Western blot
Samples were run in 6% or 8% polyacrylamide gels or NuPAGE Novex 4-12%
Bis-Tris Protein Gels (Life Technologies) in Tris-Gly-SDS buffer (Bio-Rad 161-0732) at 175 V. Spectra Multicolor High Range Protein Ladder (Life Technologies 26625) or
Precision Plus Protein All Blue Standard (Bio-Rad 161-0373) were used. Proteins were transferred to PVDF membrane (Bio-Rad 162-0177) in transfer buffer (25 mM
Tris, 192 mM glycine, 0.002% SDS) for 90 min at 90 V at 4°C. Membranes were blocked in blocking buffer for 3 h at room temperature. For samples from Xenopus oocytes, the blocking buffer is 10% newborn calf serum (NCS, Life Technologies
26010-066), 1% fish gelatin (Sigma-Aldrich G7041) and 0.2% Tween-20
(Sigma-Aldrich P2287) in PBS. For other samples, the blocking buffer contains 2 volumes of Odyssey Blocking Buffer (LI-COR Biosciences 927-40000) and 1 volume of PBS (Life Technologies) supplemented with 0.2% Tween-20. Membranes were
92 incubated with primary antibody diluted with blocking buffer overnight at 4°C.
Membranes were washed with PBS (Life Technologies AM9625, 1X) containing 0.2%
Tween-20 (PBST) 4 times, 5 min each time. Then membranes were incubated with secondary antibody diluted with blocking buffer for 1 h at room temperature. After four 5-min washes with PBST, membranes were rinsed with PBS, and then the protein bands were visualized using enhanced chemiluminescence reagents (Life
Technologies 34079 and 34094) on X-ray films (HyBlot and Kodak).
The following primary antibodies were used: rabbit polyclonal anti-1A II-III loop (1/1000, Sigma C1353), rabbit polyclonal anti-1A N-terminus (1/1000, Abcam ab32642), rabbit polyclonal anti-actin (1/1500, Sigma A2066), rabbit monoclonal anti-Na/K-ATPase (1/10000, Abcam ab76020), mouse monoclonal anti-spectrin
(1/1000, EMD Millipore MAB1622), rabbit polyclonal anti-PARP (1/1000, Cell
Signaling 9542), mouse monoclonal anti-FLAG (1/500, 2 g/mL, Sigma F1804), and mouse monoclonal anti-HA (1/1000, BioLegend 901502). Secondary antibodies used are goat anti-rabbit IgG-HRP(1/2000, Santa Cruz Biotechnology sc-2004) and goat anti-mouse IgG-HRP (1/2000, Santa Cruz Biotechnology sc-2005).
2.5 Transfection
Cultured rat hippocampal neurons were transfected on day in vitro (DIV) 4. 6 L of Lipofectamine 2000 (Life Technologies 11668030) was diluted into 400 L of
Opti-MEM (Life Technologies 31985062) and incubate at room temperature for 5 min.
93
Meanwhile, 8 g of DNA was diluted into 400 L of Opti-MEM. The diluted
Lipofectamine 2000 was mixed with the diluted DNA and incubated at room temperature for 25-30 min. Meanwhile, coverslips (usually four) with neurons were transferred to a 60-mm Petri dish with prewarmed Neurobasal medium without supplements. The DNA-Lipofectamine mixture was then added to the neurons. The neurons were incubated in the incubator for 40 min before being electroporated using an electroporator (BTX Harvard Apparatus) to facilitate transfection. The neurons were then moved back to the original cultural medium. 16 h after transfection, half the culture medium was replaced with Neurobasal medium supplemented with B-27 and
GlutaMAX. Further experiments were performed 36-48 h after transfection.
HEK 293T cells were propagated on the day before transfection so that the cells will be 85~95% confluent on the day of transfection. For each 35-mm Petri dish with coverslips, 5 L of Lipofectamine 2000 was diluted into 125 L of Opti-MEM and incubated for 5 min at room temperature. Meanwhile, 4 g of DNA was diluted into
125 L of Opti-MEM. For each 60-mm Petri dish, the amount of each reagent was doubled. The diluted Lipofectamine 2000 was mixed with the diluted DNA and incubated at room temperature for 25-30 min. The DNA-Lipofectamine mixture was then added to the cells. For imaging experiments with cells on coverslips, the cells were incubated in the incubator for 40 min, before the culture medium was replaced with DMEM supplemented with 10% fetal bovine serum and 1% penicillin and streptomycin. For biochemical experiments, the cells were incubated overnight before
94 the culture medium was replaced. Further experiments were done 24-48 h after transfection.
2.6 RNA synthesis
Template DNAs were linearized overnight and purified using the
Phenol:Chloroform:Isoamyl Alcohol (25:24:1) reagent (pH 8.0, Sigma-Aldrich
P2069) and sodium acetate/ethanol precipitation. cRNAs were synthesized using an in vitro transcription reaction containing the purified DNA template, 2 L T7 RNA
7 polymerase (100 units, New England Biolabs M0251S), 15 L m G(5')ppp(5')G RNA
Cap Structure Analog (10 mM, New England Biolabs S1404S), 4 L rNTP (100 mM,
Roche), 5 L DTT (100 mM, Life Technologies R0861), 7.2 L MgCl2 (50 mM), 4
L RNAPol Reaction Buffer (New England Biolabs), 1.5 L ribonuclease inhibitor
(15 units, Life Technologies 15518-012), and 10.8 L DEPC-treated water. The reaction mixture was incubated at 37°C for 2.5-3 h. The cRNAs were purified using the Phenol:Chloroform:Isoamyl Alcohol (25:24:1 reagent (pH 5.2, Fisher Scientific
BP1753I-100) and sodium acetate/ethanol precipitation. The cRNAs were then dissolved in DEPC-treated water. The RNA concentration was determined by agarose-formaldehyde gel electrophoresis with an RNA ladder for comparison (Life
Technologies 15623200).
95
2.7 Oocyte preparation and cRNA injection
Female Xenopus laevis (Xenopus I) were anesthetized in a carbon dioxide chamber and euthanized by decapitation followed by pithing of the spinal cord.
Ovarian lobes were divided into small parts and treated with 0.4 mg/mL collagenase A
(Roche 11088793001) in OR2 solution (82.4 mM NaCl, 2.5 mM KCl, 1 mM MgCl2, 5 mM HEPES, pH 7.6) for 1.5-2.5 h under 250 rpm shaking. The oocytes were then
rinsed with ND96 solution (96 mM NaCl, 2.5 mM KCl, 1 mM MgCl2, 5 mM HEPES,
1.8 mM CaCl2, 100 units/mL penicillin, 100 μg/mL streptomycin, pH 7.6).
Defolliculated stage VI oocytes were selected. cRNAs were injected into oocytes
(1.25 ng/oocyte for 1 or subunits, 0.2 ng/oocyte for 2 subunit). Experiments were performed 3-5 d after injection.
2.8 Brain slices
21-day old C57BL/6 mice were euthanized using a carbon dioxide chamber and decapitated. Brain dissection was performed in a sucrose-saline solution (27 mM
NaHCO3, 1.5 mM NaH2PO4, 2.55 mM KCl, 222 mM sucrose) bubbled with 95% O2 and 5% CO2 at 4°C. Coronal slices of 500 μm were obtained from the cerebral cortex of each hemisphere using a brain slicer instrument while bubbling in ice-cold sucrose solution. Slices were gently transferred to artificial cerebrospinal fluid (ACF, in mM:
NaCl 119, NaHCO3 26, NaH2PO4 1.25, KCl 2.5, glucose 15, myo-inositol 1, pyruvate
2, N-acetyl-L-cystein 0.06) of 35-37°C, bubbled with 95% O2 and 5% CO2, to recover
96 for 60 min before any treatment.
2.9 Surface biotinylation
For surface biotinylation of brain slices, slices were rapidly moved to ice-cold
ACF containing 1 mg/mL sulfo-NHS-SS-biotin from the Pierce Cell Surface Protein
Isolation Kit (Life Technologies 89881), bubbled with 95% O2 and 5% CO2, and incubated for 45 min. Slices of the mock biotinylation control group were treated in parallel using the same kind of solution in which sulfo-NHS-SS-biotin was omitted.
Quenching solution (from the kit) of 20% of the volume of the ACF was added and incubated for 5 min. Slices were then rapidly rinsed with phosphate-buffered saline
(PBS, in mM: KH2PO4 1.06, NaCl 155.17, Na2HPO4 2.97, pH 7.4) with 1:250 Halt protease inhibitor cocktail (Life Technologies 78430). Slices were homogenized in lysis buffer (from the kit) with 1:50 Halt protease inhibitor cocktail using a syringe with a 22G needle, and incubated for 1 h at 4°C with vortexing for 40 s every 2 min.
The homogenates were centrifuged at 10,000 g for 2 min. A small portion of each supernatant was set aside as the whole-cell lysate sample. The remaining supernatants were mixed with NeutrAvidin agarose beads in spin columns (from the kit) and rotated for 1 h at room temperature. The flow-through samples were collected by centrifugation. The beads were washed 4 times with wash buffer (from the kit) with
1:50 Halt protease inhibitor cocktail, and then incubated with Blue Loading Buffer
(1X, diluted with PBS, New England Biolabs B7703S) with 50 mM of extra DTT for
97
1 h at 37°C. The eluates were collected by centrifugation.
For surface biotinylation of Xenopus oocytes, ~30 healthy oocytes from each group were selected 5 days after cRNA injection and washed 3 times in ice-cold OR2.
Oocytes were incubated in 2.5 ml of 1.2 mg/ml sulfo-NHS-SS-Biotin (from the kit) in ice-cold OR2 for 30 min with gentle shake. 500 l quenching solution (from the kit) was added to each group and incubated for 5 min. Oocytes were then washed with 3 ml of Tris Buffer Saline (from the kit) with 1:500 Halt protease inhibitor cocktail 3 times, and homogenized using a syringe with a 22G needle in oocyte lysis buffer (mM:
Tris 5, EDTA 1, EGTA 1, pH 8.0), 10 l per oocyte, with 1:50 Halt protease inhibitor cocktail. The homogenates were centrifuged at 1,000 g for 5 min at 4°C. Supernatants were then subjected to ultracentrifugation at 100,000 g for 40 min at 4°C. Meanwhile,
NeutrAvidin agarose beads, 125l per group, were washed three times with Pierce IP lysis buffer in a spin column. After ultracentrifugation, the pellets were resuspended in Pierce IP lysis buffer, 5l per oocyte, with 1:50 Halt protease inhibitor cocktail, mixed with the NeutrAvidin agarose beads, and gently rotated at 4°C for 4 h.
Flow-through samples were collected by centrifugation. The beads were washed 4 times with high-salt wash buffer (300 mM NaCl and 1% (v/v) Triton X-100 in Pierce lysis buffer) and incubated with Blue Loading Buffer (1X, diluted with PBS, New
England Biolabs) with 50 mM of extra DTT for 1 h at 37°C. The eluates were collected by centrifugation.
98
2.10 Immunocytochemistry and confocal microscopy
2.10.1 Fixed-cell imaging
Immunocytochemistry for fixed-cell imaging was performed at room temperature. Fixed-cell imaging of hippocampal neurons was performed 36-48 hours post-transfection. For immunocytochemistry under non-permeabilizing conditions, the culture medium was removed and the neurons were gently rinsed with PBS (Life
Technologies). Neurons were fixed for 15 min in fixation solution (2% paraformaldehyde and 4% sucrose in PBS, pH7.4) and then washed 4 times with PBS.
Neurons were blocked for 2 h with blocking buffer (10% goat serum in PBS, goat serum: Life Technologies 16210-064), and then stained with a mouse anti-HA antibody (1:100, Biolegend 901502) in the same kind of blocking buffer for 1 h.
Neurons were washed 4 times with PBS before being stained with an
Alexa594-conjugated goat anti-mouse secondary antibody (1:750, Life Technologies
A-11005) in the same kind of blocking buffer. Neurons were finally washed 4 times with PBS before confocal microscopy.
Fixed-cell imaging of HEK 293T cells was performed 24 h post-transfection. The protocol is the same as above except that the blocking buffer for HEK 293T is 5% goat serum, 1% bovine serum albumin (Jackson ImmunoResearch Laboratories
001-000-161) and 0.5% fish gelatin (Sigma-Aldrich G7041) in PBS.
The protocol of fixed-cell imaging under permeabilizing conditions is the same as the protocol of fixed-cell imaging under non-permeabilizing conditions, except that
99 the cells were permeabilized using 0.25% Triton X-100 (v/v) in PBS for 15 min before blocking.
2.10.2 Live-cell antibody feeding
Live-cell antibody feeding of hippocampal neurons was performed 36-48 hours post-transfection. Neurons were incubated in mouse anti-HA antibody (1:80,
Biolegend 901502) diluted in conditioned medium for 45 min in the incubator.
Neurons were washed 4 times with ice-cold PBS before being fixed with fixation solution (2% paraformaldehyde and 4% sucrose in PBS, pH7.4) for 40 min at room temperature. The following steps were all performed at room temperature. Neurons were washed 4 times with PBS, and then blocked for 1.5 h in 1.5 mL of blocking buffer (0.5% fish gelatin and 2% goat serum in PBS) supplemented with 4 drops of signal enhancer (Life Technologies I36933). Neurons were then stained for 1.5 h with the Alexa594-conjugated goat anti-mouse secondary antibody (1:500, Life
Technologies A-11005) diluted in the same kind of blocking buffer with signal enhancer. Neurons were finally washed 4 times with PBS before confocal microscopy.
2.11 Two-electrode voltage clamp (TEVC)
The bath solution contained 4 mM Ba(OH)2, 96 mM NaOH, 2 mM KCl and 5 mM HEPES (pH 7.4 with methanesulfonic acid, osmolarity 195 mmol/kg). Electrodes were filled with 3 mM KCl and had a resistance of 0.2-1 M. For I-V curves, currents
100 were evoked by step depolarizations to various levels from a holding potential of -100 mV. To determine the steady-state inactivation, the oocyte was given a 20-ms normalization pulse to the voltage level where the peak current was, followed sequentially by a 5-s conditioning pulse to various levels and a 50-ms test pulse to the same level as the normalization pulse. The interval between sweeps is 50 s.
2.12 Data analysis
Densities of bands in Western blot were analyzed using ImageJ. TEVC data were analyzed using Clampfit and presented as mean ± SEM. For I-V curves, currents of each oocyte at different voltage levels were normalized to the peak current of that cell.
At each voltage level, the normalized currents of all the oocytes in the same group were averaged. For steady-state inactivation curves, current evoked by the test pulse was normalized by the current evoked by the normalization pulse. At each voltage level, the normalized currents of all the oocytes in the same group were averaged and plotted on the graph against the conditioning potentials.
101
Chapter 3
Biochemical Studies of the 190-kDa Fragment of the P/Q-Type
Calcium Channel
I performed all of the experiments, except for the Western blot analysis of the lysate of a whole mouse brain in section 3.2 and the experiments in section 3.7, which were done by Ye Gong. The analyses in 3.9 were done by Dr. Lawrence Chasin.
102
3.1 Introduction
The CaV1 subunits of voltage-gated calcium channels have been found to exist not only in full-length forms, but also in the forms of shorter lengths—fragment channels. A number of fragment channels of L-, P/Q- and N-type CaV1 have been found in Western blots (5, 8, 10-12, 49, 131, 132, 141, 176, 177, 226-230). Our lab found that 1C in the rat brain can be proteolytically cleaved at a site presumably in the II-III loop, resulting in two fragments of 90 kDa and 150 kDa, respectively (4).
This “mid-channel” proteolysis increases significantly with calcium activity and with age. The mechanism involves the calcium-dependent protease calpain and the ubiquitin-proteasome system.
Proteolysis has also been shown to be the mechanism for the production of other fragments of the L-type calcium channels (see 1.6.1.1) and sodium channels (see
1.6.2). In contrast, the mechanisms that cause the various short forms of the P/Q-type calcium channel have not been thoroughly studied. A previous study reported that a
75-kDa C-terminal fragment of 1A is generated by an internal ribosome entry site
(IRES) (7). Another study demonstrated that a 190-kDa fragment of 1A in the rat brain contains both the N- and C-termini, suggesting that this fragment does not result from proteolysis, but from alternative splicing (8). However, known alternative splicing events can only reduce the molecular weight of rat 1A to about 220 kDa
(see 1.7.2.5).
It is noteworthy that in most of the previously reported Western blot results of
103 endogenous 1A, the apparent molecular weights of the putative full-length 1A, the highest band, are smaller than the theoretical molecular weights of 1A. In addition, they also saw shorter bands with molecular weights between 140-190 kDa (5, 8, 10-12,
176, 177). In contrast, the apparent molecular weights of 1A expressed in heterologous systems were shown to be closer to the theoretical size, and there are no bands present between 140-190 kDa (177, 178). This indicates that native 1A undergoes some processing events that do not happen in heterologous systems.
In this chapter, I report the existence of a 190-kDa fragment of 1A and investigate the origin of this fragment.
3.2 The 190-kDa fragment of 1A in the mouse brain
The cerebella from a wild-type (WT) male mouse and an 1A-knockout male mouse of postnatal day 16 (P16) were lysed in Pierce IP Lysis Buffer and the lysates were run on a gel for Western blot, as described in Materials and Methods. An anti-1A antibody (Sigma-Aldrich C1353) was used to detect 1A. This antibody is against the peptide CNA1, corresponding to residues 865-881 in the II-III loop of rat brain 1A (UniProtKB - P54282), and thus will be hereafter referred to as the “II-III loop antibody”. According to Sigma-Aldrich, the antibody also recognizes mouse
1A, whose residues 865-881 (UniProtKB - P97445) have a sequence identity of
16/17 with rat 1A. The WT sample shows a “doublet” band containing two similar-sized bands around 190 kDa (Figure 5A). The knockout sample does not have
104 the doublet band, indicating that the doublet band is from 1A. Surprisingly, full-length mouse 1A, 270 kDa in theory, is almost invisible. In some other experiments, the full-length band is more visible (Figure 8A).
To investigate whether the lack of full-length 1A is an age-specific phenomenon, lysates of cerebella from mice of different ages were tested using
Western blot. The result shows that the lysates of newborn, 19-day and 9-month old mice all have the 190-kDa doublet band as the predominant form of 1A (Figure 5B).
To see whether the 190-kDa doublet band also exists in other regions of the mouse brain, lysates of the cerebellum, hippocampi and the cerebral cortex from a mouse were run on the same gel for Western blot. The result shows that the lysates of the cerebral cortex and hippocampi also have the 190-kDa doublet band (Figure 5C).
In the lysate of a whole mouse brain, the amount of the 190-kDa doublet band is much higher than the full-length 1A, as detected by the II-III loop antibody (Figure 5 D and E).
In some experiments, the lower-molecular-weight band of the doublet band is almost invisible. Thus, the 190-kDa doublet band or single band—if only one band is visible—will be referred to as “the 190-kDa fragment” or “the 190-kDa species”.
105
A B 1A- WT KO 1 2 3 4
kDa kDa
250 250
150 150
100 100
75 75
WT KO
anti-Na,K-ATPase
D C E 1 2 3 kDa kDa 315 315 kDa 250 250 180 250 180 130 130 150 95 95 100
Figure 5. Western blot analysis of the mouse brain with the II-III loop antibody reveals the 190-kDa fragment of 1A in the mouse brain. (A) Lysates of the cerebella from a wild-type and an 1A-knockout P16 mouse. Probing with anti-Na,K-ATPase reveals that comparable amounts of protein were loaded. (B) Lysates of the cerebella from mice of different ages: 1) newborn; 2) 19 days; 3) 9 months; 4) negative control: P14 1A-KO. (C) Lysates of different brain regions: 1) cerebellum; 2) hippocampus; 3) cortex. (D) Whole-brain lysate of a P10 mouse. (E) Whole-brain lysate of a P21 mouse.
3.3 The 190-kDa fragment of 1A in cultured mouse cortical neurons
Neurons in the brain and neurons in culture are exposed to drastically differently environments. To investigate whether the 190-kDa fragment also exists in primary
106 neuron cultures, cultured cortical neurons from WT and 1A-knockout mice were lysed on day in vitro (DIV) 14 in Pierce IP Lysis Buffer, as describe in Materials and
Methods. The lysates were run on a gel for Western blot. Two primary antibodies were used and one was the II-III antibody as in 3.2. The other was an antibody (Abcam ab32642) against the N-terminal tail (residues 1-100) of rat 1A, hereafter referred to as the “N-terminus antibody”. According to Abcam, this antibody also works with mouse 1A, whose N-terminus is completely the same as rat 1A.
The II-III loop antibody detected a band with an apparent molecular weight of
~270 kDa, the same as the theoretical molecular weight of full-length mouse 1A
(Figure 6A). Unlike many previous reports showing a low apparent molecular weight of the full-length 1A (5, 8, 10-12, 176, 177), this result shows that the full-length
1A has a normal apparent molecular weight. The II-III loop antibody also detected a doublet band around 190 kDa, similar to the 190-kDa fragment in mouse brain tissues.
Like the 190-kDa fragment found in mouse brain tissues, the 190-kDa fragment in cultured mouse cortical neurons is detected sometimes as a doublet band, and sometimes as a single band (Figure 6A and 8C), and thus will be referred to simply as
“the 190-kDa fragment” or “the 190-kDa species”.
The N-terminus antibody also detected the full-length 1A of 270 kDa, along with some other bands (Figure 6B). Note that the band with an apparent molecular weight of ~100 kDa might be the fragment complementary to the 190-kDa fragment, which should be ~80 kDa in theory.
107
A B
KO WT KO WT kDa kDa 300 300 250 250
180 180
130 130
100 100
Figure 6. Western blot analysis of 1A in cultured mouse cortical neurons. (A) Lysates of 1A-knockout and wild-type mouse cortical neurons probed with the II-III loop antibody. (B) Same samples as in (A), probed with the N-terminus antibody.
3.4 1A expressed in heterologous systems shows normal molecular weight
To find supporting evidence that the 270-kDa band in the above results is the full-length 1A, HEK 293T cells were transfected with mouse 1A (Addgene 26573) and the auxiliary subunits 3 and 2 as described in Materials and Methods.
Another group of HEK 293T cells were transfected with only 3 and 2 as a negative control. The cells were lysed in Pierce IP Lysis Buffer as described in
Materials and Methods, and the lysates were run for Western with the II-III loop antibody. A 270-kDa band clearly corresponds to the full-length mouse 1A and no specific bands of 190 kDa or any other low molecular weight was detected (Figure
7A).
108
A B C - + - - + + kDa kDa kDa
300 300 300
250 250 250 180 180 180
130 130 130
100 100 100
Figure 7. Western blot analysis of 1A expressed in heterologous systems. (A) Mouse 1A expressed in HEK 293T cells probed with the II-III loop antibody. (B) Human 1A expressed in Xenopus oocytes probed with the II-III loop antibody. (C) Human 1A expressed in Xenopus oocytes probed with the N-terminus antibody. In all three panels, the negative control and the sample with 1A are indicated with “-” and “+”, respectively.
Besides mouse 1A, human 1A was also expressed and tested. Oocytes of the
African clawed frog (Xenopus laevis) were injected with the cRNAs of human 1A and the auxiliary subunits 3 and 2 as described in Materials and Methods. A control group of oocytes were injected with the cRNAs of 3 and 2 only. On the fourth day after injection, the oocytes were lysed, and the membrane system was isolated and run on a gel for Western blot, as described in Materials and Methods. The
II-III loop antibody was tried, but the result is ambiguous (Figure 7B). A thick non-specific band of ~300-kDa makes it hard to tell whether there is a specific band of ~270-kDa. No specific bands of other molecular weights were detected. Note that this antibody (Sigma-Aldrich C1353) is meant to be used only for mouse and rat 1A subunits, which have identical N-terminal tail sequences. 10% of the human 1A
N-terminal tail sequence is different from that of mouse or rat, so an inconclusive 109 result with human 1A is not surprising. The N-terminus antibody (Abcam 32642) was then tried. According to Abcam, this antibody is predicted to work also with human 1A, whose N-terminal tail has a sequence identity of 91/100 with rat 1A. A specific band at ~280 kDa was detected (Figure 7C). Additionally, there are two non-specific bands near 190 kDa. However, they are both of comparable intensity with those in the control group, and thus it is unlikely that they have covered up any specific bands.
In summary, heterologously expressed mouse and human 1A did not show anomalous migration or bands of smaller molecular weights.
3.5 Evidence that the 190-kDa fragment of 1A exists in live brains and live cells
Could the 190-kDa fragment be a product of protein degradation during sample preparation, caused by a protease which does not cut 1A in live cells? Although all the sample preparation was carried out at 4°C or on ice with protease inhibitors, some residual protease activity might still exist. In order to further reduce possible protein degradation during sample preparation, mouse cerebellum and cultured mouse cortical neurons were lysed with a lysis buffer containing 2% SDS (hereafter referred to as
SDS lysis buffer) at 95°C as described in Materials and Methods. The samples were run on a gel for Western blot with the II-III loop antibody in parallel with samples prepared using the methods described in 3.2 and 3.3. When the samples are homogenized with SDS at high temperature, most proteins including most proteases,
110 will be denatured rapidly, so that proteolytic degradation can be minimized. If the
190-kDa fragment was due to proteolytic degradation during sample preparation, the new sample preparation method should reduce the amount of the 190-kDa fragment and increase the amount of full-length 1A.
As shown in Figure 8A, the 190-kDa fragment is the dominant species in both samples, prepared using Pierce IP Lysis Buffer at 4°C and SDS lysis buffer at 95°C, respectively. A 270-kDa band, presumably the full-length 1A, is also visible in both samples. The ratio of the intensity of the 270-kDa band to that of the 190-kDa fragment is very similar between the two samples: 0.12 and 0.086, respectively.
To confirm the specificity of the staining, the II-III loop antibody was incubated with a blocking peptide with the sequence of residues 865-881 of rat 1A, as in the immunogen of the antibody, prior to the immunoblotting. The antibody incubated with the blocking peptide does not recognize the 270-kDa band and the 190-kDa fragment anymore (Figure 8B). This indicated that both the 270-kDa band and the 190-kDa fragment are from mouse 1A, instead of non-specific staining.
With cultured mouse cortical neurons, a similar result was obtained (Figure 8C).
The sample prepared using SDS lysis buffer at 95°C even has a higher amount of the
190-kDa fragment than the sample prepared using Pierce IP Lysis Buffer at 4°C, while the two samples have similar amount of full-length 1A. These results suggest that the 190-kDa fragment exists in the live mouse brain and neurons, and is not an artifact due to proteolytic degradation during sample preparation.
111
blocked control A B 1 2 1 2 1 2 kDa kDa 300 300
250 250
180 180
130 130
100 100
C D E 1 2 1 2 kDa kDa kDa 1 2
300 300 300
250 250 250
180 180 180
130 130 130
100 100 100
70 70 70
Figure 8. Homogenization in SDS lysis buffer at 95°C does not reduce the amount of the 190-kDa species. In all panels, lane 1 is the sample prepared using ice-cold Pierce IP lysis buffer, and lane 2 is the sample prepared using SDS lysis buffer at 95°C. (A) Mouse cerebellum lysates probed with the II-III loop antibody. The ratios of the intensity of the 270-kDa band to that of the 190-kDa band in the two samples are 0.12 and 0.086, respectively. (B) The II-III loop antibody preincubated with the blocking peptide does not recognize the 270-kDa and 190-kDa species any more. 200 nM of blocking peptide was used for per nM of antibody. In the control group, water was substituted for the blocking peptide solution. (C) Lysates of cultured mouse cortical neurons probed with the II-III loop antibody. (D) Lysates of cultured mouse cortical neurons probed with the N-terminus antibody. (E) Lysates of cultured mouse cortical neurons as in C and D, probed with anti-Na,K-ATPase antibody, showing that sample 1 and sample 2 have comparable amounts of protein loaded.
112
3.6 The 190-kDa fragment of 1A contains the C-terminal tail but not the
N-terminal tail
Another question that we wanted to address was which part of the mouse 1A is included in the 190-kDa fragment. At the time, we already knew that the fragment included residues 865-881, since the II-III loop antibody bound to this region. Next, I investigated whether the 190-kDa fragment contained the C-terminal tail of the mouse
1A.
The C-terminal tail of the mouse 1A contains an evolutionarily conserved
“histidine tract”, which consists of 10-11 histidine residues and, according to a previous study, can be recognized by an anti-penta-HIS antibody (11). We surmised that the histidine tract can bind to immobilized bivalent nickel ions like a polyhistidine-tag. Therefore, if the 190-kDa fragment contains the C-terminal tail, it should be able to bind to immobilized bivalent nickel ions. A WT mouse cerebellum was lysed in lysis buffer, and then a nickel affinity gel (Ni Sepharose High
Performance, GE Healthcare 17-5268-01) was used to purify 1A and any 1A fragment that contains the histidine tract, as described in Materials and Methods. An
1A-knockout mouse cerebellum was treated the same way as a negative control.
As shown in Figure 9A, the eluate from the nickel affinity gel that was incubated with the lysate of WT mouse cerebellum contained a much higher concentration of the
190-kDa fragment than the lysate itself. It also contained a species of a higher molecular weight, presumably the full-length 1A, which in this case was not
113 detected in the lysate of the cerebellum. No bands were detected in the eluate of the
1A-knockout group. These findings suggest that the 190-kDa fragment has a high affinity to the nickel affinity gel.
B C A WT
KO WT KO WT WT KO WT KO elu
lys lys elu elu lys lys elu elu kDa kDa 300 kDa 250 250 250 150 180 150
100 130 100 75 75 100
WT WT WT WT WT D E F elu lys elu lys elu kDa kDa kDa
300 300 300
250 250 250
180 180 180
130 130 130
100 100 100
70 70 70
Figure 9. The 190-kDa fragment of 1A contains the C-terminal tail but not the N-terminal tail. (A) Lysates (lys) of wild-type and 1A-knockout mouse cerebella and eluates (elu) from nickel affinity gel that has been incubated with the lysates, probed with the II-III loop antibody. (B) Same samples as in (A) probed with an anti-Na,K-ATPase antibody. Volumes of the four samples in (A) are 23, 25, 20 and 25 L, respectively. Volumes of the four samples in (B) are 3, 7, 4, 5 L, respectively. (C) The II-III loop antibody recognizes the 190-kDa species. This experiment was done in parallel with (D). (D) The N-terminus antibody does not recognize the 190-kDa species. (E) Lysate of cultured mouse cortical neurons and eluate from nickel affinity gel that has been incubated with the lysate, probed with the II-III loop antibody. (F) Same samples as in (E), probed with the N-terminus antibody.
To find supporting evidence that the affinity of the 190-kDa fragment to the nickel affinity gel is due to the histidine tract, the samples were also run on a Western 114 blot with an antibody against Na/K-ATPase, a membrane protein without a histidine tract. The eluates of the two groups were shown to contain a much lower concentration of Na/K-ATPase than the cerebellum lysates (Figure 9B), indicating that
Na/K-ATPase has very little affinity to the nickel affinity gel.
These results suggest that the 190-kDa fragment in the mouse cerebellum contains the histidine tract in the C-terminal tail. The part of the channel from the
N-terminus to the histidine track is approximately 246 kDa. Since the fragment that we observe is only 190-kDa, we can infer that if the 190-kDa fragment is a product of proteolytic processing, it is unlikely to contain the N-terminal tail. However, other mechanisms such as alternative splicing could produce a shorter fragment with both the N-terminal and C-terminal tails. Thus we also wanted to determine whether the
190-kDa fragment contained the N-terminal tail.
The eluate from the nickel affinity gel that was incubated with the lysate of WT mouse cerebellum was run on a Western blot with the II-III loop antibody and the
N-terminus antibody respectively. The 190-kDa fragment was recognized by the II-III loop but not by the N-terminus antibody (Figure 9 C and D), indicating that the
190-kDa fragment in the mouse cerebellum does not contain the N-terminal tail. Next, the lysate of cultured mouse cortical neurons was also purified using the nickel affinity gel and analyzed using Western blot with both the II-III loop antibody and the
N-terminus antibody. The result is similar to the result obtained from the mouse cerebellum. The eluate from nickel affinity gel that was incubated with the lysate of
115 cultured mouse cortical neurons had a higher concentration of the full-length 1A, as detected by both the II-III loop antibody and the N-terminus antibody (Figure 9 E and
F). The eluate also contained a higher concentration of the 190-kDa fragment than the lysate, as detected by the II-III loop antibody. The N-terminus antibody detected a band of 180-kDa in the lysate. We know that this band is non-specific according to the result in Figure 6B. In the eluate, there was only a sparse amount of this 180-kDa band. Judging from the intensity, apparent molecular weight, and shape of the band, it was unlikely for the II-III loop antibody to detect the 190-kDa fragment; instead, it was probably the residue of the non-specific 180-kDa band, which was not thoroughly washed off from the nickel affinity gel.
These results suggest that the 190-kDa fragment of mouse 1A in both the mouse cerebellum and cultured mouse cortical neurons contains the histidine tract in the C-terminal tail but not the N-terminal tail. Therefore, the 190-kDa fragment appears to contain most of the II-III loop, repeat III, III-IV loop, repeat IV and the
C-terminus.
3.7 Is the 190-kDa fragment of 1A on the plasma membrane?
The main function of voltage-gated calcium channels is to convert the electrical signal of the depolarization of the plasma membrane into an intracellular calcium signal, and this function is carried out by those channels on the plasma membrane (1).
Therefore, we wanted to know whether the 190-kDa fragment of mouse 1A exists on
116 the plasma membrane.
Cell surface biotinylation is an important method for isolating and analyzing cell surface proteins and has been widely used to study the cell surface expression of ion channels (4, 108, 158). Surface biotinylation was performed on mouse cortical slices as described in Materials and Methods. Briefly, coronal slices of mouse cerebral cortices were made and divided into two groups. One group of brain slices was subjected to biotinylation, and the other group was subjected to mock biotinylation, in which all treatments were the same as biotinylation except that no sulfo-NHS-SS-Biotin was added. The brain slices were lysed and a small portion of the lysate was set aside for the analysis of whole-cell 1A. NeutrAvidin Agarose was used to pull down the biotinylated proteins from the lysates. Both the whole-cell lysate and the eluate from the NeutrAvidin Agarose were analyzed using Western blot with the II-III loop antibody. -actin was also analyzed as both a loading control and an indicator of biotinylation of intracellular protein. In some experiments, the flow-through instead of the eluate was analyzed.
Figure 10 shows two typical results of this experiment. The eluate from the biotinylation group contained a considerable amount of the 190-kDa fragment of 1A, while in the mock biotinylation group, the 190-kDa fragment of 1A was almost or completely undetectable. Surprisingly, the whole-cell lysate or flow-through in the biotinylation group also consistently contained a higher amount of the 190-kDa fragment of 1A than the mock biotinylation group. This suggests that either
117 sulfo-NHS-SS-Biotin can increase the amount of the 190-kDa fragment of 1A in the cells, or the biotinylated 190-kDa fragment of 1A can somehow be more easily detected by the II-III loop antibody. If sulfo-NHS-SS-Biotin can increase the amount of the 190-kDa fragment of 1A in the cells, then the difference between the amounts of the 190-kDa fragment in the eluates from the two groups can be explained by non-specific binding of the 190-kDa fragment to the NeutrAvidin Agarose. If the biotinylated 190-kDa fragment of 1A can be more easily detected by the II-III loop antibody, then the difference between the amounts of the 190-kDa fragment in the eluates from the two groups can be explained by biotinylation of intracellular protein of damaged cells. A higher amount of -actin was detected in the eluate from the biotinylation group than the mock biotinylation group (Figure 10 B and D); this indicates that some cells were indeed damaged and some intracellular protein was biotinylated. With the complexity of these results, it is inclusive whether the 190-kDa fragment of 1A exists on the plasma membrane.
118
A B 1 2 3 4 kDa kDa 1 2 3 4 315 72 250 55 180 130 36
95
C D kDa 1 2 3 4 kDa 1 2 3 4 250 72 130 55 36 95
Figure 10. Western blot analysis of 1A in mouse cortical slices after surface biotinylation. Results of two independent experiments are shown in panels (A and B) and panels (C and D), respectively. Note that 10% of the eluate was loaded on the gel, while only 1% of the whole-cell lysate or flow-through was loaded. (A) Western blot with the II-III loop antibody. (1) Eluate from the NeutrAvidin Agarose of the mock biotinylation group. (2) Lysate of cortical slices of the mock biotinylation group. (3) Eluate of the biotinylation group. (4) Lysate of the biotinylation group. (B) Same samples as in (A) probed with the anti--actin antibody. (C) Western blot with the II-III loop antibody. (1) Eluate of the biotinylation group. (2) Flow-through of the biotinylation group. (3) Eluate of the mock biotinylation group. (4) Flow-through of the mock biotinylation group. (D) Same samples as in (C) probed with the anti--actin antibody.
3.8 Effect of calcium activity on the relative amount of the 190-kDa fragment of
1A
The finding that the 190-kDa fragment of mouse 1A contains the C-terminal tail but not the N-terminal tail (see 3.6), is consistent with the possibility that this fragment is caused by mid-channel proteolysis. Previous findings in our lab suggest that the mid-channel proteolysis of L-type calcium channel 1 subunit is upregulated by increased calcium activity, through the calcium-dependent protease calpain and the
119 ubiquitin-proteasome system (4). Calpain has also been reported to mediate proteolysis of the voltage-gated sodium channel subunit (158). Therefore, we wanted to determine whether the relative amount of the 190-kDa fragment of mouse
1A can be altered by increased calcium activity.
Three dishes of cultured mouse cortical neurons were prepared. The first dish was treated for 2 h with 65 mM of KCl and 14 M of Bay K8644, which can cause an increase of intracellular calcium concentration. The second dish was treated for 2 h with DMSO (vehicle of Bay K8644, 1/2000 diluted in culture medium). The third dish was not treated since it acted as a control. Then the dishes of neurons were lysed in
SDS lysis buffer at 95°C and the lysates were analyzed using Western blot.
Western blot stained with an anti--spectrin antibody (EMD Millipore,
MAB1622) showed that the neurons treated with KCl and Bay K8644 had a decreased amount of full-length (180-kDa) -spectrin and an increased amount of a 150-kDa species (Figure 11A), a known calpain/caspase 3-mediated proteolytic product (158).
Poly ADP ribose polymerase (PARP) was also analyazed. Proteolytic cleavage of the
113-kDa full-length PARP into an 89-kDa fragment has been reported to be indicative of apoptosis (231-233). As shown in Figure 11B, the anti-PARP antibody (Cell
Signaling 9542) recognized two bands of ~120 kDa and ~100 kDa, respectively, in each sample. The intensity ratio of the 100-kDa band relative to the 120-kDa band in the group treated with KCl and Bay K8644 is about twice as high as that in the other two samples. This result indicates that the treatment with KCl and Bay K8644
120 increased apoptosis.
A B 1 2 3 kDa 1 2 3 kDa
300 180 250 130 180
130 100
70 100
C D kDa 1 2 3 1 2 3 kDa 300 300
250 250 180 180
130 130
100 100 70 70
Figure 11. Effects of calcium activity on the proteolysis of -spectrin, PARP and 1A in cultured mouse cortical neurons. The three groups of neurons were treated with KCl and Bay K8644 (1), DMSO (2) or nothing (3). (A) Western blot with the anti--spectrin antibody. (B) Western blot with the anti-PARP antibody. (C) Western blot with the II-III loop antibody. (D) Western blot with the N-terminus antibody.
Interestingly, staining with the II-III loop antibody shows that the intensity ratio of the 190-kDa fragment relative to the full-length 1A is lower in the DMSO treated neurons than in both the other two groups (Figure 11C). This result signifies that increased calcium activity can increase the ratio of the amount of the 190-kDa fragment to that of the full-length 1A, while DMSO has an effect of decreasing that ratio. Western blot stained with the N-terminus antibody also showed a higher amount
121 of full-length 1A in the DMSO treated group than in the other two groups (Figure
11D). However, the difference is not as dramatic as in the Western blot stained with the II-III loop antibody. Because of time constraints, this experiment was performed only once, so the effect of calcium activity on the relative amount of the 190-kDa fragment is not completely conclusive.
3.9 Possible alternative splicing of mouse 1A
Numerous splice variants of 1A have been discovered, but none can account for an isoform as small as 190-kDa (see 1.7.2.5). Nevertheless, it cannot be ruled out that
190-kDa species is due to an undiscovered type of alternative splicing. Given that the
190-kDa species cannot be recognized by the antibody against the N-terminal tail, the
190-kDa species either is missing the whole N-terminal tail or has only part of it, which is too small to be recognized by the antibody. The N-terminal tail of 1A is almost entirely encoded by exon 1 of CACNA1A (See NCBI Reference Sequence
NM_007578.3 and UniProt P97445). Therefore, if the 190-kDa species is a splice variant, the splice donor site (5’ splice site) should be in exon 1, either upstream of or shortly downstream of the start codon. In addition, the acceptor site (3’ splice site) should be upstream of exon 17, 18 or 19, which encodes the proximal part of the II-III loop. We know that acceptor sites are present in the above-mentioned areas, since the splicing of the full-length 1A mRNA relies on them. However, it is unknown whether there are splice donor sites within exon 1.
122
To identify potential splice donor sites in exon 1 of CACNA1A, the nucleotide sequence of exon 1 was analyzed using the method of Shapiro and Senapathy (234,
235). A sequence of 9 nucleotides (9-mer) with a GT at fourth and fifth positions is highly conserved at the splice donor site. A consensus value (CV) score is calculated for every 9-mer in exon 1 of CACNA1A with a GT at fourth and fifth positions based on the following formula:
ffini,,min CV ii, ff iiii,max,min
In this formula, i = -3, -2, -1, +1, +2, +3, +4, +5 and +6 (positions in the 9-mer);
fi,min is the frequency (in 126,480 known human splice donor sites, similarly hereinafter) of the rarest nucleotide (among A, T, G and C) at position i. Additionally,
fi,max is the frequency of the most common nucleotide at position i, and fi,n is the frequency of the nucleotide at position i in the 9-mer starting from nucleotide n in exon 1 of CACNA1A. The maximum CV score that a 9-mer can have is 1.
The CV scores of 24 9-mers with a GT at fourth and fifth positions are displayed in Figure 12. All the 9-mers downstream of the start codon have CV scores under 0.6.
Two 9-mers upstream of the start codon have CV scores higher than 0.6 (0.68 and
0.66, respectively). Exon 1 of CACNA1A was also analyzed using the MaxEntScan method of Yeo and Burge (236). The two 9-mers with the highest CV scores also have the highest MaxEnt scores (Figure 12). Therefore, these two 9-mers are the best candidates for potential splice donors.
123
Figure 12. CV and MaxEnt scores of potential splice donor sites in exon 1 of CACNA1A. The brown regions of each bar indicate where MaxEnt and CV overlap.
The distribution of the CV scores of the 126,480 known human splice donor sites is shown in Figure 13. As can be seen in the figure, only a small percentage of the total 126,480 donor sites have CV scores below 0.7. Among all the 126,480 donor sites, only 1,444 (1%) have a CV score of 0.70. Therefore, we can infer that the probability that the above-mentioned two 9-mers are spice donors is very low.
Figure 13. Distribution of the CV scores of 126,480 known human splice donor sites. The curve was smoothed using 5-point moving average.
124
3.10 Discussion
3.10.1 The presence of the 190-kDa species in live brains and cultured neurons
We detected a 190-kDa form of 1A in mouse brain and cultured mouse cortical neurons. The 190-kDa species is recognized by the antibody against a 17-amino acid segment (CNA1) in the II-III loop of 1A, and was shown to contain the C-terminal tail but not the N-terminal tail. Similar short forms of 1A in animal brains have been reported. Westenbroek et al. reported a result very similar to ours: an anti-CNA1 antibody recognized a very faint 250-kDa band and a 190-kDa major band in rat brain membrane glycoprotein preparation (9). Leenders et al. reported a 210-kDa band
(major form) and a 190-kDa band recognized by an anti-CNA1 antibody mouse forebrain synaptosome lysate (10). The apparent molecular weights of proteins in
SDS-PAGE are often inaccurate because of anomalous migration of the protein samples and/or markers. Thus the two bands reported by Leenders et al. might be the same species as the 190-kDa doublet band that we found.
Scott et al. used an antibody against a long epitope that has 191 amino acid residues and includes CNA1 and flanking regions, and detected two major bands of
95 kDa and 190 kDa, respectively (5). They identified the 95-kDa species as a two-domain truncated form of 1A including repeat I, repeat II and part of the II-III loop (5). The 190-kDa species, which was less discussed by the authors, may be the complementary fragment containing repeat III and repeat IV, i.e., the same fragment as the 190-kDa fragment found in our experiments.
125
Similar short forms of 1A have also been reported to be recognized by antibodies against the C-terminal tail of 1A, in accordance with our finding that the
190-kDa species contains the C-terminal tail. Kordasiewicz et al. reported a 220-kDa band and a 170-kDa band of similar intensities from the mouse cerebellum recognized by an antibody against part of the C-terminal tail of 1A (11). Using a similar antibody, Ishiguro et al. found a 160-kDa strong band in the human cerebellum, along with additional bands including a 250-kDa faint band, which may be the full-length
1A (12).
Several groups tried to address the question whether the short forms of 1A are present in live brains or arise from proteolytic degradation during sample preparation.
Scott et al. reported that the banding pattern (with two major bands of 95-kDa and
190-kDa, respectively) is highly reproducible in different preparations (5). As stated above and in 1.7.2.1, short forms of 1A of 190 kDa or similar sizes recognized by antibodies against the II-III loop or the C-terminal tail were observed by many groups using different preparations from brains of different animals. These observations support the likelihood that the 190-kDa species is present in live brains. It is unlikely that so many conditions of sample preparation all happen to activate the same mechanism of proteolytic degradation or different mechanisms that produce fragments of similar sizes. Other lines of evidence against the likelihood of proteolytic degradation during sample preparation have also been reported. Scott et al. reported that 1A was detected as multiple bands, while 1B, which has 65% sequence
126 homology to 1A in the same preparation, was not (5). Two independent studies reported that inclusion of high concentrations of several protease inhibitors and handling of samples at 0°C did not alter the relative quantities of the bands, including a 190-kDa band (5, 9).
In our experiments, sample preparation (except sample preparation with SDS lysis buffer at high temperature) was always performed on ice or at 4°C in a lysis buffer containing high concentrations of protease inhibitors, including inhibitors of serine proteases, cysteine proteases, aminopeptidases, and aspartyl peptidases. To further decrease possible proteolytic degradation during sample preparation, homogenization with SDS lysis buffer at 95°C was performed. Hot SDS buffer has been shown to effectively minimize protein degradation during sample preparation
(237). In addition, Nucleolin, an extremely labile protein, has been shown to undergo autocatalyzed proteolytic degradation in nondividing T cells (237). When T cell samples were prepared in Triton X-100 buffer on ice, the degradation was so pronounced that the 105-kDa full-length nucleolin was almost completely invisible and short fragments of nucleolin were the most intense (237). However, when the samples were prepared in 1% SDS buffer at 100°C, the 105-kDa full-length nucleolin had the most intense band while the short forms had less intense bands (237). These results signify that the degradation of nucleolin continues during sample preparation in Triton X-100 buffer, while hot SDS buffer can better preserve the in vivo steady-state level of nucleolin. Heat itself also can effectively reduce protein
127 degradation. It has been shown that heat stabilization of mouse brain tissue at 95°C within 1 min after decapitation effectively reduces protein degradation and preserves the proteome of the sample better than snap freezing methods, even when protease inhibitors are used (238). Therefore, we expect the hot SDS lysis buffer to protect proteins much better than the Pierce IP Lysis Buffer. The amount of the 190-kDa species relative to the full-length 1A was not decreased by the hot SDS lysis buffer
(Figure 8 A and C), suggesting that the 190-kDa species is not due to degradation of the full-length 1A during sample preparation. Compared with Pierce IP Lysis Buffer, hot SDS lysis buffer even increased the relative amount of the 190-kDa species in the lysate of cultured mouse cortical neurons (Figure 8C). This may be because the
190-kDa species is subjected to faster proteolytic degradation than the full-length
1A, in accordance with the findings that truncated and full-length forms of proteins have different rate of degradation (239-241). Sometimes a truncated version of a protein undergoes faster proteasomal degradation than the full-length version, most likely because it is quickly detected as non-functional (242). However, it is doubtful whether this is what happens to the 190-kDa fragment of 1A. Our results and many other previous results (see 1.7.2.1) show that the 190-kDa fragment of 1A had the most intense band, while the full-length 1A had a lesser intense band. Thus it would be surprising if the 190-kDa is simply a non-functional degradation product of 1A.
Nevertheless, it is possible that the 190-kDa is indeed a non-functional degradation product present mainly in the cytoplasm, while most 1A on the cell surface is in the
128 full-length form.
It is interesting that 1A is processed into the 190-kDa form only in the brain and in cultured neurons, but not in heterologous systems such as Xenopus oocytes and
HEK 293T cells. One possible explanation for this is that neuron-specific factors, e.g., proteases, are necessary for the production of the 190-kDa species. It has been reported that endogenous 1S and1C in muscle and brain tissues are present in two size forms (one missing part of the C-terminal tail), while those expressed in heterologous systems are only present in the full-length form (see 1.6.1.1).
Moreover, our lab has found that 1C subunits in rat brain and cultured rat hippocampal neurons are subjected to mid-channel proteolysis (4, 50), but those expressed in Xenopus oocytes and HEK 293T cells are not (50). Calpain has been shown to be involved in the mid-channel proteolysis of rat brain 1C (4), and also in the proteolysis of the C-terminal tail of skeletal muscle 1S (135) and brain 1C
(136). Calpain is present in Xenopus oocytes and HEK 293 cells. In Xenopus oocytes, calpain has been found to play an important role in osmostress-induced apoptosis
(243). In HEK 293 cells, calpain has also been shown to be an important participant in apoptosis (244). It would be interesting to investigate whether activation or overexpression of calpain in Xenopus oocytes and HEK 293 cells can cause proteolysis of the 1 subunits of L- and P/Q-type calcium channels. Alternatively, we could try to knockdown calpastatin, an endogenous calpain inhibitor present in
Xenopus oocytes (245) and HEK 293 cells (246). However, a negative result does not
129 mean that calpain does not participate in the proteolysis of these channels, because the heterologous expression systems may lack other indispensible factors that are expressed in neurons. Alternatively, the 190-kDa species of 1A may not be a product of proteolysis at all, but a product of a pre-translational process, which is present in neurons but not the heterologous systems.
Another interesting phenomenon is that in the mouse brain, the full-length 1A is almost undetectable and the 190-kDa form is predominant, while in cultured mouse cortical neurons the levels of the full-length form and the 190-kDa form are comparable. Neurons in the brain and neurons in culture are exposed to drastically differently environments. Neurons in the brain are surrounded by a large number of glial cells. Astrocytes, comprising about half of the volume of the adult mammalian brain, are the most predominant glial cell type (247). Astrocytes not only serve supportive and trophic functions for the neurons, but also display excitability and communicate with neurons (247). Astrocytes regulate the extracellular environment of neurons, offer metabolic support to neurons, and release growth factors such as nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF) and fibroblast growth factor 2 (FGF-2) (248).
It is not surprising that neurons in the culture environment, where no astrocytes play these supportive roles, display physiological differences from neurons in the brain. A typical example of this is that neurons in the brain can live as long as the organism lives (249), while cultured neurons usually survive for no more than a few
130 weeks (250). It is possible that under the trophic and metabolic support of glial cells, the mechanism for the production of the 190-kDa species is more engaged in neurons in the brain than in neurons in culture. It would be interesting to look at the relative amount of the 190-kDa species in neuron-astrocyte co-culture and see whether astrocytes influence the production of the 190-kDa species.
3.10.2 Identity of the 190-kDa species
Our results indicate that the 190-kDa form of 1A contains the part of the II-III loop equivalent to CNA1 in rat 1A (residues 865-881 in mouse 1A, UniProt
P97445) and the histidine tract (residues 2162-2171) in the C-terminus, but does not contain the N-terminal tail. Thus, we hypothesize that the 190-kDa species is an
N-terminally truncated 1A, starting somewhere in the II-III loop before residue 865.
In fact, in order for the fragment to be exactly 190-kDa, it would need to start in S6 of repeat II. However, we think that this fragment is likely a product of mid-channel proteolysis (see 3.10.4), which is more likely to happen in an intracellular loop.
Therefore, we think the fragment starts in the II-III loop, making it a few kDa smaller than 190-kDa.
Note that the histidine tract is 197 residues away from the C-terminal end. We do not have evidence that these 197 residues are present in the 190-kDa species. If the
190-kDa species has a truncation right after the histidine tract, it would need to start from S1 of repeat II to have the exact molecular weight of 190 kDa. Because
131 mid-channel proteolysis is more likely to happen in an intracellular loop, if the
190-kDa species is a product of mid-channel proteolysis, it is more likely to start somewhere in the I-II loop and end right after the histidine tract. The resulting cleavage would make it a few kDa larger than 190-kDa. Moreover, it is also possible that the 190-kDa species is simply an N-terminally truncated 1A starting somewhere near the end of the I-II loop. This will make its molecular weight at least 213 kDa.
Actually when the 190-kDa is detected as a doublet band, the molecular weight of the upper band is 205-210 kDa (Figure 8A and 9E). Moreover, because the apparent molecular weights of both the marker and the samples may be inaccurate, it is possible that our “190-kDa” species has a real molecular weight of 213 kDa.
Taken together, the 190-kDa species may include either repeats III and IV or repeats II, III, and IV. We favor the former possibility for the following reasons. If the
190-kDa fragment is a product of mid-channel proteolysis in the II-III loop, the proteolysis should also produce a complementary fragment of 80-kDa. Western blot analyses with the antibody against the N-terminus of 1A revealed a band slightly above 100-kDa in the lysate of cultured mouse cortical neurons (Figure 8D), as well as also in the lysate of mouse cerebellum (data not shown).
Considering that the apparent molecular weights of the marker and the protein samples might be inaccurate, this 100-kDa species might be the complementary fragment. In fact, when compared with another protein marker (BioRad #1610373), this species is closer to 90 kDa. Moreover, the apoptosis-related proteolytic product of
132
PARP has been reported to be 89-kDa (231-233) (also see UniProt Q921K2), but it appears to be 100 kDa in our Western blot analysis, indicating that the apparent molecular weight of the 100-kDa marker band is inaccurate. Taken together, our
N-terminus antibody and II-III loop antibody seem to detect a complementary pair of truncated 1A, which may be due to mid-channel proteolysis in the II-III loop.
On the other hand, if the 190-kDa species is produced by a pre-translational process, e.g., alternative slicing, it is also more likely to contain repeats III and IV instead of repeats II, III, and IV. Evidence suggests that during evolution, the four-domain sodium and calcium channels originated from a single domain ancestral channel, which would undergo two rounds of gene duplication (180, 181). In modern organisms, expression of the two-domain isoforms still occurs through alternative splicing (180, 182), probably as an evolutionarily throwback. Therefore, the 190-kDa species may also be a two-domain splice isoform as an evolutionarily throwback, although no known splice variants of 1A can account for a fragment-channel as small as 190-kDa (see 1.7.2.5).
As for whether the 190-kDa species contains the last 197 residues distal to the histidine tract, although we do not have direct evidence for that, previous studies reported that antibodies against epitopes distal to the histidine tract detected species of molecular weights around 190-kDa (160-220 kDa) in mouse and human cerebellum lysates (11, 12). There is no evidence that the C-terminal tail of 1A is subjected to proteolytic cleavage. Evidence shows that C-terminal fragments of L-type calcium
133 channels can be produced through proteolysis (see 1.6.1.1) and cryptic promoters (see
1.7.1). C-terminal fragments of 1A, however, have only been shown to be produced by an internal ribosome entry site (IRES) (see 1.7.2.3). Therefore, we surmise that the
190-kDa species that we found contains the complete C-terminal tail.
More experiments would be needed to confirm the identity of the 190-kDa form of 1A. Immunoblot analyses with antibodies against the I-II loop and the C-terminus would be able to address the questions raised above. Moreover, it would be interesting to analyze the 190-kDa species using N-terminal sequencing (5, 51) or mass spectrometry (130) to not only confirm what parts of the channel it contains but also identify its N-terminal sequence. This will give us a clue to the possible origin of the
190-kDa species. For example, a methionine residue on the N-terminus would suggest that the 190-kDa species might be the product of an independent translational event due to alternative splicing, IRES, or a cryptic promoter. A different residue on the
N-terminus would indicate that the 190-kDa species has undergone proteolytic processing, but the other three possibilities cannot be ruled out because a truncated
1A produced by alternative splicing, IRES, or a cryptic promoter might further undergo proteolytic processing.
The 190-kDa species in mouse brain tissue is almost always detected as a doublet band (Figure 5, 8A, 9A and 9C), while that in cultured mouse cortical neurons is sometimes detected as a doublet band and sometimes almost as a single band (Figure
6A and 9E) but with a smear to the lower molecular weight side (Figure 8C and 11C).
134
The apparent molecular weights of the two bands in the doublet band are approximately 175 kDa and 210 kDa, respectively, in both mouse cerebellar tissue
(Figure 8A) and cultured mouse cortical neurons (Figure 6A). When it is detected almost as a single band, it shows an apparent molecular weight of ~190 kDa (Figure
11C). Our protein samples were all denatured with the presence of 50-100 mM of dithiothreitol (DTT), which is much more concentrated than in most previously reported immunoblot analyses of the P/Q-type channel (8, 9, 12). Therefore, it is unlikely that the doublet band is caused by insufficient reduction of disulfide bonds.
The doublet band is probably due to one or more of the following reasons: different post-translational modifications, proteolysis at different sites, and alternative splicing
(251).
As discussed in 1.5.1.2, CaV2.1 can be regulated by PKA- and PKC-mediated
phosphorylation. PKC has been shown to modulate CaV2 through phosphorylation of the I-II loop and the synprint site in the II-III loop (1). Phosphorylation decreases the mobility of proteins in SDS-PAGE (252, 253). The doublet band of 1A may correspond to a phosphorylated and a non-phosphorylated form, or a more phosphorylated and a less phosphorylated form of truncated 1A. Analysis of the mouse 1A peptide sequence (UniProt P97445) using NetPhos 3.1 (254, 255) predicts a number of phosphorylation sites distal to repeat II, including 15 PKA sites (2 in the
II-III loop, 7 in the C-terminal tail), 29 PKC sites (11 in the II-III loop, 11 in the
C-terminal tail) and 1 CaMKII site (in the C-terminal tail). Phosphorylation of any of
135 these sites may increase the apparent molecular weight of fragment-channel. In theory, each phosphate group adds 80 Da to the molecular weight (256). In reality however, the degree of shift in mobility is hard to predict. It has been reported that one phosphate group could cause a 2-kDa shift (252). The molecular weight difference between the two bands in out 1A doublet band is about 35 kDa. It is possible that this molecular weight difference is caused by phosphorylation of some of the putative phosphorylation sites. Note that the lysate of cultured mouse cortical neurons prepared using hot SDS lysis buffer shows almost only a single 1A band (Figure 8C), while that prepared using Pierce IP Lysis Buffer shows a doublet band (Figure 6A). A possible reason for this phenomenon is that after cells were lysed using Pierce IP
Lysis Buffer, some of the 190-kDa fragment-channels were dephosphorylated by phosphatases, producing the lower molecular weight band. On the other hand, in the hot SDS lysis buffer, phosphatases were denatured immediately and almost all the
190-kDa fragment-channels were kept in their phosphorylated form.
Phos-tag polyacrylamide gels (NARD Institute Ltd.) can be used to explore the phosphorylation status of the doublet band. Phos-tag is a metal complex that binds specifically phosphates (257). In SDS-PAGE with Phos-tag-conjugated polyacrylamide, the mobilities of phosphorylated proteins are significantly slowed down compared with the corresponding dephosphorylated proteins (253, 257). If the doublet band of 1A is due to phosphorylation, on a Phos-tag polyacrylamide gel the mobility of one or both of the two bands should be slowed down, and the distance
136 between the two bands should be increased, compared with that on a regular polyacrylamide gel. Moreover, we can add phosphatases into the lysis buffer to increase dephosphorylation (257) or phosphatase inhibitors to suppress dephosphorylation (253). If the doublet band is due to dephosphorylation of some of the 190-kDa fragment-channels, phosphatases should decrease the abundance of the upper band and increase that of the lower band, while phosphatase inhibitors should do the opposite. Sites of phosphorylation can be identified using N-terminal sequencing (51) or mass spectrometry (258).
Another common type of post-translational modification that may account for the doublet band is glycosylation. Scott et al. reported that a 95-kDa C-terminally truncated form of rabbit 1A is glycosylated (5). This 95-kDa fragment contains the
N-terminus, repeats I and II and part of the II-III loop (5), and is possibly the fragment complementary to our 190-kDa fragment produced by mid-channel proteolysis. Scott et al. identified two potential glycosylation sites in rabbit 1A, N283 (in the S5-S6 loop of repeat I) and N1665 (in the S3-S4 loop of domain IV); with the former being present in the 95-kDa fragment (5). The two sites are conserved in mouse 1A as
N285 and N1607, respectively (UniProt P27884 and P97445). Analysis of mouse 1A peptide sequence (P97445) by NetNGlyc 1.0 (255) identified 13 potential N-linked glycosylation sites, among which only N285 may really be glycosylated because it is extracellular. Interestingly, N1607 is not identified as a potential glycosylation site by
NetNGlyc 1.0. However, another glycosylation site prediction tool, GlycoEP (259),
137 recognized N1607 as a potential glycosylation site. N1607 may be glycosylated, causing an increase in the molecular weight and resulting in the upper band in the doublet band. To investigate glycosylation of the 190-kDa species, peptide
N-glycosidase and/or endoglycosidase F can be used to treat the protein samples (5,
260, 261). If the upper band in the doublet band is due to glycosylation, treatment with the enzymes should cleave off the oligosaccharides, causing a decrease in the abundance of the upper band and an increase in that of the lower band.
Ubiquitination can also alter the mobility of a protein in SDS-PAGE.
Ubiquitination usually results in ladder-like protein bands or smears with increased molecular weight (84, 114). Our 190-kDa species, however, has a strong and clear upper band. A smear sometimes is present on the lower molecular weight side, instead of the higher molecular weight side (Figure 8C and 11C). Therefore, the doublet band is unlikely to result from ubiquitination. Because of time constraints, we did not do experiments to investigate the post-translational modifications of the 190-kDa species.
As discussed above, it is possible that the 190-kDa species is due to a truncation in the I-II loop, although it is more likely due to a truncation in the II-III loop. It is also possible that the doublet band corresponds to two different truncated forms of
1A, the upper band (205-210 kDa) with the truncation site in the I-II loop, and the lower band (175 kDa) with the truncation site in the II-III loop. However, we think this is unlikely, since sometimes the fragment-channel is detected almost as a single band, with an intermediate molecular weight of 190-200 kDa.
138
Another possible reason that makes the 190-kDa species a doublet band is alternative spicing. Several loci of alternative splicing have been found in and distal to the II-III loop of 1A. Among these loci, two can cause relatively big changes in molecular weight. The exon 43/44 alternative splicing and the exon 47 alternative splicing can shift the molecular weight for 6 kDa and 16 kDa, respectively (218).
Immunoblot analyses using antibodies against the regions encoded by these exons can be used to explore which splice variants the two bands in the double band belong to.
Moreover, PCR using a transcript-scanning strategy (218, 262, 263) can be used to systematically screen the cDNA of mouse 1A for alternative splicing.
It has been reported that HLA-H, (a member of the major histocompatibility complex class I family), expressed in HEK 293 cells is detected as a doublet band in
Western blot analyses, and the two bands are enriched in different fractions of the cell lysate (264). The upper 49-kDa band is mainly associated with the plasma membrane, while the lower 46-kDa band is mainly present in the Golgi complex, suggesting incomplete post-translational processing or modification of the 46-kDa species (264).
It would be interesting to use density gradient centrifugation to isolate different organelles and study the subcellular distribution of the two bands in the 1A doublet band so as to better understand the functions and fates of the two bands.
3.10.3 Is the 190-kDa species on the plasma membrane?
It is important to know whether the 190-kDa fragment of 1A is expressed on
139 the plasma membrane. Presence of the fragment-channel on the plasma membrane would suggest that the fragment-channel contributes to and/or regulates channel activities, whereas absence of the fragment-channel on the surface would suggest the possibility that the fragment-channel is merely a degradation product of unneeded channels. Unfortunately, because the biotinylation reagent sulfo-NHS-SS-biotin was found to increase either the expression level or detectability of the 190-kDa species
(see 3.7), it is inconclusive whether the 190-kDa species is expressed on the plasma membrane.
Possible influence of biotinylation on protein expression has also concerned other groups. Volz et al. (1995) analyzed the half-lives of biotinylated and non-biotinylated surface transferrin receptor (TfR) and dipeptidylpeptidase IV
(DPPIV) using pulse-chase experiments, and did not find any influence of biotinylation on the metabolic stability of these two proteins (265). To our knowledge, there has been no report of biotinylation influencing protein levels. However, it cannot be excluded that biotinylation may influence the protein level of the 190-kDa 1A fragment. In our experiments, surface biotinylation substantially increased the immunoreactivity of the 190-kDa 1A fragment, but not -actin, in the subsequent
Western blot analyses (Figure 10). The reproducibility of this result and the different responses of 1A and -actin indicate that the influence of biotinylation on the immunoreactivity of the 190-kDa 1A fragment is not an incident or artifact.
The only difference between the biotinylation group and the mock biotinylation
140 group is that sulfo-NHS-SS-biotin was added into the biotinylation group, but not the mock biotinylation group. Sulfo-NHS-SS-biotin may increase the immunoreactivity of the 190-kDa species in two possible ways. It may increase the protein level of the
190-kDa species, or it may increase the affinity of the 190-kDa species to the II-III loop antibody. Because of time constraints, we did not investigate the reason for the increase in the immunoreactivity. A simple experiment that can test the two possibilities is to replace the II-III loop antibody with an antibody against another region of 1A, e.g., the C-terminal tail. The N-terminus antibody should not be used since the 190-kDa species does not contain the N-terminus. If the new antibody also detects an increase in the immunoreactivity of the 190-kDa species in response to biotinylation, we can infer that sulfo-NHS-SS-biotin probably increases the protein level of the 190-kDa species, since it is unlikely that biotinylation of the protein can increase the affinity of two different epitopes to their antibodies.
To explore the reason sulfo-NHS-SS-biotin increases the protein level of the
190-kDa species, pulse-chase analyses can be used to measure the protein synthesis and decay rates (266) of the 190-kDa species. A rise in the protein synthesis rate may be due to an increase in the rate of translation or an increase in the mRNA level. The mRNA level can be measured by methods such as Northern blotting and real-time
PCR. The mRNA level is determined by the synthesis and decay rates of the mRNA.
The mRNA synthesis rate can be measured using transcription run-on analysis, in which newly synthesized mRNAs are labeled with radioactive nucleotides and then
141 the relative transcription rate of the mRNA of interest is measured by hybridization of the mRNA to a sequence-specific probe (267). The mRNA decay rate can be measured by inhibiting transcription using thiolutin or actinomycin D, and then measuring the amount of the mRNA remaining at different time points (268).
As will be presented in Chapter 5, hPQB2, an engineered truncated two-domain form of human 1A equivalent to the putative truncated mouse 1A corresponding to the 190-kDa species, showed a hint of expression on the plasma membrane in
Xenopus oocytes. However, this experiment did not yield completely conclusive results. This suggests the possibility that the 190-kDa species might also be expressed on the plasma membrane of neurons in the brain and in culture. However, we cannot draw a definitive conclusion because the experiment in Xenopus oocytes may not effectively reflect what happens in the neurons. Firstly, the engineered fragment-channel, hPQB2, is unlikely an exact equivalent of the 190-kDa, because we do not know the exact truncation site and had to estimate its position. Secondly,
Xenopus oocytes and mouse neurons have substantial physiological differences. For instance, in the mouse brain, the major form of 1A is the 190-kDa species and the full-length channel is only minor, whereas in Xenopus oocytes, mouse 1A is only expressed into the full-length channel (Figure 7C). Moreover, it has been reported that the ER retention machinery in Xenopus oocytes is weak for misfolded mammalian proteins (212). Therefore, it is possible that the “leaky” ER retention machinery in
Xenopus oocytes allows hPQB2 to traffic to the plasma membrane, while in the
142 mouse brain the 190-kDa fragment is recognized as a misfolded protein and retained in the ER.
Alternative methods are needed to study the expression of the 190-kDa species on the cell surface. One possible method is to isolate plasma membranes using sucrose gradient centrifugation, which has been shown to be able to effectively enrich plasma membranes of HEK 293 cells (264) and Xenopus oocytes (269). A caveat of this method is that the purified sample, albeit enriched in plasma membrane proteins, will also contain intracellular proteins from organelles such as the Golgi complex
(264), mitochondria , and lysosomes (269). Our biotinylation experiments showed that most (or all) of the 190-kDa species is intracellular (Figure 10). Therefore, the plasma membrane samples purified by sucrose gradient centrifugation will probably contain a much smaller amount of the 190-kDa species than the samples enriched for intracellular membranes. This will make the result ambiguous as to whether the
190-kDa is present on the plasma membrane, because the sample enriched for plasma membranes will also contain intracellular membranes.
Another possible method involves radioiodination of surface proteins. Surface proteins of live cells are first radioiodinated (270, 271). Next, the cells are lysed and the protein of interest is purified using immunoprecipitation, with the antibody either directly coupled to Sepharose beads or bound to protein A-Sepharose (271). Then the protein sample is analyzed using SDS-PAGE followed by autoradiography (271). In our case, the II-III loop antibody can be used to purify the 190-kDa species.
143
Eventually if the 190-kDa species can be detected by autoradiography, we can infer that it is expressed on the cell surface.
The third possible method makes use of peptide toxins that specifically bind to
1A on its extracellular surface. -agatoxin IVA is a peptide toxin that inhibits the
P/Q-type channel by altering channel gating (272). The glutamate residue at the end of S3 of repeat IV is crucial for the inhibition by the toxin (273), thus the binding site of -agatoxin IVA in1A is thought to be in the S3-S4 loop of repeat IV. Another peptide toxin, -conotoxin MVIIC, inhibits the P/Q- and N-type channels as a pore blocker (274). The precise binding sites of -conotoxin MVIIC has not been clearly identified (274). It has been suggested that -conotoxin GVIA, -conotoxin MVIIC and -agatoxin-IIIA bind to a similar site in 1B, and-conotoxin MVIIC and
-agatoxin-IIIA bind to a similar site in 1A (275). In 1B, each of the four repeats contributes to the binding of -conotoxin GVIA, especially repeat III, in which the extracellular loop between S5 and the “pore loop” is critical for the binding of the toxin (276). Therefore, -conotoxin MVIIC probably binds to similar sites in 1A.
To study surface expression of the 190-kDa species, live cells would be incubated with radioiodinated -agatoxin IVA or -conotoxin MVIIC (176). Next, the inhibitor and the channel are cross-linked using membrane impermeable cross-linkers such as BS3 (277) (for more cross-linkers, see https://www.thermofisher.com/). Then the cells would be lysed and the lysate analyzed using SDS-PAGE and autoradiography. If the 190-kDa species can be detected by autoradiography, we can
144 infer that it is expressed on the cell surface.
We can also try other biotinylation reagents, which may not have the effect of increasing 1A protein level as sulfo-NHS-SS-biotin does. Amino-oxy-biotin and biocytin hydrazide, which specifically biotinylate glycoproteins, are good candidates because they have been previously shown to lead to higher purity of plasma membrane proteins than sulfo-NHS-SS-biotin (278). Additionally, electrophysiology
(see 3.10.4) and fluorescent imaging (see Chapter 4) may also offer indirect evidence for the presence of the 190-kDa fragment-channel on the plasma membrane.
3.10.4 Origin of the 190-kDa species
The Catterall group reported that an anti-CNA3 (peptide sequence
SEPQQREHAPPREHV in the II-III loop of 1A) antibody recognizes a major band of 190 kDa and two minor bands of 220 kDa and 160 kDa, respectively, in the rat brain (8). This result has some resemblance to our findings: the 220-kDa band corresponds to the full-length channel, and the 190-kDa and 160-kDa bands resemble the 190-kDa doublet band that we found. However, their double immunoprecipitation experiments revealed that the 220-kDa band and the 190-kDa band contain both the
N-terminus and the C-terminus of 1A, indicating that these two bands result from alternative splicing (8). Our results, however, suggest that the 190-kDa doublet band contains the C-terminus but not the N-terminus. This result is compatible with several possible origins of the 190-kDa species: mid-channel proteolysis, alternative splicing,
145
IRES, and cryptic promoters. The difference between our result and the result of the
Catterall group is probably due to the different species used.
As presented in section 3.9, the two best candidates for potential splice donors in exon 1 are both upstream of the start codon. This means that the splice variants produced by these two potential splice donors would not contain any part of the
N-terminal tail. While the possibility that 190-kDa fragment of 1A is produced by alternative splicing cannot be excluded, such possibility is very small according to the analysis of consensus value (CV) scores presented in 3.9.
Preliminary results show that an increase in intracellular Ca2+ concentration increases the amount of the 190-kDa species (Figure 11C) as well as the relative amount of the 110-kDa species recognized by the N-terminus antibody, the putative complementary fragment to the 190-kDa species (Figure 11D). Similar results have
2+ been observed for CaV1.2 (4) and NaV1.2 (158). Increased intracellular Ca
concentration promotes the mid-channel proteolysis in the II-III loop of CaV1.2 (4). In
NaV1.2, mid-channel proteolysis happens in both the I-II loop and II-III loop, and both can be promoted by increased intracellular Ca2+ concentration (158). Calpain
was shown to partially participate in the mid-channel proteolysis of CaV1.2 (4) and
required for the mid-channel proteolysis of NaV1.2 (158). Therefore, it would not be
surprising if calpain-mediated mid-channel proteolysis also happens to CaV2.1 and can be promoted by increased intracellular Ca2+ concentration. However, it is also possible that the 190-kDa of 1A is not produced by proteolysis, but by alternative
146 splicing, IRES, or a cryptic promoter that can be regulated by intracellular Ca2+. More experiments are necessary to explore the origin of the 190-kDa species.
Electrophysiological recordings can be used to investigate whether the 1A is subjected to mid-channel proteolysis. Mid-channel proteolysis can either upregulate or downregulate channel activity, depending on the type of channel. Mid-channel
proteolysis of CaV1.2 reduces current amplitude (4). In contrast, proteolysis of the voltage-gated sodium channel subunit prevents normal inactivation and thus leads to persistent influx of Na+ (160-163). To examine the acute effect of mid-channel
proteolysis on the functional properties of CaV2.1, tobacco etch virus protease (TEVp) cleavage sites (with the sequence ENLYFQG) were inserted into the I-II loop and
II-III loop of 1A (4). Inside-out macropatch recordings from Xenopus oocytes showed that a 2-min treatment with purified TEVp did not abolish CaV2.1 currents.
However, it caused an irreversible left shift of the activation curve (4), in accordance with an increase in the coupling between voltage sensing and channel opening.
Two-electrode voltage clamp (TEVC) recordings in Xenopus oocytes show that two-domain truncated forms of 1A do not conducted currents when expressed individually or coexpressed (see 5.2). These results suggest that after mid-channel proteolysis, the fragment-channels can remain associated, at least temporarily, to keep functioning as a channel with altered electrophysiological properties (also see discussion in 5.5).
The experiments on engineered CaV2.1 with TEVp sites reveal the putative
147 changes in the electrophysiological properties of the channel after mid-channel proteolysis. Next we can study the changes in electrophysiological properties of
endogenous CaV2.1 after treatments that presumably promote mid-channel proteolysis
(e.g., KCl and Bay K8644 treatment). If the changes following the treatments are similar to the changes that TEVp causes on the engineered channels, we can infer that the 190-kDa fragment is likely due to mid-channel proteolysis. One caveat of this experiment is that if the treatment (e.g., KCl and Bay K8644 treatment) causes a
decrease in CaV2.1 currents, it is hard to tell whether it is caused by a change in the electrophysiological properties of the channel or internalization of the channels. In this case, single-channel recordings on cell-attached patches may be a better choice.
In single-channel recordings, a complete loss of currents would most likely be due to internalization, while changes in the current amplitude or open probability indicate other alterations in the channel. Single-channel recordings on cell-attached patches also avoid the decrease in channel activity due to rundown, which happens in inside-out patch recordings and, to a lesser extent, in whole-cell recordings (279).
We can also explore indirect evidence for mid-channel proteolysis by trying to induce mid-channel proteolysis of 1A in Xenopus oocytes and HEK 293 cells. As discussed in 3.10.1, mid-channel channel may be induced by activation or overexpression of calpain, or knockdown of calpastatin, in Xenopus oocytes and HEK
293. Another method for investigating mid-channel proteolysis is fluorescent imaging, which will be discussed in Chapter 4.
148
Additionally, the N-terminal truncation site of the 190-kDa species could be identified using N-terminal sequencing (5, 51) or mass spectrometry (130). The
N-terminal sequence of the truncated channel may help us identify the possible protease(s) that causes the mid-channel proteolysis, or reveal whether the truncated channel is the product of an independent translational event due to alternative splicing,
IRES or a cryptic promoter.
Considering the possibility that the 190-kDa species may be produced by a short transcript due to alternative splicing or a cryptic promoter, it would be useful to screen for short splice variants of 1A that may produce the 190-kDa species. Efforts have been made to find splice variants of 1A, and several splice variants have been discovered (23, 218, 224), but none can account for a truncated form as short as
190-kDa. The methods used in these studies, however, have limitations and would not be able identify the transcript that produces the 190-kDa species. Bourinet et al. identified two splice variants of 1A by screening a cDNA library of rat brain using a probe corresponding to domain I of previously cloned neuronal 1 subunits (23). This method would not be able to identify the putative transcript for the 190-kDa species, which does not contain domain I. Soong et al. systematically scanned the whole cDNA of human 1A using PCR primers, and each pair of primers spans 1-4 exons
(218). This method would not be able to ascertain the putative transcript for the
190-kDa species either, because the putative transcript is missing more than 10 exons
(see 3.9) and none of their primer pairs spans so many exons. Rajapaksha et al. used
149 nested PCR with primers targeting the sequence of rat CaV2.1 II-III loop to identify splice variants that differs in the sequence of the II-III loop (224). Because these primers do not span all the missing exons in the putative transcript for the 190-kDa species, the putative transcript would not be identified either. Because we would like to identify a transcript for an N-terminally truncated protein and find out the origin of that transcript, it is important to obtain the sequence of 5’ end of the transcript.
Therefore, 5’ RACE-PCR (280) would be an effective method. The putative transcript for the 190-kDa species would be missing at least exons 2-16 (see 3.9). A transcript with part of exon 1 would be a splice variant, while one completely without exon 1 would be a product of a cryptic promoter.
150
Chapter 4: Visualization of 1C and 1A in Cell Cultures
151
4.1 Introduction
The finding that the 190-kDa fragment of mouse 1A contains the C-terminal tail but not the N-terminal tail (see 3.6) implies the possibility that this fragment is a product of mid-channel proteolysis. Unfortunately, the surface biotinylation experiment could not answer the question whether the 190-kDa fragment exist on the plasma membrane (see 3.7). Therefore, we tried to address this question using immunofluorescence.
In a previous study of the mid-channel proteolysis of the L-type calcium channel,
“LGH3”, an 1C subunit tagged with EGFP on the N-terminus and HA on an extracellular loop of repeat III, was expressed in cultured rat hippocampal neurons.
LGH3 was then labeled with an anti-HA antibody and Alexa594 under non-permeabilizing conditions and the neurons were visualized by confocal microscopy (4). The result revealed that LGH3 went onto the plasma membrane, and in some locations red (HA) did not colocalize with green (EGFP). These red-only clusters are consistent with the hypothesis that 1C can be proteolytically cleaved in the II-III loop, and further suggests that the cleavage products can dissociate from one another on the plasma membrane. Moreover, confocal microscopy with two additional
EGFP/HA-tagged 1C subunits, LGH1 and LGH2 (HA-tagged extracellularly on repeat I and II, respectively), suggests that there might be also a cleavage site in the
I-II loop.
If the 190-kDa fragment of 1A exists on the plasma membrane, then very likely
152 it can dissociate from the complementary 80-kDa fragment, and immunofluorescence with EGFP/HA-tagged 1A will reveal the separation of fragment channels on the plasma membrane.
4.2 Fixed-cell imaging in cultured neurons under non-permeabilizing conditions
Four constructs of EGFP/HA-tagged human 1A were made. The four constructs are all tagged with EGFP on the N-terminus, and are tagged with HA extracellularly on different repeats. Thus they are named hPQGH1, hPQGH2, hPQGH3, and hPQGH4, respectively (Figure 14 A-D). Two-electrode voltage clamp recordings in
Xenopus oocytes shows that all four constructs are able to be expressed on the plasma membrane and produce currents (Figure 14E).
hPQGH3 were expressed in cultured rat hippocampal neurons and visualized by fixed-cell imaging as described in Materials and Methods. Briefly, cultured rat hippocampal neurons were transfected with hPQGH3 and 3. 36-48 h after transfection, neurons were fixed and subjected to immunocytochemistry under non-permeabilizing conditions with an anti-HA primary antibody (BioLegend 901502) and an Alexa594-conjugated secondary antibody (Life Technologies A-11005). What we expected to see was a red “ring” around each neuron that has green fluorescence, i.e., a membrane associated pattern of Alexa594, labeling the HA tag of hPQGH3 expressed on the plasma membrane. However, the staining was highly nonspecific
(Figure 15A). The fluorescence of Alexa594 appeared in neurons with and without
153 green fluorescence. Moreover, although the cells were stained under non-permeabilizing conditions, they seemed to be permeabilized. The fluorescence of
Alexa594 can be seen in the cytoplasm and sometimes even the nuclei of the neurons.
To troubleshoot, different conditions were tried in immunocytochemistry (Table 2) but could not eliminate the nonspecific staining and unwanted permeabilization. In an attempt to increase the expression level of hPQGH3 or improve the folding of hPQGH3, 2 was cotransfected with hPQGH3 and 3 in one group of neurons, and another group of neurons was kept at 30°C instead of 37°C after transfection before fixation. Neither of the conditions improved the staining, although keeping the neurons at 30°C did seem to increase the intensity of the green fluorescence (data not shown). hPQGH4 was also tried but the same problems arose. LGH3, previously shown to display distinct membrane-associated pattern of staining (4), was used as a positive control. The staining of LGH3 is much more specific and it showed a hint of membrane-associated pattern, although there still seemed to be some permeabilization
(Figure 15B).
154
A B
C D
E 50
A 40
30 20 10
current/- peak 0
hPQ hPQGH1hPQGH2hPQGH3hPQGH4
Figure 14. Four EGFP/HA-tagged human 1A constructs. Green square represents the EGFP on the N-terminus of the channel. Red dot represents the extracellular HA tag. (A-D) Membrane topology of hPQGH1, hPQGH2, hPQGH3 and hPQGH4, respectively. (E) Currents conducted by the wild-type human P/Q channel (hPQ) and the four tagged P/Q channels, measured by TEVC in Xenopus oocytes. Data are represented as mean ± SEM.
155
A EGFP Alexa594 Overlay Bright field
B EGFP Alexa594 Overlay Bright field
Figure 15. Fixed-cell imaging in cultured rat hippocampal neurons under non-permeabilizing conditions. (A) Neurons transfected with hPQGH3 labeled with an anti-HA primary antibody and an Alexa594-conjugated secondary antibody. (B) Neurons transfected with LGH3 labeled with the same antibodies. Note that the Alexa594 signal is more associated with the plasma membrane (arrow), and a non-transfected neuron does not have EGFP or Alexa594 signal (arrowhead).
Table 2. Different conditions tried for fixed-cell imaging with cultured neurons under non-permeabilizing conditions. Fixation conditions 2% paraformaldehyde solution 4% paraformaldehyde solution Quenching after fixation Quenching with 0.1 M glycine solution No quenching Blocking conditions 0.5% fish gelatin and 2% goat serum in PBS 10% goat serum in PBS Wash between primary and secondary Wash with PBS incubations Wash with PBS with 300 mM extra NaCl Auxiliary subunits 3 3 + 2 Culture temperature 37°C 30°C after transfection
156
4.3 Fixed-cell imaging in HEK 293T cells
The experiments in 4.2 could not lead to a conclusion whether hPQGH3 can be expressed on the plasma membrane, since the red signal in the cytoplasm might have concealed the tiny amount of signal from the plasma membrane. Thus HEK 293T cells were tried for immunocytochemistry with hPQGH3 and other tagged 1A constructs. Although we know that mouse 1A is not expressed into the 190-kDa fragment in HEK 293T cells (Figure 7A), immunocytochemistry with HEK 293T might give us a clue whether the tagged 1A constructs can go to the plasma membrane and be recognized by the antibody.
LGH3, hPQGH1-4, and hPQGH4x2 (a tagged human 1A with one more extracellular HA tag in repeat IV than hPQGH4), were transfected into different groups of HEK 293T cells along with 3 and 2. hPQGN, a human 1A subunit tagged with EGFP on the N-terminus without HA tags, was also cotransfected with 3 and 2 as a negative control. Immunocytochemistry was done as described in
Materials and Methods, with the same primary and secondary antibodies as in 4.2. All the tagged channel constructs were expressed, as indicated by the green fluorescence in some of the cells of each group. LGH3 showed distinct membrane-associated pattern of staining (Figure 16A). However, none of the tagged 1A constructs showed such pattern (Figure 16B). The EGFP/HA-tagged 1A constructs had very low fluorescence of Alexa594 in general, just like hPQGN. The high background staining and unwanted permeabilization seen in cultured neurons were not seen in HEK 293T
157 cells. These results indicate that the tagged P/Q channels expressed in HEK 293T cells either cannot adequately traffic to the plasma membrane or cannot be stained on the plasma membrane because of steric hindrance.
A EGFP Alexa594 Overlay
B EGFP Alexa594 Overlay
Figure 16. Fixed-cell imaging in HEK 293T cells under non-permeabilizing conditions. (A) HEK 293T cells transfected with LGH3 labeled with an anti-HA primary antibody and an Alexa594-conjugated secondary antibody. Note the plasma membrane-associated staining pattern. (B) HEK 293T cells transfected with hPQGH3 showed poor labeling of the HA tag. hPQGH1, hPQGH2, hPQGH4, and hPQGH4x2 showed similar results.
158
4.4 Live-cell antibody feeding in cultured neurons
The lack of staining of the tagged P/Q channels in HEK 293T cells cannot rule out the possibility that these tagged channels can go to the plasma membrane and be stained on the plasma membrane in culture neurons. However, the high background staining and unwanted permeabilization hinders further experiments in culture neurons with fixed-cell imaging. Thus the live-cell antibody feeding method was implemented to visualize tagged P/Q channels.
Two groups of cultured rat hippocampal neurons were transfected with hPQGH3 and hPQGH4, respectively, along with 3. The positive control group was transfected with 3 and L-SEP2-HA6, an 1C subunit tagged with a super-ecliptic pHluorin (SEP) extracellularly on repeat II and two HA tags extracellularly on repeat III. 36-48 h after transfection, live-cell antibody feeding was done as described in Materials and
Methods. Briefly, live neurons were incubated with the anti-HA primary antibody before being washed, fixed and stained with the Alexa594-conjugated secondary antibody under non-permeabilizing conditions. All the tagged channels were able to be expressed in some neurons, as indicated by the green fluorescence. L-SEP2-HA6 showed a distinct membrane-associated pattern of staining on the neurons with green fluorescence (Figure 17A). hPQGH3 and hPQGH4, however, had no staining at all
(Figure 17B).
159
A EGFP Alexa594 Overlay Bright field
B EGFP Alexa594 Overlay Bright field
Figure 17. Live-cell antibody feeding in cultured rat hippocampal neurons. (A) Neurons transfected with L-SEP2-HA6 labeled with an anti-HA primary antibody and an Alexa594-conjugated secondary antibody. Note the plasma membrane-associated staining pattern and the absence of staining in non-transfected neurons. (B) Neurons transfected with hPQGH3 stained with the same antibodies. Note that the HA tag cannot be detected. hPQGH4 showed a similar result.
4.5 Fixed-cell imaging in cultured neurons under permeabilizing conditions
Neither fixed-cell imaging nor live-cell antibody feeding enabled us to visualize the tagged 1A subunits on the plasma membrane of culture rat hippocampal neurons.
Do the tagged 1A subunits not traffic to the plasma membrane, or are they not able to be stained by the anti-HA antibody because of steric hindrance? To address these questions, two groups of cultured rat hippocampal neurons were transfected with
LGH3 and hPQGH3, respectively, and subjected to immunocytochemistry under permeabilizing conditions with the same antibodies as mentioned above. This time, the patterns of staining in LGH3 and hPQGH3 are very similar. Although there is some nonspecific staining, especially in the nuclei, neurons with green fluorescence
160 consistently show stronger staining in the cytoplasm than the neurons without green fluorescence (Figure 18). This indicates that hPQGH3, like LGH3, can be stained by the anti-HA antibody at least under permeabilizing condition.
A EGFP Alexa594 Overlay Bright field
B EGFP Alexa594 Overlay Bright field
Figure 18. Fixed-cell imaging in cultured rat hippocampal neurons under permeabilizing conditions. (A) Neurons transfected with LGH3 labeled with an anti-HA primary antibody and an Alexa594-conjugated secondary antibody. (B) Neurons transfected with hPQGH3 stained with the same antibodies. Note that transfected neurons (neurons with EGFP signal) have stronger Alexa594 signal than non-transfected neurons do.
4.6 Discussion
The imaging experiments with tagged 1A were performed to address two questions: 1) whether the 190-kDa fragment is expressed on the plasma membrane of neurons, and 2) whether the 190-kDa species dissociate from the complementary
80-kDa fragment on the plasma membrane. Unfortunately, the tagged 1A constructs did not work well in the imaging experiments.
In our fixed-cell imaging experiments under non-permeabilizing conditions and live-cell antibody feeding experiments, cell cultures transfected with tagged 1C
161 constructs consistently showed specific labeling on the plasma membrane (Figure 15B,
16A and 17A), while those transfected with tagged 1A constructs did not show a membrane-associated pattern of staining (Figure 15A, 16B and 17B). This difference between tagged 1C and tagged 1A has been reported in previous studies.
Watschinger et al. engineered a human 1A construct tagged with EGFP on the
N-terminus and two consecutive HA tags in the extracellular loop adjacent to S5 in repeat IV, (only 2 residues away from where we inserted the HA tag in hPQGH4).
They applied to live-cell antibody feeding experiments in culture mouse hippocampal neurons using a protocol similar to ours, and no staining of the channel by anti-HA antibodies was observed (281). In contrast, an extracellularly HA-tagged 1C could be stained with the same protocol (282). Another group reported that the above-mentioned tagged 1A construct “was not well exposed” on the surface of
Neuro2A cells. However, an extracellularly HA-tagged 1B construct (with two consecutive HA tags in S3-S4 loop of repeat II) (283) could be stained by an anti-HA antibody in a membrane-associated manner in fixed Neuro2A cells under non-permeabilizing conditions (190).
Our fixed-cell imaging experiments under permeabilizing conditions clearly indicates expression of both the tagged 1C and tagged 1A constructs in cultured rat hippocampal neurons (Figure 18). If the tagged 1A constructs are expressed, why can they not be stained on the plasma membrane like tagged 1C and 1B constructs?
There are two possible reasons: 1) the tagged 1A does not traffic to the plasma
162 membrane, and 2) the HA tags in the tagged 1A constructs cannot be stained under the conditions used. Expression of tagged 1A on the plasma membrane can be confirmed using whole-cell patch recordings. The Tsien group reported that HEK 293 cells transfected with human 1A (with 1C and 2) displayed robust whole-cell
2+ Ba currents (~370 pA/pF), while those transfected with a nonfunctional mutant 1A displayed no detectable whole-cell Ba2+ currents (211). The Tsien group also reported that cultured 1A-knockout mouse hippocampal neurons transfected with human 1A displayed robust whole-cell Ba2+ currents (~110 pA/pF), of which more than half was conducted by P/Q-type channels (211, 284). They also showed that in cultured wild-type mouse hippocampal neurons, heterologous expression of human 1A doubled the whole-cell Ba2+ currents to ~150 pA/pF, of which about half was conducted by P/Q-type channels. However, in control neurons, which were transfected with EGFP, only ~1/5 of the currents were conducted by P/Q-type channels (211). These results indicate that heterologously expressed 1A subunits in
HEK 293 cells and cultured mouse hippocampal neurons traffic to the plasma membrane and form functional channels. Note that in cultured mouse hippocampal neurons, the functional expression of human P/Q-type channels on the plasma membrane does not required extra auxiliary subunits (211, 284). The P/Q-type channel plays an essential role in synaptic transmission (1, 17), and it has been reported that the expression level of the P/Q-type channel is higher in the dendrites than in the somata of cerebellar Purkinje cells (42, 43). Thus it is reasonable to
163 assume that some, if not all, of the heterologously expressed P/Q channels are expressed on the plasma membrane of axon terminals and dendrites. Then can any heterologously expressed P/Q channels traffic to the surface of the somata?
Watschinger et al. reported that the extracellularly HA-tagged human 1A could be labeled by anti-HA antibodies in cultured mouse hippocampal neurons after fixation under non-permeabilizing conditions. Yet, intracellular proteins such as synapsin and microtubule-associated protein 2 (MAP2) could also be stained by corresponding antibodies under the same conditions (281). These results suggest that the neurons were actually permeabilized, probably by paraformaldehyde during fixation, and the labeled 1A subunits might be intracellular instead of on the surface.
The Tsien group reported that a human 1A construct with an extracellularly HA tag could be labeled by anti-HA antibodies in HEK 293 cells (coexpressing 1C and 2) under non-permeabilizing conditions after fixation, and the labeling displayed a membrane-associated pattern (225). A control experiment showed that the intracellular protein calmodulin (CaM) could not be stained by an anti-CaM antibody under the same condition (225). These results offer convincing evidence that heterologously expressed tagged human 1A can be expressed and stained on the surface of HEK 293 cells. Interestingly, our hPQGH2 construct, which has an extracellular HA tag in the exact same position as the HA tag in the construct used by the Tsien group, could not be stained in HEK 293T cells under non-permeabilizing conditions after fixation. The discrepancy is probably due to methodological
164 differences such as the different subunits used (1C vs. 3) and the different methods of primary antibody incubation (4°C overnight vs. room temperature for 1 h).
There is no doubt that heterologously expressed 1A constructs can be expressed in the cytoplasm of the somata of cultured hippocampal neurons from both mouse
(281) and rat (Figure 18). Additionally, there is no doubt that heterologously expressed 1A constructs can form functional channels on the plasma membrane of the neurons (211, 284). However, whole-cell patch recordings cannot reveal in which parts of the neuron the channels are expressed on the plasma membrane. Nevertheless, because whole-cell patch clamp poorly controls voltage in the dendritic tree of neurons (285, 286), the whole-cell Ba2+ currents recorded in cultured mouse hippocampal neurons heterologously expressing human 1A (211, 284) are unlikely to come from merely the dendrites and axon terminals of the neurons. Moreover, single-channel P/Q-type currents were recorded on cell-attached patches on the soma of acutely isolated rat hippocampal CA1 pyramidal cells. However, P/Q-type currents were rarely observed on 9 of 68 patches, compared with L-type currents which were observed on 267 of 502 patches (287). Therefore it is reasonable to surmise that 1A constructs heterologously expressed in cultured hippocampal neurons can traffic to the plasma membrane of the somata. However, to our knowledge, there has been no report of successful labeling of endogenous or heterologous 1A in neurons under true non-permeabilizing conditions.
Considering that the lack of labeling of the tagged 1A constructs on the plasma
165 membrane might be due to inadequate surface expression of the tagged channels, we have tried several strategies to increase surface expression of the tagged 1A constructs. Although it has been reported that heterologous expression of human 1A on the surface of cultured hippocampal neurons does not require extra auxiliary subunits (211, 284), we tried cotransfecting 2 along with hPQGH3 and 3, which did not improve the labeling of surface hPQGH3 (see 4.2). We also tried substituting
2a for 3 in fixed-cell imaging experiments in HEK 293T cells under non-permeabilizing conditions, and obtained no improvement of the labeling of hPQGH3 (data not shown). It has been shown that temperature influences the folding efficiency of GFP (288-290) and even the functions of GFP fusion proteins (289). At
37°C, GFP has poor folding efficiency in mammalian cells (288, 289) and E. coli
(290), and a GFP-fused transcription factor gives no transactivation (289). At 15-30°C, however, the folding efficiency of GFP (288-290) and the function of the GFP-fused transcription factor (289) are significantly improved. Although EGFP has been shown to fold efficiently at both 37°C and 28°C (290), our results show that incubating transfected neurons at 30°C enhances the green fluorescence of hPQGH3 (data not shown), suggesting that a low temperature improves the folding efficiency of the
EGFP tag in hPQGH3. Labeling of hPQGH3 under non-permeabilizing conditions, however, is not improved by the low temperature (data not shown). Low temperature of 15-20°C has been shown to have a dual effect of improving the folding efficiency of GFP and arresting vesicular protein transport at the trans-Golgi network in HeLa
166 cells (288). Restoring the temperature to 37°C after low temperature treatment ensures both efficient folding and secretion of a GFP-fused secretory protein (288). Incubation at 30°C might inhibit trafficking of hPQGH3 to the plasma membrane, and restoring the temperature to 37°C might be necessary to obtain robust surface expression of hPQGH3.
Moreover, single-channel recordings on cell-attached patches revealed that on the plasma membrane of the somata of acutely dissociated mouse cerebellar Purkinje cells, P/Q-type channels are robustly expressed (recorded on 21 of 24 patches) and the
P/Q-type current comprises ~85% of the whole-cell Ba2+ current (291). In contrast, on the soma of acutely isolated rat hippocampal CA1 pyramidal cells, single-channel
P/Q-type currents were rarely observed (on 9 of 68 patches), compared to L-type currents (observed on 267 of 502 patches) (287). Because of the relatively high surface expression of P/Q-type channels in cerebellar Purkinje cells, substituting cerebellar Purkinje cells for hippocampal neurons in the imaging experiments may improve the labeling of tagged 1A on the surface. However, poor surface expression may not be the reason or the only explanation for the lack of labeling of tagged 1A on the surface. The lack of labeling may be due to poor exposure of the HA tag to the extracellular space or steric hindrance of antibody binding. It is surprising that of several tagged 1A constructs with the HA tags on different positions that our lab
(Figures15,17) and other groups (281) tried, none displayed a distinct membrane-associated pattern of staining in hippocampal neurons under true
167 non-permeabilizing conditions. Moreover, hPQGH1-4 and hPQGH4x2 all failed to be labeled in HEK 293T cells under non-permeabilizing conditions, although it has been reported that a construct similar to hPQGH2 displayed a distinct membrane-associated pattern of staining in HEK 293 cells under non-permeabilizing conditions (225).
These results suggests that if the lack of labeling is due to steric hindrance, the steric hindrance is not localized to a certain extracellular loop, but is present in the whole extracellular portion of the 1A subunit.
Taken together, imaging experiments that our group and others performed with
HA-tagged 1A have raised a lot of questions about surface P/Q-type channels. More studies are needed to answer these questions. Compared with immunocytochemistry, electrophysiology is a more direct method to detect ion channels on the plasma membrane. Therefore, it would be best to first use single-channel, inside-out or outside-out patch recordings to confirm that our tagged 1A constructs can be expressed on the surface of cultured neurons, and in which parts of the neuron the channels are expressed on the surface. Once the expression of the tagged 1A constructs on the plasma membrane is confirmed, we can further explore how to make the channels stainable from the extracellular side. Other tags, such as polyhistidine-tags and bungarotoxin-binding site, can be tried and may overcome the steric hindrance. Using antibodies directly against extracellular loops of 1A is also an alternative. In addition, structural studies of 1A may shed light on the mechanisms of potential steric hindrance.
168
Chapter 5: Functional Properties of Fragment-Channels
169
5.1 Introduction
The experiments in Chapter 3 indicate that the 190-kDa fragment of mouse 1A contains part of the II-III loop, repeat III, III-IV loop, repeat IV and all or part of the
C-terminal tail, including the histidine tract, and is missing all or part of the
N-terminal tail. Thus, it is probable that the 190-kDa fragment is a product of mid-channel proteolysis, whose cleavage site is in the II-III loop. This mid-channel proteolysis would also produce a product of 80-kDa, N-terminal to the cleavage site.
Western blot analysis with the II-III loop antibody revealed that most of the 1A subunits in the mouse brain are truncated forms instead of the full-length form (Figure
5). Why do the cells produce these “fragment channels”? Do these fragment channels conduct currents? Do they interact with full-length channels? Do they alter the electrophysiological properties of full-length channels?
5.2 Functional properties of fragment-channels produced by cleavage in the
II-III loop of 1A
Two constructs of truncated human 1A were made as if they were the products of a proteolytic cut in the II-III loop (Figure 19A). The cleavage site was chosen to be between N719 and A720. The two fragments are named B1 and B2, and have molecular weights of 81 kDa and 201 kDa, respectively. An equivalent cleavage site in mouse 1A would produce two fragments of 81 kDa and 187 kDa, respectively, corresponding to our finding in the Western blot experiments described in Chapter 3.
170 cRNAs of B1 and B2, either individually or together, were injected into oocytes of
Xenopus laevis along with the cRNAs of the 3 and 2 subunits. Full-length human
1A was used as a positive control. Current-voltage (I-V) relationship was determined using two-electrode voltage clamp (TEVC) as described in Materials and Methods.
Full length human 1A produced currents with an average peak value of about -8 A, while B1 and B2 did not produce currents when they are expressed individually or coexpressed (Figure 19B).
Next, the influences of the B1 and B2 fragments on full-length 1A were tested.
Full-length human 1A was expressed alone or coexpressed with B1 or B2, along with the auxiliary subunits, in Xenopus oocytes. I-V relationship and voltage dependence of steady-state inactivation were determined using TEVC. The result shows that B1 reduces the average peak current of the full-length channel for about
50%, while B2 does not have a significant effect on the peak current amplitude of the full-length channel (Figure 19C). Neither B1 nor B2 had an obvious effect on the shape of the I-V curve or the steady-state inactivation curve of the full-length channel
(Figure 19 D and E).
171
A hPQB1 (81 kDa) hPQB2 (202 kDa) I II III IV
B C 10 6 *
A A *
8 5 4 6 3 4 2 2 1 n=23 n=20 n=23
peak current/- peak
Peakcurrent/- 0 0
hPQ hPQ (n=8)B1 (n=8)B2 (n=8) hPQ+B1hPQ+B2 B1+B2 (n=9)
hPQ (n=6) hPQ (n=23) D E hPQ+B1 (n=5) hPQ+B1 (n=20) StandardhPQ+B2 Err(n=5) hPQ+B2 (n=23) 1.0 0.2 0.0 0.8 -0.2 0.6 -0.4 -0.6 0.4 -0.8 0.2 -1.0 0.0
normalized current normalized normalized current -1.2
-60 -40 -20 0 20 40 60 -80-60-40-20 0 V/mV V/mV
Figure 19. Functional properties of fragment-channels hPQB1 and hPQB2. (A) Schematic structures of hPQB1 and hPQB2. Red segment denotes the synprint site. (B) hPQB1 and hPQB2 do not conduct currents when expressed individually or together. (C) hPQB1 significantly reduces currents of full-length P/Q channel, while hPQB2 does not have a significant effect on the current amplitude of full-length P/Q channel. (D) Neither hPQB1 nor hPQB2 alters the I-V relationship of steady-state inactivation of full-length P/Q channel. (E) Neither hPQB1 nor hPQB2 alters the voltage-dependence of steady-state inactivation of full-length P/Q channel. All experiments were done using bath solution containing 4 mM of Ba2+, except the experiment in (B), which was done using bath solution containing 40 mM of Ba2+. Data are represented as mean ± SEM.
172
5.3 Functional properties of fragment-channels produced by cleavage in the I-II loop and the III-IV loops of 1A
It is possible that the I-II loop and III-IV loop of the mouse 1A may also be subject to proteolytic cleavage. The 1 subunits of the L-type calcium channel and the voltage-gated sodium channel have both been shown to have more than one cleavage site (4, 158). The 140-kDa band detected by the N-terminus antibody in cultured mouse cortical neurons (Figure 6B) might be the N-terminal product of a proteolytic cut in the III-IV loop, whose theoretical molecular weight is ~170 kDa. Thus, we would also like to know the functional properties of the proteolytic products of these cleavage sites. Therefore, two more sets of 1A fragments were made (Figure 20 A and B). Fragments A1 and A2 correspond to a proteolytic cut in the I-II loop, while fragments C1 and C2 correspond to a proteolytic cut in the III-IV loop. These fragments were also expressed in the Xenopus oocytes and characterized using TEVC.
None of these four fragment channels produced currents by themselves. When coexpressed, however, A1 and A2 did produce currents comparable to those conducted by full-length channels (Figure 20C). The I-V curve of the currents conducted by A1 and A2 is also very similar to that of the full-length channel (Figure
20D). When coexpressed with the full-length channel, none of the four fragments had a significant effect on the current amplitude of the full-length channel, although the average current conducted by full-length channels with A2 is significantly larger than with A1 (Figure 1 E and F). None of the fragments changed the I-V relationship or the
173 voltage dependence of steady-state inactivation of the full-length channel (Figure 20
G-J).
A hPQA1 (50 kDa) hPQA2 (233 kDa) B hPQA1 (172 kDa) hPQA2 (111 kDa) I II III IV I II III IV
C D hPQ (n=10) E 2D GraphA1+A2 (n=13) 2 5 10
A
A
4 0.0 8 -0.2 3 6 -0.4 2 -0.6 4 -0.8 1 2 -1.0
Peak current/- Peak
Peak current/- Peak
0 normalized current -1.2 0
A1 A2 C1 C2 -60-40-200 204060 hPQ hPQ A1+A2 C1+C2 V/mV hPQ+A1hPQ+A2
F G hPQ (n=15) hPQ (n=13) H 2D Graph 4 2DhPQ+A1 Graph (n=16) 2 hPQ+C1 (n=15) hPQ+A2 (n=14) hPQ+C2 (n=15) 10 0.2
A 0.2 8 0.0 0.0 -0.2 -0.2 6 -0.4 -0.4 4 -0.6 -0.6 -0.8 -0.8 2 -1.0 -1.0
current/- Peak
normalized current 0 -1.2 normalized current -1.2 -60 -40 -20 0 204060 -60 -40 -200 20 40 60 80 hPQ hPQ+C1hPQ+C2 V/mV V/mV
I 2D hPQGraph (n=5) 5 J 2D GraphhPQ (n=5) 6 hPQ+A1 (n=4) hPQ+C1 (n=5) hPQ+A2 (n=5) hPQ+C2 (n=5) 1.0 1.0 0.8 0.8 0.6 0.6 0.4 0.4 0.2 0.2
normalized current normalized normalizedcurrent 0.0 0.0 -80 -60 -40 -20 0 -80 -60 -40 -20 0 V/mV V/mV Figure 20. Functional properties of fragment-channels produced by cleavage in the I-II loop and the III-IV loops of 1A. (A) Schematic structures of hPQA1 and hPQA2. Red segment denotes the -interaction domain (AID). (B) Schematic structures of hPQC1 and hPQC2. (C) hPQA1 and hPQA2, when coexpressed, conduct currents comparable to those conducted by full-length P/Q channel. (D) I-V relationship of the channel formed by hPQA1 and hPQA2 is similar to that of the full-length P/Q channel. (E-J) None of the four fragment-channels significantly alters the current amplitude (E and F), I-V relationship (G and H) or voltage-dependence of steady-state inactivation (I and J) of full-length P/Q channel.
174
5.4 Expression of fragment-channels
Previous results from our lab shows that all three complementary pairs of L-type fragment channels are functional (4). Among the three complementary pairs of
P/Q-type fragment channels, however, only one pair produced currents. We would like to know why the other two pairs cannot produce currents. Is it because the proteins of the fragment channels cannot be synthesized? Can the reason be that the fragment channels do not go to the plasma membrane? Is it because the complementary pair of fragment channels cannot form a functional channel?
To answer these questions, tagged P/Q fragment channels were expressed in
Xenopus oocytes and analyzed using Western blot, as described in Materials and
Methods. In brief, the N-terminal fragments, (A1, B1, and C1), were tagged with
FLAG, and the C-terminal fragments, (A2, B2, and C2), were tagged with HA. cRNAs of each complementary pair of tagged 1A subunits were coinjected into
Xenopus oocytes, along with 3 and 2. Because A1 and A2 produced currents when coinjected, we know that they can be expressed and form functional channels on the plasma membrane, and thus A1 and A2 were used as a positive control. Uninjected oocytes were used as a negative control. Five days after injection, cell surface protein was isolated using surface biotinylation and protein samples were prepared for
Western blot, as described in Materials and Methods. Briefly, the oocytes were subjected to either surface biotinylation or mock biotinylation, in which no sulfo-NHS-SS-biotin was added. The membrane fraction was isolated using
175 ultracentrifugation, and NeutrAvidin Agarose was used to pull down the biotinylated proteins. Both the flow-through and the eluate from the NeutrAvidin Agarose were analyzed using Western blot with anti-FLAG and anti-HA antibodies.
In the first experiment, FLAG-B1 and HA-B2 were analyzed, with FLAG-A1 and HA-A2 as positive controls. FLAG-A1 and HA-A2 were both detected in the flow-through, as expected (Figure 21 A and C). Additionally, FLAG-B1 and HA-B2 were also detected in the flow-through (Figure 21 A and C). These results suggest that both fragments were expressed. In the eluates, we expected to see FLAG-A1 and
HA-A2 detected in the biotinylated group, but not in the mock-treated group.
However, the two fragments were detected in both the biotinylated group and the mock-treated group (Figure 21 B and D). Therefore, the fragments can non-specifically bind to the NeutrAvidin Agarose.
Nevertheless, more HA-A2 was detected in the eluate of the biotinylated group than in the eluate of the mock-treated group. This indicates that the HA-A2 on the plasma membrane was able to be biotinylated. Unfortunately, the bands of FLAG-A1 in the biotinylated group and the mock-treated group have almost the same intensity, which is probably since the non-specific binding of FLAG-A1 to the NeutrAvidin
Agarose was so strong that it concealed the effect of biotinylation. HA-B2 was also detected in both the biotinylated group and the mock-treated group (Figure 21D). Like
HA-A2, more HA-B2 was detected in the eluate of the biotinylated group than in the eluate of the mock-treated group, which suggests that HA-B2 can go to the surface.
176
Very little FLAG-B1 were detected in the eluates (Figure 21B), indicating that B1 does not adequately traffic to the plasma membrane.
A flow-through B eluate C flow-through D eluate IB: FLAG IB: FLAG IB: HA IB: HA
E flow-through F eluate G flow-through H eluate IB: FLAG IB: FLAG IB: HA IB: HA
I
177
Figure 21. Expression of fragment-channels in the cytoplasm and on the plasma membrane of Xenopus oocytes. (A-D) Western blot analysis of oocytes expressing FLAG-B1 and HA-B2, with oocytes expressing FLAG-A1 and HA-A2 as positive controls and uninjected oocytes as negative controls. All four fragment-channels were detected in the flow-through (A and C). Similar amounts of FLAG-A1 were detected in the eluates of both the biotinylated and mock-treated groups (B). FLAG-B1 is not detected in the eluates (B). More HA-A2 and HA-B2 were detected in the eluate of the biotinylated group than in the eluate of the mock-treated group (D). (E-H) Western blot analysis of oocytes expressing FLAG-C1 and HA-C2, with oocytes expressing FLAG-A1 and HA-A2 as positive control and uninjected oocytes as negative control. All fragment-channels were detected in the flow-through, except FLAG-C1 (E and G). More FLAG-A1, HA-A2 and HA-C2 were detected in the eluate of the biotinylated group than in the eluate of the mock-treated group (F and H). (I) In the whole-cell lysate, FLAG-C1 could not be detected either, while FLAG-A1 could be detected. Because in some samples multiple bands were detected, bands of interested were indicated by red arrows. Small pictures show the bands in the red squares in the pictures above with a shorter exposure time. Theoretical molecular weights of the fragment-channels: FLAG-A1, 51 kDa; HA-A2, 234 kDa; FLAG-B1, 82 kDa; HA-B2, 203 kDa; FLAG-C1, 173 kDa; HA-C2, 112 kDa.
In the second experiment, FLAG-C1 and HA-C2 were analyzed, with FLAG-A1 and HA-A2 as positive control. Again, FLAG-A1 and HA-A2 were both detected in the flow-through and the eluates of both the biotinylated group and the mock-treated group (Figure 21 E-H). This time, both FLAG-A1 and HA-A2 have a stronger band in the biotinylated group than in the mock-treated group, indicating that surface
FLAG-A1 and HA-A2 could be biotinylated. The biotinylated FLAG-A1 and HA-A2 could be pulled down by the NeutrAvidin Agarose, although there is also non-specific binding of FLAG-A1 and HA-A2 to the NeutrAvidin Agarose. HA-C2 was also detected in both the flow-through and the eluates (Figure 21 G and H), and more
HA-C2 was detected in the eluate of the biotinylated group than in the mock-treated group, indicating that HA-C2 was expressed and transported to the plasma membrane.
Surprisingly, FLAG-C1 was not detected in the flow-through or in the eluates (Figure
178
21 E and F). Was FLAG-C1 not expressed? Note that in this experiment, only the membrane fraction of the oocytes was isolated and analyzed using Western blot.
Could the FLAG-C1 be only in the cytosol? In another experiment, we tested the whole-cell lysate instead of the membrane fraction, and found that FLAG-C1 cannot be detected either (Figure 21I).
In summary, the above biochemical experiments show that all the 1A fragments can be expressed except C1, which can explain the lack of currents when C1 and C2 are coexpressed. Because of non-specific binding of non-biotinylated 1A fragments to the NeutrAvidin Agarose, surface expression of the fragment channels is ambiguous. Nevertheless, C2 and B2 did show a hint of surface expression, while B1 seemed to lack surface expression. This finding can explain the lack of currents when
B1 and B2 are coexpressed.
5.5 Discussion
5.5.1 Why is C1 not expressed?
By injecting oocytes with cRNAs, we obtained protein expression of all of the six engineered truncated 1A subunits except C1 (Figure 21). The absence of C1 protein might be due to several reasons. Firstly, the cRNA might be unstable in oocytes and was degraded, which can be confirmed by Northern blot analysis. Most eukaryotic mRNAs require a 7-methyl guanosine (m7G) cap structure at the 5’ end and a poly (A) tail at the 3’ end to be stable and efficiently translated. RNA Cap
179
Structure Analog m7G(5')ppp(5')G (New England Biolabs S1404S) is always included in our in vitro RNA synthesis system, and the DNA templates for RNA synthesis always include a poly (A) stretch downstream of the open reading frame. Thus the lack of a cap structure or a poly (A) tail is unlikely to be the reason for any RNA instability. However, because C1 is a truncated protein and its cRNA contains a premature stop codon, RNA degradation may result from nonsense-mediated mRNA decay (NMD) (292). NMD is a quality control process that degrades mRNAs containing premature stop codons, thereby repressing the production of truncated and potentially harmful proteins (292).
Secondly, the cRNA of C1 may not be able to be translated. There are two main methods of improving translation of cRNA in Xenopus oocytes (293). One is to remove the RNA secondary structures in the regions flanking the coding sequence.
Another method is to use a vector containing the 5’ and 3’ untranslated regions (UTRs) of the Xenopus -globulin gene, such as the pSP64T vector (Addgene 15030) (293).
Thirdly, C1 may be recognized as an unfolded or misfolded protein by the endoplasmic reticulum (ER) and induce the unfolded protein response (UPR) or endoplasmic reticulum-associated degradation (ERAD) (94). UPR results in translation arrest and mRNA destabilization (214, 215). A certain amount of unfolded or misfolded protein is required for UPR to take place, and UPR blocks the translation of a broad range of proteins (214). Because the C1 protein was completely absent while the C2 protein in the same oocytes was robustly expressed (Figure 21 E and G),
180
UPR is unlikely to be the reason for the lack of expression of C1. Misfolded proteins can also be retained in the ER and degraded by the proteasome, through a mechanism called endoplasmic reticulum-associated degradation (ERAD) (94). This mechanism has been shown in HEK 293 cells to be the way in which degradation of the
EA2-causing truncated 1A, R1281X occurs (178). A simple way to test the presence of ERAD is to apply proteasome inhibitors and see whether C1 protein can be detected.
Interestingly, the other two engineered C-terminally truncated 1A, A1 and B1, were both able to be expressed. Why C1 is the only fragment-channel that was unable to be expressed is unknown. What is more interesting is that C1 resembles two
EA2-causing truncation mutants with premature stop codons in the III-IV loop,
R1549X (206) and Q1564X (207) (see GenBank AF004884). However, a lack of protein expression, or haploinsufficiency can hardly explain the pathogenesis of these truncation mutants (210). Actually R1549X has been studied and was shown to be able to be expressed in COS-7 cells (202) and HEK 293T cells 141 (199). R1549X
was also shown to have a dominant negative effect on full-length CaV2.1 in Xenopus oocytes (203) and HEK 293T cells (199). C1, however, not only lacked protein
expression, but also failed to influence the currents of full-length CaV2.1 (Figure 20 F,
H and J). Additionally, the truncation sites of C1 and R1549X are only 24 amino residues apart. Why do two truncation mutants so similar to each other display such different properties? It is crucial to study C1, R1549X and Q1564X in parallel,
181 confirm their different properties, and explore the causes of their different properties.
5.5.2 A1 and A2 form a functional channel
The complementary fragment-channels, A1 and A2, when coexpressed, conduct
2+ Ba currents as strong as those conducted by full-length CaV2.1, indicating that the complementary pair of fragment-channels can associate with each other and form a functional channel. The phenomenon that complementary fragment-channels form a function channel is common among VGCCs. The two-domain forms of 1B,
CaV2.2-Dom I-II and CaV2.2-Dom III-IV, when expressed individually along with auxiliary subunits 1b and 2-1 in COS-7 cells, do not conduct currents, but when coexpressed, they form functional channels that conduct low-density whole-cell
currents with properties similar to full-length CaV2.2 (187). Six engineered truncated
1C subunits corresponding to the six engineered truncated 1A subunits were studied in Xenopus oocytes (4). None of the fragment-channels conducts currents by itself, but each complementary pair of fragment-channels conducts whole-cell currents of lower density than the full-length channel (4). Interestingly, the currents conducted by each complementary pair of truncated L-type channels display a positive shift in the I-V relationship and the voltage dependence of steady-state inactivation (4). In contrast, the currents conducted by coexpressed A1 and A2, have an I-V curve similar to that of the full-length channel (Figure 20D) (steady-state inactivation was not tested), and the currents conducted by the complementary pair of
182 truncated N-type channels showed a slight positive shift in the I-V curve but no shift in the voltage-dependence of steady-state inactivation (187). These results have revealed that mid-channel proteolysis may influence the coupling between voltage sensing and gating of different channels in different manners.
It is not surprising that two complementary fragments of a protein can assemble into a complex with a similar function to the full-length protein. Such phenomena have been observed elsewhere too. The mechanism of the inactivation of the
Drosophila melanogaster rapidly inactivating potassium channel, shaker (ShB), could be well explained by the ball and chain model originally proposed for voltage-gated sodium channels (294). The first 20 amino acid residues in the N-terminus of ShB channel form the channel-blocking ball, which is connected to the rest of the channel by a chain of 60 or more amino acid residues (295). A mutant ShB channel without amino acid residues 6-46, ShB6-46, does not inactivate with a rapid time course
(295, 296). Interestingly, application of a peptide corresponding to the channel-blocking ball to the cytoplasmic side of ShB6-46 restores the rapid inactivation (295). Later studies revealed that the ball occludes the channel through a long-range electrostatic interaction, which increases the rate of diffusion of the peptide to the binding site (297). The genes encoding four-domain ion channels, such
as CaV and NaV channels, might have evolved by two rounds of gene duplication of a
gene encoding an ancestral single-domain voltage-gated potassium (KV) channel,
which contains six transmembrane helices (TMs) like each domain (repeat) of a CaV
183 or NaV channel (180, 181). Although a CaV or NaV 1 subunit and a KV 1 subunit have different numbers of domains, the structures of the three types of channels are similar. While a CaV or NaV channel contains a single 1 subunit organized into a pseudo-tetrameric structure (30, 298, 299), a KV channel contains four identical 1 subunits organized into a homotetrameric structure (300). The tetramerization of KV
1 subunits relies on the specific interaction between the 1 subunits. The T1 domain formed by the N-termini of the four 1 subunits contains characteristic residues in the subunit interface, which are crucial for subfamily-specific tetramerization of the KV
1 subunits (301). Before the high-resolution crystal structure of CaV1.1 was solved
(30), structural interpretation of CaV or NaV channels had mainly been based on the crystal structures of homotetrameric channels formed by four single-domain 1 subunits. Therefore it is not surprising that a complementary pair of truncated CaV1 or
CaV2 channels can associate to form a functional channel, just like the four single-domain KV 1 subunits can.
5.5.3 Why do B1 and B2 not conduct currents when coexpressed?
It is strange that no current was detected in Xenopus oocytes coexpressing B1
and B2 (Figure 19B), while equivalent pairs of truncated CaV1.2 and CaV2.2 could form functional channels that conduct currents (4, 187). A simple explanation for this finding is that B1 did not traffic to the plasma membrane, as shown in the results of the surface biotinylation experiments (Figure 21B). However, P1219X (see Table 1), a
184 truncated 1A resembling B1, has been shown to be able to traffic to the plasma membrane when expressed in mouse skeletal muscle myotubes (152) or tsA-201 cells
(6), although it could not conduct currents (152). It has also been shown that P1219X
has a suppressive effect on full-length CaV2.1 currents, probably by competing with the full-length 1A for the subunits (6). B1 also suppresses full-length CaV2.1 currents (Figure 19C). Its lack of expression on the plasma membrane is an interesting difference from P1219X.
Western blot analysis shows that the expression level of B1 protein in the cytoplasm is relatively low among all the fragment-channels expressed in Xenopus oocytes (Figure 21A). Thus it is possible that B1 was expressed on the plasma membrane, but the amount of B1 on the plasma membrane was too low to be detected.
We can try to increase the expression level of B1 by adjusting the amount of cRNA injected. Substituting an anti-reverse cap analog (ARCA, 3’-O-Me-m7G(5')ppp(5')G,
New England Biolabs S1411) for the standard cap analog (m7G(5')ppp(5')G, New
England Biolabs S1404S) also can increase the expression level. The standard cap analog can be incorporated into cRNA in two alternative orientations, resulting in only
50% of capped cRNA that is translatable. In contrast, ARCA is incorporated into cRNA exclusively in the correct orientation, resulting in twice the amount of translatable cRNA.
If increasing the expression level cannot give rise to detectable surface expression of B1, it would be interesting to study whether surface expression of B1
185 can be induced by fusing B1 with a membrane-targeting sequence and whether this can facilitate the formation of functional channels with B2. It has been reported that
CD4 fused to a short fragment of CaV1 subunit can facilitate the transport of the fragment to the plasma membrane of mammalian cells (84, 193). CD4 is a glycoprotein expressed on the surface of T lymphocytes and other immune cells, and is a single-pass transmembrane protein with an extracellular N-terminus and an intracellular C-terminus (302). It has been reported that a fusion protein containing the proto-oncogene protein Wnt-1 whose C-terminus is fused to CD4 can be transported to the plasma membrane as a transmembrane protein in Xenopus embryos (303).
Fusing CD4 to the N-terminus of our B1 fragment-channel may promote the expression of B1 on the plasma membrane and subsequently facilitate the association of B1 and B2 to form functional channels. Moreover, a C-terminal CAAX motif promotes prenylation and palmitoylation of the C-terminus of the protein, thereby allowing the protein to associate with intercellular membrane and subsequently the plasma membrane in both mammalian cells (304, 305) and Xenopus oocytes (306).
Adding a CAAX motif to the C-terminus of non-membrane proteins, such as GFP and
the N-terminal tail of CaV2.2, effectively induces the association of the proteins with the plasma membrane (189, 190). Therefore, adding a CAAX motif to the C-terminus of B1 may also promote the expression of B1 on the plasma membrane and subsequently facilitate the association of B1 and B2 to form functional channels. A palmitoylation sequence (MTLESIMACCL) also has also been shown to target a
186 protein to the plasma membrane (116), and thus can also be tried.
If promoting the surface expression of B1 cannot induce the formation of functional channels, we can infer that B1 and B2 either cannot form a complex or form a complex without channel function. The formation of the B1-B2 complex can be tested by coimmunoprecipitation (co-IP). If B1 and B2 cannot form a complex, it would be interesting to study whether adding specific sequences to B1 and B2 can induce the association of B1 and B2 and subsequently lead to the formation of a functional channel. Coiled coils are a common structure responsible for protein complex formation. For example, trimerization of a coiled-coil domain in the
C-terminus of TRPP2 is crucial for the formation of the TRPP2 trimer and the
TRPP2/PKD1 complex, which contains 3 TRPP2 and 1 PKD1 (307). We can fuse the
C-terminus of B1 and the N-terminus of B2 respectively to the two strands of a coiled coil, which can be either from a natural protein, such as the transcriptional activator
GCN4 (308), or designed de novo (309). Coexpression of these two fusion proteins may lead to association of B1 and B2 and formation of a functional channel.
Our lab has previously examined the acute effect of mid-channel proteolysis on
the functional properties of CaV2.1 by inserting tobacco etch virus protease (TEVp) cleavage sites (with the sequence ENLYFQG) into the I-II loop and II-III loop of 1A.
We then measured the changes in currents upon TEVp treatment using macropatch recordings on inside-out patches from Xenopus oocytes (4). The result showed that a
2-min treatment with purified TEVp did not abolish CaV2.1 currents, but caused an
187 irreversible left shift of the activation curve (4). These findings were in accordance with an increase in the coupling between voltage sensing and channel opening. In contrast, when the complementary fragment-channels B1 and B2 were coexpressed in
Xenopus oocytes, no currents were detected (Figure 19B). These two results are compatible with each other. In the first case, the full-length 1A was already on the plasma membrane before it was cleaved, and thus the resultant fragment-channels will likely stay on the plasma membrane for some time, although the further fate of the fragment-channels is yet to be studied. In the second case, the two fragment-channels were expressed separately and would need to associate with each other properly and traffic to the plasma membrane to form a functional channel. Failure in any of these steps would lead to an absence of functional channels on the plasma membrane.
5.5.4 Possible mechanisms underlying the dominant negative effect of B1
Of all the engineered P/Q-type fragment-channels (except C1 which was not
expressed), only B1 significantly altered full-length CaV2.1 currents. B1 decreases the
current amplitude of full-length CaV2.1 without altering the I-V relationship or voltage dependence of steady-state inactivation. This result differs from those obtained with L-type and N-type fragment-channels. Six L-type fragment-channels were engineered, similarly to the six P/Q-type fragment-channels (4). Of the six engineered L-type fragment-channels, B1 (1C) and B2 (1C) are the only ones that does not show any significant effect on full-length CaV1.2 currents, and C2 (1C) is
188 the only one that suppresses full-length CaV1.2 currents (4). All the other L-type fragment-channels only influence the I-V relationship or voltage dependence of
steady-state inactivation of full-length CaV1.2 currents (4). In the case of N-type
fragment-channels, CaV2.2-Dom I-II, CaV2.2-Dom III-IV(187), CaV2.2-Dom I (188),
CaV2.2 without the N-terminal tail and even the N-terminal tail itself (189) all have a
suppressive effect on full-length CaV2.2 currents without affecting the I-V or steady-state inactivation curves.
The suppressive (dominant negative) effect of B1 on full-length CaV2.1 currents
may be due to the following reasons: 1) B1 decreases full-length CaV2.1 mRNA level;
2) B1 decreases full-length CaV2.1 protein level; 3) B1 suppresses the trafficking of
full-length CaV2.1 to the plasma membrane; 4) B1 alters the single-channel properties
of full-length CaV2.1. The dominant negative effects of several truncated forms of
CaV2.1 and CaV2.2 have been studied and three different mechanisms have been suggested to underlie the dominant negative effects (see 1.7.2.4). Firstly, some
fragment-channels (e.g., CaV2.2-Dom I and CaV2.1-P1217fs) suppress the synthesis of the cognate full-length channel by associating with the full-length channel and initiates the unfolded protein response (UPR). This subsequently leads to translation arrest and mRNA destabilization of the full-length channel and other proteins. UPR leads to a decrease in both the mRNA and protein levels of the full-length channel.
Secondly, some fragment-channels (e.g., the EA2-causing truncated 1A R1281X) promote the degradation of full-length channel protein though a “destructive
189 interaction mechanism” (DIM) involving ER-associated degradation (ERAD). This mechanism leads to a decrease in the protein level of the full-length channel. Thirdly, some other fragment-channels (e.g., 1A truncation mutants 402X and P1219X) compete with the full-length channel for subunits. This mechanism suppresses the trafficking of the full-length channel to the plasma membrane and may also alter the single-channel properties of the full-length channel.
The fragment-channel B1 (A720X) is truncated near the beginning of the II-III loop, remotely resembling a previously studied P/Q-type fragment-channel, P1219X
(6, 152) (equivalent to R1213X in AF004884), which is truncated near the end of the
II-III loop. Evidence suggests that P1219X exerts its dominant negative effect by competing with full-length 1A for subunits (6). It is possible that B1 also suppresses full-length CaV2.1 currents through this mechanism. However, A1, which also contains the -interaction domain (AID) and thus can presumably bind to subunits, does not have a dominant negative effect (Figure 20E). Therefore, we surmise that competition with full-length 1A for subunits is unlikely the mechanism underlying the dominant negative effect of B1.
It has been shown that two-domain truncated forms of 1A and 1B not only suppress the cognate full-length channels, but also cross-suppress the currents of other
members of the CaV2 family (188). These findings indicate that the dominant negative effects of two-domain truncated forms of 1A and 1B may share the same mechanism. Of the truncated forms of 1A and 1B studied, none has a biologically
190 significant effect on the activation and inactivation parameters of the corresponding
full-length channel (Table 1) (187-189). Moreover, CaV2.2-Dom I-II, despite
suppressing full-length CaV2.2 currents, does not alter the single-channel properties of
full-length CaV2.2 (187). Therefore, it is unlikely that B1 suppresses full-length
CaV2.1 currents by binding to the full-length channel and altering the single-channel properties.
Studies from the Dolphin group suggest that the mechanism underlying the
dominant negative effects of CaV2.1-P1217fs, CaV2.2-Dom I and CaV2.2-Dom I-II is the unfolded protein response (UPR). This response involves the endoplasmic reticulum resident RNA-dependent protein kinase (PERK) and results in translation
arrest and mRNA destabilization of full-length CaV2 and other proteins (187, 188).
They also provided evidence that the mechanism does not rely on the sequestration of
subunits or the ubiquitin-proteasome pathway (187, 188).
However, other studies revealed that EA2 truncation mutants Q681fs and
R1281X exert their dominant negative effects through a “destructive interaction mechanism” (DIM) involving ER-associated degradation (ERAD), which relies on ubiquitin-proteasome pathway (178, 199, 203). Q681fs and R1281X both remotely resemble P1217fs studied by the Dolphin group (Figure 2). Their different mechanisms of dominant negative effects may be due to the slight differences in their truncation sites, but may also be due to methodological differences between different studies. Our fragment-channel B1, which remotely resembles all the above three
191 mutants, may exert its dominant negative effect through UPR and/or DIM. Further studies will be needed to reveal the mechanism for the dominant negative effect of
B1.
The following experiments can be performed to explore the mechanisms underlying the dominant negative effect of B1. Co-immunoprecipitation can be used to find out whether B1 interacts with subunits or full-length 1A. If B1 suppresses full-length 1A currents by sequestration of subunits, B1 should be found to interact with subunits. Additionally, mutations in the -interaction domain (AID) of B1 that abolish the interaction with subunits (6) should also abolish the dominant negative effect. Interaction of B1 with full-length 1A would be compatible with all the other possible mechanisms. In this case, single-channel recordings should be performed to see whether B1 alters the single-channel properties of 1A.
Moreover, imaging or cell-surface biotinylation assay should be performed to examine whether B1 reduces surface expression of full-length 1A. The influence of
B1 on the protein and mRNA levels of full-length 1A should also be examined.
Eventually, the involvement of the ubiquitin-proteasome pathway and endoplasmic reticulum resident RNA-dependent protein kinase (PERK) will be the most important indicators of DIM and UPR, respectively. The expected results for each mechanism are summarized in Table 3.
192
Table 3 Possible mechanisms underlying the dominant negative effect of B1 and the expected phenomena Sequestration Alteration of UPR DIM of subunits single-channel properties Interaction with Yes Maybe Maybe Maybe Interaction with Maybe Yes Yes Yes full-length 1A Mutation in AID Abolishes the No influence on No influence No influence on (6) dominant the dominant on the the dominant negative effect negative effect dominant negative effect negative effect Changes in Maybe Yes No No single-channel properties Decrease in Maybe No Yes Yes full-length 1A protein level Decrease in No No Yes No full-length 1A mRNA level Decrease in Yes No Yes Yes full-length 1A surface expression Effect of No effect No effect No effect Reduces the proteasome dominant inhibitor (187, negative effect 188) Effect of dominant No effect No effect Reduces the No effect negative PERK dominant (188) negative effect UPR: unfolded protein response DIM: destructive interaction mechanism AID: -interaction domain (AID) PERK: endoplasmic reticulum resident RNA-dependent protein kinase
5.5.5 Additional concerns
Our cell-surface biotinylation assays provided some preliminary results about the surface expression of the fragment-channels. However, the results are not convincing 193 enough because of the binding of non-biotinylated fragment-channels to the
NeutrAvidin agarose. To confirm the surface expression of the fragment-channels, the cell-surface biotinylation assay needs to be optimized in order to remove the nonspecific binding. Before being mixed with the NeutrAvidin agarose, the membrane preparation can be processed using one or a combination of the following methods to improve the purity of plasma membrane proteins: 1) wash with a hypertonic sucrose
buffer, 2) wash with 100 mM Na2CO3, and 3) pre-clear with control agarose (278). 0.5%
(w/v) SDS can be included in the lysis buffer (278) to decrease the binding of non-biotinylated proteins to the NeutrAvidin agarose. Harsher washes may also be needed. A wash buffer containing 450 mM of NaCl was used in our experiments (see
Materials and Methods). Increasing the NaCl concentration to 0.5-5 M and including
0.5% (w/v) SDS and 100 mM Na2CO3 in the wash buffer (278) may help eliminate non-biotinylated proteins from the NeutrAvidin agarose.
Interestingly, in the immunoblot analysis of Xenopus oocytes coexpressing
FLAG-A1 and HA-A2, the anti-FLAG antibody not only detected a 50-kDa band corresponding to FLAG-A1, but also detected a band of ~260 kDa (Figure 21 A and
B), which is approximately the same molecular weight as full-length 1A (282 kDa).
A possible explanation is that the 260-kDa band is the complex formed by FLAG-A1 and HA-A2, which was not broken apart by SDS and DTT. Alternatively, the 260-kDa band might just be aggregated FLAG-A1. Moreover, the apparent molecular weight of
HA-A2 is ~280-kDa, which is higher than it theoretical molecular weight (233 kDa)
194 and closer to the molecular weight of full-length 1A (Figure 21 C and D). Note that the expression level of HA-A2 is relatively low.
Therefore, a possible reason we do not see HA-A2 of its theoretical molecular weight is that HA-A2 is saturated with FLAG-A1 and the 280-kDa band recognized by the anti-HA antibody is the complex containing both FLAG-A1 and HA-A2.
However, it is also possible that the 280-kDa band is HA-A2 only and its high apparent molecular weight is due to anomalous migration. A similar phenomenon can be observed in Xenopus oocytes coexpressing FLAG-B1 and HA-B2. HA-B2 has an apparent molecular weight of 270 kDa, which is more than the theoretical molecular weight of HA-B2 (202 kDa) and closer to the molecular weight of full-length 1A
(Figure 21 C and D). FLAG-B1, however, did not show any band of similar molecular weight to full-length 1A (Figure 21A).
All the protein samples from Xenopus oocytes were incubated with 2% SDS at
37°C in the presence of high concentration (50-90 mM) of DTT, in order to thoroughly dissociate protein complexes and denature the proteins. However, it cannot be excluded that some protein complexes might be extremely stable and resistant to denaturation. Superoxide dismutase (SOD) is one of the most denaturation-resistant proteins (310). SOD from Suvolobus acidocaldarius retains almost all its activity after
24 h incubation at 95°C, and is also extremely resistant to 6 M guanidinium hydrochloride or 0.1% SDS at 20°C (310). Bovine SOD retains much of its activity in the presence of 10 M urea or 4% SDS (311). Bovine SOD is a homodimer whose two
195 subunits are associated by non-covalent interactions (312). The SOD homodimer is highly resistant to dissociation and does not dissociate even in 8 M urea for 72 h at
25°C (313). Nevertheless, bovine SOD homodimer can be dissociated by incubation at 37°C for 2 h in 1% SDS and 1% -mercaptoethanol (314, 315). The
-mercaptoethanol is necessary for the dissociation (314), even though no intersubunit disulfide bond has been found. There is not enough evidence to conclude whether our method of sample preparation can disrupt an intersubunit interaction as strong as that in SOD. It is possible that our denaturation conditions were not harsh enough to dissociate the complementary pair of fragment-channels.
If the complementary fragment-channels can indeed be shown to form stable complexes highly resistant to dissociation, this will provide great insight into the functional properties of fragment-channels. To address this issue, we should express the fragment-channels individually in separate groups of Xenopus oocytes and analyze them using immunoblot blot. If the high molecular weight species are absent and all the fragment-channels display their theoretical molecular weights, this indicates that the high molecular weight species are complexes formed by the complementary fragment-channels. If the high molecular weight species appear as in the oocytes coexpressing the complementary fragment-channels, we can infer that the high molecular weight species is due to aggregation (in the case of A1) or anomalous migration (in the case of A2 and B2).
196
Chapter 6: Summary and Future Directions
197
6.1 Functional importance of the 190-kDa truncated 1A
A 190-kDa fragment of 1A was found in mouse brain tissue and cultured mouse cortical neurons, but not in heterologous systems expressing full-length 1A. In the brain, the 190-kDa truncated channel is the predominant form of 1A, while in cultured cortical neurons the amount of the 190-kDa truncated channel is comparable to that of the full-length channel. The 190-kDa truncated channel contains part of the
II-III loop, repeat III, repeat IV and the C-terminal tail. Evidence suggests that the
190-kDa truncated channel is not due to degradation of the full-length channel during sample preparation. Our immunoblot experiments also revealed a 110-kDa short form of 1A, which is presumably the truncated channel complementary to the 190-kDa truncated channel.
It is interesting that the brain produces much more truncated P/Q-type calcium channels than full-length ones. We would like to know whether the 190-kDa truncated channel has functional importance or is only a product of routine protein degradation in the cells. Because ion channels carry out their functions mainly on the plasma membrane, expression of the 190-kDa truncated channel on the plasma membrane would suggest functional importance. It is even possible that most of the functional
P/Q-type channels on the plasma membrane are formed by complementary pairs of truncated channels, instead of full-length channels. We tried two methods to explore the expression of the truncated channel on the plasma membrane, but the results were inconclusive. The cell surface biotinylation experiment with mouse forebrain slices
198 could not reveal whether the 190-kDa truncated channel is on the plasma membrane, because the biotinylation reagent increases the immunoreactivity of the truncated channel. The immunocytochemistry experiments show that extracellularly HA-tagged
1A expressed in cultured rat hippocampal neurons or HEK 293T cells cannot be effectively labeled under non-permeabilizing conditions. In contrast, extracellularly
HA-tagged 1C could be robustly labeled on the plasma membrane under the same conditions. The lack of labeling of HA-tagged 1A might be due to either inadequate surface expression or poor exposure of the HA tags. Therefore the immunocytochemistry experiments cannot give us a clue about the surface expression of the truncated channel either.
To continue exploring the expression of the 190-kDa truncated channel on the plasma membrane, we can try the following methods: 1) using another antibody, such as an antibody against the C-terminal tail of 1A, in the cell surface biotinylation experiment, 2) using other biotinylation reagents such as amino-oxy-biotin and biocytin hydrazide, 3) isolating cell surface protein using radioiodination, and 4) isolating cell surface 1A using peptide toxins that specifically bind to 1A on its extracellular surface. To trouble-shoot the immunocytochemistry experiment, single-channel, inside-out or outside-out patch recordings should be used to evaluate cell-surface expression of the HA-tagged 1A constructs on the soma. If the constructs display low cell-surface expression on the soma, the following methods could be tried to boost cell-surface expression: 1) culturing transfected cells first at
199
30°C to improve protein folding and then at 37°C to promote trafficking to the plasma membrane, and 2) substituting cerebellar Purkinje cells for hippocampal neurons. If the HA-tagged 1A constructs display adequate cell-surface expression on the soma, this means poor exposure of the HA tags is the reason for the lack of labeling. In this case we should try the following methods: 1) inserting long flanking sequences beside the HA tag to improve the exposure of the tag, 2) using other types of tags such as
FLAG-tag, His-tag and Myc-tag, and 3) directly staining the channel using an antibody against an extracellular loop.
The engineered fragment-channel B1, presumably resembling the truncated channel complementary to the 190-kDa truncated channel, suppresses full-length P/Q channel currents. The fragment-channel B2, equivalent to the 190-kDa truncated channel, does not significantly affect full-length P/Q channel currents. B1 and B2 cannot form a functional channel when coexpressed in Xenopus oocytes. However, these results cannot rule out the possibility that the 190-kDa truncated channel has important functions. The 190-kDa truncated channel may regulate other types of
VGCCs. TEVC recordings should be performed to test the influence of the B2 fragment-channel on L- and N-type channels.
Another possible function of the 190-kDa truncated channel is to regulate the suppressive effect of its complementary fragment-channel on the full-length channel.
Several other truncated forms of 1A, including EA2 mutants and non-disease-related truncated forms, have been reported to have a dominant negative effect (see Table 1).
200
The dominant negative effect is believed to be an important mechanism for the pathogenesis of EA2. It is possible that the 190-kDa truncated channel downregulates the suppressive effect of its complementary fragment on the full-length channel, thereby preventing it from causing EA2 or a similar disorder. To test this hypothesis, we need to use TEVC to measure the dominant negative effect of the B1 fragment-channel in the presence and absence of B2. The interaction between B1 and
B2 should be examined using co-IP.
6.2 Origin of the 190-kDa truncated 1A
Our preliminary data show that the relative abundance of the 190-kDa truncated channel and the putative complementary fragment-channel are upregulated by increased intracellular [Ca2+]. These findings are reminiscent of the mid-channel proteolysis of 1C found by our group (4), and therefore suggest that the 190-kDa truncated channel may also be produced by mid-channel proteolysis. Like the mid-channel proteolysis of 1C, mid-channel proteolysis of 1A may also serve as a negative feedback regulatory mechanism, which downregulates channel activity when intracellular [Ca2+] increases.
To further explore the possibility of mid-channel proteolysis, we can insert a
TEVp site in the putative cleavage site in the II-III loop of 1A, and then perform single-channel recordings to find out how the electrophysiological properties of the channel changes upon proteolysis. Next we use single-channel recordings to measure
201 the changes in electrophysiological properties of endogenous CaV2.1 following intracellular [Ca2+] increase. If the changes following intracellular [Ca2+] increase are similar to the changes that TEVp causes on the engineered channels, we can infer that the 190-kDa fragment is likely due to mid-channel proteolysis.
N-terminal sequencing or mass spectrometry should be performed to confirm the identities of the 190-kDa truncated channel and the complementary fragment of 110 kDa. Considering the possibility that the 190-kDa truncated channel may be produced by a short transcript due to alternative splicing or a cryptic promoter, 5’ RACE-PCR should be used to screen for short splice variants of 1A that may produce the
190-kDa species. The putative transcript for the 190-kDa species would be missing at least exons 2-16 (see 3.9). A transcript with part of exon 1 would be a splice variant, while one completely without exon 1 would be a product of a cryptic promoter. 3’
RACE-PCR can be used to screen for splice variants of 1A that may produce the
110-kDa species.
My research has expanded our knowledge of VGCCs. The pore-forming subunit of a VGCC is usually described as a protein with a single peptide chain containing four transmembrane domains. However, my research shows that most of the P/Q-type channel 1A subunits in the brain are truncated. These truncated channels may play various physiological roles both on the plasma membrane and intracellularly. My research also added complexity to our understanding of the pathogenesis of EA2. My results suggest that the presence of an EA2 truncation
202 mutant 1A and the absence of a complementary truncated 1A may both be contributing factors to EA2. Future studies on the P/Q-type fragment-channels may further shed light on the pathogenesis of EA2 and help discover potential therapeutic tools.
203
References 1. Catterall WA. Voltage-gated calcium channels. Cold Spring Harb Perspect Biol. 2011;3(8):a003947. PMCID: 3140680.
2. Catterall WA. Regulation of Cardiac Calcium Channels in the Fight-or-Flight Response. Curr Mol Pharmacol. 2015;8(1):12-21. PMCID: 4664455.
3. Gomez-Ospina N, Panagiotakos G, Portmann T, Pasca SP, Rabah D, Budzillo A, et al. A promoter in the coding region of the calcium channel gene CACNA1C generates the transcription factor CCAT. PLoS One. 2013;8(4):e60526. PMCID: 3628902.
4. Michailidis IE, Abele-Henckels K, Zhang WK, Lin B, Yu Y, Geyman LS, et al. Age-related homeostatic midchannel proteolysis of neuronal L-type voltage-gated Ca(2)(+) channels. Neuron. 2014;82(5):1045-57. PMCID: 4052215.
5. Scott VE, Felix R, Arikkath J, Campbell KP. Evidence for a 95 kDa short form of the alpha1A subunit associated with the omega-conotoxin MVIIC receptor of the P/Q-type Ca2+ channels. J Neurosci. 1998;18(2):641-7.
6. Arikkath J, Felix R, Ahern C, Chen CC, Mori Y, Song I, et al. Molecular characterization of a two-domain form of the neuronal voltage-gated P/Q-type calcium channel alpha(1)2.1 subunit. FEBS Lett. 2002;532(3):300-8.
7. Du X, Wang J, Zhu H, Rinaldo L, Lamar KM, Palmenberg AC, et al. Second cistron in CACNA1A gene encodes a transcription factor mediating cerebellar development and SCA6. Cell. 2013;154(1):118-33. PMCID: 3939801.
8. Sakurai T, Hell JW, Woppmann A, Miljanich GP, Catterall WA. Immunochemical identification and differential phosphorylation of alternatively spliced forms of the alpha 1A subunit of brain calcium channels. J Biol Chem. 1995;270(36):21234-42.
9. Westenbroek RE, Sakurai T, Elliott EM, Hell JW, Starr TV, Snutch TP, et al. Immunochemical identification and subcellular distribution of the alpha 1A subunits of brain calcium channels. J Neurosci. 1995;15(10):6403-18.
10. Leenders AG, van den Maagdenberg AM, Lopes da Silva FH, Sheng ZH, Molenaar PC, Ghijsen WE. Neurotransmitter release from tottering mice nerve terminals with reduced expression of mutated P- and Q-type Ca2+-channels. Eur J Neurosci. 2002;15(1):13-8.
204
11. Kordasiewicz HB, Thompson RM, Clark HB, Gomez CM. C-termini of P/Q-type Ca2+ channel alpha1A subunits translocate to nuclei and promote polyglutamine-mediated toxicity. Hum Mol Genet. 2006;15(10):1587-99.
12. Ishiguro T, Ishikawa K, Takahashi M, Obayashi M, Amino T, Sato N, et al. The carboxy-terminal fragment of alpha(1A) calcium channel preferentially aggregates in the cytoplasm of human spinocerebellar ataxia type 6 Purkinje cells. Acta Neuropathol. 2010;119(4):447-64. PMCID: 2841749.
13. Simms BA, Zamponi GW. Neuronal voltage-gated calcium channels: structure, function, and dysfunction. Neuron. 2014;82(1):24-45.
14. Pietrobon D. CaV2.1 channelopathies. Pflugers Arch. 2010;460(2):375-93.
15. Baloh RW. Episodic ataxias 1 and 2. Handb Clin Neurol. 2012;103:595-602.
16. Berridge MJ, Lipp P, Bootman MD. The versatility and universality of calcium signalling. Nat Rev Mol Cell Biol. 2000;1(1):11-21.
17. Catterall WA. Structure and regulation of voltage-gated Ca2+ channels. Annu Rev Cell Dev Biol. 2000;16:521-55.
18. Nowycky MC, Fox AP, Tsien RW. Three types of neuronal calcium channel with different calcium agonist sensitivity. Nature. 1985;316(6027):440-3.
19. Nowycky MC, Fox AP, Tsien RW. Long-opening mode of gating of neuronal calcium channels and its promotion by the dihydropyridine calcium agonist Bay K 8644. Proc Natl Acad Sci U S A. 1985;82(7):2178-82. PMCID: 397516.
20. Fox AP, Nowycky MC, Tsien RW. Kinetic and pharmacological properties distinguishing three types of calcium currents in chick sensory neurones. J Physiol. 1987;394:149-72. PMCID: 1191955.
21. Fox AP, Nowycky MC, Tsien RW. Single-channel recordings of three types of calcium channels in chick sensory neurones. J Physiol. 1987;394:173-200. PMCID: 1191956.
22. Tsien RW, Lipscombe D, Madison DV, Bley KR, Fox AP. Multiple types of neuronal calcium channels and their selective modulation. Trends Neurosci. 1988;11(10):431-8.
23. Bourinet E, Soong TW, Sutton K, Slaymaker S, Mathews E, Monteil A, et al.
205
Splicing of alpha 1A subunit gene generates phenotypic variants of P- and Q-type calcium channels. Nat Neurosci. 1999;2(5):407-15.
24. Richards KS, Swensen AM, Lipscombe D, Bommert K. Novel CaV2.1 clone replicates many properties of Purkinje cell CaV2.1 current. Eur J Neurosci. 2007;26(10):2950-61.
25. Llinas R, Sugimori M, Lin JW, Cherksey B. Blocking and isolation of a calcium channel from neurons in mammals and cephalopods utilizing a toxin fraction (FTX) from funnel-web spider poison. Proc Natl Acad Sci U S A. 1989;86(5):1689-93. PMCID: 286766.
26. Randall A, Tsien RW. Pharmacological dissection of multiple types of Ca2+ channel currents in rat cerebellar granule neurons. J Neurosci. 1995;15(4):2995-3012.
27. Newcomb R, Szoke B, Palma A, Wang G, Chen X, Hopkins W, et al. Selective peptide antagonist of the class E calcium channel from the venom of the tarantula Hysterocrates gigas. Biochemistry. 1998;37(44):15353-62.
28. Kimm T, Bean BP. Inhibition of A-type potassium current by the peptide toxin SNX-482. J Neurosci. 2014;34(28):9182-9. PMCID: 4087201.
29. Dolphin AC. Voltage-gated calcium channels and their auxiliary subunits: physiology and pathophysiology and pharmacology. J Physiol. 2016.
30. Wu J, Yan Z, Li Z, Yan C, Lu S, Dong M, et al. Structure of the voltage-gated calcium channel Cav1.1 complex. Science. 2015;350(6267):aad2395.
31. Buraei Z, Yang J. The beta subunit of voltage-gated Ca2+ channels. Physiol Rev. 2010;90(4):1461-506. PMCID: 4353500.
32. Jay SD, Sharp AH, Kahl SD, Vedvick TS, Harpold MM, Campbell KP. Structural characterization of the dihydropyridine-sensitive calcium channel alpha 2-subunit and the associated delta peptides. J Biol Chem. 1991;266(5):3287-93.
33. Takahashi M, Seagar MJ, Jones JF, Reber BF, Catterall WA. Subunit structure of dihydropyridine-sensitive calcium channels from skeletal muscle. Proc Natl Acad Sci U S A. 1987;84(15):5478-82. PMCID: 298881.
34. Black JL, 3rd. The voltage-gated calcium channel gamma subunits: a review of the literature. J Bioenerg Biomembr. 2003;35(6):649-60.
206
35. Green PJ, Warre R, Hayes PD, McNaughton NC, Medhurst AD, Pangalos M, et al. Kinetic modification of the alpha(1I) subunit-mediated T-type Ca(2+) channel by a human neuronal Ca(2+) channel gamma subunit. J Physiol. 2001;533(Pt 2):467-78. PMCID: 2278624.
36. Bezanilla F. Voltage-gated ion channels. IEEE Trans Nanobioscience. 2005;4(1):34-48.
37. Catterall WA. Ion channel voltage sensors: structure, function, and pathophysiology. Neuron. 2010;67(6):915-28. PMCID: 2950829.
38. Yang J, Ellinor PT, Sather WA, Zhang JF, Tsien RW. Molecular determinants of Ca2+ selectivity and ion permeation in L-type Ca2+ channels. Nature. 1993;366(6451):158-61.
39. Perez-Reyes E, Schneider T. Molecular biology of calcium channels. Kidney Int. 1995;48(4):1111-24.
40. Chen RS, Deng TC, Garcia T, Sellers ZM, Best PM. Calcium channel gamma subunits: a functionally diverse protein family. Cell Biochem Biophys. 2007;47(2):178-86.
41. Craig PJ, McAinsh AD, McCormack AL, Smith W, Beattie RE, Priestley JV, et al. Distribution of the voltage-dependent calcium channel alpha(1A) subunit throughout the mature rat brain and its relationship to neurotransmitter pathways. J Comp Neurol. 1998;397(2):251-67.
42. Hillman D, Chen S, Aung TT, Cherksey B, Sugimori M, Llinas RR. Localization of P-type calcium channels in the central nervous system. Proc Natl Acad Sci U S A. 1991;88(16):7076-80. PMCID: 52236.
43. Llinas R, Sugimori M, Hillman DE, Cherksey B. Distribution and functional significance of the P-type, voltage-dependent Ca2+ channels in the mammalian central nervous system. Trends Neurosci. 1992;15(9):351-5.
44. Urbano FJ, Piedras-Renteria ES, Jun K, Shin HS, Uchitel OD, Tsien RW. Altered properties of quantal neurotransmitter release at endplates of mice lacking P/Q-type Ca2+ channels. Proc Natl Acad Sci U S A. 2003;100(6):3491-6. PMCID: 152320.
45. Jurkat-Rott K, Lehmann-Horn F. The impact of splice isoforms on voltage-gated calcium channel alpha1 subunits. J Physiol. 2004;554(Pt 3):609-19. PMCID: 1664792.
207
46. Mintz IM, Adams ME, Bean BP. P-type calcium channels in rat central and peripheral neurons. Neuron. 1992;9(1):85-95.
47. Wheeler DB, Randall A, Tsien RW. Roles of N-type and Q-type Ca2+ channels in supporting hippocampal synaptic transmission. Science. 1994;264(5155):107-11.
48. Sather WA, Tanabe T, Zhang JF, Mori Y, Adams ME, Tsien RW. Distinctive biophysical and pharmacological properties of class A (BI) calcium channel alpha 1 subunits. Neuron. 1993;11(2):291-303.
49. Westenbroek RE, Hell JW, Warner C, Dubel SJ, Snutch TP, Catterall WA. Biochemical properties and subcellular distribution of an N-type calcium channel alpha 1 subunit. Neuron. 1992;9(6):1099-115.
50. Henckels K. Mid-channel proteolysis of the L-type voltage gated calcium channel and the potential role of amyloid- (Sb (Bprecursor protein [Ph D]: Columbia University; 2014.
51. De Jongh KS, Murphy BJ, Colvin AA, Hell JW, Takahashi M, Catterall WA. Specific phosphorylation of a site in the full-length form of the alpha 1 subunit of the cardiac L-type calcium channel by adenosine 3',5'-cyclic monophosphate-dependent protein kinase. Biochemistry. 1996;35(32):10392-402.
52. Hulme JT, Westenbroek RE, Scheuer T, Catterall WA. Phosphorylation of serine 1928 in the distal C-terminal domain of cardiac CaV1.2 channels during beta1-adrenergic regulation. Proc Natl Acad Sci U S A. 2006;103(44):16574-9. PMCID: 1637623.
53. Yang L, Liu G, Zakharov SI, Morrow JP, Rybin VO, Steinberg SF, et al. Ser1928 is a common site for Cav1.2 phosphorylation by protein kinase C isoforms. J Biol Chem. 2005;280(1):207-14.
54. Gao T, Yatani A, Dell'Acqua ML, Sako H, Green SA, Dascal N, et al. cAMP-dependent regulation of cardiac L-type Ca2+ channels requires membrane targeting of PKA and phosphorylation of channel subunits. Neuron. 1997;19(1):185-96.
55. Fuller MD, Emrick MA, Sadilek M, Scheuer T, Catterall WA. Molecular mechanism of calcium channel regulation in the fight-or-flight response. Sci Signal. 2010;3(141):ra70. PMCID: 3063709.
56. Emrick MA, Sadilek M, Konoki K, Catterall WA. Beta-adrenergic-regulated
208 phosphorylation of the skeletal muscle Ca(V)1.1 channel in the fight-or-flight response. Proc Natl Acad Sci U S A. 2010;107(43):18712-7. PMCID: 2972969.
57. Dai S, Hall DD, Hell JW. Supramolecular assemblies and localized regulation of voltage-gated ion channels. Physiol Rev. 2009;89(2):411-52. PMCID: 2733249.
58. Mikami A, Imoto K, Tanabe T, Niidome T, Mori Y, Takeshima H, et al. Primary structure and functional expression of the cardiac dihydropyridine-sensitive calcium channel. Nature. 1989;340(6230):230-3.
59. Snutch TP, Tomlinson WJ, Leonard JP, Gilbert MM. Distinct calcium channels are generated by alternative splicing and are differentially expressed in the mammalian CNS. Neuron. 1991;7(1):45-57.
60. Navedo MF, Nieves-Cintron M, Amberg GC, Yuan C, Votaw VS, Lederer WJ, et al. AKAP150 is required for stuttering persistent Ca2+ sparklets and angiotensin II-induced hypertension. Circ Res. 2008;102(2):e1-e11.
61. Gross RA, Uhler MD, Macdonald RL. The cyclic AMP-dependent protein kinase catalytic subunit selectively enhances calcium currents in rat nodose neurones. J Physiol. 1990;429:483-96. PMCID: 1181711.
62. Gross RA, Macdonald RL. Cyclic AMP selectively reduces the N-type calcium current component of mouse sensory neurons in culture by enhancing inactivation. J Neurophysiol. 1989;61(1):97-105.
63. Surmeier DJ, Bargas J, Hemmings HC, Jr., Nairn AC, Greengard P. Modulation of calcium currents by a D1 dopaminergic protein kinase/phosphatase cascade in rat neostriatal neurons. Neuron. 1995;14(2):385-97.
64. Fournier F, Bourinet E, Nargeot J, Charnet P. Cyclic AMP-dependent regulation of P-type calcium channels expressed in Xenopus oocytes. Pflugers Arch. 1993;423(3-4):173-80.
65. Fukuda K, Kaneko S, Yada N, Kikuwaka M, Akaike A, Satoh M. Cyclic AMP-dependent modulation of N- and Q-type Ca2+ channels expressed in Xenopus oocytes. Neurosci Lett. 1996;217(1):13-6.
66. Swartz KJ. Modulation of Ca2+ channels by protein kinase C in rat central and peripheral neurons: disruption of G protein-mediated inhibition. Neuron. 1993;11(2):305-20.
209
67. Swartz KJ, Merritt A, Bean BP, Lovinger DM. Protein kinase C modulates glutamate receptor inhibition of Ca2+ channels and synaptic transmission. Nature. 1993;361(6408):165-8.
68. Doerner D, Abdel-Latif M, Rogers TB, Alger BE. Protein kinase C-dependent and -independent effects of phorbol esters on hippocampal calcium channel current. J Neurosci. 1990;10(5):1699-706.
69. Tamse CT, Xu Y, Song H, Nie L, Yamoah EN. Protein kinase A mediates voltage-dependent facilitation of Ca2+ current in presynaptic hair cells in Hermissenda crassicornis. J Neurophysiol. 2003;89(3):1718-26.
70. Solem M, McMahon T, Messing RO. Protein kinase A regulates regulates inhibition of N- and P/Q-type calcium channels by ethanol in PC12 cells. J Pharmacol Exp Ther. 1997;282(3):1487-95.
71. Wu L, Bauer CS, Zhen XG, Xie C, Yang J. Dual regulation of voltage-gated calcium channels by PtdIns(4,5)P2. Nature. 2002;419(6910):947-52.
72. Li Y, Wu Y, Zhou Y. Modulation of inactivation properties of CaV2.2 channels by 14-3-3 proteins. Neuron. 2006;51(6):755-71.
73. Zamponi GW, Bourinet E, Nelson D, Nargeot J, Snutch TP. Crosstalk between G proteins and protein kinase C mediated by the calcium channel alpha1 subunit. Nature. 1997;385(6615):442-6.
74. Hamid J, Nelson D, Spaetgens R, Dubel SJ, Snutch TP, Zamponi GW. Identification of an integration center for cross-talk between protein kinase C and G protein modulation of N-type calcium channels. J Biol Chem. 1999;274(10):6195-202.
75. Jarvis SE, Zamponi GW. Distinct molecular determinants govern syntaxin 1A-mediated inactivation and G-protein inhibition of N-type calcium channels. J Neurosci. 2001;21(9):2939-48.
76. Jarvis SE, Magga JM, Beedle AM, Braun JE, Zamponi GW. G protein modulation of N-type calcium channels is facilitated by physical interactions between syntaxin 1A and Gbetagamma. J Biol Chem. 2000;275(9):6388-94.
77. Yokoyama CT, Sheng ZH, Catterall WA. Phosphorylation of the synaptic protein interaction site on N-type calcium channels inhibits interactions with SNARE proteins. J Neurosci. 1997;17(18):6929-38.
210
78. Yokoyama CT, Myers SJ, Fu J, Mockus SM, Scheuer T, Catterall WA. Mechanism of SNARE protein binding and regulation of Cav2 channels by phosphorylation of the synaptic protein interaction site. Mol Cell Neurosci. 2005;28(1):1-17.
79. Hudmon A, Schulman H, Kim J, Maltez JM, Tsien RW, Pitt GS. CaMKII tethers to L-type Ca2+ channels, establishing a local and dedicated integrator of Ca2+ signals for facilitation. J Cell Biol. 2005;171(3):537-47. PMCID: 1343528.
80. Jiang X, Lautermilch NJ, Watari H, Westenbroek RE, Scheuer T, Catterall WA. Modulation of CaV2.1 channels by Ca2+/calmodulin-dependent protein kinase II bound to the C-terminal domain. Proc Natl Acad Sci U S A. 2008;105(1):341-6. PMCID: 2224214.
81. Abriel H, Staub O. Ubiquitylation of ion channels. Physiology (Bethesda). 2005;20:398-407.
82. Scott PM, Bilodeau PS, Zhdankina O, Winistorfer SC, Hauglund MJ, Allaman MM, et al. GGA proteins bind ubiquitin to facilitate sorting at the trans-Golgi network. Nat Cell Biol. 2004;6(3):252-9.
83. Puertollano R, Bonifacino JS. Interactions of GGA3 with the ubiquitin sorting machinery. Nat Cell Biol. 2004;6(3):244-51.
84. Altier C, Garcia-Caballero A, Simms B, You H, Chen L, Walcher J, et al. The Cavbeta subunit prevents RFP2-mediated ubiquitination and proteasomal degradation of L-type channels. Nat Neurosci. 2011;14(2):173-80.
85. Gandini MA, Sandoval A, Zamponi GW, Felix R. The MAP1B-LC1/UBE2L3 complex catalyzes degradation of cell surface CaV2.2 channels. Channels (Austin). 2014;8(5):452-7. PMCID: 4594605.
86. Pickart CM. Mechanisms underlying ubiquitination. Annu Rev Biochem. 2001;70:503-33.
87. Einbond A, Sudol M. Towards prediction of cognate complexes between the WW domain and proline-rich ligands. FEBS Lett. 1996;384(1):1-8.
88. Kasanov J, Pirozzi G, Uveges AJ, Kay BK. Characterizing Class I WW domains defines key specificity determinants and generates mutant domains with novel specificities. Chem Biol. 2001;8(3):231-41.
211
89. Staub O, Dho S, Henry P, Correa J, Ishikawa T, McGlade J, et al. WW domains of Nedd4 bind to the proline-rich PY motifs in the epithelial Na+ channel deleted in Liddle's syndrome. EMBO J. 1996;15(10):2371-80. PMCID: 450167.
90. Rotin D, Staub O, Haguenauer-Tsapis R. Ubiquitination and endocytosis of plasma membrane proteins: role of Nedd4/Rsp5p family of ubiquitin-protein ligases. J Membr Biol. 2000;176(1):1-17.
91. Jespersen T, Membrez M, Nicolas CS, Pitard B, Staub O, Olesen SP, et al. The KCNQ1 potassium channel is down-regulated by ubiquitylating enzymes of the Nedd4/Nedd4-like family. Cardiovasc Res. 2007;74(1):64-74.
92. Henke G, Maier G, Wallisch S, Boehmer C, Lang F. Regulation of the voltage gated K+ channel Kv1.3 by the ubiquitin ligase Nedd4-2 and the serum and glucocorticoid inducible kinase SGK1. J Cell Physiol. 2004;199(2):194-9.
93. Velez P, Schwartz AB, Iyer SR, Warrington A, Fadool DA. Ubiquitin ligase Nedd4-2 modulates Kv1.3 current amplitude and ion channel protein targeting. J Neurophysiol. 2016;116(2):671-85. PMCID: 4982906.
94. Gelman MS, Kopito RR. Cystic fibrosis: premature degradation of mutant proteins as a molecular disease mechanism. Methods Mol Biol. 2003;232:27-37.
95. Amerik AY, Hochstrasser M. Mechanism and function of deubiquitinating enzymes. Biochim Biophys Acta. 2004;1695(1-3):189-207.
96. Fakitsas P, Adam G, Daidie D, van Bemmelen MX, Fouladkou F, Patrignani A, et al. Early aldosterone-induced gene product regulates the epithelial sodium channel by deubiquitylation. J Am Soc Nephrol. 2007;18(4):1084-92.
97. Ruffieux-Daidie D, Poirot O, Boulkroun S, Verrey F, Kellenberger S, Staub O. Deubiquitylation regulates activation and proteolytic cleavage of ENaC. J Am Soc Nephrol. 2008;19(11):2170-80. PMCID: 2573013.
98. Krzystanek K, Rasmussen HB, Grunnet M, Staub O, Olesen SP, Abriel H, et al. Deubiquitylating enzyme USP2 counteracts Nedd4-2-mediated downregulation of KCNQ1 potassium channels. Heart Rhythm. 2012;9(3):440-8.
99. Rougier JS, Albesa M, Syam N, Halet G, Abriel H, Viard P. Ubiquitin-specific protease USP2-45 acts as a molecular switch to promote alpha2delta-1-induced downregulation of Cav1.2 channels. Pflugers Arch. 2015;467(9):1919-29. PMCID: 4537497.
212
100. Meir A, Bell DC, Stephens GJ, Page KM, Dolphin AC. Calcium channel beta subunit promotes voltage-dependent modulation of alpha 1 B by G beta gamma. Biophys J. 2000;79(2):731-46. PMCID: 1300973.
101. Neely A, Garcia-Olivares J, Voswinkel S, Horstkott H, Hidalgo P. Folding of active calcium channel beta(1b) -subunit by size-exclusion chromatography and its role on channel function. J Biol Chem. 2004;279(21):21689-94.
102. Bichet D, Cornet V, Geib S, Carlier E, Volsen S, Hoshi T, et al. The I-II loop of the Ca2+ channel alpha1 subunit contains an endoplasmic reticulum retention signal antagonized by the beta subunit. Neuron. 2000;25(1):177-90.
103. Kamp TJ, Perez-Garcia MT, Marban E. Enhancement of ionic current and charge movement by coexpression of calcium channel beta 1A subunit with alpha 1C subunit in a human embryonic kidney cell line. J Physiol. 1996;492 ( Pt 1):89-96. PMCID: 1158863.
104. Leroy J, Richards MW, Butcher AJ, Nieto-Rostro M, Pratt WS, Davies A, et al. Interaction via a key tryptophan in the I-II linker of N-type calcium channels is required for beta1 but not for palmitoylated beta2, implicating an additional binding site in the regulation of channel voltage-dependent properties. J Neurosci. 2005;25(30):6984-96.
105. Cornet V, Bichet D, Sandoz G, Marty I, Brocard J, Bourinet E, et al. Multiple determinants in voltage-dependent P/Q calcium channels control their retention in the endoplasmic reticulum. Eur J Neurosci. 2002;16(5):883-95.
106. Fang K, Colecraft HM. Mechanism of auxiliary beta-subunit-mediated membrane targeting of L-type (Ca(V)1.2) channels. J Physiol. 2011;589(Pt 18):4437-55. PMCID: 3208217.
107. Hu Z, Wang JW, Yu D, Soon JL, de Kleijn DP, Foo R, et al. Aberrant Splicing Promotes Proteasomal Degradation of L-type CaV1.2 Calcium Channels by Competitive Binding for CaVbeta Subunits in Cardiac Hypertrophy. Sci Rep. 2016;6:35247.
108. Rougier JS, Albesa M, Abriel H, Viard P. Neuronal precursor cell-expressed developmentally down-regulated 4-1 (NEDD4-1) controls the sorting of newly synthesized Ca(V)1.2 calcium channels. J Biol Chem. 2011;286(11):8829-38. PMCID: 3059038.
109. Catalucci D, Zhang DH, DeSantiago J, Aimond F, Barbara G, Chemin J, et al.
213
Akt regulates L-type Ca2+ channel activity by modulating Cavalpha1 protein stability. J Cell Biol. 2009;184(6):923-33. PMCID: 2699149.
110. Rechsteiner M, Rogers SW. PEST sequences and regulation by proteolysis. Trends Biochem Sci. 1996;21(7):267-71.
111. Canti C, Nieto-Rostro M, Foucault I, Heblich F, Wratten J, Richards MW, et al. The metal-ion-dependent adhesion site in the Von Willebrand factor-A domain of alpha2delta subunits is key to trafficking voltage-gated Ca2+ channels. Proc Natl Acad Sci U S A. 2005;102(32):11230-5. PMCID: 1183569.
112. Felix R, Gurnett CA, De Waard M, Campbell KP. Dissection of functional domains of the voltage-dependent Ca2+ channel alpha2delta subunit. J Neurosci. 1997;17(18):6884-91.
113. Gandini MA, Henriquez DR, Grimaldo L, Sandoval A, Altier C, Zamponi GW, et al. CaV2.2 channel cell surface expression is regulated by the light chain 1 (LC1) of the microtubule-associated protein B (MAP1B) via UBE2L3-mediated ubiquitination and degradation. Pflugers Arch. 2014;466(11):2113-26.
114. Marangoudakis S, Andrade A, Helton TD, Denome S, Castiglioni AJ, Lipscombe D. Differential ubiquitination and proteasome regulation of Ca(V)2.2 N-type channel splice isoforms. J Neurosci. 2012;32(30):10365-9. PMCID: 3428229.
115. Page KM, Rothwell SW, Dolphin AC. The CaVbeta Subunit Protects the I-II Loop of the Voltage-gated Calcium Channel CaV2.2 from Proteasomal Degradation but Not Oligoubiquitination. J Biol Chem. 2016;291(39):20402-16. PMCID: 5034038.
116. Waithe D, Ferron L, Page KM, Chaggar K, Dolphin AC. Beta-subunits promote the expression of Ca(V)2.2 channels by reducing their proteasomal degradation. J Biol Chem. 2011;286(11):9598-611. PMCID: 3059031.
117. Bell TJ, Thaler C, Castiglioni AJ, Helton TD, Lipscombe D. Cell-specific alternative splicing increases calcium channel current density in the pain pathway. Neuron. 2004;41(1):127-38.
118. Opatowsky Y, Chen CC, Campbell KP, Hirsch JA. Structural analysis of the voltage-dependent calcium channel beta subunit functional core and its complex with the alpha 1 interaction domain. Neuron. 2004;42(3):387-99.
119. Arias JM, Murbartian J, Vitko I, Lee JH, Perez-Reyes E. Transfer of beta subunit
214 regulation from high to low voltage-gated Ca2+ channels. FEBS Lett. 2005;579(18):3907-12.
120. Van Petegem F, Clark KA, Chatelain FC, Minor DL, Jr. Structure of a complex between a voltage-gated calcium channel beta-subunit and an alpha-subunit domain. Nature. 2004;429(6992):671-5. PMCID: 3076333.
121. Tanabe T, Takeshima H, Mikami A, Flockerzi V, Takahashi H, Kangawa K, et al. Primary structure of the receptor for calcium channel blockers from skeletal muscle. Nature. 1987;328(6128):313-8.
122. De Jongh KS, Merrick DK, Catterall WA. Subunits of purified calcium channels: a 212-kDa form of alpha 1 and partial amino acid sequence of a phosphorylation site of an independent beta subunit. Proc Natl Acad Sci U S A. 1989;86(21):8585-9. PMCID: 298327.
123. De Jongh KS, Warner C, Colvin AA, Catterall WA. Characterization of the two size forms of the alpha 1 subunit of skeletal muscle L-type calcium channels. Proc Natl Acad Sci U S A. 1991;88(23):10778-82. PMCID: 53014.
124. Lai Y, Seagar MJ, Takahashi M, Catterall WA. Cyclic AMP-dependent phosphorylation of two size forms of alpha 1 subunits of L-type calcium channels in rat skeletal muscle cells. J Biol Chem. 1990;265(34):20839-48.
125. Hell JW, Westenbroek RE, Warner C, Ahlijanian MK, Prystay W, Gilbert MM, et al. Identification and differential subcellular localization of the neuronal class C and class D L-type calcium channel alpha 1 subunits. J Cell Biol. 1993;123(4):949-62. PMCID: 2200142.
126. Hell JW, Yokoyama CT, Wong ST, Warner C, Snutch TP, Catterall WA. Differential phosphorylation of two size forms of the neuronal class C L-type calcium channel alpha 1 subunit. J Biol Chem. 1993;268(26):19451-7.
127. Gao T, Puri TS, Gerhardstein BL, Chien AJ, Green RD, Hosey MM. Identification and subcellular localization of the subunits of L-type calcium channels and adenylyl cyclase in cardiac myocytes. J Biol Chem. 1997;272(31):19401-7.
128. Chien AJ, Zhao X, Shirokov RE, Puri TS, Chang CF, Sun D, et al. Roles of a membrane-localized beta subunit in the formation and targeting of functional L-type Ca2+ channels. J Biol Chem. 1995;270(50):30036-44.
129. Puri TS, Gerhardstein BL, Zhao XL, Ladner MB, Hosey MM. Differential
215 effects of subunit interactions on protein kinase A- and C-mediated phosphorylation of L-type calcium channels. Biochemistry. 1997;36(31):9605-15.
130. Hulme JT, Konoki K, Lin TW, Gritsenko MA, Camp DG, 2nd, Bigelow DJ, et al. Sites of proteolytic processing and noncovalent association of the distal C-terminal domain of CaV1.1 channels in skeletal muscle. Proc Natl Acad Sci U S A. 2005;102(14):5274-9. PMCID: 555994.
131. Hell JW, Westenbroek RE, Elliott EM, Catterall WA. Differential phosphorylation, localization, and function of distinct alpha 1 subunits of neuronal calcium channels. Two size forms for class B, C, and D alpha 1 subunits with different COOH-termini. Ann N Y Acad Sci. 1994;747:282-93.
132. Hell JW, Yokoyama CT, Breeze LJ, Chavkin C, Catterall WA. Phosphorylation of presynaptic and postsynaptic calcium channels by cAMP-dependent protein kinase in hippocampal neurons. EMBO J. 1995;14(13):3036-44. PMCID: 394364.
133. Rotman EI, De Jongh KS, Florio V, Lai Y, Catterall WA. Specific phosphorylation of a COOH-terminal site on the full-length form of the alpha 1 subunit of the skeletal muscle calcium channel by cAMP-dependent protein kinase. J Biol Chem. 1992;267(23):16100-5.
134. Rotman EI, Murphy BJ, Catterall WA. Sites of selective cAMP-dependent phosphorylation of the L-type calcium channel alpha 1 subunit from intact rabbit skeletal muscle myotubes. J Biol Chem. 1995;270(27):16371-7.
135. De Jongh KS, Colvin AA, Wang KK, Catterall WA. Differential proteolysis of the full-length form of the L-type calcium channel alpha 1 subunit by calpain. J Neurochem. 1994;63(4):1558-64.
136. Hell JW, Westenbroek RE, Breeze LJ, Wang KK, Chavkin C, Catterall WA. N-methyl-D-aspartate receptor-induced proteolytic conversion of postsynaptic class C L-type calcium channels in hippocampal neurons. Proc Natl Acad Sci U S A. 1996;93(8):3362-7. PMCID: 39613.
137. Gerhardstein BL, Gao T, Bunemann M, Puri TS, Adair A, Ma H, et al. Proteolytic processing of the C terminus of the alpha(1C) subunit of L-type calcium channels and the role of a proline-rich domain in membrane tethering of proteolytic fragments. J Biol Chem. 2000;275(12):8556-63.
138. Schroder E, Byse M, Satin J. L-type calcium channel C terminus
216 autoregulates transcription. Circ Res. 2009;104(12):1373-81. PMCID: 2728034.
139. Wei X, Neely A, Lacerda AE, Olcese R, Stefani E, Perez-Reyes E, et al. Modification of Ca2+ channel activity by deletions at the carboxyl terminus of the cardiac alpha 1 subunit. J Biol Chem. 1994;269(3):1635-40.
140. Gao T, Cuadra AE, Ma H, Bunemann M, Gerhardstein BL, Cheng T, et al. C-terminal fragments of the alpha 1C (CaV1.2) subunit associate with and regulate L-type calcium channels containing C-terminal-truncated alpha 1C subunits. J Biol Chem. 2001;276(24):21089-97.
141. Hulme JT, Yarov-Yarovoy V, Lin TW, Scheuer T, Catterall WA. Autoinhibitory control of the CaV1.2 channel by its proteolytically processed distal C-terminal domain. J Physiol. 2006;576(Pt 1):87-102. PMCID: 1995633.
142. Hulme JT, Ahn M, Hauschka SD, Scheuer T, Catterall WA. A novel leucine zipper targets AKAP15 and cyclic AMP-dependent protein kinase to the C terminus of the skeletal muscle Ca2+ channel and modulates its function. J Biol Chem. 2002;277(6):4079-87.
143. Hulme JT, Lin TW, Westenbroek RE, Scheuer T, Catterall WA. Beta-adrenergic regulation requires direct anchoring of PKA to cardiac CaV1.2 channels via a leucine zipper interaction with A kinase-anchoring protein 15. Proc Natl Acad Sci U S A. 2003;100(22):13093-8. PMCID: 240750.
144. Fu Y, Westenbroek RE, Yu FH, Clark JP, 3rd, Marshall MR, Scheuer T, et al. Deletion of the distal C terminus of CaV1.2 channels leads to loss of beta-adrenergic regulation and heart failure in vivo. J Biol Chem. 2011;286(14):12617-26. PMCID: 3069462.
145. Fu Y, Westenbroek RE, Scheuer T, Catterall WA. Phosphorylation sites required for regulation of cardiac calcium channels in the fight-or-flight response. Proc Natl Acad Sci U S A. 2013;110(48):19621-6. PMCID: 3845157.
146. Fu Y, Westenbroek RE, Scheuer T, Catterall WA. Basal and beta-adrenergic regulation of the cardiac calcium channel CaV1.2 requires phosphorylation of serine 1700. Proc Natl Acad Sci U S A. 2014;111(46):16598-603. PMCID: 4246329.
147. Orrenius S, Zhivotovsky B, Nicotera P. Regulation of cell death: the calcium-apoptosis link. Nat Rev Mol Cell Biol. 2003;4(7):552-65.
148. Gomez-Ospina N, Tsuruta F, Barreto-Chang O, Hu L, Dolmetsch R. The C
217 terminus of the L-type voltage-gated calcium channel Ca(V)1.2 encodes a transcription factor. Cell. 2006;127(3):591-606. PMCID: 1750862.
149. Moyer JR, Jr., Thompson LT, Black JP, Disterhoft JF. Nimodipine increases excitability of rabbit CA1 pyramidal neurons in an age- and concentration-dependent manner. J Neurophysiol. 1992;68(6):2100-9.
150. Thibault O, Landfield PW. Increase in single L-type calcium channels in hippocampal neurons during aging. Science. 1996;272(5264):1017-20.
151. Thibault O, Porter NM, Chen KC, Blalock EM, Kaminker PG, Clodfelter GV, et al. Calcium dysregulation in neuronal aging and Alzheimer's disease: history and new directions. Cell Calcium. 1998;24(5-6):417-33.
152. Ahern CA, Arikkath J, Vallejo P, Gurnett CA, Powers PA, Campbell KP, et al. Intramembrane charge movements and excitation- contraction coupling expressed by two-domain fragments of the Ca2+ channel. Proc Natl Acad Sci U S A. 2001;98(12):6935-40. PMCID: 34456.
153. Kruger LC, Isom LL. Voltage-Gated Na+ Channels: Not Just for Conduction. Cold Spring Harb Perspect Biol. 2016;8(6).
154. Paillart C, Boudier JL, Boudier JA, Rochat H, Couraud F, Dargent B. Activity-induced internalization and rapid degradation of sodium channels in cultured fetal neurons. J Cell Biol. 1996;134(2):499-509. PMCID: 2120887.
155. Iwata A, Stys PK, Wolf JA, Chen XH, Taylor AG, Meaney DF, et al. Traumatic axonal injury induces proteolytic cleavage of the voltage-gated sodium channels modulated by tetrodotoxin and protease inhibitors. J Neurosci. 2004;24(19):4605-13.
156. Jette N, Coderre E, Nikolaeva MA, Enright PD, Iwata A, Smith DH, et al. Spatiotemporal distribution of spectrin breakdown products induced by anoxia in adult rat optic nerve in vitro. J Cereb Blood Flow Metab. 2006;26(6):777-86.
157. Onwuli DO, Beltran-Alvarez P. An update on transcriptional and post-translational regulation of brain voltage-gated sodium channels. Amino Acids. 2016;48(3):641-51. PMCID: 4752963.
158. von Reyn CR, Spaethling JM, Mesfin MN, Ma M, Neumar RW, Smith DH, et al. Calpain mediates proteolysis of the voltage-gated sodium channel alpha-subunit. J Neurosci. 2009;29(33):10350-6. PMCID: 2799681.
218
159. von Reyn CR, Mott RE, Siman R, Smith DH, Meaney DF. Mechanisms of calpain mediated proteolysis of voltage gated sodium channel alpha-subunits following in vitro dynamic stretch injury. J Neurochem. 2012;121(5):793-805.
160. Vassilev PM, Scheuer T, Catterall WA. Identification of an intracellular peptide segment involved in sodium channel inactivation. Science. 1988;241(4873):1658-61.
161. Stuhmer W, Conti F, Suzuki H, Wang XD, Noda M, Yahagi N, et al. Structural parts involved in activation and inactivation of the sodium channel. Nature. 1989;339(6226):597-603.
162. Benz I, Beck W, Kraas W, Stoll D, Jung G, Kohlhardt M. Two types of modified cardiac Na+ channels after cytosolic interventions at the alpha-subunit capable of removing Na+ inactivation. Eur Biophys J. 1997;25(3):189-200.
163. Brocard C, Plantier V, Boulenguez P, Liabeuf S, Bouhadfane M, Viallat-Lieutaud A, et al. Cleavage of Na(+) channels by calpain increases persistent Na(+) current and promotes spasticity after spinal cord injury. Nat Med. 2016;22(4):404-11.
164. Li Y, Gorassini MA, Bennett DJ. Role of persistent sodium and calcium currents in motoneuron firing and spasticity in chronic spinal rats. J Neurophysiol. 2004;91(2):767-83.
165. Gorassini MA, Knash ME, Harvey PJ, Bennett DJ, Yang JF. Role of motoneurons in the generation of muscle spasms after spinal cord injury. Brain. 2004;127(Pt 10):2247-58.
166. Higuchi M, Tomioka M, Takano J, Shirotani K, Iwata N, Masumoto H, et al. Distinct mechanistic roles of calpain and caspase activation in neurodegeneration as revealed in mice overexpressing their specific inhibitors. J Biol Chem. 2005;280(15):15229-37.
167. Schoch KM, Evans HN, Brelsfoard JM, Madathil SK, Takano J, Saido TC, et al. Calpastatin overexpression limits calpain-mediated proteolysis and behavioral deficits following traumatic brain injury. Exp Neurol. 2012;236(2):371-82. PMCID: 3392428.
168. Schoch KM, von Reyn CR, Bian J, Telling GC, Meaney DF, Saatman KE. Brain injury-induced proteolysis is reduced in a novel calpastatin-overexpressing transgenic mouse. J Neurochem. 2013;125(6):909-20. PMCID: 3676438.
219
169. Kharatmal SB, Singh JN, Sharma SS. Calpain inhibitor, MDL 28170 confer electrophysiological, nociceptive and biochemical improvement in diabetic neuropathy. Neuropharmacology. 2015;97:113-21.
170. Beltran-Alvarez P, Tarradas A, Chiva C, Perez-Serra A, Batlle M, Perez-Villa F, et al. Identification of N-terminal protein acetylation and arginine methylation of the voltage-gated sodium channel in end-stage heart failure human heart. J Mol Cell Cardiol. 2014;76:126-9.
171. Cox JJ, Reimann F, Nicholas AK, Thornton G, Roberts E, Springell K, et al. An SCN9A channelopathy causes congenital inability to experience pain. Nature. 2006;444(7121):894-8.
172. Cox JJ, Sheynin J, Shorer Z, Reimann F, Nicholas AK, Zubovic L, et al. Congenital insensitivity to pain: novel SCN9A missense and in-frame deletion mutations. Hum Mutat. 2010;31(9):E1670-86. PMCID: 2966863.
173. Remacle AG, Kumar S, Motamedchaboki K, Cieplak P, Hullugundi S, Dolkas J, et al. Matrix Metalloproteinase (MMP) Proteolysis of the Extracellular Loop of Voltage-gated Sodium Channels and Potential Alterations in Pain Signaling. J Biol Chem. 2015;290(38):22939-44. PMCID: 4645627.
174. Moore KB, O'Brien J. Connexins in neurons and glia: targets for intervention in disease and injury. Neural Regen Res. 2015;10(7):1013-7. PMCID: 4541216.
175. Ge J, Ju Y, Xue Z, Feng Y, Huang X, Liu H, et al. Distal C terminus of CaV1.2 channels plays a crucial role in the neural differentiation of dental pulp stem cells. PLoS One. 2013;8(11):e81332. PMCID: 3836819.
176. Woppmann A, Ramachandran J, Miljanich GP. Calcium channel subtypes in rat brain: biochemical characterization of the high-affinity receptors for omega-conopeptides SNX-230 (synthetic MVIIC), SNX-183 (SVIB), and SNX-111 (MVIIA). Mol Cell Neurosci. 1994;5(4):350-7.
177. Kubodera T, Yokota T, Ohwada K, Ishikawa K, Miura H, Matsuoka T, et al. Proteolytic cleavage and cellular toxicity of the human alpha1A calcium channel in spinocerebellar ataxia type 6. Neurosci Lett. 2003;341(1):74-8.
178. Mezghrani A, Monteil A, Watschinger K, Sinnegger-Brauns MJ, Barrere C, Bourinet E, et al. A destructive interaction mechanism accounts for dominant-negative effects of misfolded mutants of voltage-gated calcium channels. J Neurosci. 2008;28(17):4501-11.
220
179. Snutch TP, Leonard JP, Gilbert MM, Lester HA, Davidson N. Rat brain expresses a heterogeneous family of calcium channels. Proc Natl Acad Sci U S A. 1990;87(9):3391-5. PMCID: 53906.
180. Plummer NW, McBurney MW, Meisler MH. Alternative splicing of the sodium channel SCN8A predicts a truncated two-domain protein in fetal brain and non-neuronal cells. J Biol Chem. 1997;272(38):24008-15.
181. Moran Y, Barzilai MG, Liebeskind BJ, Zakon HH. Evolution of voltage-gated ion channels at the emergence of Metazoa. J Exp Biol. 2015;218(Pt 4):515-25.
182. Malouf NN, McMahon DK, Hainsworth CN, Kay BK. A two-motif isoform of the major calcium channel subunit in skeletal muscle. Neuron. 1992;8(5):899-906.
183. Zhang C, Zhou Z. Ca(2+)-independent but voltage-dependent secretion in mammalian dorsal root ganglion neurons. Nat Neurosci. 2002;5(5):425-30.
184. Hirasawa M, Kombian SB, Pittman QJ. Oxytocin retrogradely inhibits evoked, but not miniature, EPSCs in the rat supraoptic nucleus: role of N- and P/Q-type calcium channels. J Physiol. 2001;532(Pt 3):595-607. PMCID: 2278576.
185. Okagaki R, Izumi H, Okada T, Nagahora H, Nakajo K, Okamura Y. The maternal transcript for truncated voltage-dependent Ca2+ channels in the ascidian embryo: a potential suppressive role in Ca2+ channel expression. Dev Biol. 2001;230(2):258-77.
186. Ebihara T, Komiya Y, Izumi-Nakaseko H, Adachi-Akahane S, Okabe S, Okamura Y. Coexpression of a Ca(v)1.2 protein lacking an N-terminus and the first domain specifically suppresses L-type calcium channel activity. FEBS Lett. 2002;529(2-3):203-7.
187. Raghib A, Bertaso F, Davies A, Page KM, Meir A, Bogdanov Y, et al. Dominant-negative synthesis suppression of voltage-gated calcium channel Cav2.2 induced by truncated constructs. J Neurosci. 2001;21(21):8495-504.
188. Page KM, Heblich F, Davies A, Butcher AJ, Leroy J, Bertaso F, et al. Dominant-negative calcium channel suppression by truncated constructs involves a kinase implicated in the unfolded protein response. J Neurosci. 2004;24(23):5400-9.
189. Page KM, Heblich F, Margas W, Pratt WS, Nieto-Rostro M, Chaggar K, et al. N terminus is key to the dominant negative suppression of Ca(V)2 calcium channels:
221 implications for episodic ataxia type 2. J Biol Chem. 2010;285(2):835-44. PMCID: 2801285.
190. Dahimene S, Page KM, Nieto-Rostro M, Pratt WS, D'Arco M, Dolphin AC. A Ca2.1 N-terminal fragment relieves the dominant-negative inhibition by an Episodic ataxia 2 mutant. Neurobiol Dis. 2016;93:243-56.
191. Jiang M, Tseng-Crank J, Tseng GN. Suppression of slow delayed rectifier current by a truncated isoform of KvLQT1 cloned from normal human heart. J Biol Chem. 1997;272(39):24109-12.
192. Zhuchenko O, Bailey J, Bonnen P, Ashizawa T, Stockton DW, Amos C, et al. Autosomal dominant cerebellar ataxia (SCA6) associated with small polyglutamine expansions in the alpha 1A-voltage-dependent calcium channel. Nat Genet. 1997;15(1):62-9.
193. Marqueze-Pouey B, Martin-Moutot N, Sakkou-Norton M, Leveque C, Ji Y, Cornet V, et al. Toxicity and endocytosis of spinocerebellar ataxia type 6 polyglutamine domains: role of myosin IIb. Traffic. 2008;9(7):1088-100.
194. Subramony SH, Schott K, Raike RS, Callahan J, Langford LR, Christova PS, et al. Novel CACNA1A mutation causes febrile episodic ataxia with interictal cerebellar deficits. Ann Neurol. 2003;54(6):725-31.
195. Raike RS, Kordasiewicz HB, Thompson RM, Gomez CM. Dominant-negative suppression of Cav2.1 currents by alpha(1)2.1 truncations requires the conserved interaction domain for beta subunits. Mol Cell Neurosci. 2007;34(2):168-77. PMCID: 3236250.
196. Jen J, Kim GW, Baloh RW. Clinical spectrum of episodic ataxia type 2. Neurology. 2004;62(1):17-22.
197. van den Maagdenberg AM, Kors EE, Brunt ER, van Paesschen W, Pascual J, Ravine D, et al. Episodic ataxia type 2. Three novel truncating mutations and one novel missense mutation in the CACNA1A gene. J Neurol. 2002;249(11):1515-9.
198. Denier C, Ducros A, Vahedi K, Joutel A, Thierry P, Ritz A, et al. High prevalence of CACNA1A truncations and broader clinical spectrum in episodic ataxia type 2. Neurology. 1999;52(9):1816-21.
199. Jeng CJ, Sun MC, Chen YW, Tang CY. Dominant-negative effects of episodic ataxia type 2 mutations involve disruption of membrane trafficking of human
222
P/Q-type Ca2+ channels. J Cell Physiol. 2008;214(2):422-33.
200. Scoggan KA, Chandra T, Nelson R, Hahn AF, Bulman DE. Identification of two novel mutations in the CACNA1A gene responsible for episodic ataxia type 2. J Med Genet. 2001;38(4):249-53. PMCID: 1734855.
201. Ophoff RA, Terwindt GM, Vergouwe MN, van Eijk R, Oefner PJ, Hoffman SM, et al. Familial hemiplegic migraine and episodic ataxia type-2 are caused by mutations in the Ca2+ channel gene CACNL1A4. Cell. 1996;87(3):543-52.
202. Jen J, Wan J, Graves M, Yu H, Mock AF, Coulin CJ, et al. Loss-of-function EA2 mutations are associated with impaired neuromuscular transmission. Neurology. 2001;57(10):1843-8.
203. Jeng CJ, Chen YT, Chen YW, Tang CY. Dominant-negative effects of human P/Q-type Ca2+ channel mutations associated with episodic ataxia type 2. Am J Physiol Cell Physiol. 2006;290(4):C1209-20.
204. Wappl E, Koschak A, Poteser M, Sinnegger MJ, Walter D, Eberhart A, et al. Functional consequences of P/Q-type Ca2+ channel Cav2.1 missense mutations associated with episodic ataxia type 2 and progressive ataxia. J Biol Chem. 2002;277(9):6960-6.
205. Yue Q, Jen JC, Thwe MM, Nelson SF, Baloh RW. De novo mutation in CACNA1A caused acetazolamide-responsive episodic ataxia. Am J Med Genet. 1998;77(4):298-301.
206. Jen J, Yue Q, Nelson SF, Yu H, Litt M, Nutt J, et al. A novel nonsense mutation in CACNA1A causes episodic ataxia and hemiplegia. Neurology. 1999;53(1):34-7.
207. Spacey SD, Materek LA, Szczygielski BI, Bird TD. Two novel CACNA1A gene mutations associated with episodic ataxia type 2 and interictal dystonia. Arch Neurol. 2005;62(2):314-6.
208. Imbrici P, Jaffe SL, Eunson LH, Davies NP, Herd C, Robertson R, et al. Dysfunction of the brain calcium channel CaV2.1 in absence epilepsy and episodic ataxia. Brain. 2004;127(Pt 12):2682-92.
209. Jouvenceau A, Eunson LH, Spauschus A, Ramesh V, Zuberi SM, Kullmann DM, et al. Human epilepsy associated with dysfunction of the brain P/Q-type calcium channel. Lancet. 2001;358(9284):801-7.
223
210. Jun K, Piedras-Renteria ES, Smith SM, Wheeler DB, Lee SB, Lee TG, et al. Ablation of P/Q-type Ca(2+) channel currents, altered synaptic transmission, and progressive ataxia in mice lacking the alpha(1A)-subunit. Proc Natl Acad Sci U S A. 1999;96(26):15245-50. PMCID: 24805.
211. Cao YQ, Piedras-Renteria ES, Smith GB, Chen G, Harata NC, Tsien RW. Presynaptic Ca2+ channels compete for channel type-preferring slots in altered neurotransmission arising from Ca2+ channelopathy. Neuron. 2004;43(3):387-400.
212. Denning GM, Anderson MP, Amara JF, Marshall J, Smith AE, Welsh MJ. Processing of mutant cystic fibrosis transmembrane conductance regulator is temperature-sensitive. Nature. 1992;358(6389):761-4.
213. Morello JP, Petaja-Repo UE, Bichet DG, Bouvier M. Pharmacological chaperones: a new twist on receptor folding. Trends Pharmacol Sci. 2000;21(12):466-9.
214. Harding HP, Calfon M, Urano F, Novoa I, Ron D. Transcriptional and translational control in the Mammalian unfolded protein response. Annu Rev Cell Dev Biol. 2002;18:575-99.
215. Mazumder B, Seshadri V, Fox PL. Translational control by the 3'-UTR: the ends specify the means. Trends Biochem Sci. 2003;28(2):91-8.
216. Page KM, Canti C, Stephens GJ, Berrow NS, Dolphin AC. Identification of the amino terminus of neuronal Ca2+ channel alpha1 subunits alpha1B and alpha1E as an essential determinant of G-protein modulation. J Neurosci. 1998;18(13):4815-24.
217. Canti C, Page KM, Stephens GJ, Dolphin AC. Identification of residues in the N terminus of alpha1B critical for inhibition of the voltage-dependent calcium channel by Gbeta gamma. J Neurosci. 1999;19(16):6855-64.
218. Soong TW, DeMaria CD, Alvania RS, Zweifel LS, Liang MC, Mittman S, et al. Systematic identification of splice variants in human P/Q-type channel alpha1(2.1) subunits: implications for current density and Ca2+-dependent inactivation. J Neurosci. 2002;22(23):10142-52.
219. Gray AC, Raingo J, Lipscombe D. Neuronal calcium channels: splicing for optimal performance. Cell Calcium. 2007;42(4-5):409-17. PMCID: 2001240.
220. Tsunemi T, Saegusa H, Ishikawa K, Nagayama S, Murakoshi T, Mizusawa H,
224 et al. Novel Cav2.1 splice variants isolated from Purkinje cells do not generate P-type Ca2+ current. J Biol Chem. 2002;277(9):7214-21.
221. Sakurai T, Westenbroek RE, Rettig J, Hell J, Catterall WA. Biochemical properties and subcellular distribution of the BI and rbA isoforms of alpha 1A subunits of brain calcium channels. J Cell Biol. 1996;134(2):511-28. PMCID: 2120867.
222. Rettig J, Sheng ZH, Kim DK, Hodson CD, Snutch TP, Catterall WA. Isoform-specific interaction of the alpha1A subunits of brain Ca2+ channels with the presynaptic proteins syntaxin and SNAP-25. Proc Natl Acad Sci U S A. 1996;93(14):7363-8. PMCID: 38990.
223. Oguro-Okano M, Griesmann GE, Wieben ED, Slaymaker SJ, Snutch TP, Lennon VA. Molecular diversity of neuronal-type calcium channels identified in small cell lung carcinoma. Mayo Clin Proc. 1992;67(12):1150-9.
224. Rajapaksha WR, Wang D, Davies JN, Chen L, Zamponi GW, Fisher TE. Novel splice variants of rat CaV2.1 that lack much of the synaptic protein interaction site are expressed in neuroendocrine cells. J Biol Chem. 2008;283(23):15997-6003. PMCID: 3259657.
225. Barrett CF, Cao YQ, Tsien RW. Gating deficiency in a familial hemiplegic migraine type 1 mutant P/Q-type calcium channel. J Biol Chem. 2005;280(25):24064-71.
226. Brawley RM, Hosey MM. Identification of two distinct proteins that are immunologically related to the alpha 1 subunit of the skeletal muscle dihydropyridine-sensitive calcium channel. J Biol Chem. 1992;267(25):18218-23.
227. Schiff ML, Siderovski DP, Jordan JD, Brothers G, Snow B, De Vries L, et al. Tyrosine-kinase-dependent recruitment of RGS12 to the N-type calcium channel. Nature. 2000;408(6813):723-7.
228. Hell JW, Appleyard SM, Yokoyama CT, Warner C, Catterall WA. Differential phosphorylation of two size forms of the N-type calcium channel alpha 1 subunit which have different COOH termini. J Biol Chem. 1994;269(10):7390-6.
229. Leenders AG, Lin L, Huang LD, Gerwin C, Lu PH, Sheng ZH. The role of MAP1A light chain 2 in synaptic surface retention of Cav2.2 channels in hippocampal neurons. J Neurosci. 2008;28(44):11333-46. PMCID: 2596948.
225
230. Ramakrishnan NA, Drescher MJ, Sheikhali SA, Khan KM, Hatfield JS, Dickson MJ, et al. Molecular identification of an N-type Ca2+ channel in saccular hair cells. Neuroscience. 2006;139(4):1417-34.
231. Lazebnik YA, Kaufmann SH, Desnoyers S, Poirier GG, Earnshaw WC. Cleavage of poly(ADP-ribose) polymerase by a proteinase with properties like ICE. Nature. 1994;371(6495):346-7.
232. Nicholson DW, Ali A, Thornberry NA, Vaillancourt JP, Ding CK, Gallant M, et al. Identification and inhibition of the ICE/CED-3 protease necessary for mammalian apoptosis. Nature. 1995;376(6535):37-43.
233. Oliver FJ, de la Rubia G, Rolli V, Ruiz-Ruiz MC, de Murcia G, Murcia JM. Importance of poly(ADP-ribose) polymerase and its cleavage in apoptosis. Lesson from an uncleavable mutant. J Biol Chem. 1998;273(50):33533-9.
234. Shapiro MB, Senapathy P. RNA splice junctions of different classes of eukaryotes: sequence statistics and functional implications in gene expression. Nucleic Acids Res. 1987;15(17):7155-74. PMCID: 306199.
235. Senapathy P, Shapiro MB, Harris NL. Splice junctions, branch point sites, and exons: sequence statistics, identification, and applications to genome project. Methods Enzymol. 1990;183:252-78.
236. Yeo G, Burge CB. Maximum entropy modeling of short sequence motifs with applications to RNA splicing signals. J Comput Biol. 2004;11(2-3):377-94.
237. Chen CM, Chiang SY, Yeh NH. Increased stability of nucleolin in proliferating cells by inhibition of its self-cleaving activity. J Biol Chem. 1991;266(12):7754-8.
238. Svensson M, Boren M, Skold K, Falth M, Sjogren B, Andersson M, et al. Heat stabilization of the tissue proteome: a new technology for improved proteomics. J Proteome Res. 2009;8(2):974-81.
239. Havens KA, Guseman JM, Jang SS, Pierre-Jerome E, Bolten N, Klavins E, et al. A synthetic approach reveals extensive tunability of auxin signaling. Plant Physiol. 2012;160(1):135-42. PMCID: 3440190.
240. Dolan PJ, Johnson GV. A caspase cleaved form of tau is preferentially degraded through the autophagy pathway. J Biol Chem. 2010;285(29):21978-87. PMCID: 2903354.
226
241. Geng J, Xia L, Li W, Dou F. The C-terminus of tau protein plays an important role in its stability and toxicity. J Mol Neurosci. 2015;55(1):251-9.
242. Lichius A, Bidard F, Buchholz F, Le Crom S, Martin J, Schackwitz W, et al. Genome sequencing of the Trichoderma reesei QM9136 mutant identifies a truncation of the transcriptional regulator XYR1 as the cause for its cellulase-negative phenotype. BMC Genomics. 2015;16:326. PMCID: 4409711.
243. Ben Messaoud N, Yue J, Valent D, Katzarova I, Lopez JM. Osmostress-induced apoptosis in Xenopus oocytes: role of stress protein kinases, calpains and Smac/DIABLO. PLoS One. 2015;10(4):e0124482. PMCID: 4395108.
244. Lawal AO, Ellis EM. Phospholipase C mediates cadmium-dependent apoptosis in HEK 293 cells. Basic Clin Pharmacol Toxicol. 2012;110(6):510-7.
245. Lorca T, Galas S, Fesquet D, Devault A, Cavadore JC, Doree M. Degradation of the proto-oncogene product p39mos is not necessary for cyclin proteolysis and exit from meiotic metaphase: requirement for a Ca(2+)-calmodulin dependent event. EMBO J. 1991;10(8):2087-93. PMCID: 452893.
246. Kepplinger KJ, Forstner G, Kahr H, Leitner K, Pammer P, Groschner K, et al. Molecular determinant for run-down of L-type Ca2+ channels localized in the carboxyl terminus of the 1C subunit. J Physiol. 2000;529 Pt 1:119-30. PMCID: 2270188.
247. Agulhon C, Petravicz J, McMullen AB, Sweger EJ, Minton SK, Taves SR, et al. What is the role of astrocyte calcium in neurophysiology? Neuron. 2008;59(6):932-46. PMCID: 3623689.
248. Sidoryk-Wegrzynowicz M, Wegrzynowicz M, Lee E, Bowman AB, Aschner M. Role of astrocytes in brain function and disease. Toxicol Pathol. 2011;39(1):115-23.
249. Magrassi L, Leto K, Rossi F. Lifespan of neurons is uncoupled from organismal lifespan. Proc Natl Acad Sci U S A. 2013;110(11):4374-9. PMCID: 3600460.
250. Ray J, Peterson DA, Schinstine M, Gage FH. Proliferation, differentiation, and long-term culture of primary hippocampal neurons. Proc Natl Acad Sci U S A. 1993;90(8):3602-6. PMCID: 46349.
251. Rainbow L, Platt GM, Simpson GR, Sarid R, Gao SJ, Stoiber H, et al. The
227
222- to 234-kilodalton latent nuclear protein (LNA) of Kaposi's sarcoma-associated herpesvirus (human herpesvirus 8) is encoded by orf73 and is a component of the latency-associated nuclear antigen. J Virol. 1997;71(8):5915-21. PMCID: 191847.
252. Wegener AD, Jones LR. Phosphorylation-induced mobility shift in phospholamban in sodium dodecyl sulfate-polyacrylamide gels. Evidence for a protein structure consisting of multiple identical phosphorylatable subunits. J Biol Chem. 1984;259(3):1834-41.
253. Hosokawa T, Saito T, Asada A, Fukunaga K, Hisanaga S. Quantitative measurement of in vivo phosphorylation states of Cdk5 activator p35 by Phos-tag SDS-PAGE. Mol Cell Proteomics. 2010;9(6):1133-43. PMCID: 2877975.
254. Blom N, Gammeltoft S, Brunak S. Sequence and structure-based prediction of eukaryotic protein phosphorylation sites. J Mol Biol. 1999;294(5):1351-62.
255. Blom N, Sicheritz-Ponten T, Gupta R, Gammeltoft S, Brunak S. Prediction of post-translational glycosylation and phosphorylation of proteins from the amino acid sequence. Proteomics. 2004;4(6):1633-49.
256. McLachlin DT, Chait BT. Analysis of phosphorylated proteins and peptides by mass spectrometry. Curr Opin Chem Biol. 2001;5(5):591-602.
257. Kinoshita E, Kinoshita-Kikuta E, Takiyama K, Koike T. Phosphate-binding tag, a new tool to visualize phosphorylated proteins. Mol Cell Proteomics. 2006;5(4):749-57.
258. Parker CE, Mocanu V, Mocanu M, Dicheva N, Warren MR. Mass Spectrometry for Post-Translational Modifications. 2010.
259. Chauhan JS, Rao A, Raghava GP. In silico platform for prediction of N-, O- and C-glycosites in eukaryotic protein sequences. PLoS One. 2013;8(6):e67008. PMCID: 3695939.
260. Freeze HH, Varki A. Endo-glycosidase F and peptide N-glycosidase F release the great majority of total cellular N-linked oligosaccharides: use in demonstrating that sulfated N-linked oligosaccharides are frequently found in cultured cells. Biochem Biophys Res Commun. 1986;140(3):967-73.
261. Singh MB, Hough T, Theerakulpisut P, Avjioglu A, Davies S, Smith PM, et al. Isolation of cDNA encoding a newly identified major allergenic protein of rye-grass pollen: intracellular targeting to the amyloplast. Proc Natl Acad Sci U S A.
228
1991;88(4):1384-8. PMCID: 51022.
262. Mittman S, Guo J, Agnew WS. Structure and alternative splicing of the gene encoding alpha1G, a human brain T calcium channel alpha1 subunit. Neurosci Lett. 1999;274(3):143-6.
263. Mittman S, Guo J, Emerick MC, Agnew WS. Structure and alternative splicing of the gene encoding alpha1I, a human brain T calcium channel alpha1 subunit. Neurosci Lett. 1999;269(3):121-4.
264. Feder JN, Tsuchihashi Z, Irrinki A, Lee VK, Mapa FA, Morikang E, et al. The hemochromatosis founder mutation in HLA-H disrupts beta2-microglobulin interaction and cell surface expression. J Biol Chem. 1997;272(22):14025-8.
265. Volz B, Orberger G, Porwoll S, Hauri HP, Tauber R. Selective reentry of recycling cell surface glycoproteins to the biosynthetic pathway in human hepatocarcinoma HepG2 cells. J Cell Biol. 1995;130(3):537-51. PMCID: 2120536.
266. Fierro-Monti I, Racle J, Hernandez C, Waridel P, Hatzimanikatis V, Quadroni M. A novel pulse-chase SILAC strategy measures changes in protein decay and synthesis rates induced by perturbation of proteostasis with an Hsp90 inhibitor. PLoS One. 2013;8(11):e80423. PMCID: 3842330.
267. McPheeters DS, Wise JA. Measurement of in vivo RNA synthesis rates. Methods Enzymol. 2013;530:117-35.
268. Hu W, Coller J. Method for measuring mRNA decay rate in Saccharomyces cerevisiae. Methods Enzymol. 2013;530:137-55.
269. Hill WG, Southern NM, MacIver B, Potter E, Apodaca G, Smith CP, et al. Isolation and characterization of the Xenopus oocyte plasma membrane: a new method for studying activity of water and solute transporters. Am J Physiol Renal Physiol. 2005;289(1):F217-24.
270. Richardson JC, Simmons NL. Demonstration of protein asymmetries in the plasma membrane of cultured renal (MDCK) epithelial cells by lactoperoxidase-mediated iodination. FEBS Lett. 1979;105(2):201-4.
271. Peltz GA, Gallis B, Peterlin BM. Monoclonal antibody immunoprecipitation of cell membrane glycoproteins. Anal Biochem. 1987;167(2):239-44.
272. Nimmrich V, Gross G. P/Q-type calcium channel modulators. Br J Pharmacol.
229
2012;167(4):741-59.
273. Winterfield JR, Swartz KJ. A hot spot for the interaction of gating modifier toxins with voltage-dependent ion channels. J Gen Physiol. 2000;116(5):637-44. PMCID: 2229484.
274. Sousa SR, Vetter I, Lewis RJ. Venom peptides as a rich source of cav2.2 channel blockers. Toxins (Basel). 2013;5(2):286-314. PMCID: 3640536.
275. McDonough SI, Boland LM, Mintz IM, Bean BP. Interactions among toxins that inhibit N-type and P-type calcium channels. J Gen Physiol. 2002;119(4):313-28. PMCID: 2311392.
276. Ellinor PT, Zhang JF, Horne WA, Tsien RW. Structural determinants of the blockade of N-type calcium channels by a peptide neurotoxin. Nature. 1994;372(6503):272-5.
277. Tomaska L, Resnick RJ. Suppression of platelet-derived growth factor receptor tyrosine kinase activity by unsaturated fatty acids. J Biol Chem. 1993;268(7):5317-22.
278. Weekes MP, Antrobus R, Lill JR, Duncan LM, Hor S, Lehner PJ. Comparative analysis of techniques to purify plasma membrane proteins. J Biomol Tech. 2010;21(3):108-15. PMCID: 2922835.
279. Zamponi GW. Voltage-gated calcium channels. Georgetown, Tex.
New York, N.Y.: Landes Bioscience/Eurekah.com ;
Kluwer Academic/Plenum Publishers; 2005.
280. Zhang Y. Rapid amplification of cDNA ends. Methods Mol Biol. 2003;221:13-24.
281. Watschinger K, Horak SB, Schulze K, Obermair GJ, Wild C, Koschak A, et al. Functional properties and modulation of extracellular epitope-tagged Ca(V)2.1 voltage-gated calcium channels. Channels (Austin). 2008;2(6):461-73.
282. Obermair GJ, Szabo Z, Bourinet E, Flucher BE. Differential targeting of the L-type Ca2+ channel alpha 1C (CaV1.2) to synaptic and extrasynaptic compartments in hippocampal neurons. Eur J Neurosci. 2004;19(8):2109-22.
230
283. Cassidy JS, Ferron L, Kadurin I, Pratt WS, Dolphin AC. Functional exofacially tagged N-type calcium channels elucidate the interaction with auxiliary alpha2delta-1 subunits. Proc Natl Acad Sci U S A. 2014;111(24):8979-84. PMCID: 4066493.
284. Cao YQ, Tsien RW. Effects of familial hemiplegic migraine type 1 mutations on neuronal P/Q-type Ca2+ channel activity and inhibitory synaptic transmission. Proc Natl Acad Sci U S A. 2005;102(7):2590-5. PMCID: 548328.
285. Bar-Yehuda D, Korngreen A. Space-clamp problems when voltage clamping neurons expressing voltage-gated conductances. J Neurophysiol. 2008;99(3):1127-36.
286. Williams SR, Mitchell SJ. Direct measurement of somatic voltage clamp errors in central neurons. Nat Neurosci. 2008;11(7):790-8.
287. Marrion NV, Tavalin SJ. Selective activation of Ca2+-activated K+ channels by co-localized Ca2+ channels in hippocampal neurons. Nature. 1998;395(6705):900-5.
288. Kaether C, Gerdes HH. Visualization of protein transport along the secretory pathway using green fluorescent protein. FEBS Lett. 1995;369(2-3):267-71.
289. Ogawa H, Inouye S, Tsuji FI, Yasuda K, Umesono K. Localization, trafficking, and temperature-dependence of the Aequorea green fluorescent protein in cultured vertebrate cells. Proc Natl Acad Sci U S A. 1995;92(25):11899-903. PMCID: 40510.
290. Patterson GH, Knobel SM, Sharif WD, Kain SR, Piston DW. Use of the green fluorescent protein and its mutants in quantitative fluorescence microscopy. Biophys J. 1997;73(5):2782-90. PMCID: 1181180.
291. Dove LS, Abbott LC, Griffith WH. Whole-cell and single-channel analysis of P-type calcium currents in cerebellar Purkinje cells of leaner mutant mice. J Neurosci. 1998;18(19):7687-99.
292. Ottens F, Gehring NH. Physiological and pathophysiological role of nonsense-mediated mRNA decay. Pflugers Arch. 2016;468(6):1013-28.
293. Theodoulou FL, Miller AJ. Xenopus oocytes as a heterologous expression system. Methods Mol Biol. 1995;49:317-40.
294. Armstrong CM, Bezanilla F. Inactivation of the sodium channel. II. Gating
231 current experiments. J Gen Physiol. 1977;70(5):567-90. PMCID: 2228472.
295. Zagotta WN, Hoshi T, Aldrich RW. Restoration of inactivation in mutants of Shaker potassium channels by a peptide derived from ShB. Science. 1990;250(4980):568-71.
296. Hoshi T, Zagotta WN, Aldrich RW. Biophysical and molecular mechanisms of Shaker potassium channel inactivation. Science. 1990;250(4980):533-8.
297. Holmgren M, Jurman ME, Yellen G. N-type inactivation and the S4-S5 region of the Shaker K+ channel. J Gen Physiol. 1996;108(3):195-206. PMCID: 2229322.
298. Dudley SC, Jr., Chang N, Hall J, Lipkind G, Fozzard HA, French RJ. mu-conotoxin GIIIA interactions with the voltage-gated Na(+) channel predict a clockwise arrangement of the domains. J Gen Physiol. 2000;116(5):679-90. PMCID: 2229485.
299. Li RA, Ennis IL, French RJ, Dudley SC, Jr., Tomaselli GF, Marban E. Clockwise domain arrangement of the sodium channel revealed by (mu)-conotoxin (GIIIA) docking orientation. J Biol Chem. 2001;276(14):11072-7.
300. Long SB, Campbell EB, Mackinnon R. Crystal structure of a mammalian voltage-dependent Shaker family K+ channel. Science. 2005;309(5736):897-903.
301. Bixby KA, Nanao MH, Shen NV, Kreusch A, Bellamy H, Pfaffinger PJ, et al. Zn2+-binding and molecular determinants of tetramerization in voltage-gated K+ channels. Nat Struct Biol. 1999;6(1):38-43.
302. Littman DR, SpringerLink (Online service). The CD4 Molecule Roles in T Lymphocytes and in HIV Disease. Berlin, Heidelberg: Springer Berlin Heidelberg; 1996. Available from: http://dx.doi.org/10.1007/978-3-642-79798-9.
303. Parkin NT, Kitajewski J, Varmus HE. Activity of Wnt-1 as a transmembrane protein. Genes Dev. 1993;7(11):2181-93.
304. Choy E, Chiu VK, Silletti J, Feoktistov M, Morimoto T, Michaelson D, et al. Endomembrane trafficking of ras: the CAAX motif targets proteins to the ER and Golgi. Cell. 1999;98(1):69-80.
305. Takahashi K, Nakagawa M, Young SG, Yamanaka S. Differential membrane localization of ERas and Rheb, two Ras-related proteins involved in the
232 phosphatidylinositol 3-kinase/mTOR pathway. J Biol Chem. 2005;280(38):32768-74.
306. Kloc M, Li XX, Etkin LD. Two upstream cysteines and the CAAX motif but not the polybasic domain are required for membrane association of Xlcaax in Xenopus oocytes. Biochemistry. 1993;32(32):8207-12.
307. Yu Y, Ulbrich MH, Li MH, Buraei Z, Chen XZ, Ong AC, et al. Structural and molecular basis of the assembly of the TRPP2/PKD1 complex. Proc Natl Acad Sci U S A. 2009;106(28):11558-63. PMCID: 2710685.
308. O'Shea EK, Klemm JD, Kim PS, Alber T. X-ray structure of the GCN4 leucine zipper, a two-stranded, parallel coiled coil. Science. 1991;254(5031):539-44.
309. Woolfson DN. The design of coiled-coil structures and assemblies. Adv Protein Chem. 2005;70:79-112.
310. Kardinahl S, Schmidt CL, Petersen A, Schafer G. Isolation, characterization and crystallization of an iron-superoxide dismutase from the crenarchaeon Sulfolobus acidocaldarius. FEMS Microbiol Lett. 1996;138(1):65-70.
311. Forman HJ, Fridovich I. On the stability of bovine superoxide dismutase. The effects of metals. J Biol Chem. 1973;248(8):2645-9.
312. Abernethy JL, Steinman HM, Hill RL. Bovine erythrocyte superoxide dismutase. Subunit structure and sequence location of the intrasubunit disulfide bond. J Biol Chem. 1974;249(22):7339-47.
313. Malinowski DP, Fridovich I. Subunit association and side-chain reactivities of bovine erythrocyte superoxide dismutase in denaturing solvents. Biochemistry. 1979;18(23):5055-60.
314. Keele BB, Jr., McCord JM, Fridovich I. Further characterization of bovine superoxide dismutase and its isolation from bovine heart. J Biol Chem. 1971;246(9):2875-80.
315. Weber K, Osborn M. The reliability of molecular weight determinations by dodecyl sulfate-polyacrylamide gel electrophoresis. J Biol Chem. 1969;244(16):4406-12.
233