Tetraspanin-7 regulation of L-type voltage-dependent calcium channels controls pancreatic β-cell insulin secretion

Abstract

Glucose-stimulated insulin secretion (GSIS) is regulated by calcium (Ca²⁺) entry into pancreatic β-cells through voltage-dependent Ca²⁺ (Ca_V) channels. Tetraspanin (TSPAN) transmembrane proteins control Ca²⁺ handling, and thus they may also modulate GSIS. TSPAN-7 is the most abundant islet TSPAN and immunostaining of mouse and human pancreatic slices shows that TSPAN-7 is highly expressed in β- and α-cells; however, the function of islet TSPAN-7 has not been determined. Here, we show that TSPAN-7 knockdown (KD) increases glucose-stimulated Ca²⁺ influx into mouse and human β-cells. Additionally, mouse β-cell Ca²⁺ oscillation frequency was accelerated by TSPAN-7 KD. Because TSPAN-7 KD also enhanced Ca²⁺ entry when membrane potential was clamped with depolarization, the effect of TSPAN-7 on Ca_V channel activity was examined. TSPAN-7 KD enhanced L-type Ca_V currents in mouse and human β-cells. Conversely, heterologous expression of TSPAN-7 with Ca_V1.2 and Ca_V1.3 L-type Ca_V channels decreased Ca_V currents and reduced Ca²⁺ influx through both channels. This was presumably the result of a direct interaction of TSPAN-7 and L-type Ca_V channels because TSPAN-7 coimmunoprecipitated with both Ca_V1.2 and Ca_V1.3 from primary human β-cells and from a heterologous expression system. Finally, TSPAN-7 KD in human β-cells increased basal (5.6 mM glucose) and stimulated (45 mM KCl + 14 mM glucose) insulin secretion. These findings strongly suggest that TSPAN-7 modulation of β-cell L-type Ca_V channels is a key determinant of β-cell glucose-stimulated Ca²⁺ entry and thus the set-point of GSIS.

Introduction

Glucose-stimulated insulin secretion (GSIS) from pancreatic β-cells serves a critical role in maintaining glucose homeostasis and preventing hyperglycaemia (Prentki & Matschinsky, 1987; Rhodes & White, 2002; Röder et al. 2016). It is well established that calcium (Ca²⁺) influx into β-cells, primarily through L-type voltage-dependent Ca²⁺ (Ca_V) channels (e.g. Ca_V1.2 and Ca_V1.3), is required for GSIS (Curry et al. 1968; MacDonald et al. 2005; Rorsman et al. 2012; Reinbothe et al. 2013). Glucose metabolism increases the β-cell ATP:ADP ratio leading to closure of ATP-sensitive K⁺ (K_ATP) channels; the resulting membrane potential (V_m) depolarization activates β-cell Ca_V channels. Although Ca_V channel α₁-subunits contain the pore-forming domain and voltage sensing domain (VSD), interactions with auxiliary proteins greatly affect channel surface expression, voltage gating, as well as activation/inactivation kinetics (Hofmann et al. 1999; Catterall, 2011; Dolphin, 2016). Thus, β-cell Ca_V auxiliary protein composition governs Ca_V channel control of Ca²⁺ handling and insulin secretion. Furthermore, Ca_V auxiliary protein expression patterns are altered during conditions of glucolipotoxicity, which contributes to perturbations in islet Ca²⁺ oscillations and pulsatile insulin secretion (Mastrolia et al. 2017; Lee et al. 2018; Luan et al. 2019). Although it is clear that Ca_V auxiliary proteins serve an important role in tuning insulin secretion, the identity and function of these channel subunits is incompletely understood.

The tetraspanin superfamily of transmembrane proteins includes key regulators of Ca_V channel function (Andronache et al. 2007; Chen et al. 2007; Wu et al. 2015; Luan et al. 2019). The tetraspanin superfamily is categorized by the presence of four transmembrane domains (Charrin et al. 2014; Termini & Gillette, 2017). Two extracellular domains result from this structural architecture; disulphide bonding between cysteine residues in the larger extracellular domain play a prominent role in determining protein structure and function, as well as facilitating formation of tetraspanin-enriched signalling microdomains (Charrin et al. 2014; Termini & Gillette, 2017). In addition, the intracellular N- and C-termini of tetraspanin proteins can facilitate protein–protein interactions, as well as guide intracellular trafficking (Charrin et al. 2014; Termini & Gillette, 2017). One prominent tetraspanin subgroup comprises the Ca_V γ-subunits (encoded by CACNG1-8), which control Ca_V channel activation and inactivation kinetics through interactions with channel VSDs (Andronache et al. 2007; Chen et al. 2007; Sandoval et al. 2007; Wu et al. 2015; Luan et al. 2019). Of these, Ca_Vγ₄ is expressed in β-cells where it increases GSIS by enhancing L-type Ca_V channel activity (Luan et al. 2019). Another subgroup of tetraspanin proteins, explicitly designated as TSPANs (encoded by TSPAN1-33) like Ca_V γ-subunits influence Ca²⁺ handling via interactions with Ca_V channels (Kahle et al. 2011; Mallmann et al. 2013). For example, TSPAN-13 interacts with the VSD of Ca_V2.2 leading to acceleration of activation and inactivation kinetics (Mallmann et al. 2013). However, the roles of TSPANs in controlling islet Ca²⁺ handling and hormone secretion have not been investigated.

TSPAN7 is the most abundant TSPAN transcript in mouse and human islets and is highly expressed in all major islet cell types (i.e. α-, β and δ-cells) (Blodgett et al. 2015; DiGruccio et al. 2016; Segerstolpe et al. 2016). Additionally, antibody-based proteomics have identified TSPAN-7 as a surface marker of mature pancreatic endocrine cells (Lindskog et al. 2010; Hald et al. 2012). Although many TSPAN proteins are ubiquitously expressed, TSPAN-7 protein is primarily restricted to islets and a subset of neurons (Uhlén et al. 2015). Although RNA sequencing of human islets suggests that TSPAN7 transcript is not altered during diabetic conditions (i.e. palmitate treatment, low grade inflammatory cytokine exposure) (Eizirik et al. 2012; Cnop et al. 2014), after 16 weeks on a high-fat diet, TSPAN-7 protein was reduced in the islets of non-obese diabetic and New Zealand obese mice compared to C57Bl6 wild-type animals (Mitok et al. 2018). Therefore, it is possible that expression of human TSPAN-7 is regulated post-transcriptionally or post-translationally during diabetes pathogenesis. Interestingly, autoantibodies against TSPAN-7 have been identified in type 1 diabetic serum and are predictive of deteriorating β-cell function during latent autoimmune iabetes in adults (LADA), linking TSPAN-7 and diabetes (McLaughlin et al. 2016; Walther et al. 2016; Shi et al. 2019). Moreover, a protein interaction study has shown that TSPAN-7 may interact with Ca_V channels (Ca_V2.1), and thus it is possible that TSPAN-7 modulates islet Ca_V channel function (Kahle et al. 2011). Taken together, these findings indicate that TSPAN-7 may control islet Ca²⁺ handling and hormone secretion.

Here, to the best of our knowledge, we show for the first time that TSPAN-7 is an auxiliary protein of β-cell L-type Ca_V1.2 and Ca_V1.3 channels, which reduces glucose-stimulated Ca²⁺ influx, decelerates Ca²⁺ oscillation frequency, and limits basal and glucose-stimulated insulin secretion. L-type Ca_V channel interactions with TSPAN-7 decreased the Ca²⁺ conductance of both Ca_V1.2 and Ca_V1.3 channels. TSPAN-7 also reduced the rate of Ca_V1.3 channel inactivation, slowed Ca_V1.2 channel activation and sped up Ca_V1.2 channel recovery from voltage-dependent inactivation (VDI). Because TSPAN-7 interacted with both Ca_V1.2 and Ca_V1.3 channels but did not influence channel surface localization, TSPAN-7 probably directly modulates channel function. Thus, these findings suggest that TSPAN-7 control of Ca_V channel activity serves a key role in controlling β-cell Ca²⁺ handling and defining the set-point of insulin secretion during periods of secretagogue stimulation.

Methods

Ethical approval

All mice were 12–18-week old, age-matched males on a C57Bl6/J background that were bred in-house (Stock #: 000664; The Jackson Laboratory, Bar Harbor, ME, USA). Animals were housed in a Vanderbilt University Institutional Animal Care and Use Committee (IACUC) approved facility under a 12:12 h light/dark cycle with access to water and standard chow (5L0D; Lab Diets, St Louis, MO, USA) ad libitum. All experiments were carried out in accordance with guidelines established by the Vanderbilt University IACUC and conform to the principles and regulations for the use of animals outlined in Grundy et al. (2015). As approved in IACUC protocol #M1600063-01 (SOP.AWEL.01), mice were humanely euthanized by cervical dislocation followed by exsanguination by experienced personnel; to preserve islet ion channel function mice were not treated with anaesthesia (IACUC protocol #M1600063-01; SOP.AWEL.01).

All studies detailed here were approved by the Vanderbilt University Health Sciences Committee Institutional Review Board (IRB# 110164). Healthy human islets were provided from multiple isolation centers by the Integrated Islet Distribution Program (IIDP). Deidentified human donor information is provided in Table 1. The IIDP obtained informed consent for deceased donors in accordance with NIH guidelines prior to reception of human islets for our studies.

Table 1.

Summary of human islet donor information

RRID	Age (years)	Sex	BMI	Assay performed
SAMN08611211	45	F	29.8	KDE
SAMN08617638	49	M	34.0	KDE
SAMN08744571	32	M	32.3	KDE; IS
SAMN08799544	45	F	32.1	IS
SAMN08971735	50	M	32.3	IS
SAMN09001166	42	F	32.1	KDE
SAMN09768368	56	F	26.6	IS
SAMN09929594	51	F	29.7	IS
SAMN10023853	25	F	33.5	IS
SAMN10176536	49	M	28.2	IS
SAMN10252228	55	F	35.7	IS
SAMN11250012	57	F	29.2	CR
SAMN11476721	50	M	32.8	CI
SAMN11522709	51	M	26.7	CI
SAMN11578698	57	M	25.8	CR; CI
SAMN11864195	53	F	40.0	CR
SAMN12011371	33	M	33.2	CR
SAMN12227196	51	M	32.8	CR
SAMN13067487	57	M	21.3	CoIP
SAMN13108021	42	M	25.0	CoIP
SAMN13134368	62	M	31.3	CoIP
SAMN14121359	64	F	17.5	CI

BMI, body mass index; KDE, TSPAN-7 shRNA knockdown efficiency; IS, insulin secretion; CR, Ca_V current recording; CI, intracellular Ca²⁺ imaging; CoIP, TSPAN-7 coimmunoprecipatation.

Chemicals and reagents

Unless otherwise noted all chemicals and reagents were purchased from Sigma-Aldrich (St Louis, MO, USA) or Thermo-Fisher (Waltham, MA, USA). DNA plasmids expressing Ca_V1.2 α₁-HA, Ca_V1.3 α₁-HA, Ca_Vβ_2A and Ca_Vα₂δ₁ were provided by the laboratory of Roger Colbran (Wang et al. 2017 b,c). The DNA plasmid expressing TSPAN-7-pIRES-EGFP was provided by the laboratory of Maria Passafaro (Bassani et al. 2012). A stable cell line with tetracycline-inducible Kir6.2/SUR1 expression was provided by the laboratory of Jerod Denton (Raphemot et al. 2014).

Islet isolation and cell culture

Mouse pancreata were digested with collagenase P (Roche; Basel, Switzerland) and islets were isolated using density gradient centrifugation as described previously (Vierra et al. 2015; Dickerson et al. 2017; Dickerson et al. 2018). Mouse islets were dispersed into single cells by titrating in 0.005% trypsin or maintained as whole islets. Human islets were dispersed into single cells by gently titrating in Accutase™ (Innovative Cell Technologies, Inc., San Diego, CA, USA) for 2 min at 37°C or maintained as whole islets. Whole mouse islets and dispersed mouse islet cells were cultured in RPMI-1640 medium (RPMI) with 11 mM glucose supplemented with 15% fetal bovine serum (FBS), 100 IU · ml⁻¹ penicillin and 100 mg · ml⁻¹ streptomycin at 37°C, 5% CO₂. Whole human islets and dispersed human islet cells were also cultured in RPMI with 5.6 mM glucose.

Generation of TSPAN-7 small hairpin RNA (shRNA) lentiviruses

Binding domains in the coding sequence of TSPAN-7 predicted to facilitate small interfering RNA-mediated gene silencing were identified and utilized to produce third-generation lentiviral plasmids that express shRNAs targeted toward mouse TSPAN-7 (mTSPAN-7shRNA: GCTGACTTTGGGAACCTATAT), human TSPAN-7 (hTSPAN-7shRNA 1: ACTTACTCTGGGCACCTATAT, hTSPAN-7shRNA 2: GTAAATGCCACACACCTTTAA, hTSPAN-7shRNA 3: GTTAACCAGAAGGGTTGTTAT), or a scramble shRNA with no mouse or human target (CCTAAGGTTAAGTCGCCCTCG) driven by a U6 promoter as well as an mCherry fluorescent reporter driven by a hPGK promoter (VectorBuilder, Chicago, IL, USA). A GCaMP6s fluorescent Ca²⁺ indicator driven by an optimized rat insulin promoter was cloned into a third-generation lentiviral plasmid (RIP-GCaMP6s) (Chen et al. 2013; Wang et al. 2017a). HEK293FT cells were grown to 80% confluence in T175 tissue cultures flasks in Dulbecco’s modified Eagle’s medium GlutaMax-I (Thermo Fisher) supplemented with 10% FBS, 100 IU·ml⁻¹ penicillin and 100 mg·ml⁻¹ streptomycin (DMEM) at 37°C, 5% CO₂. On the day of transfection, media was replaced with 25 mL of antibiotic free DMEM supplemented with 10% heat inactivated FBS. Cells were transfected with the following DNA plasmids utilizing calcium phosphate transfection in the presence of 70 μM chloroquine at 37°C, 5% CO₂: 53.5 μg of lentivirus plasmid, 40.5 μg of psPAX2 lentivirus packaging plasmid (plasmid #: 12 260; Addgene; Watertown, MA, USA) and 16.2 μg of pMD2.G lentivirus envelope plasmid (plasmid #: 12 259; Addgene). After 6 h, transfection mixtures were replaced with 25 mL of fresh DMEM and transfected cells were cultured for 48 h at 37°C, 5% CO₂. Lentivirus-containing media was then collected, centrifuged at 300 × g for 5 min, and filtered through 0.45 μm filters to remove cell debris. Lentiviruses were purified with PEG-it™ Virus Precipitation Solution (System Biosciences, Mountain View, CA, USA) for 48 h at 4°C in accordance with the manufacturer’s instructions. Purified lentiviruses were resuspended in 250 μL of DMEM containing 25 mM HEPES buffer, aliquoted and stored at −80°C. To confirm the efficacy of lentiviral shRNA-mediated TSPAN-7 knockdown (KD), 300 mouse or human islets were dispersed into single cells, transduced for 6 h (5 μL of lentivirus stock, 0.008 mg·mL⁻¹ polybrene, 100 μL of FBS-free RPMI), and cultured for 48 h in RPMI with 11 mM glucose (5.6 mM glucose for human islet cells). Mouse and human islet cell lysates were isolated and resolved on a nitrocellulose membrane. Immunoblots were probed with rabbit anti-TSPAN-7 (dilution 1:200; catalogue no. NBP1-90310; Novus, Phoenix, AZ, USA) and TSPAN-7 protein bands visualized with anti-rabbit HRP-conjugated secondary (dilution 1:2500; catalogue no. 711-036-152; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) and SuperSignal™ Femto Substrate (Thermo Fisher) utilizing a ChemiDoc Imaging System (Bio-Rad, Hercules, CA, USA). Immunoblots were stripped, probed with mouse anti-α-tubulin (dilution 1:1000; catalogue no. A11126; Thermo Fisher) and α-tubulin protein bands visualized with anti-mouse HRP-conjugated secondary (dilution 1:2500; catalogue no. 715-036-150; Jackson ImmunoResearch Laboratories) and SuperSignal™ Femto Substrate.

Immunofluorescence imaging

Paraffin-embedded mouse and human pancreas sections were processed and probed as previously described (Vierra et al. 2015; Dickerson et al. 2017; Vierra et al. 2017). Rehydrated pancreas sections were stained with primary antibodies (dilution 1:300 guinea pig anti-insulin (Dako, Santa Clara, CA, USA) or rat anti-C-peptide (dilution 1:200; catalogue no. GN-ID4; DSHB; Iowa City, IA, USA), mouse anti-glucagon (dilution 1:200; ab10988; Abcam, Cambridge, MA, USA) and rabbit anti-TSPAN-7 (dilution 1:200; catalogue no. NBP1-90310; Novus, Phoenix, AZ, USA)) followed by secondary antibodies (dilution 1:500 anti-guinea pig or anti-rat Alexa Fluor 488, dilution 1:500 anti-mouse Alexa Fluor 647, and 1:500 anti-rabbit Alexa Fluor 546). Mouse islets were dispersed, transduced with lentiviral shRNAs for 6 h, and cultured for 72 h at 37°C, 5% CO₂. Transduced mouse islet cells were fixed on ice with 4% paraformaldehyde in PBS for 20 min then stained with rabbit anti-TSPAN-7 (dilution 1:200) followed by anti-rabbit Alexa Fluor 650 (dilution 1:500). Immunofluorescence images were collected with a LSM 780 multiphoton confocal microscope (Carl Zeiss, Oberkochen, Germany), a LSM 880 confocal microscope with AIRY scan (Carl Zeiss) (both equipped with Zeiss Zen software; 20× and 63× magnification), or a Ti2 epifluorescence microscope (Nikon, Yokyo, Japan) equipped with a Prime 95B camera (Teledyne Photometrics, Tucson, AZ, USA) with 25 mm complementary metal oxide semi-conductor sensors and Elements software (Nikon) (20× magnification; Nikon Ti2 microscope).

Intracellular Ca²⁺ imaging

Mouse and human islets or partially dispersed mouse islet clusters (100–200 cells) were transduced with lentiviral shRNAs for 6 h then cultured in RPMI with 11 mM glucose (5.6 mM for human islets) for 48 h at 37°C, 5% CO₂. Human islets were cotransduced with lentiviral RIP-GCaMP6s. Mouse islets were loaded with 10 μM Cal-630 or Cal 520 (AAT Bioquest; Sunnyvale, CA) for 1 h before the start of an experiment. Mouse islet clusters were loaded with 2 μM Fura-2 AM for 30 min before the start of an experiment. Islets and islet clusters were incubated in RPMI with 2 mM glucose for 30 min then washed and perifused with Krebs-Ringer HEPES buffer (KRHB) containing (mM) 119.0 NaCl, 4.7 KCl, 2.0 CaCl₂, 1.2 MgSO₄, 1.2 KH₂PO₄, and 10.0 HEPES (pH 7.35 adjusted by NaOH) supplemented with 2 mM glucose. β-cell glucose-stimulated Ca²⁺ influx was triggered with KRHB containing the glucose concentrations indicated in figure legends. Human islet K_ATP channels were subsequently activated with KRHB supplemented with 125 μM diazoxide and V_m depolarized using KRHB containing 45 mM KCl to investigate K_ATP independent TSPAN-7 effects on β-cell Ca²⁺ handling. Transduced mouse β-cells were identified by mCherry fluorescence; transduced human β-cells were identified by a combination of mCherry and GCaMP6s fluorescence. Mouse β-cell Cal-630 fluorescence was measured every 5 s as an indicator of intracellular Ca²⁺ using the Nikon Ti2 microscope. Human β-cell GCaMP6s fluorescence was measured every 5 s as an indicator of intracellular Ca²⁺ using a Zeiss Observer Z1 confocal microscope equipped with a Yokogawa CSU-X1 spinning disk head and PerkinElmer Volocity software.

HEK293 cells were grown to 80% confluence in 35 mm tissue culture dishes then transfected for 24 h with the following DNA plasmids utilizing lipofectamine 3000: 625 ng Ca_V1.2 α₁ (plasmid #: 26572; Addgene), 625 ng Ca_V β_2A, 625 ng Ca_V α₂δ₁, and 625 ng TSPAN-7-pIRES-EGFP or pIRES-EGFP (catalogue no. 6029-1; Takara Bio USA; Mountain View, CA). As Ca_V1.3 channels undergo VDI at the resting V_m of HEK293 cells (~ −50 mV), a stable cell line with tetracycline inducible TALK1 K2P leak channels (T3H16) was utilized to tonically hyperpolarize V_m and thus enhance Ca_V1.3 activity (Dickerson et al. 2017). T3H16 cells were transfected for 24 h with the following DNA plasmids utilizing lipofectamine 3000:625 ng Ca_V1.3 α₁ (plasmid #: 49 333; Addgene), 625 ng Ca_V β_2A, 625 ng Ca_V α₂δ₁, and 625 ng TSPAN-7-pIRES-EGFP or pIRES-EGFP. Transfected cells were dispersed into single cells and cultured overnight in DMEM (T3H16 culture media was supplemented with 1 μg·ml⁻¹ tetracycline to induce TALK-1 expression) at 37°C, 5% CO₂. Cells were loaded with 10 μM Cal-590 (AAT Bioquest) for 1 h then washed and perifused with buffer containing (mM) 150.0 NaCl, 5.0 KCl, 2.5 CaCl₂, 2.0 MgCl₂, 10.0 HEPES, and 10.0 glucose (pH 7.35 adjusted by NaOH) until Cal-590 fluorescence signal stabilized. The cells were then perifused for at least 10 min with buffer containing (mM) 105.0 NaCl, 50.0 KCl, 2.5 CaCl₂, 2.0 MgCl₂, 10.0 HEPES, and 10.0 glucose (pH 7.35 adjusted by NaOH) to stimulate Ca²⁺ influx through Ca_V channels. Transfected cells were identified by green fluorescent protein (GFP) fluorescence and Cal-590 fluorescence was measured every 5 s as an indicator of intracellular Ca²⁺ with the Nikon Ti2 microscope.

Whole-cell patch clamp electrophysiology

Mouse and human islets were dispersed into single cells, transduced with lentiviral shRNAs for 6 h (human islet cells were cotransduced with lentiviral RIP-GCaMP6s to identify β-cells) and cultured for 48 h. Patch electrodes (3–4 MΩ) utilized for Ca_V current recordings were backfilled with intracellular solution containing (mM) 132.0 CsCl, 10.0 tetraethylammonium chloride (TEA-Cl), 10.0 EGTA, 3.0 Na-ATP and 5.0 HEPES (pH 7.25 adjusted by CsOH). Mouse and human β-cells were patched in KRHB; after a whole-cell configuration was established (seal resistance >1 GΩ) the bath was exchanged for buffer containing (mM) 82.0 NaCl, 5.0 CsCl, 30.0 CaCl₂, 1.0 MgCl₂, 0.1 EGTA, 20.0 TEA-Cl, 0.1 tolbutamide and 17.0 glucose (pH 7.35 with NaOH). Mouse and human β-cells were perifused with this solution for 3 min prior to measuring Ca_V currents. A whole-cell patch clamp technique was utilized to record Ca_V currents in single β-cells using an Axopatch 200B amplifier with pCLAMP10 software (Molecular Devices, Sunnyvale, CA, USA). Sequential 10 mV steps ranging from −60 to 70 mV (500 ms) were applied from a holding potential of −80 mV to generate Ca_V currents; linear leak currents were subtracted using a P/4 protocol. The L-type Ca_V channel inhibitor isradipine (10 μM) was perifused into the bath solution and isradipine-insensitive currents were subtracted from total Ca_V currents to determine mouse and human β-cell L-type Ca_V currents.

HEK293 cells were grown to 80% confluence in 35 mm tissue culture dishes then transfected with the DNA plasmids utilizing lipofectamine 3000 and cultured for 24 h at 37°C, 5% CO₂: 625 ng Ca_V1.2 α₁ or Ca_V 1.3 α₁, 625 ng Ca_V β_2A, 625 ng Ca_V α₂δ₁, and 625 ng TSPAN-7-pIRES-EGFP or pIRES-EGFP. Alternatively, a cell line with stable expression of SUR1 and tetracycline-inducible Kir6.2 expression (functional K_ATP channels) was transfected with 2500 ng of TSPAN-7-pIRES-EGFP or pIRES-EGFP (Raphemot et al. 2014). Cells were dispersed and cultured for 24 h in DMEM at 37°C, 5% CO₂ prior to an experiment. Culture media was supplemented with 1 μg·ml⁻¹ tetracycline to induce Kir6.2 expression for K_ATP current recordings. Cells utilized for Ca_V current recordings were patched in extracellular solution containing (mM) 112.0 NaCl, 5.0 CsCl, 10.0 CaCl₂, 1.0 MgCl₂, 10.0 HEPES, 5.0 glucose and 20.0 TEA-Cl (pH 7.35 adjusted by NaOH) and current densities through heterologously expressed Ca_V channels were measured as a function of V_m. Maximum Ca_V current traces (Ca_V1.2_max acquired at 10 mV; Ca_V1.3_max acquired at −20 mV) were divided into an activation phase (e.g. period of increasing current) and an inactivation phase (e.g. period of decreasing current). Ca_V current activation curves were fit to a model of one-phase exponential association (Eqn (1)) and Ca_V current inactivation curves were fit to a model of one-phase exponential decay (Eqn (2)) using Prism, version 8 (GraphPad Software Inc., San Diego, CA, USA), then the associated kinetic parameters determined. I_act (pA/pF) is current density as a function of time during Ca_V channel activation, I₀ (pA/pF) indicates Ca_V current density at time equal to zero, plateau (pA/pF) refers to a steady state Ca_V current density as time approaches infinity, K_act (s⁻¹) is the Ca_V channel activation rate constant, I_inact (pA/pF) is current density as a function of time during Ca_V channel inactivation, K_inact (s⁻¹) is the Ca_V channel inactivation rate constant and t (s) is time.

$$I_{\text{act}}(t)=I_0+(plateau-I_0)\ast \left(1-e^{-K_{\text{act}}t}\right)$$ (1)

$$I_{\text{inact}}(t)=(I_0-plateau)\times e^{-K_{\text{inact}}t}+plateau$$ (2)

Patch electrodes employed for K_ATP current recordings were backfilled with intracellular solution containing (mM) 140.0 KCl, 0.5 MgCl₂, 10.0 EGTA, 0.1 Mg-ATP and 5.0 HEPES (pH 7.25 adjusted with KOH). Cells utilized for K_ATP current recordings were patched in KRHB. Immediately after a whole-cell configuration was established currents were recorded every 15 s as a function of V_m utilizing a voltage ramp from −120 mV to 60 mV (Vierra et al. 2015; Dickerson et al. 2017; Dickerson et al. 2018). This voltage clamp protocol was continued until intracellular ATP equilibrated with ATP in the recording pipette (0.1 mM Mg-ATP) and thus maximum whole-cell currents were reached. K_ATP currents activated by a reduction in intracellular ATP were calculated by subtracting initial current traces from maximum current traces. Finally, VDI of heterologously expressed Ca_V channels was evaluated by applying sequential 10 mV prepulse steps ranging from −80 to 10 mV (5 s), then measuring the magnitude of Ca_V currents (Ca_V1.2_max at 10 mV; Ca_V1.3_max at −20 mV).

Coimmunoprecipitation of TSPAN-7 with L-type Ca_V channels

Cell lysates were isolated from healthy human islets, combined with 5 μg·ml⁻¹ rabbit anti-Ca_V1.2 (catalogue no. ACC-003; knockout (KO) validated (Jeon et al. 2010); Alomone Labs, Jerusalem, Israel) or rabbit anti-Ca_V1.3 (catalogue no. ACC-005; KO validated (Shi et al. 2017); Alomone Labs) and mixed with Pierce protein A/G magnetic beads (catalogue no. 88803; 30 μL per sample) overnight at 4°C with agitation as described previously (Dickerson et al. 2017). Purified anti-Ca_V1.2 (or anti-Ca_V1.3)/islet protein immunocomplexes were isolated and resolved on nitrocellulose membranes. Immunoblots were probed with mouse anti-TSPAN-7 (dilution 1:200; catalogue no. NBP2-52894; Novus Biologicals, Centennial, CO, USA) and TSPAN-7 protein bands visualized with anti-mouse HRP-conjugated secondary (dilution 1:2500) and SuperSignal™ Femto Substrate.

HEK293 cells were grown to 80% confluence in 100 mm tissue culture dishes, transfected with DNA plasmids utilizing calcium phosphate transfection (5 μg each of Ca_V1.2 α₁-HA or Ca_V1.3 α₁-HA, Ca_V β_2A, Ca_V α₂ä₁, and TSPAN-7-pIRES-EGFP or pIRES-EGFP, and cultured for 48 h at 37°C, 5% CO₂. HEK293 cell lysates were isolated, equal amounts of each lysate were combined with 5 μg·ml⁻¹ mouse anti-HA (catalogue no. VAPR12CA5; Vanderbilt Antibody and Protein Resource, Nashville, TN, USA) or mouse IgG isotype control (catalogue no. 31903; Invitrogen; Carlsbad, CA) and incubated for 1 h at 4°C with agitation. Anti-HA/protein immunocomplexes were immunoprecipitated at 4°C overnight with protein A/G magnetic beads (30 μL per sample) and resolved on a nitrocellulose membrane. Immunoprotein complex western blots were probed with anti-TSPAN-7 (dilution 1:200) and protein bands were visualized with donkey anti-rabbit HRP-conjugated secondary (dilution 1:2500) and SuperSignal™ Femto Substrate. Alternatively, surface proteins were biotinylated and purified utilizing neutravidin agarose resin in accordance with the manufacturer’s instructions (catalogue no. A44390; Cell Surface Biotinylation and Isolation Kit; Pierce, Rockford, IL, USA); equal amounts of each lysate were loaded and resolved on a nitrocellulose membrane. Surface protein western blots were probed with mouse anti-HA (dilution 1:500) and Ca_V1.2-HA or Ca_V1.3-HA protein bands visualized with anti-mouse HRP-conjugated secondary (dilution 1:2500) and SuperSignal™ Femto Substrate.

Insulin secretion assays

Upon reception human islets were dispersed into single cells and cultured overnight in 24-well plates at 37°C, 5% CO₂. The islet cells were transduced with lentiviral shRNAs for 6 h the nest day then cultured for at least 72 h in RPMI supplemented with 5.6 mM glucose prior to an experiment. On the day of an experiment, the culture media was replaced with 500 μL of DMEM containing 0.5 mg·mL⁻¹ BSA, 0.5 mM CaCl₂ and 10.0 mM HEPES (DMEM*) supplemented with 10% FBS and 5.6 mM glucose and incubated for 1 h at 37°C, 5% CO₂. On ice the culture media was replaced with 500 μL of FBS-free DMEM* supplemented with the glucose concentrations and/or treatments as indicated where appropriate. Insulin secretion was measured over a 1 h period at 37°C in 5% CO₂, after which the supernatants were collected, supplemented with mammalian protease inhibitor (catalogue no. 4693132001; Roche, Basel, Switzerland) and stored at −20°C until analysed. Human islet cells were gently washed twice with cold 1× phosphate-buffered saline and collected in 100 μL of acid-ethanol supplemented with mammalian protease inhibitor for analysis of total islet insulin content. The insulin concentrations of islet secretion supernatants and insulin extracted from islet cells were analysed by the Vanderbilt Hormone Assay and Analytical Services Core (Nashville, TN, USA). Secreted insulin was normalized to total insulin content.

Statistical analysis

Islet TSPAN-7 localization and KD, Ca_V1.2/1.3 surface expression and intracellular Ca²⁺ imaging data were quantified using Zen software and the ImageJ Fiji image processing pack (NIH, Bethesda, MD, USA); Ca_V currents and K_ATP currents were quantified with Axon Clampfit software (Molecular Devices). Statistical analysis was carried out using Excel (Microsoft Corp., Redmond, WA, USA) and Prism, version 8 (GraphPad Software Inc.) as indicated where appropriate. Data were normalized when appropriate. All data are presented as the mean ± 95% confidence interval (CI) for the specified number of samples (n). P < 0.05 was considered statistically significant.

Results

TSPAN-7 is expressed in both pancreatic β- and α-cells

To investigate islet TSPAN-7 protein expression, mouse and human pancreas sections were stained for insulin (or C-peptide), glucagon (GCG) and TSPAN-7. TSPAN-7 was found to be expressed in β- and α-cells, although not in surrounding acinar tissue (Fig. 1A). TSPAN-7 displayed greater heterogeneity in staining intensity in α-cells compared to in β-cells. Interestingly, although TSPAN-7 is a transmembrane protein (Lindskog et al. 2010; Hald et al. 2012), significant cytoplasmic staining with a punctate distribution was also observed, which may indicate that TSPAN-7 serves additional intracellular roles (Fig. 1B). Punctate distribution of TSPAN-7 at the plasma membrane and in the cytosol was confirmed in dispersed mouse islet cells (Fig. 2A). TSPAN-7 staining was observed in islet cells that were negative for both insulin and GCG staining, which indicates that TSPAN-7 is also expressed in other islet cell types such as δ-cells and/or pancreatic polypeptide cells.

Figure 1. TSPAN-7 is highly expressed in mouse and human pancreatic islets.

A, representative immunofluorescence images of mouse (upper) and human (lower) pancreas sections (TSPAN-7, green; insulin, red; glucagon, magenta; nuclei, blue; scale bars = 50 μm). B, representative immunofluorescence images of a human pancreas section (63× magnification; TSPAN-7, green; C-peptide, red; nuclei, blue; scale bars = 10 μm).

Figure 2. Knockdown of islet TSPAN-7 utilizing lentiviral shRNA delivery.

A, representative mouse islet cell immunofluorescence imaging showing TSPAN-7 KD relative to scramble shRNA controls (mCherry, red; TSPAN-7, green). B, average TSPAN-7 immunofluorescence of mouse islet cells expressing mTSPAN-7 shRNA (dark grey; n = 28 islet cells) relative to scramble shRNA controls (white; n = 37 islet cells); P = 5.0 × 10⁻¹¹. C, representative western blot showing TSPAN-7 KD in dispersed mouse islet cells (upper left) relative to α-tubulin (lower left) and representative western blot of TSPAN-7 KD in dispersed human islet cells (upper right) relative to α-tubulin (lower right). D, average TSPAN-7 protein in dispersed mouse islet cells expressing mTSPAN-7 shRNA (dark grey; n = 4 mice) relative to scramble shRNA controls (white; n = 4 mice); values normalized to α-tubulin protein; P = 0.0005. E, average TSPAN-7 protein in dispersed human islet cells expressing hTSPAN-7 shRNA (light grey; n = 4 islet donors) relative to scramble shRNA controls (white; n = 4 islet donors); hTSPAN-7 shRNA refers to hTSPAN-7 shRNA 3; values normalized to α-tubulin protein; P = 0.0003. Statistical analysis was conducted using an unpaired two-sample t test (B) or one-sample t tests (D and E); uncertainty is expressed as the 95% CI.

TSPAN-7 KD in mouse and human β-cells enhances glucose-stimulated Ca²⁺ influx

To investigate the function of TSPAN-7 in primary mouse and human β-cells, lentiviral shRNA constructs with an mCherry fluorescent reporter were used to KD TSPAN-7 protein expression. Quantification of TSPAN-7 immunofluorescence in TSPAN-7 KD mouse islet cells showed a significant decrease in fluorescence intensity (53.1 ± 11.1% reduction; P = 5.0 × 10⁻¹¹) (Fig.2A and B) compared to scramble controls. TSPAN-7 KD efficiency was confirmed in dispersed islet cells utilizing western blot analysis. Mouse islet cells expressing TSPAN-7 shRNA showed significant loss of TSPAN-7 protein compared to scramble shRNA controls (63.2 ± 7.6% reduction; P = 0.0005) (Fig. 2C and D). All three tested lentiviral hTSPAN-7 shRNA constructs knocked down TSPAN-7 in dispersed human islets; because hTSPAN-7 shRNA 3 was the most effective at knocking down TSPAN-7 expression (64.2 ± 6.5% reduction; P = 0.0003) (Fig. 2C and E), it was utilized for all subsequent experiments (hTSPAN-7 shRNA). Because α-cell lentiviral transduction efficiency is very low in our hands, the overall TSPAN-7 KD efficiency was probably greater in transduced β-cells compared to total islet TSPAN-7 KD. Interestingly, two TSPAN-7 bands were observed for human islet immunoblots. Because TSPAN-7 is glycosylated and palmitoylated, it is possible that multiple bands represent differential post-translational modification of TSPAN-7 (Skaar et al. 2015).

TSPANs have been shown to regulate Ca²⁺ handling; thus, the effect of TSPAN-7 KD on β-cell glucose-stimulated Ca²⁺ influx was evaluated. When mouse islets were stimulated with 14 mM glucose, TSPAN-7 KD enhanced Ca²⁺ influx into β-cells compared to scramble controls (38.3 ± 3.7% increase in Cal-630 area under the curve (AUC); P = 0.002) (Fig. 3A and B). TSPAN-7 did not affect β-cell Ca²⁺ at low (2 mM) glucose (P = 0.8); however, under basal (7 mM) glucose conditions, TSPAN-7 KD β-cell Ca²⁺ was modestly elevated compared to scramble shRNA controls (7.2 ± 2.7% increase in Fura-2 AM AUC; P = 0.04) (Fig. 2C and D). As pulsatile insulin secretion is controlled by oscillations in β-cell intracellular Ca²⁺, we next assessed whether TSPAN-7 affects glucose-mediated islet Ca²⁺ oscillation frequency. TSPAN-7 KD mouse islets stimulated with 10 mM glucose exhibited more rapid Ca²⁺ oscillations compared to control islets expressing scramble shRNA (29.4 ± 19.8% increased Ca²⁺ oscillation frequency; P = 0.002) (Fig. 3E-G). The impact of TSPAN-7 on glucose-stimulated Ca²⁺ influx into human islets was also measured by quantifying AUC of GCaMP6s fluorescence. During stimulation with 10 mM glucose, TSPAN-7 KD increased Ca²⁺ influx into human β-cells compared to scramble shRNA controls (82.3 ± 11.2% increase in GCaMP6s AUC; P = 0.001) (Fig. 3H and I). To determine whether TSPAN-7 control of β-cell Ca²⁺ is a result of effects on Ca_V channel function, V_m was clamped in a depolarized state (diazoxide and KCl). Under these conditions, we observed significantly more Ca²⁺ influx into TSPAN-7 KD β-cells compared to scramble shRNA controls (37.8 ± 10.8% increase in GCaMP6s; P = 0.006) (Fig. 3H and J) suggesting a role for TSPAN-7 in controlling Ca_V channel activity. TSPAN-7 did not alter Ca²⁺ in human β-cells at 2 mM glucose and the time required to reach minimum Ca²⁺ following diazoxide-induced K_ATP channel activation was unaffected by TSPAN-7. Interestingly, TSPAN-7 KD had a larger impact on glucose-stimulated Ca²⁺ influx than on depolarization-stimulated Ca²⁺ entry.

Figure 3. TSPAN-7 controls islet glucose-stimulated Ca²⁺ entry and Ca²⁺ oscillation frequency.

A, representative Cal-630 recordings as an indicator of intracellular Ca²⁺ in mouse islets expressing scramble shRNA (black) or mTSPAN-7 shRNA (dark grey); treatments are as indicated above Ca²⁺ traces. B, average AUC of glucose-stimulated Ca²⁺ influx into mouse islets expressing scramble shRNA (white; n = 3 mice) or mTSPAN-7 shRNA (dark grey; n = 3 mice); P = 0.002. C, average ratio-metric Fura-2 AM recordings as an indicator of intracellular Ca²⁺ in mouse islets expressing scramble shRNA (black; n = 5 mice) or mTSPAN-7 shRNA (dark grey; n = 5 mice). D, average AUC of islet Ca²⁺ expressing scramble shRNA (white; n = 5 mice) or mTSPAN-7 shRNA (dark grey; n = 5 mice) at 2 mm and 7 mm glucose; P = (2 mm: 0.8, 7 mm: 0.04). E, representative Cal-520 recordings as an indicator of glucose-stimulated (10 mM glucose) Ca²⁺ oscillations in mouse islets expressing scramble shRNA. F, representative Cal-520 recordings as an indicator of glucose-stimulated Ca²⁺ oscillations in mouse islets expressing mTSPAN-7 shRNA. G, average Ca²⁺ oscillation frequency of mouse islets expressing scramble shRNA (white; n = 60 islets from 3 mice) or mTSPAN-7 shRNA (dark grey; n = 53 islets from 3 mice); P = 0.002. H, representative GCaMP6s recordings as an indicator of intracellular Ca²⁺ in human β-cells expressing scramble shRNA (black; upper) or hTSPAN-7 shRNA (light grey; lower). I, average normalized glucose-stimulated Ca²⁺ influx into human β-cells within intact islets expressing hTSPAN-7 shRNA (light grey; n = 4 human islet donors) relative to scramble shRNA controls (white; n = 4 human islet donors); P = 0.001. J, average normalized depolarization-stimulated Ca²⁺ influx into human β-cells within intact islets expressing hTSPAN-7 shRNA (light grey; n = 4 human islet donors) relative to scramble shRNA controls (white; n = 4 human islet donors); P = 0.006. Statistical analysis was conducted using a one-sample t test (B, I and J) or an unpaired two-sample t test (D and G); uncertainty is expressed as 95% CI.

TSPAN-7 decreases β-cell L-type Ca²⁺ currents

Because TSPAN-7 KD increased Ca²⁺ influx into human β-cells when V_m was clamped in a depolarized state, control of β-cell L-type Ca_V channel activity by TSPAN-7 was measured. Ca_V currents were generated by applying sequential depolarizing steps (Fig. 4A) to transduced mouse and human β-cells; maximum Ca_V current densities were plotted as a function of V_m. The L-type Ca_V channel inhibitor isradipine was utilized to isolate the L-type component of β-cell Ca_V currents (Fig. 4B). TSPAN-7 KD significantly increased L-type Ca_V currents in mouse β-cells across a wide range of V_m (Fig. 4C). TSPAN-7 KD also augmented L-type Ca_V currents in human β-cells (Fig. 4D). TSPAN-7 KD did not affect non-L-type Ca_V currents in mouse or human β-cells (Fig. 4E and F). These results suggest that TSPAN-7 limits L-type Ca_V channel activity, which diminishes glucose-stimulated Ca²⁺ entry into β-cells.

Figure 4. TSPAN-7 regulates β-cell isradipine sensitive voltage-dependent Ca²⁺ currents.

A, summary of whole-cell voltage step protocol utilized to generate Ca_V currents (upper) and representative β-cell Ca_V current recording (lower). B, representative human β-cell Ca_V currents before and after the addition of isradipine. L-type Ca_V currents were calculated by subtracting isradipine insensitive component from total current density. C, average isradipine sensitive (L-type) Ca_V current densities as a function of V_m recorded from dispersed mouse β-cells expressing scramble shRNA (black; n = 12 cells from 4 mice) or mTSPAN-7 shRNA (dark grey; n = 14 cells from 4 mice); P = (−60 mV: 0.8, −50 mV: 0.7, −40 mV: 0.8, −30 mV: 0.9, −20 mV: 0.6, −10 mV: 0.4, 0 mV: 0.04, 10 mV: 0.005, 20 mV: 0.007, 30 mV: 0.002, 40 mV: 0.005, 50 mV: 0.004, 60 mV: 0.008, 70 mV: 0.3). D, average isradipine sensitive Ca_V current densities as a function of V_m recorded from dispersed human β-cells (identified by GCaMP6s fluorescence) expressing scramble shRNA (black; n = 23 cells from four human islet donors) or hTSPAN-7 shRNA (light grey; n = 22 cells from four human islet donors); P = (−60 mV: 0.6, −50 mV: 0.5, −40 mV: 1.0, −30 mV: 0.6, −20 mV: 0.4, −10 mV: 0.1, 0 mV: 0.08, 10 mV: 0.02, 20 mV: 0.01, 30 mV: 0.008, 40 mV: 0.003, 50 mV: 0.001, 60 mV: 0.02, 70 mV: 0.2). E, average isradipine insensitive (non-L-type) Ca_V current densities as a function of V_m recorded from dispersed mouse β-cells expressing scramble shRNA (black; n = 12 cells from four mice) or mTSPAN-7 shRNA (dark grey; n = 15 cells from four mice); P = (−60 mV: 0.5, −50 mV: 0.3, −40 mV: 0.4, −30 mV: 0.6, −20 mV: 0.8, −10 mV: 0.5, 0 mV: 0.9, 10 mV: 0.8, 20 mV: 0.2, 30 mV: 0.3, 40 mV: 0.2, 50 mV: 0.1, 60 mV: 0.5, 70 mV: 0.2). F, average non-L-type Ca_V current densities as a function of V_m recorded from dispersed human β-cells (identified by GCaMP6s fluorescence) expressing scramble shRNA (black; n = 23 cells from four human islet donors) or hTSPAN-7 shRNA (light grey; n = 22 cells from four human islet donors); P = (−60 mV: 0.2, −50 mV: 0.2, −40 mV: 0.3, −30 mV: 0.7, −20 mV: 0.9, −10 mV: 1.1, 0 mV: 1.4, 10 mV: 1.3, 20 mV: 1.0, 30 mV: 0.8, 40 mV: 0.5, 50 mV: 0.4, 60 mV: 0.3, 70 mV: 0.1). Statistical analysis was conducted using unpaired two-sample t tests; uncertainty is expressed as 95% CI.

TSPAN-7 interacts with and limits Ca²⁺ entry through L-type Ca_V channels in human β-cells

Although TSPAN-7 regulates β-cell Ca²⁺ handling in part by controlling L-type Ca_V channel activity, it is unclear whether TSPAN-7 modulates Ca_V channel function directly or indirectly. To determine whether TSPAN-7 interacts with β-cell Ca_V channels, Ca_V1.2 and Ca_V1.3 immunoprotein complexes were pulled down from human islet cell lysates (n = 3 human islet donors) and coimmunoprecipitated TSPAN-7 was analysed by western blot analysis. Bands corresponding to TSPAN-7 were observed for all Ca_V1.2 and Ca_V1.3 immunoprotein complex preparations, indicating a direct interaction between TSPAN-7 and L-type Ca_V channels in human islets (Fig. 5A). TSPAN-7 also coimmunoprecipitated with heterologously expressed Ca_V1.2-HA and Ca_V1.3-HA immunoprotein complexes (Fig. 5B and C). TSPAN-7 only pulled down with immunoprotein complexes containing anti-HA and not with immunocomplexes containing an IgG isotype control, which shows that the observed TSPAN-7 bands are not a result of nonspecific protein binding.

Figure 5. TSPAN-7 regulation of Ca²⁺ handling is mediated via interactions with Ca_V1.2 and Ca_V1.3.

A, representative western blot of Ca_V1.2 (left; n = 3 human islet donors) or Ca_V1.3 (right; n = 3 human islet donors) immunocomplexes isolated from human islet preparations and probed with anti-TSPAN-7. B, representative western blot of heterologously expressed Ca_V1.2-HA immunocomplexes purified with anti-HA (left) or an IgG isotype control (right) from HEK293 cells; heterologously expressed, coimmunoprecipitated TSPAN-7 detected using anti-TSPAN-7. C, representative western blot of heterologously expressed Ca_V1.3-HA immunocomplexes purified with anti-HA (left) or an IgG isotype control (right) from HEK293 cells; heterologously expressed, coimmunoprecipitated TSPAN-7 detected using anti-TSPAN-7. D, average normalized plasma membrane Ca_V1.2 heterologously expressed with TSPAN-7 + GFP (light grey; n = 3 biological replicates) relative to GFP alone (white; n = 3 biological replicates); P = 0.9. E, average normalized plasma membrane Ca_V1.3 heterologously expressed with TSPAN-7 + GFP (light grey; n = 3 biological replicates) relative to GFP alone (white; n = 3 biological replicates); P = 1.0. F, average Cal-590 traces normalized to minimum fluorescence intensity (F_min) as an indicator of depolarization-stimulated Ca²⁺ entry into HEK293 cells coexpressing Ca_V1.2 and GFP (left; black; n = 3 biological replicates) or TSPAN-7 + GFP (light grey; n = 3 biological replicates); quantification of AUC of depolarization-stimulated Ca²⁺ influx into GFP-positive HEK293 cells (right); P = 0.01. G, average Cal-590 traces normalized to F_min as an indicator of depolarization-stimulated Ca²⁺ entry into induced T3H16 cells coexpressing Ca_V1.3 and GFP (left; black; n = 5 biological replicates) or TSPAN-7 + GFP (light grey; n = 5 biological replicates); quantification of AUC of depolarization-stimulated Ca²⁺ influx into GFP-positive T3H16 cells (right); P = 0.008. Statistical analysis was conducted using unpaired two-sample t tests; uncertainty is expressed as 95% CI.

TSPAN proteins serve diverse biological functions, including the control of protein trafficking and regulation of ion channel activity; thus, the mechanism underlying TSPAN-7 modulation of β-cell Ca_V channel function was examined. TSPAN-7 was heterologously expressed in HEK293 cells with Ca_V1.2 or Ca_V1.3 L-type Ca_V channels, plasma membrane proteins were biotinylated and captured on neutravidin-conjugated beads, and the impact of TSPAN-7 on Ca_V channel surface expression analysed by western blot analysis. When normalized to total surface protein, TSPAN-7 expression had no effect on the number of Ca_V1.2 or Ca_V1.3 channels at the plasma membrane (P = 0.9 and P = 1.0, respectively) (Fig. 5D and E). Furthermore, overexpression of TSPAN-7 significantly decreased the AUC of depolarization-stimulated Ca²⁺ entry through both Ca_V1.2 (51.0 ± 13.9% decrease in Cal-590 fluorescence; P = 0.01) (Fig. 5F) and Ca_V1.3 (28.4 ± 14.7% decrease in Cal-590 fluorescence; P = 0.008) (Fig. 5G) channels. Taken together, these data show that TSPAN-7 interacts with and directly modulates Ca²⁺ entry through L-type Ca_V channels.

TSPAN-7 differentially regulates L-type Ca_V1.2 and Ca_V1.3 channels

Because TSPAN-7 reduced Ca²⁺ influx through Ca_V1.2 and Ca_V1.3 channels, the impact of TSPAN-7 expression on Ca_V channel activation/inactivation kinetics was examined. Heterologous TSPAN-7 expression significantly decreased Ca²⁺ conductance through both Ca_V1.2 and Ca_V1.3 channels over a range of V_m (Fig. 6A and B). Maximum Ca_V1.2 (measured at 10 mV) and Ca_V1.3 (measured at −20 mV) channel current traces were also fit to a model of one-phase exponential association (or decay) to investigate the influence of TSPAN-7 on Ca_V channel activation and inactivation kinetics, respectively. TSPAN-7 expression decreased peak Ca_V1.2 currents (32.8 ± 21.2% decrease; P = 0.03) (Fig. 6C) and slowed channel activation kinetics (K_act: 52.7 ± 10.6% decrease; P = 6.0 × 10⁻⁶) (Fig 6C, inset). TSPAN-7 expression also decreased peak Ca_V1.3 currents (65.4 ± 15.3% decrease; P = 0.001) (Fig. 6D) and slowed channel inactivation kinetics (K_inact: 32.8 ± 9.4% decrease; P = 8.0 × 10⁻⁶) (Fig 6D, inset). Because TSPANs are structurally similar to Ca_V γ-subunits, which modulate Ca_V channel VDI kinetics (Klugbauer et al. 2000), we explored the effect of TSPAN-7 on L-type Ca_V channel VDI. Five second depolarizing steps were applied sequentially from −80 to 10 mV and L-type Ca_V current densities were measured after each voltage step (Fig. 6E); TSPAN-7 expression significantly accelerated recovery of Ca_V1.2 channels from VDI following depolarization to −20 mV or greater (Fig. 6F), although it did not affect VDI of Ca_V1.3 channels (Fig. 6G).

Figure 6. TSPAN-7 controls Ca_V1.2 and Ca_V1.3 activity in a channel specific manner.

A, average Ca_V current densities as a function of V_m recorded from HEK293 cells coexpressing Ca_V1.2 and GFP (black; n = 19 cells) or TSPAN-7 + GFP (light grey; n = 13 cells); P = (−60 mV: 1.0, −50 mV: 0.8, −40 mV: 0.1, −30 mV: 0.07, −20 mV: 0.06, −10 mV: 0.04, 0 mV: 0.03, 10 mV: 0.03, 20 mV: 0.05, 30 mV: 0.07, 40 mV: 0.1, 50 mV: 0.3, 60 mV: 0.2, 70 mV: 0.4). B, average Ca_V current densities as a function of V_m recorded from HEK293 cells coexpressing Ca_V1.3 and GFP (black; n = 17 cells) or TSPAN-7 + GFP (light grey; n = 18 cells); P = (−60 mV: 0.2, −50 mV: 0.4, −40 mV: 0.03, −30 mV: 0.006, −20 mV: 0.001, −10 mV: 0.002, 0 mV: 0.03, 10 mV: 0.2, 20 mV: 0.5, 30 mV: 0.7, 40 mV: 1.0, 50 mV: 0.6, 60 mV: 0.2, 70 mV: 0.4). C, average maximum (measured at 10 mV) Ca_V1.2 current traces recorded from HEK293 cells coexpressing Ca_V1.2 and GFP (black; n = 19 cells) or TSPAN-7 + GFP (light grey; n = 13 cells). Inset table shows activation (K_act) and inactivation (K_inact) rate constants for Ca_V1.2; P = (K_act: 6.0 × 10⁻⁶, K_inact: 0.6). D, average maximum (measured at −20 mV) Ca_V1.3 current traces recorded from HEK293 cells coexpressing Ca_V1.3 and GFP (black; n = 17 cells) or TSPAN-7 + GFP (light grey; n = 18 cells). Inset table shows K_act and K_inact for Ca_V1.3; P = (K_act: 0.7, K_inact: 8.0 × 10⁻⁶). E, summary of whole-cell voltage step protocol utilized to measure rate of Ca_V1.2 and Ca_V1.3 recovery from VDI (upper) and a representative Ca_V current recording (lower); inset shows a representative Ca_V current generated after a 5 s depolarizing step. F, average normalized Ca_V currents recorded from HEK293 cells coexpressing Ca_V1.2 and GFP (black; n = 19 cells) or TSPAN-7 + GFP (light grey; n = 13 cells) relative to maximum Ca_V1.2 currents measured after a 5 s depolarizing step; P = (−50 mV: 0.5, −40 mV: 0.4, −30 mV: 0.2, −20 mV: 0.03, −10 mV: 0.01, 0 mV: 0.03, 10 mV: 0.007). G, average normalized Ca_V currents recorded from HEK293 cells coexpressing Ca_V1.3 and GFP (black; n = 16 cells) or TSPAN-7 + GFP (light grey; n = 13 cells) relative to maximum Ca_V1.3 currents measured after a 5 s depolarizing step; P = (−70 mV: 0.1, −60 mV: 0.2, −50 mV: 0.6, −40 mV: 0.9, −30 mV: 0.8, −20 mV: 0.4, −10 mV: 0.1, 0 mV: 0.08, 10 mV: 0.06). H, average K_ATP current densities as a function of V_m recorded from induced KATP cells expressing GFP (black; n = 12 cells) or TSPAN-7 + GFP (light grey; n = 12 cells); P = (−120 mV: 0.4, −100 mV: 0.3, −80 mV: 0.8, −60 mV: 0.5, −40 mV: 0.4). Statistical analysis was conducted using unpaired two-sample t tests; uncertainty is expressed as 95% CI.

Because the impact of TSPAN-7 on glucose-stimulated Ca²⁺ influx was greater than on depolarization-mediated Ca²⁺ influx, we went on to examine the effect of TSPAN-7 on glucose-regulated K_ATP channels. Heterologous expression of TSPAN-7 with K_ATP channels did not affect K_ATP current density (Fig. 6H). This suggests that the glucose-dependent modulation of β-cell Ca²⁺ by TSPAN-7 is primarily mediated through interactions with L-type Ca_V channels.

TSPAN-7 KD augments insulin secretion from human β-cells

Regulation of β-cell L-type Ca_V channel activity by TSPAN-7 would be predicted to influence GSIS; therefore, basal and stimulated insulin secretion from human β-cells were measured. Interestingly, under basal (5.6 mM) glucose conditions, where Ca_V channel activity is low, TSPAN-7 KD β-cells secreted significantly more insulin compared to scramble shRNA controls (113.4 ± 90.2% increase in insulin secretion per hour; P = 0.04) (Fig. 7A). Because TSPAN-7 KD increased intracellular Ca²⁺ in mouse β-cells under basal glucose conditions, this could indicate a left-shift in glucose concentration required for Ca_V channel activation. Insulin secretion was also increased from β-cells stimulated with 14 mM glucose and 45 mM KCl (36.4 ± 23.1% increase in insulin secretion per hour; P = 0.02) (Fig. 7B). This suggests that TSPAN-7 controls not only human β-cell glucose sensitivity, but also maximum secretory capacity.

Figure 7. TSPAN-7 regulation of β-cell L-type voltage-dependent Ca²⁺ channel activity limits GSIS.

A, average normalized basal (5.6 mM glucose) insulin secretion from dispersed human islets expressing hTSPAN-7 shRNA (light grey; n = 8 human islet donors) relative to scramble shRNA (white; n = 8 human islet donors); P = 0.04. B, average normalized stimulated (14 mM glucose + 45 mM KCl) insulin secretion from dispersed human islets expressing hTSPAN-7 shRNA (light grey; n = 8 human islet donors) relative to scramble shRNA (white; n = 8 human islet donors); P = 0.02. Statistical analysis was conducted using one-sample t tests; uncertainty is expressed as 95% CI.

Discussion

Proper control of insulin secretion is critical for prevention of hyperglycaemia and maintenance of blood glucose homeostasis (Prentki & Matschinsky, 1987; Rhodes & White, 2002; Röder et al. 2016). It is well established that Ca²⁺ entry into β-cells through L-type Ca_V channels is a critical regulator of GSIS (Curry et al. 1968; MacDonald et al. 2005; Rorsman et al. 2012; Reinbothe et al. 2013); however, the mechanisms that control β-cell L-type Ca_V channel function are incompletely understood. Here, we demonstrate that TSPAN-7 regulates β-cell Ca²⁺ handling and GSIS in part by modulating L-type Ca_V channel activity. TSPAN-7 interaction with both Ca_V1.2 and Ca_V1.3 channels decreased Ca²⁺ conductance. TSPAN-7 also altered activation/inactivation kinetics in a channel-specific manner and accelerated recovery of Ca_V1.2 channels from VDI. The importance of TSPAN-7 control of Ca_V channel function was exemplified in β-cells with TSPAN-7 KD, which displayed enhanced glucose-stimulated Ca²⁺ influx, accelerated Ca²⁺ oscillation frequency and increased insulin secretion. This suggests that TSPAN-7 tunes the set-point of GSIS and regulates the rate of Ca²⁺ oscillations and thus pulsatile insulin release. Furthermore, during sustained secretagogue stimulation, a TSPAN-7-mediated decrease in Ca_V1.2 channel VDI could increase β-cell Ca²⁺ influx and therefore help maintain GSIS.

We provide the first evidence indicating that TSPAN-7 controls L-type Ca_V channel function. However, previous studies have reported that TSPAN proteins can interact with and control Ca_V channel activity (Kahle et al. 2011; Mallmann et al. 2013). For example, TSPAN-13 regulates Ca_V2.2 through an interaction with the channel VSD in membrane-spanning domain IV of the α₁-subunit (Mallmann et al. 2013). Similarly, tetraspanin Ca_V γ-subunits control Ca_V channel function through interactions with the channel VSD (Wu et al. 2015). Because TSPAN-13 modulation of Ca_V2.2 function is caused by diminished coupling efficiency between voltage sensor activation and pore opening (Mallmann et al. 2013), it is possible that TSPAN-7 limits movement of L-type Ca_V channel VSDs and/or allosterically controls pore opening. A yeast two-hybrid assay also showed that TSPAN-7 interacts with P/Q-type Ca_V channels (Kahle et al. 2011). Therefore, TSPAN-7 may affect the function of non-L-type Ca_V channels. However, in β-cells, this is probably not the case because non-L-type Ca_V currents were unaffected by TSPAN-7 KD. TSPAN-7 regulation of Ca_V channel activity would also be predicted to influence all islet cell types where it is expressed. Single cell RNA sequencing and immunofluorescence staining reveal that TSPAN-7 expression is not restricted to β-cells, although it is also expressed in α- and δ-cells (Blodgett et al. 2015; DiGruccio et al. 2016; Segerstolpe et al. 2016). Activation of P/Q-type Ca_V channels is required for GCG secretion; thus, it is probable that an interaction between α-cell TSPAN-7 and P/Q-type Ca_V channels regulates α-cell function (Ramracheya et al. 2010; Rorsman et al. 2012). Heterogenous TSPAN-7 staining in α-cells may also suggest that relative TSPAN-7 expression contributes to differences in α-cell Ca²⁺ handling. In addition, R- and L-type Ca_V channels control δ-cell function (Zhang et al. 2007); therefore, an interaction of TSPAN-7 with δ-cell Ca_V channels presumably impacts glucose-stimulated Ca²⁺ entry and somatostatin (SST) secretion. Taken together, these findings suggest an important role for TSPAN-7 in regulating islet Ca²⁺ handling.

In addition to direct regulation of L-type Ca_V channels TSPAN-7 modulates phospholipid signalling through an interaction with phosphatidylinositol 4-kinase, a key regulator of phosphatidylinositol 4,5-bisphosphate (PIP2) production (Yauch & Hemler, 2000; Bassani et al. 2012; Lee et al. 2017). PIP2 is an important signalling molecule that influences the function of many islet ion channels (Baukrowitz et al. 1998; Shyng et al. 2000; Wu et al. 2002). For example, PIP2 binds to and holds K_ATP in an open conformation through an association with Kir6.2 close to the SUR1 binding site (Martin et al. 2017). However, because heterologous expression of TSPAN-7 did not affect K_ATP activity, this suggests that PIP2 control of K_ATP channel activity is not the primary mediator of TSPAN-7 regulation of β-cell Ca²⁺ handling. PIP2 also supports Ca_V channel function but, instead, decreases open probability by right-shifting the voltage threshold for channel activation (Wu et al. 2002). Because TSPAN-7 did not affect the voltage threshold for Ca_V activation, PIP2 modulation of Ca_V channel function probably does not play a role in the effect of TSPAN-7 on β-cell Ca²⁺ handling. Although our data do not suggest that TSPAN-7 control of PIP2 influences β-cell L-type Ca_V channel or K_ATP channel activity, it is possible that modulation of PIP2 locally in TSPAN-enriched microdomains impacts β-cell function.

TSPAN-7 control of receptor signalling could also directly or indirectly modulate β-cell Ca²⁺ handling and insulin secretion. For example, TSPAN-7 regulates the function and trafficking of ionotropic AMPA receptors (AMPARs) in hippocampal neurons (Bassani et al. 2012; Murru et al. 2017). Because glutamate stimulation of β-cell AMPARs has been shown to enhance glucose-stimulated Ca²⁺ influx and GSIS by directly depolarizing V_m and by indirectly inhibiting K_ATP activity (Wu et al. 2012), it is possible that TSPAN-7 regulation of AMPAR activity also influences β-cell Ca²⁺ handling and hormone secretion. However, other studies have reported that expression of GRIA1-4 (i.e. genes encoding AMPAR subunits) is low in β-cells and that glutamate only modestly affects β-cell Ca²⁺ handling and insulin secretion (Cabrera et al. 2008). Therefore, TSPAN-7 regulation of AMPAR function would be expected to serve a more prominent role in pancreatic α-cells, which display higher GRIA1-4 expression and a larger response to glutamate stimulation (Cabrera et al. 2008; Cho et al. 2010). TSPAN-7 has also been shown to interact with D2 dopamine receptors (DRD2s), which promotes DRD2 endocytosis (Lee et al. 2017). DRD2s are D2-like dopamine receptors (e.g. DRD2, DRD3 and DRD4), which are G_αi-coupled G protein-coupled receptors. DRD2s, as well as DRD3s, are expressed in β-cells, and thus dopamine secreted from β-cells at high glucose reduces intracellular cAMP and activates hyperpolarizing G protein-coupled inwardly-rectifying K⁺ channels; this results in decreased Ca_V channel activity and GSIS (Rubí et al. 2005; Ustione & Piston, 2012; Farino et al. 2019). Therefore, TSPAN-7 control of G_αi-coupled DRD2 turnover in β-cells could influence receptor desensitization kinetics and/or surface expression, and thus tune dopamine inhibition of GSIS. Importantly, because AMPAR and DRD2/D3 expression is not restricted to β-cells, it is probable that TSPAN-7 regulation of one or both affects Ca²⁺ handling and hormone secretion in other islet cell types (i.e. α- and/or δ-cells). Additional experiments will examine whether TSPAN-7 regulation of AMPAR activity and/or DRD2/D3 signalling influences islet cell Ca²⁺ handling and hormone secretion.

TSPAN-7 also regulates filamentous (F)-actin formation, which controls β-cell insulin granule trafficking, membrane access and exocytosis (Bassani et al. 2012; Ménager & Littman, 2016; Ménager, 2017). Under low glucose conditions, a dense subplasmalemmal F-actin network is generated by the actin-related protein 2/3 (ARP2/3) complex, which restricts insulin secretion; glucose stimulation induces β-cell F-actin remodelling that facilitates GSIS (Goley et al. 2010; Tran et al. 2015; Ménager & Littman, 2016). Cytosolic protein interacting with C kinase 1 (PICK1) acts as an allosteric switch for F-actin polymerization by inhibiting the ARP2/3 complex (Ménager & Littman, 2016; Ménager, 2017). PICK1 exists as a homodimer containing spatially separated PDZ and lipid binding BAR domains; this arrangement facilitates plasma membrane tethering and interactions with as many as two binding partners. TSPAN-7 contains a C-terminal PDZ-binding domain that interacts with the PDZ domain of PICK1, which locks PICK1 in an inactive conformation and results in the formation of highly branched F-actin networks (Bassani et al. 2012; Ménager & Littman, 2016; Ménager, 2017). Therefore, β-cell TSPAN-7 may play an important role in limiting insulin secretion under low glucose conditions by stabilizing F-actin. This is supported by our data showing that TSPAN-7 KD leads to an increase in basal (5.6 mM glucose) insulin secretion. F-actin also influences hormone granule fusion and exocytosis in α-cells (Hutchens & Piston, 2015; Hughes et al. 2018) and presumably δ-cells; therefore, TSPAN-7 modulation of actin dynamics probably does not affect GCG and/or SST secretion. Interestingly, PICK1 also localizes to the surface of insulin granules and global PICK1 KO impairs proinsulin processing and insulin granule maturation (Cao et al. 2013; Li et al. 2018). This results in decreased GSIS and elevated secretion of proinsulin from immature insulin granules. Based on a strong cytosolic staining pattern in islets, it is probable that TSPAN-7 serves additional intracellular functions possibly through interactions with PICK1. Therefore, improved proinsulin processing and insulin granule maturation resulting from TSPAN-7 KD could contribute to the observed increase in basal insulin secretion and GSIS. Future studies will aim to investigate the properties of islet TSPAN-7/PICK1 complexes and examine whether these interactions influence hormone secretion.

The importance of β-cell TSPAN-7 is further demonstrated by the presence of autoantibodies against TSPAN-7 in the serum of 35% of type 1 diabetic patients and in 21% of LADA patients, which could indicate a link between TSPAN-7 and diabetes (McLaughlin et al. 2016; Shi et al. 2019). Because the majority of islet autoantigens associated with type 1 diabetes (T1D) pathogenesis are important components of the β-cell secretory pathway (i.e. insulin, zinc transporter 8, islet antigen 2/2β, islet cell autoantigen 69) (Pihoker et al. 2005; Roep & Peakman, 2012), it is probable that TSPAN-7 also plays a part in this process. Moreover, L-type Ca_V channels form clusters that interact with and facilitate insulin granule plasma membrane fusion and exocytosis (Gandasi et al. 2017). Because we show that TSPAN-7 directly modulates L-type Ca_V channel activity, TSPAN-7 also would be predicted to localize to and could influence formation of Ca_V channel clusters where insulin secretion occurs. Interestingly, TSPAN-7 has also been shown to localize to β-cell exosomes along with another known T1D autoantigen, glutamate decarboxylase 65 (Sheng et al. 2011). Exosomes are small (50–100 nm) extracellular vesicles that originate from exocytosis of multivesicular bodies, and which transport protein, lipid and/or nucleic acid cargo to neighbouring or distant cells. Importantly, exosomes are secreted in response to stressful conditions and have been identified as targets of autoimmune response in non-obese diabetic mice (Sheng et al. 2011). Thus, it is possible that secretion of proinflammatory exosomes during T1D pathogenesis contributes to development of TSPAN-7 autoantibodies.

In conclusion, these finding show that β-cell L-type Ca_V channel activity is controlled by TSPAN-7, which limits glucose-stimulated Ca²⁺ influx, reduces the speed of Ca²⁺ oscillations and decreases insulin secretion. Our data indicate that TSPAN-7 interacts with and directly regulates L-type Ca_V channel function. Binding of TSPAN-7 reduces Ca_V1.2 and Ca_V1.3 channel Ca²⁺ conductance, modulates activation/inactivation kinetics in a channel specific manner, and accelerates the recovery of Ca_V1.2 channels from VDI. These results suggest that TSPAN-7 serves important roles in determining the set-point at which GSIS occurs, tuning the speed of Ca²⁺ oscillations and pulsatile insulin secretion, as well as sustaining insulin secretion during periods of secretagogue stimulation. Furthermore, these results can be extended to other cell types that express TSPAN-7 and where Ca²⁺ serves as an important intracellular signal, such as in hippocampal neurons, as well as in pancreatic α- and δ-cells.

Key points.

• Tetraspanin (TSPAN) proteins regulate many biological processes, including intracellular calcium (Ca²⁺) handling. TSPAN-7 is enriched in pancreatic islet cells; however, the function of islet TSPAN-7 has not been identified.

• Here, we characterize how β-cell TSPAN-7 regulates Ca²⁺ handling and hormone secretion. We find that TSPAN-7 reduces β-cell glucose-stimulated Ca²⁺ entry, slows Ca²⁺ oscillation frequency and decreases glucose-stimulated insulin secretion.

• TSPAN-7 controls β-cell function through a direct interaction with L-type voltage-dependent Ca²⁺ channels (Ca_V1.2 and Ca_V1.3), which reduces channel Ca²⁺ conductance. TSPAN-7 slows activation of Ca_V1.2 and accelerates recovery from voltage-dependent inactivation; TSPAN-7 also slows Ca_V1.3 inactivation kinetics.

• These findings strongly implicate TSPAN-7 as a key regulator in determining the set-point of glucose-stimulated Ca²⁺ influx and insulin secretion.

Biography

Matthew T. Dickerson is a Research Instructor in the Molecular Physiology and Biophysics (MPB) Department at Vanderbilt University in Nashville, TN, USA. He has a BS (2006) and PhD (2012) in Chemical Engineering from the University of Kentucky in Lexington, KY. His continuing research focuses on tetraspanin regulation of pancreatic islet function.

Data availability statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.