Phosphorylation of the Cav3.2 T-type calcium channeldirectly regulates its gating propertiesIulia Blesneaca,b,c, Jean Chemina,b,c, Isabelle Bidauda,b,c, Sylvaine Huc-Brandta,b,d, Franck Vandermoerea,b,and Philippe Lorya,b,c,1

aUniversité de Montpellier, CNRS UMR 5203, Département de Neuroscience & Biologie des Canaux Ioniques, Institut de Génomique Fonctionnelle, MontpellierF-34094, France; bINSERM, Montpellier F-34094, France; cLabEx Ion Channel Science and Therapeutics, Montpellier F-34094, France; and dLaboratoireDynamique des Interactions Membranaires Normales et Pathologiques, Université de Montpellier, CNRS UMR 5235, Montpellier F-34095, France

Edited by William A. Catterall, University of Washington School of Medicine, Seattle, WA, and approved September 25, 2015 (received for review June16, 2015)

Phosphorylation is a major mechanism regulating the activity ofion channels that remains poorly understood with respect to T-typecalcium channels (Cav3). These channels are low voltage-activatedcalcium channels that play a key role in cellular excitability andvarious physiological functions. Their dysfunction has been linked toseveral neurological disorders, including absence epilepsy andneuropathic pain. Recent studies have revealed that T-type channelsare modulated by a variety of serine/threonine protein kinasepathways, which indicates the need for a systematic analysis ofT-type channel phosphorylation. Here, we immunopurified Cav3.2channels from rat brain, andwe used high-resolutionMS to constructthe first, to our knowledge, in vivo phosphorylation map of avoltage-gated calcium channel in a mammalian brain. We identifiedas many as 34 phosphorylation sites, and we show that the vastmajority of these sites are also phosphorylated on the human Cav3.2expressed in HEK293T cells. In patch-clamp studies, treatment ofthe channel with alkaline phosphatase as well as analysis ofdephosphomimetic mutants revealed that phosphorylation regu-lates important functional properties of Cav3.2 channels, includingvoltage-dependent activation and inactivation and kinetics. We alsoidentified that the phosphorylation of a locus situated in the loop I-IIS442/S445/T446 is crucial for this regulation. Our data show thatCav3.2 channels are highly phosphorylated in the mammalian brainand establish phosphorylation as an important mechanism involvedin the dynamic regulation of Cav3.2 channel gating properties.

T-type calcium channel | Cav3.2 subunit | patch clamp |mass spectrometry | phosphorylation

Voltage-gated calcium channels (L-, N-, P/Q-, R-, and T-types)mediate calcium entry in many different cell types in response

to membrane depolarization and action potentials. Calcium influxthrough these channels serves as an important second messenger ofelectrical signaling, initiating a variety of cellular events and phys-iological functions (1, 2). Among the family of voltage-gated cal-cium channels, T-type calcium channels (Cav3 family) have uniqueelectrophysiological properties, because they display low voltage-activated calcium currents with rapid activation/inactivation kinet-ics. In neurons, small changes in the membrane potential near theresting potential can activate T-type channels, favoring furthermembrane depolarization and repetitive firing of action potentials(3–5). These unique gating properties of T-type channels makethem important in many different processes, including neuronalspontaneous firing and pacemaker activities, rebound burst firing,sleep rhythms, sensory processing, and neuronal differentiation, aswell as in pathological conditions, such as epilepsy and neuropathicpain (6).To properly assure this plurality of physiological functions, a

tight control of T-type calcium channels is necessary. An im-portant regulatory mechanism is phosphorylation, the fastest andmost frequent posttranslational modification for a protein. Ionchannels, especially voltage-gated channels, are critically regulatedby phosphorylation. Voltage-dependent sodium and potassium

channels have been shown to be the target of multiple phosphor-ylation events, regulating different channel functions and beinginvolved in pathological states, like epilepsy (7–11). Also, severalstudies have shown the crucial role of the L-type/Cav1 phosphor-ylation in important physiological functions, like the fight or flightresponse (12–15).Regarding T-type channels, phosphorylation remains poorly

understood. There are three Cav3 pore-forming proteins (Cav3.1,Cav3.2, and Cav3.3 subunits) all displaying typical properties ofT-type channels when expressed in heterologous cell systems (3, 16).Among them, the Cav3.2 channel seems to be particularly sensitiveto various types of regulation, including phosphorylation. To date,several serine/threonine kinases, like PKA, PKC, or CamKII, havebeen shown to regulate Cav3.2 activity (reviewed in refs. 17–20);however, in most cases, this regulation is tissue-dependent, and littleis still known about its molecular basis. Cav3.2 channels bear morethan 100 multiple intracellular serine and threonine residues thatare predicted to be phosphorylated by common prediction algo-rithms, like NetPhos2.0 (21). However, which of these residues areactually phosphorylated and what functional impact this phos-phorylation will have remain to be determined.In this study, we investigate the phosphorylation pattern of the

Cav3.2 isoform of the T-type channels and its role in Cav3.2channel properties. Using an MS approach, we have establishedthe first, to our knowledge, phosphorylation map of the Cav3.2channel in a mammalian brain and a human cell line. Then, byusing alkaline phosphatase (AP) and dephosphomimetic mutantsin patch-clamp experiments, we reveal the importance of phos-phorylation in modulating Cav3.2 gating properties. We have

Significance

Ion channels are membrane proteins essential for signal gen-eration and transmission in the nervous system. They are finelyregulated, and even small changes in their activity can triggerimportant physiological or pathological consequences on brainfunction. A regulatory mechanism of particular importance isphosphorylation. In this study, we assess the impact of phos-phorylation on the activity of a specific ion channel, the T-typecalcium channel Cav3.2. We show for the first time, to ourknowledge, that Cav3.2 is highly phosphorylated in vivo and ina mammalian brain as well as in a human cell line, and weidentify many phosphorylation sites critical for the way thatthe channel opens in response to changes in membrane po-tential and neuronal excitability.

Author contributions: I. Blesneac, F.V., and P.L. designed research; I. Blesneac, J.C.,I. Bidaud, and F.V. performed research; S.H.-B. contributed new reagents/analytic tools;I. Blesneac, J.C., F.V., and P.L. analyzed data; and I. Blesneac and P.L. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: philippe.lory@igf.cnrs.fr.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1511740112/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1511740112 PNAS | November 3, 2015 | vol. 112 | no. 44 | 13705–13710

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

May

12,

202

0

also identified a phosphorylation hot spot situated in the loopconnecting domains I and II of the channel that plays a crucialrole in this regulation. Altogether, this study provides importantinsights regarding how phosphorylation regulates Cav3.2 channels.

ResultsCav3.2 Is Highly Phosphorylated in Brain and Heterologous Cells. Toestablish the phosphorylation status of Cav3.2 in brain tissue, weimmunopurified Cav3.2 from rat brain lysate, digested the pu-rified protein with proteases, and further analyzed the resultingpeptides by high-resolution MS (details are in Materials andMethods); 62% of the protein could be detected by this method.This percentage of protein coverage was even higher (78%) whenconsidering only the major intracellular domains, the principaltarget of kinases and phosphatases (Fig. S1A).Based on the MS/MS spectra, in total, 34 different phosphor-

ylation sites were identified (Fig. S2 and Dataset S1 show therepresentative spectra), 26 of which have not been described else-where to our knowledge. Among the others, two of them, S1104and S1203, have been previously identified as targeted by PKA andCamKII, respectively (22, 23), in adrenocortical carcinoma cells,and six others (S445/T446, S541, S1171, T1172, S2201, and S2360)were described in high-throughput studies using automatic assig-nation of phosphorylation sites from MS/MS spectra (24, 25). Ex-cept the loop connecting the domains III and IV (LIII-IV), which isthe smallest, all intracellular loops contained several phosphory-lated threonine and serine residues. No tyrosine phosphorylationwas detected.In view of further studying the functional impact of this phos-

phorylation, we also assessed the phosphorylation status of Cav3.2expressed in HEK293T cells. This human cell line is a widely usedsystem for studying ion channel function, and it is important toknow whether the phosphorylation pattern of the recombinantCav3.2 protein is similar to the one observed in native tissue(brain). For expression in HEK293T cells, we used the humanisoform of Cav3.2, bearing an HA epitope that allows an efficientpurification of the protein. The rat and the human Cav3.2 isoforms

are highly similar, having a sequence identity of 84%. After ex-pression and immunoprecipitation, we subjected the recombinantCav3.2 to the same type of analysis as the native protein from ratbrain. The overall coverage and the coverage of the major intra-cellular loops were 71% and 87%, respectively (Fig. S1B). Theseexperiments allowed us to identify of a total of 43 phosphosites (Fig.1B, Fig. S2, and Dataset S1 show representative MS/MS spectra).Comparing these sites with those identified in rat brain revealedthat the vast majority of phosphorylation sites found in the brainwere also phosphorylated in HEK293T cells (27 sites from a total of34). Among the remaining ones, some (S18, S541, S767, andS2354) were not conserved between rat and human Cav3.2 iso-forms, and only three were conserved but not detected as phos-phorylated in HEK293T cells. More details about phosphositecorrespondence between the rat and the human isoforms are givenin Table S1. These results suggest that the phosphorylation pat-terns of Cav3.2 are mostly conserved between the HEK293T cellline and the brain as well as between the human and rat isoforms.

Phosphorylation Regulates the Biophysical Properties of Cav3.2Channels. The large number of phosphorylated sites found onCav3.2 suggests extensive modulation of the channel by phos-phorylation. To rapidly evaluate the impact that this phosphory-lation could have on channel activity, we used a phosphatase, theAP, in patch-clamp studies. This phosphatase was already suc-cessfully used in others studies to show the role that phosphory-lation has on Kv channels (26). We, therefore, expressed thehuman isoform of Cav3.2 in HEK293T cells and measured itsactivity in a whole-cell patch-clamp configuration. Subsequently,we allowed AP to dialyze into the cell through the patch pipette toinduce channel dephosphorylation. The AP dialysis (100 U/mL for30 min) led to a significant shift toward more negative potentialsof the steady-state inactivation and activation curves (Fig. 2).The half-inactivation potential (V1/2) was shifted ∼19 mV from−70.12 ± 0.7 to −89.53 ± 2.2 mV (P < 0.001) (Fig. 2 A and B), andsimilarly, the half-activation potential was shifted ∼16 mV from−46.02 ± 0.4 to −61.6 ± 1 mV (P < 0.001) (Fig. 2 C and D). Also,

A

B Fig. 1. Schematic representation of the Cav3.2 MSanalysis. (A) Cartoon of the membrane topology ofCav3.2 showing the phosphorylation sites identifiedin rat brain. (B) Cartoon of the membrane topologyof human Cav3.2 showing the phosphorylation sitesidentified in in HEK293T cells. The red circles indicatethe sites identified as phosphorylated in both ratbrain and HEK293T cells. Correspondence betweenthe amino acid numbering of the rat and humanCav3.2 isoform is given in Table S1. The unambiguoussites are bold, and the ambiguous sites are bold anditalicized. Liquid chromatography–MS/MS sequencecoverage of the Cav3.2 protein and theMS/MS spectraof the Cav3.2 peptides can be found in Figs. S1 and S2and Dataset S1.

13706 | www.pnas.org/cgi/doi/10.1073/pnas.1511740112 Blesneac et al.

Dow

nloa

ded

by g

uest

on

May

12,

202

0

the activation and inactivation kinetics became significantly fasteron AP treatment. For instance, at −40 mV, the τinact decreasedfrom 28.9 ± 3.3 to 15.5 ± 1.5 ms (P < 0.01) (Fig. S3A), and the τactdecreased from 6.5 ± 0.7 to 2.4 ± 0.2 ms (P < 0.001) (Fig. S3B).No significant changes were observed regarding other biophysicalproperties of the Cav3.2 channel, including recovery from in-activation and deactivation kinetics (Fig. S3 C and D). A set ofexperiments designed to determine the reversal potential showedno significant effect of the AP treatment on this parameter either(Fig. S3E).These data suggest that the gating properties of Cav3.2 channels

are greatly influenced by the phosphorylation status. To determinewhether the AP effect is because of a direct dephosphorylation ofthe channel, we mutated the serine and threonine residues that wepreviously identified as phosphorylated to alanine, thus mimickingthe dephosphorylation.In a first series of experiments, six constructs were generated,

each of them designed to assess the functionality of the identifiedphosphoserine and phosphothreonine residues from a particularintracellular region of human Cav3.2 (Table 1 shows detailsconcerning each construct). No significant effect on the Cav3.2steady-state activation and inactivation properties or kinetics wasobserved when all of the identified phosphorylated sites of thecytoplasmic N-terminal part of the protein were mutated to al-anine (the construct named N-ter Ala is shown in Fig. 3 andTable 1). The same result was obtained when the phosphorylatedsites from LII-III and LIII-IV were mutated (constructs LII-IIIAla and LIII-IV Ala, respectively, are shown in Fig. 3 and Table1). To identify a role in gating of the C-terminal part of Cav3.2, weconstructed a deletion mutant by introducing a stop codon atQ1886 (C-terΔ). No change in the biophysical properties of theT-type current was observed with the C-terΔ mutant (Fig. 3 andTable 1). To evaluate the role of the large intracellular loop

connecting domains I and II (LI-II ; 357 aa), we generated twoconstructs where all of the identified phosphorylated sites in theproximal part (LI-IIA) or the distal part (LI-IIB), respectively,were mutated to alanines (LI-IIA Ala and LI-IIB Ala are shown inFig. 3 and Table 1). We did not observe any effect on Cav3.2current properties with the LI-IIB Ala construct. However, theLI-IIA Ala construct displayed steady-state activation and inacti-vation curves significantly shifted toward more negative potentials.The extent of the shift was −7 mV for the inactivation curve (V1/2shifted from −69.02 ± 0.27 to −76.59 ± 0.91 mV; P < 0.001)(Fig. 3B) and −9 mV for the activation curve (V1/2 shifted from−45.07 ± 0.34 to −54.71 ± 1.59 mV; P < 0.001) (Fig. 3C). Inaddition, the activation and inactivation kinetics were also fasterfor the LI-IIA Ala mutant (Table 1). No significant change in thecurrent density was seen for any of the mutants.Because the LI-IIA Ala construct has seven mutated serine

and threonine residues, we sought to determine which of themwas responsible for the gating effect. Therefore, three additionalconstructs were generated: LI-IIA1 (carrying the mutationsS442A, S445A, and T446A), LI-IIA2 (S532A and S535A), andLI-IIA3 (S558A and S561A). Whereas LI-IIA2 and LI-IIA3 hadcurrent properties similar to the WT, the LI-IIA1 mutant exhibi-ted a shift in both the steady-state half-activation and half-inacti-vation potentials and faster activation and inactivation kineticssimilar to the LI-IIA Ala construct (Table 1), indicating that thephosphosites from this region are important for the voltage-dependent gating of the channel. Furthermore, treatment of theLI-IIA1 mutant with AP resulted in a hyperpolarizing shift involtage-dependent activation and inactivation to the same endpoints as the AP-treated WT Cav3.2 channel (Table 1). The shiftobserved for the LI-IIA1 mutant in the presence of AP was sig-nificantly reduced compared with the one obtained for the WT.For the mutant channel, the steady-state activation and inactivationcurves were shifted by 10 and 13 mV, respectively, compared withthe 16- and 19-mV shifts measured for the WT. These experimentsfurther support that the Cav3.2 gating properties are modulated bythe phosphorylation of the S442/S445/T446 cluster.The LI-IIA1 construct is a triple mutant carrying the following

mutations: S442A, S445A, and T446A. Individual mutation ofthese sites led to three single alanine mutants each exhibiting a shiftin the steady-state activation and inactivation properties as well asaccelerated kinetics, similar to those observed for LI-IIA Ala or thetriple mutant LI-IIA1 (Fig. 4 A and B). Importantly, these specificvoltage-dependent biophysical properties of the LI-IIA1 mutantcould also be observed in another cell line, the neuroblastomaNG108-15 cell line, which constitutively expresses Cav3.2. Inthis neuronal environment, expression of LI-IIA1 also yielded ahyperpolarizing shift of the steady-state activation and inactivationcurves as well as faster activation and inactivation kinetics (Fig.S4). In addition, we have performed action potential voltage-clamp experiments. Because the role of T-type channels in pro-moting burst activity in thalamic neurons is well documented (27),these experiments were done using a burst firing activity typical ofthe thalamic reticular neurons as a voltage-clamp command (28).Using this protocol on cells displaying similar T-type currentdensity, we observed that, at a physiological membrane potential(−76 mV), the LI-IIA1 mutant generated significantly smallercalcium current compared with the WT channel (Fig. 4C). Thisresult was further confirmed by the quantification of both the peakcurrent density (Fig. 4D) and the inward current area (Fig. 4E),which accounts for total calcium entry during burst activity pro-tocol. To evaluate the impact of Cav3.2/T-type channel phos-phorylation on neuronal excitability, we have used a NEURONmodel of thalamic reticular neuron (28). Introduction of our ex-perimental values for the WT, LI-IIA1 mutant, and AP-treatedWT channels revealed marked differences in firing patterns (Fig.4F). We found that both mutant and AP-treated WT channelsled to increased firing. The number of spikes was increased for the

A B

Membrane poten al (mV)

-100 -80 -60 -40 -20 0 0.0

0.2

0.4

0.6

0.8

1.0

Membrane poten al (mV)

g/g

max

25 ms

100 pA

25 ms

100 pA

5 ms

100 pA

5 ms

100 pA

Non-treated AP-treated

C D Non-treated AP-treated

-120 -100 -60 -40

0.2

0.4

0.6

0.8

1.0

I/I m

ax

-80 0.0

Fig. 2. AP effect on biophysical properties of Cav3.2 channels. (A) Represen-tative currents elicited by test pulses from different holding potentials (−110,−100, −90, −85, −80, −75, −70, −65, −60, and −50 mV) for 5 s to −30 mV for thecells nontreated or treated with AP. HEK293T cells expressing human Cav3.2were treated by adding AP into the patch pipette (100 U/mL) and letting it di-alyze into the cell for 30 min. For the control cells, the current was recorded after30 min of dialysis of the internal solution without AP. Note the change in thecurrent amplitude for the trace in blue (holding potential was −75 mV) in thepresence of AP as well as the changes in the kinetics. (B) Steady-state inactivationcurve fit from traces in A for the control cells (black circles; n = 5) and the AP-treated cells (red squares; n = 11). (C) Representative currents elicited from aholding potential of −100 mV to different test pulse potentials (−90, −80, −70,−65, −60, −55, −50, −45, −40, −35, −30, −25, −20, and −10 mV) for the cellsnontreated or treated with AP. Note the change in the amplitude of the currentfor the trace in blue (test pulse potential was −50 mV) in the presence of AP aswell as the changes in the kinetics. (D) Steady-state activation curve fit fromtraces in C for the control cells (black circles; n = 4) and the AP-treated cells (redsquares; n = 12).

Blesneac et al. PNAS | November 3, 2015 | vol. 112 | no. 44 | 13707

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

May

12,

202

0

LI-IIA1 mutant compared with the WT channel, an increase thatwas even stronger for the AP-treated WT channel. Also, the la-tency for the first spike was shorter for the dephosphorylatedchannels (50.7 ms for AP-treated WT and 55.7 ms for LI-IIA1mutant compared with 82.9 ms for the WT), and the averageinterspike interval was decreased (36.1 ms for AP-treated WT and36.8 ms for LI-IIA1 mutant compared with 45.1 ms for the WT).

DiscussionPhosphorylation is a major mechanism regulating the activity ofion channels that remains poorly understood for T-type calciumchannels (17–20). Taking advantage of the MS approach, weprovide here the first, to our knowledge, phosphorylation map ofthe Cav3.2 T-type calcium channel purified from both nativetissue (rat brain) and recombinant system (human HEK293Tcells). Importantly, we show that phosphorylation highly modu-lates channel gating properties, and we report that phosphory-lation of a locus within the intracellular loop connecting domainsI and II is directly involved in this gating control.Our MS experiments reveal that a large number of serine and

threonine residues of the Cav3.2 protein is phosphorylated: 34for the native channel in the rat brain and 43 for the recombinanthuman Cav3.2 channel in the HEK293T cells. At least 27 phos-phosites identified in vivo were also found to be phosphorylated inthe heterologous expression system (Table S1), suggesting that thephosphorylation pattern of the Cav3.2 protein is broadly con-served among species (rat and human) and cellular systems(recombinant channel in HEK293T and native channel in thebrain). Our work has validated 8 sites identified elsewhere (TableS1) and identified 26 never described in vivo phosphosites. Amongthese discovered sites, some (S18, S647, S687, and S2193 in the rat

isoform) were not even predicted to be phosphorylated by widelyused computer algorithms, like NetPhos2.0. This underscoresthe limitations of algorithm-based predictions and shows the

Table 1. Electrophysiological properties of WT and mutant Cav3.2 channels

Constructions MutationsActivation V1/2 (mV)

and K (mV)Inactivation V1/2 (mV)

and K (mV)Kinetics at −45 mV τact (ms)

and τinact (ms)

WT — −45.07 ± 0.34 (65) −69.02 ± 0.27 (68) 9.476 ± 0.33 (62)5.26 ± 0.07 (65) 4.06 ± 0.06 (68) 29 ± 0.7 (62)

N-ter Ala S29A, S32A, S49A, S51A, S53A −45.97 ± 1.0 (8) −69.97 ± 1.02 (8) 9.036 ± 1.07 (7)4.963 ± 0.15 (8) 4.199 ± 0.2 (8) 26.14 ± 2.75 (8)

LI-IIA Ala S442A, S4445A, T446A, S532A, S535A,S558A, S561A

−54.71 ± 1.59(7)* −76.59 ± 0.914(7)* 4.347 ± 0.43 (5)*5.354 ± 0.41(7) 3.609 ± 0.07(7) 17.96 ± 1.67(5)*

LI-IIB Ala S650A, S653A, S687A, S715A, S717A, S749A,T751A, S758A

−44.24 ± 0.6 (6) −67.84 ± 0.25 (7) 12.14 ± 0.42 (5)5.029 ± 0.16 (6) 4.021 ± 0.09 (7) 32.75 ± 3.5 (5)

LII-III Ala T1034A, S1035A, T1061A, S1071A, S1073A,S1091A, S1090A, S1099A, S1103A, S1107A,S1127A, S1144A, S1174A, S1175A, S1198A, S1246A

−45.6 ± 1.52 (5) −68.97 ± 0.87 (5) 8.235 ± 0.94 (4)5.264 ± 0.16 (5) 4.102 ± 0.05 (5) 25.83 ± 3.6 (4)

LIII-IV Ala S1587A, T1588A, S1591A −46.03 ± 1.62 (8) −70.35 ± 0.78 (10) 8.203 ± 0.64 (11)5.604 ± 0.18 (8) 3.873 ± 0.08 (10) 29.76 ± 2.1 (11)

C-terΔ Q1886Stop −47.79 ± 1.4 (7) −69.09 ± 0.97 (8) 8.558 ± 1.08 (5)4.961 ± 0.34 (7) 4.087 ± 0.1 (8) 26.02 ± 2.16 (5)

LI-II A1 S442A, S4445A, T446A −54.05 ± 0.7 (11)* −76.17 ± 0.7 (11)* 4.561 ± 0.24 (11)*5.297 ± 0.1 (11) 3.509 ± 0.11 (11)† 23.49 ± 1.06 (11)‡

LI-II A2 S532A, S535A −45.69 ± 1.03 (5) −69.09 ± 0.4 (5) 7.927 ± 0.48 (5)5.326 ± 0.35 (5) 4.189 ± 0.18 (5) 27.34 ± 2.17 (5)

LI-II A3 S558A, S561A −44.04 ± 1.69 (6) −66.4 ± 0.866 (6) 11.28 ± 1.20 (7)5.356 ± 0.36 (6) 4.045 ± 0.25 (6) 27.2 ± 2.8 (7)

WT AP — −61.6 ± 1.04 (12)* −89.53 ± 2.2 (11)* 2.919 ± 0.26 (11)*6.124 ± 0.14 (12) 4.224 ± 0.04 (11) 15.82 ± 1.5 (11)*

LI-II A1 AP S442A, S4445A, T446A −64.58 ± 1.08 (8)* −89.3 ± 1.3 (9)* 2.584 ± 0.211 (8)*5.572 ± 0.32 (8) 3.763 ± 0.06 (9) 21.33 ± 2.3 (8)†

V1/2 represents the half-activation and half-inactivation potentials, respectively; K is the slope factor; and τact and τinact are the activation and inactivationkinetics, respectively. The kinetics values were obtained by fitting the current traces obtained using the I-V curve protocol with a double-exponential function.The number of cells is indicated in parentheses.*P < 0.001 compared with the WT channels.†P < 0.01 compared with the WT channels.‡P < 0.05 compared with the WT channels.

A

Membrane Poten al (mV)

g/g

max

-90 -70 -50 -30 -100.0

0.2

0.4

0.6

0.8

1.0C

N-terLI-IIA LI-IIB

LII-III C-ter

LIII-IV

I II III IV

-40-100 -80 -600.0

0.2

0.4

0.6

0.8

1.0

Membrane Poten al (mV)

I/Im

ax

B

Fig. 3. Electrophysiological analysis of the phosphorylation mutants.(A) Cartoon of the membrane topology of Cav3.2 showing the region mutatedfor the different constructions (Table 1 shows details regarding the mutatedresidues for each construction). (B and C) Steady-state inactivation curves andsteady-state activation curves, respectively, for the WT (black circles), N-ter Ala(blue triangles), LI-II A Ala (red squares), LI-II B Ala (orange diamonds), LII-III Ala(purple squares), LIII-IV Ala (brown squares), and C-terΔ (gray triangles).

13708 | www.pnas.org/cgi/doi/10.1073/pnas.1511740112 Blesneac et al.

Dow

nloa

ded

by g

uest

on

May

12,

202

0

importance of undertaking MS-based proteomic studies to char-acterize the phosphorylation status of a protein.The striking extent of Cav3.2 phosphorylation raises questions as

to its role in Cav3.2 regulation. The remarkable hyperpolarizingshift in the activation and inactivation curves of the channel (16and 19 mV, respectively) obtained by treatment with a phosphatase(AP) indicates that the phosphorylation level of the cell will greatlyinfluence the biophysical properties of the channel. Mutating all ofthe identified phosphosites of human Cav3.2 shows that de-phosphorylation of S442, S445, and /or T446 is, at least partially,responsible for the shift observed in the presence of AP. Futurestudies are needed to determine if the additional shift observedwith AP (7 mV for the activation and 12 mV for the inactivationcurve) is because of an indirect action through channel partners orthe direct dephosphorylation of other Cav3.2 phosphosites thatcould not be identified in this study. Indeed, some phosphorylationsites, particularly those with low stoichiometry, might have been

missed by our MS approach. Either way, our data show that thephosphorylation level of the cell can drastically influence thevoltage-dependent gating of Cav3.2 channel.Our results also show that the phosphorylation locus S442/S445/

T446 has a particular importance for the voltage-dependent gating ofCav3.2. These three amino acids were detected by MS on theARHLS442NDS445T446LASFSEPGSCYEELLK peptide (spectra areshown in Dataset S1) for the human isoform expressed in HEK293Tcells and the YLS442NDS445T446LASFSEPGSCYEELLK peptide(Fig. S2) for the rat brain isoform. The latter was detected in themonophosphorylated as well as the diphosphorylated forms. For themonophosphorylated peptide, the MS/MS data showed a hybridspectrum with b and y ions supporting both the phosphorylation ofS442 and S445 or T446 (Fig. S2). Analysis of theMS/MS spectrum ofthe human peptide allowed us to determine that one of the residuesS442, S445, or T446 is phosphorylated, but data were not sufficient topinpoint exactly which one of the three is actually phosphorylated.From a functional point of view, individual mutations of S442, S445,and T446 to alanine led to single mutants exhibiting the same bio-physical properties as the triple-alanine mutant, suggesting thatmutation/dephosphorylation of either one of these residues is suf-ficient to trigger functional changes in Cav3.2 activity.S442, S445, and T446 are located within the loop connecting

domains I and II (I-II loop) in a region that was previously de-scribed as a “gating brake” (29–31). Indeed, Vitko et al. (29)showed that deletion of the first 62 aa of the I-II loop (429–491)allows the Cav3.2 channel to open at more negative membranepotential by shifting the steady-state activation and inactivationcurves, a phenotype recapitulated in our LI-IIA1 mutant. It istempting to postulate that phosphorylation/dephosphorylation ofthe residues S442, S445, and/or T446 would affect the conforma-tion of this region, leading to a change in channel gating. Our dataare, therefore, consistent with the critical role of this region in thegating of the Cav3.2 channel and provide a possible physiologicalregulation of the gating brake. It is interesting to notice that thevast majority of the described Cav3.2 absence epilepsy-linkedmutations are also on the I-II loop (32). One of these mutations,C456S, which induces a 5-mV shift in the activation curveand an increase in surface expression of the protein (29, 33),has been shown to increase the spontaneous firing rate ofhippocampal neurons and thus, an increase in seizure suscep-tibility (34). This underscores the importance of Cav3.2 gatingproperties in neuronal firing, a property that could be dynam-ically regulated by loop I-II phosphorylation/dephosphorylationas documented in this study.Our modeling experiments predict that neuronal firing pat-

terns depend on T-type channel phosphorylation. The LI-IIA1mutant and the AP-treated WT channels could trigger an in-creased firing in a thalamic reticular neuron model, suggestingthat, in their native environment, these neurons would be moreexcitable under dephosphorylated conditions. It remains, how-ever, possible that the complex interplay between the gatingproperties of the Cav3.2 channel would give a different output inanother neuronal environment. It is interesting to notice that thevalues reported in the literature for the half-activation and half-inactivation potentials (V1/2) of the T-type current markedlyvary between neurons. For instance, the half-activation potentialranges from −45 to −60 mV (35), values that are consistent withour findings (−45 mV for WT and −61 mV for AP-treated WT).Although different contributions of T-type channel isoforms ormethodological differences could explain this wide range in half-activation potentials, it could also be caused by different phos-phorylation states. Our MS approach is a global one, not takinginto account the cell type specificity. Future experiments should bedesigned to assess the phosphorylation pattern of the channelprotein in a more specific set of neurons, allowing better corre-lation with the gating properties of the channel.

A B

-90 -70 -50 -30 -10 0.0

0.2

0.4

0.6

0.8

1.0

Membrane Poten�al (mV)

g/g

max

Membrane Poten�al (mV) -100 -80 -60 -40

0.0

0.2

0.4

0.6

0.8

1.0 WT LI-IIA1

T446A S445A S442A

I/I m

ax

C

0.0

0.1

0.2

0.3

0.4 **

Pea

k cu

rren

t (A.

U)

0

2

4

6

8

10 *

Are

a (A

.U)

D E

25 ms

WT

LI-IIA1

100 pA

100 mV

F

20 mV

100 ms

WT LI-IIA1 WT AP-treated

Fig. 4. Effect of mutation of the phosphorylation loci S442, S445, and T446.(A and B) Steady-state inactivation curves and steady-state activation curves,respectively, for the WT (black circles; n = 62–68), LI-IIA1 (red squares; n = 11),S442A (purple triangles; n = 8), S445A (gray diamonds; n = 7), and T446A (bluesquares; n = 8). (C) Representative calcium current trace for the WT (blacktrace) and the LI-IIA1 mutant (red trace) recorded during a voltage-clamp firingpattern of a thalamic reticular neuron generated by the NEURON model. Pleasenote that these representative recordings were obtained in cells exhibitingsimilar current density when recorded at −30 mV from a holding potential (HP)of −110 mV. (D) Calcium current area quantification from traces in C. (E) Peakcalcium current quantification from traces in C. The WT is represented in black(n = 3), and the LI-IIA1 mutant is represented in red (n = 3). (F) Reticular thalamicneuron firing simulation using values obtained with WT (black trace), LI-IIA1mutant (red trace), and AP-treated WT (blue trace) channels. The current clampresponses were elicited by a 0.037-nA current injection for 150 ms in a dendriteof the detailed cell model from a holding potential of −70 mV.

Blesneac et al. PNAS | November 3, 2015 | vol. 112 | no. 44 | 13709

NEU

ROSC

IENCE

Dow

nloa

ded

by g

uest

on

May

12,

202

0

Except for S442, S445, and T446, alanine mutations of all ofthe other identified phosphosites resulted in no apparent effecton Cav3.2 biophysical properties. Most of our alanine mutationswere done concomitantly for all of the phosphosites found in aparticular loop, and it is, indeed, possible that some phosphositesindividually shift gating into different directions, thereby result-ing in the absence of a detectable biophysical effect. A functionalrole of some of the other Cav3.2 phosphosites may be related toother modes of channel regulation not investigated in this studyor depend on a particular channel environment. For instance,Hu et al. (22) showed that phosphorylation of S1107 (S1104 inthe rat isoform) could induce an inhibition of the T-type currentonly in the presence of G protein βγ-subunits (Gβ2γ2), which areabsent in HEK293T cells. It is, therefore, possible that proteinpartners are needed to reveal the functional role of a particularphosphorylation site of Cav3.2 channel. It could also be possiblethat a functional effect may be observed when two or severalresidues from distinct intracellular domains are concomitantlyphosphorylated/dephosphorylated. The shift observed with APtreatment may have more complex underpinnings that arecomprised of multiple components going in both directions. Fi-nally, it should also be noticed that the phosphorylation mapsprovided in this study correspond to the basal state of thechannel in brain tissue and HEK293T cells. It may, therefore, bepossible that other serine and threonine residues of the channelcould be phosphorylated in response to physiological or patho-

logical stimuli and signaling pathways that are not constitutivelyactive in brain or HEK293T cells.Taken together, our results show that the Cav3.2 T-type calcium

channel is highly phosphorylated both in vivo (brain) and in vitro(HEK293T cells) and that phosphorylation greatly contributes tothe biophysical properties of the channel, likely leading to im-portant physiological consequences. Our findings pave the way toa better understanding of the dynamic regulation of Cav3.2 activityby phosphorylation in physiology and pathology.

Materials and MethodsHEK-293T cells were transfected with human Cav3.2-pcDNA3.1 constructsusing standard protocols. Mutagenesis was performed using the QuikChangeII XL Kit (Agilent). For MS experiments, Cav3.2 channel immunoprecipitationwas performed from 6-wk-old rat brain and transfected HEK-293T cells.T-type calcium currents were recorded from HEK-293T cells using the whole-cell patch-clamp technique. AP experiments were conducted by adding100 U/mL enzyme (Roche) into the patch pipette solution with cells dialyzed for30 min. Details on constructs, immunoprecipitation experiments, MS, electro-physiology, and NEURON modeling are provided in SI Materials and Methods.Data are presented as mean ± SEM.

ACKNOWLEDGMENTS. We thank Dr. Maithé Corbani for technical help andDr. François Rassendren, Dr. Nathalie Guérineau, and Dr. Chris Jopling forhelpful comments and critical reading of the manuscript. This work wassupported by CNRS, INSERM, the Université de Montpellier, Agence Natio-nale pour la Recherche Grants ANR-10-BLAN-1601 and ANR-09-MNPS-035,and LabEx Ion Channel Science and Therapeutics.

1. Catterall WA, Perez-Reyes E, Snutch TP, Striessnig J (2005) International Union ofPharmacology. XLVIII. Nomenclature and structure-function relationships of voltage-gated calcium channels. Pharmacol Rev 57(4):411–425.

2. Simms BA, Zamponi GW (2014) Neuronal voltage-gated calcium channels: Structure,function, and dysfunction. Neuron 82(1):24–45.

3. Chemin J, et al. (2002) Specific contribution of human T-type calcium channel isotypes(alpha(1G), alpha(1H) and alpha(1I)) to neuronal excitability. J Physiol 540(Pt 1):3–14.

4. Perez-Reyes E (2003) Molecular physiology of low-voltage-activated t-type calciumchannels. Physiol Rev 83(1):117–161.

5. Perez-Reyes E (2006) Molecular characterization of T-type calcium channels. CellCalcium 40(2):89–96.

6. Cheong E, Shin HS (2013) T-type Ca2+ channels in normal and abnormal brain func-tions. Physiol Rev 93(3):961–992.

7. Berendt FJ, Park KS, Trimmer JS (2010) Multisite phosphorylation of voltage-gatedsodium channel alpha subunits from rat brain. J Proteome Res 9(4):1976–1984.

8. Baek JH, Rubinstein M, Scheuer T, Trimmer JS (2014) Reciprocal changes in phos-phorylation and methylation of mammalian brain sodium channels in response toseizures. J Biol Chem 289(22):15363–15373.

9. Park KS, Mohapatra DP, Misonou H, Trimmer JS (2006) Graded regulation of the Kv2.1potassium channel by variable phosphorylation. Science 313(5789):976–979.

10. Yang JW, Vacher H, Park KS, Clark E, Trimmer JS (2007) Trafficking-dependentphosphorylation of Kv1.2 regulates voltage-gated potassium channel cell surfaceexpression. Proc Natl Acad Sci USA 104(50):20055–20060.

11. Cerda O, Trimmer JS (2010) Analysis and functional implications of phosphorylationof neuronal voltage-gated potassium channels. Neurosci Lett 486(2):60–67.

12. Hulme JT, Westenbroek RE, Scheuer T, Catterall WA (2006) Phosphorylation of serine1928 in the distal C-terminal domain of cardiac CaV1.2 channels during beta1-ad-renergic regulation. Proc Natl Acad Sci USA 103(44):16574–16579.

13. Fuller MD, Emrick MA, Sadilek M, Scheuer T, Catterall WA (2010) Molecular mechanismof calcium channel regulation in the fight-or-flight response. Sci Signal 3(141):ra70.

14. Emrick MA, Sadilek M, Konoki K, Catterall WA (2010) Beta-adrenergic-regulatedphosphorylation of the skeletal muscle Ca(V)1.1 channel in the fight-or-flight re-sponse. Proc Natl Acad Sci USA 107(43):18712–18717.

15. Fuller MD, Fu Y, Scheuer T, Catterall WA (2014) Differential regulation of CaV1.2channels by cAMP-dependent protein kinase bound to A-kinase anchoring proteins15 and 79/150. J Gen Physiol 143(3):315–324.

16. Cain SM, Snutch TP (2010) Contributions of T-type calcium channel isoforms to neu-ronal firing. Channels (Austin) 4(6):475–482.

17. Zhang Y, Jiang X, Snutch TP, Tao J (2013) Modulation of low-voltage-activated T-typeCa²⁺ channels. Biochim Biophys Acta 1828(7):1550–1559.

18. Iftinca MC, Zamponi GW (2009) Regulation of neuronal T-type calcium channels.Trends Pharmacol Sci 30(1):32–40.

19. Huc S, et al. (2009) Regulation of T-type calcium channels: Signalling pathways and

functional implications. Biochim Biophys Acta 1793(6):947–952.20. Chemin J, Traboulsie A, Lory P (2006) Molecular pathways underlying the modulation

of T-type calcium channels by neurotransmitters and hormones. Cell Calcium 40(2):

121–134.21. Blom N, Gammeltoft S, Brunak S (1999) Sequence and structure-based prediction of

eukaryotic protein phosphorylation sites. J Mol Biol 294(5):1351–1362.22. Hu C, Depuy SD, Yao J, McIntire WE, Barrett PQ (2009) Protein kinase A activity

controls the regulation of T-type CaV3.2 channels by Gbetagamma dimers. J Biol

Chem 284(12):7465–7473.23. Yao J, et al. (2006) Molecular basis for the modulation of native T-type Ca2+ channels

in vivo by Ca2+/calmodulin-dependent protein kinase II. J Clin Invest 116(9):2403–2412.24. Trinidad JC, et al. (2012) Global identification and characterization of both

O-GlcNAcylation and phosphorylation at the murine synapse. Mol Cell Proteomics 11(8):

215–229.25. Huttlin EL, et al. (2010) A tissue-specific atlas of mouse protein phosphorylation and

expression. Cell 143(7):1174–1189.26. Murakoshi H, Shi G, Scannevin RH, Trimmer JS (1997) Phosphorylation of the Kv2.1 K+

channel alters voltage-dependent activation. Mol Pharmacol 52(5):821–828.27. Cain SM, Snutch TP (2013) T-type calcium channels in burst-firing, network synchrony,

and epilepsy. Biochim Biophys Acta 1828(7):1572–1578.28. Destexhe A, Contreras D, Steriade M, Sejnowski TJ, Huguenard JR (1996) In vivo,

in vitro, and computational analysis of dendritic calcium currents in thalamic reticular

neurons. J Neurosci 16(1):169–185.29. Vitko I, et al. (2007) The I-II loop controls plasma membrane expression and gating of

Ca(v)3.2 T-type Ca2+ channels: A paradigm for childhood absence epilepsy mutations.

J Neurosci 27(2):322–330.30. Arias-Olguín II, et al. (2008) Characterization of the gating brake in the I-II loop of

Ca(v)3.2 T-type Ca(2+) channels. J Biol Chem 283(13):8136–8144.31. Perez-Reyes E (2010) Characterization of the gating brake in the I-II loop of CaV3 T-type

calcium channels. Channels (Austin) 4(6):453–458.32. Chen Y, et al. (2003) Association between genetic variation of CACNA1H and child-

hood absence epilepsy. Ann Neurol 54(2):239–243.33. Vitko I, et al. (2005) Functional characterization and neuronal modeling of the effects

of childhood absence epilepsy variants of CACNA1H, a T-type calcium channel.

J Neurosci 25(19):4844–4855.34. Eckle VS, et al. (2014) Mechanisms by which a CACNA1H mutation in epilepsy patients

increases seizure susceptibility. J Physiol 592(Pt 4):795–809.35. Huguenard JR (1996) Low-threshold calcium currents in central nervous system neu-

rons. Annu Rev Physiol 58:329–348.

13710 | www.pnas.org/cgi/doi/10.1073/pnas.1511740112 Blesneac et al.

Dow

nloa

ded

by g

uest

on

May

12,

202

0