Mitochondrial potassium channel Kv1.3 mediatesBax-induced apoptosis in lymphocytesIldiko Szabo*, Jurgen Bock†‡, Heike Grassme†, Matthias Soddemann†, Barbara Wilker†, Florian Lang§, Mario Zoratti¶,and Erich Gulbins†�

*Department of Biology and ¶Consiglio Nazionale della Ricerche Institute of Neuroscience and Department of Biomedical Sciences, University of Padova,Viale G. Colombo 3, 35121 Padua, Italy; †Department of Molecular Biology, University of Essen, Hufelandstrasse 55, 45122 Essen, Germany;and §Department of Physiology, University of Tuebingen, Gmelinstrasse 5, 72076 Tuebingen, Germany

Communicated by Tullio Pozzan, University of Padua, Padua, Italy, May 13, 2008 (received for review December 14, 2007)

The potassium channel Kv1.3 has recently been located to the innermitochondrial membrane of lymphocytes. Here, we show that mouseand human cells that are genetically deficient in either Kv1.3 ortransfected with siRNA to suppress Kv1.3-expression resisted apo-ptosis induced by several stimuli, including Bax over-expression.Retransfection of either Kv1.3 or a mitochondrial-targeted Kv1.3restored cell death. Bax interacted with and functionally inhibitedmitochondrial Kv1.3. Incubation of isolated Kv1.3-positive mitochon-dria with recombinant Bax, t-Bid, or toxins that bind to and inhibitKv1.3 successively triggered hyperpolarization, formation of reactiveoxygen species, release of cytochrome c, and marked depolarization.Kv1.3-deficient mitochondria were resistant to Bax, t-Bid, and thetoxins. Mutation of Bax at K128, which corresponds to a conservedlysine in Kv1.3-inhibiting toxins, abrogated its effects on both Kv1.3and mitochondria. These findings suggest that Bax mediates cyto-chrome c release and mitochondrial depolarization in lymphocytes, atleast in part, via its interaction with mitochondrial Kv1.3.

ion channels � mitochondria

Mitochondria are key organelles of the intrinsic apoptoticpathway, mediating cell death in pathological and stress

conditions via the release of proapoptotic factors (1). This releaseis often a direct consequence of the activation of proapoptotic Bcl-2family proteins, in particular Bax and Bak (2), although release canalso be induced in cells lacking these proteins (3). However, atpresent, it is unknown how Bax and Bak mediate these apoptoticevents.

We investigated the role of mitochondrial ion channels in apo-ptosis mediated by Bax. Recent pharmacological data demon-strated an important role of calcium-dependent (4) and ATP-regulated (5) mitochondrial potassium channels in the control ofapoptotic death. Although apparently conflicting data exist on thefunction of these channels in apoptosis, their activation appears toprotect against cell death induced by massive ischemia (4, 5).Furthermore, LETM1 (leucine zipper-, EF-hand-containing trans-membrane protein 1), a putative component of the K�/H� anti-porter of the inner mitochondrial membrane (IMM), has recentlybeen shown to control cell viability (6). Thus, potassium fluxesacross the IMM are emerging as regulators of cell death in varioussystems.

Here, we define the molecular role of the mitochondrial potas-sium channel Kv1.3 in apoptosis. Activation of plasma membraneKv1.3 is intimately involved in T cell proliferation and maturation(7), but studies employing actinomycin D to induce cell death alsosuggest the importance of Kv1.3 in apoptosis (8). We have recentlyreported that, like other channels expressed in multiple subcellularlocations (9–12), the potassium channel Kv1.3 is present in afunctionally active form not only in the plasma membrane, but alsoin the IMM of lymphocytes (13). The present study demonstratesthat IMM Kv1.3 interacts with Bax in a toxin-like mode to triggercell death.

ResultsMitochondrial Kv1.3 is Critically Involved in Apoptosis. To address therole of Kv1.3 in apoptosis, we used a genetic model in which

Kv1.3-deficient CTLL-2 cells (14) were transfected with either anexpression vector for Kv1.3 (pJK-Kv1.3) (cells designated CTLL-2/Kv1.3) or control vector (pJK) (designated CTLL-2/pJK) (13, 15).

Incubation of Kv1.3-reconstituted CTLL-2/Kv1.3 cells, Jurkatcells, or activated human peripheral blood lymphocytes (PBL) withTNF�, staurosporine, sphingomyelinase, or C6-ceramide resultedin DNA fragmentation, cytochrome c release, mitochondrial de-polarization, and morphological alterations typical of apoptosis,whereas Kv1.3-deficient CTLL-2/pJK cells were resistant (Fig. 1A,and data not shown). Prolonged (24–36 h) incubation of CTLL-2/pJK cells with these stimuli finally resulted in apoptosis, suggestingthat Kv1.3 is involved in an amplification loop, such as the mito-chondrial pathway that mediates apoptosis.

To confirm the role of Kv1.3 in apoptosis, we transfected eitheractivated human PBL or Jurkat cells with either siRNA, whichalmost completely suppressed both the expression and activity ofKv1.3 in the transfected population [supporting information (SI)Fig. S1], or control siRNA, respectively, which was without effect onKv1.3. Suppression of Kv1.3 inhibited the proapoptotic effect ofstaurosporine that was used as a paradigmatic drug to stimulate themitochondrial pathway of apoptosis (Fig. S1 A and B). Almostidentical results were obtained for the induction of apoptosis inJurkat cells transfected with siRNA to target Kv1.3 (data notshown).

To specifically address the role of intracellular Kv1.3 in apoptosis,we transfected CTLL-2/Kv1.3 and control CTLL-2/pJK cells withan expression vector for Bax and determined whether apoptosis wastriggered by over-expression of Bax (Fig. 1C). Bax over-expressiontriggered massive apoptosis of Kv1.3-positive cells (CTTL-2/Kv1.3and Jurkat), whereas Kv1.3-deficient CTLL-2 cells were resistant toBax. These data establish a critical role of Kv1.3 in the induction ofapoptosis and suggest an important role of intracellular Kv1.3 inBax-induced apoptosis.

To directly test the role of mitochondrial Kv1.3 in apoptosis, wetransfected Kv1.3-deficient CTLL-2 cells with an expression vectorfor an EYFP-Kv1.3 construct that specifically targets the expressionof Kv1.3 in mitochondria (pJK/mito-EYFP-Kv1.3). Transfectionwith this mitochondrial Kv1.3 construct restored apoptosis inCTLL-2 cells when treated with either TNF� or staurosporine (Fig.1B), whereas cells lacking Kv1.3 did not show significant apoptosis12 h after stimulation. Mitochondrial expression (Fig. S2A) andorientation of Kv1.3 (Fig. S2B), mito-EYFP-Kv1.3 (Fig. S2C), or

Author contributions: I.S., J.B., F.L., M.Z., and E.G. designed research; I.S., J.B., H.G., M.S.,B.W., M.Z., and E.G. performed research; I.S., M.Z., and E.G. analyzed data; and I.S., M.Z.,and E.G. wrote the paper.

The authors declare no conflict of interest.

‡Present address: Department of Internal Medicine, University of Regensburg, 93053Regensburg, Germany.

�To whom correspondence should be addressed. E-mail: erich.gulbins@uni-duisburg-essen.de.

This article contains supporting information online at www.pnas.org/cgi/content/full/0804236105/DCSupplemental.

© 2008 by The National Academy of Sciences of the USA

www.pnas.org�cgi�doi�10.1073�pnas.0804236105 PNAS � September 30, 2008 � vol. 105 � no. 39 � 14861–14866

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

020

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

020

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

020

translocation of Bax (Fig. S2 D and E) were confirmed in controlexperiments.

To test the role of an additional Kv channel for apoptosistriggered by staurosporine, we transfected CTLL-2 cells that lack

expression of all known Kv channels with an expression vector ofeither Kv1.1 or control vector. The results (Fig. S3) demonstratethat Kv1.1 is localized in mitochondria after transfection of CTLL-2cells and, most importantly, that expression of Kv1.1 in CTLL-2

8 nM GST-Bax

9 nM GST

Control

Who

le-c

ell c

urre

nt (

pA)

Who

le-c

ell c

urre

nt (

pA)

0

200

400

600

800

1000

Who

le-c

ell p

eak

curr

ent a

t +70

mV

Control

200 pA

100 ms

0 2 4 6 8 10 12 14 16 180

20

40

60

80

100

Nor

mal

ised

cur

rent

(%)

GST-Bax (nM)

500 pA

100 ms

400 pA

100 ms

100 pA

200 ms

17 nM GST-Bax7 nM GST-Bax

3.5 nM GST-Bax

Control

Who

le-c

ell c

urre

nt (

pA)

Who

le-c

ell c

urre

nt (

pA)

Control

8 nM GST-Bax

pH 6.7

Control50 nM GST-Bax

pH 6.0

8 nM GST-Bax

Fig. 2. Bax physically interacts with and inhibits Kv1.3. Patch-clamp studies reveal that recombinant GST-Bax inhibits Kv1.3. (Upper) (Left) Whole-cell currents(voltage from �70 to �70 mV at 45-sec intervals) in Jurkat lymphocytes. GST-Bax (8 nM) added to the same patch inhibited current within 135 sec; controlGST-protein (9 nM) did not decrease the current within 450 sec. (Center) Currents elicited at �70 mV at various GST-Bax concentrations. Arrows indicate peakcurrents after addition of GST-Bax. (Right) whole-cell mean peak currents (at �70 mV) for controls (n � 12) and GST-Bax-treated cells (n � 8). (Lower) (Left)Dose–response curve (n � 3) yielding an IC50 of 4 nM. (Center and Right) A change of pH to 6.7 strongly reduces the inhibitory effects of GST-Bax on Kv1.3; atpH 6.0 of the bath solution, inhibition of Kv1.3 is abolished even with 50 nM GST-Bax. Currents are recorded (at �70 mV) in the absence (black line) and presence(gray line) of Bax. IC50 for GST-Bax at pH 6.7: 12 nM (n � 3) and at pH 6.0: ��50 nM (n � 4). In another set of experiments, currents at �70 mV were measuredin control conditions at pH 6.7 (1,076 � 127 pA, n � 14) and with 20 nM GST-Bax at pH 6.7 (320 � 76 pA, n � 4).

Ap

opto

sis

[%]

5

15

25

0 12

35

time of stimulation [hrs]

45

55

65

untr

eate

dC

-

Cer

amid

e T

NF

Stau

rosp

orin

eSp

hing

omye

linas

e

6

untr

eate

dC

-

Cer

amid

eT

NF

Stau

rosp

orin

eSp

hing

omye

linas

e

6

untr

eate

dC

-

Cer

amid

eT

NF

Stau

rosp

orin

eSp

hing

omye

linas

e

6

untr

eate

dC

-

Cer

amid

eT

NF

Stau

rosp

orin

eSp

hing

omye

linas

e

6

CTLL-2 / pJK CTLL-2 / Kv 1.3

*

*

*

*

0 12

A

CTLL-2/mito-Kv1.3

CTLL-2/control transfected

Staurosporine

EYFP-mito-Kv1.3

70.37%

17.79%

1.59%

10.25%

0.96%

97.92%

0.77%

0.35%

unstimulated 1.3%

97.61%

0.68%

0.41%101 103

101

103

TNF-3.78%

94.31%

1.21%

0.7%101

101

103

103

101

101

103

103

101

101

103

103

72.5%

16.9%

1.1%

9.5%101

101

103

103

0.79%

24.44%

0.15%

74.62%

101

101

103

103

Cy3

-Ann

exin

V

Cy3

-Ann

exin

V

B

Fig. 1. Kv1.3 is required for Bax-, TNF�, staurosporine-, sphingomyelinase-, and C6-ceramide- induced apoptosis. (A) Treatment of CTLL-2/Kv1.3 cells with 20�M C6-ceramide, 1 �M staurosporine, 100 ng/ml TNF�, or 1 u/ml sphingomyelinase results in DNA fragmentation indicative of apoptosis. CTLL-2/pJK cells lackingKv1.3 were resistant. Means � SD are shown (n � 3; *, P � 0.05, t test). (B) Transfection of Kv1.3-deficient CTLL-2 cells with a construct that specifically targetsKv1.3 expression to mitochondria (pJK/EYFP-mito-Kv1.3) restores apoptosis in these cells when stimulated with 100 ng/ml TNF� or 1 �M staurosporine. Cells werestained with fluorescent Annexin to determine apoptosis. Shown are representative data of each three independent experiments.

14862 � www.pnas.org�cgi�doi�10.1073�pnas.0804236105 Szabo et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

020

cells was sufficient to restore cell death in these cells after treatmentwith staurosporine.

These studies as well as published data (16) suggest that, inaddition to Kv1.3, other Kv channels may be expressed in mito-chondria and, thereby, may mediate cell death upon induction ofthe mitochondrial death pathway.

Bax and Kv1.3 Physically Interact. To elucidate the molecular mech-anisms by which apoptosis is mediated via Kv1.3, we investigated

whether Kv1.3 interacts with proapoptotic Bax. First, we performedpatch-clamp experiments in a whole-cell configuration employingfull-length Bax and GST-Bax that lacks the C-terminal transmem-brane domain (17). GST-Bax or full-length Bax (both added to thebath) inhibited plasma membrane Kv1.3 with an IC50 in the low nMrange (Fig. 2 and Fig. S4). Both recombinant GST-Bcl-2 andGST-Bcl-xL were without effect (data not shown). Seals on themitochondrial membrane were unstable when the pipette wasbackfilled with GST-Bax. However, we expected this behavior

Orange = 0 minRed = 5 min treatmentBlue = 15 min treatmentGreen = CCCP

CTLL/Kv1.3 GST

CTLL/pJKGST

CTLL/pJK untreated

100

200

Cou

nts

CTLL/pJKGST-Bax

mm

mmm

CTLL/pJKBid

101

103

101

103

101

103

101

103

CTLL/Kv1.3untreated

100

200

Cou

nts

101

103

101

103

CTLL/Kv1.3GST-Bax

101

103

Jurkatuntreated

100

200

Cou

nts

101

103

Jurkat GST

101

103

CTLL/Kv1.3Bid

CTLL/Kv1.3Bid

101

103

CTLL/Kv1.3full length Bax

60

120

101

103

CTLL/pJKfull

length Bax

60

120

101

103

JurkatGST-Bax

101

103

unst

imul

ated

unst

imul

ated

CTLL/pJK CTLL/Kv1.3

GST

GST

-Bax

unst

imul

ated

GST

GST

-Bax

GST

GST

-Bax

Jurkat

Cytochrome c (mitochondrial)

Cytochrome c (released)

CTLL/pJK

CTLL/Kv1.3

unst

imul

ated

unst

imul

ated

t-B

id

t-B

id

1519

1519

kDa

CTLL/Kv1.3

CTLL/pJK

full

leng

th B

ax

unst

imul

ated

full

leng

th B

ax

unst

imul

ated

Cytochrome c(released)

Cytochrome c(mitochondrial)

18kDa

18

A

BC

TL

L-2

/pJK

CT

LL

-2/K

v1.3

Ju

rkat

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0+ GST-Bax

noadd.

RO

S p

rod

uct

ion

[a.

u.]

*

**

Ju

rkat

+ AA

CT

LL

-2/p

JK

CT

LL

-2/K

v1.3

Ju

rkat

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0+ GST-Bax

noadd.

RO

S p

rod

uct

ion

[a.

u.]

** *

Ju

rkat

+ AA

Amplex Red Cytochrome c absorptionC

Fig. 3. Bax triggers apoptosis via mitochondrial Kv1.3. (A) GST-Bax, t-Bid (each 5 nM), or full-length Bax (20 nM) induce initial hyperpolarization (right shift)followed by depolarization of Kv1.3-positive mitochondria isolated from CTLL-2/Kv1.3 and Jurkat cells, whereas the mitochondrial membrane potential ofKv1.3-negative mitochondria from CTLL/pJK cells was unaffected. Control GST proteins were without effect on any of the mitochondria. Full-length Bax wasadded with 0.1 nM t-Bid to facilitate activation of full-length Bax. At this concentration, t-Bid was without effect on ��m (data not shown). Shown arerepresentative FACS-plots of three independent experiments. Complete depolarization upon application of CCCP served as a control for mitochondria integrity.(B) Incubation of purified, Kv1.3-positive mitochondria with 5 nM recombinant GST-Bax, 5 nM t-Bid, or 20 nM full-length Bax results in release of cytochromec from mitochondria expressing Kv1.3, whereas mitochondria isolated from Kv1.3-deficient CTLL/pJK cells were resistant. Control GST was without effect;full-length Bax was added with 0.1 nM t-Bid, a dose of t-Bid that did not induce any cytochrome c release when added alone (data not shown). Blots arerepresentative of 10 similar experiments. (Upper) Cytochrome c released from mitochondria. (Lower) Cytochrome c remaining in mitochondria. (C) Fluorescencetraces reveal production of ROS in suspensions of Jurkat mitochondria. Mitochondria were treated with either 3.5 nM GST-Bax or 1.8 �M Antimycin A (AA). ROSproduction was evaluated with Amplex Red-assays (Left) or by the change of cytochrome c absorption (Right). The means � SD of the relative rates of ROS fromthree independent experiments are shown. *, differences are significant (P � 0.05).

Szabo et al. PNAS � September 30, 2008 � vol. 105 � no. 39 � 14863

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

020

because the orientation of the channel in mitochondria is the sameas in the plasma membrane as suggested by the sensitivity of themitochondrial channel to MgTx (13). Previous studies indicatedthat the binding of the highly specific Kv1.3 inhibitor toxin (e.g.,MgTx), which docks in the outer-facing vestibule of Kv1.3, is pHdependent, and that protonation of Histidine 404 of Kv1.3 weakensthe pore-toxin interaction (18). Accordingly, the IC50 of GST-Baxincreased from 4 to 12 nM when the bath solution pH was loweredto pH 6.7 and to a value �� 50 nM at pH 6.0, indicating a toxin-likeinteraction between Bax and the external vestibule of Kv1.3 (Fig. 2).The Bax-Kv1.3 interaction involves parts of the channel that facethe extracellular (or IMM) space (Fig. S5).

Coimmunoprecipitation experiments confirmed that Kv1.3 andBax physically interacted in both mouse and human cells only uponinduction of apoptosis (Fig. S6).

Finally, we isolated mitochondria from either untreated orstaurosporine-treated PBL or CTLL-2/Kv1.3 cells. Bax associatedwith mitochondrial Kv1.3 only in those mitochondria that wereisolated from apoptotic cells (Fig. S6). In the unstimulated sample,a small fraction of Bax was present in mitochondria as the proteinhas been previously shown to be loosely associated with the outermitochondrial membrane, but was not inserted into the membrane(19) and does not interact with IMM-located Kv1.3. In summary,the data indicate a direct interaction of Kv1.3 with Bax.

Bax Plus t-Bid Triggers Kv1.3-Mediated Hyperpolarization, Depolar-ization, and Cytochrome c Release in Isolated Mitchondria. To deter-mine whether the interaction of Kv1.3 with Bax is critically involvedin apoptosis, we incubated isolated, purified mitochondria with

recombinant full-length Bax, recombinant GST-Bax, or controlGST. Whereas GST-Bax is active without further modification,full-length Bax requires t-Bid-mediated activation. The latter wasadded together with a very low concentration of t-Bid (0.1 nM),which was previously shown to be without effect if added alone (18,20). We also treated the mitochondria with 5 nM t-Bid which waspreviously shown to trigger activation of Bak and Bax, which areeither constitutively present in mitochondria or, to some degree,loosely bound to the outer mitochondrial membrane, respectively(20, 21). Recombinant full-length Bax, GST-Bax, and t-Bid (5 nM)triggered hyperpolarization, followed by depolarization of the mi-tochondrial membrane (Fig. 3A) only in Kv1.3-positive mitochon-dria. Hyperpolarization is compatible with the inhibition of Kv1.3that carries a depolarizing K�-influx and was prevented by substi-tution of external potassium with sodium (Fig. S7). Incubation ofisolated Kv1.3-positive mitochondria with either Bax or t-Bid alsoresulted in a rapid release of cytochrome c from isolated Kv1.3-positive mitochondria, indicating apoptotic changes in these or-ganelles (Fig. 3B). GST was without effect on mitochondria (Fig. 3A and B). Importantly, similar changes, both qualitatively andquantitatively, were detected when either full-length Bax or GST-Bax were added to mitochondria that were isolated from geneticallymanipulated (CTLL-2/Kv1.3) and nonmanipulated (Jurkat) cells.

Next, we measured whether GST-Bax induces a Kv1.3-dependent increase in ROS release (see also SI Text). The datashow that GST-Bax induced a 1.5- to 2-fold increase in the rate ofROS production in Kv1.3-expressing mitochondria, whereas mito-chondria from CTLL-2/pJK cells did not respond (Fig. 3C). ROSrelease induced by GST-Bax was similar to that observed in the

0.2

0.4

0.6

0.8

1

1.2

1.4

RO

Spr

oduc

tion

[a.u

.]

CTLL/pJK

no a

ddit

ion

GST

-wt

Bax

GST

-K12

8E B

ax

CTLL/Kv1.3

no a

ddit

ion

GST

-wt

Bax

GST

-K12

8E B

ax

*A

101 103

GST

101 103

GST-Bax

101 103

GST- K128E Bax

101 103

GST-Bcl-xL

Orange: 0 min after treatmentRed: 5 min after treatmentBlue: 15 min after treatmentGreen: CCCP

100

200

Cou

nts

100

200

Cou

nts

101 103

untreated

m

mm

mm

B

untr

eate

d

GST

GST

-Bax

GST

-K12

8E B

ax

Cytochrome c (released)

Cytochrome c (mitochondrial)

15

19

15

19

kDa

C 200 ms

300 pA

a

b

200 ms

300 pA

abcd

D

Fig. 4. K128 of Bax is critical to block Kv1.3. (A–C) Mutation of Bax at K128 to glutamate (5 nM GST-Bax K128E) abrogated proapoptotic effects of Bax andprevented the release of ROS (A), changes in ��m (B), and release of cytochrome c (C), which occurred after incubation with 5 nM GST-Bax. (*, P � 0.05 comparedwith control, n � 3, t test). (D) (Left) K128E Bax (10 nM) does not decrease Jurkat whole-cell Kv1.3 currents (at �70 mV). Black: control; gray: 405 sec after additionof K128E Bax to bath. In five similar experiments, addition of K128E Bax did not significantly alter the current (�12 � 7%, P � 0.663). Preincubation with K128EBax did not induce a significant change in peak current at 70 mV [693 � 68 pA for control (n � 15) and 583 � 64 pA for K128E Bax-treated cells (n � 11), P �0.274]. (Right) Current traces obtained from the same patch in trace a: control conditions; trace b: after addition of 3 nM K128E Bax. Traces c and d were obtainedafter washing and subsequent addition of wild-type Bax to a final concentration of 3 nM in trace c and 12 nM in trace d.

14864 � www.pnas.org�cgi�doi�10.1073�pnas.0804236105 Szabo et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

020

presence of the respiratory chain inhibitor Antimycin A, previouslyshown to trigger ROS production and apoptosis (22) (Fig. 3C). Theantioxidants DTT, Tiron, and BHT and the PTP inhibitor CSA didnot affect hyperpolarization but prevented mitochondrial depolar-ization (Fig. S8 A and B). These antioxidants also inhibited therelease of cytochrome c (Fig. 3D and data not shown).

Kv1.3 Toxins Trigger Apoptotic Changes in Isolated Mitochondria. Toconfirm that inhibition of Kv1.3 by Bax mediates apoptotic changesin mitochondria, we tested whether pharmacological inhibitors ofKv1.3 (i.e., ShK, MgTx, and Psora-4) induce the same alterations asBax. ShK and MgTx are positively charged peptide toxins thatinteract with negatively charged residues in the Kv1.3 proteinvestibule to block the pore. Psora-4 [5-(4-phenylbutoxy-psoralen](23) is the most potent synthetic inhibitor of Kv1.3 available. Allthree compounds triggered the same changes as Bax in Kv1.3-positive isolated mitochondria (Fig. S9). Kv1.3-deficient mitochon-dria from CTLL/pJK cells did not respond to the toxins (Fig. S9),indicating the specificity of the effects of the inhibitors. Depolar-ization induced by toxins was also prevented by DTT and cyclo-sporin A (data not shown).

Bax Interacts with Kv1.3 Via a Toxin-Like Mechanism. All toxins thatblock Kv1.3 contain a lysine residue, which is critical for theinteraction with Kv1.3 (7, 24) (Fig. S10). A model of the structureof the membrane-integrated Bax monomer indicates that at leastamino acids 127 and 128, located between the 5th and 6th helicesof Bax, protrude from the outer mitochondrial membrane into theintermembrane space (25). The amino acid in position 128 is ahighly conserved, positively charged lysine, which may mimic theaction of the critical lysine in Kv1.3-blocking toxins (Fig. S10) bybinding to the ring of four aspartate residues (24) of the channelvestibule, which faces the intermembrane space in mitochondria(13). Furthermore, antiapoptotic proteins contain a negative chargein the corresponding position (amino acid 158 for Bcl-xL) (Fig. S10).Mutation of lysine 128 in GST-Bax to a negatively charged gluta-mate (GST-K128E Bax) prevented ROS release (Fig. 4A), hyper-polarization/depolarization of the mitochondrial membrane poten-tial (Fig. 4B), and release of cytochrome c (Fig. 5C); this mutationalso prevented the inhibition of Kv1.3 in whole-cell patch clampexperiments (Fig. 4D). The mutant GST-K128E Bax still integratedinto the mitochondrial membrane and formed ion channels whenreconstituted into black lipid bilayers (M.Z., unpublished results),indicating that the protein conformation is not significantly alteredby the mutation. GST, Bcl-xL, and Bcl-2 were without effect in allof these experiments.

DiscussionOur results identify mitochondrial Kv1.3 as a target for Bax andindicate that Kv1.3 is required for induction of apoptosis by Bax, atleast in lymphocytes. The amino acid residue K128 in Bax seems tobe particularly important for the interaction of Bax with Kv1.3because mutation of this lysine to a glutamic acid abrogatedinhibition of Kv1.3 by Bax and the proapoptotic activity of Bax. Thefinding that mutation of a single, critical amino acid controls theinteraction of a protein/peptide with a channel finds a precedent inthe interaction of agitoxin with KcsA (26). Our data suggest thatBax and the toxins ShK and MgTx act on Kv1.3 in a functionallysimilar way.

Bax-mediated inhibition of Kv1.3 results in hyperpolarization andROS release, which are upstream of both cytochrome c release (27)and PTP activation as indicated by the experiments with radicalscavengers. ROS may also play multiple roles in apoptosis, i.e., toinduce the dissociation of cytochrome c from the IMM, PTPactivation, and the release of cytochrome c from mitochondria uponformation of pores by oligomerized Bax (28).

Because the inactivation of Kv1.3 by Bax, ShK, MgTx, andPsora-4 provides a positive signal, i.e., hyperpolarization of themitochondrial membrane and ROS release, which are absent inKv1.3-deficient cells, deficiency of Kv1.3 is clearly not equivalent toinhibition of the channel. A similar situation was found in rodentbeta islet cells, in which either glucose- or drug-induced KATPchannel closure led to insulin secretion, whereas the lack of afunctional channel resulted in greatly reduced rather than increasedglucose-induced insulin release (29).

The interaction of Bax with ion channels may not be restricted toKv1.3; other mitochondrial Kv channels, such as Kv1.5 (16), mayhave similar functions. Furthermore, because up-regulation ofKv1.1 (and of a chloride channel) occurs in genetically modifiedKv1.3 knockout mice (30), Kv1.1 up-regulation may reconstituteapoptosis in these animals as suggested in the present studies.CTLL-2 cells do not express Kv1.1 (13, 14) enabling us to study therole of Kv1.3 in apoptosis. Conversely, mitochondrial KATP chan-nels (31) are not central to Bax-induced apoptosis in CTLL-2/Kv1.3cells, although this does not exclude a function of KATP channels inother cells and contexts.

In summary, our studies establish a direct link in lymphocytesbetween mitochondrial potassium conductance and the proapo-ptotic machinery in the cytoplasm, indicating a channel-blockingaction of Bax (Fig. 5).

MethodsCell Culture, Recombinant Proteins, Transient Transfections, and Cellular Apo-ptosis Assays. All cells were cultured as previously described and detailed inthe SI Text. Bax (amino acids 1–170) was cloned into pGEX-3X, expressed inBL21A1, and purified from bacterial lysates using glutathione-Sepharose(for details see SI Text). The K128E mutant of Bax was obtained by asite-directed mutagenesis PCR technique (Stratagene), cloned into pGEX-3X, and expressed and purified as a GST-fusion protein as above. Controlsconfirm that K128E Bax still integrates into the mitochondrial membrane,indicating that the protein conformation is not significantly altered by themutation.

Full-length recombinant hBax (32) was kindly provided by J. C. Martinou(University of Geneva, Geneva). Purified t-Bid was purchased from Axxora.

For thesiRNAexperiments, cellsweretransiently transfectedwith20nMAlexaFluor 488- or Cy3-coupled siRNA molecules by electroporation at 400 V with 5pulses,3mseceach,usingaBTXelectroporator.DeadcellswereremovedviaFicollpurification after 24 h. The cells were used for apoptosis or patch-clamp assays36–48 h after transfection.

CTLL-2/pJK or CTLL-2/Kv1.3 cells (2 107 cells per sample) were cotransfectedwith 10 �g of an expression vector of GFP-tagged actin (pcDNA-EGFP-actin) and50 �g of pcDNA-Bax (kindly provided by M. Weller, University of Tuebingen,Germany). Cells were electroporated as above and cultured for an additional 24 hwith IL-2. IL-2 was then removed as above, cells were incubated for 8 h, stainedwith Cy3-Annexin, and analyzed by fluorescence microscopy. Transfected cellswere identified by the expression of GFP-tagged actin and at least 400 GFP-positive cells per sample were scored for apoptosis.

Fig. 5. Model for action of mitochondrial Kv1.3 during apoptosis. Baxinhibits Kv1.3 via interaction of lysine 128 with the channel pore vestibule,resulting in hyperpolarization of the IMM. Hyperpolarization interferes withrespiration and triggers release of ROS, favoring detachment of cytochromec. Release of cytochrome c may be mediated by oxidation of membrane lipids,whereas activation of PTP might be caused by oxidation of cysteine residues.

Szabo et al. PNAS � September 30, 2008 � vol. 105 � no. 39 � 14865

CELL

BIO

LOG

Y

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

020

Transient transfection of mito-EYFP-Kv1.3. Expression of Kv1.3 that was spe-cifically targeted to mitochondria was achieved by electroporation of 107 CTLL-2cells with 25 �g of an EYFP-tagged Kv1.3 construct that was fused to the mito-chondrial targeting sequence of subunit VIII of human cytochrome c oxidase(Clontech) in the expression vector pJK (pJK-mito-EYFP-Kv1.3).

Apoptosis was determined by DNA fragmentation, cytochrome c release, andmorphological analysis as described in SI Text.

Patch-Clamp Experiments. Whole-cell currents were recorded with an EPC-7amplifier (List) (filter: 1 kHz, sampling rate: 5 kHz), as described in ref. 13 andSI Text.

Coimmunoprecipitations, Anti-Kv1.3 Antibodies. In the coimmunoprecipitationstudies, either cells or mitochondria were lysed in 4% CHAPS, 5 mM MgCl2, 137mM KCl, 1 mM EDTA, 1 mM EGTA, 20 mM Tris�HCl (pH 8.0), and 10 �g/ml bothaprotinin and leupeptin for 5 min at 4°C, insoluble material was cleared bycentrifugation for 5 min, and the supernatants were subjected to coimmunopre-cipitation experiments with either rabbit anti-Kv1.3 or anti-Bax antibodies. West-

ern blots were developed with the corresponding antibodies using the ECLsystem as detailed in SI Text.

Cytochrome c and ROS Release From, and Membrane Potential in, IsolatedMitochondria. Cells were homogenized by using a Dounce homogenizer and thesoluble fraction isolated. Cytochrome c was analyzed by Western blotting asdescribed in SI Text. Amplex Red fluorimetric assays and the change of cyto-chrome c absorption were used to determine ROS production by isolated mito-chondria. The membrane potential of isolated mitochondria was determined byFACS analysis of DioC6 (3)-stained mitochondria treated as indicated.

ACKNOWLEDGMENTS. The authors thank Drs. O. Pongs (University of Hamburg,Hamburg, Germany) and J. C. Martinou for important reagents and L. Scorrano,P. Bernardi, N. Mueller, G. Zanotti, U. De Marchi, L. Biasutto, G. Walton, and V.Teichgraeberforusefuldiscussions.ThisworkwassupportedinpartbyDFG-grantGu 335/13–3 and the International Association for Cancer Research (to E.G.),Italian Association for Cancer Research grants (to M.Z. and I.S.), and an EuropeanMolecular Biology Organization Young Investigator Program and a Progetti diRilevante Interesse Nazionale grant (to I.S.).

1. Green D, Kroemer G (2004) The pathophysiology of mitochondrial cell death. Science305:626–629.

2. Antignani A, Youle RJ (2006) How do Bax and Bak lead to permeabilization of the outermitochondrial membrane? Curr Opin Cell Biol 18:685–689.

3. Scorrano L, et al. (2003) BAX and BAK regulation of endoplasmic reticulum Ca2�: Acontrol point for apoptosis. Science 300:135–139.

4. Xu W, et al. (2002) Cytoprotective role of Ca2�-activated K� channels in the cardiacinner mitochondrial membrane. Science 298:1029–1033.

6. Ardehali H, O’Rourke B (2005) Mitochondrial K(ATP) channels in cell survival and death.J Mol Cell Cardiol 39:7–16.

6. Dimmer KS, et al. (2008) LETM1, deleted in Wolf–Hirschhorn sindrome is required fornormal mitochondrial morphology and cellular viability. Hum Mol Gen 17:201–217.

7. Chandy KG, Wulff H, Beeton C, Pennington M, Gutman GA, Cahalan MD (2004) K�

channels as targets for specific immunomodulation. Trends Pharmacol Sci 25:280–289.8. Bock J, Szabo I, Jekle A, Gulbins E (2003) Actinomycin D-induced apoptosis involves the

potassium channel Kv1.3. Biochem Biophys Res Com 295:526–531.9. Annis MG, Yethon JA, Leber B, Andrews DW (2004) There is more to life and death than

mitochondria: Bcl-2 proteins at the endoplasmic reticulum. Biochim Biophys Acta1644:115–123.

10. Shoshan-Barmatz V, Israelson A (2005) The voltage-dependent anion channel inendoplasmic/ sarcoplasmic reticulum: Characterization, modulation and possible func-tion. J Membr Biol 204:57–66.

11. Siemen D, Loupatatzis C, Borecky J, Gulbins E, Lang F (1999) Ca2�-activated K channelof the BK-type in the inner mitochondrial membrane of a human glioma cell line.Biochem Biophys Res Commun 257:549–554.

12. Kottgen M, et al. (2005) Trafficking of TRPP2 by PACS proteins represents a novelmechanism of ion channel regulation. EMBO J 24:705–716.

13. Szabo I, et al. (2005) A novel potassium channel in lymphocyte mitochondria. J BiolChem 280:12790–12798.

14. Deutsch C, Chen LQ (1993) Heterologous expression of K� channels in T lymphocytes:Functional consequences for volume regulation. Proc Natl Acad Sci USA 90:10036–10040.

15. Grassme H, et al. (1997) Acidic sphingomyelinase mediates internalization of Neisseriagonorrhoeae into non-phagocytic cells. Cell 91:605–615.

16. Vicente R, et al. (2006) Association of Kv1.5 and Kv1.3 contributes to the majorvoltage-dependent K� channel in macrophages. J Biol Chem 281:37675–37685.

17. Antonsson B, Montessuit S, Lauper S, Eskes R, Martinou JC (2000) Bax oligomerizationis required for channel-forming activity in liposomes and to trigger cytochrome crelease from mitochondria. Biochem J 345:271–278.

18. Aiyar J, et al. (1995) Topology of the pore-region of a K� channel revealed by theNMR-derived structures of scorpion toxins. Neuron 15:1169–1181.

19. De Marchi U, Campello S, Szabo I, Tombola F, Martinou JC, Zoratti M (2004) Bax doesnot directly participate in the Ca2�-induced permeability transition of isolated mito-chondria. J Biol Chem 279:37415–37422.

20. Eskes R, Desagher S, Antonsson B, Martinou JC (2000) Bid induces the oligomerization andinsertion of Bax into the outer mitochondrial membrane. Mol Cell Biol 20:929–935.

21. Wei MC, et al. (2000) tBID, a membrane-targeted death ligand, oligomerizes BAK torelease cytochrome c. Genes Dev 14:2060–2071.

22. Skulachev VP (2006) Bioenergetic aspects of apoptosis, necrosis and mitoptosis. Apo-ptosis 11:473–485.

23. Vennekamp J, et al. (2004) Kv1.3-blocking 5-phenylalkoxypsoralens: A new class ofimmunomodulators. Mol Pharmacol 65:1364–1374.

24. Rauer H, Pennington M, Cahalan M, Chandy KG (1999) Structural conservation of thepores of calcium-activated and voltage-gated potassium channels determined by a seaanemone toxin. J Biol Chem 274:21885–21892.

25. Annis MG, et al. (2005) Bax forms multispanning monomers that oligomerize topermeabilize membranes during apoptosis. EMBO J 24:2096–2103.

26. Lange A, et al. (2006) Toxin-induced conformational changes in a potassium channelrevealed by solid-state NMR. Nature 440:959–962.

27. Liu XS, Kim CN, Yang J, Jemmerson R, Wang X (1996) Induction of apoptotic programin cell free extracts: Requirement for dATP and cytochrome c. Cell 86:147–157.

28. Scorrano L, et al. (2002) A distinct pathway remodels mitochondrial cristae andmobilizes cytochrome c during apoptosis. Dev Cell 2:55–67.

29. Miki T, et al. (1998) Defective insulin secretion and enhanced insulin action in KATPchannel-deficient mice. Proc Natl Acad Sci USA 95:10402–10406.

30. Koni PA, et al. (2003) Compensatory anion currents in Kv1.3 channel-deficient thymo-cytes. J Biol Chem 278:39443–39451.

31. Dahlem YA, Horn TF, Buntinas L, Gonoi T, Wolf G, Siemen D (2004) The humanmitochondrial KATP channel is modulated by calcium and nitric oxide: A patch-clampapproach. Biochim Biophys Acta 1656:46–56.

32. Montessuit S, Mazzei G, Magnenat E, Antonsson B (1999) Expression and purificationof full-length human Bax alpha. Protein Expr Purif 15:202–206.

14866 � www.pnas.org�cgi�doi�10.1073�pnas.0804236105 Szabo et al.

Dow

nloa

ded

by g

uest

on

Janu

ary

30, 2

020

CELL BIOLOGY. For the article ‘‘Mitochondrial potassium channelKv1.3 mediates Bax-induced apoptosis in lymphocytes,’’ byIldiko Szabo, Jurgen Bock, Heike Grassme, Matthias Sodde-mann, Barbara Wilker, Florian Lang, Mario Zoratti, and ErichGulbins, which appeared in issue 39, September 30, 2008, of ProcNatl Acad Sci USA (105:14861–14866; first published September25, 2008; 10.1073�pnas.0804236105), the authors note that dueto a printer’s error, the e-mail address for corresponding authorErich Gulbins appeared incorrectly. The correct address iserich.gulbins@uni-duisburg-essen.de. The online version hasbeen corrected. In addition, in the Abstract, line 2, ‘‘Here, weshow that mouse and human cells that are genetically deficientin either Kv1.3 or transfected with siRNA’’ should instead read:‘‘Here, we show that mouse and human cells either geneticallydeficient in Kv1.3 or transfected with siRNA.’’ Also on page14861, right column, third full paragraph, line 4, ‘‘Fig. 1C’’ shouldappear as ‘‘Fig. S1C.’’ On page 14863, right column, last line,‘‘However, we expected this behavior’’ should instead read:‘‘However, we expected the same behavior.’’ On page 14865, leftcolumn, in line 6, ‘‘Fig. 3D’’ should appear as ‘‘Fig. S8 A and B.’’On the same page, left column, 8 lines from the bottom, ‘‘Fig.5C’’ should appear as ‘‘Fig. 4C.’’ Finally, on page 14866, in thelist of references, the reference number 6 appears twice, and thefirst instance should instead be numbered 5. The correctedreferences appear below.

5. Ardehali H, O’Rourke B (2005) Mitochondrial K(ATP) channels in cell survival and death.J Mol Cell Cardiol 39:7–16.

6. Dimmer KS, et al. (2008) LETM1, deleted in Wolf–Hirschhorn syndrome is required fornormal mitochondrial morphology and cellular viability. Hum Mol Gen 17:201–217.

www.pnas.org�cgi�doi�10.1073�pnas.0809894105

PERSPECTIVE. For the article ‘‘Ecosystem Services Special Feature:An ecosystem services framework to support both practicalconservation and economic development,’’ by Heather Tallis,Peter Kareiva, Michelle Marvier, and Amy Chang, which ap-peared in issue 28, July 15, 2008, of Proc Natl Acad Sci USA(105:9457–9464; first published July 9, 2008; 10.1073�pnas.0705797105), the authors note that on page 9459, rightcolumn, line 7, ‘‘1999’’ should have appeared as ‘‘1998.’’

www.pnas.org�cgi�doi�10.1073�pnas.0809375105

PNAS � November 11, 2008 � vol. 105 � no. 45 � 17587

CORR

ECTI

ON

S