palmitoylation of stathmin family proteins domain a controls golgi versus mitochondrial subcellular...

Biol. Cell (2008) 100, 577–589 (Printed in Great Britain) doi:10.1042/BC20070119 Research article

Palmitoylation of stathmin familyproteins domain A controls Golgiversus mitochondrial subcellulartargetingStephanie Chauvin*†‡, Fabienne E. Poulain*†‡, Sylvie Ozon*†‡ and Andre Sobel*†‡1

*Inserm, U839, 75005 Paris, France, †Universite Pierre et Marie Curie – Paris 6, 75005 Paris, France, and ‡Institut du Fer a Moulin,

75005 Paris, France

Background information. Precise localization of proteins to specialized subcellular domains is fundamental forproper neuronal development and function. The neural microtubule-regulatory phosphoproteins of the stathminfamily are such proteins whose specific functions are controlled by subcellular localization. Whereas stathmin iscytosolic, SCG10, SCLIP and RB3/RB3′/RB3′′ are localized to the Golgi and vesicle-like structures along neuritesand at growth cones. We examined the molecular determinants involved in the regulation of this specific subcellularlocalization in hippocampal neurons in culture.

Results. We show that their conserved N-terminal domain A carrying two palmitoylation sites is dominant over theothers for Golgi and vesicle-like localization. Using palmitoylation-deficient GFP (green fluorescent protein) fusionmutants, we demonstrate that domains A of stathmin proteins have the particular ability to control protein targetingto either Golgi or mitochondria, depending on their palmitoylation. This regulation involves the co-operation of twosubdomains within domain A, and seems also to be under the control of its SLD (stathmin-like domain) extension.

Conclusions. Our results unravel that, in specific biological conditions, palmitoylation of stathmin proteins mightbe able to control their targeting to express their functional activities at appropriate subcellular sites. They, moregenerally, open new perspectives regarding the role of palmitoylation as a signalling mechanism orienting proteinsto their functional subcellular compartments.

IntroductionNeuronal development and plasticity rely on theproper formation of the axon and dendrites, whosefunctional diversity is provided by a correct sortingand delivery of each constituent to its specific subcel-lular domain. Among these constituents, peripheral-membrane-bound proteins use a hierarchy of signalsto be transported to the axon, dendrites or growth

1To whom correspondence should be addressed ([email protected]).Key words: Golgi, mitochondria, neuron, palmitoylation, stathmin family,subcellular targeting.Abbreviations used: GAP-43, growth-associated protein-43; GFP, greenfluorescent protein; HBSS, Hanks balanced salt solution;PAT, palmitoyltransferase; PSD-95, postsynaptic density 95; SLD,stathmin-like domain.

cones. One major signal is protein palmitoylationthat represents a common lipid modification of neur-onal proteins (El-Husseini and Bredt, 2002). Thispost-translational change involves addition of thesaturated 16-carbon palmitate lipid in a thioesterlinkage to specific cysteine residues, with no well-defined surrounding consensus sequence. Palmitoyla-tion is known to take place on numerous neur-onal proteins, including GAP-43 (growth-associatedprotein-43) (Skene and Virag, 1989), PSD-95 (post-synaptic density 95) protein (Topinka and Bredt,1998) or G-protein α-subunits (Linder et al., 1993;Wedegaertner et al., 1993). Numerous studiesdemonstrated that this specific lipidation is involvedin protein trafficking (Smotrys and Linder, 2004): forinstance, dual palmitoylation of PSD-95 is necessary

www.biolcell.org | Volume 100 (10) | Pages 577–589 577

Bio

log

y o

f th

e C

ell

w

ww

.bio

lcel

l.org

S. Chauvin and others

for appropriate postsynaptic localization to dendriticspines (El-Husseini and Bredt, 2002; Kanaani et al.,2002) and dual palmitoylation of GAP-43 directsthe protein to growth cone membranes (Sudo et al.,1992). Importantly, palmitoylation occurs in a revers-ible fashion, allowing the dynamic regulation of pro-tein functions, and therefore participates in diverseaspects of neuronal signalling (Smotrys and Linder,2004).

A good example of proteins whose function ismodulated by subcellular localization is providedby members of the stathmin family, which includesstathmin, SCG10, SCLIP, RB3 and its splice variantsRB3′/RB3′′. This protein family is preferentially ex-pressed in the nervous system (Di Paolo et al., 1997a;Ozon et al., 1997, 1998) and is known to modulatemicrotubule dynamics by sequestering free tubulin(Curmi et al., 1997; Jourdain et al., 1997; Charbautet al., 2001). Similarly to stathmin, the microtubule-destabilizing activity of SCG10 is regulated by mul-tiple phosphorylations (Grenningloh et al., 2004;Togano et al., 2005) that can be induced by neur-onal activity and/or plasticity (Morii et al., 2005).Despite sharing the same microtubule-destabilizingactivities, these proteins do not share the same subcel-lular localization. Stathmin is cytosolic, whereas itsrelated proteins are found at the Golgi and vesicle-like structures, enriched in growth cones (Di Paoloet al., 1997a, 1997b; Lutjens et al., 2000; Gavet et al.,2002) mostly along dynamic microtubules (Poulainand Sobel, 2007). We previously hypothesized thatthe various proteins of the stathmin family fulfil atleast partially distinct and specific biological roles(Ozon et al., 1998). Indeed, in spite of their similarsubcellular localization, SCLIP and SCG10 were re-cently shown to have independent and complement-ary regulatory roles during neuronal differentiation(Poulain and Sobel, 2007).

Membrane anchorage of stathmin-related proteinsis attributed to the presence of an N-terminal exten-sion (Di Paolo et al., 1997b; Lutjens et al., 2000)composed of one to three domains. Domain A, withits two cysteine palmitoylation sites, is well con-served between all stathmin-related proteins, with70% identity for SCLIP and 57% for RB3 whencompared with SCG10. This conserved domain is re-sponsible for the Golgi and vesicle-like localizationof all stathmin-related proteins including RB3. RB3and its splice variants possess two other N-terminal

domains (A′ and A′′) that contribute to the func-tional variety of stathmin family proteins possibly byinfluencing the control of their subcellular localiz-ation. We previously showed that the co-operationof two distinct subdomains within domain A is re-sponsible for the Golgi and vesicular targeting ofstathmin-related proteins (Charbaut et al., 2005).Importantly, the reversible nature of palmitoylationprovides an additional layer of regulation that is prob-ably crucial to control specific and local functions ofstathmin-related proteins either on microtubule dy-namics or in relation to other potential partners (Liuet al., 2002; Greka et al., 2003; Nixon and Casey,2004; Kang et al., 2005; Ng et al., 2006). Elu-cidation of the molecular mechanisms involved inthe subcellular localization of stathmin family pro-teins, particularly understanding the regulation oftheir palmitoylation, will contribute to better un-derstanding of their physiological roles in neurons.To this end, we examined and compared the role ofpalmitoylation of domains A of SCG10, SCLIP andRB3 in their specific subcellular localization. Trans-fection of palmitoylation-deficient mutants in cul-tured hippocampal neurons showed that palmitoyla-tion regulates localization between Golgi and mi-tochondria, with some differences between domainsA of the various stathmin proteins. In addition, wedemonstrated that this targeting regulation involvesthe co-operation of two of the three domain A sub-domains, and is likely further regulated by theirstathmin-like extensions.

ResultsStathmin-related proteins are localized to the Golgiand vesicular structures present in the cell body andalong both dendrites and axons (Gavet et al., 2002).Understanding the molecular mechanisms respons-ible for their subcellular targeting is crucial to un-ravel the functional roles of the various proteins of thefamily. Each member of the family other thanthe soluble stathmin possesses one to three specificN-terminal domains (Figure 1A) that may influ-ence its localization. We showed that only domainA promotes the specific Golgi/vesicle-like targetingof stathmin-related proteins (Charbaut et al., 2005).The other domains do not directly participate in thesubcellular targeting of stathmin family proteins (seeSupplementary Figure S1 at http://www.biolcell.org/

578 C© The Authors Journal compilation C© 2008 Portland Press Ltd

Stathmin family subcellular targeting Research article

Figure 1 Domains and constructs investigated in thepresent study(A) Stathmin family proteins: except for stathmin, each pro-

tein possesses, in addition to its SLD, an N-terminal exten-

sion including the conserved domain A (black), domain A′ for

RB3/RB3′/RB3′ ′ (grey) and domain A′ ′ only for RB3′ ′ (white).

(B) Sequence alignment of domains A from SCG10, SCLIP and

RB3/RB3′/RB3′ ′: conserved cysteine residues are in black.

Each cysteine residue or both cysteine residues were mutated

into alanine to generate three mutants for each domain A:

mut1, mut2 or mut1,2.

boc/100/boc1000577add.htm) but may contribute indiversifying their functional properties, for exampleby interacting with specific partners in neuronalcells.

The conserved N-terminal domains A carry twopalmitoylation sites (Figure 1B) that may contributeto their specific subcellular localization. Importantly,protein palmitoylation being labile and reversible,it can provide a significant regulatory mechanism inresponse to physiological stimuli. Therefore we in-vestigated and compared the role and possible sub-cellular targeting regulation of each N-terminal do-main A through its palmitoylation, using GFP (greenfluorescent protein) fusion proteins overexpressed inhippocampal neurons.

Palmitoylation of domain A regulates Golgi versusmitochondrial subcellular targetingIt has been shown for SCG10 that palmitoyla-tion occurs on two cysteine residues (Cys22 andCys24) within domain A (Di Paolo et al., 1997b)and that, in PC12 cells, their mutation promotedSCG10 solubilization and prevented its transportto growth cones (Lutjens et al., 2000). To eval-uate and compare the role(s) of palmitoylation ofdomains A within the various members of thestathmin family, we chose to examine its contribu-tion to subcellular protein targeting in actual neur-ons in primary culture. In embryonic rat hippocam-pal neurons in culture, transfected ASCG10 exhib-ited, as previously described (Charbaut et al. 2005),a similar subcellular localization as the full-lengthSCG10, with an accumulation at the Golgi, and atvesicle-like structures present along neurites andwithin the growth cone (Figures 2A and 3Cand Supplementary Figure S2 at http://www.biolcell.org/boc/100/boc1000577add.htm). We mutatedCys22, Cys24 or both to alanine (Figure 1B) to pre-vent their palmitoylation and explore the role(s) ofpalmitoylation in the subcellular targeting of ASCG10.This strategy allowed us to visualize a specific re-localization of the mutated fusion proteins withincultured hippocampal neurons. Mutation of Cys22

within domain A of SCG10 (ASCG10-mut1) only par-tially affected the Golgi-like localization (shown bya partial perinuclear co-localization with endogenousSCG10) but abolished the punctate, vesicle-like loc-alization along neurites (Figure 2A). In these struc-tures, this mutant was instead localized mostly to mi-tochondria, as revealed by an extensive co-localizationwith the mitochondrial marker cytochrome c (Fig-ure 2B). Mutation of the second (Cys24, ASCG10-mut2) or both (Cys22,24, ASCG10-mut1,2) cysteineresidues to alanine within ASCG10 (Figures 2A and2B) produced a more dramatic effect on protein relo-calization. Indeed, ASCG10-mut2 and ASCG10-mut1,2completely lost the Golgi localization for a mi-tochondrial one, associated with a weak cytoso-lic distribution both in the cell body and alongneurites.

Altogether, our results following mutation of eachcysteine residue or both cysteine residues within do-main A of SCG10 to affect its level of palmitoylation(Di Paolo et al., 1997b) suggest that, in primaryneurons, regulation of ASCG10 palmitoylation also



Figure 2 Subcellular localization of cysteine mutants of domain A of SCG10 in neuronsDomain A of SCG10 fused to GFP mutated on Cys22 (ASCG10-mut1), Cys24 (ASCG10-mut2) or Cys22,24 (ASCG10-mut1,2) was

transfected in cultured hippocampal neurons. Double immunolabelling with anti-GFP antibody and either anti-SCG10 (A) or

anti-cytochrome c (cyt.C) (B) antibody was performed to visualize either a co-localization with endogenous SCG10, in particular

at the Golgi (A), or with mitochondria (B). Mutation of Cys22 preserved the Golgi localization (co-localization with endogenous

SCG10) but abolished the punctate, vesicle-like staining in the neurites, replaced by a mitochondrial localization (co-localization

with cytochrome c). Mutation of Cys24 or both cysteine residues Cys22 and Cys24 led to a solely mitochondrial localization, as

revealed by the absence of coincidence with endogenous SCG10 (A) and perfect coincidence with cytochrome c (B). For each

merged image, an enlarged view of the cell body (upper square) and of a neurite (lower square) region of the transfected neurons

is shown. Results of confocal microscopy are shown. Scale bars, 10 μm.



controls its subcellular targeting between Golgi, ves-icles and mitochondria.

The palmitoylation motif present within ASCG10

is highly conserved within domains A of SCLIP andRB3 (Figure 1B). Therefore we checked whethermutation of the two corresponding cysteine residueswithin domain A of SCLIP (Cys22, Cys24) or RB3(Cys20, Cys22) would have the same relocalizationeffect on the subcellular localization of the fusionproteins. First, we transfected domain A of SCLIPor RB3 (ASCLIP or ARB3 respectively) fused to GFPin hippocampal neurons and verified, similarly toASCG10, that these fusion proteins reproduced thesubcellular localization of the endogenous proteins.By confocal microscopy, we observed a very similar la-belling of ASCLIP and endogenous SCLIP at the Golgiand at punctate, vesicle-like structures present alongneurites and within the growth cone (Figure 3A andSupplementary Figure S2). Regarding ARB3, we per-formed a co-labelling with the anti-SCG10 antibodysince the endogenous RB3 was not efficiently detec-ted with our antibodies. Like the other domains A ofthe stathmin family, ARB3 was present at the Golgiand on vesicle-like structures along neurites andwithin the growth cone (Figures 3A and 3C and Sup-plementary Figure S2). However, co-localization withthe endogenous SCG10 was not perfect, especiallyon the vesicle-like structures present along neurites,suggesting that RB3 and SCG10 do not have exactlythe same targeting and trafficking along neurites.

Interestingly in some neurons, we observed in somecases that ASCLIP and ARB3 were not only localized atthe Golgi and vesicles (visualized by a co-localizationwith endogenous SCG10), but also to a small ex-tent at mitochondria (shown by a co-localization withcytochrome c labelling) (Figure 3B, ASCLIP as an ex-ample). In order to quantify the degree to whichdomains A of SCG10, SCLIP or RB3 localized GFPto mitochondria, we counted individual transfectedneurons according to three groups of compartmentsdescribing the repartition of the GFP fusion pro-teins between Golgi and vesicle-like structures com-pared with Golgi and mitochondria or mitochondriaalone. As shown in Figure 3(C), most of the cells dis-played, as expected, the fusion proteins in both theGolgi and vesicle-like structures (over 80% ofthe transfected neurons for the three domains). How-ever, similarly to what we have previously observedwith the palmitoylation-deficient mutants of ASCG10,

wild-type domains A also exhibited both Golgi andmitochondrial localization in 5–7% of transfectedhippocampal neurons. A unique mitochondrial loc-alization of each fusion protein was also observed tovarious proportions depending on the domain con-sidered; ARB3 and ASCLIP were present at mitochon-dria in 13 and 10% of the transfected neurons re-spectively, as compared with 5% for ASCG10. Thevarious domains A thus seem to be capable of beingtargeted under some conditions to mitochondria. In-terestingly, in Drosophila, a single gene produces sol-uble stathmin and non-soluble splice variants withan N-terminal extension (Ozon et al., 2002). Thenon-soluble, membrane-bound forms were found es-sentially in the low-speed cytochrome c-containing‘mitochondrial’ fraction following subcellular frac-tionation of Drosophila embryos (Figures 3D and3E). When quantitatively analysed together with aGolgi marker, membrane-bound D-stathmin actu-ally followed the mitochondrial rather than the Golgimarker throughout the various fractions (Figure 3E),which clearly indicates that mitochondria may indeedbe physiological targets of stathmin family proteinsin vivo.

To test the possible role of protein palmitoyla-tion in mitochondrial targeting of the domains A ofSCLIP and RB3, we constructed and analysed theirpalmitoylation-deficient mutants, as for ASCG10 (Fig-ure 1B). We found similar results when we transfec-ted ASCLIP or ARB3 palmitoylation-deficient mutantsin hippocampal neurons; therefore only the resultswith ASCLIP mutants are illustrated here (Figure 4).Similarly to ASCG10-mut1 (Figure 2), mutation ofthe first cysteine residue (ASCLIP-mut1) did not dis-rupt the perinuclear localization (identified as theGolgi complex) observed for the wild-type fusionprotein, but abolished the vesicle-like localizationusually present along neurites. A co-labelling withcytochrome c revealed that ASCLIP-mut1 was in-stead present along neurites at mitochondria. Sim-ilarly to ASCG10 mutants, this mitochondrial loc-alization became more exclusive when we mutatedthe second (ASCLIP-mut2) or both cysteine residues(ASCLIP-mut1,2), since these two mutants were local-ized only to mitochondria both in the cell body andalong neurites, as shown by the co-localization withcytochrome c.

Since a high conservation of the SCG10 palmitoyla-tion motif is observed between all stathmin-related



Figure 3 Comparison of subcellular localization of domains A of SCG10, SCLIP and RB3 with endogenous neuronalstathmin proteins(A) Domains A of SCLIP (ASCLIP) or RB3 (ARB3) were fused to GFP and transfected in cultured hippocampal neurons. Double

labelling with an anti-GFP and an anti-endogenous SCLIP or SCG10 revealed that the localization of the fusion proteins reflects

that of the corresponding endogenous ones, but also some local differences between RB3 and SCG10 vesicle-like staining.

(B) Double labelling of transfected domain A of SCLIP (ASCLIP) fused to GFP with anti-SCG10 or anti-cytochrome c (cyt.C)

(mitochondria). For each merged image, an enlarged view of the cell body (upper square) and of a neurite (lower square) region of

the transfected neurons is shown. Confocal microscopy results are shown. Scale bars, 10 μm. (C) For each domain A–GFP fusion

construct, quantification of transfected neurons shows either Golgi/vesicle-like structures (light grey), Golgi and mitochondria

(dark grey) or only mitochondrial localization (black). More than 80% of transfected neurons showed a Golgi and vesicle-like

localization. A significant number of transfected neurons (5–13% of transfected neurons depending of the construct) also

depicted a unique mitochondrial localization. (D) A Drosophila embryo extract (H) was fractionated by differential centrifugation:

low-speed supernatant (S1) and pellet (P1, nuclear fraction ‘N’); 5000 g supernatant (S2) and pellet (P2, mitochondrial fraction

‘Mi’); and 400 000 g supernatant (S3, cytosolic fraction ‘cyt’) and pellet (P3, membranous fraction ‘Mb’). Equivalent amounts

of each fraction were analysed by Western blotting with an anti-D-stathmin detecting all endogenous forms of stathmins in

Drosophila (Ozon et al., 2002), and an anti-cytochrome c used as a mitochondrial marker. Most of the non-soluble forms of

D-stathmin co-fractioned with the mitochondrial marker, cytochrome c. (E) Quantitative analysis of the distributions of D-stathmin,

cytochrome c and a Golgi marker following a similar fractionation to that in (D) shows that D-stathmin follows essentially the

mitochondrial (cytochrome c) rather than the Golgi marker.

proteins and mutation of these cysteine residues inall domains A relocalizes the fusion proteins to mito-chondria, regulation of protein palmitoylation might,

at least under some conditions (see the Discussion sec-tion), modulate the subcellular targeting of stathmin-related proteins between the Golgi and mitochondria.



Figure 4 Subcellular localization of cysteine mutants of domain A of SCLIP in neuronsDomain A of SCLIP fused to GFP was mutated on Cys22 (ASCLIP-mut1), Cys24 (ASCLIP-mut2) or both Cys22 and Cys24

(ASCLIP-mut1,2) and transfected in cultured hippocampal neurons. Double immunolabelling with anti-GFP and anti-cytochrome

c (cyt.C) antibodies was performed to visualize a potential localization to the mitochondria. As has already been observed for

SCG10 mutants (Figure 2), mutation of Cys22 preserved the Golgi localization (corresponding to the perinuclear staining) but

abolished the punctate, vesicle-like staining in neurites, replaced by a mitochondrial localization (co-localization with cytochrome

c). Mutation of Cys24 or both cysteine residues Cys22 and Cys24 promoted a solely mitochondrial localization as revealed by

perfect coincidence with cytochrome c staining. For each merged image, an enlarged view of the cell body of the transfected

neurons is shown. Confocal microscopy results are shown. Scale bar, 10 μm.

For full-length stathmin proteins, transfection of apalmitoylation-deficient RB3 mutant, or treatmentof hippocampal neurons with 2-bromopalmitate forpreventing palmitoylation, resulted in their cytosolicdistribution, as described for SCG10 in PC12 cells(Di Paolo et al., 1997b). Surprisingly, it did not res-ult in their mitochondrial localization as observedwith the A–GFP fusion protein, either mutated orafter treatment with 2-bromopalmitate (results notshown). These observations might reflect a negativecontrol of the SLD (stathmin-like domain) (Figure 1)over the mitochondrial targeting signal present inthe non-palmitoylated N-terminal domain, whereasin Drosophila the mitochondrial targeting would beconstitutive.

Two motifs within domain A are responsible forthe mitochondrial localizationSince the capacity to shuttle GFP to mitochon-dria is conserved between all domains A when non-

palmitoylated, we looked for the smallest conservedregion within domain A that would drive GFPto mitochondria. As described by Charbaut et al.(2005) and shown in Figure 5, domain A can be di-vided into three subdomains: a perfectly conservedGolgi-specifying subdomain an, a highly conservedpalmitoylated subdomain am and a poorly conservedsubdomain ac. We have previously shown that an andam co-operate in promoting the specific Golgi local-ization of stathmin family proteins (Charbaut et al.,2005).

Therefore we designed and overexpressed in neur-ons different constructs encoding mutated forms(C22A/C24A, mut1,2) of ASCG10 subdomains fusedto GFP. A double labelling using anti-cytochromec or anti-SCG10 antibodies to visualize mitochon-dria or Golgi respectively allowed us to concludeabout the subcellular localization of the fusion pro-teins. Our observations are summarized in Figure 5:whereas wild-type am was present at all membranes



(already described by Charbaut et al., 2005), thepalmitoylation-deficient am mutant became soluble,suggesting that the mitochondrial targeting signalis composed of the association of at least two sub-domains (am and an and/or ac). Therefore we pro-duced palmitoylation-deficient mutants of anm andamc of ASCG10 fused to GFP. As expected (Charbautet al., 2005), wild-type amc was present at all cellmembranes, whereas anm was restricted to the Golgiand vesicle-like structures. Mutation of both cysteineresidues within anm–GFP did not relocalize the fu-sion protein to mitochondria but rather promoted itssolubilization (Figure 5B). A mitochondrial localiz-ation only appeared when we mutated both cysteineresidues within amc (Figure 5B). However, ac alonewas present in the cytosol as well as an. These res-ults suggest that the mitochondrial targeting signalrequires both am, when non-palmitoylated, and ac

subdomains.Altogether, our results reveal that within domain

A two subcellular targeting signals coexist, one sig-nal becoming dominant over the other dependingon the level of protein palmitoylation. When am ispalmitoylated, its association with an drives GFP tothe Golgi and vesicle-like structures, whereas its non-palmitoylated form directs the protein, in associationwith ac, to the mitochondria.

DiscussionPrecise localization of proteins to specialized sub-cellular domains is fundamental for proper neuronaldevelopment and function. Among stathmin familyproteins, several stathmin-related proteins display aspecific Golgi and vesicle-like subcellular targeting,including vesicles accumulated in neuronal growthcones. In the present study, we investigated the poten-tial role of each N-terminal domain of the stathmin-related proteins and examined the importance of theirN-terminal palmitoylation as a regulator of this spe-cific localization.

Membrane anchorage of stathmin-related proteinsis attributed to the presence of an N-terminal exten-sion (Di Paolo et al., 1997b; Lutjens et al., 2000)composed of one to three domains. Our results showthat the conserved and common domain A is domin-ant over the others for the specific Golgi and vesicularprotein targeting. It carries two conserved cysteinesites of palmitoylation, a post-translational modi-

fication known to play a central role in regulatingthe functions of diverse neuronal proteins as high-lighted by several corresponding neuronal dysfunc-tions (Huang and El-Husseini, 2005). Here, we in-vestigated and compared the role of the two conservedcysteine residues in subcellular targeting of SCG10,SCLIP and RB3. Domain A can be subdividedinto a well-conserved subdomain am (containing twocysteine residues), a perfectly conserved subdomainan and a poorly conserved subdomain ac (Charbautet al., 2005). We investigated the potential role(s) ofpalmitoylation of the targeting domains of the variousstathmin family proteins directly in primary neur-ons by transfecting palmitoylation-deficient mutantsgenerated from domains A of SCG10, SCLIP or RB3in fusion with GFP. Wild-type constructs reflec-ted the endogenous subcellular localization with aGolgi and a vesicle-like localization present in thecell body, along neurites and within the growthcones (Di Paolo et al., 1997a; Lutjens et al., 2000;Gavet et al., 2002). Interestingly, small and variableproportions of ASCG10, ASCLIP and ARB3 fusion pro-teins were able to localize also to the mitochon-dria, which remarkably appears to be the ma-jor subcellular target of non-soluble stathmins inDrosophila.

The role of palmitoylation in this differential loc-alization is validated by the use of palmitoylation-deficient mutants. Indeed, mutation of the firstcysteine residue of ASCG10, ASCLIP or ARB3 preservedthe Golgi localization but abolished the vesicle-likestaining. Instead, we observed a mitochondrial la-belling along neurites. Mutation of the second or bothcysteine residues accentuated this effect by inducing acomplete (for SCLIP and RB3) or a partial mitochon-drial localization (for the SCG10 construct, numer-ous neurons showed also a cytosolic staining) in thecell body and within neurites. The dramatic effect ob-served by mutating the second cysteine residue is con-sistent with previous data revealing that this residuewas the major site of palmitoylation for SCG10(Di Paolo et al., 1997b).

We previously showed that the co-operation of sub-domains am and an is crucial to specific localizationof stathmin family proteins to Golgi membranes(Charbaut et al., 2005). Our present results arein favour of the membrane-trapping model, whichstates that proteins cycle on and off membranes un-til they encounter a membrane with an appropriate



Figure 5 Subcellular localization of GFP fusion proteins containing wild-type or palmitoylation-deficient mutants ofSCG10 domain A or its derived subdomains(A) Sequence alignment of domains A from SCG10, SCLIP and RB3/RB3′/RB3′ ′: conserved residues are highlighted in grey, and

the two palmitoylated cysteine residues in black. The delineation of the three regions, ‘n’ (dark grey), ‘m’ (white) and ‘c’ (black),

was driven by sequence conservation pattern (Charbaut et al., 2005). (B, C) Wild-type or Cys22 and Cys24 mutants of domain A

of SCG10 or derived subdomains fused to GFP were overexpressed in neurons. Subcellular localization of each construct was

determined by double labelling using anti-GFP and either anti-SCG10 (to stain the Golgi and vesicles) or anti-cytochrome c (cyt.C)

antibody (to stain mitochondria). (B) A summary of the results from confocal microscopy observations with each construct, some

of which are illustrated in (C): double immunolabelling with anti-GFP and anti-cytochrome c antibodies of neurons expressing

C22A/C24A (mut1,2) mutants of domain A and subdomain mc, showing their partial mitochondrial targeting, with an enlarged

view of the cell body region; and GFP immunolabelling for the double-mutants of nm and m fused to GFP, showing their fully

cytosolic localization. Confocal microscopy results are shown. Scale bar, 10 μm.

membrane-targeting receptor (Smotrys and Linder,2004). We can imagine that an, when associated witham, interacts with a specific receptor present at theGolgi that might actually be a PAT (palmitoyltrans-ferase). On the other hand, lack of palmitoylation oncysteine residue(s) within domain A prevents its usualtargeting, resulting in the conversion of vesicle-likeand Golgi localization into mitochondrial localiza-tion.

This mitochondrial localization depends on thepresence of subdomain ac in combination with am,in contrast with the requirement of an for Golgi andvesicular targeting of the palmitoylated proteins.When proteins are palmitoylated (especially on Cys24

for SCG10), an and am co-operate in shuttling theprotein to the Golgi and vesicles, whereas when non-palmitoylated, am does not co-operate with an butwith ac in directing the protein to the mitochondria.



This is of particular interest within the stathminfamily, since subdomain ac is poorly conservedand might thus be responsible for the differencesbetween the three domains A for the propensity ofpalmitoylation-deficient mutants and for the wild-type fusion proteins to be targeted to mitochondria.Nonetheless, two well-conserved tyrosine residuesare present within all subdomains ac that mightbe involved also in mitochondrial localization. Onecan speculate, for example, that the absence ofpalmitoylation could allow phosphorylation of thesetwo residues and then promote an interaction witha specific partner that would trigger the protein tomitochondria. The observed differences might be ofbiological relevance to specify the individual rolesand actions of the various proteins of the stathminfamily.

Interestingly, in Drosophila embryos, mitochondriarepresent the major subcellular targeting compart-ment for the endogenous non-soluble stathmin forms,which gives weight to the fact that mitochondria arelikely physiological targets of stathmins also in ver-tebrates under some conditions. The difference withvertebrates might result from an evolution of mem-brane targeting of stathmins from invertebrates tovertebrates, the latter having possibly acquired theability to be anchored at other membranes such asGolgi and vesicle-like structures. The lack of mito-chondrial relocalization of the full-length stathminproteins upon mutation of their palmitoylation sitesor inhibition of palmitoylation by 2-bromopalmitatesuggests then the existence of a negative control ofthe SLD (Figure 1) over the mitochondrial targetingsignal present in the non-palmitoylated N-terminaldomain. In fact, it is the replacement of the SLDs ofstathmin proteins by GFP that allowed us to revealthe differential and potentially regulated protein-targeting properties of the various domains A. Itis likely that biological signals control this inhibi-tion under specific physiological conditions that re-main to be identified. Interaction with specific part-ners, possibly other than tubulin, and/or structuralmodifications of the SLDs might control the activityof targeting signals present in the N-terminal do-main. Identifying the biological conditions and regu-lations potentially allowing the palmitoylation/non-palmitoylation-mediated relocalization of stathminproteins will be of great interest. Indeed, under spe-cific stimuli, this SLD negative control could be re-

leased, allowing mitochondrial localization of thefull-length protein in its non-palmitoylated form.The stimuli that control SLD inhibition might be thesame as or related to the ones that control PATactivity. Therefore the identification of the PAT(s)responsible for stathmin family protein palmitoyla-tion will help us to understand their functional reg-ulations.

The diversity of proteins modified by palmitatecombined with the absence of common consensus se-quences for palmitoylation makes precise predictionsvery difficult. Besides, it is to date unknown whetherthese PATs are functionally redundant or whe-ther they are differentially specialized in thepalmitoylation of a subset of proteins. However,among the 23 mammalian cloned PATs (containing aconserved DHHC motif) (Fukata et al., 2004), many,like HIP14 (huntingtin-interacting protein 14; alsocalled DHHC-17) (Huang et al., 2004) or GODZ(Golgi apparatus-specific protein with DHHC zincfinger domain; also called DHHC-3), have been de-tected in the brain and described at the Golgi com-plex, and can thus be potential PATs for stathminfamily proteins (Ohno et al., 2006). Therefore itwould be of great interest to determine whetherstathmin family proteins are substrates of one or moreof these PATs.

Our results indicate that, probably in responseto specific regulatory signals, stathmin-related pro-teins might be switched from a Golgi/vesicle sub-cellular localization to a mitochondrial localizationwhen they are non-palmitoylated. This is of par-ticular interest in view of the known tubulin in-teraction and microtubule-regulatory properties ofstathmin proteins. Indeed, like other organelles suchas the Golgi and derived vesicles, mitochondria aredelivered along microtubules to their functional loc-ation, for example areas where metabolic demand ishigh, such as synapses (Peters et al., 1991) or activegrowth cone and branches (Ruthel and Hollenbeck,2003). It is tempting to propose that specific localiz-ation of the microtubule-regulating stathmin familyproteins to mitochondria might contribute to theregulation of mitochondrial motility and then parti-cipate in neuronal functionality.

Altogether, our results open new perspectives re-garding the mechanisms by which stathmin fam-ily proteins are targeted to their functional com-partments and, more generally, the role of protein



palmitoylation in regulating subcellular targetingbetween the Golgi and mitochondria. Elucidationof the complex mechanisms by which stathmin fam-ily proteins are regulated will help us to understandtheir physiological roles and activities in neurons.

Materials and methodsPlasmid constructsStandard recombinant DNA techniques were carried out as de-scribed by Sambrook et al. (1989). For fusion proteins ASCG10–GFP, ASCLIP–GFP and ARB3–GFP, a DNA fragment correspond-ing to each domain A was amplified by PCR, using a plasmid en-coding rat SCG10, mouse SCLIP and rat RB3 respectively (Gavetet al., 1998) as a template. For A–A′–GFP and A–A′′–A′–GFPfusion proteins, a DNA fragment corresponding to domains A–A′ and A–A′′–A′ was amplified by PCR, using a plasmid encod-ing rat RB3 and RB3′′ respectively (Gavet et al., 1998). For theconstructs encoding either domain A′ or A′′ fused to GFP, com-plementary 5′-phosphorylated oligonucleotides (synthesized byGenset, Paris, France) were designed to form a double-strandedDNA fragment containing the whole sequence of the insert, withKpnI and BamHI cohesive ends at 5′- and 3′-ends respectively.The annealing of complementary oligonucleotides and ligationin the plasmid pEGFP-N1 (ClonTech, Palo Alto, CA, U.S.A.)were performed as previously described (Charbaut et al., 2005).For all constructs, the ligation reactions were used to transformEpicurian Coli® XL1-Blue supercompetent cells (Stratagene, LaJolla, CA, U.S.A.). Kanamycin-resistant clones were cultivatedin liquid Luria–Bertani medium containing 25 μg/ml kanamy-cin, and plasmid DNA was prepared using the Nucleobond kit(Macherey-Nagel, Duren, Germany). The fusion protein openreading frames were checked by DNA sequencing (Genome Ex-press, Meylan, France).

Palmitoylation-deficient mutants were generated by site-directed mutagenesis using the QuikChange® mutagenesis kit(Stratagene) with pairs of complementary mutagenic primers.All constructs were confirmed by DNA sequencing.

When the sequence was correct, large preparations of plas-mid DNA were carried out using the QiaTip-550 kit (Qiagen,Hilden, Germany) or the Nucleobond AX2000 kit (Macherey-Nagel), and their concentrations were assessed by spectrometryat 260 nm.

Primary hippocampal neuron cultureEmbryos [E18 (embryonic day 18)] from a pregnant female rat(OFA, Charles River, France) were dissected in HBSS (Hanksbalanced salt solution)/Hepes medium (1× HBSS and 20 mMHepes) to isolate hippocampi. Hippocampi dissection and cellsdissociation were performed as previously described (Poulainand Sobel, 2007). Cells (6 × 104 per well) were then seededon 14 mm glass coverslips coated with 3 μg/ml poly-ornithine(Sigma) in an appropriate volume of Neurobasal medium (Invit-rogen Life Technologies, Carlsbad, CA, U.S.A.) supplementedwith 4% supplement B27 (Invitrogen Life Technologies) and0.5 mM L-glutamine. Cells were cultured at 37◦C in a humidi-fied atmosphere containing 5% CO2.

HeLa cellsHeLa cells were grown in DMEM (Dulbecco’s modified Eagle’smedium)–Glutamax with 10% decomplemented fetal calf serumand 1% penicillin/streptomycin (Invitrogen Life Technologies),at 37◦C in a humidified atmosphere containing 5% CO2. Twice aweek, cells were dissociated in 0.25% trypsin and 1 mM EDTA(Invitrogen Life Technologies), and then diluted 1:10 beforereplating.

TransfectionNeurons were transfected on d.i.v. (day in vitro) 4 by usingLipofectamineTM 2000 according to the manufacturer’s instruc-tions (Invitrogen Life Technologies). Briefly, for each 14 mmglass coverslip, 0.5 μl of LipofectamineTM 2000 was diluted in50 μl of OptiMEM (Invitrogen Life Technologies) and, after5 min of incubation (22◦C), added to 50 μl of OptiMEM con-taining 0.5 μg of DNA. The DNA–lipid complex was allowedto form for 20 min at room temperature (22◦C), and the 100 μlmixture was then added to the culture.

HeLa cells were seeded at 12500 cells/cm2 in 35 mm dishescontaining three 12 mm glass coverslips. They were transfected16 h later using the liposoluble reagent FuGENETM 6 (RocheDiagnostics), according to the manufacturer’s instructions: 3 μlof reagent and 1 μg of plasmid DNA were used to transfect onedish.

Immunocytochemistry and fluorescence microscopyHeLa cells were fixed 24 h after transfection with 2% (w/v)paraformaldehyde and 30 mM sucrose in PBS for 15 min atroom temperature. Primary hippocampal neurons were fixed for48 h after transfection with 4% paraformaldehyde prewarmed at37◦C in PBS for 10 min. After fixation, coverslips were washedthree times in PBS before being processed for immunochemistryor directly mounted in a Mowiol solution supplemented withthe antifading agent, DABCO (1,4-diazadicyclo[2.2.2]octane).

Immunocytochemistry was performed as follows: cells werepermeabilized with 0.1% Triton X-100 in PBS for 5 min, andthen blocked for 1 h with 3% (w/v) BSA in PBS. The coverslipswere next incubated for 1 h with the appropriate antibodiesdiluted in blocking buffer: either the monoclonal antibodyCTR433 (a marker of the median Golgi; a gift from Dr M.Bornens, Institut Curie, Paris, France) diluted 1:2), or themonoclonal anti-cytochrome c (1:400; a marker of mitochon-dria; BD Biosciences PharMingen), or a purified polyclonal an-tiserum raised against SCG10 (1:400) or SCLIP (1:400) (Ozonet al., 1998; Poulain and Sobel, 2007), or a monoclonal anti-body against GFP (1:500; Roche Diagnostics). After five washeswith 0.1% Tween 20 in PBS, the coverslips were incubatedfor 1 h with an Alexa Fluor® 488-conjugated anti-mouse IgGantibody and an Alexa Fluor® 546-conjugated anti-rabbitIgG antibody (Molecular Probes, Eugene, OR, U.S.A.) bothdiluted 1:1000 in blocking buffer. The coverslips were finallywashed five times with 0.1% Tween 20 in PBS and mounted asdescribed above.

Fluorescence microscopy was performed in the Institut duFer a Moulin Cell Imaging facility either with a Provis fluores-cence photomicroscope (Olympus, Tokyo, Japan) equipped witha digital camera (Princeton Scientific Instruments, MonmouthJunction, NJ, U.S.A.), or with an SP2 confocal microscope(Leica Microsystems, Wetzlar, Germany). Confocal images inFigures 1–5 are single confocal sections from high-resolution



stacks. For quantification of subcellular localization of variousA–GFP constructs, a series of random, single confocal sectionsfrom high-resolution stacks were analysed. Double labellingwith either the mitochondrial marker cytochrome c (mitochon-drial compartment group) or endogenous SCG10, known to betargeted to the Golgi and vesicle-like structures (Di Paolo et al.,1997, Lutjens et al., 2000; Gavet et al., 2002) (Golgi+vesiclescompartment group), allowed us to count and classify the num-ber of cells in each category manually. A total of 100 transfectedneurons were counted for each construct.

Drosophila embryo extractsDechorionated 0–24 h embryos were homogenized in a Douncehomogenizer in 20 mM Hepes and 250 mM sucrose (pH 7.3)with an antiprotease cocktail (CompleteTM protease inhibitorcocktail; Roche Diagnostics) and fractionated by differentialcentrifugation. Briefly, S1/P1, S2/P2 and S3/P3 are, respectively,supernatant and pellet of a 200 g centrifugation for 5 min of theinitial homogenate (H), of a 5000 g centrifugation for 10 min ofS1 and of a 100000 rev./min ultracentrifugation (TL100 rotor,Beckman) for 6 min of S2 (Figure 3). Each pellet was washedand centrifuged again.

All forms of stathmin proteins were analysed by Westernblotting with a polyclonal antibody diluted at 1:10 000 (Ozonet al., 2002). Mitochondria and Golgi membranes were detectedusing anti-cytochrome c (1:1000; 7H8.2C12; BD BiosciencesPharMingen) and ‘anti-Golgi’ (1:500; 7H6D7C2; Calbiochem)monoclonal antibodies respectively. Primary antibodies were de-tected using IRDye 800- or Alexa Fluor® 688-conjugated sec-ondary antibodies, which were visualized using the Odyssey Ima-ging System (LI-COR Biosciences). The respective percentagesof non-soluble D-stathmin, cytochrome c and the Golgi markerin P1 (N, nucleus), P2 (Mi, mitochondria), P3 (Mb, membranes)and S3 (Cyt, cytosol) fractions were quantified using the Odysseysoftware (Figure 3).

AcknowledgmentsWe thank Dr E. Charbaut (Merck-Serono, Corsier-sur-Vevey, Switzerland) for her initial contributionto this study, and Dr A. Guichet (Institut JacquesMonod, Paris, France) and M. Lebois for their con-tribution to the Drosophila fractionation experiment.Fluorescence microscopy and confocal microscopywere performed in the Institut du Fer a Moulin Ima-ging facility. This work was supported by the Inserm,UPMC, ARC (Association pour la Recherche contrele Cancer), AFM (Association Francaise contre la My-opathie) and ARSEP (Association pour la Recherchesur la Sclerose en Plaques). F.E.P. is supported by agraduate fellowship of the FRM (Fondation pour laRecherche Medicale).

ReferencesCharbaut, E., Curmi, P.A., Ozon, S., Lachkar, S., Redeker, V. and

Sobel, A. (2001) Stathmin family proteins display specific molecularand tubulin binding properties. J. Biol. Chem. 276, 16146–16154

Charbaut, E., Chauvin, S., Enslen, H., Zamaroczy, S. and Sobel, A.(2005) Two separate motifs cooperate to target stathmin-relatedproteins to the Golgi complex. J. Cell Sci. 118, 2313–2323

Curmi, P.A., Andersen, S.S., Lachkar, S., Gavet, O., Karsenti, E.,Knossow, M. and Sobel, A. (1997) The stathmin/tubulin interactionin vitro. J. Biol. Chem. 272, 25029–25036

Di Paolo, G., Lutjens, R., Osen-Sand, A., Sobel, A., Catsicas, S. andGrenningloh, G. (1997a) Differential distribution of stathmin andSCG10 in developing neurons in culture. J. Neurosci. Res. 50,1000–1009

Di Paolo, G., Lutjens, R., Pellier, V., Stimpson, S.A., Beuchat, M.H.,Catsicas, S. and Grenningloh, G. (1997b) Targeting of SCG10 tothe area of the Golgi complex is mediated by its NH2-terminalregion. J. Biol. Chem. 272, 5175–5182

El-Husseini, A.D. and Bredt, D.S. (2002) Protein palmitoylation:a regulator of neuronal development and function.Nat. Rev. Neurosci. 3, 791–802

Fukata, M., Fukata, Y., Adesnik, H., Nicoll, R.A. and Bredt, D.S.(2004) Identification of PSD-95 palmitoylating enzymes. Neuron44, 987–996

Gavet, O., Ozon, S., Manceau, V., Lawler, S., Curmi, P. and Sobel, A.(1998) The stathmin phosphoprotein family: intracellularlocalization and effects on the microtubule network. J. Cell Sci.111, 3333–3346

Gavet, O., El Messari, S., Ozon, S. and Sobel, A. (2002) Regulationand subcellular localization of the microtubule-destabilizingstathmin family phosphoproteins in cortical neurons. J. Neurosci.Res. 68, 535–550

Greka, A., Navarro, B., Oancea, E., Duggan, A. and Clapham, D.E.(2003) TRPC5 is a regulator of hippocampal neurite length andgrowth cone morphology. Nat. Neurosci. 6, 837–845

Grenningloh, G., Soehrman, S., Bondallaz, P., Ruchti, E. and Cadas,H. (2004) Role of the microtubule destabilizing proteins SCG10 andstathmin in neuronal growth. J. Neurobiol. 58, 60–69

Huang, K. and El-Husseini, A. (2005) Modulation of neuronal proteintrafficking and function by palmitoylation. Curr. Opin. Neurobiol.15, 527–535

Huang, K., Yanai, A., Kang, R., Arstikaitis, P., Singaraja, R.R., Metzler,M., Mullard, A., Haigh, B., Gauthier-Campbell, C., Gutekunst, C.A.et al. (2004) Huntingtin-interacting protein HIP14 is a palmitoyltransferase involved in palmitoylation and trafficking of multipleneuronal proteins. Neuron 44, 977–986

Jourdain, L., Curmi, P., Sobel, A., Pantaloni, D. and Carlier, M.F.(1997) Stathmin: a tubulin-sequestering protein which forms aternary T2S complex with two tubulin molecules. Biochemistry 36,10817–10821

Kanaani, J., El-Husseini, A.D., Aguilera-Moreno, A., Diacovo, J.M.,Bredt, D.S. and Baekkeskov, S. (2002) A combination of threedistinct trafficking signals mediates axonal targeting andpresynaptic clustering of GAD65. J. Cell Biol. 158, 1229–1238

Kang, S.W., Shin, Y.J., Shim, Y.J., Jeong, S.Y., Park, I.S. and Min,B.H. (2005) Clusterin interacts with SCLIP (SCG10-like protein) andpromotes neurite outgrowth of PC12 cells. Exp. Cell Res. 309,305–315

Linder, M.E., Middleton, P., Hepler, J.R., Taussig, R., Gilman, A.G.and Mumby, S.M. (1993) Lipid modifications of G proteins: alphasubunits are palmitoylated. Proc. Natl. Acad. Sci. U.S.A. 90,3675–3679

Liu, Z., Chatterjee, T.K. and Fisher, R.A. (2002) RGS6 interacts withSCG10 and promotes neuronal differentiation. Role of the Ggamma subunit-like (GGL) domain of RGS6. J. Biol. Chem. 277,37832–37839

Lutjens, R., Igarashi, M., Pellier, V., Blasey, H., Di Paolo, G., Ruchti,E., Pfulg, C., Staple, J.K., Catsicas, S. and Grenningloh, G. (2000)Localization and targeting of SCG10 to the trans-Golgi apparatusand growth cone vesicles. Eur. J. Neurosci. 12, 2224–2234



Morii, H., Yamada, T., Nakano, I., Coulson, J.M. and Mori, N. (2005)Site-specific phosphorylation of SCG10 in neuronal plasticity:role of Ser73 phosphorylation by N-methyl D-aspartic acidreceptor activation in rat hippocampus. Neurosci. Lett. 396,241–246

Ng, D.C., Lin, B.H., Lim, C.P., Huang, G., Zhang, T., Poli, V. andCao, X. (2006) Stat3 regulates microtubules by antagonizing thedepolymerization activity of stathmin. J. Cell Biol. 172,245–257

Nixon, A.B. and Casey, P.J. (2004) Analysis of the regulation ofmicrotubule dynamics by interaction of RGSZ1 (RGS20) with theneuronal stathmin, SCG10. Methods Enzymol. 390, 53–64

Ohno, Y., Kihara, A., Sano, T. and Igarashi, Y. (2006) Intracellularlocalization and tissue-specific distribution of human and yeastDHHC cysteine-rich domain-containing proteins.Biochim. Biophys. Acta 1761, 474–483

Ozon, S., Maucuer, A. and Sobel, A. (1997) The stathmin family –molecular and biological characterization of novel mammalianproteins expressed in the nervous system. Eur. J. Biochem. 248,794–806

Ozon, S., Byk, T. and Sobel, A. (1998) SCLIP: a novel SCG10-likeprotein of the stathmin family expressed in the nervous system.J. Neurochem. 70, 2386–2396

Ozon, S., Guichet, A., Gavet, O., Roth, S. and Sobel, A. (2002)Drosophila stathmin: a microtubule-destabilizing factor involved innervous system formation. Mol. Biol. Cell 13, 698–710

Peters, A., Palay, S. and Websters, H. (1991) The Fine Structure ofthe Nervous System: The Neurons and Supporting Cells, OxfordUniversity Press, New York

Poulain, F.E. and Sobel, A. (2007) The ‘SCG10-LIke Protein’ SCLIP isa novel regulator of axonal branching in hippocampal neurons,unlike SCG10. Mol. Cell Neurosci. 34, 137–146

Ruthel, G. and Hollenbeck, P.J. (2003) Response of mitochondrialtraffic to axon determination and differential branch growth.J. Neurosci. 23, 8618–8624

Sambrook, J., Fritsch, E.F. and Maniatis, T. (1989) Molecular Cloning:A Laboratory Manual, Cold Spring Harbor Laboratory Press,Plainview, NY

Skene, J.H. and Virag, I. (1989) Posttranslational membraneattachment and dynamic fatty acylation of a neuronal growth coneprotein, GAP-43. J. Cell Biol. 108, 613–624

Smotrys, J.E. and Linder, M.E. (2004) Palmitoylation of intracellularsignaling proteins: regulation and function. Annu. Rev. Biochem.73, 559–587

Sudo, Y., Valenzuela, D., Beck-Sickinger, A.G., Fishman, M.C. andStrittmatter, S.M. (1992) Palmitoylation alters protein activity:blockade of G(o) stimulation by GAP-43. EMBO J. 11,2095–2102

Togano, T., Kurachi, M., Watanabe, M., Grenningloh, G. and Igarashi,M. (2005) Role of Ser50 phosphorylation in SCG10 regulation ofmicrotubule depolymerization. J. Neurosci. Res. 80, 475–480

Topinka, J.R. and Bredt, D.S. (1998) N-terminal palmitoylation ofPSD-95 regulates association with cell membranes and interactionwith K+ channel Kv1.4. Neuron 20, 125–134

Wedegaertner, P.B., Chu, D.H., Wilson, P.T., Levis, M.J. and Bourne,H.R. (1993) Palmitoylation is required for signalling functions andmembrane attachment of Gq alpha and Gs alpha. J. Biol. Chem.268, 25001–25008

Received 6 September 2007/14 April 2008; accepted 21 April 2008

Published as Immediate Publication 21 April 2008, doi:10.1042/BC20070119


Biol. Cell (2008) 100, 577–589 (Printed in Great Britain) doi:10.1042/BC20070119

Supplementary online data

Palmitoylation of stathmin family proteins domain A controlsGolgi versus mitochondrial subcellular targeting

Stephanie Chauvin*†‡, Fabienne E. Poulain*†‡, Sylvie Ozon*†‡ and Andre Sobel*†‡1

*Inserm, U839, 75005 Paris, France, †Universite Pierre et Marie Curie - Paris 6, 75005 Paris, France, and ‡Institut du Fer a Moulin,

75005 Paris, France

N-terminal domain A has a dominanteffect over domains A′ and A′′ for theGolgi specific localization ofRB3/RB3′/RB3′′Membrane association of stathmin-related proteins isachieved by their N-terminal extension composed ofone or several specific domains (Figure 1A). Unlikedomain A that is known to be responsible for theGolgi- and vesicle-like localization, the role of do-main A′, specific to RB3/RB3′/RB3′′, and domainA′′, restricted to RB3′′, in subcellular localizationis still unknown. To assess whether domains A′ andA′′ confer different subcellular localization inform-ations, we transfected HeLa cells with domain A′or A′′ fused to GFP and inspected their distribu-tion by fluorescence microscopy. Figure S1(A) showsthat domain A′–GFP was entirely localized to thecytosol and the nucleus, similar to GFP alone. Do-main A′′–GFP also partially followed the distribu-tion of soluble GFP; however, a fraction of the fusionprotein was also detected at the plasma membrane(arrows), but not at the Golgi and vesicles, as observedfor ARB3-GFP. This plasma membrane distributioncould be the result of the presence in recombinantdomain A′′ of an N-terminal glycine residue thatmight initiate myristoylation. To rule out this pos-sibility, we mutated this residue to alanine and stillobserved a similar, partial plasma membrane local-ization (not shown). To investigate how domains A′

1 To whom correspondence should be addressed ([email protected]).

and A′′ combine their effects with the N-terminaltargeting sequence A of RB3 or RB3′′, we studiedthe subcellular localization of combinations presentin RB3, RB3′ and RB3′′, namely domain A–domainA′ and domain A–domain A′′–domain A′ fused toGFP. As shown in Figure S1(B), these two fusionproteins were absent from the plasma membrane orfrom the cytosol, but were present at the Golgi, asshown by their co-localization with the CTR433 me-dian Golgi marker. They were also localized, similarto domain ARB3–GFP, on multiple puncta through-out the cytosol. These data reveal that domain A hasa dominant effect for subcellular targeting comparedwith domains A′′ and A′, since their association doesnot target GFP to the plasma membrane or to thecytosol. Domain A′′, however, also contains a mem-brane targeting signal that might be exposed in con-ditions when domain ARB3 is inactivated, such asafter a cleavage. To verify this hypothesis, we con-structed domain A′′–domain A′ fused to GFP and,surprisingly, observed a cytosolic distribution of thefusion protein as with the domain A′–GFP construct(data not shown).

In conclusion, our results demonstrate that onlydomain A promotes the specific Golgi/vesicle-liketargeting of stathmin-related proteins. The other do-mains do not directly participate in the subcellulartargeting of stathmin family proteins, but may con-tribute in diversifying their functional properties.

www.biolcell.org | Volume 100 (10) | Pages 577–589


Figure S1 Dominant effect of domain A over domains A′

and A′ ′ of RB3 variants for Golgi and vesicle-likelocalization(A) Fusion proteins generated from RB3/RB3′/RB3′ ′. Domain A

from RB3 (ARB3, black), domain A′ (grey) and domain A′ ′ (white)

alone, or in combination, were fused to GFP and transfected

in HeLa cells. As shown previously (Charbaut et al., 2005),

domain A alone is able to shuttle GFP to the Golgi and ves-

icle-like structures. In contrast, domains A′ and A′ ′ displayed

a soluble distribution similar to the classical localization of

GFP alone, but a fraction of domain A′ ′ was also present at

the plasma membrane (arrows). (B) Subcellular localization of

domains A, A–A′ (corresponding to RB3/RB3′) and A–A′ ′–A′

(corresponding to RB3′ ′) fused to GFP in HeLa cells. Co-loc-

alization with a Golgi marker, CTR433. Domain A alone or

in fusion with other domains shows a Golgi and vesicle-like

localization. Scale bar, 10 μm.

Figure S2 Comparison of subcellular localization ofdomains A of SCG10, SCLIP and RB3 with endogenousneuronal stathmin proteins within growth cones byconfocal microscopyDomains A of SCG10 (ASCG10), SCLIP (ASCLIP) or RB3 (ARB3)

were fused to GFP and transfected into cultured hippocampal

neurons. Double labelling with an anti-GFP and an anti-en-

dogenous SCG10 or SCLIP antibody revealed the presence

of the fusion proteins similar to the endogenous ones in the

growth cones of cultured neurons. The localization of the fu-

sion proteins essentially reflects that of the corresponding en-

dogenous ones, with some local differences, mostly between

RB3 and SCG10 vesicle-like staining. Scale bar, 5 μm.

Reference

Charbaut, E., Chauvin, S., Enslen, H., Zamaroczy, S. and Sobel, A.(2005) Two separate motifs cooperate to target stathmin-relatedproteins to the Golgi complex. J. Cell Sci. 118, 2313–2323

Received 6 September 2007/14 April 2008; accepted 21 April 2008

Published as Immediate Publication 21 April 2008, doi:10.1042/BC20070119

C© The Authors Journal compilation C© 2008 Portland Press Ltd

palmitoylation of stathmin family proteins domain a controls golgi versus mitochondrial subcellular...

Documents