Molecular organization of the cullin E3 ligase adaptor KCTD11

lable at ScienceDirect

Biochimie xxx (2011) 1e10

Contents lists avai

Biochimie

journal homepage: www.elsevier .com/locate/biochi

Research paper

Molecular organization of the cullin E3 ligase adaptor KCTD11

Stefania Correale a,b,1, Luciano Pirone a,b,1, Lucia Di Marcotullio c, Enrico De Smaele d, Azzura Greco c,Daniela Mazzà c, Marta Moretti d, Vincenzo Alterio a, Luigi Vitagliano a, Sonia Di Gaetano a,*,Alberto Gulino c, Emilia Maria Pedone a,*

a Institute of Biostructures and Bioimaging, CNR, 80134 Napoli, ItalybDepartment of Biological Sciences, “Federico II” University, 80134 Napoli, ItalycDepartment of Molecular Medicine, La Sapienza University, 00161 Roma, ItalydDepartment of Experimental Medicine La Sapienza University, 00161 Roma, Italy

a r t i c l e i n f o

Article history:Received 19 August 2010Accepted 28 December 2010Available online xxx

Keywords:UbiquitinationProteineprotein interactionsMedulloblastomaHedgehog pathwayStructureefunction relationshipsLight scattering

Abbreviations: KCTD, potassium channel tetramproteins; PDB, protein data bank; sKCTD11 and lKCTforms, respectively; Cul3, cullin 3; HDAC1, histoneKCTD5 BTB, and KCTD21 BTB, POZ/BTB domains of lrespectively.* Corresponding authors. Institute of Biostructure

Mezzocannone 16, 80134, Napoli, Italy. Tel.:þ39 08125E-mail addresses: [email protected] (S. Di Gaetan

Pedone).1 These authors equally contributed.

0300-9084/$ e see front matter � 2011 Elsevier Masdoi:10.1016/j.biochi.2010.12.014

Please cite this article in press as: S. Correalej.biochi.2010.12.014

a b s t r a c t

The family of human proteins containing a potassium channel tetramerization domain (KCTD) includes21 members whose function is largely unknown. Recent reports have however suggested that theseproteins are implicated in very important biological processes. KCTD11/REN, the best-characterizedmember of the family to date, plays a crucial role in the ubiquitination of HDAC1 by acting, in complexwith Cullin3, as an E3 ubiquitin ligase. By combining bioinformatics and mutagenesis analyses, here weshow that the protein is expressed in two alternative variants: a short previously characterized form(sKCTD11) composed by 232 amino acids and a longer variant (lKCTD11) which contains an N-terminalextension of 39 residues. Interestingly, we demonstrate that lKCTD11 starts with a non-canonical AUUcodon. Although both sKCTD11 and lKCTD11 bear a POZ/BTB domain in their N-terminal region, thisdomain is complete only in the long form. Indeed, sKCTD11 presents an incomplete POZ/BTB domain.Nonetheless, sKCTD11 is still able to bind Cul3, although to much lesser extent than lKCTD11, and toperform its biological activity. The heterologous expression of sKCTD11 and lKCTD11 and their individualdomains in Escherichia coli yielded soluble products as fusion proteins only for the longer form. Incontrast to the closely related KCTD5 which is pentameric, the characterization of both lKCTD11 and itsPOZ/BTB domain by gel filtration and light scattering indicates that the protein likely forms stabletetramers. In line with this result, experiments conducted in cells show that the active protein is notmonomeric. Based on these findings, homology-based models were built for lKCTD11 BTB and for itscomplex with Cul3. These analyses indicate that a stable lKCTD11 BTB-Cul3 three-dimensional modelwith a 4:4 stoichiometry can be generated. Moreover, these models provide insights into the determi-nants of the tetramer stability and into the regions involved in lKCTD11-Cul3 recognition.

� 2011 Elsevier Masson SAS. All rights reserved.

1. Introduction

Potassium channel tetramerization domain containing proteins(KCTD) constitute a family of 21 human proteins that share aconserved domain at their N-terminus [1]. The name of the protein

erization domain containingD11, short and long KCTD11deacetylase 1; lKCTD11 BTB,KCTD11, KCTD5, and KCTD21,

s and Bioimaging, CNR, Via34521; fax:þ39 0812534574.o), [email protected] (E.M.

son SAS. All rights reserved.

, et al., Molecular organization

family derives from the sequence similarity of their N-terminalregion with the tetramerization domain detected in some voltage-gate potassium channel. Comparative sequence analyses suggestthat this domain adopts the common POZ/BTB (Bric-a-brack,Tram-track, Broad complex) fold. POZ/BTB are widespread andwell-characterized domains detected in several diversified systemsinvolved in a variety of biological processes [2e4]. Recentcomparative analyses of different POZ/BTB domains have clearlyindicated that their structures are characterized by variations ona conserved structural theme. Indeed, in all structures a moduleformed by a three-stranded b-sheet and five a helices is present [4].

Despite the similarity of the N-terminal regions among KCTDproteins, their C-terminus is highly variable [5]. Although thebiological role of these proteins is yet to be determined, recentinvestigations suggest that theyare involved in important biological

of the cullin E3 ligase adaptor KCTD11, Biochimie (2011), doi:10.1016/

mailto:[email protected]

mailto:[email protected]

www.sciencedirect.com/science/journal/03009084

http://www.elsevier.com/locate/biochi

http://dx.doi.org/10.1016/j.biochi.2010.12.014



S. Correale et al. / Biochimie xxx (2011) 1e102

processes. It has been shown that KCTD5 interacts with the humanGolgi reassembly stacking protein [6], while amissense mutation ofKCTD7 is linked to neurodegeneration and progressive myoclonicepilessy [7]. Moreover, KCTD15 variants have been associated withadult obesity risk [8]. Very recently it has been shown that KCTD12,previously characterized as an important factor involved in thematuration of ear neurons, along with KCTD8 and KCTD16 act asauxiliary subunits of GABA(B) receptors that determine the phar-macology and kinetics of the receptor response [9,10].

The tumor suppressor KCTD11, previously denoted as REN, is animportant antagonist of the Hedgehog (Hh) pathway which isfrequently deleted in human medulloblastoma [11e15] and widelydown-regulated in other human cancers [16]. KCTD11 has beenidentified as a 232 amino acids protein present in severalmammalian genomes. From the biological point of view, we haverecently shown that this protein plays a major role in the mecha-nisms of deacetylation and activation of Gli1 and Gli2, two keytranscription factors of Hh signaling [17]. Indeed, Gli deacetylationis regulated by a multiprotein module involving HDAC and an E3ubiquitin ligase, formed by Cullin3 (Cul3) and KCTD11. Gli1 isdeacetylated by histone deacetylase 1 (HDAC1), which is ubiq-uitined for degradation by the E3 ubiquitin ligase, thereby sup-pressing Hedgehog signaling [17]. We here report a molecular andstructural characterization of KCTD11. We show that two differentforms of the protein can be expressed in transfected HEK293 cells.Information on the protein structure, on its oligomeric organizationand on the interaction with Cul3 have been obtained by combiningspectroscopic data and molecular modeling.

2. Materials and methods

2.1. Database sequence/structure surveys

Sequence analyses were conducted by using the tools of theserver ExPaSy (http://www.expasy.org/). These analyses were alsoperformed by using data available at the POZ/BTB database (http://btb.uhnres.utoronto.ca/index.html) developed by Privé and co-workers [4]. The sequence of KCTD11 c-DNAwas retrieved from thesite http://www.ensembl.org/.

2.2. Protein expression, purification and mutagenesis

All materials used for gene amplification were supplied byStratagene Cloning Systems. All synthetic oligonucleotides werepurchased from Sigma (Italy). All aqueous solutions were madeusing water purified by a Milli-Q water system (Millipore). All thechemicals used were of the highest grade available. Expressionvector pRSET was from Invitrogen; pETM11, pETM20, pETM41 andpET32 were from Novagen. E. coli DH5a, E. coli BL21(DE3)codonplus RIL were purchased from Stratagene. BL21(DE3), BL21(DE3)STAR strains were supplied by Invitrogen. The identity of theinserts in the resulting recombinant plasmids was confirmedby DNA sequencing (MWG-Biotech). All resulting expressionplasmids were used to transform different competent strains foranalytical expression purposes. The construct “N-module” 40e145of KCTD5 was a gift from Prof. S. A. N. Goldstein (University ofChicago, USA).

The molecular mass of each recombinant protein was evaluatedby mass spectrometry. A LCQ DCA XP Ion Trap mass spectrometer(ThermoElectron, Milan, Italy) equipped with an OPTON ESI sourceand a complete Surveyor HPLC system (including an MS pump, anauto-sampler and a photo diode array) was used. Analyses wereperformed using a 300 Å narrow bore 250 mm � 2 mm C4 Jupitercolumn (Phenomenex, Torrance, CA) and applying a gradient of

Please cite this article in press as: S. Correale, et al., Molecular organizationj.biochi.2010.12.014

solvent B (0.05% TFA in CH3CN) in solvent A (0.08% TFA in H2O) from5 to 70%, over a period of 40 min.

Asp68, Arg73, Asn77 and Arg80 residues of the POZ/BTB domainwere mutated in Alanine by using the QuikChange Site DirectedMutagenesis Kit (Stratagene).

2.3. Design of E. coli constructs derived from sKCTD11

Different constructs deriving from sKCTD11 nucleotidesequence containing BamHI and EcoRI restriction sites at the N- andC-terminus, respectively were amplified by PCR. The amplified DNAfragments (sKCTD11 BTB and sKCTD11), opportunely digested, wereinserted into pRSET expression vector with a cleavable N-terminalpolyHis tag.

2.4. Design of E. coli constructs derived from lKCTD11

The full length lKCTD11 nucleotide sequence was amplified byusing specific oligonucleotides containing NcoI and XhoI restrictionsites at the N- and C-terminus, respectively. The amplified DNAfragment, opportunely digested, was inserted into pETM20,pETM30, pETM41 plasmids containing an N-terminal poly His-fusion partner tag (Trx, GST and MBP, respectively).

Based on the lKCTD11 BTB nucleotide sequence, different oligo-nucleotides were designed and used as primers in the PCR geneamplification procedure containing all NcoI and XhoI restrictionsites. pETM20 (Novagen) expression vector, which gives a proteinwith anN-terminal polyHis-Trx tag,was utilized. The amplifiedDNAfragments, opportunely digested, were inserted into the pETM20plasmid. The DBTB (113e271) construct was subcloned from pRSETinto pET32 plasmid with an N-terminal polyHis-Trx tag.

2.5. Expression and purification of recombinant proteins

After a first screening of small-scale expression cultures, per-formed using different strains, temperatures, IPTG concentrationand induction length, recombinant constructs were transferredinto E. coli strain that assured the best expression level in solublephase.

Cell pellets from 500ml cultures were re-suspended in 10ml of20 mM TriseHCl pH 7.5 (buffer A), supplemented with a proteaseinhibitor cocktail (Complete EDTA-FREE, Roche). Crude extractswere prepared by disrupting the cells with 10 min pulses at 20 Hz(Misonix Sonicator 3000) and centrifugating lysates at 16 000 rpmfor 30 min. The crude extracts were applied on a HisTrap HP (GEHealthcare) equilibrated with buffer A containing 0.3 M NaCl and10 mM imidazole. Proteins were eluted with the same buffer Asupplemented with 300 mM imidazole. The active fractions werepooled, extensively dialyzed against 50 mM Tris/HCl pH 7.5,0.5 mM EDTA and successively digested with the specific protease(TEV or Enterokinase). The digested samples were applied ontoa HisTrap performed as above described. The flow-throughsamples were applied onto a 1.6 cm � 60 cm column (HiLoadSuperdex 200, GE Healthcare) connected to an AKTA system(GE Healthcare) and eluted with Buffer A containing 0.15 M NaClat a flow rate of 1 ml/min.

2.6. Analytical methods for protein characterization

Protein concentration was determined by the Bradford methodusing BSA as the standard. Protein homogeneity was estimated bySDS/PAGE. In addition, the proteins were analyzed by a non-dena-turing electrophoresis 10% (w/v) polyacrylamide slab gel at pH 4.5.

The molecular mass of the proteins was determined usinga SEC-LS system consisting of a semi-preparative size-exclusion


http://www.expasy.org/

http://btb.uhnres.utoronto.ca/index.html

http://btb.uhnres.utoronto.ca/index.html

http://www.ensembl.org/

S. Correale et al. / Biochimie xxx (2011) 1e10 3

chromatography column (Superdex200 10/30, GE Healthcare)coupled to a light scattering detector (miniDAWN TREOS-WyattTechnology) and a differential refractive index detector (ShodexRI-101). The Astra (5.3.4 version, Wyatt Technology Corporation)software allowed us to the collect, record, and process the scatteringdata. The standards used for the calibrationwere aldolase (158 kDa),ovalbumin (43 kDa) and carbonic anidrase (29 kDa). The molecularmass of KCTD11 DBTB was determined using a Superdex75 PC 3.2/30 (GE Healthcare). The standards used for the calibrationwere BSA(66 kDa), ovalbumin (43 kDa) and chymotrypsinogen (22.8 kDa).

Far UV-CD spectra were recorded on a Jasco J-710 spec-tropolarimeter equipped with a Peltier thermostatic cell holder(Jasco, model PTC-343) using 1 mm path length cell at 190e260 nmat 20 �C. Thermal denaturation analyses were conducted byfollowing the CD signal at 223 nm.

Recordings were carried out at 20 �C under constant N2 flow byusing the following parameters: scanning speed of 20 nm min�1,band width of 2 nm, and response time 4 s. The sample was heatedat the rate of 1 �C min�1.

2.7. Cell culture, plasmids, mutagenesis, antibodies, transfections

HEK293T were cultured in DMEM plus 10% FBS. PCR-amplifiedcDNAs of different human KCTD11 and mutated forms, of Cul3were cloned into various tagged-vectors. Antibody sources andconcentrations used were: Cul3 (1:500; Zymed/Invitrogen); mousemonoclonal against HA andMyc (western blotting 1:1000; IP 1:500;Santa Cruz) and Flag (M2; western blotting 1:2000; IP 1:200;Sigma); the rabbit polyclonal anti-KCTD11 antibody was generatedusing a recombinant C-terminal fragment of the protein as previ-ously described in Gallo et al. [18].

KCTD11 residues were mutated by the Quickchange site-directed or multi-site mutagenesis kit (Stratagene). Transfectionswere performed using Lipofectamine 2000 or Plus (Invitrogen).

Fig. 1. Canonical structure and representative sequences of POZ/BTB domains. (A) A typical mfile 1NN7. Secondary structure elements which are missing in sKCTD11 are reported in greyKv4.2. KCTD BTB sequences are from Homo sapiens whereas the Kv4.2 BTB one is from rat. Thand Kv4.2 a-helical and b-sheet regions are shown in magenta and cyan, respectively.


Extract preparation, immunoprecipitation and immunoblottingwere performed as previously described [19].

2.8. Immunoblotting, immunoprecipitation and antibodies

Transfected cells were lysed in 50 mM TriseHCl pH 7.6, 150 mMNaCl, 1% NP-40 and protease inhibitors. For immunoblotting (IB),protein extract was separated by SDS-PAGE. Immunoprecipitation(IP) were carried out as described previously [17,19]. Antibodysources used were: anti-HA (sc-7392), anti-Myc (sc-40AC), anti-Actin (sc-1616) purchased from Santa Cruz Biotechnology; anti-Flag(M2) from Sigma.

2.9. Molecular modelling

A three-dimensional model for lKCTD11 BTB was generated byusing standardmolecularmodeling techniques. The structure of theKv4.2 BTB (Protein Data Bank code 1NN7) domain determined at2.1 Å [20] was used as template. Computermodelingwas conductedon a Silicon Graphics Octane workstation using the O program [21].Several cycles of constrained energy minimization by using theInsight/Discover program package regularized the structure andgeometrical parameters. Minimizations were performed using theCVVF force field until a final energy gradient of 0.1 kcal/mol Å.

The overall architecture of the complex between KCTD11 BTBand Cul3 was generated by taking into account the crystal structureof the complex between Skp1 and Cul1 (PDB code 1LDK) [22]. Themodel of Cul3 was generated from the structure of the closelyrelated Cul4 (PDB code 2HYE) [23]. The larger complex formed byKCTD11 BTB tetramer with Cul3, Rbx1, and E2 was generated byconsidering the crystal structure of the complex Skp1 e Cul1 e

Rbx1 by docking of E2 on Rbx1 taking into account the architectureof the related complex E2-cCbl [24].

odel of the structure of a POZ/BTB domain generated using the coordinates of the PDB(B) Alignment of KCTD11 sequence with KCTD21 and the BTB domains of KCTD5 ande N-terminal residues present only in KCTD11 long form are colored in grey. For KCTD5



3. Results

3.1. Surveys of sequence and structure databases suggestthat an alternative form of KCTD11 may exist

All structures of POZ/BTB domains so far reported are character-ized by a conservative module formed by a three-stranded b-sheetand five a-helices (Fig. 1A) [4]. Alignments of POZ/BTB sequences ofmembers with unknown 3D-structures also suggest that theseelements are strictly conserved among all proteins of the family [4].Intriguingly, the POZ/BTB domain of KCTD11 is peculiar in thisgeneral framework. InKCTD11onlya portionof the conserved core ofthe POZ/BTB domain is present. Sequence alignments show that the

Fig. 2. Nucleotide sequence of KCTD11 ORF. The ATG triplet encoding the starting Met of


part of the domain corresponding to the b-strands B1 and B2 and tothe a-helix A1 ismissing (Fig.1B). This unusual scenario promptedusto check the possibility of an alternative initiation point of KCTD11RNA translation. The analysis of the KCTD11 translated c-DNA clearlyindicates that the entire POZ/BTB conserved core could be generatedby simply considering an alternative upstream starting point.Indeed, the DNA triplets preceding those corresponding to theputative first residue of the reported sequence (Met1) encode forresidues that are very similar to those observed in the other POZ/BTBdomains. The generation of the entire POZ/BTB requires at least 24extra amino acids at the N-terminal side of the protein (Fig. 1). Thisobservation prompted us to perform experiments aimed at verifyingthe occurrence of KCTD11 alternative forms.

sKCTD11 and the ATT triplet encoding the starting Ile of lKCTD11 are shown in bold.



3.2. An alternative translation start in KCTD11 mRNA

Taking into account the prediction provided by the survey ofprotein structure/sequence containing BTB domains, the entireKCTD11 ORF was transfected in HEK293 mammalian cells (Fig. 2).The expression of this c-DNA yielded two distinct forms of theprotein with different molecular weights (Fig. 3A left). The shortestone (hereafter denoted as sKCTD11) shows a molecular weight thatis comparable to the previously characterized and denoted asKCTD11/REN [11e15]. The longest one presents an approximatemolecular weight of 30 kDa that wewill describe as long (lKCTD11).This latter form appears to be also more stable than the short form,which is stabilized by ZLLL mediated block of proteasomal inhibi-tion (Fig. 3B). Since no classical AUG start was present on theupstream sequence of the mRNA, the alternative translation initi-ation site of lKCTD11 is a non-AUG start. In order to identify thisstart site we have analyzed the upstream region in mRNAsequences from human, mouse, rat and bovine KCTD11, in searchfor nucleotide conservation. We have observed that there is goodconservation in other organisms through a sequence starting atnucleotide 656. The sequence upstream does not appear to besufficiently conserved (Supplementary Fig. S1). Looking for alter-native non-AUG start in this region and taking into account non-AUG coding triplets reported in literature, we identified threepossible initiation points, the triplet CUU (656e658 encodinga leucine), AUU (665e667 encoding an isoleucine), and GUG(680e682 encoding a valine). The latter starting point wasexcluded by considering that a plasmid construct containing thehuman KCTD11 ORF starting at nucleotide 674 in the 50 sequence,when transfected in HEK293 human cell line, did not generate thelong form, but only the short one (not shown). We then replacedthe triplets 656e658 and 665e667 with the triplet GCT encodingan alanine, which has never been reported as a start translationpoint. The replacement of the triplet 656e658 did not produce anyvariation in the expression profile with both forms present. On theother hand the replacement of the triplet 665e667 abolished theexpression of the long form (Fig. 3C). Altogether these findingssuggest that the AUU (665e667) represents the start point oflKCTD11. Of relevance, the short form was still generated from 665

Fig. 3. Non-AUG start of the long form of KCTD11. (A) HEK293 cells were transfected with665e667 encoding an Ile residue (ISP). After 24 h from transfection the cells were lysed andvisualized by rabbit anti-KCTD11 antibody. (B) sKCTD11 protein is more unstable and increncoding the complete KCTD11 ORF. 24 h after transfection, cells were treated for 4 h with orin (A). (C) HEK293 cells were transfected with plasmids encoding the complete KCTD11 ORF (Ala mutation in the triplet 665e667). Cell lysates were analyzed as in (A). (D) HEK293 cells,or without ZLLL (50 mM), lysed and analyzed by Western blot as in (A).


to 667 replacement, and was stabilized by proteosome inhibitionby ZLLL (Fig. 3D).

3.3. Heterologous expression of KCTD11 and its variants

Initial experiments were focused on the expression of sKCTD11and lKCTD11 full length in E. coli. All attempts made to expresssKCTD11 were unsuccessful. On the other hand, lKCTD11 could beexpressed into BL21(DE3) STAR cells as a fusion protein with MBP.SDS-PAGE analysis of the cytosolic extract revealed that MBP-tag-ged KCTD11 was expressed as a soluble product in large amounts.The fusion protein was initially purified by affinity chromatographyby exploiting the presence of the N-terminal His-tag. The proteinwas further purified by size-exclusion chromatography (Fig. 4). Thedifferent results obtained in the expression of these two forms arelikely related to the instability of the incomplete BTB domainpresent in sKCTD11.

Although theMBP-tagged form of lKCTD11 could be purified, theremoval of the fusion partner MBP leads to the precipitation of theprotein, thus preventing any further characterization. In thisscenario, both variants were dissected in smaller putative domainsthat were designed by performing sequence alignments and byexploiting previous functional data on KCTD11 fragments. Inparticular, the attention was focused on the characterization ofPOZ/BTB domain and the C-terminal region of the protein (DBTB).Different variants for the BTB domain, differing both at the N- andC-terminus were tried (Supplementary Fig. S2). The N-terminusconsidered for the constructs was either the one corresponding tothe long (lKCTD11 BTB) or to the short (sKCTD11 BTB) form. Variantsat the C-terminus were generated by considering several BTBconstructs of different lengths (Fig. S2).

In line with the results obtained on sKCTD11, all attempts madeto express its BTB domain in E. coliwere unsuccessful. On the otherhand, its alternative longer variants of lKCTD11 BTB could beobtained in soluble form as a fusion product with TrxA. These BTBdomains have been purified to homogeneity after the removal ofthe fusion protein (Fig. 4). The yield of the recombinant proteinwasabout 40 mg/l. Minor effects were produced by extensions at theC-terminus or by the presence of the first fourteen residues at the

plasmids encoding the complete KCTD11 ORF (ORF) or the c-DNA starting from tripletprotein extracts analyzed by Western blot. The long and short forms of KCTD11 were

eases following proteasomal inhibition. HEK293 cells were transfected with plasmidwithout the proteasome inhibitor ZLLL (50 mM), lysed and analyzed by Western blot aswt) or the mutants KCTD11L/A (Leu to Ala mutation in the triplet 656e658) or I/A (Ile totransfected with empty or L/A or I/A KCTD11 mutants vectors, were treated for 4 h with


Fig. 4. SDS-PAGE analysis of different purified constructs. Lane 1 low range proteinmarker (sigma); lane 2 MBP-lKCTD11, lane 3 lKCTD11 BTB domain, lane 4 KCTD11DBTB, lane 5 KCTD5 BTB domain.


N-terminus. Therefore, thereafter we will describe the BTB domaincorresponding to residues 15e126 of lKCTD11 (Fig. S2).

The DBTB construct (residues 114e271 of lKCTD11), cloned intopRSET vector, was always found in inclusion bodies. On thecontrary, the fusion protein TrxA-DBTB was obtained in solubleform. DBTB has been successfully purified after the removal of thefusion protein.

3.4. Solution studies on KCTD11 and its variants

The production of KCTD11 variants as soluble products providedthe opportunity to perform solutions studies aimed at gaininginformation on the molecular and structural organization of theprotein. As expected on the basis of literature data on homologousdomain, the BTB domain of KCTD11 shows a far-UV CD spectrumcharacteristic of a/b proteins (Fig. 5A). The ability of the isolatedBTB domain to assume a well-defined state is confirmed by theobservation that crystals of this construct could be obtained usingPEG as precipitating agent (Fig. 5B). However, the resolution of the

Fig. 5. Spectroscopic characterization and crystallization of lKCTD11 fragments. (A) Far-UVKCTD5 BTB (dotted line) and lKCTD11 BTB (continuous line): CD signal was followed at 223


data collected on these crystals (w4 Å) prevents hitherto highresolution structural studies on the domain.

The structural stability of the domain was checked by per-forming thermal denaturation experiments following the CD signalat 223 nm. Although a detailed thermodynamic characterizationwas hampered by protein precipitation upon unfolding, thisstudy indicates that the domain is stable up to 45 �C (Fig. 5C). Acomparative analysis of the thermal stability of KCTD5 and lKCTD11BTB domains indicates that the former is significantly more stable.

The analysis of the CD spectrum of the DBTB domain, that doesnot display any significant identity with proteins with knownstructure, indicates that it assumes a folded structure. The locationof the minimum (w214 nm) in the curve suggests that KCTD11DBTB possesses predominantly a b-structure (Fig. 5D). Taking intoaccount that the DBTB domain of KCTD5 has a b-structure [5], it ispossible that the C-terminal ends of KCTD5 and KCTD11 sharesimilar structures despite the lack of any detectable sequencesimilarity.

These proteins were also used for gel filtration and light scat-tering analyses aimed at determining their oligomeric state insolution. Gel filtration experiments indicate that MBP-lKCTD11 hasan apparent molecular weight of about 254 kDa, which is compat-ible with a tetrameric state (Fig. 6A). To identify the role of proteindomain in the oligomerization process we also analyzed both BTBand DBTB domains. While DBTB is monomeric (molecular weightof w15 kDa), the BTB domain appears to be tetrameric (w46 kDa).Since it has been recently reported that the closely related BTBdomain of KCTD5 (sequence identity 40%) is pentameric [5], we alsoexpressed and purified this domain for comparative purposes. Theapparent molecular weight of this latter domain is w64 kDacompatible with a pentameric state (Table 1). In order to furthercheck these results we performed static light scattering experi-ments on MBP-lKCTD11, lKCTD11 BTB and KCTD5 BTB. RegardingKCTD11 DBTB, SEC-LS experiments could not be performed due tothe low solubility of this form.

The hydrodynamic radius (9.2 nm) and the molar mass detectedthrough SEC-LS analyses (274.8 kDa) of MBP-lKCTD11 confirmedthe tetrameric organization of the protein. Light scattering

spectrum of lKCTD11 BTB; (B) a crystal of lKCTD11 BTB; (C) Thermal denaturation ofnm. (D) Far-UV spectrum of lKCTD11 DBTB.


Fig. 6. Gel filtration elution profiles of: (A) MBP-lKCTD11 (S200 10/30); (B) KCTD11 DBTB (Superdex75 PC 3.2/30); (C) lKCTD11 BTB (S200 10/30); (D) KCTD5 BTB (S200 10/30). ForS200 10/30 column the formula was: y ¼ �0.2952x þ 0.9006 while for Superdex75 PC 3.2/30 was: y ¼ �1.2733x þ 2.0698.


experiments confirmed the gel filtration results also for lKCTD11BTB and KCTD5 BTB. Indeed, lKCTD11 BTB displayed a hydrody-namic radius (4.5 nm) and a molar mass (w56.6 kDa) compatiblewith a tetrameric state, while the values exhibited by KCTD5 BTB(molecular weight w64.1 kDa and hydrodynamic radius 6.0 nm)are in line with the pentameric state of this assembly.

3.5. Functional characterization of lKCTD11 and sKCTD11

Previous investigations have shown that KCTD11 is involved inthe formation of an E3 ubiquitin ligase complex with Cul3. Inanalogy with other POZ/BTB containing proteins, the POZ/BTBdomain mediates the interactions between KCTD11 and Cul3. Wechecked whether the two forms of the protein lKCTD11 andsKCTD11 could be able to form such a complex. As shown in Fig. 7both KCTD11 variants are able to interact with Cul3. This is some-what surprising for sKCTD11, taking into account that its incom-plete POZ/BTB domain lacks important structural elements.

By using ad-hoc flagged variants of lKCTD11 we also analyzed itsaggregation state in cells. In line with the results obtained in thecharacterization of the protein expressed in E. coli we found thatthe protein is not monomeric being able to form higher assemblies(Fig. 8).

3.6. Homology modeling of the lKCTD11 BTB domain

The analysis of the protein structures deposited in the PDBreveals the presence of POZ/BTB domain models displaying

Table 1Evaluation of the protein masses. The experimental data derived from the calibra-tion formula.

Proteins Calculated massesof the monomers

Experimental/calculatedmass ratios

lKCTD11 BTB 12.4 kDa 3.7KCTD5 BTB 12.4 kDa 5.16MBP- lKCTD11 69 kDa 3.68KCTD11 DBTB 16 kDa 0.94


significant sequence identities with lKCTD11 BTB (w35e40%). Thisfinding opens the possibility for homology modeling studies of thedomain. The highest sequence similarity to lKCTD11 BTB detectedin the PDB is displayed by pentameric KCTD5 BTB (w40%). A slightlylower value (w33%) is exhibited by POZ/BTB tetramerizationdomain of the potassium channels Kv4.3. Since lKCTD11 BTB wasshown to be a tetramer in solution we used this latter model asa template (PDB code 1NN7) (Fig. 9). The analysis of the 3D modelgenerated for lKCTD11 BTB suggests analogies and differences withthe Kv4 model used as a template. In POZ/BTB Kv4 the tetramerinterface is stabilized by two different types of interactions. Themost important ones are mediated by a zinc ion that is coordinatedby cysteine and histidine residues belonging to different subunits.This feature is however not present in lKCTD11 BTB domain. Furtherstability to the POZ/BTB tetramer of Kv4 is provided by hydrogenbonding and electrostatic interactions among residues located atthe inter-subunit interface. Interestingly, some of these residues areconserved in KCTD11. In particular, the conserved residues includeArg93 (Kv4 sequence numbering, Arg73 in KCTD11 sequence),Asp88 (Asp68), Asn97 (Asn77), and Arg100 (Arg80). The analysis ofKCTD11 BTB model shows that these charged residues are close atthe tetramer interface. This indicates that the stability of thelKCTD11 BTB tetramer essentially relies on electrostatic interactionsat the intersubunit interface. This observation is corroborated bythe finding that the tetra-mutant BTB (Asp68Ala, Arg73Ala,Asn77Ala, Arg80Ala), expressed in the same conditions describedas for wild-type, could not be obtained in soluble form also whenfused with TrxA (data not shown).

3.7. Modeling the interaction of lKCTD11 BTB tetramer with Cul3

As reported above the BTB domain mediates the interaction oflKCTD11 in the formation of a complex with Cul3. Since structuresdescribing the interactions between proteins of the Cul family withPOZ/BTB domains have been reported, we generated by usinghomologymodeling techniques a 3Dmodel of the complex betweenCul3 and lKCTD11 BTB. It is important to note that all structuresof POZ/BTB-Cul complexes hitherto reported contain a single


Fig. 7. Interaction of sKCTD11 (A) and lKCTD11 (B) with Cul3. HEK293 cells, co-transfected with Myc-Cul3 and Flag-sKCTD11 or Flag-lKCTD11, were lysed and immunoprecipitatedwith a Flag antibody. The interactions of sKCTD11 or lKCTD11 with Cul3 were detected with a Myc antibody. The blot was reprobed with anti-Flag.


monomeric BTB domain interacting with a single Cul molecule (1:1stoichiometry). The experimental evidence that lKCTD11 BTB likelyforms stable tetramer in solution prompted us to check the possi-bility to generate larger POZ/BTB-Cul3 complexes. We initiallyconsidered the structure of the complex between Cul1 and Skp1 tobuild a one to one complex of lKCTD11 BTB-Cul3 (see Methods). Insuch a model, the BTB domain interacts with its partner essentiallythrough the b-strand B3 of the b-sheet and the helix A4.

A convincing complex between lKCTD11 BTB and Cul3with a 4:4stoichiometry could be generated by considering the structure ofthe BTB tetramer and the interactions of each subunit with Cul3(Fig. 10). Interestingly, an a-helix of Cul3 docks into a cavity formedat the inter-subunit interface within the tetramer. The analysis ofthe model also shows that, in this complex, a larger portion of thePOZ/BTB domain is involved in the complex interface if comparedto the 1:1 model. Indeed, each Cul3 molecule interacts with twodistinct BTB subunits (Fig. 10B). As shown in Fig. 10C, residuePhe102 of lKCTD11 BTB directly interacts with a cluster of aromaticside chains of Cul3 made of Tyr58, Tyr62 and Tyr125. This is in linewith the observation that the mutation of these residues abolishesthe Cul3-lKCTD11 interaction ([17] and Fig. S3).

4. Discussion

Recent investigations have unveiled that members of the KTCDfamily are involved in important and diversified cellular processes

Fig. 8. KCTD11 is not monomeric in transfected HEK293 cells. HEK293 cells weretransfected with HA-lKCTD11 in presence of Flag-lKCTD11 or Flag-ΔBTB/lKCTD11. Thelysates were immunoprecipitated with an HA antibody. The interactions were detectedwith a Flag antibody. The blot was reprobed with anti-HA.


[1e17]. Despite their denomination, the functional characterizationof the members of this family unveils that these proteins are notimplicated in potassium transport across the membranes. We herecharacterized at molecular level KCTD11, an important regulatorof HDAC1 and of the Hedgehog pathway. By combining bio-informatics, biochemical and functional investigations we showthat the protein is expressed in two distinct forms: a short previ-ously characterized (sKCTD11) and a longer one (lKCTD11). Theshort form is characterized by an incomplete POZ/BTB domain,which lacks important secondary structure elements (two strandsand one a-helix) of the domain scaffold. Expression of this form intransfected HEK293 cells has shown that sKCTD11 is highlysusceptible to degradation by proteosome. On the other hand, thelong form possesses a POZ/BTB domain endowed with all of thecanonical secondary structure elements and an enhanced stabilityto degradation. We here unveiled a peculiar feature of lKCTD11,which presents a non-canonical AUU translation start. Indeed,

Fig. 9. Structural model of lKCTD11 BTB tetramer. Different colors have been used forthe four subunits of the assembly.


Fig. 10. Complex of lKTCD11 BTB with Cul3. (A, B) Two different views of the complexbetween the KTCD11 POZ/BTB tetramer with four molecules of Cul3 (red). (C) Asnapshot of the lKTCD11 BTBeCul3 interface.

Fig. 11. Complex between the lKTCD11 BTB tetramer with Cul3 (red), Rbx1 (yellow)and E2 (green).


eukaryotic translation is rather stringent in using an AUG codon asinitiation site [25].

However, it has been reported that, in approximately 10% ofeukaryotic mRNA, ribosomes may initiate at an upstream non-AUGcodon in addition to initiating at the first AUG. In recent years,growing evidences have confirmed that alternative aminoacidssuch as leucine, isoleucine, or valine can also be translated asinitiator amino acids [26]. Alternative translation initiation signal,first identified among viral genes, have been demonstrated also ineukaryotic genes, such as proto-oncogenes, transcription factorkinases and growth factors. Significantly, the use of alternativetranslation signals, to generate multiple proteins from a singlegene, contributes to the generation of protein heterogeneity,thereby allowing an organism to increase its level of complexity[27]. It has also been proposed that this process is used either as amean to produce two functional proteins or for regulatory reasons.


Data here presented indicate that KCTD11 expression fits into thisscheme, although the functional implications of this finding are yetto be elucidated. Alignments of KCTD11 sequence against proteinsequence databases revealed the presence of a closely relatedprotein within the KCTD family denoted as KCTD21 (Fig. 1A). Thetwo proteins exhibit an overall sequence identity of 40%, whichincreases to 60% when the comparison is restricted to the BTBdomain. KCTD21 is the only protein within the Swiss Prot databasethat exhibits a significant identity with KCTD11 C-terminus. Incontrast with lKCTD11/sKCTD11, KCTD21 has a standard AUGtranslation start and a complete BTB domain. A functional charac-terization of KCTD21 may provide intriguing information on therole and on the regulations of these two homologous proteins.

The characterization of lKTCD11 in solution clearly indicates thatthe protein is tetrameric. As expected, the protein oligomerization isdriven by POZ/BTB domain being the DBTB monomeric. The closesimilarity of lKTCD11 BTB with the corresponding domain of thepentameric KCTD5 [5] provides evidence for versatility of thesedomains in assuming different oligomeric states. The tetramericstate of the protein is fully compatible with the binding of Cul3.Indeed, a reliable three-dimensional model of (lKTCD11 BTB-CUL3)4could be generated by homology modeling techniques. The analysisof the tetramer indicates that a single Cul3 molecule interacts withtwo subunits of lKTCD11 BTB. The strongest interface involves thestrand B3 and the helix A4 of lKCTD11 and the helix of Cul3, whichalso stabilizes the 1:1 complexes of POZ/BTB domains and cullinshitherto characterized. It isworthmentioning that B3 andA4 are alsopresent in the sequence of KCTD11 short form. This may account forthe observed functionality of this form in the regulation of theHedgehog pathway in cells overexpressing sKCTD11. Indeed, it hasbeen demonstrated that sKCTD11 is able to inhibit medulloblastomagrowth by negatively regulating the Hedgehog pathway [11].Although efficient binding of POZ/BTB to Cul3 likely requires thepresence of the missing parts, it is important to note that the pres-ence in the short form sequence of these regions is per se sufficientfor the anchoring of the protein to the Cul3. This suggests that simplepeptide-based molecules opportunely designed on the basis of thelocal KCTD11 sequence might be able to modulate this interaction.

Starting from this (lKTCD11 BTB-CUL3)4 larger aggregateswere generated by including other partners for which structural



information is available (i.e., Rbx1 and E2). By following theprocedure reported in the methods section, a tetrameric lKTCD11BTB-Cul3-Rbx1-E2 assembly has been obtained (Fig. 11). Thismodel clearly indicates that the C-terminal of lKTCD11 BTB, towhich the DBTB portion of KTCD11 and the ligase substrate(HDAC1) are likely connected, is on the same side of the enzyme E2(Fig. 11) [28]. In other words, the organization of this giant complexis compatible with the proposed functional role of KTCD11.Intriguingly, the tetrameric nature of lKTCD11 BTB makes, from thestructural point of view, the traditionally complicated ubiquitin-protein ligases even more intricate. This consideration leads to thesuggestive hypothesis that the ubiquitination process of specificsubstrates, which involves KTCD11 and other related proteins,relies on assemblies endowed with a large structural complexity.The identification of the reasons that require the involvement inthese processes of non-monomeric ligases represents an importantchallenge for future investigations.

Acknowledgement

The authors thank the Italian MIUR for financial support (PRIN2007) and the Pasteur Institute-Cenci Bolognetti Foundation.

Appendix. Supplementary data

Supplementary data associated with this article can be found inthe online version, at doi:10.1016/j.biochi.2010.12.014.

References

[1] W. Li, M.H. Bengtson, A. Ulbrich, A. Matsuda, V.A. Reddy, A. Orth, S.K. Chanda,S. Batalov, C.A. Joazeiro, Genome-wide and functional annotation of human E3ubiquitin ligases identifies MULAN, a mitochondrial E3 that regulates theorganelle’s dynamics and signaling, PLoS One 3 (2008) e1487.

[2] V.J. Bardwell, R. Treisman, The POZ domain: a conserved protein-proteininteraction motif, Genes Dev. 8 (1994) 1664e1677.

[3] P.J. Stogios, G.G. Privé, The BACK domain in BTB-kelch proteins, Trends Bio-chem. Sci. 29 (2004) 634e637.

[4] P.J. Stogios, G.S. Downs, J.J. Jauhal, S.K. Nandra, G.G. Privé, Sequence andstructural analysis of BTB domain proteins, Genome Biol. 6 (2005) R82.

[5] I.S. Dementieva, V. Tereshko, Z.A. McCrossan, E. Solomaha, D. Araki, C. Xu,N. Grigorieff, S.A. Goldstein, Pentameric assembly of potassium channel tet-ramerization domain-containing protein 5, J. Mol. Biol. 387 (2009) 175e191.

[6] Y. Bayón, A.G. Trinidad, M.L. de la Puerta, M. Del Carmen Rodríguez, J. Bogetz,A. Rojas, J.M. De Pereda, S. Rahmouni, S. Williams, S. Matsuzawa, J.C. Reed,M.S. Crespo, T. Mustelin, A. Alonso, KCTD5, a putative substrate adaptor forcullin3 ubiquitin ligases, FEBS J. 275 (2008) 3900e3910.

[7] P. Van Bogaert, R. Azizieh, J. Désir, A. Aeby, L. De Meirleir, J.F. Laes,F. Christiaens, M.J. Abramowicz, Mutation of a potassium channel-relatedgene in progressive myoclonic epilepsy, Ann. Neurol. 61 (2007) 579e586.

[8] C.E. Elks, R.J. Loos, S.J. Sharp, C. Langenberg, S.M. Ring, N.J. Timpson, A.R. Ness,G. Davey Smith, D.B. Dunger, N.J. Wareham, K.K. Ong, Genetic markers of adultobesity risk are associated with greater early infancyweight gain and growth,PLoS Med. 7 (2010) e1000284.

[9] J. Schwenk, M. Metz, G. Zolles, R. Turecek, T. Fritzius, W. Bildl, E. Tarusawa,A. Kulik, A. Unger, K. Ivankova, R. Seddik, J.Y. Tiao, M. Rajalu, J. Trojanova,V. Rohde, M. Gassmann, U. Schulte, B. Fakler, B. Bettler, Native GABA(B)receptors are heteromultimers with a family of auxiliary subunits, Nature 465(2010) 231e235.


[10] T. Bartoi, K.T. Rigbolt, D. Du, G. Köhr, B. Blagoev, H.C. Kornau, GABAB receptorconstituents revealed by tandem affinity purification from transgenic mice,J. Biol. Chem. 285 (2010) 20625e20633.

[11] L. Di Marcotullio, E. Ferretti, E. De Smaele, I. Screpanti, A. Gulino, Sup-pressorsof hedgehog signaling: linking aberrant development of neuralprogenitors and tumorigenesis, Mol. Neurobiol. 34 (2006) 193e204.

[12] L. Di Marcotullio, E. Ferretti, E. De Smaele, B. Argenti, C. Mincione, F. Zazzeroni,R. Gallo, L. Masuelli, M. Napolitano, M. Maroder, A. Modesti, F. Giangaspero,I. Screpanti, E. Alesse, A. Gulino, REN(KCTD11) is a suppressor of hedgehogsignaling and is deleted in humanmedulloblastoma, Proc. Natl. Acad. Sci. U S A101 (2004) 10833e10838.

[13] E. De Smaele, L. Di Marcotullio, E. Ferretti, I. Screpanti, E. Alesse, A. Gulino,Chromosome 17p deletion in human medulloblastoma: a missing checkpointin the hedgehog pathway, Cell Cycle 3 (2004) 1263e1266.

[14] B. Argenti, R. Gallo, L. Di Marcotullio, E. Ferretti, M. Napolitano, S. Canterini,E. De Smaele, A. Greco, M.T. Fiorenza, M. Maroder, I. Screpanti, E. Alesse,A. Gulino, Hedgehog antagonist REN(KCTD11) regulates proliferation andapoptosis of developing granule cell progenitors, J. Neurosci. 25 (2005)8338e8346.

[15] E. Ferretti, E. De Smaele, L. Di Marcotullio, I. Screpanti, A. Gulino, Hedgehogcheckpoints in medulloblastoma: the chromosome 17p deletion paradigm,Trends Mol. Med. 11 (2005) 537e545.

[16] M.M. Mancarelli, F. Zazzeroni, L. Ciccocioppo, D. Capece, A. Po, S. Murgo, R. DiCamillo, C. Rinaldi, E. Ferretti, A. Gulino, E. Alesse, The tumor suppressor geneKCTD11REN is regulated by Sp1 and methylation and its expression isreduced in tumors, Mol. Cancer 9 (2010) 172.

[17] G. Canettieri, L. Di Marcotullio, A. Greco, S. Coni, L. Antonucci, P. Infante,L. Pietrosanti, E. De Smaele, E. Ferretti, E. Miele, M. Pelloni, G. De Simone,E.M. Pedone, P. Gallinari, A. Giorgi, C. Steinkühler, L. Vitagliano, C. Pedone,M.E. Schinin, I. Screpanti, A. Gulino, Histone deacetylase and Cullin3-REN(KCTD11) ubiquitin ligase interplay regulates hedgehog signalling through Gliacetylation, Nat. Cell Biol. 12 (2010) 132e142.

[18] R. Gallo, F. Zazzeroni, E. Alesse, C. Mincione, U. Borello, P. Buanne,R. D’Eugenio, A.R. Mackay, B. Argenti, R. Gradini, M.A. Russo, M. Maroder,G. Cossu, L. Frati, I. Screpanti, A. Gulino, REN: a novel, developmentallyregulated gene that promotes neural cell differentiation, J. Cell. Biol. 158(2002) 731e740.

[19] L. Di Marcotullio, E. Ferretti, A. Greco, E. De Smaele, A. Po, M.A. Sico,M. Alimandi, G. Giannini, M. Maroder, I. Screpanti, A. Gulino, Numb isa suppressor of hedgehog signalling and targets Gli1 for Itch-dependentubiquitination, Nat. Cell Biol. 8 (2006) 1415e1423.

[20] M.H. Nanao, W. Zhou, P.J. Pfaffinger, S. Choe, Determining the basis ofchannel-tetramerization specificity by x-ray crystallography and a sequence-comparison algorithm: family values (FamVal), Proc. Natl. Acad. Sci. U S A 100(2003) 8670e8675.

[21] T.A. Jones, J.Y. Zou, S.W. Cowan, M. Kjeldgaard, Improved methods for buildingprotein models in electron density maps and the location of errors in thesemodels, Acta Crystallogr. a47 (1991) 110e119.

[22] N. Zheng, B.A. Schulman, L. Song, J.J. Miller, P.D. Jeffrey, P. Wang, C. Chu,D.M. Koepp, S.J. Elledge, M. Pagano, R.C. Conaway, J.W. Conaway, J.W. Harper,N.P. Pavletich, Structure of the Cul1-Rbx1-Skp1-F boxSkp2 SCF ubiquitinligase complex, Nature (2002) 703e709.

[23] S. Angers, T. Li, X. Yi, M.J. MacCoss, R.T. Moon, N. Zheng, Molecular architec-ture and assembly of the DDB1-CUL4A ubiquitin ligase machinery, Nature 443(2006) 590e593.

[24] N. Zheng, P. Wang, P.D. Jeffrey, N.P. Pavletich, Structure of a c-Cbl-UbcH7complex: RING domain function in ubiquitin-protein ligases, Cell 102 (2000)533e539.

[25] M. Kozak, Initiation of translation in prokaryotes and eukaryotes, Gene 234(1999) 187e208.

[26] D.S. Peabody, Translation initiation at non-AUG triplets in mammalian cells,J. Biol. Chem. 264 (1989) 5031e5035.

[27] C. Touriol, S. Bornes, S. Bonnal, S. Audigier, H. Prast, A. Prats, S. Vagner,Generation of protein isoform diversity by alternative initiation of trans-lational at non-AUG codons, Biol. Cell. 95 (2003) 169e178.

[28] X. Tang, S. Orlicky, Z. Lin, A. Willems, D. Neculai, D. Ceccarelli, F. Mercurio,B.H. Shilton, F. Sicheri, M. Tyers, Suprafacial orientation of the SCFCdc4 dimeraccommodates multiple geometries for substrate ubiquitination, Cell 129(2007) 1165e1176.



Documents

Molecular organization of the cullin E3 ligase adaptor KCTD11