Biochimica et Biophysica Acta 1833 (2013) 1199–1211

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta

j ourna l homepage: www.e lsev ie r .com/ locate /bbamcr

The highly conserved, N-terminal (RXXX)8 motif of mouse Shadoomediates nuclear accumulation

E. Tóth a,b, P.I. Kulcsár a, E. Fodor a, F. Ayaydin c, L. Kalmár d, A.É. Borsy e, L. László b, E. Welker a,e,⁎a Institute of Biochemistry, Biological Research Centre, Hungarian Academy of Sciences, 62 Temesvari str, Szeged, H-6726, Hungaryb Department of Anatomy, Cell and Developmental Biology, Institute of Biology, Faculty of Science, Eötvös Loránd University, 1/A Pazmany P., H-1117, Hungaryc Cellular Imaging Laboratory, Biological Research Center, Hungarian Academy of Sciences, 62 Temesvari str, Szeged, H-6726, Hungaryd Institute of Enzymology, Research Centre for Natural Sciences, Hungarian Academy of Sciences, 29 Karolina str., Budapest, H-1113, Hungarye Institute of Molecular Pharmacology, Research Centre for Natural Sciences, Hungarian Academy of Sciences, 59-67 Pusztaszeri str., Budapest, H-1025, Hungary

Abbreviations: aa, amino acid; ATP, adenosine tripho(Cherry)2, tandemCherry dimer; (Cherry)4, tandemCherrsystem; DEG, 2-deoxy-D-glucose; DMEM, High-glucose DulECL, enhanced chemiluminescence; ER, endoplasmatic risothiocyanate labelled bovine serum albumin; GFP, enhaGPI anchor, glycosylphosphatidylinositol anchor; HD, hydlocalization signal; NRS, nuclear retention signal; PBS, pprion protein (C: cellular, Sc: scrapie); QKI-5, quaking-I 5Park Memorial Institute Medium; RT, room temperature;the prion protein); snls, short predicted nuclear localizat(ER-targeting); SV40, simian vacuolating virus 40; TTF-1YFP, enhanced yellow fluorescent protein⁎ Corresponding author at: Institute of Molecular Pha

Natural Sciences, HAS, 59-67, Pusztaszeri str., H-1025, Bu4158500.

E-mail addresses: t.esyter@enzim.hu (E. Tóth), kulcsfodor@brc.hu (E. Fodor), Ferhan@nucleus.szbk.u-szegedlkalmar@enzim.hu (L. Kalmár), borsy@enzim.hu (A.É. B(L. László), welker@brc.hu (E. Welker).

0167-4889/$ – see front matter © 2013 Published by Elhttp://dx.doi.org/10.1016/j.bbamcr.2013.01.020

a b s t r a c t

a r t i c l e i n f o

Article history:Received 1 November 2012Received in revised form 29 December 2012Accepted 15 January 2013Available online 27 January 2013

Keywords:ShadooPrion proteinRGG-box(RXXX)n motifNucleic acid bindingNuclear localization signal

The prion protein (PrP)—known for its central role in transmissible spongiform encephalopathies—has beenreported to possess two nuclear localization signals and localize in the nuclei of certain cells in various forms.Although these data are superficially contradictory, it is apparent that nuclear forms of the prion protein canbe found in cells in either the healthy or the diseased state. Here we report that Shadoo (Sho)—a member ofthe prion protein superfamily—is also found in the nucleus of several neural and non-neural cell lines asvisualized by using an YFP-Sho construct. This nuclear localization is mediated by the (25-61) fragment ofmouse Sho encompassing an (RXXX)8 motif. Bioinformatic analysis shows that the (RXXX)n motif (n=7-8)is a highly conserved and characteristic part of mammalian Shadoo proteins. Experiments to assess if Sho en-ters the nucleus by facilitated transport gave no decisive results: the inhibition of active processes that requireenergy in the cell, abolishes nuclear but not nucleolar accumulation. However, the (RXXX)8motif is not able tomediate the nuclear transport of large fusion constructs exceeding the size limit of the nuclear pore for passiveentry. Tracing the journey of various forms of Sho from translation to the nucleus and discerning the potentialnuclear function of PrP and Sho requires further studies.

© 2013 Published by Elsevier B.V.

1. Introduction

Transmissible spongiform encephalopathies are a group of rare,infectious, lethal neurodegenerative disorders in mammals, includingCreutzfeldt–Jakobdisease inhumans, bovine spongiformencephalopathyin cattle, scrapie in sheep and goat, and chronic wasting disease in mule

sphate; β-Gal, β-Galactosidase;y tetramer; CNS, central nervousbecco's Modified Eagle medium;eticulum; FITC-BSA, fluoresceinnced green fluorescent protein;rophobic domain; NLS, nuclearhosphate buffered saline; PrP,protein; RPMI medium, RoswellSho, Shadoo protein (shadow ofion signal; SS, signal sequence, thyroid transcription factor 1;

rmacology, Research Centre ofdapest, Hungary. Tel.: +36 30

pet@gmail.com (P.I. Kulcsár),.hu (F. Ayaydin),orsy), laszlo@cerberus.elte.hu

sevier B.V.

deer and elk [1,2]. The infectious agent contains an improperly foldedform of the host-encoded prion protein (PrPC). Conversion of PrPC intothe disease-associated isoform, termed PrPSc, is thought to be the primarypathogenic event, although the mechanisms by which the conversion toPrPSc causes disease are poorly understood [3]. Deciphering the physio-logical role of PrP may hold a key to this question; however, despiteintensive research over decades, the exact physiological function of theprion protein remains unclear.

PrP is expressed throughout the body, but in varying amounts;the highest levels occur in the central nervous system (CNS) andthe heart [4]. It has been alleged to associate with more than fiftyinteracting partner molecules [5,6] and its involvement in copperhomeostasis, neuroprotection and signaling, memory, proliferation,differentiation, apoptosis, myelination, circadian rhythm, cell adhe-sion, synaptic activity, etc. [5,7,8] has been proposed. However, it isnot clear which, if any, of these functions can explain the conservedoccurrence of PrP across many species including mammals, birds,fishes and amphibians. To attribute such a function to PrP, Shadoo,the newestmember of the prion superfamilymayprovide somehint [9].

The Shadoo protein (Shadow of Prion Protein, SPRN gene), wasdiscovered in silico by Premzl et al. in 2003. They noted that the struc-ture of the predicted Shadoo protein loosely resembles the flexiblydisordered N-terminal part of the prion protein. In particular there

1200 E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

are two features which are present in both Shadoo's and PrP's nativelyunstructured N-terminus: i) a series of tandem repeats of short se-quences with positively charged residues (from aa. 26 to aa. 48 inmouse Shadoo, termed XXRG tetrarepeats, and from aa. 51 to aa. 90 inmouse PrP); ii) and a hydrophobic domain (HD) (from aa. 63 to aa. 82inmouse Shadoo and fromaa. 111 to aa. 130 inmouse PrP). Thepredictedsubcellular localization of both proteins is the same—i.e. both possessglycophosphatidyl-inositol (GPI) anchor signal sequences and one ortwoN-glycosylation sites that are suggestive of their release to the secre-tory pathways—laying further emphasis on their similarity [9,10].

The presence of the GPI anchor and N-glycosylation were experi-mentally demonstrated on recombinant Sho constructs with orwithout a tag as well as on the endogenous protein in brainhomogenates [11,12].

Besides these structural similarities, several other analogies havebeen identified between PrP and Sho.

Their expression reaches its highest levels in the CNS [4,9,10,12].Endoproteolysis close to the center of PrP generates C1 and N1fragments which likely have important biological properties [13,14].C-terminal fragment of Sho resembling PrP's C1 is readily detectedin brain samples and in the cell-associated fraction from lysed cul-tured cells [12]. Besides these data, PrP endoproteolysed adjacent tothe GPI anchor that releases a nearly full length form of the moleculewhich can be reproduced in vitro by ADAM 10 [15,16]. Likewise, aglycosylated form of Sho is easily detectable within the conditionedmedium of Sho expressing cells [17].

Functional analogies have also been confirmed: the toxic effect ofthe third prion protein superfamily member, Doppel, and internal dele-tion mutant forms of PrP could be counteracted by the co-expressionof either PrP or Sho in transfected granule cell neurons. This rescueeffect was abolished by the deletion of the HD region [12]. Tatzelt andco-workers reached to a similar conclusion investigating the rescueeffect of the expression of PrP or Sho on kainate or glutamate inducedapoptosis in SH-SY5Y cells. These rescue effects could also be abolishedby the deletion of the HD region [18,19]. In addition, some experimentson PRNP-knockoutmice suggest that the two proteins have overlappingfunctions and can complement each other [20,21], although this issue israther controversial [22,23].

To further strengthen the argument for the similarity betweenthese two proteins, formation of homodimers of both PrP and Shohave been demonstrated when expressed in cell culture. These asso-ciations are facilitated by the HD region in the case of both proteins.Furthermore the formation of heterodimers between PrP and Shomediated also by the HD region has also been shown [6,18,19,24].In addition to this protein-protein interaction, more than a dozenproteins outside the prion protein superfamily have been identifiedthat might interact with both PrP and Sho [6,25]. The similarity be-tween their binding partners appears to go beyond protein-proteininteractions: nucleic acids have also been shown to bind PrP [26–28]and it has also been proposed that they bind Sho as well [29]. Sho'stetrarepeats include an RGG box motif which is present in RNA-binding proteins involved in various aspects of RNA processing as wellas mediating interactions to other proteins [29].

An overlap in the subcellular distribution pattern of minor pop-ulations of PrP and Sho, would further buttresses the case for thefunctional analogy between these two proteins. PrP has been shownto have two short sequences, referred to as NLS1 and NLS2 that,under some conditions, function as nuclear localization signals (NLS)[30,31]. Various forms of PrP, such as PrPSc [32], truncated PrPsrepresenting prion proteins with familial mutations [33], or PrPexpressed without endoplasmic reticulum (ER)-targeting signal se-quences and GPI anchor by genetic engineering [34] or by alternativetranslation initiation [35,36] have been demonstrated to appear in boththe cytosol and nucleus. In addition, cellular PrP has been reported inthe nucleus of proliferating epithelial cells [37] as well as in endocrineand neuronal cells [38]. In line with these observations, Sho has a

potential arginine-methylation site that is a common post-translationalmodification in RGG-box domains and occurs on proteins that arefound in the nucleocytoplasmic compartment [29]. More interestingly,the endogenous Shadoo appears in the cell bodies of some of the hypo-thalamic neurons and it was found in nuclear fractions of both brainand cultured cells [22].

Here we observed a fluorescent tagged form of Sho in the nucleusof several cell lines, and found nuclear localization signals in the se-quence of Sho by PredictNLS [39] and by NLStradamus [40]. Thesehave prompted us to experimentally validate whether any of thesesequences functions as an NLS: which in turn would not only addone more item to the growing list of similarities between PrP andSho, but might also point to a common function of PrP and Sho inthe nucleocytoplasmic compartment.

2. Materials and methods

Chemicals: The chemicals usedwere purchased fromSigma-Aldrich,unless otherwise stated.

2.1. Plasmid constructions

To generate YFP-, and GFP-chimeric Shadoo constructs, generallythe N-terminal NheI and AgeI restriction sites (to insert ER-targetingsignals or Shadoo core sequences), and the C-terminal EcoRI and BamHIrestriction sites (to insert Shadoo, GPI-anchor signal, potential NLSsequences) were used in pEYFP-C1 and pEGFP-C1 (Clontech) respec-tively. GFP-β-Gal andNLSx3SV40-GFP-β-Galwere kind gifts of Y. Lazebnik.To generate Sho-GFP-β-Gal and (25-61)Sho-GFP-β-Gal XhoI and XbaIrestriction sites were used. More details for plasmid constructions areprovided in the supplementary documents (Supp. Table 1). Restrictionenzymeswere brought from Fermentas. The deletionmutants were gen-erated by Quickchange Mutagenesis (Stratagene). All constructs wereverifiedby sequencing before beingused. The synthesis of the oligonucle-otides used and the sequencing of the DNA constructs were carried outby Microsynth AG (http://microsynth.ch/).

2.2. Cell culture

Cells (HeLa, SH-SY5Y, Cos-7, Zpl2-1, Zw3-5, Hpl3-4 and Hw13-3)were grown at 37 °C in a humidified atmosphere of 5% CO2 in high-glucose Dulbecco's Modified Eagle medium (DMEM, Lonza) sup-plemented with 10% heat inactivated fetal bovine serum (Lonza) andcontaining 100 units/ml penicillin and 100 μg/ml streptomycin (Lonza).

K562 cellswere cultured in Roswell ParkMemorial InstituteMedium(RPMI)-1640 medium supplemented with 2.0 g/l sodium-bicarbonate,10% fetal bovine serum (Lonza) and 1% GlutaMAX (Gibco).

Zpl2-1, Zw3-5 [41], Hpl3-4 and Hw13-3 [42] are immortalized,hippocampal cell lines, they were kind gifts of Y. S. Kim and T. Onoderarespectively.

2.3. Transfection

Cells were cultured on Labtek II 8 well chambers (Labtek) or 6well plates (Greiner Bio-One), seeded a day before the transfectionat a density of 2×104 cells/well or 3×105 cells/well respectively.The next day, at around 50% confluence, cells were transientlytransfected with plasmid constructs using Turbofect (Fermentas),briefly as follows: DMEM was changed on the cells to fresh one,containing 10% FBS. Typically 500 ng/4000 μg plasmid and 0.5 μl/4 μlTurbofect wasmixed in 50 μl/400 μl serum free DMEM and themixturewas incubated for 40 min at RT prior adding to cells. Transfectionmedi-umwas changed to fresh supplemented DMEM after 3 h of incubation.

1201E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

2.4. Chilling and deoxy-glucose treatment

Chilling and deoxy-glucose treatment were performed as describedby Chatterjee and Stochaj [43]. Briefly, cellswerewashedwith PBS, thenchilled for 2 h at 4 °C followed by 2 h on ice. For 2-deoxy-D-glucose(DEG) treatment, after washing with PBS, cells were incubated in PBScontaining 10 μM DEG and 10 μM NaN3 (Fluka) for 1 h at RT.

2.5. Confocal microscopy

Cells were fixed using 4% paraformaldehyde in PBS for 10 min, thecells were then washed with PBS containing Hoechst33342 (5 μM)(Invitrogen).

Confocal laser scanning microscopy was performed using anOlympus FV1000 confocal laser scanning microscope (Olympus LifeScience Europe GmbH, Hamburg, Germany). Microscope settings aresummarized in the supplementary documents. Picture capturing andanalysis were performed using Fluoroview software (Olympus).

About 20 photos were made to count 100–300 transfected cellsfrom each well. All experiments were made with at least 3 replicatesfor every construct.

2.6. Western blotting

HeLa cells were cultured on 6-well plate and transfected asdescribed above. The day after transfection cells were washed withice-cold PBS, then scraped, centrifuged at 110 rcf for 5 min at 4 °C.Pellet was resuspended in Harlow buffer (50 mM Hepes pH 7.5;0.2 mM EDTA; 10 mM NaF; 0.5% NP40; 250 mM NaCl; ProteinaseInhibitor Cocktail 1:100). Cell lysates in selected samples were di-gested by PNGase F (Fermentas) for 2 h at 37 °C.

Before SDS gel loading, samples were boiled in Protein Loading Dyefor 5 min at 95 °C then kept at 4 °C for 5 min. Proteins were separatedby SDS-PAGE using 10 or 12% polyacrylamide gels and transferred toPVDF membrane. Membranes were blocked by 5% non-fat milkfor 2 h. Blots were incubated overnight with primary antibody(anti-GFP/YFP 1:3,500, Central European Biosystems), and incubatedwith anti-rabbit HRP 1:40,000 (Pierce). The signal from detected pro-teins was visualized by ECL (Immobilon Western HRP SubstrateLuminol Reagent, Millipore).

2.7. Bioinformatics

Sho protein sequences were obtained from the UniProtKnowledgebase, using “shadoo” and “shadow” search terms. Our finaldataset contains Sho proteins from 33 different vertebrate species(UniProt ACs: A2BDG9, Q5TJB9, Q5BIV6, A2BDJ8, Q1JPW9, A2BDJ9,A2BDG5, A2BDG0, A2BDG4, C6ETC7, C6ETD1, A0RZB4, C6ETD7, C6ETD2,C6ETC9, C6ETC6, C7DLK9, C6ETD0, C6ETC8, A6XN32, C5H878, A2BDG2,A2BDJ4, A2BDG7, Q8BWU1, C6ETC5, Q5BIV7, A2BDG8, A2BDJ5,A2BDG6, A2BDH0, A2BDG1, Q5BIV9). Initial multiple alignmenton the dataset was performed using Clustal Omega software(pmid: 21988835). For further refinements and visualization weused Jalview software (pmid: 19151095), sequence logos were createdby Weblogo online software (http://weblogo.berkeley.edu/logo.cgi,pmid: 15173120). We used different searching techniques to findhomologies to the (RXXX)8 motif region, (i) BLAST search on theUniProtKB proteins, (ii) pattern based search using PERL scripts,(iii) pattern based search combined with residue composition filter-ing using PERL scripts, (iv) using profile hidden Markov models(HMMER software package, pmid: 22039361).

2.8. Statistical analysis

In each experiment, the subcellular localization of the constructswas scored in at least 100 transfected cells in 3 replicates.

Subcellular localizations were scored as follows:

• exclusive nuclear: nuclear enchancement with nearly no cytosolicsignal

• enchanced nuclear: nuclear enhancement with cytosolic signal• nucleus: exclusive nuclear+enhanced nuclear signal• homogenous: nuclear and cytosolic intensities are equal• plasma membrane

The normality of the distributions was tested with Kolmogorov–Smirnov test; since the distribution of the data was normal, Student'st-test for independent samples was used to check their significance(SPSS 9.0 Statistica Program).

p value: * 0.01bpb0.05; ** 0.001bpb0.01; *** pb0.001.

3. Results and discussion

3.1. Dual subcellular localization: YFP-tagged Shadoo in the secretorypathway and in the nucleus

To follow the traffic of Sho, fluorescent protein [enhanced yellowfluorescent protein (YFP)]was fused to either the N-terminal (YFP-Sho)or the C-terminal (Sho-YFP) side of the Shadoo protein downstreamfrom the ER-targeting- or preceding the GPI-signal sequence of Sho,respectively (Fig. 1A). The original methionine of YFP in YFP-Sho waschanged to leucine in order to eliminate the expression of a signalpeptide-less protein form that is observed in several cell lines (Nyesteand Welker, unpublished results).

Western blot analysis of cell lysates from transfected HeLa cells in-dicates that a full length fusion protein is detectable in both YFP-Shoand Sho-YFP-expressing cells. It should be noted that Sho-YFP showedbands with lower mobility and seems to be more prone to proteolyticcleavage as observed consistently during repeated experiments(Fig. 1B). When analyzed by laser-scanning confocal microscopy,cells expressing Sho-YFP exhibited a localization pattern which isexpected for a GPI-anchored protein: the YFP signal appeared pre-dominantly in the plasma membrane and in intracellular vesiclesthat most likely correspond to ER and Golgi-like structures. Thesame localization was observed for the YFP-GPI control protein. Bycontrast, YFP-Sho showed a dual localization; beside the plasmamembrane and ER/Golgi-like structures we observed a nuclear signalin a subset of the cells (Fig. 1B). Interestingly, while some of thesecells exhibited both membrane and nuclear signals, other showedonly nuclear or nuclear/cytosolic ones.

We examined the expression of these constructs exploring severalneural (human: SH-SY5Y, mouse: Zpl 2-1, ZW 3-5, Hpl 3-4, HW 13-3)and non-neural (human: HeLa, Cos-7, K562) mammalian cell lines,and observed the same localizations for these constructs as in HeLacells (Supp. Fig. 1; data not shown in the case of Sho-YFP andYFP-GPI). However, the percentage of the cells showing nuclear accu-mulation in the case of YFP-Sho varied among the cell lines, beinghighest in HeLa cells (about 30% of the cells).

In order to facilitate the identification of the segment that isresponsible for Shadoo's nuclear accumulation, signal peptide-less(ΔSS) Shadoo constructs were also generated to direct these fusionproteins straight to the cytosol, and thus, enhance their nuclear accu-mulation (Fig. 2A). This strategy is based on our observation thatYFP-Sho's nuclear distribution appears as a diffuse signal throughoutthe nucleus with no specific localization to membrane-like structures,and on the report that the cytosolic form of PrP accumulates in thenucleus of N2a cells [34].

Interestingly, both ΔSS-YFP-Sho and ΔSS-Sho-YFP exhibited nearexclusive nuclear localization in the cell lines investigated (Fig. 2B).The presence or absence of the GPI signal sequences did not affectthe localization of the ER-targeting signal-less constructs (data notshown). Summing up these results, the position of the YFP tag appar-ently influenced the subcellular distribution of the fusion proteins

YFP-ShoA

Sho-YFP YFP-GPI

SS YFP Shadoo SS SSYFP YFPShadoo

BYFP-Sho Sho-YFP YFP-GPI

28

36

+ -PNGase F

7295

55

+ - + -

kDa

28

17

Fig. 1. Subcellular localization and immune blot pattern of YFP-Sho, Sho-YFP and YFP-GPI. (A) Schematic structures of YFP-Sho, Sho-YFP and YFP-GPI fusion proteins. SS stands forthe ER-targeting signal peptide; GPI-anchor signal sequence is represented by a small light blue box at the C-terminus. Full length (−) and PNGase F digested (+) forms of YFP-Sho,Sho-YFP and YFP-GPI detected byWestern blot using 12% SDS-PAGE and anti-GFP/YFP antibody dilution of 1:3,500. Sho-YFP consistently showed slightly slower mobility bands andseemed to be more prone to proteolytic cleavage than YFP-Sho in repeated experiments (B) Confocal microscopy images of YFP-Sho, Sho-YFP and YFP-GPI expressing HeLa cells.Only YFP-Sho exhibits dual localization (plasma membrane and nuclear). Signals from YFP or Hoechst 33342 dye are shown in green or blue colours, respectively, and the trans-mitted light images of the corresponding areas are presented as gray pictures. Scalebar represents 10 μm.

1202 E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

(Fig. 1B). While Sho-YFP exhibited a distribution that is consistent witha protein committed to the secretory pathway, YFP-Sho demonstrated adual, plasma membrane and nuclear localization (Fig. 1B, Supp. Fig. 1).Since in the absence of the signal peptide both fusion proteins accumu-lated in the nucleus, the position of the YFP tag (N- or C-terminal to Sho)does not affect directly its nuclear targeting (Fig. 2B). Rather, the loca-tion of the YFP tag seems to be associated with the ability of the fusionconstructs—that are supposedly targeted to the secretory pathway—toenter the cytosol. One possible explanation for this difference is that

ΔΔ SS-YFP-ShoA

Δ SS-Sho-YFP

YFP Shadoo YFPShadoo

36

+PNGase F

55

+

kDa

72

28

95

17

- -

Fig. 2. Subcellular localization and immune blot pattern of ΔSS-YFP-Sho and ΔSS-Sho-YFP. (Athe absence of the ER-targeting signal peptide; GPI-anchor signal sequence is represented(+) forms of ΔSS-YFP-Sho and ΔSS-Sho-YFP detected by Western blot analysis (12% SDS-PAmobility bands than ΔSS-YFP-Sho in repeated experiments; (B) Confocal microscopy imaΔSS-Sho-YFP exhibits near exclusive nuclear localization. The presence or absence of the GP(data not shown). Signals from YFP or Hoechst 33342 dye are shown in green or blue cpresented as gray pictures. Scalebar represents 10 μm.

Sho-YFP may be more prone to proteolytic degradation than YFP-Sho (see Fig. 1B). Alternatively, a specific proteolytic cleavage onShadoo may separate its potential NLS from the fluorescent protein.Indeed, such proteolytic cleavages have been reported to take placein the case of endogenous Shadoo protein [12]. Experiments to testthis explanation by tagging this proposed N-terminal fragmentwith an unrelated epitope were not conclusive (data not shown).However, when an NLS was inserted between Sho and YFP, followingthe putative protease cleavage site, the YFP signal—in agreement

BΔSS-ShoYFPΔSS-YFP-Sho

) Schematic structures of ΔSS-YFP-Sho and ΔSS-Sho-YFP fusion proteins. ΔSS indicatesby a small light blue box at the C-terminus. Full length (−) and PNGase F digestedGE, anti-GFP/YFP antibody dilution 1:3,500). ΔSS-Sho-YFP consistently showed slowerges of ΔSS-YFP-Sho and ΔSS-Sho-YFP expressing HeLa cells. Both ΔSS-YFP-Sho andI signal sequences did not affect the localization of the ER-signal peptide-less constructsolours, respectively, and the transmitted light images of the corresponding areas are

1203E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

with this explanation—appeared in the nucleus (data not shown).The clarification of this issue needs further research.

3.2. In silico identification of NLS sequences in Sho

To identify the segment of Sho that is responsible for its nuclearaccumulation we first analyzed its amino acid sequences by severalsoftware programs designed for the prediction of nuclear localizationsignals.

PredictNLS (available through the PredictProtein server, http://www.predictprotein.org/ [39]) identified a stretch in the mouse Shadooprotein from amino acid 28 (arginine) to amino acid 37 (glycine) thatcorresponds to one of the 214 NLS sequences in the database with aconsensus sequence: R G/A X(0-2) G/A R G/A X G/A R G/A. This pre-dicted NLS (short NLS; snls) is a GR-motif type NLS, and was describedin the large fibroblast growth factor (FGF)-2 isoform [44,45]. Inthemouse Shadoo sequence, NLStaradamus [40] was also able to identi-fy an NLS stretch, which is longer but totally overlapping withthe PredictNLS's one: from amino acid 26 (glycine) to amino acid 53(alanine).

However, PSORT II [46] and PROSITE [47] found no NLS sequencesin Shadoo.

This information was used for designing experiments in Sections3.3.1 and 3.3.2 aimed at identifying the NLS of Sho.

3.3. Experimental identification of a segment in Sho that is responsible forits nuclear targeting

The identification of NLS(s) that direct a protein to the nuclearcompartment of the cells could be a tricky task involving several ca-veats. Deletion or mutation of a sequence may diminish nuclear local-ization even when the sequence is not an NLS or not a complete NLS,by affecting the traffic, maturation or structure of the protein or theaccessibility of the real NLS. Additionally, proteins frequently havemore than one NLS. Thus, a putative NLS may be effective in directinganother protein to the nucleus although its deletion or mutation hasno apparent impact on the nuclear transport of the protein of interestdue to the possible masking effect by another NLS present in the pro-tein [48,49].

Testing a sequence for its ability to direct a non-related protein tothe nucleus is another complementary approach having its own caveats.An NLS may not be effective in the context of a non-related protein ifthat protein interferes with the accessibility of the NLS. Conversely, asequence may be effective in facilitating nuclear localization when at-tached to another protein, although it does not function in the contextof its own original protein because its accessibility to the nuclear trans-port machinery is restricted.

The complexity of this task became apparent when one of theselected positive controls, the extreme N-terminal NLS of the thyroidtranscription factor 1 (NLSTTF1) did not function when attached to aGFP (data not shown). The sequence of NLSTTF1 was identified onlyby deletions [50] and mutations [51] in the cited references. Whichof the possible pitfalls discussed above caused this discrepancy ispresently not clear. In further experiments the NLS of quaking I-5protein (qki5) and Simian Vacuolating Virus 40 (SV40) was used asa positive control.

3.3.1. Deletion constructsTo identify the segment of the Shadoo protein that is responsible for

the nuclear targeting of the YFP-Sho construct, the sequence betweenaa. 28 and 38, identified by PredictNLS (snls) (see 3.2) was deleted(YFP-ShoΔsnls). In YFP-ShoΔRGG the deletion was extended to includethe full RGG box (from aa. 28 to aa. 41). The RGG-box, the arginineand glycine rich N-terminal segment of Sho might also result nuclearlocalization by virtue of its reported nucleic acid [52] and/or proteinbinding capability [53]. In a third construct, we extended the deletion

further to the whole positively charged N-terminus of Sho (from aa.25 to aa. 61), that spans the long NLS predicted by NLStaradamus(YFP-ShoΔ(25–61)). The same segments were deleted in the signalpeptide-less protein ΔSS-YFP-Sho as well.

In YFP-Sho, increasing the length of the deletionswas correlatedwitha decrease in the number of cells showing nuclear localization that be-came significant (p=0,037, *) with the deletion of the (25–61) segmentthat reduced it from about 30% to 0% (Supp. Fig. 2). Deleting the samesegments in the signal peptide-less variant (ΔSS-YFP-Sho), the exclusivenuclear localization was practically abolished with all three constructs,although the deletion of the snls or the RGG-box still resulted in about50% of the transfected cells showing enhanced nuclear signal. The en-hanced nuclear signal decreased to the control GFP level only whenthe whole (25–61) region of Sho was deleted (Fig. 3). Such results aremore straightforward to interpret for a cytosolic protein. However, Shois supposed to be synthesized on the rough ER and a deletion could in-terfere with its ability to get to the cytosol from (or instead of) the secre-tory pathway, and thus, could indirectly abolish its nuclear transport.Since the deletion of the (25–61) fragment of Sho causes the same effectin YFP-Sho and in the ΔSS-YFP-Sho construct that is—in the absence of asignal sequence—directed directly into the cytosol, this possibility can beruled out.

These results therefore suggest that the segment which is respon-sible for the nuclear targeting of Sho is confined within the (25-61)region of Sho.

3.3.2. Potential NLS sequences attached to GFP reveal that snls and theRGG motif do not function as an NLS

To see which, if any, of these segments can function independentlyas an NLS, all three deleted segments [snls, RGG-box and (25–61)]were also fused to the C-terminal of GFP, generating GFP-snlsSho,GFP-RGGSho and GFP-(25–61)Sho. As a control, NLS of the quakingI-5 protein [54] was also attached to GFP (GFP-NLSqki5). The vectorswere transfected toHeLa cells alongwithGFP andΔSS-YFP-Sho. Near ex-clusive nuclear localization were observed only with GFP-(25–61)Sho,GFP-NLSqki5 and ΔSS-YFP-Sho. Little to no difference to the control GFPwas observable with GFP-snlsSho, GFP-RGGSho (Fig. 4).

One possible pitfall in the interpretation of these experiments(as was pointed out previously) is that the snls or RGG-box frag-ments may fail to function in these constructs due to their positionrelative to GFP or due to their spacing from GFP. However, aa. 25 to61 in GFP-(25–61)Sho construct which also contains the snls andthe RGG-box sequences in the same orientation and spacing, doesfunction as a nuclear targeting signal. These results indicate thatthe snls and the RGG-box of Sho are insufficient to direct the fusionprotein to the nucleus, although they are likely to be part of the com-plete nuclear localization signal of Sho which is contained in the(25-61) segment.

3.3.3. Role of snls and RGG-box in the nuclear targeting of YFP-ShoSince the second half (last 24 amino acids) of the (25–61) seg-

ment alone is able to cause YFP to accumulate in the nucleus inof about 50% of the cells even in the absence of the first part(ΔSS-YFP-ShoΔsnls in Fig. 3), it is possible that only this second partis involved directly in the nuclear targeting of Sho. Thus the onlyfunction of the first part (snls or RGG-box) in these experimentsmay be to keep the second part at an appropriate distance from GFPin constructs where the (25–61) fragment or the full length proteinis attached. We addressed this question by changing the position ofSho and of all the three Sho deletion mutants from the C-terminalto the N-terminal end of YFP (Fig. 5). Fig. 5 shows that the last 20 or24 amino acids of (25–61) in the absence of snls or the RGG-box, re-spectively, have a reduced ability to drive the fusion protein to thenucleus. The deletions greatly diminish Sho's nuclear localization, al-though the distance and orientation of the second half of the (25–61)segment was not altered in these constructs indicating that the

A100

exclusive nuclear enhanced nuclear

** *60

80n

um

ber

of

cells

(%

)

*** ***

20

40

B ΔΔsnls ΔRGG Δ(25-61)

YFP Shadoo YFP YFPShadoo YFPShadoo Shadoo YFP

0

ΔSS-YFP-Sho ΔSS-YFP-ShoΔsnls ΔSS-YFP-ShoΔRGG ΔSS-YFP-ShoΔ(25-61) YFP

Fig. 3. Subcellular localization of the deletion constructs of ΔSS-YFP-Sho by confocal microscopy. (A) Nuclear localization of ΔSS-YFP-Sho deletion constructs in HeLa cells. At leastone hundred cells were counted in each of three replicates; * 0.05>p>0.01, ** 0.01>p>0.001, *** 0.001>p. Standard deviations and significances were calculated for total nuclearlocalizations (exclusive nuclear+enhanced nuclear) and were compared to that of ΔSS-YFP-Sho; (B) Schematic structures of ΔSS-YFP-Sho, ΔSS-YFP-ShoΔsnls, ΔSS-YFP-ShoΔRGG,ΔSS-YFP-ShoΔ(25–61) and YFP and representative confocal microscopy images of HeLa cells transfected with the above mentioned constructs. The nuclear signal decreased to thecontrol YFP level only when the whole (25–61) region of Sho was deleted. (Green: enhanced YFP, blue: Hoechst 33342, gray: transmitted light). Scalebar represents 10 μm.

1204 E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

function of the first part is more than being just a spacer to the fluo-rescent protein (Fig. 5). These data point to the whole (25–61) seg-ment of Sho as being the element responsible for Sho's full nucleartargeting.

It is worth to note that in this region (between aa. 28 and aa. 58) theprecise positions of arginine residues are conspicuous. Between two ar-ginines, generally three small, predominatly hydrophobe amino acids[(RXXX)n motif, where X is usually glycine, alanine, valine, serine,Supp. Fig. 3) hold the distance. Negative amino acids are completelymissing from this region. Eight of these tetrarepeats [(RXXX)8 motif]could be observed in the N-terminal part of Shadoo in mouse.

3.4. The nature of the nuclear-targeting (RXXX)8 motif of Sho

The NLS of Sho we have identified resembles to the substrateof transportin (Karyopherin-β2) that binds natively unstructuredsequences, about 30-amino-acid-long with an overall positive

(or hydrophobic) character and an R/K/H X(2-5) P Y sequence in theirC-terminal region. The (RXXX)8 motif of mouse Sho seems to fulfilthese requirements except that there is a RPAPRY sequence (aa. 51–56)instead of a RPAPY terminus. The individual amino acids in aKaryopherin-β2 substrate make only a small contribution to thebinding energy and the sequence as a whole is relatively insensi-tive to mutations except for the PY terminus mutation of whichabolishes the binding to Karyopherin-β2 [55]. In order to see ifSho has a similarly functioning NLS, the contribution of amino acids52–56 of Sho to the nuclear transport of the fusion constructs wasexamined by substituting this sequence with 5 consecutive alanines.The mutations cause no alteration in the nuclear targeting of YFP-Sho, ΔSS-YFP-Sho or GFP-(25–61)Sho (data not shown) and give nosupport to the putative involvement of this element in nucleartargeting despite the apparent sequence similarity.

The other most prevalent feature of the (RXXX)8 motif is theperiodically reoccurring arginines. To test if these arginines play a

A*** ******100

exlusive nuclear enhanced nuclear

60

80

nu

mb

er o

f ce

lls (

%)

20

40

BGFP GFP GFP GFP YFP GFPsnlsSho RGGSho (25-61)Sho Shadoo NLSqki5

0

GFP GPF-NLS GFP-RGG GFP-(25-61) ΔΔSS-YFP-Sho GFP-NLSqki5

Fig. 4. Subcellular localization of potential NLS fragments of Sho in fusion to GFP by confocal microscopy. (A) Nuclear localization of GFP fused potential NLS fragments or of GFP alone inHeLa cells. At least one hundred cells were counted in each of three replicates; * 0.05>p>0.01, ** 0.01>p>0.001, *** 0.001>p. Standard deviations and significanceswere calculated fortotal nuclear localizations (exclusive nuclear+enhanced nuclear) and were compared to that of GFP; (B) Schematic structures of GFP, GFP-snlsSho, GFP-RGGSho, GFP-(25–61)Sho,ΔSS-YFP-Sho and GFP-NLSqki5 (NLSqki5 in red box) and representative confocal microscopy images of HeLa cells transfected with the indicated constructs. Only the (25–61) fragmentof Sho was able to direct the GFP fusion protein to the nucleus. (Green: enhanced YFP, blue: Hoechst 33342, gray: transmitted light). Scalebar represents 10 μm.

1205E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

pivotal role in nuclear targeting of Sho, they were mutated to gluta-mines [GFP-(25–61)RQSho]. The mutations eliminated the nuclear andnucleolar accumulation of GFP-(25–61)Sho resulting in a homogenous,GFP-like localization in all of the cells (Supp. Fig. 4) suggesting thatthe positively-charged arginines play a crucial role in Shadoo's nuclearenhancement.

3.5. Bioinformatic analysis shows high evolutionary conservation to thecharacteristic (RXXX)n motif of Sho

We observed high sequence conservation and high residue com-position conservation in the entire (RXXX)n (n=7–8) motif amongthe analyzed mammalian Shadoo proteins. Eight of this tetrarepeat[(RXXX)8] could be observed in the N-terminal part of Shadoo inthe known rodent and homininid Sho sequences; seven repeats[(RXXX)7] could be observed in the known ruminant sequences, inchicken's and in Western clawed frog's Sho (Supp. Fig. 3). The multi-ple alignment data we present show that the first part (RGG box)of the motif is the most conserved part of the whole protein andaccording to the sequence logo, the characteristic basic amino acidbias and residue composition is well conserved in the entire motif

(Fig. 6). The previously known functional modules of Sho (ER-targetingsignal sequence, repeat region, RGG box, HD region, GPI signalsequence) show the same or lower conservation compared to theidentified (RXXX)n motif. We found, that (RXXX)n is not only awell conserved region in Shadoo proteins, but also an obligatorycharacteristic of these proteins. Using different search strategies,we were not able to find any other protein with a region similar tothe entire motif. If we separated the two parts, and search for simi-larities to each of them, we found a few proteins with high similarityto the RGG-box part of the (RXXX)n motif, but the second part re-mains completely unique. The eukaryotic proteins we found homol-ogous for the RGG-box part are annotated as having nuclear and/ornucleolar localization and function (e.g. SwissProt IDs: P83214,Q6Z1C0, D1Z6B1).

3.6. Transport through the nuclear pore complex driven by the (25–61)fragment of Sho

Some smaller nuclear proteins (cca. below 60 kDa) can getthrough the nuclear pore by diffusion (passive entry) and accumulatethere by binding to a partner. The protein segment involved in the

A (25-61)

RGG: 28-41snls: 28-37

YFP Shadoo GPIX(25-61)

Shadoo

RGG: 28-41snls: 28-37

GPIX

100

exclusive nuclear enhanced nuclearB

60

80

nu

mb

er o

f ce

lls (

%)

*****

20

40

ΔΔsnls ΔRGG Δ(25-61)

Shadoo YFP Shadoo YFP Shadoo YFP Shadoo YFP YFP

******

0

ΔSS-Sho-YFP ΔSSS-hoΔsnls-YFP ΔSS-ShoΔRGG-YFP ΔSS-ShoΔ(25-61)-YFP YFP

YFP

Fig. 5. Deletion of potential NLSs in ΔSS-Sho-YFP: position of the deletions and the effect on the subcellular distribution of ΔSS-YFP-Sho. (A) Comparison of ΔSS-YFP-Sho deletionconstructs and ΔSS-Sho-YFP deletion constructs. In the case of ΔSS-YFP-Sho (upper scheme), the deletion alters the distance between YFP and the rest of Shadoo, therefore therelative position of YFP to Sho is changed that might be able to interfere with the nuclear targeting of Sho. In the case of ΔSS-Sho-YFP (lower scheme), the deletion occurs atthe N-terminal, leaving the distance between YFP and the rest of Sho untouched. The red cross indicates the deleted region. The orange–coloured box indicates the remainingpart of the (25–61) fragment of Sho after snls or RGG box deletion. (B) Nuclear localization of ΔSS-Sho-YFP deletion constructs in HeLa cells. At least one hundred cells were countedin each of three replicates; * 0.05>p>0.01, ** 0.01>p>0.001, *** 0.001>p. Standard deviations and significances were calculated for total nuclear localizations (exclusivenuclear+enhanced nuclear) and were compared to that of ΔSS-Sho-YFP. The deletions greatly diminish Sho's nuclear localization, although the distance and orientation of the sec-ond half of the (25–61) segment was not altered in these constructs indicating that the function of the first part is more than just being a spacer to the fluorescent protein. Schematicstructures of ΔSS-Sho-YFP, ΔSS-Sho(Δsnls)-YFP, ΔSS-Sho(ΔRGG)-YFP, ΔSS-ShoΔ(25-61)-YFP and YFP are showed below the graph. GPI-anchor signal sequence is represented by asmall light blue box at the C-terminus.

1206 E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

binding is termed the nuclear retention signal (NRS). Since this pro-cess results in a net accumulation of the protein in the nucleus, it isalso referred to as passive transport. In contrast to passive transport,many proteins that accumulate in the nucleus exploit an energy-dependent transport process frequently facilitated by the importinprotein family (the process is referred to as active or facilitated nucleartransport). Some proteins use both active and passive means to enterthe nucleus. Thus, an NLS sequence which mediates facilitated nucleartransport may be masked by the presence of an NRS or vice versa.

The same situation may be relevant in the case of Sho. Since, Sho issmall enough to enter freely into the nucleus by passive entry, it couldaccumulate there by binding e.g. to nucleic acids through its reportedRGG-box [29]. This effect may not become apparent with the con-structs we explored above, since deletions that might abolish Sho's

active nuclear transport might also abolish its nucleic acid binding,thus its passive transport.

To avoid such pitfalls, we used two approaches: (i) where an NRScan be detected even in the presence of a facilitated transport-mediating NLS (see in Section 3.6.1) and (ii) where a facilitatedtransport-mediating NLS can be detected even in the presence ofan NRS (see in Section 3.6.2). Both the full-length Sho protein andits (25–61) fragment in fusion to a fluorescent protein were usedfor these purposes.

3.6.1. Testing if the (25–61) segment of Sho contains an NRSIn the first approach, chilling the cells for a few hours effectively

slows down all the active transport processes of the cells including nu-clear transport [43]. Thus proteins small in size that are accumulated in

Fig. 6. Evolutionary conservation of the mammalian Shadoo protein and the newlydescribed (RXXX)n motif. (A) Five residue long sliding window was used to visualizethe average sequence conservation in the mammalian Sho protein. Known motifs areindicated by blue boxes above the conservation diagram, red box and grey shadingindicates the position of the identified (RXXX)n motif (n=7–8). We observed high se-quence conservation and high residue composition conservation in the entire (RXXX)n(n=7–8) motif among the analyzed mammalian Shadoo proteins. The number of the(RXXX) repeats varies among the analyzed taxons. For example the ruminant Shadoosequences contain seven (RXXX) tetrarepeat, by contrast the rodent and hominidShadoo sequences contain eight (RXXX) tetrarepeats. This difference resulted in agap on the A graph, although it does not affect the whole motif conservation. (B) Se-quence logo of the (RXXX)n motif shows high conservation of tandem basic residues(mostly arginines), and conservation of the overall residue composition. A moredetailed comparison of the known vertebrate (RXXX)n motifs in different species issummarized in Supp. Fig. 3.

1207E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

the nucleus will freely equilibrate between the nucleus and the cytosolby diffusion unless they are retained by binding to a nuclear partner.

HeLa cells were transfected with (i) GFP, i.e. control showing nonuclear accumulation (data not shown), (ii) GFP-(Cherry)4-NLSSV40,i.e. control showing the intactness of the nuclear pore during treatment(iii) GFP-NLSSV40, i.e. positive control showing nuclear accumulationwithout possessing an NRS, (iv) NLSx3SV40-GFP, i.e. positive controlshowing nuclear accumulation and possessing an NRS (unpublishedresult), (v) along with ΔSS-Sho-YFP, ΔSS-YFP-Sho and GFP-(25–61)Sho. The cells were chilled for 4 h, then fixed and were analyzed byconfocal microscopy for the localization of the fusion proteins (Fig. 7).

Each of the control constructs showed localizations as expected.All constructs, except GFP showed nuclear accumulation at 37 °C.However, after 4 h incubation at 4 °C, only GFP-(Cherry)4-NLSSV40

and NLSx3SV40-GFP exhibited a nearly unchanged nuclear accumula-tion (Fig. 7). Since the NLSSV40 has no binding partner in the nucleus,the fact that the nuclear accumulation of GFP-(Cherry)4-NLSSV40—thatis too large to freely diffuse in and out of the nucleus—was unchanged,indicates that the intactness of the nuclear pore was not altered by thetreatment. Interestingly, accumulation of ΔSS-Sho-YFP, ΔSS-YFP-Shoand GFP-(25–61)Sho in the nucleuswas diminished except in the nucle-olus which retained its fluorescence and appeared as bright spots(Fig. 7). The identity of nucleolar localization was confirmed by theco-localization of the GFP signal with anti-fibrillarin antibody staining(data not shown).

To avoid any direct effect of the chilling on the binding or on theconformations of the binding partners, the same experiments wererepeated using 2-deoxy-D-glucose (DEG) and Na-azide treatment

that slows down the active processes by causing ATP depletion.These data gave the same results as the chilling experiments (Fig. 7).

3.6.2. Testing whether the (RXXX)8 motif mediates active transportIn a second approach, to check if the (RXXX)8 motif is able to

mediate facilitated nuclear transport, we applied a seemingly easytest to assess whether Sho's nuclear transport is an active or a passiveprocess, by increasing the dimensions (molecular weight and shape)of the fusion proteins above the cut off size of the nuclear pore. Thus,the passive transport of the fusion proteins is blocked. If a fragment isstill able to mediate the transport of the fusion protein, it must betransported by means of an active transport process. However, thereare a few considerations with this approach too. The fusion proteinitself, without an NLS should be totally excluded from the nucleusto rule out passive transport albeit with a slower kinetics. A GFPdimer works well in yeast that has a smaller size limit for passivetransport through the nuclear pore [49,56]. However, in HeLa cellswe found even three consecutively joined fluorescent proteins[GFP-(Cherry)2] in the nucleus of a fraction of the cells. For thatreason, we chose 5 consecutive fluorescent proteins GFP-(Cherry)4for C-terminal attachment along with GFP-β-Gal fusion constructsfor N-terminal attachment [57].

Thus to minimize the probability of an unnoticed interferencefrom the fusion protein on the function of an NLS we used bothN-terminal and C-terminal attachments, and two different fusion pro-teins [(Cherry)4 and β-Gal]. Furthermore we made the respective GFPcontrol construct where the putative NLS was connected by the samespacer and in the same orientation as to GFP-(Cherry)4 or to GFP-β-Galin order to reveal a possible interference with GFP.

To test whether the transport of the fusion constructs is an activeor passive process, we coupled the (25–61) fragment of Sho, the fulllength Shadoo protein, NLSqki5 or the NLSx3SV40 to GFP-β-Gal as wellto the control GFP constructs (Fig. 8A) and examined their localiza-tion in transiently transfected HeLa cells. We could not detect anyproteolytic cleavage that could separate GFP from β-Gal on Westernblots (Fig. 8A). The (25–61) fragment of Sho or the full length Showere not able to facilitate the nuclear targeting of the fusion construct(data not shown in the case of Sho-GFP-β-Gal). Since the correspond-ing ΔSS-Sho-YFP and (25–61)Sho-GFP proteins exhibited nuclear ac-cumulation, these results do not support the involvement of Shoin facilitated nuclear transport. By contrast, the positive controlNLSx3SV40-GFP-β-Gal exhibited nuclear targeting (Fig. 8B). Inter-estingly, one of the positive controls, NLSqki5 showed no nucleartargeting, and the corresponding GFP fusion protein also failed toaccumulate in the nucleus indicating an interference of NLSqki5

with GFP when attached N-terminally (data not shown).Experiments with 5 consecutive fluorescent proteins [GFP-

(Cherry)4 fusion constructs] with C-terminal attachments of theputative NLS sequences along with the corresponding controlsgave the same results. No nuclear targeting was observed in thecase of GFP-(Cherry)4-(25–61)Sho or the negative control GFP-(Cherry)4, however the controls—GFP-(Cherry)4-NLSSV40 and GFP-(Cherry)4-NLSqki5—showed nuclear signals (Supp. Fig. 5).

Taking the outcome of the experiments in Sections 3.6.1 and 3.6.2together, the results are rather surprising. The results of the experi-ments with the large fusion proteins leave little chance for Sho to betransported through the nuclear pore by active processes. None of thethree different constructs (with both Sho and the (25–61) fragment,in two orientations, with two different fusion partners) were activelytransported to the nucleus although each of the corresponding GFP con-structs showed nuclear accumulation. Thus, interference by the fusionproteins that mask the NLS function of Sho in all three different con-structs is rather unlikely.

However, the chilling and ATP-depletion experiments cause Sho toequilibrate between the cytosol and the nucleus (except the nucleoluswhere Sho remained bound) indicating that under these condition

un

trea

ted

chill

edD

EG

A

B

0

20

40

60

80

100

ΔSS-YFP-Sho ΔSS-Sho-YFP GFP-(25-61)Sho GFP-(Cherry)4-NLSSV40 GFP-NLSSV40 NLSx3SV40-GFP

untreated chilled DEG

nu

mb

er o

f ce

lls (

%)

*** *** ******

Fig. 7. Chilling and deoxy-glucose treatment of Shadoo transfected HeLa cells. (A) Effect of chilling and DEG treatment on Shadoo transfected HeLa cells (p=0.000, ***).GFP-(Cherry)4-NLSSV40 served as a control for nuclear lamina intactness. As this construct is too large for passive entry or exit, it maintained its nuclear accumulation during treatmentsindicating that the nuclear pores were intact. NLSSV40 does not mediate binding to any nuclear components in the cells that could inhibit its exit from the nucleus when the pores aredamaged—as demonstrated by a GFP-NLSSV40 construct. NLSx3SV40-GFP served as a control that possesses a nuclear retention signal and maintained its nuclear accumulation duringtreatment. Accumulation ofΔSS-Sho-YFP,ΔSS-YFP-Sho and GFP-(25–61)Sho in the nucleuswas diminished except in the nucleoluswhich retained its fluorescence and appeared as brightspots; (B) Representative confocal microscopy images of HeLa cells expressing of ΔSS-YFP-Sho, ΔSS-Sho-YFP, GFP-(25–61)Sho, GFP-(Cherry)4-NLSSV40, GFP-NLSSV40 and NLSx3SV40-GFP.Green: enhanced YFP/GFP, yellow: GFP and mCherry merged together. Scalebar represents 10 μm.

1208 E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

Sho no longer has binding partner(s) elsewhere in the nucleus thatwould facilitate its accumulation. Seemingly, Sho's nuclear accumula-tion is mediated by neither an active, nor a passive transport. Our inter-pretation of these data is that Sho accumulates in the nucleus by apassive transport but its binding or the presence of its binding partnerin the nucleus requires active processes (note that chilling or ATP-depletion slows down all active processes in the cells, not just thenuclear transport). Alternatively, under stress conditions (chilling orATP-depletion) modifications (such as by (de)phosphorylation or (de)methylation) may occur on Sho or on its binding partner that preventsthem binding.

Neither newly-synthesized proteins, nor phosphorylation seemsto play a role in Sho's accumulation as neither cycloheximide norstaurosporine treatment diminished it (data not shown). Therefore,the mechanism of the accumulation of Sho and the identification of

its putative binding partner in the nucleus and in the nucleolusneeds further research.

3.7. Localization of the endogenous Sho protein

Tagged versions of the Shadoo protein consistently appearedwithin the nucleus in our study, however, the localization of the en-dogenous protein is not demonstrated here. Answering this questionis rather difficult in the absence of appropriate antibodies. We havegenerated antibodies against five different peptides of Sho in rabbitsand hen, but only one of them seemed to show some specificityagainst Shadoo (EG62). Four antibodies [EG62 and the three commer-cially available antibodies: R-12 (Santa Cruz), S-12 (Santa Cruz),SPRN antibody (C-term) (Abgent)] gave strong nuclear staining. Weanalyzed the specificity of these antibodies in peptide competition

ANLSx3SV40- (25-61)Sho-

B

GFP-ββ-GalGFP-β-Gal GFP-β-Gal

GFP β-Gal GFP β-Gal GFP β-GalNLS (25-61)

140250

kDa

42

7295

55

42

26

17

GFP-β-GalNLSx3SV40-GFP-β-Gal

(25-61)Sho-GFP-β-Gal

GFP NLSx3SV40-GFP (25-61)Sho-GFP

Fig. 8. Subcellular localization and immune blot pattern of GFP-β-Gal fusion- and control constructs (A). Schematic structures and western blot analysis of GFP-β-Gal,NLSx3SV40-GFP-β-Gal (NLSx3SV40 in red box) and (25-61)Sho-GFP-β-Gal (10% SDS-PAGE; anti-GFP/YFP AB 1:3.500). (B) Confocal microscopy images of HeLa cells expressingGFP-β-Gal, NLSx3SV40-GFP-β-Gal, (25–61)Sho-GFP-β-Gal (upper images), GFP, NLSx3SV40-GFP and (25–61)Sho-GFP (lower images) [merged (blue and green): Hoescht 33342 andGFP]. The (25–61) fragment of Sho was not able to facilitate the nuclear and nucleolar targeting of the large GFP-β-Gal fusion construct. Scalebar represents 10 μm.

1209E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

experiments and by examining their co-localization with the YFP sig-nal in cells transfected with YFP-Sho or Sho-YFP constructs (data notshown), but the results of these experiments leave room for severalalternative interpretations. Considering these results together withthe low specificity of these antibodies as judged by Western blotting(data not shown), we deemed these data inconclusive.

However, Daude et al. [22] could reach a conclusion regarding the sub-cellular localization of the Shadoo protein by their antibodies using braintissues from Shadoo knockout mice as negative controls. Beside theexpected membrane localization, they acquired data showing definednucleocytoplasmic staining of the endogenous Shadoo in some of the hy-pothalamic neurons of wild-type mice. Furthermore, in contrast to thenegative controls, a portion of wild-type Shadoo was found in histonecontaining fractions from subcellular separations of both brain and cul-tured cells. During the review process of this manuscript a study waspublished by Lau et al. [58] where, in line with these observations,they proved that Shadoo indeed binds nucleic acids, further supportingits possible role in the nuclear compartment. In addition, severalinteracting partners from the nucleocytoplasmic compartment wereidentified for Shadoo in a recent study [6] that is also consistent withthe nuclear accumulation of a fraction of Shadoo.

3.8. The traffic of Sho to non-secretory compartments

There are a growing number of proteins that are detected indual (both secretory and non-secretory) localizations, however, thetransport mechanisms by which a secretory protein appears in thenucleocytoplasmic compartment has not been clarified for many ofthem yet [59].

The inefficiency of signal peptides in directing newly-synthesizedpolypeptides to the ER causes a fraction of the secretory proteins to ap-pear in the cytosol [60]. Furthermore, retrotranslocation or escape fromubiquitin–proteasome degradation on retrograde transport from the ERto the cytosol has also been suggested for proteinswith dual localization[61]. Such “mislocalized” proteins could be observed in the nucleus/cytosol as the removal of secretory proteins from the cytosol is appar-ently not completely effective [59]. Thus, a fraction of the endogenousSho likewise may appear in the nucleocytoplasmic compartment.

We also considered the effect of the overexpression on Shadoo'snuclear localization. The overexpression of Sho may overload the

ER-translocation machinery or proteosomes and lead to a non-naturalnuclear localization for these Shadoo variants. However, the over ex-pression of similar ER-targeted constructs (YFP-PrP, data not shown,YFP-GPI or Sho-YFP) does not result in such strong cytosolic/nuclearaccumulation, arguing against a simple overload explanation. In addi-tion, the overload effect should lead to the appearance of nuclear signalsadditional to themembrane, ER and Golgi staining. By contrast, in theseexperiments we commonly observed nuclear localization in cells inwhich the membrane (or ER/Golgi) signals were absent seeming topoint to a distinct, specific mechanism that directs Sho to the nucleus.

3.9. Implications for the function of the prion protein family

Apparently, PrP has minor populations of different forms withdifferent subcellular localizations that are associated with specificdistinct functions [38,62]. For example, some 2–5% [63,64] of theprion protein appears in a transmembrane form and exposes a partof the otherwise extracellular GPI-anchored prion protein to thecytosol. Although this small fraction was not detectable by im-munocytochemistry and its existence could be revealed only byoverexpressing the prion protein [65,66] a biological function wasattributed to this small fraction [62].

This multifunctionality, which is linked to multiple subcellular lo-cations might be a common feature of some natively disordered pro-teins [59,67–69], such as the prion and Shadoo proteins [38,62,64,70].

Since proteins' subcellular compartmentalization is not an all ornone process [60], a few percentage of Shadoo unavoidably appearsin the nucleocytoplasmic compartment. Our finding that Shadoo hassequences that mediate its nuclear accumulation indicates that wehave to calculate on a considerable nuclear fraction of Shadoo evenin those cells where this nuclear form of Shadoo does not reach a de-tectable level. Further studies are needed to discern if this small frac-tion of Shadoo with altered subcellular localization has a specificfunction like the transmembrane form of the prion protein.

PrP binds to polyanions, thus its nuclear accumulation may berationalized on the basis of its nucleic acid binding ability. Sho alsobinds nucleic acids (unpublished results and [58]), thus, its positivelycharged arginine residues (Suppl. Fig. 4) may also mediate nuclearaccumulation in a similar manner. However, there are two factsthat suggest alternative explanations. First, the second half of the

1210 E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

(RXXX)8 motif seems to exert a more profound effect on Sho nuclearaccumulation than the first part containing the RGG box that encom-pass four arginines (Figs. 3 and 5). Second, Sho seems to be involvedin different interactions in the nucleolus compared to the rest of thenucleus (Fig. 7). These observations suggest that the interactionsthat are responsible for its nuclear accumulation are rather specificin nature.

It is intriguing that although PrP and Sho have no apparent se-quence similarities other than about 25 amino acids around the HDregion share a feature that is apparently not related to this segment.There are forms of and/or conditions for both proteins where theyappear in the nucleocytosolic compartment and they both containsequences that are capable of mediating nuclear accumulation. Thismay provide considerable support to the argument that both PrPand Sho have a function in the nucleus. A detailed analysis and compar-ison of PrP's and Sho's nuclear localizations may reveal whether thesecommonalities are circumstantial or not.

4. Conclusions

In conclusion, we detected tagged-forms of the Shadoo protein inthe nucleus of several cell lines. This nuclear accumulation seems notto be mediated by a facilitated transport. However, under conditionsthat block facilitated transport nuclear accumulation is diminished.This may indicate a specific regulatory mechanism for Sho's nuclearaccumulation. Our results imply, in accord with data found in theliterature, that a fraction of the endogenous Shadoo protein is likelyto locate to the nucleus either of certain types of cells and/or undercertain conditions, however, whether this fraction has any function,and if so what, remains unclear.

Sho also accumulates in the nucleolus, even under conditions thatdiminish its accumulation in other parts of the nucleus suggesting aspecific binding partner there.

The nuclear accumulation of Shadoo is mediated by an N-terminalsegment of the protein encompassing (RXXX)n repeats. The (RXXX)nmotif is highly conserved and unique among the Shadoo proteinsfrom mammals. Whether this motif has evolved to enable a nuclearfunction for these proteins is not clear.

Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.bbamcr.2013.01.020.

Acknowledgement

We would like to thank to J. Baunoch for maintaining the cellcultures, K. Kovács, I. Vida and A. Mészáros for plasmid constructions.G. Antalffy's help with the confocal microscope is greatly appreciated.

GFP-β-Galactosidase and NLSx3SV40-GFP-β-Galactosidase werekind gifts of Y. Lazebnik.

Zpl2-1, Zw3-5, Hpl3-4 and Hw13-3 were kind gifts of Y. S. Kim andT. Onodera, respectively.

This research was supported by the Hungarian Scientific ResearchFund (K-82090) and a TÁMOP-4.2.2/B-10/1-2010-0012 grant. E. Tóthhad a Habilitas Scholarship from MFB Zrt.

References

[1] A. Aguzzi, Prion diseases of humans and farm animals: epidemiology, genetics,and pathogenesis, J. Neurochem. 97 (2006) 1726–1739.

[2] J. Collinge, Prion diseases of humans and animals: their causes and molecularbasis, Annu. Rev. Neurosci. 24 (2001) 519–550.

[3] B. Caughey, P.T. Lansbury, Protofibrils, pores, fibrils, and neurodegeneration:separating the responsible protein aggregates from the innocent bystanders,Annu. Rev. Neurosci. 26 (2003) 267–298.

[4] J.G. Fournier, Nonneuronal cellular prion protein, Int. Rev. Cytol. 208 (2001) 121–160.[5] A. Aguzzi, F. Baumann, J. Bremer, The prion's elusive reason for being, Annu. Rev.

Neurosci. 31 (2008) 439–477.[6] J.C. Watts, H. Huo, Y. Bai, S. Ehsani, A.H. Jeon, T. Shi, N. Daude, A. Lau, R. Young, L. Xu,

G.A. Carlson, D. Williams, D. Westaway, G. Schmitt-Ulms, Interactome analyses

identify ties of PrP and its mammalian paralogs to oligomannosidic N-glycans andendoplasmic reticulum-derived chaperones, PloS Pathog. 5 (2009) e1000608.

[7] L. Westergard, H.M. Christensen, D.A. Harris, The cellular prion protein (PrP(C)):its physiological function and role in disease, Biochim. Biophys. Acta 1772 (2007)629–644.

[8] V. Zomosa-Signoret, J.D. Arnaud, P. Fontes, M.T. Alvarez-Martinez, J.P. Liautard,Physiological role of the cellular prion protein, Vet. Res. 39 (2008) 9.

[9] N. Daude, D. Westaway, Biological properties of the PrP-like Shadoo protein,Front. Biosci. 16 (2011) 1505–1516.

[10] M. Premzl, L. Sangiorgio, B. Strumbo, J.A. Marshall Graves, T. Simonic, J.E. Gready,Shadoo, a new protein highly conserved from fish to mammals and with similar-ity to prion protein, Gene 314 (2003) 89–102.

[11] M.Miesbauer, T. Bamme, C. Riemer, B. Oidtmann, K.F.Winklhofer,M. Baier, J. Tatzelt,Prion protein-related proteins from zebrafish are complex glycosylated and containa glycosylphosphatidylinositol anchor, Biochem. Biophys. Res. Commun. 341 (2006)218–224.

[12] J.C. Watts, B. Drisaldi, V. Ng, J. Yang, B. Strome, P. Horne, M.S. Sy, L. Yoong, R.Young, P. Mastrangelo, C. Bergeron, P.E. Fraser, G.A. Carlson, H.T. Mount, G.Schmitt-Ulms, D. Westaway, The CNS glycoprotein Shadoo has PrP(C)-likeprotective properties and displays reduced levels in prion infections, EMBO J. 26(2007) 4038–4050.

[13] J.B. Oliveira-Martins, S. Yusa, A.M. Calella, C. Bridel, F. Baumann, P. Dametto, A.Aguzzi, Unexpected tolerance of alpha-cleavage of the prion protein to sequencevariations, PLoS One 5 (2010) e9107.

[14] S.G. Chen, D.B. Teplow, P. Parchi, J.K. Teller, P. Gambetti, L. Autilio-Gambetti, Trun-cated forms of the human prion protein in normal brain and in prion diseases,J. Biol. Chem. 270 (1995) 19173–19180.

[15] N. Stahl, M.A. Baldwin, A.L. Burlingame, S.B. Prusiner, Identification of glycoinositolphospholipid linked and truncated forms of the scrapie prion protein, Biochemistry29 (1990) 8879–8884.

[16] D.R. Taylor, E.T. Parkin, S.L. Cocklin, J.R. Ault, A.E. Ashcroft, A.J. Turner, N.M.Hooper, Role of ADAMs in the ectodomain shedding and conformational conver-sion of the prion protein, J. Biol. Chem. 284 (2009) 22590–22600.

[17] N. Daude, V. Ng, J.C. Watts, S. Genovesi, J.P. Glaves, S. Wohlgemuth, G. Schmitt-Ulms, H. Young, J. McLaurin, P.E. Fraser, D. Westaway, Wild-type Shadoo proteinsconvert to amyloid-like forms under native conditions, J. Neurochem. 113 (2010)92–104.

[18] V. Sakthivelu, R.P. Seidel, K.F. Winklhofer, J. Tatzelt, Conserved stress-protectiveactivity between prion protein and shadoo, J. Biol. Chem. 286 (2011) 8901–8908.

[19] A.S. Rambold, V. Muller, U. Ron, N. Ben-Tal, K.F. Winklhofer, J. Tatzelt, Stress-protective signalling of prion protein is corrupted by scrapie prions, EMBO J. 27(2008) 1974–1984.

[20] R. Young, B. Passet, M. Vilotte, E.P. Cribiu, V. Beringue, F. Le Provost, H. Laude, J.L.Vilotte, The prion or the related Shadoo protein is required for early mouseembryogenesis, FEBS Lett. 583 (2009) 3296–3300.

[21] B. Passet, R. Young, S. Makhzami, M. Vilotte, F. Jaffrezic, S. Halliez, S. Bouet, S.Marthey, M. Khalife, C. Kanellopoulos-Langevin, V. Beringue, F. Le Provost, H.Laude, J.L. Vilotte, Prion protein and shadoo are involved in overlapping embry-onic pathways and trophoblastic development, PLoS One 7 (2012) e41959.

[22] N. Daude, S. Wohlgemuth, R. Brown, R. Pitstick, H. Gapeshina, J. Yang, G.A.Carlson, D. Westaway, Knockout of the prion protein (PrP)-like Sprn gene doesnot produce embryonic lethality in combination with PrP(C)-deficiency, Proc.Natl. Acad. Sci. U.S.A. 109 (2012) 9035–9040.

[23] N. Daude, D. Westaway, Shadoo/PrP (Sprn (0/0)/Prnp (0/0)) double knockoutmice: more than zeroes, Prion 6 (2012) 420–424.

[24] W. Jiayu, H. Zhu, X. Ming, W. Xiong, W. Songbo, S. Bocui, L. Wensen, L. Jiping, M.Keying, L. Zhongyi, G. Hongwei, Mapping the interaction site of prion protein andSho, Mol. Biol. Rep. 37 (2009) 2295–2300.

[25] D. Rutishauser, K.D. Mertz, R. Moos, E. Brunner, T. Rulicke, A.M. Calella, A. Aguzzi,The comprehensive native interactome of a fully functional tagged prion protein,PLoS One 4 (2009) e4446.

[26] C. Gabus, E. Derrington, P. Leblanc, J. Chnaiderman, D. Dormont, W. Swietnicki, M.Morillas, W.K. Surewicz, D. Marc, P. Nandi, J.L. Darlix, The prion protein has RNAbinding and chaperoning properties characteristic of nucleocapsid protein NCP7of HIV-1, J. Biol. Chem. 276 (2001) 19301–19309.

[27] A. Grossman, B. Zeiler, V. Sapirstein, Prion protein interactions with nucleic acid:possible models for prion disease and prion function, Neurochem. Res. 28 (2003)955–963.

[28] C. Gabus, S. Auxilien, C. Pechoux, D. Dormont, W. Swietnicki, M. Morillas, W.Surewicz, P. Nandi, J.L. Darlix, The prion protein has DNA strand transfer proper-ties similar to retroviral nucleocapsid protein, J. Mol. Biol. 307 (2001) 1011–1021.

[29] S.M. Corley, J.E. Gready, Identification of the RGG box motif in Shadoo: RNA-binding and signaling roles? Bioinf. Biol. Insights 2 (2008) 383–400.

[30] A. Jaegly, F.Mouthon, J.M. Peyrin, B. Camugli, J.P. Deslys, D. Dormont, Search for a nu-clear localization signal in the prion protein, Mol. Cell. Neurosci. 11 (1998) 127–133.

[31] Y. Gu, J. Hinnerwisch, R. Fredricks, S. Kalepu, R.S. Mishra, N. Singh, Identificationof cryptic nuclear localization signals in the prion protein, Neurobiol. Dis. 12(2003) 133–149.

[32] K. Pfeifer, M. Bachmann, H.C. Schroder, J. Forrest, W.E. Muller, Kinetics of expres-sion of prion protein in uninfected and scrapie-infected N2a mouse neuroblasto-ma cells, Cell Biochem. Funct. 11 (1993) 1–11.

[33] H. Lorenz, O. Windl, H.A. Kretzschmar, Cellular phenotyping of secretory and nu-clear prion proteins associated with inherited prion diseases, J. Biol. Chem. 277(2002) 8508–8516.

[34] C. Crozet, J. Vezilier, V. Delfieu, T. Nishimura, T. Onodera, D. Casanova, S. Lehmann,F. Beranger, The truncated 23–230 form of the prion protein localizes to the

1211E. Tóth et al. / Biochimica et Biophysica Acta 1833 (2013) 1199–1211

nuclei of inducible cell lines independently of its nuclear localization signals andis not cytotoxic, Mol. Cell. Neurosci. 32 (2006) 315–323.

[35] M.E. Juanes, G. Elvira, A. Garcia-Grande, M. Calero, M. Gasset, Biosynthesis ofprion protein nucleocytoplasmic isoforms by alternative initiation of translation,J. Biol. Chem. 284 (2009) 2787–2794.

[36] C. Lund, C.M. Olsen, S. Skogtvedt, H. Tveit, K. Prydz, M.A. Tranulis, Alternativetranslation initiation generates cytoplasmic sheep prion protein, J. Biol. Chem.284 (2009) 19668–19678.

[37] E. Morel, S. Fouquet, C. Strup-Perrot, C. Pichol Thievend, C. Petit, D. Loew, A.M. Faussat,L. Yvernault, M. Pincon-Raymond, J. Chambaz, M. Rousset, S. Thenet, C. Clair, The cel-lular prion protein PrP(c) is involved in the proliferation of epithelial cells and in thedistribution of junction-associated proteins, PLoS One 3 (2008) e3000.

[38] A. Strom, G.S. Wang, D.J. Picketts, R. Reimer, A.W. Stuke, F.W. Scott, Cellular prionprotein localizes to the nucleus of endocrine and neuronal cells and interacts withstructural chromatin components, Eur. J. Cell Biol. 90 (2011) 414–419.

[39] M. Cokol, R. Nair, B. Rost, Finding nuclear localization signals, EMBO Rep. 1 (2000)411–415.

[40] A.N. Nguyen Ba, A. Pogoutse, N. Provart, A.M. Moses, NLStradamus: a simpleHidden Markov Model for nuclear localization signal prediction, BMC Bioinforma.10 (2009) 202.

[41] B.H. Kim, J.I. Kim, E.K. Choi, R.I. Carp, Y.S. Kim, A neuronal cell line that does notexpress either prion or doppel proteins, NeuroReport 16 (2005) 425–429.

[42] C. Kuwahara, A.M. Takeuchi, T. Nishimura, K. Haraguchi, A. Kubosaki, Y. Matsumoto,K. Saeki, T. Yokoyama, S. Itohara, T. Onodera, Prions prevent neuronal cell-line death,Nature 400 (1999) 225–226.

[43] S. Chatterjee, U. Stochaj, Diffusion of proteins across the nuclear envelope of HeLacells, Biotechniques 24 (1998) 668–674.

[44] R. Dono, D. James, R. Zeller, A GR-motif functions in nuclear accumulation ofthe large FGF-2 isoforms and interferes with mitogenic signalling, Oncogene 16(1998) 2151–2158.

[45] D. Christophe, C. Christophe-Hobertus, B. Pichon, Nuclear targeting of proteins:how many different signals? Cell. Signal. 12 (2000) 337–341.

[46] K. Nakai, P. Horton, PSORT: a program for detecting sorting signals in proteins andpredicting their subcellular localization, Trends Biochem. Sci. 24 (1999) 34–36.

[47] C.J. Sigrist, L. Cerutti, N. Hulo, A. Gattiker, L. Falquet, M. Pagni, A. Bairoch, P.Bucher, PROSITE: a documented database using patterns and profiles as motifdescriptors, Brief. Bioinform. 3 (2002) 265–274.

[48] A. Lange, R.E. Mills, C.J. Lange, M. Stewart, S.E. Devine, A.H. Corbett, Classical nuclearlocalization signals: definition, function, and interaction with importin alpha, J. Biol.Chem. 282 (2007) 5101–5105.

[49] A. Lange, L.M. McLane, R.E. Mills, S.E. Devine, A.H. Corbett, Expanding the definitionof the classical bipartite nuclear localization signal, Traffic 11 (2010) 311–323.

[50] M. Ghaffari, X. Zeng, J.A. Whitsett, C. Yan, Nuclear localization domain of thyroid tran-scription factor-1 in respiratory epithelial cells, Biochem. J. 328 (Pt 3) (1997) 757–761.

[51] C. Christophe-Hobertus, V. Duquesne, B. Pichon, P.P. Roger, D. Christophe, Criticalresidues of the homeodomain involved in contacting DNA bases also specifythe nuclear accumulation of thyroid transcription factor-1, Eur. J. Biochem. 265(1999) 491–497.

[52] C.G. Burd, G. Dreyfuss, Conserved structures and diversity of functions of RNA-binding proteins, Science 265 (1994) 615–621.

[53] R. Lukasiewicz, B. Nolen, J.A. Adams, G. Ghosh, The RGG domain of Npl3p recruitsSky1p through docking interactions, J. Mol. Biol. 367 (2007) 249–261.

[54] J. Wu, L. Zhou, K. Tonissen, R. Tee, K. Artzt, The quaking I-5 protein (QKI-5) has anovel nuclear localization signal and shuttles between the nucleus and the cyto-plasm, J. Biol. Chem. 274 (1999) 29202–29210.

[55] B.J. Lee, A.E. Cansizoglu, K.E. Suel, T.H. Louis, Z. Zhang, Y.M. Chook, Rules fornuclear localization sequence recognition by karyopherin beta 2, Cell 126(2006) 543–558.

[56] M.T. Harreman, T.M. Kline, H.G. Milford, M.B. Harben, A.E. Hodel, A.H. Corbett,Regulation of nuclear import by phosphorylation adjacent to nuclear localizationsignals, J. Biol. Chem. 279 (2004) 20613–20621.

[57] L. Faleiro, Y. Lazebnik, Caspases disrupt the nuclear-cytoplasmic barrier, J. CellBiol. 151 (2000) 951–959.

[58] A. Lau, C.E. Mays, S. Genovesi, D. Westaway, RGG repeats of PrP-like Shadoo pro-tein bind nucleic acids, Biochemistry 51 (2012) 9029–9031.

[59] E.J. Arnoys, J.L. Wang, Dual localization: proteins in extracellular and intracellularcompartments, Acta Histochem. 109 (2007) 89–110.

[60] C.G. Levine, D. Mitra, A. Sharma, C.L. Smith, R.S. Hegde, The efficiency of proteincompartmentalization into the secretory pathway, Mol. Biol. Cell 16 (2005)279–291.

[61] N. Afshar, B.E. Black, B.M. Paschal, Retrotranslocation of the chaperone calreticulinfrom the endoplasmic reticulum lumen to the cytosol, Mol. Cell. Biol. 25 (2005)8844–8853.

[62] D. Gibbings, P. Leblanc, F. Jay, D. Pontier, F. Michel, Y. Schwab, S. Alais, T. Lagrange,O. Voinnet, Human prion protein binds Argonaute and promotes accumulation ofmicroRNA effector complexes, Nat. Struct. Mol. Biol. 19 (2012) 517–524, (S511).

[63] N.S. Rane, O. Chakrabarti, L. Feigenbaum, R.S. Hegde, Signal sequence insufficiencycontributes to neurodegeneration caused by transmembrane prion protein, J. CellBiol. 188 (2010) 515–526.

[64] A.B. Emerman, Z.R. Zhang, O. Chakrabarti, R.S. Hegde, Compartment-restrictedbiotinylation reveals novel features of prion protein metabolism in vivo, Mol.Biol. Cell 21 (2010) 4325–4337.

[65] C.S. Yost, C.D. Lopez, S.B. Prusiner, R.M. Myers, V.R. Lingappa, Non-hydrophobicextracytoplasmic determinant of stop transfer in the prion protein, Nature 343(1990) 669–672.

[66] R.S. Hegde, J.A. Mastrianni, M.R. Scott, K.A. DeFea, P. Tremblay, M. Torchia, S.J.DeArmond, S.B. Prusiner, V.R. Lingappa, A transmembrane form of the prion pro-tein in neurodegenerative disease, Science 279 (1998) 827–834.

[67] A.M. Villamil Giraldo, M. Lopez Medus, M. Gonzalez Lebrero, R.S. Pagano, C.A.Labriola, L. Landolfo, J.M. Delfino, A.J. Parodi, J.J. Caramelo, The structure ofcalreticulin C-terminal domain is modulated by physiological variations of calci-um concentration, J. Biol. Chem. 285 (2010) 4544–4553.

[68] S.J. Wijeyesakere, A.A. Gafni, M. Raghavan, Calreticulin is a thermostable proteinwith distinct structural responses to different divalent cation environments,J. Biol. Chem. 286 (2011) 8771–8785.

[69] M.M. Babu, R. van der Lee, N.S. de Groot, J. Gsponer, Intrinsically disordered pro-teins: regulation and disease, Curr. Opin. Struct. Biol. 21 (2011) 432–440.

[70] M. Miesbauer, A.S. Rambold, K.F. Winklhofer, J. Tatzelt, Targeting of the prionprotein to the cytosol: mechanisms and consequences, Curr. Issues Mol. Biol. 12(2010) 109–118.