Báo cáo khoa học: Drosophila proteins involved in metabolism of uracil-DNA possess different types of nuclear localization signals pdf

For this latter mecha-nism, which is the major one for proteins larger than 30–35 kDa, specific and direct nuclear targeting requires the presence of a nuclear localization signal NLS, wh

possess different types of nuclear localization signals

Ga´bor Mere´nyi1, Emese Ko´nya1 and Bea´ta G Ve´rtessy1,2

1 Institute of Enzymology, Biological Research Center, Hungarian Academy of Sciences, Budapest, Hungary

2 Department of Applied Biotechnology, Budapest University of Technology and Economics, Budapest, Hungary

Introduction

In eukaryotic organisms, proteins with cognate nuclear

function must penetrate the nuclear envelope after

translation in the cytoplasm Nuclear import and

export of proteins can proceed by active or passive

transport, or as a member of protein complex actively

targeted into the nucleus [1–3] For this latter mecha-nism, which is the major one for proteins larger than 30–35 kDa, specific and direct nuclear targeting requires the presence of a nuclear localization signal (NLS), which is the relevant sequence information in

Keywords

cellular trafficking; Drosophila melanogaster;

dUTPase; nuclear localization signal;

uracil-DNA degrading factor

Correspondence

B.G Ve´rtessy, Institute of Enzymology,

Biological Research Center, Hungarian

Academy of Sciences, Karolina u´t 29,

H-1113 Budapest, Hungary

Fax: +36 1 466 5465

Tel: +36 1 279 3116

E-mail: vertessy@enzim.hu

(Received 4 August 2009, revised 23

February 2010, accepted 1 March 2010)

doi:10.1111/j.1742-4658.2010.07630.x

Adequate transport of large proteins that function in the nucleus is indis-pensable for cognate molecular events within this organelle Selective pro-tein import into the nucleus requires nuclear localization signals (NLS) that are recognized by importin receptors in the cytoplasm Here we inves-tigated the sequence requirements for nuclear targeting of Drosophila pro-teins involved in the metabolism of uracil-substituted DNA: the recently identified uracil-DNA degrading factor, dUTPase, and the two uracil-DNA glycosylases present in Drosophila For the uracil-DNA degrading factor, NLS prediction identified two putative NLS sequences [PEKRKQE(320– 326) and PKRKKKR(347–353)] Truncation and site-directed mutagenesis using YFP reporter constructs showed that only one of these basic stretches is critically required for efficient nuclear localization in insect cells This segment corresponds to the well-known prototypic NLS of SV40 T-antigen An almost identical NLS segment is also present in the Drosophila thymine-DNA glycosylase, but no NLS elements were pre-dicted in the single-strand-specific monofunctional uracil-DNA glycosylase homolog protein This latter protein has a molecular mass of 31 kDa, which may allow NLS-independent transport For Drosophila dUTPase, two isoforms with distinct features regarding molecular mass and subcellu-lar distribution were recently described In this study, we characterized the basic PAAKKMKID(10–18) segment of dUTPase, which has been pre-dicted to be a putative NLS by in silico analysis Deletion studies, using YFP reporter constructs expressed in insect cells, revealed the importance

of the PAA(10–12) tripeptide and the ID(17–18) dipeptide, as well as the role of the PAAK(10–13) segment in nuclear localization of dUTPase We constructed a structural model that shows the molecular basis of such rec-ognition in three dimensions

Abbreviations

NLS, nuclear localization signal; LD-DUT, long isoform of dUTPase; NTT-DUT, N-terminally truncated short isoform of dUTPase; SMUG1, single-strand-specific monofunctional uracil-DNA glycosylase 1; T-ag, T-antigen; TDG, thymine-DNA glycosylase; UDE, uracil-DNA degrading factor.

sequence motifs have been identified to date, and there

is no unique well-defined consensus amino acid

sequence for all NLS [4,5] However, major common

characteristics of these sequences are (i) a high content

of basic amino acid residues such as lysine (K) and

arginine (R), and (ii) the presence of conserved

pro-line(s) (P) potentially involved in breaking secondary

structural elements within the NLS One group of

sim-ple NLS includes monopartite motifs, generally defined

as a short amino acid region consisting of 4–6 basic

residues in a row, like the classic NLS of SV40 large

T-antigen (SV40 T-ag) [6] Another type of NLS, such

as the NLS of nucleoplasmin in Xenopus laevis,

com-prises bipartite motifs, which contain two distinct

stretches of positively charged clusters separated by a

mutation-tolerant linker region [7] In addition,

sequences containing several neutral or even negatively

charged conserved residues may also act as functional

monopartite NLS, with the negatively charged

aspar-tate⁄ glutamate (D ⁄ E) also contributing to NLS

func-tion [8] Interestingly, the NLS of human RanBP3 [9]

is an unusual signal with close homology to the NLS

of c-Myc [10]

Nuclear proteins containing NLS motifs could enter

into the nucleus via the nuclear pore, utilizing a strictly

organized mechanism maintained by karyopherin

mol-ecules and the nuclear pore complex [11] The nuclear

pore complex is a large protein complex consisting of

multiple subunits and located in the nuclear

mem-brane It is also the main possibility for exchange of

small particles, e.g ions, nucleotides, etc., between the

nuclear and cytosolic compartments Importin b,

a type of karyopherin molecule, is a nuclear transport

receptor, which can bind its molecular cargo either

directly or indirectly through adaptor proteins such as

importin a Importin b is unable to bind directly to

classical nuclear targeting motifs such as the NLS of

SV40 T-ag or the NLS of nucleoplasmin, but could

mediate nuclear import indirectly in association with

importin a Importin a possesses two major domains

for its adaptor function, the importin b binding (IBB)

domain in its N-terminus and the C-terminal

NLS-binding domain In the absence of importin b, an

auto-inhibiting part of the IBB domain forms an

intra-molecular interaction with the NLS-binding domain,

preventing the association with NLS on the cargo

pro-tein Thus, the presence or absence of importin b

regu-lates the NLS binding ability of importin a The

relatively large NLS-binding domain of importin a

consists of ten armadillo repeats, each constituting

three a-helices In association with each other, the

armadillo repeats form a large concave superhelical

in extended conformation to the binding pockets of the superhelical surface of importin a These binding pockets contain several conserved residues (e.g aspara-gine, tryptophan and negatively charged residues) involved in hydrophobic and electrostatic interactions with the positively charged residues of the NLS (see [1] for recent review)

Here, we wished to identify and characterize NLS for Drosophila melanogaster proteins involved in uracil-DNA metabolism Four such major proteins have been described to date: (i) the newly identified uracil-DNA degrading factor (UDE) [12,13], (ii) dUT-Pase, which is responsible for prevention of uracil incorporation into DNA [14], and (iii) two DNA gly-cosylases, thymine-DNA glycosylase (TDG) [15] and the single-strand-specific monofunctional uracil-DNA glycosylase 1 (SMUG1) homolog protein

The UDE protein, encoded by the CG18410 gene in the D melanogaster genome, was recently identified in

a pull-down screen on uracil-DNA from larval extracts [12] In vitro studies have shown that this protein spe-cifically degrades uracil-containing DNA, but lacks any appreciable homology to previously described ura-cil-DNA-recognizing proteins CG18410 gene expres-sion may be under developmental control, and the protein has been suggested to play a role in metamor-phosis in Drosophila The subcellular localization of this protein had not been characterized

dUTPase catalyzes the cleavage of dUTP into dUMP to control cellular dUTP⁄ dTTP ratios, and is

an essential enzyme in both prokaryotes and eukary-otes [16,17] Lack of dUTPase leads to uracil-substi-tuted DNA that perturbs base excision repair, resulting in DNA fragmentation and thymine-less cell death [14] Most dUTPases are homotrimers with native molecular masses of approximately 50–65 kDa [18–24] Both human and D melanogaster cells contain

a nuclear isoform of dUTPase, and the NLS segment

of the human enzyme has been investigated in detail [25] In D melanogaster dUTPase, a similar N-terminal segment was recently proposed as the NLS region [26]

In D melanogaster, two physiological isoforms of the enzyme were identified, with apparent molecular masses of 69 and 63 kDa for the native homotrimers (termed long isoform, LD-DUT, and the N-terminally truncated short isoform, NTT-DUT, respectively) [27] Only LD-DUT contains the complete putative NLS sequence [PAAKKMKID(10–18)], while NTT-DUT lacks 14 residues at the N-terminus This segment shows a high degree of flexibility and cannot be located in the 3D structure of the protein determined

by X-ray crystallography (PDB ID 3ECY) [21])

Uracil-DNA glycosylases are the key repair enzymes

that remove uracil from DNA by catalyzing cleavage

of the N-glycosidic bond [28] To perform this function

in eukaryotic cells, these enzymes must reside in the

nuclear or mitochondrial compartments ([29] There

are four or five major families of uracil-DNA

glycosy-lases, but only two of these are encoded in the

D melanogaster genome [30] The molecular mass of

these two glycosylases, based on reported sequences

[15], are 191 kDa for TDG and 31 kDa for the

SMUG1 homolog No quantitative data are available

indicating potential oligomerization for the monomeric

species, and the family member uracil-DNA glycosylase

is a monomer [31]

In the present study, we aimed to (i) determine the

subcellular distribution of UDE, (ii) identify sequence

determinants essential for nuclear translocation in

pro-teins involved in uracil-DNA metabolism in

Drosoph-ila, and (iii) functionally characterize these NLS Based

on in silico prediction, we fused various sequence

seg-ments from the ORF of UDE and dUTPase to the

yellow fluorescent protein (YFP) and generated

chime-ric reporter constructs In addition, to characterize the

essential and sufficient amino acids of the NLS, we

performed deletion studies and site-directed

mutagene-sis on the putative NLS regions For transient

transfec-tion studies, we used the Sf9 homogeneous insect cell

line, which has superior characteristics for subcellular

sorting analysis compared with the Drosophila

Schnei-der 2 cell line, including convenient generation time,

and its morphology allows straightforward microscopic

detection of cellular compartments

Results and Discussion

Subcellular targeting of UDE

Nuclear targeting of UDE may be critical for

perfor-mance of the suggested degradation function on

geno-mic DNA containing uracil [12] In silico prediction

(using PSORTII [32]; http://psort.ims.u-tokyo.ac.jp/)

suggested two individual clusters of residues as a

puta-tive NLS region, separated by 21 amino acids, in the

C-terminus of the protein (Fig 1A, and Tables 1 and

2) The first cluster (NLS1), PEKRKQE(320–326),

consists of both positively and negatively charged

resi-dues The second stretch (NLS2), PKRKKKR(347–

353)E, is located at the very end of the C-terminus and

has a high proportion of positively charged amino

acids Underlined residues are predicted to be part of

the NLS Each sequence starts with the neutral amino

acid proline and ends its context with glutamic acid

We fused the full-length UDE, containing these two

predicted sequences, to the N-terminus of YFP After Sf9 cell transfection using the chimera construct, fluo-rescence was observed on samples of fixed cells The 22.2 kDa YFP alone, used as a control, could pene-trate non-selectively through the nuclear pore, most probably because its smaller molecular mass allows passive diffusion Fluorescence microscopy analysis showed that the YFP-tagged UDE has an exclusive nuclear localization in Sf9 cells (Fig 2A and Table 3)

In the control experiment, YFP alone was observed throughout the cell (Fig 2K and Table 3) These data demonstrate that the wild-type UDE is targeted specifi-cally and exclusively into the nucleus, in agreement with its putative nuclear function in insect cells

Subcellular distribution of C-terminal truncated forms of UDE

To test whether the nuclear import of UDE requires any or both of the predicted signals, various C-termi-nally truncated UDE species were linked to the N-ter-minus of the YFP reporter (Fig 1) In the first construct, UDED(316)355)–YFP, a large part of the C-terminus was deleted, including both putative NLS segments In the second construct, UDED(346)355)– YFP, the last ten residues of the C-terminus were removed, including the PKRKKKR(347–353) sequence The reporter constructs were introduced into Sf9 cells and subsequently analyzed by fluorescent microscopy The results show that lack of the full-length flexible C-terminal region, containing both of the predicted signals, totally abolished the nuclear dis-tribution, causing significant cytoplasmic retention of UDE (Fig 2B and Table 3) When the last ten residues

of the C-terminus, including only the second predicted NLS, were deleted, the pattern of subcellular distribu-tion was also exclusively cytoplasmic (Fig 2C) These results suggest that the PEKRKQE(320–326) sequence

on its own is not able to translocate the protein into the nuclear compartment In contrast, the presence of the PKRKKKR(347–353) sequence, consisting of six contiguous positively charged amino acids, is critical for exclusive nuclear localization of UDE The PKRKKKR(347–353) segment is almost identical to the NLS of SV40 T-ag, indicating a powerful capabil-ity for function as an NLS

Subcellular targeting of UDE containing specific site mutations in the NLS sequence

To extend our investigations, we generated separate mutations to identify amino acids responsible for the nuclear targeting function of the PKRKKKR sequence

(Fig 1) The K350A⁄ K351A double mutation slightly

altered the pattern of subcellular distribution,

indicat-ing attenuation of the nuclear targetindicat-ing effect (Fig 2D

and Table 3) The K350A⁄ K351A ⁄ K352A ⁄ R353A

quadruple mutation also perturbed the exclusive

nuclear targeting of UDE, resulting in significant

cyto-plasmic retention (Fig 2E) Based on these results, the

PKRKKKR(347–353) sequence is suggested to be a

strong NLS sequence with high mutation tolerance In

accordance with the putative segments defined by

in silico prediction (Table 1), it was found that the

presence of the KPKR(346–349) segment is sufficient

for partial nuclear localization of the protein

Subcellular targeting potential of the predicted

UDE NLS1 and NLS2 sequences

To determine whether either of the two predicted NLS

sequences possess strong nuclear targeting potential on

their own, the PEKRKQE (NLS1) and PKRKKKR

(NLS2) coding sequences were fused as a C-terminal tag

to YFP protein (Fig 1) The constructs YFP–UDE-NLS1 and YFP–UDE-NLS2 were transiently

transfect-ed into Sf9 cells Expression and intracellular appear-ance of the fluorescent proteins were observed by fluorescent microscopy The results show that the NLS2 segment has selective and powerful targeting potential for accumulation of YFP in the nucleus (Fig 2F and Table 3) The pattern of subcellular distribution of YFP–UDE-NLS1 was not exclusively nuclear or cyto-plasmic, although some accumulation was observed within the nuclear compartment compared to the YFP control (compare Fig 2G and K) Further, the C-termi-nal portion of UDE was fused to YFP and expressed in Sf9 cells This UDED(1)319)–YFP reporter construct containing both predicted NLS sequences was exclusively retained in the nucleus (Fig 2H) After introducing quadruple mutations (K350A⁄ K351A ⁄ K352A⁄ R353A) into this construct UDED[1)319 (350AAAA353)]–YFP, the exclusive nuclear distribu-tion was highly perturbed, but increased nuclear accu-mulation was observed compared to YFP alone

B

C

UDEΔ(316–355)-YFP UDE WT -YFP

UDEΔ(346–355)-YFP

UDEΔ(1–319)-YFP UDEΔ[1–319 (350AAAA353)]-YFP

YFP-UDE-NLS2Δ350–353 YFP-UDE-NLS2 YFP-UDE-NLS1

UDE(350AA351)-YFP UDE(350AAAA353)-YFP

Fig 1 Scheme of D melanogaster UDE constructs used in the present study (A) Position and context of putative nuclear localization sequences (underlined) within the flexible C-terminus of D melanogaster UDE are indicated (B) Schematic representation of various UDE– YFP reporter constructs The wild-type (wt), flexible C-terminally truncated [D(316 )355)] and the NLS truncated [D(346)355)] coding sequences were fused in-frame to the N-terminus of YFP protein, resulting in UDE WT –YFP, UDED(316 )355)–YFP and UDED(346)355)–YFP reporter constructs The UDE(350AA351)–YFP reporter construct contains the K350A and K351A mutations, and the UDE(350AAAA353)– YFP reporter construct contains the K350A, K351A, K352A and R353A mutations The truncated reporter constructs UDED(1 )319)–YFP and UDED[1 )319(350AAAA353)]–YFP are also indicated The relevant regions, positions and mutations of the NLS of UDE are indicated by differ-ently shaded boxes (C) The predicted NLS sequences (NLS1 and NLS2) and the deleted variant of NLS2 were fused in-frame to the C-termi-nus of the YFP ORF generating the YFP–UDE-NLS1, YFP–UDE-NLS2 and YFP–UDE-NLS2D(350 )353) reporter constructs Establishment of vector constructs was performed as described in Experimental procedures.

(Fig 2I,K) The last examined reporter construct YFP–

UDE-NLS2D(350)353), which possesses only three

basic residues [KPKR(346–349)] from the NLS2

seg-ment fused to YFP, also showed localization in the

nucleus and the cytoplasm, with some accumulation

within the nucleus (Fig 2J)

These observations indicate that the NLS2 segment

is a strong monopartite NLS, and that the

contribu-tion of the predicted NLS1 to nuclear localizacontribu-tion is

negligible Within the NLS2 segment, both the KPKR

and the KKKR tetrapeptides contribute to nuclear

localization

Prediction of NLS signals in Drosophila

uracil-DNA glycosylases

Table 1 lists the predicted NLS signals for the TDG

protein Several clusters of putative localization

signals were observed Among these, the PKKRG RKKK(711–719) sequence is almost identical to the NLS of the SV40 T-ag and also to the UDE NLS segment As the SV40 T-ag has been extensively char-acterized [33] and we also found in our present experi-ments that such a sequence has very strong nuclear localization potential, we propose that this sequence also acts as an NLS in the TDG protein For the SMUG1 homolog protein, no nuclear localization sig-nal was predicted by the PSORTII program (Table 1) Lack of predicted signals cannot be taken as evidence for the actual absence of NLS segments, as prediction performs well only for classical NLS It is also worth-while noting that the molecular size of SMUG1 may allow passive translocation to the nucleus

Subcellular distribution of the D melanogaster dUTPase isoforms

For D melanogaster dUTPase, prediction identified the underlined segment within PAAKKMK(10–16)ID

as a conventional NLS comprising a short cluster of non-polar and basic residues (Fig 3, and Tables 1 and 2) To determine the subcellular distribution of

D melanogaster dUTPase isoforms in the Sf9 cell line,

Table 1 In silico predictions of putative nuclear localization signals

of Drosophila dUTPase, UDE, TDG and SMUG1 homolog proteins.

To identify the putative nuclear localization sites, the full length

open-reading frame sequences of the proteins were obtained from

the UniProt database (http://www.uniprot.org) and analyzed using

PSORTII (http://psort.ims.u-tokyo.ac.jp/) Putative signal sequences,

defined as potential NLS regions, are shown, with the number in

parentheses indicating the number of the first residue.

Protein

Uniprot ID

of protein

(UniProtKB ⁄

TrEMBL)

ORF length (amino acids

Sequences defined

as putative NLS segments

PKRK (347) KRKK (348) RKKK (349) KKKR (350) PEKRKQE (320) PKRKKKR (347)

RKKK (716) RKKH (760) KKKR (1088) RPKK (1093) PKKK (1141) KKKR (1142) RPKK (1147) PNNRKRQ (114) PMPKKRG (709) PKKRGRK (711) PKERKKH (757) PLEKKKR (1085) PKKIKGQ (1094) PKKKRGR (1141) PKKLKPA (1148)

Table 2 Comparison of UDE, dUTPase and TDG NLS segments with NLS sequences of various proteins The monopartite sequences listed show close similarity to either the SV40 T-ag NLS

or the c-Myc NLS segments The NLS sequences of UDE and TDG show close homology to the SV40 T-ag NLS, but the D melanogas-ter dUTPase NLS belongs to the c-Myc group Inmelanogas-terestingly, the NLS segment of human dUTPase is more similar to the first group

of sequences For comparison, the classic bipartite NLS sequence

of X laevis nucleoplasmin is shown, which possesses an additional short cluster of basic residues separated by 10 amino acids from the basic stretch, which has close homology with the NLS of SV40 T-ag SV40 T-ag, simian virus 40 large T-antigen [6]; v-Jun, sarcoma virus 17 oncogene homolog [39]; H2B, histone 2B [40]; UDE, uracil-DNA degrading factor; human dUTPase [25]; c-Myc, myelocytoma-tosis cellular oncogene [10]; RanBP3, Ran binding protein 3 [9].

Monopartite

H2B of Saccharomyces cerevisiae GKKRSKV

dUTPase of D melanogaster PAAKKMKID

Bipartite

reporter constructs were created by N-terminal fusion

to YFP (Fig 3B and Table 3) Cellular targeting of

both isoforms was subsequently determined via cell

transfection experiments followed by fluorescent

micro-scopic detection The results show that the long

iso-form of dUTPase (LD-DUT) is specifically targeted

into the nucleus, but the short one (NTT-DUT) was

not able to enter into the nuclear compartment and remained exclusively in the cytoplasm (Fig 4A,B) This is in agreement with studies performed in Dro-sophila Schneider S2 cells [26]

These results indicated that the presence of the pre-dicted complete targeting sequence is necessary and sufficient for exclusive nuclear targeting of the long

F

G

H

I

J

K

B

A

C

D

E

Fig 2 Subcellular localization of D melanogaster UDE protein and its various sequence derivatives Fluorescence microscopy observations show the subcellular distribution of chimeric UDE constructs (A) Wild-type UDE (UDE WT –YFP) was targeted exclusively to the nucleus (B,C) Deletion studies showed that removal of the entire flexible C-terminus or the last ten residues of the C-terminus of the UDE ORF results in exclusive cytoplasmic localization of chimeric constructs UDED(316 )355)–YFP and UDED(346)355)–YFP, respectively (D) The reporter construct UDE(350AA351)–YFP, which contains a double K ⁄ A mutation, is predominantly located in the nucleus and slightly in the cytoplasm (E) Quadruple mutations in the reporter construct [UDE(350AAAA353)–YFP] have an attenuating effect on nuclear localization, with most of the construct accumulating within the nucleus, although cytoplasmic localization was also observed (F) The YFP–UDE-NLS2 reporter localized almost exclusively in the nucleus (G) The YFP–UDE-NLS1 construct was seen in both the nuclear compartment and the cytoplasm (H) The UDED(1 )319)–YFP reporter, which contains both predicted NLS sequences, was exclusively retained in the nucleus (I) The UDED[1 )319(350AAAA353)]–YFP construct was seen in both the nucleus and the cytoplasm, but seemed to accumulate in the nucleus (J) The reporter construct YFP–UDE-NLS2D(350 )353), which possesses only three basic residues from the NLS segment, did not show any selective compartmentalization, and was distributed almost equally in the nucleus and the cytoplasm (K) YFP alone was used as

a negative control The cellular distribution of YFP was approximately the same within the nuclear and cytoplasmic compartments.

Table 3 Summary of results for the subcellular distributions of reporter constructs Details of the reporter constructs for dUTPase and UDE are shown in the first three columns The observed subcellular localizations of reporter constructs are indicated by plus and minus signs Two plus signs indicate distribution between the nuclear and cytoplasmic compartments; one plus sign indicates exclusion from either the nucleus or the cytoplasm.

Protein Name of reporter construct NLS sequence present in reporter construct

Localization

A

B

C DUT-NLS-YFP

DUT-NLSΔ(10–12)-YFP DUT-NLSΔ(10–13)-YFP DUT-NLSΔ(17–18)-YFP DUT-NLSΔ(10–12,17–18)-YFP DUT-NLSΔ(10–13,17–18)-YFP

Fig 3 Scheme of D melanogaster dUTPase constructs used in the present study (A) The position and context of putative nuclear localiza-tion signals (underlined) are indicated in the N-terminus of the long isoform of D melanogaster dUTPase (B) The long (LD-DUT WT ) and short (NTT-DUT WT ) isoforms of the D melanogaster dUTPase coding sequences were fused in-frame to the N-terminus of the YFP ORF to gener-ate the LD-DUT–YFP and NTT-DUT–YFP chimeric constructs, respectively The relevant motifs, regions and positions of the NLS of dUTPase are indicated by differently shaded boxes (C) The NLS sequence (PAAKKMKID) and its truncated sequence variants (KKMKID and KMKID) were fused in-frame to the N-terminus of the YFP ORF generating the DUT-NLS–YFP, DUT-NLSD(10 )12)–YFP and the DUT-NLSD(10)13)– YFP reporter constructs Further reporter constructs, DUT-NLSD(17 )18)–YFP, DUT-NLSD(10)12,17)18)–YFP and DUT-NLSD(10)13,17)18)– YFP, are also indicated, which were generated in the way, but all lack the ID(17–18) dipeptide Establishment of vector constructs was performed by the general cloning method described in Experimental procedures.

isoform (LD-DUT) The partial segment MKID(15–18),

present on the short isoform, cannot drive nuclear

import In the case of the short isoform (NTT-DUT),

absence of the first 14 residues of the N-terminus,

including the PAAKK(10–14) segment, dramatically

alters the translocation pattern of dUTPase

Nuclear targeting potential of the dUTPase NLS

sequence and its truncated derivatives

To confirm that the complete putative NLS sequence

has nuclear targeting potential of its own, the

PAAKKMKID coding sequence was fused as an

N-terminal tag to YFP protein (Fig 3) The construct

(DUT-NLS–YFP) was transiently transfected into Sf9 cells After cell fixation, the expression and intracellu-lar localization of the fluorescent protein were observed by fluorescent microscopy The results show that this putative NLS sequence was able to confer nuclear localization to the YFP protein (Fig 4C and Table 3) DUT-NLS–YFP is found predominantly in the nuclear compartment, demonstrating that this sequence, which possesses a cluster of basic amino acids flanked by non-polar and acidic residues, is a powerful NLS

In order to identify amino acid residues that are essential for NLS function, we constructed trun-cated derivatives of the NLS sequence linked to the

A

B

C

D

E

F

G

H

I

(10–12,17–18) -YFP

(10–13,17–18) -YFP

Fig 4 Subcellular localization of the isoforms of D melanogaster dUTPase and its various NLS sequence derivatives Fluorescence micros-copy observations reveal the subcellular distribution of chimeric constructs (A,B) The long isoform of dUTPase (LD-DUT WT –YFP) was local-ized to the nucleus exclusively, and the short isoform (NTT-DUT WT –YFP) was present exclusively in the cytoplasm (C) NLS sequence studies show that, in the presence of the complete nuclear localization signal, the reporter construct DUT-NLS–YFP is located in the nucleus (D) Deletion of the first three residues (PAA), producing construct DUT-NLSD(10 )12)–YFP) slightly perturbed exclusive nuclear localization, with some cytoplasmic localization observed (E) Deletion of the first four residues (PAAK), producing the reporter construct DUT-NLSD(10 )13)–YFP, resulted in localization to the nucleus and the cytoplasm in an approximately equal ratio (F) The subcellular localization of the reporter construct DUT-NLSD(17 )18)–YFP, lacking the ID(17–18) dipeptide, was nuclear, with some infiltration into the cytoplasm (G) The DUT-NLSD(10 )12,17)18)–YFP construct, which lacks the tripeptide PAA and the ID(17–18) dipeptide, shows an almost equal distribu-tion in the nucleus and the cytoplasm (H) The subcellular targeting of the DUT-NLSD(10 )13,17)18)–YFP reporter was also not selective, showing close to equal distribution in the nucleus and the cytoplasm (I) YFP alone was used as a negative control The cellular distribution

of YFP was approximately the same within the nuclear and cytoplasmic compartments.

N-terminus of the YFP reporter In the first construct,

DUT-NLSD10)12–YFP, the neutral PAA tripeptide

was removed and the remaining part of the sequence,

KKMKID, was fused to the YFP reporter In the

sec-ond construct, the PAAK residues were deleted and

the KMKID stretch was fused to the reporter,

result-ing in the chimeric fluorescent construct

DUT-NLSD(10)13)–YFP After transfection and subsequent

fixation of Sf9 cells, the NLS potential of the

individ-ual truncated derivatives was monitored by fluorescent

microscope Observations show that deletion of the

PAA tripeptide slightly perturbs nuclear localization,

as cytoplasmic fluorescence was also observed

(Fig 4D) Although the PAA neutral tripeptide alone

may not define subcellular compartmentalization for

proteins, its position upstream of the short cluster of

basic residues may be essential to relax the secondary

structure of polypeptide chain, facilitating the

molecu-lar interaction with importins Removal of these three

non-basic residues of the dUTPase NLS resulted in

moderate perturbation of nuclear import and

accumu-lation In the truncated construct lacking the PAAK

segment, we observed greatly increased cytoplasmic

localization of the fluorescent reporter construct

(Fig 4E) This observation indicates that removal of

only one positively charged residue in addition to the

PAA tripeptide strongly alters recognition

characteris-tics within the nuclear import machinery

Furthermore, we established and examined three

additional NLS–reporter constructs lacking the ID(17–

18) dipeptide of the putative NLS sequence The

subcel-lular distribution of the DUT-NLSD(17)18)–YFP

construct was nuclear, with some infiltration in the

cyto-plasm (Fig 4F) The DUT-NLSD(10)12,17)18)–YFP

construct, which lacks the first PAA tripeptide, shows

an almost equal distribution within the nucleus and the

cytoplasm (Fig 4G) The subcellular targeting of the

third reporter construct, DUT–NLSD(10)13,17)18)–

YFP, which lacks the PAAK residues, was also close

to equal distribution between the nucleus and the

cyto-plasm (Fig 4H) These results indicate that the lack of

ID(17–18) might slightly decrease the exclusive nuclear

localization potential of the predicted NLS sequence

Additional oligopeptide deletions (PAA and PAAK)

have a further negative effect on the nuclear targeting

potential of the NLS sequence examined

Structural model of the Drosophila dUTPase NLS

segment in complex with importin a protein

Binding of the NLS segment to importin a has been

characterized by in-depth structural studies that allow

molecular insight into the specific interactions Based on

the published 3D structure of yeast importin a in com-plex with the c-Myc NLS segment peptide (PDB ID 1EE4) [34], and the close similarity between the NLS segments of c-Myc and Drosophila dUTPase (Table 2),

we modeled this latter peptide onto the c-Myc peptide

in the NLS peptide–yeast importin a structure Fig-ure 5A shows the alignment of the yeast and Drosophila importin a protein sequences, which show 69% similar-ity and 54% identsimilar-ity within the ten armadillo domains responsible for NLS recognition Figure 5A also shows the aligned sequence of a mammalian importin a (mouse importin a, which is 94% identical to the human sequence) (PDB ID 1IAL) [35] For mammalian impor-tins, 3D structures of complexes with other types of NLS peptides have been reported [36–38] The align-ments in Fig 5A show the high degree of conservation

of helical structure and residues interacting with NLS peptides Figure 5B shows the structural models of the two NLS peptides in complex with yeast importin a (c-Myc NLS peptide in turquoise, Drosophila dUTPase NLS peptide in green), indicating very close superposi-tion of the two NLS segments The close overlap is indi-cated by the observation that the two colors (green and turquoise) overlap considerably, and it is mostly the green color that is seen as the dUTPase NLS peptide was selected to be the ‘upper’ one in pymol Conse-quently, most of the molecular interactions are equally present in both NLS peptides Importantly, all impor-tin a amino acids that contain atoms within 4 A˚ of the NLS peptides (displayed in orange in Fig 5A,B) are conserved between the yeast and Drosophila importin a proteins, strengthening the assumption that the modeled recognition does take place in the physiological com-plex There are two noteworthy differences between the NLS peptides of c-Myc and D melanogaster dUTPase: lysine at position 14 in the dUTPase NLS is an arginine

in c-Myc, while methionine at position 15 in the dUTPase NLS is a valine in c-Myc With regard to the important role of the PAAK(10–13) segment in the NLS peptide, it is noteworthy that the e-NH2 group of the lysine residue at position 13 makes numerous con-tacts: it is within H-bonding distance to three oxygen atoms of conserved amino acids within importin a (the main-chain oxygen of glycine at position 168, the side-chain hydroxyl oxygen of threonine at position 173, and the side-chain carboxylate oxygen of aspartate at posi-tion 210; the numbering of the Drosophila sequence is used) However, the subsequent lysine residue at posi-tion 14 (arginine in c-Myc) cannot establish polar inter-actions with the carboxylate oxygen of aspartate at position 237 (the electrostatic bonding partner of the arginine residue in the c-Myc peptide) due to its shorter side chain The methinone residue at position 15,

Fig 5 Modeling the interactions between

the D melanogaster dUTPase NLS segment

and importin a protein (A) Sequence

align-ment for armadillo domains of Mus

muscu-lus (M mus.), D melanogaster (D mel.)

and yeast importin a Residues within the

a-helices constituting the armadillo domains

are shown on a pink background; residues

that contain atoms within 4 A ˚ of the NLS

peptides of c-Myc or D melanogaster

dUT-Pase (see Fig 5B) are on an orange

back-ground Asterisks indicate identical residues,

semicolons and dots show highly conserved

or conserved replacements, respectively.

Ten armadillo domains (ARM) are shown.

(B) Three-dimensional structural model of

the NLS peptide–importin a complex The

protein surface is shown for the first five

armadillo domains in either pink (for the

a-helices) or brown (for other protein parts).

The NLS peptides of c-Myc or D

melanog-aster dUTPase and importin a residues that

contain atoms within 4 A ˚ of the peptides

are shown as stick models with atomic

coloring (red, oxygen; blue, nitrogen; yellow,

sulfur; orange, green or turquoise, carbon

atoms of importin a, dUTPase NLS and

c-Myc NLS, respectively) For orientation,

most residues of the dUTPase NLS are

labeled, together with four residues of

importin a (see text for details) Note that

the dUTPase NLS peptide can adopt a

dock-ing conformation equivalent to that of the

c-Myc peptide on the importin protein

surface.