Identification of a nuclear localization motif in the serine/arginine protein kinase PSRPK of physarum polycephalum

Background Serine/arginine (SR) protein-specific kinases (SRPKs) are conserved in a wide range of organisms, from humans to yeast. Studies showed that SRPKs can regulate the nuclear import of SR proteins in cytoplasm, and regulate the sub-localization of SR proteins in the nucleus. But no nuclear localization signal (NLS) of SRPKs was found. We isolated an SRPK-like protein PSRPK (GenBank accession No. DQ140379) from Physarum polycephalum previously, and identified a NLS of PSRPK in this study. Results We carried out a thorough molecular dissection of the different domains of the PSRPK protein involved in its nuclear localization. By truncation of PSRPK protein, deletion of and single amino acid substitution in a putative NLS and transfection of mammalian cells, we observed the distribution of PSRPK fluorescent fusion protein in mammalian cells using confocal microscopy and found that the protein was mainly accumulated in the nucleus; this indicated that the motif contained a nuclear localization signal (NLS). Further investigation with truncated PSPRK peptides showed that the NLS (318PKKGDKYDKTD328) was localized in the alkaline Ω-loop of a helix-loop-helix motif (HLHM) of the C-terminal conserved domain. If the 318PKKGDK322 sequence was deleted from the loop or K320 was mutated to T320, the PSRPK fluorescent fusion protein could not enter and accumulate in the nucleus. Conclusion This study demonstrated that the 318PKKGDKYDKTD328 peptides localized in the C-terminal conserved domain of PSRPK with the Ω-loop structure could play a crucial role in the NLS function of PSRPK.


Background
Serine/arginine (SR) protein-specific kinases (SRPKs) represent a class of evolutionarily conserved kinases that specifically phosphorylate the arginine/serine-rich (RS) domains of the SR splicing factor [1]. After the identifica-tion of SRPK1-the first SRPK-by Gui et al. [2,3], other SRPKs such as SRPK2 [4], mouse SRPK1 and SRPK2 [5], yeast Dsk1 [6] and Sky1p [7], nematode SPK-1 [8], Trypanosoma cruzi TcSRPK [9], and Arabidopsis thaliana SRPK4 [10] have been subsequently identified. Some studies have shown that SRPKs are mainly localized in the cytoplasm, with only a few present in the nucleus [ [2,5], and [11]]. The SRPKs that are localized in the cytoplasm regulate the nuclear import of SR proteins via phosphorylation [12], while those in the nucleus regulate the nuclear localization of the SR splicing proteins via phosphorylation [13][14][15]. Ding et al. [11] have discovered that the spacer sequences present between the conserved domains of mammalian SRPKs have cytoplasmic anchoring function. However, there has been no report on the nuclear localization signal (NLS) of SRPKs.
Model organism Physarum polycephalum is mitochondriacontaining primitive eukaryotes. Its life cycle includes a single-celled amoeba, plasmodium (the main life form), and sporulation stages. The nuclei in the same plasmodium proliferate by way of synchronization of mitosis. In our previous study, we identified an SRPK containing 426 amino acids (aa) from P. polycephalum; this kinase was termed PSRPK (GenBank accession No. DQ140379). Similar to other SRPKs, PSRPK also has 2 conserved domains and can phosphorylate human SR protein alternate splicing factor/splicing factor 2 (ASF/SF2). However, it differs from other SRPKs in that the divergent motif ( N) in its N-terminus is rich in acidic amino acids; the spacer sequence of PSRPK between two conserved domains is shorter than that of other SRPKs. In this study, the distribution of PSRPK fluorescent fusion protein in mammalian cells was observed using laser scanning confocal microscopy. Maximum fluorescence was detected in the nucleus, suggesting the existence of an NLS in PSRPK. When the distribution of truncated PSPRK peptides was observed, a putative NLS was found in the C-terminal of PSRPK based on its homology to the classic NLS. When the sequence was deleted or when lysine 320 in the sequence was substitute by threonine, the cytoplasmic distribution of PSRPK was observed.

Construction of expression plasmids containing PSRPK and truncated PSPRK peptides
pMD18-psrpk had been constructed previously [16]. PSRPK fragments were obtained by polymerase chain reaction (PCR), using pMD18-psrpk as the template and PSRPK-specific primers ( Table 1). The resulting PCR fragments and the mammalian expression vectors pDsRed1-N1 and pECFP-C1 (Clontech) were digested with EcoRI and BamHI. The digested fragments were subjected to agarose gel electrophoresis, recovered from the gel, purified, and ligated using T4 DNA ligase; this resulted in the expression plasmids pDsRed-psrpk and pECFP-psrpk.

Construction of expression plasmids containing mutant and default PSRPK peptides
In order to understand the effect of the loop motif on nuclear localization, we designed default PSRPK, namely, PSRPK d ( 318 PKKGDK 323 deletion) and mutant PSRPK, namely, PSRPK m (K 320  T 320 mutation). The DNA fragments containing psrpk d or psrpk m up-and downstream were obtained using the primers listed in Table 2 and pMD18-psrpk as the template. The 2 overlapping PCR products were mixed in a ratio of 1:1, followed by denaturing at 94°C for 3 min and annealing at 55°C for 3 min.
Since there were 18 complementary bases (underlined parts of primers, Table 2) in the 2 PCR fragments, they would be complementary to the cohesive ends during annealing at 55°C. The DNA with 5' and 3' overhangs could be extended by DNA polymerase (Hot Start Taq Polymerase; Qiagen) to form a complete DNA fragment at 72°C. Primer pair F4/R3 were introduced in the PCR reaction for the amplification of the psrpk d and psrpk m sequences. The 2 PCR products, psrpk d and psrpk m , were inserted into pDsRed1-N1, resulting in the recombinant plasmids pDsRed-psrpk d and pDsRed-psrpk m , respectively.

Transfection of mammalian cells with lipofectamine
The abovementioned recombinant plasmids were used to transform E. coli DH5a cells. The positive recombinant products were grown in Luria-Bertani (LB) medium containing 30 g/ml kanamycin. The plasmids were isolated using the alkaline lysis method, dissolved in Tris-EDTA (TE) buffer, and quantified using GeneQuant pro (Amersham Bioscience, GE Healthcare). The purified plasmid (A 260 /A 280 > 1.8) was diluted to 16 g/ml in serum-free RPMI-1640 medium (Gibco). Lipofectamine™ 2000 (Invitrogen) was diluted in serum-free RPMI-1640 medium (40 l/ml) and then mixed with an identical volume of plasmid solution, resulting in lipid-DNA complexes.
HeLa and L929 cells (ATCC) were cultured in RPMI-1640 medium supplemented with 10% fetal bovine serum (FBS; Hyclone), 100 IU/ml penicillin, and 100 g/ml Distribution of fluorescent fused PSRPK in mammalian cells streptomycin at 37°C in a 5% CO 2 incubator. The cells were harvested while they were in the logarithmic phase and 1.5 ml culture (approximately 1 × 10 5 cells) was seeded into a 35-mm plate with a coverslip and cultured for 24 h. The medium was replaced with serum-free RPMI-1640 medium, and the cells were cultured for 1 h to initiate transfection. The cells were overlaid with 200 l lipid-DNA complexes and cultured for 5 h. The medium was replaced with 1.5 ml of 10% FBS medium, and the cells were further cultured for 48 h.

Three-Dimensional Structure Modeling of PSRPK
The three-dimensional (3-D) PSRPK structure was modeled by the Swiss Institute of Bioinformatics program SWISS-MODEL, available at http://swiss model.expasy.org/, and the results analyzed by the Deep-View program [17].

PSRPK was mainly localized in the nucleus of mammalian cells
The DAPI-stained nucleus of cells HeLa and L929 transfected with pECFP-psrpk were clearly observed under UV excitation (Figure 2a, b). Confocal microscopy showed that the distribution of cyan fluorescent protein (CFP)-PSRPK was non-uniform and that this protein was Schematic diagrams of truncated PSRPKs compared with PSRPK

Existence of an NLS on the C-terminal conserved domain of PSRPK
The expression and distribution of CFP fusion TP1, TP2, TP3, and TP4 in mammalian cells were observed by con-focal microscopy. Compare with the DAPI staining of the nucleus (Figure 3a ure 3g, i) and L929 (Figure 3h, j) cells were similar to those of CFP-PRPK and CFP-TP1, i.e., both mainly accumulated in the nucleus. This indicated that PSRPK continued to accumulate in the nucleus despite the deletion of  N, CD1, and the spacer sequence and that CD2 contains an NLS peptide. Unlike the distributions of CFP-PSRPK, CFP-TP1, CFP-TP3, and CFP-TP4, CFP-TP2 in HeLa (Figure 3e) and L929 (Figure 3f) cells exhibited diffuse distribution. The results above indicated that the  N, CD1, and spacer sequence deletions did not influence nuclear localization of PSRPK and that an NLS was located in the C-terminal conserved domain of PSRPK.

PSRPK NLS was located in the 318 PKKGDKYDKTD 328 sequence
Confocal microscopy revealed that RFP-TP5 and RFP-TP7 were uniformly distributed in the L929 cells (Figure 4b, d), indicating that there was no NLS in the TP5 and TP7 sequences. However, RFP-TP6 mainly accumulated in the nucleus (Figure 4c), indicating that an NLS existed in TP6. By comparing the amino acid sequences of TP5, TP6, and TP7, we primarily confirmed that PSRPK NLS was located in 318 PKKGDKYDKTD 328 . RFP-TP8 and RFP-TP9 were mainly distributed in the cytoplasm (Figure 4e, f); this further indicated that PSRPK and TP3 lost the ability of nuclear localization after the deletion of the abovementioned sequence. Thus, we could confirm that 318 PKKGDKYDKTD 328 was the NLS sequence of PSRPK.

The NLS sequence of PSRPK contained a -loop motif
Analysis of the secondary structure of 318 PKKGDKYDKTD 328 within PSRPK by DNASIS v2.5 Demo revealed that it was a -turn ( Figure 5A). Based on the crystal structure of SRPK1 and Sky1p [17], the tertiary structure of PSRPK ( Figure 5B) was predicted using SWISS-MODEL software [18][19][20]. Figure 5B shows that the A: Secondary structure of PSRPK NLS, as predicted by DNASIS v2.5 Demo Figure 5 A: Secondary structure of PSRPK NLS, as predicted by DNASIS v2.5 Demo. The capital H, S and T blow the sequence represent the corresponding amino acid above has more probability to form helix, sheet or turn. B: The overall structure of PSRPK that mimics the structures of SRPK1 and Sky1p, as predicted by SWISS-MODEL online and displayed by SWISS-Pdb viewer compared with the overall structure of SRPK1 and Sky1p [17][18][19][20]. C: NLS, the NLS of PSRPK; NLS d , the NLS of PSRPK without 318 PKKGDK 323 ; NLS m , the NLS of PSRPK with the K 320  T 320 mutation. structure 318 PKKGDKYDKTD 328 mimics that of the corresponding sequences of SRPK1 and Sky1p. All the 3 sequences are located in the C-terminal loop motif. The spatially near K 320 and D 325 may form a salt-bridge via their side chains, causing the 318 PKKGDKYDKTD 328 to form a stable -loop. This structure of 318 PKKGDKYDKTD 328 changes after 318 PKKGDK 323 deletion or K 320  T 320 mutation, as predicted by SWISS-MODEL software ( Figure 5C). 318 PKKGDK 323 deletion causes damage to the -loop motif, and K 320  T 320 mutation damages the salt-bridge in the -loop.
The fluorescent fusion proteins PSRPK d and PSRPK m in the L929 cells are shown in Figure 6. Compare with the DAPI staining of the nuclear (Figure 6B1a, B2a), the fluorescent signal of RFP-PSRPK d and RFP-PSRPK m was observed in cytoplasm (Figure 6B1b, B2b). Further, confocal microcopy revealed that RFP-PSRPK d and RFP-PSRPK m were mainly distributed in the cytoplasm ( Figure  6B1c, B2c); this indicated that the 318 PKKGDK 323 deletion or K 320  T 320 mutation destroyed the structure of PSRPK NLS, resulting in PSRPK losing its ability of nuclear localization. Ding et al. [11] studied the cell localization of mammalian SRPKs and found that almost all SRPKs without spacer sequences are accumulated in the nucleus, indicating that there is a cytoplasm localization signal in the spacer sequence between the conserved domains. Kuroyanagi et al. [5] speculated that mouse SRPK1 might have 2 potential NLSs that are located in 11~21 aa and 265~277 aa; thus, mSRPK2 may have a potential NLS in 264~276 aa. However, direct evidence of NLS sequences in SRPK family members has been lacking thus far. Laser scanning confocal microscopy revealed that RFPs of default PSRPKs containing 318 PKKGDKYDKTD 328 mainly accumulated in the nucleus of mammalian cells, while the PSRPKs in which the abovementioned sequence was deleted did not accumulate in the nucleus. This indicated that the NLS of PSRPK was located in the 318 PKKGDKYDKTD 328 sequence of the C-terminal conserved domain.

PSRPK has an NLS sequence
Further, Ding et al. [11] showed that mammalian SRPKs were mainly distributed in the cytoplasm. However, the A: Schematic diagrams of default PSRPK and mutant PSRPK compared with PSRPK results obtained in our study showed that PSRPK was mainly expressed in the nucleus of the mammalian cells. A comparison of the primary structure of PSRPK and other SRPKs revealed that the conserved domains were almost identical; however, the nonconserved  N and spacer sequence differed among the kinases. A cytoplasm localization signal was present in the spacer sequences of mammalian SRPKs. Further, the spacer sequence of PSRPK was considerably smaller than that of mammalian SRPKs. This difference might be the major reason why PSRPK cannot anchor itself in the mammalian cytoplasm.

Function of the PSRPK NLS is related to the loop motif
A classic NLS (cNLS) comprises a monopartite or bipartite signal. A monopartite NLS contains 1 cluster of basic residues, while a bipartite NLS contains 2 clusters of basic residues [21]. Similar to the sequence feature of a monopartite NLS, the NLS of PSRPK is rich in basic residues. A monopartite NLS sequence that is located at the terminal of a protein is usually in a coil, while that located in the interior of a protein is usually in a loop. The PSRPK NLS is also located in the basic loop between 2 -helixes ( Figure 5). The close side chains of K 320 and D 325 form a salt bridge, thus forming a stable NLS -loop motif (Figure 5C). The experimental results showed that PSRPK lost its nuclear localization ability following the deletion of 318 PKKGDK 323 or mutation from K 320 to T 320 . The deletion or mutation destroyed the -loop motif of PSRPK, thus suggesting that the loop structure of PSRPK NLS controls the nuclear localization of PSRPK. The 318 PKKGDKYDKTD 328 sequence, which corresponds to the nonconserved sequences of SRPK1 and Sky1p, is located in the loop of an HLHM [18,19] (Figure 5B). Therefore, studies on NLSs of SRPKs should focus on the loops in HLHMs.

Conclusion
In this study, by truncation of PSRPK protein, deletion of and single amino acid substitution in a putative NLS and transfection of mammalian cells, we demonstrated that the 318 PKKGDKYDKTD 328 peptides localized in the C-terminal conserved domain of PSRPK with the -loop structure could play a crucial role in the NLS function of PSRPK.