Evolutionary and Transcriptional Analysis of Karyopherin β Superfamily Proteins

Abstract

In eukaryotes, karyopherin β superfamily proteins mediate nucleocytoplasmic transport of macromolecules. We investigated the evolutionary and transcriptional patterns of these proteins using bioinformatics approaches. No obvious homologs were found in prokaryotes, but an extensive set of β-karyopherin proteins was found in yeast. Among 14 β-karyopherins of Saccharomyces cerevisiae, eight corresponded to their human orthologs directly without diversification, two were lost, and the remaining four proteins exhibited gene duplications by different mechanisms. We also identified β-karyopherin orthologs in Caenorhabditis elegans, Drosophila melanogaster, Danio rerio, Xenopus tropicalis, Gallus gallus, and Mus musculus. β-Karyopherins were ubiquitously but nonuniformly expressed in distinct cells and tissues. In yeast and mice, the titer of some β-karyopherin transcripts appeared to be regulated both during the cell cycle and during development. Further virtual analysis of promoter binding elements suggested that the transcription factors SP1, NRF-2, HEN-1, RREB-1, and nuclear factor Y regulate expression of most β-karyopherin genes. These findings emphasize new mechanisms in functional diversification of β-karyopherins and regulation of nucleocytoplasmic transport.

Nucleocytoplasmic transport of proteins, RNAs, and ribosomes is essential in eukaryotic cells. In this process, the β-karyopherins (or importin-β superfamily members) play a central role as they transport cargoes across the nuclear pore complex (1–9). Members of this family of 95–145-kDa proteins have a so-called importin-β N-terminal (IBN_N)¹ domain (or RanGTP binding domain) at their N terminus and several “HEAT repeat” motifs that mostly occupy the C-terminal portion of the structure. The HEAT repeat motifs are able to assume different conformations in different functional states. The flexibility of these HEAT repeat motifs facilitates the accommodation of their binding partners by an induced fit type of mechanism (10). These features distinguish β-karyopherins from transporters that deliver cargoes to other subcellular organelles. β-Karyopherins have been found from yeast to humans. In Saccharomyces cerevisiae,14 β-karyopherins have been shown to transport corresponding cargoes in and out of the nucleus (11, 12), whereas in mammalian cells, about 20 β-karyopherins participate in these events (7). The family includes both import and export receptors.

Nuclear proteins larger than the “diffusion limit” of the nuclear pore complex (∼50 kDa) generally have a nuclear localization signal for nuclear import, whereas smaller proteins that are actively excluded from the nucleus have a nuclear export signal (3). The transport of these cargoes is mediated by karyopherins in conjunction with the small GTPase Ran, which cycles between its GDP-bound (in the cytoplasm) and GTP-bound (in the nucleus) forms. RanGTP confers directionality upon transport by controlling the conformation of β-karyopherins and their ability to bind cargo (13, 14). Although most karyopherins are constitutively expressed in cells (6), emerging evidence shows that expression of at least some karyopherin genes is regulated (15, 16) presumably due to varying requirements for transport, e.g. during development (17). For example, expression of Kpnb1 and Ranbp5 is regulated during rodent spermatogenesis (18). Moreover the expression of IPO13 in rats and humans is regulated both hormonally and developmentally (16, 19). Although β-karyopherins have common features, the biological importance of their functions in nuclear transport is impressively varied. A little explored aspect of β-karyopherin biology is their evolution. In particular, it is unclear how the 14 yeast β-karyopherins gave rise to 20 members in human. Moreover it is unknown whether the cell type expression of β-karyopherin genes reflects corresponding functional requirements.

To address the above questions, we used bioinformatics approaches. We also investigated the expression of each β-karyopherin member through the systematic analysis of transcriptional profiles and identification of putative transcription factor binding sites.

EXPERIMENTAL PROCEDURES

Identification of Orthologous Proteins—The assignment of orthologous proteins is critical for evolutionary analyses of β-karyopherins. For this purpose, the orthologous proteins were searched and compared in eight species: Homo sapiens, Mus musculus, Gallus gallus, Danio rerio, Xenopus tropicalis, Drosophila melanogaster, Caenorhabditis elegans, and S. cerevisiae. Each assignment of orthologs was made on the basis of multiple lines of evidences. 1) As the fundamental method, the putative orthologs were identified or verified by BLAST searching against assembled genomes, adopting the default parameters of the National Center for Biotechnology Information (NCBI) blastp for sequence alignment. 2) Most β-karyopherin orthologs can be identified by searching the NCBI Clustering of Orthologous Groups (COG) database (20). 3) The “Orthologue Prediction” function of the Ensembl database also helps to identify putative orthologs of β-karyopherins in whole genomes (21).

Construction of Phylogenetic Trees—The protein sequences and exon information of karyopherin β superfamily members of eight organisms were derived from either the Ensembl genome database release 46 (21) or GenBank^™. Corresponding protein sequences were extracted from GenBank and Ensembl (supplemental Table 1). To illustrate the evolution of β-karyopherins from yeast to human, phylogenetic trees were generated. Multiple sequence alignment on protein sequences was first demonstrated using MUSCLE3.6 (22), then refined manually with consideration of the secondary structure, and finally visualized by ClustalX 1.83 (23). 14 individual phylogenetic trees (Fig. 1) were then constructed based on the multiple alignments using the neighbor-joining method provided by the software MEGA4 (24) under the Poisson correction amino acid substitution model with uniform rates among sites (25, 26). Bootstrap analysis was conducted using 1,000 replicates to test the robustness of these phylogenetic trees.

Fig. 1.

Phylogenetic trees constructed on the full-length protein sequences of karyopherin β superfamily members in eight organisms using the neighbor-joining methods provided by the software MEGA4 under the Poisson correction amino acid substitution model with uniform rates among sites. Bootstrap values calculated by MEGA4 are beside their nodes. The β-karyopherins are arranged into 14 trees as shown. Yeast β-karyopherin genes are marked with solid circles. The names of species are abbreviated as follows: Hs, H. sapiens; Mm, M. musculus; Gg, G. gallus; Xt, X. tropicalis; Dr, D. rerio; Dm, D. melanogaster; Ce, C. elegans; and Sc, S. cerevisiae.

Investigation of Transcription Patterns—The gene expression microarray data sets GNF1H_GCRMA and GNF1M_GCRMA used in this study were retrieved from GNF SymAtlas (27), which covers 24,277 human genes in 79 tissues and 32,905 mouse genes in 61 tissues. The gene expression patterns of β-karyopherins were analyzed quantitatively as described in previous work (28). Geometric comparison (similarity measure) was used to indicate how similar two gene expression profiles are. A value of similarity measure close to 1 indicates high similarity of gene expression patterns regardless of their expression levels. Such genes may have related biological roles. Specificity measure (SPM) was calculated to compare expression in distinct tissues (28). Such information is helpful for understanding the physiological significance of genes. Three different microarray data sets of S. cerevisiae (based on treatment with α factor and analysis of cdc15 and cdc28 mutants) (29, 30) were adopted to investigate possible cell cycle-dependent expression of β-karyopherins.

Investigation of Regulatory Elements—400 bp of nucleotide sequences from −300 to +100 bp of the transcript start of β-karyopherin genes in eight species were derived from the Ensembl database (21) or GenBank for the transcription factor binding sites (TFBSs) analysis. The MAPPER search engine is a program developed to aid the molecular biologist in determining what eukaryotic transcription factor binding elements may exist in a given DNA sequence (31). In this study, the putative TFBSs were identified by screening the upstream regulatory sequences of β-karyopherin genes. Default score and E-value thresholds (0 and 10, respectively) of the MAPPER search engine were specified. To reduce the false positives, only species-specific TFBS models were adopted for predictions, e.g. TFBS models built from human, mouse, or fly data were only applied to human, mouse, or fly promoter sequences, respectively. These predictions were further combined with information on gene regulation to identify possible regulatory patterns.

RESULTS AND DISCUSSION

Evolution of the Karyopherin β Superfamily

The Sequence Alignments—Previous studies have identified 14 β-karyopherins in bakers’ yeast (S. cerevisiae) and about 20 in human (32). A few studies that have investigated β-karyopherin divergence and multiplication during evolution were based on limited data (33, 34). Through ortholog analyses, 138 β-karyopherin genes (132 annotated and six novel) were identified in human (20 genes), S. cerevisiae (14 genes), and six other organisms: M. musculus (20 genes), G. gallus (17 genes), X. tropicalis (18 genes), D. rerio (23 genes), D. melanogaster (16 genes), and C. elegans (10 genes) (Table I). The IDs of these sequences in Ensembl or GenBank can be found in the supplemental Table 1.

Table I. Summary of β-karyopherin genes of eight organisms, including H. sapiens, M. musculus, G. gallus, X. tropicalis, D. rerio, D. melanogaster, C. elegans, and S. cerevisiae.

Orthologous gene groups are separated by shading. Note that IPO7 and IPO8 may come from the same common ancestor gene, NMD5, in S. cerevisiae; IPO13 and TNPO3 may come from the same common ancestor gene, MTR10, in S. cerevisiae; RANBP6 and RANBP5 may come from the same common ancestor gene, PSE1, in S. cerevisiae; TNPO1 and TNPO2 may come from the same common ancestor gene, KAP104, in S. cerevisiae; and XPO7 and RANBP17 may come from the same common ancestor gene, C35A5.8, in C. elegans. Slashes represent no homologous genes found.

S.cerevisiae	C.elegans	D.melanogaster	D. rerio	X.tropicalis	G.gallus	M.musculus	H.sapiens
CSE1	IMB-5	CAS	CSE1L	CSE1L	CSE1L	CSE1L	CSE1L
KAP123	/	CG32164 CG32165	IPO4	IPO4	/	IPO4	IPO4
NMD5	/	MSK	IPO7	IPO7	IPO7	IPO7	IPO7
			IPO8a IPO8b	IPO8	IPO8	IPO8	IPO8
KAP114	/	RANBP9	IPO9	IPO9	IPO9	IPO9	IPO9
KAP120	/	RANBP11	IPO11	IPO11	IPO11	IPO11	IPO11
MTR10	T16G12.6	CDM	IPO13	IPO13	IPO13	IPO13	IPO13
	TSR-1	TRN-SR	TNPO3	TNPO3	TNPO3	TNPO3	TNPO3
KAP95	IMB-1	FS(2)KET	KPNB1	KPNB1	KPNB1	KPNB1	KPNB1
PSE1	IMB-3	KARYBETA3	RANBP5	RANBP5	RANBP5	RANBP5	RANBP5
						RANBP6	RANBP6
KAP104	IMB-2	TRN CG8219	TNPO1	TNPO1	TNPO1	TNPO1	TNPO1
			TNPO2a TNPO2b	TNPO2	/	TNPO2	TNPO2
CRM1	IMB-4	EMB	XPO1a XPO1b	XPO1	XPO1	XPO1	XPO1
/	Y69A2AR.16	/	XPO4a XPO4b	XPO4	XPO4	XPO4	XPO4
MSN5	/	RANBP21	XPO5	XPO5	XPO5	XPO5	XPO5
/	/	EXP6	XPO6	XPO6	XPO6	XPO6	XPO6
/	C35A5.8	RANBP16	XPO7a XPO7b	XPO7	XPO7	XPO7	XPO7
					RANBP17	RANBP17	RANBP17
LOS1	IMB-6	/	XPOT	XPOT	XPOT	XPOT	XPOT
SXM1	/	/	/	/	/	/	/
KAP122	/	/	/	/	/	/	/

Multiple sequence alignment of full-length protein sequences indicated the extensive diversity of β-karyopherins within a species (data not shown). Because the full-length sequence alignment within a species cannot provide an inclusive description of β-karyopherin evolution, local structural clues were searched to provide complementary information. The IBN_N domain (Pfam PF03810) plays a pivotal role in cargo release and thus carries distinctive “structural signatures” of the karyopherin β family (33, 35). We then compared the IBN_N domains of β-karyopherins within a species. The conservation of IBN_N domains within a species is comparatively poor (Fig. 2). This diversity suggests that β-karyopherins may not all be derived from one or a small number of ancestors. In other words, it is inappropriate to probe the kinship of β-karyopherins on the basis of multiple sequence alignment of the IBN_N domain or full-length sequence within a species. Exceptions were found in some pairs of β-karyopherins that share domain composition, length, and start sites (e.g. IPO7-IPO8, IPO13-TNPO3, RANBP5-RANBP6, TNPO1-TNPO2, and XPO7-RANBP17), suggesting that they are paralogous pairs. Although it is difficult to identify an altogether constant IBN_N domain among human β-karyopherins, several amino acid residues in this region are relatively conserved (Val/Ile/Leu⁴⁶, Val/Ile/Leu⁷⁴, and Lys/Arg⁷⁵) (Fig. 2). These amino acids may be critical to maintain domain structure and bind RanGTP. The comparison of IBN_N domains among species was also undertaken using KPNB1 proteins as an example (Fig. 3). By contrast, the IBN_N domain shows strong conservation among species, especially after D. melanogaster, which agrees well with the results of multiple sequence alignment of full-length proteins (supplemental materials).

Fig. 2.

Comparison of the IBN_N domain (Pfam PF03810) in human β-karyopherins. Multiple sequence alignment of the IBN_N domain residues was performed using MUSCLE3.6 and refined manually on the basis of a reference alignment (importin-β N-terminal domain profile, PROSITE PS50166) and the secondary structure of human KPNB1 (Protein Data Bank 1QGR) (79). The numbers under “Start site” are positions of the first residues of IBN_N domain in each protein; numbers under “Length” are the number of residues in the IBN_N domain in each protein. A conservation estimate for each position in the alignment is plotted under the alignment. Highly conserved positions in the alignment will get a high score (the peaks), whereas low conservation or exceptional residues at a partially conserved position will lower the score (the valleys). The “:” character indicates that one of the “strong” groups of amino acids is fully conserved, whereas “.” indicates that one of the “weaker” groups of amino acids is conserved as described in ClustalX. Consensus secondary structure predicted using the software JPred (80) is given above the alignment. Note the conservation of the IBN_N domain in β-karyopherin pairs IPO7-IPO8, IPO13-TNPO3, RANBP5-RANBP6, TNPO1-TNPO2, and XPO7-RANBP17. Sequences were shaded based on the alignment consensas, which was calculated automatically by the ClustalX program.

Fig. 3.

Comparison of the IBN_N domain (Pfam PF03810) in KPNB1 orthologs.Hs, H. sapiens; Mm, M. musculus; Gg, G. gallus; Xt, X. tropicalis; Dr, D. rerio; Dm, D. melanogaster; Ce, C. elegans; Sc, S. cerevisiae. The sequences were aligned using MUSCLE3.6, and the definitions of the symbols are described in Fig. 2. “*” indicates positions that are fully conserved. RanGTP binding sites for yeast Kap95p as described previously (81) are labeled with arrows. Note the conservation of the IBN_N domain through different organisms especially after D. rerio.

The Ancestors ofβ-Karyopherins—On the basis of the multiple sequence alignments of full-length protein sequences, the β-karyopherins were allocated to 14 separate trees with high confidence (with bootstrap values ≥50%). They are CSE1L, IPO4, IPO7-IPO8, IPO9, IPO11, IPO13-TNPO3, KPNB1, RANBP5-RANBP6, TNPO1-TNPO2, XPO1, XPO4-XPO7-RANBP17, XPO5, XPO6, and XPOT. It is interesting that these 14 phylogenetic trees do not all correspond to ancestors in S. cerevisiae. As summarized by phylogenetic analyses (Fig. 1) and ortholog analyses (Table I), only 12 yeast β-karyopherin genes have orthologs in multicellular organisms judging from multiple sequence alignments and phylogenetic analysis. The other two genes, KAP122 and SXM1,do not have orthologs. Actually before the diversification of β-karyopherins stabilized, gain and loss of genes was common. Thus, seven β-karyopherin genes of S. cerevisiae (KAP114, KAP120, KAP122, KAP123, MSN5, NMD5, and SXM1) are absent in C. elegans, although two new genes were gained: C35A5.8 (the possible ancestor of human RANBP17 and XPO7) and Y69A2AR.16 (ancestor of human XPO4). Gain and loss were also observed in D. melanogaster where Y69A2AR.16 (ancestor of human XPO4) and LOS1 (ancestor of human XPOT) are absent, and a new β-karyopherin gene, Exp6 (ancestor of human XPO6), emerges. These changes presumably equip distinct species with an ample but minimal repertoire of carriers to transport critical cargoes.

Sequence analysis of seven other yeasts, Candida glabrata (14 genes), Debaryomyces hansenii (13 genes), Eremothecium gossypii (14 genes), Kluyveromyces lactis (13 genes), Yarrowia lipolytica (12 genes), Schizosaccharomyces pombe (13 genes), and Cryptococcus neoformans (nine genes), and two plants, Arabidopsis thaliana (17 genes) and Oryza sativa (14 genes), showed that the absence of selected β-karyopherin genes is not rare during evolution. The causes of gene loss may be related to unique aspects of physiology, behavior, and development. As no new types of β-karyopherin genes were detected in D. rerio (although there are duplicates of some known β-karyopherin genes), the diversification and amplification of β-karyopherin genes became stable no later than D. rerio. However, this does not mean that no further gain or loss of genes occurred. For example, orthologs of IPO4 and TNPO2 are absent in G. gallus.

It is still a mystery why no very rudimentary set of “urkaryopherins” has been detected. However, simplified sets of karyopherins have been found in some protozoa, e.g. Giardia lamblia and in C. elegans. G. lamblia has only four β-karyopherin genes (36). It is not possible to estimate how few β-karyopherins might be sufficient for survival because knowledge of their transport specificity is incomplete and because there is redundancy: some cargoes can be recognized and transported by more than one β-karyopherin, and many β-karyopherins recognize multiple cargoes. Nevertheless it is reasonable to postulate that the presence or absence of some β-karyopherins is critical for survival for all eukaryotes.

To search for possible ancestors in prokaryotes, we BLAST searched both the RNA and protein sequences of all yeast β-karyopherins against 327 bacterial and 27 archaebacterial genomes. Mild similarity (sequence identity, 26%) was found for Sxm1p to an ∼143-amino acid fragment (fragment of sensory transduction histidine kinase in Clostridium acetobutylicum). Other β-karyopherins also show mild similarity (no more than 30% sequence identity) to some short prokaryotic sequences, many of which are fragments of uncharacterized hypothetical proteins. These sequences are remotely homologous to HEAT repeats.

Gene Duplications—The expansion of the karyopherin β family is accompanied by repeated gene amplification. Sequence alignment shows that there are five pairs of human β-karyopherins (10 proteins) that closely resemble each other: IPO7-IPO8, IPO13-TNPO3, RANBP5-RANBP6, TNPO1-TNPO2, and XPO7-RANBP17. Because each pair of β-karyopherins may be derived from a common ancestor, exon analysis within species and among species was undertaken (Figs. 4 and 5 and the supplemental materials). We found that these β-karyopherin genes are well conserved over species judging from both the length and nucleotide acid composition of each exon. Such conservation can be traced back to the earliest common ancestor, the PSE1 gene of S. cerevisiae. Comparison of exon arrangements among β-karyopherin genes indicates that for IPO7-IPO8, TNPO1-TNPO2, and XPO7-RANBP17 both members of each pair have almost the same exon number, length, and order (Fig. 4). This observation strongly supports the hypothesis that these β-karyopherin pairs are derived from the same ancestors by gene duplication.

Fig. 4.

The exon arrangements of 20 human β-karyopherin genes. The transcript IDs in Ensembl are ENST00000261412, ENST00000356861, ENST00000379719, ENST00000256079, ENST00000252512, ENST00000377354, ENST00000265388, ENST00000372343, ENST00000290158, ENST00000354464, ENST00000361565, ENST00000389352, ENST00000389538, ENST00000255305, ENST00000265351, ENST00000304658, ENST00000389161, ENST00000262982, ENST00000357602, and ENST00000259569, respectively. The shaded blocks are translated regions, whereas the transparent blocks are non-translated regions. The numbers under the blocks are the length of the exon (in bp), and the numbers in parenthesis indicate the length of translated regions when an exon contains a non-translated region. Note the similarity between TNPO1-TPNP2, IPO7-IPO8, and XPO7-RANBP17. RANBP6 has only one exon and is thought to originate from RANBP5 through retroposition.

Fig. 5.

The exon arrangements of RANBP5 and RANBP6 in different organisms, including H. sapiens (Hs), M. musculus (Mm), G. gallus (Gg), X. tropicalis (Xt), D. rerio (Dr), D. melanogaster (Dm), C. elegans (Ce), and S. cerevisiae (Sc). The transcript IDs of these β-karyopherin genes in Ensembl or GenBank are ENST00000357602, ENSMUST00000032898, XM_416978.2, ENSXETT00000050713, XM_692846.2, CG1059-RA, and C53D5.6 YMR308C for RANBP5 and ENST00000361966 and ENSMUST00000046742 for RANBP6. The shaded blocks are translated regions, whereas the transparent blocks are non-translated regions. The numbers under the blocks are the length of exons, and numbers in parentheses are the length of translated regions when an exon contains a non-translated region. Note the conservation of exon arrangements of RANBP5 after D. rerio and the single exon RANBP6 in human and mouse after retroposition.

Pairs of duplicated β-karyopherin genes were also found in D. melanogaster (CG32164-CG32165 and Trn-CG8219) and in D. rerio (ipo8a-ipo8B, tnpo2a-tnpo2b, xpo1a-xpo1b, xpo4a-xpo4b, and xpo7a-xpo7b) (Table I). These gene pairs exhibit high identity of protein sequence: 99.9% for ipo8a-ipo8b, 97.7% for CG32164–CG32165, 94.8% for xpo1a-xpo1b, 91.9% for tnpo2a-tnpo2b, 84.1% for xpo7a-xpo7b, 68.6% for Trn-CG8219, and 63.9% for xpo4a-xpo4b. There nevertheless is no evidence that the duplications persist in higher organisms, and some duplicated β-karyopherin gene pairs may diverge to the point that they acquire specialized neofunctions (37). Thus gene duplication in D. melanogaster and D. rerio could in fact contribute to speciation (37).

Retroposition—As discussed previously, gene duplication is one of several mechanisms that could have made it possible for the karyopherin β family to expand. RANBP5 and RANBP6 are very closely related judging from sequence alignment and ortholog analysis. To draw the evolutionary path of RANBP5 and RANBP6, their exon arrangement was studied. Fig. 5 shows, remarkably, that the exon number grows from one exon for PSE1, the putative ancestor of human RANBP5 in S. cerevisiae, to 26 exons in D. rerio and finally to 29 exons in H. sapiens. The multiplication of exons presumably reflects the high phenotypic complexity of mammals (38). The enormous increase of the untranslated regions (0 bp in S. cerevisiae to about 2,709 bp in human) is likely to contribute to regulation of gene transcription (39, 40). However, unlike RANBP5 and other β-karyopherins, the RANBP6 gene has only one exon in both human and mouse (Fig. 5). Further BLAST searches against the Ensembl genome database found that RANBP6 orthologs exist only in mammals. We therefore suggest that the emergence of RANBP6 may result from the retroposition of RANBP5 gene, which leads to production of a new functional pseudogene with only a single exon (37). Judging from known RANBP6 orthologs, the retroposition of RANBP6 from ancient RANBP5 may have happened before the divergence of mammals and birds. This inference is supported by the previous evidence that RANBP6 is a functional retrocopy of RANBP5(41). Additional ortholog analysis found that some RANBP6 orthologs contain several exons, e.g. Canis familiaris (three exons), Oryctolagus cuniculus (six exons), and Sorex araneus (12 exons). Because no ortholog was found other than in mammals, it is reasonable to conclude that the retroduplication event happened in the speciation of mammals. After the retroposition, the RANBP6 is further diversified in mammals: introns are species-specifically inserted into the only exon. This may explain the multiple exons of RANBP6 orthologs. Similar phenomena have been reported previously (42, 43).

In conclusion, karyopherin β superfamily proteins are not found in prokaryotes. β-Karyopherin evolution is suggested to proceed along 14 lines from yeast, although two genes (KAP122 and SXM1) are lost as summarized in Table I. Some yeast β-karyopherins (KAP95, KAP114, KAP120, KAP123, CSE1, MSN5, CRM1, and LOS1) can be directly linked to their orthologs in man (KPNB1, IPO9, IPO11, IPO4, CSE1L, XPO5, XPO1, and XPOT), respectively. Other β-karyopherins (PSE1, NMD5, MTR10, and KAP104) underwent diversification during their evolution.

The selective pressure for efficient communication between the nucleus and the cytoplasm likely prompts the multiplication and diversification of karyopherins to such an extent that only weak sequence homology is often observed between β-karyopherins in a species. Considering the evolutionary history and functional importance of β-karyopherins, reclassification of the karyopherin β superfamily may become necessary. Classification should consider not only their overall structural characteristics but their entire sequence, their cargo specificity, and their interactions with both Ran and nucleoporins.

Expressions of the Karyopherin β Superfamily in Cells and Tissues

The functional importance of β-karyopherins determines their expression. Motivated by this premise, we therefore examined the gene expression patterns of β-karyopherins by statistically analyzing gene expression microarray data, classifying them as comparatively high expression (SPM > 4), above average (SPM > 2), and below average (SPM < 2).

β-Karyopherin genes are thought to be “housekeeping” genes, and Table II shows that they are ubiquitously expressed. Nevertheless they are not evenly expressed in all human tissues. 12 of 20 human β-karyopherin genes (CSE1L, IPO4, IPO7, IPO9, KPNB1, RANBP5, TNPO1, TNPO3,XPO1, XPO4, XPO5, and XPOT) show comparatively high expression in tissues that proliferate actively, e.g. lymphocytes, tumors, testis, and stem cells. The high expression of these genes may be due to their ability to carry cargoes needed for cell proliferation such as histones, glucocorticoid receptors, and RNAs (7). For example, CSE1L, which transports importin-α(44), is overexpressed in colon cancer cell lines, breast cancers, and liver neoplasms (45), and excess CSE1L may reduce importin-α/β-dependent import. Higher expression of CSE1L in tumors may reflect deregulation of nuclear transport. Among the overexpressed genes, KPNB1, RANBP5, CSE1L, XPO1, XPO5, and TNPO3 are more highly expressed in lymphoblasts than in other proliferating tissues. A second group of karyopherin genes (IPO11, IPO13, RANBP6, TNPO2, and XPO6) is expressed more strongly in brain and spinal cord than in other tissues. Different expression patterns of β-karyopherin genes among tissues were also documented in mouse microarray data; however, the patterns are different from those in man. Tissue expression screening of β-karyopherin genes in the mouse revealed the restricted expression of Ipo8 in both oocytes and fertilized eggs, whereas, curiously, such restriction was not observed in man. Mouse Ranbp17 mRNA is abundantly expressed in testis and pancreas (46). This finding is supported by our ongoing analysis of microarray data. We also observed that Ranbp17 has high expression in brown fat and skeletal muscle.

Table II.

Summmary of gene expression ofβ-karyopherins in adult human tissues and during mouse embryonic development (from fertilized egg to days 10.5)

The expression levels are symbolized by the number of “+” symbols according to the log₂ values of their average difference value. More + symbols indicate a higher expression level (27). — denotes that the expression data are not available. BM, bone marrow

Genes	Human data			Mouse data
	High expression	Above average	Below average	Fertilized egg	Blastocysts	6.5 dpc	7.5 dpc	8.5 dpc	9.5 dpc	10.5 dpc
KPNB1	721_B_lymphoblasts	Lymphocytes, BM-stem cells, cancerous cells, smooth muscle	Neurons, internal organs, ovary, etc.	++++	+++	+++++	+++++	+++++	+++++	+++++
IPO7		Cancer cells, BM-CD34+ cells, thyroid	Internal organs, testis, neurons, etc.	++	+++++	++++	++++	++++	++++	++++
IPO9		Lymphocytes, prefrontal cortex, BM-stem cells	Neurons, internal organs, testis, muscle, pancreas, etc.	+	+	+	+	+	+	+
RANBP5	721_B_lymphoblasts	Testis, BM-stem cells, cancer cells, bronchial epithelial cells, smooth muscle	Neuron, internal organ, uterus, etc.	+++	++++	++++	++++	++++	++++	++++
TNPO1		BM-34+, lymphocytes, smooth muscle, fetal thyroid, pituitary	Brain, neuron, internal organs, skin, colorectal adenocarcinoma, etc.	+	+	+	+	+	++	++
CSE1L	721_B_lymphoblasts	Testis, cancer cells, bronchial epithelial cells, BM-CD105+ endothelial	Brain, neurons, internal organs, glands, etc.	+++	++++	++++	++++	++++	++++	++++
XPO1	Lymphoblasts, BM-CD34+	Lymphocytes, stem cells, cancer cells, prefrontal cortex	Brain, neurons, internal organs, testis, ovary, pancreas	+++	++++	++++	++++	++++	++++	++++
XPO4		BM-CD71+ early erythroid, pancreas, testis, lymphocytes, lymphoma testis, cancer cells	The other tissues	+	+	++	++	+++	+++	++
XPO5	BM-CD34+, 721_B_lymphoblasts		Neuron, internal organs, muscles, glands, ovary	+	+	++	++	++	+++	++
XPO6	Blood	Lymphocytes, stem cells	The other tissues	—	—	—	—	—	—	—
XPOT	Cancer cells	BM-CD34+, lymphoblasts, bronchial epithelial cells	Brain, neuron, internal organs, testis, ovary	++	++++	++++	+++	+++	++++	+++
TNPO3	721_B_lymphoblasts and other lymphocytes	Leukemia, smooth muscle, BM-stem cells, testis	The other tissues	+++	+++++	+++++	++++	++++	++++	++++
IPO4		Testis, lymphoblasts, cancer	The other tissues	++	++	+++	+++	++	+++	++
IPO11		Trigeminal ganglion, testis, BM-stem cells	Lymphocytes and lymphoma	++	+++	+++	+++	+++	+++	+++
IPO13		Brain, spinal cord, testis, heart	Stem cells, lymphocytes, uterus	+	+++	+++	+++	+++	+++	+++
RANBP6	Prefrontal cortex	Brain, lymphocytes, BM-stem cells	Cancer, testis, tissues, etc.	+	++	++	++	++	++	++
TNPO2	Prefrontal cortex	Whole brain	Ganglia, lymphocytes, muscles	++	+++	+++	+++	+++	+++	+++
XPO7	BM-CD71+ early erythroid		No expression in the other tissues	+++	++	+++	+++	+++	+++	+++
IPO8			No obvious expression in all tissues	+++++	+++	+++	+++	+++	+++	+++
RANBP17			No obvious expression in all tissues	+	+	+	+	+	+	+

Developmental Regulation of KaryopherinβSuperfamily Gene Expression—During the cell cycle and during development, there is a continuous need for precise transport of large molecules in and out of the nucleus. Regulation of expression of β-karyopherins could be important for these processes (16, 19). We therefore examined the gene expression patterns of β-karyopherins during mouse development (27) by analyzing gene microarray data (Fig. 6). Most β-karyopherin genes increase their expression gradually during development, reach their peak expression by day 6.5–9.5, and then drop by day 10.5. Through the similarity measure of gene expression profiles in the early embryonic period (from fertilized egg to day 10.5), we grouped 19 mouse β-karyopherin genes (Xpo6 data are not available) into six expression patterns. The pattern that was found most frequently is characteristic of 10 genes (Cse1l, Ipo4, Ipo7, Ipo11, Ipo13, Ranbp5, Tnpo2, Tnpo3, Xpo1, and Xpot) (Fig. 6a) whose expression increases significantly from fertilized egg to blastocysts and then fluctuates slightly until day 10.5. Both Ranbp6-Xpo5 and Tnpo1-Xpo4 pairs show expression peaks; however, the former pair is at day 6.5 and day 9.5 (Fig. 6b), and the latter pair is primarily at day 9.5 (Fig. 6c). Ipo9 and Ranbp17 are nearly constant (Fig. 6d), whereas Xpo7, Kpnb1, and Ipo8 decrease in blastocysts, and all but Ipo8 then rebound (Fig. 6e). Expression of Xpo7 and Kpnb1 rebounds on day 6.5 and remains high until day 10.5 (Fig. 6f). In conclusion, β-karyopherin transcript levels do vary during development. Differences in their cargo specificity may explain these changes.

Fig. 6.

Gene expression patterns of β -karyopherins in mouse early development. The vertical axis is the log₂ ratio of gene expression levels; the horizontal axis is the developmental period abbreviated as follows (days postcoitus (dpc)): T1, fertilized egg; T2, blastocysts; T3, dpc 6.5; T4, dpc 7.5; T5, dpc 8.5; T6, dpc 9.5; and T7, dpc 10.5. The gene expression of mouse β-karyopherin genes was grouped into six patterns: a, Cse1l, Ipo4, Ipo7, Ipo11, Ipo13, Ranbp5, Tnpo2, Tnpo3, Xpo1, and Xpot; b, Ranbp6 and Xpo5; c, Tnpo1 and Xpo4; d, Ipo9 and Ranbp17; e,Kpnb1 and Xpo7; and f, Ipo8.

Cell Cycle-dependent Transcriptional Regulation of KaryopherinβSuperfamily Genes—As nuclear import of macromolecules can be regulated during the cell cycle due to alterations in the nuclear pore complex (47), it is of interest to ask whether expression of β-karyopherin genes is regulated in a cell cycle-dependent manner. Different approaches have been adopted to identify, correct, and complete microarray data sets of S. cerevisiae synchronized with mating factor or in cell cycle mutants. KAP95 shows the most obvious changes during cell cycle progression (29, 30). In addition, a periodic least square regression suggests cell cycle regulation of KAP123 and NMD5(48), and an approach based on cubic splines suggests modulation of MTR10(49). Bayesian modeling techniques also indicate that KAP114 and KAP95 are modulated (50). Thus, although it would be useful to have additional data sets, cell cycle-dependent regulation of β-karyopherin expression could impact nuclear import and export during the cell cycle.

Mechanism of Regulation of KaryopherinβSuperfamily Genes—DNA regulatory elements are crucial for understanding gene expression because the binding of corresponding factors determines the timing, location, and level of gene expression. Computational methods have been developed to identify and localize regulatory elements in a high throughput manner (51, 52). Moreover similar clusters of TFBSs can be found in the promoter regions of orthologous genes (53). A computational tool, MAPPER (31), was therefore adopted to identify putative TFBSs of human β-karyopherin genes (Fig. 7) and their orthologs in eight model species (supplemental data and supplemental Table 2).

Fig. 7.

Transcription factor binding sites of human β -karyopherin genes predicted by the MAPPER search engine. The regulatory analyses of the 20 β-karyopherin genes are illustrated line by line. The numbers under the lines are the positions (in bp) from the start of the transcript (position +1). The TFBSs are indicated above or beneath the lines. Five mostly potential TFBSs (SP1, NRF-2, Hen-1, RREB-1, and CAAT box) are highlighted by symbols. USF, upstream stimulatory factor; PPAR, peroxisome proliferator-activated receptor; COUP-TF, chicken ovalbumin upstream promoter-transcription factor; CREB, cAMP-response element-binding protein; RXR, retinoid X receptor; VDR, vitamin D receptor; ER, estrogen receptor; NF, nuclear factor; SRY, sex-determining region Y-chromosome protein; FTF, α-1-fetoprotein transcription factor; HLF, hepatic leukemia factor; ATF, activating transcription factor; CDP, CCAAT displacement protein; NF-GMa, nuclear factor for granulocyte/macrophage colony-stimulating factor; HLTF, helicase-like transcription factor.

We observed that the predicted promoter binding sites of β-karyopherin genes lack TATA boxes but are rich in GC content (Fig. 7 and supplemental materials). Although the TATA box is often required for transcription by nuclear RNA polymerases, several in vivo studies show that the TATA box is more specifically important for cellular proliferation, transformation, and control of the cell cycle (54, 55) than for regulation of housekeeping genes (56, 57). Mammalian promoters lacking a TATA box often contain SP1 (GGGCGG)/NF-1 (TGGNNNNNNGCCAA) binding sites and rely on them to initiate gene expression (58 – 60). SP1 is ubiquitously expressed and plays a key role in maintaining basal transcription of housekeeping genes (61). SP1 is also implicated in mediating development-specific gene expression (62). It is predicted that the SP1 binding site is one of the most common TFBSs for β-karyopherin genes: 13 of 20 human β-karyopherin genes (CSE1L, IPO8, IPO9, IPO13, KPNB1, RANBP5, RANBP6, RANBP17, TNPO1, TNPO2, TNPO3, XPO1, and XPOT) are predicted to have one or multiple SP1 binding sites in their proximal regulatory regions (Table III and Fig. 7). The CAAT box is another frequent candidate binding site for β-karyopherin genes: seven of 20 human β-karyopherin genes (CSE1L, IPO9, KPNB1, RANBP5, RANBP6, TNPO2, and XPO1) and 10 of 20 mouse β-karyopherin genes (Cse1l, Ipo9, Ipo11, Kpnb1, Ranbp5, Tnpo1, Tnpo2, Xpo1, Xpo4, and Xpo6) were predicted to posses this CAAT box in their upstream regulatory regions. The CAAT box is also ubiquitously distributed, being present in about 30% of eukaryotic promoters (63, 64). In higher eukaryotes, it is involved in many types of promoters: developmentally controlled (65), cell-cycle regulated (66), and housekeeping (67). It is interesting that sometimes SP1 can interact with nuclear factor Y (NF-Y), also known as CAAT box DNA-binding protein, to regulate downstream genes cooperatively (68, 69). Nuclear respiratory factor 2 (NRF-2), known as GA-binding protein, was also reported to cooperate with SP1 in the activation of several widely expressed housekeeping genes and genes that control cell cycle, differentiation, development, and other key cellular functions (70). The TFBS for NRF-2 is also found frequently (12 of 20 genes) in promoters of human β-karyopherin genes (IPO4, IPO7, IPO8, IPO9, IPO11, IPO13, RANBP17, TNPO1, TNPO3, XPO5, XPO6, and XPOT). There are some TFBSs, e.g. the binding sites for Ras-responsive element-binding protein 1 (RREB-1), helix-loop-helix protein 1 (HEN-1), aryl hydrocarbon receptor nuclear transporter, and B-cell-specific activator protein, that are also predicted to be potential TFBSs of human or mouse β-karyopherin genes. These transcription factors are known to participate in the regulation of development, differentiation, cell cycle, and other karyopherin-related cellular processes (38, 71, 72). Moreover several further TFBSs are predicted to be frequently involved in the regulation of β-karyopherin genes of different experimental models (Table III), e.g. five of 18 X. tropicalisβ-karyopherin genes have binding sites for Staf, 12 of 16 D. melanogasterβ-karyopherin genes have binding sites for Broad-complex isoform 4, three of 10 C. elegansβ-karyopherin genes have binding sites for UNC-86, and three of 14 S. cerevisiaeβ-karyopherin genes have binding sites for Ste12p.

Table III.

The top putative TFBSs of β-karyopherin genes in eight organisms: H. sapiens (Hs), M. musculus (Mm), G. gallus (Gg), X. tropicalis (Xt), D. rerio (Dr), D. melanogaster (Dm), C. elegans (Ce), and S. cerevisiae (Sc)

The number following the abbreviated taxa indicates the number of β-karyopherin genes in the designated organism, whereas the number after a TFBS indicates the number of β-karyopherin genes that were predicted by MAPPER search engine to be regulated by the TFBS. ARNT, aryl hydrocarbon receptor nuclear transporter; Bsap, B-cell-specific activator protein.

Species	TFBSs
Hs (20)	SP1 (13), NRF-2 (12), HEN-1 (10), RREB-1 (9), CAAT box (7)
Mm (20)	CAAT box (10), ARNT (9), Bsap (9)
Gg (17)	NF-κB (2)
Xt (18)	Staf (5)
Dr (23)	HNF-1 (11)
Dm (16)	Broad-complex_1 (12), Broad-complex_2 (7), Broad-complex_3 (7), Broad-complex_4 (13), Hunchback (6)
Ce (10)	UNC-86 (3)
Sc (14)	STE12 (3)

Despite our conclusion that selected TFBSs are shared among β-karyopherin genes within a species (Fig. 8 and supplemental materials), many of the putative TFBSs seem to be species-specific. For example, putative binding sites of SP1, NRF-2, HEN-1, RREB-1, and NF-Y emerge frequently in upstream regions of β-karyopherin genes in human and mouse, whereas Broad-complex serial binding sites are preferred in the fly (Table III and supplemental materials). These results suggest that functional regulatory factors change quickly during evolution and that the gain and loss of functional TFBSs are frequent during evolution (73). These and related concepts will be refined as the β-karyopherins and β-karyopherin genes of additional species are characterized.

Fig. 8.

Transcription factor binding sites of XPO1 orthologous genes predicted by the MAPPER search engine. The regulatory analyses of XPO1 over species are illustrated line by line: Hs, H. sapiens; Mm, M. musculus; Gg, G. gallus; Xt, X. tropicalis; Dr, D. rerio; Dm, D. melanogaster; Ce, C. elegans; and Sc, S. cerevisiae. The numbers under the lines are the positions (in bp) from the start of the transcript (position +1). The TFBSs are indicated above or beneath the lines. TF, transcription factor; PPAR, peroxisome proliferator-activated receptor; NF, nuclear factor; Bsap, B-cell-specific activator protein. NF-GMa, nuclear factor for granulocytes/macrophage colony-stimulating factor; DREF, DNA replication-related element binding factor; Su(H), suppressor of hairless protien; SU_h, suppressor of hairless protein.

Conclusions

We analyzed karyopherin β superfamily members with regard to their evolution and expression. Although bacteria do encode HEAT repeat proteins, such as CpcE and CpcF (36), the karyopherin β family itself appears to be absent from prokaryotes. Interestingly yeast Ran proteins are extremely similar to mammalian Ran or “Ran proteins of other species.” (74). Moreover most nucleoporins have homologs in all extant eukaryotic lineages, and the existence of distant prokaryotic homologs of several nucleoporins has been proposed (75). By contrast, there is more variability among the β-karyopherins.

Given the incomplete evolutionary record, we found the seemingly sudden appearance of 14 β-karyopherin members in S. cerevisiae. Yet these 14 yeast β-karyopherins are not uniformly represented in higher organisms. Moreover they are not all required for eukaryotic cells considering that C. elegans and some protozoa (e.g. G. lamblia) lack many of them. From yeast to man, eight β-karyopherin genes appear to have evolved directly without diversification, but KAP104, MTR10, NMD5, and PSE1 diverged by gene duplication or retroposition. Although the number of β-karyopherins increases from 14 in yeast to 20 in human, only 10 β-karyopherins have been identified so far in C. elegans although the size of the genome of C. elegans (97 Mbp) is much larger than that of S. cerevisiae (12 Mbp). It will be of interest to know how higher organisms that have lost β-karyopherins still maintain their normal physiological functions.

Expression analysis of β-karyopherins showed that their transcripts vary with cell differentiation and proliferation. Moreover in yeast, the titer of some β-karyopherin transcripts appears to be regulated during the cell cycle, and in mice, titers vary during development, suggesting changing roles of nucleocytoplasmic transport during cell differentiation. Virtual analysis of promoter binding elements showed that the combinations of transcription factors SP1, NRF-2, HEN-1, RREB-1, and NF-Y regulate expression of most β-karyopherin genes.

We thus used bioinformatics techniques to give a systematic overview of β-karyopherins: their sequence and structural features, their evolution, their transcriptional patterns, and their gene regulatory motifs. Future studies will need to address how a full set of β-karyopherins first appeared and diversified (from nine to 14 members) in yeast, whether there are additional functional β-karyopherins that have not been found, and how the β-karyopherins have co-evolved along with nucleoporins, components of the Ran GTPase cycle, and cargo diversity (70, 76–78).