A major effort in the nucleocytoplasmic transport field has been directed towards the analysis of all members of the importin β family, in particular with the aim of identifying specific transport cargoes. This has led to the characterization of a large number of related proteins (both importins and exportins) in all eukaryotic species analyzed. New members have primarily been identified by sequence homology or biochemically via their interaction with the small GTPase Ran. Strikingly, the roles of many of these receptors is conserved from yeast to humans. There are 14 putative members of the importin β family in the completed Saccharomyces cerevisiae genome, nine of which have been shown to function as importins and four as exportins [1]. The genes encoding yeast transport receptors are dispersed throughout the genome and none of them contains introns. Higher eukaryotes contain an even larger number of importin-β-like proteins. It has been proposed that there are more than 22 putative members in mammals [1,2,3]. Little is known about the gross structure of the genes encoding these nuclear transport receptors, and our knowledge of the chromosomal localization or the organization of the individual genes encoding members of this family is very poor.

The relative molecular masses of members of importin-β-like proteins vary between 90 kDa and 130 kDa, but all are characterized by an acidic isoelectric point. The overall sequence similarity between various transport receptors is low (less than 20% amino acid identity) and, in many cases, is restricted to the amino-terminal domain. Work mainly on importin β has demonstrated that these receptors bind RanGTP via the amino-terminal domain and cargo via the carboxy-terminal domain [4,5,6,7,8]. To permit shuttling through the nuclear pore complex (NPC), transport receptors also contain one or multiple binding domains for components of the NPCs, called nucleoporins. Truncation studies using importin β indicate that the binding site for nucleoporins containing FxFG repeats (in the single-letter amino acid code, where x is, in many cases, a small polar residue or glycine) is located in an amino-terminal/central region of importin β (residues 152-352) [4,5]. This was recently confirmed in the crystal structure of an amino-terminal fragment of importin β in a complex with five FxFG nucleoporin repeats [9]. Additional information comes from the crystal structures of importin β and of transportin 1 (also known as karyopherin β2), which were solved either in a complex with RanGTP [7,8], with a cargo [6], or in the free form [10].

Overall, the structures of importin-β-like receptors are characterized by a very similar series of helical HEAT repeats (19 in importin β and 18 in transportin 1; Figure 1). HEAT repeats are approximately 40 residues in length and are found in many eukaryotic proteins such as the PR65/A subunit of protein phosphatase 2A [11]. The fundamental repeat unit is a right-handed superhelical structure consisting of a hairpin made up of two β helices, named A and B, separated by a sharp turn. Each hairpin is connected to the next by a linker region. In transportin 1, almost all linkers contain a third helix, but there are very few linker helices in importin β. In both receptors, one turn is extended into a long acidic loop, which has been suggested to be important for RanGTP-mediated cargo release. Full-length importin β complexed with the importin-β-binding domain of importin α (IBB) forms a snail-like superhelical structure wrapping tightly around the IBB domain. The structure of the uncomplexed amino terminus of importin β reveals a different superhelical architecture with a much steeper helical pitch than the cargo-bound or RanGTP-bound forms [10]. This suggests that importin β undergoes twisted conformational changes in its HEAT-repeat helix stacking, which could be essential for the regulation of cargo binding and release and/or for protein interactions during the translocation through the NPC. No structure of an exportin has yet been reported. Although the sequence homology is limited, it is expected that exportins will fold in a similar way to the reported importin structures. It still remains unclear, however, why RanGTP is required for binding of cargo to exportins but causes cargo dissociation from importins.

A major function of transport factors of the importin β family is to mediate the transport between the nucleus and cytoplasm of macromolecules that contain nuclear import or export signals [1,12]. To this end, all transport factors constantly shuttle between the nucleus and the cytoplasm. At steady state, they can be found in the nucleus, at the NPC or in the cytoplasm. At present, very little is known about the tissue distribution of this family of proteins. All members have the ability to recognize and bind specific cargoes, either directly or via adaptor molecules, to bind RanGTP and to interact with nucleoporins at the NPC. Interactions between the proteins of the importin β family and nucleoporin repeats have been shown both in vitro [4,5,13,14,15,16,17] and in vivo [18]. These interactions contribute to the import or export of importin β family members and their cargoes through the central transporter of the NPC.

Import and export are multistep processes that are initiated by the recognition of nuclear localization signals (NLSs) and nuclear export signals (NESs). The most thoroughly studied import signals are the 'classical' and the bipartite NLSs, first identified in SV40 large T antigen and nucleoplasmin, respectively [19]. Their transport is mediated by importin β, the first-characterized member of this protein family. Importin β indirectly associates with these NLS motifs via the adaptor molecule importin α [20,21,22,23,24]. Additional importin-β-dependent adaptors in vertebrates include snurportin 1 (involved in import of m3G-capped small nuclear ribonucleoproteins, snRNPs [25]) and XRIPα, (involved in the import of replication protein A, RPA [26]). Importin β can also form a complex with another importin-β-like factor, importin 7, in order to transport the linker histone H1 into the nucleus [27]. In addition, importin β is able to interact directly with a large variety of different cargoes, including the T-cell protein tyrosine phosphatase [28], the human immunodeficency virus (HIV) TAT and Rev proteins [29], human T-cell leukemia virus Rex protein [30], ribosomal proteins L23a, S7, and L5 [31], cyclin B1 [32,33], Smad [34] and the parathyroid-hormone-related protein [35]. Table 1 details known importins, exportins and their cargoes and adaptors.

The import of ribosomal proteins seems to rely on at least partially redundant mechanisms. In yeast, the importin Kap123p/Yrb4p has been shown to be an important mediator of ribosomal-protein import [36,37], but the related protein Kap121p/Pse1p can functionally substitute for Kap123p in vivo [36]. In mammalian cells, at least four importin-β-like transport factors are able to mediate import of ribosomal proteins [31]. Interestingly, both the β-like importin receptor binding (BIB) domain and some ribosomal proteins can be imported by any of the four receptors importin β, transportin 1, importin 5 or importin 7 [31].

The yeast importin Kap104p mediates nuclear import of the mRNA-binding proteins Nab2p and Nab4p [38]. Its vertebrate homolog, transportin 1 (Kapβ2), also mediates import of the RNA-binding proteins hnRNP A1 and hnRNP F, but also of ribosomal proteins [31,39,40,41,42]. Mammalian transportin-SR mediates nuclear import of a group of abundant arginine/serine-rich proteins, which are essential pre-mRNA splicing factors [43,44]. In yeast, the TATA-binding protein (TBP) is imported into the nucleus by Kap114p [45].

Other members of the importin-β family have been found to mediate nuclear export events. Exportin 1 (Crm1p, Xpo1p) has been identified in both yeast and higher eukaryotes as an export receptor for leucine-rich NES-containing proteins [46,47,48,49]. Human exportin 1 has been found to export protein kinase inhibitor α (PKIα) and HIV Rev [46], IκBα [50], snurportin 1 [51], HTLV Rex [52], the small nuclear RNA [46,53], cyclin B1 [54] and the transcription factor NF-AT4 [55]. Targets for exportin in S. cerevisiae include the transcription factors Yap1p [56], and Ace2p [57], the mitogen-activated protein kinase Hog1p [58], and the heat-shock protein Ssb1p [59]. CAS (yeast Cse1p) functions in the export and recycling of importin α [60,61,62,63]. Msn5p was first identified as a yeast exportin that exports the phosphorylated form of the transcription factor Pho4p [64]. Interestingly, Msn5p was recently shown to function as an import receptor for the trimeric RPA [65], suggesting that individual members of the importin-β-like protein family can function both as import and export receptors. The first cellular RNA export receptor to be discovered, named exportin-t in higher eukaryotes or Los1p in yeast, mediates nuclear export of tRNAs [66,67,68].

Many of the nuclear transport factors identified in S. cerevisiae are not essential for viability, even though they transport essential cargoes. This phenomenon can be explained by the fact that cargoes can use alternative transport pathways. This is probably best exemplified in the import pathway of ribosomal proteins, which is mediated by at least four different import receptors (see above).

Despite the large amount of progress that has been made in the nucleocytoplasmic transport field in recent years, many important questions remain unsolved. Many import and export receptors have now been characterized and their first cargoes have been identified. The further characterization of new transport receptors and adaptors, and the identification of new import and export substrates, will lead to a more complete picture of nucleocytoplasmic transport. A big challenge for the future will be to understand how translocation through the NPC occurs, and how nucleocytoplasmic transport is regulated. How does the NPC achieve its tremendous amount of selectivity? What type of changes in NPC conformation are required for passage through the NPC? To gain insight into these questions, a quantitative analysis of interactions between transport receptors and nucleoporins will be required. It will be also interesting to see whether an increasing affinity gradient of receptors for nucleoporins along the NPC exists and, if so, whether it makes an important contribution towards the direction of transport.

Other issues that remain to be solved include the structural differences between members of the importin and exportin family. How does RanGTP dissociate import complexes in the nucleus but promote binding of export cargoes to exportins? A key mechanistic topic that is poorly understood is the export of RNAs and RNPs. Although a tRNA-export factor has been identified, the mechanism of rRNA or mRNA export is still poorly understood. Several proteins in S. cerevisiae, such as Mex67p and Yra1p, and their metazoan counterparts TAP and Aly, have been indicated to play an important role in mRNA export. Mex67p/TAP does not belong to the family of importin-β-like proteins, suggesting that there are alternative translocation pathways through the NPC. Future studies will provide insights into these important questions.