The Structure of Importin-ß Bound to SREBP-2: Nuclear Import of a Transcription Factor

See allHide authors and affiliations

Science  28 Nov 2003:
Vol. 302, Issue 5650, pp. 1571-1575
DOI: 10.1126/science.1088372


The sterol regulatory element–binding protein 2 (SREBP-2), a nuclear transcription factor that is essential for cholesterol metabolism, enters the nucleus through a direct interaction of its helix-loop-helix leucine zipper domain with importin-β. We show the crystal structure of importin-β complexed with the active form of SREBP-2. Importin-β uses characteristic long helices like a pair of chopsticks to interact with an SREBP-2 dimer. Importin-β changes its conformation to reveal a pseudo-twofold symmetry on its surface structure so that it can accommodate a symmetric dimer molecule. Importin-β may use a similar strategy to recognize other dimeric cargoes.

One of the most important steps in the control of cellular gene expression in eukaryotic cells is the nucleocytoplasmic transport of functional molecules through the nuclear envelope. In response to extracellular signals or intracellular changes, certain transcription factors or their cofactors enter the nucleus to control gene expression. Expression is regulated by posttranslational modifications of the factors such as proteolysis, phosphorylation, and ligand binding. SREBP-2 is an example of a transcription factor for which the nuclear import and activity are regulated by proteolysis.

SREBP-2 is a member of the SREBP family of transcription factors, which are synthesized as precursor molecules and bound to the endoplasmic reticulum membrane and the outer nuclear envelope (1). The cytosolic N-terminal segment includes the basic helix-loop-helix leucine zipper (bHLHZ) domain, whereas the C-terminal regulatory segment interacts with the SREBP cleavage-activating protein (SCAP), a polytopic membrane protein (2, 3). When cells are deprived of cholesterol, SCAP escorts SREBPs to the Golgi apparatus, where they are proteolytically processed to liberate a transcriptionally active N-terminal fragment of 480 residues (3). This fragment enters the nucleus and activates the transcription of genes that control the synthesis and uptake of cholesterol and unsaturated fatty acids.

The transport of molecules into and out of the nucleus is mediated by the importin-β superfamily [for reviews, see (49)]. Importin-β plays a role in transporting both cargo that carries a classical nuclear localization signal (NLS) and cargo with no obvious signal sequence. It must also interact with molecules that regulate transport such as Ran guanosine 5′-triphosphate (RanGTP). Importin-β recognizes classical NLS-containing proteins by means of the adaptor molecule importin-α. The adaptor-independent cargoes for importin-β include ribosomal proteins (10), the human immunodeficiency virus (HIV) Rev and Tat (11), the Rex protein of human T cell leukemia virus type 1 (HTLV-1) (12), GAL4 (13), the parathyroid hormone–related protein (PTHrP) precursor protein (14), cyclin B1 (15), Smad3 (16), and SREBP-2 (17, 18).

Crystal structures have been demonstrated for importin-β complexed with cargo molecules (19, 20), RanGTP (21), an Phe-Gly (FG)–repeat nucleoporin (22), and in the free form (23). However, how importin-β directly recognizes the active form of transcription factors such as Smad3 and SREBP-2 remained unknown. We have previously shown that importin-β interacts directly with a dimeric SREBP-2 HLHZ region and mediates the import in a Ran-dependent manner (17, 18).

We determined a 3.0 Å resolution x-ray structure of the complex of full-length 876–amino acid importin-β and a dimer of the SREBP-2 HLHZ region comprising residues 343 to 403 (Fig. 1A). The x-ray structure is consistent with previous evidence showing that SREBP-2 is transported into the nucleus as a dimer (17, 18). The binding site for the SREBP-2 dimer is completely separated from those for importin-α, the Phe-x-Phe-Gly (FxFG) repeats of nucleoporin (where x is any amino acid), and RanGTP (Fig. 2). The SREBP-2 dimer binds between the long heat repeats 7 and 17 of importin-β, whereas importin-α binds in the concave face of the importin-β C-terminal half (19). The HLHZ dimer is inserted into the importin-β superhelix at an orientation perpendicular to the central axis of the superhelix, whereas the importin-β–binding (IBB) domain of importin-α is parallel to the central axis (Fig. 2). The PTHrP NLS peptide binds to the extended cargo-binding site on the concave surface of the N-terminal portion of importin-β (20).

Fig. 1.

(A) X-ray crystal structure of the complex of full-length 876–amino acid importin-β and a dimer of SREBP-2 residues 343 to 403 is shown. One of the two complexes in an asymmetric unit is depicted, composed of the importin-β A molecule and SREBP-2 HLHZ motif C and D chains. Columns and curved wires represent α helices and loops, respectively. The importin-β molecule is drawn in three colors, with heat repeats 1 to 4 in dark brown, heat repeats 5 to 12 half in light blue and half in blue, and C and D molecules of the SREBP-2 dimer colored in yellow and pink, respectively. Amino acids involved in major interactions for complex formation are emphasized using a ball-and-stick model. Major interaction sites are mainly present in the helix areas of heat repeats 7 and 17. The C and D chains of the HLHZ dimer have a pseudo-twofold symmetry axis. The helical rod locations of heat repeats of the 5 to 12 half and those of the 19 to 12 half in the reverse order are roughly related by twofold symmetry. However, each of the two similar helices has a reverse orientation from the other. Two complexes in an asymmetric unit have an almost identical structure except for heat repeats 1 to 4, in which the N-terminal (N-ter) segment exhibits structural variation between the two complexes. This is because the N-terminal area is not involved in complex formation, and thus, the importin-β molecule has relatively more freedom in these parts. The atomic coordinates have been deposited in the Protein Data Bank with the accession code 1UKL. C-ter, C terminus. (B) A schematic drawing of the 19 heat repeats of importin-β and the SREBP-2 HLHZ dimer. Importin-β and SREBP-2 dimer is depicted in the same colors as in (A). The importin-β molecule has a twofold symmetric arrangement of heat repeats 5 to 19, with heat repeat 12 in the center. In the importin-β molecule, the characteristic four long helices (heat repeats 7, 11, 13, and 17) are present. Major interactions occur between heat repeats 7 and 17 of importin-β and helix 2 of SREBP-2 (black). Importin-β uses the characteristic long helices 7 and 17 as a pair of chopsticks to pick up SREBP-2.

Fig. 2.

SREBP-2, importin-α, and RanGTP recognition schemes by the importin-β and nucleoporin FxFG binding sites are depicted. Importin-β is illustrated by a wire colored in the same manner as in Fig. 1A. RanGTP is represented by orange wires. The SREBP-2 dimer is colored as in Fig. 1A, and importin-α is colored in black. The ellipsoid ball indicates the predicted spatial arrangement of SREBP-2 and importin-α domains. Nucleoporin FxFG binding sites are colored in light brown. The cargo orientations of SREBP-2 in the importin-β:SREBP-2 complex are quite different from those of IBB in the importin-β:IBB complex structure.

On the basis of the structure of the SREBP-2 HLHZ dimer, we can predict the location of the large N-terminal domain (residues 1 to 342) and the small C-terminal domain (residues 404 to 480) (Fig. 2). The N-terminal domains bind concave surfaces near the N- or C-terminal domains of importin-β. This is well separated from the nucleoporin-binding sites. The C-terminal domains of SREBP-2 also do not interfere with nucleoporin binding; thus, the importin-β: SREBP-2 complex can bind to and move through the nuclear pore complex (21).

The importin-β:SREBP-2 complex crystal structure shows two importin-β:SREBP-2 dimer complexes in an asymmetric unit with the two importin-β molecules, referred to as A and B. The average root mean square deviation (RMSD) values of Cα atoms between importin-βs A and B were 2.01 Å when fitting the entire importin-β molecule and 1.16 Å when only residues 170 to 830 were fitted. This reflects higher conformational flexibility in the N-terminal region, which may be required to accomodate the various ligands that bind in this region (FxFG, RanGTP, and PTHrP NLSs).

Several x-ray crystal structures have been determined for bHLHZ family members bound to DNA (2426). Excluding the loop region, the RMSDs of Cα atoms of bHLHZ dimers in the DNA-complex structures are between 1.2 and 1.7 Å. The structure is maintained in the complex with importin-β, with a 1.1 Å RMSD between the HLHZ Cα atoms in importin-β: SREBP-2 and SREBP-1:DNA (fig. S1A) (26).

Previously, we showed that heat repeats 7 and 17 are required for importin-β binding to SREBP-2, but not to IBB. The structure is consistent with this result, showing that the HLHZ dimer binds mainly to two long helices in heat repeats 7 and 17. To accommodate the dimer molecules, importin-β moves heat repeats 7 and 17 to adopt a more twisted open conformation than importin-β:IBB. This confers a pseudo-twofold symmetry to importin-β that allows it to bind a dimeric substrate. The two long helices of heat repeats 7 and 17 act like chopsticks and grip the HLHZ dimer around its pseudo-twofold axis (Fig. 1B). When the importin-β molecules in the complexes with IBB and SREBP-2 were fitted by only Cα atoms of residues 449 to 876, the N-terminal area moved by almost 19 Å to accomodate the SREBP-2 dimer. To clarify the hinge site in the importin-β molecule, these two importin-β conformations were analyzed by the DynDom program (27). This did not reveal a mechanical hinge site, but there was a 22° rotation of the N-terminal area of importin-β (heat repeats 1 to 12). Thus, the rotation occurs close to the center of the pseudo-twofold symmetry, which is at heat repeat 12. This notable movement reveals the intrinsic flexibility of importin-β molecules for binding many types of cargo or RanGTP. Thus, it is of general interest to know how this group of receptors recognizes very different types of ligands.

The residue interactions between importin-β and the SREBP-2 dimer are represented schematically in Fig. 3A. Binding is mediated mainly by hydrophobic interactions. The conserved residue F752 (28) in importin-β heat repeat 17 packs against aromatic groups Y376 and Y379 of the SREBP-2 C chain (Fig. 3B). These residues, together with L380 and V383 of SREBP-2 and V755 of importin-β, create a hydrophobic core that stabilizes the complex. Other important interactions occur at the opposite site of the above interactions (Fig. 3A). Hydrophobic interactions between residues L380 of the SREBP-2 D chain and I295 of importin-β are important at the pseudosymmetric site. Electrostatic interactions also occur in the vicinity of the hydrophobic interactions (Fig. 3B). Although the loop regions of the SREBP-2 dimer are embedded in importin-β, only six loop residues—K363, H365, and R371 of the C chain and K363, H365, and K366 of the D chain—interact with importin-β. Heat repeats 7, 9, 11, 14, 15, 17, and 18 recognize the dimeric transcription factor (Fig. 3A), consistent with our previous study (17), indicating that importin-β residues 226 to 876 (heat repeats 6 to 19) are required for the nuclear import of SREBP-2.

Fig. 3.

(A) Overall interactions between importin-β and SREBP-2 dimer molecules are represented schematically. The ubiquitous hydrophobic residues that are involved in the hydrophobic core are represented in yellow. Hydrophobic close contacts between importin-β and the HLHZ dimer are shown as yellow lines, and electrostatic interactions are shown in black. SREBP-2 binding to importin-β is mainly dependent on hydrophobic interactions and supported by many other hydrophilic interactions. Residue F752 of importin-β heat repeat 17 interacts with residues Y376 and Y379 of the SREBP-2 C chain by close contact of aromatic rings in a rectangular orientation. Other important interactions occur at the opposite sites of the interaromatic ring interactions. This interaction site is also involved in hydrophobic interactions between residues I295 of importin-β and L380 of the SREBP-2 D chain. Electrostatic interactions are abundant in the vicinity of these hydrophobic residues. (B) Hydrophobic cores are depicted with a σA weighted 2Fo-Fc (32) electron density map contoured at 1.0σ. Residue names from importin-β are shown in light blue single-letter notation with the residue number, and those from SREBP-2 are in red. L380 and V383 of SREBP-2 and V755 of importin-β are crowded around F752 of importin-β heat repeat 17 and residues Y376 or Y379 of the SREBP-2 C chain. These six residues create a hydrophobic core and stabilize the binding of SREBP-2 to importin-β.

Although several types of proteins are direct-binding cargoes for importin-β, a consensus sequence required for direct recognition has not been determined. Polypeptide segments required for the nuclear import of cargo molecules range widely from 9 for HIV-1 Tat to the 120-residue dimer for SREBP-2 (11, 17). Several cargo molecules are rich in basically charged residues but do not have a conserved alignment (fig. S3). Binding between importin-β and the IBB-domain peptide occurs through an extended network of electrostatic interactions between the acidic cavity of importin-β and the basic side chains of the IBB domain. Thus, short sequences rich in basic amino acids such as HIV-1 Tat NLS may bind to importin-β in a similar fashion as the IBB domain. Although it also uses electrostatic interactions, the nonclassical NLS of PTHrP binds to a second extended cargo-binding site on importin-β that is distinct from the IBB-domain binding site. Unlike these direct-binding cargoes, the SREBP-2 NLS sequence has more hydrophobic residues and no consecutive basic charged residues (fig. S3), and hydrophobic residues play the major role in the interaction (17).

SREBP homodimers recognize not only the E box but also the sterol regulatory element (SRE), whereas cMyc, Max, and upstream stimulatory factor (USF), which belong to the same bHLHZ superfamily as SREBPs, form homo- or heterodimers that only recognize the E box. In addition, the critical residues of the SREBP-2 HLHZ dimer that interact closely with importin-β are markedly conserved in SREBP-1 but are not highly conserved in the HLHZ motifs of cMyc, Max, and USF (fig. S1B). Thus, it is likely that importin-β also directly recognizes SREBP-1 but not cMyc, Max, or USF. cMyc and USF have two classical NLSs, one within the basic region of the bHLHZ and the other upstream of the basic region (2931). Thus, in addition to their function in DNA binding, the basic regions of cMyc and USF bHLHZs probably act as classical basic NLSs, implying that their transport is importin-α/β–dependent. Max contains only one classical NLS in its C-terminal domain.

We propose that the bHLHZ superfamily transcription factors can be divided into at least two subgroups, the SREBP group and the cMyc group, as judged from the nuclear import mechanism. The nuclear transport of SREBPs is regulated by proteolysis. SREBPs are embedded in membranes of the endoplasmic reticulum. When the cell senses a low sterol level, a two-step proteolytic process releases the transcriptionally active 480-residue N-terminal domain from the membranes. Because the translational freedom of SREBP molecules on the membrane is restricted to lateral motion, it is likely that SREBPs are embedded into the membranes as dimers. Processed SREBPs preserve their dimeric active structure and are imported into the nucleus by importin-β. The SREBPs can bind to SRE immediately after the SREBP dimers are released from importin-β through the binding of nuclear RanGTP to the N-terminal domain of importin-β. The RanGTP-bound form of importin-β would open the chopsticks, releasing SREBPs in the nucleus. Thus, we suggest that transcriptional regulation by SREBPs—including proteolysis, nuclear import, and SRE recognition—is efficiently and accurately performed without the dissociation and reassociation reactions of SREBP dimers. In contrast, cMyc, Max, and USF recognize the E box as homo- or heterodimers. These E box–binding bHLHZ proteins have classical NLSs and may be constitutively imported into the nucleus as monomers by complexes immediately after they are synthesized on free ribosomes. Instead of transcriptional regulation through release of the cargo, regulation may be at the level of homo- or heterodimerization in the nucleus. Thus, despite its structural similarity to other bHLHZ motifs, SREBP appears to have a different nuclear transport mechanism and possibly a different mechanism for regulating transcription.

Supporting Online Material

Materials and Methods

Figs. S1 to S4


References and Notes

Table 1.

Crystallographic data and refinement statistics. Values in parentheses are for the highest resolution shell (3.0 to 3.1 Å for data 1 and 3.2 to 3.1 for data 2). I/σ, signal-to-noise ratio of observed intensities.

Crystal data Se (SREBP-2) (data 1) Se (importin-β) (data 2)
Space group P212121P212121
Unit cell parameters (Å)
    a 101.1 101.3
    b 113.3 113.4
    c 240.0 240.1
Wavelength (Å)View inline 0.97939 0.97958
Resolution (Å) 80-3.0 (3.1-3.0) 40-3.1 (3.2-3.1)
Unique reflections 55607 (5720) 49373 (4693)
Redundancy 5.9 (2.2) 2.7 (2.0)
Completeness (%) 99.7 (97.5) 97.9 (94.2)
R merge View inline 0.064 (0.300) 0.077 (0.408)
I/σ 27.7 (3.0) 18.2 (1.7)
Phasing statistics View inline
Phasing powerView inline
Anomalous (acentric) 0.96 1.06
    (acentric) 0.49
    (centric) 0.53
Figure of merit [after SHARP (View inline)]View inline
    (centric) 0.11
    (acentric) 0.29
Figure of merit [after DM (View inline)] 0.82
Refinement statistics
Number of protein atoms 15612
R factor (%)View inline 23.7 (35.2)
Rfree (%)View inline 29.6 (38.9)
RMSDs from ideal
Bond lengths (Å) 0.008
Bond angles (°) 1.427
  • View inline* The wavelength is calibrated by fluorescence data of a crystal for Se-Met data.

  • View inline Rmerge = ΣhΣi||F(h,i)|2 - 〈|F(h,i)|2〉|/ΣhΣi|F(h,i)|2.

  • View inline Phases were calculated using the data 1 and the data 2 with SHARP. The data 1 was treated as the reference.

  • View inline§ Phasing power = Σ|FH|/|E|, where FH and E are the calculated structure factor of heavy atoms and the lack of closure error, respectively.

  • View inline Figure of merit is the weighted mean of the cosine of the deviation of the phase angles from the best phases (αbest).

  • View inline R factor = Σh|F(h) - Fc(h)|/Σh|F(h)|, where Fc(h) is a calculated structure factor.

  • View inline Rfree, R factor evaluated for the 5% of reflections that were excluded from the refinement.

  • View Abstract

    Stay Connected to Science

    Navigate This Article