Drosophila TIM Binds Importin α1, and Acts as an Adapter to Transport PER to the Nucleus

To determine if any member of the importin α family is required for the nuclear entry of PER/TIM, we used the UAS/GAL4 system to knock down the expression of each importin α via RNAi in clock cells. We obtained UAS-importin α RNAi lines from the Vienna-based Drosophila stock center (VDRC) or the Bloomington stock center, and crossed each one to Pdf-GAL4 and tim-UAS-GAL4 (TUG) to achieve knockdown in central clock neurons, the ventral lateral neurons (LNvs), and all clock cells, respectively. Dicer was also coexpressed to increase the efficacy of RNAi. We assayed the locomotor rhythms of these flies under continuous dark conditions (DD) and found that knockdown of importin α1 in PDF-positive LNvs resulted in weak rhythms, as determined through fast fourier transform (FFT) analysis (Table 1). Use of stronger drivers (TUG and tim-GAL4) to express importin α1 RNAi in all clock cells rendered all flies arrhythmic in DD (Table 1 and Fig. 1A). We used two independent RNAi lines for importin α1, one from VDRC (α1 RNAi-1) and the other from the Bloomington stock center (α1 RNAi-2). α1 RNAi-1 expressed by TUG produced arrhythmic behavior, which was accompanied by lower mRNA levels of importin α1 (~ 64%). TUG driven α1 RNAi-2 lines did not show a behavioral phenotype, which was consistent with lack of an effect on importin α1 mRNA (S1 Fig.).

We found that Pdf-GAL4- and TUG-driven importin α2 RNAi also produced arrhythmia or weak rhythmicity, although only after 4–5 days in DD (Fig. 1A). In addition, knockdown of importin α3 by TUG in clock cells resulted in arrhythmia in 50% of the flies after 5 days in DD, and knockdown of importin α3 with a stronger tim-GAL4 driver induced lethality (Table 1). Overall, the behavioral phenotype produced by knocking down these other two importins was weaker than that seen for importin α1, even though mRNA levels of the relevant importin were reduced in the flies tested (S1 Fig.). Knockdown of importin α4 knockdown did not produce any behavioral phenotype.

To determine whether the arrhythmic behavioral phenotype of importin α1 knockdown flies was caused by blocked nuclear entry of PER and TIM, we examined PER/TIM protein localization in adult LNvs. In control flies, PER and TIM were nuclear at zeitgeber time 0 (ZT0) (ZT0 = lights on and ZT12 = lights off), as has been reported in several previous studies [24–26]. However, both proteins were cytoplasmic at ZT0 in TUG driven α1 RNAi-1 flies (Fig. 1B). We did not observe any anatomical defects in the PDF+ cell bodies or in the PDF projections of LNvs in these flies, suggesting that the observed behavioral and molecular phenotype is not due to gross developmental defects (S2 Fig.).

As mentioned above, flies in which expression of importin α2 or α3 was reduced in adult LNvs or in all clock neurons were also arrhythmic 4–5 days after transferring to DD. To investigate whether importins α2 or α3 also play a role in the nuclear entry of clock proteins, we immunostained whole-mount brains of α2 and α3 knockdown flies using a PER antibody at ZT1. In both fly lines, PER was expressed in the nucleus at ZT1 (), indicating that neither importin α2 nor α3 is required for the nuclear translocation of PER in LNvs. However, the distal layer of the optic medulla, including the PDF projections of large LNvs into the medulla, was impaired in these flies, suggesting that importin α2 and α3 contribute to the development of the LNvs. Thus, the arrhythmic phenotype observed after a few days in DD in flies with reduced importin α2 and α3 may derive from developmental problems. We infer that only importin α1 affects the nuclear entry of PER and TIM in central clock cells. S2 Fig.

Given the limitations with RNAi technology, such as nonspecific and off-target effects [27], we sought to verify the results of RNAi knockdown using a complete loss-of-function allele of importin α1- Df(3L)α1S1. This allele consists of a deletion, which removes the entire importin α1 gene and several other genes, and is homozygous viable [28]. Consistent with the phenotype obtained by knocking down importin α1, Df(3L)α1S1 flies were arrhythmic in DD. These flies also showed reduced rhythmicity in the presence of light:dark cycles, with ~60% of flies scored as arrhythmic (Table 2 and Fig. 2A). As the deletion in Df(3L)α1S1 flies includes multiple genes besides importin α1, we sought to determine whether the mutant phenotype derived exclusively from the loss of importin α1. We tagged a cDNA encoding wild type importin α1 with an HA epitope and cloned it under control of a UAS sequence and then expressed in the Df(3L)α1S1 background using the GAL4 system. In DD, expression of importin α1 in clock neurons using TUG restored rhythmicity in 84% of Df(3L)α1S1 flies (Table 2 and Fig. 2B). We were also able to rescue rhythmicity in 70% of mutant flies using the Pdf-GAL4 driver, suggesting that expression of importin α1 is predominantly required in LNvs for free running locomotor rhythms (Table 2). These results demonstrate that importin α1 is essential for maintaining rhythms in DD and LD.

To investigate whether nuclear translocation of PER/TIM was affected in Df(3L)α1S1 mutant flies, we performed immunostaining. At ZT0, PER and TIM were completely cytoplasmic in mutant flies while they were nuclear in wild-type flies, recapitulating the result obtained with importin α1 RNAi flies (TUG > importin α1 RNAi-1) (Fig. 2C). Over a 24 hour cycle, as expected, the localization and expression of PER was rhythmic in wild type flies, such that it was nuclear at ZT0 and ZT6 (with lower levels at ZT6 than ZT0), cytoplasmic at ZT18 and practically undetectable at ZT12 (S3A Fig.). In mutant flies, PER was cytoplasmic whenever detected, but expression was undetectable at ZT12, indicating that its levels are regulated in the cytoplasm. Under DD conditions, in wild type flies, the pattern of localization and expression of PER were similar to those in LD. However, in mutants PER was cytoplasmic at all times and levels did not cycle (S3B Fig.). Likewise, the TIM expression pattern in wild-type flies (nuclear at ZT0, expressed at very low levels at ZT6 and 12 and cytoplasmic at ZT18) was similar in LD and DD except that expression was slightly higher at CT12 than ZT12 (S3C/D Fig.). In mutant flies, TIM was cytoplasmic at ZT0 and ZT18 and undetectable at ZT6 and ZT12, presumably due to its degradation by light. However, in DD TIM was detected in the cytoplasm at all times (S3D Fig.). These data are also supported by studies of temporal patterns of PER and TIM protein accumulation in fly heads as discussed below.

The deletion mutant of importin α1, Df(3L)α1S1, confirmed that importin α1 is important for mediating the nuclear entry of clock proteins, PER and TIM, and thus maintaining behavioral rhythms. We also determined if IMPα1 produces an over-expression phenotype in impα1 heterozygotes (Df(3L)α1S1/+), which show wild type behavior, or in period-altering per mutants, per^L and per^S. The per^L mutation, which increases circadian period, was of particular interest as it is associated with delayed nuclear entry of PER and TIM. However, IMPα1 expression did not affect rhythms of “wild type” flies (S4A Fig.) or the periodicity of per mutants (S4B/C Fig.). These data suggest that while IMPα1 is required for nuclear expression of PER and TIM, it is typically not a limiting factor in flies.

From the late night to early morning, nuclear PER and TIM inhibit their own transcription, resulting in low levels of both mRNAs. As nuclear TIM is degraded in response to light at daybreak and then PER is hyperphosphorylated and also degraded, repression activity is released and levels of per and tim mRNA rise [1,2]. We examined whether blocked nuclear entry of PER and TIM in Df(3L)α1S1 mutant flies affects cycling of their mRNAs. We expected that RNA levels of per and tim in mutant flies would not cycle and levels would be higher than those of wild-type flies at all times over the course of a day due to reduced repression. Consistent with our expectation, mRNA levels did not cycle. However, we noticed higher levels of per and tim mRNA, relative to wild type, only during the normal trough of mRNA expression, which corresponds to nuclear expression of TIM and PER (ZT21-ZT1) (Fig. 3A). At other times, RNA levels were lower in the mutant. It is possible that release of transcriptional repression in wild type flies leads to more efficient gene expression than constitutive transcription, which could account for this effect. We also tested whether mRNA levels of importin α1 are expressed cyclically. Quantitative PCR (qPCR) revealed that importin α1 transcript levels do not cycle.

We then assayed the expression of PER and TIM proteins in whole heads of Df(3L)α1S1 mutant flies. Western blots of fly head extracts from different time points indicated 24 hour oscillations of PER and TIM protein levels in LD. However, the amplitude of the oscillation was blunted relative to wild type (Fig. 3B). As reported previously [25], hyperphosphorylated forms of PER, detected as low mobility forms on the gel, are observed predominantly at times of nuclear expression e.g. at ZT1. These are the forms that are subsequently targeted for degradation. PER in mutant flies was less phosphorylated at ZT1 and more stable at ZT7 than in wild-type (Fig. 3B). In DD, it appeared that PER stability, but not its phosphorylation, still cycled in mutant flies, albeit with an aberrant and variable phase (Fig. 3C). Thus, the stability of PER may fluctuate in the cytoplasm. It is also possible that the translocation of PER is regulated in an importin α1-independent manner in some cells that contribute to the signal on whole head western blots. Based upon the lack of cyclic phosphorylation, however, we favor the explanation that PER remains in the cytoplasm.

As for PER, TIM cycling was dampened in importin α1 mutants. TIM levels were reduced at daybreak in mutant flies, suggesting that light can degrade cytoplasmic TIM; indeed, the weak cycling seen for PER could be driven by light-induced cycling of TIM, as TIM stabilizes PER [10,11,29]. However, some TIM remained in mutant flies at ZT1 while TIM in wild type flies was completely gone (Fig. 3B and S5A Fig.). It is possible that cytoplasmic TIM is less sensitive to light-driven degradation. On the other hand, as TIM degradation around dawn is also regulated by circadian clock mechanisms, which persist in constant darkness [30], we speculate that clock-dependent degradation of TIM requires nuclear expression. Consistent with this idea, TIM cycling was completely abolished in DD (Fig. 3C and S5B Fig.).

Given the importance of IMPα1 in the nuclear translocation of PER/TIM in clock cells, we sought to determine if its effect on the subcellular localization of clock proteins could be recapitulated in S2 cells. Both per and tim were tagged with CFP and YFP respectively, and their expression was induced by heat-shock as previously described [9,25]. Previous studies have shown that either PER or TIM alone is cytoplasmic when transfected into S2 cells, but each shows more nuclear expression when the other protein is co-expressed [12,22,25]. However, even under these conditions, neither protein becomes completely nuclear in the S2 cells we use [25]. As these cells likely express very low levels of IMPα1, we wondered if it was the missing component, and so added IMPα1 in these experiments. The addition of IMPα1 had no significant effect on PER localization by itself, but it increased uniform (nuclear and cytoplasmic) expression of TIM up to almost 50% (Fig. 4A/B). As previously reported [22,25], co-expression of PER and TIM also increased nuclear and uniform expression of both proteins (40–50%). Interestingly, co-transfection of all three proteins (PER, TIM and IMPα1) resulted in strong and specific nuclear localization of PER/TIM in ~70% of cells, with most other cells showing uniform expression (Fig. 4A/B). These results suggested that IMPα1 primarily facilitates nuclear entry of TIM in S2 cells.

Previous studies have found a delay in the nuclear translocation of PER/TIM following induction of their expression in S2 cells. Thus, they remain in the cytoplasm for 5–6 hours and then transfer into the nucleus [9,22]. This delay is reminiscent of the delay in nuclear entry observed in clock cells, a delay that is thought to be critical for clock function [24]. To investigate whether IMPα1 might be a rate-limiting factor that causes the delay in nuclear transport of PER/TIM in S2 cells, we followed subcellular localization every 2hr after heat-shock induction of both proteins. When PER and TIM were expressed together without IMPα1, they were mainly cytoplasmic 2hr after induction and showed increased uniform and nuclear expression 4hr after induction. However, increased nuclear and uniform expression of PER and TIM was detected even 2hr after induction when IMPα1 was co-expressed with PER/TIM (Fig. 4C). Thus, the delay in S2 cells may typically result from low expression levels of endogenous IMPα1.

As S2 cells vary from batch to batch, we also examined a different sub-culture that shows higher nuclear expression (~85%) of co-expressed PER and TIM [9]. Interestingly, these cells express detectable levels of IMPα1 (S6A Fig.), which may account for the increased nuclear expression. To determine if this was the case, we down-regulated IMPα1 via RNAi, and found that it confined the expression of PER and TIM to the cytoplasm; however, it did not affect nuclear expression of dCLK in S2 cells (S6A/B Fig.). Together these data support the idea that IMPα1 is specifically required for the nuclear translocation of PER/TIM in S2 cells.

Importins typically drive nuclear transport of cargoes in concert with a number of other proteins, including nuclear envelope proteins (nucleoporins). Interestingly, we have frequently noticed nuclear envelope association of TIM in LNvs of adult flies at ZT20 (S7A Fig.). This expression of TIM at the nuclear rim was also observed for TIM transfected in S2 cells (Fig. 4B and S7B Fig.). To investigate whether nucleoporins (NUPs) are part of the circadian mechanism, we knocked several NUPs (NUP214, 88, 153, 154 and Megator) down in all clock cells using TUG with dicer and assayed rest:activity behavior. None of these showed significant change in circadian behavior compared to control flies. However, knockdown of NUP153 resulted in lethality. To circumvent the lethality, we restricted the expression of nup153 RNAi to adult clock neurons by coupling the TUG driver with tublin-GAL80^ts. We found that these flies also died 3–4 days after being moved to 29°C. However, they were arrhythmic before they died (S7C Fig.). Use of the weaker Pdf-GAL4 driver (along with dicer) to downregulate NUP153 expression dampened rhythmicity of flies without causing any lethality (Fig. 5A and S1 Table). As with the importins, we assayed expression of PER in the PDF⁺ cells of flies that had NUP153 knocked down. Knockdown of NUP153 disrupted the s-LNvs, and so we assayed PER in l-LNvs and found that it was cytoplasmic at ZT1, when PER is nuclear in control flies (Fig. 5B). These data indicate that while important for development, NUP153 is also required for the nuclear entry of clock proteins.

Given the nuclear envelope localization of TIM, sought to determine if NUP153 binds TIM. Co-immunoprecipitation assays in S2 cells revealed a specific interaction between V5-tagged TIM and endogenous NUP153, which we confirmed through GST-pull down assays. A GST-NUP153FG fusion protein, in which phenylalanine glycine (FG) repeats of NUP153 were fused to GST and expressed in E. coli, interacted with TIM-V5 expressed in S2 cells (Fig. 5C).

Another key regulator of importin-mediated nuclear transport is the small GTPase Ran, which is predominately GDP-bound in the cytoplasm and GTP-bound in the nucleus. In the nucleus Ran-GTP dissociates the IMPα/IMPβ/cargo complex by binding to IMPβ [17,31]. To examine the role of Ran in flies, we expressed a dominant negative form of Ran (RanDN) in adult PDF⁺ neurons using the inducible PdfGeneSwitch (Pdf-GS) driver. RU486-induced expression of RanDN caused arrhythmia under DD conditions (Fig. 5D and S1 Table). To investigate whether this behavioral phenotype was related to altered nuclear translocation of clock proteins, we determined the localization of PER in Pdf-GS > UAS-RanDN flies at ZT1 on the 3^rd day after RU486 treatment. In these flies PDF expression and the dorsal projection from sLNvs were severely impaired. Also, PER was difficult to detect, presumably due to low expression levels. To minimize these effects of Ran, we assayed the localization of PER at ZT1 on the 1^st day after RU486 treatment, as morphological effects were not visible at this time. PER was cytoplasmic at ZT1 in RanDN flies whereas it was nuclear in control flies (Fig. 5F), suggesting that Ran is needed for nuclear translocation of clock proteins. Interestingly, we found that TIM expressed in S2 cells also interacts with GST-Ran and this interaction is enhanced when Ran is in the GTP-bound form (Fig. 5E).

Our cell culture data indicated that TIM was required for the nuclear localization of PER by IMPα. This could reflect co-transport of two cargoes, or perhaps even an IMPβ like function for TIM. To further test this idea, we sought to determine whether TIM interacts with IMPα1 in a manner that would be predicted for an IMPβ. IMPα proteins bind IMPβ through the Importin-β-binding (IBB) domain and this interaction modulates the binding of cargo to IMPα [32]. Specifically, the IBB domain folds over and interacts with the NLS-binding site on IMPα and so it competes with cargo for binding to this site [32]. Cargo binding is facilitated when the IBB domain binds IMPβ and releases the NLS-binding site [33,34]. We hypothesized that if TIM functioned as an IMPβ it would not bind to an IMPα1 that lacked the IBB domain. Contrary to our expectations, V5-tagged wild-type TIM showed strong binding to VSV-tagged IMPα1∆IBB whereas it did not bind to wild-type IMPα1 (Fig. 6A). Previous work has shown that the binding affinity of IMPα for cargo increases when the IBB domain is mutated [35]. This result suggests that TIM is cargo for IMPα1 rather than an IMPβ- like molecule.

The idea that TIM is classical cargo for importins is also supported by the finding that mutation of the TIM NLS lengthens the circadian period in flies (~30hr), significantly delays nuclear accumulation of both TIM and wild-type PER in S2 cells, and affects molecular oscillations of PER and TIM in the same manner as the importin α1 mutation [23]. Also, previous work indicated that the nuclear entry of mCRY2, a mammalian clock component that acts as a PER partner, is mediated by the importin α/β system through a bipartite NLS [36]. To determine if the interaction between TIM and IMPα1∆IBB was dependent on the TIM NLS, we mutagenized basic residues of the NLS to alanines. TIM carrying this mutated NLS did not bind IMPα1∆IBB (Fig. 6A), suggesting, as expected for a typical cargo protein, that the NLS mediates the interaction with IMPα1. In addition, we found that mutation of the TIM NLS prevented nuclear expression of TIM and PER co-expressed in S2 cells with IMPα1 (S8A Fig.).

We also assayed interactions between PER and IMPα1 by immunoprecipitating PER with an anti-HA antibody (against PER) and assaying the pellets for expression of IMPα1. Wild type IMPα1 was not pulled down with PER regardless of the presence of TIM. IMPα1∆IBB was pulled down by PER, but only when TIM was co-expressed, suggesting that TIM serves as a bridge between PER and IMPα1∆IBB (Fig. 6B). Based upon sequence analysis, PER has at least two monopartite NLSs: one near the N-terminus (aa 73–77) and the other in the middle of the protein (aa 788–791). However, mutations in these NLSs have little to no effect on the nuclear localization of PER [6]. Chang et al. identified a functional bipartite NLS near the C-terminus between amino acids 813–840. However, mutations in this NLS did not affect the nuclear entry of PER and TIM co-expressed with IMPα1 (S8A Fig.). Thus, TIM is likely the primary target of IMPα1 and its nuclear entry is regulated by the canonical nuclear entry mechanism described above; however, at the same time TIM acts as a carrier for PER.

To assay the interaction between IMPα1 and PER/TIM in flies, we immunoprecipitated HA-tagged IMPα1 from head extracts of flies expressing IMPα1-HA under control of the TUG driver. TIM was pulled down in the early night as well as the late night (). PER was only detected over background in the late night, but we note that PER levels are also substantially higher at this time. S8B Fig.

Although the TIM structure has not been experimentally determined yet, it was previously proposed to belong to an ARM/HEAT protein superfamily [37]. ARM (for Armadillo) and HEAT (for Huntingtin, Elongation factor 3, regulatory subunit A of Protein phosphatase 2A, and Target of rapamycin) repeats are structural units of two (HEAT) or three (ARM) α-helices which form one turn of a superhelix [38]. Intriguingly, karyopherins (both importins and exportins) and even some nuclear pore complex proteins belong to the ARM/HEAT family. Using diverse up-to-date homology modeling programs, we also observed that TIM is weakly homologous to proteins in the ARM/HEAT superfamily or more specifically to karyopherins (S8C Fig.). In addition, TIM shuttles in and out of the nucleus, as do importin molecules [12]. Taken together, these data (requirement of TIM for PER nuclear entry, TIM expression at the nuclear rim, TIM binding with NUP153 and Ran, and homology modeling of TIM) support the idea that TIM serves as a karyopherin-like protein for PER.

We sought to determine if the mechanism uncovered here contributes to the cytoplasmic expression of TIM and PER caused by the mutant tim^PL allele. This proline 115 to leucine mutation leads to behavioral and molecular phenotypes similar to those observed in IMPα1 mutant flies. In addition, mutation of a nearby threonine residue at 113 (TIM^TA) recapitulates the cytoplasmic expression of TIM^PL in S2 cells and the behavioral/molecular phenotypes of tim^PL mutant flies [25]. We assayed interactions of each of these proteins with IMPα1∆IBB. In transfected S2 cells, we observed interactions between wild-type TIM and IMPα1∆IBB, but neither mutant TIM, TIM^PL or TIM^TA, bound IMPα1∆IBB (Fig. 6C and S9 Fig.). Lack of an interaction with IMPα1 likely accounts for the cytoplasmic localization, and thereby arrhythmic behavior, of the tim^PL mutant. These data suggest that the interaction with nuclear transport machinery is dependent on the phosphorylation state of TIM. Phosphorylation likely drives a conformational change in TIM, exposing the NLS for interaction with the cargo-binding site on IMPα1.

Together these data indicate that TIM is primary cargo for IMPα1 and it is transported through a classical nuclear entry mechanism, but at the same time, it co-transports PER and thus acts as a karyopherin-like molecule.

RNAi stocks were obtained from VDRC and Bloomington stock centers. Df(3L)α1S1, importin α1 null mutant, flies were kindly provided by Dr. R. Fleming [28]. RanDN flies were kindly provided by Dr. W. Odenwald [50]. The UAS-importin α1-HA construct was generated by inserting the coding region of importin α1 tagged with HA epitope into the pUAST-attB vector. Transgenic fly lines carrying this construct were generated by the site-specific PhiC31 Integration System (Rainbow Transgenics) using the attP on the 2^nd chromosome [51]. These transgenic flies were outcrossed 5–6 times into an isogenic w¹¹¹⁸ (iso³¹) strains. For the behavioral assay, male flies were entrained to 12 h light/dark cycles at 25°C for a least 3 days. Locomotor activity rhythms were measured for 7 to 12 days in DD or LD as previously described [52].

3–5 day old flies were collected at indicated Zeitgeber times (ZT) on the 4th day of LD entrainment or on the indicated day of LD after drug (RU486 or EtOH) treatment. Fly heads of each genotype were dissected open, and brains were immediately fixed with 4% paraformaldehyde (in 1× PBS) for 15–20 minutes, followed by dissection in 1× PBST (0.3% Triton X-100 in PBS) at room temperature. After a 30min-wash with 1x PBST at room temperature, brains were blocked with 5% normal donkey serum (NDS) for 20 minutes, and then incubated overnight at 4°C with primary antibody: rat anti-PER (UPR34, 1:1000), anti-TIM (UPR42, 1:500), and mouse anti-PDF (C7, Developmental Studies Hybridoma Bank, 1:500). After a 30-min wash in 1× PBST at room temperature, brains were incubated with secondary antibodies (Jackson ImmunoResearch Laboratories, 1:500) for two hours in NDS at room temperature, followed by an extensive 30min-wash. Samples were imaged using a Leica TCS SP5 confocal microscope. Eight to ten brains were examined for each time point.

Fly heads were collected on dry ice at indicated time points in LD. Total RNA was isolated using the Trizol isolation system (Life Technologies). After DNase treatment, cDNAs were synthesized by using a high-capacity cDNA Reverse Transcription kit (Applied Biosystems). RT-PCR was performed on an ABI prism 7100 using a SYBR Green kit (Applied Biosystems). The sequences of primers used for real-time PCR are as follows: per (fwd: 5'- CCAGATTCCCGAACGTCCGT-3'; rev: 5'- GCAGGAGTGGTGACCGAGTG-3'), tim (fwd: 5'-CCAATGGACAAAAAGGAGCTTAGA-3'; rev: 5'-GTAACCCTTG AGGAGGAAATCCAC-3'), importin α1 (fwd: 5'-CCAATGATAAAATCCAGGCT GTAA-3'; rev: 5'-GGCTAATGCAGGT CAAAGCGTTGT-3'), importin α2 (fwd: 5'-GGCACAGATCAACAGACTGACG-3'; rev: 5'- TGCTTCTGGTTACCTGCTGT GA-3'), importin α3 (fwd: 5'- ACCTTGATCAAGGAGGGCG TCATT-3'; rev: 5'- TTCCTCAATGCAGTTGGCCACCGC-3'), importin α4 (fwd: 5'- CAAAATTC GAGCCGACGCCGCAGA-3'; rev: 5'- AATATACCCTTTTCGCAGACTTCA-3'), and actin (fwd: 5'-GCGCGGTTACTCTTTCACCA-3'; rev: 5'-ATGTCACG GACGATTTCACG-3').

At the indicated time points in LD or DD, eight to ten fly heads were collected on dry ice and homogenized in EB1 lysis buffer (1X CompleteEDTA-free Protease Inhibitor (Roche), 20mM HEPES pH 7.5, 100mM KCl, 5% glycerol, 2.5mM EDTA, 5mM DTT, 0.1% Triton X-100, 25mM NaF, 0.5mM PMSF) [53]. After 30-minute incubation on ice, fly head extracts were spun down, boiled and resolved on 4–12% Tris-Glycine gels (Novex; Life Technologies). Gels were transferred to nitrocellulose membrane and probed with the following antibodies: guinea pig anti-PER (PA1141, 1:1000), rat anti-TIM (UPR42, 1:1000), and mouse anti-Hsp70 (sigma, 1:5000). Western blot assays were repeated two or three times with similar results.

The PCR-amplified-coding region of importin α1-VSV was cloned into a pIZ/V5-His vector (Invitrogen). pCaspeR-hs-per-CFP and pCaspeR-hs-tim-YFP were used as previously described [9], and also with advice of Taichi Hara (former Sehgal lab member). S2 cells were cultured in a standard Schneider medium and transfected with these constructs as indicated using an Effectene kit (Qiagen) according to manufacturer's protocol. For RNA interference of importin α1 in S2 cells, the first 400 bp and last 400 bp of a full-length Drosophila importin α1 cDNA was PCR amplified using standard PCR techniques and primers encoding a T7 transcription promoter sequence in both directions. In vitro transcription followed by DNAse I digestion was used to generate double stranded RNA (Ambion), as per manufacturer's protocols. Medium from cells transfected in a 6-well plate was replaced with 1 ml Schneider's medium lacking FBS a day after transfection and 1 ug of each RNA preparation was added to each well. After 1 hour of shaking incubation at room temperature, 1 ml of Schneider's medium with 20% FBS was added to the cells to bring the FBS levels to 10%. After 48 h, cells were treated with heat shock (37°C) for 30 minutes and then fixed 8 h or at indicated times after heat shock induction. S2 cells were scored as previously described [25]. At least 100 cells were scored for each condition in two or three independent blind tests.

For immunoprecipitation assays in S2 cells, cDNAs encoding per and tim were cloned into the expression vectors pAc-HA and pIZ/V5-His, respectively. NLS mutants for each construct (pAc-per_mNLS-HA and pIZ-tim_mNLS-V5) were generated by site-directed mutagenesis with the Quick Change mutagenesis kit (Stratagene). The cDNAs of importin α1 lacking the IBB domain (importin α1∆IBB) and wild-type were amplified and tagged with a VSV epitope using PCR and cloned into a pIZ/V5-His vector. Both pIZ-tim^PL-V5 and pIZ-tim^TA-V5 were generated as previously described [25]. S2 cells were transfected with various combinations as indicated. After 60 h, cells were collected and lysed in lysis buffer (1X CompleteEDTA-free Protease Inhibitor (Roche), 50 mM Tris-HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.5% Triton X-100). Extracts were incubated with anti-V5 or anti-VSV antibodies at 4°C for 4 h and followed by 2 h-incubation with Protein G or A Dynabeads (Invitrogen). After washing, bound proteins were eluted and subjected to western blot analysis with the following antibodies: rabbit anti-NUP153 antibody (a gift from Dr. M. Capelson [54], 1:1000), mouse anti-V5 (Invitrogen, 1:1000), anti-HA (Sigma, 1:1000) and anti-VSV (Sigma, 1:1000). To perform the immunoprecipitation from fly head extracts, fly heads were homogenized in EB1 lysis buffer. Extracts were incubated overnight with anti-HA beads (sigma) at 4°C and washed three times with lysis buffer. Proteins were eluted and subjected to western blot analysis. For the GST pull-down assay, the FG repeat of Nup153 or Ran (for GDP-bound) or Ran Q69L (for GTP-bound) was cloned into a pGEX-4T-1 vector (GE Healthcare). These GST-fusion proteins were expressed in E. coli and purified using glutathione-Sepharose 4B beads (GE Healthcare). GST-Ran proteins were loaded with GDP or GTP γS using the Ran activation assay kit from New East Biosciences. GST alone (control) or GST-fusion proteins on beads were incubated with TIM-V5 expressed in S2 cells for 4 h at 4°C. After washing, bound TIM-V5 was detected by using a mouse anti-V5 antibody (Invitrogen).

Secondary structure prediction programs detected two alpha-helical regions in TIM. These two regions were threaded with different homology modeling programs: PHYRE [55], Robetta [56], ITasser [57], pGenThreader [58], HHPred [59] and Raptor X [60].

All relevant data are within the paper and its Supporting Information files.