Intersectin 1-short (ITSN1-s) is a 1220 amino acid ubiquitously expressed scaffold protein presenting a multidomain structure that allows to spatiotemporally regulate the functional interaction of a plethora of proteins. Besides its well-established role in endocytosis, ITSN1-s is involved in the regulation of cell signaling and is implicated in tumorigenesis processes, although the signaling pathways involved are still poorly understood. Here, we identify ITSN1-s as a nucleocytoplasmic trafficking protein. We show that, by binding to importin (IMP)α, a small fraction of ITSN1-s localizes in the cell nucleus at the steady state, where it preferentially associates with the nuclear envelope and interacts with lamin A/C. However, upon pharmacological ablation of chromosome region maintenance 1 (CRM-1)-dependent nuclear export pathway, the protein accumulates into the nucleus, thus revealing its moonlighting nature. Analysis of deletion mutants revealed that the coiled coil (CC) and Src homology (SH3) regions play the major role in its nucleocytoplasmic shuttling. While no evidence of nuclear localization signal (NLS) was detected in the CC region, a functional bipartite NLS was identified within the SH3D region of ITSN1-s (RKKNPGGWWEGELQARGKKRQIGW-1127), capable of conferring energy-dependent nuclear accumulation to reporter proteins and whose mutational ablation affects nuclear import of the whole SH3 region. Thus, ITSN1-s is an endocytic protein, which shuttles between the nucleus and the cytoplasm in a CRM-1- and IMPα-dependent fashion.
Scaffold proteins (scaffolds) are important components of several cellular processes and signaling systems. Usually, scaffolds are soluble proteins devoid of enzymatic activity, containing several modular protein interaction domains within their structure, together with sites for inducible posttranslational modifications . Scaffolds control the assembly of multiprotein complexes, thus contributing to localize signaling molecules to specific cell compartments or/and to regulate the efficiency of signaling pathways . For example, Grb2-associated binding protein 2 is involved in the assembly of signaling systems downstream of receptor tyrosine kinases and non-receptor tyrosine kinases , while Shc proteins have roles in signaling via many different types of receptors, such as growth factor receptors, antigen receptors, cytokine receptors, G-protein-coupled receptors, hormone receptors, and integrins .
Accumulating evidence suggests that several cytoplasmic adaptor proteins involved in endocytosis, such as clathrin, adaptor protein containing a pleckstrin homology domain, a phosphotyrosine-binding domain, and a leucine zipper motif 1 (APPL1) and Beta arrestin1, are capable of shuttling between the nucleus and the cytoplasm, being involved in nuclear signaling and transcriptional events in response to extracellular signals [5–7]. Therefore, a detailed understanding of the cellular compartmentalization dynamics of adaptor proteins is crucial to gain insights regarding their function.
In eukaryotes, nucleocytoplasmic transport of cargoes larger than 50–60 kDa is a signal- and energy-dependent process, which takes place across aqueous channels, delimited by nuclear envelope (NE)-embedded nuclear pore complexes (NPCs). Members of the karyopherin superfamily recognize specific nuclear targeting signals (NTSs) responsible for targeting cargoes either into or out of the nucleus, nuclear localization or nuclear export signals (NLSs or NESs), respectively . Usually, nuclear import is mediated by importins (IMPs), such as IMPβ1 or one of its homologs, after recognition of cargo-bearing NLSs, either directly or through the adaptor molecule IMPα. IMPα recognizes short basic NLSs — also named ‘classical’ NLS (cNLS). cNLSs can be classified as monopartite — matching the consensus K-(K/R)-X-(K/R), or bipartite — matching the consensus: [(K/R)(K/R)−X10–12−(K/R)3/5], where X is any amino acid, and (K/R)3/5 represents three lysine or arginine residues out of five consecutive amino acids. Subsequently, complexes are translocated through the NPCs into the nucleus, whereby binding of RanGTP to IMPβ promotes their dissociation and cargo release .
On the other hand, proteins are exported from the nucleus by exportins such as chromosome region maintenance 1 (CRM-1), the so far best-characterized exportin. In the nucleus, RanGTP-complexed CRM-1 recognizes cargoes bearing short NESs containing four nonconsecutive hydrophobic residues , and it translocates them to the cytoplasm, where RanGTP–CRM1–cargo complexes are dissociated and RanGTP is hydrolyzed to RanGDP and this allows CRM1 to be recycled back to the nucleus .
Intersectin 1 (ITSN1) is a ubiquitously expressed scaffold protein present in a long and short (ITSN1-s) isoform of 190 and 145 kDa, respectively. ITSN1-s contains two N-terminal Eps15 Homology domains (EHs), a coiled coil (CC) region, and five SRC homology 3 (SH3A–E) domains. Owing to its multimodular architecture, it interacts with several proteins involved in clathrin- and caveolin-mediated endocytosis, rearrangements of the actin cytoskeleton, cell signaling and survival [12,13]. Indeed, EH domains recognize the Asn-Pro-Phe (NPF) motif of many endocytic machinery proteins, including Epsin [14,15], while the CC region mediates protein dimerization and interacts with Eps15 and Eps15R , as well as with SNAP23, SNAP25, and HIP . The SH3 domains, typical of cytoskeleton proteins and proteins involved in signal transduction, recognize proline-rich motifs of endocytic proteins such as Dynamin, Synapsin, and Synaptojanin , as well as the SH3 domain of Endophilin .
Analysis of HeLa nuclei phosphoproteome revealed the presence of peptides derived from ITSN1-s, suggesting that ITSN1-s could have access to the nucleus and play a role therein, similarly to other scaffold adaptor proteins .
In the present study, we report that a small, but significant fraction of ITSN1-s is present at the steady state in the nucleus of HeLa and HEK 293 cells, where it accumulates on the NE. We show here, for the first time, that ITSN1-s is able to bind to IMPα and shuttle between the cytoplasm and the nucleus in a CRM-1-dependent fashion. We also identify a bipartite NLS located at residues 1104–1127, capable of conferring energy- and Ran-dependent nuclear import abilities to a reporter protein.
Cell culture and transfections
HeLa, HEK 293-A, and HEK 293-T cells were maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) fetal bovine serum (FBS), 50 U/ml penicillin, 50 U/ml streptomycin, and 2 mMl-glutamine as described previously [21–23]. For confocal laser scanning microscopy (CLSM) experiments, 3 × 104 HEK 293-A or 2.5 × 104 HeLa cells were seeded onto polylisinated 12 mm glass coverslips in 24-well plates the day before transfection using Lipofectamine 2000 (Thermo Fisher) according to the manufacturer's recommendations . For live cell imaging experiments, 6 × 105 HeLa cells were seeded in a glass bottom six-well plate (P06G-0-10F, MatTec) the day before transfection using Lipofectamine 3000 (Thermo Fisher) according to the manufacturer's recommendations. For cytosol/nucleus separation of endogenous and FLAG-ITSN1-s, and for immunoprecipitation between GFP-IMPαΔIBB- and FLAG-tagged fusion proteins, 5 × 105 HeLa cells and HEK 293-T cells were seeded in 60 mm diameter dish the day before transfection using Lipofectamine 2000. All experiments were carried out 48 h post transfection.
Mammalian expression plasmids were generated using the Gateway™ technology (Invitrogen). ITSN1-s regions of interest were amplified with appropriate primer pairs containing attB sites using plasmid FLAG-ITSN1-s  (a kind gift of Peter S. McPherson, McGill University, Montreal, Quebec, Canada) and cloned into plasmid vector pDNR207 (Invitrogen) via BP recombination, to generate entry clones, as previously described . Entry clones were then used to generate C-terminal YFP fusion mammalian expression vectors following LR recombination reactions with the pDESTnYFP  Gateway compatible vector, as described previously . All vectors were confirmed by sequencing. Point mutant derivatives carrying amino acid substitutions of ITSN-NLS (RKKNPGGWWEGELQARGKKRQIGW-1127 to aatNPGGWWEGELQARGatsQIGW-1127) were generated using appropriate oligo pairs and the QuikChange Mutagenesis Kit (Agilent Technologies), as described previously . As positive and negative controls for response to Leptomycin B (LMB) treatment, plasmids GFP-Rev (2–116), encoding the Rev protein from HIV-1, which shuttles between the nucleus and the cytoplasm in a CRM-1-dependent manner, and GFP-UL44ΔNLS driving the expression of an exclusively cytosolic version of human cytomegalovirus DNA polymerase processivity factor UL44, were used in addition to the pEGFP-C1 (Clontech) expression vector [30,31]. Plasmid pEGFP-C1-mIMPαΔIBB, encoding a GFP-tagged deletion mutant of mouse IMPα6 lacking the autoinhibitory IMPβ binding (IBB) domain, therefore binding to cNLSs with high affinity , was described elsewhere . pDESTnFLAG-UL44, a plasmid mediating expression of a FLAG-tagged version of UL44 which is known to be transported into the nucleus by the IMPα/β heterodimer, was used as a positive control in immunoprecipitation experiments with GFP-mIMPαΔIBB . Plasmid pDESTnYFP-NLS[R], encoding a fusion protein between YFP and a minimal cNLS derived from Simian Virus 40 Large Tumor antigen, mediating nuclear targeting via IMPα/β, was described elsewhere (Smith et al., unpublished observations).
Confocal laser scanning microscopy
HeLa cells were treated as described in ref. . Briefly, cells were washed with PBS, fixed with 4% paraformaldehyde (PFA) for 10 min at room temperature (RT), permeabilized with 0.2% Triton X-100, 2 mg/ml BSA, and 1 mM NaN3 in PBS on ice three times for 10 min. Cells were subsequently incubated with blocking solution (0.02% Triton X-100, 3% BSA, and 1 mM NaN3 in PBS). Primary antibodies α-FLAG (Sigma) and α-lamin A/C (Thermo Fisher) were incubated for 30 min in blocking solution and washed three times for 10 min with wash buffer (PBS 0.02% Triton X-100, 1.5% BSA, and 1 mM NaN3). Secondary antibodies (goat α-mouse Cy3, goat α-rabbit Cy2, Jackson ImmunoResearch) were incubated for 45 min and washed as described above. Coverslips were mounted using an antifade mounting medium (ProLong Gold—Invitrogen) on a glass slide. CLSM was performed on a ZEISS LSM 510 META confocal laser scanning microscope using the 63× or the 100× Plan-NEOFLUAR oil immersion objective. To analyze the subcellular distribution of spontaneously fluorescent fusion proteins expressed in HEK 293-A cells, cells were transfected and, after 48 h, washed twice in PHEM buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, and 4 mM MgSO4) to preserve the cellular cytoskeleton , before being fixed with 3% PFA in PHEM buffer for 15 min at RT and mounted onto glass coverslips with FluoromountG (Southern Biotech). Samples were processed by CLSM using a Leica TCT-SP2 system, equipped with a Planapo fluor 63× oil immersion objective (Leica). At least four randomly chosen fields were acquired, and a total of at least 30 cells, expressing the fusion proteins of interest to similar levels, were analyzed for each repetition. The Fn/c values were determined using the NIH ImageJ 1.62 public domain software, from single-cell measurements for each of the nuclear (Fn) and cytoplasmic (Fc) fluorescence, after the subtraction of fluorescence due to autofluorescence/background . Data were plotted and statistically analyzed using Prism 6 (GraphPad) software.
Live cell imaging
HeLa cells were transfected to express the spontaneously fluorescent fusion protein of interest. Forty-eight hours later, the medium was substituted with complete phenol red-free DMEM (A18967, Life Technology). Cells were imaged with a DMi8 inverted microscope (Leica), equipped with a 40× NA 0.6 objective, a motorized stage, and a heated/humidified chamber, at 37°C and 5% CO2. Several positions have been recorded for each sample. Phase-contrast image served as a reference for cytoplasm vs. nuclei masking. When required, the Fn/c values relative to each fusion protein were calculated as described above.
In indicated cases, LMB (Sigma L2913; 2.9 ng/ml) was added to cells 8 h before processing samples for imaging, as previously described . Intracellular ATP was depleted by incubating cells for 2 h at 37°C in DMEM lacking phenol red and glucose (Thermo Fisher), supplemented with 10 mmol/l sodium azide and 6 mmol/l 2-deoxy-d-glucose (Sigma), as described previously .
Identification of putative NTSs on ITSN1-s
Samples were heated in reducing SDS sample buffer (80 mM Tris, pH 6.8, 2% SDS, 7.5% glycerol, and 0.01% bromophenol blue) supplemented with 2% 2-mercaptoethanol for 5 min at 95°C and separated by SDS–PAGE on acrylamide/bisacrylamide gels and analyzed by Western blot (WB), as previously described . Briefly, samples were transferred for 1 h onto the PVDF membrane, blocked for 1 h at 37°C with 5% fat-free milk and 0.05% Tween-20 in PBS. PVDF membranes were incubated with specific antibodies: α-ITSN (Abcam), α-FLAG (Sigma), α-lamin A/C (Thermo Fisher), α-ERK 1/2 (Santa Cruz), anti-histone H3 (Santa Cruz), α-tubulin (Santa Cruz), or α-GFP/GST (a generous gift from Prof. Höning, University of Cologne) overnight in PBS Tween 0.05% + 1% fat-free milk. The membrane was washed three times for 10 min with PBS Tween 0.05% and incubated 1 h with one of the following secondary antibodies horseradish peroxidase-conjugated: goat α-mouse and goat α-rabbit (Bethyl), and rabbit α-goat (Pierce). Blots were detected using Immobilon Western Classico or Forte (Millipore). Images were acquired using a G:Box Chemi XT Imaging system (Syngene) .
Co-immunoprecipitation of protein complexes
HeLa or HEK 293-T cells were washed with PBS 1× and harvested 48 h post transfection, as described in ref. . Cells were centrifuged at 800 g for 10 min, and pellet was resuspended in ice-cold lysis buffer, 50 mM HEPES, 100 mM NaCl, and 1% NP-40 + protease inhibitor (PI) cocktail [1 : 1000 (Roche)] for 10 min on ice. Cells were then sonicated at 15% of instrument power (Sonopuls, Bandelin) for 10 s. After clarification, supernatants were incubated with 4 µg of the α-ITSN, α-Lamin, α-FLAG antibodies or without antibody as negative control as indicated, overnight at 4°C, with gentle rocking. The following day, 50 µl of protein A/G beads (Santa Cruz Biotechnology) were added and the mixtures were incubated for 4 h. After three washes with PBS 1×, beads were resuspended in SDS sample buffer for 5 min at 95°C and centrifuged at 800 g for 3 min before the supernatant was collected. For immunoprecipitation of GFP-mIMPαΔIBB in the presence of FLAG-ITSN1-s or not, the protocol has been adapted from ref. [33, 42]. HEK 293-T cells were washed with PBS 1× and harvested 48 h post transfection. Cells were centrifuged at 800 g for 10 min and the pellet was resuspended in ice-cold lysis buffer, 50 mM HEPES, 100 mM NaCl, 1 mM MgCl2, and 1% NP-40 + PI cocktail [1 : 1000 (Roche)] for 10 min on ice. Cells were then sonicated, as described above. After clarification, supernatants were incubated with 4 µg of the α-FLAG as indicated, overnight at 4°C, with gentle rocking. The following day, 50 µl of protein A/G beads were added and the mixtures were incubated for 4 h. After three washes with 50 mM HEPES, 100 mM NaCl, 1 mM MgCl2, and 1% NP-40, beads were resuspended in SDS sample buffer and treated as described above. Samples were subjected to SDS–PAGE/WB analysis.
To separate nuclei from the cytoplasm, a protocol from Nabbi and Riabowol has been adapted from ref. . Briefly, HeLa cells grown to 90% confluency in 10 cm dishes were washed with ice-cold PBS and scraped. Cells were centrifuged at 10 000 g for 10 s. Cell pellets were resuspended in PBS with 0.1% NP-40 and triturated with a P1000 micropipette. The lysed cell suspension was centrifuged at 10 000 g for 10 s. The supernatant representing the cytoplasmic fraction was isolated, and the nuclei were gently resuspended in PBS with 0.1% NP-40 and centrifuged again. Nuclear pellet was resuspended in SDS sample buffer and sonicated twice at 15% of instrument power (Sonopuls, Bandelin) for 6 s on ice. Fractions were analyzed by SDS–PAGE/WB, as described above.
To separate the nucleoplasm and the NE, nuclei obtained as described above were resuspended in NP-40 buffer [PBS 0.5×, 10 mM MgCl2, 50 mM MOPS (pH 7.4), 0.5% NP-40, DNAase 5 U/ml, and PI 1 : 1000] and incubated on ice for 5 min. Samples were centrifuged at 20 000 g for 10 min and supernatants were collected and considered as nucleosol. Pellets (NE) were resuspended in RIPA buffer [150 mM NaCl, 1% NP-40, 50 mM Tris–HCl (pH 8), 0.1% SDS, 0.5% sodium desossicolate, and PI] and tip-sonicated 12 s at 15% of instrument power (Sonopuls, Bandelin). Fractions were analyzed by SDS–PAGE/WB, as described above.
Chemical cross-linking to detect DNA-binding proteins
To visualize if ITSN1-s is a DNA-binding protein, a protocol from Qiu and Wang  has been adapted. HeLa cells (10 cm dish, 90% confluency) were trypsinized and collected by centrifugation at 300 g for 5 min at 4°C. Cells were washed with ice-cold PBS to remove culture medium and FBS. In vivo cross-linking was achieved by adding PFA to 1 ml of cell suspension in PBS to obtain a final concentration of 1% (w/v). After incubating at RT, PFA was quenched by glycine to a final concentration of 125 mM and incubated at RT for 5 min. The cross-linked cells were collected by centrifugation (300 g at 4°C for 5 min), and the cell pellet was washed twice with cold PBS. Nuclei isolation was carried out using a protocol adapted from that reported by Henrich et al. . The cross-linked HeLa cell pellet (2 × 10 cm dishes, 70% confluent) was resuspended in 10 volumes of ice-cold hypotonic lysis buffer A containing 10 mM HEPES (pH 7.4), 10 mM KCl, 1.5 mM MgCl2, 1 mM DTT, and PI 1 : 1000. After incubation on ice for 30 min, 0.5% (v/v) NP-40 was added to the lysis buffer. Cells were then gently lysed with a Dounce homogenizer and the nuclear fraction was collected by centrifugation. The nuclear fraction resuspended in buffer B [250 mM sucrose, 10 mM MgCl2, 20 mM Tris–HCl (pH 7.4) and 1 mM DTT] was layered over a two-step sucrose gradient cushion [1.3 M sucrose, 6.25 mM MgCl2, 20 mM Tris–HCl (pH 7.4), 0.5 mM DTT above 2.3 M sucrose in 2.5 mM MgCl2, and 20 mM Tris–HCl (pH 7.4)] and centrifuged subsequently at 19 000 g at 4°C for 45 min. The isolated nuclei were washed with buffer A and collected by centrifugation at 1000 g.
Isolation of DNA–protein complexes, cross-linking reversal, and DNA removal were performed, as described previously . The purified DNA-binding proteins were separated using a NuPage 4–12% Bis–Tris protein gel (Invitrogen).
Statistically significant differences between datasets were determined with Student's t-test (Graphpad software, Inc.). p-values of less than 0.05 were considered statistically significant with *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001. Values shown are the mean ± standard error of the mean (SEM) relative to at least three independent experiments [46,47].
A fraction of ITSN1-s localizes to the nucleus
To investigate ITSN1-s subcellular localization, HeLa cells were transfected with a FLAG-ITSN1-s expression plasmid and subjected to biochemical separation of nuclear and cytosolic fractions. Mock-transfected cells were used as a negative control. Then, ITSN1-s distribution was analyzed in the two fractions by SDS–PAGE/WB. ERK1/2 and lamin A/C were used as reference proteins for cytosolic and nuclear compartments, respectively. Both FLAG-tagged and endogenous ITSN1-s were enriched in the cytosol, but detectable in both fractions, while reference proteins were localized exclusively in the specific fraction, proof of a clean compartment separation (Figure 1A). We also confirmed our results by immunofluorescence (IF), limiting our analysis to FLAG-ITSN1-s due to the lack of suitable commercial antibodies. The protein presented mainly a cytoplasmic localization, although it was also visible in the cell nucleus (Figure 1B). Thus, our data demonstrate the existence of a nuclear pool of ITSN1-s.
A fraction of endogenous and overexpressed ITSN1-s localizes within the cell nucleus.
ITSN1-s is enriched in the NE
The relative amount of nuclear FLAG-ITSN1-s detectable in the nucleus after subcellular fractionation appears higher than that observable by IF (Figure 1), raising the possibility that the majority of nuclear FLAG-ITSN1-s are associated with the NE rather than in the nucleoplasm. To verify this hypothesis, nucleoplasmic and NE fractions from purified nuclei of HeLa cells transfected to express FLAG-ITSN1-s were analyzed by SDS–PAGE/WB to detect the cytosolic marker ERK1/2, the NE marker lamin A/C, as well as FLAG-ITSN1-s. As expected, while ERK1/2 was not detectable in the nuclear fraction, lamin A/C localized exclusively in the NE fraction (Figure 2A). FLAG-ITSN1-s was detectable in the nucleoplasmic and the NE fractions, but highly enriched in the latter, consistent with the idea that ITSN1-s could associate with the NE. IF analysis revealed that FLAG-ITSN1-s partially colocalizes with lamin A/C on the NE (Figure 2B). Subsequently, we investigated the interaction between ITSN1-s and lamin A/C by co-immunoprecipitation (co-IP) analysis. HeLa cells were lysed, and ITSN1-s complexes pulled down using either an α-ITSN or an α-lamin A/C antibody. SDS–PAGE/WB analysis revealed the presence of two bands corresponding to lamin A and C in the ITSN1-s pull-down lane and of a ITSN1-s band in the lamin A/C pull down (Figure 2C).
Nuclear ITSN1-s is associated with the NE.
Finally, since some endocytic proteins were found to be part of transcription complexes , we tested the possibility that ITSN1-s could bind DNA. To this end, we adapted a protocol which includes chemical cross-linking of cells followed by nuclei isolation and purification of covalently bound DNA–protein complexes . DNA–protein complexes were isolated and analyzed by a SDS–PAGE/WB using α-ITSN1, in addition to α-tubulin and α-Histone H3 antibodies as negative and positive controls, respectively . As expected, Histone H3 was detectable in the DNA–protein complexes (DPC) lane as well as in the cell homogenate (H), whereas tubulin, which is incapable of binding to DNA, was detectable only in the cell homogenate (Figure 2D). Importantly, FLAG-ITSN1-s could not be detected in the DPC, suggesting that ITSN1-s does not interact with cellular DNA. Taken together, our results indicate that a small, but detectable amount of ITSN1-s localizes within the cell nucleus at the steady state, where it preferentially associates with the NE, possibly interacting with lamin A/C.
ITSN1-s interacts with IMPα
The fact that a certain fraction of the 145 kDa protein ITSN1-s can gain access to the nucleus implies that the protein is actively translocated into the nucleus in an NLS-dependent fashion by IMPs. A bioinformatics analysis using the NLS prediction software ‘cNLS MAPPER’ revealed the presence of a putative bipartite cNLS (RKKNPGGWWEGELQARGKKRQIGW-1127), located in the C-terminal, SH3 portion of ITSN1-s (Figure 3A,B). Such sequence strongly resembles the prototype bipartite cNLS originally described on nucleoplasmin (KRPAATKKAGQAKKKK-170) and perfectly matches the consensus for such signals. This finding raised the possibility that ITSN1-s could bind to the IMPα/β heterodimer. To assess this hypothesis by co-IP experiments, HEK 293-T cells were transfected to express GFP-IMPαΔIBB either individually or in the presence of FLAG-ITSN1-s. The former is a GFP-tagged derivative of IMPα that lacks the autoinhibitory IBB domain, and that therefore binds to NLSs with an affinity comparable to that of the IMPα/β heterodimer . As a positive control, GFP-IMPαΔIBB was also expressed in the presence of FLAG-UL44, a protein known to be recognized by the IMPα/β heterodimer . Proteins were subsequently immunoprecipitated in the presence or in the absence of the FLAG antibody. As expected, FLAG-UL44 could be co-immunoprecipitated with GFP-IMPαΔIBB, whereas no GFP-IMPαΔIBB was obtained after incubation of cell lysates in the absence of the αFLAG antibody, indicating that GFP-IMPαΔIBB did not interact unspecifically with protein A/G beads (Figure 4A). Importantly, GFP-IMPαΔIBB could be also co-immunoprecipitated by the α-FLAG antibody from cells co-expressing FLAG-ITSN1-s, but not from cells expressing GFP-IMPαΔIBB alone, indicating that GFP-IMPαΔIBB did not interact unspecifically with the α-FLAG antibody (Figure 4B). Taken together, these results show that ITSN1-s is able to interact with IMPα, thus being actively transported into the nucleus.
ITSN1-s contains several putative NTSs.
ITSN1-s interacts with IMPα.
ITSN1-s is a nucleocytoplasmic trafficking protein
Despite its ability to bind to IMPα, at the steady state, ITSN1-s preferentially localizes in the cytosol. It is therefore possible that, as described for other endocytic proteins , ITSN1-s shuttles between the nucleus and the cytoplasm thanks to the simultaneous presence of NLS and NES. A bioinformatics analysis using the NES prediction software ‘NES finder’, integrated by visual inspection based on a systematic analysis of NES consensus performed in eukaryotic cells , identified six putative hydrophobic-rich NESs [NES A–F] distributed throughout the three regions of ITSN1-s (EH-like homology; EHs, aa 1–380; CC, aa 381–680; and SH3s, aa 681–1220) (Figure 3A,B). The presence of putative hydrophobic NESs within ITSN1-s indeed suggested that the protein could undergo nucleocytoplasmic shuttling in a CRM-1-dependent manner.
To confirm this hypothesis, the subcellular localization of ITSN1-s was tested in HeLa cells upon LMB-mediated inhibition of CRM1 activity. FLAG-ITSN1-s transiently transfected cells were treated or not with LMB, before CLSM analysis to quantify their levels of nuclear accumulation. In addition to FLAG-ITSN1-s, cells were also transfected with GFP, GFP-Rev, a protein known to accumulate in the nucleus after LMB treatment, as well as with GFP-UL44ΔNLS, an ∼180 kDa dimeric cytosolic protein which is excluded from the nucleus due to the lack of a functional NLS. The addition of LMB did not affect the subcellular localization of either GFP or GFP-UL44ΔNLS, implying that the treatment did not affect cell viability and morphology as well as NPC permeability, while it caused GFP-Rev to strongly accumulate in the cell nucleus (Supplementary Figure S1 and Table S1). Importantly, after treatment with LMB, FLAG-ITSN1-s nuclear staining became significantly more evident (Fn/c from 0.15 to 0.45), although the protein did not accumulate in the cell nucleus to high levels (Figure 5A,B and Supplementary Table S2). Thus, ITSN1-s can shuttle between the nucleus and the cytosol in an IMPα/β- and CRM1-dependent fashion.
ITSN1-s shuttles between the nucleus and the cytoplasm in a CRM-1-dependent fashion.
ITSN1-s-CC and SH3 regions can independently undergo nucleocytoplasmic shuttling
In an effort to identify ITSN1-s functional NTSs among the putative ones predicted, we generated several ITSN1-s deletion mutants fused to the C-terminus of YFP. We initially verified that the addition of a YFP-tag did not interfere with ITSN1-s subcellular localization and nucleocytoplasmic shuttling abilities. To this end, HeLa cells were transfected to express YFP-ITSN1-s, whose subcellular localization was initially monitored by live cell imaging every min for a period of 30 min. As expected, YFP-ITSN1-s localized mainly in the cytoplasm, with evident dots reminiscent of endocytic vesicles, which appeared to be highly mobile (Supplementary Figure S2 and Movie S1). In a second series of experiments, HeLa cells were transfected to express either YFP-ITSN1-s or GFP-UL44ΔNLS, treated with either LMB or with solvent, and the subcellular localization of the fusion protein of interest visualized every 15 min for 10 h. As expected, YFP-ITSN1-s gradually accumulated into the nucleus, reaching a maximum ∼5 h after LMB addition, whereas the negative control GFP-UL44ΔNLS remained mainly cytosolic for the whole duration of the experiment (Supplementary Figure S3 and Movie S2). Therefore, YFP-ITSN1-s could be used to study the nucleocytoplasmic properties of ITSN1-s. Beside full-length ITSN1-s (1–1220), we generated plasmids mediating the expression of five additional deletion mutants: YFP-ITSN1-s-ΔEHs (316–1220), lacking the N-terminal epsin-like domain; YFP-ITSN1-s-ΔSH3s (1–680), lacking the C-terminal SH3 domains; YFP-ITSN1-s-SH3s (681–1220); YFP-ITSN1-s-EHs (1–315); and YFP-ITSN1-s-CC (316–680) (Figure 6A). As a first step toward the characterization of such variants, their molecular masses were verified by SDS–PAGE/WB using an α-GFP antibody (Figure 6B). Fusion proteins migrated at the expected molecular masses (YFP-ITSN1-s: 160 kDa, YFP-ITSN1-s-ΔEHs: 125 kDa, YFP- ITSN1-s-ΔSH3s: 101 kDa, YFP-ITSN1-s-SH3s: 85 kDa, YFP-ITSN1-s-EHs: 61 kDa, YFP-ITSN1-s-CC: 66 kDa). Multiple bands of higher molecular mass could be detected for the YFP-ITSN1-s-CC fusion, most likely due to protein multimerization/aggregation. Secondly, we decided to investigate their subcellular localization either in the absence or in the presence of LMB. HEK 293-A cells were transiently transfected with appropriate expressing vectors and the subcellular localization of YFP-ITSN1-s was compared by quantitative CLSM with that of the above-described deletion mutants. Along with YFP-ITSN1-s fusions, the previously described controls (GFP, GFP-Rev and GFP-UL44ΔNLS) were used to verify the functionality of LMB treatment (Figure 6C). As expected, GFP evenly distributed between the nucleus and the cytoplasm both in the absence and presence of LMB (Fn/c 1.27 ± 0.14 vs. 1.50 ± 0.28). Similarly, GFP-UL44ΔNLS, having a molecular mass of an ∼180 kDa, was retained in the cytosol in both cases (Fn/c 0.10 ± 0.09 vs. 0.17 ± 0.08). Importantly, LMB addition significantly enhanced nuclear accumulation of GFP-Rev (Fn/c 0.16 ± 0.03 vs. 16.97 ± 9.87), which strongly accumulated in the nucleoli. Taken together, our data indicate that LMB treatment functionally inhibited CRM-1-mediated nuclear export without affecting NE permeability (see Figure 6C,D, left panel and Supplementary Table S3). As expected, in the absence of LMB, YFP-ITSN1-s mainly localized in the cytoplasm of untreated cells (Figures 2D, right panels and 6C; Supplementary Table S4; Fn/c 0.11 ± 0.08) with a punctuate pattern present in the cytoplasm and in close proximity to the plasma membrane, thus resembling that of endocytic vesicles. Deletion of EH domains (YFP-ITSN1-s-ΔEH) altered ITSN1-s localization pattern, in that the punctuate structures observed for FL-ITSN1-s were not detectable, and large protein aggregates were often present in strongly expressing cells. However, the protein was still mainly detectable in the cytoplasm (Fn/c 0.07 ± 0.04). Removal of both EH and CC regions (YFP-ITSN1-s-SH3s) caused the protein to localize in the cytosol with a mainly diffuse pattern, although a few vesicle-like dots were still detectable (Fn/c 0.10 ± 0.04). On the contrary, removal of ITSN1-s-SH3 region (YFP-ITSN1-s-ΔSH3s) did not change the punctate pattern described for ITSN1-s but caused a significant increase in nuclear signal (Fn/c 0.42 ± 0.09). When both the CC and SH3 regions were simultaneously deleted (YFP-ITSN1-s-EHs), the subcellular localization was completely altered: this protein equally distributed between the nucleus and the cytoplasm with a diffuse pattern (Fn/c 1.13 ± 0.35). In addition, vesicle-like dots or protein aggregates were not detectable. Finally, YFP-ITSN1-s-CC fusion, devoid of both EH and SH3 regions, localized mainly in the cytoplasm, but entered the nucleus slightly more efficiently with respect to YFP-ITSN1-s (Fn/c 0.24 ± 0.08). The protein distributed with a diffuse pattern in the cytoplasm, with the presence of large protein aggregates in the cytosolic compartment of highly expressing cells.
ITSN1-s-CC and SH3 regions can independently undergo nucleocytoplasmic shuttling.
Importantly, the addition of LMB increased nuclear levels of YFP-ITSN1-s (Fn/c 0.11 ± 0.08 vs. 0.41 ± 0.11), as well as of all YFP-ITSN1-s deletion mutants, apart from YFP-ITSN1-s-EHs (Fn/c 1.13 ± 0.35 vs. 1.34 ± 0.32, see Figure 6C,D). Indeed, both ITSN1-s-CC and ITSN1-s-SH3 responded to LMB treatment (Fn/c of 0.24 vs. 0.72 and 0.10 vs. 0.78, respectively) similarly to ITSN1-s-ΔEH and ITSN1-s-ΔSH3 (Fn/c of 0.07 vs. 0.32 and 0.42 vs. 1.32, respectively). Therefore, both ITSN1-s-CC and SH3 regions were sufficient to confer LMB-dependent nucleocytoplasmic shuttling properties to YFP. Taken together, the subcellular localization of ITSN1-s and its deletion mutants at the steady state suggests that the N-terminal EH domain (to a larger extent) and the C-terminally located SH3 regions play a role in ITSN1-s localization to endocytic vesicles-like dots, while the CC domain can cause protein aggregation, when expressed outside the physiological context of FL-ITSN1-s. In addition, LMB experiments suggest that ITSN1-s could contain multiple NTSs located in the CC and SH3 regions, whereas ITSN-EH domain does not actively contribute to the protein's nucleocytoplasmic shuttling ability, and NESs [A–C] are not functional in terms of mediating CRM-1-dependent nuclear export.
ITSN1-s residues 1104–1127 represent a functional cNLS
As a first attempt toward the characterization of ITSN1-s nucleocytoplasmic shuttling process, we decided to identify its NLS. Our bioinformatics analysis could not detect any cNLS in ITSN1-s-CC region, but revealed the presence of a putative bipartite cNLS in the SH3 region, within the ITSN1-s-SH3D domain (RKKNPGGWWEGELQARGKKRQIGW-1127) (Figure 3). To validate its functionality, we generated an expression plasmid encoding such sequence C-terminally fused to YFP (YFP-ITSN1-s-NLS) and analyzed the ability of the YFP-ITSN1-s-NLS fusion protein to accumulate in the nucleus in an energy-dependent manner. We also analyzed the subcellular localization of YFP alone or of the control YFP-NLS[R] fusion protein, known to localize to the cell nucleus through interaction with the IMPα/β heterodimer (Smith et al., unpublished observations). As expected, when cells were maintained in normal media, YFP-NLS[R] accumulated into the nucleus of transfected cells (Fn/c 2.8) in significantly higher levels than YFP alone (Fn/c 1.1; Figure 7 and Supplementary Table S5). Depletion of intracellular ATP significantly decreased the nuclear accumulation of YFP-NLS[R], without affecting the subcellular localization of YFP alone (Fn/c of 1.5 and 1.1, respectively; Figure 7). Importantly, YFP-ITSN1-s-NLS accumulated into the nucleus at significantly higher levels when compared with YFP alone when cells were maintained in normal media (Fn/c 1.8), and its nuclear accumulation was significantly impaired after incubation in the energy-depletion media (Fn/c 1.3). These data clearly show that ITSN1-s residues 1104–1127 represent a basic NLS capable of conferring energy-dependent nuclear localization to heterologous proteins.
ITSN1-s residues 1104–1127 represent a bipartite cNLS able to confer energy-dependent nuclear targeting properties to heterologous proteins.
ITSN1-s-NLS is essential for nuclear targeting of ITSN1-s C-terminal domain
We decided to test the impact of ITSN1-s-NLS on nuclear targeting of the protein. We compared the subcellular localization of YFP-ITSN1-s with that of its derivative YFP-ITSN1-sΔNLS, where basic residues of its NLS (RKKNPGGWWEGELQARGKKRQIGW-1127) were substituted with hydrophobic ones (ΔNLS; aatNPGGWWEGELQARGatsQIGW-1127; see Figure 8A), either in the absence or in the presence of LMB. Under both conditions, the NLS defective derivative localized with a very similar pattern when compared with the wild-type protein (Figure 8B). This was not surprising, since deletion of the whole ITSN1-s-SH3 region, comprising ITSN1-s-NLS, did not impair the ability of the protein to enter the nucleus upon LMB treatment, implying a contribution of ITSN1-s-CC region in nucleocytoplasmic shuttling of ITSN1-s (Figure 6). To study the contribution of ITSN1-s-NLS to nuclear import of the protein independently of the contribution of ITSN1-s-CC region, we analyzed the subcellular localization of a series of ITSN1-s C-terminal domain deletions, including YFP-ITSN1-s-SH3s, encompassing residues 681–1220; YFP-ITSN1-s-SH3(A–D) comprising residues 681–1173; and YFP-ITSN1-s-SH3(D–E) comprising ITSN1-s residues 1074–1220, either carrying the wild-type or the mutated NLS (Figure 8). In the absence of LMB, both YFP-ITSN1-s-SH3 and YFP-ITSN1-s-SH3(A–D) localized mainly in the cytosol, and inactivation of ITSN-NLS did not affect the protein subcellular localization (Fn/c 0.2, see Figure 8). Upon LMB treatment, both YFP-ITSN1-s-SH3 and YFP-ITSN1-s-SH3(A–D) significantly accumulated into the nucleus (Fn/c 0.8 and 1.1, respectively), while their NLS-mutated counterparts were mainly retained in the cytosol (Fn/c 0.4), clearly showing the importance of ITSN1-s residues 1104–1127 for nuclear targeting. The contribution of ITSN1-s-NLS to nuclear import was even more evident when analyzing the subcellular localization of YFP-ITSN1-s-SH3(D–E). In the absence of LMB, this protein was detectable in the nucleus, compatible with its ability to passively enter the nucleus by passive diffusion (Fn/c 0.9), while the addition of LMB increased its nuclear accumulation (Fn/c 1.8). Importantly, in the absence of LMB, YFP-ITSN1-s-SH3(D–E)ΔNLS localized to the nucleus to lower levels (Fn/c of 0.5) when compared with its NLS bearing counterpart, further confirming the importance of ITSN1-s-NLS in nuclear import of the protein (Figure 8B,C and Supplementary Table S6). Overall, our data also suggest that ITSN1-s-NESE — located within the SH3D domain — is the functional NES within the SH3 region, while NESF — located within the SH3E domain — does not contribute to CRM-1 nuclear export. Indeed, both YFP-ITSN1-s-SH(A–D) and YFP-ITSN1-s-SH3(D–E) accumulate to the nucleus in the presence of LMB. Furthermore, site-specific mutagenesis for NESF did not abolish LMB responsiveness, nor increased steady-state nuclear localization either in the SH3(A–E) or SH3(D–E) context (data not shown). However, attempts to mutate NESE hydrophobic residues resulted in increased nuclear accumulation in the presence of LMB (data not shown), probably by interfering with the nearby-located NLS, thus precluding formal proof that NESE is a functional NES.
ISTN-NLS is important for ITSN-SH3 region nuclear accumulation.
ITSN1-s-CC shuttles between the nucleus and the cytoplasm independently of the presence of putative NTSs
Our data indicate that ITSN1-s-CC is capable of nucleocytoplasmic shuttling. However, no evident NTSs are present within its sequence, with the exclusion of a stretch of aa containing four L residues, which could resemble the NES originally described on HIV-1 Rev (NESD: LELEKQLEKQREL-419, see Figure 3). Since NESD is located at the N-terminal portion of the CC region, we decided to test its functionality by analyzing the subcellular localization of the N-terminal (residues 316–456, containing NESD) and of the C-terminal (residues 457–680) portions of ITSN1-s-CC region, either in the absence or in the presence of LMB (Figure 9A–C and Supplementary Table S7). Strikingly, in the absence of LMB, both YFP-ITSN1-s(316–456) and YFP-ITSN1-s(457–680) mainly localized in the cytosol of transfected cells (Fn/c of 0.7 and 0.3, respectively), but accumulated in the nucleus upon LMB treatment (Fn/c of 1.4 and 1.2, respectively). Surprisingly, mutation of ITSN1-s-NESD hydrophobic residues (LELEKQLEKQREL-419 to qEaEKQqEKQREL-419) did not affect protein subcellular localization either in the absence or in the presence of LMB. Therefore, ITSN1-s-CC appears capable of shuttling between the nucleus and the cytosol in the absence of evident NTSs.
ITSN1-s-CC possesses an intrinsic capability to undergo nucleocytoplasmic shuttling.
In the present work, we showed for the first time that the adaptor scaffold protein ITSN1-s, known to be involved in several signaling and endocytic processes [1,13,50], and interacting with a plethora of factors involved in endocytosis, cytoskeleton rearrangements, cell signaling, and survival [51–53], is a nucleocytoplasmic shuttling protein, which accumulates in the cell nucleus upon pharmacological ablation of CRM1-dependent nuclear export. At the steady state, a fraction of ITSN1-s localizes at the nucleus and is enriched at the NE, where it interacts with lamin A/C.
The identification of ITSN1-s nucleocytoplasmic shuttling sequences has been hampered by the evidence that both ITSN1-s-CC and SH3 regions can shuttle between the nucleus and the cytosol independently of each other (Figure 6). Moreover, the ITSN-1s-CC region is apparently devoid of evident NTSs, bearing no putative cNLSs and only one putative leucine-rich NES (NESD: LELEKQLEKQREL-419), whose mutation to qEaEKQqEKQREL-419 did not affect LMB responsiveness (Figure 9). Therefore, it is not known how ITSN-1s-CC could undergo cytoplasmic shuttling. One possibility is that the α-helix-rich CC domain can interact with the FG repeats of the NPC to allow passage in and out of the nucleus, in analogous fashion as it has been demonstrated for HEAT repeats containing proteins such as IMPβ . Another possibility is that one of the several cellular proteins capable of interacting with ITSN1-s-CC region is responsible for transport across the NPCs through a ‘piggy-back’ mechanism. Such proteins include the nucleocytoplasmic shuttling Eps15 and HIP1 proteins, as well as endogenous ITSN1-s, ITSN1-L, and ITSN2-s [16,17,55–57]. Despite the confounding effect due to the presence of the CC-region, we were able to clearly demonstrate the existence of a functional bipartite cNLS in the SH3 region of ITSN1-s (ITSN1-s-NLS: RKKNPGGWWEGELQARGKKRQIGW-1127). ITSN1-s-NLS is capable of conferring ATP-dependent nuclear targeting to YFP (Figure7), and mutation of its basic residues impairs nuclear targeting of several YFP-ITSN1-s-SH3 deletion mutants in the presence of LMB (Figure 8). However, mutation of ITSN1-s-NLS in the context of the full-length protein does not abolish its ability to enter the nucleus, most likely due to the fact that the CC region can mediate nuclear import and that deletion of the SH3 region of ITSN1-s similarly does not abolish the protein's nucleocytoplasmic shuttling properties (see Figure 6). The bipartite nature of the ITSN1-s-NLS was confirmed by CLSM quantitative analysis, showing that deletion of upstream basic cluster alone only partially affected nuclear accumulation when compared with mutation of both clusters, in the context of the YFP-ITSN1-s-SH3(D–E) fusion protein (not shown), implying that both basic stretches of aa are required for optimal NLS activity .
Similar to ITSN1-s, several endocytic adaptor proteins undergo nucleocytoplasmic trafficking, mainly to perform additional, specialized tasks within the nucleus, thus being dubbed ‘moonlighting’ proteins [55,59–61]. Many moonlighting proteins, such as Paxilin and EHD2, migrate to the nucleus to regulate cellular proliferation and transcription processes [62,63]. Our data, showing that ITSN1-s concentrates on the NE, where it interacts with lamin A/C (Figure 2A–C) and does not bind to cellular DNA (Figure 2D), rather suggest the possibility that ITSN1-s might play a role at the NE. It has been recently shown that a fourth endosomal route, besides recycling endosomes, endolysosomes, or Golgi apparatus, transports cell surface receptors to the nucleoplasm through docking and membrane fusion of a population of endosomes with the NE . Such nuclear envelope-associated endosomes route may be an alternative mechanism by which external stimuli can influence cellular activity independently of the conventional signaling cascades that operate in the cytosol, and ITSN1-s could therefore be a new player in such route, helping to transport molecules from the plasma membrane to the nucleoplasm, as it has been hypothesized for epidermal growth factor receptor. It is very likely that this process comprises a series of tightly regulated events, and hence, experiments to unravel the mechanisms regulating ITSN1-s shuttling to the nucleus are undergoing in our laboratories. Furthermore, we do not exclude that the process of nuclear localization might be dependent on cell cycle phase, as it happens for other proteins [6,65–67].
Since modifications of the endocytic process have been recently linked to malignancy [68,69], it is likely that rerouting of endocytic proteins to other pathways or compartments due to moonlighting functions could be functionally linked to tumorigenesis. In this context, a previous study conducted very recently showed that APPL1 and APPL2, Rab5 effector proteins, and multifunctional adaptors containing different domains, implicated in several signaling pathways, and recently discovered as nucleocytoplasmic shuttling proteins , are required for the nuclear translocation of type I serine/threonine kinase receptors intracellular domain (TβRI-ICD), thereby promoting progression of prostate cancer cells . Our results might have important implications for the process of carcinogenesis. ITSN1-s is highly expressed in pancreatic, lung, liposarcomas, and Wilm's tumors, as shown in the ONCOMINE database. Furthermore, ITSN1-s is necessary for malignant glioma cell proliferation and for in vitro and in vivo tumorigenic properties of primary human neuroblastoma tumors [71,72]. It is evident that ITSN1-s plays a critical role in this process due to its tertiary structure, allowing its domains to make contacts with many specific targets. Furthermore, its role in tumorigenesis has been linked to signaling regulation rather than endocytosis, although the signaling pathways involved have only been started to be unveiled.
In conclusion, our results suggest a new scenario that foresees the nucleocytoplasmic shuttling of ITSN1-s as an important clue for understanding the physiological and disease-related role of this scaffold protein.
leucine zipper motif 1
confocal laser scanning microscopy
chromosome region maintenance 1
Dulbecco's modified Eagle's medium
fetal bovine serum
- Fn and Fc
nuclear and cytoplasmic fluorescence
nuclear export signal
nuclear localization signal
nuclear pore complex
nuclear targeting signals
A.R. and D.R. conceived the project. A.R. and G.A. designed the experiments. A.R., G.A., L.P., N.M., A.C., V.D.A., C.Z., and M.T. performed the experiments. A.R., G.A., and L.P. wrote the paper. All authors analyzed the data and proofread the paper prior to submission.
This work was supported by the University of Brescia Research fund (ex 60%) to A.R., D.R., and L.C., and by the University of Padua [grant 60A07-1024/15] to G.A.
We thank Yavuz Mercan (University of Padua) for skilled technical assistance, Peter McPherson (McGill University, Montreal, Quebec, Canada) for sharing the pRK1S plasmid and Stephan Höning (University of Cologne, Germany) for sharing the rabbit α-GFP/GST antibody.
The Authors declare that there are no competing interests associated with the manuscript.