Mechanism of Microtubule-facilitated “Fast Track” Nuclear Import

Abstract

Although the microtubule (MT) cytoskeleton has been shown to facilitate nuclear import of specific cancer-regulatory proteins including p53, retinoblastoma protein, and parathyroid hormone-related protein (PTHrP), the MT association sequences (MTASs) responsible and the nature of the interplay between MT-dependent and conventional importin (IMP)-dependent nuclear translocation are unknown. Here we used site-directed mutagenesis, live cell imaging, and direct IMP and MT binding assays to map the MTAS of PTHrP for the first time, finding that it is within a short modular region (residues 82–108) that overlaps with the IMPβ1-recognized nuclear localization signal (residues 66–108) of PTHrP. Importantly, fluorescence recovery after photobleaching experiments indicated that disruption of the MT network or mutation of the MTAS of PTHrP decreases the rate of nuclear import by 2-fold. Moreover, MTAS functions depend on mutual exclusivity of binding of PTHrP to MTs and IMPβ1 such that, following MT-dependent trafficking toward the nucleus, perinuclear PTHrP can be displaced from MTs by IMPβ1 prior to import into the nucleus. This is the first molecular definition of an MTAS that facilitates protein nuclear import as well as the first delineation of the mechanism whereby cargo is transferred directly from the cytoskeleton to the cellular nuclear import apparatus. The results have broad significance with respect to fundamental processes regulating cell physiology/transformation.

Introduction

Nuclear localization of many cancer-regulatory proteins is integral to their function (1–6) in the same way that efficient subcellular trafficking of viruses/viral particles is crucial for pathogenicity (7, 8). Nuclear protein import is generally accepted to be independent of the cytoskeleton (9), conventionally being mediated by soluble transport factors (importins (IMPs)),³ which promote the passage of cargoes bearing nuclear localization signals (NLSs) through the nuclear envelope-localized nuclear pore complex. It is becoming increasingly apparent, however, that specific cancer-related proteins, such as the tumor suppressors p53 (4) and retinoblastoma protein (Rb) (9), and the paracrine/autocrine signaling molecule parathyroid hormone-related protein (PTHrP) (10), rely on the MT network for efficient nuclear import as do certain transcription factors (e.g. STATs 1 and 5) (11, 12). PTHrP is a crucial factor in the early stages of development of various tissues as well as in the biology of adult tissues (13, 14). Importantly, it is present as a circulating factor in numerous types of tumors (15) with its malignant actions linked to its nuclear localization (2, 3), which significantly is known to be dependent on the IMP family member IMPβ1, as well as MT integrity (10).

Specific nuclear localizing proteins, such as PTHrP, appear to possess the ability to interact with MTs for facilitated transport through the cytoplasm prior to conventional IMP-dependent import through the nuclear pore complex into the nucleus (9, 16, 17). However, neither the sequences responsible for mediating MT association and facilitation of nuclear import (microtubule association sequences (MTASs)) nor the mechanisms underlying cargo uncoupling from the MTs for IMP interaction and subsequent nuclear import have been defined. Here we used site-directed mutagenesis, live cell imaging approaches/fluorescence recovery after photobleaching (FRAP) analysis, and novel in vitro binding assays to define for the first time the minimal sequence within PTHrP that is functional both in terms of MT association and MT-dependent enhancement of nuclear import. This represents the first MTAS described for any protein.

Significantly, our results provide the first insight into the mechanism by which MTAS-containing proteins can interact with MTs to enhance delivery to the perinuclear region of the cell prior to displacement from the MTs for subsequent nuclear transport. The findings have clear significance to basic processes in cell biology with relevance both to cellular physiology and transformation.

EXPERIMENTAL PROCEDURES

Generation of Expression Constructs for GFP-PTHrP Fusion Cassettes and Site-directed Mutagenesis

PTHrP coding sequences were inserted into the C terminus of enhanced green fluorescent protein (EGFP). PTHrP full length (amino acids 1–141) and its truncated derivatives, 1–108, 38–108, and 38–141, were inserted into the episomally replicating Gateway^TM-compatible pEPI-DESTC plasmid (18) using the Gateway recombination system (Invitrogen). PTHrP full length and the truncated derivatives 38–94, 66–108, 82–108, 66–94, and 94–108 were inserted into pEGFP-C1 (Clontech). Point mutants in the PTHrP-(66–108) sequence, BASIC-1mut (KKKK⁹¹ → TNTN⁹¹), LINKERmut (KEQE¹⁰¹ → AAAA ¹⁰¹), and BASIC-2mut (KKRR¹⁰⁶ → AAAA¹⁰⁶), were generated using the QuikChange site-directed mutagenesis kit (Stratagene).

PTHrP residues 66–108 and mutant derivatives thereof, 82–108 and 66–94, were also inserted into plasmid pMH830 (19) between the coding sequences of GFP and β-galactosidase (β-gal) and at the C terminus of His₆-tagged GFP in the Gateway-compatible plasmid pGFPRfB (20) for bacterial expression and protein purification.

Cell Culture, Transfection, and Drug Treatments

COS-7 and U2 OS cells were maintained in Dulbecco's modified Eagle's medium (DMEM) and McCoy's medium, respectively, supplemented with 10% FCS in a 5% CO₂ atmosphere at 37 °C. COS-7 cells at 70–80% confluence were transfected using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. In some experiments, cells were grown in the absence or presence of 5 μg/ml nocodazole (NCZ) (Sigma), 1 μg/ml Taxol (Sigma) for 4 h, or 0.5 μg/ml cytochalasin D (CD) (Sigma) for 3 h prior to analysis.

Confocal Laser Scanning Microscopy (CLSM) and Image Analysis

Cells were routinely imaged 20–24 h post-transfection by CLSM using a Bio-Rad MRC-600 microscope. Live cell imaging was performed using a 40× water immersion objective on a 37 °C heated stage, whereas fixed cells were imaged using a 60× oil immersion objective. Image analysis was carried out on the digitalized confocal files using the ImageJ 1.38 public domain software (National Institutes of Health) (9, 21–23). An area of the nucleolus or nucleus (not including the nucleolus) and cytoplasm was selected to determine the relative nucleolar (Fnu), nuclear (Fn), and cytoplasmic (Fc) concentrations of GFP fusion proteins after correction for background fluorescence, enabling the nucleolar (Fnu/c) or nuclear (Fn/c) to cytoplasmic fluorescence ratios to be calculated (10, 24, 25). Results were expressed for a minimum of 40 cells, and experiments were repeated at least twice. For MT or actin staining, cells were fixed and stained with anti-β-tubulin (1:500 dilution; Cytoskeleton, Inc.) followed by Alexa Fluor^TM 568-coupled secondary antibodies (1:1000 dilution; Invitrogen) or with Alexa Fluor 594-labeled phalloidin (1:850 dilution; Invitrogen), respectively.

To investigate the subcellular localization of endogenous proteins, cells were fixed and immunostained as described previously (9) using specific anti-PTHrP (1:500 dilution; GenWay) or anti-SV40 large tumor antigen (T-ag) (1:1000 dilution; Santa Cruz Biotechnology) primary and Alexa Fluor 488-coupled secondary antibodies. Quantitative analysis was performed as above with background values obtained from cells stained only with secondary antibody (26). For statistical analysis, raw data were analyzed using the GraphPad InStat 3 software, and a two-tailed unpaired t test was applied for samples with similar standard deviation (S.D.), or a Mann-Whitney test was applied for samples with significantly different S.D. values where p values <0.05 were considered significant.

FRAP

FRAP was performed essentially as described previously (9). Briefly, COS-7 cells transiently expressing GFP-PTHrP fusion proteins were imaged using an Olympus Fluoview 1000 microscope equipped with an argon ion laser (40 milliwatts) and 100× oil immersion lens (Nikon) in combination with a heated stage. Prior to bleaching, three images were collected using 3% of total laser power with excitation at 488 nm, scanning an area of 126 μm² at a rate of 8 μs/pixel. Nuclear bleaching was performed in an area covering almost the whole nucleus (∼12 μm²), scanning the area 10 times at a rate of 12.5 μs/pixel and applying 80% of the laser power. After bleaching, the cells were immediately scanned, and fluorescence recovery was monitored by acquiring images at 20-s intervals over a period of 9 min using detector and laser settings identical to those used prior to photobleaching. Image analysis was performed as above to determine the Fn and Fc values at all time intervals during the photobleaching experiment. Results were then expressed as fractional recovery of Fn/c (Fn/c of respective time points divided by the corresponding prebleach value), and data were fitted exponentially according to the formula y = a(1 − e^−bx) as described previously (9, 16, 22, 27–31) to determine the maximal recovery and half-time (t½). Results for Fc between 20 and 120 s were used to determine the rate of cytoplasmic decay (31).

Protein Expression and Purification

All His₆GFP-PTHrP fusion proteins were expressed in Escherichia coli host strain BL21(DE3)/pREP4. Cultures grown at 28 °C to an A₆₀₀ of 0.6 were induced with 1 mm isopropyl 1-thio-β-d-galactopyranoside for 4 h. Pellets were lysed in native buffer (50 mm NaH₂PO₄, 600 mm NaCl, pH 8.0) containing 10 mm imidazole, 5 mg/ml lysozyme, 1 unit/ml DNase, 1% Triton, and Complete^TM EDTA-free protease inhibitors (Roche Applied Science). Proteins were purified using the nickel-nitrilotriacetic acid system (Qiagen) by incubating nickel-nitrilotriacetic acid beads with cleared lysate for 2 h at 4 °C before elution with native buffer containing 500 mm imidazole. The imidazole was removed by dialysis.

His₆GFP alone (expressed from plasmid pTRCA-EGFP) (32), His₆GFP-T-ag (T-ag residues 110–135 including the NLS and flanking protein kinase CK2 phosphorylation site), and His₆GFP-PTHrP derivatives to be used for MT binding assays were purified under denaturing conditions as described previously (20, 32) and slowly refolded on the column by gradually decreasing the urea concentration. His₆GFP-PTHrP derivatives were eluted in 200 mm NaCl, 50 mm NaH₂PO₄ (pH 8.0) using 200 mm imidazole, dialyzed to remove imidazole, and subjected to affinity chromatography using a HiTrap heparin HP column (GE Healthcare) by FPLC with elution by salt gradient. The protein fractions were transferred to 450 mm KCl, 80 mm HEPES buffer (pH 8.0) and concentrated by microfiltration.

IMPs were purified as GST fusion proteins as described (21, 33). For ALPHAScreen assays, IMP proteins were biotinylated using the sulfo-NHS-biotin reagent (Pierce). Briefly, 3.5 mg of each GST-IMP protein was incubated with 250 μl of sulfo-NHS-biotin (1 mg dissolved in 150 μl of H₂O) on ice for 2 h, and free biotin was removed via a PD-10 column (Amersham Biosciences).

ALPHAScreen Assay

To determine the binding affinities of different truncated PTHrP proteins and mutants thereof with IMPs, an ALPHAScreen binding assay was performed as previously described (34). Briefly, His₆GFP-PTHrP fusion proteins (30 nm) were bound to Ni²⁺-chelated acceptor beads and incubated with increasing concentrations of biotinylated GST-tagged IMPs or GST alone bound to streptavidin-coated donor beads. Binding interactions were then detected using a FUSIONα (PerkinElmer Life Sciences) plate reader, and the curves were fitted sigmoidally using SigmaPlot software to calculate the maximal binding values (B_max) and the dissociation constants (K_d).

In Vitro MT Polymerization and Association Assay

MTs were polymerized using the MT-binding protein spin-down assay kit (BK029, Cytoskeleton, Inc.) according to the manufacturer's instructions. Briefly, 10 μl of tubulin at 5 mg/ml was incubated with 100 μm GTP and 1.25 μl of cushion buffer (80 mm PIPES (pH 7), 1 mm MgCl₂, 1 mm EGTA, 50% glycerol) for 20 min at 35 °C to generate polymerized MTs. MTs were stabilized by incubation for 30 min at RT in 40 μl of prewarmed tubulin general buffer (80 mm PIPES (pH 7), 1 mm MgCl₂, 1 mm EGTA) containing 50 μm Taxol. 15 μm PTHrP and control proteins were incubated with 10 μl of polymerized MTs for 1 h before the mixture of MTs and GFP fusion proteins was visualized on a slide using a Leica AF6000 LX microscope with 40× objective. In some experiments, PTHrP was preincubated with 3 μm IMPs at room temperature for 30 min prior to MT incubation, or IMPs were added 10 min prior to imaging following 1-h preincubation of PTHrP with MTs. In some cases, MTs were visualized by CLSM using a Bio-Rad MRC-600 microscope with a 60× oil immersion lens, and image analysis on the digitalized confocal files was performed using ImageJ 1.38 where a line plot was used to quantify the specific MT-associated fluorescence with background fluorescence subtracted (35).

MT Association in Vivo Assay

MT and associated proteins were extracted from COS-7 cells expressing GFP-PTHrP fusion proteins using the MT/Tubulin in Vivo Assay kit (BK038, Cytoskeleton, Inc.) as described (36). Briefly, COS-7 cells transfected to express recombinant GFP-PTHrP constructs, GFP alone, and/or human myc-IMPβ1 were lysed using MT stabilizing buffer containing 100 μm GTP, 1 mm ATP, and 10 μl/ml protease inhibitor mixture by incubation for 10 min at 37 °C. MTs and associated proteins (pellet (P)) were harvested by ultracentrifugation at 55,000 × g for 30 min at 37 °C and separated from the supernatant (S) (containing soluble tubulin and non-MT-associated proteins). After the pellet was washed three times with MT stabilizing buffer before resuspension. The P, wash, and S fractions were analyzed by Western blotting using anti-GFP (1:1000; Roche Applied Science), anti-β-tubulin (1:500 dilution; Cytoskeleton, Inc.), anti-IMPβ1 (1/1000; Cell Signaling Technology), or anti-myc (1/1000; Santa Cruz Biotechnology) antibody. The blots were developed using chemiluminescence (PerkinElmer Life Sciences), and densitometric analysis was performed using ImageQuant TL software (GE Healthcare). The signal intensities for GFP fusion proteins in the P and S fractions were normalized to the internal control protein, tubulin. For single transfections, the normalized values were expressed as percentage of the intensity for the S fraction. In experiments where cells were cotransfected to express GFP-PTHrP-(66–108) (or BASIC-1mut) and IMPβ1, the ratio of MT-associated to non-associated PTHrP protein (P/S) was calculated, and to quantify the effect of IMPβ1 coexpression, results were shown as a percentage of the P/S ratio from cells expressing GFP-PTHrP-(66–108) alone.

RESULTS

PTHrP Amino Acids 82–108 Are Sufficient to Confer MT-enhanced Nuclear Accumulation

Although previous studies implicate an MT-facilitated nuclear import pathway for PTHrP (9, 10) as well as for other specific proteins (9), neither the mechanistic basis of MT enhancement nor the sequence (MTAS) responsible has been defined in any case. We first set out to map the MTAS of PTHrP by generating a number of constructs encoding a series of GFP fusion protein derivatives of PTHrP (Fig. 1A). The nuclear accumulation of the derivatives was examined by live cell imaging of transfected cells without or with pretreatment with NCZ (Fig. 1, B and C). Cells treated in parallel were also fixed and stained for tubulin, confirming that NCZ treatment caused large scale disassembly of the MT cytoskeleton (Fig. 1B, lower panels).

FIGURE 1.

Efficient nuclear accumulation of PTHrP-(82–108) is dependent on intact MTs. A, schematic representation of the various GFP-PTHrP truncated derivatives used in this study with parathyroid hormone (PTH)-like, NLS, and nuclear export sequence (NES) domains indicated. B, CLSM images of live COS-7 cells transfected to express the indicated GFP fusion proteins and treated without or with NCZ as indicated. Left lower panels show fixed cells treated without or with NCZ and immunostained for tubulin. For GFP-PTHrP derivatives 1–141, 38–141, 1–108, 38–108, 66–108, and 82–108 that are predominantly nuclear/nucleolar in the absence of NCZ treatment, the cell cytoplasm is outlined with yellow dotted lines. Scale bars represent 20 μm. C, results for quantitative analysis of digitized images, such as those in B, to determine the Fnu, Fn, or Fc corrected for background or the extent of nucleolar and nuclear accumulation expressed as Fnu/c or Fn/c, respectively. Results represent the mean (n > 40), and error bars represent the standard error of the mean. Significant differences are denoted by the p values. †, unpaired t test; *, Mann-Whitney test.

CLSM imaging revealed an increase in the cytoplasmic localization of GFP-PTHrP full length (amino acids 1–141) and the truncated derivatives 38–141, 1–108, 38–108, 66–108, and 82–108 in cells lacking intact MTs (NCZ-treated cells) compared with non-treated cells containing intact MTs (Fig. 1B). No change was observed for GFP-PTHrP-(38–94), GFP-PTHrP-(66–94), GFP-PTHrP-(94–108), or GFP itself. Quantitative analysis confirmed this observation, indicating a significant (p < 0.003) increase in Fc of the former group of proteins following NCZ treatment (up to almost 9-fold in some cases; Fig. 1C, middle histogram). This was accompanied by decreases in the Fn (Fig. 1C, left, lower histogram), although the Fn was still higher than the Fc due to the presence/activity of the conventional NLS.

The Fn/c and Fnu/c were calculated as described previously (see “Experimental Procedures” and Roth et al. (9)), revealing that GFP-PTHrP-(1–141), GFP-PTHrP-(38–141), GFP-PTHrP-(1–108), GFP-PTHrP-(38–108), GFP-PTHrP-(66–108), and GFP-PTHrP-(82–108) all showed significantly (p < 0.0001) reduced Fn/c and Fnu/c values in the absence compared with the presence of functional MTs (Fig. 1C, right histograms). In contrast, NCZ treatment had no effect on the extent of nuclear/nucleolar accumulation of GFP-PTHrP-(38–94), GFP-PTHrP-(66–94), GFP-PTHrP-(94–108), or GFP itself (Fig. 1C), implying a lack of MT-enhanced nuclear import and suggesting that the MTAS is within the region 82–108. Significantly, none of the latter accumulated in the nucleus/nucleolus to any great extent (Fn/c and Fnu/c values of 2–3 and 3–5, respectively) even in the absence of NCZ. This was especially evident when compared with the data for the former group of proteins (e.g. GFP-PTHrP-(66–108) has Fn/c and Fnu/c values of ∼50 and ∼140, respectively; see Fig. 1C, right), implying that MT-enhanced nuclear import contributes strongly to the overall nuclear accumulation efficiency of PTHrP.

The fact that GFP-PTHrP-(38–94) and GFP-PTHrP-(66–94) contain the previously identified IMPβ1-recognized NLS of PTHrP (24) but accumulate to a much lower extent than GFP-PTHrP-(66–108) implies that the NLS may require additional residues (94–108) for optimal nuclear import efficiency in vivo. The additional sequence required also appears to confer NCZ sensitivity to nuclear/nucleolar accumulation. Thus, the sequence appears to be part of the putative MTAS (amino acids 82–108) of PTHrP, the minimal region able to confer MT-enhanced nuclear import by working in combination with the NLS to achieve maximal nuclear accumulation (see also below).

Nuclear Accumulation of PTHrP-(82–108) Requires MT Dynamics

MTs are intrinsically dynamic filaments essential for many cellular events. Drugs that affect MT dynamic behavior and functionality, such as Taxol, an MT-stabilizing agent, have been shown to impair efficient nuclear import of several nuclear proteins, such as p53 (4) and Rb (9). The effect of NCZ (above) is to disassemble the MT network. Thus, to test the role of MT dynamics more directly, we examined the effect of Taxol on the nuclear accumulation of the GFP-PTHrP constructs. As a control, we also tested the effect of the actin-depolymerizing agent CD. The efficacy of these drug treatments in affecting MTs and actin was confirmed by treating cells in parallel and fixing/staining cells using an anti-β-tubulin antibody or phalloidin, respectively (see Fig. 2A, lower panels).

FIGURE 2.

Efficient nuclear accumulation of PTHrP-(82–108) is dependent on MT dynamics but not on actin filaments. A, CLSM images of live COS-7 cells transfected to express the indicated GFP fusion proteins and treated without (No add.) or with Taxol or CD as indicated. Lower panels show fixed cells treated without or with Taxol or CD and immunostained for tubulin (left) or actin filaments (right). Scale bars represent 20 μm. B, results for quantitative analysis of the levels of nucleolar or nuclear accumulation (see legend to Fig. 1) in COS-7 cells expressing the indicated GFP fusion proteins without or with Taxol or CD pretreatment. Results represent the mean (n > 40), and error bars represent the standard error of the mean. Significant differences are denoted by the p values. *, Mann-Whitney test.

Importantly, we found that the effect of Taxol treatment on nuclear/nucleolar accumulation of the various GFP-PTHrP derivatives closely resembled that of NCZ, resulting in an increase in cytoplasmic localization of GFP-PTHrP-(1–141), GFP-PTHrP-(38–141), GFP-PTHrP-(1–108), GFP-PTHrP-(38–108), GFP-PTHrP-(66–108), and GFP-PTHrP-(82–108) fusions when compared with that observed in non-treated cells (Fig. 2A). The localization of GFP-PTHrP-(66–94) or GFP alone was unaffected.

Quantitative analysis supported this observation with significantly (p < 0.0002) reduced nuclear/nucleolar accumulation evident for all constructs following Taxol treatment except for GFP-PTHrP-(66–94) and GFP alone (Fig. 2B). These results clearly imply that efficient nuclear translocation conferred by PTHrP residues 82–108 requires a functional, dynamic MT network. CD treatment, in contrast, had no effect on nuclear localization of any of the proteins examined (Fig. 2, A and B), underlining the fact that efficient nuclear accumulation of PTHrP depends specifically on MTs and not on actin filaments.

Two Clusters of Basic Amino Acids Are Critical for MT-dependent PTHrP Nuclear Import

To determine the specific residues within PTHrP amino acids 66–108 responsible for MT-dependent nuclear import, we performed site-directed mutagenesis of several specific sequences within the NLS/MTAS-containing region (see Fig. 3A). PTHrP-(66–108) contains two main basic regions (KKKK⁹¹) and (KKRR¹⁰⁶), which are separated by a linker of 10 residues (GKPGKRKEQE¹⁰¹), an arrangement that is characteristic of a bipartite NLS. Within the linker region, there is also a sequence (KEQE¹⁰¹) that has some resemblance to the association motif ((K/R)XTQT) for the dynein MT motor protein light chain 8 (DLC8) (37–39). To test the potential role of these three regions in PTHrP nuclear import, we generated derivatives of GFP-PTHrP-(66–108) containing mutations in each of the basic amino acid clusters, KKKK⁹¹ and KKRR¹⁰⁶ (denoted BASIC-1mut and BASIC-2mut, respectively), as well as in the linker region, KEQE¹⁰¹ (denoted LINKERmut) (see “Experimental Procedures” for details).

FIGURE 3.

Efficient MT-dependent nuclear accumulation of PTHrP requires KKKK⁹¹ and KKRR¹⁰⁶. A, schematic representation of the GFP-PTHrP-(66–108) region showing the previously identified NLS (amino acids 66–94). The two basic regions and the linker regions that were mutated in this study are also indicated with residues selected for mutation in bold. B, schematic representation of PTHrP showing the modular NLS (residues 66–108) and MTAS (82–108) sequences identified in this study. C, CLSM images of live COS-7 cells transfected to express the indicated GFP-PTHrP-(66–108) derivatives and treated without or with NCZ as indicated. Scale bar represent 20 μm. D, results for quantitative analysis for the levels of nucleolar or nuclear accumulation (according to Fig. 2) in COS-7 cells expressing the indicated GFP-PTHrP-(66–108) derivatives treated without or with NCZ. Results represent the mean (n > 60), and error bars represent the standard error of the mean. Significant differences are denoted by the p values. *, Mann-Whitney test.

COS-7 cells were transfected to express the various mutant derivatives as well as wild type (WT) GFP-PTHrP-(66–108) prior to treatment without or with NCZ and imaging 20–24 h post-transfection. In the absence of NCZ treatment, the BASIC-1mut and BASIC-2mut derivatives both showed increased cytoplasmic localization compared with WT, which was exclusively nuclear/nucleolar (Fig. 3C). Quantitative analysis confirmed significantly (p < 0.0001) reduced nuclear/nucleolar accumulation for the BASIC-1mut and BASIC-2mut derivatives (Fn/c and Fnu/c values of 2–3 and 4–6, respectively) (Fig. 3D) compared with WT (Fn/c and Fnu/c values of 50 and 135, respectively).

That MT-dependent transport was directly affected by the basic cluster mutations was indicated by the fact that NCZ treatment did not further reduce nucleolar or nuclear accumulation (Fig. 3, C and D) in contrast to the case of WT where NCZ treatment caused a significant reduction of both. Thus, it appears that both KKKK⁹¹ and KKRR¹⁰⁶ are critical for MT-enhanced nuclear import of PTHrP.

In contrast to the results for the basic cluster mutants, the LINKERmut derivative showed levels of nucleolar/nuclear accumulation more similar to that of WT with values of Fnu/c and Fn/c of 60% that of the WT. This reduction, however, did not appear to reflect an essential role for this sequence in efficient MT-facilitated import as the nuclear accumulation of the LINKERmut derivative remained sensitive to NCZ. In fact, NCZ treatment significantly reduced its nuclear/nucleolar accumulation, bringing it to the same level as that observed following NCZ treatment of WT (Fig. 3, C and D).

PTHrP-(66–108) Can Confer Efficient MT-facilitated Nuclear Targeting of Large Cargo Protein Dependent on KKKK⁹¹ and KKRR¹⁰⁶

Molecules smaller than ∼45 kDa can enter the nucleus through the nuclear pore by passive diffusion (40). With a molecular mass of 33 kDa, GFP-PTHrP-(66–108) is below this limit. To exclude the possibility that the various GFP-PTHrP fusion proteins may enter the nucleus by passive diffusion and accumulate in the nucleus by binding to nuclear components (24), the sequences encoding PTHrP amino acids 66–108 (WT and mutant derivatives) and 82–108, 66–94, and 94–108 were inserted between the sequences encoding GFP and β-gal in the plasmid vector pMH830 (19). The resulting tetrameric GFP-β-gal fusion proteins (∼580 kDa) absolutely require active nuclear transport for nuclear entry/accumulation (19).

Strikingly, fusion proteins containing PTHrP-(66–94) or PTHrP-(94–108) were completely cytoplasmic in transfected cells, comparable in all respects to GFP-β-gal alone (Fn/c of ∼0.3; Fig. 4, A and B), implying that neither PTHrP region 66–94 nor PTHrP region 94–108 are sufficient to target a large cargo protein to the nucleus. In contrast, PTHrP-(66–108) conferred strong nuclear/nucleolar accumulation (Fn/c and Fnu/c values of 16 and 23, respectively; Fig. 4A). PTHrP-(82–108) also showed significant nuclear/nucleolar accumulation of the β-gal fusion, although this was ∼4–5-fold lower than that of PTHrP-(66–108) (Fn/c and Fnu/c values of 3 and 6, respectively). The results clearly imply that PTHrP residues 66–108 represent the NLS conferring optimal PTHrP nuclear accumulation, consistent with previous data showing that the residues 66–82 contribute to efficient nuclear import conferred by PTHrP-(82–108) (41).

FIGURE 4.

Dependent on KKKK⁹¹ and KKRR¹⁰⁶, PTHrP-(66–108) can direct efficient MT-dependent nuclear import of large cargo protein. A, CLSM images of live COS-7 cells transfected to express the indicated GFP-PTHrP-β-gal fusion proteins and treated without (No add.) or with NCZ or Taxol as indicated. The scale bar represents 20 μm. B, results for quantitative analysis for the levels of nucleolar or nuclear accumulation (according to Fig. 2) in COS-7 cells expressing the indicated GFP-PTHrP-β-gal fusion proteins without or with NCZ or Taxol pretreatment. Results represent the mean (n > 60), and error bars represent the standard error of the mean. Significant differences are denoted by the p values. *, Mann-Whitney test.

Mutation of either of the basic clusters severely reduced nuclear import of GFP-PTHrP-(66–108)-β-gal, resulting in an absence of nucleolar accumulation and 2-fold lower nuclear accumulation (Fn/c of 8) in the case of BASIC-1mut or total abolition of nuclear accumulation (Fn/c of 0.2) in the case of BASIC-2mut (Fig. 4, A and B). This indicates that both basic clusters contribute to active nuclear targeting of PTHrP but that KKRR¹⁰⁶ is absolutely essential. Again, in contrast to the effect of the basic cluster mutations, mutation of KEQE¹⁰¹ (LINKERmut) had a negligible effect on nucleolar/nuclear localization of GFP-PTHrP-(66–108)-β-gal (Fig. 4, A and B), consistent with the idea that this sequence is not essential for nuclear/nucleolar targeting (see above) (Fig. 3).

Importantly, treatment with either NCZ or Taxol significantly (p ≤ 0.05) reduced nuclear/nucleolar accumulation of GFP-β-gal fusions carrying PTHrP residues 66–108 and 82–108 as well as LINKERmut, resulting in a >50% decrease (Fig. 4, A and B). This implies dependence of all of these proteins on MT functionality/dynamics for optimal nuclear accumulation. In contrast, NCZ or Taxol treatment had no effect on nuclear accumulation of GFP-PTHrP-(66–108)-β-gal-BASIC-1mut, confirming that MT-dependent nuclear import of PTHrP requires KKKK⁹¹.

MT Integrity Is a Critical Determinant of Rate of PTHrP Nuclear Import

To confirm that the NLS/MTAS of PTHrP enhances its nuclear import rate in living cells, we performed FRAP analysis as described previously (9, 16). For this approach, the fluorescent signal in the nucleus is photobleached, enabling monitoring of the nuclear import rate by quantifying the return of fluorescence to the nucleus, which results from translocation of non-bleached, cytoplasmic fluorescent protein into the nucleus (22, 31, 42, 43).

COS-7 cells were transfected to express GFP-PTHrP-(66–108)-β-gal or the BASIC-1mut derivative and treated without or with NCZ before analysis of nuclear import using FRAP (Fig. 5). Fig. 5C shows typical traces of the increase in Fn subsequent to bleaching together with the corresponding decline in Fc. Non-treated WT GFP-PTHrP-(66–108)-β-gal showed a significant increase in Fn in parallel with a rapid decline in Fc. Upon treatment with NCZ, however, the increase of Fn was reduced concomitantly with slowed Fc decay, consistent with the idea that efficient PTHrP nuclear import requires MT integrity. In contrast, even in the absence of NCZ, the BASIC-1mut derivative showed a less robust postbleach increase in Fn and a slower decay in Fc, whereas NCZ treatment had little effect (Fig. 5C). Thus, MT-dependent transport appears to be impaired in the mutant protein.

FIGURE 5.

MT-facilitated nuclear import of PTHrP in vivo is dependent on residues KKKK⁹¹. CLSM visualization of the return of nuclear fluorescence after photobleaching (see “Experimental Procedures”) in COS-7 cells expressing GFP-PTHrP-(66–108)-β-gal-WT (A) and GFP-PTHrP-(66–108)-β-gal-BASIC-1mut (B) treated without (No add.) or with NCZ (+NCZ). Scale bars represent 20 μm. C, changes in Fn and Fc after subtraction of background fluorescence plotted over time postnuclear bleach. D, quantification of the recovery over time of nuclear fluorescence after photobleaching expressed in terms of fractional recovery of Fn/c (Fn/c of respective time points divided by the prebleach value). The regression values for the exponential fits were 0.83 and 0.88 for WT with no addition and +NCZ, respectively, and 0.86 and 0.92 for BASIC-1mut with no addition and +NCZ, respectively. Pooled data represents mean, n ≥ 10, and error bars represent standard error of the mean for the initial rate of cytoplasmic decay (Fc s⁻¹) (E) or the half-time of return of fluorescence (t½) (F) for GFP-PTHrP-(66–108)-β-gal-WT and GFP-PTHrP-(66–108)-β-gal-BASIC-1mut for non-treated and NCZ-treated cells are shown. Significant differences between cells treated without and with NCZ are denoted by the p values. †, unpaired t test; *, Mann-Whitney test. NS, not significant.

Fig. 5D shows the results in terms of the fractional recovery of Fn/c (see “Experimental Procedures” for details). After photobleaching, GFP-PTHrP-(66–108)-β-gal accumulated rapidly in the nucleus of non-treated cells with a half-time for fractional recovery (t½) of 51 s. Importantly, BASIC-1mut derivative showed a significantly (p < 0.0001) slower rate of recovery (t½ of 93 s; see also Fig. 5F), confirming that residues KKKK⁹¹ are necessary for rapid and efficient nuclear import of PTHrP. Results for the initial rate of decay of Fc through transport to the nucleus (Fig. 5E) confirmed this, indicating that the BASIC-1mut derivative had a ∼2-fold slower rate of nuclear import compared with WT (Fc s⁻¹ of ∼0.7 compared with 1.5).

Cells treated with NCZ showed a significantly reduced rate of nuclear import of GFP-PTHrP-(66–108)-β-gal (t½ of 118; Fig. 5F) with the initial rate of decay of Fc similarly reduced (Fig. 5E; Fc s⁻¹ of 0.60), consistent with the idea that efficient/rapid nuclear import of PTHrP-(66–108) is dependent on MT integrity. In contrast, NCZ treatment had essentially no effect on the nuclear import rates of GFP-PTHrP-(66–108)-β-gal-BASIC-1mut (t½ values of 91 and 93s in NCZ-treated and non-treated cells, respectively), and the initial rates of cytoplasmic decay were identical in both conditions (Fc s⁻¹ of ∼0.65). These results are consistent with the idea that MT-dependent transport is essentially eliminated by this mutation. Clearly, WT PTHrP can traffic efficiently to the nucleus dependent on MTs, and mutation of residues KKKK⁹¹ impairs this, indicating that residues KKKK⁹¹ are crucial for the MT-enhanced nuclear trafficking mechanism of PTHrP.

KKKK⁹¹ and KKRR¹⁰⁶ Are Critical for PTHrP Interaction with IMPβ1

The experiments above indicate that both KKKK⁹¹ and KKRR¹⁰⁶ are necessary for efficient nuclear accumulation of PTHrP. To test their contribution to IMPβ1 binding by PTHrP, we performed an ALPHAScreen binding assay (34) using bacterially expressed His₆GFP-PTHrP derivatives and GST-tagged human or mouse IMPβ1 with His₆GFP-PTHrP-(38–108) fusion protein as a positive control (41).

Human IMPβ1 was found to bind to GFP-PTHrP-(66–108) with high affinity (an apparent dissociation constant (K_d) of 10.9 nm) with maximal binding (B_max) of 76,000 arbitrary units). These values were comparable with those for GFP-PTHrP-(38–108) (B_max of 109,000 arbitrary units; K_d of 9 nm; see Fig. 6 and Table 1). GFP-PTHrP-(82–108) showed 50% reduced maximal binding (B_max of 42,900 arbitrary units) with a lower binding affinity (K_d of 13.1 nm) compared with GFP-PTHrP-(66–108), confirming that residues 66–82 are important for interaction with human IMPβ1, although they are not sufficient to bind to IMPβ1 (41, 44).

FIGURE 6.

IMP-β1 binding to PTHrP requires both KKKK⁹¹ and KKRR¹⁰⁶. GFP-PTHrP truncated proteins or GFP alone as indicated were incubated with increasing concentrations of biotinylated GST-human IMPβ1, and an ALPHAScreen assay was performed (see “Experimental Procedures”). Each data point represents the average of triplicate results of a single typical experiment. Sigmoidal curves were fitted using SigmaPlot software to determine the B_max and K_d values; the regression coefficients for the curve fits in all cases were >0.94 with the sole exception of the fits where low binding is in evidence (regression coefficients of 0.86 and 0.82 for GFP-PTHrP-(66–94) and GFP, respectively). ND, not able to be determined.

TABLE 1.

Summary of K_d and B_max values obtained from ALPHAScreen assays using human IMPβ1

B_max, maximum binding; K_d, dissociation constant; ND, not able to be determined; AU, arbitrary units. Values shown are the average of three individual experiments ±S.E. according to Fig. 6.

	Kd	Bmax	Bmax relative to PTHrP-(66–108)
	nm	AU	%
GFP-PTHrP-(38–108)	9.0 ± 0.4	109,200 ± 20,000
GFP-PTHrP-(66–108)	10.9 ± 0.7	76,100 ± 3,000	100
GFP-PTHrP-(66–108)-BASIC-1mut	ND	12,500 ± 2,300	16
GFP-PTHrP-(66–108)-BASIC-2mut	ND	11,800 ± 2,900	15
GFP-PTHrP-(82–108)	13.1 ± 0.9	42,000 ± 9,000	55
GFP-PTHrP-(66–94)	ND	4,300 ± 1,100	6
GFP	ND	1,900 ± 500	2.5

GFP-PTHrP-(66–108)-BASIC-1mut and GFP-PTHrP-(66–108)-BASIC-2mut as well as GFP-PTHrP-(66–94) all showed lower binding (B_max values of 16, 15, and 6%, respectively, of those for GFP-PTHrP-(66–108); see Fig. 6 and Table 1), clearly indicating that residues KKKK⁹¹ and KKRR¹⁰⁶ are critical for IMPβ1 binding. Similar results for the PTHrP derivatives were observed for mouse IMPβ1, whereas only very low binding was observed for the GFP-PTHrP fusion proteins with human IMPα1 or GST alone (not shown), underlining the specificity of the assay. Comparable results were also obtained using native gel electrophoresis/fluoroimaging (not shown) (45). The clear implication of these data is that GFP-PTHrP-(66–108) specifically binds IMPβ1 with high affinity with both KKKK⁹¹ and KKRR¹⁰⁶ required for this interaction.

PTHrP-(82–108) Confers Association with MTs in Vivo and in Vitro Dependent on KKKK⁹¹ and KKRR¹⁰⁶

To test whether PTHrP can interact directly with intact MTs and determine the possible role of IMPβ1 in this interaction, tubulin was polymerized in vitro as described previously (10). The in vitro polymerized MT filaments were then incubated with bacterially expressed GFP-PTHrP fusion proteins (amino acids 38–108, 66–108, 82–108, and 66–94) as well as mutant derivatives thereof and negative controls (GFP and GFP-T-ag NLS). Some assays were performed in the presence of GST-IMPs. The binding of the GFP fusion proteins to the MTs was analyzed by CLSM (see “Experimental Procedures”). Clear association of GFP-PTHrP-(82–108), GFP-PTHrP-(38–108), and GFP-PTHrP-(66–108) but not GFP-PTHrP-(66–94) with the MTs filaments was observed, consistent with the idea that PTHrP-(82–108) (Fig. 7A, left panels) represents the minimal MTAS region of PTHrP (see above). Quantitative analysis to determine the specific MT-associated fluorescence relative to background fluorescence supported these findings (Fig. 7C), indicating significantly reduced (p < 0.0001; ∼10-fold less) binding to MTs by GFP-PTHrP-(66–94) compared with the other truncation derivatives. Mutation of KKKK⁹¹ or KKRR¹⁰⁶ in GFP-PTHrP-(66–108) reduced MT association (Fig. 7, A, right panels, and C) by about 50%, demonstrating the dependence of MT association on these residues and indicating that both basic regions are required for optimal MT association. The clear implication is that the NLS and MTAS are collinear/overlap. GFP and GFP-T-ag (which carries the same net positive charge as GFP-PTHrP-(66–108)) both failed to associate with MTs (Fig. 7A, lower right panels), confirming the specificity of the results. Interestingly, preincubation of PTHrP with GST-IMPβ1 but not GST-IMPα (see Fig. 7A, left panels) or GST alone (not shown) impaired the ability of all MT-associating PTHrP derivative proteins to interact with MTs, consistent with colinear/overlapping NLS and MTAS sequences and with the idea that both basic clusters are involved in IMP and MT binding.

FIGURE 7.

PTHrP-(82–108) confers binding to intact MTs in vitro and in vivo dependent on KKKK⁹¹ and KKRR¹⁰⁶. A and B, visualization of the indicated GFP-fused proteins incubated with in vitro intact assembled MTs as indicated (see “Experimental Procedures”); corresponding transmission images of the MTs are shown below with the exception of the right panels in A where the transmission images are on the right. Where indicated, GST-IMPβ1 (+IMPβ) or GST-IMPα (+IMPα) were preincubated (30 min) with GFP-PTHrP fusion proteins prior to MT binding for 1 h followed by CLSM imaging. Where indicated in B, GST-IMPβ1 was added 10 min prior to CLSM imaging after (“post”) 1-h preincubation of GFP-PTHrP with MTs. C, results for quantitative analysis for the levels of MT fluorescence (MT-associated fluorescence (Fmt) after the subtraction of background fluorescence) due to binding of fluorescent proteins. Results represent the mean (n > 100), and error bars represent standard error of the mean from two separate experiments where significant differences between the GFP-PTHrP proteins are denoted by the p values. D–G, COS-7 cells expressing the indicated proteins were lysed in MT stabilizing buffer and subjected to ultracentrifugation. The MT pellet/associated proteins (P) were separated from supernatant (S) containing soluble proteins and tubulin as described under “Experimental Procedures” and washed (W) three times prior to resuspension. Fractions were then subjected to Western analysis using the indicated antibodies. Densitometric analysis for the protein levels in D and F is presented in E and G where the signal intensity above background for the GFP fusion proteins in S and P are expressed relative to the corresponding values for tubulin, and the values are then expressed as a percentage relative to that derived for the supernatant (100%) (in E) or the ratio of pellet/supernatant relative to the maximum value (for wild type, single expression) (in G). No add., no addition.

Further investigation indicated that the MT association of PTHrP could be disrupted by a short incubation (10 min) of GST-IMPβ1 even after prebinding (1 h) of PTHrP to MTs, indicative of competitive binding (Fig. 7B, right panels). Importantly, biotinylated GST-IMPβ1, detected using FITC-streptavidin, was found not to associate directly with polymerized MT filaments (not shown). This suggested that displacement of PTHrP from MTs is due to IMPβ1 binding to PTHrP and not to MTs at the same site as PTHrP. Overall, the results imply that the role of IMPβ1 in PTHrP trafficking in the cell may be to release PTHrP from MTs in the region of the MT-organizing center near the nuclear envelope where the local IMPβ1 concentration is high (46) prior to translocating soluble PTHrP through the nuclear pore into the nucleus.

To confirm the association of PTHrP with MTs from living cells and the potential role of IMPβ1 in regulating this interaction, intact MTs were extracted from COS-7 cells expressing GFP-PTHrP-(66–108) (and mutant derivatives thereof), GFP-PTHrP-(82–108), GFP-PTHrP-(66–94), and GFP-PTHrP-(94–108) fusions, GFP alone, or myc-tagged human IMPβ1. In some cases, cells were cotransfected to express PTHrP derivatives with myc-IMPβ1. The presence of GFP-PTHrP and/or human IMPβ1 in isolated MT fractions was assessed by Western blotting and quantified by densitometry (Fig. 7, D–G). GFP-PTHrP-(66–108), GFP-PTHrP-(82–108), LINKERmut, and BASIC-2mut were all found to associate with MTs, although BASIC-2mut did so to a much lesser extent (see Fig. 7, D and E), consistent with the results for in vitro MT binding (see above and Fig. 7C). In contrast to GFP-PTHrP-(66–94), GFP-PTHrP-(94–108), BASIC-1mut, and GFP alone showed no association with MTs (Fig. 7, D and E).

Neither endogenous cellular (not shown) nor ectopically expressed human IMPβ1 was found in the MT fraction (Fig. 7, D and E), indicating that IMPβ1 does not associate with MTs in vivo, consistent with the in vitro findings (see above). IMPβ1 was also not detectable in the MT fractions in which PTHrP was detected, consistent with the idea that the binding of PTHrP to MTs is mutually exclusive with its binding to IMPβ1.

To confirm that IMPβ1 could displace PTHrP from intact MTs in living cells, MTs were extracted from cells coexpressing PTHrP and myc-IMPβ1. GFP-PTHrP-(66–108) association with MTs was found to be ∼2-fold lower in cells coexpressing myc-IMPβ1 than in cells expressing GFP-PTHrP-(66–108) alone. The specificity of this effect to PTHrP-(66–108) was indicated by the fact that no such effect of IMPβ1 coexpression on MT association was observed for either BASIC-1mut or GFP alone (see Fig. 7, F and G).

The results clearly imply that the ability of IMPβ1 to inhibit PTHrP binding to MTs (Fig. 7) is through IMPβ1 binding to PTHrP and not through IMPβ1 competing for PTHrP binding sites on MTs. The absence of IMPβ1 in the PTHrP MT fraction indicates that although PTHrP is able to bind both IMPβ1 and MTs it does not associate with IMPβ1 when bound to MTs in intact cells, consistent with the mutual exclusivity of binding observed in the in vitro assays (Fig. 7, A–C). Overall, these results indicate that PTHrP-(82–108) is likely to be the MTAS that mediates interaction of PTHrP with MTs in vivo and in vitro. The fact that PTHrP-(66–94) and PTHrP-(94–108) alone are unable to confer MT binding implies that both KKKK⁹¹ and KKRR¹⁰⁶ are essential constituents not only of the IMPβ1-recognized NLS but also of the MTAS.

Endogenously Expressed PTHrP Relies on MT Functionality for Optimal Nuclear Localization

To confirm that the results for ectopically expressed GFP-PTHrP fusion proteins are relevant to endogenous PTHrP, COS-7 and U2 OS osteosarcoma cell lines, which express PTHrP endogenously, were treated with or without NCZ or Taxol before fixation and staining for PTHrP, and quantitative CLSM analysis of its subcellular localization was performed (Fig. 8). As expected, increased cytoplasmic localization of PTHrP was evident in the cells treated with NCZ or Taxol (Fig. 8A).

FIGURE 8.

Optimal nuclear accumulation of endogenous cellular PTHrP is dependent on functional MTs. A, CLSM images of COS-7 and U2 OS osteosarcoma cells treated without (No add.) or with NCZ or Taxol prior to fixation and immunostaining for PTHrP or T-ag as indicated. Scale bars represent 20 μm. B and C, results for quantitative analysis of images, such as those in A, for the levels of Fn and Fc (expressed relative to non-treated cells) (B) or the extent of nuclear accumulation of PTHrP or endogenous T-ag (Fn/c) (according to Fig. 1) (C). Results represent the mean (n ≥ 110), and error bars represent standard error of the mean, where significant differences are denoted by p values. *, Mann-Whitney test.

Quantitative analysis confirmed a significant (p < 0.05) 30–50% increase in cytoplasmic fluorescence (Fig. 8B, middle) and a corresponding decrease in nuclear fluorescence in COS-7 cells treated with NCZ or Taxol compared with non-treated cells (Fig. 8B, left). In both cell lines, this resulted in a significant (p < 0.0001) 50–60% reduction in the Fn/c value following MT depolymerization by NCZ or MT stabilization by Taxol treatments (Fig. 8, B and C). These treatments produced no detectable effect on the nuclear accumulation of endogenously expressed T-ag in COS-7 cells, strongly implying that the effects observed for PTHrP were not attributable to general effects of the drugs on nuclear transport efficiency (see also Ref. 9). Overall, the results indicate that endogenously expressed PTHrP relies on functional/intact MTs for optimal nuclear accumulation.

DISCUSSION

Here we characterize for the first time the molecular mechanism underlying MT-dependent nuclear import of PTHrP. In so doing, we also define the first modular MTAS able to confer physical and functional MT association in nuclear import. Importantly, we establish the detailed mechanism of its action, i.e. that it functions through the intriguing interplay between overlapping NLS and MTAS sequences to ensure efficient transfer of PTHrP from MTs to IMPβ1 and subsequent transport into the nucleus. MT-dependent nuclear import has been reported previously for a number of proteins (e.g. p53 and Rb) (4, 9, 16, 17), but the sequences responsible as well as the details of the trafficking mechanisms involved have remained unresolved. In particular, an MTAS sufficient to confer MT association and enhancement of nuclear import per se has never previously been identified. The work described here defines such an MTAS for the first time, thereby shedding new light on the mechanisms underlying MT-facilitated nuclear transport.

We show here that optimal nuclear accumulation of PTHrP requires specific residues within 66–108, which includes the overlapping IMPβ1-recognized NLS (66–108) and MTAS (82–108) sequences (see Fig. 3B). The fact that the NLS and MTAS share key residues integral to their function ensures that IMPβ1 binding and MT association are mutually exclusive. This is documented here not only in our in vitro binding studies but also by the subcellular fractionation experiments using transfected cells (Fig. 7, D–G) where overexpression of IMPβ1 resulted in reduced association of PTHrP with MTs.

These results identify a mechanism by which PTHrP can be released from MTs subsequent to MT-dependent delivery to the perinuclear region. Specifically, it appears that PTHrP traffic to the nuclear region is accelerated by its association with MTs until arrival at the MT-organizing center/perinuclear region where the local concentration of IMPβ1 is highest (46). This high concentration enables IMPβ1 to compete with MTs for PTHrP binding, thereby releasing PTHrP from the MT cytoskeleton prior to ferrying it through the nuclear pore into the nucleus. That IMPβ1 is not present in the MT fraction of intact cells (Fig. 7) is consistent with the idea that IMPβ1 acts in addition to/subsequent to MT-dependent transport of PTHrP. Importantly, the results explain for the first time how, through dynamic interactions with MTs, MTAS-dependent MT interaction can facilitate nuclear import rather than resulting in cytoplasmic sequestration and consequently inhibition of nuclear import (47–49).

Efficient MT-dependent nuclear import of p53, Rb, and several other cellular and viral proteins, such as rabies virus P-protein (4, 9, 16), is dependent on dynein/dynactin with preliminary data (not shown) indicating that MT-facilitated transport of PTHrP toward the nucleus also involves dynein. Because optimal PTHrP nuclear import is impaired in Taxol-treated cells, the clear implication is that an intact MT network in itself is insufficient for facilitated import, but rather a dynamic MT network may be required. A possible explanation is that MT dynamics contribute to the transfer of cargo between MTs and dynein where PTHrP may first associate with MTs and then pass to dynein for trafficking. Whether the same mechanism relates to similar observations for MT-facilitated nuclear transport of p53 (4) and Rb (9) remains to be determined.

Interestingly, the MTAS responsible for MT association/facilitated nuclear import, which is within PTHrP residues 82–108 and is critically dependent on the KKKK⁹¹ and KKRR¹⁰⁶ basic clusters, is quite distinct from consensus motifs for DLC association (e.g. KSTQT and GIQVD for DLC8 and SKCSR for Tctex-1), which mediate MT-dependent nuclear import of proteins including rabies virus P-protein (16, 37–39, 50–54). Although the latter appear to be able (out of context) to enhance conventional NLS-dependent nuclear import (16), such sequences do not directly confer MT interaction in contrast to the PTHrP MTAS examined here (55, 56). Clearly, the PTHrP basic motifs responsible for PTHrP MT association differ markedly from defined consensus motifs for DLC association, and mutation of PTHrP KEQE¹⁰¹, which can be postulated to resemble a DLC8 association motif, had no effect on PTHrP MT-dependent nuclear import. PTHrP fusion proteins also do not associate with DLC8 (not shown), further underlining the fact that MTASs and consensus motifs for DLC association are distinct.

In summary, this study has enhanced our understanding of the nuclear trafficking of cellular proteins with key roles in vital processes in physiology and transformation. In addition, it has established the mechanism of MT-dependent nuclear protein cargo delivery. This has clear implications for the design of vectors for non-viral gene delivery to the cell nucleus for which cytoplasmic diffusion of large DNA molecules is a significant limiting factor (57, 58). The fact that PTHrP-(66–108) is a sequence module able to target a 580-kDa fusion protein (GFP-β-gal) rapidly and efficiently into the nucleus dependent on MT integrity (e.g. Fig. 4) implies that combining MTASs with NLSs should have great potential for future gene delivery approaches.