Beyond Nuclear Transport: Ran-Gtp as a Determinant of Spindle Assembly

Ran is an abundant GTP-binding protein that is required for the trafficking of proteins and RNA in and out of the nucleus (Moore and Blobel 1993). In general, it forms a complex with nuclear transport signal (NLS) receptors and their cargoes and directs their movement through nuclear pores. Like other small GTP-binding proteins (e.g., Ras), Ran's GTP hydrolysis activity and guanine nucleotide affinity are modulated by accessory factors. GTP hydrolysis is stimulated by a GTPase-activating protein known as Ran-GAP1 (Bischoff et al. 1994) and its accessory factor RanBP1 (Bischoff et al. 1995), while replacement of GDP with GTP is accomplished by the guanine nucleotide exchange factor (GEF) RCC1 (Bischoff and Ponstingl 1995; Ohtsubo et al. 1989). Together, these proteins comprise an enzymatic cycle by which Ran binds GTP, hydrolyzes it to GDP (due to the activity of Ran-GAP1), releases the GDP (due to RCC1 activity), and rebinds GTP (due to the presence of a relatively high GTP concentration in the cell).

Ran's function as a nuclear transport factor depends upon the nature of its bound guanine nucleotide. Specificity of binding (i.e., to import or export receptors) and direction of movement is determined by whether Ran is bound to GTP or GDP. Ran-GDP typically associates with nuclear import receptors (e.g., importin α/β) and directs their movement from the cytoplasm into the nucleus. Conversely, Ran-GTP generally associates with nuclear export receptors (e.g., CRM1) and directs movement of their cargoes from the cytoplasm into the nucleus (see Koepp and Silver 1996; Mattaj and Englmeier 1998).

According to this scheme, the Ran protein should be predominantly bound to GTP while in the nucleus and to GDP while in the cytoplasm. This is achieved by compartmentalization of Ran's GAP and GEF. Whereas Ran is localized throughout the cell, RCC1 is bound to chromatin in the nucleus (Ohtsubo et al. 1989), while Ran-GAP1 and RanBP1 are found exclusively in the cytoplasm (Matunis et al. 1996; Richards et al. 1996). Consequently, Ran-GDP is prevalent in the cytoplasm due to stimulation of GTPase activity by Ran-GAP1 and RanBP1, whereas Ran-GTP is prevalent in the nucleus due to the nucleotide exchange activity of RCC1 (Fig. 1 A) (Koepp and Silver 1996; Mattaj and Englmeier 1998). Thus, the localization of the GAP and GEF is paramount in regulating the proper function of Ran in nuclear transport.

Whereas the nuclear transport activity of Ran has been extensively characterized, evidence for its role in microtubule regulation has, until recently, been largely circumstantial. An initial hint for Ran involvement in microtubule regulation was derived from the observation that, in the budding yeast Saccharomyces cerevisiae, overexpression of the RCC1 homologue Prp20 can suppress the toxic effects of certain hyperstable α-tubulin mutants (Kirkpatrick and Solomon 1994). Moreover, yeast strains harboring temperature-sensitive mutant alleles of the yeast Ran-binding protein Yrb1 exhibit spindle misorientation defects due to a lack of astral microtubules (Ouspenski 1998). Thus, at a phenotypic level, some mutant forms of Ran-associated proteins can have an effect on microtubule structure in vivo.

Well beyond this initial suggestion, a series of recent papers have converged upon the discovery that Ran and its cohorts are key components of aster formation and spindle assembly, especially for spindles assembled in the absence of centrosomes. An initial insight into this process came from the identification (Nakamura et al. 1998) of a novel mammalian Ran-binding protein, RanBPM, that could elicit microtubule polymerization. Isolated on the basis of its interaction with Ran-GTP in a two-hybrid assay, RanBPM has been shown to associate with centrosomes, the microtubule-nucleating centers of mammalian cells. Interestingly, it has been observed that overexpression of RanBPM can induce ectopic aster formation in transfected cells. Such asters are structurally similar to normal centrosomal asters in that they contain centrosomal proteins such as γ-tubulin at their foci. Moreover, inhibition of RanBPM or Ran activity can prevent in vitro aster formation from a mixture of purified centrosomes and tubulin. Taken together, these results suggest that Ran and RanBPM somehow act together to effect microtubule nucleation.

It has long been known that asters will readily grow from sperm centrioles in Xenopus egg extracts arrested in meiotic metaphase II. Surprisingly, this centriole-mediated process has now been found to be Ran-GTP–dependent. Reduction in the relative levels of Ran-GTP by immunodepletion of RCC1 (Ohba et al. 1999) or by addition of a mutant form of Ran that favors GDP binding (T24N; see Table; Carazo-Salas et al. 1999; Kalab et al. 1999; Ohba et al. 1999; Wilde and Zheng 1999), severely inhibits aster formation from centrioles. Therefore, the GTP-bound form of Ran is necessary for centriole-dependent aster formation in these extracts.

Conversely, addition of high levels of purified Ran-GTP, GTP-locked forms of Ran (e.g. Ran-GTPγS; Ran G19V, Ran Q69L; RanL45E; see Table), or high levels of RCC1 strongly stimulates centriole-associated aster growth in sperm-treated extracts as well as de novo aster formation in extracts lacking added chromatin or centrioles (Carazo-Salas et al. 1999; Kalab et al. 1999; Ohba et al. 1999; Wilde and Zheng 1999; Zhang et al. 1999). Like the asters formed by overexpression of RanBPM (Nakamura et al. 1998), the asters formed by artificially raising the levels of Ran-GTP include typical centrosome-associated proteins (e.g., γ-tubulin, NuMA, XGRIP109, and XMAP215) at their foci (Ohba et al. 1999; Wilde and Zheng 1999). Strikingly, centrosome-free asters formed by high levels of Ran L45E are capable of forming into bipolar spindle-like structures (Wilde and Zheng 1999). In contrast, asters formed without Ran (i.e., by chemically stimulating tubulin polymerization with agents such as dimethyl sulfoxide) do not require γ-tubulin or XMAP215 and do not assemble into higher-order structures (Wilde and Zheng 1999). In sum, the findings that loss of Ran function blocks aster formation while overexpression of Ran-GTP induces ectopic formation of microtubule structures clearly implicate the Ran-GTPase cycle in microtubule assembly in M phase.

Although it can induce aster assembly in Xenopus extracts, Ran-GTP does not stimulate microtubule polymerization from purified tubulin subunits (Wilde and Zheng 1999). Thus, there must exist additional factors that more directly affect polymerization properties. The most plausible candidate now known for such an effector is RanBPM, which both associates with centrosomes and Ran-GTP in mammalian cells. In the absence of centrosomes, the simplest view is that a soluble (i.e., non-centrosomal) portion of RanBPM (or an as yet undiscovered relative) may associate with Ran-GTP generated by chromosomal RCC1, thereby promoting microtubule polymerization adjacent to chromosomes.

Is there a role for GTP hydrolysis by Ran in microtubule assembly? It has been suggested that during nuclear transport, the nucleotide-bound state of Ran acts simply as a switch to delineate the direction of movement and that the energy of GTP hydrolysis is not strictly required (Richards et al. 1997; Schwoebel et al. 1998). The observation that forms of Ran that do not hydrolyze GTP (e.g. Ran L45E, Ran-GTPγS) can induce aster formation in Xenopus extracts (Carazo-Salas et al. 1999; Kalab et al. 1999; Ohba et al. 1999; Wilde and Zheng 1999), suggests that a similar situation exists for microtubule assembly. However, while GTPγS-bound Ran can induce aster formation, such asters are considerably smaller than those formed by Ran-GTP (Ohba et al. 1999). Thus, GTP hydrolysis by Ran may have some secondary role in the elongation of previously nucleated microtubules.

The recognition that Ran-GTP may be a key component of microtubule nucleation allows resolution of an old cell biological problem: how chromatin can drive spindle assembly, especially in the absence of centrosomes (Karsenti et al. 1984). What now seems likely is that in the absence of a nucleus, chromatin-bound RCC1 and cytoplasmic RanGAP1 produce a natural gradient of Ran-GTP that is most concentrated at the chromosome surface (Fig. 2 A). Consequently, a Ran-GTP–dependent factor such as RanBPM, would stimulate microtubule assembly adjacent to the chromosomes. Then, other cellular factors could act to organize these microtubules into spindles. For instance, it has been shown that the chromosome-associated, kinesin-like protein XKLP1 is required for the stable attachment of microtubules to chromatin (Vernos et al. 1995; Walczak et al. 1998). Such chromatin-bound, plus end–directed kinesins could, thus, attach to the newly formed microtubules and draw the rapidly growing plus ends to the surface of the chromosomes (Fig. 2 B). Finally, a complex of cytoplasmic dynein, dynactin and NuMA, previously shown to be required for maintenance of focused microtubule arrays both in the presence or absence of centrosomes (Yang and Snyder 1992; Merdes et al. 1996), could organize the chromosome-bound microtubules into spindles (Fig. 2 C). This would arise through the microtubule cross-linking activity of NuMA and the retrograde motility of dynein acting together to draw the microtubules into poles at their minus ends. Such a model could explain the inside-out assembly of spindles during vertebrate oogenesis and plant mitosis, both of which are accomplished without centrosomes.

Indeed, the works described herein offer some evidence in support of such a model. First there is evidence for the influence of chromatin-dependent components. While it had previously been shown that random segments of DNA attached to a solid-phase support (e.g., polystyrene beads) could initiate centrosome-free spindle assembly in CSF-arrested Xenopus extracts (Heald et al. 1996), Carazo-Salas et al. 1999 have now shown that this process requires generation of Ran-GTP by chromatin-associated RCC1. Second, concerning a role in centrosome-free pole formation, it has been demonstrated that microtubules formed by addition of purified Ran-GTP require the activity of cytoplasmic dynein for organization into aster-like structures (Ohba et al. 1999; Wilde and Zheng 1999).

The linkage of Ran-GTP to M phase microtubule nucleation reinforces a principle long understood from nuclear transport. Compartmentalization of the Ran GAP and GEF modulate the function of Ran itself. Because of its association with chromatin, RCC1 can establish a high concentration of Ran-GTP exclusively in the vicinity of chromosomes during M phase. This, in turn, can serve as the initial positional cue for spindle formation in the absence of centrosomes. Thus, the observation that the GTP-bound form of Ran can stimulate microtubule polymerization offers a significant insight into the process by which chromosomes drive spindle assembly, especially in the absence of microtubule-organizing centers.