Two protein 4.1 domains essential for mitotic spindle and aster microtubule dynamics and organization in vitro

Multifunctional structural proteins belonging to the 4.1 family are components of nuclei, spindles, and centrosomes in vertebrate cells. Here we report that 4.1 is critical for spindle assembly and the formation of cen-trosome-nucleated and motor-dependent self-organized microtubule asters in metaphase-arrested Xenopus egg extracts. Immunodepletion of 4.1 disrupted microtubule arrays and mislocalized the spindle pole protein NuMA. Remarkably, assembly was completely rescued by supplementation with a recombinant 4.1R isoform. We identified two 4.1 domains critical for its function in microtubule polymerization and organization utilizing dominant negative peptides. The 4.1 spectrin-actin binding domain or NuMA binding C-terminal domain peptides caused morphologically disorganized structures. Control peptides with low homology or variant spectrin-actin binding domain peptides that were incapable of binding actin had no deleterious effects. Unexpectedly, the addition of C-terminal domain peptides with reduced NuMA binding caused severe microtubule destabilization in extracts, dramatically inhibiting aster and spindle assembly and also depolymerizing preformed structures. However, the mutant C-terminal peptides did not directly inhibit or destabilize microtubule polymerization from ice for 10 min prior to the initiation of assembly. Reactions in three independent experiments were sampled during 15–45-min incubation periods for the assembly of spindles, centrosomes, or asters. The exper- iment presented was performed in parallel using the same depleted extract as described (19). remained entirely in the supernatant. However, in reactions with tu- bulin and C-terminal domain mut3V peptides ( 4.1R 20–21 mut3V ) or C-terminal domain ( 4.1R 20–21 ) but not SADB peptides, a fraction of the peptides was detected in microtubule pellets.

Shipping and Delivery: UPS ground within the United States (1-5 days delivery) is included in the reprint prices, except for orders in excess of 1000 copies. Orders shipped to authors outside the United States are mailed via an expedited air service. Your order will be shipped within 2 weeks of the journal print date.

NOTICE TO AUTHORS
To avoid delay in publication, please provide complete information to the following questions.

Category
The Table of Contents is arranged by Category. Please indicate which of the following is the appropriate Category for your paper.
Genes: Structure and Regulation 3.
Protein Structure and Folding 7.
Enzyme Catalysis and Regulation 8.
Glycobiology and Extracellular Matrices 10. Lipids and Lipoproteins 11. Membrane Transport, Structure, Function, and Biogenesis 12. Mechanisms and Signal Transduction 13. Molecular Basis of Cell and Developmental Biology Authors of Accelerated Publication papers: If your paper exceeds 4 pages in length and you choose to run it as a regular article, please indicate category.

Reprints
Please indicate the number of reprints ordered:

No Color
We recommend that articles with gels and electron micrographs be printed on heavy paper for higher quality reproduction. Cost for heavy paper is $120: cost for each halftone is $20 regardless of paper choice. For reproduction of fi gure(s) on heavy paper, total cost would be $ Please circle your choice of paper. If you have no halftones, your paper will automatically be printed on regular stock paper.

HEAVY PAPER REGULAR PAPER
Color Your article will automatically be printed on heavy paper.

Revised Page Proofs
Revised page proofs can be faxed to you if you have a specifi c concern regarding the makeup of your paper or if there are extensive alterations. Please note that revised page proofs are used for confi rmation purposes only and are not considered as a second opportunity for corrections.

COLOR ARTICLES ONLY
We CANNOT publish your article until we receive approval of the reproduction quality for your color fi gure(s). Please view your color images on the computer screen and fax or e-mail your approval to me within 24 hours, referring to the article number. P.S. When you return the page proofs, please be sure to indicate the total number of reprints desired. It is also important to return your reprint order form as quickly as possible to make sure the color reprints are ordered promptly.

New-From the Publishers of the Journal of Biological Chemistry
Visit the Web Site www.mcponline.org Molecular and Cellular Proteomics will have an emphasis placed on determining how the presence or absence of proteins affects biological responses and how the interaction of proteins with relevant cellular partners allows them to function. Articles utilizing or advancing protein identification technology -such as multi-dimensional electrophoresis and/or mass spectrometry -protein and nucleic acid arrays, and computational assessments will be particularly appropriate.
• In addition to manuscripts describing research advances in proteomics, articles concerning technological advances will also be accepted. In addition, MCP will publish large data sets as either appendices to regular manuscripts or as stand alone contributions. The latter must include a summary, not to exceed two printed pages, describing the germane points and importance of the information. The data sets themselves (either as appendices or as separate articles) will appear only in the on-line version. A letter of intent describing the extent and format of this supplemental material must precede submission of the manuscript.
• Electronic Manuscript Submission -Manuscript submission, review, and initial appearance will all be accomplished electronically (the e-version will be published as a member of the HighWire consortium).
• Immediate Publication -All papers accepted for publication will appear immediately as a Paper in Press.
• Printed Monthly -The print version will appear on a monthly basis (without supplemental information). Multifunctional structural proteins belonging to the 4.1 family are components of nuclei, spindles, and centrosomes in vertebrate cells. Here we report that 4.1 is critical for spindle assembly and the formation of centrosome-nucleated and motor-dependent self-organized microtubule asters in metaphase-arrested Xenopus egg extracts. Immunodepletion of 4.1 disrupted microtubule arrays and mislocalized the spindle pole protein NuMA. Remarkably, assembly was completely rescued by supplementation with a recombinant 4.1R isoform. We identified two 4.1 domains critical for its function in microtubule polymerization and organization utilizing dominant negative peptides. The 4.1 spectrin-actin binding domain or NuMA binding C-terminal domain peptides caused morphologically disorganized structures. Control peptides with low homology or variant spectrinactin binding domain peptides that were incapable of binding actin had no deleterious effects. Unexpectedly, the addition of C-terminal domain peptides with reduced NuMA binding caused severe microtubule destabilization in extracts, dramatically inhibiting aster and spindle assembly and also depolymerizing preformed structures. However, the mutant C-terminal peptides did not directly inhibit or destabilize microtubule polymerization from pure tubulin in a microtubule pelleting assay. Our data showing that 4.1 is a crucial factor for assembly and maintenance of mitotic spindles and self-organized and centrosome-nucleated microtubule asters indicates that 4.1 is involved in regulating both microtubule dynamics and organization. These investigations underscore an important functional context for protein 4.1 in microtubule morphogenesis and highlight a previously unappreciated role for 4.1 in cell division.
Protein 4.1, formerly characterized solely as a crucial membrane skeletal protein in mature red cells, is now also recognized to be an important multifunctional structural protein family in nucleated cells. Although protein 4.1 can be plasma membrane-associated in nucleated cells, it also is detected at diverse and interesting subcellular locations during the cell cycle. Protein 4.1 isoforms localize within the nucleus and at centrosomes during interphase, at spindle poles during mitosis, in perichromatin at anaphase, and in the midbody at telophase (1)(2)(3). Whereas mature red cells express only 80-kDa 4.1, the complex subcellular localization patterns of 4.1 in mammalian cells likely results from expression of several 4.1 isoforms, post-translational modifications and expression of multiple 4.1related genes (4.1R, G, B, and N) (4 -14).
Beyond the characterization of its localization, the current challenge is to decipher functions of 4.1 in various subcellular structures. Although the list remains incomplete, a number of protein 4.1 binding partners have been identified to interact with a specific 4.1 domain in red cells and/or nucleated cells, providing some clues as to potential 4.1 functions (see Fig. 1). Prototypical protein 4.1 (R, red cell) contains several functional domains. An N-terminal extension present only in some isoforms in nucleated cells has been found to interact with the centrosomal protein CPAP (centrosome protein-4.1 associated protein) (15). The FERM (4.1/erzin, radixin/moesin) domain interacts with plasma membrane-binding proteins and was recently discovered to also contain a microtubule binding site (16). The spectrin-actin binding domain (SABD 1 ) is capable of forming ternary complexes with spectrin and actin. The Cterminal domain is of particular interest, because it has been found to interact with NuMA (nuclear mitotic apparatus protein) (17). These observations suggest that 4.1 plays diverse roles within the cytoskeleton.
Recently we showed (18,19) that 4.1 is essential for the assembly of functional nuclei in interphasic Xenopus egg extracts and requires its capacity to bind actin, which is found closely associated with 4.1 on nuclear filaments in mammalian cells. This latter observation was recently confirmed and extended in a study (20) that characterized an extensive system of nuclear pore-linked filaments in Xenopus oocytes that contain actin and 4.1 epitopes. Here we provide evidence that 4.1 is also essential during mitosis in Xenopus egg extracts for proper polymerization and organization of the microtubule cytoskeleton.
During interphase, centrosomes nucleate and organize a radial array of microtubules. We characterized protein 4.1 previously as an integral centrosome component, resisting vigorous salt/detergent extraction and present in centrosomes independent of microtubules. By immunofluorescence and cell whole mount electron microscopy 4.1 epitopes localized on centrioles, in the pericentriolar matrix, and on the fibers connecting the centriolar pair (1). During mitosis duplicated centrosomes become the poles of the mitotic spindle apparatus, contributing to the organization of microtubules into a bipolar array with their minus ends focused at the poles and their plus ends interacting with chromosomes and overlapping within the center of the spindle. Ultimately responsible for accurate chromosome segregation, spindle function is also dependent on a variety of microtubule-based motor proteins including dynein and kinesin-related proteins that cross-link and sort microtubules according to their structural polarity, and mediate chromosome interactions within the spindle. Proper spindle pole organization is known to depend on the function of NuMA, which interacts with dynein and contributes to microtubule minusend cross-linking to maintain spindle pole structure (as reviewed in Ref. 21) (22)(23)(24)(25)(26). Taken together, the observations that protein 4.1 binds to NuMA and localizes to centrosomes and mitotic spindle poles raise the intriguing hypothesis that protein 4.1 is involved in cell division.
In this report we used Xenopus egg extracts, a powerful system for in vitro dissection of mitotic microtubule assembly, organization, and function. Xenopus eggs are laid arrested in the metaphase of meiosis II by cytostatic factor, until fertilization triggers a calcium wave promoting entry into the first mitotic cell cycle. Extracts prepared from unfertilized eggs maintain the cytostatic factor metaphase arrest. Mitotic microtubule asters analogous to spindle poles can be assembled in cytostatic factor extracts through the addition of purified centrosomes that directly nucleate microtubules, or by the addition of microtubule stabilizing agents such as Me 2 SO or taxol that induce microtubule polymerization. In the absence of centrosomes, asters are progressively organized into a focused array of minus ends by motor proteins including dynein and are similarly dependent on the activity of NuMA (24,27). Complete mitotic spindle assembly can be induced by adding demembranated Xenopus sperm nuclei, as the centriole-containing basal body of the flagellum remains tightly attached to the sperm and becomes competent to nucleate microtubules in the extract, defining the spindle poles. Thus, the Xenopus system can be used to probe both centrosome-nucleated and motor-dependent organization of microtubules into asters, as well as the more complex events of mitotic spindle assembly.
Taking advantage of the open nature of this cell-free system, we present data establishing that protein 4.1 is essential for assembly of these microtubule-based structures by demonstrating that 4.1 depletion of extracts produces aberrant structures. Proper assembly can be restored by the addition of recombinant 4.1. Furthermore we have identified two 4.1 domains critical for spindle, centrosome, and microtubule aster assembly in assays utilizing dominant negative peptides corresponding to 4.1 domains. Our results indicate that 4.1 is involved in regulating both microtubule dynamics and organization and underscore an important functional context for protein 4.1 in cell division.

EXPERIMENTAL PROCEDURES
Materials-Expression vectors for His 6 fusion peptides were either pMW172 (the gift of Dr. M. Way, European Molecular Biology Laboratory, Heidelberg, Germany) or pET 28 (Novagen). The antibody against NuMA was a very generous gift of Dr. A. Merdes (U. Edinburgh, Scotland). IgGs against 4.1R SABD and 4.1R C-terminal domain were described (2). Fluorescent secondary antibodies were from Molecular Probes. Bovine brain tubulin was prepared according to Ashford et al. (28).
Xenopus Extracts and Assembly Reactions-10,000 ϫ g cytoplasmic Xenopus egg extracts and demembranated sperm nuclei were prepared as described (29). For spindle assembly, demembranated Xenopus sperm were added to 20 l of egg extract on ice with 0.2 mg/ml Texas Red-labeled tubulin, reactions incubated at 20°C for 30 -45 min., diluted with BRB80 (80 mM PIPES, 2 mM MgCl 2 , 1 mM EGTA, pH 6.8) containing 30% glycerol and 1% Triton X-100 and spun through BRB80 cushions with 40% glycerol onto coverslips (30). Self-organized microtubule asters were assembled by addition of Me 2 SO (final concentration 5%) to the egg extract and incubation for 15 min at 20°C (31). Centrosome asters were assembled for 15 min at 20°C after the addition of 1 l of KE37 centrosomes at 2 ϫ 10 8 /ml, prepared according to Moudjou and Bornens (32), to 20 l of extract that had been centrifuged for 30 min at 60,000 rpm in a TLA 100.3 rotor. Centrosome or self-organized aster reactions were diluted with BRB80 containing 15% glycerol and 1% Triton X-100 and spun onto coverslips through a cushion of BRB80 containing 30% glycerol.
Indirect Immunofluorescence-In vitro assembled structures on coverslips were fixed in Ϫ20°C MeOH and probed by immunofluorescence as described (33). The concentrations of primary antibodies were: SABD IgG, 5 g/ml; C-terminal domain IgG, 10 g/ml; anti-NuMA, 1:50 dilution. Secondary antibodies were used at a 1:100 dilution. Samples probed with equal amounts of control non-immune IgG or without primary antibody or sera showed no fluorescent patterns. Images were captured using a Nikon Eclipse 2000 microscope equipped with a CCD camera and processed using Adobe PhotoShop. Under the imaging conditions used, the limits of resolution of overlap between two fluorophores (e.g. superimposition of red and green signals to generate yellow coloration) was estimated to be ϳ300 nm.
Expression and Purification of His 6 -tagged Proteins-Protein 4.1related peptides were expressed and purified as described (19). Assays with His 6 peptides (1-8 g) were preincubated on ice for 10 min then incubated at 20°C for assembly of spindles, centrosome asters, or self-organized asters as indicated. Although a range of 1-8 g was tested for each peptide, the data presented are from experiments using 8 g of the indicated peptide.
Microtubule  (34). Immunodepletion and Rescue-For 4.1 depletion from Xenopus extracts, protein G-coupled magnetic beads (Dynal) from 100 l of slurry were mixed with 15 g of 4.1R SABD, C-terminal domain IgGs or non-immune rabbit IgG for 1 h at 4°C, the beads were washed twice with 0.1 M sodium phosphate buffer, pH 7.0 (57.7% Na 2 HPO 4 and 42.3% NaH 2 PO 4 , v/v), and three times with XB buffer (29) then divided into three aliquots. Extract (100 l) was successively depleted three times by rotation with IgG-coupled beads at 4°C for 1 h, beads were collected magnetically, and extract was used for the assembly of spindles, centrosome asters, or self-organized asters. Extract depletion was estimated by densitometry of Western blots using an Alpha Imager 2200 and software. In rescue experiments, 1-9 g of purified bacterially expressed 80-kDa 4.1R was added to 20-l reactions and incubated on ice for 10 min prior to the initiation of assembly. Reactions in three independent experiments were sampled during 15-45-min incubation periods for the assembly of spindles, centrosomes, or asters. The experiment presented was performed in parallel using the same depleted extract as described (19).

Localizes to Mitotic Spindles, Centrosome Asters, and Self-organized Asters Reconstituted in Xenopus Egg Extracts-
Previous reports (1, 2, 16, 17) using a variety of mammalian cells established that 4.1 is localized to centrosomes and mitotic spindle poles. As the Xenopus laevis 4.1 sequence has many highly homologous regions relative to mammalian family members, including the SABD and C-terminal domains (Fig. 1) (19), we anticipated a similar 4.1 localization in Xenopus. Furthermore, 4.1 function was shown to be conserved between the frog and mammals in studies reconstituting properties of 4.1-deficient human erythrocyte membranes using a recombinant Xenopus 4.1 domain (35). We first verified that 4.1 epitopes could be detected in centrosomes and mitotic spindles of cultured Xenopus fibroblasts (data not shown). Next we incubated cytostatic factor-arrested Xenopus egg extracts either with demembranated Xenopus sperm to assemble mitotic spindles or with centrosomes from KE37 cells to reconstitute centrosome-based microtubule arrays (referred to as centrosome asters). We also induced the motor-dependent assembly of stabilized microtubules into asters by adding 5% Me 2 SO to the egg extract (referred to as self-organized asters). This allowed us to examine both centrosome-nucleated and motor-organized structures. In each reaction, microtubules were stained by the addition of trace amounts of rhodamine-labeled tubulin. When each of these structures was probed by immunofluorescence microscopy using affinity-purified antibodies against the 4.1R SABD or C-terminal domain or against the 80-kDa 4.1R, strong 4.1 signals were concentrated at the minus ends of microtubules focused at the spindle poles and also in the center of both centrosome and self-assembled asters (Fig. 2).

Immunodepletion of 4.1 Compromises Assembly of Microtubule-based Structures and Can Be Rescued by Supplementation with Recombinant 4.1R-
To test whether the protein 4.1 itself is essential for the assembly of spindles, centrosome asters, and self-organized asters, we depleted 4.1 from Xenopus extracts using 4.1 domain-specific affinity-purified IgGs bound to protein G magnetic beads. Previously (19) we showed by Western blotting that Xenopus extracts contain protein bands from ϳ47 to 110 kDa detected by SABD and C-terminal domain IgGs, which were reduced by 50 -100% after three rounds of antibody depletion. Using 4.1-immunodepleted extracts, we evaluated the morphology of spindles, centrosomes, and self-organized microtubule asters, and as an additional indicator of structural integrity we localized the 4.1 binding partner/spindle pole protein NuMA by immunofluorescence. In controls, NuMA largely localized with 4.1 epitopes in a tight focus at the spindle poles, and in the centers of centrosome and self-organized microtubule asters (Figs. 2 and 3).
Normal spindles, centrosome asters, and self-organized asters assembled in egg extracts mock-depleted with nonspecific IgG and protein G beads (Fig. 3A). In contrast, there was a dramatic morphological disruption of all microtubule structures assembled in extracts depleted with either 4.1 domainspecific SABD or C-terminal domain IgGs. In spindles assembled in either depleted extract, chromosomes were not aligned equidistant from the poles but were looped out of the spindle midzone (Fig. 3A). Spindles formed in SABD-depleted extracts most often were multipolar, whereas those from C-terminal domain-depleted extracts generally had large unfocused poles. Centrosome asters were disorganized microtubule arrays without an obvious focal center revealed by NuMA staining (Fig.  3B). In SABD-depleted extracts, centrosomes asters often contained multiple small NuMA foci radiating several bundles of microtubules. Similarly, self-organized asters assembled in either the C-terminal domain or SABD-depleted extracts had disorganized microtubules, and NuMA was mislocalized (Fig.  3C). Therefore, depleting SABD-and C-terminal domain-con-taining 4.1 proteins disrupted spindle, centrosome aster, and self-organized microtubule aster assembly.
The aberrant reconstitution of microtubule structures observed in depleted extracts could result either from loss of 4.1 function itself or from loss of an essential 4.1 protein binding partner co-depleted in the reaction. To address this issue we added back-purified recombinant 80-kDa 4.1R to depleted extracts. Strikingly, spindle, centrosome, and aster reconstitution was completely restored by supplementation with recombinant 4.1R in extracts depleted with either SABD or C-terminal domain IgGs, producing structures with morphology and NuMA distribution comparable with controls (Fig. 3, AЈ, BЈ, and CЈ). The rescue of assembly by recombinant 4.1R shows directly that protein 4.1 is essential for the assembly of mitotic spindles, centrosome, and self-organized microtubule asters. Furthermore, this result indicates that a single isoform containing both SABD and C-terminal domains is sufficient to mediate all of the functions of 4.1 necessary for its role in organizing mitotic microtubule arrays in Xenopus egg extracts.

FIG. 2. Protein 4.1 epitopes detected by immunofluorescent microscopy in spindles, centrosome asters, and self-organized microtubule asters assembled in vitro in Xenopus egg extracts.
Localization of DNA in spindles was detected by 4,6-diamidino-2phenylindole (blue), and microtubules were labeled with rhodamine tubulin (red). Protein 4.1 epitopes were visualized by indirect immunofluorescence using affinity-purified IgG directed against the 4.1 SABD and C-terminal domain as described previously (2). Both IgGs produced protein 4.1 signals (green) at spindle poles and at the minus ends of microtubules oriented toward the center of centrosome asters and self-organized asters. The yellow coloration in these merged images indicates apparent colocalization of 4.1 epitopes and microtubules. Bar ϭ 20 m

.1 Peptides Distort Assembly of Spindles, Centrosomes, and Microtubule Asters in Vitro-Having
established that 4.1 is required for proper formation of microtubule structures, we next wanted to test the functions of specific 4.1 domains in spindle, centrosome, and microtubule aster assembly. To this end we added to in vitro assembly reactions bacterially expressed peptides with amino acid sequences corresponding to 4.1 domains. We reasoned that the peptides might act competitively to disrupt 4.1 complexes or to sequester important 4.1 binding partners during the assembly process. We analyzed peptide effects both on morphology and localization of the 4.1 binding partner NuMA as another measure of functional disruption because in controls NuMA localized in a tight focus at spindle poles, in the pericentriolar area of centrosomes and at the centers of self-organized asters (Fig. 3).
Initially we analyzed the effects of peptides related to 4.1 SABD and C-terminal domains. We focused on these domains because they (a) have important defined functions, (b) are highly conserved between frog and mammals, (c) were present in recombinant 4.1R used to rescue extracts immunodepleted by SABD and C-terminal domain IgGs, and (d) were demonstrated to profoundly distort nuclear assembly in vitro in egg extracts (19). For our experiments, we expressed His 6 -tagged peptides encoded by either exons 16 -17 (amino acids 644 -705) in the 4.1R spectrin-actin binding domain (SABD) or by exons 20 -21 (amino acids 800 -858) of the 4.1R C-terminal domain (Fig. 4). As controls for the SABD peptide, we used a variant 4.1N SABD peptide with low amino acid sequence homology to the 4.1R 16 -17 peptide and a 4.1R 16 -17⌬NF peptide with a deletion of two amino acids within its actin binding domain rendering it unable to bind actin but retaining spectrin binding. As a control for the 4.1R C-terminal domain peptide we used a C-terminal domain peptide in which three valines were changed to alanines, an alteration previously shown to reduce its binding affinity to NuMA Tail I fragment by ϳ60-fold (19).
Addition of the 4.1R 16 -17 SABD peptide produced spindles with less focused microtubules at the poles and more dispersed NuMA in the peripolar area. Often there were multipolar structures. In these spindles, chromatin appeared less condensed at the metaphase plate (Fig. 5A, top row). Centrosome asters, while retaining a small central focus of NuMA staining, had only a few bundled microtubules radiating outward (Fig. 5A, middle row). Self-organized asters also appeared to contain bundled microtubules and had a very diffuse central area of NuMA with additional NuMA epitopes distributed along microtubule bundles (Fig. 5A, bottom row). However, an equal In the merged images, DNA in spindles was detected with 4,6-diamidino-2-phenylindole (blue), microtubules are red, and NuMA is green. Yellow indicates an overlap of red and green signals. A, the products of spindle assembly reactions in Xenopus extracts depleted using SABD IgG (⌬SABD) or C-terminal domain IgG (⌬C-terminal domain) were severely perturbed relative to spindles assembled in control extracts (IgG). Bar, 20 m. B, centrosome asters assembled in Xenopus extracts depleted using SABD IgG or C-terminal domain IgG were aberrant both with respect to microtubules and NuMA distribution relative to centrosome asters assembled in control extracts. C, self-organized microtubule asters formed after Me 2 SO addition to extracts depleted using SABD IgG or C-terminal domain IgG were disoriented and had a scattered distribution of NuMA epitopes. AЈ, BЈ, CЈ, spindles, centrosome asters, and self-organized asters assembled after the addition of recombinant 80-kDa 4.1 (ϩ4.1) to depleted extracts. With the addition of 9 g of 80-kDa 4.1 spindles, centrosome asters, and self-organized asters had similar morphology, size, and distribution of NuMA as to their respective controls.  concentration of a deletion mutant in the 4.1R SABD peptide-(4.1R 16 -17⌬NF) added to extracts did not affect the assembly of spindles, centrosomes, or asters with respect to the morphology or distribution of NuMA (Fig. 5A). A variant 4.1N SABD peptide also did not cause any apparent perturbation (Fig. 5A). Therefore the dominant negative effects of the 4.1R SABD peptide on proper structural assembly are sequence-specific and require its capacity to bind actin. The C-terminal domain was also critical for the assembly of normal spindles, centrosomes, and microtubule asters. C-terminal domain peptide addition resulted in structures even more distorted than those treated with SABD peptides. Spindles had unfocused poles or were multipolar and were also bent in most cases. Centrosome-nucleated and self-organized asters were not radiating from a discernable center, but appeared as microtubule "mats" with irregularly distributed NuMA (Fig.  5A). Although the importance of NuMA in proper spindle formation is well documented, these observations imply that the interaction of NuMA and 4.1 is also crucial for centrosome and microtubule aster formation in vitro. To test this, the mutant C-terminal domain peptide with decreased NuMA affinity was added to the assembly reactions. Surprisingly this peptide had dramatic and unpredicted effects. In both centrosome and aster assembly reactions, no microtubules of normal length were observed but only small NuMA foci were detected with extremely short and sparse microtubules (Fig. 5A). In spindle reactions, Xenopus sperm DNA was decondensed and had juxtaposed a single polar remnant. Quantification of these effects shows that dominant negative effects of 4.1R SABD and Cterminal domain peptides on microtubule-based structures are profound and sequence-specific (Fig. 5B).  could be added at any time to assembly reactions. We tested whether microtubules in bipolar spindles and selforganized asters could survive exposure to C-terminal domain mut3V peptides by adding C-terminal domain mut3V peptides to extracts containing normal preformed structures. After a 15-min incubation, few microtubules remained in spindles. These microtubules appeared to extend from pole remnants demarcated by very small NuMA-staining foci and were markedly "slackened." Chromatin was decondensed and not positioned at a spindle equator, extending beyond the microtubules (Fig. 6A). Incubation with wild type C-terminal domain peptide also compromised structure but much less severely than Cterminal domain mut3V mutant peptides. Spindle microtubules became less rigidly and symmetrically organized with smaller less focused NuMA staining in the pole areas. Some NuMA was also unevenly distributed on chromatin (Fig. 6A). Incubation with SABD peptides dispersed peripolar NuMA, whereas SABD⌬NF peptides did not appear to compromise spindle morphology or NuMA localization.
To test more stringently the effects of 4.1 C-terminal domain peptides on microtubules arrayed independently of centrosomes, self-organized asters stabilized by Me 2 SO were incubated with 4.1 peptides. After the exposure of microtubule asters to C-terminal domain mut3V peptides, there remained on average 3-8 microtubule bundles emanating from a NuMAcontaining area similar in size to controls (Fig. 6B). Incubation with wild type C-terminal domain peptides produced a different structure phenotype having markedly truncated microtubules symmetrically arrayed around a NuMA region less tightly focused than in controls. In some orientations, a central hollow area or ring-like distribution of NuMA was detected (Fig. 6B). Because most mature microtubules in spindles as well as self-organized asters did not survive exposure to either mutant or wild type C-terminal domain peptides, it appears that continuous 4.1 function is required to maintain proper microtubule polymerization and organization.
The deleterious effects on mature spindles and asters could result from direct or indirect microtubule destabilization mechanisms. To investigate whether 4.1 peptides directly bind or destabilize microtubules, we performed a sedimentation assay using purified tubulin polymerized in vitro that compares the polymerized tubulin polymer (in the pellet) with unpolymerized tubulin (in the supernatant fraction). In control reactions without tubulin, none of the 4.1-related peptides pelleted under microtubule polymerizing conditions (Fig. 7, Ϫtub lanes). When added to microtubule polymerization reactions, SABD peptides remained entirely in the supernatant and did not appreciably alter the amount of microtubules pelleted. In reactions with either Cterminal domain or C-terminal domain mut3V peptides, each peptide was detected both in the microtubule pellet and in the supernatant (Fig. 7, ϩtub lanes). However, the amount of tubulin in the supernatant and pellet was similar to controls or when SABD peptide was added. Therefore, whereas both C-terminal domain-related peptides showed some binding to microtubules and SABD peptides did not, the 4.1 peptides did not have measurable direct microtubule destabilizing activity.

Protein 4.1 Is Essential for Spindle and Centrosome
Assembly-Protein 4.1 interactions in mammalian red cells serve to link and stabilize structural components in the membrane skeleton with integral membrane proteins. Protein 4.1 is also widely distributed in many tissues and localizes at several distinct subcellular sites crucial for cell division in nucleated cells. By analogy to its critical role in red cell membrane mechanics, protein 4.1 may link or stabilize components in nuclei, centrosomes, and mitotic spindles, providing both structural organization and flexibility necessary for dynamic cytoskeletal assembly and disassembly during the cell cycle. To begin to address 4.1 function in microtubule morphogenesis, we exploited the power of in vitro reconstitution using Xenopus egg extracts after first establishing 4.1 localization at in vitro assembled spindles, centrosomes, and self-organized microtubule asters. In depletion/add-back experiments, markedly aberrant structures formed in depleted egg extracts, but the addition of purified recombinant 80-kDa 4.1R protein restored normal centrosome and spindle assembly demonstrating that 4.1 is essential. Because 4.1 was required for arraying microtubule asters by motor-driven self-organization as well as by microtubule FIG. 6. Effects of 4.1 peptides on structural maintenance of mature spindles and self-organized microtubule asters. A, after assembly of spindles in extracts 4.1 peptides were added and incubated for an additional 15 min. Structures were spun onto coverslips and analyzed by immunofluorescence. In the merged images, DNA was stained by 4,6diamidino-2-phenylindole (blue), microtubules are red, and NuMA is green. Bar, 20 m. B, microtubule asters formed after an addition of 5% Me 2 SO to extracts were further incubated for 15 min with 4.1 peptides, and structures were spun onto coverslips. Coincidence of red (microtubules) and green (NuMA) signals generates a yellow coloration. Bar, 15 m.

FIG. 7. Analysis of in vitro association of 4.1 peptides and microtubules.
Western blots of equivalent aliquots of pellets (P) and supernatants (S) from co-pelleting assays were probed with antibodies against tubulin and His 6. In all tubes containing tubulin (ϩtub), microtubules readily formed as detected in pellet fractions regardless of the presence of various peptides. When tubulin was omitted, the peptides remained entirely in the supernatant. However, in reactions with tubulin and C-terminal domain mut3V peptides (4.1R 20 -21 mut3V) or C-terminal domain (4.1R 20 -21) but not SADB peptides, a fraction of the peptides was detected in microtubule pellets. nucleating centrosomes, the requirement for 4.1 to properly organize and orient microtubules is not dependent exclusively on centrioles.
Although rescue was with recombinant 80-kDa 4.1R containing both a SABD and C-terminal domain, it is possible that 4.1 rescue may not be exclusive to that isoform or even to the 4.1R family. Multiple 4.1R splice variants exist, and recent transfection studies (16) showed that apparent co-localization of 4.1R with microtubules or disruption of microtubule organization may be isoform-and tissue-specific. Furthermore, another generally expressed 4.1 family member (4.1G) also contains an SABD region that can form a ternary complex with spectrin and actin (36) and a highly homologous 4.1G C-terminal domain that can bind NuMA (37). Because 4.1G colocalizes with NuMA at spindle poles (37) and has been detected at centrosomes, 2 future experiments will test the rescue of spindle and centrosome assembly in 4.1-depleted extracts using specifically engineered 4.1 recombinant proteins from both the 4.1R and 4.1G families. This approach will facilitate functional mapping of 4.1 sequences required for microtubule organization in addition to the microtubule binding site in the 4.1R FERM (16).

Peptides From Two Independent Domains Act as Dominant Negatives in Assembly
Reactions-To begin to define 4.1 domains critical for aster/spindle assembly and structure we added peptides to egg extracts, initiated assembly and examined morphology of the structures formed. One domain that behaved as a dominant negative was the SABD peptide. Although neither F-actin nor spectrin is reported to be a centrosome or spindle component, an SABD peptide mutant for actin binding did not exert deleterious effects on morphology. This suggests several of the following possibilities. 1) The SABD may be able to interact with centractin/Arp1, a protein Ͼ50% homologous to actin required for centrosomal microtubule anchoring and/or focusing (38,39). Centractin/Arp1 is also a major component of dynactin, an activator of the minus enddirected microtubule motor cytoplasmic (40,41). This is a particularly attractive hypothesis because disruption of dynein-dynactin complexes or overexpression of centractin/Arp1␣ perturbs centrosomes and spindles (25,42,43). 2) Interactions between microtubules and actin may contribute to aspects of spindle and centrosome formation. Proteins like MACF have been identified that cross-link these two filaments (44). 3) Non-filamentous actin, present in nuclei (45,46) could play a role in spindle and centrosome assembly. It was reported that one molecule of actin is present for each dynactin complex (47). Future experiments could test whether actin inhibitors such as latrunculin interfere with spindle and centrosome assembly. 4) The two deleted amino acids in SABD peptides, critical for actin binding (36), may also be part of another interaction site for an unidentified 4.1 binding partner.
The discovery of a NuMA binding site within exons 20 -21 of the 4.1R C-terminal domain and that NuMA, dynein, and the dynactin subunit p150 glued co-precipitate with 4.1 (17) provided us with insights about possible mechanisms responsible for the dominant negative effects of C-terminal domain peptides on spindle and centrosome assembly. Transport of NuMA toward microtubule minus ends is required for assembly and maintenance of focused spindle poles (25); aster C-terminal domain peptides may disengage NuMA/dynactin as cargo from the dynein motor. This is not unlikely, because the transient nature of the interaction of dynein with dynactin/NuMA has been reported (25,40). Additionally, NuMA also directly binds to and stabilizes microtubules (48), but stabilization is regulated by other proteins such as LGN (49). Although the neighboring 4.1 binding site on NuMA does not directly overlap the microtubule/LGN binding site of NuMA, 4.1 C-terminal domain peptides may indirectly alter the modulation of microtubule dynamics integral to the assembly of spindle poles and asters. The dense microtubule mats with mislocalized NuMA observed after centrosome or aster assembly with C-terminal domain peptides may reflect NuMA oligomerization and stabilization of microtubules (48,49).
Protein 4.1 and the Minus Ends of Microtubules-Localization of 4.1 at the minus ends of radially arrayed microtubules in spindles and centrosomes may occur by two mechanisms: 1) transport by minus end-directed motors and/or 2) association with other minus end-associated proteins already on site (reviewed in Ref. 50). At spindle poles/centrosomes 4.1 could function not only as a crucial structural protein but also to regulate interactions within scaffolding complexes such as those containing NuMA (17), ␥-tubulin (15), 2 and pericentrin (1). Some of these interactions appear to be dynamic, because as reported here 4.1 function is required for continuous maintenance of proper mitotic structures. However, many key functions of 4.1 related to microtubules remain to be identified as suggested by another recent study (37) as well as demonstrated by our own data.
In mitotic egg extract assembly reactions, few microtubules survived exposure to C-terminal domain peptides containing changes in only three amino acids critical for NuMA binding. When added to nuclear assembly reactions in interphasic egg extracts, this peptide also produced markedly truncated centrosome microtubule arrays (19). However, C-terminal domain mut3V peptides did not directly destabilize microtubules polymerized from purified tubulin and, in fact, partially co-pelleted with them. This observation indicates the involvement of other extract proteins in co-regulation of microtubule dynamics by 4.1.
Our investigations predict that multiple protein 4.1 domains regulate microtubule dynamics and/or release as well as microtubule organization. Apparently this is accomplished by both direct and indirect mechanisms. Supporting the hypothesis that 4.1 impacts cell division, depletion of 4.1 or addition of 4.1 dominant negative peptides produced multipolar asymmetric spindles and strongly disrupted centrosome organization. Clearly 4.1 misregulation or disruption in vivo could lead to spindle and centrosomal aberrations, abnormal chromosome segregation, and cell cycle perturbations, hallmarks of potential pathology.