The DYNLT3 Light Chain Directly Links Cytoplasmic Dynein to a Spindle Checkpoint Protein, Bub3*

Cytoplasmic dynein is the motor protein responsible for the intracellular transport of various organelles and other cargoes toward microtubule minus ends. However, it remains to be determined how dynein is regulated to accomplish its varied roles. The dynein complex contains six subunits, including three classes of light chains. The two isoforms of the DYNLT (Tctex1) family of light chains, DYNLT1 and DYNLT3, have been proposed to link dynein to specific cargoes. However, no specific binding partner had been found for the DYNLT3 light chain. We find that DYNLT3 binds to Bub3, a spindle checkpoint protein. Bub3 binds exclusively to DYNLT3 and not to the other dynein light chains. Glutathione S-transferase pull-down and co-immunoprecipitation assays demonstrate that Bub3 interacts with the cytoplasmic dynein complex. DYNLT3 is present on kinetochores at prometaphase, but not later mitotic stages, demonstrating that this dynein light chain, like Bub3 and other checkpoint proteins, is depleted from the kinetochore during chromosome alignment. Knockdown of DYNLT3 with small interference RNA increases the mitotic index, in particular, the number of cells in prophase/prometaphase. These results demonstrate that dynein binds directly to a component of the spindle checkpoint complex through the DYNLT3 light chain. Thus, DYNLT3 contributes to dynein cargo binding specificity. These data also suggest that the subpopulation of dynein, containing the DYNLT3 light chain, may be important for chromosome congression, in addition to having a role in the transport of checkpoint proteins from the kinetochore to the spindle pole.

Cytoplasmic dynein 1 is a large multisubunit ATPase that moves toward the minus ends of microtubules. It is involved in multiple important cellular processes, including mitosis, nuclear migration, Golgi and centrosome localization, organelle transport, and viral transport (1,2). This dynein complex has a molecular mass greater than 1 MDa and contains two identical heavy chains, each containing motor domains that generate the force necessary to move along the microtubule (3)(4)(5). The cytoplasmic dynein cargo binding domain consists of five subunits that assemble into the complex as dimers: the intermediate chain, light intermediate chain, and three classes of light chains; DYNLL, DYNLT, and DYNLRB. 2 There are at least two isoforms of each of these five subunits (6). The DYNLT light chain family has two members, DYNLT1 (previously called Tctex1) and DYNLT3 (previously called rp3) (2,(7)(8)(9). The two family members share ϳ55% amino acid identity. Both light chains are expressed in all the cells and tissues so far examined, and they incorporate into dynein complexes as homodimers (7,10). 3 Thus, a dynein complex contains either DYNLT1 or DYNLT3 but not both.
While the dynactin complex is responsible for binding dynein to many cargoes, there is increasing evidence that the isoforms of cargo binding subunits are important in binding dynein to specific cargoes (reviewed in Ref. 1). DYNLT1 has been shown to specifically interact with a number of proteins. For example, DYNLT1 binds to rhodopsin and Doc2, whereas DYNLT3 does not (1,11,12). The discovery of a large number of functionally unrelated proteins that bind to DYNLT1 led to the suggestion that the DYNLT light chains function as adaptors to link specific cargoes to the dynein motor (1,6,(13)(14)(15). Although this hypothesis is supported by the report that both DYNLT3 and DYNLT1 bind to the herpes simplex virus 1 caspid protein (16), to date no specific cargo for the DYNLT3 light chain has been identified. To gain insight into the cargo binding function of the DYNLT3 light chain, we performed a yeast two-hybrid screen to identify novel DYNLT3 binding proteins. One of the putative DYNLT3 interacting proteins was Bub3, a component of the spindle checkpoint complex (17). The Bub3 gene (budding uninhibited by benzimidazoles 3) was first identified in yeast (18).
The spindle checkpoint is a signal transduction pathway that blocks mitosis until all kinetochores are attached to spindle microtubules (reviewed in Ref. 19 -22). Bub3 is a conserved essential component of the checkpoint, and it is a scaffold that * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  promotes interactions between the other spindle checkpoint proteins, including Bub1, Mad2, Cdc20, and BubR1/Mad3 (17,(23)(24)(25)(26). Recently, RNA interference has been used to reduce the level of spindle checkpoint proteins in cells. These experiments have shown that some of the proteins have important roles early in mitosis, prior to their involvement in the checkpoint complex. For example, when Bub1 is depleted, cells are more likely to have misaligned chromosomes, and this defect has been linked to the role of Bub1 in maintaining centromeric cohesion (33,34). Also, RNA interference reduction of Bub3 levels leads to the accumulation of cells in prophase (24). Cytoplasmic dynein is thought to contribute to the inactivation of the spindle checkpoint by transporting checkpoint proteins from kinetochores to the spindle poles along the newly attached kinetochore microtubules (27)(28)(29)(30)(31)(32). However, the molecular basis underlying the association of checkpoint proteins with dynein was unknown.
We find that the cytoplasmic dynein light chain DYNLT3 binds to Bub3 in vivo and in vitro and that Bub3 does not bind to the related light chain DYNLT1 or any other dynein light chains. We further show that GST 4 -Bub3 binds to the dynein complex and that Bub3 and Mad2 co-immunoprecipitate with dynein from mitotic cell extracts. DYNLT3 co-localizes with Bub3 at kinetochores in prometaphase but not at later mitotic stages. siRNA knockdown of the DYNLT3 light chain results in a 3-fold increase in the mitotic index and an increase in the number of cells in prophase/prometaphase. Our data thus strongly support the hypothesis that the two DYNLT light chain family members contribute to the differential regulation of dynein cargo binding and that the interaction of Bub3 and DYNLT3 links a checkpoint protein complex to a subpopulation of dynein.

EXPERIMENTAL PROCEDURES
Plasmids-pBluescript-SK-DYNLT3 encoding mouse DYNLT3 (Open Biosystem) served as a template for the PCR amplification. To minimize auto-activation by DYNLT3 in yeast two-hybrid assays (16), 3 a flexible linker, -Ser-Gly-Gly-Ser-Gly-Gly-, was inserted in front of the open reading frame of DYNLT3 by PCR, and the product was cloned into the BamHI/ XhoI sites of the pGBKT7 vector (Clontech). For expression of the rat DYNLT1, DYNLT3, DYNLL1 (previously called LC8a), DYNLL2 (previously called LC8b), DYNLRB1 (previously called Roadblock1), and DYNLRB2 (previously called Road-block2) light chains in cultured mammalian cells, their entire coding regions were individually cloned into the pCMV-Myc vector (Clontech). Human DYNLT3-GFP was made by subcloning the DNA fragment into pEGFP vector (Clontech). An His 6 N-terminal tag was added to human DYNLT3 by cloning it into pET-14b (Novagen Inc.). The entire coding region for human Bub3 was PCR-amplified using Bub3-p3ϫFLAG-CMV (the gift of Dr. Tim Yen, Fox Chase Cancer Center, Philadel-phia, PA) as a template. The PCR product was cloned into EcoRI and SalI sites of GST pGEX4T-3 vector (Amersham Biosciences). Mouse Bub3 N-terminal fragment (amino acids 1-166) isolated from the yeast two-hybrid screen was excised from NotI sites of the Gal4 activation domain vector, pVP16, and subcloned into the GST fusion vector, pGEX4T-1 (Amersham Biosciences), GST-Bub3NT. All constructs were sequenced to confirm gene sequence and correct reading frame.
Yeast Two-hybrid Screen-DYNLT3 in pGBKT7 vector was used to screen ϳ1 ϫ 10 6 clones from a mouse embryonic cDNA library constructed in pVP16 vector (35) (the gift of Dr. Ian Macara, University of Virginia). The yeast strain (AH109) used for the screening assay contained both HIS3 and lacZ reporter genes under control of a Gal4-responsive upstream activation site. Bait and library vectors were co-transformed into AH109 and positive interactions initially identified by growth of cells on media lacking leucine, tryptophan, and histidine (referred to as -His) with 5 mM 3-AT, an inhibitor of histidine synthase, at 30°C for 3-4 days. Positive colonies obtained from the screen were confirmed by ␤-galactosidase activity assay. To minimize the number of false positives and to increase the likelihood of obtaining pure clones, the initial isolates were grown in -His media with 30 mM 3-AT. Inserts from individual positive clones were isolated by direct PCR screening using flanking primers specific for the pVP16 plasmid. All of the PCR products were sequenced, and potential DYNLT3-interacting clones were identified using NCBI Blast. The colony inhibition assay using 3-AT described previously (36) was used to confirm the specificity of the putative interactions.
Expression and Purification of Fusion Proteins-Bacterial expression of GST-recombinant proteins was carried out using Escherichia coli BL21(DE3) as host cells. To minimize the formation of inclusion bodies, protein expression continued for 10 -12 h at 18°C before harvesting by centrifugation. After lysis, soluble proteins were purified by using a glutathione-Sepharose (GSH) affinity column (Amersham Biosciences) following the manufacturer's instructions. The purified GST fusion proteins were kept on beads in phosphate-buffered saline (PBS), including protease inhibitors (described below) and were directly used for binding assay experiments. GST fusion protein concentrations were estimated by comparing the intensity of the GST fusion protein to a series of different concentration bovine serum albumin protein standards. Expression of the His-tagged DYNLT3 in bacteria was performed according to the manufacturer's instruction (Qiagen).
GST Pull-down Experiments-The GST pull-down assays between GST-Bub3 and overexpressed light chains were performed and described previously (37,38). Briefly, at least 10 g of GST or GST fusion proteins prebound to GSH beads were incubated with 100 -120 mg of the cell lysates in appropriate binding buffers. After incubation for 3 h at 4°C, the pelleted beads were washed extensively with the binding buffers and subsequently boiled with SDS-PAGE sample buffer. The proteins were first resolved by SDS-PAGE and subsequently analyzed by immunoblot analysis. Buffer A (PBS, pH 7.4, 0.1% Triton X-100, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 mM benzamidine, and 1 g/ml each leupeptin and pepstatin) was used to characterize the interaction between GST fusion proteins and the overexpressed light chains. Buffer B (50 mM PIPES, 50 mM HEPES, 1 mM EGTA, 1 mM MgCl 2 , 1 mM phenylmethylsulfonyl fluoride, 0.1 mM benzamidine, and 3 g/ml each of leupeptin and pepstatin) was used to characterize the interaction between GST fusion proteins and endogenous dynein subunits. For the direct in vitro binding assay, 10 g of GST or GST-tagged proteins prebound to GSH beads were incubated with 1 ml of bacterial lysates containing His-DYNLT3 in the buffer A. After incubation at 4°C for overnight, the beads were extensively washed with the buffer A, and the bound proteins were subsequently released by boiling with SDS-PAGE sample buffer. The released proteins were resolved by SDS-PAGE and detected on Western blots.
PAGE and Immunoblotting-SDS-PAGE and immunoblotting were performed as described previously (37,38).
Cell Culture and Transfection-COS7 (monkey kidney) and NRK (normal rat kidney) cells were grown in Dulbecco's modified essential medium containing 10% calf serum (HyClone). LLCPK (porcine kidney epithelial) cells were cultured in F-12 medium containing 10% fetal bovine serum. For transient transfection, cells were transfected with the FuGENE 6 transfection kit (Roche Applied Science) or the nucleofector (Amaxa Biosystems), following the manufacturers' instructions. Transfected cells were allowed to grow for 24 -36 h. Transfections were performed at 80% confluence.
Immunoprecipitation-A mitotic NRK cell lysate was prepared by the method of Nakagawa (39). Briefly, cells were incubated with 2 mM thymidine for 10 -12 h to arrest the cells in S phase. The cells were then arrested in mitosis by incubating with (1.7 M) nocodazole for another 12 h. The mitotic cells were harvested by a "shake-off " procedure. The cells were washed and then lysed in Buffer B by repeated passages through a needle. The mitotic extract was recovered by ultracentrifugation of the lysed cells. To obtain an interphase extract, unsynchronized NRK cells were harvested by scraping. The lysates were clarified by centrifugation and then used for immunoprecipitation with antibodies, as indicated, and protein A-Sepharose beads (Zymed Laboratories Inc.). After extensively washing the beads with Buffer B, the precipitated proteins were resolved by SDS-PAGE and detected by Western blotting with the indicated antibodies.

RESULTS
Cytoplasmic Dynein Light Chain, DYNLT3, Interacts with the Spindle Checkpoint Protein, Bub3-In preliminary experiments using DYNLT3 in pairwise yeast two-hybrid assays it was found that DYNLT3 fused to the DNA binding domain was a strong auto-activator of histidine synthase (16). 3 Therefore, to use a yeast two-hybrid library screen to identify DYNLT3 binding partners, it was first necessary to engineer a construct that minimized the auto-activation. We found that adding a flexible linker -Ser-Gly-Gly-Ser-Gly-Gly-between the DNA binding domain and DYNLT3 reduced the auto-activation to very low levels (data not shown). The screening of a mouse embryonic library resulted in the isolation of 99 independent clones. 51 clones from the primary isolate were identified as strong interactors with the ␤-galactosidase activity assay. Of these, 27 were able to grow in -His medium with high concentration of 3-AT (30 mM). Sequence analysis of these 27 clones indicated that 9 were membrane associated proteins, 6 were novel proteins, and 11 were enzymes/structural proteins. DYNLT3 was also identified in the screen consistent with the finding that it forms homodimers (10). One candidate DYNLT3 binding partner identified in the yeast two-hybrid library screen was the N-terminal portion of mouse Bub3 (amino acids 1-166). The strong specificity of the interaction of the two proteins was confirmed by using the full-length human Bub3 in a pair-wise yeast twohybrid screen with DYNLT3 (Fig. 1A)  assay (36). The specificity was further confirmed by assaying for the independent expression of ␤-galactosidase in the transformants (Fig. 1B).
To demonstrate that the interaction between Bub3 and DYNLT3 observed in the yeast two-hybrid assay was not mediated by a third yeast protein, the ability of bacterially expressed DYNLT3 and Bub3 to interact was examined. GST-Bub3 was purified from a bacterial expression system and mixed in vitro with His-tagged DYNLT3 also from a bacterial expression system. As can be seen in Fig. 1C, the His-DYNLT3 binds to and co-pellets with GST-Bub3 demonstrating that the purified proteins interact directly. We next sought to determine if Bub3 interacted with the other DYNLT family member, DYNLT1. DYNLT1 and DYNLT3 were tagged with the Myc epitope and individually expressed in COS7 cells. The cells were lysed and incubated with GST-Bub3 bound to GSH beads in a pull-down assay. The data presented in Fig. 2 show that Bub3 bound exclusively to DYNLT3. The same approach was used to show that Bub3 does not interact with any of the other dynein light chains, further demonstrating the specificity of the interaction between DYNLT3 and Bub3 (Fig. 2).
Bub3 and Mad2 Bind to the Cytoplasmic Dynein Complex-To demonstrate that the DYNLT3 light chain is a link between Bub3 and the dynein complex, we employed two approaches. GST-Bub3 and GST were incubated with lysates from mitotic normal rat kidney (NRK) cells, and the bound proteins were analyzed by SDS-PAGE and Western blot (Fig. 3A). As expected, DYNLT3 from the mitotic extract bound to the Bub3 beads but not the GST beads. The dynein intermediate and heavy chains were also bound to the GST-Bub3 beads. The intermediate chain binds both DYNLT3 and the heavy chain. Therefore, the presence of both the intermediate chain and heavy chain in the GST pull-down assays is diagnostic for the presence of an intact dynein complex. This demonstrates that the dynein complex from mitotic cell extracts is capable of interacting with Bub3. To further confirm interaction of dynein with Bub3, an antibody to the intermediate chain was used to immunoprecipitate the entire dynein complex (40,45). Bub3 from mitotic cell lysates was present in the dynein immunoprecipitate (Fig.  3B) further supporting the conclusion that Bub3 interacts with the dynein complex. Bub3 also forms a complex with other checkpoint proteins, including Mad2. We also found that Mad2 co-immunoprecipitated with dynein from mitotic cell extracts (Fig. 3B). To test whether the interaction of DYNLT3 and Bub3 was specific for mitotic cells, we analyzed dynein immunoprecipitated from unsynchronized, NRK FIGURE 1. The cytoplasmic dynein light chain, DYNLT3, binds to the spindle checkpoint protein Bub3. A, yeast two-hybrid assay. Yeast transformants expressing the indicated constructs were grown on plates with (ϩHis) or without histidine (ϪHis) and the indicated concentrations of the histidine biosynthesis inhibitor 3-AT. Only those co-transformants expressing interacting constructs were able to grow in the absence of histidine and in the presence of 3-AT. DYNLT3 and human Bub3 are in the pGBKT7 and pGADT7 vectors, respectively. B, ␤-galactosidase expression assay. The positive interaction of DYNLT3 and Bub3 was confirmed with a semi-quantitative ␤-galactosidase assay: ϩϩϩ, substantial growth; ϩ, marginal growth; and (Ϫ), no growth. C, in vitro binding assay. Bacterial lysates expressing His-DYNLT3 were incubated in a pull-down assay with either GST-Bub3NT or GST, bound to GSH beads. Right panel, His-DYNLT3 bound to GST-Bub3NT beads but not to GST control beads. His-DYNLT3 was detected with Western blots using anti-His antibody. The input lane shows 2% of the His-DYNLT3 used in the pull-down assays (1.5 ml per reaction), and the experimental lanes were loaded with 50% of the pellets.  cells, most of which are in interphase. The data in Fig. 3B demonstrates that Bub3 was not found in the dynein immunoprecipitate.
DYNLT3 Co-localizes with Bub3 at Prometaphase Kinetochores-The fact that Bub3 binds to DYNLT3 and not to DYNLT1 suggested that DYNLT3 serves as the link between a subpopulation of cytoplasmic dynein and a checkpoint protein complex. We therefore analyzed the distribution of DYNLT3-GFP in cultured cells to determine if its location was consistent with this role. During interphase, the DYNLT3-GFP fluorescence signal was distributed throughout the cytoplasm similar to the cytoplasmic distribution observed when cells are transfected with a GFP-tagged dynein intermediate chain (not shown, and see Ref. 46). During prometaphase, DYNLT3 was observed at kinetochores and co-localized with Bub3 (Fig. 4A). We were unable to identify any DYNLT3 light chain associated with kinetochores of metaphase, anaphase, or telophase cells (Fig. 4B). Rather, DYNLT3 was found along astral and spindle  microtubules (Fig. 4, C and D). Identical results were obtained regardless of the cell line used (NRK) or the use of Myc and HA epitope tags (data not shown). We were also unable to detect Bub3 on kinetochores after prometaphase (Fig. 4D). These observations indicate that the dynein variant with the DYNLT3 light chain associates with kinetochores during prometaphase but not during other phases of the cell cycle, and they are consistent with the hypothesis that the subpopulation of dynein with DYNLT3 is moving off of the kinetochore prior to metaphase.
Knockdown of Endogenous DYNLT3 with siRNA Slows Progress through Mitosis-To determine if the dynein variant with the DYNLT3 light chain has a role in mitosis consistent with disruption of checkpoint protein function, we used siRNA to reduce the expression of this light chain in NRK cells. We confirmed with reverse transcription-PCR that both DYNLT1 and DYNLT3 are expressed in this cell line (not shown). The efficient depletion of DYNLT3 protein by treatment of cells with siRNA oligonucleotides, compared with the control scrambled oligonucleotide, was demonstrated by Western immunoblot analysis (Fig. 5A). DYNLT1 remains in the cells after DYNLT3 siRNA (data not shown). Therefore, only DYNLT3 is depleted by this procedure. We then compared the relative number of mitotic cells in siRNA-treated and control cells. siRNA knockdown of DYNLT3 resulted in a 3-fold increase in the mitotic index (Fig.  5B, p Ͻ 0.002, Student's t test) and an almost 5-fold increase in the number of prophase/prometaphase cells relative to controls (Fig. 5C, p Ͻ 0.0002, Student's t test). Interestingly, we observed no significant increase in the number of abnormal spindles in siRNA compared with control cells. These results indicate that dynein with DYNLT3 is necessary for efficient progression through mitosis and suggest that it has a role in an early stage of mitosis.

DISCUSSION
It has been proposed that the two members of the DYNLT light chain family, DYNLT1 and DYNLT3, are important for binding dynein to specific cargoes (1,7,10). Supporting this model were two observations: that both light chains were expressed in all tissues and cultured cells examined and that a wide range of proteins bound to DYNLT1, including rhodopsin (11), voltage-gated calcium channel (47), Doc2␤ (12), heat shock protein PBP74 (48), and CD155, the polio virus receptor (49). The major weakness of the model was that no specific cargo for DYNLT3 had been identified. In this study, we demonstrate that DYNLT3 links cytoplasmic dynein directly to Bub3, a spindle checkpoint protein. This conclusion is supported by a yeast two-hybrid pair-wise interaction assay and an in vitro binding assay. The interaction of bacterially expressed Bub3 and DYNLT3 indicates that the binding of the two proteins is direct rather than through a third polypeptide. GST-Bub3 also bound to DYNLT3 overexpressed in cultured cells, not DYNLT1 or the other four dynein light chains. Furthermore, Bub3 and its binding partner Mad2 co-immunoprecipitated with cytoplasmic dynein from mitotic cell extracts, and dynein from mitotic cell extracts co-pelleted with GST-Bub3. Bub3 did not co-immunoprecipitate with dynein from interphase extracts. We also found that DYNLT3, like Bub3, is enriched on mammalian prometaphase kinetochores, but not metaphase or anaphase kinetochores. Depletion of DYNLT3 with siRNA increased the mitotic index, with an accumulation of cells in prophase/ prometaphase. Our data demonstrates that dynein with DYNLT3 has functional importance for progression through mitosis. Our data also strongly support the model for dynein binding to different cargoes though the two DYNLT light chains.
The knockdown of DYNLT3 levels with siRNA suggests that the dynein subpopulation with DYNLT3 has a role in prophase/ pro-metaphase, and possibly in chromosome congression. Interestingly, it has recently been observed that RNA interference knockdown of Bub3 leads to an accumulation of cells in prophase (24). Misaligned chromosomes are also observed, suggesting that that Bub3 may be involved in kinetochore-microtubule attachment (33,34). 6 Kinetochore dynein is thought to be responsible for moving newly attached chromosomes along astral microtubules to the spindle poles (50). But the two sets of RNA interference knockdown data suggest that dynein, together with Bub3, may also be involved in the formation of stable bipolar kinetochore attachments to microtubules and congression. An increase in the number of cells in metaphase was not observed after DYNLT3 depletion. This may be because the accumulation of cells in prometaphase masked putative effects on metaphase. GFP-Mad2 has been observed moving from kinetochores along microtubules to spindles poles in a dynein dependent manner (27). Our immunocytochemical data and the finding that Bub3, Mad2, and dynein co-immunoprecipitate may provide a molecular explanation for dynein-mediated checkpoint protein translocation from the kinetochore upon microtubule attachment. Bub3 and DYNLT3 could function as scaffolds directly linking dynein to the checkpoint protein complex for kinetochore to pole transport.
Cytoplasmic dynein participates in several other stages of the mitotic process. Dynein is important for spindle assembly (51). In Drosophila it is involved in chromosome to pole movement during anaphase A (52). Dynein in the cell cortex is positioned to orient the spindle and to pull the spindle poles apart in anaphase B (53)(54)(55). Our immunocytochemical and siRNA data suggest that different subpopulations of dynein may be responsible for some of these roles. The dynein variant with DYNLT3 does not appear to be involved in spindle assembly, because we observed no difference in the number of abnormal spindles after siRNA reduction of DYNLT3 levels. This is in contrast to the substantial increase in the number of abnormal spindles observed when dynein is inhibited by overexpression of either the p50 (dynamitin) subunit of dynactin or the CC1 fragment of the p150 subunit of dynactin (56,57). We were also unable to detect any DYNLT3 on kinetochores after prometaphase. This suggests that dynein with DYNLT3 is not on kinetochores and so would not participate in chromosome to pole movement.
In conclusion, we have shown that the cytoplasmic dynein light chain DYNLT3 and Bub3 co-localize on prometaphase kinetochores; that DYNLT3, binds to the checkpoint protein Bub3 in vitro; and that GST-pull-down and co-immunoprecipitation assays demonstrate the interaction between Bub3 and Mad2 with cytoplasmic dynein in mitotic cell extracts. These data support models by which DYNLT3 links cytoplasmic dynein to Bub3 containing spindle checkpoint complexes and contribute to checkpoint complex function during congression and/or during the inactivation of the checkpoint by transporting the complexes from kinetochores to spindle poles.