PKD2 Interacts and Co-localizes with mDia1 to Mitotic Spindles of Dividing Cells ROLE OF mDia1 IN PKD2 LOCALIZATION TO MITOTIC SPINDLES*

Mutations in pkd2 result in the type 2 form of autosomal dominant polycystic kidney disease, which ac-counts for (cid:1) 15% of all cases of the disease. PKD2, the protein product of pkd2 , belongs to the transient receptor potential superfamily of cation channels, and it can function as a mechanosensitive channel in the primary cilium of kidney cells, an intracellular Ca 2 (cid:1) release channel in the endoplasmic reticulum, and/or a nonselective cation channel in the plasma membrane. We have identified mDia1/Drf1 (mammalian Diaphanous or Di-aphanous-related formin 1 protein) as a PKD2-interact-ing protein by yeast two-hybrid screen. mDia1 is a mem-ber of the RhoA GTPase-binding formin homology protein family that participates in cytoskeletal organization, cytokinesis, and signal transduction. We show that mDia1 and PKD2 interact in native and in transfected cells, and binding is mediated by the cytoplasmic C terminus of PKD2 binding to the mDia1 N terminus. The interaction is more prevalent in dividing cells in which endogenous PKD2 and mDia1 co-localize to the mitotic spindles. RNA interference experiments reveal that endogenous mDia1 knockdown in HeLa cells results in the loss of PKD2 from mitotic spindles and alters intracellular Ca 2 (cid:1) release. that mDia1 facilitates the movement of PKD2 to a centralized position during cell division and has a positive effect on intracellular Ca 2 (cid:1) release during mitosis. may be important to ensure equal segregation of to the daughter cell to maintain a necessary level of channel activity. PKD2 channel activity may be important in the cell division process or in cell fate after division.

Autosomal dominant polycystic kidney disease affects 1 in 1,000 individuals primarily by the development of large, fluidfilled renal cysts that ultimately may lead to renal failure. Extrarenal manifestations such as cyst formation in the liver, pancreas, and spleen may also occur with autosomal dominant polycystic kidney disease as well as cranial aneurysms and secondary hypertension (1,2). Positional cloning identified pkd2 as one of the genes mutated in ϳ15% of affected families (3)(4)(5). Polycystin-2 (PKD2) has significant sequence homology to the transient receptor potential channel proteins (6). PKD2 was found on male-specific sensory neurocilia in Caenorhabditis elegans (7) and later in the primary cilia of kidney epithelial cells (8,9). The apparent function of nonmotile cilia is to act as a sensor of the extracellular environment and transmit this information to the cell body. Consistent with this idea, PKD2 has been shown to have an essential role in mediating Ca 2ϩ entry in response to flow rate changes, suggesting that it may be part of the mechanosensing machinery residing in the primary cilium of terminally differentiated epithelial cells (10). However, in addition to its expression in ciliary structures, PKD2 expression has been reported in the endoplasmic reticulum of LLC-PK1 cells (11), in the plasma membrane of Madin-Darby canine kidney and mouse inner medullary collecting duct (IMCD) 1 cells (12,13), and in apical membranes of human term syncytiotrophoblasts (14). The pool of PKD2 expressed in the endoplasmic reticulum is believed to function as an intracellular Ca 2ϩ release channel, whereas plasma membrane-anchored PKD2 functions as a nonselective cation channel. Whereas all of these functions are likely to represent physiological functions, its role as a mechanosensor is more likely to be associated with the cystic phenotype seen in pathogenic mutants of PKD2. This idea is supported by recent findings that loss-of-function mutations in proteins required for ciliary formation and function often result in cystic diseases (8,7,(15)(16)(17). Alternatively, PKD2 may have a role in cell division during development regulating the differentiation process of kidney tubular cells. Loss of PKD2 in pathogenic mutants may account for the less differentiated phenotype of cystic epithelial cells, thus indirectly affecting ciliary function. Ciliary function of PKD2 in terminally differentiated cells may be more closely related to disease progression (18,19). These findings highlight the fact that PKD2 is a multifunctional protein with several possible roles in cell physiology and kidney development; the exact mechanisms regulating these apparently divergent cellular functions and subcellular localizations are largely unknown. mDia1 (mammalian diaphanous 1) was originally identified as a RhoA-interacting protein in a yeast two-hybrid screen using activated RhoA as bait (20), and it was later shown to function as a downstream effector of RhoA in signal transduction and cytoskeletal organization (21)(22)(23). mDia1 is also referred to as Drf1 (diaphanous-related formin 1) because it has homology to formin, the limb deformity (ld) gene product. Formin has a proline-rich domain and a 130-amino acid region that have been named FH1 and FH2 domains, respectively (24). A third region, the FH3 domain, is an amino-terminal domain that is the least conserved FH domain among formin family members (25). Diaphanous-related formin homology proteins such as mDia1, mDia2, and yeast Bni1p contain all three FH (24) as well as an amino terminus GTPase binding domain (RhoA binding domain) (20) and a carboxyl terminus interaction domain termed Dia autoregulatory domain (DAD domain) (26). The DAD domain loops around to bind the N terminus of mDia in the RhoA GTPase binding domain to hold mDia1 in a closed and inactive state. When Rho protein binds mDia1, the DAD domain releases the N terminus to form an active mDia1 to expose its internal domains for interactions with effector molecules (27,26). The FH1 domain binds profilin, an actin-binding protein (20), and Src homology 3 domains containing proteins such as Src (28) and IRSp53/BAIAP2 (29). The FH2 domain has been reported to be involved in coordinating microtubules and the actin cytoskeleton during cell movement (30). mDia1 contains a 173-amino acid region in the C terminus of its FH3 domain that is necessary for mDia1 localization to mitotic spindles in HeLa cells (31). A similar FH3 domain function has been seen in Schizosaccharomyces pombe when FH3 domains from Fus1 or Cdc12 formin-related proteins were fused to GFP, and GFP was subsequently localized to the projection tip of mating cells. These results suggest that FH3 domains may serve to specifically localize forminrelated proteins to specific subcellular sites (25). The diaphanous subfamily appears to primarily function in cytokinesis and in cytoskeletal rearrangement and stabilization. Consistent with its role in actin polymerization in hair cells of the ear, naturally occurring mutations in mDia1 cause nonsyndromic deafness in humans (32). mDia1 was also recently found to be essential for mechanotransduction in response to integrin activation in SV-80 human fibroblast cells (33) but is not involved in the regulation of integrin-mediated cell adhesion in Jurkat cells (34). In conclusion, mDia1 appears to serve several basic cellular functions ranging from mechanosensation to cytokinesis through a possible effect on cytoskeletal organization.
In the present study, we identified and characterized an interaction of PKD2 and mDia1 using a yeast two-hybrid screen and co-immunoprecipitation experiments. The interaction was further supported by indirect immunofluorescence and confocal microscopy of the endogenous proteins in mitotic cells. RNAi experiments in mitotic HeLa cells have shown a concomitant loss of PKD2 from the mitotic spindle in cells lacking mDia1, and we have demonstrated diminished intracellular Ca 2ϩ release in these cells. Our results suggest that a novel function of mDia1 is to bind the PKD2 channel protein possibly contained in an ER vesicle, facilitate its movement to mitotic spindles during cell division, and modulate intracellular Ca 2ϩ release during mitosis.
Yeast Two-hybrid Screen-The L40 yeast strain containing both his and lacZ reporters under the control of LexA binding sites was sequentially transformed with a bait plasmid in pLexA containing a portion of the C-terminal cytoplasmic region of human PKD2 (Ser 851 -Val 968 ; accession number NM_00297) and a whole mouse day 9.5 embryonic library in pVP16. From 4 ϫ 10 6 independent clones screened, several clones were identified that contained inserts interacting with PKD2 Ser 851 -Val 968 . All inserts were in-frame with the DNA binding domain of LexA, and their interaction with PKD2 was further verified by reintroducing the rescued plasmids back to the L40 yeast strain that was originally used in the screen.
Cell Culture-LLC-PK1 cells were maintained in M199 medium, HeLa cells were maintained in RPMI 1640, HEK293T cells were grown in Dulbecco's modified Eagle's medium, and Madin-Darby canine kidney and IMCD cells were grown in Dulbecco's modified Eagle's medium/ F-12 medium. All media were supplemented with 10% fetal bovine serum except LLC-PK1, which required 3% fetal bovine serum. All cell culture media were from Invitrogen. Cells were cultured at 37°C in a 5% CO 2 atmosphere. Cells used for confocal microscopy were plated onto 12-mm glass coverslips in 24-well plates and maintained as above. HeLa cell coverslips were coated with CellTak (BD Biosciences) per manufacturer's instructions for some confocal microscopy experiments. Cells grown for cilia staining were allowed to grow until confluent.
Purified antibody titers were determined by ELISA using purified MBP-PKD2(Ile 679 -Arg 742 ). The specificity of the antibodies was evaluated by immunoblotting lysates of transfected HEK293T cells with an expression plasmid for HA-tagged wild type PKD2 (HA-PKD2) or a truncation mutant lacking the antigenic region (HA-PKD2-(1-643)).
Transient Transfection Procedures-HEK293T cells were transfected by a standard calcium transfection protocol as described previously (6). HeLa cells were transfected with Myc-PKD2 or GFP-mDia1 by Effectene (Qiagen), according to the manufacturer's instructions.
Mitotic Spindle Preparations-Mitotic spindles from HeLa cells transfected with Myc-PKD2 or GFP mDia1 were isolated as previously reported (31). Isolated microtubules from each 15-cm dish were resuspended in 200 l of Laemmli sample buffer, and 30 l was run on SDS-PAGE, subjected to Western blot, and probed with a polyclonal ␣-Myc antibody at 1:500 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA), polyclonal antibody to GFP at 1:500 (Santa Cruz Biotechnology), or a monoclonal anti-␤-tubulin antibody at 1:5,000 (Sigma; Clone 2.1). Horseradish peroxidaseconjugated secondary antibodies were used at 1:10,000 followed by enhanced chemiluminescence detection (Pierce Supersignal).
Immunoprecipitation, in Vitro Binding, and Western Blotting-Cells were lysed in 1 ml/10-cm dish of 1% Triton X-100, 150 mM NaCl, 10 mM Tris-HCl, pH 7.5, 1 mM EGTA, 1 mM EDTA, 10% sucrose, and Complete protease inhibitors (Roche Applied Science). Lysates were cleared by centrifugation at 12,000 ϫ g for 20 min, and protein assay (BCA; Pierce) was done on the supernatants. Co-immunoprecipitation of PKD2 and mDia1 (Transduction Laboratories) was performed in the HEK293T cleared lysate from a 15-cm dish, whereas all other co-immunoprecipitations were from a volume less than or equal to a single 10-cm dish of cleared cell lysate. GST isotype control antibody was from Santa Cruz. After an 8-h incubation with primary antibody, 30 l of 50% slurry of Protein A-or G-agarose beads was added for 2 h to capture immune complexes. The beads were washed five times with lysis buffer and then mixed with Laemmli buffer for SDS-PAGE. For immunoblotting, mDia1 antibody and rabbit affinity-purified PKD2 antibody were both used at 1:500. Horseradish peroxidase-conjugated secondary antibodies were used at 1:10,000 followed by enhanced chemiluminescent detection after Western blotting. Immunoprecipitations using IgG-tagged constructs (sIg.7, sIg.7-PKD2/Gln 743 -Glu 871 , or sIg.7-PKD2/Ile 679 -Val 968 ) were done as described earlier (6,33). GFP-tagged mDia1 fusion proteins were detected with a rabbit polyclonal antibody against GFP at 1:500.
RNA Interference-To inactivate pig PKD2 in LLC-PK1 cells, we designed two PKD2-specific constructs, PKD2 KD2-2 and PKD2 KD2-3 . PK-D2 KD2-2 targeted specifically porcine PKD2, whereas PKD2 KD2-3 targeted both porcine and human PKD2. These constructs were made in our previously described RNAi vector, pUB/H1RNAi vector (36). Porcine PKD2 cDNA sequences were determined based on a partial pig expressed sequence tag clone (GenBank TM number BI338275). The PK-D2 KD2-2 sequence was 5Ј-gatcccccctgttctgtgtgatcaggttcaagagacctgtcacacagaacaggtttttggaaa-3Ј, whereas the PKD2 KD2-3 sequence was 5Ј-gatccccccgcttggatctacacaagttcaagagacttgtgtagatccaagcggtttggaa-3Ј. Both sequences represent the sense strand of the double-stranded DNA cloned into pUB/H1RNAi. LLC-PK1 cells were transfected with 1.9 g of either silencing construct, along with 0.1 g of GFP-tubulin to indicate cells that had taken up the DNA complex and to also identify dividing cells. After 48 h, these cells were replated to coverslips for confocal microscopy.
To silence human mDia1, we generated two inactivating constructs, pSUPER retro-mDia1 KD1 and pSUPER-retro-mDia1 KD2 , in the pSUPER-retro vector (Oligoengine, Inc.). The pSUPER-retro-mDia1 KD1 construct was designed to silence specifically human mDia1, whereas pSUPER-retro-mDia1 KD2 was designed to silence both human and mouse mDia1. Neither of these constructs would silence mDia2 or homologous genes. The mDia1 KD1 forward primer was 5Ј-atccccgcatgagatcattcgctgcttcaagagagcagcgaatgatctcatgctttttggaaa-3Ј, whereas the mDia1 KD2 forward primer was 5Ј-gatccccaggcagagccacacttcctttcaagagaaggaagtgtggctctgccttttttggaaa-3Ј. Specific oligomers were cloned into pSUPER-retro plasmid to generate recombinant retroviruses directing the expression of short hairpin RNAs. Vesicular stomatitis virus-G pseudotyped retroviruses were made in the HEK293T cell line following transient co-transfection with vesicular stomatitis virus-G, gag-pol, and pSUPER-retro-mDia1 KD1 or pSUPER-retro-mDia1 KD2 plasmids. After 48 h, retroviral supernatants of transfected HEK293T cells were diluted 1:1 with fresh media and incubated with HeLa cells in the presence of 4 g/ml polybrene. After 12 h, fresh medium was added, and cells were allowed to recover for 24 h. Pools of infected cells were selected in the presence of 1 g/ml puromycin. A control vector for nonspecific silencing effects was constructed in the pSUPER-retro vector directing the expression of a mouse I-mfa-specific short hairpin RNA (5Ј-gatccccgttgcagacgcatccatctttcaagagaagatggatgcgtctgcaactttttggaaa-3Ј). Recombinant virus containing this short hairpin RNA was generated and used to infect HeLa cells as described above.
Confocal Microscopy-Prior to fixation, cells plated on coverslips were washed in 37°C cytoskeletal stabilizing buffer (60 mM PIPES, 25 mM HEPES, 10 mM EGTA, 2 mM MgCl 2 , adjusted to pH 6.9 with KOH) followed by fixation with 3.7% formaldehyde in cytoskeletal stabilizing buffer (PHEM buffer). Fixation was at room temperature for 15 min followed by three rinses in PHEM buffer. The cells were extracted with 0.1% Triton X-100 in PHEM buffer for 90 s. Blocking was overnight at 4°C or room temperature for 1 h in PHEM buffer containing 0.1% Triton X-100 and 10% normal goat serum. Primary antibodies were diluted in PHEM containing 0.1% Triton X-100 and 10% normal goat serum for overnight incubation. Rabbit affinity-purified antibody to PKD2 was used at 1:100, mouse antibody to ␣-tubulin (DM1A; Sigma) was used at 1:2,000, mouse antibody to mDia1 (Transduction Laboratories) was used at 1:100, and mouse anti-acetylated tubulin (6-11B-1; Sigma) was used at 1:1,000 for cilia staining. The fluorescence-tagged secondary antibodies (goat anti-rabbit Alexa 488, goat anti-mouse Alexa 568, and goat anti-chicken Alexa 488; Molecular Probes, Inc., Eugene, OR) were applied for 1 h at 1:2,000 in phosphate-buffered saline containing 0.1% Triton X-100 and 10% normal goat serum, followed by five washes in phosphate-buffered saline. ToPro3 (Molecular Probes) was included in the next to the last wash at 2 M for 5 min to visualize DNA. Coverslips were mounted with ProLong (Molecular Probes). ToPro3 was excited with a HeNe633 laser, and antibody staining was visualized with laser excitation from argon-488 and krypton-568 lasers and viewed on a Leica TCS NT Microscope. Images were processed with Leica Lite software.
Ca 2ϩ Fluorescence Spectroscopy-HeLa cells in 10-cm dishes were treated with 40 ng/ml nocodazole for 12 h prior to mitotic shake off. Mitotic cells were harvested and washed twice in Ca 2ϩ loading buffer (140 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 1.8 mM CaCl 2 , 10 mM glucose, 0.1% bovine serum albumin, 15 mM HEPES, 0.023% pluronic acid) containing 40 ng/ml nocodazole. Cells were loaded with 2 M Indo AM (Molecular Probes) in Ca 2ϩ buffer with nocodazole for 60 min at 37°C in a 5% CO 2 atmosphere. The cells were again washed twice in Ca 2ϩ buffer without nocodazole and then resuspended in 2 ml of Ca 2ϩ loading buffer for spectroscopy. Normal HeLa cells were handled as above, except that nocodazole was omitted in every step, and they were harvested with 1ϫ trypsin (Invitrogen). Ca 2ϩ spectroscopy measurements for HeLa cells done in the absence of extracellular [Ca 2ϩ ] o were performed on cells washed and assayed in the loading buffer described above except that CaCl 2 was omitted and 0.5 M EGTA was added.
Ca 2ϩ release was assayed by adding 10 M histamine to each 2-ml cell suspension and recording fluorescence ratios with a Photon Technology International fluorescence spectrophotometer. Excitation was at 350 nm, and emission data were collected at 405 and 485 nm and analyzed with Felix software.

Identification of mDia1 as an Interacting
Partner of PKD2-To better understand the function and regulation of PKD2 we employed a yeast two-hybrid screen using human PKD2 C-terminal residues Ser 851 to Val 968 as bait. This region of PKD2 was chosen because it would enable us to identify interacting proteins with the region of PKD2 that is mostly deleted in the longest truncation pathogenic mutant described to date, PKD2(871X) (37). PKD2(Ser 851 -Val 968 ) was cloned into the pLexA vector and used to screen a mouse embryonic day 9.5 library fused to the activation domain of VP16. Several clones that contained inserts interacting with PKD2(Ser 851 -Val 968 ) were identified from the 4 ϫ 10 6 independent clones screened. All inserts were in-frame with the DNA binding domain of LexA, and their interaction with PKD2 was further verified by reintroducing the rescued plasmids back to the L40 yeast strain originally used in the screen. One of the clones matched amino acids Leu 143 to Leu 260 of mouse p140 Diaphanous (p140mDia, mDia1, accession number NM_007858). The PKD2-interacting region of mDia1 lies in the C-terminal portion of an already described functional domain of mDia1 that is called the RhoA binding domain and partially overlaps with the N terminus of the mDia1 FH3 domain.
Production and Characterization of PKD2-specific Antibodies-To confirm the interaction between PKD2 and mDia1, we generated polyclonal antibodies to PKD2 and used a commercially available monoclonal antibody to mDia1. We selected a region of PKD2 composed of amino acids Ile 679 -Arg 742 as antigen for the production of polyclonal antibodies in chicken and rabbit. Rabbit or chicken immune serum was affinity-purified by passage over a column of immobilized MBP-PKD2-(679 -742). Antibody characterization by immunoblotting is shown in Fig. 1. Specifically, we have expressed HA-PKD2 that should be recognized by the polyclonal antibodies and a truncated form, HA-PKD2-(1-643), which does not contain the antigenic site. Cell lysates containing the tagged proteins were analyzed by Western blot followed by detection with chicken anti-PKD2 (Fig. 1A) or rabbit anti-PKD2 (Fig. 1B). In both cases, the antibody did not recognize HA-PKD2-(1-643) (Fig. 1, A and B, lane 1) but did recognize a 110-kDa protein (Fig. 1, A and B, lane 2), which corresponds to the expected molecular weight of HA-PKD2. The rabbit antibody also recognized an smaller molecular weight protein to a much lesser extent (Fig. 1B, lane 2), and we would suggest that this is a proteolytic fragment of the larger HA-PKD2 molecule because it was not detected in HA-PKD2-(1-643)-transfected lysates (Fig. 1B, lane 1). Relatively equal expression of HA-PKD2-(1-643) and HA-PKD2 proteins in the same lysates is shown in Fig. 1, C and D. The specificity of the rabbit affinity-purified antibody to PKD2 was further confirmed by indirect immunofluorescence staining of IMCD (Fig. 1E) and LLC-PK1 (Fig. 1F) cells, where PKD2 is known to co-localize with stabilized tubulin in ciliated cells. These results indicate that we have generated two polyclonal antibodies suitable for PKD2 detection by Western blotting.
In Vivo Interaction of PKD2 and mDia1 in Native Cells-To determine whether native PKD2 and mDia1 are associated in mammalian cells, we performed co-immunoprecipitation experiments in native human embryonic kidney cells (HEK293 cells) because they have been shown to express a significant amount of endogenous PKD2 (37), and mDia1 was expected to be present in HEK293 cells due to its ubiquitous expression in mammalian tissues (20). Fig. 2A shows that immunoprecipitation of mDia1 with a commercially available monoclonal antibody to mDia1 immunoprecipitated PKD2 from HEK293 cell lysates ( Fig. 2A, lane 1). PKD2 was not immunoprecipitated when an isotype-matched monoclonal antibody (anti-GST) was used as a mDia1-dependent Localization of PKD2 to Mitotic Spindles control ( Fig. 2A, lane 2). The specificity of our antibody for endogenous PKD2 in HEK293 cell lysates is evident in Fig. 2A, lane 3, where it is shown to detect a single band with a molecular mass of ϳ110 kDa. We also confirmed the immunoprecipitation of endogenous mDia1 with the mDia1 antibody (Fig. 2B, lane 1) and its expression in HEK293 lysates ( Fig. 2A, lane 3) by stripping the membrane shown in Fig. 2A and reprobing for mDia1. These results confirmed the yeast two-hybrid results and further provided in vivo evidence that mDia1 is an interacting protein with PKD2, suggesting that the two proteins exist in a native complex in HEK293 cells. Although we were able to sufficiently immunoprecipitate endogenous mDia1 to detect an interaction with PKD2, the commercial mDia1 antibody is not recommended for immunoprecipitation, and this technical limitation required a large amount of HEK293 cell lysate for our immunoprecipitation experiments. This was also evident by the small amount of immunoprecipitated mDia1 (Fig. 2B, lane 1). In light of this, we have used our PKD2 antibody as an immunoprecipitating antibody in Nalm 6 (pre-B cell line) and in mIMCD3 (mouse kidney inner medullary collecting duct) lysates. Endogenous mDia1 was detected in these immunoprecipitants (data not shown), indicating that the mDia1-PKD2 interaction is common to more than one cell type.
In Vitro Interaction between PKD2 and mDia1-To determine whether the C-terminal cytoplasmic tail of PKD2 was sufficient to mediate the interaction with mDia1 and to test whether the longest truncation pathogenic mutant of PKD2 could still interact with mDia1, we IgG-tagged the entire Cterminal cytoplasmic tail (sIg.7-PKD2/Ile 679 -Val 968 ) or a portion of the C-terminal tail of human PKD2 terminated at residue Glu 871 (sIg.7-PKD2/Gln 743 -Glu 871 ). It should also be noted that these constructs were designed to be forwarded to the secretory pathway and eventually inserted in the plasma membrane by the addition of the leader sequence of CD45 and the transmembrane segment of CD7 (6). Next, we co-transfected HEK293T cells with control sIg.7, sIg.7-PKD2/Gln 743 -Glu 871 , or sIg.7-PKD2/Ile 679 -Val 968 with full-length GFP-tagged mDia1 (GFP-mDia1) or an N-terminal deletion mutant of GFP-mDia1 (GFP-⌬N3mDia1) that lacks the PKD2 binding site based on our yeast two-hybrid results. Protein A was used to capture the IgG fusions and their interacting proteins from transfected cell lysates. The results shown in Fig. 2C are a Western blot of the Protein A pull-downs using an antibody to GFP to detect the presence of GFP-mDia1 or GFP-⌬N3mDia1. Fig. 2C, lanes 1 and 2, serve as controls to indicate that nonspecific binding between sIg.7 control and GFP-mDia1 or GFP-⌬N3mDia1 did not occur; nor was there specific binding between GFP-mDia1 or GFP-⌬N3mDia1 with the sIg.7-PKD2/ Gln 743 -Glu 871 (lanes 4 and 6). Specific binding occurred when sIg.7-PKD2/Ile 679 -Val 968 and GFP-mDia1 were present to- were stained with rabbit affinity-purified antibody to PKD2 (green, left) and mouse (Ms) anti-acetylated tubulin (red, middle), indicating co-localization of these proteins on cilia (yellow; an arrow designates one of several cilia) in the merged images on the right. Primary antibodies were detected with secondary antibody to goat anti-rabbit and goat anti-mouse conjugated to Alexa 488 (green) or Alexa 568 (red).

mDia1-dependent Localization of PKD2 to Mitotic Spindles
gether in the lysate (Fig. 2C, lane 3), with the middle band being at the expected molecular weight for GFP-mDia1. We also observed a band of GFP reactivity in lane 5, indicating that GFP-⌬N3mDia1 may associate with sIg.7-PKD2/Ile 679 -Val 968 . This could occur via a homotypic interaction between GFP-⌬N3mDia1 with unlabeled, endogenous mDia1, since mDia1 has been reported to form multimers in vitro when an mDia1 N terminus construct was shown to bind to a separate mDia1 C terminus construct (27). The molecular weight of the GFPimmunoreactive band in lane 6 did not correspond to any of the GFP-tagged constructs or to the in vitro proteolytic fragment evident in lane 3 (see below), making it likely to be an artifact.
In Vitro Proteolytic Cleavage of GFP-mDia1-We also note a smaller size GFP-reactive band associating with sIg.7-PKD2/ Ile 679 -Val 968 (Fig. 2C, lane 3). The size of this band is nearly identical to GFP-⌬N3mDia1, but it could not represent GFP-⌬N3mDia1, because the only GFP-tagged construct transfected in these cells was full-length GFP-mDia1. A recent publication (28) has shown that in vitro proteolytic digests can generate stable, active C-terminal fragments of both mDia1 and mDia3. These results prompted us to reason that mDia1 could undergo proteolytic cleavage to generate an N-terminal fragment capable of interacting with PKD2 because the PKD2-interacting domain remains intact. Because the C-terminal tail of mDia1 has been shown to interact with its N terminus, we tested whether deletion of the 52 most C-terminal residues (GFP-mDia1⌬C-52) could affect the proteolytic cleavage of wild type mDia1. We chose to delete the 52 C-terminal residues of mDia1-dependent Localization of PKD2 to Mitotic Spindles mDia1, because a naturally occurring mutation truncating these residues results in deafness (32). We have analyzed cell lysates transfected with either GFP-⌬N3mDia1 (Fig. 2D, lane  1), GFP-mDia1⌬C-52 (Fig. 2D, lane 2), the entire GFP-mDia1 molecule (Fig. 2D, lane 3), or GFP alone (Fig. 2D, lane 4) by Western blotting and detection with GFP antibody. The correct molecular weight for GFP-⌬N3mDia1 is shown in lane 1. The GFP-reactive band in the second lane represents mDia1⌬C-52 and is slightly smaller than the full-length GFP-mDia1, which is the higher molecular weight band shown in lane 3. GFP is a 27-kDa protein and is shown in lane 4. However, in Fig. 2D, lane 3, there is a smaller band that runs at a molecular weight almost identical to GFP-⌬N3mDia1, and this band is also present in the cell lysates shown in Fig. 2C, lanes 7, 9, and 10. We have consistently seen this smaller GFP-reactive protein band in lysates containing full-length GFP-mDia1, and we have considered that this may represent a form of mDia1 that has undergone a proteolytic cleavage only when in the closed conformation, since we did not see this band in the slightly shorter mDia1⌬C-52-containing lysates. The C terminus 52-amino acid truncation shortens the C-terminal intramolecular interaction domain (39), presumably leaving the molecule in an open position that is not susceptible to proteolysis. This could generate a functional mDia1 isoform lacking the FH1 and FH2 domains but maintaining the N-terminal Rho binding domain and FH3 domain that overlaps the PKD2 binding region.
Co-localization of PKD2 and mDia1 in Dividing Cells-To determine the subcellular localization of PKD2 and mDia1, we first employed indirect immunofluorescence and confocal microscopy on porcine kidney LLC-PK1 cells, since we have found PKD2 to be present on their nonmotile cilia (Fig. 1, E and F), in keeping with the published findings of others who have reported PKD2 on kidney epithelial cell cilia (8,9). We theorized that cilia might be a possible site for interaction between PKD2 and mDia1, since cilia are microtubule-based structures and mDia1 and mDia2 have both been reported to interact with microtubules (31,32). None of our efforts to demonstrate the presence of mDia1 on cilia by confocal microscopy were successful. Because mDia1 has already been reported to localize to the mitotic spindles of HeLa cells (31), we looked to see if this could be reproduced in conjunction with PKD2 immunoreactivity on LLC-PK1 mitotic spindles (Fig. 3A). Fig. 3A shows co-localization of endogenous PKD2 (green, left) and mDia1 (red, middle) in LLC-PK1 cells (merged, right). Fig. 3B shows LLC-PK1 cells similarly treated with normal rabbit and mouse IgG as nonspecific binding controls. These data suggest that mitotic spindles are a possible site of interaction for mDia1 and PKD2 in vivo.
Whereas the localization of mDia1 to the mitotic spindles of dividing HeLa cells has been shown (31), the finding that PKD2 localizes in these cellular structures is novel. Therefore, we examined PKD2 subcellular localization during cell division in other cell lines. Fig. 4A indicates that PKD2 immunoreactivity (left, green) overlaps with that of ␣-tubulin (middle, red) on the microtubule-based spindles of mitotic LLC-PK1 cells. We repeated this in human cervical carcinoma HeLa cells (Fig. 4, B and C) and mouse IMCD cells (Fig. 4D) to demonstrate that this is a specific localization of PKD2 in dividing cells in general and not restricted solely to LLC-PK1 cells. We have also analyzed LLC-PK1 cells with our chicken polyclonal antibody to PKD2 (green) and with ToPro3 DNA (blue) staining to show chromosome alignment (Fig. 4, E and F). Characteristic PKD2 immunoreactivity on spindle microtubules was evident with this second antibody to PKD2.
To directly that confirm PKD2 is in contact with mitotic spindles, we isolated mitotic spindles from HeLa cells transiently

mDia1-dependent Localization of PKD2 to Mitotic Spindles
transfected with Myc-tagged PKD2 (PKD2-myc) in the presence of taxol, a microtubule-stabilizing drug, or in the presence of the microtubule-destabilizing drug nocodazole. A PKD2-myc control (Fig. 5A, lane 1) and aliquots of the isolated spindles were subjected to Western blot and detected with antibody to Myc (Fig.  5A, lanes 2 and 3) or to ␤-tubulin (Fig. 5A, lanes 4 and 5). The levels of PKD2-myc in cell lysates from the spindle preparations were also detected with Myc antibody (Fig. 5A, lanes 6 and 7) to show that PKD2-myc expression levels were similar in both preparations. Although the ␤-tubulin level was less in the nocodazolestabilized preparation (Fig. 5A, lane 4 versus lane 5), PKD2-myc immunoreactivity in both nocodazole-and taxol-stabilized spindle preparations appeared similar. Fig. 5B shows mitotic spindle preparations from HeLa cells transfected with GFP-mDia1 in the presence of nocodazole or taxol (Fig. 5B, lanes 2 and 3, respectively), subjected to Western blot, and detected with antibody to GFP. Lane 1 is an aliquot of HEK293 cell lysate expressing GFP-mDia1 for molecular weight comparison. These results indicate that mDia1 and PKD2 may be associated with tubulin in pericentriolar material contained in the nocodazole-stabilized preparations, as well as with mitotic spindle microtubules in the taxol-stabilized spindle preparations. A second possibility is that PKD2 and Dia1 could be associated with a nonmicrotubule com-ponent of the spindle that remains in the presence of nocodazole and taxol.
To provide additional evidence that our PKD2-specific antibodies detected a PKD2-specific signal on mitotic spindles, we knocked down endogenous PKD2 in LLC-PK1 cells by RNAi and tested whether the spindle-specific signal would be eliminated or reduced. We generated two RNAi silencing constructs for PKD2, with one construct (PKD2 KD2-2 ) designed to specifically silence porcine PKD2 and the other (PKD2 KD2-3 ) designed to silence both porcine and human PKD2. Fig. 6A demonstrates the effectiveness of these constructs in HEK293T cells expressing human HA-PKD2 with PKD2 running as a 110-kDa size band and multimers of PKD2 at Ͼ200 kDa. Fig. 6A, lane 1, indicates the control expression level of HA-PKD2, and lane 2 shows a reduction in HA-PKD2 because the PKD2 KD2-3 silencing construct is capable of silencing both pig and human PKD2. Lane 3 does not show a reduction in HA-PKD2, since PKD2 KD2-2 is a pig-specific knockdown construct and should not silence human HA-PKD2. These results show that PKD2 KD2-3 can reduce but not completely abolish expression of HA-PKD2 in a transient transfection assay. Slow PKD2 protein turnover may account for the protein detection in transient transfection-based assays. Next, we co-transfected each of the PKD2 silencing constructs with GFP-tagged tubulin (GFP-tubulin) into LLC-PK1 cells. GFP-tubulin was used to mark transfected cells and to label mitotic spindles in dividing cells. Fig. 6, B and C, show that significant silencing of endogenous PKD2 occurs in co-transfected cells, as judged by PKD2 immunoreactivity (red, middle) on the mitotic spindles, thereby providing strong supporting evidence that our antibodies detected endogenous PKD2 in mitotic spindles.
mDia1-dependent Localization of PKD2 to the Mitotic Spindles-Since we had shown that PKD2 physically associates and co-localizes with mDia1 in the spindles and mDia1 has been shown to organize and stabilize microtubules, we tested whether PKD2 expression in the spindles requires mDia1 activity. Therefore, we elected to knock down mDia1 in HeLa cells and test whether PKD2 could still be detected in the spindles. We generated two mDia1-inactivating constructs (pSUPERretro-mDia1 KD1 and pSUPER-retro-mDia1 KD2 ) and a control construct containing irrelevant DNA to mouse I-mfa (pSUPERretro KD control ) in the pSUPER-retro vector. We prepared vesicular stomatitis virus-G recombinant retroviruses with these constructs that would allow permanent silencing of human mDia1 in cells infected with the mDia1-silencing constructs and no effect with the irrelevant DNA. Pools of infected HeLa cells were selected in the presence of 1 g/ml puromycin and analyzed for PKD2 localization in the spindles by indirect immunofluorescence and confocal microscopy. The results of these experiments are shown in Fig. 7. Efficient silencing levels of endogenous mDia1 in cells infected with mDia1 KD1 or mDia1 KD2 virus are shown in Fig. 7A. Fig. 7, B and C, represents the complete stack of confocal images and shows that both silencing constructs result in diffuse, indistinct, and nonspecific PKD2 staining throughout the cytosol but no defined staining on mitotic spindles (green, left). However, tubulin staining (red, middle) remains well defined. The merged image (right) reflects the overlay of well defined tubulin staining (red) on spindles with diffuse and cytosolic PKD2 staining (green). Fig. 7D shows that there is no effect on PKD2 localization to mitotic spindles in HeLa cells by the silencing vector control DNA. These results suggest that mDia1 activity is required for the localization of PKD2 to the mitotic spindles.
Fluoresence Spectroscopy of HeLa and HeLa mDia1 KD1 Cells- Fig. 8A shows intracellular Ca 2ϩ release assays in Ca 2ϩ buffer in response to 10 M histamine in mDia1 knockdown  2, 4, and 6) or taxol (lanes 3, 5, and 7) were immunoblotted (IB) with rabbit (Rb) anti-Myc (lanes 1, 2, 3, 6, and 7) or mouse (Ms) anti-␤ tubulin (lanes 4 and 5). Lanes 6 and 7 represent expression of PKD2-myc in supernatants following microtubule isolation. Lane 1 shows expression of PKD2-myc in crude HEK293T cell lysates as a molecular weight reference. B, lane 1, GFP-mDia1 in crude HEK293T cell lysates as a molecular weight reference; lanes 2 and 3, GFP-mDia1 in nocodazole-and taxol-stabilized mitotic HeLa preparations. mDia1-dependent Localization of PKD2 to Mitotic Spindles cells (HeLa mDia1 KD1 , red) and in HeLa cells (black). The inset bar graph representing normal HeLa cells (black, n ϭ 4) and HeLa mDia1 KD1 cells having reduced mDia1 expression (red, n ϭ 4) shows that there is no significant difference in histamine responsiveness in these cell types. Additionally, the inset confocal images indicate that PKD2 immunoreactivity (green) is concentrated in a perinuclear position at the endoplasmic reticulum in both cell types. This suggests that in nonmitotic normal or mDia1 knockdown HeLa cells, the localization of PKD2 has remained concentrated at the endoplasmic reticulum, and these findings suggest that PKD2 channel activity or ER localization is not altered by decreased mDia1 expression in nonmitotic cells. Previous studies by others have generated profiles of intracellular Ca 2ϩ release in normal and mitotic HeLa cells (40,41), which have enabled us to assess the quality of our mitotic cell preparations by analyzing the reproducibility of their earlier results with our own. Fig. 8B represents a typical Ca 2ϩ release assay in mitotic HeLa and HeLa mDia KD1 cells in contrast to nonmitotic HeLa cells in the presence of Ca 2ϩ and in response to 10 M histamine. As expected, normal interphase HeLa cells (green) were responsive to histamine by exhibiting a rapid intracellular Ca 2ϩ release followed by a prolonged period of extracellular Ca 2ϩ entry that remains above base-line values. Normal mitotic HeLa cells (black) given an identical amount of histamine released intracellular Ca 2ϩ followed by little or no extracellular Ca 2ϩ entry, whereas mitotic mDia1 knockdown cells had decreased intracellular calcium release under these conditions. These findings are con-sistent with the earlier studies and verify that we have obtained suitable preparations of interphase and mitotic cells. The mitotic HeLa mDia1 KD1 cells (red) responded to histamine in a manner similar to mitotic HeLa (black) cells, with the exception that each assay reflected a decrease in overall intracellular Ca 2ϩ release in the cells with decreased mDia1 expression. This difference, as measured by peak increase over baseline values, approaches a decrease of ϳ50% for each experiment. To determine the histamine response in the absence of extracellular Ca 2ϩ , we repeated this assay with wild type HeLa and HeLa mDia1 knockdown cells in buffer containing EGTA. Fig. 8C represents one of these assays, with a bar graph to the right showing the results of three assays. The mitotic HeLa knockdown (red) cells have a 60% decrease (0.05 Ϯ 0.004) in Ca 2ϩ release versus normal mitotic HeLa cells (black, 0.083 Ϯ 0.007). These findings suggest that Ca 2ϩ release is occurring with little or no Ca 2ϩ entry during mitosis and that mitotic mDia1 knockdown cells do not release Ca 2ϩ as efficiently as wild type mitotic HeLa cells. These results imply that mDia1 can modulate intracellular Ca 2ϩ release in mitotic cells, although our assay cannot determine whether this is a direct or indirect effect only on PKD2. DISCUSSION Using a yeast two-hybrid screen, we have identified a protein-protein interaction between the C-terminal tail of human PKD2 and the N terminus of mDia1. The interaction was verified by co-immunoprecipitation experiments in vivo in na-

mDia1-dependent Localization of PKD2 to Mitotic Spindles
tive cell lines and in vitro in transfected cells. Our results suggest that the interaction is more prevalent in the mitotic spindle of dividing cells and that loss of mDia1 activity results in the loss of PKD2 from the spindle. These results prompt us to conclude that mDia1 facilitates the localization of PKD2 to the spindle during cell division. We propose that the interaction is likely to have physiological significance, because it occurs in cells under physiological conditions. In addition, the interaction is likely to be conserved between mammalian homologs of PKD2 but not in C. elegans and Drosophila PKD2 homologs, since the homology at the very C terminus is quite low among different organisms. Similarly, PKD2L or the disease-causing mutant PKD2(871X) would not be expected to interact with mDia1, since they lack all or most of the mDia1 binding site. Conversely, the mDia1-related proteins, mDia2 and mDia3, do not have significant homology to mDia1 in the PKD2 binding region. The implication of these results is 2-fold. First, PKD2 may be symmetrically divided between mother and daughter cell, and second, PKD2 activity may be essential during or immediately following cell division.
The PKD2 binding site (amino acids 143-260) is in the mDia1 RhoA binding domain (amino acids 63-260) that over-laps with the mDia1 FH3 domain (amino acids 157-456), which contains a 173-amino acid region that binds polymerized tubulin (31). This suggests that PKD2 could compete with or block RhoA binding to generate a unique open and functional form of mDia1 still capable of interacting with its effector molecules via its coiled-coil, FH1, and FH2 domains. Thus, PKD2 may activate mDia1 independently of RhoA, and this may have implications for mDia1 functions that do not require RhoA. A second scenario is that PKD2 keeps mDia1 in a closed, nonfunctional form in terms of FH1 and FH2 domain interactions but still permits mDia1 to transport PKD2 to the spindle. mDia1-dependent Localization of PKD2 to Mitotic Spindles terminus of the protein. We are currently investigating whether such cleavage occurs in vivo and whether it depends on the presence of an intact C terminus. The functional implication of mDia1 cleavage is that it may generate a part of the molecule that would still be capable of mediating some of the functions of full-length mDia1 independently of RhoA binding.
Our results indicate that PKD2 localizes to the spindle during cell division. This was supported by indirect immunofluorescence staining of endogenous PKD2 in several cell types using two independent antibodies, RNAi, and biochemical evidence showing PKD2 association with purified spindles in transfected cells. Based on these results, we conclude that the mitotic spindle is a physiological subcellular localization of PKD2 during cell division. However, we presently do not have evidence for the physiological role of PKD2 localization at the mitotic spindle. We would like to propose two not necessarily mutually exclusive possibilities. The simpler one would be that PKD2 is proportionally divided between mother and daughter cells through symmetric loading on the spindle. Second, PKD2 channel activity is required during or immediately following cell division. There is precedence for both of these possibilities, since there are a number of proteins that can be defined as chromosomal passenger proteins or spindle-associated proteins. For example, survivin is necessary for chromosome alignment and spindle checkpoint assembly arrest (42), PRC1 is a microtubule-associated protein that maintains the midzone (43), and p23 transiently associates with mitotic spindles dur-ing cell cycle with a currently unknown function (44). The Drosophila transmembrane Axs protein co-localizes with ER proteins and is in a structure that ensheaths spindles during meiosis, and loss of Axs function impairs meiotic progression (45). Therefore, it is not necessarily unique to find a resident ER transmembrane protein bound to microtubule spindles during cell division. It remains possible that PKD2 is taken to the spindle microtubules primarily to ensure its coordinated segregation into the mother-daughter cells so that Ca 2ϩ homeostasis can be maintained in the new cell.
The idea that PKD2 has a role in cell division by regulating Ca 2ϩ homeostasis is an attractive one and is supported by previous findings that PKD2 can function as an intracellular Ca 2ϩ release channel. A previous study has examined mitotic Chinese hamster ovary cells for positioning of the IP 3 receptor (IP 3 R) and endoplasmic reticulum Ca 2ϩ stores and found that Ca 2ϩ stores localized to the future cleavage cortex prior to the onset of cleavage furrow formation, and the IP 3 R protein was associated with microtubule bundles (46). We have verified that the ER-resident IP 3 receptor protein co-localizes with tubulin on the mitotic spindles of HeLa cells by indirect immunofluorescence (data not shown). Another study has shown that microinjection of Ca 2ϩ store-enriched microsomes into dividing newt eggs induced a new cleavage furrow at ϳ70% of the injection sites via an IP 3 Rinduced Ca 2ϩ release (47). Taken together, these findings suggest that ER-resident proteins such as the IP 3 R and now the PKD2 calcium channel are positioned with Ca 2ϩ stores in divid- FIG. 8. Ca 2؉ ratiometric fluorescence spectroscopy of interphase and mitotic HeLa cells. Data are expressed as the ratio of bound Ca 2ϩ over unbound Ca 2ϩ in the presence of 1.8 mM Ca 2ϩ or 0.5 M EGTA and 2 M Indo 1 AM Ca 2ϩ indicator dye. The arrow indicates a 10 M histamine addition to initiate Ca 2ϩ release. A, Ca 2ϩ release in nonmitotic HeLa (black) and HeLa mDia1 KD1 (red) cells in response to histamine in the presence of Ca 2ϩ . The inset bar graph represents four identical assays and indicates that there is no significant difference between cell types. Confocal microscope images (right) show PKD2 immunoreactivity (green) in the endoplasmic reticulum around the nucleus (blue; ToPro3 DNA stain) in both HeLa and HeLa mDia1 KD1 cells used in these assays. B, representative Ca 2ϩ release measurements of normal interphase HeLa cells (green), mitotic normal HeLa cells (black), and mitotic mDia1 KD1 (red) knockdown HeLa cells in response to histamine in the presence of Ca 2ϩ . C, a representative graph of Ca 2ϩ release measurements of normal mitotic (black) and mitotic mDia1 KD1 (red) knockdown HeLa cells (left) in response to histamine in the absence of Ca 2ϩ . Inset bar graph (right), the results of three Ca 2ϩ release experiments in the absence of Ca 2ϩ , p ϭ 0.03. ing cells in a manner that utilizes these proteins for localized Ca 2ϩ release during mitosis. Our finding that decreased mDia1 expression in mitotic cells diminishes intracellular Ca 2ϩ release, taken together with our binding and localization studies, strongly suggests that PKD2 and mDia1 are active players in Ca 2ϩ release during mitosis. The decrease in intracellular Ca 2ϩ release in mDia1 knockdown cells may be a consequence of fewer PKD2 channels being localized to spindles, or it may reflect a direct effect of mDia1 on channel activity due to changes in cytoskeletal architecture or be the result of a loss of direct interaction with the channel protein.
Alteration of Ca 2ϩ signaling during cell division may have a significant impact on the differentiation process of kidney epithelial precursors. Given the fact that pathogenic mutations in PKD1 and PKD2 result in dedifferentiated epithelial cell types (48,49), our results prompt us to propose that loss of PKD2 may indirectly affect ciliary function by altering the differentiation status of tubular cells. This idea is further supported by a previous study showing that PKD1 and PKD2 can modulate the activity of the AP-1 transcription factor via activation of the mitogen-activated protein kinases p38 and JNK1 and protein kinase C⑀ (49) and the Jun-activated kinase/signal transducer and activator of transcription pathways (50). It would appear that any deficiencies in Ca 2ϩ homeostasis may have multiple levels of complexity, leading to altered gene expression. Therefore, our findings reported here that PKD2 is positioned along mitotic spindles during cell division may be a physiologically relevant mechanism to ensure that the newly divided cells can initiate necessary gene transcription leading to a fully differentiated cell from a very early time point. Loss of PKD2 by inability to segregate properly along spindles or by loss of normal function after cell division could lead to the undifferentiated phenotype of cystic cells. mDia1 has been shown earlier to localize to the spindles of HeLa cells. In addition, the mDia1 localization to the spindle was independent of RhoA binding and was mediated by a region within the FH3 domain, a region overlapping the PKD2 binding site. We found that endogenous PKD2 and mDia1 co-localized to these structures of LLC-PK1 cells. These data prompted us to test whether mDia1 may regulate the expression of PKD2 at the spindles. Knockdown of mDia1 in HeLa cells resulted in a diffuse and nondiscrete expression of PKD2 in the spindle, supporting the idea that mDia1 is essential for symmetrical loading of PKD2 onto the spindle. Although this effect appears specific to PKD2 because we did not observe changes in the spindle expression of the IP 3 R type 3 (data not shown), we cannot exclude the possibility that loss of mDia1 may have resulted in global changes of ER arrangements affecting PKD2 localization and function. Nevertheless, the finding that mDia1 affected the localization of an ER resident protein may have implications for an effect of mDia1 on ER partition during cell division (ER inheritance).
The fact that HeLa cells containing mDia1 knockdown constructs continued to grow well in culture is not an entirely unexpected result, since mouse embryonic stem cells that are deficient or null for mDia1 continue to proliferate and show an increased level in the Drf2/p134mDia2 protein (51). Therefore, mDia2 may be performing the role of mDia1 in cytokinesis in our mDia1 knockdown HeLa cell lines, since it is normally present in HeLa cells (28). A second consideration is that protein silencing does not necessarily yield a complete cessation of protein expression, and this could also allow cytokinesis to continue, particularly if a second formin protein could be assisting in a redundant fashion. However, our confocal images of HeLa mDia1 knockdown cells do not indicate that a second protein is capable of quantitatively bringing PKD2 to the mitotic spindles, since the loss of PKD2 staining is almost complete in some HeLa cells and having varying levels of reduction in others, consistent for a population of cells. Therefore, our results suggest that mDia1 has an additional function beyond its well documented role as a cytoskeletal organizing protein in that it partitions PKD2 and perhaps other proteins to mitotic spindles.
In conclusion, mDia1 is generally considered to be a cytoskeletal organizing protein capable of forming actin stress fibers and regulating the formation and orientation of stable microtubules. We now suggest that a novel function for mDia1 is to facilitate PKD2 localization to mitotic spindles, possibly by binding the C terminus of PKD2 that protrudes into the cytoplasm from the membrane of a subset of ER vesicles and to modulate intracellular Ca 2ϩ release in mitotic cells. In terms of autosomal dominant polycystic kidney disease, some mutations in PKD2 could permit equal distribution among dividing cells via mDia1, although truncation mutants of PKD2 unable to interact with mDia1 would generate mother-daughter spindles largely devoid of PKD2 and presumably having less intracellular Ca 2ϩ release. Other PKD2 loss of function mutations may segregate evenly during cell division but result in subsequent cellular abnormalities resulting in cyst formation due to impaired PKD2 activity.