Interaction of the DYNLT (TCTEX1/RP3) Light Chains and the Intermediate Chains Reveals Novel Intersubunit Regulation during Assembly of the Dynein Complex*

The cytoplasmic dynein 1 cargo binding domain is formed by five subunits including the intermediate chain and the DYNLT, DYNLL, and DYNLRB light chain families. Six isoforms of the intermediate chain and two isoforms of each of the light chain families have been identified in mammals. There is evidence that different subunit isoforms are involved in regulating dynein function, in particular linking dynein to different cargoes. However, it is unclear how the subunit isoforms are assembled or if there is any specificity to their interactions. Co-immunoprecipitation using DYNLT-specific antibodies reveals that dynein complexes with DYNLT light chains also contain the DYNLL and DYNLRB light chains. The DYNLT light chains, but not DYNLL light chains, associate exclusively with the dynein complex. Yeast two-hybrid and co-immunoprecipitation assays demonstrate that both members of the DYNLT family are capable of forming homodimers and heterodimers. In addition, both homodimers of the DYNLT family bind all six intermediate chain isoforms. However, DYNLT heterodimers do not bind to the intermediate chain. Thus, whereas all combinations of DYNLT light chain dimers can be made, not all of the possible combinations of the isoforms are utilized during the assembly of the dynein complex.

Cytoplasmic dynein 1 is a microtubule-based molecular motor that functions to generate force for cargo transport to microtubule minus-ends (1)(2)(3). It is involved in numerous eukaryotic cell processes including the trafficking of membranous vesicles, viruses, and other intracellular particles. Cytoplasmic dynein 1 is a large multisubunit complex (ϳ1.5 MDa) containing two copies of six subunits, the heavy chain (DYNC1H), the intermediate chain (DYNC1I), the light intermediate chain (DYNC1LI), and three distinct light chains, DYNLT (previously called Tctex1), DYNLRB (previously called roadblock), and DYNLL (previously called LC8) (2, 4 -6). The motor domains of cytoplasmic dynein 1 are located in the C-terminal globular heads of the two identical heavy chains (7)(8)(9). The heavy chains dimerize via their N-terminal stalks, and the stalks also contain the light intermediate chain and intermediate chain binding sites (3, 10 -13). The three light chains bind to different locations on the N terminus of the intermediate chain (14 -16).
Whereas there is only a single heavy chain isoform, there are multiple isoforms of the five subunits that make up the cargo binding domain (2,3,17). In mammals, at least six intermediate chain isoforms are produced by the alternative splicing of two genes, and there are at least two genes for each of the other four subunits (2,3,5,18). Assembly of individual subunit isoforms into the dynein complex creates different populations of the motor protein that are thought to be involved in specific cargo binding and regulation (2, 19 -22). For example, pericentrin is transported to the centrosome exclusively by the dynein complexes that contain the light intermediate chain isoform DYNC1LI-1 (23). The DYNLT and DYNLL light chains have been shown to interact with numerous functionally unrelated proteins (3, 14, 15, 24 -27).
The two members of the DYNLT family, DYNLT1 (previously called Tctex1) 2 and DYNLT3 (previously called rp3) are found in all cultured cells and adult and fetal tissues so far examined (19,28,29). Unlike the DYNLT isoforms, the expression of the six intermediate chains isoforms are tissue and cell type-specific (18, 30 -32). One intermediate chain isoform, DYNC1I-2C (IC-2C), 3 is found in all cells and it is often the only isoform found in cultured cells (18). Most tissues express only * This work was supported by National Institutes of Health NINDS Grants NS29996 (to K. K. P.) and GM 51293 (to S. M. K.). 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 To whom correspondence should be addressed: P. O. Box 800732, University of Virginia, Charlottesville, VA 22908-0732. the IC-2C and IC-2B isoforms. All six isoforms are found only in brain, where their expression levels are regulated during  development, and changes in the expression levels of the IC-2C  and IC-2B isoforms and the light intermediate chains are also  observed upon nerve growth factor-induced PC12 cell differen-tiation (18, 30 -32). The intermediate chain and light intermediate chain isoforms define distinct pools of dynein in axons (18,33). However, much remains to be learned about the subunit organization of the dynein cargo binding domain. To investigate the contributions of different subunit isoforms to the structural organization of the dynein complex, we investigated the interactions of two members of the DYNLT light chain family, DYNLT1 and DYNLT3, with one another and with other subunits of the dynein cargo binding domain. The DYNLT light chain family was chosen for this study because the two isoforms bind to different proteins and mediate their transport by dynein (20,27,34). Also, structural studies indicate that the DYNLT1 and DYNLT3 light chains contain unique domains that are predicted to be utilized for specific cargo binding, whereas many of the proteins that bind to DYNLL1 and DYNLL2 compete with the intermediate chain for binding to the light chains (35)(36)(37)(38). We found that dynein with the DYNLT light chain also contained the DYNLL light chain, and that whereas all the DYNLT isoforms co-purified with the dynein complex, most of the DYNLL polypeptides did not. Using yeast two-hybrid and co-immunoprecipitation assays, we found that DYNLT1 and DYNLT3 form homodimers and heterodimers, and that the homodimers bound to all six of the intermediate chain isoforms in vivo. Most importantly, we found that the DYNLT1-DYNLT3 heterodimer did not bind to the intermediate chain. These results demonstrate that binding to the intermediate chain regulates the assembly of light chain dimers into the dynein cargo binding domain.
Antibodies-The antibodies used were 74.1, a mouse monoclonal antibody that reacts with all isoforms of the intermediate chain (42); CT199, a rabbit polyclonal antibody that reacts with all the DYNLRB family members, and R4058, a rabbit polyclonal antibody that reacts with all the DYNLL family members (43,44); HC8, a rabbit polyclonal antibody that reacts with the cytoplasmic dynein heavy chain; and R1B2, a rabbit polyclonal antibody that reacts with the isoforms of the LIC subunit, both from Dr. R. Vallee (23); and anti-HA (12C5) and anti-Myc (9E10) antibodies obtained from the Lymphocyte Culture Center, University of Virginia.
Yeast Two-hybrid Studies-Pairwise yeast two-hybrid analysis was performed as described previously with slight modification (16,34). Briefly, AH109 yeast strains were co-transformed with the constructs as indicated in the figures according to the manufacturer's instructions (Clontech). The transformed yeast cells were resuspended in 100 l of sterile water and 20 l were dropped onto both Ϫ2 and Ϫ3 plates. The Ϫ2 plates, lacking both leucine (Leu) and tryptophan (Trp), showed co-transformation. The Ϫ3 plates, lacking Leu, Trp, and histidine (His), demonstrated protein-protein interactions. To eliminate false positives arising from "leaky" HIS3 expression, 3 mM 3-amino-1,2,4-triazole, a competitive inhibitor of histidine synthase, was added to the Ϫ3 plates. The plates were incubated at 30°C for 5 days. Positive interactions were confirmed with a ␤-galactosidase assay that screened for the independent expression of the lacZ reporter gene. Co-transformation with the pGBK-p53 and pGAD-T-antigen vectors was used as a control for positive interaction.
Cell Culture, Transfection, and Co-immunoprecipitations-293T cells were maintained in Dulbecco's modified Eagle's medium and transfected as described previously (41). For the co-immunoprecipitation assays, cells were resuspended in lysis buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, 1 g/ml leupeptin), and spun in a microcentrifuge for 10 min at 4°C, and lysates were incubated 3 h at 4°C with anti-HA antibody (10 g) pre-bound to protein A beads (Zymed Laboratories Inc.). The beads were then washed extensively with 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 1 mM EDTA. The co-immunopreciptates were analyzed by SDS-PAGE and immunoblotting with anti-Myc antibody as described previously (41,45). To study the equilibrium association of the members in DYNLL and DYNLT families, 293T cells were individually transfected with the HA-or Myc-tagged light chains and equal volumes of the resulting lysates with homodimers of the light chains were mixed together for 3 h at 4°C. The lysates were then incubated for an additional 3 h at 4°C with 10 g of anti-HA antibody prebound to Protein A beads, and the coimmunoprecipitating proteins were analyzed by SDS-PAGE and immunoblotting.
Dynein Purification-Cytoplasmic dynein was immunoprecipitated from rat brain lysates with antibodies to the dynein intermediate chain and the light chains as described previously (33,46) except that the wash buffer was as described above. Rat brain and 293T cytosols were fractionated on 5-20% sucrose density gradients as described previously for microtubule affinity purified dynein (19,42). Microtubule pellets were prepared as described previously (42,47).
DYNLT Heterodimer Binding to the Intermediate Chain-To determine whether DYNLT1-DYNLT3 light chain heterodimers were competent to bind to the intermediate chain, 293T cells were co-transfected with His-DYNLT3 and Myc-DYNLT1 light chains. Cell lysates were prepared as described above, except that EDTA was omitted from the buffer. The lysates were incubated with Ni 2ϩ -nitrilotriacetic acid beads (Novagen) for 1 h at 4°C. Protein complexes bound to the beads were eluted with 0.25 M imidazole buffer, 0.1% bovine serum albumin, pH 8.0, and the eluates were then incubated with cytosol from cells expressing Myc-tagged intermediate chain (DYNC1I-2C) for 2 h at 4°C and diluted with lysis buffer. The mixtures were then incubated for an additional 3 h at 4°C with 74.1 antibody bound to Protein A beads. The beads were pelleted and unbound proteins remaining in the supernatant were collected and concentrated by precipitation with 10% trichloroacetic acid. The beads were washed extensively and immunoprecipitated proteins were eluted with 0.1 M glycine, 0.1% bovine serum albumin, pH 3.0. The pH of the eluate buffer was neutralized with 1 M Tris, pH 8.1, and the supernatant and immunoprecipitated proteins were analyzed by SDS-PAGE and immunoblotting.
Preparation and Characterization of Monoclonal Antibodies to DYNLT1 and DYNLT3-The two light chain-myosin basic protein fusion proteins were expressed in bacteria and purified with amylose column chromatography following the manufacturer's instructions as described previously (19). The purified fusion proteins were injected into separate mice and monoclonal antibodies were prepared by the University of Virginia Lymphocyte Culture Center as described previously (48,49). Positive hybridoma colonies producing antibodies specific for each light chain were identified by enzyme-linked immunosorbent assay screening of tissue culture supernatants against purified DYNLT1-and DYNLT3-GST fusion proteins. Antibodies from positive colonies were screened in immunoprecipitation and immunoblot assays with lysates of the bacteria expressing the GST fusion proteins, as described previously (34,42), and positive hybridomas were cloned.

Specificity of DYNLT1 and DYNLT3 Antibodies-
The specificities of four antibodies to the DYNLT isoforms were demonstrated by immunoprecipitation or probing Western blots of bacterial cell extracts expressing GST-light chain fusion proteins ( Fig. 1). Two antibodies, R1 and T1, reacted specifically with the DYNLT3 and DYNLT1 fusion protein bands, respectively (Fig. 1A). These antibodies also detected single bands of the appropriate size from brain and tissue culture cell lysates and microtubule pellets prepared from rat brain (Fig. 1B). The T1 antibody has a higher affinity for human DYNLT1 than the rat DYNLT1 (not shown). Because, the two proteins differ in only two conserved amino acid substitutions at the N terminus of the polypeptide; it is likely that the antibody is directed toward the DYNLT1 N terminus. Two antibodies, R2 and R3, immunoprecipitated both the DYNLT3 and DYNLT1 GSTlight chain fusion proteins (Fig. 1C). Thus, these antibodies recognize both members of the DYNLT light chain family. The R1 antibody also immunoprecipitated recombinant DYNLT3. However, R1 did not immunoprecipitate the dynein complex ( Fig. 6) suggesting that the epitope is masked when DYNLT3 is incorporated into the dynein complex.
Homodimer and Heterodimer Formation by Members of the DYNLT Light Chain Family in Eukaryotic Cells-Two different approaches were used to determine whether DYNLT1 and DYNLT3 exist as heterodimers or homodimers in eukaryotic cells: a pairwise yeast two-hybrid assay and co-immunoprecipitation assay. In the yeast two-hybrid analyses, positive results were obtained when either DYNLT light chain isoform was present in both the pGBKT7 and pGADT7 vectors ( Fig. 2A). Transformation with individual light chain constructs did not activate the His gene. The specificity of the interaction was confirmed with the ␤-galactosidase assay (data not shown). These data indicated that members of the DYNLT family can associate in all possible combinations of homodimers and heterodimers. The co-immunoprecipitation assay further demonstrated that both homodimers and heterodimers are formed between members of the DYNLT family in cultured mammalian cells (Fig. 2B). Cells were co-transfected with two light chains, each marked with a different epitope tag. Immunoprecipitation with antibodies to one tag invariably co-immunoprecipitated the light chain with the second tag. To determine whether the heterodimers detected in the co-immunoprecipitation assay were formed after cell lysis, cytosols prepared from cells expressing either DYNLT1 or DYNLT3 homodimers were mixed. After incubation, no co-immunoprecipitation of the two isoforms was detected (Fig. 2C). The absence of heterodimer formation in the mixed cytosols demonstrates that there is no exchange of subunits between the DYNLT1 and DYNLT3 homodimers.  (Fig. 3). The interactions were further confirmed with ␤-galactosidase assays (data not shown). With the yeast two-hybrid and mammalian expression data showing that the DYNLT light chains form homodimers (Fig. 2), and the DYNLT1 structural data derived from bacterially expressed proteins (35,36,50), we conclude that homodimers of both members of the DYNLT light chain family are capable of directly interacting with all six intermediate chain isoforms.
DYNLT1-DYNLT3 Heterodimers Do Not Bind the Intermediate Chain-We next sought to determine whether the DYNLT1-DYNLT3 heterodimers bind to intermediate chains. When light chains with different tags are co-transfected into cultured cells, three populations of dimers are created: DYNLT1 homodimers, DYNLT3 homodimers, and DYNLT1-DYNLT3 heterodimers. To separate the heterodimers and the DYNLT3 homodimers from the DYNLT1 homodimers, we expressed His-DYNLT3 and Myc-DYNLT1 in cultured cells and purified the His-DYNLT3 homodimers and His-DYNLT3-Myc-DYNLT1 heterodimers on nickel beads. As a control, the nickel bead binding confirmed that the His tag did not interfere with the formation of His-DYNLT3-Myc-DYNLT1 heterodimers (Fig. 4) or His-DYNLT3-Myc-DYNLT3 homodimers (data not shown). After the His-DYNLT3 homodimers and His-DYNLT3-Myc-DYNLT1 heterodimers were eluted from the nickel beads, they were incubated with Myc-tagged intermediate chains. When the intermediate chains were immunoprecipitated, DYNLT3 was found in the immunoprecipitate, but DYNLT1 was not (Fig. 4). This demonstrated that DYNLT3 homodimers but not DYNLT3-DYNLT1 heterodimers bound to the intermediate chain. Myc-DYNLT1 and His-DYNLT3  were found in the supernatant after the immunoprecipitation, confirming that the heterodimer was not immunoprecipitated (Fig. 4). Therefore, whereas DYNLT1 and DYNLT3 form heterodimers in vivo, the heterodimers do not bind intermediate chains and are thus not incorporated into dynein complexes.
Because the DYNLT1-DYNLT3 heterodimer was unable to bind to the intermediate chain, we sought to determine the size of the endogenous pools of DYNLT1 and DYNLT3 subunits that were not incorporated into dynein complexes. Cytosol from rat brain or cultured human cells was applied directly to sucrose density gradients without prior selection of intact dynein complexes by binding to microtubules (Fig. 5). All the DYNLT3 in rat brain cytosol co-purified with the other dynein subunits in the 18 S peak. In addition, all the DYNLT1 in cultured human cell cytosol co-purified exclusively with the dynein complex. No DYNLT1 or DYNLT3 was found at the top of the sucrose gradient where unincorporated homodimers and heterodimers would have been located. In contrast, the fractionation pattern of the DYNLL light chain subunit was distinct. DYNLL light chains were found in all the sucrose gradient fractions, demonstrating that most of the DYNLL was not assembled into the dynein complex.
Dynein Complexes with the DYNLT Light Chains Contain the DYNLL and DYNLRB Light Chains-The DYNLT, DYNLL, and DYNLRB light chains bind to separate regions of the intermediate chain N terminus (14,15,51). Whereas these data sug-gested that the light chains bind to the intermediate chain simultaneously, that fact has not yet been demonstrated with purified dynein. To determine whether dynein with the DYNLT family members also contained either the DYNLL or DYNLRB light chains, endogenous dynein complexes were immunoprecipitated from rat brain lysates using antibodies that recognize both members of the DYNLT families (Figs. 1 and 6). We found that both DYNLL and DYNLRB light chains co-immunoprecipitated with the antibodies to the DYNLT family. As a control to demonstrate that intact dynein complexes were immunoprecipitated, we confirmed the presence of the heavy and intermediate chain subunits (Fig. 6). In addition, whereas R1 immunoprecipitated recombinant DYNLT3 (Fig.  1), it did not immunoprecipitate DYNLT3 from cytosol (Fig. 6). This provides further evidence that there is no free pool of DYNLT3 in cytosol.
DYNLL Isoform Dimerization and Intermediate Chain Isoform Binding-Using the approaches described in the previous sections to characterize the DYNLT family, we found that the DYNLL isoforms formed both homodimers and heterodimers (Fig. 7, A and B). Interestingly, when cytosols from cells expressing DYNLL homodimers were mixed, the two isoforms co-immunoprecipitated (Fig. 7C). Thus, in contrast to the two DYNLT homodimers, there was exchange of subunits between the two DYNLL homodimers. We further found that both DYNLL isoforms bound to all intermediate chain isoforms (Fig.  7D). However, we were unable to determine whether the DYNLL heterodimers bound to intermediate chains, in part due to nonspecific binding of an overexpressed protein to the purification matrix.

DISCUSSION
To better understand the functional significance of different subunit isoforms to the dynein complex, we characterized the   DECEMBER 21, 2007 • VOLUME 282 • NUMBER 51 organization and assembly interactions of the DYNLT isoforms with other subunits in the dynein cargo binding domain. We found that the two isoforms of DYNLT light chain family, DYNLT1 and DYNLT3, form all possible combinations of homodimers and heterodimers in yeast and cultured mammalian cells (Fig. 2). The structure of DYNLT1 has been the subject of considerable study. The NMR and crystal structures have been solved and show that two monomers associate across a strand-switched ␤-sheet interface to form a symmetrical dimer (35,36). Using bacterially expressed protein, the Barbar group (50) was unable to identify conditions to detect structured monomers, and they calculated that the dimer exists at concentrations greater or equal to 10 Ϫ14 M. This is consistent with our finding that there is no exchange of DYNLT subunits between homodimers in cytosol and together these data suggest that DYNLT dimer formation may be coupled to assembly into the dynein complex. DYNLT1 and DYNLT3 share 74% sequence similarity (55% identity) and sequence analysis shows that the regions that make the dimerization interface are highly conserved. The only non-conservative amino acid difference between DYNLT1 and DYNLT3 is at the periphery of the inter-face (36). Thus, there is no structural impediment to the formation of the DYNLT1-DYNLT3 heterodimers or the DYNLT3 homodimers that we observed in yeast and tissue culture cells.  Ϫ2 co-transformation and Ϫ3 interaction plates. Only yeast strains carrying a DYNLL isoform in both vectors (spots 1, 2, 5, and 8) were able to grow on the Ϫ3 plate, indicating that members in DYNLL family are capable of forming homodimers and heterodimers in vivo. There was no interaction when the light chains were co-transformed with their respective empty partner plasmids (spots 3, 4, 6, and 7). B, co-immunoprecipitation assay. 293T cells were co-transfected with the pairs of DYNLL light chain isoforms tagged with the HA and Myc epitopes indicated by the plus sign (ϩ). The HA-tagged light chain isoforms in the cell lysates were immunoprecipitated with anti-HA antibody. The lysates (Myc Inputs) and immunoprecipitates (Anti-HA IP) were analyzed by SDS-PAGE and immunoblotting, probed with anti-Myc antibody. Lysates are shown to verify expression of DYNLL proteins. Control immunoprecipitations were performed without transfecting HA-tagged light chain and in these no Myc-tagged light chains were immunoprecipitated by the HA antibody. Consistent with the yeast two-hybrid assay, all combinations of DYNLL homodimers and heterodimers were detected in the co-immunoprecipitation assay. C, DYNLL heterodimer formation by exchange of subunits between homodimers. 293T cells were transfected individually with Myc-DYNLL1 and HA-DYNLL2. The Myc-DYNLL1 containing lysates (input) were then mixed with HA-DYNLL2 containing lysates (input) for 3 h at 4°C. The HA-DYNLL2 was immunoprecipitated with anti-HA antibody and immunoprecipitate was analyzed by SDS-PAGE and the immunoblot (IB) was probed with anti-Myc antibody. Myc-tagged DYNLL1 (IB) was co-immunoprecipitated with HA-tagged DYNLL2, demonstrating an exchange of the HA-and Myc-tagged DYNLL isoforms. D, interactions of DYNLL homodimers with the intermediate chain isoforms. Yeast strains containing various combinations of vectors (left panel) were plated on the Ϫ2 co-transformation and Ϫ3 interaction plates. Both members in the DYNLL family interacted with all intermediate chain isoforms (spots 1-6). Co-transformation with pGBKT7 and pGADT7 (spot 9) was used as a positive control for the Ϫ2 co-transformation plate and a negative control for the Ϫ3 interaction plate. Co-transformation with p53 and T-antigen vectors (spot 8) was used as a positive control for the Ϫ3 interaction plate. There was no interaction when the light chains (spot 7) were co-transfected with the empty partner vector.

DYNLT Light Chain Isoform Interactions within Cytoplasmic Dynein
The two members of the second light chain family, DYNLL1 and DYNLL2, also formed both homodimers and heterodimers in the yeast and mammalian expression systems (Fig. 7). However, in contrast to the DYNLT light chain homodimers, we observed an exchange of subunits between the DYNLL homodimers. It has previously been observed that, unlike the DYNLT dimers, the DYNLL dimers undergo a pH-dependent shift to the monomer (14,35,52,53). The ability of the DYNLL light chains to form a structured monomer may facilitate the exchange of subunits we observed between the DYNLL homodimers.
We further found that the DYNLT1 and DYNLT3 isoforms bind all six intermediate chain isoforms in a yeast two-hybrid binding assay (Fig. 3). Having shown that both the DYNLT light chains form homodimers in yeast two-hybrid and mammalian expression assays (Fig. 2), and with the structural data from bacterially expressed proteins discussed above, we concluded that homodimers of both members of the DYNLT light chain family were capable of directly interacting with all six intermediate chain isoforms. However, as shown is Fig. 4, the DYNLT1-DYNLT3 heterodimers were incapable of binding to the intermediate chain and thus only the homodimers assemble into the dynein complex. This is the first report that only a subset of the light chain dimers are utilized during the assembly of the dynein complex. Whereas it has previously been shown that the two DYNC1LI (light intermediate chain) subunits also incorporate into dynein as homodimers, not heterodimers, the light intermediate chains did not form heterodimers in vivo, in contrast to the DYNLT light chains (13). Our data are also in agreement with the report that the DYNLT1-DYNLT3 heterodimer was not found when dynein was purified from brain (54).
Crystal and NMR structural studies have shown that the two intermediate chain binding domains of the symmetrical DYNLT dimer are formed by contributions from both monomers, thus the DYNLT1 dimer is required for intermediate chain binding (36,38). From the crystal structure of the DYNLT1 dimer bound to an intermediate chain peptide, two DYNLT1 amino acids, His 34 and Asn 38 , were identified that were important for intermediate chain binding (38). Whereas sequence analysis shows that both of these amino acids are conserved in DYNLT3, several of the amino acids near them are not conserved (36,38). This suggests that the differences in the neighboring amino acids may produce an asymmetric environment in the DYNLT1-DYNLT3 heterodimer that does not favor intermediate chain binding.
The DYNLT and DYNLL light chains bind near one another on the N terminus of the intermediate chain, and the structures of the DYNLL1 and DYNLT1 homodimers are very similar (15,35,36,55). This suggested that their presence in dynein might be redundant and that dynein complexes would incorporate only one or the other of these light chains (15). Supporting this hypothesis were genetic studies showing that DYNLT1 is not an essential cytoplasmic dynein component in Drosophila (56). In addition, the Schroer group (4) isolated a part of the dynein complex whose mass indicated the presence of two intermediate chains and only four light chains. Previously, when dynein was purified either by immunoaffinity purification with anti-bodies to the intermediate chain, or by utilizing microtubule binding by the heavy chain, it was not possible to demonstrate that the same dynein complexes contained both the DYNLT and DYNLL light chains (44,46). Our data, obtained by immunoaffinity purification with antibodies to the DYNLT isoforms (Fig. 6), shows that purified mammalian dynein contains the different light chain families. Thus, the DYNLT and DYNLL light chains are present in the same complex. Whereas we were unable to detect a pool of free soluble DYNLT1 or DYNLT3 isoforms (Figs. 1, 5, and 6), we cannot rule out the possibility that they were present in the insoluble fraction or in a very small soluble pool that we could not detect.
When the sedimentation profiles of all the subunits of cytosolic dynein were compared, the fractionation pattern of the DYNLL subunit was atypical (Fig. 5). Consistent with our previous report, most of the endogenous DYNLL did not copurify with dynein (44). These data further support the hypothesis that DYNLL has many roles independent of the dynein complex (22,37,57,58). In particular, it has been suggested that DYNLL is necessary for the dimerization of various proteins (57). Our finding that both DYNLL isoforms bind to all the intermediate chain isoforms is consistent with our finding that both DYNLL isoforms are present when dynein is immunoprecipitated with antibodies to the intermediate chain, but does not support the hypothesis that one DYNLL isoform is associated with dynein and the other with myosin V (46,59).
Assembly of individual subunit isoforms into the dynein complex creates different populations of the motor protein.
Given two genes for each of the five cargo binding subunits, and a total of six intermediate chain alternative splice variants, and assuming that only homodimers are assembled into the complex, then 96 variants of the dynein complexes are possible. One possible functional role for the different populations is to allow specific dynein regulation or cargo binding (2, 19 -22). Our finding that the DYNLT1 and DYNLT3 homodimers, but not the heterodimer, bind to all the intermediate chains has important implications for models of dynein regulation or cargo binding. Because dynein complexes will not contain both light chain isoforms, the light chain homodimers define two different dynein complexes. Many cells express only one intermediate chain isoform, IC-2C, so the ability of both DYNLT homodimers to bind this intermediate chain doubles the number of different dynein complexes in such cells. Whereas there is no specificity in the interactions of the DYNLT homodimers with the intermediate chains, functional specificity of the intermediate chain-defined dynein complexes may be obtained through the limited expression patterns of subunit isoforms. For example, whereas DYNLT1 and DYNLT3 are found in almost all cells and tissues, the expression of the intermediate chain isoforms is cell-and tissue-specific. Intermediate chains encoded by gene 1 are expressed only in neurons (or testis) (18,30,31). Similarly, the DYNLRB2 light chain is not expressed in brain and the light intermediate chains are also not uniformly expressed in all cells and tissues (32,43,46).
In conclusion, we have shown that dynein complexes have multiple light chain families, that the DYNLT and DYNLL isoforms can form all combinations of homodimers and heterodimers, and that all the homodimers bind the six interme-diate chain isoforms. However, the DYNLT heterodimer does not bind the intermediate chain and thus will not assemble into the dynein complex. This result shows that intermediate chain binding limits the incorporation of DYNLT heterodimers into the dynein complex and demonstrates that only a subset of the possible combinations of subunit isoforms can be utilized during the assembly of the dynein cargo binding domain.