790 VOLUME 14 NUMBER 9 SEPTEMBER 2007 NATURE STRUCTURAL & MOLECULAR BIOLOGY

in productive transcription is considerably larger at endogenous genes than the value of ~1% measured in this study. Nonetheless, the results of Darzacq et al.1 are striking and suggest that a substantially larger number of RNA Pol II molecules abort transcription than previously suspected. It should also be noted that the mathematical approach used by Darzacq et al.1 requires assumptions to be made in fitting the data to a specific model. The authors make a very plausible argument in favor of the assumptions inherent in their model, which appears to be the simplest model that is consistent both with their experimental observations and with the known biological properties of transcription. Nonetheless, it remains to be determined whether the model will continue to explain future data when other genes are analyzed using similar methods.

An important goal for the future is therefore to repeat methods of in vivo analysis similar to those reported by Darzacq et al.1 with other gene loci, to determine how general their findings prove to be. Ideally, this will eventually be done using either much smaller gene arrays or, if possible, single-

copy genes, preferably with endogenous DNA sequences and chromatin structure. We look forward also to future technical developments that will facilitate the imaging of single RNA Pol II molecules in living cells. One possibility is that fluorescent speckle microscopy, a live-cell technique that allows visualization of single fluorescent molecules, will enable analysis of DNA transcription by single polymerases5. This would avoid the need to derive molecular models from analysis of populations of molecules. Single-molecule data directly reflect the heterogeneity in populations and thus would provide information about the distribution of paused and transcribing RNA Pol II molecules. Combining single-molecule analysis with either RNA interference–based knockdown or genetic depletion of specific factors could allow the analysis of both paused and transcribing states, thus providing a more detailed understanding of the properties of RNA Pol II during initiation and elongation.

Despite the limitations discussed above, this study represents a crucial milestone on the

way to building a quantitative understanding of the mechanism of transcription in single live cells. Regardless of how general the results presented ultimately prove to be for most endogenous cellular genes, the work of Darzacq et al.1 demonstrates what is already possible in the rapidly emerging field of live-cell fluorescence imaging and points the way to what should become possible in future. It will be important now to determine whether RNA Pol II’s initiation efficiency and pausing frequency are modulated as a regulatory mechanism during transcriptional activation. Once such mechanistic understanding in live cells matures, we will begin to understand properly the systems biology of DNA transcription in vivo.

COMPETING INTERESTS STATEMENTThe authors declare no competing financial interest.

1. Darzacq, X. et al. Nat. Struct. Mol. Biol. 14, 796–806 (2007).

2. Janicki, S.M. et al. Cell 116, 683–698 (2004).3. Sprague, B.L. & McNally, J.G. Trends Cell Biol. 15,

84–91 (2005).4. Dundr, M. et al. Science 298, 1623–1626 (2002).5. Danuser, G. & Waterman-Storer, C.M. Annu. Rev.

Biophys. Biomol. Struct. 35, 361–387 (2006).

When it comes to couple(r)s, do opposites attract?Sharsti Sandall & Arshad Desai

A recent study using electron microscopy provides a detailed view of the oligomerization of a protein complex on the surface of a microtubule polymer. The findings point to a new type of interaction that may be well suited to couple the movement of cargo to dynamic cytoskeletal polymers.

Sharsti Sandall and Arshad Desai are at the Ludwig Institute for Cancer Research, Department of Cellular & Molecular Medicine, University of California, San Diego, CMM East, Room 3052, 9500 Gilman Dr., La Jolla, California 92093-0653, USA. e-mail: abdesai@ucsd.edu

The budding yeast Dam1 complex, comprised of ten subunits, localizes to centromeres and is centrally important for chromosome–spindle microtubule interactions1. Inhibition of this complex leads to extensive chromosome mis-segregation and cell lethality because replicated chromosomes cannot properly connect to opposite poles of the mitotic spindle. Wang et al.2 have used electron microscopy to investigate the binding

of the Dam1 complex to microtubules. Single-particle analysis of the free complex and helical reconstruction of polymer-bound oligomeric complexes reveal a striking conformational change between the monomeric and oligomeric forms. More significantly, the structure of oligomeric Dam1 complex on the microtubule highlights a new mode of association between accessory proteins and cytoskeletal polymers that has broad implications for the functions of the cytoskeleton.

The work of Wang et al.2 was made possible by the biochemical reconstitution of the Dam1 complex, accomplished by coexpression of all ten subunits in bacteria3. This technical breakthrough has laid the groundwork for in vitro biophysical as well as structural studies of Dam1 complex

interaction with microtubules3–8. In prior work, negative-stain electron microscopy has revealed multiple ten-subunit Dam1 complexes associated on the surface of a microtubule, forming both ring-shaped and spiral structures3,6. Assembly of the oligomeric structures is markedly promoted by the microtubule surface, although the free complex does oligomerize at high concentrations. Wang et al.2 extended these earlier studies by first obtaining a 28-Å-resolution view of the free Dam1 complex using single-particle analysis. The free complex exists mostly as dimers of the ten-subunit monomer; each of the monomers has a projecting arm that extends out of a long base (see schematic in Fig. 1a). Reconstructions of mutant-containing complexes reveal that the arm is comprised of

N E W S A N D V I E W S©

2007

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

smb

NATURE STRUCTURAL & MOLECULAR BIOLOGY VOLUME 14 NUMBER 9 SEPTEMBER 2007 791

the C-terminal region of the Dam1 subunit, which is known to be important for Dam1 complex function in vivo and is probably closest to the microtubule. The notion of a protein arm projecting from the complex core is also supported by a biochemical study examining proteolysis and partial complex assembly after the controlled removal of one of the ten subunits8.

Wang et al.2 performed helical reconstructions by exploiting the propensity of the Dam1 complex to form spirals and rings on the microtubule surface. The resulting 30-Å-resolution view reveals striking conformational changes between the free complex and the complex present in the spirals on the microtubule (Fig. 1). This indicates that interaction with the microtubule promotes a conformational change that may in turn facilitate oligomerization. The reconstruction also reveals the presence of two antiparallel spirals comprising the larger spiral, each formed by head-to-tail association of the Dam1 complex. At lower complex concentrations, by contrast, mostly single rings were observed. Although it is unclear whether the antiparallel structure is relevant in vivo, the

structure does indicate that, at least in vitro with purified components, Dam1 oligomer formation is insensitive to the polarity of the underlying microtubule.

The most remarkable finding from the reconstruction is that the symmetry of the Dam1 complex spirals does not match the symmetry of the microtubule—the helical lattices of the two structures could not be related to each other (Fig. 1b). On average, approximately 16 Dam1 complexes were present in a spiral or ring that encircled a 14-protofilament microtubule. A different approach, using scanning transmission electron microscopy to measure the mass of Dam1 complex rings, has suggested that ~25 complexes encircle a microtubule8. The important conclusion from these results is that, unlike all other characterized microtubule-binding motor and nonmotor proteins, the polymer surface of the Dam1 complex does not form a complementarity-mediated interaction with the microtubule. Instead, it seems that microtubules promote the intrinsic propensity of the Dam1 complex to self-associate on their surface, and the resulting structure is not directly docked onto the microtubule. In vivo,

interactions with centromere-localized proteins probably restrict ring formation to the vicinity of the centromere and may even promote ring formation. In photobleaching analyses, the lack of turnover of the Dam1 complex at centromeres, despite rapid turnover in the rest of the mitotic spindle, supports this idea9. The Dam1 complex is also a major target of the centromere-localized Aurora B kinase regulatory pathway, which ensures connection of replicated chromosomes to opposite spindle poles10. Wang et al.2 shed some light on how Aurora B phosphorylation affects Dam1 complex function by analyzing a complex containing a phosphomimic mutant Dam1 subunit. Although no change was detected in the structure of the unbound mutant complex, its propensity to oligomerize was reduced, suggesting that Aurora kinase phosphorylation modulates Dam1 complex self-association.

What is responsible for the interaction between the Dam1 complex and the microtubule, if it does not involve docking onto a surface feature of the lattice? The present view on this is confusing, as two groups have reached opposite conclusions

Microtubule(cross-section)

Dam1 complexspiral/ring

Dam1C terminus

FreeMicrotubule-

bounda

b

c

Dam1 complex

αβ -tubulin

Centromere

Linkercomplex

Direction ofchromosomemovement

Coupling todepolymerization

Tension-promotedpolymerization

Lowtension

Hightension

Figure 1 Structural analysis of the Dam1 complex defines a new type of interaction with microtubules. (a) The majority of free heterodecameric Dam1 complexes exist as dimers. Each ten-subunit monomer has a projecting arm composed of the C terminus of the Dam1 subunit. Helical reconstruction of spirals formed by the complex on microtubules reveals that striking conformational changes, involving distinct domains of the ten-subunit monomer, are needed to fit the structure of the free complex into that of the microtubule-bound complex. (b) Schematic view of one turn of a Dam1 complex spiral on the microtubule. The symmetries of the Dam1 oligomer and the microtubule are, surprisingly, not related to each other. (c) Lack of a fixed docking site for the Dam1 complex on the microtubule lattice suggests an appealing mechanism for coupling microtubule dynamics to chromosome movement. Left, coupling to depolymerization, with the intrinsic instability of the microtubule driving movement. Right, tension-dependent promotion of polymerization, which was observed during chromosome alignment in vivo (tension is generated by the physically connected sister centromere, which is not depicted, moving in the opposite direction). Both activities were observed with pure Dam1 complex bound to beads in vitro. In vivo, a linker complex must connect the Dam1 oligomer to the centromere to move chromosomes. The depicted linker complex, most probably the four-subunit Ndc80 complex, is shown associating with the microtubule lattice and with the Dam1 complex, as indicated by other work (reviewed in ref. 1).

N E W S A N D V I E W S©

2007

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

smb

792 VOLUME 14 NUMBER 9 SEPTEMBER 2007 NATURE STRUCTURAL & MOLECULAR BIOLOGY

from the same experiment (giving rise to the question mark in the title of this piece). The heterodimeric subunits of microtubules are formed from α- and β-tubulin, which have highly charged acidic C-terminal tails. These tails are important for mediating electrostatic interactions with basic regions of microtubule-associated proteins. Studies using tail-less microtubules generated by protease treatment as binding substrates have yielded differing opinions on whether an electrostatic interaction between the projection arm of the Dam1 complex and the highly charged acidic tubulin tails is important for association of the Dam1 complex with the microtubule surface3,6,8. Although the electrostatic mechanism favored by Wang et al.2 and supported by their previous study is appealing, Miranda et al.8 reached opposite conclusions, leading to a troubling discrepancy that needs to be resolved.

What is the biological relevance of the new type of association documented by the structural analysis of the Dam1 complex? GTP hydrolysis on β-tubulin accompanies the polymerization of tubulin11. A majority of the energy of hydrolysis is stored in the polymer lattice and is released during rapid depolymerization12. Effective harnessing of depolymerization to drive movement requires

couplers than can transduce the released energy to a cargo, such as a chromosome. During chromosome alignment, it is also important for each chromatid to remain bound to a polymerizing microtubule that is under tension, and a coupler that promotes polymerization while under tension must provide this activity. The Dam1 complex, which localizes to the interface between the centromere and the spindle in vivo, both couples movement to depolymerization and shows tension-dependent stimulation of polymerization in vitro4,5,7. The type of structure described by Wang et al.2 is ideally suited to perform such functions, as the ring or spiral on the surface is not directly docked onto the tubulin subunits of the polymer, making it feasible for the entire oligomer to slide along the polymer lattice (Fig. 1c). Future biophysical, modeling and structural studies will reveal whether this type of mechanism explains the intriguing activities of the Dam1 complex, and mutational studies will help relate in vitro findings to cellular functions.

The results of Wang et al.2 extend our understanding of the Dam1 complex, but the more exciting implications relate to the nature of the association that they have discovered. Although no counterparts of the Dam1 complex have yet been

identified outside of fungi, the idea that a polymer surface can promote oligomerization to form a coupler is very appealing and will influence analysis of the large number of accessory proteins that interface with the cytoskeleton.

COMPETING INTERESTS STATEMENTThe authors declare no competing financial interest.

1. Westermann, S., Drubin, D.G. & Barnes, G. Annu. Rev. Biochem. 76, 563–591 (2007).

2. Wang, H.W. et al. Nat. Struct. Mol. Biol. 14, 721–726 (2007).

3. Miranda, J.J., De Wulf, P., Sorger, P.K. & Harrison, S.C. Nat. Struct. Mol. Biol. 12, 138–143 (2005).

4. Asbury, C.L., Gestaut, D.R., Powers, A.F., Franck, A.D. & Davis, T.N. Proc. Natl. Acad. Sci. USA 103, 9873–9878 (2006).

5. Franck, A.D. et al. Nat. Cell Biol. 9, 832–837 (2007).

6. Westermann, S. et al. Mol. Cell 17, 277–290 (2005).

7. Westermann, S. et al. Nature 440, 565–569 (2006).

8. Miranda, J.J., King, D.S. & Harrison, S.C. Mol. Biol. Cell 18, 2503–2510 (2007).

9. Joglekar, A.P., Bouck, D.C., Molk, J.N., Bloom, K.S. & Salmon, E.D. Nat. Cell Biol. 8, 581–585 (2006).

10. Cheeseman, I.M. et al. Cell 111, 163–172 (2002).11. Desai, A. & Mitchison, T.J. Annu. Rev. Cell Dev. Biol.

13, 83–117 (1997).12. Grishchuk, E.L., Molodtsov, M.I., Ataullakhanov, F.I.

& McIntosh, J.R. Nature 438, 384–388 (2005).

Chloride finds its place in the transport cycleSusan G Amara

Neurotransmitter:sodium symporters, which use sodium gradients for the coupled uptake of sodium and neurotransmitters during synaptic transmission, often display a chloride dependence. New data from two separate groups identify the chloride-binding site for this family of symporters and suggest that the chloride charge facilitates sodium binding and substrate transport.

Neurons and glial cells accumulate neurotransmitters by a sodium-coupled transport process that is in many respects similar to systems used by other cells and organisms for concentrating amines and amino acids. By cotransporting a solute with one or more sodium ions, the energy

stored in the transmembrane electrochemical gradient can be harnessed to drive the solute into the cell. However, many members of the SLC6 family of neurotransmitter:sodium symporters have an additional ionic requirement in that the transport activity of these carriers also depends on chloride ions1. Mechanistic understanding of this particular family of transporters is of interest because they are often targets for antidepressants and other psychoactive drugs2. Two recent reports, by Forrest et al.3 and Zomot et al.4, illuminate the basis for the curious, less

well-understood chloride dependence by identifying structural elements crucial for the binding of chloride. Although the two groups emphasize different transporters and approaches, both arrive at remarkably similar explanations for the coupling of chloride and sodium ions during transport: a negative charge contributed either by a chloride ion or by the transporter itself appears necessary for binding of a sodium ion to its primary binding site.

The new work was inspired by pioneering structural studies of a number of membrane

Susan G. Amara is at the Department of Neurobiology, University of Pittsburgh School of Medicine, Pittsburgh, PA 15261, USA. e-mail: amaras@pitt.edu

N E W S A N D V I E W S©

2007

Nat

ure

Pub

lishi

ng G

roup

ht

tp://

ww

w.n

atur

e.co

m/n

smb