Conserved Structures of Mediator and RNA Polymerase II Holoenzyme

See allHide authors and affiliations

Science  12 Feb 1999:
Vol. 283, Issue 5404, pp. 985-987
DOI: 10.1126/science.283.5404.985


Single particles of the mediator of transcriptional regulation (Mediator) and of RNA polymerase II holoenzyme were revealed by electron microscopy and image processing. Mediator alone appeared compact, but at high pH or in the presence of RNA polymerase II it displayed an extended conformation. Holoenzyme contained Mediator in a fully extended state, partially enveloping the globular polymerase, with points of apparent contact in the vicinity of the polymerase carboxyl-terminal domain and the DNA-binding channel. A similarity in appearance and conformational behavior of yeast and murine complexes indicates a conservation of Mediator structure among eukaryotes.

Mediator was isolated as a complex of nearly 20 proteins required to support transcriptional activation in a fully reconstituted yeast RNA polymerase II transcription system (1,2). Mediator subunits fall into three groups: Srb proteins (3), a Sin4/Rgr1 group shown to form a distinct module that functions in repression as well as in activation (4, 5), and a group of proteins termed “Med” for Mediator (5). A related complex, containing four homologs of yeast Mediator subunits, has been isolated from murine cells (6). Yeast and murine complexes interact with RNA polymerase II through the polymerase's COOH-terminal domain (CTD) to form a 1.5-MD holoenzyme (2, 3, 5). We report here on the structures of Mediator and the holoenzyme as seen in projection.

Yeast Mediator, resolved to homogeneity as described (7,8), was adsorbed to carbon-coated grids and negatively stained with uranyl acetate for imaging in the electron microscope. Fields of many similar particles were observed, which suggests that Mediator exists as a discrete entity, as was previously inferred from biochemical studies. A large fraction of the particles appeared to be nearly identical, which is indicative of a preferred orientation on the grid (9). Images of such particles were aligned and averaged with the use of the SPIDER suite of programs (10, 11). The resulting view of the Mediator structure in projection revealed details to about 40 Å resolution (Fig. 1). The structure measured about 400 Å in height and appeared roughly triangular in shape, with a distinct dense domain at the base.

Figure 1

Yeast Mediator in projection. Images of 54 individual particles were aligned and averaged. Scale bar, 200 Å.

When Mediator and RNA polymerase II were mixed in equimolar (0.05 μM) amounts, about 75% of the Mediator associated with polymerase to form holoenzyme. Many images of the remaining unassociated Mediator exhibited a partially extended conformation. Three domains, barely distinguishable in the compact form of Mediator found in the absence of polymerase, were linearly arrayed, with variable angles between them (Fig. 2).

Figure 2

Contrast-enhanced images of single yeast Mediator particles. Closed and progressively more extended conformations of Mediator (A through D) are shown. Part of an RNA polymerase II molecule is included alongside the Mediator particle in the lower righthand corner of (D). The background around the Mediator particles was partially blocked for ease of viewing.

Alignment and averaging of holoenzyme images revealed Mediator in a fully extended form (Fig. 3). Mediator was identified as an arc of protein density on the periphery, on the basis of the similarity in appearance and dimensions to the partially extended form (Fig. 2). The three Mediator domains were clearly distinguishable in the holoenzyme, the largest of which is referred to here as the “head,” presumably corresponding to the dense domain at the base of Mediator in the compact form, followed by “middle” and “tail” domains.

Figure 3

Holoenzyme formed by yeast Mediator and RNA polymerase II in projection. Images of 73 particles were aligned and averaged to calculate the map. Mediator is in an extended conformation, with head (h), middle (m), and tail (t) domains clearly distinguishable. The globular density embraced by Mediator is identified as RNA polymerase II. The outline of a projection of the previously determined polymerase 3D structure is superimposed (dark line), with the point of attachment of the CTD (dark circle) and the location of the DNA-binding channel (c) indicated. This alignment is speculative because of the low (∼40 Å) resolution of the holoenzyme map and possible conformational changes of the polymerase upon interaction with Mediator.

The central globular density in the image of the holoenzyme was identified as RNA polymerase II by comparison with images of the polymerase alone and with the three-dimensional (3D) structure of the enzyme, previously determined at 16 Å resolution from two-dimensional (2D) crystals in negative stain (12). A view of the polymerase structure in projection was found, whose outline (dark line in Fig. 3) roughly matched that of the polymerase density in the holoenzyme and that placed the CTD region of the polymerase structure, also known from 2D crystallography (13) (black dot in Fig. 3), adjacent to the Mediator density of the holoenzyme. With the polymerase in this orientation, the CTD contacted the Mediator middle domain. A second contact with Mediator was formed between a region of the polymerase in the vicinity of the DNA-binding channel (arrow inFig. 3) and the Mediator head domain. There was no apparent contact between the polymerase and the Mediator tail domain, which was nonetheless well localized, suggesting conformational rigidity of the extended Mediator structure.

As a control, Mediator was mixed with RNA polymerase II lacking a CTD and examined in the electron microscope (14). No holoenzyme was observed, although some Mediator particles appeared to be partially extended. Inasmuch as unfolding only occurred in the presence of polymerase, unfolding in the absence of a CTD was consistent with a second site of Mediator-polymerase interaction. Also, no unfolding was observed upon incubation of Mediator with a CTD peptide, and neither holoenzyme formation nor unfolding of yeast Mediator occurred in the presence of mammalian RNA polymerase II. An activator protein that functions through Mediator, Gal4-VP16, also failed to cause unfolding. Mediator persisted in the compact configuration under a range of solution conditions, except for prolonged incubation at 40°C, which caused the apparent loss of the headpiece from many Mediator particles, and exposure to a pH above 8.5, which caused extensive unfolding, even in the absence of polymerase. Unfolding at elevated pH was reversed by restoration of neutrality, although the refolded Mediator appeared slightly different from the original compact form.

The related murine complex, containing homologs of yeast Mediator subunits (6), appeared similar to yeast Mediator in the electron microscope. Fields of well-defined particles gave evidence of a discrete complex, and selection, alignment, and averaging yielded a view in projection. The murine complex was not only the same size and shape as yeast Mediator but also bore a close resemblance in internal structural detail (Fig. 4). The very occurrence of such a view, deriving from a preferred orientation on the surface of the electron microscope grid, points to a similarity in 3D structure of the two complexes, leading to a similar manner of adsorption to the grid.

Figure 4

Murine Mediator in projection (A), compared with its yeast counterpart (B). Images of 66 individual particles, purified from murine hybridoma cells as described (6), were aligned and averaged. The maps are displayed as contour plots to emphasize their main features. Scale bar in (B), 200 Å.

The murine complex also underwent a conformational change to an extended form at high pH or in the presence of RNA polymerase II (14). Unfolding was again species specific, occurring only with mammalian but not with yeast RNA polymerase II. The proportion of extended particles was, however, smaller, and the images of holoenzyme were more variable, precluding alignment and averaging. The similarity to yeast holoenzyme was nonetheless apparent from views of individual particles.

Parallel investigation of the murine and yeast complexes was of particular interest because of the apparent differences between them. Although four components of the murine complex are homologous to yeast Mediator subunits, there is as yet no evidence of further homology, with a murine equivalent of the Sin4/Rgr1 repression-activation module being conspicuously lacking (6). The structural similarities noted here point to a correspondence between the two complexes: Both occur in a compact conformation of about the same size and shape and exhibit similar internal structural detail; both unfold to an extended conformation at elevated pH or in the presence of RNA polymerase II; and both interact with polymerase in the extended conformation, forming holoenzymes that are similar in general appearance. Moreover, both complexes include a tail domain, as identified in a preliminary study of yeast Mediator from asin4 deletion strain as the Sin4/Rgr1 repression-activation module (15). Thus, the murine complex appears to be a true counterpart of yeast Mediator.

Unfolding of Mediator in the presence of RNA polymerase II may reflect the equilibrium association of the two proteins detected previously by biochemical means (16). Mediator unfolds upon binding polymerase and fails to refold immediately after dissociation. The evidence for at least two sites of interaction with polymerase was unexpected. It was previously thought that the interaction occurred solely through the polymerase CTD. The species specificity of Mediator unfolding, as well as its persistence in the absence of the CTD, provide evidence for a second site. Apparently both sites are required for a stable interaction.

  • * Present address: Department of Cell Biology MB 25, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA.

  • Present address: Institute of Medical Biochemistry, Göteborg University, Medicinaregatan 13, 405 30 Göteborg, Sweden.

  • To whom correspondence should be addressed. E-mail: kornberg{at}


View Abstract

Navigate This Article