The molecular architecture of the Dam1 kinetochore complex is defined by cross-linking based structural modelling

Abstract

Accurate segregation of chromosomes during cell division is essential. The Dam1 complex binds kinetochores to microtubules and its oligomerization is required to form strong attachments. It is a key target of Aurora B kinase, which destabilizes erroneous attachments allowing subsequent correction. Understanding the roles and regulation of the Dam1 complex requires structural information. Here we apply cross-linking/mass spectrometry and structural modelling to determine the molecular architecture of the Dam1 complex. We find microtubule attachment is accompanied by substantial conformational changes, with direct binding mediated by the carboxy termini of Dam1p and Duo1p. Aurora B phosphorylation of Dam1p C terminus weakens direct interaction with the microtubule. Furthermore, the Dam1p amino terminus forms an interaction interface between Dam1 complexes, which is also disrupted by phosphorylation. Our results demonstrate that Aurora B inhibits both direct interaction with the microtubule and oligomerization of the Dam1 complex to drive error correction during mitosis.

Figure 1. Structural model of the Dam1 complex.

Structural model of the Dam1 complex in the (a) absence and (b) presence of MTs. Each model represents the centre of the single cluster found in the top 1,000 models. Four orientations are shown. Each panel shows the Dam1 EM structure EMD-1372 for reference (left) followed by 3 differently coloured versions of the Dam1 structural model. The left version shows all the beads coloured according to protein (key given at the bottom of the figure). The centre version highlights only those residues that cross-link to MTs. Transparency is used to make the MT-binding region visible. The right-hand version removes the transparency so that the proximity of the MT-binding region to the model's surface can be clearly seen. It is noteworthy that in the absence of MTs, the MT-binding region is in the interior of the structure.

Figure 2. The number and overlap of distance restraints generated by cross-linking experiments on the Dam1 complex in the presence and absence of MTs.

Per cent fit of distance restraint data with the minus MT and plus MT models generated is shown in boxes. Data are shown for peptides with Percolator assigned q-values ≤0.01. A distance restraint is considered a ‘fit' if the modelled bead surfaces of the two beads that included the cross-linked residues are within 10 Å. The minus MT model was generated using the 678 UDRs observed in the absence of MTs. The plus MT model was generated using the 458 UDRs observed in the presence of MTs. Per cent fit is shown for the data used to generate the models (green arrows) and for the data unique to the alternate condition (red arrows).

Figure 3. The precision of our modelling approach in determining the positions of all beads.

(a,b) Density maps expressing the probability that a given region of space is occupied by a bead in the top 1,000 scoring models. The space was discretized in bins of 2 Å size. The density maps of each bead are shown at half their maximum value. It is noteworthy that almost all beads are well determined and therefore look similar in the probability density map and the representative model. An exception would be the Dad2p beads (salmon color) in the presence of MTs (bottom right image). Only a Dam1 complex monomer is shown for clarity. (c) A cumulative plot of bead variability.

Figure 4. Dam1 complex to MT cross-links.

Peptide sequence coverage (coloured boxes), mono-links (coloured vertical lines and circles) and lysines (vertical white lines) are also shown. Dam1p to Duo1p and tubulin to tubulin cross-links are hidden for clarity. Data are shown for peptides with Percolator assigned q-values ≤0.01.

Figure 5. Cross-link based flexible docking of the Dam1 complex model onto the MT specifies a preferred orientation.

(a) The model of the Dam1 complex on the MT with beads coloured by protein (Hsk1p is not visible at this angle). (b) Only those beads found cross-linked to the MT are highlighted. (c) The orientation of the Dam1 complex on the MT with respect to the orientation of the Ndc80 complex. The C-terminal MT-binding region of the Ndc80 complex is nearer the minus end of the MT. Aravamudhan et al. measured FRET between N-Nuf2 and six Dam1 complex C termini. The proteins in the key to the left and right of the model represent these results, which indicates proximity between the C termini of Dam1p, Dad4p and Dad3p to the N terminus of Nuf2, while suggesting that the C termini of Spc34p, Ask1p and Dad1p are further away. Only a Dam1 complex monomer is shown for clarity. (d) Attempts to dock the model of the Dam1 complex in the minus MT conformation resulted in a poor fit and an orientation incompatible with a ring.

Figure 6. The Dam1 complex MT-binding domain.

Structural models show that the C-terminal regions of Dam1p and Duo1p interact with the N-terminal regions of Dam1p and Duo1p in the absence of MTs. On binding MTs, these interactions are lost, freeing the C-terminal regions of Dam1 and Duo1, and allowing them to bind the MT. (a) Dam1N (light yellow) to Dam1C (dark yellow); (b) Duo1N (turquoise) to Duo1C (blue); (c) Dam1C (dark yellow) to Duo1N (turquoise); (d) Dam1N (light yellow) to Duo1C (blue). The Dam1 complex is shown as a monomer.

Figure 7. The Dam1 complex to Dam1 complex interface.

In all panels, the Dam1 complex is shown as a dimer, with one monomer in red and the other in grey. (a) The interface between the two monomers is formed by multiple interactions between Spc19p and Spc34p in both the presence and absence of MTs. (b) In the presence of MTs, Duo1p more than doubles its interactions across the interface by binding to Spc19p and Ask1p. (c) On binding to MTs, Dam1N gains interactions with Ask1p and Dam1M, and Dam1C lose interactions with Ask1N. (d) The Aurora B kinase phosphorylation site Dam1p S20 lies at the interface between the two Dam1 complex monomers. Dam1p S20 beads are coloured yellow.

Figure 8. Aurora B kinase (Ipl1) phosporylation controls both MT binding and oligomerization of the Dam1 complex.

(a) Schematic of phospho-blocking Dam1 complex constructs. Both constructs were phosphorylated by Ipl1 kinase in the presence of ATP. Mock-treated Dam1 complex had ATP substituted with distilled water. (b) Survival curve summary of single-molecule Dam1-GFP TIRF experiments. Dam1 NP mock (n=638), Dam1 NP phos (n=1009), Dam1 CP mock (n=777) and Dam1 CP phos (n=646) were each incubated at 40 pM for single-molecule Dam1–GFP complex imaging. Inset: GFP fluorescence (A.U.) distributions for Dam1 NP mock (3,500±1,200), Dam1 NP Phos (3,300±1,100), Dam1 CP mock (4,300±1,500) and Dam1 CP phos (3,900±1,400). (c) Survival curve summary of tracer Dam1-GFP TIRF experiments. GFP-tagged Dam1 NP mock (20 pM; n=360), Dam1 NP phos (n=587), Dam1 CP mock (n=294) and Dam1 CP phos (n=454) constructs were incubated with 2 nM corresponding non-tagged Dam1 constructs. Inset: GFP fluorescence (A.U.) distributions for Dam1 NP mock (3,700±1,300), Dam1 NP phos (3,700±1,300), Dam1 CP mock (3,600±1,400) and Dam1 CP phos (3,500±1,400) in tracer experiments. Inset panels show initial brightness of Dam1–GFP complex are similar across the whole TIRF data set, indicating that all data are from imaging single molecules of Dam1–GFP complexes either alone or incorporated into oligomers of unlabelled Dam1 complexes (in the tracer experiments). Vertical dashed line in inset panels represents the average height of single-step photobleach events under identical conditions (3,900±1,900 A.U., n=287). Average single-step photobleach duration under identical conditions is 130.9±5.7 s (n=287). All errors are s.d. of the mean.