Structure of the Blm10-20S Proteasome Complex by Cryo-electron Microscopy. Insights into the Mechanism of Activation of Mature Yeast Proteasomes

Abstract

The 20S proteasome is regulated at multiple levels including association with endogenous activators. Two activators have been described for the yeast 20S proteasome: the 19S regulatory particle and the Blm10 protein. The sequence of Blm10 is 20% identical to the mammalian PA200 protein. Recent studies have shown that the sequences of Blm10 and PA200 each contain multiple HEAT-repeats and that each binds to the ends of mature proteasomes, suggesting a common structural and biochemical function. In order to advance structural studies, we have developed an efficient purification method that produces high yields of stoichiometric Blm10-mature yeast 20S proteasome complexes and we constructed a three-dimensional (3D) model of the Blm10-20S complex from cryo-electron microscopy images. This reconstruction shows that Blm10 binds in a defined orientation to both ends of the 20S particle and contacts all the proteasome α subunits. Blm10 displays the solenoid folding predicted by the presence of multiple HEAT-like repeats and the axial gates on the α rings of the proteasome appear to be open in the complex. We also performed a genetic analysis in an effort to identify the physiological role of Blm10. These experiments, however, did not reveal a robust phenotype upon gene deletion, overexpression, or in a screen for synthetic effects. This leaves the physiological role of Blm10 unresolved, but challenges earlier findings of a role in DNA repair.

INTRODUCTION

The 20S proteasome in yeast is a cylindrical particle of 28 subunits arranged as four heptameric stacked rings.^1,2,3 The two rings at the ends, the α rings, comprise seven catalytically inactive subunits that are similar but distinct proteins, named α1 to α7. Similarly, the two central rings, the β rings, are formed by seven different β subunits, three of which house active sites at their N-termini.³ In order to fulfill all its different functions, the 20S proteasome binds to a number of proteins that deliver substrates, stimulates the peptidase/protease activity and may target the proteasome complex to specific locations in the cell. These include the 19S regulatory particle (RP),^4,5 the 11S activator (PA28/PA26/REG)^6,7,8,9 (not present in yeast) and PA200,¹⁰ all of which bind to the ends of the 20S proteasome. The two ends of the cylindrically shaped 20S proteasome can bind either the same or different regulatory particles forming homogeneous or hybrid proteasomes respectively.^10,11,12,13

The PA200 activator was initially found in mammals¹⁰ and electron microscopy (EM) studies showed that it binds to the ends of the 20S proteasome and opens the axial channel into the proteolytic cavity.¹⁴ The PA200 sequence contains multiple HEAT-like repeats suggesting that the protein adopts a solenoid structure.¹⁵ The sequences of PA200 and the yeast Blm10 (formerly known as Blm3^16,17) proteins are 20% identical (41% similar) over the 1715 C-terminal residues of each protein. Conservation of this class of proteasome activators from yeast to humans suggests an important and evolutionary conserved biological role. Like PA200, Blm10 binds to the ends of mature 20S proteasomes and is predicted to contain HEAT repeats.¹⁸ Despite the obvious similarities of Blm10 and PA200, published data present a confusing picture. Blm10 was initially identified as a factor important for repairing specific types of DNA damage,^17,19 and some features of mammalian PA200 are consistent with such a function.¹⁰ A truncated fusion protein was found to be mislocalized and cells with this mutant protein showed to be somewhat sensitive to the DNA damaging agent bleomycin, but deletion mutations did not share this sensitivity.¹⁸ Recently, it was reported that PA200 is essential for spermatogenesis in mice.²⁰ It has also been shown that Blm10 interacts with nascent but not mature 20S proteasomes and that it functions in the coordination of proteasome assembly in yeast.¹⁶ Finally, yeast cells lacking the ability to upregulate proteasome subunit expression due to deletion of the RPN4 gene have been reported to be more sensitive than normal to low levels of canavanine, consistent with a defect in processing misfolded proteins.¹⁸ Thus, the physiological role(s) of PA200 and Blm10 and the extent to which the functional roles have been conserved throughout evolution remain uncertain.

Here, we report an efficient purification method able to yield abundant stable Blm10 complexes with mature yeast 20S proteasomes that are stoichiometrically assembled and suitable for structural studies. This allowed us to obtain images of Blm10-20S complexes by cryo-electron microscopy and to calculate a three-dimensional model of the yeast 20S proteasome with a Blm10 molecule bound at each end. This extends the resolution of the structural model beyond that available from negatively stained samples.¹⁸ The 3D-map shows that Blm10 binding to the ends of the 20S proteasome opens the axial gates on the α rings in a manner similar to its mammalian homolog PA200. This result shows that both activators share a similar proteasome activation mechanism. Moreover, the map is consistent with the solenoid predicted by the presence of multiple HEAT-like repeats in the Blm10 sequence.¹⁵ We also describe genetic experiments to investigate the role of Blm10 in DNA damage repair and in growth at high temperatures. In contrast to earlier reports,^17,18,19 our results do not support a significant role for Blm10 in DNA damage repair or resistance to stresses that cause misfolding of proteins.

RESULTS

Purification of stable and active Blm10-20S proteasome complexes

Recombinant N-terminally histidine-tagged Blm10 was expressed in yeast (S. cerevisiae) and purified by Nickel affinity chromatography. Incubation of Blm10 with yeast 20S proteasome that was bound to IgG-sepharose beads by a protein A affinity tag, resulted in formation of stable Blm10₂-20S (doubly capped) complexes that were released from the beads by incubation with TEV protease (Figure 1(a) and 1(b)).

Figure 1. Analysis of active Blm10-20S proteasome complexes.

(a) Representative Blm10₂:20S samples were compared by denaturing polyacrylamide gel conditions one and 21 days after completing the preparation. Upon storage, purified Blm10 becomes processed, predominantly to two large fragments that roughly correspond to the N and C-terminal halves of the protein. Asterisks indicate a proteasome 20S subunit that also becomes processed during storage. (b) Native gel analysis of the same samples indicates that the covalently separated products remain associated in a stable complex with 20S proteasome. (c) Equivalent samples run in parallel are shown developed in an in gel peptidase assays show approximately the same level of Blm10-mediated activation for one and 21 day old complexes. Although not apparent in this sample, the position to which singly capped Blm10p-20S complexes migrate is also indicated (arrow). This assay was modified from⁴³ using 4% nondenaturing gels run for 200 Volt-Hours in 0.5X Tris-Borate. The gel was incubated in 10 ml of 0.01M sLLVY-AMC substrate in assay buffer (20 mM Na/K phosphate pH 7.5, 50 mM NaCl, 1 mM DTT) for 30 min at 30°C prior to visualization using a UVP BioDocIt system. (d) Solution assays also show equivalent levels of activation for complexes one and 21 days after preparation. This assay was optimized from²¹ to accommodate a 96 well format and a Polarstar Optima (BMG Labtech) fluorimeter. 100 ng or 169.8 ng of double headed complex in 90 μl of assay buffer were pre-incubated at 30°C for 12 minutes prior to adding 10 μl of 1 mM sLLVY-AMC substrate for a final concentration of 100 μM. After addition of substrate, incubation at 30° was continued and fluorescence was read every 5 minutes for 1 hour, then SDS was added to a final concentration of 0.01% and fluorescence was followed for another hour to verify that the 20S was still capable of being activated (data not shown). Fluorescence/minute was calculated using values within the linear range (CV=1.0, n=6 for each sample).

Blm10 is a single polypeptide chain of 2144 residues. It undergoes proteolytic cleavage upon storage to yield predominantly two large fragments visualized by SDS-PAGE. Each of the two large Blm10 fragments (Figure 1(a)) were shown by N-terminal sequencing to comprise at least two different starting points and represent roughly the N-terminal (N-terminally sequenced as Asn¹⁸⁸AsnThrAlaGlu and Gly²²⁸LysIleThrAla) and the C-terminal (Ile¹⁰⁸⁷GluAlaGly, Ser¹¹⁴⁰AspLeuAlaPhe) halves of Blm10. Due to the broad appearance of these bands, it is likely that they comprise additional similar N-terminal and C-terminal fragments that were not detected by sequencing.

This clipping of Blm10 into covalently separated sections presumably occurs at exposed surface regions and does not appear to significantly alter the structure as judged by native gel analysis (Figure 1(b)). This is consistent with inspection of amino acid sequences; the N-terminal ~500 residues of Blm10 are poorly conserved and residues from ~1040 to 1140 lack counterparts in other species, suggesting that they form an exposed loop. It therefore seems likely that these regions can be processed without impacting structurally important interactions within Blm10 or between Blm10 and the 20S proteasome.

Solutions of 20S proteasome or Blm10-20S complex that contain equal amounts of 20S were tested for their ability to hydrolyze a fluorogentic peptide substrate (Figures 1(c) and 1(d)). We find that Blm10 stimulates hydrolysis of a peptide substrate approximately 4-fold. This level of activation is lower than that reported for members of the PA28/11S family of proteasome activators,^21,22,23 but it is reproducible and it is seen both in gel overlay (Figure 1(c)) and in solution (Figure 1(d)) assays. The level of activation is essentially identical in one-day old preparations that show only slight proteolysis of Blm10 and in 21-day old preparations that show almost complete processing to the two major intermediate species. This further supports the conclusion that processing of Blm10 does not substantially alter the structure in the complex with 20S proteasome.

Cryo-electron microscopy of Blm10-20S proteasome complexes

The purified complexes were analyzed by cryo-electron microscopy. Nearly 95% of the observed particles were oriented with the longitudinal axis parallel to the grid, as a consequence of pretreatment of the grids with polylysine.²⁴ This orientation made it possible to visually distinguish between Blm10₂-20S (doubly capped) and Blm10-20S (singly capped) complexes.

EM images most frequently showed Blm10₂-20S complexes that contained the 20S proteasome with a Blm10 molecule bound to each end (Figure 2). A smaller number of Blm10-20S complexes were observed, as well as rare free 20S proteasome particles.

Figure 2. Cryo-electron microscopy of Blm10₂-20S proteasome complexes.

Representative micrograph of Blm10-20S and Blm10₂-20S proteasome complexes embedded in vitreous ice.

These results further demonstrate that our purification method was very efficient at producing high yields of stable Blm10 complexes with mature 20S proteasomes.

Three-dimensional structure of the Blm10₂-20S complex

Recent cryo-electron microscopy studies have shown that PA200 opens the axial channel of the 20S proteasome upon binding to its ends in a fashion similar to that previously described for PA26.⁸ To determine whether Blm10 protein activates the 20S proteasome in similar fashion, we used cryo-EM to build a 3D model of the Blm10₂-20S proteasome complex.

The reconstruction of the Blm10₂-20S complex was calculated by the projection matching approach from several initial models and attempting many variations of the reconstruction process. Regardless of the initial map and the detailed protocol for the refinement process, we consistently obtained essentially the same final result. The resolution calculated for the refined 3D model was 18 Å.

The Euler angle distribution of aligned particle images within the asymmetric triangle revealed some degree of preferential orientation. The large initial number of particles used for the reconstruction (86,425) assured that most class averages contained at least 100 particle images and therefore, a complete description of the complex was obtained. However, the preferential orientation of the particles produced anisotropic resolution in the three-dimensional model with views described by overrepresented projections showing slightly higher resolution than others (Figure 3(a) versus 3(b), respectively). The number of particles in overpopulated projections was also limited to 400 particles to avoid over-representation of images in class averages, a problem that is difficult to compensate for with weighting factors used to place the class averages within a 3D Fourier volume. The number of particles incorporated in the final three-dimensional model was 47,567.

Figure 3. Three-dimensional reconstruction of the Blm10-20S proteasome complex.

Surface representation of the reconstruction of the Blm10₂-20S proteasome complex at 18 Å resolution compared with the x-ray structure of yeast 20S proteasome (PDB code:1RYP) limited at similar resolution. From (a) to (d) different side views of the complex are shown side by side with the analogous view of the 20S proteasome. The contour used for the representation of the Blm10₂-20S complex displays a molecular volume corresponding to a complex of 1.2 MDa. Blm10 are the non-symmetrical caps (colored in light brown) bound to the ends of the proteasome (colored in green). The two molecules of Blm10 contact the α rings at both ends of the proteasome in the same manner. All the α subunits are contacted by the activator.

One molecule of Blm10 binds to each end of the yeast 20S proteasome as an asymmetric cap (Figure 3). Viewed along the longitudinal axis, Blm10 showed the solenoid folding predicted by bioinformatics studies (Figure 4(a), left and center panels) and the thickness of the solenoid is very similar to the one observed in other HEAT repeat containing proteins such as karyopherin-β2²⁵ (Figure 4(a), center and right panels). The solenoid is bound to the α-ring, contacts each of the α subunits, and appears to spiral along the 7-fold pseudo-symmetry axis of the 20S proteasome. This visualization of the solenoidal structure bound to Blm10 is consistent with and extends the resolution of the 2D negative stain analysis of Schmidt et al.¹⁸ Although no symmetry was applied during the construction of the model, Blm10 caps at both ends were very similar and had the same shape (Figure 3).

Figure 4. Visualization of the solenoid folding of the Blm10 protein. Opening of the proteasomal axial channel upon Blm10 binding.

(a) Surface representation of the axial view of the two Blm10 caps in a Blm10₂-20S proteasome complex (left & center panels) compared with the HEAT-repeat containing structure of karyopherin-β2 protein (PDB code:1QBK) blurred to 18 Å resolution (right panel). Double-headed arrows span the thickness of the solenoid in both proteins. Part of the atomic structure of karyopherin-β2 (residues Tyr 3 to Leu 394) was docked into the solenoid path of the Blm10 cap (center panel). (b) Sagital cut of the surface rendering representation of the Blm10₂-20S proteasome complex. The light brown Blm10 caps at both ends of the 20S proteasome feature an internal cavity communicated with the digestion chamber of the proteasome by an opening of the axial channel. (c) Central section of the density maps of the Blm10₂-20S (left panel) and Blm10-20S (right panel) complexes. Density in the center of the α rings in contact with Blm10 (white arrows) in both complexes is significantly lower than in the free α ring (yellow arrow) in the asymmetric complex.

When the three-dimensional reconstruction was cut sagitally, the Blm10 caps featured a cavity that connects with the pre-chambers of the 20S proteasome by an opening on the axial channel. The opening of the axial channel was evident both in a surface rendering representation (Figure 4(b)) and in a central section of the density map (Figure 4(c), left panel, white arrow). The center of both α rings clearly showed decreased density compared to a map of free 20S obtained from the x-ray structure of the yeast proteasome³ limited to 18 Å resolution (not shown). An internal control for our sample was provided by the contaminant particles representing the Blm10-20S complex in our data set. A low-resolution three-dimensional reconstruction was obtained from these particles (~3,500). In a sagital section, the Blm10-free end of the 20S proteasome showed the same level of density all across the α ring, suggesting that the axial channel of the yeast 20S proteasome is closed in solution when Blm10 is not bound (Figure 4(c), right panel, yellow arrow).

Orientation of the 20S proteasome in the Blm10-20S proteasome complex

We determined the orientation of the 20S proteasome within the map of the Blm10₂-20S complex to identify the seven α subunits and define how Blm10 binds to the end of the proteasome. All α subunits in the yeast 20S proteasome have a similar fold,³ and we therefore needed to determine first whether we could distinguish them at the calculated resolution of our reconstructed map (18 Å). Our ability to distinguish the different α subunits at 18 Å was confirmed by comparing the 7-fold symmetrized and the non-symmetrized atomic structure of the yeast 20S proteasome³ limited at 10 Å resolution. Fourier Shell Correlation (FSC) between these two maps started to drop at about 28 Å resolution and it reached 0.7 at 13 Å (Figure 5) suggesting that these two maps are significantly different in this spatial frequency range.

Figure 5. Spatial frequency range in which the seven α subunits of the 20S proteasome can be differentiated.

Comparison of the 20S proteasome density map with a D7 symmetrized version to establish a spatial frequency range of significant difference for distinguishing the 7 quasi-equivalent orientations. A 10 Å resolution 20S proteasome map was generated from the x-ray structure (PDB code:1RYP), and D7 (dihedral 7-fold) symmetrized. The two maps were compared using Fourier shell correlation. The two maps differed at spatial frequencies beyond 30 Å. The orientation of the 20S proteasome in the Blm10-20S complex shown in Figure 3 by cross-correlation was done at 18 Å.

To determine whether during the refinement the 20S proteasome model in the complex either remained in the same orientation or rotated around its longitudinal axis, we calculated the cross-correlation coefficient between the Blm10₂-20S complex density map and the x-ray crystal structure limited to 15 Å that was used as starting model. This calculation was performed for all the other six symmetry related orientations of the x-ray crystal structure and restricted to the 20S part of the density map in the complex. The highest cross-correlation coefficient was obtained for the orientation of the 20S in the starting model suggesting that the 20S proteasome in the complex did not rotate during the refinement process and that the α subunits can be assigned directly from the starting model (Figure 3). An identical result was obtained when the x-ray crystal structure of the 20S proteasome was limited to other resolutions such as 10 Å, 15 Å and 25 Å indicating the robustness of the result. This experiment shows that Blm10 binds to the α ring in a defined orientation and in the symmetric complex both caps are bound identically to the α ring. Importantly, this experimentally determined orientation places the proteasome’s 2-fold axis coincident with the 2-fold axis relating density of the two independent Blm10 molecules at either end of the proteasome complex.

BLM10 mutants are not sensitive to DNA damage conditions

Previous publications reported that BLM10 plays a role in DNA repair and is required to support growth at elevated temperatures.^17,19,26 To examine the role of Blm10 in yeast, we deleted the gene and examined the resulting cells.

Surprisingly, deletion of BLM10 in the A364a strain background had no effect on the rate of growth or viability when cells were challenged with DNA-damaging agents such as bleomycin or phleomycin (Figure 6(a)). The drugs were effective, as they severely restricted the growth of an isogenic strain with the dna2-2 mutation, which causes sensitivity to a range of DNA-damaging agents (Figure 6(a)).^27,28 Ploidy can affect the repair strategies used by yeast cells, as homologous repair is more readily available in diploids. We therefore tested sensitivity to these drugs in a homozygous deletion diploid strain, but this strain was not more sensitive to bleomycin or phleomycin than a WT diploid strain. Deletion of BLM10 did not cause sensitivity to other forms of DNA damage, including exposure to gamma-irradiation (Figure 6(b)), methyl methane sulfonate (MMS), UV irradiation, camptothecin, and hydroxyurea (not shown). These treatments caused inviability of sensitive control strains but the blm10-Δ mutant strain paralleled the growth rate and viability of the WT control in all cases (Figure 6(b) and not shown).

Figure 6. Deletion of blm10 does not cause sensitivity to DNA damage factors and high temperature.

(a) Cultures of haploid yeast strains from the A364a strain background 4053-5-2 (WT), 7883-6-3 (ctf4-Δ), 7757-3-3 (dna2-2), 8015-4-1 (blm10-Δ), or diploids 7860 (WT), and 8052 (blm10-Δ/blm10-Δ) (Table 2) were grown to saturation in rich medium. Aliquots of 10-fold serial dilutions were spotted to complete synthetic medium, complete medium with 12 μg/ml bleomycin (Sigma), or rich medium with 8 μg/ml phleomycin (Sigma), then incubated at 30°C. (b) Cultures were grown and processed as in panel (a), except strains 7860-6-3 (WT), 7373-1-4 (ctf4-Δ), 7757-6-3 (dna2-2), and 8015-4-3 (blm10-Δ) with the MATα mating type allele were used (Table 2). The plate in the right panel was exposed to about 100 kRad of gamma-irradiation from a Cs¹³⁷ source after cells were placed on the plate. (c) Cultures of the A364a strains 4053-5-2 (WT, haploid), 7860 (WT, diploid), 8015-4-1 (blm10-Δ haploid), and 8052 (blm10-Δ/blm10-Δ diploid), or the diploid BY4743 (Table 2) and its isogenic derivatives with deletions of YFL006w or YFL007w obtained from Research Genetics³⁰ were processed as in panel (a), then incubated on rich medium at 30°C, 37°C, or on rich medium containing 8 μg/ml phleomycin at 30°C. Incubation of all plates was terminated when the more phleomycin resistant BY4743 (Table 2) strains grew to full size, so the A364a strains appear to be more sensitive to phleomycin here than in panel (a).

Meiosis involves extensive recombination between homologous chromosomes, thus mutations that cause defects in DNA damage repair often cause defects in sporulation in yeast. However, diploids lacking both copies of BLM10 were able to form viable spores at rates qualitatively the same as WT, indicating normal ability to progress through the meiotic program in the absence of BLM10 (not shown). Therefore, we conclude that BLM10 is not needed for repair of various forms of DNA damage or to perform meiotic recombination.

Strains lacking BLM10 also grew at normal rates at temperatures ranging from 13°C to 38°C (Figure 6(c) and data not shown), indicating that Blm10 is not needed for survival at extreme temperatures.

We performed additional controls to determine whether the previously observed phenotypes are dependent on the genetic background and the mating type of the strain tested. The same deletion allele was introduced into W303 and S288C strains and similar results to those described above were obtained in these backgrounds (not shown). No defects in DNA repair were observed in MATa haploids, Matα haploids, or MATa/Matα diploids lacking BLM10 (Figure 6(a) and (b), and not shown). However, as previously reported, deletion of the gene encoding the DNA Polymerase α binding protein Ctf4 used as a control, caused sensitivity to bleomycin and gamma irradiation specifically in strains expressing the MATα mating type allele²⁹ (Figure 6(a) and (b)). Therefore, the conclusion that BLM10 is not needed to promote growth at extreme temperatures or for DNA damage repair seems to be independent of the genetic backgrounds and the mating type of the strain.

Finally, we attempted to reproduce the published results that seem to implicate BLM10 in DNA damage repair, allowing growth at high temperatures, or surviving exposure to low levels of canavanine. We obtained strains from the standard BY4743 deletion collection affecting BLM10,³⁰ a background isogenic with the strains described by Schmidt et al.¹⁸ Due to a sequencing error in the original genomic database release, BLM10 was initially thought to be two open reading frames, YFL007w and YFL006w.²⁶ Deletion of YFL007w removes the initiating methionine and about 80% of the BLM10 ORF, while deletion of YFL006w should allow expression of a C-terminal truncation protein that is missing the most highly conserved region of BLM10. In our experiments, the BY4743 background was significantly more resistant to phleomycin than A364a strains, but neither deletion within the BLM10 region affects growth at high temperatures or on media containing phleomycin (Figure 6(c)). Schmidt et al¹⁸ reported that a strain lacking both Blm10 and Ecm29 displayed increased sensitivity to the amino acid analog canavanine, but our results do not confirm this result (supplemental information). We did find that overexpression of Blm10 is detrimental to growth, and may have a slightly greater negative effect on media containing bleomycin, but these overexpression effects are likely to be indirect results of the sequestering of 20S proteasomes (supplemental information). These results support the conclusion that BLM10 is not needed for growth at high temperature or for various forms of DNA damage repair.

DISCUSSION

The sequence conservation between mammalian PA200 and yeast Blm10 suggested that they are functional homologs. This assumption has been obscured, however, by apparently conflicting functional data.^10,16,18 The 23 Å resolution reconstruction of bovine PA200 in complex with bovine 20S proteasome has been reported,¹⁴ and Blm10 also binds to the ends of mature proteasomes and stimulates their peptidase activity.¹⁸ Efforts to advance structural studies have been hindered by the low natural abundance and difficulty in separating Blm10 complexes from other activated proteasomes. We therefore developed an efficient purification method for the Blm10₂-20S complex. The high yield of complexes obtained facilitated the higher resolution reconstruction reported here, and further established the structural and biochemical relationship between yeast Blm10 and mammalian PA200.

The structural similarity is shown by comparing the previously described 3D-reconstruction of the bovine PA200-20S complex¹⁴ with the one from yeast Blm10-20S complex described here. This comparison reveals significant similarities between both models including the observation that both PA200 and Blm10 proteins bind to the ends of the proteasome as hollow asymmetric caps (Figure 7(a)) and that both proteins cause opening of the axial channel, further suggesting a conserved mechanism for activation of the 20S proteasome.

Figure 7. Comparison of the three-dimensional reconstruction of the yeast Blm10₂-20S and bovine PA200-20S proteasome complexes.

Two analogous views of the surface representation of yeast Blm10₂-20S proteasome (left) and bovine PA200-20S¹⁴ complexes (right). View perpendicular (a) to and along (b) the longitudinal axis.

Some interesting differences are also observed. For example, the Blm10 solenoid appears to be tightly packed and to contact all of the α subunits, whereas PA200 features an opening in its surface as it does not contact the α7 subunit in the 20S proteasome¹⁴ (Figure 7(b)). The absence of large openings in the surface of Blm10 raises the question of how peptides can access or exit the digestion chamber once the Blm10 cap is bound to the α ring. Because Blm10-20S proteasome complexes are labile and dynamic,¹⁸ one possibility is that Blm10 binds substrates before docking onto the 20S proteasome ends, whereupon Blm10 would open the gate of the axial channel and substrates diffuse into the digestion chamber. Alternatively, our reconstruction shows a narrow entrance pore in the Blm10 cap close to the longitudinal axis. This pore could expand under certain conditions and allow the traffic of peptides. Another possibility is that some portion of the Blm10 structure is sufficiently flexible to allow passage of peptides yet also sufficiently occupied to display significant density in our reconstructions. “Flexible curtains” of this nature appear to be present at the axial opening of archaeal 20S proteasomes² and in a central loop of the PA26 activator.⁶

The original blm3-1 mutation which was reported to be suppressed by BLM10 caused sensitivity to gamma-irradiation and to the oxidizing agent hydrogen peroxide.¹⁹ Moreover, a deletion of the N-terminal region of BLM10 was reported to cause sensitivity to phleomycin.^17,19,26 These results suggested that Blm10 functions in DNA repair. Subsequently, BLM10 deletion was found to make growth of yeast strains temperature sensitive.¹⁶ In apparent conflict with these reports, our genetic data indicate that BLM10 is not needed for repair of various forms of DNA damage or for survival at extreme temperatures. Furthermore, our results are consistent with published data from a number of high throughput screens that have been performed using the standard BY4743 deletion collection or isogenic relatives,³⁰ none of which have revealed phenotypes caused by the yfl006w or yfl007w deletions. Therefore, although conservation of sequence, structure, and proteasome activation argue strongly for a conserved function of Blm10 and PA200, the actual physiological role remains obscure. Definition of the functional roles and the extent to which they overlap for PA200 and Blm10 remain important questions for future studies.

MATERIALS AND METHODS

Construction of BLM10 plasmid to overexpress the protein in yeast cells

Due to the large size of BLM10, a gap repair strategy was used to construct a plasmid to overexpress the protein in yeast cells. The 681 bp BamHI-EcoRI fragment containing the GAL1-GAL10 promoter region from pMTL³¹ was inserted into YEplac195.³² This was digested with BamHI and HindIII and two overlapping oligos (PA-10 and PA-11; Table 1) were annealed and inserted to make pTF153. Primers PA-06 and PA-09 (Table 1) were used to amplify a 227 bp fragment containing the 5′ end of the BLM10 ORF by PCR, and primers PA-07 and PA-08 (Table 1) were used to amplify a 574 bp fragment containing the 3′ end of the ORF and extending 448 bp downstream of it. The two fragments were digested with KpnI and BamHI or BamHI and XhoI, respectively, then ligated with pTF153 digested with KpnI and XhoI to produce pTF155i. After digestion with BamHI, the linearized pTF155i with 223 bp of homology to the 5′ end of BLM10 and 549 bp of homology to the 3′ flanking region was used to transform the FY2/S288C strain BY4743 (Table 2).³⁰ Plasmids which had repaired the 6083 bp gap to reconstitute the BLM10 ORF under control of the GAL1 promoter and with 12 histidines fused to the N terminus were recovered. The initiating methionine of the 2143 residue BLM10 ORF is replaced with MH₁₂G… in this construct, making a protein of 2157 residues. Sequencing confirmed the accuracy of the construction and the gap repair.

Preparation of Blm10-20S proteasome complex

N-terminally 12X-histidine tagged Blm10 protein was recombinantly expressed in the yeast (S. cerevisiae) strain Bcy213 (Table 2). Cells were grown to an optical density (600 nm) of ~0.7 in synthetic medium containing 2% raffinose, 1.3 g/l amino acid mix^−URA, and 1% ammonium sulfate. Expression of Blm10 was induced by addition of 20% galactose to the culture to a final concentration of ~1.1%. Cells were harvested by centrifugation 12–18 hours after induction, washed with dH₂O, re-suspended in a volume of dH₂O equal to the pellet weight and flash frozen in liquid nitrogen. With the exception of weighing/thawing and cell disruption, all subsequent steps were performed at 4°C.

Typical preparations started with 80 g of frozen cells, which were thawed and added to an equal volume of 2X lysis buffer to a final concentration of 20 mM Tris pH 7.5, 1 M NaCl, 5 mMβ-mercaptoethanol, 1 mM EDTA, 0.1% NP40, 20 mM imidazole, and protease inhibitors (leupeptin 0.5 μg/ml, pepstatinA 0.7 μg/ml, aprotinin 0.5 μg/ml, and PMSF 0.25 mg/ml added during the lysis step). Cells were disrupted using a gaulin homogenizer, the lysate clarified by centrifugation, and batch-bound for 1 hour to a Ni-NTA slurry equilibrated with 20 mM Tris pH 7.5, 50 mM NaCl, 50 mM imidazole, and 0.01% NP40. The resin was packed into a column and washed with 50 ml of 20 mM Tris pH 7.5, 2 M NaCl, 50 mM imidazole, 0.01% NP40 followed by 50 ml of 20 mM Tris pH 7.5, 50 mM NaCl, 50 mM imidazole, 0.01% NP40. Blm10 was eluted with 100 ml of 20 mM Tris pH 7.5, 50 mM NaCl, 500 mM imidazole, 0.01% NP40. The eluted solution was collected into fresh β-mercaptoethanol to yield a finalβ-mercaptoethanol concentration of 5 mM.

Blm10-proteasome complex was prepared using yeast strain SDL135 in which the genomic copy of the Pre1 (β4) subunit gene encodes a domain of protein A at its C-terminus.³³ About 80 g of frozen cell pellets (expressing Pre1-proteinA 20S) were prepared and bound to IgG-sepharose resin as described.³⁴ The immobilized 20S proteasome was equilibrated with 50 mM Tris pH 7.5, 1 mM EDTA and all of the Blm10 previously purified from 80 g of cell pellet (typically 10–50 mg of Blm10) was incubated with 20S-IgG for 1hour. Blm10-20S complex (bound to IgG) was washed first with 500 ml of 50 mM Tris pH 7.5, 1 mM EDTA, and second with 500 ml of 50 mM Tris pH 7.5, 1 mM EDTA, 100 mM NaCl. The resin was collected into a glass column and washed with 20 ml of 50 mM Tris pH 7.5, 1 mM EDTA, 50 mM NaCl, 0.5 mM dithiothreitol. Blm10-20S complex was released from the IgG resin with TEV protease at a final concentration of 2–8 ug/ml 20 ml of 50 mM Tris pH 7.5, 1mM EDTA, 50 mM NaCl, and 0.5 mM dithiothreitol overnight. The purified Blm10-20S complex was collected by draining the column and washing with an additional 20 ml of TEV buffer. The protein was concentrated in a spin filtration conical vial and dialyzed against 20 mM Tris pH 7.5, 25 mM KCl, 10 mM NaCl, 1 mM MgCl₂, 0.1 mM EDTA, 1 mM dithiothreitol, and 0.5% glycerol for EM and SAXS analysis.

Cryo-electron microscopy

Freshly prepared holey carbon grids (400-mesh copper) were washed with acetone for 15 minutes and their surface was glow discharged in air for 30 seconds.³⁵ Grids were then floated on a five μl drop of polylysine hydrobromide (0.1% w/v aqueous solution; MW 60–120 kDa; Polysciencies, Inc) for two minutes. Excess polylysine solution was blotted away and a 3.5 μl drop of sample at a protein concentration of 100μg/ml was deposited on the grid. After blotting, the grid was plunged into liquid ethane and transferred to a Philips CM200 FEG electron microscope operated at 120 kV with a Gatan 626 cryo-holder. Focal pair images were acquired with low-dose techniques using nominal defocus values between 0.9 μm and 2.1 μm and 38,000X magnification. Kodak SO-163 film was used to record the images and they were digitized with a SCAI scanner (Z/I Imaging, Huntsville, AL) and binned twofold giving a sampling of 3.68 Å/pixel. Only images showing no drifting or astigmatism and with visible Thon rings in the power spectrum up to at least 12 Å⁻¹ were used for the image processing.

Three-dimensional reconstruction and visualization of maps

The program Boxer (EMAN)³⁶ was used to interactively pick 86,425 particles.

A rotationally averaged power spectrum was obtained from particles of each micrograph and used to determine the Contrast Transfer Function (CTF) and envelope function parameters of the micrograph with ctfit (EMAN).³⁶ In this process an x-ray scattering profile of the Blm10₂-20S complex was used (see supplementary materials). CTF-phase flipping was performed directly on the particle images and amplitude correction for the CTF was done during the generation of class-averages.³⁷

A total of 86,425 projection images representing side views (with the longitudinal axis parallel to the grid) were selected. Only side views were selected since these views enable us to visually distinguish whether Blm10 was bound to either one or both ends of the 20S proteasome. This set of views completely covers the Fourier space and provides all of the necessary information to build an accurate 3D model.

Orientations of the particles were determined by the projection matching approach provided by EMAN. Sets of projections were calculated from the initial maps using an angular spacing of 4.3o and particles were classified by comparison to these projections. After two cycles of refinement the resulting map was used to produce a Blm10-20S complex with a cap only on one end. These Blm10₂-20S and Blm10-20S models were used for a classification step in which projections from different view angles were calculated from both maps and each particle image was cross-correlated to projections from both maps. Particles assigned to the Blm10-20S map were discarded and the refinement continued with the particles assigned to the Blm10₂-20S map. This classification step allowed us to eliminate the particles in our data set representing Blm10-20S complexes that were erroneously picked. Convergence of the reconstruction was obtained upon four or five refinement cycles and symmetry was not imposed at any stage.

Non-model bias of the 3D-reconstruction was proven by using several initial maps including the symmetrized and non-symmetrized x-ray crystal structure of the yeast 20S proteasome (1RYP.pdb) and crystal structure of PA26 from Trypanosome Brucei in complex with yeast 20S proteasome (1FNT.pdb).⁸ Both structures were limited at 15 Å resolution.

The correct handedness of the model was imposed by the yeast 20S proteasome x-ray crystal structure³ used as starting model because it cannot be determined from our untilted images.

Orientation determination of the 20S proteasome in the symmetric Blm10-20S complex was performed with the Foldhunter program.³⁸

Resolution of the final map was estimated by two methods. First, the data set was split in two halves and a 3-D reconstruction was calculated from both halves independently using the same starting model and following the same refinement approach. Then Fourier Shell Correlation (FSC) was calculated between both maps applying the 0.5 FSC threshold. Second, we performed the even-odd test in EMAN^37,39 in which two maps were obtained following the last cycle of refinement from the even and odd numbered particles and compared by FSC.

The program Chimera^40,41 was used to generate surface rendering representations of the model and the rendering threshold selected included 100% of the mass of the complex. The Bshow program from the Bsoft package⁴² was used to visualize the density maps.

Deletion of BLM10

BLM10 was deleted by amplifying fragments from the 5′ and 3′ flanks of the gene using PCR with primers PA-01 and PA-02 (digested with KpnI and EcoRI) and PA-03 and PA-04 (digested with KpnI and BamHI), respectively (Table 1). These were ligated with YIplac204³² digested with EcoRI and BamHI, and the product (pTF156) was digested with KpnI. Transformation of yeast strains with linearized pTF156 leads to replacement of the BLM10 ORF with YIplac204 (TRP1), leaving only the first 4 and the last 14 bases of the 6429 bp ORF intact. Accuracy of the PCR products was confirmed by sequencing, and accuracy of integration was confirmed by Southern blotting.