Spatial definition of the human progesterone receptor-B transcriptional complex.

The transcriptional activity of recombinant human PR-B protein, which was purified from baculovirus, was tested using an in vitro transcription assay using a chromatinized growth regulating estrogen receptor binding-1 (GREB1) composite DNA template ( Panigrahi et al., 2018 ). The GREB1 composite fragment was constructed using GREB1 enhancer and promoter regions that contain progesterone response elements (PREs). Addition of increasing concentrations of purified recombinant PR-B protein in the presence of progesterone gradually increased the levels of in vitro -transcribed GREB1 mRNA ( Figure S1 A), confirming that the purified PR-B protein is functionally active. For single-particle cryo-EM experiments, recombinant PR-B protein was incubated with a short PRE oligonucleotide (32 bp) in the presence of progesterone. The liganded DNA-bound PR-B sample was directly applied to cryo-EM grids and subjected to standard single-particle cryo-EM data collection and analyses (see STAR Methods section). Representative raw images of DNA-bound PR-B are shown in Figure S1 B. We resolved a structure at ∼11 Å resolution ( Figures 1 A and S1 C) based on the gold standard Fourier shell correlation (FSC) ( Henderson et al., 2012 ). In the presence of progesterone, PR-B forms a homodimeric structure that binds to DNA. Similar to our previous DNA-bound AR dimer structure ( Yu et al., 2020 ), the DNA-bound PR-B dimer map contains a DNA-like density that defines the location of PRE-DNA bound by the two DBDs ( Figure 1 B). The reconstructed structure presents a pseudo 2-fold symmetry reflecting formation of a dimer. Following a similar segmentation procedure reported for complexes of DNA-bound liganded ERα and AR ( Yi et al., 2015 ; Yu et al., 2020 ), we performed segmentation on the reconstructed PR-B map and assigned densities using a combination of relative molecular mass, known binding domains, and antibody-binding results ( Figures 1 B and S2 ). The NTD density map was further segmented into subregions (AF1, AF3, and the rest of the NTD located between the AF1 and AF3 labeled as rNTD). To confirm the assignment of AF3, we incubated the PRE/PR-B sample with a PR-B-specific antibody fragment (N-Fab) which recognizes the 164 aa N-terminal extremity that is present in PR-B but not in PR-A. The reconstructed Fab-bound PRE/PR-B structure is similar to the PRE/PR-B structure with the exception of two protruding densities indicating the binding of an N-Fab to each PR-B monomer ( Figures S2 A and S2B). The two N-Fab densities are restricted to the AF3 region in the segmented density map, thereby confirming the AF3 annotation. We assigned the region connecting AF1 and AF3 as rNTD-a and rNTD-b for each PR-B-monomer. Furthermore, we labeled the LBD using another Fab fragment (C-Fab) generated from an antibody that selectively recognizes the C-terminal end of PR-B (919-933aa), next to the AF2 region (904-919aa) ( Danielian et al., 1992 ). As shown in Figures S2 C and S2D, two C-Fabs bind to the region assigned as the LBD in the segmented map and are positioned at the interaction interface between the LBD and the AF3. This result is consistent with an earlier finding that the 30-aa C-terminal end of the PR-B isoform is essential for contact between the LBD and the AF3 ( Dong et al., 2004 ). Collectively, the above antibody-labeling results confirm our PR-B dimer domain segmentation analyses. Figure 1 Density map and segmentation of DNA-bound liganded PR-B dimer (A) Cryo-EM map of PRE-DNA/PR-B at a resolution ∼10.9 Å viewed from different orientations. (B) Segmentation of PRE-DNA/PR-B. The segments of two PR-Bs are shown as -a and -b separately. The functional domains of PR-B are annotated in different colors. The NTD (dash circled) contains AF3 (Brown and light Brown), AF1 (Purple and light Purple), and the rNTD (the NTD region located between the AF1 and AF3 colored in Green and light Green). LBD: ligand binding domain (Orange and Red); DBD: DNA binding domain (Blue and light Blue). A DNA-like feature is colored in White. See also Figures S1 and S2 and Table S1 . The scale bars represent 100 angstrom. Density map and segmentation of DNA-bound liganded PR-B dimer (A) Cryo-EM map of PRE-DNA/PR-B at a resolution ∼10.9 Å viewed from different orientations. (B) Segmentation of PRE-DNA/PR-B. The segments of two PR-Bs are shown as -a and -b separately. The functional domains of PR-B are annotated in different colors. The NTD (dash circled) contains AF3 (Brown and light Brown), AF1 (Purple and light Purple), and the rNTD (the NTD region located between the AF1 and AF3 colored in Green and light Green). LBD: ligand binding domain (Orange and Red); DBD: DNA binding domain (Blue and light Blue). A DNA-like feature is colored in White. See also Figures S1 and S2 and Table S1 . The scale bars represent 100 angstrom. The DNA-bound PR-B dimer structure reveals that the two LBD densities are in the middle of the dimerization interface with the two NTDs located on each side of the LBDs. However, the two NTDs do not come into strong contact with each other. This NTD-LBD organization is strikingly different from our previously reported AR dimer structure ( Yu et al., 2020 ) ( Figure 2 ) in which the AR NTDs wrap around the LBDs before contacting each other, essentially encasing the LBDs. On the other hand, the ERα dimer density from the ERα/SRC-3/p300 complex map is comparable to the PR-B dimer density except that ERα contains a significantly shorter NTD ( Figure 2 ). Although the structure of the PR-B LBD dimer is similar to the AR LBD dimer, the PR-B LBD adopts a different orientation toward the DNA and DBD ( Figure S4 ). Comparison of these three steroid receptor dimer structures demonstrates that they do not adopt the same dimerization mechanisms despite having similar domain organization. Figure 2 Structural comparison of the PR-B, AR, and ERα dimers on DNA PR-B, AR, and ERα densities in two different views show different arrangements of the LBD and NTD. PR-B and AR are segmented DNA bound dimer density while ERα is extracted from ERE-DNA/ERα/p300/SRC-3 complex. The cartoon illustrations of the three receptors are shown at the right. All three densities show LBD in the center. The NTDs of PR-B and ERα are packed at two sides of the LBDs while the LBDs of AR are encircled by its large NTDs. See also Figure S4 . Structural comparison of the PR-B, AR, and ERα dimers on DNA PR-B, AR, and ERα densities in two different views show different arrangements of the LBD and NTD. PR-B and AR are segmented DNA bound dimer density while ERα is extracted from ERE-DNA/ERα/p300/SRC-3 complex. The cartoon illustrations of the three receptors are shown at the right. All three densities show LBD in the center. The NTDs of PR-B and ERα are packed at two sides of the LBDs while the LBDs of AR are encircled by its large NTDs. See also Figure S4 . We previously demonstrated that AR recruits primary and secondary coactivators through its NTD while ERα recruits two SRC-3 through its LBD ( Yi et al., 2021 ). Here, we examined how PR-B recruits core coactivators. Because of its importance in a number of progesterone-dependent physiological processes ( Kommagani et al., 2013 , 2014 ; Mukherjee et al., 2006 ), SRC-2 was chosen as the primary coactivator in these structural studies. In our in vitro transcription assay, we found that immunodepleting SRC-2 and p300 abolished PR-B-mediated GREB1 transcription ( Figure 3 A), demonstrating that SRC-2 and p300 are essential primary and secondary coactivators, respectively, for affecting PR-B-mediated transactivation. Adding back both purified recombinant SRC-2 and p300 (but not each individually) largely restored PR-B-mediated transcriptional activity ( Figure 3 A). This result confirms that our purified recombinant SRC-2 and p300 are functionally active. The assembly of the DNA-bound PR-B/SRC-2/p300 complex was accomplished by incubating the three purified recombinant proteins with progesterone and biotinylated PRE-DNA, followed by DNA pull-down with streptavidin beads and the enzymatic release of the complexes from the beads ( Yi et al., 2015 , 2017 ; Yu et al., 2020 ). Following a standard single-particle cryo-EM pipeline, we collected raw images of the complex as shown in Figure S1 D. Using RELION ( Punjani et al., 2017 ; Scheres, 2012 ), we resolved a density map of the PRE-DNA-bound PR-B/SRC-2/p300 complex ( Figure 3 B). The complex density map has a dimension of ∼160 x 220 x 240 Å with a resolution of ∼19 Å based on FSC ( Figure S1 E) ( Henderson et al., 2012 ). The density map was then segmented, and each component was identified ( Figure 3 B). Our results indicate that the PR-B dimer recruits only one SRC-2 and one p300, which resembles the AR-coactivator complex but not the ERα-coactivator complex ( Yi et al., 2015 ; Yu et al., 2020 ) ( Figure S4 ). However, the overall shape of the complexed SRC-2 resembles the SRC-3 densities in the AR and ERα transcriptional complexes. The p300 density is also similar to the p300 density in the ERα-coactivator complex or the free antibody-labeled p300 structure ( Yi et al., 2015 ). Alignment of these p300 structures indicates that the SRC-interaction domain (SRCID) at the C-terminal end of p300 is positioned at the p300-SRC-2 interaction interface ( Figure S5 ), confirming previous reports that the SRCID is necessary for p300-SRC interaction ( Kamei et al., 1996 ). Next, we docked the segmented PRE-DNA/PR-B structure into the PR-B density within the complex ( Figures 3 D and S6 ) to understand how each domain of PR-B is engaged in coactivator binding. To ensure correct docking, we labeled the DNA-bound PR-B/SRC-2/p300 complex with a PR-B N-Fab antibody ( Figures S3 A and S3B). Aligning the position of the N-Fab antibody in the complex map with its position in the segmented PR-B dimer map further guided the replacement. As shown in Figure 3 C, p300 has multiple contacts with the NTD and LBD of PR-B. The AF3 is the primary region of the NTD that interacts with p300. We found that the C-Fab-labeled region ( Figure S2 D), which is adjacent to the AF2-a, is located at the p300 docking site, suggesting that the AF2 also contributes to the interaction with p300. Together, the two AF3s and AF2-a form a docking site for p300 binding. A part of AF1 is also involved in interacting with p300. SRC-2 was shown to mainly interact with the AF1 region of the NTD with a modest interaction with the LBD-a. A close contact between SRC-2 and p300 within the complex was detected, which comprised a long interaction interface. Altogether, through multiple contacts between each other, the three proteins form a stable DNA-bound complex. Figure 3 The structure of PRE-DNA-bound PR-B/SRC-2/p300 complex (A) Purified recombinant PR-B, SRC-2, and p300 proteins activate PR-targeted gene transcription in vitro . Chromatinized PR-targeted gene GREB1 composite template with endogenous enhancer and promoter elements (CompF template) was transcribed in vitro with HeLa nuclear extract (NE) after control immunodepletion (ΔCtr) and immunodepletion of both SRC-2 and p300 (ΔSRC-2/p300). Reactions were then supplemented with recombinant purified PR-B, SRC-2, and p300 as indicated. HeLa NE lacking SRC-2 and p300 fails to support PR-B-dependent transcription activation, which is rescued by addition of recombinant SRC-2 and p300 demonstrating that our purified PR-B, SRC-2, and p300 are functionally active. The error bars represent standard error of the mean. (B) Cryo-EM density map of the PRE-DNA-bound PR-B/SRC-2/p300 complex at 19.1 Å resolution. Shown are 4 different angles of the map rotating every 90 degrees. (C) Segmentation of PRE-DNA/PR-B/SRC-2/p300. Each component was segmented to annotate different proteins: PR-B, Green; SRC-2, Orange; p300, Blue, respectively. (D) The assembled PRE-DNA/PR-B/SRC-2/p300 structure with segmented PRE-DNA/PR-B density ( Figure 1 B) replacing the PR-B density. The green outline represents the replaced density from the PRE-DNA/PR-B complex. See also Figures S1 , S3 , S5 , and S6 and Table S1 . The scale bars represent 100 angstrom. The structure of PRE-DNA-bound PR-B/SRC-2/p300 complex (A) Purified recombinant PR-B, SRC-2, and p300 proteins activate PR-targeted gene transcription in vitro . Chromatinized PR-targeted gene GREB1 composite template with endogenous enhancer and promoter elements (CompF template) was transcribed in vitro with HeLa nuclear extract (NE) after control immunodepletion (ΔCtr) and immunodepletion of both SRC-2 and p300 (ΔSRC-2/p300). Reactions were then supplemented with recombinant purified PR-B, SRC-2, and p300 as indicated. HeLa NE lacking SRC-2 and p300 fails to support PR-B-dependent transcription activation, which is rescued by addition of recombinant SRC-2 and p300 demonstrating that our purified PR-B, SRC-2, and p300 are functionally active. The error bars represent standard error of the mean. (B) Cryo-EM density map of the PRE-DNA-bound PR-B/SRC-2/p300 complex at 19.1 Å resolution. Shown are 4 different angles of the map rotating every 90 degrees. (C) Segmentation of PRE-DNA/PR-B/SRC-2/p300. Each component was segmented to annotate different proteins: PR-B, Green; SRC-2, Orange; p300, Blue, respectively. (D) The assembled PRE-DNA/PR-B/SRC-2/p300 structure with segmented PRE-DNA/PR-B density ( Figure 1 B) replacing the PR-B density. The green outline represents the replaced density from the PRE-DNA/PR-B complex. See also Figures S1 , S3 , S5 , and S6 and Table S1 . The scale bars represent 100 angstrom. We next examined the contributions of each PR-B functional domain in recruiting SRC-2 and p300 to confirm the cryo-EM structure. We generated three GST-fused PR-B fragments, AF1 (401–546 aa), AF3 (1–170 aa), and LBD (688–933 aa, which contains the AF2). These fragments were expressed in E . coli ., purified, and then incubated with purified recombinant full-length SRC-2 or p300 protein. Progesterone was added to the PR-B LBD fragment and SRC-2 or p300 mixture to facilitate ligand-dependent interaction. A GST pull-down experiment was then performed to examine PR-B fragments associated with SRC-2 or p300. As shown in Figure 4 A, SRC-2 interacts strongly with the AF1 of PR-B. This is consistent with the cryo-EM structure of the PR-B/SRC-2/p300 complex where SRC-2 was observed to contact the AF1 region ( Figure 3 D). The p300 protein interacted with all three fragments of PR-B ( Figure 4 A), confirming our structural conclusions that p300 contacts both the NTD and LBD of PR-B within the core transcriptional complex ( Figure 3 D). Figure 4 Domain interactions between PR-B, SRC-2, and p300 (A) The interaction between PR-B functional domains and SRC-2, p300. Upper and Middle panels are the Western blot results of GST-fused PR-B fragments-associated p300 or SRC-2 protein, respectively. The bottom panel is the Coomassie staining of PR-B fragments pulled-down using the glutathione beads. (B and C) The SRC-2 domains involved in interacting with PR-B (B) or p300 (C). GST-fused SRC-2 functional domains were incubated with purified PR-B or p300 protein followed by GST pull-down experiments. Shown are Western blot results of associated PR-B or p300. The positions of fragment bands on the coomassie staining are labeled by stars. See also Figures S1–S3 and S5 . Domain interactions between PR-B, SRC-2, and p300 (A) The interaction between PR-B functional domains and SRC-2, p300. Upper and Middle panels are the Western blot results of GST-fused PR-B fragments-associated p300 or SRC-2 protein, respectively. The bottom panel is the Coomassie staining of PR-B fragments pulled-down using the glutathione beads. (B and C) The SRC-2 domains involved in interacting with PR-B (B) or p300 (C). GST-fused SRC-2 functional domains were incubated with purified PR-B or p300 protein followed by GST pull-down experiments. Shown are Western blot results of associated PR-B or p300. The positions of fragment bands on the coomassie staining are labeled by stars. See also Figures S1–S3 and S5 . It is known that the LXXLL motifs present in the SRC receptor interaction domain (RID) interact with NR LBDs in a ligand-dependent manner. However, we observed that SRC-2 interacts with AF1, suggesting that a region other than the RID of SRC-2 may be involved in the interaction with the PR-B NTD. To test this, we generated five GST-fused SRC-2 fragments, corresponding to the bHLH/PAS (basic helix-loop-helix/Per-Arnt-Sim), S/T (Ser/Thr rich region), RID, CID (CBP/p300 interaction domain), and HAT (Histone acetyltransferase) domains ( Figure 4 B). A GST pull-down experiment was performed to examine the interaction between these fragments and purified PR-B. We found that the HAT domain interacts with PR-B in addition to the known RID interaction ( Figure 4 B), suggesting that the HAT domain is involved in the AF1 interaction. There is a substantial contact surface between SRC-2 and p300 in addition to the p300-PR-B and SRC-2-PR-B interactions. Through the GST pull-down experiments, we demonstrate that several domains of SRC-2, including S/T and HAT, interact with p300 in addition to the known CID ( Figure 4 C). This result suggests that the two coactivators recruited by PR-B have strong connections with one another via multiple domain interactions, which help to enhance and sustain the complex integrity. In summary, our cryo-EM and biochemical studies demonstrate the functional role of AF1 in recruiting SRC-2 and cooperative action of AF3 and AF2 in recruiting p300. Each of the protein components has tight contacts with the other two proteins within the complex to form a core functional active transcription unit ( Figure 5 ). We also found that the three steroid receptors (PR-B, ERα, and AR) have different complex assembly mechanisms as summarized in Table 1 . Figure 5 Model of full-length PR-B domain organization and the contribution of each AF in SRC-2 and p300 coactivator recruitment Shown is the illustration of the assembled PRE-DNA/PR/SRC-2/p300 structure demonstrating the strong interaction between p300 (transparent) and two AF3s & LBD-a (AF2-a) (top panel), and the strong interaction between SRC-2 (transparent) and AF1-b (bottom panel). See also Figure S5 . The scale bars represent 100 angstrom. Table 1 Comparison of three steroid receptor dimer structures and the complex assembly mechanisms with coactivators ERα, AR, PR-B complexes comparison ERα AR PR-B Dimer 2-fold symmetric structure Yes No Yes LBDs at the center of dimer interface Yes Yes Yes NTD-LBD spatial organization Each NTD is at one side of LBD The NTD wraps around the LBD Each NTD is at one side of LBD Coactivator binding Two SRC-3 One p300 One SRC-3 One p300 One SRC-2 One p300 Receptor-SRC contact site Mainly LBD (AF2) N-terminal end close to the FXXLF motif AF1 close to the DBD Receptor-p300 direct interaction No Yes (AF1) Yes (AF3, AF2 and AF1) See also Figures S2–S5 and Table S1 . Model of full-length PR-B domain organization and the contribution of each AF in SRC-2 and p300 coactivator recruitment Shown is the illustration of the assembled PRE-DNA/PR/SRC-2/p300 structure demonstrating the strong interaction between p300 (transparent) and two AF3s & LBD-a (AF2-a) (top panel), and the strong interaction between SRC-2 (transparent) and AF1-b (bottom panel). See also Figure S5 . The scale bars represent 100 angstrom. Comparison of three steroid receptor dimer structures and the complex assembly mechanisms with coactivators See also Figures S2–S5 and Table S1 .

The AF3 domain not only distinguishes PR-B from other members of the NR superfamily but also has long been assumed to underpin the distinct transactivational properties that differentiate the PR-A and PR-B isoforms from each other ( Sartorius et al., 1994 ; Tung et al., 2001 , 2006 ). Using isolated functional domain fragments, biochemical studies have implicated AF3 as a major contributor to PR-B’s tertiary structure; as an intramolecular interface for other regions of PR-B, including AF1, AF2, and the DBD; and as a key interacting surface for accessory proteins, such as nuclear coactivators. In the case of the AF2 containing LBD, early investigations demonstrated that the more conserved C-terminal region of NRs undergoes a ligand-dependent conformational change that results in formation of an AF2 hydrophobic pocket that docks with a RID containing LXXLL motif within primary SRCs ( Hill et al., 2012 ). Mapped to an intrinsic disordered region just upstream of the DBD within the PR-B NTD, the ligand-independent AF1 has been implicated in the molecular recognition and assembly of primary coactivators ( Goswami et al., 2014 ; Kumar et al., 2013 ; Onate et al., 1998 ; Wardell et al., 2005 ; Woo et al., 2019 ). While multiple biochemical investigations have uncovered important aspects of their transactivational roles, a structural framework for providing a mechanistic understanding of how these functions are integrated within the context of full-length PR-B assembled with full-length coactivators as an active and intact core transcriptional complex has been missing. Here, we describe the cryo-EM structures of the DNA-bound full-length PR-B homodimer as well as its higher order assembly with full-length SRC-2 and p300. Our structural model indicates a 2-fold symmetrical structure in which the two PR-B LBDs (LBD-a and LBD-b) mediate a strong interaction between the PR-B monomers within the dimer’s center. This finding agrees with previously published crystal structures that underscored the importance of both LBDs in PR-B dimerization ( Williams and Sigler, 1998 ). Comprising nearly half of the PR-B protein, the intrinsically disordered NTD of each monomer closely associates not only with their corresponding LBDs but also with the LBD of their monomer partner. The structural consequences of this NTD-LBD association are close intermolecular positioning of the NTD-AF3a and NTD-AF3b with LBD-AF2b and LBD-AF2a, respectively. This spatial arrangement supports intermolecular “functional synergy” between AF3 and AF2 within the PR-B dimeric complex in addition to physically contributing to a tighter dimer. The exposed location of both AF3 regions along with interposed LBD regions demonstrates that a prominent interface surface exists for coactivator recruitment. The structure of the liganded PR-B dimeric complex bound to its target DNA element reveals both topological similarities and differences with the corresponding dimeric complexes for ERα and AR ( Yi et al., 2015 ; Yu et al., 2020 ). While a similar spatial orientation is shared by the PR-B and ERα homodimers—LBDs in the center of the complex with flanking NTDs on the outside—the significantly longer PR-B NTD with its additional AF3 extremity leads to a different coactivator assembly mechanism for PR-B compared to ERα when forming their respective core transcriptional complexes. While PR-B and ERα adopt a tail-to-tail 2-fold symmetry on DNA, the AR monomers form a unique head-to-head and tail-to-tail arrangement ( Yu et al., 2020 ), with their long NTDs encircling their LBDs. The AR NTD encirclement not only masks the AF2 containing LBD but also results in direct contact of the AR NTD at their extremities ( Yu et al., 2020 ). In the case of the PR-B homodimer structure, the monomer NTDs neither mask AF2 nor make direct contact with each other. The net result of this spatial configuration is that the PR-B can avail three AFs for coactivator recruitment whereas AR depends heavily on one. Therefore, the PR-B complex structure reinforces our original finding that dimeric NRs co-opt different topologies and spatial arrangements on DNA, despite their similar functional domain organizations ( Yi et al., 2021 ). Such structural differences predict that subsequent coactivator assembly mechanisms also will be markedly different, leading to core transcriptional complexes with significantly distinct quaternary structures. The core PR-B transcriptional complex comprises the PR-B dimer assembled with one SRC-2 and one p300. The stoichiometry of the PR-B core transcriptional complex significantly differs from the corresponding ERα complex, which instead recruits two primary coactivators and one p300 ( Yi et al., 2015 ). As predicted from our PR-B dimer structure above, both AF3s directly contact p300 in the PR-B core transcriptional complex. However, only the PR-B LBD-a directly contacts p300, suggesting that LBD-b is free to interact with other coactivators as the core transcriptional complex further nucleates prior to initiation of transcription ( O'Malley, 2003 ). Based on C-Fab recognition of the region adjacent to AF2, we found that PR-B AF2 and AF3 are involved in binding p300 ( Figure S2 D). Mutation of the key residue in this region (E911A) inhibits PR transcriptional activity ( Gong et al., 1997 ; Wen et al., 1994 ). Unlike PR-B and AR ( Yu et al., 2020 ), the ERα dimer does not directly contact p300 but instead uses two SRCs to indirectly enlist p300, which indicates that absence of a large multi-domain NTD in ERα necessitates the recruitment of an additional primary SRC to bridge the gap between the NR and a secondary coactivator. The AF2 of ERα is required for interacting with SRCs ( Feng et al., 1998 ; Shiau et al., 1998 ; Yi et al., 2015 ) instead of p300 as observed in the PR-B complex. Although the AR directly contacts p300 within its core transcriptional complex, direct contact is made exclusively through its N-terminal AF1 ( Yu et al., 2020 ). Within the PR-B core transcriptional complex, a single SRC-2 molecule is wedged between the PR-B dimers and p300. The AF1-b in the PR-B’s NTD-b region forms a key interface between PR-B and SRC-2. While this observation agrees with our biochemical data here and previous studies elsewhere ( Goswami et al., 2014 ; Kumar et al., 2013 ; Onate et al., 1998 ; Woo et al., 2019 ) that support a pivotal role for AF1 in primary coactivator recruitment, these findings underscore a “division of labor” between the PR-B monomers when it comes to both primary and secondary coactivator recruitment during formation of the core transcriptional complex. The quaternary structures for the PR-B dimeric and core transcriptional complexes on DNA now provide the pretext for similar structural studies on the PR-A isoform in the future. Structures of each isoform complex will be required to delineate the structural similarities and differences to explain their common and divergent transactivational responses to progesterone ligand shown in vitro and in vivo ( Jacobsen and Horwitz, 2012 ). Our structural analysis underscores a pivotal role for the PR-B-specific AF3 region as a direct contact with the secondary p300 coactivator, suggesting that the absence of AF3 in the PR-A isoform will block direct secondary coactivator contact. We also predict that without the AF3, the tertiary conformation of PR-A will markedly differ from PR-B, resulting in isoform-specific differences in intramolecular interdomain allosteric crosstalk within their respective transcriptional complexes. Apart from elucidating the fundamental structural differences of the core transcriptional complexes that underpin the distinct transactivational properties of each member of the NR triad that is responsible for mediating sex steroid hormone action in reproduction, the PR-B structures—their protein-protein interaction regions in particular—may offer a paradigm shift in rational drug design that takes into account the intact receptor complexed with full-length coactivators rather than just focusing on one domain. Following a holistic-based structure-based design framework would significantly expand the options for progestin therapy currently available to treat a spectrum of gynecological disorders, from uterine fibroids to endometriosis ( Critchley and Chodankar, 2020 ; Vannuccini et al., 2021 ). In this study, we identified the contributions of each AF and demonstrate their structural basis in forming the PR-B dimer interface and PR-B/coactivator complex. Based on the structural information, we proposed a mechanism of how each of the protein components contact with the other two proteins within the complex to form a core functional active transcription unit. Nevertheless, there are limitations of structural resolution associated with the endogenous flexibility of each component. The PR-B structure was roughly segmented into regions based on Segger in Chimera with watershed algorithm and validated by antibodies recognition. The atomic-level structural details like side chain orientations or exact boundaries of regions are missing. Thus, the deficiency of atomic structure details for interface residues leads to limitation for further discussion of interactions between PR-B and its coactivators. Besides, the structures of another functional isoform PR-A without the AF3 remain unknown. PR-A will obviously differ from PR-B, resulting in isoform-specific differences in intramolecular interdomain allosteric crosstalk within their respective transcriptional complexes. However, all the differences of PR-A and PR-B were indirectly analyzed based on known structures for the PR-B. Further structural investigation of PR-A is needed.

As the apex hormone of pregnancy, progesterone is indispensable for female fertility and overall reproductive health ( DeMayo and Lydon, 2020 ). Along with infertility and early pregnancy loss, dysregulation of the progesterone response promotes myriad pathologies, from endometriosis and leiomyoma to endometrial and mammary gland tumorigenesis ( Diep et al., 2015 ; Horwitz and Sartorius, 2020 ; Ishikawa et al., 2010 ; Kim et al., 2020 ; Patel et al., 2015 ). The majority of these cellular and tissue responses to progesterone are mediated by the progesterone receptor (PR). The PR is a key member of the nuclear receptor (NR) superfamily of transcription factors ( O'Malley, 2020 ; Tsai and O'Malley, 1994 ), which also includes the estrogen and androgen receptors (ERα and AR, respectively). With ERα and AR, PR completes the triad of NRs for sex steroid hormones that control female and male reproductive biology. The PR is a multidomain allosteric transcription factor that shares with NR family members a common functional domain organization, which includes an N-terminal domain (NTD), a centrally located DNA-binding domain (DBD) with a hinge region, followed by a ligand-binding domain (LBD) at the C-terminus ( Grimm et al., 2016 ). Like other NR superfamily members, PR contains two activation function (AF) domains: AF1 and AF2 ( Grimm et al., 2016 ). Located within the NTD, the ligand-independent AF1 domain is intrinsically disordered, modular, and lacks evolutionary conservation with other NR members. Mapped to the LBD, the ligand-dependent AF2 is significantly more conserved to other NR members and is highly structured. Early biochemical studies on PR indicated that additional steroid receptor coactivator (SRC) interaction surfaces reside within the PR NTD, of which only a portion maps to the AF1 ( Onate et al., 1998 ). Although lacking structural support at the time, these findings suggested that optimum NR transcriptional activity requires functional synergy between AF domains ( Tetel et al., 1999 ). Unique to the NR superfamily, the PR exists naturally as two functionally distinct receptor isoforms (PR-A and PR-B), which are identical to each other except that the PR-A isoform lacks the first 164 amino acids (aa) present in the N-terminus of PR-B ( Dong et al., 2004 ; Sartorius et al., 1994 ; Tung et al., 2001 , 2006 ). While both PR isoforms possess AF1 and AF2 domains, the PR-B isoform contains a context-dependent AF3 domain, which is located within the N-terminal 164 aa sequence. Early cell-based reporter assays suggested that the AF3 domain contributes in part to promoter-specific transactivation differences between the two isoforms ( Takimoto et al., 2003 ); however, it remains unclear how the addition of AF3 regulates PR transcriptional activity. DNA-bound NRs—through a sequential and processive recruitment process—interact with members of the SRC/p160 family of primary coactivators, which in turn recruit secondary coactivators (i.e. p300/CREB-binding protein [CBP]) to form their respective core coregulator transcriptional complexes ( O’Malley, 2003 ). Our previous cryo-EM structural studies demonstrated that ERα and AR utilize different functional domains to recruit coactivators, and their highly variable NTDs support formation of the dimerization interface and are key determinants of receptor-coactivator interaction mechanisms ( Yi et al., 2021 ; Yu et al., 2020 ). Compared to ERα, AR has a longer NTD that surrounds the LBD to mask the LBD-coactivator interaction surface, resulting in a stronger AF1 function. PR-B has an NTD that is comparable with AR in length. In contrast to AR, the AF2 of PR is a potent transcriptional activator ( Meyer et al., 1990 ) but the underlying reason for its differential activation function is not well understood. Structural insights that define the stoichiometry and topological arrangement of the individual components of DNA-bound NR core coactivator transcriptional complexes have been valuable for gaining fundamental insights into their control of gene expression ( Yi et al., 2021 ). Here, we present structures of a DNA-bound PR-B dimer and a PR-B/SRC-2/p300 core complex by single-particle cryo-EM, further validated through functional interaction experiments. Our studies demonstrate that PR-B adopts a distinct dimerization and coactivator assembly architecture compared to ERα and AR. Thus, structural analyses reveal that ERα, AR, and PR-B utilize distinct conformations to recruit coactivators and activate transcription despite having a similar, modular arrangement of their AF1, DBD, and LBD/AF2 domains.

REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies PRB-N Ab Santa Cruz Cat#Sc-811; RRID: AB_628173 PRB-C-Ab Santa Cruz Cat# sc-53943; RRID: AB_831674 HA Ab Santa Cruz Cat# sc-805, RRID: AB_631618 Flag M2 beads Sigma Cat#F2426; RRID: AB_2616449 P300 Santa Cruz Cat#Sc-584; RRID: AB_2293429 Flag-HRP Sigma Cat#A8592; RRID: AB_439702 Bacterial and virus strains E . coli BL21 (DE3) Agilent Cat#200131 Chemicals, peptides, and recombinant proteins SRC-2 Monoclonal Antibody/Recombinant Protein Expression Core at BCM N/A P300 Monoclonal Antibody/Recombinant Protein Expression Core at BCM N/A PR-B Monoclonal Antibody/Recombinant Protein Expression Core at BCM N/A Dynabeads™ M-280 Streptavidin Thermo Fisher Cat#11205D Progesterone Sigma Cat#P0130 Glutathione Sepharose 4B GE Life Sciences Cat#17075601 Pierce Fab MicroPreparation Kit Thermo Fisher Cat#44685 Deposited data CryoEM maps of PRE-DNA/PR-B This study EMDB: EMD-27537 CryoEM maps of PRE-DNA/PR-B/SRC-2/p300 This study EMDB: EMD-27540 CryoEM maps of PRE-DNA/PR-B/N-Fab This study EMDB: EMD-27537 CryoEM maps of PRE-DNA/PR-B/C-Fab This study EMDB: EMD-27537 CryoEM maps of PRE-DNA/PR-B/SRC-2/p300/N-Fab This study EMDB: EMD-27540 Experimental models: Cell lines HEK 293T/17 Tissue Culture Core (BCM) Cat#ATCC® CRL-11268™ Sf9 Monoclonal Antibody/recombinant Protein Expression Core (BCM) Cat# ATCC® CRL-1711™ Oligonucleotides PRE PCR forward primer Thermo Fisher N/A PRE PCR reverse primer Thermo Fisher N/A ARE/PRE oligonucleotide Santa Cruz Cat#sc-2551 Recombinant DNA pSG5-HA-PRB This study N/A pCMV-flag-SRC-3 WT/Mut This study N/A pGEX-SRC-2 fragments This study N/A 3XARE/PRE-E4 Yu et al.,2020 N/A Software and algorithms EMAN2.3 https://cryoem.bcm.edu/cryoem/downloads/view_eman2_versions N/A Relion3.1 https://www3.mrc-lmb.cam.ac.uk/relion/index.php/Download_%26_install N/A Gctf https://www2.mrc-lmb.cam.ac.uk/research/locally-developed-software/zhang-software/ N/A Chimera https://www.cgl.ucsf.edu/chimera/download.html N/A ChimeraX https://www.rbvi.ucsf.edu/chimerax/download.html N/A Other SerialEM http://bio3d.colorado.edu/SerialEM/ N/A Requests for reagents and further information should be directed to the lead contact, Bert W O'Malley ( [email protected] ). This study did not generate new unique reagents. Plasmids generated in this study will be available upon request. Sf9 insect cell line is a clonal isolate of Spodoptera frugiperda Sf21 cells (IPLB-Sf21-AE). Sf9 cells were cultured in ESF 921 Insect Cell Culture Medium. The protein purification methods are similar as the methods we published previously ( Yi et al., 2015 ; Yu et al., 2020 ). Sf9 insect cells were infected with His-PR-B expressing baculoviruses (that were produced in BCM Monoclonal Antibody/recombinant Protein Expression Core Facility). Cells were harvested 48 hours post-infection. Cells were washed and spined down at 5,000 rpm for 10 min at 4°C. Sf9 cells were further resuspended in a lysis buffer (50 mM Tris-HCl, pH 8.0; 150 mM NaCl; 100 nM progesterone; 0.5% NP40) and lysed using a homogenizer. After 40 min centrifugation at 15,000 rpm, the cleared lysate was incubated with Ni-NTA (Qiagen) resins, washed three times with wash buffer (50 mM Tris-HCl, pH8.0; 150 mM NaCl; 100 nM R1881; 0.05% NP40; 25 mM imidazole) and finally eluted with elution buffer (50 mM Tris-HCl, pH8.0; 150 mM NaCl; 100 nM progesterone; 0.05% NP40; 300 mM imidazole). Eluted protein was further applied on a gel-filtration column (Superdex 200 Increase 10/300 GL, GE Healthcare) pre-equilibrated with gel filtration buffer (50 mM Tris-HCl, pH8.0; 150 mM NaCl; 100 nM progesterone; 0.05% NP40). Purified protein fractions were pooled together (∼0.5 mg/mL) and for all the studies. Baculovirus-expressing His- and flag-tagged human recombinant SRC-2 purified through Nickel affinity purification followed by flag affinity purification. His-tagged human recombinant p300 protein was expressed in baculoviruses and purified through Nickel affinity purification. GST-fused SRC-2 fragments were expressed in E.coli and purified using glutathione sepharose beads. The methods are similar as the methods we published previously ( Yi et al., 2015 ; Yu et al., 2020 ). Different PR-B or SRC-2 fragments were fused to GST in a pGEX-4T1 vector and the proteins were expressed in E . coli . Bacterial lysates containing expressed different GST-fused proteins were incubated with 6 ul of glutathione sepharose 4B beads (GE Healthcare Life Sciences) for 1 hour and then washed 3 times with wash buffer (20mM HEPES pH 7.6, 150mM KCl, 1mM DTT, 0.1% NP40, 8% glycerol and protease inhibitor cocktail). The beads were then incubated with full-length recombinant proteins purified from baculovirus (SRC-2, p300 for GST-PR-B; PR-B, p300 for GST-SRC-2) for 3 hours. After extensive wash, the beads were boiled in a 2x SDS sample buffer and then loaded to a 4–15% SDS-PAGE. Chromatin reconstitution and in vitro transcription (IVT) of the GREB1 composite fragment (CompF) was performed as described ( Panigrahi et al., 2018 ). Briefly, each IVT reaction contained 0.2 pmoles of chromatinized templates and 10-fold molar excess of recombinant PR-B, and 50 μg of HeLa nuclear extracts (control as well as SRC-2/p300 immunodepleted) were used in each IVT reaction as indicated. Two pmoles of purified recombinant SRC-2 and p300 were used as indicated. After 50 min of IVT reaction, RNA was extracted with TriReagent, digested with Turbo-DNA-free DNase kit (Invitrogen AM1907), and analyzed by 1-step RT-qPCR using primers specific for GREB1 mRNA ( Panigrahi et al., 2018 ). Immunodepletion was essentially performed as detailed ( Panigrahi et al., 2018 ). One mg of HeLa NE was immunodepleted with control rabbit + mouse IgG (EMD-Millipore 12–370 and 12–371, respectively; 5 μg each) or antibodies against SRC-2 (Cell Signaling #96687) and p300 (Santa Cruz sc-48343), 5 μg each. For the structure of DNA-bound PR, purified His-tagged PR was incubated with a 32 bp PRE/ARE-consensus oligo (Santa Cruz, sc-2551) in the presence of 1 μM progesterone. For the structure of DNA-bound PR-B/SRC-2/p300 complex, a 324 bp long biotinylated PRE/ARE-containing DNA was used ( Yu et al., 2020 ). 0.6 μg of recombinant PR-B, SRC-2 and p300 proteins were incubated with 200 ng of PREDNA in the presence of 1 μM progesterone on ice for 1 hour. The mixture was then incubated with 15 μL Dynabeads M280 streptavidin (Invitrogen) for 15 min at room temperature followed by restriction enzyme digestion to release the DNA-bound protein complex from the beads. The sample was kept on ice before vitrification ( Dubochet et al., 1988 ) on the grid. A 3.5 μL sample was applied onto a 200-mesh R1.2/1.3 Quantifoil holy carbon grid covered with graphene oxide (GO)-coated grid following the same preparation pipeline ( Yu et al., 2020 ). After applying the sample, the grid was blotted and rapidly frozen in liquid ethane using a Vitrobot IV (FEI), with constant temperature and humidity during the process of blotting. The grid was stored in liquid nitrogen before imaging. 7,011 movie stacks for PRE-DNA/PR-B were collected at 200 kV on the Falcon 4 Electron Detector on the Glacios™ Cryo-TEM (Thermo Fisher). Images of PRE-DNA/PR-B were collected in dose fractionation super-resolution counting mode at 150,000x magnification, corresponding to a calibrated physical pixel size of 0.968 Å, with a defocus range from −1.0 to −2.6 μm. The total exposure time for the dataset was 8s, leading to a total accumulated dose of 55 electrons∗Å −2 on the specimen. Each image stack was fractionated into 40 subframes, each with an accumulation time of 0.2 s per frame. 1,201 movie stacks for PRE-DNA/PR-B/SRC-2/p300 complex at 300 kV on K2 summit on the NCEF Titan Krios. Images of PRE-DNA/PR-B/SRC-2/p300 complex were collected in dose fractionation super-resolution counting mode at 105,000x magnification, corresponding to a calibrated physical pixel size of 1.32 Å, with a defocus range from −1.0 to −2.5 mm. The total exposure time for the dataset was 10 s, leading to a total accumulated dose of 50 electrons∗Å −2 on the specimen. Each image stack was fractionated into 50 subframes, each with an accumulation time of 0.2 s per frame. Following the same procedure with PRE-DNA/PR-B, grids of the PRE-DNA/PR-B sample with N-Fab or C-Fab and PRE-DNA/PR-B/SRC-2/p300 sample with N-Fab were prepared. The images of these four samples were taken in the same procedure as PRE-DNA/PR-B mentioned above. The total image numbers of these three samples are shown in Table S1 . The final frame average of PRE-DNA/PR-B and PRE-DNA/PR-B/SRC-2/p300 were computed from averages of every three consecutive frames to correct beam-induced motion correction during exposure by MotionCor2 ( Zheng et al., 2017 ) CTF parameters of the particles in each frame average was determined by program e2ctf . py in EMAN2 ( Tang et al., 2007 ) and CTFFIND ( Rohou and Grigorieff, 2015 ). A total of 341,819 particles images for PRE-DNA/PR-B and 194,158 particles images for PRE-DNA/PR-B/SRC-2/p300 were automatically boxed out by newly developed e2boxer.py in EMAN2 ( Bell et al., 2016 ) with a box size of 280 × 280 pixels using the averaged sum of 40 and 50 raw frames (representative) per specimen area. The particle intensity in each frame was weighted according to a radiation damage model ( Guo et al., 2020 ; Wang et al., 2014 ; Zheng et al., 2017 ) (courtesy of B. Bammes of Direct Electron, LP). 2D reference free class averages were computed by RELION 3.1 ( Scheres, 2012 ). Initial models for every reconstruction were generated from scratch by e2initialmodel . py program using selected good quality 2D averages based on the 2D averages results. Refinements were carried out by RELION 3.1 and cryoSPARC, low pass filtered to 60Å, with search angle of 7.5 degrees for 11 iterations and subsequently with an angular sampling of 0.9375 degrees for 14 iterations. 42,619 particles images for PRE-DNA/PR-B and 15,779 particles images for PRE-DNA/PR-B/SRC-2/p300 were applied to final reconstruction yielding a resolution of 10.9 Å and 19.1 Å, respectively, at 0.143 of the Fourier shell correlation (FSC). The PR-B binds to the PREDNA at the center of the complex density. During refinement and classification, approximately 60% of particles were observed without bound DNA, which were excluded from the final refinement, but the DNA density is still somewhat weaker than expected due to particle classification accuracy in ice. Following the same procedure, other antibody related structures were carried out. The particle numbers and final resolution are shown in Table S1 . Full segmentation of the complex followed the same protocol as ERα complex and AR complex using Segger in Chimera with watershed algorithm ( Pintilie and Chiu, 2012 ). Regions of each segment were identified by specific antibody labeling first. The ratio between each segmented component volume is consistent with the ratio of corresponding domain masses. The boundary of PRE-DNA/PR-B segmented domains was determined according to the rigid body docking result of LBD crystal structures and the relative ratios of each domain. The boundary between each component in the PRE-DNA/PR-B/SRC-2/p300 complex was refined based on the previous knowledge of PR-B, p300 and SRC-2. The ratio between each component was validated by the molecular weight ratio as described in the previous ERα and AR structures ( Yi et al., 2015 ; Yu et al., 2020 ). The quantification and statistical analyses are integral parts of the software and algorithms used. Details are described in the main text and STAR Methods sections.