Tumor Susceptibility Gene-101 Regulates Glucocorticoid Receptor Through Disorder-Mediated Allostery

bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license.

Tumor susceptibility gene-101 regulates glucocorticoid receptor through disorder-mediated allostery

Jordan T. White1, James Rives2, Marla E. Tharp1, James O. Wrabl1, E. Brad Thompson1,3, Vincent J. Hilser*1,4

1Department of Biology at Johns Hopkins University, Baltimore, MD 21218; 2Department of Chemistry at Johns Hopkins University; 3Sealy Center for Structural Biology and Molecular Biophysics in the Department of Biochemistry and Molecular Biology at Univ. of Texas Medical Branch, Galveston, TX; 4T. C. Jenkins Department of Biophysics at Johns Hopkins University

Running title: TSG101 allosterically regulates GR

*To whom correspondence should be addressed: Vincent Hilser: Department of Biology, Johns Hopkins University, Baltimore MD 21218; [email protected]; Tel.(410)-516-6072

Keywords: glucocorticoid receptor, tumor susceptibility gene-101, allostery, allosteric regulation, transcription, transcription coregulator, transcription regulation, thermodynamics, biophysics

Abstract Introduction Tumor Susceptibility Gene-101 (TSG101) is Transcription factors (TFs) are multi- involved in endosomal maturation and has been domain, allosteric proteins, in which the binding implicated in the transcriptional regulation of of one effector ligand to its specific domain causes several steroid hormone receptors (SHRs), a conformational change in one or more other although a detailed characterization of such domains, so as to alter transcription. Most TFs regulation has yet to be conducted. Here we contain intrinsically disordered regions (IDRs), directly measure binding of TSG101 to one SHR, which though critical for TF function, are still not glucocorticoid receptor (GR). Using biophysical well understood (1). Generally, TF IDRs consist and cellular assays, we show that the coiled-coil of large ensembles of rapidly interconverting domain of TSG101; 1) binds and folds the conformers, including in many cases, conformers disordered N-terminal domain (NTD) of GR, 2) that contain some degree of structure. By upon binding, improves DNA-binding of GR in definition then, allostery arises due to an effector- vitro, and 3) enhances the transcriptional activity driven shift in the distribution of states such that of GR in vivo. Our findings suggest that TSG101 states that are capable of initiating transcription is a bona fide transcriptional co-regulator of GR. are either promoted or repressed, thus affecting transcriptional activity accordingly (2,3). We have adopted the glucocorticoid receptor (GR) as a model system for our studies because the NTD of GR is known to be intrinsically disordered (ID)

1 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

and also to be critical for the TF function of the which adopts the canonical zinc-finger binding receptor. motif (17,18). Conversely, we showed that The GR is a 3-domain protein comprised although the NTD of GR appears in its entirety to of (from N to C terminus): N-terminal (NTD), be an IDR, it actually consists of two adjacent, but DNA-binding (DBD), and ligand-binding domains functionally distinct IDRs. The first IDR, which (LBD), where L in this case refers to steroid or we call the functional (or F-domain) includes steroid substitute. However, if the LBD is deleted, residues 99-420. We showed that the F-domain the 2-domain NTD-DBD is constitutively active exists as an equilibrium between a disordered (or due to a potent activation function-1 (AF1) region unfolded) state and a folded ordered state, and that located in its NTD, a fact that makes the GR NTD- the folded state of the F-domain is responsible for DBD amenable to exploring the allosteric impact transcriptional activation. of DNA binding on transcriptional activation, as The second functionally distinct IDR of well as the effects of changes to the NTD on DNA the NTD consists of residues 1-98 (Fig. 1A), and binding (Fig. 1A). We and others have shown that this region serves as a variable-length regulator (or the ID NTD can be made to cooperatively fold R-domain). The R-domain influences both the F- into conformers containing both 2º and 3º domain of the NTD as well as the DBD, providing structure, by use of; 1) protective organic GR with genetically “tunable” energetic osmolytes (4-7), 2) known binding partner frustration (13), whereby DNA binding to the proteins (8-10), 3) phosphorylation of key serines DBD both directly stabilizes (and thus activates) (11,12), and 4) addition of the DBD (13,14). the F-domain, but also indirectly destabilizes Previously, we hypothesized, using a (therefore, repressing) the F-domain through rigorous thermodynamic model, that under a wide stabilization of the R-domain. Thus, our results range of initial energetic conditions, inclusion of revealed that all the GR NTD-DBD isoforms that an IDR in a protein could enhance its allosteric contain the AF1 region (e.g., A, B1, B3 C1, and behavior (15). Specifically, in a 2-domain model C2), with the exception of the C3 isoform, actually in which each domain is capable of binding its consist of three domains; the DBD, the F-domain, own specific ligand, when one or both domains is and variable lengths of the R-domain. Because the ID, binding of ligand to one domain can greatly GR C3 is the only isoform lacking the R-domain influence the probability of the second domain (19), it represents an ideal system to interrogate being in a favorable folded state for binding its the direct impact of changes to the NTD on the ligand. Further, we showed this ensemble DBD, as the absence of any R-domain should allosteric model (EAM) could be applied to greatly simplify interpretation. proteins with multiple domains, allowing us to As a transcriptional cofactor for these probe the role of ID in mediating allosteric control studies we selected the Tumor Susceptibility Gene in numerous domain topologies (16). Indeed, in 101 (TSG101), a protein that is primarily known validating this model, we demonstrated that DNA for its role in the ESCRT-I complex (i.e., binding affinity and transcriptional activity for a Endosomal Sorting Complex Required for series of isoforms of GR could be described in the Transport-I), promoting endosomal maturation context of inter-domain coupling in the GR (13), (19,20). However, it has also been reported to thus, establishing one tenet of the EAM, that bind to the GR, and to possibly act as a cofactor stabilization of an input domain (i.e. the DBD) for GR and other TF’s (Table S1 and (21-25)). The could affect the function of the activation domain. domains of TSG101 that are critical to these Here, we set out to test the reciprocal interactions are shown in Figure 1B. Herein, we prediction of the model; that binding a investigate the binding of TSG101 to GR and the transcriptional cofactor ligand to the NTD could ability of this binding to allosterically affect GR either positively or negatively (depending on the DBD binding to its canonical palindromic DNA sign of the coupling) tune the affinity of the DBD. binding site (i.e., the GR response element To address this question, we utilized the 2-domain (GRE)), as well as influence subsequent GR system described above (13). Importantly, transcriptional activity. within this construct, the DBD is known to be a structured, albeit highly dynamic, globular protein, Results

2 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

TSG101 binds within the GR NTD Activation (i.e.~120-200) and following (i.e.~240-300) Function-1 (AF1) region residues in the AF1 core. Interestingly, the It has been reported previously that the orientation of the cross-links is suggestive of an coiled-coil region of TSG101 (TSG101cc) binds to inversion in the register of the interaction, the AF1 region of the GR NTD (23). Here we set whereby the GR NTD is aligned parallel with out to establish whether this binding is specific TSG101cc but then doubles back on TSG101cc, and whether it affects the functional coupling we aligning in antiparallel (Fig. 4 inset). Importantly, previously revealed in GR (13). Importantly, we these results are consistent with a model whereby confirmed previous results and extended our the TSG101cc—which is known to form a understanding using quantitative binding studies heterotrimeric coiled-coil in its functional ESCRT and protease protection assays. First, we complex (27)—interacts with the GR NTD in the examined the binding of TSG101cc to the single- two regions before and after the AF1 core residues domain GR C3 NTD and to the two-domain GR (5), thus mimicking the trimeric coiled-coil seen in C3 NTD-DBD (Fig. 2). Our results reveal that the ESCRT complex. We note that this proposed TSG101cc binds to the two constructs with 1:1 interaction model is also consistent with our stoichiometry, with TSG101cc binding more structural thermodynamic analysis of GR using the tightly to the isolated C3 NTD (Kd = 19.6 ± 1.0 COREX algorithm, whereby the two regions of the µM, see Table 1) than to the two-domain C3 GR NTD that interact with TSG101cc are NTD-DBD (Kd = 72.6 ± 14.6 µM). Similar results predicted to be highly helical (See Fig. 4, were obtained with the GR A isoform (Fig. S1), Supporting Information, Fig. S6, Supporting Excel and importantly, TSG101cc does not bind the Spreadsheet, and (28)). isolated DBD (data not shown). To determine the location on the GR that TSG101cc alters GR NTD-DBD affinity for DNA interacts with TSG101cc, we conducted limited sequences corresponding to the GRE proteolysis of the bound complex. As seen in the Previously, we presented an ensemble electrophoretic gels (Fig. 3A), trypsin digestion of based thermodynamic model that allowed us to GR C3 NTD in the presence TSG101cc resulted in investigate the role of ID in mediating allostery protection of a series of 14.8-20.9 kDa peptides (15,16). Using this model, we demonstrated the from the GR AF1 region. Mass spectroscopic importance of ID in the allosteric behavior of 2 analysis of the protected bands identified a 9.1 and 3-domain proteins (15,16). Applied to the kDa fragment of GR (Fig. 3B, also see Figs. S2-3 GR, the model predicts that the DBD and the AF1- and Table S2), leaving about 10 kDa unaccounted containing region of the NTD (termed the F- for (although see below). These results agree with domain in (4)) are positively coupled, meaning previous yeast-two hybrid analysis (23), and stabilization of the high affinity state of the DBD suggest specific binding to the NTD, consistent increases the stability of the active conformation with the observed binding effects (Fig. 2) of the F-domain. If this model behaves To gain insight into the structural reciprocally, binding of a ligand (e.g. TSG101cc) organization of the bound TSG101cc:GR C3 to the active state of the F-domain should also NTD, the cross-linker disuccinimidyl tartrate enhance the binding of the GR DBD to its ligand, (DST) was employed, resulting in cross-linking the palindromic sequence of DNA that forms a between TSG101cc and a ~19 kDa stretch of the GRE. Consistent with this prediction, the DNA GR C3 NTD (Fig. S4 and Tables S3-8). The binding affinity of the TSG101cc-bound GR C3 results indicate that the GR binding site stretches isoform is increased relative to the GR C3 alone, from approximately residue 121 to 301. Of note, when tested on DNA sequences corresponding to fluorescence anisotropy suggests that the AF1 core several symmetric GREs, although the magnitude region (a.a. 187—244), which binds TATA-box of the increase appears to be sequence dependent Binding Protein (TBP (26)), does not appear to (Fig. 5, Table 1, and Fig. S7-8). Importantly, the interact with TSG101cc (Fig. S5). This finding relative increase in affinity of GR C3 NTD-DBD taken together with the protease protection and for DNA when bound to TSG101cc is also cross-linking results, indicates that the interactions observed for the GR A NTD-DBD isoform (Fig. with TSG101cc involve the amino acids prior to 5), suggesting that the presence of the repressor R-

3 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

domain does not qualitatively affect the ability of TSG101 co-localizes with GR in the nucleus, and TSG101cc to allosterically regulate the DBD both TSG101 and TSG101cc enhance through its interaction with the F-domain. transcription from glucocorticoid response elements (GREs) The effect of energetic frustration on TSG101:GR To determine whether the in vitro effects A binding reported above are consistent with the in vivo The above experiments reveal an overall impact, GR-negative U2OS cells were transfected positive allosteric coupling between TSG101cc with TSG101 and GR. TSG101 was distributed binding to the GR NTD and DNA binding to the throughout both the cytoplasm and the nucleus, in GR DBD. We also tested the reciprocal effect both endogenous and overexpressed conditions which would stipulate that DNA-bound GR should (Fig. 7 and Fig. S9). GR NTD-DBD localized to bind more tightly to TSG101cc. By titrating the nucleus, as is known to occur with GR lacking TSG101cc binding in a series of experiments with an LBD (13), and co-localized with both GR bound to palindromic GRE, we obtained an endogenous and overexpressed TSG101 in the apparent binding affinity for TSG101cc and DNA- nucleus (Fig. 7). bound GR (Fig. 6A-B). The results confirmed the Transfection of full-length TSG101 prediction: the apparent TSG101cc binding together with GR A or C3 NTD-DBD increased affinity is more than ten-fold tighter in this the expression of Gaussia luciferase driven by a context. GRE, in a dose-dependent manner (Fig. 8). The Importantly, the binding of TSG101cc to TSG101cc alone marginally increased GR GR with and without DNA also illuminated the transcriptional activity, but deletion of the mechanistic impact of the regulatory R-domain TSG101 coiled-coil domain abrogated all that is present in the full-length GR A isoform, but stimulation (Fig. 8). Thus, TSG101 is clearly a which is missing in the GR C3 isoform. co-regulator for GR gene induction, with the Specifically, thermodynamic analysis of the coiled-coil domain being both necessary and stability and DNA binding affinities of various sufficient for TSG101-mediated stimulation of isoforms of GR (13), revealed that the GR A GR-driven transcription (although additional isoform exists in a frustrated state, wherein the R- TSG101 regions appear to enhance the effect). domain (Fig. 6A) decreases the probability of the Finally, to test whether the increased GR A being in its active conformation (29). This activity was mediated through the AF1 region of occurs because the R-domain is both positively the GR NTD, each of the eight translational coupled and, therefore, stabilized by the DBD, and isoforms of the GR were screened for activation also negatively coupled, therefore destabilizing, to by TSG101 in U2OS cells. Consistent with the the F-domain, which is responsible for interaction effect being mediated by the AF1-containing with TSG101cc. Consequently, the impact of region, the activity of all isoforms containing AF1 TSG101cc should have an overall greater impact (i.e., GR-A, B, C1, C2, and C3) were enhanced by on the GR C3 isoform, as the frustration present in TSG101, while those without AF1 (i.e. GR-D1, the GR A isoform prevents the DBD from fully D2, and D3) showed no response (Fig. S10). populating the high affinity state for DNA binding. Further, when tested with full-length GR A (α), In ensemble terms, under apparent saturating which contains the steroid binding LBD, TSG101 conditions of TSG101cc, only a subset of the was able to hyper-activate GR so long as an ensemble is competent to bind DNA with high activating steroid was administered (Fig. 9). affinity. As such even though the DBD is identical Administration of a weak, mixed agonist (i.e., in the GR A and GR C3 isoforms, the change in RU486) caused nuclear localization of GR (Fig. binding affinity with increasing TSG101cc S11-12), but inhibited hyper-activation of GR by saturates at a lower level for GR A NTD-DBD TSG101 and inhibited activation by than it does for the GR C3 NTD-DBD (Fig. 6C). dexamethasone (Fig. 9). These effects are likely related to RU486’s competition for the LBD’s steroid binding site and known recruitment of corepressor proteins (30,31).

4 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

the nucleus. However, it is clear that TSG101 can Discussion enter the nucleus and upon GR binding DNA, the We have examined the allosteric behavior interaction with TSG101 is greatly promoted (Fig. of ID-containing regulatory proteins by use of a 10A). Importantly, our data are consistent with a well-studied ID transcription factor, GR, and the model of GR and TSG101cc forming a dimeric effect of TSG101 on its DNA binding affinity and triple helix, wherein the GR NTD folds back on activity. First, we confirm and extend earlier the TSG101cc in a manner reminiscent of the anti- findings that the TSG101 coiled-coil domain binds parallel ESCRT-I triple helical coiled-coil (20). the ID AF1-containing region of the GR (i.e. the F Such a binding model would likely mimic the domain). We have improved the mapping of the burial of hydrophobic surface that is seen in the interactions of TSG101cc and the GR NTD, and heterotrimeric coiled-coil of the ESCRT-I we provide novel evidence that binding between complex, as shown previously (27). Confirmation TSG101cc and GR causes tertiary structure to of this hypothesis awaits further study. form in the NTD. Furthermore, we show that GR Finally, it is also important to note that the and TSG101 co-localize in the nucleus, supporting proposed model agrees with two other previously the view that they interact in vivo. Importantly, published reports (Fig. 10B). First, Copik, et al and counter to what was observed previously (21), (26) found that the TATA box binding protein TSG101 causes increased transcriptional activity (TBP), which binds to the TATA box as a of the GR. Hence, in the systems studied here, monomer (33), recruits and interacts with two TSG101 is clearly a GR co-regulator that enhances different AF1 cores (Fig. 10B, right). Second, the transcription. anti-parallel alignment of the two regions of GR With regard to the functionally important NTD that interact with TSG101cc yields a regions of TSG101, we find that the coiled-coil compelling and structurally reasonable model for domain is both necessary and sufficient for how the R-region of GR, which is the most N- activation, although additional as yet undefined terminal region of the NTD, can interact with its regions of TSG101 are needed for maximal effect. own DBD, an interaction recently reported by Li, Toward this end, studies on the androgen receptor et al. (13) (Fig. 10B, left). (AR) suggest that the TSG101 UEV and S-box domains may also be important for interactions Conclusion with transcriptional machinery (25,32). We have shown that TSG101cc interacts We also found that interaction of TSG101 with the GR NTD, confirming previous findings, with GR increases the affinity of the GR DBD for and that the interaction enhances activity in vivo. palindromic GRE DNA sequences, results that Furthermore, we show that TSG101cc directly confirm and support the predictions of our allosterically affects DBD binding of DNA and model for allosteric proteins containing ID regions that DNA binding allosterically affects NTD (16). Namely, ligand-induced binding and folding binding of TSG101cc. Our results thus indicate of an ID domain of a protein can allosterically that TSG101cc is a bona fide transcriptional co- influence the binding to a structured domain. Also regulator of GR function. consistent with our previous work, we found that the GR A isoform is energetically frustrated by its Experimental Procedures R-domain; thereby lowering the degree to which Cloning of Plasmids Expressing Human GR or TSG101cc is able to promote GR A binding to TSG101 DNA. The functional implications of these All E. coli expression was achieved with isoform dependent effects are currently not the pJ411 plasmid of DNA2.0 (now ATUM, CA known. USA), and the cloning of most constructs was In total, our results suggest a model described previously (4,13). For the largest GR whereby TSG101cc interacts with a helical region NTD constructs (A = a.a. 1—420, C3 = a.a. 99— of GR AF1 (Fig. 10). Given the known 420), the coding sequence was tagged with a 9x functionally relevant binding partners unique to His tag on both the N and C-termini along with both TSG101 and GR in the cytosol, it is likely TEV cut sites to remove the tags. The smaller AF1 that TSG101 and GR interact weakly outside of core of GR (a.a. 187—244) was tagged with an N-

5 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

terminal 9xHis-TEV tag. The NTD-DBD from Thermo-Fisher Scientific or ProteoChem. constructs of GR (A = a.a. 1—525, C3 = a.a. 99— BS3 and EDC proved to be unusable in these 525) were tagged on just the N-terminus with a 6x circumstances, either because of too much or no His-Thrombin sequence (expression problems cross-linking, respectively. The rest of this were observed with TEV or longer His tags). The manuscript only refers to DST cross-linking. The open reading frame of the TSG101 coiled coil (a.a. DST cross-linking buffer was the same as the 229—304) was tagged with a 9x His tag followed fluorescence anisotropy buffer (below). Initial by a TEV site and a serine on the N-terminus (27). reactions were carried out according to the See Supporting Information for sequences used. manufacturer protocols, but because TSG101cc Besides luciferase, all U2OS expression of has a large number of lysines (11 out of 78 a.a.) TSG101 or GR NTD-DBD was achieved with the there was an excessive amount of TSG101cc self pJ603 plasmid of DNA2.0, and the cloning of the cross-linking. After lowering the concentration of GR constructs was described previously ((13); A = DST to about one-sixth or one-twelfth the a.a. 1—525, B = a.a. 27—525, C1 = a.a. 86—525, recommended amount (0.5 mM used here) we C2 = a.a. 90—525, C3 = a.a. 99—525, D1 = a.a. were able to detect a faint GRxTSG101cc cross- 316—525). Note that the pJ603 plasmid uses a link without TSG101cc background signal. See the constitutive promoter (cytomegalovirus) to express Supporting Information for cross-linking and the inserted gene. DNA2.0 generated the full- MALDI mass spectrometry details. length TSG101 expression plasmid with codon optimization for human cell expression. See the Thermodynamic Fold Recognition Supporting Information for sub-cloning details. To gain insight into which structure(s) For studies of full-length GR, pEGFP GR was a might be energetically compatible with the GR- gift from Alice Wong (Addgene plasmid #47504). NTD, the amino acid sequence (residues 1-420) was threaded against all domains contained in the Protein Expression and Purification Protein Data Bank (PDB) (34), as annotated in the Expression and purification of TSG101cc ECOD database version 88 (35). The basic fold and the various GR constructs used here has recognition algorithm used is described in (28), largely been described before (4,13,27) and with the addition of three modifications specific to detailed methods can be found in the Supporting a database search using an IDP as query. First, Information text. See Figure S13 for examples of thermodynamic environments were obtained using GR and TSG101 protein purity. the eScape modification of COREX (36) instead of the original COREX algorithm (37), since the latter Protease Protection, Cross-Linking, and Mass requires a crystal structure as input. Second, Spectrometry sequence:structure matches were scored using Protease digestions were carried out in denatured state thermodynamic information in both the fluorescence anisotropy buffer (below) addition to native state information, as described and 20 mM Na2HPO4 + 50 mM NaCl pH 7.2, with in (38). Third, to increase sensitivity, only matches no observable differences. Saturation of GR achieving reciprocally significant scores were required at least a thirty-fold excess of TSG101cc. considered, meaning that compatible environments To control for the total amount of protein present between PDB and GR NTD sequence in each digest, the ratio of trypsin and targeted simultaneously exhibited compatible environments protein was held constant at 1000 µg protein:1 µg between GR NTD and PDB sequence. As trypsin, except where noted. All digests occurred described in (28), the significance (p-value) of a at room temperature after allowing the protein match was estimated by integrating a normal solutions at least 20 minutes to equilibrate. Each distribution between the limits of -∞ and the time point was taken by quenching an aliquot of observed max[native + denatured] scores of the reaction in Laemmli sample buffer and heating at match, where the mean µ and standard deviation σ 95º C for ~ 5 minutes. of the distribution were respectively defined as: µ Three cross-linkers were tested as follows, = 0.00525 – 0.326*L and σ = 0.818*sqrt(L), where with their linker lengths in parentheses: BS3 (11.4 L is the length of the PDB domain. Å), DST (6.4 Å), and EDC + sulfo-NHS (0 Å)

6 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

To determine secondary structure, the (streptomycin and penicillin). Cells were grown in structural coordinates of the aligned PDB hits 15 cm plates (Falcon) at 37º C in a 5% CO2 were subsequently run through the DSSP atmosphere and the medium was changed every 2– algorithm built into Chimera (39,40). Of the entire 3 days until passaging. Trypsin with EDTA was PDB, 125 domains cleared the threshold for used for detaching cells during passaging. For statistical significance (P < 10-5) and were used transfection, cells were plated at a density of 3x104 here. “Helix” described here is a combination of 3- cells per well on 96-well plates or 3x105 cells per 10, π, and α-helix. The resulting per-residue well on 6-well plates and grown to ~ 90% secondary structure counts were then summed and confluence before transfecting. Xtreme GENE HP normalized as a percentage of the total number of (Roche) was used to transfect DNA per the hits at a given residue. For presentation purposes, manufacturer directions with a resulting percentages were smoothed by a five amino acid transfection efficiency of about 20%. For studies rolling average with a step size of two. of full-length GR, cells were passaged using media supplemented with charcoal stripped FBS the day Fluorescence Anisotropy before transfection. The following buffer was used for all fluorescence anisotropy experiments shown here: Luciferase Assays 13.2 mM HEPES, 16 mM K Acetate, 67.2 mM Most luciferase assays were done using 2 NaCl, 4.2 mM MgCl2, and 0.84 mM TCEP all 96-well plates (0.3 cm wells, Falcon), except pH'd to 7.56 using 10 M KOH. It is difficult to comparison of TSG101 mutants proceeded by 6- purify large amounts of any GR construct that well plates (9.4 cm2 wells, Sigma). The NEB includes the disordered NTD. In order to Gaussia and Cypridina plasmids were used at 40 maximize the high concentration end of our ng each per well for each 96-well assay (or ten experiments, all the titrations presented here are times as much for 6-well plates). Two GR dilution experiments: The fluorophore (pyrene- response elements were cloned into the promoter TSG101cc or 6FAM-DNA) was diluted into a of the Gaussia plasmid and the Cypridina plasmid concentrated solution of GR and the GR was then served as a control (13). GR expression plasmids diluted out with a solution of fluorophore, with the were transfected at 0.6 ng per well and TSG101 concentration of fluorophore being kept the same plasmids were transfected from 0 up to 9.4 ng per throughout. The dilution solution was actively well, with the empty pJ603 plasmid balancing the held at 20º C during the experiments to prevent total transfection to 10 ng of pJ603 DNA (96-well thermal disequilibrium. plates). All experiments were at the very least For studies of full-length GR, 0.6 ng of duplicated with independent protein preparations, eGFP-GR expression plasmid was co-transfected and all replicate data sets were fit globally to with either 9.4 ng of empty pJ603 plasmid or 8 ng produce the presented values, except where noted. of TSG101 pJ603 and 1.4 ng of empty pJ603. Equilibration times were varied to check for About 16—20 hours later, 1 mM hormone stocks thermodynamic equilibrium, and because no in ethanol were diluted to 1.5 µM using media obvious effects were discerned from 1 to ~ 30 with charcoal stripped FBS. This hormone dilution minutes, most of the data presented were collected or diluted ethanol control was then added with 2 minute incubations between measurements. immediately to the cells for a final hormone The only sensitive incubation step was the first concentration of 100 nM and final ethanol measurement, which needed to be about 20 concentration of 0.02%. In order to maintain the minutes long to establish thermal equilibrium. See same concentration of ethanol across all Supporting Information for instrumental and experiments, when only one hormone was diluted fitting details. to 1.5 µM, an equivalent volume of ethanol was added to yield the same final ethanol concentration Cell Culture as in dual-hormone experiments. Supernatants U2-osteosarcoma cells (U2OS) were were collected about 24 hours after hormone maintained in McCoy's 5A medium, supplemented addition. with 10% fetal bovine serum and antibiotics

7 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

Gaussia and Cypridina substrates were are constitutively excreted, our luciferase obtained from NEB or its original supplier, measurements are largely a time average of Targeting Systems. Either a Berthold or Promega everything occurring before the shown artifact. luminometer were used for automated data We also observed repression of GR in some collection. groups of older, high-passage cells, and the presented data are with low-passage cells only (< Confocal Microscopy 7). Cells were passaged onto German glass TSG101-tdTom was used as an alternative cover slips in 6-well plates. For the verification that TSG101 is in U2OS cell nuclei. immunostaining data, there were two transfection For expression of TSG101-tdTom, 105 cells were procedures, both using 3 x 105 cells per well. The passaged and the expression plasmid was first addressed whether or not TSG101 is in the transfected at 15 ng per well and sonicated salmon nucleus of U2OS cells, with or without NTD-DBD sperm DNA was used to bring the total DNA GR, and was also used to determine full-length amount per well to 500 ng. The TSG101-tdTom GR localization. Each pJ603 plasmid (TSG101 cells were fixed ~ 24 hours after transfection. and/or GR) was transfected at 100 ng and All immunostaining was similar to our sonicated salmon sperm DNA was used to bring previous work and details can be found in the the total amount of DNA to 1000 ng per well, and Supporting Information (13). the cells were fixed about 40 hours after All microscopy was done using the transfection. Transfection and induction of full- Integrated Imaging Center of Johns Hopkins length GR proceeded as stated above. University. The data shown were collected using The second procedure addressed how either a Zeiss LSM 700 (immunostaining) or 780 TSG101 localization changes in response to the (TSG101-tdTom) confocal microscope. DAPI was luciferase assay conditions, over time. excited with a 405 nm diode laser, AlexaFluor 488 Transfection proceeded in the same manner as and eGFP (GR) were excited with a 488 nm laser, with luciferase assays. Cells were then collected and both AlexaFluor 594 and tdTom (TSG101) for fixation 16 or 40 hours later. A cytosolic were excited with a 561 nm laser. All imaging was aberration appeared about 40 hours post- done at room temperature with 400x or 630x transfection in the group with the highest amount magnification oil objectives (1.4 NA) using ZEN of TSG101 plasmid (Fig. S14-15). This coincided acquisition software. The gain and digital offset with cell death, as seen by floating cells in the were optimized for the best signal-to-noise in each associated wells. Overexpression of TSG101 image; thus, the absolute intensities are not previously caused both of these effects in literature necessarily comparable. Analysis of images was data (41,42). Because we collected cells ~ 40 done using the Fiji version of ImageJ (43,44). hours post transfection and both of our luciferases

Acknowledgements: Professors Brand, Schroer, Hedgecock, Wendland, and Moudrianakis for sharing instruments, supplies, and advice. Dr Katie Tripp and the Center for Molecular Biophysics at Johns Hopkins University.

Conflict of Interest: The authors declare that they have no conflicts of interest with the contents of this article.

Author Contributions: JTW, JR, JOW, EBT, and VJH designed experiments. JTW, JR, MT, and JOW executed experiments. JTW and JOW analyzed data. All authors contributed to the final paper.

References

1. Liu, J., Perumal, N. B., Oldfield, C. J., Su, E. W., Uversky, V. N., and Dunker, A. K. (2006) Intrinsic Disorder in Transcription Factors. Biochemistry 45, 6873-6888

8 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

2. Uversky, V. N. (2012) Intrinsically disordered proteins and novel strategies for drug discovery. Expert opinion on drug discovery 7, 475-488 3. Kumar, R., and Thompson, E. B. (2003) Transactivation functions of the N-terminal domains of nuclear hormone receptors: protein folding and coactivator interactions. Mol Endocrinol 17, 1-10 4. Li, J., Motlagh, H. N., Chakuroff, C., Thompson, E. B., and Hilser, V. J. (2012) Thermodynamic Dissection of the Intrinsically Disordered N-terminal Domain of Human Glucocorticoid Receptor. J Biol Chem 287, 26777-26787 5. Baskakov, I. V., Kumar, R., Srinivasan, G., Ji, Y. S., Bolen, D. W., and Thompson, E. B. (1999) Trimethylamine N-oxide-induced cooperative folding of an intrinsically unfolded transcription- activating fragment of human glucocorticoid receptor. J Biol Chem 274, 10693-10696 6. Kumar, R., Serrette, J. M., Khan, S. H., Miller, A. L., and Thompson, E. B. (2007) Effects of different osmolytes on the induced folding of the N-terminal activation domain (AF1) of the glucocorticoid receptor. Arch Biochem Biophys 465, 452-460 7. Kumar, R., Lee, J. C., Bolen, D. W., and Thompson, E. B. (2001) The conformation of the glucocorticoid receptor af1/tau1 domain induced by osmolyte binds co-regulatory proteins. J Biol Chem 276, 18146-18152 8. Kumar, R., Volk, D. E., Li, J., Lee, J. C., Gorenstein, D. G., and Thompson, E. B. (2004) TATA box binding protein induces structure in the recombinant glucocorticoid receptor AF1 domain. Proc Natl Acad Sci U S A 101, 16425-16430 9. Kumar, R., Baskakov, I. V., Srinivasan, G., Bolen, D. W., Lee, J. C., and Thompson, E. B. (1999) Interdomain signaling in a two-domain fragment of the human glucocorticoid receptor. J Biol Chem 274, 24737-24741 10. Khan, S. H., Awasthi, S., Guo, C., Goswami, D., Ling, J., Griffin, P. R., Simons, S. S., Jr., and Kumar, R. (2012) Binding of the N-terminal region of coactivator TIF2 to the intrinsically disordered AF1 domain of the glucocorticoid receptor is accompanied by conformational reorganizations. J Biol Chem 287, 44546-44560 11. Garza, A. M., Khan, S. H., and Kumar, R. (2010) Site-specific phosphorylation induces functionally active conformation in the intrinsically disordered N-terminal activation function (AF1) domain of the glucocorticoid receptor. Mol Cell Biol 30, 220-230 12. Kumar, R., and Thompson, E. B. (2019) Role of Phosphorylation in the Modulation of the Glucocorticoid Receptor's Intrinsically Disordered Domain. Biomolecules 9 13. Li, J., White, J. T., Saavedra, H., Wrabl, J. O., Motlagh, H. N., Liu, K., Sowers, J., Schroer, T., Thompson, E. B., and Hilser, V. J. (2017) Genetically tunable frustration controls allostery in an intrinsically disordered transcription factor. eLife 6 14. Kumar, R., and Thompson, E. B. (2010) Influence of flanking sequences on signaling between the activation function AF1 and DNA-binding domain of the glucocorticoid receptor. Arch Biochem Biophys 496, 140-145 15. Hilser, V. J., and Thompson, E. B. (2007) Intrinsic disorder as a mechanism to optimize allosteric coupling in proteins. Proc Natl Acad Sci U S A 104, 8311-8315 16. Motlagh, H. N., and Hilser, V. J. (2012) Agonism/antagonism switching in allosteric ensembles. Proc Natl Acad Sci U S A 109, 4134-4139 17. Hard, T., Kellenbach, E., Boelens, R., Maler, B. A., Dahlman, K., Freedman, L. P., Carlstedt- Duke, J., Yamamoto, K. R., Gustafsson, J. A., and Kaptein, R. (1990) Solution structure of the glucocorticoid receptor DNA-binding domain. Science 249, 157-160 18. Baumann, H., Paulsen, K., Kovacs, H., Berglund, H., Wright, A. P. H., Gustafson, J. A., and Hard, T. (1993) Refined Solution Structure of the Glucocorticoid Receptor DNA-binding Domain. Biochemistry 32, 13463-13471 19. Babst, M., Odorizzi, G., Estepa, E. J., and Emr, S. D. (2000) Mammalian tumor susceptibility gene 101 (TSG101) and the yeast homologue, Vps23p, both function in late endosomal trafficking. Traffic 1, 248-258

9 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

20. Kostelansky, M. S., Schluter, C., Tam, Y. Y., Lee, S., Ghirlando, R., Beach, B., Conibear, E., and Hurley, J. H. (2007) Molecular architecture and functional model of the complete yeast ESCRT-I heterotetramer. Cell 129, 485-498 21. Watanabe, M., Yanagi, Y., Masuhiro, Y., Yano, T., Yoshikawa, H., Yanagisawa, J., and Kato, S. (1998) A putative tumor suppressor, TSG101, acts as a transcriptional suppressor through its coiled-coil domain. Biochem Biophys Res Commun 245, 900-905 22. Chua, H. H., Lee, H. H., Chang, S. S., Lu, C. C., Yeh, T. H., Hsu, T. Y., Cheng, T. H., Cheng, J. T., Chen, M. R., and Tsai, C. H. (2007) Role of the TSG101 gene in Epstein-Barr virus late gene transcription. Journal of virology 81, 2459-2471 23. Hittelman, A. B., Burakov, D., Iniguez-Lluhi, J. A., Freedman, L. P., and Garabedian, M. J. (1999) Differential regulation of glucocorticoid receptor transcriptional activation via AF-1- associated proteins. The EMBO journal 18, 5380-5388 24. Muromoto, R., Sugiyama, K., Yamamoto, T., Oritani, K., Shimoda, K., and Matsuda, T. (2004) Physical and functional interactions between Daxx and TSG101. Biochem Biophys Res Commun 316, 827-833 25. Burgdorf, S., Leister, P., and Scheidtmann, K. H. (2004) TSG101 interacts with apoptosis- antagonizing transcription factor and enhances androgen receptor-mediated transcription by promoting its monoubiquitination. J Biol Chem 279, 17524-17534 26. Copik, A. J., Webb, M. S., Miller, A. L., Wang, Y., Kumar, R., and Thompson, E. B. (2006) Activation function 1 of glucocorticoid receptor binds TATA-binding protein in vitro and in vivo. Mol Endocrinol 20, 1218-1230 27. White, J. T., Toptygin, D., Cohen, R., Murphy, N., and Hilser, V. J. (2017) Structural Stability of the Coiled-Coil Domain of Tumor Susceptibility Gene (TSG)-101. Biochemistry 56, 4646-4655 28. Hoffmann, J., Wrabl, J. O., and Hilser, V. J. (2016) The role of negative selection in protein evolution revealed through the energetics of the native state ensemble. Proteins 84, 435-447 29. White, J. T., Li, J., Grasso, E., Wrabl, J. O., and Hilser, V. J. (2018) Ensemble allosteric model: energetic frustration within the intrinsically disordered glucocorticoid receptor. Philos Trans R Soc Lond B Biol Sci 373 30. Zhang, X., Jeyakumar, M., Petukhov, S., and Bagchi, M. K. (1998) A nuclear receptor corepressor modulates transcriptional activity of antagonist-occupied steroid hormone receptor. Mol Endocrinol 12, 513-524 31. Schulz, M., Eggert, M., Baniahmad, A., Dostert, A., Heinzel, T., and Renkawitz, R. (2002) RU486-induced glucocorticoid receptor agonism is controlled by the receptor N terminus and by corepressor binding. J Biol Chem 277, 26238-26243 32. Felten, A., Brinckmann, D., Landsberg, G., and Scheidtmann, K. H. (2013) Zipper-interacting protein kinase is involved in regulation of ubiquitination of the androgen receptor, thereby contributing to dynamic transcription complex assembly. Oncogene 32, 4981-4988 33. Nikolov, D. B., Chen, H., Halay, E. D., Hoffman, A., Roeder, R. G., and Burley, S. K. (1996) Crystal structure of a human TATA box-binding protein/TATA element complex. Proc Natl Acad Sci U S A 93, 4862-4867 34. Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., Shindyalov, I. N., and Bourne, P. E. (2000) The Protein Data Bank. Nucleic Acids Res 28, 235-242 35. Cheng, H., Schaeffer, R. D., Liao, Y., Kinch, L. N., Pei, J., Shi, S., Kim, B. H., and Grishin, N. V. (2014) ECOD: an evolutionary classification of protein domains. PLoS computational biology 10, e1003926 36. Gu, J., and Hilser, V. J. (2008) Predicting the energetics of conformational fluctuations in proteins from sequence: a strategy for profiling the proteome. Structure 16, 1627-1637 37. Hilser, V. J., Garcia-Moreno, E. B., Oas, T. G., Kapp, G., and Whitten, S. T. (2006) A statistical thermodynamic model of the protein ensemble. Chemical reviews 106, 1545-1558

10 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

38. Wang, S., Gu, J., Larson, S. A., Whitten, S. T., and Hilser, V. J. (2008) Denatured-state energy landscapes of a protein structural database reveal the energetic determinants of a framework model for folding. Journal of molecular biology 381, 1184-1201 39. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C., and Ferrin, T. E. (2004) UCSF Chimera--a visualization system for exploratory research and analysis. Journal of computational chemistry 25, 1605-1612 40. Kabsch, W., and Sander, C. (1983) Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22, 2577-2637 41. Zhong, Q., Chen, Y., Jones, D., and Lee, W. H. (1998) Perturbation of TSG101 protein affects cell cycle progression. Cancer research 58, 2699-2702 42. Goila-Gaur, R., Demirov, D. G., Orenstein, J. M., Ono, A., and Freed, E. O. (2003) Defects in human immunodeficiency virus budding and endosomal sorting induced by TSG101 overexpression. Journal of virology 77, 6507-6519 43. Schindelin, J., Arganda-Carreras, I., Frise, E., Kaynig, V., Longair, M., Pietzsch, T., Preibisch, S., Rueden, C., Saalfeld, S., Schmid, B., Tinevez, J. Y., White, D. J., Hartenstein, V., Eliceiri, K., Tomancak, P., and Cardona, A. (2012) Fiji: an open-source platform for biological-image analysis. Nature methods 9, 676-682 44. Schindelin, J., Rueden, C. T., Hiner, M. C., and Eliceiri, K. W. (2015) The ImageJ ecosystem: An open platform for biomedical image analysis. Mol Reprod Dev 82, 518-529 45. Meijsing, S. H., Pufall, M. A., So, A. Y., Bates, D. L., Chen, L., and Yamamoto, K. R. (2009) DNA binding site sequence directs glucocorticoid receptor structure and activity. Science 324, 407-410 46. Wallberg, A. E., Neely, K. E., Gustafsson, J. A., Workman, J. L., Wright, A. P., and Grant, P. A. (1999) Histone acetyltransferase complexes can mediate transcriptional activation by the major glucocorticoid receptor activation domain. Mol Cell Biol 19, 5952-5959 47. Almlof, T., Wallberg, A. E., Gustafsson, J. A., and Wright, A. P. (1998) Role of important hydrophobic amino acids in the interaction between the glucocorticoid receptor tau 1-core activation domain and target factors. Biochemistry 37, 9586-9594 48. Lu, Q., Hope, L. W., Brasch, M., Reinhard, C., and Cohen, S. N. (2003) TSG101 interaction with HRS mediates endosomal trafficking and receptor down-regulation. Proc Natl Acad Sci U S A 100, 7626-7631 49. Kim, B. Y., Olzmann, J. A., Barsh, G. S., Chin, L. S., and Li, L. (2007) Spongiform neurodegeneration-associated E3 ligase Mahogunin ubiquitylates TSG101 and regulates endosomal trafficking. Molecular biology of the cell 18, 1129-1142 50. Garrus, J. E., von Schwedler, U. K., Pornillos, O. W., Morham, S. G., Zavitz, K. H., Wang, H. E., Wettstein, D. A., Stray, K. M., Cote, M., Rich, R. L., Myszka, D. G., and Sundquist, W. I. (2001) Tsg101 and the vacuolar protein sorting pathway are essential for HIV-1 budding. Cell 107, 55- 65 51. Lee, H. H., Elia, N., Ghirlando, R., Lippincott-Schwartz, J., and Hurley, J. H. (2008) Midbody targeting of the ESCRT machinery by a noncanonical coiled coil in CEP55. Science 322, 576-580 52. Sundquist, W. I., Schubert, H. L., Kelly, B. N., Hill, G. C., Holton, J. M., and Hill, C. P. (2004) Ubiquitin recognition by the human TSG101 protein. Mol Cell 13, 783-789 53. Katoh, K., Suzuki, H., Terasawa, Y., Mizuno, T., Yasuda, J., Shibata, H., and Maki, M. (2005) The penta-EF-hand protein ALG-2 interacts directly with the ESCRT-I component TSG101, and Ca2+-dependently co-localizes to aberrant endosomes with dominant-negative AAA ATPase SKD1/Vps4B. The Biochemical journal 391, 677-685 54. Muromoto, R., Sugiyama, K., Takachi, A., Imoto, S., Sato, N., Yamamoto, T., Oritani, K., Shimoda, K., and Matsuda, T. (2004) Physical and functional interactions between Daxx and DNA methyltransferase 1-associated protein, DMAP1. Journal of immunology 172, 2985-2993 55. Cheng, H., Liao, Y., Schaeffer, R. D., and Grishin, N. V. (2015) Manual classification strategies in the ECOD database. Proteins 83, 1238-1251

11 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

56. Dahlman-Wright, K., Baumann, H., McEwan, I. J., Almlof, T., Wright, A. P., Gustafsson, J. A., and Hard, T. (1995) Structural characterization of a minimal functional transactivation domain from the human glucocorticoid receptor. Proc Natl Acad Sci U S A 92, 1699-1703 57. Benjamini, Y., and Hochberg, Y. (1995) Controlling the False Discovery Rate - a Practical and Powerful Approach to Multiple Testing. J Roy Stat Soc B Met 57, 289-300 58. Luisi, B. F., Xu, W. X., Otwinowski, Z., Freedman, L. P., Yamamoto, K. R., and Sigler, P. B. (1991) Crystallographic analysis of the interaction of the glucocorticoid receptor with DNA. Nature 352, 497-505

12 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

Footnotes This work was supported by an NIH training grant to the Cell, Molecular, Development, and Biophysics program in the biology department at JHU (T32-GM007231), and an NIH grant to VJH (R01-GM- 126130). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

The abbreviations used are: GR, glucocorticoid receptor; NTD, N-terminal domain; DBD, DNA-binding domain; LBD, ligand-binding domain; SHR, steroid hormone receptor; TSG101, tumor susceptibility gene-101; TSG101cc, TSG101 coiled-coil.

13 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

Tables

GR TSG101cc 6FAM-GRE Kd A NTD + pyrene N/A 20.9 ± 9.4 µM (3) C3 NTD + pyrene N/A 19.6 ± 2.0 µM (3) A NTD-DBD + pyrene N/A 33.6 ± 15.2 µM (3) C3 NTD-DBD + pyrene N/A 72.6 ± 14.6 µM (3) A NTD-DBD – Pal 212 ± 12 nM (4) A NTD-DBD + Pal 100.3 ± 8.0 nM (5) A NTD-DBD – GilZ 241 ± 19 nM (2) A NTD-DBD + GilZ 183 ± 23 nM (2) A NTD-DBD – Tat 287 ± 37 nM (2) A NTD-DBD + Tat 128 ± 43 nM (2) C3 NTD-DBD – Pal 229 ± 24 nM (3) C3 NTD-DBD + Pal 94.5 ± 5.1 nM (3) C3 NTD-DBD – GilZ 238 ± 26 nM (2) C3 NTD-DBD + GilZ 183 ± 19 nM (2) C3 NTD-DBD – Tat 298 ± 43 nM (2) C3 NTD-DBD + Tat 192 ± 55 nM (2) Table 1. Globally fitted binding constants Apparent binding constants are shown, all collected in the context of our fluorescence anisotropy buffer (methods). The first four rows refer to binding of GR constructs and TSG101cc in Fig. 2 and Fig. S1, and the rest of the table refers to binding of GR NTD-DBD and DNA in Figure 5 and S7. Parentheses next to the fitted Kd's indicate the number of replicate data sets that were globally fit. Errors are ± 95% confidence intervals of the non-linear least squares fits. 6-FAM construct acronyms: Pal = palindromic consensus, GilZ = GRE from glucocorticoid induced leucine zipper, Tat = tyrosine amino-transferase derived (45). For details of GREs used, see Supporting Information and Fig. S7.

14 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

Figures

Figure 1. The domains and functions of GR and TSG101 Domains are rectangles oriented from N- to C-terminus, scaled with respect to size. A) Two of GR’s translational isoforms are indicated above the GR diagram. The A NTD isoform contains a regulatory (R) domain that alters the stability and transcriptional activity of the GR (13). The functional domain (F) binds transcriptional cofactors and is also called activation function 1 (AF1) in the literature. AF1 has been shown to bind: TSG101, Med14, TBP, CBP, Ada2, and many other cofactors (7,23,46,47). B) TSG101 domain functions are listed at the top, and the coiled-coil's nuclear involvement is highlighted. Some examples of protein binding partners are listed at the bottom, see references for a more complete listing (20-23,25,48-54).

15 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

1.0 ● C3 NTD ● C3 NTD−DBD ● ●● ● ●●●● ● 0.8 ● ● ● ● ●● ● ● ●●● ● ● ●●●● ● ● 0.6 ● ●● ● ● ●● ● ●●● ● ● ● ● ● ● ● ● ● 0.4 ●●● ● ● ● ●● ●●● ● ● ● ●●●● ●● 0.2 ● ● ● ● ● ● ● ● ● ● ● ● ● ● ● ●●●● ●●● ● ● ● ● ● ●● ●●● ●●●● ●● ● ● ● ● ● ● ● Fraction of TSG101cc Bound Fraction 0.0 ● ● ● ● ● ● ● ●

10-7 10-6 10-5 10-4 GR C3 construct (Molar)

Figure 2. Quantitative binding of TSG101cc to C3 constructs of GR Data from fluorescence anisotropy, globally fitted and converted to fraction of TSG101cc bound. Fluorescence anisotropy of pyrene labeled TSG101cc was followed while titrating either the GR C3 NTD (pale blue circles) or C3 NTD-DBD (dark blue circles). The data were fit to a single-site binding model (Kd’s: C3 NTD = 19.6 ± 1.0 µM, C3 NTD-DBD = 72.6 ± 14.6 µM). See Supporting Information for details of fitting and Figure S1 for similar data on the GR A isoform.

16 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

A B Potential fragments from C3-NTD alone +TSG101cc C3-NTD of GR (bars = R/K) Minutes Minutes kDa 0 5 10 30 0 5 10 30 40 F-region MS of full length C3-NTD 30 25 20 Protected MS of protected band 15 Fragment

10 TSG- Two hybrid by Hittelman et al. 101cc

Using a constant 1:1000 µg of trypsin to µg of protein

Figure 3. TSG101cc protects the N-terminus of GR from limited protease digestion A) Polyacrylamide gel stained with Coomassie dye. Digestion of GR C3 NTD alone (left) produces numerous fragments. Addition of TSG101cc (right) causes formation of a protected fragment. Note that far more trypsin was added to the +TSG101cc reaction to maintain a constant ratio of trypsin:total protein. B) Illustrated results of MALDI mass spectrometry of bands cut from a gel, as shown in part A, and fully trypsinized overnight. Possible fragments are shown at the top, with trypsin cut sites (Arg or Lys) as black bars. In descending order, the next two sets of boxes represent the fragments actually observed for unprotected full-length GR C3 NTD and the protected band, respectively. The box at the bottom represents the sequence used as bait in the yeast two-hybrid experiment of Hittelman et al. (23). See Figures S2-3 and Table S2 for mass spectrometry, controls, and an example of GR A NTD digestion.

17 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

Size of TSG101cc Amines *** GR Helical GR Compatibility C3-NTD 100% (323 a.a.) X-linking Connectivity

50% Highly Protease Protected X-linked Site 0% TBP binding site C (AF-1 core) N N/A N

Figure 4. Cross-linking results and model for GR C3:TSG101cc binding Representation of GR C3-NTD and TSG101cc cross-linking data. Amines within TSG101cc and GR C3 NTD are shown as blue dots (lysines or N-termini). An asterisk (*) indicates a cross-linked position on GR (raw data in Fig. S4 and Tables S3-8). The GR NTD sequence was also computationally analyzed for helical compatibility (28,35,55), represented by heat map coloring of the GR box in the center. “N/A” indicates a sequence around GR a.a. 300 that did not match any known structure and it is colored black. See Figure S6 and Supporting Excel file for the helical compatibility data. Around the GR box are three bars: the blue bar is in reference to Fig. 3, the red bar is the AF1 core of the literature (56), and the black bar represents the a.a. length of the TSG101cc. Bottom Right Inset: Model of the heterodimer, based on the following: 1) flipped cross-links, 2) helical compatibility of the GR sequence, 3) 1:1 binding of GR NTD and TSG101cc (Fig. 2 and Fig. S4), and 4) our finding that the GR AF1 core does not bind TSG101cc (Fig. S5). The TSG101 coil is in gray and the cross-linked lysines are shown as blue dots with lines going roughly to the appropriate site on GR. One cross-link was ambiguous because of adjacent lysines—indicated by a dashed circle. The GR protease protected site and TBP binding site are colored as in the main figure. The black spheres “N” and “C” indicate N and C-termini.

18 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

Figure 5. Binding of GR A/C3 NTD-DBD to GRE constructs with or without TSG101cc The raw data were fit globally to extract Kd’s, then converted to ΔΔG relative to the case without

TSG101cc. Error bars are the additive 95% confidence intervals from two fits (Kd with versus without

TSG101cc). See Table 1 for sample sizes and the fitted Kd values. See Fig. S7 for the fitted data. Pal = palindromic consensus, GilZ = GRE from glucocorticoid induced leucine zipper, Tat = tyrosine amino- transferase derived (45). See Supporting Information for the DNA sequences used.

19 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

Figure 6. Energetic frustration in GR A is revealed by TSG101cc binding A) Schematic representation of the interdomain thermodynamic coupling in the GR A NTD-DBD (left) and GR C3 NTD-DBD (right). The domains are labeled and portrayed as colored shapes connected by gray bars representing their thermodynamic coupling, either positive or negative. Binding of a cofactor (small gray box) to the GR F-domain can directly stabilize the DBD and promote binding of DNA via the F:DBD coupling; however, for GR A it will simultaneously destabilize the DBD and decrease DNA binding through its interaction with the R-domain of GR A. B) TSG101cc was titrated in multiple Pal GRE binding experiments, and each data set was fit individually, not globally. The ΔΔG of the y-axis was calculated as in Fig. 5. The lines of best fit are a simple binding equation: (ΔΔGMax × TSG101cc ÷ Kd) / (1 + [TSG101cc ÷ Kd]). The points are the average of duplicate binding datasets ± one standard deviation, except at 100 µM TSG101cc five GR A and three GR C3 isoform datasets were used. Fitted values for TSG101cc Kd with GRE bound to GR: GR A Kd = 2.0 ± 2.9 µM (7.6 ± 0.5 kcal/mol), ΔΔGMax = 360 ± 93 cal/mol; GR C3 Kd = 3.2 ± 2.3 µM (7.4 ± 0.3 kcal/mol), ΔΔGMax = 550 ± 80 cal/mol (± 95% CI). C) A simulation of Part A using the EAM with values similar to our previous work (13). Attempts to directly fit our data proved the problem too unconstrained. See Supporting Information for explanation of the calculations here.

20 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

Figure 7. TSG101 and GR co-localize in U2OS nuclei GR A/C3 NTD-DBD and TSG101 were detected by immunostaining. The expression plasmids transfected are indicated on the left of each image series. All scale bars are 25 µm. *In order to view the GR channel of cells without a GR expression plasmid, the microscope gain had to be increased about 1.4- fold. Examples shown are typical from six or more fields randomly examined (experimental groups) or three fields (blank).

21 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

GR:A C3 A C3 A C3

Figure 8. TSG101's coiled-coil is necessary for enhancement of GR-driven transcriptional activation Signal is from a reporter luciferase expressed from GRE promoter:luciferase DNA transfected into U2OS cells. For GR containing samples, a constant 6 ng of a given GR NTD-DBD plasmid was transfected. Error bars represent mean ± 95% CI of four samples. *P < 0.05, **P < 0.01, black diamond is P < 0.001. T-tests were one-sided with the alternative hypothesis being that the 0 ng group mean < experimental mean. Benjamini-Hochberg corrections were used throughout (57).

22 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

P 1.1×10-6 C

P 2.2×10-4 C B

B B

A A B A A A

GR GR GR GR GR Transfection: TSG101 TSG101 TSG101 TSG101 TSG101

GR+TSG101 GR+TSG101 GR+TSG101 GR+TSG101 GR+TSG101 Empty Plasmid Empty Plasmid Empty Plasmid Empty Plasmid Empty Plasmid 100 nM 100 nM 100 nM 100 nM 0.02% EtOH Cortisol Dex RU486 RU+Dex Figure 9. TSG101 enhances transcriptional activation driven by full-length GR A (α) Transcription of luciferase reporter from a GRE PAL-driven construct transfected into U2OS cells. Each data set was normalized to its respective empty plasmid transfection plus EtOH steroid vehicle; a higher ratio corresponds to more GR transcriptional activity. These example data were run in sextuplicate and are shown as means ± 95% confidence intervals. A Tukey HSD test was applied to each drug treatment, with at least a P < 0.01 defining different groups, labeled A, B, or C for each treatment. The actual P- values between B and C are listed in the figure. Dex = dexamethasone and RU+Dex = 100 nM of both RU486 and dexamethasone. RU486 is a partial agonist well known to partially block dexamethasone activation by binding to the steroid ligand binding site.

23 bioRxiv preprint doi: https://doi.org/10.1101/2020.02.02.931485; this version posted February 3, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-NC-ND 4.0 International license. TSG101 allosterically regulates GR

A ESCRT-I Cytosol Nucleus

Strong Ka Strong K Weak Ka TSG101 a

GR GR+GRE

B This Study

Copik et al., 2006 (TBP binds two AF1 cores) GRE TSG101cc

GR NTD TBP GR DBD dimer

GR NTD

TSG101cc

Li et al., 2017 (R region binds DBD)

DNA Looping to Promoter

Figure 10. Model: TSG101cc in the nucleus directly binds to the GR NTD to enhance transcriptional activation A model based on the results of this study is schematized in A) and the dash-lined boxed area of B). A) A cell is partially sketched to illustrate that cytosolic TSG101 binds weakly to GR and better to other binding partners. Nuclear TSG101 binds GRE-bound GR more strongly than free GR. B) In the boxed area, GR (blue) bound to its GRE (grey) is shown. Each GR NTD forms a pseudo-trimer with one TSG101cc (grey coils, pdb:3iv1). The individual GR NTDs do not interact (our data only show heterodimers of GR NTD and TSG101cc). The R region of GR (N-term of the GR A NTD) interacts with the DBD (13), and this schematic model allows for that association. The data of Fig. S6 suggests that the R-region forms a helix, hence our depiction of R here. TSG101's interaction with GR does not depend on the known TBP binding site (pink box), possibly leaving it open for interaction with TBP. On the right, TBP is shown in pink. Others demonstrated that TBP binds two AF1 core elements of GR (26). Structures: (33,58).