The UBAP1 subunit of ESCRT-I interacts with ubiquitin via a SOUBA domain

Abstract

The endosomal sorting complexes required for transport (ESCRTs) facilitate endosomal sorting of ubiquitinated cargo, MVB biogenesis, late stages of cytokinesis, and retroviral budding. Here we show that ubiquitin associated protein 1 (UBAP1), a subunit of human ESCRT-I, coassembles in a stable 1:1:1:1 complex with Vps23/TSG101, VPS28, and VPS37. The X-ray crystal structure of the C-terminal region of UBAP1 reveals a domain that we describe as a solenoid of overlapping UBAs (SOUBA). NMR analysis shows that each of the three rigidly arranged overlapping UBAs making up the SOUBA interact with ubiquitin. We demonstrate that UBAP1-containing ESCRT-I is essential for degradation of antiviral cell-surface proteins, such as tetherin (BST-2/CD317), by viral countermeasures, namely, the HIV-1 accessory protein Vpu and the Kaposi sarcoma-associated herpesvirus (KSHV) ubiquitin ligase K5.

Graphical abstract

Figure 1

UBAP1 Forms a Stable Heterotetrameric Complex with ESCRT-I Subunits TSG101, VPS28, and VPS37 In Vitro and In Vivo (A) Schematic representation of the ESCRT-I heterotetramer illustrating the putative UBAP1 arrangement relative to other ESCRT-I subunits. The top sequence alignment shows the UMA domain, also present in the C-terminal part of MVB12A and MVB12B. The bottom part shows a detailed view of the conserved sequence corresponding to the SOUBA domain in the C-terminal region of UBAP1 from various species. Hallmark UBA residues homologous to the conserved (M/L)-G-(Y/F) motif for the three successive UBA domains are identified by blue, pink, and orange triangles, respectively. Positions of the seven α helices mapped from the structure are shown above the alignment. (B) Gel filtration of recombinant ESCRT-I containing UBAP1-His6 on a HiLoad 16/60 Superdex 200 column. The inset shows a Coomassie blue-stained SDS-PAGE gel of the peak fraction containing the tetrameric ESCRT-I complex including UBAP1. See also Table S1 and Figure S1. (C) ESCRT-I protein complexes from cells stably expressing One-Strep tagged TSG101 (OSHA-TSG101) were affinity purified on a Strep-Tactin matrix and visualized by western blot. One percent of the starting cell lysate and 10% of the volume eluted from the matrix (pull-down) were analyzed by western blot with anti-TSG101, anti-VPS37A, and anti-UBAP1 antibodies. As a control for the specificity of the binding, purification from cells stably expressing an empty vector (OSHA) is also shown. (D) Deletion analysis to determine the minimal ESCRT-I binding region on UBAP1. Recombinant UBAP1, either full length (lane 4) or N-terminal fragments (1–92, lane 3; 1–77, lane 2; or 1–68, lane 1), with a His6 tag at the C terminus were coexpressed with TSG101/VPS28/VPS37A ESCRT-I components in E. coli and purified by affinity chromatography and gel filtration. A Coomassie-stained SDS-PAGE of the purified complexes is shown. (E) Coprecipitation studies to map the interaction of UBAP1 with ESCRT-I. 293T Cells were transfected with plasmids expressing GST-TSG101, Myc-VPS28, HA-VPS37A/B, and HA-UBAP1 wild-type (WT) or mutant (M1 to M4), followed by purification using glutathione-coated beads. One percent of the starting cell lysate and 10% of the volume eluted from the beads (pull-down) were analyzed by western blot with anti-HA antibody. See also Figure S1 and Table S1.

Figure 2

UBAP1 Is Essential for Degradation of Tetherin Triggered by Viral Countermeasures (A) Tetherin was immunoprecipitated from 293T cells stably expressing tetherin. The cells were uninfected (uninf), infected with HIV-1 (HIV-1 wt), or Vpu-defective HIV-1 (HIV-1 delVpu) and were treated with irrelevant control siRNA (Cont) or with siRNAs against UBAP1 (UBAP1Q3 and UBAP1Q4). Western blots show the immunoprecipitated tetherin (upper panel), siRNA-mediated silencing of the endogenous UBAP1 (middle panel), and the HSP90 as a protein loading control (lower panel). (B) Tetherin was immunoprecipitated from cells infected with HIV-1 (HIV-1 wt) and transfected with a plasmid expressing HA-tagged Ubiquitin and either control (cont) or UBAP1 specific siRNA (UBAP1 Q3). Tetherin ubiquitination was visualized with anti-HA antibody (top panel, THN-Ubn). As a control, the same analysis was performed in cells infected with a Vpu-defective HIV-1 (HIV-1 delVpu) and treated with an irrelevant siRNA (Cont). Bottom panels show immunoprecipitated tetherin, UBAP1 depletion and tubulin as a loading control. (C) Infectious HIV-1 virion production was measured by inoculation of TZM-bl indicator cells and is expressed as relative luminescence units (R.L.U.). Error bars indicate the standard deviation from the mean of three independent experiments. See also Table S2.

Figure 3

UBAP1 Depletion Prevents K5-Mediated Tetherin Degradation (A) Western blots of cell lysates from HT1080 cells stably expressing HA-tetherin alone (HT1080:HA-THN) or in combination with K5 (HT1080:HA-THN/K5) and treated with siRNAs against an irrelevant control (cont), TSG101 or UBAP1. THN-HA was detected with anti-HA antibody, with tubulin as a loading control and visualized using Li-Cor fluorescently coupled 650 and 800 nm secondary antibodies. The graphs show the percentage of mature tetherin levels, normalized to tubulin loading. Error bars indicate the standard deviation from the mean of three independent experiments. Lower panels show depletion of TSG101 and UBAP1, and tubulin as a loading control. (B) Confocal immunofluorescence showing localization of tetherin, CD63, and ubiquitin in cells treated with either control siRNA-, TSG101-, or UBAP1-specific siRNA. A higher magnification of the boxed areas is shown in TSG101- and UBAP1-treated panels. In the overlay panels, DNA is shown in blue, tetherin in green, and CD63 or ubiquitin in red. See also Table S3.

Figure 4

Role of UBAP1 in ESCRT-I Mediated HIV-1 Release and Cytokinesis (A) Infectious virus release upon coexpression of an HIV-1 provirus with an irrelevant siRNA (Cont), siRNA against TSG101 or two different siRNAs against UBAP1. Western blots show intracellular TSG101, UBAP1, and HSP90 (protein loading control) as well as intracellular (cell lysates) and virion-associated (virions) HIV Gag protein. (B) Quantification of cells with multiple nuclei after treatment with the irrelevant control siRNA (Cont), siRNA against hIST1 (positive control), or with either of the two different siRNAs against UBAP1. As for (A), western blots show siRNA-mediated silencing of the endogenous hIST1 and UBAP1 and HSP90 as a protein loading control. Error bars indicate the standard deviation from the mean of three independent experiments.

Figure 5

Ubiquitin Binding to UBAP1 Binding of human UBAP1 SOUBA domain (389–502) or UBAP1-containing ESCRT-I complex consisting of full-length UBAP1, VPS28 (full-length), TSG101 (198–390, i.e., lacking the UEV domain) and VPS37A (229–397, i.e., lacking the UEV domain). The experimental data (squares) were fit using a binding model for a single set of independent sites. The mean Kd and standard deviation were calculated from three independent measurements. (A) ITC titration of monoubiquitin into a cell containing the UBAP1 SOUBA domain (389–502), with the thermogram shown on top and integrated peaks shown on bottom. (B) ITC titration of monoubiquitin into a cell containing ESCRT-I with full-length UBAP1: thermogram (top) and integrated peaks (bottom). (C) ITC titration of the SOUBA domain into a cell containing K63 diubiquitin: thermogram (top) and integrated peaks (bottom). (D) ITC titration of the SOUBA domain into a cell containing K48 diubiquitin: thermogram (top) and integrated peaks (bottom).

Figure 6

Overlapping UBAs Build a Compact SOUBA Domain (A) Overview of UBAP1 SOUBA domain (389–502), showing the tandem arrangement of the three overlapping UBA domains rainbow colored from N terminus (blue) to C terminus (red). (B) Sequence alignment of each of the three UBA domains of UBAP1 with three other well-characterized UBA domains. UBA signature motif residues are shown in red and represented as sticks in (A). Residues mutated in the SOUBA crystal structure are shown with a blue background. (C) Structural alignment of each of the three UBAP1 UBA domains with the three previously reported UBA domains present in the sequence alignment (B), namely Ubiquilin1 (2JY5) in light gray, DSK2 UBA (2BWB) in medium gray, and HHR23A UBA2 (1DV0) in dark gray. See also Table 1.

Figure 7

UBAP1 Interactions with Monoubiquitin Detected by NMR Chemical Shift Perturbations (A) Overlay of HSQC spectra of ¹⁵N-labeled monoubiquitin in the absence (black) and presence (red) of 2.5 molar equivalents of unlabeled UBAP1 SOUBA. Amide resonances with significant chemical shift changes have been annotated. (B) Plot of chemical shift perturbations (CSPs) for ¹⁵N-labeled monoubiquitin as a function of residue number. Proline residues (for which no data are available) have been assigned a value of 0 ppm and are marked by an asterisk. (C) CSPs for monoubiquitin mapped onto its structure. (D) Overlay of HSQC spectra of ¹⁵N-labeled SOUBA in the absence (black) and presence (red) of three molar equivalents of unlabeled monoubiquitin. (E) Plot of CSPs for ¹⁵N-labeled SOUBA as a function of residue number. (F) CSPs for SOUBA mapped onto its structure. (G–J) Paramagnetic relaxation enhancement effects in the SOUBA/Ub complex induced by the spin label attached to the K6C (G) and K48C (I) mutant ubiquitins. The graphs plot experimental PREs observed in SOUBA upon addition of the SL-ubiquitin against residue number. Significant paramagnetic effects are illustrated in green on a model of SOUBA bound to three molecules of ubiquitin for K6C-SL (H) and K48C-SL (J). The side chains of K6 and K48 (mutated to a Cys for spin labeling) are shown as salmon sticks.

Figure 8

Model of Monoubiquitin Binding to the UBAP1 SOUBA The UBA1, UBA2, and UBA3 are colored blue, yellow, and red, respectively. UBAP1 residues showing a CSP >0.2 ppm upon binding to monoubiquitin are shown as green sticks. Three monoubiquitins have been docked on the basis of superposition of each UBA domain with the ubiquilin UBA domain in complex with ubiquitin (PDB ID 2JY6). Ubiquitin residues with CSPs >0.2 ppm upon binding to the UBAP1 SOUBA are shown as magenta sticks. The PRE and CSP measurements show that all three UBAs bind ubiquitin. The gradual addition of substoichiometric amounts of ubiquitin causes chemical shifts for residues in all three UBA motifs, suggesting that there is no preference for the binding of ubiquitin to one site versus another. The NMR data indicated that binding occurs at each of the sites, but cannot distinguish whether one, two, or three molecules are bound at the same time. Nevertheless, there appear to be no steric clashes in the model that would prevent all three ubiquitins binding simultaneously.