The C-Terminal Portion of the Hrs Protein Interacts with Tsg101 and Interferes with Human Immunodeficiency Virus Type 1 Gag Particle Production

Abstract

The human immunodeficiency virus type 1 (HIV-1) Gag protein recruits Tsg101 to facilitate HIV-1 particle budding and release. In uninfected cells, the Hrs protein recruits the ESCRT-I complex to the endosome, also through an interaction with Tsg101, to promote the sorting of host proteins into endosomal vesicles and multivesicular bodies. Here, we show that the overexpression of the C-terminal fragment of Hrs (residues 391 to 777) or Hrs mutants lacking either the N-terminal FYVE domain (mutant dFYVE) or the PSAP (residues 348 to 351) motif (mutant ASAA) all efficiently inhibit HIV-1 Gag particle production. Expression of the dFYVE or ASAA mutants of Hrs had no effect on the release of Moloney murine leukemia virus. Coimmunoprecipitation analysis showed that the expression of Hrs mutant dFYVE or ASAA significantly reduced or abolished the HIV-1 Gag-Tsg101 interaction. Yeast-two hybrid assays were used to identify two new and independent Tsg101 binding sites, one in the Hrs coiled-coil domain and one in the proline/glutamic acid-rich domain. Scanning electron microscopy of HeLa cells expressing HIV-1 Gag and the Hrs ASAA mutant showed viral particles arrested in “lump-like” structures that remained attached to the cell surface. Together, these data indicate that fragments of Hrs containing the C-terminal portion of the protein can potently inhibit HIV-1 particle release by efficiently sequestering Tsg101 away from the Gag polyprotein.

Retroviral Gag polyproteins contain all the information necessary to carry out the processes of assembly, budding, and release of virus-like particles similar in size and morphology to those of infectious viral particles (54). To achieve efficient budding and exocytosis of virus from the cell, retroviral Gag polyproteins require an interaction, via their late-domain motifs, with proteins that function in the class E protein-sorting (Vps) pathway (38). The class E Vps proteins are organized into three endosomal complexes termed ESCRT-I, -II, and -III (for endosomal sorting complex required for transport) (3, 4, 24); these multiprotein complexes are believed to function normally at the surface of endosomal membranes where they are recruited by a protein called the hepatocyte growth factor-regulated tyrosine kinase substrate (Hrs) protein via its interaction with components of the ESCRT-I complex, Tsg101 and HCRP1/Vps37A (7, 9, 17, 25, 48). Hrs seems to play the role of a scaffold protein that sets the stage for a chain of multiple interactions that take place at the endosomal membrane (25, 42). These events appear to be triggered by the recognition of ubiquitin that has been conjugated to cellular cargo en route to the lysosome (11, 25, 42, 44). The disassembly of these complexes and the release of ubiquitin are then catalyzed by an AAA ATPase called Vps4 (5, 6). Retroviral Gag proteins contain so-called late or L domains that recruit the ESCRT complexes by binding directly to Vps proteins or indirectly through proteins involved in the regulation of the Vps pathway. Thus, the human immunodeficiency virus type 1 (HIV-1) Gag protein contains an L domain, a PTAP motif located in the p6 domain that interacts directly with Tsg101, a member of the ESCRT-I complex (18, 52). A PPPY motif found in Rous sarcoma virus, Moloney murine leukemia virus (MoMLV), human T-cell leukemia virus type 1, and Mason-Pfizer monkey virus Gag proteins binds Hect ubiquitin ligases of the Nedd4 family of proteins (13, 14, 19, 22, 26, 55); these Nedd4-like Hect E3 ligases probably bridge Gag to other Vps components (33). An LXXL/YPDL motif in the p9 protein of equine infectious anemia virus and the p12 protein of MoMLV recruits AIP-1/Alix (33, 34, 43, 47, 53), which binds components of both ESCRT-I and -III complexes (34, 47, 53). Interference with Vps pathway function using dominant-acting fragments of several ESCRT factors or Vps4 has been shown to disrupt retroviral budding and release (18, 34, 47, 50, 53). Collectively, these findings indicate that retroviruses require access to multiple proteins naturally involved in the multivesicular body sorting pathway to efficiently bud and release viral particle from the cell.

The Hrs protein contains several well-defined domains: an N-terminal VHS domain whose function is yet to be determined; an FYVE domain that binds phosphatidyl-inositol-3,4,5-triphosphate (PIP3) lipids at endosomal membranes (27, 41, 46, 51); a ubiquitin interaction motif (UIM); a well-conserved proline-rich motif, the PSAP motif (residues 348 to 351); a coiled-coil (CC) domain; a proline/glutamic acid (P/Q)-rich domain; and a clathrin binding (CB) domain. In mammalian cells, Hrs binds Eps15 and STAM proteins (2, 8, 10, 49) and is found within a large complex with an approximate molecular weight of 500 kDa (12). Hrs is believed to recognize ubiquitinated cargo through the UIM (37, 40, 44) and recruits the ESCRT-I complex to the surface of the early endosome through an interaction with Tsg101 (25, 39). This initiates the sorting of cargo proteins and is subsequently followed by the recruitment of ESCRT-II and ESCRT-III complexes in a series of events that ultimately results in the invagination and internalization of cargo proteins into multivesicular bodies (40, 42). Hrs thus appears to be a master coordinator of endosomal protein sorting (42).

Because Gag and Hrs proteins both carry a PT/SAP motif and bind Tsg101, it has been proposed that HIV-1 Gag mimics the Hrs protein to bind Tsg101 and recruit the ESCRT machinery to facilitate virion particle release (39). We overexpressed several mutants and fragments of Hrs to compete with HIV-1 Gag binding to Tsg101. The data obtained here showed that Hrs fragments and mutants can indeed exert a potent inhibition of particle release by interfering with the Gag interaction with Tsg101. Fine mapping of the Tsg101 interaction with Hrs showed, surprisingly, that both proteins carry multiple and independent binding sites for one another. The expression of an Hrs mutant carrying a mutation in PSAP (substitution to ASAA) caused HIV-1 particles to arrest and cluster in a “lump-like” structure at the cell surface, demonstrating that modified Hrs proteins can potently block particle release.

MATERIALS AND METHODS

Cells and transfection.

Human 293T cells were maintained in Dulbecco modified Eagle medium containing 10% fetal bovine serum at 37°C and 5% CO₂. Transfections were performed (2 × 10⁶ cells in a 10-cm dish) with either the calcium phosphate precipitation method or Fugene 6 transfection reagent (Roche) according to the manufacturer's recommendations.

Yeast two-hybrid and mammalian expression plasmids.

The human Hrs gene used in this study was obtained from Asao Hironobu (2). The Hrs gene was cloned into plasmid pKU-Hrs in frame with a hemagglutinin (HA) tag sequence; pKU-Hrs and pKU-Hrs dFYVE that contain the full-length Hrs gene and Hrs gene lacking the FYVE domain, respectively, were described previously (2, 36). The pKU-Hrs ASAA mutant was obtained by a replacement of the proline residues in the PSAP motif (residues 348 to 351) with alanines using the QuikChange site-directed mutagenesis kit (Stratagene). The full-length Hrs gene, a fragment carrying residues 1 to 502 (N-term), and a fragment carrying residues 391 to the stop codon (C-term) were also cloned in frame with a Flag tag sequence as described previously (21). The deletion mutants HrsΔCC and HrsΔP/Q were also described previously (21), and they carry deletions of regions in Hrs containing residues 430 to 500 and 499 to 604, respectively. MoMLV was expressed from plasmid pNCS (15). The HIV-1 Gag-Pol-expressing plasmid pgp-RRE-r and Rev-expressing plasmid pCMV-rev were previously described (45). HIV-1 Gag-green fluorescent protein (GFP) was a gift from Marilin Resh (23). The Tsg101 Myc-tagged protein was expressed using plasmid pLLD as described previously (29) (a gift from Limin Li and Stanley Cohen, Stanford University, CA).

Yeast two-hybrid plasmids used in this study were pGADNOT and pSH2-1 (20, 32). Hrs fragments containing residues 1 to 166, 1 to 215, 1 to 290, 1 to 390, 390 to 511, and 511 to 777 were amplified by PCR using the human Hrs gene (2), the full-length wild-type Hrs gene, an Hrs gene lacking the FYVE domain, and a gene carrying a replacement of the PSAP motif with the ASAA motif, and all fragments and mutated genes were cloned into plasmid pGADNOT between the NotI and SalI sites. Similarly, the entire Tsg101 gene, the ubiquitin enzyme variant (UEV) region of Tsg101 (residues 1 to 256), a fragment containing N-terminal residues 1 to 271 named N-Tsg101, a Tsg101 fragment carrying C-terminal residues 256 to 391 named C-Tsg101, and a fragment containing residues 256 to 311 called CC-Tsg101 were all cloned into pGADNOT. Tsg101 and Hrs genes were both cloned into plasmid pSH2-1 between the BamHI and SalI sites; N-Tsg101 and C-Tsg101 were also inserted into plasmid pSH2-1 between the BamHI and SalI sites. The entire Hrs gene as well as the Hrs ASAA mutant were also cloned into pSH2-1. Full-length Tsg101 and N-Tsg101 and C-Tsg101 fragments were also cloned in frame with the 5′ end of the LexA gene in plasmid pN-LexA (Origne) between EcoRI and BamHI; the entire Hrs gene fragments containing residues 1 to 290, 1 to 391, and 1 to 511 were inserted into pNLexA between NotI and BamHI. The AIP-1 gene (47) was cloned into plasmids pSH2-1 and pGADNOT and used as controls. The pSH2-1MLVGag, pMLVGADNOT, pSH2-1 HIV-1 Gag, and pHIV-1 Gag GADNOT constructs were previously described (1, 30).

Two-hybrid assay.

Protein-protein interactions in Saccharomyces cerevisiae were performed with plasmids pGADNOT and pSH2-1 in strains CTY10-5d and GGY::171 (generously provided by R. Sternglanz, SUNY, Stony Brook, NY). Plasmid pSH2-1 encodes an N-terminal LexA DNA domain (LexADB), and pGADNOT encodes an N-terminal Gal4 activation domain (Gal4AD). The assay was performed as described previously (32).

Correlative SEM.

HeLa cells were plated onto coverslips with grids (MatTek Corporation) and transfected with a mixture of two plasmids, pRRE-r DNA, the HIV Gag-Pol expression plasmid (45), and the HIV Gag-GFP plasmids (23) at a ratio of 2:1. In this experiment, an excess of untagged HIV Gag-Pol DNA was used to rescue the Gag-GFP budding defect (28). Cells were fixed 12 to 20 h posttransfection with 4% paraformaldehyde in phosphate-buffered saline (PBS) for 15 min. After fluorescence images were captured and the locations of individual cells on the grid were recorded, cells were fixed with 2.5% glutaraldehyde in 0.2 M sodium phosphate buffer (pH 7.2), postfixed with 2% osmium tetroxide (Electron Microscopy Sciences), dehydrated through a graded series of ethanol washes, critical-point dried, and sputter coated with gold-palladium. The same individual cells were located and visualized by scanning electron microscopy (SEM) by using a model LEO 1550 field emission scanning electron microscope at 3 kV. In most cases, multiple digital images were captured at the same magnification and pasted together using Adobe Photoshop.

Western blots and antibodies.

Cells were washed with PBS and lysed on ice in 500 μl of lysis buffer (50 mM Tris [pH 8], 150 mM NaCl, 1% NP-40, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulfate [SDS]) containing Complete Mini EDTA-free protease inhibitors (Roche). Lysates were clarified by centrifugation for 30 min (13,000 rpm at 4°C). Twenty percent of the total lysate (100 μl) was mixed with 25 μl 5× SDS sample buffer, heated to 100°C for 5 min, and separated on SDS-12% polyacrylamide gels. Virions and virus-like particles were collected from the culture supernatant, which was filtered through a 0.45-μm filter, by ultracentrifugation for 2 h (25,000 rpm at 4°C) through a 20% sucrose cushion. Pelleted virions were resuspended in 50 μl SDS sample buffer and analyzed on SDS-12% polyacrylamide gels. Proteins were transferred onto an Immobilon-P membrane (Millipore) according to the manufacturer's instructions. For Western blot analysis, membranes were blocked with 5% nonfat skim milk in 1× Tris-buffered saline-Tween (TBST) for 1 h at room temperature and incubated with the appropriate primary antibody for 1 h and a relevant peroxidase-conjugated secondary antibody for 45 min. Peroxidase activity was detected by reacting the membrane with ECL Plus (Amersham Biosciences), followed by exposure and development of autoradiographic film. The monoclonal anti-HIV-1 capsid antibody was obtained from NEN and used at a 1:1,000 dilution. Polyclonal rabbit anti-HIV-1 capsid antibody was obtained from Carol Carter, SUNY, Stony Brook, NY, and used at a 1:1,000 dilution. Polyclonal goat anti-murine leukemia virus capsid antibody (NCI 79S-804) was used at a 1:5,000 dilution. The monoclonal antibody anti-Na⁺/K⁺ ATPase (Biomol) or anti-HA (Roche) or the polyclonal anti-Flag antibody (Sigma) was used at a dilution of 1:1,000. Horseradish peroxidase-conjugated polyclonal anti-mouse, anti-rabbit (Amersham Biosciences), and anti-goat (Roche) antibodies were used at a dilution of 1:10,000. Primary antibodies were maintained in a solution of 2% bovine serum albumin and 0.02% sodium azide in TBST; secondary antibodies were diluted into 5% milk in TBST.

For the complementation assay with wild-type Tsg101 and Hrs proteins, 293T cells were transfected with HIV-1 Gag-Pol-expressing plasmid pRRE-r (∼10 μg) and pCMV-rev (∼ 2 μg) alone, with increasing amounts of dFYVE Hrs mutant DNA (3 and 6 μg) or ASAA Hrs mutant DNA (3 and 6 μg); in parallel, a set of DNA samples that are identical to those described above was prepared, and increasing amounts of either wild-type Hrs-HA DNA (pKUHrs construct) (3 and 6 μg) or Myc-tagged Tsg101 (pLLD construct) (29) (2 and 4 μg) were transfected with the DNA mixes. Cells and particles were harvested 48 h posttransfection, and their protein content was analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and Western blotting using an anti-HIV CA antibody (NEN). The expression of Hrs mutants was detected using a mouse monoclonal anti-HA antibody; the expression of Hrs could not be assessed because the signal of the full-length Hrs protein overlapped with that of the ASAA-HA Hrs mutant. The expression of Tsg101 was detected using an anti-Myc mouse monoclonal antibody. Yeast was lysed in extraction buffer (25 mM Tris [pH 7.5], 150 mM NaCl or KCl, 10% glycerol, 1 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 7 mM beta-mercaptoethanol) with 0.2-mm glass beads (Sigma). Gal4AD fusion proteins were detected using a mouse monoclonal anti-Gal4AD antibody (Santa Cruz).

Coimmunoprecipitation assay.

293T cells were transfected with HIV-1 Gag-Pol-expressing plasmids pRRE-r (∼10 μg) and pCMV-rev (∼ 2 μg), with HIV-Gag-Pol-expressing plasmids and Tsg101-Myc-expressing plasmid pLLD (∼4 μg), with Tsg101-Myc-expressing plasmid pLLD (∼ 4 μg), and with either wild-type Hrs-HA-expressing plasmid pKU-Hrs (∼5 μg), the dFYVE Hrs-HA mutant construct (∼ 5 μg), or the ASAA-Hrs-HA construct (∼ 5 μg). Cells were harvested 48 h after transfection, washed with cold PBS buffer, and lysed with a cold solution containing 150 mM NaCl, 1% NP-40, 50 mM Tris (pH 8), and a cocktail of proteasome inhibitors (Roche).

Cells lysates were cleared by high-speed centrifugation and then incubated overnight with anti-HIV-1 p6 rabbit polyclonal antibody that was captured by agarose beads coated with protein A. The beads carrying the protein complexes were washed six times with the lysis buffer and boiled with 1× sample buffer to dissociate the proteins complexes. The content of the complexes was analyzed by SDS-PAGE and Western blot analysis by first using an anti-Myc mouse monoclonal antibody to search for Tsg01-Myc-tagged proteins. The membrane was subsequently probed with both the anti-Myc mouse monoclonal antibody and a mouse monoclonal anti-HIV-1 CA antibody (NEN) to detect the HIV-1 Gag protein. Cell lysates were also analyzed by SDS-PAGE and Western blotting for the expression of wild-type Hrs-HA and the mutant proteins using an anti-HA mouse monoclonal antibody; the expression of Tsg101-Myc-tagged proteins was detected using a mouse monoclonal anti-Myc antibody. Gag particle release was also sought in the culture medium by concentrating the virus-like particles through a 20% sucrose cushion, and the protein content of the pellet was analyzed by SDS-PAGE and Western blotting using mouse monoclonal anti-CA antibodies (NEN) or anti-HIV-1 CA 183-H12-5C antibody (NIH AIDS Reagent Program).

RESULTS

Fragments and mutants of Hrs interfere with HIV-1 Gag particle production.

In HIV-1-infected cells, HIV-1 Gag competes Tsg101 away from its natural partner, Hrs, to gain access to the ESCRT machinery. We hypothesized that the overexpression of Hrs proteins could interfere with the Tsg101-HIV-1 Gag interaction. To test this hypothesis, we expressed various fragments and mutants of Hrs (Fig. 1A) and tested them for their ability to interfere with HIV-1 particle production. Constructs expressing epitope-tagged versions of the full-length Hrs gene, an N-terminal fragment containing residues 1 to 502 (N-term), a C-terminal fragment containing residues 391 to 777 (C-term), full-length Hrs carrying the PSAP (residues 348 to 351)-to-ASAA replacement (ASAA mutant), or an Hrs gene with a deletion of the FYVE domain (dFYVE mutant) was used in these experiments. The various forms of Hrs were coexpressed with an HIV-1 Gag-Pol expression construct in 293T cells, and the accumulation of Gag polyprotein in the cytoplasm and its release from the cell were examined (Fig. 1B to E). When HIV-1 Gag-Pol was expressed alone (Fig. 1B, lane 2), p24CA was readily detected in the culture medium, indicating that Gag was properly expressed, released, and processed; similar amounts of CA were released when coexpressed with increasing amounts of the N-terminal fragment of Hrs (Fig. 1B, lanes 5 and 6). In contrast, the overexpression of increasing amounts of the full-length Hrs protein (Fig. 1B, lanes 3 and 4), the C-terminal fragment (Fig. 1B, lanes 7 and 8), dFYVE (Fig. 1C, lanes 3 to 6), or ASAA (Fig. 1C, lanes 7 to 10) resulted in significant and dose-dependent decreases in the release of virion-associated CA protein into the culture medium. In all cases, proteins of the expected sizes were detected with antibodies directed to the epitope tags present on the Hrs constructs (Fig. 1F and G). Although the C-terminal fragment was expressed at lower levels, it showed a strong inhibitory effect on HIV-1 Gag-Pol particle production. Gag precursor protein expression and accumulation in the cytoplasm were largely unaffected (Fig. 1D and E). Other intracellular proteins such as the plasma membrane Na⁺/K⁺ ATPase remained unaffected, as shown by the constant amount of its alpha subunit (∼100 kDa) (Fig. 1H and I). To examine whether Hrs or Hrs mutants caused the characteristic Gag processing defects associated with late-domain phenotypes, the wild type, Hrs fragments, and mutants were expressed with HIV-1 Gag-Pol, and Gag accumulation, processing, and release were examined. In agreement with the data shown in Fig. 1B and C, the expression of Hrs and Hrs mutants had no effect on the accumulation of Gag in the cell (even when lower amounts of Gag were examined) (data not shown). However, the processing of Gag was severely affected when Hrs, the C-terminal fragment, the dFYVE mutant, or the ASAA mutant was coexpressed with Gag (Fig. 1J, lanes 3 and 4 and lanes 6 and 7). This defect was more pronounced than the one caused by the expression of the N-terminal Tsg101 fragment (Fig. 1J, lane 8), a fragment of Tsg101 that causes a characteristic L domain defect. Collectively, the data suggest that the overexpression of wild-type Hrs protein, C-term, and mutants of Hrs containing an intact C-terminal region can potently interfere with HIV-1 Gag particle production.

FIG. 1.

Hrs and Hrs mutants inhibit HIV Gag release. (A) Schematic representation of the Hrs protein and the Hrs mutants and fragments used in this study. Full-length Hrs contains several domains (from the N- to the C-terminal end): VHS, FYVE, UIM, a PSAP motif, and the CC, P/Q-rich, and CB domains. (B and C) 293 T cells expressing HIV-1 Gag-Pol alone or with wild-type Hrs, N-term or C-term of Hrs, or the Hrs dFYVE or ASAA mutant were harvested 48 h after transfection, and Gag particles released into the culture medium were collected through a 20% sucrose cushion and analyzed by SDS-PAGE and Western blotting using a mouse monoclonal anti-HIV CA antibody. No DN, no “dominant-negative” DNA. (D and E) HIV Gag expression was detected by immunoprecipitation using a rabbit anti-p6 antibody followed by Western blot analysis using a mouse monoclonal anti-HIV CA antibody. (F and G) Expression of Flag-tagged full-length Hrs, N-term and C-term fragments, and HA-tagged Hrs dFYVE and ASAA mutants was detected using either a mouse monoclonal anti-Flag antibody or a mouse monoclonal anti-HA antibody. (H and I) The Na⁺/K⁺ ATPase, a cellular membrane protein located at the plasma membrane, was used as a loading control in this experiment. (J and K) Hrs and Hrs mutants inhibited Gag processing and release. wt, wild type.

MoMLV release is resistant to Hrs.

To test whether Hrs and Hrs mutants affect production of other retroviral particles, we tested the effect of full-length Hrs and Hrs mutants on MoMLV particle production. The overexpression of full-length Hrs (Fig. 2C, lanes 3 to 5), N-term (Fig. 2C, lanes 6 to 8), dFYVE (Fig. 2C, lanes 12 to 14), or ASAA (Fig. 2C, lanes 15 to 17) had no or very little discernible effect on MoMLV particle production (Fig. 2A) or on Gag expression and accumulation in the cytoplasm (Fig 2B). The full-length Hrs protein and the C-terminal fragment had only a very small effect on MoMLV Gag intracellular accumulation (Fig. 2B, lanes 9 to 11) and a small effect on virion release (Fig. 2A, lanes 9 to 11) at the highest levels of expression. Moreover, Hrs and Hrs mutants had no effect on the virion release of an MoMLV chimera, MoMLV-PTAP (56), in which the L domain PPPY motif located in the p12 protein was replaced with 12 residues of the HIV-1 p6 protein containing the L domain motif PTAP (Fig. 2E and F and data not shown). This chimeric Gag protein binds Tsg101 and depends on Tsg101 for virion release because of its HIV-1 PTAP motif (35); nevertheless, this virus is not inhibited by the Hrs constructs. The data suggest that Hrs and Hrs mutants had no significant effect on MoMLV virion production and release and imply that their strong effect on HIV-1 particle production is specific and does not reflect a general toxicity toward the host.

FIG. 2.

Hrs and Hrs mutants have no effect on MoMLV release. (A) 293T cells were transfected with pNCS plasmid alone to generate MoMLV particles (no dominant-negative DNA [No DN]) or with pNCS and plasmids expressing Hrs, N-term or C-term fragments, or dFYVE and ASAA Hrs mutants. Virions were collected through a 20% sucrose cushion, and their protein content was analyzed by SDS-PAGE and Western blotting using an anti-MoMLV CA (p30^CA) goat antibody. (B) Expression and accumulation of p65 MoMLV Gag was sought in the cells extracts by using a goat anti-MoMLV CA (p30^CA) antibody. (C) Expression of the dominant-negative proteins was detected using an anti-HA antibody. (D) The Na⁺/K⁺ ATPase was used as a loading control. (E and F) The dFYVE and ASAA mutants had no effect on the release of the murine leukemia virus-PTAP chimera virus.

Hrs contains multiple Tsg101 binding sites.

To probe the mechanism of inhibition of HIV-1 Gag particle release by Hrs, we mapped the regions involved in the binding of Hrs to Tsg101. The full-length Tsg101 protein or the N-terminal half (residues 1 to 271) (N-Tsg101) or the C-terminal half (residues 256 to 391) (C-Tsg101) of Tsg101 was expressed as a fusion with the LexA domain in plasmid pSH2-1 or pNLexA. To confirm the functionality of these constructs, these same Tsg101 fragments were also fused to Gal4AD and tested for their interaction with each other in yeast. Full-length Tsg101 dimerized with itself strongly and also bound to N-Tsg101 and C-Tsg101 in all settings (Fig. 3A). These results show that all the Tsg101 constructs were expressed and able to form transcriptional activator complexes in yeast. The ability of Tsg101, N-Tsg101, or C-Tsg101 to bind HIV-1 and MoMLV Gag was also tested. As expected, Tsg101 and N-Tsg101, but not C-Tsg101, bound HIV-1 Gag; also, as expected, Tsg101 did not bind or only weakly bound MoMLV Gag.

FIG. 3.

Yeast two-hybrid interaction between Tsg101 and Hrs. (A) Full-length Tsg101 and N-Tsg101 (residues 1 to 271) or C-Tsg101 (residues 256 to 391) interact in agreement with results reported previously (18, 35, 52). ND, not determined. (B) Hrs fragments and mutants indicated on the left side were cloned in fusion with Gal4-activating domain protein, and Tsg101, the N-terminal fragment (1 to 271 amino acids), or the C-terminal region (251 to 391 amino acids) of Tsg101 was cloned in fusion with the LexA DNA binding domain. (C) Two Hrs fragments (residues 391 to 777 or 511 to 777) were cloned in fusion with the LexA binding domain, and their interaction with two fragments of Tsg101 (residues 1 to 256 or 251 to 311) fused to the Gal4 activation domain was tested. The plus (+) symbol indicates the detection of interactions as shown by β-galactosidase assays (blue colonies), and the minus (−) symbol indicates its absence (white colonies). The number of symbols (+) indicates the strength of the interaction between the two proteins tested.

Various fragments of Hrs were fused to Gal4AD in a yeast expression plasmid (Fig. 3B), and the strength of the interaction with the LexA-Tsg101, LexA-N-Tsg101, or LexA-C-Tsg101 construct was tested in yeast (Fig. 3B). As previously observed (31, 39), full-length Hrs bound Tsg101 strongly. N-terminal fragments of Hrs carrying VHS alone, both the VHS and the FYVE domains, or the VHS-FYVE domains and the UIM all showed no binding to Tsg101. Even when an Hrs construct was extended to include the region downstream of the UIM and PSAP motifs (residues 1 to 390), no interaction with Tsg101 was observed. The binding of Hrs fragments to Tsg101 was observed only when an Hrs fragment including the CC domain was used (residues 1 to 509); the isolated CC domain (residues 391 to 511) retained a partial ability to bind Tsg101. The P/Q-rich CB domains (residues 511 to 777) of Hrs contained the major binding activity and interacted with Tsg101 as well as full-length Hrs. A deletion of the N-terminal FYVE domain or a mutation of the PSAP motif in the full-length Hrs had almost no effect on the Hrs-Tsg101 interaction. These results indicate that the P/Q-rich CB domain interacts more strongly with Tsg101 than was suggested by previous studies (39).

Tests of the interaction with Tsg101 fragments allowed further mapping of binding sites. The C-Tsg101 fragment showed the same binding abilities for Hrs, Hrs fragments, and dFYVE and ASAA mutants as the full-length Tsg101 protein. This C-Tsg101 fusion protein showed much stronger binding ability for the CC domain of Hrs than did the full-length Tsg101 protein. N-Tsg101 interacted with wild-type Hrs and the P/Q-rich CB domain and retained the ability to bind Hrs mutants ASAA and dFYVE. In contrast, N-Tsg101 lost the ability to bind the CC domain of Hrs. These results suggest that the Hrs CC domain interacts with C-Tsg101 and that the Hrs P/Q-rich CB domain interacts with C-Tsg101 and also with N-Tsg101.

To further investigate the ability of the Hrs C terminus to bind Tsg101, two Hrs fragments containing residues 391 to 777 or 511 to 777 were fused to LexABD, and their interactions with Tsg101 and two fragments of Tsg101 were tested. The data (Fig. 3C) confirmed that the Hrs C terminus binds Tsg101 and can bind to two regions of Tsg101: the N-terminal UEV region (residues 1 to 256) and the CC domain (residues 251 to 311). Thus, Tsg101 contains two independent binding sites for the Hrs C terminus.

The PSGP and PSMP motifs located in the P/Q-rich domain of Hrs bind Tsg101.

The observation that the UEV domain of Tsg101 binds to the C terminus of Hrs suggests that proline-rich motifs located in the C-terminal end of Hrs might serve as binding sites; the UEV domain is known to bind the PT/SAP motif in HIV-1 Gag. We searched for similar PxxP motifs in the P/Q-rich domain of Hrs and identified two motifs: a PSGP motif (residues 583 to 586) and a PSMP motif (residues 620 to 623) (Fig. 4). These PSAP-like motifs have been suggested to be involved in binding the Tsg101 UEV domain (39). To test this possibility, the proline residues in these motifs were replaced with alanine residues, and the effect of these mutations on the binding of full-length Hrs or the isolated P/Q-rich domain to Tsg101 was assessed using yeast two-hybrid assays (Fig. 4). Mutations of PSGP and/or PSMP motifs to ASGA or ASMA in Hrs had no effect on Hrs binding to Tsg101. It is possible that the retention of the PSAP motif in the center of Hrs was providing a binding site for the Tsg101 UEV domain and thus suppressing the effects of the C-terminal motifs. To test this notion, the C-terminal mutations were tested in the context of a PSAP mutation (to ASAA). Even here, strong binding was observed, and there was no effect of mutations of the C-terminal PSGP and/or PSMP motif, implying that the CC region of Hrs was providing an interaction with Tsg101. To completely isolate the C-terminal Hrs interaction, the separate C-terminal P/Q-rich domain of Hrs was tested for Tsg101 binding. While the wild-type C terminus of Hrs bound Tsg101 strongly, replacements of either or both PSGP and PSMP motifs eliminated binding of the isolated domain to Tsg101. Importantly, mutations in either or both PSGP and PSMP motifs had no detectable effect on the expression of the Gal4AD fusion proteins in yeast (Fig. 4B, lanes 2 to 4). This result indicates that both the PSGP and PSMP motifs in the P/Q-rich domain in Hrs are required for the binding of the isolated C terminus to Tsg101.

FIG. 4.

PSXP motifs in the P/Q-rich domain bind Tsg101. (A) Hrs fragments and mutants indicated on the left side were cloned in fusion with Gal4-activating domain protein, and Tsg10l was cloned in fusion with the LexA DNA binding domain. The positions of the PSGP and the PSMP motifs are indicated with small rectangles; the mutant motifs are underlined. The plus (+) symbol indicates the detection of an interaction as shown by β-galactosidase assay (blue colonies), and the minus (−) symbol indicates its absence (white colonies). The number of symbols (+) indicates the strength of the interaction between the two proteins tested. (B) Expression of wild-type (Wt) Pro/Gln or the ASGA, ASMA, or ASGA/ASMA mutant in yeast was analyzed by Western blotting using anti-Gal4AD (α-Gal4AD) antibody.

An Hrs mutant lacking the P/Q-rich domain fails to inhibit HIV-1 particle release.

The overexpression of Hrs inhibited HIV-1 particle production (Fig. 1B); to test whether this inhibition required the C-terminal P/Q-rich domain, we expressed the HIV-1 Gag-Pol region with wild-type Hrs, Hrs lacking the CC domain, or Hrs lacking the P/Q-rich domain (Fig. 5C) and tested for the accumulation of Gag in the cytoplasm and the release of virus-like particles. Wild-type Hrs potently inhibited HIV-1 particle production, and the mutant lacking the CC domain showed equally potent inhibition (Fig. 5A, lanes 2 and 3). In contrast, the deletion of the P/Q-rich domain of Hrs completely abrogated the inhibitory effect of Hrs on HIV-1 particle production (Fig. 5A, lane 4). None of the Hrs constructs (Fig. 5C, lanes 3 to 5) had any detectable effect on the expression of Gag in the cytoplasm (Fig. 5B). This result indicates that the P/Q-rich domain of Hrs plays a key role in the Hrs inhibitory effect on HIV-1 particle production.

FIG. 5.

An Hrs mutant lacking the P/Q-rich motif failed to inhibit HIV VLP release. 293 T cells were transfected with HIV-1 Gag-Pol alone (lane 1) or HIV-1 Gag-Pol with either full-length Hrs (lane 2), an Hrs mutant lacking the CC domain (Hrs DelCC) (lane 3), or an Hrs mutant Hrs lacking the P/Q-rich domain (Hrs delP-Q) (lane 4). (A) HIV-1 particles (VLPs) released in the culture medium were collected through a 20% sucrose cushion. (B and C) Cells were harvested 48 h posttransfection, and the protein contents of both VLPs and intracellular extracts were analyzed by SDS-PAGE and Western blotting using a mouse monoclonal anti-CA (α-CA) (B) or anti-Flag (C) antibody. No DN, no dominant-negative DNA.

Hrs mutants prevent HIV-1 Gag binding to Tsg101.

To understand the basis for the inhibition of virus production by Hrs, the intracellular interaction between HIV-1 Gag and Tsg101 in the presence or absence of Hrs and Hrs mutants was examined. HIV-1 Gag-Pol was expressed in 293T cells either alone, with Myc epitope-tagged Tsg101 protein (Tsg101-Myc), with Tsg101-Myc and an HA epitope-tagged Hrs (Hrs-HA), or with Tsg101-Myc and various other Hrs constructs (Fig. 6). Cells were lysed, and p55^Gag was immunoprecipitated using an anti-p6 antibody. The protein complexes were collected and analyzed by SDS-PAGE and Western blotting using the indicated antibodies (Fig. 6). When HIV-1 Gag was coexpressed with the Tsg101-Myc construct, Tsg101-Myc (∼46 kDa) was readily detected as coimmunoprecipitating with Gag (Fig. 6A, lane 3); no such band was detected when cells were transfected with the empty vector (lane 1) or with the HIV-1 Gag construct alone (lane 2). High amounts of Tsg101-Myc were also coimmunoprecipitated with Gag when cells expressed HIV-1 Gag, Tsg101-Myc, and full-length Hrs-HA constructs (Fig. 6A, lane 4). In contrast, no or very little Tsg101-Myc bound to Gag protein was detected when the proteins were coexpressed with either the dFYVE Hrs-HA mutant (Fig. 6A, lane 5) or the ASAA Hrs-HA mutant (lane 6).

FIG. 6.

Hrs mutants dFYVE and ASAA interfere with Tsg101 binding to HIV-1 Gag. (A) 293 T cells expressing HIV-1 Gag-Pol alone, HIV-1 Gag-Pol and Tsg101-Myc protein, HIV-1 Gag-Pol, Tsg101-Myc protein, and wild-type HA-Hrs protein, or the HA-Hrs dFYVE mutant or HA-Hrs ASAA mutant, were lysed with NP-40 lysis buffer (see Materials and Methods); Gag was immunoprecipitated with an anti-p6 rabbit antibody (IP:αp6); the protein complexes that were immunoprecipitated were analyzed by SDS-PAGE; and the membrane was probed first with an anti-Myc mouse monoclonal antibody (αMyc). (B) The membrane was then probed with both the anti-Myc mouse monoclonal antibody and an anti-HIV CA mouse monoclonal antibody; the positions of Tsg101-Myc and HIV-1 p55Gag are indicated. (C) HIV Gag particles released in the culture medium were collected by centrifugation through a 20% sucrose cushion and analyzed by SDS-PAGE and Western blotting using an anti-HIV CA antibody. (D) The expression and cytoplasmic accumulation of HA-Hrs, the HA-Hrs dFYVE or ASAA mutant, and Tsg101-Myc proteins were detected using an anti-HA mouse monoclonal antibody and an anti-Myc mouse monoclonal antibody, respectively.

The expression of Tsg101-Myc or Hrs-HA had little effect on the accumulation of Gag in the cytoplasm; however, somewhat lower amounts of cytoplasmic Gag were detected when the Hrs mutants were expressed with HIV-1 Gag (Fig. 6A, compare lane 2 to lanes 3 to 6). HIV-1 Gag particle production was severely diminished by the expression of Tsg101-Myc (Fig. 6B, lane 3) and completely abolished when both Tsg101-Myc and Hrs (lane 4) or Hrs mutants dFYVE and ASAA (lanes 5 and 6) were expressed with Gag. More importantly, the expression of Hrs or Hrs mutants (Fig. 6C, lanes 4 to 6) had no effect on the expression and intracellular accumulation of Tsg101-Myc (Fig. 6D, lanes 2 to 6). Taken together, the data indicate that mutants of Hrs lacking the N-terminal-domain FYVE or carrying a mutated PSAP motif (ASAA) can efficiently interfere with the HIV-1 Gag-Tsg101 intracellular interaction.

Tsg101 rescues Hrs mutant-induced inhibition of HIV-1 Gag particle release.

The data described above suggest that Hrs mutants dFYVE and ASAA potently inhibit HIV-1 Gag particle production by sequestering Tsg101 away from HIV-1 Gag and thus preventing its efficient release from the cell. Based on this finding, we hypothesized that the overexpression of Tsg101 could restore the proper release of HIV-1 Gag to the inhibited cells. To test this, HIV-1 Gag was expressed alone or with either dFYVE Hrs-HA or ASAA Hrs-HA, and Gag intracellular expression and particle production were examined when either Hrs-HA or Tsg101-Myc was provided in trans. An anti-p24 antibody was used to detect HIV-1 CA in the culture medium of 293T cells (Fig. 7A, lane 2). When cells were transfected with HIV-1 Gag-Pol constructs and increasing amounts of the dFYVE Hrs-HA or ASAA Hrs-HA mutant (Fig. 7C), very little or no p24^CA protein was detected in the culture medium (Fig. 7A, lanes 3 to 6). The ability of Tsg101 or Hrs to rescue the effect of Hrs mutants on HIV-1 Gag particle production was tested by expressing increasing amounts of either Hrs-HA or Tsg101-Myc. The expression of Hrs-HA had no detectable effect on the loss of particle production resulting from the expression of the dFYVE and ASAA Hrs-HA mutants (Fig. 7A, compare lanes 3 to 6 to lanes 7 to 10). In contrast, the expression of Tsg101-Myc (Fig. 7B, lanes 10 to 14) significantly relieved the inhibition caused by the expression of the dFYVE Hrs-HA mutant (Fig. 7A, compare lane 2 to lanes 3 and 11); up to ∼30% of the particle release was recovered. The expression of Tsg101-Myc also relieved the block caused by the ASAA Hrs-HA mutant (Fig. 7A, compare lane 2 to lanes 5 and 13); here, up to ∼100% of particle production was restored. The proper processing of Gag into mature CA was also recovered (Fig. 7D, compare lanes 2, 5, and 13), indicating that the expression of Tsg101-Myc relieved the processing defect caused by the ASAA Hrs-HA mutant and restored efficient budding and release. In all cases, Gag expression and accumulation in the cytoplasm remained largely unaffected (Fig. 7D). The data indicate that providing Tsg101-Myc in trans can rescue the inhibitory effect of Hrs mutants dFYVE and ASAA on HIV-1 Gag particle production. This is consistent with the notion that Hrs mutants dFYVE and ASAA interfered with the Tsg101-Gag interaction and prevented HIV-1 Gag particle production.

FIG. 7.

Tsg101 rescues the phenotype caused by dFYVE and ASAA Hrs mutants. 293 T cells were induced to express HIV-1 Gag-Pol alone, HIV-1 Gag-Pol and increasing amounts of either dFYVE or ASAA Hrs mutant protein, or HIV-1 Gag and increasing amounts of either dFYVE or ASAA Hrs mutant protein in addition to increasing amounts of low doses of either wild-type Hrs-HA protein or Tsg101-Myc protein. (A) HIV-1 particles released into the culture medium were collected through a 20% sucrose cushion and analyzed by SDS-PAGE and Western blotting using a mouse monoclonal anti-CA antibody (α-CA). Cells were harvested 48 h posttransfection, and their protein content was analyzed by SDS-PAGE and Western blotting using either an anti-Myc antibody (B) to detect the Tsg101-tagged protein, an anti-HA antibody (C) to detect the HA-tagged dFYVE and ASAA mutants, or anti-CA antibody (D) to detect the p55 HIV-1 Gag polyprotein.

An Hrs mutant causes arrest of HIV-1 budding in “lump-like” structures at the cell surface.

SEM was used to further characterize the block to particle production caused by Hrs. Two HIV-1 Gag-expressing plasmids were used: an HIV-1 Gag-Pol construct (45) and HIV-1 Gag-GFP (23). In control experiments, HeLa cells were transfected with both HIV Gag-Pol and HIV-1 Gag-GFP at a DNA ratio of 2:1; individual cells were visualized first by confocal fluorescence microscopy and then by SEM. HIV-1 Gag-GFP was distributed in punctate staining at the cell surface (Fig. 8A). SEM analysis of the same cell verified that the same fluorescent spots observed by confocal microscopy directly correlated with numerous virus-like particles that accumulated at the cell membrane of transfected cells (Fig. 8B and C, arrows), thus verifying that most of the Gag was efficiently budding from the cell. No virus-like particles were seen on the surface of mock-transfected cells (Fig. 8D). When cells were transfected with a mix of HIV-1 Gag-Pol and HIV-1 Gag-GFP and the ASAA Hrs-HA mutant at a DNA ratio of 1:2, large fluorescent structures were observed at the cell surface by confocal microscopy (Fig. 8E). SEM analysis of the same cells showed that cells contain viral particles arrested in “lump-like” structures that remained attached to the cell surface (Fig. 8F, G, H, and I). Thus, an ASAA Hrs mutant arrests HIV-1 Gag particle production in late stages of budding in delimited spots of the membrane where particles are concentrated in large numbers. This pattern of accumulation is distinct from the classical “L domain” mutant phenotype in which individual virion bud structures accumulate at the cell surface.

FIG. 8.

ASAA Hrs mutant blocks HIV-1 Gag particle release in “lump-like” structures at the plasma membrane. (A) Fluorescence-merged z sections of a HeLa cell transfected with a mixture of HIV Gag-Pol and HIV Gag-GFP DNAs at a 2:1 ratio. (B and C) Corresponding SEM images of the surfaces of the same individual cell shown in A. (D) HeLa cells that were mock transfected. (E) Fluorescence-merged z sections of a HeLa cell transfected with a mixture of HIV Gag-Pol and HIV Gag-GFP DNAs at a 2:1 ratio in addition to the ASAA HA-Hrs mutant construct expressed in trans. (F, G, and H) Corresponding SEM images of the surfaces of the same individual cell shown in E. (I) Field of HeLa cells carrying “lumps” of particles caused by the expression of the ASAA Hrs mutant.

DISCUSSION

The studies presented here show that Hrs and Tsg101 make multiple independent contacts through multiple regions of both proteins. The major binding sites in Hrs were found to lie in two independent binding domains located in the C terminus, the CC and P/Q-rich domains. Each of these separately expressed Hrs domains could independently bind Tsg101, with the P/Q-rich domain displaying a higher affinity for Tsg101 than the CC domain. Further analysis showed that the Hrs P/Q-rich domain bound to the N-terminal UEV domain of Tsg101. Mutations in motifs in the P/Q-rich domain similar to PSAP, a PSGP motif (residues 583 to 586) and a PSMP motif (residues 620 to 623), eliminated the binding of the isolated Hrs C-terminal domain to Tsg101. These findings suggest that the Tsg101 UEV can bind motifs that contain the consensus sequence PSXP, where the prolines and serine residues are conserved and the X residue could be an alanine, a glycine, or a methionine residue. The yeast two-hybrid data also showed that the CC domain of Hrs strongly bound to the CC domain located in the C-terminal portion of Tsg101. Together, these results indicate that Hrs carries multiple Tsg101 binding sites concentrated at the C-terminal region of the protein.

The data further show that the overexpression of wild-type Hrs and mutants of Hrs can potently interfere with HIV-1 Gag particle production. Fragments of Hrs that include only the N-terminal region of the protein, including the PSAP motif, showed no inhibition of particle release. Thus, the PSAP motif is not sufficient for the block. In contrast, expression of the C-terminal region containing the CC and P/Q-rich domains efficiently inhibited particle release. Moreover, two Hrs mutants, Hrs lacking the FYVE domain (dFYVE mutant), rendering the protein unable to bind endosomal membrane (41), and a mutant carrying mutations in the PSAP motif (ASAA mutant), both exerted a potent block of HIV-1 Gag particle release. Thus, these motifs and any interactions that they mediate are not essential for inhibition. Interestingly, a deletion of the C-terminal P/Q-rich domain rendered the Hrs protein incapable of inhibiting HIV-1 Gag particle production. These findings indicate that the retention of the intact C terminus is required to efficiently inhibit HIV particle production.

The data presented here showed that dFYVE and ASAA Hrs mutants could efficiently interfere with the Tsg101-HIV Gag interaction in vivo. Expression of Tsg101 in trans (at a Gag-to-Tsg101 ratio of 5:1) overcame the block to HIV-1 Gag particle release, which is consistent with the notion that the Hrs block was mediated by the titration of the available Tsg101. The overexpression of larger amounts of Tsg101 had an inhibitory effect on the release of HIV-1 virus-like particles (VLPs) (Fig. 6), suggesting a tight modulation of VLP release by Tsg101. The accumulation of a high quantity of Tsg101 in the cytoplasm may inhibit HIV-1 particle release by binding up other members of the ESCRT machinery required for HIV-1 budding and release. The results further show that the disruption of the Tsg101-HIV Gag interaction did not absolutely require an intact PSAP motif in the interfering Hrs proteins, since the ASAA Hrs mutant efficiently interfered with Tsg101-HIV Gag binding and abolished HIV Gag particle production. However, the Hrs dFYVE mutant, containing an intact PSAP motif, exerted an even stronger inhibition of HIV Gag particle release and completely abolished Tsg101-HIV Gag binding, suggesting that an intact PSAP motif may have increased the ability of this Hrs mutant to compete with Tsg101 binding to HIV-1 Gag. The deletion of the dFYVE domain in Hrs may have exposed motifs that became available to capture Tsg101 when the protein was no longer capable of binding the endosomal membrane. Unlike wild-type Hrs and ASAA, the dFYVE Hrs mutant is also “free” in the cytoplasm; its new localization may have contributed to its potent effect on HIV-1 particle release. The result is consistent with previous reports that the PSAP motif (residues 348 to 351) located in the center of Hrs may contribute to the binding of Tsg101 in the cell (31, 39). It should be noted, however, that the overexpression of wild-type Hrs did not disrupt the Tsg101-HIV-1 Gag interaction but still inhibited virion production. Thus, wild-type Hrs acts through a distinct mechanism.

The effect of Hrs mutants and fragments was specific to HIV-1 in that these same constructs caused no inhibition of MoMLV particle production. The primary L domain of MoMLV is a PPPY motif in p12^Gag (56), which binds members of the Hect E3 ligase Nedd4 family (33). The use of this L domain may somehow bypass the ability of Hrs fragments to block virion production. The Hrs constructs also had no effect on a MoMLV-PTAP chimera provirus, which uses the HIV-1 PTAP motif (35, 56). This result suggests the existence of additional late domains in MoMLV Gag that act even in the face of inhibitory Hrs expression. Alternatively, other aspects of MoMLV budding may differ from HIV-1 budding and account for the lack of inhibition of the chimeric Gag protein.

Scanning electron microscopy showed that the overexpression of the Hrs ASAA mutant blocked HIV Gag particle production at late stages of budding; the cell surface showed clusters of particles that seemed to be budding from the same area of the membrane without properly detaching from each other or from the cell. The particle arrest weakly resembles the budding blocks observed when Tsg101 is depleted by RNA interference or when dominant fragments of Tsg101 are overexpressed (16, 18). However, the Hrs phenotype differs from a typical late-domain block because the arrested particles are gathered in “lump-like” structures at the cell surface, suggesting that the expression of the Hrs ASAA mutant seemed to arrest HIV Gag budding in a localized “island” of the plasma membrane.

The Hrs P/Q-rich region, containing the PSGP and PSMP motifs, seems to play the critical role in binding Tsg101 UEV and in blocking virus release. Consistent with this notion, Pornillos et al. (39) previously showed that the fusion of a C-terminal fragment of Hrs containing the CC and P/Q-rich domains to HIV-1 Gag could substitute for the HIV-1 PTAP late domain, possibly because it can serve as a docking site for Tsg101. For this function, the CC domain and the PSAP motif of Hrs were shown to be dispensable, and the presence of an intact P/Q-rich domain was required (39). Our data are in agreement with these findings and show that the P/Q-rich domain of the Hrs protein contains strong and independent Tsg101 binding sites that can efficiently interfere with HIV-1 Gag-Tsg101 interactions by sequestering Tsg101.

In summary, this study has defined domains of the Hrs protein that can potently interfere with HIV Gag particle production by sequestering Tsg101 away from HIV Gag. New Tsg101 binding sites were identified in the C-terminal end of Hrs and include new proline-rich motifs, the PSGP and PSMP motifs. Mutants and fragments of Hrs that contained these new Tsg101 binding sites exerted a strong inhibition of HIV particle release and arrested HIV-1 particles in late stages of budding but in a distinctive late-domain phenotype.