|Home | About | Journals | Submit | Contact Us | Français|
The human immunodeficiency virus type 1 (HIV-1) accessory protein Nef directs virus escape from immune surveillance by subverting host cell intracellular signaling and membrane traffic to down-regulate cell-surface major histocompatibility complex class I (MHC-I). The interaction of Nef with the sorting proteins PACS-1 and PACS-2 mediates key signaling and trafficking steps required for Nef-mediated MHC-I down-regulation. Little is known, however, about the molecular basis underlying the Nef–PACS interaction. Here we identify the sites on Nef and the PACS proteins required for their interaction and describe the consequences of disrupting this interaction for Nef action. A previously unidentified cargo subsite on PACS-1 and PACS-2 interacted with a bipartite site on Nef formed by the EEEE65 acidic cluster on the N-terminal domain and W113 in the core domain. Mutation of these sites prevented the interaction between Nef and the PACS proteins on Rab5 (PACS-2 and PACS-1)- or Rab7 (PACS-1)-positive endosomes as determined by bimolecular fluorescence complementation and caused a Nef mutant defective in PACS binding to localize to distorted endosomal compartments. Consequently, disruption of the Nef–PACS interaction repressed Nef-induced MHC-I down-regulation in peripheral blood mononuclear cells. Our results provide insight into the molecular basis of Nef action and suggest new strategies to combat HIV-1.
After human immunodeficiency virus type 1 (HIV-1) transmission, the acute phase of infection is manifest by a massive burst in viremia and viral cytopathicity that triggers the antiviral response (McMichael et al., 2010 ). Initially, the innate immune response is mobilized, which induces expression of restriction factors to impede the virus (Geijtenbeek and Gringhuis, 2009 ). The innate immune response also shapes the subsequent action of the adaptive immune response, which requires major histocompatibility complex class I (MHC-I) molecules to present viral antigens on the surface of infected cells, targeting them for destruction by cytotoxic T lymphocytes (CTLs; Lieberman, 2003 ).
HIV-1 uses several mechanisms to combat the antiviral response. The accessory proteins Vif and Vpu blunt the innate immune response by targeting destruction of APOBEC 3G and tetherin, respectively (Kirchhoff, 2010 ). HIV-1 uses two strategies to combat the adaptive immune response. One is the production of CTL escape mutants, which are generated by the low fidelity of HIV-1 replication (Sarmady et al., 2011 ). This escape mechanism, however, is ultimately exhausted by the unsustainable cost to viral fitness (Friedrich et al., 2004 ). The accompanying strategy involves the down-regulation of cell-surface MHC-I. Indeed, the extent of MHC-I down-regulation in simian immunodeficiency virus-1–infected macaques correlates directly with the severity of disease progression, suggesting an important role for this immune-evasive mechanism in vivo (Friedrich et al., 2010 ).
HIV-1 relies on the 27-kDa N-myristoylated accessory protein Nef to down-regulate MHC-I (Peterlin and Trono, 2003 ). Nef is required for the onset of acquired immune deficiency syndrome and can affect cells in many ways, including alteration of T cell activation and maturation, promotion of viral infectivity, subversion of the apoptotic machinery, and down-regulation of cell-surface molecules, including MHC-I (Fackler and Baur, 2002 ). Surprisingly, Nef controls these diverse processes using a limited number of sites that combine to bind >40 cellular proteins and multienzyme complexes, enabling the virus to subvert posttranscriptional programs as well as key steps in intracellular signaling and endocytic membrane traffic (Sarmady et al., 2011 ). However, little is known about the biochemistry underlying most Nef–host protein interactions, precluding an understanding of how Nef drives HIV-1 disease.
To down-regulate MHC-I, Nef directs a temporally regulated program that uses at least four evolutionarily conserved sites—M20, the EEEE65 acidic cluster, and the PXXP75 SH3-domain binding site, all of which are located in the N-terminal flexible region, and P78 (Lee et al., 1996 ; Casartelli et al., 2006 ; Dikeakos et al., 2010 ). During the first 2 d after infection, Nef down-regulates MHC-I by an endocytic program called the signaling mode (Dikeakos et al., 2010 ). This mode is initiated by the EEEE65-dependent binding to the sorting protein PACS-2, which targets Nef to the paranuclear region, enabling PXXP75 to direct assembly of an SFK/ZAP-70/PI3K complex that accelerates endocytosis of cell-surface MHC-I (Blagoveshchenskaya et al., 2002 ; Hung et al., 2007 ; Atkins et al., 2008 ). Internalized MHC-I is then sequestered in paranuclear compartments by subversion of endocytic trafficking steps, which requires the EEEE65-dependent binding to PACS-1 as well as Nef M20 and the heterotetrameric adaptor AP-1 (Blagoveshchenskaya et al., 2002 ; Roeth et al., 2004 ; Noviello et al., 2008 ; Dikeakos et al., 2010 ). By 3 d postinfection, Nef switches to the stoichiometric mode of down-regulation that prevents delivery of newly synthesized MHC-I molecules to the cell surface (Dikeakos et al., 2010 ). Pharmacological inhibition of the signaling mode blocks onset of the stoichiometric mode, suggesting a key role for the PACS proteins in Nef-induced MHC-I down-regulation. This time course of the switch in Nef action suggests that HIV-1 may use the signaling and stoichiometric modes to down-regulate MHC-I in a reservoir-specific manner. For example, the ~1.5-d lifespan of an HIV-1–infected CD4+ T cell suggests that Nef relies on the signaling mode to down-regulate MHC-I in this short-lived reservoir, whereas in longer-lived reservoirs, including cells of the monocytic lineage, Nef may sequentially employ the signaling mode, followed by the stoichiometric mode (Dikeakos et al., 2010 ). The role of P78 has not been reported.
The PACS proteins are multifunctional homeostatic regulators expressed in metazoans that integrate secretory pathway traffic and interorganellar communication in healthy cells, with key steps in death ligand–induced apoptosis in diseased cells (Aslan et al., 2009 ; Youker et al., 2009 ). Lower metazoans express a single PACS gene, whereas higher metazoans express the PACS-1 and PACS-2 genes (Youker et al., 2009 ). Caenorhabditis elegans PACS localizes to early endosomes and was identified in a genetic screen as a mediator of synaptic transmission (Sieburth et al., 2005 ). In mammalian cells, PACS-1 mediates the trans-Golgi network (TGN) localization of furin and other cargo proteins that contain acidic cluster binding sites (Wan et al., 1998 ; Crump et al., 2001 , 2003 ; Bouard et al., 2007 ; Youker et al., 2009 ). PACS-1 also mediates recycling of internalized cargo from early endosomes to the cell surface and trafficking to the primary cilium (Molloy et al., 1998 ; Schermer et al., 2005 ; Chen et al., 2009 ; Jenkins et al., 2009 ). PACS-2 mediates the localization of protein cargo to the endoplasmic reticulum and trafficking of endocytosed cargo from early endosomes to the TGN (Kottgen et al., 2005 ; Atkins et al., 2008 ; Myhill et al., 2008 ).
The 150–amino acid PACS-1 and PACS-2 cargo (furin)-binding regions (FBRs) bind cargo proteins, sorting adaptors, and cellular kinases (Youker et al., 2009 ). Whereas sites that bind sorting adaptors and kinases have been identified, the sites that bind cargo have not. Moreover, mutation of Nef EEEE65 only reduces binding to PACS proteins, suggesting that their interaction is multivalent (Atkins et al., 2008 ). Here we identify a conserved site in the PACS-1 and PACS-2 FBRs that interacts with a bipartite site on Nef formed by the EEEE65 acidic cluster on the N-terminal domain and W113 in the Nef core domain. Mutation of this bipartite site disrupts the interaction with PACS proteins and correspondingly blocks Nef trafficking and MHC-I down-regulation. We also report the role of Nef P78 in binding PACS proteins and mediating Nef function. Our results provide new insight into the molecular basis of Nef action and suggest new strategies to combat HIV-1.
Ab initio modeling predicts that the PACS-1 FBR (residues 117–266) is a loosely packed structure composed of long, flexible loops interspersed with several defined secondary structure elements (Youker et al., 2009 ). The sites in PACS-1 important for binding the AP-1 heterotetramer (ETELQLTF175), the monomeric adaptor GGA3 (KIY251), or the protein kinase CK2 β-subunit (RRKRY200) reside in ordered regions located on different surfaces of the predicted structure (Crump et al., 2001 ; Scott et al., 2006 ; Youker et al., 2009 ; Figure 1A). The cargo-binding sites, however, have not been identified. The PACS-1 FBR binds the furin cytosolic domain at the CK2-phosphorylated acidic cluster EECPpS773DpS775EEDE (Wan et al., 1998 ). By contrast, the HIV-1 Nef acidic cluster EEEE65, which is required for maximal binding of Nef to PACS-1 or PACS-2, lacks a CK2 site, suggesting that PACS proteins may bind Nef and furin differently. Consistent with this possibility, glutathione S-transferase (GST)–tagged PACS-1117-266 bound hexahistidine (His6)-tagged furin containing S773S775→DD phosphomimetic mutations (furin-DD) more robustly than His6-Nef in vitro (Figure 1C). Correspondingly, a yeast two-hybrid analysis showed PACS-1117-266 interacted with furin-DD but not Nef (Figure 1B).
The relatively weak binding between Nef and PACS-1117-266 in vitro, together with the lack of a detectable interaction between these two molecules using the two-hybrid assay, raised the possibility that misfolded sequences within the PACS-1 FBR selectively interfered with the Nef-binding site. Consistent with this possibility, protein stability algorithms predicted the PACS-1 FBR is largely disordered but contains a stable region between residues 181 and 240 (Guruprasad et al., 1990 ). To determine whether this stable region would be sufficient to interact with Nef, we tested the ability of PACS-1 FBR segments spanning residues 117–180, 181–240, and 241–266 to interact with Nef or furin-DD (Figure 1B). We found that Nef and furin-DD interacted robustly with PACS-1181-240 but not with PACS-1117-180 or PACS-1241-266, suggesting that disordered regions flanking PACS-1 FBR residues 181–240 selectively interfered with binding to Nef. Further truncation analyses revealed that Nef interacted with PACS-1181-195. Deletion of the C-terminal Q195, however, blocked the interaction of PACS-1181-194 with Nef. As controls, we analyzed the ability of the PACS-1 FBR segments to interact with CK2β or GGA3 (Figure 1B). In agreement with previous studies, CK2β interacted with PACS-1181-204 but not PACS-1181-195, consistent with the importance of the PACS-1 RRKRY200 polybasic segment for the interaction of PACS-1 with CK2β (Scott et al., 2006 ). In addition, GGA3 interacted only with PACS-1241-266, which contains the KIY251 GGA3-binding site (Scott et al., 2006 ).
To test whether PACS-1181-195 could bind Nef, we conducted an in vitro protein capture assay (Figure 1C). His6-tagged Nef or furin-DD was mixed with GST-tagged PACS-1117-266 or PACS-1181-195 or with GST alone. GST proteins were captured, and bound Nef or furin-DD was detected by Western blot. As expected, PACS-1117-266 bound furin-DD approximately sevenfold greater than Nef. By contrast, PACS-1181-195 bound Nef but not furin-DD. Together these experiments suggest that PACS-1 robustly binds furin and Nef and that disordered regions in the FBR selectively interfere with the Nef subsite.
Next we conducted an alanine scan to identify which residues in PACS-1181-195 were required for interaction with full-length Nef (Nef1-206; Figure 2A, top). Polyalanine substitution of D186-K189 but not H181-R185 blocked the interaction between PACS-1181-195 and Nef. Accordingly, a single alanine scan from PACS-1 D186 to Q195 showed that N188, K189, and Q195 were required for the interaction between PACS-1181-195 and Nef. To determine whether mutational inactivation of the Nef:PACS-1181-195 interaction in the two-hybrid assay reflected a direct block in protein–protein binding, we conducted in vitro protein capture assays. In agreement with the yeast two-hybrid analysis, His6-Nef bound GST-tagged PACS-1181-195 but not GST-tagged PACS-1181-195 containing N188→A or K189→A substitutions (Figure 2B and Supplemental Figure S1). Because the PACS-1 N188→A and K189→A substitutions blocked Nef binding equally, subsequent analyses focused on the N188→A mutation. To determine whether PACS-1 Asn188 was required for interaction between the corresponding full-length proteins in mammalian cells, we conducted coimmunoprecipitation analyses (Figure 2C). We coexpressed Nef-eYFP with hemagglutinin (HA)-tagged PACS-1 or PACS-1N188A. PACS-1 proteins were immunoprecipitated, and coprecipitating Nef was detected by Western blot. In agreement with the yeast two-hybrid analyses, Nef interacted more efficiently with PACS-1 than with PACS-1N188A. Moreover, PACS-1 Asn188 appeared to be selectively required for the interaction of PACS-1 with Nef compared with other client proteins, since the N188→A substitution blocked the interaction of PACS-1181-240 with Nef but not with CK2β or furin-DD (Supplemental Figure S2 and Figure 1).
The PACS-1181-195 sequence is nearly identical to the corresponding sequence located between residues 102 and 116 in the PACS-2 FBR, except that DA187 in PACS-1 is substituted by EG108 in PACS-2 (Figure 1A). To test whether Nef could interact with the PACS-2 FBR, we generated a PACS-1181-195 construct containing the DA187→EG substitution present in PACS-2. We found no difference in the interaction between Nef and PACS-1181-195 or PACS-2102-116, suggesting that Nef binds the same sites in the PACS-1 and PACS-2 FBRs (Figure 2A, bottom). Because PACS-1 N188 was required for the interaction of PACS-1 with Nef, we then asked whether PACS-2 N109, which corresponds to PACS-1 N188, was similarly required for the interaction with Nef in vivo. A coimmunoprecipitation analysis showed that Nef-eYFP interacted with HA-tagged PACS-2 but not with PACS-2N109A (Figure 2C).
The repressed interaction between Nef and PACS-1N188A or PACS-2N109A raised the possibility that one or both of these mutant PACS proteins would interfere with Nef-induced down-regulation of MHC-I. To test this, we transfected H9 T cells with plasmids expressing eYFP (vector) or Nef-eYFP alone or cotransfected them with plasmids expressing Nef-eYFP and either PACS-1, PACS-1N188A, PACS-2, or PACS-2N109A (Figure 2D). We found that Nef-induced MHC-I down-regulation was markedly repressed by PACS-1N188A, as well as by PACS-2N109A, suggesting that these mutants disturbed Nef action by interfering with binding of Nef to endogenous PACS proteins.
Identification of residues in the PACS-1 or PACS-2 FBRs required for the interaction with Nef provided an opportunity to visualize the subcellular location of the interaction between Nef and PACS-1 or PACS-2 using bimolecular fluorescence complementation (BiFC; Kerppola, 2008 ). Cells coexpressing BiFC reporter proteins composed of PACS-1 fused to the nonfluorescent yellow fluorescent protein (YFP) N-terminal fragment (PACS-1-Yn) or Nef fused to the nonfluorescent YFP carboxy-terminal fragment (Nef-Yc) displayed a dispersed cytoplasmic punctate staining pattern (Figure 3A). By contrast, coexpression of Nef-Yc with PACS-1N188A-Yn generated a markedly reduced BiFC signal despite similar levels of expression, suggesting that the fluorescent signal generated by coexpression of Nef-Yc with PACS-1-Yn represented a bona fide protein–protein interaction (Figure 3, A and B). Next we tested whether the interaction between Nef and PACS-2 could by visualized by BiFC. We detected a punctate BiFC signal in the cytoplasm of cells that coexpressed Nef-Yc with PACS-2 fused to the nonfluorescent YFP N-terminal fragment (PACS-2-Yn) but not with PACS-2N109A-Yn (Figure 3, A and B). Singular expression of Nef-Yc, PACS-1Yn, or PACS-2Yn alone generated background BiFC signals similar to that obtained by coexpression of Nef-Yc with PACS-1N188A-Yn or with PACS-2N109A-Yn, consistent with the determination that the mutant PACS proteins generate only background interaction with Nef (unpublished data).
To test whether the punctate BiFC staining pattern represented the interaction of Nef and PACS proteins on endosomes, we coexpressed the BiFC reporter proteins with mcherry-Rab5, which decorates early endosomes, or mcherry-Rab7, which decorates multivesicular bodies (MVBs)/late endosomes (Figure 3C). We found that the BiFC signal generated by the interaction of Nef-Yc with PACS-1-Yn overlapped predominantly with mcherry-tagged Rab7 but not Rab5. By contrast, the BiFC signal generated by the interaction of Nef-Yc with PACS-2Yn overlapped with both mcherry-tagged Rab5 and Rab7. These findings suggest that Nef interacts with PACS-1 on MVB/late endosomes and with PACS-1 or PACS-2 on early endosomes.
The Nef acidic cluster EEEE65 is required for MHC-I down-regulation and for maximal binding to PACS-1 or PACS-2 (Greenberg et al., 1998 ; Piguet et al., 2000 ; Atkins et al., 2008 ). However, the EEEE65→AAAA65 (NefE4A) mutation only partially reduces binding to PACS proteins, suggesting that additional sites in Nef are required for maximal binding (Atkins et al., 2008 ). To systematically identify Nef residues required for binding PACS proteins, we conducted a yeast two-hybrid screen to analyze the interaction between PACS-1181-195 and a series of Nef truncation mutants (Figure 4A). As expected, PACS-1181-195 interacted with Nef1-65 but not with Nef1-61 or Nef1-65E4A (Figure 4B). Surprisingly, however, PACS-1181-195 also interacted with Nef71-206, suggesting that PACS-1 binds a second site in the Nef core domain (Figure 4C). An N-terminal truncation analysis showed that PACS-1181-195 interacted with Nef77-206 but not Nef80-206, suggesting that PACS-1181-195 binding to Nef requires a site in the polyproline region C-terminal to the PXXP75 site. Stepwise alanine substitution of Nef residues 77-79 revealed that only P78 was required for interaction of the Nef C-terminal region with PACS-1181-195.
We conducted several experiments to test the extent to which Nef EEEE65 and P78 are required for interacting with PACS-1 or PACS-2 and for mediating PACS-dependent Nef action. We found in the yeast two-hybrid assay that PACS-1181-195 interacted with full-length Nef (Nef1-206) but not Nef-E4AP78A, which contains the EEEE65→AAAA and P78→A double mutation (Figure 4D). Correspondingly, coimmunoprecipitation experiments in mammalian cells showed that PACS-1 and PACS-2 interacted with Nef but not Nef-E4AP78A (Figure 4E). To test the effect of the E4AP78A mutation on Nef-induced MHC-I down-regulation, we incubated H9 T-cells expressing eYFP-tagged Nef or Nef-E4AP78A or eYFP alone with an anti-MHC-I antibody and analyzed them by flow cytometry. In agreement with the coimmunoprecipitation analysis, we found that cell-surface MHC-I was efficiently down-regulated by Nef but not by Nef-E4AP78A (Figure 4F).
The failure of Nef-E4AP78A to interact with PACS-1 and PACS-2 suggested that, in addition to an inability to down-regulate MHC-I, this Nef mutant would be blocked in PACS-dependent trafficking. In agreement with previous studies (Atkins et al., 2008 ; Dikeakos et al., 2010 ), Nef-eYFP expressed in HeLa cells accumulated in the Golgi region (Figure 4G). By contrast, Nef-E4AP78A-eYFP accumulated largely in a dispersed population of transferrin receptor (TfR)–positive early endosomes. The subcellular distribution of Nef-E4AP78A-eYFP was similar to the mislocalization of Nef-eYFP caused by PACS-2 small interfering RNA (siRNA) knockdown, which also caused Nef-eYFP to redistribute to a population of dispersed, TfR-positive early endosomes (Figure 4G and Supplemental Figure S3; Atkins et al., 2008 ; Dikeakos et al., 2010 ).
To gain insight into how Nef may bind PACS-1 and PACS-2, we used the ClusPro 2.0 server with unconstrained binding parameters to dock PACS-1181-195 onto Nef (Protein Data Bank [PDB] ID: 2Nef; Grzesiek et al., 1997 ; Kozakov et al., 2010 ). The docked cluster with the most populated poses is shown in Figure 5A. The α-helical structure of PACS-1181-195 portrayed in this analysis was consistent with ab initio modeling of the PACS-1 FBR and was supported empirically by circular dichroism (CD) spectroscopy of a synthetic peptide containing the PACS-1 residues N188, K189, and Q195 required for interaction with Nef (Youker et al., 2009 ; Supplemental Figure S4). Because a peptide corresponding to PACS-1 181–195 was insoluble at neutral pH, we determined the secondary structure of the slightly offset 17-mer peptide corresponding to PACS-1 184–200. CD analysis showed that PACS-1184-200 contained a small amount of helical structure that was abolished by the denaturant urea and increased by the α-helix stabilizer 2,2,2-trifluorethanol (TFE; Luo and Baldwin, 1997 ).
The docked poses generated by the ClusPro 2.0 server suggested that PACS-1181-195 could interact with a bipartite site in Nef composed of EEEE65 and helix αB in the core domain but not with P78 (Figure 5A). Of interest, the HIV-1 Nef structure reveals that P78 packs against Y120, which stabilizes the Nef core region by capping helix αB (Lee et al., 1996 ). Consistent with the structural data, a yeast two-hybrid analysis showed that P78→A or Y120→A substitutions each blocked the interaction of Nef77-206 with PACS-1181-195 (Figure 5B, left). By contrast, neither the P78→A nor the Y120→A substitution interfered with the interaction of Nef77-206 with Arf1T31N, which interacts with Nef EE156 (Figure 5B, right; Faure et al., 2004 ). Conversely, a Nef EE156→AA substitution blocked interaction of Nef with Arf1T31N but not PACS-1181-195. Together these findings suggest that P78 promotes interaction of Nef with PACS proteins not by binding PACS-1 but by interacting with Y120, which stabilizes the structural integrity of the Nef core domain required for binding PACS-1.
In support of the possibility that PACS-1 interacts with Nef helix αB, we found that PACS-1181-195 interacted with Nef77-125 but not Nef120-206 (Figure 5C). Further analyses showed that PACS-1181-195 interacted robustly with Nef104-119, which corresponds to helix αB, but not with Nef77-96, which corresponds to helix αA. To identify specific residues in Nef helix αB required for interaction with PACS-1181-195, we conducted a single-alanine scan of Nef residues 104–119 and found that mutation of L112, W113, or I114 blocked interaction with PACS-1181-195 (Figure 5D). To test whether this hydrophobic cluster was required for binding Nef to PACS-1 in vitro, we incubated GST-tagged PACS-1181-195 with His6-tagged Nef or Nef-E4AW113A. As a control, we tested the binding of GST-tagged Arf1T31N to His6-tagged Nef or Nef-E4AW113A. In agreement with the yeast two-hybrid analysis, we found that Nef bound Arf1T31N and PACS-1181-195, whereas Nef-E4AW113A bound Arf1T31N but not PACS-1181-195 (Figure 5E).
To test whether the hydrophobic cluster in Nef helix αB was required for interacting with full-length PACS proteins, we coexpressed HA-tagged PACS-1 or PACS-2 with Nef, NefE4A, or Nef-E4AW113A. The PACS proteins were immunoprecipitated, and coprecipitating Nef proteins were detected by Western blot. In agreement with previous studies, the PACS proteins interacted with Nef and to a lesser extent with NefE4A (Figure 6A; Atkins et al., 2008 ). However, similar to our results for Nef-E4AP78A (Figure 4), the PACS proteins failed to coimmunoprecipitate Nef-E4AW113A. Because PACS-2-mediated trafficking of Nef to the Golgi region is required for the PXXP75-dependent interaction of Nef with SFKs (Atkins et al., 2008 ), we asked whether the E4AW113A mutation would affect the interaction with Hck. We found that Hck was readily coimmunoprecipitated with Nef but not Nef-E4AW113A (Supplemental Figure S5). To test the effect of the E4AW113A mutation on MHC-I down-regulation, we conducted two experiments. First, H9 T cells expressing eYFP-tagged Nef or Nef-E4A W113A or eYFP alone (vector) were incubated with an anti-MHC-I antibody and analyzed by flow cytometry. In agreement with the coimmunoprecipitation analysis, we found that cell-surface MHC-I was efficiently down-regulated by Nef but not by Nef-E4AW113A (Figure 6B). To determine whether the EEEE65 and W113 sites were required for Nef-induced MHC-I down-regulation in primary target cells, PBMCs were nucleofected with plasmids expressing eYFP-tagged Nef or Nef-E4AW113A or eYFP alone (vector), and the extent of HLA-A2.1 down-regulation was monitored by flow cytometry (Supplemental Figure S6). In agreement with the studies in H9 T cells, the E4AW113A mutation markedly repressed Nef-induced down-regulation of HLA-A2.1.
We asked to what extent the failure of Nef-E4AW113A to interact with PACS proteins would affect its trafficking and found that this Nef mutant failed to localize to the Golgi region (Figure 7, A and B). Whereas Nef-E4AW113A-eYFP and Nef-E4AP78A-eYFP were both redistributed to cortical TfR- and mcherry-Rab5–positive early endosomes, Nef-E4AW113A-eYFP expression induced accumulation of tubular- and donut-shaped endosomal structures (compare Figures 4G and and7A).7A). We found that Nef-E4AW113A-eYFP colocalized predominantly with mcherry-Rab7 in donut-shaped compartments, suggesting that they were a population of MVB/late endosomes (Figure 7, A and C). By contrast, Nef-E4AW113A-eYFP was predominantly found in Rab5-positive tubular structures, possibly reflecting a population of tubulated early endosomes. Structured illumination microscopy was used to more precisely resolve the subcellular localization of Nef-E4AW113A-eYFP (Figure 7D, left). We found that tubule-localized Nef-E4AW113A-YFP was frequently concentrated in varicosities that extended linearly from mcherry-Rab5–containing endosomes. By contrast, donut-localized Nef-E4AW113A-YFP was frequently found apposed to or surrounding endosomal mcherry-Rab7 (Figure 7D, right). Together these findings suggest that EEEE65 and W113 combine to mediate the interaction of Nef with PACS-1 or PACS-2, which is crucial for directing the TGN/endosomal trafficking of Nef and for the ability of Nef to down-regulate MHC-I.
The identification of specific sites on the PACS proteins and HIV-1 Nef necessary and sufficient for their interaction and for Nef-induced MHC-I down-regulation prompted us to ask to what extent the molecular docking analysis was consistent with the empirical data. Of interest, one Nef:PACS-1181-195 pose suggested that PACS-1 NK189 (corresponding to PACS-2 NK110) was included in an α-helical segment with aliphatic chains abutted against W113 in HIV-1 Nef (Figures 8A and Supplemental Figure S4). Moreover, the docking model was also consistent with the ability of PACS-1181-195 containing a conservative N188→Q substitution to interact with Nef104-119 containing a conservative W113→F substitution (Figure 8, B–D). These findings suggest that the failure of the PACS-1 N188→A and Nef W113→A substitutions to support protein–protein binding likely resulted from the small Ala side-chain substitution rather than from a global disruption of protein structure (Figures 2 and and55).
The docking model further predicted that PACS-1 KR185 (KR106 in PACS-2) could interact with HIV-1 Nef EEEE65. Because our initial analysis of the sites in PACS-1181-195 required for interaction with Nef were conducted using full-length Nef (Nef1-206), it raised the possibility that the interaction between Nef W113 and PACS-1 NK189 may have masked a weaker interaction between Nef EEEE65 and PACS-1 KR185. To test this, we systematically analyzed the interaction of a series of PACS-1181-195 constructs containing Ala substitutions at each residue with either Nef1-65, which contains only the EEEE65 site, or Nef104-119, which contains only the W113 hydrophobic site. Mutation of PACS-1 N188, K189, or Q195 blocked the interaction with Nef104-119 but not Nef1-65 (Figure 8E and Supplemental Figure S7). By contrast, mutation of PACS-1 KR185 blocked interaction of PACS-1181-195 with Nef1-65 but not Nef104-119 (Figure 8F and Supplemental Figure S7). Together these findings suggest that PACS proteins mediate Nef action by interaction between bipartite binding sites—one formed by KR185 and NK189 (PACS-1 numbering) located within a small, 15–amino acid helical segment in the PACS-1 and PACS-2 FBRs and the other formed by EEEE65 in the Nef N-terminal flexible domain and W113 in helix αB in the Nef core.
The 150–amino acid PACS-1 and PACS-2 FBRs contain sites required for binding secretory pathway trafficking regulators, including AP-1, COPI, CK2, and GGA3, as well as a large number of cargo proteins that contain acidic cluster sorting motifs (Youker et al., 2009 ). Despite the markedly different acidic cluster motifs on furin and Nef, these two cargo proteins interact with overlapping subsites in the PACS-1 and PACS-2 FBRs distinct from sites that interact with adaptors or CK2 (Figure 1). Here we identified a 15–amino acid segment of PACS-1 (residues 181–195) or PACS-2 (residues 102–116) sufficient to interact robustly with Nef; this interaction required residues N188 (N109 in PACS-2), K189, and Q195. Mutation of the essential asparagine residue in the Nef-binding site on PACS-1 or PACS-2 (PACS-1N188A or PACS-2N109A) blocked the interaction of PACS proteins with Nef on endosomal compartments and repressed Nef-induced MHC-I down-regulation (Figures 2 and and3).3). Correspondingly, the NefE4AP78A and NefE4AW113A mutants unable to bind PACS-1 or PACS-2 were blocked in PACS-dependent trafficking of Nef to the paranuclear region and in MHC-I down-regulation (Figures 4–7). Protein interaction assays showed that PACS-1 KR185 was required for interaction with Nef EEEE65, whereas N188, K189, and Q195 were required for interaction with a hydrophobic site in the Nef core that included L112, W113, and I114 in helix αB (Figure 8). The conservation of the EEEE65 and W113 sites from multiple HIV clades suggests a broad and essential role for the Nef–PACS interaction in driving Nef function (Lee et al., 1996 ; Kirchhoff et al., 1999 ; Ali et al., 2010 ).
The PACS family of sorting proteins arose late in evolution and is expressed only in metazoans (Youker et al., 2009 ). Of interest, C. elegans PACS localizes to early endosomal compartments, where it mediates trafficking steps controlling cell–cell communication (Sieburth et al., 2005 ). Thus the interaction of HIV-1 Nef with PACS proteins on endosomal compartments as determined using BiFC suggests an evolutionarily conserved role for PACS-1 and PACS-2 in endosomal traffic (Figure 3). The colocalization of Nef with PACS-1 on a subpopulation of Rab7-positive endosomes is consistent with the role of PACS-1 in sequestering MHC-I molecules after their Nef-induced endocytosis and the determination that down-regulated MHC-I accumulates in Rab7-positive compartments (Schaefer et al., 2008 ; Dikeakos et al., 2010 ). By contrast, Nef interacted with PACS-2 on Rab5-positive early endosomes (Figure 3), which is consistent with the role of PACS-2 in trafficking Nef from peripheral early endosomes to the Golgi region, where it can assemble the multikinase complex that triggers MHC-I down-regulation (Atkins et al., 2008 ).
The Nef proline-rich region mediates multiple steps in Nef-induced MHC-I down-regulation. P72 and P75 form a type II polyproline helix that binds the SH3 domains of SFKs (Lee et al., 1996 ; Grzesiek et al., 1997 ). This binding directly activates a subset of SFKs—Hck, Lyn, and Src—each of which can be assembled by Nef into the SFK/ZAP-70/PI3K multikinase complex that triggers MHC-I down-regulation (Dikeakos et al., 2010 ). By contrast, P78 is not required for SH3 binding but is required for mediating MHC-I down-regulation (Yamada et al., 2003 ; Casartelli et al., 2006 ). However, the mechanism has been unknown. Thus our determination that P78 is required for interaction with PACS-1 and PACS-2 explains one role for this site in Nef-induced MHC-I down-regulation (Figure 4). Our results, together with structural studies, suggest, however, that P78 does not directly bind the PACS proteins but instead interacts with Y120 to stabilize helix αB, enabling W113 to interact with the PACS-1 and PACS-2 FBRs (Figures 4–6; Lee et al., 1996 ; Grzesiek et al., 1997 ).
Consistent with their failure to interact with PACS proteins, both Nef-E4AP78A and Nef-E4AW113A were mislocalized from the paranuclear region to early or late endosomal compartments (Figures 4 and and7).7). Strikingly, however, Nef-E4AW113A was frequently observed in varicosities along tubular structures, suggesting the Nef-E4AP78A and Nef-E4AW113A mutants are phenotypically distinct (Figure 7). The extensive tubulation in Nef-E4AW113A–expressing cells was similar to that observed in cells knocked down for retromer-interacting proteins or expressing constitutively active SNX18 (Haberg et al., 2008 ; Harbour et al., 2010 ). The colocalization of SNX18 with PACS-1 and AP-1 on endosomes suggests that interaction of PACS proteins with their cargo may modulate vesicle formation and point to how Nef uses PACS-1 and AP-1 to sequester internalized MHC-I (Dikeakos et al., 2010 ). The extensive tubulation in Nef-E4AW113A–expressing cells may also result from defects in the early endosome recruitment of COPI, which interacts with PACS-2, or disruption of the Rab5/Rab7 switch, perhaps compounded by the intrinsic ability of Nef to induce tubulation (Kottgen et al., 2005 ; Gerlach et al., 2009 ; Huotari and Helenius, 2011 ). These findings suggest new avenues to investigate how membrane trafficking steps regulate Nef action.
The ClusPro 2.0 protein–protein docking server generated a speculative model of the Nef:PACS bipartite binding site that was consistent with the results determined empirically using yeast two-hybrid interaction, in vitro protein binding, and CD spectroscopy (Figure 8). In this model, PACS-1 KR185 (KR104 in PACS-2) is in the proximity of Nef EEEE65 and PACS-1 NK189 (NK110 in PACS-2) is vicinal to W113 in helix αB of the Nef core. Recent benchmarking studies demonstrate that the ClusPro 2.0 server competes with the best human predictor groups to generate reliable receptor:ligand (i.e., Nef:PACS-1) complexes using independently solved structures, provided that there is limited backbone mobility (Kozakov et al., 2010 ). Thus one caveat of this docking model is location of the EEEE65 subsite within the flexible Nef N-terminal region, which can adopt multiple conformations beyond the single fixed receptor conformation used for docking studies. In addition, the structural importance of PACS-1 Q195 for binding Nef remains to be determined. Of interest, interaction with furin also requires PACS-1 Q195, albeit not KR185, suggesting this residue is uniquely important for maintaining the distinct cargo-binding subsites in the PACS FBRs defined by Nef and furin (Figure 1 and Supplemental Figure S2; Dikeakos and Thomas, unpublished data). Thus PACS proteins bind at least two subclasses of cargo molecules: proteins like furin that contain CK2 phosphorylatable acidic clusters and those like Nef that contain a bulky hydrophobic residue within an α-helical segment. Because many pathogenic viruses express proteins containing CK2 phosphorylatable acidic cluster motifs that mimic the PACS-binding site in furin and other cellular cargo, we suspect that Nef evolved to bind the adjacent subsite used by as-yet-unidentified cellular proteins (Youker et al., 2009 ). Together these results illustrate the exquisite cargo selectivity of the PACS proteins. Structural analyses will be required to elucidate the nature of the cargo subsites on the PACS-1 and PACS-2 FBRs with their respective cargoes.
The multiple pathways by which Nef drives HIV-1 disease are achieved in part by the context-specific protein–protein interactions of a limited number of conserved sites in Nef with cellular protein partners (Geyer et al., 2001 ; Sarmady et al., 2011 ). For example, in addition to mediating binding to PACS proteins, EEEE65 contributes to the ability of Nef to form a complex with AP-1 and the MHC-I cytosolic domain, suggesting that the acidic cluster mediates multiple trafficking steps in the sequestering phase of MHC-I down-regulation (Singh et al., 2009 ). The EEEE65 site is also required for incorporation of Nef into exosomes, for mediating cell-to-cell transfer of Nef through tunneling nanotubes, and for interacting with TRAF2 to promote proinflammatory cytokine signaling (Xu et al., 2009 ; Ali et al., 2010 ; Mangino et al., 2011 ). These findings, together with the location of the EEEE65 site in the flexible N-terminal domain, suggest that this site may mediate multiple low-affinity protein–protein interactions that promote high-affinity interactions, such as the interaction between W113 in the Nef core and PACS-1 NK189. In this manner, these low- and high-affinity sites may combine to drive the diverse actions of Nef. Within helix αB, L112 not only mediates interaction of Nef with PACS-1 and PACS-2, but it also mediates Nef dimerization by combining with Y115 to form a hydrophobic surface that stabilizes the interaction between opposite protomers (Poe and Smithgall, 2009 ). Similarly, whereas W113 is required for the interaction of helix αB with PACS-1 and PACS-2, it also combines with F90 on helix αA to delimit a hydrophobic pocket that stabilizes the interaction of Nef with SH3 domains (Arold et al., 1997 ; Poe and Smithgall, 2009 ; Horenkamp et al., 2011 ). The determination that the interaction of Nef with PACS-2 or SFKs is mutually exclusive is consistent with the dual role of W113 in controlling whether Nef interacts with PACS proteins or with SFKs (Atkins et al., 2008 ). Whether the W113→A mutation disrupts the Nef–Hck interaction by interfering solely with the PACS-2–dependent trafficking of Nef required for interaction with SFKs or also blocks binding of Nef to the Hck SH3 domain warrants additional studies. It will be important to determine how Nef combines a limited number of binding surfaces to interact with the large set of cellular proteins that enable Nef to drive HIV-1 disease and to use this knowledge to develop new approaches to combat HIV-1.
HeLa and H9 T cells were cultured as described (Hung et al., 2007 ; Atkins et al., 2008 ). To prepare peripheral blood mononuclear cells (PBMCs), peripheral blood was obtained from healthy HLA-A*0201+ volunteers by leukapheresis or by venipuncture using protocols approved by the Oregon Health and Science University Institutional Review Board (protocols IRB00004039 and IRB00002251). Written informed consent was obtained from all subjects according to the Declaration of Helsinki. PBMCs were then cultured in RPMI 1640 containing 10% fetal bovine serum (FBS), nonessential amino acids, and pyruvate and supplemented with IL-2 (50 U/ml; Sigma, St. Louis, MO). Plasmids expressing HA-tagged PACS-1 or PACS-2 or Nef-eYFP were previously described (Hung et al., 2007 ) and were used to generate site-directed mutations in PACS-1, PACS-2, or Nef mutants using QuikChange Site-Directed Mutagenesis Kit (Agilent Technologies, Santa Clara, CA). Plasmids expressing mcherry-Rab5 or mcherry-Rab7 were provided by T. Weber (Mount Sinai School of Medicine, New York, NY). For the yeast two-hybrid analyses, the Nef, CK2β, and GGA3 sequences were cloned into pBTM116 bait vector, which contains the LexA DNA–binding domain (Liu et al., 1997 ). The pBTM116 vector expressing furin-DD was described previously (Wan et al., 1998 ). The PACS sequences were cloned into the pVP16 vector containing the HSV activation domain (Liu et al., 1997 ). For the protein-binding assays, pGEX-4T PACS-1117-266, and pET3.2a furin-DD were as previously described (Crump et al., 2001 ; Simmen et al., 2005 ). The PACS-1181-195 DNA sequence was synthesized and cloned into pGEX-4T (Celtek Genes, Nashville, TN) and was used to generate PACS-1181-195N188A– and PACS-1181-195K189A–expressing plasmids by QuikChange Mutagenesis. pET-32a His6-Nef was provided by T. Smithgall (University of Pittsburgh School of Medicine, Pittsburgh, PA) and was used to generate His6-Nef-E4AW113A using QuikChange Site-Directed Mutagenesis Kit. The ARF1T31N cDNA was provided by T. Roberts (Dana-Farber Cancer Institute, Boston, MA; Addgene [Cambridge, MA] plasmid 10833; Furman et al., 2002 ) and was subcloned into pGEX-4T. For the BiFC analyses, Nef-Yc and a backbone vector containing the amino portion of YFP (Yn) were provided by T. Smithgall. The PACS sequences were inserted 5′ to the Yn fragment by standard cloning techniques. The vaccinia virus recombinant expressing FLAG-tagged Nef (VV:Nef/f) was described previously (Blagoveshchenskaya et al., 2002 ). The vaccinia virus recombinant expressing FLAG-tagged Nef-EF4W113A (VV:Nef-EF4W113A/f) was generated as previously described (Dascher et al., 1995 ). Control (nonspecific) siRNA and siRNAs specific for PACS-2 (Smartpool; Dharmacon, Boulder, CO) were transfected (Lipofectamine 2000; Life Technologies, Grand Island, NY) into cells according to manufacturer's instructions.
H9 cells or PBMCs were nucleofected (Amaxa, Gaithersburg, MD) with the plasmids indicated in the figure legends as previously described (Hung et al., 2007 ). At 40 h postnucleofection, cells were fixed in 2% paraformaldehyde, washed, and resuspended in FACS buffer (phosphate-buffered saline [PBS], pH 7.2, containing 0.5% FBS) and incubated with monoclonal antibody (mAb) BB7.2-PE (anti-HLA-A2; BD Biosciences, San Jose, CA) or mAb W6/32 (anti-MHC-I, 1:4000), followed by PE-conjugated donkey anti–mouse immunoglobulin G (1:500; Jackson ImmunoResearch Laboratories, West Grove, PA) as indicated in the figure legends. Cells were analyzed by listmode acquisition on a FACSCalibur (BD Biosciences) using CellQuest acquisition/analysis software (BD Biosciences), and data were analyzed using CellQuest or FCS express (De Novo Software, Los Angeles, CA).
Cells were transfected using Fugene (Roche, Indianapolis, IN) with the indicated plasmids for 48 h and subsequently harvested in PBS containing 1% NP-40, protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride [PMSF] and 0.1 μM aprotinin, E-64, and leupeptin) and phosphatase inhibitors (1 mM Na3VO4, 20 mM NaF). HA-tagged PACS constructs were immunoprecipitated with protein G Sepharose (Sigma-Aldrich, St. Louis, MO). The Nef/PACS-1 coimmunoprecipitates were washed in 50 mM Tris (pH 7.4), 175 mM NaCl, and 1% Nonidet P-40, 0.3% deoxycholate, and the Nef/PACS-2 coimmunoprecipitates were washed in 50 mM Tris (pH 7.4), 200 mM NaCl, 1% Nonidet P-40, and 1% deoxycholate. Coimmunoprecipitating proteins were detected by Western blot. The following antibodies were used: anti-His6 (GenScript, Piscataway, NJ), anti–FLAG mAb M2 (Sigma-Aldrich), anti–HA mAb HA.11 (Covance, San Diego, CA), anti–Nef 2949 (AIDS Research and Reference Reagent Program, National Institutes of Health, Germantown, MD), anti–PACS-1 703 (Atkins et al., 2008 ), anti–PACS-2 193 (Atkins et al., 2008 ), anti–Hck sc-72 (Santa Cruz Biotechnology, Santa Cruz, CA), and anti-actin (Chemicon, Bedford, MA).
The L40 yeast strain (MATa his3, trp, ade2, LYS::(lexAop)4-HIS3 URA::(lexAop)8-lacZ GAL4) was cotransformed with a prey plasmid (Leu− selection) and a bait plasmid (Trp− selection) as previously described (Liu et al., 1997 ) and then selected for growth on Leu−/Trp− minimal plates. Colonies from the Leu−/Trp− plates were streaked onto His− plates supplemented with 5–50 mM of 3-aminotriazole (3AT) as indicated in the figure legends.
Cells grown on coverslips were either nucleofected (Nef trafficking experiments) or transfected (BiFC experiments, Lipofectamine 2000) according to the manufacturer's instructions. Samples were washed and fixed immediately with 4% paraformaldehyde or, for BiFC analyses, were preincubated for 3 h at room temperature before fixation. Anti-TfR antibody was provided by C. Enns (Oregon Health and Science University) and was detected using an Alexa Fluor 546 secondary antibody (Life Technologies). Nuclei were stained with 4′,6-diamidino-2-phenylindole (DAPI; Dapi-Fluoromount G; Southern Biotech, Birmingham, AL).
Images were acquired on a high-resolution, wide-field Core DeltaVision system (Applied Precision, Issaquah, WA). This system is an Olympus IX71 inverted microscope (Olympus, Center Valley, PA) with a proprietary XYZ stage enclosed in a controlled environment chamber, differential interference contrast transmitted light, and a solid-state module for fluorescence. The camera is a Nikon CoolSNAP ES2 HQ (Nikon, Melville, NY). Each image was acquired as Z- stacks in a 1024 × 1024 or 512 × 512 format with a 60×, 1.42 numerical aperture, PlanApo objective in three colors—YFP, DAPI, and mcherry. The pixel size was 0.107 × 0.107 × 0.4 μm for 512 × 512 × 8 size images and 0.16 × 0.16 × 0.4 μm for 1024 × 1024 × 8 size images. The images were deconvolved with the appropriate optical transfer function using an iterative algorithm of 10 iterations. The histogram was optimized for the most positive image and applied to all the other images for consistency before saving the images as 24-bit merged TIFF files.
Images were captured on a Zeiss ELYRA PS.1 system (Carl Zeiss, Jena, Germany). Superresolution images were acquired in three colors with five rotations and reconstructed using Zen 2010 multidimensional software (Gustafsson, 2000 ). Three-dimensional (3D) image processing was performed using Imaris 7.3.1 (Bitplane, Zurich, Switzerland) to recreate volumes. Surfaces were selected based on the 3D intensity data and represented as pseudocolored images.
Statistical analyses for Figures 3C and and4G4G and the colocalization of Nef-E4AW113A-YFP with endocytic markers in Figure 7A, which are presented in Figure 7B, were compiled using Imaris contour and masking techniques. A precise region of interest (ROI) was selected, and then automated thresholding was used to calculate colocalization statistics. Colocalization was defined as the overlap in the voxels from the red and green channels. The degree of colocalization in each voxel was represented by the percentage of each channel above the threshold and determined to be significant using the unpaired Student's t test. Statistics for Figure 3A and the overlap of Rab proteins with Nef-E4AW113A-YFP–containing tubule or donut structures in Figure 7A were compiled using softWoRx Explorer 2.0 (Applied Precision). For the analysis in Figure 3A, a mean intensity reading of the BiFc (YFP) signal was obtained from the cytoplasmic ROI in Nef-Yc–positive cells (identified with anti-Nef 2949 [AIDS Research and Reference Reagent Program], followed by staining with an Alexa Fluor 546–coupled secondary antibody [Life Technologies]) and from Nef-Yc-negative cells (background). After subtraction of the averaged background signal, the specific BiFC signal was found to be significant using the unpaired Student's t test. To determine the association of mcherry-Rab5 or mcherry-Rab7 with Nef-E4AW113A-YFP–containing tubule or donut structures in Figure 7A, the structures were first selected in the green (enhanced YFP [eYFP]) channel with the red (mcherry) channel turned off. Nef-E4AW113A-eYFP–containing structures were designated as either “tubule” or “donut.” The red channel was illuminated, the mcherry-positive spots that associated with each eYFP structure were counted, and the percentage of Nef-E4AW113A-eYFP structures (tubule or donut) positive for mcherry-Rab5 or mcherry-Rab7 was determined, with the statistics presented in Figure 7C.
Plasmids expressing GST, GST-PACS-1181-195, GST-PACS-1181-195N188A, GST-PACS-1181-195K189A, GST-PACS-1117-266, GST-ARF1T31N, His6-Nef, His6-Nef-E4AW113A, or His6-furin-DD were transformed in BL21 Escherichia coli, and protein expression was induced with 1 mM isopropyl-β-d-thiogalactoside (Calbiochem, Gibbstown, NJ) for 4 h at 37°C. Bacterial pellets were resuspended in lysis buffer (50 mM Tris, pH 7.6, 1.5 mM EDTA, 100 mM NaCl, 0.5% Triton X-100, 0.1 mM dithiothreitol, 10 mM MgCl2) containing protease inhibitors (0.5 mM PMSF and 0.1 μM each of aprotinin, E-64, and leupeptin) and lysed using a French Press (Aminco, Rockville, MD). Soluble material was collected after 1 h of a 25,000 × g spin and subsequently bound to glutathione Sepharose 4B (GE Healthcare, Uppsala, Sweden) of Ni2+– nitriloacetic acid (Qiagen, Valencia, CA) for 1 h at 4°C. For the His6-Nef or His6-Furin-DD interaction with GST-PACS the proteins were mixed at a 1:1 ratio for 30 min at 4°C in binding buffer (20 mM Tris, pH 7.9, containing 150 mM NaCl, 0.1 mM EDTA, and 0.1% NP-40). After incubation, glutathione Sepharose 4B was added to the reaction and incubated for an additional 30 min at 4°C. The resin was subsequently washed three times in binding buffer and resuspended in SDS–PAGE sample buffer.
The 17-mer peptide KRDANKLQIMLQRRKRY (Creative Peptides, Shirley, NY) corresponding to PACS-1184-200 was resuspended in 10 mM potassium phosphate (pH 7.4) at a final concentration of 5 mg/ml. CD spectra were recorded in a Aviv Model 215 spectropolarimeter (Aviv Biomedical, Lakewood, NJ) in a 0.01-cm-thick rectangular quartz cell in a total volume of 30 μλ. In some experiments, the protein denaturant urea or the α-helix stabilizer TFE (Luo and Baldwin, 1997 ) was added as indicated in the figure legend. Spectra were measured at 0.25-nm intervals from 190- to 260-nm wavelength at 25°C. A background spectrum of buffer alone was measured and used for subtraction from all measured CD spectra. All spectra were averaged from three independent measurements and smoothened by locally weighted polynomial regression. The spectrum in aqueous buffer was fit by a linear combination of the spectra in 3 M urea and 30% TFE using a least-squares algorithm.
The peptide corresponding to PACS-1181-195 was docked on to the solution structure of Nef (PDB ID: 2Nef) using ClusPro 2.0 (http://cluspro.bu.edu), which automates two steps of protein–protein docking (Grzesiek et al., 1997 ; Comeau et al., 2004 ; Kozakov et al., 2010 ). The first step involves docking of two rigid bodies using a fast Fourier transformation correlation approach that exploits pairwise interaction potentials, and the second step clusters the 1000 best energy conformations, retaining the top 30 largest clusters. ClusPro 2.0 provides the results for docking based on four different coefficients: hydrophobic, electrostatic, balanced, and combined. The balanced coefficient was selected for modeling of Nef:PACS-1181-95 since the PACS-1 structure was based upon ab initio modeling and CD spectroscopy without prior knowledge of the nature of the protein:peptide complex. The subset of docking poses generated using the balanced coefficients are displayed in Figure 5A using surface rendering with 80% transparency (PACS-1181-95) and a ribbon representation (Nef), whereas the model consistent with the experimental data is depicted as a ribbon in Figure 8A. All structures were visualized using PyMOL (www.pymol.org).
We thank T. Smithgall, T. Weber, T. Roberts, C. D. Hu, and C. Enns for reagents and T. Dillon, T. Smithgall, and L. L. Thomas for critically reading the manuscript. We thank A. Snyder of the Jungers Center Advanced Light Microscopy Core (Oregon Health and Science University) and K. M. Atkins, L. L. Thomas, and L. Larsen for experimental assistance. This work was supported by Canadian Institutes of Health Research Fellowship HFE-87760, Fonds de recherche du Québec-Santé Fellowship 23037, the Collins Medical Trust (J.D.D.), National Science Foundation Grant MCB0746589 and the American Heart Association (U.S.), National Institutes of Health Grants S10-RR023432 (to the Advanced Light Microscopy Core), and DK37274 and CA151564 (G.T.).
This article was published online ahead of print in MBoC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E11-11-0928) on April 11, 2012.