Structural and functional analysis of human cytomegalovirus US3 protein

Abstract

Human cytomegalovirus (HCMV) unique short region 3 (US3) protein, a type I membrane protein, prevents maturation of class I major histocompatibility complex (MHC) molecules by retaining them in the endoplasmic reticulum (ER) and thus helps inhibit antigen presentation to cytotoxic T cells. US3 molecules bind to class I MHC molecules in a transient fashion but retain them very efficiently in the ER nonetheless. The US3 luminal domain is responsible for ER retention of US3 itself, while both the US3 luminal and transmembrane domains are necessary for retaining class I MHC in the ER. We have expressed the luminal domain of US3 molecule in Escherichia coli and analyzed its secondary structure by using nuclear magnetic resonance. We then predicted the US3 tertiary structure by modeling it based on the US2 structure. Unlike the luminal domain of US2, the US3 luminal domain does not obviously interact with class I MHC molecules. The luminal domain of US3 dynamically oligomerizes in vitro and full-length US3 molecules associate with each other in vivo. We present a model depicting how dynamic oligomerization of US3 may enhance its ability to retain class I molecules within the ER.

FIG. 1.

Structure of US3. (a) Kyte-Doolittle hydropathy plot of the US3 molecule, representing the signal peptide, luminal domain, transmembrane, and cytoplasmic tail domain. An index of 1 or higher represents a hydrophobic/transmembrane domain. (b) Silver-stained SDS-PAGE gel of US3L-His₆ after size exclusion chromatography. This sample was used for NMR analysis. (c) Sequence alignment between US2 and US3. Identical residues are indicated in red; homologous residues are indicated in green. The yellow bars above the US2 sequence indicate the positions of the β-strands. The circles below the US3 sequence indicate residues assigned by NMR; filled circles indicate residues in β-sheet conformation as confirmed by NMR. The asterisks indicate US2 residues that are involved in binding to the class I MHC molecule HLA-A2.

FIG.2.

NMR and modeling of US3L-His₆. (a) Schematic diagram of US2 and US3 secondary structures. The β-sheet assigned based on the US2 secondary structure are blue, whereas the assigned US3 segments are orange. The intramolecular disulfide bond is represented by a yellow line. (b) Ribbon structure representing a homology model of US3 based on the US2 structure. The left panel shows modeled US3 with the GFCC′ face in front, and the right panel shows the same model rotated 90° clockwise. The residues S58, E63, and K64, mutation of which abolishes ER retention, are shown in blue. The glycosylation site N60 is shown in red, and the disulfide bond is shown in yellow. (c) NMR two-dimensional ¹⁵N-¹H-correlated HSQC spectra comparing ¹⁵N-labeled US3L-His₆ in the absence (left panel) or presence (right panel) of the class I MHC molecule HLA-A2 (0.17 mM). These spectra are essentially identical.

FIG. 3.

Surface electrostatic potential maps comparing the US3 model (a) and the US2 structure (b). The positively charged regions are blue, and the negatively charged regions are red. Figures on the left panel are in the same orientation as the ribbon structure shown in Fig. 2b (left). The surfaces on the right are rotated 180° with respect to the surfaces on the left. The figure was generated by using Molmol (13).

FIG. 4.

In vitro light-scattering and cross-linking studies suggest that US3L-His₆ oligomerizes in solution. (a) In the left panel is shown an elution profile of the concentrated US3L-His₆, used in NMR spectroscopy, from the analytical sizing column. US3L-His₆ elutes in a single peak with no evidence of aggregation. In the right panel is shown the multiangle light-scattering profile of the peak shown in the left panel. US3L-His₆ scatters with molecular mass of 14.480 kDa with a 0.9% error rate, which is in agreement with the size of a monomer. (b) In vitro cross-linking of US3L-His₆ in presence of 0, 10, 100, or 1,000 μM BS³. US3L-His₆ at 1 mg/ml was incubated for 2 h with the indicated concentrations of BS³, and the samples were subjected to SDS-PAGE and Coomassie blue staining.

FIG. 5.

US3-HA molecules are functional and can oligomerize with US3 molecules in vivo. (a) Expression of US3-HA in U373-MG cells. Cells infected with pLNW/US3-HA retrovirus vector were pulsed with [³⁵S]methionine for 20 min and subjected to lysis in 1% SDS. US3 and US3-HA molecules were immunoprecipitated by using 4 μl of anti-US3 and/or anti-HA antibodies followed by SDS-PAGE and autoradiography. (b) US3-HA retains the class I MHC heavy chain molecules in the ER. U373 (control), US3⁺, and US3-HA⁺ cells were pulsed with [³⁵S]methionine for 20 min and chased for 90 min. Cells were lysed as done for panel a, and 5 μl of anti-HC antibody was added to each lysate to immunoprecipitate the class I MHC heavy-chain molecules. Samples were treated with Endo H and subjected to SDS-PAGE and autoradiography. (c) US3 and US3-HA interact in HEK-293 cells. HEK-293 cells were transiently transfected with pcDNA3.1(+) vectors expressing US3, US3-HA, or both. Cells were pulsed with [³⁵S]methionine for 20 min and lysed in the presence of 1% digitonin. US3 and US3-HA molecules were immunoprecipitated by addition of 3 μl of the relevant antibodies and were subjected to SDS-PAGE and autoradiography.

FIG. 6.

US3 oligomerization is independent of the residues involved in ER retention and occurs even when US3 is stably expressed. (a) In vivo cross-linking shows association between US3 and US3-HA. Cells expressing both US3 and US3-HA were pulsed for 1 h with [³⁵S]methionine and incubated with or without DSP for 1 h. Cells were then lysed in 1% SDS, and the lysate was diluted 10-fold with NP-40 buffer. The lysate was divided equally and subjected to immunoprecipitation with 4 μl of the indicated antibodies, followed by SDS-PAGE and autoradiography. (b) US3-HA* also associates with the US3 molecules. Cells were pulsed with [³⁵S]methionine for 45 min and lysed with 1% digitonin. The lysates were normalized for equal amounts of radioactivity, and 4 μl of the anti-HA antibody was used to immunoprecipitate US3-HA molecules. The precipitate was resuspended in 1% SDS and boiled to disrupt all interactions. The samples were diluted as described above and subjected to reimmunoprecipitation with 3 μl of the anti-US3 antibody, followed by SDS-PAGE and autoradiography.

FIG. 7.

Schematic model depicting how US3 oligomerization may enhance its function. US3 molecules (purple) bind to the class I MHC trimeric complex (red) via their transmembrane domains (represented by blue and red arrows). The interaction between US3 and class I MHC transmembrane domains is transient. US3 retains itself in the ER by interacting weakly with a putative ER receptor protein X (green) through its Ser58, Glu63, and Lys64 residues (represented by green arrows). The model illustrates dynamic oligomerization of US3 molecules via their interacting luminal domains in the ER (represented by blue arrows). Dynamic oligomerization of US3 molecules may enhance their ability to retain class I MHC molecules. More US3 transmembrane domains in proximity to class I transmembrane domain may make it more difficult for class I MHC molecules to escape the ER. In addition, by recruiting putative ER receptor protein X to the site of retention, both US3-US3 interactions and US3-class I MHC interactions may become more stable due to minimization of lateral diffusion within the ER membrane.