Probing the Structure of Rotavirus NSP4: a Short Sequence at the Extreme C Terminus Mediates Binding to the Inner Capsid Particle

Abstract

The rotavirus nonstructural glycoprotein NSP4 functions as the receptor for the inner capsid particle (ICP) which buds into the lumen of the endoplasmic reticulum during virus maturation. The structure of the cytoplasmic domain of NSP4 from rotavirus strain SA11 has been investigated by using limited proteolysis and mass spectrometry. Digestion with trypsin and V8 protease reveals a C-terminal protease-sensitive region that is 28 amino acids long. The minimal sequence requirements for receptor function have been defined by constructing fusions with glutathione S-transferase and assessing their ability to bind ICPs. These experiments demonstrate that 17 to 20 amino acids from the extreme C terminus are necessary and sufficient for ICP binding and that this binding is cooperative. These observations are consistent with a model for the structure of the NSP4 cytoplasmic region in which four flexible regions of 28 amino acids are presented by a protease-resistant coiled-coil tetramerization domain, with only the last ∼20 amino acids of each peptide interacting with the surface binding sites on the ICP.

Rotavirus is the single most important cause of severe, dehydrating diarrhea in young children worldwide and accounts for as many as 125 pediatric deaths per year in the United States and 873,000 childhood deaths each year in developing countries (10). The mechanism by which rotavirus induces intestinal secretion of fluid and electrolytes is unknown, but proposals include destruction of enterocytes in the small intestine (11), alterations in transepithelial fluid balance (5), a toxin-like effect by a viral nonstructural protein, NSP4 (3), or activation of the enteric nervous system (15). The virus particle consists of a three-layer protein capsid that contains a segmented, double-stranded RNA genome. The assembly of mature rotavirus involves a unique budding process in which cytoplasmically assembled immature virus particles bud into the lumen of the rough endoplasmic reticulum (ER). The transfer is mediated by an interaction between the double-layered rotavirus inner capsid particle (ICP; approximate molecular mass, 5 × 10⁷ Da) and the cytoplasmic tail of NSP4 (1, 2, 19, 24). This region of NSP4 contains a putative coiled-coil stem structure and a C-terminal receptor domain that is sensitive to proteolysis (26), but the precise size of the domain that interacts with the ICP and the number of interactions involved remain to be established.

Here, we have used a combination of limited proteolysis and mass spectrometry to probe the structure of the C-terminal region of NSP4. The minimal receptor domain and characteristics of the NSP4-ICP interaction have been defined by using site-directed mutagenesis and quantitative analysis of binding kinetics. The results demonstrate that 28 residues form an exposed C-terminal domain, of which the last 17 to 20 (amino acids 156 to 175) are required for binding of the ICP to the receptor. In solid-phase binding assays that reflect some aspects of the in vivo interaction, multiple receptors appear to interact with the ICP surface.

Protease digestion of the cytoplasmic domain of NSP4.

Previous biophysical studies of the cytoplasmic domain of NSP4 showed that the protein is a tetramer in which a putative coiled-coil domain forms the subunit interface. This region was found to be resistant to limited trypsin digestion while, under identical conditions, the C terminus was removed (26). We refined this analysis to probe the structure of NSP4 with emphasis on the size of the protease-sensitive C-terminal region while monitoring the integrity of the coiled-coil domain.

This was achieved by expressing cDNA encoding the C-terminal 90 amino acids of the cytoplasmic domain as a glutathione S-transferase (GST) fusion protein (GSTC90) in Escherichia coli DH5α cells (23, 25) by using the pGex-2TK vector (Pharmacia), which incorporates a kinase site adjacent to the thrombin cleavage site. Fusion protein adsorbed to 500 μl of glutathione-agarose beads (Sigma) was washed with protein kinase buffer (20 mM Tris-HCl [pH 7.5], 100 mM NaCl, 12 mM MgCl₂) and radiolabelled by addition of 150 μl of protein kinase reaction mixture (protein kinase buffer containing 50 U of bovine heart kinase [Sigma] in 40 mM dithiothreitol and 50 μCi of [γ-³²P]ATP). The mixture was vortexed and incubated for 30 min at 4°C. The reaction was terminated by the addition of 5 ml of stop solution (10 mM sodium phosphate [pH 8.0], 10 mM sodium pyrophosphate, 10 mM EDTA, 1 mg of bovine serum albumin per ml). The beads were centrifuged at 500 × g for 2 min and washed extensively with 0.1 M Tris-HCl (pH 8.0). The radioactively labelled fusion protein was then washed with thrombin cleavage buffer (50 mM Tris-HCl [pH 8.0], 150 mM NaCl, 2.5 mM CaCl₂) and incubated with 2 U of bovine thrombin (Sigma) per mg of fusion protein for 90 min at 37°C with shaking. After centrifugation, the supernatant was collected, concentrated by using a 3.5-ml Microsep concentrator (10K cutoff; Filtron Technology Corporation), and incubated with glutathione-agarose beads on ice for 10 min to remove contaminating GST. Thrombin was removed by incubating the preparation with para-aminobenzamidine–agarose beads (Sigma) on ice for 10 min. This procedure yielded the cytoplasmic domain of NSP4 (C90) with a ³²P label at the N terminus.

Purified receptor protein (10 μg) in thrombin cleavage buffer was incubated with protease (0.1 μg) in a total volume of 20 μl at temperatures ranging from 0 to 32°C. Proteolysis was terminated by precipitation of digested protein with cold 10% trichloroacetic acid (TCA) followed by centrifugation for 5 min. The precipitated protein was dissolved in Laemmli loading buffer, neutralized with 1 M Tris (pH 11), and boiled for 3 min prior to analysis by gel electrophoresis as described by Schägger and von Jagow (22). Limited digestion of radiolabelled C90 with either trypsin or V8 protease produced smaller fragments (∼8 kDa) that retained the radioactive label, indicating that the most accessible cleavage sites are those located in the C-terminal region of the molecule (Fig. 1).

FIG. 1.

Limited proteolysis of ³²P-labelled C90 by using trypsin or V8 protease. ³²P-labelled C90 (10 μg) in thrombin cleavage buffer was digested with 0.1 μg of trypsin at 0°C (A) or V8 protease at 32°C (B) for 0 to 60 min. Reactions were terminated by precipitation of the protein with ice-cold 10% TCA followed by centrifugation for 5 min, and the digestion products were resolved by SDS-PAGE. Gels were dried and exposed to X-ray film for 1 h to produce autoradiographs. Lane 1, nonradioactive protein size markers (Sigma); lane 2, undigested C90; remaining lanes, C90 digested for the time indicated.

Mass spectrometry of C90 and its digestion products.

Mass spectra for undigested and digested C90 (Fig. 2) were obtained by using the matrix-assisted laser desorption/ionization (MALDI) process (13). Preparations of the C90 variant or C90 limited digestion products (25 μg) were purified by reversed-phase high-performance liquid chromatography (HPLC) using a C₈ Brownlee cartridge column (Applied Biosystems) and elution in a 10 to 70% acetonitrile gradient containing 0.08% trifluoroacetic acid. Samples were mixed in a 1:1 ratio with sinapinic acid (3,5-dimethoxy-4-hydroxycinnamic acid; Hewlett-Packard Corp.), vortexed briefly, and loaded onto individual mesas on the tip of an inert probe. Peptide standards with known molecular masses (goosefish angiotensin I, 1,281.49 kDa; somatostatin, 1,637.90 kDa; human insulin, 5,807.7 kDa; equine cardiac cytochrome c, 12,359 kDa; Hewlett-Packard Corp.) were analyzed in the same fashion. Molecular masses were determined in a Hewlett-Packard G2025A time-of-flight mass spectrometer by comparison of the spectra for unknown samples with those obtained for standards (G2025A Software A.01.00; Hewlett-Packard Corp.). Undigested C90 yielded a molecular mass of 11,649 Da (cf. predicted monomeric value of 11,657 Da), whereas trypsin-digested C90 contained four distinct fragments (unresolved in Fig. 1) ranging in size from 8,372 to 8,903 Da. The spectrum for the major V8 protease-resistant product indicated a molecular mass of 8,347 Da.

FIG. 2.

Mass spectrometry of proteolytic fragments of NSP4. Purified C90 (50 μg) was digested with trypsin (0.5 μg) for 2 h at 0°C or V8 protease (0.5 μg) for 60 min at 32°C. The reactions were terminated by the addition of protease inhibitors (0.5 mM Pefabloc for trypsin and 5 μM dichloroisocumarin for V8 protease). Mass spectra for digested and undigested samples were obtained by MALDITOF mass spectrometry. Shown are undigested C90 (A), trypsin-digested C90 (B), and V8 protease-digested C90 (C).

The origins of the protease-resistant NSP4 fragments generated by trypsin and V8 protease digestion were then deduced by inspection of the amino acid sequence of C90 (Fig. 3). The N-terminal serine residue bearing the radioactive phosphate label (bold type in Fig. 3) clearly was retained because all the major proteolytic fragments were detectable by autoradiography. Trypsin cleavage could occur at the arginine residues in the kinase labelling site as well as at the C-terminal cleavage sites. Potential V8 protease cleavage sites are confined to the C terminus.

FIG. 3.

Origins of protease-resistant NSP4 fragments. The origins of protease-resistant fragments of C90 generated by limited digestion with trypsin or V8 protease were deduced from the amino acid sequence of NSP4 (4). The sizes determined for these fragments by mass spectrometry (Fig. 2) were compared with the sizes of C90 peptides terminated at known trypsin (solid arrows) and V8 protease (open arrows) cleavage sites and calculated by using GPMAW software (Hewlett-Packard Corp.). The top bar represents the cytoplasmic domain of NSP4, and the numbers refer to amino acid positions in the full-length NSP4 sequence of rotavirus strain SA11 (shown in part). Additional residues derived from the pGex-2TK vector are shown at the N terminus, and residues inferred to be sensitive to limited proteolysis are indicated (∗). Boundaries of the α-helical region are those identified by Taylor et al. (26). The small bars represent the protease-resistant fragments, with their predicted molecular masses, which agreed (± 0.5%) with those measured by mass spectrometry.

The predicted sizes of four trypsin-resistant fragments terminating at either Lys 151 or Arg 154 agreed closely (±0.5%) with the molecular masses of the proteolytic fragments obtained by mass spectrometry. Small peptides corresponding to C-terminal tryptic fragments also were found (data not shown), indicating that the C-terminal region that spans residues 151 to 175 is sensitive to trypsin under conditions of limited digestion. The size of the major V8 protease-resistant product (8,347 Da) is consistent with the molecular mass expected for a fragment cleaved at Glu 147 (8,384 Da). Other possible cleavages yield fragments with molecular masses of 7,895 Da (cleavage at Asp 143) or 9,580 Da (cleavage at Glu 157), but fragments of these sizes were not detected. N-terminal sequencing (18) and mass spectrometry of the products released from C90 by V8 protease digestion revealed the presence of C-terminal fragments corresponding precisely to amino acids 148 to 175, 148 to 160, 148 to 157, and 161 to 175 (Table 1). Thus, it is clear from the results of limited digestion with V8 protease that residues 147 to 175 are exposed.

TABLE 1.

Identities of C-terminal fragments of NSP4 produced by limited V8 protease digestion

Fragmenta	N-terminal sequence	Predicted molecular mass (Da)	Observed molecular mass (Da)
148-175	INQ …	3,290.7	3,285.4
148-160	Not determined	1,658.8	1,657.0
148-157	Not determined	1,214.4	1,215.0
161-175	SGK …	1,649.8	1,650.4

^aNumbers refer to amino acid positions in the wild-type SA11 NSP4 sequence (4). 

When the results of limited proteolysis with both enzymes are considered, it is clear that Lys 146 represents the C-terminal boundary of the protease-resistant region of the cytoplasmic domain, because this residue is resistant to trypsin digestion, whereas Glu 147 is sensitive to cleavage with V8 protease. Overall, this result suggests that the C-terminal 28 amino acids of NSP4 adopt a less ordered conformation that renders this region of the protein more accessible to proteolytic enzymes.

ICP-binding activity of truncated variants of NSP4.

Previous studies of the binding of NSP4 and rotavirus ICPs revealed the importance of the extreme C terminus for receptor activity (2, 24) and demonstrated that GST fusion proteins containing as little as the C-terminal 20 amino acids retain the ability to bind ICPs when immobilized on the surface of glutathione-agarose beads (25, 26). To assign a minimal ICP-binding domain more confidently, further variants retaining fewer C-terminal amino acids were constructed. A C16 variant was constructed by using the protocol described previously for the C90 to C20 variants (26). The C10, C12, and C14 variants were constructed by annealing complementary oligonucleotides encoding the corresponding amino acids of NSP4 flanked by 5′ BamHI and 3′ EcoRI sites. These cDNA fragments were cloned into the pGEX-2T vector (Amersham Pharmacia Biotech), and the identities of the constructs were confirmed by DNA sequencing using an Applied Biosystems automated sequencer.

GST fusion proteins were purified from cultures of E. coli DH5α cells (50 ml) expressing the receptor variants as follows. Cultures were centrifuged (3,500 rpm, 10 min, 4°C, Sorvall RT7 benchtop centrifuge), and cell pellets were resuspended in buffer (2 ml) containing 25 mM Tris-HCl (pH 8.0), 10 mM EDTA, and 50 mM glucose. The cell suspension was sonicated (10 s, Branson sonicator, level 6) after the addition of DNase (50 μg/ml), RNase (50 μg/ml), lysozyme (100 μg/ml), and Pefabloc [4(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride; Boehringer Mannheim; 0.5 mM]. Cell debris was pelleted by microcentrifugation for 1 min. Cell lysates were then incubated with glutathione-agarose beads (100 μl) on ice for 30 min, after which fusion proteins were eluted with 10 mM glutathione (200 μl) in 100 mM Tris-HCl (pH 8.0) for 30 min at room temperature. Dilutions (1/50) of the eluted fusion protein preparations in 100 mM Tris-HCl (pH 8.0) were adsorbed to the wells of a microtiter plate, and the receptor activity of the adsorbed protein was measured by enzyme-linked immunosorbent assay as described previously (25). The relative levels of adsorbed receptor were compared by detecting adsorbed fusion protein by using a rabbit polyclonal anti-GST antiserum. As anticipated, fusions incorporating 90, 44, or 20 amino acids exhibited strong binding activity (Fig. 4). In contrast, no detectable binding activity was found for the C10 to C16 fusions even though the levels of adsorbed receptor varied by no more than ±11% of the mean level for all variants (data not shown). This result indicates that the minimal receptor domain consists of 17 to 20 amino acids at the extreme C terminus. Truncated variants in which methionine 175 was mutated to isoleucine were also receptor negative, confirming the result demonstrated earlier that, in the full-length protein, the C-terminal methionine residue is required for ICP-binding activity (24).

FIG. 4.

Receptor activity of GST-NSP4 fusions. GST fusion proteins were purified from cultures of cells (50 ml) expressing N-terminally truncated variants of the cytoplasmic domain of NSP4. The receptor activity for each variant was measured by enzyme-linked immunosorbent assay. ∗, variants in which the C-terminal methionine was mutated to isoleucine. The results represent the average of duplicate measurements, and error bars indicate the range measured for duplicate samples. OD, optical density.

The dissociation constant for NSP4-ICP binding was measured for mutants C90 to C20 by Scatchard analysis of the ICP-binding activity of GST-NSP4 fusions immobilized on the surface of glutathione-agarose beads. Receptor assays used ¹²⁵I-labelled ICPs and glutathione-agarose-bound fusion protein (25). Bound-to-free ratios were calculated from the ratio of bound counts to input counts, and the concentration of bound ICPs was calculated from the counts bound and the specific radioactivity of the labelled ICPs. A value of 5 × 10⁷ Da was used for the molecular mass of the ICP that was based on the molecular mass of the reovirus core particle (6). All the Scatchard plots were concave downwards rather than linear (Fig. 5, left panels), indicating that the binding event involved positive cooperativity (20). The binding data was further analyzed by using Hill plots (Fig. 5, right panels). The fraction of binding sites filled (Y) was calculated as the ratio of counts bound to the maximum counts bound in the assay. The dissociation constant (K_d) was obtained from the intercept of the x axis {log [Y/(1−Y)] = 0}, and the Hill coefficient n was obtained from the slope of the plot. The parameters of the linear fit confirmed the cooperative nature of the binding of ICPs to bead-immobilized receptor (7). Derived values for K_d and the Hill coefficient n are summarized in Table 2.

FIG. 5.

Scatchard and Hill plot analysis of ICP-binding activity of bead-immobilized GST-NSP4. (Left) ICP-binding activity was measured for each variant of NSP4, but only results for C90 and C20 are shown. Assays contained 10 μl of bead-bound receptor and 0.3 to 15 μg of ¹²⁵I-labelled ICPs (0.3 μg/μl, 1.5 × 10⁴ to 2.1 × 10⁴ cpm/μg). For Scatchard analysis, bound-to-free ICP ratios were calculated from the ratio of bound counts to input counts for the assay. The concentration of bound ICPs was calculated from the counts bound and the specific radioactivity of the labelled ICPs. (Right) Data for Scatchard analysis of ICP-binding activity (left) were further analyzed to generate Hill plots. Fractional saturation (Y) for each point was estimated from the ratio of counts bound to the maximum counts bound in the assay. For all plots, lines of best fit were applied to the data points.

TABLE 2.

ICP binding by variants of NSP4

Variant	Kd (M)	Hill coefficient n
C90	7.83 × 10−11	1.55
C78	6.95 × 10−11	1.66
C69	7.18 × 10−11	1.53
C53	6.43 × 10−11	1.57
C44	6.52 × 10−11	1.65
C32	7.68 × 10−11	1.52
C20	7.51 × 10−11	1.69

The K_d results confirm that each variant displayed comparable ICP-binding ability. Hill coefficients of >1 imply that binding of one ligand molecule facilitates binding of others to the same receptor. For the NSP4-ICP interaction, the values derived for the Hill coefficients indicate that when ICPs bind to bead-immobilized receptor, the binding of the first virus particle positively influences the next binding event. A likely mechanism is that the binding of the immobilized receptor to the first site on the surface of the ICP increases the effective ICP concentration for subsequent interactions.

Isolation of NSP4-ICP complexes.

NSP4-ICP complexes were obtained by incubating ICPs with bead-immobilized ³³P-labelled GSTC90 fusion protein and then releasing receptor-ICP complexes from beads by thrombin cleavage. Briefly, glutathione-agarose beads (50-μl packed volume) were washed twice with receptor assay buffer (10 mM Tris-HCl [pH 7.0], 150 mM NaCl, 5 mM MgCl₂, 5 mM CaCl₂) and incubated on ice for 45 min with 100 μg of ³³P-labelled GSTC90 or its Met175→Ile equivalent in the same buffer (radiolabelling was performed as described above for ³²P-labelled fusion protein). The beads were pelleted by centrifugation for 5 s, washed twice, and incubated with ICPs (100 μg) for 90 min at room temperature in receptor assay buffer containing 0.1% octyl glucoside (final volume, 400 μl). The samples were then centrifuged, the supernatants were discarded, and the beads were washed three times with thrombin cleavage buffer. NSP4-ICP complexes were released from the beads by incubation with thrombin (2 U) in 85 μl of thrombin cleavage buffer for 2 h at 37°C. The beads were removed by centrifugation (700 × g, 10 s), and the supernatants were centrifuged through MicroSpin S400 columns (Amersham Pharmacia Biotech) for 2 min at ∼700 × g to resolve the complexes from unbound NSP4.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis of the complexes (Fig. 6) showed that the GSTC90-ICP sample (lane 2) contained Coomassie-stained bands that comigrated with either those of the ICP (lane 1) or C90 (lane 6). The 14-kDa band was identified as bound receptor by autoradiography (Fig. 6, lower panel). The amount of NSP4 associated with the ICP was quantified from the absorbency of the ICP (lanes 1 and 2) and C90 (lanes 2 and 6) bands by using the NIH Image program. The calculation was based on the assumption that the molecular mass of the rotavirus ICP is close to that of the reovirus core particle (5 × 10⁷ Da [6]). This analysis yielded a value of approximately 40 C90 tetramers per particle. However, given the likelihood of steric hindrance of binding, this figure almost certainly represents only the level possible with bead-immobilized receptor rather than true saturation of all the binding sites on the surface of the ICP. Decoration of ICPs with soluble rather than immobilized NSP4 would eliminate this problem, but specific binding could not be demonstrated when soluble receptor was incubated with ICPs. This failure may reflect a lower affinity of binding for soluble receptor and is consistent with the inability of soluble variants shorter than C90 to block binding of ICPs in solid-phase assays (data not shown) as well as with the suggested mechanism of positive cooperativity. The isolation of complexes containing approximately 40 tetrameric C90 molecules per ICP does, however, have implications for the nature of potential receptor binding sites on the surface of the ICP. The icosahedral symmetry of rotavirus (21, 27) dictates that it possesses 12 vertices, 20 faces, and 30 edges (five-, three-, and twofold axes of symmetry, respectively). If 40 or more molecules bind to each particle, then NSP4 probably does not bind to the 12 fivefold vertex positions or to the 20 threefold faces of the particle. However, if the receptor binding sites are located on twofold symmetry axes (the 30 edges plus 390 local twofold axes between the VP6 trimers on each face), the decoration protocol using bead-immobilized C90 may have achieved an occupancy of only approximately 10% of the available binding sites.

FIG. 6.

Analysis of NSP4-ICP complexes. ICPs were incubated with bead-immobilized ³³P-labelled GSTC90 or its Met175→Ile equivalent (GSTC90∗). After centrifugation of the receptor-ICP complexes through MicroSpin S400 columns, the eluates were incubated with ice-cold 10% TCA. The precipitated proteins were pelleted by centrifugation and analyzed by SDS-PAGE. The Coomassie-stained gel (upper panel) was dried and exposed to X-ray film overnight (lower panel). Lane 1, SA11 ICPs (10 μg); lane 2, C90-ICP eluate; lane 3, C90-no ICP eluate; lane 4, C90∗-ICP eluate; lane 5, C90∗-no ICP eluate; lane 6, ³³P-labelled C90 (4.4 μg). The arrowed band corresponds to the anticipated position of C90 (lane 6).

The results of this study confirm that the C terminus of NSP4 is more sensitive to proteolysis than is the rest of the cytoplasmic domain, suggesting that this region of the protein may be disordered and flexible. Sequence alignments of NSP4 genes from human and animal rotavirus strains (14) indicate that the C-terminal 20 amino acids are reasonably well conserved, and circular dichroism spectroscopy has demonstrated the presence of a high proportion of random conformation in this region of the protein (26). A number of potential trypsin and protease V8 cleavage sites are contained within the upstream α-helical region (Fig. 3, residues 95 to 137) but presumably are protected by the high degree of secondary structure in this region.

Our solid-phase binding assays indicate that the functional ICP binding domain of NSP4 requires at least 17 of the C-terminal 20 amino acids (156 to 175), including the final methionine residue. Somewhat conflicting results have been reported by Au et al. (2), who measured rotavirus ICP binding by using membranes from Spodoptera frugiperda cells containing NSP4 mutants. These workers found that loss of residues 161 to 175 abolished ICP binding activity, while deletion of the last three amino acids (173 to 175) diminished but did not abolish ligand binding. Since Met175→Ile versions of each N-terminal truncation consistently reinforced the requirement for the C-terminal methionine residue, the result obtained by Au et al. is difficult to reconcile with the results obtained in this study.

Quantitative analysis of the binding of ICPs to GST-NSP4 fusion proteins immobilized on the surface of glutathione-agarose beads showed comparable levels of binding activity for all receptor-positive variants. The K_d for the interaction was estimated to be on the order of 7 × 10⁻¹¹ M, which agrees closely with the figure of 5 × 10⁻¹¹ M previously measured for membrane-anchored, full-length NSP4 expressed in eukaryotic cells (19). Although the solid-phase binding assays used here are unlikely to achieve saturation of the ICP with receptor, they nevertheless do reflect the situation likely to occur in vivo, in which multiple NSP4 molecules are anchored in the ER membrane with their C-terminal cytoplasmic domains exposed and available for interaction with the ICP. According to the model first proposed for animal enveloped viruses by Garoff and Simons (8), the budding process is thought to be driven by multiple interactions between the receptor and the binding sites on the surface of the particle. Furthermore, lateral interactions between peripheral or integral membrane proteins seem now to be a general feature that drives the budding processes of different viruses (9). For rotavirus, NSP4 has been shown to occur in membranes associated with two other viral proteins, VP7 and VP4 (16). Thus, lateral interactions between all three proteins might supplement the positive cooperativity demonstrated here for the NSP4-ICP interaction and together drive the transfer of the immature virus particle across the ER membrane.

In summary, this work supports a model for NSP4 in which a disordered C-terminal domain of 28 amino acids is presented to the cytoplasm by each subunit of the tetrameric receptor. Only the last 17 to 20 residues of this region interact with the binding sites on the ICP. Future access to a high-resolution structure of the ICP (B. McClain, S. C. Harrison, and A. R. Bellamy, unpublished observations) should provide the means by which the precise molecular details of this unusual receptor interaction might be clarified. Of particular interest will be the reasons for the relatively precise size limitation for the ICP-binding domain and the requirement for the C-terminal methionine residue. Features of this interaction may reflect those of others known to involve large proteins and short peptide domains—the presentation of peptide antigens in the cleft of the class I major histocompatibility protein (17) and the binding of the absolute C-terminal ends of target molecules to the PDZ domains of membrane proteins (12) are two well-known examples.