Structure of the Receptor Binding Domain of EnvP(b)1, an Endogenous Retroviral Envelope Protein Expressed in Human Tissues

Organisms can access genetic and functional novelty by capturing viral elements within their genomes, where they can evolve to drive new cellular or organismal processes. We demonstrate that a retroviral envelope gene, EnvP(b)1, has been maintained and its fusion activity preserved for 40 to 71 million years. It is expressed as a protein in multiple healthy human tissues. We determined the structure of its inferred receptor binding domain and compared it with the same domain in modern viruses. We found a common conserved architecture that underlies the varied receptor binding activity of divergent Env genes. The modularity and versatility of this domain may underpin the evolutionary success of this clade of fusogens.

mutations that erode the coding potential of individual genes and render these loci incapable of producing infectious viruses. However, some ERV genes can retain their function in an otherwise degraded provirus, an indication that selective pressures preserve the function of these genes. Over time, ERV genes can be coopted to perform beneficial or essential host functions. For example, the cell-cell fusion events that produce the placental syncytiotrophoblast layer in eutherian mammals are driven by multiple independently acquired and independently coopted ERV envelopes (Envs), known as syncytins (2). Primate genomes harbor a number of ERV Env open reading frames (ORFs) with unknown functions (3,4). Many of these ORFs have been maintained for the duration of primate evolution, time periods exceeding 40 million years.
The human genome contains four copies of ERV-P(b), located on different chromosomes; one of these, located within an intron of the RIN3 gene, retains an intact ORF termed EnvP(b)1 (5). The intact EnvP(b)1 ORF has been identified in Old World primates and must have arisen as the result of integration of a then-extant virus into the RIN3 locus of a common primate ancestor. The primate EnvP(b)1 ORF is reportedly evolving under purifying selection, and various dating methods estimate EnvP(b)1 to be between 30 and 55 million years old (5,6). EnvP(b)1 has the characteristics of a classical gammaretroviral envelope protein. All the sequence motifs necessary for processing and cell surface expression remain intact (5,7). When ectopically expressed, human EnvP(b)1 can mediate fusion of cells in culture (7,8). EnvP(b)1 RNA has been detected in human tissues; however, protein expression and processing have not been previously reported (5,6). It is unclear what function(s) EnvP(b)1 may have that would explain the fact that it has maintained a functional ORF over tens of millions of years.
Here, we show that EnvP(b)1 protein is expressed in multiple human tissues in a form that appears to be fusion competent. We establish that EnvP(b)1 is more ancient than previously reported by identifying orthologs in the RIN3 locus of apes, Old World monkeys and New World monkeys (simian species). Its absence in the syntenous locus of prosimian species indicates that EnvP(b)1 was acquired between 40 and 71 million years ago (MYA). The membrane fusogen activity of EnvP(b)1 has been preserved for the duration of simian evolution. The host factors required for fusion are conserved and are ubiquitously expressed in the cultured cells of primates. Thus, an activated, functional, ancient fusogen of unknown function is expressed in multiple healthy human tissues. We have determined the structure of the EnvP(b)1 inferred receptor binding domain and compared it with available structures of the homologous domain from extant retroviruses. Their similarities define a common structural architecture that has been retained, despite selection for different functions and divergence over timescales greatly exceeding 40 to 71 million years. Modularity within this domain demonstrates the flexibility of the architecture of this domain as a versatile scaffold for receptor binding.

RESULTS
EnvP(b)1 protein is expressed in human tissues. EnvP(b)1 is the only remaining intact ORF in the ERV-P(b) provirus in the human RIN3 locus, ERV-P(b)1. All other ERV genes at this locus have been disrupted. We produced a phylogeny to describe the evolutionary relationships of EnvP(b)1 with other retroviral Envs (Fig. 1A). EnvP(b)1 occupies a clade of viruses that comprises endogenous mammalian and extant avian examples. Based upon its phylogenetic relationships with other Envs and characteristic motifs in its surface (SU) and transmembrane (TM) domains, EnvP(b)1 has been classified as a gamma-type Env (5). Notable elements include an intact furin cleavage site and a CXXC/CX 6 CC disulfide bond motif (Fig. 1B).
EnvP(b)1 RNA has previously been detected in human tissues (5,7). To determine whether EnvP(b)1 protein is also present, we expressed the genetically unique (no similar sequences in human or rabbit genomes) inferred receptor binding domain (iRBD) and used this iRBD protein to immunize a rabbit and produce antiserum. We enriched for EnvP(b)1 iRBD-binding IgGs with a series of affinity resins and confirmed by Western blotting that the enriched iRBD antisera detected EnvP(b)1 in transfected 293T cells (Fig. 1C). Two major bands were observed, one corresponding to the predicted molecular weight of unprocessed EnvP(b)1 SU-TM and a second corresponding to the furin-processed SU domain. The specificity of the SU band was confirmed by expressing EnvP(b)1 in which the furin cleavage site was mutated [EnvP(b)1 CSϪ ]. For this mutant, only the band corresponding to the unprocessed Env was observed. The reagent is specific for EnvP(b)1, as it failed to detect human syncytin-1 (HERV-W Env), and Western blots of mock-transfected 293T lysate did not show bands in the appropriate locations.
To determine whether EnvP(b)1 protein is expressed in healthy human tissues, we assayed tissue lysates by Western blotting under reducing conditions (Fig. 1D). We chose tissues that were similar to those in which RNA transcripts were previously detected (5,7). Protein expression was strongest in placenta, ovary, and thymus. In these tissues, we detected a doublet band for SU that was most prominent in placenta. This may represent an additional proteolytic processing event or glycosylation heterogeneity. EnvP(b)1 protein was also detected at lower levels in testis and small intestine. A band of a different molecular weight was present in spleen lysate, which we therefore cannot conclude corresponds to EnvP(b)1 protein. Tissues from both healthy males and females showed clear EnvP(b)1 expression, indicating that expression is not restricted to females (Table S1). We did not observe a higher-molecular-weight species corresponding to unprocessed SU-TM in any tissue, suggesting that EnvP(b)1 is efficiently processed in these tissues. In contrast, overexpression of EnvP(b)1 in 293T cells produced primarily SU-TM (Fig. 1C).
Under nonreducing conditions SU remains covalently linked to TM via a disulfide bond, and we observed EnvP(b)1 as SU-TM and higher-molecular-weight species (Fig. S1A). In tissue lysates under nonreducing conditions, EnvP(b)1 was mainly detected as SU-TM and these higher-molecular-weight species (Fig. S1B). In these human tissues, SU-TM are covalently linked and EnvP(b)1 is likely to be in a state that is primed for fusion.
EnvP(b)1 ORF is conserved in simian primates. The EnvP(b)1 ORF has been identified previously in the RIN3 locus of select ape and Old World monkey genomes (5). We surveyed primate genomes in publicly available databases for the EnvP(b)1 ORF. We identified 35 orthologs in simian species (New and Old World monkeys and apes). The presence of each within the RIN3 locus was confirmed by mapping flanking sequences outside the ERV locus. The phylogenetic relationships between the simian EnvP(b)1s generally follow the same clustering as their hosts (Fig. 2). This pattern of branching is consistent with insertion of an ancestral ERVP(b)1 into the RIN3 locus of a common ancestor of all extant simian primates, followed by divergence of the Env sequences together with their respective host species. Based on the reported divergence times of New World and Old World monkey lineages (9-11), the EnvP(b)1 insertion arose at least 40 to 47 MYA, considerably earlier than previous synteny-based estimates (5).
We were unable to identify the EnvP(b)1 ORF in prosimian lineages (tarsiers, lemurs, and lorisoids). We were also unable to find evidence of the ERV (or solo long terminal repeat) in prosimian genome assemblies of the RIN3 locus. These observations suggest that the EnvP(b)1 insertion in simians occurred after their divergence from prosimians EnvP(b)1-related sequences, of any length or coding capacity, were also absent in all but one prosimian species, aye-ayes. We identified a number of short reads from an incomplete genome, which were similar to EnvP(b)1 and the surrounding ERV (Fig. S2B). These could not be unambiguously assigned to a specific ERV-P(b) locus. How these sequences entered the aye-aye genome is unknown. Aye-ayes diverged from other extant lemurs ϳ55 MYA (9-12), shortly after lemurs colonized Madagascar and became geographically isolated between 55 and 75 MYA (13,14). EnvP(b)1-related sequences were also absent in all publicly available databases of mammalian, avian, reptilian, amphibian, piscine, and insect genomes.
EnvP(b)1 orthologs fuse cells of diverse tissue and species origin. EnvP(b)1 has evolved under at least three consecutive stages comprising distinctly different selective pressures: (i) as the Env protein of a circulating virus prior to endogenization, 40 to 71 MYA; (ii) as a captured ERV during the process of cooption by the host; and (iii) as a gene under purifying selection in all simian species. The fact that the EnvP(b)1 ORF has been retained throughout simian evolution suggests that it may perform a host function(s). Expression of human EnvP(b)1 has been shown to drive the formation of multinucleated syncytia in cultured cells, an indication of cell-cell fusion (7,8). We examined the expression, proteolytic processing, and ability to mediate cell-cell fusion of 11 divergent simian EnvP(b)1 orthologs that were expressed from cells of human, Old World monkey, and New World monkey origin. All EnvP(b)1s were expressed in all cell lines, as shown by Western blotting (Fig. S3). Similar to our observations in 293T cells, and unlike in human tissue lysate, the dominant form of the protein was unprocessed SU-TM. EnvP(b)1 orthologs from New World monkeys (marmoset and owl monkey) were detected at lower levels than those from apes and Old World monkeys, possibly due to sequence differences in the iRBD compared to the human ortholog Active Retroviral Fusogen Expressed in Human Tissues ® (against which the serum was raised). All 11 EnvP(b)1 orthologs mediated cell-cell fusion in each of the cell lines tested, with the exception of the pigtail macaque ortholog in human A549 cells ( Fig. 3A and B). Fusion efficiency varied between EnvP(b)1 orthologs and cell lines ( Fig. 3B and Fig. S4). EnvP(b)1 from great ape species (human, chimpanzee, and gorilla) were generally the most fusogenic across all cell lines. In these assays, the extent of syncytium formation did not correlate with the extent of apparent EnvP(b)1 expression or processing. Proteolytic processing is essential for fusion, as we observed complete loss of fusogenicity with a variant of the human EnvP(b)1 ortholog in which the predicted furin cleavage site was mutated (Fig. 3C).

Structure of the EnvP(b)1 inferred receptor binding domain.
We determined the crystal structure of the human EnvP(b)1 iRBD. Single anomalous dispersion (SAD) from a platinum derivative was used to obtain experimental phases and ultimately produce a model (Fig. 4A). The iRBD's core is defined by a ␤ sandwich that shares some similarity with a V-type Ig fold, although it lacks the first ␤ sheet at the N terminus. Computational analysis suggests that the rest of the N-terminal region of EnvP(b)1 is helical and that it is cleaved by signal peptidase. Extensions between the ␤ sheets form a series of loops and ␣-helical regions that decorate the core. There is weak sequence identity between these features and the gammaretrovirus variable regions A, B, and C (VRA, VRB, and VRC, respectively), which are believed to dictate Env-receptor interactions for gammaretroviruses (15,16).
The iRBD is conserved in sequence among the 35 simian EnvP(b)1 orthologs (Fig. S5), with insertions and deletions present only in the unstructured C terminus. We mapped amino acid diversity to the iRBD structure by scoring the number of different amino acids that were present at each position (Fig. 4B). For the majority of the domain, only one or two amino acids are present at a given site. Substitutions within the core were often conservative. Most of the sites that tolerate amino acid diversity are in loops or surface exposed residues on helices. In particular, the sequence of VRC has diverged   the most rapidly, suggesting that its sequence is not subject to the same constraints as other regions of the iRBD.
We compared the iRBD structure to the two other known gammaretrovirus RBD structures. These are from Friend murine leukemia virus (FrMLV) and feline leukemia virus (FeLV), both extant circulating viruses (15,17). The iRBD shares little (ϳ20%) sequence identity with these examples, yet it shows extremely strong structural similarities. The EnvP(b)1 iRBD is more compact with shorter, less prominent loop structures extending from its core (Fig. 4C). All three VRs are shorter in EnvP(b)1: VRA and VRB are dramatically shorter, and the VRC, also shorter, adopts a different conformation as it is liberated from constraints by the absence of the paired Cys residues that form a disulfide in FrMLV and FeLV. Additional connecting regions also exhibit changes in length and corresponding structural differences. As a consequence, the potential receptor-interacting interfaces of EnvP(b)1 are quite different from those for FrMLV and FeLV. However, there is clearly a structurally conserved architecture that is common among gammaretroviruses (Fig. S6). The ␤ sandwich core, a central helix that runs perpendicular to it, and an 8-amino-acid loop between the last two ␤ sheets are all structurally conserved. This conserved scaffold structure serves as a platform for the presentation of unique elements that determine receptor binding specificity.

DISCUSSION
Captured viral diversity has produced host genetic novelty. Viral Env proteins have at least two activities: receptor binding and membrane fusion. In cells, Env-receptor interactions can result in receptor blockade or downregulation of the receptor from the cell surface. Receptor engagement is also often a trigger that potentiates membrane fusion. Over the course of animal evolution, these viral functions have been coopted in multiple instances to perform host functions. ERV Envs that have been coopted are diverse in their ages, sequences, and, where known, binding partners (18). It appears that capturing and maintaining a variety of genes from divergent viruses provides evolutionary benefits.
EnvP(b)1 has all the hallmarks of a viral fusogen that has been repurposed to perform a host function. In contrast to other ERV-P(b)1 genes, the EnvP(b)1 ORF has been maintained as a coopted gene for the entirety of simian evolution, and likely longer. At least one of its functions, membrane fusion, has also been retained in every simian clade. Syncytin-2 (ERV-FRD Env), which is thought to play a key role in the formation of the syncytiotrophoblast layer of the placenta, is the only other ERV Env known to have fusogenic members in all simian clades (19)(20)(21). Our transfection and fusion experiments, performed in cells derived from divergent primate species, indicate that the unidentified cellular factors required for fusion, such as receptors and proteases, are also conserved. In these assays, ape and New World monkey EnvP(b)1 orthologs tended to behave more similarly to each other, while Old World monkey EnvP(b)1s were overall less fusogenic. These lineage-specific differences may reflect the consequences of divergence and coevolution of EnvP(b)1 orthologs with the factors that are required for cell fusion.
EnvP(b)1 protein is expressed in multiple healthy human tissues. In these tissues, it exists in a processed, presumably fusion-competent form. We detected the fully processed SU form and no unprocessed SU-TM in all EnvP(b)1-expressing tissues. In contrast, SU-TM was the major species in transfected cells. We speculate that in tissues, expression of EnvP(b)1 and the proper furin-like protease may be coordinated. Under nonreducing conditions, EnvP(b)1 is predominantly found in higher-molecular-weight species in both tissues and transfected cells, similar to oligomeric species observed for syncytin-1 (ERV-W Env) under nonreducing conditions (22,23), indicating that disulfidelinked and fusion-primed SU-TM is present in these tissues. Our detection of protein was generally in agreement with the previously reported RNA abundances, with the exception of splenic tissue, where we observed only a band of a different molecular weight (5,7). Placental tissue had the highest expression of EnvP(b)1 protein. In other tissues, expression did not vary with pregnancy or gender. The function of EnvP(b)1 is unknown, but its maintenance among simians for at least 40 to 71 million years and its expression and processing in human tissues indicate that it must in some way contribute to an important physiological process, likely involving membrane fusion.
Among human examples, selective preservation or loss of furin processing has accompanied cooption. The two syncytins are activated by furin-like cleavage events and act as membrane fusogens (2). HERV-T is no longer processed, is not fusogenic, and has evolved to downregulate its receptor (24). HEMO, of unknown function, is not classically processed by proteases but has evolved to be shed from cells via an alternative cleavage event (25). Others, like recently integrated HERV-K copies, may have been too recently acquired to have gained or lost functions. EnvP(b)1 has clearly evolved to retain its processing and fusogenicity.
EnvP(b)1 and the original infectious progenitor virus are quite ancient. We have determined that the EnvP(b)1 ORF has been maintained within the RIN3 locus for the duration of simian primate evolution. Insertion into this locus occurred after the divergence of simians from prosimians and before the radiation of extant simian species, or between 71 and 40 million years ago. One study, using LTR dating approaches, has estimated ERV-P(b)1 to be as old as 55 million years (6). We identified a number of related sequences in the genome of an aye-aye, a species which has evolved in isolation on Madagascar for the past 55 to 75 million years (13,14). Due to their short length and an incomplete aye-aye genome, we cannot resolve the relationship of these sequences with EnvP(b)1 or other related loci in simian genomes. Nonetheless, it appears likely that interactions between primates and related circulating virus are ancient and were widely distributed.
EnvP(b)1 has been preserved in simian genomes for at least 40 to 71 million years of evolution. We determined a structure of the iRBD from this ancient viral element. The iRBD amino acid sequence is largely conserved among the EnvP(b)1 loci found in host genomes. Though the EnvP(b)1 ORF as a whole is under purifying selection, some diversity has accumulated in the variable regions. These sites evolve rapidly in circulating viruses, and their divergence in EnvP(b)1 indicates that such sites can tolerate substitution without compromising function even in the context of a coopted protein. This conservation and the lack of insertions/deletions suggest that the current iRBD may closely approximate that of the infectious progenitor virus that circulated 40 to 71 MYA.
Comparisons with RBD structures from two extant gammaretroviruses highlight the exquisite evolutionary plasticity of this domain. All three share a conserved architecture that depends little on sequence conservation. Upon this common scaffold, each Env has evolved a series of unique features. EnvP(b)1 is the most compact, while the cores of FeLV and Fr-MLV are decorated by large structured extensions. The lack of sequence conservation prevents us from tracing the origins of these features. Whether the elaborate elements reflect a consequence of ongoing virus-host evolution (e.g., antibody escape or receptor evolution) over timescales greatly exceeding 40 to 71 million years or whether contemporary Envs retain RBD domains that are equivalently compact can only be resolved through additional structures. It does, however, underscore the versatility of this fold as a platform for recognizing many different receptor molecules. This versatility in receptor recognition may underlie the evolutionary success of gammaretroviruses, and of the Env proteins themselves, as evidenced by their wide distribution as viral fusogens, ERV elements, and coopted genes.
Identification of EnvP(b)1 sequences. Human, chimpanzee, white-cheeked gibbon, African green monkey, and pigtail macaque sequences were previously published (5). All other full-length EnvP(b)1 ORFs were identified through querying publicly available databases, including NCBI GenBank, Ensmbl, and the UCSC genome browser, using the full-length human EnvP(b)1 nucleotide sequence. Databases queried included fully assembled genomes, NCBI nucleotide database, and whole-genome shotgun reads identified through the basic local alignment tool (BLAST). Details of these sequences are shown in Table S2.
Recombinant EnvP(b)1 iRBD expression and purification. The inferred receptor binding domain, residues 51 to 210, of human EnvP(b)1, was deduced as follows. A region of homology to murine and feline leukemia virus RBDs was identified using the Phyre2 server (30). The signal peptide cleavage site was predicted using the SignalP-4.1 server (31) and Phyre2 secondary structure prediction (30). This defined the iRBD N terminus. The C-terminal boundary was defined by the transition from a predicted unstructured region to the proline-rich region. iRBDs were synthesized and cloned into a modified pVRC8400 mammalian expression vector. We produced a variant with a C-terminal human rhinovirus 3C protease cleavage site followed by a 6ϫHis tag and a variant with the following C-terminal tag configuration: an HA-tag followed by a rhinovirus 3C protease cleavage site, an 8ϫHis tag, and a 6ϫHis tag. Short Gly-Gly-Ala linkers separate the HA, 3C, and poly-His tags. iRBD variants were produced by polyethylenimine (PEI)-facilitated transient transfection of 293F cells that were maintained in FreeStyle 293 expression medium. Transfection complexes were prepared in Opti-MEM and added to cells. Supernatants were harvested 4 to 5 days posttransfection and clarified by low-speed centrifugation. iRBDs were purified by passage over cobalt-nitrilotriacetic acid (Co-NTA) agarose (Clontech) followed by gel filtration chromatography on Superdex 200 (GE Healthcare) in 10 mM Tris-HCl, 150 mM NaCl at pH 7.5 (buffer A).
For immunization of rabbits and production of antisera enriched for iRBD-directed IgGs, we used the version with the 3C protease site and 6ϫHis tag. Both were removed by treatment with PreScission protease (MolBioTech; ThermoScientific). The protein was repurified on Co-NTA agarose followed by gel filtration chromatography on Superdex 200 (GE Healthcare) in buffer A to remove the protease, tag, and uncleaved protein. Crystallographic studies used the hemagglutinin (HA)-tagged version, which was not treated with protease to remove its poly-His tags.
Generation and enrichment of EnvP(b)1 iRBD-specific antisera. Bulk anti-iRBD serum was generated by Covance. Purified recombinant EnvP(b)1 iRBD was used to immunize a single rabbit. Serum from the terminal bleed was tested for detection of EnvP(b)1 by Western blot. Crude rabbit serum was diluted 1:50 in buffer A. Bulk IgGs were captured by protein A agarose resin (GoldBio) at 4°C with rotating overnight. The resin was collected in a chromatography column, washed with a column volume of buffer A, and eluted with 0.1 M glycine buffer (pH 2.5), which was immediately neutralized by 1 M Tris (pH 8.0). Antibodies were then dialyzed against phosphate-buffered saline (PBS) (pH 7.4). To produce an EnvP(b)1 iRBD affinity resin, EnvP(b)1 was expressed and purified as described above. Cleaved iRBD was then concentrated, dialyzed against phosphate-buffered saline (PBS) (pH 7.4), and arrested to Pierce N-hydroxysuccinimide (NHS)-activated agarose beads per the manufacturer's protocol. The reaction was then quenched by the addition of Tris (pH 7.5). Rabbit antibodies were then incubated with the iRBD resin at 4°C with rotating overnight. The iRBD resin was collected in a chromatography column, washed with a column volume of PBS (pH 7.4), and eluted with 0.1 M glycine buffer (pH 2.5), which was immediately neutralized by 1 M Tris (pH 8.0). Antibodies were then dialyzed against PBS (pH 7.4).
Crystallization, structure determination, and refinement. EnvP(b)1 iRBD, with its one native N-linked glycan and a C-terminal tag, was concentrated to 10 to 12 mg/ml. Crystals were grown in hanging drops over a reservoir of 0.5 M sodium chloride, 0.1 M bis-Tris propane (pH 7.0), 20% (wt/vol) polyethylene glycol 4000. Platinum was incorporated by incubating crystals in a solution of 0.08 M bis-Tris propane, 0.4 M sodium formate, 16% (wt/vol) polyethylene glycol 4000, and 10 mM potassium tetrachloroplatinate(II) for approximately 48 h. All crystals were cryoprotected in 22 to 25% (vol/vol) glycerol. We recorded diffraction data at the Advanced Light Source on beamline 8.2.2 and on the Advanced Photon Source on beamlines 24-ID-E and 24-ID-C.
Using XDS (32), data were processed for one native crystal and the single anomalous dispersion (SAD) data were obtained for one platinum derivative crystal. The merged data used for phasing had an anomalous signal to about 3.5 Å resolution. Using SHELX (33), we found 5 platinum sites, refined them, and calculated phase probabilities. Initial maps were obtained after statistical density modification using RESOLVE (34). The Matthew coefficient suggested that there were multiple iRBD copies in the asymmetric unit. We flooded the initial map with dummy atoms and carved out a single copy of iRBD (the one with the best-resolved density) by inspection in PyMol (Schrödinger, LLC). With this we obtained a search model to locate two additional copies of iRBD in the asymmetric unit by molecular replacement using PHASER (35). After further optimizing the phases by 3-fold noncrystallographic symmetry (NCS) averaging, we obtained a map of reasonable quality to build an initial model that was then used to phase the native data set.
We carried out refinement calculations with PHENIX (36) and model modifications with COOT (37). Refinement of atomic positions and B factors was followed by translation-liberation-screw (TLS) parameterization and placement of water molecules. Final coordinates were validated with the MolProbity server (38). Data collection and refinement statistics are in Table S3. Figures were made with PyMol.
Comparison of RBD structures. Topology diagrams of the EnvP(b)1 iRBD, Friend murine leukemia virus RBD (PDB code 1LCS) (15), and feline leukemia virus RBD (PDB code 1AOL) (17) were modified from those created by PDBSum (39). We used the Dali protein structure comparison server (40) to determine the percent identity, root mean square deviation (RMSD), and DALI Z-scores for these three proteins.
Human tissue lysates and cell lysate Western blots. Tissue protein lysates were purchased from TaKaRa. Details for each lysate are shown in Table S1. Tissue lysate was mixed with Laemmli buffer with or without (for nonreducing conditions) 2-mercaptoethanol and boiled for 5 min. A 40-g portion of tissue lysate was run on a 4 to 20% acrylamide gel (Bio-Rad).
For all Western blots, proteins were transferred onto nitrocellulose membrane. Membranes were blocked with 5% milk in PBS-0.1% Tween 20 and probed with anti-EnvP(b)1 serum or mouse anti-GAPDH antibody (Genscript), followed by horseradish peroxidase (HRP)-conjugated goat anti-mouse immunoglobulin (Invitrogen) or HRP-conjugated goat anti-rabbit immunoglobulin (Sigma-Aldrich) antibodies. Membranes were incubated with enhanced chemiluminescence (ECL) reagent (Thermo), and signal was detected using an Amersham imager 600 device (GE Healthcare). Images were processed using ImageJ software (National Institutes of Health).
Data availability. Coordinates and diffraction data have been deposited at the Protein Data Bank (PDB code 6W5Y).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only.