The Antiviral Activity of the Cellular Glycoprotein LGALS3BP/90K Is Species Specific

Abstract

Cellular antiviral proteins interfere with distinct steps of replication cycles of viruses. The galectin 3 binding protein (LGALS3BP, also known as 90K) was previously shown to lower the infectivity of nascent human immunodeficiency virus type 1 (HIV-1) virions when expressed in virus-producing cells. This antiviral effect was accompanied by impaired gp160Env processing and reduced viral incorporation of mature Env glycoproteins. Here, we examined the ability of 90K orthologs from primate species to reduce the particle infectivity of distinct lentiviruses. We show that 90K's ability to diminish the infectivity of lentiviral particles is conserved within primate species, with the notable exception of 90K from rhesus macaque. Comparison of active and inactive 90K orthologs and variants uncovered the fact that inhibition of processing of the HIV-1 Env precursor and reduction of cell surface expression of HIV-1 Env gp120 are required, but not sufficient, for 90K-mediated antiviral activity. Rather, 90K-mediated reduction of virion-associated gp120 coincided with antiviral activity, suggesting that 90K impairs the incorporation of HIV-1 Env into budding virions. We show that a single "humanizing" amino acid exchange in the BTB (broad-complex, tramtrack, and bric-à-brac)/POZ (poxvirus and zinc finger) domain is sufficient to fully rescue the antiviral activity of a shortened version of rhesus macaque 90K, but not that of the full-length protein. Comparison of the X-ray structures of the BTB/POZ domains of 90K from rhesus macaques and humans point toward a slightly larger hydrophobic patch at the surface of the rhesus macaque BTB domain that may modulate a direct interaction with either a second 90K domain or a different protein.IMPORTANCE The cellular 90K protein has been shown to diminish the infectivity of nascent HIV-1 particles. When produced in 90K-expressing cells, particles bear smaller amounts of the HIV-1 Env glycoprotein, which is essential for attaching to and entering new target cells in the subsequent infection round. However, whether the antiviral function of 90K is conserved across primates is unknown. Here, we found that 90K orthologs from most primate species, but, surprisingly, not from rhesus macaques, inhibit HIV-1. The introduction of a single amino acid exchange into a short version of the rhesus macaque 90K protein, consisting of the two intermediate domains of 90K, resulted in full restoration of antiviral activity. Structural elucidation of the respective domain suggests that the absence of antiviral activity in the rhesus macaque factor may be linked to a subtle change in protein-protein interaction.

Keywords: antiviral; human immunodeficiency virus; interferons; restriction factor; virus infectivity.

FIG 1

Species-specific antiviral activity of 90K proteins. (A) Phylogenetic tree of 90K proteins, generated with phylogeny.fr (40). (B) HEK293T cells were cotransfected with 1.3 μg HIV-1 proviral DNA and 1.3 μg of the indicated constructs. The abbreviations for the individual species are defined in Materials and Methods. At 48 h posttransfection, the infectivity of cell-free particles was calculated as infectivity per nanogram of p24 capsid antigen in the supernatant. Particle infectivity in the absence of 90K was set to 100%. Shown are mean values from 3 to 4 independent experiments ± SEM. Lysates of the producer cells were immunoblotted with anti-myc and anti-MAPK antibodies. Asterisks indicate statistical significance (P < 0.05); n.s., not significant.

FIG 2

90K inhibits HIV-1 by impairing Env incorporation into particles. (A) Particles for which results are shown in Fig. 1B were sucrose cushion-purified and analyzed by immunoblotting for gp120 incorporation using anti-gp120 and anti-p24 antibodies. gp120 incorporation was calculated as the ratio of gp120 signal intensity to p24 signal intensity. Values obtained in the absence of 90K were set to 100%. Asterisks indicate statistical significance (P < 0.05); n.s., not significant. (B) Producer cell lysates were analyzed by immunoblotting for gp160 processing using anti-gp120 and for Gag expression using anti-p24. Relative gp160 processing was determined by expressing the level of mature gp120 as a percentage of the whole-Env (gp120 + gp160) signal. Band intensities were quantified by infrared-based imaging. The bar diagrams show mean values from 3 to 7 independent experiments ± SEM. (C) Correlative analysis of gp120 Env content per p24 CA and HIV-1 particle infectivity (infectivity per nanograms of p24) in the supernatants of cells expressing the indicated 90K-myc protein or the vector. (D) Correlative analysis of HIV-1 particle infectivity and levels of mature gp120 per whole-Env expression in producer cell lysates expressing the indicated 90K-myc protein or the vector. (E) Correlative analysis of levels of gp120 viral incorporation and levels of mature gp120 per whole-Env expression in producer cell lysates. In panels C to E, symbols represent the values for individual transfections. The Pearson correlation coefficient r and the corresponding P value were calculated using GraphPad Prism software. (F) HEK293T cells were mock transfected or cotransfected with the indicated HIV-1 proviral DNAs and constructs. At 48 h posttransfection, intact cells were immunostained for HIV-1 Env surface expression and analyzed by flow cytometry. In the dot plots, the HIV-1 Env levels are plotted against GFP. Numbers inside the boxes indicate the mean fluorescence intensity (MFI) of the HIV-1 Env signal in the respective gates. (G) The levels of cell-surface-exposed HIV-1 Env in each sample were quantified on GFP-positive cells in the R3 gate relative to the MFI of GFP-negative cells in the R2 gate, which serve as an internal reference. The relative cell surface expression levels of HIV-1 Env are depicted, with values for HIV-1 Env expression in vector-transfected cells set to 100%. Shown are mean values from six independent experiments ± SEM. (H) HEK293T cells were either mock treated or transfected with an empty vector or 90K-encoding plasmids. Supernatants were harvested at days 1 and 2 posttransfection and were subjected to Luminex-based multiplex assays according to the manufacturer's instructions. Secretion into supernatants was analyzed for MIF, M-CSF, and MCP-1 (CCL2); all concentrations are given in picograms per milliliter. (I) HEK 293T cells were either mock treated or brefeldin A treated for 6 h. Supernatants were then subjected to Luminex-based multiplex assays as outlined for panel H.

FIG 3

Specificity of antilentiviral activity of 90K. HEK293T cells were cotransfected with 1.3 μg HIV-1_NL4.3 (A), HIV-2_7312A (B), SIV_mac239 (C), SIV_smm (D), SIV_agm (E), or SIV_cpz (F) proviral DNA and 90K-myc-encoding expression plasmids of the indicated species. Asterisks indicate statistical significance (P < 0.05); n.s., not significant. (A to C) At 48 h posttransfection, the infectivities of cell-free HIV-1 (A), HIV-2 (B), and SIV_mac239 (C) were calculated as infectivity per capsid antigen in the supernatant. Panel A is identical to Fig. 1B and is shown as a reference. (D to F) Due to the lack of availability of specific anti-CA antibodies, the mere relative infectivity is shown. Values obtained in the absence of 90K were set to 100%. Shown are mean values ± SEM from 4 to 7 independent experiments.

FIG 4

Toward a minimal antiviral 90K variant. (A) Scheme of the 90K protein domain organization and the analyzed truncated mutants. Numbers indicate amino acid positions. (B) HEK293T cells were cotransfected with HIV-1 proviral DNA and the indicated 90K constructs. At 48 h posttransfection, the infectivity of cell-free particles was measured as infectivity per nanogram of p24 capsid antigen in the supernatant. Values obtained in the absence of 90K were set to 100%. Shown are arithmetic means ± SEM from 3 to 8 independent experiments. Asterisks indicate statistical significance (P < 0.05); n.s., not significant. Lysates of the producer cells were immunoblotted with anti-myc and anti-MAPK antibodies. (C) Sucrose cushion-purified progeny virions were analyzed by immunoblotting, and band intensities were quantified by infrared-based imaging. gp120 incorporation, defined as the ratio of gp120 signal intensity to p24 signal intensity, was calculated from 3 to 6 independent experiments. Values obtained in the absence of 90K were set to 100%. (D) gp160 processing in cell lysates was determined by immunoblotting using the indicated antibodies. Band intensities were quantified by infrared-based imaging, and the percentage of mature gp120 from total Env protein is depicted. The bar diagrams show the arithmetic means ± SEM from 3 to 5 independent experiments. (E) Relative levels of gp120 viral incorporation and particle infectivity, obtained in the presence of the human 90K variants used, are plotted against each other. (F) Levels of mature gp120 per whole-Env expression in producer cell lysates and relative levels of particle infectivity are plotted against each other. (G) Relative levels of gp120 viral incorporation and levels of mature gp120 per whole-Env expression in producer cell lysates are plotted against each other. For these values, the Pearson correlation coefficient r and the corresponding P value are indicated. The open circles depict the values obtained in the absence of 90K expression.

FIG 5

Genetic humanization renders a minimal version of rhesus macaque 90K antiviral. (A) HEK293T cells were cotransfected with HIV-1 proviral DNA and the indicated 90K constructs. At 48 h posttransfection, the infectivity of cell-free particles was measured as infectivity per nanogram of p24 capsid antigen in the supernatant. Values obtained in the absence of 90K were set to 100%. Shown are arithmetic means ± SEM from 3 to 8 independent experiments. Asterisks indicate statistical significance (P < 0.05); n.s., not significant. (B) The human and rhesus macaque BTB/POZ and IVR domains were aligned using Clustal Omega (41) and ESPript (42). Conserved secondary-structure elements, including α-helices, 3₁₀-helices, β-strands, and β-turns taken from the human BTB domain structure, are shown above the alignment (α, η, β, and TT, respectively). Nonconserved residues are highlighted by a red background; a green background indicates conserved cysteines, and their disulfide connectivity is drawn in green below the alignment. (C) HEK293T cells were first cotransfected with HIV-1 proviral DNA and constructs encoding the indicated interspecies chimeras and then processed as for panel A. Shown are arithmetic means ± SEM from 3 independent experiments. (D) HEK293T cells were cotransfected with HIV-1 proviral DNA and the indicated constructs and were processed as for panel A. Shown are arithmetic means ± SEM from 3 independent experiments. (E) HEK293T cells were cotransfected with HIV-1 proviral DNA and combinations of plasmids encoding for short versions of wild-type rhesus macaque 90K-myc [127-408 (wt)] and of rhesus macaque 127-408 (V142A). Both 90K variants were titrated in the opposite direction, covering plasmid ratios of 0+4, 1+3, 1.5+2.5, 2+2, 2.5+1.5, 3+1, and 4+0. Transfected cells were processed as for panel A.

FIG 6

Crystal structure of human and rhesus macaque BTB. (A and C) Ribbon diagrams of the human BTB homodimer, with one subunit shown in gray and the other colored in gradations from the N to the C terminus (blue to red) according to the color scales in the bars underneath. Disulfide bonds are shown as green sticks. (B) Superposition of human BTB and rhesus macaque BTB, with the two subunits displayed in gray (human BTB) and green (rhesus macaque BTB). Side chains of residues that are divergent between the human and rhesus macaque 90K proteins are shown as sticks and are colored according to atom type (gray and green for carbon of human BTB and rhesus macaque BTB, respectively; red and blue for oxygen and nitrogen, respectively). (D) Closeup view of the homodimerization interface of superposed human BTB and rhesus macaque BTB. Side chains of residue 142, which provides a gain of antiviral activity in the context of the short version of rhesus macaque 90K-myc, and the conserved residue A¹⁷⁸, which makes an additional hydrophobic contact with V¹⁴² (dashed line), are shown as sticks. (E and F) The molecular surfaces of the human (E) and rhesus macaque (F) 90K BTB domain are displayed in gray and green, respectively, as in panel B. The surface area corresponding to A¹⁴² (human BTB) or V¹⁴² (rhesus macaque BTB) is shown in red.