Schlafen 12 restricts HIV-1 latency reversal by a codon-usage dependent post-transcriptional block in CD4+ T cells

Abstract

Latency is a major barrier towards virus elimination in HIV-1-infected individuals. Yet, the mechanisms that contribute to the maintenance of HIV-1 latency are incompletely understood. Here we describe the Schlafen 12 protein (SLFN12) as an HIV-1 restriction factor that establishes a post-transcriptional block in HIV-1-infected cells and thereby inhibits HIV-1 replication and virus reactivation from latently infected cells. The inhibitory activity is dependent on the HIV-1 codon usage and on the SLFN12 RNase active sites. Within HIV-1-infected individuals, SLFN12 expression in PBMCs correlated with HIV-1 plasma viral loads and proviral loads suggesting a link with the general activation of the immune system. Using an RNA FISH-Flow HIV-1 reactivation assay, we demonstrate that SLFN12 expression is enriched in infected cells positive for HIV-1 transcripts but negative for HIV-1 proteins. Thus, codon-usage dependent translation inhibition of HIV-1 proteins participates in HIV-1 latency and can restrict the amount of virus release after latency reversal.

Fig. 1. Differential RNA-seq analysis of cultured naive CD4+ T cells to identify candidate HIV-1 restriction factors.

a Scheme of naïve CD4+ T cell cultures. Naïve CD4+ T cells from 3 different HIV-1-negative blood donors were maintained for 13 days under 4 different conditions: IL-7+IL-15 alone (A-HSP) or with anti-CD3/CD28 activation at day 12 (B-HSP+ TCR); IL-2 after anti-CD3/CD28 activation alone (C-TCR) or with a second activation by anti-CD3/CD28 at day 12 (D-TCR+ TCR). On day 13, total RNA was isolated and processed for transcriptome analysis to identify candidate HIV-1 restriction factors. b Phenotypes of CD4+ T cells maintained under HSP or TCR condition. A representative result of one of the three blood donors utilized in a is shown. Isolated naive CD4+ T cell purity was above 91%, and 96.5% for the donor shown in the upper panel. After 12-day cultivation, a naïve T cell phenotype is partially maintained in HSP conditions (CD27+/CD45RO−, 45.1%; on average >30%), while TCR conditions increased cells with central memory phenotype (CD27+/CD45RO+, 35.1%; on average >35%). T_N, naïve T cells; T_CM, central memory T cells; T_EM, effector memory T cells; T_TM, terminally differentiated memory T cells. c Heatmap of the top 2,000 variable genes clustered by k-means analysis (red: upregulated, blue: downregulated). Four main clusters were identified: Cluster I, genes upregulated in both HSP and HSP + TCR conditions compared to either TCR or TCR + TCR conditions (n = 384). Cluster II, genes upregulated in HSP, HSP + TCR and TCR conditions (n = 677). Cluster III, genes with a trend to be upregulated in TCR condition (n = 205). Cluster IV, genes upregulated by TCR stimulation at day12 (n = 734). These differences in expression are visualized on the right panel for clarity. d Top 5 gene ontology terms of the cluster I genes. e Expression patterns of two known inhibitors of HIV-1 replication/function in the cluster I according to the NIH HIV interaction database^–. Mean counts per million reads (CPM) and the standard error of the mean (SEM; n = 3) are shown. f Venn diagram shows overlap among (i) cluster I genes, (ii) differentially expressed genes in HSP vs TCR and in HSP + TCR vs TCR + TCR conditions (DEGs), and (iii) members of gene families that contain a known restriction factor (RF gene family members). g Candidate restriction factors identified in f and their corresponding gene families. The asterisks show gene families that could potentially be involved in post-transcriptional and/or translational events. h Expression pattern of SLFN11 and SLFN12 from 3 blood donors. Plots represent mean CPM ± SEM.

Fig. 2. SLFN12 affects HIV-1 latency reversal from ACH2 cells.

a Relative transcript levels of the human SLFN family members 5, 11, 12 and 13 in different cell lines (mean values of two repeated measurements) as well as in naive CD4+ T cells under HSP or TCR culture conditions. Naïve CD4+ T cells were isolated from five independent HIV-1-negative blood donors and cultivated for 13 days with IL-7 + IL-15 (HSP condition) or with anti-CD3/CD28 antibodies + IL-2 (TCR condition) as in Fig. 1a. Transcripts were quantified by RT-qPCR with specific primer pairs and normalized to cellular 18S rRNA. Dot plots of the individual values and p-values were provided in Supplementary Fig. 2. b Flow chart of SLFNs knockdown and HIV-1 reactivation. HIV-1 latently infected ACH2 cells were transduced with retroviral vectors expressing specific shRNAs against SLFN11 (shSLFN11) or SLFN12 (shSLFN12#1/shSLFN12#2), or with a scrambled shRNA (shSc) as a negative control. The cells were then treated with SAHA to reactivate HIV-1. DMSO at a concentration that equals the activation mix served as the vehicle control. Cell lysates and supernatants (SN) were harvested at 48- and 72-h post-reactivation, respectively, and analyzed. c Relative RNA levels of SLFN11 (left panel) and SLFN12 (right panel) in the knockdown ACH2 cells (n = 4 biological replicates; mean ± SD). *** represents p < 0.01 by Student’s t test. d SAHA-induced HIV-1 reactivation from ACH2 cells after SLFN knockdown. Fold HIV-1 reactivation was determined by titration of HIV-1-containing supernatants on TZM-bl cells and normalized to the basal HIV-1 level in DMSO-treated samples. Given is the mean ± SD for three independent samples. e Knockdown of SLFN11 and SLFN12 in ACH2 cells increases translation efficiency of HIV-1 Gag-Pr55. Plots represent fold change (FC) translation efficiency of Gag-Pr55 in SAHA-treated samples to DMSO-treated samples. Pr55 translation efficiency was calculated as a ratio of cellular Pr55 protein levels to gag RNA levels. The cellular Pr55 protein and HIV-1 gag RNA levels were quantified by Western blot and RT-qPCR, respectively. The mean FC efficiency of the control knockdown cells (shSc) was set to 1. Results show the mean ± SD of three independent experiments. Ratio paired t-test was used to calculate statical significance (*p < 0.05, **p < 0.005).

Fig. 3. SLFN12 inhibits HIV-1 at the level of translation.

a Experimental outline. HEK 293T cells were co-transfected with an expression plasmid encoding mCherry-fused SLFN11 or SLFN12, or mCherry alone, and the HIV-1 vector pNL-E. At 48 h post-transfection, supernatants (SN) and cell lysates were collected and processed for further analyses as indicated. b Expression of recombinant SLFN proteins in the transfected HEK 293T cells was studied by Western blots with an anti-mCherry antibody. c SLFN12 expression strongly diminishes HIV-1 production from transfected HEK 293T cells. Supernatants from the co-transfected cells were titrated by the TZM-bl assay (upper panel; n = 3 for each SLFN11 (denoted as rectangles) or SLFN12 (triangles) independent transfection; mean ± SD; *p < 0.05 by Student’s t test) and analyzed by Western blot (lower panel; a representative example of three independent experiments is shown). d Dose-dependent inhibition of HIV-1 production by SLFN12. HEK 293T cells were co-transfected with pNL-E and increasing amounts of pmCherry-SLFN12 (0, 0.2, 0.4, 0.8 or 1.6 µg). The empty mCherry plasmid was added to maintain constant DNA amounts for all transfections. SNs were harvested 48 h after transfection and titrated by TZM-bl assay (n = 3 biological replicates; mean ± SD; *p < 0.05 by Student’s t test). SLFN12 expression did not significantly affect the levels of HIV-1 gag RNA (e) or total HIV-1 RNA (f). RNA levels were quantified by RT-qPCR and normalized to 18S rRNA levels (n = 4 biological replicates; mean ± SD). There were no significant differences amongst the indicated samples (Student’s t test). g Western blot from cell lysates of co-transfected HEK 293T cells with indicated antibodies. An arrow and asterisk highlight the bands of Nef and a nonspecific protein, respectively. Representative results of three independent experiments are shown.

Fig. 4. SLFN12 slows down translation elongation.

a Outline of polysome profiling. HEK 293T cells were co-transfected with expression plasmids encoding mCherry-fused SLFN11 or SLFN12, or mCherry alone, and the HIV-1 vector pNL-E. At 48 h post-transfection, cells were lysed and subjected to a 10–50% sucrose gradient centrifugation. The gradient was fractionated, and samples were processed for specific RT-qPCR shown in d and e. b UV absorbance profiles of lysed transfected HEK 293T cells after sucrose gradient fractionation (representative result of three independent samples). c SLFN12 expression does not influence the global ratio between monosomes and polysomes (mean ± SD; n = 3 biological replicates). d, e SLFN12 expression shifts the HIV-1 RNA distribution towards heavy polysomes. Distribution of GAPDH RNA (d) or HIV-1 gag RNA (e) in different fractions of HEK 293T cells expressing SLFN11 or SLFN12. Pooled fractions represent free proteins (mRNPs fraction; 1–7), single ribosomal subunits (40S + 60S; 8–12), 80S monosome (13–17), light- (18–23) and heavy -polysomes (24–29). The total amount of the RNAs in all fractions was set to 100% (mean ± SD; n = 2 biological replicates).

Fig. 5. SLFN12-mediated inhibition of HIV-1 translation is codon-usage dependent.

a Histogram of codon adaptation indices (CAIs) for the human RefSeq transcript hg38. CAIs of HIV-1 sequences, GAPDH and EGFP are highlighted for comparison. The mean CAI of human transcripts was 0.77. Gag-opt, codon-optimized gag sequence as in plasmid pGag-opt. b Experimental outline. HEK 293T cells were co-transfected with expression plasmids encoding mCherry-fused SLFN11 or SLFN12, or mCherry alone, and an HIV-1 vector expressing Gag-p24 with either wild-type (pGag-wt) or optimized codon usage (pGag-opt). At 48 h post-transfection, transfected cells were lysed and analyzed by Western blot and RT-qPCR. c, d Upper panel: Western blot detection of p24 from HEK 293T cells expressing wild-type Gag (Gag-wt, c) or codon-optimized Gag (Gag-opt, d). Lower panel: Relative HIV-1-gag RNA levels from indicated expression vectors (n = 3 biological replicates; mean ± SD). There were no significant differences between the RNA levels amongst the indicated samples (Student’s t test). e Heatmap of relative synonymous codon usage of human (H. sapiens), HIV-1, HIV-2 and Chikungunya virus (CHIKV) transcripts. Codons Leu-UUA (Leu(TTA) in the graph) and Leu-CUG (Leu(CTG)) are highlighted with arrows. f SLFN11 and SLFN12 specifically inhibit Leu-UUA codon-swapped EGFP expression (n = 3 biological replicates; mean ± SD). The upper panel shows the outline of the experiment. Wild-type EGFP expression vector (WT) or Leu codon-swapped vectors (Leu-CUG, Leu-CUC, Leu-UUA, Leu-CUU, Leu-UUG, and Leu-CUA) were individually transfected together with mCherry (MOCK), mCherry-fused-SLFN11 or 12 expression vector into HEK 293T cells. After 48 h, relative mean fluorescence intensities (MFI) were measured by flow cytometry. The fold change MFI values are shown; the MFI of mock transfection was set to 1. The asterisks represent statistical significance (*p < 0.02, by one sample t-test).

Fig. 6. SLFN12-mediated HIV-1 suppression depends on a putative tRNase cleavage domain.

a Conformational similarity of human SLFN11 (beige), SLFN12 (cyan) and SLFN13 (magenta). Monomer structures retrieved from the cryo-EM structures of SLFN12 (7LRE) and SLFN11 (7ZEL), and from AlphaFold predicted model of SLFN13 (downloaded from Uniprot, code Q68D06) were superimposed and compared in ribbon plates obtained with Chimera. SLFN12 has a short C-terminal domain. b Structural models of SLFN12 dimer/tRNA complexes. The cryo-EM structure of SLFN12 dimer (7LRE) docked with the structure of type II tRNA (selenocysteine tRNA, 3HL2) and relaxed, adding the relaxed structure of type I tRNA^Phe (5AXM) for comparison, are shown as surface representations (chain A of SLFN12 in light blue, chain B of SLNF12 in cyan, type II tRNA in yellow and type I tRNA in green). In the upper right is shown in ribbon plates a comparison of the structures of SLFN11 (7ZEL) and the relaxed model of SLFN12 docked with type I (green) and type II (yellow) tRNAs, highlighting in van der Waals spheres the side-chain atoms of E200 and E205 of SLFN12, and E209 and E214 of SLFN11, of both chains. In the bottom are shown the surface of the relaxed structure of SLFN12, colored by Coulombic potential the positively (red) and negatively (blue) charged areas, plus a ribbon plate of the structures of type I and type II tRNAs, with arrows pointing to the positions of E200 and E205 in both chains and a detailed view of the active site in chain A. The positively charged surface of SLFN12 was responsible of the binding of the tRNA helping to locate the tRNA in a cleft close to the active site formed by glutamates E200 and E205. c Experimental outline. mCherry-fused SLFN expression vectors or their respective active site mutants E209A and E214A of SLFN11, and E200A and E205A of SLFN12 were individually co-transfected with HIV-1 pNL-E vector into HEK 293T cells. 48 h post-transfection, cells and supernatants were harvested and processed for further analyses as indicated. The following results are from three independent experiments (d, e). d Mutants of the tRNase cleavage sites of SLFN12 lose their anti-HIV-1 activity (mean ± SD). Asterisks indicate significant differences compared to MOCK-transfected cells calculated by Student’s t test. The lower panel shows expression levels of the SLFN proteins in a Western blot using an anti-mCherry antibody. e Western blot showing intracellular HIV-1 Gag Pr55 and p24 levels in SLFN-expressing cells. EGFP and GAPDH were used as controls. f The Leu-UUA codon swapped vector was transfected with indicated SLFN mutant expression vectors into HEK 293T cells. After 48 h, relative mean fluorescence intensity (MFI) was measured by flow cytometry. The fold change MFI values are shown; the MFI of mock transfection was set to 1 (n = 2 biological replicates; mean ± SD).

Fig. 7. Involvement of SLFN12 gene in HIV-1 post-transcriptional restriction in patients.

a Expression level of SLFN12 in PBMCs from HIV-1-infected individuals without ART. SLFN12 RNA levels were quantified using PBMCs from HIV-1 high viremic- (High, n = 16) and low viremic- (Low, n = 30) patients, and viremic- (VC, n = 11) and elite-controllers (EC, n = 12). SLFN12 levels were normalized by TBP RNA levels and CD4+ T cell counts per µl. Each point represents the values of a single individual. p-values were calculated by Mann–Whitney test. b Correlation between SLFN12 expression levels and viral RNA loads. Each point represents the values of a single individual. The correlation was calculated by Spearman’s rank test. c Correlation between SLFN12 expression levels and HIV-1 provirus loads. Each point represents the values of a single individual. Spearman’s rank test was applied to analyze the correlation. d–f SLFN12 expressing CD4+ T cells are enriched in HIV-1 RNA+/p24- cells. Gating strategy to measure SLFN12 expression in CD4+ T cells from patients on suppressive ART (d). Gating strategy consisted of selecting lymphocytic cells (“Lymphocytes” on the plots) by FSC- and SSC-scatter, followed by a double doublet exclusion (“Singlets1” and “Singlets2”), dead cells exclusion (“Live cells”) and finally an HIV-1-RNA+ and p24+ gate from where SLFN12+ cells were identified. The proportion of SLFN12 expressing cells (e) or relative MFI of SLFN12 (f) in p24-/HIV-1 RNA−, p24-/HIV-1 RNA+, or p24+ cells. Blue circles, patient samples treated with Romidepsin plus Ingenol (RMD + ING in d) ex vivo (n = 3); White circles, untreated patient samples (R10 in d, n = 3).