Sp140L функционирует как фактор ограничения герпесвируса, подавляющий вирусную транскрипцию и активирующий интерферонально-стимулированные гены.

Herpesviruses are double-stranded DNA viruses that cause lifelong infections due to their ability to establish viral latency and avoid immunodetection ( 1 ). Clinically relevant herpesviruses include herpes simplex virus (HSV), human cytomegalovirus (HCMV), and Kaposi’s sarcoma–associated herpesvirus (KSHV), as well as Epstein–Barr virus (EBV), which is associated with numerous lymphomas and carcinomas and is implicated in several autoimmune disorders – particularly multiple sclerosis ( 2 – 4 ). The B cell biology of EBV infection and associated diseases can be modeled in vitro, as infection of primary B cells leads to cellular transformation and the generation of immortalized lymphoblastoid cell lines (LCLs). Given the ability of herpesviruses to establish lifelong latency, initial infection is a crucial stage where cellular sensing and restriction can prevent the virus from successfully establishing latency. Yet, the mechanisms of intrinsic restriction of herpesviruses have not been fully elucidated.

One key mediator of defense against DNA viruses during initial infection is the PML nuclear body (PML-NB). Core PML-NB proteins implicated in restricting DNA viruses include PML, DAXX, ATRX, and Sp100. PML serves as a scaffold to mediate the interaction of other core proteins, while death domain associated protein XX (DAXX) and alpha-thalassemia mental retardation syndrome X-linked (ATRX) together load the repressive histone variant H3.3 onto viral genomes during initial infection – preventing the accumulation of active histone marks on the incoming viral genome and suppressing the transcription of viral genes ( 5 – 9 ). The speckled protein Sp100 can restrict the replication and viral gene expression of several DNA viruses ( 10 – 14 ). The speckled protein genetic locus includes additional family members SP110 , SP140 , and – exclusively in primates – SP140L ( 15 ), whose roles in viral infection are largely unknown. The speckled family genes all encode key domains: i) an N-terminal CARD domain involved in multimerization, ii) the DNA-binding SAND domain, and iii) the histone-reading PHD/Bromodomains ( 15 ). Genetic mutations in SP110 and SP140 are associated with autoimmune disorders and increased susceptibility to intracellular bacterial infections ( 15 ). And while Sp110 and Sp140 repress the Type I Interferon (IFN) response in mice ( 16 , 17 ), they appear to promote Type I IFN responses in humans ( 18 ). The function of Sp140L, however, remains unexplored.

To overcome PML-NB restriction, successful herpesviruses encode proteins that antagonize various PML-NB components of their hosts. These include viral tegument proteins pp71 in HCMV ( 19 , 20 ) and BNRF1 in EBV ( 21 ) which target DAXX and ATRX; and ORF3 of herpesvirus saimiri (HVS) – a KSHV-like rhadinovirus of squirrel monkeys – which degrades Sp100 ( 22 ). Some immediate-early viral proteins also antagonize PML-NBs, including ICP0 in HSV which targets PML, Sp100, DAXX, and ATRX ( 23 , 24 ); IE1 in HCMV targeting PML and Sp100 ( 12 , 25 ); and Epstein-Barr virus nuclear antigen-leader protein (EBNA-LP) in EBV, which transiently displaces Sp100 from PML-NBs during early infection ( 26 , 27 ). While the molecular mechanism by which these viral proteins antagonize PML-NB components may be distinct, the functional consequences are similar as evidenced by the ability of ICP0 or various combinations of BNRF1 or pp71 with IE1 or EBNA-LP to complement an ICP0-null HSV and pp71-deficient HCMV ( 28 – 30 ). The role of Sp100 in EBV infection has not been directly investigated in primary infection ( 26 , 27 ).

EBNA-LP is critical for the transformation of naïve B cells by EBV infection ( 31 ), but its function in early infection is not well defined compared to other PML-NB-disrupting herpesvirus proteins mentioned above. The architecture of EBNA-LP is unique in that it is composed of N-terminal, identical, 66 amino acid repeats (W domains) and a C-terminal 45 amino acid Y domain. While the structure of EBNA-LP is unknown, it is predicted to be highly disordered ( 32 ). Based on transient transfection assays, EBNA-LP was thought to primarily function as a coactivator of another early EBV protein, EBNA2 to enhance expression of the essential viral gene Latent Membrane Protein 1 (LMP1) and some cellular genes ( 33 – 37 ). The mechanism by which EBNA-LP coactivates EBNA2 has been variously attributed to displacement of the transcriptionally repressive HDAC4 and NCoR proteins from viral and cellular promoters ( 38 , 39 ). However, B cells infected with an EBNA-LP Knockout EBV (LPKO) exhibit reduced expression of not only LMP1 but also other viral genes that are not EBNA2-dependent, while EBNA2-induced host gene expression is not reduced, and sometimes even increased after LPKO infection ( 31 ) suggesting that EBNA-LP’s relationship to EBNA2 is more complex than a simple “coactivator.” Additionally, our recent work identified highly conserved, hydrophobic leucine-rich motifs in both the W and Y domains of EBNA-LP that associate with the DNA-looping factor YY1 ( 40 ). Thus, EBNA-LP likely has EBNA2-independent roles that are critical for establishing latency in, and transformation of, naïve B cells.

Given the essential role of immediate-early herpesvirus proteins in overcoming intrinsic viral restriction, and particularly the gap in knowledge as to how and why EBNA-LP is necessary for EBV infection and transformation of naïve B cells, we used single-cell RNA sequencing (scRNAseq) to compare the trajectories of wild type (WT) and LPKO virus-infected B cells during the first days of infection. This assay revealed key facets of the restriction of EBV in the absence of EBNA-LP. We then defined the cellular factors and mechanisms responsible for the restriction of the LPKO virus, and finally the breadth of how these factors are restrictive across human DNA viruses.

Results

EBNA-LP Influences the Fate Trajectories of EBV-Infected B Cells.

To assess the role of EBNA-LP in EBV-driven B cell fate trajectories, we performed single-cell RNA sequencing (scRNAseq) of primary human B cells prior to infection (Day 0), and at 2, 5, and 8 d following WT and LPKO infection as depicted in Fig. 1 A ( 41 ). B cell purity ( SI Appendix , Fig. S1 A ) and infection efficiency ( SI Appendix , Fig. S1 B ) were confirmed by expression of CD19 and CD23 respectively at the time of collection. Single-cell RNA-seq revealed that infected samples contained more messenger RNAs (mRNA) ( SI Appendix , Fig. S1 C ) and more unique mRNA features ( SI Appendix , Fig. S1 D ) per cell than uninfected (Day 0) B cells. Dimensional reduction integrating all seven samples generated a single Uniform Manifold Approximation and Projection (UMAP) for further analysis ( Fig. 1 B ). Uninfected cells largely cluster separately from infected cells ( Fig. 1 C ) and 2 d postinfection LPKO- and WT-infected clusters are nearly identical to each other ( SI Appendix , Fig. S1 D and E ). Differences in cellular states between WT- and LPKO-infected B cells emerge by 5 d ( SI Appendix , Fig. S1 F and G ) and are accentuated 8 d postinfection ( SI Appendix , Fig. S1 H and I ).

LPKO-infected B cells have altered cellular trajectories compared to WT EBV. ( A ) Schematic of collected samples. ( B ) UMAP for Dimension Reduction of the combined seven samples. Colors correspond to samples in ( A ). ( C ) UMAP with uninfected B cells highlighted. ( D ) Cells 2 d postinfection with WT. ( E ) Cells 2 d postinfection with LPKO. ( F ) Cells 5 d postinfection with WT. ( G ) Cells 5 d postinfection with LPKO. ( H ) Cells 8 d postinfection with WT. ( I ) Cells 8 d postinfection with LPKO. ( J ) UMAP with cells grouped into 9 unique clusters based on differential gene expression. ( K ) Model of the identified clusters in the cellular trajectories of EBV-infected B cells. ( L ) Frequency of each cluster within each sample. ( M ) Model of the trajectories of WT- and LPKO-infected B cells at each time point based on frequencies in ( L ). Circle area is proportional to the frequency of clusters in each sample. Transparent circles represent clusters which are absent.

Our prior work defined a set of distinct B cell fate trajectories following EBV infection that largely mimic the states of B cell activation and germinal center maturation ( 42 ). In comparing WT- and LPKO-infected cells, we have identified 9 distinct cell states that largely recapitulated those characterized previously ( Fig. 1 J and Dataset S1 ). We used the top cluster-defining genes in our previous study to identify the corresponding cell states in this dataset ( SI Appendix , Fig. S2 A – I ). Cell states were further distinguished on the basis of viral gene expression ( SI Appendix , Fig. S2 J – U ) and cell cycle phase ( SI Appendix , Fig. S2 V – X ). Cluster 0 (Uninfected—c0) contained uninfected naïve and memory B cells ( Fig. 1 K ). Two days postinfection, prior to cellular proliferation, which begins around 3 d postinfection ( 43 ), a portion of cells exist in a gene expression state associated with antiviral sensing (Sensed – c1) ( Fig. 1 K ). Cells that avoid this initial barrier transition toward an activated state, resembling pregerminal center B cells (Early Activation – c2), and toward a state of DNA replication enriched for cells in the S phase (S Phase – c3) ( Fig. 1 K ). By 5 d postinfection, cells advanced to a state of hyperproliferation, with markers of dark zone germinal center B cells (Hyperproliferating – c4) ( Fig. 1 K ). Likely as a result of the cellular stress associated with EBV-induced hyperproliferation ( 43 , 44 ), by this time point some cells also arrest in a state of cellular stress (Stressed – c5) ( Fig. 1 K ). Cells also transition toward a second, but distinct, activated B cell state (Late Activation – c6) ( Fig. 1 K ). Compared to the Early Activation state (c2), this Late Activation state (c6) is more highly enriched in naïve B cells based on IgD expression ( SI Appendix , Fig. S2 X ), has lower expression of Myc and E2F targets ( SI Appendix , Fig. S2 Y and Z ) likely a consequence of reduced expression of the viral transactivator EBNA2 in c6 compared to c2 ( 45 ) ( SI Appendix , Fig. S2 R ), and is likely dependent upon cells having undergone proliferation based on its appearance only after day 2 ( Fig. 1 K ). EBV-infected cells also enter a NFκB-High state, likely driven by the viral protein LMP1, whose expression is highest in this population ( SI Appendix , Fig. S2 T ), that is characterized by markers of light zone germinal center B cells (NFκB High—c7) ( Fig. 1 K ). The final population is a differentiated plasmablast-like state (Differentiated—c8) ( Fig. 1 K ).

While this model held true for WT-infected cells, LPKO-infected cells had altered trajectories. First, unlike WT-infected cells, LPKO-infected cells remain in the early (preproliferation) states of activation (c2 and c3) beyond 2 d postinfection ( Fig. 1 L ). Second, LPKO-infected cells were also more likely than WT-infected cells to enter states of antiviral sensing (c1) or cellular stress (c5) after day 2 ( Fig. 1 L ). Finally, the LPKO-infected cells largely failed to establish the NFκB-High state (c7) ( Fig. 1 L ), consistent with the previously reported low expression of LMP1 in LPKO infections ( 31 ) ( SI Appendix , Fig. S2 U ). Surprisingly, the LPKO virus efficiently established the plasmablast state (Differentiated – c8) ( Fig. 1 L ), despite the absence of the NFκB-High state which is critical in the emergence of plasmablasts from the germinal center ( 46 ). Interpreted as trajectories in Fig. 1 M , LPKO-infected cells persist in early infection states of activation (c2 and c3) and cellular stress prior to proliferation (c1) longer than WT. LPKO-infected cells that do hyperproliferate have a higher propensity to arrest with signs of cellular stress (c5) and to enter a germinal center- or light zone-independent route toward the plasmablast state (c8), similar to germinal center-independent B cell maturation ( 47 ).

To validate our transcriptome data, we first identified RNA changes in cell surface markers defining the key cell populations. RNA levels of CCR6 ( Fig. 2 A ) and CD23 ( Fig. 2 B ) in the scRNA-seq data distinguish many cell states, similar to previous studies ( 42 ). Early Activation states (c2 and c3) are characterized by CCR6 hi/ CD23 lo expression, while the Late Activation state (c6) is CCR6 hi/ CD23 hi . The NFκB-High state (c7) is CCR6 lo /CD23 hi and is further resolved by high expression of ICAM1 ( Fig. 2 C ), a proxy marker for LMP1 expression ( 48 ). The remaining states are CCR6 lo /CD23 lo with the Stressed (c5) distinguishable from the Differentiated state (c8) by low expression of CD38 ( Fig. 2 D ) and reduced cellular proliferation. Therefore, these markers along with the proliferation tracker CellTrace should allow quantitation of the main states that differ between LPKO and WT infections: persistence in states of activation (c2, c3, c6), reduced transition to the NFκB-high state (c7), and increased cellular arrest prior to hyperproliferation (c5) as outlined in Fig. 2 E .

Altered cell populations between WT- and LPKO-infected B cells identified in scRNAseq data are validated by flow cytometry 8 d postinfection. Expression levels of CCR6 ( A ), CD23 ( B ), ICAM1 ( C ), and CD38 ( D ) in scRNA-seq clusters. ( E ) Legend of flow panel markers for the key states identified. Flow staining for CCR6 and CD23 8 d postinfection in representative Donor 2 for WT ( F ) and LPKO ( G ) infected B cells. Quadrant color corresponds to cell states that can be distinguished as depicted in ( E ). Percent of WT- and LPKO-infected cells 8 d postinfection in the CCR6 hi /CD23 lo (Early Activation) states ( H ), the CCR6 hi /CD23 hi (Late Activation) state ( I ), and the CCR6 lo /CD23 hi (NF?B-High) states ( J ). ( K ) Histogram comparing expression of ICAM1 protein (proxy for LMP1) between WT and LPKO infected cells in CCR6 lo /CD23 hi population. ( L ) Quantified Median Fluorescence Intensity (MFI) of ICAM1 signal in CCR6 lo /CD23 hi population in WT and LPKO infections (n = 4). ( M ) Histogram of CellTrace stain, diluted during each cell division, 8 d postinfection in WT and LPKO infected cells in representative Donor 1. Gray area indicates cells that have not proliferated. ( N ) MFI of CellTrace (n = 4). Lower MFI indicates more cell proliferation. ( O ) Percent of WT and LPKO infected cells 8 d postinfection in the CCR6 lo /CD23 lo states (n = 4). Separation of CCR6 lo /CD23 lo populations in WT ( P ) and LPKO ( Q ) infection by CD38 and CellTrace to distinguish Differentiated cells from Stressed cells. ( R ) Percent of CCR6 lo /CD23 lo cells that are also CD38 lo and CellTrace hi (indicating low proliferation) representing the stressed cell cluster. Significance determined with the ratio-paired t test, N = 4 biological replicates. *Indicates P value less than or equal to 0.05, ** indicates P value less than or equal to 0.01. Histograms are scaled as percent of maximum count (modal).

Using flow cytometry to analyze protein expression of identified markers in a new set of infected samples, we classified cells by CCR6 and CD23 levels ( Fig. 2 F and G and SI Appendix , Fig. S3 A and B ) and confirmed the prolonged presence of LPKO-infected cells in states of early activation (c2 and c3) (CCR6 hi /CD23 lo ) ( Fig. 2 H ) and late activation (c6) (CCR6 hi /CD23 hi ) ( Fig. 2 I ) compared to WT infection. Second, we confirmed that LPKO-infected cells have reduced transition to NF?B-high state (c7), as LPKO-infected cells have reduced CCR6 lo /CD23 lo populations ( Fig. 2 J ) and the cells that do enter this state express significantly less ICAM1 ( Fig. 2 K and L and SI Appendix , Fig. S3 C ) compared to WT. Furthermore, compared to WT, LPKO-infected cells are less proliferative ( Fig. 2 M and N and SI Appendix , Fig. S3 D ). And, upon further separating CCR6 lo /CD23 lo cells based on CD38 expression and proliferation, we confirmed that compared to WT, LPKO-infected cells are more frequently in states of cellular stress (c5) (CCR6 lo /CD23 lo /CellTrace hi /CD38 lo ) ( Fig. 2 O – R and SI Appendix , Fig. S3 E and F ). Altogether, these data support a model in which EBNA-LP is important in preventing the arrest, stress response, and immune activation after EBV infection.

LPKO-Infected Cells Fail to Suppress the Antiviral Response and Fail to Induce Cellular Metabolic Remodeling.

We next sought to characterize the molecular mechanisms underlying the cell fate restrictions following LPKO infection. We first pooled all time points for each WT and LPKO scRNAseq dataset into a “pseudobulk” dataset and performed gene set enrichment analysis (GSEA) to identify differentially regulated pathways between WT and LPKO infection. In WT-infected cells, we observed significant enrichment of transcripts involved in NFκB signaling ( Fig. 3 A ) in agreement with the failure of LPKO-infected cells to generate an NFκB-high state ( Fig. 1 L ). We also observed an increase in expression of genes controlling cellular metabolism in WT- relative to LPKO-infected cells ( Fig. 3 A ). In contrast, LPKO-infected cells were enriched for transcripts of interferon-stimulated genes (ISGs) and antiviral responses ( Fig. 3 B ).

EBNA-LP promotes expression of proproliferative genes, while reducing expression of antiviral genes.Pathways from GSEA Hallmark gene sets or Gene Ontology datasets that are enriched in ( A ) WT-infected over LPKO-infected or ( B ) LPKO-infected over WT-infected cells, based on our scRNA data. Representative top-ranking pathways with false discovery rates (FDR q-values) less than 0.05. Higher normalized enrichment score indicates stronger enrichment. Dot plot of representative NFκB signaling-related genes ( C ) and metabolism genes ( D ) identified in GSEA pathways enriched in WT-infected cells for each sample in the scRNAseq time course. ( E ) Basal rate of adenosine triphosphate (ATP) derived from glycolysis (glycoATP) in WT- and LPKO-infected naïve B cells 4 d postinfection in two donors. ( F ) Basal rate of ATP derived from oxidative phosphorylation (mitoATP) in WT- and LPKO-infected naïve B cells 4 d postinfection in two donors. ( G ) Dot plot of representative Interferon Stimulated Genes enriched in LPKO-infected cells for each sample in the time course. ( H ) Top 10 predicted transcription factors identified by ChEA3 associated with expression of the genes enriched in WT-infected cells in scRNAseq samples compared to LPKO or ( I ) LPKO-infected cells compared to WT. Lower mean rank indicates a stronger correlation.

We next used time-resolved pseudobulk analysis to better define the dynamic nature of EBNA-LP-dependent gene regulation. NFκB targets were modestly decreased 2 d postinfection in both WT and LPKO-infected cells, but then diverged at 5 d, and more so at 8 d, postinfection ( Fig. 3 C ). Cellular metabolic gene expression diverged by day 5, following the first EBV-induced cell division ( Fig. 3 D ). While loss of NFκB signaling is concurrent with failure to efficiently express LMP1 ( 31 ) ( SI Appendix , Fig. S2 U ), failure to upregulate cellular metabolism in the absence of EBNA-LP was previously unexplored. Therefore, we further examined cellular metabolic activity during early infection and confirmed naïve B cells infected with LPKO virus had reductions in both basal glycolysis and oxidative phosphorylation compared to WT-infected cells at 4 d postinfection ( Fig. 3 E and F ).

Transcripts downregulated upon expression of EBNA-LP (i.e., induced in LPKO-infected cells) were primarily ISGs. Intriguingly, though, interferon gene mRNAs were not elevated in LPKO-infected cells ( SI Appendix , Fig. S4 A ), indicating ISG induction may occur through other mechanisms. These ISG transcripts are widely induced by 2 d postinfection in both WT- and LPKO-infected cells ( Fig. 3 G ). In the presence of EBNA-LP, WT-infected cells effectively down-regulated expression of these ISGs, whereas in LPKO-infected cells, levels of these ISG transcripts were sustained or increased ( Fig. 3 G ). These trends in differentially expressed pathways observed in the scRNAseq data were confirmed by bulk RNAseq of WT- and LPKO-infected cells in additional donors at 8 d postinfection ( SI Appendix , Fig. S4 B – F ).

Because the ISG induction appeared to be independent of IFN induction, we sought to identify which cellular factors could be mediating the observed gene expression differences in WT- and LPKO-infected cells. We used ChEA3 ( 49 ) (ChIP-X Enrichment Analysis Version 3) to identify transcription factors whose target genes were enriched in the transcriptome of WT- and LPKO-infected cells. Genes upregulated in WT-infected cells were associated with factors including NFκB subunits; CENPA—a centromere protein and transcriptional regulator that forms a complex with YY1 ( 50 ) which we recently identified as an EBNA-LP-interacting protein ( 40 ); and IRF4, which can suppress ISG transcription ( 51 ), induce expression of Myc ( 51 , 52 ), and function in cooperation with BATF and BATF3 that are also suggested mediators of these expression changes by the analysis ( 53 , 54 ) ( Fig. 3 H ). Genes upregulated in the absence of EBNA-LP were instead associated with all four human speckled proteins (Sp100, Sp110, Sp140, and Sp140L) ( Fig. 3 I ).

Sp100 and Sp140L, But Not Other PML-NB Proteins, Are Required to Restrict EBV Transformation in the Absence of EBNA-LP.

To examine whether the ChEA3-identified speckled proteins or other PML-NB proteins specifically restrict EBV infection in the absence of EBNA-LP, we assessed the effect of knocking out these target genes on rescuing the inability of LPKO virus to transform naïve B cells. Naïve B cells isolated from adult peripheral blood ( SI Appendix , Fig. S5 A ) were directly transfected with Cas9-guide RNA ribonucleoprotein (Cas9/RNP) complexes targeting genes encoding the main components of PML bodies, including the speckled proteins, along with CD46, a nonessential cell surface protein used as a proxy for knockout of the target gene ( 55 ) ( Fig. 4 A and SI Appendix , Table S1 ). Transfected cells were then infected with either the LPKO or WT virus, and outgrowth of edited, CD46-negative cells was assessed by flow cytometry weekly postinfection, for 4 wk ( Fig. 4 A ). As expected, control LPKO-infected cells in which only CD46 was targeted failed to sustain proliferation to generate LCLs, whereas WT EBV still transformed the naïve B cells into LCLs ( Fig. 4 B – D ). While knockout of most of the PML-NB-associated proteins that were targeted – including PML – had no significant impact on the outgrowth of LPKO-infected cells, which decreased in number over time like the CD46 -only knockout, mutation of either SP100 or – even more emphatically – SP140L rescued outgrowth ( Fig. 4 B and C ). Indeed, knockout of SP100 or SP140L was sufficient to consistently generate LPKO LCLs from naive B cells by 28 d postinfection ( Fig. 4 D ), and target genes were efficiently knocked out in these LCLs as validated by genomic DNA sequencing ( SI Appendix , Fig. S5 B ) and protein expression where applicable ( SI Appendix , Fig. S5 C ). Additional LPKO LCLs were generated after a longer time in culture with inconsistent target knockout ( SI Appendix , Fig. S5 D ), likely arising from contaminating memory B cells. Targeting the same factors in WT-infected cells had no impact on outgrowth ( SI Appendix , Fig. S5 E and F ), confirming Sp100 and Sp140L restrict EBV infection only in the absence of EBNA-LP.

Knockout of SP140L or SP100 rescues transformation of LPKO-infected naïve B cells. ( A ) Schematic of Cas9/RNP screening approach. ( B ) Log(10) Fold Change from Day 2 of the total number of CD46 negative cells for each condition 2, 7, 14, 21, and 28 d postinfection. The dark turquoise line represents the mean number of CD46 negative cells in CD46 only control WT- or LPKO-infected cells 28 d postinfection. SD = standard deviation. ( C ) Total number of CD46 negative cells 28 d postinfection for each condition plotted as log fold change from 2 d postinfection. P values calculated by one-way ANOVA with multiple comparisons to compare to LPKO infected cells transfected with guide targeting CD46 only as control. *Indicates P -values < 0.05, ** indicates P- values < 0.01, **** indicates P value < 0.0001. ( D ) Total number of LPKO LCLs in each donor for each target out of four replicates per donor. # Indicates conditions in which at least one LCL was generated after an additional 3 wk in culture with variable knockout efficiency. ^Indicates conditions in which at least one LCL was generated in the same time frame at WT virus, but the target gene was not knocked out. P values calculated using Fischer’s exact test to compare outcomes to LPKO control condition ** indicates P -values < 0.01, **** indicates P value < 0.0001.

Structural Modelling of Sp140L and Sp100 Interactions with EBNA-LP Defines a Critical Interface Important for B Cell Transformation.

To better understand the nature of the putative interaction between EBNA-LP and Sp140L and Sp100, we used AlphaFold3 to predict the structure of a complex between these proteins ( 56 ). First, we found that despite EBNA-LP being predicted to have high intrinsic disorder ( 32 ), AlphaFold3 consistently predicted that EBNA-LP forms an extended beta sheet, with loops connecting helices derived from the repeated W domains and a loop extending the unique 45 amino acid Y domain containing a short alpha helix ( Fig. 5 A and B ). The interaction sites with Sp140L and Sp100 are consistently predicted to involve this Y domain alpha helix engaging a CARD domain–derived helical bundle of Sp140L or Sp100 ( Fig. 5 C and D ), the latter being consistent with prior biochemical findings ( 26 ). Specifically, the hydrophobic interface of these interactions relies on a conserved leucine-rich motif (LRM) within EBNA-LP ( 40 ) ( Fig. 5 E and F ). Indeed, modification of the leucines in this helix to alanines was sufficient to disrupt the association between EBNA-LP and Sp100 by coimmunoprecipitation ( Fig. 5 G ).

EBNA-LP is predicted to engage Sp140L and Sp100 through an alpha-helix containing a leucine-rich motif in the C-terminal Y domain. ( A ) Schematic of EBNA-LP (encoding four repeated W domains) predicted structure W (magenta) and Y domain (orange) contributions to beta strands and alpha helices based on ( B ). ( B ) AlphaFold3-predicted model of EBNA-LP three-dimensional structure.Predicted model of EBNA-LP (orange) in complex with (C) Sp140L or (D) Sp100. Speckled protein domains are highlighted: CARD domain (blue), SAND domain (green), and PHD/BROMO domain (yellow). Enlarged view of EBNA-LP Y domain helix (spanning amino acids 276 through 287) encoding the leucine-rich motif in contact with ( E ) Sp140L and ( F ) Sp100 CARD domains. EBNA-LP residues with side chains predicted to contact CARD domains are indicated. ( G ) FLAG coimmunoprecipitation of FLAG-tagged EBNA-LP wild type or mutant with leucines in Y domain alpha-helix modified to alanine (Y LRM Mutant, L276,280,281A) with HA-Sp100A performed in 293T cells. EBNA-LP constructs encode 4 W domains. 1% of input material and 25% of coimmunoprecipitation were used for imaging. ( H ) Total tdTomato positive cells in each condition of cord blood B cells infected with LPKO and transcomplemented with indicated DNA construct 21 d postinfection and ( I ) 35 d postinfection (performed in two donors, with three replicates each). Untransfected cells indicate background signal threshold. P values calculated by ordinary one-way ANOVA with Dunnett’s multiple comparisons test. **** Indicates P value < 0.0001. ( J ) Number of tdTomato positive LCLs generated per condition. # Indicates a single replicate in which cells were transformed but were not tdTomato positive. P values were calculated using Fisher’s exact test to compare transformation outcomes to LPKO-infected cells transcomplemented with empty vector. ** Indicates P value < 0.01.

To further examine the role of the Y domain LRM at the speckled protein interface in the transformation of naïve B cells, we used a transcomplementation assay in cord blood B cells infected with LPKO virus as previously described ( 40 ). Prior to infection, cells were transfected with an episomal plasmid encoding tdTomato and either wild type or an EBNA-LP mutant in which leucines were mutated to alanine. We first confirmed expression of the mutant protein in LPKO-transformed total B cells, since EBNA-LP is not required for the transformation of the memory B cell component, and demonstrated that both wild type and mutant EBNA-LP are able to be expressed ( SI Appendix , Fig. S6 A ) and had the same subcellular localization profile as EBNA-LP expressed from WT virus ( SI Appendix , Fig. S6 B ). EBNA-LP with a mutated Y domain LRM failed to rescue transformation of LPKO-infected cord blood B cells compared to wild-type EBNA-LP ( Fig. 5 H – J ), indicating that the Y domain LRM of EBNA-LP involved in Sp100 and likely Sp140L interaction is essential during EBV infection.

SP100 or SP140L Knockout Rescues Cellular and Viral Gene Expression Critical for Naïve B Cell Outgrowth.

We then sought to investigate the mechanism by which Sp100 and Sp140L restrict EBV infection in the absence of EBNA-LP. We used bulk RNAseq to examine the effect of SP100 and SP140L knockout on viral and cellular gene expression 5 d after LPKO infection of naïve B cells ( SI Appendix , Fig. S7 A ). Knockout of SP100 or especially SP140L reduced induction of IFN-related genes in LPKO-infected cells ( Fig. 6 A – D ). Knockout of either SP100 or SP140L also rescued the expression of pathways important for proliferation including MYC targets, E2F targets, and particularly upon SP140L knockout, metabolic pathways ( Fig. 6 A – C and E ). NF?B signaling was also rescued by SP100 and SP140L knockout in LPKO infected cells ( Fig. 6 A – C and F ), with knockout of SP140L again exhibiting a more complete rescue of gene expression toward wild-type levels – in agreement with the transformation screen ( Fig. 4 B ). Since LMP1 and other viral genes are inefficiently activated during LPKO infection ( 31 ), we examined viral latency gene expression and found that LMP1 and LMP2 RNA levels in LPKO infected cells were somewhat elevated upon loss of SP100 and largely recovered in SP140L knockouts ( Fig. 6 G ). This was further confirmed at a functional level by flow cytometry using ICAM1 as a proxy for LMP1, as knockout of SP100 or – to a greater extent – SP140L rescued surface ICAM1 protein to levels comparable to WT-infected cells ( Fig. 6 H and SI Appendix , Fig. S7 B ).

SP100 and SP140L Knockout in LPKO-infected cells promotes transcription of pathways required for outgrowth and inhibits the antiviral response. GSEA Hallmark pathway analysis of bulk RNA-seq samples 5 d postinfection comparing LPKO-infected cells with CD46 knockout versus ( A ) SP100 knockout or ( B ) SP140L knockout or ( C ) LPKO- versus WT virus-infected cells (n = 2). GSEA terms with nominal P value less than 0.05 are shown. Higher absolute value of the normalized enrichment score indicates stronger enrichment. Heatmap of genes related to the significantly enriched Hallmark pathways ( D ) “IFN Alpha Response,” ( E ) “E2F Targets,” and ( F ) “TNFA Signaling by NFκB” for each condition. Average z-score of two samples is plotted. P values determined by One-way Repeated Measures ANOVA with multiple comparisons to LPKO control. **** indicates P value less than 0.0001. ( G ) Average Z-score for mRNA encoding LMP1 and LMP2A for each sample. ( H ) Histogram of flow cytometry signal for LMP1 proxy gene ICAM1 in each sample.

Sp140L Also Restricts ORF3-Deficient HVS.

As Sp140L has not previously been studied as a viral restriction factor, we sought to determine whether Sp140L can restrict herpesviruses other than EBV, and in cell types other than B cells, by studying Sp140L in HVS infection. As the HVS tegument protein ORF3 targets Sp100 for degradation ( 22 ), we generated two independent WT HVS and ORF3-KO HVS viruses ( SI Appendix , Fig. S8 A and B ) carrying a CMV promoter-driven luciferase expression cassette. We confirmed that infection of human fibroblast HFF cells with WT HVS, but not ORF3-KO HVS, resulted loss of Sp100 protein ( SI Appendix , Fig. S8 C ). We then compared virus transcription after initial infection in two human fibroblast cell lines (HFF and MRC5 cells) carrying knockout of either SP100 or SP140L alongside the proxy gene CD46 ( Fig. 7 A ). Consistent with a previous report with an ORF3-KO of a C-strain HVS ( 22 ), knockout of SP100 rescued viral transcription after ORF3-KO HVS infection ( Fig. 7 B and C ). Consistent with our EBV data, knockout of SP140L was also sufficient to rescue luciferase expression from the ORF3-KO virus in both cell types ( Fig. 7 B and C ). Loss of SP100 or SP140L in the context of WT HVS infection had little or no effect on viral transcription ( SI Appendix , Fig. S8 D – F ) suggesting that ORF3 antagonizes both Sp100 and Sp140L function during WT HVS infection. These findings suggest Sp140L is a restriction factor of diverse gammaherpesviruses in primates, that is at least as important as Sp100.

Sp100 and Sp140L restrict HVS in the absence of ORF3. ( A ) Schematic of experiment. Fold change in luciferase expression HFF cells (n = 7) ( B ) or MRC5 cells (n = 5) ( C ) infected with HVS-ORF3 Knockout or WT virus encoding luciferase with knockout of control CD46 , SP100 , or SP140L . Two virus preparations ( A and B ) were used. P values calculated by Dunnett’s multiple comparisons test using an ordinary two-way ANOVA. *Indicates P- values < 0.05, **indicates P- values < 0.01, ****indicates P value < 0.0001.

Discussion

Herpesviruses infect nearly all animal species and result in lifelong latent and persistent infections ( 57 ). As DNA viruses, herpesviruses must overcome cellular barriers to establish infection, including the detection and chromatinization of their foreign DNA genomes by intrinsic restriction factors such as those in PML-NBs, including Sp100. Here, we have identified Sp140L, an Sp100-homolog, as a restriction factor of herpesvirus infection that is counteracted by the EBV latency protein EBNA-LP and the HVS tegument protein ORF3. Our data are consistent with a mechanistic model whereby both Sp100 and Sp140L restrict EBV infection by preventing the expression of essential viral latency genes that are expressed after EBNA-LP and EBNA2, and by altering cellular gene expression—both inducing an ISG response and reducing B cell metabolic capacity. While the core antiviral PML-NB proteins have been thought to include PML, Sp100, DAXX, and ATRX, our work adds the evolutionarily recent Sp140L to the list of key cellular intrinsic restriction factors of DNA viruses in primates.

Based on studies of HSV and CMV mutants that lack abilities to inhibit PML-NB functions ( 5 , 19 , 21 , 58 , 59 ), we can conclude that PML-NBs have two parallel repressive pathways in humans: one driven by DAXX/ATRX and another by Sp100 and Sp140L. In the LPKO virus, its DAXX-counteracting tegument protein BNRF1 is still intact, suggesting that factors in our screen that failed to rescue naïve B cell transformation may be part of the DAXX-associated pathway. While DAXX/ATRX load repressive histones (e.g., H3.3) onto the incoming viral genome ( 5 ), Sp100 and Sp140L may be responsible for the addition of repressive histone marks such as H3K9me3 and/or H3K27me3. For example, during latent KSHV infection of endothelial cells, loss of Sp100, but not DAXX nor PML, correlates with H3K27me3 deposition on the viral genome ( 60 ), although Sp100’s ability to bind heterochromatin protein 1 (HP1) and high-mobility group protein B1 (HMGB2) ( 61 ) suggests Sp100 may also be linked to H3K9me3 ( 62 ).

While PML-NB components are known to restrict DNA virus gene expression, we found that Sp100 and Sp140L are also important in suppressing cellular proliferation. One possible mechanism is the failure to express the LMPs during LPKO infection, although LMP1 is not essential for the initial cell divisions after EBV infection ( 63 ). A more convincing alternative is that these factors integrate intrinsic DNA sensing with an antiproliferative ISG response. Recent evidence suggests Sp100 can localize to ISG promoters ( 64 ) indicating Sp100 and potentially Sp140L induce ISGs through direct transcriptional regulation or chromatin modification at ISG promoters – as observed with other PML-NB components including HIRA ( 65 , 66 ), ATRX ( 67 ), and PML ( 68 ) in mutant HSV- and HCMV-infected cells. We also observed restored expression of proliferation-linked genes upon knockout of SP100 or SP140L , suggesting that foreign DNA detection may repress proproliferative genes. For example, RSAD2, an ISG highly expressed during LPKO compared to WT EBV infection, influences cellular metabolism and mitochondrial function ( 69 , 70 ). IRF proteins can also target metabolic genes ( 71 , 72 ), and several IRFs were identified as potential mediators of gene changes in LPKO-infected cells. These findings suggest that Sp100 and Sp140L impose an antiviral state by regulating key viral genes and repressing cell cycle progression to create an antiproliferative environment unfavorable for viral infection.

AlphaFold3 models predicted that an essential alpha helix in the C-terminal Y domain of EBNA-LP associates with the CARD domains of Sp100 and Sp140L. Our previous work identified leucine-rich motifs present in the W and Y exons as important for EBNA-LP to associate with the cellular transcription factor YY1 ( 40 ). This motif is similar in sequence to both the LXXLL motif found in many cellular proteins that regulate transcription ( 73 ) and the “flexiNR” encoded in many viral proteins with transcription regulation functions ( 74 ). Our findings support the concept that viral proteins have evolved short linear motifs, generally less than 10 residues long in which a few key residues define affinity and specificity to mimic or hijack cellular protein functions ( 75 , 76 ).

This study was strengthened by the use of primary cells rather than cancer cell lines, as PML-NBs functions can be disrupted in cancer cells, potentially affecting physiological relevance ( 65 ). However, limitations exist in performing knockouts in primary B cells. To reduce cell death, B cells are infected with EBV only 2 h posttransfection. Since EBNA-LP is one of the earliest expressed EBV latency proteins, turnover of targeted proteins may be incomplete when EBNA-LP is expressed, even if genomic editing has occurred. While the HVS assay using fibroblasts to generate knockout cell lines prior to infection avoids this temporal disadvantage, achieving complete knockout remained challenging even in cells where the proxy target, CD46, was lost. Despite these limitations, our assays clearly demonstrate that both SP100 and SP140L restrict these diverse herpesviruses, with knockout of SP140L showing stronger phenotypic rescue.

The SP140L gene was formed by a gene duplication and cross-over event between SP100 and SP140 shortly before primate emergence ( 77 ). The Sp140L N-terminal CARD domain is over 90% identical to that of Sp100, while its C-terminal SAND and PHD/Bromodomains resemble those of Sp140 ( 77 ). Sp140L lacks the exons that in Sp100 encode SUMOylation and HP1 binding motifs but retains the exons for binding the high mobility group protein HMGB2. Given that the CARD domain is responsible for dimerization ( 78 , 79 ) or multimerization ( 80 ) of Sp100, and that Sp100 and Sp140L colocalize in cells ( 77 ), and that Sp100 can form homotypic interactions with the more distantly related protein Sp110 ( 81 ), we anticipate they interact interchangeably, through this domain. This is supported by our findings that knockout of either gene alone was sufficient for rescue of transformation. Whether Sp140L arose from an evolutionary arms race with viral antagonists of Sp100 remains a possibility. As knockout of SP140 did not enhance transformation of either WT- or LPKO-infected B cells, further analysis of the molecular differences between Sp140L and Sp140 – and the key domains responsible for DNA virus restriction - require further investigation. Additionally, expression differences between Sp100 and Sp140L may contribute to the dependency of EBNA-LP in naïve, but not memory B cell transformation as at the RNA level, naïve B cells more highly express both Sp100 and Sp140L ( 82 ).

Based on Sp100’s properties, we propose that Sp100 and Sp140L sense and suppress viral genomes through multiple mechanisms. The SAND domain of Sp100 preferentially binds unmethylated CpG dinucleotides ( 83 ), while its PHD/Bromodomain contains bulky residues where other PHD/Bromodomains have binding pockets for histone modifications ( 84 ). CARD domains commonly multimerize to activate immune signaling ( 85 ). Speckled proteins Sp100, Sp110 ( 81 ), and Sp140 ( 86 , 87 ) also associate on cellular chromatin at gene promoters, for instance, Sp100A at ISGs ( 64 ), with Sp140 localizing to sites enriched in H3K27me3 to maintain heterochromatin ( 86 , 87 ). We hypothesize therefore that the SAND domain of Sp100 and Sp140L recognize unmethylated incoming viral DNA as foreign, perhaps in combination with the PHD/Bromodomain recognizing features of newly assembled chromatin. Multimerization of Sp100 and Sp140L along the viral genome could block transcription either directly or by recruiting repressors to promote heterochromatinization of the viral genome. Simultaneously, these proteins may relocate to ISG promoters or nucleate signaling that activates ISG transcription and cell cycle checkpoints.

In conclusion, we identify Sp140L as a restriction factor of herpesviruses, and likely all DNA viruses, targeted by critical viral proteins of primate viruses, but may represent a significant barrier to any virus with a nuclear DNA phase of its lifecycle that transmits from nonprimate mammals. Sp140L not only restricts viral infection by preventing transcription of key viral genes, but also promotes an ISG response that suppresses cellular proliferation.

Materials and Methods

WT and LPKO viruses were prepared from 293-EBV producer lines and Raji Green Unit titer was obtained as previously described ( 31 ). In brief, LPKO virus refers to LPKO w from Szymula et al. ( 31 ) which contains a stop codon in exon W2 of every internal W repeat and WT EBV refers to WT w which contains a repaired stop codon in the first W1 exon found in the parental B95-8 virus that (unlike B95-8) allows robust EBNA-LP expression ( 88 ). Isolated B cells were infected by adding WT or LPKO virus at a ratio of 0.2 Raji Green Units per cell and incubating at 37 °C for 1 h. Following incubation, cells were pelleted and resuspended in Roswell Park Memorial Institute medium 1640 supplemented with 20% heat-inactivated fetal bovine serum (Corning) (R20 media).

For HVS viruses, a bacterial artificial chromosome (BAC) clone of the HVS A11-Sac strain (which lacks the HVS transforming genes) ( 89 ) was modified using RecA-mediated recombineering to replace the AgeI-BlpI fragment of dsRed1 ORF with a mNeonGreen-T2A-Luciferase fusion. This generated two BAC clones (A and B) that were each recombineered to delete the ORF3 open reading frame, introducing a STOP codon (alongside SalI and Cla I restriction sites) after the 5th amino acid of ORF3, and deleting all but the last 237 nucleotides of the ORF3 coding sequence, thereby generating two independent HVS-ORF3KO BACs (A-ORF3 KO and B-ORF3 KO) (as validated in SI Appendix , Fig. S8 A ). To generate virus, HVS BAC is transfected into – using a peptide–lipofectin complex ( 90 ) – or inoculated onto Owl Monkey kidney cells. The supernatant is harvested when the cell monolayer is largely destroyed (typically after 4 to 7 d) and centrifuged to remove cell debris. Detailed methodologies used in these studies are given in SI Appendix .

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)