Abstract

   U6 small nuclear RNA (U6 snRNA), a critical spliceosome component
   primarily found in the nucleus, plays a vital role in RNA splicing. Our
   previous study, using the simian immunodeficiency virus (SIV) macaque
   model, revealed an increase of U6 snRNA in plasma extracellular
   vesicles (EVs) in acute retroviral infection. Given the limited
   understanding of U6 snRNA dynamics across cells and EVs, particularly
   in SIV infection, this research explores U6 snRNA trafficking and its
   association with splicing proteins in the nucleus, cytoplasm, and EVs.
   We observed a redistribution of U6 snRNA from the nucleus to EVs
   post-infection, accompanied by distinct protein profile changes and
   alterations in nucleic acid metabolism and spliceosome pathways. In
   addition, U6 machinery proteins changed in cells and EVs in a
   contrasting manner. The redistribution of U6 and related proteins we
   observed could be part of a viral strategy to redirect host splicing
   machinery, suggesting that U6 may have regulatory roles and be part of
   retroviral infection signature.
     __________________________________________________________________

   In HIV infection, U6 snRNA leaves the nucleus and is packaged into
   extracellular vesicles to aid in intercellular communication.

INTRODUCTION

   Extracellular vesicles (EVs) are nanosized membranous vesicles released
   from most cells, playing crucial roles in intercellular communication
   ([38]1, [39]2). They have emerged as critical factors in HIV
   pathogenesis and host-virus interactions, functioning similarly to
   retroviruses by transporting various bioactive molecules between cells,
   including both host and viral components ([40]3–[41]5). EVs encapsulate
   diverse RNA species, such as mRNAs ([42]6, [43]7), microRNAs (miRNAs)
   ([44]8, [45]9), and other noncoding RNAs ([46]10, [47]11), alongside
   viral RNA fragments ([48]12). Recent research indicates that these RNAs
   within EVs from different sources can either facilitate or inhibit HIV
   infection ([49]13). Our group’s latest high-throughput sequencing
   studies also identified EV-associated RNA profiles involved in immune
   regulation and inflammatory responses in HIV infection ([50]14).

   HIV manipulates host cellular machinery to integrate its genetic
   material into the host’s genome, producing spliced and unspliced mRNA
   forms essential for viral protein production and replication ([51]15,
   [52]16). In addition, HIV also alters the splicing patterns of host
   genes, leading to altered gene and RNA levels in both host cells
   ([53]8, [54]17) and the EVs ([55]14, [56]18, [57]19) they release.
   Evidence indicates that HIV also encapsulates host RNAs. Several
   polymerase III (Pol III) transcripts such as Y RNAs, 7SK RNA, U6 snRNA,
   and cellular mRNAs are reported to be selectively enriched in HIV-1
   particles ([58]20–[59]22). Given their overlapping contents and
   physical characteristics, distinguishing between EVs and virions (or
   “host EVs” and “viral EVs”) after release can be challenging ([60]23,
   [61]24). Consequently, RNAs that have previously been reported in
   association with EVs or virions may not be specific to either class of
   particle.

   Previously, we revealed an association between plasma EV U6 small
   nuclear RNA (U6 snRNA) levels and the viral peak during acute SIV
   infection in pigtailed macaques (Macaca nemestrina) co-inoculated with
   the neurovirulent SIV-17EFr clone and the immunosuppressive SIV-B670
   swarm ([62]18). Although several studies have indicated alterations in
   EV-associated U6 snRNA in cancer ([63]25–[64]27), the dynamics of EV U6
   snRNA in other disease states, including viral infection, remain
   largely unexplored. As a crucial spliceosome component typically
   located in the nucleus, U6 snRNA is also known to interact transiently
   with ribonucleoproteins ([65]28) or U bodies ([66]29) in the cytoplasm.
   Since U6 snRNAs play a vital role in pre-mRNA splicing ([67]30,
   [68]31), particularly in catalyzing splicing reactions after
   spliceosome assembly, changes in the quantities and localization of U6
   snRNAs may contribute to altered transcript alternative splicing and
   increased intron retention ([69]31) during HIV infection. Furthermore,
   the release of these RNAs into EVs is still poorly defined.

   Our study thus aims to elucidate how U6 snRNA and its associated
   proteins are distributed across various cellular compartments and EVs
   in retroviral infection. In SIV-17EFr– and SIV-B670–infected human
   CEMx174 cells, we noted a unique distribution pattern of U6 snRNA,
   marked by its reduction in the nucleus and increase in EVs. This
   pattern was specifically found in SIV and not in hepatitis B virus
   (HBV) infection. Additionally, changes in proteins related to nucleic
   acid metabolism, spliceosome pathways, and U6 snRNA machinery were
   detected in both the nucleus and EVs of SIV-infected cells. These
   findings suggest that retroviruses like HIV and SIV may use a strategy
   to redirect the host splicing machinery from the nucleus to EVs during
   infection, potentially affecting RNA splicing and the viral life cycle.
   This study thus provides valuable insights into the interplay between
   HIV infection and cellular RNA processing mechanisms.

RESULTS

Characterization of EVs derived from uninfected and SIV-infected cells

   Infection of CEMx174 cells with SIV-17EFr and SIV-B670 resulted in a
   decrease in cell viability 3 days post-infection, with the reduction
   being more pronounced in SIV-B670–infected cells (fig. S1A). The
   infectivity of virions produced from these cells, assessed by measuring
   luciferase activity in Tat-responsive LuSIV indicator cells ([70]32),
   increased progressively for both viral strains, with SIV-B670 showing
   consistently higher luciferase activity induction compared with
   SIV-17EFr from day 1 to day 3 post-infection (fig. S1B).

   EVs from uninfected (UI), SIV-17EFr–infected, and SIV-B670–infected
   cells were separated and characterized. Medium processing controls (MC)
   were included to assess potential contamination of proteins and bovine
   serum–derived EVs from the cell culture medium (CCM). Particle counts
   of EVs were normalized based on both particles per million viable cells
   and milliliter (ml) of medium. More particles were observed in the
   three B670-infected samples compared with the UI samples, although the
   difference did not reach statistical significance due to the small
   sample size and high variability ([71]Fig. 1A, left). The MC showed a
   significantly lower particle concentration (particles/ml medium)
   compared with EVs ([72]Fig. 1A, right). Particle size distribution
   analysis revealed a higher number of smaller particles from
   B670-infected cells ([73]Fig. 1B). Transmission electron microscopy
   (TEM) images revealed oval/round particles consistent with EV
   morphology, as well as smaller protein aggregates in all groups
   ([74]Fig. 1C). The levels of proteins that are commonly reported to be
   associated with intracellular components, fetal bovine serum (FBS),
   EVs, and immune cells were examined by mass spectrometry (MS) ([75]Fig.
   1D). Intracellular proteins were more enriched in the cytoplasm than in
   EVs and MC in all three groups. Conversely, EV general and immune cell
   proteins were more enriched in EVs than cytoplasm and MC. FBS markers
   were highly enriched in MC compared with cytoplasm and EVs.

Fig. 1. Characterization of EVs from uninfected and SIV-infected cells.

   [76]Fig. 1.
   [77]Open in a new tab

   (A) Particle concentrations of EV preparations from UI,
   SIV-17EFr–infected, and SIV-B670–infected cells were measured by NFCM.
   Particle concentration for each group was normalized by particles per
   million viable cells (left) and per milliliter of CCM (right). (B) Size
   distributions of EVs (uninfected and infected) were measured by NFCM
   and presented as particle numbers in specific size ranges. (C) EVs and
   MC were visualized by negative staining TEM (scale bar ,  100 nm). TEM
   is representative of 10 images taken of each sample from three
   independent samples. (D) Unsupervised hierarchical clustering of
   cytoplasm, EVs, and MC samples based on the levels of intracellular,
   FBS, EV, and immune cell–related markers. (E) The levels of CD9, CD63,
   and CD81 in EV and MC samples were detected by MACSplex. (F)
   Fluorescence detection of CD63 and CD81 on EVs captured by CD63 (left)
   and CD81 (right) antibody spots using SP-IRIS. The percent of CD63,
   CD81, and double-positive particles were calculated versus the total
   particle number captured on each spot. [(A) to (F)] Data are presented
   as mean ± SD. ns (no significant difference, P  >  0.05), *P ≤ 0.05 by
   unpaired t test.

   The levels of canonical EV surface markers CD9, CD63, and CD81 were
   detected on EVs and MC by MACSplex ([78]Fig. 1E) and single-particle
   interferometric reflectance imaging sensor (SP-IRIS) ([79]Fig. 1F and
   fig. S2) platforms to evaluate EV presence and characteristics. CD63
   and CD81 were significantly more abundant in EVs compared with MC,
   while CD9 exhibited similar levels in both, indicating an association
   of CD9 with CCM contamination ([80]Fig. 1E and fig. S2). Furthermore,
   particles captured on the SP-IRIS chip through CD63 and CD81 were
   subsequently probed with fluorescently labeled CD63 and CD81 antibodies
   to confirm EV subtype differences based on tetraspanin profiles between
   uninfected and infected cells ([81]Fig. 1F). More signal for both CD63
   and CD81 was observed in B670 EVs when captured on CD63 and CD81
   captured spots, respectively. Additionally, more captured particles
   were positive for either CD63 or CD81, while a small percentage
   exhibited dual positivity for both CD63 and CD81 in all samples.

U6 snRNA level and distribution alteration post-SIV infection: Increased in
EVs and reduced in the nuclei

   RNA concentrations of EVs and cell fractions were measured and
   normalized to ensure consistent RNA input within each batch for the
   subsequent quantitative polymerase chain reaction (qPCR) assays (fig.
   S3A). The levels of U6 and U1 snRNAs were assessed using the SYBR-based
   qPCR method in EVs obtained from UI, cells infected with SIV-17EFr and
   SIV-B670, and in the MC ([82]Fig. 2A and fig. S3B). Both U6 and U1
   snRNAs were significantly lower in MC, amplifying at approximately five
   cycles later on average (equivalent to a 32-fold reduction) compared
   with EVs, indicating minimal snRNA contribution from the CCM (fig.
   S3B). The abundance of EV U6 snRNAs was significantly higher in both
   17EFr and B670 compared with UI, while U1 snRNA levels were similar
   ([83]Fig. 2A). The elevated abundance of U6 snRNA following SIV
   infection was confirmed by both TaqMan- and SYBR-based qPCR methods,
   with an average fold change >3 for both virus strains ([84]Fig. 2B).
   Additionally, U6 and U1 snRNA levels were found to be higher in EVs
   from HIV Rf– and BAL-infected cells versus UI by the TaqMan-based qPCR
   method (fig. S4).

Fig. 2. U6 and U1 snRNA detection in cell fractions and EVs during SIV
infection.

   [85]Fig. 2.
   [86]Open in a new tab

   (A) qPCR validation of U6 (left) and U1 snRNA (right) in EVs from UI,
   cells infected with SIV-17EFr and B670, and in MC using the SYBR-based
   qPCR method. Lines connect data points within each batch to illustrate
   results using equal RNA amounts. (B) Fold change of EV U6 snRNA in
   infection compared with no infection using TaqMan- and SYBR-based qPCR
   methods. (C) qPCR validation of U6 (left) and U1 (right) snRNAs in
   cytoplasm and nucleus of UI, 17EFr-infected, and B670-infected groups
   using the SYBR-based qPCR method. Lines connect data points within each
   batch to illustrate results using equal RNA amounts. (D) Confocal
   microscopy images (scale bar, 20 μm; magnification, ×40) of uninfected
   and infected cells stained with TR for U6 snRNA and DAPI for nuclear
   DNA. Imaris was used to identify and count voxels, where TR and DAPI
   signals overlap. The number of colocalized voxels was quantified, and
   the percentage of colocalized voxels relative to the total number of
   voxels in the defined cell area (Colocalization %) was calculated and
   presented. [(A) to (D)] Data are presented as mean ± SD. ns (no
   significant difference, P  >  0.05), *P ≤ 0.05, ****P ≤ 0.0001 by
   unpaired t test.

   To determine whether SIV infection affects the cellular abundance and
   localization of snRNAs, uninfected and infected cells were fractionated
   into cytosolic and nuclear fractions. Distribution of tubulin and
   glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (cytoplasm) and lamin
   B (nucleus) confirmed successful separation (fig. S5). Both U6 and U1
   snRNAs exhibited significantly lower abundance in the nuclei of cells
   infected with 17EFr and B670, with no difference in the cytoplasm, as
   measured by the SYBR-based qPCR method ([87]Fig. 2C). The reduction of
   U6 snRNA in the nucleus was further confirmed using the TaqMan-based
   qPCR method (fig. S3C). Cq values were then normalized to the geometric
   mean of internal reference miRNAs (miR-29a, miR-342, and miR-181a) for
   additional validation (fig. S3D). Colocalization analysis of U6 snRNA
   and DAPI (4′,6-diamidino-2-phenylindole) staining (nuclear DNA) showed
   a reduction in the overlapping signal between U6 snRNA and DAPI signals
   post-infection with both virus strains ([88]Fig. 2D).

   HBV-infected plasma EVs were also assessed to determine whether the
   elevation of EV U6 snRNA is a phenomenon common to blood-borne viruses.
   Despite a higher concentration of particles in EV preparations from
   HBV-infected plasma with a higher viral load (10 × 10^8 copies/ml)
   (fig. S6, A and B), the level of U6 snRNA remained similar among
   uninfected individuals, HBV patients with a lower viral load (10 × 10^3
   copies/ml), and those with a higher viral load (fig. S6C).

SIV infection and proteins in cytoplasm, nucleus, and EVs

   We subsequently used label-free MS to explore protein profile
   alterations following SIV infection in the cytoplasm, nuclei, and EVs,
   seeking potential explanations for the observed changes in U6 snRNA.
   Proteins (3158, 28.6%) were detected in all three compartments, 2339
   proteins (21.2%) were found in both the cytoplasm and nuclei, and a
   subset of proteins was exclusively detected in a single fraction
   ([89]Fig. 3A). Gene ontology (GO) analyses by STRING were used to
   determine cellular compartments ([90]Fig. 3B, left) and biological
   processes ([91]Fig. 3B, right) associated with fraction-exclusive
   proteins. Proteins exclusive to EVs were primarily enriched for terms
   associated with vesicles, EVs, and the extracellular matrix. Terms such
   as cytoplasm and cytosol were enriched in both EVs and cytoplasm, while
   organelle-related terms and intracellular anatomical structures were
   enriched in both cytoplasm and nucleus. Mitochondria and organelle
   membrane terms were specifically enriched in cytoplasm, whereas
   nucleus-related terms like nuclear lumen and nucleoplasm were
   exclusively enriched in nucleus ([92]Fig. 3B, left). Additionally,
   proteins exclusively detected in different fractions showed enrichment
   for distinct functions. For instance, metabolic processes were enriched
   in the cytoplasm, DNA and RNA processes in the nucleus, and multiple
   cellular processes in EVs ([93]Fig. 3B, right).

Fig. 3. Proteomic differences in SIV infection in cytoplasm, nucleus, and
EVs.

   [94]Fig. 3.
   [95]Open in a new tab

   (A) Venn diagrams of all proteins identified by proteomics in
   cytoplasm, nucleus, and EVs. (B) Top 10 cellular compartments and
   biological processes ranked by identified exclusive protein number in
   the cytoplasm, nucleus, and EVs by the STRING. (C) PCA based on the
   proteome of samples. (D) Number of DE proteins in SIV-17EFr and B670
   infection compared with uninfected groups in cytoplasm, nucleus, and
   EVs. (E) The UpSetR analysis presents DE proteins across cytoplasm,
   nucleus, and EVs. Each row represents a different comparison, with
   intersections marked by black dots connected by vertical lines in the
   matrix. The bottom charts show the total DE proteins in each
   comparison, while the top charts indicate the number of proteins
   overlapping between comparisons. STRING enrichment analysis identified
   biological processes associated with proteins dysregulated in the
   intersection of cytoplasm and nucleus and nucleus and EVs.

   Principal components analysis (PCA) of the proteome revealed distinct
   separation among various fractions, including cytoplasm, nucleus, EVs,
   and MC. Within each fraction, the proteome of UI samples exhibited a
   separation from both infected groups ([96]Fig. 3C). Differentially
   expressed (DE) proteins in response to SIV infection were further
   examined in the cytoplasm, nuclei, and EVs ([97]Fig. 3D). Nuclei
   exhibited the greatest number of proteins with altered abundance,
   followed by EVs. Comparing the two virus strains, B670 infection
   induced a higher number of altered proteins than 17EFr, but with a
   substantial portion of them being common to both virus strains. The DE
   proteins induced by both virus strains were visualized to show unique
   and overlapping proteins among three fractions by UpSetR analysis
   ([98]Fig. 3E). Most of these proteins were exclusively identified in a
   single fraction, with a small subset demonstrating overlap between
   fractions. These included proteins that exhibited higher abundance in
   SIV infection, with 136 overlapping cytoplasm and nuclei and 108 in EVs
   and nuclei. GO biological process enrichment revealed that all of these
   proteins were associated with metabolic pathways. Additionally,
   proteins increased in infected nuclei and EVs were linked to nucleic
   acid processing terms, including RNA transportation and DNA
   replication.

Proteins related to nucleic acid metabolic and spliceosome pathways in
SIV-infected nucleus and EVs

   Considering that most proteins affected by SIV infection showed
   alterations exclusively in one fraction, we used a stepwise pipeline
   ([99]Fig. 4A) to identify potential pathways associated with DE
   proteins in the cytoplasm, nucleus, and EVs. DE proteins in the three
   fractions were clustered into protein modules using weighted gene
   correlation network analysis (WGCNA), and these modules were further
   interpreted through protein-protein interaction (PPI) networks.
   Specifically, 5, 16, and 10 correlation modules were constructed in the
   cytoplasm, nucleus, and EVs, respectively, based on DE gene expression
   levels. Distinct color labels assigned by the WGCNA software were used
   to differentiate these modules, as shown in [100]Fig. 4 (B to D). The
   proteins associated with each module are listed in table S1.

Fig. 4. Protein profiling of SIV infection reveals nucleic acid metabolic and
spliceosome pathways.

   [101]Fig. 4.
   [102]Open in a new tab

   (A) Bioinformatics pipeline for DE protein correlation analysis,
   clustering, and functional enrichment. (B to D) Left panels:
   Identification of protein modules associated with sample traits in
   cytoplasm (B), nucleus (C), and EVs (D). Protein modules were
   identified and assigned distinct colors by the WGCNA algorithm to
   visually differentiate clusters of coexpressed proteins. Module-trait
   relationships were determined by bi-weight mid-correlation between
   module eigenprotein expression and sample traits, including infection,
   U6 snRNA, and EV count. Correlation coefficients (r) are indicated on
   the top with corresponding P values in brackets below. Positive
   correlations are highlighted in yellow, while negative correlations are
   in blue. Right panels: Enriched pathways of proteins in modules
   significantly correlated with infection (r > 0.9 or < −0.9, P < 0.05)
   in the cytoplasm (B), nucleus (C), and EVs (D). Positively correlated
   modules are highlighted in yellow, while negative correlations are
   shown in blue. Significantly enriched pathways by KEGG and STRING
   analysis (FDR < 0.05) are shown. (E) Protein number involved in nucleic
   acid processing pathways in cytoplasm, nucleus, and EVs. (F) Detected
   protein interaction modules involved in spliceosome pathway in nucleus
   and EVs.

   To explore the physiological significance of the color-labeled modules
   and explore their correlation with U6 snRNA differences, WGCNA was used
   to calculate their correlation with sample traits, including SIV
   infection, U6 snRNA level, and EV count ([103]Fig. 4, B to D, left
   panels). In the cytoplasm, all protein modules displayed a significant
   correlation (P < 0.05) with SIV infection but were not significantly
   correlated with U6 snRNA levels (P > 0.05) ([104]Fig. 4B, left panels).
   In the nuclei, all protein modules exhibited significant correlations
   with both SIV infection and U6 snRNA levels. Notably, modules
   positively correlated with SIV infection (r > 0) had a negative
   correlation with U6 snRNA levels (r < 0), whereas modules negatively
   correlated with SIV infection showed a positive correlation with U6
   levels ([105]Fig. 4C, left panels). For EVs, all modules exhibited a
   significant correlation with SIV infection, with 6 of 10 also
   correlated with U6 snRNA levels, while none showed a correlation with
   EV count. In contrast with the nuclei, modules positively or negatively
   correlated with SIV infection also exhibited a corresponding
   positive/negative correlation with U6 snRNA levels ([106]Fig. 4D, left
   panels).

   Significant modules, with a correlation coefficient > 0.9 (versus SIV
   infection), were selected for pathway enrichment analysis across all
   three fractions ([107]Fig. 4, B to D, right panels). Comparing common
   pathways in the three fractions, modules positively correlated with SIV
   infection (colored yellow) in the cytoplasm and nucleus were enriched
   for metabolic processes and immune system processes. Modules negatively
   correlated with SIV infection (colored blue) in the cytoplasm and
   nucleus but positively correlated with SIV infection in EVs were
   enriched for nucleic acid metabolic processes, particularly
   spliceosome, RNA transport, and mRNA surveillance pathways in both the
   nucleus and EVs. DE proteins enriched for common pathways between EVs
   and cells exhibited opposite changes in response to SIV infection in
   EVs compared with both cytoplasm and nucleus ([108]Fig. 4E). Given that
   U6 snRNA is a crucial component of the spliceosome, we uncovered
   potential U6 snRNA–related proteins affected by SIV infection by
   identifying PPI networks for the spliceosome pathway, in both the
   nucleus (n = 9, purple and light cyan modules) and EVs (n = 35, yellow,
   green, and turquoise modules) ([109]Fig. 4F).

U6 snRNA machinery proteins in SIV-infected cells and EVs

   Proteins reported to regulate U6 snRNA cellular level and location were
   further detected in both cells and EVs through proteomic analysis
   ([110]Fig. 5A) and Western blot (WB) ([111]Fig. 5B and fig. S7). On the
   basis of proteomics data, log[2] fold change (infected versus
   uninfected) was calculated for LSM1, LSM8, USP39, SART3, SMN, DIS3, and
   CD63 for cytoplasm, nucleus, and EVs ([112]Fig. 5A). Most of these
   proteins exhibited opposite changes in EVs compared with both cytoplasm
   and nucleus following SIV infection, including LSM8 (up-regulated in
   EVs but down-regulated in the nucleus), USP39, SART3, and DIS3
   (up-regulated in EVs but down-regulated in the cytoplasm), and CD63
   (down-regulated in EVs but up-regulated in cytoplasm and nucleus). SMN
   was significantly down-regulated only in the cytoplasm. Further WB
   analysis of whole-cell lysate, cytoplasm, and nucleus fractions
   confirmed the dysregulation of LSM8, USP39, SART3, SMN, DIS3, and CD63
   in response to SIV infection ([113]Fig. 5B). While proteomics indicated
   the presence of these U6 snRNA machinery proteins in EVs, WB analysis
   detected LSM1, DIS3, and CD63 (fig. S7), but not SMN and LSM8, possibly
   due to antibody sensitivity and limitations in sample volume (since
   most EV samples were used for proteomics). Of the 37 EV surface
   proteins related to potential cellular sources examined using the
   MACSplex platform, five showed significant differences between infected
   and uninfected states ([114]Fig. 5C). Specifically, T cell and B cell
   markers CD4 and CD19 decreased in infection, indicating that cell death
   limited EV release from these specific cell types. Conversely, immune
   activation–related markers CD69, CD2, and CD25 increased on the surface
   of EVs, reflecting potential changes in the immune cell response.

Fig. 5. U6 snRNA–associated proteins and EV surface markers in SIV infection.

   [115]Fig. 5.
   [116]Open in a new tab

   (A) Log[2] fold change of U6 snRNA–associated proteins between SIV-B670
   infection and no infection in cytoplasm, nucleus, and EVs (proteomics).
   ns (no significant difference, P  >  0.05), *adjusted P values < 0.05
   (Benjamini-Hochberg method). (B) WBs of U6 snRNA–associated proteins in
   whole-cell, cytoplasm, and nucleus lysates; representative of three
   independent cell fractionation experiments. (C) Log[2] fold change of
   CD4, CD19, CD69, CD2, and CD25 (SIV-17EFr and SIV-B670 infection
   compared with uninfected controls) in EVs (MACSplex).

DISCUSSION

   On the basis of previous findings from the SIV-infected macaque model
   indicating increased levels of extracellular U6 snRNA in infection
   ([117]18), we examined changes in U6 snRNA and associated proteins in
   cytoplasmic and nuclear cellular compartments as well as EVs of human
   immune cell lines infected with two virus strains. Decreased U6 snRNA
   in the nucleus corresponded with increased U6 in EVs, suggesting
   relocation and cellular export during infection.

   U6 snRNA is a critical spliceosome component, primarily responsible for
   catalyzing the splicing of nuclear pre-mRNA ([118]33). Its essential
   role in maintaining normal splicing processes has been demonstrated by
   studies showing that cellular U6 snRNA knockdown leads to increased
   intron retention for specific genes ([119]31). Moreover, transient
   exogenous expression of U6 snRNA has been shown to slightly reduce
   mis-splicing rates of certain transcripts, further underscoring its
   role in regulating splicing fidelity ([120]34). These findings suggest
   that reduced U6 snRNA levels within cells during infection could
   contribute to splicing dysregulation and altered gene expression
   patterns. In contrast, the observed increase of U6 in EVs raises
   important questions about its mechanism of incorporation, whether
   through selective sorting or as a consequence of infection-related
   cellular damage. Additionally, the potential functional impact of U6
   snRNA in recipient cells remains unknown, highlighting the need for
   further investigations.

   Protein profiling, in addition to revealing broad shifts across
   compartments and during infection, further suggested that U6 snRNA
   distribution changes are part of wider alterations in virus-induced
   splicing dysregulation. Aligned with previous studies, the proteome of
   EVs was notably different from that of intracellular compartments
   ([121]35, [122]36). Furthermore, the proteomes of EVs, the nucleus, and
   the cytoplasm were all distinctively altered in infected cells,
   consistent with previous reports on HIV infection ([123]37, [124]38).
   Several U6-related proteins were affected during infection, notably
   LSM8, essential for U6 snRNA nuclear localization ([125]39), and SART3,
   which forms an intermediate complex with U6 snRNA and is essential for
   spliceosome assembly ([126]40). In addition, other key components of
   RNA processing and spliceosome were also altered during infection, such
   as USP39, a component protein of the U4/U6.U5 small nuclear
   ribonucleoprotein (snRNP) complex ([127]41), and DIS3, which plays an
   important role in regulating the expression of specific spliceosome
   snRNAs ([128]42). These findings underscore the notable impact of viral
   infection on U6 snRNA and broader RNA processing pathways.

   In parallel, changes in U1 snRNA levels in the nucleus also reflect
   broader disruptions in RNA processing and spliceosome dynamics during
   infection. Both U1 and U6 snRNAs exhibited decreased nuclear levels;
   however, only U6 increased in EVs, suggesting distinct regulatory or
   sorting mechanisms. Previous studies have indicated tissue-dependent
   expression and regulation of snRNAs ([129]31, [130]43), and the
   recycling of snRNPs could affect their cellular reuse and differential
   release into EVs ([131]30, [132]44). Furthermore, variations in
   subtypes and levels of snRNA-associated proteins observed in the EV
   proteome, such as Prp24 and Lsm2-8 for U6, and SNRP70 for U1, may play
   a role in the selective packaging and transport of these snRNAs. To
   further explore these mechanisms, future studies should include
   investigations into RNA recycling and degradation pathways, and
   functional analyses of specific snRNP-associated proteins.

   A question that arises from these results is whether they are due to
   general inflammatory or antiviral responses or are, instead, specific
   to retroviruses. While several studies have highlighted the impact of
   bacterial infections on the spliceosome and RNA splicing ([133]45,
   [134]46), research into how these components change during viral
   infections ([135]47), especially in HIV ([136]47, [137]48), remains
   limited. While we cannot rule out general responses, the increases in
   EV U6 snRNA that we report here and in our previous study ([138]18) do
   not hold for HBV-infected plasma EVs. Thus, at least some viral
   infections do not appear to produce the same shift in U6 distribution.

   Possible effects of techniques and artifacts from CCM components should
   be carefully considered in evaluating results, especially when
   low-abundance materials like EV samples are involved. To be sure, our
   EV separation by size exclusion chromatography (SEC) has been
   previously validated ([139]49), and we also rigorously evaluated the EV
   samples per MISEV guidelines ([140]1, [141]2). As recommended, we
   implemented MC. These controls revealed the presence of EVs, likely
   originating from FBS, that were characterized by high concentrations of
   FBS components. However, these EVs lacked detectable levels of
   cell-specific or typical EV markers, such as the tetraspanins CD63 and
   CD81. Additionally, both U6 and U1 snRNAs were significantly lower in
   these control samples, as indicated by the very high cycle of
   quantitation numbers in qPCR. Thus, although FBS contaminants were
   present, their consistency across the uninfected and infected groups,
   only trace amounts of measurable U6, supports the validity of our
   conclusions.

   The relationship between host EVs and virions also bears discussion.
   The increase of U6 in EV-enriched fractions isolated by SEC was also
   found in EV subtypes as well as virion-enriched fractions that were
   further separated by density gradient ultracentrifugation ([142]18).
   The difficulty of fully separating EVs from virions, due to their
   overlapping contents and physical characteristics, means that previous
   reports focusing on one category or the other may not be exclusive to
   either entity ([143]13, [144]23, [145]24). Specifically, Pol III
   transcripts, including U6 snRNAs, were reported to be preferentially
   loaded into both EVs separated from uninfected cells ([146]50) as well
   as virions that might be copurified with EVs in gradient-based
   isolation methods ([147]20, [148]22, [149]51). This overlap aligns our
   findings with previous reports, demonstrating an increased transfer of
   U6 snRNA to extracellular compartments during infection, potentially
   affecting both EVs and virions. Notably, CD63, a common marker for both
   EVs and HIV ([150]52, [151]53), showed altered expression in our
   samples, reinforcing the connection between EVs and HIV. Packaging of
   snRNAs into retroviral virions can lead to snRNA reverse transcription
   and integration into the host genome ([152]21). Potentially, EVs
   produced during infection might perform similar roles. Our findings
   highlight the complex interplay between EVs and virions in the context
   of HIV infection and potential shared contributions to RNA transfer and
   viral replication processes.

   In summary, our research demonstrates a connection between EVs, U6
   snRNA, RNA processing, and SIV infection. We characterized the
   increased packaging of nuclear proteins and RNAs into EVs or viral
   particles post-SIV infection. The changes observed in U6 snRNA and its
   associated proteins suggest that EVs play an important role in the
   dysfunctions and redirection of RNA processing that may be triggered by
   HIV infection. Future studies may include HIV-infected cell models and
   clinical samples, as well as examination of other retroviruses.
   Investigating the functions of U6 snRNA and its associated proteins in
   viral replication, particularly through the use of small-molecule
   inhibitors to disrupt SIV replication, could provide valuable insights
   into how viral replication influences host RNA processing.
   Additionally, exploring the mechanisms whereby EVs encapsulate RNA
   processing elements from the nucleus may provide deeper insights into
   the interaction between HIV and host cell RNA processing mechanisms.

MATERIALS AND METHODS

Cell culture

   The B cell/T cell hybrid CEMx174 (gift from J. Hoxie, University of
   Pennsylvania) ([153]54) was maintained in RPMI 1640 (Life Technologies)
   supplemented with 10% FBS (Atlanta Biologicals, Norcross, GA), 2 mM
   sodium pyruvate, 10 mM Hepes, gentamicin (0.5 mg/ml), and 2
   mM l-glutamine (sR10 medium). H9 and PM1 (T lymphocytic lines) were
   obtained from AIDS Reagent and American Type Culture Collection (ATCC)
   and maintained in complete 10% FBS RPMI medium with 10 mM Hepes, 1
   mM l-glutamine, and penicillin-streptomycin (1 mg/ml). Cells were grown
   at 37°C in a humidified 5% CO[2] incubator ([154]55).

Virus stocks and infections

   Simian immunodeficiency virus (SIV)/17EFr and Delta B670 virus swarms
   were prepared as previously described and stored at −80°C ([155]56).
   The CEMx174 cells were counted by Muse Cell Analyzer (Millipore
   Corporation), washed one time in RPMI, and resuspended to 1 × 10^6
   cells per well. SIV/17EFr and SIV/Delta B670 stocks were added
   separately to cells at 1 × 10^3 plaque-forming units (PFU) per well and
   incubated at 37°C for 72 hours. The infected CCM and cells were further
   split, incorporating uninfected cells to expand the infected culture.
   SIV infection was confirmed by cell syncytia formation, SIV p27 assay
   (ZeptoMetrix 0801169), and luciferase assay ([156]32).

   HIV-1 Rf and BAL stocks were generated from infected H9 and PM1 T
   lymphocytic cell lines, respectively, and stored at −80°C. HIV Rf or
   BAL stocks were added to H9 and PM1 cells at a concentration of 500
   ng/ml (p24) and incubated at 37°C for 4 to 6 hours. Cells were then
   rinsed several times with phosphate-buffered saline (PBS) and spun
   twice at 400g for 5 min each ([157]55).

Cell fractionation

   Infected and uninfected cells (around 2 × 10^7) were washed twice with
   PBS and subsequently lysed in 2 ml of ice-cold 1% Triton in Dulbecco’s
   PBS (DPBS) for 2 min. After pipetting up and down 10 times, 600 μl of
   the lysate was collected as whole-cell lysate. The remaining lysate was
   then subjected to centrifugation at 5000g for 5 min. The supernatant
   (cytosolic fraction) was collected (600 μl) and centrifuged again to
   eliminate any residual nuclei. The pelleted nuclear fraction was lysed
   once more in 1 ml of ice-cold 1% Triton. Following a 5000g, 5-min spin,
   supernatant was removed. To fully dissolve the nuclei, 600 μl 1% Triton
   was added to the pellets, followed by sonication in an ultrasonic ice
   bath at 20 kHz for three cycles of 2 min each, with a 30-s interval
   between each sonication.

EV separation

   CCM (90 ml) from uninfected, SIV/17EFr-infected, and SIV/Delta
   B670–infected cells, as well as medium-only controls (MC), were
   collected and differentially centrifuged at 1000g for 10 min and 2000g
   for 15 min at 4°C. Supernatants were then transferred into 10-kDa
   protein concentrators (Millipore-Sigma, UFC701008) and concentrated to
   10 ml. Retentate was then applied on the top of qEV10/70nm SEC columns
   (Izon Science, IC10-70) pre-rinsed with 150 ml of PBS. Fractions (5 ml)
   were collected by Izon automated fraction collectors (Izon Science).
   Fractions 1 to 4 (20 ml total) after the void volume (22.9 ml) were
   pooled as EV-enriched fractions and concentrated by 10-kDa molecular
   weight cutoff protein concentrators (Millipore-Sigma, UFC901008) to 550
   μl. For EV separation from HBV-infected clinical plasma samples, see
   Supplementary Text.

Nanoflow cytometry

   Concentration and size of EVs (1 μl, 1:1000 dilution) and MC (1 μl,
   1:100 dilution) were measured for 1 min by side scatter using nanoflow
   cytometry (NFCM) calibrated for concentration and size with 200-nm
   silica beads and nanosphere mixture (diameters of 68, 91, 113, and 151
   nm) as previously described ([158]57).

Transmission electron microscopy

   EV preparations and MC (8 μl, 1:2 diluted in DPBS) were absorbed onto
   glow-discharged 400-mesh ultrathin carbon-coated grids (EMS
   CF400-CU-UL) for 2 min, followed by three rinses in tris-buffered
   saline (TBS). Staining was performed using 1% uranyl acetate with 0.05%
   Tylose. After aspiration and drying, the grids were examined using a
   Philips CM120 instrument at 80 kV, and images were captured with an
   XR80 charge-coupled device (8 megapixels; AMT Imaging, Woburn, MA,
   USA).

Single-particle interferometric reflectance imaging sensor

   SP-IRIS was performed as described ([159]57). EVs and MC (1 μl) were
   diluted in 44 μl of incubation buffer (IB) and incubated overnight on
   ExoView tetraspanin chips (NanoView Biosciences). After 4 × 3 min
   washes in IB, chips were imaged (ExoView, NanoView) and analyzed
   (NanoViewer 2.8.10, NanoView).

EV analysis by bead-based multiplex flow cytometry assay (MACSplex)

   EV fractions (5 μl) were subjected to bead-based multiplex analysis
   [Miltenyi, MACSPlex EV Kit IO (for immuno-oncology), 130-122-209] by
   flow cytometry (BD LSRFortessa). Thirty-seven EV surface epitopes plus
   two isotype controls were tested according to the manufacturer’s
   protocol. The average APC medians of individual bead populations were
   used to determine the expression of EV surface markers.

Total RNA extraction

   RNA was extracted from 250 μl each of EVs and MC with TRIzol LS (Thermo
   Fisher Scientific, 10296028), and 200 μl each of whole-cell lysates,
   cytoplasm, and nucleus using TRIzol (Thermo Fisher Scientific,
   15596018). After homogenization, RNA was isolated by miRNeasy solutions
   (Qiagen, 217004) with Zymo-Spin columns (Zymo Research, C1003–50) per
   the manufacturer’s instruction. The concentration of extracted RNA was
   measured using a NanoDrop spectrophotometer. To ensure consistency
   across samples, the RNA concentration was adjusted to be equivalent
   within the uninfected, SIV-17EFr, and SIV-B670 groups within each
   experimental batch.

Individual quantitative PCR assays

   qPCR assays were performed using equal amounts of RNAs within each
   batch for U6 snRNA using TaqMan MicroRNA Assays (Thermo Fisher
   Scientific, Assay ID 001973) and miRCURY LNA miRNA PCR Assay (Qiagen,
   GeneGlobe ID YP00203907). qPCR assays for U1 snRNA were performed by
   miRCURY LNA miRNA PCR Assay (Qiagen, Assay GeneGlobe ID YP00203909).

Fluorescence imaging

   The U6 snRNA sequence was retrieved from the National Center for
   Biotechnology Information (NCBI) Reference Sequence Database (RefSeq
   accession number [160]NR_004394.1). Five probes targeting regions
   across the U6 snRNA sequence (atgtgctgccgaagcgagc, gtatcgttccaattttagt,
   ggggccatgctaatcttct, tttgcgtgtcatccttgcg, aaatatggaacgcttcacg) were
   designed using Stellaris Probe Designer (Biosearch Technologies). Each
   probe was synthesized with a 3′ amino modification to facilitate
   conjugation. The probes were pooled in equimolar concentrations and
   conjugated with Texas Red (TR) fluorophore (Invitrogen, T6134) followed
   by high-pressure liquid chromatography (HPLC) purification as
   previously described ([161]58). CEM174 cells were fixed using 4%
   formaldehyde and subsequently stored in 70% ethanol at 4°C. Before
   hybridization, cells were washed with a buffer containing 2× saline
   sodium citrate (SSC) solution (Ambion, AM9763), 20% formamide (Ambion,
   AM9342), and 2 mM ribonucleoside vanadyl complex (New England Biolabs,
   S1402S). Hybridization was performed by incubating the cells with U6
   snRNA probe sets (25 ng/μl) diluted in 100 μl of hybridization buffer
   [containing 10% dextran sulfate (Sigma, D8906), yeast tRNA (1 μg/μl;
   Invitrogen, 15401-029), 2 mM ribonucleoside-vanadyl complex, 0.02%
   ribonuclease-free bovine serum albumin (BSA) (Ambion, AM2618), and 20%
   formamide dissolved in 2× SSC] on coverslips in a humidified chamber
   overnight at 37°C. Post-hybridization, cells attached to coverslips
   were washed and stained with DAPI to visualize nuclear DNA (Sigma,
   D9542). The prepared samples were then mounted using established
   protocols ([162]59). The images were taken with 40×/1.30 PlanNeofluar
   oil objective on a Zeiss LSM 800 Confocal Laser Scanning Microscope.
   Initial overview images were acquired at ×20 magnification, followed by
   high-resolution z-stack imaging at ×40 across four to five distinct
   positions per slide. The acquired images were analyzed using Imaris
   software to assess colocalization. The “Colocalization” tool in Imaris
   was used to identify and count voxels, where TR and DAPI signals
   overlap. The number of colocalized voxels was quantified, and the
   percentage of colocalized voxels relative to the total number of voxels
   in the defined cell area (Colocalization %) was calculated and
   presented.

Mass spectrometry

Sample preparation, protein digestion, and peptide TMT labeling

   Cytoplasm and nucleus samples were resuspended in 1×
   radioimmunoprecipitation assay (RIPA) (Cell Signaling Technology) with
   protease inhibitors and incubated on ice for 5 min. The samples were
   sonicated for 5  min in the ice water bath before centrifugation at
   14,000g at 4°C for 15 min. The supernatant was then collected as lysed
   proteins. EVs and MC were resuspended in DPBS and lysed with 0.05% SDS.
   Protein concentration was assessed by bicinchoninic acid assay (BCA)
   assay (Thermo Fisher Scientific, 23250). Proteins (50 μg) of cytoplasm,
   nucleus, and EVs samples, as well as 30 μg of MC samples, were dried in
   a speed vacuum and reconstituted in triethylammonium bicarbonate (TEAB)
   buffer. Samples were then treated with dithiothreitol (DTT) and
   iodoacetamide for reduction and alkylation. SP3 beads were used for
   protein purification, followed by overnight digestion with trypsin at
   37°C directly on the beads. Post-digestion, peptides were labeled with
   TMT10plex isobaric label reagent for quantitative analysis.

Liquid chromatography separation and tandem mass spectrometry

   For peptide fractionation, the tandem mass tag (TMT)-labeled peptides
   were reconstituted, cleaned to remove any excess labels, lipids, and
   small molecules, and subjected to basic reversed-phase chromatography
   using an Agilent HPLC system. This process involved an 85-min gradient
   to separate peptides, transitioning from 100% solvent A (10 mM TEAB in
   water) to 100% solvent B (90% acetonitrile/10 mM TEAB). Fractionated
   peptides were then analyzed on a nano-LC-Orbitrap system. The system
   used a reversed-phase chromatography gradient from 2% to 90%
   acetonitrile in 0.1% formic acid over 82 min. Eluted peptides were
   sprayed into an Orbitrap-Fusion-LumosIC mass spectrometer through a
   1-μm emitter tip (New Objective) at 2.4 kV. Initial survey scans (full
   MS) were acquired in the Orbitrap analyzer within a mass-to-charge
   (m/z) range of 375 to 1600. Top 15 most intense ions from the survey
   scan were selected for further analysis, using a dynamic exclusion
   duration of 15 s. Precursor ions were individually isolated with a
   window of 0.7 Da, fragmented (MS/MS) using a higher-energy collisional
   dissociation (HCD) with a set collision energy of 40. Precursor (MS1)
   and fragment (MS2) ions were subsequently analyzed at a resolution of
   120,000 and 50,000, respectively.

MS data processing and analysis

   Tandem MS2 spectra (signal/noise > 2) were processed by Proteome
   Discoverer (v2.4, Thermo Fisher Scientific) using the Files RC option.
   MS/MS spectra were searched with Mascot v. 2.8.0 (Matrix Science,
   London, UK) against RefSeq2021_204_Human database and a small database
   containing enzymes, BSA. Trypsin specificity was used for the
   digestion, with up to two missed cleavages. Methionine oxidation and
   deamidation were selected as the variable modifications.
   Carbamidomethylation of cysteines and TMT-label modifications of N
   termini and lysine were set as fixed modifications. Peptide
   identifications from the Mascot searches were processed within the
   Proteome Discoverer and Percolator to identify peptides with a
   confidence threshold of 1% false discovery rate (FDR). Only top ranked
   (rank 1) peptides were considered. Quantitative analyses were
   normalized across all identified peptides, while quantification ratios
   were computed exclusively from unique peptides. Quality control
   thresholds were applied, requiring a reporter signal-to-noise ratio
   (SN) > 5 and isolation interference < 25%. Protein identifications were
   classified into high, medium, or low confidence levels based on the
   Percolator, with inclusion criteria requiring at least one
   high-confidence peptide (1% FDR, rank 1) per protein.

   The threshold for significant differential abundance of proteins was
   defined at adjusted P values <0.05 (Benjamini-Hochberg method), as
   calculated using a non-nested analysis of variance (ANOVA) test per
   Proteome Discoverer 2.4 recommendations. Nonscaled proteomics data were
   visualized with a PCA plot. The protein interaction, protein cluster
   function prediction, cellular component, and biological process
   annotations were done by Protein-Protein Interaction Networks
   Functional Enrichment Analysis (STRING) in Cytoscape ([163]60). Venn
   diagrams were generated in R by VennDiagram and UpSetR packages.

Weighted gene coexpression network analysis (WGCNA)

   WGCNA R package ([164]61) was carried out to establish protein
   correlation modules for DE proteins identified in both 17EFr- and
   B670-infected samples compared with uninfected samples, across
   cytoplasm, nucleus, and EVs ([165]Fig. 4A). Briefly, gene coexpression
   networks were constructed based on a correlation matrix of protein
   levels in all samples. The matrix was then subjected to specific module
   cutting parameters (power = the optimal power defined by
   pickSoftThreshold, minimum module size = 30, deepSplit = 3, threshold
   of correlation = 0.01) to define protein modules based on hierarchical
   clustering. Protein modules with distinct coexpression patterns were
   assigned unique color labels by the software. These arbitrary colors
   aid in visually distinguishing each module in network plots and
   facilitate subsequent functional enrichment annotations. The eigengene
   network was then constructed to calculate the correlation of protein
   modules to sample traits. PPI networks and function enrichment analysis
   of significantly correlated modules were predicted using STRING in
   Cytoscape ([166]60).

Western blotting

   Whole-cell lysates, cytosolic fractions, nuclear fractions, and EVs
   were lysed in 1× RIPA buffer (Cell Signaling Technology, 9806). Protein
   concentrations were determined by BCA protein assay (Thermo Fisher
   Scientific, 23250). Equivalent amounts of total protein from cellular
   samples and equivalent volumes from EV samples were separated on 4 to
   15% stain-free SDS–polyacrylamide gel electrophoresis (PAGE) gradient
   gels (Bio-Rad, 5678083) under nonreducing conditions and transferred
   onto polyvinylidene difluoride (PVDF) membranes (Sigma-Aldrich,
   IPVH00010). After 1-hour blocking in 5% nonfat milk solution (Bio-Rad,
   170–6404), membranes were incubated with anti-SMN (1:2000 dilution),
   anti-CD63 (1:500 dilution) (BD Biosciences, 610646 and 556019,
   respectively), anti–lamin B1 (1:1000), anti-GAPDH (1:1000) (Abcam,
   ab16048, ab8245), anti-tubulin (1:1000, Novus Biologicals, NB100-690),
   anti-LSM1 (1:1000, LSBio, LS-C97364-400), anti-LSM8 (1:200, Santa Cruz
   Biotechnology, sc-81315), anti-SART3 (1:2000), anti-DIS3 (1:500), and
   anti-USP39 (1:500) (Proteintech, 18025-1-AP, 14689-1-AP, 23865-1-AP)
   overnight at 4°C. The membrane was then incubated with horseradish
   peroxidase (HRP)–conjugated secondary antibody (1:10,000 dilution)
   (Santa Cruz Biotechnology, sc-2357, sc-516102) at room temperature for
   1 hour after washing in PBST. After washing in PBST again, SuperSignal
   West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific,
   34580) was applied, and blots were visualized using a Thermo Fisher
   Scientific iBright 1500 imaging system.

Statistical analysis, data availability, and EV-TRACK

   The statistical significance of differences was assessed variously by
   ANOVA, Tukey’s multiple comparisons test, and unpaired t test, as
   indicated in figure legends for individual tests. Proteomics data files
   are available on request. We have submitted all relevant data of our
   experiments to the EV-TRACK knowledgebase (EV-TRACK ID: [167]EV240032)
   ([168]62).

Acknowledgments