The early macrophage response to pathogens requires dynamic regulation of the nuclear paraspeckle

Significance

To mount appropriate immune responses and fight infection, macrophages need to sense and respond to pathogen-associated signals with incredible precision. Membraneless organelles (MLOs) are complexes of RNAs and proteins that change in size, shape, and abundance in response to extracellular signals. We hypothesized that an MLO called the nuclear paraspeckle helps macrophages initiate and calibrate innate immune gene expression during infection. We found that paraspeckles rapidly aggregate and then dissolve in macrophages following pathogen sensing. Macrophages lacking paraspeckles cannot properly induce inflammatory genes, resulting in a failure to control replication of intracellular bacterial and viral pathogens. These data suggest that altered paraspeckle dynamics may dysregulate inflammatory gene expression in a variety of human diseases.

Abstract

To ensure a robust immune response to pathogens without risking immunopathology, the kinetics and amplitude of inflammatory gene expression in macrophages need to be exquisitely well controlled. There is a growing appreciation for stress-responsive membraneless organelles (MLOs) regulating various steps of eukaryotic gene expression in response to extrinsic cues. Here, we implicate the nuclear paraspeckle, a highly ordered biomolecular condensate that nucleates on the Neat1 lncRNA, in tuning innate immune gene expression in murine macrophages. In response to a variety of innate agonists, macrophage paraspeckles rapidly aggregate (0.5 h poststimulation) and disaggregate (2 h poststimulation). Paraspeckle maintenance and aggregation require active transcription and MAPK signaling, whereas paraspeckle disaggregation requires degradation of Neat1 via the nuclear RNA exosome. In response to lipopolysaccharide treatment, Neat1 KO macrophages fail to properly express a large cohort of proinflammatory cytokines, chemokines, and antimicrobial mediators. Consequently, Neat1 KO macrophages cannot control replication of Salmonella enterica serovar Typhimurium or vesicular stomatitis virus. These findings highlight a prominent role for MLOs in orchestrating the macrophage response to pathogens and support a model whereby dynamic assembly and disassembly of paraspeckles reorganizes the nuclear landscape to enable inflammatory gene expression following innate stimuli.

Biologists have long been interested in the functions of the membrane-bound organelles that define eukaryotic cells. More recently, membraneless organelles (MLO) have captured the attention of many. MLOs are biomolecular condensates that form through the process of liquid–liquid phase separation (LLPS). Through sequestration of proteins and RNAs, MLOs regulate and compartmentalize a variety of cellular processes in both the cytosol and the nucleus (1–3). A common theme of MLOs is their ability to change number, size, structure, and composition in response to cellular stress (4, 5). While phenomena related to condensate assembly have been extensively described, links between MLO dynamics and their function remain poorly understood.

One nuclear condensate with well-established links to stress responses is the nuclear paraspeckle (6–8). Discovered in HeLa cells in 2002, nuclear paraspeckles were first defined as nuclear domains enriched for paraspeckle protein 1 (PSP1) found in proximity to SC35-containing nuclear speckles (7). These highly ordered MLOs organize on a lncRNA called nuclear paraspeckle assembly transcript 1 (Neat1) (9). The Neat1 gene encodes two isoforms, Neat1_1 and Neat1_2. While both are found in paraspeckles, only Neat1_2, the long isoform (22.7 kb in humans, 21.2 kb in mice), is required for paraspeckle assembly. Although the two isoforms share the same promoter, their processing is distinct; instead of being polyadenylated, the 3′ end of Neat1_2 is stabilized by an atypical triple helix structure (8, 10). The short Neat1_1 isoform, on the other hand, is spliced and polyadenylated (1, 9). Paraspeckles are composed of ~50 copies of Neat1_2 and a cohort of RNA-binding proteins (RBPs).

Because of links between Neat1 and cancer, neurodegenerative disease, and inflammatory disorders, there is growing interest in how Neat1 and paraspeckles control cellular homeostasis and stress responses (2, 11). Specialized cell types, like neurons and immune cells, are constantly receiving and responding to environmental inputs that trigger remarkable changes to their transcriptomes and proteomes. Innate immune cells like macrophages are particularly exemplary of this behavior. Several studies have begun to link paraspeckles to innate immune gene expression and antiviral responses. We know that Neat1-deficient mice mount reduced inflammatory responses during models of peritonitis and pneumonia (12) and that Neat1 itself can be up-regulated in response to DNA or RNA viral infection (13–15). Upregulation of Neat1 has been shown to promote expression of inflammatory genes like IL-8 via sequestration of repressive SFPQ from the IL-8 promoter (16), and Neat1 knockdown has been linked to reduced dengue virus replication (17). Despite these intriguing links between paraspeckles and inflammation, it remains to be seen how paraspeckles function in bona fide immune cells like macrophages.

We hypothesized that nuclear MLOs like paraspeckles help the macrophage nucleus respond to pathogen sensing. Here, we report that paraspeckles undergo rapid changes in macrophages following several innate stimuli and demonstrate a critical role for Neat1 and the paraspeckle in mounting a balanced innate immune gene expression program to control bacterial and viral replication.

Results

Paraspeckles Are Rapidly Up- and Down-Regulated in Response to Innate Agonist Treatment of Macrophages.

To begin to define the dynamics of paraspeckle formation during macrophage activation, we employed a technique to simultaneously detect the Neat1 lncRNA by RNA-FISH and paraspeckle proteins like PSP1 by immunofluorescence staining (FISH-IF). Our FISH probes only anneal to sequences in the paraspeckle-forming Neat1_2 lncRNA, which we will refer to as Neat1 from now on. Using FISH-IF for Neat1 and PSP1, we observed that resting RAW 264.7 macrophages maintain two clear paraspeckles, consistent with previous reports demonstrating cotranscriptional paraspeckle formation at each Neat1 genomic locus (18, 19). We then treated macrophages with lipopolysaccharide (LPS) (100 ng/mL) and performed FISH-IF - (SI Appendix, Fig. S1A). We observed dramatic changes to the paraspeckle over a 4 h time course of LPS treatment (Fig. 1A). At 0.5 h post-LPS treatment, paraspeckles were up-regulated in RAW 264.7 macrophages. At 1 h, paraspeckle area remained high, but qualitatively, paraspeckles became more dispersed throughout the nucleus. Paraspeckle aggregation was concomitant with an approximately twofold to threefold increase in total Neat1_2 transcript as measured by qRT-PCR (Fig. 1B). Interestingly, we observed no change to Neat1_1 levels over the 4 h LPS time course, suggesting that the two isoforms of Neat1 are differentially responsive to signaling downstream of TLR4 (Fig. 1B). At 2 h post-LPS, paraspeckle signal was virtually undetectable, with no observable Neat1 or PSP1 puncta. This phenotype was pervasive, as virtually no cells had detectable paraspeckles at 2 h post-LPS (SI Appendix, Fig. S1C). Paraspeckle disaggregation was concomitant with a loss of Neat1_2 RNA signal by qRT-PCR (Fig. 1B). Compared to our normal Zymo DirectZol column extractions, Neat1_2 extraction efficiency improved with mechanical shearing or incubation at 55° (SI Appendix, Fig. S1D), but the pattern of Neat1_2 up- and downregulation at each time point post-LPS treatment remained similar (Fig. 1B and SI Appendix, Fig. S1D). By 4 h post-LPS, paraspeckles started to reform although total Neat1 RNA remained low. At 6 h post-LPS, cells were heterogeneous in their paraspeckle numbers, with most cells having 0, 1, or 2 paraspeckles and about 20% of cells maintaining higher numbers (SI Appendix, Fig. S1E). By 8 h post-LPS, the percentage of cells maintaining >2 paraspeckles increased to ~50%, suggestive of cells restarting the cycle of paraspeckle aggregation (SI Appendix, Fig. S1F).

Fig. 1.

Nuclear paraspeckles are dynamically regulated following innate immune agonist treatment of macrophages. (A) RNA-FISH of Neat1 (red) and immunofluorescence staining of PSP1 (green) in RAW 264.7 macrophages post-LPS treatment (100 ng/mL unless otherwise noted). Quantitation of paraspeckle area/nucleus on Right. (B) qRT-PCR of Neat1_2 (Top) and Neat1_1 (Bottom) transcript levels in RAW 264.7 macrophages post-LPS treatment shown relative to Actb. (C) As in (A) but with BMDMs treated with 10 ng/mL LPS. Quantitation of paraspeckle area/nucleus on Right. (D) RNA-FISH of Neat1 (red) in RAW 264.7 macrophages post-Pam3CSK4 treatment (100 ng/mL). Quantitation of paraspeckle area/nucleus below. (E) As in (D) but post-dsDNA transfection (ISD) (1 µg/mL). (F) As in (D) but post-dsRNA transfection (poly I:C) (500 ng/mL). (G) RNA-FISH of Neat1 (red) in RAW 264.7 macrophages post-LPS treatment, following overnight polarization into M1- (+IFN-γ; 50 ng/mL) or M2 (+IL-4; 25 ng/mL)-like macrophages. Quantitation of paraspeckle area/nucleus below. Statistical tests: Data are presented as the mean n = 3 unless otherwise noted. Statistical significance was determined using a one-way ANOVA. ***P < 0.001, ****P < 0.0001.

We observed similar paraspeckle dynamics in primary BMDMs (Fig. 1C and SI Appendix, Fig. S1G), with peak paraspeckle area at 0.5 h post-LPS and no paraspeckles detectable at 2 h post-LPS. Compared to reports in other cell types, paraspeckle hyperaggregation following pathogen sensing in macrophages is fast [0.5 h vs. 3 h or more (16, 20)]. The phenomena of paraspeckle disappearance at 2 h and subsequent recovery at 4 h post-LPS treatment are not reported to occur in other mammalian cell types. To begin to determine whether these dynamics are unique to Neat1/paraspeckles, we used RNA FISH to detect another abundant nuclear lncRNA, Malat1, which is encoded in the genome directly downstream of Neat1. Apart from a modest increase in signal intensity at 0.5, 1, and 2 h post-LPS, we observed no major changes to the size or number or Malat1 aggregates [nuclear speckles (21)] in response to LPS (SI Appendix, Fig. S1 H and I). Total Malat1 transcript abundance was also unchanged over an 8 h LPS treatment (SI Appendix, Fig. S1J).

We next asked whether paraspeckle dynamics triggered by LPS, which activates pathogen sensing cascades via TLR4, were unique. Having seen very similar paraspeckle dynamics between RAW 264.7 macrophages cells and primary BMDMs, we chose to continue studies with the genetically tractable RAW 264.7 cell line. Likewise, having seen identical patterns for the Neat1 lncRNA by FISH and the PSP1 protein by immunofluorescence microscopy across multiple experiments, we opted to track paraspeckles by Neat1 FISH alone. Thus, we treated RAW 264.7 macrophages with a panel of innate immune agonists and measured paraspeckle formation by Neat1 RNA-FISH. Treatment of cells with the TLR2 agonist Pam3CSK4 (Fig. 1D and SI Appendix, Fig. S1 K and L) or the cGAS agonist cytosolic dsDNA (ISD) (Fig. 1E and SI Appendix, Fig. S1 M and N) triggered paraspeckle dynamics that closely followed those induced by LPS. Surprisingly, transfection of poly I:C, a dsRNA agonist of TLR3 (endosomal) and RIG-I/MDA5 (cytosolic) RNA sensing cascades, as well as protein kinase R (PKR), ablated Neat1 signal in the nucleus by 0.5 h (Fig. 1F and SI Appendix, Fig. S1 O and P). Our poly I:C results contrast a previous report that showed paraspeckle upregulation in HeLa cells post-poly I:C transfection (22).

Macrophage polarization is an important determinant in dictating innate immune outcomes (23). Macrophages can take on a classical proinflammatory M1 state when treated with IFN-γ or can be alternatively activated to a wound-healing M2 state after treatment with IL-4. To determine whether macrophage polarization impacts paraspeckle dynamics, we treated RAW 264.7 macrophages overnight with IFN-γ (M1) or IL-4 (M2) and confirmed polarization by measuring canonical M1/M2 transcripts by qRT-PCR (SI Appendix, Fig. S1Q). We did not see dramatic upregulation of PS at 0.5 h post-IFN-γ or IL-4 treatment alone, suggesting that treatment with these cytokines is not sufficient to up-regulate paraspeckles (SI Appendix, Fig. S1R). We then repeated our LPS time course in these M1 or M2 macrophages. In both cases, we qualitatively observed hyperaccumulation of Neat1 by RNA-FISH at 0.5 and 1 h, suggesting that polarized macrophages are primed to up-regulate paraspeckles (Fig. 1G). Downregulation of paraspeckles occurred with similar kinetics in M1 and M2-polarized macrophages. Together, these data identify nuclear paraspeckles as immune-responsive MLOs in macrophages that are dynamically regulated downstream of multiple innate sensing cascades.

Macrophage Paraspeckles Are Compositionally Distinct from Those in HeLa Cells.

Paraspeckles contain many copies of the Neat1 RNA and an array of RNA-binding proteins (24). To better understand the nature of the macrophage paraspeckle, we next asked whether canonical paraspeckle proteins colocalize with Neat1 in the macrophage nucleus and whether these associations are altered by LPS treatment. Consistent with reports in other cell types, we observed significant colocalization between Neat1 and PSP1 (Fig. 2A), SFPQ (Fig. 2B), FUS (Fig. 2C), and RBM14 (Fig. 2D) in both resting and LPS-stimulated macrophages. Association between NONO and Neat1 appeared to follow distinct dynamics, with approximately 30% more colocalization measured in LPS-treated compared to resting macrophages (Fig. 2E). As total cellular levels of paraspeckle proteins remained constant over a 1 h time course of LPS treatment, these data support a model whereby already synthesized paraspeckle proteins in the nucleus are brought into newly aggregating paraspeckles (SI Appendix, Fig. S2A). Components of the SWI/SNF nucleosome remodeling complex, which play an important role in activating expression of secondary response genes like Il6 (25) have been found in paraspeckles in other model cell types. Surprisingly, BRG1 and BRM, deemed essential for paraspeckle assembly in HeLa cells (26), did not display obvious punctate staining reminiscent of paraspeckles pre- or post-LPS treatment (SI Appendix, Fig. S2 B and C), despite clear aggregation of BRM in resting macrophages. Lack of evidence for BRM and BRG1 enrichment in macrophage paraspeckles begins to suggest that the composition of these condensates may be cell-type specific.

Fig. 2.

Paraspeckle upregulation in macrophages sequesters nuclear RNA-binding proteins. (A) RNA-FISH of Neat1 (red) and immunofluorescence staining of PSP1 (green) in RAW 264.7 macrophages at 0.5 h and 1 h post-LPS treatment (100 ng/mL). Correlation coefficient between Neat1 and PSP1 quantified below. (B) As in (A) but for SFPQ. (C) As in (A) but for FUS. (D) As in (A) but for RBM14. (E) As in (A) but for NONO. (F) As in (A) but for hnRNP M. Statistical tests: Data are presented as the mean n = 3 unless otherwise noted. Statistical significance was determined using a one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001.

As over 20 RBPs involved in a variety of nuclear processes (pre-mRNA splicing, RNA editing, mRNA export, etc.) have been shown to localize to and/or purify with the paraspeckle in nonimmune cells (20), we posited that additional RBPs and immune-associated proteins could be brought into paraspeckles during macrophage activation. One RBP that reportedly interacts with paraspeckle proteins (27) and has links to innate immune gene expression is the splicing factor hnRNP M (28). By FISH-IF, we saw a marked increase in colocalization between hnRNP M and Neat1 at 0.5 h post-LPS treatment (Fig. 2F). We also tested whether paraspeckles sequester TLR4-activated innate immune transcription factors. Overall, colocalization between Neat1 and the two factors queried, NFκB and STAT1, was low (SI Appendix, Fig. S2 D and E), although we did measure a slight, but statistically significant, increase in colocalization between Neat1 and NFκB in LPS-treated macrophages. This could represent enrichment of NFκB at the site of Neat1 transcription and would be consistent with ChIP-seq experiments that show enrichment of the NFκB subunit RelA at the Neat1 promoter in response to LPS (SI Appendix, Fig. S2 F and G). Together, these data hint at compositional differences between macrophage paraspeckles and those previously described in other cell types. They also suggest that RBPs previously linked to posttranscriptional regulation of innate immune gene expression (e.g., hnRNP M) have links to the paraspeckle in macrophages.

Transcription and MAPK Signaling Are Required to Maintain and Up-Regulate Paraspeckles in Macrophages.

Given the rapid up- and downregulation of paraspeckles during the early macrophage response to LPS, we set out to investigate the cellular pathways that control paraspeckle maintenance and aggregation in these cells. First, we asked whether transcription was required for paraspeckle upregulation after LPS treatment. Briefly, RAW 264.7 macrophages were treated with the transcription inhibitor actinomycin D (ActD) for 0.5 h and Neat1 was monitored by FISH and qRT-PCR ± a 0.5 h LPS treatment. ActD not only prevented LPS-induced paraspeckle upregulation, but also inhibited paraspeckle maintenance all together, as we could no longer detect Neat1 puncta in resting macrophages after 0.5 h of ActD (Fig. 3A and SI Appendix, Fig. S3A). Importantly, ActD-mediated loss of paraspeckle signal did not directly correspond to total cellular levels of Neat1 lncRNA, which remained high for at least 1 h after ActD (Fig. 3B) and ActD/LPS treatment (Fig. 3C). Because ActD has poor selectivity for eukaryotic polymerases (29), we treated macrophages with α-amanitin, a selective inhibitor of RNA polymerase II at low concentrations, and again measured paraspeckles by RNA FISH and Neat1 levels by qRT-PCR. We observed a similar response to α-amanitin, albeit with slightly slower kinetics [paraspeckle ablation at 1 h vs. 0.5 h with ActD (SI Appendix, Fig. S3D)], consistent with slow uptake of the drug (30). Together, these findings suggest that active transcription is required to maintain paraspeckles in macrophages.

Fig. 3.

Active transcription and basal MAPK signaling maintain paraspeckles in macrophages. (A) RNA-FISH of Neat1 (red) in RAW 264.7 macrophages after actinomycin D treatment (ActD) (5 µg/mL) followed by LPS stimulation (100 ng/mL unless otherwise noted) for time points indicated. Quantitation of paraspeckle area/nucleus on Right. (B) qRT-PCR of Neat1_2 transcript levels in RAW 264.7 macrophages after ActD treatment, (5 µg/mL) shown relative to Actb. (C) As in (B) but with LPS treatment following 0.5-h ActD. (D) As in (B) but for Gapdh. (E) As in (B) but for Kcnq1OT1. (F) RNA-FISH of Neat1 (red) in RAW 264.7 macrophages post-MAPK inhibitor treatment [MEK inhibitor (U0126; 25 µM), JNK inhibitor (SP600125; 25 µM), p38 inhibitor (SB203580; 10 µM)] for 0.75 h. Quantitation of paraspeckle area/nucleus on right. (G) As in (F) but with the addition of LPS for 0.5 h following ActD. (H) RNA-FISH of Neat1 (red) in RAW 264.7 macrophages after treatment with the TBK1 inhibitor (GSK-8612; 10 µM) at 0 and 0.5 h post-LPS stimulation. Quantitation of paraspeckle area/nucleus on right. (I) As in (H) but with CLK1 inhibitor (Cpd 23; 10 µM). Quantitation of paraspeckle area/nucleus on right. Statistical tests: Data are presented as the mean n = 3 unless otherwise noted. Statistical significance was determined using one-way ANOVA. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

After transcription shut off, we began to see loss of total Neat1 transcript signal after 2 h in RAW 264.7 macrophages, both in response to ActD (Fig. 3 B and C and SI Appendix, Fig. S3E) and α-amanitin (SI Appendix, Fig. S3 H and I). Based on our ActD experiment, we calculate the half-life of Neat1_2 to be 109 min (±5 min) in RAW 264.7 macrophages. Degradation of the Neat1 RNA after ActD treatment far exceeded that of a stable housekeeping gene like Gapdh (Fig. 3D and SI Appendix, Fig. S3F) and was even faster than that of another previously identified short lived noncoding RNA, Kcnq10t1 (31) (Fig. 3E and SI Appendix, Fig. S3G), consistent with earlier reports of Neat1 being an unstable/short-lived RNA (31). Neat1 was comparably more stable in α-amanitin-treated cells, but we believe this is due mainly to poor uptake of the peptide into macrophages, as even a 4-h treatment was not sufficient to completely block induction of RNAPII transcription in response to LPS (SI Appendix, Fig. S3J). Together, these data argue that de novo transcription is required both to maintain paraspeckles in macrophages, as was previously reported in myoblasts (19), and to aggregate paraspeckles upon LPS treatment.

We next sought to identify the signaling cascades that promote Neat1 transcription and/or upregulation upon LPS treatment. LPS stimulation of TLR4, and TLR signaling in general, activate a complex network of kinase cascades, including MEK, JNK, and p38 MAP kinases, AKT and PI3-K, IκB kinase, and the noncanonical IκB kinase homologs IKK-ε and TBK1 (32, 33). To begin to test the role of MAPKs in regulating PS dynamics in macrophages, we LPS-treated RAW 264.7 macrophages for 0.5 h following pretreatment for 0.75 h with a MEK inhibitor (U0126), a JNK inhibitor (SP600125), or a p38 inhibitor (SB203580) (SI Appendix, Fig. S3K). Remarkably, not only did we see paraspeckles fail to accumulate in response to LPS after all three MAPK inhibitor treatments, but we also saw loss of paraspeckles in non-LPS-treated macrophages, suggesting that basal MAPK signaling is important for maintaining paraspeckles in resting cells (Fig. 3 F and G and SI Appendix, Fig. S3L). Inhibiting other potentially relevant cellular kinases like TBK1 (Fig. 3H and SI Appendix, Fig. S3 M and N), which is activated downstream of TLR4 to phosphorylate the transcription factor IRF3, or CLK1, which is activated downstream of AKT signaling to regulate phosphorylation of SR proteins (34, 35), did not ablate paraspeckles (Fig. 3I and SI Appendix, Fig. S3 O and P). In fact, treatment with the CLK1 inhibitor resulted in modest hyperaggregation of Neat1 at 0.5 h post-LPS treatment. These results implicate MAPKs as positive regulators of paraspeckle maintenance and hint at SR protein phosphorylation negatively regulating paraspeckle aggregation in macrophages.

The Neat1 lncRNA Is Targeted to the Nuclear Exosome by the NEXT Complex to Regulate PS Dynamics in Macrophages.

We next wanted to investigate the mechanisms driving paraspeckle ablation and loss of Neat1 transcript that we report at 2 h post-LPS treatment. In the nucleus, RNA turnover is controlled by the RNA exosome, a multiprotein complex responsible for 3′ end processing and/or degradation of RNAs (36). The exosome forms a barrel structure and has two associated 3′ to 5′ exoribonucleases: EXOSC10/RRP6 and DIS3 (Fig. 4A). RNAs are targeted to the exosome for processing or degradation by one of three accessory protein complexes: NEXT, TRAMP, or PAXT/PCC. NEXT is involved in turnover of introns released by pre-mRNA splicing and unstable RNAs from pervasive transcription. TRAMP degrades RNAs like pre-rRNAs, cryptic unstable transcripts, as well as a variety of aberrant small RNAs (tRNAs, ncRNAs, snoRNAs, snRNAs). PAXT/PCC is responsible for bringing nuclear ncRNAs with long polyA tails to the exosome. To begin to implicate the exosome in regulation of Neat1 and paraspeckles in macrophages, we transfected siRNAs designed against Mtr4, alongside a nontargeted control (NC), into RAW 264.7 macrophages. MTR4 is a member of the SKI2 family of RNA helicases that is common to all three nuclear exosome targeting complexes. At 48 h posttransfection, we achieved >90% knockdown of Mtr4 transcript (SI Appendix, Fig. S4A) and protein (SI Appendix, Fig. S4B). By RNA-FISH, we observed a dramatic increase in Neat1 signal in resting Mtr4 knockdown macrophages compared with NC siRNA control cells (Fig. 4B). This correlated with total Neat1_2 transcript accumulation (Fig. 4C) and is consistent with previous reports of Neat1 instability and a role for the exosome in controlling Neat1 turnover (37). Mtr4 knockdown did not result in accumulation of other abundant mRNAs or nuclear snRNAs (SI Appendix, Fig. S4C).

Fig. 4.

The NEXT complex targets the Neat1 lncRNA to the nuclear exosome to regulate paraspeckle dynamics in macrophages. (A) Model of nuclear RNA exosome and the NEXT and TRAMP targeting complexes. Colored shapes denote factors knocked down in Figure 4 experiments. (B) RNA-FISH of Neat1 (red) in Mtr4 knockdown RAW 264.7 macrophages 48 h after Silencer Select siRNA transfection, alongside a negative control (siNC), over a time course of LPS treatment (100 ng/mL unless otherwise noted). Quantitation of paraspeckle area/nucleus on Right. (C) qRT-PCR of Neat1_2 transcript levels in siMtr4 and siNC RAW 264.7 macrophages post-LPS treatment shown relative to Actb. (D) RNA-FISH of Neat1 (red) and immunofluorescence microscopy of PSP1 (green) in siNC RAW264.7 macrophages. (E) As in (D), but in siMtr4-treated RAW 264.7 macrophages. (F) As in (D), but in siZcchc8-treated macrophages. (G) As in (D), but in siZcchc7-treated macrophages. (H) As in (D), but in siExosc10-treated macrophages. (I) As in (D), but in siDis3-treated macrophages. (J) Quantitation of paraspeckle area/nucleus of (D–I). (K) Quantitation of Neat1 aggregates per nuclei in (D–I). (L) RNA-FISH of Neat1 (red) in siNC and siMtr4-treated RAW 264.7 macrophages in untreated cells, +LPS (0.5 h) or +ActD (5 µg/mL; 0.5 h) followed by LPS (0.5 h). Quantitation of paraspeckle area/nucleus on Right. Statistical tests: Data are presented as the mean n = 3 unless otherwise noted. Statistical significance was determined using a Student’s t test (A–C) or one-way ANOVA (J–L). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Neat1 aggregates that form in the absence of MTR4 are bona fide structured PSs, as they each show significant enrichment with the paraspeckle protein PSP1 (Fig. 4 D and E). Even though Mtr4 KD macrophages have high numbers of paraspeckles at rest, these numbers still increase after LPS treatment (siMtr4 time 0 vs. 0.5 h vs. 1 h post-LPS; Fig. 4C). We interpret this to mean that aggregation of paraspeckles following LPS stimulation occurs independently of the exosome. We can implicate the RNA exosome in paraspeckle disassembly at 2 h post-LPS, as we do not see the characteristic loss of Neat1 signal at 2 h post-LPS in Mtr4 KD macrophages (Fig. 4C). Interestingly, although bulk measurements of Neat1_2 cellular transcripts reflect this phenotype at 2 h (i.e., Neat1 is higher in siMtr4 relative to siNC cells), by 4 h, Neat1_2 transcript levels are very low regardless of whether cells have MTR4. This discrepancy at 4 h may stem from incomplete knockdown of Mtr4, where our bulk measurements reflect a mixture of cells, some of which have Mtr4 knocked down and others that have normal levels of Mtr4 (Fig. 3K). It is also possible that some kind of exosome-independent transcriptional shut-off further limits Neat1 abundance post-LPS treatment. Regardless, we can conclude from these data that TLR4 engagement signals exosome-mediated dissolution of paraspeckles and turnover of Neat1 at 2 h post-LPS treatment in macrophages.

Having implicated the MTR4 RNA exosome helicase in Neat1 turnover and paraspeckle aggregation (Fig. 4 B and E), we next sought to pin down the targeting complex and exonuclease that regulate Neat1 stability in macrophages. Since Neat1_2 is not polyadenylated, we ruled out a role for the PAXT/PCC targeting complex. To implicate either NEXT or TRAMP in Neat1 turnover, we transfected RAW 264.7 macrophages with siRNAs directed against Zcchc8 (NEXT) (SI Appendix, Fig. S4 D and E) and Zcchc7 (TRAMP) (SI Appendix, Fig. S4 F and G) and observed paraspeckles by IF-FISH in resting cells. Loss of Zcchc8 up-regulated paraspeckles in a similar fashion to loss of Mtr4 (Fig. 4 E, F, J, and K), while loss of Zcchc7 had no impact on paraspeckle number or size (Fig. 4 G, J, and K). We performed a similar experiment to determine the exoribonuclease that degrades Neat1, comparing paraspeckles in resting Exosc10 (SI Appendix, Fig. S4 H and I) versus Dis3 knockdown macrophages (SI Appendix, Fig. S4 J and K). We observed a clear upregulation of paraspeckles in cells transfected with siRNAs directed against Dis3 but not Exosc10 (Fig. 4 H–K). Together, these data suggest that Neat1 is targeted to the exosome in macrophages by the NEXT complex, although it is possible that the lack of Neat1 accumulation in Zcchc7 and/or Exosc10 siRNA-treated macrophages is attributable to incomplete knockdown (SI Appendix, Fig. S4 F and G and Fig. 4 J–K).

Next, we asked whether cotranscriptional PS assembly and exosome targeting of Neat1 in macrophages are separable mechanisms. To do so, we treated siNC- and siMtr4-transfected macrophages with LPS in the presence or absence of ActD, as in Fig. 3A. Whereas ActD completely ablated Neat1 signal after 0.5 h in siNC cells (as we observed for wild-type macrophages in Fig. 3A), ActD had no impact on paraspeckles in siMtr4 KD cells (Fig. 4L). Therefore, blocking the exosome allows for paraspeckle maintenance even in the absence of de novo transcription, suggesting a tug-of-war between transcription and exosome turnover in maintaining Neat1 and paraspeckles in macrophages. By demonstrating regulated turnover of Neat1 by the nuclear RNA exosome in response to LPS treatment, our data hint at undescribed links between pattern recognition receptor engagement and exosome activity.

Neat1 Is Required to Activate the Innate Immune Response in Macrophages.

Having observed dramatic, regulated reorganization of paraspeckles in macrophages following LPS treatment, we set out to test whether ablating Neat1 and disrupting this paraspeckle cycle impacts the macrophage innate immune response. We acquired mice that do not express Neat1 due to incorporation of a lacZ cassette at the 5′ end of the Neat1 gene (38) (SI Appendix, Fig. S5A). We differentiated BMDMs from WT and Neat1 KO mice and isolated RNA for next generation sequencing at 0, 2 and 4 h post-LPS stimulation (10 ng/mL) (39). Loss of Neat1 in these cells was confirmed by RNA-FISH and qRT-PCR (Fig. 5A and SI Appendix, Fig. S5 B and C). To enable identification of noncoding RNAs and incompletely processed RNAs whose abundance may be altered in the absence of Neat1, sequencing libraries were generated using ribodepletion. Differential expression analysis uncovered hundreds of misregulated genes in Neat1 KO macrophages (208 genes at rest, 348 genes at 2 h post-LPS, and 352 genes at 4 h post-LPS; log₂FC >± 0.5; adj. P-value < 0.05) (SI Appendix, Table S1 and Fig. S5D). Approximately one-third of Neat1-dependent genes reach this differential expression threshold in all three conditions (i.e., 0, 2, 4 h post-LPS) but most exhibited Neat1 dependence only after LPS treatment (SI Appendix, Fig. S5D). We observed both up and downregulation of genes in a Neat1-dependent fashion and these genes segregate into distinct cellular pathways. Specifically, Ingenuity Pathway Analysis identified upregulation of genes with functional connections to hepatic fibrosis, pulmonary fibrosis, and wound healing in Neat1 KO macrophages (Fig. 5 B and D). These are functions typically attributed to alternatively activated or M2-like macrophages, which specialize in cellular proliferation and tissue repair. Many genes involved in fibrosis and wound healing are normally down-regulated as part of the proinflammatory macrophage response to LPS, which polarizes macrophages toward an M1 state (SI Appendix, Fig. S3E). Our findings suggest that loss of Neat1 results in a failure to fully down-regulate M2 genes in the early process of M1 polarization.

Fig. 5.

Neat1 is required for proper up- and downregulation of innate genes during the early macrophage response to LPS. (A) RNA-FISH of Neat1 (red) and immunofluorescence microscopy of PSP1 (green) in WT and Neat1 KO BMDMs over a 1 h time course of LPS treatment (10 ng/mL). Quantitation of paraspeckle area/nucleus over 4 h time course below. (B) Ingenuity pathway analysis (IPA) for differentially expressed up-regulated genes in Neat1 KO vs. WT BMDMs (log₂FC > 0.5; P < 0.05) at 0, 2, and 4 h post-LPS treatment (10 ng/mL). (C) As in (B) but for down-regulated genes. (D) Log₂FC of differentially expressed genes in Neat1 KO BMDMs categorized as “wound healing signaling pathway” by IPA. (E) As in (D) but for “role of hypercytokinemia in the pathogenesis of influenza.” (F) As in (E) but for “pathogen-induced cytokine storm signaling pathway.” (G) Integrated Genomics Viewer tracks of RNA-seq reads from a representative WT and Neat1 KO sample at the Il6 genomic locus (mm10). (H) As in (G) but for Cxcl9. (I) Measurements of IL-6 levels in the supernatants of WT and Neat1 KO BMDM at 6 and 12 h post-LPS treatment (10 ng/mL) via cytokine array. (J) As in (I) but for IL-12p40. (K) As in (I) but for KC (CXCL1). (L) As in (I) but for MIP-2 (CXCL2). (M) As in (I) but for MCP-1 (CCL2). (N) As in (I) but for MIG (CXCL9). (O) As in (I) but for VEGF-A. (P) As in (I) but for G-CSF (CSF3). Statistical tests: Data are presented as the mean n = 3 unless otherwise noted. Statistical significance in (A and I–P) was determined using a Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

We also detected many macrophage genes that were down-regulated by loss of Neat1. Down-regulated genes are enriched in pathways related to interferon signaling, hypercytokinemia in influenza pathogenesis, pathogen induced cytokine storm signaling, and pattern recognition receptor activation (Fig. 5 C, E, and F). These pathways are related to proinflammatory or M1-like responses in macrophages and are composed of genes that are activated in response to LPS. We validated downregulation of several key inflammatory and antiviral transcripts (Rsad2, Il6, Cxcl9) in Neat1 KO BMDMs by qRT-PCR (SI Appendix, Fig. S5 F–H). As we can detect Neat1-dependent differences in reads aligning to both intronic and exonic sequences, our results suggest a defect in transcription and not pre-mRNA splicing (Fig. 5 G and H). Together, our data argue that Neat1 KO macrophages fail to properly polarize to a proinflammatory state when they encounter a M1-polarizing stimulus like LPS.

To determine whether defective innate immune transcript accumulation in Neat1 KO macrophages is borne out at the level of cytokine and chemokine secretion, we collected supernatants from WT and Neat1 KO BMDMs and measured cytokine and chemokine secretion by multiplex cytokine array. Although we saw little evidence for differences in secretion of cytokines in resting Neat1 KO and WT BMDMs (SI Appendix, Fig. S5 I–R), we observed defective expression/secretion of several cytokines and chemokines at 6 and 12 h post-LPS treatment in Neat1 KO BMDMs (Fig. 5 I–P and SI Appendix, Fig. S5 Q–R). Generally, altered cytokine/chemokine secretion was well-correlated with altered transcript abundance at 2 h and 4 h post-LPS, suggesting that early gene expression changes in Neat1 KO BMDMs are sufficient to impact downstream protein output (SI Appendix, Fig. S5S). Factors that were down-regulated by loss of Neat1, such as IL-6, KC/CXCL1, MIP-2/CXCL2, and MCP-1/CCL2, play critical roles in promoting inflammation and infiltration of neutrophils and lymphocytes to sites of infection (Fig. 5 I–N). Cytokines up-regulated by loss of Neat1, such as VEGFA and G-CSF (encoded by the Csf3 gene), are involved in cell proliferation and differentiation (Fig. 5 O–P). Taken together, these data implicate Neat1 in two critical aspects of the macrophage response to LPS: eliciting a balanced inflammatory response and differentiation into classical vs. alternative activation states.

Neat1 CRISPRi iBMDMs Fail to Execute Antibacterial and Antiviral Defenses.

As a strategy to overcome breeding limitations of the Neat1 KO mice (40) and to expand upon our findings in primary cells, we designed a CRISPR interference-based approach to ablate Neat1 from immortalized bone marrow derived macrophages. Using lentiviral transduction, we introduced a guide RNA directed against the Neat1 promoter, alongside an untargeted control, into immortalized BMDMs expressing an endonuclease-deficient form of Cas9 (deactivated (dCas9) (41). We confirmed loss of Neat1 expression in these CRISPRi cell lines by qRT-PCR (SI Appendix, Fig. S6A) and by RNA-FISH. Over a 4 h LPS treatment, paraspeckle dynamics in iBMDMs recapitulate those that we report in RAW 264.7 macrophages and BMDMs (Fig. 6 A and B).

Fig. 6.

Neat1 is required for antimicrobial responses in macrophages. (A) RNA-FISH of Neat1 (red) and immunofluorescence staining of PSP1 (green) at 0, 0.5, 1, 2, and 4 h post-LPS treatment (100 ng/mL unless otherwise noted) of dCas9-expressing iBMDMs transduced with an off-target (OT) gRNA lentiviral construct or a Neat1 promoter-targeted gRNA lentiviral construct. Zoom in shown for 0.5 and 1 h time points. (B) Quantitation of percent cells with paraspeckles in (A). (C) qRT-PCR of Oas2 transcript levels in off-target gRNA and Neat1 gRNA dCas9-expressing iBMDMs post-LPS treatment, shown relative to Actb. (D) As in (C), but for Socs1. (E) As in (C), but for Il6. (F) As in (C), but for Rsad2. (G) Viperin (encoded by Rsad2) protein expression in off-target gRNA and Neat1 gRNA dCas9-expressing iBMDMs post-LPS treatment. (H) Quantitation of viperin protein levels, relative to tubulin, as in (G). n = 4 (I) qRT-PCR of Vsvg and Vsvm transcript levels in off-target gRNA and Neat1 gRNA dCas9-expressing iBMDMs after VSV infection (MOI = 1) for 0, 2 and 4 h, shown relative to Actb. (J) VSV-GFP at 0, 0.5, 1, 2, and 4 h postinfection (MOI = 1) in off-target gRNA and Neat1 gRNA dCas9-expressing iBMDMs. (K) Colony-forming units (CFUs) of Salmonella recovered from off-target gRNA and Neat1 gRNA dCas9-expressing iBMDMs at 2 and 20 h postinfection (MOI = 10). (L) Immunofluorescence of Salmonella (anti-LPS antibody; red) at 1 and 6 h post-infection (MOI = 10) in off-target gRNA and Neat1 gRNA dCas9-expressing iBMDMs. Quantitation of Salmonella on right. Statistical tests: Data are presented as the mean n = 3 unless otherwise noted. Statistical significance was determined using a Student’s t test. *P < 0.05, **P < 0.01, ***P < 0.001.

Consistent with our RNA-seq data collected from BMDMs, many LPS-induced genes, including Oas2, Socs1, Il6, and Rsad2, failed to be effectively up-regulated in Neat1 gRNA iBMDMs (Fig. 6 C–F). Additional inflammatory genes whose differential expression was not deemed statistically significant in our RNA-seq data, e.g., Cxcl1 and Il1b, were in fact down-regulated in Neat1 gRNA iBMDMs (SI Appendix, Fig. S6 B and C). Failure to induce Rsad2 transcript expression resulted in significantly lower Viperin (RSAD2) protein levels in Neat1 gRNA iBMDMs compared to off-target gRNA controls (Fig. 6 G and H).

Based on these data and our Neat1 KO BMDM transcriptomics (Fig. 5), we hypothesized that Neat1 gRNA iBMDMs may be compromised in their ability to restrict viral replication. To test this, we infected Neat1 gRNA and off-target control iBMDMs with vesicular stomatitis virus (VSV) at an MOI of 1. VSV is a single-stranded negative-sense RNA virus whose replication is exquisitely sensitive to type I interferon and interferon stimulated gene (ISG) expression. We detected a dramatic increase in VSV replication in Neat1 gRNA iBMDMs at early time points postinfection, as measured by qRT-PCR of two VSV genes (Vsvg and Vsvm) (Fig. 6I). We also directly visualized enhanced viral protein expression (via VSV-GFP) in Neat1 gRNA iBMDM monolayers compared to off-target gRNA controls (Fig. 6J). Failure to control VSV replication in Neat1 gRNA iBMDMs was concomitant with low expression of antiviral genes like Rsad2, Mx1, and Oas2 (SI Appendix, Fig. S6D). Although these data support an important role for Neat1 in restricting viral replication in macrophages, it is possible that Neat1 contributes to VSV assembly, budding, attachment, entry, and pathogenesis in additional ways not assayed here.

As M2-like macrophages are known to provide a niche that supports replication of the intracellular bacterial pathogen Salmonella enterica serovar Typhimurium (42–44), we hypothesized that Neat1 gRNA iBMDMs may also be permissive to Salmonella. To test this, we infected Neat1 gRNA and off-target control iBMDMs with Salmonella (MOI = 10) and enumerated CFUs from cells at 2 and 20 h postinfection. We observed a twofold to threefold increase in Salmonella replication in iBMDMs lacking Neat1 (Fig. 6K). Immunofluorescence microscopy using an anti-LPS antibody confirmed significantly enhanced Salmonella replication in Neat1 gRNA iBMDMs, without a defect in early internalization (Fig. 6L and SI Appendix, Fig. S6E). While the precise mechanism through which Salmonella can survive and replicate in Neat1 gRNA iBMDMs remains to be elucidated, these phenotypes are consistent with dysregulated expression of M1/M2 genes like nitric oxide synthase (Nos2), a key antimicrobial mediator in macrophages (45, 46), and arginase, Arg1, which has been shown to negatively regulate nitric oxide (47, 48) (SI Appendix, Fig. S6F).

Discussion

Despite enthusiasm surrounding the phenomenon of liquid–liquid phase separation and the structure of MLOs, the function of condensates in cellular homeostasis and stress responses remains poorly understood. Our current knowledge of paraspeckle structure and assembly far exceeds that of paraspeckle function. Here, we investigated a role for paraspeckles in activating the macrophage innate immune response following an infection-relevant stimulus (e.g., LPS). Our experiments identified several unique aspects of the paraspeckle lifecycle in macrophages. First, our data clearly demonstrate that maintenance and upregulation of paraspeckles in activated macrophages requires transcription (Fig. 3). As loading of paraspeckles proteins on Neat1 has been shown to occur cotranscriptionally (19), we know that transcription of Neat1 itself is needed for PS maintenance. However, whether Neat1 itself is transcriptionally up-regulated as part of the response to LPS remains unclear. By qRT-PCR, we typically measure a twofold to threefold increase in Neat1 transcript levels at 1 h post-LPS treatment (Fig. 1B). Neat1 transcript abundance peaks later than PS aggregation, which occurs at 0.5 h, suggesting that early PS aggregation is driven by sequestration of already synthesized Neat1 as opposed to de novo transcription of new Neat1 transcripts. This makes sense, given the rapid aggregation of PS observed (~0.5 h) and the length of Neat1 itself (21.1 kb), which will take some time to fully transcribe. Previously published RNA sequencing data and ChIP-seq data for the NFκB transcription factor subunit RelA (49) show some evidence for RelA binding at the Neat1 promoter following LPS stimulation of BMDMs (SI Appendix, Fig. S2F), although this was not concomitant with significantly increased Neat1 sequencing reads (SI Appendix, Fig. S2G). These data and our findings together support a model whereby paraspeckle aggregation in macrophages does not require cells to make more Neat1. With that said, it remains a possibility that bulk measurements of Neat1 RNA from a population of cells obscures our ability to accurately measure Neat1 transcriptional activation. Promoter fusion constructs and/or single-cell transcriptomics may help answer the question of whether Neat1 is transcriptionally induced downstream of pattern recognition receptor engagement and help reconcile our findings with other studies whose models invoke Neat1 transcriptional upregulation following PAMP sensing in nonimmune cells (16).

We can also conclude that the RNA exosome, specifically the NEXT targeting complex, is involved in turning over Neat1 and regulating paraspeckle dynamics in macrophages (Fig. 4). This is consistent with another report linking MTR4 and NEXT to Neat1 turnover, although the kinetics of Neat1 turnover reported in HeLa cells (loss of detectable Neat1 transcript by 6 to 8 h posttranscription shut-off) are different from what we see in macrophages (2 h post-LPS) (50). Paraspeckle upregulation in resting Mtr4, Zcchc8, and Dis3 knockdown macrophages demonstrates a role for the exosome in constitutively controlling paraspeckle size/numbers. The fact that we see paraspeckles maintained in exosome-knockdown macrophages even at time points when paraspeckles disaggregate in wild-type cells (2 h post-LPS) can mean two things: 1) exosome targeting of Neat1 is enhanced at 2 h post-LPS treatment, suggesting that the exosome itself is regulated downstream of pattern recognition receptor engagement or 2) exosome turnover of Neat1 is constant and Neat1 transcription is turned off in a regulated fashion. Potential mechanisms for such a shut off could include competition between the long and short isoforms of Neat1, histone modification/chromatin remodeling at the Neat1 promoter, and altered association of the Neat1 genomic locus with enhancer elements. Again, promoter fusions and/or cell lines that enable rapid loss of exosome components [like the auxin degron system in ref. 51] could help tease apart the role of transcription versus turnover in regulating Neat1 levels post-LPS treatment. In future experiments, it will also be important to monitor loss of exosome function via another readout, as failure to sufficiently inhibit complex components could result in false negative phenotypes.

A particularly curious aspect of the macrophage paraspeckle life cycle is its disintegration at 2 h post-LPS treatment. At 2 h post-LPS treatment, the macrophage innate response is still ramping up, with maximum transcript accumulation for many inflammatory and antimicrobial genes seen at 4 to 6 h poststimulation. Breakdown of paraspeckles at 2 h post-LPS could aid in this amplification by, for example, releasing factors involved in nucleosome remodeling at secondary response genes [e.g., SWI/SNF components (25)] or posttranscriptional processing of innate transcripts (e.g., RBPs). Such a model would make paraspeckle disintegration a critical step in activation of inflammatory responses and may position the paraspeckle as a stopgap to prevent this “ramping up” (i.e., continue to sequester innate activating proteins) in the absence of a strong stimulus.

On the other hand, it is possible that early paraspeckle aggregation at 0.5 to 1 h post-LPS sequesters inhibitory factors to help activate the innate response. Such a model has been proposed for paraspeckle-mediated relief of SFPQ repression of IL8 transcription in HeLa cells (16). Of note, because LPS and polyI:C elicit different transcriptional responses, it is difficult to directly compare our findings to those of Imamura et al., who reported that Neat1 aggregation can promote innate immune gene expression by sequestering SFPQ from promoter of genes like Il8 (16). Il8 is not expressed by macrophages in response to LPS at the time points we queried (SI Appendix, Table S1). However, consistent with the idea of paraspeckles sequestering repressive factors upon immune activation, we observed increased colocalization between the repressive RNA-binding protein hnRNP M and Neat1 following LPS treatment (Fig. 2F). Our previous work showed that hnRNP M can repress innate gene expression in macrophages by slowing intron removal in inflammatory transcripts like Il6 (28). Thus, is it possible that dampened IL-6 expression in Neat1 KO macrophages (Figs. 5 G and I and 6E and SI Appendix, Fig. S5G) is driven in part by failure to sequester hnRNP M from nascent pre-mRNAs. HnRNP M is likely one of many RBPs whose paraspeckle association is stimulated by macrophage activation. The dual action of Neat1 in promoting expression of genes turned on by LPS and dampening expression of genes turned off by LPS may mean that paraspeckles can sequester RBPs with opposing functions to simultaneously activate and inhibit expression of innate genes. Alternatively, it is possible that by redistributing components of the transcription/RNA processing machineries to inflammatory M1 transcripts, paraspeckles could indirectly down-regulate M2 gene expression, by virtue of denying them access to the nucleus’ limited gene expression machinery. Future experiments that add Neat1 back to Neat1 KO macrophages, as well as experiments designed to test whether knockdown of paraspeckle proteins phenocopy loss of Neat1 will overcome current limitations of our approach and provide important insights into the molecular mechanisms driving paraspeckle-mediated innate immunity. Together these findings further our understanding of how biomolecular condensate assembly can be regulated by environmental cues and how dynamic control of these complexes enables rapid calibration of gene expression in the nucleus.

Materials and Methods

RNA FISH and Immunofluorescence Microscopy.

Fluorescence in situ hybridization–immunofluorescence (FISH-IF) was used to simultaneously visualize Neat1 lncRNA and proteins. See extended methods for details. A pool of 48 Neat1 smFISH (single-molecule fluorescence in situ hybridization) RNA probes were purchased from Stellaris® (SMF-3010-1, Mouse Neat1 Middle Segment with Quasar® 570 Dye). For immunofluorescence microscopy, primary antibodies (1:200) against SFPQ (Abcam, ab177149), PSP1 (Abcam, ab214012), NONO (Abcam, ab133574), HNRNP M (Abcam, ab177957), FUS (Abcam, ab243880), BRM (Abcam, ab240648), BRG1 (Abcam, ab110641), NF-κB (Active Motif, 40916), RBM14 (Abcam, ab70636), STAT1 (CST, #9167), and DIS3 (Proteintech, 14689-1-AP) were used along with 10% BSA (NEB #B9200) as a blocking agent.

Primary Cell Culture.

Bone marrow–derived macrophages (BMDMs) were differentiated in DMEM containing 20% fetal bovine serum (FBS), 1 mM sodium pyruvate, and 10% myeloid colony stimulating factor (MCSF)-conditioned media.

Cell Lines and Cell Culture.

Low passage stocks of RAW 264.7 macrophages obtained from ATCC (TIB-71), were cultured at 37 °C/5% CO₂ in complete media containing high glucose DMEM, with 10% FBS and 0.2% HEPES. dCas9 iBMDMs obtained from the Carpenter Lab at University of California, Santa Cruz (UCSC), were cultured at 37 °C/5% CO₂ in complete media containing high glucose DMEM, with 10% FBS and 0.2% HEPES.

Agonist Treatments.

LPS stimulations were done with 100 ng/mL LPS (InvivoGen) for RAW 264.7 macrophages and 10 ng/mL LPS for BMDMs for times indicated. Additional treatments were carried out with 100 ng/mL Pam3CSK4 (InvivoGen), 1 μg/mL ISD, or 500 ng/mL poly(I:C).

Salmonella Infections.

A bacterial inoculum was prepared by growing the bacteria to log phase with an optical density at 600 nm (OD600) of 1.0. Bacteria were washed and suspended in HBSS and added to the macrophages (at MOI 10), followed by a brief centrifugation step. The cells were then incubated at 37 °C for 20 min, washed three times with HBSS, and fresh media containing gentamycin (100 µg/mL) was added for an hour, then gentamycin (10 µg/mL) for the remainder of the infection.

Vesicular Stomatitis Viral (VSV) Infections.

Neat1 gRNA or off-target gRNA iBMDMs were infected with VSV-GFP at MOI of 1 in serum-free DMEM. After 1 h of incubation with media containing virus, the supernatant was removed, and complete media was added. Cells were harvested with TRIzol at time points indicated.

siRNA Transfections.

Transfection was carried out using Fugene SI reagent along with 50 μM of ThermoFisher siRNA stock against [Skiv2l2 (Mtr4); s90745), Zcchc8; s89034, Zcchc7; s38623, Dis3; s91196, and Exosc10; s78572] according to the manufacturer’s instructions. Silencer Select Negative Control #1 (ThermoFisher, 4390843) served as a negative control.

RNA Sequencing.

Ribodepletion library preparation was carried out by the Baylor College of Medicine Genomic and RNA Profiling Core (GARP) in biological triplicate. RNA sequencing (150 bp paired-end reads) was performed on an Illumina NovaSeq 6000 with S4 flow cell.

Data Analysis and Presentation.

GraphPad Prism software (Version 10) was used to perform statistical analysis of data and generate graphs. RNA and protein colocalization was measured using Coloc 2 plugin in Fiji (ImageJ) software.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Data, Materials, and Software Availability

All study data are included in the article and/or supporting information. RNA-seq data from Neat1 KO BMDMs has been deposited at NCBI GEO (GSE255322) (39).