Host‐Directed Antiviral Activity of SB2960 Through Selective Induction and Remodeling of Stress Granules

Abstract

The increasing prevalence of emerging infectious diseases highlights the urgent need for broad‐spectrum antiviral therapeutics. Here, the discovery of SB2960, a small molecule, is reported that promotes stress granule (SG) formation and exhibits potent antiviral activity across diverse viral species. SB2960 suppresses viral replication with minimal cytotoxicity by modulating host antiviral immune responses. Target identification studies reveal that SB2960 engages receptor for activated protein C kinase 1 (RACK1), a key regulator of SG dynamics. Integrated transcriptomic and proteomic analyses further demonstrate that SB2960 alters the composition of SG and modulates the expression of antiviral genes. These findings establish SB2960 as a promising host‐directed broad‐spectrum antiviral agent and a valuable probe for investigating SG biology in the context of viral infection.

Amid the ongoing threat of emerging viral pathogens, host‐directed antivirals offer a strategy to overcome viral mutation and drug resistance. SB2960, a small‐molecule inducer of stress granules (SGs), exhibits potent broad‐spectrum antiviral activity with minimal cytotoxicity. By selectively remodeling SG composition and engaging receptor for activated protein C kinase 1 (RACK1), SB2960 provides a chemical tool to dissect SG biology and a promising lead for antiviral therapeutics.

1. Introduction

The persistent threat of emerging infectious diseases underscores the urgent need for broad‐spectrum antiviral therapies. The global impact of the COVID‐19 pandemic revealed critical gaps in our worldwide preparedness and the inherent limitations of pathogen‐specific antiviral strategies. More recently, outbreaks of mpox and avian influenza have further emphasized the unpredictability and rapid spread of novel viral pathogens.^{[ 1} , ^{2 ]}

RNA viruses, in particular, present a formidable challenge due to their high mutation rates and genetic variability, which enable them to evade immune surveillance and develop resistance to conventional antivirals.^{[ 3} , ⁴ , ^{5 ]} Compounding this issue, novel viral pathogens often emerge unpredictably, making it difficult to maintain an effective and adaptive antiviral arsenal. Without broad‐acting countermeasures, these rapidly evolving viruses pose a substantial risk of causing severe and uncontrolled outbreaks.

Conventional antiviral development has primarily focused on targeting viral enzymes or structural proteins.^{[ 6 ]} However, these virus‐specific approaches are often undermined by even minor genomic changes.^{[ 7} , ⁸ , ^{9 ]} Consequently, there is growing interest in host‐directed antiviral strategies that modulate cellular pathways essential for viral replication or immune regulation.^{[ 10} , ^{11 ]} By targeting conserved host processes, these approaches offer the potential for durable efficacy across diverse viral families, regardless of viral mutation or the emergence of new variants.

One host mechanism that has recently garnered significant attention is the formation of stress granules (SGs). SGs are dynamic, membraneless cytoplasmic organelles composed of untranslated mRNAs, RNA‐binding proteins, translation initiation factors, and regulatory components.^{[ 12} , ^{13 ]} They form reversibly in response to various cellular stresses, including viral infection, and function to halt translation, conserve energy, and promote cell survival under stress.^{[ 14 ]} During viral infection, SG formation is often triggered by double‐stranded RNA (dsRNA), a replication intermediate of many RNA viruses. dsRNA is recognized by pattern recognition receptors (PRRs) such as RIG‐I‐like receptors (RLRs) and protein kinase R (PKR), which activate downstream antiviral signaling cascades. While these pathways are critical for initiating host defense, excessive or sustained activation can lead to cellular dysfunction and apoptosis.^{[ 15} , ¹⁶ , ^{17 ]} SGs modulate these responses by sequestering viral and host factors, suppressing viral translation, and buffering signaling pathways, thereby functioning as both antiviral effectors and cytoprotective regulators (Figure 1 ).^{[ 18 ]}

Figure 1.

Cytoprotective and antiviral roles of stress granules (SGs). Viral double‐stranded RNA (dsRNA) activates innate immune sensors, including RIG‐I‐like receptors (RLRs) and protein kinase R (PKR). PKR‐mediated phosphorylation of eIF2α triggers SG assembly, which suppresses excessive antiviral responses to prevent cell death while sequestering viral and host factors to inhibit translation and replication.

Building on this concept, we previously demonstrated that pharmacological enhancement of SG formation confers antiviral activity.^{[ 19 ]} Using live‐cell imaging in G3BP1‐GFP U‐2 OS cells treated with poly(I:C), a synthetic mimic of viral dsRNA, we screened a structurally diverse chemical library built using the privileged substructure‐based diversity‐oriented synthesis (pDOS) strategy^{[ 20 ]} and identified benzopyranylpyrazole derivatives that induced SG formation. Among them, SB2901 (previously C01) exhibited the most robust SG‐inducing activity.^{[ 19 ]} In SARS‐CoV‐2‐infected cells, SB2901 promoted SG assembly and suppressed the expression of viral nucleocapsid (N) protein, supporting the therapeutic potential of SG induction. However, its cytotoxicity upon prolonged exposure limited its translational prospects.

Here, we present a host‐directed antiviral strategy based on pharmacological enhancement of SG formation. Through structure‐activity relationship (SAR) optimization of benzopyranylpyrazole scaffolds, we developed SB2960, a low‐toxicity SG modulator that activates antiviral immune signaling and displays broad‐spectrum antiviral efficacy. Using a molecular glue degrader probe,^{[ 21 ]} we identified receptor for activated protein C kinase 1 (RACK1) as a mechanistically relevant protein associated with SB2960‐mediated SG regulation. Transcriptomic analyses further revealed that SB2960‐induced SGs are compositionally distinct from those induced by sodium arsenite, a canonical inducer of oxidative stress.^{[ 22} , ^{23 ]} Together, these results highlight SB2960 as a promising therapeutic candidate and a chemical tool to dissect SG‐mediated antiviral defense.

2. Results

2.1. SAR Optimization of Benzopyranylpyrazoles

To enhance SG induction while minimizing cytotoxicity, we undertook a comprehensive SAR campaign (Figure 2A). Our previous findings indicated that benzopyranylpyrazole derivatives bearing para‐trifluoromethyl‐phenyl (p‐CF₃‐Ph) group at the R¹ position consistently promoted SG formation more effectively than analogs with alternative aryl or alkyl groups, which exhibited negligible activity.^{[ 19 ]} Consequently, we fixed the R¹ position as p‐CF₃‐Ph to maintain SG‐enhancing activity.

Figure 2.

Structure‐activity relationship (SAR) analysis of benzopyranylpyrazole‐based stress granule (SG) inducers. A) SAR strategy highlighting modification at R¹, R², and R³ positions on the benzopyranylpyrazole scaffold. B,C) Quantification of SG formation in Vero cells treated with each analog (20 µm) or sodium arsenite (NaAs, 250 µm) as a positive control. SGs were visualized by G3BP1 immunofluorescence, and SG number per cell was quantified. Data represent mean ± SD (n ≥ 3).

Next, we optimized the R² position (Figure 2A,B; Figure S1, Supporting Information). Substitutions on the piperazine nitrogen had previously compromised aqueous solubility and SG‐inducing efficacy. To address this limitation, we evaluated a series of small substituents, including methyl, acetyl, and trifluoroacetyl groups (SB2902, SB2903, SB2906), as well as a hydroxyethyl moiety (SB2905) (Figure S2, Supporting Information), but these analogs still showed poor solubility and failed to dissolve even at 20 µm. Introducing the morpholine or acetic acid group improved solubility but did not enhance SG formation (SB2908 and SB2909). Notably, N,N‐dimethylamino (SB2907), proline‐conjugated (SB2915), and (1H‐imidazol‐4‐yl)acetyl (SB2916) analogs modestly induced SGs but retained or increased cytotoxicity.

A key breakthrough arose with the incorporation of heteroaromatic rings at the piperazine nitrogen, which improved SG induction while reducing toxicity. The N‐methylimidazole analog SB2913 demonstrated optimal properties, combining robust SG formation with favorable solubility. The (pyridin‐3‐yl)acetyl analog (SB2904) also induced SGs at 40 µm (Figure S3, Supporting Information). In contrast, analogs lacking either carbonyl or methylene linkers (SB2910–SB2912) or bearing a carboxylic acid substituent on the pyridine (SB2914) exhibited poor solubility or activity. The (1H‐imidazol‐4‐yl)acetyl analog (SB2916) was active but cytotoxic at higher doses. To explore alternatives to the piperazine scaffold, we installed morpholine directly onto the core structure (SB2921–SB2925). Most showed poor solubility, and notably, SB2923—bearing an N‐methylimidazole group at the R³—retained solubility but lacked SG‐inducing activity, suggesting that complete replacement of the piperazine core is unfavorable. These findings highlight N‐methylimidazole as the optimal R² substituent for balancing potency, solubility, and safety.

We then explored SAR at the R³ position using SB2913 as the core scaffold (Figure 2A,C). A series of non‐aromatic and alkyl substituents, including methyl, dimethyl, acetyl, α‐chloroacetyl, hydroxyethyl, morpholine, proline (SB2951–SB2955, SB2959, SB2968), and benzyl (SB2956), failed to induce SGs. Aromatic substitutions such as 4‐methoxycarbonylbenzoyl (SB2958) or its bicyclo[1.1.1]pentyl analog ^{[ 24 ]} (SB2957) also lacked activity. In contrast, introducing a 2‐pyridoyl group resulted in a highly active analog, SB2960. Notably, 3‐ and 4‐pyridoyl analogs (SB2962, SB2963) showed reduced activity, and additional linker atom or polar substituents on the pyridine ring (SB2964, SB2965, SB2969) eliminated SG induction. The pyrazinoyl analog (SB2967) suffered from poor solubility, while the N‐methylimidazolyl analog (SB2961) was cytotoxic (Figure S4A, Supporting Information). Similar trends were observed in analogs with a (pyridin‐3‐yl)acetyl R² group (SB2904, SB2931–SB2936; Figure S4B, Supporting Information), suggesting conserved R³ effects across different R² substituents. Together, these findings identified SB2960 as the most promising analog, demonstrating potent SG‐inducing activity, low cytotoxicity, and good aqueous solubility, making it suitable for further evaluation as an antiviral agent.

2.2. Functional Characterization of SB2960

Next, we characterized SB2960 using a series of functional assays to assess SG formation, antiviral signaling, and cytotoxicity. Immunofluorescence in Vero cells stained with anti‐G3BP1 revealed a dose‐dependent increase in SG formation, with an EC₅₀ of 12.31 µm, whereas the closely related inactive analog SB2910 did not induce detectable SGs even at 100 µm (Figure 3A; Figure S5, Supporting Information). Time‐lapse imaging in G3BP1‐GFP U‐2 OS cells showed that SG formation peaked ≈5 h post‐treatment (Figure S6, Supporting Information).

Figure 3.

SB2960 induces stress granules (SGs), modulates innate immunity, and exhibits low cytotoxicity. A) Dose‐dependent SG induction in Vero cells by SB2960, visualized by G3BP1 staining. SB2910 served as an inactive control and did not induce SG formation even at 100 µm. B) qRT‐PCR analysis of antiviral genes in poly(I:C)‐stimulated G3BP1‐GFP U‐2 OS cells treated with SB2960 (20 µm). Data represent mean ± SD (n = 3–4). Statistical values were calculated using one‐way ANOVA. **P < 0.01; ***P < 0.001; **** P < 0.0001. C) Proteomic profiling comparing poly(I:C) versus poly(I:C) + SB2960 in G3BP1‐GFP U‐2 OS cells. The line graph (top) shows the average expression of proteins upregulated by poly(I:C) and downregulated by SB2960 (n = 3). The bar graph (bottom) presents the top seven Gene Ontology (GO) biological processes enriched in this protein cluster, ranked by p‐value. D,E) Annexin V/PI flow cytometry and LDH release assay in Vero cells showing minimal cytotoxicity up to 100 µm SB2960, in contrast to sodium arsenite (NaAs) (n = 3). F) Co‐immunofluorescence staining for G3BP1 and SARS‐CoV‐2 nucleocapsid (N) protein in infected Vero cells treated with SB2960. Scale bar is 30 µm.

To determine whether SB2960 acts through canonical SG nucleators, we silenced G3BP1 using siRNA in Vero cells and quantified SB2960‐induced SGs by immunofluorescence using TIA‐1, another well‐established SG marker. G3BP1 knockdown markedly reduced SG counts compared with the non‐targeting control, although the SG EC₅₀ values remained similar between the two conditions (Figure S7, Supporting Information). In a complementary experiment, SB2960 treatment in cells expressing V5–G3BP1–miniSOG led to robust co‐localization of endogenous G3BP2 with SB2960‐induced puncta, confirming recruitment of both G3BP paralogs by SB2960 (Figure S8, Supporting Information). Consistent with a physiological SG response, SB2960‐induced stress granules disassembled after compound washout and returned close to baseline by ≈20 h (Figure S9, Supporting Information). These results indicate that SB2960 promotes SG assembly via the G3BP1/2 axis.

To determine its effect on antiviral responses, we performed quantitative real‐time PCR (qRT‐PCR) on key innate immune genes in poly(I:C)‐stimulated cells (Figure 3B). SB2960 significantly enhanced IFN‐β expression—an essential component of type I interferon response—in the presence of poly(I:C), but had minimal impact in unstimulated cells. Interestingly, SB2960 attenuated expression of downstream inflammatory genes, including RIG‐I, ISG56, and RANTES under the same conditions, suggesting selective amplification of antiviral signaling while suppressing excessive inflammation.

We next performed global quantitative proteomic profiling using tandem mass tags (TMTs) and LC‐MS/MS. Cells were either untreated, poly(I:C)‐stimulated with DMSO, or poly(I:C)‐stimulated with SB2960. Among 8154 quantified proteins, poly(I:C) alone altered 149 proteins (1.8%), whereas co‐treatment with SB2960 resulted in 851 differentially expressed proteins (10.4%), the majority being downregulated (800 proteins, 9.8%) (Figure S10, Supporting Information). Gene ontology (GO) analysis revealed that poly(I:C) activated pathways related to type I interferon response, cytokine signaling, and inflammation. In contrast, SB2960 co‐treatment broadly reversed these transcriptional signatures, shifting the expression profiles toward baseline (Figure 3C). These data support that SB2960 dampens broad inflammatory responses while preserving essential antiviral signaling. We also assessed cytotoxicity using Annexin V/propidium iodide (PI) staining after 24 h of treatment (Figure 3D). Even at 100 µm—far exceeding its EC₅₀ for SG induction—SB2960 caused minimal apoptosis or necrosis. In contrast, sodium arsenite at 500 µm, a concentration near its SG EC₅₀ (441 µm; Figure S11, Supporting Information), significantly increased Annexin V/PI double‐positive populations. Lactate dehydrogenase (LDH) assays showed consistent results in both Vero and G3BP1‐GFP U‐2 OS cells (Figure 3E; Figure S12, Supporting Information). SB2960 caused only baseline membrane damage, whereas sodium arsenite markedly increased extracellular LDH. Based on LDH release, the CC₅₀ of sodium arsenite was 1345 µm (Figure S11, Supporting Information). Thus, SB2960 exhibits a substantially wider safety margin (CC₅₀/EC₅₀ > 8.1) than sodium arsenite (CC₅₀/EC₅₀ ≈ 3.0).

Finally, we assessed antiviral efficacy through co‐immunofluorescence staining for G3BP1 and SARS‐CoV‐2 N protein (Figure 3F). SB2960 treatment (3.7 µm) robustly induced cytoplasmic G3BP1 puncta and markedly decreased N protein levels, from 57.30% (DMSO) to 20.65% at 1.2 µm and 10.59% at 3.7 µm (Figure S13, Supporting Information), demonstrating a strong association between SG induction and viral suppression.

2.3. Broad‐Spectrum Antiviral Activity of SB2960

Given that SG formation is a conserved host defense mechanism against viral infection,^{[ 25 ]} we hypothesized that a potent SG inducer, SB2960, may exhibit broad‐spectrum antiviral activity. To test this, we evaluated the antiviral efficacy of SB2960 and selected analogs against a panel of clinically relevant viruses, including SARS‐CoV‐2 (ancestral, alpha, beta, gamma, delta, and omicron variants), SARS‐CoV (both positive‐strand RNA viruses), and SFTSV (severe fever with thrombocytopenia syndrome virus, a negative‐strand RNA virus) (Table 1 ), representing distinct viral families.

Table 1.

Dose‐response analysis of benzopyranylpyrazole analogs against multiple viral pathogens. Half‐maximal inhibitory concentrations (IC₅₀) and cytotoxic concentration (CC₅₀) values were determined in Vero cells using immunofluorescence‐based quantification of viral antigens: nucleocapsid protein for SARS‐CoV‐2 and SFTSV, and spike protein for SARS‐CoV. Data represent mean values from duplicate experiments.

Cpd	R2	R3	IC50 [µM]								CC50 [µM]
			SARS‐CoV‐2 variants						SARS‐CoV	SFTSV	SARS‐CoV‐2 variants						SARS‐CoV	SFTSV
			Ancestral	Alpha	Beta	Gamma	Delta	Omicron			Ancestral	Alpha	Beta	Gamma	Delta	Omicron
SB2901			3.516	4.343	3.918	4.798	3.305	2.206	2.773	7.13	14.18	9.869	7.243	11.74	7.386	5.308	7.532	7.48
SB2904			6.99	6.714	6.753	6.903	4.725	2.639	6.173	7.02	>100	>100	>100	>100	>100	37.67	>100	>100
SB2907			3.565	5.552	3.98	6.471	3.543	2.048	2.844	9.91	15.26	15.55	15.58	16.92	15.69	8.493	14.88	14.91
SB2910			76.02	54.99	83.48	>100	80.25	36.54	39.26	40.03	>100	>100	>100	>100	>100	>100	>100	>100
SB2913			5.361	5.798	5.698	5.543	5.367	4.698	4.639	5.16	>100	>100	>100	>100	>100	56.5	>100	>100
SB2915			2.759	4.106	3.245	4.572	2.567	0.979	2.267	3.71	7.6	7.458	7.608	7.892	7.448	3.719	11.56	3.92
SB2935			5.408	5.232	5.142	4.922	5.025	4.242	7.128	1.73	>100	>100	>100	>100	>100	>100	>100	>100
SB2936			8.481	7.119	7.855	7.137	5.907	5.973	8.353	6.12	>100	>100	>100	>100	>100	71.67	>100	>100
SB2960			5.128	4.07	4.45	3.774	4.234	5.449	4.436	3.01	>100	>100	>100	>100	>100	>100	>100	>100
SB2961			3.739	3.42	3.546	3.633	3.669	3.571	4.916	2.91	6.846	5.759	6.127	5.802	6.656	3.625	7.008	7.34
SB2962			19.48	23.02	22.53	19.3	22.69	22.1	26.1	7.46	41.57	38.82	37.03	38.02	38.47	23.34	54.04	40.31
SB2965			33.47	12.05	11.93	10.84	16.35	9.639	13.51	16.81	>100	>100	>100	>100	>100	73.84	>100	>100

Antiviral activity was assessed by immunofluorescence‐based assays, and IC₅₀ and CC₅₀ values were determined based on viral protein expression and total cell counts. SB2960 (highlighted in sky blue, Table 1) demonstrated potent antiviral activity against all tested viruses, with IC₅₀ values ranging from 3.01 to 5.45 µm and negligible cytotoxicity (CC₅₀ >100 µm). A similar effect was observed in Calu‐3 cells (Table S1, Supporting Information). In addition, we tested SB2960 against dengue virus (DENV) in Huh7 cells and observed significant antiviral activity (Figure S14, Supporting Information). These data support SB2960's strong SG‐inducing profile and its potential as a host‐directed, broad‐spectrum antiviral. In contrast, although comparably potent, the parent compound SB2901 exhibited considerable cytotoxicity (CC₅₀ = 5.3–14.2 µm), which limits its therapeutic applicability. A structurally similar analog, SB2910, which lacks SG‐inducing activity (shaded in gray, Table 1), exhibited substantially weaker antiviral effects (IC₅₀ = 36.54 to >100 µm).

In addition to the immunofluorescence‐based assay, we quantified infectious SARS‐CoV‐2 in culture supernatants by TCID₅₀ (50% tissue culture infectious dose). SB2960 reduced viral titers in a dose‐dependent manner, achieving a 5.42 log₁₀ reduction at 12.5 µm, whereas remdesivir under the same conditions produced a 2.67 log₁₀ reduction at 25 µm (Figure S15A, Supporting Information).

Furthermore, intracellular viral genomic RNA measured by qRT‐PCR showed a 4.16 log₁₀ reduction at 12.5 µm SB2960 at 24 h post‐infection, whereas 25 µm remdesivir resulted in only a 1.24 log₁₀ reduction relative to the DMSO vehicle (Figure S15B, Supporting Information), indicating that SB2960 suppresses SARS‐CoV‐2 replication across independent quantitative readouts.

To further evaluate the link between SG induction and antiviral efficacy, we compared SG EC₅₀ values with antiviral activity across a panel of benzopyranylpyrazole analogs (Figure S16A, Supporting Information). A strong correlation was observed between SG‐inducing potency and antiviral efficacy against ancestral SARS‐CoV‐2 (Figure 4A). This relationship was further validated with the omicron variant, SARS‐CoV, and SFTSV, which exhibited similar trends (Figure S16B, Supporting Information). These findings support SG induction as a key mechanism underlying the antiviral activity of this compound series.

Figure 4.

Antiviral activity of stress granule (SG) inducers and synergy with remdesivir. A) Correlation between SG‐inducing potency (EC₅₀) and antiviral activity (IC₅₀ against ancestral SARS‐CoV‐2) across benzopyranylpyrazole analogs. B) Immunofluorescence images showing enhanced antiviral effect by SB2960 and remdesivir combination against ancestral SARS‐CoV‐2. C) ZIP synergy analysis using SynergyFinder+ demonstrating strong synergistic interaction in SB2960–remdesivir combination.

We next investigated whether SB2960 could synergize with approved antiviral agents. Combination assays with remdesivir—a nucleoside analog targeting viral RNA‐dependent RNA polymerase ^{[ 26 ]}—revealed pronounced synergy against SARS‐CoV‐2 (Figure 4B,C). Co‐treatment significantly enhanced antiviral activity, reducing the IC₅₀ of SB2960 more than tenfold at higher remdesivir concentrations. A ZIP synergy score of 95.34 was observed at 1.11 µm SB2960 and 1.56 µm remdesivir. Comparable synergy was also observed against the omicron variant (Figure S17, Supporting Information). A similarly active analog, SB2935, exhibited strong synergy in both ancestral and omicron SARS‐CoV‐2 infection models (Figure S18, Supporting Information). Likewise, co‐treatment with nirmatrelvir, an FDA‐approved SARS‐CoV‐2 protease inhibitor,^{[ 27 ]} resulted in robust synergy, yielding a maximum ZIP synergy score of 94.25 (Figure S19, Supporting Information). These findings highlight the therapeutic potential of combining SG inducers with direct‐acting antivirals to enhance efficacy and reduce resistance.

2.4. Cellular Target Profiling of SB2960

To elucidate the mechanism of action of SB2960, we aimed to identify its direct cellular targets. Given that SB2960 functions via host‐directed mechanisms, identifying its binding partners is essential to understanding its antiviral effects. However, existing target ID methods such as UV‐induced crosslinking (FITGE)^{[ 28 ]} or thermal stability shift (TS‐FITGE)^{[ 29 ]} approaches are poorly suited for this context, as SGs are sensitive to physical perturbation, including heat and UV, which may confound native interactions.^{[ 30} , ^{31 ]}

We therefore employed a chemical proteomic strategy using a molecular glue degrader approach. We incorporated a fumarate electrophilic handle—previously validated for E3 ligase RNF126 recruitment—into SB2960 at the modifiable piperazine position (Figure 5A; Figure S20, Supporting Information). While derivatives containing 2‐pyridoyl or 3‐hydroxy‐2‐pyridoyl group were insoluble or unstable, the N‐methylimidazole‐conjugated analog retained SG‐inducing activity despite some cytotoxicity. This probe, termed SB2991, enabled proteasomal degradation of SB2960‐interacting proteins under near‐physiological conditions.

Figure 5.

Identification of RACK1 as a mechanistic mediator of SB2960. A) Chemical structure of SB2991, a molecular glue degrader probe derived from SB2960 for chemoproteomic target identification. B) 2D differential gel electrophoresis (2D‐DIGE)‐based identification of proteins selectively degraded by SB2991. Green spots indicate SB2991‐depleted proteins and were analyzed by LC‐MS. C,D) Western blot showing dose‐dependent RACK1 degradation by SB2991 and rescue by SB2960 treatment. Numbers below each blot represent mean values from independent experiments, normalized to the DMSO control. E) Cellular thermal shift assay (CETSA) showing thermal stabilization of RACK1 upon SB2960 treatment in G3BP1‐GFP U‐2 OS cells; α‐tubulin served as control. Data represent mean ± SD (n = 2). F) G3BP1 immunofluorescence analysis in Vero cells with or without RACK1 knockdown and SB2960 treatment. RACK1 knockdown enhanced SB2960‐induced SG formation. Scale bar is 30 µm. Data represent mean ± SD (n = 3).

2D difference gel electrophoresis (2D‐DIGE) was used to compare the proteomes of SB2991‐ and DMSO‐treated cells. Lysates were fluorescently labeled (Cy5 for SB2991 and Cy3 for DMSO) and combined. Proteins unaffected by treatment appeared as yellow spots (merged Cy3/Cy5), whereas selectively depleted proteins appeared as green‐only spots (Cy3) (Figure 5B). A distinct green spot was detected uniquely in SB2991‐treated samples. Mass spectrometry identified this protein as RACK1, a scaffold protein involved in ribosomal regulation and antiviral signaling (Figure S21, Supporting Information).^{[ 32} , ^{33 ]} RACK1 is known to facilitate replication of several RNA viruses, including flaviviruses and coronaviruses,^{[ 34} , ³⁵ , ³⁶ , ^{37 ]} suggesting a mechanistic role in SB2960's antiviral effects.

Western blot analysis confirmed dose‐dependent RACK1 degradation by SB2991 (Figure 5C), which was competitively blocked by excess SB2960 (Figure 5D), indicating specific engagement. Cellular thermal shift assays (CETSA) further validated this interaction, showing increased thermal stability of RACK1 upon SB2960 treatment in Vero and G3BP1‐GFP U‐2 OS cells (Figure 5E; Figure S22, Supporting Information), while α‐tubulin served as a negative control. Surface plasmon resonance (SPR) analysis also revealed concentration‐dependent binding of SB2960 to immobilized RACK1, following a 1:1 binding model (K _d 5.65 µm; k_on 2.89 × 10² M⁻¹·s⁻¹; k_off 8.78 × 10⁻⁴ s⁻¹), whereas the inactive analog SB2910 showed no detectable binding (Figure S23, Supporting Information). These data provide direct biophysical evidence that SB2960 engages RACK1.

To assess the functional role of RACK1 in SG formation, we performed siRNA‐mediated knockdown in Vero cells (Figure 5F; Figure S24, Supporting Information), as complete knockout of RACK1 is lethal.^{[ 38 ]} Partial RACK1 depletion did not induce SGs alone but markedly enhanced SB2960‐induced SG formation at submaximal doses. These results support RACK1 as a negative regulator of SG assembly and a mechanistically relevant target of SB2960.

RACK1 knockdown modestly increased SB2960‐induced SG puncta number without changing puncta area (Figure S25, Supporting Information), consistent with a model in which RACK1 primarily limits SG nucleation efficiency rather than granule growth. Comparative docking studies to the human RACK1 (PDB 4AOW) using active (SB2960, SB2935) versus inactive (SB2963, SB2910) analogs revealed a conserved hydrogen bonding between Gln20 and the carbonyl adjacent to the piperazine and a π–sulfur interaction of heteroaromatic rings at R² with Cys153 in active compounds (Figure S26, Supporting Information). Thr199 also has a carbon‐hydrogen contact that stabilized the piperazine orientation. These modeled interactions support a structure‐dependent RACK1 binding mode, consistent with observed SAR.

2.5. Transcriptomic Profiling of SB2960‐Induced Stress Granules

To investigate how SB2960 modulates the RNA composition of SGs, we employed CAP‐seq, a photocatalytic RNA proximity labeling method based on a G3BP1‐miniSOG fusion protein (Figure 6A).^{[ 39 ]} Upon blue‐light exposure, miniSOG generates singlet oxygen that oxidizes nearby guanosines, which, in the presence of propargyl amine (PA), are converted into alkyne‐bearing products. This enables selective biotin conjugation via click chemistry and enrichment of SG‐associated RNAs. Cells expressing G3BP1‐miniSOG were treated with or without SB2960, illuminated, and subjected to copper‐catalyzed azide‐alkyne cycloaddition (CuAAC). Biotinylated RNAs were captured with streptavidin beads and analyzed by next‐generation sequencing (NGS).

Figure 6.

Transcriptomic profiling of SB2960‐induced stress granules (SG) using CAP‐seq. A) Schematic of CAP‐seq workflow for proximity labeling of SG‐associated RNAs using G3BP1‐miniSOG. B) Comparative transcriptomic analysis of poly(I:C)‐ and SB2960‐induced SGs. Venn diagrams (left) show overlapping and distinct transcript subsets; GO Biological Process (GOBP) enrichment analysis (right) highlights the major functional categories identified in SG‐enriched fractions. C) Differential SG transcriptomic profiles between sodium arsenite (NaAs)‐ and SB2960‐treated cells.

SG‐localized transcripts were defined from three comparisons: 1) post‐ versus pre‐enrichment samples, 2) reactions with versus without PA, and 3) targeted G3BP1‐miniSOG versus untargeted miniSOG controls. Consistent depletion across 1) and 3) defined SG‐excluded RNAs. CAP‐seq has been validated previously to preserve SG integrity,^{[ 39 ]} confirming its suitability for interrogating SG transcriptomes. Differential analysis showed that in the presence of poly(I:C) and SB2960, CAP‐seq enriched 1480 transcripts and depleted 1109, which, after control filtering, refined to 339 SG‐proximal and 198 SG‐excluded transcripts (Figure 6B; Figure S27A, Supporting Information). SG CAP‐seq with poly(I:C) alone enriched only 132 transcripts, with GO analysis highlighting terms related to cell cycle, organelle organization, and biogenesis pathways. In comparison, the presence of SB2960 substantially broadens the SG transcriptome, with GO analysis revealing additional RNA categories linked to DNA repair, protein localization, and nucleobase metabolism, suggesting expanded SG functional capacity.

To compare pharmacological and oxidative stress‐induced SGs, we analyzed sodium arsenite‐treated CAP‐seq data. Of the 339 transcripts enriched under SB2960, 253 (74.6%) overlapped with the arsenite‐induced SG transcripts (1135 transcripts) (Figure 6C). Both arsenite‐induced and SB2960‐modulated SGs sequester transcripts involved in cell cycle regulation and DNA metabolic processes (Figure S27B, Supporting Information). However, the top GO terms diverge: oxidative stress enriches RNA processing, indicating broad transcriptional repression, whereas SB2960 preferentially enriches chromosomal segregation. This pattern suggests that SB2960 directs translational repression toward mitosis‐related transcripts, thus supporting safe, reversible cell‐cycle arrest, while sparing RNAs essential for metabolic balance and cellular homeostasis. Metascape analysis^{[ 40 ]} likewise revealed distinctions in excluded transcripts, revealing differences in ubiquitin‐related pathways (Figure S27C, Supporting Information). Because SG proteins frequently display elevated ubiquitination under stress,^{[ 31} , ⁴¹ , ^{42 ]} these findings suggest that SB2960‐induced SGs maintain essential metabolic and cellular functions while minimizing cytotoxic stress programs, distinguishing them from canonical oxidative stress‐induced SGs. Taken together, the above analysis indicates that SB2960 may fine‐tune SG composition to protect cell viability and facilitate recovery rather than enforcing a global translational shutdown.

3. Discussion

This study identified and characterized SB2960, a small‐molecule modulator that promotes SG formation and exerts host‐directed, broad‐spectrum antiviral activity. In contrast to classical SG inducers, such as sodium arsenite, SB2960 elicits SG formation without triggering apoptosis or membrane damage, thereby enabling a pharmacological approach to antiviral immunity with minimal cytotoxicity. Pharmacokinetic profiling of SB2960 further confirmed its therapeutic potential, demonstrating a long half‐life (T_1/2 = 4.59 h) and high oral bioavailability (F = 67.8%) (Table S2, Supporting Information).

Through quantitative proteomics, we found that poly(I:C) stimulation robustly activated antiviral and inflammatory pathways, whereas SB2960 treatment attenuated these responses and restored protein expression toward homeostatic levels. These results suggest that SB2960 fine‐tunes the innate immune response, enhancing beneficial antiviral signals (e.g., type I interferon induction) while suppressing pro‐inflammatory cascades that can contribute to tissue damage.

Transcriptomic profiling using CAP‐seq revealed that SB2960‐induced SGs largely overlap with those formed during oxidative stress, yet still display distinct differences across their transcriptomes. Both conditions sequestered transcripts related to cell‐cycle and chromosomal organization, consistent with a conserved role of SGs as dynamic mRNA triage centers.^{[ 43 ]} However, arsenite strongly enriched RNA‐processing and protein‐modification pathways, including ubiquitination (Figure S27, Supporting Information), indicative of widespread stress responses. By contrast, SB2960 preferentially enriches pathways linked to chromosomal segregation and metabolic homeostasis, reflecting a more controlled and reversible translational checkpoint. Moreover, SB2960‐treated cells selectively excluded transcripts related to mitochondrial translation, rRNA maturation, and global protein synthesis—functions critical for maintaining cellular viability. Together, these findings indicate that SB2960 induces SGs that preserve essential metabolic programs while maintaining antiviral capacity.

Mechanistically, our molecular glue degrader probe identified RACK1 as a direct interactor of SB2960. RACK1 is a multifunctional scaffold protein associated with ribosomal regulation and viral replication. Recent work indicates that loss of specific RACK1 functions does not globally suppress translation but instead perturbs ribosome‐linked regulatory complexes that influence SG dynamics.^{[ 44 ]} Consistent with this, partial RACK1 knockdown alone did not induce SGs but significantly potentiated SB2960‐mediated SG assembly, particularly at suboptimal doses. Combined with the observation that complete RACK1 loss is lethal,^{[ 38 ]} and that SB2960 does not induce toxicity under SG‐inducing conditions, these findings suggest that SB2960 modulates, rather than abolishes, RACK1 activity in a context‐dependent manner. While additional protein targets may contribute, RACK1 represents a key mediator.

In summary, SB2960 represents a novel chemical probe that induces SG formation with low cytotoxicity, modulates SG composition, and calibrates antiviral immune response. By engaging RACK1 and reconfiguring SG transcript content, SB2960 provides a powerful platform for dissecting SG biology and developing host‐directed antiviral strategies. More broadly, these findings highlight the therapeutic potential of selectively modulating SG assembly to reprogram host defense against diverse viral pathogens.

4. Experimental Section

Materials for Chemical Synthesis

All commercially available reagents for chemical synthesis were purchased from Sigma–Aldrich, Tokyo Chemical Industry Co., Ltd, or ThermoFisher Scientific and used without further purification unless otherwise specified. Solvents were purchased from commercial vendors and used without further purification unless otherwise mentioned. Dry solvents were prepared using the ultimate solvent purification system CT‐SPS‐SA [Glass Contour]. The progress of the reaction was monitored using thin‐layer chromatography (TLC) (silica gel 60, F₂₅₄ 0.25 mm) or liquid chromatography‐mass spectrometry (LC‐MS) LCMS‐2020 [Shimadzu]. Components on TLC were visualized by UV light (254 and 365 nm) or by treating the TLC plates with KMnO₄ or phosphomolybdic acid, followed by heating. Compounds were purified by medium‐pressure liquid chromatography (MPLC) Isolera One [Biotage] or reverse‐phase preparative high‐performance liquid chromatography (HPLC) with LC‐6AD [Shimadzu] using YMC‐Pack ODS‐A AA20S05‐2520WT (S‐5 µm, 12 nm, 20 × 250 mm) [YMC]. Eluents used for purification were reported in parentheses.

Compound Characterization

¹H and ¹³C NMR spectra were obtained on Agilent 400‐MR DD2 Nuclear Magnetic Resonance System [400 MHz, Agilent], Varian/Oxford As‐500 [500 MHz, Varian Associates], and Bruker AVANCE III HD 500 [500 MHz, Bruker]. Chemical shifts were reported in parts per million (δ, ppm). ¹H NMR spectra were calibrated using tetramethylsilane (TMS, 0.00 ppm) as an internal standard. ¹³C NMR spectra were calibrated using the residual solvent peak (CDCl₃ ¹³C: 77.00 ppm; (CD₃)₂CO ¹³C: 29.84 ppm; CD₃OD ¹³C: 49.00 ppm; (CD₃)₂SO ¹³C: 39.52 ppm). Multiplicities were indicated as follows: s (singlet); d (doublet); t (triplet); q (quartet); m (multiplet), dd (doublet of doublet), dt (doublet of triplet), tt (triplet of triplet), br s (broad singlet), and so on. Coupling constants were reported in Hz. Low‐resolution mass spectrometry (LRMS) was performed using an LCMS‐2020 [Shimadzu] system with electrospray ionization (ESI). High‐resolution mass spectrometry (HRMS) was confirmed by Orbitrap Exploris 120 [ThermoFisher] using ESI from the Research Facilities Center at Seoul National University (Department of Chemistry).

Cell Culture

Vero kidney epithelial cells derived from the African green monkey were obtained from the Korean Cell Line Bank [KCLB] (KCLB No.10081). Prof. Jin‐A Lee at Hannam University generously provided G3BP1‐GFP‐expressing U‐2 OS human osteosarcoma epithelial cells. Both G3BP1‐GFP‐expressing U‐2 OS and Vero cells were cultured in DMEM [Welgene], supplemented with 10% heat‐inactivated fetal bovine serum (FBS) [Gibco] and 1× antibiotic‐antimycotic solution [Welgene] under a humidified atmosphere containing 5% CO₂ at 37 °C.

For experiments involving authentic viruses, Vero cells were obtained from the American Type Culture Collection [ATCC] (ATCC CCL‐81). Huh‐7 cells were purchased from the Japanese Collection of Research Bioresources Cell Bank [JCRB] (JCRB0403). Both cells were cultured in DMEM [Welgene], supplemented with 10% heat‐inactivated FBS and 1× antibiotic‐antimycotic solution [Gibco] at 37 °C with 5% CO₂. Calu‐3 cells used in this study were clonal isolates with an enhanced proliferation rate compared to the parental Calu‐3 cell line obtained from ATCC [ATCC HTB‐55]. Calu‐3 cells were cultured in EMEM [ATCC] supplemented with 20% heat‐inactivated FBS, 1% MEM non‐essential amino acid solution [Gibco], and 2% antibiotic‐antimycotic solution [Gibco] at 37 °C with 5% CO₂.

For CAP‐seq experiments, U‐2 OS [NSTI‐BMCR] cells were cultured in McCoy's 5A medium [VivaCell] supplemented with 10% fetal bovine serum [Gibco] at 37 °C with 5% CO₂. U‐2 OS cell lines expressing G3BP1‐miniSOG and untargeted miniSOG were generated from previous work.^{[ 39 ]}

Stress Granule Imaging by Immunofluorescence

Vero cells were seeded at a density of 1 × 10⁴ cells per well in 96‐well black‐sided, clear‐bottom plates. Following complete adhesion, cells were incubated with complete medium containing 1.25 µg mL⁻¹ poly(I:C) (HMW) pre‐complexed with LyoVec [InvivoGen] for 5 h, followed by 5‐h incubation in complete medium containing the indicated compounds. Cells were then rinsed with PBS and fixed with 4% paraformaldehyde (PFA) in PBS for 20 min at room temperature (r.t.). After fixation, cells were washed three times with PBS and permeabilized with 0.5% Triton X‐100 in PBS for 10 min at 4 °C. Subsequently, cells were rewashed with PBS three times, blocked with 3% BSA in PBS for 1 h at r.t., and incubated overnight at 4 °C with anti‐G3BP1 primary antibodies (1:200 dilution in PBS with 1% BSA) [Santa Cruz, #sc‐81940]. After washing three times with PBS, cells were incubated with Alexa 488‐conjugated anti‐mouse secondary antibody (1:300 dilution in PBS with 1% BSA) [Abcam, #ab150113], and nuclei were stained with Hoechst 33342 (1:5000 dilution in PBS) [ThermoFisher] for 30 min. For the determination of EC₅₀ for sodium arsenite, cells were treated with various concentrations of sodium arsenite (1000–0.49 µm) using 1/2 serial dilutions. After 5‐h incubation in a humidified 5% CO₂ atmosphere at 37 °C, cells were washed with PBS and fixed with 4% PFA in PBS for 15 min at r.t. Cells were then rinsed three times with PBS and permeabilized with 0.5% Triton X‐100 in PBS at 4 °C for 10 min. Cells were washed three times with PBS, blocked with 3% BSA in PBS for 10 min at r.t., and incubated overnight at 4 °C with anti‐G3BP1 (1:300 dilution) [Santa Cruz, #sc‐41940] in PBS containing 1% BSA. Cells were washed three times with PBS and incubated with goat mAb anti‐mouse secondary antibody conjugated with Alexa 488 (1:300 dilution) [Abcam, #ab150113] for 1 h at r.t. Nuclei were stained with Hoechst 33342 (5000:1 dilution) [Invitrogen, H3570]. Cells were imaged using the INCell Analyzer 2500 [Cytiva]. The number of stress granule puncta per cell was quantified using Developer software [Cytiva].

Stress Granule Monitoring Using G3BP1‐GFP‐Expressing U‐2 OS Cell Line

G3BP1‐GFP‐expressing U‐2 OS cells were seeded at a density of 1 × 10⁴ cells per well in 96‐well black‐sided, clear‐bottom plates and incubated for 24 h. Cells were incubated with complete medium containing 1.25 µg mL⁻¹ poly(I:C) (HMW) pre‐complexed with LyoVec [InvivoGen] for 5 h, prior to well imaging. Nuclei were stained with Hoechst 33342 (1:7000 dilution in culture medium) [ThermoFisher] for 30 min. Images at the 0 min time point were acquired after staining and before compound treatment using INCell Analyzer 2500 [Cytiva]. Subsequently, either 20 µm SB2960 or 0.5% DMSO, serving as a vehicle control, was added in the same order as image acquisition. After compound addition, plates were scanned after 40, 100, 160, 220, 280, 460, 1230, and 1440 min. Using Developer software [Cytiva], images were analyzed to quantify total stress granule count per cell. Each value was normalized to that of 0 min and DMSO controls.

RNA Extraction and Quantitative Real‐Time PCR

G3BP1‐GFP‐expressing U‐2 OS cells were seeded at a density of 1 × 10⁵ cells per well in 12‐well plates. After complete adhesion, cells were incubated with complete medium containing 1.25 µg mL⁻¹ poly(I:C) (HMW) pre‐complexed with LyoVec [InvivoGen] for 5 h. After treatment, the medium was removed and replaced with fresh medium containing either 0.5% DMSO (vehicle) or the designated compound at a final concentration of 20 µm. Cells were further incubated for 24 h, washed with PBS, and then harvested.

Total RNA was extracted using TRIzol Reagent [Ambion/Invitrogen, #15596018] according to the manufacturer's instructions. Reverse transcription was performed using AccuPower CycleScript RT PreMix (dT20) [Bioneer, #K‐2044]. Quantitative RT‐PCR was conducted using KAPA SYBR Green ABI Prism 2× qPCR Master Mix [Kapa Biosystems, #KK4605]. Each 20 µL qPCR reaction contained 10 µL of 2× SYBR premix, 0.8 µL of 5 µm each of forward and reverse primers mix, 8.2 µL of nuclease‐free water, and 1 µL of cDNA. Each qPCR reaction was performed in technical duplicates. Gene expression was quantified using the comparative Ct method and normalized to human beta‐actin (hATCB).

Primer sequences used in this study were as follows:

Human IFNB1 (IFN‐β): forward 5′‐TGACCAACAAGTGTCTCCTCC‐3′,

Reverse 5′‐AGCCTCCCATTCAATTGCCA‐3′;

Human IL6 (IL‐6): forward 5′‐ACTCACCTCTTCAGAACGAATTG‐3′,

Reverse 5′‐CCATCTTTGGAAGGTTCAGGTTG‐3′;

Human DDX58 (RIG‐I): forward 5′‐AGAGCACTTGTGGACGCTTT‐3′,

Reverse 5′‐TCAGCAACTGAGGTGGCAAT‐3′;

Human CCL5 (RANTES): forward 5′‐TACACCAGTGGCAAGTGCTC‐3′,

Reverse 5′‐TGTACTCCCGAACCCATTTC‐3′;

Human ACTB (β‐actin): forward 5′‐AGAGCTACGAGCTGCCTGAC‐3′,

Reverse 5′‐AGCACTGTGTTGGCGTACAG‐3′.

Sample Preparation and Quantitative Proteome Analysis

Proteins were reduced and alkylated with 5 mM DTT [Sigma‐Aldrich, #43815] and 15 mM IAA [Sigma‐Aldrich, I1149], respectively. The SP3 protocol was applied with Sera‐Mag Carboxylate‐Modified SpeedBead [Cytiva, #45152105050250, #65152105050250]. Briefly, the proteins were mixed with Sera‐Mag Carboxylate SpeedBead at 24 °C for 5 min at 1000 rpm and washed using 80% ethanol [Merck, #1009832511]. Trypsin solution [ThermoFisher, #90057] in 50 mM TEAB [Sigma‐Aldrich, #18597] was added and incubated at 37 °C for 16 h at 1000 rpm. For tandem mass tag (TMT) labeling‐based quantitative proteomic analysis, the peptide samples were labeled with TMT 11‐131C label reagent [ThermoFisher; #A34808] according to the manufacturer's instructions. The labeled peptide samples were desalted using Pierce C18 Spin Tips [ThermoFisher; #84850] and dried using a GeneVac evaporator.

Subsequently, TMT‐labeled peptide analysis was performed using a high‐resolution mass spectrometer, Q Exactive HF‐X [ThermoFisher], coupled with an Ultimate 3000 nanoLC system [ThermoFisher]. The samples were dissolved in 0.1% FA [Sigma‐Aldrich, #85178] and loaded onto a trap column, PepMap C18 [ThermoFisher, #164535, 3 µm, 2 cm × 75 µm]. The loaded peptides were separated using an EASY‐Spray C18 column [ThermoFisher, #ES903, 2 µm, 50 cm × 75 µm] with a gradient of 5‒25% acetonitrile (ACN) [Merck, #1000291000] in 0.1% FA for 2 h at a flow rate of 250 nL min⁻¹ and a column temperature of 55 °C. Using a data‐dependent acquisition method, the Q Exactive HF‐X was set to acquire data in positive ion mode. Each cycle consisted of one full MS scan, followed by a maximum of 20 MS MS⁻¹ scans. Full MS scans were collected at a resolution of 120 000, with an AGC target of 3e6, a scan range of 375‒1500 m z⁻¹, and a maximum injection time of 50 ms. All higher‐energy collision‐induced dissociation (HCD) MS/MS spectra were acquired at a resolution of 45 000, with an AGC target of 1e5, a maximum injection time of 96 ms, an isolation window of 0.7 m z⁻¹, a collision energy of 32%, and a dynamic exclusion time of 45 sec.

MS raw files were processed using Proteome Discoverer software [ThermoFisher, version 2.4] and searched against the UniProtKB human database. The SEQUEST HT search engine was used with the following parameters: fully tryptic specificity; a maximum of two missed cleavages; a minimum peptide length of six amino acids; static modifications for TMT tags on lysine residues and peptide N‐termini, as well as carbamidomethylation of cysteine residues; dynamic modifications for oxidation of methionine; a precursor mass tolerance of 10 ppm; and a fragment mass tolerance of 0.02 Da. Peptide scoring was performed using Percolator with an identification threshold of 1% false discovery rate (FDR). No imputation of missing values was performed.

Determination of Apoptosis and Necrosis Using Flow Cytometry

Vero cells were seeded at a density of 8 × 10⁴ cells per well in 12‐well plates and incubated overnight. Cells were then treated with SB2960 (20 or 100 µm), sodium arsenite (NaAs) (500 µm), or 0.5% DMSO (vehicle control). Plates were incubated at 37 °C in a humidified 5% CO₂ incubator for 24 h. Cells were then harvested and subjected to flow cytometry analysis to assess apoptotic and necrotic cell death using the Annexin V‐FITC Apoptosis Staining/Detection Kit [Abcam, ab14085].

Flow cytometry was conducted using a CytoFLEX LX [Beckman Coulter] equipped with laser lines at 405, 488, 561, 638, and 375 nm. FITC emission was detected using a 525/40 nm BP (band pass) filter, excited by the 488 nm laser. The PE signal was generated with the 561 nm laser and detected through a 585/42 nm BP filter. For each sample, at least 10 000 cells were analyzed. Data acquisition and analysis were performed by CytExpert software [Beckman Coulter].

Lactate Dehydrogenase (LDH) Release Assay

Vero or G3BP1‐GFP‐expressing U‐2 OS cells were seeded at a density of 9 × 10³ cells per well in 96‐well plates and incubated overnight. After complete adhesion, cells were treated with various concentrations of SB2960 (0.39–100 µm) or sodium arsenite (NaAs) (3.9–1000 µm) using 1/2 serial dilutions in medium containing 0.5% DMSO. After 24‐h incubation in a humidified 5% CO₂ atmosphere at 37 °C, LDH release was measured using EZ‐LDH Cell Cytotoxicity Assay Kit [DoGen, DG‐LDH1000] according to the manufacturer's instructions. The absorbance at 450 nm was measured using a BioTek Synergy HT Microplate reader. The percentage of cytotoxicity was calculated by the following equation: Cytotoxicity (%) = (absorbance of treated wells – absorbance of low control)/(absorbance of high control – absorbance of low control) × 100. Corresponding background absorbance was subtracted for each well.

Viruses

The SARS‐CoV‐2 variants were obtained from the Korea Disease Control and Prevention Agency (KDCA). The SARS‐CoV‐2 variants used in this study were as the following; ancestral virus (βCoV/KOR/KCDC03/2020, NCCP43326), alpha variant (hCoV‐19/Korea/KDCA51463/2021, NCCP43381), beta variant (hCoV‐19/Korea/KDCA55905/2021, NCCP43382), gamma variant (hCoV‐19/Korea/KDCA95637/2021, NCCP43388), delta variant (hCoV‐19/Korea/KDCA119861/ 2021, NCCP43390) and omicron (BA.1) variant (hCoV‐19/Korea/KDCA447321/2021, NCCP43408). SARS‐CoV‐1 (HK39849) was a kind gift from Dr. Malik Peiris from Hong Kong University. SFTSV KADGH strain (NCCP43261) was obtained from KDCA, and DENV‐2 BR/01‐01 (GenBank JX073928) was a generous gift from Dr. Claudia N. Duarte dos Santos from Instituto Carlos Chagas. All viruses except DENV‐2 were propagated and quantified by plaque assays in Vero cells. DENV‐2 was propagated in C6/36 cells. All experiments involving infectious viruses were conducted at Institut Pasteur Korea in accordance with the guidelines of the KNIH, utilizing enhanced Biosafety Level 2 (BSL‐2) or Biosafety Level 3 (BSL‐3) containment procedures in laboratories approved by the KDCA.

Dose‐Response Curve (DRC) Assay with Coronaviruses

Vero cells were seeded at a density of 1.0 × 10⁴ cells per well in DMEM, supplemented with 2% FBS and 1× antibiotic‐antimycotic solution [Gibco/ThermoFisher] in black, 384‐well, µClear plates [Greiner Bio‐One], 24 h before the infection. Ten‐point, twofold serial dilutions were made from compounds to generate DRCs, with compound concentrations ranging from 0.2 to 100 µm. Serially diluted compounds were added to the cells ≈30 min before the infection. For viral infection, plates were transferred into the BSL‐3 containment facility, and the virus was inoculated at a multiplicity of infection (MOI) ranging from 0.03 to 0.1. The MOI for each virus was as follows: SARS‐CoV‐2 ancestral, gamma, and delta variant at 0.03 MOI, SARS‐CoV‐2 alpha, beta, and SARS‐CoV‐1 at 0.05 MOI, and SARS‐CoV‐2 omicron (BA.1) variant at 0.1 MOI. The MOIs for different coronaviruses were determined through independent experiments to achieve maximal infectivity and cell viability. The infected cells were fixed at 24 h post‐infection (hpi) except for the omicron variant, which was fixed at 48 hpi. Immunofluorescence assays further analyzed fixed cells. Cells were fixed with 4% PFA for 30 min at r.t. and permeabilized with 0.25% Triton X‐100 for 10 min. After blocking with 5% goat serum, cells were incubated with anti‐SARS‐CoV‐2 nucleocapsid (N) primary antibody (1:2000 dilution in PBS with 5% goat serum) [Sino Biological, #40143‐T62] or anti‐SARS‐CoV Spike S1 primary antibody (1:2000 dilution in PBS with 5% goat serum) [Sino Biological, #40150‐MM02] at 37 °C for 1.5 h. Alexa 488‐conjugated goat anti‐rabbit IgG secondary antibody (1:2000 dilution in PBS with 5% goat serum) [Molecular Probes, #MOP‐A‐11034] and Hoechst 33342 (1:10 000 dilution) [Molecular Probes] were co‐incubated at 37 °C for 1 h. Images were acquired using the Operetta CLS high content analysis system [PerkinElmer]. The acquired images were analyzed using the Columbus image analysis system provided by the same manufacturer. Infectivity and cell viability were measured using the Columbus software, and the results were normalized to positive (mock) and negative (0.5% DMSO) controls in each assay plate. DRCs were generated using the Prism7 software [GraphPad] and fitted by variable slope dose‐response models, with the following equation: Y = Bottom + (Top ˗ Bottom)/(1 + 10^((LogIC₅₀‐X)*Hillslope)). IC₅₀ values were calculated from the normalized activity dataset‐fitted curves. All IC₅₀ and CC₅₀ values were measured in duplicate experiments, and the quality of each assay was controlled by a Z'‐factor higher than 0.5.

Calu‐3 cells were seeded at a density of 2.0 × 10⁴ cells per well in DMEM, supplemented with 2% FBS and 1× antibiotic‐antimycotic solution [Gibco/ThermoFisher] in black, 384‐well, µClear plates [Greiner Bio‐One], 24 h before the infection. Ten‐point, twofold serial dilutions were made from compounds to generate DRCs, with compound concentrations ranging from 0.2 to 100 µm. Serially diluted compounds were added to the cells ≈30 min before the infection. For viral infection, plates were transferred to the BSL‐3 containment facility, and the virus was inoculated at the following MOIs: 1 MOI for the SARS‐CoV‐2 ancestral strain and 2 MOI for the SARS‐CoV‐2 omicron (BA.1) variant. The MOIs for different coronaviruses were determined through independent experiments to achieve maximal infectivity and cell viability. The infected cells were fixed at 24 hpi and were further analyzed by immunofluorescence assays. Image analysis and the DRC generation method were identical to the method with Vero cells.

DRC Assay in SFTSV

Vero cells were seeded at a density of 1.0 × 10⁴ cells per well in DMEM, supplemented with 2% FBS and 1× antibiotic‐antimycotic solution [Gibco/ThermoFisher] in black, 384‐well, µClear plates [Greiner Bio‐One], 24 h before the infection. Ten‐point, twofold serial dilutions were prepared from the compounds to generate the DRC, with compound concentrations ranging from 0.2 to 100 µm. Serially diluted compounds were added to the cells ≈30 min before the infection. For viral infection, plates were transferred into the BSL‐3 containment facility, and SFTSV was inoculated at 1 MOI. The infected cells were fixed at 24 hpi, and immunofluorescence assays were further analyzed on the fixed cells. Cells were fixed with 4% PFA for 30 min at r.t. and permeabilized with 0.25% Triton X‐100 for 10 min. After blocking with 5% goat serum, cells were incubated with anti‐SFTSV nucleoprotein primary antibody (1:1000 dilution in PBS with 5% goat serum) [Native Antigen, #MAB12414‐100] at 37 °C for 1.5 h. Alexa 488‐conjugated goat anti‐mouse IgG1 secondary antibody (1:1000 dilution in PBS with 5% goat serum) [Invitrogen, #A‐21121] and Hoechst 33342 (1:10 000 dilution) [Molecular Probes] were co‐incubated at 37 °C for 1 h. Images were acquired using the Operetta imaging system. Image analysis and the DRC generation method were identical to the method for coronaviruses.

DRC Assay in DENV‐2

Huh‐7 cells were seeded at 5.0 × 10³ cells per well in DMEM, supplemented with 2% FBS and 1× antibiotic‐antimycotic solution [Gibco/Thermo Fisher Scientific] in black, 384‐well, µClear plates [Greiner Bio‐One, Kremsmünster] at 24 hpi. Ten‐point, twofold serial dilutions were prepared from the compounds to generate the DRC. The concentration ranges were 0.2–100 µm for testing compounds. Serially diluted compounds were added to the cells ≈30 min prior to infection. For viral infection, DENV‐2 was infected at 0.08 MOI. The infected cells were fixed at 48 hpi, and immunofluorescence assays were further analyzed on the fixed cells. Image analysis and the DRC generation method were identical to the method for coronaviruses.

Co‐Immunofluorescence

Vero cells were seeded at a density of 3.5 × 10⁴ cells per well in 96‐well µClear plates [Greiner Bio‐One] at 37 °C for 16 h. 30 µL of compounds were added to cells in each well, and the cells were infected with SARS‐CoV‐2 at 5 MOI. The infected cells were fixed at 5 hpi with 4% PFA [Electron Microscopy Sciences, 32% aqueous solution] and permeabilized with 0.25% Triton X‐100 for 10 min. After blocking with 5% goat serum, cells were co‐incubated at 37 °C for 1 h with the following primary antibodies diluted in 5% goat serum: rabbit anti‐SARS‐CoV‐2 N protein antibody (1:1000) [Sino Biological, #40143‐T62] and mouse anti‐G3BP1 antibody (1:200) [Santa Cruz, #sc‐81940]. After washing, cells were co‐incubated at 37 °C for 1 h with Alexa Fluor 488‐conjugated goat anti‐rabbit IgG (1:2000) [Invitrogen, #A‐11034], Alexa Fluor Plus 555‐conjugated goat anti‐mouse IgG (1:500) [Invitrogen, #A32727], and Hoechst 33342 (1:8000) [Invitrogen, #H‐3570], all diluted in 5% goat serum. Images were acquired using the Operetta CLS high content analysis system [PerkinElmer].

Drug Combination Assay

Vero cells were seeded at a density of 1.0 × 10⁴ cells per well in DMEM, supplemented with 2% FBS and 1× antibiotic‐antimycotic solution [Gibco/ThermoFisher] in black, 384‐well, µClear plates [Greiner Bio‐One], 24 h prior to the experiment. For the combination study, an 8‐point DRC of SB2935 or SB2960 and remdesivir [MedChemExpress] was combined in an 8×8 matrix format. The concentration ranges were 0–30 µm for SB2935 or SB2960 and 0–50 µm for remdesivir. After compound treatment, the cells were infected with SARS‐CoV‐2 ancestral or omicron (BA.1) variant with the following MOIs: 0.03 MOI for SARS‐CoV‐2 ancestral, 0.1 MOI for SARS‐CoV‐2 omicron variant. The infected cells were fixed at 24 hpi, except for the omicron variant, which was fixed at 48 hpi. Immunofluorescence assays further analyzed fixed cells. Images were acquired using the Operetta CLS high content analysis system [PerkinElmer]. The acquired images were analyzed using the Columbus image analysis system provided by the same manufacturer. Infectivity and cell viability were measured using the Columbus software, and the results were normalized to positive (mock) and negative (0.5% DMSO) controls in each assay plate. All combination experiments were conducted in quadruplicate. The average data was presented and used to calculate synergy with SynergyFinder+.^{[ 45 ]}

Viral Load Quantification

Vero cells were seeded in a 96‐well plate [Thermo Fisher Scientific] with DMEM, supplemented with 2% FBS and 1× antibiotic‐antimycotic solution [Gibco/Thermo Fisher Scientific], 24 h prior to the infection. SB2960 and remdesivir were prepared as three‐point, fourfold serial dilutions, with top concentrations of 12.5 and 25 µm, respectively. These top concentrations were determined based on their IC₉₀ values. The diluted compounds were added to the cells, and ancestral SARS‐CoV‐2 was inoculated at 0.02 MOI. Supernatants were harvested at 24 hpi. All experiments were performed in triplicate. Viral titers of cell supernatants were determined by 50% tissue culture infectious dose (TCID₅₀) assay, and intracellular viral RNA copy numbers were quantified by real‐time reverse transcription PCR (qRT‐PCR) as previously described.^{[ 46 ]} For the TCID₅₀ assay, Vero cells were plated in clear 96‐well plates [Thermo Fisher Scientific] ≈24 h prior to infection and cultured in Opti‐PRO SFM [Thermo Fisher Scientific] supplemented with 4 mM l‐glutamine and 1× antibiotic‐antimycotic solution at 37 °C with 5% CO₂. For infection, supernatants were serially diluted 10‐fold in SFM and applied to the cell monolayers. The plates were then incubated under the same conditions for 96 h, until cytopathic effects (CPE) were observed; CPE were scored for each well. All samples were tested in quadruplicate, and the median TCID₅₀ was calculated using Spearman's & Kärber's algorithm.^{[ 47 ]} For qRT‐PCR, viral RNA was isolated from Vero cells using a MagMAX mirVana Total RNA Isolation Kit [Applied Biosystems]. Complementary DNA (cDNA) was synthesized from the extracted RNA using the SuperScript IV Reverse Transcriptase kit [Invitrogen], following the manufacturer's instructions. qRT‐PCR was carried out with TaqMan Universal PCR Master Mix [Applied Biosystems] on a ViiA 7 Real‐Time PCR System [Applied Biosystems]. The primer and probe sequences were obtained from the Centers for Disease Control and Prevention (CDC), and the 2019‐nCoV_N2 primer set was employed for this experiment.

The sequences of the primers were as follows:

2019‐nCoV_N2_F, TTACAAACATTGGCCGCAAA;

2019‐nCoV_N2_R, GCGCGACATTCCGAAGAA;

2019‐nCoV_N2_P, ACAATTTGCCCCCAGCGCTTCAG.

A standard curve was established using cDNA templates containing the identical sequence applied in the real‐time PCR. Viral RNA copy numbers were then quantified based on this standard curve, and each reaction was performed in triplicate.

Molecular Modeling and Docking Analysis

Molecular structures were optimized using semi‐empirical PM6 calculation in Gaussian 09 W. Computational binding analyses of SB2960, SB2935, SB2963, and SB2910 with RACK1 were performed using Discovery Studio Client v16.1.0.15350 [BIOVIA]. The X‐ray crystal structure of human RACK1 (PDB: 4AOW) was prepared by the Prepare protein module. The ligand binding site was defined based on the receptor cavities of the prepared protein. Molecular docking was performed using the CDOCKER module, and the CDOCKER energy value was analyzed to determine the optimal binding mode. The 2D Interaction module was used to visualize potential between the compounds and RACK1.

Surface Plasmon Resonance (SPR) Assay

The binding affinity between the human RACK1 protein [Sino Biological, 12498‐H110E] and compounds was determined by SPR using a Biacore 1K+ instrument [Cytiva] at the Danaher‐SNU Discovery Center, Seoul National University. For the immobilization process, a CM5 sensor chip [Cytiva, BR100530] was activated by injecting a combination of 1‐ethyl‐3‐(3‐dimethylaminopropyl)‐carbodiimide (EDC) and N‐hydroxysuccinimide (NHS) onto flow cells 3 and 4. RACK1 was immobilized on the flow cell 4 in 10 mM sodium acetate buffer (pH 5.0), followed by quenching of unreacted NHS esters on both flow cells with 1 m ethanolamine‐HCl (pH 8.0). A total of 11 692 RU of RACK1 was immobilized in PBS buffer containing 0.005% (v/v) surfactant P20. After immobilization, compounds were injected for 120 s at a flow rate of 30 µL·min⁻¹ in various concentrations ranging from 2.5 to 15 µm in PBS running buffer (pH 7.4, 0.005% P20, 5% DMSO). Dissociation was monitored for 300 s under the same flow rate, and all binding measurements were conducted at 25 °C. Data were analyzed using the Biacore Insight Evaluation Software [Cytiva]. Sensorgrams were fitted to a 1:1 binding model to calculate the dissociation constant (K_D).

Western Blot

Vero or G3BP1‐GFP‐expressing U‐2 OS cells were seeded and incubated overnight to allow complete adhesion. Cells were treated under the indicated conditions and incubated in a humidified 5% CO₂ atmosphere at 37 °C. After treatment, cells were rinsed with PBS and lysed with RIPA buffer (50 mM Tris‐HCl, pH 7.5, 150 mM NaCl, 1% (w/v) sodium deoxycholate, 2 mM Na₃VO₄, 5 mM NaF, protease inhibitor cocktail, 1% (v/v) IGEPAL CA‐630) to extract proteins. Lysates were clarified by centrifugation at 20000 g for 20 min at 4 °C. Protein concentrations in the supernatants were determined using the Pierce BCA Protein Assay Kit [ThermoFisher, #23225] with BSA as a standard. Samples were mixed with 5× SDS‐PAGE loading buffer [Biosesang, SF2002‐110‐00], denatured at 95 °C for 5 min, and subjected to SDS‐PAGE. The separated proteins were transferred to PVDF membranes [Bio‐Rad, #1620177] and blocked with a 2% BSA in TBST buffer (Tris‐Buffered Saline with 0.01% Tween‐20) for 1 h at r.t. Membranes were incubated overnight at 4 °C with specific primary antibodies diluted in TBST containing 1% BSA. After three washes with TBST buffer, membranes were incubated for 1 h at r.t. with HRP‐conjugated secondary antibodies diluted in TBST containing 1% BSA. Following four additional washes with TBST buffer, protein bands were detected using EZ‐Western Lumi Pico Alpha kit [DoGenBio, DG‐WPAL250]. Chemiluminescent signals were visualized with the ChemiDoc MP imaging system [Bio‐Rad]. The following antibodies were used in this study: α‐tubulin [CST, #3873], RACK1 [CST, #5432], anti‐mouse IgG, HRP‐linked [CST, #7076], and anti‐rabbit IgG, HRP‐linked [CST, #7074].

4.1. 2D Differential Gel Electrophoresis (2D‐DIGE)

Vero cells were cultured to ≈80% confluency in 10‐cm culture dishes and incubated with complete medium containing either 10 µm SB2991 or 0.5% DMSO as a vehicle control for 12 h in a humidified 5% CO₂ atmosphere at 37 °C. After treatment, cells were rinsed with PBS, lysed in RIPA buffer containing 1× PIC [Roche, #11873580001], and the concentration was measured by Pierce BCA Protein Assay Kits [ThermoFisher, #23225]. Next, acetone was added to 50 µg of protein and incubated at −20 °C for 20 min. The mixture was centrifuged at 20 000 g for 7 min at 4 °C. The supernatant was discarded, and the pellet was washed twice with cold acetone. Then, the pellet was labeled with either 40 µm of Cy3‐NHS or Cy5‐NHS, along with labeling buffer (30 mM Tris‐HCl (pH 8.6), 2 m thiourea, 7 m urea, 4% CHAPS (w/v)), at 4 °C for 45 min. The dye‐conjugated protein was precipitated using cold acetone at −20 °C for 20 min, followed by centrifugation at 20000 g, 4 °C for 7 min. After four acetone washes, the pellets were re‐suspended with 50 µL of rehydration buffer (7 m urea, 2 m thiourea, 2% (w/v) CHAPS, 40 mM DTT [Millipore, #1.11474.0025], and 1% pH 3–10 IPG buffer [Cytiva, #17‐6000‐87]).

An equal amount of Vehicle and SB2991‐treated samples were mixed and loaded on a pH 3–10, 24‐cm Immobiline Drystrip gel [Cytiva, #17‐6002‐44]. The primary dimension was separated using an isoelectric focusing system [Cytiva, Ettan IPGphor 3], and the secondary dimension was separated by a polyacrylamide gel electrophoresis (PAGE) system [Cytiva; Ettan DALTsix]. The resulting 2D gel was scanned with a fluorescent gel scanner [Azure Biosystems; Sapphire].

In‐Gel Digestion and Mass Spectrometry

The candidate protein spots were excised from silver‐stained gels, de‐stained, and digested using trypsin.^{[ 48 ]} The resulting peptides were analyzed by peptide sequencing using nanoAcquity UPLC‐ESI‐Q‐TOF mass spectrometry [Waters, SYNAPT G2‐Si HDMS]. Peptides were eluted with a linear gradient of 5–40% buffer B (ACN/formic acid, 100:0.1 (v/v)) with buffer A (water/formic acid, 100:0.1 (v/v)) over 80 min. Each MS scan cycle was composed of one MS scan followed by MS/MS scans of the 10 most abundant ions in each MS scan. The resulting MS data were processed with Protein Lynx Global Server (PLGS) 2.3 data processing software [Waters]. The peaklists were converted to pkl files and searched using the global search engine Mascot (version 2.2.0) with the protein sequence database SwissProt (version 51.6, 257 964 entries), regarding primates as the taxonomy filter. The maximum number of 1 missed cleavage was permitted, and no fixed modifications were included. The following variable modifications were considered for protein spot identification: carbamidomethylation of Cys, oxidation of Met, phosphorylation of Ser or Thr, acetylation and formylation of Lys, N‐terminal pyroglutamylation of Gln and Glu, and acrylamide adduct propionamide of Cys.

CETSA

Vero or G3BP1‐GFP‐expressing U‐2 OS cells were cultured and harvested in two 10‐cm culture dishes to ≈80% confluency. Cells were treated with 40 µm SB2960 or 0.5% DMSO as a vehicle control for 2 h in a humidified 5% CO₂ atmosphere at 37 °C. After treatment, cells were washed with PBS and divided so each sample contained 10% of the harvested cells from one 10‐cm dish. Samples were heated at the designated temperature for 3 min, washed with PBS, resuspended in PBS containing 0.4% IGEPAL CA‐630, and lysed by three freeze‐and‐thaw cycles using liquid nitrogen. Soluble fractions were collected by centrifugation at 20000 g for 20 min at 4 °C. The resulting supernatants were mixed with 5× SDS‐PAGE loading buffer [Biosesang, SF2002‐110‐00], heated at 95 °C for 5 min, and subjected to western blot analysis.

Knockdown

Vero cells were seeded at a density of 2 × 10³ cells per well in 96‐well black‐sided, clear‐bottom plates. After complete adhesion, cells were transfected with either RACK1 siRNA [Bioneer, SDO‐1001, siRNA No 10399‐1], G3BP1 siRNA [Bioneer, sc‐1021], or negative siRNA [Bioneer, SN‐1003] using Lipofectamine RNAiMAX reagent [Invitrogen, #13778075] according to the manufacturer's protocol, and incubated for 72 h in a 5% CO₂ atmosphere at 37 °C.

Quantification of Stress Granule Count and Area of RACK1 Knockdown

Vero cells were seeded at a density of 4 × 10³ cells per well in an 8‐well chamber slide [ThermoFisher, #155409]. Cells were incubated for 16 h and transfected with either Negative or siRACK1 siRNA using Lipofectamine RNAiMAX reagent according to the manufacturer's protocol, and incubated for 72 h in a humidified 5% CO₂ atmosphere at 37 °C. Cells were then incubated with poly(I:C) (HMW) for 5 h followed by 20 µm SB2960 treatment. Stress granule imaging by immunofluorescence was performed according to the method described above using the DeltaVision Elite system [Cytiva].

Quantification of Stress Granule Count of G3BP1 Knockdown

Vero cells were seeded at a density of 2.0 × 10³ cells per well in 96‐well black‐sided, clear‐bottom plates. Cells were then incubated with poly(I:C) (HMW) for 5 h followed by 0.16–80 µm SB2960 or DMSO treatment using 1:2 serial dilution. Immunofluorescence was performed according to the method described above using anti‐TIA1 antibody (1:100 dilution) [Santa Cruz, sc‐166247]. Stress granules were imaged with INCell Analyzer 2500.

Stress Granule Monitoring After SB2960 Washout

G3BP1‐GFP‐expressing U‐2 OS cells were seeded at a density of 1 × 10⁴ cells per well in 96‐well black plates. After complete adhesion, cells were treated with poly(I:C) (HMW) for 5 h, washed with DMEM, and stained with Hoechst 33342 for 30 min. Media was exchanged with DMEM containing 20 µm SB2960 or DMSO at the corresponding concentration, and the cells were immediately imaged using an INCell Analyzer 2500 (0 h after compound treatment). After 5 h, cells were imaged again (5 h after compound treatment) and washed three times with DMEM. 19 h after washout, cells were stained again with Hoechst 33342 for 30 min, and stress granules were imaged using the INCell Analyzer (24 h after compound treatment, 19 h after washout).

Drug Treatment and CAP‐Seq Labeling in Living Cells

U‐2 OS cells stably expressing G3BP1‐miniSOG and untargeted miniSOG were cultured to 90% confluency in 10‐cm culture dishes and incubated with complete medium containing 1.25 µg mL⁻¹ poly(I:C) for 5 h at 37 °C with 5% CO₂, followed by 2 h 40 min incubation in complete medium containing either 20 µm SB2960 or 0.5% DMSO as a vehicle control. Cells were then rinsed with 1× HBSS, and the medium was replaced with HBSS containing 10 mM propargyl amine (PA) [Accela, #SY002930] with 20 µm SB2960 or 0.5% DMSO. After 5 min of incubation at 37 °C, cells were illuminated with blue LED (emission peak 465‐475 nm, 24 mW cm⁻²) for 15 min at r.t.

Following CAP‐seq labeling, cells were washed twice with PBS before being lysed with TRIzol Reagent [Invitrogen, #15596018]. RNAs were extracted following the manufacturer's instructions, digested with 1 µL DNase I [NEB, #M0303] at 37 °C for 30 min, and then incubated with click reagents containing 0.1 mM biotin‐azide [Click Chemistry Tools, #AZ104], 2 mM THPTA [Click Chemistry Tools, #1010], 0.5 mM CuSO₄ [Aladdin, #C112401] and 5 mM sodium ascorbate for 10 min on shaker at r.t. for CuAAC reaction. RNAs were then purified with RNA Clean & Concentrator kit [Zymo Research, #R1019] and eluted with pre‐warmed nuclease‐free water. The RNA integrity was detected by the Fragment Analyzer [Agilent]. ≈1 µg of RNA was reserved as pre‐enrichment samples, while the remaining RNA was purified using streptavidin beads.

Affinity‐Based Purification of RNA

20 µL of Dynabeads MyOne Streptavidin C1 [Life Technologies, #65002] were washed three times with 100 µL B&W buffer (5 mM Tris pH 7.5, 1 m NaCl, 0.5 mM EDTA, 0.1% (v/v) Tween‐20), twice with solution A (0.1 m NaOH, 0.05 m NaCl in nuclease‐free water), once with solution B (0.1 m NaCl in nuclease‐free water), and re‐suspended in 100 µL blocking buffer (1 mg mL⁻¹ BSA, 1 mg mL⁻¹ Yeast‐tRNA in B&W buffer) on a shaker (1200 rpm) for 2–3 h at 25 °C. Thereafter, pre‐blocked beads were washed three times with 100 µL B&W buffer. Extracted RNAs were mixed with an equal volume of 2× B&W buffer (10 mM Tris pH 7.5, 2 m NaCl, 1 mM EDTA, 0.2% (v/v) Tween‐20) and pre‐blocked beads on a shaker (1200 rpm) for 1 h at 25 °C to allow biotinylated RNAs to bind. The supernatant was discarded, and the beads were washed three times with 100 µL B&W buffer, twice with 100 µL Urea buffer (4 m Urea, 0.1% (w/v) SDS in PBS), and twice with 100 µL PBS. The beads were finally re‐suspended in 50 µL Elution buffer (95% formamide, 10 mM EDTA, 1.5 mM d ‐biotin), heated at 50 °C for 5 min, and then at 90 °C for 5 min. The supernatant containing eluted biotinylated RNAs was transferred to a 1.5 mL Eppendorf tube with 1 mL TRIzol reagent to extract RNA according to manufacturer's instructions. The enriched RNAs were dissolved into 10 µL of nuclease‐free water.

Library Construction

100 ng pre‐enrichment and 5 µL post‐enrichment (labeled or omitting probe) RNAs were used for cDNA library construction with NEBNext Ultra II RNA Library Prep Kit [NEB, #E7770] for Illumina according to the manufacturer's instructions. Total RNAs were fragmented to ≈300 nt for library construction. After construction, two rounds of size selection were performed with 0.6–0.7× and 0.3× DNA Clean Beads according to the instructions of VAHTS DNA Clean Beads [Vazyme, #N411‐02]. The final cDNA libraries of 150 bp paired‐end were sequenced with ≈40 m reads on the Illumina NovaSeq 6000 platform.

Sequencing Data Analysis

The adaptors sequence in reads were removed by Cutadapt (v.1.18)^{[ 49 ]} and quality controlled by FastQC (v0.11.8). Then reads were mapped by hisat2 (v2.1.0)^{[ 50 ]} to the human genome assembly GRCh38 (hg38) with gene annotation (v.87) downloaded from the Ensembl website. The mapped reads were counted by htseq‐count (v0.7.2)^{[ 51 ]} with the option “–stranded no”.

A series of differential analyses was performed by the R package DESeq2 (v1.34.0)^{[ 52 ]} to define the SG datasets captured by CAP‐seq. The genes with log₂(Fold Change) > 0.3 and log₂(Fold Change) < −0.3, along with an adjusted p value (p_adj) < 0.05, were defined as enriched and depleted, respectively, in each DESeq2 analysis. The SG‐proximal or G3BP1‐proximal RNAs in the aforementioned drug‐treated cells were defined as the overlap between the enriched targets in three differential analyses: 1) post‐ versus pre‐enrichment of RNA labeled with G3BP1‐miniSOG; 2) post‐enriched RNA labeled with G3BP1‐miniSOG versus RNA from negative control omitting PA; and 3) post‐enriched RNA labeled with G3BP1‐miniSOG versus untargeted miniSOG under the same treatment. The SG/G3BP1‐excluded RNAs were defined as the overlap between the depleted targets in 1) and 3), as mentioned above. Gene Ontology (GO) enrichment analysis of SG‐proximal/excluded transcripts was performed using Ensembl release 114 and ShinyGO (v0.82).^{[ 53 ]} P‐value <0.05 was set as the threshold. Three biological replicates of U‐2 OS cells stably expressing G3BP1‐miniSOG and untargeted miniSOG of each treatment and labeling were used in the analysis. The analysis used two biological replicates of U‐2 OS cells stably expressing G3BP1‐miniSOG and untargeted miniSOG with drug treatment, omitting PA. The NaAs‐induced SG‐proximal/excluded transcriptome dataset was acquired from previous work.^{[ 39 ]}

GO Enrichment Analysis Using Metascape

To compare enriched GO terms between SB2960‐ and NaAs‐treated conditions, SG‐proximal or SG‐excluded transcript lists were analyzed using the Multiple Gene Lists option in Metascape^{[ 40 ]} with default settings. Heatmaps displaying the top 20 enriched clusters for each condition were generated.

Immunofluorescence of V5‐G3BP1‐miniSOG–Expressing U‐2 OS Cells

U‐2 OS cells stably expressing G3BP1‐miniSOG were cultured to 70–90% confluency on glass coverslips pre‐coated with Matrigel. The incubation of poly(I:C) and SB2960 was performed as described above. Cells were then rinsed three times with PBS, fixed with 4% (v/v) paraformaldehyde in PBS for 15 min at r.t., and permeabilized with 0.1% Triton X‐100 and 0.05% Tween‐20 in PBS for 15 min on ice. Thereafter, the cells were rinsed three times with PBS and blocked with 3% (w/v) BSA in PBS for 1 h. For subsequent immunostaining, cell samples were incubated with primary antibodies (mouse anti‐V5, Biodragon, 1:1000 dilution; rabbit anti‐G3BP2, Abcam, 1:1000 dilution) overnight at 4 °C. After PBS washing thrice (5 min each), cells were incubated with fluorescent secondary antibodies and DAPI (goat anti‐mouse‐Alexa Fluor 488, ThermoFisher, 1:1000 dilution; goat anti‐rabbit‐Alexa Fluor 647, ThermoFisher, 1:1000 dilution; DAPI, 1:1000 dilution) for 45 min at r.t. The cell samples were then washed thrice with PBS and were ready for imaging. Immunofluorescence images were collected with a ZEISS LSM 980 confocal microscope, controlled by a customized software written in Micromanager. A 40× oil immersion lens was used to acquire fluorescence images. All images shown were maximum intensity projections from Z‐stacks (ImageJ).

Liver Microsomal Stability

Liver microsomal stability of compounds was examined in both mouse and human systems. A 500 µm spiking solution of SB2960 and a reference were prepared from a 10‐mM stock solution. 1.5 µm spiking solution in microsome (0.75 mg mL⁻¹) was made with 500 µm spiking solution, 20 mg mL⁻¹ liver microsome, and PBS. 6 mM NADPH stock solution diluted in PBS was also prepared. Then, 30 µL of 1.5 µm spiking solution was dispensed to the assay plates designated for each time point (0, 5, 15, 30, and 45 min). The assay plates were pre‐incubated at 37 °C for 5 min. Then, 15 µL of NADPH stock solution was added to the wells to start the reaction and timing. 150 µL of ACN:methanol (1:1) containing internal standard was added to each sample at the designated time (0, 5, 15, 30, and 45 min) to quench the reaction. After shaking the assay plates for 10 min at 600 rpm, the samples were centrifuged for 15 min at 6000 rpm. 80 µL of the resulting supernatant from each well was transferred into a 96‐well plate containing 140 µL of pure water for LC‐MS/MS (LCMS‐8050) [Shimadzu] using positive electro spray ionization and a reverse‐phase column (ACQUITY UPLC HSS T3 1.8 µm, 50 mm × 2.10 mm) [Waters].

Animal Studies

The animals used in this study were all ICR male mice (6–8 weeks old, SPF grade). All mice were bred and maintained in an adjusted environment. The mice were fasted overnight prior to oral administration and resumed food supply 4 h after the post‐dose. All animal experiments were performed with the approval of the Institutional Animal Care and Use Committee at Medicilon (Approval No. 19205‐23031‐NG).

Intravenous and Oral Administration of SB2960

SB2960 was administered by intravenous (i.v.) injection (5 mg kg⁻¹) or oral (p.o.) administration (10 mg kg⁻¹) in ICR male mice (n = 3). SB2960 was prepared as a solution. For intravenous administration, SB2960 (5 mg kg⁻¹) was dissolved in DMSO/Tween‐80/PBS (5:5:90, v/v/v) at a concentration of 1 mg mL⁻¹ with a dose of 5 mg/5 mL kg⁻¹ to ICR male mice (n = 3). For oral administration, SB2960 (10 mg kg⁻¹) was dissolved in DMSO/Tween 80 /PBS (5:10:85, v/v/v) at a concentration of 1 mg mL⁻¹, and orally administered to ICR male mice (n = 3) at a dose of 10 mg/10 mL kg⁻¹ by using a gastric gavage tube. 30 µL of blood samples were collected via submandibular vein or other suitable vein in mice at 0.083, 0.5, 1, 2, 4, 8, and 24 h after the start of administration of the drug. Blood samples were centrifuged at 6800 g for 6 min, and then aliquots of plasma samples were stored at −80 °C until LC‐MS/MS‐27 (TQ6500+) [SCIEX] analysis of SB2960.

LC‐MS/MS Analysis of SB2960

For plasma preparation, 240 µL of methanol containing warfarin (internal standard: 100 ng mL⁻¹) was added to 12 µL of each biological sample. The mixture was vortexed for 1 min. Then, samples treated with tubes were centrifuged at 14000 rpm for 7 min, while samples treated with 96‐well plates were centrifuged at 4000 rpm for 10 min, and 200 µL of the supernatant was transferred to the 96‐well plates. An aliquot of 1 µL supernatant was injected for LC‐MS/MS analysis. The running time was 1.8 min for each sample.

The LC‐MS/MS‐27 system consisted of a SHIMADZU 6460 triple quadrupole mass spectrometer with an Agilent 1260 LC system [Agilent]. Control of the equipment and data acquisition was performed using MassHunter Workstation software (Version B. 04. 01) [Agilent]. Chromatographic separations were performed on a C₁₈ HPLC column (ACQUITY UPLC HSS T3, 1.8 µm, 2.1 mm i.d. × 50 mm) [Waters]. The mobile phase A and B consisted of 0.1% formic acid in distilled water and ACN, respectively, and were pumped at a flow rate of 0.6 mL min⁻¹ with the oven temperature of 40 °C under gradient conditions (B(%): 20 (0.01 min), 90 (0.5–1.1 min), 20 (1.11–1.4 min)). Autosampler was maintained at 4 °C. A carryover was removed by a needle wash solution of a mixture of isopropyl alcohol/ACN/methanol/0.1% formic acid in water at a ratio of 1:1:1:1 (v/v/v/v) for a strong wash or 50% methanol for a weak wash. The total running time was 4 min for each sample. The eluent was monitored using the triple quadrupole tandem mass spectrometer with positive electrospray ionization and multiple reaction monitoring modes. The precursor to product ion transitions for SB2960 and the IS were m/z 657.2 ([M+H]⁺) → 152.1 and m/z 309.1 ([M+H]⁺) → 163.1, respectively. The retention times for SB2960 and the IS were 0.98 and 0.88 min, respectively. The calibration ranges of SB2960 in plasma samples were 5–5000 ng mL⁻¹.

Statistical Analysis

All statistical analyses were conducted using Prism software [GraphPad]. Unless indicated otherwise, data were presented as the mean ± standard deviation. The corresponding figure legends state the number of biological replicates (n) and the type of used statistical tests. The statistical significance was reported in figures.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

W.G.B. and S.S. contributed equally to this work. W.G.B., S.S., and S.B.P. conceived and designed the study. W.G.B. performed biological experiments and contributed to data analysis. S.S. designed and synthesized the compounds and conducted structural characterization. J.L., M.K., and S.J. performed virological experiments under the supervision of S.K. S.Z. conducted CAP‐seq experiments under the supervision of P.Z. J.H.K. performed LC‐MS analyses. K.V. contributed to chemical synthesis. M.L., J.H.K., J.H.L., and D.Y. assisted with biological experiments. S.B.P. supervised the overall project, and S.K. supervised the virological work. W.G.B., S.S., and S.B.P. wrote the manuscript with input from all authors. All authors contributed to manuscript editing and approved the final version.

Supporting information

Supporting Information

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.