Introduction

Sepsis, a life-threatening condition resulting from dysregulated host responses to infection, remains a significant global health challenge with alarmingly high mortality rates1,2. Despite advances in medical care, effective therapeutic interventions for sepsis are limited, underscoring the urgent need for innovative treatment strategies. In recent years, the nucleotide-binding oligomerization domain-like receptor pyrin domain-containing protein 3 (NLRP3) inflammasome has been proven to be a central mediator of the inflammatory cascade in sepsis pathophysiology3. The NLRP3 inflammasome, a multiprotein complex predominantly expressed in immune cells like macrophages, plays a pivotal role in sensing microbial products and endogenous danger signals, culminating in the activation of pro-inflammatory cytokines such as interleukin-1β (IL-1β) and interleukin-18 (IL-18)4,5. Dysregulated NLRP3 activation is implicated in the systemic inflammation and tissue damage characteristic of sepsis6.

Recent advances in molecular biology have highlighted the significance of RNA editing, a post-transcriptional modification process, in modulating gene expression and protein function. Specifically, RNA editing of the NLRP3 transcript has emerged as a promising avenue for regulating inflammasome activity7. Of particular interest is the development of the CRISPR-based RNA editing system, CasRx, which allows precise and programmable manipulation of RNA sequences. Unlike CRISPR/Cas9 systems which can permanently cause DNA double-stranded breaks, CasRx is an RNA-targeting CRISPR system, and has shown high efficiency and specificity in editing mRNA8. Compared to RNA interference (RNAi) technology, one of the most popular tools for targeting RNA and inhibiting transcripts, CRISPR-CasRx typically shows higher efficacy and lower off-target activity when used in vivo. For non-congenital genetic diseases, CRISPR-CasRx for RNA targeting is also a suitable candidate, as it can avoid genotoxicity associated with altering DNA sequences. Despite these merits, several critical requirements remain for effective therapeutic RNA editing. Particularly, safe and efficient in vivo delivery of the CRISPR-CasRx system remains challenging. Recent studies have shown that chemical control using small molecules is an effective method to regulate Cas9 activity at both transcriptional and post-transcriptional levels, and this control can be regulated reversibly and repeatedly. For example, we have previously shown that Cas9 can be designed into a deliverable prodrug form that only becomes active in response to small molecules that are released from the delivery carrier under the elevated reactive oxygen species signals or polyamines in vivo9,10. Despite exhibiting low off-target editing in non-targeted organs, the system still suffers from low genome-editing efficiency in the target cells. Therefore, the development of strategies for efficient RNA editing with high cell-type specificity is critical.

In view of above challenges, we propose a therapeutic strategy for sepsis treatment by harnessing the chemogenetic activation of repeatable CasRx RNA editing of NLRP3 mRNA in macrophages through bioengineered lentiviruses. Specifically, calcified lentiviruses were engineered to deliver a transgene construct encoding destabilized CasRx editor (dsCasRx) driven by a macrophage-specific promoter. The expressed editor can be activated by the small molecule antibacterial drug, trimethoprim (TMP), for the precise editing of NLRP3 mRNA in macrophages during sepsis. First, a transgene construct encoding destabilized CasRx protein, which was fused by unstable dihydrofolate reductase (DHFR) domains, was constructed. Second, the construct is further installed with a macrophage-specific promoter, by which the expression of CasRx system was only initiated in the macrophages. After the transgene construct is packed into lentivirus, a biocompatible layer of calcium phosphate (Cap) is applied to the outer surface to camouflage the virus (Lenti@Caps), which may evade the systemic antibody neutralization and potential inflammation. Finally, to prepare the M1 macrophage-targeting biomineralized lentivirus (termed as MTBL), Lenti@Caps were functionalized with folic acid (FA), a ligand for folate receptor beta (FR-β), which is overexpressed on the membranes of M1 macrophages. TMP encapsulated by 1,2-distearoyl-sn-glycero-3-phosphoethanolamine (DSPE) were prepared to stabilize dsCasRx and mediate target and repeated RNA knockdown by means of chemogenetic activation. As shown in Fig. 1, after the i.p. administration, the biomineralized lentivirus can target the inflammatory M1 macrophages, where the repeated administration of TMP@DSPE complexes can stabilize dsCasRx enzyme to knockdown NLRP3 mRNA. Nevertheless, in the absence of TMP, structurally unstable dsCasRx is rapidly degraded via the ubiquitin-dependent proteasomal degradation route. Moreover, due to the role of macrophage-specific promoter, the gene expression is only initiated in the macrophages and merely occurred in other types of cells. Collectively, eliminating pathogens and control of inflammation through biomineralized and bioengineered lentivirus combined with TMP@DSPE mediated cell type-specific delivery represent a promising and effective strategy for sepsis treatment.

Fig. 1: Schematic illustration of process of preparation of the Lenti@Cap-FAs and targeted delivery and M1 macrophages-specific RNA editing mediated by Lenti@Cap-FAs.
Fig. 1: Schematic illustration of process of preparation of the Lenti@Cap-FAs and targeted delivery and M1 macrophages-specific RNA editing mediated by Lenti@Cap-FAs.The alternative text for this image may have been generated using AI.
Full size image

a A DNA construct encoding destabilized CasRx protein, which was fused by unstable dihydrofolate reductase (DHFR) domains, was constructed. The transgene construct is further installed with a macrophage-specific promoter. After the DNA construct is packed by lentivirus, a biocompatible layer of calcium phosphate (Cap) is applied to the outer surface to camouflage the virus. The Lenti@Caps was functionalized with folic acid (FA). TMP encapsulated by DSPE were prepared to stabilize dsCasRx and mediate target RNA knockdown by means of chemogenetic activation. b After the i.p. administration, the biomineralized lentivirus can target the inflammatory M1 macrophages, where the repeated administration of TMP@DSPE complexes can stabilize dsCasRx enzyme to knockdown NLRP3 mRNA. In the absence of TMP, structurally unstable dsCasRx is rapidly degraded via the ubiquitin-dependent proteasomal degradation route, and since other type cells do not express dsCasRx due to the macrophage-specific promoter, the system loses its ability to knockdown the target gene. Collectively, eliminating pathogens and control of inflammation through biomineralized and engineered lentivirus combined with TMP@DSPE mediated subset delivery represent an effective strategy for sepsis treatment. Created in BioRender. Yan, X. (2025) https://BioRender.com/feoajap.

Results

Characterization of biomineralized lentiviral nanoparticles

The in vivo application of lentiviral vectors is frequently hindered by host immune responses, which generate neutralizing antibodies and other clearance mechanisms that compromise the efficacy of repeated administrations. To evaluate whether pre-existing anti-lentiviral immunity affect subsequent transduction efficiency, we pre-injected the right hind limb of separate mouse groups with EGFP-encoding lentivirus (EGFP-lentivirus) at different time points—day −1, −2, −7, and −15—relative to the Luc-lentivirus injection on day 0 (Supplementary Fig. 2). We observed a significant reduction in luciferase expression upon rechallenge as the pre-exposure interval increased. Notably, a substantial decrease was evident following a 7-day pre-exposure, with an almost complete loss of luciferase signal in the 15-day pre-exposure group. These results imply that pre-existing immunity to lentiviruses inactivates the lentiviruses upon the repeated injections. To detect antibody-dependent clearance responses, we assessed the levels of specific IgG generated in response to lentivirus administration. The specific IgG titer obviously increased with time after multiple administration of lentiviruses (Supplementary Fig. 3), which explains why luciferase intensity decreased after the multiple injection. These results suggest that the repeated injection of lentiviruses cause antibody-dependent clearance in vivo. Next, by intravenous administration different doses of lentivirus C57BL/6 mice in vivo (5 µg, 0.9 × 108 TU lentivirus; 25 µg, 4.5 × 108 TU lentivirus), we observed that the high dose (25 µg) induced liver toxicity, as evidenced by elevated level of blood urea nitrogen (BUN), lactate dehydrogenase (LDH), acetylsalicylic acid (ASP) (Supplementary Fig. 4). Together, these results highlight that repeated administration and systemic delivery of unmodified lentiviruses can trigger strong immune responses and dose-dependent toxicity.

Biomimetic mineralization represents a promising strategy to reduce antibody-dependent clearance responses during systemic circulation11. We propose that lentivirus can be biomineralized using calcium ions with the phosphate radical to reduce its immune clearance. Therefore, lentiviruses were first suspended in Dulbecco’s modified Eagle medium (DMEM), after which CaCl₂ was added and the mixture was incubated at 37 °C. Acidic residues on the viral surface are capable of binding Ca²⁺ ions, establishing initial coordination sites that trigger and support the onset of mineral nucleation (Fig. 2a). Using transmission electron microscopy (TEM), we found that the biomineralized lentiviruses exhibited a core-shell structure, with a layer of Caps being coated onto the lentiviruses (Fig. 2b,c), forming Caps-coated lentivirus (Lenti@Caps). As revealed by DLS results, the average hydrodynamic diameter of lentivirus was 133 ± 12 nm, which increased to 226 ± 9 nm after coating with Caps (Fig. 2f). The ζ-potential of Lenti@Caps (−21 ± 4 mV) was lower than that of uncoated lentiviruses (−15 ± 2 mV). Energy-dispersive X-ray spectroscopy (EDX) characterization indicated that the composition of Lenti@Caps nanoparticles mainly contains a substantial amount of Ca elements, derived from calcium phosphate. Furthermore, other elements, including phosphorus (P), carbon (C), nitrogen (N) and sulfur (S), are derived from lentiviruses (Fig. 2e). The TGA analysis suggested the weight proportion of lentiviruses in Lenti@Caps nanoparticles was around 40%. Next, SDS-PAGE analysis of the protein composition of Lenti@Caps indicated that the major protein compositions were identical as those in lentiviruses (Fig. 2i), suggesting that the proteins of lentiviruses were well reserved after the biomineralization process. In the meantime, we found that vesicular stomatitis virus G (VSV G) protein, a typical type III viral fusion protein, was shielded after the coating of Ca3(PO4)2 layer (Supplementary Fig. 5). Both size and ζ-potential of Lenti@Cap remain stable within 7 days at room temperature in saline solution (Fig. 2j); however, at an acidic pH of 6.5, the dissolution of Ca3(PO4)2 layer was clearly observed as indicated by the decrease in size (Fig. 2i), with 80% of Ca2+ released within 24 hours (Fig. 2k). Taken together, these results suggested that the biomineralization resulted in a stable layer of Ca3(PO4)2 coating over lentiviruses at physiological pH, whereas this layer could quickly degrade to expose lentivirus under the mild acidic pH conditions, such as in the endosome.

Fig. 2: Characterizations of biomineralized lentivirus (Lenti@Caps).
Fig. 2: Characterizations of biomineralized lentivirus (Lenti@Caps).The alternative text for this image may have been generated using AI.
Full size image

a Scheme illustration of biomineralization process of Lenti@Caps. Created in BioRender. Yan, X. (2025) https://BioRender.com/feoajap. b Representative TEM images of lentivirus (scale bar: 200 nm). c Representative TEM images of Lenti@Caps (scale bar: 200 nm). d Representative SEM images of Lenti@Caps (scale bar: 200 nm). e Elemental mapping analysis of Lenti@Caps (scale bar: 200 nm). f Hydrodynamic sizes of Lentivirus and Lenti@Caps measured by DLS. g Zeta potentials of Lentivirus and Lenti@Caps measured by DLS. h TG-DSC analysis of Lenti@Caps. i SDS-PAGE resolves of the protein bands in lentivirus and Lenti@Caps (M, marker; Lenti, lentivirus; Cap, Lenti@Caps, Representative images). j Stability investigation of Lenti@Caps on size and zeta potential in NS buffer. k In vitro Ca2+ release of Lenti@Caps in different pH NS buffer. l Size changes of Lenti@Caps after 16 h incubation in different pH NS buffer. Data in (f–g) and (j–l) are presented as the mean ± SD (n = 3 biologically independent samples). Analysis performed once with a randomly chosen batch (b–d, i).

Next, we sought to evaluate whether the Caps shell affects the lentiviral transduction. Cell viability after treatment with Lenti@Caps remained comparable to that observed with uncoated lentiviruses (Supplementary Fig. 6), and luciferase expression mediated by viral transduction showed only a slight reduction 48 h post-transfection (Supplementary Fig. 7). A similar reduction in luciferase expression was observed in the abdominal cavity of C57BL/6 mice 48 h post-treatment in vivo (Supplementary Fig. 8). Meanwhile, we performed Tn5 LM-seq to analyze genome-wide integration events, which confirmed that Lenti@Cap retains lentiviral integration capability comparable to native lentivirus (Lenti) and none of the identified integration sites were located in exonic regions of oncogenes (ONCOGENEs) or tumor suppressor genes (TSGs) (Supplementary Fig. 9a). These results suggested that the biological properties of lentiviruses largely remain unchanged after the biomineralization. In addition, the biomineralization greatly enhances the storage stability of the lentivirus at room temperature. Whereas the transduction efficiency of Lenti@Caps slightly dropped over a period of 7 days, the efficiency of uncoated lentiviruses dropped significantly over the same period (Supplementary Fig. 10). More importantly, the biomineralized lentivirus (Lenti@Cap) showed significantly reduced adverse effects compared to native lentivirus, as evidenced by decreased inflammatory cytokine levels (IL-6, TNF-α) and lower liver enzyme markers (ALT, AST) in treated animals (Supplementary Fig. 9b).

Conditional activation of CRISPR-CasRx by biomineralized lentiviral nanoparticles in vitro

To achieve the specific expression of the editing tool in macrophages, we first assessed whether transcription of CasRx could be exclusively initiated in macrophages by using the human SP146-C1 promoter, a macrophage-specific promoter12 (Fig. 3a). As shown in Fig. 3b, lentiviral transduction with the DNA construct SP146-C1-GFP resulted in the stronger GFP expression in RAW264.7 cells (a macrophage cell line), whereas very weak GFP expression was observed in 293T cells (an embryonic kidney cell line). Quantitative analysis by flow cytometry suggested that there were approximately 40% GFP-positive cells in RAW264.7 group, which was much higher than that in 293T group (around 2%).

Fig. 3: Intracellular delivery of RNA editor mediated by Lenti@Caps in vitro.
Fig. 3: Intracellular delivery of RNA editor mediated by Lenti@Caps in vitro.The alternative text for this image may have been generated using AI.
Full size image

a Scheme illustration for the fabrication of M1 macrophage targeting nanoparticles. Created in BioRender. Yan, X. (2025) https://BioRender.com/feoajap. b Representative fluorescence images of 293T (left) and RAW264.7 cells (right) transducted with Lenti@Caps. Scale bar, 10 μm. c Quantitative analysis of EGFP-positive cells by flow cytometry (FCM). d UV/vis absorption spectra of Lenti@Cap-FAs. e Investigation of cellular uptake mechanism of Lenti@Cap-FAs in RAW264.7 by FCM. f  Representative images fluorescence images of RAW 264.7 cells and activated RAW 264.7 cells (by LPS) treated with DiD-labelled Lenti@Cap-FAs for 2 h. Scale bar, 20 μm. g Mean fluorescence intensity (MFI) of Lenti@Cap-FAs in RAW264.7 cells analyzed by FCM. h Endosomal escape of Lenti@Cap-FAs in RAW264.7 cells 2 or 6 h after the transfection, as evaluated by confocal laser scanning microscopy (CLSM). Scale bar, 20 μm. Analysis performed once with a randomly chosen batch. i The level of TNF-α, IL-1β and IL-6 after the exposure to lentivirus and Lenti@Cap-FA encoding CasRx and sgNLRP3. Data represent means ± SD; n = 3 biologically independent samples in (c, e, g, i). N.D. not detectable. Unpaired two-tailed Students’ t-test was used in c, one-way ANOVA with a Tukey post hoc test was used in (e, g, i), NS denotes no significant difference (P > 0.05).

The proinflammatory M1 macrophages produce proinflammation cytokines that sustain and aggravate sepsis inflammation, and highly express folate receptor-β (FR-β) over their membrane13,14. Thus, targeting FR-β represents a promising approach for delivering therapeutic agents specifically to inflammatory M1 macrophages15. To this end, we further functionalized the Lenti@Caps with FA by interacting Ca2+ with carboxyl groups of FA during calcium biomineralization as reported previously (Fig. 3a)11. The obatained nanoparticles, termed as Lenti@Cap-FA, exhibited a similar morphology and size to Lenti@Cap, and showed an absorbance peak at 280 nm, indicating that FA was successfully modified on the surface of Lenti@Cap nanoparticles (Fig. 3d). To investigate the targeting efficacy of Lenti@Cap-FA to M1 macrophages, DiD-labeled Lenti@Cap-FAs were used to observe the cellular uptake. The M1 macrophages treated with Lenti@Cap-FAs appeared obvious red fluorescence after 2 h, indicating substantial cellular uptake (Fig. 3f). In contrast, the M1 macrophages treated with Lenti@Caps showed relatively weaker red fluorescence (Fig. 3g). For the normal macrophages (without LPS stimulation), both the Lenti@Caps and Lenti@Cap-FAs treated groups exhibited similar weak red fluorescence in the cells (Supplementary Fig. 14). These observations were further confirmed by flow cytometry analysis (Fig. 3g). Furthermore, we observed that cellular uptake of Lenti@Cap-FA was markedly reduced at 4 °C, indicating that the process requires active, energy-dependent transport (Fig. 3e). Inhibiting clathrin assembly with chlorpromazine or disrupting lipid rafts using methyl-β-cyclodextrin substantially diminished internalization. In contrast, blocking caveolae-mediated trafficking with genistein caused only a minor reduction, suggesting that clathrin- and raft-associated pathways play predominant roles in Lenti@Cap-FA entry. Together, the cellular uptake of Lenti@Cap-FA by RAW264.7 cells was primarily mediated by energy-dependent pathway (Fig. 3e), followed by clathrin-dependent endocytosis pathway16. The endosomal escape dynamics of Lenti@Cap-FA nanoparticles within cells were further investigated (Fig. 3h). DiO-labelled Lenti@Cap-FA nanoparticles were exposed to RAW264.7 cells. After 2 h of exposure, the nanoparticles were primarily localized in the endo/lysosomes (red), indicating efficient cellular uptake. By 6 hours, substantial nanoparticles had successfully escaped from the lysosomal compartments, demonstrating effective endosomal disruption and cytoplasmic release—a key feature for subsequent functional delivery. To assess in vivo targeting, macrophages expressing EGFP were quantified to evaluate viral uptake and transduction efficiency. Our results demonstrated that the Lenti@Cap achieved comparable macrophage transduction efficiency to the native lentivirus (Lenti), confirming that the biomineralization process does not compromise the viral delivery capability in septic mice (Supplementary Fig. 15). Importantly, Lenti@Cap-FA exhibited significantly higher macrophage-targeting efficiency, attributable to the overexpression of folate receptor-β on inflammatory M1 macrophages17,18,19. These results suggested that Lenti@Cap-FAs had an excellent targeting effect on macrophages in vitro and in vivo.

The activation of NOD-like receptor thermal protein domain-associated protein 3 (NLRP3) inflammasome is closely related to the progression of inflammation20. Therefore, the knockdown or knockout of genes encoding NLRP3 would contribute to a reduction in the level of proinflammatory cytokines and alleviate inflammation21,22. We first evaluated whether the biomineralization process affects the functional capability of lentiviruses to mediate RNA editing in macrophages. Following sgRNA optimization (Supplementary Table 1 and Supplementary Fig. 16) and transduction, we confirmed that Lenti@Cap-FA effectively downregulated NLRP3 mRNA, leading to a significant reduction in pro-inflammatory cytokines (TNF-α, IL-1β, and IL-6) in LPS-stimulated macrophages (Fig. 3i). These results indicate that the biomineral coating well preserves bioactivity of lentiviral vectors in inducing therapeutic RNA editing. We next investigated whether TMP could activate destabilized CasRx (dsCasRx)—a fusion protein incorporating an unstable domain derived from E. coli dihydrofolate reductase—to enable chemogenetic control of NLRP3 mRNA editing in macrophages. Notably, no significant differences were observed in the transcriptional level of CasRx between the dsCasRx lentiviral transduction groups in the presence or absence of TMP (Fig. 4a). We further investigated the level of FLAG-tagged dsCasRx protein after the Lenti@Cap-FA-mediated transduction. Following the transduction, the protein level of dsCasRx increased in the presence of TMP (Fig. 4b, c), whereas the expression of dsCasRx protein was hardly observable in the absence of TMP. These results were further confirmed by the quantitative analysis of CasRx protein (Fig. 4c), where the band intensity of dsCas9 in TMP-treated group was approximately 80% relative to that in the wild-type group. Furthermore, the stabilization of dsCasRx by TMP resulted in a significant reduction in NLRP3 mRNA levels in LPS-stimulated RAW264.7 macrophages, as compared with the unstabilized dsCasRx in the absence of TMP (Fig. 4d). The reduction in NLRP3 mRNA levels also decreased the levels of NLPR3 protein (Fig. 4e) and pro-inflammatory cytokines, including TNF-α, IL-1β and IL-6 (Fig. 4f). Since dsCasRx is unstable in the absence of TMP, we next examined the underlying degradation mechanism. To confirm that the degradation of dsCasRx occurs through the ubiquitin-dependent proteasomal pathway, we employed the proteasome inhibitor MG132. As shown in Supplementary Fig. 17, both the Lenti@Cap-FA + MG132 group (with inhibitor) and the Lenti@Cap-FA + TMP group (with stabilization) exhibited significantly reduced NLRP3 mRNA expression, in contrast to groups without TMP stabilization or inhibitor and the untreated negative controls. In terms of safety, we incorporated shRNA-mediated NLRP3 knockdown (MTBL-shNLRP3) as a critical experimental control to systematically evaluate off-target effects. This approach enables a direct comparison at the RNA level under identical lentiviral delivery conditions, thereby excluding the impact of the delivery carrier. Our results demonstrate that although both CasRx and shRNA-mediated systems achieved comparable knockdown efficiency of NLRP3 mRNA, the CasRx-based editor exhibited markedly superior specificity. Global transcriptomic analysis revealed only 53 differentially expressed genes in the CasRx-treated group, compared to 1215 in the shRNA group, underscoring the enhanced precision and reduced off-target activity of our RNA editing strategy (Supplementary Fig. 18).

Fig. 4: Conditional activation of RNA editor mediated by Lenti@Caps in vitro and in vivo.
Fig. 4: Conditional activation of RNA editor mediated by Lenti@Caps in vitro and in vivo.The alternative text for this image may have been generated using AI.
Full size image

a CasRx mRNA expression of RAW264.7 cells after Lenti@Cap-FAs-mediated transfection. CasRx mRNA expression level was normalized to that of the control group. WT: CasRx protein was not‌ fused with unstable dihydrofolate reductase (DHFR) domains. b A schematic diagram of TMP-induced dsCasRx protein expression. Created in BioRender. Yan, X. (2025) https://BioRender.com/feoajap. c Western blot analysis and relative level of the expression of CasRx protein of RAW264.7 cells after transduction with/without stimulation by 10 μM TMP. d qPCR analysis of NLRP3 of RAW264.7 cells after Lenti@Caps transduction. e Western blot analysis of NLRP3 level in RAW264.7 cells after Lenti@Caps transduction. f The level of TNF-α, IL-1β and IL-6 of RAW264.7 cells after Lenti@Caps transduction. g Dose-responsive of NLRP3 gene downregulation in cells transfected with Lenti@Cap-FAs. h Cells were transfected and treated with 10 μM TMP to downregulation NLRP3. After 18 h of TMP treatment, cells were provided with fresh media containing or lacking TMP before harvesting and analysis by qPCR. i, (left) Representative luciferase expression of C57BL/6 mice after treated with TMP or lacked TMP before Lenti@Cap-FAs was injected (+: injected TMP for the first time; ++: injected TMP for the second time). (right) Luciferase expression level of C57BL/6 mice after treated with TMP (0.5 μg/g) at 48 h and 120 h or lacked TMP before Lenti@Cap-FAs was injected. Analysis performed once with a randomly chosen batch. Data represent means ± SD; n = 3 biologically independent samples in (a, d, f, g, h, i). N.D. not detectable. One-way ANOVA with a Tukey post hoc test was used in (a, d, f), NS denotes no significant difference (P > 0.05).

Dose-dependent and temporal control of nuclease activity in vitro and in vivo

The precise control of CasRx activity may be conducive to reducing off-target RNA editing23. Therefore, the dose-dependent effect for CasRx-mediated RNA editing was firstly investigated. As shown in Fig. 4g, the mRNA level of NLRP3 decreased with the increasing doses of TMP, suggesting the stabilization of dsCasRx following the NLRP3 disruption by Lenti@Cap-FA transduction. When the concentrations of TMP reached 104 nM, the level of NLRP3 mRNA decreased by almost 60% as compared with the control group. Although the removal of TMP in the cell culture after 48 hours resulted in a rapid increase in NLRP3 mRNA transcript levels, the reintroduction of TMP could rapidly reduce NLRP3 mRNA level by 60%, similar to that before the removal of TMP (Fig. 4h). Additionally, to evaluate the durability of TMP-responsive RNA editing in lentivirus-transduced macrophages, we conducted subculture experiments over multiple passages. The results confirm that the subcultured macrophages maintained efficient and inducible NLRP3 mRNA editing upon TMP stimulation, demonstrating the stability and sustained functionality of the chemogenetic editing system (Supplementary Fig. 19). Taken together, these results indicated that our system possesses tunable on-demand RNA editing in vitro. To validate whether the dose-dependent activation works in vivo, the destabilized luciferase-encoding Lenti@Cap-FAs were injected into C57BL/6 mice into abdominal cavity. Following the initial local injection of TMP (0.5 µg/g), we observed the strong bioluminescence intensity during the first 48 h (Fig. 4i). The intensity diminished significantly after 96 h due to the TMP excretion. Nevertheless, the bioluminescence intensity recovered after a second injection, indicating reactivation of luciferase protein in vivo.

Therapeutic effect of Lenti@Cap-FAs in the CLP model of sepsis

Finally, the therapeutic efficacy of Lenti@Cap-FAs was examined in vivo. To improve the in vivo uptake, TMP was encapsulated by DSPE which is often used to improve the water solubility of lipophilic small molecules to form TMP@DSPE micelles. The micelles displayed uniform spherical morphology with average diameters of 118 ± 15 nm (Supplementary Fig. 21), and the encapsulated TMP can be readily released from micelles (Supplementary Fig. 22). The sepsis model was established in C57BL/6 mice by cecum ligation and puncture (CLP) (Fig. 5a), which generates the similar pathogenesis and progression of human sepsis24. Following 24 h of i.p. administration, we found the distribution of indocyanine green (ICG)-labeled Lenti@Cap-FAs nanoparticles were primarily located in the cecum of the septic mice where FA receptors overexpressed M1 macrophages accumulate11. As compared with the administration of Lenti@Caps without FA targeting ligands, Lenti@Caps nanoparticles were also distributed in the liver in addition to cecum in the healthy mice following the same route of administration (Supplementary Fig. 23). The biomineralization of lentiviruses allowed for up to five repeated administrations, in sharp contrast with lentiviruses without biomineralization (Supplementary Figs. 3 and 24).

Fig. 5: Therapeutic effect of Lenti@Cap-FAs in the CLP model of sepsis.
Fig. 5: Therapeutic effect of Lenti@Cap-FAs in the CLP model of sepsis.The alternative text for this image may have been generated using AI.
Full size image

a Experimental procedures for the CLP model and treatment schedule. Created in BioRender. Yan, X. (2025) https://BioRender.com/feoajap. b Percentage survival of mice, c, clinical score, and (d) body weight of mice in different treatment groups monitored for 8 days. e, f Bacterial burden in peritoneal or blood of each surviving mouse treated by CLP. g, h mRNA level and protein level of each surviving mouse treated by CLP, Data represent means ± SD; n = 4 biologically independent samples in (g), n = 6 biologically independent samples in (c–f) or n = 10 biologically independent samples in (b). Kaplan-Meier method and the log-rank test were used in (b), one-way ANOVA with a Tukey post hoc test was used in (g). Boxplots indicate median (middle line), 25th, 75th percentile (box), and 5th and 95th percentiles (whiskers) as well as all data points.

To further investigate the therapeutic potential in septic mice, Lenti@Cap-FA nanoparticles were injected into the septic mice via i.p. administration, and the treatment schedule was shown in Fig. 5a. After the treatment, a significant improvement in mouse survival was observed. As shown in Fig. 5b, all septic mice treated with PBS died within 48 h, whereas the treatment by LCFT-SP146 (SP146-CasRx encoding Lenti@Cap-FA treatment combined with TMP@DSPE) extended the survival rate of septic up to 80% in 8 days. Treatment with TMP@DSPE and Lenti@Cap-FAs (driven by CMV promoter), termed LCFT-CMV, exhibited the moderate therapeutic efficacy against sepsis. TMP treatment alone showed lower efficacy than TMP@DSPE (Supplementary Fig. 25), and neither LCFT-NT (TMP+CasRx-NT-sgRNA) nor DSPE significantly improved survival. The clinical score for evaluating sepsis indicates that the septic mice recovered physical and mental states after LCF-TMP treatment (Fig. 5c). In the meantime, we observed the increase in the body weight of septic mice 2 days after treatment with LCFT-SP146, suggesting its substantial therapeutic efficacy (Fig. 5d). We also evaluated the infection as another important indicator of inflammation in sepsis therapy. TMP, known for its antibiotic activity against gram-positive bacterial infections, could also contribute to bacteria eradication in addition to its role in stabilizing dsCasRx. After the septic mice were sacrificed, the bacterial level in peritoneal exudates and blood were analyzed. As expected, the treatment with TMP@DSPE significantly decreased the total bacteria numbers in the peritoneal cavity and blood, and LCFT-SP146 treatment simultaneously eradicated the pathogens in the blood and alleviating the inflammation in peritoneal cavity (Fig. 5e, f). We further analyzed the disruption of NLRP3 mRNA of M1 macrophages isolated from peritoneal cavity by qPCR assay and found LCFT-SP146 treatment resulted in a rapid depletion of NLRP3 mRNA (Fig. 5g), suggesting LCFT-SP146 treatment can target M1 macrophages and exerts its RNA editing missions. Following LCFT-SP146 treatment, the decreased NLRP3 protein is associated with a reduction in NLRP3 mRNA levels (Fig. 5h and Supplementary Fig. 26).

Sepsis often leads to a sharp increase in leukocytes within the peritoneal cavity25. To evaluate the recruitment of leukocytes to the peritoneal cavity in the septic mice after the indicated treatments, the peritoneal lavage was collected after the septic mice were sacrificed. As expected, LCFT-SP146 treatment showed the lowest level of leukocytes recruitment to the peritoneal cavity among all treatment groups (Fig. 6a, b). Additionally, LCFT-SP146 treatment resulted in the decreased levels of NLRP3 protein, which corresponded with reduced NLRP3 mRNA levels. This was accompanied with the decreased levels of various pro-inflammatory cytokines, including TNF-α, IL-6, and IL-1β, in peritoneal cavity and blood (Fig. 6c-h). Moreover, M2-type macrophage populations (F4/80+/CD206+) in the peritoneal cavity markedly increased following LCFT-SP146 treatment (Fig. 6i), indicating a reduction in inflammation.

Fig. 6: Pathological indicators of in vivo treatment.
Fig. 6: Pathological indicators of in vivo treatment.The alternative text for this image may have been generated using AI.
Full size image

a Total peritoneal protein in peritoneal of each surviving mouse treated by CLP. b Total peritoneal leukocytes in peritoneal of each surviving mouse treated by CLP. c, e, g Proinflammatory cytokines TNF-α, IL-1β and IL-6 level in serum of each surviving mouse treated by CLP. d, f, h Proinflammatory cytokines TNF-α, IL-1β and IL-6 level in peritoneal of each surviving mouse treated by CLP. i M2 polarization ratio of macrophages in the peritoneal cavity. Data shown in this figure are representative of three independent experiments. j Representative H&E of lung of animals after the indicated treatment. The black arrowheads indicate the infiltration of immune cells. The scale bar is 100 μm. Data represent means ± SD; n = 4 biologically independent samples in (a–h). One-way ANOVA with a Tukey post hoc test was used in a-h, NS denotes no significant difference (P > 0.05).

To evaluate the therapeutic efficacy of LCFT-SP146 in preventing multiple organ failure, blood biochemical analysis and histopathological assessments were performed to determine its effect on sepsis. The serum index of liver and kidney injuries, alanine aminotransferase (ALT), blood urea nitrogen (BUN), aspartate aminotransferase (AST), and alkaline phosphatase (ALP), significantly elevated in the septic mice treated by PBS, whereas LCFT-SP146 effectively protected the organs from CLP-induced damage and reduced the ALT, BUN, AST, and ALP to normal level (Supplementary Fig. 27). Histopathological analysis indicated that leukocyte infiltration and tissue destruction were observed in multiple organs in PBS treated group, including the lung, kidney, heart, and liver, revealed typical multiple organ injury in septic mice (Fig. 6j and Supplementary Fig. 28). Although TMP@DSPE treatment provided slight protection against organ damage, LCFT-SP146 treatment remarkably reduced the inflammation. These results confirmed the protective effects of LCFT-SP146 in preventing organ failure. Secondary infections in sepsis often trigger renewed waves of inflammatory responses, posing a major challenge in clinical management due to their role in exacerbating organ dysfunction and worsening patient outcomes26,27. To evaluate the potential of our system in addressing this issue, we performed a secondary infection challenge in septic mice. Following initial CLP, approximately 80% of LCFT-SP146–treated mice survived and entered a stable recovery phase. To mimic clinical secondary infection, these survivors were subjected to an intraperitoneal challenge with a low dose of LPS (10 mg/kg). Subsequent administration of TMP@DSPE nanoparticles effectively reactivated the destabilized CasRx system, yielding a pronounced survival advantage—75% of mice survived the secondary challenge, in contrast to rapid mortality observed in PBS-treated controls (Supplementary Fig. 29). These results underscore the capability of our platform to facilitate sustained and on-demand CasRx activation, effectively mitigating secondary septic insults and highlighting its promise for the long-term clinical management of sepsis.

To better understand the potential mechanism of LCFT-SP146-mediated sepsis alleviation, we performed RNA sequencing (RNA-seq) of peritoneal lavage fluid after LCFT-SP146 treatment. As shown in Supplementary Fig. 30, several inflammation-related mRNA expression significantly decreased after the LCFT-SP146 treatment, such as NLRP3, IL1β, IL6, and TNFA. Furthermore, KEGG analysis of differential expression genes also identified several inflammation-related pathway as the most enriched pathways, such as NOD-like receptor signaling pathway, TNF signaling pathway and Toll-like receptor signaling pathway. With these data, we conclude that LCFT-SP146 can efficiently regulate the gene expression and pathways related to the inflammation, leading to the sepsis alleviation.

Discussion

In this study, we developed M1 macrophage-targeting, biomineralized lentivirus (MTBL) with reduced immunogenicity to address the current issues in repeated administration of viral delivery systems. Beyond the immunogenicity, the safety of gene therapy is strongly influenced by the precision of targeted delivery to specific cell types. Previous studies indicated that folate modification of the delivery system could enhance the endocytosis by M1 macrophages due to the overexpression of folate receptors13,14,15. Our design incorporates folate ligands into the biomineralized lentivirus through electrostatic interactions, enabling the active targeting of M1 macrophages in vivo. Moreover, the incorporation of a macrophage-specific promoter into the upstream of CasRx transgene constuct effectively minimizes unwanted expression in non-targeted cells. Thus, our approach well emphasizes dual macrophage-specific targeting through both “extracellular” targeted delivery and “intracellular” cell-specific gene expression, greatly ensuring the precision of macrophage-specific RNA editing. Overall, such a design is critical for precise macrophage-specific RNA editing in vivo.

Compared to genome-editing tools such as Cas9, CasRx selectively edits targeted RNA under the guidance of gRNA, avoiding the direct cleavage of the genome and mitigating potential genotoxicity. Previous studies demonstrate that CasRx-based RNA editing offers a more precise method for on-target transcript silencing, significantly reducing off-target transcriptional silencing commonly encountered by RNA interference (RNAi)8. Moreover, CasRx-mediated RNA editing achieves comparable knockdown efficiency as compared to RNAi. In our work, when combined with the destabilized CasRx and the stabilizer TMP, we successfully performed localized RNA editing with minimal off-target effects in other organs (Supplementary Fig. 31). Beyond RNAi, several emerging RNA editing platforms have also been reported, yet they present limitations that our system is designed to overcome. Compared to other emerging RNA editing technologies such as REPAIR (ADAR-based)28 and Cas7-11 systems29, our platform offers unique advantages in specificity, controllability, and translational potential. While REPAIR systems (based on Cas13b) can achieve A-to-I editing with minimal off-target effects, they lack the precise targeting capability of CasRx and require extensive engineering of guide RNAs for optimal efficiency. Cas7-11 is highly specific but exhibits poor editing efficiency in mammalian cells and poses delivery challenges owing to its large size. Our system addresses these limitations through three key innovations: (1) the combination of bioengineered lentivirus with biomineralization enables efficient and repeated in vivo delivery while reducing immunogenicity; (2) the chemogenetic control via TMP allows precise temporal regulation of editing activity, overcoming the constitutive activation issues seen in other systems; and (3) the macrophage-specific targeting strategy achieves cell-type precision unmatched by current RNA editing platforms. Furthermore, whereas most existing RNA editors focus solely on editing function, our integrated approach simultaneously provides antibacterial activity through TMP, addressing both inflammatory and infectious components of sepsis. This multifunctional capability represents a significant advance over conventional mono-therapeutic approaches. Together, these advantages establish a safe, efficient, and versatile framework for in vivo RNA editing. We demonstrate the therapeutic potential of this framework by specifically targeting NLRP3 mRNA in M1 macrophages, which effectively inhibits inflammasome assembly and alleviates inflammation in sepsis. MTBL-mediated NLRP3 mRNA editing decreases the expression of immune activation-related genes (such as IL-1β and TNF-α, Supplementary Fig. 30), while enhances the expression of immune suppression-related genes including IL-10 (Supplementary Fig. 30). As demonstrated in this study, these changes are particularly advantageous when therapeutic RNA editing is combined with other treatments (Fig. 5b, e), such as antibiotic therapies that often underperform in infectious diseases. We anticipate that targeting inflammatory proteins through RNA editing may enhance various therapeutic modalities, including anti-inflammatory and cell therapies, to improving the overall therapeutic efficacy in vivo. Furthermore, the primary treatment for severe inflammation relies on systemic antibody treatments30. Our MTBL-mediated RNA editing strategy offers a straightforward, universal solution to inhibit pro-inflammatory protein expression, avoiding the complex engineering and screening processes associated with antibody development. This approach can also be easily adapted for editing other RNA targets, such as TNFA and IL1B, by designing the appropriate gRNA. Beyond RNA editing platforms, therapeutic strategies targeting NLRP3 also include small-molecule inhibitors currently under clinical evaluation (e.g., MCC950, OLT1177, DFV890). While these compounds act on ATP hydrolysis or ASC speck formation to broadly suppress inflammasome activation, they often lack cell-type specificity and are limited by systemic toxicity and transient pharmacodynamics. In contrast, our MTBL system enables macrophage-specific NLRP3 downregulation at the mRNA level, providing spatial precision and sustained yet controllable editing activity through TMP induction. Moreover, unlike small molecule inhibitors that solely modulate inflammation, our platform concurrently delivers antibacterial function via TMP, addressing both inflammatory and infectious components of sepsis. This dual mechanism may provide a therapeutic benefit particularly relevant in polymicrobial sepsis models where anti-inflammatory interventions alone are often insufficient.

Despite these promising results, further efforts are crucial to overcome the current limitations of the MTBL system. First, the use of lentiviral vectors faces the issues of the integration into the host genome31. Although the integration of dsCasRx into macrophages by lentiviruses is beneficial for the temporal control of RNA activation by administering TMP@DSPE, such an integration may lead to the genotoxicity of macrophages. Additionally, the integration of genetic payload into the host genome would result in long-term expression of dsCasRx. Although the conditional activation of dsCasRx by TMP mitigates the continuous RNA editing, the down-regulation of NLRP3 mRNA is still observed in the absence of TMP (Fig. 4a). In this aspect, the replacement of lentiviral vectors with integrase-deficient lentiviruses (IDLVs) is expected to reduce the risk of genotoxicity. Recent studies, such as those demonstrating successful in vivo application of IDLVs32,33, support the feasibility of this approach. Our ongoing work is focused on combining such non-integrating vectors with our biomineralization strategy to enhance targeting and safety while preserving high transduction efficiency. Second, although we found that the modification with folate ligands enhances M1 macrophage targeting in our delivery system, the administration route is limited to intraperitoneal injection. This limitation arises from the fact that folate-modified nanoparticles in the bloodstream can easily absorb natural IgM, leading to off-target effects and rapid clearance34. Thus, a more rational design of targeting moieties is essential to improve macrophage targeting in vivo. Third, although biomineralization reduces the immunogenicity of lentiviral vectors, serum antibodies against the lentivirus were still detected after repeated administration (Supplementary Fig. 3). Looking forward, several promising research directions emerge from our findings. First, the modular nature of our system allows easy retargeting to other inflammatory mediators (e.g., TNF-α, IL-1β) by simple sgRNA replacement35, facilitating its potential application in a broader range of chronic inflammatory diseases beyond sepsis. Second, combining our RNA-editing platform with emerging delivery technologies, such as lipid nanoparticles or extracellular vesicles36,37, could enhance cell-specific targeting and clinical translatability. Third, integrating bioresponsive elements into viruses could create smarter systems that autonomously respond to inflammatory biomarkers in real time38,39, representing an innovative frontier in precision immunotherapies. Finally, exploring inhaled delivery formats could address acute lung inflammation and sepsis-associated acute respiratory distress syndrome, where targeted intervention is urgently needed40. These directions align with the growing emphasis on RNA-based therapeutics and could significantly advance the treatment of inflammatory disorders.

In summary, we developed a chemical-inducible RNA editing strategy to simultaneously disrupt NLRP3 mRNA and kill pathogens in overcoming the key challenges in sepsis therapy. The MTBL system could target and deliver dsCasRx into inflammation-related M1 macrophages to disrupt NLRP3 mRNA in response to TMP. This system not only plays an important role in activating dsCasRx to downregulate NLRP3 mRNA in the inflammatory M1 macrophages, but also serves as an antibiotic to eliminate bacteria in septic mice. The additional incorporation of a macrophage-specific promoter and dose-dependent activation enables the precise RNA editing in M1 macrophages, significantly avoiding the off-target editing at non-targeted RNA transcripts and in non-targeted tissues (Supplementary Fig. 23). Without the permanent alteration of DNA sequences, our current strategy is capable of eliminating bacteria and inhibiting inflammation simultaneously, offering a promising approach for sepsis treatment and potential application in other inflammatory disorders.

Methods

Mice and ethics

C57BL/6 (male, 6- or 8-week-old) were purchased from SLAC Laboratory Animal Co., Ltd. All animal experiments were performed according to the NIH guidelines for the care and use of experimental animals and were approved by the ethics committee of Zhejiang University (ZJU20241146). The animals were maintained in a controlled environment with a 12 h light/dark cycle. The ambient temperature was maintained at 22–24 °C, and the humidity level was kept at 40–60%.

Materials

All reagents used in the experiments were purchased from commercial sources and used without further purification. Dimethyl sulfoxide (DMSO), and TMP were purchased from Energy Chemical Co. Ltd. (Shanghai, China). HEK-293T and RAW 264.7 cell lines were kindly gifted by Prof. Haifang Wang. Dulbecco’s modified Eagle’s medium, and FBS were purchased from Sigma-Aldrich (USA). Lipo8000 transfection reagents and CCK-8 assay kit were obtained from Beyotime Co. Ltd. (Shanghai, China). Dialysis bag [molecular weight cutoff (MWCO): 1000 Da] was obtained from Yuanye Bio-Technology Co. Ltd. (Shanghai, China). Ultrapure water was generated using a Milli-Q system. The plasmid encoding CMV-3XFLAG-NLS-DHFR-CasRx-NLS-DHFR, SP146-C1-3XFLAG-NLS-CasRx-NLS-P2A-EGFP and SP146-C1-3XFLAG-NLS-DHFR-CasRx-NLS-DHFR were constructed in our laboratory. Target sgRNA (sgNLRP3) was designed by online tools (https://cas13design.nygenome.org). pMD2.G (plasmid #12259) and psPAX2 (plasmid #12260) were obtained from Addgene. Details of all plasmids and oligonucleotides are provided in Supplementary Data 1 and Supplementary Data 2. Primary antibodies used in this project were as follows: Anti-NLRP3 (1:1000; ER1706-72) antibody was obtained from HUABIO (Hangzhou, China). Anti-IL-1β (1:1000; ab200478) was obtained from Abcam (Shanghai, China). Anti-FLAG antibody was obtained from HuaAn Biotechnology (HUABIO, Hangzhou, China). Anti-GAPDH (glyceraldehyde-3-phosphate dehydrogenase) antibody was obtained from Abcam (Shanghai, China). All enzyme-linked immunosorbent assay (ELISA) kits used in this project were supplied by Multisciences (Hangzhou, China). The following antibodies were purchased from BioLegend (San Diego, CA, USA): Anti-mouse CD45-APC (cat. 157605), anti-CD45-APC/A750 (cat. 103153), anti-CD206-FITC (cat. 141703), anti-CD11b-PerCP (cat. 101229), anti-CD86-APC/Cyanine7 (cat. 159217), anti-F4/80-PE (cat. 157303), and anti-F4/80-PB450 (cat. 123123).

Preparation and characterizations of Lenti@Caps

For the preparation of Lenti@Caps, lentivirus (1 mg total protein) was dispersed in 1 mL of DMEM medium (Gibco) and equilibrated overnight with a vertical mixer at 4 °C. Subsequently, 10 µL of CaCl2 (5 mmol) was added to the mixture and incubated at 37 °C for an additional 2 h. After incubation, the biomineralized Lentivirus (Lenti@Caps) was washed 3 times with ultrapure (UP) water via centrifugation at 12,000 g for 10 min. The resulting pellet was resuspended in either normal saline solution (0.9% sodium chloride) or UP water for further characterization.

The nanoparticles suspended in UP water were evaluated for size and zeta potential using Dynamic Light Scattering (DLS) with a Nanoseries device (Malvern). The obtained data were analyzed with Zetasizer software. For Transmission Electron Microscopy (TEM) imaging, the nanoparticle suspension was dropped onto a carbon film supported by a copper grid and dried at room temperature. The structure of the nanoparticles was observed directly without any stain using an HT-7700 transmission electron microscope (Hitachi, Japan). For Scanning Electron Microscopy (SEM) imaging, the nanoparticle suspension was dropped onto aluminum foil and dried at room temperature. The samples were then coated with gold using a gold sputter in a high-vacuum evaporator, and the morphology of the nanoparticles was observed using a GeminiSEM 300 scanning electron microscope (Zeiss, Germany). For element mapping and Energy Dispersive Spectrometer (EDS) analysis, the sample preparation was the same as for SEM, and the data were obtained using an EDS attachment equipped with the SEM. For thermogravimetric and differential scanning calorimetry (TG-DSC) analysis, the nanoparticles were lyophilized and then measured using a TGA2 at a heating rate of 10 °C/min from 25 °C to 800 °C in a nitrogen atmosphere. For in vitro Ca²⁺ release assays, Lenti@Caps were dissolved in normal saline (NS) solution with different pH values (6.5 and 7.0). The supernatants were collected at different time points via centrifugation and analyzed using a calcium detection kit.

Tn5 LM-seq for lentiviral integration site analysis

Lentiviral integration sites were mapped using Tn5 LM-seq, a sensitive method combining Tn5 transposase fragmentation with unique molecular identifiers (UMIs) to enhance accuracy and reduce artifacts41. Genomic DNA was fragmented and ligated to UMI-containing adapters using Tn5 transposase (GeneRulor Co., Ltd). Viral-host junctions were enriched through a three-round PCR strategy: initial amplification with virus-specific primers targeting 3’ LTR regions and P5 adapters, nested PCR with LTR-specific primers containing P7 sites, and a final indexing PCR for library preparation. Libraries were quality-checked on an Agilent 2100 Bioanalyzer and sequenced on an MGI platform (150 bp PE). A custom bioinformatic pipeline developed and implemented by GeneRulor Co., Ltd was used to identify integration sites, filter artifacts, and annotate genomic locations, with UMI-based deduplication applied to eliminate PCR duplicates.

In Vitro NLRP3 gene disruption with Lenti@Cap-FA

To evaluate the knockdown efficiency of target RNA, samples were lysed and total RNAs were extracted by Trizol (TIANGEN Biotech Co., Ltd). Briefly, Raw264.7 (2 × 105 cells/well) were seeded in six-well plates for 24 h and then incubated with lentivirus/sgNLRP3. After incubation for 6 h, the medium was replaced with fresh medium. Then the cells were harvested by trypsinization after further incubation for 48 h. 1 µg extracted RNA was reverse-transcribed into cDNA and subjected to quantitative real time PCR (qRT-PCR) using a master-mix with SYBR-green (Yeasen Biotech Co., Ltd) and the QuantStudio 3 Real-Time PCR Systems (Thermo Fisher Scientific, USA). All primers for qRT-PCR were listed in Supplementary Table 2.

Conditional activation of Lenti@Cap by TMP in vitro

The small-molecule-mediated regulation of the Lenti@Cap gene editing systems was investigated by qPCR. Briefly, HEK293T cells (1.3 × 105 cells per well in a 24-well plate) were transiently transfected with Lenti@Cap (5.2 × 105 TU). The cells were incubated with or without appropriate DD-stabilizing small molecules (TMP: 0, 1, 10, 100, 1000, or 10,000 nM) for 72 h post-transfection at 37 °C. After cell harvesting, the total RNAs were extracted by Trizol. qRT-PCR was performed as described above.

RNA extraction, sequencing and bioinformatic analysis

Total RNA was extracted from tissue samples using Trizol reagent (TIANGEN Biotech), with quality and quantity assessed using a NanoDrop spectrophotometer and an Agilent 2100 Bioanalyzer. mRNA was enriched using oligo(dT) magnetic beads, fragmented, and reverse-transcribed into cDNA with random hexamer primers. After end repair, A-tailing, and adapter ligation, the libraries were purified with Ampure XP Beads, amplified by PCR, and quality-validated. DNA nanoballs generated via rolling circle amplification were sequenced on the DNBSEQ-T7 platform (MGI Tech) to produce 150 bp paired-end reads.

Raw sequencing reads were preprocessed with fastp (v0.20.1) to remove adapters and low-quality sequences. Clean reads were aligned to the reference genome using HISAT2 (v2.2.1). Transcript assembly and quantification were performed with StringTie (v2.0.4), and differential expression analysis was conducted using DESeq2 (v1.26.0). Functional enrichment analyses (GO and KEGG) were carried out to interpret the biological relevance of differentially expressed genes.

CLP model

C57BL/6 mice (male, 6- or 8-week-old) were purchased from SLAC Laboratory Animal Co., Ltd. Cecal ligation and puncture (CLP)-induced sepsis was established according to a previously described procedure24. Briefly, mice were anesthetized with isoflurane anesthesia, and the abdominal hair was shaved, followed by disinfection of the abdominal area with alcohol prep pads. Then a midline 1 cm incision was made, and the cecum was gently removed from the abdominal cavity, leaving the remainder of the small and large bowel within the peritoneal cavity. After ligation with 4-0 silk at the designated position for severe-grade sepsis, the cecum was punctured with a 21-gauge needle and the cecal contents were extruded through the perforation. The cecum was gently placed back into the peritoneal cavity and the incision was stitched and closed. Mice were divided into seven groups and subjected to the different treatments. Sham mice had only the abdominal laparotomy procedure without CLP. Untreated CLP group underwent the procedure as described above without further treatment. LCFT-CMV, TMP@DSPE, LCFT-SP146, treatment groups, the initial dose was administered at 6 hours after CLP. Subsequently, TMP@DSPE was injected every two days throughout the 7-day survival observation period. The survival rate, clinical scores, and body weight were monitored for 8 days (The dose of lentivirus administered to each mouse was 0.9 × 108 TU in LCFT-CMV treatment groups and LCFT-SP146 treatment groups). The criteria for clinical score were as follows: 0, no symptoms; 1, piloerection and huddling; 2, piloerection, diarrhea, and huddling; 3, lack of interest in surroundings and severe diarrhea; 4, decreased movement and listless appearance; and 5, loss of self-righting reflex. Mice were killed humanely when they exhibited a score of 5. At the experimental endpoint, surviving mice mice were sacrificed, the blood was collected for hematology analysis (Haoke Biotech Co., Ltd), and the colon tissues were photographed and collected to ELISA, qRT-PCR, Western blotting, histological.

Statistics & reproducibility

All data were analyzed by Graphpad prism 8.0. Biological replicates were used in all experiments unless stated otherwise. No data were excluded from the analyses. The statistical significance was analyzed using Students’ t-test and one-way analysis of variance (ANOVA). Unpaired two-tailed Students’ t-test was used for comparison of two groups. One-way ANOVA with a Bonferroni post-hoc test was used when both time and treatment were compared. The Kaplan-Meier method and the log-rank test were used to estimate the survival rates. All results were calculated by expressing the mean ± standard deviation (S.D.). A P-value less than 0.05 was considered significant. All P-values < 0.0001 are labeled as P < 0.0001.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.