Evaluation of on‐ and off‐target effects of self‐assembled epidermal growth factor receptor small interfering RNA delivery system

RNAi therapy holds inherent potential to address limitations associated with targeted anticancer drugs. However, the clinical advancement of small interfering RNA (siRNA)-based therapeutics faces a significant challenge in the realm of in vivo delivery, particularly for extrahepatic targets. To overcome this obstacle, we have recently developed an innovative siRNA delivery strategy. This approach involves reprogramming the host liver using genetic circuits. The reprogramming triggers the synthesis of siRNAs, which subsequently self-assemble into secretory small extracellular vesicles (sEVs) delivered to specific tissues through the circulating system. While the efficiency and efficacy of this delivery system have been demonstrated across various diseases,1-5 a critical aspect that remains insufficiently characterized is the on-target and off-target toxicity effects. Taking the epidermal growth factor receptor (EGFR) gene as a representative, we assessed the on-target effects in tumour and normal cells after self-assembled siRNA treatment both in vitro and in vivo. Further, we comprehensively evaluated the off-target effects in tumours and main tissues with the prospect of addressing the potential concerns of the systemic siRNA delivery approach. Following the described strategy,1 we constructed the cytomegalovirus (CMV)-siRE circuit, featuring an EGFR siRNA-expressing module downstream of a CMV promoter. To demonstrate the inhibitory potential of EGFR siRNA-encapsulating sEVs, HEK293T cells were transfected with CMV-siRE or CMV-scrR, a control expressing scramble RNA. sEVs obtained from the culture medium of HEK293T cells using differential centrifugation were analyzed using dynamic light scattering, revealing a peak at 132 nanometers (Figure S1). The separation of sEVs into distinct layers was achieved through gradient ultracentrifugation (Figure S2A). Upon transmission electron microscopy analysis, vesicles exhibiting the characteristic size and morphology of sEVs were only observed in the middle layer, but not in the low and high layers (Figure S2B). Western blotting analysis indicated that the middle layer showed high enrichment for CD9, CD63 and TSG101, distinguishing it from the low and high layers where these sEV markers were absent (Figure S2C). In addition, a majority of the EGFR siRNAs were localized within the middle layer (Figure S2D). Furthermore, the sEVs were subjected to anti-CD9 or anti-CD63 immunoprecipitation, revealing a notably higher recovery of EGFR siRNA from the CD9- or CD63-positive sEVs (Figure S2E,F). These results demonstrated the EGFR siRNA was mainly encapsulated in sEVs assembled from CMV-siRE. Next, we incubated the cell-assembled sEVs with human lung adenocarcinoma cells (H1975) or human bronchial epithelial cells (BEAS-2B) (Figure 1A). A reduction in EGFR expression was observed in both cell lines along with the increase of sEVs (Figure 1B,C and Figure S3A,B) or synthetic EGFR siRNA doses (Figure 1D,E and Figure S3C,D). Notably, even a minimal amount of EGFR siRNAs (0.3 fM) encapsulated in sEVs downregulated EGFR protein in H1975 and BEAS-2B cells by over 50%, mirroring the interference achieved by a large quantity of synthetic EGFR siRNAs (0.625 nM). Analysis of EGFR siRNA concentrations in sEV-treated cells versus EGFR siRNA-transfected cells revealed lower levels of siRNA in cells after co-culture with sEVs (Figure 1F). Despite the EGFR siRNA delivered by sEVs being approximately 1/2,000,000 of synthetic transfection, it exhibited potent interference on EGFR protein. We hypothesized that the majority of transfected synthetic siRNAs, not actively loaded into functional Argonaute proteins, act at supraphysiological levels.6 In contrast, sEV-enclosed siRNAs, processed endogenously and packaged into sEVs, were taken up in a physiological state, readily functioning in recipient cells. In alignment with the “oncogene addiction” theory,7 inhibition of EGFR protein expression by ∼50% induced significant apoptosis in H1975 cells but not in BEAS-2B cells (Figure 2A–D). In contrast, synthetic EGFR siRNAs caused increased apoptosis in both cell lines (Figure 2E–H). These results indicate that the lower concentration of sEV-enclosed EGFR siRNAs from the CMV-siRE circuit demonstrates superior biosafety to somatic cells compared to synthetic EGFR siRNAs. To assess the in vivo therapeutic efficacy and biosafety of sEV-enclosed EGFR siRNAs, we utilized an orthotopic lung cancer mouse model established with the H1975 cell line. Following tumour confirmation via micro-computed tomography (micro-CT) scanning, mice were subjected to intravenous injection with phosphate-buffered saline, CMV-scrR or CMV-siRE circuit (10 mg/kg every 2 days) seven times (Figure 3A). In line with in vitro findings, sEV-encapsulated EGFR siRNAs demonstrated potent anti-tumour effects in serum at concentrations of approximately 1000–2000 fM, akin to endogenous circulating microRNAs (miRNAs) (Figure S4A). Micro-CT imaging and 3-D reconstructions revealed substantial attenuation of lung tumour growth in mice treated with the CMV-siRE circuit (Figure 3B,C). This result was corroborated by histopathological examination of lung tissues (Figure 3D). The examination of the H&E-stained sections revealed a distinct alleviation in tumour size and a noticeable decrease in the nuclear-to-cytoplasmic ratio following the administration of the CMV-siRE circuit in comparison with control groups. Immunohistochemistry (IHC) staining confirmed lowered EGFR protein levels in CMV-siRE-treated tumours, accompanied by increased Caspase-3 (Figure 3E–G). Both EGFR protein and mRNA were significantly downregulated in lung tumours treated with the CMV-siRE circuit (Figure 3H–J) with a substantial amount of EGFR siRNA detected in tumour tissue (Figure S4B). Evaluation of potential side effects on normal tissues revealed no signs of damage in the lung, liver, brain, heart and muscle upon repeated injection of the CMV-siRE circuit (Figure S5). IHC staining of Caspase-3 in these tissues showed no apparent apoptotic signals (Figure S6). These results demonstrated significant anti-EGFR effects and tumour apoptosis while exhibiting minimal toxicity in normal tissues. Another major concern about the siRNA treatment is the off-target risk. We analyzed transcriptome changes in tumours and normal tissues, distinguishing between miRNA-like off-target effects and on-target effects of EGFR silencing. We identified 165, 206, 307 and 111 significantly altered transcripts (mean reads > 500, fold change > 2 and p < .05) in the lung, liver, spleen and kidney respectively, in CMV-siRE circuit-treated mice in comparison with the control ones (Figure 4A). Next, we evaluated whether the downregulated transcripts were directly targeted by EGFR siRNA in a miRNA-like manner. No statistically significant enrichment for perfect matches between the EGFR siRNA seed region and 3′-UTRs of downregulated transcripts was observed in normal tissues (Figure 4B). In contrast, tumour tissues from CMV-siRE circuit-treated mice showed substantial transcriptome alterations (941 differentially expressed transcripts), with only two of the 269 downregulated transcripts having direct binding potential to the EGFR siRNA seed sequence. Gene ontology (GO) enrichment analysis in tumour cells revealed that six of the top 20 GO terms were closely linked to EGFR signalling (Figure 4C). This enrichment of GO terms associated with EGFR signalling pathways suggests that widespread transcriptome alterations primarily resulted from direct EGFR target silencing, accompanied by indirect blockade of EGFR downstream signalling cascades by the CMV-siRE circuit. The data indicates that the off-target effects in normal tissues are minimal compared to the on-target effects observed in tumour cells, supporting the specificity and safety of the CMV-siRE circuit in EGFR siRNA delivery. On the other hand, self-assembled siRNAs utilize mammalian cellular machinery for gene regulation. The liver, central to EGFR siRNA-encapsulating sEV production, raised concerns about the potential disruption to host RNAi machinery. Thus, concerns about hepatotoxicity, specifically perturbing miR-122,8 a major hepatocyte-specific miRNA, were addressed. Analysis of miR-122 target genes post-CMV-siRE circuit treatment revealed only 13 significant changes among 1566 potential targets (Figure 4D). Most miR-122 targets exhibited similar expression patterns with CMV-scrR circuit treatment (Figure 4D), indicating that in vivo self-assembled EGFR siRNAs maintain physiological levels in the liver. This ensures proper processing and normal secretion and avoids competition with endogenous miRNAs for RISC loading. Our previous studies have proved the potential of a self-assembled siRNA delivery system in disease treatment.1-4 In this study, we comprehensively evaluated the potential on-target and off-target toxicity of self-assembled EGFR siRNAs. In lung cancer models, we characterized the robust bioactivity and minimal side effects of self-assembled siRNA delivery systems both in vitro and in vivo. Utilizing the self-assembled siRNA delivery system, we provide a promising solution to the challenges of incorporating RNAi therapeutics into the medical realm: how to strike a delicate balance between efficient encapsulation of RNAi drugs and mitigating off-target risks. Our research also provided a paradigmatic scheme for the evaluation of potential on- and off-target toxicity which could benefit the design of in vivo self-assembled siRNAs targeting any other genes. Xi Chen, Zhen Zhou, Xu Guo and Chen-Yu Zhang conceived and designed the experiments. Hongyuan Guo, Yuanyuan Su, Xiao Hu, Hao Zhu and Ruoyan Zhang performed the experiments. Hongyuan Guo, Yuanyuan Su, Xu Guo and Zhen Zhou contributed the materials. Xin Yan analyzed all the results of H & E in a single-blind manner. Hongyuan Guo and Zhen Zhou analysed the data. Xi Chen and Hongyuan Guo wrote the manuscript. All authors read and approved the final manuscript. This work was supported by grants from the National Natural Science Foundation of China (Nos. 32022015, 31871295 and 92268120), the Starry Night Science Fund of Zhejiang University Shanghai Institute for Advanced Study (No. SN-ZJU-SIAS-008), the CAMS Innovation Fund for Medical Sciences (No. CIFMS-2021-I2M-5-015), the Fundamental Research Funds for the Central Universities (Nos. 020814380146 and 020814380162), the Excellent Postdoctoral Program of Jiangsu Province (2023ZB697) and Natural Science Foundation of Jiangsu Province (BK20230786). The authors declare no conflict of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.