Suppression of PARP1 enhances PTEN mRNA therapy in castration-resistant prostate cancer by glycolysis disruption

Abstract

Current treatments for prostate cancer (PC) inevitably lead to the development of resistance to androgen deprivation therapy, resulting in the emergence of castration-resistant prostate cancer (CRPC), which is currently considered incurable. In this study, we discovered that the loss of phosphatase and tensin homolog (PTEN) function, in combination with elevated poly(ADP-ribose) polymerase 1 (PARP1) expression, significantly shortens the survival of PC patients. Motivated by this finding, we developed an RNA-based therapeutic agent consisting of PARP1-targeting small interfering RNA (siRNA) (siPARP1) and PTEN-expressing mRNA (mPTEN), which were co-encapsulated in an ionizable lipid nanoparticle, named mPsiP@miLAND. Both siPARP1 and mPTEN were shown to individually attenuate tumor cell growth. Moreover, the simultaneous regulation of these two targets nearly completely suppressed proliferation and robustly induced apoptosis and necrosis in CRPC both in vitro and in vivo. Mechanistically, the restoration of PTEN inhibited glycolysis via the PI3K-Akt signaling pathway in CRPC cells, and the silencing of PARP1 further enhanced this effect. Overall, the mPsiP@miLAND developed in this study effectively inhibited the growth of PTEN-deficient prostate tumors, providing a promising strategy to overcome CRPC by exploiting the synergistic effects of PI3K-Akt and PARP inhibition.

Graphical abstract

A novel therapeutic formulation was conceived by generating PTEN mRNA (mPTEN) and small interfering RNA (siRNA) targeting PARP1 (siPARP1) for castration-resistant prostate cancer treatment. Preclinical studies demonstrated siPARP1 further enhanced mPTEN efficacy in CRPC by glycolysis disruption, presenting a feasible strategy for treating CRPC through AR-independent mechanisms.

Introduction

Prostate cancer (PC) is the second most common male malignancy disease.^1,2 Androgen plays a crucial role in the occurrence and development of PC^3,4,5; accordingly, the current standard treatments are often accompanied by androgen deprivation therapy (ADT).⁶ With the disease progression, most patients eventually become resistant to ADT, developing castration-resistant prostate cancer (CRPC). Unfortunately, the standard treatments for CRPC are generally palliative.⁷ Exceptionally, clinical trials have demonstrated the effectiveness of poly(ADP-ribose) polymerase (PARP) inhibitors,^8,9 represented by olaparib, against metastatic CRPC. These reagents induce cell death by inhibiting DNA damage response in tumor cells¹⁰ and have been successfully commercialized for the treatment of BRCA-mutated advanced breast and ovarian cancers.^11,12 Although the effectiveness of olaparib has been established with an extension in progression-free survival of CPRC patients, the duration was only extended for 2.3 months in average.¹³ To further prolong the progression-free survival of patients, clinical trials have explored the combination treatment of olaparib with cabazitaxel or radium-223.^14,15 However, the safety risks of chemotherapy or radiotherapy cannot be overlooked.¹⁶

Additionally, studies have revealed that about 40% of metastatic CRPC patients harbor a phosphatase and tensin homolog (PTEN) deletion.^17,18,19,20 Functionally, PTEN encodes lipid phosphatase activities that negatively regulate the phosphatidylinositol 3-kinase (PI3K)-Akt signaling by catalyzing the dephosphorylation of phosphatidylinositol (3,4,5)-trisphosphate (PIP3).^21,22 The loss of PTEN function results in the sustained activation of PI3K-Akt signaling, enhancing the survival, proliferation,²³ migration,²⁴ angiogenesis,²⁵ and anti-apoptotic ability of tumor cells.²⁶ These findings have led to the widespread recognition of PTEN as a tumor suppressor gene.^27,28 Previous work has demonstrated the feasibility of inhibiting CPRC tumor growth by restoring PTEN activity.²⁹ In addition, it has been reported that NVP-BKM120,³⁰ another PI3K inhibitor, demonstrated synergistic effects in breast cancer treatment with olaparib. Given the significant prevalence of PTEN deletion in CRPC patients, we hypothesize that suppressed PARP can amplify the therapeutic effects on restoring PTEN in the treatment of CRPC.

In this work, we discovered that the loss of PTEN in combination with elevated PARP1, which dominates the function in PARP family, significantly shortens the survival of patients with PC. This finding encouraged us to employ an in vitro-transcribed PTEN mRNA (mPTEN) and a modified small interfering RNA (siRNA) against PARP1 (siPARP1). The therapeutic formulations were obtained by encapsulating RNAs in a novel lipid nanoparticle (LNP) termed miLAND. Studies have shown that miLAND possesses excellent stability, protecting the encapsulated mRNA from degradation for 14 days. Besides, intratumorally injected miLAND mediated sustained and specific expression of target proteins within the tumor. It was proven that individually used mPTEN or siPARP1 is capable of inducing tumor cell apoptosis and necrosis, while the efficacy was significantly enhanced when these RNAs were used in combination. Mechanistic explorations revealed that inhibited PI3K-Akt signaling reduced glycolysis in PC-3 cells, and the suppressed PARP1 significantly enhances this reduction. Moreover, the inhibited glycolysis further hindered cellular transcription and translation. Overall, this work validated the feasibility of treating CRPC by inhibiting PARP1 to enhance the efficiency of PTEN mRNA, offering valuable insights into RNA-based therapies for CRPC with PTEN deficiency.

Results

Preparation and characterization of miLAND

In this work, the lipid nanoparticle, termed miLAND, was composed of the laboratory-designed ionizable lipid A1-D1-5,³¹ 1,2-Dioleoyl-sn-glycero-3- phosphoethanolamine (DOPE), cholesterol, and 1,2-dimyristoyl-rac-glycero-3-methoxy polyethylene glycol-2000 (PEG-DMG2000) and was used for mRNA and siRNA co-delivery (Figure 1A); the molar percentages of the four lipids were 35.1%, 11.7%, 52.6%, and 0.6%. Previously, iLAND was developed as an LNP platform only for siRNA delivery and showed excellent in vivo efficiency and safety.³² In order to compare the co-delivery efficiency of mRNA and siRNA between iLAND and miLAND, luciferase mRNA (mLuc) and siRNA targeting PARP1 (siPARP1-1) were simultaneously encapsulated into each LNP. After the transfection into PC-3 cells, the luminescence intensity mediated by mLuc and the suppression efficiency of siPARP1 were measured. It was discovered that miLAND exhibited significantly higher mRNA delivery efficiency than iLAND (Figure S1A), while both demonstrated essentially equivalent PARP1 inhibition efficiency, with iLAND showed only a marginal advantage (Figure S1B). Collectively, miLAND was chosen as the preferred delivery platform for this work.

Figure 1.

The assembling process and the physicochemical characterization of miLAND and its formulation

(A) Schematic representation of the assembling process for mLsiN@miLAND. (B) The TEM image of mLsiN@miLAND, scale bar: 200 nm. (C) The size, PDI, E.E., and L.E. of miLAND before and after RNA encapsulation. (D) The cell viabilities of mLsiN@miLAND, the concentration of RNA ranging from 1 to 300 ng/well. The stability of mLsiN@miLAND at 4°C, reflected by (E) size, PDI, and (F) transfection efficiency. (G) Fluorescence microscopy images of GFP expression, scale bar: 50 μm. (H) Subcellular localization of mLsiN@miLAND, with Lipo 2000 as a positive control (RNA concentration: 50 nmol/L), scale bar: 20 μm. (I) The coefficients between LNPs and endosomes or lysosomes. (J) The quantification of cellular uptake efficiency for Lipo 2000 and miLAND. (K) Western blot analysis of PTEN expression delivered by miLAND. ∗p < 0.05, each bar chart represents the mean ± SD.

Consequently, the physicochemical properties of miLAND and its RNA formulation were characterized. After encapsulating mLuc and non-targeting siRNA (siNC) into miLAND (mLsiN@miLAND), the morphology was first observed using transmission electron microscopy. Figures S2 and 1B separately illustrated that miLAND and its RNA formulation presented typical nanoparticle morphology, with uniform particle size of about 130 and 150 nm, respectively. Consequently, the hydrodynamic size and polydispersity index (PDI) of miLAND and mLsiN@miLAND were accurately measured using dynamic light scattering. It was shown that the particle size and PDI of miLAND were 129.6 nm and 0.134, respectively (Figure 1C). After the encapsulation of RNAs, the two indicators were slightly increased to 146.7 nm and 0.231. Subsequently, the RNA-loading capacity of miLAND was determined; the encapsulation efficiency (E.E.) and loading efficiency (L.E.) were recorded as 97.4% and 71.9%, indicating that miLAND is highly effective in encapsulating RNAs. In addition, the cytotoxicity of the mLsiN@miLAND was also tested. It was considered that the encapsulated mLuc and siNC have no adverse effects on cells, and thus the cell viability reflects the safety of miLAND. Gratifyingly, the results (Figure 1D) indicated that mLsiN@miLAND did not reduce viability at the highest transfection concentration, underscoring the satisfactory safety of miLAND. Stability is another important metric for delivery carriers; therefore, the stability of mLsiN@miLAND was evaluated. Figure 1E depicted that the size and PDI of mLsiN@miLAND remained largely unchanged within 14 days (stored at 4°C), implying that the structure of the formulation did not change significantly. Additionally, the delivery efficiency of mLsiN@miLAND was tested to evaluate the integrity of mLuc within miLAND. Herein, the FDA-approved Dlin-MC3-DMA (MC3) LNP was introduced as control. The findings (Figure 1F) demonstrated that mLuc activity remained stable throughout the test, indicating that miLAND well protected the internal mRNA. Notably, miLAND exhibited a higher delivery efficiency compared to MC3 LNP, indicating that miLAND was more efficient than MC3 LNP at least in PC-3 cells. Collectively, miLAND showed outstanding safety, stability, and efficacy, positioning it as a promising carrier for RNA delivery.

Evaluating the delivery and escape capabilities of miLAND

After exploring the physicochemical properties of miLAND, we then examined its transfection efficiency by monitoring green fluorescent protein (GFP) synthesis in PC-3 cells. To this end, GFP@miLAND was prepared by encapsulating a GFP-encoding plasmid into miLAND nanoparticles, followed by transfection into PC-3 cells. After 6 h, the fluorescence generated by GFP was observed by confocal microscopy. Figure 1G showed that only miLAND mediated efficient GFP expression within cells, whereas the free plasmid failed to do so, underscoring the necessity of miLAND for endocytosis of RNA. To assess endosome/lysosome escape efficiency, the commercial transfection reagent Lipofectamine 2000 (Lipo 2000) served as control. Both Lipo 2000 and miLAND were used to transfect PC-3 cells with equal mass of Cy5-labeled siNC (Cy5-siNC) and mLuc mRNA. Four hours after transfection, the transfected cells were stained by Hoechst 33342 and Lysotracker green to label nucleus and endosomes/lysosomes, respectively. The endocytosis and coefficient between Cy5-siNC and endosomes/lysosomes was evaluated by confocal microscopy. It can be seen that the fluorescence intensity mediated by Cy5-mLsiN@miLAND and Cy5-mLsiN@Lipo 2000 in cells was found to be comparable (Figure 1H); however, the co-localization between Cy5-siNC and endosomes/lysosomes transfected by Lipo 2000 was significantly lower than that transfected by miLAND (Figure 1I). Besides, the endocytosis efficiency was further quantitatively determined by flow cytometry, and similar delivery efficiencies were reconfirmed between miLAND and Lipo 2000 (Figure 1J). Moreover, the result (Figure 1J) also revealed that both carriers achieved high positive endocytosis rates exceeding 98%. The above results led us to conclude that miLAND mediates the same endocytosis efficiency as Lipo 2000, but with a lower endosome/lysosome escape capacity. Nevertheless, considering Lipo 2000 is incapable for in vivo applications, we still believe that a more excellent delivery system was chosen in this work.

Finally, we aim to examine whether miLAND mediates PTEN expression in cells. To this end, a plasmid encoding PTEN gene was first constructed using pcDNA3.1, and the PTEN mRNA (mPTEN) was obtained through in vitro transcription technology. After encapsulating mPTEN with miLAND, the formulation was transfected to PC-3 cells. The western blot assay demonstrated that the mPTEN was effectively delivered into cells and subsequently translated into the target protein (Figure 1K).

Exploring the in vivo distribution characteristics of miLAND

Afterward, we aimed to investigate the distribution and metabolic characteristics of miLAND in mice. First, mLuc@miLAND (0.5mg/kg) or PBS was intravenously administered to healthy mice, and the living images were collected at 1, 3, 6, 9, and 24 h post-administration after all animals received luciferin. It was found that luminescence was predominantly collected in liver and spleen (Figure S3A). To more accurately understand the expression profile of mLuc@miLAND, one animal from each group was randomly selected then sacrificed for organ imaging at 3, 6, and 24 h post-administration (Figure S3B). The highest signal was recorded in the spleen, followed by the liver; no signal was detected in the other tissues or organs. Quantitative analysis provided a deeper understanding of the metabolic features of miLAND. It was revealed that the luminescence intensity in the liver was generally comparable within 6 h post-administration and then sharply decreased from 9 h post-administration, with only a very weak signal detected at 24 h post-administration (Figure S3C). In contrast, the luminescence intensity in spleen showed a linear downward trend (Figure S3D).

Second, the biodistribution and tumor retention of miLAND were explored. Here, a PC-3 subcutaneous tumor model was established using BALB/c nude mice, and an RNA cocktail composed of mLuc and Cy5-siNC was intratumorally injected into the mice with or without the encapsulation of miLAND (mLuc = 0.25 mg/kg and Cy5-siNC = 0.25 mg/kg). It was observed that Cy5-siNC delivered by miLAND highly enriched in the tumor throughout the test, whereas the siRNA without a carrier only retained in the tumor for no more than 3 h (Figure 2A). After quantitatively analyzing the living images, it was found that miLAND substantially enhanced the tumor retention of siRNA, with a 2.6- to 6.2-fold increase in fluorescence intensity from 3 to 24 h post-administration. (Figure 2B). Meanwhile, the expression pattern of mLuc was also analyzed. Only the miLAND-encapsulated mLuc mediated luminescence (Figures 2A and 2C), suggesting that miLAND is essential for mRNA expression within cells. To gain further insights into the distribution characteristics of miLAND, organ imaging was also conducted. Here, one animal from each group was randomly selected at 3, 6, and 24 h post-dose. Organs and tumors were then isolated after euthanasia and subjected to imaging. Figure 2D showed that miLAND almost exclusively enriched in tumors, which is advantageous for reducing the occurrence of adverse effects in non-targeting tissues. Contrastingly, free Cy5-siNC highly enriched both in tumor and kidneys, proving that miLAND effectively prevented RNA from entering circulation and elimination. Besides, organ imaging re-confirmed that free-delivered mLuc failed to mediate target protein expression.

Figure 2.

Evaluation of the tumor retention capability of Cy5-siNC@miLAND

(A) In vivo observation of Cy5-siNC distribution (top) and mLuc expression (bottom) at different time points. The quantification of (B) Cy5 fluorescence intensity and (C) luminescence intensity in tumors at different time points. (D) The Cy5-siNC distribution (top) and mLuc expression (bottom) in different tissues and organs at different time points. H, heart; Lu, lung; Li, liver; S, spleen; K, kidney; T, tumor. Each bar chart represents the mean ± SD.

Exploring the role of PARP1 and PTEN on PC oncogenesis

In order to explore the role of PARP and PTEN in the development of PC, we analyzed the relationship between target mRNA levels and median survival of PC patients from The Cancer Genome Atlas (TCGA) dataset.³³ Herein, PARP1 was selected as the target gene since it dominates the function in PARP family. The analysis revealed that PTEN mutations significantly reduced patients’ median survival (p = 0.0315, Figure 3A) and the level of PARP1 negatively correlated with the development of PC (p = 0.391, Figure 3B). Notably, a synergistic effect between PTEN mutations and high PARP1 expression on patients’ survival was observed (Figure 3C), but no significant difference between the two subsets (p = 0.106) was observed due to the limited number of samples with both PTEN mutations and high PARP1 expression in this dataset. Nevertheless, the analysis still strongly indicated that the levels of PARP1 and PTEN synergistically affect the development of PC.

Figure 3.

The effects of individually used siPARP1 and mPTEN on PC-3 cells

The impacts of (A) PTEN deletion mutation, (B) high PARP1 expression, and (C) the combination of both on the survival of patients with PC; the data were obtained from the TCGA database, experiments:³⁴n = 240, analyzed using cBioPortal website. (D) The comparison of the activity of two siRNAs targeting PARP1. The effects of siPARP1 on (E) genome stability and (F) cell cycle (RNA concentration: 50 nmol/L). (G) siPARP1 exhibited dose-dependent cytotoxicity in PC-3 cells. (H) mPTEN efficiently triggered PC-3 apoptosis (RNA concentration: 800 ng/mL). (I) The cell viability after treatment with different concentrations of mP@miLAND. (J) The effects of mPTEN on PI3K-Akt signaling, cellular transcription, glycolysis, and protein synthesis, evaluated by western blot assay. ∗p < 0.05, considered as statistically significant; ∗∗p < 0.01, each bar chart represents the mean ± SD.

Evaluating the efficacy of individually used siPARP1and mPTEN

Consequently, the efficacy of single used siPARP1and mPTEN on PC-3 cells was evaluated. First, we sought to identify an optimal siRNA sequence targeting PARP1 for further use. To this end, we selected two siRNAs from the literature,³⁵ which were then chemically modified and tested for their silencing efficiency using quantitative real-time polymerase chain reaction (RT-qPCR). It was shown that siPARP1-2 silenced mRNA expression by 63% and 70% at concentrations of 50 and 100 nmol/L, respectively, while siPARP1-1 only reduced mRNA expression by 33% at both concentrations (Figure 3D). Thus, siPARP1-2 was chosen as the preferred sequence, and its name was simplified to siPARP1. Subsequently, the effects of siPARP1@miLAND (siP@miLAND) on PC-3 cells were explored. Since PARP1 is the most abundant and contributes most homologous recombinations in PARP family,³⁶ the effects of siPARP1 on PARP family and genomic stability were first investigated using western blot assay. siPARP1 treatment resulted in a significant decrease in PARP levels and an increase in phosphorylated H2AX (p-H2AX), a marker of DNA damage (Figure 3E). Next, we reasoned that the unstable genome might affect the cell cycle and found that siPARP1 induced more treated cells to be arrested in the S phase and a smaller population in G2/M phase (Figure 3F), indicating that siPARP1 hindered DNA replication in tumor cells. Last, the cytotoxicity of siP@miLAND was determined. Compared with siNC treatment, siPARP1 displayed a dose-dependent reduction in cell viability reduction, with the half-maximal inhibitory concentration (IC50) on PC-3 cells being 22,831 ng/mL (Figure 3G). Collectively, the above data led us to conclude that siPARP1 leads to accumulation of damage by inhibiting DNA repair, thereby inducing the cell to be arrested in the S phase and eventually producing synthetic lethality.

Besides, the effects of mPTEN@miLAND (mP@miLAND) on PC-3 cells were also investigated. First, it can be seen that mPTEN significantly boosted necrosis or late apoptosis from 4.08% in untreated cells to 26.5% (Figure 3H), proving that the restorative PTEN efficiently triggered tumor apoptosis and necrosis. Next, the cytotoxicity of mPTEN on PC-3 was evaluated, and the result showed that mPTEN displayed a dose-dependent effect, with the IC₅₀ being 2,169 ng/mL (Figure 3I). To further understand the mechanisms underlying the activity of mP@miLAND, another western blot assay was performed. Herein, Akt, Foxo3a, the platelet isoform of phosphofructokinase (PFK) (PFKP), and 4E-BP1 were selected as markers to reflect the activity of PI3K-Akt signaling, cellular transcription, glucose metabolism, and protein synthesis, respectively. The result (Figure 3J) illustrated that mP@miLAND substantially inhibited the activity of PI3K-Akt signaling, which, in turn, affected cellular transcription, glucose metabolism, and protein synthesis pathways.

Exploring the benefits and mechanisms of simultaneously regulating PARP1 and PTEN in the treatment of CRPC

After confirming that siPARP1 and mPTEN effectively induced PC-3 cell death, the RNA cocktail encapsulated by miLAND was prepared. First, the key physicochemical properties of the final formulation were characterized. The results showed that the RNA mixture was efficiently loaded by miLAND, for which the L.E. and E.E. were 69.9% and 95.2%, respectively (Figure S4A). In addition, we found that mPsiP@miLAND maintained a compact spherical morphology (Figure S4B) and demonstrated stability over 14 days, with the particle size was around 140 nm (Figure S4C). The pKa of mPsiP@miLAND was determined since it is a critical parameter in realizing efficient nucleic acid delivery.³¹ The measured pKa of 6.33 suggests that the formulation can undergo protonation in the endosomal compartment, thereby facilitating effective endosomal escape (Figure S4D).³⁷ Afterward, we proceeded to test the cytotoxicity of mPsiP@miLAND by using PC-3 cells. The IC₅₀ of mPsiP@miLAND was calculated as 657 ng/mL, whereas the IC₅₀ values of siP@miLAND and mP@miLAND were 18,204 ng/mL and 2,984 ng/mL, respectively (Figure 4A), demonstrating a significant synergistic effect between the two RNAs. Furthermore, we calculated the combination index (CI) between two RNA therapeutics by using the Chou-Talalay method.³⁸ The results showed that the CI for the two RNAs in combination was 0.256, showing a strong synergistic effect. Next, the impacts of formulations on cell proliferation were also analyzed; it can be seen that the number of cells treated with PBS increased by 2.8 times in 4 days, while mPTEN or siPARP1 alone rendered cells incapable of proliferation (Figure 4B). Further, mPsiP@miLAND reduced the relative number of survival cells by 84% after 4 days’ treatment. In addition, similar results were also observed in cell cloning experiment (Figure 4C). In summary, the above cellular experiments demonstrated synergistic effects between mPTEN and siPARP1 on PC-3 cell killing.

Figure 4.

The effects of simultaneously regulating PARP1 and PTEN on PC-3 cells

The impacts of three formulations on (A) cell viability, (B) cell growth, and (C) clonal formation. The synergistic effects of mPsiP@miLAND on changing (D) cell cycle and on triggering (E) cellular apoptosis. (F) Western blot analysis of PI3K-Akt signaling, glycolysis, and apoptosis in PC-3 cells after treatment with siPARP1, mPTEN, or both. The effects of the formulations on (G) glucose uptake (H) ATP production, and (I) lactic acid metabolism in PC-3 cells. ∗p < 0.05, considered as statistically significant; ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, each bar chart represents the mean ± SD.

To investigate the broader applicability of mPsiP@miLAND, we extended our evaluation to the LNCaP, a commonly used androgen-sensitive PC cell line. Western blot analysis first validated the PTEN deficiency in the cells. Furthermore, all three formulations successfully modulated their target genes in LNCaP cells (Figure S5A). We then treated the cells with a range of concentrations of each formulation and calculated the IC₅₀ based on cell viability assays. Consistent with our previous findings, mPsiP@miLAND demonstrated the most potent cytotoxicity, with an IC₅₀ of 1,996 ng/mL. This value corresponds to a 5.5- and 2.7-fold increase in potency compared to siP@miLAND and mPs@miLAND, respectively (Figure S5B).

Subsequently, we sought to explore the potential mechanisms of mPsiP@miLAND on tumoricidal effects. First, it was confirmed that mPsiP@miLAND affected cell cycle. Our findings revealed that the cocktail pronouncedly induced cells arrested in the S phase, while mPTEN and siPARP1 modestly increased this proportion (Figures 4D and S6). In addition, experiments revealed that the enhanced antitumor efficacy of mPsiP@miLAND was associated with apoptosis and necrosis induction (Figure 4E). It was found that both siP@miLAND and mPs@miLAND mediated moderate cellular necrosis or late apoptosis (39.7% and 39.4%, respectively), while the cocktail treatment induced necrosis or late apoptosis in a higher percentage (49.0%). The western blot assay demonstrated that siPARP1 and mPTEN correctly regulated the levels of PARP and PTEN proteins within PC-3 cells (Figure 4F), respectively. This regulation further led to genomic instability and reduced the activity of PI3K-Akt signaling. Concurrently, the same effects were also observed in cells treated with the cocktail. In addition, we noted a decrease in the levels of pro-caspase 3 within the cells upon treatment with siPARP1 or mPTEN, and the RNA cocktail made this change more pronounced, which was consistent with previous data (Figure 4E). It was reported that the suppressed PI3K-Akt signaling inhibits cellular glycolysis.³⁹ Hence, the effects of formulations on PFKP were evaluated, since it is the rate-limiting enzyme of glycolysis and serves as an important regulatory factor in cellular energy metabolism. It was observed that mPTEN slightly reduced the level of PFKP, while siPARP1 did not. Interestingly, it was found that the cocktail significantly reduced the level of PFKP, suggesting that the inhibition of PARP1 amplified the impact of mPTEN on glycolysis. Such observation was further supported by other experiments. We reconfirmed that mP@miLAND inhibited the cellular glucose uptake from medium (Figure 4G), while also reducing the production of lactate (Figure 4H) and intracellular ATP (Figure 4I); siP@miLAND only slightly reduced the level of intracellular lactate, whereas the RNA cocktail substantially hindered glucose consumption as well as the generation of lactate and intracellular ATP. Collectively, mPsiP@miLAND showed higher antitumor efficiency by mediating stronger apoptosis, necrosis, and glycolytic disruption.

Evaluating the antitumor capability and safety of mPsiP@miLAND in CRPC model

Encouraged by the promising outcomes observed in the in vitro tests, we further evaluated the antitumor capabilities of mPsiP@miLAND by establishing a CRPC model by subcutaneously inoculating PC-3 cells into BALB/c nude mice. After that, the animals were randomly divided into four groups (n = 6 each): (1) PBS; (2) siP@miLAND, 0.6 mg/kg; (3) mP@miLAND, 0.6 mg/kg; and (4) mPsiP@miLAND, 0.6 mg/kg. Herein, the treatments were intratumorally injected into tumors when their volume reached approximately 100 mm³ (Figure 5A). The tumor volumes were monitored throughout the study to evaluate the antitumor outcomes of the treatments. It was observed that the tumors in the PBS-treated group grew 2.4 times during the entire study period. As expected, both siP@miLAND and mP@miLAND significantly slowed tumor growth, with the tumors increasing by 1.2 and 1.3 times, respectively. Furthermore, mPsiP@miLAND demonstrated the highest antitumor efficiency, with the treated tumors increasing by only 0.2 times throughout the entire study (Figure 5B). Additionally, the body weights of the animals were also recorded; Figure 5C showed no significant changes during the study, suggesting that all tested formulations are generally well tolerated.

Figure 5.

The antitumor effects of mPsiP@miLAND on PC-3 subcutaneous tumor model

(A) Brief description of the experimental process and grouping information. (B) Tumor growth curves and (C) bodyweight changes of the treated animals during the entire study period. (D) Tumor weights, (E) serum biochemical indicators, and (F) relative mRNA expression of PARP1, recorded or analyzed at the end of the study. Seven indicators including creatine kinase (CK, U/liter), aspartate aminotransferase (AST, U/liter), alkaline phosphatase (ALP, U/liter), total protein (TP, g/liter), urea nitrogen (UREA, mmol/L), serum creatinine (CREA, mmol/L), and triglyceride (TG, mmol/L) were recorded. (G) Analysis of PI3K-Akt signaling, glycolysis, apoptosis, transcription, and translation in tumors after treatment. (H) The impacts of the formulations on tumor proliferation and apoptosis, as measured by Ki67 (top) and TUNEL (bottom) staining. ∗p < 0.05, considered as statistically significant; ∗∗∗∗p < 0.0001, each bar chart represents the mean ± SD.

After completing the treatment, all animals were euthanized after 3 days to collect tumors, serum, and major organs for subsequent evaluations. Figure 5D more accurately reflected the tumor suppression efficiency of the three formulations. It is evident that siP@miLAND and mP@miLAND decreased tumor weight by 28.4% and 39.8% compared to PBS, respectively, while mPsiP@miLAND achieved a 63.6% reduction in tumor weight. Concurrently, the levels of key serum biochemical markers were also measured to evaluate the toxicity of the formulations. It was found that all three formulations led to aspartate aminotransferase elevation, indicating that such treatment strategy led to potential liver dysfunction (Figures 5E and S7). Histopathological analysis was then conducted to further evaluate the effects of formulations on organs, and it was found that the three RNA therapeutics only induced structural changes in tumors but not in organs (Figure S8), suggesting that all formulations showed efficient tumoricidal effects and were generally well tolerated. The safety of formulations were also evaluated by calculating the weight coefficients of the liver and spleen to body (Figure S9), and it was found that the three formulations did not result in obvious changes in either the liver/body or spleen/body weight ratios. Besides, consistent with in vitro experiments, siPARP1 and mPTEN effectively regulated the level of PARP and PTEN. Figure 5F showed that two formulations containing siPARP1 markedly suppressed the relative expression of the target gene in tumors compared to PBS. Specifically, siP@miLAND suppressed PARP1 expression by 85.1%, while mPsiP@miLAND resulted in comparable PARP1 inhibition, which was 86.8%. Such result led us to conclude that 0.3 mg/kg siPARP1 is close to the maximum inhibition efficiency, and further increasing the dosage would not significantly improve gene suppression. This conclusion also suggests that the dosage could be further reduced in this study.

To reveal how mPsiP@miLAND inhibits tumor growth, the effects of the three formulations on tumor tissue were thoroughly evaluated by western blot assay (Figure 5G). It is evident that mP@miLAND and mPsiP@miLAND efficiently inhibited PI3K-Akt signaling by rescuing the activity of PTEN. Besides, it was also confirmed that the two formulations containing siPARP1 significantly reduced the level of PARP. Markedly, we noticed that individually used siPARP1 did not affect PI3K-Akt signaling. However, when combined with mPTEN, it further reduced the activity of the signaling. This suggests that siPARP1 does not directly target the PI3K-Akt signaling but rather enhances the efficacy of mPTEN. A similar trend was observed in the changes of PFKP, the downstream factor of the PI3K-Akt signaling, which further corroborates the conclusion. Moreover, the increased p-H2AX indicated that all three formulations induced genomic instability in tumor cells, which was consistent with in vitro findings. A more pronounced downregulation of p-Foxo3a and p-4E-BP1 was discovered in mPsiP@miLAND-treated tumors, suggesting that abnormal glycolysis in tumor cells further affects transcription and translation activities. Overall, siPARP1 enhances the efficacy of mPTEN rather than directly targeting PI3K-Akt signaling. The excellent antitumor activity of mPsiP@miLAND is primarily attributed to the synergistic effects of mPTEN and siPARP1 on disrupting glycolysis and energy-dependent activities. Finally, terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end labeling (TUNEL) assays and hematoxylin and eosin (H&E) staining were conducted. Figure 5H visually demonstrated that siP@miLAND and mP@miLAND significantly inhibited tumor proliferation, while mPsiP@miLAND almost completely inhibited this activity by inducing the most significant apoptosis.

Antitumor study of mPsiP@miLAND in androgen-sensitive PC model

Furthermore, we evaluated the antitumor efficacy of formulations using an LNCaP subcutaneous xenograft model. Given the formulation-induced liver abnormalities observed in previous antitumor study, the administered dose of RNA was reduced from 0.6 to 0.3 mg/kg in this study to avoid such adverse effects. Tumor-bearing nude mice were randomly assigned to five groups: (1) PBS, n = 5, (2) free mPsiP (the RNA mixture without miLAND encapsulation, n = 6), (3) siP@miLAND, n = 6; (4) mP@miLAND, n = 6; and (5) mPsiP@miLAND, n = 6. The animals received five intratumoral injections on study day 1, 4, 7, 10, and 13, respectively (Figure 6A). During the entire experimental period, it was observed that the average tumor volume in the PBS group increased by 3.5 times. As anticipated, free mPsiP showed no therapeutic effect due to the absence of a delivery system, with the treated tumors growing 3.3 times. In contrast, both siP@miLAND and mP@miLAND exhibited antitumor activity, resulting in tumor volume increase by 2.0 and 1.7 times, respectively. Notably, mPsiP@miLAND demonstrated the most potent therapeutic efficacy, maintaining the overall tumor volume without change (Figure 6B). In this study, animals with tumor volumes exceeding 600 mm³ were considered to have reached the endpoint. Based on this rule, survival analysis revealed that all animals in the PBS and free mPsiP groups were found dead by study day 39, whereas only two animals in the siP@miLAND group and one in the mP@miLAND group were found dead within 45 days. Owing to the remarkable antitumor efficacy, mPsiP@miLAND ensured the survival of all animals throughout the study period (Figure 6C), and even resulted in complete tumor eradication in two animals Additionally, the body weight of animals in all groups remained stable during the study, suggesting good overall tolerability to the treatments (Figure 6D).

Figure 6.

The antitumor effects of mPsiP@miLAND in LNCaP-bearing mice model

(A) Brief description of the experimental process and grouping information. (B) Tumor growth curves, (C) animal survival, and (D) bodyweight changes of the treated animals during the entire study period. On study day 16, three mice in each group were sacrificed; (E) tumor image and (F) tumor weight were recorded. (G) RT-qPCR and (H) western blot assay were performed to analyze the change of PARP1 and PTEN, respectively. (I) Intratumoral concentration of siPARP1 antisense strand, measured using stem-loop RT-qPCR technology. (J) Seven indicators were monitored to evaluate the safety of the tested formulations. Creatine kinase (CK, U/L), aspartate aminotransferase (AST, U/L), alkaline phosphatase (ALP, U/L), alanine transaminase (ALT, U/L), serum creatinine (CREA, mmol/L), urea nitrogen (UREA, mmol/L), and triglyceride (TG, mmol/L). The coefficient of (K) liver to body weight and (L) spleen to body weight. (M) Histological alterations in tumors visualized by H&E staining. ∗p < 0.05, considered as statistically significant; ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001, each bar chart represents the mean ± SD.

Meanwhile, another 15 LNCaP tumor-bearing mice were randomly assigned and received the same treatments. Three days after the final treatment, all animals were euthanized, and their blood, tumors, and major organs were collected for further evaluation. Figures 5E and 5F demonstrate the antitumor effects across treatment groups. Subsequent RT-qPCR and western blot analysis revealed that siPARP1 effectively suppressed PARP1 gene expression (Figure 6G), while mPTEN restored PTEN protein levels via mRNA translation (Figure 6H). To accurately quantify the RNA accumulation in tumor tissues until mouse endpoint, stem-loop RT-qPCR technology was employed to determine the concentration of siPARP1 (Figure 6I).⁴⁰ It was found that the free RNA cocktail yielded only 0.70 ng/g of siPARP1 in tumor tissue, while miLAND increased this level by 12.2-fold to 8.52 ng/g. The level of siPARP1 in siP@miLAND-treated tumors was 14.88 ng/g, showing a correlation between the administration concentration and tumor enrichment of siPARP1. In addition, siPARP1 was barely detected in PBS- and mP@miLAND-treated tumors, since these two formulations lacks siPARP1. Finally, we evaluated the safety of all formulations. Serum biochemical analysis showed that all formulations did not cause any dysfunction in the heart, liver, and kidney (Figure 6J), with all factors being within normal ranges (Figure S10). The organ coefficient of liver to body (Figure 6K) and spleen to body (Figure 6L) further indicated that the tested formulations led no dysfunction in the liver and spleen. Immunohistochemical sections demonstrated no noticeable damage to normal tissues across all treatment groups (Figure S11). Importantly, it was observed that mPsiP@miLAND induced more pronounced structural changes in tumors compared to siP@miLAND and mP@miLAND (Figure 6M), highlighting the synergistic antitumor effect of combining siPARP1 and mPTEN. Collectively, these findings demonstrate that reducing the mPsiP@miLAND dosage from 0.6 mg/kg to 0.3 mg/kg retains potent antitumor efficacy while enhancing safety. Furthermore, the study confirms that the efficacy of mPsiP@miLAND extends beyond CRPC to other PC cell lines, underscoring its broad therapeutic potential.

Discussion

In this study, an RNA cocktail comprising siPARP1 and mPTEN was constructed and showed significant synergistic effects in inhibiting CRPC and androgen-dependent PC. The mPsiP@miLAND was developed to enhance the therapeutic efficacy on PC, our findings revealed that the cocktail increased tumor apoptosis and necrosis, and more effectively disrupted cellular glycolysis, as well as following energy-dependent transcription and translation activities.

The relevant targets have been proven to be effective in treating PC in preclinical or clinical trials. Studies have shown that more than 40% of mCRPC patients experience PTEN loss, which significantly promotes tumor cell survival. Islam et al. showed that the restoration of PTEN in PTEN-null PC inhibits the PI3K-AKT signaling and enhances apoptosis.²⁹ Furthermore, olaparib showed benefits on progression-free survival in mCRPC patients, regardless of whether BRCA1/2 is mutated.¹³ However, to our knowledge, suppressing PC tumors by simultaneously regulating two such targets has not been previously shown. This work revealed that inhibitors of PARP1 and PI3K-Akt signaling synergistically halted tumor growth. It was reported that inhibited PI3K-Akt signaling affects genomic stability,³⁰ and our finding confirmed this conclusion, leading us to speculate that it is the potential mechanism for the synergistic effect of the cocktail. However, we further discovered that siPARP1 and mPTEN more significantly disrupted glycolysis in tumor cells. Given that the presence of the Warburg effect is an important characteristic of tumor cells, which is characterized by high aerobic glycolysis,⁴¹ inhibiting glycolysis is meaningful for tumor treatment. In addition, overcoming ADT resistance after long-term treatment is a pivotal challenge in the effective treatment of PC. The drug resistance mechanism of CRPC involves complex biological processes. Although we cannot definitively prove whether the cocktail can avoid drug resistance, we still demonstrated its remarkable effectiveness in CRPC mice model, hence proposing a new option that may avoid ADT-dependent drug resistance during PC treatment.

The development of RNA therapeutics necessitates overcoming obstacles including the inherent instability of RNAs.⁴² In this work, we have chemically modified siPARP1 and introduced 5′ARCA cap and 3′ poly (A) tail to the coding sequence of mPTEN. These modifications were designed to enhance efficiency and confer resistance to exonucleases.^43,44,45 Furthermore, LNP was selected as carrier since it has risen as a versatile delivery system.⁴⁶ Herein, we fabricated a novel LNP termed miALND, based on laboratory-designed thermostable ionizable lipids A1-D1-5. It efficiently encapsulated and delivered RNA cocktails. Our results demonstrated that miALND maintained a uniform and stable morphology for up to 14 days post-preparation. It effectively protected the activity of the encapsulated mRNA from degradation, thereby significantly improving the formulation’s stability and availability. Additionally, miLAND exhibited a relatively small size (around 120 nm), an exceptionally high mRNA E.E. (higher than 95%), and a compact spherical shape, all of which contributed to high transfection efficiency and remarkable tumor suppression capability. In summary, this study presents a highly efficient therapeutic strategy for CRPC utilizing a novel RNA cocktail. This approach effectively disrupts glycolysis in tumor cells by concurrently modulating the levels of PARP1 and PTEN, offering a promising avenue for the treatment of advanced PC.

Materials and methods

Materials

DOPE (4004-05-1), cholesterol (57-88-5), and DMG-PEG₂₀₀₀ (160743-62-4) were purchased from AVT Pharmaceutical Co. Ltd (Shanghai, China). BCA Protein Assay Kit (CW0014S) and GoldHi plasmid extraction kit (CW2108M) were purchased from CWBIO Co., Ltd (Jiangsu, China). BstZ17I-HF endonuclease (R3594L), rCutSmart Buffer (B6004S), and HiScribe T7 ARCA mRNA in vitro transcription kit (E2060S) were purchased from NEB (MA, USA). Phenol:chloroform DNA extraction (0883-100ML) was purchased from Amresco (MA, USA). F12 medium (ZQ-400) was purchased from Shanghai Zhong Qiao Xin Zhou Biotechnology Co., Ltd. Fetal bovine serum (FBS) was purchased from PAN Biotech (Aidenbach, Germany). Dual Luciferase Reporter Assay Kit (DL101-01), RNA isolater (R401-01-AA), and HiScript III 1st Strand cDNA synthesis Kit (R312-01) were purchased from Vazyme Biotech (Nanjing, China). Passive Lysis 5X Buffer (E1941), beetle luciferin, and potassium salt (E1605) were purchased from Promega (Wisconsin, USA). Cell Counting Kit-8 (CA1210) and Lactic Acid Detection kit (BC2235) were purchased from Solarbio Life Science (Beijing, China). qPCR SYBR Green Mix (11201ES03), Cell Cycle and Apoptosis Analysis Kit (40301ES60), and Annexin V-FITC/PI Apoptosis Analysis Kit (40302ES20) were purchased from Yeasen biotechnology (Shanghai, China). Quant-it RiboGreen RNA Kit (R11491), Lipofectamine 2000 (11668-019), Opti-MEM (11058-021), and RNA Later (AM7021) were purchased from Thermo Fisher (MA, USA). Penicillin-streptomycin and trypsin (15070063) was purchased from Gibco (MA, USA). Glucose Assay Kit with o-toluidine (S0201), Cell and Tissue Lysis Buffer for Glucose Assay (S3062), and Enhanced ATP Assay Kit (S0027) were purchased from Beyotime (Shanghai, China). PTEN mAb (9188), PARP mAb (9532), p-Akt (Ser 473) mAb (4060), PFKP mAb (8164), phospho-histone H2A.X (Ser139) mAb (9178), p-Foxo3a (Ser318/321) mAb (9465), and 4E-BP1 (Thr37/46) mAb (2855) were purchased from Cell Signaling Technology (MA, USA). Caspase 3 (ab184787) was purchased from Sbcam. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) mAb (60004-1-Ig) was purchased from Proteintech (Illinois, USA).

The preparation of miLAND and its formulations

A1-D1-5, DOPE, cholesterol, and DMG-PEG₂₀₀₀ were dissolved in ethanol at a concentration of 20 mg/mL, respectively. Then, the organic components with given volume were mixed into a single tube. Concurrently, sodium citrate buffer (0.05 mol/L, pH 4.0) was prepared in a separate tube, with a volume three times that of the organic phase. Finally, the organic phase was transferred into a syringe and rapidly injected into the buffer-containing tube with stirring to form miLAND.

The concentration of RNAs was adjusted to 1,000 ng/μL by diethyl pyrocarbonate (DEPC)-treated water and then mixed with aliquots of 50% (v/v) ethanol. Subsequently, appropriate amount of miLAND and RNA solution were mixed at a mass ratio of 15:1 and immediately incubated at 50°C for 15 min. Finally, the formulation was dialyzed in PBS for 2 h.

For mPsiP@miLAND preparation, siP@miLAND and mP@miLAND were first separately formulated. The L.E. of miLAND for siPARP1 and mPTEN was then determined using Quant-it RiboGreen RNA Assay Kit. Based on the measurement result, siP@miLAND and mP@miLAND were mixed at an appropriate ratio to achieve a 1:1 mass ratio of siPARP1 to mPTEN in the final formulation.

Physicochemical characterization

Morphological observations were conducted using a transmission electron microscope (Tecnai G2 F20 U-TWIN, FEI, USA). The particle size and PDI of LNP were determined using Zetasizer 3000HS (Malvern, UK). The measurements were performed at a wavelength of 677 nm with a fixed angle of 173° at room temperature.

Encapsulation determination

The L.E. and E.E. of RNAs were determined by using Quant-it RiboGreen RNA Assay Kit following the protocol. Briefly, the dialyzed formulation was diluted 20 times by Tris-Ethylenediaminetetraacetic acid (TE) buffer and then added into a 96-well light-proof plate with the volume of 100 μL to determine free RNA. Meanwhile, 15 μL sample was added to a well containing 35 μL TE buffer and 50 μL 2% (w/v) Triton X-100 to determine total RNA. The mixture was incubated at room temperature for 30 min. Finally, 100 μL of RiboGreen RNA quantitation working solution was added to each well, and the fluorescence intensity of the samples was immediately measured using a microplate reader. The concentration of RNA was calculated by its standard curve. E.E. and L.E. were calculated according to the following equations:

$$\text{Encapsulation}\text{efficiency}(\%)=\frac{\mathrm{m}\left(\text{siRNA}\text{after}\text{dialysis}-\text{free}\text{RNA}\right)}{\mathrm{m}\left(\text{RNA}\text{after}\text{dialysis}\right)}\times 100$$ Equation 1

$$\text{Loading}\text{efficiency}(\%)=\frac{\mathrm{m}\left(\text{total}\text{RNA}\right)-\mathrm{m}\left(\text{free}\text{RNA}\right)}{\mathrm{m}\left(\text{total}\text{RNA}\right)}\times 100\text{.}$$ Equation 2

Cell culture

Human PC cells PC-3 were purchased from Shanghai Zhong Qiao Xin Zhou Biotechnology Co., Ltd. and identified by STR profiling. The cells were F12 medium containing 10% (v/v) FBS, and 1% (v/v) penicillin and streptomycin was used as the cell culture medium. Cells were grown in an incubator at 37°C with 5% CO₂. When the cells reached approximately 80% confluence in the culture dish, they were digested with trypsin at 37°C for 1–2 min. Cell suspensions were collected and centrifuged, and then the cells were re-suspended in fresh medium. Finally, half of the cells were transferred to a new dish and continued to be cultured in the incubator.

Cell viability

PC-3 cells were seeded into a 96-well plate at a density of 10,000 cells per well. On the subsequent day, siNC and mLuc were mixed together and then encapsulated by miLAND. The transfection was performed with concentrations of 1, 3, 10, 30, 100, and 300 ng per well. After 24 h, the culture medium was discarded and replaced with fresh medium containing CCK8. The cells were further cultured in the incubator for additional 1 h. The absorbance was measured using microplate reader at 450 and 630 nm. The relative cell viability of treated cells was referring to control.

Stability determination

PC-3 cells were seeded in 96-well plate on the day prior to transfection. MC3 LNPs were prepared according to the method described for miLAND, with a molar ratio of 50:10:38.5:1.5 for Dlin-MC3-DMA, DSPC, cholesterol, and DMG-PEG₂₀₀₀. mLuc was encapsulated into miLAND and MC3 LNP and then dialyzed. Subsequently, cells were transfected by two formulations with 100 ng/well on days 0, 4, 7, 10, and 14 post-preparation. During this period, the samples were stored at 4°C. The luminescence intensity mediated by the two formulations was measured using Dual Luciferase Reporter Assay Kit at 6-h post-transfection. The relative stability of the samples was referred to the luminescence intensity mediated by mLuc@MC3 LNP on day 0 post-preparation.

pKa determination

The pKa values of the LNPs were determined using the fluorescent probe 2-(p-toluidino)-6-naphthalene sulfonic acid (TNS). First, a stock buffer solution was prepared containing NaCl (1300 mmol/L), CH₃COONH₃ (100 mmol/L), HEPES (100 mmol/L), and MES (100 mmol/L). A TNS stock solution (100 μmol/L) was prepared by dissolving TNS in deionized water. Both miLAND and LungtiLAND were then diluted to 100 μmol/L using the stock buffer and divided into 19 aliquots. The pH of each aliquot was adjusted across a range from 2.5 to 11. Subsequently, TNS was added to each aliquot to a final concentration of 1 μmol/L. The fluorescence intensity of each sample was measured immediately at room temperature using a microplate reader with an excitation wavelength of 321 nm. Finally, the pKa values were calculated from the fluorescence-pH titration curve by applying the Henderson–Hasselbalch equation in GraphPad Prism (v.8).

Transfection efficiency and endosome/lysosome escape

PC-3 cells were seeded into 3.5-cm dishes at a concentration of 1×10⁵ cells/mL. On the next day, plasmid expressing green fluorescent protein (GFP) was encapsulated by miLAND. After that, the cells were transfected by the formulation at a concentration of 1,600 ng GFP per well. After 6 h, the culture medium was discarded and the cells were washed with PBS three times and then observed and photographed under a confocal microscope.

Cy5-siNC and mLuc were encapsulated by miLAND and then dialyzed. PC-3 cells were transfected by the formulation at a concentration of 1,600 ng/well. After 4 h, the medium was discarded and the cells were washed by PBS three times. After that, cells were stained by Hoechst 33342 and LysoTracker Green for 15 min. Finally, the cells were observed and photographed under a confocal microscope. The Pearson’s correlation coefficient was analyzed using NIS_Viewer software. Besides, the fluorescence intensity of transfected cells was measured by FACSverse flow cytometer. Analysis was performed using FlowJo 10 software.

mRNA and protein expression analysis

For mRNA analysis, the tumor samples that received different formulations were cut into 1 mm³ pieces and then soaked in RNAlater and stored at 4°C for short-term preservation. Upon initiating the assay, the samples were transferred to RNA isolater lysate and thoroughly ground using a KZ-Ⅲ-F tissue grinding machine (Servicebio, China). Total RNA extraction was performed using chloroform and isopropanol, and 75% ethanal (v/v) was employed to remove impurities. cDNA was reverse transcribed using HiScript III 1st Strand cDNA Synthesis Kit. Quantitative PCR (qPCR) was conducted with qPCR SYBR Green Mix. The procedures referred to the kits’ introduction. The relative expression of PARP1 mRNA was normalized against GAPDH. The sequences for siPARP1 and the primers for PARP1 and GAPDH can be found in the supplemental information. Western blot assay was performed for proteins analysis. The samples of cells or tumors were lysed using passive lysis buffer supplemented with a protease inhibitor at room temperature for 20 min. Consequently, the samples were transferred to tubes and centrifuged at 12,000 rpm for 10 min. The supernatant was then collected and placed into new tubes, which were stored on ice. Protein concentrations were determined using BCA Protein Assay Kit, following the manufacturer’s protocol. Samples containing 20 μg of total protein were mixed with 5× loading buffer and denatured at 99°C for 10 min. The targeting proteins were separated by SDS-PAGE electrophoresis and immunoblotted with indicated antibodies.

Cell cycle and apoptosis assay

PC-3 cells were seeded into 6-well plate. The sample preparation and transfection are carried out as previously described. Twenty-four hours post-transfection, the cells were digested with trypsin. The cells were collected and then washed three times with PBS and fixed in 70% ethanol at 4°C for 2 h. The treated cells were then washed, centrifuged, and stained by propidium iodide (PI). Meanwhile, RNase A was also used to eliminate the contamination from RNA. Subsequently, the fluorescence intensity of PI-labeled cells was measured using FACSverse flow cytometer. For apoptosis assay, the treated cells were collected and then washed, centrifuged, and stained by a solution containing Annexin V-FITC and PI. Subsequently, the fluorescence intensity of FITC and PI on cells was measured using a FACSverse flow cytometer. The data were analyzed using FlowJo 10 software.

Glucose consumption, ATP production, lactic acid production

The treatment procedures have been described previously. The testing procedures were carried out in accordance with the manufacturers’ instructions.

Colony formation

PC-3 cells were seeded into 6-well plates at a density of 1,000 cells per well. The treatments were done following the aforementioned description. Thereafter, the culture medium was replaced with different formulations every 3 days. After 2 weeks, the cells were fixed by methanol at room temperature for 30 min. After that, those cells were stained by a solution containing 0.1% crystal violet at room temperature for 3 min. Finally, the cells were carefully washed and photographed by digital camera.

Animal housing

All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Beijing Institute of Technology and performed in accordance with the guidelines and policies (approval number BIT-EC-SCXK (Beijing) 2016-0006-M-202016). BALB/c mice and BALB/c nude mice were purchased from SPF Biotechnology Co., Ltd., with age range of 6–8 weeks. The mice were housed in a temperature-controlled, ventilated room with an artificial light source that alternated between 12 h of light and 12 h of darkness. Animals were provided with free access to food and water.

In vivo biodistribution

Eight shaved BALB/c mice were randomly divided into two groups (n = 4). The mice were then intravenously injected with 0.5 mg/kg of either naked mLuc or mLuc@miLAND. Subsequently, at 1, 3, 6, 9, and 24 h post-injection, the animals were intraperitoneally administered 10 mg/kg of luciferin. After 10 mins, the animals were anesthetized with isoflurane and imaged using an IVIS in vivo imaging system. Furthermore, one animal from each group was randomly selected and euthanized after the imaging at the 3-, 6-, and 24-h time points. Organs and tissues including the submandibular gland, thymus, heart, lungs, liver, spleen, and kidneys were harvested for ex vivo imaging.

PC-3 cells were subcutaneously inoculated into the right hind leg of BALB/c nude mice at a density of 5 × 10⁶ cells per mouse. When the tumor volume reached approximately 200 mm³, six animals were randomly divided into two groups. Equivalent weight of mLuc and Cy5-siNC were mixed and encapsulated by miLAND and then intratumorally injected into three nude mice at 0.5 mg/kg (mLuc = 0.25mg/kg and Cy5-siNC = 0.25mg/kg). Meanwhile, equivalent un-encapsulated mLuc and Cy5-siNC were also intratumorally injected into another three nude mice. The living images were collected at pre-dose and at 1, 3, 6, and 24 h post-dose. After completing living images at 3-, 6-, and 24-h post-dose, one animal from each group was randomly selected for euthanasia, and organs including heart, lungs, liver, spleen, kidneys, and tumor were isolated for organ imaging. The fluorescence and luminescence intensity of the regions of interest were analyzed using LivingImage software.

Antitumor test

PC-3 cells were subcutaneously inoculated into BALB/c nude mice. After that, tumor volume was continuously measured. When the tumor volume reaches approximately 100 mm³, 24 animals were randomly divided into 4 groups, (1) PBS, n = 6; (2) siP@miLAND, 0.6 mg/kg, n = 6; (3) mP@miLAND, 0.6 mg/kg, n = 6; and (4) 0.6 mg/kg mPsiP@miLAND, 0.6 mg/kg, n = 6. The animals received the corresponding formulations intratumorally on days 1, 4, 7, 10, 13, and 20. The body weight and tumor volume were continuously monitored throughout the test. All animals were euthanized on day 23, and serum was collected from the whole blood for serum biochemistry analysis. The tumors were isolated, weighed, and divided into three parts for preservation. One part was preserved in RNAlater for gene expression analysis, another part was frozen in liquid nitrogen for protein expression analysis, and the final part was preserved in 4% formalin for subsequent H&E, TdT-mediated dUTP nick end labeling (Tunel), and Ki67 staining.

In a separate study, LNCaP cells were subcutaneously inoculated into BALB/c nude mice. The animals were randomly assigned into 5 groups: (1) PBS, n = 8; (2) Free mPsiP, 0.3 mg/kg, n = 9; (3) siP@miLAND, 0.3 mg/kg, n = 9; (4) mP@miLAND, 0.3 mg/kg, n = 9; and (5) mPsiP@miLAND, 0.3 mg/kg, n = 9. The intratumoral treatment was initiated when the tumor volume reaches approximately 100 mm³. The day of the first treatment was defined as study day 1. The following treatments were performed on study days 3, 7, 10, and 13. On study day 16, three animals of each group were euthanized, and their blood, tumor, and major organs were collected for serum biochemical, RT-qPCR, western blot, and stem-loop RT-qPCR assays. The remaining mice were kept for survival monitoring and euthanized when tumors reached 600 mm³.

Stem-loop RT-PCR

20 mg of untreated tumor tissue was accurately weighed into a 1.5-mL tube, 190 μL of 0.25% Triton-PBS was added, and the tissue was homogenized completely. After that, 10 μL of 20 ng/μL siPARP1 was added into the tube for further use. In addition, the treated tumors were also accurately weighed for 20 mg and then added into tubes containing 500 μL of 0.25% Triton-PBS and homogenized thoroughly. All samples were heated at 95°C for 5 min and then cooled rapidly on ice to thoroughly release siPARP1. After that, the samples were centrifuged at 12,000 × g, 4°C for 20 min, and the supernatant was isolated to serve as standard and determinants, respectively. The stock standard was serially diluted to generate a calibration curve. Reverse transcription and RT-PCR were performed for both standards and samples. The concentration of siPARP1 antisense was calculated by its calibration curve. The primer and probe information were provided in the supplemental information.

Statistical analysis

All data were analyzed using GraphPad Prism 8.0 software. Bilateral Student’s t test was used for comparing two groups, and univariate or two-factor analysis of variance (ANOVA) and Tukey’s correction were used to analyze differences among the groups. ∗p < 0.05, considered as statistically significant; ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001. Each bar chart represents the mean ± S.D.

Data and code availability

All data are available in the main text or supplemental information.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (grant nos. 82202338, 32171394, 32471514, and 22304106), the China Postdoctoral Science Foundation (2024M764115), the Fundamental Research Funds for the Central Universities (grant no. 2022CX01013), and the Taishan Scholars Program (tsqnz20221157). The authors thank the Biological and Medical Engineering Core Facilities, and Analysis & Testing Center, Beijing Institute of Technology, for supporting experimental equipment and the staffs for valuable help with technical support. All animal procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of Beijing Institute of Technology and performed in accordance with the guidelines and policies (approval number BIT-EC-SCXK (Beijing) 2016-0006-M-202016).

Author contributions

B.H. performed the most of experiments. B.H., P.W., L.Z., H.Y., and Y.F. performed animal study. J.Y. participated in figure organization. S.K. synthesized all siRNAs. C.Z. gave advices for survival analysis. L.T., M.Y., and Q.L. participated in manuscript revising. B.H. and Y.H. conceived the project, designed research, and wrote the manuscript. J.Z. and Y.H. supervised experiments and critically reviewed the manuscript.

Declaration of interests

Y.H. is the founder of Rigerna Therapeutics and declares no competing interests. S.K. is the founder of Suzhou Biosyntech Co., Ltd and declares no competing interests.