Myeloid cell-targeted lipid nanoparticles for B cell lymphoma immunotherapy

Abstract

We previously demonstrated that blocking tolerogenic signal transducer and activator of transcription 3 (STAT3) signaling in the tumor microenvironment can unleash Toll-like receptor 9 (TLR9)-mediated antitumor immunity. To enable systemic administration of minimally modified CpG-siSTAT3, we developed a panel of MC3-based lipid nanoparticle (LNP) formulations optimized for targeting immune cells and B cell lymphoma cells. The selected LNP2(CpG-siSTAT3) induced potent type I interferon (IFN) production in human peripheral blood mononuclear cells (PBMCs) and resulted in >50% STAT3 knockdown in human cancer cells at low oligonucleotide concentrations. In vivo, LNP2(CpG-siSTAT3) showed a 10-fold improvement in potency against B cell lymphoma xenotransplants compared to the naked oligonucleotide. Further changes in chemical composition yielded LNP2.1, which preferentially targeted human monocytes and dendritic cells (DCs). In A20 lymphoma-bearing mice, the fluorescently labeled LNP2.1(CpG-siSTAT3) quickly drained to local tumor-draining lymph nodes (TDLNs) after subcutaneous injection and was taken up by activated DCs and macrophages. Furthermore, LNP2.1(CpG-siSTAT3) administration significantly reduced A20 tumor growth by rapidly activating DCs and macrophages in TDLNs, thereby promoting T cell activation and specifically increasing tumor-infiltrating cytotoxic CD8 T cells secreting IFNγ and tumor necrosis factor alpha. The LNP2.1 formulation offers an effective vehicle for targeting tolerogenic myeloid cells in the B cell lymphoma microenvironment and potentially in solid tumors.

Graphical abstract

Lipid nanoparticle formulation delivering dual-functional CpG-siSTAT3 to B cell lymphoma and associated myeloid cells achieves enhanced drainage to tumor-draining lymph nodes, a broadened therapeutic window through improved potency at lower doses, and activation of dendritic cells and tumor-infiltrating T cells, offering an effective strategy to boost antitumor immunity.

Introduction

B cell lymphoma-associated myeloid cells, such as macrophages and myeloid-derived suppressor cells (MDSCs), support lymphoma progression by creating an immunosuppressive tumor microenvironment (TME) that dampens the development of antitumor immune responses.^1,2,3,4 The density and polarization state of tumor-associated myeloid cells (TAMs) serve as important prognostic markers in B cell lymphomas, with M2-like TAM accumulation generally predicting poorer outcomes, thus establishing myeloid cells as crucial therapeutic targets. Despite multiple approaches to target TAMs and MDSCs, such as antibodies or small-molecule inhibitors of CSF1R, CXCR2,⁴ SIRPα,⁵ and Siglec,⁶ the successful clinical outcomes of these therapies are limited thus far. These therapeutic limitations result from the challenges of selectively depleting tumorigenic myeloid cells, such as TAMs or MDSCs, without affecting the majority of myeloid cells that are necessary for sustaining the patient’s innate immune responses to microbial or fungal infections (e.g., dermal macrophages and hepatic macrophages).

The cytokine-mediated cross-talk between B cell lymphoma cells and myeloid cells often propagates oncogenic and tolerogenic signaling through signal transducer and activator of transcription 3 (STAT3) from cancer cells to TAMs.^7,8 In the activated B cell subtype of diffuse large B cell lymphoma (ABC-DLBCL), genetic mutations augment Toll-like receptor 9 (TLR9)/MyD88 signaling, thereby leading to the secretion of cytokines that induce STAT3 in lymphoma cells and in TAMs and MDSCs.⁹ The autocrine and paracrine STAT3 activation in cancer cells and in lymphoma-associated myeloid cells enhances STAT3-mediated tumorigenicity and immunosuppression. We previously developed an oligonucleotide strategy to target STAT3 specifically in TLR9+ B cell lymphoma cells and lymphoma-associated myeloid cells using conjugates of STAT3 inhibitors (siRNA or decoy DNA) with CpG oligodeoxynucleotides (CpG-ODNs) targeting TLR9. CpG-STAT3siRNA/decoy inhibitors demonstrate a direct cytotoxic effect toward STAT3-dependent lymphoma B cells while promoting antigen presentation via dendritic cells (DCs) and antitumor T cell-mediated responses.^10,11 More specifically, the therapeutic effects of CpG-STAT3 inhibitors rely on the synergy of blocking STAT3 in lymphoma cells and myeloid cells to unleash TLR9 immune stimulation.^8,12

The success of coronavirus disease 2019 mRNA vaccines stimulated efforts to develop myeloid cell-targeted tumor neoantigen mRNA vaccines for cancer immunotherapy using lipid nanoparticle (LNP) delivery systems. The LNPs are formulations assembled mainly from four core lipid components, including ionizable lipid, cholesterol, helper lipid, and PEG lipid, which complex with polyanionic oligonucleotides to form colloidal structures.¹³ The ionizable lipid component, crucial for encapsulation and endosomal escape, can potentially trigger the production of proinflammatory mediators such as interferon (IFN)-I, interleukin (IL)-1, IL-6, tumor necrosis factor alpha (TNF-α), CXCL1, CXCL2, or prostaglandins.^14,15 While the levels of these immunostimulatory mediators are not likely to overcome tumor immune suppression, they can promote inflammatory processes in the TME and augment the tolerogenic effects of STAT3.^12,16,17

Here, we developed specialized LNP formulations designed to enhance delivery of immunostimulatory oligonucleotides to myeloid cells while maintaining their direct cytotoxicity toward B cell lymphoma cells for cancer immunotherapy. We selected the CpG-STAT3siRNA conjugate (hereafter referred to as CpG-siSTAT3) as our proof-of-concept molecule due to its clinical relevance and approved IND status.⁸ The unformulated CpG-siSTAT3 conjugates demonstrated promising antitumor efficacy against various preclinical models of both hematological malignancies and solid tumors.^10,18 Although the conjugate’s Dicer-substrate siRNA design enhanced gene silencing, it also resulted in greater sensitivity to serum endonucleases, limiting its serum stability and therapeutic potency.^19,20 We investigated whether encapsulation of CpG-siSTAT3 within myeloid cell-targeted LNPs would preserve its dual immunostimulatory and gene-silencing functions while enhancing its antitumoral immunity. This approach could potentially eliminate the need for extensive chemical modifications that may compromise the immunostimulatory or gene-regulatory properties of oligonucleotide therapeutics.

Results

LNP selection for the delivery of dual-function CpG-siSTAT3

We selected a standard four-component LNP formulation that includes the D-Lin-MC3-DMA (MC3) ionizing lipid, fully hydrogenated soy phosphatidylcholine (HSPC),²¹ cholesterol, and 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG2000),¹³ as a potential vehicle for the delivery of dual-function CpG-siSTAT3 to target DLBCL cells and myeloid cells. First, we tested LNP formulations with different N-to-P ratios (amines in ionizable lipid to phosphates in the CpG-siSTAT3 ratio) and reduced percentages of DMG-PEG2000, given that higher PEG content can reduce target cell uptake (Table S1).^22,23 The LNPs were assembled using a microfluidic system with a controlled flow rate (9 mL/min) and then characterized in terms of their morphology, size (dynamic light scattering and nanosight tracking), and encapsulation efficiency (>90%). All three parameters suggested that the generated LNPs were monodispersed, intact lipid complexes (Figures 1A, 1B, and S1A) with an intact oligonucleotide cargo (Figure S1B). In general, the LNP3 and LNP4 formulations with lower PEG content (≤0.5%) formed larger-sized particles (>100 nm vs. 60–80 nm).

Figure 1.

Functional validation of dual moieties from LNP-encapsulated CpG-siSTAT3

(A) Transmission electron microscopy images (negative staining) of LNP2 at 21,000× (scale bars, 200 nm). (B) Median hydrodynamic size evaluation of LNP1–4 (nanoparticle tracking analysis) and encapsulation efficiency determined through Ribogreen fluorescence-based assay (n = 2–5, shown are mean ± SEM). (C and D) Human PBMC secretion of IFN-α and IL-6. hPBMC were incubated with CpG-ODN or LNP1–4 (CpG-siSTAT3) at 100, 200, or 500 nM (500 nM only for IL-6) for 48 h. Secreted cytokines in the supernatant were measured/quantified using ELISA (n = 3; shown are mean ± SEM). (E) OCI-Ly3 cells were treated with 100 nM of LNP1–4 encapsulating CpG-siSTAT3 or siSTAT3 for 72 h, and protein lysates were analyzed using western blotting with β-actin as a loading control. Levels of total STAT3 were digitally quantified, normalized to β-actin, and shown as a ratio relative to the untreated sample. Summarized results are shown on the right (n = 3, mean ± SEM). (F) In vitro cytotoxic effect of LNP2(CpG-siSTAT3) in OCI-Ly3, U2946, and OCI-Ly18 after 72 h incubation (n = 3, means ± SD). (G) OCI-Ly3 cells were treated with 50 nM of indicated LNP treatments for 72 h, and cell apoptosis was analyzed via flow cytometry (Annexin V and Aqua staining). Representative flow cytometry quadrant plots show the percentage of OCI-Ly3 cells that were live, apoptotic, dead, or necrotic after 72 h of indicated treatments (50 nM). (H) Summarized bar graph showing the proportion of live cells (green), dead (gray), apoptotic (black), or necrotic (light gray) among different treatment groups. Student’s t tests were used to analyze significance between groups in data presented in this figure (n = 3),∗p< 0.05, ∗∗p < 0.005, ∗∗∗p < 0.0005.

Next, we assessed the immunostimulatory properties of LNP-encapsulated CpG-siSTAT3 and siSTAT3 compared to naked CpG-ODN. All four LNP(CpG-siSTAT3) formulations stimulated production of IFNα (Figure 1C) and IL-6 (Figure 1D) in cultured human peripheral blood mononuclear cells (PBMCs) at levels comparable to or higher than the unformulated CpG-ODN used as a positive control. Minimal to no detectable cytokine levels were induced by the LNPs containing only siSTAT3 without CpG-ODN, used as a negative control (Figure S1C). Nuclear factor κB (NF-kB) reporter assay in macrophages revealed higher immunogenicity of LNP3 and LNP4, potentially due to increased sizes and potential formulation instability (Figure S1D). To verify the gene-silencing potential of the various formulations, we tested LNPs 1–4 encapsulating siSTAT3 and CpG-siSTAT3 in multiple cancer cell lines. As shown in Figure 1E, at 100 nM, LNP1 and notably LNP2 were more effective in silencing STAT3 protein levels in OCI-Ly3 B cell lymphoma cells compared to LNP3-4 with lower PEG content. In addition, siSTAT3 led to more pronounced target silencing than CpG-siSTAT3 conjugates when delivered in LNPs (though only statistically significant in LNP4), likely due to the size of the conjugate (Figures 1E and S1E).¹⁹ Finally, we confirmed that LNP1 and LNP2 encapsulated CpG-siSTAT3 formulations remain functionally active during long-term storage. The efficacy of STAT3 silencing in human OCI-Ly3 B cell lymphoma cells (Figure S1F) or U251 glioma cells (Figure S1G) did not significantly differ between LNP (CpG-siSTAT3) used fresh and those stored for 3 months at 4°C. The LNP2 formulation was consistently superior to the other tested LNP variants and thus was selected for further studies. Next, we verified effective STAT3 silencing and cell growth inhibition in a panel of three human DLBCL cell lines—OCI-Ly3, OCI-Ly18, and U2946—treated with CpG-siSTAT3 formulated in LNP2 (Figures 1F and S1H). The conjugate with a single 2′-O-methyl modification of the siRNA was consistently more effective than more extensively chemically modified CpG-siSTAT3 variants (Figure S1H). In addition, LNP2 (CpG-siSTAT3) treatment induced growth inhibition and apoptosis (Figures 1G and 1H) in the OCI-Ly3 lymphoma cells.

Finally, the therapeutic efficacy of LNP-encapsulated CpG-siSTAT3 was assessed in two B cell lymphoma models in immunodeficient mice. Intratumoral (i.t.) treatments of OCI-Ly3 and U2946 tumors using LNP2 (CpG-siSTAT3) significantly inhibited growth of both lymphoma models, whereas naked CpG-siSTAT3 (Figures 2A and 2B) or LNP2 (scrRNA-1) treatments did not show any significant antitumor activity (Figure S1I). Correspondingly, LNP2 (CpG-siSTAT3), but not control LNP2 (CpG-scrRNA), reduced target STAT3 protein expression in whole tumors (Figures 2C and 2D) while augmenting apoptosis within the treated tumors (Figures 2E and 2F). Overall, we found that the LNP2 formulation preserves the dual functionality of CpG-siSTAT3 in vitro while augmenting the potency of the oligonucleotide in vivo.

Figure 2.

Intratumoral treatment of low-dose LNP2(CpG-siSTAT3) induces human lymphoma tumor growth arrest

(A and B) 10⁷ indicated tumor cells were implanted s.c. in 6–8-week-old NSG mice. After tumors reached 100 mm³, 0.5 mg/kg of the indicated treatments was administered i.t. every other day for a total of 6 doses. Tumor volumes were monitored using a caliper (n = 5–7, shown are mean ± SEM; tumor volume: equal to length∗width∗height). Two-way ANOVA was used to analyze significance. Data shown are representative of two independent experiments. For short-term target gene and downstream analysis, treatments were given i.t. every other day for three doses; OCI-Ly3 tumors were collected and processed 24 h after the last treatment. (C) Total STAT3 protein in tumors were analyzed using western blot. (D) Quantified results showing STAT3 silencing in vivo (n = 3, mean ± SEM). (E) Tumor cell apoptosis was analyzed using flow cytometry (Annexin V and Aqua staining). Representative contour plots show induced apoptosis. (F) Quantified apoptosis results from all tumor samples (n = 5, mean ± SEM). Statistical significance was determined using Student’s t test, ∗p < 0.05, ∗∗p < 0.005, ∗∗∗p < 0.0005.

Optimization of LNP2 formulation to enhance targeting of myeloid cells and B cells

Previous studies suggested that incorporating anionic lipids into LNP formulations can enhance tropism to extrahepatic target organs such as lymphoid tissues and potentially improve myeloid cell targeting.^24,25 To optimize our formulations for enhanced myeloid targeting, we modified the LNP2 formulation by incorporating different molar ratios of anionic helper lipids, such as cholesterol hemisuccinate (CHEMS) or 1,2-dipalmitoyl-sn-glycero-3-phosphoglycerol (DPPG) (Table S1). Among the four variants tested, the LNP2.1 (10% DPPG) and LNP2.2 (20% CHEMS) formulations retained the stability of the original LNP2, with a detectable reduction in LNP2.1 size compared to LNP2 or LNP2.2 (Table 1; Figure S2A). Compared to LNP2, both LNP2.1 and LNP2.2 exhibited reduced surface charge (zeta potential), which could potentially promote scavenger receptor (SR)-mediated uptake by myeloid cells and certain cancer cells.^25,26 Thus, we evaluated the uptake of fluorescently labeled CpG-siSTAT3^Cy3 in various LNP formulations by PBMCs from healthy donors (Figures 3A, 3B, and S2B). Interestingly, the LNP2.1 and LNP2.2 formulations differed clearly in their patterns of immune cell-mediated uptake. Compared to LNP2, LNP2.1 improved CpG-siRNA delivery to human monocytes and, to a lesser extent, DCs and B cells (Figures 3A and 3B). In contrast, LNP2.2 preferentially targeted B cells with minimal uptake by myeloid cells (monocytes), except for myeloid DCs (mDCs) (Figures 3A and 3B). Although all three LNP formulations effectively targeted human and mouse myeloid leukemia and B cell lymphoma cells, the selectivity of uptake was less consistent with the myeloid or B cell cancer cell origin (Figure S3C). Next, we verified that selective uptake of primary human CD14⁺ monocytes by LNP2 and LNP2.1 effectively reduced STAT3 activity, as measured by decreased Y705-phosphorylated STAT3 (pSTAT3) (Figures 3C and 3D). The inhibitory effect of LNP2.1(CpG-siSTAT3) was superior to that of LNP2(CpG-siSTAT3), consistent with the improved internalization of the LNP2.1 formulation. Neither LNP reduced activated STAT3 levels in nonmonocytic CD14^– cells, consistent with their limited uptake (Figure 3E). In addition, we verified that CpG-siSTAT3 delivered using LNP2.1 showed similar stability and immunostimulatory potential but enhanced STAT3 knockdown potency compared to the LNP2 formulation (Figures S2D–S2G). Additionally, LNP2.1 did not reduce monocyte viability, as the percentage of CD14⁺ cells among total human PBMCs did not change significantly after 48 h of incubation (Figure S3).

Table 1.

Characterizations of the LNP formulations

	DLS (nm)	Zeta (mV)
LNP2	99.7 ± 1.1	−5.64 ± 0.82
LNP2.1	74.4 ± 0.4	−30.23 ± 1.21
LNP2.2	99.5 ± 0.4	−23.59 ± 1.25

Hydrodynamic size and charge analysis of LNP2, LNP2.1, and LNP2.2 encapsulating CpG-siSTAT3, measured by dynamic light scattering (DLS) and zeta potential (shown as mean ± SD).

Figure 3.

Enhanced myeloid cell-selective uptake when DPPG is incorporated into LNP2 formulation

(A and B) Healthy human PBMCs were incubated with the indicated LNPs encapsulating CpG-siSTAT3^Cy3 at 100 nM for 4 h. Uptake by immune cell subtypes was analyzed using flow cytometry (monocytes, CD14⁺; pDC, CD303a⁺; mDC, CD1c⁺; B cells, CD19⁺; T cells, CD3⁺). Shown are representative results from two independent experiments. (A) Summarized mean fluorescent intensity value (n = 3; mean ± SEM). (B) Representative overlaid histogram showing different uptake patterns of LNP2, LNP2.1, and LNP2.2 across various immune subtypes. (C–E) Freshly isolated healthy human PBMCs were incubated with indicated treatments at 100 nM for 48 h, followed by a 15-min human IL-6 (20 ng/mL) stimulation with immediate fixation. Activated STAT3 (pSTAT3%) was analyzed in CD14⁺ monocytes. (C) Representative histogram overlay of untreated (black open histogram), stimulated (gray filled histogram), and LNP-treated (yellow/red open histogram) cells. (D) CD14+ monocytic population. (E) CD14⁻ nonmonocytic population using flow cytometry (n = 3, shown are mean ± SEM). Statistical significance was determined using Student’s t test, ∗p < 0.05, ∗∗p < 0.005, ∗∗∗ p < 0.0005.

LNP2.1 formulation improved the antitumor activity of CpG-siSTAT3 against syngeneic B cell lymphoma in mice

Subcutaneous (s.c.) injection near the tumor (hereafter referred to as s.c. administration) serves as a local delivery method since nanoparticle-sized particulates mainly pass through the lymphatic system, which is essential for antigen recognition and antitumoral immunity.^27,28 We set out to evaluate the biodistribution of locally administered LNP2- and LNP2.1-formulated CpG-siSTAT3 in a syngeneic B cell lymphoma model. DiD-labeled LNP formulations were administered s.c. in A20-bearing BALB/c mice. Within 2 h, we observed accumulation of LNP2 and LNP2.1(CpG-siSTAT3) within the tumor-draining lymph nodes (TDLNs), specifically in subcapsular sinus macrophages (sMAC), mDCs, and freshly recruited monocytes (Figures 4A, 4B, and S4A). These results support the feasibility of using the LNP2.1 formulation for local delivery of immunotherapeutic oligonucleotides such as CpG-siSTAT3 and demonstrate enhanced selectivity of myeloid cells for LNP2.1 compared to the LNP2 formulation.

Figure 4.

LNP2.1 (CpG-siSTAT3) induces local and abscopal tumor regression

10⁷ A20 cells were implanted subcutaneously (s.c.) in 6–8-week-old BALB/C mice. After tumors reached 100 mm³, indicated DiD-labeled LNPs encapsulating CpG-siSTAT3 (0.5 mg/kg) were administered s.c. near the tumor site. After 2 h, draining lymph nodes were collected, single-cell suspensions were generated, and cellular uptake of LNPs in vivo was analyzed using flow cytometry (subcapsular sinus macrophages [sMAC], CD11b⁺/CD169⁺; monocyte, CD11b⁺/Ly6C^hi; myeloid DCs (mDC), CD11b⁺/CD11c⁺). (A) Representative flow histogram showing uptake of LNP2.1 in various immune subtypes compared to control. (B) Summarized results for LNP2 and LNP2.1 (n = 4, mean ± SEM). Statistical significance was analyzed using Student’s t tests. (C) Indicated treatments (0.5 mg/kg of encapsulated ODN) were given s.c. every other day after tumor sizes reached 100 mm³, and tumor volumes were monitored using caliper measurements (tumor volume: length∗width∗height). Shown are summarized tumor growth curves for each individual treatment (n = 5–6, mean ± SEM). (D) Tumor weights were measured at the end of the study (n = 5–6, mean ± SEM). (E and F) A20 tumors were implanted on both sides of the flank; after tumors reached 100 mm³, treatments (0.5 mg/kg of encapsulated ODN) were given i.t. at one tumor site (E) (treated side) every other day. Tumor volumes at the treated and untreated sides (F) were measured using calipers (n = 6, mean ± SEM). Two-way ANOVA was performed to analyze statistical significance, ∗p < 0.05, ∗∗p < 0.005, ∗∗∗p < 0.0005.

Next, we assessed whether efficient myeloid cell-targeting would promote the antitumor activity of LNP-formulated CpG-siSTAT3. Both s.c. administrations of LNP2- and LNP2.1-encapsulated CpG-siSTAT3 in A20 lymphoma-bearing mice resulted in significant tumor growth inhibition, although the effect of the LNP2.1 formulation was more pronounced (Figures 4C and 4D). We did not observe any activity with the control LNP2.1(scrRNA-1) or with naked CpG-siSTAT3 (due to 10× lower than optimal dosing), and the treatments were well tolerated (Figure S4B). Since CpG-siSTAT3 showed similar dose-dependent direct cytotoxicity in A20 lymphoma cells in vitro (Figure S4C), the improved efficacy of LNP2.1 in vivo may at least partly depend on enhanced uptake by lymphoma-associated myeloid cells and their contribution to antitumor immune responses. To assess the potential for generating abscopal effects against distant tumors, we used a dual-tumor model of A20 lymphoma engrafted in both flanks of the mice. As shown in Figure 4E, i.t. injections of LNP2.1(CpG-siSTAT3) and naked CpG-siSTAT3 induced complete tumor regressions at the directly treated tumor sites because of the sensitivity of A20 tumor cells to CpG-siSTAT3 (Figure S4C). However, only LNP2.1(CpG-siSTAT3) controlled A20 lymphoma progression at the distant, noninjected site (Figure 4F). These results suggest that the LNP2.1 formulation can better engage systemic antitumoral effects of CpG-siSTAT3s, likely mediated through activation of T cell-mediated immune responses.

To determine whether these formulations can also facilitate systemic CpG-siSTAT3 administration, we evaluated the biodistribution of intravenously (i.v.) injected LNP2, LNP2.1, and LNP2.2 formulations. As expected, all tested formulations localized primarily to the liver, followed by the spleen and bone marrow (Figure S5AB). At the cellular level, both LNP2 and LNP2.1 efficiently targeted myeloid cells within the spleen and bone marrow, but LNP2.1 excelled in targeting macrophages and DC subsets in peripheral lymph nodes (Figure S5C). These results underscore the potential of using the LNP2.1 formulation for both local and systemic in vivo siRNA delivery.

Local administration of LNP2.1(CpG-siSTAT3) resulted in tumor recruitment and activation of cytotoxic T cells

We next investigated the mechanism underlying the improved in vivo activity of LNP2.1(CpG-siSTAT3), specifically the proinflammatory effects on the maturation of DCs and effector T cell activity. A single s.c. injection of LNP2.1(CpG-siSTAT3), but not LNP2.1(scrRNA-1), induced maturation of the TDLNs’ cDC1 (CD8a⁺, CD11c⁺) and mDCs (CD11b⁺, CD11c⁺), as the TDLNs serve as the primary draining site (Figures 5A, 5B, and S4A). This is evidenced by significant elevations in major histocompatibility complex (MHC) class II complexes with CD80 costimulatory molecules, which are essential for successful activation of CD8⁺ T cell.^29,30 In addition, the treatment led to rapid activation of T cells, as indicated by elevation of the early activation marker CD69 in TDLNs (Figures 5C and 5D) and in tumors (Figure 5E). Repeated treatments augmented the percentages of activated DCs in tumors, while the effects of control LNP2.1(scrRNA-1) treatments were limited (Figure 5F). More detailed intracellular staining studies confirmed the elevated expression of cytokines and mediators essential for the immunostimulatory (IFNɣ⁺ or IFNɣ⁺TNFα⁺) functions of effector T cells (Figure 5G). Overall, these results underscore the feasibility of using the LNP2.1 formulation to boost the potency of CpG-siSTAT3 as immunotherapy in B cell lymphoma.

Figure 5.

LNP2.1(CpG-siSTAT3) primes the tumor-draining lymph nodes to enable T cell-mediated tumor control

10⁷ A20 cells were implanted s.c. in 6–8-week-old BALB/c mice. After tumors reached 100 mm³, indicated LNPs encapsulating 0.5 mg/kg ODN were administered s.c. After 24 h, tumors and tumor-draining lymph nodes (TDLN) were collected and processed for immune cell activation using flow cytometry. (A) Representative flow contour plots showing dendritic cell activation (cDC1) in TDLNs. cDC1 (CD11c⁺/CD8a⁺) and myeloid DCs (CD11b⁺/CD11c⁺) were analyzed; activation was determined by co-expression of CD80 and MHC class II complexes. (B) Summarized bar graph showing activation of myeloid DCs and cDC1 (% CD80⁺, MHCII⁺ on viable DCs) in TDLNs. (C) Representative flow contour plot showing percentage of CD8 T cell activation (CD69⁺) in TDLNs. (D-E) Summarized bar graphs showing percentage of T cell activation at indicated locations: T helper cells (CD3⁺/CD4⁺/FoxP3) and cytotoxic T cells (CD3⁺/CD8⁺) (n = 5, shown are mean ± SEM). (F) Long-term treatment effects were analyzed 24 h after 4 s.c. treatments. Tumors and TDLNs were collected and analyzed by flow cytometry. Activated DCs (CD11b⁺/CD11c⁺/CD80⁺/MHC class II complexes⁺) were analyzed in tumors. (G) Following tumor cell suspension preparation after long-term treatments, single-cell suspensions generated from tumors were plated at 2 million/200 μL and stimulated with phorbol myristate acetate/ionomycin (w/GOLGI plug) for 6 h and the percentage of CD8 T cell cytokine secretion was analyzed using flow cytometry. Shown is the total percentage of CD8 T cells secreting single cytokine IFNγ+, TNFα+, or both cytokines IFNγ+ and TNFα+ (n = 5–6, shown are mean ± SEM). Student’s t tests were used to assess statistical significance,∗p < 0.05, ∗∗p < 0.005, ∗∗∗p < 0.0005.

Discussion

In this study, we successfully developed an LNP formulation optimized for targeted delivery to myeloid immune cells, including DCs and macrophage subsets, suitable for application in cancer immunotherapy. The LNP2.1 formulation was used for delivery of the clinically relevant, immunostimulatory CpG-siSTAT3 oligonucleotide. Encapsulation of CpG-siSTAT3 demonstrated remarkable efficacy in directly inhibiting human DLBCL progression in immunodeficient mice while simultaneously generating potent immune-mediated antitumor responses in immunocompetent models.

A key innovation in our approach was the co-delivery of two therapeutic oligonucleotide moieties, CpG conjugated to siSTAT3, within a single LNP system. Briefly, the exact loading quantification for each component, if done separately, may require more sophisticated analysis such as thermogravimetric analysis or HPLC-MS.³¹ Encapsulation of the conjugated form ensured a precise 1:1 ratio of the therapeutic components, addressing a significant challenge in the field, where few studies have successfully encapsulated both agents within nanoparticle-based delivery systems.^31,32,33,34 In addition, this dual-action approach also accommodated the distinct subcellular targeting requirements: CpG-induced TLR9 activation occurs in endosomes,³⁵ while siSTAT3 requires cytosolic delivery to achieve gene silencing.³⁶

The physicochemical optimization of our LNP formulation revealed important insights about structure-function relationships. Although modulating PEG content did not significantly impact IFNα release from plasmacytoid DCs (pDCs), it substantially influenced the nanoparticles’ physicochemical properties and gene-silencing efficiency. This is consistent with the known role of PEG content in the lipid organization of LNP structure.^37,38 However, reductions in PEG content did not influence the functional activity of the CpG component when tested in vitro in our studies.^22,23 A significant advantage of our approach lies in addressing a fundamental challenge in CpG-based immunotherapy: the tetrameric structure formed by CpG-ODN drives strong type I IFN cytokine release to augment favorable antitumoral immune responses in recent clinical trials.^39,40 However, these CpG molecules have limited stability and reduced preferential uptake by B cells due to their nanoparticulate structures.⁴¹ Our LNP2.1 formulation effectively overcomes this limitation by enabling efficient delivery of CpG-ODN to the appropriate cellular compartments, thereby addressing the inherent instability of CpG-ODN.⁴² Notably, encapsulation of CpG-siSTAT3 within LNPs preserves a favorable CpG-ODN nanostructure that prouces substantially more robust type I IFN induction compared with unencapsulated oligonucleotides, confirming the importance of maintaining CpG secondary structures for effective pDC activation in early endosomes.⁴³ This enhanced interferon response represents a critical advancement for generating effective antitumor immunity while simultaneously silencing STAT3-mediated immunosuppression.⁴⁴

Targeted delivery of therapeutic oligonucleotides to extrahepatic tissues remains a major challenge for delivery systems. Recent studies in mRNA delivery have shown that altering ionizable or helper lipids can redirect cargo expression to other organs, particularly the lung or spleen.^24,45 In our formulations, the incorporation of anionic helper lipids (LNP2.1) is likely to enhance steric hindrance, thereby reducing nonspecific interactions while promoting uptake by myeloid cells through engagement of SRs. SR-A and SR-B1 are well-established SRs expressed on myeloid cells,^46,47 which are professional phagocytes.^48,49 B cells, by contrast, are traditionally considered nonphagocytic and uptake mainly through the B cell receptor (BCR) in an antigen-specific manner.⁵⁰ However, a small subset of circulating B cells also has the capacity to phagocytose.⁵¹ Nevertheless, the reduced uptake of LNP2.2 by monocytes suggests an opportunity for indications that require minimal myeloid uptake. The incorporation of a specific percentage of DPPG (10% mol/mol vs. 5% mol/mol) enhanced uptake by human monocytes and pDCs but not by CHEMS or neutral-charged HSPC, also suggesting that lipid arrangement (core versus surface) can significantly impact cellular tropism.⁵² Our comparative analysis of LNP2.1 and LNP2.2, DPPG- and CHEMS-modified, respectively, revealed distinct cellular uptake patterns across immune cell populations that have not been previously reported. These observations highlight the potential for fine-tuning formulations for specific cellular targets.

While the self-adjuvant effects of LNPs are well documented, it is important to note that such effects are less likely to boost antitumoral immune responses against human cancers despite their potently immunosuppressive TME.^3,53,54 The dual-functional CpG-siSTAT3 can simultaneously induce immune stimulation and remove tolerogenic STAT3 activation in the B cell lymphoma cells and in the TME.^{10,11,17,18,55} Although the lead LNP formulations in our study were well tolerated, further safety studies will assess in greater detail, for example, the potential immunostimulatory and/or immunotoxic LNP effects.⁵⁶

In the present study, we focused on local LNP administration to promote delivery of the immunostimulatory oligonucleotide to the TDLNs while minimizing liver toxicity.²⁷ LNP2.1(CpG-siSTAT3) injected s.c. significantly outperformed the benchmark formulation (LNP2), leading to enhanced tumor growth inhibition and increased apoptosis within treated lymphomas. The negative charge characteristics of LNP2.1 also likely contributed to improved retention in the TDLNs, enhancing therapeutic efficacy.^57,58,59 Critically, this formulation not only enabled uptake by lymphoma cells but also activated antigen-presenting cells (APCs), including the cDC1 subset, which is important for the recruitment and activation of effector CD8⁺ T cells into the tumor.²⁹ Simultaneously delivered immunostimulatory agents, such as CpG-siSTAT3, silence oncogenic pathways in tumor cells while boosting the activity of APCs and T cells, augmenting therapeutic efficacy. In conclusion, the encapsulation of potently immunostimulatory CpG-siSTAT3 in LNP2.1 can provide a comprehensive strategy to improve outcomes for patients with B cell lymphoma and potentially other malignancies where myeloid cell targeting offers a crucial therapeutic benefit.

Materials and methods

Cell culture

PBMCs from anonymous healthy donors were collected in accordance with the Declaration of Helsinki under the institutional review board (IRB) protocol 13378 (City of Hope; COH). Human OCI-Ly3 cells were from DSMZ; U2932 cells were kindly provided by Dr. G. Inghirami (Weill Cornell Medicine, NY, USA); and OCI-Ly18, U2946, and mouse A20 cells were kindly provided by Dr. Larry Kwak (City of Hope, CA). OCI-Ly3 cells were cultured in RPMI 1640 supplemented with 20% fetal bovine serum (FBS), whereas the other human cell lines were cultured in RPMI 1640 with 10% FBS. Human leukemia cell line KG1a was originally purchased from the American Type Culture Collection (ATCC) and cultured in Iscove's Modified Dulbecco's Medium (IMDM) with 20% FBS. Mouse BV2, RAW264.7, and RAW BLUE cells were originally purchased from ATCC and maintained in DMEM with 10% FBS.

Mice

All animal experiments were conducted in accordance with institutional guidelines and an approved protocol from the Institutional Animal Care and Use Committee. NOD/SCID/IL-2RgKO(NSG) were originally purchased from the Jackson Laboratory and bred at COH. C57/B6 and BALB/c mice (6–8 weeks old) were purchased from the Jackson Laboratory.

Oligonucleotide sequences

The oligonucleotides were synthesized by the DNA/RNA Synthesis Core Laboratory at City of Hope as described previously.¹⁸ The sequence names used in this study are as follows:

CpG-hSTAT3siRNA:

5′ G∗G∗TGCATCGATGCAGG∗G∗G∗G∗GxxxxxUCAGUCGUAUCUUUCUGCAGCUUCCmGU 3′

dAdGUCAGCAUAGAAAGACGUCGAAGG.

CpG-ScrRNA:

5′ G∗G∗TGCATCGATGCAGG∗G∗G∗G∗GxxxxxAUUUAGCCUUAAUACACGCCmAA 3′ UCUAAAUCGGAAUUAUGUGCGG.

CpG-mSTAT3siRNA:

5′ G∗G∗TGCATCGATGCAGG∗G∗G∗G∗GxxxxxUUAGCCCAUGUGAUCUGACACCCUGmAA 3′

dAdAUCGGGUACACUAGACUGUGGGAC.

GpC-hSTAT3siRNA (Scr-RNA-2):

5′ G∗G∗TGCATGCATGCAGG∗G∗G∗G∗GxxxxxUCAGUCGUAUCUUUCUGCAGCUUCCmGU 3′

dAdGUCAGCAUAGAAAGACGUCGAAGG.

ScrRNA-1:

UCUAAAUCGGAAUUAUGUGCGG.

Preparation and characterization of LNP formulations

Fully HSPC, 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), DPPG, and DMG-PEG2000 were purchased from Avanti. Cholesterol and cholesteryl succinate were purchased from Sigma, and MC-3 was purchased from MedChem. LNPs were prepared using a microfluidic system (Precision Nanosystem). Briefly, pre-dissolved lipids (in ethanol) were mixed according to the lipid composition listed in Table 1. Oligonucleotides were diluted in 50 mM citrate buffer (Fisher Scientific). The lipid phase and aqueous phase were mixed at a ratio of 1:3 with a total flow rate of 9 mL/min. Upon production, LNPs were immediately mixed with sterile 1× PBS and kept on ice until purification. Diluted LNPs (∼15 mL) were transferred to an Amicon Ultra-15 centrifugal filter unit with a 10 kDa cutoff (EMD Millipore) and centrifuged at 2,000×g for 20–30 min. The concentrated LNPs were then diluted with ∼15 mL PBS and purified again. The final LNPs were stored at 4°C.

Each batch of LNPs was characterized by dynamic light scattering (DLS) or nanosight tracking analysis (NTA), and zeta potential was acquired using a Brookhaven 90 Plus/BI-MAS instrument (Brookhaven Instruments). Samples (20 μL) were diluted with PBS (180 μL), and DLS measurements were obtained by performing 5 runs at 30 s per run. Samples (20 μL) were diluted with 10 mM NaCl, and zeta potential measurements were obtained by performing 10 runs at 30 cycles per run.

To quantify or verify the loading of oligonucleotides into LNPs, 10 μL of prepared LNPs were mixed with 10 μL of 2% Triton X-100 and incubated at 37°C for at least 15 min. Following that, the mixture was diluted 10-fold with H₂O and further diluted with TE buffer. Standard curves were prepared using known-concentration oligo stocks following the exact dilution scheme. General procedures were carried out according to the Quanti-iT RiboGreen RNA assay kit manufacturer’s protocols.

In parallel, approximately 1 μg of oligo-loaded LNPs was lysed with Triton X or diluted with PBS at a 1:1 ratio. Intact LNPs were mixed with DNA loading buffer, lysed LNPs were mixed with 2× RNA loading buffer, and oligonucleotides were resolved by gel electrophoresis and imaged using the ChemiDoc system (Bio-Rad).

Electron microscopy

The LNPs were placed onto 300-mesh carbon-formvar-coated grids and allowed to absorb into the formvar for a minimum of 1 min. The grids were rinsed with double-distilled water and stained for contrast using 1% uranyl acetate. The samples were viewed with an FEI Tecnai T12 transmission electron microscope at 120 keV, and images were taken with a Gatan Ultrascan 2K charge-coupled device (CCD) camera.

Human peripheral mononuclear cell assays

Healthy donor whole blood was layered with an equal volume of room-temperature Histopaque 1077 (Sigma-Aldrich) and centrifuged at 1,500 rpm for 20 min with 0 brake and full acceleration. The middle (mononuclear) layer was washed twice with PBS and counted for further use. For in vitro uptake assays, ∼2e⁶ cells were plated in a 96-well plate with 200 μL media (RPMI with 10% FBS) and incubated with fluorescently labeled oligo or oligo-loaded LNPs at 37°C for 4 h. The cells were then centrifuged and washed with PBS for further analysis. For in vitro release of cytokines, ∼2e⁶ cells were plated in a round-bottom 96-well plate with 400 μL media, and the outer layer of the 96-well plate was loaded with excess PBS to prevent evaporation of the cell culture media. hPBMCs were incubated at 37°C for 48 h, and the cells were centrifuged at ×400g for 5 min. Supernatants were immediately aliquoted and froze at −80°C or analyzed following the human IFNα and IL-6 ELISA (Invitrogen) protocols for further analysis. For in vitro analysis of STAT3 silencing in human monocytes, ∼2e⁶ cells were incubated with the indicated LNPs for 24–48 h, followed by direct stimulation with 20 ng/mL human IL-6 (R&D Systems) added to the cultured media and mixed well. After 15 min of stimulation, the cells were immediately fixed with 2% paraformaldehyde (PFA) (final concentration) at 37°C for 20 min. The cells were then transferred to flow cytometry tubes for further analysis.

Cell viability assays

Cells were counted and plated at a seeding density of 10,000 cells/200 μL in 96-well TC-treated plates. For suspension cells, treatments were pre-diluted with culture media and added to the cell culture. For adherent cells, the cells were allowed to attach overnight, and the supernatants were replaced with treatment-containing media. After 24 to 72 h, 10× cell counting kit-8 (Donjindo) detection solution was added directly to each well, with media-only wells serving as the blank control. The cells were then incubated at 37°C for 1 to 3 h, and absorbance was read at 450 nm using a plate reader (BioTek Cytation 3, Aglient).

RAW-blue reporter assay

RAW-Blue cells were plated at 5,000 cells per well in a 96-well plate. After the cells were completely attached, the media was replaced with 200 μL of the desired treatment reagents (100 ng/mL lipopolysaccharide (LPS) as the positive control) at various concentrations. After 24–48 h, 20 μL of treatment supernatants were added to 180 μL of QuantiBlue reagent (InvivoGen) and incubated at 37°C for 45 min, and absorbance was read at 625 nm using a plate reader (BioTek Cytation 3, Aligent). NF-kB activation was calculated relative to untreated wells.

Western blot

Cellular levels of target proteins were analyzed using western blotting as previously described.¹⁰ STAT3(124H6) from Cell Signaling Technology (1:3000), IRAK1 (D51G7) from Cell Signaling Technology (1:3000), TRAF6 (EP592) from Abcam (1:3000), and β-actin-Horseradish Peroxidase (HRP) (1:100,000) from Millipore Sigma were incubated with the blots overnight in 2.5% milk at 4°C. Mouse-specific IgG (GE Healthcare) or rabbit-specific IgG (Abcam) conjugated to HRP was used, HRP signals were detected using SuperSignal West Femto Maximum Sensitivity Substrate (Thermo Fisher Scientific), and the blots were imaged using a ChemiDoc imaging system. Target protein bands were quantified using Image Lab Software 6.0.1 (Bio-Rad).

Flow cytometry

For in vitro assays, cells were washed once with PBS and incubated with Aqua (Invivogen) at a 1:100 ratio in PBS for 20 min. For in vivo samples (i.e., biodistribution and immunophenotyping analysis), tumor, spleen, or lymph node tissues were digested with collagenase D and DNAse I for 10–20 min (depending on tissue type) and passed through a 70 μm strainer using the back of a syringe plunger. Red blood cells were lysed with 1× ACK lysis buffer for 5–10 min and washed with PBS. Prepared single-cell suspensions were first stained with LIVE/DEAD Fixable Aqua Dead Cell Stain Kit (Invitrogen) following the manufacturer’s instructions. This was followed by staining of surface markers—hCD14 (M5E2), hCD19 (H1B18), hCD3 (OKT3), hCD303 (Z01A), hCD1c (L161), pSTAT3 (pY705) (Thermo Fisher Scientific)—with FcR block (1:100) (Miltenyi) on ice for 20 min. Stained cells were then washed twice with FACS buffer. For pSTAT3 staining, cells were first fixed with 2% PFA and then permeabilized with True-Phospho Permeabilization Buffer (BioLegend). The cells were subsequently stained with pSTAT3-conjugated antibody (1:20) with blocker for 1 h. For mouse staining panels, h/mCD11b (M1/70), mCD11c (N418), mCD3 (145-2C11), mCD8 (53–6.7), mCD4 (RM4-5), FoxP3 (FJK-16s), mCD163 (10D6), mCD80 (16-10A1), and MHC class II complex (M5/114.15.2) with CD16/32 antibodies were used. For ex vivo T cell stimulation analysis, single-cell suspensions from each individual mouse tumor were counted and plated at a density of 2 × 10ˆ6 cells/200 μL in a round-bottom 96-well plate. Cells were first rested for ∼2 h, and a cell activation cocktail (contains phorbol myristate acetate and ionomycin) without brefeldin A (BioLegend) and Golgi Plug protein transport inhibitor (BD Biosciences) was then added to the culture for 6 h. After live/dead staining, an antibody cocktail containing mCD3, mCD8, mCD4, and CD16/32 in FCS buffer was added to stain for surface markers on ice for 20 min. Stained cells were washed, fixed, and permeabilized using the FoxP3/transcription factor staining buffer set (BD Biosciences). Intracellular cytokines IFNγ (XMG1.2), TNFα (MP6-XT22), and Granzyme B (GB11) were stained for at least 1 h and then washed and resuspended at the appropriate volume for flow cytometry analysis. Results were collected using an Attune flow cytometer (Thermo Fisher Scientific) and analyzed using FlowJo v.10.6 (i.e., example gating strategies are included in Figure S3).

Animal experiments

For therapeutic efficacy analysis in human lymphoma xenografts, NSG mice were implanted s.c. with 10⁷ OCI-Ly3 and U2946 cells. Tumor engraftment and progression were monitored with calipers every other day. Tumor volumes were calculated as length × width × height (mm³). Treatments were given i.t. every other day after the average tumor size reached 100 mm³.

For biodistribution analysis of LNPs administered i.v., 10⁶ Cbfb-MYH11/Mpl+.eGFP (CMM.eGFP) acute myeloid leukemia cells were implanted systemically into C57/B6 mice. 10 days after implantation (>1% CMM signal in blood), 1 mg/kg of DiD-labeled LNP formulations was administered i.v. 3 h later, the liver, spleen, lung, kidney, and peripheral lymph nodes were subjected to fluorescent imaging (Spectral Instruments) with excitation at 605 nm and emission at 690 nm, 20% power, and 5 s of exposure. Blood, spleen, bone marrow, and lymph nodes were subjected to flow cytometry analysis for cellular uptake.

For biodistribution, therapeutic efficacy, or mechanistic experiments in syngeneic mouse models, 10⁷ parental A20 tumor cells were implanted s.c. After tumors reached 100 mm³, animals were grouped based on measured tumor sizes. For biodistribution analysis, 0.5 mg/kg of DiD-lableled LNP(CpG-siSTAT3) were administered, and 2 h later, tumors and TDLNs (inguinal and axillary) were collected for downstream analysis. For efficacy studies, controls and LNPs (0.5 mg/kg ODN) were administered s.c. near the tumor, or i.t., and tumors were monitored with caliper measurements every other day. Immunophenotyping analysis after ×1 treatment, or ×4 treatments, was performed to understand the immune-mediated antitumoral effect.

Statistics

To estimate the statistical significance of differences between two treatment groups, a two-tailed p value was calculated using an unpaired t test. One- or two-way ANOVA with Tukey’s multiple-comparison post-test was applied to assess differences between multiple groups or in tumor growth kinetics experiments. Statistically significant p values are indicated in the figures as follows: ∗∗∗p < 0.0005; ∗∗p< 0.005; and ∗p < 0.05. Data were analyzed using Prism software v.8.1.0 (GraphPad).

Data availability

All reagents and data generated from this study are available from the corresponding author upon reasonable request.

Acknowledgments

This work was supported in part by the National Cancer Institute/National Institutes of Health award numbers R01CA284593 (M.K.) and P30CA033572 (COH) as well as by a sponsored research agreement with Duet Biotherapeutics. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. We are grateful to the staff at the DNA/RNA Synthesis, Analytical Cytometry, Electron Microscopy, Analytical Pharmacology, and Animal Resource Cores (COH). We thank Dr. Tristan Scott for providing materials and access to equipment for formulation optimization and Jeremy Hall for proofreading and providing input on manuscript writing. The graphical abstract was created with BioRender.com.

Author contributions

Conceptualization, M.K. and E.Y.K.; experiment design and execution, E.Y.K., D.K., D.W., and A.D.; investigation, E.Y.K., D.W., D.K., and W.T.; oligo synthesis, P.S.; writing original draft, E.Y.K. and M.K.; writing – review and editing, M.K.; funding acquisition and resources, M.K.; supervision, M.K.

Declaration of interests

M.K. is the inventor of the oligonucleotide design presented in this report. M.K. was a scientific advisor to Duet Biotherapeutics and is currently a co-founder and scientific advisor to Aptadir Therapeutics and Twin Peaks Biotherapeutics, holding stock options.