Targeted Delivery of miR‐34a via Anti‐CD47 Antibody Conjugates for Enhanced Cancer Immunotherapy in Triple Negative Breast Cancer

Abstract

Triple‐negative breast cancer (TNBC) is a highly aggressive subtype of breast cancer that lacks estrogen receptor, progesterone receptor, and HER2 expression, which limits the efficacy of targeted therapies. MicroRNA‐34a‐5p (miR‐34a), a tumor‐suppressor miRNA known for regulating oncogenic pathways, initially appeared promising as a therapeutic avenue. However, the clinical translation of miR‐34a has been hindered by challenges such as poor stability, inefficient cytoplasmic delivery, and immune‐related toxicities, as evidenced by the failure of MRX34 in trials. To address these limitations, this study developes a novel antibody‐oligonucleotide conjugate (AOC) platform anti‐CD47‐miR‐34a conjugate (aCD47‐C‐miR34a). The aCD47‐C‐miR34a system combines the anti‐CD47 antibodies with miR‐34a using a bioreducible linker, ensuring targeted cytoplasmic delivery via CD47‐mediated endocytosis and endosomal escape. CD47, an immune checkpoint protein overexpressed in TNBC, facilitates immune evasion, making it an attractive therapeutic target. In preclinical TNBC models, aCD47‐C‐miR34a successfully restored miR‐34a's tumor‐suppressive functions by downregulating oncogenic pathways including PD‐L1, while modulating the tumor microenvironment. This dual mechanism promoted macrophage phagocytosis, enhanced CD8⁺ T‐cell activation, and induced apoptosis, resulting in significant tumor inhibition without systemic toxicity. These findings demonstrate the transformative potential of aCD47‐C‐miR34a in overcoming TNBC's oncogenic and immune‐evasive mechanisms, paving the way for innovative treatments in TNBC and other heterogeneous, aggressive cancers.

aCD47‐C‐miR34a, an antibody‐oligonucleotide platform combining an anti‐CD47 antibody with miR‐34a, is designed for targeted therapy in TNBC. This approach improved delivery and restored the tumor‐suppressing function of miR‐34a, enhanced CD8⁺ T‐cell activation, and inhibited tumor growth, providing a potential strategy to overcome immune evasion and tumor progression in triple‐negative breast cancer.

1. Introduction

The discovery of microRNAs (miRNAs) has opened new frontiers in cancer therapeutics, offering unique opportunities to target oncogenic pathways through post‐transcriptional regulation of gene networks.^{[ 1 ]} By silencing key genes involved in tumorigenesis, miRNAs act as powerful modulators of cellular processes such as proliferation, apoptosis, and immune evasion. Among these, miRNA‐34a‐5p (miR‐34a) stands out as a well‐characterized tumor‐suppressor that targets a range of oncogenic proteins, including sirtuin 1 (SIRT1), programmed death ligand‐1 (PD‐L1), Bcl‐2, Notch1, and other regulators critical for cell survival and immune escape.^{[ 2 ]} Through these interactions, miR‐34a suppresses tumor growth, promotes apoptosis, and enhances antitumor immune responses, making it a compelling candidate for therapeutic intervention.^{[ 3 ]}

Despite its therapeutic promise, miR‐34a expression is frequently downregulated in triple‐negative breast cancer (TNBC), an aggressive breast cancer subtype associated with poor prognosis.^{[ 4 ]} Reintroducing miR‐34a into TNBC cells holds substantial therapeutic potential but faces substantial challenges. Naked miR‐34a is inherently unstable in biological environments, exhibits poor pharmacokinetics, and risks off‐target effects, which compromise its therapeutic efficacy.^{[ 5 ]} The clinical translation of miR‐34a therapeutics has also faced setbacks. For instance, MRX34, a liposomal formulation of miR‐34a, showed initial promise in preclinical studies but failed during clinical trials due to immune‐related toxicities, off‐target effects, and severe lymphocytopenia.^{[ 6 ]} This failure underscores the critical need for safe and efficient delivery systems tailored to TNBC's unique molecular and pathological landscape.

Unlike other breast cancer subtypes, TNBC lacks the expression of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2), which limits the availability of established molecular targets.^{[ 7 ]} This heterogeneity complicates therapeutic design and highlights the necessity for receptor‐independent or highly selective approaches. Ligand‐directed miRNA delivery, though promising, is constrained by TNBC's limited receptor expression and the critical challenge of endosomal entrapment.^{[ 8 ]} Without effective mechanisms for endosomal escape, internalized miRNAs are at risk of lysosomal degradation, undermining their therapeutic efficacy.^{[ 9 ]} Strategies such as cell‐penetrating peptides, fusogenic agents, and chemical enhancers have shown potential but face translational barriers due to systemic toxicity, emphasizing the need for safer alternatives.^{[ 10 ]}

In this study, we propose antibody‐oligonucleotide conjugates (AOCs) as a novel platform for the targeted delivery of miR‐34a to TNBC cells. AOCs combine the specificity of antibody‐mediated targeting with the therapeutic potential of oligonucleotides, offering a promising solution to address the limitations of conventional miRNA delivery systems.^{[ 11 ]} AOCs are strategically designed to optimize intracellular delivery through key mechanisms such as targeted binding to cancer cells, efficient internalization via receptor‐mediated endocytosis, and precise release of therapeutic oligonucleotides into the cytoplasm. Notably, incorporating cleavable linkers into the AOC design enables controlled release in response to specific intracellular stimuli, such as protease activity or redox changes, thereby enhancing therapeutic efficacy while reducing off‐target effects.^{[ 12 ]} Given these advances, the development of AOCs highlights a paradigm shift in the field of therapeutic materials: integrating biological specificity with material responsiveness to achieve programmable, disease‐targeted delivery. Our AOC platform exemplifies this new generation of smart biomaterials, offering a clinically actionable solution to overcome critical challenges in miRNA‐based therapeutics for TNBC.

To address the challenges of TNBC therapy, we developed an anti‐CD47‐miR‐34a conjugate (aCD47‐C‐miR34a) as a prototype AOC (Figure 1 ). CD47, often referred to as the “don't eat me” signal, is an immune checkpoint protein overexpressed in many cancers, including TNBC, and associated with poor clinical outcomes. Therefore, targeting CD47 offers an advantage that enables selective delivery of therapeutic agents to TNBC cells. Our aCD47‐C‐miR34a consists of an anti‐CD47 antibody (aCD47), a bioreducible (Cleavable, C) linker, and miR‐34a. Incorporating a bioreducible linker, the conjugate facilitates cytoplasmic release of miR‐34a via CD47‐mediated endocytosis and subsequent endosomal escape.

Figure 1.

Strategy and mechanism of action of aCD47‐miR‐34a (Antibody‐Oligonucleotide Conjugate) for cancer immunotherapy. 1) Targeted delivery of aCD47‐C‐miR34a to CD47 overexpressed TNBC cells. 2) The aCD47‐C‐miR34 initiates binding to CD47 on tumor cell surfaces and is internalized via CD47‐mediated endocytosis. 3) In early endosomes, the linker of aCD47‐C‐miR34a is cleaved by GSH, releasing intact miR‐34a. 4) The released miR‐34a directly targets and downregulates SIRT1 and PD‐L1 mRNAs, both of which are associated with tumorigenesis. 5,6) Consequently, apoptotic cell‐induced DC activation promotes CD8⁺ T‐cell infiltration, enhancing antitumor immunity.

The aCD47‐C‐miR34a employs dual mechanisms of action, restoring miR‐34a's tumor‐suppressive functions while modulating the tumor microenvironment (TME). By blocking CD47, the conjugate promotes macrophage‐mediated phagocytosis, while miR‐34a delivery downregulates PD‐L1 expression, enhancing CD8⁺ T‐cell activation. These combined effects mitigate immune escape and induce apoptosis, achieving potent antitumor activity in the preclinical TNBC model without observable systemic toxicity. In addition to addressing TNBC‐specific challenges, this platform also demonstrates the broader potential of ligand‐directed miRNA delivery in precision oncology. Compared to conventional packaged vehicles, the strategy circumvents issues such as non‐specific biodistribution and systemic toxicity.

In summary, the aCD47‐C‐miR34a represents a promising advance in miRNA‐based therapeutics, addressing TNBC's oncogenic and immune‐evasive mechanisms. This work not only highlights the transformative potential of AOCs for TNBC but also paves the way for broader applications of miRNA therapeutics in combating heterogeneous and aggressive cancers.

2. Results and Discussion

2.1. Preparation and Characterization of Antibody‐Oligonucleotide Conjugate (aCD47‐C‐miR34a)

As the first step in synthesizing the aCD47‐miR34a AOC conjugate, the aCD47 was functionalized with an azide group by covalently attaching either N‐hydroxysuccinimide (NHS)‐PEG3‐SS‐Azide, a glutathione (GSH)‐cleavable linker, or NHS‐PEG4‐Azide, a non‐cleavable linker. Both linkers are bifunctional, featuring an NHS‐activated ester at one terminus and an azide group at the other, allowing attachment to primary amines on lysine residues of the antibody. Afterward, azide‐functionalized aCD47 was conjugated with a dibenzocyclooctyne (DBCO)‐functionalized miR‐34a mimic through a strain‐promoted cycloaddition reaction between an azide and DBCO functional groups. Owing to site‐specific and rapid reaction between azide and DBCO, it could be readily prepared a novel AOC platform, aCD47‐miR34a, that could deliver therapeutic miRNAs with high specificity and minimal off‐target effects (Figure 2a; Figure S1, Supporting Information).

Figure 2.

Characterization of aCD47‐C‐miR34a (Antibody‐Oligonucleotide Conjugate) a) Schematic of the Antibody‐Oligonucleotide Conjugate designed with a covalently attached GSH‐cleavable linker (aCD47‐C‐miR34a). b) Representative 12% reducing SDS‐PAGE gel images showing the conjugation of non‐cleavable linker‐containing aCD47 with miR‐34a at different molar ratios. The results are presented for both the heavy chain (left) and the light chain (right) of aCD47. The gels were stained with Coomassie dye (upper panel) to show protein bands and imaged with fluorescence (lower panel) to detect miR‐34a. c) Representative native gel (4‐15%) images showing the conjugation of cleavable linker‐containing aCD47 with miR‐34a at various molar ratios. d) Whole protein mass analysis of aCD47, aCD47‐linker, miR‐34a, and aCD47‐C‐miR34a. The conjugation reaction between aCD47 and miR‐34a was analyzed at a 1:4 molar ratio. MW, molecular weight. e) The DoC was calculated using the relative peak intensities obtained from whole protein mass analysis. The Coomassie‐stained native gel (4–15%) analysis also verified the conjugation of miR‐34a to the aCD47‐linker. f) Representative native gel (4‐15%) images of aCD47‐C‐miR34a, visualized by Coomassie staining. g, h) Cy5 detection on the same gel, with quantitative analysis of fluorescence band intensities to evaluate the release of Cy5‐labeled miR‐34a following DTT reduction.

To determine the optimal aCD47‐miR34a AOC formulation, a range of molar ratios of aCD47 to miR‐34a (1:1, 1:2, 1:4, 1:8, and 1:10) was first assessed using both non‐cleavable (NC) and cleavable (C) linkers. As shown in Figure 2b,c, the fluorescence intensity of miR‐34a increased proportionally with the feed ratio, and the fluorescence signals consistently aligned with protein bands, confirming the successful conjugation of miR‐34a to aCD47. At higher molar ratios (1:8 and 1:10), no additional band shifts or fluorescence changes were observed, indicating saturation of conjugation efficiency. Furthermore, a comparative analysis of various molar ratios revealed that the 1:4 ratio achieved the highest efficiency at the miRNA expression level (Figure S2, Supporting Information). These findings establish the 1:4 molar ratio as the optimal formulation for aCD47‐C‐miR34a, ensuring efficient and effective delivery of miR‐34a. Based on these results, we determined a 1:4 molar ratio between the antibody and miR‐34a to be optimal, which was subsequently utilized in all subsequent experiments. To further validate successful conjugation, intact mass spectrometry analysis detected a mass peak at 147,261 Da for aCD47 and 151,862 Da for the aCD47‐linker, which confirms the efficient attachment of an average of 9‐linker molecules to aCD47. Following conjugation with miR‐34a (8,395 Da), the aCD47‐C‐miR34a exhibited a mass peak shift to 169 554 Da, demonstrating the successful conjugation of miR‐34a to aCD47 (Figure 2d). Moreover, the degree of conjugation (DoC) is a key parameter for bioconjugates, representing the average number of oligonucleotide molecules attached to each antibody. Accordingly, we determined the DoC of aCD47‐C‐miR34a to be 2.01, based on intact mass spectrometry analysis, and confirmed this result through non‐reducing gel electrophoresis (Figure 2e). To assess the stability under physiological conditions, we compared the degradation profiles of the aCD47‐C‐miR‐34a conjugate and free miR‐34a after serum incubation. The conjugate demonstrated markedly improved oligonucleotide stability, retaining ≈72.8% of its conjugated oligonucleotide integrity after 72 h of incubation, in contrast to the complete degradation observed for free miR‐34a (Figure S3, Supporting Information). To further evaluate its stability under physiological enzymatic conditions, we assessed the proteolytic resistance of aCD47‐C‐miR‐34a upon incubation with matrix metalloproteinase‐9 (MMP‐9) and cathepsin B, proteases abundantly present in mouse serum, at concentrations mimicking physiological levels.^{[ 13} , ^{14 ]} Under the same incubation conditions, the conjugate maintained ≈99.8% and 79.8% of its original form after 72 h in the presence of MMP‐9 and cathepsin B, respectively, demonstrating sustained resistance to enzymatic degradation (Figure S4, Supporting Information). These results further support the structural robustness of the aCD47‐C‐miR‐34a conjugate and reinforce its potential as a stable and effective platform for systemic in vivo delivery.

A key rationale for utilizing a cleavable PEG‐based crosslinker in conjugation is to enable selective cleavage by GSH within cell endosomes or lysosomes, thereby promoting the controlled release of miRNAs. The concentration of GSH in blood is significantly lower than in cancer cells, which allows the linker to remain stable in circulation while selectively releasing the active payload in cancer cells with elevated GSH levels. From this point of view, we investigated the release of miR‐34a from aCD47‐C‐miR34a under the reducing conditions of a disulfide‐containing linker. As shown in Figure 2f, the protein band and fluorescence band of aCD47‐C‐miR34a gradually reduced into small fragments, exhibiting similar band locations. Furthermore, miR‐34a was partially released in a conjugated form with both the heavy and light chains of the antibody, with complete release occurring at Dithiothreitol (DTT) concentrations above 50 mm. Notably, the fluorescence signal of the free form of miR‐34a gradually increased in a DTT concentration‐dependent manner, indicating miR‐34a could be released from aCD47‐C‐miR34a under reducing conditions (Figure 2g,h). These results suggest that our approach facilitated simple and efficient conjugation between aCD47 and miR‐34a while ensuring the effective release of miR‐34a under reducing conditions, such as endosomes and lysosomes in the cancer cells. This structural design provides a critical advantage by maintaining conjugate stability in the extracellular environment, thereby minimizing premature degradation and off‐target effects during systemic circulation. Upon internalization, the elevated intracellular GSH levels cleave the disulfide bond, enabling the cytoplasmic release of functional miR‐34a. This controlled, stimulus‐responsive release is essential for efficient RNA‐induced gene expression. Thus, our AOC platform exemplifies a next‐generation biomaterial that unites targeted delivery with intracellular activation, addressing key limitations in current miRNA therapeutics and paving the way for broader application in precision oncology.

2.2. CD47‐Mediated Intracellular Delivery and Endosomal Escape of aCD47‐C‐miR34a

To evaluate the role of aCD47 as the intracellular transport of miR‐34a cargoes, we investigated the intracellular delivery of aCD47‐miR34a using 4T1 TNBC cells. We confirmed cancer cell lines, such as 4T1 and 4T1‐Luc cells, highly expressed CD47 on the cell surface compared to normal cell lines such as BALB 3T3 and RAW264.7 (Figure S5, Supporting Information). As expected, aCD47‐C‐miR34a demonstrated a significantly higher intracellular uptake efficiency of 81.6% in 4T1 cells, whereas miR‐34a alone showed a much lower uptake efficiency of 6.74%, underscoring the efficacy of the antibody‐mediated targeting strategy in enhancing cellular internalization (Figure 3a). Pre‐treatment of 4T1 cells with aCD47 to block the CD47 receptors on the cell surface resulted in a significant reduction in cellular uptake, demonstrating that aCD47‐C‐miR34a is specifically internalized through CD47 receptor‐mediated endocytosis. To further verify the CD47‐mediated internalization mechanism, we performed uptake analysis in BALB 3T3 and RAW264.7 cells, which express markedly lower levels of CD47. The internalization efficiency of aCD47‐C‐miR34a was significantly reduced in these low‐CD47‐expressing cells – 30.0% in BALB 3T3 and 14.8% in RAW264.7 – compared to 70.7% in 4T1 cells (Figure S6, Supporting Information). These results align with the differential CD47 expression profiles (Figure S5, Supporting Information) and provide strong evidence that cellular uptake of the conjugate is predominantly mediated by CD47. Additionally, aCD47‐C‐miR34a could deliver miR‐34a into 4T1 cells in a dose‐dependent manner, confirming CD47‐targeted miR‐34a delivery efficacy (Figure S7, Supporting Information). Consistent with flow cytometry analysis, confocal microscopy images of 4T1 cells also showed that aCD47‐C‐miR34a significantly improved the intracellular delivery of miR‐34a by 7.02‐fold compared to miR‐34a alone (Figure 3b). Pre‐treatment of 4T1 cells with aCD47 resulted in a significant 6.89‐fold reduction in cellular uptake compared to untreated cells.

Figure 3.

CD47‐mediated intracellular delivery and endosomal escape of aCD47‐miR34a in vitro. a) Flow cytometry analysis showing the cellular uptake of 25 nM Cy5‐labeled miR‐34a or aCD47‐C‐miR34a in 4T1 cells. The cells were pre‐incubated with or without 1 mg mL⁻¹ anti‐CD47 antibodies (n = 3). b) Representative confocal images of cellular uptake after treatment with 150 nM Cy5‐labeled miR‐34a or aCD47‐C‐miR34a in 4T1 cells. The cells were pre‐incubated with or without 1 mg mL⁻¹ anti‐CD47 antibodies. c,d) Representative confocal images of aCD47‐miR34a endosomal escape in 4T1 cells after treatment with 150 nm aCD47‐miR34a. Colocalization of Cy5‐labeled aCD47‐miR34a (red) and early endosome (green) or lysosome (green) was observed at 1, 3, and 6 h post‐treatment. Scale bar: 50 µm. The degree of the colocalization between aCD47‐miR34a and early endosome or lysosome signals was quantified as Pearson's correlation coefficient (r). Fluorescence intensity per cell nucleus was calculated using ImageJ software and presented as normalized fluorescence intensity (n = 3). Data are presented as the mean ± SD (^* p <0.05, ^** p <0.01, ^*** p <0.001, ^**** p <0.0001). a, b) One‐way ANOVA with Tukey's post‐hoc test. c,d) two‐tailed unpaired Student's t‐test.

Given that intracellular internalization mechanisms of AOC significantly impact miRNA delivery, we investigated how the aCD47‐C‐miR34a is retained within the endo‐lysosomal compartments and subsequently released into the cytoplasm. This is particularly relevant in the therapeutic applications of AOCs, where the choice of linkers plays a vital role in ensuring stability in circulation and precise payload release within the tumor cells. To explore the payload release mechanism, we employed two distinct linkers: non‐cleavable and cleavable. Initially, we observed the co‐localization of aCD47‐C‐miR34a conjugates with endosomes and lysosomes. Confocal microscopy images exhibited that the miR‐34a fluorescence signals (red) from both non‐cleavable and GSH‐cleavable linker‐containing aCD47‐miR34a predominantly co‐localized (yellow) with early endosomes (green) at 1 h post‐treatment. By 6 h post‐treatment, the fluorescence signal of aCD47‐C‐miR34a exhibited a marked decrease in co‐localization with early endosomes. However, aCD47‐NC‐miR34a showed no significant changes in the co‐localization pattern (Figure 3c; Figure S8a, Supporting Information). The degree of co‐localization between aCD47‐miR34a conjugates and early endosomes at 1 h was quantified using Pearson's correlation coefficients, with values of 0.468 and 0.601 for aCD47‐NC‐miR34a or aCD47‐C‐miR34a, respectively. Notably, by 6 h, aCD47‐C‐miR34a had separated from the endosomes, indicating reduced co‐localization. In contrast, the aCD47‐NC‐miR34a conjugates remained trapped within the early endosomes, maintaining a higher Pearson's correlation coefficient of 0.541 compared to aCD47‐C‐miR34a.

Typically, early endosomes mature into late endosomes and subsequently fuse with lysosomes. Accordingly, we analyzed the co‐localization of the conjugates with lysosomes 24 h post‐treatment (Figure 3d; Figure S8b, Supporting Information). As expected, the fluorescence signal from aCD47‐C‐miR34a showed a significantly lower degree of co‐localization with lysosomes, with a Pearson's correlation coefficient of 0.512. In contrast, the fluorescence signal from aCD47‐NC‐miR34a predominantly co‐localized with that of lysosomes, yielding a higher correlation coefficient of 0.812. These results suggest that aCD47‐C‐miR34a was effectively internalized into CD47‐expressing 4T1 cells and facilitated the efficient release of miR‐34a into the cytoplasm upon cellular internalization, thereby bypassing lysosomal degradation. This mechanism underscores the potential of cleavable linkers in enhancing miRNA delivery efficiency.

2.3. Impact of Effective Endosomal Escape of aCD47‐C‐miR34a in 4T1 TNBC

Endosomal escape is a critical intracellular process for the successful delivery of miRNAs as efficient release from the endosomal compartment enhances and prolongs their therapeutic efficacy.^{[ 15 ]} Next, we monitored the endosomal escape of miR‐34a using calcein, a membrane‐impermeable green fluorophore, as a co‐localization marker.^{[ 16 ]} Upon co‐treatment of calcein with the aCD47‐NC‐miR34a, miR‐34a fluorescence signal in 4T1 cells remained co‐localized with calcein for up to 6 h (Figure 4a). In contrast, aCD47‐C‐miR34a showed initial co‐localization with calcein at 1 h post‐treatment but exhibited progressive separation of miR‐34a fluorescence from calcein at 3 and 6 h, suggesting successful endosomal escape and cytoplasmic release of miR‐34a (Figure 4b). Notably, the co‐administered calcein, typically retained within endosomes, exhibited diffuse cytosolic green fluorescence at 6 h post‐treatment with aCD47‐C‐miR34a, further supporting successful endosomal disruption. We propose that GSH‐mediated cleavage of disulfide linkages releases miR‐34a from the aCD47 within endosomes. This cleavage is followed by the generation of free thiols and a sudden increase in osmotic pressure due to the release of negatively charged oligonucleotides, collectively contributing to physicochemical stress within the endosomal lumen.^{[ 17} , ¹⁸ , ¹⁹ , ^{20 ]} These combined effects destabilize the endosomal membrane by inducing transient pore formation or partial rupture, thereby promoting efficient endosomal escape and enhancing the cytosolic delivery of miR‐34a.^{[ 21 ]} Statistical analysis of Pearson's correlation coefficient confirmed these observations, with aCD47‐NC‐miR34a showing no significant change over time, whereas aCD47‐C‐miR34a exhibited a significant decrease in correlation coefficients (0.784 to 0.495 at 3 and 6 h post‐treatment, respectively) (Figure 4c; Figure S9a–c, Supporting Information). This efficient escape was attributed to the cleavage of disulfide bonds in the cleavable linker by intracellular GSH, facilitating the release of intact miR‐34a during endosomal maturation.

Figure 4.

Impact of effective endosomal escape of aCD47‐miR34a in 4T1 breast cancer cell line in vitro. a,b) Representative confocal images of aCD47‐miR34a tracking the endosomal escape in 4T1 cells after treatment with 150 nm aCD47‐miR34a. Colocalization of Cy5‐labeled aCD47‐miR34a (red) and calcein (green) was observed at 1, 3, and 6 h post‐treatment. Scale bar: 25 µm. c) The degree of colocalization between aCD47‐miR34 and calcein signals was compared at various time points, quantified by Pearson's correlation coefficient (r). Fluorescence intensity per cell nucleus was calculated using ImageJ software and presented as normalized fluorescence intensity (n = 6). d) Relative miR‐34a expression levels in 4T1 cells measured by RT‐qPCR. All samples were normalized to U6 expression (n = 3). e, f) qRT‐PCR analysis of relative PD‐L1 and SIRT1 mRNA expression in 4T1 cells following treatmnt. All samples were normalized to 18s RNA expression (n = 3). g) Western blot analysis of SIRT1 and PD‐L1 abundance in 4T1 cells after treatment with 130 nm aCD47‐C‐miR34a for 24 h, and h) quantification of protein expression levels normalized to GAPDH expression (n = 3). i) Flow cytometry analysis showing the Annexin V/PI staining in 4T1 cell after treatment with PBS, 130 nM miR‐34a, or 130 nM aCD47‐C‐miR34a for 48 h (n = 3). Data are presented as the mean ± SD (^* p <0.05, ^*** p <0.001, ^**** p <0.0001). c,e,f,i) Two‐way with Sidak's multiple comparison test. d, h) One‐way ANOVA with Tukey's post‐hoc test.

Unlike MRX34, a first‐in‐human clinical trial of a miRNA‐based cancer therapy,^{[ 6a ]} which relies on passive or pH‐responsive mechanisms and often suffers from lysosomal entrapment and off‐target release,^{[ 22 ]} aCD47‐C‐miR34a ensures that the miR‐34a is released from the antibody after uptake into the target tumor cell, thereby promoting more efficient cytosolic delivery than non‐cleavable linkers. Its modular design integrates material engineering principles with antibody‐guided targeting, minimizing premature release and improving pharmacological specificity.

Given miR‐34a's role as a tumor suppressor, frequently downregulated in various malignancies, including myeloma, colorectal cancer, lung adenocarcinoma, and TNBC, we next examined its therapeutic potential in the 4T1 TNBC model. As shown in Figure 4d, treatment with aCD47‐C‐miR34a significantly increased miR‐34a expression levels, leading to reduced expression of tumorigenesis‐related proteins SIRT1 and PD‐L1,^{[ 4a ]} by directly targeting their 3′ UTRs. To determine whether miR‐34a suppressed SIRT1 and PD‐L1 at the transcriptional level, we quantified their mRNA expression by qRT‐PCR. Treatment with aCD47‐CmiR‐34a resulted in a significant downregulation of both PD‐L1 and SIRT1 mRNA levels, whereas treatment with antibody alone had no effect (Figure 4e,f). Consistently, western blot analysis showed that SIRT1 and PD‐L1 protein levels decreased by 2.31‐fold and 3.02‐fold, respectively, compared to the phosphate‐buffered saline (PBS) group (Figure 4g,h). Furthermore, repression of SIRT1, a key regulator of apoptosis, enhanced apoptotic cell death rates (9.67%) compared to PBS (1.42%) and miR‐34a treatment alone (2.59%) (Figure 4i). In vitro cytotoxicity assays further confirmed the concentration‐dependent anti‐cancer activity of aCD47‐C‐miR34a, underscoring its therapeutic potential (Figure S10, Supporting Information).

In addition, we explored the ability of aCD47‐C‐miR34a to enhance macrophage‐driven phagocytosis in vitro. Overexpression of CD47 on tumor cells allows immune evasion by acting as a “don't eat me” signal.^{[ 23 ]} Treatment of aCD47‐C‐miR34a synergistically disrupted this signal and promoted macrophage recognition of tumor cells by blocking CD47 and delivering miR‐34a to modulate SIRT1 levels.^{[ 24 ]} Bone marrow‐derived macrophages (BMDMs) co‐treated with aCD47‐C‐miR34a exhibited significantly enhanced macrophage‐mediated phagocytosis compared to aCD47 alone, as validated by fluorescence microscopy and flow cytometry (Figure S11a–c, Supporting Information). These findings emphasize the therapeutic potential of aCD47‐C‐miR34a, which not only targets oncogenic pathways but also modulates the TME, providing a promising approach for TNBC immunotherapy.

2.4. Tumor‐Targeted miR‐34a Delivery and Therapeutic Efficacy of aCD47‐C‐miR34a in a 4T1 TNBC Model

To investigate the tumor‐targeted miR‐34a delivery by aCD47‐C‐miR34a in vivo, orthotopic and xenograft models were established in BALB/c nude mice using 4T1 murine TNBC cell line. Tumor‐targeted miR‐34a delivery was monitored following intravenous injection of Cy5‐labeled miR‐34a, IgG‐C‐miR34a, or aCD47‐C‐miR34a in an orthotopic 4T1 TNBC tumor model. In vivo near‐infrared fluorescence (NIRF) images revealed significant fluorescence signal retention at tumor sites in aCD47‐C‐miR34a‐treated mice, with fluorescence intensity at tumor sites being 3.89‐ and 1.74‐fold higher than that observed for IgG‐C‐miR34a‐ and free miR‐34a‐treated groups, respectively (Figure  5a,b).

Figure 5.

Tumor‐targeting ability and therapeutic efficacy of aCD47‐C‐miR34a in an orthotopic 4T1 TNBC model. a) In vivo fluorescence images of orthotopic 4T1 tumor‐bearing TNBC BALB/c nude mouse models following systemic administration of IgG‐C‐miR34a, miR‐34a, or aCD47‐C‐miR34a (n = 3). b) Quantitative fluorescence intensities of IgG‐C‐miR34a, miR‐34a, or aCD47‐C‐miR34a in the tumor tissues at 0, 1, 3, and 6 h. c) Ex vivo fluorescence images of excised six major organs including the liver (Li), lung (Lu), spleen (Sp), kidney (Ki), heart (He), and tumor (Tu) harvested at 6 h after systemic administration of IgG‐C‐mR34a, miR‐34a, or aCD47‐C‐miR34a (n = 3). d) Quantification of organ distribution of IgG‐C‐miR34a, miR‐34a, or aCD47‐C‐miR34a (n = 3). e–g) Ex vivo fluorescence images and quantification of average radiant efficiency in 4T1 tumors harvested at 6 h after systemic administration of IgG‐C‐miR34a, miR‐34a, or aCD47‐miR34a (n = 3). h,i) Quantification of tumor Cy5 radiant efficiency with normalized to that of liver or kidney (n = 3). j) Schematic of the experimental design for the aCD47‐C‐miR34a administration in 4T1‐Luc tumor‐bearing TNBC mouse models. k) Bioluminescence imaging and l) total bioluminescence flux measured via IVIS Lumina, in orthotopic 4T1‐Luc tumor‐bearing mice on days 6, 9, 12, and 15 (n = 6). m) Average tumor growth curves in orthotopic 4T1‐Luc tumor‐bearing mice treated with PBS, aCD47, or aCD47‐C‐miR34a (n = 6). n) Image of excised orthotopic 4T1‐Luc tumors along with individual tumor growth curves for each group. o) Tumor weight on day 17 after tumor inoculation. Data are presented as the mean ± SD (^* p <0.05, ^** p <0.01, ^*** p <0.001, ^**** p <0.0001). Statistical significance was calculated by b, d, l, m) two‐way ANOVA with Sidak's multiple comparison test and g, h, i, o) one‐way ANOVA followed by Tukey's multiple comparisons tests.

Ex vivo NIR fluorescence imaging of major organs (liver, lung, spleen, kidney, and heart) and tumor showed that miRNA fluorescence was primarily observed in the kidney (Figure 5c,d). This pattern is likely the kidney's natural role in processing monoclonal antibodies, which are reabsorbed by the proximal tubules and returned to the systemic circulation.^{[ 25 ]} Consequently, the IgG‐C‐miR34a and aCD47‐C‐miR34a groups exhibited the highest renal signals due to antibody reabsorption. Furthermore, although the molecular weight of aCD47‐C‐miR34a (162–178 kDa) far exceeds the 60 kDa glomerular filtration threshold, the released miR‐34a from aCD47 is expected to undergo metabolic degradation and be reabsorbed by proximal tubular epithelial cells.^{[ 26 ]} Since the proximal tubule is known to concentrate antisense oligonucleotides at high levels, this mechanistic pathway provides a plausible explanation for the observed renal localization of miRNA fluorescence.^{[ 27 ]} Nevertheless, analysis of miR‐34 a distribution revealed that tumor‐specific accumulation was significantly higher in the aCD47‐C‐miR34a group compared to other groups (Figure 5e,f). Quantitative analysis revealed that aCD47‐C‐miR34a achieved 2.5‐fold greater tumor accumulation than IgG‐C‐miR34a and 1.5‐fold higher accumulation than free miR‐34a (Figure 5g). While the liver and kidney were identified as the primary off‐tumor accumulation sites, normalization of ex vivo tumor radiant efficiency to the levels in the liver or kidney revealed that aCD47‐C‐miR34a maintained a significantly higher tumor‐to‐liver/kidney ratio compared to both IgG‐C‐miR34a and free miR‐34a (Figure 5h,i). Consistent results were observed in NIR fluorescence imaging of 4T1 TNBC xenograft models (Figure S12, Supporting Information), emphasizing the superior tumor selectivity of aCD47‐C‐miR34a.

Finally, we evaluated therapeutic efficacy in a 4T1‐Luc orthotopic tumor model to determine its potential for tumor suppression. When the tumor volumes reached ≈40–50 mm³, the mice were intravenously injected with PBS, aCD47, or aCD47‐C‐miR34a at an equivalent antibody dose of 5 mg kg⁻¹, according to the treatment schedule presented in Figure 5j. Tumor progression was monitored using bioluminescence imaging. As shown in Figure 5k, a bioluminescence signal showed no observable changes in the aCD47‐C‐miR34a‐treated group over 15 days, whereas signals gradually increased in the other treatment groups (Figure 5l). Tumor volume analysis confirmed significant tumor growth inhibition in the aCD47‐C‐miR34a‐treated group (Figure 5m,n). Correspondingly, the weight of excised tumor tissues in the aCD47‐C‐miR34a‐treated group (74.6 ± 13.5 mg) was significantly lower than those in PBS (199.7 ± 44.3 mg) or aCD47‐treated groups (192.6 ± 45.6 mg) (Figure 5o). These results showed the therapeutic potential of aCD47‐C‐miR34a, which combines targeted miR‐34a delivery to the tumor site with potent anti‐tumor efficacy, making it a promising strategy for TNBC treatment.

2.5. In Vivo Antitumor Immune Response of aCD47‐C‐miR34a in an Orthotopic 4T1 TNBC Model

Consistent with in vitro results, aCD47‐C‐miR34a treatment restored miR‐34a expression in the tumor tissues (Figure  6a). As miR‐34a exerts its therapeutic effects by suppressing PD‐L1 and SIRT1 mRNA expression, the immunomodulatory effects of CD47‐C‐miR34a were further evaluated. Immunofluorescence imaging revealed a significant reduction in PD‐L1 expression within the TME following aCD47‐C‐miR34a treatment (Figure 6b). Concurrently, Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining demonstrated a marked increase in apoptotic cells, with apoptosis rates reaching 58.6% in the aCD47‐C‐miR34a‐treated group, underscoring its dual mechanism of action in modulating the TME and inducing cancer cell apoptosis (Figure 6c).

Figure 6.

In vivo antitumor immune response of aCD47‐C‐miR34a in an orthotopic 4T1 TNBC model. a) Relative miR‐34a expression levels in tumor tissues measured by RT‐qPCR. All samples were normalized to U6 expression (n = 3). b,c) Representative immunofluorescence images and quantification of relative fluorescence intensities in tumor tissues stained with APC‐conjugated PD‐L1 antibody or TUNEL staining (n = 3). Scale bar: 200 µm. d–f) The percentage of mature DCs (CD11c⁺CD40⁺, CD86⁺, or CD80⁺) in TDLN (n = 3). g) The percentage of total T‐cells (CD45.2⁺CD3⁺CD8⁺) and h–j) activated CD8⁺ T‐cells (CD45.2⁺CD3⁺CD8⁺Granzyme B⁺, CD44⁺, or Ki67⁺) in tumor tissues (n = 3). k) Representative immunofluorescence images of whole tumor sections to visualize the distribution and activation status of CD8⁺ T‐cells, stained with APC‐conjugated CD8 antibody and FITC‐conjugated Granzyme B antibody. Scale bar, 1 mm. Data are presented as the mean ± SD (^* p <0.05, ^** p <0.01, ^*** p <0.001, ^**** p <0.0001). Statistical significance was calculated by a,d,e,f,g–j) one‐way ANOVA followed by Tukey's multiple comparisons test.

Tumor antigen presentation by dendritic cells (DCs) is critical for activating antigen‐specific cytotoxic CD8⁺ T‐cells, particularly via apoptotic tumor cells.^{[ 28 ]} To assess the impact of miR‐34a‐induced apoptosis on DC activation, DC maturation in tumor‐draining lymph nodes (TDLNs) was analyzed. In the aCD47‐C‐miR34a‐treated group, mature DC markers (CD11c⁺CD40⁺, CD86⁺, or CD80⁺) were significantly upregulated, indicating enhanced DC maturation within the TME (Figure 6d–f; Figure S13, Supporting Information).

This elevated DC activation correlated with a notable increase in tumor‐infiltrating CD8⁺ T‐cells (CD45.2⁺CD3⁺CD8⁺) in the TME (Figure 6g). Additionally, aCD47‐C‐miR34a treatment substantially elevated the activation marker CD44 (CD45.2⁺CD3⁺CD8⁺CD44⁺) on tumor‐infiltrating CD8⁺ T‐cells (Figure 6h). Flow cytometry further demonstrated increased expression of Granzyme B and Ki67 in CD8⁺ T‐cells (CD45.2⁺CD3⁺CD8⁺Granzyme B⁺ or Ki67⁺), reflecting enhanced effector function and proliferative activity (Figure 6i,j). Whole‐tumor immunofluorescence imaging corroborated these findings, showing co‐localization of CD8⁺ and Granzyme B within TME, further confirming robust infiltration and activation of CD8 T‐cells in the aCD47‐C‐miR34a‐treated group (Figure 6k). In summary, aCD47‐C‐miR34a therapy effectively enhances DC maturation, promotes infiltration of CD8⁺ T‐cells into the TME, and boosts their functional activation. These findings collectively underscore the potential of aCD47‐C‐miR34a as a potent immunotherapeutic strategy for TNBC.

2.6. Safety Profile and Comparative Advantage of aCD47‐C‐miR34a Over Conventional miR‐34a Therapy

Previous clinical trials with miR‐34a‐based therapies failed due to challenges including off‐target effects, poor delivery efficiency, and systemic toxicities. To address these limitations, we evaluated the in vivo safety profile of aCD47‐C‐miR34a, which incorporates targeted delivery and immune modulation, offering a potential advantage over traditional approaches.

Mice were intravenously injected with PBS, aCD47, or aCD47‐C‐miR34a at an equivalent antibody dose of 5 mg kg⁻¹, following the treatment schedule shown in Figure 5j. Throughout the study, no significant changes in body weight were observed in any treatment group (Figure  7a), indicating the absence of overt systemic toxicity. Blood chemistry analysis revealed that mice treated with aCD47‐C‐miR34a had white blood cell (WBC) counts within the normal range (5.69–14.84) compared to PBS‐ or aCD47‐treated groups.^{[ 29 ]} Additionally, we confirmed that aCD47‐C‐miR34a did not cause lymphocytopenia or other abnormalities commonly linked to excessive immune activation (Figure 7b). Histological analysis of major organs (liver, spleen, lung, heart, and kidney) using hematoxylin and eosin (H&E) staining showed no significant abnormalities or pathological changes, such as fibrosis, necrosis, or inflammation in mice (Figure 7c). Therefore, our findings highlight that aCD47‐C‐miR34a delivers miR‐34a with high tumor selectivity, circumventing the systemic toxicities that plagued earlier miR‐34a therapies. The conjugate achieves a robust antitumor immune response without inducing immune‐related adverse effects, underscoring its promise as a safe and effective therapeutic platform for clinical application.

Figure 7.

H&E‐stained images from a tumor growth inhibition study showing sections of extracted visceral organs from an orthotopic 4T1 TNBC model. a) Average body weight curves in orthotopic 4T1‐Luc tumor‐bearing mice treated with PBS, aCD47, or aCD47‐C‐miR34a (n = 6). b) Total WBC and percentages of different blood cell types were analyzed after mice sacrifice. Normal range c) H&E‐stained images from a tumor growth inhibition study showing sections of the five major visceral organs extracted from the 4T1 orthotopic TNBC model. Scale bar: 200 µm. Data are presented as the mean ± SD (^* p <0.05, ^** p <0.01, ^*** p <0.001, ^**** p <0.0001). a) One‐way ANOVA followed by Tukey's multiple comparisons test.

3. Conclusion

In conclusion, our study demonstrates the potential of aCD47‐C‐miR34a as a novel and comprehensive therapeutic strategy for TNBC. This AOC effectively addresses key limitations of traditional miR‐34a therapies, which failed in clinical trials due to off‐target effects, poor delivery efficiency, and systemic toxicities. By leveraging CD47‐mediated targeting and bioreducible linker, aCD47‐C‐miR34a successfully delivered miR‐34a into the cytoplasm of TNBC cells, restoring tumor‐suppressive functions while simultaneously modulating the immune microenvironment.

The aCD47‐C‐miR34a conjugate not only inhibited tumor growth but also promoted CD8⁺ T‐cell activation through a synergistic mechanism of miR‐34a‐related apoptosis and PD‐L1 suppression. Importantly, preclinical studies showed that this targeted therapeutic approach elicited robust antitumor immune responses without systemic toxicity or adverse effects, overcoming critical safety challenges associated with prior miR‐34a therapies.

These findings position aCD47‐C‐miR34a as a promising candidate for clinical application, offering a safe and effective solution to the intrinsic heterogeneity and immune evasion commonly observed in TNBC. By integrating precise targeting and immune modulation, this strategy provides a robust framework for advancing treatment options for patients with TNBC and potentially other immune‐suppressive malignancies.

4. Experimental Section

Materials

The anti‐mouse CD47 antibody was purchased from BioXCell (Republic of Korea). Cleavable Azide‐PEG3‐SS‐NHS linker was purchased from MedChemExpress (USA). Non‐cleavable NHS‐PEG4‐Azide linker was purchased from Thermo Fisher Scientific (USA). L‐Glutathione reduced was purchased from Sigma–Aldrich (USA). The DBCO‐functionalized miR‐34a mimic comprising DNA (DBCO‐5′‐TGGCAGTGTCTTAGCTGGTTGT‐3′), Cy5‐labeled miR‐34a, DBCO‐miR‐34a with backbone modifications (DBCO‐5′‐TGGCAGT GTCTTAGCTGGTTGT‐3′), underlined identify locked nucleic acids (LNAs), and italics represent phosphorothioate (ps) linkages, and Cy5‐labeled miR‐34a with backbone modifications, were purchased from Bioneer (Republic of Korea).

Preparation and characterization of aCD47‐miR34a AOC—General Procedure for Antibody‐Linker Conjugation

A aCD47 antibody (100 µg, 0.67 nmol) of 1 mg mL⁻¹ was reacted with 25 equiv. of Azide‐PEG3‐SS‐NHS (7.3 µg, 16.75 nmol) or NHS‐PEG4‐Azide (6.5 µg, 16.75 nmol) in PBS buffer (pH 8.0) at 25 °C for 2 h with shaking at 700 rpm. After the reaction, the aCD47 antibody‐linker conjugate was purified, and buffer exchange was performed by ultrafiltration using an Amicon Ultra 30K MWCO (MERCK Millipore).

Preparation and Characterization of aCD47‐miR34a AOC–General Procedure for Antibody‐Oligonucleotide Conjugation

The azide‐functionalized aCD47 antibody‐linker conjugate (100 µg, 0.67 nmol) was reacted with a DBCO‐functionalized miR‐34a mimic (21 µg, 2.68 nmol) in PBS buffer (pH 7.4) at 37 °C for 2 h with shaking at 1100 rpm. After the reaction, the aCD47‐miR34a conjugate was purified with ultrafiltration using an Amicon Ultra 10K MWCO (MERCK Millipore) to remove free miRNA. The conjugates were characterized by native gel electrophoresis (Bio‐Rad, USA), SDS‐PAGE (Bio‐Rad, USA), and whole protein mass analysis using Autoflex Speed LRF (Bruker, USA). The DoC was calculated from the areas of major peaks in the whole protein mass analysis using the following formula (1). To ensure accurate quantification, the relative intensities of major peaks were analyzed in the mass spectrum, each corresponding to antibody conjugates with a defined number of miR‐34a molecules.

$$DoC=\frac{\sum n.I_n}{\sum I_n}$$ (1)

I_n represents the relative peak intensity of conjugates with n additional molecules conjugated per antibody.

Preparation and characterization of aCD47‐miR34a AOC—GSH‐Mediated Linker Cleavage Assay

For evaluating the release of miR‐34a from aCD47‐C‐miR34a, the disulfide bridges in aCD47‐C‐miR34a (0.4 mg mL⁻¹) were reduced with various concentrations of DTT (0.05, 0,5, 5, or 50 mm dissolved in PBS pH 7.4) at 37 °C for 30 min. All samples were confirmed using 4–15% native gel, followed by Coomassie blue staining and Cy5‐labeled miR‐34a detection. The relative fluorescence intensity of the released Cy5‐labeled miR‐34a was quantified.

Serum Stability of aCD47‐C‐miR34a AOC

To assess the serum stability of the aCD47‐C‐miR‐34a conjugate, free miR‐34a and the aCD47–C–miR‐34a conjugate were incubated in 20% mouse serum at 37 °C for up to 72 h. At designated time points, samples were collected and analyzed by native gel (Bio‐Rad, USA) and 2% agarose gel electrophoresis to assess miRNA integrity.

To evaluate the proteolytic stability of the aCD47‐C‐miR‐34a conjugate, the conjugates were incubated with MMP‐9 (18 ng mL⁻¹) or cathepsin B (1 µg mL⁻¹) at 37 °C for up to 72 h. At designated time points, the integrity of the conjugates was assessed using by native gel (Bio‐Rad, USA), followed by EtBr and Coomassie blue staining.

Cell Culture

Murine breast cancer cell lines 4T1 and 4T1‐Luc were obtained from the American Type Culture Collection (ATCC, USA) and cultured in Roswell Park Memorial Institute (RPMI)‐1640 (Welgene, Republic of Korea) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% antibiotic‐antimycotic (Gibco, USA), and maintained at 37 °C with 5% CO₂. Fibroblast murine cell line BALB 3T3 and macrophage cell line RAW264.7 were obtained from the American Type Culture Collection (ATCC, USA) and cultured in Dulbecco's Modified Eagle's Medium (DMEM; Welgene, Republic of Korea) supplemented with 10% fetal bovine serum (FBS; Gibco, USA) and 1% antibiotic‐antimycotic (Gibco, USA), and maintained at 37 °C with 5% CO₂.

To induce differentiation into BMDMs, bone marrow cells were isolated from the hind limbs of 7‐week‐old male BALB/c mice and plated in 100‐mm culture dishes overnight to remove unwanted cells. Following overnight incubation in cell growth media, suspended bone marrow cells were re‐plated at a density of 2 × 10⁵ cells per dish and cultured for 7 days in growth media supplemented with 20 ng mL⁻¹ macrophage colony‐stimulating factor (M‐CSF; Peprotech, USA).

Detection of CD47 Expression on Various Cell Lines

To assess CD47 expression levels on the surface of various cell types, the 4T1, 4T1‐Luc, BALB 3T3, and RAW264.7 cells were seeded into 35‐mm glass‐bottom dishes at density of 3 × 10⁵ cells/well and incubated at 37 °C for 24 h. Following day, cells were pre‐blocked with an Fc Block to avoid nonspecific binding and stained with and stained with an APC‐anti‐CD47 antibody (clone miap301, #127514) 15 min at 4 °C. These cells were then washed three times with PBS and analyzed using CytoFLEX flow cytometer (Beckman Coulter, USA) to quantify CD47 surface expression.

Cellular Uptake of aCD47‐C‐miR34a

For fluorescence imaging of aCD47‐C‐miR34a, 2 × 10⁵ 4T1 cells were seeded into 35‐mm glass‐bottom dishes and incubated at 37 °C for 24 h. After stabilization, cells were incubated in a serum‐free medium containing 150 nM Cy5‐labeled miR‐34a or aCD47‐C‐miR34a for 6 h. Following treatment, cells were washed three times with PBS, fixed with 4% paraformaldehyde (PFA; Biosesang, Republic of Korea) for 10 min, and the nuclei were stained with Hoechst 33342 (Invitrogen, USA) for 10 min. Fluorescence imaging was performed using a Leica TCS SP8 confocal laser scanning microscope (Leica Microsystems, Germany), and mean fluorescence intensity was quantified using Image J software (National Institutes of Health, Bethesda, MD, USA).

For flow cytometry analysis of cellular uptake, 2 × 10⁵ 4T1 cells were seeded into 35‐mm glass‐bottom dishes and incubated at 37 °C for 24 h. After stabilization, cells were incubated in a serum‐free medium containing 25 nM Cy5‐labeled miR‐34a or aCD47‐C‐miR34a for 30 min. Following treatment, cells were washed three times with PBS to remove any nonspecific binding. Subsequently, the cellular uptake was assessed using a CytoFLEX flow cytometer (Beckman Coulter, USA). Additionally, the cellular uptake of aCD47‐C‐miR34a was analyzed across a range of concentrations to evaluate dose‐dependent uptake. For blocking CD47, 4T1 cells were pre‐treated with anti‐CD47 antibody (1 mg mL⁻¹, BioXCell, USA) for 2 h to block CD47 on the cell membrane.

Tracking Endosomal Escape of aCD47‐miR34a

To observe colocalization with aCD47‐miR34a and Early endosomes or lysosomes, depending on the presence of the cleavable linker, 4T1 cells were seeded in 35‐mm glass‐bottom confocal dishes at a density of 1 × 10⁵ cells/well and labeled with the following reagents: CellLight™ Early endosomes‐GFP and Lysosomes‐GFP (Invitrogen, USA) for overnight. The next day, cells were treated with 150 nm Cy5‐labeled aCD47‐miR34a, synthesized with non‐cleavable or cleavable linker in a serum‐free media for 1, 6, and 24 h. After treatment, the cells were washed three times with PBS. Live cell fluorescence images were acquired by a TCS SP8 confocal laser microscope (Leica Microsystems, Germany).

To measure calcein‐dependent trafficking of endosomal escape of aCD47‐miR34a, 4T1 cells were seeded under the condition described above. The next day, cells were treated with Cy5‐labeled aCD47‐miR34a synthesized with a non‐cleavable or cleavable linker, in a calcein solution (0.1 mg mL⁻¹) for 1, 3, and 6 h. After treatment, the cells were washed three times with Opti‐MEM medium (Gibco, USA). Live cell fluorescence images were acquired by a TCS SP8 confocal laser microscope (Leica Microsystems, Germany).

In Vitro Cytotoxicity Assay

To evaluate the cytotoxic effects of aCD47‐C‐miR34a, 5 × 10³ 4T1 cells were seeded into 96‐well plates and incubated at 37 °C for 24 h. After stabilization, cells were treated with aCD47‐miR‐34a at concentrations ranging from 0 to 5 µm in serum‐free media for 48 h. Following treatment, a Cell Counting Kit‐8 (CCK‐8; Dojindo Laboratory, Japan) solution was added to each well. Cell viability was assessed by measuring the colorimetric change (absorbance at 450 nm) of the CCK‐8 solution using a microplate reader (SpectraMax 34, Molecular Devices, USA).

In Vitro Phagocytosis Assay

To assess the phagocytic activity of aCD47‐C‐miR‐34a, BMDMs were isolated from BALB/c mice using an established protocol. The successful differentiation of BMDMs was identified by the high expression levels (97%) of the F4/80 and CD11b markers. For flow cytometry analysis of phagocytic activity, 2 × 10⁵ BMDMs were stained with CellTracker Green CMFDA (1 µm; Thermo Fisher Scientific, USA) for 30 min at room temperature and seeded into a 35‐mm glass‐bottom petri dishes. Separately, 2.5 × 10⁵ 4T1 cells were seeded in 60‐mm glass‐bottom dishes and incubated at 37 °C for 24 h. Following stabilization, 4T1 cells were treated with 150 nM aCD47 or aCD47‐C‐miR34a in a serum‐free medium for 24 h. Treated 4T1 cells were subsequently stained with CellTracker Deep Red (1 µm; Thermo Fisher Scientific, USA) for 30 min at room temperature and then co‐cultured with stained BMDMs at a 1:4 ratio for 2 h. After co‐culture, cells were washed three times with PBS to prepare single‐cell suspensions. Phagocytic activity was quantified by flow cytometry using a CytoFLEX flow cytometer (Beckman Coulter, USA).

For microscopic evaluation of phagocytosis activity, 2.5 × 10⁵ 4T1 cells were seeded in 60‐mm glass‐bottom dishes and incubated at 37 °C for 24 h. Following stabilization, 4T1 cells were treated with 150 nm aCD47 or aCD47‐C‐miR34a in a serum‐free medium for 24 h. Treated 4T1 cells were subsequently stained with pHrodo‐SE (Thermo Fisher Scientific, USA) for 30 min at room temperature and then co‐cultured with stained BMDMs at a 1:4 ratio for 2 h. After co‐culture, cells were washed with pH 10 PBS to remove cells non‐engulfed cells. Fluorescence images were acquired using an Evos M5000 fluorescence microscopy (Thermo Fisher Scientific, USA).

Apoptosis Assay

To evaluate the induction of cell death by aCD47‐C‐miR34a, 1 × 10⁵ 4T1 cells were plated in 6‐well plates and incubated at 37 °C for 24 h. After stabilization, cells were treated with 130 nm miR‐34a or aCD47‐C‐miR34a for 48 h. Following incubation, cells were washed, resuspended in 500 µL of binding buffer, and stained with FITC‐conjugated Annexin V and propidium iodide (PI) for 5 min at room temperature in the dark. Apoptotic cells were quantified using a CytoFLEX flow cytometer (Beckman Coulter, USA) and analyzed with FlowJo software (v10).

Quantitative Reverse Transcription Polymerase Chain Reaction (qRT‐PCR)

To analyze the expression level of miR‐34a, 1.5 × 10⁵ 4T1 cells were seeded into 6‐well plates and incubated at 37 °C for 24 h. After stabilization, cells were treated with aCD47‐C‐miR34a at a final concentration of 130 nm in serum‐free medium for 20 h, with varying molar ratios of the conjugate. The miRNeasy mini kit (Qiagen, Germany) was utilized for total miRNA extraction. miRNA samples were reverse transcribed into cDNA by the Mir‐X miRNA First Strand Synthesis kit (Takara Bio, Japan) and subsequently eluted with 40 µL of RNase‐free water. qRT‐PCR was conducted using the StepOnePlus Real‐Time PCR System (Applied Biosystems, USA) and TOPreal SYBR Green qPCR Kits (Enzynomics, Republic of Korea) to amplify cDNA. Ct values were normalized to U6 RNA. The relative levels of miR‐34a were quantified using the 2^−ΔΔCt method. All qRT‐PCR procedures for miRNAs were performed following the manufacturer's instructions. The primers used in this study were as listed in the table below.

Gene	Sequence (5′‐3′)
miR‐34a	F: 5′‐CGC GTG GCA GTG TCT TAG CT‐3′ R: 5′‐AGT GCA GGG TCC GAG GTA TT‐3′
U6	F: 5′‐GCT TCG GCA CAT ATA CTA AAA T‐3′ R: 5′‐CGC TTC ACG AAT TTG CGT GTC AT‐3′
PD‐L1	F: 5′‐CCA TCC TGT TGT TCC TCA TTG‐3′ R: 5′‐CAC TGC TTA CGT CTC CTC G‐3′
SIRT1	F: 5′‐GAC GAT GAC AGA ACG TCA CAC‐3′ R: 5′‐CGA GGA TCG GTG CCA ATC A‐3′
18s	F: 5′‐TTC CGA TAA CGA ACG AGA CTC T‐3′ R: 5′‐TGG CTG AAC GCC ACT TGT C‐3′

(F: Forward; R: Reverse)

Western Blot Analysis

For western blot analysis, 1.5 × 10⁵ 4T1 cells were seeded into 6‐well plates and incubated overnight at 37 °C. The following day, cells were treated with 130 nm aCD47‐C‐miR34a in a serum‐free medium for 24 h. An equal amount of cell lysate protein (30 µg) was quantified using the BCA Protein Assay Kit (Bio‐Rad, USA) and electrophoresed on a 10% sodium dodecyl sulfate‐polyacrylamide gel electrophoresis (SDS‐PAGE) gel for 90 min at 100 V. The proteins were subsequently transferred onto nitrocellulose membranes, blocked with 5% skim milk for 1 h, and incubated with primary antibodies overnight at 4 °C. Next day, the membranes were washed three times with TBS‐T (pH 7.4, 20 mm Tris, 150 mm NaCl, and 0.1 Tween 20) for 10 min, and then incubated with secondary antibodies at room temperature for 1 h. Finally, protein signals were visualized using an ECL substrate (Bio‐Rad, USA) and detected via chemiluminescence on the iBright Imaging System (Invitrogen, USA). The antibodies used in this study were as follows: PD‐L1 (Abcam, UK, ab213480; dilution 1:1000), SIRT1 (Cell signaling, USA, 8469T; dilution 1:1000), glyceraldehyde 3‐phosphate dehydrogenase (GAPDH) (GeneTex, USA, TX100118; 1:1000), anti‐rabbit IgG‐horseradish peroxidase (HRP) antibody (GeneTex, USA, GTX213110–01; 1:2000), and anti‐mouse IgG‐HRP antibody (GeneTex, USA, GTX213111 01; 1:2000).

In Vivo Biodistribution Study

All animal care and experimental procedures were performed in accordance with guidelines of the Institutional Animal Care and Use Committee (IACUC; approved number of KIST‐2022‐04‐072) in Korea Institute of Science and Technology (KIST). The mice were purchased from Orient Bio (Republic of Korea) and bred under pathogen‐free conditions at KIST. For in vivo biodistribution studies, orthotopic TNBC models were established by implanting 1 × 10⁵ 4T1 cells into the mammary fat pad of 7‐week‐old female BALB/c nude mice. To generate a 4T1 tumor‐xenograft mouse model, 2 × 10⁶ 4T1 cells were subcutaneously inoculated into the left flank of 7‐week‐old female BALB/c nude mice. When the tumor volumes reached 100–150 mm³, the mice administered an intravenous injection of IgG‐C‐miR34a‐Cy5, miR‐34a‐Cy5, or aCD47‐C‐miR34a‐Cy5 (4 nmol/mouse). In vivo and ex vivo noninvasive NIRF imaging were conducted using an In Vivo Imaging System (IVIS) Lumina Series III system device (PerkinElmer, USA) after 1, 3, and 6 h of post‐injection. For ex vivo fluorescence imaging, 4T1 tumor‐bearing mice were sacrificed 6 h post‐injection. Tumors and major organs, including the liver, lung, spleen, and heart, were then harvested for analysis. Finally, harvested tumors were embedded in an Optimal Cutting Temperature compound to facilitate cryo‐sectioning and stained with Hoechst 33 342. Fluorescence images of the sectioned tissues were then obtained using a TCS SP8 confocal laser microscope (Leica Microsystems, Germany).

In Vivo Antitumor Effect of aCD47‐C‐miR34a

To investigate the antitumorigenic effects of aCD47‐C‐miR34a, 1 × 10⁵ 4T1‐Luc cells were implanted on a female BALB/c mouse mammary fat pad. When the average tumor volume reached ≈40–50 mm³ on day 6 following inoculation, all mice (n = 6) were randomly divided into three treatment groups (PBS, aCD47, and aCD47‐C‐miR34a). Mice were intravenously injected via the tail vein with aCD47 (5 mg kg⁻¹) or aCD47‐C‐miR34a (5 mg kg⁻¹) every 3 days, for a total of four doses administered on days 6, 9, 12, and 15 post‐tumor inoculation. Tumor sizes and body weights were measured every 2 days. The volume of the tumor was calculated as (width²× length)/2. Luminescence imaging was conducted after intraperitoneal injection of luciferin (30 mg mL⁻¹, Promega, USA) on days 6, 9, 12, and 15.

The antitumor efficacy was evaluated by sacrificing mice 17 days after the final injection. Tumor tissues were dissociated into single‐cell suspensions using a MACS Tumor Dissociation Kit (Miltenyi Biotec, USA) and a gentleMACS Octo Dissociator with Heaters (Miltenyi Biotec, USA). To remove red blood cells (RBCs), an RBC lysis buffer (BioLegend, USA) was applied to the cell suspensions. After cell counting, FcBlock was introduced to the cell suspensions to prevent nonspecific binding for 15 min at 4 °C. The collected cells were washed three times with PBS to resuspend the pellets as single cells. All samples were analyzed using a CytoFLEX flow cytometer (Beckman Coulter, USA), and data were processed using the FlowJo (v10) software.

The antibodies used for the analysis of activated CD8⁺ T‐cells were obtained from BioLegend (USA): PE‐anti‐CD45.2 (clone 104, #109807), PerCp Cy5.5‐anti‐CD3 (clone 17A2, #100218), APC‐anti‐CD8 (clone 53‐6.7, #100711), PE‐anti‐Ki67 (clone 11F6, #151209), FITC‐anti‐CD44 (clone IM7, #103005), and FITC‐anti‐Granzyme B (clone GB11, #515403). The antibodies used for the analysis of DC maturation and activation were obtained from BioLegend (USA): PE‐anti‐CD11c (clone N418, #117 307), APC‐anti‐CD40 (clone 3/23, #124611), FITC‐anti‐CD86 (clone GL‐1, #105005), and PerCP Cy5.5 anti‐CD80 (clone 16‐10A1, #104722).

Histological Analysis

For histological analysis, tumor tissues were harvested from the 4T1 tumor‐bearing mouse model on the day after the final injection and fixed in a 4% PFA solution (Biosesang, Republic of Korea). For the analysis of the PD‐L1 inhibition and CD8⁺ T‐cell activation induced by aCD47‐C‐miR34a, immunofluorescence staining for PD‐L1, CD8, and Granzyme B was conducted on tumor tissues harvested from the 4T1 TNBC model one day after the final injection. The tissues were fixed in 4% PFA, embedded in paraffin, and sectioned into 8‐µm thick slices using a rotary microtome (Accu‐cut SRM200 Rotary Microtome, Japan). The sections were pre‐blocked with a solution containing 1% BSA and 0.5% Triton X‐100 in PBS to avoid nonspecific binding for 15 min at room temperature. After three washes with TBS‐T, the slides were incubated overnight at 4 °C with APC‐anti‐PD‐L1 (clone 10F.9G2, Cat #: 124311), APC‐ anti‐CD8 (clone 53‐6.7, #100711), and FITC‐anti‐Granzyme B antibodies (clone GB11, #515403). Nuclei were stained with Hoechst 33342 for 10 min.

A fluorescent TUNEL assay (Promega) was performed following the manufacturer's instructions to detect apoptotic cells within the tumor tissues. The resulting images were analyzed with ImageJ software to measure relative fluorescence intensity in comparison to the PBS group.

In Vivo Toxicity Assessment

The major organs were fixed in 4% PFA, embedded in paraffin, and sectioned at 6‐µm thick for analysis of histology. To detect any histopathological changes, these sections were stained with H&E and examined using an Olympus BX51 microscope (Tokyo, Japan). Blood samples were collected from the abdominal vena cava into an EDTA‐coated vacutainer (Becton Dickinson, USA) for serum biochemical profiling. The counts of total WBC, including neutrophil, lymphocyte, and monocyte levels, were analyzed by DKKorea Co. (Republic of Korea), a certified nonclinical contract research organization.

Statistical Analysis

All statistical analyses were performed using Prism 8 (GraphPad) and presented as the mean ± standard deviation (SD). Statistical significance between groups was calculated using student's t‐tests or one‐way/two‐way ANOVA followed by Tukey's post‐hoc test for multiple comparisons. p values<0.05 were considered statistically significant.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Ryu Y., Kim E. H., Jang H., et al. “Targeted Delivery of miR‐34a via Anti‐CD47 Antibody Conjugates for Enhanced Cancer Immunotherapy in Triple Negative Breast Cancer.” Small 21, no. 36 (2025): 21, e04468. 10.1002/smll.202504468

Data Availability Statement

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