Abstract

   The immunosuppressive tumor microenvironment (TME) critically
   undermines the efficacy of T cell-based tumor immunotherapy by impeding
   CD8^+ T cell infiltration and cytotoxic function, primarily through
   tumor-associated macrophages (TAMs) and immune checkpoint molecules
   such as programmed death ligand 1 (PD-L1). Here, we present a
   multifunctional nanoplatform, IN@OMV-PDL1nb, designed to simultaneously
   inhibit TAM-derived immunosuppressive metabolite itaconic acid (ITA) by
   targeting immune-responsive gene 1 (IRG1) and block PD-L1 within the
   TME. Engineered outer membrane vesicles (OMVs) serve as precision
   delivery vehicles for the IRG1 inhibitor IRG1-IN-1 (IN) and as carriers
   for PD-L1 nanobody release, activated by matrix metalloproteinase-2
   (MMP-2). IN@OMV-PDL1nb effectively inhibits IRG1 expression in TAMs,
   thus reducing the accumulation of ITA, restoring chemokines (CXCL9 and
   CXCL10) secretion, and enhancing CD8^+ T cells infiltration within
   tumors. The released PD-L1 nanobody protects CD8^+ T cells, preserving
   their tumoricidal activity. In murine tumor models, IN@OMV-PDL1nb
   significantly inhibited tumor growth, increased survival, and enhanced
   antigen presentation and T cell recruitment. Additionally,
   IN@OMV-PDL1nb induced robust adaptive immunity, facilitating
   antigen-specific immune memory that prevented tumor recurrence and
   metastasis. This dual-targeting approach offers a promising strategy to
   overcome TME-driven immunosuppression in tumor immunotherapy.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12951-025-03507-7.

   Keywords: Tumor immunotherapy, Tumor microenvironment, Immune
   checkpoint blockade, Tumor-associated macrophages, Outer membrane
   vesicles

Graphical Abstract

   [62]graphic file with name 12951_2025_3507_Figa_HTML.jpg
   [63]Open in a new tab

   A multifunctional nanoplatform, IN@OMV-PDL1nb@CaP, is engineered
   through genetic modifications to express a matrix metalloproteinase-2
   (MMP-2)-responsive peptide (PLGLAG) and the PD-L1 nanobody (PD-L1nb) on
   its surface, while IRG1-IN-1 is loaded into its internal cavity.
   Additionally, calcium phosphate (CaP) is biomineralized on the surface.
   This platform effectively alleviates immune suppression induced by
   tumor-associated macrophages (TAMs) and the immune checkpoint molecule
   PD-L1 within the tumor microenvironment (TME), offering a promising
   approach for tumor immunotherapy

Introduction

   The recruitment, infiltration, and functional activity of T cells
   within tumor tissues are critical determinants of the success of
   anti-tumor immunotherapy [[64]1, [65]2]. However, the efficacy of T
   cell-mediated immune responses is severely constrained by the
   immunosuppressive tumor microenvironment (TME) [[66]3–[67]5].
   Tumor-associated macrophages (TAMs), the most abundant immune cells
   within tumors, are key players in establishing this suppressive milieu
   [[68]6, [69]7]. Tumor cells induce high expression of IRG1 in TAMs via
   activation of the nuclear factor-κB (NF-κB) pathway. The downstream
   metabolite of IRG1, ITA, suppresses the production of
   macrophage-derived chemokines such as CXCL9 and CXCL10, reducing CD8^+
   T cell infiltration into tumor tissues [[70]8, [71]9]. Furthermore, the
   infiltrating T cells often fail to execute their cytotoxic functions
   due to the high expression of immune checkpoints, such as PD-L1, within
   the TME. This leads to programmed exhaustion and functional impairment
   of T cells [[72]10–[73]12]. Thus, a strategy that combines IRG1
   inhibition in TAMs with PD-L1 blockade to remodel the TME holds
   significant potential for improving anti-tumor immunotherapy.

   One major challenge in targeting IRG1 lies in the lack of cellular
   specificity of its inhibitors, making selective delivery to TAMs
   crucial for regulating their functions. OMVs, naturally secreted
   spherical nanoparticles from Gram-negative bacteria (20–250 nm in
   size), offer a promising solution [[74]13–[75]15]. OMVs possess unique
   biological properties, including their composition of
   pathogen-associated molecular patterns (PAMPs) such as
   lipopolysaccharide (LPS), flagellin, and peptidoglycan, which render
   them highly immunogenic and capable of activating multiple toll-like
   receptor (TLR) signaling pathways [[76]16–[77]18]. OMVs are readily
   recognized and phagocytosed by macrophages and, due to their internal
   cavity structure, serve as ideal carriers for delivering therapeutic
   agents specifically to TAMs [[78]19, [79]20]. This makes OMVs an ideal
   delivery vehicle for selectively targeting TAMs and modulating their
   immunosuppressive functions within the TME.

   In this study, we developed a multifunctional nanoplatform,
   IN@OMV-PDL1nb, to simultaneously target IRG1 in TAMs and block the
   PD-L1 in TME, thereby remodeling the TME and enhancing anti-tumor
   immunity. Using genetically engineered Escherichia coli (E. coli), OMVs
   were designed to express MMP-2-responsive peptide (PLGLAG) and PD-L1
   nanobody (PDL1nb) on their surface (OMV-PDL1nb) [[80]21, [81]22]. Next,
   we used electroporation to package the IRG1 inhibitor IRG1-IN-1 into
   the internal cavity of OMV-PDL1nb to form IN@OMV-PDL1nb (Fig. [82]1A).
   In the tumor site, the abundant MMP-2 cleaves its substrate peptide
   between OMVs and PDL1nb, releasing PDL1nb to block PD-L1 and restore
   CD8^+ T cell cytotoxicity. Simultaneously, the activation of TAMs by
   IN@OMVs is accompanied by efficient phagocytosis, releasing IRG1-IN-1
   within these cells. The inhibition of IRG1 reduces the production of
   ITA, restoring the secretion of chemokines such as CXCL9 and CXCL10,
   which facilitate the recruitment and infiltration of CD8^+ T cells into
   the tumor (Fig. [83]1B). Subsequently, neoantigens released by dead
   tumor cells were drained into lymph nodes, which induced stronger
   adaptive immunity and long-term immune memory that prevented tumor
   recurrence and metastasis in the mice (Fig. [84]1C). This
   dual-targeting strategy effectively overcomes immune suppression within
   the TME, promoting synergistic anti-tumor responses and offering a
   promising platform for advancing tumor immunotherapy.

Fig. 1.

   [85]Fig. 1
   [86]Open in a new tab

   Schematic illustration of construction and antitumor strategy of
   IN@OMV-PDL1nb. (A) The preparation of IN@OMV-PDL1nb involved
   engineering E. coli to express the MMP-2-responsive peptide (PLGLAG)
   and the PD-L1 nanobody (PDL1nb) on the surface protein ClyA, resulting
   in OMV-PDL1nb. Subsequently, the IRG1 inhibitor IRG1-IN-1 was loaded
   into OMV-PDL1nb via electroporation to form IN@OMV-PDL1nb. (B) In the
   tumor site, IN@OMV-PDL1nb facilitates the recruitment and infiltration
   of CD8^+ T cells by increasing the secretion of chemokines. It also
   blocks PD-L1 through the released PDL1nb, thereby restoring CD8^+ T
   cell cytotoxicity, which can then attack tumor cells effectively. (C)
   Neoantigens released from dead tumor cells are drained into the lymph
   nodes, which can induce stronger adaptive immunity and promote
   long-term immune memory in mice. MMP-2, matrix metalloproteinase-2.
   TLR4, toll-like receptor 4. IRG1, immune-responsive gene 1. ITA,
   itaconic acid

Materials and methods

Materials

   The following antibodies for flow cytometry were purchased from
   BioLegend Inc. (USA): anti-mouse CD3ε-PerCP/Cyanine5.5 (100327),
   anti-mouse CD8α-FITC (100706), anti-mouse CD45-PE/Cyanine7 (103114),
   anti-mouse CD11b-PE (101208), anti-mouse F480-FITC (123108), anti-mouse
   CD86-APC (105012), anti-mouse CD11b-PerCP/Cyanine5.5 (101227),
   anti-mouse Gr-1-APC (108411), anti-mouse IFN-γ-PE (505807), anti-mouse
   MHCI-OVA-PE (141603), anti-mouse CD44-PE (103007), anti-mouse CD62L-APC
   (108411), anti-mouse MHC-I-APC (116517), anti-mouse MHC-II-Brilliant
   Violet421 (107631), anti-mouse CD40-APC (124611), anti-mouse CD11c-FITC
   (117305), anti-mouse CD80-Brilliant Violet421 (104725), anti-mouse
   PD-1-PE (135205), anti-mouse LAG-3- APC (125209), anti-mouse TIM-3-APC
   (119705), anti-mouse Ly6G-FITC (127605). Isopropyl-β-d-thiogalactoside
   (I8070), kanamycin (K8020) and TRIzol (R1100) were purchased from
   Solarbio Biotechnology Co., Ltd. (China). Anti-FLAG antibody (M185-3L)
   was purchased from MBL (Japan). And the secondary antibody (7076 S)
   used in the western blot procedure was purchased from Cell Signaling
   Technology Inc. (USA). Enhanced chemiluminescence reagents (SQ202L-2)
   was purchased from Epizyme (USA). IRG1-IN-1 (HY-148335) was purchased
   from MedChemExpress (USA). αPD-L1 (BP0101) was purchased from Bio X
   cell (USA). Cy5.5-NHS ester (A8103) was purchased from APExBIO (USA).
   Phalloidin-iFluor 594 conjugate (ab176757) was purchased from Abcam
   (UK). Hoechst 33,342 (C1022) was purchased from Beyotime (China).
   Recombinant MMP-2 Protein (902-MP-010) was purchased from R&D Systems
   (USA). Murine macrophage colony-stimulating factor (M-CSF, 315-02-10),
   murine IL-4 (214-14-20), and murine IFN-γ (315-05-20) were purchased
   from Peprotech (USA). Mouse IFN-γ precoated enzyme-linked immunospot
   assay (ELISPOT) kit (2210005) was purchased from BioLegend Inc. (USA).
   Erythrocyte lysate (BL503B) was purchased from Biosharp (China).
   C[13]-labeled ITA (I931004) was purchased from Toronto Research
   Chemicals (Canada). ELISA kit of CXCL9 (70-EK2143) and CXCL10 (EK268)
   were purchased from Multisciences (China). ELISA kit of MMP-2 (EK0460)
   were purchased from Boster Biological Technology Co., Ltd. (China).
   PD-L1 protein and ELISA detection reagent kit of PD-L1 (SEKM-0259) were
   purchased from Solarbio Biotechnology Co., Ltd. (China). ELISA kit of
   TNF-α (1217202) and IL-6 (1210602) were purchased from Dakewe
   Biotechnology Co., Ltd. (China). The antibody used for
   immunofluorescence analysis of CD8^+ T cells infiltration within tumor
   tissues was the following: Anti-CD8α antibody (Abcam, USA, ab217344).
   The antibody used for immunohistochemical analysis of PD-L1expression
   within tumor tissues was the following: PD-L1 polyclonal antibody
   (Proteintech, USA, 17952-1-AP). OVA[257–264] peptide (SIINFEKL) was
   purchased from Qiangyao Biotechnology Co., Ltd. (China).

Plasmid construction and OMV-PDL1nb Preparation

   The gene encoding ClyA-FLAG-PLGLAG-PDL1nb was cloned into the pET28a
   vector to construct the expression plasmid (GENEWIZ, China). Then, the
   plasmid was transformed into Escherichia coli (E. coli) BL21 (DE3) to
   form engineered bacteria. The engineered bacteria were inoculated into
   LB medium (50 µg mL^− 1 kanamycin was added) at 37 °C with shaking
   (180 rpm). When the OD[600] reached 0.6, IPTG (1 mM in LB medium) was
   added to induce protein expression at 16 °C for 16 h with shaking
   (180 rpm). The collected bacterial suspension was centrifuged at 5000
   ×g for 15 min at 4 °C to remove bacteria, and the supernatant was then
   filtered through 0.45 μm and 0.22 μm filters (BIOFIL, China, FPE404030
   and FPE204030). Next, the filtrate was ultracentrifuged at 150,000 ×g
   for 3 h at 4 °C to collect OMV-PDL1nb. The pellet was resuspended in
   phosphate buffered saline (PBS), filtered using a 0.22 μm filter and
   stored at -80 °C until use. OMV-PDL1nb was characterized using TEM
   (HITACHI, Japan, HT7700) and DLS (Malvern, UK, Zetasizer Nano ZS90).

Western blot analysis

   OMVs were resuspended in IP lysis buffer and lysed at 4 °C for 30 min,
   followed by centrifugation at 13,000 × g for 15 min at 4 °C, and the
   supernatant was collected. The protein concentration of the supernatant
   was quantified using the BCA protein assay Kit (Thermo Scientific, USA,
   23228 and 1859078). Protein samples were electrophoresed using sodium
   dodecyl sulfide-polyacrylamide gels and subsequently transferred to a
   polyvinylidene fluoride membrane. The membranes were blocked in 5%
   nonfat milk and then incubated with an anti-FLAG antibody overnight at
   4 °C. The immunoreactive proteins were visualized using enhanced
   chemiluminescence reagents.

Preparation of IN@OMVs, IN@OMV-PDL1nb and IN@OMV-PDL1nb@CaP

   IRG1-IN-1 was loaded into OMVs or OMV-PDL1nb by electroporation to
   prepare IN@OMVs and IN@OMV-PDL1nb, respectively. Briefly, 100 µg
   IRG1-IN-1 and 100 µg OMVs or 100 µg OMV-PDL1nb were separately mixed
   and suspended in electroporation cuvettes, and electroporation was
   performed using a MicroPulser Electroporator (Bio-Rad, USA) at 700 V
   and 50µF. Then, the free IRG1-IN-1 was removed by ultracentrifugation
   at 150,000 ×g for 3 h at 4 °C. The encapsulation efficiency of
   IRG1-IN-1 in IN@OMVs or IN@OMV-PDL1nb was determined by UV-vis
   spectrum. To prepare IN@OMV-PDL1nb@CaP, IN@OMV-PDL1nb (1 mg vesicle
   protein) was suspended in 1 mL of DMEM overnight at 4 °C, and then
   calcium chloride (CaCl[2], the final concentration:1 mmol/L) was added
   to the reaction, followed by incubation at 37 °C for 2 h. Next, the
   biomineralized OMV-PDL1nb was washed 2–3 times using ultrapure water
   and following centrifugated at 14,000 × g for 10 min at 4 °C to obtain
   IN@OMV-PDL1nb@CaP. IN@OMVs, IN@OMV-PDL1nb, and IN@OMV-PDL1nb@CaP were
   characterized using TEM and DLS.

Animals and cell culture

   Female C57BL/6 mice (6–8-week-old) were purchased from SPF
   Biotechnology Co., Ltd. (Beijing, China). All animal studies were
   approved by the Ethics Committee of Tianjin Medical University Cancer
   Institute and Hospital. B16-F10, B16-OVA, Pan02, MC38 and MC38-OVA
   Cells were obtained from the National Center for Nanoscience and
   Technology. B16-F10, B16-OVA, Pan02, MC38 and MC38-OVA cells were
   cultured in DMEM supplemented with 10% fetal bovine serum, 100 U mL^− 1
   penicillin-streptomycin. These cells were tested mycoplasma-negative.
   The cells were cultured at 37 °C in a humidified atmosphere incubator
   with 5% CO[2]. To prepare for tumor cell-conditioned medium, MC38,
   Pan02 or B16-F10 tumor cells were cultured for 24 h, then the
   supernatant was collected and filtered through 0.45 μm filters and
   stored at -20 °C until use.

Analysis of the binding specificity and affinity between PD-L1nb and PD-L1,
and the cleavage efficiency of substrate peptide PLGLAG on OMV-PDL1nb by
MMP-2

   Firstly, OMVs (12.5 µg vesicle protein per well) and OMV-PDL1nb
   (12.5 µg vesicle protein per well) were separately spread into 96 well
   plates and incubated overnight at 4 ℃ to prepare coated ELISA plates.
   Then, the plates were blocked by 5% nonfat milk. And the concentration
   of MMP-2 in mouse MC38-OVA tumors was detected using ELISA. Next, MMP-2
   was added into the OMV-PDL1nb coated wells and incubated for 0 h,
   0.5 h, 1 h, and 2 h at 37 ℃, respectively. Finally, the binding
   specificity and affinity between PD-L1nb and PD-L1, and the cleavage
   efficiency of substrate peptide PLGLAG in OMV-PDL1nb by MMP-2 were
   measured using ELISA as recommended by the manufacturer’s protocol.

The released PD-L1 nanobody’s function in binding PD-L1

   In brief, OMV-PDL1nb was mixed with MMP-2 for 0 h, 0.5 h, 1 h, and 2 h
   at 37 ℃, respectively. Then, the supernatant which containing the
   released PD-L1 nanobody on OMV-PDL1nb cleaved by MMP-2 was collected
   using ultracentrifugation at 150,000 × g for 3 h at 4 °C. Next, the
   released PD-L1 nanobody was separately spread into 96 well plates to
   prepare coated ELISA plates, followed by detection of the released
   nanobody’s function in binding PD-L1 as mentioned above.

Cell binding assay in vitro

   To assess the binding of OMV-PDL1nb onto tumor cells, OMVs and
   OMV-PDL1nb were labeled using Cy5.5-NHS. The free Cy5.5-NHS was removed
   by ultracentrifugation at 150,000 ×g for 3 h at 4 °C. Then,
   Cy5.5-labeled vesicles (10 µg vesicle protein mL^− 1) were incubated
   with MC38 cells for 2 h at 4 °C. For blockade of PD-L1, MC38 cells were
   pre-incubated with anti-PD-L1 (10 µg mL^− 1 αPD-L1) for 12 h at 4 °C.
   In addition, to detect the expression of MMP-2 substrate peptide
   PLGLAG, MMP-2 (0.1 µg mL ^− 1) was added after the binding of
   OMV-PDL1nb onto MC38 cells. Next, the cells were washed three times
   with PBS, and then the binding state was detected by flow cytometry
   (BECKMAN, USA). To further visualize OMV-PDL1nb binding onto tumor
   cells, MC38 cells were treated with the formulations previously
   mentioned, followed by staining for cytoskeleton actin with
   phalloidin-iFluor 594 conjugate and labeling of cell nuclei with
   Hoechst 33,342. Finally, the binding state was detected using CLSM
   (Zeiss, LSM700, Germany).

BMDMs isolation and culture

   Briefly, bone marrow cells in the femur and tibia of mice were
   harvested and then cultured in DMEM medium containing 10% fetal bovine
   serum, 100 U mL^− 1 penicillin-streptomycin, 0.05 mM β-ME and 20 ng
   mL^− 1 M-CSF. Half of the medium was replaced every 3 days until the
   bone marrow cells differentiated into M0-type BMDMs on day 7. To obtain
   M2-polarized BMDMs, M0-type BMDMs were treated with 40 ng mL^− 1 IL-4
   for 48 h.

Cellular uptake of IN@OMVs and IN@OMV-PDL1nb in vitro

   To assess cellular uptake, IN@OMVs and IN@OMV-PDL1nb were prepared and
   labeled using Cy5.5-NHS as previously mentioned. IN@OMVs (12.5 µg
   vesicle protein mL^− 1) or IN@OMV-PDL1nb (12.5 µg vesicle protein
   mL^− 1) were incubated with BMDMs at 37 °C for 4 h and then the cell
   nuclei of BMDMs were stained with Hoechst 33,342. Afterward, the
   cellular uptake was detected by CLSM. In addition, to detect the
   cellular time-uptake efficiency of IN@OMVs and IN@OMV-PDL1nb by
   macrophages, Cy5.5-NHS labeled IN@OMVs (12.5 µg vesicle protein mL^− 1)
   and IN@OMV-PDL1nb (12.5 µg vesicle protein mL^− 1) were incubated with
   BMDMs at 37 °C for 0 h, 0.5 h, 1 h, 2 h and 4 h at 37℃, respectively.
   Next, these BMDMs were stained with Hoechst 33,342. Finally, the
   cellular time-uptake efficiency was quantified using flow cytometry.

Analysis of BMDMs polarization and antigen presentation capacity

   For BMDMs polarization assay, M0-type BMDMs or M2-polarized BMDMs were
   incubated with PBS, OMVs (12.5 µg vesicle protein mL^− 1) or IN@OMVs
   (12.5 µg vesicle protein mL^− 1) for 24 h, respectively. Then, the
   M1-related marker (CD86) in the CD11b^+ F4/80^+ macrophages was
   evaluated by flow cytometry. For BMDMs antigen presentation capacity
   assay, BMDMs were incubated with PBS, OMVs (12.5 µg vesicle protein
   mL^− 1) or IN@OMVs (12.5 µg vesicle protein mL^− 1) for 24 h,
   respectively. Then, the antigen presentation capacity-related markers
   (MHC-I, MHC-II, CD40) in the CD11b^+ F4/80^+ macrophages were detected
   by flow cytometry.

Quantitative real-time PCR analysis

   BMDMs were incubated with TCM, OMVs (12.5 µg vesicle protein mL^− 1),
   TCM + IRG1-IN-1 (3.25 µg mL^− 1), TCM + OMVs (12.5 µg vesicle protein
   mL^− 1) or TCM + IN@OMVs (12.5 µg vesicle protein mL^− 1) for 6 h,
   respectively. Then, total RNA of cells was extracted using TRIzol
   reagent and reversely transcribed to synthesize the DNA, followed by
   qRT-PCR reactions to detect the IRG1, CXCL9 and CXCL10 mRNA expression,
   respectively. The qRT-PCR primer sequences were as follows: IRG1-F:
   AGTTTTCTGGCCTCGACCTG, IRG1-R: AGAGGGAGGGTGGAATCTCT; CXCL9-F:
   GAGCAGTGTGGAGTTCGAGG, CXCL9-R: TCCGGATCTAGGCAGGTTTG; CXCL10-F:
   AATGAGGGCCATAGGGAAGC, CXCL10-R: AGCCATCCACTGGGTAAAGG; Rn18S-F:
   CGCGGTTCTATTTTGTTGGT, Rn18S-R: AGTCGGCATCGTTTATGGTC.

Intracellular ITA quantification by LC-MS/MS

   Briefly, BMDMs were incubated with TCM, TCM + IRG1-IN-1 (3.25 µg
   mL^− 1), TCM + OMVs (12.5 µg vesicle protein mL^− 1) or TCM + IN@OMVs
   (12.5 µg vesicle protein mL^− 1) for 12 h, respectively. Next, the
   cells were washed three times with PBS. After that, the cells were
   fixed by adding 1 ml of 80% (v/v) chilled (-80 °C) methanol containing
   C[13]-labeled ITA as an internal standard. The cell extracts were
   analyzed by ultrahigh performance liquid chromatography (Thermo
   scientific, UAS, vanquish UHPLC) coupled to an orbitrap Mass
   Spectrometer (Thermo scientific, UAS, Orbitrap Exploris 480). And the
   intracellular ITA quantification of macrophages within tumor sites was
   analyzed in the same way.

Enzyme-linked immunosorbent assay (ELISA)

   BMDMs were incubated with TCM, TCM + OMVs (12.5 µg vesicle protein
   mL^− 1) or TCM + IN@OMVs (12.5 µg vesicle protein mL^− 1) for 24 h, and
   then the cell-free supernatants were obtained for the cytokines
   assessment by ELISA kit of CXCL9 and CXCL10 as recommended by the
   manufacturer′s protocol.

T cell migration assay

   The migration ability of CD8^+ T cells was detected by Transwell assay.
   In brief, 1 × 10^6 BMDMs were incubated with TCM, TCM + OMVs (12.5 µg
   vesicle protein mL^− 1) or TCM + IN@OMVs (12.5 µg vesicle protein
   mL^− 1) for 24 h at the bottom chamber. At the same time, the CD8^+ T
   cells were isolated from the spleens of wild-type C57BL/6 mice (female,
   6–8-week-old) by MojoSort™ Mouse CD8^+ T Cell Isolation Kit (BioLegend,
   USA, 480008). Then, the CD8^+ T (1 × 10^6) cells were placed in the top
   chamber. The number of migrated CD8^+ T cells from the top to the
   bottom chamber was counted after incubation for 12 h at 37 °C in a
   humidified atmosphere incubator with 5% CO[2].

RNA sequencing

   Firstly, 1 × 10^6 MC38 cells were injected subcutaneously into the
   right back of 6–8-week-old female C57BL/6 mice to generate mouse tumor
   models. On day 9, mice were randomly divided into three experimental
   groups: Control, OMVs (625 µg vesicle protein/kg body weight) and
   IN@OMVs (625 µg vesicle protein/kg body weight). OMVs, IN@OMVs were
   injected intratumorally, the control mice were injected intratumorally
   with an equal volume of solvent, and tumors were harvested from mice
   after treatment. Next, and the CD11b^+ F4/80^+ -labeled cells (TAMs)
   were sorted by FACS. Cellular RNA was isolated from TAMs using the
   TRIzol reagent and then the samples were immediately shipped frozen to
   BGI Genomics Co., Ltd. China for transcriptome sequencing.

Evaluation of antitumor efficacy in vivo

   To establish mouse tumor models, 1 × 10^6 MC38-OVA cells were injected
   subcutaneously into the right back of 6–8-week-old female C57BL/6 mice.
   On day 9, mice were randomly divided into six experimental groups:
   Control, αPD-L1(10 mg/kg body weight), IRG1-IN-1 (162.5 µg /kg body
   weight), IN@OMVs (625 µg vesicle protein/kg body weight), OMV-PDL1nb
   (625 µg vesicle protein/kg body weight), IN@OMV-PDL1nb (625 µg vesicle
   protein/kg body weight). IRG1-IN-1 and OMV-based formulations were
   injected intratumorally, αPD-L1 was injected intraperitoneally in mice,
   and the control mice were injected intratumorally with an equal volume
   of solvent. The treatment was given every 3 days for a total of 3
   rounds. Tumor size was measured during treatment using a vernier
   caliper, and the volume of the tumor was calculated as 1/2 × (length ×
   width^2). When the tumor volume exceeded 2000 mm^3, mice were
   sacrificed and the data were used to generate survival curves. On day
   16, a fraction of the mice was sacrificed for adaptive immunity
   analysis. On day 22, some mice were sacrificed and the tumors were
   harvested for further immune response analysis. For safety evaluation,
   the main organs (heart, liver, spleen, lung, and kidneys) of mice were
   harvested for the H&E staining, while the levels of ALT, AST, ALP, BUN,
   CREA, and LDH in serum were analyzed. The remaining animals continued
   to monitor tumor volume up to day 60, and the data were used to
   generate survival curves.

Immune response analysis in vivo

   For adaptive immunity analysis, splenocytes were harvested and cultured
   in RPMI1640 medium containing 10% fetal bovine serum. The splenocytes
   were stimulated with OVA[257–264] peptide (10 µg mL^− 1) at 37 °C in a
   humidified atmosphere incubator with 5% CO[2] for 16 h, and the IFN-γ^+
   cells in CD3^+CD8^+ T cells were detected by flow cytometry, while
   IFN-γ secretion was analyzed using the mouse IFN-γ precoated ELISPOT
   kit as recommended by the manufacturer′s protocol. Concurrently, the
   levels of cytokines (TNF-α, IFN-γ and IL-6) in tumor tissues and the
   serum of mice were detected to evaluate immune activation. On day 16,
   the tumor tissues and serum of some mice were harvested. Tumor tissues
   were homogenized, and the homogenates were centrifuged at 12,000 × g
   for 15 min at 4 °C to collect the supernatant. Then, the levels of
   TNF-α, IFN-γ and IL-6 in tumor tissues and the serum of mice were
   detected using ELISA as recommended by the manufacturer’s protocol. To
   investigate potential off-target inflammatory effects of IN@OMV-PDL1nb
   treatment, the proportions of CD3^+ T lymphocytes and neutrophils
   within spleens of mice were analyzed by flow cytometry. And the levels
   of lymphocytes and neutrophils in blood of mice were detected using
   automatic hematology analyzer. In addition, lymphocytes from the
   tumor-side draining lymph nodes were isolated and then prepared as
   single cell suspensions. The DCs were stained with anti-mouse CD11c and
   MHCI-OVA antibodies, followed by detection of antigen presentation in
   DCs using flow cytometry. On day 22, subcutaneous tumor tissues were
   prepared as single cell suspensions, and then tumor cells were stained
   using antibodies. Next, the proportions of CD8^+ T cells, M1-phenotype
   macrophages, MHC-II^+ macrophages and MDSCs in the tumor tissues were
   analyzed by flow cytometry. In addition, part of the tumor tissues was
   sectioned for detection of CD8^+ T infiltration and PD-L1 expression by
   immunofluorescence and immunohistochemistry, respectively.

Evaluation of antitumor efficacy of IN@OMV-PDL1nb@CaP treatment

   The mouse tumor models were established as mentioned above. On day 5,
   mice were randomly divided into three experimental groups: Control,
   IN@OMV-PDL1nb (625 µg vesicle protein/kg body weight) and
   IN@OMV-PDL1nb@CaP (2500 µg vesicle protein/kg body weight).
   IN@OMV-PDL1nb was injected intratumorally, IN@OMV-PDL1nb@CaP was
   injected intravenously in mice, and the control mice were injected with
   an equal volume of solvent. The treatment was given on day 5 and day 9
   for a total of 2 rounds. Tumor size was measured during treatment every
   two days. The intracellular levels of ITA in macrophages within tumors
   after treatment measured by LC-MS/MS on day 10. On day 15, the mice
   were sacrificed and the tumors were harvested to evaluate the antitumor
   efficacy of IN@OMV-PDL1nb@CaP treatment.

Evaluation of immune memory

   Female C57BL/6 mice (6–8-week-old) were injected subcutaneously into
   the right back with 1 × 10^6 MC38-OVA cells. On day 7, mice were
   randomly divided into two experimental groups: Control and
   IN@OMV-PDL1nb (625 µg vesicle protein/kg body weight). Mice were
   injected intratumorally with IN@OMV-PDL1nb or the same solvent for 3
   rounds. On day 18, tumors in these two groups were removed by surgery.
   On day 60, the mice were rechallenged by subcutaneously inoculating
   with 1 × 10^6 MC38-OVA cells or injecting intravenously with 2 × 10^5
   B16-OVA cells. The subcutaneous tumor volume was monitored every four
   days. The mice were sacrificed for immune memory analysis and
   examination for lung metastasis nodules on day 80. For evaluation of
   memory T cells, the splenocytes were harvested from the mice and
   stained with anti-CD3ε, anti-CD8α, anti-CD44 and anti-CD62L antibodies
   followed by detection using flow cytometry. The proportion of IFN-γ^+
   cells in CD3^+CD8^+ T cells within splenocytes was detected as
   previously mentioned.

Statistical analysis

   Data were analyzed using GraphPad Prism software version 8 and
   presented as the mean ± standard deviation (SD) and represented at
   least three independent experiments. The P values were determined using
   a two-tailed unpaired t test or one-way ANOVA with a Tukey post-hoc
   test. The P values of survival analysis were calculated using log-rank
   (Mantel-Cox) test. And the values of P < 0.05 were considered
   statistically significant. *P < 0.05, **P < 0.01, ***P < 0.001,
   ****P < 0.0001, n.s., no significance.

Results

Preparation and characterization of OMV-PDL1nb, in@omvs and IN@OMV-PDL1nb

   To obtain OMV-PDL1nb, the genes encoding ClyA (one of the most abundant
   proteins on the surface of OMVs), PLGLAG and PDL1nb were recombined in
   the expression plasmid pET28a. The resulting fusion protein was then
   expressed in E.coli BL21 (DE3) (Fig. [87]2A). Western blot analysis
   confirmed the expression of ClyA-PLGLAG-PDL1nb in OMV-PDL1nb following
   Isopropyl-β-d-thiogalactoside (IPTG) induction, as evidenced by the
   detection of the FLAG tag (Fig. [88]2B). Transmission electron
   microscopy (TEM) and dynamic light scattering (DLS) were employed to
   characterize the OMVs generated from the original E.coli and OMV-PDL1nb
   generated from the engineered E.coli. Both OMVs and OMV-PDL1nb
   exhibited regular spherical double-layer membrane structures, with
   average diameters of 20–30 nm (Fig. [89]2C). The zeta potential of OMVs
   was approximately − 20 mV, while that of OMV-PDL1nb measured around
   − 24 mV (Fig. [90]2D). We then performed an ELISA to evaluate the
   binding specificity and affinity of PD-L1nb (on OMV-PDL1nb) for PD-L1.
   Our results showed that PD-L1 can efficiently bind to OMV-PDL1nb,
   significantly more than to OMVs. Additionally, the substrate peptide
   PLGLAG on OMV-PDL1nb was rapidly cleaved by MMP-2, with a cleavage rate
   of approximately 74% at 0.5 h, and increasing to nearly complete
   cleavage (99%) at 2 h (Supplementary Fig. [91]1). Concurrently, we
   collected the PD-L1 nanobody released from OMV-PDL1nb cleaved by MMP-2
   at different time points and assessed its functionality in binding
   PD-L1 using ELISA. The results demonstrate that the released PD-L1
   nanobody retains full functionality and effectively binds to PD-L1
   (Supplementary Fig. [92]2). Subsequently, an in vitro tumor cell
   binding assay was performed to evaluate the affinity between PD-L1nb
   (on OMV-PDL1nb) and PD-L1 (expressed on tumor cells). In this
   experiment, both OMVs and OMV-PDL1nb were labeled with Cy5.5. Utilizing
   flow cytometry and confocal laser scanning microscopy (CLSM), it was
   demonstrated that OMV-PDL1nb effectively adhered to the surface of MC38
   cells. Furthermore, pre-incubation with anti-PD-L1 antibodies
   significantly inhibited the binding of OMV-PDL1nb to the MC38 cell’s
   surface, thereby confirming the specificity of the interaction between
   PD-L1nb (on OMV-PDL1nb) and PD-L1 (on tumor cells). Additionally,
   OMV-PDL1nb bound to tumor cells was effectively cleaved by MMP-2,
   further validating the expression of ClyA-PLGLAG-PDL1nb in OMV-PDL1nb
   (Fig. [93]2E-F).

Fig. 2.

   [94]Fig. 2
   [95]Open in a new tab

   Characterization of OMV-PDL1nb, IN@OMVs and IN@OMV-PDL1nb. (A)
   Schematic representation of the pET28a-ClyA-PLGLAG-PDL1nb expression
   plasmid construct. (B) Western blot analysis of ClyA-PLGLAG-PDL1nb
   expression in OMV-PDL1nb using an anti-FLAG antibody.
   Isopropyl-β-d-thiogalactoside (IPTG) was utilized as the inducer for
   protein expression. (C) TEM images and size distribution of OMVs and
   OMV-PDL1nb. Scale bar = 100 nm. d., diameter. (D) The zeta potentials
   of OMVs and OMV-PDL1nb measured by DLS (n = 3). (E-F) Flow cytometry E)
   and CLSM F) analysis of the affinity between PDL1nb (on OMV-PDL1nb) and
   PD-L1 (on MC38 cells). MC38 cells were treated with the indicated
   formulations for 2 h. In group III, MC38 cells were pre-incubated with
   anti-PDL1 for 12 h to block PD-L1. In group IV, MMP-2 was added after
   the binding of OMV-PDL1nb to tumor cells. Cell nuclei of MC38 were
   stained with Hoechst 33342 (blue). The cytoskeleton actin of MC38 cells
   were stained with phalloidin-iFluor 594 conjugate (green) and OMVs,
   OMV-PDL1nb were labeled with Cy5.5(red). Scale bar = 20 μm. (G) TEM
   images and size distribution of IN@OMVs and IN@OMV-PDL1nb. Scale bar
   = 100 nm. d., diameter. (H) The zeta potentials of IN@OMVs and
   IN@OMV-PDL1nb measured by DLS (n = 3). (I) IRG1-IN-1 encapsulation
   efficiency of IN@OMVs and IN@OMV-PDL1nb were determined using
   UV-visible spectrum (n = 3). (J-K) Cellular uptake of IN@OMVs and
   IN@OMV-PDL1nb in BMDMs were detected using CLSM. Cell nuclei of BMDMs
   were stained with Hoechst 33342 (blue), while IN@OMVs and IN@OMV-PDL1nb
   were labeled with Cy5.5 (red). Scale bar = 50 μm. The average Cy5.5
   fluorescence intensity per macrophage of the ELISA experiment in the
   IN@OMVs group was used as the control. Data were processed by GraphPad
   Prism software version 8 and presented as the mean ± SD. The P value
   was calculated using a two-tailed unpaired t test. n.s., no
   significance

   Based on the structural characteristics of OMVs with internal cavity
   structure, we encapsulated IRG1-IN-1 into OMVs or OMV-PDL1nb to form
   IN@OMVs or IN@OMV-PDL1nb via electroporation, respectively. TEM
   observations and DLS analysis revealed that both IN@OMVs and
   IN@OMV-PDL1nb exhibited regular spherical double-layer membrane
   structures, with diameters of approximately 20–30 nm (Fig. [96]2G). The
   zeta potential of IN@OMVs was approximately − 23 mV, while that of
   IN@OMV-PDL1nb was around − 27 mV (Fig. [97]2H). Furthermore, UV-visible
   spectrum determined that the encapsulation efficiency of IRG1-IN-1 in
   both IN@OMVs and IN@OMV-PDL1nb was about 26% (Fig. [98]2I).
   Subsequently, the cellular uptake efficiency of IN@OMVs and
   IN@OMV-PDL1nb by bone marrow-derived macrophages (BMDMs) was assessed
   by CLSM, it was demonstrated that IN@OMVs and IN@OMV-PDL1nb could be
   effectively uptaken by BMDMs, and the uptake of IN@OMV-PDL1nb was
   higher, indicating that the PD-L1nb might enhance its macrophage
   targeting (Fig. [99]2J-K). Moreover, flow cytometry analysis revealed a
   time-dependent increase in cellular uptake of both IN@OMVs and
   IN@OMV-PDL1nb by macrophages, with uptake efficiencies of 19% and 21%
   at 0.5 h, respectively, and 84% and 88% at 4 h (Supplementary
   Fig. [100]3). It was evident that the rate of cellular uptake was
   significantly slower than the cleavage rate of PD-L1nb from OMV-PDL1nb,
   indicating that PD-L1nb on OMV-PDL1nb could be efficiently released
   before phagocytosis by macrophages.

Evaluation of the role of IN@OMVs in vitro

   Firstly, we treated BMDMs with different types of tumors conditioned
   medium (TCM) and OMVs and found that both of these treatments could
   induce high IRG1 mRNA expression in BMDMs (Supplementary Fig. [101]4)
   [[102]23]. Next, we investigated the IRG1 mRNA expression and ITA
   accumulation in BMDMs treated by IRG1-IN-1 and IN@OMVs using qRT-PCR
   and liquid chromatography-tandem mass spectrometry (LC-MS/MS),
   respectively. It was demonstrated that MC38-TCM highly induced IRG1
   expression and ITA accumulation in BMDMs, and IRG1-IN-1 treatment
   significantly reduced their expression levels. However, the combination
   of MC38-TCM and OMVs treatment further increased the expression levels
   of IRG1 and ITA. Encouragingly, treatment with IN@OMVs significantly
   reduced their expression levels, closed to the TCM + IRG1-IN-1
   treatment group, indicating the advantages of OMV-based nanoplatforms
   for drug delivery (Fig. [103]3A-B).

Fig. 3.

   [104]Fig. 3
   [105]Open in a new tab

   The effect of IN@OMVs on macrophages in vitro. BMDMs were treated with
   the indicated formulations. TCM, MC38 cell-tumor conditioned medium.
   (A-B) IRG1 mRNA expression A) and ITA accumulation B) in BMDMs measured
   by qRT-PCR (n = 4) and LC-MS/MS (n = 3), respectively. (C-D) Transwell
   assay was used to analyze the number of CD8^+ T cells migration
   (n = 3). (E-H) CXCL9 E-F) and CXCL10 G-H) expression in BMDMs measured
   by qRT-PCR (E, G; n = 4) and ELISA (F, H; n = 3). (I-K) Analysis of
   antigen presentation of macrophages (n = 3). Flow cytometry evaluated
   the antigen presentation-related markers (MHC-I, MHC-II, CD40) in the
   CD11b^+F4/80^+ BMDMs. (L-M) Analysis of macrophages M1 polarization
   (n = 3). Flow cytometry evaluated the M1-related marker (CD86) in the
   CD11b^+F4/80^+ BMDMs. Data were processed by GraphPad Prism software
   version 8 and presented as the mean ± SD. The P values were determined
   using one-way ANOVA with a Tukey post-hoc test. **P < 0.01,
   ***P < 0.001, and ****P < 0.0001, n.s., no significance

   Subsequently, we examined the effect of macrophages treated with OMVs
   and IN@OMVs on the migration ability of CD8^+ T cells by transwell
   assay, which showed that OMVs treatment increased the number of CD8^+ T
   cells migrating, while macrophages treated with IN@OMVs further
   increased CD8^+ T cell migration (Fig. [106]3C-D). The results of
   quantitative real-time PCR (qRT-PCR) and enzyme-linked immunosorbent
   assays (ELISA) showed that OMVs treatment increased macrophage-derived
   CXCL9 and CXCL10 mRNA expression and extracellular secretion. Moreover,
   the mRNA expression and extracellular secretion of CXCL9 and CXCL10 in
   macrophages were further increased after IN@OMVs treatment
   (Fig. [107]3E-H). These results indicate that IN@OMVs can efficiently
   inhibite IRG1 expression and ITA accumulation in macrophages, thereby
   increasing the expression of macrophage-derived CXCL9 and CXCL10 and
   enhancing the ability of macrophages to promote CD8^+ T cell migration.

   As an important antigen presenting cell, macrophages play a crucial
   role in antigen presentation [[108]24]. The antigen presentation
   capacity of macrophages was evaluated by flow cytometry, and the
   results showed that compared with the PBS (control) and OMVs-treated
   groups, IN@OMVs treatment significantly increased the antigen
   presentation capacity of macrophages, as indicated by higher expression
   of antigen presentation-related markers MHC-I, MHC-II and CD40 on the
   surface of macrophages (Fig. [109]3I-K, Supplementary Fig. [110]5). In
   addition, the results of flow cytometry demonstrated that IN@OMVs
   treatment could significantly promote the M0-type and M2-polarized
   macrophages to M1-phenotype macrophages (Fig. [111]3L-M).

   These results indicate that IN@OMVs treatment can reduce IRG1 mRNA
   expression and ITA accumulation levels in macrophages induced by tumor
   cells and OMVs, thereby reversing macrophage dysfunction caused by
   excessive ITA accumulation.

Transcriptome profiling of TAMs after IN@OMVs treatment

   To further investigate the mechanisms by which IN@OMVs affect
   macrophages, we isolated TAMs from the tumors of control, OMVs and
   IN@OMVs treated mice using fluorescence-activated cell sorting (FACS)
   for transcriptome analysis (Fig. [112]4A). The results indicated that
   OMVs treatment led to an increased expression of IRG1 in TAMs compared
   with the control group, whereas IN@OMVs treatment effectively inhibited
   the expression of IRG1 in these cells. Additionally, OMVs treatment
   enhanced the expression of genes associated with chemotaxis, including
   Ccl2, Ccl7, Ccl8, Ccl12, Cxcl9, and Cxcl10. Notably, these genes were
   even more significantly upregulated following treatment with IN@OMVs.
   Simultaneously, IN@OMVs treatment markedly increased the expression of
   antigen presentation-related genes in TAMs, such as Cd40, Cd86, Icam1,
   H2-M2, H2-Aa, H2-Ab1, Tap1, and Tap2. Furthermore, the results showed
   that OMVs treatment reduced the expression of genes associated with M2
   polarization, such as Cd163 and protein kinase AMP-activated catalytic
   subunit alpha 1 (Prkaa1), with IN@OMVs treatment leading to even more
   significant downregulation of these genes [[113]25]. IN@OMVs treatment
   markedly decreased the expression of Mmp8 and Mmp9, genes that promote
   tumor metastasis [[114]26], compared to other treatment groups. In
   addition, IN@OMVs treatment significantly reduced the expression of
   carcinoembryonic antigen-related cell-adhesion molecule 1 (ceacam1) in
   tumor-associated macrophages (TAMs), a molecule involved in
   angiogenesis and tumorigenesis (Fig. [115]4B) [[116]27, [117]28]. These
   findings corroborate the results of previous experiments in vitro. Gene
   Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG)
   pathway enrichment analyses revealed that OMVs treatment enhanced the
   activation of antigen processing and presentation pathways, as well as
   the NF-κB pathway and response to interferon-β (IFN-β) in TAMs compared
   to the control group (Fig. [118]4C-E). In comparison with the control
   and OMVs treatment, the IN@OMVs treatment significantly enhanced the
   activation of pathways related to the response to IFN-β, antigen
   processing and presentation, and NF-kB in TAMs (Fig. [119]4F-H,
   Supplementary Fig. [120]6). Concurrently, GO pathway enrichment
   analysis revealed that OMVs treatment downregulated cell growth and
   cell adhesion pathways in TAMs compared to the control group
   (Supplementary Fig. [121]7A). Notably, IN@OMVs treatment further
   decreased the activation of pathways related to angiogenesis, cell-cell
   adhesion, and growth factor activity in TAMs (Supplementary
   Fig. [122]7B-C). Additionally, Gene Set Enrichment Analysis (GSEA)
   demonstrated downregulation of hypoxia and epithelial-mesenchymal
   transition (EMT) pathways after IN@OMVs treatment compared to OMVs
   treatment, while Reactome gene set analysis indicated suppression of
   IL-4, IL-13, and IL-10 signaling pathways (Supplementary Fig. [123]7D,
   and F-G). Similar results were observed in the comparison between
   IN@OMVs and the control group, with angiogenesis also being
   downregulated (Supplementary Fig. [124]7E, and H-I). These findings
   suggest that IN@OMVs treatment modulates the tumor microenvironment to
   inhibit tumor progression and immune escape. Previous studies have
   indicated that the activation of the NF-kB pathway in TAMs can lead to
   the expression of downstream genes such as Irg1, Cxcl9, and Cxcl10.
   ITA, the downstream metabolite of IRG1, has the capacity to inhibit the
   persistent activation of the NF-kB pathway, resulting in decreased
   expression of CXCL9 and CXCL10, which ultimately hinders T cell
   infiltration [[125]8, [126]9]. In this study, we demonstrated that OMVs
   can also activate the NF-kB pathway in TAMs, leading to elevated IRG1
   expression and subsequent suppression of CXCL9 and CXCL10. IN@OMVs,
   leveraging its efficient drug delivery capabilities, maintained the
   activation of the NF-kB pathway while delivering IRG1-IN-1 into TAMs,
   thereby reducing ITA accumulation. This reduction eliminated the
   inhibition of sustained NF-kB pathway activation and the expression of
   CXCL9 and CXCL10 by ITA (Fig. [127]4I). Furthermore, IN@OMVs also
   facilitated the activation of pathways related to antigen processing
   and presentation in TAMs.

Fig. 4.

   [128]Fig. 4
   [129]Open in a new tab

   Transcriptome analysis of TAMs in mice after treatment. (A) Schematic
   illustration of the experiment schedule. (B) Gene expression heatmap in
   TAMs from the MC38 tumor-bearing mice (n = 3). (C-E) Transcriptomic
   changes in TAMs after OMVs treatment compared with the control group.
   C) The volcano plot showed the number of differentially expressed genes
   distribution. P value (-Log[10], y axis) and the fold change (Log[2], x
   axis). Significant changes (P value < 0.05,|Log[2] Fold change| > 0.3)
   were indicated in red (upregulation) and blue (downregulation). D) Gene
   ontology (GO) enrichment pathways of up-regulated genes. Ontology:
   Biological process (BP), Cell component (CC) and Molecular function
   (MF). E) Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment
   pathways of up-regulated genes. (F-H) Transcriptomic changes in TAMs
   after IN@OMVs treatment compared with OMVs treatment group. F) The
   volcano plot showed the number of differentially expressed genes
   distribution. P value (-Log[10], y axis) and the fold change (Log[2], x
   axis). Significant changes (P value < 0.05,|Log[2] Fold change| > 0.3)
   were indicated in red (upregulation) and blue (downregulation). G) GO
   enrichment pathways of up-regulated genes. H) KEGG enrichment pathways
   of up-regulated genes. (I) Schematic illustration of IN@OMVs triggering
   TLR4-NF-κB pathway activation in macrophages

Antitumor effects of and TME remodeling by IN@OMV-PDL1nb

   In order to evaluate the antitumor effects of IN@OMV-PDL1nb, we
   constructed a subcutaneous tumor-bearing mouse model using MC38-OVA
   cells (murine colon cancer cells expressing ovalbumin). The mice
   received three intratumoral injections of either αPD-L1, IRG1-IN-1,
   IN@OMVs, OMV-PDL1nb, or IN@OMV-PDL1nb, as illustrated in Fig. [130]5A.
   Initially, we assessed the adaptive immunity induced by the
   IN@OMV-PDL1nb treatment on day 16. Flow cytometry analysis revealed a
   higher proportion of MHCI-OVA^+ DCs in the tumor-draining lymph nodes
   of the IN@OMV-PDL1nb treatment group on the seventh day post-treatment,
   indicating that IN@OMV-PDL1nb effectively enhanced antigen presentation
   by DCs (Fig. [131]5B). Additionally, we observed a significant increase
   in both the volume and weight of the spleen in the IN@OMV, OMV-PDL1nb,
   and IN@OMV-PDL1nb treatment groups, which may be attributed to the
   systemic immune enhancement provided by OMVs (Supplementary
   Fig. [132]8A-B). Furthermore, analysis of IFN-γ secretion through flow
   cytometry and enzyme-linked immunospot (ELISPOT) assays demonstrated
   that IN@OMV-PDL1nb treatment significantly activated OVA-specific CD8^+
   T cells within splenocytes (Fig. [133]5C, Supplementary
   Fig. [134]9A-B). These specific CD8^+ T cells can effectively recognize
   and kill MC38-OVA cells after they infiltrating into the tumor tissues.
   And cytokine levels (TNF-α, IFN-γ, and IL-6) were measured in tumor
   tissues and serum of mice following treatment. IN@OMV-PDL1nb treatment
   significantly enhanced the secretion of these cytokines compared to
   other treatments, suggesting its ability to promote immune activation
   (Supplementary Fig. [135]10). Simultaneously, we assessed immune cell
   populations in non-tumor tissues to evaluate potential off-target
   inflammatory effects of IN@OMV-PDL1nb treatment, and the results showed
   mild elevations in systemic immune cell levels, all remaining within
   the normal range, indicating no off-target inflammatory effects while
   activating the immune system (Supplementary Fig. [136]11).
   Additionally, the volume of the draining lymph nodes on the tumor side
   in mice significantly increased following IN@OMV-PDL1nb treatment,
   indicating an accumulation of immune cells in these lymph nodes
   (Supplementary Fig. [137]12). These results suggest that after
   IN@OMV-PDL1nb treatment, tumor cells are eliminated, releasing tumor
   antigens that are subsequently taken up and processed by DCs in the
   tumor-draining lymph nodes. These processed antigens are then presented
   to T cells, thereby inducing a robust tumor antigen-specific adaptive
   immune response. Furthermore, the LC-MS/MS assay demonstrated that
   IN@OMV-PDL1nb treatment significantly reduced ITA levels in macrophages
   within the tumors, which is critical for promoting the infiltration of
   CD8^+ T cells into the tumor tissues (Supplementary Fig. [138]13). At
   the endpoint of this experiment, we assessed the antitumor effects of
   the indicated treatment approaches. The results indicated that the
   IN@OMV-PDL1nb treatment was more effective in inhibiting tumor growth
   compared with αPD-L1, IRG1-IN-1, IN@OMVs, and OMV-PDL1nb treatments.
   Mice treated with IN@OMV-PDL1nb exhibited significantly smaller tumor
   volumes than those in the other groups (Fig. [139]5D-E, Supplementary
   Fig. [140]14). To investigate the role of IN@OMV-PDL1nb treatment in
   the TME, we assessed the infiltration of immune cells in the tumor
   tissue post-treatment using flow cytometry. The IN@OMV-PDL1nb treatment
   notably increased the proportion of infiltrating CD8^+ T cells in the
   tumor compared with other treatments (Fig. [141]5F). Concurrently,
   IN@OMV-PDL1nb treatment significantly reduced the proportion of
   exhausted CD8^+ T cells in the tumor microenvironment, enhancing their
   functional state (Supplementary Fig. [142]15), thereby reinforcing its
   robust anti-tumor efficacy. Additionally, this treatment significantly
   enhanced the proportion of M1-phenotype macrophages within the tumor
   tissues and elevated the expression of markers associated with
   macrophage antigen presentation (Fig. [143]5G-H). Further flow
   cytometry analysis revealed that IN@OMV-PDL1nb treatment reduced the
   proportion of infiltrating myeloid-derived suppressor cells (MDSCs) in
   the tumor tissues (Fig. [144]5I). And the survival time of mice treated
   with IN@OMV-PDL1nb was significantly prolonged (Fig. [145]5J).
   Furthermore, the assay of immunofluorescence revealed that
   IN@OMV-PDL1nb treatment markedly increased the proportion of
   infiltrating CD8^+ T cells in tumor tissues (Fig. [146]5K).
   Immunohistochemical analyses indicated that IN@OMV-PDL1nb treatment
   significantly reduced PD-L1 expression in tumor tissues (Fig. [147]5L).
   In addition, we evaluated the systemic toxicity associated with
   IN@OMV-PDL1nb treatment. The hematoxylin and eosin (H&E) staining
   images of the main organs in mice demonstrated that IN@OMV-PDL1nb
   treatment did not result in noticeable organ damage (Supplementary
   Fig. [148]16). Concurrently, the assessment of serum biochemical
   indices, including alanine transaminase (ALT), aspartate transaminase
   (AST), alkaline phosphatase (ALP), blood urea nitrogen (BUN),
   creatinine (CREA), and lactate dehydrogenase (LDH), further confirmed
   the safety of the IN@OMV-PDL1nb treatment (Supplementary Fig. [149]17).
   In summary, these experimental results suggest that IN@OMV-PDL1nb
   treatment effectively induces adaptive immunity, thereby remodeling the
   TME and generating potent anti-tumor effects.

Fig. 5.

   [150]Fig. 5
   [151]Open in a new tab

   The antitumor effects of IN@OMV-PDL1nb in MC38-OVA subcutaneous tumor
   model. (A) Schematic illustration of the experiment schedule. (B) The
   proportion of MHCI-OVA^+ cells in CD11c^+ DCs in the tumor side
   draining lymph nodes on day 16 (n = 5). (C) The proportion of IFN-γ^+
   cells in CD3^+CD8^+ T cells in spleen on day 16 (n = 5). (D) The images
   of tumors after treatment (n = 6). Scale bar = 1 cm. (E) Tumor weight
   (n = 6). (F-I) The proportions of CD8^+ T cells, M1-phenotype
   macrophages, MHC-II^+ macrophages and MDSCs within tumor tissues
   (n = 6). (J) Survival curves (n = 6). (K) Immunofluorescence (IF)
   staining of CD8^+ cells in tumors of mice. Scale bar = 20 μm. (L)
   Immunohistochemical (IHC) images showing the expression of PD-L1 in
   tumor tissue of mice. Scale bar = 50 μm. Data were processed by
   GraphPad Prism software version 8 and presented as the mean ± SD. The P
   values of survival analysis were calculated using log-rank (Mantel-Cox)
   test. And other P values were determined using one-way ANOVA with a
   Tukey post-hoc test. *P < 0.05, **P < 0.01, ***P < 0.001, and
   ****P < 0.0001, n.s., no significance

Long-term immune memory induced by IN@OMV-PDL1nb

   The challenge of tumor immunotherapy lies not only in whether the
   selected treatment regimen can achieve clinical cure or remission, but
   also in the effective control of long-term tumor recurrence and
   metastasis. Previous studies have reported that OMVs can induce
   long-term immune memory effects, thanks to their natural size
   advantages and efficient immune activation ability on DCs. OMVs can be
   efficiently drained into lymph nodes. In addition, the abundant PAMPs
   molecules on the surface of OMVs can effectively promote the rapid
   maturation of DCs in lymph nodes. Mature DCs can process, present tumor
   antigens, and activate T cells, thereby inducing strong adaptive
   immunity and long-term immune memory effects. The adaptive immune and
   immune memory effects induced by OMVs are significantly superior to
   conventional immunotherapy strategies [[152]29, [153]30]. Thus, we
   investigated the immune memory effects induced by IN@OMV-PDL1nb
   treatment (Fig. [154]6A). We constructed the subcutaneous MC38-OVA
   tumor-bearing mice model and treated the mice for three times. Compared
   with the control group, IN@OMV-PDL1nb treatment significantly inhibited
   the tumor growth (Fig. [155]6B). Following this treatment, we
   surgically excised subcutaneous tumors from the mice. Then we
   inoculated mice subcutaneously with MC38-OVA or intravenously with
   B16-OVA cells on day 60. The results indicated that in mice treated
   with IN@OMV-PDL1nb, the re-challenged subcutaneous tumors regressed
   almost completely (Fig. [156]6C). Moreover, the mice treated with
   IN@OMV-PDL1nb had significantly fewer metastasis nodules in the lungs
   compared with the control group (Fig. [157]6D-E). Additionally, flow
   cytometry analysis revealed that the proportions of effector memory T
   cells and central memory T cells in the splenocytes of IN@OMV-PDL1nb
   treated mice were significantly elevated compared with the control mice
   on day 80 (Fig. [158]6F-H). Compared with the control group, the
   proportion of IFN-γ^+ cells in CD3^+CD8^+ T cells within spleens of the
   IN@OMV-PDL1nb treated mice was significantly increased, suggesting that
   the immune system of the mice treated with IN@OMV-PDL1nb could still
   recognize and response to tumor antigens efficiently (Fig. [159]6I).
   These findings imply that IN@OMV-PDL1nb treatment can elicit
   antigen-specific immune memory, which contributes to a long-term tumor
   prevention effect against tumor recurrence and metastasis.

Fig. 6.

   [160]Fig. 6
   [161]Open in a new tab

   Evaluation of long-term immune memory induced by IN@OMV-PDL1nb. (A)
   Schematic illustration of the experiment schedule. (B) Tumor growth
   curves of the first subcutaneously implanted MC38-OVA cells in each
   mouse (n = 5). (C) Tumor growth curves of the re-challenge of
   subcutaneous inoculation with MC38-OVA cells in each mouse (n = 5).
   (D-E) The images of lung and numbers of metastatic nodules in each
   mouse re-challenged with intravenous injection of B16-OVA cells
   (n = 5). (F-H) The proportions of effector memory T cells and central
   memory T cells in spleen (n = 5). (I) The proportion of IFN-γ^+ cells
   in CD3^+CD8^+ T cells in splenocytes (n = 5). Data were processed by
   GraphPad Prism software version 8 and presented as the mean ± SD. The P
   value was calculated using a two-tailed unpaired t test. **P < 0.01,
   ****P < 0.0001

Evaluation of antitumor efficacy by IN@OMV-PDL1nb@CaP treatment in vivo

   In our study, we initially employed intratumoral injection of
   IN@OMV-PDL1nb, which demonstrated potent anti-tumor effects. Building
   upon this, following the methodology established in our previous work
   [[162]31], we encapsulated IN@OMV-PDL1nb in CaP to form
   IN@OMV-PDL1nb@CaP, which retained the spherical double-layer membrane
   structure with an average diameter of approximately 120 nm
   (Fig. [163]7A-C), and a zeta potential of around − 15 mV
   (Fig. [164]7D). To further evaluate its potential for systemic
   administration, we developed a subcutaneous tumor-bearing mouse model
   using MC38-OVA cells. The mice were administered two intratumoral
   injection of IN@OMV-PDL1nb or intravenous (i.v.) injections of
   IN@OMV-PDL1nb@CaP. The results showed that both formulations
   effectively inhibited tumor growth, with similar tumor volumes observed
   in the IN@OMV-PDL1nb@CaP and IN@OMV-PDL1nb treatment groups
   (Fig. [165]7E-H). Furthermore, LC-MS/MS analysis revealed that
   IN@OMV-PDL1nb@CaP treatment significantly reduced ITA levels in
   macrophages within the tumor tissues (Fig. [166]7I), confirming the
   therapeutic potential of this formulation. These results suggest that
   while intratumoral injection offers direct therapeutic benefits,
   encapsulation of IN@OMV-PDL1nb in CaP also allows for enhanced tumor
   targeting via intravenous administration, expanding the potential
   application of this dual-target nanoplatform.

Fig. 7.

   [167]Fig. 7
   [168]Open in a new tab

   The antitumor effects of IN@OMV-PDL1nb@CaP in MC38-OVA subcutaneous
   tumor model. (A) The Schematic illustration of the preparation of
   IN@OMV-PDL1nb@CaP. (B-C) TEM image and size distribution of
   OMV-PDL1nb@CaP. Scale bar = 100 nm. d., diameter. (D) The zeta
   potential of OMV-PDL1nb@CaP measured by DLS (n = 3). (E) Schematic
   illustration of IN@OMV-PDL1nb@CaP treatment. (F) Tumor growth curves in
   each mouse (n = 5). (G) The images of tumors after treatment (n = 5).
   (H) Tumor weight (n = 5). (I) The intracellular levels of ITA in
   macrophages within tumors after treatment measured by LC-MS/MS on day
   10 (n = 3). Data were processed by GraphPad Prism software version 8
   and presented as the mean ± SD. The P values were determined using
   one-way ANOVA with a Tukey post-hoc test. ***P < 0.001,****P < 0.0001,
   n.s., no significance

Discussion

   Tumor immunotherapy, particularly through the use of PD-1/PD-L1
   inhibitors, has demonstrated clinical efficacy in certain tumor types;
   however, the overall therapeutic effect of these inhibitors remains
   limited [[169]32–[170]34]. The low response rates associated with
   PD-1/PD-L1 inhibitors may be attributed to the immunosuppressive nature
   of the tumor microenvironment, which mainly includes inadequate T cell
   infiltration in tumor sites [[171]35, [172]36]. Consequently,
   activating immune remodeling within the tumor microenvironment, in
   conjunction with PD-1/PD-L1 inhibitor therapy, represents a promising
   strategy for enhancing tumor immunotherapy [[173]37–[174]39].

   ITA, a byproduct of the Krebs cycle, accumulates within macrophages
   following the activation of the NF-κB pathway by tumor cells. ITA
   inhibits TET DNA dioxygenase, subsequently reducing the secretion of
   chemokines, which ultimately disrupts CD8^+ T cells infiltration into
   tumor sites [[175]8, [176]9, [177]40, [178]41]. Therefore, the targeted
   inhibition of ITA production within macrophages is crucial for
   remodeling the TME.

   OMVs contain abundant PAMPs and possess a strong capacity for immune
   activation, effectively engaging TLRs such as TLR2, TLR4, and TLR5,
   among other innate immune pathways [[179]42–[180]45]. Furthermore, OMVs
   are readily obtainable and amenable to genetic engineering
   modifications, which has garnered increasing interest in the field of
   tumor immunotherapy. OMVs can serve as vectors for tumor vaccines or
   immune activators to elicit anti-tumor responses [[181]46, [182]47]. In
   addition, their internal cavity structure and easily phagocytosis by
   macrophages, positioning them as excellent drug delivery vehicles. In
   our previous research, we modified OMVs to investigate their potential
   as therapeutic tumor vaccines and activators of tumor immunotherapy,
   yielding promising results [[183]29, [184]30, [185]48, [186]49]. As a
   significant component of OMVs, LPS can stimulate tumor immunity via the
   TLR4-NF-kB pathway [[187]50]. Notably, in this study, we observed that
   OMVs induced IRG1 overexpression and ITA accumulation within
   macrophages while activating the expression of chemokine-related genes.
   This finding suggests that LPS-mediated immune activation in OMVs may
   represent a double-edged sword; while LPS activates the TLR4-NF-kB
   pathway in macrophages to enhance immunity, it concurrently promotes
   the production of immunosuppressive factors, such as ITA, which
   accumulate in macrophages and potentially inhibit the efficacy of
   immune activation. This phenomenon may explain the limited
   effectiveness of OMVs monotherapy in tumor treatment. Currently,
   efforts to reduce the endotoxin levels of LPS in OMVs are primarily
   achieved through genetic modification of the parent strain, which
   introduces new challenges, particularly regarding the diminished
   capacity of LPS to activate immunity [[188]29, [189]51, [190]52].
   Therefore, a strategic approach that preserves LPS-induced immune
   activation while mitigating the associated immunosuppression could
   enhance the anti-tumor efficacy of OMVs.

Conclusions

   In conclusion, IN@OMV-PDL1nb nanoplatform effectively inhibits
   TAM-derived ITA through IRG1 suppression while concurrently blocking
   PD-L1 to restore CD8^+ T cell infiltration and cytotoxic functions. The
   platform leverages the unique biological properties of engineered OMVs
   for precise delivery of therapeutic agents to TAMs, ensuring targeted
   modulation of the TAMs in the TME. In preclinical models, IN@OMV-PDL1nb
   not only inhibited tumor growth and enhanced survival but also promoted
   robust antigen presentation, adaptive immune activation, and long-term
   immune memory. These findings underscore the potential of IN@OMV-PDL1nb
   to overcome critical barriers in T cell-based tumor immunotherapy,
   offering a promising strategy for improving tumor treatment outcomes.

Electronic supplementary material

   Below is the link to the electronic supplementary material.
   [191]Supplementary Material 1^ (2.5MB, pdf)

Acknowledgements