Potentiating cancer immunotherapies with modular albumin-hitchhiking nanobody-STING agonist conjugates

Abstract

The enhancement of antitumour immunity via agonists of the stimulator of interferon genes (STING) pathway is limited by pharmacological barriers. Here we show that the covalent conjugation of a STING agonist to anti-albumin nanobodies via site-selective bioconjugation chemistries prolongs the circulation of the agonist in the blood and increases its accumulation in tumour tissue, stimulating innate immune programmes that increased the infiltration of activated natural killer cells and T cells, which potently inhibited the growth of mouse tumours. The technology is modular, as demonstrated by the recombinant integration of a second nanobody domain targeting programmed death-ligand 1 (PD-L1), which further increased the accumulation of the agonist in tumours while blocking immunosuppressive PD-1/PD-L1 interactions. The bivalent nanobody-STING agonist conjugate stimulated robust antigen-specific T-cell responses and long-lasting immunological memory and conferred enhanced therapeutic efficacy. It was also effective as a neoadjuvant treatment to adoptive T-cell therapy. As a modular approach, hitchhiking STING agonists on serum albumin may serve as a broadly applicable strategy for augmenting the potency of systemically administered cancer immunotherapies.

Fig. 1. Design, synthesis and in vitro characterization of an anti-albumin nanobody for site-selective conjugation of STING agonists.

a, Scheme depicting the concept of an albumin-hitchhiking nanobody–STING agonist conjugate for cancer immunotherapy. Anti-albumin nanobodies conjugated to STING agonists bind to circulating albumin in situ, resulting in improved pharmacokinetics and increased biodistribution to tumour sites that stimulates antitumour innate and adaptive immune responses. b, Computational model of the anti-albumin nanobody (nAlb) binding at domain IIB of HSA. c, ITC traces (top) and binding isotherms (bottom) of nAlb binding to human and mouse serum albumin at pH 7.5 with calculated dissociation constant (K_d). d, Reaction scheme for generating molecularly homogeneous nAlb conjugates through site-selective enzymatic ligation of an amine-PEG₃-azide followed by conjugation of agonist or dye cargo through strain-promoted azide-alkyne cycloaddition (SPAAC). e, Structure of diABZI STING agonist conjugated to a DBCO-PEG₁₁ handle for ligation to azide-functionalized nanobodies via SPAAC. f,g, ESI–MS (f) and SDS–PAGE (g) showing nanobody conjugate purity and molecular weight (see Source Data for uncropped gel in ref. ). h,i, Dose–response curves in A549-Dual (n = 3) (h) and THP1-Dual type I interferon reporter cell lines (n = 3) (i) with estimated EC₅₀ values indicated in the legends; RLU, relative light unit. j, qPCR analysis of gene expression in mouse BMDMs treated in vitro with 0.25 µM of free diABZI or nAlb–diABZI conjugate (n = 3). P values determined by one-way ANOVA with Dunnett’s multiple comparison test with groups compared to PBS. Replicates are biological, and data are shown as mean ± s.e.m. Panel a created with BioRender.com.

Fig. 2. Anti-albumin nanobodies increase cargo delivery to tumour sites to promote uptake by cancer cells and tumour-associated myeloid cells.

a, Representative dose–response curves for nanobody–Cy5 conjugate surface binding and intracellular uptake at 37 °C and 4 °C measured by flow cytometry in THP-1 cells in vitro. b, MFI (Cy5) of RAW 264.7 (n = 5), EMT6 (n = 4) and BMDM (n = 3) cells treated with nAlb–Cy5 (2 µM) with (+EIPA) or without (−EIPA) the macropinocytosis inhibitor EIPA. P values determined by two-sided Student’s t-test. c,d, Representative confocal micrographs showing colocalization of Cy5 (red) with lysotracker green (green) in RAW 264.7 cells; Hoechst nuclear stain (blue) (scale bars, 100 µm) (c) with percentage colocalization determination for nAlb–Cy5 and nGFP–Cy5 in RAW 264.7 (n = 9) and EMT6 (n = 6) cells (d). P values determined by two-sided Student’s t-test. e, Pharmacokinetics of free DBCO–Cy5 dye and indicated nanobody–Cy5 conjugates injected intravenously at 2 mg kg⁻¹ in healthy female C57BL/6 mice (n = 5). Elimination phase half-life and AUC are indicated in the legend. f,g, Representative IVIS fluorescence images of excised tumours and major organs (f) and quantification of average radiant efficiencies 24 h following intravenous administration of DBCO–Cy5 (n = 5) and nAlb–Cy5 (n = 8) at 2 mg kg⁻¹ to female Balb/c mice with orthotopic EMT6 breast tumours (g). P values determined by one-way ANOVA with Dunnett’s multiple comparison test with each organ compared to tumour. h,i, Quantification of percentage of injected dose per gram of tissue (%ID per g) 24 h following intravenous administration of nAlb–Cy5 at 2 mg kg⁻¹ or PBS (vehicle) to female Balb/c mice with orthotopic EMT6 breast tumours (n = 5) (h) and female C57BL/6 mice with subcutaneous B16.F10 tumours (n = 5) (i). P values determined by two-way ANOVA with post hoc Tukey’s correction for multiple comparisons. j, Representative fluorescence microscopy images of EMT6 tumour sections stained for DAPI (blue), CD45 (green) and CD31 (red) 24 h following administration of nAlb–Cy5 (yellow) alone or in combination with nAlb–diABZI. Scale bars, 200 µm. k,l, Flow cytometric analysis of nAlb–Cy5 cellular uptake in EMT6 tumours evaluated as the percentage of indicated cell type comprising all Cy5⁺ live cells (k) or as the percentage of Cy5⁺ cells (cell type of all Cy5⁺ cells) within an indicated live cell population (l) 24 h following administration of nAlb–Cy5 alone (n = 7) or nAlb–Cy5 co-administered with nAlb–diABZI (n = 8); MFI for each cell population is shown in Supplementary Fig. 13. Inset of k: percentage of indicated cell population in the tumour as measured by flow cytometry. DC, dendritic cell; MΦ, macrophage; NK cell, natural killer cell. P values determined by two-way ANOVA with Šídák’s test for multiple comparisons. Replicates are biological, and data are shown as mean ± s.e.m.

Fig. 3. Albumin-hitchhiking STING agonist inhibits breast tumour growth by shifting the immune cell profile of the TME.

a, Schematic of EMT6 tumour inoculation, treatment schedule and study end point for gene expression and flow cytometry analysis. b,c, Tumour growth curves (b) and spider plots of individual tumour growth curves (c) for each mouse with EMT6 tumours treated with nAlb–diBZI (n = 9), diABZI (n = 8) or PBS (n = 8). P value determined by two-way ANOVA with post hoc Tukey’s correction for multiple comparisons with comparison to PBS on day 17 shown. d–j, Flow cytometric analysis of breast tumours and spleen 24 h following final dose of nAlb–diABZI or PBS (n = 6). d, t-Distributed stochastic neighbour embedding (tSNE) plots of live cells in EMT6 tumours coloured by cell population with relative expression level of Ki67, CD69 and PD-1 as indicated on heat map. e,f, Heat maps summarizing the fold change in the percentage of indicated cell population (e) and fold change in the frequency of NK cells, CD8⁺ T cells and CD4⁺ T cells expressing the indicated marker or marker combination in EMT6 breast tumours (f). g, Quantification of Ki67⁺CD69⁺ and Ki67⁺PD1⁺ CD8⁺ and CD4⁺ T cells in EMT6 tumours following treatment with nAlb–diABZI or PBS. h, Quantification of frequency of major histocompatibility complex-II (MHC-II)⁺ and PD-L1⁺ macrophages in EMT-6 tumours following treatment with nAlb–diABZI or PBS. i, Heat map summarizing fold change in the frequency of NK cells, CD8⁺ T cells and CD4⁺ T cells expressing activation markers within splenic populations. j, Quantification of Ki67⁺CD69⁺ and Ki67⁺PD1⁺ CD8⁺ and CD4⁺ T cells in spleens. P values determined by two-tailed Student’s t-test. Replicates are biological, and data are shown as mean ± s.e.m. Panel a created with BioRender.com.

Fig. 4. Design, synthesis and testing of bivalent nanobody–STING agonist conjugate for albumin hitchhiking and targeting of PD-L1.

a, Scheme for the cloning, expression and bioconjugation of small molecule cargo to generate the AP–diABZI conjugate. b,c, SDS–PAGE (b) and ESI–MS (c) confirming the purity and molecular weight of AP conjugates (see Source Data for uncropped gels in ref. ). d,e, Dose–response curves for indicated nanobody–diABZI conjugate in A549-Dual (n = 3) (d) and THP1-Dual type I interferon reporter cell lines (n = 3) (e) with estimated EC₅₀ values indicated in the legends. f, qPCR analysis of genes associated with STING activation in BMDMs in response to treatment at discrete time points with indicated agonist at 0.25 µM (n = 3). g,h, Dose–response curve for nAlb–Cy5 and AP–Cy5 conjugate intracellular uptake and surface binding at 37 °C and 4 °C as measured by flow cytometry in B16.F10 cells (n = 2 at 4 °C and n = 3 at 37 °C) (g) and EMT6 cells (n = 3) (h). i, MFI for nAlb–Cy5 and AP–Cy5 conjugate surface binding at 2 µM compared to PBS (0 µM) for EMT6 WT and EMT6 PD-L1 KO cell lines at 37 °C (n = 3). KO, knock-out; WT, wild type. j, Pharmacokinetics of indicated nanobody–Cy5 conjugate in healthy Balb/c female mice (n = 4 for nPD-L1–Cy5; n = 5 for all other groups). Elimination phase half-life and AUC are indicated in the legend. k, Representative IVIS fluorescence images of excised tumours and major organs (left) and quantification of average radiant efficiencies (right) of tumours and major organs 48 h after administration of nPD-L1–Cy5 and AP–Cy5 in mice with EMT6 breast tumours (n = 4). P values determined by repeated measures ANOVA with Dunnett’s multiple comparison test for tumour compared to indicated tissue. l, Comparison of Cy5 radiant efficiencies in tumour tissue 48 h following administration of indicated nanobody–Cy5 conjugate (n = 6 for PBS and nAlb–Cy5; n = 4 for AP–Cy5; n = 3 for nPD-L1–Cy5). P values determined by one-way ANOVA with post hoc Tukey’s correction for multiple comparisons with comparisons between all groups and PBS and between nAlb–Cy5 and AP–Cy5 as indicated. m, Representative IVIS fluorescence images of excised tumours and major organs (left) and quantification of average radiant efficiencies (right) of tumours and major organs 48 h after administration of AP–Cy5 in mice with wild-type EMT6 (WT) and PD-L1 knock-out EMT6 (PD-L1 KO) breast tumours (n = 5). P values determined by repeated measures ANOVA with Dunnett’s multiple comparison test for WT versus PD-L1 KO groups. Replicates are biological, and data are shown as mean ± s.e.m. Panel a created with BioRender.com.

Fig. 5. Systemic administration of AP–diABZI conjugates enhance antitumour immune and therapeutic responses in EMT6 breast cancer model.

a, Schematic of EMT6 tumour inoculation and treatment schedule; nanobody–diABZI conjugates and PBS (vehicle) were administered intravenously, and ICB (anti-PD-L1 IgG) was injected intraperitoneally. b–d, Tumour growth curves (b), spider plots of individual tumour growth curves (c) and Kaplan–Meier survival plots (d) for mice with EMT6 tumours treated as indicated (n = 10). CR, complete responder. P values in b determined by one-way ANOVA with Dunnett’s multiple comparison test for each group compared to PBS on day 22. In d, end-point criteria of 1,500 mm³ tumour volume with P value determined by log-rank test compared to PBS group or between nAlb–diABZI and AP–diABZI as indicated. e,f, Spider plots of individual tumour growth curves (e) and Kaplan–Meier survival curves (f) of mice challenged or re-challenged (for complete responders to the treatment regimen) with EMT6 cells (n = 10 for treatment-naive and re-challenge of mice treated with AP–diABZI; n = 9 for re-challenge of mice treated with nAlb–diABZI + ICB); end-point criteria of 1,500 mm³ tumour volume with P value determined by log-rank test compared to treatment-naive group. g, Scheme of EMT6 WT and EMT6 PD-L1 KO tumour inoculation and treatment schedule. h, Kaplan–Meier survival plots for mice with EMT6 WT (n = 13) or PD-L1 KO (n = 5) tumours treated with AP–diABZI or PBS; end-point criteria of 1,500 mm³ tumour volume with P value determined by log-rank test compared to PBS (WT) group or between WT and PD-L1 KO groups as indicated in the legend. i,j, Volcano plots representing −log₁₀(significance) and log₂(fold change) for gene expression analysis in nAlb–diABZI versus PBS (n = 4) (i) and AP–diABZI versus PBS (n = 4) (j). k–m, Heat maps of NanoString gene cluster matrices showing Z score fold changes for functional gene annotations (k), biological signatures (l) and cell types (n = 4 for PBS and AP–diABZI; n = 3 for nAlb–diABZI) (m). Replicates are biological, and data are shown as mean ± s.e.m. Panels a and g created with BioRender.com.

Fig. 6. AP–diABZI activates a tumoricidal NK and T-cell response.

Flow cytometric analysis of orthotopic EMT6 breast tumours 24 h following two intravenous doses of AP–diABZI (n = 8) or PBS (n = 7). a, tSNE plots of live cells in EMT6 tumours coloured by cell population with relative expression level of Ki67, CD69, PD-1 and PD-L1 as indicated on heat map. b, Heat map summarizing the fold change in the percentage of indicated cell populations in EMT6 tumours. c, Bar plots showing an increase in CD8⁺ cells and the ratio of CD8⁺ to CD4⁺FoxP3⁺ cells (% of CD3⁺ tumour cells). d, Quantification of Ki67⁺CD69⁺ and Ki67⁺PD1⁺ CD8⁺ T cells in EMT6 tumours. e, Spleen phenotyping heat map of frequency of NK cells, CD8⁺ T cells and CD4⁺ T cells (n = 7). In b–e, P values determined by two-tailed Student’s t-test. f, Schematic of EMT6 tumour inoculation and treatment schedule with depletion antibodies anti-Asialo GM1 (αNK) IgG, anti-CD8 IgG and anti-CD4 IgG (n = 13 for PBS and AP–diABZI and n = 7 for AP–diABZI combined with anti-Asialo GM1, anti-CD8 or anti-CD4 IgG). g,h, Tumour growth curves (g) and Kaplan–Meier survival plots (h) for mice with EMT6 tumours treated as indicated. In g, P values determined by two-way ANOVA with post hoc Tukey’s correction for multiple comparisons for all groups compared to PBS on day 22. In h, end-point criteria of 1,500 mm³ tumour volume with P values determined by log-rank test compared to PBS. Replicates are biological, and data are shown as mean ± s.e.m. Panel f created with BioRender.com.

Fig. 7. Nanobody–STING agonist conjugates stimulate antitumour immunity in B16.F10 melanoma tumour model.

a, Schematic of B16.F10 tumour inoculation and treatment schedule; nanobody–diABZI conjugates and PBS (vehicle) were administered intravenously, and ICB (anti-PD-L1 IgG) was injected intraperitoneally. b–d, Tumour growth curves (b), spider plots of individual tumour growth curves (c) and Kaplan–Meier survival plots (d) (n = 15 for PBS; n = 10 for all other groups). In b, P values determined by two-way ANOVA with post hoc Tukey’s correction for multiple comparisons for all groups compared to PBS on day 18. In d, end-point criteria of 1,500 mm³ tumour volume with P values determined by log-rank test compared to PBS control or between nAlb–diABZI and AP–diABZI as indicated. e, Schematic of B16.F10-OVA tumour inoculation, treatment schedule and study end point for flow cytometry analysis (n = 12). f, Tumour weight on day 15 for mice with B16.F10-OVA tumours treated with AP–diABZI or PBS. g, Frequency of CD4⁺ and CD8⁺ T cells in the spleen at study end point. h–k, Flow cytometric analysis of the frequency of CD69⁺ CD8⁺ and CD4⁺ T cells (h), CD44⁺CD62L⁻ effector memory T cells (i), CD44⁻CD62L⁺ naive T cells (j) and CD44⁺CD62L⁺ central memory T cells (k). l, Representative flow cytometry dot plots (left) and analysis of the frequency of SIINFEKL/H-2Kb tetramer⁺ ((PE) (MFI)) CD8⁺ T cells ((FITC) (MFI)) (right) in the spleen at study end point. m, Representative flow cytometry dot plots showing the distribution of CD8⁺ T_EM (CD44⁺CD62L⁻) and T_CM (CD44 + CD62L⁺) (CD44: (PE/Cy5) (MFI); CD62L: (BV711) (MFI)) within the OVA-specific (tetramer⁺) and non-OVA-specific (tetramer⁻) populations. P values determined by two-tailed Student’s t-test. Replicates are biological, and data are shown as mean ± s.e.m. Panels a and e created with BioRender.com.

Fig. 8. Albumin-hitchhiking STING agonists improve immunotherapy responses in a model of lung metastatic melanoma and adoptive T-cell transfer therapy.

a, Schematic of B16.F10-Luc intravenous tumour inoculation, treatment schedule and study end point for analysis of lung tumour burden; nanobody–diABZI conjugates and PBS (vehicle) were administered intravenously, and ICB (anti-PD-L1 IgG) was injected intraperitoneally (n = 15 for AP–diABZI; n = 14 for PBS; n = 12 for ICB and nAlb–diABZI + ICB; n = 11 for nAlb–diABZI). b,c, Representative images of lungs (b) and lung weights (c) of mice treated as indicated. d,e, Representative IVIS luminescence images (d) and quantification of average radiance from luciferase expressing B16.F10 cells within isolated lung tissue (e). P values determined by one-way ANOVA with post hoc Tukey’s correction for multiple comparisons for all groups versus PBS or nAlb–diABZI versus AP–diABZI as indicated. f–i, Evaluation of AP–diABZI as an adjuvant therapy for adoptive OT-I T-cell transfer therapy in a B16.F10-OVA model. f, Schematic of B16.F10-OVA tumour inoculation and of treatment schedule with OT-I T cells (0.5 million cells) on either day 9 (OT-I alone or single dose AP–diABZI pre-treatment) or day 15 (three-dose AP–diABZI pre-treatment). g–i, Tumour growth curves (g), spider plots of individual tumour growth curves (h) and Kaplan–Meier survival curves (i) (n = 15 for PBS; n = 12 for all other treatments). In g, P values determined by two-way ANOVA with post hoc Tukey’s correction for multiple comparisons for all groups compared to PBS on day 17. In i, end-point criteria of 1,500 mm³ tumour volume with P value determined by log-rank test for comparison to PBS group or for the comparisons indicated in the legend. Replicates are biological, and data are shown as mean ± s.e.m. Panels a and f created with BioRender.com.

Extended Data Fig. 1. In vitro analysis of nanobody internalization.

(a) Median fluorescence intensity (MFI) of BMDMs treated with AlexaFluor647-labeled mouse serum albumin (MSA-AF647) at 1 µM or PBS in serum-containing ( + serum) or serum -deficient (-serum) media as measured by flow cytometry (n = 2). P values determined by ANOVA with post-hoc Tukey’s correction for multiple comparisons. (b) MFI of BMDM cells treated with 2 µM MSA-AF647 in serum containing ( + serum) or serum deficient (-serum) media as measured by flow cytometry (n = 2). P values determined by ANOVA with Šídák’s multiple comparison test. MFI of (c) BMDM cells, (d) BMDC cells, and (e) EMT6 cells treated with 1 µM nAlb-Cy5 or nGFP-Cy5 at 37 °C and 4 °C as measured by flow cytometry (n = 3). P values determined by two-tailed Student’s t-test. (f) MFI of EMT6 cells treated with 2 µM nAlb-Cy5 or nGFP-Cy5 in serum containing ( + serum) or serum deficient (-serum) media as measured by flow cytometry (n = 4). P values determined by ANOVA with Šídák’s multiple comparison test. (g-h) MFI of (g) EMT6, and (h) RAW 264.7 cells treated with nAlb-Cy5 (2 µM) with (+EIPA) or without (-EIPA) the macropinocytosis inhibitor EIPA as measured by flow cytometry (n = 3). P values determined by two-tailed Student’s t-test. (i) Integrated pixel intensity of Gal9-mCherry puncta per cell for cells treated DBCO-PEG₁₁-diABZI (DBCO-diABZI), nAlb-diABZI, and AP-diABZI at 0.25 µM (n = 9 for PBS; n = 3 for all other groups). P values determined via ANOVA with Dunnett’s multiple comparisons test for all groups vs. PBS; ns: not-significant (P > 0.05). (j-k) Colocalization analysis (j) of Cy5 and LysoTracker in RAW264.7 and EMT6 cells, and (k) fluorescent micrographs of RAW 264.7 cells treated with 2 μM nAlb-Cy5 or nGFP-Cy5 for analysis of colocalization of Cy5 (red) with lysosomes (LysoTracker Green; green); nuclei are stained with Hoechst (blue) (scale bar: 100 µm). Replicates are biological, and data are shown as mean ± SEM.

Extended Data Fig. 2. Evaluation of nAlb-diABZI and AP-diABZI toxicity.

(a) Scheme for treating healthy C57BL/6 female mice with nAlb-diABZI or AP-diABZI (1.25 µg diABZI) or PBS (vehicle). (b) Body weight change of mice in response to indicated treatment (n = 3 for PBS; n = 5 for other groups). (c) Quantification of serum cytokines 4 and 24 h after the first treatment (n = 6). P values determined by one-way ANOVA with Dunnett’s multiple comparisons test for each group compared to PBS. (d) After 3 treatments, mice were euthanized and blood samples were collected to determine changes in red blood cells (RBCs), white blood cells (WBCs), neutrophils, platelets, and lymphocytes. Serum samples were also used to analyze liver and kidney function by measuring changes in alanine aminotransferase (ALT), aspartate transferase (AST), blood urea nitrogen (BUN), and creatinine (n = 4 for AP-diABZI; n = 5 for other groups). P values determined by ANOVA with Dunnett’s multiple comparison test each group compared to PBS. (e) Representative microscopy images of H&E stained tissue sections from healthy C57BL/6 mice treated as indicated (scale bar: 100 µm for heart, lung, liver, kidney, and spleen; 50 µm for pancreas and bone marrow). Replicates are biological, and data are shown as mean ± SEM. Panel a created with BioRender.com.

Extended Data Fig. 3. Flow cytometric immunophenotyping of EMT6 tumors following nAlb-diABZI treatment.

tSNE plots of live cells in EMT6 tumors after three doses of (a) PBS or (b) nAlb-diABZI, colored by cell population with relative expression levels. DC: dendritic cell; Mφ: macrophage; NK: natural killer cell; MDSC: myeloid-derived suppressor cells.

Extended Data Fig. 4. Evaluation of AP-diABZI in a spontaneous breast cancer model.

Female FVB/N-Tg (MMTV-PyVT)^634Mul mice with breast tumors were treated with AP-diABZI or PBS (vehicle) once a week for 3 weeks starting at approximately 8–10 weeks of age. (a) Growth rate of first palpable mammary tumor during treatment until the study was terminated on day 22 (n = 8 for PBS; n = 7 for AP-diABZI). At necropsy, all breast tumors were removed and weighed (b) (n = 8 for PBS; n = 7 for AP-diABZI; P value determined by two-tailed Mann-Whitney test) and histological analysis of lungs was performed to quantify lung metastasis (c) (n = 8 for PBS; n = 6 for AP-diABZI; P value determined by two-tailed Student’s t-test). Replicates are biological, and data are shown as mean ± SEM.

Extended Data Fig. 5. NanoString gene expression analysis of EMT6 tumors following treatment with nAlb-diABZI and AP-diABZI.

Annotated matrices for (a) functional gene annotations, (b) biological signatures, and (c) cell types from IO360 Pan Cancer NanoString gene expression panel comparing PBS, nAlb-diABZI, and AP-diABZI 24 h after three doses (n = 3 for nAlb-diABZI; n = 4 for all other groups). P values determined by one-way ANOVA with post-hoc Tukey’s correction for multiple comparisons with comparison to PBS indicated. Replicates are biological.

Extended Data Fig. 6. Flow cytometric analysis of EMT6 tumors following single dose of nAlb-diABZI or AP-diABZI.

EMT6 tumor bearing female Balb/c mice were treated with a single dose of nAlb-diABZI (n = 8), AP-diABZI (n = 8), or PBS (n = 7) and tumors isolated 48 h later for flow cytometric analysis. (a) tSNE plots of live cells in EMT6 tumors, colored by cell population with relative expression level of Ki67, CD69, PD-1, and PD-L1 as indicated on heat map. DC: dendritic cell; Mφ: macrophage; NK: natural killer cell; MDSC: myeloid-derived suppressor cell. (b) Analysis of the frequency of live CD45^- cells and frequency of PD-L1⁺, Ki67⁺, and PD-L1⁺Ki67⁺ expressing CD45^- cells in the tumor. (c-d) Heat maps summarizing the fold change in the percentage of (c) indicated cell population and (d) frequency of NK cells, CD8⁺ T cells, and CD4⁺ T cells expressing the indicated marker in EMT6 tumors. (b-d) P values determined by one-way ANOVA with Dunnett’s multiple comparisons test for all groups vs. PBS. (e-f) Representative flow cytometry dot plots characterizing the expression of CD69 and PD-1 on Ki67⁺CD4⁺ and Ki67⁺CD8⁺ T cells in EMT6 tumors. Replicates are biological, and data are shown as mean ± SEM.

Extended Data Fig. 7. Flow cytometric analysis of EMT6 tumors following two doses of AP-diABZI.

EMT6 tumor bearing female Balb/c mice were treated with two intravenous doses of AP-diABZI (n = 8) or PBS (n = 7) and tumors isolated 24 h later for flow cytometric analysis of (a) frequency of live and Ki67⁺ breast cancer cells (CD45^-), (b) frequency of NK cells, CD8⁺ T cells, and CD4⁺ T cells expressing the indicated markers in EMT6 tumors, and (c) frequency of NK cells, CD8⁺ T cells, and CD4⁺ T cells expressing the indicated markers in the spleen. (d) Frequency of Ki67⁺CD69⁺ and Ki67⁺PD-1⁺ CD8⁺ T cells and CD4⁺ T cells in the spleens of mice treated as indicated. P values determined by two-tailed Student’s t-test. Replicates are biological, and data are shown as mean ± SEM.

Extended Data Fig. 8. Evaluation of serum cytokines induced by nanobody-diABZI conjugates.

Serum cytokine concentration in B16.F10 tumor bearing C57BL/6 female mice 4 h after the first treatment represented as (a) heat maps and (b) bar plots (n = 10). ICB: Anti-PD-L1 IgG. P values determined by one-way ANOVA with (a) Dunnett’s multiple comparison test for all groups vs. PBS and (b) post-hoc Tukey’s correction for multiple comparisons. Replicates are biological, and data are shown as mean ± SEM.

Extended Data Fig. 9. Effect of NK and T cell depletion on AP-diABZI efficacy in B16.F10 model.

(a) Schematic of B16.F10 tumor inoculation and treatment schedule with depletion antibodies (n = 9 for PBS, anti-CD4, and AP-diABZI; n = 8 for all other groups). Anti-NK1.1 IgG, anti-CD8 IgG, and anti-CD4 IgG were injected I.P. at 200 µg and AP-diABZI was injected I.V. at 1.25 µg of diABZI per injection. (b) Tumor growth curves and (c) Kaplan-Meier survival plots for mice with B16.F10 tumors treated as indicated. (b) P values determined by two-way ANOVA with post-hoc Tukey’s correction for multiple comparisons; comparisons to PBS on day 19 are shown. (c) Endpoint criteria of 1500 mm³ tumor volume with P value determined by log-rank test for comparison to PBS group and for the comparisons indicated. Replicates are biological, and data are shown as mean ± SEM. Panel a created with BioRender.com.

Extended Data Fig. 10. Characterization of T cell memory response to AP-diABZI in B16.F10-OVA model.

Quantification of flow cytometric analysis presented in Fig. 7m showing the distribution of memory of SIINFEKL-specific (SIINFEKL/H-2kB tetramer⁺) or non-SIINFEKL-specific (tetramer^-) CD8⁺ T cells (n = 12). Represented bar plots include effector memory T_EM (CD44⁺CD62L^-), activated CD69⁺ effector memory T_EM (CD44⁺CD69⁺CD62L^-), naïve (CD44^-CD62L⁺), and central memory T_CM (CD44 + CD62L⁺). P values determined by by two-tailed Student’s t-test. Replicates are biological, and data are shown as mean ± SEM.