Dear Editor,

As a successful drug for inflammatory diseases, the application of TNF-α inhibitor on cancer therapy is limited by repeated administration and off-target effects.1 A body of evidence indicated that the anti-tumor efficacy of TNF-α inhibitor is unsatisfactory, though repeated administration was used to improve its efficacy in tumor-treating fields, it will also lead to severe side effects and high cost.2,3,4 Hence, an efficient and highly targeted TNF-α antibody delivery system is worth developing.

The genetically modified strain of attenuated Salmonella typhimurium VNP20009 (VNP) not only has super tumor-targeting capacity and genetic stability in vivo, but also has thousands of times higher enrichment in tumors than that of liver and spleen.5 Thus, in this work, we built a novel VNP delivery system expressing anti-TNF-α nanobody (VNPαTNF-α) (Fig. 1a and Supplementary Fig. S1a–g), which could significantly improve the delivery efficiency by continuous release of the nanobody under a hypoxic tumor environment (Fig. 1b and Supplementary Fig. S2a–d). Moreover, another strain of VNPαTNF-α/mCherry was constructed with TNF-α nb fused to mCherry for the visualization of expressed TNF-α nb (Supplementary Fig. S1h–j). The TNF-α nb secreted by VNP had a similar particle size (75.27 ± 14.08 nm) and affinity compared with pure nanobody synthesized in our previous work6 (Supplementary Fig. S2e, f). VNPαTNF-α induced about 40% apoptosis of B16F10 which was similar to that of VNP, while pure TNF-α nb couldn’t induce cell apoptosis, the results confirmed the antitumor activity of VNP and VNPαTNF-α in vitro (Supplementary Fig. S3b, c). In addition, VNPαTNF-α stimulated dendritic cells (DCs) activation and cytotoxic CD8+ T cell production in vitro (Fig. 1c, d). VNPαTNF-α stimulated CD8+ T cell production immediately by activating macrophage antigen presentation (Fig. 1e, Supplementary Fig. S3d). To evaluate the neutralization ability of VNPαTNF-α, the supernatants of M1-type RAW264.7 were collected to measure the level of TNF-α. The result indicated that VNPαTNF-α treatment significantly neutralized secreted TNF-α (sTNF-α), thus, decreasing the level of sTNF-α (Supplementary Fig. S3e).

Fig. 1
figure 1

a The TNF-α nb expression inside bacteria or secreted into LB was detected by western blot (n = 3). b Expression of TNF-α nb-mCherry under normoxic and hypoxic (1% O2) conditions (n = 3). c Ratio of splenic DCs after incubated with different groups for 2 h or 6 h (n = 3). d Proportion of granzyme B+CD8+ T cells after incubated with different groups for 2 h or 6 h (n = 3). e Counted CD8+ T cell ratio in splenic lymphocytes after RAW264.7 cells incubation (n = 3). f The organ burden of tumor-bearing mice 5 days after injection (n = 3). g MFI analysis of representative fluorescent image of colonized VNPαTNF-α/mCherry (Red; 1 × 108 CFU/mouse) and TNF-α nb-mCherry (Red; 150 µg/kg) in tumors (n = 3). h Tumor growth curves of B16F10 administered with PBS, VNP, or VNPαTNF-α (n = 6). i The calculated tumor doubling time of different groups (n = 6). j Overall survival of the tumor-bearing mice (n = 6). k, l H&E staining of tumors after VNPαTNF-α treatments. Scale bar, 200 µm. N: necrotic region, nN: nonnecrotic region. m The distribution of tumor-infiltrating immune cells after VNPαTNF-α administration. 1000 μm scale bars shown. (Neutrophil: Ly6g+; Macrophage: F4/80+; CD8+; CD4+; active T cells: CD69+; VNPαTNF-α: FITC+) n Quantification of DC cells in tumor and other related immune organs by flow cytometry (peripheral blood, spleen, and TdLNs) of tumor-bearing mice (n = 4). o Ratio of CD86+ in DCs (n = 4). p Expression of CD80, CD86, and PDL1 in DC2.4 cells after VNPαTNF-α treatment (n = 3). q Ratio of CD8+ and CD4+ T cells in the lower chambers (n = 3). r, s Apoptosis ratio of B16F10-OVA cells after VNPαTNF-α-stimulated splenocyte treatments (n = 3). t, u Collected unstimulated splenocytes (naïve T) and splenocytes activated by cytokine (mature T) and identified splenic T cells as naïve T cells, cytotoxic T cells, or dysfunctional T cells by RT-PCR. The bar is the relative expression level (n = 3). v Quantification of granzyme B+ CD8+ T cells in blood, spleen, TdLNs and tumor (n = 4). w Quantification of Ki67+granzyme B+ CD8+ T cells in blood, spleen, TdLNs and tumor (n = 4). x Flow cytometry assessment of CD8 Treg cells by CD25 and CD122 markers (n = 5). y Schematic diagram of therapeutic mechanisms of B16F10-bearing mice by VNPαTNF-α. Illustrations was made with BioRender. Data are shown as the mean ± SD. ****p < 0.0001, ***p < 0.001, **p < 0.01, *p < 0.05, ns no significance

Full size image

To evaluate the tumor targeting ability of VNPαTNF-α, the organ burdens of bacteria were measured (Supplementary Fig. S4a). It was indicated that both VNP and VNPαTNF-α showed tumor targeting ability as expected, which were hundreds to thousands of times higher than other tissue (Fig. 1f, Supplementary Fig. S4c). To further study the tumor residence time of TNF-α nb in tumor, we injected TNF-α nb-mCherry (150 µg/kg) and VNPαTNF-α/mCherry (1 × 108 CFU, the amount of secreted TNF-α nb-mCherry was equivalent to that of TNF-α nb-mCherry group) according to our data. At first, the MFI of TNF-α nb-mCherry in tumor tissue was 1.3 times higher than that of VNPαTNF-α/mCherry within 4 hours. After 12 hours, pure TNF-α nb was depleted slowly, while TNF-α nb-mCherry of VNPαTNF-α/mCherry increased to 2.3 times higher than pure TNF-α nb as VNP proliferated continuously (Fig. 1g, Supplementary Fig. S4b).

Next, the antitumor effect of VNPαTNF-α in vivo was evaluated (Supplementary Fig. S5a). The tumor growth curve indicated that VNPαTNF-α effectively inhibited melanoma progression (Fig. 1h). In addition, delayed tumor doubling time (TDT) was 2.38 days in the VNP group and 3.35 days in the VNPαTNF-α group, and the TDT ratio of VNPαTNF-α to PBS or VNP reached 1.67 times or 1.4 times respectively (Fig. 1i). VNPαTNF-α also prolonged tumor-bearing mouse survival significantly than that of VNP (Fig. 1j). These results suggested that VNPαTNF-α had an excellent therapeutic effect. Moreover, the H&E analysis of tumor section showed that more than 75% of the tumor was necrotic after VNPαTNF-α treatment (Fig. 1k, l). Next, we preliminary explored the therapeutic mechanism of VNPαTNF-α. Firstly, it is indicated that the level of transmembrane TNF-α (tmTNF-α) in VNPαTNF-α was reduced, which is lower than the baseline of the PBS group (Supplementary Fig. S6a). Since it is reported that low-dose TNF-α induces angiogenesis while high-dose TNF-α lead to thrombosis within tumor vasculature,7 we then assessed the distribution and gene expression of tumor vessel by CD31 and vascular endothelial growth factor (VEGF) staining. It is shown in Supplementary Fig. S6b, c that VEGF and CD31 was downregulated in the VNPαTNF-α group, suggesting that VNPαTNF-α inhibited tumor progression by reducing the density of tumor vessels. Therefore, VNPαTNF-α induced more cell apoptosis in the tumor tissue (Supplementary Fig. S6d).

To further elucidate the therapeutic mechanisms, the distribution of tumor-infiltrating immune cells was detected. The proportion of CD8+ T cells and CD69+ cells were significantly increased, approximately 11 and 7%, while the ratio of CD4+ T cells were reduced in the VNPαTNF-α group (Fig. 1m). In addition, the proportions of neutrophils and macrophages were significantly increased both in the VNP and VNPαTNF-α group (Fig. 1m, Supplementary Fig. S7a). Next, the proportion and state of DCs in vivo were investigated. It is shown that the ratio of DCs and activated DCs (CD86+DCs) were significantly increased in immune organs (Fig. 1n, o). The results were consistent with in vitro experiment, where VNPαTNF-α induced the upregulation expression of CD86, CD80, and PDL1 of DC2.4 cells in vitro (Fig. 1p). As for in vivo experiment, the elevated level of CD86 and CD80, and the decrease level of PD1 and PDL1 in the tumor mixed pool were observed (Supplementary Fig. S5b). In addition, CD11b+ in DCs was upregulated 1.6 times higher than that of VNP in tumor, which means DCs were activated by VNPαTNF-α8 (Supplementary Fig. S5c). Together, these results indicated that VNPαTNF-α stimulated transformation form “cold” tumor with immune suppression to “hot” tumor with anti-tumor immune activation.

We further investigated whether VNPαTNF-α could stimulate CD8+ T cell activation. Therefore, we firstly stimulated splenic T cells in vitro and cocultured them with B16F10-OVA cells as illustrated in Supplementary Fig. S5d, the cells in lower chambers were collected and tumor cell-recruited CD8+ T cells and B16F10-OVA cells apoptosis were analyzed. The results revealed that the highest proportion of CD45+cells in the lower chamber was T cells, which was approximately 90%, and the proportion of CD8+ T cells increased from 32 to 40%, in contrast, CD4+ T cells decreased after VNPαTNF-α incubation (Fig. 1q), which indicated that VNPαTNF-α stimulated CD8+ T cell chemotaxis and activation. As a result, significant tumor apoptosis was induced from 15 to 25% by VNPαTNF-α-stimulated splenic T cells (Fig. 1r, s). Further detection of relative expression of markers by RT-PCR showed that VNPαTNF-α induced CD8+ T cell polarization into cytotoxic T cells, according to the upregulated TNF-α, IFN-γ, IL-2, perforin, and granzyme B as well as downregulated markers of exhausted cells, such as PD1 and TIM3. More importantly, in vivo experiments showed that splenic CD8+ T cells matured after stimulation because the markers of naïve T cells were downregulated (CCR7 and TCF7) (Fig. 1t, u). In addition, the percentage of granzyme-B+ CD8+ T cells of VNPαTNF-α group was increased in immune organs, particularly in tumor, which was four times higher than the control group (Fig. 1v). And VNPαTNF-α stimulated more Ki67+ cytotoxic CD8+ T cell, which was five times higher than the control group (Fig. 1w). These results indicated that VNPαTNF-α mobilized the systemic immune response. Furthermore, it is noteworthy that VNPαTNF-α reduced CD8+ T cell death approximately two-fold (Supplementary Fig. S5f). Notably, the same results were previously reported that anti-TNF-α inhibitor lessened CD8+ T cell death in vivo.9 Meanwhile, CD8 Tregs were reduced in the tumor draining lymph nodes (TdLNs) and tumor after VNPαTNF-α administration (Fig. 1x, Supplementary Fig. S5g, i). As expected, the percentage of Annexin V-positive CD8 Tregs was increased approximately twice after VNPαTNF-α administration (Supplementary Fig. S5h), which means that the tumor immunosuppression was alleviated.

Based on our strategy, TNF-α nb could be delivered into tumor tissue by VNP safely and efficiently, and this system exerted robust antitumor effects with controllable TNF-α nb secretion in melanoma with only one dosage, which could also avoid the side effects and high costs of TNF-α inhibitors. Moreover, VNPαTNF-α promoted antitumor immune responses in a melanoma tumor microenvironment by mobilizing tumor immune response as follows, (1) VNPαTNF-α reduced tumor angiogenesis. (2) VNPαTNF-α stimulated DCs maturation manifesting as elevated CD86. DCs activated CD8+ T cells by antigen-presentation and induced CD8+ T cells to upregulate Granzyme-B and Ki67, stimulated cytotoxic CD8+ T cell induced tumor apoptosis. (3) CD8 Treg reduced after administration of VNPαTNF-α. (4) VNPαTNF-α directly induced tumor apoptosis in vivo and in vitro (Fig. 1y). In addition, VNPαTNF-α causes acceptable splenomegaly but has better biosafety than VNP (Supplementary Fig. S8a–f).