Anti-mesothelin immunotoxin induces mesothelioma eradication, anti-tumor immunity, and the development of tertiary lymphoid structures

Significance

Malignant mesothelioma has a poor prognosis and new therapies are needed. Mesothelioma expresses mesothelin (MSLN), which is an attractive target for antibody-based therapies. LMB-100 is a recombinant immunotoxin composed of an MSLN-targeted Fab linked to a toxin. Using an orthotopic mesothelioma model, LMB-100 alone induces complete tumor remissions and tumor immunity accompanied by the development of tertiary lymphoid structures (TLS) consisting of clusters of immune cells with a distinct core of B cells. Depletion of B cells as well as CD8 and CD4 T cells blocks the anti-tumor activity of LMB-100. Strategies to induce TLS formation could be useful in treating mesothelioma.

Abstract

LMB-100 is a recombinant immunotoxin composed of a Fab linked to a toxin. It kills cells expressing human mesothelin (hMSLN), which is highly expressed on the surface of mesothelioma and many other cancer cells. Clinically, we observed some patients had delayed responses to an anti-hMSLN immunotoxin treatment, suggesting the induction of anti-tumor immunity. We aimed to develop a mouse model to investigate whether immunotoxin alone can induce anti-tumor immunity and to study the mechanism of this immunity. An immunocompetent transgenic mouse was used to grow mouse mesothelioma AB1 cells expressing hMSLN in the peritoneal cavity. Mice were treated with LMB-100, and mice with complete responses (CRs) were rechallenged with tumor cells to determine whether anti-tumor immunity developed. Changes in gene expression profiles were evaluated by Nanostring, and changes in cytokines and chemokines were checked by protein arrays. The distribution of various immune cells was assessed by immunohistochemistry. Our results show that the mice with tumor reached CRs and developed anti-tumor immunity after LMB-100 treatment alone. The primary response requires CD8⁺ T cells, CD4⁺ T cells, and B cells. Transcriptional profiling shows that LMB-100 treatment reshapes the tumor immune microenvironment by upregulating chemotaxis signals. LMB-100 treatment upregulates genes associated with tertiary lymphoid structures (TLS) development and induces TLS formation in tumors. In sum, immunotoxin-mediated cell death induces anti-tumor immunity and the development of TLS, which provides insights into how immunotoxins cause tumor regressions.

Malignant mesothelioma is a rare cancer with a poor prognosis. Treatment options include chemotherapy with pemetrexed plus cisplatin and the recommended frontline immunotherapy with nivolumab plus ipilimumab (1, 2). Mesothelin (MSLN) is a membrane glycoprotein highly expressed in a variety of human cancers, including mesothelioma, non-small cell lung cancer, pancreatic cancer, and ovarian cancer (3, 4), and also expressed on a few non-essential normal tissues (5). This expression pattern makes MSLN an attractive target for antibody-based immunotherapies (5–7).

Immunotoxins are engineered antibody-toxin chimeric molecules that are being developed for cancer therapy. Moxetumomab pasudotox, a Pseudomonas exotoxin A (PE)-based immunotoxin targeting CD22, demonstrated a high rate of durable complete responses (CRs) for the treatment of hairy cell leukemia and was approved by the US Food and Drug Administration for hairy cell leukemia treatment in 2019 (8). SS1P (SS1-dsFv-PE38) is a MSLN-targeted immunotoxin and consists of the disulfide-stabilized heavy chain Fv (VH) and light chain Fv (VL) of the antibody fragment SS1 as well as a 38 kDa truncated form of PE (PE38) (9). As SS1P has high immunogenicity which limits its clinical application, a second-generation recombinant immunotoxin with lower immunogenicity, LMB-100, was engineered. LMB-100 consists of a humanized Fab that targets human mesothelin (hMSLN) and a B-cell epitope silenced, 24 kD fragment of PE (PE24) (10). PE-based immunotoxins have a unique mechanism of cell killing. The toxin catalyzes the irreversible ADP-ribosylation of the diphthamide residue of EF2 and arrests protein synthesis. This mechanism of cell killing differs from that of currently approved chemotherapies for solid tumors (11).

Though it is generally believed that the anti-tumor effects of immunotoxins are attributed to their direct tumor cell killing through arresting protein synthesis, there is growing evidence that immunotoxins may also induce anti-tumor immune responses. For example, delayed CR was observed in some patients treated with Moxetumomab pasudotox 6 mo after the end of treatment, suggesting that the immunotoxin does more than direct tumor cells killing (8, 12). A delayed response was also observed in some SS1P-treated mesothelioma patients (13). Additionally, a recent study reveals that treatment with LMB-100 results in systemic inflammatory response in some patients with mesothelioma (14). Preclinical animal studies in a 66C14-M murine breast cancer model (15) and AE17-M murine mesothelioma model (16), showed that mice achieved CRs and developed anti-tumor immunity after combination treatment of anti-MSLN immunotoxin injected into the tumor and anti-CTLA-4 given systemically. However, the subcutaneous tumor model in these studies does not resemble the microenvironment of human mesothelioma. To study the immune response induced by immunotoxin, a clinical-relevant animal model is needed.

In the present study, we established an orthotopic mouse mesothelioma model and use it to study how immunotoxin-mediated cell death produces immune-related signals and reshapes the tumor immune microenvironment. We implanted mouse mesothelioma cells expressing hMSLN (AB1-L9) into the peritoneal cavity of Balb/c mice that are tolerant to hMSLN (TPO-MSLN mice) and treated the mice with LMB-100 given intraperitoneally. We showed here that immunotoxin-mediated cell death reshapes the tumor immune microenvironment, induces tertiary lymphoid structures (TLS) development, and contributes to the establishment of anti-tumor immunity.

Results

LMB-100 Completely Eradicates Orthotopic Mesothelioma and Produces an Anti-Tumor Immunity.

To determine whether immunotoxins alone can induce anti-tumor immunity and to study the mechanism of this immunity, we generated an orthotopic mouse model in which the mouse mesothelioma cell line AB1 expressing hMSLN (AB1-L9) grows in the peritoneal cavity of mice. To ensure the tumor cells expressing hMSLN are not rejected, the mice, termed TPO-MSLN mice, express a human transgene (MSLN) in their thyroid gland (17). These mice have a normal immune system, do not recognize hMSLN as foreign, and can safely be used to study anti-tumor immune responses to LMB-100.

The data in Fig. 1A show that AB1-L9 cells are killed by LMB-100 with an IC₅₀ of 19 ng/ml. AB1 cells not expressing hMSLN are resistant to LMB-100 with high viability up to 1,000 ng/ml. AB1-L9 cells are not killed by LMB-255, an inactive anti-hMSLN immunotoxin that has a deletion of the furin cleavage site and a mutation at residue 553 of the toxin. AB1-L9 cells are also not killed by LMB-34, an immunotoxin containing the same toxin moiety as LMB-100, but target human B-cell maturation antigen (hBCMA), which is not expressed in the tumors. These data indicate a specific killing effect of LMB-100.

Fig. 1.

Test the therapeutic effect of LMB-100 on an orthotopic mesothelioma TPO-MSLN mouse model. A. Cytotoxicity of LMB-100, non-targeting immunotoxin LMB-34 (BM24-Fab-LO10R-456A) and inactive immunotoxin LMB-255 (LMB-100-del fur-del553) on hMSLN expressing AB1-L9 cells, and cytotoxicity of LMB-100 on hMSLN non-expressing AB1 cells. B. Representative immunohistochemistry (IHC) image of tumor at day 5, stained by hMSLN. (Scale bar: 4 mm.) C. Percentage of hMSLN⁺ neoplasm cells among all cells in each tumor section at day 5 (n = 3). Mean ± SD. D–F. TPO-MSLN Tg mice were inoculated with 5 * 10⁶ AB1-L9 cells in abdominal cavity at day 1. Mice were treated with either LMB-100 (i.p. 3 mg/kg) or PBS (i.p. 200 μl) at day 5. Mice were euthanized and opened at day 5 (before the treatment), 6 and 7. Mouse plasma was collected immediately before execution. D. Weights of tumor in mice. Mean ± SD. ***P < 0.005, ****P < 0.001 compared with LMB-100 group. E. Plasma hMSLN concentration in mice treated by either LMB-100 (i.p. 3 mg/kg, day 5) or PBS (i.p. 200 μl, day 5). Mean ± SD. ****P < 0.001 compared with LMB-100 group. F. Tumor weight at days 5, 6, and 7 in mice treated by PBS (i.p. 200 μl, day 5) correlated with plasma hMSLN.

After intraperitoneal injection, AB1-L9 cells are found growing attached to the gastro-splenic ligament (GS-ligament, SI Appendix, Fig. S1A) and on the surface of many other organs (liver, spleen, stomach, diaphragm, and mesentery), as would be expected for peritoneal mesothelioma. Autopsy of mice on day 17 indicates tumors can also spread into the thoracic cavity. The earliest observable and the largest tumors are found growing on the GS-ligament. Fig. 1B shows a representative IHC image of an AB1-L9 tumor stained for MSLN that is growing on the GS-ligament 5 d after implantation. On day 5, 90% of the cells on the GS-ligament are hMSLN positive (Fig. 1 B and C). These tumors are average 48 ± 17 mg in weight (Fig. 1D).

To assess the anti-tumor activity of LMB-100, mice were treated with one dose of LMB-100 (3 mg/kg, i.p., on day 5) or with PBS, and tumor size and blood MSLN levels which reflect tumor burden was measured (Fig. 1 D and E). The tumors in the PBS-treated mice grew rapidly and their weight significantly correlated with MSLN levels in the blood (Fig. 1F). In contrast, the tumors in the LMB-100-treated mice were significantly reduced in size (Fig. 1D, P_D7 < 0.01). The size decrease was accompanied by a fall in MSLN levels (Fig. 1E, P_D7 < 0.0001). The correlation between tumor size and MSLN blood levels (Fig. 1F and SI Appendix, Fig. S1B) allowed us to calculate the size of the tumors growing in the peritoneal cavity without sacrificing the mice.

To determine an effective dose strategy, we tried the various number and scheduling of doses (Table 1). The effect of giving 1 dose of LMB-100 on day 5 or 2 doses on days 5 and 9 is shown in Fig. 2 A and B. With one dose there were 9/19 CRs (47%) and with 2 doses there were 11/22 CRs (50%); the additional dose did not significantly improve mice’s survival (Fig. 2B and Table 1). Then, we explored the effect of delaying dosing (Fig. 2 C and D and Table 1). Survival was significantly decreased if the two-dose treatment was given on day 7 and day 11 (3/11 CRs; 27%) or on day 9 and day 13 (1/11 CRs; 9%). In addition, treatment efficacy was diminished if mice only received a single dose of LMB-100 on day 7 or day 9 (SI Appendix, Fig. S2 A and B). The present result indicates the importance of early treatment.

Table 1.

Summary of mice that reach CR and tumor rejection in each experiment

Treatment	Total mice	Mice with CR	Mice with anti-tumor immunity
LMB-100 (D5)	19	9	9#
LMB-100 (D5 + D9)	42	25	17# + 5*
LMB-100 (D5 + D7 + D9)	10	4	3*
LMB-100 (D7 + D11)	11	3	3#
LMB-100 (D9 + D13)	11	1	1#
PBS (D5, D9)	23	0	0#

TPO-MSLN Tg mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶, day 1) and were treated with either LMB-100 (i.p., 2.5–3 mg/kg) or PBS (i.p.) at the date shown in the table. The number of mice in each group was pooled from all the experiments in Fig. 2. Mice with completely regressed abdominal tumor at experimental endpoint were considered as CR mice. All survived mice were re-challenged by either 2 * 10⁶ AB1 (*) or AB1-L9 cells (#) subcutaneously after 1.5 mo of initial cell inoculation. Mice that rejected the re-challenged tumor cells are mice with anti-tumor immunity.

Fig. 2.

hMSLN-targeted LMB-100 eradicated orthotopic mesothelioma and induced anti-tumor immunity. A. Scheme of using one or two doses of LMB-100 to treat AB1-L9 tumor. Mice bearing AB1-L9 tumors were treated with either one (i.p., 3 mg/kg, day 5, n = 19) or two doses of LMB-100 (i.p., 3 mg/kg, day 5 and 9, n = 22). Control mice were treated with PBS (i.p., day 5 and 9, n = 10). B. Survival of mice in each group before the re-challenge assay. Data were pooled from two experiments. ****P < 0.001 compared with LMB-100 two doses group. C. Scheme of delayed LMB-100 treatment start at day 5, day 7, or day 9. Mice bearing AB1-L9 tumors were treated with LMB-100 (i.p., 3 mg/kg) at either day 5 + 9 (n = 13), day 7 + 11 (n = 11) or day 9 + 13 (n = 11). Control mice were treated with PBS (i.p., day 5 and 9, n = 13). D. Survival of mice in each group before the re-challenge assay. *P < 0.05, ***P < 0.005, ****P < 0.001 compared with LMB-100 D5 + D9 group. E. Scheme of using LMB-100 or inactive LMB-255 to treat AB1-L9 tumor. Mice bearing AB1-L9 tumors were treated with LMB-100 (i.p., 3 mg/kg, days 5, 9 and 13, n = 10), LMB-255 (i.p., 3mg/kg, days 5, 9 and 13, n = 10) or PBS (i.p., days 5, 9 and 13, n = 5). F. Survival of mice in each group before the re-challenge assay. ***P < 0.005, ****P < 0.001 compared with LMB-100 group. G. Scheme of using LMB-100 or non-targeting LMB-34 to treat AB1-L9 tumor. Mice bearing AB1-L9 tumors were treated with LMB-100 (i.p., 3 mg/kg at day 5, 2.5 mg/kg at day 9, n = 7), LMB-34 (i.p., 3 mg/kg at day 5, 2.5 mg/kg at day 9, n = 9) or PBS (i.p., day 5 and 9, n = 5). H. Survival of mice in each group before the re-challenge assay. ***P < 0.005 compared with LMB-100 group. I and J. LMB-100 induced CR mice were re-challenged with either AB1-L9 cells (I, s.c., 2 * 10⁶ cells) or hMSLN non-expressing AB1 cells (J, s.c., 2 * 10⁶ cells) at least 1.5 mo after the initial tumor inoculation. Control mice were TPO-MSLN mice injected with either AB1-L9 cells (s.c., 2 * 10⁶ cells) or AB1 cells (s.c., 2 * 10⁶ cells). X-axis shows days after the re-challenge assay.

To test whether tumor-specific cell targeting and killing were required for generating CRs, the anti-tumor activity of non-active immunotoxin (LMB-255) and non-targeting immunotoxin (LMB-34) were evaluated. As shown in Fig. 2 E–H, neither LMB-255 (Fig. 2 E and F) nor LMB-34 (Fig. 2 G and H) produce any CRs, and all mice had tumors in their abdominal cavity at experimental endpoints. This result established the role of cell targeting and killing in the anti-tumor response generated from LMB-100.

To determine whether a long-term anti-tumor immunity had developed in the LMB-100-treated mice that had achieved CRs, the surviving mice were re-challenged by AB1-L9 cells given subcutaneously about 1.5 mo after the initial tumor inoculation. The re-challenged AB1-L9 cells failed to form tumors in 9/9 mice with one-dose treatment and 17/19 mice in the group that received two doses of LMB-100 indicating anti-tumor immunity was induced (Fig. 2I).

To determine whether the anti-tumor immunity was directed at MSLN or other tumor antigens in the mouse tumor, we also challenged 10 mice with AB1 cells not expressing MSLN. Using the re-challenge assay, 8/10 CR mice rejected the hMSLN-negative AB1 cells (Fig. 2J), indicating the anti-tumor response was directed at other antigens in the mouse tumor.

These results indicate both tumor cell targeting and killing are required for LMB-100-induced tumor CRs. Importantly, the hMSLN-specific killing by LMB-100 induced an anti-tumor immunity that is capable of rejecting tumor cells without hMSLN expression, suggesting the treated cells provide other antigens, beyond hMSLN, to induce systemic anti-tumor immunity.

LMB-100 Reprograms the Tumor Microenvironment and Induces Systemic Anti-Tumor Immunity.

To investigate whether LMB-100 induced changes in immune-related transcripts in the tumor, we employed a Nanostring nCounter platform (nCounter® PanCancer Immune Profiling Panel). We collected tumor samples 12, 24, and 48 h post-LMB-100 treatment. As shown in Fig. 3A, we found that changes in transcripts could be detected as early as 12 h. Pathway scores related to cell cycle, pathogen responses, and senescence were increased at 12 h, peaked at 24 h, and decreased at 48 h. Later at 24 h, multiple immune-related pathways were upregulated, which include pathways associated with chemotactic-related activities (e.g., cytokine/chemokines and receptors), antigen presentation-related activities (e.g., macrophage/dendritic cell functions), as well as tumor-infiltrated lymphocytes (TILs)-based cell activities (e.g., T cell and B cell functions). The result shows that LMB-100 treatment has a profound effect on immune-related pathways, and these pathways were activated in a time-dependent manner.

Fig. 3.

LMB-100 reshape tumor immune transcriptional microenvironment A–E. TPO-MSLN mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶, day 1) and were treated with either -LMB-100 (i.p., 3 mg/kg, day 5), LMB-255 (i.p., 3 mg/kg, day 5) or PBS (i.p., 200 μl, day 5). Tumors were collected 12 h, 16 h, 24 h or 48 h after the treatment. Immune-related genes were analyzed by Nanostring using a PanCancer Immune Profiling Panel. Signature sets of gene expression associated with different pathways were scored with NanoString Advanced Analysis modules. A. Z score heatmap of pathway score for each tumor treated with PBS for 12 h (n = 4), LMB-100 for 12 h (n = 4), LMB-100 for 24 h (n = 4) and LMB-100 for 48 h (n = 4). B. Z score heatmap of pathway score for each tumor treated with PBS (n = 4), LMB-100 (n = 4) or LMB-255 (n = 4) for 16 h. C. Volcano plot showing differentially expressed mRNA at 16 h in tumors treated with LMB-100 (n = 6) comparing tumors treated with LMB-255 (n = 6). The corresponding types of immune responses of the genes that were significantly increased over fivefold were highlighted with different colors: Cytokine & chemokine (Red); inflammation (Green); pathogen response (Blue), apoptosis (Purple), and T cell function (Brown). D and E. Volcano plot showing differentially expressed mRNA at 16 h in either mice tumors or in AB1-L9 cells treated with LMB-100 in vitro comparing tumors treated with LMB-255. The AB1-L9 cells were treated with either LMB-100 or LMB-255 for a same period as in vivo experiment (16 h) as described in Methods. Genes that were significantly increased over twofold in LMB-100 vs. LMB-255-treated tumor in mice were marked in both plots. Red dots represent genes upregulated both in vivo and in vitro; blue dots represent genes upregulated specifically in vivo. F and G. The colony-forming capability of AB1-L9 cells (5 * 10⁶ cells) pre-treated with LMB-100. F. Representative images of cell colonies at day 15. Image for the control showed 5 * 10⁴ cells cultured for 7 d. G. Number of colonies in each dish counted at day 15. Mean ± SD. *P < 0.05 compared with 48 h. H–J. AB1-L9 cells (5 * 10⁶ cells) were pre-treated with LMB-100 (60 μg/2 mL for 30 min, then 60 μg/8 mL for either 6 h or 48 h) and were injected (i.p.) into the TPO-MSLN mice. At day 33, all survived mice were re-challenged by 2 * 10⁶ AB1-L9 cells (s.c.). H. Scheme of vaccination assay. I. Size of the re-challenged subcutaneous tumors (not the initial tumor) in each mouse. J. Endpoint weight of initial tumor in each mouse. Endpoint tumor weight = 0 means tumor in abdominal cavity has completely regressed. Mean ± SD. *P < 0.05 compared with Con.

Because the toxin domain of LMB-100 is immunogenic in mice (18), we examined whether the upregulated immune responses in the tumor were caused by an anti-drug immune reaction rather than associated with the killing of tumor cells. To do this, we treated mice with LMB-100 to kill cells or with an inactive immunotoxin, LMB-255 that, except for a few inactivating mutations, is identical to LMB-100 and should be equally immunogenic. Changes in immune-related pathway scores are shown in Fig. 3B; we found that AB1-L9 tumors treated with LMB-255 had very few changes relative to tumors treated with LMB-100. This data clearly shows that the immune responses in the tumor microenvironment require immunotoxin-mediated tumor cell killing and are not an immune reaction to a foreign protein.

To identify the most upregulated gene transcripts induced by LMB-100-mediated tumor cell killing, the relative change of genes influenced by LMB-100 vs. LMB-255 and the P-value are shown in the volcano plot in Fig. 3C. There are 27 genes that are increased over fivefold and are statistically meaningful. They correspond to five types of immune responses: cytokine and chemokine (Ccl20, Cxcl2, Ifnb1, Cxcl1, Ifna4, Csf2, Cxcl3, Il6, Cxcl5, and Lif); inflammation (Ccl20, Cxcl22, Cxcl1, Cxcl3, Il6, Tnfaip3, Ptgs2, Cxcl5, Fos and Creb5); pathogen response (Ifnb1, Il6, Ptgs2, Nfkbia, Jun, and Fos); apoptosis (Tnfaip3, Ptgs2, Txnip, Muc1, and Jun); and T cell function (Csf2, Egr1, Gata3, Icam1, Ifnb1, Il6, and Relb).

To determine whether non-malignant cells in the tumor microenvironment are responsible for the increased expression of these genes, we treated tumor cells in culture and compared their transcriptional profiles change with tumors treated in mice. We first determined the genes which were significantly changed in mice after treating with LMB-100 by using LMB-255 as the control. As shown in Fig. 3D, among the 770 tested genes, 252 genes were significantly increased (P < 0.05), and 98 were increased more than twofold (Table 2 and SI Appendix, Table S1). Next, we compared our tumor data with changes that occurred when AB1-L9 cells were treated in culture in the absence of stroma. As shown in Fig. 3 D and E, among the 98 significantly increased genes in mice tumors, 40 genes were only increased in tumor tissue (Fig. 3 D and E, blue dots), indicating non-malignant cells in tumor tissue contribute to the change, directly or indirectly. We further analyzed the top 10 significantly upregulated genes in tumor tissue (Table 2) and found 7 genes (Ccl20, Cxcl2, Ifnb1, Ifna4, Csf2, Cxcl3, and Il6) encoded proteins associated with chemokine and cytokine activities and were only highly expressed in tumor tissue. Ccl20, which is the gene with the highest fold-change after LMB-100 treatment (807-fold), encodes a chemokine and the ligand-receptor pair Ccl20-Ccr6 is responsible for the chemotaxis of dendritic cells (DC), effector/memory T-cells and B-cells (19). These results indicate that besides changes in immune-related genes in tumor cells, non-malignant cells also contribute to the change in gene profiles in response to LMB-100 treatment. These data indicate that signals from LMB-100 intoxicated cells act upon adjacent non-malignant cells and contribute to the development of the anti-tumor immune response.

Table 2.

Summary of Top 10 significantly upregulated genes

Region
Tissue	Cell	Gene	Fold change (Log10)	P-value	Adjusted P-value
√		Ccl20	2.907	2.46E-07	1.86E-05
√		Cxcl2	2.728	6.88E-08	6.16E-06
√		Ifnb1	2.290	4.38E-10	1.12E-07
√	√	Cxcl1	2.000	2.38E-09	3.68E-07
√		Ifna4	1.587	3.08E-08	3.08E-06
√		Csf2	1.459	4.87E-05	0.00179
√	√	Egr1	1.369	9.30E-13	1.65E-09
√		Cxcl3	1.356	9.13E-03	0.173
√	√	Serpinb2	1.336	2.43E-06	0.000135
√		Il6	1.283	3.10E-08	3.08E-06

TPO-MSLN mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶, day 1) and were treated with either LMB-100 (i.p., 3 mg/kg, day 5) or LMB-255 (i.p., 3 mg/kg, day 5). Tumors were collected 16 h after the treatment. For in vitro experiment, AB1-L9 cells were treated with either LMB-100 or LMB-255 as described in Method part, and cells were harvested at 16 h. Immune-related genes of both tumor tissues and cells in each group were analyzed by Nanostring. The top-10 significantly upregulated genes in tumor tissue were listed here. Check in “Tissue” represents that the gene was significantly upregulated in LMB-100 than LMB-255 in treated tumor tissue; Check in “Cell” represents that the gene was significantly upregulated in LMB-100 than LMB-255 in treated AB1-L9 cells. “Fold Change (Log₁₀)” and “P-value” were calculated by comparing mRNA counts between LMB-100- and LMB-255-treated tumor tissues.

The Anti-Tumor Immunity Induced by LMB-100 Requires the Presence of Non-Malignant Cells at the Treatment.

The results from immune transcriptional profiling suggest that non-malignant cells in tumors also play important roles in LMB-100-induced anti-tumor activity. To determine whether tumor cells killed by LMB-100 without the presence of tumor stroma can induce anti-tumor immunity, we treated tumor cells with LMB-100 before injecting them into the mice. To do this, five million AB1-L9 cells were treated with LMB-100 for 6 h or 48 h. Then the medium containing LMB-100 was removed and replaced with normal growth medium, and the cells were maintained for 15 d. At the end of this period, the number of colonies was counted and used to calculate how many cells had survived and had been killed. Although most of the cells treated with LMB-100 for 6 h were still attached to the flask when the fresh medium was added, 99.99% of tumor cells eventually died, and only around 140 colonies formed from an initial seeding of five million cells (Fig. 3 F and G). This result indicates that a 6-h treatment is sufficient for LMB-100 to initiate irreversible cell death. Prolonged treatment with LMB-100 to 48 h slightly decreased the number of colonies to around 110 (Fig. 3G).

For the vaccine studies (Fig. 3H), cells were treated for 6 h or 48 h with LMB-100 and were then inoculated (i.p.) into TPO-MSLN mice. On day 33, all mice inoculated with LMB-100 pre-treated cells were alive and appeared healthy, whereas all mice receiving untreated cells had died with large tumors. The surviving mice were then subcutaneously challenged with AB1-L9 cells to determine whether anti-tumor immunity was generated. We found that none of the mice that received cells treated with LMB-100 in vitro rejected the subcutaneous re-challenged tumor (Fig. 3I). The abdominal cavity of each mouse was examined to determine if any of the LMB-100-treated cells had grown, and we found that in 7/8 mice in the 6-h treatment group and 4/10 mice in the 48-h group, small tumors were present (Fig. 3J). The data shows that signals from LMB-100-induced tumor cell killing without the presence of non-malignant cells in tumor stroma are insufficient to generate anti-tumor immunity to reject the re-challenged cells. In addition, the upregulation of some important genes that are associated with lymphoid formation (e.g., Ccl21a, Ccl19, Ccr7, Cxcl13, and Cxcr5) was specifically detected in LMB-100-treated tumors, rather than in tumor cells treated in culture (Fig. 3E), supporting the critical roles of tumor stroma in the induction of anti-tumor immunity.

LMB-100-Mediated Tumor Cell Death Induces the Development of TLS.

Since immune cells are important components of the tumor stroma, we examined the hypothesis that they are essential in LMB-100-induced anti-tumor immunity. To determine which types of immune cells were influenced by LMB-100, we analyzed the tumor samples by Nanostring at 12, 24, and 48 h post-LMB-100 treatment. The untreated tumor at 12 h was used as the control. The results show a time-dependent increase in the abundance of multiple immune cells (Fig. 4A). These cells include B cells, T cells, NK cells, DC, macrophages, and neutrophils. Among these cells, B cells are greatly increased with upregulated absolute and relative cell type scores at 48 h (Fig. 4 A and B and SI Appendix, Fig. S3 A and B, P = 0.0002, P = 0.0287, respectively). To confirm the increase of B cells is induced by treatment instead of tumor progression, transcripts profiles of LMB-100-treated tumors at 48 h were compared to that of the control group at 48 h. A significantly higher B cell score was observed in the LMB-100 group than in the control group (Fig. 4 C and D, P = 0.0004, P = 0.0063 respectively). In addition, we observed a proportional decrease in NK cells, DCs, CD8⁺ T cells, and Th1 cells among all TILs at 48 h (Fig. 4B), whereas the absolute scores of these cells were either increased (DCs and Th1 cells) or not changed (CD8⁺ T cells and NK cells). The decrease is probably due to the greatly enlarged B cells population among all TILs in the tumor.

Fig. 4.

LMB-100 induced CD8⁺ T cell enrichment and B cell aggregation in tumor. A–D. TPO-MSLN mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶, day 1) and were treated with either LMB-100 (i.p., 3 mg/kg, day 5), LMB-255 (i.p., 3 mg/kg, day 5) or PBS (i.p., 200 μl, day 5). Tumors were collected 12 h, 24 h, or 48 h after the treatment. Immune-related genes were analyzed by Nanostring using a PanCancer Immune Profiling Panel. Signature sets of gene expression associated with different cell types were scored with NanoString Advanced Analysis modules. A and B. Z score heatmap of cell type score (A) and relative cell type score (B) for each tumor treated with PBS for 12 h (n = 4), LMB-100 for 12 h (n = 4), LMB-100 for 24 h (n = 4) and LMB-100 for 48 h (n = 4). P<0.05 is statistically significant. C and D. Violin graph of either B cell score or relative B cell score for each tumor treated with either PBS (n = 4) or LMB-100 (n = 4) and collected at 48 h. **P < 0.01, ***P < 0.001 compared with Con 48 h. E–G. TPO-MSLN Tg mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶, day 1) and were treated with either LMB-100 (i.p., 3 mg/kg, day 5) or PBS (i.p., 200 μl, day 5). Tumors were collected 24 or 48 h after each treatment. Sections of tumors were analyzed by IHC. E. Representative IHC image of tumor at 48 h after each treatment. CD8⁺ T cells were stained by anti-CD8a. B cells were stained by anti-CD45R. (Scale bar: 3 mm.) Histogram showing density of CD45R⁺ B cells (F) and CD8⁺ T cells (G) per tissue area in each slide. Mean ± SD. * P < 0.05.

We assessed the enrichment and spatial distribution of B cells and other immune cells in the tumor histologically. The percentage of hMSLN⁺ tumor cells was significantly decreased (67%, P = 0.001) in the LMB-100 treatment group when compared to the control group at 24 h (SI Appendix, Fig. S4 A and B). This decrease in tumor cells was accompanied by an increase in Casp3⁺ apoptotic cells (P = 0.0002, SI Appendix, Fig. S4 A and C). Also, at 48 h, the infiltrating CD45R+ B cells were increased 4.9-fold (P = 0.014) and CD8+ T cell infiltration was increased 2.95-fold (P = 0.014) (Fig. 4 E–G) Fig. 4 E and F. Most importantly, the tumor-infiltrating B cells were aggregated in lymphoid follicles called TLSs which were distributed throughout the tumor (Figs. 4E and 5A).

TLSs consist of clusters of immune cells with a distinct core of B cells. Analysis of consecutive sections of tumors treated with LMB-100 shows that regions rich in B cells are surrounded by CD4+ and CD8+ T cells. Also, regions rich in follicular dendritic cell-like CD11c+ cells are present within the core of B cells (Fig. 5A). In contrast, though B cell clusters are detectable in untreated tumors, they are smaller and fewer in number, and most do not have an intact TLS structure (Fig. 5A). The number of TLSs per tissue area is increased by 1.5-fold (P = 0.021) at 24 h and 7.9-fold (P = 0.014) at 48 h in response to LMB-100 (Fig. 5B). The overall number of TLSs is also higher than in the control group at 24 h (1.78-fold, P = 0.28) and 48 h (4.27-fold P = 0.014) (Fig. 5C).

Fig. 5.

LMB-100 mediated tumor cell death induced development of TLS in tumor. A–C. TPO-MSLN Tg mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶, day 1) and were treated with either LMB-100 (i.p., 3 mg/kg, day 5), LMB-255 (i.p., 3 mg/kg, day 5) or PBS (i.p., 200 μl, day 5). Tumors were collected 24 h or 48 h after each treatment. A. Representative IHC image that shows the architecture of TLS in either tumor treated with LMB-100 or PBS at 24 h (Scale bar: 2 mm). Sections were taken consecutively to spatially analyze the different immunostainings. Populational characterization of TLS in tumor treated with LMB-100 was shown in the magnified image (On the right, Scale bar: 200 μm). Cells were stained by CD8a (cytotoxicity T cell), CD4 (T helper cell), CD45R (B cell), CD11c (Dendritic cell), hMSLN (tumor cell) and CASP3 (apoptosis cell). B. Histogram showing density of TLS per tissue area in each slide at 24 h or 48 h. Mean ± SD. *P < 0.05. C. Histogram showing total number of TLS in each tumor in each slide at 24 h or 48 h. Mean ± SD. *P < 0.05. D–L. Histogram showing counts of genes that are associated with TLS development. TPO-MSLN Tg mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶, day 1) and were treated with either LMB-100 (i.p., 3 mg/kg, day 5) or PBS (i.p., 200 μl, day 5). Tumors were collected 16 h or 48 h after each treatment. Genes were analyzed by Nanostring. Mean ± SD. *P < 0.05, ***P < 0.005.

To understand whether the induced TLS is a benefit to the anti-tumor activity, we performed a principal component analysis (PCA) with the 11 TLS signature genes using the RNAseq data extracted from a previous publication (20). As shown in SI Appendix, Fig. S5 A and B, mice that respond to immune checkpoint blockade (ICB) therapies (red dots) are well separated from non-responders (blue dots), indicating that the TLS signature gene expression profiles are intrinsically different among the two groups. The expression of 8 TLS signature genes, highlighted in SI Appendix, Fig. S5C, increased more than twofold in responders compared to the non-responders. The complete data is in SI Appendix, Table S2. This result is consistent with our findings that upregulation of TLS gene expression improves treatment efficacy.

To determine whether TLSs are induced in the absence of cell-killing activity, tumors were treated with inactive LMB-255. No change in the percentage of hMSLN+ cells (P = 0.683, SI Appendix, Fig. S6 A and B) or the number of TLSs in the tumor (P = 0.927, SI Appendix, Fig. S6C) was detected, indicating that TLS formation is induced by LMB-100 mediated tumor cell killing.

LMB-100 Induces TLS Formation by Influencing Chemokines/Cytokines Levels in Tumor.

To understand the mechanism of LMB-100-induced TLS formation, we analyzed genes closely associated with TLS development. LTα₁β₂ and LIGHT produced by lymphoid tissue inducer (LTi) cells are crucial to promoting the maturation of LTβR⁺ lymphoid tissue organizer (LTo) cells, which further produces Ccl19, Ccl21, and Cxcl13 to recruit more LTi to form a positive feedback loop (21). Cxcl13-Cxcr5 and Ccl19/Ccl21-Ccr7 pairs are important in the recruitment of T cells and B cells to lymphoid structures (22, 23) and are known to be molecular markers of TLS formation (24). As shown in Fig. 5 D–L, the expressions of Lymphotoxin β, Light, Ccl21a, Ccl19, Ccr7, Cxcl13, as well as Cxcr5 were significantly upregulated in LMB-100-treated tumors at 16 h or 48 h. The upregulation of Ccl21a was confirmed at the protein level (SI Appendix, Fig. S8C). Importantly, the expression of these genes was not detected in tumor cells treated in culture (16 h, Fig. 3E), indicating that tumor stroma is required for the increase of these mRNA transcripts.

Further experiments have shown that 10/11 of TLS-signature chemokines (25) (i.e., Ccl2, Ccl3, Ccl4, Ccl5, Ccl8, Cxcl9, Cxcl10, Cxcl11, Ccl21a, Ccl19, and Cxcl13) are significantly upregulated by LMB-100 at either 16 h or 48 h or both (Fig. 5 H, I, and K and SI Appendix, Fig. S7 A–H), and the increase in Ccl4, Ccl5, Ccl8, and Ccl21a were confirmed at the protein level (SI Appendix, Fig. S8 C–M). Lymphangiogenic factor VEGF-C, as well as adhesion molecules Icam-1 and Vcam-1 that are associated with the formation of high endothelial venule (HEV) (25), a specialized blood vessel mediating lymphocyte trafficking to TLS, are also upregulated in mRNA in tumor treated with LMB-100 (SI Appendix, Fig. S7 I–K), and the increase in Icam-1 at 48 h was confirmed by the cytokine assay (SI Appendix, Fig. S8 N–P). Together, these results support the conclusion that immunotoxin-mediated tumor cell death stimulates TLS development by influencing cytokines and chemokines that are associated with lymphoid formation, lymphocyte chemotaxis, and HEV development. In addition, changes in TLS-associated genes and proteins are time-dependent. For example, the upregulation of Ccl5 lasts from 16 h to 48 h (SI Appendix, Fig. S8I), whereas an increase in Ccl4 is only observed at 16 h (SI Appendix, Fig. S8H).

CD8⁺ T Cells, CD4⁺ T Cells, and B Cells Are Required for the LMB-100-Induced Anti-Tumor Immunity.

We found that depletion of CD8⁺ T cells by anti-CD8a antibody shortened the survival time of mice treated with LMB-100 (Fig. 6 A and B) (P < 0.0001). The CR rate decreased from 64% (14/22) with the control antibody to 12% (2/17), with anti-CD8a (Table 3). This indicates CD8 cells are important for the initial response to LMB-100, probably by eliminating those tumor cells not killed by LMB-100. Depletion of B cells also effectively reduced the number of complete remissions in mice treated with LMB-100 from 12/17 (71%) to 3/13 (23%) and significantly decreased survival (P = 0.002) (Fig. 6 C and D and Table 3). Depletion of CD4⁺ T cells by anti-CD4 antibody also significantly reduced survival of mice (P = 0.0196), though to a lesser extent than of the anti-CD8a (Fig. 6B). The CR rate of mice with anti-CD4 treatment was 32% (7/22). Depletion of NK cells by anti-asiaio-GM1 antibody did not influence the survival of mice (P = 0.28, Fig. 6 E and F).

Fig. 6.

LMB-100-induced tumor complete regression needs CD8⁺ T cells, CD4⁺ T cells, and B cells. A. Scheme of the CD4⁺ and CD8⁺ T cells depletion assay. TPO-MSLN mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶ cells, day 1) and were treated with LMB-100 (i.p., 3 mg/kg, days 5 and 9) together with anti-CD8a antibody (i.p., 200 μg/dose, days 5, 9, and 13, n = 17), anti-CD4 antibody (i.p., 200 μg/dose, days 5, 9, and 13, n = 22) or Rat IgG2b as control (i.p., 200 μg/dose, days 5, 9, and 13, n = 22). Data were pooled from 3 experiments. B. Survival of mice in each group before the re-challenge assay. All CR mice were re-challenged by AB1-L9 cells (s.c., 2 * 10⁶ cells, days 51–67). *P < 0.05, ****P < 0.001 compared with LMB-100 + Rat IgG2b group. C. Scheme of the B cells depletion assay. TPO-MSLN mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶ cells, day 1) and were treated with LMB-100 (i.p., 3 mg/kg, days 5 and 9). B cells were depleted by injecting with a mixture of monoclonal antibodies including rat anti-mouse CD19 (clone 1D3, i.p., 150 μg), rat anti-mouse B220 (clone RA3-6B2, i.p., 150 μg), and mouse anti-mouse CD22 (clone CY34, i.p., 150 μg) at days 2, 10, and 18; and mouse anti-rat k chain antibody (clone i.p., 150 μg) at days 4, 12, and 20. Data were pooled from two experiments. D. Survival of mice in each group. **P < 0.01 compared with LMB-100 group. E. Scheme of the NK cell depletion assay. TPO-MSLN mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶ cells, day 1) and were treated with LMB-100 (i.p., 3 mg/kg, days 5 and 9, n = 9). Anti-asialo-GM1 antibody was administrated for NK cell depletion. (i.p., 30 μl/dose, days 4, 6 and 11, n = 13). F. Survival of mice in each group. Ns, not significantly different from LMB-100 group.

Table 3.

Summary of mice that reach CR in each experiment

Treatment†	Total mice	Mice with CR
LMB-100 + IgG 2b	22	14
LMB-100 + anti-CD8a	17	2
LMB-100 + anti-CD4	22	7
Treatment‡	Total mice	Mice with CR
LMB-100 + PBS	17	12
LMB-100 + Mix-Ab*	13	3
Treatment§	Total mice	Mice with CR
LMB-100 + PBS	9	4
LMB-100 + anti-asialo-GM1	13	9

TPO-MSLN mice were inoculated with AB1-L9 cells (i.p., 5 * 10⁶ cells, day 1) and were treated with LMB-100 (i.p., 3 mg/kg, days 5 and 9). CD8⁺ T cells, CD4⁺ T cells, B cells and NK cells were depleted as described in Method part as well as in Fig. 6. Mice with completely regressed abdominal tumor at experimental endpoint were considered as CR mice.

^*Mix-Ab: anti-CD22, anti-CD19, anti-B220, and anti-Rat IgG.

^†Data in Fig. 6B.

^‡Data in Fig. 6D.

^§Data in Fig. 6F.

Discussion

We have used an immunocompetent mouse model to show that immunotoxin-mediated tumor cell death induces anti-tumor immunity by reprograming the tumor immune microenvironment, which is critical for the complete elimination of tumors. The induction of anti-tumor immunity is dependent on both tumor targeting and killing activities of LMB-100 and the participation of multiple immune cells including CD8⁺ T cells, CD4⁺ T cells, and B cells. The CR mice develop a systemic anti-tumor immunity that rejects tumor cells with or without hMSLN expression. Transcriptional profiling of the tumor immune microenvironment indicates a time-dependent enrichment of many types of immune cells (B cells, T cells, NK cells, DC, macrophages, and neutrophils) within 48 h. B cells account for a large proportion of the total immune cells in tumor tissues. Notably, tumor stroma is critical for the development of immunotoxin-induced anti-tumor immunity, since the absence of non-malignant cells in the tumor microenvironment at the early stage of treatment failed to produce anti-tumor immunity in mice. Histological analysis shows that TLSs consisting of clusters of immune cells with a distinct core of B cells are induced and significantly increase in response to LMB-100 treatment. Genes encoding proteins associated with TLS development, such as Ccl19/Ccl21a-Ccr7 and Cxcl13-Cxcr5, are exclusively increased in non-malignant cells in the tumor microenvironment, indicating a role of these cells in lymphocyte recruitment in response to immunotoxin-mediated tumor cell death.

New Animal Model to Study the Immunomodulatory Effects Induced by LMB-100.

Previous animal studies showed that hMSLN-targeted immunotoxins synergize with anti-PD-1/anti-CTLA-4 in the treatment of lung adenocarcinoma (anti-PD-1), breast cancer (anti-CTLA-4) and mesothelioma tumors (anti-CTLA-4) which grow subcutaneously. In these studies, immunotoxin by itself did not upregulate immune-related transcripts or increase CD8⁺ T cells in tumors (14, 15). In this paper, we show that immunotoxin treatment alone killed many cells and upregulated immunomodulatory signals in a peritoneal mesothelioma model – a more clinically relevant model. In this model, the orthotopic tumor leads to death of mice within ~20 d, so treatment on day 5 is at about 25% of the time to death from the cancer. Though less effective, late treatments (starting at D7 or D9) also significantly improve the survival of mice (Fig. 2D). Since LMB-100 alone induces anti-tumor immunity, the addition of checkpoint inhibitors, which also induce immunity provide a way to improve the therapeutic activity of LMB-100 in mesothelioma.

One factor which may contribute to the difference in response is the location where tumor cells grow. Peritoneal tumors have an environment that is rich in immune cells, including B cells, macrophages, DC, and T cells (26), allowing their direct interaction with tumor cells during tumor development. A previous report (27) showed that melanoma growing in the peritoneal cavity contained intra-tumoral TLSs, but the same tumor in subcutaneous melanoma did not, indicating a more immune active microenvironment in the peritoneal cavity. This is also consistent with our results and clinical observations that peritoneal mesothelioma contains TLSs-like lymphoid aggregates (28). Considering the important role of tumor stroma in the induced anti-tumor immunity by the LMB-100 monotherapy, we believe the presence of immune cells and stromal cells in the tumor environment are the critical factors which influence CR.

The Anti-Tumor Immunity Induced by LMB-100.

We have observed that LMB-100-induced anti-tumor immunity helps mice reach initial tumor CRs as well as prevents the growth of the re-injected cells. Among the pathways affected by LMB-100 treatment, cell cycle, pathogen responses, and senescence pathways were changed at the earliest time at 12 h (Fig. 3 A and B), followed by increase in immune-related transcriptional pathways at 16 h (cytokine & chemokine, inflammation), TLSs boosting at 24 h (Fig. 5B) and enrichment of CD8⁺ T cells and CD45R⁺ B cells at 48 h post-treatment (Fig. 4 E – G)). Depletion of either CD8⁺ T cells or B cells significantly reduces mice’s survival. Though the absolute transcriptional score of NK cell was also increased (Fig. 4A), depletion of NK cell did not influence the therapeutic effect of LMB-100 (Fig. 6F). Our data suggest that LMB-100 induced tumor cell death and reprogrammed the transcriptional profiles of cells in the tumor microenvironment which then promoted the recruitment of immune cells (including B cells) to cluster in TLSs, facilitating the CD8⁺ T cells to kill the malignant cells.

Immunotoxin-Induced TLS Formation.

We show here for the first time that LMB-100-mediated cell death induces the recruitment of B cells and the development of TLSs (Fig. 5A). TLSs are ectopic lymphoid organs that form in non-lymphoid tissues, including tumors (25). Generally, anti-tumor immune responses occur in secondary lymphatic organs (SLO) and require the migration of DCs from the tumor to the SLO to achieve the antigen presentation to prime B cells and T cells (25). After antigen stimulation, B cells undergo intense proliferation and some of them differentiate into antibody-producing plasma cells which produce anti-tumor antibodies (29); the antigen-primed naïve T cells develop into effector T killer cells which recognize tumor antigens and migrate back to the tumor site to destroy the cancer cells (30). The development of local TLSs in tumor tissue raises the possibility that all these steps can occur in the tumor rather than in the SLOs. This could shorten the distance and time for antigen presentation and T cell homing and contribute to a more effective anti-tumor response.

In the present study, very few TLSs were found in the untreated mesothelioma and their size and number were increased after LMB-100 treatment (Figs. 4E and 5 A–C). Direct tumor cell killing by LMB-100 was also shown by the increased Casp3⁺ apoptosis cells (SI Appendix, Fig. S4C). There were CD11c⁺ DC cells present in the center of the B cell-rich regions, forming a germinal center (GC)-like regions, consistent with the role of DCs in promoting B-cell maturation and activation. The GC-like regions were surrounded by CD4⁺ and CD8⁺ T cells (Fig. 5A). Both increased B cells and CD8⁺ T cells are favorable prognostic markers for mesothelioma patients (31–33), and the presence of TLSs is a portent of improved responsiveness to immunotherapy in many kinds of cancers (melanoma, sarcoma, and renal cell carcinoma) (24, 34, 35). Taken together, the present study suggests immunotoxin might be more effective if it is used as an adjuvant in combination with ICB in the treatment of mesothelioma.

The modulation of TLS is an attractive option to induce anti-tumor immunity in an immunosuppressive tumor microenvironment. Clinically, there are reports showing chemotherapies will not influence the TLS density and both neoadjuvant radio- or chemotherapy are reported to impair TLS maturation in lung squamous cell carcinoma patients (36). In preclinical animal studies, a recent study shows that intratumorally delivered low-dose stimulator of interferon genes agonist supports TLS formation in mouse melanoma (37). In addition, an anti-CD40 agonist also induces the formation of TLS in the proximity of meningeal tissue in a preclinical glioma model (38). As far as we know, immunotoxin is the only agent that promoting targeted cell death while inducing TLS formation, probably due to it having a unique cell-killing mechanism.

Non-Malignant Cells in the Development of Anti-Tumor Immunity.

Our previous study has shown that immunotoxin induces tumor cell death and the release of immunogenic cell death markers (ATP and Calreticulin) in vitro (16). In the present study, we further showed genes associated with LTi and LTo chemotaxis (LTβR ligands, Ccl19, Ccl21a, and Cxcl13), which are crucial in TLS development, were upregulated specifically in non-malignant cells in the tumor microenvironment, in response to LMB-100 treatment, whereas tumor cell won’t produce these signals (Fig. 5 D–L). The presence of non-malignant cells in the tumor at the time of treatment is required to initiate sufficient CTL-dependent immune responses against mesothelioma in vivo (Fig. 3 H–J). In addition, genes associated with HEV development were also upregulated by LMB-100, and interestingly, these signals could be detected both in vitro and in vivo (SI Appendix, Fig. S7 I–K and Fig. 3 D–E).

Some reports show that stromal cells are important in forming TLSs (21, 25). We therefore propose that TLSs induced by LMB-100 are important components of the anti-tumor immunity realized through the participation of tumor stromal cells. It is hypothesized that LMB-100-mediated tumor cell death first sends signals to tumor stromal cells which secret cytokines to recruit lymphocytes to develop into TLSs. These TLSs then contribute to the establishment of efficient anti-tumor immunity by developing timely and localized immune responses. Further studies are still in need to understand which type of non-malignant cells in the tumor microenvironment responds for cytokine secretion, and which specific signals are transferred from tumor cells to these cytokine-producing cells.

In sum, we show immunotoxin-mediated cell death induces orthotopic mesothelioma CR and the development of anti-tumor immunity in TPO-MSLN mouse. The regression of the primary tumor requires CD8⁺ T cells, CD4⁺ T cells as well as B cells. LMB-100 reshapes the tumor immune microenvironment by upregulating immune-related transcripts in both tumor cells and non-malignant cells in tumor stroma. The induced anti-tumor immunity requires the presence of tumor stroma. At last, LMB-100-mediated tumor cell death induces the development of TLS in the solid tumor microenvironment by upregulating chemokines and cytokines that are associated with TLS formation. Our results suggest agents that stimulate B cells, tumor stromal cells, and TLSs have the potential to synergize with LMB-100 to improve its therapeutic activity in mesothelioma.

Materials and Methods

Additional details about the methods we used in this study are provided in SI Appendix, Materials and Methods.

Cytotoxicity Assay.

AB1 is a mouse mesothelioma cell line purchased from Sigma-Aldrich (39). AB1-L9 cells are AB1 cells transfected with MSLN to stably express hMSLN on the cell membrane (40). hMSLN-targeted immunotoxin LMB-100 (SS1-Fab-LO10R 456A) and inactivated LMB-100 (LMB-255, LMB-100-del fur-del553) were manufactured by Roche (Basel, Switzerland). Endotoxin levels for LMB-100 and LMB-255 are below 5 EU/mg. Immunotoxin LMB-34 (BM24-Fab-LO10R-456A) targeting hBCMA was produced in our laboratory.

The cells were seeded at 3000 cells/well in 96-well plates. The medium was replaced with fresh medium with various concentrations of LMB-100, LMB-255, or LMB-34 after cell attachment. Normal medium and medium with 0.1 mg/mL cycloheximide (Sigma-Aldrich, St. Louis, MO) were used as the blank and positive controls, respectively. Cells were incubated for 3 d before cell viability was assessed using a WST-8 cell counting kit (Dojindo Molecular Technologies, Kumamoto, Japan).

Mouse Experiments.

All mouse experiments were approved by the NCI Animal Care and Use Committee (Protocol number: LMB-014). The BALB/c TPO-MSLN transgenic mouse expressing hMSLN in the thyroid under the control of TPO promoter was produced in our lab as described previously (17). The TPO-MSLN mice do not reject genetically modified mouse tumor cells expressing hMSLN.

Five million AB1-L9 tumor cells were injected (i.p.) in TPO-MSLN mice (10–20 wk) on day 1. Plasma hMSLN concentration of each mouse before tumor injection (day 1) and before the first treatment was measured (MSD R-Plex ECL assay). Mice that did not show an increase in plasma hMSLN concentration were excluded from further analysis. Dosage and treatment schedule of immunotoxins, anti-CTLA-4 (Clone 9D9, mouse IgG2a, Absolute Antibody), or PBS are listed in each figure legend. Mice were euthanized if their weight loss exceeded 15% or looked morbid. The survival of mice in each treatment group was recorded. After the mice were euthanized, the tumors in the abdominal cavity were collected. Usually, tumors attached to the gastrosplenic ligament (GS-ligament) were the largest in the abdominal cavity and were used for further analysis. The weight of the tumor and the attached GS ligament were recorded.

In re-challenge assays, the mice with CRs were re-challenged by either 2 × 10⁶ AB1 or AB1-L9 cells given subcutaneously 1.5 mo after initial cell inoculation. The size of the tumors was measured 2–3 times a week, and mice were euthanized if tumor size exceed 400 mm³ or if the mice looked morbid. Tumors in the abdominal cavity of each mouse in the re-challenge assay were checked at each experimental endpoint (euthanize day or last day in the record). Mice that died with both subcutaneous tumor and tumor in the abdominal cavity were not considered as CR from the treatment. Mice rejecting the subcutaneous tumor were considered to have developed anti-tumor immunity.

Cell Depletion Assay.

Anti-mouse CD8a antibody (clone 2.43, BioxCell) and anti-CD4 antibody (clone GK1.5, BioxCell) were used to deplete CD8⁺ T cells and CD4⁺ T cells as described in a previous study (15). Briefly, mice with tumors were treated with LMB-100 (i.p., 60 μg/dose) on days 5 and 9, and each antibody or isotype control (rat IgG2b) was administered (i.p., 200 μg/dose) on days 5, 9, and 13. NK cells were depleted with anti-asialo-GM1 antibody (BioLegend, Clone Poly21460) on days 4, 6, and 11 as described previously (i.p., 30 μl with 170 μl PBS/dose) (41). B cell depletion was performed as previously described (42). Mice were injected with a mixture of monoclonal antibodies: rat anti-mouse CD19 (clone 1D3, i.p., 150 μg), rat anti-mouse B220 (clone RA3-6B2, i.p., 150 μg), and mouse anti-mouse CD22 (clone CY34, i.p., 150 μg) on days 2, 10, and 18; after 48 h of each injection, the mice were further injected with anti-rat kappa immunoglobulin light chain antibody (clone MAR 18.5, i.p., 150 μg). To check whether the antibody treatment impaired the development of anti-tumor immunity, the mice with CRs were re-challenged with 2 × 10⁶ AB1-L9 cells about 1.5 mo after the initial cell inoculation.

Nanostring.

Mouse tumors at the GS-ligament were collected at 12, 16, 24, or 48 h post-treatment with LMB-100, LMB-255, or PBS. The tumor tissue of each mouse was snap-frozen, and RNA was isolated using the RNeasy Mini kit (Qiagen) according to the manufacturer’s guide. Tumor responses were monitored by plasma hMSLN which reflects tumor size, and the top 50% responsive tumors were used to be further analyzed. AB1-L9 cells (5 × 10⁶) were treated with 2 ml medium containing either 60 μg LMB-100 or 60 μg LMB-255 in a 50-mL tube for 30 min. Cells were then transferred into a 100-mm dish with another 6 ml fresh medium. After 16 h, harvested all the cells in each flask and centrifuged at 3,000 rpm for 5 min. Total RNA of cells in each dish was isolated using the RNeasy Mini kit (Qiagen).

The quantity and quality of RNA were analyzed by a Nanodrop, ND-100 spectrophotometer. Samples of 50 ng of total RNA were analyzed on the nCounter platform with a PanCancer Immune Profiling Panel which contains 770 immune-related genes (NanoString Technologies). For each batch of samples to be analyzed, the raw data were normalized to the geometric mean of the internal positive controls as well as to housekeeping genes selected by nSolver Analysis software. The same housekeeping genes were used, when we compare the RNA data between in vitro and in vivo samples. “DE analysis” was conducted using an “optimal” mode to generate the volcano plots to show fold-change vs. P-value of each gene. “Cell Type Profiling Module” and “Pathway Scoring Module” were used to calculate scores for the expression of gene sets of different immune cell types and pathways. The increase in pathway scores was confirmed by the directed global significance scores analysis. Z scores of cell type and pathway scores were calculated across all tumor samples before generating heat maps for visualization. In cell type analysis, cell type score represents the abundance of each cell type, whereas relative cell type abundance measures the abundance of each immune cell relative to overall lymphocytes.

Cytokine/Chemokine Analysis.

Tumor samples at 16 h or 48 h of treatment were flash-frozen with liquid nitrogen and stored at −80°C. Tissue homogenates were prepared from frozen tissue with EZ-Grind™ (G-bioscience) based on the manufacturer’s description and diluted to 800 μg/mL (total protein) to be used for cytokine detection. Multi-Plex Discovery Assay (Eva Technologies, Calgary, AB, Canada) was used to detect TLS-associated cytokines and chemokines which include TLS development-associated genes (Lymphotoxin α, Light, Ccl21a, Ccl19, and Cxcl13), TLS signature gene (Ccl2, Ccl3, Ccl4, Ccl5, Ccl8, Cxcl9, Cxcl10, Cxcl13, Ccl21a, and Ccl19) (25), and genes associated with HEV activity (Icam-1 and Vegf-c). For TLS signature genes, Ccl18 was not analyzed because it does not have a rodent homologue. Cxcl11 and Vcam-1 (sample 1:500 diluted) were detected with traditional ELISA assays (Thermo Fisher).

Bioinformatic and Statistical Analyses.

Previously published mouse sensitization to checkpoint blockage RNA-seq data (20) were downloaded from Gene Expression Omnibus (GEO) database, accession number GSE117358. Briefly, Balb/c mice were bilaterally inoculated with AB1 tumors and were divided into two groups based on whether they responded to anti-CTLA-4 and anti-PD-1 treatment or not; one tumor was surgically removed before the treatment was analyzed using bulk RNA-seq. Downloaded gene counts were normalized and the fold change comparing the responders against non-responders was calculated by using DESeq2 (43). The PCA was performed on the 11 TLS signature genes listed before.

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.