NQO1 targeting prodrug triggers innate sensing to overcome checkpoint blockade resistance

Lack of proper innate sensing inside tumor microenvironment (TME) limits T cell-targeted immunotherapy. NAD(P)H:quinone oxidoreductase 1 (NQO1) is highly enriched in multiple tumor types and has emerged as a promising target for direct tumor-killing. Here, we demonstrate that NQO1-targeting prodrug β-lapachone triggers tumor-selective innate sensing leading to T cell-dependent tumor control. β-Lapachone is catalyzed and bioactivated by NQO1 to generate ROS in NQO1high tumor cells triggering oxidative stress and release of the damage signals for innate sensing. β-Lapachone-induced high mobility group box 1 (HMGB1) release activates the host TLR4/MyD88/type I interferon pathway and Batf3 dendritic cell-dependent cross-priming to bridge innate and adaptive immune responses against the tumor. Furthermore, targeting NQO1 is very potent to trigger innate sensing for T cell re-activation to overcome checkpoint blockade resistance in well-established tumors. Our study reveals that targeting NQO1 potently triggers innate sensing within TME that synergizes with immunotherapy to overcome adaptive resistance.

In the present paper, the authors study the effect of NQO1-based tumor targeting prodrug, βlapachone on induction or triggering of innate sensing to overcome the checkpoint blockade resistance in cancer cells. Several interesting observations reported in the manuscript show that βlapachone induces immunogenic cell death, leading to the TLR4/MyD88 signaling pathway leading to type 1 interferon signaling and inducing Batf3 dendritic cell dependent cross-priming, resulting in innate and adaptive immune responses. All the experimental procedures are clearly explained, and the results shown in figures support the authors, interpretations and conclusions. Overall the study is robust and well done. However, there are several concerns which need to be clarified and addressed.
Major comments: 1. In this manuscript, there are rather many statements like 'data not shown'. I think it would be better to delete the expression 'data not shown' or show them.
2. As described in introduction section, β-lapachone is NQO1 bioactivatable drug. In Figure 1a and h, β-lapachone caused cell death in NQO1-deficient B16 cells, but β-lapachone did not show cancer cell death in Figure 1d. The authors should clarify the data.
3. The first title in the Result is "NQO1-targeting prodrug β-lap suppresses murine tumor growth in a NQO1 dependent manner both in vitro and in vivo". It is not accurate to refer the in vitro cell death as suppression of tumor growth. The authors and other investigators have already published numerous papers showing that βlapachone suppresses tumor growth in NQO1 dependent manner. Given that this is already a wellknown fact, this reviewer fell it is not necessary to discuss this in detail consuming one and half pages of the manuscript. I recommend to make this section shorter and use this space to show "data not shown" or the "supplementary figures". 4. The authors obtained data using mainly NQO1-expressing MC38 and TC-1 cell lines. The authors should perform a loss-of-function and gain-of-function studies to clarify the data in 6. In Figure 4G, the authors should clarify whether β-lapachone induces HMGB-DAMP signal pathway to activate of CD8+ T cells using cells which is knockdown of HMGB1.
Reviewer #2 (Remarks to the Author): NQO1-based tumor targeting prodrug triggers innate sensing to overcome checkpoint blockade resistance The authors have done a heroic amount of work in this manuscript I have some specific questions that need addressing In general, I would like to see the cell number used in each in vivo tumor model (s.c) displayed at the top of each figure. Some of the tumor growth curves are different from figure to figure using the same cell lines and this is sometimes due to injected cell number and would help the reader greatly.
Also, "Spaghetti plots" of each tumor grown in individual mice should be shown in the supplemental data so growth of each tumor can be visualized. Error bars are impressively small given the low number of mice used in each group as these lines are known to have sizable variation in B6 mice. Spaghetti plots would help here.  Fig. 2A

Response:
We appreciated the reviewer's suggestion. In the revised manuscript, we deleted the expression 'data not shown' and provided more details of the data in the supplementary figures. Comment: 2. As described in introduction section, β-lapachone is NQO1 bioactivatable drug. In Figure 1a and h, β-lapachone caused cell death in NQO1-deficient B16 cells, but β-lapachone did not show cancer cell death in Figure 1d. The authors should clarify the data.

Response:
We agree with the reviewer that we should clarify the data that were generated from different assays and with different treatment times. We examined several cancer cell lines with or without NQO1 expression to confirm the tumor targeting effect of β-lap. We also used cell lines that were reconstituted or knocked out of NQO1 expression to confirm the targeting effect.
In Fig. 1a, 4000 cells (per well) in 48-well plates were treated with β-lap (0-8 μM) for a 3-hr followed by washing and replacing medium. 4 days later, cell survival was determined by SRB cytotoxicity assay. The results showed that tumor cell lines (MC38, TC-1 and Ag104Ld) that express high level of NQO1 were sensitive to β-lap exposure with a lethal dose around 2-4μM.
In contrast, NQO1 deficient cell lines, B16 and Panc02 were resistant to β-lap exposure. Only much higher concentration of β-lap (≥8 μM) had obvious cytotoxicity which might result from "off target" effect. In Fig. 1d, 4000 cells (per well) in 96-well plates were exposed to β-lap for 3 hr and cell survival was assessed 48 hr later with SRB cytotoxicity assay. Here, we did not use 8 µM of β-lap treatment because of the potential off target effect in this experiment. In both figure   1a and 1d, 4µM of β-lap treatment almost killed all the NQO1 positive cells but had no effect on the negative cell lines. The variation for 6 µM treatment might be because of the different treatment time and procedure (48 well vs 96 well, 4 days vs 2 days). In our revised manuscript, we deleted the 8 µM treatment group in figure 1a. Response: Thanks for your suggestion. We agree with the reviewer that the first title should be revised. In page 5, lines 2-3, the first title was changed to "NQO1-targeting prodrug β-lap induces murine tumor cell death in vitro and suppresses murine tumor growth in mice in a NQO1 dependent manner".
We agree with the reviewer that numerous papers have shown that β-lap suppresses tumor growth in NQO1 dependent manner. However, all the previous studies, including our own studies, mainly focused on the tumor-direct killing effect in vitro and in vivo by using human tumor lines and human xenograft immunodeficient model, with no mechanisms and evidence related to the host immune responses. Our current study is actually the first time to report the antitumor efficacy of β-lap in vivo through NQO1-dependent immune-mediated killing in immune competent hosts. Syngeneic murine tumor models allow us to better understand the role of host immunity during β-lap treatment. Therefore, we need to screen and test various murine lines that allow us to do gain-of-function or loss-of-function studies in immune competent hosts.
We would like to clarify the NQO1-dependent killing effect in the murine tumor lines in vitro first, then we are able to use NQO-1 positive lines to study the immunologic mechanisms of βlap in tumor killing with immune competent hosts. tumor bearing mice ( Fig. 1j and fig.2a) and MC38 tumor bearing immunodeficient mice ( Fig.2a,b). Considering the nonresponse of NQO1 knockout MC38 tumor in immunocompetent mice, we did not further deplete the CD8 + T cells in WT mice and test the antitumor effect of βlap. Instead, we used B16 model to do gain of function study. B16 tumor was not response to β- Response: We appreciate the reviewer's suggestion. As suggested, we detected the CD8 + T cell infiltration by immunofluorescence staining. Similarly, we observed that only few CD8 + T cell had infiltrated primary tumor in vehicle treatment group, β-lap or anti-PD-L1 monotherapy induced more CD8 + T cells infiltration, and combination treatment dramatically magnified the infiltration of CD8 + T cells (supporting figure R2).

Comment: 4. The authors obtained data using mainly NQO1-expressing MC38 and
The figure (Fig.R2) has been added to the Supplementary Fig. 9. In pages 11, lines 32-33 "Immunostaining of CD8 + T cells further confirmed the increased tumor infiltrating CD8 + T cells in TME ( Supplementary Fig. 9)." will be added to describe this.
We agree with the reviewer that we should determine the cytotoxicity of β-lap to tumors in vivo to support the tumor growth data. According to this comment, we have further detected the  priming of CD8 + T cells, BMDCs were exposed to OVA protein for 4 hours, then naïve OT1 CD8 + T cells and the supernatants from β-lap-treated tumor cells (with/without αHMGB1 antibody) were added. IFNγ secretion was measured to evaluate the capability of DCs to prime the antigen specific T cells. The results showed that β-lap treatment greatly increased the IFNγ secretion and this effect was diminished when HMGB1 was neutralized or knocked down (Supporting figure R4c). Together, these data further confirmed that β-lap treatment induced dendritic cell-mediated T cell cross-priming and activation was in a tumor-derived HMGB1 dependent manner.
The figure (Fig.R4a and c) has been added to the Supplementary Fig. 7. In pages 10, lines 8-15 "HMGB1 was found to be critical for β-lap-induced immunogenicity in three experiments: (i) BMDCs were exposed to OVA protein for 4 hours, then naïve OT1 CD8+ T cells and the supernatants from β-lap-treated tumor cells (with/without αHMGB1 antibody or knockdown the HMGB1 in MC38 cells). IFNγ secretion was measured to evaluate the capability of BMDCs to prime the antigen specific T cells. The results showed that β-lap-treatment greatly increased the IFNγ secretion, and this effect was significantly diminished when HMGB1 was neutralized or knocked down ( Supplementary Fig. 7a,b)." will be added to describe this. with β-lap (4 μM) for 3 hr followed by washing and replacing fresh medium, BMDCs with/without αHMGB1(20μg/ml) were cocultured with β-lap-treated tumor cells for another 24 hours. CD40 and CD86 expression on BMDCs were determined by flow cytometry analysis. (c) MC38 sheGFP or MC38 shHMGB1 cells were treated with β-lap (4 μM) for 3 hr followed by replacing fresh medium for another 24 hours. BMDCs were exposed to 40µg/ml OVA protein for 4 hours, then naïve OT1 CD8 + T cells and the supernatants from β-lap-treated tumor cells (with/without αHMGB1 antibody) were added and allowed to incubate for another 48 hours. IFN-γ was determined by cytometric bead array assay.

Supporting figure R5:
The DNA fragments containing the target sequence for NQO1 knockout were amplified by PCR from genomic DNA from MC38 sgNQO1 5# cells. Then the PCR product was purified and sent out for sequencing. Allele 1 lost 6 base pairs, 2 amino acids, but no shift mutation. Allele 2 lost 1 base pair which cause shift mutation. Fig 2A,B -Why do MC38 grow so differently in WT Vehicle mice in Fig. 2A

Response:
Thank the reviewer for pointing this out. In Figure 2A and 2B, 6×10 5 MC38 cells were subcutaneously transplanted into C57BL/6 WT mice (n=5). Tumor volumes were measured at least twice weekly and calculated as 0.5 × length × width × height and shown as mean±SEM.
"Spaghetti plots" of each tumor grown in mice from vehicle group was shown (Supporting figure   R6). As the reviewer mentioned, there are some growth variation in the vehicle groups in Figure   2A and  Fig 6C and 4C (Fig. 6 a-c). Similarly as noted above, in Fig. 6 a-c, β-lap monotherapy could induce the tumor regression and rejection when treated at Day 7 with a relative small tumor mass (around 50 mm 3 ); however, the therapeutic effect was largely impaired when treated at Day 12 with an established larger tumor mass (150-200 mm 3 ).

Comment: The MC38 growth in the b-Lap treated group between
The advanced tumor was also refractory to immunotherapy (Fig. 6c). Next, we tested whether combined local β-lap treatment with anti-PD-L1 treatment could control the established large MC38 tumors (Fig.6c). The results showed that β-Lap could eradicate large established and checkpoint blockade refractory tumors by combination with anti-PD-L1 therapy (Fig.6c). So the difference of therapeutic effect in smaller and advanced tumor should because of the different status of immunosuppressive tumor microenvironment.

Comment: Fig 6 E and F legends are swapped around. E is the survival plot.
Response: We apologize for the mistake of the figure sequence. In the revised manuscript, we have corrected the sequence of Fig. 6e Fig.1a). Briefly, 4T1 tumor bearing BALB/c mice were locally treated with β-lap (15 mg/kg, i.t.) every other day for four times. The growth curve of primary tumor was measured twice a week.
Five weeks after tumor inoculation, tumor bearing mice was were euthanized and the lungs were excised. As shown, β-lap treatment greatly inhibited the primary tumor growth (Supporting Figure R7a). Moreover, compared with the vehicle treatment group, the number of pulmonary metastases was significantly reduced in the β-lap treatment group (Supporting Figure R7b, c).
These data indicate that β-lap is very potent in limiting tumor metastasis.
Supporting Figure R7: Local β-lap treatment inhibits 4T1 tumor growth and limits the pulmonary metastasis. BALB/c mice bearing 4T1 tumor (n=5-6/group) were locally treated with vehicle or β-lap (15 mg/kg, i.t.) every other day for four times. (a) tumor growth was monitored twice a week. b Day 37 after tumor inoculation, lungs were isolated for metastasis detection. Whole lung pictures were shown (b, upper panel), single cell suspensions were prepared from 100mg lung tissues, 1/300 of total cells were plated into 6 well plates. 10µg/ml 6-Thioguanine was added into the culture medium for 4T1 cell selection. 5 days later, crystal violet staining was used to detect the 4T1 clones (b, lower panel). Number of the 4T1 clones was counted (c). Data is shown as mean ± SEM, **P < 0.01, unpaired student t-test was used to analyze the significance of changes.

Reviewers' comments:
Reviewer #1 (Remarks to the Author): The authors have addressed many of the comments, but the critical experiment has not been done.
Comment 2 asked for clarification that β-lapachone caused cell death in NQO1-deficient B16 cells in Figure 1a, but not cell death in Figure 1d. The authors explained that the high-concentration βlapachone had an "off target" effect and revised the manuscript by deleting the 8 μM treatment group in Figure1a. This reviewer thinks that it is necessary to re-exam Figure 1a and d under the same conditions.
The authors obtained data using mainly NQO1-expressing MC38 and TC-1 cells in the mechanism studies. Comment 4 asked for the analysis of β-lapachone-induced innate sensing "in NQO1deficient cells" using overexpression of NQO1 in the mechanism studies in Figure 2-6. Therefore, the experiments are needed to clarify the data.
Comment 5 asked for an analysis of the infiltration of T cells using immunohistochemistry. The authors demonstrated CD8+ T cell infiltration in combination with β-lapachone monotherapy or anti-PD-L1 therapy using immunofluorescence in Figure R2. Rather, in the Figure R2, this reviewer confuses a correlation between inhibition of PD-L1 and infiltration of cytotoxic T cells.
Therefore, this reviewer feels that the authors need to do a lot more experimental work to consolidate their findings.

Response to reviewer:
Reviewer #1: 1. Comment 2 asked for clarification that β-lapachone caused cell death in NQO1-deficient B16 cells in Figure 1a, but not cell death in Figure 1d. The authors explained that the highconcentration β-lapachone had an "off target" effect and revised the manuscript by deleting the 8 μM treatment group in Figure1a. This reviewer thinks that it is necessary to re-exam Figure 1a and d under the same conditions.

Response:
We appreciated the reviewer's suggestion. In the revised manuscript, we re-exam the Figure 1a and Figure 1d under the same conditions. Briefly, 4000 cells (per well) in 96-well plates were exposed to β-lap with or without NQO1 inhibitor DIC for 3 hr and cell survival was assessed 48 hr later with SRB cytotoxicity assay. The results showed that tumor cell lines (MC38, TC-1 and Ag104Ld) that express high level of NQO1 were sensitive to β-lap exposure with a lethal dose around 2-4μM. In contrast, NQO1 deficient cell lines, B16 and Panc02 were resistant to β-lap exposure. Overexpression of NQO1 in B16 cells led to sensitivity to β-lap, and inhibition of NQO1 by dicoumarol spared β-lap lethality (Fig. 1d). Figure 1a and Figure 1d have been replaced with new data.

The authors obtained data using mainly NQO1-expressing MC38 and TC-1 cells in the
mechanism studies. Comment 4 asked for the analysis of β-lapachone-induced innate sensing "in NQO1-deficient cells" using overexpression of NQO1 in the mechanism studies in Figure 2 -6. Therefore, the experiments are needed to clarify the data.

Response:
We appreciated the reviewer's suggestion. In our study, we have proved that the βlapachone could induce innate sensing and antitumor immune response in NQO1 highly expressing MC38 tumor models. In this revised manuscript, we have some additional data to confirm that β-lapachone could trigger innate sensing in NQO1 overexpressed B16 models.
Firstly, we carried out direct killing assay by in vitro assays in NQO-null and NQO1overexpressing B16 cells. B16 (NQO1-null) was resistant to β-lap exposure, and overexpression of NQO1 in B16 cells led to sensitivity to β-lap ( Figure R1a and Figure 1d). Inhibition of NQO1 by dicoumarol spared β-lap lethality in NQO1-overexpressing B16 cells ( Figure R1a and Figure   1d). Next, we examined the antitumor efficacy of β-lap in NQO-null and NQO1-overexpressing B16 cells. NQO1-null B16 tumors didn't respond to β-lap treatment ( Figure R1b and Supplementary Fig. 1k). In sharp contrast, NQO1 overexpressing (clone #1 and #4) tumor bearing mice showed a dramatic tumor suppression after β-lap treatment ( Figure R1b and Supplementary Fig. 1k). To determine whether β-lapachone-induced innate sensing was response for the antitumor efficacy of β-lap in NQO1-overexpressing B16 cells, we analyzed the innate danger protein HMGB1secretion in β-lap treated tumor cells in vitro. Indeed, like in MC38 cells, we observed a dose-dependent secretion of HMGB1 in NQO1-overexpressing B16 cells but not in NQO1-null B16 ( Figure R1c and Figure 5a). To determine whether β-lapachone induced innate sensing and activated type I IFNs in B16NQO1 tumor cells and whether this effect was dependent on HMGB1/TLR4/MyD88 pathway, we co-cultured β-lapachone treated B16 NQO1 cells and BMDCs from WT or MyD88-/-mice. Indeed, BMDCs greatly increased the production of IFNβ protein after co-culture, and MyD88 deficiency completely abolished the IFNβ production ( Figure R1d and Supplementary Fig. 5c). To prove the β-lapachone induced innate sensing and type I IFNs' production were required for the antitumor immune response in vivo, we inoculated NQO1-overexpressing B16 cells separately in WT mice, immune deficient mice and Supplementary Fig. 6c). All these results above were consistent with that from MC38 tumor model. Moreover, we further determined where β-Lap eradicates checkpoint blockade refractory B16 NQO1 tumors by combination with anti-PD-L1 therapy. NQO1 overexpressing B16 tumors failed to respond to anti-PD-L1 Ab alone. By contrast, β-lap monotherapy largely inhibited the growth of the B16-NQO1 tumors. Strikingly, when combined with PD-L1 blockade, β-lap had a markedly synergetic antitumor effect ( Figure R1e and 1j and Supplementary Fig. 8f). Similar synergetic effect was also observed in MC38 models.

Comment 5 asked for an analysis of the infiltration of T cells using immunohistochemistry.
The authors demonstrated CD8+ T cell infiltration in combination with β-lapachone monotherapy or anti-PD-L1 therapy using immunofluorescence in Figure R2. Rather, in the  Fig.9). We have clarified this argument in our revised manuscript. In page 14, lines 27-29 "Encouragingly, PD-1/PD-L1 blockade has shown promising capacity to increase the proliferation and function of tumor infiltrating CD8 + T cells, and enhance the antitumor efficacy in several cancer types." has been added to describe this.