The Safety and Efficacy of Systemic Delivery of a New Liver-de-targeted TGFβ Signaling Inhibiting Adenovirus in an Immunocompetent Triple Negative Mouse Mammary Tumor Model

Aberrant TGFβ signaling is linked to metastasis and tumor immune escape of many cancers including metastatic triple negative breast cancer (mTNBC). Previously, we have found that oncolytic adenoviruses expressing a TGFβ signaling inhibitory protein (sTGFβRIIFc) induced immune activation in a mouse TNBC (4T1) immunocompetent subcutaneous model with intratumoral injection. Systemic administration of adenoviruses can be a superior route to treat mTNBC but faces the challenges of increased toxicity and viral clearance. Thus, we created a liver-de-targeted sTGFβRIIFc- and LyP-1 peptide-expressing adenovirus (mHAdLyp.sT) with enhanced breast cancer cell tropism. Its safety and immune response features were profiled in the 4T1 model. Our data showed that the systemic administration of mHAdLyp.sT resulted in reduced hepatic and systemic toxicity. mHAdLyp.sT was also effective in increasing Th1 cytokines and anti-tumor cell populations by cytokine analysis, spleen/tumor qRT-PCR, and flow cytometry. We further tested the therapeutic effects of mHAdLyp.sT alone and in combination with immune checkpoint inhibitors (ICIs). mHAdLyp.sT alone and with all ICI combinations elicited significant inhibition of lung metastasis by histological analysis. When mHAdLyp.sT was combined with both anti-PD-1 and anti-CTLA-4 antibodies, primary 4T1 tumor growth was also significantly inhibited. We are confident in advancing this new treatment option for mTNBC.


INTRODUCTION
Triple-negative breast cancer (TNBC) is a highly heterogeneous disease and often associated with a poor prognosis, particularly for patients diagnosed with late-stage metastatic cancers.Recent advances in immunotherapy with the development of immune checkpoint inhibitors (ICIs), have changed the treatment paradigm of TNBC, with new FDA approvals and many new clinical trials in progress [1][2][3].
Although current TNBC immunotherapy strategies are effective in early stage TNBC patients, only modest clinical responses are observed in patients with metastatic disease.This may be because late-stage tumors are often highly immunosuppressive, thus reestablishing a favorable immune environment; therefore, more effective and long-lasting anti-cancer responses with combination therapies is essential [3].In recent years, many studies have shown that TGFβ signaling plays a central role in tumor immune evasion and resistance to ICIs [4,5].We have previously shown that tumor stromal expression of TGFβ-1 is associated with TNBC and is a poor prognostic marker of overall survival in breast cancer patients, which is consistent with other studies that linked TGFβ signaling to drug resistance and poor survival rate in TNBC patients [6][7][8].Therapeutic strategies to block TGFβ signaling in advanced cancers by inhibitory antibodies or fusion proteins have been the focus of several clinical investigations, and some of them involve patients with metastatic TNBC (mTNBC).[9][10][11].Thus, investigations combining TGFβ pathway inhibitors with ICIs holds promise for the treatment of mTNBC.
We previously reported that TGFβ blockade by direct inoculation of an oncolytic adenovirus expressing a fusion protein with soluble TGFβ receptor II and the human IgG Fc fragment (sTGFβRIIFc) into subcutaneous mouse TNBC (4T1) tumors can inhibit protumorigenic signals and induce immune activation [12].It can also enhance the antitumor responses of ICIs (anti-PD-1 and anti-CTLA-4 antibodies) in this immunocompetent mouse model.Although direct inoculation of adenovirus showed favorable results, the preferred route to deliver adenoviral vectors would be via systemic administration for metastatic cancers [13,14].Key challenges in the use of Ad5-based adenoviruses for systemic administration are increased liver/systemic toxicities, quick viral clearance, and limited tumor tropism.To limit hepatic and systemic toxicities, we previously created a liver de-targeted oncolytic adenovirus (mHAd.sT)expressing sTGFβRIIFc that has Ad5/48 chimeric hexon, which has a reduced capacity to bind with blood coagulation Factor X (FX) when administered intravenously [15,16].Because Ad5-FX complex is the major mechanism of hepatic sequestration, systemic toxicity, and viral clearance in mouse models [17,18], we were able to use this adenovirus at the optimal dose to achieve better inhibitory effects on skeletal metastases of both human breast cancer and prostate cancer cells in immunode cient bone metastasis mouse models [15,16].Recently, to enhance tumor tropism, we further engineered our sTGFβRIIFc expressing adenoviruses with a 9-amino acid-long tumor homing-cell penetrating peptide (LyP-1) into the HI loop of Ad5 adenoviral ber to generate AdLyp.sTfrom Ad.sT, and mHAdLyp.sTfrom mHAd.sT [6].Both AdLyp.sT and mHAdLyp.sTbind to LyP-1 receptor, which has been shown to be expressed on the breast cancer cell surface, tumor macrophages, and tumor lymphatics, but not readily detectable on normal tissues [19,20].
The main goal of this study is to further test hepatic/systemic toxicity, tumor tropism, immunomodulatory factor expression, immune cell response, and the therapeutic e cacy of mHAdLyp.sT in the immunocompetent mouse TNBC (4T1) model.We use Ad.sT, AdLyp.sT, and mHAd.sT as the controls for our toxicity and immune mechanism studies.To evaluate mHAdLyp.sT'seffects on 4T1 tumor growth and metastasis we compare it with ICIs (anti-PD-1 and anti-CTLA-4 antibodies) and assess if it can synergize with ICIs to inhibit tumor growth and metastasis.We report here that intravenous delivery of mHAdLyp.sT has reduced hepatic uptake and hepatic/system toxicity but retains tumor tropism.mHAdLyp.sT is also potent in generating anti-tumor cytokine and immune cell responses, both systemically and in the tumor microenvironment (TME).It can enhance anti-tumor and antimetastasis e cacy of anti-PD-1/anti-CTLA-4 therapy as well.Overall, our studies suggest that mHAdLyp.sT in combination with ICIs is suitable to be developed into clinical trials as a new systemic treatment option for mTNBCs, which is still particularly devastating to many patients.

Animal studies
All animal experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) at the NorthShore University HealthSystem.

Tumor formation, adenoviral treatments, and sample preparation
To establish mouse mammary tumor syngeneic mouse model, we injected 2 × 10 6 4T1 cells per mouse subcutaneously (day 0) into the dorsal right ank of sixty female BALB/c mice (6-8 weeks old).The subcutaneous tumors were usually apparent after day 6.On days 7 and 12 post tumor cell inoculation, mHAdLyp.sT,AdLyp.sT,mHAd.sT,Ad.sT, Ad(E-).null, or the vehicle control (PBS buffer) were individually administered intravenously via tail vein (2.5 × 10 10 VPs or 100µl of buffer per mouse for each injection, ten mice per treatment group) to the tumor bearing mice.Several additional mice that were not injected with any tumor cells and viruses served as the normal control.Mice were monitored carefully every day and some mice with signi cant stress/sick sign and severely ulcerated tumor were euthanized early according to our animal protocol.On day 14, two days (48 hours) after the second viral injection, whole blood (n = four to ve mice per treatment group) was withdrawn via cardiac puncture in anesthetized animals before they were euthanized to prepare mouse serum and blood cells for toxicity biomarker, circulating cytokine, and ow cytometry analysis, respectively.Then, subcutaneous tumors, liver, lung, and spleen tissues were removed.They were either frozen for qRT-PCR analysis or processed immediately for ow cytometry analysis for selected groups.On day 25, all remaining mice were euthanized to obtain the samples described above and processed accordingly.Liver and spleen tissues were used for all assays but blood and tumor samples were used for a subset of assays depending on the nal volumes collected.The sample sizes for each assay are indicated in the relevant gures and gure legends.
Toxicity studies and blood immune marker analysis DNA was extracted from liver samples and viral DNA copy numbers were measured by quantitative PCR (qPCR) using the method previously described [6, 15,16].Serum lactate dehydrogenase (LDH), alanine transaminase (ALT), and aspartate aminotransferase (AST) levels were measured with serum samples using commercially available kits as described in the previous publications [6,15,16].Blood sTGFβRIIFc and TGFβ-1 expression were measured using the lab ELISA method described in previous publications [6,12,15,16], or a TGFβ-1 ELISA kit from R&D Systems (Minneapolis, MN) following manufacturer's instruction, respectively.A subset of sera was also submitted for additional cytokine analysis using a MSD custom U-Plex assay kit from PBL Assay Science (Piscataway, NJ) to quantify the serum levels of IFN-γ, TNF-a, IL-2, IL-4, IL-6, IL-10, IL-12p70, IL-17A, and GM-CSF.

Quantitative real-time RT-PCR analysis of gene expression
Total RNA was isolated from mouse spleen and tumor tissues and cDNA was synthesized using the qScript cDNA SuperMix (VWR International, Inc. Radnor, PA) according to the manufacturer's instructions.mRNA expression pro les of various genes in spleen and tumor tissues were determined by qRT-PCR on the StepOnePlus real-time PCR system (Applied Biosystems/Life Technologies, Foster City, CA).The primers used for each target were described in the Supplementary Table S1 of our previous publication [12], except for Ad5 genome primer (sense: 5' cagcgtagccccgatgtaag 3'; anti-sense: 5' tttttgagcagcaccttgca 3').Mouse GAPDH (sense: 5' cagaggtagttatggcgtagf 3'; anti-sense: 5' gaggctgtgatgggaagt 3') was used as the endogenous gene control and relative expression (RQ-fold changes) of the target genes was calculated using the ΔΔCT method with normal or buffer samples as the calibrators.

Therapeutic analysis
Subcutaneous 4T1 tumors were established in forty-eight female BALB/c mice by the method described above.On day 6 post tumor cell inoculation, tumor dimension was measured (in mm) by using a caliper.
Tumor volumes were calculated by the following formula: (width 2 x length)/2.Then, tumor bearing mice were divided into eight groups, without statistical differences of tumor volume between each group (n = 6 per group).On day 6 and 8 post tumor cell inoculation, mHAdLyp.sTor the vehicle control (PBS buffer) was administered intravenously via tail vein (2.5 × 10 10 VPs or 100µl of buffer per mouse each time).On days 7, 9, 11, and 13, anti-PD-1 and/or anti-CTLA-4 antibodies were administered intraperitoneally (0.2mg per mouse per antibody each time) in the groups as indicated in Fig. 7.The tumor volumes were monitored again on day 9, 13, 16, 20, and 23, together with their body weights.Four additional mice that were not injected with any tumor cells served as the normal control for mouse health conditions.Mice were monitored carefully every day for signi cant stress/sick signs or severely ulcerated tumors according to our animal protocol.On day 25, all remaining mice were euthanized, and the blood, spleen, lung and tumor tissues were collected according to lab procedures [6,12], and then either processed immediately or frozen for later use.Half of each lung was xed and prepared for H&E staining according to our published protocol [6,12].Lung sections were examined by a Nikon Eclipse TE200 Inverted Microscope.Nikon DS-Fi3 microscope camera and NIS-Elements BR 5.41.02 were used for documenting micrographs.Pulmonary metastatic burden was quanti ed by using ImageJ software for each lung section with low magni cation images, but higher magni cation images were used when necessary for identi cation of micrometastases.

Statistical Analyses
All statistical analyses were performed using GraphPad Prism software 9 (version 9.3.1 (471)).One-Way ANOVA (Kruskal-Wallis or ordinary) with Dunn's or Bonferroni's multiple comparisons tests were used for group statistical analyses, and unpaired t tests were used to determine the difference between two categorical groups when required.For outcomes with repeated measurements over time (e.g., tumor volume growth), two-way ANOVA with Tukey's multiple comparisons tests were used.Signi cant difference is shown as: * = p < 0.05, ** = p < 0.01, *** = p < 0.001, **** = p < 0.0001.

RESULTS
We previously reported that systemic administration of mHAdLyp.sT in immunode cient nude mice resulted in reduced uptake in the liver and spleen, reduced hepatotoxicity and systemic toxicity, and attenuated innate immune response [6].In this study we used immunocompetent BALB/c mice bearing 4T1 TNBC tumors to determine viral tropism, toxicity, and immune responses by the systemic administration of mHAdLyp.sT.First of all, liver samples collected 48 hours after adenovirus injection were evaluated for viral genomic DNA copy numbers.Unlike the control vectors lacking hexon modi cation for liver-de-targeted tropism (AdLyp.sT,Ad.sT and Ad(E-).null),mHAdLyp.sTtreatment didn't lead to a signi cant increase of liver viral genomic DNA copy numbers compared to the buffer group (Fig. 1A).mHAdLyp.sTtreatment also did not signi cantly increase viral uptake in liver on day 25, although a signi cant increase of liver viral DNA remained in the Ad.sT and Ad(E-).nullgroups (Fig. 1B).Thus, we concluded that mHAdLyp.sThad reduced liver uptake in immunocompetent mice as well.We further evaluated the tumor viral uptake by qRT-PCR of viral genome and sTGFβRIIFc expression.We observed signi cant viral genome tumor expression for all of our adenoviruses (AdLyp.sT,mHAd.sT and Ad(E-).nullvs buffer: P < 0.05; mHAdLyp.sTand Ad.sT vs buffer: P < 0.01) in day 14 tumor samples (Fig. 1C), although these changes were no longer observed in day 25 tumor samples (Fig. 1D).We also found that mHAdLyp.sT and Ad.sT treatment led to a signi cant increase in sTGFβRIIFc mRNA expression when compared to the Ad(E-).nulltreated group (mHAdLyp.sT:P < 0.01; Ad.sT: P < 0.05) in day 14 tumor samples (Fig. 1E).In addition, a signi cant increase of blood sTGFβRIIFc protein expression was observed in serum samples on day 14 for mHAdLyp.sT,mHAd.sT, and Ad.sT (Fig. 1F, P < 0.05, 0.01, or 0.001 vs Ad(E-).null,respectively).These data suggest that mHAdLyp.sTmaintained tumor tropism in 4T1 mouse TNBC tumors but had reduced liver uptakes when administered systemically.
We used mouse sera obtained 48 hours after adenovirus injection to examine the short-term hepatotoxicity, systemic toxicity, and systemic in ammatory responses in this 4T1 syngeneic mouse model.The quanti cation of serum LDH, ALT, and AST levels suggested that the systemic administration of mHAdLyp.sT and another adenovirus containing Ad5/48 chimeric hexon (mHAd.sT)resulted in no signi cant systemic and hepatic toxicity (Fig. 2A-C).Also, while the replicating adenovirus with Ad5 hexon (Ad.sT) stimulated a signi cant increase of both Th2 cytokines (IL-4, IL-6, and IL-10) (Fig. 2D-F, Ad.sT vs buffer: P < 0.05 or 0.01) and Th1 cytokines (IL-12p70, IL-2, TNF-α and IFN-γ) (Fig. 2G-J, Ad.sT vs buffer: P < 0.01 or 0.001), mHAdLyp.sTonly elicited the signi cant increase of IL-12p70 and IFN-γ Th1 cytokines (Fig. 2G and 2J, mHAdLyp.sTvs buffer: P < 0.05).For the other two adenoviruses we tested (AdLyp.sTand mHAd.sT),AdLyp.sT is similar to Ad.sT but mHAd.sTseems less effective in eliciting Th1 cytokine response than mHAdLyp.sTas only a signi cant change in IL-12p70 was detected (Fig. 2G, mHAd.sTvs buffer: P < 0.05).Severe systemic toxicity and in ammatory responses to adenoviruses upon systemic delivery is a major obstacle for their potential clinical application [13].Thus, mHAdLyp.sT is more likely to be applicable for future clinical trials as it didn't induce detectable hepatotoxicity and systemic toxicity but maintained some critical anti-tumor Th1 cytokine response such as IL-12p70 and IFN-γ.
We also tested serum GM-CSF and TGFβ-1 levels on day 14 to examine changes in these immunosurveillance molecules where different expression levels can either promote anti-tumor immune responses (GM-CSF) or lead to immunosuppression (TGFβ-1) [29,30,10].All replicating adenoviruses expressing sTGFβRIIFc almost equally prompted GM-CSF production while hindering immune inhibitory TGFβ-1 secretion (Fig. 2K-L).It is not surprising to see reduced TGFβ-1 levels in mouse sera by these therapeutic adenoviruses since they all express sTGFβRIIFc that can bind TGFβ-1 and neutralize its downstream signaling events via Type II receptors, but the increase of serum GM-CSF levels are also encouraging because many GM-CSF based treatment strategies are currently used in the clinic or hold promise in clinical trials for several cancer types, including breast cancer.Additionally, the expression levels of these biomarkers in sera obtained on day 25 were also analyzed, but the results did not support persistent changes (data not shown).
In tumor-bearing mice, spleen is the major immunomodulatory organ and tumor is where all actions are propelled.Therefore, we analyzed the expression levels of the biomarkers above on both day 14 and day 25 samples of spleen and tumor by qRT-PCR.Only mHAdLyp.sTinhibited TGFβ-1 expression in spleen on both day 14 and day 25 (Fig. 3A, day 14; Fig. 3C, day 25, both the rst panel, P < 0.05 or 0.01 vs buffer).We also analyzed the expression of other Th2 cytokines besides TGFβ-1 (IL-4 and IL-6) in spleen and tumor samples, but treatment with mHAdLyp.sTdidn't lead to any signi cant changes when compared to the buffer group (data not shown).For Th1 cytokines of interest (IL-12, IL-2, TNF-α and IFN-γ), in spleen, we detected increased levels of IL-2 and IFN-γ expression by mHAdLyp.sTtreatment on day 25 (Fig. 3C, day 25; IL-2, the second panels; IFN-γ, the fourth panels; both P < 0.01 vs buffer).It is interesting that no signi cant changes of them were observed in the mHAdLyp.sTtreatment group in day 14 spleen samples (Fig. 3A, the second to fourth panels), suggesting mHAdLyp.sTmay be more likely to have delayed but persistent systemic immunomodulatory effects in spleen.In the tumor microenvironment, mHAdLyp.sTtreated group had increased expression of IL-12 on both day 14 and day 25, and IFN-γ on day 25 (Fig. 3B, day 14; Fig. 3D, day 25; IL-12, the third panels; IFN-γ, the fourth panels; P < 0.05 or 0.01 vs buffer).No signi cant changes of TNF-α in both spleen and tumor was seen at any of our ending points (data not shown), but the localized stimulation of two critical Th1 cytokines (IL-12 and IFN-γ) in tumors by mHAdLyp.sTsuggests that mHAdLyp.sT is able to help prime antitumor immunity directly in the TME, speci cally with IL-12, which was elevated by mHAdLyp.sT at both early and late stages of tumor resistance.
Next, we pro led the immune cell population changes in blood, spleen, and tumor samples from both day 14 and day 25 samples.On day 14 of blood samples, mHAdLyp.sTtreatment led to a signi cant increase in the percentage of central memory cells (T CM , CD44 + CD62L + ) among CD8 + T lymphocytes (Fig. 4A, the right panel, P < 0.01 vs buffer), even though the CD8 + proportion of total T cells remained similar in blood (Fig. 4A, the left panel).In spleen and tumor, we observed signi cantly elevated percentages of CD8 + T lymphocytes by mHAdLyp.sT(Fig. 4B and 4C, the left panels, P < 0.05 vs buffer).The control adenovirus we used in the ow studies, mHAd.sT,only increased the percentage of CD8 + T cells in spleen (Fig. 4B, the left panel).Importantly, in tumor, mHAdLyp.sTtreatment also increased the percentage of CD8 + central memory cells (T CM , CD44 + CD62L + ) signi cantly on day 14 (Fig. 4C, the right panel, P < 0.05 vs buffer).
Interestingly, in spleen, the percentage of CD8 + effector memory cells (T EM , CD44 + CD62L − ) was raised signi cantly by both adenoviruses (Fig. 4B, the right panel, P < 0.05 vs buffer).In mouse models, both CD8 + T CM and T EM cells are vital components of the anti-tumor response [31].Based on these data mHAdLyp.sTmay depend on different microenvironments of speci c tissues to help CD8 + T cells acquire distinct anti-tumor memory cell characteristics.
The same set of immune cell analyses were conducted with the Day 25 samples of blood, spleen and tumor.We did not detect any meaningful differences systemically at this late tumor resistant stage (data not shown).However, an immune in amed TME seems largely retained in adenovirus treated groups, especially for those with mHAdLyp.sTtreatment.Within tumors, the percentage of CD8 + T lymphocytes and IFN-γ producing CD8 + T cells were both signi cantly increased by adenovirus treatments (Fig. 6A, P < 0.05 vs buffer).Furthermore, only mHAdLyp.sTled to a signi cant increase of m-MDSCs (Ly6C hi Ly6G − CD11b + , Fig. 6B, P < 0.05 vs buffer) and M1 macrophage (CD11C + F4/80 + , Fig. 6C, P < 0.05 vs buffer) in day 25 tumors.Taken together, our immune cell analysis data in the tumor-bearing immunocompetent mouse model supports mHAdLyp.sTas a potent primer towards the anti-tumor phenotype that could synergize with other immunomodulators to achieve better therapeutic results.
In the past, we showed that systemic administration of mHAdLyp.sTinhibited bone metastases in a human TNBC cell line (MDA-MB-231) immunode cient mouse model.Therefore, we applied mHAdLyp.sTalone and together with immune checkpoint inhibitors (ICIs) to test their treatment e cacy in this 4T1 immunocompetent model.Our tumor volume analysis indicated that the combination of mHAdLyp.sT,anti-PD-1 and anti-CTLA-4 antibodies (triple treatment) was the only group that signi cantly inhibited primary tumor progression when compared to the buffer group using a two-way ANOVA analysis at the end of this experiment (Day 23) (Fig. 7A, P < 0.001 vs buffer).This was further supported by tumor weight analysis because the average tumor weight of the triple treatment group was lowest and was signi cantly different from two other groups by t tests (Fig. 7B, P < 0.05 vs mHAdLyp.sTand mHAdLyp.sT+ anti-CTLA-4).Most importantly, all treatment groups, except for anti-PD-1 alone, were almost equally effective in inhibiting lung metastasis by our H&E staining microscopy analysis (Fig. 7C, P < 0.001 or P < 0.0001 vs buffer; Fig. 7D, representative images of H&E-stained lung sections).Although mHAdLyp.sTdidn't alleviate the growth of the primary tumor, no signi cant difference in inhibiting lung metastasis between mHAdLyp.sTalone and the triple treatment when comparing them directly by t test was observed.Because mHAdLyp.sT is much safer to be used systemically, currently we are working on new treatment experiments with increased doses of mHAdLyp.sT to enhance its anti-tumor activities in this model.

DISCUSSION
We report here that the new liver-de-targeted TGFβ signaling inhibiting adenovirus: mHAdLyp.sT is safe to be used in the TNBC tumor-bearing immunocompetent mouse model and retains its tumor tropism while applied systemically.mHAdLyp.sT is also potent in stimulating Th1 cytokine production and priming an immune in amed phenotype, as supported by our cytokine, qRT-PCR and ow cytometry analysis.Thus, mHAdLyp.sTcould restore anti-tumor immune responses in highly immunosuppressive cancers, such as mTNBC, and be used synergistically with other systemic immunotherapy approaches, such as ICIs and CAR T cells, etc.
There are several distinct advantages for using mHAdLyp.sTas a potential new systemic treatment option for metastatic cancers.First of all, compared to several FDA-approved oncolytic virus approaches and most of others in the current development [35,36], mHAdLyp.sTcan be safely delivered systemically because it has reduced hepatotoxicity and systemic toxicity.Also, because mHAdLyp.sTexpresses Ad48 Hexon HVRs (1-7), the most dominant anti-Ad5 antibody epitopes [13], mHAdLyp.sT is expected to circumvent pre-existing neutralizing Ad5 antibodies in human sera.We will carefully screen human serum samples to test if mHAdLyp.sT will be neutralized in the presence of anti-adenoviral antibodies in our future pre-clinical studies.Secondly, mHAdLyp.sT is engineered with two levels of tumor selective replication/targeting abilities.On one side, the viral backbone of our Ad series adenoviruses only allows them to selectively replicate in human cancer cells but can induce cell lysis for all tumors from different species when used with a high dose [21][22][23][24][25][26].On the other side, mHAdLyp.sThas higher binding a nity to tumor and tumor tissues expressing LyP-1 receptors via the LyP-1 peptide sequence inserted into the adenoviral ber [6].LyP-1 receptor expression has been shown to be present in many cancers, including breast, prostate, melanoma, glioblastomas, and pancreas, etc.Therefore, mHAdLyp.sT-induced cell death would release tumor-related antigens, such as tumor-associated antigens (TAAs), pathogen-or damage-associated molecular patterns (PAMPs and DAMPs), etc, and trigger antitumor immunity speci cally [6,37].
Third, systemic delivery of mHAdLyp.sT will produce a large amount of sTGFβRIIFc, the TGFβ decoy, both systemically and in the TME.sTGFβRIIFc will bind with TGFβ to inhibit aberrant TGFβ signaling, relieving TGFβ induced immune suppression.Besides TNBCs, aberrant TGFβ signaling has also been shown to promote tumor growth and metastases of many other cancers, including prostate, kidney, and gastrointestinal cancers [4,38].More importantly, aberrant TGFβ signaling has been identi ed as the key mediator for immune evasion in late stage cancers and their poor responses to cancer immunotherapy [10].Thus, we expect that the combination therapy with mHAdLyp.sT and ICIs would be able to augment the response rate in immunogenic tumors, and make immune suppressed tumors responsive to immunotherapy, even in patients with a highly immunosuppressive phenotype.Our preliminary data in 4T1 TNBC mouse model in this study showed encouraging results, although we did not observe complete remission of the 4T1 tumor, which is notoriously non-immunogeneic and di cult to treat.Since viral toxicity is the major barrier in the systemic delivery of oncolytic adenoviruses, for mHAdLyp.sT,we will be able to apply it with a higher dose due to indications of low toxicity in this study, and hopefully we will observe enhanced therapeutic e cacy in the future.In addition, since both aberrant TGFβ signaling and LyP-1 receptor expression are presented in several cancer types, mHAdLyp.sT-basedimmunotherapy approaches have the potential of targeting several other malignancies, too.
To be noted, TNBCs are a group of highly heterogeneous and fundamentally different diseases with distinct histologic, genomic, and immunologic pro les, which are concentrated under the operational term that stemmed from the fact that they don't have estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER-2) expression [39].But for late stages of TNBCs, such as mTNBC.a shift towards more immunosuppressive setting have often been observed [1,3,36], As a result, the response to ICIs for mTNBCs generally are low (5% if selection for PD-L1 positivity was not used) [3,36].The major focus of many current pre-clinical combination immunotherapy trials is to induce a more immune-in amed phenotype for better outcomes.In this study, a mouse TNBC (4T1) model was used.The 4T1 tumor, although di cult to be treated, is ideal for pre-clinical evaluation of mTNBCs immunotherapies, because it resembles several key genome, transcriptome, and immunome signatures of human mTNBCs [40].We will further explore signature changes of mHAdLyp.sT and its combination with ICIs by using large-scale data, such as RNA-Seq of the whole transcriptome and single cell mass cytometry (CyTOF) of immune cell subsets in blood, spleen, and tumor tissues of this model in the future.These studies will not only further con rm the discoveries we reported here, but also give us more information about other important alterations of functional signi cance, including extracellular matrix genes, tumor vasculature, metabolic pathways, epithelial-mesenchymal transition (EMT), and other prometastases genes, besides in ammatory and immune signature pro les systemically and in the TME.In the current study, we have observed a shift from a Th2 to Th1 effector phenotype, an increase in the frequency of CD8 + T cells, an increase in CD8 + effector memory cells, an increase in CD86 + DCs, an increase in m-MDSCs with a decrease of g-MDSCs, and an increase in cancer cell-killing M1 macrophages.We have not investigated some other important cell types such as Tregs, TCRγδ cells, Th17 cells, natural killer (NK) T cells, and tumor-in ltrating B cells.We realize the enormously complex nature of the in ammatory milieu, and several important studies such as TCR sequencing to determine alterations of T-cell repertoire within the tumor microenvironment should be performed in the future as well [41][42][43].In the end, all of our ndings in the 4T1 models should be veri ed in our future studies with TNBC patient samples.To do so, rst we will screen TNBC patient samples for all important immune signatures to see how it can guide us for possible future clinical trials with mHAdLyp.sT and its combinations.As we move into the commercial development of mHAdLyp.sT in the future, we should be ready to select patients with proper phenotypic characteristics and molecular features to optimize treatment strategies for mTNBC.
In conclusion, our studies described here are critical to bring forward our novel LyP-1 modi ed Ad5/48 chimeric hexon oncolytic virus mHAdLyp.sTtargeting TGFβ, in combination with immune checkpoint inhibitors, for clinical evaluation in TNBC patients in the future.We are con dent that mHAdLyp.sTbased combination therapy has the real potential to produce effective immune responses in mTNBC patients who are generally non-responders or respond poorly to current treatments.

Supplementary Files
This is a list of supplementary les associated with this preprint.Click to download.

CGT230578originaldata.xlsx
administration of mHAdLyp.sT in immunocompetent 4T1 mouse model exhibited reduced liver uptake but retained viral gene and sTGFβRIIFc expression in tumor tissues and/or blood serum.(A) Viral DNA copy numbers in the liver on day 14 of tumor cell inoculation were measured and shown (n= 4 or 5 for each group); (B) Viral DNA copy numbers in the liver on day 25 were measured and shown (n= 3 to 5 for each group); (C) Viral genome expression in the tumor on day 14 was measured by qRT-PCR and shown with RQ-fold changes to the buffer group (n= 4 or 5 for each group); (D) Viral genome expression in the tumor on day 25 were measured and shown (n= 3 to 5 for each group); (E) sTGFβRIIFc expression in the tumor on day 14 was measured by qRT-PCR and shown (n= 4 or 5 for each group); (F) sTGFβRIIFc levels in mouse serum on day 14 were measured by ELISA and shown (n= 3 or 4 for each group).Signi cant differences (compared to the buffer and/or Ad(E-).nullby One-Way ANOVA with Dunn's multiple comparisons tests) are shown as: * = p < 0.05, ** = p < 0.01, *** = p < 0.001, or **** = p < 0.0001.

Figure 7 A
Figure 7