Background Immune checkpoint blockade (ICB) has revolutionized cancer treatment. However, ICB alone has demonstrated only benefit in a small subset of patients with breast cancer. Recent studies have shown that agents targeting DNA damage response improve the efficacy of ICB and promote cytosolic DNA accumulation. However, recent clinical trials have shown that these agents are associated with hematological toxicities. More effective therapeutic strategies are urgently needed.
Methods Primary triple negative breast cancer tumors were stained for cytosolic single-stranded DNA (ssDNA) using multiplex immunohistochemical staining. To increase cytosolic ssDNA, we genetically silenced TREX1. The role of tumor cytosolic ssDNA in promoting tumor immunogenicity and antitumor immune response was evaluated using murine breast cancer models.
Results We found the tumorous cytosolic ssDNA is associated with tumor-infiltrating lymphocyte in patients with triple negative breast cancer. TREX1 deficiency triggered a STING-independent innate immune response via DDX3X. Cytosolic ssDNA accumulation in tumors due to TREX1 deletion is sufficient to drastically improve the efficacy of ICB. We further identified a cytosolic ssDNA inducer CEP-701, which sensitized breast tumors to ICB without the toxicities associated with inhibiting DNA damage response.
Conclusions This work demonstrated that cytosolic ssDNA accumulation promotes breast cancer immunogenicity and may be a novel therapeutic strategy to improve the efficacy of ICB with minimal toxicities.
- breast neoplasms
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
WHAT IS ALREADY KNOWN ON THIS TOPIC
Checkpoint blockade immunotherapy for breast cancer is developing rapidly. However, only a small subset of patients with breast cancer has robust and durable responses to immune checkpoint blockade (ICB) monotherapy, as breast cancer is typically immunologically quiescent. The modest response of breast cancer to immunotherapy has directed the efforts towards novel combination therapy strategies to sensitize breast tumors to immunotherapy. Despite the significant benefit of the use of combination therapy, the toxicity issues and resistance to combination therapy highlight the urgent need for developing novel therapeutic strategies.
WHAT THIS STUDY ADDS
This study demonstrates that inducing cytosolic single-stranded DNA (ssDNA) accumulation in breast cancer promotes tumor immunogenicity. We found that the tumorous cytosolic ssDNA is associated with tumor-infiltrating lymphocyte (TIL) in patients with triple negative breast cancer. For a more tolerable alternative approach to increase cytosolic ssDNA, we genetically inhibited TREX1 and found that loss of TREX1 activity triggered a STING-independent innate immune response. Inducing cytosolic ssDNA accumulation by TREX1 depletion in tumors is sufficient to drastically improve the efficacy of ICB. We further discovered a novel pharmacological approach to induce cytosolic ssDNA accumulation using CEP-701, which sensitizes breast cancer to immunotherapy and circumvents the toxicities.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Our discoveries uncover the direct role of cytosolic ssDNA in regulating breast cancer immunogenicity. This may inspire researchers to use cytosolic ssDNA as a potential biomarker for predicting the response of breast cancer to immunotherapy. In addition, we provide a novel therapeutic strategy to overcome ICB resistance and minimize toxicities by inducing the accumulation of cytosolic ssDNA in breast cancer. This could expand the benefits of immunotherapy to more patients with cancer.
The recent advances in immunotherapy have offered a new opportunity for the treatment of patients with breast cancer. However, the majority of breast cancers do not respond to immune checkpoint blockade (ICB) therapy alone, as breast cancers are typically immunologically quiescent compared with other tumor types such as melanoma and non-small cell lung cancer.1 2 The lack of response of patients with breast cancer to immunotherapy has directed research toward novel combination immunotherapy strategies that aim to convert the non-responders to responders.
In contrast to cancers such as melanoma and non-small cell lung cancer, breast cancer immunogenicity does not associate with mutation or neoantigen burden.3 4 Instead, we found that immunogenicity of breast and other less mutated cancers was associated with defects in DNA replication stress response that corresponded with accumulation of cytosolic DNA.5 Recent studies have established that mis-localized cytoplasmic DNA can act as a potent mediator that stimulates an innate immune response.6–8 Based on this, we and others have shown that inhibitors targeting DNA damage response to induce accumulation of exceptionally high amounts of nuclear-derived cytosolic single-stranded DNA (ssDNA) in cancer cells can promote the efficacy of ICB.5 9 However, these studies cannot differentiate effects of the cytosolic ssDNA directly versus other signaling alterations induced by these inhibitors. Several ongoing clinical trials are evaluating combinations of ICB with agents targeting DNA damage response, such as ATR inhibitor, DNA-PK inhibitor or WEE1 inhibitor, in breast cancer,10–13 but their associated toxicities and resistance in patients have limited their application. More effective therapies are urgently needed.
Here, we hypothesized that inducing cytosolic ssDNA accumulation continually produced by cancer cells may phenocopy the immunostimulation of DNA damage response inhibition with fewer toxicities, providing a promising therapeutic approach to potentiate the efficacy and expand the patient population to be treated by ICB therapy in breast cancer. Cytosolic DNA is degraded by two primary mechanisms. DNase II preferentially degrades double-stranded DNA (dsDNA) in lysosomes, whereas TREX1 preferentially degrades free cytosolic ssDNA. TREX1, is a 3’−5’ exonuclease that degrades endogenous DNA rapidly before it accumulates, preventing the aberrant activation of the cytoplasmic DNA sensing pathway.14 15 It has been reported that the accumulation of cytosolic ssDNA derived from DNA replication byproducts in TREX1-deficient cells can stimulate immune response.16
In this study, we demonstrated how induction of cytosolic ssDNA accumulation in breast cancer cells promotes tumor immunogenicity. We first show that tumorous cytosolic ssDNA is associated with tumor-infiltrating lymphocyte (TIL) in patients with triple negative breast cancer (TNBC). To induce the accumulation of cytosolic ssDNA, we suppressed TREX1, resulting in activation of STING-independent immunostimulatory pathway via DDX3X. We further demonstrated that TREX1 deletion in tumor cells increased immune cell recruitment in preclinical breast cancer models, which was mirrored in TREX1-deficient patient tumors, and drastically improved the response of breast cancer to ICB therapy. We leveraged the transcriptional rewiring induced by TREX1 loss to discover novel pharmacological approaches to induce cytosolic ssDNA accumulation, and found that CEP-701 increased the tumorous cytosolic ssDNA level and improved the efficacy of ICB in vivo. These findings provide insight into the function of cytosolic ssDNA in regulating tumor immunity and provide a novel therapeutic strategy to overcome ICB resistance by increasing the levels of cytosolic ssDNA and tumor adjuvanticity in breast cancer.
Cytosolic ssDNA levels correlate with tumor-infiltrating lymphocyte levels in patients with TNBC
The immunogenicity of breast tumors has shown to be independent of neoantigen load and tumor mutation burden, but numerous studies have demonstrated that inhibition of the DNA damage response can promote recruitment of immune cells in preclinical models.3 17 18 However, the direct relationship between cytosolic DNA and immune infiltration remains unclear as it has not been studied in patient tumors, and inhibition of the DNA damage response in preclinical models induces numerous effects outside of generation of cytosolic ssDNA. Since TIL levels reflect antitumor immune response and are predictive of responses to immunotherapy, we examined the relationship between cytosolic ssDNA and TILs in tumors of patients with TNBC using quantitative fluorescent multiplex immunohistochemistry (mIHC). Tissue microarrays (TMAs) comprised of tumor samples obtained from 77 patients with TNBC were stained for ssDNA and tumor cells identified by pan-cytokeratin (figure 1A,B). We found that the mean cytosolic ssDNA intensity was significantly increased in TIL-high compared with TIL-low tumors (figure 1C), with TIL-high defined as >20% TILs as determined by H&E staining. The percentage TIL was determined according to the international TIL Working Group Guidelines.19 20 These results indicate that cytosolic ssDNA levels may be associated with enhancing the antitumor immune response in patients with TNBC.
Accumulation of cytosolic ssDNA due to TREX1 depletion triggers STING-independent innate immune response
Although inducing DNA release from the nucleus to cytoplasm by inhibiting DNA damage response can prime tumor for immune therapies,5 21 clinical trials of these inhibitors have demonstrated the hematological toxicities precluding the use of these drugs in combination with ICB.13 22 23 TREX1, a major cytosolic exonuclease that acts preferentially on ssDNA, can prevent the aberrant activation of the cytoplasmic DNA sensing pathway24 and function as a crucial component of the innate immune response.25 Thus, inducing cytosolic DNA accumulation by targeting TREX1 may be a novel strategy for sensitizing patients with cancer to ICB therapy.
We first sought to increase the levels of cytosolic ssDNA in TNBC cells by silencing TREX1 using small interfering RNAs (siRNAs). Immunofluorescent staining of cytosolic ssDNA confirmed that siRNA-induced knockdown of TREX1 in CAL51 and MDA-MB-468 cells was sufficient to induce dramatic accumulation of ssDNA in cytosol (figure 2A, online supplemental figure 1A–C). Direct quantification of cytosolic ssDNA via cellular fractionation further confirmed that TREX1-deficient CAL51 and MDA-MB-468 cells had higher levels of cytosol ssDNA than cells transfected with control siRNA (figure 2B, online supplemental figure 1D).
Accumulation of cytosolic DNA has been shown to stimulate the production of type I interferon (IFN) and consequently promote antitumor immune response.26 27 C-X-C motif chemokine ligand 10 (CXCL10) and chemokine ligand 5 (CCL5), induced by type I IFN activation, play key roles in recruitment of CD4+ and CD8+ T lymphocytes in breast cancer and other cancers.28 29 These chemokines are also known transcriptional targets of IRF3. We found silencing TREX1 resulted in significant increase of CCL5, CXCL10, and IFNβ mRNA expression (figure 2C, online supplemental figure 1E).
To confirm the relationship between loss of TREX1 and immunogenicity in patient tumors, we analyzed patient tumors with genetic TREX1 loss-of-function (LOF) across cancers from The Cancer Genome Atlas (TCGA). Consistent with our in vitro models, we found that TREX1 LOF was associated with significantly increased expression of CXCL10, CCL5, and eight other cytokines that promote antitumor immune responses across all cancer types (figure 2D). Further analysis of immune cell infiltration by RNA sequencing (RNA-seq) deconvolution using gene signature from Bindea et al30 revealed increased immune cell recruitment in TREX1 LOF tumors (figure 2E). Further analysis in patients with TNBC from TCGA using gene set enrichment analysis (GSEA) to identify pathways associated with TREX1 found that TREX1 expression is correlated with negative regulation of type I IFN production and immune response (figure 2F, (online supplemental figure 1F). These results strongly suggest that TREX1 plays an important role in antagonizing immune response.
Immunostimulatory signaling from cytosolic DNA has been primarily shown to be activated by cGAS acting as a cytosolic DNA sensor to produce cyclic GMP-AMP on binding to DNA, which in turns activates STING.26 27 However, despite induction of immunostimulatory gene expression following TREX1 suppression (figure 2C, online supplemental figure 1E), we observed no induction of STING or TBK1 following TREX1 suppression in CAL51 and MDA-MB-468 cells, although there was induction of IRF phosphorylation (figure 2G, online supplemental figure 1G). Consistent with this lack of STING activation, we found no detectable expression of cGAS (figure 2G, online supplemental figure 1G) or alternative cytosolic DNA sensors IFI16 and AIM2 known to activate STING31 32 likewise were not expressed at detectable levels in either cell line (online supplemental figure 1H,I) . Additionally, silencing STING in TREX1-deficient CAL51 cells has no significant effect on the expression of CCL5, CXCL10 or IFN beta (figure 2H,I). TREX1 depletion in STING-deficient cell lines also promoted the production of CCL5, CXCL10, and IFNβ (figure 2J,K). Together, these results indicate that inducing cytosolic ssDNA accumulation in TNBC cells by silencing TREX1 triggers the DNA sensing pathway in a cGAS-STING independent fashion.
TREX1 ssDNA substrates activate cytosolic DNA sensing immune response by DDX3X
Activation of immune response from TREX1 suppression independent of STING raised questions about the nature and the source of the ssDNA and how they are being sensed to activate inflammatory signaling. To investigate the identity of these cytosolic ssDNA, we sequenced cytosolic ssDNA extracted from either control or TREX1 depleted CAL51 cells (figure 3A). Exclusion of genomic DNA was confirmed by PCR for an intron of GAPDH (online supplemental figure 2A). Comparing the abundance of the cytosolic ssDNA recovered from wild-type and TREX1-deficient CAL51 cells, we noticed several features of these cytosolic ssDNA fragments that collectively provided clues to the origin and nature of TREX1 ssDNA substrates. Over half of the recovered ssDNA fragments in TREX1-deficient samples are in the range of 20 to 50 nucleotides (figure 3B). We selected three different ssDNA fragments in this range to further investigate the significance of these DNA fragments. Oligonucleotide 1 (Oligo1) is the ssDNA fragment recovered in both wild-type and TREX1-deficient samples (online supplemental figure 2B). Oligonucleotide 2 and 3 (Oligo2 and Oligo3) are recovered only in TREX1-deficient samples (online supplemental figure 2B). Interestingly, we found that Oligo3 are 100% identical to FRA1L fragile site using a database of Human Chromosomal Fragile Sites.33
We next investigated the role of these ssDNA fragments in the activation of cytosolic DNA sensing immune response. We first examined whether these DNA fragments are present in cytosol. Using fluorescence in situ hybridization (FISH), we detected a robust increase in Oligo2 and Oligo3 staining in the cytoplasm in TREX1-deficient CAL51 and MDA-MB-468 cells (figure 3C,D, (online supplemental figure 2C,D), whereas the staining intensity and distribution of Oligo1 in TREX1-deficient cells were similar to cells transfected with control siRNA (figure 3C,D, online supplemental figure 2C,D). These DNA fragments were susceptible to S1 nuclease but not dsDNase digestion (figure 3E), suggesting that these DNA fragments exist as ssDNA. Importantly, we found that transient transfection of Oligo2 and Oligo3 in TREX1-deficient CAL51 cells significantly promotes the expression of these immunostimulatory genes (figure 3F). Immunofluorescent staining shows that these transfected oligos are distributed mainly in cytoplasm of CAL51 cells (online supplemental figure 2E). Thus, TREX1 ssDNA substrates released from fragile sites to cytoplasm probably may play a significant role in the activation of innate immune signaling.
After identification of these TREX1 ssDNA substrates, we sought to determine which DNA sensor can bind to these DNA fragments and mediate innate immune signaling. We performed ssDNA pull-down assays by incubating biotin-labeled oligos with cytoplasmic extracts of CAL51. All of the three oligos were found to interact with RPA70 (figure 3G). The Oligo2 and Oligo3 were found to specifically interact with DDX3X, but not Ku70 (figure 3G). We also probed for binding of other DNA sensors, which are known to induce a broad range of innate immune responses including RIG-1, DDX41, DHX9, and HMGB1. However, none of these DNA sensors were found to interact with these oligos (figure 3G). To confirm that DDX3X was sensing cytosolic ssDNA to activate immunostimulatory signaling, we depleted DDX3X in TREX1-deficient cells, and found TREX1-deficient cells no longer overexpressed CCL5, CXCL10 or IFN beta (figure 3H, online supplemental figure 2F). The secretion of IFN beta in culture supernatant was measured by a human ELISA kit (online supplemental figure 2G). Together, these results suggest that the increased cytosolic ssDNA fragments due to TREX1 deficiency are sensed by DDX3X to trigger the innate immune response.
TREX1 deficiency improves the response of TNBC to ICB therapy and alters tumor microenvironment
Our next goal was to determine the role of TREX1 in regulating the response of TNBC to ICB therapy. We used CRISPR-mediated genome editing to delete TREX1 from murine TNBC 4T1 cells (figure 4A). In agreement with our results described above, TREX1-deficient 4T1 cells accumulated higher levels of cytosolic ssDNA compared with the 4T1 cells expressing a control guide RNA (gRNA) (online supplemental figure 3A,B). To assess whether chemokines secreted from cancer cells regulate lymphocytes in vitro, cells from mouse spleen were isolated. T cells were activated by anti-CD3/CD28, expanded in interleukin 2 and cocultured with control or TREX1 deficient 4T1 cells. The secretion of CCL5 was assessed by a mouse ELISA kit (online supplemental figure 3C). The T-cell killing assay was conducted using Incucyte 5. The results show that silencing TREX1 in 4T1 cells enhanced the antitumorous activity of T lymphocytes (online supplemental figure 3D,E). We next determine the effect of TREX1 deficiency on antitumor immune response using murine models. TREX1-deficient 4T1 tumors grew markedly slower than control tumors in immunocompetent mice given isotype control antibodies (figure 4B, online supplemental figure 3F), however no growth defect was observed for TREX1-deficient cells in vitro (online supplemental figure 3G). These results imply that accumulation of cytosolic ssDNA may drive innate immune responses, which in turn inhibit tumor growth even without ICB treatment. Additionally, we found that TREX1-deficient tumors were more sensitive to ICB therapy than control tumors treated with ICB inhibitors (figure 4B, online supplemental figure 3F). Strikingly, all mice with TREX1-depleted tumors treated with ICB had complete responses and improved overall survival, compared with only 10–20% of mice with control tumors treated with ICB (figure 4C, online supplemental figure 3H). These results suggested that manipulating cytosolic ssDNA levels by TREX1 depletion in TNBC cells may drive the sensitivity of tumors to ICB therapy.
To identify the mechanisms by which TREX1 deficiency improves the sensitivity of TNBC cells to ICB therapy, we performed RNA-seq with control and TREX1-deficient 4T1 tumors treated with isotype control antibodies or ICB. We found that TREX1 deletion increased an IFNγ/T cell inflamed gene expression signature which has been shown to be increased in tumors responsive to ICB (figure 4D),34–36 including many genes we observed to be associated with TREX1 loss in patient tumors (figure 2D,E). Profiling of wild-type TREX1 and TREX1-deficient tumors treated with isotype control antibodies or ICB also showed that the expression of genes in response to type I IFN stimulation was considerably higher in tumors lacking TREX1 than in control tumors (figure 4E). Collectively, these results established that cytosolic ssDNA may serve as a regulator that drives innate immune activation in a TREX1-dependent manner.
We next sought to understand how TREX1 inactivation altered the tumor microenvironment using mIHC. We found that loss of TREX1 significantly increased global immune cell infiltration in tumors, including lymphocytes, macrophage and dendritic cells (figure 5). TREX1 depletion also led to enhanced ICB-induced CD4+ and CD8+ T-cell infiltration (figure 5A–E, online supplemental figure 4A–D). Additionally, loss of TREX1 promoted macrophage polarization toward an antitumor M1 phenotype (F4/80+, CD11c+, CD206−) (figure 5F–I, online supplemental figure 4E–G) and increased the infiltration of dendritic cells (F4/80−, CD11c+) (figure 5J, online supplemental figure 4H). Taken together, these results demonstrated that TREX1 inactivation improves the response of TNBC tumor to ICB therapy by altering the tumor immune microenvironment, including activation of the cytosolic ssDNA-sensing pathway and increased infiltration of antitumor immune cells.
The cytosolic ssDNA inducer CEP-701 sensitizes TNBC tumors to ICB
Having observed that TREX1 deficiency improved the response of TNBC tumors to ICB and increased infiltration of antitumor immune cells, we hypothesized that pharmacological inducing cytosolic ssDNA accumulation might potentiate the efficacy of ICB therapy with less toxicity than other current approaches such as DNA damage response inhibitors. To this end, we looked for drugs that transcriptionally rewire cells to mirror gene expression changes following TREX1 KO using connectivity map (CMAP) analysis (figure 6A). We next evaluated the efficiency of the top 14 candidates in inducing cytosolic ssDNA by cellular fractionation. Consistent with TREX1 depletion, the treatment with three drugs, including CEP-701, alone significantly increased the accumulation of cytosolic ssDNA (online supplemental figure 5A).
After initial identification of three hits, a dose response assay was performed for each positive compound to filter out false positives. Among the three compounds, only CEP-701 showed effect on inducing cytosolic ssDNA accumulation in a dose-dependent manner in different TNBC cells (figure 6B,C, online supplemental figure 5B,C). Having observed that TREX1 depletion promoted the production of CCL5, CXCL10 and IFNβ, we next examine whether the treatment of cells with these compounds have similar effects. RT-PCR analysis shows that CAL51 cells treated with CEP-701 (0.1 µM) showed the highest potency on increasing the expression of these genes compared with cells treated with quetiapine or G89686 (figure 6D, online supplemental figure 5D). Thus, we chose to focus on CEP-701, which has shown clinical activity in patients with relapsed or refractory acute myeloid leukemia,37 in this study. CEP-701 is an inhibitor of FLT3, JAK2 and TRK. To confirm the increased levels of cytosolic ssDNA we observed were not through FLT3, JAK2 or TRK pathways, we analyzed the expression levels of FTL3, JAK2 and TRK in 4T1 tumors treated with or without ICB. Only JAK2 was highly expressed in 4T1 tumors (online supplemental figure 5E). To examine whether CEP-701 inhibits the activity of JAK2 and promotes cytosolic ssDNA accumulation in vivo, we treated 4T1 tumors with CEP-701 for 4 days. We found that CEP-701 was unable to inhibit the activity of JAK2 (online supplemental figure 5F), but it induced the accumulation of cytosolic ssDNA in tumor cells (online supplemental figure 5G,H).
Due to the observed effect of TREX1 deficiency in regulating ICB response, we next sought to evaluate whether CEP-701 has a similar effect. Following treatment of 4T1 tumors with ICB, CEP-701, or combination thereof, the combination of ICB inhibitors with CEP-701 significantly decreased tumor growth (figure 6E) and improved the survival time of mice (figure 6F) compared with monotherapy. Treatment and control mice had similar body weights suggesting that the treatment was well tolerated (online supplemental figure 5I). Together, these data indicate that pharmacological inducing the accumulation of cytosolic ssDNA using CEP-701 potentiates the efficacy of ICB therapy.
As mentioned above, therapies targeting DNA damage response, such as ATR inhibitors and CHK1 inhibitors, can also induce the accumulation of cytosolic DNA and enhance the response to ICB.5 21 38 However, recent clinical trials indicated hematological toxicities in patients with cancer treated with these drugs may both limit tolerability and potentially inhibit immune system function.13 We next aimed to compare the hematologic side effects of CEP-701 and ATR inhibitor AZD6378. Compared with CEP-701, a higher dose of AZD6378 was needed to stimulate the similar level of CCL5 expression (online supplemental figure 6A). Additionally, we evaluated hematologic toxicity through the complete blood count assay. Blood samples were collected from mice with 4T1 tumors treated with CEP-701 or AZD6378 for 14 days. Neither of the treatments showed significant effect on bodyweight changes (online supplemental figure 6B). However, AZD6378 significantly decreased reticulocyte count, whereas CEP-701 showed a slight increase in reticulocyte count (online supplemental figure 6C). Impaired erythropoiesis in mice treated with AZD6738 was confirmed by an increase in red cell distribution width, which was not altered in mice treated with CEP-701 (online supplemental figure 6D). Together, these results indicate that the anemia and other hematological side effects observed when using DNA damage response inhibitors such as AZD6378 to induce cytosolic DNA may be circumvented by accumulation of cytosolic ssDNA with CEP-701, suggesting CEP-701 may serve as a better agent than ATR inhibitors for ICB combination therapy.
Currently, there is tremendous interest in using immunotherapy to treat breast cancer. However, ICB has demonstrated only modest benefit against breast cancer, as breast tumors are typically immunologically quiescent. We and others have shown that the accumulation of immunostimulatory cytosolic DNA induced by agents targeting DNA damage response, but not tumor mutation burdens, plays an essential role in driving the response of patients with non-hypermutated cancers to ICB.5 9 Despite these encouraging findings, recent clinical trials have shown that these agents can cause severe hematological toxicities, which could lead to severe anemia and neutropenia.13 Therefore, it is highly important to identify novel effective therapeutics, which could trigger the accumulation of immunostimulatory cytosolic DNA and consequently enhance the ICB response.
This study established, for the first time, the role of cytosolic ssDNA accumulation in breast cancer in antitumor immunity. First, we found that cytosolic ssDNA levels correlate with tumor-infiltrating lymphocyte levels in patients with TNBC. In addition, using a preclinical murine breast cancer model, we demonstrated that induction of cytosolic ssDNA accumulation in breast cancer improves the response of TNBC to ICB therapy. It is previously reported that tumors with lower TREX1 expression levels are associated with poor prognosis.39 However, the increased mRNA expression they observed may be due to the loss of protein function. In the present study, we analyzed the correlation of TREX1 LOF with cytokines and with immune cell type specific gene signature. We found that TREX1 LOF may facilitate the antitumor immune response across diverse malignancies.
Mechanistically, we identified the existence of STING-independent cytosolic ssDNA sensing signaling that mediates the innate immune response. To date, many cytosolic DNA sensors have been proposed. Most of these sensors are thought to act upstream of STING, the key adaptor protein for DNA sensing pathways.6 The STING-independent DNA sensing pathways remain largely unknown. Here we reported the unexpected finding that the accumulation of cytosolic ssDNA due to TREX1 deficiency triggers DNA sensing signaling in a STING independent manner. We identified DDX3X as the sensor of these cytosolic DNA. Our findings demonstrated the possibility of harnessing this STING-independent DNA sensing to trigger innate immune response in the tumor microenvironment, which could broaden the toolkit of sophisticated adjuvant immunotherapies.
Another key to understanding the contribution of cytosolic ssDNA to innate immune response is the nature of the cytosolic ssDNA. It is particularly interesting that DNA fragments from chromosomal fragile sites were detected as accumulating cytosolic ssDNA in our studies. We found that transit transfection of these DNA fragments into cancer cells was sufficient to trigger activation of innate immune signaling. The identification of these DNA fragments has profound significance. For instance, the molecular events leading to the formation of these DNA fragments are an interesting area of further investigation. Also, the identification of these fragments would help uncover novel mechanics of cytosolic DNA sensing pathway. Additionally, it will be of great interest to develop DNA vaccines constructed by these DNA fragments for cancer immunotherapy.
Although the role of TREX1 in enhancing antitumor immunity has been reported and several inhibitors targeting TREX1 to sensitize tumors to ICB have been identified,40 41 targeting TREX1 may cause auto-immune diseases,16 42 senescence-associated secretory and inflammation response.15 43 Our study identified CEP-701 as a compound capable of triggering cytosolic ssDNA accumulation and enhancing the response of TNBC tumors to ICB treatment. However, we cannot exclude the possibility that this compound may contribute to antitumor immunity through other mechanisms. More studies are needed to clearly delineate the effects of cytosolic ssDNA inducers on immunotherapy.
Taken together, the study described here is the first to illustrate the essential role of cytosolic ssDNA accumulation in promoting breast cancer immunogenicity. Our findings may inspire researchers to consider the role of cytosolic ssDNA as a potential biomarker for predicting breast cancer response to immunotherapy. In addition, this study opens new horizons for cancer immunotherapy by combining ICB therapy with selected cytosolic ssDNA agonists or cytosolic ssDNA-inducing agents to expand the benefits of ICB to more patients.
Materials and methods
All animal experiments were approved by the MD Anderson Institutional Animal Care and Use Committee (00000467-RN03). Female BALB/c mice were randomly assigned, and tumor measurements were performed by a blinded investigator. For the mIHC of tumor tissues from patients with TNBC, the human TMA used in this study were approved by the institutional review board at the University of Texas MD Anderson Cancer Center (MDACC). The study population consisted of 77 patients with TNBC who had surgical resection at MDACC. The TNBC TMAs were constructed from formalin-fixed paraffin-embedded (FFPE) blocks of archived TNBC specimens. Representative areas of tumor were selected based on the review of H&E-stained slides. For each patient, two 1.0 mm cores from representative areas of the tumor were used for TMA construction.
TREX1 LOF analysis and GSEA
TCGA data were downloaded from the TCGA Pan-Cancer Atlas.44 TREX1 LOF was defined as mutation or deep deletion and wild type defined as tumors without mutation or deep/shallow deletion. As TREX1 LOF was not highly prevalent in all tumor types, we restricted our analysis to lineages with at least five LOF events (CESC, COAD, OV, LUSC, HNSC, BRCA, STAD, LGG, SKCM, BLCA, KIRC). In total, this resulted in 124 TREX LOF tumors and 2422 TREX1 wild type tumors. Changes in gene expression associated with TREX1 LOF were then assessed using a generalized linear mixed effects model controlling for tumor lineage. Immune signatures were calculated using signatures from Bindea et al.30 For GSEA analysis, processed TCGA RNA-seq data were downloaded from the TCGA Pan-Cancer Atlas,44 and patients with TNBC in the Atlas were identified for GSEA.45 Using an absolute value of 0.3 as the cut-off for Pearson correlation coefficients, we identified genes whose expression was correlated with TREX1 expression level before GSEA.
Cells were maintained in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 100 U/mL penicillin and streptomycin, and 2 mM L-glutamine in 5% CO2 at 37°C. Cells were either purchased fromthe American Type Culture Collection (ATCC) or obtained from collaborators.
CAL51 or MDA-MB-468 cells were plated in a 12-well tissue culture plate at a density of 100,000 cells per well. After 12 hours at 37°C, cells were transfected with 100 nM siRNA oligonucleotides using Lipofectamine 3000 (cat. #L3000015; Thermo Fisher Scientific). SiRNAs were purchased from Sigma, and the information about these siRNAs is summarized in online supplemental table 1. At 72 hours after transfection, the efficiency of knockdown was assessed using qRT-PCR or western blot.
Immunofluorescence of cytosolic ssDNA
Cytosolic ssDNA in cells was stained immunofluorescently as described previously.5 Cells were fixed in 5% formalin and permeabilized with 0.5% saponin before blocking with 5% normal horse serum. Primary antibody incubation with a mouse anti-ssDNA antibody (MAB3868, 50 µg/mL in blocking buffer; MilliporeSigma) was performed overnight at 4°C followed by incubation with an Alexa Fluor 488 anti-mouse antibody (Invitrogen) or Alexa Fluor 568 anti-mouse antibody (Invitrogen) for 1 hour at room temperature and DAPI staining. Representative images were captured under a Nikon Eclipse TE2000-E confocal microscope.
Quantification of cytosolic DNA by cell fragmentation
Quantification of cytosolic DNA was performed as described previously.5 Briefly, cells were trypsinized, washed in phosphate-buffered saline (PBS), and resuspended in a pre-cold hypotonic lysis buffer (20 mM HEPES, pH 7.4, 10 mM KCl, 2 mM MgCl2, 1 mM EDTA, 1 mM EGTA, 1 mM DTT, 1× protease inhibitors). Nuclei, mitochondria, and other debris were cleared from the samples via centrifugation at 15,000 g for 15 min at 4°C. Cytosolic ssDNA and dsDNA in the samples were quantified using a Qubit 4 fluorometer with Qubit dsDNA HS and Qubit ssDNA assay kits (cat. #Q32851, cat. #Q10212; Thermo Fisher Scientific) according to the manufacturer’s instructions. Cytosolic DNA concentrations were normalized to cytosolic protein concentrations as determined using a Protein Assay dye reagent (Bio-Rad).
RNA extraction and qRT-PCR
Total RNA was isolated from cultured cells as described previously.46 Isolated total RNA (250 ng) was used for reverse transcription to produce cDNA using a Bio-Rad iScript cDNA synthesis kit. For qRT-PCR, Bio-Rad iQ SYBR Green Supermix was used with a Bio-Rad CFX instrument. Gene expression levels were normalized to 18s. The primer sequences are listed in online supplemental table 1.
Western blot analysis
Cells were lysed in either radio-immunoprecipitation assay (RIPA) buffer or urea buffer and samples were cleared from lysates via centrifugation. The protein concentration was determined using a Bradford quantification kit (Bio-Rad). The antibodies used for western blotting are listed in online supplemental table 1. Nuclear and cytoplasmic extraction was conducted using NE-PER Nuclear and Cytoplasmic Extraction Reagents (cat. #78833; Thermo Fisher Scientific) according to the manufacturer’s instructions.
Generation of stable cell lines
Single gRNA oligos (online supplemental table 1) were annealed and cloned into the lentiCRISPRv2 plasmid (Addgene). The gRNA vectors were co-transfected with the packaging plasmids psPAX2 and pMD2.G into HEK293FT cells using Lipofectamine 3000 (Invitrogen). Target cells were infected with lentivirus and then selected using puromycin. Single cell clones were selected for high-efficiency editing. To generate stable cell lines expressing control EGFP or wild-type TREX1 BT-549 cells (107 cells/mL) were mixed with 2 µg of plasmids and electroporated using an Amaxa Nucleofector II electroporation machine (Lonza). Cells stably expressing the reporters were selected using G418 for 2 weeks followed by sorting EGFP positive cells via flow cytometry for downstream experiments.
Cloning of cytosolic ssDNA
Briefly, live cells were harvested and washed with PBS. Cytoplasmic fractions were extracted in an extraction buffer containing 50 mM HEPES, 150 mM NaCl, 10% Glycerol, 200 µM digitonin, 1 mM DTT, 2 mM EDTA and Complete protease inhibitors. Cleared extracts were treated with 100 µg/mL proteinase K for 30 mins at 45°C. DNA was precipitated by adding the buffer containing 0.02M MgCl2, 0.16M Ammonium acetate and cold 66% v/v ethanol followed by incubation at 4°C overnight and centrifugation for 15 mins at 15,000 rpm in a bench-top centrifuge. The pellet was washed with 70% ethanol, air-dried and resuspended in water. Purified cytosolic DNA was amplified for an intron of human Gapdh gene to exclude the possibility of genomic DNA contamination. Primers used are shown in online supplemental table 1. For cytosolic ssDNA cloning, purified cytosolic DNA was treated with dsDNase and CIP at 37°C for 10 min. Samples were tailed with TdT (NEB) and 92 µM dTTP/8 µM ddCTP in TdT buffer with 0.75 mM CoCl2 for 25 min at 37°C according to a published method.47 The tailed DNA was melted, primed with a poly-dA primer containing a unique 5’sequence tag and copied with Klenow polymerase (NEB) and dNTPs for 15 min at 25°C. The copied DNA fragments were cloned into pCR-Blunt II-TOPO vector according to the manufacturer’s instructions (Invitrogen).
FISH detection of cytosolic DNA fragments
FISH detection of cytosolic DNA fragments was performed as described previously.48 The DNA oligonucleotides with 5’ biotin modification were synthesized by Sigma. The DNA sequences were summarized in online supplemental table 1. Cells were cultured on coverslips and fixed with freshly prepared 5% paraformaldehyde solution for 20 min at room temperature, and then permeabilized with 0.5% Triton X-100 solution for 20 min on ice. The coverslips were then treated with 400 µg/mL RNase A in 2xSSC for 30 min at 37°C. To determine the effect of the S1 nuclease, cells were incubated with S1 nuclease together with RNase A. For dsDNase treatment, cells were incubated with dsDNase in 2xSSC for 5 min at 37°C before treatment with RNase A. The samples were then treated with 0.1N HCl for 15 min, washed in 2xSSC buffer and incubated in 50% formamide in 2xSSC for 30 min. Biotinylated DNA probes (5 ng/mL) were added and samples were denatured at 85°C for 7 min. Hybridization was performed in a humidified chamber at 37°C for 16 hours. After hybridization, samples were washed with 50% formamide in 2xSSC 3 times at 45°C, 1xSSC 2 times at 45°C and once at room temperature. Then, samples were incubated with Alexa Fluor 488 streptavidin (Invitrogen) for 30 min at room temperature and DAPI staining. Representative images were captured under a Nikon Eclipse TE2000-E confocal microscope.
This protocol was adapted from a previously described method.49 Briefly, Streptavidin beads were washed with 1×PBS once, followed by immobilizing with biotinylated DNA oligos (400 pmol) at 4°C for 2 hours. Beads were incubated with 200 µg CAL51 cytoplasmic extraction in 300 µl binding buffer (20 mM Tris, 200 mM NaCl, 6 mM EDTA, 5 mM potassium fluoride, 5 mM b-glycerophosphate, 2 mg/mL aprotinin at pH 7.5) at 4°C for 2 hours. Beads were washed with a binding buffer three times and boiled in a 60 µl sodium dodecyl sulfate buffer. The samples (10 µl) were analyzed by western blot.
For ICB assay, 5×104 4T1 cells (control and TREX1-knockout) were injected directly into the mammary fat pad in female BALB/c mice (age, 6–8 weeks; The Jackson Laboratory). Seven days after injection, the mice were randomized into two groups and intraperitoneally injected three times weekly with either an isotype control antibody (rat IgGa clone 2A3, polyclonal hamster IgG; cat. #BE0087) or ICB (200 µg of an anti-programmed cell death protein 1 antibody and 100 µg of an anti-cytotoxic T-lymphocyte-associated protein 4 antibody). For tissue multispectral staining, mice were given treatment for 1 week before collection of tumors. To evaluate the effect of CEP-701 on enhancing ICB response, mice with 4T1 tumors were randomized into four groups and treated with respective controls, monotherapies, or combination therapy. All treatments were administered for 4 weeks, after which any remaining mice were monitored continually. Tumor volumes were measured twice a week using digital calipers, and overall survival was monitored.
Total RNA from control and TREX1-deficient 4T1 tumors was extracted using a Total RNA Kit I (Omega Bio-tek). RNA from three different tumors per treatment group was extracted according to the kit manufacturer’s instructions. After confirming the RNA quality, RNA-seq was performed by Novogene. Raw FASTQ files were quantified to generate transcripts per million using Kallisto (80) (V.0.44.0). RNA sequencing data has been deposited under accession number GSE244296.
Multiplex immunohistochemical staining
To evaluate the relationship between the levels of cytosolic ssDNA and TIL, TMA slides of patients’ TNBC tumors were stained for ssDNA (MAB3034, MilliporeSigma) and pan-Cytokeratin (MA5-13203, Thermo Fisher). Two TMA tumor tissue cores collected from different locations of an FFPE block were analyzed per patient. Cores were excluded if no analyzable tissue were present. Pan-CK and DAPI were used to identify epithelial cancer cells in tumor cores and nuclear stains, respectively. Cytoplasmic area was identified as the part of each cell excluding the nucleus and the mean intensity of cytosolic ssDNA was quantified using an image analysis software Visiopharm. The percentage of TIL in tumors was determined by H&E stain of the TMA slides. Tumors were grouped as high TIL and low TIL based on the percentage of TIL (20% cut-off). To determine the effect of TREX1 deficiency in altering tumor microenvironment, mouse sections of FFPE blocks were stained using an Opal 7-Color IHC Kit (cat. # NEL811001KT; PerkinElmer) along with primary antibodies against cytokeratin 7 (ab181598: Abcam) and against CD3 (clone D4V8L), CD4 (clone D7D2Z), CD8 (clone D4W2Z), FoxP3 (clone D6O8R), CD11c (clone D1V9Y), F4/80 (clone D2S9R), and CD206 (clone E6T5J; all from Cell Signaling Technology). All samples were deparaffinized in xylene, rehydrated through an ethanol gradient, and fixed with 10% neutral buffered formalin. Microwave antigen retrieval was performed using the PerkinElmer AR6 antigen retrieval buffer (pH 6) for 15 min. Slides were subjected to four or five sequential rounds of staining, each including a protein block with PerkinElmer Antibody Diluent/Block buffer followed by a primary antibody incubation and corresponding secondary horseradish peroxidase-conjugated polymer incubation. Additional microwave antigen retrieval was performed using the AR6 antigen retrieval buffer (pH 6) for 15 min to remove bound antibodies from the tissue section before the next step in the sequence of staining. After all sequential reactions, sections were counterstained with Spectra DAPI solution and mounted.
Slide scanning and analysis for multiplex immunohistochemical staining
mIHC slides were scanned using VECTRA V.3.0 software (PerkinElmer). The whole section on each slide was imaged and analyzed. Tissue detection and segmentation, nuclear detection, and cellular subpopulation identification and quantification were processed using the Visiopharm 2020.09 image analysis software program. For each mouse tumor model, five independent tumors were imaged.
Connectivity map analysis
We used the Clue.io Graphical interface (https://clue.io/) to identify compounds that transcriptionally rewire cells to mirror gene expression changes following TREX1 KO. According to the CMAP guidelines, we used the top 100 upregulated and 120 downregulated differentially expressed genes from RNA-seq data. Compounds identified were filtered based on their normalized connectivity score. The 14 compounds with the highest score and commercially available are present and discussed in this study.
All statistical analyses were performed appropriately using Prism 8 software (GraphPad Software). Data are presented as mean±SEM. Two-tailed Student’s t-test, one-way analysis of variance (ANOVA) and two-way ANOVA were performed to determine significant differences among groups. Survival data were plotted using the Kaplan-Meier method and analyzed using a log-rank test. P value less than 0.05 was considered significant.
Data availability statement
All data relevant to the study are included in the article or uploaded as supplementary information.
Patient consent for publication
We thank the MD Anderson Flow Cytometry and Cellular Imaging Core facility for providing access to the PE Vectra Polaris imaging system.
Contributors JZ, DJM and S-YL were responsible for the overall project concept and design. JZ and DJM performed computational and statistical analysis. JZ and HD performed in vitro and in vivo experiments. LH and JKB provided guidance and advice on multispectral tissue staining and analysis. JZ, DJM and S-YL wrote and edited the manuscript, with critical comments from all authors. S-YL accepts full responsibility for the work and/or the conduct of the study, had access to the data and controlled the decision to publish.
Funding This study was supported by the National Institutes of Health grant R01CA247862 to S-YL. Additional funding was provided by National Institutes of Health grant R00CA240689 to DJM.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.