cGAS puts gas on anti-tumor effect of checkpoint inhibitor

This week PNAS published new article explaining mechanism of action of checkpoint inhibitors, such as anti-PDL1, in tumor immunity. It shows that cytoplasmic DNA sensor cGAS/STING pathway synergize with anti-PDL1 therapy in mouse model of melanoma.

While this study provides noteworthy observation, it is poorly done. For example, when comparing WT and KO mice (cGAS or STING KO mice), the authors did not mention if they used littermate control in these experiments.

In addition, when the authors used exogenous cGAMP (product of cGAS activity detected by STING), they did not use it on cGAS or STING KO mice as controls to verify relationship between cause and effect.   

    

In summary, the connection between DNA recognition system and checkpoint inhibitors is very interesting. Whether it is a simple generic augmentation of T cell priming or specific stimulation of tumor-specific T cells is to be seen.   

David Usharauli

Sensing of RNA:DNA heteroduplexes by cGAS/STING pathway mirrors human autoimmune Aicardi–Goutières syndrome

Autoimmune diseases occur when bodies immune system continues to respond to self-antigens. Ordinarily, such [mis]directed responses are ascribed to adaptive immune system which possesses myriads of randomly generated receptors. So it is quite reasonable to assume that some of those receptor equipped clones (T or B cells) could go haywire and attack their own body's particular antigen or set of antigens. There are few autoimmune diseases known thus far that fall in this category (e.g. in Graves' disease, thyroid stimulating immunoglobulin [over]stimulates the same receptor that normally responds to only thyroid stimulating hormone). This is what is called lack or break of tolerance [mostly due to cross-reactivity]. 

However, much larger category of autoimmune diseases are due to genetic defects of different kind. These conditions harbor mutations in "housekeeping" molecules that are required for proper cellular homeostasis. Such mutations leads to immunopathologies that target broad spectrum of antigens or could even be restricted solely to the activation of innate immune system.    

For example, mutations in enzymes that ordinarily degrades self-DNA could lead to systemic immunopathologies due to activation of DNA sensing molecules cGAS or STING.

One such genetic mutation in ribonuclease (RNase) H2B (a enzyme that degrades cellular RNA:DNA heteroduplexes) leading to pathological cGAS/STING activation was recently described in the EMBO Journal.

This study revealed that mutation in RNase H2B that renders it less active led to up-regulation of interferon‐stimulated gene (ISG) transcripts. 

Follow-up experiments confirmed that interferon "signature" in RNase H2B KO cells was driven by cGAS (deletion via CRISPR/Cas9 genome editing) and STING activation (STING KO).

Exactly what type of self-DNA is detected that leads to cGAS and STING activation in cells with RNase H2B mutation is not yet clear. It could be (a) a cytosolic RNA:DNA heteroduplexes accumulated as a consequence of reduced RNA:DNA degradation, or it could be (b) a cytoplasmic DNA with embedded ribonucleotides that may accumulate due to impaired ribonucleotide excision repair (RER).

With the advance of human genomic testing, it is becoming evident that large portion of autoimmune diseases / immunopathologies may have a clear genetic basis and could be specifically targeted for therapeutic purpose.

David Usharauli

    

DNA vaccine immunogenicity requires STING but not cGAS activity

Intracellular DNA sensor STING has become a major player in mediating the range of immune responses from viruses to tumors to autoimmunity.

From practical point of view, a therapeutic application of STING pathway would involve development of DNA vaccines both for infectious diseases or cancers. In general, DNA vaccine has a long history of testing, but if I am not mistaken there is still no FDA-approved human DNA vaccine [though I think there is approved DNA vaccines for domestic animals].  

So it was interested to read this new paper in Journal of Immunology where the authors found that priming of adaptive immune response by DNA vaccine expressing Flu virus H1 antigen required STING but surprisingly not cGAS pathway.

This is a simple paper. First, the authors found that two i.m injections of H1HA DNA vaccine could prime anti-Flu H1-specific CD8 T cell and IgG response in WT but not in STING KO mice.

However, surprisingly, cGAS KO mice that are deficient of enzyme upstream of STING pathway showed normal response to this DNA vaccine.

Finally, mice deficient for IRF7, but not IRF3, showed similar [to STING KO] reduction of DNA vaccine-induced adaptive immune response (earlier this year journal Science has published research article describing IRF7 deficient patient who showed severe susceptibility to H1N1 viral infection).

In summary, this paper showed that DNA vaccine immunogenicity required STING/IRF7 pathway rather than classical cGAS/STING/IRF3 pathway.

Of note, since the authors have not done virus challenge experiment, the results in this study do not necessarily tell whether this DNA vaccine was effective in providing clinically relevant protection against Flu virus [one of the reasons this study was published in Journal of Immunology and not in a more prestigious journal, in my opinion].

David Usharauli

Tumor endothelial cells, not DCs, initiate anti-tumor IFN-β response to STING agonist

PNAS has published very interesting study in tumor immunity. Some of you might know endogenous DNA recognition complex made of cGAS-STING-IRF3 axis plays an important role in spontaneous anti-tumor activity observed in clinics in some cancer patients.

Immunologists already knew few years back that type I IFN system, rather than TLR system, facilitated natural, spontaneous priming of endogenous anti-tumor T cells. However, they did not know how it worked. Only after discovery of STING-cGAS the scientists made a connection between endogenous [tumor] DNA recognition and type I IFN in anti-tumor activity. Naturally it was assumed that antigen-presenting cells, such as DCs or macrophages, were the major players in STING anti-tumor "initiation" pathway.

Now, Swiss scientists provided evidence that when tumor-bearing mice were injected with STING agonist, cyclic dinucleotide GMP-AMP (cGAMP), it was endothelial cells rather than DCs that were responding to cGAMP by secreting IFN-β.

Initially, the authors showed that in murine B16 melanoma model STING was important for anti-cancer effect of intra-tumorally delivered cGAMP. 1/3 of cGAMP treated mice survived long-term. In addition, cGAMP injection showed synergy with anti-CTLA4/PD1 immunotherapy.

Next, the authors showed that anti-tumor CD8 T cell priming with cGAMP required functional type I IFN system.

In addition, these experiments revealed that anti-tumor effect of injected cGAMP was entirely depended on type I IFN responsiveness.

Surprisingly, analysis of tumor tissue showed that it was tumor vasculature endothelial cells, not DCs, that were IFN-β positive after cGAMP injection.

In vitro experiments confirmed that endothelial cells were specifically responding to cGAMP and tumor DNA.

In summary these results points to a complex interplay between tumor vasculature, tumor DNA and local anti-tumor T cell priming. It must be emphasized here that the authors had conducted short-term tumor challenge experiments here in most part and did not show whether cGAMP injection primed functional anti-tumor memory response in long term, for example by secondary tumor challenge.

David Usharauli

Cytosolic DNA sensor cGAS controls systemic autoimmunity to self-DNA

Type 1 IFN system (IFN-α, IFN-β, etc) is a complex and powerful immuno-regulatory circuit that show both anti-viral (or general pro-inflammatory) or anti-inflammatory activity. Similar to IL-12p70, type I IFNs were shown to prime naive T cells for anti-tumor activity. Very few cytokines have such direct priming effect on naive T cells. 

However, it was not clear how tumors were detected by IFN circuit. Later, cytosolic DNA sensor cGAS and its adaptor molecule STING were discovered and implicated in tumor DNA sensing and T cell priming via IFN circuit.

Unlike viral or bacterial DNA, self-DNA are not ordinarily accessible to cGAS in the cytoplasm. But in disease state when there is an excess of self-DNA, cytoplasmic cGAS could become activated and mistakenly initiate "anti-viral" cascade to self-DNA leading to debilitating immunopathology.

This is a story told by two recent papers, one published in Journal of Immunology and another in PNAS. In these papers, the authors showed that mice knockout for self-DNA degradation enzymes (Trex1^-/- or DNase-II^-/-) succumb to immunopathology, but could be rescued by absence of cGAS (or STING).     

Both papers showed very similar results so I will going to mix them in my analysis. In brief, the authors showed that immunopathology in mice deficient for Trex1 (an exonuclease that degrades cytosolic DNA) could be completely reversed by absence of DNA sensor cGAS.

Mechanistically, Trex1^-/-cGAS^-/- double knockout mice showed reduced level of auto-antibodies, comparable to WT mice (immunoblots against heart Ag are shown).

Similarly, peripheral immunopathology in [DNase-II^-/- mice] that lack another cytosolic DNA degradation enzyme DNase-II could be completely rescued by simultaneous absence of cGAS.

In summary, these two studies clearly support the hypothesis that abnormal sensing of endogenous self-DNA by cGAS is a major priming step leading to immunopathology in diseases such as systemic lupus erythematosus (SLE), Aicardi-Goutieres syndrome, familial chilblain lupus, retinal vasculopathy and cerebral leukodystrophy.

In general, I agree that modulation of cGAS-STING pathway will provide enormous benefits in both tumor therapy and systemic autoimmunity.

David Usharauli

Mitochondrial DNA abnormality activates cytosolic type I IFN system

Mitochondria is a cell's energy hub. It is of a prokaryotic (simple bacterial) origin. It has its own DNA. However, prokaryotic DNA can activate innate immune system. So, what would happen if our mtDNA undergo misfolding stress?

This is exactly what the new Nature paper tried to uncover. This study led by Gerald Shadel at the Yale School of Medicine, has examined the outcome of mitochondrial DNA (mtDNA) packaging abnormality by studying mouse model heterozygous for mitochondrial transcription factor A (TFAM^-/+), a protein responsible for mtDNA higher-order nucleoid organization.

Gene expression profiling revealed that cells in TFAM^-/+ mice showed enrichment in IFN-stimulated pathway genes.

The authors showed that this type I IFN response was significantly abrogated in cGAS, STING, IRF3 and TBK1 deficient backgrounds indicating that in TFAM^-/+ mice aberrant mtDNA released into cytosol activates cGAS-STING-IRF3 pathway leading to type IFN production.

Additional experiments showed that such heightened type I IFN response in TFAM^-/+ mice reduced viral cellular burden.

Finally, the authors showed that viruses from herpesviridae family could induce mtDNA misfolding (by TFAM depletion via viral-encoded UL12 M185 molecule) and mtDNA dependent type I IFN response in wild type mice, mimicking TFAM^-/+ mice, suggesting physiological function of mtDNA stress response.

In summary, these results highlight how mitochondrial DNA stress could activate innate immune system and cause inflammation (in human disorders such as systemic lupus erythematosus, cancer, multiple sclerosis, etc).

This study provides additional support for recently described immunological function of mtDNA and its detection by cGAS-STING-IRF3-type I IFN pathway. It is of note that many cellular abnormalities analysed from immunological point of view are converging on type I IFN response.

David Usharauli

       

Interruption of a programmed cell death leads to cell autonomous IFN-beta production

The second paper discussing the role of death caspases in cell-intrinsic IFN-beta production and heightened anti-viral state came from Richard Flavell's lab at the Yale University School of Medicine.

Initially, the authors made unexpected observation that mice deficient in death caspases in epithelial and hematopoietic cells (caspase 9 fl/fl Tie2Cre+ and caspase 3 fl/fl + caspase 7KO Tie2CRE+ DKO) showed increase viral resistance both in vivo and in vitro. 

They showed that this increased resistance against viral infection was type I IFN dependent since it was abolished on IFN-alpha receptor 1 KO background or when using anti-IFN alpha/beta blocking antibodies. 

In vitro experiments with a chemical inhibition of caspase pathway confirmed that IFN-beta was induced in cells when pro-apoptotic stimulus was coupled with caspase inhibition. Similar observations were made when using caspase 9 KO or caspases 3/7 DKO primary (MEF) cells.

Here too, the authors found that IFN-beta response were abolished on cGAS and STING knockout backgrounds suggesting the role of endogenous DNA recognition.

Finally, the authors found IFN-beta response was abolished when cells were depleted of mitochondrial DNA (mtDNA).

In summary, the authors proposed that in conditions when caspase-dependent programmed cell death is interrupted (in viral infection [or tumors?], for example), infected cell can secrete IFN-beta as a endogenous danger signal and alert neighboring cells for viral presence and confer paracrine anti-viral resistance. 

Still, there are few results from these two papers that require additional explanation: (1) both groups have reported that death caspase deficient mice have high levels of serum IFN-beta but do not show any obvious immunopathology; (2) neither group have examined the impact of caspase deficiency on programmed cell death using physiological stimuli, like FasL, TRAIL, etc; (3) does IFN-beta, spontaneously secreted from caspase-deficient cells, activates tissue resident dendritic cells?

David Usharauli

Reverse engineering the T-independent type 2 (TI-2) antibody responses

A very interesting and thought-provoking research in humoral immunity was published in journal Science this week. It came from Bruce Beutler's lab at the UT Southwestern Medical Center. Professor Beutler, as many may know, is a 2011 recipient of Nobel Prize in Physiology or Medicine for his discovery of endotoxin receptor (TLR4).

Actually, before I review this paper I would like to highlight that since 2005 when Ruslan Medzhitov's lab has published the first paper in Nature that suggested the role of TLR in B cell responses and then later in 2006, David Nemazee's lab has published new study in Science questioning the validity of 2005 paper, there was a controversy in this field. In fact, in 2009, Ruslan Medzhitov's lab has published another study that confirmed that haptenated proteins used in 2006 study could promote antibody response independent of TLR signaling. However the mechanism remained elusive.

It appears that this new study in Science may be a first glimpse in the darkness to reconcile the differences.

I also would like to point out that Professor Beutler's lab is famous for studying ENU-induced random mutations in mice to generate unbiased molecular signature of immune deficiencies.

Humoral immunity is divided into T-dependent and T-independent (TI) antibody response. TI responses, are in turn segregated into two categories: TI-1 and TI-2. This new study focus on TI-2 response.

Initially, using NP-Ficoll immunization, as a TI-2 model antigen, the authors observed that NP-specific IgM response was diminished in single STING-KO, cGAS-KO, MAVS-KO mice but not in several TLR signaling KO mice. 

Mice, double deficient in cGAS/MAVS, had almost complete absence of NP-specific IgM response.

This was strange observation since STING, cGAS and MAVS are known to detect the presence of DNAs or RNAs and NP-Ficoll contained none of them.

Using adoptive B cell transfer from these three KO mice into RAG-KO hosts, the authors showed that this effect of NP-Ficoll was B cell intrinsic (and the number of B1 or MZ B cells were normal in non-responding KOs).

The authors concluded that DNAs and RNAs were generated as a result of B cell stimulation by NP-Ficoll. Indeed, several endogenous retroviruses were up-regulated in NP+ B cells as compared to NP- B cells.

Furthermore, NP-ficoll immunization induced Reverse Transcriptase (RT) activity in NP+ B cells.

Parallel experiments revealed that mice deficient in NF-kB signaling also lacked the ability to respond to NP-Ficoll immunization or express endogenous retroviruses.

The authors observed that B cells from MAVS-KO, but not from cGAS KO mice, showed reduced phosphorylation of NF-kB proteins after anti-IgM stimulation, implying that MAVS played more critical role in sustained activation of NF-kB.

In summary, the authors proposed the new model of TI-2 antibody response. According to this model, when antigens with repetitive epitopes engage specific B cells, this activates initial wave of NF-kB activity leading to expression of endogenous retroviruses as a RNA that are subsequently converted into cDNA by RT activity. These newly generated RNAs and cDNAs are recognized by MAVS and cGAS/STING pathways, respectively. In turn, MAVS activation induces second wave of NF-kB activation to sustain TI-2 antibody response.

As the authors correctly pointed out, at this stage, it is not clear whether expression of endogenous retroviruses in activated B cells indeed play a critical role in TI-2 response. Their presence could be coincidental to this process. It would require B cell devoid of endogenous retroviruses to definitely test their precise role (since both TI-1 and T-dependent antibody response were shown to be STING, MAVS and cGAS independent).

For me this results represent new concept how we should assess the role of endogenous viruses present among our genes. Not everyone carries them, even among laboratory mice strains there is significant differences. Why would nature develop such system where TI-2 response would depend on endogenous viruses? It is just a fascinating idea.  

David Usharauli