Other Articles published in this series Mouse models for Aicardi–Goutières syndrome provide clues to the molecular pathogenesis of systemic autoimmunity. Clinical and Experimental Immunology 2014, 175: 9–16. Aicardi–Goutières syndrome: a model disease for systemic autoimmunity Clinical and Experimental Immunology 2014, 175: 17–24.
Aicardi–Goutières syndrome (AGS) is a genetically determined disorder, affecting most particularly the brain and the skin, characterized by the inappropriate induction of a type I interferon-mediated immune response. In most, but not all, cases the condition is severe, with a high associated morbidity and mortality. A number of important recent advances have helped to elucidate the biology of the AGS-related proteins, thus providing considerable insight into disease pathology. In this study, we outline the clinical phenotype of AGS, paying particular attention to factors relevant to therapeutic intervention. We then discuss the pathogenesis of AGS from a molecular and cell biology perspective. Finally, we suggest possible treatment strategies in light of these emerging insights.
Lessons from the natural history of Aicardi–Goutières syndrome
‘Classical’ and ‘non-classical’ phenotypes
We have previously published a description of the genotype–phenotype correlation in 121 patients with Aicardi–Goutières syndrome (AGS) . Based on that work, and an ongoing exercise to assimilate clinical and laboratory data from >250 cases (http://www.nimbl.eu/ni/Home), the natural history of AGS is becoming clearer.
In a significant minority of patients with AGS, problems are recognized at birth, i.e. the disease process begins in utero. Over time, severe neurological dysfunction manifests as progressive microcephaly, spasticity, psychomotor retardation and, in approximately 35% of cases, death in early childhood. Typical clinico-radiological features include intracranial calcification, white matter changes and raised numbers of white cells in the cerebrospinal fluid (CSF). To a remarkable degree this form of the disease, seen most consistently in association with mutations in TREX1, RNASEH2A and RNASEH2C, mimics the sequelae of congenital, transplacentally acquired infection (hence the tag: ‘pseudo-TORCH’ syndrome – Toxoplasmosis, Rubella, Cytomegalovirus and Herpes) . More frequently, a later-onset presentation of AGS is seen, occurring in some cases after several months of normal development [3, 4]. This form of the disease, which otherwise shares the similar disease characteristics described above, may be associated with a lesser degree of neurological dysfunction, and is observed most commonly in the context of mutations in RNASEH2B, SAMHD1 and ADAR1.
Although the majority of recognized patients conform to the relatively stereotyped ‘classical’ phenotype just described, there is now an extensive literature reporting a broader spectrum of disease presentation, progression and outcome. These ‘non-classical’ cases highlight a remarkable paradox relating to the diagnosis of AGS; that is, patients with mutations in the AGS-associated genes are observed frequently to demonstrate the absence of one or more, and even all in rare cases, of the original diagnostic criteria as outlined by Aicardi and Goutières in their 1984 paper . Thus, neurological dysfunction is not always severe nor, indeed, necessarily present at all; microcephaly is not invariable; onset is not always in the first year of life; intracranial calcification and white matter changes are not inevitable; and a CSF lymphocytosis is often absent. Importantly, disparity in the clinical phenotype can be seen even within the same family, thus highlighting the role of modifying factors .
With the integration of new sequencing technologies into standard clinical practice, we predict that the spectrum of phenotypes associated with mutations in the AGS-related genes will broaden further. These observations beg the question as to whether such cases should actually be referred to as AGS. The important point is that these phenotypes will probably all relate to a common pathology, and therefore potentially benefit from similar therapeutic strategies.
Is AGS a progressive or non-progressive disease?
At least relating to the classical presentation of AGS, the period of neurological damage appears to be limited to an initial encephalopathic phase, generally lasting for a period of a few months, after which further disease progression is apparently unusual. This important statement is based on the testament of many families with affected children, and the follow-up of a number of children into adulthood. Thus, although we are aware of some patients seeming to experience intermittent ‘decompensations’, we believe that in most cases AGS can be considered to follow a non-regressive course.
Although, in our view, AGS is generally non-progressive, it is of note that chilblains, seen in approximately 40% of cases, frequently persist/recur, particularly so in the winter months [7, 8]; and an inflammatory intracranial large-vessel phenotype, which has so far been recorded only in patients with mutations in SAMHD1, seems to constitute an ongoing disease risk [9-12]. We have also observed frank autoimmune disease, albeit in a minority of cases. Thus, at least some aspects of the AGS phenotype appear to be ongoing (Fig. 1).
How the AGS-associated disease process is triggered, and apparently ‘abates’ neurologically (while the skin disease is frequently recurrent), and whether or not certain patients (e.g. those with SAMHD1-related disease) are, in fact, at risk of disease progression in later life, is still unclear.
Implications of these clinical observations for treatment strategies
Although we would not expect to be able to reverse neurological damage already accrued at the time of initiating treatment, a fact of particular relevance to children affected in utero and displaying signs of disease at birth, the following points deserve to be highlighted:
- The majority of children with AGS demonstrate the onset of disease at a variable time postnatally
- Clinical observation suggests that there is frequently an early period of ‘active regression’, occurring seemingly over several months
- Some disease features can present later (most particularly chilblains and the SAMHD1-related intracranial vascular disease)
- ‘Extreme’ intrafamilial variability can occur
These observations are important because they suggest that:
- Treatment in the early stages of the disease might result in attenuation of the associated inflammation and consequent tissue damage
- It might be possible to discontinue treatments after the subacute encephalopathic period subsides
- In certain cases, e.g. where chilblains are a particular problem and in the context of some of the recognized later-presenting SAMHD1-associated phenotypes, treatment beyond the subacute encephalopathic phase might be necessary/beneficial (even where there is significant neurological damage)
- Determining the efficacy of an intervention has to take account of already recognized phenotypic variability
The pathogenesis of AGS
Up-regulation of interferon is seen in AGS and is probably central to disease pathogenesis
Type I interferon activity was described originally more than 50 years ago as a soluble factor produced by cells treated with inactivated, non-replicating viruses that blocked subsequent infection with live virus. Although the rapid induction and amplification of the type I interferon system is highly adaptive in terms of virus eradication, aberrant stimulation or unregulated control of the system could lead to inappropriate and/or excessive interferon output. Thus, we have recently discussed the concept of type I interferonopathies as a group of inborn errors of metabolism in which an up-regulation of type I interferons is central to disease pathology .
An association of raised levels of CSF and serum interferon-alpha with AGS was first described by Lebon and colleagues in their seminal paper published in 1988 . This remarkable observation led not only to the provision of a highly consistent diagnostic marker of the disease, it also presaged a series of fundamental insights into the pathogenesis of AGS. Various lines of clinical and experimental evidence suggest that type I interferon is toxic to the central nervous system, especially during early neurological development, so that the raised levels of interferon seen in AGS patients probably represent a primary pathogenic factor rather than an epiphenomenon. Of particular note in this regard, Akwa et al. engineered a transgenic mouse with a glial fibrillary acidic protein–interferon-alpha fusion gene resulting in chronic astrocyte-specific over-production of interferon in the central nervous system . These mice developed a progressive inflammatory encephalopathy with neuropathological features closely recapitulating those observed in AGS. Considering these data, although not proven beyond doubt, we predict that limiting the exposure of the infant brain to an AGS-related type I interferon immune response will attenuate the disease-associated brain damage.
The genetic basis of AGS
AGS is a genetically heterogeneous disease resulting from mutations in any one of the genes encoding (i) the 3-prime repair exonuclease TREX1  with preferential activity on single-stranded (ss) DNA; (ii) the three non-allelic components of the RNASEH2 endonuclease complex  acting on ribonucleotides in RNA : DNA hybrids; (iii) the Sam domain and HD domain containing protein (SAMHD1) , which functions as a deoxynucleoside triphosphate triphosphohydrolase; and (iv) adenosine deaminase acting on RNA (ADAR1) , which catalyses the hydrolytic deamination of adenosine to inosine in double-stranded (ds) RNA (Table 1). It is possible that at least one further genetic subtype of AGS is yet to be defined.
|Gene name||Locus||Chromosome||Inheritance||Phenotypes||Mouse phenotype||Protein function|
|TREX1||AGS1||3p21||AR (rare AD cases reported)||AGS, FCL, RVCL, SLE||Inflammatory myocarditis with features of autoimmune activation||3′-5′ exonuclease with preference for ssDNA|
|RNASEH2A||AGS4||19p13||AR||AGS||Not published||Catalytic component of RNase H2 complex|
|Acts on RNA portion of RNA/DNA hybrids and removes ribonuceotides embedded in DNA|
|RNASEH2B||AGS2||13q14||AR||AGS||Embryonic lethal; DNA damage response (not obviously interferon-related)||Non-catalytic component of RNase H2 complex|
|RNASEH2C||AGS3||11q13||AR||AGS||Embryonic lethal||Non-catalytic component of RNase H2 complex|
|SAMHD1||AGS5||20q11||AR||AGS, FCL||Not published||dNTP triphosphohydrolase triphosphatase|
|ADAR1||AGS6||1q21||AR, AD||AGS, DSH||Embryonic lethal E11.5; associated with interferon signature||Hydrolytic deamination of adenosine to inosine in dsRNA|
Although most cases of AGS demonstrate an autosomal recessive pattern of inheritance, rare examples due to de-novo dominant TREX1 mutations have been reported [20-23]. Moreover, the same heterozygous D18N mutation in TREX1 has been seen to cause both (dominant) AGS and familial chilblain lupus (effectively, ‘non-neurological AGS’), thus highlighting the role of unknown, modifying factors (which might be genetic or environmental) and/or stochastic mechanisms.
The cellular pathology of AGS
The proteins defective in AGS are all associated with nucleic acid metabolism. The finding of mutations in TREX1 and the genes encoding the RNASEH2 complex in 2006, in the context of a clinical phenotype mimicking congenital infection, led us to hypothesize that (i) these proteins might be involved in clearing cellular nucleic acid ‘debris’; and (ii) that a failure of such waste removal could result in immune activation, specifically triggering an innate immune response more normally induced by viral nucleic acid  (Fig. 2).
At least with regard to TREX1, cogent evidence has emerged in support of this hypothesis. Thus, Yang et al.  demonstrated that TREX1 deficiency results in the intracellular accumulation of abnormal ssDNA species. This finding was confirmed by Stetson and colleagues , who showed that in Trex1-null mice, ssDNA activation of a Toll-like receptor (TLR)-independent cytosolic pathway involving IRF3, TBK1 and STING results in the induction of a type I interferon response, and a recruitment of the adaptive immune system requiring functional lymphocytes. Further ground-breaking studies in the same mouse model  have defined a process that is initiated in non-haematopoietic cells, perhaps because these cells are the first to accumulate sufficient endogenous DNA substrates to trigger an interferon response, which then signal to haematopoietic cells, leading to T cell-dependent inflammation and a disease-relevant autoantibody response.
The precise mechanisms relating RNASEH2, SAMHD1 and ADAR1 dysfunction to the AGS phenotype remain to be clarified. Of particular note, unlike the other AGS-related proteins, the RNASEH2 complex is not induced by interferon, and the RNaseH2B knock-out mouse does not demonstrate an obvious up-regulation of innate immune signaling . However, clinical and biochemical (see below) overlap observed in human studies across the six disease-associated genotypes leads us to predict that the pathogenesis of all forms of AGS relates to inappropriate stimulation of the innate immune system by nucleic acids.
How to assess treatment efficacy?
Because of already-accrued neurological damage, and also because of recognized intrafamilial variability, it will be difficult to monitor treatment efficacy using only clinical/radiological criteria in the context of early, proof-of-principle clinical trials. Rather, it would be ideal to assess the effects of therapy by assaying a reactive biomarker.
As discussed above, AGS is associated with increased levels of interferon alpha in the CSF and serum. Interferon alpha levels and white cell counts in the CSF of AGS patients have been reported to fall during the first few years of life, perhaps corresponding with the apparent ‘burning-out’ of the encephalopathic period already described . However, due to the obvious difficulties of repeat CSF sampling, very few serial data are available (i.e. systematic interferon alpha profiling beyond infancy has not been undertaken).
Of significance, in currently unpublished data we have observed that >90% of AGS patients, of any genotype, sampled at any age, demonstrate a so-called ‘interferon signature’, i.e. increased expression of multiple type I interferon-stimulated genes (ISGs), in whole blood. Beyond the interesting biological questions that our findings raise, most particularly why we observe a persistent interferon signature when the disease is, apparently, ‘clinically quiescent’ (see earlier), we propose that the level of ISGs measured in blood samples from patients with AGS might be used as a biomarker of disease activity, and potentially of treatment efficacy. Other cytokines and chemokines are also increased in the CSF and serum of AGS subjects and may, similarly, be considered as possible biomarkers for the future assessment of therapeutic effect.
Of note, for some patients/families, chilblains are a major disease-associated problem (e.g. precluding the use of splinting for the prevention of contractures). Because of their visibility, chilblain status could possibly also serve as an indicator of treatment efficacy.
Therapeutic strategies in AGS
It is clear that AGS is a disorder of inappropriate immune activation, demonstrating some characteristics of both autoinflammatory and autoimmune disease. As such, a survey of the literature and patient data, together with information from personal communications, confirms that immune-modulating therapies have been tried empirically, on an ad-hoc basis, in a limited number of AGS patients. In those cases known to us, involving treatments which have included prednisone with azathioprine , intravenous (i.v.) methylprednisolone with i.v. immunoglobulin (IVIG) , methylprednisolone  or IVIG alone , neurological improvement was variable. In reality, judging the efficacy of these interventions is difficult, considering the small numbers involved, the different stages of the disease process at which treatments were started and the different regimens employed, as well as differences in genotype. Such limitations highlight the urgent need to define coherent treatment strategies and monitoring protocols. Below, we outline three approaches to treatment which we think are of immediate interest, although we predict that others will present themselves as our understanding of the pathophysiology of AGS advances.
Anti-interferon alpha antibody therapy
Considering a possible primary role of exposure to type I interferons in AGS pathogenesis, a treatment strategy in which interferon alpha activity is blocked using monoclonal antibodies is worthy of consideration. Clinical trials of such agents, targeted against interferon alpha subtypes and the type I interferon receptor, are already being undertaken in the context of systemic lupus erythematosus , and the results are eagerly awaited in relation to AGS.
Reverse transcriptase inhibition in AGS
What is the source of the nucleic acid inducing the immune disturbance in AGS? Intriguingly, Stetson and colleagues presented data to show that Trex1 can metabolize reverse-transcribed DNA, and that single-stranded DNA derived from endogenous retro-elements accumulates in Trex1-deficient cells . Retro-elements account for close to half of the human genome, and there is evidence to indicate that such elements are more active than recognized previously [34-37]. These observations suggest that mechanisms must exist to limit such activity, the function of which might plausibly involve TREX1, the RNASEH2 complex, SAMHD1 and ADAR1 (Fig. 3).
Considering the above, it is of particular interest that both TREX1 and SAMHD1 have been implicated in the metabolism of nucleic acid derived from exogenous retrovirus. Thus, Lieberman and colleagues have shown that cytoplasmic TREX1 digests non-productive human immunodeficiency virus infection 1 (HIV-1) reverse transcripts in CD4 T cells and macrophages, so that early HIV-1 infection does not trigger a type I interferon response in these cells . Furthermore, the groups of Benkirane , Skowronski  and Keppler  showed that SAMHD1 is a restriction factor for HIV-1 in cells of the myeloid lineage and in CD4+ T cells, and that silencing of SAMHD1 in non-permissive cell lines is associated with a significant accumulation of viral DNA. Although a role for RNASEH2 in retro-element metabolism is so far unproven, we note that the RNASEH2 complex acts on RNA : DNA hybrids, a nucleic acid substrate generated during reverse transcription .
Taken together, the available data suggest that AGS might be treated with reverse transcriptase inhibitors (RTIs: compounds that can potentially disrupt the replication cycle of both exogenous retroviruses and endogenous retro-elements). Indeed, considering this possibility, Stetson et al.  dosed the Trex1-null mouse with the nucleoside analogue RTI azidothymidine (AZT) – but without obvious effect on the lethal phenotype. However, Doitsh et al.  showed, in the context of HIV-1 infection of CD4+ T cells, that AZT inhibits DNA elongation but not early DNA synthesis, indicating that it might be necessary to block reverse transcription at an earlier stage in order to avoid accumulation of immunostimulatory DNA. Taking this insight into account, Beck-Engeser et al.  have rescued the lethal Trex1-null murine phenotype by treatment with a combination of RTIs. On the assumption of no ‘off-target’ mechanism, this truly remarkable experiment indicates that the accumulation of cytosolic DNA in Trex1-null cells can be ameliorated by inhibiting endogenous retro-element cycling. Importantly, we are aware of these results having been recapitulated in an independent laboratory.
RTIs are prescribed worldwide to children and adults (with HIV-1 infection), so that their pharmacodynamic, safety and toxicity profiles are already well characterized. There is no reason to predict that patients with AGS will demonstrate a distinct safety/toxicity profile when treated with these drugs, and so we are actively considering a trial of RTIs in AGS patients. One thing to note here is that any regimen employed will need to incorporate drugs capable of crossing the blood–brain barrier, an issue of no relevance in the Trex1-null mouse which does not demonstrate a neurological phenotype.
Autoantibody production in AGS and disease-related effects
The production of autoantibodies against nucleic acids has been variably documented in AGS. Of note, Trex1-deficient mice  develop organ-targeted autoantibodies against cytosolic cardiac proteins, probably related to the lethal inflammatory myocarditis seen in these animals. Furthermore, a possible role of autoantibodies in AGS pathogenesis is indicated by substantial rescue of the Trex1-null mouse after crossing onto a B cell-deficient background . Notably, these double knock-out mice demonstrate sustained increased levels of interferon, suggesting that interferon alone is not sufficient, on its own, to drive disease.
The implication of lymphocytes and autoantibody production in AGS pathogenesis suggests possible therapeutic strategies, including the use of already licensed agents to deplete B cells. Other compounds of possible interest might include the use of medications, alone or as adjuvants, directed toward the probable presence of autoreactive T cells, such as mycophenolate mofetil. That such agents are established and often already approved for use in children – albeit for other indications – may facilitate clinical trial design and development. However, the side effects of these medicines can be significant, and thus have to be balanced against uncertainty regarding the course of the disease in an individual case.
AGS is a Mendelian disorder of aberrant immune activation. Growing evidence suggests that an accumulation of endogenous nucleic acid species, perhaps derived from retro-elements, provokes a type I interferon response with subsequent recruitment of the adaptive immune system. The disease is associated with significant morbidity and a high rate of mortality. Designing effective therapeutic approaches will be enhanced by an improved understanding of disease pathophysiology. Following proof-of-principle studies in the Trex1-null mouse, treatment strategies of immediate interest include type I interferon blockade, interruption of the generation of the products of reverse transcription and a depletion of B and T cells. Therapies already exist relating to each of these strategies. In the future, inhibition of components of the relevant cytosolic signalling pathways (for example, in the case of TREX1 – cGAS, TBK1, STING and IRF3) might also represent attractive targets.
The difficulties of randomization and controlled studies in rare disorders with small populations are relevant to AGS. It may be useful to consider using an historical cohort as a control population in a treatment trial; to that end, careful attention to natural history is crucial at this time. Additionally, outcome measures to determine the effectiveness of treatments need to be established, and their best use carefully considered. Disease manifestations, e.g. radiological findings and clinical outcomes, are frequently difficult to measure objectively. Thus, the relevance and specificity of biomarkers needs to be established in anticipation of clinical trials. Combinations of outcomes may prove to be the most useful.
Therapy is most likely to be beneficial in the early stages of the disease, making rapid diagnosis of the utmost importance. However, ongoing disease and later-onset phenotypes mean that treatment will also probably have a role in at least some older patients. Unanswered questions as to whether one therapy will be appropriate for disease due to any genotype will become clearer as our understanding of AGS-related protein function improves and other animal models are developed. For example, the possibilities of using treatment with hydroxyurea to deplete the pool of deoxyribonucleotide triphosphates (dNTPs) might be relevant in the context of SAMHD1-related disease, but not other subtypes of AGS. Finally, it will be interesting to determine if treatments developed in the context of AGS are germane to other phenotypes including familial chilblain lupus, retinal vasculopathy with cerebral leucodystrophy and some cases of systemic lupus erythematosus.
We thank sincerely the families and clinicians who have contributed to our collective work. Y.J.C. would like to thank Diana Chase for her expert proof-reading. Y.J.C. acknowledges the Manchester Biomedical Research Centre and the Greater Manchester Comprehensive Local Research Network. The research leading to these results has received funding from the European Union's Seventh Framework Programme (FP7/2007–2013) under grant agreement 241779, and the European Leukodystrophy Association. The NIMBL Consortium comprises David Bonthron, Genetics Section, Leeds Institute of Molecular Medicine (LIMM), St James's University Hospital, Leeds, UK; Antonio Celada, Institute for Research in Biomedicine (IRB) Barcelona, Spain; Yanick Crow, Genetic Medicine, Manchester Academic Health Science Centre, Manchester, UK; Taco Kuijpers, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands; Arn van den Maagdenberg, Departments of Human Genetics and Neurology, Leiden University Medical Centre, Leiden, The Netherlands; Simona Orcesi, Department of Child Neurology and Psychiatry, IRCCS C. Mondino Institute of Neurology Foundation, Pavia, Italy; Dan Stetson, Department of Immunology, University of Washington, Seattle, WA, USA; Adeline Vanderver, Children Research Institute, Washington DC, USA.
All authors report no disclosures.