Immune signatures in disease and visions for their future use
This essay was written by Anne O’Garra and was first published in the 2011/12 Mill Hill Essays.
When we become infected with a pathogen, for example a virus, bacteria, fungus or parasite, there are rapid changes in gene activity in the cells of our immune system, at the site of infection. The genes help to produce a large number of molecules, including soluble proteins called cytokines, which contribute to the eradication of the infection. During infection the specialised cells of the innate and the adaptive immune systems (described in detail later in this essay) are exposed to infectious agents in the tissue and to other components of the immune and inflammatory response. Thus these cells, moving through the blood to and from the site of disease, can reflect the responses of the cells in the tissues to infectious organisms. Collectively we refer to global changes in gene activity resulting from infection as the ‘immune signature’ of the infection. Immune signatures are generated by infection with pathogens, or by inflammatory and autoimmune diseases such as colitis and systemic lupus erythematosus (SLE). In the future the analysis of immune signatures should play a key role in supporting the detection and identification of inflammatory and infectious diseases and their monitoring and treatment. Analysing immune signatures can give us a composite picture of how the host is responding during disease, how this response will change with drug therapy, and about the processes contributing to disease.
Immune signatures can be determined by measuring RNA expression, which specifies the proteins that cells make, in the blood from infected individuals. In the future it may be possible to sample other relevant tissues, e.g. sputum in respiratory infections. Changes in gene expression can be monitored by measuring relative increases or decreases in host RNA expression, using a technique called microarray. Essentially a microarray consists of a glass slide known as a ‘chip’, onto which an organised grid (or array) of probes specific for a given RNA can be fixed. The probes may be single strand DNA molecules or, more commonly now, shorter sequences of a key section of the DNA molecule complementary to the RNA molecule to be detected. The RNA is processed and labelled so that it can be visualised when it binds to its complementary DNA probe. The scanned image is processed to transform it into numerical data for each target, that is then used for data analysis. Microarray technology allows tens of thousands of probes to be incorporated into a single array and thus microarrays permit the simultaneous measurement of very many RNA transcripts, enabling a complete assay of the transcriptome (the whole genome) of the sample of interest. This technology thus allows the measurement of the composite immune signature of infectious, inflammatory, or autoimmune diseases.
The usefulness of immune signatures is based on the premise that these signatures are different depending on the type of infectious pathogen, the nature of the host’s immune response, and the nature of the disease. Studies have shown the potential of this approach for a number of autoimmune diseases and for viral and bacterial infectious diseases. One advantage of using immune signatures for the detection of infectious diseases is that they can provide valuable information even when the pathogen is below the level of conventional detection methods, or resides in tissues that are difficult to sample. For example, the diagnosis of pulmonary tuberculosis (TB) relies on proof of the identity in sputum by culture of the causative Mycobacterium tuberculosis (Mtb) organism which infects the lung. This can take weeks and in some cases the Mtb organism is not culturable. More recently a new gene test for the Mtb organism has been introduced but this test still requires that the Mtb be detectable in sputum. However, 30% of patients with pulmonary TB are unable to produce sputum neither spontaneously nor with assistance. In other cases of TB the organism resides in organs that are difficult to sample, including the abdomen or the spine. These conditions make it difficult to distinguish whether the disease is caused by infection with Mtb or by another pathogen. Thus, the use of immune signatures reflective of an immune response to the Mtb organism during TB disease may speed up diagnosis and therapy. Finally, immune signatures may be very useful in cases of infectious disease caused by previously unknown pathogens, for example new emerging viruses. Here, the immune signature may allow the identification of the class of pathogen and thus may be of use in rapidly choosing the most effective therapy. However, if immune signatures are to become an effective diagnostic tool we need to study samples from patients with different diseases, and analyse the immune response in various diseases and before and after effective treatment to assemble databases of immune signatures for each type of disease.
The innate immune response
On infection through many routes the invading microorganism is first taken up by immune cells called macrophages and dendritic cells, which are present in tissues and in blood (Figure 1). The dendritic cell is a key cell in initiating immune responses to infections and was discovered by Ralph Steinman, who received the Nobel Prize in 2011 for his seminal work. Dendritic cells present the microbe’s products to other cells of the immune system to activate them to fight the pathogen. Dendritic cells also make soluble proteins called cytokines, which fur ther activate the cells either to directly kill the pathogen, or to activate other immune cells, which then also attack the pathogen (Figures 1 and 2). Various other cells in addition to dendritic cells, such as macrophages, neutrophils, eosinophils and mast cells, are also activated by many microbial products in both animals and man. This first phase of the immune and inflammatory response is called the innate immune response. (Figure 2). Our immune system recognises potential pathogens by detecting characteristic molecular patterns, through binding to pattern recognition receptors (PRRs). Studies of these receptors have highlighted common recognition strategies among mammals. However, they have also shown differences with respect to the nature of the receptors involved and in the exact molecular patterns recognised. In animals, microbially-derived products are recognised by specific PRRs, such as the Toll-like receptors (TLRs). The importance of TLRs was recognised in 2011 by the award of the Nobel Prize to Bruce Beutler and Jules Hoffman for their early work on TLRs. TLRs were also first demonstrated in mammals by Charlie Janeway and Ruslan Medzhitov. Different pathogens stimulate different TLRs and hence cause distinct gene expression changes in cells of the human’s or animal’s innate immune system.
The adaptive immune response
In animals and man, dendritic cells that have been activated by microbial products migrate to lymphoid organs, e.g. lymph nodes and spleen, in response to soluble mediators called chemokines, which may be induced upon infection (Figure 1). In the lymphoid organs dendritic cells process the microbial molecules and present them in a specific form that leads to the activation of key cells called lymphocytes. This results in the swelling of lymph nodes and the discomfort that people experience when they get infections. Lymphocytes are found in lymphoid tissue and in blood but on activation they migrate, under the influence of other soluble proteins called chemokines, to the infected tissue. Here the dendritic cells produce more proteins, called cytokines, which in turn activate further components of the immune response. Cytokines can directly arrest or kill the pathogen. This specific immune response, which is influenced strongly by the immune mediators induced during the earlier innate immune response, is referred to as the adaptive immune response. It is this adaptive response, consisting of the activation of specific functions in the immune cells called lymphocytes, that retains immunological memory to a specific pathogen, such that on reinfection the organism can be swiftly and specifically eliminated.
This memory response is what makes vaccines work. To achieve successful vaccination, eliminating or preventing infection with minimum undesirable side-effects, such as tissue damage to the host, it is essential to understand the molecular basis by which immune mechanisms either eradicate a pathogen or lead to harmful immune pathology. Hence methods for monitoring the immune molecules induced during vaccination and/or infection are important.
The key cells of the adaptive immune response are the B and T lymphocytes (Figures 1 and 2). B lymphocytes produce proteins called antibodies which directly recognise molecules of the infectious organism. These antibodies can be extremely effective in neutralising the molecules of the pathogen leading to its eradication. Some antibodies can also serve to activate other immune cells such as macrophages that will kill the pathogen, while others can act to shut off the immune response, once the pathogen is eradicated, thus contributing to a controlled immune response and limiting immune damage to the host. Vaccines provide protection against viruses by the induction of long-term antibody responses, i.e. immunological memory. Thus, the ability to detect antibodies in the serum of animals or man that are specific for a panel of known pathogen-derived proteins or antigens can help to define the infectious microorganism. Furthermore, there are a number of classes of antibody reflective of particular kinds of immune responses which can also help to define whether the infection is a first infection or a reinfection. Some of these classes of antibodies, also called immunoglobulins (subdivided into IgM, IgG, IgE, IgA etc.), can also help to reflect the type of infection that may have caused the disease. For example, some antibodies are elicited during infections of the lung or gut (IgA), whereas others are produced in response to parasitic (IgE) but not bacterial or viral infections. The immunoglobulin (IgE) associated with parasitic infections is also associated with allergic manifestations such as asthma, but other immune indicators will differ such that they can be used to distinguish between these diseases.
T lymphocytes can be subdivided functionally according to the presence on their surface of one of two proteins, called CD4 and CD8 proteins (Figure 2). T cells with CD4, also called T helper (Th) cells, produce a number of different combinations of soluble proteins called cytokines, otherwise referred to as interleukins (Figure 2). The particular combinations determine the type of immune response induced and thus the type of pathogen that can be eliminated, as well as the kind of host damage that may be caused. Initially two main subsets of Th cells were discovered (termed Th1 and Th2 cells) but more were discovered later on. The different subsets produce different cytokines specialized to eradicate different pathogens (Figure 2).
Cytokines – key proteins in the immune response
The Th1 subset of CD4+ T helper (Th) cells produces a distinct set of cytokines, including one called interferon gamma (IFNg), which activate macrophages to kill intracellular pathogens such as bacteria, parasites and viruses. For example, a Th1 response is protective against mycobacterial infections, such as MTb, the causative pathogen of tuberculosis (TB). A Th1 response is often accompanied by another response which can eliminate cells infected by either viruses or bacteria. This is called the CD8 T killer cell response. In addition to producing a similar profile of soluble chemokines and cytokines to Th1 cells, CD8 T killer cells also generate a number of molecules that can kill pathogens. The presence of the CD8 T killer cell, and the Th1 cellular response, can serve as an effective immune signature of some viral infections. However, if Th1 cytokines are dysregulated they can cause damage to the host such as in multiple sclerosis and colitis.
The Th2 subset of CD4 T helper cells produce a different set of cytokines. Some of these are growth factors for B lymphocytes, and also result in antibody-producing B cells. Th2 cells also produce hallmark cytokines such as IL-4, and are activators of mast cells and smooth muscle cells. Others of these cytokines (IL-13, IL-5) activate cells that in turn secrete further distinct immune mediators typical of aTh2 immune response. The cytokines produced during aTh2 response play an important role in the eradication of helminth worm parasites (Figure 2), but they can also be over-produced and dysregulated during allergic and atopic manifestations such as asthma, thereby contributing to disease.
More recently another Th cell subset, Th17, was described which produces a distinct cytokine, IL-17, which helps to combat extracellular bacteria, such as Klebsiella pneumonia, and fungi such as Candida albicans. Th17 cells play a major role in the activation of innate cells called neutrophils and their migration to infected tissue where they can assist in the killing of invading pathogens. Whilst essential for controlling fungal infections in humans, Th17 responses can go wrong and also like Th1 cells can contribute to autoimmune and inflammatory pathologies such as multiple sclerosis and rheumatoid arthritis.
The cytokine called TNF is produced by a number of cells of the innate and adaptive immune responses, including dendritic cells, macrophages and T cells. It is critical in controlling Mtb infection in both mouse and man (Figure 2). However, TNF is also a key cytokine causing disease in rheumatoid arthritis and Crohn’s disease. Treatment of patients with either of these diseases using anti-TNF antibodies decreases inflammation and symptoms in a large number of cases. However, treatment of those individuals who had been previously infected/exposed to Mtb (latent asymptomatic individuals – explained in detail later) may cause reactivation and lead to active TB disease. Thus, TNF production can contribute to inflammatory pathologies such as rheumatoid arthritis and Crohn’s disease, but is also produced in response to Mtb and other bacterial infections, where it is protective. It is thought that the pattern of molecules induced by the bacterial pathogens during infection, but not in rheumatoid arthritis and Crohn’s disease, could be used to distinguish each type of disease.
Type I interferons (IFNs), which were first discovered by Alick Isaacs at NIMR, are produced by dendritic cells and many other immune and epithelial cells early in the immune response, and are fundamental in the protective reaction to a number of viruses. Indeed IFNs are actually used to treat some chronic viral infections such as hepatitis C. Conversely, dysregulated production of Type I IFNs, and Type I IFN-inducible molecules, has been associated with the autoimmune syndrome systemic lupus erythematosus (SLE), and is diagnostic of disease in SLE.
Accompanying most immune responses that are caused by infection is the simultaneous induction of both regulatory T cells and molecules that act to limit the immune response and prevent damage to the host. One important immunosuppressive cytokine, called IL-10, is produced by many cells of the immune response as a feedback regulator to stop host damage. In its absence mice develop the disease colitis since they over respond to harmless bacteria in their gut. IL-10 dampens immune responses to pathogens, and thus minimises immune damage. An absence of IL-10 in experimental models of infection such as Toxoplasma, Malaria and Helicobacter, will result in an over-exuberant immune response and disease, ultimately death. On the other hand, during infections such as Mtb, over-production of IL-10 will contribute to chronic infection. Therefore balanced IL-10 production during immune responses to infections is required to resolve the infection with minimum damage to the host. Additional regulation is also provided at many levels, including the action of specialised regulatory T cells (Treg). These cells are thought to perform a specialised role in controlling both the innate and the adaptive immune system to prevent inflammatory diseases. Treg cells are characterised by a particular protein which is key for their development and function. Mutations in this protein in humans results in severe endocrine, inflammatory and autoimmune diseases, due to an uncontrolled immune response. Thus, mediators such as IL-10 and other factors produced by Treg cells regulate the immune response to avoid immune mediated damage to the host, however their overproduction may result in under-representation of the inflammatory cytokines described above, and contribute to chronic infection. The expression of unique genes by Treg cells and the over-production of IL-10 may contribute to the composite immune response and immune signature for such chronic infections.
Thus during a protective immune response against an infectious organism, cytokines are produced by CD4+ helper, Th1, Th2, Th17 and CD8 lymphocytes and by other immune cells including those of the innate immune response. This is a dynamic process involving several steps that help to eliminate the infection. All of these immune responses are set in place upon infection of a host, resulting in the simultaneous activation of multiple genes involved in the diverse defence mechanisms. A number of cytokines produced by these different cells act to inhibit the opposite cell type, thereby limiting the immune response and avoiding host damage. This results in such a high complexity that to understand the composite immune response, particularly in disease situations, requires the ability to simultaneously analyse many thousands of genes, and changes in their activity levels, during an immune response. This can be accomplished with the use of microarray technology. Measurement of the global immune response by microarray analysis of blood or tissues can thus provide important information as to what type of immune response may be leading to protection against an infectious agent or disease.
Immune signatures for the detection and identification of infectious diseases
Blood acts as a reservoir of immune cells that migrate to and from different tissues of the body. White blood cells, called leukocytes, contain immune cells which have been recently exposed to either infectious organisms, allergens, tumours, transplants or autoimmune reactions. They can thus be used to establish immune signatures of disease. Immune signatures of infection may additionally be detectable in leukocytes from other fluids such as saliva, sputum, tears or urine but the number of cells in these fluids may be too few for effective analysis. Blood is easily accessible and available in sufficient amounts for study of blood immune signatures by microarray analysis.
The composite immune response can be measured by microarray analysis of the blood of patients with infectious, autoimmune or inflammatory diseases, and compared with that in blood from healthy control individuals. This will help to establish immune signatures that are characteristic of different infections, and could in turn be used for detecting, identifying and monitoring infectious disease. Such immune signatures could be based on a combination of gene activity patterns in circulating white blood cells, which include innate cells and lymphocytes, as well as by the levels of key soluble immune mediators in the serum, (and possibly in the sputum and urine) such as antibodies, cytokines and chemokines (Figures 1 and 2). The signatures may be informative with respect to what is causing the disease or conversely the factors that are successfully protecting against infection.
Conventional approaches for the detection and identification of infectious disease are likely to fail in cases where the organism is hidden in inaccessible tissues, has already been eliminated, or is a novel microorganism for which no test exists. In such circumstances, transcriptional immune signatures can be a valuable tool. Moreover, measurement of immune signatures may also establish the type of response required to clear an infection, whether the response is sub-optimal and chronic infection may set in, or whether undesirable immune damage or pathology is occurring either in infection or in autoimmune disease. Even where the pathogen can readily be identified, determining the immune signature of an infected individual will be very valuable for following the nature of the immune response itself and hence informing decisions on suitable therapy. Furthermore, immune signatures may be able to distinguish whether the disease results from infection or from non-infectious causes, such as allergy or autoimmunity. Immune signatures will undoubtedly change over time after infection with a pathogen or following vaccination. This may be of use in choosing suitable therapeutic strategies.
It will be important to record base level immune signatures in individuals of different ethnicities and residing in different geographical locations, since they may differ with respect to their own genetic composition and also by their exposure to environmental non-pathogenic organisms. Furthermore, immune signatures generated in response to a particular pathogen may be altered by a co-infection, and thus it will be essential to collect immune signature data from patients carrying both single and multiple infections. It is likely that it will be ultimately possible to diagnose coinfections by immune signature analysis.
Gene activity profiles reveal immune signatures for different diseases
The introduction of microarray technology to simultaneously measure the activity of all genes in a genome has proved particularly valuable in the diagnosis and prognosis of different types of cancer and leukaemias. The technology has also influenced decision-making about chemotherapy in patients with breast cancer. Furthermore, researchers at the Baylor Institute of Immunology Research (BIIR) in Dallas, USA have performed microarray analysis of blood from patients with the autoimmune disease SLE, and have shown a unique signature of Type I IFNs andType I IFN-inducible genes. This has led to a better understanding of the mechanisms of disease onset as well as improved diagnosis and treatment monitoring. More recent studies by BIIR have shown that the serum of patients with a particular form of arthritis turns on some innate immune response genes, including a cytokine (IL-1), in blood cells from healthy donors. Based on these data, seven out of nine of the arthritis patients who did not respond to other therapies showed complete remission when treated with a drug that blocks the cytokine IL-1, with a partial response obtained in the other two patients. Thus, the use of microarray technology can predict immune signatures responsible for immune pathologies and help in the design of efficacious therapies.
Emerging data suggest that each class of infectious pathogen elicits a different host response manifested by distinct gene expression patterns in blood. Researchers at the BIIR have shown distinct patterns of gene expression in blood leukocytes from patients with acute infections caused by four common human pathogens, including Influenza A virus, and different types of bacteria such as Staphylococcus aureus, Streptococcus pneumoniae, and Escherichia coli.
Recently my laboratory, together with clinical collaborators in hospitals in London and South Africa and collaborators at BIIR, have demonstrated a blood transcriptional signature in patients with active tuberculosis (TB) that was absent in the blood of healthy control individuals and the majority of individuals with asymptomatic latent TB. Tuberculosis is still a major cause of morbidity and mortality worldwide, leading to 9.27 million new cases and 1.7 million deaths in 2007, and its incidence has more than doubled in London over the last 20 years. Control of this global epidemic has been impaired by variability of the BCG vaccine, emergence of drug resistant forms of Mtb and the lack of optimal and rapid diagnosis or ways to monitor whether drug treatment is working. The immune response to Mtb is complex and incompletely characterised, hindering development of new diagnostics, treatments and vaccines. Although it is known that T cells and cytokines, such as TNF and IFNγ, are important for immune control of Mtb infection, there remains an incomplete understanding of the host factors determining protection or pathogenesis in TB. Interestingly, although there are still 1.7 million deaths currently reported annually in patients with TB, a third of the world’s population has been estimated to have been infected with Mtb but remain well, with no symptoms of the disease. These individuals are defined as having latent TB and only about one-tenth of them will develop active TB disease during their lifetime (Figure 3).
Diagnosis of latent TB is based solely on evidence of immune sensitisation, classically by the skin reaction to Mtb antigens, or by a test based on Mtb-specific induction of the cytokine IFNγ in blood cells. However, there is currently no test to differentiate latent from active disease, nor to clearly identify those patients who may progress to active TB disease. Identification of those latent individuals most at risk of developing active disease would allow preventative therapy to be better targeted. This is important since drug treatment is lengthy and potentially toxic.
Our study based on whole genome array analysis of the immune signature in the blood of TB patients has provided new insights into the immune damage of human TB (Figure 4 and 5). We have identified a whole blood 393-gene signature for active TB.The extent of the signature completely correlates with the extent of disease detected in the lungs of individuals, as shown by X-ray. Four individuals characterised by existing methods as having active TB showed a blood transcriptional signature similar to healthy controls and these individuals were shown to be healthy by X-ray. Conversely, although the signature was absent in the majority of individuals with latent TB, 10-20% of the patients in the latent TB group had a transcriptional signature similar to that in active TB patients (Figure 4). This molecular signature suggests that these latent individuals may be harbouring subclinical active TB. Whether these latent patients have subclinical disease, or are at a high risk of developing active disease, requires prospective evaluation to determine whether this signature could act as a predictor of progression to active disease. If so it would greatly assist in directing their treatment. The 393-gene signature was dominated by a relative increase of IFN- induced genes, both IFN-γ and Type 1 IFN. Although IFN-γ has been shown to be protective against Mtb infection, Type I IFNs have detrimental effects on the immune response to bacteria. Since the immune signature was significantly changed following successful drug treatment (Figure 5), this suggests that Type I IFN and Type I IFN-inducible genes may be contributing to active TB disease. This immune signature of TB was shown to be distinct from that in patients suffering with other diseases. It will be useful to compare the TB signature with that from patients who have inflammatory and other lung diseases that can present clinically similarly to TB in order for these findings to be useful in the clinic as a potential diagnosis. To be useable in the clinic, however, the signature would need to be reduced to a small gene number (around 30 – 80 genes) and converted into an easy test that can be used in the clinic.
The measurement of immune signatures composed of gene expression profiles in blood leukocytes, together with the analysis of specific immune mediators in serum, offers an attractive method for the rapid detection and identification of an infectious pathogen. These immune signatures, once identified, will allow us to identify the class of pathogen, and possibly to identify the specific microorganism. Thus, determinations of immune signatures would allow rapid administration of appropriate therapies within hours of admission of a patient to a clinic, and the ability to determine whether the therapy is effective. The genetic background of the patient will affect the immune signature generated in response to any given pathogen, thus it would be valuable to simultaneously obtain genetic information about the patient.
It remains to be seen whether immune signatures can reflect the specific pathogen causing infection, or whether they can just indicate a class of pathogen. Regardless, this global approach will provide useful, previously unrecognised information about each disease and may still be useful in diagnosis and prognosis of infectious diseases where detection of the microorganism itself is difficult. Once a set of genes is chosen for study, changes in expression of smaller numbers of key genes may also be monitored by more rapid and inexpensive quantitative techniques.
For the use of immune signatures to become a routine technique that can be applied globally in the field by laypersons, their analysis will require the development of sophisticated bioinformatics and statistical methods that can handle large quantities of data, comparing them to known immune signatures in databases. Robust automatic and semi-automatic techniques will need to be developed to allow accurate diagnosis with minimal human intervention. This will require acquisition of samples of blood from patients infected with every known pathogen at different stages after infection and/or cure and their analysis by microarray.
A major research undertaking is envisaged over the next few years to determine immune signatures reflective of a large variety of infections and autoimmune diseases before, during and after infection or in response to immunotherapy or drug treatment. Data on immune signatures generated by vaccination will also become available, and may predict if the vaccine is going to work or not. Databases containing such immune signatures will enable diagnosis of the type of infection for example by viruses, bacteria, parasites and will be able to distinguish different stages and types of immune responses to pathogens, e.g. ongoing infection, successful clearance, chronic infection, immune damage to the individual and severity of the disease. This information will be of use in determining how the patient should be treated. Importantly, immune signatures will also be of use in detecting the effects of therapy in terms of efficacy and may be able to predict side-effects upon treatment of infectious as well as autoimmune and inflammatory diseases.