Viral infections play a major role in human and animal health. Exemplarily, chronic viral infections caused by human immune deficiency virus (HIV) are still an increasing burden on human health worldwide (Cohen et al., 201 1 ).

Despite improvements in antiretroviral therapy (ART) for HIV infection significantly reducing viral burden and replication but also morbidity and mortality, it has been clearly established that immune dysfunction and chronic inflammation persist in HIV patients (Deeks et al., 2013). Moreover, in absence of ART therapy, HIV infection still progresses to AIDS in the large majority of infected individuals (Cohen et al., 201 1 ). While research into immunodeficiency caused by HIV was of central interest for a long time, more recently it has been appreciated that even under effective ART chronic inflammation might be as important as immunodeficiency promoting HIV persistence by numerous mechanisms such as causing virus production, infecting new target cells, or preventing viral clearance mechanisms (Deeks et al., 2013).

Notably, in chronic viral infections such as chronic HIV infections, the T cells increasingly loose their anti-viral effectiveness in the course of time.

Therefore, restoring effectiveness of the immune system is of crucial importance for a successful treatment of a viral infection, in particular in a longer-lasting viral infection.

In murine models, it was found that the loss of anti-viral effectiveness of the immunologic response may be induced by PD-1 expression (Barber et al., 2006) as well as by multiple other inhibitory receptors (CTLA-4, T cell Ig domain and mucin domain 3 (TIM-3), LAG3, CD244, CD160) or immuno-regulatory cytokines like interleukin 10 (IL-10) and transforming growth factor beta 1 (TGF-beta1 ). In human patients suffering from HIV, viral persistence was shown to be required for PD-1 expression (Streeck et al., 2008) and its expression correlated with viral load and inversely with CD4+ T cell numbers (D'Souza et al., 2007). So far, however, no effective therapy for restoring effectiveness of the immune system is known.

Therefore, there is still an unmet need for improved means to treat a patient suffering from a viral infection.

Surprisingly, it was found that tumor necrosis factor alpha (TNF-alpha) antagonists can be effectively used for treating a viral infection in a patient.

In a first aspect, the present invention relates to a tumor necrosis factor alpha (TNF-alpha) antagonist for use in a method of treating a viral infection in a patient.

As used herein, the terms "tumor necrosis factor alpha", "tumor necrosis factor", "cachexin" and "cachectin" may be understood interchangeably. TNF-alpha is an adipokine involved in systemic inflammation and is a member of a group of cytokines that stimulate the acute phase reaction. It is produced chiefly by activated macrophages (M1 ), although it can be produced by many other cell types such as CD4+ lymphocytes, NK cells and neurons. The primary role of TNF- alpha is in the regulation of immune cells and is capable to induce fever, apoptotic cell death, cachexia, inflammation and to inhibit tumorigenesis and viral replication. Human TNF is produced as a 233 amino acid protein (NCBI Reference Sequence: NP_000585) and arranged in stable homotrimers as a primarily 212- amino acid-long type II transmembrane protein. From this membrane-integrated form the soluble homotrimeric cytokine (sTNF) is released via proteolytic cleavage by the metalloprotease TNF alpha converting enzyme. The soluble 51 kDa trimeric sTNF tends to dissociate at concentrations below the nanomolar range, thereby losing its bioactivity. The secreted form of human TNFa takes on a triangular pyramid shape, and weighs around 17-kD. Both the secreted and the membrane bound forms are biologically active, but, both forms do have overlapping and distinct biology activities.

A TNF-alpha antagonist or inhibitor as used herein may be any compound or composition comprising one or more compounds that antagonizes or inhibits TNF- alpha biological activity in the body, particularly its activity in the immune systems. A TNF-alpha antagonist or inhibitor is a substance or composition that binds to TNF-alpha and decreases its activity, e.g. suppresses the cellular response to TNF-alpha. The binding of an inhibitor hinders TNF-alpha from acting in the signal cascade and reduces its downstream signal relative to a situation in which the antagonist is absent. Inhibitor binding is either reversible or irreversible. Irreversible inhibitors usually react with the target and change it chemically. These inhibitors modify key amino acid residues needed for activity. In contrast, reversible inhibitors bind non-covalently and different types of inhibition are produced depending on whether these inhibitors bind the target.

A number of drugs are known as TNF-alpha antagonists or inhibitors. An inhibitor used as pharmaceutical is often judged by its specificity (its lack of binding to other proteins) and its potency (its dissociation constant, which indicates the concentration needed to inhibit the enzyme). A high specificity and potency are desirable in order to minimize side effects and toxicity.

The inhibitor can be a full or partial inhibitor. A full inhibitor is capable of completely blocking the activity (100 % inhibition) at a suitable concentration, whereas a partial inhibitor may inhibit the enzyme's activity to a certain extend (e.g. 60 % inhibition), but not to 100 %. However, inhibition in the context of the present invention is based on a specific interaction between the inhibitor and TNF- alpha and not unspecific mechanisms (e.g. denaturation).

Exemplarily, the TNF-alpha antagonist may bind to TNF-alpha in the extracellular space (e.g., in the blood, the lymph etc.), may bind and block the TNF-alpha receptor (TNFR) located at cellular plasma membrane, may decrease the expression and/or exocytosis of TNF-alpha, and/or may inhibit the intracellular signal transduction triggered by TNF-alpha binding to its receptor. As used herein, a TNF-alpha antagonist binding to TNF-alpha may preferably bind to TNF-alpha with a dissociation constant (Kd) of not more than 100 μΜ, more preferably not more than 10 μΜ, even more preferably not more than 1 μΜ, in particular not more than 100 nM or even not more than 50 nM. Treatment or treating is the attempted remediation of a health problem, usually following a diagnosis. A treatment deals with an existing medical problem, and may lead to its cure, but often ameliorates a problem only for as long as the treatment is continued, especially in chronic diseases. Cures are a subset of treatments that reverse illnesses completely or end medical problems permanently. A treatment or cure is applied after a medical problem has already started.

As used in the context of the present invention, the term "patient" may be understood in the broadest sense as any subject or individual bearing a viral infection, irrespective whether clinical symptoms occur or do not occur. The patient may be any animal, including humans. Preferably, the patient is a mammal (e.g., a human, a mouse, a rat, a cow, a pig, a dog, a cat, a horse, a donkey, a goat, etc.), most preferably a human or a mouse, in particular a human.

A viral infection as used herein may be an infection by any virus, particularly a disease causing virus. The viral infection may be an acute viral infection or a chronic viral infection. In the present context, a viral infection is the invasion of a host organism's bodily tissues by a disease-causing virus, its multiplication, and the reaction of host tissues to these organisms and optionally their toxins. Infectious diseases comprise clinically evident illness (i.e., characteristic medical signs and/or symptoms of disease) resulting from the infection, presence and growth of pathogenic biological agents in an individual host organism. Host can fight infections using their immune system. Mammalian hosts react to infections with an innate response, often involving inflammation, followed by an adaptive response. Examples of clinically important virus families and species include - without limitation - Adenovirus, Herpes simplex, type 1 , Herpes simplex, type 2, Varicella-zoster virus, Epstein-barr virus, Human cytomegalovirus, Human herpesvirus, type 8, Human papillomavirus, BK virus, JC virus, Smallpox, Hepatitis B virus, Human bocavirus, Parvovirus B19, Human astrovirus, Norwalk virus, coxsackievirus, hepatitis A virus, poliovirus, rhinovirus, Severe acute respiratory syndrome virus, Hepatitis C virus, yellow fever virus, dengue virus, West Nile virus, Rubella virus, Hepatitis E virus, Human immunodeficiency virus, Influenza virus, Guanarito virus, Junin virus, Lassa virus, Machupo virus, Sabia virus, Crimean-Congo hemorrhagic fever virus, Ebola virus, Marburg virus, Measles virus, Mumps virus, Parainfluenza virus, Respiratory syncytial virus, Human metapneumovirus, Hendra virus, Nipah virus, Rabies virus, Hepatitis D, Rotavirus, Orbivirus, Coltivirus, and Banna virus. In a preferred embodiment, the viral infection is a chronic viral infection.

A chronic infection is an infection that is persistent or otherwise long-lasting in its effects. The opposite of chronic is acute. In the present invention, a chronic infection also includes an infectious disease with a recurrent course; i.e. a recurrent disease relapsing repeatedly, with periods of remission in between.

Accordingly, as used herein, the term "chronic viral infection" may be understood in the broadest sense as any viral infection that lasts for at least a week, at least a month, at least three months, at least six month, at least a year or even several years. In humans, a chronic viral infection is usually understood as lasting for at least one month or preferably at least three months. This understanding may also be applied to the present invention. The chronic viral infection may or may not be associated with a phenotype, may be latent or may cause symptoms. Preferably, the chronic viral infection is associated with an essentially permanent activation of the immune system.

In a more preferred embodiment, the chronic viral infection is characterized by chronic inflammation.

As used herein, the term "chronic inflammation" may be understood in the broadest sense as any inflammatory effect that lasts for at least a week, at least a month, at least three months, at least six month, at least a year or even several years. In humans, a chronic inflammation is usually understood as lasting for at least one month or preferably at least three months. This understanding may also be applied to the present invention. An inflammation in the context of the present invention may preferably be associated with an increase in the NF-κΒ activity, C- reactive protein (CRP) level, interferon-gamma (IFN-gamma) level, interleukin 1 (IL-1 ) level and/or interleukin 8 (IL-8) level. Preferably, an inflammation is associated with at least two of the aforementioned factors (i.e., increase of NF-KB activity and CRP level, of NF-κΒ activity and IFN-gamma level, of NF-κΒ activity and IL-1 level, of NF-κΒ activity and IL-8 level, of CRP level activity and IFN- gamma level, of CRP level and IL-1 level, of CRP level and IL-8 level, of IFN- gamma level and IL-1 level, of IFN-gamma and IL-8 level, or IL-1 level and IL-8 level). More preferably, an inflammation is associated with at least three of the aforementioned factors. The person skilled in the art will know more factors typically associated with a viral infection. As the viral infection in the context of the present invention is preferably a chronic viral infection associated with chronic inflammation, in a preferred embodiment, the viral infection is characterized by an exhausted T cell phenotype, preferably exhausted CD4+ T cells.

T cell exhaustion is well-known in the art. Typically, T cell exhaustion is associated with immunodeficiency as the T cells loose effectiveness. The transition from acute to chronic viral infection is typically accompanied by hierarchical loss of antiviral T cell effector functions that has been termed T cell exhaustion (Wherry, 201 1 ) leading to a loss of anti-viral effector function of the T cells. Furthermore, exhausted CD4+ T cells may be considered to be characterized by functional defects including loss of interleukin-2 (IL-2) secretion and decreased proliferation capacity (Brooks et al., 2005). As used throughout the present invention, the terms "loss of and "diminished" in the context of exhausted T cells may be understood as a decrease by at least 10 % compared to the control, reference or healthy state, preferably by at least 20 % compared to the control, reference or healthy state, more preferably by at least 30 % compared to the control, reference or healthy state, even more preferably by at least 50 % compared to the control, reference or healthy state, in particular by at least 60 % or even more compared to the control, reference or healthy state, wherein the healthy state reflects a corresponding non-exhausted functional T cell.

As used throughout the present invention, the term "increased" in the context of exhausted T cells may be understood as an increase by at least 10 % compared to the control, reference or healthy state, preferably by at least 20 % compared to the control, reference or healthy state, more preferably by at least 40 % compared to the control, reference or healthy state, even more preferably by at least 60 % compared to the control, reference or healthy state, in particular by at least 80 % or even more compared to the control, reference or healthy state, wherein the healthy state reflects a corresponding non-exhausted functional T cell.

In the view of the above and the literature recited herein, the person skilled in the art will therefore know how to characterize an exhausted T cell.

The exhausted T cells may be a CD4+ T cells and/or CD8+ T cells.

In a particularly preferred embodiment, the T cells are exhausted CD4+ T cells T cells with diminished expression of the Klrdl gene and/or the Ly6c1 gene. Therefore, the T cells may be exhausted CD4+ T cells with diminished expression of the Klrdl gene. Additionally or alternatively, the T cells may be exhausted CD4+ T cells with diminished expression of the Ly6c1 gene. Additionally or alternatively, the T cells may be exhausted CD4+ T cells with diminished expression of the Klrdl gene and the Ly6c1 gene.

Additionally or alternatively, such exhausted CD4+ T cell phenotype may be associated with an increase in the level of a molecule selected from the group consisting of

(i) programmed death factor (PD-1 ),

(ii) cytotoxic T-lymphocyte antigen 4 (CTLA-4),

(iii) prostaglandin E2 (PGE ₂ ),

(iv) transforming growth factor beta (TGF-beta),

(v) interleukin 10 (IL-10), and

(vi) a combination of two or more of (i)-(v).

Therefore, exemplarily, the T cells may be exhausted CD4+ T cells associated with an increase in the level of (i) PD-1 and CTLA-4, (ii) PD-1 and PGE ₂ , (iii) PD-1 and TGF-beta, (iv) PD-1 and IL-10, (v) CTLA-4 and PGE ₂ , (vi) CTLA-4 and TGF- beta, (vii) CTLA-4 and IL-10, (viii) PGE ₂ and TGF-beta, (ix) PGE ₂ and IL-10, or (x) TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with an increase in the level of (xi) PD-1 , CTLA-4 and PGE ₂ , (xii) CTLA-4, PGE ₂ and TGF-beta, (xiv) PGE ₂ , TGF-beta and IL-10, (xv) PD-1 , PGE ₂ and TGF-beta, or (xvi) PD-1 , TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with an increase in the level of (xvii) PD-1 , CTLA-4, PGE ₂ and TGF-beta, (xviii) PD-1 , CTLA-4, PGE ₂ and IL-10, (xix) PD-1 , CTLA-4, TGF-beta and IL-10, (xx) PD-1 , PGE ₂ , TGF-beta and IL-10, or (xxi) CTLA-4, PGE ₂ , TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with an increase in the level of (xxii) PD-1 , CTLA-4, PGE ₂ , TGF-beta and IL-10.

Exemplarily, the T cells may be exhausted CD4+ T cells associated with a diminished expression of the Klrdl gene and an increase in the level of (i) PD-1 and CTLA-4, (ii) PD-1 and PGE ₂ , (iii) PD-1 and TGF-beta, (iv) PD-1 and IL-10, (v) CTLA-4 and PGE ₂ , (vi) CTLA-4 and TGF-beta, (vii) CTLA-4 and IL-10, (viii) PGE ₂ and TGF-beta, (ix) PGE ₂ and IL-10, or (x) TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Klrdl gene and an increase in the level of (xi) PD-1 , CTLA-4 and PGE ₂ , (xii) CTLA-4, PGE ₂ and TGF-beta, (xiv) PGE ₂ , TGF-beta and IL-10, (xv) PD-1 , PGE ₂ and TGF-beta, or (xvi) PD-1 , TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Klrdl gene and an increase in the level of (xvii) PD-1 , CTLA-4, PGE ₂ and TGF-beta, (xviii) PD-1 , CTLA-4, PGE ₂ and IL-10, (xix) PD-1 , CTLA-4, TGF-beta and IL-10, (xx) PD-1 , PGE ₂ , TGF-beta and IL-10, or (xxi) CTLA-4, PGE ₂ , TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Klrdl gene and an increase in the level of (xxii) PD-1 , CTLA-4, PGE ₂ , TGF-beta and IL-10.

Exemplarily, the T cells may be exhausted CD4+ T cells associated with a diminished expression of the Ly6c1 gene and an increase in the level of (i) PD-1 and CTLA-4, (ii) PD-1 and PGE ₂ , (iii) PD-1 and TGF-beta, (iv) PD-1 and IL-10, (v) CTLA-4 and PGE ₂ , (vi) CTLA-4 and TGF-beta, (vii) CTLA-4 and IL-10, (viii) PGE ₂ and TGF-beta, (ix) PGE ₂ and IL-10, or (x) TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Ly6c1 gene and an increase in the level of (xi) PD-1 , CTLA-4 and PGE ₂ , (xii) CTLA-4, PGE ₂ and TGF-beta, (xiv) PGE ₂ , TGF-beta and IL-10,

(xv) PD-1 , PGE ₂ and TGF-beta, or (xvi) PD-1 , TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Ly6c1 gene and an increase in the level of (xvii) PD-1 , CTLA-4, PGE ₂ and TGF-beta, (xviii) PD-1 , CTLA-4, PGE ₂ and IL-10, (xix) PD-1 , CTLA-4, TGF-beta and IL-10, (xx) PD-1 , PGE ₂ , TGF-beta and IL-10, or (xxi) CTLA-4, PGE ₂ , TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Ly6c1 gene and an increase in the level of (xxii) PD-1 , CTLA-4, PGE ₂ , TGF-beta and IL-10.

Therefore, exemplarily, the T cells may be exhausted CD4+ T cells associated with a diminished expression of the Klrdl gene and the Ly6c1 gene and an increase in the level of (i) PD-1 and CTLA-4, (ii) PD-1 and PGE ₂ , (iii) PD-1 and TGF-beta, (iv) PD-1 and IL-10, (v) CTLA-4 and PGE ₂ , (vi) CTLA-4 and TGF-beta, (vii) CTLA-4 and IL-10, (viii) PGE ₂ and TGF-beta, (ix) PGE ₂ and IL-10, or (x) TGF- beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Klrdl gene and the Ly6c1 gene and an increase in the level of (xi) PD-1 , CTLA-4 and PGE ₂ , (xii) CTLA-4, PGE ₂ and TGF-beta, (xiv) PGE ₂ , TGF-beta and IL-10, (xv) PD-1 , PGE ₂ and TGF-beta, or

(xvi) PD-1 , TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Klrdl gene and the Ly6c1 gene and an increase in the level of (xvii) PD-1 , CTLA-4, PGE ₂ and TGF- beta, (xviii) PD-1 , CTLA-4, PGE ₂ and IL-10, (xix) PD-1 , CTLA-4, TGF-beta and IL- 10, (xx) PD-1 , PGE ₂ , TGF-beta and IL-10, or (xxi) CTLA-4, PGE ₂ , TGF-beta and IL-10. Exemplarily, the T cells may also be exhausted CD4+ T cells associated with a diminished expression of the Klrdl gene and the Ly6c1 gene and an increase in the level of (xxii) PD-1 , CTLA-4, PGE ₂ , TGF-beta and IL-10.

The chronic inflammation may be longer-lasting.

In a preferred embodiment, the chronic inflammation lasts for at least a week, at least a month, at least six month or at least a year.

In humans, a chronic inflammation is preferably understood as lasting for at least one month or at least three months. This understanding may also be applied to the present invention.

As indicated above, the TNF-alpha antagonist as used herein may be any TNF- alpha antagonist known in the art.

In a preferred embodiment, the TNF-alpha antagonist comprises or is a compound selected from the group consisting of:

(i) an anti-TNF-alpha antibody, in particular a monoclonal anti-TNF-alpha

antibody;

(ϋ) a TNF-alpha-binding antibody fragment of (i);

(iii) a TNF receptor (TNFR);

(IV) a TNF-alpha-binding fragment of (iii);

(v) etanercept;

(vi) a xanthine derivative, in particular pentoxifylline or bupropion;

(νϋ) an 5-hydroxy tryptamine 2A agonist, in particular (R)-2,5-dimethoxy-4- iodoamphetamine, TCB-2, lysergic acid diethylamide or lysergic acid 2,4- dimethylazetidide;

(viii) curcumin;

(iv) a catechin;

(x) an antagonist of cannabinoid receptor CB1 or CB2; and

(xi) a combination of two or more of (i)-(x).

An anti-TNF-alpha antibody is an antibody directed against and binding to TNF- alpha. An anti-TNF-alpha receptor is a receptor (naturally occurring or artificial) for TNF-alpha. Alternatively or additionally, a TNF-alpha-binding fragment of the antibody or receptor may be used. In the context of the present invention, the antibody receptor or fragment thereof binds to TNF-alpha to inhibit or antagonize TNF-alpha's action in the body, particularly its downstream signaling. The present invention includes, for example, monoclonal (preferred) and polyclonal antibodies, chimeric, single chain, and humanized antibodies, as well as Fab fragments, or the product of a Fab expression library. It is within the present invention that the antibody may be chimeric, i. e. that different parts thereof stem from different species or at least the respective sequences are taken from different species.

For preparation of monoclonal antibodies, any technique known in the art which provides antibodies produced by continuous cell line cultures can be used.

Techniques described for the production of single chain antibodies (U.S. Patent No. 4,946,778) can be adapted to produce single chain antibodies. Also, transgenic mice, or other organisms such as other mammals, may be used to express humanized antibodies according to this invention.

Examples or monoclonal antibody include infliximab (Remicade), adalimumab (Humira), certolizumab pegol (Cimzia), and golimumab (Simponi). Known antibodies to inhibit TNF Alpha related diseases such as Crohn's disease, inflammation and rheumatoid arthritis, which are also contemplated for use in the present invention, include e.g. ozoralizumab (Ablynx NV), adalimumab (AbbVie Inc.), DLX-105 (Delenex Therapeutics AG), YHB-141 12 (Yuhan Corporation), APX-001 (Apexigen, Inc.), AG-014 (ActogeniX NV), AME-532 (Trimer Biotech, LLC), HL-036 (HanAII Biopharma Co., Ltd.), MP-1031 (MetrioPharm AG), OPK- 13031 (TheraKine Ltd.), OPK-30020 (TheraKine Ltd.), CYT007-TNFQb (Cytos Biotechnology AG), DLX-015 (Delenex Therapeutics AG), AZD-9773 (BTG pic), CDP-571 (UCB S.A.), pentoxifylline (Walter Reed Army Medical Center), thalidomide (National Institute of Nursing Research (NINR)), afelimomab also referred to as MAK 195F, Certolizumab pegol (UCB), and Golimumab (MSD/Centocor Inc.). Further anti-TNF-alpha compounds include ALKS-693 (Acceleron Pharma, Inc.), CDX-085 also referred to as Heptax (ardax Pharmaceuticals, Inc.), COVA-322 (Covagen AG), CTP-221 (Concert Pharmaceuticals, Inc.), UTL-5G (21 st Century Therapeutics, Inc.), TA-101 (TechnoPhage SA), adalimumab Next Generation (rEVO Biologies), ISIS-104838 (Isis Pharmaceuticals, Inc.), OA-1 (Weizmann Institute of Science), BVX-10 (Biocon Limited), siRNAs to Inhibit TNF-alpha for Inflammation (RXi Pharmaceuticals Corporation) and ABT-122 (AbbVie Inc.) Etanercept (Enbrel) is a fusion protein of the soluble TNF receptor 2 and the Fc end of the lgG1 antibody. However, also some simple molecules including, e.g., xanthine derivatives in particular pentoxifylline or bupropion; 5-hydroxy tryptamine 2A agonists, in particular (R)-2,5-dimethoxy-4-iodoamphetamine, TCB-2, lysergic acid diethylannide or lysergic acid 2,4-dimethylazetidide; curcumin; catechin as well as a antagonist of cannabinoid receptor CB1 and CB2 such as SR141716A and SR144528 are well known TNF-alpha antagonists. Also combinations of the above compounds are contemplated.

More preferably, the TNF-alpha antagonist comprises a compound selected from the group consisting of:

(i) an anti-TNF-alpha antibody, in particular a monoclonal anti-TNF-alpha

antibody;

(ii) a TNF-alpha-binding antibody fragment of (i);

(iii) a TNF receptor (TNFR);

(iv) a TNF-alpha-binding fragment of (iii); and

(v) etanercept.

Even more preferably, the TNF-alpha antagonist comprises a compound is an anti-TNF-alpha antibody, in particular a monoclonal anti-TNF-alpha antibody, or a TNF-alpha-binding antibody fragment thereof.

In an even more preferred embodiment, the TNF-alpha antagonist is a monoclonal anti-TNF-alpha antibody selected from the group consisting of infliximab, adalimumab, certolizumab, and golimumab. These monoclonal antibodies are commercially available.

In a particularly preferred embodiment, the TNF-alpha antagonist is infliximab.

The TNF-alpha antagonist may be for use in the patient in need thereof. Therefore, it may be preferably for a mammal such as, e.g., for a human or a mouse.

In a particularly preferred embodiment, the TNF-alpha antagonist is for use in a human.

The chronic viral infection may be any chronic infection known in the art. Preferably, it is an infection with viruses bearing an RNA genome. In a highly preferred embodiment, the chronic viral infection is a human immunodeficiency virus (HIV), Hepatitis B virus (HBV), Hepatitis C virus (HCV), Epstein-Barr virus (EBV), cytomegalovirus (CMV) and/or a lymphocytic choriomeningitis virus (LCMV) infection.

The above viral infections are all well-known to the person skilled in the art. HIV has been characterized above and below. Hepatitis is an inflammation of the liver, most commonly caused by a viral infection. There are five main hepatitis viruses, referred to as types A, B, C, D and E. These five types are of greatest concern because of the burden of illness and death they cause and the potential for outbreaks and epidemic spread. In particular, types B and C lead to chronic disease in hundreds of millions of people and, together, are the most common cause of liver cirrhosis and cancer. About 1 million deaths per year are attributed to viral hepatitis infections. Together, hepatitis B virus (HBV) and hepatitis C (HCV) are the leading cause of liver cancer in the world, accounting for 78 percent of cases. Accordingly, HBV and HCV infections account for a substantial proportion of liver diseases worldwide.

EBV, also called human herpesvirus 4 (HHV-4), is a virus of the herpes family, and is one of the most common viruses in humans. It is best known as the cause of infectious mononucleosis (glandular fever). It is also associated with particular forms of cancer, such as Hodgkin's lymphoma, Burkitt's lymphoma, nasopharyngeal carcinoma, and conditions associated with human immunodeficiency virus (HIV) such as hairy leukoplakia and central nervous system lymphomas. There is evidence that infection with the virus is associated with a higher risk of certain autoimmune diseases especially dermatomyositis, systemic lupus erythematosus, rheumatoid arthritis, Sjogren's syndrome, and multiple sclerosis. EBV infects B cells of the immune system and epithelial cells. Once the virus's initial lytic infection is brought under control, EBV latently persists in the individual's B cells for the rest of the individual's life.

CMV that infects humans is commonly known as human CMV (HCMV) or human herpesvirus-5 (HHV-5), and is the most studied of all cytomegaloviruses. Within Herpesviridae, CMV belongs to the Betaherpesvirinae subfamily, which also includes the genera Muromegalovirus and Roseolovirus (HHV-6 and HHV-7). It is related to other herpesviruses within the subfamilies of Alphaherpesvirinae that includes herpes simplex viruses (HSV)-1 and -2 and varicella-zoster virus (VZV), and the Gammaherpesvirinae subfamily that includes . All herpesviruses share a characteristic ability to remain latent within the body over long periods. Although they may be found throughout the body, CMV infections are frequently associated with the salivary glands in humans and other mammals. HCMV infection is typically unnoticed in healthy people, but can be life-threatening for the immunocompromised, such as HIV-infected persons, organ transplant recipients, or new born infants. It can cause hydrops fetalis in infants.

The patient may be administered with the TNF-alpha antagonist once or more then once, thus, repeatedly. In a preferred embodiment, the patient is administered with the TNF-alpha antagonist repeatedly.

Therefore, the patient may be administered with the TNF-alpha antagonist at least twice, at least three times, at least four times, at least five times, at least six times, at least seven times, at least eight times, at least nine times, at least ten times, at least 15 times, at least 20 times, at least 30 times, at least 50 times, at least 100 times or more often.

In a more preferred embodiment, the patient is administered with the TNF-alpha antagonist repeatedly at least once per week, in particular at least once every second day.

Such treatment may last for several days, several weeks, several months or even several years.

In a preferred embodiment, the TNF-alpha antagonist is administered via injection.

Exemplarily, the TNF-alpha antagonist may be administered intravenously {i.v.). In case the TNF-alpha antagonist is an antibody or fragment thereof, it may optionally be combined with one or more adjuvant(s).

In a more preferred embodiment, the TNF-alpha antagonist is an anti-TNF-alpha antibody or a TNF-alpha-binding antibody fragment of which a dose of 1 -50 mg/kg body weight is administered per injection.

Herein, it will be understood that a TNF-alpha-binding antibody fragment is also suitable to bind the TNF-alpha with a Kd of not more than 10 μηηοΙ/Ι, not more than 1 μηηοΙ/Ι, in particular less than 100 nmol/l . In a particularly preferred embodiment, the TNF-alpha antagonist is an anti-TNF- alpha antibody or a TNF-alpha-binding antibody fragment of which a dose of 2-25 mg/kg body weight is administered per injection. Such anti-TNF-alpha antibody or a TNF-alpha-binding antibody fragment may be used as the sole compound for treating the viral infection or in combination with one or more further anti-viral compound(s).

In a preferred embodiment, the TNF-alpha antagonist is administered concomitantly, previously or subsequently with a second compound for treating the viral infection.

The person skilled in the art will know a number of such anti-viral, in particular anti- HIV compounds. Such second compound may exemplarily be a compound selected from the group consisting of an entry or fusion inhibitor, a nucleoside/nucleotide reverse transcriptase inhibitor, a non-nucleoside reverse transcriptase inhibitor, an integrase inhibitor, and a protease inhibitor such as, e.g., a compound is selected from the group consisting of T20, AZT, efavirenz, raltegravir and darunavir.

As used herein, a concomitant administration may be an administration in a single composition or in two separate compositions that may also, optionally, be administered via the same or different routes of administration (e.g., via injection, orally, nasally, percutaneously, etc.).

As used herein, when administering the TNF-alpha antagonist previously or subsequently, there may be a time interval between the administration of the TNF- alpha antagonist and the second compound of less than one hour, one hour or more, three hours or more, six hours or more, twelve hours or more, 24 hours or more, two days or more or a week or more.

It will be noticed that often the TNF-alpha antagonist for use according to the present invention forms part of a pharmaceutical composition. Therefore, a further aspect of the present invention relates to a pharmaceutical composition for use in a method of treating a viral infection in a patient comprising the TNF-alpha antagonist for use according to the present invention and a pharmaceutically acceptable carrier. In this aspect, the definitions as laid out in detail above also apply mutatis mutandis.

A pharnnaceutically acceptable carrier according the present invention may be any additive that is pharmaceutically acceptable, therefore, any additive that is nontoxic to the patient. Exemplarily, a pharmaceutically acceptable carrier may comprise a solvent such as, e.g., water, dimethyl sulfoxide (DMSO), ethanol, vegetable oil, paraffin oil or combinations thereof. Furthermore, a carrier may contain one or more detergents, one or more foaming agents (e.g., sodium lauryl sulfate (SLS)/ sodium doceyl sulfate (SDS)), one or more coloring agents (e.g., ΤΊΟ ₂ , food coloring), one or more vitamins, one or more salts (e.g., sodium, potassium, calcium, zinc salts), one or more humectants (e.g., sorbitol, glycerol, mannitol, propylene glycol, polydextrose), one or more enzymes, one or more preserving agents (e.g., benzoic acid, methylparabene), one or more texturing agents (e.g., carboxymethyl cellulose (CMC), polyethylene glycol (PEG), sorbitol), one or more emulsifiers , one or more bulking agents, one or more glacing agents, one or more separating agents, one or more antioxidants, one or more herbal and plant extracts, one or more stabilizing agents, one or more polymers (e.g., hydroxypropyl methacrylamide (HPMA), polyethylene imine (PEI), carboxymethyl cellulose (CMC), polyethylene glycol (PEG)), one or more uptake mediators (e.g., polyethylene imine (PEI), dimethyl sulfoxide (DMSO), a cell-penetrating peptide (CPP), a protein transduction domain (PTD), an antimicrobial peptide, etc.) one or more antibodies, one or more sweeteners (e.g., sucrose, acesulfame K, saccharin Na, stevia), one or more counterstain dyes (e.g., fluorescein, fluorescein derivatives, Cy dyes, an Alexa Fluor dyes, S dyes, rhodamine, quantum dots, etc.), one or more homeopathic ingredients one or more gustatory substances and/or one or more fragrances. The combination of the TNF-alpha antagonist as therapeutic agent with a pharmaceutical acceptable carrier may also be understood in the broadest sense as drug or medicine.

In a preferred embodiment, the pharmaceutical composition further comprises a second compound suitable for treating the viral infection.

The properties of such second compound evidently depend on the individual viral infection.

In a more preferred embodiment, the viral infection is HIV and the second compound is suitable for treating an HIV infection. In an even more preferred embodiment, the second compound is selected from the group consisting of an entry or fusion inhibitor, a nucleoside/nucleotide reverse transcriptase inhibitor, a non-nucleoside reverse transcriptase inhibitor, an integrase inhibitor, and a protease inhibitor.

The person skilled in the art will know a number of such anti-viral, in particular anti- HIV compounds.

In a particularly preferred embodiment, the second compound is selected from the group consisting of T20, AZT, efavirenz, raltegravir and darunavir.

In a further aspect, the present invention relates to a method for treating a viral infection in a patient comprising administering to said patient an amount of a tumor necrosis factor alpha (TNF-alpha) antagonist sufficient for treating said viral infection in said patient.

In this aspect, the definitions as laid out in detail above also apply mutatis mutandis. A still further aspect of the present invention relates to a method for treating a viral infection in a patient comprising administering to said patient an amount of a pharmaceutical composition for use comprising the TNF-alpha antagonist for use according to the present invention and a pharmaceutically acceptable carrier sufficient for treating said viral infection in said patient.

In this aspect, the definitions as laid out in detail above also apply mutatis mutandis.

The disclosure is not limited to the particular methodology, protocols, and reagents described herein because they may vary. Further, the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of the present disclosure. As used herein and in the appended claims, the singular forms "a", "an", and "the" include plural reference unless the context clearly dictates otherwise. Similarly, the words "comprise", "contain" and "encompass" are to be interpreted inclusively rather than exclusively.

Unless defined otherwise, all technical and scientific terms and any acronyms used herein have the same meanings as commonly understood by one of ordinary skill in the art in the field of the disclosure. Although any methods and materials similar or equivalent to those described herein can be used in the practice as presented herein, the specific methods, and materials are described herein.

The disclosure is further illustrated by the following figures and examples, although it will be understood that the figures and examples are included merely for purposes of illustration and are not intended to limit the scope of the disclosure unless otherwise specifically indicated.

Brief Description of the Figures

Figure 1 shows the application of inhibitory RNA-fingerprints to in vivo RNA profiles from CD4+ T cells from HIV-infected individuals. A, B, Flow cytometric analysis of PD-1 expression on CD4+ T cells from HIV ^|0W /PD-1 ^|0W or HIV ^hi /PD-1 ^hi patients. A, Representative flow cytometry dot plots from one HIV ^|0W /PD-1 ^|0W and one HIV ^hi /PD-1 ^hi patient. B, Proportion of PD-1 -expressing CD4+ T cells from HIV ^|0W /PD-1 ^|0W (n=26) or HIV ^hi /PD-1 ^hi patients (n=37). Bounds of boxes denote interquartile range; lines within boxes denote mean; whiskers indicate interdecile range. Dots represent outliers. C, Schematic overview of analysis of gene expression data for the contribution of RNA-fingerprints to the differences between CD4+ T cells from HIV ^|0W /PD-1 ^|0W (n=10) or HIV ^hi /PD-1 ^hi patients (n= 10). D, RNA- fingerprint enrichment analysis of inhibitory molecules PD-1 , CTLA-4, PGE ₂ , TGF- betal , and IL-10 for HIV ^hi /PD-1 ^hi vs. HIV ^|0W /PD1 ^|0W HIV patient samples. Barplots indicate percentage of probes within a specific fingerprint, which are significantly different (fold change > ± 2, p-value < 0.05) between the two patient groups. ^* P- value < 0.05 (χ ² test). E, Group prediction analysis using support vector machines (SVM). Mean predicted probability of samples being classified as HIV ^hl /PD-1 ^hl vs. HIV ^|0W /PD-1 ^|0W based on the RNA-fingerprint genes. F, Relative mRNA expression of CD4+ T cells from HIV ^|0W /PD-1 ^|0W (n=5) or HIV ^hi /PD-1 ^hi patients (n=5) for DHRS3, CD79b, OAS1 , TRADD, and ALDOA by qPCR. G, Proportion of CD49d, CD79b, IL-4R, and CCR4 expressing CD4+PD-I + T cells from HIV ^low PD-1 ^low (n= 10) or HIV ^hi /PD-1 ^hi patients (n=8). E, F, Mean ± SEM. ^* P < 0.05 (Student's t-test). See also Figure S1 (A, Flow cytometric analysis of PD-1 expression on CD4 ⁺ T cells from HIV ^|0W /PD-1 ^|0W or HIV ^hi /PD-1 ^hi patients. Mean PD-1 expression of CD4 ⁺ T cells from HIV ^|0W /PD-1 ^low (n=26) or HIV ^hi /PD-1 ^hi patients (n=37). B, Relative PD-1 mRNA expression of CD4 ⁺ T cells from HIV ^|0W /PD-1 ^|0W (n=5) or HIV ^hi /PD-1 ^hi patients (n=7) by qPCR. C-E, Flow cytometric analysis of PD-1 expression on CD8 ⁺ T cells from HIV ^|0W /PD-1 ^|0W or HIV ^hi /PD-1 ^hi patients. C, Representative flow cytometry dot plots from one HIV ^|0W /PD-1 ^|0W and one HIV ^hi /PD-1 ^hi patient. D, Proportion of PD-1 -expressing CD8 ⁺ T cells from HIV ^|0W /PD-1 ^|0W (n=10) or HIV ^hl /PD-1 ^hl patients (n=8). Bounds of boxes denote interquartile range; lines within boxes denote mean; whiskers indicate interdecile range. Dots represent outliers. E, Mean PD-1 expression of CD8 ⁺ T cells from HIV ^|0W /PD-1 ^|0W (n=10) or HIV ^hi /PD-1 ^hi patients (n=8). F-H, Correlation between CD4 ⁺ T-cell count of each individual and HIV-RNA (F), CD4 ⁺ T-cell count and CD4 ⁺ T-cell PD-1 expression (G), and HIV-RNA and CD4 ⁺ T-cell PD-1 expression (H). Open circles: HIV ^|0W /PD- 1 ^|0W patients; closed circles: HIV ^hl /PD-1 ^hl patients. I, Representative flow cytometry dot plots from one HIV ^|0W /PD-1 ^|0W or one HIV ^hi /PD-1 ^hi patient for CTLA-4 expression on CD4 ⁺ T cells. J, Proportion of CTLA-4-expressing CD4 ⁺ T cells from HIV ^|0W /PD-1 ^|0W (n=10) or HIV ^hi /PD-1 ^hi patients (n=8). K, Mean CTLA-4 expression of CD4 ⁺ T cells from HIV ^|0W /PD-1 ^|0W (n=10) or HIV ^hi /PD-1 ^hi patients (n=8). L, Representative flow cytometry dot plots from one HIV ^|0W /PD-1 ^|0W or one HIV ^hi /PD- 1 ^hi patient for CTLA-4 expression on CD8 ⁺ T cells. M, Proportion of CTLA-4- expressing CD8 ⁺ T cells from HIV ^|0W /PD-1 ^|0W (n=10) or HIV ^hi /PD-1 ^hi patients (n=8). N, Mean CTLA-4 expression of CD8 ⁺ T cells from HIV ^|0W /PD-1 ^|0W (n=10) or HIV ^hl /PD-1 ^hl patients (n=8). O, Gene set enrichment analysis (GSEA) using genes regulated by PD-1 , CTLA-4, PGE ₂ , TGF-βΙ , and IL-10 as the gene set in CD4 ⁺ T cells from HIV ^hi /PD-1 ^hi vs. HIV ^|0W /PD-1 ^|0W patients. ES: enrichment score, FDR: False-discovery rate. Significant enrichment is called with a FDR <0.25. ^* P < 0.05 (Student's f-test). n.s. not significant. A, B, E, K, N Mean ± SEM.).

Figure 2 shows the in vivo evidence for TNFR-signaling in late-stage HIV- infection. A, Contribution network derived from gene ontology analysis performed on HIV ^hi /PD-1 ^hi vs. HIV ^low /PD-1 ^low samples built based on 177 significant pathways derived from the gene class testing approach. Shown are genes, which occur in the 20 most significant pathways out of 177. The number of shared GO terms was chosen as the maximal value to obtain a sufficiently connected network. B, Probability of TNFR-signaling in HIV ^hi /PD-1 ^hi vs. HIV ^|0W /PD-1 ^|0W samples was computed using a TNF RNA-fingerprint derived from a comparison of untreated and TNF-treated CD4+ T cells. Barplot indicates percentage of probes derived from the fingerprint to be significantly different (fold change > ± 2, p-value < 0.05). ^* P-value < 0.05 (χ ² test). C, Gene set enrichment analysis (GSEA) using genes up-regulated by TNF as the gene set in CD4+ T cells from HIV ^hi /PD-1 ^hi vs. HIV ^|0W /PD-1 ^|0W patients. ES: enrichment score, FDR: False-discovery rate. Significant enrichment is called with p<0.05 and FDR <0.25. D, group prediction analysis of the TNF RNA-fingerprint. Prediction probability for each sample being classified as HIV- positive (HIV+) vs. uninfected control (HIV-) based on the TNF RNA-fingerprint using an additional publicly available dataset comparing HIV- infected and uninfected controls (GSE9927). E, Relative mRNA expression of CD4+ T cells from HIV ^|0W /PD-1 ^|0W (n=5) or HIV ^hi /PD-1 ^hi patients (n=5) for TNF by qPCR. Mean ± SEM. F, Representative flow cytometry dot plots from one _H l _V iow _{/p D 1} iow _{or one H} |v ^hi /PD-1 ^hi patient for TNF expression gated on CD4+ T cells. G, Proportion of TNF-expressing CD4+ T cells from HIV ^|0W /PD-1 ^|0W (n= 10) or HIV ^hi /PD-1 ^hi patients (n=8). H, Relative mRNA expression of CD4+ T cells from HIV ^|0W /PD-1 ^|0W (n=5) or HIV^/PD-I IV patients (n=5) for OX40, CD74, and TNFAIP3 by qPCR. Mean ± SEM. I-J, Proportion of I, OX40 and J, CD74-expressing CD4+PD-1 + T cells from HIV ^|0W /PD-1 ^|0W (n=10) or HIV ^hi /PD-1 ^hi patients (n=8). K, Serum TNF levels in HIV ^|0W /PD-1 ^|0W (n=20) and HIV ^hi /PD-1 ^hi patients (n=31 ). L, Correlation between CD4+ T cell PD-1 expression of each individual and TNF serum levels. Open circles: HIV ^|0W /PD-1 ^|0W patients; closed circles: HIV ^hi /PD-1 ^hi patients. D, E, G, H, I, J, K, ^* P < 0.05 (Student's t-test). See also Figure S2 (A, Output from Ingenuity Pathway Analysis (IPA) upstream regulator analysis. The mechanistic network of TNF is shown. Looking at the transcriptional network of all upstream regulators, TNF is placed upstream of all regulators, indicating that TNF is the major modulator of the changes seen in HIV ^|0W /PD-1 ^|0W vs. HIV ^hi /PD-1 ^hi patients. B, Probability of TNFR-signaling in HIV ^hi /PD-1 ^hi vs. HIV ^|0W /PD-1 ^|0W patients was computed using a TNF RNA-fingerprint derived from a comparison of untreated and TNF-treated Jurkat cells (GSE2504). The barplot indicates percentage of probes derived from the RNA-fingerprint to be significantly different between CD4 ⁺ T cells from HIV ^hi /PD-1 ^hi and HIV ^|0W /PD-1 ^|0W patients (fold change > ± 2, p-value < 0.05). ^* P-value < 0.05 (χ ² test). C, Group prediction analysis of the TNF-fingerprint. Predicted probability of samples being classified as HIV ^hl /PD-1 ^hl vs. HIV ^|0W /PD-1 ^low based on the Jurkat-derived TNF RNA-fingerprint. Mean ± SEM. ^* P-value < 0.05 (Student's f-test)).

Figure 3 shows the PD-1 expression in CD4+ T cells is dependent on TNFR- signaling in HIV-infected individuals. A, A luciferase reporter construct driven by the predicted human PD-1 promoter -5.0 kb upstream of the transcriptional start site (TSS) and a control construct were transfected into HEK293 cells and luciferase activity was assessed after 24 hours in unstimulated cells and cells stimulated with TNF. Mean ± SD of triplicate cultures. Data are representative of three independent experiments. B-D, Flow cytometric analysis of PD-1 and OX-40 expression on CD4+ T cells from HIV ^|0W /PD-1 ^|0W (upper row) or HIV ^hi /PD-1 ^hi patients (lower row) after treatment of PBMC with CD3 ± Infliximab for two days. B, Representative flow cytometry dot plots from one HIV ^|0W /PD-1 ^|0W and one HIV ^hi /PD-1 ^hi patient. B continued, Relative PD-1 (left) and OX40 expression (right) of CD4 ⁺ T cells from HIV ^|0W /PD-1 ^|0W (n=5) or HIV ^hi /PD-1 ^hi patients (n=5).. C, Relative PD-1 expression of CD4+ T cells from HIV ^|0W /PD-1 ^|0W (n=5) or HIV ^hi /PD-1 ^hi patients (n=5). Mean ± SEM. D, Relative OX-40 expression of CD4+ T cells from HIV ^|0W /PD-1 ^|0W (n=5) or HIV ^hi /PD-1 ^hi patients (n=5). Mean ± SEM. A, C, D ^* P < 0.05 (Student's t-test). n.s. not significant. See also Figure S3 (A, Schematic representation of the human PD-1 locus. The PD-1 promoter predicted by analysis using Genomatix is shown. B, Luciferase reporter constructs driven by the Genomatix-predicted human PD-1 promoter (-5.0 kb), the region directly upstream of the transcriptional start site (-0.5 kb), and an intronic enhancer in intron 4 (Prokunina et al., 2002) (intron 4) were transfected into HEK293 cells and luciferase activity was assessed after 24 hours in unstimulated cells and cells stimulated with TNF. Control represents the empty pGL4.24 construct. Mean ± SD of triplicate cultures are shown. Data are representative of three independent experiments. ^* P < 0.05 (Student's i-test)) Figure3S C, Relative mRNA expression of CD4+ T cells from HIVIow/PD-1 low (n=10) or HIV ^hi /PD-1 ^hi patients (n=10) for TNFRI and TNFRII by microarray analysis. Figure3S D, Expression of TNFRI and TNFRII on CD4+ T cells from HIV ^low /PD-1 ^low (n=5) or HIV ^hi /PD-1 ^hi patients (n=6).

Figure 4 shows mice chronically infected with LCMV have high levels of PD-1 expressing T cells. A, B, Flow cytometric analysis of PD-1 expression on A, CD4+ and B, CD8+ T cells from control (n=6) and chronic LCMV-infected mice (n=4). C, Proportion of gp33-specific CD8+ T cells from control (n=3) and chronic LCMV- infected mice (n=7). D, LCMV serum titers in chronic LCMV-infected mice. E, F, ALT (E) and TNF serum levels (F) in control (n=3) and chronic LCMV-infected mice (n=3). G, Differentially expressed genes between CD4+PD-1 + and CD4+PD- 1 - T cells from mice chronically infected with LCMV. Top 10 differentially up- (left) or downregulated genes (right) visualized as heatmaps. H, GSEA using genes from the murine chronic exhaustion RNA fingerprint as the gene set in CD4+ T cells from HIV ^hi /PD-1 ^hi vs. HIV ^|0W /PD-1 ^|0W patients. ES: enrichment score, FDR: False-discovery rate. Significant enrichment is called with p < 0.05 and FDR <0.25. A, B, C Mean ± SEM. ^* P < 0.05 (Student's t-test). See also Figure S4 (A, Left, sorting strategy to isolate PD-1 expressing CD4 ⁺ T cells from control- and infliximab-treated acute and chronic LCMV-infected mice for gene expression analysis. Right, analysis of purities of isolated cell populations. B, Prediction probability of the murine chronic exhaustion RNA-fingerprint in HIV ^hl /PD-1 ^hl vs. HIV ^|0W /PD-1 ^|0W patients was computed. The barplot indicates percentage of probes derived from the fingerprint to be significantly different between CD4 ⁺ T cells from HIV ^hi /PD-1 ^hi and HIV ^|0W /PD-1 ^|0W patients (fold change > ± 2, p-value < 0.05). ^* P- value < 0.05 (χ ² test). C, Group prediction analysis of the murine chronic exhaustion RNA-fingerprint. Predicted probability of samples being classified as HIV ^hi /PD-1 ^hi vs. HIV ^|0W /PD-1 ^|0W based on the murine chronic exhaustion RNA- fingerprint. Mean ± SEM. ^* P-value < 0.05 (Student's i-test)). I, Proportion of gp66- specific CD4+ (upper left) and gp33-specific CD8+ T cells (upper right) from control (n=4), acute WE (n=4), and chronic neonatal WE LCMV-infected mice (n=4). Box plots showing 25th, 50th and 75th percentiles (horizontal bars), 10th and 90th percentage (error bars), and outliers (dots). Flow cytometric analysis of PD-1 expression on gp66-specific CD4+ (lower left) and gp33-specific CD8+ T cells (lower right) from acute WE LCMV (n=4) and chronic neonatal WE LCMV- infected mice (n =4).

Figure 5 shows TNF-blockade in chronic LCMV-infected mice restores immunity against LCMV. A, Model for Infliximab treatment of mice chronically infected with LCMV. B-D, Left, representative flow cytometric analysis of B, PD-1 expression on CD4+ T cells, C, PD-1 expression on CD8+ T cells, and D, gp33-specific CD8+ T cells from control and infliximab- treated chronic LCMV-infected mice. Right, cumulative data from at least 4 mice per group are shown. E, Immunoblotting for plkk-alpha/beta (Ser176/180) (top) and beta-actin (bottom) in control- and Infliximab-treated CD4+ and CD8+ T cells from mice chronically infected with LCMV. Data shown are representative of three mice each. F, ALT serum levels in control (n=3) and Infliximab-treated chronically LCMV-infected mice (n=3). G, quantification of LCMV titers in serum over time. H, I, LCMV titers in the liver (H) and spleen (I) of control (n=3) and Infliximab-treated chronic LCMV-infected mice (n=3). Mean ± SEM. B, C, D, H, I, ^* P < 0.05 (Student's t-test). See also Figure S5 (A-D, Wild-type and Tnf' ^~ mice were infected with LCMV Docile (2 x 10 ³ pfu). A, gp33- and np396-specific CD8 ⁺ T cells from wild-type and Tnf ^~ mice were analyzed 8 days p.L Cumulative mean data from 7 wild-type T and 8 Tnf ^~ mice per group are shown. B, IFN-γ production of CD8 ⁺ T cells is shown after restimulation of splenocytes from wild-type and Tnf ^~ mice with the LCMV specific epitopes gp33 and np396 (n= 7 wild-type and 8 Tnf ^~ mice). C, LCMV titers in spleen and liver tissue of wild-type and Tnf ^~ mice were determined 8 days after infection with LCMV Docile (n=7 wild-type and 8 Tnf ^~ mice). D, Survival of wild- type and Tnf ^~ mice was monitored after infection (n= starting with 14 wild-type and 15 Tnf ^/_ mice). E-l, Wild-type mice were acute infected with LCMV WE (2 x 10 ⁴ pfu) and measured after 10 days. E, PD-1 expression on CD4 ⁺ T cells, F, PD-1 expression on CD8 ⁺ T cells, and G, gp33-specific CD8 ⁺ T cells from control- and infliximab-treated acute LCMV-infected mice. Right, cumulative mean data from at least 3 mice per group are shown. H, I, Immunoblot analysis of plkka/β (Ser176/180) (top) and β-actin (bottom) in H, CD4 ⁺ and I, CD8 ⁺ T cells from control- and infliximab-treated acute LCMV-infected mice. Data shown are representative of three mice each. E-G, Mean ± SEM. ^* P < 0.05 (Student's t- test)).

Figure 6 shows infliximab-treatment of chronic LCMV-infected mice reverts CD4+PD-1 + T cell gene expression. A, Principle component analysis of CD4+PD- 1 - and CD4+PD-1 + T cells from control- and infliximab-treated chronic LCMV- infected mice. B, Unsupervised hierarchical clustering of variable genes of CD4+PD-1 - and CD4+PD-1 + T cells from control- and Infliximab-treated chronic LCMV-infected mice. C, Overlap of differentially expressed genes (FC > 2, p-value < 0.05) between CD4+PD-1 - and CD4+PD-1 + T cells from control- and infliximab- treated chronic LCMV-infected mice shown as Venn-diagram differ from another.

D, Fold-change-fold-change plot showing the influence of Infliximab-treatment on gene expression in CD4+PD-1 - and CD4+PD-1 + T cells. The y axis compares the expression profiles between control- and Infliximab-treated CD4+PD-1 - T cells mice, whereas the x axis compares the expression profiles between control- and Infliximab-treated CD4+PD-1 + T cells. Highlighted in red are genes assessed in

E. E, Relative mRNA expression of CD4+PD-1 - and CD4+PD-1 + T cells from control- and Infliximab-treated chronic LCMV-infected mice for Ly6c\ and K\rd\ by qPCR. Mean ± SEM of at least triplicates, representative of two independent experiments. ^* P < 0.05 (Student's t-test). F, GSEA using the murine chronic exhaustion RNA-fingerprint as the gene set in CD4+PD-1 + T cells in control- and Infliximab-treated mice. G, GSEA using a murine TNF RNA-fingerprint (GSE2504) as the gene set in CD4+PD-1 + vs. CD4+PD-1 " T cells in control- and Infliximab- treated mice. See also Figure S6 (A, GSEA using genes up- or down-regulated between effector and exhausted CD8 ⁺ T cells (GSE9650) as the gene set in CD4 ⁺ PD-1 ⁺ T cells from infliximab and control -treated chronically LCMV-infected mice. B, GSEA using the human TNF RNA-fingerprint genes defined in CD4 ⁺ T cells as the gene set in CD4 ⁺ PD-1 ⁺ T cells in infliximab and PBS-treated mice. ES: enrichment score, FDR: False-discovery rate. Significant enrichment is called with p<0.05 and FDR <0.25).

Figure 7 shows TNF induces PD-1 expression and T cell exhaustion in T cells adoptively transferred into chronic LCMV-infected mice. A, Protocol used to determine the effect of TNF on T cells in mice chronically infected with LCMV. 2 x 10 ⁶ CD4+ and 2 x 10 ⁶ CD8+ T cells from Thy1 .2 congenic wild-type or TNFRI/II ^-/- mice 8 days after acute infection with LCMV were transferred to Thy1 .1 congenic mice chronically infected with LCMV and assessed after 10 days. B, Left, representative flow cytometry analysis of PD-1 expression on Thy1 .2 CD4+ T cells from mice receiving wild- type or TNFRI/II ^-/- CD4+ and CD8+ T cells. Right, cumulative data. C-E, Flow cytometric analysis of PD-1 expression on Thy1 .1 CD4+ T cells from mice receiving no transfer, wild-type, or TNFRI/II ^-/- CD4+ and CD8+ T cells. C, representative flow cytometric analysis. D, cumulative percentage of Thy1 .1 +CD4+PD-1 + T cells. E, Cumulative numbers of splenic Thy1 .1 +CD4+PD-1 + T cells. F, Left, representative flow cytometric analysis of gp33- specific Thy1 .2 CD8+ T cells from mice receiving wild-type or TNFRI/II ^-/- CD4+ and CD8+ T cells. Right, cumulative data. G-l, flow cytometric analysis of gp33-specific Thy1 .1 + CD8+ T cells from mice receiving no transfer, wild-type, or TNFRI/II ^-/- CD4+ and CD8+ T cells. G, representative flow cytometric analysis. H, cumulative percentage of splenic gp33-speeifie CD8+ T cells. I, Cumulative numbers of splenic gp33-speeifie CD8+ T cells. J, K, Flow cytometric analysis of gp33-specifie Thy1 .1 + CD8+ T cells from mice receiving wild-type or TNFRI/II ^-/- CD8+ T cells. J, cumulative percentage of splenic gp33-specific CD8+ T cells. K, Cumulative numbers of splenic gp33-specific CD8+ T cells. L, ALT serum levels in mice receiving no transfer, wild-type or TNFRI/II ^-/- CD4+ and CD8+ T cells, or wild-type or TNFRI/II ^-/- CD8+ T cells. M, Quantification of LCMV titers in serum in mice receiving no transfer, wild-type or TNFRI/II ^-/- CD4+ and CD8+ T cells, or wild-type or TNFRI/II ^-/- CD8+ T cells. Data from at least 3 mice per group are shown. ^* P < 0.05 (Student's l-test). See also Figure S7 (A, B, Left, sorting strategy to isolate Thy1 .2 CD4 ⁺ and CD8 ⁺ T cells from acute LCMV-infected A, wild-type and B, TNFRI/II ^-/- mice for adoptive transfer in chronic LCMV-infected Thy1 .1 mice. Right, analysis of purities of isolated cell populations. C, Left, representative flow cytometric analysis of PD-1 expression on Thy1 .2 CD8 ⁺ T cells from mice receiving wild-type or TNFRI/II ^-/- CD4 ⁺ and CD8 ⁺ T cells. Right, cumulative data. D- F, Flow cytometric analysis of PD-1 expression on Thy1 .1 CD8 ⁺ T cells from mice receiving no transfer, wild-type, or TNFRI/II ^-/- CD4 ⁺ and CD8 ⁺ T cells. D, Representative flow cytometric analysis. E, Cumulative percentage of splenic CD8 ⁺ PD-1 ⁺ T cells. F, Cumulative numbers of splenic CD8 ⁺ PD-1 ⁺ T cells. Data from at least 3 mice per group are shown).

Figure S8 shows a workflow for screening of the HIV patient population. Boxes in black indicate the patient groups chosen for further studies. Samples from patients with at least 10 ⁶ /ml CD4 ⁺ T cells were used for further studies ( ^* ). Patients were excluded when RNA amount and quality did not reach necessary quality standards for genomic analysis ( ^** ).

Figure S9 shows generation of RNA-fingerprints. A, Prior to assessment of transcriptional changes the functional impact of all components on purified CD4 ⁺ T cells was analyzed. Freshly isolated primary human CD4 ⁺ T cells were labeled with CFSE and left unstimulated or were stimulated as indicated. After 4 days CFSE dilution was analyzed by flow cytometry. The overall percentage of dividing cells is displayed in the corresponding gate. For each condition at least four individual experiments were performed. Shown here are representative results. B, CD4 ⁺ T cells were stimulated as above. After four days the concentration of IFN-γ was determined using flow cytometric bead assays. For each condition at least four individual experiments were performed. Mean ± SD. C, Visualization of fold changes and amount of genes significantly altered in CD4 ⁺ T-cell transcription profiles after indicated stimulations of three different healthy blood donors defining the RNA-fingerprint of the particular analyzed component.

Figure 10 shows TNF neutralization in chronic neonatal WE LCMV-infected mice restores immunityagainst LCMV. a, Model for TNF neutralization of chronic neonatal WE LCMV-infected mice, b, Left, representative, right, cumulative flow cytometric analysis of gp66 expression in splenic CD4+ T cells, c, Total numbers of splenic gp66+ CD4+ T cells, d, PD-1 expression on splenic gp66+ CD4+ T cells, e, Left, representative, right, cumulative flow cytometric analysis of gp33 expression in splenic CD8+ T cells, f, Total numbers of splenic gp33+ CD8+ T cells, g, PD-1 expression on splenic gp33+ CD8+ T cells, h, Immunoblotting for plkk□/□ (Ser176/180) (top) and □-actin (bottom) in CD4+ and CD8+ T cells from mice with chronic neonatal WE LCMV infection after TNF neutralization. Data shown are representative of three mice each, i, ALT serum levels in chronic neonatal WE LCMV infection after TNF neutralization (each n=3). j, Quantification of LCMV titers in serum over time (each n=3). k, I, LCMV titers in the liver (k) and spleen (I) after TNF neutralization (each n=3). Box plots showing 25th, 50th and 75 ^th percentiles (horizontal bars), 10th and 90th percentage (error bars), and outliers (dots), m, IL-2 and n, IFN-γ expression on splenic gp33+ CD8+ T cells, o, IL-21 , p, IL-2, q, IFN-γ, and r, CD40L expression on splenic gp66+ CD4+ T cells, b, c, d, e, f, g, k, I, m, n, o, p, q, r * P < 0.05 (Student's i-test). b, c, d, e, f, g, m, n, o, p, q, r each n=5. Data are representative for two independent experiments.

Figure 11 shows TNF induces PD-1 expression and loss of T-cell function in adoptively transferred T cells into chronic neonatal WE LCMV-infected mice a, GSEA using a murine TNF RNA-fingerprint (GSE2504) as the gene set in CD4+PD-1 + vs. CD4+PD-1 - (left) and CD8+PD-1 + vs. CD8+PD-1 - T cells (right), b, Protocol used to determine the effect of TNF on T cells in mice with chronic neonatal WE LCMV infection. 2 x 10 ⁶ CD8+ T cells +/- 2 x 10 ⁶ CD4+ from Thy1 .2 congenic wildtype or TNFRI/II-/- mice 8 days after acute infection with WE LCMV were transferred to Thy1 .1 mice with chronic neonatal WE LCMV infection and assessed after 10 days, c, d, Flow cytometric analysis of gp33-specific Thy1 .2+ CD8+ T cells from mice receiving wildtype (n=3) or TNFRI/II-/- CD8+ T cells (n=3). c, Cumulative percentage of splenic gp33-specific CD8+ T cells, d, Cumulative percentage of splenic PD-1 + gp33-specific CD8+ T cells, e, ALT serum levels in mice receiving no transfer (n=4), wildtype (n=3), or TNFRI/II-/- CD8+ T cells (n=3). f, Quantification of LCMV titers in serum in mice receiving no transfer (n=4), wildtype (n=3), or TNFRI/II-/- CD8+ T cells (n=3). g, Left, representative flow cytometric analysis of gp66 expression on Thy1 .2 CD4+ T cells from mice receiving wildtype (n=4) or TNFRI/II-/- CD4+ and CD8+ T cells (n=4). Right, cumulative data, h, Cumulative percentage of Thy1 .2+ gp66+CD4+PD-1 + T cells (each n=4). i, Left, representative flow cytometric analysis of gp33 expression on Thy1 .2 CD4+ T cells from mice receiving wildtype (n=4) or TNFRI/II-/- CD4+ and CD8+ T cells (n=4). Right, cumulative data, j, Cumulative percentage of Thy1 .2+gp33+CD4+PD-1 + T cells (each n=4). k-m, Flow cytometric analysis of gp33-specific Thy1 .1 + CD8+ T cells from mice receiving no transfer (n=4), wildtype (n=4), or TNFRI/II-/- CD4+ and CD8+ T cells (n=4). k, Cumulative percentage of splenic gp33-specific CD8+ T cells. I, Cumulative numbers of splenic gp33-specific CD8+ T cells, m, Cumulative percentage of splenic PD-1 + gp33-specific CD8+ T cells, n, ALT serum levels in mice TNFR-signaling impairs CD4+ T-cells in chronic viral infection 42 receiving no transfer, wildtype or TNFRI/II-/- CD4+ and CD8+ T cells (each n=4). o, Quantification of LCMV titers in serum in mice receiving no transfer, wildtype or TNFRI/II-/- CD4+ and CD8+ T cells (each n=4). p, TNF serum levels in mice receiving no transfer,b wildtype or TNFRI/II-/- CD4+ and CD8+ T cells (each n=4). d, g, h, i, j ^* P < 0.05 (Student's t- test), k, I, m ^* P < 0.05 vs. wildtype (One-way ANOVA with Dunnett correction), n.s. not significant. Data are representative for two independent experiments.

Figure 12 shows the analysis of murine chronic LCMV infection as model for late- stage HIV-infection. a, LCMV serum titers in chronic clone 13 LCMV-infected mice (n=5). b, ALT serum levels in control (n=3) and chronic neonatal WE LCMV- infected mice (n=3). c, Left, sorting strategy to isolate PD-1 expressing CD4+ T cells from acute WE and chronic neonatal WE LCMV-infected mice for gene expression analysis. Right, analysis of purities of isolated cell populations, d, GSEA using genes from the murine chronic clone 13 LCMV-infected gp66 CD4+ T cell RNA fingerprint.2 as the gene set in CD4+PD-1 + and CD4+PD-1 - T cells from chronic neonatal WE LCMV-infected mice. ES: enrichment score, FDR: False- discovery rate. Significant enrichment is called with p < 0.05 and FDR <0.25. e, Prediction probability for each sample being classified as HIV-positive (HIV+) vs. uninfected control (HIV-) based on group prediction analysis of the murine chronic neonatal WE LCMV-infected Th cell RNA fingerprint using an additional publicly available dataset comparing HIV-infected and uninfected controls (GSE9927). Various probabilities for the murine chronic exhaustion are indicated by an RNA fingerprint.

Figure 13 shows TNF neutralization in chronic neonatal WE LCMV-infected mice restores immunity against LCMV. a, Left, representative, right, cumulative flow cytometric analysis of np396 expression in splenic CD8+ T cells (each n=5). b, Total numbers of splenic np396+ CD8+ T cells (each n=5). c, PD-1 expression on splenic np396+ CD8+ T cells (each n=5). d, Left, representative, right, cumulative flow cytometric analysis of PD-1 expression in splenic CD4+ T cells after control treatment (n=5) or TNF neutralization (n=4). e, Left, representative, right, cumulative flow cytometric analysis of PD-1 expression in splenic CD8+ T cells from animals after control treatment (n=6) or TNF neutralization (n=5). a, b, c, d, e ^* P < 0.05 (Student's i-test). Data are representative for two independent experiments.

Figure 14 shows the role of TNFR-signaling in acute murine LCMV infection, a-e, Wildtype mice were acute infected with LCMV WE (2 x 10 ⁴ pfu) and analyzed after 10 days, a, b, Immunoblot analysis of plkk · / · (Ser176/180) (top) and β-actin (bottom) in a, CD4+ and b CD8+ T cells from animals after control treatment or TNF neutralization during acute WE LCMV infection. Data shown are representative of three mice each, c, PD-1 expression on CD4+ T cells, d, PD-1 expression on CD8+ T cells, and e, gp33-specific CD8+ T cells from animals after control treatment or TNF neutralization during acute WE LCMV infection, c-e, Mean ± SEM. ^* P < 0.05 (Student's i-test). n.s. not significant.

Figure 15 shows that TNF neutralization in chronic neonatal WE LCMV-infected mice partially restores immunity against LCMV. a, Model for TNF neutralization of chronic neonatal WE LCMV-infected mice, b, Left, representative, right, cumulative flow cytometric analysis of gp66 expression in splenic CD4+ T cells (each n=8). c, Total numbers of splenic gp66+ CD4+ T cells (each n=8). d, PD-1 expression on splenic gp66+ CD4+ T cells (each n=8). e, Left, representative, right, cumulative flow cytometric analysis of PD-1 expression in splenic CD4+ T cells (each n=8). c, Total numbers of splenic gp66+ CD4+ T cells (each n=8). f, Left, representative, right, cumulative flow cytometric analysis of gp33 expression in splenic CD8+ T cells (Control n=5, anti-TNF n=9). g, Total numbers of splenic gp33+ CD8+ T cells (Control n=5, anti-TNF n=9). h, PD-1 expression on splenic gp33+ CD8+ T cells (Control n=5, anti-TNF n=9). i, Left, representative, right, cumulative flow cytometric analysis of PD-1 expression in splenic CD8+ T cells (Control n=5, anti- TNF n=9). j, ALT serum levels in chronic clone 13 LCMV infection after TNF neutralization (each n=5). k, Quantification of LCMV titers in serum over time (Control n=5, anti-TNF n=9). k, I, LCMV titers in the liver (I), kidney (m), and lung (n) after TNF neutralization (Control n6, anti-TNF n=1 1 ). Box plots showing 25th, 50th and 75 ^th percentiles (horizontal bars), 10th and 90th percentage (error bars), and outliers (dots), m, IL-2 and n, IFN-γ expression on splenic gp33+ CD8+ T cells (Control n=5, anti-TNF n=9). o, IL-21 , p, IL-2, q, IFN-γ, and r, CD40L expression on splenic gp66+ CD4+ T cells (Control n=5, anti-TNF n=9). b, c, d, e, f, g, h, i, I, m, n, o, p, q, r, s, t ^* P < 0.05 (Student's i-test). n.s. not significant. Data are representative for two independent experiments.

Figure 16 shows that TNFR-signaling in chronic neonatal WE LCMV-infected mice induces PD-1 expression and loss of T-cell helper function, a, b, Left, sorting strategy to isolate Thy1 .2 CD4+ and CD8+ T cells from acute LCMVinfected a, wildtype and b, TNFRI/II-/- mice for adoptive transfer in chronic neonatal WE LCMV-infected Thy1 .1 mice. Right, analysis of purities of isolated cell populations, c-e, Flow cytometric analysis of gp66-specific Thy1 .1 + CD4+ T cells from mice receiving no transfer (n=4), wildtype (n=4), or TNFRI/II-/- CD4+ and CD8+ T cells (n=4). c, Cumulative percentage of splenic gp66-specific CD4+ T cells, d, Cumulative numbers of splenic gp66-specific CD4+ T cells, e, Cumulative percentage of splenic PD-1 + gp66-specific CD4+ T cells, f, Left, representative flow cytometric analysis of PD-1 expression on Thy1 .2 CD4+ T cells

from mice receiving wild-type or TNFRI/II-/- CD4+ and CD8+ T cells. Right, cumulative data, g, h Flow cytometric analysis of PD-1 expression on Thy1 .1 CD4+ T cells from mice receiving no transfer, wild-type, or TNFRI/II-/- CD4+ and CD8+ T cells, g, Cumulative percentage of splenic CD4+PD-1 + T cells, h, Cumulative numbers of splenic CD4+PD-1 + T cells, i, Left, representative flow cytometric analysis of PD-1 expression on Thy1 .2 CD8+ T cells from mice receiving wild-type or TNFRI/II-/- CD4+ and CD8+ T cells. Right, cumulative data, j, k Flow cytometric analysis of PD-1 expression on Thy1 .1 CD8+ T cells from mice receiving no transfer, wild-type, or TNFRI/II-/- CD4+ and CD8+ T cells, j, Cumulative percentage of splenic CD8+PD-1 + T cells, k, Cumulative numbers of splenic CD8+PD-1 + T cells. Data from 4 mice per group are shown, f , i ^* P < 0.05 (Student's i-test). c, d, e, g, h. j, k ^* P < 0.05 vs. wildtype (One-way ANOVA with Dunnett correction), n.s. not significant. Data are representative for two independent experiments. Examples

Materials and Methods Recruitment of patients

HIV-positive patients were recruited from our outpatient cohort (n=1304) (Table S1 and Figure S8). 37 patients with increased CD4+ T cell PD-1 expression (HIV ^hl /PD-1 ^hl ) were identified by flow cytometry. For a patient to be defined as HIV ^hi /PD-1 ^hi , the cutoff was set at 15% of PD-1 positive cells in the CD4+ T cell population with HIV virus copies >1007ml. 26 HIV-infected patients with PD-1 expression in CD4+ T cells below 3% and HIV copies <50/ml were used as controls. In 20 of all assessed patients (10 HIV ^hi /PD-1 ^hi and 10 HIV ^|0W /PD-1 ^|0W ) we were able to collect enough RNA for whole-genome transcriptome analysis (Figure S8). To define RNA-fingerprints for inhibitory pathways in healthy CD4 ⁺ T-cells, blood samples in form of standard buffy coat preparations were collected from four different healthy blood donors at the Center for Transfusion Medicine and processed immediately as described below. All blood samples were collected after informed consent following Institutional Review Board approval (07-093). Isolation of CD4+ T cells from healthy donors and HIV-infected individuals

CD4 ⁺ T cells from buffy coat samples were isolated as described previously (Chemnitz et al., 2007). Peripheral blood mononuclear cells (PBMC) from HIV- infected patients were prepared using Ficoll (Amersham Biosciences, Buckinghamshire, UK). CD4 ⁺ T cells were isolated by positive selection using Miltenyi magnetic cell sorting columns (Miltenyi Biotech, Bergisch Gladbach, Germany) after depletion of CD14 ⁺ monocytes.

Antibodies

Analysis of human CD4 ⁺ and CD8 ⁺ T cells by flow cytometry was performed using the following antibodies (all from BD Biosciences unless indicated otherwise): phycoerythrin-cyanine 7-conjugated or phycoerythrin-conjugated anti-PD-1 (Biolegend, EH12.2H7), fluorescein isothiocyanate-conjugated anti-OX40 (e- Bioscience, ACT35), phycoerythrin-conjugated anti-TNFRII (hTNFR-M1 ), fluorescein isothiocyanate-conjugated anti-TNF (Biolegend, MAb1 1 ), phycoerythrin-conjugated anti-IFN-γ (Biolegend, B27), allophycocyanin-conjugated anti-CTLA-4 (BNI3), peridinin chlorophyll protein-cyanine 5.5 or phycoerythrin- conjugated CD4 (Biolegend, RPA-T4), fluorescein isothiocyanate or V450- conjugated CD3 (UCHT1 ), peridinin chlorophyll protein-cyanine 5.5 or V500- conjugated CD8 (RPA-T8), fluorescein isothiocyanate-conjugated CD74 (e- Bioscience, 5-329), fluorescein isothiocyanate-conjugated CD49d (9F10), fluorescein isothiocyanate-conjugated CD79b (e-Bioscience, CB3-1 ), phycoerythrin-conjugated anti-IL4-R (hlL4R-M57), and phycoerythrin-conjugated CCR4 (1 G).For murine samples the following antibodies were used: fluorescein isothiocyanate- or phycoerythrin-conjugated PD-1 (e-Bioscience, J43), phycoerythrin-cyanine 7-conjugated CD8 (e-Bioscience, 53-6.7), allophycocyanin- cyanine 7-conjugated CD4 (e-Bioscience, RM 4-5), peridinin chlorophyll protein- cyanine 5.5-conjugated Thy1 .2 (e-Bioscience, 30-H12), V450-conjugated Thy1 .1 (BD Biosciences, OX-7), allophycocyanin-conjugated Dextramers specific for gp33 (Immudex) were used.

Flow cytometry

Flow cytometry was performed on a FACSCalibur, Canto II or LSR II (all BD Biosciences) and FlowJo software (TreeStar) was used for data analysis. CD4+ and CD8+ T cells were sorted with a FACSAria III (BD Biosciences).

In particular, for murine samples antibody staining was done in presence of Fc receptor blockade (monoclonal antibody 2.4G2 to mouse CD16-CD32 (10 μg/ml); prepared in-house) in PBS. Human samples were stained in PBS. A FACSCalibur, Canto II or LSR II (all BD Biosciences) and FlowJo software (TreeStar) were used for acquisition and data analysis. For sorting of CD4 ⁺ and CD8 ⁺ T cells by flow cytometry for adoptive transfer or gene expression analysis of PD-1 ⁺ CD4 ⁺ , PD-1 ^" CD4 ⁺ , gp33-Dextramer ⁺ CD8a ⁺ , and or gp33-Dextramer ^" CD8a ⁺ T cells, splenocytes were isolated from mice, depleted of non-T cells using the Pan T Cell Isolation Kit II (Miltenyi Biotech), followed by staining with the respective antibodies and filtering through a 10Ό-μηη mesh. Next, cells were sorted with a FACSAria III (BD Biosciences). In the LCMV Docile experiments, tetramer and intracellular cytokine stainings were performed as previously described (Lang et al., 2013). Briefly, blood samples were incubated with gp33-tet and np396-tet for 15 minutes at 37°C. Next, anti-CD8 (eBioscience) antibody was added followed by lysis of erythrocytes and analysis by flow cytometry. For intracellular cytokine staining splenocytes were incubated with the LCMV specific peptides gp33 and np396. After 1 h Brefeldin A (eBiosciences) was added, followed by additional 5h incubation at 37°C. After surface staining with anti-CD8 (eBiosciences) cells were fixed with 2% formalin and permeabilized with 0.1 % Saponin and stained with anti- IFN-γ (eBiosciences) for 30 min at 4°C.

CD4+ T cell stimulation for the generation of RNA-finqerprints

Pathway-specific genome-wide transcriptional changes (RNA-fingerprints) were determined following short term stimulation of primary CD4 ⁺ T cells concomitant with signaling via these inhibitory pathways (for details see Figure S9). For the CTLA-4 and PD-1 fingerprints pure CD4 ⁺ T-cells from buffy coat samples were stimulated by mixing with artificial antigen presenting cells (aAPC) comprised of magnetic beads (Dynal Biotech, Oslo, Norway) coated with the following antibodies: anti-CD3 (OKT3), anti-CD28 (9.3), anti-PD-1 (PD-1 -17), anti-CTLA-4 (ER5.3D6) and anti-MHC-l (W6/32), as previously described (Chemnitz et al., 2007). For defining the PGE _2, TGF-βΙ and IL-10 fingerprints, CD4 ⁺ T-cells were stimulated with CD3/CD28/MHC-I and PGE ₂ , TGF-βΙ or IL-10 were added in concentrations of 1 μΜ, 30 ng/ml or 50 ng/ml, respectively. For the generation of the TNF RNA-fingerprint of CD4 ⁺ T cells, CD4 ⁺ T-cells were left unstimulated or incubated with 100 ng/ml TNF (Sigma-Aldrich) for 5 days.

Assessment of CD4 ⁺ T-cell function for the generation of RNA-fingerprints

T-cell proliferation was analyzed by FACS using CFSE staining, as previously described (Chemnitz et al ., 2007). The concentration of IFN-γ in cell culture supernatants was measured using the human Thi/Th ₂ Cytokine kit II (BD Pharmingen) following the manufacturer's instructions.

RNA preparation, microarrav hybridization and primary data analysis

To define RNA-fingerprints CD4 ⁺ T cells of four different healthy donors were either left unstimulated or stimulated via inhibitory receptors respectively soluble factors as described above. After 8h magnetic beads were removed and cells were lysed in TRIzol reagent (Invitrogen Life Technologies). CD4 ⁺ T cells of HIV- infected individuals were lysed in TRIzol immediately after isolation to retrieve the in vivo state of transcriptional regulation in these cells. RNA isolation, cDNA and cRNA transcription, as well as microarray hybridization and analysis of lllumina Bead array expression data was performed as described previously (Chemnitz et al ., 2007). All statistical and bioinformatic analysis was performed using R language and Bioconductor packages. All microarray samples are listed below. Microarray data can be accessed under GSE52185.

ID n species donor subset cell type Experim. condition

1 10 human chronic PD-1 low CD4+ unstimulated

HIV

2 10 human chronic PD-1 high CD4+ unstimulated

HIV

3 4 human healthy resting Tconv CD4+ unstimulated

4 4 human healthy activated Tconv CD4+ 8 hrs CD3CD28

5 4 human healthy activated Tconv CD4+ 8 hrs CD3CD28 + TGF- β1

6 4 human healthy activated Tconv CD4+ 8 hrs CD3CD28 + IL10

7 4 human healthy activated Tconv CD4+ 8 hrs CD3CD28 + PGE2

8 4 human healthy activated Tconv CD4+ 8 hrs CD3CD28 + PD1

9 4 human healthy activated Tconv CD4+ 8 hrs CD3CD28 +

CTLA4

10 3 human healthy resting Tconv CD4+ 72 hrs cultivated

1 1 3 human healthy resting Tconv CD4+ 72 hrs cultivated + TNF

12 5 murine chronic PD-1 IOW CD4+ 10 days PBS

LCMV

13 5 murine chronic PD-I I0W CD4+ 10 days Infliximab

LCMV

14 5 murine acute PD-I I0W CD4+ 10 days PBS

LCMV

15 5 murine acute PD-I I0W CD4+ 10 days Infliximab

LCMV

16 4 murine chronic PD-1 high CD4+ 10 days PBS

LCMV

17 3 murine chronic PD-1 high CD4+ 10 days Infliximab

LCMV

18 4 murine acute PD-1 high CD4+ 10 days PBS

LCMV

19 3 murine acute PD-1 high CD4+ 10 days Infliximab

LCMV

Establishment of RNA-fingerprints for inhibitory pathways

Specific RNA-fingerprints for an inhibitory molecule were identified by a two-step differential expression analysis of resting and activated CD4 ⁺ T-cells in the absence or presence of each inhibitory signal (Chemnitz et al., 2007). Transcripts present in more than one RNA-fingerprint were subtracted. Transcripts were cross-annotated using gene symbols.

Establishment of the TNF RNA-fingerprints

A specific TNF RNA-fingerprint was established by comparing the gene expression of unstimulated CD4 ⁺ T cells and TNF-treated CD4 ⁺ T cells after 5 days of culture. A second RNA-fingerprint for TNF was taken from GSE2504, where untreated and TNF-treated Jurkat cells were compared (Lewinski et al., 2005). Transcripts were cross-annotated using gene symbols. qPCR

Total RNA extracted using TRIzol from CD4 ⁺ T cells was used to generate cDNA along with the Transcriptor First Strand cDNA synthesis kit (Roche Diagnostics). qPCR was performed using the LightCycler Taqman master kit and the Universal Probe Library assay specific for CD74, TRADD, AOS1 , ALDOA, CD79b, DHRS3, TNF, TNFAIP3, OX40, PD-1 , and beta-2 microglobulin (B2M; Roche Diagnostics) on a LightCycler 480 II (Roche Diagnostics). Results were normalized to B2M expression. Examples for possible primers (Human qPCR oligonucleotides, forward (Fw) primers and reverse (Rv) primers) applicable for qPCR of the aforementioned biomolecules are enlisted in the following:

CD74 Forward primer: 5'-ATGAGCAACTGCCCATGC-3' (SEQ ID NO 1 )

CD74 Reverse primer: 5'-CAGGATGGAAAAGCCTGTGT-3' (SEQ ID NO 2)

TRADD Forward primer: 5'-CAGAAGGTGGCAGTGTACAGG-3' (SEQ ID NO 3)

TRADD Reverse primer: 5'-CAGCATCTGCAGCACGTC-3' (SEQ ID NO 4)

OAS1 Forward primer: 5'-GGTGGAGTTCGATGTGCTG-3' (SEQ ID NO 5)

OAS1 Reverse primer: 5'-AGG I I I ATAGCCGCCAGTCA-3' (SEQ ID NO 6)

ALDOA Forward primer: 5'-TGCCAGTATGTGACCGAGAA-3' (SEQ ID NO 7)

ALDOA Reverse primer: 5'-GCCTTCCAGGTAGATGTGGT-3' (SEQ ID NO 8)

CD79b Forward primer: 5'-GGACAGAGCGGTGACCAT-3' (SEQ ID NO 9)

CD79b Reverse primer: 5'-GATCTGGCTGCTGGTACTGG-3' (SEQ ID NO 10)

DHRS3 Forward primer: 5'-CACACCAGCACCGAGATG-3' (SEQ ID NO 1 1 )

DHRS3 Reverse primer: 5'-TTCAGTGGGGGAAAGAGGTT-3' (SEQ ID NO 12)

TNF Forward primer: 5'-CAGCCTCTTCTCCTTCCTGAT-3' (SEQ ID NO 13)

TNF Reverse primer: 5'-GCCAGAGGGCTGATTAGAGA-3' (SEQ ID NO 14)

TNFAIP3 Fw primer: 5'-TGCACACTGTG I I I CATCGAG-3' (SEQ ID NO 15)

TNFAIP3 Rv primer: 5'-ACGCTGTGGGACTGAC I I I C-3' (SEQ ID NO 16)

OX40 Forward primer: 5'-CCTGCACGTGGTGTAACCT-3' (SEQ ID NO 17)

OX40 Reverse primer: 5'-AGCGGCAGACTGTGTCCT-3' (SEQ ID NO 18)

Beta-2 microglobulin Fw: 5'-TTCTGGCCTGGAGGCTAT-3' (SEQ ID NO 19)

Beta-2 microglobulin Rv: 5'-TCAGGAAATTTGACTTTCCATTC-3'(SEQ ID NO 20) Serum TNF levels in HIV patients

Patient serums were analyzed for TNF serum levels by ELISA (e-Bioscience) according to the manufacturer's instructions.

Application of RNA-finqerprints to the in vivo signature of HIV-patients

To access evidence for an in vivo contribution of RNA-fingerprints three different methods were used. First, association between significantly changed RNA- fingerprint transcripts of lllumina gene expression data from HIV ^|0W /PD-1 ^|0W vs. HIV ^hl /PD-1 ^hl patients were modeled and tested via x ² -statistic (fingerprint enrichment analysis). Therefore, the proportion of significant RNA-fingerprint transcripts (fold change > 2, p-Value < 0.05, Student's i-test) in the fingerprint were compared to the overall proportion and tested in a contingency table. Furthermore, 1000 random fingerprints comprising the same number of transcripts as each individual RNA-fingerprint were generated and a p-value for each random RNA-fingerprint was calculated in a sampling strategy. Second, leave-one-out supervised classification using support vector machines was performed {group prediction analysis). Prediction probability of a single sample to be classified as either HIV ^|0W /PD-1 ^|0W or HIV ^hi /PD-1 ^hi is reported for each individual RNA-fingerprint. Third, Gene Set Enrichment Analysis was used to determine whether RNA fingerprints were statistically significant between HIV ^|0W /PD-1 ^|0W or HIV ^hi /PD-1 ^hi patients.

One additional publicly available dataset reporting HIV ⁺ infections in human CD4 ⁺ T-cells (Sedaghat et al., 2008) was obtained from the GEO database (GSE9927) and RNA-fingerprint activation was tested as described above.

Gene ontology analysis

We developed an algorithm based on gene ontology analysis, which takes the approach of gene class testing one step further in order to detect key players within the identified gene classes. The algorithm will be described in more detail elsewhere (Maisel et al., manuscript in preparation). Here, we used it to find genes, which play a central role in the differentiation of HIV ^|0W /PD-1 ^|0W vs. HIV ^hl /PD-1 ^hl samples. Briefly, in a first step, gene classes based on Gene Ontology (GO) terms are used to calculate pairwise centroid distances (Euclidean distance) between the two groups. To obtain a statistically significant measure of the calculated distances, class permutations followed by recalculation of the distances are performed. This first step of finding significant GO terms can be substituted by any other gene class testing approach like Gene Set Enrichment Analysis. In a second step, the list of significant GO terms is further analyzed by constructing a network of contributing genes. Here, all genes composed within the significant GO terms are extracted and an edge is drawn between gene x and gene y if x and y belong to at least one of the GO terms. By increasing the number of shared GO terms the network gets more and more specific towards genes which can be found in several GO terms from the list and are therefore key players within the identified gene classes. For network visualization, Cytoscape 2.2 was utilized. The number of shared GO terms was chosen as the maximal value to obtain a sufficiently connected network.

IPA (Ingenuity Pathway Analysis) upstream regulator analysis To perform upstream regulator analysis in IPA, differentially expressed genes between HIV ^|0W /PD-1 ^|0W and HIV ^hi /PD-1 ^hi patients were calculated using a fold- change of +/- 2 and a p-value of 0.05 as the cutoff criteria. Fold changes and p- values of the respective genes were uploaded to IPA and the upstream regulator analysis was performed. Details on the method can be found at www.ingenuity.com.

Assessment of TNF-dependent PD-1 expression in HIV patients

PBMC of HIV ^|0W /PD-1 ^|0W or HIV ^hi /PD-1 ^hi patients were stimulated with plate-bound CD3 (OKT3, 0.5 Mg/ml). To analyze TNF-dependent PD-1 expression, samples were incubated including infliximab (10 μg/ml) for 48 hours before determining PD- 1 expression on CD4 ⁺ T cells by flow cytometry. Additionally, OX40 expression was analyzed serving as a known TNF-dependent control. Cloning of PD-1 constructs with potential NF-κΒ binding regions.

The corresponding PD-1 genomic regions were amplified by PCR using human genomic DNA as source material. The full-length NF-κΒ binding site constructs were amplified with the primers listed below. After digestion with Acc65 I and Hind III the fragments were cloned into the pGL4.24 vector with a minP element upstream of the potential binding motif and a destabilized downstream Firefly luciferase (Promega). Examples for possible primers for cloning of the human PD- 1 genomic regions are enlisted in the following:

-5kb promoter Forward primer

5'-GCAGCGGTACCTGTACACACTGGGAGATTT-3' (SEQ ID NO: 21 )

-5kb promoter Reverse primer

5'-GCTTGAAGCTTATCACTGTTTCATTCCAGC-3' (SEQ ID NO: 22)

-0.5kb CNS Forward primer

5'-GCAGCGGTACCCAGCAGGGGCAGAGGCTG-3' (SEQ ID NO: 23)

-0.5kb CNS Reverse primer

5'-GCTTGAAGCTTGTCCCAGGTCAGGTTGAAG-3' (SEQ ID NO: 24)

Intronic enhancer Forward primer

5'-GCAGCGGTACCCCCAGGCAGCAACCTCAA-3' (SEQ ID NO: 25) Intronic enhancer Reverse primer

5'-GCTTGAAGCTTATGACCAAGCCCACCCCA-3' (SEQ ID NO: 26) Luciferase assays.

Human embryonic kidney (HEK) 293T (ATCC CRL-1 1268) cells were maintained in DMEM containing 10% heat-inactivated fetal bovine serum. To assess regulation of PD-1 expression by signaling via TNF mediated by binding of N F-KB to the genomic PD-1 - locus, constructs containing the three potential NF-KB binding regions were transfected separately into HEK293T cells in 96-well plates together with a plasmid encoding renilla luciferase for normalization. To assess TNFR-signaling transfected cells were stimulated with TNF. Lysis and analysis were performed 24 h post-transfection using the Promega Dual Luciferase Kit as described previously (Beyer et al., 201 1 ). Luciferase activity was counted on a MicrolumatPlus LB 96V plate reader (Berthold).

Mice

C57BL/6J, Thy1 .1 , Tnf ^-/- and Tnfrsf1a ^-/- Tnfrsf1b ^-/- (TNFR1 -TNFR2-deficient) mice were maintained under specific pathogen-free conditions in single ventilated cages in the animal facility of the University of Bonn or Dusseldorf. All mice used in this study were maintained on the C57BL/6 genetic background. TNF-deficient mice were purchased from the Jackson Laboratory. Health status of animals was checked daily and after observations of severe disease symptoms animals were sacrificed and counted as dead. All animal experiments were done in accordance with German legislation governing animal studies and the Principles of Laboratory Animal Care guidelines (US National Institutes of Health publication 85-23, 1996 revision).

Chronic and acute LCMV infections

To establish a C57BL/6J LCMV carrier colony, newborn mice (1 day old) were infected with 1 x 10 ⁶ PFU of LCMV (WE strain). The resultant carrier mice were then interbred to establish new carrier mice through vertical transmission (i.e., passage of virus from mother to offspring). Chronic infection with LCMV in CD90.1 mice was induced by infection of newborn mice with 1 x 10 ⁶ plaque-forming units of LCMV (WE strain). At 8 weeks after infection, mice were used for experiments. Acute infections were established by infecting mice with 2 x 10 ⁴ plaque-forming units of LCMV WE. LCMV Docile strain was a generous gift from Rolf M. Zinkernagel and was propagated in L929 cells as described (Lang et al., 2012). For LCMV Docile infections, mice were infected i.v. with 2 x 10 ³ plaque-forming units of LCMV Docile. For adoptive transfer of CD47CD8 ⁺ T cells into chronically LCMV-infected animals, T cells were obtained from spleens at day 8 after acute infection with LCMV WE (2 x 10 ⁴ plaque-forming units).

Titers of virus for the chronic vertical transmission LCMV model were determined by plaque assay on Vera cells as previously described (Ahmed et al., 1984). For the LCMV Docile model, virus titers were measured using a plaque forming assay as previously described (Lang et al., 2012; Lang et al., 2013). Briefly, organs were harvested into HBSS and homogenized using a Tissue Lyzer (Qiagen). MC57 cells (0.8 x 10 ⁶ /ml) were added to 10 fold serial dilutions of virus samples on 24 well plates. After 3 hrs, overlay medium containing 1 % methylcellulose was added. After 48 hrs, plates were fixed (4% formalin), permeabilized (1 % Triton X HBSS), and stained with anti-LCMV-NP antibody (clone: VL-4) and Peroxidase anti rat antibody. Plaques were developed using color reaction solution containing OPD, 50mM Na ₂ HPO and 25mM citric acid.

Infliximab treatment of mice chronically infected with LCMV

To block TNFR-signaling in mice chronically infected with LCMV, infliximab (Janssen Biotech, Germany) or PBS was administered intraperitoneally each second day for 10 days (25 mg/kg). Mice were sacrificed on day 10 and CD4 and CD8 T cells were assessed.

Adoptive transfer of CD4 ⁺ and CD8 ⁺ T cells

To assess the role of TNFR-signaling on the induction of PD-1 expression in CD4 ⁺ T cells and gp33-specific CD8 ⁺ T cells, 2 x 10 ⁶ CD4 ⁺ and 2 x 10 ⁶ CD8 ⁺ T cells from Thy1 .1 congenic wild-type or TNFRI/II ^-/- mice 10 days after acute infection with LCMV were isolated as described above and transferred to Thy1 .2 congenic mice chronically infected with LCMV. After 10 days mice were sacrificed and assessed.

Immunoblot analysis.

Cell lysates from purified murine CD4 ⁺ and CD8 ⁺ T cells from control- or infliximab- treated LCMV-infected mice were prepared as previously described (Abdullah et al., 2012) followed by immunoblotting with murine plkkb (Ser176/180, Cell Signaling) as well as anti β-tubulin polyclonal antibody (Licor Bioscience) as loading control.

Serum alanine aminotransferase (ALT) determination

Serum ALT was analyzed from whole blood using ALT strips from Roche according to the manufacturer's instructions. Measurement was performed in a Reflovet machine from SCIL animal care. Values above assay range were diluted prior to measurement.

Serum cytokine analysis

Mouse serum was analyzed for TNF cytokine secretion by ELISA (e-Bioscience) according to the manufacturer's instructions.

Whole-genome gene expression in murine cells and bioinformatic analysis

Prior to array based gene expression profiling total RNA was further purified using the MinElute Reaction Cleanup Kit (Qiagen). Biotin labeled cRNA was generated using the TargetAmp Nano-g Biotin-aRNA Labeling Kit for the lllunnina System (Epicentre). Biotin labeled cRNA (1 .5 μg) was hybridized to MouseWG-6 v2.0 Beadchips (lllunnina) and scanned on an lllunnina iScan system. Raw intensity data were processed and exported with BeadStudio 3.1 .1 .0 (lllunnina). Subsequent analyses were performed using Partek Genomics Suite V6.6 (PGS) (Downey, 2006). Non-normalized data were imported from BeadStudio using the default PGS report builder. Data were then quantile normalized and transcripts with variable expression within the dataset as well as differentially expressed (DE) genes between the different conditions were calculated using two- and three-way ANOVA models including batch correction. If not stated otherwise, differentially expressed genes were defined by a fold change (FC) greater 2 and an unadjusted p-value<0.05. For visualization of sample relationships we applied principle component analysis (PCA) using all transcripts and default settings in PGS. Hierarchical clustering with default settings was used in PGS to visualize the transcripts with most significant variance with the dataset. Microarray data can be accessed under GSE52185.

Gene set enrichment analysis

Gene Set Enrichment Analysis (GSEA; http://www.broadinstitute.org/ gsea/index.jsp) was used to determine whether a defined set of genes is statistically significant between two different states (Subramanian et al., 2005). Gene sets for comparison of murine cells are based on a TNF RNA-fingerprint published by Ruan et al. (Ruan et al., 2003) and association with exhaustion on profiles of murine CD8 ⁺ T cells (Wherry et al., 2007). Cross annotation of the different datasets was performed based on gene symbols. qPCR of murine T cells.

Total RNA was extracted with TRIZOL reagent from flow-sorted T cells. cDNA was synthesized with the Transcriptor First Strand cDNA synthesis kit (Roche Diagnostics). qPCR was performed for Ly6c1 , Klrdl , and β-actin using the LightCycler Taqman master kit and the Universal Probe Library assay (Roche Diagnostics) on a LightCycler 480 II (Roche Diagnostics). Results were normalized to β-actin expression. Examples for possible primers applicable for qPCR of the aforementioned biomolecules are enlisted in the following:

Ly6c1 Forward primer: 5'-TCTTGTGGCCCTACTGTGTG-3' (SEQ ID NO: 27) Ly6c1 Reverse primer: 5'-GCAATGCAGAATCCATCAGA-3' (SEQ ID NO: 28) Klrdl Forward primer: 5'-GGATTGGAATGCATTATAGTGAAAA-3'

(SEQ ID NO: 29)

Klrdl Reverse primer: 5'-TGCTCTGGCCTGATAACTGAG-3' (SEQ ID NO: 30) b-actin Forward primer: 5'-CTAAGGCCAACCGTGAAAAG-3' (SEQ ID NO: 31 ) b-actin Reverse primer: 5'-ACCAGAGGCATACAGGGACA-3' (SEQ ID NO: 32) Statistical analysis.

All statistical analysis except analysis of gene expression data were performed with SPSS 21 .0 software.

Results

Exhausted phenotype of CD4+ T cells from HIV infected individuals with high viral load

We assessed the role of inhibitory signaling on CD4+ T cell exhaustion in a large cohort of patients suffering from HIV infection. In this patient cohort (n=1304) (Table S1 ), we first identified HIV patients with high viral titers and correlated this to PD-1 expression levels on CD4+ T cells (Figure 1A,B and Figure S1A,B). The majority of patients were successfully treated with highly active antiretroviral therapy resulting in plasma viral load below detection level and absence of PD-1 expression in CD4+ T cells. However, in 37 patients with high viral load, significantly increased expression of PD-1 (HIV ^hi /PD-1 ^hi ) on CD4+ but also CD8+ T cells was observed (Figure 1A,B and Figure S1A-E). We compared this group to 26 HIV- infected individuals with undetectable viral load and low PD-1 expression (HIV ^|0W /PD-1 ^|0W ). As expected, PD-1 expression on CD4+ T cells from HIV-infected patients correlated with reduced CD4+ T cell counts and increased viral load while CD4+ T cell numbers were inversely correlated to viral load (Figure S1 F-H). In patients with high viral load, CTLA-4 expression was increased in CD4+ and CD8+ T cells (Figure S1 l-N) consistent with an exhausted T cell phenotype. Table S1. Clinical characteristics of the analyzed patients.

Multiple inhibitory signaling pathways besides PD-1 are present in CD4+ T cell exhaustion during HIV infection

As HIV infection is associated with exhausted CD4+ T cell responses we investigated which inhibitory pathways operate in CD4+ T cells during chronic HIV infection. To investigate the relevance of these inhibitory pathways in an unbiased fashion, we generated genome-wide transcriptional profiles of CD4+ T cells from 20 HIV-patients, i.e. 10 patients with high virus titer (HIV ^hi /PD-1 ^hi ) and 10 patients with low virus titer (HIV ^|0W /PD-1 ^|0W ). Then, we established RNA-fingerprints for five inhibitory signaling pathways (PD-1 , CTLA-4, PGE ₂ , IL-10, and TGF-beta1 ) under defined in vitro conditions in CD4+ T cells from healthy individuals. This allowed us to analyze the contribution of particular inhibitory pathways (RNA-fingerprints) to the transcriptional profile of CD4+ T cells from HIV patients, an approach successfully applied for identification of inhibitory mechanisms in tumor-infiltrating CD4+ T cells in Hodgkin's lymphoma (Chemnitz et al., 2007). RNA-fingerprint enrichment analysis as well as gene set enrichment analysis (GSEA) (Subramanian et al., 2005) revealed that signaling from PD-1 , CTLA-4, PGE ₂ , IL- 10, and TGF-beta1 stimulation were significantly enriched in CD4+ T cells from HIV ^hi /PD-1 ^hi patients (Figure 1 C,D, Figure S1 O, and Table S2). We substantiated these findings using a group prediction analysis based on each of the inhibitory RNA-fingerprints to classify each sample as being PD-1 ^high or PD-1 ^|0W . Application of these inhibitory RNA-fingerprints led to accurate classification in most cases with individual mean prediction probabilities for each RNA-fingerprint shown in Fig. 1 E. Utilization of RNA fingerprint analysis to an additional dataset (GSE9927) of transcriptional profiles of CD4+ T cells from chronic HIV- infected individuals vs. uninfected controls (Sedaghat et al., 2008) showed that the inhibitory RNA- fingerprints identified by us were also active in CD4+ T cells from HIV-infected patients in that study. We confirmed the expression of prominent members of the respective RNA-fingerprints in CD4+ T cells from HIV ^hl /PD-1 ^hl patients by quantitative PCR showing significant upregulation of DHRS3 (for PD-1 ), CD79b (for CTLA-4), OAS1 (for PGE ₂ ), and TRADD (for TGF-beta1 ) (Fig. 1 F). Importantly, we found co-expression of all these markers of inhibitory RNA- fingerprints - CD49d for PD-1 , CD79b for CLTA4, IL-4R for PGE ₂ , and CCR4 for TGF-beta1 with the exception of the IL-10 RNA fingerprint that did not include a marker accessible to flow cytometric analysis - in PD-1 +CD4+ T cells from HIV ^hi /PD-1 ^hi patients (Fig. 1 G). Taken together, these data firmly establish that multiple inhibitory pathways were active in CD4+ T cells from HIV ^hl /PD-1 ^hl patients that are likely contributing to the exhausted CD4+ T cell phenotype. Table S2. RNA-fingerprint enrichment analysis: Significant changes of the analyzed fingerprints and controls between HIV ^hl /PD-1 vs. HIV ^|0W /PD-1 ^|0W samples.

Total number of genes included in each RNA-fingerprint or control pathway (symbols used for cross-annotation)

Total number of RNA-fingerprint genes that are significant differentially expressed (p-value <0.05) in the gene expression comparison HIV ^hi /PD-1 ^hi vs. HIV ^low /PD-1 ^low

Total number of probes included on the microarray (lllumina Beadarray WG6v2)

Total number of significantly differentially expressed probes (p-value <0.05) in the gene expression comparison HIV ^h 7PD-1 ^hl vs. HIV ^low /PD-1 ^low

p-value from a one-sided Fisher's exact test of the contingency table derived from (a-d)

Percentage of significant genes from all genes in the RNA-fingerprint

p-value of 1000 random probes comprising the same amount of probes compared to each specific fingerprint

TNF involvement in CD4+ T cell exhaustion in HIV infection

The activity of multiple inhibitory pathways in CD4+ T cells from HIV ^hl /PD-1 ^hl patients in vivo raised the question whether a hierarchical organization of these pathways exists or whether a major molecular driver enforces the complex molecular signature that we found associated with CD4+ T cell exhaustion. To identify potential upstream molecules involved in the induction of these inhibitory processes we developed an algorithm that combines a gene class testing approach with a network construction approach. Applying gene ontology analysis, TNF was found to be the central molecule in the resulting contribution network (Figure 2A). In line with this result, upstream regulator analysis further revealed TNF as a potential key upstream molecule that may explain the observed changes in gene expression between HIV ^hi /PD-1 ^hi and HIV ^|0W /PD-1 ^|0W patients (Figure S2A), suggesting that TNFR-signaling may be the major determinant for development of an exhausted T cell response. To corroborate this network analysis, we established a RNA- fingerprint from TNF-stimulated CD4+ T cells isolated from healthy individuals and observed a substantial enrichment in T cells from HIV ^h '/PD- 1 ^hl patients using RNA-fingerprint enrichment and GSEA analysis (Fig. 2B,C and Table S2). Applying prediction probability based on the TNF RNA-fingerprint from healthy individuals classified HIV ^hl /PD-1 ^hl patients correctly (Fig. 2D). Furthermore, it allowed discrimination between HIV-positive (HIV+) and healthy individuals (HIV- , dataset GSE9927). Using a TNF RNA-fingerprint derived from TNF-treated Jurkat cells (Lewinski et al., 2005), we could further corroborate these findings (Figure S2B-C). This supports the notion that TNF-induced genes may be involved in CD4+ T cell exhaustion in HIV infection. In line with the enrichment in TNF- induced gene expression, we detected increased TNF production in CD4+ T cells from HIV ^hi /PD-1 ^hi patients (Fig. 2E-G). To further investigate the impact of TNF for development of CD4+ T cell exhaustion in HIV patients, we assessed expression levels of molecules downstream of TNFR-signaling by qPCR. We detected increased mRNA expression of OX40, CD74, and TNFAIP3 in CD4+ T cells from HIV ^hi /PD-1 ^hi patients (Fig. 2H). Also at the protein level, we observed co- expression of PD-1 + and OX40 or CD74 in CD4+ T cells from HIV1 1 i/PD-1 hi patients (Fig. 2I,J). Along this line, the TNF serum concentration differed significantly between HIV ^hi /PD-1 ^hi and HIV 10 ^|OW /PD-1 ^|OW patients (Fig. 2K) and correlated with PD-1 expression on CD4+ T cells (Fig. 2L). These results suggest a key role for TNF in the induction of CD4+ T cell exhaustion in HIV infection. Table S3. Experimental conditions for microarray analysis.

Influence of TNF on PD-1 expression in CD4' T cells during HIV infection

NF-KB is one of the most important downstream transcription factors activated by TNFR-signaling. As gene expression analysis strongly suggested that TNF might act as a transcriptional activator for the expression of PD-1 , we performed bioinformatic in silico prediction to detect NF-κΒ binding regions at the genomic PD-1 locus. We identified three potential NF-κΒ binding regions (-5kb upstream of the transcriptional start site (TSS), -0.5 kb upstream of the TSS, and within intron 4) in the genomic locus of PD-1 (Figure S3A). To probe for functional consequences of NF-κΒ binding to the PD-1 locus, we performed luciferase reporter assays for the three predicted NF-κΒ binding regions. NF-κΒ binding regions were cloned in between a minP promoter element and a luciferase reporter gene (Ishihara et al., 201 1 ). TNF stimulation of cells transfected with these reporter-constructs and measurement of luciferase activity revealed that only the region -5 kb upstream of the TSS is responsive to TNF stimulation while the -0.5 kb region has high constitutive activity (Fig. 3A and Figure S3B) indicating that PD-1 expression can be increased by TNF-induced NF-κΒ activation. This led us to reason that PD-1 expression on CD4+ T cells was controlled by TNF in HIV- infected patients. To directly assess the role of TNF on PD-1 expression, we neutralized TNF using the neutralizing anti-TNF antibody, infliximab, in anti-CD3 activated PBMCs and determined PD-1 expression levels. While blockade of TNFR-signaling did not change the expression of PD-1 on CD4+ T cells from HIV ^|0W /PD-1 ^|0W patients, we observed a significant downregulation on CD4+ T cells from HIV ^hi /PD-1 ^hi patients (Fig. 3B,C). Expression of a bona fide target of TNF, i.e. OX40, was downregulated in the presence of infliximab in CD4+ T cells from both patient groups, suggesting that TNFR-signaling was efficiently blocked in both CD4+ T cell populations (Fig. 3D). These results support the assumption that TNFR- signaling plays an important role in PD-1 expression on CD4+ T cells in chronic HIV infection and may thereby contribute to T cell exhaustion. Chronic LCMV infection is associated with high levels of TNF and CD4+ PD-1 + T cells

To gain mechanistical insight into the relevance of TNF for induction of CD4+ T cell exhaustion in chronic viral infection, we employed the chronic murine LCMV infection model (Huang et al., 2013). Similar to HN-infected patients, we observed an increase in CD4+ and CD8+ T cells expressing PD-1 in mice with chronic LCMV infection compared to non-infected mice (Fig. 4A,B). T cell exhaustion was evident in mice with chronic LCMV infection because of the persistently high viral load in the presence of LCMV gp33-specific CD8+ T cells (Fig. 4C,D). Mice with chronic LCMV infection showed signs of intermittent liver damage as determined by periodic increases of serum ALT (Fig. 4E). Notably, we observed increased serum levels of TNF in mice with chronic LCMV infection (Fig. 4F), suggesting that the chronic LCMV infection model is a suitable model to study the influence of TNF on CD4+ T cell exhaustion. Consistent with presence and activity of virus-specific T cells during chronic infection we detected liver damage (Fig. 12b). Indeed, neonatal chronic LCMV infection did not lead to clonal deletion of LCMV-specific CD4+ or CD8+ T cells as these cells were detected in peripheral blood and spleen (Fig. 4I, upper row). Furthermore, LCMV-specific T cells detected in mice with neonatal chronic infection showed high PD-1 expression levels compared to T cells from mice that cleared acute LCMV WE infection (Fig. 4I, below). To assess the similarity between HIV infection and neonatal chronic LCMV infection for Th cells we performed whole genome transcriptome analysis of sorted PD-1 - and PD- 1 + CD4+ T cells from animals with chronic or acute LCMV infection (Fig. 12c). First, we identified a set of 556 differentially expressed genes between PD-1 + and PD-1 - Th cells isolated from mice with neonatal chronic LCMV infection. From these genes we subtracted those genes being differentially expressed between PD-1 + and PD-1 - CD4+ T cells during acute infection. With this analysis, we established a set of 201 genes specific for CD4+ T cells during chronic viral infection that partially overlaps with reported signatures of exhausted Th cells during chronic LCMV clone 13 infection (Fig. 12d). Using the CD4+ T-cell RNA- fingerprint from chronic neonatal LCMV infection for GSEA and RNA fingerprint enrichment analysis in human Th cells from HIV patients clearly showed enrichment of this gene set in human Th cells from HIV ^hl /PD-1 ^hl patients. Furthermore the murine RNA-fingerprint classified HIV patients correctly into HIV ^hl /PD-1 ^hl and HIVIow/PD-1 low groups. It further allowed discrimination between HIV-positive (HIV+) and healthy individuals (HIV-) (Fig. 12e). Taken together, the increased TNF TNFR-signaling impairs CD4+ T-cells in chronic viral infection serum level and the close similarity in the gene expression of CD4+ T cells establish that the neonatal chronic LCMV infection reciprocates the key elements of Th cell changes during HIV infection and that it can be used to study the role of TNF in chronic viral infection.

Chronic exhaustion of CD4+ PD-1 + T cells in LCMV-infected mice is similar to chronic CD4+ T cell exhaustion in HIV

To assess the global transcriptional changes in CD4+PD-I + T cells in comparison to CD4+PD-1 - T cells from LCMV-infected animals we performed whole genome transcriptome analysis of sorted CD4+PD-1 - and CD4+PD-1 + T cells from animals with chronic or acute LCMV infection (Figure S4A). First, we identified a set of 556 differentially expressed genes between CD4+PD-1 + and CD4+PD-1 - T cells in chronic LCMV-infected mice. From these genes we subtracted those being differentially expressed between CD4+PD-1 + and CD4+PD-1 - T cells in acute infection in order to identify genes associated only with CD4+ T cell exhaustion in chronically infected animals. With this analysis, we established a set of 201 genes specific for CD4+ T cell exhaustion during chronic viral infection (Fig. 4G and Table S3). This set of genes distinguishing exhausted CD4+PD-1 + T cells from CD4+PD-1 - T cells in chronic viral infection is distinct from genes differentiating virus-specific exhausted and memory CD8+ T cells (data not shown) (Doering et al., 2012). When assessing expression of this murine CD4+ T cell exhaustion RNA-fingerprint in human CD4+ T cells during HIV infection by GSEA and RNA- fingerprint enrichment analysis, we observed an enrichment of this RNA-fingerprint in CD4+ T cells from HIV ^hi /PD-1 ^hi patients (Fig. 4H and Figure S4B). Furthermore, applying prediction probability based on the murine chronic exhaustion RNA- fingerprint classified HIV patients correctly (Figure S4C) and allowed discrimination between HIV-positive and healthy individuals. Taken together, these data strongly support the notion that the changes in CD4+ T cells in HIV ^hl /PD-1 ^hl patients are mirrored in CD4+PD-I + T cells in chronic LCMV infection.

Relevance of TNFR-signaling in vivo for PD-1 expression and chronicity of viral infection To establish a link between TNFR-signaling and CD4+ T cell exhaustion as suggested by the gene expression analysis in CD4+ T cells from HIV patients and LCMV-infected mice, we characterized the impact of TNF on anti-viral T cell immunity in vivo. First, we infected TNF-deficient and corresponding wild-type mice with LCMV Docile and observed increased numbers of virus-specific CD8+ T cells with highly improved effector function in absence of TNF when compared to wild-type animals (Figure S5A,B). Consequently, viral titers were reduced 8 days after infection (Figure S5C). Consistent with increased CD8+ T cell effector function, TNF-deficient animals succumbed in part during infection with a chronic strain of LCMV (Figure S5D). Next, we neutralized TNF by administration of Infliximab in mice with chronic LCMV (WE) infection through vertical transmission (Fig. 5A). TNF neutralization resulted in reduced percentages ofPD-1 -expressing CD4+ and CD8+ T cells after treatment (Fig. 5B,C) and led to an increase of LCMV gp33+-specific CD8+ T cells (Fig. SD). Activation of NF-κΒ downstream of TNFR-signaling was more prominent in CD4+ compared to CD8+ T cells in mice with chronic LCMV infection as shown by reduced phosphorylation of Ikk- alpha/beta following Infliximab treatment (Fig. 5E). In contrast, in mice with acute LCMV infection TNF-blockade did not lead to reduction of PD- I expression on CD4+ and CD8+ T cells, gp33-specific CD8+ T cells or phosphorylation of Ikk- alpha/beta (Figure S5G-1 ). Upon TNF neutralization serum ALT levels increased suggesting activation of virus- specific immunity that caused immune mediated hepatitis (Fig. 5F). TNF neutralization was also associated with control of chronic viral infection as shown by clearance of LCMV from serum (Fig. 5G), liver (Fig. 5H), and spleen at day 10 after initiation of infliximab treatment (Fig. 5I). These data suggest that TNF during chronic viral infection leads to exhaustion of CD4+ and CD8+ T cells and that blockade of TNF can overcome such exhaustion and trigger activation of antiviral immunity leading eventually to clearance of chronic viral infection. TNF blockade during chronic LCMV infection reverts exhaustion-associated transcriptional changes in murine CD4+ T cells

To assess how blockade of TNFR-signaling by Infliximab treatment alters gene expression of CD4+ T cells we performed whole genome transcriptome analysis of sorted CD4+PD-1 - and CD4+PD-1 + T cells from control- and Infliximab-treated animals with chronic or acute LCMV infection. Principal component analysis (PCA) (Fig. 6A), hierarchical clustering (Fig. 6B), and analysis of fold-changes (Fig 6C,D) demonstrated specific changes in gene expression patterns in CD4+PD-1 + T cells post infliximab treatment. K\rd\ and Ly6c\ , two genes previously reported as hallmarks for regain of effector function for CD8+ T cells (Allie et al., 201 1 ; Lin et al., 2007; McMahon et al., 2002) or linked to TH1 -like memory function in CD4+ T cells (Graham et al., 2007) were among the genes downregulated during chronic exhaustion (see Fig. 4G). Restored expression levels of K\rd\ and Ly6c\ in CD4+PD-1 + T cells were detected by transcriptome analysis after infliximab treatment (Fig. 6D, highlighted) and confirmed by qPCR (Fig. 6E). Next, we asked whether the exhaustion signature identified in CD4+PD-1 + T cells from chronically LCMV-infected animals would be changed after blockade of TNFR-signaling. Indeed GSEA showed enrichment of the exhaustion signature in control -treated CD4+PD-1 + T cells (Control) in comparison to infliximab-treated CD4+PD-1 + T cells (infliximab, Fig. 6F). Using a gene set distinguishing effector from exhausted CD8+ T cells for analysis we can even show by GSEA that the genes associated with reversal of CD8+ T cell exhaustion are enriched in CD4+PD-1 + T cells after Infliximab treatment (Figure S6A). Next, we addressed if a murine TNF RNA- fingerprint established by Ruan et al. (Ruan et al., 2003) is also present in CD4+PD-1 + T cells from mice with chronic LCMV infection. GSEA showed an enrichment of this TNF signature in murine CD4+ T cells (Fig. 6G, left panel). We postulated that TNF-blockade should abrogate the TNF-fingerprint. Indeed, infliximab treatment of mice led to a loss of this particular gene set in CD4+ T cells (Fig. 6G, right panel). As genomic analysis allows cross-sped es comparison, we also assessed if the human TNF RNA-fingerprint is enriched in CD4+PD-1 + T cells in mice with chronic LCMV infection and whether this is changed after Infliximab administration. We confirmed that the human TNF RNA-fingerprint is indeed present in murine CD4+PD-1 + T cells in chronic LCMV-infected animals and that this RNA-fingerprint is lost by Infliximab treatment (Figure S6B). These analyses indicated a similar regulation of TNFR-signaling associated genes in CD4+ T cells in patients with high level HIV viremia and CD4+ T cells in mice with chronic LCMV infection. This raised the question whether TNFR-signaling in CD4+ T cells was causally involved in the induction of immune exhaustion during chronic viral infection.

TNFR-signaling in CD4+ T cells is responsible for exhaustion during chronic viral infection

To directly address the relevance of TNFR-signaling in CD4+ T cells for exhaustion in vivo, we used T cells from TNFRI/II double-deficient mice for adaptive transfer into mice with chronic LCMV infection. Thy1 .2+ TNFRI/1 1 double-deficient or wild-type mice were experimentally infected with LCMV and CD4+ and CD8+ T cells were isolated at d 8 p.i. during resolution of acute infection (Figure S7A,B). Adaptive transfer of these cells into Thy1 .1 + mice with chronic LCMV infection allowed us to follow the fate of Thy1 .2+ TNFRI/II double- deficient T cells (Fig. 7).

At d 10 after adoptive transfer of Thy1 .2+ CD4+ and CD8+ T cells into Thy1 .1 + mice with chronic LCMV infection, Thy1 .2+ TNFR-competent CD4+ T cells showed uniform high levels of PD-1 whereas TNFRI/II double-deficient Thy1 .2+ CD4+ T cells had much lower levels of PD-1 on their surface (Fig. 7B). We also observed a reduction in the size of the PD-1 + population among the endogenous Thy1 .1 + CD4+ T cells (Fig. 7C-E), suggesting that transfer of TNFRI/II double- deficient Thy1 .2+ T cells also helped endogenous T cells to recover from exhaustion. Importantly, numbers of LCMV-gp33-specific Thy1 .2+ CD8+ T cells increased among adoptively transferred TNFRI/II double-deficient T cells (Fig. 7F) and also among endogenous Thy1 .1 + CD8+ T cells (Fig. 7G-I) while PD-1 expressing CD8+ T cells remained constant (Figure S7C-F). As this suggested that blockade of TNFR-signaling might be important for CD8+ T cell effector function to overcome chronic viral infection, we adoptively transferred TNFRI/II double-deficient CD8+ T cells alone without CD4+ T cells. Transfer of TNFRI/II double-deficient CD8+ T cells resulted in an expansion of Thy1 .1 + LCMV-gp33- specific CD8+ T cells (Fig. 7J,K). However, sustained elevation of serum ALT was only observed after transfer of CD4+ together with CD8+ TNFRI/II double-deficient T cells but not after transfer of TNFRI/II double-deficient CD8+ T cells alone or TNFR- competent T cells (Fig. 7L). We then addressed the functional relevance of TNFR-signaling in T cells for control of chronic viral infection. Only transfer of TNFRI/II double-deficient CD4+ and CD8+ T cells led to robust reduction in LCMV serum titers in mice with chronic infection (Fig. 7M). Transfer of TNFRI/II double- deficient CD8+ T cells or TNFR-competent T cells (CD4+/CD8+ or CD8+ T cells) led to a transient antiviral effect that eventually failed to control viral replication beyond d 5 after transfer (Fig. 7M). Taken together, these results identify a crucial role of CD4+ T cell help for CD8+ T cell function in sustained control of chronic viral infection as the transfer of CD8+ T cells from TNFRI/II double-deficient mice led to an expansion of gp33-specific CD8+ T cells but did not result in functional competence of the cells against the virus. Moreover, we identify TNF as a key negative regulator of CD4+ T helper cell function for CD8+ T cell antiviral immunity in chronic infection that induces CD4+ T cell exhaustion, which may serve to limit immunopathology but also may curtail efficient CD8+ T cell responses required for controlling chronic viral infection.

Summary Summarized, it could be demonstrated that the disruption of the immunoregulatory properties of TNF restored protective anti-viral T cell immunity without the need to interfere with any of the downstream regulatory pathways so far associated with T cell exhaustion. The blockade of TNFR-signaling by administration of neutralizing antibodies against TNF in vivo or infection of TNF-deficient mice resulted in restoration of effective anti-viral immunity.

In detail, it was revealed that a lack of TNF or TNFR-signaling during chronic LCMV infection causes increased immunity and even immunopathology, indicating that TNF indeed restrains T cell immunity to protect infected tissues from excessive effector functions. Using LCMV infection as a model for chronic viral infections in humans, the link between chronic viral infection TNF exposure, and T cell exhaustion could be strengthened. It was further found that a lack of PD-1 signaling leads to exacerbated autoimmune liver damage and death during LCMV infection (Iwai et al., 2003). Thus, the above data also show that the simultaneous activity of inhibitory signaling pathways in exhausted T cells might be the result of an auto-regulatory negative feedback loop to prevent development of organ-damaging T cell immunity.

TNF neutralization rescues T-cell function and promotes control of chronic viral infection To investigate the relevance of TNF, we neutralized TNF by administration of anti- TNF antibodies to mice with neonatal chronic LCMV infection (Fig. 10a). Such TNF neutralization led to an increase in LCMV-specific CD4+ and CD8+ T cells as well as a reduction of their PD-1 expression (Fig. 10b-g and Fig. 13a-c). We also observed a decreased PD-1 expression on total CD4+ and CD8+ T cells after TNF neutralization (Fig. 13d,e). In addition, high phosphorylation of ΙΚΚα/β reflecting high NF-KB signaling downstream of the TNFR, was detected in Th and CD8+ T cells in neonatal chronic LCMV infection, which was reduced more prominently in the Th cell compartment after TN F neutralization (Fig. 10f). In contrast, we did not find phosphorylation of ΙΚΚα/β in T cells during acute LCMV infection and TN F neutralization did not lead to reduction of PD-1 expression on total CD4+/CD8+ T cells or LCMV-specific CD8+ T cells (Fig. 14a-e). Since we observed a stronger TNF effect in Th cells we assessed global changes in this cell compartment by transcriptome analysis (Fig. 6A-D). Several approaches including principal component analysis, hierarchical clustering, an ANOVA model to detect differentially expressed genes, and FC/FC plots clearly revealed that the PD1 + Th cells showed more changes then PD1 - Th cells post TNF neutralization. For two hallmark genes associated with low expression in T cells in chronic viral infection, we found restoration of expression after TNF neutralization (Fig. 6E). Moreover, the PD-1 +CD4+ T-cell signature from chronic neonatal LCMV infection was lost after TNF TNFR-signaling impairs CD4+ T-cells in chronic viral infection neutralization (Fig. 6F) as well as the major transcription factors associated with chronic clone 13 infection, indicating that TNFR signaling caused this gene expression signature. Moreover, the increase of TNF signaling in PD-1 + Th cells was also lost (Fig. S6A). Even more, genes downregulated in human CD4+ T cells in response to TNF are elevated in murine PD-1 +CD4+ T cells after TNF neutralization (Fig. S6B) indicating similar regulation of TNF target genes in human and murine chronic viral infection. Upon TNF neutralization in mice chronic LCMV infection serum ALT levels increased suggesting activation of virus-specific T-cell immunity that caused immune mediated hepatitis (Fig. 10i). Importantly, TNF neutralization was followed by control of chronic viral infection as shown by clearance of LCMV from serum (Fig. 10j), liver (Fig. 10k), and spleen within ten days (Fig. 101). Although we did not detect increased TNF serum levels in LCMV clone 13 infected mice (see Fig. 4a), we observed a small increase in virus- specific T cells with less PD1 expression after TNF neutralization. However, this did not result in control of LCMV infection (Fig. 15a-n). Taken together, our data support the notion that TNF neutralization rescues the antiviral T-cell response during chronic LCMV infection that is accompanied by increased TNF serum levels. Next, we investigated whether TNF neutralization affected the functionality of virus-specific T cells. We found more LCMV-specific CD8+ T cells after TNF neutralization in mice with neonatal chronic LCMV infection that produced IFNy and/or IL-2 after re-stimulation (Fig. 10m, n) consistent with a gain of T-cell functionality following inhibition of TNF. Furthermore, the expanded population of LCMV-specific CD4+ T cells showed more helper function as evidenced by increased numbers of IL-21 , IL-2, IFNy and CD40Lexpressing CD4+ T cells (Fig. 10o-r), which could also be observed after TNF neutralization in TNFR-signaling impairs CD4+ T-cells in chronic viral infection clone 13 infection (Fig. 15o-t). Together these data indicate that inhibition of TNF improves both functionality of virus-specific Th cells as well as CD8+ T cells.

TNFR-signaling in CD4+ T cells curtails anti-viral immunity

So far, we established a beneficial role of TNF neutralization on both Th cells and CD8+ T cells. We next wanted to better understand the contribution of both cell compartments. As a first step, we determined the magnitude of TNF signaling on both compartments. Interestingly, a previously described murine TNF signature 44 was only significantly enriched in PD-1 +CD4+ T cells but not PD-1 +CD8+ T cells during neonatal chronic LCMV infection (Fig. 1 1 a). To address the relevance of TNFR-signaling for T-cell immunity in vivo, we next isolated CD8+ T cells from Thy1 .2+ TNFRI/II knockout mice or wild type mice at d8 after infection with LCMV- WE, i.e. during resolution of acute infection, for adoptive transfer into Thy1 .1 + mice with chronic LCMV infection (Fig. 1 1 b and 16a,b). After adoptive transfer, we observed no change in transferred virus-specific Thy1 .2+ TNFRI/l l-deficient CD8+ T cells compared to wild type CD8+ T cells (Fig. 1 1 c) while PD-1 expression was reduced (Fig. 1 1 d). Notwithstanding the development of increased immune- mediated hepatitis and an initial drop in viral titers, Thy1 .2+ TNFRI/l l-deficient virus-specific CD8+ T cells failed to control viremia in mice with neonatal chronic LCMV infection (Fig. 1 1 e,f) suggesting that Th cells are critically involved in beneficial effects post TNF neutralization. To determine the influence of Th cells we transferred CD4+ T cells together with CD8+ T cells that were isolated from TNFRI/l l-deficient or wildtype mice treated as described above. At d10 after adoptive transfer we detected an increase in virus-specific TNFRI/l l-deficient CD4+ T cells (Fig. 1 1 g). These virus-specific TNFRI/l l-deficient CD4+ T cells expressed lower PD-1 levels compared to TNFRI/l l-competent cells (Fig. 1 1 h). Importantly, TNFR-signaling impairs CD4+ T-cells in chronic viral infection adoptively transferred virus-specific CD8+ T cells were increased and showed a reduction in PD-1 expression levels (Fig. 1 1 i,j). We also observed an increase in endogenous virus-specific CD8+ T cells and a decrease in their PD-1 expression levels after transfer of TNFRI/l l-deficient but not TNFRI/l l-competent CD4+ and CD8+ T cells (Fig. 1 1 k-m) suggesting that adoptively transferred TNFRI/l l-deficient T cells helped to sustain anti-viral immunity from endogenous TNFRI/l l-competent CD8+ T cells. In accordance with improved anti-viral activity we detected an increased immune-mediated hepatitis after transfer of TNFRI/l l-deficient CD4+ and CD8+ T cells (Fig. 1 1 n). Importantly, transfer of TNFRI/l l-deficient T cells was accompanied by sustained reduction in LCMV serum titers (Fig. 1 1 o). These reduced viral titers together with the reduced levels of TNF (Fig. 1 1 p) might also explain further observations demonstrating increased levels of endogenous virus- specific Th cells (Fig. 16c,d) with reduced PD-1 expression (Fig. 16e), reduced overall PD-1 expression on transferred Thy1 .1 + and endogenous Thy1 .2+ Th cells (Fig. 16f-h) as well as transferred Thy1 .1 + and endogenous Thy1 .2+ CD8+ T cells (Fig. 16i-k). In line with these findings, we could detect similar results after transfer of Th cells and CD8+ T cells from TNFRI/l l-deficient animals into clone 13 infected mice (data not shown). Taken together, these results demonstrate that prevention of TNFRI/ll-signaling in T cells reinstalls impaired anti-viral T-cell responses during chronic viral infection.

References

Abdullah, Z., M. Schlee, S. Roth, M. A. Mraheil, W. Barchet, J. Bottcher, T. Hain,

S. Geiger, Y. Hayakawa, J. H. Fritz, F. Civril, K. P. Hopfner, C. Kurts, J. Ruland, G.

Hartmann, T. Chakraborty and P. A. Knolle (2012). RIG-I detects infection with live Listeria by sensing secreted bacterial nucleic acids. EMBO J 31 (21 ): 4153-4164.

Ahmed, R., A. Salmi, L. D. Butler, J. M. Chiller and M. B. Oldstone (1984).

Selection of genetic variants of lymphocytic choriomeningitis virus in spleens of persistently infected mice. Role in suppression of cytotoxic T lymphocyte response and viral persistence. J Exp Med 160(2): 521 -540.

Allie, S.R., Zhang, W., Fuse, S., and Usherwood, E.J. (201 1 ). Programmed death

1 regulates development of central memory CD8 T cells after acute viral infection.

J Immunol 186, 6280-6286.

Barber, D.L., Wherry, E.J., Masopust, D., Zhu, B., Allison, J. P., Sharpe, A.H., Freeman, G.J., and Ahmed, R. (2006). Restoring function in exhausted CD8 T cells during chronic viral infection. Nature 439, 682-687.

Beyer, M., Y. Thabet, R. U. Muller, T. Sadlon, S. Classen, K. Lahl, S. Basu, X. Zhou, S. L. Bailey-Bucktrout, W. Krebs, E. A. Schonfeld, J. Bottcher, T. Golovina, C. T. Mayer, A. Hofmann, D. Sommer, S. Debey-Pascher, E. Endl, A. Limmer, K. L. Hippen, B. R. Blazar, R. Balderas, T. Quast, A. Waha, G. Mayer, M. Famulok, P. A. Knolle, C. Wickenhauser, W. Kolanus, B. Schermer, J. A. Bluestone, S. C. Barry, T. Sparwasser, J. L. Riley and J. L. Schultze (201 1 ). Repression of the genome organizer SATB1 in regulatory T cells is required for suppressive function and inhibition of effector differentiation. Nat Immunol 12(9): 898-907.

Brooks, D.G., Teyton, L, Oldstone, M.B., and McGavern, D.B. (2005). Intrinsic functional dysregulation of CD4 T cells occurs rapidly following persistent viral infection. Journal of virology 79, 10514-10527.

Chemnitz, J.M., Eggle, D., Driesen, J., Classen, S., Riley, J.L., Debey-Pascher, S., Beyer, M., Popov, A., Zander, T., and Schultze, J.L. (2007). RNA fingerprints provide direct evidence for the inhibitory role of TGFbeta and PD-1 on CD4+ T cells in Hodgkin lymphoma. Blood 1 10, 3226-3233.

Cohen, M.S., Shaw, G.M., McMichael, A.J., and Haynes, B.F. (201 1 ). Acute HIV-1 Infection. The New Englandjournal ofmedicine 364, 1943-1954.

D'Souza, M., Fontenot, A.P., Mack, D.G., Lozupone, C, Dillon, S., Meditz, A., Wilson, C.C., Connick, E., and Palmer, B.E. (2007). Programmed death 1 expression on HIV-specific CD4+ T cells is driven by viral replication and associated with T cell dysfunction. J Immunol 179, 1979-1987.

Deeks, S.G., Tracy, R., and Douek, D.C. (2013). Systemic Effects of Inflammation on Health during Chronic HIV Infection. Immunity 39, 633-645.

Doering, T.A., Crawford, A., Angelosanto, J.M., Paley, M.A., Ziegler, C.G., and Wherry, E.J. (2012). Network analysis reveals centrally connected genes and pathways involved in CD8+ T cell exhaustion versus memory. Immunity 3 7, 1 130- 1 144.

Downey, T. (2006). Analysis of a multifactor microarray study using Partek Genomics Solution. Method Enzymol 41 1 , 256-270.

Graham, CM., Christensen, J.R., and Thomas, D.B. (2007). Differential induction of CD94 and NKG2 in CD4 helper T cells. A consequence of influenza virus infection and interferon- gamma? Immunology 121 ,238-247.

Huang, L.R., Wohlleber, D., Reisinger, F., Jenne, C.N., Cheng, R.L., Abdullah, Z., Schildberg, F.A., Odenthal, M., Dienes, H.P., van Rooijen, N., et al. (2013).

Intrahepatic myeloid-cell aggregates enable local proliferation of CD8(+) T cells and successful immunotherapy against chronic viralliver infection. Nature immunology 14, 574-583.

Iwai, Y., Terawaki, S., Ikegawa, M., Okazaki, T., and Honjo, T. (2003). PD-1 inhibits antiviral immunity at the effector phase in the liver. The Journal of experimental medicine

198, 39-50.

Lang, P. A., K. S. Lang, H. C. Xu, M. Grusdat, I. A. Parish, M. Recher, A. R. Elford, S. Dhanji, N. Shaabani, C. W. Tran, D. Dissanayake, R. Rahbar, M. Ghazarian, A. Brustle, J. Fine, P. Chen, C. T. Weaver, C. Klose, A. Diefenbach, D. Haussinger, J. R. Carlyle, S. M. Kaech, T. W. Mak and P. S. Ohashi (2012). Natural killer cell activation enhances immune pathology and promotes chronic infection by limiting CD8+ T-cell immunity. Proc Natl Acad Sci U S A 109(4): 1210-1215.

Lang, P. A., H. C. Xu, M. Grusdat, D. R. Mcllwain, A. A. Pandyra, I. S. Harris, N. Shaabani, N. Honke, S. K. Maney, E. Lang, V. I. Pozdeev, M. Recher, B. Odermatt, D. Brenner, D. Haussinger, P. S. Ohashi, H. Hengartner, R. M. Zinkernagel, T. W. Mak and K. S. Lang (2013). Reactive oxygen species delay control of lymphocytic choriomeningitis virus. Cell Death Differ 20(4): 649-658. Lewinski, M.K., Bisgrove, D., Shinn, P., Chen, H., Hoffmann, C, Hannenhalli, S., Verdin, E., Berry, C.C., Ecker, J.R., and Bushman, F.D. (2005). Genome-wide analysis of chromosomal features repressing human immunodeficiency virus transcription. J Virol 79, 6610-6619. Lin, S.J., Peacock, CD., Bahl, K., and Welsh, R.M. (2007). Programmed death-1 (PD-1 ) defines a transient and dysfunctional oligoclonal T cell population in acute homeostatic proliferation. The Journal of experimental medicine 204, 2321 -2333. McMahon, C.W., Zajac, A.J., Jamieson, A.M., Corral, L, Hammer, G.E., Ahmed, R., and Raulet, D.H. (2002). Viral and bacterial infections induce expression of multiple NK cell receptors in responding CD8(+) T cells. J Immunol 169, 1444- 1452.

Prokunina, L, C. Castillejo-Lopez, F. Oberg, I. Gunnarsson, L. Berg, V.

Magnusson, A. J. Brookes, D. Tentler, H. Kristjansdottir, G. Grondal, A. I. Bolstad, E. Svenungsson, I. Lundberg, G. Sturfelt, A. Jonssen, L. Truedsson, G. Lima, J.

Alcocer-Varela, R. Jonsson, U. B. Gyllensten, J. B. Harley, D. Alarcon-Segovia, K.

Steinsson and M. E. Alarcon-Riquelme (2002). A regulatory polymorphism in

PDCD1 is associated with susceptibility to systemic lupus erythematosus in humans. Nature Genetics 32(4): 666-669.

Ruan, H., Pownall, H.J., and Lodish, H.F. (2003). Troglitazone Antagonizes Tumor

Necrosis Factor-a-induced Reprogramming of Adipocyte Gene Expression by

Inhibiting the Transcriptional Regulatory Functions of NF-KB. Journal of Biological

Chemistry 278, 28181 -28192.

Sedaghat, A.R., German, J., Teslovich, T.M., Cofrancesco, J., Jr., Jie, C.C., Talbot, C.C., Jr., and Siliciano, R.F. (2008). Chronic CD4+ T cell activation and depletion in human immunodeficiency virus type 1 infection: type I interferon- mediated disruption of T cell dynamics. J Virol 82, 1870-1883.

Streeck, H., Brumme, Z.L., Anastario, M., Cohen, K.W., Jolin, J.S., Meier, A., Brumme, C.J..Rosenberg, E.S., Alter, G., Allen, T.M., el al. (2008). Antigen load and viral sequence diversification determine the functional profile ofH IV- 1 -specific CD8+ T cells. PLoS Med 5, e100.

Subramanian, A., Tamayo, P., Mootha, V.K., Mukheijee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., et al. (2005). Gene set enrichment analysis: a knowledge-based approach for interpreting genome- wide expression profiles. Proc Natl Acad Sei USA 102, 15545-15550.

Wherry, E.J. (201 1 ). T cell exhaustion. Nature immunology 12, 492-499.

Wherry, E.J., Ha, S.J., Kaeeh, S.M., Haining, W.N., Sarkar, S., Kalia, V., Subramaniam, S., Blattman, J.N., Barber, D.L., and Ahmed, R. (2007). Molecular signature of CD8+ T cell exhaustion during chronic viral infection. Immunity 2 7, 670-684.