The concept that stressors can have a negative impact on immunity and ultimately induce a reduction in host resistance to infectious pathogens and even to tumour progression is not new. As a followup of the early studies of Hans Selye on stress, several scientists demonstrated in the 1950s and 1960s that laboratory rodents exposed to various stressors, including inescapable painful electric shocks, displayed an altered resistance to viral, bacterial, and parasitic infections. These effects of stressors were accompanied by decreases in antibody responses to the specific microbial pathogen under study. In view of the already demonstrated immunosuppressive effects of glucocorticoids and the pivotal role of glucocorticoids in the stress response, pathophysiological mechanisms of the immunosuppressive effects of stressors were easy to determine. However, it was already clear at that time that mechanisms of the effects of stressors on immunity were not that simple since decreases as well as increases in immunity could be observed, depending on the immune response under study, the type of stressor and the time point at which the stressor took place during the mounting of the immune response. Furthermore, administration of glucocorticoids at physiological instead of pharmacological doses had little effect on some aspects of immunity and evolution of the disease process.

There was little innovation in this field until the 1980s when a few pioneer immunologists and neuroscientists decided to work together in order to understand how the central nervous system communicates with the immune system. One impetus for this was the demonstration that immune responses can be submitted to Pavlovian conditioning in apparently the same way as the salivary response, as discovered by Robert Ader at the University of Rochester.⁽¹⁾ Mice exposed to a new taste paired with an immunosuppressive agent such as cyclophosphamide during the development of an antibody response were found to display further decreases in antibody titres when re-exposed to the taste alone in
the absence of any immunomodulating agent. For this to occur, there must be pathways of communication from the brain to the immune system that are activated by the taste paired with cyclophosphamide. The search for these pathways of communication resulted in the demonstration of innervation of the primary (thymus) and secondary (spleen and lymphoid nodes) lymphoid organs by the sympathetic nervous system.⁽²⁾ Sympathetic efferent nerves enter lymphoid organs with the vasculature but ultimately separate from blood vessels to innervate the parenchyma, where both B and T lymphocytes reside and can proliferate. Sympathetic fibres innervating lymphoid organs contain all the neurotransmitter machinery of other sympathetic neurones, including noradrenaline and neuropeptides such as substance P, neuropeptide Y (NPY), and calcitonin gene-related peptide (CGRP). This implies that the chemical composition of the microenvironment in which lymphocytes are present ultimately depends on activity of the autonomic nervous system. These findings gained in prominence when it was discovered that specific subsets of leukocytes have receptors for these neuronal communication signals.

In addition to cytoplasmic receptors that bind steroid hormones including glucocorticoids and sex hormones, lymphocytes, and other cells of the immune system have been found to have membrane receptors that bind and respond to most neurotransmitters and neuropeptides and are quasi-identical to brain neurotransmitter and neuropeptide receptors. As supported by an important body of literature, activation of these receptors in leukocytes has functional consequences on immune responses whether immunity is measured in vivo or in vitro.⁽³⁾ As an typical example, growth hormone (GH), a pituitary hormone known for its growthpromoting activity and with no known immune function, was shown to restore the resistance of hypophysectomized rats to an infection with Salmonella typhimurium, with an efficacy comparable to that of a tetracycline antibiotic or the macrophage-stimulating factor interferon-gamma (IFN-γ).^(4,5) These results obtained in vivo were replicated in vitro.⁽⁶⁾ GH activated highly purified populations of pulmonary macrophages in the same way as IFN-γ. Both factors were able to prime macrophages triggered with opsonized zymosan to secrete superoxide anion O^–₂, an index of macrophage activation, even if GH was less active in this system than IFN-γ. Antibody blocking studies demonstrated that the priming activity of GH was independent of IFN-γ, and vice versa the activity of IFN-γ was distinct from that of GH. This priming had functional consequences since both IFN-γ and GH increased the capability of macrophages to kill Pasteurella multocida. Since most of the effects of GH on its target cells are mediated by the local production of insulin-growth factors (IGF), the capability of IGF-I to prime alveolar macrophages in vitro was also tested and found to be similar to that of GH, although the priming effects of GH were independent of the local production of IGF-I.⁽⁷⁾ Other studies were showing at the same time that non-stimulated as well as immuneactivated leukocytes were able to produce a GH-like peptide that was identical to pituitary GH,⁽⁸⁾ conferring credibility to the important hypothesis that communication signals originally identified in the neuroendocrine system can actually be used by immune cells. In the same vein, corticotropin-releasing hormone (CRH), the main regulator of the hypothalamic-pituitary-adrenal axis, has been identified in the immune system in which it functions as an autocrine/paracrine mediator of inflammation.⁽⁹⁾ In particular, CRH causes degranulation of mast cells and the release of histamine and several proinflammatory mediators.

The neuropeptides that are contained in sympathetic nerve endings that innervate lymphoid organs can play an important role in the modulation of the fine balance between the different populations of T helper (Th) cells that regulate cellular and humoral immunity. Th1 cells normally produce IFN-γ and interleukin-2 (IL-2), and both promote cellular immunity. In contrast, Th2 cells normally produce IL-4 and IL-10 that down-regulate cellular immunity and promote humoral immunity. CGRP and NPY drive Th1 cells towards the production of IL-4 whereas Th2 cells are driven by somatostatin and CGRP to produce IL-2 and IFN-γ.⁽¹⁰⁾ If these effects that were observed in vitro are also true under in vivo conditions, they provide a possible mechanism by which stress can polarize immune responses in the direction of either Th1 or Th2 cells.

A recent potentially important discovery is that of the inhibition exerted by the parasympathetic nervous system on the production of proinflammatory cytokines by macrophages. Direct electrical stimulation of the peripheral vagus nerve that innervates the liver inhibited the production of proinflammatory cytokines by Kupffer cells in response to a lethal dose of endotoxin and prevented development of septic shock.⁽¹¹⁾ This vagal function was termed the cholinergic anti-inflammatory pathway⁽¹²⁾ and it is mediated by nicotinic acetylcholine receptors containing an alpha-7 subunit.

Since the immune system makes use of communication signals and receptors that are identical to those used by the central nervous system, the immune system should be very sensitive to neural influences. Besides the cholinergic anti-inflammatory pathway already mentioned, many data attest to the fact that brain events have an impact on immune responses. For instance, lesions in the neuroendocrine brain have profound influences on immunity. As an example, destruction of the tubero-infandibular region of the hypothalamus in mice persistently abrogates natural killer cell cytotoxic activity without altering T and B cell populations, but cortical and sham lesions had only a short-lived effect.⁽¹³⁾ In other studies, ablation of the left sensori-motor cortex decreased cellular immunity whereas ablation of the right sensory-motor cortex increased it, showing that brain influences on immunity are lateralized.⁽¹⁴⁾ This lateralization phenomenon was later demonstrated to exist in the absence of any lesion since left-handed mice, labelled as such based on their predominant use of the left paw to reach a food pellet in a tube that only enabled them to use one paw, displayed higher cellular immune responses than right-handed mice.⁽¹⁵⁾ The mechanisms for this lateralized influence of the brain on immunity are still elusive.

The impact of stressors on immune responses represents another example of the influence of brain events on immunity. At the time these studies were carried out it was already well known that the influence of psychosocial stressors on the hypothalamic-pituitary-adrenal axis are not simply a function of the intensity and duration of the stressors but also depend upon their psychological features. Novelty, predictability, and controllability are the key factors that ultimately determine the neuroendocrine impact of stressors. It was therefore not surprising that the same psychological features were pivotal in the influence of stress on immunity. For instance,
rats exposed to inescapable electric shocks 24 h after injection of syngenic tumour cells displayed more rapid tumour growth and a higher mortality rate than rats exposed to controllable electric shocks, despite the fact that the intensity and duration of electric shocks were exactly the same in both groups.⁽¹⁶⁾ Lack of control had the same influence on the rejection of non-syngenic tumours⁽¹⁷⁾ and cellular immunity as measured by the proliferative response of lymphocytes to T-cell mitogens.⁽¹⁸⁾ It cannot be inferred that uncontrollability is always immunosuppressive. Lack of control over the occurrence of electric shocks was later found to increase rather than decrease humoral immunity, as measured by antibody titres against sheep red blood cells injected into rats that were submitted chronically to controllable or uncontrollable electric shocks.⁽¹⁹⁾ The same difference in the way the immune system responds to an uncontrollable stressor was confirmed in an experiment in which mice were exposed to the odour of a stressed congener. A 24 h exposure to this stressor decreased the cellular immune response, as measured by proliferation of T cells to mitogens and natural killer cell cytotoxicity, but increased antibody titres against keyhole limpet haemocyanin.⁽²⁰⁾

Studies of the influence of stress on the immune system have also been carried out in human subjects in experimental settings or in real-life conditions. In these studies, the immune end points are either measured on blood lymphocytes or deduced from the result of an already existing pathological process. The group of Janice Kiecolt-Glaser at Ohio State University in Columbus is certainly the pioneer in this field. For instance, first year medical students were shown to display a reduction in the production of IFN-γ by circulating leukocytes and a reduced cytotoxicity of natural killer cells during the end of the year examination period, and these changes were independent on lifestyles.⁽²¹⁾ As a result of extensive studies on different populations at risk, such as spouses experiencing marital conflict, caregivers of patients with Alzheimer’s disease, and aged subjects, Kiecolt-Glaser’s group proposed that negative emotions and stressful experiences can contribute to prolonged infection and delayed wound healing. In addition, negative emotions were proposed to directly produce the production of proinflammatory cytokines and therefore increase the risk for a spectrum of conditions associated with ageing, including cardiovascular disease, osteoporosis, arthritis, type 2 diabetes, certain cancers, frailty and functional decline, and periodontal disease.⁽²²⁾