Contents - Previous - Next

Effect of protein-energy interaction with reference to immune function and response to disease

G.T. KEUSCH*

* Division of Geographic Medicine and Infectious Diseases, New England Medical Center Hospitals, Tufts University School of Medicine, 750 Washington Street Boston, MA 02111, U.S.A.

Abstract
1. Introduction
2. Outlining the issues
3. Host metabolism and host defense
4. The metabolic profile of the infected host
5. The role of cytokines
6. Cytokine regulation: Natural antagonists and biological modulators
7. The future
References

Abstract

Protein-energy malnutrition is associated with reversible and clinically significant immunosuppression of multiple host defense mechanisms. Infections also lead to nutrient wasting, and may trigger nutritional deterioration in initially well-nourished persons or exacerbate pre-existing malnutrition. Host metabolism and host defense are interlinked by a common set of peptide mediators (cytokines) that regulate both the altered metabolism and the activation of the immune system in infection. This regulation proceeds by transcriptional and translational control of specific genes, and the molecular mechanisms by which these events occur are currently under intensive investigation and are being unraveled. The levels of dietary intake of calories, protein and various micronutrients to optimize these responses is not known. In order to use these regulatory pathways to clinical advantage, additional information will be required. There are five major areas of relevant contemporary research, as the focus of study shifts from in vitro experiments to in vivo studies of the whole animal, tissue, cell, and gene levels: (1) development of methods to assess cytokines in vivo, (2) study of tissue- and cell-specific responses to single and multiple cytokines given together, (3) genetic regulation of cytokine responses, (4) impact of nutritional status on cytokine responses in vivo, and (5) identification of cytokine inhibitors and their role in modulating cytokine responses in vivo.

1. Introduction

It is now over three decades since the publication of the seminal paper by SCRIMSHAW, TAYLOR and GORDON (1959), describing the interaction of nutrition and infection. In this paper, and later more fully developed in a WHO monograph (1968) by the same authors, the concept was introduced that nutritional status affected immune function and conditioned the host response to infection. Over the ensuing years, as the result of considerable effort into the study of these mutual interactions, many phenomena have been described, and plausible explanations have in a few situations been put forward. However, it remains a great disappointment that the molecular mechanisms by which malnutrition affects the immune system by and large remain as uncertain today as ever.

The reasons for this situation are several: (1) Malnutrition is far too complicated to permit simple physiological models or even universally applicable classification schemes. Human malnutrition is characterized by multiple nutrient deficiencies of differing extent in individual patients, varying over prolonged periods of time, and frequently complicated at the same time by the presence of infections, each with its own significant impact on immune function. (2) Animal models do not faithfully reproduce what occurs in the human because of the necessary and rigorous experimental control exerted in making these models, resulting in defined or single nutrient deficiencies that are uniformly imposed in the absence of known infection. (3) The organization and regulation of the immune system in animals, while a paradigm for understanding human immunity, is not identical to the human system, and important, albeit sometimes subtle, differences exist. (4) The hypotheses concerning nutrient effects on immunity and host defense are often gross oversimplifications, and the interpretations of observed data at times are rather naive. (5) Our understanding of the dynamics of regulation of immune function is still fragmentary. (6) Patients with malnutrition are either in developing countries lacking sophisticated investigative capacity, or located in settings in which the most modern methods can be applied but the patients are commonly affected by other serious and confounding illnesses.

2. Outlining the issues

The two fundamental questions relevent to this conference, what levels of protein and energy intake are required for normal immune function and host defense, and second, are there immunologic markers that will serve as surrogate measures of host nutritional state, remain difficult to answer in definitive terms. Most studies have been conducted in severely malnourished individuals, and the clinical immunological tests in common use lack the requisitive sensitivity to answer these questions. Moreover, abnormal test results do not necessarily translate into clinically relevant decreases in host defenses. (KEUSCH, 1990a).

Review of the interactions of nutrition and immunity in animals reveals a number of general issues of potential relevance to the human situation. Two recent and well referenced reviews of animal data have appeared (BURKHOLDER and SWECKER, 1990; REDDY and FREY, 1990). The first of these is particularly noteworthy, for it examines the effect of nutritional deficiency in domestic food animals rather than in models created in small laboratory animals. While it is hard to consider the conditions under which feed animals are reared as natural, they are certainly distinct from those of the laboratory rodent and at least provide a new perspective for evaluating the impact of dietary changes on host immune responses and infection over time. Unfortunately, there are few studies of protein-energy malnutrition in large animals, and the available studies have yielded conflicting results.

The first issue emerging from animal studies is that early malnutrition, whether induced in utero or within the first few weeks of life, may result in prolonged or permanent impairment of immune response (JOSE, STUTMAN and GOOD, 1973). Moreover, the offspring of protein- or energy-deprived mice also manifest reduced immune responses, even when the offspring themselves are given adequate nutrition after birth (CHANDRA, 1975; BEACH, GERSHWIN and HURLEY, 1983). Weanling mice receiving diets restricted to 60% of normal (the standard generally being considered to be the usual intake of animals fed ad libitum) demonstrate a reduction in the number of antibody forming cells, in mitogenic (proliferative) responses, and in delayed type hypersensitivity (DTH) skin responses, for as long as two years (WALFORD et al., 1973). Later on, antibody forming cells and mitogen responses may increase over control levels, although DTH still remains abnormal (GEBRASE-DeLIMA et al., 1975). Thus, early malnutrition may affect diverse maturational events, including the development or selection of T- and B-cell repertoires, which comprise the preprogrammed immune responsivity of the host. Deletions in this repertoire are thought to lead to autoimmunity.

The second issue, raised by a number of studies in mice, is that when animals are fed energy-restricted diets they experience a decreased incidence of spontaneous neoplasms and autoimmune disease compared to controls fed ad libitum, and the 'deprived' animals apparently live longer (WEINDRUCH and WALFORD, 1982; WEINDRUCH et al., 1982; 1983; BEACH, GERSHWIN and HURLEY, 1982; FERNANDES et al., 1978). Mildly restricted protein diets also lead to enhanced cell-mediated immune (CMI) responses, even though serum immunoglobulin levels and specific antibody production may be diminished (KRAMER and GOOD, 1978; PRICE and BELL, 1973; BARRY et al., 1979; REINHARDT and STUART, 1979). The antibody defect can be corrected by the transfer of thymocytes from protein-adequate mice to deficient animals (MATHUR, RAMALINGASWAMI and DEO, 1972). These studies raise the question of what is 'normal' or optimal protein and energy intake.

It must be quickly noted that the concept that nutrient restriction in animals boosts the immune response is based on the assumption that nutrient consumption during ad libitum feeding of caged animals is the normal standard for comparison. But if ad libitum dietary intake actually represents a significant excess over physiological needs, that is if it leads to relative obesity, it would not be surprising to find that the ad libitum diet might lead to adverse consequences on immune function, disease occurence, and survival. It may well be that energy- or protein-restricted lab rodent diets are optimal rather than restricted in the sense that they optimize physiological responses and promote health in the rodent. In this case the results could not be directly extrapolated to the human, i.e., one could not suggest that energy or protein intake be restricted to 60% of present recommended levels, and that induction of a state of mild malnutrition is good for the individual. Comparative studies, perhaps impossible to achieve, would be necessary to prove this.

The third issue is that there may be major differences in the effects of feeding different fat diets on immediate responses such as immune function and long-term consequences such as development of neoplasms, at least in rodents. Diets deficient in essential fatty acids increase lymphocyte blastogenic responses, while the incidence of neoplasia and autoimmune disease diminishes. In contrast, feeding an excess of polyunsaturated fatty acids (PUFA) suppresses immune responses and increases neoplasia (ERICKSON, 1983; HWANG, 1989). The data could be explained by the known impact of dietary lipids on the ratio of saturated and unsaturated fatty acids in biomembranes which, in turn, affects the fluidity and functional organization of the plasma membrane of most cells in the immune response. These membrane changes not only potentially act as critical variables in determining immune function and, ultimately, tumor surveillance, but also result in altered production of eicosanoids which exert regulatory effects on the immune system.

The available in vivo data remain rather limited and are primarily descriptive, since mechanisms are ordinarily not addressed. For example, it is reported that diets rich in linolenic acid impede rejection of skin grafts in rats (RING et al., 1974), while guinea pigs given fish oil emulsions intravenously prior to challenge with endotoxin (lipopoly-saccharide, LPS) survive better than controls given safflower oil (MASCIOLI et al., 1988).

3. Host metabolism and host defense

It has long been known that nutritional deterioration occurs in the course of acute infections, with dramatic changes in both energy and protein metabolism (KEUSCH, 1979). More recent studies have shown that the common metabolic response to infection is coupled to activation of immune host defenses (KEUSCH and FARTHING, 1986). To fully address the complex relationship between protein and energy require meets and host immune responses during infection, however, it is not only necessary to describe these interactions, but also to determine how these events are mechanistically linked. Only then will it become possible to design biochemically and biologically sound interventions to control these events for the benefit of the host. Investigators have previously focused on the phenomenology of events during acute infections; today we are interested in the proximate or intermediate causes of these events. The future, no doubt, will be concerned with the molecular biology of these critical host responses, bringing us to the level of gene regulation.

Numerous descriptive studies over the past 25 years have demonstrated that individuals with protein-energy malnutrition present functional abnormalities in one or more host defense system. These phenomena have been reviewed elsewhere and will not be considered here in detail (MASCIOLI et al., 1988; KEUSCH, 1979; KEUSCH, 1990b). Briefly, patients with PEM usually have diminished numbers of mature circulating and tissue T lymphocytes. This is apparently due at least in part to a lack of peptide hormones derived from the thymus gland which influence T-cell differentiation and maturation (KEUSCH, 1992). The resulting deficit in mature T cells predicts a decrease in normal T-cell functional capacity, leading to impairment of all immune mechanisms requiring T-cell involvement. These include not only cell-mediated immune mechanisms, but also B-cell functions such as recognition and antibody production to most antigens, and the isotype switch from IgM to IgG production (KEUSCH, 1983).

Other host defense systems are also affected (KEUSCH, 1990b). For example, complement activity in plasma is usually low, probably due to both in vivo consumption of complement and an inadequate increase in synthesis of complement component proteins which normally occurs during the acute phase response. In vitro studies also show that phagocytic cells from PEM patients often do not generate normal amounts of oxygen radicals with bactericidal activity. The combination of immunoglobulin and complement defects will impede in vivo opsonization of microorganisms, and thereby reduce the normal ingestion and killing of pathogens by host phagocytic cells, even though these functions appear relatively preserved in the in vitro assay systems, for which normal human serum is ordinarily used.

With such a diversity of host defense abnormalities in PEM, there should be, and in fact is, susceptibility to a wide range of infectious agents of different organ systems. PEM patients develop more frequent, more prolonged and more severe infections.

4. The metabolic profile of the infected host

The metabolic response to systemic infection is generally similar regardless of the nature of the infecting agent, and is geared towards the consumption of endogenous stores of energy and protein, as if anticipating decreased food intake due to anorexia and fatigue during infection (BEISEL, 1984). Thus, carbohydrate stores in the form of liver glycogen are quickly consumed, the oxidation of fat commences, and skeletal muscle proteins are catabolized and fed into gluconeogenic pathways to make glucose. Depletion of stores of carbohydrate, fat, and protein develops while glucose flow, plasma glucose concentration, and glucose oxidation rates increase. These changes in carbohydrate utilization are often designated 'the glucose intolerance of infection', but appear to have the physiological purpose to maintain the supplies of energy for the host during the period of reduced food intake, and they are therefore supported by the necessary and appropriate alterations in glucogenic hormones, including insulin, glucagon, and growth hormone (BEISEL, 1984).

Table 1. Interleukin-1-induced gene activation

The consumption and loss of energy and protein stores from the body are reflected in net negative energy and protein balances. These must result in weight loss, which is precisely what happens to the host during acute and, even more dramatically, during chronic infections. On top of this metabolic wasting, the presence of fever requires an additional increase in energy utilization. This was appreciated over 50 years ago when an increase in the basal metabolic rate of approximately 13% was measured for each degree centigrade of fever above normal, regardless of the cause of the temperature elevation (DuBOIS, 1937). During sepsis, measured resting energy requirements may increase by as much as 40% above normal (LONG, 1977).

As already noted, catabolism of muscle protein is clinically noticeable as the loss of muscle mass, and is reflected in negative nitrogen balances and elevated nitrogen excretion in urine. Serum levels of albumin, prealbumin, transferrin, and other proteins also diminish rapidly and predictably. At the same time, major anabolic responses occur, including biosynthesis of proteins associated with the acutephase protein response (such as ceruloplasmin, µ2-macroglobulin, µ1-antitrypsin, and the third component of complement), cells involved directly in host defenses, and immunoglobulin, interferons, and other soluble proteins needed for immune responses and tissue repair. Catabolic and anabolic events occur simultaneously during infection as a result of transcriptional control of specific genes (Table 1), while the immune system and host defenses are being activated. Observing these diverse events, it certainly does appear as if there was a preprogrammed and highly coordinated host mechanism to cope with the stress of infection.

5. The role of cytokines

One of the major breakthroughs in understanding the mechanisms underlying the alterations described above is the discovery that certain endogenous signals regulate both the metabolic and the immunologic responses (DINARELLO, 1984). These signals appear to be largely carried by a series of small peptides produced by inflammatory cells known collectively as cytokines, including interleukin (IL)-1, IL-6, tumor necrosis factor (TNF), and no doubt others. With these discoveries just a few years ago it seemed that the mechanisms underlying the metabolic alterations associated with infection would be quickly unraveled with sufficient clarity to permit targeted interventions (KEUSCH and FARTHING, 1986). This concept was largely based on information derived from in vitro study of a few cytokines. While the view that these peptides are the keys to reorganization of host metabolic responses and priorities remains the most likely mechanism to explain the observed events, this has been complicated by the identification of many more mediators with overlapping properties, providing both up- and down-regulatory controls, and the finding that responses in vivo may not parallel events in vitro. Even the well-known fever inducing activity of IL-1, which led to its designation as the endogenous pyrogen (DINARELLO, RENFER and WOLFF, 1977), is in question now (KLUGER, 1991), because other cytokines also cause fever by similar mechanisms (DINARELLO et al., 1986; 1991).

In the beginning of the cytokine era, it was shown that a single injection of as little as 10-100 ng/kg of purified IL-1 resulted in fever, neutrophilia, increases in the level of other cytokines (such as leukocyte colony stimulating factors and IL-6), decreased serum iron and zinc levels, increased protein synthesis in the liver but diminished albumin production, and produced changes in glucoregulatory hormone levels in circulation in experimental animals (DINARELLO, 1991). It truly seemed that the master key to a control box sitting at the cross roads of immune activation and metabolic adaptation to infection had been identified. All was good and the world was simple, needing only further study for full understanding.

However, a few years later, a product of LPS-stimulated macrophages was found to have potent effects on lipid metabolism in T3T adipocyte culture and in vivo (KAWAKAMI et al., 1982). The LPS-stimulated culture supernatant sharply decreased synthesis of lipoprotein lipase (LPL) and lipid synthetic enzymes in these cells, thus resulting in a failure to take up and store lipids. These changes were characterized as 'cachexia at the cellular level', and the mediator was named cachectin (BEUTLER et al., 1985a). It was suggested that cachectin was the mechanism of the hyperlipemia accompanying many acute infections and the mechanism of fat depletion during infection. Purification and sequencing of the mediator, however, revealed its identity with the previously described TNF (BEUTLER et al., 1985b). All still seemed quite simple, except that now differential production of these two mediators, IL-1 and TNF, needed to be invoked to explain the patterns of response to different infections, although both the signal and the producing cell were the same for the two peptides.

Today, things no longer look so simple. IL-1 and TNF have been shown to mediate the same multiple events, with a few notable exceptions, and another distinct cytokine, IL-6, has been found which overlaps in activity in a major way with these other two (Table 2; DINARELLO, 1989). IL-1 and TNF can also induce the synthesis of one another, and this is enhanced in the presence of another common soluble mediator, interferon-g (IFN-g). At the same time, IFN-g inhibits the activation of IL-1 synthesis by IL-1, making it clear that regulation of the production of these cytokines in vivo may be very complex, where it may be conditioned by the relative amounts present of a number of molecules with regulatory properties.

Intensive study of IL-1 has also demonstrated that transcription and translation of IL-1 genes are two separable and distinct processes. Different activators affect these two events with differential impact and distinct kinetics. Transcription without translation is found under certain conditions, for example adherence of blood monocytes to surfaces or incubation of monocytes in the presence of C5a derived from the activation of complement, without alteration in the rate of degradation for the message. These cells, containing small amounts of IL-1 mRNA, can therefore be considered to be 'primed' and ready to rapidly enter translation when the appropriate signal comes along. For example, heat-killed Staphylococcus epidermidis serves primarily as a translational signal to 'primed' monocytes (SCHINDLER, CLARK and DINARELLO, 1990). Calcium ionophores also selectively increase processing and secretion of IL-1 (SUTTLES, GIRI and MIZEL, 1990), while the transcription and translation of IL-1 are inhibited by corticosteroids, which are most effective when present before a stimulus is applied (KNUDSEN, DINARELLO and STROM, 1987). 13-lipoxygenase products appear to be involved in regulation as well; non-specific blockers of the lipoxygenase pathway, but not specific inhibitors of the 5-lipoxygenase pathway, inhibit IL-1 production (DINARELLO et al., 1984; SIRKO et al., 1990). It should also be noted that there are two related yet structurally unique IL-1 molecules under separate gene control, IL-1µ and IL-1b (FURUTANI et al., 1986; BENSI et al., 1987), which have overlapping but not identical effects on cells. There are also two different cell receptors, types I and II, with distinct distribution on cells (Table 3) and capacity for binding the two IL-1 species (BOMSZTYK et al., 1989; CHIZZONITE et al., 1989). Although both receptors bind both forms of IL-1, differences in the binding epitopes for the µ and b forms may exist on certain cells, depending on nearby groups or conformational features. This may permit distinct patterns of response in the host, according to the amounts of each mediator present and the nature of the extracellular matrix in which the responding cells exist at the time (NATHAN and SPORN, 1991).

Table 2. Comparison of biological activities of some cytokines

Table 3. Distribution of interleukin-1 receptors

Additional controls over responses may be exerted by regulation of surface expression of cytokine receptors. Partial occupancy of IL-1 receptors on some target cells can significantly down-regulate the availability of the receptor for hours (MATSUSHIMA et al., 1986) and alter the expression of receptors for other cytokines, such as TNF (HOLTMANN and WALLACH, 1987). While it is still difficult to put together a cohesive picture of what happens when cytokines are released in vivo, or even in vitro, the complexities already noted make it abundantly clear that the presence of multiple potentially modulatory interactions in the system should allow for considerable fine tuning in vivo of the stimulus-response coupling.

While TNF has come to resemble IL-1 more and more as, for example, the common biological activities of these two mediators have been demonstrated (DINARELLO, 1989), the very rapid rise in TNF levels in vivo following LPS administration, its ability to induce IL-1, and the apparent correlation of TNF levels with clinical events in certain circumstances, for example in septic shock, bacterial meningitis, or cerbral malaria, has led some authorities to speculate that its role in disease might be pre-eminent (ZIEGLER, 1988). For example in one recent study, TNF, but not IL-1 serum levels, correlated with fever at diagnosis in children with solid tumors (ISHII, OHGA and UEDA, 1990). However, TNF and other cytokines may play a role in health and host defense that may be helpful at some level, but harmful if allowed to progress too far (TRACEY and LOWRY, 1990).

Recent studies of the effects of TNF on lipid metabolism have demonstrated significant distinctions between in vitro and in vivo events (GRUNFELD et al., 1989b). While the previously described (PEKALA, LANE and CERAMI, 1982) inhibitory effect of TNF on LPL in fat cells in culture could be confirmed, in vivo diminished LPL activity was detected in the epididymal fat pad of rats given infusions of TNF, but not in other adipose tissues or muscle, while post-heparin plasma lipase activity, in fact, increased. In addition, the kinetics of the response showed that it took several hours for the fat pad LPL to drop, while hypertriglyceridemia was demonstrable within 45 minutes to 1 hour.

There appear to be several mechanisms by which hyperlipidemia is mediated, which at least in the rat can be altered by diets affecting basal triglyceride levels. One is the TNF-induced increase in hepatic content of citrate previously reported in chow-fed animals. Citrate is an allosteric activator of acetyl-CoA carboxylase, a limiting enzyme for triglyceride synthesis (GRUNFELD et al., 1988). A second mechanism is TNF-induced lipolysis, which results in an increase in circulating free fatty acid levels, thus stimulating triglyceride synthesis. An additional mechanism has been described in rats receiving a high-sucrose diet. In these animals, there is no increase in plasma fatty acids; rather, TNF stimulates hepatic fatty acid synthesis in a more direct manner, providing the substrate for increased triglyceride production. Interestingly, TNF-induced lipid changes persist, although there is a demonstrable tachyphylaxis to its effects on appetite and weight loss (GRUNFELD et al., 1989a).

Hyperlipidemia has also demonstrated to be a consequence of administration of other cytokines, including IL-1, interferons (FEINGOLD et al., 1989) and IL-6 (GRUNFELD et al., 1990), and is blocked by inhibitors of prostaglandin synthesis (FEINGOLD, DOERRLER and DINARELLO, 1992). The common end result of cytokine lipid alterations, the acute hypertriglyceridemia, suggests that this, like so many cytokine-mediated responses, may be of some importance to host defenses and that it might have survival value. Indeed, in an experimental model of endotoxemia in mice, the addition of triglyceriderich VLDL and chylomicrons, as well as cholesterol-rich LDL and HDL, or a synthetic cholesterol-free lipid emulsion to the LPS dose protected the animals from lethal doses of LPS (HARRIS et al., 1990).

Contents - Previous - Next