Interactions of Adipose and Lymphoid Tissues
1. INTRODUCTION Signal Molecules
Interactions of Adipose and Lymphoid
proteins in chemical diversity and physiological importance in the immune system and elsewhere (5,6). They are synthesized from long-chain polyunsaturated fatty acids (PUFA) that are dietary essentials (i.e., they cannot be synthesized de novo) and are
“stored” as components of membranes and triacylglycerols.
In addition to acting as precursors for lipid-derived messenger molecules, lipids also modulate the synthesis, secretion, and reception of peptide messengers. Nanotubes derived from the cell membrane have been suggested as a means by which immune cells exchange surface proteins and perhaps larger particles (7). Nonesterified fatty acids can act as regulators of surface receptors (8) and gene transcription (9). The incorporation of fatty acids into complex lipids does not have the equivalents of mRNA, tRNA, and ribosomes that ensure that the correct precursors are assembled into proteins, so many, including phospholipids, are structurally varied. Recent experiments on Caenorhabiditis elegans have demonstrated that fatty acid insufficiency alters neurotransmission and behavior (10). Lipids rafts require appropriate fatty acids that are believed to have a major role in cell signaling in lymphoid cells (11,12). Insufficiencies would impair cell signaling and so disrupt immune function. Caveolae are also numerous in adipocyte membranes and are essential for their responses to insulin (13), and prob- ably other signal molecules.
Adipocytes can selectively sequester and release fatty acids that differ in molecular weight and degree of unsaturation (13) and may supply specific fatty acid precursors to contiguous lymphoid cells (14). By ensuring that appropriate precursors are available where and when they are needed for synthesis of complex molecules, adipocytes can fulfill for lipids the role of some of the protein synthesis machinery for proteins. But the anatomical relations between adipocytes and lymphoid cells are important for the effi- ciency of such interactions.
1.2. Cellular Interactions
The next step was the demonstration of immune cells, especially macrophages (15,16) and dendritic cells (17,18), in adipose tissue. Although representing only a tiny fraction of the total mass of a depot, these intercalated cells are the source of large pro- portion of the TNF-F (and probably of many other signal molecules) synthesized in whole adipose tissue (19). The fatty acid compositions of lipids in intercalated dendritic cells closely resembles those of contiguous adipocytes (Fig. 1). Changing the dietary lipids alters the fatty acid composition of both types of cells, but the correla- tion between values for cells that were adjacent in life remains. The simplest explana- tion for this similarity is that the dendritic cells take up their lipids from the contiguous adipocytes, rather than from the blood or lymph, as was previously assumed (14). The site-specific differences in adipocyte composition (20) are thus conferred on interca- lated dendritic cells, adding another source of diversity to these cells that hitherto have been classified by genes activated and proteins synthesized. It also may be an essential and hitherto unrecognized link in the mechanisms by which dietary lipids modulate immune function. Some properties of preadipocytes in vitro are strikingly similar to those of macrophages, suggesting common pathways of early development of macrophages and adipocytes (21,22). Preadipocytes injected into mice acquire surface antigens characteristic of macrophages (23). Receptors for bacterial products such as lipopolysaccharide have been demonstrated in 3T3-L1 adipocytes in vitro (24,25).
These properties are found only in preadipocytes extracted from white adipose tissue, not those of brown adipose tissue (26), and disappear as soon as mitosis ends and the cells differentiate.
Dendritic cells interact with adjacent adipocytes (27). Those extracted from the adipose tissue stimulate lipolysis, whereas those from an adjacent lymph node inhibit the process, although the effects are strong only in perinodal and milky spot-rich samples and minimal in the adipocytes extracted from sites more than 1 cm from lymph nodes. Switching from antilipolytic to lipolytic secretions seems to be among Fig. 1. The correlations between mean unsaturation indices (UI =[% monoenoic fatty acids] +2[%
dienoic]+3[% trienoic]…etc.) of fatty acids extracted from complex lipids (mostly membrane phos- pholipids) in dendritic cells (DCs) and those of adipocytes (mostly triacylglycerols) isolated from the corresponding sample of adipose tissue (mesenteric perinodal and remote, omental with many or few milky spots, and popliteal remote from node) from adult male rats fed on unmodified chow (circles);
rats fed on chow +20% sunflower oil for 6 wk (squares). N=6 sets of homologous samples from 3 similarly treated cage-mate rats for each dietary group. The standard errors of each mean are shown as bars. Data simplified from ref. 14.
the transformations that dendritic cells undergo as they migrate between the lymph nodes and the adjacent adipose tissue, and thus should be considered as part of the maturation process (27). Inducing mild inflammation by injection of lipopolysaccharide amplifies both effects, suggesting that they are integral to immune responses. Paracrine interactions between adipocytes and macrophages have also been demonstrated (28).
The significance of these lymphoid cells in intact animals remains unclear; a role in defense against pathogens has been suggested (29) and they may contribute to insulin resistance and other chronic metabolic disorders (30).
1.3. Paracrine Interactions Between Adipose Tissue and Other Tissues 1.3.1. LYMPHNODES ANDOMENTALMILKYSPOTS
In most animals, including murid rodents, adipose depots (such as perirenal, epididy- mal, and parametrial) that consist of “pure” adipose tissue are larger—in obese speci- mens, much larger—than those with embedded lymphoid structures. These depots are often the only ones providing enough tissue in mice and other small species, and are preferred for study, especially for experiments in which the tissue undergoes large changes in relative mass, as in obesity research or for the study of gene action in trans- genic mice (31). To avoid collateral damage, biopsy sites from humans and larger ani- mals are always chosen for their remoteness from lymph nodes and vessels.
Consequently, most of the data come from samples of large adipose depots chosen for their surgical accessibility (32–36) rather than for known site-specific properties. Much of the confusion and paradoxical findings about the relationship between adipose stores and immune function can be clarified by taking account of site-specific properties of adipose tissue and its interactions with adjacent lymphoid cells.
Recent reviews (37,38) summarize the evidence demonstrating that perinodal adipocytes selectively take up and store dietary lipids (and perhaps other precursors) that are released in response to local, paracrine signals from adjacent lymphoid cells that take them up. The depots that incorporate lymphoid tissue, especially the perinodal regions, are specialized to support the growth and metabolism of adjacent leukocytes.
Many hitherto unexplained site-specific differences in cytokine metabolism in adipose tissue are consistent with the hypothesis of paracrine interactions between lymphoid and adipose TNF-F than those from the nodeless epididymal or perirenal depots, and their secretions are less readily modulated by dietary lipids (39). The role of embedded lym- phoid tissues in this finding was never investigated. Genes specific to adipocytes are upregulated in the stomach wall of mice successfully immunized against Helicobacter, probably arising from paracrine interactions between adipose and lymphoid tissues (40).
In vitro studies (Figs. 2,3) show that perinodal adipocytes respond much more strongly than others to TNF-F, interleukin (IL)-4, IL-6, and probably other cytokines (41). Figure 2 shows that the response of adipocytes isolated from perinodal adipose tissue contiguous to nodes differ from those from a few millmeters distant, and the site-specific differences are amplified by the neurotransmitter norepinephrine (41).
Furthermore, the action of combinations of cytokines is not predictable from that of single agonists: although IL-6 is prolipolytic, especially for perinodal adipocytes (Fig. 2), when combined with IL-4 it suppressed lipolysis (Fig. 3). The same is true of interactions between IL-4 and TNF-F(41). The data in Figs. 2 and 3 report only lipolysis, but similar site-specific differences in the responses to signal molecules probably occur for other
metabolic pathways. Greater sensitivity to cytokines is also implicated in the larger responses of perinodal adipocytes to the presence of dendritic cells (29). The interactions between the effects of cytokines and neurotransmitters on perinodal adipocytes (Figs. 2,3) suggest mechanisms by which the nervous system may modulate immune processes.
IL-6 was found to be 100 times more concentrated in the interstitial fluid of human superficial adipose tissue than in the blood plasma of the same subjects, strongly suggesting a paracrine role for this cytokine in adipose tissue (42). Nonetheless, the nodeless epididymal or perirenal depots remain the favorite choice of adipose tissue for cytokine studies (34,43,44). Even where depots are compared, the largest rather than those most strongly involved in cytokine secretion are chosen for study (45). Matters are Fig. 2. Rates of glycerol release over 1 h from mesenteric perinodal (darker bars) and far from (paler bars) lymph node(s) adipocytes from adult guinea pigs (N=12) with and without supramaxi- mal norepinephrine, after preincubation for 24 h with and without interleukin-6 at two different concentrations. Asterisks denote comparison of pairs of samples from the same depot under the same conditions: ** significantly different at p < 0.01; * significantly different at p < 0.05. Daggers denote comparison of corresponding sample incubated with and without cytokines: ††† significantly different at p < 0.001; †† significantly different at p < 0.01; † significantly different at p < 0.05.
Data simplified from ref. 41.
further complicated by erroneously classifying the intra-abdominal perirenal/retroperi- toneal and gonadal (i.e., epididymal or parametrial) depots in the same category (“visceral”) as the omentum and mesentery (46–50), although the latter have many distinctive properties arising from the embedded lymphoid structures not found in the former. Data from the larger, nodeless depots may elucidate the metabolic changes associated with obesity (34,51), but could be misleading as well as irrelevant to under- standing the source and action of adipocytes and the nutrition of the immune system.
By releasing fatty acids only to lymphoid cells and only when and where they are required, the perinodal adipose tissue partially emancipates immune function from Fig. 3. Rates of glycerol release over 1 h from mesenteric perinodal (darker bars) and far from (paler bars) lymph node(s) adipocytes from adult guinea pigs (N=12) with and without supramaxi- mal norepinephrine, after preincubation for 24 h with and without interleukin-4 (10 ng/mL) alone or with interleukin-6 (0.5 ng/mL) or TNF-F(10 ng/mL). Daggers compare corresponding samples incubated with or without cytokines: ††† significantly different at p < 0.001; † significantly differ- ent at p < 0.05. For clarity, symbols indicating within-depot differences, and those indicating that all the values from “near node” adipocytes, are significantly different at p < 0.001 from those from the corresponding control samples incubated without cytokines are not shown. Double daggers compare samples incubated with two cytokines with the corresponding sample incubated with IL-4 alone: ‡‡‡ significantly different at p < 0.001; ‡‡ significantly different at p < 0.01. Data simplified from ref. 41.
dependence on the quantity and lipid composition of food (38). Paracrine interactions may also account for some features of the anatomy of lymph vessels and nodes (52–54).
The branching of lymph vessels near nodes would slow the passage of lymph and bring a greater surface area of vessels into contact with adipocytes, thus facilitating the exchange of signals and metabolites.
1.3.2. BONEMARROW
Red bone marrow has much in common with lymph nodes in mammals, as it is the origin of many of the lineages of lymphoid cells found in nodes (55). Adipocytes in
“yellow” marrow are smaller than those of other depots, but the roles of cell prolifera- tion and cell expansion in tissue growth follow a similar ontogenetic time course (56).
Independent studies of the contribution of bone marrow adipocytes to the control of lymphohematopoiesis in vitro (57,58) reveal paracrine interactions similar in principle to those we have described between perinodal adipose tissue and lymph-node lymphoid cells (37,38). Suggested mediators for paracrine interactions between bone marrow adipocytes and lymphoid cells include both cytokines (59) and polyunsaturated fatty acids (60). Bone marrow adipocytes may also interact locally with osteoblasts in bone formation (154). However, most of the data come from in vitro experiments, and exactly how such processes determine the course of osteoporosis in elderly people remains to be established (61). Transgenic mice without detectable adipocytes in the bone marrow develop normal skeletons (62).
1.3.3. OTHERPARACRINEINTERACTIONS
Local paracrine interactions between minor depots of physiologically specialized adipocytes and anatomically contiguous tissues are increasingly recognized. Such effects have been described for major blood vessels (63,64), skeletal muscle (65), car- diac muscle (66,67), bone (57), and lymph nodes (37,68). All these concepts are based on experiments using laboratory animals. Site-specific differences and paracrine interac- tions are difficult to detect in humans, as they have little or no blood manifestation, but they should be considered in interpreting data from the emerging field of lipidomics (69,70) as well as the actions of dietary and blood-borne lipids on the composition of lymphoid cells (71,72).
The concept of local, paracrine exchange of signals and nutrients between function- ally specialized contiguous adipose tissue and other tissues provides an explanation for why adipose tissue in mammals is partitioned into a few large and many small depots, most of the latter associated with lymphoid structures (38). Lipid reserves that are dedicated to supplying the immune system may be essential to combining fever with immune responses in defense against pathogens and to enabling both processes to be combined with anorexia. Many of the polyunsaturated fatty acids selectively accumu- lated in triacylglycerols of perinodal adipocytes (20) and in phospholipids in lymph node lymphoid cells (68) and dendritic cells (14) are dietary essentials, so they can be scarce, especially during periods of high metabolic demand (e.g., pregnancy, lactation) and reduced food intake (e.g., while disabled by injury or disease). Local control of lipolysis by the immune system manages supplies of these essential fatty acids effi- ciently, and minimizes competition with other tissues (e.g., pyrogenic tissues such as muscles and liver) for circulating lipids.
1.4. Growth of Adipose Tissue Associated With Lymph Nodes
Control of proliferation of preadipocytes and of their maturation into adipocytes has long been studied, because of its implications for obesity in humans and domestic live- stock, but emphasis has been on diet and energy balance (73) and angiogenic factors (74,75). Local growth of adipose tissue associated with long-standing lymphatic dis- orders has long been known, though its cause is poorly understood (76). More recently, the increasing prevalence of certain human disorders has directed attention to the pos- sibility that inflammatory cytokines and other immune-derived factors also have a role in regulating adipogenesis (51,77).
The actions of dietary lipids and hormones can be simulated in vitro, but demonstrat- ing a role for the immune system requires in vivo experiments. Unfortunately, adipo- cytes form and die very slowly, and preadipocytes cannot be recognized unambiguously, so measuring adipocyte turnover in vivo is difficult (78,79). The total adipocyte com- plement seems to differ substantially between individuals of similar body composition, even among inbred mammals (80), making the detection of experimentally induced increases in numbers of adipocytes difficult unless they are very large. Major systemic immune responses induce anorexia, and eventually cachexia, which deplete the adipose tissue anyway, especially in small laboratory animals. However, by comparing the num- bers of adipocytes in the popliteal depots, and locally stimulating the lymph nodes in one of them with small doses of lipopolysaccharide, the formation of more adipocytes in response to chronic inflammation could be demonstrated, corresponding to the decrease in average volume of adipocytes throughout the depot (81). After experimen- tal lipectomy of the epididymal fat pads of adult rats, compensatory regrowth of adipose tissue is significant 16 wk later in the node-containing mesenteric and inguinal depots but not in perirenal depots (82).
Local hypertrophy of adipose tissue associated with lymph nodes is not reversed within 3 mo of the end of experimental inflammation (83). Increases in infiltrating den- dritic cells and in adipocyte apoptosis were found, as expected, in the adipose tissue around the inflamed lymph node, and also in the perinodal mesenteric and milky spot- rich parts of the omentum. This finding is another case in which adipocytes in these intra-abdominal depots have well-developed properties associated with paracrine inter- actions with lymphoid cells and actively participate in immune responses that in other respects are confined to remote parts of the body.
The quantities of perivascular and intermuscular adipose tissue are so small and so variable in laboratory rodents that systematic study of their growth in vivo has not yet been possible. Human postmortem data suggest the presence of cardiac depot hyper- trophy in chronic heart disease (66,67).
2. PARACRINE INTERACTIONS OF ADIPOSE TISSUE IN HUMAN