Loss of Lymphotoxin-α but Not Tumor Necrosis Factor-α Reduces Atherosclerosis in Mice*

Inflammatory processes are involved with all phases of atherosclerotic lesion growth. Tumor necrosis factor-α (TNFα) is an inflammatory cytokine that is thought to contribute to lesion development. Lymphotoxin-α (LTα) is also a proinflammatory cytokine with homology to TNFα. However, its presence or function in lesion development has not been investigated. To study the role of these molecules in atherosclerosis, the expression of these cytokines in atherosclerotic lesions was examined. The presence of both cytokines was observed within aortic sinus fatty streak lesions. To determine the function of these molecules in regulating lesion growth, mice deficient for TNFα or LTα were examined for induction of atherosclerosis. Surprisingly, loss of TNFα did not alter lesion development compared with wild-type mice. This brings doubt to the generally held concept that TNFα is a “proatherogenic cytokine.” However, LTα deficiency resulted in a 62% reduction in lesion size. This demonstrates an unexpected role for LTα in promoting lesion growth. The presence of LTα was observed in aortic sinus lesions suggesting a direct role of LTα in modulating lesion growth. To determine which receptor mediated these responses, diet-induced atherosclerosis in mice deficient for each of the TNF receptors, termed p55 and p75, was examined. Results demonstrated that loss of p55 resulted in increased lesion development, but loss of p75 did not alter lesion size. The disparity in results between ligand- and receptor-deficient mice suggests there are undefined members of the TNF ligand and receptor signaling pathway involved with regulating atherogenesis.

Inflammatory processes are an integral component to atherosclerotic lesion development. Cytokine-mediated proinflammatory responses such as endothelial cell activation and leukocyte recruitment are thought to positively contribute to the atherogenic process. One of the best-studied proinflammatory cytokines is tumor necrosis factor-␣ (TNF␣) 1 that is expressed in both human and rodent atherosclerotic plaques (1)(2)(3)(4)(5). However, the physiological role of TNF ligand and receptor family members in the atherogenic process remains unclear.
TNF␣ and lymphotoxin-␣ (LT␣) are two predominant members of the TNF ligand family. Their structural genes are located on human chromosome 6 within the major histocompatibility complex (6). TNF␣ and LT␣ proteins are structurally similar and display 50% amino acid homology (7). TNF␣ is first synthesized as a type II transmembrane protein and is subsequently cleaved to form circulating homotrimeric TNF␣ (8). TNF␣ is synthesized primarily by activated macrophages (9), although under appropriate stimulation other cells can express this cytokine (10 -12). TNF␣ influences the function of macrophages, smooth muscle cells, and endothelial cells (13), which are major cell types observed in plaques. LT␣ is synthesized primarily by activated T and B lymphocytes (6,7) and is also found in the circulation as a homotrimer. Unlike TNF␣, membrane-bound homotrimeric LT␣ has not been observed. The presence or function of LT␣ in atherosclerotic lesions has not been previously investigated.
Homotrimeric TNF␣ and LT␣ elicit responses through two receptors termed p55 and p75 (6,14). The p55 receptor activates the majority of responses associated with TNF␣ including induction of adhesion molecule expression (15,16), apoptosis (17,18), and resistance to bacterial infection (19,20). In an earlier report we showed that p55 receptor deficiency in mice results in increased atherosclerotic lesion development, demonstrating that signaling through this receptor is atheroprotective (21). Activities associated with p75 activation include induction of T cell proliferation (22,23), induction of TNF␣mediated skin tissue necrosis (24), and modulation of TNF␣mediated pulmonary inflammation (25).
In this report, we investigated whether TNF␣-or LT␣-mediated responses alter lesion growth. Control mice or mice deficient for either TNF␣ or LT␣ were fed an atherogenic diet, and the presence of these ligands within the lesions was examined. Confirming other reports, we observed TNF␣ in the atherosclerotic lesions. Surprisingly though TNF␣ deficiency did not alter lesion size. This brings into question the generally held concept that TNF␣ promotes atherogenesis. Furthermore, loss of LT␣ resulted in a 3-fold decrease in lesion size. These findings demonstrate that LT␣ is the predominant member of the TNF ligand family that elicits proatherogenic responses. Since loss of LT␣ decreased lesion growth but loss of the major TNF receptor, p55, resulted in increased atherosclerosis, we hypothesized that the p75 receptor was involved with regulating LT␣ responses to promote atherogenesis. However, we show that loss of p75 did not alter lesion development. The disparity between the results obtained with ligand-deficient versus receptor-deficient mice illustrates the complexity of this cellular signaling system and suggests that there are alternative TNF ligand/receptor molecules involved with regulating lesion growth.

EXPERIMENTAL PROCEDURES
Mice-Female C57BL/6CR mice, age of 6 weeks, were purchased from Charles River Breeding Laboratories and used as the wild-type control strain for studies involving receptor-deficient mice. Mice lacking either the TNF receptor p55 (p55Ϫ/Ϫ), p75 (p75Ϫ/Ϫ), or both receptors (p55Ϫ/Ϫp75Ϫ/Ϫ) have been described previously (25). The p55Ϫ/Ϫ mice were developed directly in C57BL/6CR and are an inbred strain. The p75Ϫ/Ϫ and p55Ϫ/Ϫp75Ϫ/Ϫ animals represent 4 -5 backcrosses onto C57BL/6CR, respectively. C57BL/6J tumor necrosis factor-␣-deficient mice (TNF␣Ϫ/Ϫ) and lymphotoxin-␣-deficient mice (LT␣Ϫ/Ϫ) were purchased from The Jackson Laboratory. The C57BL/6J mice were used as the wild-type control strain for the studies involving ligand-deficient mice as both the TNF␣Ϫ/Ϫ and LT␣Ϫ/Ϫ mice are maintained on the C57BL/6J genetic background. Mice were bred to generate colonies of each gene knockout strain here at the University of Washington. F2 and F3 offspring were used for the experiments presented in this report. Apolipoprotein E-deficient male mice (apoEϪ/Ϫ) maintained on the C57BL/6J genetic background were obtained from The Jackson Laboratory. Mice were maintained in a temperature-controlled (25°C) facility with a strict 12-h light/dark cycle and given free access to food and water. Blood was collected after a 4-h fast from the retro-orbital sinus into tubes containing 1 mM EDTA, and plasma was stored at Ϫ20°C prior to analysis.
Study Design-Two experiments were performed for this study. In experiment 1, wild-type, TNF␣Ϫ/Ϫ or LT␣Ϫ/Ϫ female mice were fed an "atherogenic diet" containing 15% fat, 1.25% cholesterol, and 0.5% sodium cholate (diet No. TD90221, Harlan Teklad) (26). This diet induces fatty streak lesions in the aortas of susceptible mice (27,28). Mice were fed the diet for 16 weeks before quantifying lesion areas. In experiment 2, wild-type and TNF receptor-deficient mice were fed the atherogenic diet for 18 weeks before quantifying lesion development. For both studies, an additional set of female mice were fed a rodent chow diet (Wayne Rodent BLOX 8604, Harlan Teklad) for 16 -18 weeks prior to analyzing plasma lipids and lesion areas. ApoEϪ/Ϫ mice were maintained on a rodent chow diet until 22 weeks of age before they were evaluated for aortic sinus lesions and the presence of TNF␣ or LT␣.
Plasma Analysis-Total cholesterol and triglycerides were determined using established colorimetric assays as described (29,30) (kits 1127578 and 450032, Roche Molecular Biochemicals). Plasma lipoproteins were separated by FPLC gel filtration using a Superose 6 column (Amersham Biosciences). A 100-l aliquot of plasma from each of 3-4 mice per diet group was separated at a flow rate of 0.2 ml/min using phosphate-buffered saline. 100-l aliquots from each of 0.5-ml fractions were used for cholesterol determinations.
Aortic Sinus Lesion Quantification-Aortic sinus lesion areas were quantified as described (21,29). Hearts were removed from mice, perfused with phosphate-buffered saline, and formalin fixed using a 4% neutral formalin solution. After removing peripheral fat, the heart was sectioned directly under and parallel to the atrial leaflets. The upper section was incubated in phosphate-buffered saline containing 30% sucrose for 18 h and then embedded in O.C.T. embedding medium and frozen. Every other section (10-m thick) throughout the aortic sinus was taken for analysis. Sections were evaluated for fatty streak lesions following lipid staining with oil red O and nuclei staining using hematoxylin. Lesion area measurements were analyzed using the Optimas Image Analysis Software Package (BioScan).
Immunohistochemistry-Frozen sections were fixed in acetone and endogenous peroxidase quenched by incubating slides in 3% hydrogen peroxide. Samples were then incubated with either a polyclonal anti-TNF␣ antibody (RDI-mTNFAabrP, Research Diagnostics, Inc.) at 1:30 dilution or with a polyclonal anti-LT␣ antibody (RDI-TNFBabr1, Research Diagnostics, Inc.) at 1:30 dilution. After rinsing samples were incubated with a biotinylated secondary antibody (BA1000, Vector Laboratories) at 1:200 and binding detected using streptavidin-horseradish peroxidase followed by AEC colorimetric product formation using the Histomouse kit (95-9541, Zymed Laboratories Inc.). To stain for macrophages, a rat anti-mouse CD11b antibody was used (01711D, BD PharMingen) at 1:00 dilution. To stain for T cells, a rat anti-mouse CD3 antibody was used (28001D, BD PharMingen) at 1:100 dilution. Binding for CD11b and CD3 was detected using a biotinylated secondary antibody (B7139, Sigma-Aldrich) followed by streptavidin-horseradish peroxidase and AEC detection as described above. Deletion of either the primary or secondary antibody resulted in minimal to no staining.
Statistical Analysis-Values are reported as mean Ϯ S.E. Nonparametric Wilcoxon signed ranks tests were used to determine differences in lesion areas. The Student's t test was used to compare independent means in some cases. p Ͻ 0.05 was accepted as statistically significant.

LT␣ Is Expressed in Atherosclerotic
Lesions-Aortic sinus lesions from atherogenic diet-fed wild-type female mice and chow-fed apoEϪ/Ϫ male mice were evaluated for the expression of TNF␣ and LT␣. Lesions from atherogenic diet-fed mice are comprised predominantly of macrophages with small amounts of T and B cells present (31)(32)(33). In contrast, lesions from apoEϪ/Ϫ mice contain macrophages, smooth muscle cells, and T and B cells (34,35). Thus, these two systems provide examples of early simple fatty streak lesions and more complex lesions as observed in human atherosclerosis (36). TNF␣ immunostaining was observed in lesions from both of these models (Fig. 1). Staining was punctate indicating cell-associated protein expression and was also observed within the medial smooth muscle cell layer under the lesioned sites. Surprisingly however, LT␣ staining was also observed in lesions from both of these models (Fig. 1). Immunostaining was observed throughout the lesions and medial layers. In general, the staining was more diffuse than the staining observed for TNF␣. We next confirmed that the immunoreactivity we observed did not simply reflect cross-reactivity of these antibodies to the different ligands. Lesions from atherogenic diet-fed TNF␣Ϫ/Ϫ and LT␣Ϫ/Ϫ mice were immunostained for TNF␣ and LT␣ (Fig. 2). Results demonstrate that lesions from TNF␣Ϫ/Ϫ mice did not show immunoreactivity against the TNF␣ antibody, and lesions from LT␣Ϫ/Ϫ mice did not show reactivity against the LT␣ antibody. Furthermore, loss of one ligand did not impede the expression of the other ligand within these lesions.
B and T cells but not monocyte/macrophages are the predominant cell types to secrete LT␣ (37). We investigated whether LT␣ expression was limited to a specific cell type within the lesions. Lesions from apoEϪ/Ϫ mice were immunostained for macrophages using the CD11b marker, for T cells using the CD3 marker, and for LT␣ (Fig. 3). Results show that lesions from apoEϪ/Ϫ mice contain both of these cell types and that LT␣ expression is not confined to a specific cell type. The diffuse nature of LT␣ immunostaining suggests that this protein is secreted from one of these cell types and has become trapped within the extracellular matrix and necrotic regions of these lesions. Further investigation about the cellular source of LT␣ within the lesion remains to be determined.
Loss of LT␣ but Not TNF␣ Decreases Atherosclerosis in C57BL/6 Mice-To determine whether TNF␣ or LT␣ has a physiological role in lesion development we determined whether loss of these genes would influence diet-induced atherosclerosis. Aortic lesion areas were quantified for wild-type, TNF␣Ϫ/Ϫ and LT␣Ϫ/Ϫ female mice fed the atherogenic diet for 16 weeks (Fig. 4). Surprisingly, loss of TNF␣ did not alter lesions sizes (10.9 Ϯ 2.4 m 2 ϫ 10 3 , n ϭ 18) as compared with wild-type mice (11.5 Ϯ 2.3 m 2 ϫ 10 3 , n ϭ 11). In contrast, lesion areas in LT␣Ϫ/Ϫ mice were reduced (4.3 Ϯ 1.4 m 2 ϫ  (top and bottom, respectively). Lesions were stained for lipids using oil red O (ORO, A and B), for TNF␣ (C and D), or for LT␣ (E and F). Original magnification was ϫ200. 10 3 , n ϭ 13 and p ϭ 0.0128). These data are consistent with the idea that LT␣ is the primary TNF family ligand involved with atherosclerosis.
Lipid Levels Are Lower in LT␣Ϫ/Ϫ Mice-To understand why lesion areas were reduced in LT␣Ϫ/Ϫ mice, plasma lipid and lipoprotein profiles were analyzed. As compared with atherogenic diet-fed wild-type mice, cholesterol levels were 46% higher in TNF␣Ϫ/Ϫ mice (p ϭ 0.10 versus wild-type mice) and 20% lower in LT␣Ϫ/Ϫ mice (p ϭ 0.05). Final values were 189 Ϯ 19 mg/dl for wild-type, 276 Ϯ 42 mg/dl for TNF␣Ϫ/Ϫ, and 151 Ϯ 6 mg/dl for LT␣Ϫ/Ϫ mice. Total cholesterol levels did not correlate with lesion areas for any of the strains fed the atherogenic diet. Triglyceride levels were ϳ11-16 mg/dl for all atherogenic diet-fed mice, and no significant differences were observed among strains.
To determine whether lipids were redistributed into different lipoprotein particles in atherogenic diet-fed TNF␣Ϫ/Ϫ or LT␣Ϫ/Ϫ mice, FPLC profiles of lipoproteins were examined (Fig. 5). No differences in the proportion of cholesterol in the VLDL/LDL to HDL fractions was observed between wild-type and TNF␣Ϫ/Ϫ mice. That is, the increase in total cholesterol levels observed in TNF␣Ϫ/Ϫ mice reflects a proportional increase in both VLDL/LDL and HDL lipoprotein fractions. In contrast the relative amount of total cholesterol found in the HDL fraction tended to be increased for LT␣Ϫ/Ϫ mice. The combination of lower total cholesterol and higher relative amounts of cholesterol found in the HDL fraction likely accounts at least partially for the reduced lesion development observed for LT␣Ϫ/Ϫ mice.
Loss of p55 but Not p75 Increased Atherosclerosis in C57BL/6 Mice-Mice deficient for p55 receptors display a 2.3-fold increase in diet-induced atherosclerosis as compared with wildtype mice (21). Since neither the LT␣Ϫ/Ϫ nor the TNF␣Ϫ/Ϫ mice recapitulated these results, we hypothesized that signaling via the p75 receptor influences lesion development. Wildtype, p55Ϫ/Ϫ, p75Ϫ/Ϫ, and p55Ϫ/Ϫp75Ϫ/Ϫ female mice were fed the atherogenic diet for 18 weeks, and aortic sinus lesion areas were quantified (Fig. 6). Results demonstrate that loss of p55 resulted in a 2.4-fold increase in lesion size, confirming our previous result (21). Loss of p75 did not alter lesion development, and loss of both p55 and p75 receptors resulted in lesion areas comparable with those observed for p55Ϫ/Ϫ mice. No significant differences in plasma total cholesterol, HDL cholesterol, or triglycerides were observed between genotypes (data not shown). Thus the p55 receptor signals events that retard lesion development, whereas p75 signaling does not influence lesion development. DISCUSSION This study provides several important new findings about the role of TNF signaling pathways in regulating atherosclerotic lesion development. This is the first report demonstrating that LT␣ is expressed in atherosclerotic lesions and that loss of this cytokine reduces lesion size. Surprisingly, loss of TNF␣, which is involved with numerous proinflammatory responses, did not alter lesion development in atherogenic diet-fed mice. Loss of p55 receptors, but not p75 receptors, resulted in increased diet-induced atherosclerosis showing that the p55 receptor has the predominant role in regulating lesion growth. Taken together these results illustrate the complexity of TNF ligand and receptor interactions in modulating inflammatory responses such as those observed during lesion growth. Furthermore, since the ligand deficiency did not recapitulate the responses we observed with the receptor deficiency it suggests that there are undefined members of the TNF ligand or receptor signaling pathway involved with regulating atherogenesis.
The observations that LT␣ is expressed within atherosclerotic lesions and deficiency of this protein retards lesion development suggests that there may be a direct function of LT␣ in  (top and bottom, respectively). Lesions were stained for lipids using oil red O (ORO, A and B), for TNF␣ (C and D), or for LT␣ (E and F). Original magnification was ϫ200.  promoting atherogenesis. LT␣ is produced primarily from T and B cells (37). Both of these cell types have been identified in atherosclerotic lesions with the extent of their presence dependent on the animal model and on the stage of lesion size or complexity (35, 38 -41). The role of these cell types in lesion development has been intensively studied. For example RAGϪ/Ϫ mice have impaired T and B cell function such that T lymphocytes do not mature into CD4 ϩ helper or CD8 ϩ suppressor cells, and B cells are unable to synthesize immunoglobulin (42). When atherosclerosis-susceptible apoEϪ/Ϫ mice were crossed with RAG-1Ϫ/Ϫ mice a 2-fold decrease in atherosclerotic lesions was observed, suggesting that T and B cell function promotes lesion growth (43). However, this response was not observed when RAG-2Ϫ/Ϫ mice were studied (44). When T cell-deficient nude (nu/nu) mice were fed an atherogenic diet lesion areas were reduced 90% as compared with control mice (45). Therefore, most but not all studies implicate a proatherogenic role for lymphocytes. Our results are consistent with this concept and suggest that T and/or B cells are secreting LT␣ within the developing lesion to promote atherogenesis.
Several important functions of LT␣ have recently been identified that demonstrate a significant role for this cytokine in lymphocyte activation and proliferation. LT␣ promotes inflammatory and chemoattractant responses (46 -48) and B cell proliferation (49). LT␣ is also involved with CD8 ϩ T cell activation (50). LT␣-deficient mice have altered immune function including an absence of peripheral lymph nodes, abnormal Peyer's patches, and no germinal centers (51,52). The decreased atherosclerosis we observed in the LT␣Ϫ/Ϫ mice supports the concept that normal leukocyte activity promotes atherogenesis. By identifying which LT␣-mediated processes are actually proatherogenic we will have a more focused target for designing antiatherogenic therapies.
The loss of LT␣ also resulted in improving total cholesterol levels and lipoprotein profiles. These findings suggest that a primary mechanism of reduced lesion growth in the LT␣Ϫ/Ϫ mice may be through changes in plasma lipid levels. The influence of LT␣ in regulating plasma lipid levels has not yet been investigated but should provide us with new insights about how cytokines influence lipid metabolism.
Our results suggest that LT␣ may be signaling through receptors unique from the p55 or p75 receptor to promote atherogenesis. LT␣ is expressed in two forms. It is found in the circulation as a homotrimer, or it can form heterotrimers with membrane-bound LT␤ a third member of the TNF ligand family (7,53). In fact Mackay et al. (54) have shown that LT␣ homotrimers displayed a 50 -200-fold lower K d for p55 receptor binding as compared with TNF␣. This suggests that circulating LT␣ is less efficient at activating the p55 receptor. In contrast LT␣LT␤ heterotrimers effectively bind and activate the LT␤ receptor to induce inflammatory and cytotoxic responses (54 -56). We propose that the proatherogenic responses mediated by LT␣ result from signaling events mediated primarily by the LT␤ receptor. The LT␤ receptor is expressed on multiple tissues and has a distribution pattern similar to that observed for the p55 receptor (7). Inhibition of this pathway by deleting the LT␣ gene may reduce LT␤ receptor-induced inflammatory events resulting in reduced lesion development.
The lack of effect of TNF␣ deficiency on lesion development was surprising given the many roles of TNF␣ on mediating proliferative and inflammatory responses (13). Our findings suggest that the presence of TNF␣ within atherosclerotic lesions may simply represent a marker of inflammatory responses but may not be the actual inducing mediator of inflammatory events within the growing lesion. These findings lead us to question the commonly held notion that the presence of TNF in lesions signifies proatherogenic responses. In fact, based on our findings we suggest that caution be used in attributing TNF␣ presence to a deteriorated atherosclerotic environment.
The observation that p55 receptors but not p75 receptors are involved with regulating lesion growth is consistent with the concept that p55 is the primary receptor for mediating many TNF ligand responses. Signaling via the p55 receptor is associated with induction of adhesion molecule expression (15,16), apoptosis (17,18), and leukocyte chemotaxis (15). In contrast p75 signaling is associated with activation induced cell death of T cells (57), TNF-mediated skin necrosis (24), and suppression of TNF-mediated inflammatory responses (25). The findings presented here are consistent with our earlier report (21) demonstrating that p55 signaling attenuates lesion growth. The actual events elicited by p55 signaling and the cells involved with this response remain undefined. However, because TNF␣Ϫ/Ϫ mice did not recapitulate these findings the responses may be generated independently of TNF ligand binding or indicate that there are other unidentified ligands mediating p55 atheroprotective responses.
In conclusion we show that members of the TNF ligand and receptor superfamily are involved with regulating lesion growth in mice fed high amounts of cholesterol. The results presented here demonstrate that there are multiple members of this ligand/receptor system involved with regulating events in the atherogenic process. The disparate results obtained between ligand versus receptor-deficient mice reflects the complex interaction between the members of this system. By separating the function of each member in lesion growth a clear target for pharmacological intervention will be achieved.