Detection and localization of tumor necrosis factor in human atheroma

Detection and Localization of Tumor Necrosis Factor in Human Atheroma Peter Barath, MD, PhD, Michael C. Fishbein, MD, Jin Cao, MD, James Berenson, MD, Richard H. Helfant, MD, and James S. Forrester. MD

Tumor necrosis factor (TNF) is a secretory product of normal macrophages that can cause cell necrosis, new blood vessel formation and thrombosis. These are also 3 characteristic features of the progression of stable atheroma to endothelial disruption. Accordingly, an immunohistochemical method was developed to detect TNF in human tissue. Udng this method TNF positivity was demonstrated in 57 of 6S (66%) of tissue secttons classified as atherosclerotic and in 5 of 11(4S%) sections dasstfled as minimally atherosclerotic. TNF was absent in 6 sections classified as normal. TNF posltlvlty was found not only in the cytoplasm of macrophages, but also in the cytoplasm and attached to the cell membrane of smooth muscle cells and endothelial cells of the human atheroma. Because TNF is known to cause new vessel fermation, hemorrhagic necrosis and increased thrombogenecity, it may play a role in the evolution of uncomplicated to complex atheroma. (Am J Cardioi 193@65:297-302)

umor necrosis factor (TNF), which is produced principally by activated macrophages,1-6 activates endothelial cell~,‘-~~ stimulates angiogenesis3 and induces hemorrhagic necrosis.4,5 Because central necrosis and new vessel formation in the presence of macrophage accumulation characterize evolving atheroma l lo* we hypothesized that TNF might be detected in theie lesions. There are no published reports of morphologic localization of TNF; the purpose of this study, therefore, was to develop a method to detect and localize immunoreactive TNF in atherosclerotic human blood vessels, and to analyze its distribution in human atheroma.

T

METHODS

We used arteries from freshly amputated legs and autopsies. There were 50 anterior and posterior tibial, femoral, carotid and coronary vessels. Exclusion criteria were coexisting neoplasia, immunologic disease and acute or chronic infectious diseases. Tissue was fmed in 10% neutral buffered formalin. Vessels taken from amputated legs were fixed within minutes of amputation; autopsy vessels were fixed within 6 hours. We made 82 sets of paraffin-embedded tissue sections: 12 from tibia1 arteries, 60 from coronary arteries and 10 from carotid and femoral arteries (Table I). Each set consisted of 6 consecutive sections. The first section was stained by hematoxylin and eosin, the second by immunohistochemistry for TNF, and then 4 sections were made for immunohistochemical identification of cell types. Each hematoxylin and eosin-stained section was classified as normal, having intimal thickening or being significantly atherosclerotic. We defined endothelial ulceration as rupture of the fibrous cap of the atheroma or presence of in vivo thrombus. Because postmortem change can cause loss of continuity changes on the endothelial surface, this finding alone was not defined as endothelial ulceration. The histologic classification was made independent of the subsequent immunohistochemical findings. We also performed immunohistochemical staining on a number of other tissue samples. As a positive control, we stained 5 necrotizing colon cancers obtained at surgical resection. As negative control we stained 5 samples from brain frontal lobe cortex. Immunehistochemicalidentiftcation of turner necrosis factor: After deparaffinization and dehydration, the

From the Division of Cardiology, Cedars-Sinai Medical Center, Los Angeles, California. Manuscript received June 7, 1989; revised manuscript received September 25, 1989, and accepted September 26. Address for reprints: Peter Barath, MD, PhD, Cedars-Sinai Medical Center, Division of Cardiology, 8700 Beverly Boulevard, Los Angeles, California 90048.

sections were incubated in 3% methanolic peroxidase for 10 minutes at room temperature, followed by trypsin digestion (0.1% of type III bovine pancreas trypsin in 0.5 M Tris buffer, pH 7.8) at 37’C for 10 minutes. Sections were incubated for 10 minutes in bovine serum

THE AMERICAN

JOURNAL

OF CARDIOLOGY

FEBRUARY

1, 1990

297

diluted 1:20 in phosphate-buffered saline (PBS) (0.1 M phosphate-buffered 0.9% saline, pH 7.4) in a humidity chamber at room temperature. The antiserum used for immunohistochemistry of TNF was a monoclonal antibody produced against recombinant human TNF-alpha (rhTNF) (gift of Dr. M. Narachi, Amgen). The antibody was purified with protein A-Sepharose chromatography.13 We used 1:lOOO to

298

THE AMERICAN

JOURNAL

OF CARDIOLOGY

VOLUME

65

1:2000 dilution of the primary antibody with 1:20 swine serum:PBS and an incubation time of 30 minutes at room temperature in a humidity chamber. The secondary antibody, rabbit antiserum to mouse immunoglobulin G (Dako), was diluted to 150 with swine serum:PBS. After PBS washing we incubated the sections in swine antiserum to rabbit immunoglobulin G conjugated with horseradish peroxidase (Dako). Sections

were then washed in PBS and stained with 0.05% 3.3’diaminobenzidine (DAB, Sigma). Sections were counterstained with hematoxylin. We used 4 types of negative controls: the primary TNF antiserum deleted from the staining procedure; the primary antiserum absorbed with excess rhTNF; the primary antiserum replaced with an antiserum produced in mouse but directed to rabbit immunoglobulin; and brain tissue (considered to be negative for TNF). We used necrotic colon carcinomas as positive controls, and corroborated the presence of TNF in these lesions by enzyme-linked immunosorbent assay (ELISA). Immunohistochemical

identification

immunosorbent

assay

Normal

lntimal Thickening

Arteriosclerotic

Total

Coronary Arteries Sections

arteries* 2 2

5 5

27 53

34 60

Peripheral Arteries Sections

arteries 4* 4*

6* 6’

6+ 12+

16 22

* Autopsy, tamputaton

of cell types:

The identity of individual cells within the atheroma was established by cell-specific immunohistochemical stains performed on sections adjacent to those stained for TNF. For identification of macrophages HAM56 (Enzo Biochem Inc.), for vascular smooth muscle cells HHF35 (gift of Dr. A. M. Gown, Department of Pathology, University of Washington), and for T lymphocytes, Pan T (Tl 1, Dako) monoclonal antibodies were used. For endothelial cells antihuman factor VIII-related rabbit antiserum (Dako) was used. For studies using monoclonal antibodies we used the above described system; for factor-VIII related antigen we used the PAP (peroxidase-antiperoxidase, Dako) system.14 Test of monoclonal antibody specificity: Human recombinant TNF alpha from 2 different sources (Amgen and T Cell Sciences, Inc.) underwent electrophoresis according to Laemmli.15 The bands were visualized with Coomassie brilliant blue. The proteins were transferred to nitrocellulose sheetsi in a Novex Western transfer apparatus (Novel). Immunoblotting17 was performed making cross-reaction between the rhTNF-s and monoclonal antibodies of 2 sources: Amgen (the antibody used in the immunohistochemical reactions) and T Cell Sciences. Enzyme-linked tumor and brain

TABLE I Frequency Distribution of Vessels and Sections Analyzed

of vascular,

tissue: To establish that TNF was present in measurable quantity in tissues with TNF positivity, we extracted the protein from one of the colon tumors and from 2 of the carotid arteries that were immunohistochemically positive for TNF. The tissues were homogenized directly into 10% volume/weight of sodium dodecyl sulfate sample buffer containing 2%

beta-mercaptoethanol. A commercially available TNF ELISA kit (Biokine, T Cell Sciences, Inc.) was used to measure the TNF concentration in the extracted tissue. The plates were read at varying time periods on a Perkin Elmer reader at 490 nm wavelength. Results were expressed in ng/mg wet weight. Data were analyzed with a Perkin Elmer analytical database software on a PC2 computer. Induction of immunoreactive tumor necrosis factor expresskw by low density lipoprotein incubation in vascular smooth muscle cells in cutture: Aortic smooth

muscle cells (from normal human aorta collected from heart explant) were isolated by combination of collagenase (type CLS II) and elastase (type I) digestion and mechanical dissection. The cells were incubated in medium 199 containing 20% fetal bovine serum (both from Gibco) at 37’C. Confluent primary cultures were incubated with 100 pg/ml LDL (Sigma) for 24 hours. Cultures without LDL incubation served as controls. The cultures were stained to detect TNF by the already described immunohistochemical method using the same internal controls. Statistical analysis: Data were analyzed using 2 X 2 contingency tables with the Fisher’s exact test. RESULTS Sensitivity, speciffcity and cross-reactivity of the monochal antibody: Our monoclonal antibody reacted

with a pure 17-18 kd rhTNF alpha from a second independent source in Western blot analysis. Indirect support that our antibody was recognizing TNF alpha was that in immunohistochemically positive arteries and tumors we could also detect by ELISA, whereas there was

complicated plaque) we observed the same pattern of increasing TNF positivity as the ulcerated plaque area was approached. lntimal We also tried to determine if TNF positivity was Atherosclerotic Thickening Normal a generalized vascular phenomenon. In TNF-positive Smooth muscle cells 43/65 3/11 O/6 samples, we examined both adjacent veins and sections Macrophages 31/65 o/11 ‘J/f5 of the gastrocnemius containing normal small muscular Endothelial cells 12/65 2/11 O/6 arteries. None of the sections was positive. Tissue section 57/65 3/11 ‘J/6 CeAular distribution of tumor necrosis factor: TNF The differences between normal and diseased vessels were statistically significant. In the atherosclerotic category the diierence in prevalence between both smooth positive staining was found in 3 types of cells: smooth muscle cells and macrophages versus endothelial cells was statistically significant. muscle cells, endothelial cells and macrophages. Smooth muscle cells rather than macrophages were the most no detectable TNF present in immunohistochemically frequently positive. Figure lE, a magnified portion of negative brain tissue. T lymphocytes (as identified by Figure lD, reveals that positive smooth muscle cells Tl l), source of TNF beta, did not show immunohistowere found in the media, vasa vasora and newly formed chemical positivity for TNF alpha, suggesting that vessels. The distribution and extent of smooth muscle TNF beta was not a cross-reactor with our antibody. cell positivity varied from scattered to confluent. As a positive control, we used colon cancer samples, Table II summarizes the frequency of TNF positiviwhere TNF is frequently present in high quantity. Fig- ty in the 3 cell types. Smooth muscle cells were TNF ure 1A shows the immunohistochemical detection of positive in 43 of 65 (66%) of the sections classified as TNF in a section of a colonic carcinoma. Although the significantly atherosclerotic by hematoxylin and eosin positive TNF stain in macrophages was expected, some staining, and 3 of 11 (27%) segments classified as intiof the cancer cells were also found to stain positively. ma1 thickening (p
300

THE AMERICAN

JOURNAL

in Three Different Categories

OF CARDIOLOGY

VOLUME

Cell

65

Figure 1J and K show a high power view of the TNF positive macrophages. TNF positive macrophages were found both in adventitial inflammatory infiltrates and in the atheroma as foam cells. In contrast to the surface localization of TNF in smooth muscle cells, TNF positivity was always found in the cytoplasm of the macrophages. DISCUSSION

The most significant finding in this study is the discovery of TNF in atherosclerotic arteries. We also report for the first time that TNF appears in both smooth

muscle and endothelial cells, using the immunohistochemical technique that we developed. To document the specificity of the TNF antibody, we performed Western blotting with rhTNF alpha, and measured TNF by ELISA on extracted fresh vascular homogenates. The tissue extract study revealed a measurable amount of TNF. For negative controls we used a standard approach. Our immunohistochemical negative controls (antibody-deleted and TNF-negative tissue) did not show any immunohistochemical staining. Macrophages are known as the principal source of TNF. The macrophage is also a prominent cell type in

flGURE 2.khMcath of TNF positive intimal ceNs as smooth muscle ce& by HHF35, an alpha and gamma actin specitk monodo~l antibody in v sections. hnmunohistochemii staining used indii peroxidase-labeled antibody techkpe. lmmunonactive TNF and actin appears in brown. Hematoxylin nuclear staining. A, group of inthnal (stellate) cells (5) with cytoplasmic positlvity for TNF (X 550). 8, group of intimal (stellate) cells (5) in an adjacent section to A staii by HHF35. same group of cells shows cytoplasmic HHF35 positivity, suppohng the smooth muscle character of these cells (X 550).

FtGURE 3. Expra~M with 100 &ml LDL bow inahtion with these cells (X 1,000).

The

of imbue TNF (in brown) in primary vascular smooth muscle cell adture after inahath of cells fur 24 hours. lmmunohistochemii staining for TNF and hematoxylin nuclear timing. A, culture after 24LDL. The cytoplasm of the cek appears in brown, supporting that im munoreactlve TNF is expressed in 6, cutWe without LDL incubation are TNF negative (lack of brown colar in cytoplasm) (X 1,000).

THE AMERICAN

JOURNAL

OF CARDIOLOGY

FEBRUARY

1. 1990

301

the necrotic core of atheroma.18J9 There are at least 2 stimuli that, in theory, might induce a tissue macrophage in an atheroma to produce TNF. These stimuli, lipid ingestion and decreased oxygen tension, both cause cultured monocytes to produce TNF (P. Barath, unpublished data). The possibility that postmortem anoxia stimulated TNF production by macrophages seems unlikely for 2 reasons. First, there was no postmortem TNF positivity in the macrophages of normal arteries. Second, fresh tissue obtained at surgery was strongly positive for immunoreactive TNF. The cytoplasmic staining suggests that in atheroma the macrophage is a source of TNF. In contrast to the macrophage data, the finding of TNF positivity in smooth muscle and endothelial cells was not anticipated. Although a number of mesenchyma1 and epithelial cell lines produce TNF,20-22 TNF positivity has not previously been detected in situ vascular smooth muscle or endothelial cells. The peripheral staining pattern may represent receptor-ligand interaction, endocytosis or extracellular ma&rix binding. Our method of preparation and the resdlution of light microscopy do not allow us to separate these possibilities, but it may represent membrane-associated TNF recently described in monocytes. 23The localization of TNF in the cytoplasm of some smooth muscle cells suggests that these cells also may be capable of producing TNF. Recently, capability of smooth muscle cells to express the TNF gene has been demonstrated in cultures by Northern analysis24 and in human vascular tissue by in situ hybridization (P. Barath, submitted for publication). In a pilot study we demonstrated that LDL uptake induces the expression of immunoreactive TNF in vascular smooth muscle cell culture. This mechanism may be operative in natural induction of TNF expression during atheroma evolution. Endothelial cell TNF positivity at the site of new vessel formation is potentially important, because neovascularization is a prominent part of atheroma evolution, and TNF is an unusually potent stimulus to new vessel formation. Potential rekvance to elkal coronary disemw: We have demonstrated a high occurrence of endothelial ulceration and thrombus formation in coronary arteries of unstable angina patients.25*26 The demonstration of TNF in human atheroma, combined with its known biologic effects in other tissues, suggests some speculation as to its biologic role. TNF is a powerful angiogenic factor, even in low concentration, and is capable of causing necrosis within 24 to 48 hours in cells that are sensitive to its actionsS2 Both new vessel formation and central necrosis are known to precede atheromatous endothelial disruption.i1J2 Thus, while our data do not establish a causeeffect relation, the presence of TNF and its previously established cellular actions suggest that it could be involved in the evolution of atheroma.

302

THE AMERICAN

JOURNAL

OF CARDIOLOGY

VOLUME 65

REFERENCES 1.Mannel D, Moore R, Magenhagen

S. Macrophages as a source of tumoricidal activity (tumor necrosis factor). Infect Immun 1980;30:523-530. 2.Beutler BA, Cerami A. Cache&n and tumor necrosis factor as two sides of the same biological coin. Nature 1986;320:584-588. 3.Leibovich SJ, Polverini PJ, Shipard HM, Wiseman DM, Shiveley V, Nuseir N. Macrophage-induced angiogeneais is mediated by tumor necrosis factor. Narure 1987:329:630-632. 4. Kawai T, Satomi N, Sato N, Sakurai A, Haranaka K, Goto T, Suzuki M. Necrotizing activity of tumor necrosis factor: histopathological investigation using Meth A sarcoma and granulation tissue. Virchows Arch B 1987;53:353-358. 5. Watanabe N, Nitsu Y, Umeno H, Kuriyama H, Neda H, Yamauchi N, Meada M, Urushazaki I. Toxic effect of tumor necrosis factor on tumor vasculature in mice. Cancer Res 1988;48:2179-2183. 6. Van der Meer JWM, Endrea S, Lonnemann G, Cannon JG, Ikejima T, Oktisawa S, Gelfand JA, Dinarello CA. Concentrations of immunoreactive human tumor necrosis factor alpha produced by human mononuclear cells in vitro. J Leukocyte Biol 1988;43:216-223. 7. Nawroth PP. Stern DM. Modulation of endothelial cell hemostatic properties by tumor necrosis factor. Exp Med 1986;740-745. 8. Schleef RR, Bevilaqua MP, Sawdey M, Gimbrone MA Jr, Loskutoff J. Cytokine activation of vascular endothelium. Effects on tissuetype plasminogen activator and type 1 plasminogen activator inhibitor. J Biol Cbem 1988;263: 5797-5803. 9. Bevilaqua MP, Pober JS, Majeau GR, Fiers W, Cotran R, Gimbrone MA Jr. Recombinant tumor necrosis factor induces procoagulant activity in cultured human vascular endothelium: characterization and comparison with the actions of interleukin 1. Proc Acad Sci USA 1986;83:4533-4537. 10. Cybulsky MI, Chan MKW, Movat HZ. Biology of disease. Acute inflammation and microthrombosis induced by endotoxin, interleukin-1, and tumor necrosis factor and their implication in gram-negative infection. Lab Iwest 1988;58:365378. Il. Barger AC, Beeuwkes R, Lainey LL, Silverman KJ. Hypothesis: vasa vasorum and neovascularization of human coronary arteries. A possible role in pathophysiology of atherosclerosis. N Eng/ J Med 1984;310:175-177. 12. Falk E. Plaque rupture with severe pre-existing stenosis precipitating corenary thrombaris: characteristics of coronary atherosclerotic plaque underlying fatal occlusive thrombi. Br Heart J 1983;50:127-134. 13. Ey PL, Prowse SJ, Jet&in CR. Isolation of pure IgG, IgGh and IgG2b immunoglobulins from mouse serum using protein A-sepharose. Immumxhemistry 1978;15:429-436. 14. Stemberger LA. Immunocytcchemistry: Second Edition. New York: John Wifey and Sons, 1979. IS. Laemmli UK. Cleavage of structural proteins during the assembly of the head of bacteriophage T4. Nature 1970:227:680-685. 16.Towbin H, Staehelin T, Gordon J. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc Narl Acad Sci USA 1979;76:4350-4354. 17. Burnette WH. Western blotting: Electrophoretic transfer of proteins from SDS-polyacrylamide gels to unmodified nitrocellulose and radiographic detection with antibody and radioiodinated protein A. Anal Biorhem 1981;112:195-203. 18.Ross R, Wight TN, Strandneas E, Thiele B. Human atherosclerosis. I. Cell constitution and characteristics of advanced lesions of the superficial femoral artery. Am J Pathol 1984;l l4:79-93. 19.Gown AM, Tsukada T, Ross R. Human atherosclerosis. Immunohitochemical analysis of cellular composition of human atherosclerotic lesions. Am J Parho/ 1986;125:191-207. 20. Steffen M, Ottmann OAG, Moore MA. Simultaneous production of tumor necrosis factor-alpha and lymphotoxin by normal T cells after induction with IL-2 and anti-T3. J Immunoll988;140:2261-2624. 21.Spriggs DR, Imamura K, Rodriguez C, Sariban E, Kufe DW. Tumor necrcsis factor expression in human epithelial cell lines. J Clin Invest 1988;81:14551460. 22. Pennica D, Nedwin GS, Hayflick JS, Seeburg PH, Derinck R, Palladino MA, Kohr WJ, Aggarwal BB, Gwddel DV. Human tumor necrosis: precursor structure, expression and homology to lymphotoxin. Nature 1984;312:724-729. 23. Chensue SW, Remick DG, Shmyr-Forsch C, Beals TF, Kunkel SL. Immune+ histwhemical demonstration of cytoplasmic and membrane-associated tumor necrosis factor in murine macrophagea. J Pathol 1988;133:564-572. 24. Warner SJC, Libby P. Human vascular smooth muscle cells. Target and source of tumor necrosis factor. J Immunol 1989;142:100-109. 25. Sherman CT, Litvack F, Grundfest W, Lee M, Hickey A, Chaux A, Kass R, Blanche C, Kass R, Matloff J, Morgenstern L, Ganz W, Swan HJC, Forrester SJ. Demonstration of thrombus and complex atheroma by in viva angioscopy in patients with unstable angina pectoris. N Engl J Med 1986;315:913-919. 26. Forrester JS, Litvack F. Grundfest W, Hickey A. A perspective of coronary disease seen through the arteries of living man. Circularion 1987;75:505-513.