Interleukin-l and Tumor Necrosis Factor Synergistically Stimulate Lung Fibroblast Interleukin-lo Production Jack A. Elias and Margaret M. Reynolds Pulmonary Section, Department ofInternal Medicine, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania
We determined whether normal human lung fibroblasts expressed cell-associated thymocyte-stimulating activity in response to recombinant interleukin-l (rIL-l) (a and (3) and recombinant tumor necrosis factor (rTNF). Individually, rIL-l and rTNF induced fibroblast expression of thymocyte-stimulating activity, with rIL-l being significantly more potent. Importantly, combining rIL-l and rTNF resulted in a synergistic increase in fibroblast thymocyte-stimulating activity. This synergistic interaction was dose dependent for both cytokines and was not noted when '"(-interferon was combined with rIL-l or rTNF. In all cases, the thymocyte-stimulating activity was the result of an IL-la-like moiety whose maximal production required protein synthesis. IL-la activity could be detected after as little as 4 h, peaked after 24 h, and returned toward normal with longer periods of cytokine-fibroblast incubation. However, cytokinestimulated fibroblasts that no longer expressed IL-la activity could be induced to re-express this activity with repeat cytokine challenge. Induction of fibroblast IL-la by IL-l and/or TNF may be an important mechanism amplifying IL-l-mediated biologic events at sites of local inflammation.
The interleukin-l (IL-l) family of proteins are important regulators of inflammation, fibrosis, coagulation, and hemostasis (1, 2). The most studied member of this family is IL-l (3 , which was initially described as a soluble product of stimulated mononuclear phagocytes (3). It is now known that IL1(3 can be produced by a wide variety of nucleated cells (4-7). It is also known that the biologic effects of IL-I are mediated by complex networks of interacting cytokines in which cytokines regulate target tissues directly, that cytokines in combination have quantitatively and/or qualitatively different effects than they have individually (8-10), and that cytokines induce target tissues to produce bioactive cytokines (4, 5, 11, 12). Particularly important in this regard are the demonstrations from this (8, 13, 14) and other laboratories (15-17) that IL-l and tumor necrosis factor (TNF) can synergize in regulating target cell function and that IL-l can induce fibroblasts (14), endothelial cells (5), monocytes (18), and smooth muscle cells (4) to produce IL-l(3-like moieties. IL-la, the other major member of the IL-l family, is produced by appropriately stimulated mononuclear phagocytes (19, 20), B lymphocytes (21), dendritic cells (22), keratino-
Key Words: interleukin-1, tumor necrosis factor, fibroblast (Received in original form October 3, 1989 and in revised form February 7, 1990) Address correspondence to: Jack A. Elias, M.D., Pulmonary and Critical Care Medicine, Department of Internal Medicine, Yale University School of Medicine, 105 LCI, P. O. Box 3333, 333 Cedar Street, New Haven, CT 06510-8757. Abbreviations: interleukin, IL; recombinant, r; tumor necrosis factor, TNF; thymocyte-stimulating activity, TSA. Am. J. Respir. Cell Mol. BioI. Vol. 3. pp. 13-20, 1990
cytes (23), and fibroblasts (24). It has a similar spectrum of biologic activity and shares a common receptor with IL-l(3 (1, 2). However, IL-la and IL-l(3 may serve different functions in the inflanunatory response since their production can be differentially regulated (25, 26) and they are found in different locations in cells and supernatants (19, 20). The role IL-la plays in the cytokine networks that regulate biologic homeostasis has not been as extensively studied as that ofIL-l(3. Specifically, it is not known whether IL-l induces nonmononuclear phagocytes to produce functional IL-Ia and whether TNF alters this inductive effect. To address these issues, we determined whether recombinant (r) IL-l and rTNF caused fibroblasts to express a cell-associated thymocyte-stimulating activity (TSA) and investigated the mediator(s) of this effect. These studies demonstrate that rIL-l (a and (3) and rTNF individually stimulate fibroblasts to express a cell-associated TSA and that combining rIL-l and rTNF results in a synergistic increase in the expression of this activity. They also demonstrate that this cell-associated TSA is largely mediated by an It-Io-Iike cytokine.
Materials and Methods Cells Three adult human lung fibroblast strains were used in these studies. Two were established from histologically normal areas of lungs removed at surgery for standard clinical reasons. The techniques used to prepare these strains and their proliferative and biosynthetic characteristics have been described (27, 28). Normal lung fibroblasts, strain CCL-202, were also obtained from the American Type Culture Collection (Rockville, MD). All cells were free of mycoplasma contamination and responded similarly in these studies.
14
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 3 1990
Recombinant Human Cytokines Recombinant human TNF (sp act, 5 x 107 U/mg protein) was a gift of Dr. H. Michael Shepard (Genentech, Inc., San Francisco, CA). Recombinant IL-liJ (sp act, 3.3 X 107 Ulmg protein) was obtained courtesy of Dr. Philip L. Simon (Smith Kline and French, Gulph Mills, PA). Recombinant IL-la (sp act, 4 X 107 Ulmg protein) was obtained courtesy of Dr. Peter Lomedico (Hoffmann-LaRoche, Nutley, NJ) and r interferon-y (lFN--y) (sp act, 1.4 x 108 IU/mg) was obtained courtesy of Dr. Peter Sorter (Hoffmann-LaRoche). Recombinant interleukin-6 (IL-6) (sp act, 1 to 2 X 106 half maximal Ulmg protein in a bone marrow colony-forming assay) was obtained courtesy of Dr. Steven Clark (Genetics Institute, Cambridge, MA). Neutralizing Antibodies Monospecific polyclonal antiserum against rIL-liJ and monospecific polyclonal antiserum against rll.-Io were kindly provided by Dr. Philip L. Simon (Smith Kline and French). These antisera were specific for the moieties against which they were raised and did not cross-react with IL-2 or IFN--y. Monospecific polyclonal antiserum against rIL-6 was kindly supplied by Drs. Pravinkumar B. Sehgal and Lester T. May (Rockefeller University, New York, NY). This antiserum was specific for IL-6 and did not cross-react with IL-la, ILliJ, or IFN--y. Monoclonal anti-rTNF was kindly supplied by Dr. H. Michael Shepard (Genentech). Assessment of Cell-associated TSA Cell-associated TSA was quantitated by determining whether paraformaldehyde-fixed fibroblasts or fibroblast lysates stimulated thymocyte proliferation. Paraformaldehyde-fixed fibroblasts. As described by Le
and associates (12), fibroblasts were grown to confluence in flat-bottom microtiter wells (Falcon; Becton-Dickinson, Oxnard, CA) in complete medium (DMEM supplemented with nonessential amino acids [GIBCO, Grand Island, NY] and penicillin and streptomycin [GIBCO]) supplemented with 10% heat-inactivated FBS (GIBCO). Once confluent, the medium was gently aspirated and replaced with 0.2 ml of serum-free complete medium with or without added cytokine(s) and the cells were incubated in 5% CO 2 and air for varying periods of time. At the end of the incubation period, the supernatants were removed and the cells washed and fixed by incubating for 25 min at 3r C with 0.5 % paraformaldehyde (Polysciences, Warrington, PA). The paraformaldehyde was then removed, the cells were washed, and TSA was assessed by performing a standard mouse thymocyte assay in these fibroblast-containing wells (see below). Fibroblast lysates. The TSA of fibroblast lysates was assessed using modifications of the procedures described by Lepe-Zuniga and Gery (29). Fibroblasts were grown to confluence in complete medium supplemented with 10% FBS in 60-mm plastic dishes and then incubated in serum-free complete medium in the presence and/or absence of cytokine(s) as noted above. They were then washed, mechanically detached, counted, frozen, thawed, and sonicated. Mouse thymocyte assay. TSA was assessed using the standard mouse thymocyte costimulator assay as previously described (30). Thymocytes from 3- to 6-wk-old C3H/HEJ mice were cultured in flat-bottom microtiter wells (1.5 x 106/well) in the presence of paraformaldehyde-fixed fibroblasts, serially diluted fibroblast lysates, control solutions, or rll.-Io, A suboptimal concentration of phytohemagglutinin (25 Ilg/ml) (Difco, Detroit, MI) was added to all wells, and the cultures were incubated for 72 h in 5 % CO 2 and air. Tritiated thymidine «(3H]Tdr) was added during the final
1200
1000 :::J
0 a:
800
0
600
Figure 1. Time course of induction of fibroblast TSA by rIL-liJ (5 ng/ml) (circles), rTNF (20 ng/ ml) (triangles), and the two cytokines in combination (rectangles). Values represent the mean ± SEM of six experiments comparing the TSA of cytokine-stimulated fibroblasts to fibroblasts incubated in media alone (control).
IZ
o
~ ~
«
Cf)
I-
400
200
0 0
20
40
HOURS
60
80
Elias and Reynolds: Synergistic Stimulation of IL-Ia by IL-I and TNF
16 h of the incubation period. At the end of the incubation, the thymocytes were harvested onto glass filter strips and (3H]Tdr incorporation was assessed by scintillation counting. The cell-associated TSA detected on/in paraformaldehyde-fixed cells was quantitated by comparison to a standard curve prepared with rIL-la. The TSA in Iysates was quantitated using dilutional probit analysis as previously described (30, 31).
Results Regulation of Fibroblast TSA by Individual Cytokines Thymocytes cocultured with fibroblasts that had been incubated under control conditions prior to fixation did not incorporate more (3H]Tdr than thymocytes incubated under control conditions in the absence of fibroblasts (data not shown). In contrast, fibroblasts incubated with rIL-I (a or (3) or rTNF before paraformaldehyde fixation expressed significant TSA. Induction of this TSA was seen after as little as 4 h, peaked at 24 h, and returned toward normal after 72 h of fibroblastcytokine incubation (Figure I). This return toward normal did not indicate that the fibroblasts were refractory to further stimulation since fibroblasts that had been preincubated for 72 h with rIL-I (a or (3) or rTNF were still able to express TSA if restimulated with these cytokines (Figure 2). In all cases, the effects of these cytokines were dose dependent, with rIL-I (a or (3) being more potent than rTNF (Figure 3). Maximal thymocyte proliferation occurred when fibroblasts
15
were incubated for 24 h with 2.5 ng/ml rIL-Ia, 5 ng/ml rIL1{3, or 20 ng/ml rTNF (the highest doses tested) (Figure 3). These doses of rIL-la , rIL-I{3, and rTNF caused thymocytes to incorporate 4.9-, 4.7-, and 1.8-fold, respectively, more (3H]Tdr than thymocytes incubated with fixed control fibroblasts (P < 0.01 for all, Student's t test). The activity of 10' fibroblasts stimulated with these doses of rIL-I (a or (3) or rTNF before fixation was comparable to that caused by 4 to 7 Vlml and 0.5 to 1.5 U/rnl, respectively, of rIL-la in the standard mouse thymocyte assay (Table I). Similar results were noted with lysates from unstimulated, rIL-I-stimulated, and rTNF-stimulated fibroblasts (Table I). In contrast, rIFN-'Y (103 to 1 IV/ml) did not induce fibroblast TSA, demonstrating that the ability to induce this activity was not a property of all inflammatory cytokines (data not shown). Regulation of Fibroblast TSA by rIL-l and rTNF in Combination Experiments were performed comparing the effects of the cytokines individually and in combination, These studies demonstrated that fibroblasts stimulated simultaneously with rIL-I and rTNF expressed more TSAthan could be accounted for by the sum of the effects of the individual cytokines (Figure 4 and Tables 1 and 2). Isobolograms demonstrated that this interaction was synergistic in nature (data not shown) (32). It was also at least partially specific for IL-I and TNF since rIFN-'Y did not augment the ability of rIL-l or rTNF
80000
60000
40000
•
CO'ITR)l
1\1
IL-1
II
TNF
r:a
IL-1+TNF
20000
o MEDIUM
IL-1
lNF
IL-1+TNF
PREINCUBATION CONDITIONS Figure 2. Demonstration that cytokine-stimulated fibroblasts can be restimulated to produce TSA. Fibroblasts were preincubated for 72 h in complete medium only (MEDIUM), rIL-I{3 (5 ng/rnl) (IL-I), trNF (20 ng/rnl) (TNF), or rIL-I{3 plus trNF (IL-I + TNF) as noted. They were then washed and incubated for 24 h with complete medium alone (control) (solid bars), rIL-I{3 (5 ng/rnl) (darkly shaded bars), trNF (20 ng/rnl) (criss-crossed bars), or the two cytokines together (lightly shaded bars) as indicated in the insert. After an additional washing and fixation, TSA was assessed as described.
16
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 3 1990
600
500
400
::J' 0
a:
Figure 3. Comparison of the thymocyte-
fZ
stimulating activity of fibroblasts incubated for 24 h in media alone (control) and varying doses of rIL-la (circles), rIL111 (triangles), and ffNF (rectangles). Values represent the mean ± SEM of 17 experiments at each dose of cytokine.
300
0 0
~
~
< C/)
200
f-
i
100
0 .001
.01
.1
10
100
LOG CONCENTRATION (NGlML)
to induce fibroblast TSA (data not shown). The TSA of fibroblasts incubated with rIL-l plus rTNF was noted after as little as 4 h, peaked at 24 h, and returned toward baseline with longer periods of fibroblast-cytokine incubation (Figure 1). However, these fibroblasts were not refractory to repeat cytokine stimulation (Figure 2). The induction of fibroblast TSA caused by rIL-l and rTNF in combination was dose dependent for both cytokines. When incubated with 20 ng/ml rTNF, doses of rIL-l (ex or (3) as low as 0.005 ng/ml synergistically stimulated fibroblast TSA (Figure 4 and Table 2).
TABLE I
Recombinant /L-1 and rTNF stimulation of fibroblast thymocyte-stimulating activity IL-I,. Activity (equivalent Ulml)t Fibroblast Culture Conditions*
A
B
Control
0
0
rIL-la rIL-113 rTNF rIL-la rIL-113
+ +
rTNF rTNF
6.8 6.5
± 1.0 ± 0.51
5.5 7.1
0.75
± 0.2
1.7 ± 0.03
40.7 39.8
± 4.8
± 2.7
45.5 48.7
± 1.3
± 0.9 ± 3.1
± 3.3
* Fibroblasts were incubated for 24 h in complete medium alone (control) or in the presenceof rll.-!« (2.5 ng/ml; 100 U/ml), rIL-113 (5 ng/ml; 165 U/ml), and/or rTNF (20 ng/ml; 103 U/ml), as noted. They were then washedand TSA was assessed after fixation (column A) or lysate preparation (column B) as described. t TSA of 105 fibroblasts compared to rIL-I,. in standard mouse thymocyte assay. Values represent the mean ± SEM of values with nine experiments using paraformaldehyde-fixed fibroblasts and five experiments with fibroblast Iysates.
Similarly, when incubated with 2.5 to 5 ng/rnl rIL-l, doses of rTNF as low as 0.02 ng/ml synergistically stimulated fibroblast TSA (Figure 4 and Table 2). Maximal thymocyte proliferation occurred when fibroblasts were incubated for 24 h with 20 ng/ml of rTNF and 2.5 ng/ml of rIL-lex or 5 ng/ml of rIL-l{3 (the highest doses tested). Fibroblasts incubated under these conditions caused cocultured thymocytes to incorporate 11- to 14-fold more pH]Tdr than thymocytes incubated with similarly fixed control fibroblasts (P < 0.001, Student's t test) (Figure 4 and Table 2). The activity of 105 fibroblasts stimulated in this fashion was comparable to that caused by 26 to 50 Vlml of rIL-lex (Table 1). Role of Recombinant Cytokine Adherence, Cytokine Contaminants, and Cell Number To be sure that the TSA of cytokine-stimulated fibroblasts was not due to rIL-l adherence to membrane IL-l receptors, modifications of the procedures of Bakouche and colleagues (33) were employed. In these experiments, fibroblasts were incubated with rIL-l (ex or (3) for 24 h, washed thoroughly, and then incubated in complete medium at pH 3.0 or in a 1 M salt solution for 2 h. The cells were then washed and their TSA evaluated after fixation or lysate preparation. Exposure to pH 3.0 or high salt did not cause a significant decrease in the TSA of cytokine-stimulated fibroblasts (data not shown). In addition, (1) the TSA expressed by rIL-l{3-stimulated fibroblasts was largely mediated by an IL-lex-like moiety (see below), (2) the TSA expressed by rIL-lex-stimulated fibroblasts was protein synthesis-dependent (see below), and (3) rIL-l (ex or (3) did not adhere significantly to unstimulated fibroblasts that had been fixed with paraformaldehyde (data not shown). Thus adherent rIL-l was not contributing significantly to the TSA of cytokine-stimulated fibroblasts.
Elias and Reynolds: Synergistic Stimulation of IL-la by IL-I and TNF
17
rTNF(ng/ml)
1200
20
1000 2 800
e...
-
Figure 4. Demonstration of the ability of rIL1a and rTNF, alone and in combination, to stimulate fibroblast TSA. Fibroblasts were incubated for 24 h in complete medium only (control) or with rlL-la and/or rTNF as noted. They were then washed and fixed and their TSA assessed as described. Values represent the mean ± SEM of eight experiments.
0.2
c:
a
u
0~ ~
600
«
0.02
C/)
~
400
0
200
I"
I 111110 I 111110 1111 1 1111111 I 1 1111" OL........L....1,I~WL.......L...L.LI.llW---l--LI...lll.lIIL..-..J.....I..J.JJ.l1JL---l....Ll.ll.WJ
o
.001
.01
.1
1
10
r1L-1 (ng/ml)
Three different approaches were taken to determine whether contaminants in the cytokine preparations were responsible for the stimulation of fibroblast TSA. First, we compared the ability of the cytokines to induce fibroblast TSA before and after preincubation with their respective neutralizing antibodies. These experiments demonstrated that specific neutralizing antibodies against rIL-la, rIL-l~,
TABLE 2
Synergistic stimulation offibroblast thymocyte-stimulating activity by rIL-l;3 plus rTNF rTNF (ng/mlJ* rIL-II3* (ng/ml)
0
0.02
0.2
2
20
0.0 0.005 0.05 0.5 5.0
3,27It 4,179 7,335 18,414 28,721
3,365 4,331 9,457 21,433 41,317
3,814 4,242 10,377 27,334 45,157
4,179 8,317 18,331 39,414 56,633
6,218 11,133 29,337 51,113 67,576
* Fibroblasts were incubated for 24 h with rIL- 113 and/or rTNF at the concentrations noted. They were then washed and fixed and their TSA assessed as described. t Thymocyte [3H]Tdr incorporation (cpm/well): values are means of triplicate determinations that were within 10% of each other.
and trNF abrogated the ability of their respective cytokines to stimulate fibroblast TSA (Table 3). Secondly, we determined whether polymyxin B altered the ability of recombinant cytokines to stimulate fibroblast TSA. At a dose of 10 Itg/ml polymyxin B abrogated the ability of lipopolysaccharide (LPS) to stimulatemonocyte IL-6 production but did not alter the ability of rIL-I and/or trNF to induce fibroblast TSA (data not shown), Lastly, we determined whether LPS, at concentrations well in excess of those in our cytokine preparations (5 to 0.005 Itg/ml), stimulated fibroblast TSA. At these concentrations, LPS did not induce significant cellassociated TSA (data not shown). When viewed together, these studies demonstrate that contaminants were not responsible for the stimulation of TSA caused by rIL-I (a or ~) and/or trNF. Identification of the Cytokine-induced Thymocyte Stimulator Specific neutralizing antibodies against rIL-Ia, rIL-I~, and rIL-6 were used to determine ifIL-la-, IL-l~-, or IL-6-like cytokines mediated the TSA of cytokine-stimulated fibroblasts. These studies demonstrated that> 90% of the cytokine-stimulated TSA detected on/in fixed fibroblasts and fibroblast lysates was neutralized by antiserum against rIL-
18
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 3 1990
TABLE 3
Recombinant cytokine stimulation offibroblast TSA: effect of preincubation with neutralizing antibodies Thymocyte [3H]Tdr Incorporation Fibroblast Culture Conditions*
Control Control Control Control rIL-1{3 rIL-1{3 rIL-1{3 rIL-1{3 rTNF rTNF rTNF rTNF
+ +
+ +
+ + + +
+ rIL-1{3 + rIL-1{3 rIL-1{3 rIL-111
+
+ +
anti-IL-1{3 anti-IL-1a anti-TNF
anti-IL-1{3 anti-IL-1a anti-TNF
+
+ +
2,707 3,404 3,316 3,114 34,474 2,941 33,437 35,531 8,741 8,499 8,981 2,437
anti-IL-1{3 anti-IL-1a anti-TNF rTNF rTNF rTNF rTNF
(cpm/well)t
anti-IL-1{3 anti-IL-1a anti-TNF
63,307 8,847 66,109 33,717
± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
212 317 161 144 3,191 221 228 1,519 737 319 643 201 6,144 801 5,531 3,714
* Fibroblasts were incubated for 24 h in complete medium only (control) or with rIL-IJ3 (5 ng/ml) and/or rTNF (20 ng/ml) as noted. They were then washed and fixed and their TSA assessed as described. The stimulating cytokines were used without modification or after preincubation (37 0 C, I h) with antiserum against rIL-Ie< (1:200 dilution), antiserum against rIL-IJ3 (1:200 dilution), or monoclonal anti-TNF (103 neutralizing units). t Mean ± SEM.
10( (Table 4). Anti-rIL-Ifj and anti-rIL-6 did not neutralize the TSA induced by these cytokines (Table 4). Thus, an ILl o--like moiety was the major mediator of the TSA of cytokine-stimulated fibroblasts. Role of Protein Production To determine whether fibroblast expression ofIL-IO( activity required protein production, we compared the TSA of fibroblasts incubated for 24 h with rIL-I and/or rTNF in the presence and absence of cycloheximide (50 j.tg/ml). Fibroblasts incubated with rTNF in the presence of cycloheximide expressed approximately 50% as much TSA as fibroblasts incubated with frNF in the absence of cycloheximide (Table 5). Similarly, fibroblasts incubated with rIL-I alone or rIL-I and rTNF in combination in the presence of cycloheximide expressed approximately 20 % as much TSAas fibroblasts incubated with these cytokines in the absence of cycloheximide (Table 5). These effects were not due to residual cycloheximide-inhibiting thymocyte proliferation since fibroblasts that had been incubated with cycloheximide prior to washing and fixation did not inhibit rlf.-Ice-induced thymocyte proliferation (Table 5). Thus, maximal induction of fibroblast IL-IO( activity by rIL-I and/or rTNF required protein synthesis.
Discussion To further understand the cytokine networks regulating local inflammatory and fibrotic events, we determined whether rIL-I and rTNF stimulated fibroblasts to express cellassociated TSA. These studies demonstrated that rIL-l and rTNF individually induce fibroblast TSA, with rIL-l being more potent. Importantly, they also demonstrated that rIL-l
and rTNF interact in a synergistic fashion to further upregulate fibroblast TSA. In all cases, this TSA was shown to be mediated by an Ils-l ce-Iike cytokine whose maximal expression required protein synthesis. Lastly, we demonstrated that, unlike LPS-stimulated monocytes (1,2, 30), fibroblasts that have been incubated with rIL-l and/or rTNF are not refractory to repeat stimuation with these cytokines. Together these findings suggest that cytokines released at sites of inflammation can repetitively stimulate local fibroblasts to produce an Il-Io-Iike moiety that could have important local regulatory effects. IL-I and TNF are products of different genes that bind to different membrane receptors (1, 2). Despite this, they are often elaborated simultaneously (34-36) and have similar spectra of biologic activity (1, 2). Their overlapping functional profiles are due, in part, to the ability of one cytokine to induce target cells to produce the other cytokine, as demonstrated in this report and others (12, 24). In addition,
TABLE 4
Identification of cytokine-induced thymocyte stimulator Neutralizing Abt Experiment
2
3
Thymocyte Culture Conditions*
None
AntiIL-Ie
We determined whether normal human lung fibroblasts expressed cell-associated thymocyte-stimulating activity in response to recombinant interleukin-l (rIL-l) (a and (3) and recombinant tumor necrosis factor (rTNF). Individually, rIL-l and rTNF induced fibroblast expression of thymocyte-stimulating activity, with rIL-l being significantly more potent. Importantly, combining rIL-l and rTNF resulted in a synergistic increase in fibroblast thymocyte-stimulating activity. This synergistic interaction was dose dependent for both cytokines and was not noted when '"(-interferon was combined with rIL-l or rTNF. In all cases, the thymocyte-stimulating activity was the result of an IL-la-like moiety whose maximal production required protein synthesis. IL-la activity could be detected after as little as 4 h, peaked after 24 h, and returned toward normal with longer periods of cytokine-fibroblast incubation. However, cytokinestimulated fibroblasts that no longer expressed IL-la activity could be induced to re-express this activity with repeat cytokine challenge. Induction of fibroblast IL-la by IL-l and/or TNF may be an important mechanism amplifying IL-l-mediated biologic events at sites of local inflammation.
The interleukin-l (IL-l) family of proteins are important regulators of inflammation, fibrosis, coagulation, and hemostasis (1, 2). The most studied member of this family is IL-l (3 , which was initially described as a soluble product of stimulated mononuclear phagocytes (3). It is now known that IL1(3 can be produced by a wide variety of nucleated cells (4-7). It is also known that the biologic effects of IL-I are mediated by complex networks of interacting cytokines in which cytokines regulate target tissues directly, that cytokines in combination have quantitatively and/or qualitatively different effects than they have individually (8-10), and that cytokines induce target tissues to produce bioactive cytokines (4, 5, 11, 12). Particularly important in this regard are the demonstrations from this (8, 13, 14) and other laboratories (15-17) that IL-l and tumor necrosis factor (TNF) can synergize in regulating target cell function and that IL-l can induce fibroblasts (14), endothelial cells (5), monocytes (18), and smooth muscle cells (4) to produce IL-l(3-like moieties. IL-la, the other major member of the IL-l family, is produced by appropriately stimulated mononuclear phagocytes (19, 20), B lymphocytes (21), dendritic cells (22), keratino-
Key Words: interleukin-1, tumor necrosis factor, fibroblast (Received in original form October 3, 1989 and in revised form February 7, 1990) Address correspondence to: Jack A. Elias, M.D., Pulmonary and Critical Care Medicine, Department of Internal Medicine, Yale University School of Medicine, 105 LCI, P. O. Box 3333, 333 Cedar Street, New Haven, CT 06510-8757. Abbreviations: interleukin, IL; recombinant, r; tumor necrosis factor, TNF; thymocyte-stimulating activity, TSA. Am. J. Respir. Cell Mol. BioI. Vol. 3. pp. 13-20, 1990
cytes (23), and fibroblasts (24). It has a similar spectrum of biologic activity and shares a common receptor with IL-l(3 (1, 2). However, IL-la and IL-l(3 may serve different functions in the inflanunatory response since their production can be differentially regulated (25, 26) and they are found in different locations in cells and supernatants (19, 20). The role IL-la plays in the cytokine networks that regulate biologic homeostasis has not been as extensively studied as that ofIL-l(3. Specifically, it is not known whether IL-l induces nonmononuclear phagocytes to produce functional IL-Ia and whether TNF alters this inductive effect. To address these issues, we determined whether recombinant (r) IL-l and rTNF caused fibroblasts to express a cell-associated thymocyte-stimulating activity (TSA) and investigated the mediator(s) of this effect. These studies demonstrate that rIL-l (a and (3) and rTNF individually stimulate fibroblasts to express a cell-associated TSA and that combining rIL-l and rTNF results in a synergistic increase in the expression of this activity. They also demonstrate that this cell-associated TSA is largely mediated by an It-Io-Iike cytokine.
Materials and Methods Cells Three adult human lung fibroblast strains were used in these studies. Two were established from histologically normal areas of lungs removed at surgery for standard clinical reasons. The techniques used to prepare these strains and their proliferative and biosynthetic characteristics have been described (27, 28). Normal lung fibroblasts, strain CCL-202, were also obtained from the American Type Culture Collection (Rockville, MD). All cells were free of mycoplasma contamination and responded similarly in these studies.
14
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 3 1990
Recombinant Human Cytokines Recombinant human TNF (sp act, 5 x 107 U/mg protein) was a gift of Dr. H. Michael Shepard (Genentech, Inc., San Francisco, CA). Recombinant IL-liJ (sp act, 3.3 X 107 Ulmg protein) was obtained courtesy of Dr. Philip L. Simon (Smith Kline and French, Gulph Mills, PA). Recombinant IL-la (sp act, 4 X 107 Ulmg protein) was obtained courtesy of Dr. Peter Lomedico (Hoffmann-LaRoche, Nutley, NJ) and r interferon-y (lFN--y) (sp act, 1.4 x 108 IU/mg) was obtained courtesy of Dr. Peter Sorter (Hoffmann-LaRoche). Recombinant interleukin-6 (IL-6) (sp act, 1 to 2 X 106 half maximal Ulmg protein in a bone marrow colony-forming assay) was obtained courtesy of Dr. Steven Clark (Genetics Institute, Cambridge, MA). Neutralizing Antibodies Monospecific polyclonal antiserum against rIL-liJ and monospecific polyclonal antiserum against rll.-Io were kindly provided by Dr. Philip L. Simon (Smith Kline and French). These antisera were specific for the moieties against which they were raised and did not cross-react with IL-2 or IFN--y. Monospecific polyclonal antiserum against rIL-6 was kindly supplied by Drs. Pravinkumar B. Sehgal and Lester T. May (Rockefeller University, New York, NY). This antiserum was specific for IL-6 and did not cross-react with IL-la, ILliJ, or IFN--y. Monoclonal anti-rTNF was kindly supplied by Dr. H. Michael Shepard (Genentech). Assessment of Cell-associated TSA Cell-associated TSA was quantitated by determining whether paraformaldehyde-fixed fibroblasts or fibroblast lysates stimulated thymocyte proliferation. Paraformaldehyde-fixed fibroblasts. As described by Le
and associates (12), fibroblasts were grown to confluence in flat-bottom microtiter wells (Falcon; Becton-Dickinson, Oxnard, CA) in complete medium (DMEM supplemented with nonessential amino acids [GIBCO, Grand Island, NY] and penicillin and streptomycin [GIBCO]) supplemented with 10% heat-inactivated FBS (GIBCO). Once confluent, the medium was gently aspirated and replaced with 0.2 ml of serum-free complete medium with or without added cytokine(s) and the cells were incubated in 5% CO 2 and air for varying periods of time. At the end of the incubation period, the supernatants were removed and the cells washed and fixed by incubating for 25 min at 3r C with 0.5 % paraformaldehyde (Polysciences, Warrington, PA). The paraformaldehyde was then removed, the cells were washed, and TSA was assessed by performing a standard mouse thymocyte assay in these fibroblast-containing wells (see below). Fibroblast lysates. The TSA of fibroblast lysates was assessed using modifications of the procedures described by Lepe-Zuniga and Gery (29). Fibroblasts were grown to confluence in complete medium supplemented with 10% FBS in 60-mm plastic dishes and then incubated in serum-free complete medium in the presence and/or absence of cytokine(s) as noted above. They were then washed, mechanically detached, counted, frozen, thawed, and sonicated. Mouse thymocyte assay. TSA was assessed using the standard mouse thymocyte costimulator assay as previously described (30). Thymocytes from 3- to 6-wk-old C3H/HEJ mice were cultured in flat-bottom microtiter wells (1.5 x 106/well) in the presence of paraformaldehyde-fixed fibroblasts, serially diluted fibroblast lysates, control solutions, or rll.-Io, A suboptimal concentration of phytohemagglutinin (25 Ilg/ml) (Difco, Detroit, MI) was added to all wells, and the cultures were incubated for 72 h in 5 % CO 2 and air. Tritiated thymidine «(3H]Tdr) was added during the final
1200
1000 :::J
0 a:
800
0
600
Figure 1. Time course of induction of fibroblast TSA by rIL-liJ (5 ng/ml) (circles), rTNF (20 ng/ ml) (triangles), and the two cytokines in combination (rectangles). Values represent the mean ± SEM of six experiments comparing the TSA of cytokine-stimulated fibroblasts to fibroblasts incubated in media alone (control).
IZ
o
~ ~
«
Cf)
I-
400
200
0 0
20
40
HOURS
60
80
Elias and Reynolds: Synergistic Stimulation of IL-Ia by IL-I and TNF
16 h of the incubation period. At the end of the incubation, the thymocytes were harvested onto glass filter strips and (3H]Tdr incorporation was assessed by scintillation counting. The cell-associated TSA detected on/in paraformaldehyde-fixed cells was quantitated by comparison to a standard curve prepared with rIL-la. The TSA in Iysates was quantitated using dilutional probit analysis as previously described (30, 31).
Results Regulation of Fibroblast TSA by Individual Cytokines Thymocytes cocultured with fibroblasts that had been incubated under control conditions prior to fixation did not incorporate more (3H]Tdr than thymocytes incubated under control conditions in the absence of fibroblasts (data not shown). In contrast, fibroblasts incubated with rIL-I (a or (3) or rTNF before paraformaldehyde fixation expressed significant TSA. Induction of this TSA was seen after as little as 4 h, peaked at 24 h, and returned toward normal after 72 h of fibroblastcytokine incubation (Figure I). This return toward normal did not indicate that the fibroblasts were refractory to further stimulation since fibroblasts that had been preincubated for 72 h with rIL-I (a or (3) or rTNF were still able to express TSA if restimulated with these cytokines (Figure 2). In all cases, the effects of these cytokines were dose dependent, with rIL-I (a or (3) being more potent than rTNF (Figure 3). Maximal thymocyte proliferation occurred when fibroblasts
15
were incubated for 24 h with 2.5 ng/ml rIL-Ia, 5 ng/ml rIL1{3, or 20 ng/ml rTNF (the highest doses tested) (Figure 3). These doses of rIL-la , rIL-I{3, and rTNF caused thymocytes to incorporate 4.9-, 4.7-, and 1.8-fold, respectively, more (3H]Tdr than thymocytes incubated with fixed control fibroblasts (P < 0.01 for all, Student's t test). The activity of 10' fibroblasts stimulated with these doses of rIL-I (a or (3) or rTNF before fixation was comparable to that caused by 4 to 7 Vlml and 0.5 to 1.5 U/rnl, respectively, of rIL-la in the standard mouse thymocyte assay (Table I). Similar results were noted with lysates from unstimulated, rIL-I-stimulated, and rTNF-stimulated fibroblasts (Table I). In contrast, rIFN-'Y (103 to 1 IV/ml) did not induce fibroblast TSA, demonstrating that the ability to induce this activity was not a property of all inflammatory cytokines (data not shown). Regulation of Fibroblast TSA by rIL-l and rTNF in Combination Experiments were performed comparing the effects of the cytokines individually and in combination, These studies demonstrated that fibroblasts stimulated simultaneously with rIL-I and rTNF expressed more TSAthan could be accounted for by the sum of the effects of the individual cytokines (Figure 4 and Tables 1 and 2). Isobolograms demonstrated that this interaction was synergistic in nature (data not shown) (32). It was also at least partially specific for IL-I and TNF since rIFN-'Y did not augment the ability of rIL-l or rTNF
80000
60000
40000
•
CO'ITR)l
1\1
IL-1
II
TNF
r:a
IL-1+TNF
20000
o MEDIUM
IL-1
lNF
IL-1+TNF
PREINCUBATION CONDITIONS Figure 2. Demonstration that cytokine-stimulated fibroblasts can be restimulated to produce TSA. Fibroblasts were preincubated for 72 h in complete medium only (MEDIUM), rIL-I{3 (5 ng/rnl) (IL-I), trNF (20 ng/rnl) (TNF), or rIL-I{3 plus trNF (IL-I + TNF) as noted. They were then washed and incubated for 24 h with complete medium alone (control) (solid bars), rIL-I{3 (5 ng/rnl) (darkly shaded bars), trNF (20 ng/rnl) (criss-crossed bars), or the two cytokines together (lightly shaded bars) as indicated in the insert. After an additional washing and fixation, TSA was assessed as described.
16
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 3 1990
600
500
400
::J' 0
a:
Figure 3. Comparison of the thymocyte-
fZ
stimulating activity of fibroblasts incubated for 24 h in media alone (control) and varying doses of rIL-la (circles), rIL111 (triangles), and ffNF (rectangles). Values represent the mean ± SEM of 17 experiments at each dose of cytokine.
300
0 0
~
~
< C/)
200
f-
i
100
0 .001
.01
.1
10
100
LOG CONCENTRATION (NGlML)
to induce fibroblast TSA (data not shown). The TSA of fibroblasts incubated with rIL-l plus rTNF was noted after as little as 4 h, peaked at 24 h, and returned toward baseline with longer periods of fibroblast-cytokine incubation (Figure 1). However, these fibroblasts were not refractory to repeat cytokine stimulation (Figure 2). The induction of fibroblast TSA caused by rIL-l and rTNF in combination was dose dependent for both cytokines. When incubated with 20 ng/ml rTNF, doses of rIL-l (ex or (3) as low as 0.005 ng/ml synergistically stimulated fibroblast TSA (Figure 4 and Table 2).
TABLE I
Recombinant /L-1 and rTNF stimulation of fibroblast thymocyte-stimulating activity IL-I,. Activity (equivalent Ulml)t Fibroblast Culture Conditions*
A
B
Control
0
0
rIL-la rIL-113 rTNF rIL-la rIL-113
+ +
rTNF rTNF
6.8 6.5
± 1.0 ± 0.51
5.5 7.1
0.75
± 0.2
1.7 ± 0.03
40.7 39.8
± 4.8
± 2.7
45.5 48.7
± 1.3
± 0.9 ± 3.1
± 3.3
* Fibroblasts were incubated for 24 h in complete medium alone (control) or in the presenceof rll.-!« (2.5 ng/ml; 100 U/ml), rIL-113 (5 ng/ml; 165 U/ml), and/or rTNF (20 ng/ml; 103 U/ml), as noted. They were then washedand TSA was assessed after fixation (column A) or lysate preparation (column B) as described. t TSA of 105 fibroblasts compared to rIL-I,. in standard mouse thymocyte assay. Values represent the mean ± SEM of values with nine experiments using paraformaldehyde-fixed fibroblasts and five experiments with fibroblast Iysates.
Similarly, when incubated with 2.5 to 5 ng/rnl rIL-l, doses of rTNF as low as 0.02 ng/ml synergistically stimulated fibroblast TSA (Figure 4 and Table 2). Maximal thymocyte proliferation occurred when fibroblasts were incubated for 24 h with 20 ng/ml of rTNF and 2.5 ng/ml of rIL-lex or 5 ng/ml of rIL-l{3 (the highest doses tested). Fibroblasts incubated under these conditions caused cocultured thymocytes to incorporate 11- to 14-fold more pH]Tdr than thymocytes incubated with similarly fixed control fibroblasts (P < 0.001, Student's t test) (Figure 4 and Table 2). The activity of 105 fibroblasts stimulated in this fashion was comparable to that caused by 26 to 50 Vlml of rIL-lex (Table 1). Role of Recombinant Cytokine Adherence, Cytokine Contaminants, and Cell Number To be sure that the TSA of cytokine-stimulated fibroblasts was not due to rIL-l adherence to membrane IL-l receptors, modifications of the procedures of Bakouche and colleagues (33) were employed. In these experiments, fibroblasts were incubated with rIL-l (ex or (3) for 24 h, washed thoroughly, and then incubated in complete medium at pH 3.0 or in a 1 M salt solution for 2 h. The cells were then washed and their TSA evaluated after fixation or lysate preparation. Exposure to pH 3.0 or high salt did not cause a significant decrease in the TSA of cytokine-stimulated fibroblasts (data not shown). In addition, (1) the TSA expressed by rIL-l{3-stimulated fibroblasts was largely mediated by an IL-lex-like moiety (see below), (2) the TSA expressed by rIL-lex-stimulated fibroblasts was protein synthesis-dependent (see below), and (3) rIL-l (ex or (3) did not adhere significantly to unstimulated fibroblasts that had been fixed with paraformaldehyde (data not shown). Thus adherent rIL-l was not contributing significantly to the TSA of cytokine-stimulated fibroblasts.
Elias and Reynolds: Synergistic Stimulation of IL-la by IL-I and TNF
17
rTNF(ng/ml)
1200
20
1000 2 800
e...
-
Figure 4. Demonstration of the ability of rIL1a and rTNF, alone and in combination, to stimulate fibroblast TSA. Fibroblasts were incubated for 24 h in complete medium only (control) or with rlL-la and/or rTNF as noted. They were then washed and fixed and their TSA assessed as described. Values represent the mean ± SEM of eight experiments.
0.2
c:
a
u
0~ ~
600
«
0.02
C/)
~
400
0
200
I"
I 111110 I 111110 1111 1 1111111 I 1 1111" OL........L....1,I~WL.......L...L.LI.llW---l--LI...lll.lIIL..-..J.....I..J.JJ.l1JL---l....Ll.ll.WJ
o
.001
.01
.1
1
10
r1L-1 (ng/ml)
Three different approaches were taken to determine whether contaminants in the cytokine preparations were responsible for the stimulation of fibroblast TSA. First, we compared the ability of the cytokines to induce fibroblast TSA before and after preincubation with their respective neutralizing antibodies. These experiments demonstrated that specific neutralizing antibodies against rIL-la, rIL-l~,
TABLE 2
Synergistic stimulation offibroblast thymocyte-stimulating activity by rIL-l;3 plus rTNF rTNF (ng/mlJ* rIL-II3* (ng/ml)
0
0.02
0.2
2
20
0.0 0.005 0.05 0.5 5.0
3,27It 4,179 7,335 18,414 28,721
3,365 4,331 9,457 21,433 41,317
3,814 4,242 10,377 27,334 45,157
4,179 8,317 18,331 39,414 56,633
6,218 11,133 29,337 51,113 67,576
* Fibroblasts were incubated for 24 h with rIL- 113 and/or rTNF at the concentrations noted. They were then washed and fixed and their TSA assessed as described. t Thymocyte [3H]Tdr incorporation (cpm/well): values are means of triplicate determinations that were within 10% of each other.
and trNF abrogated the ability of their respective cytokines to stimulate fibroblast TSA (Table 3). Secondly, we determined whether polymyxin B altered the ability of recombinant cytokines to stimulate fibroblast TSA. At a dose of 10 Itg/ml polymyxin B abrogated the ability of lipopolysaccharide (LPS) to stimulatemonocyte IL-6 production but did not alter the ability of rIL-I and/or trNF to induce fibroblast TSA (data not shown), Lastly, we determined whether LPS, at concentrations well in excess of those in our cytokine preparations (5 to 0.005 Itg/ml), stimulated fibroblast TSA. At these concentrations, LPS did not induce significant cellassociated TSA (data not shown). When viewed together, these studies demonstrate that contaminants were not responsible for the stimulation of TSA caused by rIL-I (a or ~) and/or trNF. Identification of the Cytokine-induced Thymocyte Stimulator Specific neutralizing antibodies against rIL-Ia, rIL-I~, and rIL-6 were used to determine ifIL-la-, IL-l~-, or IL-6-like cytokines mediated the TSA of cytokine-stimulated fibroblasts. These studies demonstrated that> 90% of the cytokine-stimulated TSA detected on/in fixed fibroblasts and fibroblast lysates was neutralized by antiserum against rIL-
18
AMERICAN JOURNAL OF RESPIRATORY CELL AND MOLECULAR BIOLOGY VOL. 3 1990
TABLE 3
Recombinant cytokine stimulation offibroblast TSA: effect of preincubation with neutralizing antibodies Thymocyte [3H]Tdr Incorporation Fibroblast Culture Conditions*
Control Control Control Control rIL-1{3 rIL-1{3 rIL-1{3 rIL-1{3 rTNF rTNF rTNF rTNF
+ +
+ +
+ + + +
+ rIL-1{3 + rIL-1{3 rIL-1{3 rIL-111
+
+ +
anti-IL-1{3 anti-IL-1a anti-TNF
anti-IL-1{3 anti-IL-1a anti-TNF
+
+ +
2,707 3,404 3,316 3,114 34,474 2,941 33,437 35,531 8,741 8,499 8,981 2,437
anti-IL-1{3 anti-IL-1a anti-TNF rTNF rTNF rTNF rTNF
(cpm/well)t
anti-IL-1{3 anti-IL-1a anti-TNF
63,307 8,847 66,109 33,717
± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±
212 317 161 144 3,191 221 228 1,519 737 319 643 201 6,144 801 5,531 3,714
* Fibroblasts were incubated for 24 h in complete medium only (control) or with rIL-IJ3 (5 ng/ml) and/or rTNF (20 ng/ml) as noted. They were then washed and fixed and their TSA assessed as described. The stimulating cytokines were used without modification or after preincubation (37 0 C, I h) with antiserum against rIL-Ie< (1:200 dilution), antiserum against rIL-IJ3 (1:200 dilution), or monoclonal anti-TNF (103 neutralizing units). t Mean ± SEM.
10( (Table 4). Anti-rIL-Ifj and anti-rIL-6 did not neutralize the TSA induced by these cytokines (Table 4). Thus, an ILl o--like moiety was the major mediator of the TSA of cytokine-stimulated fibroblasts. Role of Protein Production To determine whether fibroblast expression ofIL-IO( activity required protein production, we compared the TSA of fibroblasts incubated for 24 h with rIL-I and/or rTNF in the presence and absence of cycloheximide (50 j.tg/ml). Fibroblasts incubated with rTNF in the presence of cycloheximide expressed approximately 50% as much TSA as fibroblasts incubated with frNF in the absence of cycloheximide (Table 5). Similarly, fibroblasts incubated with rIL-I alone or rIL-I and rTNF in combination in the presence of cycloheximide expressed approximately 20 % as much TSAas fibroblasts incubated with these cytokines in the absence of cycloheximide (Table 5). These effects were not due to residual cycloheximide-inhibiting thymocyte proliferation since fibroblasts that had been incubated with cycloheximide prior to washing and fixation did not inhibit rlf.-Ice-induced thymocyte proliferation (Table 5). Thus, maximal induction of fibroblast IL-IO( activity by rIL-I and/or rTNF required protein synthesis.
Discussion To further understand the cytokine networks regulating local inflammatory and fibrotic events, we determined whether rIL-I and rTNF stimulated fibroblasts to express cellassociated TSA. These studies demonstrated that rIL-l and rTNF individually induce fibroblast TSA, with rIL-l being more potent. Importantly, they also demonstrated that rIL-l
and rTNF interact in a synergistic fashion to further upregulate fibroblast TSA. In all cases, this TSA was shown to be mediated by an Ils-l ce-Iike cytokine whose maximal expression required protein synthesis. Lastly, we demonstrated that, unlike LPS-stimulated monocytes (1,2, 30), fibroblasts that have been incubated with rIL-l and/or rTNF are not refractory to repeat stimuation with these cytokines. Together these findings suggest that cytokines released at sites of inflammation can repetitively stimulate local fibroblasts to produce an Il-Io-Iike moiety that could have important local regulatory effects. IL-I and TNF are products of different genes that bind to different membrane receptors (1, 2). Despite this, they are often elaborated simultaneously (34-36) and have similar spectra of biologic activity (1, 2). Their overlapping functional profiles are due, in part, to the ability of one cytokine to induce target cells to produce the other cytokine, as demonstrated in this report and others (12, 24). In addition,
TABLE 4
Identification of cytokine-induced thymocyte stimulator Neutralizing Abt Experiment
2
3
Thymocyte Culture Conditions*
None
AntiIL-Ie
