Transforming growth factor-~2 prevents preterm delivery induced by interleukin-la and tumor necrosis factor-a in the rabbit

Kristina Bry, MD, and Mikko Hallman, MD

Irvine, California

OBJECTIVES: The purpose of the study was to determine whether preterm parturition in the rabbit can be induced by intraamniotic injection of proinflammatory 'cytokines, interleukin-1 a and tumor necrosis factor-a, and whether transforming growth factor-(32, an inhibitor of the cytokine-induced prostaglandin synthesis, modifies the effect of these cytokines. STUDY DESIGN: New Zealand White rabbits were injected in each amniotic cavity on day 24 of gestation with one of the following: a combination of interleukin-1 a (150 ng) and tumor necrosis factor-a (1.25 J.Lg), 50 ng of transforming growth factor-(32 concomitantly with interleukin-1a and tumor necrosis factor-a, or vehicle. In the first study the animals were observed for signs of delivery until day 29 of gestation. In the second study the effect of transforming growth factor-(32 (50 ng/fetus) on the rate of premature delivery was evaluated. In the third study the concentrations of prostaglandin E2 and 13,14-dihydro-15-keto-prostaglandin F2a were measured in the amniotic fluid on day 27 of gestation. The statistics used were Fisher's exact test, the x2 test, and the Mann-Whitney U test. RESULTS: Altogether 61 of 191 fetuses (32%) were born prematurely in the interleukin-1a-tumor necrosis factor-a group, whereas only two of 161 fetuses (1.2%) (p = 0.0001) and one of 159 (0.6%) (p = 0.0001) were born prematurely in the interleukin-1a-tumor necrosis factor-a-transforming growth factor-(32 group and in the control group, respectively. Of the 23 animals injected with interleukin-1a and tumor necrosis factor-a, six (26%) delivered all of their fetuses prematurely versus none in the other groups (p = 0.02). None of the 88 fetuses in the transforming growth factor-(32 group were born prematurely. The prostaglandin E2 concentrations in the amniotic fluids were higher in the interleukin-1 a-tumor necrosis factor-a group than in the interleukin-1 a-tumor necrosis factor-a-transforming growth factor-(32 group (p = 0.05) or in the control group (p = 0.02). CONCLUSIONS: Preterm parturition can be provoked in the rabbit by intraamniotic injections of interleukin-1a and tumor necrosis factor-a. Transforming growth factor-(32 prevents the cytokine-induced increase in premature delivery. (AM J OSSTET GYNECOL 1993;168:1318-22.)

Key words: Cytokine, preterm labor, rabbit model, interleukin-I, tumor necrosis factor, transforming growth factor-~

Increased concentrations of interleukin-I (IL-I) and tumor necrosis factor (TNF) are present in amniotic fluid in preterm labor in the presence of intrauterine infection. I

-' IL-I and TNF stimulate the production of prostaglandins by fetal membrane cells."' 4-7 Because prostaglandins play an important role in parturition,8 IL-I and TNF have been proposed as signals for the initiation of preterm labor. I

-'

Transforming growth factor-~ (TGF-~) is a group of cytokines participating in immunoregulation and extra­cellular matrix formation. 9 TGF -~ is present in the

From the Department of Pediatrics, University of California, Irvine. Supported by the Sigrid Juselius Foundation and the Foundation for Maternal and Infant Care. Received for publication June 27, 1992; revised October 13, 1992; accepted October 31, 1992. Reprint requests: Kristina Bry, MD, University of California, Irvine, Department of Pediatrics, Med. Sci. I, Rm C202, Irvine, CA 92717. Copyright © 1993 by Mosby-Year Book, Inc. 0002-9378193 $1.00 + .20 6/1/43964

1318

decidua and other tissues at the maternal-fetal inter­face. 10. 11 We have recently shown that TGF-~ inhibits the IL-I-and TNF-induced prostaglandin E2 (PGE2) synthesis by amnion cells in culture. I.

Systemic injection of IL-Ia induces preterm delivery in the mouse. 13 The purpose of the current study was to determine whether preterm delivery in the rabbit can be induced by injecting IL-Ia and TNF-a into the amniotic fluid and whether the effects of IL-I a and TNF-a can be prevented by TGF-~2.

Material and methods Experimental design. The protocols of the studies

were approved by the Animal Research Committee of the University of California, Irvine. The animals used in these studies were timed-pregnant New Zealand White rabbits (Irish Farms, Orange County, Calif.). The mat­ing day was defined as day 0 of gestation. The animals were housed in the university animal facility with a

Volume 168, Number 4 Am J Obstet Gynecol

Table I. Outcome by treatment group

Does Premature delivery* Fetuses in utero day 24 Fetuses in utero day 29 Dead fetuses in utero day 29

Control

19 o

159 158

2

IL-la-TNF-a

23 6t

191 130t

6§

Bry and Hallman 1319

IL-l a-TNF-a-TGF-{32

19 o

161 159

1211

*Expulsion of all fetuses from the uterus before day 29. In addition, three does in the IL-la-TNF-a group delivered five fetuses. Two does in the IL-la-TNF-a-TGF-j32 group and one doe in the control group delivered one fetus each.

tp = 0.024 compared with the control group and p = 0.024 compared with the IL-la-TNF-a-TGF-j32 group (Fisher's exact test). tp = 0.0001 compared with the control group and p = 0.0001 compared with the IL-la-TNF-a-TGF-j32 group (X 2 test). §p = 0.085 compared with the control group (X 2 test). lIP = 0.0065 compared with the control group and p = 0.44 compared with the IL-la-TNF-a group (X2 test).

12-hour-day-12-hour-night cycle. They were allowed free access to food and water.

Study 1. On day 24 of pregnancy the rabbits were randomized to one of three treatment groups. All the surgical procedures were performed by the same investi­gator, who was unaware of the treatment allocation. The rabbits were anesthetized with an intramuscular injec­tion ofketamine hydrochloride (40 mg/kg) and xylazine hydrochloride (3 mglkg). The abdomen was shaved be­fore surgery. The skin, subcutaneous tissue, and perito­neum were infiltrated with lidocaine (5 mglml) with epi­nephrine. A midline incision was made, and the uterine horns were exposed. One of the following three drug combinations was injected in 50 ILl of carrier to the am­niotic fluid of all fetuses in each litter: 150 ng of human recombinant IL-la (a gift from the National Cancer In­stitute, Bethesda, Md.) and 1.25 ILg of human recombi­nant TNF-a (a gift from Genentech, South San Fran­cisco) (IL-la-TNF-a group); 50 ng of porcine TGF-~2 (Research and Diagnostic Systems, Minneapolis) in addi­tion to IL-la and TNF-a (IL-la-TNF-a-TGF-~2

group); or carrier (Hanks' balanced salt solution contain­ing 0.4% rabbit serum and 0.4 mmoVL hydrochloric acid) (control group). The concentrations of the cyto­kines used were based on in vitro studies on the effects of cytokines on amnion cell PGE2 production7

. 12 and on a preliminary study on the effects of different concentra­tions of cytokines on pregnancy outcome in rabbits. Ut­most care was taken to inject the drugs into the translu­cent amniotic fluid and not into the fetus, whose shadow was visible through the uterine wall. The number of fe­tuses in each horn was recorded. After the injections, the peritoneum, the subcutis, and the skin were sutured. All surgical procedures were done under strictly aseptic con­ditions. On the day of operation and 4 days thereafter, the does were given procaine penicillin G (15,000 unitslkg) and dihydrostreptomycin sulfate (equivalent to 20 mg/kg of dihydrostreptomycin base) intramus­cularly.

The does were observed for signs of labor or delivery until day 29 of gestation, at which time they were killed.

The fetuses present in utero at that time were delivered by hysterotomy. They were weighed and allowed to breathe for 1 hour to check for their viability before they were killed.

Study 2. This study was performed to investigate the effects of TGF -~2 on the viability of the fetuses and on pregnancy outcome. A total of 11 does were studied. Fifty nanograms of TGF-~2 was injected into the am­niotic fluid of each fetus. The protocol was otherwise the same as for study I.

Study 3. This study was performed with the same protocol as in study I, but the does were killed at 27 days of gestation to analyze amniotic fluid samples for PGE2 (radioimmunoassay kit from NEN-Dupont, Boston) and 13,14-dihydro-15-keto-prostaglandin F2o:

(PGFM) (radioimmunoassay kit from Advanced Mag­netics Inc., Cambridge, Mass.). For PGFM assay the amniotic fluid samples were acidified with hydrochloric acid and extracted with ethyl acetate. For PGE2 the samples were not extracted. The does' sera on days 24 and 27 of gestation were analyzed for progesterone (radioimmunoassay kit from Advanced Magnetics).

Statistics. All results are expressed as means ± SD. The tests used are Fisher's exact test, the X2 test, and the Mann-Whitney U test, as appropriate. All signifi­cance levels reported are for two-sided tests.

Results

Study 1 Premature deliveries. Altogether 61 pregnant rabbits

were studied. The results are summarized in Table I. Six does of 23 (26.0%) in the IL-la-TNF-a group delivered all their 56 fetuses prematurely. Two of these deliveries occurred on day 28 and four on day 29 of pregnancy. In the two other groups no doe delivered all her fetuses. The difference in the number of deliveries between the IL-la-TNF-a group and the other two groups is significant (p = 0.024, Fisher's exact test). In addition to the deliveries where all the fetuses were expelled from the uterus, one doe in the IL-la-TNF-o: group delivered one fetus of her total of five fetuses,

1320 Bry and Hallman

5 88

...... ............. -....... _ ... E 4 D til

0 -A--c - 3 .. _._.§ .... __ . 0 N t::.t::.t::.

w D CJ 2

Q.

o~-----------------------------Control IL-1/TNF IL-1/TNF/TGF-B

Fig. 1. Amniotic fluid PGE2 in different treatment groups. Each point represents mean of amniotic fluid PGE2 concen­trations in one doe. PGE2 concentration in IL-Ia-TNF-a group is significantly higher than in IL-Ia-TNF-a-TGF-f32 group (p = 0.05, Mann-Whitney U test) and in control group (p = 0.02). IL-Ia-TNF-a-TGF-f32 group does not differ sig­nificantly from control group (p = 0.4).

another rabbit in this group delivered two of her seven fetuses, and a third rabbit in this group delivered two of her eight. One doe in the IL-la-TNF-a-TGF-~2 group delivered one of her four fetuses soon after the opera­tion, and another doe in this group delivered one fetus of her nine. One control doe delivered one of her ten fetuses. The differences in the numbers of fetuses de­livered prematurely between the IL-la-TNF-a group and the other two groups are highly significant (p = 0.0001, X2 test).

Fetal deaths in utero. There were 20 intrauterine deaths. Of these, six occurred in two litters in the IL-la-TNF-a group. There were 12 intrauterine deaths in four does in the IL-la-TNF-a-TGF-~2 group. Five of these were in one rabbit having seven fetuses and grossly purulent and foul-smelling amniotic fluid. In spite of the chorioamnionitis this doe did not deliver before day 29. One control doe had two dead fetuses. There were significantly more intrauterine deaths in the IL-la-TNF-a-TGF-~2-treated animals than in the controls (p = 0.0065, X2 test). A similar trend was observed in the IL-la-TNF-a-treated animals as well (p = 0.085), whereas the difference in the number of dead fetuses between the IL-l a-TNF -a and the IL-l a­TNF-a-TGF-~2 group was not significant (p = 0.44).

Fetal weights. All fetuses that were alive at birth sur­vived for the observation period of 1 hour. The mean weight of live fetuses on day 29 of gestation was 40.1 ± 5.3 gm in the IL-la-TNF-a group, 42.0 ± 5.9 gm in the IL-la-TNF-a-TGF-~2 group, and 43.5 ±

5.9 gm in the control group. These differences are not significant.

Study 2. None of the 11 rabbits in the TGF-~2 group delivered before day 29 of pregnancy. There were 88 fetuses in utero on day 24 and 88 on day 29. Of these,

20

...... E 15 til c - D

:2 10 D u.

<CJ ---~ a.. 5 D

D

0 Control

0

0

§ 0

IL-1/TNF

April 1993 Am J Obstet Gynecol

t::.

t::.t::. ---.~~--.

IL-1/TNF/TGF-B

Fig. 2. Amniotic fluid PGFM concentrations in different treat­ment groups. Each point represents mean of PGFM concen­trations in one doe. PGFM concentration in IL-Ia-TNF-a group differs from IL-Ia-TNF-a-TGF-f32 group with signifi­cance level of 0.05 (Mann-Whitney U test). IL-Ia-TNF-a group does not differ significantly from the control group (p = 0.4).

87 were alive at birth and survived the observation period of 1 hour. Thus the number of intrauterine deaths did not differ significantly from that of the control group in study 1. The mean weight of the surviving fetuses was 41.6 ± 6.3 gm.

Study 3. Of the 23 rabbits participating in the third study, eight were treated with IL-la and TNF-a, eight with IL-la, TNF-a, and TGF-~2, and seven with vehi­cle. One doe in the IL-la-TNF-a group delivered all her fetuses before day 27. No other fetuses in this study delivered prematurely.

Prostaglandin concentrations in the amniotic fluid on day 27 of pregnancy. Grossly bloody amniotic fluid was ex­cluded from the prostaglandin assays. The mean PGE2

concentrations in the does' amniotic fluid were 4.13 ± 0.86 ng/ml (n = 6) in the IL-la-TNF-a group, 3.13 ± 0.44 ng/ml (n = 8) in the IL-la-TNF-a-TGF-~2 group, and 2.82 ± 0.58 ng/ml (n = 6) in the control group. The PGE2 concentrations in the amniotic fluid of the does in the IL-la-TNF -a group were significantly higher than in the IL-la-TNF-a-TGF-~2 group (p = 0.05) and in the control group (p = 0.02). There was no difference in the PGE2 concentrations between the control group and the IL-la-TNF-a-TGF-~2 group (Fig. 1).

The PGFM concentrations in the amniotic fluid were 9.68 ± 0.55 ng/ml in the IL-la-TNF-a group (n = 6), 5.21 ± 0.23 ng/ml in the IL-la-TNF-a-TGF-~2 group (n = 8) (p = 0.05 relative to the IL-la-TNF-a group, Mann Whitney U test), and 7.31 ± 0.31 ng/ml in the control group (n = 6) (Fig. 2).

Progesterone concentrations in the does' sera. Because parturition in the rabbit is progesterone dependent,14 we measured the progesterone concentrations in the does' sera in the beginning and at the end of the study (Fig. 3). On day 24 of pregnancy the progesterone

Volume 168. Number 4 Am J Obstet Gynecol

-E C, c ~

CD c 0 .. CD iii CD CI 0 .. D-

Bry and Hallman 1321

Control IL-1/TNF IL-1/TNF/TGF-B 18

16 14 0 0

12

~ 10

8 6

4 2

o 24 27 24 27 24 27

Day of GestatIon

Fig. 3. Progesterone concentrations in does' sera at time of operation (day 24) and at death (day 27). eThere is no significant difference between three groups. One doe that was delivered is marked with black dot.

concentrations in the does' sera were 10.6 ± 3.2 ng/ml (n = 8) in the IL-1o:-TNF-o: group, 9.0 ± 2.8 ng/ml (n = 8) in the IL-1o:-TNF-o:-TGF-~2 group, and 9.6 ± 2.9 ng/ml (n = 7) in the control group. On day 27 of gestation the progesterone levels were 11.3 ± 4.3 ng/ml, 10.2 ± 3.0 ng/ml, and 10.0 ± 2.7 ng/ml in the IL-1o:-TNF-o:, the IL-1o:-TNF-o:-TGF-~2, and the con­trol animals, respectively. These differences are not significant. In the IL-1o:-TNF-o: group there were two rabbits whose progesterone levels declined consider­ably. One of these was the only rabbit in the series that was delivered before day 27 (Fig. 3).

Comment

According to the current study, IL-1o: and TNF-a injected to the amniotic fluid increase the risk of pre­term delivery in the rabbit. This effect was associated with an increase in the concentration of PGE2 in the amniotic fluid. TGF-~2, on the other hand, inhibited preterm delivery induced by IL-1o: and TNF-a. TGF-~2 also prevented the increase in amniotic fluid PGE2 concentration associated with IL-1o: and TNF-a.

In this study we used the combination of IL-1 and TNF because of their synergistic effects on the PGE2 synthesis by amnion cells in vitro.' To our knowleoge, this is the first study where cytokines have been used to induce pregnancy loss in rabbits. Intrauterine inocula­tion of bacteria induces preterm pregnancy loss in the rabbit. 15. 16 Preterm delivery has been induced in mice by subcutaneously given IL_1o:.I3 In these animals par­turition occurred within 24 hours of IL-1 administra­tion. In the current study, in contrast, the time interval between cytokine injections and delivery was in most cases 4 to 5 days. In addition, the rate of preterm delivery was lower than in mice, where systemic admin­istration ofIL-1a resulted in abortion in the majority of cases. I3 These differences may be caused by the consid­erably lower dose of IL-1o: per kilogram of body weight

in the current study, by a different route of administra­tion of the drugs, or by a difference in animal species. No behavioral changes were observed in the does, whereas lethargy was noted in the IL-1a-treated mice. I3

There was a tendency to an increased incidence of fetal deaths in the animals injected with cytokine com­binations compared with the controls. In some cases this may be due to inadvertent injection of cytokines directly to the fetuses. In a separate series of experi­ments we tested the effects of TGF-~2 on pregnancy outcome and on the viability of the fetuses. No cases of preterm delivery occurred in the TGF-~2-injected ani­mals, and the rate of intrauterine deaths among the fetuses did not differ from that of the controls.

In several in vitro systems TGF-~ has been shown to modify the action of other cytokines. TGF-~ attenuates the IL-1-or TNF-stimulated PGE2 production by am­nion cells in culture I2 and inhibits the TNF-a-induced expression ofIL-1 receptors on amnion cells. I7 TGF-~2

also inhibits the IL-1-induced release of PGE2 and of phospholipase A2 in rat mesangial cells. 18

IL-1 receptor antagonist has been recently reported to prevent the induction of preterm delivery in mice by IL_1o:.I9 The current results show that TGF-~2 is also potentially effective as an inhibitor of the IL-1o:-and TNF-o:-induced premature delivery.

REFERENCES I. Romero R. Brody DT, Oyarzun E, et al. Infection and

labor. III. Interleukin-I: a signal for the onset of parturi­tion. AM] OBSTET GYNECOI. 1989; 160: 1117-23.

2. Casey ML. Cox SM, Beutler B. Milevich L, MacDonald PC. Cachectin/tumor necrosis factor-a formation in human decidua. ] Clin Invest 1989;83:430-6.

3. Romero R. Manogue KR. Mitchell MD, et al. Infection and labor. IV. Cachectin-tumor necrosis factor in the amniotic fluid of women with intraamniotic infection and preterm labor. AM] OBSTET GYNECOL 1989;161:336-41.

4. Romero R, Durum S, Dinarello CA, Oyarzun E, Hobbins ]C, Mitchell MD. Interleukin-I stimulates prostaglandin

1322 8ry and Hallman

biosynthesis by human amnion. Prostaglandins 1989;37: 13-22.

5. Romero R, Mazor M, Wu YK, Avila C, Oyarzun E, Mitchell MD. Bacterial endotoxin and tumor necrosis factor stim­ulate prostaglandin production by human decidua. Pros­taglandins Leukot Essent Fatty Acids 1989;37:183-6.

6. Mitchell MD, Edwin S, Romero RJ. Prostaglandin biosyn­thesis by human decidual cells: effects of inflammatory mediators. Prostaglandins Leukot Essent Fatty Acids 1990; 41:35-8.

7. Bry K, Hallman M. Synergistic stimulation of amnion cell prostaglandin E2 synthesis by interleukin-l, tumor necro­sis factor, and products from activated human granulo­cytes. Prostaglandins Leukot Essent Fatty Acids 1991;44: 241-5.

8. Novy M], Liggins GC. Role of prostaglandins, prostacy­din, and thromboxanes in the physiologic control of the uterus and in parturition. Semin Perinatol 1980;4:45-66.

9. Massague J. The transforming growth factor-j3 family. Annu Rev Cell Bioi 1990;6:597-641.

10. Kauma S, Matt D, Strom S, Turner T. Interleukin-lj3, human leukocyte antigen HLA-DRa, and transforming growth factor-j3 expression in endometrium, placenta, and placental membranes. AM ] OSSTET GYNECOL 1990;163: 1430-7.

11. Graham CH, LysiakJJ, McCrae KR, Lala PK. Localization of transforming growth factor-beta at the human fetal­maternal interface - role in trophoblast growth and differ­entiation. Bioi Reprod 1992;46:561-72.

12. Bry K, Hallman M. Transforming growth factor-j3 opposes

April 1993 Am J Obstet Gynecol

the stimulatory effects of interleukin-l and tumor necrosis factor on amnion cell prostaglandin E2 production: impli­cation for preterm labor. AM] OSSTET GYNECOL 1992;167: 222-6.

13. Romero R, Mazor M, Tartakovsky B. Systemic administra­tion ofinterleukin-l induced preterm parturition in mice. AM] OSSTET GYNECOL 1991;165:969-71.

14. Challis ]RG, Davies ], Ryan KJ. The concentrations of progesterone. estrone and estradiol-17 in the plasma of pregnant rabbits. Endocrinology 1973;93:971-6.

15. Dombroski RA, Woodard DS, Harper M]K, Gibbs RS. A rabbit model for bacteria-induced preterm pregnancy loss. AM] OSSTET GYNECOL 1990; 163: 1938-43.

16. McDuffie RS ]r, Blanton S], Shikes RH, Gibbs RS. A rabbit model for bacterially induced preterm pregnancy loss: intervention studies with ampicillin-sulbactam. AM ] OSSTET GYNECOL 1991;165:1568-74.

17. Bry K, Hallman M. Modulation of interleukin-l (IL-l) receptor expression on amnion cells by tumor necrosis factor (TNF) and transforming growth factor-j3 (TGF-j3). Pediatr Res 1992;31:197A.

18. Schalkwijk C, pfeilschifter ], Marki F, Vandenbosch H. Interleukin-lj3-induced and forskolin-induced synthesis and secretion of group II phospholipase-~ and prosta­glandin-E2 in rat mesangial cells is prevented by trans­forming growth factor-j32.] Bioi Chern 1992;267:8846-51.

19. Romero R, Mazor M, Tartakovsky B. The natural inter­leukin-l receptor antagonist prevents interleukin-l-in­duced preterm parturition. AM ] OSSTET GYNECOL 1992; 166:358.