In vivo changes in plasma acute phase protein levels in the rat induced by slow release of IL-1, IL-6 and TNF

Administration of large doses of cytokines by injection is required to induce changes in acute phase protein levels. Comparisons were made in the rat of the effects of administering recombinant human cytokines by injection with continuous release from implanted osmotic minipumps. Continuous release of interleukin-1β (0.2–2.1 ng h-1) induced dose-related changes in the plasma levels of albumin, seromucoid proteins, haptoglobin and caeruloplasmin; interleukin-1α had similar effects but required higher doses (2–21 ng h-1). Tumour necrosis factor α (50 ng h-1) only significantly increased seromucoid levels, whereas IL-6 (3–30 ng h-1) induced haptoglobin and caeruloplassynthesis. This method provides a better technique for studying the in rive effects of cytokines which may be relevant to the release mechanisms in inflammation.


Introduction
Acute phase proteins (APPs) are defined as those proteins produced by the liver whose levels change during the course of inflammation or trauma. 1'2 APP are heterogeneous in nature with different physiological functions, protein synthesis and often electrophoretic mobility and cannot be classified under one simple chemical heading. APP include proteins such as C-reactive protein, fibrinogen, haptoglobin, caeruloplasmin, albumin etc. In man, during the course of severe inflammation, levels of CRP can increase up to 1 000-fold and fibrinogen twoto three-fold. However, albumin levels can decrease, thus specific APPs also have a variable degree of change. The change in levels of APPs has been shown to correlate to the degree of inflammation in patients with rheumatoid arthritis 3 and for this reason, APPs are routinely used as markers of inflammation and the effectiveness of drugs modifying arthritis. In the early 1970s it was shown that a plasma-borne mediator released from cells at the site of inflammation altered the rate of liver synthesis of the acute phase proteins. 4 This mediator was later purified, defined as a cytokine and named interleukin-1. Subsequently, a number of cytokines involved in inducing changes in liver synthesis have been identified, purified and cloned. The main cytokines recognised as inducing changes in liver synthesis of APPs are interleukin-1 (IL-1), interleukin-6 (IL-6) and tumour necrosis factor (TNF).
All of these cytokines have been shown to induce APP changes in cultured hepatocytes, hepatoma (C) 1992 Rapid Communications of Oxford Ltd. cells 6'7 and also induce changes in the plasma levels of animals after injection. 8'9 The concentrations of the cytokines required to induce changes in APPs in vivo have been much greater than those used in vitro because of the short half-life of the cytokines in blood. 1 '11 In inflammatory conditions, release of the cytokines is more likely to be continuous rather than spasmodic, therefore to model these effects in vivo a method of sustained release of the cytokines is required. The aim of this study was to investigate the effects of interleukin-lo on APP levels in the rat after single and multiple injection and to compare these effects with IL-I administered by slow release from implanted osmotic minipumps. Further studies were made using slow release of recombinant human IL-lfl, TNFo and IL-6 in the rat in order to determine the role of these cytokines in the induction of individual APPs.

Materials and Methods
Animals: In all the experiments described, female, Allen and Hanbury Hooded strain rats were used. The rats weighed between 90 and 160g and were aged between 8 and 12 weeks old. The animals were provided with tap water and Spillers PCD diet ad libitum throughout the experiment.
Collection ofplasma: Ether anaesthetized rats were bled from the dorsal tail artery using a heparin rinsed (Liquemine, Roche; 500 U/ml) 1 ml syringe fitted with a 23G x 30 mm needle. Blood was collected and transferred to a 1.5 ml eppendorf tube and centrifuged for 5 min at 1500 g in a MSE microcentrifuge. Plasma was either stored at 4C if analysed within 3 days or frozen at -20C to be analysed at a later date. Formulation of cytokines for administration: The recombinant human cytokines were kind gifts from Dr P Lomedico, Hoffman La Roche, Nutley, New York (rhlL-l and rhlL-lfl); Dr Hooculi, Hoffman La Roche, Basel (rhTNF) and Professor W Fiers, University of Ghent, Belgium (rhIL-6). In all experiments the concentrations of the cytokines required were achieved by diluting stock concentrations with a sterile saline solution containing 0.2% bovine serum albumin, which was used as a carrier protein for the cytokine. Stock solutions were kept frozen at --20C and thawed immediately before use. The specific activity of the recombinant human cytokines were :-rhlL-l 2.1 x 10 U/mg; rhlL-lfl 2.4 108U/mg; rhTNF--I.0 x 108 U/mg; and rhlL-6 3.1 108 U/mg as determined by the appropriate cellular assays. Administration of cytokines: Injections of rhIL-lcz were made using dilutions of the stock solutions in 0.2% bovine serum albumin and rats were injected with 2 ml kg of body weight -1. Alzet osmotic minipumps (Model 2001; 1/,1 h-1) were filled in a sterile air flow cabinet with the cytokine. Before the pumps were implanted the rats were anaesthetized and their backs shaved with animal clippers. A small incision (10-15 mm) was made through the skin and a pair of blunt ended scissors was inserted through the incision on one side and the tissue connecting the skin layer to the muscle body wall layer was gently teased apart to form a channel of about 3 cm wide to a depth of 3 cm. An osmotic minipump was implanted through the incision to the end of the channel on one side of the backbone. The incision was then closed using two or three Michel clips (6 mm) and the area swabbed with 1% Hibitane (R) in 70% ethanol. Assay of plasma APP" Plasma obtained from the rats was analysed using a Cobas-bio centrifugal analyser. The methods used to assay APPs have been previously described in detail 12 but, in brief, albumin levels were determined using the bromocresol green dye binding assay; 13 seromucoid by protein precipitation, 14 haptoglobin by the generation of peroxidase activity when bound to added methaemoglobinlS and caeruloplasmin by its oxidase activity using p-phenylenediamine. 16

Results
Effect of injected rhIL-la" Injections of rhIL-10 (6 ng g of animal body weight-1) were made into groups of five rats using a variety of routes of administration. The animals were bled 24 h later. Haptoglobin levels were significantly increased by injected rhlL-l and this was independent of the route of administration (Table 1). No effect of IL-1 administration was seen in plasma levels of seromucoid. Seromucoid is a crude precipitation fraction of glycosylated proteins which according to our two-dimensional immunoelectrophoretic studies is composed of zl acid glycoprotein, 1 cysteine protease inhibitor and possibly hemopexin. Caeruloplasmin levels were unaffected whereas albumin plasma levels were significantly reduced in the groups of rats injected i.p. or i.v. but not s.c. Dose response to rhIL-l administered s.c. The effect of single s.c. doses of 2 and 6 ng g-1 of IL-I was compared to the effect of 2nggof IL-I administered at 0, 2 and 4 h. Plasma was obtained 24 h after the initial injection. The multiple dose was far more effective than the single dose with greater changes in plasma levels of albumin, seromucoid, haptoglobin and caeruloplasmin in the multiple dosed rats ( Table 2). Injection of IL-lcz (6 ng gs.c.) in both experiments (Tables 1 and 2) induced different degrees of changes in haptoglobin    Values represent the mean 4-SE from groups of five rats. Group designated as vehicle were implanted with Alzet osmotic minipumps containing 0.2% albumin in saline and releasing /1 h -1 The APP levels in plasma were determined 4 days after implantation of the pumps. Rats dosed with vehicle alone from the minipumps has significantly lower plasma levels of albumin (*p < 0.05) and greater levels of caeruloplasmin (*p < 0.05) than non-implanted rats (no minimpump). Significant changes in the levels of albumin (*p < 0.05) were obtained from rats administered with 0.6 ng h -1 Administration of IL-I at higher doeses of 2.1,6.3, and 21 n h induced dose dependent changes in all the APPs measured, most of the changes were statistically significantly different from the vehicle alone group (**p < 0.01 to ***p < 0.001 as determined by Student's t-test (unpaired, two-tailed). Other values shown were not statistically significantly different from the vehicle dosed group.
within the granuloma and the granuloma weight varied greatly between rats within each group, therefore the exact volume or weight could not be used as a parameter except for being present or absent.
Dose response studies to cytokines released from minipumps: The effects of recombinant human IL-I, IL-lfl, IL-6 and TNFz released from implanted osmotic minipumps on APP levels in the rat were studied in separate dose response experiments. The effect of the minipump released cytokine was compared to minipump released 0.2% albumin and expressed as a percentage. Implanted minipumps containing 0.2% albumin were used as a vehicle control because of the minor inflammatory effect of implanting the minipumps and because of the variation in base-line levels of APPs. The maximal % changes shown for IL-10 are consistent with the changes associated with (adjuvant) inflammation. 12 Recombinant human interleukin-lfl was the most potent cytokine tested with activity at 0. 21  and TNF (-'-I-'-) on plasma APP levels are shown after continuous release of the cytokines for 5 days. Statistical significance to the vehicle control group is shown as *p < 0.05; **p < 0.01 and ***p < 0.001 according to Student's t-test (unpaired; two-tailed). also produced a variable exudate inside the granuloma in three out of five rats. Interleukin-10 had similar effects to those seen with IL-lfl, but rhlL-lfl was ten times more active than rhlL-10 in this rat model (Fig. 1).
At the highest dose tested (30 ng h -1) rhlL-6 significantly increased haptoglobin and caeruloplasmin levels by 40% and 60% respectively when compared with the IL-1 induced levels. IL-6 did not produce granulomas or exudate fluid around the pump.
Recombinant human turnout necrosis factor at doses up to 50 ng h -1 did not induce granuloma formation or any exudate. The only change seen in the APP levels was a slight increase (20%) in the levels of seromucoid at the highest dose tested, 50 ng h-1 (Fig. 1).

Discussion
Administration of rhIL-10 by single injection with doses up to 600 ng/rat induced changes in the levels of APPs measured 24 h after injection, but these changes were not pronounced when compared with changes seen in inflammation. Administering rhlL-10 by multiple dose was more effective than the equivalent single dose with a significant change in levels of seromucoid, haptoglobin and caeruloplasmin. Similar results have been shown by De J ong et al., who demonstrated that i.p.
injections of rhlL-1 at doses of 500 #g at 0, 2 and 4h increased liver mRNA levels of zl acid glycoprotein and transferrin but decreased albumin and fibrinogen. Also sustained short term release of rhlL-1 (150#g i.v.) by a 6 h infusion in rats increased 01 acid glycoprotein but did not alter plasma levels of albumin, 01 cysteine protease inhibitor or 0 2 macroglobulin. 17 Administration of IL-10 by sustained release from the osmotic minipump (6.3 ng h -) produced a greater response than a single injection (6 ng g-) even though the total dose administered was similar (approximately 600 ng/rat). Implantation of minipumps had only a minor effect on APP levels measured 4 days after implantation, therefore the enhanced activity of the cytokine must be dependent on the continuous release overcoming the rate of cytokine degradation. Few authors have used osmotic minipumps to administer cytokines when studying changes in APP levels; Gordeuk et al., 8 have studied the et:fects of rhIL-1 on iron metabolism in mice and could not show any difference between single bolus and perfused IL-1 on the degree of decreased iron levels. 8 The dose related eects of rhlL-10 on APPs shows that the most sensitive APP of the number measured was albumin whose levels were decreased with administration of 0.6 ng h -1. This sensitivity of the liver synthesis of albumin to low dose cytokines may have a beneficial eect in vivo by reducing albumin synthesis before allowing the liver to increase synthesis of other proteins.
Separate comparative studies using osmotic minipumps to administer IL-10, IL-lfl, IL-6 and TNF0 were performed to determine which cytokine(s) modulated the synthesis of specific APPs in an in vivo model. In this model, implantation of the minipump caused a mild inflammation which induced a weak acute phase response (as seen with the increase in caeruloplasmin in the previous experiment), thus the levels of APP inducing cytokines and corticosteroids in the blood were probably raised above normal background levels. Changes in levels of the APPs were therefore compared to the APP levels from animals implanted with osmotic minipumps containing 0.2% albumin as a vehicle control.
Albumin levels were significantly reduced by both IL-10 and IL-lfl but not IL-6 or TNF0. This efl:ect of IL-1 is probably via a direct mediation because IL-1 can induce TNF0 and IL-6 release from cells 19' which could alter liver APP synthesis. In vitro studies by Baumann et al., 6 indicate that IL-1, TNF0 and IL-6 act to decrease albumin production by hepatocytes, and this data is supported by Andus et al. 21 Administration of IL-1 by repeat i.p. doses in vivo reduced albumin synthesis; IL-6 did not induce such a pronounced reduction. 9 Gresser et a/., 22 have demonstrated a weak reduction in albumin levels after single administration of rmTNF. Co-administration of dexamethasone with cytokines in human hepatoma cell cultures has been shown to reverse hypoalbuminia associated with cytokines, 6 therefore endogenous corticosteroids may act to inhibit IL-6 and TNF albumin lowering effects but not the effect of IL-1. The diflerence in potency between IL-10 and IL-lfl in the rat has been previously reported by IL-lfl and TNF0q although TNF0 only caused a minor increase in seromucoid levels (20% of total response to IL-1). Seromucoid is a crude precipitation fraction of glycosylated proteins such as 0 acid glycoprotein, 0 cysteine proteinase inhibitor and hemopexin. Interleukin-1 has been shown to increase 0 acid glycoprotein levels both in vitro 24 and in vivo. 2 Interleukin-6 had no eect on seromucoid levels which differs from the findings of Markinovic et a/., 25 but is similar to those of Geiger et a/., 24 who measured liver mRNA levels of acute phase proteins after i.p. injection of 800 ng of IL-6 but this dose only increased fibrinogen to a similar extent to that induced by injection of turpentine. In the same study 01 acid glycoprotein and 1 cysteine protease inhibitor were only marginally increased by  Haptoglobin levels were raised by IL-I, IL-lfl and IL-6. In vitro studies using hepatocytes have shown that IL-6 acts in synergy with IL-1 and glucocorticoids to induce both haptoglobin mRNA and protein. 26 In vivo studies have shown that IL-6 and IL-1 both induce haptoglobin levels in the rat. 25,27 Caerulosplasmin is not a widely studied APP because levels of this protein are only doubled in inflammatory conditions. In these experiments caeruloplasmin levels were increased by IL-1 and IL-6 administration with IL-6 inducing 60% of the change seen with IL-1, thus IL-6 may be the main inducer of caeruloplasmin.
Both IL-10 and IL-lfl produced a granulomatous response around the minipump, a similar finding has been shown by Dunn et a/., 28 who studied the effects of slow release cytokines from implanted ethylene vinyl acetate sponges. The fluid surrounding the pumps contained large numbers of red blood cells and PMNs which is consistent with the reported effects of IL-lfl in inducing a Shwartzmanlike reaction in rabbit skin after intradermal injections. 29 The infiltration of PMNs into the skin has also been reported to occur in rabbit skin after i.d. injection of IL-lfl and IL-10 at doses of 0.1 ng 3 and 2.5 ng/injection site respectively. 31 No granulomatous response was seen with implanted osmotic minipumps containing vehicle, IL-6 or TNF0.
Overall, the results suggest that in vivo in the rat where there would be continuous release of IL-1, IL-6, TNF and glucocorticoids at basal levels, IL-6 may be the major inducer of caeruloplasmin and act in synergy with IL-1 and glucocorticoids to induce haptoglobin. IL-1 is probably the main inducer of the seromucoid proteins synthesis and hypoalbuminia. It would appear that TNF has only a minor role in the production of the seromucoid proteins.
The use of osmotic minipumps to infuse cytokines into animals has shown that greater effects can be obtained with 'low' concentrations of cytokines and also that this method may be physiologically more relevant than single bolus dose since in inflammatory conditions cytokines would be released continuously over the period of the inflammation.