Effects of keratinocyte growth factor (KGF) on gut growth and repair

, . 184: 316–322 (1998)

EFFECTS OF KERATINOCYTE GROWTH FACTOR(KGF) ON GUT GROWTH AND REPAIR

. *, 1, 2, 3, 4, . 4, . 4 . 2

1Department of Medicine and Therapeutics, University of Leicester, Leicester, U.K.2Histopathology Unit, Imperial Cancer Research Fund, London U.K.3Department of Physiology, University of Liverpool, Liverpool, U.K.

4Department of Medicine, Royal Postgraduate Medical School, London, U.K.

SUMMARY

Keratinocyte growth factor (KGF) is a mitogen found throughout the gastrointestinal tract, but its role in gastrointestinalpathophysiology is unclear. The effect of recombinant KGF on gut growth and repair has been examined using a variety of in vivomodels. Rats receiving total parenteral nutrition had co-infusions of KGF or control for 6 days. Changes in gut growth (wet weight andvincristine-induced metaphase arrest) were then assessed. The effects of KGF on gastric repair and acid secretion in rats were determinedusing an indomethacin (20 mg/kg)/restraint model and animals fitted with chronic gastric fistulae. KGF at 0·1, 1, and 3 mg/kg increasedgut growth as assessed by wet weight throughout the gastrointestinal tract and increased vincristine-induced accumulation of metaphasesin the stomach and small intestine but not in the colon. Plasma gastrin, peptide YY, enteroglucagon, and glucagon-like peptide-1 wereall increased, whereas insulin was lowered by KGF (all P<0·01). KGF was ineffective in reducing indomethacin-induced gastric damagebut caused a reduction in basal acid secretion of about 35 and 50 per cent when administered at 0·2 or 5 mg/kg (P<0·05). These studiessupport the idea that KGF is involved in the control of proliferation of the gastrointestinal tract. They do not provide evidence, however,for a role in the early reparative process invoked during short-term models of gastrointestinal injury. ? 1998 John Wiley & Sons, Ltd.

J. Pathol. 184: 316–322, 1998.

KEY WORDS—gastrointestinal tract; epithelium; cell division; cell proliferation; growth control; intravenous (parenteral) nutrition

INTRODUCTION

Keratinocyte growth factor (KGF), a member of thefibroblast growth factor family (FGF7), was originallyisolated from conditioned medium derived from ahuman lung fibroblast cell line and is a potent stimulantof keratinocyte proliferation.1 It has subsequently beenshown to stimulate proliferation of a variety of epi-thelial cell lines by binding to a splice variant of FGFreceptor 2.2 KGF mRNA and peptide are produced bymesenchymal tissue and the KGF receptor is expressedby epithelial cells of a number of embryonic and adultorgans including the respiratory, urogenital, andgastrointestinal systems,3 suggesting a role in epithelial–mesenchymal interactions.4 The presence of the KGFreceptor and its peptide throughout the gastrointestinaltract suggests a role in the maintenance of mucosalintegrity or in stimulating repair. This idea is supportedby the findings that recombinant KGF stimulatesgrowth of the gastrointestinal tract when administeredsystemically5 and that KGF ameliorates damage in anexperimental model of colitis in rats.6 The physiologicalrole of KGF in the control of upper gastrointestinalfunction is, however, largely unexplored. We thereforeexamined its effects on gut growth, gastrointestinal

hormone release, and gastric acid secretion, and itsability to influence gastric damage using a variety ofin vivo models.

MATERIALS AND METHODS

Materials

Recombinant human KGF, expressed in Escherichiacoli, was a gift from Dr B. Boyle of Amgen Inc.,Thousand Oaks, CA, U.S.A. All chemicals wereobtained from Sigma U.K. Ltd., Poole, Dorset, U.K.unless otherwise stated.

Ethical approval

All animal facilities and protocols were approved andperformed under appropriate project licences issued bythe United Kingdom Home Office in accordance withthe Animals (Scientific procedures) Act 1986.

Preparation of animals and experimental protocol forin vivo study examining the effect of KGF on gut growthin rats receiving total parenteral nutrition

Male August rats (200–250 g) were anaesthetized withHypnorm and Diazepam, and a silastic cannula was tiedinto the right external jugular vein. The cannula wasconnected through a stainless steel skin button and

*Correspondence to: Professor R. J. Playford, University Divisionof Gastroenterology, Leicester General Hospital, Gwendolen Road,Leicester LE5 4PW, U.K.Contract grant sponsors: Medical Research Council; Wellcome

Trust.

CCC 0022–3417/98/030316–07 $17.50 Received 20 June 1997? 1998 John Wiley & Sons, Ltd. Accepted 22 September 1997

tether to a fluid swivel joint (SMA, Barnett, U.K.). Therats were housed individually in wire-bottomed cages.The total parenteral nutrition (TPN) diet was pumpedinto the rats, from a refrigerator, by a multichannelperistaltic pump, at a rate of 60 ml/rat per day, giving1·8 g N, 6·0 g lipid, 8·5 g glucose, and 1047 kJ per kg perday.7Four groups of rats (N=6 per group) received TPN

solution alone or TPN supplemented with various dosesof KGF. A further group was orally fed throughout theexperiment. All rats received the different treatments for6 days and were then injected with 1 mg/kg of vincristinesulphate intraperitoneally (David Bull Laboratories,Warwick, U.K.), anaesthetized 2 h later with pento-barbitone, and then exsanguinated by cardiac puncturefor subsequent hormone assay.The wet weight of various sections of the gastro-

intestinal tract was recorded and samples of thestomach, small intestine, and colon (defined by theirpercentage length) were fixed in Carnoy’s fluid for 3 hand then stored in 70 per cent (v/v) ethanol. Tissues weresubsequently stained using the Feulgen reaction and thecrypts displayed by microdissection.7 The numbers ofarrested metaphases in 20 crypts per animal per site werecounted.To examine the relative contribution of mucosal

growth to the total changes in gut growth seen in thecolon, cross-sections of colon were examined under amicroscope fitted with a Weibel II graticule (GraticulesLtd., Morely Road, Kent, U.K.). The number of timesthat the end of the graticule lines intersected a particulararea of the intestine was counted and the number ofthese ‘hits’ reflects the volume fraction occupied by thetissue component.8

Hormone assays

Plasma concentrations of gut hormones weremeasured by radioimmunoassays for enteroglucagon,9glucagon-like polypeptide I (GLP-I),10 peptide YY(PYY),11 and gastrin.12 Enteroglucagon was calculatedby subtracting specifically measured pancreatic glucagon(C-terminal immunoreactivity measured with the anti-serum RCS5) from total N-terminal glucagon immuno-reactivity (measured with antiserum R59). The assayswere capable of detecting pancreatic glucagon 2 pmol/l,enteroglucagon 5 pmol/l, GLP-I 6 pmol/l, PYY2·5 pmol/l, and gastrin 2 pmol/l with 95 per centconfidence.

Effect of KGF on indomethacin-induced gastric damage

The ability of KGF to decrease gastric damagewas tested using a rat model in which mucosal lesionsare induced by a combination of indomethacin andrestraint.13 Epidermal growth factor (EGF), a well-established cytoprotective agent, was used as a positivecontrol. Briefly, under light ether anaesthesia, rats hadtwo subcutaneous cannulae inserted into the nape of theneck and were then placed in Bollman-type restraintcages. Animals then received a continuous subcutaneousinfusion of KGF at 0·3, 1 or 3 mg/kg per h, EGF at

5 ìg/kg per h (positive control) or control infusion(saline). Thirty minutes later, all animals received20 mg/kg of indomethacin subcutaneously via thesecond cannula. Three hours later, the animals werekilled and the stomachs were removed and inflated with4 ml of 10 per cent formalin.The next day, the stomachs were opened and placed in

fresh formalin prior to assessment. The stomachs wererandomly coded and all analyses of gastric damage wereassessed blind. The total ulcerated area (mm2/stomach)was assessed using a dissecting microscope (#10) withthe aid of a square grid. The stomachs were thenembedded in wax, sections were cut and stained, and thedepth of damage was assessed microscopically and givena microscopic ulcer score, as previously described.14

Acid secretion studies

To determine if KGF might influence gastric mucosalintegrity by altering acid secretion, rats fitted withchronic gastric cannulae were injected with variousdoses of KGF and the effect on basal acid secretion wasmeasured using a standard technique.13 These animalsdid not receive indomethacin. Following an overnightfast, rats fitted with gastric cannulae were lightlyrestrained in Bollman-type cages; the cannulae caps wereremoved; and their stomachs were flushed with isotonicsaline (37)C) to remove food debris. The stomachs werethen allowed to drain for 1 h prior to the start of thestudy. Gastric secretions were collected and pooledevery 15 min for the first hour before giving a sub-cutaneous injection of saline or KGF (N=6 per dose).The concentration of H+ ions was determined bytitration to pH 7·0 with 0·02 NaOH using an auto-titrator assembly (RTS 822, Radiometer, Copenhagen,Denmark). Acid secretion was measured over 15-minperiods, and the data presented are the mean of theresults from the two 15-min collections of the middle ofthe hour before injection, and similarly for the middle ofthe first and second hour after injection.

Statistics

All results are presented as the group mean&standard error of that mean. Data were tested as appro-priate by two-sided t-test or by analysis of variance.When there was a statistically significant result (P<0·05)using the analyses of variance, individual treatmentswere compared against control using Dunnets test(based on the group means and residual derived fromthe ANOVA).

RESULTS

KGF infusion

KGF caused significant dose-related increases in theweight of the stomach, small intestine, caecum, andcolon, with the stomach response being the most promi-nent (2·4-fold greater than TPN alone, Fig. 1). KGF hadno significant effect on the length of the small intestine,

317GASTROINTESTINAL EFFECT OF KGF

? 1998 John Wiley & Sons, Ltd. , . 184: 316–322 (1998)

but was associated with a 10 per cent increase in thelength of the colon at the highest dose (18·2&0·4 cmwhen KGF was infused at 3 mg/kg compared with16·6&0·3 cm in control animals, P<0·05). The increasein colonic gut weight was associated with a significantincrease in the volume fraction of the mucosa (53·9&5·0per cent in control vs. 62·9&3·8 per cent in rats receiv-ing 1 mg/kg of KGF, P<0·05). Increased numbers ofgoblet cells were seen in both the small and the largeintestine, as previously described by Housley et al.5 (datanot shown).A highly significant dose-related increase in prolifer-

ation was seen in the stomach and small intestine, withproliferation raised to values greater than those seen inthe orally fed rats (Fig. 2). No significant increase,however, was seen in response to the KGF in the colon(Fig. 2).Plasma gastrin, PYY, enteroglucagon, and GLP-I

were all significantly increased and plasma insulin wassignificantly decreased in response to infusion of KGF(Fig. 3).

Effect of KGF on indomethacin-induced gastric damage

As expected, the positive control (EGF) significantlyreduced the amount of macroscopic damage induced byindomethacin. There was no reduction, however, in thedegree of macroscopic damage seen in animals receivingKGF (Fig. 4). Similar results were seen when usinghistological scoring (data not shown).

Acid secretion

For the first hour following administration, KGFdecreased basal acid secretion by about 35 per cent wheninjected subcutaneously at a dose of 0·2 mg/kg and by50 per cent when administered at 5 mg/kg (Fig. 5).

DISCUSSION

We have used a variety of in vivo models of growthand repair to examine the role of KGF, a member ofthe FGF family, in the gastrointestinal tract. KGFstimulated growth of the entire gastrointestinal tract andreduced basal gastric acid secretion. KGF was ineffec-tive, however, in reducing the amount of gastric injuryinduced by indomethacin and restraint.

KGF and gut growth

The in vivo total parenteral feeding rat model is arobust, reproducible model to examine proliferativeresponses throughout the gastrointestinal tract.7 Feed-ing the animals using total parenteral nutrition has themajor advantage that the intestine becomes hypoplasticwithout recourse to starvation. The intestine is thus in asteady state of basal cell proliferation, as are endo-genous secretions and gut hormones. In addition, theseveral variables associated with altered food intake andfood intake patterns are removed.

Fig. 1—Effect of keratinocyte growth factor (KGF) on proliferation ofthe gastrointestinal tract as assessed by wet weight. Rats were fed usingtotal parenteral nutrition for 6 days and received co-infusions ofvarious doses of KGF. Results are expressed as mean&SEM of N=6per group. * and ** signify P<0·05 and 0·01 vs. animals receiving TPNalone. Values seen in chow-fed animals (which were not given KGF)are also shown but were not incorporated into the statistical analyses

318 R. J. PLAYFORD ET AL.


Our studies confirm a previous report by Housleyand co-workers examining the effect of dailyintraperitoneal injections of KGF in rats. Using a dailydose of KGF similar to that used in the present study,they found that KGF induced growth of the entiregastrointestinal tract.5 The most marked response inour study was seen in the stomach and, in combinationwith our finding that KGF influences acid secretion,this suggests that KGF may be of particular impor-tance to this organ. As expected, the increases in theweight of the stomach and small intestine were associ-ated with increased proliferation, as assessed by 2 hmetaphase determination. Somewhat surprisingly, theincrease in the wet weight of the colon was notassociated with increased accumulation of metaphaseswithin the crypts of the colonic mucosa. We thereforeconsidered the possibility that changes in tissue massof the muscle layer might be responsible for thisincrease in wet weight, but morphometric assessmentshowed no increase in the volume fraction of thislayer, in contrast to the increase in volume fraction ofthe mucosal layer.Tissue mass is dependent on the equilibrium estab-

lished between cell production, and loss, includingapoptosis. Housley and co-workers have reported anincrease in the number of bifurcating crypts in KGF-treated animals, supporting the idea that some of thesechanges in gut growth were mediated through cryptfission.5 The finding that growth factors can influencecrypt fission is important, as one of the standard ways ofquantitating gut growth is to express results in termsof the number of mitoses seen per crypt. Crypt fissionwill change the denominator (i.e., number of crypts),potentially leading to paradoxical results. The effect ofKGF on apoptosis is unknown and further work isrequired to examine these areas in more detail.In addition to direct actions of KGF on the crypt stem

cells, trophic responses might have been mediated viachanges in circulating gut hormones. Administration ofKGF resulted in increased plasma gastrin, which isknown to stimulate proliferation of the stomach, but therole of gastrin in proliferation in the rest of the gastro-intestinal tract is uncertain. Increased plasma con-centrations of the gut hormones PYY, GLP-I andenteroglucagon were also seen in response to admin-istration of KGF. Raised PYY concentrations inhumans have been associated with increased gutgrowth, although direct infusions of PYY into rats havenot demonstrated a trophic effect.15 Uncertainties alsoexist over the potential gastrointestinal trophic effectof members of the proglucagon-derived family ofpeptides.16

Fig. 2—Effect of KGF on proliferation of the gastrointestinal tractas assessed by metaphase accumulation. The same animals as thoseused in Fig. 1 were given vincristine 2 h before killing. Accumulatedmetaphases were then counted and expressed on a per crypt basis.Symbols as in Fig. 1. Results show changes similar to those foundmeasuring wet weight (Fig. 1), except that no increase in metaphaseaccumulation was found in the colon



Effects of KGF on gastric acid secretion and mucosalrepairThere are many well-validated acute models of gastric

injury. We chose the indomethacin/restraint model

because peptic ulceration due to non-steroidal anti-inflammatory drugs is a major source of morbidity andmortality in humans. We have previously used thismodel to demonstrate the cytoprotective activity of the

Fig. 3—Plasma hormone levels were measured in the animals used in Fig. 1; symbols as in Fig. 1



trefoil peptides.13 This model is pH-dependent andwe therefore also used our well-validated protocolinvolving rats fitted with chronic gastric fistulae13 toassess the effects of KGF on acid secretion. This modelallows the accurate analysis of changes in acid secretion

without the need for administering anaesthesia, whichcan independently influence gastric secretion.KGF did not affect the degree of damage induced by

indomethacin and restraint. The mechanism by whichthese factors induce damage is likely to be multifactorialand include alteration in mucosal prostaglandin levels,blood flow, and neutrophil activation. The admin-istration of several peptide factors thought to be import-ant in mediating intestinal repair, such as EGF and thetrefoil peptides, has been shown to reduce the amount ofdamage using this model (e.g., ref. 13). Protection bythese peptides is independent of their proliferativeactivity or their effect on acid secretion.KGF reduced basal acid secretion by about 35–50 per

cent, which may well be of physiological relevance. Amuch greater amount of acid suppression is, however,required to influence the degree of damage induced byindomethacin.17 Further studies are required to examinewhether this acid suppressant effect of KGF wasmediated via direct effects on parietal cells, or throughindirect actions such as increasing local somatostatinrelease from gastric D cells.In contrast to our findings involving the adminstra-

tion of damaging agents to the stomach, Zeeh et al. haverecently shown that KGF, in doses similar to those usedby us, decreases the degree of damage induced bytrinitrobenzesulphonic acid (TNB) and ethanol in thecolon.6 Importantly, a single injection of KGF prior tothe TNB failed to affect the degree of damage, whereaschronic administration (daily injections) followingadministration of the TNB did reduce the amount ofinjury. These results suggest that it is a delayed action ofKGF, such as increasing proliferation or increasinggoblet cell numbers, which is important in mediating thishealing response. This contrasts with the ‘cell stabiliz-ation’ effect seen with some other peptides involvedin mucosal repair, such as EGF, where administrationof the peptide prior to injury reduces the degree ofsubsequent damage.It is possible that the role of KGF varies depending on

the tissue examined and also on the experimental modelused. In the skin, for example, KGF and its receptorhave been reported to be markedly upregulated at sitesof wounding18 and recombinant KGF has been shownto stimulate the migration of human keratinocytes.19These studies suggest that KGF may be involved in theearly stages of re-epithelialization in the skin. In con-trast, no change in KGF expression was seen in ratgastric mucosal damage using indomethacin or aceticacid.20 Brauchle et al., however, have reported markedoverexpression of KGF at sites of inflammatory boweldisease.21 One possible explanation for this apparentdiscrepancy is that upregulation of KGF in the gastro-intestinal tract may only occur at sites of chronicinflammation. Our present studies, however, provide noevidence for a role of KGF in the early stages ofgastrointestinal repair invoked during short-termmodels of gastric injury. There is at present little infor-mation regarding endogenous concentrations of KGF inthe gastrointestinal tract. We are therefore not able todetermine if the effects of KGF demonstrated in thepresent studies are of physiological or pharmacological

Fig. 4—(Lack of) Effect of KGF on the degree of gastric damageinduced by indomethacin and restraint. Each rat received indometh-acin (20 mg/kg, subcutaneously) and was restrained for 3 h. Animalsalso received continuous subcutaneous infusions of saline (negativecontrol), or various doses of KGF or EGF (positive control). Resultsare expressed as mean&SEM of N=6 per group. **signifies P<0·01vs. animals receiving saline

Fig. 5—Rats fitted with chronic gastric fistulae received control, orKGF at 0·2 or 5 mg/kg subcutaneously. Changes in acid output werethen determined over the next 2 h. *signifies P<0·05 vs. acid outputseen during the initial baseline period



significance. Further research, including the produc-tion of transgenic KGF overexpressors and knockoutmodels, is required to provide further insight into therole of KGF in vivo.

ACKNOWLEDGEMENTS

Thanks are due to Graham Dockray for assistancewith the acid secretion studies and to the MedicalResearch Council and the Wellcome Trust for funding.

REFERENCES1. Rubin JS, Osada H, Finch PH, Taylor WG, Rudikoff S, Aaronson SA.

Purification and characterization of a newly identified growth factor specificfor epithelial cells. Proc Natl Acad Sci USA 1989; 86: 802–806.

2. Miki T, Bottaro DP, Fleming TP, et al. Determination of ligand-bindingspecificity by alternative splicing: two distinct growth factor receptorsencoded by a single gene. Proc Natl Acad Sci USA 1992; 89: 246–250.

3. Mason IJ, Fuller-Pace F, et al. FGF-7 (keratinocyte growth factor) expres-sion during mouse development suggests roles in myogenesis, forebrainregionalisation and epithelial–mesenchymal interactions. Mech Dev 1994;45: 15–30.

4. Finch PW, Cunha GR, et al. Pattern of keratinocyte growth factor andkeratinocyte growth factor receptor expression during mouse fetal develop-ment suggests a role in mediating morphogenetic mesenchymal–epithelialinteractions. Dev Dyn 1995; 203: 223–240.

5. Housley RM, Morris CF, et al. Keratinocyte growth factor inducesproliferation of hepatocytes and epithelial cells throughout the rat gastro-intestinal tract. J Clin Invest 1994; 94: 1764–1777.

6. Zeeh JM, Procaccino F, Hoffmann P, et al. Keratinocyte growth-factorameliorates mucosal injury in an experimental model of colitis in rats.Gastroenterology 1996; 110: 1077–1083.

7. Goodlad RA, Boulton R, Playford RJ. Comparison of the mitogenicactivity of human EGF1–53 and EGF1–48 in vitro and in vivo. Clin Sci 1996;91: 503–507.

8. Aherne WA, Dunhill MS. In: Aherne WA, Dunhill MS, eds. London:Arnold Publishing, 1982; 33–44.

9. Ghatei MA, Uttenhall LO, et al. Molecular forms of human enteroglucagonin tissue and plasma: plasma response to nutrient stimuli in health anddisorder of the upper gastrointestinal tract. J Clin Endocrinol Metab 1983;57: 488–495.

10. Uttenhall LO, Ghiglione M, George SK, Bishop AE, Polak JM, Bloom SR.Molecular forms of glucagon-like peptide-1 in human pancreas andglucagonomas. J Clin Endocrinol Metab 1985; 61: 472–479.

11. Adrian TE, Ferri GI, et al. Human distribution of a putative new guthormone, peptide YY. Gastroenterology 1985; 89: 1070–1077.

12. Bryant MG, Adrian TE. Gastrin. In: Bloom SR, Long RG, eds. Radio-immunoassay of Gut Regulatory Peptides. London: W. B. Saunders, 1982;51–59.

13. Playford RJ, Marchbank T, Chinnery R, Thim L, Hanby AH. Humanspasmolytic polypeptide is a cytoprotective agent which stimulates cellmigration. Gastroenterology 1995; 108: 108–116.

14. Playford RJ, Vesey DA, et al. Dose-dependent effects of fentanyl onindomethacin-induced gastric damage. Digestion 1991; 49: 198–203.

15. Goodlad RA, Ghatei MA, Domin J, Bloom SR, Wright NA. Is PYYtrophic to the intestinal epithelium of parenterally fed rats? Digestion 1990;46: 177–181.

16. Johnson LR, McCormack SA. Regulation of gastrointestinal mucosalgrowth. In: Johnson LR, ed. Physiology of the Gastrointestinal Tract. 3rdedn. New York: Raven Press, 19XX; 626–627.

17. Elliott SL, Ferris RJ, Giraud AS, Cook GA, Skeljo MV, Yeomans ND.Indomethacin damage to rat gastric mucosa is markedly dependent onluminal pH. Clin Exp Pharmacol Physiol 1996; 23: 432–434.

18. Marchese C, Chedid M, Dirsch OR, et al. Modulation of keratinocytegrowth-factor and its receptor in reepithelializing human skin. J Exp Med1995; 182: 1369– 1376.

19. Tsuboi R, Sato C, Kurita Y, Ron D, Rubin JS, Ogawa H. Keratinocytegrowth-factor (FGF-7) stimulated migration and plasminogen-activatoractivity of normal human keratinocytes. J Invest Dermatol 1993; 101: 49–53.

20. Kinoshita Y, Nakata H, Hassan S, et al. Gene expression of keratinocyteand hepatocyte growth factors during the healing of rat gastric mucosallesions. Gastroenterology 1995; 109: 1068–1077.

21. Brauchle M, Rassler R, Werner S. Suppression of keratinocyte growth-factor expression by glucocorticoids in-vitro and during wound-healing.J Invest Dermatol 1995; 105: 579–584.



Documents

Effects of keratinocyte growth factor (KGF) on gut growth and repair