J. Pathol. 186: 186–191 (1998)

VASCULAR ENDOTHELIAL GROWTH FACTOR ISCONSTITUTIVELY EXPRESSED IN NORMAL HUMAN

SALIVARY GLANDS AND IS SECRETED IN THESALIVA OF HEALTHY INDIVIDUALS

1, 2, 2, 3, 4 2,5*

1Institute of Clinical Pathology, University of Vienna Medical School, Vienna, Austria2Division of Immunology, Allergy and Infectious Diseases, Department of Dermatology, University of Vienna Medical School,

Vienna, Austria3Department of Otolaryngology, University of Vienna Medical School, Vienna, Austria

4Department of Vascular Biology and Thrombosis Research, University of Vienna Medical School, Vienna, Austria5Centre de Recherche et d’Investigation Epidermique et Sensorielle (CE.R.I.E.S.), Neuilly, France

SUMMARY

The expression of vascular endothelial growth factor (VEGF), a specific mitogen for endothelial cells, was examined in salivary glandsand in normal saliva. In normal salivary glands, VEGF mRNA and protein were strongly present in acinar cells, whereas little or noVEGF was found in ductal cells. In chronically inflamed glands, VEGF protein was in addition present in ductal elements and ininfiltrating mononuclear cells. No difference of VEGF expression was observed between benign and malignant salivary gland tumours.By ELISA, whole saliva of 24 healthy individuals contained up to 2·5 ng/ml (mean 1·4 ng/ml; SD 0·77 ng/ml) of VEGF, confirming theconstitutive secretion of this cytokine by human salivary glands. Western blot analysis of normal saliva under non-reducing conditionsdetected anti-VEGF reactive protein moieties of ~46 kD, corresponding to VEGF secreted by cells in tissue culture. Additionalanti-VEGF reactive proteins of ~60 and 90 kD were detected in the saliva of some individuals. The presence of considerable quantitiesof VEGF in normal human saliva suggests an important role for this cytokine in the maintenance of the homeostasis of mucousmembranes, with rapid induction of neoangiogenesis by salivary VEGF helping to accelerate wound healing within the oral cavity.Moreover, salivary VEGF may permeabilize intraglandular capillaries and thus participate in the regulation of saliva production itself.? 1998 John Wiley & Sons, Ltd.

KEY WORDS—VEGF; saliva; wound healing; salivary gland tumours; fenestration

*Correspondence to: Dr Erwin Tschachler, Division of Immunol-ogy, Allergy and Infectious Diseases, Department of Dermatology,University of Vienna Medical School, Währinger Gürtel 18–20,A-1090 Vienna, Austria.

Contract/grant sponsor: Austrian Science Foundation; Contract/grant number: P01437-MED.

Contract/grant sponsor: Medizinisch Wissenschaftlichen Fonds des

INTRODUCTION

Vascular endothelial growth factor (VEGF)1 repre-sents a family of secreted growth factors that act specifi-cally on vascular endothelial cells (EC). Biochemically,VEGF is a heparin-binding glycoprotein2 that occurs infour molecular forms, consisting of 121, 165, 189, and206 amino acids derived from the same gene by alterna-tive mRNA splicing.3,4 Flt-1 and flk/KDR, the twoplasma membrane tyrosine kinase VEGF receptors, arepredominantly expressed on EC.5,6

Two prominent biological effects of VEGF are toinduce mitogenesis and migration of EC1,2,7–9 and topermeabilize the microvasculature about 50 000 timesmore potently than histamine.10–12 VEGF induces ang-iogenesis under physiological conditions, as in prolifer-ating endometrium,13 corpus luteum formation,14 andembryogenesis,15 and in pathological circumstances, asduring wound healing,16,17 tissue inflammation,9,18 and

Bürgermeisters der Bundeshauptstadt Wien.

CCC 0022–3417/98/100186–06$17.50? 1998 John Wiley & Sons, Ltd.

tumour-associated neovascularization19,20 (for reviews,see ref. 21).

Besides its protective function for oral mucosa, salivacontributes directly to oral22,23 and extra-oral woundhealing.24,25 This function has been attributed tosalivary immunoglobulins, antibacterial factors,22 andcertain growth factors secreted by salivary glands,including epidermal growth factor (EGF26), transform-ing growth factor á (TGF-á27) and basic fibroblastgrowth factor (bFGF28). Because VEGF is consideredthe most important mediator of angiogenesis duringwound healing,1, 21 we investigated the expression of thiscytokine in normal and pathological salivary glandtissues and its secretion in normal saliva.

MATERIALS AND METHODS

Tissue

Formalin-fixed, paraffin-embedded specimens forimmunohistochemistry and in situ hybridization wereobtained from the files of the Institute of ClinicalPathology, University of Vienna Medical School. Sali-vary gland tumours were diagnosed according to criteriaof the Armed Forces Institute of Pathology.29 Normal

salivary glands were submandibular glands obtained

Received 22 August 1997Accepted 28 April 1998

187VASCULAR ENDOTHELIAL GROWTH FACTOR IN SALIVARY GLANDS AND SALIVA

during neck dissections (n=6) or morphologically in-conspicuous perilesional parotid tissue of Warthin’stumours (n=5).

Immunohistochemistry

Sections were dewaxed and heated in a microwaveoven in 10 m citrate buffer (pH 6·0) at 450 W for 20min. Two different mouse anti-human VEGF mono-clonal antibodies (MAbs) (clone 26503.11: IgG2b, R&DSystems, Abingdon, U.K.; and clone A4.6.1: IgG1,Genentech, San Francisco, CA, U.S.A.) were applied ata concentration of 2 and 4 ìg/ml, respectively, overnightat 4)C. Subsequently, a rabbit anti-mouse antiserum andAPAAP complex were applied (both Dako, Glostrup,Denmark), both at 1:50 dilution. An alkaline phos-phatase substrate (Vector, Burlingame, CA, U.S.A.) wasadded for 30 min. For the negative control, isotype-matched control MAbs (Coulter, Hialeah, FL, U.S.A.)were used as first-step reagents.

In situ hybridization

Non-radioactive in situ hybridization with the pks-VEGF12130 plasmid using digoxygenin (DIG)-UTP forlabelling of riboprobes was carried out as describedpreviously.31 In brief, deparaffinized sections werehybridized at 48)C overnight to 500 ng/ml of the DIGriboprobes. Post-hybridization steps included RNAsetreatment (25 ìg/ml in 2#SSC) at 37)C for 30 min andthree washes in 2#SSC/50 per cent formamide at 54)Cfor 30 min and incubation with alkaline phosphatase-labelled sheep anti-DIG F(ab)2 fragments (1:600) indilution buffer for 1 h at room temperature. Subse-quently, sections were incubated overnight in the chro-mogen solution, containing nitroblue tetrazolium (NBT)and 5-bromo-4-chloro-3-indoyl-phosisole (BCIP).

Detection of VEGF in saliva by ELISA and western blotanalyses

Whole saliva for ELISA and western blot analyseswas obtained from healthy volunteers between 8:00 and10:00 a.m. at least 1 h after the last meal. Before samplecollection, the mouth was rinsed with tap-water. Salivawas collected in 50 ml Falcon tubes (Becton Dickinson,San Jose, CA, U.S.A.) over a time period of not morethan 15 min and frozen at "20)C. For some exper-iments, saliva was collected directly from Stensen’s ductin healthy volunteers through polyethylene tubes understimulation with citric acid. ELISAs for VEGF (Cytim-mune Sciences Inc., College Park, MD, U.S.A.) andbFGF (DRG Instruments, Marburg, Germany) wereperformed according to the protocol provided by themanufacturers. Before the assay, saliva was centrifugedfor 15 min at 4400 g.

For biochemical analysis of VEGF, a combinedimmunoprecipitation/western blot protocol was used.Lysis buffer consisting of 1 per cent NP40, 1 m PMSF,2·5 ìg/ml leupeptin, and 1 per cent aprotinin inphosphate-buffered saline (PBS) was added in a 1:3 ratioto fresh saliva immediately after centrifugation (see

? 1998 John Wiley & Sons, Ltd.

above). For immunoprecipitation, 2 ìg of VEGF MAb26503.11 and 60 ìl of protein A-agarose beads(Boehringer Mannheim, Germany) were added to 15 mlof saliva for 2 h at 4)C. Beads were spun down andwashed three times with PBS. In parallel, the sameprocedure was performed on culture supernatants ofA431 cells30 as a positive control. Before SDS–PAGE,beads were resuspended in 150 ìl of non-reducingSDS–loading buffer and boiled for 5 min. SDS–PAGEand western blotting were performed as described pre-viously.32 After incubation in blocking buffer (5 per centBSA, 5 per cent non-fat dry milk in PBS) overnight at4)C, membranes were incubated with a biotinylatedanti-VEGF MAb 26503.11 at 2 ìg/ml at room tempera-ture. After washing three times with PBS, the mem-branes were incubated with StreptABComplex/HRP(Dako, Glostrup, Denmark) for 30 min. For controlpurposes, the biotinylated MAb used for western blotanalysis was preincubated with excess recombinantVEGF-121 protein (Biomol, Hamburg, Germany) (4 ngVEGF-121 per ng MAb) overnight. Membranes wereexposed to an X-OMAT-AR film (Eastman Kodak,Rochester, NY, U.S.A.) for 30 min after rinsing inwestern blotting detection reagent ECL (Amersham LifeScience, San Jose, CA, U.S.A.).

RESULTS

VEGF mRNA and protein are expressed constitutivelyin salivary glands and in salivary gland tumours

In situ hybridization revealed distinct basolaterallabelling for VEGF mRNA in serous acinar cells (Fig.1A) in each of eight morphologically normal salivaryglands (regular salivary glands n=2, peritumoural sali-vary gland tissues n=6). Mucinous cells (Fig. 1A,arrows) were consistently negative for VEGF mRNA.Intercalated and striated ducts were negative (Fig. 1A),with rare exceptions in which weak reactivity wasdetected (Fig 1A, arrow-heads).

Corresponding with the distribution of VEGFmRNA, strong VEGF protein expression was detectablein the serous acinar cells of all of 11 specimens (Fig. 1C).Whereas weak reactivity was found in a minority ofstriated ducts and scattered intercalated ducts (notshown), mucinous and myoepithelial cells were consist-ently negative for VEGF protein. When MAb 26503.11was substituted for MAb A4.6.1, additional staining of afew mucinous cells and a supranuclear Golgi-like stain-ing pattern in the striated and excretory ducts wereobserved in some specimens (not shown).

In chronic sclerosing sialadenitis (n=5), mRNA label-ling of ductal elements appeared distinctly stronger (Fig.2A, arrows) than in normal tissues. In addition, distinctsignals for VEGF mRNA were found in infiltratinground cells (Fig. 2A, arrow-heads). Corresponding withmRNA production, VEGF protein expression wasfound in residual acinar cells (Fig. 2B, arrows) with bothMAbs used. VEGF staining was distinct in residualstriated ducts (Fig. 2B, small arrows) and, to a lesserdegree, in ductular proliferations (Fig. 2B, arrow-heads)with MAb A4.6.1, whereas MAb 26503.11 detected only

J. Pathol. 186: 186–191 (1998)

188 J. PAMMER ET AL.

focally enhanced ductal VEGF expression (not shown).In addition to glandular elements, MAb A4.6.1 alsodistinctly stained infiltrating cells.

Staining of 48 salivary gland tumours for VEGFprotein revealed no differences between malignant andbenign neoplasms (Table I). In acinic cell carcinomas,staining for VEGF was moderate to strong in themalignant acinar cells, but only very weak in the inter-calated duct-type cells (Fig. 2C, arrows) and in the clearcells (not shown). VEGF staining within adenomas andcarcinomas was variable, inhomogeneous, and mostlyweaker than in normal surrounding glands (Fig. 2D,arrow-heads).

VEGF is present in the saliva of healthy individuals

When testing whole saliva from 24 healthy volunteers,high concentrations of VEGF, ranging from 0·10 to2·50 ng/ml (mean 1·4 ng/ml; SD 0·77 ng/ml), were detect-able. Parotid saliva, sampled directly from Stensen’sduct in four healthy volunteers after stimulation withcitric acid contained between 1·26 and 2·58 ng/ml VEGF(mean 1·95 ng/ml). Compared with VEGF, only smallquantities of bFGF (mean 6·4 pg/ml; SD 11·7) werepresent in the same samples.

? 1998 John Wiley & Sons, Ltd.

A combined immunoprecipitation/western blot analy-sis confirmed the ELISA data. Anti-VEGF reactiveprotein moieties of about 46 kD, corresponding to theVEGF produced by A431 cells in cell culture (Fig. 3A,lane 2), were detected in the saliva specimens from threehealthy volunteers (Fig. 3A, lanes 3–5). In some speci-mens, additional proteins of about 60 and 90 kDwere detected by the MAb. Preincubation of theanti-VEGF MAb with excess recombinant VEGF121abrogated both binding to the 46 kD moieties (Fig. 3B,lanes 2–5) and that to the larger proteins (Fig. 3B, lane5). The positive band at 200 kD in Fig. 3B representsthe anti-VEGF MAb used for immunoprecipitation(Fig. 3B, lanes 1–5).

Fig. 1—Detection of VEGF mRNA and protein in normal salivary glands. As demonstrated by the dark reaction product, VEGF mRNA wasdetected in basolateral parts of serous acinar cells by non-radioactive in situ hybridization (panel A). Weak reactivity was detected on someintercalated ducts (arrow-heads), whereas no reactivity was present in mucinous cells (panel A, arrows). No staining was found after in situhybridization with a VEGF sense RNA probe (panel B). Expression of VEGF protein assessed by immunohistochemistry was strong in the serousacinar cells but absent from mucinous and myoepithelial cells (panel C). No staining was detected when an isotype-matched control MAb was used(panel D). # 270

DISCUSSION

In the present report we demonstrate that VEGFmRNA and protein are expressed constitutively in nor-mal salivary glands and that VEGF is secreted in salivain concentrations sufficient to be angiogenic.8,33,34

An important biological activity of VEGF is its abilityto induce endothelial fenestration of post-capillary andmuscular venules and capillaries.12 Increased VEGF

J. Pathol. 186: 186–191 (1998)

189VASCULAR ENDOTHELIAL GROWTH FACTOR IN SALIVARY GLANDS AND SALIVA

Fig. 2—Expression of VEGF mRNA and protein in inflammatory and neoplastic salivary gland disorders. In chronic sclerosing sialadenitis, in situhybridization detected VEGF mRNA in ductal elements (panel A, arrows) as well as in infiltrating round cells (panel A, arrow-heads). Striated ducts(panel B, small arrows) stained weakly to moderately (compared with normal salivary gland parenchyma) for VEGF protein; ductular proliferations(panel B, arrow-heads) are negative or weakly positive. Note the strong staining of residual acinar cells (panel B, arrows). As depicted in panel C,the malignant acinar cells of an acinic cell carcinoma stained moderately to strongly for VEGF protein, similarly to normal acinar cells, whereasintercalated duct-type cells were negative or only weakly positive (arrows). In tumour cells of adenoid cystic carcinomas (panel D), VEGF wasdistinctly weaker than in normal acinar cells (arrow-heads). (A, B, D) # 270; (C) # 220

Table I—VEGF staining of salivary gland tumours

TotalNo.

Staining intensity Staining reactivity

1 2 3 + + + + + + + + + +

Warthin’s tumour 6* 5 — — 3 1 1 —Pleomorphic adenoma 13 6 6 1 5 4 3 1Basal cell adenoma 6 2 3 1 — 1 5 —Myoepithelial adenoma 4 — 4 — 1 1 2 —Adenoid-cystic carcinoma 5 1 4 — — 1 2 2Mucoepidermoid carcinoma 5 2 3 — 1 3 — 1Acinic cell carcinoma 6 1 5 — — 2 3 1Carcinoma ex mixed tumour 4 3 1 — 2 1 1 —

Staining intensity: 1=weak; 2=moderate; 3=strong.Staining reactivity (proportion of positively staining cells): + =<20 per cent; + + =20–50 per cent; + + + =50–80 per cent; + + + + =>80 per cent.*One Warthin’s tumour was negative for VEGF.

production has been reported in the kidneys, thepancreas and the choroid plexus, where it is thought toinduce and/or maintain capillary fenestration.15,35,36 Byanalogy, the fenestration observed in capillaries of sali-vary glands37 may also be due to locally producedVEGF. Thus, acinar cell-derived VEGF may participate

? 1998 John Wiley & Sons, Ltd.

in the regulation of saliva production by altering thepermeability of glandular capillaries and increasing thefluid supply to the secretory cells.12

In chronically inflamed glands, considerable amountsof VEGF are found in both glandular epithelium andinflammatory cells. The angiogenic effects of VEGF,

J. Pathol. 186: 186–191 (1998)

190 J. PAMMER ET AL.

probably not required under physiological conditions,may be responsible for the increased number of bloodvessels observed in sialadenitis.38

VEGF was initially isolated from tumour cell linesand has been implicated in tumour-associated neo-angiogenesis in the past.1,8,11,19,20 In salivary glandtumours, we did not observe increased expression ofVEGF protein as compared with normal surroundingglands. In addition, VEGF was expressed in both benignand malignant neoplasms. We concluded from thesedata that the elaboration of VEGF by salivary glandtumour cells is not a determining factor for their growthand invasiveness. This corresponds to our observationsin epithelial skin tumours, in which the levels of VEGFexpression were comparable for benign common wartsand invasive squamous cell carcinomas.31

In addition to VEGF in salivary glands, we alsodetected large amounts of VEGF in normal saliva.Western blot analyses confirmed that salivary VEGFprotein corresponds to VEGF derived from epithelialcells in tissue culture. Additional protein moieties of 60and 90 kD, specifically detected by the anti-VEGF MAbin the saliva of some volunteers, probably representmultimers of VEGF. Alternatively, they may result fromVEGF binding to other salivary proteins. Unfortu-nately, western blot analysis under reducing conditions,which might clarify this issue, was not possible with theMAbs used.

Besides its digestive function and its protective effectfor the mucosa of the upper digestive tract, saliva playsan important role in tissue repair and during woundhealing: wound-licking promotes the healing pro-cess,24,25 whereas sialadenectomy delays the formationof new connective tissue39–41 and the healing of oralwounds.42 Neoangiogenesis is a critical event in

? 1998 John Wiley & Sons, Ltd.

the formation of granulation tissue during tissueremodelling and repair16,17,21 and VEGF is an indispen-sable factor for the growth of new vessels.1,8 VEGF ispresent in normal saliva at biologically relevant concen-trations8,33,34 and could thus initiate immediate repairmechanisms by inducing EC migration and proliferationafter local injury. Since bFGF acts synergistically withVEGF,43,44 the small amounts of this cytokine present insaliva may potentiate the effects of salivary VEGF.

Previously, most of the effects of saliva on woundhealing have been attributed to EGF23,26 and TGF-á,27

which promote re-epithelialization of superficialwounds.42,45 As both EGF and TGF-á can induce ECchemotaxis in vitro,46 their effects may be additive withregard to EC migration.9,46 At the same time, themitogenic actions of VEGF and bFGF may be comple-mented by the morphogenetic effects of EGF/TGF-á,resulting in the efficient formation of new vessels andenhanced wound healing.

ACKNOWLEDGEMENTS

The authors JP and WW contributed equally to thisstudy and are listed in alphabetical order. We wish tothank Dr P. Bergstresser for his valuable discussion.This work was supported by a grant from the AustrianScience Foundation (grant No. P01437-MED) andthe Medizinisch Wissenschaftlichen Fonds des Bürger-meisters der Bundeshauptstadt Wien.

Fig. 3—Western blot analysis of salivary VEGF. Before western blot analysis, VEGF wasimmunoprecipitated from normal saliva (both panels lanes 3–5) or from supernatant of thekeratinocyte cell line A431 (lanes 2) with an anti-VEGF MAb as described in the Materials andMethods section. In the left-hand panel, western blot analysis was carried out with a biotinylatedanti-VEGF MAb followed by StreptABComplex/HRP. In the right-hand panel, the anti-VEGFMAb was reacted with an excess of recombinant VEGF prior to its use for western blot analysis.In lane 1 of both panels, only the anti-VEGF MAb was present

REFERENCES

1. Ferrara N, Davis-Smyth T. The biology of vascular endothelial growthfactor. Endocr Rev 1997; 18: 4–25.

J. Pathol. 186: 186–191 (1998)

191VASCULAR ENDOTHELIAL GROWTH FACTOR IN SALIVARY GLANDS AND SALIVA

2. Ferrara N, Henzel WJ. Pituitary follicular cells secrete a novel heparin-binding growth factor specific for vascular endothelial cells. BiochemBiophys Res Commun 1989; 161: 851–858.

3. Tischer E, Mitchell R, Hartman T, et al. The human gene for vascularendothelial growth factor: multiple protein forms are encoded throughalternative exon splicing. J Biol Chem 1991; 266: 11947–11954.

4. Houck KA, Ferrara N, Winer J, Cachianes G, Li B, Leung DW. Thevascular endothelial growth factor family: identification of a fourth molecu-lar species and characterization of alternative splicing of RNA. MolEndocrinol 1991; 6: 1808–1814.

5. de Vries C, Escobedo JA, Ueno H, Houck K, Ferrara N, Williams LT. Thefms-like tyrosine kinase, a receptor for vascular endothelial growth factor.Science 1992; 255: 989–991.

6. Quinn T, Peters KG, De Vries C, Ferrara N, Williams LT. Fetal liver kinase1 is a receptor for vascular endothelial growth factor and is selectivelyexpressed in vascular endothelium. Proc Natl Acad Sci USA 1993; 90:7533–7537.

7. Keck PJ, Hauser SD, Krivi G, et al. Vascular permeability factor, anendothelial cell mitogen related to PDGF. Science 1989; 146: 1309–1312.

8. Connolly DT, Heuvelman DM, Nelson R, et al. Tumor vascular per-meability factor stimulates endothelial cell growth and angiogenesis. J ClinInvest 1989; 84: 1470–1478.

9. Koch AE, Harlow LA, Haines GK, et al. Vascular endothelial growthfactor: a cytokine modulating endothelial function in rheumatoid arthritis.J Immunol 1994; 152: 4149–4156.

10. Senger DR, Galli S, Dvorak AM, Perruzzi CA, Harvey VS, Dvorak HF.Tumor cells secrete a vascular permeability factor that promotes accumu-lation of ascites fluid. Science 1983; 219: 983–985.

11. Senger DR, Connolly DT, Van De Water L, Feder J, Dvorak HF.Purification and NH2-terminal amino acid sequence of guinea pig tumorsecreted vascular permeability factor. Cancer Res 1990; 50: 1774–1778.

12. Roberts WG, Palade GE. Increased microvascular permeability andendothelial fenestration induced by vascular endothelial growth factor.J Cell Sci 1995; 108: 2369–2379.

13. Li XF, Gregory J, Ahmed A. Immunolocalization of vascular endothelialcell growth in human endometrium. Growth Factors 1994; 11: 277–282.

14. Kamat BR, Brown LF, Manseau EJ, Senger DR, Dvorak HF. Expressionof vascular permeability factor/vascular endothelial growth factor byhuman granulosa and theca lutein cells: role in corpus luteum development.Am J Pathol 1995; 146: 157–165.

15. Breier G, Albrecht U, Sterrer S, Risau W. Expression of vascular endo-thelial growth factor during embryonic angiogenesis and endothelial celldifferentiation. Development 1992; 114: 521–532.

16. Brown LF, Yeo KT, Berse B, et al. Expression of vascular permeabilityfactor (vascular endothelial growth factor) by epidermal keratinocytesduring wound healing. J Exp Med 1992; 176: 1375–1379.

17. Frank S, Hübner G, Breier G, Longaker MT, Greenhalgh DG, Werner S.Regulation of vascular endothelial growth factor expression in culturedkeratinocytes. J Biol Chem 1995; 270: 12607–12613.

18. Detmar M, Brown LF, Claffey KP, et al. Overexpression of vascularpermeability factor and its receptors in psoriasis. J Exp Med 1994; 180:1141–1146.

19. Senger DR, Brown LF, Claffey KP, Dvorak HF. Vascular permeabilityfactor, tumor angiogenesis and stroma generation. Invasion Metastasis1994–95; 14: 385–394.

20. Dvorak HF, Sioussat TM, Brown LF, et al. Distribution of vascularpermeability factor (vascular endothelial growth factor) in tumors: concen-tration in tumor blood vessels. J Exp Med 1991; 174: 1275–1278.

21. Dvorak HF, Brown LF, Detmar M, Dvorak AM. Vascular permeabilityfactor/vascular endothelial growth factor, microvascular hyperpermeability,and angiogenesis. Am J Pathol 1995; 146: 1029–1039.

22. Young JA, Schneyer CA. Composition of saliva in mammalia. Aust J ExpBiol Med Sci 1981; 59: 1–53.

23. Zelles T, Purushotham KR, Macauley SP, Oxford GE, Humphreys-BeherMG. Saliva and growth factors: the fountain of youth resides in us all.J Dent Res 1995; 74: 1826–1831.

? 1998 John Wiley & Sons, Ltd.

24. Hutson JM, Niall M, Evans D, Fowler R. Effects of salivary glands inwound contraction in mice. Nature 1979; 279: 793–795.

25. Bodner L, Knyszynski A, Adler-Kunin S, Danon D. The effect of selectivedesalivation on wound healing in mice. Exp Geronotol 1991; 26: 357–363.

26. Ino M, Ushiro K, Ino C, Yamashita T, Kumazawa T. Kinetics of epidermalgrowth factor in saliva. Acta Otolaryngol Suppl Stockholm 1993; 500:126–130.

27. Humphreys-Beher MG, Macauley SP, Chegini N, et al. Characterization ofthe synthesis and secretion of transforming growth factor alpha fromsalivary glands and saliva. Endocrinology 1994; 134: 963–970.

28. van Setten GB. Basic fibroblast growth factor in human saliva: detectionand physiological implications. Laryngoscope 1995; 105: 610–612.

29. Ellis GL, Auclair PL. Atlas of Tumor Pathology: Tumors of the SalivaryGlands. Washington, DC: Armed Forces Institute of Pathology, 1996.

30. Ballaun C, Weninger W, Uthman A, Weich H, Tschachler E. Humankeratinocytes express the three major splice forms of vascular endothelialgrowth factor. J Invest Dermatol 1995; 104: 7–10.

31. Weninger W, Uthman A, Pammer J, et al. Vascular endothelial growthfactor production in normal epidermis and in benign and malignantepithelial skin tumors. Lab Invest 1996; 75: 647–657.

32. Plettenberg A, Ballaun C, Pammer J, et al. Human melanocytes andmelanoma cells constitutively express the bcl-2 proto-oncogene in situ and incell culture. Am J Pathol 1996; 146: 651–659.

33. Conn C, Soderman DD, Schaefer MT, Wile M, Hatcher VB, Thomas KA.Purification of a glycoprotein vascular endothelial cell mitogen from a ratglioma-derived cell line. Proc Natl Acad Sci USA 1990; 87: 1323–1327.

34. Connolly DT, Olander JV, Heuvelman D, et al. Human vascular per-meability factor: isolation from U937 cells. J Biol Chem 1989; 264: 20017–20024.

35. Brown LF, Berse B, Tognazzi K, et al. Vascular permeability factor mRNAand protein expression in human kidney. Kidney Int 1992; 42: 1457–1461.

36. Kuroda M, Oka T, Oka Y, et al. Colocalization of vascular endothelialgrowth factor (vascular permeability factor) and insulin in pancreatic islandcells. J Clin Endocrinol Metab 1995; 80: 3196–3200.

37. Watanabe I, Koriyama Y, Yamada E. High-resolution scanning electronmicroscopic study of the mouse submandibular salivary gland. Acta Anat1992; 143: 59–66.

38. Cauli A, Yanni G, Pitzalis C, Challacombe S, Panayi GS. Cytokine andadhesion molecule expression in the minor salivary glands of patients withSjoegren’s syndrome and chronic sialoadenitis. Ann Rheum Dis 1995; 54:209–215.

39. Bodner L, Dayan D, Rothchild D, Hammel I. Extraction wound healing indesalivated rats. J Oral Pathol Med 1991; 20: 176–178.

40. Bodner L, Dayan D, Oberman M, Hirshberg A, Tal H. Healing ofexperimental wounds in sialadenectomized rats. J Clin Periodontol 1992; 19:345–347.

41. Dayan D, Bodner L, Horowitz I. Effect of salivary gland hypofunction onthe healing of extraction wounds: a histomorphometric study in rats. J OralMaxillofac Surg 1992; 50: 354–358.

42. Noguchi S, Ohba Y, Oka T. Effect of salivary epidermal growth factor onwound healing of tongue in mice. Am J Physiol 1991; 260: E620–E625.

43. Goto F, Goto K, Weindel K, Folkman J. Synergistic effects of vascularendothelial growth factor and basic fibroblast growth factor on the prolifer-ation and cord formation of bovine capillary endothelial cells withincollagen gels. Lab Invest 1993; 69: 508–517.

44. Pepper MS, Ferrara N, Orci L, Montesano R. Potent synergism betweenvascular endothelial growth factor and basic fibroblast growth factor in theinduction of angiogenesis in vitro. Biochem Biophys Res Commun 1992; 189:824–831.

45. Niall M, Ryan GB, O’Brien BM. The effect of epidermal growth factor onwound healing in mice. J Surg Res 1982; 33: 164–169.

46. Mawatari M, Kohno K, Mizoguchi H, et al. Effects of tumor necrosis factorand epidermal growth factor on cell morphology, cell surface receptors, andthe production of tissue inhibitor of metalloproteinases and IL-6 in humanmicrovascular endothelial cells. J Immunol 1989; 143: 1619–1627.

J. Pathol. 186: 186–191 (1998)