Human Reproduction vol.12 no. 11 pp.2451-2456, 1997

Human tubal fluid: production, nutrient compositionand response to adrenergic agents

J.I.Tay1'4, A.J.Rutherford\ S.R.Killick2,S.D.Maguiness2, R.J.Partridge3 and H.J.Leese3

Assisted Conception Unit, Clarendon Wing, Leeds GeneralInfirmary, Belmont Grove, Leeds LS2 9DN, 2Department ofObstetrics and Gynaecology, Princess Royal Hospital,Saltshouse Road, Hull HU8 9HE, and department of Biology,University of York, PO Box 373, York YO1 5YW, UK4To whom correspondence should be addressed at: St James'sUniversity Hospital, Department of Obstetrics and Gynaecology,Gledhow Wing, Beckett Street, Leeds LS9 7TF, UK

Vascularly perfused Fallopian tubes have been used tostudy the formation and composition of human tubal fluidand the response to adrenergic agents. An artery servingthe tube was cannulated and perfused with Medium 199supplemented with bovine serum albumin (BSA) and anti-biotics. A second cannula was attached to the fimbriatedend for native tubal fluid collection. The preparation wasviable for up to 2 h. Tubal fluid was only obtained in tubesremoved in the proliferative and early secretory phases ofthe ovarian cycle. Isoproterenol (1 mM) added to theperfusate stimulated fluid production, whereas dibutyrylcyclic AMP (1 mM) reduced fluid formation by 66%.Glucose, pyruvate and lactate concentrations in tubal fluid,measured by microfluorescence assays, were 1.11, 0.14 and5.4 mM respectively. The concentrations of 17 amino acidsin tubal fluid were measured by high performance liquidchromatography following fluorescence derivatization.Arginine (0.19 mM) > alanine (0.11 mM) > glutamate(0.09 mM) were present in highest concentration in allphases of the cycle. All 17 amino acid concentrations intubal fluid were below those in the vascular perfusate.These data provides the basis for a culture medium whosecomposition mimics the physiological environment to whichearly human embryos are exposed.Key words: human Fallopian tube/tubal fluid/vascular perfusion

Introduction

Gamete transport, maturation, fertilization and early embryodevelopment occur within the lumen of the Fallopian tube, yetthe composition of tubal fluid, its mechanism of productionand how it affects gamete and embryo survival are poorlyunderstood. Culture media used in in-vitro fertilization (IVF)programmes have attempted to mimic this environment, butwith limited success, as shown by the high failure rates ofeven the best IVF centres. Knowledge of tubal fluid formationin health and disease may help to reduce the extent of tubaldamage during pelvic inflammatory disease and, in turn,

decrease the incidence of tubal defects which currently accountfor up to 28% of female infertility (Hull et al., 1985).

Human tubal fluid constituents have been measured byLippes et al. (1972), Moghissi (1970), David et al. (1973),Shams et al. (1977), Borland et al. (1980), Lippes et al. (1981)and Gardner et al. (1996). Different techniques have been usedbut have essentially involved either sampling tubal fluid at thetime of surgery or chronic collection over 2-8 days, duringwhich time the women were kept in hospital. In some cases,it is doubtful if the fluids collected were physiological andnone of the methods enabled the mechanisms underlying tubalfluid formation to be studied.

One approach to resolving these problems involves thetechnique of vascular perfusion, devised for rabbit oviductsby Leese and Gray (1985) and used for other reproductiveorgans such as the ovary (Kobayashi et al, 1980) and placenta(Page, 1991). In this method, an artery serving the tube iscannulated and the vasculature perfused with Medium 199supplemented with bovine serum albumin and antibiotics. Thelumen may be cannulated at the same time, allowing thecollection of native oviduct fluid. Using this preparation, Gottet al. (1988) showed that the rate of formation of rabbit oviductfluid was influenced by the reproductive state of the animaland that fluid production was abolished by the addition ofdibutyryl cyclic AMP (db c-AMP) to the vascular medium.This finding was confirmed by Dickens and Leese (1994),who also found that the adrenergic agonist isoprenaline, addedto the vascular medium, stimulated tubal fluid formation, aneffect abolished by propranolol.

The vascular perfusion technique was applied to humanFallopian tubes by Dickens et al. (1995). Tubes (n = 19) wereobtained from women attending for hysterectomy and shownto remain viable for up to 90 min of perfusion. Native tubalfluid appeared in the lumen cannula in tubes from patients inthe follicular phase of their ovarian cycle. We have now usedthis technique with a larger series of patients in = 38) toinvestigate further aspects of the formation and compositionof human tubal fluid, specifically, the rate of fluid formationin the proliferative and secretory phases, the response tovascular isoproterenol and dibutyryl cyclic AMP and thecomposition of the fluid in terms of pyruvate, glucose, lactateand 17 amino acids.

Materials and methodsFallopian tubes were obtained from women attending for abdominalhysterectomy (n = 38, mean age 39.6 years ± SEM 1.08). Writteninformed consent was obtained. Blood samples were obtained pre-operatively on the day of surgery and serum concentrations of

European Society for Human Reproduction and Embryology 2451

J.I.Tay et al.

Table I. Patient details and concentrations of glucose, pyruvate and lactate in tubal fluid (n = 38)

Name Age(years)

Cycle day Menstrualphase

Reason forhysterectomy

Cannulationtime (min)

Glucose(mM)

Pyruvate(mM)

Lactate(mM)

123456789

1011121314151617181920212223242526272829303132333435363738

3443414039474835463138493237284930364534343649375240443749364848353037443532

363

17Amenn

93454

12977

171314284

1010141917317

141418142339492624152456

Mean

SEM

39

1.1

BasalBasalBasalProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeEarly secretoryEarly secretoryEarly secretorySecretorySecretorySecretoryMid-secretoryMid-secretoryMid-secretoryLate secretoryLate secretoryLate secretoryLate secretoryLate secretoryLate secretoryLate secretoryLate secretory

EndometnosisMenorrhagiaMenorrhagiaMenorrhagiaEndometriosisMenorrhagiaMenorrhagiaMenorrhagiaFibroidsSevere PMSMenorrhagiaMenorrhagiaMenorrhagiaEndometriosisPelvic painFibroidsMenorrhagiaDysmenorrhoeaMenorrhagiaEndometriosisMenorrhagiaMenorrhagiaMenorrhagiaMenorrhagiaFibroidsMenorrhagiaMenorrhagiaFibroidsFibroidsEndometriosisSevere PMSFibroidsMenorrhagiaEndometriosisEndometriosisMenorrhagiaEndometriosisMenorrhagia

121196

1777

147

1027241930187

13122511119

2288

1515766

1532121713142614

14.11

1.15

0.831.561.70.421.31.20.24-0.92------0.58-0.8high2.47-0.81-0.26-0.521.88-0.72_1.941.3-_1.021.76--

1.11

0.14

0.070.1980.510.060.10.050.01-0.21------0.134-0.120.280.09-0.04-0.203-0.40.15-0.16-0.020.12--0.90.3--

0.14

0.04

73.568.96.28.62.76

-4.6

------

4.76-

4.414.24.78

-7.2

-10.03-

8.81.6

-8.4

-8.65.6

--

9.28.8

--

5.37

0.53

oestrogen, progesterone, follicle stimulating hormone (FSH) andluteinizing hormone (LH) measured. The cycle day was calculatedfrom the first day of menstrual bleeding as recalled by the patient,and phase of the cycle was confirmed by histological dating of theendometrium from the excised uterus (Noyes et al, 1950).

The tubal perfusion technique and apparatus were similar to thatdescribed by Dickens et al. (1995). The perfusion medium, consistingof Medium 199, 4% dialysed bovine serum albumin, 4 IU/ml heparinand 50 mg/ml each of penicillin and streptomycin, contained in awater-jacketed vessel maintained at 37°C, was recirculated at a rateof 3 ml/min and oxygenated with 95% 0^5% CO2. At operation, anartery supplying the Fallopian tube was identified and clamped priorto removal of the tube. The clamp served to identify the artery, whichwas then isolated and a small glass cannula (inner diameter 0.2 mm,outer diameter 0.3 mm) tied into the artery with fine silk ligatures.Perfusion medium was immediately introduced through the cannulavia a peristaltic pump and a pressure gauge at a rate of 1 ml/min.The mean time from excision of the tube to commencement of theperfusion was 14.11 min (± 1.15 SEM, n = 38). Successfulcannulation of the artery caused blanching of the whole tube whenperfusion was started. Venous drainage from the tube was allowed torun to waste. The cannulated tube was placed on a water-jacketed

glass dish, the uterine end tied off and a wide-bore collecting cannula(inner diameter 1.6 mm, outer diameter 2.8 mm) secured into thefimbrial end with silk ligatures. The fimbrial cannula was in turnattached to a calibrated tube so that the appearance of tubal fluidcould be monitored. In experiments where no fluid appeared in thecalibrated tube, an SMI pipette was used to sample the fluid directlyfrom within the tubal lumen.

In experiments involving the addition of adrenergic agents, theperfused tubes were allowed to equilibrate for 30 min prior to theaddition of either isoproterenol (1 mM) or 2'-0-dibutyryladenosine3'5' cyclic monophosphate (db c-AMP) (1 mM) to the vascularperfusate. Isoproterenol was added (n = 19) if, during the 30 minequilibration interval, there was no appearance of tubal fluid, or ifthe amount was too small to allow a flow rate to be measured. Dbc-AMP was added to four perfusions in which a flow was measurable.At the end of the perfusion time (90-120 min), tubal fluid wascarefully removed and stored at -70°C prior to analysis. Tubal fluidswere analysed for glucose, pyruvate and lactate using ultramicro-fluorometric techniques described by Leese and Barton (1984) andGardner and Leese (1990). The assays were performed in nanolitre-sized drops on siliconized microscope slides under mineral oil. Aminoacid concentrations were measured as described by Lamb and Leese

2452

Human tubal fluid

70III! mi mi mi mi mi II

Figure 1. Light microscopy appearance of post-perfused tubalepithelium.

Figure 2. Scanning electron microscopy appearance ofpost-perfused tubal epithelium.

Cannulatlon time vs lactate concentration

10 15Time(min)

25

Figure 3. Cannulation time (min) and tubal fluid lactateconcentration (mM). Cannulation time versus lactate. x-axis: time(min); y-axis: lactate concentration (mM).

Table II. Fluid flow pattern of perfusions with detectable rates of fluidproduction (n = 17). ISO = isoprenaline

Patient Age(years)

Phase Addition Flow before(ml/h)

Flow after(ml/h)

28

172021101116479

121423222425

4335303434313849404846493749363752

BasalProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeEarly secretoryEarly secretoryEarly secretorySecretory

nononononoisoisoisoisodb c-AMPdb c-AMPdb c-AMPdb c-AMPisoisoisoiso

29.236.9151022.1

nonono

44.783.95

119.93136.8149.19nononono

98.436.0823.7639.669.7720.7441.1935.397.92

32.5613.253.46

(1994). Samples were reacted with o-phthaldialdehyde (ODA) toform fluorescent amino acid derivatives which were detected by highperformance liquid chromatography. Samples of endosalpinx fromperfused tubes were removed and placed in fixative (4% formaldehyde,2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.2) for 1.5 h atroom temperature. These samples were used for scanning electronmicroscopy (SEM) and compared with similar endosalpinx samplestaken from the contralateral, non-perfused tubes which acted ascontrols. Light histological studies of the perfused tubes were alsoperformed following fixation in 1% formaldehyde and embedding inparaffin wax prior to sectioning.

Results are expressed as means ± SEM. Differences betweenmeans were compared by Mests. Correlation was calculated usingPearson's r.

Results

A total of 38 experiments was performed. Table I shows thepatients' details, reasons for their hysterectomies and theconcentrations of glucose, pyruvate and lactate in tubal fluidobtained. Perfusion pressures remained stable for at least 2 hand perfused and non-perfused tissue had a similar appearanceunder light and scanning electron microscopy (Figures 1 and2). There was no relationship (r = -0.06) between the timetaken to cannulate the tubes and the lactate concentration inthe tubal fluid formed (Figure 3).

A total of 17 perfusions yielded sufficient fluid to allowcalculation of a rate of fluid production (Table II). In 10 ofthese 17 (patients 2, 4, 7, 8, 9, 12, 14, 17, 20 and 21),fluid was produced 'spontaneously', i.e. without the use ofisoproterenol. Mean rate of fluid production was 64.78 ml/h.Nine of these 10 patients were in the proliferative phase; the10th (patient 2) was in the basal (post-menstrual) phase of thecycle. The addition of isoproterenol to a further seven perfu-sions led to the appearance of a measurable rate of fluidproduction (Table II). Db c-AMP was added to four perfusions(Table II) (7,9,12 and 14). Unlike the addition of isoproterenol,this led to a decrease in fluid production. Mean rates of fluidflow prior to, and after addition of, db c-AMP were 122.47 ±

2453

J.I.Tay et al.

Table IV. Mean

Amino acid

ASPGLUASNSERGLNARGGLYTHRALATYRTRPMETVALPHEILELEULYS

Table III. Flow pattern after addition of isoprenaline

Patient

156

151011164

2523222428293031333 5 •

37

Mean

SEM

Age(years)

34394728313849404936375237493648353735

40

1.6

Menstrual phase

BasalProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeProliferativeEarly secretoryEarly secretoryEarly secretorySecretoryMid-secretoryMid-secretoryMid-secretoryLate secretoryLate secretoryLate secretoryLate secretory

amino acid concentrations in proliferative (n = 10)

Mean proliferative

0.017'0.0710.0630.0260.0240.1510.0320.0440.0840.0240.0120.0120.0220.0160.0170.050.042

SEM

0.0040.0190.0530.0060.0040.0320.0090.0080.0170.0040.0060.0040.0060.0050.0050.0140.009

Time beforeisoprenalineadded

32314560303160462035473230316346906060

44.68

3.96

and secretory (n =

Mean secretory

0.0410.1110.0280.0380.0510.2370.0370.0470.1390.0490.0180.0140.0320.0180.020.0670.061

Flow before(\i\Jh)

nonononononono44.7nonononononononononono

Flow after(Hl/h)

nononono98.436.0823.7639.67.92

32.5613.253.46nonononononono

38.12

10.04

7) phase human tubal fluid (mM)

SEM

0.0180.0330.0250.0080.0140.0570.0080.0080.0520.0230.010.0030.0050.0030.0020.0130.016

Overall mean

0.0270.0910.0460.0320.0380.1940.0350.0460.1120.0370.0150.0130.0260.0170.0190.0590.052

Perfusiontime

909060

1059090

100909090

105909090

10090

12090

105

93.42

2.68

SEM

0.0080.0190.0180.0060.0140.0430.0030.0020.0280.0130.0030.0010.0040.0010.0020.0090.01

P value

0.0970.2750.4390.2410.050.1780.7150.7920.2560.2460.5680.6620.2690.7140.5680.3980.261

14.17 ml/h and 41.77 ± 10.28 ml/h respectively (P < 0.005).In the remaining seven perfusions, tubal fluid was obtainedafter the addition of isoproterenol. However, the addition ofisoproterenol to one perfusion (patient 4) caused a reductionin fluid production. Six of these tubes were in the proliferativeor early secretory phases, with one (patient 25) in the secretoryphase (day 31).

Of the total of 38 experiments, isoproterenol was added to19 and the pattern of fluid production is shown in Table III.No tubal fluid was obtained from tubes in the mid- to late-secretory phases of the cycle. The mean rate of flow oftubal fluid after the addition of isoprenaline was 38.12 ±10.04 ml/h.

Seventeen amino acids were identified in native tubal fluid.Table IV shows the mean amino acid concentrations of 10tubal fluid samples from the proliferative phase, seven from

2454

the secretory phase and their overall means. In both theproliferative and secretory phases, arginine was present athighest concentration, 0.15 ± 0.03 mM and 0.24 ± 0.06 mMrespectively. Alanine and glutamate were present at the secondand third greatest concentrations respectively. All the aminoacids, except asparagine (ASN), were present at higher concen-tration in the secretory phase. The differences in concentrationbetween the two phases, however, did not reach statisticalsignificance except for glycine (P = 0.05). Overall meanconcentrations of arginine, alanine and glutamate were 0.19,0.11 and 0.09 mM respectively.

The mean concentrations of glucose, pyruvate and lactatewere 1.11 ± 0.14, 0.14 ± 0.04 and 5.37 ± 0.53 mMrespectively (Table I). The differences in concentration of thethree energy substrates between the proliferative and secretoryphases were not significant.

Human tubal fluid

Discussion

We have presented further evidence that the technique of in-vitro vascular perfusion provides a means of studying humanFallopian tubes in a physiological manner and of defining theenvironment to which the gametes and early embryo areexposed. On the basis of the maintenance of a steady perfusionpressure, linear rates of tubal fluid secretion and structuralappearance post-perfusion, tubes were viable for up to 2 h. Thisconclusion was reinforced by the absence of any relationshipbetween the time taken to cannulate the tubal artery and thesubsequent concentration of lactate (a possible marker of tubalanoxia) in the fluid formed (Figure 3). Luminal fluid appearedin tubes removed during the proliferative and early secretoryphases, in agreement with the findings of Dickens et al. (1995).Isoproterenol added to the vascular perfusate stimulated tubalfluid production while db c-AMP had an inhibitory effect;both responses being similar to those reported for rabbitoviducts (Gott et al., 1988; Dickens and Leese, 1994). Asregards their biochemical significance, these findings remainparadoxical, since isoproterenol is considered to act via thegeneration of c-AMP. We suggest that c-AMP generated inresponse to isoproterenol occupies a different compartment inthe tissue than that added exogenously. However, not all thetubes in the proliferative phase responded to isoproterenol andin the case of one patient, there was a slight decrease in tubalfluid production. This could be that as the tube was alreadyproducing fluid, possibly at its maximum rate, it could not bestimulated to increase its production further. A low, continuouslevel of fluid secretion is likely to be essential in maintainingthe endosalpinx in a viable state and ensuring tubal patency.This fluid output could, in part, be mediated via the sympatheticnervous system, as is the case for the intestine (Cooke,1994) and airways (Barnes, 1992). If the secretory flow iscompromised, it is possible that cells/cell debris derived fromsperm, bacteria, phagocytes, macrophages etc. would not becleared from the lumen and could predispose the tube tobecome blocked.

The discrepancies between cycle day and histological datingof the endometrium may be due to the patient's incorrect recallof the last menstrual period or cycles which are longer orshorter than the normal 28 days. The longer cycle days wereall in the late secretory phase.

The concentration of glucose in native tubal fluid (1.1 mM)was higher than that reported by Dickens et al. (1995)(0.5 mM); pyruvate concentrations in the two studies weresimilar (0.14 and 0.17 mM), while lactate concentration waslower in the present study (5.4 mM versus 8.58 mM). Gardneret al. (1996) sampled the human oviduct and uterine lumen insitu and found that the concentrations of these three nutrients,while generally of the same order as those reported here,varied with the state of the cycle. Of most interest was a fallin tubal lumen glucose concentration, from 3.1 mM in thefollicular phase, to 0.5 mM mid-cycle. Taken together, thesestudies indicate that while the concentration of pyruvate presentin most human embryo culture media (~0.3 mM) is reasonablyphysiological, the concentrations of glucose and lactate usedby many (2.78-5.55 mM and >20 mM respectively) are almost

Amino acid ratios, M199/tubal fluid

ASP GLU SER GLN ARG GLY THR ALA TYR MET VAL PHE ILE LEU LYS0.23 0.45 0.24 0.68 0.33 0.37 0.25 0.28 0.22 0.1 0.21 0.15 0.15 0.46 0.38

Amino acid (mM)

Figure 4. Amino acid ratios, tubal fluid/Ml99. PROLIF =proliferative phase; SECRE = secretory phase; Ml99 amino acidconcentrations shown below category, x-axis: amino acid (mM);y-axis: ratios.

certainly non-physiological. All the individual amino acidvalues were below their concentrations in Medium 199 usedas vascular perfusate. This indicates that the overall movementof amino acids from the vasculature into the lumen is by adiffusive mechanism, most likely, carrier-mediated, ratherthan by active transport. On the basis of the lumen/vascularconcentration ratios, the amino acids transported to the greatestextent were: arginine > alanine > glutamate (Figure 4). Theratio of lumen/vascular arginine concentration was 0.59 andthat for alanine and glutamate was 0.4 and 0.2 respectively.The presence of amino acids at higher concentrations in thesecretory phase suggests that the tubal epithelium is morepermeable to these constituents at a time when they are likelyto be required for the nutrition of the early embryo while it isin the tubal lumen.

Apart from contributing to our knowledge of human tubalfluid formation and composition, the data could providethe basis for a novel culture medium whose compositionapproximates that to which the gametes and early embryos areexposed within the Fallopian tube.

AcknowledgementsThe authors acknowledge support from the Medical Research Counciland Schering Health Care. We thank M.Stark for carrying out thescanning electron microscopy and Dr A.Andrew for performing thelight microscopy studies.

ReferencesBarnes, P.J. (1992) Asthma. Br. Med. Bull., 48, 231-247.Borland, R.M., Biggers, J.D., Lechene, C.P. et al. (1980) Elemental composition

of fluid in the human Fallopian tube. J. Reprod. Fertil., 58, 479^82.Cooke, H.J. (1994) Neuroimmune signalling in regulation of intestinal ion

transport. Am. J. Physiol., 266, G167-G178.David, A., Serr, D.M. and Czenobilsky, B. (1973) Chemical composition of

human oviduct fluid. Fertil. Steril, 24, 435^39.Dickens, C.J. and Leese, H.J. (1994) The regulation of rabbit oviduct fluid

formation. J. Reprod. Fertil, 100, 577-581.

2455

J.I.Tay et al.

Dickens, C.J., Maguiness, S.D., Comer, M.T. et al. (1995) Human tubal fluid:formation and composition during vascular perfusion of the Fallopian tube.Hum. Reprod., 10, 505-508.

Gardner, D.K. and Leese, HJ. (1990) Concentration of nutrients in mouseoviduct fluid and their effects in embryo development and metabolismin vitro. J. Reprod. Fertii, 88, 361-368.

Gardner, D.K., Lane, M., Calderon, I. et al. (1996) Environment of thepreimplantation human embryo in vivo: metabolite analysis of oviduct anduterine fluids and metabolism of cumulus cells. Fertii. SteriL, 65, 349-353.

Gott, A.L., Gray, S.M. James, A.F. et al. (1988) The mechanism and controlof rabbit oviduct fluid formation. Biol. Reprod., 39, 758-763.

Hull, M.R.G., Glazener, C.M.A., Kelly, J. et al. (1985) Population study ofcauses, treatment and outcome of infertility. Br. Med. J., 291, 1693-1697.

Kobayashi, Y., Santulli, R., Wright, K.H. et al. (1980) The effect ofprostaglandin synthesis inhibition by indomethacin on ovulation and ovummaturation in the in-vitro perfused rabbit ovary. Am. J. Obstet. Gynaecol.,141, 53-57.

Lamb, V.K. and Leese, HJ. (1994) Uptake of a mixture of amino acids bymouse blastocysts. J. Reprod. Fertii., 102, 169-175.

Leese, HJ. and Barton, A.M. (1984) Pyruvate and glucose uptake by mouseova and preimplantation embryos. J. Reprod. Fertii., 72, 9-13.

Leese, HJ. and Gray, S.M. (1985) Vascular perfusion: a novel means ofstudying oviduct function. Am. J. Physiol., 248, E624-E632.

Lippes, J., Enders, R.G., Pragay, D.A. etal. (1972) The collection and analysisof human Fallopian tubal fluid. Contraception, 5, 85-103.

Lippes, J., Krasner, J., Alfonso, L.A. et al. (1981) Human oviductal fluidproteins. Fertii. SteriL, 36, 623-629.

Moghissi, K.S. (1970) Human Fallopian tubal fluid. I. Protein composition.Fertii. SteriL, 21, 821-829.

Noyes, R.N., Hertig, A.T. and Rock, J. (1950) Dating the endometrial biopsy.Fertii. SteriL, 1,3-11.

Page, K.R. (1991) Perfusion of isolated human placenta. Proc. Nutr. Soc, 50,345-347.

Shams, A., Rizk, M.A., Toppozada, H.K. et al. (1977) Human tubal fluidcollection via vagina and its quantitative variations during the menstrualcycle. J. Reprod. Med., 18, 61-65.

Received on March 11, 1997; accepted on August 12, 1997

2456