Placenta (1989), IO, 263-273

Stereological Re-examination of the Effects of Varying Oxygen Tensions on Human Placental Villi Maintained in Organ Culture for up to 12 h

G. J. BURTONa’C, T. M. MAYHEWb & LESLEY A. ROBERTSONb

a Department of Anatomy, University of Cambridge, Downing Street, Cambridge CB2 3D Y, UK

b Department of Anatomy, Marischal College, Universi<y of Aberdeen, Aberdeen ABg I AS, UK

’ To whom correspondence should be addressed

Paper accepted 15.12~988

INTRODUCTION

Two decades ago, Tominaga and Page (1966) reported that acute hypoxia in vitro leads to re- versible thinning of the intervascular barrier of human placentae at term. Thinning occurred as early as 6 h of culture in 6 per cent oxygen and was accompanied by aggregation of nuclei within the syncytiotrophoblast to form clumps or ‘knots’. Both phenomena could be reversed equally rapidly by transferring to an atmosphere containing 26 per cent oxygen. In addition, a reduction in the peripheral resistance offered by the fetal vessels was observed during artificial perfusion with a hypoxic medium (Tominaga and Page, 1966). This response was attributed to vasodilation of the tertiary vessels within villi.

A more recent and rigorous morphometric study (Amaladoss and Burton, 1985) failed to reproduce these findings and the authors concluded that placental villi show no adaptations to hypoxia during in vitro organ culture. Apart from the obvious contradiction in these studies, several other factors have prompted us to investigate this problem afresh.

First, Tominaga and Page (1966) gave no indication as to the number of organs examined, whilst Amaladoss and Burton (1985) employed only three, a rather small number. Second, neither group estimated effective diffusion distances, namely, harmonic mean thicknesses (see Lega, Driscoll and Munro, 1973; Jackson et al, 1985). Third, there is increasing evidence that the thickness of the villous membrane in vivo is related to the prevailing oxygen tension within the intervillous vascular space (Teasdale, 1978; Critchley and Burton, 1987). Harmonic mean thickness is reduced significantly in the more hypoxic regions of a lobule and this cannot be explained simply by vasodilation since mean capillary diameters do not vary. Fourth, stereo- logical studies have shown that thinning of the villous membrane occurs during pregnancy at high altitude where, again, there is no dilation of fetal capillaries and no expansion of their sur- face area (Jackson, Mayhew and Haas, r987a, 1988a,b). Indeed, exchange surface areas are reduced in highland placentae owing to impoverished villous growth (Jackson, Mayhew and

0143~-4004/89/030263 + I I $05.00/o CJ 1989 BailliPre Tindall Ltd

264 Placenta (1989), Vol. IO

Haas, r987a,b) and so the possibility arises that thinning may represent a compensatory re- sponse to loss of exchange surface rather than to hypoxia directly. Finally, there is an inverse relationship between the incidence of vasculosynctial membranes and the occurrence of fetal distress or neonatal asphyxia (Fox, 1967).

In the present study, we attempt to overcome some of the practical deficiencies of earlier in- vestigations by properly randomised sampling of adequate numbers of organs and by estimat- ing the harmonic mean thickness of both the trophoblast and the villous membrane. Hypoxic and hyperoxic organ cultures were examined in order to resolve the effects of oxygen tension and time.

MATERIALS AND METHODS

Placental material and culture technique Placentae were obtained from ten normal singleton pregnancies: one spontaneous vaginal de- livery and nine caesarian sections for breech presentation or cephalo-pelvic disproportion. All mothers were under 35 years of age and had no complicating pathology such as hypertension or diabetes. Gestational ages were between 38 and 42 weeks and there were no cases of intra-uter- ine growth retardation.

Using aseptic technique, approximately I cm3 blocks of tissue were removed from the basal surface of each placenta at three equidistant sites around the disc and midway between peri- phery and centre. Areas of damage caused by tearing at delivery were avoided but, otherwise, the selection of tissue was random around the disc. Subsequently, the basal plate was trimmed away from each block and a slice of tissue 1-2 mm thick was cut parallel to the plane of the basal plate. After trimming further to roughly 5 mm x 5 mm, each slice was diced with a razor blade into a 2 x 3 matrix of cubes. Five of these tissue cubes from each placenta were then assigned to different experimental groups, by lottery, for culturing or to serve as a control (zero time in culture).

This procedure was repeated for the remaining two blocks of tissue and so each organ was represented in each experimental group by three cubes of tissue. This arbitrary scheme is satis- factory for the internal comparisons drawn in the present study but would not qualify as a random sampling scheme for estimating the morphometric composition of the placenta as a whole (see Mayhew and Burton, 1988).

The cubes of tissue were spaced around a coarse Milipore pre-filter supported on a stainless steel grid in a 35 mm Petri dish. Culture medium, pre-equilibrated with the appropriate gas mixture for I 2 h, was brought half-way up the filter so that the tissue was lying at the gas/liquid interface. Modified Medium 199 containing Earle’s Salts was used and to this was added 20 per cent neonatal calf serum, glutamine (292 mg/l), insulin (50 i.u./l) and hydrocortisone (IOO pg/l). The pH was adjusted to 7.4 and osmolality to 290 mOsm/kg. Dishes were stacked in two Fildnes-Mackintosh jars, gassed for IO min with the appropriate mixture and incubated at 37°C. Two gas mixtures were employed (BOC Special Gases): 6 per cent oxygen and 40 per cent oxygen. Both contained 5 per cent carbon dioxide with a balance of nitrogen.

Tissues were removed from each jar at 6 h and I 2 h for fixation. Immediately prior to open- ing the jars, oxygen levels within them were tested using a Beckman OM-14 analyser. In no case did the oxygen level rise above 8 per cent (hypoxic culture) or fall below 35 per cent (hyperoxic culture).

Burton, Mayhem, Robertson: Placental villi in organ culture

x-axis

mr a_

Figure I. Systematic sampling of tissue sections. The diagram depicts a slice of tissue on a microslide. For clarity, the glass coverslip is not shown. The broken lines tangent to the top and left extremities of the tissue define the position of a grid of nine squares. One square (here number four) is picked at random and provides the start (indicated by star) for systematically scanning the tissue in X- and y-directions of the microscope specimen stage. The path of the scan is marked by the arrows. In this example, two microscopical fields of view can be recorded for stereological analysis.

Tissue fixation and preparation Tissue cubes were fixed by immersion for 4 h in a mixture of z per cent glutaraldehyde/z per cent paraformaldehyde in Pipes buffer (total osmolality I 150 mOsm/kg; vehicle osmolality 310 mOsm/kg and pH 7.2). Following post-fixation in I per cent osmium tetroxide in 0.1 M

Pipes buffer for 4 h, tissue was dehydrated by passage through ascending concentrations of alcohol, briefly transferred to acetone and embedded in Taab resin. It has been shown that this protocol results in a shrinkage rate of about 2 per cent, as judged by the sizes of maternal erythrocytes (Burton and Palmer, 1988).

Sections of about I pm thick were cut on an ultramicrotome and stained with a I per cent solution of methylene blue/Azure II.

Microscopical sampling procedures Light microscopical fields of view were recorded from each tissue slice in a systematic fashion. The procedure is illustrated in Figure I. Briefly, tangents to the top and left extremities of each slice of tissue were drawn on microslides in order to define a 3 x 3 lattice of numbered squares. One square of this lattice was selected at random (by lottery) and provided the start-point (square centre) for scanning the whole tissue slice in a systematic pattern using the x,y axes of the micrometers on a microscope stage. Fields were recorded if they contained villous tissue. Roughly three fields were obtained, on average, from each of the three tissue slices per organ.

Micrographs were printed to final magnifications of between X 675 and X 850 calibrated with the aid of micrometer scale standards photographed at the same initial magnification as tissue fields of view.

Stereological estimations Micrographs were coded and analysed blind using transparent overlays of parallel straight test

a66 Placenta (1989), Vol. IO

lines spaced at 2 cm intervals (for thickness estimations) or of test points arranged in a quad- ratic grid of spacing 2 cm (for villous composition and size).

When superimposed on micrographs so as to be random and independent in position and orientation, the test lines provided the basis for sampling orthogonal intercepts across the intervascular barrier. The lengths of 70-180 orthogonal intercepts across the trophoblast or across the villous membrane (outer surface of trophoblast to inner surface of fetal capillary endothelium) were measured for each organ. Stem villi were excluded from measurements.

Using this method, the arithmetic mean intercept length, multiplied by n/4, provides an estimate of the arithmetic mean thickness of the trophoblast (Tat). In addition, the harmonic mean intercept length, multiplied by 8/3n, gives an estimate of the harmonic mean thickness of trophoblast (Tht) which exhibits minimal statistical bias and minimal variance (Jensen, Gundersen and 0sterby, 1979; Hirose et al, 1982). The ratio It = Tat/Tht provides a conveni- ent index of the variability of trophoblast thickness, the ratio decreasing as the trophoblast becomes more uniformly thick (see Jackson et al, 1985).

Identical relationships were employed to calculate arithmetic means (Tavm), harmonic means (Thvm) and thickness ratios (Ivm = Tavm/Thvm) from orthogonal intercept lengths across the villous membrane.

The volumetric composition of villi (expressed as fractional volumes of trophoblast, Vv(tr), stroma, Vv(st), and fetal capillaries, Vv(fc) within terminal villi) was determined by test point counting (Weibel, 1979). We decided not to estimate volume densities of villi in placentae because, in organ culture, these would be influenced by loss of residual blood from the inter- villous space. Instead, we chose to express the incidence of degenerating villi in each organ as a fraction of the volume of all terminal villi, Vv(dv). Finally, we estimated the mean diameter, D(v), of villi (Jackson et al., 1987~).

Statistical analyses Individual values for each placenta were used to compute the mean and coefficient of variation (c.v. = standard deviation/group mean) for each group of organs. To test for the presence or otherwise of significant differences between all five groups, we used the non-parametric Pages ‘L’ trends test for related samples (Miller, 1975). Two-way analyses of variance (Sokal and Rohlf, 1981) were employed in an attempt to resolve main and interaction effects due to oxygen tension and time in culture. We excluded the control group from these variance analyses.

RESULTS

Qualitative observations Control villi displayed intense basophilia of the syncytiotrophoblast, an inconstant cytotropho- blast, a compact stromal core and patent fetal vessels. The principal differences seen in culture, whether hypoxic or hyperoxic, were: (a) reduced syncytial basophilia, (b) more prominent trophoblastic surface processes, presumably microvilli, (c) appearance of intratrophoblastic vacuoles, (d) increased incidence of degenerating villi, and (e) appearance of subtrophoblastic dilations. Because of reduced basophilia and increased vacuolation of cytoplasm within the synctium, nuclei were more conspicuous in cultured villi.

Morphological differences between hypoxic and hyperoxic groups were also noticed. Surface processes, stromal oedema and Langhan’s cells, often with one or two nucleoli per nuclear section, were more prevalent in villi from hyperoxic versus hypoxic cultures.

Burton. Mqhem, Robertson: Placental villi in organ culture 267

Table I, Estimators of trophoblast and villous membrane thickness in placentae incubated under different conditions. Values are group means (cv.) for ten placentae.

Thickness bariablcs

Controls zero time

Tat, pm ‘l‘ht, /em It I‘avm, pm ‘Thvm, nm Ivm

3.40 (0.20) 2.35 (0.28) 1.48 (0.13) 7.42 (0.18) 5.50 (0.25) I.39 (0.15)

Hypoxia (6 per cent)

bh 12 h

3.23 (0.14) 2.94 (0.22)

2.37 (0.17) 2.23 (0.26) I.37 (0.10) 1.33 (0.06) 7.24 (0.17) 6.31 (0.21) 5.77 (0.15) 5.19 (0.22) I.26 (0.07) I .22 (0.07)

Hyperoxia (40 per cent)

bh 12 h

3.66 (0.17) 4.08 (0.20)

2.75 (0.15) 3.25 (0.29) ‘.33 (0.10) I .29 (0.09) 9.06 (0.14) 8.93 (0.13) 6.83 0.13) 7.51 (0.18) 1.33 (0.10) I .20 (0.07)

Table L. Composition, mean diameter and degree of degeneration for placental villi incubated under different con- ditions. Values are group means (c.v.) for ten placentae.

Variables Controls zero time

Hypoxia (6 per cent) Hyperoxia (40 per cent)

6h r2h 6h t2h

Vv(tr). per cent 35.6 (0.12) 30.3 (0.19) 32.5 (0.16) 30.3 (0.17) 29.8 (0.05) Vv(st). per cent 49.8 (0.10) 59.8 (0.09) 58.7 (0.10) 58.4 (0.10) 60.5 (0.05) Vv(fc), per cent 14.6 (0.14) 9.87 (044) 8.84 (0.23) I I .3 (0.30) 9.72 (0.25) Vv(dv), per cent 0.58 (1.33) I.55 (0.88) 4.43 (1.00) 3.73 (0.45) 3.73 (0.69) f>(v), /tm 38.7 (0.24) 44.2 (0.26) 36.6 (0.20) 49.8 (0.25) 54.9 (0.21)

The only alteration which was apparent with time in culture was an increase in the promin- ence of subtrophoblastic vacuolation from 6 to 12 h in hyperoxic conditions.

Quantitative findings Table I summarises the thickness variables for villous membrane and trophoblast in control, hypoxic and hyperoxic cultures at zero time, 6 h and I 2 h. Each value represents a group mean (c.v.). for ten placentae. Villous dimensions and composition are given in Table 2.

(a) Trophoblust thickness. Control organs displayed an arithmetic mean trophoblast thickness of 3.4 pm, an harmonic mean of 2.4 pm and a thickness ratio of 1.5. All three variables showed significant trends across the five experimental groups (for five related samples and ten organs per group, ‘L’ values for Tat, Tht and It were 516, 491.5 and 491 respectively, P < 0.01 in each case).

Two-way analyses of variance indicated that arithmetic and harmonic means were both in- fluenced by oxygen tension (for 1.36 degrees of freedom, F values were 14.4 and 12.5, P < 0.001 and P < 0.01 respectively). Table I shows that thicknesses tended to be greater during hyperoxia. There were no significant differences due to time in culture and no significant inter- action terms. The latter may be interpreted as indicating that time did not have a preferential effect at the one oxygen tension versus the other.

No significant main or interaction effects on trophoblast thickness ratio were demonstrable.

(b) Villous membrane thickness. At zero time and normal oxygen tension, arithmetic mean thick- ness was 7.4 pm, harmonic mean 5.5 pm and thickness ratio I .4. Variables tended to vary signi- ficantly across the five groups (‘L’ values were 522, 517 and 493.5, P < 0.01). Both arithmetic

auaJquIam snoII!A paleugsa Kaql moq aqgsap JOU p!p (9961) a%ed put: a%u!uro~ q%nOqL

.Kuyqe!JaA snolpa jo tualqoJd aq~ Joj loJluo3 01 pay aaaq a~ ‘M~!AJO splay Icydo~sorqur

ap!AoJd 01 K~p~pemalsKs paldluas uaql puv sdnolB aJnqn3 Jno JO q~ea 01 pau%!ssa Ajax anssp

jo sq3oIq lua~a!pe lvyl %u!Jnsua Ka %Jgdwes ua%Jo 01 uopuane l~!Jls dad plnoqs K%oloqdJoru

snolpn uo 1uauwaJ~ JO sl~aga aql JO Kpnls B l~ql pyuassa s! l! ‘aJojaJaq~ .(Lg61 ‘uouna put!

Kalqz~!J3 f9861 ‘oBuo~ pule lJaql!g ‘uo3a~ 3861 ‘Papa pue snaqww ‘~1~s log61 ‘LIEM~JS put?

umoJa ‘~60~ $$61 ‘ap?psea~ fP961 ‘xod) aemacwld uaurnq u!qg~ qxa aueJquIauI snolpA aql

JO ssauqqql put! K~Uolsu~ snolpa u! sawaJagp Ivuof%aJ lu~gpG3!s leql luaradde K@!seaJ3u!

s! 11 .palduIas aJaM aaluacwld aql qD!qM Kq SpoqlaLu aql JO Jo paseq aJaM sl[nsaJ aql q3!qM

uo sue%~o JO Jaqwnu aq) JO uo!~a3!pu! ou aA@ (9961) a%ed pm e%au!u~o~ ‘K~aleunlJoju~

~paluaJJaM sf slDadst! p?3!Uq3aJ aql jo Ma!AaJ ja!Jq e ‘s%u!puy Jfaql Bu!Jap!suo3 uaqM

.ua%Kxo iua3 rad 9 JO aJaqdsouw uf! u! uopeqn3u! q 9 Jalje ‘LLIT/ E.E 01 ur?f P+ lnoqe u10Jj

‘auBJqwatu snoll!A aql jo ssauq3!ql uf uoy3npaJ e paur!ep sJoqlne asaqL ‘oJl!A u! aJnlln3 uafo

u! pauyu!t?w uaqM suo!sual ua%Kxo iuaJag!p 01 ldepe 01 Klyqt! aqi ssassod op #!A lelua3

-eld leql(9961 ‘a%k!d put! e%au!uro~) u.y~ aql 1Joddns 01 Jvadde slInsaJ asaql ‘lq%!s 1s~~ 1~

‘!Il!A 1OJlUO3 UEql sa!.nq~!de3

p?laj JO Kl!SUap aurn~oh JalpwJs e peq !II!A paJngn3 ‘UOfSUal Ua%KXO aql jo aApDadsaJJ! ‘JaAa

-MOH TJa8Kxo 3ua3 Jad of pue lua3 Jad 9 it! payelu!tw saJnyn3 uaaMlaq !J!A JO uop!sodtuoD

~~JlauuqoA aql u! sa3uaJagp ou aJaM aJag lnq e!xoJadKq u! JalaaJ% SBM Jalame!p snoll!A uaaN

Tsau~s!ql u! I_uJoj!un aJocu SBM auaJqwaur snoII!A aql ‘armpI:, JO spotlad Ja%uol Jayv ‘aJnqn3

cyxoJad6q u! ldaa asoql uaql aueJqtuatu snol[!A Jauu!ql B pue lselqoqdoJ1 JO %u!JaAoJ Jauu!ql

e qloq passassod suoypuo~ D!xodKq Japun pauyu!ew !J!A ItluaDsld uemnq ‘Kpnls sty1 UI

y?AJalu! auy qloq le Jalauwp sno[pA uo a3uang

-II! J~~I_I~S E slJaxa uoisual ua%Kxo laql %uflsa%%ns ‘alqt?Jlsuotuap SBM LuJal uop3eJalu~ luwj!u

-%!s ON .( 10.0 > d ‘0.z I anlEA a) Jaiauwp snoll!A ueatu uo uo!suai ua%Kxo jo 13ajja we3y+s

f! SEM aJay ‘JaAaMoH ~a3uwj@i~s le3ps!lais uylt: IOU p!p uopeJaua%ap snolp~ jo a3uapyf

aql II! sa3uaJagp luaraddy ‘~I!A JO uop!sodwoD 3!JiauuyoA aql uo aurp x uofsua3 ua%Kxo Jo

auy ‘uo!sual uaBKxo JO sl3aBa ~ue~y!u%!s KUE It?aAaJ 01 paly aaut!!JeA JO sasdleue KEM-OM~

‘( 10’0 > d ‘I IS ‘7,) SUO!l!p

-uo3 D!xoJadKq %u!Jnp lsalaaJ8 aJaM sJalauIe!p lnq slo~luo3 II! leql 01 Jt3puys SBM JalauIe!p

snolpA ueaIu ‘SUBBJO yxoddq u! iseal 1~ ‘(10.0 > d ‘L6P ‘7,) .II. I IA aApeJaua%ap JO a3uapy

aqi u! put! (10.0 > d ‘EIS ‘3,) ewoJis jo auuqoa aApv\aJ aql u! sastlaJ3u! pul: (10.0 > d ‘EIS

‘7,) sa!Jt?lpdz3 p?laj pue (10.0 > d ‘96t ‘3,) lse1qoqdoJl JO sauuqolz ~euop3?~j aql u! saseaJ3

-ap %U!A~OAU! aJnyn3 u! sa%ueq3 aql ‘SdnOJB aql SSOJZ K.ICA 01 papual sa1qtyJEA 11~ ,(z a[qEL)

SWJOJ %upeJaua%ap 01 aIqanq!Jiie seM !II!A pw!mJal JO awnloA p201 aql JO lua3 Jad I ueql ssa?

.acw.ds Japwsea p!laj lua3 Jad SI put! SLLIOJ~S lua3 Jad OS ‘1sa~qoqdoJi lua3 Jad 9f pas!JdtuoD

put! u1ti6E inoqe JO JalauxHp uEatu I? ptq !J!A [O.I~UO~ ~.da&zuv~p puv uo,z~~so&uo~ sno~]q (3)

.paleaAaJ SBM 13aJa uo!wJalu!

lue3y!u%!s ou ‘u@v ‘uo!sual ua%Kxo Kq pal3ajjE lou SEM lnq (So.0 > d ‘kg an[EA d) aJn1

-in3 u! auy ql!M auq3ap 01 papual aueJqmatu sno[pA aql JO Opel ssaul3ql aql ‘1~~~1~03 UI

.auy pue uo!suai ua%Kxo uaaMiaq vaaa uop3aJaw! uxe~yp.~

-21s ou SBM aJaqL ~uoypuo~ D!xodKq u! Jalpws aJan\ sassauqD!qL .aJnlIn3 u! auy Kq JOU lnq

(100’0 > d ‘z+z put! 8’1 f sarqm a) uo!sual uaSKxo Kq paDuanuu! aJaM suk?atu 3!uowJaq put!

Rurton, Mqdew, Robertson: Placental villi in organ culture 269

thickness, it is reasonable to suppose from their account that minimum distances were measured. This probably accounts for the disparity between our respective estimates of villous membrane (arithmetic) mean thicknesses, i.e. 7.4 pm versus 4.4 pm for control samples.

Most estimates of intervascular barrier thicknesses in the past have been biased by artefacts of sectioning which need to be catered for (see Jackson et al, 1985). They have been biased also by the problem of vessel collapse that accompanies delivery of the placenta (Voigt, Kaufmann and Schweikhart, 1978; Burton, Ingram and Palmer, 1987). For material fixed by immersion as in the present study, Burton, Ingram and Palmer (1987) found an arithmetic mean thickness of 6.0 pm. This was associated with a mean capillary volume fraction of 26 per cent and, within the immersion-fixed placentae, there was a strong negative correlation between the mean thick- ness and the capillary volume fraction (correlation coefficient = - 0.81 for ten organs). Recently, Jackson, Mayhew and Haas (r988a) reported equivalent arithmetic means (6.22 6.6 pm) for lowland placentae and, again, these were associated with similar capillary volume fractions (Jackson, Mayhew and Haas, 1987a,b).

In the present investigations, the mean thickness of the villous membrane at time zero was higher (7.4 pm) but it should be noted that the capillary volume fraction was correspondingly lower (I 5 per cent). The discrepancies are probably due to collapse of the fetal capillary bed, particularly to profuse leakage from the capillary network as a consequence of the tissue dicing procedure. From a regression analysis based on the findings of Burton, Ingram and Palmer (1987), a figure of 7.1 pm would be predicted for a volume fraction of 15 per cent and this approximates the value estimated here rather well.

Evidently, further collapse of fetal vessels occurred during the course of the organ culture as demonstrated by the fall in capillary volume fraction. On this basis, one might predict a further increase in arithmetic mean thickness of the villous membrane. Indeed, from the data pre- sented in Table 2, the predicted thicknesses after 12 h of culture under ‘normoxic’ con- ditions would be 7.6 pm (compared to 6.3 pm for the capillary volume fraction of 9 per cent found in hypoxia) and 7.5 pm (compared to 8.9 pm for the volume fraction of IO per cent in hyperoxia).

Since one might have expected the villous membranes to have increased in thickness during organ culture as a result of the continued collapse of the fetal vascular bed, the thinning actually observed under hypoxic conditions is even more striking. However, before discussing this effect, we will consider the changes found in hyperoxia as these are possibly easier to inter- pret.

Effects of hyperoxia in organ culture Tominaga and Page (1966) found no difference in membrane thickness following incubation of villi in an atmosphere of 95 per cent oxygen but other workers have suggested that elevated levels of oxygen are toxic in organ culture. For example, Huot, Foidart and Stromberg (1979) found that the level of hCG synthesis was considerably higher in villi maintained in 21 per cent oxygen rather than 50 per cent, or than 95 per cent oxygen which was rapidly lethal.

Although there is no information available for the placenta, qualitative and quantitative changes in the structure of the lung may result from oxygen toxicity (Kistler, Caldwell and Weibel, 1967; Kapanci et al, 1969; Crapo, 1986). The first signs are an increase in capillary permeability (resulting in connective tissue oedema) and a progressive thickening of the air- blood barrier, though the timing of these changes and the sensitivity of the lung appear to be species-dependent. Similar alterations may be occurring within placental villi in our hyperoxic groups: it is noteworthy that mean villous diameter was significantly greater in these groups than in the control or hypoxic groups. Because of the rigorous sampling protocol, this can only

270 Placenta (1989), Vol. 10

indicate that villi became swollen during culture, possibly owing to stromal oedema. The oxy- gen levels used here were on the borderline for those causing pulmonary changes and the time scale was surprisingly short. However, it may be that placental villi are more vulnerable than lung tissue since they are normally exposed to considerably lower oxygen partial pressures in vivo. Ultrasound-guided intervillous blood sampling in utero has shown recently that the intervillous oxygen partial pressure is only about 40-50 torr at term (Soothill et al, 1986).

The effects of high oxygen tensions on placental villi may not have been noticed in the past because of the different measuring techniques employed. It is notable that oedema fluid in the lung never intervenes between capillary endotheliocytes and alveolar epitheliocytes at sites of their most intimate apposition (Kistler, Caldwell and Weibel, 1967), sites which are equivalent to the vasculosyncytial membranes in the placenta. If the same holds for the placenta, then it is not surprising that Tominaga and Page (1966) failed to find evidence of oxygen toxicity in their hyperoxic organs: they measured only minimum diffusion distances!

It seems reasonable to suggest that the results presented here may be reflecting the early stages of oxygen toxicity rather than a genuine structural adaptation to hyperoxia.

Effects of hypoxia in organ culture In contrast to the situation obtaining in hyperoxia, thinning of the villous membrane in re- sponse to hypoxia was not associated with any change in villous diameter. The response appears to be consistent with similar alterations observed under conditions of chronic hypoxia in vivo: in different regions of placental lobules (Critchley and Burton, 1987) and in placentae from high-altitude populations (Jackson, Mayhew and Haas 1988~). However, some import- ant differences exist for, in both these in vivo investigations, only the harmonic mean thickness of the villous membrane declined whilst the arithmetic mean remained constant. This implies that the villous membrane had become more irregular in thickness with some areas, presum- ably vasculosyncytial membranes, becoming particularly attenuated.

In this study, the arithmetic mean thickness also decreased and the lower thickness ratio indicated that the villous membrane became more uniform in thickness with increasing time in culture. Contributing to the reduction in membrane thickness was a decrease in the arithmetic mean thickness of the trophoblast, suggesting a relative loss of trophoblast volume per unit of surface. Attentuation of syncytiotrophoblast has also been noted in hypoxic culture (Fox, 1970; MacLennan, Sharp and Shaw-Dunn, 1972) where it was accompanied by involution of many of the intracytoplasmic organelles and by proliferation of cytotrophoblast. The fact that the mechanism of membrane thinning is so clearly different in vitro from that witnessed in vivo reinforces the doubt that one is observing a genuine adaptive response to hypoxia rather than merely detecting early pathology induced by the culture conditions. Preliminary ultrastruc- tural studies (Burton and Mayhew, unpublished results) confirm the presence of extensive damage to trophoblast by 12 h of culture, the damage being more severe in hypoxic versus hyperoxic groups.

The confounding effects of ischaemia Since this investigation is based on non-perfused organ culture it is important to recognize that ischaemia (lack of blood flow) may be a confounding factor.

Studies on human and guinea-pig placentae have tried to resolve the effects of ischaemia from those of anoxia (reviewed in Kaufmann, 1985). It appears that fetal ischaemia is of minor import whilst maternal ischaemia leads to ultrastructural changes in the syncytiotrophoblast in as little as I-Z min. Anoxic perfusion induces mitochondrial swelling alone, normoxic malper- fusion leads to vacuolation of rough endoplasmic reticulum (RER) cisternae whilst ischaemia

Burton. Ma,yhew, Robertson: Placental villa in organ culture 271

affects both mitochondria and RER. Additional effects of ischaemia include collapse of fetal vessels, feto-maternal fluid shift, stromal oedema, greater intervascular distances, loss of vas- culosynctyial membranes, focal loss and focal hyperplasia of trophoblastic microvilli and distortion of microvillous shape (Illsley et al, 1985; Kaufmann, 1985).

Double-sided artificial perfusion with hyperoxic medium (95 per cent oxygen, 5 per cent carbon dioxide) is also associated with stromal oedema and with the appearance of focal vacuolation below the trophoblast (Kaufmann, 1985; Miller et al, 1985; Kuhn et al, 1988). Finally, subtrophoblastic oedema occurs with avascular degeneration in villi viewed by phase- contrast microscopy after aspiration and in placentae from cases of prenatal fetal hypoxia (Aladjem, 1975).

The present study is consistent with previous studies insofar as we noted greater sub- and intra-trophoblastic vacuolation in culture than in control tissue. In addition, however, we observed expansion of the trophoblastic surface in culture and this seemed to be more exagger- ated in hyperoxic villi. It is possible that the effects of hyperoxia combine with the effects of ischaemia to cause further dilation of subtrophoblastic vacuoles and further stromal oedema. This could help to explain the increase in mean villous diameter found in hyperoxic conditions.

In accord with the findings of Kaufmann (1985), we found that total ischaemia led to fetal vessel collapse with a concomitant increase in the volume fraction of stroma. However, we did not detect any increase in the volume fraction of trophoblast or, in hypoxic culture, any in- crease in mean trophoblast thickness: this may be due to the fetal vessel collapse which pre- ceded culture. It is unlikely that the increased prominence of trophoblastic surface processes seen here by light microscopy reflects ultrastructural changes in microvillous shape and density reported by Illsley et al (1985): our own ultrastructural studies suggest that the processes are not modified microvilli but large cytoplasmic blebs from grossly damaged trophoblast.

Concluding remarks This investigation has exposed some of the problems of interpreting data from organ culture experiments and correlating them with data gleaned in vivo. At present, we feel that there is little justification for claiming that placental villi can adapt in vitro to differing oxygen tensions. Firmer evidence will be adduced when it is shown that morphometric changes can be reversed on return to normal levels of oxygenation.

SUMMARY

The effects of exposure to various oxygen tensions on villi in organ culture are re-examined. Villi from ten normal mature placentae were cultured under hypoxic (6 per cent oxygen) and hyperoxic (40 per cent oxygen) conditions for 6 or 12 h. Control tissue (zero time in culture) was also taken. Pieces of tissue were fixed by immersion and embedded in resin for semithin sectioning. Systematically sampled microscopical fields were analysed stereologically to estim- ate harmonic and arithmetic mean thicknesses for the trophoblast and for the villous mem- brane and to assess the volumetric composition and mean diameter of villi.

Trophoblast thicknesses were influenced significantly by oxygen tension, being smaller in hypoxic and greater in hyperoxic media. No significant interaction terms or effects of time in culture were detected. Villous membrane thicknesses altered in a similar fashion to trophoblast thicknesses. No significant differences in the composition of villi were detected but villi tended to be greater in diameter during hyperoxia.

Findings are discussed in the context of previously reported adaptations in vivo, in different

272 Placenta (1989), Vol. I0

regions of the placental lobule and during chronic maternal hypoxic stress at high altitude. We conclude that there is insufficient evidence to support the claim that villi can adapt in vitro to varying ambient oxygen tensions.

ACKNOWLEDGEMENTS

Parts of this study were undertaken by Miss Lesley Robertson to fulfill requirements for the Honours B.Sc. degree in Anatomy at the University of Aberdeen. We are most grateful to the Labour Ward of the Rosie Maternity Hospital (Cambridge) for help in obtaining organs and to Miss M. E. Palmer for expert assistance in preparing histological sections. This study was supported by a grant from Action Research for the Crippled Child.

REFERENCES

Aladjem, S. (1975) Examination of fresh villi. In The Placenta and Ifs Maternal Supply Line. (Ed.) Gruenwald, P. pp. 221-231. Lancaster: Medical and Technical Publishing Company.

Amaladoss, A. S. P. & Burton, G. J. (1985) Organ culture of human placental villi in hypoxic and hyperoxic con- ditions: a morphometric study. Journal of Developmental Physiology, 7, I 13~118.

Bacon, B. J., Gilbert, R. D. & Longo, L. D. (1986) Regional anatomy of the term human placenta. Placenra, 7,233- 241.

Boyd, P. A., Brown, R. A. & Stewart, W. J. (1980) Quantitative structural differences within the normal term human placenta: a pilot study. Placenta, I, 337-344.

Burton, G. J., Ingram, S. C. & Palmer, M. E. (1987) The influence of mode of fixation on morphometrical data derived from terminal villi in the human placenta at term: a comparison of immersion and perfusion fixation. Placenta, 8,37-5 I.

Burton, G. J. & Palmer, M. E. (1988) Eradicating fetomaternal fluid shift during perfusion fixation of the human placenta. Placenta, 9, 327-332.

Crapo, J. D. (1986) Morphologic changes in pulmonary oxygen toxicity. Annual Reviews of Physiology, 48,721-73x. Critchley, G. R. & Burton, G. J. (1987) Intralobular variation in barrier thickness in the mature human placenta.

Placenta, 8,185~194. Fox, H. (1964) The pattern of villous variability in the normal placenta. Journal of Obsrerrics and Gynaecology of the

British Commonwealth, 71,749758. Fox, H. (1967) The incidence and significance of vasculo-syncytial membranes in the human placenta. Journal of

Obsretrics and Gynaecology of the British Commonwealrh, 74,28-33. Fox, H. (1970) Effect of hypoxia on trophoblast in organ culture. A morphologic and autoradiographic study.

AmericanJournal of Obsterrics and Gynecology, 107, 1058-1064. Hirose, K., Osterby, R., Nozawa, M. & Gundersen, H. J. G. (1982) Development of glomerular lesions in experi-

mental long-term diabetes in the rat. Kidney International, 21,68e5. Huot, R. I., Foidart, J.-M. & Stromberg, K. (1979) Effects of culture conditions on the synthesis of human chor-

ionic gonadotrophin by placental organ cultures. In Vitro, 15,497-502. Illsley, N. P., Fox, H., van der Veen, F., Chawner, L. & Penfold, P. (1985) Human placental ultrastructure after

in vitro dual perfusion. Piaceflra, 6, 23-32. Jackson, M. R., Joy, C. F., Mayhew, T. M. & Haas, J. D. (1985) Stereological studies on the true thickness of the

villous membrane in human term placentae: a study of placentae from high-altitude pregnancies. Placenta, 6, 249- 258.

Jackson, M. R., Mayhew, T. M. & Haas, J. D. (1987~) Morphometric studies on villi in human term placentae and the effects of altitude, ethnic grouping and sex of newborn. Placenm, 8,487-495.

Jackson, M. R., Mayhew, T. M. & Haas, J. D. (19876) The volumetric composition of human term placentae: alti- tudinal, ethnic and sex differences in Bolivia.Furnal of Anatomy, 152, 173-187.

Jackson, M. R., Mayhew, T. M. & Haas, J. D. (r988a) On the factors which contribute to thinning of the villous membrane in human placentae at high altitude. I. Thinning and regional variation in thickness of trophoblast. Placenta, 9, I-8.

Jackson, M. R., Mayhew, T. M. & Hass, J. D. (19886) On the factors which contribute to thinning of the villous membrane in human placentae at high altitude. II. An increase in the degree of peripheralization of fetal capillaries. Placenta, 9, ~18.

Jensen, E. B., Gundersen, H. J. G. & Osterby, R. (1979) Determination of membrane thickness distribution from orthogonal intercepts. j’oumal of Microscopy, 1x5, 1~33.

Kapanci, Y., Weibel, E. R., Kaplan, H. P. &Robinson, D. V. M. (1969) Pathogenesis and reversibility of the pul- monary lesions ofoxygen toxicity in monkeys. Laboratory Investigarion, 20, ~OI-I 18.

Burtorr, Ma,yhem, Robertson: Placental villi in organ culture 273

Kaufmann, P. (1985) Influence of &hernia and artificial perfusion on placental ultrastructure and morphometry. In In Vitro Perfusion of Human Placental Tissue. (Ed.) Schneider, H. & Dancis, J. pp. 18~26. Basel: S. Karger.

Kistler, G. S., Cnldwell, P. R. B. & Weibel, E. R. (1967) Development of fine structural damage to alveolar and capillary lining cells in oxygen-poisoned rat lungs.3ournal of Cell Biology, 3q605-628.

Kuhn, D. C., Crawford, M. A., Gordon, G. B. & Stuart, M. J. (1988) Aspects of in vitro placental perfusion: effects of hyperoxia and phenol red. Placenta, 9,201~213.

Laga, E. M., Driscoll, S. G. & Munro, H. N. (1973) Quantitative studies of human placenta. I. ‘Morphometry. Biology of the Neonate, 23, 231-259.

MacLennan, A. H., Sharp, F. & Shaw-Dunn, J. (1972) The ultrastructure of human trophoblast in spontaneous and induced hypoxia using a system of organ culture: a comparison with ultrastructural changes in pre-cclampsia and placental insufficiency. 3oournal of Obstetrics and Gynaecology of the British Commonwealth, 79, I 13- 121.

Mayhew, T. M. & Burton, G. J. (1988) Methodological problems in placental morphometry: apologia for the use of stereology based on sound sampling practice. Placenta, 9,565-581.

Miller, R. K., Wier, P. J., Maulik, D. & di Sant’Agnese, P. A. (1985) Human placenta in vitro: characterization during 12 hours of dual perfusion. In In Vitro Perfusion of Human Placental Tissue. (Ed.) Schneider, H. & Dancis, J. pp. 77-84. Basel: S. Karger.

Miller, S. (1975) Experimental Design and Statistics. London: Methuen and Company Limited. Sala, M. A., Matheus, M. & Valeri, V. (1984) A stereologic study on the thickness of the trophoblastic layer and its

regional variation in the human term placenta. Anatomisches Anzeiger,Jeena, x57,257-260. Sokal, R. R. & Rohlf, F. J. (1981) Biometry. The Principles and Practice of Statistics in Biologtcal Research. San Fran-

cisco: W. H. Freeman and Company. Soothill, P. W., Nicolaides, K. H., Rodeck, C. H. & Cambell, S. (1986) Effect of gestational age on fetal and inter-

villous blood gas and acid-base values in human pregnancy. Fetal Therapy, I, 1688179. Teasdale, F. (1978) Functional significance of the zonal morphologic differences in the normal human placenta: a

morphometric study. AmericanJournal ofobstetrics and Gynecology, r30,7733781. Tominaga, T. & Page, E. W. (1966) Accommodation of the human placenta to hypoxia. .4merrcan3ojournal qfObstet-

rtcs and Gynecology, 94,67949 I. Voigt, S., Kaufmann, P. & Schweikhart, G. (1978) Zur Abgrenzung normaler, artefizieller und pathologischer

Strukturen in reifen menschlichen Plazentazotten. I. Morphometrische Untersuchungen zum Einfluss des Fixationsmodus. Archia fur Gynakologie, 226,3477362.

Weibel, E. R. (1979) Stereologrcal Methods, volume I. Practical Methods for Biologtcol Morphometry. New York: Academic Press.