y. Cell Sex. 44, 87-101 (1980)Printed in Great Britain © Company of BiologitU Limited 1980

THE DEVELOPMENT OF LAMPBRUSH

CHROMOSOME-TYPE TRANSCRIPTION IN THE

EARLY DIPLOTENE OOCYTES OF XENOPUS

LAEVIS: AN ELECTRON-MICROSCOPE

ANALYSIS

R. S. HILL AND H. C. MACGREGOR

Department of Zoology, University of Leicester, Leicester LEi 7RH, England

SUMMARY

Oocytes of Xenoptis laevis in pachytene and early diplotene of meiosis have been studied usingthe Miller spreading technique. Transcription first appears in germinal vesicles 25-40 /tmin diameter, when the oocyte is in early diplotene. Transcription at this stage consists of arraysof short RNP transcripts, irregularly distributed along the DNP axis. Linear regression ana-lysis has shown that many of these arrays are transcription units (Type 1) with the transcriptshaving a common point of origin. The lengths of these early transcription units (mean =706 ± 506 ftm), calculated from the linear regression data, are comparable to the lengths oftranscription units from later stages, including Dumont stage 3. The polymerase granules ofType 1 transcription units are widely and irregularly spaced, having a mean spacing of5°5±748nm. More advanced transcription units (Type II, mean length — 8-72 ± 3-77 fim)are usually found in the same chromosome set as the Type I units. Type II transcription unitshave closer and more regularly spaced polymerase granules than Type I transcription units(mean spacing = a2±4onm). Both Type I and II transcription units have comparativelyshort RNP transcripts, the mean values for the slopes of their regression lines being 0-1336and 0-1440 respectively. By the time the germinal vesicles are about 50-60/tm in diameterthe transcription units have a quite different morphology (Type III). The lengths of the TypeIII transcription units are comparable to the Type I and II units, the mean length being6-34 ±4-03 /tm. The spacing of the polymerase granules in the Type III units is closer andmore regular than the earlier stages (70140 run). Another significant difference betweenType III and Types I and II transcription units is a decrease in the foreshortening of theType III RNP transcripts. The mean slope of the regression lines for Type III transcriptionunits is 02439. The morphological appearance of the Type III transcription unit is virtuallyidentical to that of the transcription units from Dumont stage 3 oocytes, both with respect tothe length and the spacing of the polymerase granules. However, the transcripts in Type IIItranscription units are still more foreshortened than those of Dumont stage 3 oocytes, havingmean regression slopes of 0-4728. From the data obtained in the present study, it has beenconcluded that the pattern of lampbrush-type transcription is virtually fully established by thetime most germinal vesicles are about 50 /wn in diameter.

INTRODUCTION

The lampbrush chromosome stage of meiosis in amphibians has been the subjectof many biochemical and cytological studies in recent years (reviews by Davidson,1976; Sommerville, 1977; Macgregor, 1980). Nevertheless, remarkably little is knownabout how the lampbrush form develops, mainly because early diplotene stages are

88 R. S. Hill and H. C. Macgregor

difficult to study with the more conventional cytological methods. The acetic acid-orcein or Feulgen squash techniques have, for example, provided much informationon the structure of chromosomes in the leptotene, zygotene, pachytene and firstmeiotic metaphase, but are virtually useless for the study of early diplotene. Asquash preparation of an early diplotene germinal vesicle appears as a diffuse mass oflightly staining material in which individual chromosomes are not distinguishable.Similarly, the manual isolation method developed by Gall (1954), and Callan &Lloyd (i960), for preparing lampbrush chromosomes for phase-contrast microscopycannot be used in oocytes in the early lampbrush stage, because the germinal vesiclesare too small to handle.

It is well known from biochemical studies that certain important classes of RNAare synthesized in early diplotene, before the lampbrush loops are thought to bemaximally extended (Dumont stage 3, Dumont, 1972). Rosbash & Ford (1974)have reported that the bulk of poly(A)-containing RNA synthesized during oogenesisin Xenopus laevis is accumulated well before the oocytes enter Dumont stage 3.Furthermore, messenger-RNAs translatable into proteins using in vitro wheat germsystems are readily extractable from Dumont stage 1 previtellogenic oocytes of X.laevis (Darnbrough & Ford, 1976; Ruderman & Pardue, 1977). It is also known that4S and 5s RNAs are transcribed abundantly in previtellogenic oocytes of X. laevis(Thomas, 1974; Ford, 1971). As yet there is no structural information about howthis early transcriptional activity is related to the formation of lampbrush loops.

The aim of the present study was to analyse the development of transcriptionalactivity during early diplotene using the Miller spreading technique (Miller & Bakken,1972) and to search for evidence concerning the formation of lampbrush loops.Transcriptional activity has been analysed with respect to the lengths of active DNPfibres, the spacing of polymerase granules, and the structural characteristics of thelateral RNP transcripts attached to the chromosomal DNA. The data so obtainedhave been compared with the same parameters for transcription units found in Dumontstage 3 oocytes of X. laevis, where the lampbrush chromosomes are fully formed(Hill, 1979).

MATERIALS AND METHODS

Oocytes in early meiotic prophase were collected from the ovaries of young X. laevis, whichhad recently undergone metamorphosis. There were no specific problems in obtaining germinalvesicles larger than 30 fim in diameter; these were manually isolated in 5 :1 saline (5 partso-i M KC1 to 1 part o-i M NaCl). However, to obtain smaller oocytes, less than 20-25 /im indiameter, the ovaries were collagenased according to the method of Eppig & Dumont (1976).Basically, small pieces of ovary were rinsed in sterile, Ca*+-free ORS medium (Eppig & Dumont,1976) and then digested in a 0 4 % collagenase solution (S igma-Type I) made up in Ca1+-free ORa for 2-4 h. The oocytes, free of contaminating follicle cells, were sorted according tosize and then staged according to the observations and measurements made on i-^m-thicksections of immature ovaries embedded in either Spurr or methacrylate (Challenger & Hill,unpublished observations). In these sections, nuclei in pachytene had diameters of 10—15 Z"1*and were found in oocytes 15-20 fim. in diameter. In oocytes 20-30 fim in diameter (latepachytene-early diplotene), the chromosomes and the ribosomal DNA cap region, characteris-tic of pachytene in X. laevis (Macgregor, 1968), were more diffuse. In oocytes 50-60 fim in

Transcription in early diplotene oocytes 89

diameter, with germinal vesicles 30-40 /4m in diameter, no chromosomal structure was recog-nizable (diffuse diplotene). In an attempt to confirm the classification based on sections, livingoocytes 10-30 /tm in diameter were collected after collagenase digestion and examined withphase-contrast and Nomarski interference microscopy.

Chromatinfrom germinal vesicles 30—60 /tm in diameter wasprepared for electron microscopyaccording to the method of Miller & Bakken (1972), modified as described by Hill (1979). Themanually isolated germinal vesicles were rinsed in pH 9 water (glass-distilled water, broughtto pH 9 by the addition of 0-16 M borate-NaOH buffer) before being placed in a microcen-trifugation chamber. One germinal vesicle was used for each preparation. The chambercontained a lower layer of 4 % formaldehyde ino-i M sucrose at pH 8 5 and an upper layer ofpH 9 water containing 005 % of the commercial detergent, Joy (Procter & Gamble, Cincinnati,Ohio). The germinal vesicles usually disappeared within 1—2 min and were allowed to dispersefor 15 min before being centrifuged at 2390 g for 20 min. Formvar and carbon-coated coppergrids were glow discharged for 1-5 min at 0 1 Torr (1-3 x io* N m"1) to make them hydro-philic. After centrifugation, the grids were rinsed in pH 9 water and rinsed in 0-25 % Photoflo,before being air-dried and rotary-shadowed with platinum-palladium at an angle of 70.

To spread chromatin from oocytes less than 20-25 /im in diameter, 2-5 oocytes were placedin a watchglass containing 005 % Joy inpH 9 water until they ruptured and were then trans-ferred to a microcentrifugation chamber and allowed to disperse for 20 min. Alternatively,5-10 cells were rinsed in pH 9 water, placed in a drop of 0-05 % Joy in pH 9 water on a sili-conized coverslip and allowed to disperse for 20 min. The homogenate was transferred to amicrocentrifugation chamber, centrifuged and treated in the manner described above. Thepreparations were examined at magnifications of between 8000 and 25000 diameters in aPhilips 200 or Siemens 102 electron microscope operating at 80 kV. The microscopes werecalibrated at all working magnifications using an EMscope grating replica, 2160 lines per mm.

Measurements of the lengths of transcription units and RNP transcripts were made using aKontron MOP/AMO2 digitizer from prints and tracings enlarged between 2 and 6 times.Analyses of polymerase spacing, and the size and spacing of the RNP transcript beads weremade using a x 10 measuring eyepiece containing a 20-mm graticule. The irregular distribu-tion of RNP transcripts found on chromatin during the early stages of transcriptional develop-ment was analysed using linear regression analysis. Where appropriate, the data were testedusing X* and Student's t test.

RESULTS

In oocytes 10-25 Z4111 m diameter there was no evidence of transcriptional activityin the form of identifiable transcription units. In all preparations examined, thechromatin consisted of a tangled mass of inactive, nucleosomal fibres. However,because of the opacity of many of the chromatin aggregates it is not possible to statecategorically that transcriptional activity is absent in oocytes of these sizes; the oc-casional RNP transcript or transcription unit could have been overlooked or hidden.By using phase-contrast and Nomarski interference microscopy it was shownthat oocytes of 10-25 fim in diameter were in pachytene and early diplotene. Somenuclei possessed a distinctive pachytene cap, while others had rounded granules,presumably early nucleoli, scattered throughout the nucleoplasm (Figs. 1-3).

Two phases have been observed in the development of lampbrush-type trans-cription. The first, or formative phase was found in oocytes with germinal vesicles25-40 /im in diameter where putative transcription units first appear. In the secondphase, transcription units had developed to the stage where they closely resembledtranscription units found in chromatin from oocytes in Dumont stage 3.

CEL44

R. S. Hill and H. C. Macgregor

Transcription in early diplotene oocytes 91

The formative phase of lampbrush-type transcription

The stage at which lampbrush-type transcription units first appear has not beendetermined exactly, although the formative stage is complete in germinal vesicles40-60 /tm in diameter since transcription units virtually identical to those found inoocytes from Dumont stage 3 are abundant. During the formative phase the trans-criptional activity can be divided into 2 morphological categories. The first categoryconsisted of arrays of irregularly spaced RNP transcripts, the number of whichvaried from 5 to 20 in any one array. Notwithstanding the irregular and sparsedistribution of transcripts, many of these arrays looked as though they might rep-resent transcription units. To test this interpretation, 16 putative transcription unitswere studied using linear regression analysis, in the manner described by Laird,Wilkinson, Foe & Chooi (1976). The length of each RNP transcript in an array wasmeasured, as was its distance along the DNP axis from the first measurable transcriptin the array. Where possible every transcript in each array was measured, this beingrelatively easy for the early transcription units, although the occasional transcriptwhose length could not be determined accurately was not included in the analysis.However, because of the long and complex nature of their RNP transcripts this was aparticular problem with Type III transcription units. Consequently, only units withat least 10 analysable transcripts were studied. Using the sum of the least-squaresmethod, a regression line was calculated for each array, together with its correlationcoefficient (r). The 16 transcriptional arrays had correlation coefficients above 07718,the mean value being 0-9167+ 0-0379 (Table 1), indicating a strong correlationbetween the lengths of the RNP transcripts and their position with respect to a singlecommon origin. Arrays of this kind were classified as Type I transcription units. TwoType I transcription units are illustrated in Figs. 4 and 6, together with their re-gression data. However, not all the transcriptional activity could be classified clearlyas transcription units. Occasionally, 2-3 isolated RNP transcripts which could not beassigned to a transcription unit were observed.

Using the formula for a straight line, y = a + bx, where a = the intercept on they axis, b = the slope of the regression line, x, the intercept on the x axis, could be

Figs. 1-3 are photomicrographs of small oocytes from recently metamorphosedXenopus laevis, photographed with a Zeiss Universal R microscope fitted withNomarski interference-contrast equipment. Pieces of ovary from animals which hadmetamorphosed within 3 weeks were dissociated with collagenase in OR2. Individualoocytes were suspended in fresh OR2 and photographed immediately. Theoocytes were not flattened and the photographs give a true impression of the dimen-sions of the cells and their nuclei. All micrographs are produced at the same scale.Scale bar = 20 /tm.

Fig. 1. Late pachytene oocyte nucleus showing the characteristic cap (c) of extra-chromosomal DNA.

Fig. 2. Early diplotene nucleus. The cap of nucleolar DNA has disappeared and anumber of small nucleoli have formed, m, mitochondria] cluster or 'yolk nucleus'.

Fig. 3. Later diplotene nucleus showing more small nucleoli. Note that no chromo-somal threads are visible in either Fig. 2 or 3.

7-2

R. S. Hill and H. C. Macgregor

Table i. Linear regression data for the transcription units ateach stage of development

n

Correlation coefficient,M±S.D.

Slope of regression line,MiS.D.

Probability-Student's ttest

RNP packing (DNPL/RNPL)

FormativeA

Type I16

0-9167!003790-1336!00582

phase

Type II10

0-9303!003940-1440 100572

N.S.

75 69•Partly from Hill, 1979.

Established\ phase

Type III10

0-9340!

0-0492

0-2439!0-0711

1 \Y

o-oi 0

41

Dumontstage 3*

10

0-84281006340-4728 10-0493 1

1

•001

2-1

calculated, since x = — a/b, when y = o. The length of the transcription unit couldthen be estimated by adding x to the distance of the terminal transcript from thefirst measured transcript (where x = 0). This was considered to be the most reliablemethod of estimating the lengths of incomplete transcription units, since they couldnot be measured directly with a digitizer. The mean calculated length of the Type 1transcription units was 7-06 + 5-06 Jim (Table 2). The coefficient of variation(c.v. = S.D./J/X 100) of the lengths was 71-6%, indicating a wide range in the lengthsof the individual transcription units.

The spacing of the polymerase granules in Type I transcription units was alsomeasured, the mean value for the 16 units being 505 ±748 nm. The high c.v. valueof 148% illustrates the irregularity in the distribution of the RNP transcripts(Table 3).

A further characteristic of the Type I transcription unit was the high degree offoreshortening of the length of the RNP transcripts (Figs. 4, 6). This was illustratedby the comparatively low values for the slopes of the regression lines for Type I units,the mean value being 0-1336 ±0-0582 (Table 1). A simple index of foreshorteningwas obtained by calculating the packing ratio of the RNP fibres. The packing ratiowas defined as the length of the DNP fibre divided by the length of the RNP trans-cript. The mean value for all the transcripts on a transcription unit is the reciprocal

Fig. 4. Type I transcription unit from an oocyte about 35 fim. in diameter. Note thewidely and irregularly spaced RNP transcripts, which are highly foreshortened, i, t,the initial and terminal transcripts measured in the regression analysis. Scale bar= 1 /4m.

Fig. 5. Interpretive drawing of the transcription unitinFig. 4. This unit had a corre-lation coefficient of 0-9772 and a regression line slope of 0-0941. i, t, as in Fig. 4.Fig. 6. A more regularly transcribed Type I unit, from the same preparation as Fig. 4.1, t, as in Fig. 4. Scale bar = 1 fim.Fig. 7. Interpretive drawing of Fig. 6. This unit had a correlation coefficient of 0-9038and a regression line slope of o-i 187. i, t, as in Fig. 4.

Transcription in early diplotene oocytes

+r ."vrs'Ssss"?..:

•tf*

G

94 R- S. Hill and H. C. Macgregor

Table 2. The lengths of transcription units at each stage of development

Formative phase Establishedphase Dumont

Type I Type II Type III stage 3*

n 16 10 24 19Length (/tm), 706 ±5-06 872 ±377 6-34 ±4/08 628 ±609

M ± S . D . v , ' ' » ' ( » 'Probability - Student's t N.S. N.S. N.S.

testCoefficient of variation 72% 43% 64% 97%(c.v.) S.D./jf x 100

• From Hill, 1979.

Table 3. The spacing of polymerase granules in transcription unitsat each stage of development

nPolymerase spacing (nm),M±S.D.

Probability - Student's ttest

Coefficient of variation(c.v.)

Formative

Type I

16505 ± 748

N.S.

148%

• From Hill

phase

Type II

1092 ±49

0-055 3 %

. 1979-

Established-> phase

Type III

2470 ±40

N.S. N.S.

5 7 %

Dumontstage 3 #

1972 ±41. .'

57 /o

of the slope of the regression line. The mean RNP packing ratio for the Type Itranscription units was 7-5 (Table 1). The RNP transcripts themselves consisted ofrows of beads, the mean diameter of the beads being 3i-4± 5-0 nm with a centre-to-centre spacing of 38-0 ± 6-7 nm.

The second type of transcriptional activity found in the formative phase consistedof well-defined transcription units, designated as Type II (Fig. 8). Type I and IItranscription units were regularly observed within the same chromosome set and weresometimes found adjacent to one another on a common stretch of DNP axis. Theregular distribution of the RNP transcripts in Type II transcription units allowed theunits to be measured directly with a digitizer. Their mean measured length was8-72 ± 377 fim with a c.v. of 43-2% (Table 2). When the lengths of the transcriptionunits were calculated using linear regression analysis there was a discrepancy between

Fig. 8. Type II transcription unit, from the same preparation as Figs. 4, 6. Thetranscripts are more closely and regularly spaced than those of Type I units, but arestill highly foreshortened, i, t, as in Fig. 4. Scale bar = i fim.Fig. 9. Interpretive drawing of Fig. 8. This unit had a correlation coefficient of 09038and a regression line slope of 0-0844.

Transcription in early dtplotene oocytes

t - : i . , • • • - V . • • •

; : ^ ; ; ^

96 R. S. Hill and H. C. Macgregor

the results obtained by the 2 methods. The mean length calculated from linearregression analysis was 8-57 ± 3-65/tm. However, a x2 analysis of the two sets oflengths revealed that there was no significant difference between the values obtainedby the 2 methods (P > o-8). It was concluded that the lengths calculated for Type Itranscription units using linear regression analysis could confidently be comparedwith the lengths of Type II and later transcription units measured directly with adigitizer.

The mean spacing of polymerase granules in Type II transcription units was92 ± 49 nm. The mean value of the coefficients of variation was 53 %, indicating thatnot only were the RNP transcripts more closely spaced in Type II transcription units,but they were also more regularly distributed along the DNP axis (Table 3).

As with Type I transcription units the RNP transcripts showed a comparativelyhigh degree of foreshortening (Fig. 8, Table 1). The mean slope of their regressionlines was 0-1440 ±0-0572 and the mean value for the RNP packing ratio was 6-9.

The arrangement of Type I and II transcription units on the DNP axes was irregu-lar, in most cases only 1 or 2 units were found on a common stretch of axis, although inone case a sequence of 4 units was found along an 86-/tm stretch of continuous DNP.

Established lampbrush-type transcription

Germinal vesicles more than 40-60 /tm in diameter possessed quite a differentpattern of transcriptional activity from that found in the formative phase. Here thetranscription units had many of the characteristics commonly found in later lampbrushstages (Figs. 10, 11) and were designated as established, or Type III transcriptionunits. The mean measured length of the Type III units was 6-34! 4-08/tm, with amean c.v. value of 64% (Table 2). The mean measured length of transcription unitsfrom Dumont stage 3 oocytes was 6-28 ± 6-09 /4m, with a mean c.v. of 97% (Table 2,Hill, 1979). There was no significant difference in the mean lengths of the trans-cription units in any of the 4 categories studied (Table 2).

The spacing of the polymerase granules in Type III transcription units was70 + 40 nm (mean c.v. = 57%), a value significantly lower than the one calculated forType I transcription units (P < °'°5)- There was no significant difference in thespacing of polymerases found in Type II, Type III and Dumont stage 3 transcriptionunits (Table 3). The mean value for the slopes of the regression lines of Type IIItranscription units was 0-2439 ±0-0711, indicating that the transcripts are less fore-shortened than those found in Type I and II units (Table 1). The packing ratio forType III transcripts was 4-1. However, the Type III transcripts were more foreshort-ened than those found in transcription units from Dumont stage 3 oocytes, where themean slope of the regression lines was 0-4728 + 0-0493 (Table 1).

Fig. 10. Type III transcription unit. The transcripts are very closely packed andappear less foreshortened. Twenty transcripts were measured for this unit, giving acorrelation coefficient of 08655 and a regression slope of 02257. Scale bar = 1 /tm.Fig. 11. Transcription unit from a Dumont stage 3 oocyte. Note the closely packedRNP transcripts, which are comparatively long and complex. Scale bar = 1 /tm.

Transcription in early diplotene oocytes

..-. ,\.

98 R. S. Hill and H. C. Macgregor

The beads which form the Type III transcripts have a mean diameter of 29-6 +6-2 nm and a centre-to-centre spacing of 38-0 ± 67 nm (Fig. 10). These values do notdiffer significantly from those of transcripts from Type I and II transcription units.

DISCUSSION

Four main conclusions can be drawn from this study. First, during late pachyteneand very early diplotene in X. laevis, at a time when the large mass of extrachromo-somal nucleolar DNA is beginning to disperse, there is very little, if any, transcriptionof chromosomal DNA. Oocyte nuclei at this stage are between 10 and 20 /jm indiameter.

Secondly, chromosomal transcription units, comparable in length to those that arefound in fully formed lampbrush chromosomes are present in early diplotene nucleiof 25-40/im in diameter. They are sparsely covered with RNA polymerases andassociated RNP transcripts. They may represent the earliest loops to emerge, inwhich case one might suppose that the overall length of a transcriptional complex thatgives rise to a definitive lampbrush loop is not dependent upon the rate of spinningout or unfolding of axial DNA, nor upon the density of packing of RNA polymerases;it is a property that exists right at the inception, with well-defined and permanentstart and stop signals on the axial DNA strand.

The third conclusion is that the intensive transcriptional activity that is charac-teristic of a fully formed lampbrush loop is attained by an increase in the frequencyand regularity of polymerase attachment to the axial DNP fibre, although otherfactors such as change in the rate of polymerase movement and RNA transcriptelongation may also be significant. By the time germinal vesicles of X. laevis are50-60 /tm in diameter, transcription units already have the dense and regular patternof polymerase spacing that is found in definitive lampbrush material from muchlarger oocytes. It seems therefore that lampbrush loops form quickly and early inoogenesis, and that they become visible following a quite dramatic increase in theintensity of transcriptional activity and the amount of RNP transcript clothing theloop axis.

A corresponding, but later and more gradual increase in the regularity and densityof polymerase spacing has been described for extrachromosomal ribosomal DNA inX. laevis by Scheer, Trendelenburg & Franke (19766). They report that the densityof polymerase packing in ribosomal transcription units is about 3-5 times as highduring vitellogenesis as in previtellogenesis, and the distribution of polymerasesalong the DNP axis is also said to be more regular in material from vitellogenicoocytes.

The last conclusion, derived from studies of the slopes of regression lines forindividual transcription units, is that RNP transcripts are more foreshortened, by afactor of about 3:1 (Table 3) in smaller oocytes than they are in larger ones. It isnot known whether this is a consequence of differences in RNA/protein packing ordifferent patterns of processing and loss of transcript, but it is certainly a feature thatdeserves further investigation. If processing of transcripts does occur while they are

Transcription in early diplotene oocytes 99

still attached to the DNP axis, then it must be happening throughout the length ofthe transcription unit.

Foreshortening of RNP transcripts of the kind that has been observed is remarkablefor the following reasons. The most foreshortened transcripts are those that are mostwidely separated from their neighbours by stretches of nucleosomal DNA. Yet if oneassumes that the packing ratio of nucleosomal DNA is higher than that of activelytranscribing chromatin, and the length of a transcript is proportional to the length oftemplate that precedes it in the unit, then the more sparse the transcripts, the longerthey should be in relation to the preceding stretch of chromatin. In fact, the oppositeholds, so that foreshortening of sparse, early transcripts must be even greater thanthe linear regression data indicate.

On the basis of these conclusions the following scheme is proposed for the formationof an average lampbrush loop. First, the higher-order structure of the length of DNAthat is to form the loop unwinds. This probably involves an unravelling of the20-30-nm fibre (Carpenter, Baldwin, Bradbury & Ibel, 1976; Finch & Klug, 1976;Olins, 1977) into a more linear form such as the 10-nm fibre proposed by Worcel(1977). The accompanying alteration of the arrangement of nucleosomes allows RNApolymerase molecules to attach to the DNP fibre at specific initiation sites. Eachpolymerase then reads right through the entire transcription unit from start to finish.At this stage, loops are probably not identifiable as such, since most of the transcrip-tion unit is still no more than a nucleosomal fibre. The level of polymerase initiationthen increases, and the transcription unit rapidly becomes clothed in RNP transcriptsall along its length; when this happened one would expect the transcription unit tolook like a lampbrush loop if it were possible to examine it with a phase-contrastmicroscope. Ignoring for the moment the real possibility of stage-specific emergenceof transcription units and loops, it seems likely that sufficient transcription units areactive and saturated with polymerases for large numbers of definitive lampbrush loopsto be visible in all germinal vesicles of more than 50 /tm diameter in X. laevis.

Three matters now need special attention. First, for the sake of simplicity, 1transcription unit has been regarded as equivalent to 1 loop, but this is not always thecase. Some loops must include several transcription units, tandemly arranged and oflike or unlike polarities (Callan & Lloyd, i960; Scheer, Franke, Trendelenburg &Spring, 1976a). It would be interesting to know if all the transcription units in amultiple loop are switched on simultaneously, since this would give added strength tothe concept that adjacent units have related functions.

Next, Table 1 shows that the average lengths of transcription units in the ' formativephase' are the same as in later stages, notwithstanding the sparse distribution ofpolymerases in these early units. However, if indeed the transcription units are of thesame real average length with respect to axial DNA in all stages, then one mightexpect the observed length of units in the formative phase to be substantially shorterthan in later stages, because in the formative phase most of the unit is nucleosomal,whereas later on most of it is transcribing. The implication is that the lengths of thetranscription units that have been estimated by linear regression analysis in the forma-tive phase are too great. If this is so, then it may reflect a bias towards selection for

ioo R. S. Hill and H. C. Macgregor

measurement of the longest transcription units, since these would be the only onesthat had sufficient transcripts associated with them to make linear regression analysispossible. Shorter units that had one, two, or perhaps even no transcript associatedwith them at certain times may not be represented in the sample. On the other hand,it is not known to what extent the modification of the nucleosome core by polymeraseschanges the observable length of a DNP fibre. It is known that both histone and non-histone proteins remain associated with the DNA while it is being transcribed,suggesting that some compaction of the DNA is retained (Gottesfeld, Murphy &Bonner, 1975; Levy & Dixon, 1978).

The final matter that needs mentioning concerns a paradox that has led to someconfusion in recent literature on amphibian oogenesis. The essence of the paradoxis summed up in a statement made by Davidson in one of his recent reviews on thesubject (1976) where he says that 'in amphibians the peak synthetic activity andextension of lampbrush chromosomes occurs after the maternal message stockpileseems to be largely accumulated'. The basis of the paradox is the classification ofX. laevis oocytes introduced by Dumont in 1972. Following his light- and electron-microscope studies of sectioned ovarian material, Dumont published the followingscheme. Stage 1 oocytes are previtellogenic, 50—300/im in diameter, and range fromzygotene through to very early diplotene. Stage 2 oocytes are vitellogenic, 300-450 /tmin diameter, and are in diplotene with early lampbrush chromosomes. Stage 3oocytes are 450-600 fim in diameter, in mid-diplotene with fully formed lampbrushchromosomes. It is certainly true that well-developed lampbrush chromosomes arefound in Dumont stage 3 oocytes. However, unpublished observations (Hill) and thepresent study indicate that they are also present in Dumont stage 1 and 2 oocytes,and that the lampbrush type of transcription begins much earlier, in oocytes of lessthan 100 /tm diameter that might be said to be in early Dumont stage 1. Consequently,the findings of many biochemical investigators who have reported the presence ofpoly-(A)+RNA, translatable messenger RNA, and 4s and 5s RNA in previtellogenicDumont stage 1 oocytes (Rosbash & Ford, 1974; Ford, 1971; Thomas, 1974;Darnbrough & Ford, 1976; Ruderman & Pardue, 1977) are entirely reasonable.Essentially, lampbrushes seem to be active much earlier than we had hithertosupposed.

This research was supported by Science Research Council grants nos. GR/A/26417,GR/A/32128 and GR/A/86466. The authors would like to thank Professor F. Beck andDr D. Pallot for the use of their Kontron digitizer; Dr J. Bullock for his advice on statisticalmethods, and Mrs H. Horner for her technical assistance.

REFERENCES

CALLAN, H. G. & LLOYD, L. (i960). Lampbrush chromosomes of crested newts, Trituruscrislatus. Phil. Trans. R. Soc. B 243, 135-219.

CARPENTER, B. G., BALDWIN, J. P., BRADBURY, E. M. & IBEL, K. (1976). Organization of sub-units in chromatin. Nucleic Acids Res. 3, 1739—1746.

DARNBROUGH, C. & FORD, P. J. (1976). Cell-free translation of messenger RNA from oocytesof Xenopus laevis. Devi Biol. 50, 285-301.

Transcription in early diplotene oocytes 101

DAVIDSON, E. H. (1976). Gene Activity in Early Development, 2nd edn. New York & London:Academic Press.

DUMONT, J. (1972). Oogenesis in Xenopus laevis (Daudin). I. Stages of oocyte development inlaboratory maintained animals. J. Morph. 136, 153-179.

EPPIG, J. J. & DUMONT, J. (1976). A defined nutrient medium for the in vitro maintenance ofXsnopus laevis oocytes. In Vitro 12, 418-427.

FINCH, J. T. & KLUG, A. (1976). Solenoidal model for superstructure in chromatin. Proc.natn. Acad. Sci. U.S.A. 73, 1897-1901.

FORD, P. J. (1971). Non-coordinated accumulation and synthesis of 5S ribonucleic acid byovaries of Xenopus laevis. Nature, Land. 233, 561-563.

GALL, J. G. (1954). Lampbrush chromosomes from oocyte nuclei of the newt. J. Morph. 94,283-352.

GOTTESFELD, J. M., MURPHY, R. F. & BONNER, J. (197S). Structure of transcriptionally activechromatin. Proc. natn. Acad. Sci. U.S.A. 72, 4404-4408.

HILL, R. S. (1979). A quantitative electron-microscope analysis of chromatin from Xenopuslaevis lampbrush chromosomes. J. Cell Sci. 40, 145-169.

LAIRD, C. D., WILKINSON, L. E., FOE, V. E. & CHOOI, W. Y. (1976). Analysis of chromatinassociated fibre arrays. Chromosoma 58, 169-190.

LEVY, B. W. & DIXON, G. H. (1978). A subset of nucleosom.es in trout testis chromatin con-taining non-histone proteins and enriched in transcribed DNA sequences. Fedn Proc. FednAm. Socs exp. Biol. 37, 1787a.

MACCREGOR, H. C. (1968). Nucleolar DNA in oocytes of Xenopus laevis. J. CellSci. 3, 437-444.MACGREGOR, H. C. (1980). Recent developments in the study of lampbrush chromosomes.

Heredity (In Press).MILLER, O. L. & BAKKEN, A. H. (1972). Morphological studies of transcription. Acta endocr.,

Copenh. (Suppl.) 168, 155-177.OLINS, A. L. (1977). v bodies are close-packed in chromatin fibres. Cold Spring Harb. Symp.

quant. Biol. 42, 325-329.ROSBASH, M. & FORD, P. J. (1974). Polyadenylic acid containing RNA in Xenopus laevis

oocytes. J. molec. Biol. 85, 87-101.RUDERMAN, J. V. & PARDUE, M. L. (1977). Cell-free translation analysis of messenger RNA in

echinoderm and amphibian early development. Devi Biol. 60, 48-68.SCHEER, U., FRANKS, W. W., TRENDELENBURG, M. F. & SPRING, H. (1976a). Classification of

loops of lampbrush chromosomes according to the arrangement of transcriprional complexes.J. Cell Sci. 22, 503-520.

SCHEER, U., TRENDELENBURG, M. F. & FRANKE, W. W. (1976 b). Regulation of transcriptionof genes of ribosomal RNA during amphibian oogenesis. A biochemical and morphologicalstudy. J. Cell Biol. 69, 465-489.

SOMMERVILLE, J. (1977). Gene activity in the lampbrush chromosomes of amphibian oocytes.In The International Review of Biochemistry: Biochemistry of Cell Differentiation, 11, vol. 15(ed. J. Paul), pp. 79-156. Baltimore: University Park Press.

THOMAS, C. (1974). RNA metabolism in previtellogenic oocytes of Xenopus laevis. Devi Biol.39. 191-197-

WORCEL, A. (1977). Molecular architecture of the chromatin fibre. Cold Spring Harb. Symp.quant. Biol. 42, 313-324-

(Received 19 February 1980)