ABSTRACT
The primordial germ cells and the gonadal mesoderm were mapped in the Drosophila embryo by analyzing the patterns of mosaicism in ‘normal’ and ‘transformed’ gynandromorphs. Relative to the adult cuticular markers the germ cells map as the posterior most structure, which coincides with their known location in the blastoderm embryo. These data support the hypothesis that the gynandromorph map reflects the real position of the pri-mordia in the embryo. Since after the blastoderm stage the primordial germ cells migrate anteriorly these data also indicate that the map in fact corresponds to the blastoderm stage and not to a later stage of development. The genital disc maps as a single median primordium anterior and ventral to the germ cells, the gonadal mesoderm is located anterior to the genital disc and also forms a single median primordium on the ventral side of the embryo. The primordia for the genital disc and the gonadal mesoderm are unusually large in size, which presumably reflects some indeterminacy of the cell lineage leading to an ‘expansion’ of the map.
INTRODUCTION
The location, of the primordial cells for the various larval and adult structures in the early Drosophila embryo can be deduced from an analysis of gynandromorphs and other genetic mosaics (Sturtevant, 1929). The gynandromorphs generally used in these studies arise from female zygotes in which one of the two X chromosomes is lost during the first nuclear division, giving rise to flies which are mosaics of approximately 50 % female (XX) and 50 % male (XO) cells (Hinton, 1955; Lifschytz & Falk, 1969). Using appropriate recessive markers on the X chromosome which is retained in the XO cells, allows identification of these cells in any kind of tissue (Janning, 1972; Hall, Gelbart & Kankel, 1975).
In the adult gynandromorphs the male and female cells are not freely intermingled, but instead the cells of one genotype are clustered together and cover large continuous areas of the epidermis. The orientation of the genotypic boundary between XX and XO cells is largely random. The probability of this boundary passing between two cells at the blastoderm stage, when the embryo consists essentially of a monolayer of cells, is proportional to the (arc) distance between them (Sturtevant, 1929; Garcia-Bellido & Merriam, 1969). Since the location of a cell within the blastoderm largely decides upon its fate, the frequency of the genotypic boundary passing between two adult cells, can be taken as a measure for the (arc) distance between their primordial cells in the blastoderm (Hotta & Benzer, 1972, 1973). By calculating such distances and triangulation, several fate maps of this kind have been constructed (Garcia-Bellido & Merriam, 1969; Hotta & Benzer, 1972, 1973; Ripoll, 1972; Janning, 1974b; Gelbart, 1974; Baker, 1975). Such gyandromorph maps are generally in agreement with the available embryological data (Poulson, 1965), but in most cases direct evidence on the location of the primordial cells is scarce.
One cell type which provides an excellent test system for the mapping techniques is the pole cell. The pole cells can easily be identified as a group of round cells located at the posterior pole of the egg, outside the monolayer of blastoderm cells, and therefore, they should map as the most posterior structure. At the time of their formation the pole cells become determined to form the primordial germ cells (Illmensee & Mahowald, 1974) and possibly the cuprophilic cells of the midgut (Poulson & Waterhouse, 1960). In the course of development they migrate into the interior of the embryo and associate with mesodermal cells to form the gonad.
Gynandromorph mapping of the gonad gave conflicting results: Falk, Orevi & Menzel (1973) obtained a median and dorsal map position for the larval gonad. Using different reference structures, Hotta & Benzer (1973) mapped it close to the posterior end. A posterior location was also found for the adult gonad (Janning, 1974b). In any case, the map position of the gonad as determined in these studies most probably reflects the location of the primordial cells for the gonadal mesoderm rather than the pole cells. Therefore, we decided to map the two components of the gonad, germ cells and mesoderm, separately.
In the course of the present study we found that the gonads in gynandromorphs frequently lack mature germ cells. This suggested to us that in these cases the primordial germ cells were of the opposite sex to the gonad mesoderm, and therefore, they could not differentiate into mature germ cells. Meanwhile, this hypothesis has been tested directly by pole cell transplantation. It has been found that heterosexual transplantations of pole cells are unsuccessful, whereas homosexual transfers frequently lead to the formation of fertile gametes derived from the donor pole cells (Illmensee, personal communication; Van Deusen & Gehring, in preparation). Therefore, the presence or absence of germ cells in a gonad of a gynandromorph indicates whether or not the genotype of the germ cells differs from that of the gonadal mesoderm. This gives us a method to map the germ cells.
The mapping data obtained in this way were confirmed by analyzing ‘transformed gynandromorphs’ (Novitski, 1951; Seidel, 1963) which are homozygous for the recessive mutant transformer, tra (Sturtevant, 1945; Lindsley & Grell, 1968), Since tra transforms XX females into sterile males, such gynandromorphs look like male mosaics. Because they are all male, the morphology of the sexual apparatus is normal and easier to interpret. Using the white marker (see Lindsley & Grell, 1968) the genotype of the mesoderm can be recognized by the coloration of the testicular sheath. Since XX tra/tra pole cells cannot form sperm (Brown & King, 1961; Seidel, 1963), whereas XO tra/tra pole cells are able to differentiate into mature sperm (Seidel, 1963) the presence or absence of mature sperm is an indication of the genotype of the germ cells. Using both ‘normal’ and ‘transformed’ gynandromorphs very similar fate maps were obtained.
METHODS
The gynandromorphs used in this paper were obtained from zygotes of the following genotypes : In(1)wvC/y w /36a (‘normal’ gynandromorph) and In(1)wvC/ ywf36; tra/tra (‘transformed’ gynandromorphs) by loss of the unstable ring-X chromosome, In(1)wvC, during the first nuclear divisions (Hall, et al. 1975). Thus, the XO-male areas were marked with y, for yellow bristles and chitin colour, f36a, for forked bristles, and w (white) which affects the eye colour and also the colour of the testis sheath, the vasa deferentia and the Malphighian tubules (Lindsley & Grell, 1968).
About 30 % of the flies carrying the ring-X chromosome were gynandromorphs. The stocks were maintained at 25 °C on standard food and the adult flies aged for 3 days before examination, to ensure complete pigmentation of the testis sheaths. The flies were examined for external mosaicism under a dissecting microscope and the distribution of male and female tissue recorded on drawings. Subsequently, the gynandromorphs were dissected and drawings were made of the interior genitalia and the gonads. The presence or absence of germ cells in the gonads was checked under a compound microscope if necessary.
From both kinds of gynandromorphs, ‘normal’ and ‘transformed’, 209 specimens each were examined. The gynandromorphs were selected on the basis of external mosaicism. A sample of females showing no XO tissue in their cuticle was also dissected, but no mosaicism affecting the internal genitalia or the gonads was detected. As a further control 100 XO males were dissected, and no internal mosaicism was found. Of the 209 ‘normal’ gynandromorphs 109 were selected on the basis of abdominal mosaicism. This leads to a slight increase in all the sturt distances of the fate map. However, it does not significantly influence the results, since we are interested in relative rather than absolute distances of abdominal structures. If the chromosome elimination occurs during the first nuclear division, which is randomly oriented, and if the two daughter nuclei divide at the same rate, one would expect a given structure to be male in. 50 % of the gynandromorphs. For ‘normal’ gynandromorphs we found an average of 49 %, for ‘transformed’ a somewhat lower value of 39 % was obtained.
RESULTS
In gynandromorphs, the probability that the randomly oriented genotypic boundary passes between two adult structures, can be taken as a measure for their distance in. the embryonic fate map (Sturtevant, 1929; Garcia-Bellido & Merriam, 1969; Hotta & Benzer, 1972). A probability of 1 % is designated as 1 sturt unit (Hotta & Benzer, 1972). The germ cells and gonadal mesoderm were mapped relative to the abdominal tergites and sternites, each of which is formed by a group of histoblasts, and relative to the genitalia which are derived from the genital disc. All these structures are multicellular, but they originate from separate primordia. Table 1 gives the calculated distances in sturts for a given pair of structures in both ‘normal’ and ‘transformed’ gynandromorphs. These data can be summarized as follows :
Sturt distances between the germ cells, gonadal mesoderm and adult cuticular structures of the abdomen for ‘normal’ and ‘transformed’ gynandromorphs

The distance between the germ cells (GE) and the sternites (S) or the tergites (T) decreased with increasing segment number. The closest cuticular markers are the genitalia (GD) which map posterior to the last tergite and sternite. Therefore, the germ cells map as the posterior most of these structures.
In the ‘normal’ gynandromorphs GE consistently maps closer to S than to T for a given segment, but the difference is small. In ‘transformed’ gynandromorphs the GE-S distances are approximately equal to the GE-T values for a given segment, which indicates that GE is located in the median plane or slightly ventral to it.
The gonadal mesoderm (GN) also maps farthest from the first and closest to the last abdominal segment, but in this case the distances between S and GN are always considerably shorter than the distances relative to T. Therefore, the mesoderm maps close to the posterior end, on the ventral side.
- Fate maps derived from these data by triangulation are shown in Fig. 1. The triangulation method gives slightly ambiguous maps since the map distances are not strictly additive, but even when different reference points are used, the relative positions of GE and GN are the same: GE is the posterior most marker, GN is more ventral and GD is intermediate, between GE and GN (shaded areas in Fig. 1).Fig. 1.
Fate map of the germ cells, gonadal mesoderm, and cuticular structures of the abdomen derived from the analysis of (A) ‘normal’, and (B) ‘transformed’ gynandromorphs (lateral view). This map was constructed by first triangulating the cuticular structures relative to their nearest neighbour. Subsequently, the germ cells (GE) and the gonadal mesoderm (GN) were mapped relative to the three nearest cuticular markers. Depending upon which pair of cuticular markers are used for triangulation three different map positions are obtained (shaded area). For abbreviations see Table 1.
Fig. 1.Fate map of the germ cells, gonadal mesoderm, and cuticular structures of the abdomen derived from the analysis of (A) ‘normal’, and (B) ‘transformed’ gynandromorphs (lateral view). This map was constructed by first triangulating the cuticular structures relative to their nearest neighbour. Subsequently, the germ cells (GE) and the gonadal mesoderm (GN) were mapped relative to the three nearest cuticular markers. Depending upon which pair of cuticular markers are used for triangulation three different map positions are obtained (shaded area). For abbreviations see Table 1.
Information about the relative distance from the ventral or dorsal midline can be obtained by calculating left-right distances. For a pair of bilaterally symmetrical structures, the left-right distance represents twice the (arc) distance to the midline (Hotta & Benzer, 1972). In accordance with previous studies (Hotta & Benzer, 1973) we found that the left-right distances varied between 36 –45 sturts for the tergites and between 42 –50 sturts for the sternites, the last sternite and tergite giving the smallest value (Table 2). In contrast to these large left-right distances for S and T, the values for GD, GN and GE are considerably smaller ranging from 10 –13 sturts, indicating that they map very close to the midline (5 –7 sturts). This suggests that there is a single primordium near the ventral midline rather than a symmetrical pair of separate primordia. This assumption was tested further by examining the genotypic boundary separating the structures from left to right, and asking how frequently this boundary passes precisely between the left and right structure without cutting into either of the two structures. As indicated in Table 2 the frequency of perfect left-right division is high for S and T but extremely low for GD, GN and GE. Since the frequency of perfect division is related to the number of cells separating two given primordia (Wieschaus & Gehring, 1976 b) these data strongly support the assumption that there is a single median primordium for each of these three structures.
Left-right distances, frequency of perfect left-right division, and frequency of mosaicism within a given structure

For a given structure, the frequency of mosaicism is a measure for the relative size of its primordium or the number of primordial cells. The larger a primordium, the higher the probability that the genotypic boundary crosses it. Assuming a circular primordium, the radius (r) of the primordium equals the frequency of mosaicism (f) divided by π (Hotta & Benzer, 1973). For the tergites and sternites the frequency of mosaicism is smaller than 11 % (Table 2), indicating relatively small primordia of less than 7 sturts in diameter. In contrast, GD, GN and GE show much higher frequencies of mosaicism. Especially the frequencies for the gonadal mesoderm (37 %) and the genitalia (21 %) are high, which suggests unusually large primordia. Alternatively, it could be assumed that the lineage of these cells is not yet determined at the blastoderm stage, leading to an ‘expansion’ of the map (see discussion), if some of the cells within the primordium give rise to other structures as well.
DISCUSSION
Different criteria for mapping which have been used in this study need some evaluation. Autonomous recessive marker genes were available for the cuticular structures (y w/36a) which allows an unambiguous identification. In the ‘normal’ gynandromorphs the sexual phenotype was used as a marker for the inner genitalia and as an additional marker for the outer ones. Since the genitalia differentiate autonomously (Kroeger, 1959; Nöthiger, 1964) this is justified. The sex of the gonad was used as a marker for the gonadal mesoderm. In the testicular sheath the w+ marker can be used to identify XX female cells, which are capable of non-autonomous differentiation into phenotypically male cells (Dobzhansky, 1932; Janning, 1974a). The testicular sheath cells also cover the vasa deferentia (Stern & Hadorn, 1939). No such marker was available for the ovarian sheath. The germ cells were mapped on the basis of the presence or absence of mature germ cells. It is more precise to say that we mapped the site for the ability to differentiate germ cells. However, we have shown by pole cell transplantation, that the pole cells differentiate autonomously according to their chromosomal constitution and are incapable of differentiating into germ cells of the opposite sex (Van Deusen & Gehring, in preparation). Therefore, the site for the ability to form mature germ cells coincides with the location of the pole cells, provided that the germ cells are surrounded by a normal mesoderm. In all cases where the sexual phenotype was used as a marker, small mosaic areas are likely to be missed since a critical cell number may be needed for a given structure to differentiate and to be identified as being male or female. In our sample of gynandromorphs missing structures were relatively rare: only one animal had no gonads, and a single gonad was present in only 3 % of the cases.
The use of ‘transformed’ gynandromorphs has several advantages. Since they are all male in their somatic tissues, their sexual apparatus is normal and easier to interpret than normal gynandromorphs which are a complex mixture of male and female parts. All gonads are testes and the genotype of the mesoderm sheath cells can be recognized by their colour, XX showing yellow colour and XO cells being white. The chromosomal mosaicism can clearly be distinguished from white-variegation due to the wvC marker. The white-variegation results in a ‘salt and pepper’ type intermingling of phenotypically w and w+ cells, whereas the chromosomal mosaicism results in the formation of large patches of tissue of one genotype. XX tra/tra pole cells are incapable of forming mature sperm (Brown & King, 1961; Seidel, 1963). XO tra/tra germ cells give rise to mature sperm which, however, is immotile due to the absence of a Y chromosome (Seidel, 1963). Therefore, the presence or absence of sperm indicates the genotype of the pole cells. The use of transformed gynandromorphs also has some disadvantages, since the internal genitalia, which are all male and unmarked, cannot be mapped. Also the male lacks the 7th tergite and both the 6th and 7th sternites which makes mapping of the posterior structures more difficult. Nevertheless, the two methods together give consistent mapping data.
The pole cells are of particular interest since it has been shown by cytoplasmic transplantation experiments that there are ‘determinative factors’ in the egg cytoplasm near the posterior pole which induce the formation of pole cells (Illmensee & Mahowald, 1974), providing the best experimental evidence for cytoplasmic localization. The present study clearly shows that the germ cells map as the posterior most structure, which coincides with the position of the pole cells at the blastoderm stage. Therefore, there is no need to invoke a mechanism of determination operating prior to the blastoderm stage as suggested by Falk et al. (1973). Since immediately after blastoderm formation, the pole cells migrate anteriorly and penetrate into the embryo, this map location provides some direct evidence that the map does in fact relate to the blastoderm stage and not to a later stage of development. Furthermore, it indicates that the mapping data correlate well with the actual position of the primordial cells.
The gonadal mesoderm, the genital disc, and the pole cells each map as separate primordia along the ventral midline in this anterior-posterior order, behind the last abdominal sternites and tergites. A surprisingly high frequency of mosaicism was found for the gonadal mesoderm (37 %) as compared to less than 4 % for the sternites and less than 11 % for the tergites. The value for the genital disc (21 %) also appears very high, since it is known from histological studies that the disc originates from a very small number of cells (Laugé, 1967). The large frequencies of mosaicism for these structures can be explained by assuming that the cell lineage of the blastoderm cells in those regions is not strictly fixed and they can give rise to structures other than the genitalia. As pointed out clearly by Hotta & Benzer (1973), an indeterminacy of the cell lineage causes an expansion of the map. Such a map expansion has been found in ‘fine-structure’ maps of the leg discs (Wieschaus & Gehring, 1976a). The data presented above suggest that map expansion also occurs over short distances between separate primordia.
The gynandromorph mapping technique is based upon the assumption that the location of a cell within the blastoderm largely decides upon its fate, but is not known how precisely the location correlates with the fate. The fact that a self-consistent map can be constructed confirms in a general way the validity of this assumption. There is some direct evidence on this point, since we have shown by dissociation and reaggregation experiments that cells from the anterior region of the blastoderm can produce anterior adult structures but not posterior ones, and vice versa (Chan & Gehring, 1971). This was confirmed by transplantation of single blastoderm cells (Illmensee, unpublished). However, recent experiments indicate that at least some blastoderm cells can still give rise to cells from two kinds of imaginal discs and are not yet disc-specifically determined (Wieschaus & Gehring, 1976b). Such an indeterminacy of the cell lineage is expected to lead to a map expansion. The map expansion found for the genital disc and the gonadal mesoderm suggests that some indeterminacy of the cell lineage also exists for these structures.
ZUSAMMENFASSUNG
Die Mosaikmuster von ‘normalen’ und ‘transformierten’ Gynandern wurden analysiert und die Lage der Urkeimzellen und des Gonaden-Mesoderms im Anlageplan bestimmt. Im Bezug auf die adulten Kutikularstrukturen kartieren die Urkeimzellen als kaudalste Struktur im Anlageplan, was ihrer Position am Hinterpol des Blastoderm-Embryos entspricht. Dieser Befund bestätigt die Hypothese, dass die Gynanderkarte die wirkliche Lage der Primoridalzellen im Embryo angibt, und dass diese Karte dem Blastodermstadium entspricht. Die Primordialzellen für die Genitalscheibe und das Gonadenmesoderm liegen ventral in der Medianebene vor den Urkeimzellen. Die Anlagen für die Genitalscheibe und das Gonadenmesoderm umfassen relativ zu grosse Areale, was daraufhin deutet, dass diese Primordialzellen auf dem Blastodermstadium noch nicht determiniert sind, oder dass noch Primordialzellen für andere Strukturen in diesen Arealen lokalisiert sind.