Anterior determinants in embryos of Chironomus samoensis: characterization by rescue bioassay

The use of bioassays based on either ‘rescue’ or heterotopic transplantation has been instrumental in the analysis of localized cytoplasmic determinants, which are an almost universal feature of early animal development (Davidson, 1986). Many rescue bioassays use embryos that develop abnormally as a result of a maternal effect mutation. Such embryos can often be rescued, i.e. restored to normal development, by transplanting cytoplasm, RNA or molecular fractions from wild-type donors. This strategy is limited to organisms with established genetics and to the analysis of gene products for which suitable mutants already exist. Alternatively, embryos can be programmed for abnormal development by experimental manipulation. Although the resulting lesion is usually less specific than a genetic mutation, this approach can still be useful for identifying the limiting factor to normal development in the manipulated embryo.

Chironomid embryos respond to various experimental treatments with the formation of dramatically abnormal body patterns including double cephalons and double abdomens (see Sander, 1976; Yajima, 1983; Kalthoff, 1983). Double cephalons show a mirror-image duplication of cephalic structures, including mouthparts, eyes and stomodaeum, whereas thorax and abdomen are missing. In double abdomens, anterior segments are replaced by posterior segments with reversed anteroposterior polarity. Most double abdomens are symmetrical (except for the gonads), with a plane of symmetry in the third abdominal segment. There are also asymmetrical double abdomens where the posterior set of segments extends from the abdominal terminus to a thoracic or even cephalic segment while the anterior set of segments is shorter and may be limited to the terminal abdominal segment. Similar pattern abnormalities have been observed in Drosophila dicephalic and bicaudal phenotypes, as well as in other dipteran species (see Percy et al. 1986). The similarities suggest that there may be a measure of generality in the formal program, and perhaps the molecules, involved in specifying the anteroposterior body pattern.

The common denominator of the experimental treatments causing double abdomen formation in chironomid embryos appears to be the displacement, or inactivation, of cytoplasmic components required for the development of anterior segments. These components are referred to as anterior determinants. In an extended series of experiments using embryos of Smittia sp., anterior determinants were previously characterized as cytoplasmic RNP particles that are localized predominantly, but not exclusively, near the anterior egg pole (see Kalthoff, 1983). This characterization was based on a large body of indirect evidence, whereas the results reported here directly prove and extend our earlier conclusions. The new results were obtained with a rescue bioassay using embryos of Chironomus samoensis. Preliminary data have been published as a conference paper (Kalthoff & Elbetieha, 1986). Here we report more fully on the localization of anterior determinant activity, its association with poly(A)⁺RNA of a surprisingly small size, its developmental control and its function under conditions of heterospecific transplantation.

Most of the donor embryos used as sources of cytoplasm or as starting material for preparing RNA were collected from a laboratory culture of Chironomus samoensis also used in previous studies (Percy et al. 1986; Kalthoff & Elbetieha, 1986; Kuhn et al. 1987). Chironomus embryos used as donors were always derived from the N strain in which spontaneous double abdomen formation is infrequent (see Kuhn et al. 1987). Also used as donors were embryos from Drosophila melanogaster and from our laboratory culture of Smittia sp. (Ripley & Kalthoff, 1983).

Recipient embryos from our Chironomus samoensis culture were programmed for double abdomen formation by irradiation with ultraviolet light (u.v.). For this purpose,embryos were lined up over the exit slit of a monochromator with the anterior pole facing the u.v. beam (Ripley & Kalthoff, 1981). The irradiation conditions (285nm, 300 –600 J m⁻²) were gauged so that 50 –90% of the irradiated embryos developed the double abdomen pattern, while the remaining embryos developed normally. This was important for an adequate quantification of the rescue effect observed, i.e. the increase in the frequency of normal embryos at the expense of double abdomens. With less than 50% double abdomens prior to rescue, potential increases in the frequency of normal embryos of more than 50 % would have gone undetected. With more than 90% double abdomens before rescue, rescue effects would have been difficult to observe in small samples of embryos. To appreciate this limitation, rescue effects are best viewed as equivalent to a reduction of the u.v. dose applied for inducing double abdomen formation. In the exponential ‘tail’ of the u.v. dose-response curve (Kalthoff, 1971), even considerable dose reductions are reflected only by small reductions of the biological effect, in this case the frequency of double abdomens. Due to a cluster-dependent variability in the response to u.v. irradiation (Kalthoff & Elbetieha, 1986), the double abdomen frequency in u.v.-irradiated but uninjected control samples were sometimes higher or lower than desired. Such experiments were excluded when the results were tallied.

The Chironomus embryos used as recipients were preferably taken from our sda strain in which double abdomens are formed spontaneously, although with variable frequency, in about half of the clusters. (Chironomus females lay clusters of several hundred eggs, which are enclosed in a gelatinous matrix and develop in synchrony.) Embryos from the sda strain need a lower u.v. dose to induce the desired frequency of double abdomens (Kuhn et al. 1987). Presumably because unspecific u.v. damage was correspondingly less severe, these embryos tolerated the injection procedure better and could be scored more easily. However, in the absence of embryos from the sda strain, recipients were also taken from the N strain with no noticeable difference in the results. The developmental stages of Chironomus and Smittia were identified under the stereomicroscope using criteria described earlier (Kalthoff & Sander, 1968; Ripley & Kalthoff, 1983). The following stages are referred to: P_o (no pole cells), P₂ (two pole cells); P₄ (four pole cells); M₂ (late nuclear migration); Bl (cellular blastoderm); GA (germ anlage); GB (segmented germ band). Stages P₀ –P₄ probably correspond to the first five nuclear division cycles, stage M₂ to cycle 9, and stage Bl to cycle 14. Drosophila embryos used as donors of cytoplasm were shortly before or after the pole cell formation stage. Xenopus oocytes used to prepare RNA were fully grown.

In a typical experiment, a batch of approximately 60 Chironomus embryos derived from one cluster was u.v. irradiated at stage P₄. All subsequent handling of u.v.-irradiated embryos, including microinjection, was carried out in filtered light to avoid uncontrolled photoreactivation. Shortly before the irradiated embryos reached stage M₂, they were subdivided into two or more experimental samples and an ‘uninjected control’ sample. The experimental samples were injected with different types of cytoplasm or RNA solutions to be tested for anterior determinant activity. In several experiments an additional sample was injected with buffer. This ‘buffer-injected control’ was introduced when we discovered that injection itself caused a somewhat variable increase in the frequency of double abdomens as compared to the uninjected controls. In order to eliminate a possible bias due to the progressing age of the recipients during the course of the injections, each experiment was repeated several times, and the sequence of the injected materials was permutated in a cyclical fashion. The results from such a series of experiments were pooled. Anterior determinant activity was quantified as an increase in the frequency of normal embryos among the total of identifiable germ bands.

About 24h after injection, most of the injected embryos could be classified as normal embryos, symmetrical double abdomens, asymmetrical double abdomens or undeveloped eggs. The normal embryos showed, beside their typical outline, a head capsule with labrum, mouth parts and eye spots. Symmetrical double abdomens showed two sets of abdominal segments joined in mirror-image symmetry with a proctodaeum, anal papillae, procercal setae and posterior prolegs on each end. In asymmetrical double abdomens, the posterior set of segments was longer and often included thoracic segments, while the anterior set was represented by one or a few abdominal segments with reverse polarity (see Percy et al. 1986). In some instances, extremely asymmetrical double abdomens were difficult to distinguish from normal embryos with head defects. Such specimens were allowed to hatch or were removed from the chorion with tungsten needles for close inspection. If they showed the normal series of abdominal and thoracic segments with head structures but no abdominal structures at the anterior end, they were scored as normal embryos. If they showed the typical set of terminal abdominal structures on both ends, they were scored as double abdomens. We wish to emphasize, however, that most double abdomens were symmetrical (Table 3) and unmistakable, and that most normal embryos seemed to be perfect in every respect. In fact, many of them hatched spontaneously. Embryos that could be classified unequivocally as either normal embryos or double abdomens were listed together as identifiable germ bands. The average percentage of identifiable germ bands was 74 % after injection of buffer, 71 % after injection of cytoplasm and 69 % after injection of RNA (Table 3).

Microinjections were performed with a hydraulic micromanipulator (Narishige M0-102N) mounted on an inverted microscope with Nomarski optics. Glass needles with an outer tip diameter of 2 –3 μm were pulled from 1 mm capillary tubes and bevelled under an angle of 25 –30 ° on a rotating diamond-coated plate (0 –0 ·25 μm). Bevelling needles to a sharp tip was critical to the survival of injected embryos. The bevelled tips were washed with ethanol. Needles were connected with tygon tubing to a 1 ml syringe with a micrometer-driven plunger. The whole injection system was filled with paraffin oil.

Donor and recipient embryos were immobilized on a microscopic coverslip to which a piece of thin fishing line had been glued. Chironomus and Smittia embryos were pipetted onto the coverslip in a drop of adhesive medium (0 ·3 % agarose, 0 ·2 % gelatin, kept fluid in a waterbath), in which they were lined up against, and parallel to, the fishing line. As the medium was allowed to evaporate, the embryos became stuck to the coverslip and the supporting fishing line. About 15 –20 s after the medium had dried, the embryos were covered with a drop of halocarbon oil (Halocarbon Products, Hackensack, NJ). The duration of the evaporation step needed to be tightly controlled and adjusted to the humidity in the injection room. Drosophila embryos used as donors were immobilized on a piece of double-stick tape and covered with halocarbon oil.

In experiments involving cytoplasmic transplantations, cytoplasm from one donor was transferred to one recipient with a glass needle. Aqueous solutions were loaded into the needle through the tip from small drops placed under halocarbon oil. Recipients were injected perpendicular to their long axis and close to the anterior pole, unless stated differently. The tip of the needle had to penetrate into the central cytoplasm of the recipients in order to achieve an even distribution of the injected material. The injected volume was approximately 4 pl (2 % of the Chironomus egg volume), unless stated differently.

After injection, each coverslip with embryos was transferred into a small plastic dish, to which 1 –2 ml of salt solution (0 ·56% NaCl, 0 ·24% KC1, 0 ·016% CaCl) was added so that the oil would float to the surface. Embryos were gently teased off the fishing line and were allowed to develop at room temperature in light-proof boxes.

RNA was extracted from batches of several thousand embryos which had been collected, removed from the cluster jelly and stored at –70°C. The frozen embryos were thawed by adding 4vol. homogenization buffer (0 ·125 m-Tris, pH 8, 1 ·25 % SDS) and centrifuged. The supernatant was digested with proteinase K, extracted twice with phenol/chloroform/isoamyl alcohol (25:24:1; v/v/v), once with chloroform and three times with ether. After precipitation with ethanol, the RNA was dissolved in diethyl pyrocarbonate (DEPC)-treated double-distilled water. The average yield of total RNA per Chironomus embryo was about 20 ng. Poly(A)⁺RNA was selected by oligo(dT)-cellulose chromatography (Maniatis et al. 1982); the average yield of poly(A)⁺RNA was 1 ·5% of the total RNA. For microinjection, RNA was dissolved in a buffer containing 25ITIM-KCl, 2 · 5mm-Pipes, pH7 · 3, 0 · 5mm-EDTA, 5mm-DDT and 2i.u. μ l^{− 1} RNasin.

To prepare RNA fractions enriched for certain size classes, RNA was either precipitated with 4M-LIC1 or separated on a sucrose gradient. The RNA size distributions of the LiCl supernatant and the LiCl precipitate were analysed by electrophoresis in 10% polyacrylamide gels containing 90mm-Tris (pH8 · 2), 90mm-boric acid, and 7M-urea. For sucrose gradient centrifugation, linear gradients (15 – 30 %) were prepared in buffer containing 0 · 5 mm-Tris-HCl pH 7 · 5, 0 · 1 m-NaCl, 1 mm-EDTA and 0 · 5 % SDS. Total RNA dissolved in DEPC-treated water was mixed with 1vol. of buffer (5 mm-Tris-HCl, pH 7 · 5, IITIm-EDTA 0 · 5% SDS), heated at 65 ° C for 2min, cooled on ice and layered on top of the gradient. After centrifugation in a vertical rotor, four fractions of approximately equal volume were collected by side puncture near the bottom of the tube. The size distributions of RNA fractions were determined by electrophoresis through 1 % agarose gels containing formaldehyde (Maniatis et al. 1982).

The purity and integrity of the RNA preparations were assayed by their translatability in a cell-free rabbit reticulocyte lysate system using procedures (Jackson & Hunt, 1983) and components kindly shared by our colleague, Dr M. Winkler. Incorporation of [³⁵S]methionine was measured by precipitation with trichloroacetic acid and compared to parallel reactions using globin mRNA or tobacco mosaic virus RNA as positive controls and water as negative controls. Translational products were analysed by electrophoresis on 12 · 5 % and 17 · 5 % polyacrylamide gels.

Chironomus embryos programmed for double abdomen development by anterior u.v. irradiation were microinjected with cytoplasm from unirradiated donors. In an initial series of experiments, the cytoplasm was taken from different donor regions and delivered to the anterior pole region of the recipients. The frequency of normal embryos developing among the surviving recipients depended on the donor site from which the cytoplasm was taken (Table 1). Anterior cytoplasm caused a significant increase (36%; P< 0 · 001) in the frequency of normal embryos as compared to the uninjected control group. A weaker but still significant (18%; P < 0 · 05) rescue effect was also observed after injection of middle cytoplasm, whereas no rescuing effect was obtained by injection of posterior cytoplasm. In fact, injection of posterior donor cytoplasm caused a decrease in the frequency of normal embryos, and a corresponding increase in double abdomens (20%; p <0 · 01). A similar but weaker (approximately 9 %) decrease of the frequency of normal embryos was observed after injection with buffer (Figs 2 and 5).

To test whether the rescuing activity of transplanted anterior cytoplasm depended on the injection site, recipients were injected anteriorly, in the middle or posteriorly (Table 1). As in the previous experiments, anterior donor cytoplasm had a strong rescue effect if delivered to the anterior pole of the recipients. Only a slight rescue effect was observed after injecting anterior cytoplasm into the middle (6 %) or the posterior (2 %) of the recipients, as compared to controls injected with buffer into the same positions. To test whether the weak effect observed after deposition in the middle of the recipients was significant, the amount of transferred donor cytoplasm was increased from about 2 % to about 4 % of the egg volume. The large amount was tolerated better when delivered into the middle, rather than the pole regions, of the recipients. Injection of this amount of anterior cytoplasm into the middle of the recipients increased the frequency of normal embryos significantly over the uninjected control (30%; P<0 · 001), and over another sample injected with posterior cytoplasm (24%; P< 0 · 025).

The rescue experiments described above can be used as a bioassay to test the association of anterior determinant activity with defined molecular fractions. Total egg RNA was extracted from Chironomus eggs derived from the N strain and injected at varying concentrations. The injected RNA had a strong rescuing effect (Fig. 1). The frequency of normal embryos increased with the RNA concentration between 1 μ g μ l^{− 1} and 5μ g μ l^{− 1}, but showed no further increase after injection at 10 μ g μ l^{− 1}. The volume of injected RNA was 20–40pl, corresponding to 1–2 % of the egg volume (2 nl). Thus, the amount of the injected RNA at the saturating concentration (5 μ g μ l^{− 1}) averaged 100–200pg, which is less than 1 % of total RNA per Chironomus egg (about 30 ng).

In order to determine whether the rescuing activity was associated with poly(A)⁺ or poly(A)⁻RNA, the two fractions were separated by oligo(dT)-cellulose chromatography. Poly(A)⁺RNA was injected at 0·5μ g μ l^{− 1}, whereas poly(A)⁻RNA and total RNA were injected at 5μ g μ l^{− 1}. The results are summarized in Table 2. As in the previous experiment, injection of total RNA caused a significant rescue effect (58% over the level of buffer-injected controls). Poly(A)⁺RNA, although injected at a tenfold lower concentration, showed the same rescuing activity (58 %). Poly(A)⁻RNA showed much less rescuing activity (18 %). These results indicate that most of the rescuing activity was associated with the poly(A)⁺RNA fraction.

In order to determine whether the activity of anterior determinants in Chironomus eggs was associated with large or small RNA, total egg RNA was precipitated with 4M-LIC1. In one series of experiments (Fig. 2, left set of columns), the rescuing activity of LiCl precipitate was compared to the activity of total RNA saved from the same batch and injected at the same concentration (5μ g μ l^{− 1}). The LiCl precipitate showed a strong rescue effect over the level of both uninjected controls (36%) and buffer-injected control embryos (39%; P< 0·001). The frequency of normal embryos obtained after injection of the LiCl precipitate was even 6% higher than the level observed after injection of total RNA. In another series of experiments (Fig. 2, right set of columns), RNA from either the LiCl precipitate or the LiCl supernatant was injected at 5 μ g μ l^{− 1}. Clearly, the rescuing activity was retained in the precipitate, while the supernatant RNA was inactive. In fact, the RNA from the supernatant showed the same effect also observed after injection with buffer, i.e. it further increased the frequency of double abdomens at the expense of normal embryos in comparison to the uninjected controls. The results indicate that anterior determinant activity was associated with those RNA sequences that were enriched in the LiCl precipitate.

The compositions of the LiCl precipitate and the LiCl supernatant were analysed by polyacrylamide gel electrophoresis. Transfer RNA, 5S and 5·8S ribosomal RNA, and many small nuclear RNAs were present exclusively, or strongly enriched, in the supernatant, whereas the precipitate was depleted of small RNA molecules. It was estimated that RNA molecules of less than 250 nucleotides (nt) length were more prevalent in the supernatant and larger RNA molecules more prevalent in the pellet (Fig. 3).

To characterize further the size of the RNA showing anterior determinant activity, total egg RNA was divided into four fractions by centrifugation through a linear sucrose gradient (15 %−30 %). The size distribution of each fraction was analysed by electrophoresis on agarose gels containing formaldehyde (Fig. 4). While there was considerable overlap between the fractions, they showed an enrichment of size classes as follows: fraction no. 1 (>2400–1100 nt), no. 2 (1800–500nt), no. 3 (1000–200), and no. 4 (<600nt). It was also obvious that the RNA fractionation and/or electrophoresis procedure led to the disappearance of the 28S ribosomal RNA. This was probably caused by heating the RNA samples before loading onto the gradient and/or before electrophoresis (see Materials and methods). Such heating is known to cause the conversion of the 28S RNA into 18S in other insects (Ishikawa & Newburgh, 1972).

Each RNA size fraction and the pooled fractions were injected at a concentration of 5 μ g μ l^{− 1}. The results are shown in Fig. 5. The strongest rescuing activity (40 % over buffer-injected controls) was recovered in fraction no. 4, which was depleted of RNA molecules larger than 600 nt. Fraction no. 3, in which RNA molecules larger than 500 nt were more prevalent, had less rescuing activity (28 %). Fractions no. 1 and no. 2, which were depleted of RNA molecules smaller than 500 nt, had little if any rescuing effect. The activity of the pooled fractions (30 %) was close to the average activity of total RNA (34%, see Table 2).

In order to determine whether fractions no. 3 and no. 4, which contained the anterior determinant activity, also contained mRNA, the fractions were translated in a cell-free system. The stimulation by fractions no. 4 and no. 3 was about 25 % and 75 %, respectively, of total RNA added at the same mass concentration (data not shown). Gel electrophoresis showed that the translation products of fractions no. 4 and no. 3 contained discrete protein bands (data not shown), suggesting that the integrity of the RNA fractions was largely preserved during the preparation procedure.

In order to test whether the rescuing activity observed in Chironomus egg RNA was a general property of poly(A)⁺RNA and/or small (<600nt) RNA, both fractions were prepared from oocytes of an unrelated organism, Xenopus laevis. The rescuing activity of Xenopus poly(A)⁺ RNA was compared to the activity of Chironomus poly(A)⁺RNA as a positive control, and buffer as a negative control (Table 2). Only Chironomus poly(A)⁺ RNA showed a significant rescue effect, whereas Xenopus poly(A)⁺RNA injected at the same concentration (0 · 5 μ g μ l^{− 1}) was inactive. Similarly, total RNA was prepared from Xenopus oocytes and separated into four fractions by the same procedure used to prepare Chironomus RNA fractions. Electrophoresis of Xenopus RNA, in contrast to Chironomus RNA, did not show the conversion of the 28S ribosomal RNA to an 18S. Otherwise, the size distribution of Xenopus RNA was very similar to Chironomus fractions (data not shown). Injection of Xenopus small RNA (fraction no. 4) or large RNA (fraction no. 1) did not cause any rescue effect (Table 2), whereas Chironomus fraction no. 4, which was used as a positive control, showed significant rescue activity as described above. The absence of anterior determinant activity from Xenopus RNA fractions was not due to degradation, as indicated by their activity in a cell-free translation system (data not shown). The results showed that anterior determinant activity detected in poly(A)⁺RNA, as well as small (<600nt) RNA, from Chironomus was not present in corresponding RNA fractions from a widely divergent species.

To test whether the rescuing activity was due to some incidental property or a contaminant present in RNA from Chironomus, but not Xenopus, fraction no. 4 from Chironomus was u.v. irradiated in vitro prior to injection. The irradiation conditions (254 nm, 960J m^{− 2}) were gauged to dimerize, or otherwise modify, less than 2% of the pyrimidines in an RNA molecule (Jackie & Kalthoff, 1978). This degree of RNA modification interferes with translatability (Jackie & Kalthoff, 1980), but should have little or no effect on less-specific RNA functions or on contaminants. Yet most of the rescuing activity was lost after u.v. irradiation (Table 2), whereas unirradiated fraction no. 4 RNA saved from the same batch and injected at the same concentration (5 μ g μ l^{− 1}) was fully active. The loss of rescuing activity after u.v. irradiation of Chironomus fraction no. 4 RNA was statistically significant (P<0 · 05).

To determine whether the rescuing activity detected in cytoplasm and RNA from Chironomus embryos was stage-dependent, RNA or cytoplasm obtained from donors at different stages of development were tested (Fig. 6). Rescuing activity was present in total RNA and/or poly(A)⁺ RNA during intravitelline cleavage and nuclear migration stages. After blastoderm formation, i.e. shortly before gastrulation, the activity was no longer detectable in poly(A)⁺ RNA, even after the concentration was increased from 0 · 5 to 2 · 0 μ g μ l^{− 1}. This RNA preparation merely showed the negative ‘injection effect’. By contrast, a reduced but still significant (P < 0 · 05) activity was still present in anterior cytoplasm from donors at the blastoderm stage. Taken together, the results suggest that at this stage, anterior determinant activity is associated with cytoplasmic components other than RNA. After germ anlage formation (stage GA), this cytoplasmic activity was no longer detectable over the level of uninjected controls.

The poly(A)⁺ RNA from blastoderm embryos, which did not show rescuing activity in our bioassay, was tested for its ability to stimulate a cell-free translation system. Its activity did not exceed 20 % of poly(A)⁺RNA prepared with the same procedure from eggs at stage P_o (data not shown). This low activity of blastoderm RNA was not due to contamination with inhibitors because mixing this RNA with actively translated standard RNA did not decrease the efficiency of the latter. Extensive efforts to avoid any possible degradation during the extraction procedure did not significantly improve the translatability of blastoderm poly(A)⁺ RNA. Thus, its reduced efficiency may reflect an in vivo degradation of maternal messages during preblastoderm stages. This hypothesis is supported by previous results with Smittia, which showed directly that most of the maternal poly(A)⁺ was degraded before the cellular blastoderm stage (Jackie, 1979).

Since double abdomen patterns have been observed in several dipteran species, the molecular mechanisms for specifying the anteroposterior body pattern may be similar within this group of organisms. In order to test this hypothesis, cytoplasm from embryos of Smittia sp., a representative of the subfamily Orthocladiinae within the family Chironomidae, was injected into embryos of Chironomus samoensis, which belongs to the subfamily Chironominae. Similarly, egg cytoplasm from Drosophila, a representative of the suborder Brachycera, was also tested in Chironomus, which belongs to the suborder Nemato-cera. As a positive control, homospecific transplantation of cytoplasm from Chironomus embryos was carried out as described above. Injection of anterior cytoplasm from all three species had a significant rescue effect over both uninjected controls and embryos injected with posterior cytoplasm (Fig. 7). However, this effect decreased slightly from Chironomus to its close relative, Smittia, and more dramatically to its more distant relative, Drosophila.

Most of the double abdomen patterns observed in Chironomus embryos are symmetrical, showing two sets of usually six abdominal segments joined in mirror-image fashion to a midpiece. In addition, there are asymmetrical double abdomen patterns in which the anterior set of segments is shorter, and comprises fewer segments, than its posterior counterpart. The asymmetry involves a juxtaposition, at the plane of polarity, of disparate segments. The disparity may be extreme, such as a tenth abdominal segment joined, with reverse polarity, to a thoracic or even cephalic segment (Percy et al. 1986). In the experiments reported here, the proportion of asymmetrical patterns among all double abdomen embryos (asymmetry index) was about 0 · 014 in unirradiated controls and about 0 · 027 in buffer-injected controls (Table 3). The asymmetry index increased to 0 · 14 in embryos injected with cytoplasm that had a significant rescue effect and to 0 · 097 in embryos injected with RNA fractions containing anterior determinant activity. By comparison, injection of cytoplasm or RNA fractions that had no rescue effect did not lead to a significant increase in the asymmetry index.

The propensity of Chironomus eggs to form double abdomens upon u.v. irradiation has allowed us to establish a bioassay for anterior determinant activity, which is required for head and thorax formation in this species. Embryos programmed for double abdomen development by anterior u.v. irradiation were rescued, i.e. restored to normal development, by injecting cytoplasm from unirradiated donors. The rescue activity depended strongly on the donor site from which the cytoplasm was taken (Table 1). Anterior cytoplasm was more active than middle cytoplasm, whereas posterior cytoplasm led to a decrease, rather than an increase, in the frequency of normal embryos. The data indicate that anterior determinant activity in Chironomus embryos decreases from the anterior to the posterior pole region. The rescuing activity of anterior cytoplasm also decreased significantly when it was delivered to middle or posterior egg regions (Table 1). Injection of anterior cytoplasm into the middle of the recipients was followed by a significant rescue effect only after increasing the amount of transplanted material. The results show that anterior determinant activity is required especially near the anterior pole. The data further suggest that some of the components with anterior determinant activity may be diffusible. The results obtained confirm and extend conclusions drawn from earlier experiments using eggs of a related chironomid, Smittia sp. The efficiency of u.v. induction of double abdomen, as measured by the fluence per target area required to produce a given yield of double abdomens, generally decreased from anterior to posterior (Kalthoff, 1971). Moreover, the ability of Smittia embryos to form double cephalons upon centrifugation was diminished after anterior, but enhanced after posterior, u.v. irradiation (Kalthoff et al. 1982).

Transplantation experiments similar to ours were carried out using Drosophila embryos from both wildtype strains and maternal effect mutants which are defective in the determination of anteroposterior polarity (Frohnhöfer et al. 1986; Nüsslein-Volhard et al. 1987). Embryos from strong bicoid alleles lack head and thorax and show a telson at the anterior end (Frohnhöfer & Nüsslein-Volhard, 1986). Bicoid embryos were ‘rescued’, to an extent, by transplanting anterior cytoplasm from wild-type embryos, although rescue by injection of RNA has not been reported. Moreover, Frohnhöfer & Nüsslein-Volhard (1986) induced ectopic head structures by injecting anterior cytoplasm from wild-type embryos at different positions along the anteroposterior axis of Drosophila bicoid embryos. In our experiments with Chironomus embryos, transplantation of anterior cytoplasm to middle or posterior regions did not usually cause the formation of ectopic head structures. Only in rare instances did we observe, near the injection site, the formation of the red pigment which occurs in the larval eyes. The discrepancy may be explained by the larger amounts of cytoplasm (5 – 7 · 5% egg vol.) transplanted in Drosophila as compared to 2 % egg vol. in our experiments. Alternatively, the anterior determinants in Chironomus, as measured in our bioassay, may differ in function from the bicoid product in Drosophila (Kalthoff & Rebagliati, 1988).

The amounts of cytoplasm or RNA required for significant rescue effects were about 2 % of the total egg cytoplasm, or less than 1 % of an egg equivalent of RNA, respectively. This sensitivity contrasts with the scarcity of ectopic head structures after heterotopic transplantation of cytoplasm in our experiments. Also, surviving donor embryos from which anterior cytoplasm had been removed, did not normally develop into double abdomens. These results may seem paradoxical but are consistent with the sigmoid shape of fluence-response curves for u.v. induction of double abdomen formation obtained previously (Kalthoff, 1971). Up to a certain u.v. fluence, no double abdomens were induced. Only beyond this ‘shoulder’ did the yield of double abdomens increase with increasing u.v. fluence. This steep part of the fluence-response curve, which extended between about 90% and 10% normal embryos, was followed by an exponential ‘tail’ where even large increments of u.v. fluence led only to small further decreases in the percentage of normal embryos. The recipient embryos in our rescue experiments were pretreated with a u.v. fluence to place them in the steep portion of the curve, where an increase of 30% in the frequency of normal embryos required only an increase of about 2 % in anterior determinant activity. In contrast, the unirradiated donors of anterior cytoplasm were in the shoulder of the fluence-response curve, where small reductions of anterior determinant activity are not sufficient for inducing any double abdomens. By the same analogy, heterotopic transplantation of anterior cytoplasm to the posterior pole region corresponds to a small reduction of u.v. fluence in the exponential ‘tail’ of the fluence response curve which has no strong effect on the frequency of normal embryos either.

The sensitivity of the bioassay suggests that anterior determinants may have autocatalytic properties, or that they interact in a mutually inhibitory way with posterior determinants. The latter interpretation is supported by data from combined u.v. irradiation and centrifugation experiments with Smittia embryos (Kalthoff et al. 1982) and by transplantation experiments with Drosophila embryos (see Nüsslein-Volhard et al. 1987). The sensitivity of u.v.-irradiated Chironomus embryos to small perturbations is also reflected in the negative rescue effect after injection with buffer, which was observed throughout our experiments. In comparison to the uninjected controls, buffer injection caused an increase in the frequency of double abdomens at the expense of normal embryos, which averaged around 9 % throughout our experiments. This ‘injection effect’ had also been observed in earlier experiments with Smittia eggs where puncturing at the anterior pole caused double abdomen formation (Schmidt et al. 1975), although it occurred only in embryos punctured before nuclear migration and not at a stage comparable to the M₂ stage of the recipients used in the present experiments. The ‘injection effect’ observed in Chironomus embryos seems to be unspecific since it was also observed after injecting buffer into middle and posterior regions (Table 1). In some experiments, including anterior injection of posterior pole plasm (Table 1), LiCl supernatant (Fig. 2) or poly(A)⁺ RNA from blastoderm embryos (Fig. 6), we observed a negative rescue that was stronger than the average injection effect. It is not clear whether the injected agents were simply exacerbating the unspecific injection effect or were specifically inhibiting anterior determinants.

Injection of total embryonic RNA, or active fractions thereof, caused a similar rescue effect as anterior egg cytoplasm. The effects were indiscernible in terms of both the increased frequency of normal embryos and their morphology. The rescuing activity of RNA is compatible with the maternal origin of anterior determinants, which is indicated by their presence in newly deposited eggs and by the fact that the double abdomen phenotype in both Drosophila and Chironomus is caused by maternal rather than embryonic mutations (Nüsslein-Volhard et al. 1987; Kuhn et al. 1987). In chironomids, as well as in other insects with polytrophic oogenesis, oocytes receive large amounts of RNA-containing material from nurse cells through cytoplasmic bridges near the anterior pole of the oocyte. The rescuing activity of RNA from Chironomus embryos is also in accord with our earlier characterization of anterior determinants in Smittia as RNP particles. This characterization was based on several lines of independent evidence including the effects of injected enzymes and the action spectrum for u.v. induction of double abdomen formation. The action spectrum showed a major peak at 285 nm, and a shoulder at 265 nm, suggesting the involvement of protein and nucleic acid moieties, respectively (see Kalthoff, 1983). We assume that RNA injected into Chironomus embryos becomes reassociated with proteins to form functional RNP particles. This implies that complementary proteins are available in the recipients, either as independent proteins or after dissociation from u.v.-damaged RNA. Such proteins might be scarce, which would account for the saturation in the dose-response curve for injected RNA (Fig-1)

Poly(A)⁺ RNA selected from total Chironomus egg RNA showed undiminished rescuing activity (Table 1). The poly(A)^-RNA fraction prepared from the same batch and injected at tenfold higher concentration had little rescuing effect (18 % as compared to 58% with poly(A)⁺ RNA fraction). This low activity in the poly(A)⁻ RNA fraction may have resulted from incomplete poly(A)⁺ selection. We assume that poly(A)⁺ RNA would be active at much lower concentrations than used in our experiment. In fact, it might reach saturating activity at 0 · 075 μ g μ l^{− 1}, i.e. the concentration at which it is present in 5 μ g μ l^{− 1} total RNA. However, we have not tested this assumption because degradation problems suggested by the low activity of 1 μ g μ l^{− 1} total RNA (Fig. 1) would have been compounded at lower RNA concentrations. Thus, we cannot exclude the possibility that sequences in the poly(A)⁻ fraction may substitute for components of the poly(A)⁺ fraction.

The association of rescuing activity with poly(A)⁺RNA is compatible with the hypothesis that anterior determinants in Chironomus are mRNP particles. This hypothesis is supported by the persistence of anterior determinant activity in the cytoplasm at the blastoderm stage when extracted RNA was no longer active (Fig. 6), and by the u.v. sensitivity of the active RNA sequences (Table 2). The idea that anterior determinants might be mRNP particles is also supported by the development of double abdomen embryos after anterior injection of cycloheximide (25 – 40 μ g μ l^{− 1}) into Smittia embryos (Kandler-Singer, 1977). In Drosophila, where a bicoid-dependent activity in the cytoplasm acts as a localized anterior determinant (Frohnhöfer & Nüsslein-Volhard, 1986), the anterior localization of an RNA almost certainly representing a bicoid transcript was shown directly by in situ hybridization (Frigerio et al. 1986).

To obtain an estimate for the size range of the RNA molecule(s) having anterior determinant activity, two fractionation procedures were used. On the one hand, anterior determinant activity was associated with fraction no. 4, which was depleted of RNA molecules larger than 600 nt (Figs 4, 5). The active RNA molecule appeared to be even smaller than 500 nt because RNAs between 500 and 600 nt were clearly more prevalent in fraction no. 3 which, however, was less active. The same argument applies to larger contaminants in fraction no. 4, which were more prevalent in the inactive fractions no. 1 and no. 2. On the other hand, the association of anterior determinant activity with LiCl-precipitable RNA indicates that the active sequence is larger than 250 nt (Figs 2, 3). Again, the precipitate contained some RNA molecules smaller than 250 nt which were much more prevalent in the inactive supernatant (Fig. 3). Assuming that only one RNA sequence is required for the degree of rescue observed, the results suggest that the size of the Chironomus RNA with anterior determinant activity may be in a range of approximately 250-500 nt. However, given the presence of large-size contaminants in fraction no. 4 (Fig. 4), we cannot exclude the possibility that our rescue bioassay responds to the simultaneous activity of large and small RNAs. The 250 – 500 nt size class is unusual in that most eukaryotic mRNAs are longer than 500 nt, whereas most of the small nuclear RNAs and small cytoplasmic RNAs (Wolin, 1985) are shorter than 250 nt. Clearly, our preliminary size assessment needs to be refined by more stringent fractionation methods.

Earlier experiments with Smittia embryos showed that the sensitive periods for double abdomen induction by both u.v. (Kalthoff, 1971; Ripley & Kalthoff, 1983) and RNase (Kandler-Singer & Kalthoff, 1976), as well as photoreversal (Kalthoff et al. 1975), extended from egg deposition up to an early preblastoderm stage. These results suggested that anterior determinants sensitive to u.v. or RNase were no longer active by the blastoderm stage. To test the developmental control of anterior determinants in Chironomus, cytoplasm or RNA fractions from embryos at different stages were injected. As shown in Fig. 6, anterior determinant activity was present in both cytoplasm and RNA from embryos during stages P_O – M₂. At stage Bl, i.e. shortly before gastrulation, activity was still present in cytoplasm but no longer poly(A)⁺RNA. After germ anlage formation (stage GA), anterior determinant activity had also disappeared from cytoplasm. The data indicate that anterior determinant activity was associated with RNA during intravitelline cleavage and nuclear migration, but with other cytoplasmic components at the blastoderm stage. Such other components might of course be proteins translated from the RNA sequences active during the earlier stages.

Corresponding experiments with Drosophila have indicated that gene products acting as cytoplasmic determinants are developmentally controlled in a similar fashion. For example, Müller-Holtkamp et al. (1985) have shown that the rescuing activity for the pelle mutant disappeared from poly(A)⁺ RNA fraction at an earlier stage of cleavage than from the cytoplasm, suggesting that translation of the mRNA was completed by this stage. Likewise, the rescuing activity for the tube mutant was still present in the cytoplasm at the germ band stage although RNA from that stage was no longer active (Anderson & Nüsslein-Volhard, 1984).

The occurrence of double abdomen embryos in several dipteran families and the similarities in many details observed among them, suggest that the specification of the longitudinal body pattern in dipteran embryos may be governed by a similar formal ‘logic’, and possibly by homologous sets of embryonically expressed genes (see Percy et al. 1986). However, the sets of maternal gene products that determine the initial anteroposterior polarity seem to differ between Drosophila and chironomids (Kalthoff & Rebagliati, 1988). We have used our rescue bioassay with Chironomus embryos to test the activity of cytoplasm from two other dipterans, Smittia sp. and Drosophila melanogaster. As shown in Fig. 7, anterior determinant activity was present in anterior, but not posterior, cytoplasm from both species. However, the heterospecific activity decreased with the phylogenetic distance between donor and recipient. Further experiments are required to determine whether the rescuing activity of Drosophila egg cytoplasm is also present in RNA, and in particular, in size fraction no. 4. The apparent size range of anterior determinants in Chironomus (250 – 500 nt) contrasts with the sizes of cloned Drosophila cDNAs representing gene products with similar functions, such as bicoid (2 · 6 kb) and hunchback (3 · 1 kb). It will be interesting to see whether the homospecific and the heterospecific anterior determinant activities share the same molecular basis.

Similar heterospecific transplantation experiments have been carried out by other authors between different species of Drosophila. Mahowald et al. (1976) demonstrated that transplantation of pole plasm from Drosophila immigrons induced ectopic pole cell formation in Drosophila melanogaster embryos, and that these cells could produce gametes. Müller-Holtkamp et al. (1985) showed that the dorsoventral body axis in the pelle mutant was partially restored by cytoplasm from different species of Drosophila, but not from Smittia, embryos.

Several control experiments indicated that the observed anterior determinant activity was not just a general property of small RNA or degraded poly(A)⁺ RNA. As mentioned above, poly(A)⁺ RNA from Chironomus embryos at the blastoderm stage, which presumably contains large amounts of degraded maternal mRNA (Jackie, 1979), showed no activity (Fig. 6). Also inactive were intact poly(A)⁺ RNA and fraction no. 4 RNA from Xenopus (Table 2), which were prepared using the same procedure used to prepare the active fractions of Chironomus RNA. Finally, we showed that the activity of Chironomus fraction no. 4 RNA was substantially reduced after exposure to a u.v. fluence that should have dimerized, or otherwise modified, less than 2 % of the pyrimidine bases (Jackie & Kalthoff, 1978). Taken together, these results indicate that anterior determinant activity in Chironomus embryos is stage-dependent and not present in an unrelated organism. In addition, the u.v. sensitivity of the anterior determinant activity suggests that the biological function depends on the translatability, or secondary structure, of the active RNA.

Injection of cytoplasm or RNA that showed rescuing activity, i.e. increased the frequency of normal embryos at the expense of double abdomens, also had a more subtle effect. Among the remaining double abdomens, the proportion of asymmetrical specimens (asymmetry index) was significantly higher than in controls injected with buffer or nonrescuing cytoplasm or RNA fractions (Table 3). The enhanced asymmetry index under conditions promoting complete rescue suggests that asymmetrical double abdomens represent an intermediate phenotype between normal embryos and symmetrical double abdomens and, thus, partially rescued embryos. This view is supported by further observations. In spontaneously occurring double abdomens, the asymmetry index is inversely related to the double abdomen frequency in a given cluster (J. Percy & K. Kalthoff, unpublished data). Among the u.v.-induced double abdomen embryos, the asymmetry index is higher if the u.v. irradiation is carried out after nuclear migration rather than before (M. Laurel & K. Kalthoff, unpublished data). Conceivably, asymmetrical double abdomens arise when anterior determinants are inactivated to a marginal extent and/or late, so that a ‘stalemate’ between them and their posterior antagonists cannot be resolved before the activation of embryonic gap genes (Kalthoff & Rebagliati, 1988).

We are grateful to Dr H. Yajima (Ibaraki University, Mito, Japan) who provided the starting material for our laboratory culture of Chironomus samoensis. We wish to thank our colleague, Dr M. Winkler, for sharing his expertise in, and components for, in vitro translation. Drosophila melanogaster embryos and Xenopus laevis oocytes were kindly provided by our colleagues, Drs C. S. Lee and W. R. Jeffery. This work was supported by NSF grant DCB-8509517.