Differential stability of expression of similarly specified endogenous and exogenous genes in the sea urchin embryo

The early blastula of the sea urchin Strongylocentrotus purpuratus consists of five embryonic territories, each composed of a unique set of clones, and each displaying a unique developmental fate (Davidson, 1989). The aboral ectoderm is at the molecular level one of the best known of these territories, and is composed of the progeny of 11 founder cells that segregate from other lineages at 3rd–6th cleavage (Cameron et al. 1987, 1990). An early molecular marker of aboral ectoderm specification is expression of the cytoskeletal Cyllla actin gene, which begins at least as early as 8th cleavage (Hickey et al. 1987; Lee, 1986; Cox et al. 1986; Shott et al. 1984). The Cyllla gene is expressed throughout the aboral ectoderm during the whole of embryogenesis (Cox et al. 1986), and continues to be active until metamorphosis (Cameron et al. 1989). However, Cyllla transcripts are not detectable in the juvenile or adult sea urchin (Shott et al. 1984).

Regulatory interactions governing Cyllla gene expression have been studied in vivo using a fusion construct, CylllaCAT., in which the Cylllacis-regulatory domain controls the expression of the bacterial chloramphenicol acetyltransferase (CAT) reporter gene (Flytzanis et al. 1987; Hough-Evans et al. 1987, 1988,1990; Franks et al. 1990). The cri-regulatory domain required to induce the expression of Cyllla · CAT on the same temporal schedule, and in the same spatial elements of the embryo, as the endogenous Cyllla gene extends about 2.3 kb upstream from the transcriptional start site. This domain includes about 20 specific sites of DNA-protein interaction, mediated by 12 distinct factors (Thézé et al. 1990; Calzone et al. 1988 ; Thiebaud et al. 1990). Many of these factors have been partially or extensively characterized biochemically. Some are known to function positively and some negatively; some are involved with spatial, some with temporal, and others with amplitude regulation of Cyllla transcription (Franks et al. 1990; Hough-Evans et al. 1990; Calzone et al. 1991; Hö ög et al. 1991; reviewed by Davidson, 1989). An interesting experimental feature of the sea urchin embryo is that both positive and negative functions can be competitively titrated in vivo by coinjection with Cyllla · CAT of fragments of the regulatory domain containing target binding sites for given regulatory factors (Franks et al. 1990; Hough-Evans et al. 1990). If the whole Cyllla regulatory domain needed for correct expression is coinjected with Cyllla · CAT, CAT enzyme activity is decreased almost stoichiometrically, as a function of the injected molar ratio of competitor to Cyllla · CAT (Livant et al. 1988). This result at once raises the issue of whether expression of the endogenous Cyllla genes is correspondingly depressed on introduction of the excess regulatory target sites. In order to address this question experimentally, we had first to circumvent the extensive mosaicism resulting when DNA is introduced into sea urchin eggs by a single microinjection of several thousand molecules of linearized DNA, here Cyllla • CAT DNA (McMahon et al. 1985; Flytzanis et al. 1985; Hough-Evans et al. 1988). We had shown that, after a single injection, in 75 % of embryos the exogenous DNA is concatenated and stably incorporated into a blastomere following 2nd, 3rd or 4th cleavage, resulting in a number-average frequency of only about 6% of nuclei per embryo that contain exogenous DNA concatenates. In order to measure experimentally any effects of competition on expression of the endogenous Cyllla gene the competitor must obviously be stably incorporated in a far greater fraction of cells. As described below, we discovered that, after 3 or 4 separate injections postfertilization, many embryos incorporate exogenous DNA in all their nuclei, and others in most nuclei. We provide an analysis of exogenous DNA incorporation mechanics in embryos subjected to the new multiple injection procedure, and show that Cyllla · CAT expression in multiply injected embryos remains temporally and spatially correct. We then present simultaneous measurements of CAT enzyme and Cyllla actin mRNA levels in the same multiply injected embryos. These measurements showed that under conditions in which 85 % of control Cyllla · CAT expression was abolished by competition with a several-fold excess of Cyllla regulatory domain DNA, endogenous Cyllla gene expression remained entirely unaffected. In situ hybridization confirms that the endogenous Cyllla gene continues to be expressed in all aboral ectoderm cells, and no other cells, in the competition embryos. Since both endogenous and exogenous regulatory systems are specified correctly for expression, they both must participate in the same necessary regulatory interactions. However, these interactions are evidently greatly stabilized in the endogenous gene, by a mechanism that if understood might provide a clue to the phenomenon known to embryologists as ‘commitment.’

Eggs of Strongylocentrotus purpuratus were collected and prepared for injection as described previously (McMahon et al. 1985). Animals were induced to shed by 12.5 V electrical stimulation or intracoelomic injection of 0.5 m KOl. Eggs were filtered through gauze and treated for a few minutes with pH 5 filtered sea water (FSW) to remove egg jelly. Unfertilized eggs were then moved to Petri dishes that had been treated for 1 min with a 1 % solution of protamine sulfate. The eggs were placed in rows in FSW containing 82 mg 1^-1 aminotriazole (ATA), an ovoperoxidase inhibitor, in order to prevent hardening of the fertilization membrane. Eggs were fertilized by the addition of activated (diluted) sperm immediately before microinjection. The zygotes were microinjected at intervals of 4 to 8 min with approximately 2 pl of DNA solution per injection containing 0.1 pg of DNA. After injection, ATA-containing FSW was replaced by FSW containing 20 units of penicillin and 50μg of streptomycin per ml (PSSW). Embryo samples were taken at 8–24 h postfertilization for CAT assays, in situ hybridization to squashed blastula nuclear DNA, slot blot assays, and probe excess RNA titration. Embryos were cultured to 72 h (early pluteus) in PSSW at 15 °C for in situ hybridization to CAT mRNA.

Microinjection solutions contained 40% glycerol, 0.12 m KC1, the fusion gene Cyllla · CAT (Flytzanis et al. 1987) and, in some experiments, SphI-cut sea urchin DNA size selected to have an average length of 10 kb as carrier (Livant et al. 1988) and a 9.3 kb SalI fragment from the Cyllla gene. Injection solutions contained from 600 to 3000 molecules per pl of Cyllla · CAT DNA linearized with SphI. In the competition experiments, varying amounts of the 9.3 kb SalI fragment containing the entire regulatory domain of the Cyllla · CAT gene and enough carrier DNA to make the final DNA concentration 0.1 pg per 2 pl were also included in the injection solution.

CAT enzyme activity per embryo was measured in samples of 30 to 80 24 h blastulae, by a method derived from that of Gorman et al. (1982), as described by McMahon et al. (1984). The number of Cyllla · CAT DNA molecules per embryo was measured in samples of 30 to 80 24 h blastulae by quantitative slot blot hybridizations as previously described (Flytzanis et al. 1987).

Preparation of embryo sections and microscope slides were as described previously (Hough-Evans et al. 1987, based on the method of Cox et al. 1984). Embryos that developed from eggs that had been microinjected with Cyllla · CAT were fixed at the 72 h pluteus stage in 2 % paraformaldehyde, 0.5 % glutaraldehyde in buffer containing 2.5% NaCl, 25 mM phosphate buffer pH7.5. The embryos were enclosed in agarose boxes, embedded in Tissueprep 2 (Fisher Scientific), and sectioned at 5μm. The serial sections were placed on polylysine-coated slides.

Squashes were prepared by placing blastulae in PSSW on a polylysine-coated slide and flattening them under a siliconized coverslip by gradually drawing off the sea water.

The single-stranded antisense CAT RNA hybridization probe used here was transcribed as described previously (Hough-Evans et al. 1987) from a fragment of CAT coding sequence inserted into the polylinker of the pSP65 vector, using ³⁵S-labeled UTP, specific activity 800 Ci mmol^-1. In situ hybridization was carried out as in earlier experiments by the procedures of Angerer and Angerer (1981) and Cox et al. (1984), modified for ³⁵S-labeled probes by the addition of 5 him dithiothreitol and 1 % β-mercaptoethanol to the hybridization solution and 1 % β-mercaptoethanol to the wash solutions. Hybridization was carried out in an N₂ atmosphere. The photographic emulsion (Kodak NTB-2) was diluted 1:1 with distilled water. Exposure was for 4 days.

For hybridization to the DNA of squashed blastulae, CAT coding sequence inserted in the opposite orientation in the pSP64 vector was used to transcribe a sense strand RNA probe labeled with ³⁵S-UTP. The hybridization procedure included deproteinization at 10μgml^-1 of proteinase K (compared to 1 μgml^-1 in the RNA–RNA hybridization) and denaturation of the cellular DNA by immersing the slides in 95 % formamide, 0.1×SSC for 15min at 65°C (Hough-Evans et al. 1988). Cells labeled with five or more grains were considered to be positive. The fraction of cells that were labeled was estimated (as a percent) for each blastula (usually ∼25) in every experiment. Variations in grain density are to be expected, due to local differences in thickness of the squash and/or the emulsion.

Total RNA was isolated from 500–1000 blastulae, which had been counted, pelleted and frozen dry at –70°C, using a urea sarcosyl buffer (Lee et al. 1986) as described by Sucov et al. (1988). An Sp6-transcribed RNA probe labeled with 40 Ci mmol^{-1 3}H-UTP corresponding to 1300 nucleotides of unique M actin 3’ sequence (described by Lee et al. 1984, 1986) was introduced at the urea solubilization step as an internal control for preparative yield of RNA. Probe excess titration was carried out as described by Lee et al. (1986) using an Sp6-transcribed RNA probe for Cyllla mRNA. This 131 nucleotide probe is described by Lee et al. (1984, 1986).

To rationalize the exogenous DNA incorporation data shown in Figs 2 and 3, we formalized our hypothesis for the incorporation process. On the basis of evidence obtained from studies of singly-injected eggs, the following conditions are assumed: (i) that once incorporated, most exogenous DNA is retained in subsequent development in the nuclei of the lineage element descendant from a blastomere in which the incorporation event occurred, and that it replicates together with the host cell genomic DNA; (ii) that concatenates are rapidly formed and all the DNA injected is partitioned among them; and (iii) that individual incorporation events within an embryo, i.e. in different blastomeres, are independent and of low probability, so that the occurrence of these events can be described by a Poisson function. The incorporations apparently occur very early in cleavage, while our observations on stable incorporation frequencies were made on squashes of 24 h, 400-cell embryos. Thus, in the following treatment, we assume that all cell lineages divide essentially the same number of times during cleavage. While this is not precisely correct, the deviations are quantitatively insignificant for our purposes (see Davidson, 1986, pp. 213–218; Cameron et al. 1987).

We define the following parameters:

p: The probability of incorporation per concatenate, and per nucleus, p is treated as a batch-dependent biological characteristic. An analysis of single injection data (Hough-Evans et al. 1988) yields a value for p of ∼0.034, whether incorporation occurred at 2nd, 3rd or 4th cleavage.

C: The number of concatenates per egg (formed after all three injections). C is conceived as a major variable distinguishing one batch of eggs from another.

X: The number of blastomeres of a cleavage stage embryo in which a nuclear incorporation event occurs; is the average number of cells/embryo in which an incorporation is detected in a batch of embryos.

n: The cleavage stage (i.e. 1st, 2nd, 3rd, etc.) at which given incorporation events occur; the number of cells or nuclei at this stage is 2ⁿ.

G: The fraction of cells in an embryo at any given stage after incorporation has occurred which display exogenous DNA; is the average fraction of cells displaying exogenous DNA per batch.

I: Number of molecules of exogenous DNA initially injected (total for all injections) per egg.

⁠: Average number of molecules of exogenous DNA recovered per cell that contains exogenous DNA; for a given batch of embryos is calculated as average total exogenous DNA per 24 h embryo, measured by slot blot hybridization, normalized to the parameter (400 ). Here is calculated from measurements on squashed 24 h embryos.

The latter provides an estimate of the value of C for any given batch. However, for batches where there are a very large fraction of embryos in which all cells display the exogenous DNA, it is likely on random expectation that some of these cells will contain more than one stable integration. Thus the values of C calculated from eq. (3) for such cases may be larger than actually were present.

which is of the form eq.(l) of text.

This hyperbolic relation between and (Fig. 3) is expected, because as I is divided into more concatenates there is a proportional decrease in DNA/concatenate (eq. 3), but for given n and p a proportional increase in the number of cells expected to undergo incorporation (eq. 1).

In a series of preliminary experiments, we compared embryos from the same batch of eggs that developed after one, two or four rounds of microinjection. To determine the fraction of nuclei that had stably incorporated Cyllla · CATDNA after injection into the cytoplasm of fertilized or unfertilized eggs, the embryos were raised to mesenchyme blastula stage (24 h), squashed and the chromosomal DNA hybridized in situ with a sense strand RNA probe representing a portion of the CAT coding sequence (Hough-Evans et al. 1988; see Materials and methods). Fig. 1 displays squashes of four transgenic embryos. Those in panels A and B developed from fertilized eggs that had received a total of 0.4 pg of linearized Cyllla · CAT DNA (the Cyllla · CAT fusion is 14 kb long), delivered in four 0.1 pg injections. For comparison the typical embryo shown in panel C received about 0.1 pg of exogenous DNA, delivered in a single prefertilization injection. In this particular experiment, the injection aliquot contained Cyllla · CAT DNA and ∼10 kb long sea urchin DNA carrier in a molar ratio of 1:5. After a single injection, the exogenous DNA is very efficiently concatenated with the carrier, and incorporation frequency is not affected by the presence of carrier DNA (McMahon et al. 1985; Franks et al. 1988, 1990; Hough-Evans et al. 1988; data presented in this paper). Essentially all of the nuclei (>95%) in the embryos of panel A are labeled with the ³⁵S-CAT probe and thus contain Cyllla · CAT DNA. The embryo in panel B is about 50% labeled, and the embryo shown in panel C is about 6% labeled (19 out of ∼320 nuclei). As shown in Table 1, less than 10% of the injected DNA is actually incorporated and replicated in cells of the embryos that develop from multiply injected eggs.

The high frequency of stable nuclear incorporation of injected DNA shown in Fig. 1A and B is not unusual for embryos obtained from eggs injected three or four times postfertilization (hereafter designated as 3× or 4× injected eggs). These particular blastulae derived from the same batch of 4× injected eggs; 92% of the nuclei in this batch of embryos displayed exogenous DNA at 24 h of development. At least three postfertilization cytoplasmic injections are necessary to obtain this significantly enhanced incorporation of exogenous DNA. Three injections produce maximum results, and the fraction of nuclei that display incorporated exogenous DNA at the blastula stage in 3× injected eggs is indistinguishable from that of blastulae obtained from 4× injected eggs (data not shown). In other experiments (data not shown), we confirmed that enhanced incorporation does not occur after a single injection delivered 30 min after fertilization (4× injection of 50–100 eggs takes about 30min), nor does increase in the mass of DNA injected from 0.1 to 0.4 pg in itself produce enhanced incorporation. Many batches of eggs are damaged by loads of injected DNA greater than this (McMahon et al. 1985). We also discovered that multiple injection of DNA prior to fertilization yields no enhancement in the frequency of nuclear incorporation compared to single injection (data not shown). The overall mortality of embryos developing from multiply injected eggs is similar to that of embryos developing from singly injected eggs, and many of the resulting larvae survive metamorphosis to become fertile adults (B.Evans and D.Livant, unpublished data).

The eggs of individual females respond differently to multiple injection. Fig. 2 describes the results of experiments on 10 separate batches of eggs obtained from different females and injected 3× or 4× postfertilization, with 0.1 pg of linearized Cyllla-CAT and carrier per injection. As before, the blastulae were squashed at 24 h of development, and individually scored for the fraction of labeled nuclei displayed by in situ hybridization to the ³⁵S-labeled CAT probe. Panel A shows the composite nuclear incorporation profile obtained by combining all ten analyses. A total of 238 blastulae were assayed, and an average of 59% of the embryo nuclei contained Cyllla · CAT DNA. Overall, 91 % of the embryos scored had at least some nuclei containing exogenous DNA, and 29 % of all blastulae contained Cyllla · CAT DNA in >95 % of nuclei. Similar results have been obtained with constructs other than Cyllla - CAT (data not shown). For comparison, in Fig. 2B we show a similar composite nuclear incorporation profile, obtained from analyses of 24h blastulae raised from several batches of eggs injected once prior to fertilization with about 0.1 pg of linearized Cyllla · CAT plus carrier DNA (data from Hough-Evans et al. 1988). A total of 79 blastulae were assayed, and an average of 6 % of the nuclei contained Cyllla · CAT DNA. The highest fraction of nuclei bearing exogenous DNA in an embryo derived from a singly-injected egg was 32 %. Fig. 2 demonstrates that blastulae raised from 3× or 4× injected fertilized eggs have approximately ten times as many nuclei containing exogenous DNA as those obtained from singly-injected eggs. Figs 2C,D illustrate the batch variation that we observed in comparing the response of eggs of the 10 females to multiple injection. In the batch of blastulae represented in panel C, 78 % of all embryos contained CAT DNA in >95 % of nuclei, and the overall fraction of nuclei displaying the exogenous DNA was 93 %. Panel D shows a batch of blastulae in which only 9 % of all embryos contained exogenous DNA in >95 % of nuclei, and here the overall fraction of positive nuclei was just 34%. In the ten batches of embryos assayed, the average fraction of nuclei displaying CAT DNA per embryo varied from 17 % to 93 % ; and the calculated fraction of embryos in which >95 % of nuclei contained exogenous DNA varied from 2.6% to 78%.

The data from 12 out of the 13 3×injection batches are fit surprisingly well by this function (root mean square error is about 10 % of the values of ⁠; we did not pool the 4× with the 3× injection data for the analysis shown, but all the data are evidently consistent with the form depicted). Thus it appears that there is a strong inverse proportionality between the (average) amount of CAT DNA incorporated per cell and the (average) number of cells bearing this DNA, so that the largest (average) amounts of CAT DNA per cell occur in those batches in which incorporation is least widespread. We present a plausible mechanism for the incorporation process, and from this a derivation of eq.(l), in Methods. The parameters of this model are evaluated for ten multiply injected batches of embryos as summarized in Table 1. This analysis suggests that the two major variables among different batches of eggs are the number of concatenates formed from a given amount of injected DNA, and the probability of stable incorporation per concatenate nucleus. Most 3× injection batches may produce 20–100 concatenates per embryo, each having a probability of stable nuclear incorporation ranging from 0.02 to 0.14 (average 0.065) per concatenate nucleus. The most important conclusion is that for most batches, the data can be adequately accounted for on the assumption that incorporation first occurs in the nuclei of 2- to 4-cell embryos. It is unlikely that incorporation occurs before 1st cleavage because of the presence of the nuclear membrane. First or second cleavage incorporation of a concatenate should not preclude later incorporation events in the same embryos, particularly in those where large numbers of concatenates are present, but these events would involve relatively fewer cells and their effects would be superimposed on the quantitatively dominant effects of the earlier incorporation events.

Multiple injections probably produce embryos in which many or all cell nuclei bear exogenous DNA because large numbers of exogenous DNA concatenates are formed during the injection process. This might be due to physical disturbance of the egg cytoplasm, to the deposition of the DNA in several different locations in the egg, or to a physiological response to multiple injection that otherwise affects ligation efficiency. Possibly cytoskeletal changes occasioned by fertilization are important, since multiple injection into unfertilized eggs is ineffective in promoting high frequency incorporation. In any case, the result of postfertilization multiple injections is to greatly enhance the likelihood of incorporation in all or in a large fraction of early cleavage nuclei, though of smaller amounts of DNA per nucleus. The multiple injection procedure satisfies our requirement for a method that would enable exploration of the effects on endogenous genes of regulatory target site competition.

The time course of normal Cyllla transcript accumulation during embryonic development has been determined by RNA gel blot (Shott et al. 1984) and probe excess titration (Lee et al. 1986). Newly synthesized Cyllla gene transcripts are undetectable until 10 h postfertilization, but thereafter they accumulate rapidly up to about 8×10⁴ molecules per embryo (the unfertilized egg contains about 2×10³ molecules of maternal Cyllla mRNA). Flytzanis et al. (1987) showed that the time course of appearance of CAT enzyme activity in embryos raised from singly injected eggs parallels that of endogenous Cyllla mRNA accumulation. To ascertain that this would also be the case for 3× injected eggs, we carried out the experiment shown in Fig. 4. Here it is shown that the kinetics with which CAT activity increases again reproduces the measured kinetics of Cyllla mRNA accumulation in normal embryos (Shott et al. 1984). Thus there is no detectable CAT activity at 8h postfertilization; CAT expression becomes evident by 12 h; and CAT enzyme level, like Cyllla mRNA, then continues to increase until about 24 h.

Cyllla gene transcripts have been demonstrated by in situ hybridization in all aboral ectoderm cells and their precursors, but in no other embryonic cells (Cox et al. 1986). We assayed the spatial distribution of CAT transcripts by this method in 72 h plutei raised from 3× injected eggs. Each egg received a total of 3600 Cyllla · CAT molecules, plus 20400 carrier DNA molecules. In this batch of plutei, an average of 64 % of the nuclei contained Cyllla · CAT DNA. We scored complete serial sections of six >95 % transgenic embryos, four ∼50 % transgenic embryos, and four ∼25 % transgenic embryos, and a total of six ectopically labeled cells were found in different embryos, viz five single cells within the blastocoel (possibly mesenchyme cells) and one gut cell. Considering the large number of aboral ectoderm cells containing Cyllla · CAT genes in these embryos, the ectopically labeled cells would increase the number of cells expressing Cyllla · CAT by less than 0.2%. This insignificant level of ectopic expression is similar to that seen previously in embryos reared from singly injected eggs (data are not shown here as the result is essentially the same as in previous publications [Hough-Evans et al. 1987, 1988; 1990]; additional observations demonstrating appropriate expression of Cyllla · CAT in multiply injected eggs in which CAT enzyme is monitored immunologically are presented elsewhere [Zeller et al. in preparation]). We conclude from these experiments, and from Fig. 4, that as in mosaic embryos raised from singly injected eggs, in 3x injected eggs the Cyllla · CAT construct is correctly specified for expression, both spatially and temporally. Furthermore, in the highly transgenic embryos Cyllla · CAT is apparently expressed in all clones of which the aboral ectoderm is composed.

We assayed directly the transcriptional activity of the endogenous Cyllla genes of living blastulae in the presence of a saturating amount of exogenous Cyllla · CAT, and a several-fold molar excess of the Cyllla regulatory domain. In order to make this measurement, we injected fertilized eggs 3× with solutions containing 600 molecules/pl of linearized Cyllla · CAT and 0, 600, or 3360 molecules per pl of the 9.3 kb SalI fragment shown in Fig. 5A. This fragment is also present on the Cyllla · CAT construct (Fig. 5A) and it contains the complete known regulatory domain of the Cyllla gene. Thus the experimental samples contained a 0, 1, or 5.6-fold molar excess of competitor regulatory domain, with respect to Cyllla · CAT genes. A constant total mass of DNA was introduced in all injections; the balance was made up with size-selected sea urchin carrier DNA. Blastulae reared from these injected eggs were collected 24 h postfertilization, the average CAT enzyme activity per embryo was measured, and the average number of Cyllla · CAT genes per embryo (i.e. D) was determined by hybridization with a CAT probe as in the preceding incorporation experiments (see Materials and methods). The average fraction of nuclei per embryo containing the injected DNA (i.e. G) was also estimated as above, by in situ hybridization on embryo squashes. Finally, 728 blastulae of the same 5.6-fold molar excess competition sample were collected for the determination of their endogenous Cyllla mRNA content. As a control, 742 blastulae reared from the same batch of eggs, but which had not been injected, were collected to provide a quantitative comparison of the endogenous Cyllla mRNA content by the same methods. The value of i) for the embryos of this experiment was 90 copies of Cyllla · CAT DNA per nucleus containing this DNA, while the measurement of indicated that an average of 93% of the nuclei per embryo displayed CAT DNA. This sample of embryos was Batch 1 of Table 1, the incorporation distribution for which is shown in Fig. 2C, and the exogenous DNA content of which is compared to that of other batches in Fig. 3 (point C). We recall that the competitor fragments, the carrier DNA, and the Cyllla · CAT DNA are all concatenated together, and in earlier competition experiments were in fact shown to be incorporated at the expected ratios (Livant et al. 1988; Franks et al. 1990). Thus we expect an average of 500 molecules of competitor regulatory domain per nucleus bearing Cyllla · CATDNA in the 5.6-fold molar excess competition samples used for the Cyllla mRNA measurements. Note that Livant et al. (1988) showed that half saturation with respect to endogenous regulatory factors is achieved at about 100 copies of Cyllla · CAT per nucleus, and 90% saturation at about 1000 copies per nucleus. The total number of copies of exogenous Cyllla regulatory domain per nucleus in these experimental embryos, i.e. about 590/nucleus, should therefore induce about 85% saturation.

Fig. 5B shows the decline in relative CAT activity in these transgenic blastulae as a function of the molar excess of competitor fragments injected. The solid curve represents the predicted decline in relative CAT activity were the competition by excess competitor fragment perfectly stoichiometric (see Livant et al. 1988). The experimentally determined points are entirely consistent with this function. Thus we may conclude that competitive titration of limiting DNA-binding proteins necessary for Cyllla · CAT expression has taken place in these particular embryos, just as in the previous in vivo competition studies (Livant et al. 1988; Franks er al. 1990).

To measure quantitatively the amounts of Cyllla mRNA in the 728 embryos bearing Cyllla · CAT plus exogenous competitor, and in the 742 control uninjected embryos, we miniaturized RNA extraction and single-strand probe excess titration procedures, and added an internal efficiency control. Different amounts of total RNA extracted from these embryos were reacted with at least a 10-fold molar excess of a ³²P-labeled riboprobe specific for the 3’ untranslated region of the Cyllla message. These reactions achieved 95% completion, and the unreacted RNA was digested with RNAase. A known amount of a tritiated riboprobe approximately 1300 nt long derived from unique sequences 3’ of M actin mRNA (Lee et al. 1984) was introduced at the beginning of the isolation procedure as an internal standard, to provide a correction for any bias due to unequal yields of RNA in the preparation (see Materials and methods). In the event, an identical efficiency of RNA preparation (84 %) was obtained for both samples. The results of the experiment are shown in Fig. 6. The points representing measurements from the transgenic embryos (filled symbols) and those from the control embryos (open symbols) fit the same fine with a correlation coefficient of 0.92. Thus the Cyllla mRNA contents of the two sets of embryos are indistinguishable. In absolute terms, the number of Cyllla mRNAs is similar to that measured by Lee et al. (1986). Were the exogenous Cyllla regulatory domains competing as efficiently with the endogenous Cyllla regulatory domains for the necessary factors as they are with each other, we would expect to see the level of Cyllla transcript indicated by the dashed line in Fig. 6. The result shown in Fig. 6 was also obtained in a similar experiment in which, however, the resolution was not as good because the exogenous DNA was present in only 25 % of the nuclei. In this experiment, the slope of the control was 8% less than that of the injected embryos.

These measurements establish that the quantitative expression of the endogenous Cyllla gene remains unaffected in the same embryos in which quantitative expression of the exogenous Cyllla · CAT construct is severely depressed by regulatory competition. It remained possible that spatial expression of the endogenous Cyllla gene was perturbed in the competition embryos. An in situ hybridization study of endogenous Cyllla expression was carried out on embryos grown from 3× injected eggs as above, using the gene specific 3’ trailer sequence probe with which earlier studies of Cyllla expression have been carried out (Cox et al. 1986; Lee et al. 1984,1986). An antisense ³⁵S-RNA probe was prepared and hybridized to 5 μm sections of 72 h pluteus stage embryos, in which all of the cell types of the embryo are explicit, and the distinction between oral and aboral ectoderm is easily perceived, particularly in sagittal or parasagittal sections. In situ hybridization was performed as in Cox et al. (1986), with the minor modification introduced by Nisson et al. (1989). A sample of these embryos was removed in order to test the efficiency of the competition in this particular experiment. The competitor DNA was introduced at 5.6× molar excess with respect to Cyllla · CATDNA (cf. Fig. 5). The average CAT enzyme activity per embryo (99 embryos) was 15.8% of that in the control embryos (151 embryos) bearing Cyllla · CAT but no competitor. An ideal competition would predict 15.2% (i.e. [1 + n]^-1 where n is the ratio of competitor DNA to Cyllla · CAT DNA. Thus a highly effective competition against Cyllla · CAT expression was obtained. 46 of the same batch of embryos were scored for endogenous Cyllla gene expression. Of these 32 embryos were represented by a complete set of serial sections, and 14 were represented by between 6 and 15 sections each (the complete series, depending on orientation of the pluteus, consisted of 12–18 sections). All aboral ectoderm cells included in all of these sections were found to be labeled by the Cyllla probe; i.e. there were no ‘holes’ in the labelling pattern such as might have occurred were Cyllla expression silenced in the large patches of cells containing competitor DNA. No ectopic Cyllla expression was observed. In sagittal and parasagittal sections the oral ectoderm displayed no labelling whatsoever above background, and neither were gut, esophagus, intestine, mesenchyme cells nor coelomic pouches labeled above background. We conclude that spatial expression of the endogenous Cyllla gene is, so far as detection by in situ hybridization allows, entirely impervious to competition by exogenous regulatory DNA.

This work began with our discovery of a way to introduce exogenous DNA into much larger fractions of sea urchin embryo cells than tfie average 6 % of embryo cells that stably incorporate these sequences by our previous methods. As described in the initial sections, we demonstrate that multiple injection of exogenous DNA into fertilized eggs produces many embryos in which >95 % of cells contain the injected sequences. The new method was then applied to the problem of determining whether introduction of a several-fold molar excess of the Cyllla regulatory domain would affect endogenous Cyllla transcription, in the same embryos in which it stoichiometrically depresses exogenous Cyllla · CAT expression. The coinjected regulatory domain utilized in these experiments includes all of the ∼20 known factor binding sites for the Cyllla gene (Calzone et al. 1988; Thézé et al. 1990), and earlier evidence showed that this domain suffices to promote accurate specification of the Cyllla · CAT construct in the early embryo, both qualitatively and quantitatively. Thus the spatial expression of Cyllla · CAT is appropriately confined to aboral ectoderm (Hough-Evans et al. 1987, 1988, 1990); the temporal pattern of expression displays the same late cleavage activation and subsequent kinetics as does the normal Cyllla gene (Flytzanis et al. 1987) ; and when subsaturating numbers of these genes are introduced, the amount of Cyllla · CAT transcript per exogenous gene is approximately equivalent to the amount of Cyllla mRNA per gene in normal embryos (Livant et al. 1988). We confirm the normal spatial and temporal patterns of expression in embryos raised from eggs injected 3× with Cyllla · CAT in this paper. Nonetheless, we show that the endogenous Cyllla regulatory system and that driving the exogenous Cyllla · CAT construct are functionally different, in an unexpected way: both quantitative and spatial expression of the endogenous Cyllla gene are impervious to competition. This result is consistent with the often repeated observation that no adverse developmental phenotype is associated with introduction of competitively effective levels of Cyllla regulatory DNA (Livant et al. 1988; Franks et al. 1990; B. Evans, D. Livant, unpublished data). However, in itself the absence of a phenotypic effect is not interpretable. The other cytoskeletal actins could function redundantly with respect to Cyllla actin, and thus could have substituted for newly synthesized Cyllla protein even were the synthesis of the latter depressed; or the egg could be endowed with a sufficient maternal store of Cyllla protein. Our finding differs from that of Vitelli et al. (1988), who reported that microinjected exogenous early histone genes cause a reduction in the expression of the endogenous early H2A in the sea urchin Paracentrotus lividus. This difference might be understood in terms of the contrasting developmental roles of these genes. Early histone genes are actively transcribed in all the cells of the embryo during early cleavage, and are then ‘turned off.’ Thus they do not become committed to a state of activity. The Cyllla gene is activated in late cleavage in one particular set of cell lineages, and continues to be expressed in these lineages as long as the aboral ectoderm exists. Thus Cyllla mRNA is found throughout larval life up to metamorphosis when the aboral ectoderm cytolyzes.

The competitor regulatory and Cyllla · CAT reporter sequences are present in the nuclei from early in cleavage, i.e. long before Cyllla or Cyllla · CAT activation. This is indicated explicitly by the analysis of exogenous DNA incorporation included in this paper (see also Hough-Evans et al. 1988). The accurate temporal activation of Cyllla · CAT gene expression that we have observed (e.g. Fig. 4) is probably mediated by a steep developmental rise in intranuclear concentration of certain positively acting regulatory factors, whose function is specifically temporal control, and which service a number of different genes (Calzone et al. 1988; Davidson, 1989; Thiebaud et al. 1990). Among these is the factor P5. Competition with the P5 binding site (Franks et al. 1990), or deletion of this site (Flytzanis et al. 1987), obliterates Cyllla · CAT expression. Almost stoichiometric competition is observed when excess P5 target site alone is injected, as also on introduction of excess quantities of each of several other individual factor binding sites that compete effectively, or of the whole regulatory domain (e.g. Fig. 5B). This means that the relevant factor molecules are more or less equally partitioned amongst the competitive sites and the sites linked directly to the CAT reporter in the Cyllla · CAT gene. The factors are thus titrated away from the Cyllla · CAT genes, as a function of intranuclear [competitor:Cyllla · CAT] molar ratio. The endogenous and exogenous genes are expected to have the same initial access to diffusible regulatory factors. The fact that the exogenous genes function normally, in itself shows that these factors are indeed diffusible, since the exogenous sequences are not in general incorporated at the same genomic loci as the endogenous Cyllla genes. In contrast, once bound to endogenous gene target sites, the necessary regulatory factors apparently do not thereafter partition amongst the excess exogenous competitor sequences. Two conclusions follow. First, the immunity to competition of the endogenous genes means that factor-site interactions occurring in these genes must be greatly stabilized (probably > 10-fold) relative to the same interactions in the exogenous genes in the same nuclei. Second, the factors continue to redistribute amongst the exogenous sites during development, perhaps according to equilibrium parameters similar in value to those measured in vitro for many of these factors (e.g. Calzone et al. 1988). These measurements suggest that the specific factor-site complexes have a half-life of minutes to an hour or so. Continuing equilibrium redistribution of necessary factors amongst the exogenous genes is required. Thus, per nucleus the specific sites are always in excess in the competition embryo nuclei, and the exogenous DNA replicates successively along with the genomic DNA during cleavage (McMahon et al. 1985; Franks et al. 1988; Flytzanis et al. 1985). The difference between endogenous and exogenous genes is maintained throughout cleavage and thereafter, since we observe this difference at the 24h blastula stage. Thus the feature that enhances stability of factor-site interactions in the endogenous Cyllla genes must be one that is reproduced following mitosis.

In embryos in which exogenous Cyllla · CAT expression is stoichiometrically competed by excess regulatory target sites, both spatial and quantitative expression of the endogenous Cyllla gene appear to remain perfectly normal. This is shown by the in situ hybridization experiment which revealed Cyllla mRNA in all aboral ectoderm cells, but in cells of no other cell type or lineage, and by the mRNA titrations of Fig. 6, and ancillary experiments. We argue above that the imperviousness of the endogenous Cyllla gene to levels of regulatory competitor sequence that effectively titrate factors away from the exogenous Cyllla · CAT gene, requires an enhanced stability of the DNA-protein complexes specifying expression in the endogenous Cyllla gene. What could explain this difference between the exogenous and endogenous gene regulatory systems? The more likely interpretations fall into two classes, that in abbreviated form could be described as ‘historical’ and ‘locational’ explanations. Enhanced stability of the endogenous gene regulatory complexes could conceivably be affected by the developmental history of the target site if during gametogenesis the DNA had been modified in a way that can be reproducibly propagated at cell division, e.g. by methylation. The alternative is that the genomic locus in which the endogenous Cyllla gene resides includes a distant site (or sites) not contained in the Cyllla · CAT construct, the function of which is to ‘lock in’ the factor-site complexes required for proper activation of Cyllla expression. For example such a site could bind a protein that interacts cooperatively with other factors required for expression, or it could mediate a change in chromatin conformation that causes additional protein-protein interactions.

Observations on ‘genomic imprinting’ in the mouse provide a precedent for the first of these possibilities. Imprinting was discovered because the two parental genomes are not functionally equivalent in the mouse, in that both are required for normal development (Surani et al. 1984, 1986; McGrath and Softer, 1984), and specific chromosomal regions influence development according to their parental origin (reviewed by Cattanach, 1986). Genomic DNA is methylated differentially in male and female gametes, and these differences persist in early development (Sanford et al. 1987). Furthermore, Sapienza et al. (1989) showed that methylation of an exogenous sequence in a transgenic mouse depends on the sex of the parent from which it is inherited. Studies of Groudine and Conklin (1985) similarly demonstrated that in the chicken de novo methylation of specific genes occurs during spermatogenesis, generating developmentally persistent patterns, in that sites of undermethylation in sperm correspond to hypersensitive sites in the vicinity of these genes in somatic cells. However, sexually differential genomic imprinting is unlikely in the sea urchin, since normal parthenogenic development can be obtained, though only at relatively low frequency (Brandriff et al. 1975). Furthermore, the frequency of successful parthenogenic development could be affected by a phenomenon entirely distinct from differential genomic imprinting, viz that in sea urchins both centrosomes are normally contributed by the sperm (Boveri, 1900; Paweletz et al. 1987). The possibility remains that the genomes of both sexes are modified in sea urchins during gametogenesis, including the Cyllla regulatory sequences, but excluding the identical exogenous Cyllla CAT (and competitor) sequences because these are introduced only after gametogenesis. To test this category of ideas we are constructing germ line transgenics bearing Cyllla · CAT. In these animals, the exogenous genes will have undergone the same gametogenic ‘experience’, but are expected to be located in regions of the genome other than the normal domain of the endogenous Cyllla gene. If in embryos derived from such transgenic animals Cyllla · CAT expression proves to be impervious to competition by excess Cyllla regulatory sequences an ‘imprinting’ mechanism would be implied, while if it can be competed, a distant additional regulatory sequence at the Cyllla locus that is missing from the Cyllla · CAT construct would be indicated. In either case there evidently exists a mechanism, the specific function of which is to superimpose a developmental condition of irreversibility on the specification of Cyllla expression.

Cyllla gene expression occurs only in the aboral ectoderm lineages, and continues throughout the fifetime of these lineages. Our experiment shows that specification of Cyllla · CAT expression may occur normally, but remain abnormally reversible. Tirus this experiment empirically separates specification of expression per se from commitment, i.e. irreversibility of the state of specification (cf. Davidson, 1986, p.195; Davidson, 1990). Perhaps these results will lead to a molecular definition of commitment, as opposed to specification, for the Cyllla gene.

This work was supported by an NIH Grant (HD-05753) and by a Developmental Biology Grant from the Lucille P. Markey Charitable Trust.