Effect of estrogen on gene expression in the chick oviduct. Effect of estrogen on the sequence and population complexity of chick oviduct poly(A)-containing RNA.

Total cellular RNA preparations were isolated from chicken oviducts at three different development stages: (a) immature chicks which were chronically stimulated with estrogen; (b) estrogen-stimulated chicks which were then withdrawn from hormone for 12 days; and (c) laying hens. Total cellular RNA containing 3'-poly(A) sequences (poly(A)-RNA) were than isolated from these preparations using oligo(dT)-cellulose chromatography. The number average nucleotide length of the poly(A)-RNA preparations in each case was approximately 2000 nucleotides. The number average nucleotide length of the poly(A) residues at the 3'-terminal end of each RNA preparation was approximately 70 adenylate residues. Complementary DNA (cDNA) copies to each preparation of poly(A)-RNA were synthesized using avian myeloblastosis virus RNA-directed DNA polymerase. The cDNApoly(A) preparations were then utilized in DNA excess hybridization experiments to analyze the complexity of the DNA sequences from which these RNAs were transcribed. Approximately 22% of each of the total cellular poly(A)-RNAs were transcribed from repeated DNA sequences (average repeat frequency of 35 copies/genome) while the remaining majority were transcribed from single copy or unique sequence DNA. It was possible to estimate the number of different poly(A)-RNA sequences per cell by analyzing the kinetics of hybridization of these cDNApoly(A) preparations to total cellular poly(A)-RNA extracts under conditions of RNA excess. The results revealed that 41% of the poly(A)-RNA from laying hen oviduct consisted of, on the average, three different sequences/cell, each of which was present in approximately 25,000 copies/cell. The remainder of the poly(A)-RNA in this tissue consisted of approximately 25,000 different sequences/cell, which were present largely in only two or three copies/cell. A somewhat similar sequence complexity was found for oviduct cells prepared from estrogen-stimulated chicks. We estimated that there were approximately 20,000 different poly(A)-RNA sequences/cell, each represented in only one to two copies/cell. However, there were five sequences which were present, on the average, in a concentration of 5600 copies/cell. The poly(A)-RNAs from hormone-wtihdrawn tissue, on the other hand, had a lower sequence complexity. There were only approximately 10,000 different poly(A)-RNA sequences/cell, each present in about three copies/cell. Furthermore, the few sequences present in a great abundance in hen and hormone-stimulated tissues were apparently absent in oviduct tissue from hormone-wtihdrawn chicks, suggesting that the intracellular concentrations of these high frequency RNA sequences are dependent on estrogen.

J. MONAHAN For each set of hybridization data, the F-test of equality of variance of a 1, 2, or 3 component hybridization curve (i.e. for n = 1, 2, or 3) was compared with an (n-1) component curve. All curves shown gave best fits of n within 99% confidence levels over a value of (n -1). of (dT)-cellulose-bound material.

Characterization
In order to make this estimate, it was necessary to determine the percentage of poly (A) in each preparation, the number average length of these poly (A) sequences, and the number average length of the total poly(A)-RNA.
The percentage of poly (A) in total poly(A)-RNA of oviducts of hen and estrogen-stimulated and withdrawn chicks was determined by hybridizing 1 pug of RNA with an excess of [3H]poly(dT) as described under "Methods." As shown in Fig.  1 and Table  I Fig. 6, A, B, and C. The data were statistically best fit by two component curves. Also included in Fig. 6 is a curve (Fig. 6D) Fig. 4, the number of nucleotides was calculated using the relationship of Spirin (28).
mogeneity and shown to contain a sequence complexity of approximately 1900 nucleotides (22). This was very close to the number average nucleotide length of our poly(A)-RNA preparations, so no correction for the effect of length of the RNA upon the rate of hybridization was required. However, there is evidence that the kinetics of hybridization between RNA and cDNA is affected by a variation in the size of cDNA (36). In order to minimize deviations from ideal reaction kinetics, all cDNA I)olY( Aj, preparations were fractionated on alkaline sucrose gradients. Only those fractions corresponding to a nucleotide size between 280 and 900 (-5.0 to 8.0 spH la.") were used in the hybridization studies described below. A second advantage in sizing the cDNA preparation before hybridization was the achievement of a final extent of hybridization reaction exceeding 90%. We have consistently observed that if the cDNA,,,, , A, was not fractionated as described above, the back hybridization with total poly(A)-RNA seldom reached a final extent of reaction greater than 60%. It is possible that the short cDN-%ow t .t 1 hybrids are not completely stable to S, nuclease under the conditions used for assay procedures. Using a cDNA,, preparation of 280 to 900 nucleotides in length, > 95% hybridization was achieved and a R& value of 6.58 x 10m3 mol s 1-l was obtained (Fig. 7) with purified ovalbumin mRNA.
The data obtained in a hybridization experiment between cDNA,o,, / A 1 and hen total poly(A)-RNA are plotted in Fig. 8. Inspection of the range of the reaction (-5 log units) revealed that different classes existed which could be separated on the basis of sequence abundance.
An ideal pseudo-first order reaction for bimolecular reaction with one of the reactants being in a large excess will have a range of reaction of only 1 '/z log units.  Table III, fifth column.) Based on a RotH value of 0.00658 mol s 1-l for the back hybridization of ovalbumin mRNA to cDNA,v, the numbers of different 1900 long nucleotide sequences in the three different abundance classes were approximately 3, 90, and 24,500.
It was also possible to calculate, based upon our previous estimates, that the hen oviduct contains 12.7 pg of RNA/cell (6), and that 3.7% of the total cellular RNA was bound to the (dT)-cellulose column (and, of this, 42.3% was poly(A)-RNA).
Thus, there were -0.2 pg of poly(A)-RNA/cell. Using this value, we calculated that there were, on an average, 25,400 molecules/cell of length 1900 nucleotides for each of the three different sequences in the first component. In the same way, the second and third hybridization components were calculated to contain 450 and 2.8 molecules of each sequence per cell, respectively. It has been previously established that the messenger RNAs for the egg white proteins are in high concentrations in the hen oviduct tubular gland cell (6). It is likely that the first hybridization component represented these messenger RNAs.
Since there are approximately 8.69 x lo* of unique sequence nucleotide pairs of DNA per haploid chick genome by simple calculation, we could demonstrate that the first, second, and third poly(A)-RNA abundance classes represented an expression of 0.00067%, 0.019%, and 5%, respectively, of the unique DNA sequences present in the hen (Table III).
Similar studies were carried out with [SH]cDNA,,,,, Aj synthesized from poly(A)-RNA prepared from chicks stimulated with diethylstilbestrol for 14 days. Again, a three component curve provided a best fit for the data and indicated Rots values of 0.805, 33.10, and 4085 mol s l-l, respectively, for the three classes (Fig. 9). Again correcting these data for the actual percentage of poly(A)-RNA in the preparation and the fraction of the total which each component represents, the three abundance classes correspond to 5, 330, and 20,300 different nucleotide sequences, respectively. The RNA content of an estrogen-stimulated oviduct cell was calculated to be 9.23 pg/cell (6). Using similar calculations to those described above   in a region toward the 5'-terminal end of the poly(A)-RNA and was, therefore, not transcribed into cDNA,,,,, Al, then the Rot,+ value would be shifted to a higher value leading to an overestimation of the base sequence complexity. Dina et al. (45) and Firtel and Lodish (46) have suggested the presence of such a 5'terminal repeated sequence segment in Xenopus and Dictyostelium mRNAs, respectively. However, Campo and Bishop (14) have been unable to detect such "mixed molecules" (i.e. composed partly of repeated sequence transcripts and partly of nonrepeated transcripts) in mRNA preparations from rat myoblast cells. Instead, it appears that mRNA sequences that hybridize to repeated DNA sequences represent a distinct molecular entity containing repeated DNA sequence in their entirety.
Some mRNAs (e.g. histone mRNA) lack a poly(A) tail (47). Such species would not be transcribed into cDNAs and would not, therefore, take part in the above hybridization reactions. They will also lead to a slight overestimation of the base sequence complexity.
The heterogeneity of the RNA populations, i.e. not all of the RNA sequences being present in equal proportions, presents a second complication in interpreting hybridization with   (14,23,37). The base sequence complexity, therefore, of total cellular poly(A)-RNA isolated as we described clearly represents a minimum value for the base sequence complexity of total oviduct cellular mRNAs and HnRNAs.
With the above limitations in mind, it is possible, however, to reach some general conclusions concerning the base sequence and abundance complexities of poly(A)-RNAs in the whole cell of oviduct tissue. In hen oviduct tissue, three distinct transitions were seen in the hybridization curve (Fig.  8) to a large segment of the single copy DNA and would be present in the tissue at less than 1 molecule/100 cells. The excellent correlation between our kinetic data and previous DNA saturation experiments (21) suggests that we are not grossly underestimating the sequence complexity of the total poly(A)-RNA.
When comparing the number of different poly(A)-RNA sequences present in oviduct tissue with that of RNA from tissue culture cells (14)(15)(16)(17)(18)(19), it should also be remembered that were present in less than a few copies per cell. It is tempting to speculate that these RNAs were mRNAs coding for the egg white proteins, since they were not produced in hormone-withdrawn tissue (1,4,6). The reason for the slight deviation of the computer-generated curves at low R,t values from the data points in Figs. 8 (12,13). Using this method, it was possible to demonstrate that the estrogen-induced increase in oviduct chromatin transcriptive activity (12,13) was due to an increase in the number of initiation sites for RNA polymerase on the chromatin (12,13). In Table  IV