Results

Purification of GAGA and histone oligodeoxynucleotides

Samples of crude, unpurified oligodeoxynucleotides were separated on a 20% denaturing polyacrylamide gel. Starting purity was estimated at 55% and 40% for the full length (n=16 residues) GAGA and histone oligomers, respectively (Figure 9 left). Purification as described in Methods resulted in oligomer purity at 90% and 95%, respectively (Figure 9 right). Final concentrations of the purified stock oligomers were determined by UV absorption measurements to be 3.748 pmol/µL and 1.829 pmol/µL, respectively.

Equilibrium binding studies

Two Drosophila genomic sequences within supercoiled plasmids were hybridized with their respective oligodeoxynucleotide designed for third-strand binding. The two target sequences we chose have potentially different binding properties. Experiments and data investigating these properties are explained here. 0.066 pmoles (0.4 and 0.2 µg of pMJ11 and pMJ10, respectively) of plasmid were incubated with increasing molar concentrations of ³²P-end-labeled third-strand (from 0:1 to 50:1 third-strand:plasmid) for at least 20 hours at 23 °C. Both third-strand:plasmid systems showed efficient binding in that length of time. Raw phosphoimager data (in pixels of brightness) was transformed into moles of third-strand by linear regression against a series of molar standards. A BESTFIT first-order curve was used as the transformation factor between pixels and moles of ³²P-end-labeled oligomer. For each experiment, the maximum possible pixel count was calculated from the total molar volume of ³²P-end-labeled third-strand added using the BESTFIT analysis. The molar concentration of bound third-strand was then determined from the phosphoimager pixel count percentage. A plot of the molar reaction concentration of added third-strand vs. the fraction of plasmid bound was used to determine equilibrium constants, i.e., the concentration of third-strand at 50% saturation.

Histone studies

Two identical independent experiments (Histone I and Histone II) were performed using the histone third-strand and pMJ11. Data for each was evaluated independently and then compared.

Histone I binding experiment:

Spot titrations (0.5, 1.0, 2.0, 4.0 µL) of unbound ³²P-end-labeled histone third-strand were imaged concurrently with the binding gel. The BESTFIT line (Figure 12 bottom left) for the transformation between moles and pixels is

where equals the oligomer molar count and equals the phosphoimager pixel count. The maximum number of pixels expected at 100% binding was calculated from the total moles of oligomer added and the BESTFIT line. Moles of oligomer bound were used to determine the fraction of plasmid bound. Measured and calculated data is shown in Table 2.

Figure 11 (top) shows the third-strand equilibrium binding curve for this trial. Two features are immediately apparent. First, between 6.7 and 53 nmolar (2.5:1 to 20:1 third-strand:plasmid) third-strand concentration, binding is directly proportional to the molar concentration of added third-strand. At higher concentrations, this relationship disappears and the target concentration becomes a limiting factor. No binding was observed when using only the 4.3 kb parent vector pBR322. This shows that the third-strand is specific for its target sequence in pMJ11 (Figure 10, lane 1).

Second, saturation is reached at approximately 75 nmolar third-strand concentration (30:1 third-strand:plasmid). At this third-strand concentration, all target sites have bound the third-strand. The maximum observed fraction of pMJ11 bound was 1.04 (± 10%) occurring at 107 nmolar third-strand concentration. K_D (at 50% binding) can be seen to be 35 nmolar.

Histone II binding experiment:

Spot titration standards, data collection, and calculations were done as described for Histone I (Table 3). The BESTFIT line used to calculate binding molar quantities is:

This second experiment provided qualitative results essentially the same as those of the first histone experiment. The main difference between the two experiments appears to be an almost 3-fold decrease in the saturation level of binding (Figure 11 bottom). How then to explain the fact that all plasmid sites were not filled, as happened in the previous trial? Visual examination of the gels stained with ethidium bromide indicated that similar amounts of DNA were in fact used. However, there is still the possibility of other systematic errors, such as bad gel imaging or inaccurate quantitation in the phosphoimager, as a cause of the observed discrepancy. Nevertheless, the two principal observations from the first histone experiment are also apparent here. The linear portion of the binding curve occurs within a similar range of third-strand:plasmid ratios (2.5:1 to 20:1). Furthermore, a plateau in the binding curve again occurs at approximately 75 nmolar third-strand concentration (30:1 third-strand:plasmid.

The negative control (data not shown but similar to Figure 10, lane 1) using pBR322 again showed no binding. Maximum binding was observed at 107 nmolar third-strand concentrations (40:1 third-strand:plasmid and 37% of the target plasmids bound). KD can be seen to be >> 32 nmolar, in good agreement with the value from the Histone I experiment.

Combined Histone I + II data:

An arithmetic mean of the two histone binding curves (Table 4) was used to minimize the effect of random or experimental errors on final evaluation of the binding. The combined equilibrium binding curve (Figure 12 top) closely resembles the two individual binding curves (Figure 11) in shape. The curve is linear to approximately 50 nmolar third-strand concentration (20:1 third-strand:plasmid) an becomes saturated at approximately 85 nmolar. Due to the small sample size (2), the standard deviation error is very large and these results should be considered only as preliminary. However, due to the consistent shape of the curve throughout the trials, it is probable that the relative relationships previously shown between equilibrium binding, saturation, and K_D are correct. Further experiments will need to be performed to find the correct equilibrium constants.

GAGA studies

The GAGA fragment containing multiple third-strand binding targets is 309 bp long. Within this region, there are 14 complete 16 bp targets separated by a 5 bp spacer. It is therefore expected that multiple third-strands will be able to bind to the fragment when present in significantly high concentration. Phosphoimager pixel volume counts were transformed to molar counts as described above. Spot titrations of 0.0625, 0.125, 0.25, 0.5, and 1.0 µL of ³²P-end-labeled GAGA third-strand were used as standards. The BESTFIT line calculated from these spots is:

Data and calculations from the GAGA binding gel (Figure 13 top) are shown in Table 5.

The fraction of pMJ10 bound by the third-strand was calculated as above. At high third-strand:plasmid ratios (> 10:1), the apparent fraction of plasmid bound by a third-strand is greater than 1. As there are 14 target sequences per plasmid, it is thought that more than one third-strand will bind per plasmid. 200% binding therefore indicates that 2 target sequences on average have been bound per plasmid. At 105 nmolar third-strand concentration (40:1 third-strand:plasmid), one-half (7 out of 14) of the possible target sequences are bound. Because of a gel loading problem, binding data at higher third-strand concentrations must await further work.

Analysis of the equilibrium third-strand binding curve (Figure 13 bottom) shows that even at these high third-strand:plasmid ratios, saturation has not been reached. The highest point in the graph indicates only 50% binding (7 out of 14 targets occupied) and the linear nature of the curve is consistent with the numerical data. In the absence of additional experiments showing higher third-strand concentrations, this data suggests a K_D of 105 nmolar for the entire system (50% binding of target sequences occurs at 105 nmolar third-strand concentration). On average, at least one target per plasmid is bound at the low third-strand concentration of 15 nmolar. Assuming the same linearity at very low third-strand concentrations, the K_D per individual target sequence is thus estimated to be in the 7-8 nanomolar range.

Figure 9

Figure 9. Autoradiograph of 32P-end-labeled oligonucleotides. (Left) Crude unpurified oligonucleotides at 30 minute exposure. Volumes added are indicated on top. (Right top) Purified GAGA oligonucleotide at 2 minute exposure. (Right bottom) Purified histone oligonucleotide at 1 minute exposure.

Figure 10

Figure 10. 32P-end-labeled Histone third-strand binding to pMJ11 at pH 5.5. Oligomer to plasmid molar ratios are indicated on top. Control Lane 1- third-strand:pBR322 (40:1); Lanes 2-8 are third-strand:pMJ11. Lane 2 - 0:1; Lane 3 - 2.5:1; Lane 4 - 5:1: ; Lane 5 - 10:1 ; Lane 6 - 20:1; Lane 7 - 40:1; Lane 8 - 50:1.

Figure 11

Figure 11. Equilibrium binding curves for (Top) Histone I and (Bottom) Histone II third-strand:pMJ11 experiments. Error bars represent an estimated error of ± 10%.

Figure 12

Figure 12. (Top) Arithmetic mean of the two histone binding curves. Error bars are ± 1 S.D. (Bottom Left) Spot titration standards curve for the Histone I binding experiment. (Bottom Right) Spot titration standards curve for the Histone II binding experiment.

Figure 13

Figure 13. 32P-end-labeled GAGA third-strand binding to pMJ10 at pH 5.5. (Top) GAGA third-strand binding gel. Oligomer to plasmid ratios are indicated on top. Control Lane 1- third-strand:pBR322 (40:1); Lanes 2-8 are third-strand:pMJ10. Lane 2 - 0:1; Lane 3 - 2.5:1; Lane 4 - 5:1: ; Lane 5 - 10:1 ; Lane 6 - 20:1; Lane 7 - 40:1; Lane 8 - 50:1. (Bottom) Equilibrium binding curve for the GAGA third-strand:pMJ10 experiment. Error bars represent an estimated error of ± 10%.