WO1985000813A1 - Nucleic acid branched junctions with precisely defined migrational mobility - Google Patents

Abstract

Nucleic acid branched junctions with precisely defined migrational mobility comprising new compositions of matter containing semi-mobile and/or immobile branched nucleic acid junctions from which at least 3-double helices emanate are described. These compositions of matter can comprise integral parts of periodic or other networks having precise molecular dimensions, thereby permitting the constitution of exact molecular architecture on the 100 to 1,000 Angstrom scale. They can be made by dissolution of at least three polynucleotides possessing minimal sequence symmetry with each other.

Description

NUCLEIC ACID BRANCHED JUNCTIONS WITH PRECISELY DEFINED MIGRATIONAL MOBILITY

This application is a continuation-in-part of application Serial No. 519,928 filed August 3, 1983.

This invention relates to a new composition of matter comprising a polynucleotide containing at least one immobile or semi-mobile branched junction, and to the method of making it.

Background Art Naturally occurring oligomeric nucleic acids form mobile linear duplexes stabilized by hydrogen bonds between bases. Polymeric nucleic acids occasionally branch to form junction structures in nature, as shown in Figs. 1 and 1A of the drawings. These branched duplexes are unstable, and resolve to form two independent duplexes. The pre-eminent structural characteristic of stable double helical nucleic acids in nature is that the positions of all atoms in the molecule bear a well-defined relationship to a linear (although not necessar ily straight) axis which exhibits no junctions (branch points). Nevertheless, conformational variability (Kim, S.H.,

Berman, H.M., Seeman, N.C. and Newton, M.O., Acta Cryst b29, 703-710 (1973)), and backbone flexibility (Sarma, R.H., Nature London, 263, 567-572 (1976)) permit the formation of junctions which are crucial to the biological role played by nucleic acids. A replicational junction, diagram in Figure 1A is implicit in the original proposal of Watson and Crick, Nature London, 171, 737-8 (1953) for the mechanism of DNA replication. The holiday structure indicated in Figure 1, Genet Res. 5, 282-304 (1964) is a critical intermediate in genetic recombination (Broker, T. and Lehman, I.R., J. Mol. Biol. 60, 131-149 (1971)). In addition the Platt, J.R., (Proc. Nat. Acad. Sci. (USA) 41, 181-183 (1955)) and Gierer, A., (Nature London 212, 1460-1461 (1966)) cruciform structures, closely related to the holiday structure, may play an important role in the regulation of gene expression (Gellert, M., Mizuuchi, K., O'Dean, J.H. Ohmor, H. and Tomizawa, J., Cold Spring Harbor Symp. Quant Biol. 43, 35-40

(1978)). Other types of junction structures are involved as intermediates in single-strand-displacement recombination and as transcriptional intermediates, such as that shown in Fig. 1A. Heretofore, it has not been possible to study the structural and dynamic properties of these junctions in oligonucleotide model systems, where the junction will contribute a significant signal. This is due to the existence of sequence symmetry as illustrated in Figures 1 and 1A of the drawings. The strands there shown are unlikely to form junction structures in preference to double helices; if they did occasionally combine to form such structures, the process of branch point migration, shown in Fig. 1 will result in the rapid resolution of the junction structures into double helices (Thompson, B.J. Camien, M.N., and Warner, R.C., Proc. Nat. Acad. Sci. (USA) 73, 2299-2303 (1976)).

Disclosure of Invention The synthetic oligomeric nucleic acids of this invention are so constructed as to form migrationally immobile or semi-mobile branched junctions. These new stable oligomeric structures, stabilized by maximizing Watson-Crick base pairing, minimize the sequence symmetry found in their unstable analogs in living systems. The synthetic oligonucleotide sequences of the compositions of this invention contain migrationally immobile and/or semi-mobile junction structures. In a semi-mobile junction a limited degree of configurational degeneracy is introduced into the system. These DNA junctions represent nexi, from which 3 to 8 double helices may emanate. Each junction of these compositions may be treated as a macromolecular "valence cluster" containing individual clusters which may be linked together directly, or with segments of linear DNA interspersed between them. The covalently linked compositions can be formed with a high degree of specificity, using the state-of-the-art sticky-ended ligation techniques currently employed in genetic engineering. The covalently joined three-dimensional new networks of nucleic acids containing immobile or semi-mobile junctions are periodic in connectivity and may also be periodic in space, thereby generating quasicrystalline arrays of matter. All compositions are described herein in terms of DNA, but they also include RNA, RNA-DNA-Hybrids or nucleic acids in which the backbones or bases have been modified, but not so as to affect their pairing capabilities, via Watson-Crick or some other form of association. The junctions are then formed simply by dissolving the selected sequences in the desired proportions in a suitable liquid solvent, preferably an aqueous buffer, at a temperature from about 0° to about 60°C, usually at 20°-40°C, and at pH 6-9 and in the presence of a counterion. In the drawings. Fig. 1 is a schematic representation of a portion of a polymeric nucleic acid containing an unstable, mobile junction as it occurs in nature;

Fig. 1A is a schematic representation of a replicational junction;

Fig. 2 is a schematic representation of a synthetic oligomeric nucleic acid containing an immobile junction which is one embodiment of the invention; Fig. 3 is a view of a stained gel chromatogram sheet containing the composition of Fig. 2 in lane K; Fig. 4 is a view of a stained gel chromatogram sheet containing the composition of Fig. 2 in lanes A to E, inclusive; Fig. 5 is a graph showing ultraviolet absorption of the composition of Fig. 2, and of strands 1 and 2 and of strand 3, at varying temperatures;

Fig. 6 is a flow chart showing the steps for optimizing sequences of synthetic oligomeric nucleic acid strands for making compositions of the present invention; Fig. 7 illustrates an odometer analogy to an algorithm used in the optimizing procedure of Fig. 6; Fig. 8 is a schematic representation of a composition which is a second embodiment of the invention;

Fig. 9 is a view of an autoradiogram of a gel chromatographic sheet containing the composition of Fig. 8 in lanes G and H;

Fig. 10 is a schematic representation of compositions of the invention containing immobile junctions of Rank 3 to 6; Fig. 11 is a view of an autoradiogram of a gel chromatogram containing varying amounts of the composition of Fig. 8 in lanes B-E and in lanes G-J respectively; Fig. 12 is a schematic representation of a composition according to the invention containing a plurality of immobile junctions interconnected to form a lattice or network;

Fig. 13 is a schematic representation of a composition according to the invention containing a semi-mobile junction which is capable of existing in only the two interchangeable or isomeric forms shown;

Fig. 14 is a schematic representation of a composition according to the invention containing a Rank 3 immobile junction and of a composition containing a plurality of such junctions interconnected to form a lattice or network; and

Fig. 15A shows a portion of the NMR spectrum of the composition of Fig. 14; Fig. 15B shows the corresponding portion of the. NMR spectrum of a single strand of the composition of Fig. 14. The construction of new immobile junction compositions of this invention requires the ability to identify and select unique sets of sequences possessing conventional Watson-Crick base pairing patterns while at the same time minimizing sequence symmetry. Sequences containing long sequences of base pairs can be formed at higher temperatures, up to about 60°C or even higher. The probability of forming a desired junction is a function of the free energy of association of the individual strands involved. For design purposes, each strand which is chosen to participate in the formation of an immobile or semi-mobile junction can be considered to be composed of a series of overlapping segments of a given criterion length, Nc. For example, each hexadecameric strand in the immobile junction shown in Figure 2 is a series of 13 overlapping segments of length 4. Each of these segments is termed a "criton", while the complement to a criton, i.e. the sequence of bases with which it pairs, is termed its "anti-criton". Watson-Crick pairing arrangements which compete with the desired pairing must be considered from a thermodynamic point of view for lengths less than Nc. However, if the rules indicated below for minimizing continuous lengths of sequence symmetry are obeyed, there will be no competing Watson-Crick pairing interactions for segments of length Nc or longer. Two further terms must be defined to facilitate this discussion. A "bend" is a phosphodiester linkage which is flanked by bases paired to different strands. The "rank" of a junction is the number of the double helices which directly abut it. Thus, the junction shown in Figure 14 is of Rank 3, while those shown in Figures 2 and 8 are of Rank 4.

In order to have the new compositions of matter of this invention, i.e. uniquely paired stable structures with non-migratory junctions (for length Nc or greater), the following rules must be obeyed within the pairing regions prescribed by the architecture of a given junction. 1. Every criton in the individual strands forming the junction must be unique throughout all strands, regardless of frame. 2. The anti-criton to any criton which spans a bend in a strand must not be present in any strand, in any reading frame.

3. Self-complementary critons are not permitted. if Nc is an odd number, this injunction holds for all critons of size (Nc + 1).

4. The same base pair can only abut the junction twice. If it is present twice, those two occurrences must be on adjacent arms. The first three rules ensure immobility through lack of homology, except for Nc-1 base pairs on each arm belonging to the criton nearest the junction. The fourth restriction eliminates migratory possibilities for these bases; it should be applied sequentially to each base pair in the criton, and perhaps beyond. The fourth rule also limits the maximum rank of junctions: Since there are only four base pairs, A-T, T-A, G-C, C-G, and since each pair can only appear twice, junctions of rank greater than eight are not possible with the conventional bases.

These rules are limited to those cases in which Nc is less than the minimum length for stable stem-loop structures. In the unlikely event that this is not the case, further restrictions would be necessary to forbid those critons capable of forming stem-loop structures from spanning the bends which flank the junctions. Similarly, if one of the strands is not continuous at the junction, as in replication forks, the base pairs on either side of the break must be different, to avoid the sort of migration observed by Nilsen, T. and Baglioni, C. (J. Mol. Biol. 133,

319-338 (1979)). Nc is a number to be minimized, since this minimizes the strengths of competing interaction by shortening the lengths involved. The most likely reason for increasing Nc is that the desired junction cannot be generated by the 4^Nc critons available for a given value of Nc. For example, the constraints applied to the generation of the immobile junction in Figure 2 are incompatible with a value of Nc less than 4.

It is easy to generate a junction in which the limited amount of mobility is non-zero, i.e., a semi-mobile junction, such as the one shown in

Figure 13. In order to do this, the mobile bases and the phosphates which flank them must be considered part of the bend; thus, the base pairs which flank the mobile bases are now considered to abut the junction. Once this modification of the concept of a bend is in effect, the same four rules apply. Clearly, it is not possible to have more than Nc-2 bases since bends will not be properly spanned if this is not so. This junction may undergo the reactions indicated, but may not go beyond them and resolve into two linear duplexes. Thus, it constitutes a simple flip-flop. The rules of migration are satisfied for the two states shown, but the rules for non-migration come into play for any further migratory events in either direction. A FORTRAN computer program is described hereafter which generates sequences that fulfill these criteria for junctions of any design by a rapid algorithm. The program also ranks generated sequences on the basis of pairing fidelity relative to competing interactions at lengths shorter than Nc. The details of the algorithm and the way in which free energy considerations are taken into account for competing Watson-Crick pairing interactions (at lengths shorter than Nc) will be discussed elsewhere in this description.

Junctions and Networks The ability to construct junctions of rank N (N=3, 4, 5, 6, 7, 8) makes it possible to construct highly specific geometrical figures in which such junctions correspond to the vertices, while stretches of linear duplex nucleic acid correspond to the edges. These include individual polygons and polyhedra, as well as infinite N-connected networks and polyhedra

Wells, A.F., (Three Dimensional Networks and Polyhedra, John Wiley & Sons, New York (1977)) of double helical nucleic acids in two or three dimensions. These networks may be periodic or non-periodic. Examples of 3, 4, 5, and 6-connected three-dimensional periodic networks are shown in Figure 10. This construction can be accomplished by using the conventional sticky-ended ligation technology. That methodology involves the use of sequences in which the junction crossroads structure does not terminate in a "blunt ended" fashion, as shown in Figure 2. Instead, at the free end of each double strand, remote from the junction, one strand extends beyond the end of each double helix, so that a single stranded region is dangling off the end. The specificity of double helical Watson-Crick base pairing is then utilized to link up two different pieces of DNA possessing complementary sticky ends to produce a composition containing two or more immobile or semi-mobile junctions. Among the 2- and 3-dimensional networks which are possible, some are of course periodic in their connectivity. Such networks can also be periodic in space and can then comprise unique crystalline macropolymers of a size suitable for diffraction analysis using x-rays, and perhaps neutrons. An example of such a 2-dimensional network is shown in Figure 12, illustrating formation of a twodimensional lattice from an immobile junction with sticky ends. A is a sticky end and A' is complementary to it. A similar relationship exists between B and B'. Four of the monomeric junctions on the left are complexed in parallel orientation to yield the structure on the right. If the inter-junctional spacing is large enough, a ligase would be able to close the overlapped gaps to make the complex on the right a covalently bonded structure. Note that the complex has maintained open valences, so that it can be extended by the addition of more monomers. This procedure is not limited in theory to rank-4 junctions, nor is it limited to two dimensions. The relative orientation of successive junctions is a function of both junction structure and the separation of junctions, since the connecting segments between junctions are helical. By exploiting this fact it should be possible to construct 3-dimensional nucleic acid networks analogous to the networks of Figure 10, wherein the arms are composed of double helical DNA of defined length and sequence, represented by the straight lines.

Overall Strategy For Optimized Junction Sequence Generation

The architecture of a junction requires the specifications of both covalent connectivity and base pairing relationships. Because of the complementary nature of the Watson-Crick double helices which constitute the junction structure, only half of the bases must be treated as independent variables; those bases complementary to them are treated as dependent variables. Bases are encoded as numbers on to base 4. In the case of semi-mobile junctions only one out of four of the mobile bases is independent. With a computer, new sequences may be generated simply by the process of counting in base 4. If all of the arms of a junction have the same length, it is possible to fix one base at the numerical level, thereby decreasing the number of independent variables by one. The independent bases may be ordered by the rapidity of the rates of change of the digits representing their identities within the program. By "order" is meant an inverse measure of the rate at which the digit representing the base is incremented. Thus, the lowest ordered base will be changed on every pass, the next lowest ordered base will change on every fourth pass, the next lowest ordered base on every sixteenth pass, and so on. The critons themselves may be ordered according to the lowest ordered base within the criton. The critons are then tested for adherence to the rules sequentially, from highest to lowest order. Thus, if a given criton violates one of the rules, the base corresponding to the order of that criton is advanced, rather than the base of the lowest order.

Until a base at the order of violation has been changed, no changes at lower orders would correct the existence of the violation. When a base of any order is incremented, those bases of order lower than that of the incremented base are, of course, set to their lowest value. This algorithm will be easier to understand if it is noted that the procedure is analogous to the generation of configuration of numbers with defined properties, using an odometer or crowd counter, as indicated in Fig. 7. In that figure, the uniqueness of each digit is the specific property required for the numerical configurations. This property is similar to the properties involved in the criton rules for junction formation. If a program operator starts at the top of the figure, with six zeroes as an initial configuration, and increments the most rapidly changing digit sequentially, as shown on the left, it will take 12,345 steps to get the first successful numerical configuration. On the other hand, if the operator corrects the highest ordered digit which is violating the uniqueness rule, that indicated in the 10,000's place, and then proceeds accordingly, as shown on the right, it will take 15 steps to reach the same point. The way in which this algorithm is applied to the junction generation is indicated in Figure 6.

The nine logical steps in this procedure are indicated schematically in Fig. 6. The two steps in double boxes must clearly be done by the programmer while the other steps are done automatically by the programs. In the first step, the covalent connectivity and desired base pairing are selected by the programmer. Specific constraints can also be introduced at this stage. In the second step, the critons are ranked by the order of the most rapidly changing base which they contain. After that, an initial numerical sequence must be assigned. (Step 3) This numerical sequence is tested against the junction rules (Step 4) and if it fails, a new sequence is generated by the fast algorithm. If the numerical sequence obeys the rules, its base permutations are the tested, (step 5) against investigator selected constraints. If any of the eight sequences implied by the numerical procedures are acceptable, their fidelities are calculated. (Step 6), and if these are acceptably high, melting curves are calculated and plotted, (step 7). New sequences are then generated (step 8) and tested iteratively. until all possibilities have been exhausted. The programmer may then evaluate the alternatives presented by the programs. Once the strand sequences fulfill the uniqueness and nonmobility rules, thermodynamic criteria-are applied to all sequences of length less than Nc. The first question to consider is the pairing fidelity: Is the desired base pairing configuration the most probable configuration in which these particular sequences are to be found in solution? If so, what is its probability relative to other pairing configurations? This problem is treated in a pairwise fashion; the program routinely considers all alternative binary base pa ir ing conf igurations for lengths less than Nc. The stability of an olignucleotide duplex depends on its chain length, sequence and concentration, as well as on environmental variables, such as pH, ionic strength and temperature. Data on the relative stabilities of oligoribonucleic acids in conditions equivalent to 1 M NcCl, ρH7, have been accumulated by Tinoco and his co-workers. The effects of sequence can be evaluated in terms of units representing adjacent sets of two base pairs; the equilibrium constants corresponding to the association within each unit are available, as is the nucleation constant for initial strand interaction. This is denoted by β, with units M^-1. A given sequence will then be paired with its complement by a weighting factor that depends on the product of a set of numbers:

K_AB = βK₁ K₂ K₃ ... K_N-1, (1)

where N is the chain length of complementary sequences between chains A and B, and β is the nucleation constant.

The values of the K. are tabulated at 25°C by Borer et al. as:

K_i = exp ( - Δ G_i/RT). To illustrate the use of equation (1), consider the RNA tetramer (5') AGCU (3')

(3') UCGA (5') to be decomposed into the three subunit "pairs",

(5') AG GC CU (3')

(3') UC CG GU (5')

each of which has an approximate equilibrium constant assigned.

In this way, the maximum concentration of paired molecules of a duplex or arbitrary sequence can be predicted. The situation for oligodeoxynucleotides is less completely defined than for oligoribonucleotides. However, thermodynamic data are available from which primitive sets of K_{i, s} can be created, together with rough values of the ΔH°_{i, s}. Despite the uncertainties, these data make it possible to estimate the relative σontr ibutions of different sequences with reasonable accuracy, particularly if appropriately scaled values from oligoribonucleotides are used. For scaling, we alter the ΔG° values of Borer et al., so as to lower the stability of the corresponding oligodeoxynucleotide by 20°-45°C relative to the oligoribonucleotide. Comparison of the values of K_AB for each set of interactions below the criton length then permits one to estimate the relative contributions of the base pairing in each case. Junctions of maximum fideltiy will be those that contain sub-criton pairing sequences of minimal stability, relative to the stability of the complete arms. All binary Watson-Crick alternatives are checked by the program, and their stabilities are compared with those calculated for the double heices chosen for the architecture of the junction. The highest probability junctions above a selected fidelity minimum are retained for further processing. It should be pointed out that fidelity is a function of temperature. Thus, sequences must be compared for relative fidelity at a standard temperature, for which we use 25°C.

The information contained in the estimated equilibrium constants for pairing specific sequences can be used to predict approximate thermal transition profiles for junctions. In order to do this, enthalpy values, ΔH_i° corresponding to the equilibrium constants K_i, used to assess fidelity, are required. These are considerably less certain for oligodeoxynucleotides than for oligoribonucleotides, but nonetheless reasonable estimates are available, and missing values can be filled in by scaling the corresponding RNA data, as described. In the case of pairing between sequences on twδ non-identical strands. A and B, the value of K_AB and the starting concentrations of the two species uniquely characterizes the equilibrium; for starting concentrations, C_A and C_B (moles per liter).

We have discussed how to approximate K_AB; thus, C_AB can be calculated. This can be done at any temperature if the ΔH_i° for each K_i is known.

Consider next the interaction of four oligomers. A, B, C, and D, which contains uniquely complementing half sequences that can lead to formation of a 4th rank junction complex. Since at equilibrium the concentrations must be independent of the reaction pathway, it is sufficient to calculate the junction concentration resulting from any one pathway. For example, one might select:

A + B = AB (i)

C + D = CD (ii)

AB + CD = ABCD (iii)

From the values of K_AB, K_CD , and introducing a new factor, σ_R to describe the statistical weight of the central junction "loop" structure of rank R, then concentrations of junctions can be expressed in terms of known quantities. That is, C_AB and C_CD can be calculated by solving equation (2), and these values can be introduced into reaction (iii) above to give:

The value of K_ABCD is estimated as :

K_ABCD = β^-1 ((σ_Rk_BCk_DA)+k_BC+ k_DA).

This is very nearly equal to β^-1 σ_R K_BC K_DA, since σ_R is not expected to be very different from unity, while the K's are large at low temperature. It is possible that if σ_R is much less than β, only negligible concentrations of the complete junction will be detectable, as discussed more fully below. If a junction entails no strain, we anticipate that only a simple Jacobson-Stockmayer term (Jacobson, H. and Stockmayer, W. (J.Chem.Phys. 18, 1600-1606 (1950)) is involved:

σ_R = σ_O (R)^-a,

where the factor ^σ reflects the difficulty of forming the junction and the exponent a (1.5 < a < 2) is a measure of the excluded volume. Concentrations of the ternary and higher (for R > 4) intermediates can be calculated, using stepwise paths such as:

AB + C = ABC, ABC + D = ABCD.

Thus, the equilibrium concentration of each intermediate, as well as the junction itself can be calculated; a series of relations exists among these intermediate of the form:

C_AB + C_BC = C_ABC + C_B'

which simplifies the problem considerably for this approximate treatment.

In each of the examples set forth below oligodeoxynucleotide strands synthesized from appropriately blocked nucleotides or dinucleotides by standard chemical procedures are placed in a glass or plastic vessel at room temperature in a solvent containing (1) a buffer system capable of regulating pH between pH6 and pH9, so as to favor Watson-Crick base-base hydrogen bonding and (2) a source of counterions in order to reduce the electrostatic repulsion among the strands as they form a ternary, quaternary (or higher) junction or complex. Suitable buffers include phosphate, cacodylate, tris, etc. at concentrations from 0.001 to 1.0 M approximately. In the absence of buffer, the pH can be regulated by direct titration to yield an end-point in the above range; the pH so obtained is less stable to temperature and electrophoresis. Satisfactory counterions have been found experimentally to be (a) 1-2 molar NaCl, KCl, CsCl or any non-destabilizing monovalent neutral salt including Na₂SO₄, K₂SO₄, Cs₂SO₄, (b) 5-10 mM Mg₂Cl, Ca-Cl or comparable divalent neutral salt or (c) 1-5 mM spermine, spermidine or other neutral alkyl diamines NH₃+-R-NH₃+ or (d) combinations of the above agents. Lack of suitable counter ions destabilize the junctions selectively relative to simpler structures (dimers, e.g.).

Complex or junction formation can be shown to occur at temperatures below about 40°C (see the denaturatlon profile monitored by ultraviolet absorbance in Fig. 5). The junctions or complexes are stable below 0°C also. As described in the examples experimental criteria for existence of a stable base paired complex or junction include:

(1) hypochromicity of the strong (˜260nm) ultraviolet absorbance of the bases;

(2) enhanced circular dichroism of the bases in a by duplex or higher complex;

(3) characteristic electrophoretic mobility and

Ferguson behavior of the complex;

(4) existence of new proton NMR resonances corresponding to hydrogen bonds at the junction center.

Each has been applied to monitor the behavior of the complex shown in Fig. 2. Example 1 The four blunt-ended oligomeric helical strands of DNA designated in Fig. 2 were selected by means of the rules and algorithm described above, then were synthesized by conventional phosphotriester technique on a commercially available synthesizer from appropriately blocked nucleotides or dinucleotides.

Note the lack of symmetry around the center of Fig. 2., so that migration is not possible without disrupting pairing. This sequence also contains no repeating GpG sequence longer than two, in order to restrict the possibility of GG non-Watson-Crick pairing as well. These sequences were lyophilized, and equal weights of the four strands were then dissolved to form a 2 mM solution in an aqueous buffer (50 mM phosphate or tris pH 7; 10 mM MgCl₂) at room temperature, whereupon the immobile junction of Fig. 1 formed spontaneously. The buffer solution was subjected to electrophoresis on polyacrylamide gel (10-15% acrylamide) along with solutions of other materials in separate lanes which provided mobility references. The chromatogram was stained with a conventional dye (stains-All, Kodak) which colors single as well as double stranded nucleic acids; the results were as shown in Fig. 3, in which lane K contained the desired composition having an immobile junction as shown in Fig. 2.

In preparing the chromatogram, each of the other wells or lanes contained other materials to provide mobility references only. Lanes a, b and 1 contained restriction digests of PhiX-174-RF-DNA; a is the Hinf I digest, b is the Hae III digest and 1 is the Hinc II digest. The lowest molecular weight fragments in these digests are: 42,48,66 and 82

(Hinf I), 72 and 118 (Hae III) and 79 and 162

(Hinc II). Lanes c-f contained strands 1, 2, 3 and 4 respectively. Lane g contained an equimolar mixture (based on extinction coefficients derived from dry weights) of strands 1, 2, and 3; lane h, strands 1, 2, and 4; lane i, strands 1, 3, and 4; and lane j, strands 2, 3, and 4. Lane k contained an equimolar mix of strands 1, 2, 3 and 4. Lane m to r contained equimolar mixtures of pairs of strands: m contains 1 and 2, n 3 and 4, o 1 and 4, p 2 and 3, q 1 and 3, and r 2 and 4. Lane k of Fig. 3, containing an equimolar mixture of all four strands shown in Figure 2, travels as a single band with a mobility lower than any of the other oligomeric mixtures to be seen on this gel. The highest mobility band in this digest corresponds to a linear duplex DNA molecule containing 79 base pairs. The mobility of the band in lane K, containing the desired immobile junction containing 32 base pairs, corresponds to a linear DNA duplex of 44 base pairs, measured against standards not shown here. The presence of a single band with appreciable mobility in lane K containing the tetrameric complex or junction indicates that a molecular species with all four strands is present.

In addition, as shown in Fig. 4, electrophoresis of mixtures containing different ratios of two components was carried out; component (i) consisted of an equimolar mixture of strands, 1, 2 and 4 of Fig. 2 while component (ii) consisted of strand 3 alone, each in a different aliquot of the same buffer. Lanes G to J contained 8 micrograms of free strands 4, 3, 2 and 1, respectively. Lane F contained 6 micrograms of component (i). Lanes A to E each contained 6 micrograms of (i), and in addition, 0.5 micrograms (E), 1.0 micrograms (D), 2.0 micrograms (C), 3.0 micrograms (B), and 4.0 micrograms (A) of component (ii). The resulting chromatogram, after staining, is shown in Fig. 4. Each of lanes A to E contained, in varying amount, a composition containing the immobile junction of Fig. 2.

It can be seen from lanes A through E in Figure 4 that the intensity of component (i) varies with the ratio of (ii) to (i). It appears that component (i) is being chased into the higher bands as more of (ii) is added. The simplest explanation for this is that a 1:1 complex between (i) and (ii) is being formed.

Three buffer solutions were also prepared, the first containing an equimolar mixture of strands 1, 2, 3 and 4 of Fig. 2 (25 μM per strand), the second containing an equimolar mixture of strands 1 and 2 (49 μM per strand), and the third containing strand 3 (98 μM), and the ultraviolet absorbance of all three was measured at 260 nm over a range of temperatures, and the result expressed as ΔA₂₆₀=(A₂₆₀ (T)/A₂₆₀ (10°C)-1). The resulting thermal transition profiles are shown in Fig. 5. Typical ΔA₂₆₀ for high molecular weight DNA duplexes approaches 30%. Actual values depend on base composition, length and solvent. Strand 3 alone exhibited a typical non-cooperative transition characteristic of nucleic acids in the absence of base pairing. The fact that the final ΔA₂₆₀ for the four-fold complex containing the immobile junction is roughly equivalent to twice that (per mole strand) of the individual pairs suggests that the pairwise arms of the four-fold complex form additional structure of about equal extent.

These thermal denaturation results are a measure of relative stability of the different complexes of strands 1 to 4. Base paired nucleic acid duplexes are hypochromic (absorb less), relative to single strands or a mixture of their constituent mononucleotides, at wave lengths near 260 nm. From the increased hypochromism in the mixture of all four strands, it is concluded that approximately twice as much pairing pccurs in this mixture as occurs in the same concentrations of the pair 1 + 2. The results of these spectrophotometric experiments demonstrate that the oligonucleotides associate pairwise when mixed in that fashion; the form of association is clearly base pairing, as indicated by the magnitude and cooperatively of the hypochromism upon melting. Furthermore, it is concluded from the uniphasic melt of the tetrameric complex that a junction has indeed formed as in fact was predicted from the sequence selection algorithm, indicating that the base pairing associated with the formation of the junction has indeed occurred. The composition containing the immobile junction of Fig. 2 was obtained in solid crystalline form by cooling to 4°C, a buffer solution containing 1 mM of each of strands 1 to 4. The crystals were separated from the solution by filtration, and were found to melt as the temperature approached 40°C.

Example 2 The four dodecanucleotides depicted in Fig. 8 were designed by the use of the sequence symmetry minimization rules, supplemented by equilibrium distribution optimization as explained above. Note the lack of symmetry around the center so that migration is not possible. These sequences also contain no repeating GpG sequence longer than two, in order to minimize competition from this form of non-Watson-Crick pairing as well. These sequences were synthesized by conventional phosphotri ester techniques.

The four strands were then end-labelled with P radioactive phosphate at the 5' terminus. Solutions containing individual labelled strands as well as mixtures were then subjected to electrophoresis in separate lanes on polyacrylamide gel, as follows, each well containing 2.0 μg of each strand. Lanes A-D contain strands 1, 2, 3 and 4, respectively. Lane E contains an equimolar mixture (based on extinction coefficients derived from dry weights) of strands 1 and 4; lane F contains a similar mixture of strands 2 and 3. Lanes G and H each contain equimolar mixtures of strands 1, 2, 3, and 4. An autoradiogram was then made from the resulting chromatogram, as shown in Fig. 9. The composition containing the immobile junction of Fig. 8 appeared in lanes G and H and travelled as a single band with a mobility lower than that of any of strands or mixtures of the other lanes.

In interpreting Figure 9, it should be emphasized that the sequence optimization procedures described above are designed to make the junction the preferred structure when all four strands are present. Thus, the lack of self-association for individual strands has not been an explicit criterion, except insofar as it detracts from the formation of the actual junction. Strand 4 aggregates rather strongly (lane D) , but no band corresponding to the mobility of strand 4 alone results when strand 4 is mixed with equimolar quantities of strands 1, 2 & 3 (Lanes G & H). The mobility of the slowest band in lanes G and H corresponds to a linear DNA duplex of 30 base pairs based on calibration with xylene cyanol FF.

The presence of a single band with appreciable mobility in the lane corresponding to the tetrameric complex indicates that a molecular species with a well-defined stoichiometry predominates. Clearly, some dissociation of the complex occurs as well, in contrast to the situation with a larger junction involving hexadecanucleotides. Higher unclosed complexes (1:2:3:4:1:2:3:4:1...) do not represent a significant fraction of the total material present.

Additional solutions containing varying quantities of the labelled and unl-abelled dodecanucleotide strands of Fig. 8 in various combinations were subjected to electrophoresis on polyacrylamide gels, with the results shown in the autoradiogram of Fig. 11, as follows. It shows the stoichiometry of the complex; more specifically, it shows that radioactively labelled ternary complexes are chased into immobile junction bands by the addition of the appropriately complementary cold (unlabelled) missing component. Lanes A-E represent an experiment in which the only radioactive material was strands 2, 3, and 4, each of which was present in 0.6 μg quantities. Unlabelled strand 1 was present, respectively, in quantities of 0 (A), 0.15 μg, (B) 0.3 μg. (C) 0.6 μg, (D) 1.2 μg (E). The counts in the top (junction) band are maximized in lane D, indicating 1:1:1:1 stoichiometry. Lanes F-J represent a similar experiment in which the unlabelled strand was strand 4. Labelled strands 1, 2, and 3 were each present in 0.6 μg quantities in each of lanes F-J. Unlabelled strand 4 was present in quantities of: 0 (F), 0.15 μg (G), 0.3 μg (H), 0.6 μg (I) and 1.2 μg (J). Again, the junction band saturates at 1:1:1:1 stoichiometry, corresponding to lane I.

Example 3 The three strands shown in the upper part of

Fig. 14 of the drawing were designed in accordance with the rules, and algorithm set forth above and were synthesized by conventional phosphotriester technique. The vertical-appearing strand (containing 18 residues) was synthesized with blunt ends, while the ones at lower left (containing 20 residues) and at lower right (containing 22 residues) were synthesized each with a sticky end, as shown. The strand at l„ower left (20 residues) was end-labelled by enzymatic kinase reaction with T4 polynucleotide kinase enzyme using gamma ³²P labelled ATP.

Each of the three strands was dissolved in the same specimen of buffer (tris pH7 with 16 mM MgCl₂) to a concentration of 1 mM to form a composition containing the immobile junction shown in the upper part of Fig. 14 and exposed to DNA ligase, a joining enzyme, for 24 hours at room temperature, using a large excess of the enzyme and ATP, its cofactor. Subsequently, the reaction was stopped, the mix was extracted with phenol, bubbled with ether to remove the phenol, and dried by lyophilization to form a solid. When subjected to electrophoresis in buffer, a specimen of this composition exhibited several different bands, of which those appearing as 4 and 6 multiplets of the single 20 residue band were very prominent. When the composition was subjected to digestion with the exonuclease enzyme, exo III, which catalyzes hydrolysis of linear polynucleotides stepwise from their free ends, and the solution subsequently was electrophoresed on 10% polyacrylamide gel, only the two bands at the positions of the 4- and 6-multiplets remained, showing that the composition was in part in the form of the 6-multiρlet hexagonal or circular unit structure or geometric network shown in the lower portion of Fig. 14, with the 20 mer str-and running along the inside of the hexagon being covalently joined; the remainder of the composition was in the form of a 4-multiplet square (or circular) unit structure or geometrical network.

The NMR spectrum (low field region of 600 MH Η NMR spectrum of 2 mM solution) of another specimen of the lyophilized composition containing the three strands shown in Fig. 14 was determined, as shown in Fig. 15A. The extremely broad line widths are characteristic of linear duplex DNA of chain lengths in excess of 260 base pairs, indicating hydrogen bonding of the sticky ends to form larger complexes containing a number of the immobile junctions in a geometric network. In contrast, the NMR spectrum of a single strand DNA composition, shown in Fig. 15B, exhibits characteristically sharp lines.

Unlabelled specimens of the composition can be obtained by following the same procedure, using the autoradiogram as a guide to the location of the desired composition after electrophoresis. Compositions having other similar unit structures can be made containing other immobile or semi-mobile junctions of rank 3 to 8 using strands having either a single sticky end or two sticky ends. Industrial Applicability

The first utility is that these compositions containing mobile or semi-mobile nucleic acid junctions may be used as vertices of n-connected networks of nucleic acids. That is the fundamental aspect about this invention which gives it value. They can be used to make nucleic acid structures in the form of geometrical stick figures where the sticks correspond to double helical nucleic acids and the vertices are nucleic acid junctions. The utility of being able to make the geometrical figures is that this allows one to do molecular engineering of this sort on the hundred to thousand A scale. This, in turn, allows one to make various kinds of intricate figures which may have utility as appropriate surfaces upon which to condense cognate mclecules such as the protein depicted in Figure 11.

The application of translational symmetry to this case allows one to make these figures periodic in space as well as in connectivity i.e. to ligate together nucleic acids in the same form as is currently done with linear DNA in in vitro recombinant DNA work. In that case, DNA from different sources is ligated together in a specific fashion. Using well known sticky ended ligation techniques, one does the same thing and puts together periodic N-connected networks of nucleic acids which will if they're periodic in space, constitute crystals. These can be used for structural analysis of nucleic acids by nucleic acid-protein interactions and, in combination with more sophisticated architecture involving semi-mobile junctions, they can be used as well as to give informaion about nucleic-acid drug interactions. The more sophisticated application might also be necessary for nucleic acid-protein interactions.

Being a periodic network of precise architecture defined at the molecular level, it should also be possible to use this construction for doing other kinds of molecular architecture and engineering on the 100 to thousand angstrom scale. For example, these lattices may be useful as supports for micro-electronics structures. Other kinds of intricate microscopic structure are available, as well: What has been devised here is the equivalent of the polyvalent (3-8) joint in the Tinkertoy^TM while what previously existed in ligatable nucleic acid structures was only the stick of the Tinkertoy, and a divalent linear connecting piece. Thus one now has the polyvalent (3-8) connecting piece and anything that one wishes to make on that sacle, one can, in principle make, simply by exploiting the combination of the principles of nucleic acid junction formation the constraints of nucleic-acid structure, the dynamics of molecular architecture and imagination. Besides periodic structuring intricate individual structures can be constructed. These include closed cyclic polygons, such as the hexagon shown at the bottom of Figure 16, as well as open or closed polyhedra whose faces are such polygons.

Another aspect that should be considered is the fact that a periodic structural network of this form would be an appropriate substrate for the investigation of any material, be it nucleic acid. something which naturally interacts with nucleic acid, or otherwise. Once one has such an ordered array, one can use that, with a small amount of molecular engineering as the basis for the construction of an appropriate substrate on which any materials can be crystalized. For example, a reactive "hook" or "hooks" could be attached to one or more residues within the unit cell to covalently capture and identically orient the molecule of choice. Similarly, divalent antibodies covalently and/or non-covalently bound to nucleic acids or to cognate proteins could accomplish the same goal without using covalent reactions. In this vein, the junction lattic can be used as a template for crystallizing and structurally characterizing materials that otherwise may not be readily crystallizable. Thus we can look at protein folding intermediates or messenger RNA molecules that perhaps cannot crystallize at all. A great advantage of using junction lattices to look at such systems is that lattice forces would only affect the lattice molecules themselves, and not the molecules introduced into the lattice for structural study. Thus, protein-folding intermediates and long RNA molecules with readily perturbed tertiary structures would be visible structurally without perturbing them with lattice forces.

A further thing one should consider has to do with the limited mobility junction, sometimes referred to as the semi-mobile junction. These complexes have an intrinsic dynamic component to them and this intrinsic dynamic component is a natural constant relating to the material. The way in which any environmental or protenially mutagenic or teratogenic substance perturbs that fundamental dynamic constant is in fact an index of its possible effects within the living system. This also constitutes another way in which drugs can be screened: the way in which they affect the dynamics of this system, as well as in the visualization of the static way in which they interact with nucleic acids or nucleic acid junctions in the crystalline context discussed earlier.

The use of semi-mobile junctions as potential components of a computer memory is apparent because a junction which has two states, namely flipped or flopped, is obviously a prototype two state device since the equilibrium may be controlled by the supercoiling state of two arms, it is clearly possible these semi-mobile junctions can therefore be used to store information in bit-wise fashion.

What is claimed is:

Claims

1. A composition comprising a synthetic nucleic acid containing an immobile nucleic acid branched junction.

2. A composition as claimed in claim 1 further compr ising a branched junction from which at least three double-helices emanate.

3. A composition as claimed in claim 2 further comprising a branched junction from which three to eight double-helices emanate.

4. A composition as claimed in claim 3 further comprising periodic or non-periodic networks of branched junctions which at least form a substantially two-dimensional structure.

5. A composition as claimed in claim 4 wherein the networks are substantially three-dimensional structures.

6. A composition as claimed in. claim 3 further comprising a branched junction from which four double-helices emanate.

7. A composition as claimed in claim 6 comprising four hexadecanucleotides.

8. A composition as claimed in claim 6 comprising four dodecanucleotides.

9. A composition as claimed in claim 2 wherein at least a portion of the double-helices emanating from the junctions contain sticky ends.

10. A composition as claimed in claim 2 wherein at least a portion of the double-helices emanating from the junctions contains blunt ends.

11. A composition comprising synthetic nucleic acids containing a semi-mobile nucleic acid branched junction.

12. A composition as claimed in claim 11 further comprising a junction from which at least three double-helices emanate.

13. A composition as claimed in claim 12 further comprising a branched junction from which three to eight double-helices emanate.

14. A composition as claimed in claim 13 further comprising periodic or non-periodic networks of branched junctions which form substantially two dimensional structures.

15. A composition as claimed in claim 14 wherein the networks are substantially three dimensional.

16. A composition as claimed in claim 13 further comprising a branched junction from which four double-helices emanate.

17. A composition as claimed in claim 13 comprising four hexadecanucleotides.

18. A composition as claimed in claim 13 comprising four dodecanucleotides.

19. A composition as claimed in claim 12 wherein at least a portion of the double-helices emanating from the junctions contain sticky ends.

20. A composition as claimed in claim 12 wherein at least a portion of the double-helices emanating from the junction contain blunt ends.

21. A composition as claimed in claim 2 further comprising a geometric nucleic acid network containing any other chemical species bound in a specific or a non-specific manner to an immobile nucleic acid branched junction structure.

22. A composition as claimed in claim 12 further comprising a geometric nucleic acid network containing any other chemical species bound in a specific or non-specific manner to a semi-mobile nucleic acid branched structure.

23. A composition as claimed in claim 2 further comprising a geometric nucleic acid network containing any other chemical species bound in a specific or a non-specific manner to a single or double stranded nucleic acid structure emanating from an immobile nucleic acid branched junction structure.

24. A composition as claimed in claim 12 further comprising a geometric nucleic acid network containing any other chemical species bound in a specific or a non-specific manner to a single or double stranded nucleic acid structure emanating from a semi-mobile nucleic acid branched junction structure.

25. A composition as claimed in claim 2 further comprising a branched junction from which at least three double-helices emanate and two strands of at least one double helix are covalently connected to each other at the end distal to the junction.

26. A composition as claimed in claim 12 further comprising a branched junction from which at least three double-helices emanate and two strands of at least one double helix are covalently connected to each other at tl\e end distal to the junction.

27. The method of making a composition comprising a synthetic nucleic acid containing an immobile or semi-mobile branched junction, which method comprises selecting at least three double strands of nucleic acid which possess minimal sequence symmetry with each other, dissolving said strands in a buffer at pH 6 to 9 and at a temperature from 0° to 60°C, and separating the composition containing an immobile or semi-mobile branched junction from said solution.