Complex intercalators (several shown in Figure 4 and in Table 2) utilize additional "complex" types of interactions when binding to DNA and to RNA model dinucleotide and oligonucleotide systems. As a result of these additional interactions, there may or may not be alterations in the sugar-phosphate geometries (with concomitant variations in the magnitude of angular unwinding), sequence binding affinities and neighbor-exclusion volumes when binding to these oligonucleotides and to high molecular weight DNA.

Figure 4. Chemical structures for complex intercalators.

Figure 4. Chemical structures of complex intercalators.

Table 3

Table 2. Unit cell constants, space groups, and structural data for complex intercalators binding to DNA- and RNA-like self-complementary dinucleotides and oligonucleotides, as determined in early studies.

Note that C3' endo (3'-5') C3' endo and C2' endo (3'-5') C2' endo homogeneous sugar-pucker patterns in these structures allow stretching, but no unwinding at the immediate intercalation site. See text for additional discussion.


Proflavine behaves as a complex intercalator when binding to self-complementary dinucleoside monophosphates. This reflects its ability to form hydrogen bonds between its amino groups and phosphate oxygen atoms on opposite chains when intercalating in these model studies (see Figures 5a and 6a).

Figure 5a.Figure 6a.
Figure 5b.Figure 6b.

Figure 5Figure 6

View taken at a skew angle and perpendicular to base pairs and drug molecules, showing differences between (a) proflavine-iodoCpG and (b) acridine orange-iodoCpG. Dashed lines indicate hydrogen bonds. Notice that, whereas proflavine binds symmetrically, acridine orange binds asymmetrically to iodoCpG. See text for discussion.

Proflavine stabilizes a unique intercalative geometry that has all sugar residues in the C3' endo conformation [i.e., C3' endo (3'-5') C3' endo]. Although the torsional angles defining the phosphodiester geometries have values similar to the beta-structural element, it is important to realize that this structure is clearly different and should not be confused with the beta-element. Base pairs are separated by 6.8 Angstroms; however, they remain twisted 36 degrees. The angular unwinding component present in the beta-structural element is therefore absent from this structure - this reflects the homogeneous sugar puckering that is observed. 15.3 Angstroms separate phosphate oxygen atoms on opposing chains, a distance that allows proflavine to span (by hydrogen-bonding) both phosphate oxygens. This gives rise to the symmetric interaction observed in the proflavine intercalative complexes.

Acridine orange, however, binds to the beta-structural element in these model studies (see Figures 5b and 6b). Being a methylated proflavine derivative, acridine orange is unable to hydrogen bond to phosphate oxygen atoms on opposite chains. Instead, acridine orange intercalates asymmetrically into the beta-structural element, forming tight stereo-specific stacking interactions with adjacent guanine rings. This asymmetry is a feature common to several different acridines in these solid-state studies.

It is possible for proflavine to intercalate into the beta-structural element; however, it would not be able to form hydrogen bonds simultaneously with phosphate oxygens on opposite chains (i.e., distance, 17.3 Angstroms). Such a complex is expected to be asymmetric, therefore, proflavine hydrogen bonding to one of two phosphate oxygens on neighboring chains.

To investigate this type of interaction, we have synthesized N, N-dimethylproflavine and have examined three different complexes between RNA- and DNA-like self-complementary dinucleoside monophosphates.

N, N Dimethylproflavine intercalates into the beta-structural element in all three cases. One structure, N, N dimethyl-proflavine: deoxycytidylyl (3'-5') deoxyguanosine, is shown in Figure 7a and 7b. As expected, N, N dimethylproflavine intercalates asymmetrically into the beta-structural element; however, unexpectedly, its unmethylated amino- group does not hydrogen bond to a neighboring phosphate oxygen atom (distance, 3.5 Angstroms). Instead, this amino group is hydrogen bonded to a water molecule. This reflects the absence of coplanarity of both phosphate oxygen atoms with the N, N-dimethylproflavine molecular least squares plane. In this structure and in the other two structures studied, phosphate oxygen atoms lie 0.7 Angstroms above (or below) the plane of the N, N-dimethylproflavine molecule.

Figure 7a.Figure 7b.
Figure 7. Structural information from the N, N-dimethylproflavine-d (CpG) crystalline complex. (a) Skew view to base pairs and drug molecule (b) perpendicular view to base pairs and drug molecule. This figure should be compared with the information shown in Figures 5 and 6.

Although proflavine acts as a complex intercalator in these model dinucleotide studies, it is a simple intercalator when binding to high molecular weight DNA. This is indicated by the ability of proflavine to unwind superhelical DNA with a magnitude similar to ethidium, acridine orange and other simple intercalators (Waring, 1970).


Another example of a complex intercalator is daunomycin. This drug molecule forms a complex with the DNA-like hexanucleotide d (CpGpTpApCpG), this being the earliest drug-oligonucleotide complex to be analyzed by X-ray crystallography (Quigley et al., 1980). See Figure 8.

The daunomycin aglycone anthracycline chromophore is oriented at right angles to the long dimension of the DNA base pairs, while the cyclohexene ring rests in the minor groove of the hexanucleotide duplex. Substituents on this ring have hydrogen-bonding interactions with the base pairs above and below the intercalation sites. The amino- sugar lies in the minor groove, forming electrostatic interactions with phosphate groups.

Intercalation occurs in the sequence d-CpG, and is associated with all sugar residues being in the C2' endo conformation (i.e., C2' endo (3'-5') C2' endo). Base pairs above and below the anthracycline ring are twisted by 36 degrees; again, as in the proflavine studies, there is no unwinding at the immediate intercalation site. There is, however, a small amount of unwinding in neighboring regions, estimated to be about 4 degrees on either side. This decreased magnitude of angular unwinding is in good agreement with that observed from solution studies of daunomycin binding to superhelical DNA, where it is estimated to be about 11 degrees (Waring, 1970). Daunomycin, therefore, is a complex intercalator when binding to DNA.

Figure 8.
Figure 8. Daunomycin is an example of a complex intercalator that intercalates into a C2' endo (3'-5') C2' endo site. The anthracycline ring system is intercalated perpendicular to the long axes of the base pairs above and below, while a water molecule bridges an amino- group on daunomycin with carbonyl oxygen on a neighboring cytosine residue. Although base pairs have separated 6.8 Angstroms to accommodate the intercalator, there is no unwinding in this region. Base pairs above and below the anthracycline ring remain twisted 36 degrees relative to one another. There is, however, additional unwinding in neighboring regions, estimated to be about 4 degrees on either side, a total of about 8 degrees.

A large number of additional studies have since been carried out with daunomycin and related molecules (i.e., adriamycin, nogalomycin, idarubicin and doxorubicin) complexed to self-complementary hexanucleotides with related sequences. In each study, the anthracycline aglycone chromophore is found to be oriented at right angles to the long dimension of the DNA base pairs, with little or no unwinding observed at the immediate intercalation site (characterized in each case by the homogeneous sugar puckering pattern: C2' endo (3'-5') C2' endo).

Nogalomycin intercalates in much the same manner as daunomycin, in spite of the presence of bulky substituents on either side of the anthracycline ring system. The presence of these bulky substituents on nogalomycin suggests that DNA breathing events are necessary for nogalomycin to enter (and to exit from) DNA. A possible mechanism to understand this is discussed below.


Actinomycin D can be considered to be a complex intercalator when binding to DNA. Actinomycin demonstrates a d (GpX) (where X = C, T, A or G) sequence-binding affinity with oligonucleotides and synthetic and naturally occurring DNA polymers.

This sequence-binding preference reflects the presence of the cyclic pentapeptide chains on actinomycin, which form hydrogen bonds with guanine residues (on opposite chains, in the case of d (GpC) sequences) in the three-dimensional DNA interaction. Reinhardt (1976) has investigated the binding of ethidium to DNA in the presence of actinomycin and, also, actinomine, an analog of actinomycin that lacks the pentapeptide chains. His studies clearly indicate that, whereas actinomycin binds noncompetitively with ethidium for sites in DNA, actinomine competes with ethidium for these sites. These data indicate that the cyclic pentapeptide chains on actinomycin play a key role in determining the sequence-binding guanine specificity in the DNA-binding reaction.

Actinomycin behaves as a complex intercalator by virtue of its cyclic pentapeptide chains, whose interactions with DNA give rise to altered sequence-binding affinities. The enhanced neighbor-exclusion volume -- estimated to be about four base pairs -- also reflects the presence of the pentapeptide chains. Nevertheless, actinomycin unwinds DNA roughly the same as ethidium (Waring, 1970), and this suggests the phenoxazone ring to intercalate into a high-energy beta-structural element, surrounded perhaps, by lower-energy beta-structural elements on either side.

X-ray crystallographic studies of actinomycin complexed to several different self-complementary oligonucleotide sequences (including four different crystalline modifications of the same complex) have now confirmed the overall features of the actinomycin-DNA binding model. As already mentioned, the precise nature of the DNA conformation remains unknown due to problems inherent in refining large structures of this kind with limited resolution data. Further refinement of these structures (perhaps using low temperature data) could help to clarify this point.


Echinomycin and triostin A are related antibiotics that contain two quinoxaline ring systems connected together by cyclic octapeptide chains. These molecules are bifunctional intercalators that utilize bis-intercalation when they bind to DNA. They resemble actinomycin in their mechanism of action, being potent inhibitors of DNA transcription.

Crystallographic studies of these molecules complexed to different oligonucleotides have revealed the presence of a DNA conformation very similar to the high-energy beta-DNA form. Although more completely unwound, this conformation demonstrates "neighbor-exclusion" binding by the quinoxaline ring systems, which "wrap around" base-paired d (pGpC) sequences and intercalate between neighboring base pairs on either side. The cyclic octapeptide chains lie in the narrow groove of the high-energy beta-DNA structure, forming numerous van der Waals contacts with the base pairs. A particularly interesting feature of these complexes is the presence of Hoogsteen base pairing in neighboring A-T and G-C base pairs on either side of the intercalation site. This suggests that DNA breathing events must precede (or accompany) complex formation. We discuss a possible mechanism below.

Ditercalinium and flexi-di are synthetic bifunctional intercalators that contain two chromophores connected together by a flexible bridge-like chain. They form DNA complexes with the self-complementary tetranucleotide sequence d (CpGpCpG), which again appears to be in the high-energy beta-DNA form. However, as with the echinomycin and triostin A complexes, we emphasize that the detailed nature of the sugar puckering has yet to be established.

Finally, bis-daunomycin is a synthetic bifunctional intercalator that forms a complex with the hexanucleotide d (CpGpApTpCpG). The aglycone anthracycline ring systems intercalate between d-CpG sequences, again lying perpendicular to the long dimension of the base pairs. Anthracycline rings are connected together by a "bridge-like" chain, "wrapping around" four base pairs that lie internally in this complex. There appears to be little or no DNA unwinding at the intercalation sites.

D-232 is a synthetic bifunctional intercalator that forms a complex with the hexanucleotide d (CpGpTpApCpG). Although this complex has features similar to the previous complex, significant unwinding has been observed at the intercalation sites. The detailed nature of the sugar puckering in both structures has not yet been clearly established.

How do complex intercalators bind to DNA?

  1. Although simple intercalators "pin" the premelton, complex intercalators may or may not "pin" the premelton.

  2. Actinomycin "pins" the premelton, binding tightly to the beta-DNA core region. The phenoxazone ring system intercalates into the central high-energy beta-structural element, while cyclic pentapeptide chains lie in the narrow groove forming hydrogen bonds with guanine residues on opposite chains. Numerous van der Waals contacts occur between pentapeptide chains and nucleotides positioned either proximally or within neighboring low-energy beta-structural elements on either side. Water is "squeezed out" during complex formation. This gives rise to the large positive entropy change observed in the binding reaction.

  3. Daunomycin is unable to "pin" the premelton. For structural and energetic reasons (see note below), the initial complex is unstable and eventually disappears, leading to a structural rearrangement in the complex in which daunomycin remains attached to a lower energy B-like conformation. The process is reversible. A premelton spontaneously reappears. This reverses the structural arrangement of the complex, which, after a breathing motion of sufficient magnitude, allows daunomycin (or nogalomycin) to detach from DNA.

    Note: Intercalation by daunomycin and related antibiotics is atypical in that the aglycone anthracycline chromophore lies at right angles to the long dimension of the DNA base pairs. This orientation does not provide adequate stacking energies to "pin" the premelton.

  4. Synthetic bifunctional intercalators, such as ditercalinium and flexi-di, "pin" larger premeltons containing adjacent high-energy beta-structural elements in their centers.

  5. Echinomycin and triostin A "pin" still larger premeltons. Breather motions transiently focus energy into the centers of premeltons to do the work required for DNA breathing. Once polynucleotide chains come apart, deoxyadenosine and deoxyguanosine residues can isomerize to their syn conformations, allowing Hoogsteen base pairs to form. The binding by either antibiotic facilitates such a rearrangement. The process is reversible. A breathing event of sufficient magnitude causes echinomycin or triostin A to detach from DNA, facilitating a rearrangement in DNA that leads to Watson-Crick base pairing as polynucleotide chains come together again.
Echinomycin and triostin A have the same rigid octapeptide chain that holds their quinoxaline ring systems 10.2 Angstroms apart and at the proper orientation to intercalate into DNA. By doing so, this octapeptide chain provides a scaffold that holds DNA in its highly unwound beta- form. Both antibiotics "pin" a similar DNA structure in the transcription complex to inhibit chain elongation. Their mechanism of action is similar to that proposed for actinomycin D.

Some additional references:

Bond, P.J., Langridge, R., Jennette, K.W. and S. J. Lippard (1975). Proc. Nat. Acad. Sci. USA 72, 4825-4829.

Jain, S.C., Tsai, C. -C. and H.M. Sobell (1977). J. Mol. Biol. 114, 317-331.

Hoogsteen, K. (1959). Acta Cryst. 12, 822.

Jain, S.C., Bhandary, K.K. and H.M. Sobell (1979). J. Mol. Biol. 135, 813-840.

Jain, S.C. and H.M. Sobell (1984a). J. Biomolec. Struct. and Dynam. 1, 1161-1177.

Jain, S.C. and H.M. Sobell (1984b). J. Biomolec. Struct. and Dynam. 1, 1179-1194.

Krugh, T.R. and C. Reinhardt (1975). J. Mol. Biol. 97, 133-162.

Neidle, S., Achari, A., Taylor, G.L., Berman, H.M. Carrell, H.L., Glusker, J.P. and W.C. Stallings (1977). Nature 269, 304-307.

Patel, D.J., and Shen, C. (1978). Proc. Nat. Acad. Sci., 75, No. 6, 2553-2557.

Quigley, G.J., Wang, A.H., Ughetto, G., van der Marel, G., van Boom, M.J.H. and A. Rich (1980). Proc. Nat. Acad. Sci. USA 77, 7204-7208.

Reddy, B.S., Seshadri, T.P., Sakore, T.D. and H.M. Sobell (1979). J. Mol. Biol. 135, 787-812.

Reinhardt, C. (1976). Ph.D. Thesis, "Spectroscopic evidence for sequence preferences in the intercalative binding of ethidium bromide to nucleic acids", University of Rochester, Department of Chemistry.

Smith, P.J.C. and S. Arnott (1978). LALS: A linked-atom least-squares reciprocal space refinement system incorporating stereochemical restraints and constraints to supplement sparse diffraction data. Acta Crystallogr. A34, 3-10.

Sobell, H.M., Reddy, B.S., Bhandary, K. -K., Jain, S.J., Sakore, T.D. and Seshadri, T.P. 1978. Cold Spring Harbor Symposia on Quantitative Biology, 42, 87-102.

Sobell, H.M., Lozansky, E.D. and Lessen, M. 1979. Cold Spring Harbor Symposia on Quantitative Biology, 43, 11-19.

Sobell, H.M., Sakore, T.D., Jain, S.C., Banerjee, K.K., Bhandary, K.K., Reddy, B.S. and Lozansky, E.D. 1983. Spring Harbor Symposia on Quantitative Biology, 47, 293-314.

Tsai, C. -C., Jain, S.C. and H.M. Sobell (1977). J. Mol. Biol. 114, 301-315.

Wang, J.C. (1974). J. Mol. Biol. 89, 783-801.

Wang, A.H., Nathans, J., van der Marel, G., van Boom, J.H. and A. Rich (1978). Nature 276, 471-474.

Wang, A.H., Quigley, G.J. and A. Rich (1979). Nucleic Acids Res. 6, 3879-3890.

Waring, M.J. (1970). J. Mol. Biol. 54, 247-279.

Additional references in the crystallographic literature can be obtained from the Protein Data Bank (PDB), using the hyperlinks, SIMPLE INTERCALATORS and COMPLEX INTERCALATORS.