Crystal structure of the DNA sequence d(CGTGAATTCACG)₂ with DAPI

The interaction of DAPI with d(CGTGAATTCACG) results in displacement of the ordered spine waters and confers hydrophobic behaviour on the DNA.

Abstract

The structure of 4′,6-diamidine-2-phenylindole (DAPI) bound to the synthetic B-DNA oligonucleotide d(CGTGAATTCACG) has been solved in space group P2₁2₁2₁ by single-crystal X-ray diffraction at a resolution of 2.2 Å. The structure is nearly isomorphous to that of the previously reported crystal structure of the oligonucleotide d(CGTGAATTCACG) alone. The adjustments in crystal packing between the native DNA molecule and the DNA–DAPI complex are described. DAPI lies in the narrow minor groove near the centre of the B-DNA fragment, positioned over the A–T base pairs. It is bound to the DNA by hydrogen-bonding and van der Waals interactions. Comparison of the two structures (with and without ligand) shows that DAPI inserts into the minor groove, displacing the ordered spine waters. Indeed, as DAPI is hydrophobic it confers this behaviour on the DNA and thus restricts the presence of water molecules.

1. Introduction  

The d(CGTGAATTCACG) DNA duplex is interesting because it features an EcoRI restriction site (Kapuscinski, 1995 ▸; Kim et al., 1990 ▸). DAPI (4′,6-diamidino-2-phenylindole) shows strong fluorescence when bound to DNA, which led to the rapid adoption of DAPI for fluorescent staining in fluorescence microscopy (Russell et al., 1975 ▸). In addition, DAPI is a DNA-specific binding molecule which forms a fluorescent complex by attaching to the minor groove of A/T-rich sequences of DNA. When bound to double-stranded DNA (dsDNA), DAPI has an absorption maximum at a wavelength of 358 nm (ultraviolet) and its emission maximum is at 461 nm (blue). Therefore, in fluorescence microscopy DAPI is excited with ultraviolet light and is detected using a blue/cyan filter. Apart from analytical fluorescence light microscopy, DAPI is also popular for the labelling of cell cultures to detect DNA from contaminating mycoplasma or viruses (Biancardi et al., 2013 ▸). We accomplished our aim of cocrystallizing DNA with DAPI, as described in previous work (Sbirkova & Shivachev, 2016 ▸). The mechanisms of DAPI–DNA complex formation, including minor-groove binding, hydrogen bonding and hydrophobic activity, are discussed.

2. Materials and methods  

2.1. Macromolecule production  

The dry oligonucleotide sequence 5′-CGTGAATTCACG-3′ was purchased from Eurofins MWG Genomics.

2.2. Crystallization  

The DNA sequence 5′-CGTGAATTCACG-3′ (single-stranded) was dissolved in buffer to 3 mM and was annealed for 1 min at 75°C before use in order to obtain double-stranded DNA (dsDNA). The buffer solution consisted of 60 mM sodium cacodylate pH 7.0, 17 mM MgCl₂, 2 mM spermine. DAPI was dissolved in the same buffer at 3 mM. Crystals were grown from hanging drops (3 µl) equilibrated against 50%(v/v) 2-methyl-2,4-pentanediol (MPD) at room temperature. Large crystals (0.4 × 0.3 × 0.25 mm) suitable for single-crystal X-ray studies formed within a month (Fig. 1 ▸). Crystallization information is summarized in Table 1 ▸.

Figure 1.

Observed crystals of 5′-CGTGAATTCACG-3′ complexed with DAPI.

Table 1. Crystallization.

Method	Hanging drop
Plate type	24-well
Temperature (K)	290
Final DNA concentration (mM)	1
Buffer composition of DNA solution	60 mM sodium cacodylate pH 7.0, 17 mM MgCl2, 2 mM spermine
Composition of reservoir solution	50%(v/v) MPD
Volume and ratio of drop	3 µl (1:1:1 DNA:DAPI:reservoir)
Volume of reservoir (µl)	550

2.3. Data collection and processing  

Crystals were mounted on loops and flash-cooled at 130 K directly under a Cobra nitrogen cryostream (Oxford Cryosystems). All data were collected at low temperature (130 K) on an Oxford Diffraction SuperNova diffractometer using Cu Kα radiation (λ = 1.54056 Å) from a microfocus source. Determination of the unit-cell parameters, data integration, scaling and absorption corrections were carried out using CrysAlis^Pro (Rigaku).

2.4. Structure solution and refinement  

The phases were obtained by molecular replacement with Phaser (McCoy et al., 2007 ▸) using PDB entry 1d29 (Larsen et al., 1991 ▸) as a starting model. Refinement of the structure involved several cycles of refinement using REFMAC5 (Murshudov et al., 2011 ▸) and Coot (Emsley et al., 2010 ▸). The water molecules and ligand (DAPI) were positioned from the F _o − F _c difference map using the Coot interface (Emsley et al., 2010 ▸). Visual analyses of the model and the electron-density maps were carried out using Coot (Emsley et al., 2010 ▸). A summary of the most important crystal data and refinement indicators is provided in Tables 2 ▸ and 3 ▸. X3DNA (Zheng et al., 2009 ▸) was used to carry out structural analysis and geometrical calculations of DNA parameters. USFC Chimera (Pettersen et al., 2004 ▸) was used to prepare the figures. The coordinates and structure factors have been deposited in the PDB as entry 5t4w.

Table 2. Data collection and processing.

Diffraction source	SuperNova Dual, microfocus Cu (Nova)
Wavelength (Å)	1.54
Temperature (K)	130
Detector	Atlas CCD
Crystal-to-detector distance (mm)	56
Rotation range per image (°)	0.5
Exposure time per image (s)	80
Space group	P212121
a, b, c (Å)	24.50, 41.09, 65.18
α, β, γ (°)	90, 90, 90
Mosaicity (°)	1.3, 1.01, 2.28
Resolution range (Å)	34.76–2.30
R merge	0.08 (0.21)
Total No. of reflections	43577
No. of unique reflections	3229
Completeness (%)	90.1 (77.9)
Multiplicity	8.3 (5.4)
〈I/σ(I)〉	40.1 (7.9)
Overall B factor from Wilson plot (Å2)	38.04

Table 3. Structure refinement.

Resolution range (Å)	34.76–2.30 (2.36–2.30)
Completeness (%)	90.1 (77.9)
σ Cutoff	None
No. of reflections, working set	2912 (161)
No. of reflections, test set	291 (15)
Final R cryst	0.216 (0.337)
Final R free	0.258 (0.454)
Cruickshank DPI	0.378
No. of non-H atoms
DNA	486
Ligand	21
Solvent	3
Total	510
R.m.s. deviations
Bonds (Å)	0.008
Angles (°)	1.559
Average B factors (Å2)
Nucleic acid	39.5
Ligand	41.1
Water	31.6

3. Results and discussion  

Single-crystal data collection was attempted from several different crystals. It should be noted that the data set collected to a resolution of 2.3 Å was from a crystal that was harvested from the drop 5 d after it was spotted. Crystals with similar or even larger dimensions that were allowed to ‘stabilize’ for more than a week in the crystallization drop usually diffracted to resolutions of up to 2.5 Å. Attempts to collect data at room temperature (19°C) were also performed on a few crystals; however, the observed quality of the diffraction was not comparable with that for experiments conducted at 130 K. The presence of DAPI in the crystallization conditions may have played a role in destabilizing the crystal structure. It should be noted that there was no difference in the crystallization conditions used to obtain the structures of d(CGTGAATTCACG)₂ without (PDB entry 5ju4; Sbirkova & Shivachev, 2016 ▸) and with (PDB entry 5t4w) bound DAPI. In both crystallization trials crystals formed after a month and if they were allowed to equilibrate the diffraction quality was worse. We cannot provide a plausible explanation for the presence and absence of DAPI in crystals obtained under the same crystallization conditions.

The asymmetric unit of PDB entry 5t4w consists of two chemically equivalent self-complementary strands (each of 12 bases in length) forming antiparallel right-handed DNA (Fig. 2 ▸). The B-type DNA duplex is formed by classical Watson–Crick (W-C) hydrogen-bonding base-pairing inter­actions between the two strands: bases C1 to G12 from the first strand interact with bases G13 to C24 from the opposite (second) strand (the numbering corresponds to the sequence in the crystal structure). The minor groove of the present double-stranded oligonucleotide features a central TpA step (AATT) surrounded by C/G-rich regions. Its overall secondary structure (Fig. 2 ▸) is comparable to previously reported low-temperature and room-temperature structures with the same sequence: PDB entries 5ju4, 1d28 and 1d29 (Sbirkova & Shivachev, 2016 ▸; Narayana et al., 1991 ▸; Larsen et al., 1991 ▸). The base-pair morphology values for shear, stretch, stagger, buckle, opening and propeller twist obtained using X3DNA (Zheng et al., 2009 ▸) are shown in Supplementary Table S1. It is obvious from the values that the interaction of DAPI changes the conformation of the DNA when they form a complex. The variation in the shear and stretch decreases, while that in the stagger, buckle, opening and propeller twist values increases. The most pronounced differences are in the core TpA region; for example, the values of buckle are 6.05 and 8.06 for A–T pairs 5–20 in 5t4w and 5ju4, respectively, while the values of propeller twist are −22.16 and −15.28 and those of opening are 9.28 and 1.8, respectively, for T–A pairs 7–18. The interaction of DAPI destabilizes the minor-groove conformation, while in contrast the major groove and the CpG regions at the end of the sequence show significant stabilization. Consequently, owing to this compensatory behaviour, although the DNA sequence is altered the hydrogen bonds are kept unchanged and the intrastrand interactions in PDB entries 5ju4 and 5t4w produce a motif that is in agreement with the Dickerson–Drew dodecamer and has classical right-handed B-DNA duplex structural features (Wing et al., 1984 ▸; Wei et al., 2013 ▸).

Figure 2.

(a) View of the asymmetric unit of PDB entry 5t4w including the DAPI molecule (red) and (b) view of the asymmetric unit of PDB entry 5ju4 (Mg²⁺ is in green).

Both the 5t4w and 5ju4 (Sbirkova & Shivachev, 2016 ▸) crystal structures confirm the nonclassical interstrand hydrogen-bonding interactions involving G bases, as previously reported by Hossain et al. (2002 ▸). The base pairs C1–G24, G2–C23 and G12–C13, C11–G14 located at the ends of the duplexes form G⋯G bonds to the adjacent DNA duplexes. The observed G⋯G hydrogen bonding does not correspond to that of Hoogsteen (1959 ▸). Representative electron densities and hydrogen-bonding interactions are shown in Fig. 3 ▸. Based on the donor–acceptor distances (D⋯A), the observed G⋯G hydrogen bonds are slightly weaker than classical W-C hydrogen bonds (Hossain et al., 2002 ▸; Desiraju & Steiner, 1999 ▸). (The D⋯A distance for G⋯G is around 3.0 Å, while that for C⋯G is around 2.85 Å.) The G⋯G interactions are located at the ends of the DNA strands, while the DNA⋯DAPI⋯DNA interactions (Fig. 4 ▸) involve the AATT region. DAPI requires more space and thus occupies a ‘water-rich’ region in the minor groove, conferring it with a hydrophobic behaviour, and compensating the negative charge of the DNA. On the other hand, the upper and lower surfaces of the purine and pyrimidine rings are also hydrophobic and the G⋯G interactions exploit the inter­action of the edges of the bases (which are hydrophilic), thus eliminating the need for water molecules. The Mg²⁺ ions compensate for the DNA negative charge in the 5ju4 structure (Sbirkova & Shivachev, 2016 ▸). Charge compensation is achieved through Mg²⁺⋯Da6A, Dt7A (Chiu & Dickerson, 2000 ▸; Sbirkova & Shivachev, 2016 ▸). The 2F _o − F _c difference map of PDB entry 5t4w suggests the absence of heavier atoms (than water), for example Mg²⁺ ions. Indeed, in the present structure the charge is compensated by the DAPI²⁺ molecule and thus no Mg²⁺ ion is required, and it cannot be detected in the structure. The DAPI molecule is additionally hydrogen-bonded to Dt19B, Da6A, Dt7A, Dt8A and Dc9A, strengthening the DNA–DAPI interactions (Fig. 4 ▸). The asymmetric unit of PDB entry 5t4w contains only three water molecules (while 74 are located in PDB entry 5ju4). The observed reduction in the ordered waters in the hydration shell may be owing to the hydrophobic property conferred on the DNA–DAPI complex by DAPI. A similar effect of DAPI is observed when it is present in complex with the Dickerson sequence in PDB entry 1d30 (Larsen et al., 1989 ▸). Only four DNA–DAPI crystal structures are available in the PDB and NDB: PDB entries 1d30 (Larsen et al., 1989 ▸), 432d (Vlieghe et al., 1999 ▸), 3ajl (Tsunoda et al., 2010 ▸) and 3gjh (Juan et al., 2010 ▸). –­C(NH₂)₂ ⁺ groups with identical hydrogen-bonding capabilities are present at both ends of the DAPI molecule. Thus, the orientation of the DAPI molecule in the DNA minor groove is governed by the indole NH proton (Granzhan et al., 2014 ▸). Comparison of the insertion of DAPI into the DNA shows that the DAPI molecule tends to favour the formation of a hydrogen bond between the indole NH and a T base (Larsen et al., 1989 ▸; Vlieghe et al., 1999 ▸; Tsunoda et al., 2010 ▸; Juan et al., 2010 ▸). The orientation of DAPI is similar to that observed in PDB entry 1d30 (Larsen et al., 1989 ▸), while the positioning in the minor groove is shifted in the direction of the indole moiety (or towards the water molecule near to DAPI in PDB entry 1d30). Even though the position of DAPI varies slightly, most of the hydrogen-bonding interactions are retained.

Figure 3.

Representation of canonical Watson–Crick interactions within the duplex and noncanonical G⋯G interactions between the adjacent DNA duplexes. (a) G⋯G interaction I between bases Dg2 and Dg24. (b) G⋯G interaction II between bases Dg12 and Dg14.

Figure 4.

Observed interactions of DAPI with the DNA bases; hydrogen bonds are shown as dashed lines and distances are shown in Å.

Supplementary Material

PDB reference: d(CGTGAATTCACG) with DAPI, 5t4w

Funding Statement

This work was funded by Bulgarian National Science Fund grants T02/14 and DRNF 02/1.