Structural basis of B-to-Z DNA transition mediated by an anti-Z-DNA antibody
Cheng-Chung Lee, Shu-Fang Hsu, Ya-Wen Chang, Ya-Wen Chen, Meng-Ru Ho, Hiroshi Sugiyama, Wei-Chu Wang, Andrew H -J Wang
TL;DR
This study reveals how a specific antibody induces and stabilizes Z-DNA, a rare DNA structure, offering insights into its role in biology and disease.
Contribution
The first atomic-level structural analysis of antibody-mediated B-to-Z DNA transition and its stabilization mechanism.
Findings
cZ22-Fab induces a concentration-dependent B-to-Z DNA transition in CG-repeat DNA.
Crystal structures show cZ22-Fab binds Z-DNA via phosphate-clamping and base interactions.
Residues R50 and Y106 in the antibody are critical for Z-DNA recognition and stabilization.
Abstract
Z-DNA is a left-handed double-helical form of DNA that plays roles in transcription, immune responses, viral infection, bacterial biofilm formation, and autoimmune diseases. Despite its importance, the instability of Z-DNA under physiological conditions has hindered detailed structural and functional investigations. Moreover, although antibodies are known to recognize nucleic acids, the mechanisms underlying their detection and stabilization of dynamic DNA under biological conditions remain unclear. This study provides the first atomic-level structural insights into antibody-mediated B-to-Z DNA transition. Accordingly, a Z-DNA-specific chimeric Fab fragment of Z22 (cZ22-Fab) was designed and characterized using multiple biophysical approaches. cZ22-Fab mediates a concentration-dependent B-to-Z conformational transition in CG-repeat DNA, establishing a stable 2:1 Fab/DNA stoichiometry.…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6| d(CG)5 | cZ22-Fab | cZ22-Fab/d(CG)5 | |
|---|---|---|---|
| Sedimentation coefficient (S) | 1.63 | 3.62 | 6.69 |
| Frictional ratio ( | 1.28 | 1.21 | 1.31 |
| Calculated MW (kDa) | 5.92 | 43.8 | 116.8 |
| Theoretical MW (kDa) | 6.06 | 49.23 | 104.52 |
| Fab/dsDNA | – | – | 2/1 |
| cZ22-Fab | cZ22-Fab/dC(GC)3 | cZ22-Fab/d(CG)6 | |
|---|---|---|---|
| Space group |
|
|
|
| Cell dimensions | 74.0, 87.0, 149.4 90.0, 95.5, 90.0 | 195.6, 195.6, 89.4 90.0, 90.0, 120.0 | 114.9, 148.7, 172.2 90.0, 90.0, 90.0 |
| Resolution (Å) | 18.0–2.15 (2.28–2.15) | 18.0–2.35 (2.43–2.35) | 29.3–3.27 (3.39–3.27) |
| Unique reflections | 101,636 (6,578) | 156,195 (13,733) | 23,106 (2,268) |
| Rmerge (%) | 8.9 (72.1) | 7.1 (66.4) | 7.7 (34.6) |
|
| 12.58 (3.44) | 22.7 (2.4) | 8.5 (2.1) |
| Completeness | 97.3 (95.6) | 99.9 (99.8) | 99.8 (98.1) |
| Redundancy | 3.43 (3.48) | 5.30 (5.20) | 6.79 (6.98) |
| CC1/2 CC* | 0.99 (0.78) 0.99 (0.93) | 0.99 (0.81) 0.99 (0.95) | 0.99 (0.77) 0.99 (0.93) |
| Asymmetric unit | 4 Fab | 6 Fab, 3 dsDNA | 2 Fab, 1 ssDNA |
| Refinement | |||
| Resolution (Å) | 18.0–2.15 | 18.0–2.35 | 29.3–3.27 |
| No. of reflections Rwork/Rfree | 100,652/5,037 | 155,585/7,791 | 21,986/1,159 |
| Rwork/Rfree (%) | 19.2/23.8 | 16.9/20.9 | 20.0/24.2 |
| No. of atoms/avg | |||
| Protein | 11,480/44.5 | 19,661//39.1 | 6 ,684/87.7 |
| DNA | 849/38.3 | 243/56.5 | |
| Water | 1,081/38.1 | 1,734/40.9 | 69/49.5 |
| RMSD | |||
| Bond lengths (Å) | 0.0042 | 0.0111 | 0.006 |
| Bond angles (°) | 0.77 | 1.22 | 0.83 |
| Ramachandran statistics (%) | |||
| Favoured | 97.8 | 96.98 | 94.46 |
| Allowed | 1.96 | 2.78 | 5.31 |
| Outliers | 0.24 | 0.24 | 0.23 |
| Clash score | 4.29 | 4.53 | 6.72 |
| MolProbity score | 1.48 | 1.46 | 1.75 |
| dC(GC)3 | cZ22-Fab | Water mediation | Distance (Å) |
|---|---|---|---|
| Polar interaction | |||
| C3 OP1 | L-Lys93 NZ | – | 3.0 |
| C3 OP2 | L-Phe94 N | Wat | – |
| G4 OP1 | H-Tyr61 OH | – | 3.3 |
| G4 N7 | L-Lys93 NZ | – | 3.0 |
| C5 OP1 | H-Arg50 NH1 | – | 2.9 |
| H-Arg52 NE | – | 3.2 | |
| C5 N4 | H-Tyr106 OH | Wat | – |
| G6 OP1 | H-Arg52 NH1 | – | 2.8 |
| G6 O4′ | H-Arg52 NH2 | – | 3.3 |
| G6 O5′ | H-Arg52 NH2 | – | 3.1 |
| G6 N7 | H-Ser104 O | – | 3.3 |
| H-Tyr106 N | – | 3.0 | |
| C7 OP1 | H-Tyr103 OH | – | 3.1 |
| C7 N4 | H-Ser104 O | – | 3.0 |
| G2 N7* | L-Tyr32 OH | – | 3.4 |
| G2 O6* | L-Tyr91 OH | Wat | – |
| C3 N4* | L-Tyr32 OH | – | 3.2 |
| Nonpolar interaction (distance corresponds to the closest atom pair) | |||
| G4 C2′ | H-Tyr106 CE1 | – | 3.5 |
| G4 C4′ | L-Phe94 CE2 | – | 3.6 |
| G4 C5′ | H-Tyr61 CE2 | – | 4.0 |
| C5 C5 | H-Tyr106 CE2 | – | 3.9 |
| G6 C8 | H-Ser104 C | – | 3.5 |
| H-Asn105 CA | – | 3.9 | |
| H-Tyr106 CB | – | 4.0 | |
| G2 C8* | L-Tyr32 CZ | – | 3.9 |
| L-Tyr50 CE1 | – | 3.8 | |
- —National Science and Technology Council10.13039/100020595
- —NSTC10.13039/100020595
- —Taipei Medical University10.13039/501100004700
- —National Science and Technology Council10.13039/100020595
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsDNA and Nucleic Acid Chemistry · Monoclonal and Polyclonal Antibodies Research · Advanced biosensing and bioanalysis techniques
Introduction
DNA adopts multiple helical conformations, including A-, B-, and Z-forms. Among these, Z-DNA is the only left-handed double helix and is characterized by a zigzag phosphate backbone, which is distinct from the canonical right-handed B-DNA. The three-dimensional structure of Z-DNA was first resolved in 1979 by Wang et al. through X-ray crystallography of a synthetic CGCGCG hexamer, revealing an unusual left-handed conformation [1]. Although Z-DNA preferentially forms in sequences with alternating pyrimidine–purine repeats and is stabilized under high salt concentrations or through chemical modifications, these nonphysiological conditions initially raised doubts about its biological relevance in living cells. However, subsequent studies demonstrated that Z-DNA can form transiently under physiological conditions, indicating that Z-DNA has biological importance rather than being merely an experimental artifact [2, 3].
In recent years, Z-DNA has been recognized as a dynamic structural element involved in transcriptional regulation and innate immune signalling pathways. During transcription, the negative supercoiling generated behind RNA polymerase promotes the transient formation of Z-DNA near promoter regions, modulating transcription through reshaping the local chromatin environment [4]. In the immune system, several Z-DNA binding proteins specifically recognize Z-form nucleic acids. One example is adenosine deaminase acting on RNA 1 (ADAR1), an RNA-editing enzyme that recognizes Z-form nucleic acids through its Zα domain, which catalyses adenosine-to-inosine conversion within double-stranded RNA and prevents the aberrant activation of cytosolic innate immune sensors [5]. ADAR1 has also been reported to induce the DNA B-to-Z transition, and both experimental and computational studies suggest that this transition occurs through a stepwise propagation mechanism [6, 7]. Similarly, Z-DNA binding protein 1 (ZBP1) senses Z-form nucleic acids, which are generated under cellular stress or during viral infection, and triggers necroptosis and type I interferon immune responses [8, 9]. On the other hand, viral proteins such as vaccinia virus RNA binding protein E3 (E3L) use Zα domains to bind Z-form nucleic acids, blocking host sensors such as ZBP1 and enabling immune evasion [10]. In addition, influenza virus, filoviruses, and poxviruses have been reported to generate Z-form nucleic acids during replication [9, 11, 12]. Together, these studies emphasize the functional importance of the Z-conformation in gene regulation, host defences, and viral countermeasures. However, the intrinsic conformational dynamics and instability of Z-DNA have posed significant challenges to detailed mechanistic investigation. Addressing this challenge requires integrated structural and dynamic studies to clarify how protein recognition contributes to Z-DNA stabilization in vivo.
Under physiological conditions, DNA is generally tolerated by the immune system; however, abnormal exposure, impaired clearance, or structural modifications can trigger immune recognition and autoantibody production. Unlike canonical B-DNA, which is poorly immunogenic, Z-DNA is highly immunogenic and readily elicits conformation-specific antibodies. Practical applications have been demonstrated with chemically modified poly(dG–dC)·poly(dG–dC), which adopts a left-handed Z conformation under physiological conditions and has long been used to generate anti-Z-DNA antibodies in experimental animals [13–15]. The clinical relevance of Z-DNA is supported by the detection of anti-Z-DNA antibodies in several autoimmune and inflammatory diseases, including systemic lupus erythematosus (SLE) [16, 17], drug-induced lupus [18], rheumatoid arthritis (RA) [19], mixed connective tissue disease [20], and Crohn’s disease [21]. Z-DNA involvement has also been linked to infectious contexts, and low levels of anti-Z-DNA antibodies have even been observed in healthy individuals, where bacterial Z-DNA within biofilms is considered a possible source [22–24]. Beyond pathogenesis, anti-Z-DNA antibodies have been utilized as molecular probes, enabling the monitoring of disease activity in autoimmune patients and facilitating the detection of viral Z-RNA structures in infected cells [12, 25]. Therefore, antibody-based approaches hold promise for probing Z-DNA/Z-RNA dynamics in vivo. Despite their clinical and diagnostic importance, atomic-level structural information on how these antibodies specifically recognize Z-DNA remains unavailable, which limits the rational design of precision diagnostics and antibody-based therapeutics for Z-DNA-associated diseases.
In this study, we determined the crystal structure of a chimeric Fab fragment of Z22 (cZ22-Fab) in complex with Z-form DNA and conducted a detailed biophysical characterization of its binding properties. This structure provides the first atomic-resolution evidence for antibody recognition of left-handed Z-DNA. cZ22-Fab was observed to mediate the B-to-Z transition, and structural analysis revealed the molecular principles that enable antibodies to mediate this conformational change. These findings advance our understanding of antibody Z-DNA recognition and provide a structural framework for the development of antibody-based diagnostic and therapeutic applications.
Materials and methods
Materials
The cytosine–guanine repeat DNA oligonucleotide d(CG)n, an alternating pyrimidine–purine sequence known to preferentially adopt the Z-DNA conformation [26], was mainly used in this study. All DNA sequences, including unmodified oligonucleotides, 6-carboxyfluorescein (FAM)-labelled, and biotin-labelled hairpin forms, and B-form DNA controls, were purchased from Integrated DNA Technologies (Coralville, IA, USA) and are listed in Supplementary Table S1. All other reagents and chemicals were obtained from Sigma–Aldrich (St. Louis, MO, USA) unless otherwise stated.
Double-stranded DNA preparation
All DNA oligonucleotides were dissolved in Tris buffer (50 mM Tris and 150 mM NaCl, pH 8.0), and the concentrations were determined using a spectrophotometer (NanoDrop 2000, Thermo Fisher Scientific, Waltham, MA, USA). Equal molar amounts of sense and antisense oligonucleotides or a twofold molar amount of self-complementary oligonucleotides were mixed and subsequently denatured at 95°C for 10 min, followed by annealing through gradual cooling to room temperature to generate double-stranded DNA (dsDNA). The DNA duplexes were divided into aliquots and stored at −20°C. Unmodified DNA duplexes were used to examine the binding stoichiometry of cZ22-Fab and to perform crystallization with cZ22-Fab. FAM-labelled DNA duplexes were used in the gel mobility shift assay, and biotin-labelled hairpin DNA duplexes were used in binding assay and kinetic studies.
Construction of a chimeric anti-Z-DNA antibody cZ22-Fab and inactive mutants
Z22 is an anti-Z-DNA IgG2b antibody originally generated from Z-DNA immunized C57BL/6 mice [27]. A chimeric monoclonal antibody, cZ22-Fab, was constructed by grafting the Fv domain of Z22 onto the human IgG1 constant domain, as previously described [28]. The sequences encoding the Z22 light chain variable region fused with the human Cκ domain and the Z22 heavy chain variable region fused with the human CH1 domain carrying a C-terminal His-tag (Supplementary Fig. S1). The cZ22-Fab and the following designed inactive Fab mutants cZ22-FabR50A.H and cZ22-FabY106A.H, corresponding to alanine substitutions at Arg50 and Tyr106 of the heavy chain, were synthesized (BIOTOOLS, New Taipei City, Taiwan) and inserted into the mammalian expression vector pIgG under separate promoters [28].
Expression and purification of cZ22-Fab and the inactive mutants
cZ22-Fab and the inactive mutants cZ22-FabR50A.H and cZ22-FabY106A.H were expressed in Expi293F cells (A14527; Gibco, Grand Island, NY, USA) via transient transfection using an ExpiFectamine 293 transfection kit (A14525; Gibco, Grand Island, NY, USA). After six days, the culture medium containing secreted cZ22-Fab was collected, diluted 1:1 with high-salt Tris buffer (50 mM Tris and 500 mM NaCl, pH 8.0), and purified using a HisTrap Excel column (17-3712-06; Cytiva, Marlborough, MA, USA), from which cZ22-Fab was eluted with imidazole elution buffer (50 mM Tris, 150 mM imidazole, and 500 mM NaCl, pH 8.0). The purified Fab was concentrated using an Amicon Ultra centrifugal filter (UFC9010; Millipore, Burlington, MA, USA) and subjected to buffer exchange with Tris buffer for future use. The final purity of Fab was confirmed by sodium dodecyl sulfate‒polyacrylamide gel electrophoresis (SDS‒PAGE).
Electrophoretic mobility shift assay
In the gel mobility shift assay, FAM-labelled DNA duplexes (10 μM) [d(CG)n, *n *= 3–5] were mixed with cZ22-Fab at protein:DNA molar ratios ranging from 1:1 to 5:1 in a final volume of 10 μl in Tris buffer. After incubation at 20°C for 2 h, the samples were loaded onto a 2.5% agarose gel and electrophoresed at 100 V for 75 min at 4°C. DNA mobility shifts were visualized under UV exposure (Gel Doc EZ System, Bio-Rad, Hercules, CA, USA), and the same gel was subsequently stained with Coomassie blue to monitor Fab protein migration.
Circular dichroism measurements
The B-to-Z transition of d(CG)5 duplex DNA was monitored by circular dichroism (CD) spectroscopy at 20°C. Samples containing 25 μM d(CG)5 duplex were incubated with cZ22-Fab at 0, 25, 50, or 100 μM in Tris buffer at room temperature for 2 h before measurement. CD spectra were recorded on a J-815 spectrometer (Jasco, Tokyo, Japan) over a wavelength range of 230–320 nm at 0.5-nm intervals with a 2-s averaging time, and protein-only signals were subtracted from all spectra. Transition kinetics were monitored at 255 nm over 120 min. Relative transition activity was determined from the CD intensities at 255 nm, with the signal in the presence of 100 μM cZ22-Fab defined as 100%. The protein/nucleic acid (P/N) ratios are expressed as the molar ratio of Fab monomers to d(CG)5 duplex molecules.
Biolayer interferometry
Biolayer interferometry (BLI) measurements were performed using a Gator Label-Free Bioanalysis instrument (Gator Bio, Palo Alto, CA, USA) with software version 2.16.6.0130. Two SA XT probes (streptavidin-coated biosensors; one sample and one reference) were used for each measurement. Probes were prewet in Tris buffer for 300 s before loading. Sample probes were loaded with 100 nM hairpin DNA d(CG)5, while a 20-nucleotide B-form hairpin DNA (hairpin DNA 20N) served as the negative control. A threshold-based loading protocol was applied to reach a response of 0.5 nm, followed by a 300-s wash in Tris buffer. After establishing a 120-s baseline in Tris buffer, the probes were incubated in sample wells containing cZ22-Fab at concentrations ranging from 12.5 to 100 μM (twofold serial dilutions) for 7,200 s to monitor the association phase and subsequently transferred to Tris buffer for 1,800 s to monitor dissociation. Reference probes underwent the same procedure but without DNA loading. Data were processed using double reference subtraction and analysed by local fitting using a 1:1 binding model. SA XT Probes (160029), 96-well Max plates (130018), and 96-well black plates (130150) were purchased from Gator Bio (Palo Alto, CA, USA).
Sedimentation velocity analytical ultracentrifugation
Sedimentation velocity analytical ultracentrifugation (SV-AUC) was used to analyse the cZ22-Fab/d(CG)5 complex purified by size-exclusion chromatography (SEC), as well as to analyse the complex formation between cZ22-Fab and CG-repeat duplex DNA of varying lengths [d(CG)n, n = 4–7]. cZ22-Fab (6.9 μM) was mixed with duplex DNA at a Fab:DNA molar ratio of 3:1 and incubated at 20°C for 2 h. SV-AUC was performed on an XL-A analytical ultracentrifuge (Beckman Coulter, Brea, CA, USA) equipped with absorbance optics. Samples in Tris buffer were loaded into standard 12-mm, two-channel, sector-shaped centrepieces and centrifuged in an An-60 Ti rotor at 60,000 rpm at 20°C. The absorbance at 280 nm was recorded every 3 min. Data were analysed with the c(S) model in SEDFIT [29], using a buffer density of 1.0058 g/ml calculated with SEDNTERP [30].
Size-exclusion chromatography coupled with multiangle light scattering
All the Fab/DNA complexes were formed by incubating 10 μM d(CG)5 duplex with 10 to 60 μM cZ22-Fab in Tris buffer for 2 h before injection. The d(CG)5, cZ22-Fab, and preincubated cZ22-Fab/d(CG)5 complex mixtures were loaded on an SEC650 column (Bio-Rad, Hercules, CA, USA) equilibrated with Tris buffer and operated at a flow rate of 0.5 ml/min. Size-exclusion chromatography coupled with multiangle light scattering (SEC–MALS) was performed on a 1260 infinity quaternary LC system (Agilent Technologies, Santa Clara, CA, USA) coupled to four in-line detectors: laser light scattering (MiniDAWN TREOS, Wyatt Technology, Santa Barbara, CA, USA), quasielastic light scattering (QELS, Wyatt Technology, Santa Barbara, CA, USA), refractive index (Optilab T-rEX, Wyatt Technology, Santa Barbara, CA, USA), and ultraviolet detection (Agilent Technologies, Santa Clara, CA, USA). The collected data were subsequently analysed using ASTRA software version 6.1 (Wyatt Technology, Santa Barbara, CA, USA).
Size-exclusion chromatography of cZ22-Fab with DNA duplexes
cZ22-Fab and the inactive mutants cZ22-FabR50A.H and cZ22-FabY106A.H were incubated at 100 μM with annealed d(CG)5, d(TG)5, d(CG)_5_7N, or hairpin DNA d4N(CG)_5_4N at 25 μM in Tris buffer for 2 h before SEC analysis. SEC was performed on a Superdex Increase 5/150 column at a flow rate of 0.2 ml/min, with UV absorbance monitored at 280 nm. Elution profiles of Fab/DNA complexes, Fab alone, and each DNA duplex alone were recorded as retention volumes.
Crystallization, data collection, structure determination, and refinement
For Fab and Fab/DNA complex crystallization, all crystals were grown by mixing the protein or protein/DNA mixture with an equal volume of reservoir solution in a sitting drop and equilibrating by vapour diffusion at 18°C. The purified cZ22-Fab was concentrated to 12 mg/ml in Tris buffer for crystallization. Crystals belonging to the P2_1_ space group, with four Fab molecules in the asymmetric unit (ASU), were obtained in a reservoir solution containing 0.16 M ammonium sulfate, 20% (w/v) polyethylene glycol (PEG) 4,000, 20% (v/v) glycerol, and 0.08 M sodium acetate trihydrate (pH 4.6) (HR2-914-20; Hampton Research, Aliso Viejo, CA, USA). For Fab/DNA complex crystallization, cZ22-Fab was concentrated to 16 mg/ml in Tris buffer, mixed with dC(GC)3 or d(CG)6 duplex at a molar ratio of 3:1, and then incubated at room temperature for 2 h. cZ22-Fab/dC(GC)3 complex crystals were obtained in a reservoir solution of 12% (w/v) PEG 3,350, and 0.1 M sodium malonate (pH 5.0) (HR2-098-03; Hampton Research, Aliso Viejo, CA, USA) supplemented with 0.05% 1,2-diaminocyclohexane sulfate, 1,4-cyclohexanedicarboxylic acid, methylenediphosphonic acid, and sulfanilic acid; and 0.004 M HEPES sodium (pH 6.8) (HR2-996-23; Hampton Research, Aliso Viejo, CA, USA). The cZ22-Fab/dC(GC)3 complex was crystallized in the P3_1_ space group, with six Fabs and three dsDNA molecules in the ASU. cZ22-Fab/d(CG)6 complex crystals in the I2_1_2_1_2_1_ space group were obtained in a reservoir solution of 30% (w/v) PEG 400, 0.2 M sodium citrate tribasic, and 0.1 M Tris–HCl (pH 8.5) (HR2-133-13; Hampton Research, Aliso Viejo, CA, USA) with 25 mM 18-crown-6 (28125; Sigma–Aldrich, St. Louis, MO, USA). The ASU contained two Fab molecules and one single-stranded DNA (ssDNA) molecule.
All crystallography data were collected at cryogenic temperatures using an X-ray with a wavelength of 1.0 Å at the National Synchrotron Radiation Research Center (NSRRC, Hsinchu, Taiwan). Data of the cZ22-Fab crystal were collected at TPS 07A1 with a Rayonix MX300HE CCD detector (Rayonix, Evanston, IL, USA). Data of both complex crystals were collected at TPS 05A1 with a Rayonix MX300HE CCD detector (Rayonix, Evanston, IL, USA), and the cZ22-Fab/dC(GC)3 complex was cryoprotected with 25% glycerol (15523; Sigma–Aldrich, St. Louis, MO, USA). Data processing and scaling were performed with XDS [31] and HKL3000 [32]. The crystal structure of Fab was determined by molecular replacement using MOLREP in the CCP4 suite [33], with the CH1 domain of the Fab structure (7CHZ) [28] used as the search model. The complex was solved using the refined cZ22-Fab model and the Z-DNA model (2DCG) [1] as the search models. Model building and refinement were carried out using COOT [34] and PHENIX [35], respectively. Cross-validation was performed with R-free values calculated using 5% of the data, randomly selected. Solvent molecules were identified using difference Fourier (Fo–Fc) maps. The stereochemistry of the model was validated with MolProbity [36]. The final atomic coordinates and structure factors of cZ22-Fab, cZ22-Fab/dC(GC)3, and cZ22-Fab/d(CG)6 were deposited in the Protein Data Bank under the accession codes 9WRN, 9WU2, and 9WS0, respectively. The figures were prepared with UCSF Chimera (University of California, San Francisco, CA, USA) [37].
Results
Characterization of the chimeric anti-Z-DNA Fab fragment
A chimeric Fab fragment targeting Z-DNA (cZ22-Fab) was engineered from Z22, a murine IgG2b antibody originally developed from C57BL/6 mice [27]. The variable regions of the Z22 heavy (V_H_) and light (V_L_) chains were cloned and grafted onto a human IgG1 constant domain following previously described methods (Supplementary Fig. S1) [28]. An expression plasmid encoding cZ22-Fab with a C-terminal His-tag linked to the CH1 domain was transiently transfected into Expi293F cells, and the secreted Fab protein was purified from the culture supernatant via Ni²⁺ affinity chromatography. The purified cZ22-Fab displayed high purity, as confirmed by SDS–PAGE analysis (data not shown), and was used for functional assays and crystallographic studies.
First, cZ22-Fab binding to Z-DNA was validated, with a focus on CG-repeat sequences that are known to favour the Z-DNA conformation [26]. The DNA binding activity of cZ22-Fab was assessed by electrophoretic mobility shift assay (EMSA) after 2 h of incubation with FAM-labelled CG-repeat duplexes [d(CG)n, n = 3–5], with an 18-bp B-form sequence (18N-FAM) included as a negative control (Fig. 1A and Supplementary Fig. S2) [38]. Upon UV exposure, FAM-labelled d(CG)5 duplex clearly shifted towards the cathode, indicating complex formation with cZ22-Fab. Coomassie blue staining further revealed that compared with Fab alone, the Fab/DNA complexes migrated more prominently towards the anode. The extent of this shift increased with increasing protein/DNA molar ratios (1:1 to 5:1), whereas no change in mobility was observed with the negative control 18N-FAM. These results confirm that cZ22-Fab binds specifically to CG-repeat DNA duplex.
Biophysical characterization of cZ22-Fab-mediated B-to-Z DNA transition. (A) EMSA of cZ22-Fab/DNA complex formation. FAM-labelled d(CG)5 (left, Z-DNA-forming) or 18N-FAM (right, B-form sequence) duplexes were incubated for 2 h with increasing molar ratios of cZ22-Fab (1:1–5:1). DNA was visualized by UV detection, and the same gel was stained with Coomassie blue to detect protein. Free DNA, free Fab, and Fab/DNA complexes are indicated by black, blue, and red arrows, respectively. (B) CD spectra monitoring Fab-mediated B-to-Z DNA transition. Spectra were recorded from 230 to 320 nm for native d(CG)5 duplex (grey) and cZ22-Fab/d(CG)5 complexes at molar ratios of 1:1 (blue), 2:1 (green), and 4:1 (red). (C) Time-course CD analysis of the Fab-mediated B-to-Z transition at 255 nm over a 120-min period. (D) SV-AUC analysis of d(CG)5 duplex, cZ22-Fab, and the d(CG)5/cZ22-Fab complex. The sedimentation coefficients were 1.63 S (blue dashed), 3.62 S (green dashed), and 6.69 S (red) for free d(CG)5 duplex, free cZ22-Fab, and the complex, respectively. (E) SEC–MALS analysis of Fab/DNA complex formation. The flow rate was 0.5 ml/min with peaks at retention volumes of ∼12.8 ml for the cZ22-Fab/d(CG)5 complex, ∼16.2 ml for free cZ22-Fab, and ∼17.8 ml for free d(CG)5 duplex, represented in red, green dashed, and blue dashed, respectively. Estimated molar masses are indicated by the grey line. (F) SEC profiles of the Fab/DNA complexes at Fab:DNA molar ratios of 1:1, 2:1, 4:1, and 6:1, represented in blue, red, green, and grey, respectively.
Kinetics of cZ22-Fab-mediated B-to-Z DNA transition
CD spectroscopy is a primary method for monitoring the B-to-Z transition of DNA. B- and Z-DNA exhibit inverted CD spectra with isostrophic points at 255 and 295 nm, respectively. At 255 nm, B-DNA displays negative ellipticity, whereas Z-DNA is positive. Conversely, at 295 nm, the ellipticity signals are reversed [39, 40]. The ability of cZ22-Fab to mediate Z-DNA formation was examined by incubating a d(CG)5 DNA duplex with cZ22-Fab at P/N molar ratios of 1, 2, and 4, followed by CD measurements from 230 to 320 nm. The d(CG)5 duplex alone exhibited a typical B-DNA spectrum, which progressively shifted towards the Z-DNA profile with increasing cZ22-Fab concentration (Fig. 1B). A distinct isosbestic point was observed near 280 nm, indicating a direct two-state transition between the B- and Z-DNA conformation without significant accumulation of intermediates. At P/N = 4, the duplex was nearly converted to the Z-DNA spectrum, confirming the cZ22-Fab-mediated B-to-Z DNA transition. Transition kinetics were evaluated by time-course CD measurements at 255 nm over 120 min [41]. Nonlinear regression analysis revealed that the rate constants at P/N = 1, 2, and 4 were 0.0455, 0.0418, and 0.0447 min^−1^, respectively, corresponding to half-transition times of ∼15–17 min (Fig. 1C). The relative transition activity was concentration-dependent, reaching 100% at P/N = 4 and 61% and 41% at P/N = 2 and 1, respectively.
BLI was also employed to characterize the binding kinetics of cZ22-Fab to the d(CG)5 duplex. Concentration-dependent binding of cZ22-Fab (12.5–100 μM) was observed with probe-immobilized hairpin DNA d(CG)5, whereas no detectable binding was observed with the negative control B-form hairpin DNA 20N (Supplementary Fig. S3A and B). Sensorgrams displayed a slow but stable association that approached saturation within ~2 h, followed by gradual dissociation upon transfer to buffer. Local fitting of the sensorgrams yielded low association rate constants (kon) (∼10^0^ to 10^1^ M^−1 ^s^−1^), moderate dissociation rate constants (koff) (∼10^−3^ to 10^−4^ s^−1^), and apparent dissociation constants (KD) (∼10 to 30 μM), consistent with the observed binding behaviour (Supplementary Table S2). The kinetic data were analysed using a 1:1 binding model; however, systematic deviations between the experimental data and the fitted curves indicate that the model serves only to capture overall binding trends. To further define the binding stoichiometry of cZ22-Fab with the d(CG)5 duplex in solution, SV-AUC and SEC–MALS were employed.
Complex formation of cZ22-Fab with d(CG)5 DNA duplex
The binding mode and stoichiometry of cZ22-Fab with d(CG)5 DNA duplex were examined by mixing the Fab and DNA at a 3:1 molar ratio to enable observation of the DNA, protein, and complex states. After 2 h of incubation, each state was purified by SEC and analysed by SV-AUC. The c(s) distribution revealed sedimentation coefficients of 1.63, 3.62, and 6.69 S corresponding to d(CG)5 duplex, cZ22-Fab, and the complex, respectively (Fig. 1D). The cZ22-Fab/d(CG)5 complex exhibited a frictional ratio of 1.31, and the calculated molecular weight of 116.8 kDa closely matched the theoretical value for a 2:1 Fab/DNA complex (104.5 kDa) (Table 1). The complex remained intact during SV-AUC and consistently formed a stable 2:1 stoichiometry without dissociation, supporting a 2:1 binding mode. The analysis was further extended to CG-repeat DNA duplexes of different lengths [d(CG)n, n = 4–7] to assess Fab binding in a mixture without SEC purification (Supplementary Fig. S4). Based on the Fab/DNA ratio and theoretical molecular weight, cZ22-Fab/d(CG)4 and cZ22-Fab/d(CG)5 each formed a single 2:1 complex (103.3 and 104.5 kDa), cZ22-Fab/d(CG)6 produced both 2:1 (105.8 kDa) and 4:1 (204.2 kDa) complexes, and cZ22-Fab/d(CG)7 predominantly formed a 4:1 complex (205.5 kDa) (Supplementary Table S3). These results indicate that longer CG repeats support a length-dependent doubling of Fab binding capacity. SEC–MALS was simultaneously applied to validate the cZ22-Fab/d(CG)5 binding stoichiometry in the mixture. At a 2:1 molar ratio, the measured values for the d(CG)5 duplex (6.8 kDa), cZ22-Fab (50.8 kDa), and complex (112 kDa) were consistent with a 2:1 binding mode (Fig. 1E). Analysis of mixtures prepared at cZ22-Fab/d(CG)5 molar ratios of 1:1, 2:1, 4:1, and 6:1 revealed only a 2:1 complex, with increasing Fab concentrations, decreasing free DNA, and increasing complex abundance, indicating saturation at two Fab molecules per d(CG)5 duplex without higher-order species (Fig. 1F). SV-AUC and SEC–MALS analyses demonstrated that cZ22-Fab forms a stable 2:1 complex with d(CG)5 duplex, even though B-to-Z transition activity increases with increasing Fab concentration, and that the absence of a 1:1 intermediate supports a cooperative 2:1 binding mode.
Crystallographic analysis of cZ22-Fab recognition of DNA
The structural basis of DNA recognition by cZ22-Fab was examined by determining the crystal structures of the Fab alone and the Fab/DNA complex. Unliganded cZ22-Fab crystallized in space group P2_1_ with four Fab molecules per ASU (Supplementary Fig. S5A). The structure was refined to 2.15 Å, showing the canonical IgG fold with a well-conserved architecture in both the variable and constant domains (Table 2 and Supplementary Fig. S6A). For the Fab/DNA complex structures, two preannealed duplex DNA oligonucleotides, dC(GC)3 and d(CG)6, were used, and each complex crystallized in a distinct crystal form with resolutions of 2.35 and 3.27 Å, respectively. In addition, crystallization of the cZ22-Fab/d(CG)5 complex was carried out under various conditions; however, no crystals were obtained. The Fab/dC(GC)3 complex crystallized in space group P3_1_ with six Fab fragments and three dsDNA molecules per ASU, in which each DNA duplex is symmetrically bound by two Fab molecules (Supplementary Fig. S5B). DNA recognition is mediated by the complementarity-determining regions (CDRs) from the heavy chain (CDR2.H and CDR3.H) and the light chain (CDR1.L, CDR2.L, and CDR3.L), which engage the DNA from opposite sides (Table 2 and Supplementary Fig. S6B). In contrast, the Fab/d(CG)6 complex crystallized in space group I2_1_2_1_2_1_ with two Fab fragments and one ssDNA molecule per ASU (Supplementary Fig. S5C). Analysis of molecular packing revealed that a crystallographic twofold symmetry operation generated a complete double-stranded Z-DNA bound by four Fab molecules across adjacent ASUs (Table 2 and Supplementary Fig. S5C). Despite differences in crystal packing and binding stoichiometry, all the complex structures display DNA bound within the Fab antigen-binding cleft. These structural variations underscore the conformational adaptability of the Fab–DNA interface, highlighting how the DNA sequence and conformation influence Fab binding modes and provide mechanistic insight into cooperative Fab assembly, steric complementarity, and the flexible recognition of Z-DNA by antibody fragments.
Overall structure of the cZ22-Fab/dC(GC)3 complex
In the cZ22-Fab/dC(GC)3 complex structure, the short dC(GC)3 duplex is centrally positioned and symmetrically engaged by two cZ22-Fab fragments, each contacting opposite faces of the DNA duplex (Fig. 2A). The Fab variable domains (V_H_–V_L_) are oriented nearly collinearly with respect to the DNA duplex axis, forming an extended linear binding configuration. The two Fab fragments adopt an open V-shaped arrangement with an inter-Fab angle of ~140°, enabling simultaneous bivalent engagement without steric interference. This geometry represents a symmetric and sterically favourable recognition mode for the palindromic DNA sequence, dC(GC)3. Model validation with the Fo–Fc difference electron density map confirmed the Z-DNA duplex structure (Fig. 2B). The electron density, contoured at 2.5σ, is well defined along the entire DNA molecule in the cZ22-Fab/dC(GC)3 complex, supporting the placement of both the phosphate backbone and the base pairs. The map clearly reveals the helical structure and individual base stacking, which is consistent with a Z-form DNA conformation. The continuous density across both strands and at the 5′ and 3′ ends validates the CGCGCGC sequence and register, which is further emphasized by a 90° rotation that reveals uninterrupted density through the duplex core. The refined Z-form dC(GC)3 DNA duplex has an average B-factor of 36.2, closely matching that of the surrounding amino acid residues (average B-factor of 35.0), indicating that the DNA is well ordered and stably positioned within the complex.
Crystal structure of cZ22-Fab bound to dC(GC)3 Z-DNA duplex. (A) Two cZ22-Fab fragments symmetrically bind the short palindromic DNA sequence C-hanging dC(GC)3 duplex. Heavy (H) and light (L) chains are coloured purple/green and orange/cyan, respectively, for the two Fabs, with centrally positioned DNA shown as sticks. Viewed along the DNA axis, the Fabs align almost linearly; a 90° rotation reveals a V shape with an ∼140° inter-Fab angle. (B, C) dC(GC)3 Z-DNA duplex electron density map and structural features. A composite omit Fo–Fc map (green mesh, contoured at 2.5 σ) calculated from the cZ22-Fab/dC(GC)3 complex structure is shown in two orthogonal views. Close-up views show CpG and GpC Z-steps with guanine in syn and cytosine in anti conformation and alternating C2'-endo/C3'-endo sugar puckers. Red triangles indicate O4'–π interactions between the deoxyribose O4' and adjacent guanines. The full Z-DNA duplex is shown with strands in cyan and magenta; atoms are coloured by element: oxygen (red), nitrogen (blue), phosphorus (orange), and carbon (cyan/magenta). (D) Structural interaction between cZ22-Fab and dC(GC)3 Z-DNA duplex. Stick representations show key heavy (purple) and light (green) chain residues that participate in base stacking and backbone hydrogen bonding. The middle panel highlights electrostatic and hydrophobic interactions at the Fab–DNA interface. DNA phosphorus and oxygen atoms are coloured orange and red, respectively.
Z-DNA helical structure of the dC(GC)3 duplex
The crystal structure of the dC(GC)3 duplex in a sticky-ended Z-DNA conformation, as a 5′-end C-hanging Z-DNA duplex structure, was first observed in the cZ22-Fab/DNA complex, providing a unique and informative model of Z-DNA (Fig. 2C). The structure demonstrates that O4′–π interactions are a consistent structural feature that stabilizes Z-DNA at both the CpG and GpC dinucleotide steps. At these sites, guanine adopts a syn conformation while cytosine retains an anti conformation, resulting in the formation of Z-steps. All phosphate groups adopt the Z_I_ conformation, defining the zigzag phosphate backbone path of Z-DNA, and the deoxyribose sugar puckers alternate between C2′-endo and C3′-endo, reinforcing the left-handed helical geometry. The deoxyribose O4′ atoms further engage in stabilizing π-stacking interactions with adjacent guanine bases; thus, sugar–base stacking contributes to Z-DNA stability and conformation. Interestingly, the 5′-end C1 cytidine adopts an unpaired conformation despite being part of the alternating pyrimidine–purine sequence that promotes the structural bending characteristic of Z-DNA. Even unpaired, C1 still participates in base stacking and hydrogen bonding, contributing to the overall stability and structural integrity of the Z-DNA duplex.
cZ22-Fab and dC(GC)3 duplex interaction and the backbone-tracking recognition
Structural analysis of the cZ22-Fab/dC(GC)3 complex revealed a defined interaction pattern between specific Fab residues and Z-form DNA (Table 3 and Fig. 2D). The syn conformation of guanine in Z-DNA reorients the base over the deoxyribose sugar, repositioning the N7 atom into a more solvent-exposed and accessible position for protein interaction. cZ22-Fab engages in multiple Gua-specific contacts: Lys93.L forms a hydrogen bond with Gua4 N7, Ser104.H interacts with Gua6 N7, and Tyr32.L recognizes Gua2 N7*. These interactions enable the Fab to selectively recognize guanine in the syn configuration. In parallel, the Fab targets cytosine N4, with Tyr106.H interacting with Cyt5 N4, Ser104.H with Cyt7 N4, and Tyr32.L with Cyt3 N4*. These targeted contacts with Gua N7 and Cyt N4 establish base recognition that is closely related to the structural features of Z-DNA. The light chain residues Lys93 and Phe94 and the heavy chain residues Tyr61, Arg50, Arg52, and Tyr103 form a continuous interaction path along a single DNA strand, following the phosphate backbone from Cyt3 to Cyt7 (five phosphate groups) in the 5′ to 3′ direction. Among them, F94.L interacts with a DNA phosphate group via its backbone nitrogen, which is mediated by a water molecule. These residues collectively constitute the phosphate-clamping of the Fab paratope, comprising K93.L, Y61.H, R50.H, R52.H, and Y103.H, which stabilizes Z-DNA by engaging the phosphate backbone and contributes to the maintenance of its left-handed helical conformation. In addition to F94.L, two other water molecules bridge critical contacts to bases, such as those between Tyr106.H OH and Cyt5 N4 and between Tyr91.L OH and Gua2 O6*, increasing the structural integrity of the interface. Furthermore, Phe94.L, Tyr61.H, Ser104.H, Asn105.H, and Tyr106.H form nonpolar contacts with sugar or base moieties, whereas Tyr32.L and Tyr50.L stabilize the G2 base of the complementary strand. These hydrophobic interactions further stabilize the Fab/DNA complex. Overall, cZ22-Fab/DNA binding is mediated primarily by hydrogen bonding to the zigzag phosphate backbone, accompanied by selective nucleobase contacts and stabilization of the complementary strand. This binding mode suggests that cZ22-Fab relies on phosphate-clamping and base interactions to recognize the unique architecture of the Z-form DNA helix.
Z-DNA-mediated symmetric assembly of cZ22-Fab
Structural adaptability across different DNA lengths was examined by determining the crystal structure of cZ22-Fab in complex with preannealed d(CG)6 DNA duplex. In contrast to the cZ22-Fab/dC(GC)3 complex, the refined structure revealed a 2:1 stoichiometry, with two Fab molecules (black/white and yellow/green) bound to one strand of d(CG)6 duplex in ASU, the observed DNA helix adopting a Z-form conformation (Fig. 3A). Each Fab (Fab I and II) interacts primarily with one face of the DNA, forming extensive contacts along the phosphate backbone. A crystallographic twofold symmetry operation extends the assembly, generating a symmetric complex composed of four Fab molecules (Fab I, II, I*, and II*) and two Z-DNA strands, ss d(CG)6 and its symmetry mate ss d(CG)6*, which together form a double-stranded Z-DNA helix (Fig. 3B). This Z-DNA duplex bridges the Fab pairs, resulting in a compact, symmetric Fab/DNA complex stabilized by DNA-mediated multivalent interactions. A surface representation highlights the tight spatial packing and ordered architecture of this higher-order assembly within the crystal (Fig. 3C). These findings indicate that Z-DNA can also serve as a structural scaffold, promoting Fab association through DNA-mediated bridging.
Crystal structure of cZ22-Fab bound to d(CG)6 Z-DNA duplex. (A) Z-DNA-mediated Fab assembly. Two Fab molecules (black/white and yellow/green) are bound to a single-stranded Z-form DNA oligomer, ss d(CG)6 (light pink), indicating a 2:1 Fab:ssDNA binding configuration. (B) Application of the crystallographic twofold symmetry operation generates a symmetric assembly, introducing a second set of Fab molecules (orange/blue and cyan/purple, labelled Fab) and a symmetry-related ss d(CG)6* (magenta). Together, ss d(CG)6 and ss d(CG)6* form a double-stranded Z-DNA helix that bridges the Fab and Fab* molecules, thus mediating multivalent interactions and enabling the formation of a symmetric higher-order structure. (C) Surface representation of the Fab/DNA complex, highlighting the spatial organization and tight packing of the Fabs (coloured by chain). The inset (top right) shows the bound Z-DNA duplex within the Fab interface. (D) The black/white Fab interacts with the 3′ end of the Z-form DNA duplex, which is formed by ss d(CG)6 and its symmetry mate ss d(CG)6*, engaging a four-base segment. The yellow/green Fab binds to the opposite strand. Key residues from the heavy chain (black) and light chain (white) are shown as sticks. The Fab recognizes the Z-DNA primarily through interactions with the phosphate backbone of a single DNA strand, mediated by light chain residues F94.L and K93.L, and heavy chain residues R50.H, R52.H, Y61.H, and S104.H (magenta). (E) Structural comparison of cZ22-Fab bound to different Z-DNA lengths. Superimposed structures of cZ22-Fab bound to dC(GC)3 duplex (sky blue DNA, purple Fab) and d(CG)6 duplex (magenta DNA, light green Fab) reveal conserved central contacts and additional interactions in the extended duplex. The left panel shows the overall alignment, while the right panel highlights specific contacts by R53.L, and S104.H (magenta) with the 5′ region of the longer d(CG)6 Z-DNA duplex.*
Adaptation of cZ22-Fab to varying Z-DNA duplex lengths
The DNA-binding adaptability of cZ22-Fab was assessed using crystal structures of Fab/Z-DNA complexes with two different duplex lengths, dC(GC)3 and d(CG)6. In the d(CG)6 complex, the Fab II engages a four-base segment at the 3′ end of the Z-DNA helix, primarily through phosphate backbone interactions mediated by the light chain residues K93.L and F94.L and the heavy chain residues Y61.H, R50.H, and R52.H. In addition, Y106.H forms both hydrophobic and polar interactions with the bases C11 and G12, respectively (Fig. 3D and Supplementary Fig. S7). This partial DNA engagement represents a minimal binding mode, involving only a short stretch of four base pairs at the 3′ end of the Z-DNA duplex. The interaction is mediated by a subset of paratope residues, which primarily contact the DNA backbone with limited base recognition. Structural superposition of the Fab bound to dC(GC)3 and d(CG)6 duplex revealed a conserved central binding mode, with both complexes contacting the core Z-DNA segment (Fig. 3E). However, the longer d(CG)6 duplex allows additional interactions for Fab I, with residues Y32.L, Y50.L, R53.L, and S104.H specifically engaging the extended Z-DNA helix. Y50.L and R53.L interact with the 5′ region of the complementary strand, and Y50.L, in particular, helps stabilize the phosphate backbone, contributing to the maintenance of the Z-DNA conformation (Fig. 4). These findings indicate that cZ22-Fab possesses a modular, length-adaptive binding interface that accommodates extended Z-DNA conformation by recruiting additional contact residues.
Comparative analysis of Fab/Z-DNA interactions in dC(GC)3 and d(CG)6 complexes. (A) Surface representations of cZ22-Fab complexes bound to Z-DNA duplexes, namely, Fab/dC(GC)3 (left panel) and Fab I/d(CG)6 (right panel), in two distinct binding modes. Z-DNA is shown as a ribbon model. The highlighted surface regions correspond to the Fab paratopes, and the calculated paratope surface areas are shown. Key DNA contact residues are indicated, with heavy chain residues labelled in black and light chain residues in orange. (B) Schematic representation of Fab/Z-DNA contacts. DNA bases are shown as horizontal bars. Interactions are indicated by blue lines for hydrophilic interactions, green lines for hydrophobic contacts, and red circles for water-mediated interactions. Fab residues are enclosed in boxes, with a white background for light chain residues and a grey background for heavy chain residues. The asterisk () indicates the complementary strand.*
DNA length–dependent engagement of the cZ22-Fab paratope
The cZ22-Fab paratope displays modular adaptability, engaging DNA with varying extents of buried surface area and residue usage. Three distinct DNA-binding configurations were analysed (Fig. 4 and Supplementary Fig. S7). In the C-hanging dC(GC)3 duplex binding mode, the paratope buries ~580 Ų of the contact area. Light chain residues (Y32, Y50, Y91, K93, and F94) and heavy chain residues (R50, R52, Y61, Y103, S104, N105, and Y106) form a compact recognition interface. In the d(CG)6 duplex binding modes, two distinct Fab interactions were observed. In the Fab I mode, a longer DNA duplex enables deeper engagement, expanding the interface to ~650 Ų, the largest among the three binding modes. Compared with the dC(GC)3 duplex binding mode, this interaction includes the additional residues Y50 and R53 from the light chain paratope, which contribute to stabilizing the extended DNA base residue G2 of the complementary strand. This conformation highlights the antibody’s maximal adaptability in accommodating extended Z-DNA helices. In the d(CG)6 duplex Fab II binding mode (Supplementary Fig. S7), the second Fab associates with the terminal end of the same DNA. The buried surface area is smaller (∼440 Ų), indicating an auxiliary interaction. A reduced paratope set (K93.L, F94.L, R50.H, R52.H, Y61.H, S104.H, Y106.H) dominates the interface, suggesting a cooperative but less extensive contribution to overall DNA stabilization compared with that of Fab I.
Interaction networks of cZ22-Fab with Z-DNA
Interaction diagrams illustrate how the cZ22-Fab paratope adapts to Z-DNA of varying lengths (Fig. 4B). In the C-hanging dC(GC)3 duplex binding mode (left panel), the paratope engages five bases (C3–C7) and two complementary-strand bases (G2–C3) through both hydrophilic and hydrophobic interactions, with three water molecules bridging additional contacts. In the d(CG)6 duplex Fab I binding mode (right panel), a longer DNA duplex expands the interaction network. The paratope interacts with six bases (C5–G10) and three complementary-strand bases (G2–G4). The conserved heavy chain residue S104 provides an additional contact to the extended base G10, whereas the light chain residues Y50 and R53 contact bases G2 and C3 of the complementary strand, thereby stabilizing the extended helix. These extra interactions increase the buried surface area, supporting stronger DNA binding. However, the water-mediated interactions observed in the cZ22-Fab/dC(GC)3 complex were not detected in the cZ22-Fab/d(CG)6 structure because of the resolution limit. In the d(CG)6 duplex Fab II binding mode (Supplementary Fig. S7), the second Fab binds the terminal four bases (C9–G12) of the same DNA strand, with no contacts with the complementary strand. The interface is less extensive, despite involving both heavy chain residues (R50, R52, Y61, S104, and Y106) and light chain residues (K93 and F94), resulting in a smaller buried surface area. This binding pattern suggests a more auxiliary role than that of Fab I, contributing to cooperative stabilization of the Z-DNA conformation.
Inactive cZ22-Fab mutants demonstrate the functional importance of R50.H and Y106.H in Z-DNA recognition
Structural superposition of the cZ22-Fab/Z-DNA complexes with the apo Fab revealed that the overall backbone conformation remains closely aligned, while several DNA-contacting residues undergo notable side-chain rotations upon binding (Fig. 5A). Among these, R50.H and Y106.H display the largest conformational changes. R50.H rotates toward the DNA backbone to engage in a phosphate-clamping interaction, whereas Y106.H reorients to mediate hydrophobic stacking with C11 and a polar contact with G12. To assess the contribution of these interactions to Z-DNA binding, two alanine-substituted mutants, cZ22-FabR50A.H and cZ22-FabY106A.H, were generated to disrupt the phosphate-clamping and base-stacking modules, respectively. CD spectroscopy showed that d(CG)5 duplex incubated with either mutant retained a B-form DNA signature and lacked the negative ellipticity at 295 nm typically associated with Z-DNA, indicating a failure to induce B-to-Z transition or stabilize the Z-conformation (Fig. 5B and C). Consistent with these findings, SEC revealed no complex formation between either mutant, as the elution profiles showed distinct peaks corresponding to unbound protein and free DNA, without the earlier-eluting complex peak observed with wild-type cZ22-Fab (Fig. 5D and E). These results indicate that R50.H and Y106.H are critical for Z-DNA recognition and stabilization, and that alanine substitution at either position abrogates functional interaction with Z-DNA.
Structural and biophysical impacts of R50A.H and Y106A.H mutations on cZ22-Fab recognition of Z-DNA. (A) Structural comparison of the CDR conformations between the Fab/Z-DNA complexes and free-form Fabs. The CDRs of the Fab/Z-DNA complexes were superimposed on those of four free-form Fabs, yielding RMSD values of 0.47–0.73 Å across 26 Cα pairs. DNA-binding residues are shown as stick models: blue for the Fab/d(CG)6 complex and green for the Fab/dC(GC)3 complex. The corresponding residues from free-form Fabs are shown in pink. Although side chain orientations are generally conserved among free Fabs, notable conformational changes occur upon DNA binding. In particular, R50.H and Y106.H exhibit substantial side chain rotations. (B, C) CD spectra monitoring the Z-DNA formation of d(CG)5 duplex in a mixture with the inactive cZ22-Fab mutants, cZ22-FabR50A.H and cZ22-FabY106A.H, respectively. Ellipticity was recorded from 230 to 320 nm for native d(CG)5 duplex (grey) and inactive Fab with d(CG)5 duplex at a molar ratio of 4:1 (green). d(CG)5 duplex with wild-type cZ22-Fab, which formed Z-form DNA, was included as a control for comparison (red dashed). (D) SEC profile of wild-type cZ22-Fab with d(CG)5 duplex, showing the Fab/d(CG)5 complex at ∼1.6 ml (red), free Fab at ∼2.1 ml (green dashed), and free d(CG)5 duplex at ∼2.3 ml (blue dashed). (E, F) SEC profiles of R50A.H and Y106A.H inactive cZ22-Fab mutants mixed with d(CG)5 duplex, respectively. Mutant Fab and DNA mixtures (red) elute at ∼2.1 and ∼2.3 ml, overlapping with free Fab (green dashed) and free d(CG)5 duplex (blue dashed). (G) SEC profile of cZ22-Fab with d(TG)5 duplex show no complex formation, with Fab and DNA mixtures eluting at ∼2.1 and ∼2.3 ml (red), corresponding to free Fab (green dashed) and free d(TG)5 duplex (blue dashed) (H, I) SEC profiles of cZ22-Fab with d(CG)57N duplex and hairpin DNA d4N(CG)54N, respectively, showing Fab/DNA complex formation at ∼1.6 ml (red); free Fab at ∼2.1 ml (green dashed), and free DNA duplex at ∼2.1 ml or free hairpin DNA at ∼2.0 ml (blue dashed).
Antigen sequence specificity of cZ22-Fab binding oligonucleotides
cZ22-Fab recognizes and stabilizes CG-repeat DNA through phosphate-clamping and base interactions at the CGCGCGC epitope. To assess sequence specificity, d(TG)5 duplex, another Z-forming sequence [7], was pre-incubated with cZ22-Fab and analysed by SEC for complex formation. SEC was employed to assess Fab/DNA complex formation in solution under physiological salt conditions by resolving free and complexed species. However, no complex formation was detected, indicating that cZ22-Fab does not bind the TG-repeat duplex under physiological salt conditions (Fig. 5G). Additionally, the binding of cZ22-Fab to CG-repeat sequences within a mixed DNA conformational context was examined using a designed Z–B junction potential sequence, d(CG)_5_7N, resulting in the formation of a cZ22-Fab/d(CG)_5_7N complex with a retention volume comparable to that of the cZ22-Fab/d(CG)5 complex (Fig. 5H). Similarly, SEC analysis of a hairpin DNA containing an embedded CG-repeat sequence, d4N(CG)_5_4N, revealed a complex formation with cZ22-Fab and an elution profile similar to that of the cZ22-Fab/d(CG)5 complex (Fig. 5I). Collectively, these SEC analyses demonstrate that cZ22-Fab selectively forms complexes with CG-repeat DNA and that such interactions are preserved across different DNA sequence contexts.
Discussion
The B-to-Z transition is one of the most notable manifestations of the dynamic phase transition of DNA. While the right-handed B-form predominates under physiological conditions, CG-rich sequences can adopt the left-handed Z-form in response to specific environmental conditions or chemical modifications. This conformational switch requires destabilization of the energetically favoured B-form and stabilization of the less favourable Z-form, which involves a substantial energy barrier. Z-DNA is therefore generally transient and unstable; however, in recent years, increasing evidence has supported its involvement in transcriptional regulation, autoimmune diseases, and potential therapeutic applications [42–45]. Understanding how the B-to-Z transition occurs and how Z-DNA can be detected is therefore of central importance. At present, Z-DNA binding proteins with Zα domains are known to sense and enforce the left-handed conformation by triggering the B-to-Z transition, although their recognition mechanisms require further study. Antibodies are particularly valuable for detection purposes, as their adaptable binding surfaces provide a distinct strategy for probing not only proteins but also nucleic acid structures such as Z-DNA. The Z22 monoclonal antibody was employed early as a probe for Z-DNA [13, 46]; however, the detailed mechanisms of how it recognizes and stabilizes the Z-DNA conformation have remained unclear. This study presents the first structural and mechanistic insights into the antibody-mediated B-to-Z transition through the analysis of cZ22-Fab.
Our results demonstrated that the chimeric Fab fragment, cZ22-Fab, specifically recognizes CG-repeat dsDNA, as confirmed by EMSA (Fig. 1A). This finding indicates that recognition is not restricted to the IgG format but is preserved in the Fab form without Fab/IgG format-related discrepancies [47]. CD spectroscopy further revealed that cZ22-Fab mediates the B-to-Z transition of d(CG)5 duplex (Fig. 1B). The observation of a distinct isosbestic point at ∼280 nm provides compelling evidence that the cZ22-Fab-mediated B-to-Z transition follows a two-state mechanism. This spectral feature indicates that the conformational change occurs directly between B-DNA and Z-DNA with no detectable accumulation of intermediate states. Notably, a similar spectral signature has been reported for Z-DNA binding proteins such as the ADAR1 Zα domain and the viral E3L Zα domain, suggesting that cZ22-Fab can also mediate B-to-Z DNA transition in a manner comparable to that of natural Z-DNA binding proteins. To our knowledge, this is the first report of a monoclonal anti-Z-DNA antibody with the capacity to mediate the B-to-Z transition. The transition process was concentration dependent, and although the Fab/DNA ratios varied, the half-transition times remained similar, indicating that the Fab concentration governs the extent rather than the intrinsic rate of the transition (Fig. 1C). Comparison of the CD and BLI results revealed that the B-to-Z transition proceeded slowly, with an estimated duration of ~2 h (Fig. 1B and Supplementary Fig. S3). The association rate constants (kon), which are markedly lower than the typical range for protein–protein interactions (10^4^–10^6^ M^−1 ^s^−1^), are consistent with the mechanism being constrained by the two-hour DNA conformational transition rather than by normal association. In contrast, the dissociation rate constants (koff) ranged from 10⁻^3^ to 10^-4^ s^−1^, which are consistent with the formation of relatively stable complexes. The corresponding apparent binding affinities were estimated to be in the micromolar range, reflecting a moderate but specific interaction under the tested conditions (Supplementary Table S2). Although the BLI traces could be locally fitted, the experimental curves did not align perfectly with the 1:1 binding model, suggesting that binding behaviour may not follow a simple 1:1 binding mode. Nevertheless, the overall binding range and trend are similar to those of the CD measurements and those reported in previous studies [41]. The observed transition involves a pronounced conformational rearrangement of the helix and naturally proceeds on a relatively slow timescale, spanning minutes to hours [7]. These findings strongly support the identification of cZ22-Fab as a unique anti-Z-DNA monoclonal antibody that mediates the B-to-Z transition and stabilizes Z-DNA in a concentration-dependent manner.
Subsequently, the SV-AUC and SEC–MALS results demonstrated that at the end point of the B-to-Z transition, cZ22-Fab consistently formed a stable 2:1 complex with the d(CG)5 duplex, with no detectable intermediate species (Fig. 1D–F). This stoichiometry was maintained even in the presence of excess Fab, indicating that both binding sites of d(CG)5 are fully occupied by cZ22-Fab. On the basis of the complex structure, we believe that the absence of a detectable 1:1 complex suggests that initial Fab binding is unstable or energetically unfavourable unless it is accompanied by the recruitment of the second Fab (Figs 2 and 3). Relative to dC(GC)3, d(CG)6 has an extended CG-repeat length, introducing an additional Fab binding site with a minimal four-base platform at the terminus. Despite this extension, each fully accommodated binding module remains dimeric, yielding a canonical 2:1 cZ22-Fab/CG unit. The apparent 4:1 stoichiometry observed in the structure therefore arises from the extended DNA length rather than reflecting a distinct binding stoichiometry (Supplementary Fig. S4C and S5C).
Furthermore, structural analyses together with SV-AUC results demonstrate that cZ22-Fab discriminates CG-repeat sequences in a length-dependent manner, with longer repeats promoting more stable Z-DNA conformation and more uniform antibody binding. (Figs 2 and 3 and Supplementary Fig. S4). These results support a cooperative mechanism in which two Fab molecules act together to stabilize the Z-DNA conformation. This cooperative behaviour resembling an allosteric mechanism has now been demonstrated for the first time in a monoclonal anti-Z-DNA antibody, which exhibits similarity to known Zα domains of Z-DNA binding proteins such as ADAR1, ZBP1, and viral E3L, which cooperatively bind and stabilize Z-DNA in structural studies [48–50].
Structural analysis of the cZ22-Fab/Z-DNA complex revealed nucleotide site-specific multilayer recognition that integrates several key interactions, including sequential phosphate-clamping, base hydrogen bonding, and auxiliary hydrophobic and water-mediated interactions. Phosphate-clamping is the central interaction in stabilizing the Z-DNA backbone. In cZ22-Fab/dC(GC)3 complex, a primary clamp formed by K93.L, Y61.H, R50.H, R52.H, and Y103.H stabilizes one strand of the DNA, while Y32.L, Y50.L, and Y91.L reinforce the complementary strand to buckle the left-handed helix (Figs 2–4). More importantly, the clamp follows a sequential path along the DNA backbone, extending from Cyt3 to Cyt7. This sequential phosphate-clamping stabilizes the zigzag phosphate pattern of Z-DNA and is critical for maintaining the left-handed conformation. Like phosphate clamping, base recognition contributes critically to stabilization. The syn conformation of guanine repositions the N7 atom into an exposed orientation, enabling guanine-specific contacts with K93.L–Gua4 N7, S104.H–Gua6 N7, and Y32.L–Gua2 N7*. Cytosine-specific residues are recognized in parallel through hydrogen bonds to the N4 amino group, mediated by Y106.H–Cyt5 N4, S104.H–Cyt7 N4, and Y32.L–Cyt3 N4* (Figs 2–4). These guanine- and cytosine-specific interactions define a base-recognition pattern that reinforces the selectivity of the antibody for the Z-DNA conformation. Further interactions also contribute to the formation of the left-handed helix. Hydrophobic contacts involving Phe94.L–Gua4 C4′, Tyr61.H–Gua4 C5′, Ser104.H–Gua6 C8, Asn105.H–Gua6 C8, and Tyr106.H–Gua6 C8 play essential roles by occupying the hydrophobic pocket and stabilizing the helical conformation. In addition, water-mediated contacts, including F94.L–Cyt3 phosphate, Y106.H–Cyt5 N4, and Y91.L–Gua2 O6*, increase both the specificity and stability of the Fab–DNA interface while providing the spatial accommodation required for the left-handed helix (Figs 2–4). Beyond the predominant recognition, the Fab/d(CG)6 complex exhibits a different arrangement in which Y103.H adopts an alternative orientation and loses its phosphate contact. The paratope compensates by reorganizing to recruit Y32.L, Y50.L, and R53.L, with Y50.L establishing a direct phosphate interaction that broadens the interface, stabilizes the elongated Z-DNA helix, and underscores the adaptive plasticity of cZ22-Fab (Fig. 4 and Supplementary Table S4). Moreover, in the minimal binding mode of the Fab/d(CG)6 complex, Fab II contacts only four base pairs at the 3′ end, where recognition is mediated by a reduced subset of the phosphate-clamping (K93.L, Y61.H, R50.H, and R52.H) (Supplementary Fig. S7). Consistent with these structural observations, the absence of crystals for the cZ22-Fab/d(CG)5 complex is likely due to crystal-packing constraints, as each Fab engages approximately seven base pairs, and the extended d(CG)5 duplex may hinder the formation of an ordered crystal lattice. Overall, the cooperative effect of these interactions highlights an adaptive antibody strategy to sense and stabilize alternative DNA conformations, which provides mechanistic insight into the antibody-mediated stabilization of left-handed DNA.
Given that the Zα domain has been well characterized in their interactions with Z-DNA, a comparison with cZ22-Fab offers valuable context for understanding distinct recognition mechanisms. Structurally, cZ22-Fab engages a broader epitope, contacting seven base pairs (Fig. 4), whereas the Zα domain typically interacts with five bases [51]. The position along the DNA duplex at which binding initiates also differs, with cZ22-Fab initiating binding in the 5′ to 3′ direction by contacting the phosphate of cytosine, whereas the Zα domain initiates binding in the 3′ to 5′ direction by contacting the phosphate of guanine. Moreover, in cZ22-Fab, the heavy chain interacts with six bases on one strand of the Z-DNA duplex, while the light chain engages three bases on the complementary strand (Fig. 4). In contrast, the Zα domain recognizes a single strand of Z-DNA across five bases [51]. Both cZ22-Fab and Zα domain engage the DNA phosphate backbone through lysine and arginine residues. However, the Zα domain relies on conserved tryptophan and proline residues, which are absent in the cZ22-Fab antibody-binding (Fig. 4) [51]. While both proteins target left-handed Z-DNA structure, their interaction profiles are distinct. Zα domain primarily contacts the phosphate group, whereas cZ22-Fab forms additional interactions with the bases and deoxyribose (Fig. 4) [51]. Consistently, under physiological salt conditions, the absence of binding to the Z-DNA-forming TG-repeat sequence indicates that this altering sequence is insufficient for cZ22-Fab recognition (Fig. 5G) [7]. This is consistent with previous studies reporting that Z22 does not bind linear TG-repeat double-stranded oligonucleotides under low-salt conditions, whereas binding was observed for a supercoiled plasmid containing an inserted poly[(dTG)·(dCA)] sequence [52]. Z-form TG repeats are known to require a high energy barrier to form [53]; therefore, whether their recognition by antibodies requires additional structural or topological features remains to be determined. Furthermore, cZ22-Fab binding is maintained when CG-repeat sequences are presented within Z–B junction potential sequences and diverse flanking sequence contexts, supporting (CG)n as the epitope of cZ22-Fab (Fig. 5H and I) [54]. These features underscore the recognition of cZ22-Fab and reveal a fundamentally different mode of recognition compared to the Zα domain.
Structural analysis indicates that cZ22-Fab recognizes Z-DNA through specific residues. Among the paratope residues, R50.H and Y106.H play central roles in DNA recognition, mediating phosphate-clamping and hydrophobic interactions, respectively, and are identified as key contact points based on structure superposition analysis, which revealed the largest side-chain rotations at these positions upon Z-DNA binding (Fig. 5A and Supplementary Fig. S8). Substitution of Y106.H disrupts the hydrophobic contact mediated by the aromatic side chain, whereas substitution of R50.H results in the loss of key interactions with the DNA phosphate backbone, disrupting the central phosphate-clamping network and abolishing two hydrogen-bond interactions (Fig. 5A and Supplementary Fig. S8). To validate the structural observations, two inactive Fab mutants, cZ22-FabR50A.H and cZ22-FabY106A.H, at R50.H and Y106.H were designed for binding assay. Both mutants resulted in a complete loss of binding to d(CG)5 duplex, providing biochemical evidence that these residues are essential for Z-DNA interaction (Fig. 5B and C). Notably, a prior study demonstrated that substitution of Y106.H with phenylalanine did not alter Z-DNA binding, suggesting that aromaticity at this position is crucial [55, 56]. Additional residues, including S104.H and N105.H, were examined in previous studies, where substitutions corresponding to 22H-S98A and 22H-N99Q mutants were reported to significantly affect Z-DNA binding [55]. Although both heavy and light chains contribute to Z-DNA binding in the cZ22-Fab structure, previous biochemical studies indicate that the heavy chain plays a dominant role, as it retains binding activity in the absence of the light chain [57]. This also consists with earlier reports highlighting the critical role of the Z22 heavy chain CDR3 in Z-DNA recognition [55, 58].
Prior studies have shown that the B-to-Z transition requires destabilization of the B-form and stabilization of the less favourable Z-form, involving major reorganization of the phosphate backbone orientation and base-pair tilt. However, the molecular mechanisms underlying this conformational switch remain incompletely understood [59]. On the basis of the structural analysis of the cZ22-Fab/DNA complex, we propose two models to explain the B-to-Z transition (Fig. 6). In the first model, DNA duplexes in solution fluctuate between conformations, with short segments occasionally flipping into the Z-form, most likely driven by Brownian motion [3]. Although these transient Z-DNA segments constitute only a minor population, they provide accessible binding sites for cZ22-Fab. The binding of the first Fab stabilizes the left-handed conformation and is followed by the recruitment of a second Fab, thereby leading to cooperative stabilization of the Z-DNA helix. The second possible model is that the B-to-Z transition is actively driven by protein binding. In this model, the first Fab directly contacts B-DNA via charge interaction and induces structural changes that promote the B-to-Z transition. We propose that this process may proceed through four steps: (i) Lysine and arginine residues neutralize the negative charges of the B-DNA phosphate backbone, lowering the energy barrier. (ii) Hydrophobic residues promote helical rotation, shifting the duplex from right-handed B-DNA to left-handed Z-DNA. (iii) Guanine bases adopt the syn conformation, and the left-handed helix is further stabilized by phosphate-clamping and base recognition. (iv) The initial Fab binding triggers a local B-to-Z transition, which in turn promotes the recruitment of the second Fab to stabilize the extended Z-DNA helix. As the DNA is further extended, the following Fab molecules can be cooperatively accommodated, allowing increased Fab occupancy along the Z-DNA helix. The two proposed models illustrate the possible mechanisms of the B-to-Z transition; however, future studies combining time-resolved structural analysis and single-molecule biophysics will be needed to evaluate these models.
Schematic representation of models for cZ22-Fab-mediated B-to-Z DNA transition.
Finally, most antibodies are raised against antigens with well-defined tertiary structures. However, many atypical antigens, such as polysaccharides and synthetic molecules like PEG, exhibit flexible conformations yet still elicit highly specific antibody responses [60, 61]. Despite their intrinsic structural flexibility, these antigens can be recognized with remarkable specificity through conformational adaptability, either accessible in solution or induced upon antibody binding. These flexible molecules also pose challenges for structural characterization, as their conformational transitions are difficult to capture. Although current physicochemical methods such as CD, NMR, thermal stability assays, and small-angle X-ray or neutron scattering can monitor transitions and estimate kinetics, they are limited in resolving intermediate states [62–66]. Conformation-specific antibodies may serve as valuable tools to stabilize and visualize specific states of such dynamic molecules. Similarly, the study of antibodies targeting nucleic acids is critically important, as the lack of specialized anti-DNA tools presents significant challenges for detailed investigations [67]. This limitation hinders the exploration of many protein–DNA regulatory mechanisms and prevents deeper insights into gene function and biological significance. Additionally, Z-DNA can act as an autoantigen in systemic autoimmune diseases, including SLE and RA, where anti-Z-DNA antibodies are frequently detected and correlated with disease activity. Although cZ22-Fab was originally derived from mice immunized with brominated Z-DNA and not from classical autoimmune models, studies have confirmed the presence of anti-Z-DNA autoantibodies in both SLE patients and SLE model mice [22, 25, 27, 68, 69]. Notably, the original Z22 antibody has been widely applied in diagnostic assays for the detection of Z-DNA, serving as a molecular probe to visualize or quantify Z-DNA formation in various experimental studies [70–72]. In summary, we report the first high-resolution structure of an antibody–Z-DNA complex, showing that the monoclonal antibody cZ22-Fab, a specific Z-DNA tool, selectively binds and stabilizes the left-handed helix. This structural insight not only establishes a molecular basis for understanding Z-DNA recognition in autoantibody responses in autoimmune diseases but also provides a foundation for exploring how dynamic conformational transitions, such as the B-to-Z switch, contribute to immunogenicity.
Supplementary Material
gkag160_Supplemental_File
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Wang AH-J, Quigley GJ, Kolpak FJ et al. Molecular structure of a left-handed double helical DNA fragment at atomic resolution. Nature. 1979;282:680–6. 10.1038/282680 a 0.514347 · doi ↗ · pubmed ↗
- 2Rich A, Nordheim A, Wang AH-J. The chemistry and biology of left-handed Z-DNA. Annu Rev Biochem. 1984;53:791–846. 10.1146/annurev.bi.53.070184.004043.6383204 · doi ↗ · pubmed ↗
- 3Rich A, Zhang S. Timeline: Z-DNA: the long road to biological function. Nat Rev Genet. 2003;4:566–72. 10.1038/nrg 1115.12838348 · doi ↗ · pubmed ↗
- 4Wang G, Vasquez KM. Dynamic alternative DNA structures in biology and disease. Nat Rev Genet. 2023;24:211–34. 10.1038/s 41576-022-00539-9.36316397 PMC 11634456 · doi ↗ · pubmed ↗
- 5Rice GI, Kasher PR, Forte GM et al. Mutations in ADAR 1 cause Aicardi–Goutieres syndrome associated with a type I interferon signature. Nat Genet. 2012;44:1243–8. 10.1038/ng.2414.23001123 PMC 4154508 · doi ↗ · pubmed ↗
- 6Lee J, Kim YG, Kim KK et al. Transition between B-DNA and Z-DNA: free energy landscape for the B-Z junction propagation. J Phys Chem B. 2010;114:9872–81. 10.1021/jp 103419 t.20666528 · doi ↗ · pubmed ↗
- 7Kim SH, Lim SH, Lee AR et al. Unveiling the pathway to Z-DNA in the protein-induced B-Z transition. Nucleic Acids Res. 2018;46:4129–37. 10.1093/nar/gky 200.29584891 PMC 5934635 · doi ↗ · pubmed ↗
- 8Takaoka A, Wang Z, Choi MK et al. DAI (DLM-1/ZBP 1) is a cytosolic DNA sensor and an activator of innate immune response. Nature. 2007;448:501–5. 10.1038/nature 06013.17618271 · doi ↗ · pubmed ↗