Extracellular DNA: A Missing Link in the Pathogenesis of Ectopic Mineralization

Abstract Although deoxyribonucleic acid (DNA) is the genetic coding for the very essence of life, these macromolecules or components thereof are not necessarily lost after a cell dies. There appears to be a link between extracellular DNA and biomineralization. Here the authors demonstrate that extracellular DNA functions as an initiator of collagen intrafibrillar mineralization. This is confirmed with in vitro and in vivo biological mineralization models. Because of their polyanionic property, extracellular DNA molecules are capable of stabilizing supersaturated calcium phosphate solution and mineralizing 2D and 3D collagen matrices completely as early as 24 h. The effectiveness of extracellular DNA in biomineralization of collagen is attributed to the relatively stable formation of amorphous liquid droplets triggered by attraction of DNA to the collagen fibrils via hydrogen bonding. These findings suggest that extracellular DNA is biomimetically significant for fabricating inorganic–organic hybrid materials for tissue engineering. DNA‐induced collagen intrafibrillar mineralization provides a clue to the pathogenesis of ectopic mineralization in different body tissues. The use of DNase for targeting extracellular DNA at destined tissue sites provides a potential solution for treatment of diseases associated with ectopic mineralization.


Immunofluorescence
Cells seeded on 12-well chamber slides (n = 6) were fixed in 4 % paraformaldehyde and incubated with goat serum for 30 min at room temperature. The cell samples were individually exposed to collagen-I rabbit anti-mouse primary antibody (Ab34710, Proteintech, Rosemont, IL, USA) at 4 °C for 12 h. After washing, Cy3 goat anti-rabbit secondary antibodies (111-165-003, Jackson ImmunoResearch Laboratories, Inc., West Grove, PA, USA) and alizarin red S were added and incubated for 1 h in the dark. The cells were mounted in anti-fade mountant and stained with SYTOX™ Green for CLSM). For measurement of extracellular nucleic acid fluorescence intensity, areas encompassing extracellular nucleic acids in the CSLM images were manually-selected and measured via ImageJ software (National Institute of Health, Bethesda, MD, USA). [45]

Alizarin red S staining of calcification
Mineralized nodules were analyzed after the cells were cultured in osteogenic medium for a destined period (n = 3). After fixation in 4% paraformaldehyde, the cells were stained with alizarin red S. Mineralization was quantified by colorimetric detection (NanoDrop2000 ultraviolet spectroscopy, Thermo Fisher Scientific) at 405 nm after the nodules were treated with 10% acetic acid and 10% ammonium hydroxide in 96-well plates. [46] 2.5. Animals

In vivo pathological collagen mineralization models
In the heart valve calcification model, human heart valves were obtained from volunteers with cardiac valvular calcification diseases, with informed consent and the authorization from The intramuscular ectopic calcification model was created in mice (n = 6). Mouse mesenchymal stem cells (ATCC) were seeded into 3D collagen scaffolds (ACE Surgical Supply Co., Brockton, MA, USA) under sterilized conditions. [47,48] The stem cell-containing scaffolds (5 mm x 5 mm) were implanted intramuscularly into separate pockets in 4-week-old C57BL/6J mice. The mice were euthanized after 8 weeks of calcification. Specimens were fixed and embedded in Tissue-Tek OCT. Eight micrometer-thick sections were prepared using a cryogenic microtome at -20 °C. The frozen sections were equilibrated in PBS at room temperature for 15 min to remove the "optimal cutting temperature" compound and fixed in 100% cold methanol (20 °C) for 10 min. The sections were washed in PBST (PBS + 0.01% Tween20) and blocked with PBST + 3% normal goat serum. Collagen-I rabbit anti-mouse primary antibodies (Ab34710, Proteintech) were applied to the sections overnight and incubated at 4 °C. Cy 3 goat anti-rabbit secondary antibodies (111-165-003, Jackson ImmunoResearch Laboratories), alizarin red S and SYTOX™ Green were used for fluorescence detection.
The Achilles tendon ectopic calcification model was created in rats. Four-week-old rats (n = 6) were anesthetized prior to Achilles tendon laceration surgery. The full length of each rat's Achilles tendon was exposed. A complete transverse incision was made five times and no suturing was performed to the mid-point of the tendon. The subcuticular skin was closed by sutures. [49] After 3 months, animals were sacrificed and the Achilles tendons were collected for subsequent analysis.

Cell culture
MC3T3-E1 murine calvarial osteoblasts (ATCC) were cultured in α-MEM that was supplemented with 10% FBS and penicillin/streptomycin in a humidified incubator with 5% CO 2 at 37 °C. When the cells reached 80% confluence, the growth medium was change to an osteogenic medium comprising 50 mg/mL ascorbic acid, 4 mM β-glycerophosphate and 10 nM dexamethasone. The osteogenic medium was changed every 3 days. For nucleic acid collection, the cells were seeded in T-75 flasks and collected after they were incubated for 14 days.

Total DNA purity and concentration
Ultraviolet spectroscopy was used to evaluate the purity and concentration of the nucleic acids. The DNA concentration was calculated using the Beer-Lambert law and the A260/A280 ratio was used to evaluate DNA purity. For DNA, an A260/280 ratio of 1.8 is generally accepted as "pure".

Dynamic light scattering and zeta potential
The hydrodynamic diameter and the zeta potential of the DNA, DNA-CaP and CaP were measured using a Zetasizer Nano ZS detector (Malvern Instruments, Worcestershire, UK). A 4 mW He-Ne laser (633 nm) and a proper measuring angle (173°) were used for dynamic light scattering. The Smoluchowski method was used for measurement of zeta potential and the testing temperature was set at 25 °C with a voltage of 50 V. Each sample was analyzed three times.

Scanning electron microscopy (SEM) of DNA-CaP
The surface morphology of DNA-CaP was visualized using SEM (JSM-6701F, JEOL, Tokyo, Japan) operated at 15 kV. Specimens were placed on a silicon wafer, air-dried and sputter-coated with gold.

Scanning transmission electron microscopy (STEM) of DNA-ACP
The DNA-ACP was observed by STEM (JEOL 2100F) operated at 200 kV. High-angle annular dark-field scanning (HAADF-STEM) was performed. Energy dispersive X-ray spectroscopy (EDS) elemental mapping and selected area electron diffraction were also conducted to identify the mineral distribution and the crystallinity of the minerals.

Molecular dynamics (MD) simulation of DNA-ACP
To examine the polyanionic capacity of DNA, three 21 bp-dsDNA were placed in a box (a = 12.8 nm, b = 13.2 nm, c = 13.7 nm, α = 90°, β = 90°, γ = 90°), in which 252 Ca 2+ , 384 Clions were placed. The centroid distance of DNA was 1 nm. The initial model of DNA was built by packmol and AMBER tleap, and described by GAFF2 force field and AMBER ff14SB force field. The Amber 18 molecular simulation packages were employed for performing all MD simulations with a simulation time step of 2 fs. Steepest descent and conjugate gradient algorithm energy minimization methods were introduced to remove bad contact in the initial structures using sander module of Amber 18. The entire system was minimized with a positional restraint of 10 kcal mol -1 Å -2 in atoms of DNA. This was followed by equilibration at 310 K for 50 psec simulation in constant number, pressure and temperature (NVT) mode and 50 psec in constant number, pressure and temperature (NPT) mode (T = 310 K and P = 1 atm). A positional restraint of 2 kcal mol -1 Å -2 in the main chain C atoms of DNA was used to relax its side chains.
A 120 nsec MD simulation was performed on the systems in NPT ensemble (T = 310 K and P = 1 atm) with periodic boundary conditions. To further investigate how DNA attract HPO 4 2-, the DNA, Cl -, Ca 2+ of the system were extracted and placed it into a box (a = 15.7 nm, b = 16.1 nm, c = 16.5 nm). This was followed by placing 252 HPO 4 2ions and 504 K + ions into the box.

2D single-layer self-assembled collagen fibrils
Collagen fibrils were self-assembled from a rat tail tendon-derived collagen/acetic acid stock solution (5 mg/mL) using the dialysis method. [50] The self-assembled collagen solution was dropped on a 400 mesh Au TEM grid and dried at room temperature. A 0.3 M EDC/0.06 M NHS solution was used to crosslink and stabilized the reassembled collagen fibrils for 4 h. The grids were rinsed with deionized water for 3 times and air-dried.

3D demineralized dentin sections
Human caries-free third molars were extracted from the patients who provided informed consent for the use of the teeth for research, with approval from the ethics committee of the

TEM/STEM of mineralized 2D and 3D collagen fibrils
Gold grids coated with a single layer of self-assembled collagen fibrils were immersed into

SEM of mineralized 3D collagen matrices
The surfaces of the mineralized collagen scaffolds were examined by SEM (JSM-6701F, JEOL). The specimens were dehydrated in an ascending ethanol series (50-100%), immersed in hexamethyldisilane and sputter-coated with gold prior to examination.

Cryo-EM and electron tomography of mineralized single-layer collagen
Reconstituted collagen fibrils deposited on Quantifoil Jena R2/2/gold grids with 2 μm hole size were immersed into DNA-ACP mineralization medium for designated time-periods. The grids were vitrified in the manner described in Section 3.6 without further staining. The vitrified grids were examined with a Talos F200C microscope (FEI, Hillsboro, OR, USA) at 100 kV under liquid nitrogen temperature.
For electron tomography, the selected grids were tilted in 2° steps from -50° to 50° using The University of California, San Francisco TOMO software package. [51] Alignment and 3D reconstruction of a mineralized collagen fibril were performed with the IMOD package. [52] Images were Gaussian-filtered to 10 Å. Segmentation of the 3D volume was performed with the Amira software version 2019.1 (Thermo Fisher Scientific) and Matlab version R2020a.

Attenuated total reflection-Fourier transform infrared spectroscopy (ATR-FTIR)
After the collagen fibrils were incubated with DNA for designated time-periods, the specimens were rinsed with deionized water and freeze-dried in a liquid nitrogen-cooled Emitech K775X turbo freeze dryer (Quorum Technologies Ltd, Kent, UK). The specimens were characterized using a Nicolet 6700 FTIR (Thermo Fisher Scientific) equipped with an ATR set-up. Infrared spectra were acquired over the region 4000-400 cm -1 with 32 scans and a resolution of 4 cm -1 .

Ruthenium red staining
Binding of DNA to collagen was examined using ruthenium red staining. TEM grids coated with collagen were floated on top of a DNA solution for 72 h. The grids were stained with 0.02 % ruthenium red, a cationic dye, to identify the collagen-bound DNA. After ruthenium red staining, the grids were counterstained with 2% uranyl acetate for 15 min prior to TEM examination.

Immunofluorescence
Slides coated with reconstituted type I collagen fibrils were labeled with collagen-I rabbit anti-mouse antibody (Ab34710, Proteintech) and Cy 3 goat anti-rabbit secondary antibodies After the collagen fibrils were mineralized with DNA-ACP for 5 days, they were stained with alizarin red S to label the calcium ions and SYTOX™ Green for labeling the DNA.
Collagen fibrils were visualized in the CLSM bright field mode to identify bound DNA on the surface of the mineralized collagen fibrils.

Molecular dynamics simulations
Collagen and DNA models were established to analyze the binding affinity of DNA to collagen fibrils.

Molecular modeling of collagen
A collagen molecular model was constructed based on the authors' previous work. [53] Briefly, full atomistic structures of collagen were constructed using homology modelling from human FASTA sequence. The

Molecular modeling of collagen-DNA
To investigate the interaction mechanism of collagen-DNA, 64 bp-ds DNA was placed into a unit cell (a = 24.2 nm, b = 5.66 nm, c = 67.79 nm, α = 90°, β = 90°, γ = 105.58°). The DNA was placed in the extrafibrillar regions (near the overlap zone) and 2.5 nm away from the collagen fibril. In addition, 876 particles of Na + and 610 particles of Clwere included to neutralize the system. The parameters of collagen protein, DNA and water molecules were assigned with the AMBER ff14SB force field and TIP3P water model. To obtain a reasonable initial model, the system was first minimized with a positional restraint of 40 kcal mol -1 Å -2 in both the DNA and collagen (5,000 steps SD method and 5000 steps CG method). Afterwards, the entire system was minimized with a positional restraint of 20 kcal mol -1 Å -2 in the backbone atoms of collage and the heavy atoms of DNA (5,000 SD method steps and 5,000 steps CG method). The entire system was then energy-minimized without any restraints (5,000 steps SD method, 5,000 steps CG method). After minimization, the system was equilibrated with 2.5 nsec MD simulation NVT and 2.5 nsec NPT. Molecular dynamics simulation was performed for 80 nsec on the systems in NPT ensemble (T = 310 K and P = 1 atm) with periodic boundary conditions.

The molecular mechanics-generalized Born surface area (MMGBSA) method
The principle of the Molecular mechanics-Poisson-Boltzmann surface area-weighted solvent-accessible surface area (MM-PBSA-WSAS) method is well reported in the literature. [54,55] The MM-PBSA binding free energy for a ligand binding to a receptor to form a complex is expressed as: where ΔE MM is the change of molecular mechanics (MM) energy due to complexation in gas-phase, ΔG sol is the change of solvation free energy, and -TΔS is the change of conformational entropy upon ligand binding. E MM is composed of several energy terms, including internal energies E int (bond, angle, and dihedral energies), electrostatic energy E ele and van der Waals interaction energy E VDW . ΔG sol may be further decomposed into two parts, the polar contribution (electrostatic solvation energy) which is described by the Poisson-Boltzmann continuum solvation model, and the nonpolar contribution which is described by solvent-accessible surface area (SASA). As such, the MM-PBSA binding free energy has five energy terms as shown below when the "Single Trajectory" sampling protocol [51,52] is applied: Note that in Eq.
(2), the contribution from internal energies E int cancel out in the "Single Trajectory" protocol. In the present work, the polar contribution of the solvation free energy is calculated by the GB1 model using sander of Amber. The nonpolar component is calculated using a linear relationship to SASA. That is, ΔG solv_np =  ΔSASA + b, where , the surface tension, is set to 0.00720 (kcal mol -1 Å -2 ) and b to 0.0 kcal/mol, and TΔS is the entropic contribution. The latter was neglected due to high computational costs and the lack of apparent differences with the same protein system.

Micro-computed tomography (micro-CT)
Mice at 3 weeks after implantation from each group were analyzed with a high-resolution micro-CT designed for small-animal imaging (Inveon micro-CT system Siemens AG, Munich, Germany). Each site (n = 6) was scanned at 60 kV and 148 μA. Three-dimensional reconstructions were performed using the SkyScan CtAn software (Micro Photonics, Inc., Allentown, PA, USA).

Immunofluorescence
At 3 weeks post-implantation, collagen scaffolds within the implantation sites were retrieved and sectioned at a thickness of ~10 μm using a saw microtome (SP1600, Leica, Mannheim, Germany) for immunofluorescence analysis (n = 3). The collagen fibrils, minerals and nucleic acids were co-labeled using the method described in the Section 5.3.

TEM
Additional harvested post-implantation collagen scaffolds were then fixed with 2.5% glutaraldehyde for 3 days at 4 °C. After rinsing with PBS, the specimens were fixed in 1% osmic acid, dehydrated with an ascending ethanol series (50-100%), immersed in propylene oxide and embedded in epoxy resin for TEM characterization.

SEM
Additional harvested post-implantation collagen scaffolds were examined by SEM using the method described in the Section 4.5.

Inhibition of collagen mineralization with DNA-ACP-DNase
TEM grids coated with self-assembled collagen were floated upside-down over the mineralization medium. The latter comprised 3.5 mM CaCl 2 .2H 2 O, 2.1 mM K 2 HPO 4 and 128 bp dsDNA (1 mg/mL). DNase I (0.73 mg/mL) was added to the mineralization medium at room temperature. At the designated time-period, the grids were retrieved, rinsed and air-dried.

TEM of collagen fibrils
TEM grids coated with a single layer of self-assembled collagen fibrils were immersed into DNA-ACP-DNase medium for a designated time-period. Upon retrieval, each grid was rinsed, dried and examined with the JEM-1230 TEM without osmication and further staining.

SI-11. Binding of DNA with collagen fibrils [Supplementary Figure 11]
Supplementary Figure 11. Binding of DNA with collagen fibrils. a, Infrared spectra of pristine collagen (black) and DNA-bound collagen (red). Peaks at 3,000-3,700 cm -1 were assigned to hydrogen bonds. The stretching vibration of amide I (1,650 cm -1 ) of the collagen fibrils was still present after binding of nucleic acids. This indicates that DNA-collagen still possess a triple helix structure. b, Infrared spectra of pristine collagen (black) and DNA-bound collagen (red). Peaks at 3,000 cm -1 -3,700 cm -1 were assigned to the hydrogen bond. The absorption band in this region broadened in DNA binding collagen samples, indicating the formation of hydrogen bond between nucleic acids and collagen fibrils.