Single-molecule DNA visualization using AT-specific red and non-specific green DNA-binding fluorescent proteins

Jihyun Park a, Seonghyun Lee a, Nabin Won a, Eunji Shin a, Soo-Hyun Kim b, Min-Young Chun a, Jungyeun Gu a, Gun-Young Jung c, Kwang-Il Lim *b and Kyubong Jo *a
aDepartment of Chemistry and Interdisciplinary Program of Integrated Biotech, Sogang University, 1 Shinsudong, Mapogu, Seoul, 04107, Korea. E-mail:; Fax: +82-2-701-0967; Tel: +82-2-705-8881
bDepartment of Chemical and Biological Engineering, Sookmyung Women's University, Cheongpa-ro 47-gil 100, Yongsangu, Seoul, 04310, Korea. E-mail:; Tel: +82-2-2077-7627
cSchool of Material Science and Engineering, GIST, Gwangju, 61005, Korea

Received 27th July 2018 , Accepted 7th October 2018

First published on 8th October 2018

The recent advances in the single cell genome analysis are generating a considerable amount of novel insights into complex biological systems. However, there are still technical challenges because each cell has a single copy of DNA to be amplified in most single cell genome analytical methods. In this paper, we present a novel approach to directly visualize a genomic map on a large DNA molecule instantly stained with red and green DNA-binding fluorescent proteins without DNA amplification. For this visualization, we constructed a few types of fluorescent protein-fused DNA-binding proteins: H-NS (histone-like nucleoid-structuring protein), DNA-binding domain of BRCA1 (breast cancer 1), high mobility group-1 (HMG), and lysine tryptophan (KW) repeat motif. Because H-NS and HMG preferentially bind A/T-rich regions, we combined A/T specific binder (H-NS-mCherry and HMG-mCherry as red color) and a non-specific complementary DNA binder (BRCA1-eGFP and 2(KW)2-eGFP repeat as green color) to produce a sequence-specific two-color DNA physical map for efficient optical identification of single DNA molecules.


Single-cell DNA analysis has advanced so rapidly that a number of bio-analytical methods have been recently developed.1,2 Most approaches for single-cell genomic analysis are based on DNA amplification and next-generation sequencing. On the other hand, it is notable that single-cell sequencing is usually resequencing rather than de novo sequencing. Thus, DNA identification without sequencing or amplification would be more efficient for single-cell genomic analysis. As a solution, single-molecule DNA identification, such as Optical Mapping and Nanocoding systems, can be powerful for single-cell genome analysis. These systems have been developed using sequence-specific digestion by restriction enzymes,3 and sequence-specific enzymatic nicking and fluorochrome labeling by DNA polymerase.4 These approaches have been commercialized to be utilized to overcome current limitations from short-read length sequencing technologies. Recently, several other approaches have been developed, such as temperature-controlled DNA melting in nanochannels,5 A/T specific netropsin binding to inhibit YOYO-1 binding,6 covalent linking of a fluorophore using sequence-specific methyl transferase,7 and A/T specific binding synthetic polypyrrole molecule linked with a fluorophore.8

Alternatively, we speculated that fluorescent protein (FP) linked sequence-specific DNA binding protein domains would have a potential to generate sequence-specific DNA physical maps to visualize the gene information stored in a large DNA molecule, although it has not been reported so far. There are a number of DNA binding domains that have various sequence specificity. For example, histone-like nucleoid-structural protein (H-NS-mCherry) and high mobility group (2HMG-mCherry) could preferentially bind A/T-rich regions.9,10 There is a report for the binding specificity of BRCA1 on TTC(G/T)GTTG.11 Among these numerous candidates, we can choose any of them depending on the property we would like to specify. Moreover, the use of FP-fused DNA binding domains would have advantages over the existing ones due to protein properties.12 One of the advantages is no necessity to use chemical dyes that directly interact with the DNA backbone. For example, an intercalating dye of YOYO-1 is known to deform, unwind, and often break double-stranded DNA molecules induced by photons.13,14 However, FP-fused DNA binding protein could obviate this issue since fluorophore in a fluorescent protein does not directly interact with DNA backbones. In addition, this approach does not need additional biochemical reactions by using non-covalent DNA–protein interactions and hydrogen bonding with DNA backbones.

In this study, we attempted to construct various FP-fused DNA binding proteins that have a proper level of sequence specificity. FP-fused histone-like nucleoid-structural protein (H-NS-mCherry) and high mobility group (2HMG-mCherry) could preferentially bind A/T-rich regions. On the contrary, BRCA1-eGFP and 2(KW)2-eGFP worked as complementary co-staining reagents since they bind DNA backbones homogenously. Using these fluorescent proteins, we obtained genome-specific DNA molecule images without any enzymatic treatments or conventional sequencing procedures at a single-molecule level.



DNA primers were purchased from Cosmogenetech (Seoul, Korea). Biotin labeled DNA oligomer were purchased from Bioneer (Daejeon, Korea). E. coli strain BL21 (DE3) was purchased from Yeastern (Taipei, Taiwan). λ DNA (48[thin space (1/6-em)]502 bp) and single-stranded M13mp18 (7249 bp) were purchased from New England Biolabs. N-Trimethoxymethyl silyl propyl-N,N,N-trimethyl ammonium chloride in 50% methanol was purchased from Gelest. Ni-NTA agarose resin and disposable empty gravity column were purchased from Qiagen. Epoxy was from Devcon (Riviera Beach, FL). Unless noted, all enzymes were purchased from NEB (Ipswich, MA), and all chemicals were from Sigma-Aldrich (St Louis, MI).


The microscopy system consisted of an inverted microscope (Olympus IX70) equipped with 60× and 100× Olympus UPlanSApo oil immersion objectives and illuminated LED light source (SOLA SM II light engine, Lumencor, Beaverton, OR). The light was passed through corresponding filter sets (Semrock, Rochester, NY), which were for excitation and emission of lights. Fluorescence images were captured by an electron-multiplying charge-coupled device digital camera (Evolve EMCCD, Photometrics, Tucson, AZ) and stored as 16-bit TIFF format generated by the software Micro-manager. ImageJ was utilized for image processing, particularly to overlap red and green images. ImageJ plugins and python programs were written to compare DNA image intensity and sequence frequency data by calculating cross-correlation values. Python program codes were reported previously.8

DNA-binding fluorescent protein sequence construct

The plasmids were constructed for H-NS-mCherry, BRCA1-eGFP, 2HMG-mCherry,15 2(KW)2-eGFP,16 and 2(KW)2-mCherry. Each plasmid was made by overlap extension polymerase chain reaction which links fluorescent protein to C-terminal of DNA binding protein. The sequence of linker between DNA binding protein and fluorescent protein was GGSGG. For construction of H-NS-mCherry, DNA plug of E. coli MG1655 strain was used as template of H-NS DNA sequence. All of PCR primers are described in ESI. H-NS DNA was amplified by extension polymerase chain reaction using forward primer P1-HNS and reverse primer P2-HNS. The gene of mCherry was amplified with another pair of primer: forward primer P3-mCherry and reverse primer P4-mCherry. Then, H-NS was linked with mCherry by overlap polymerase chain reaction. BRCA1-eGFP was made by cloning partial BRCA1 (Addgene plasmid #71116) including 452–1079 residues linked with eGFP. DNA binding domain of BRCA1 was amplified with forward primer P5-BRCA and reverse primer P6-BRCA, and eGFP DNA was amplified with another pair of primer: forward primer P7-eGFP and reverse primer P8-eGFP. DNA binding domain of BRCA1 was linked with eGFP by overlap polymerase chain reaction. In case of 2HMG-mCherry, for tagging DNA binding parts to each terminal of mCherry, forward primer P9-HMG-mCherry and reverse primer P10-HMG-mCherry were used. The amino acid sequences for DNA binding fluorescent proteins are described in ESI.

Molecular cloning

Using standard subcloning procedures, FP-fused DNA binding protein sequences were inserted into the pET-15b vector and transformed into the E. coli BL21 (DE3) strains by using NdeI and BamHI. A single colony of the transformed cells was inoculated in a fresh LB media containing ampicillin and incubated for 1 h. After transformed cells were saturated, subsequently cultured to an optical density of ∼0.8 at 37 °C with corresponding antibiotics. Fluorescent tagging proteins over-expression was induced with a final concentration of 1 mM for IPTG overnight on a shaker at 20 °C and 250 rpm. Cells for the protein purification were harvested by centrifugation at 12[thin space (1/6-em)]000g, for 10 min (following centrifugations were performed under similar conditions), and the residual media was washed using the cell lysis buffer (50 mM Na2HPO4, 300 mM NaCl, 10 mM Imidazole, pH 8.0). The cells were lysed by ultrasonication for 30 min and cell debris were centrifuged at 13[thin space (1/6-em)]000 rpm for 10 min at 4 °C. His-tagged FP-DNA binding proteins were purified using affinity chromatography with Ni-NTA agarose resin. The mixture of cell proteins and the resin were kept on a shaking platform at 4 °C for 1 h. The lysate containing proteins bound Ni-NTA agarose resin were loaded onto the column for gravity chromatography and was further rinsed several times using the protein washing buffer (50 mM Na2HPO4, 300 mM NaCl, 20 mM Imidazole, pH 8.0) several times. Especially for H-NS-mCherry, washing buffer with 35 mM Imidazole (50 mM Na2HPO4, 300 mM NaCl, 35 mM Imidazole, pH 8.0) was used. Finally, the bound proteins were eluted using the protein elution buffer (50 mM Na2HPO4, 300 mM NaCl, 250 mM imidazole, pH 8.0). All proteins were diluted (10 μg mL−1) using 50% w/w glycerol/1× TE buffer (Tris 10 mM, EDTA 1 mM, pH 8.0).

Preparation of viral genomic double-stranded DNA

Double-stranded M13mp18 was synthesized from commercially obtained single-stranded circular DNA by Top polymerase (Bioneer, Daejeon, Korea) reaction with a primer (GGAAACCGA GGAAACGCAATAATAACGGAATACCC). Circular ds DNA was linearized by PstI. Retroviral genomic DNA was prepared from murine leukemia virus genome as previously described.17 Circular dsDNA was linearized by BmtI.

Polydimethylsiloxane (PDMS) microfluidic devices

A standard rapid prototyping method was used to make PDMS microfluidic devices for DNA elongation and deposition on a positively charged surface.18 Briefly, the patterns on a silicon wafer for microchannels (4 μm high and 100 μm wide) were fabricated using SU-8 2005 photoresist (Microchem, Newton, MA). The PDMS pre-polymer mixed with curing agent (10[thin space (1/6-em)]:[thin space (1/6-em)]1 weight ratio) was cast on the patterned wafer and cured at 65 °C for four hours or longer. Cured PDMS was peeled off from the patterned wafer, and the PDMS devices were treated in an air plasma generator for 1 min with 100 W (Femto Science Cute Basic, Korea) to make PDMS surface hydrophilic. PDMS devices were stored in water and air-dried before use.

Glass surface preparation

Glass coverslips were stacked in the Teflon rack, and soaked in piranha etching solution (30[thin space (1/6-em)]:[thin space (1/6-em)]70 v/v H2O2/H2SO4) for 2 hours and rinsed with deionized water until the pH reached the neutral (pH 7). For positively-charged glass surfaces, 350 μL of N-trimethoxymethylsilylpropyl-N,N,N-trimethyl ammonium chloride in 50% methanol was mixed with 200 mL of deionized water. The acid cleaned glass coverslips were incubated in this solution for 12 hours. To prepare glass surface for DNA tethering, 2 mL of N-[3-(trimethoxysilyl)propyl] ethylenediamine was added to 200 mL of methanol and 10 mL of glacial acetic acid to add primary amino groups. The glass coverslips were incubated in the solution for thirty minutes, sonicated for fifteen minutes, and incubated for twelve hours at room temperature. Then, they were rinsed with methanol and ethanol.

DNA imaging on a positively charged surface

For imaging DNA stained with fluorescent proteins, DNA molecules (15 ng μL−1) were mixed with FP-fused DNA binding proteins (5–100 nM) in 1× TE (10 mM Tris, 1 mM EDTA, pH 8). For competitive binding assay, two-color proteins were mixed at the ratio ranged from 1[thin space (1/6-em)]:[thin space (1/6-em)]5–1[thin space (1/6-em)]:[thin space (1/6-em)]20 before staining DNA. DNA molecules (15 ng μL−1) stained with protein mixture (100 nM) were incubated at room temperature for a while. After staining, DNA solution was loaded at the entrance of a PDMS microchannel (100 μm × 4 μm). Right after, DNA molecules were imaged on the microscope.

Flow chamber

Flow chamber preparation followed a previous report.19 Briefly, it was prepared by placing the acrylic holder on a coverslip with a spacing 100 μm between the two double-sided tapes. A yellow pipette tip was placed in the inlet port. A tubing line was fixed to the outlet port. 80 mg of mPEG-succinimidyl valerate and 1–2 mg of biotin-PEG-succinimidyl carbonate were added in freshly made 0.1 M sodium bicarbonate. After mixing and centrifuging of the solution, 50 mL of PEG solution was dropped on a clean slide glass, covered with an aminosilanized coverslip for three hours to overnight, and rinsed with water. A syringe pump was used to control the buffer delivery into the flow cell with a flow rate of 50 μL min−1 that corresponded to 2.8 mm s−1. After preparation of flow chamber, 25 mg mL−1 of NeutrAvidin in the solution of 10 mM Tris, 50 mM NaCl, pH 8.0 and 1 μM DNA overhang oligo (5′-pGGGCGGCGACCT-TEG-biotin-3′) was sequentially loaded into the flow chamber and kept for 5 minutes. The λ DNA, T4 DNA ligase, and the reaction buffer were added and incubated for 30 minutes. After washing the residual enzyme mixture, DNA binding fluorescent proteins and salty solution were flowed into the flow chamber step by step for visualization of tethered DNA.

Results and discussion

Two-color λ DNA image on the positively charged surface

Fig. 1 demonstrates λ DNA staining by combining H-NS-mCherry and BRCA1-eGFP. Consistent image patterns allowed us to align λ DNA molecules based on three distinct red spots on a green DNA backbone (Fig. 1a). FP-fused histone-like nucleoid-structural protein (H-NS-mCherry) could preferentially bind A/T-rich regions as previously reported.9,10 On the contrary, BRCA1-eGFP worked as complementary co-staining reagents since they bind DNA backbones homogenously. Although there is a report that supports the binding specificity of BRCA1 on TTC(G/T)GTTG,11 we did not observe sequence-specific binding. Therefore, we used BRCA1-eGFP as a non-specific complementary DNA binder. Fig. 1b presents a typical microscopic image in which multiple elongated λ DNA molecules were randomly deposited on a positively charged surface. Interestingly, there are fragmented DNA molecules in the middle of the image that can be interpreted based on their colored pattern. For more quantitative analysis, we converted the DNA images into the intensity profiles for red and green colors along the DNA molecules. As a comparison, we calculated A/T frequency profile of the λ genome. For this profile construction, we partitioned the whole length of the λ genome (48[thin space (1/6-em)]502 bp) into 53 domains based on the average pixel length (53 pixels) for λ DNA molecules. The characteristic of λ DNA molecule is that the entire length consists of both A/T poor region and A/T-rich region. A/T poor region covers 0–20 kb with relatively low A/T frequency about 40%. A/T rich region includes three peaks of high A/T frequency around 60% as shown in Fig. 1c. The comparison in Fig. 1c shows that the profile of red fusion proteins with AT-rich sequence preference is well matched with this in silico profile, though the experimental result does not seem to reflect minute A/T frequency variations. This profile comparison allows us to interpret the details of λ DNA images: for example, the first green-colored region corresponds to the domain of 0–20 kb, and three red spots are localized from 20 kb to 48.5 kb. Likewise, it is straightforward to interpret DNA patterns to determine the orientation of the DNA molecules and to extract detailed genomic information from the visualized DNA molecules. This profile comparison further allowed us to interpret two DNA fragments in Fig. 1b. DNA fragment (i) contains the green domain and a red peak, which corresponds to the 0–30 kb region of λ DNA. DNA fragment (ii) corresponds to the 20–40 kb region of λ DNA.
image file: c8an01426d-f1.tif
Fig. 1 Two-color FP-stained λ DNA molecules. (a) Aligned λ DNA molecules stained with H-NS-mCherry and BRCA1-eGFP. (b) Microscopic image of two-color stained λ DNA on a positively charged glass surface. The arrows indicate molecular orientations from 5′ to 3′, and the white profile represents the A/T frequency of λ DNA (scale bar: 10 μm). Annotated cross-correlation values (cc) are from individual molecules. (c) Comparison of relative intensity profiles for the red and green fusion fluorescent proteins and in silico A/T frequency profile (black). (d) Cross-correlation value between stained ten molecules in (a) and in silico A/T frequency map (0.87 ± 0.05). As a control, the DNA image profile was compared to randomly generated sequences (0.55 ± 0.14, ***p < 0.0001).

We calculated Pearson cross-correlation coefficient (cc) between in silico A/T frequency and stained DNA molecules, using lab-made Python program.8 Fluorescence intensities from H-NS-mCherry showed high average cc value of 0.87 ± 0.05 (Fig. 1d). On the contrary, we also generated randomly generated molecule images as control, which provided the cc value of 0.55 ± 0.14 between experimental data and control image. This value is significantly lower than the in silico λ DNA A/T frequency map. Therefore, we could identify individual λ DNA molecules by not only recognizing distinct patterns on the microscope but also calculating the cc and statistical p values from DNA images.

Optimal concentration for two DNA-binding proteins

As shown in Fig. 1a, distinctive color patterns were thought to be the result of a proper combination of red and green protein concentrations. Therefore, we further characterized the dependence of DNA color pattern on the relative amounts of the two staining FPs by varying their concentration ratios (Fig. 2). It was intriguing that either H-NS-mCherry or BRCA1-eGFP could stain DNA entirely and homogenously at concentrations significantly higher than that of the other FP. For example, 20 nM BRCA1-eGFP with 1 nM H-NS-mCherry generated a full green-colored DNA molecule probably due to BRCA1-eGFP's dominant staining. In contrast, the combination of 8 nM H-NS-mCherry and 10 nM BRCA-eGFP generated entirely red-colored DNA molecule. The combination of 16 nM H-NS and 20 nM BRCA produced an almost red-colored DNA molecule. Notably, these data imply that H-NS-mCherry seems to have a stronger affinity for the DNA molecules than BRCA1-eGFP. Staining DNA molecules only with H-NS-mCherry could generate a sequence-specific color pattern at 4 nM or lower. However, molecular images look like scattered spots probably because the fluorescence intensity was too low from 0 to 20 kb of the λ genome, which has A/T frequency around 40% according to Fig. 1c. The excess or insufficient A/T specific proteins did not stain DNA molecules sequence-specifically, resulting in decreased cc value quantitatively. The calculation of cc values indicated the optimal protein concentration ratio of 4 nM H-NS-mCherry and 20 nM BRCA1-eGFP when cc equals 0.91. Therefore, H-NS-mCherry stains A/T-rich regions and BRCA1-eGFP complementarily stains the parts of DNA not stained by H-NS-mCherry.
image file: c8an01426d-f2.tif
Fig. 2 Concentration combination between H-NS-mCherry and BRCA1-eGFP. The number represents cc value at each condition with the in silico map of λ DNA as shown in Fig. 1c.

Combinatorial staining of various DNA-binding proteins

Fig. 3a illustrates typical microscopic images for λ DNA stained with various combinations of A/T specific mCherry and complementary eGFP. Four out of six combinations produced A/T specific λ genome patterns, which were shown in Fig. 3b as aligned λ DNA molecules, whose cc values range from 0.84 to 0.87 for four A/T specific patterns (Fig. 3c). Previously, we attempted 2HMG-eGFP to generate A/T specific map,12 because HMG was known to have A/T specificity.10 Unfortunately, we had only observed their homogenous DNA-staining in our previous report.12 However, Fig. 2 suggested that the previous homogeneous staining might have resulted from an unoptimized concentration of 2HMG-eGFP. 2HMG-eGFP could also generate A/T specific pattern at an appropriate concentration. Thus, we attempted to optimize 2HMG concentration. In this context, we constructed 2HMG-mCherry for two-color staining. The combination of 2HMG-mCherry with BRCA1-eGFP or 2(KW)2-eGFP successfully generated the same λ genome specific color pattern to present A/T-rich regions (iii and iv) to have cc values of 0.84 and 0.86 respectively. As a contrary, two A/T specific DNA binding proteins, H-NS-mCherry and 2HMG-eGFP generated random patterns (cc = 0.61), and two non-specific DNA binding proteins also generated random patterns (cc = 0.59). These cc values were within the range of random sequences (cc = 0.55 ± 0.14). Accordingly, the combinations of A/T specific protein and non-specific protein can be a general tool to visualize A/T specific genome patterns.
image file: c8an01426d-f3.tif
Fig. 3 Various combinations of two-color FP staining. (a) Two-color stained λ DNA on positively charged surfaces (i) H-NS-mCherry & BRCA1-eGFP, (ii) H-NS-mCherry & 2(KW)2-eGFP, (iii) 2HMG-mCherry & BRCA1-eGFP, (iv) 2HMG-mCherry & 2(KW)2-eGFP (v) H-NS-mCherry & 2HMG-eGFP (vi) 2(KW)2-mCherry & BRCA1-eGFP (b) aligned λ DNA images for (i)–(iv). (c) Cross-correlation values: (control) 0.55 ± 0.14, (i) 0.84 ± 0.10, (ii) 0.87 ± 0.05, (iii) 0.84 ± 0.04, (iv) 0.86 ± 0.03 (v) 0.61 ± 0.15, and (vi) 0.59 ± 0.11. (*p < 0.02, **p < 0.005 for randomly generated sequences t-test). Each data represents the average of five measurements in (b). Scale bar: 10 μm.

Two-color staining to identify short DNA fragments

This approach can be extended to analyze other short DNA fragments. As an example, we visualized small DNA molecules obtained from two viruses, M13, a bacteriophage that infects bacteria, and murine leukemia virus (MLV), a retrovirus that infects mice.20,21 Because the M13 phage genome is a single-stranded DNA molecule that would make microscopic observation very challenging, we used DNA polymerase reaction to make it double-stranded to be readily visible on the fluorescence microscopic image. This double-stranded circular genomic DNAs were then linearized by single-cutting restriction enzyme digestions for more effective image analysis. As shown in Fig. 4, the linearized DNA molecules were two-color stained to generate genome-specific patterns. The M13 DNA images in Fig. 4a matched relatively well with A/T frequency in silico profile (the red graph in Fig. 4c), in which there are high A/T regions at both ends of the genomic DNA. Because this DNA is only 7 kb of 2.5 μm long with two symmetric red peaks and a short green internal region, it was not simple to determine the molecular orientations. More specifically, the average cc value for the forward is 0.70 ± 0.09, and that for the backward is 0.64 ± 0.16, which is only 8% lower. The MLV DNA molecules that we constructed as described in a previous report17 also exhibited two symmetrically distributed red spots with a longer green internal region (Fig. 4b and the blue graph in Fig. 4c). This DNA molecule is 8.5 kb long and more greenish due to low A/T concentration. In this case, the cc value for the forward is 0.69 ± 0.08 and the backward has 0.57 ± 0.08 (cc). Thus, it is also challenging to identify molecular orientations. However, it is clear to differentiate the M13 DNA from the MLV DNA based on the image patterns as well as molecular lengths (Fig. 4).
image file: c8an01426d-f4.tif
Fig. 4 Identification of viral genomic DNA by staining with H-NS-mCherry and BRCA1-eGFP. (a) Linear double-stranded M13mp18 (7249 bp) obtained from digestion of the circular M13mp18 DNA with Pst I. (b) Linearized double-stranded complementary DNA for the genomic parts of a retrovirus (via restriction with BmtI), murine leukemia virus (MLV, 8533 bp) (scale bar: 5 μm). (c) The magenta and cyan graphs show in silico A/T frequency profile for (a) and (b) respectively. (d) Comparison of cc for six images in (a) and (b) respectively with their corresponding in silico A/T frequency profiles. The controls are based on randomly generated sequences. The values were 0.58 ± 0.16, 0.70 ± 0.09, 0.53 ± 0.11, and 0.69 ± 0.05, from left to right, respectively (*p < 0.03, **p < 0.002 for randomly generated sequences t-test).

As demonstrated so far, two-color staining works well with λ DNA, M13mp18, and MLV viral DNA. However, we found a limitation of applying this approach to high A/T content DNA molecules, such as human genomic DNA and bacteriophage T4 DNA. They have too high A/T frequency to identify them using our program in most regions. Thus, microscopic images showed entirely stained red-color DNA molecules (ESI Fig. 1). Accordingly, it is necessary to develop more efficient DNA binding moieties to apply this approach for T4 DNA as well as the human genome.

Two-color staining for tethered DNA molecules

We applied two-color FP staining to surface tethered DNA molecules.19 One of the noticeable advantages of surface tethering is that DNA molecules could be stained with each type of FP, color by color because it is flexible to change proteins, buffers, and other chemical environments. Thus, as schematically illustrated in Fig. 5a, DNA backbones were entirely and homogeneously stained with excess H-NS-mCherry and then washed with a high salt concentration buffer solution (73 mM of NaCl). Since FP-staining is reversible, staining and destaining can be done merely by changing salt conditions.12 The next flowing solution was chosen to contain eGFP-tagged DNA-binding proteins to complementarily stain DNA molecules. In this way, we were able to obtain DNA images that matched with those on the positively charged surface. As shown in Fig. 5b, tethered DNA molecules were stained with red and green FPs to visualize their unique DNA patterns, which have cc values ranging from 0.82 to 0.93. Because we used a 3′ cos site overhang of λ DNA for tethering, the binding pattern of H-NS-mCherry went from 3′ to 5′, opposite to the orientations of DNA molecules shown in Fig. 1 and 2.
image file: c8an01426d-f5.tif
Fig. 5 Application of two-color staining to other single-molecule DNA presentation platforms. (a) Schematic illustration of multi-step staining for tethered DNA molecules (b) tethered λ DNA molecules visualized with two colors. The numbers are cc values. The flow rate was 50 μL min−1 (scale bar: 5 μm).


Here, we demonstrate sequence-specific visualization of single-molecule DNA using the combination of A/T specific red and complementary green fluorescent proteins. Since this approach utilized single DNA molecules, it would have a high potential to analyze single cellular DNA molecules that have only one copy of a long DNA molecule per cell. This two-color DNA staining quickly and effectively produces characteristic color patterns for different DNA molecules, allowing for DNA identification. Moreover, we found that excess amount of FP-fused DNA binding proteins would cover DNA molecules entirely and homogeneously, but some genome-specific patterns might appear at the optimal protein concentrations or by increased salt concentrations. These results propose that it would be very intriguing and probably useful to stain large elongated DNA molecules with a variety of DNA binding proteins linked with FPs. Accordingly, DNA staining with a novel FP-fused DNA binding protein can be a simple and straightforward analytical approach to visualize information that a single DNA molecule would contain.

Conflicts of interest

There are no conflicts to declare.


This work is supported grants from the National Research Foundation of Korea (NRF, 2017R1A2B2012665, 2017R1E1A2A02080741, 2016M3A9B6947831 and 2016R1A6A1A03012845).

Notes and references

  1. C. Gawad, W. Koh and S. R. Quake, Nat. Rev. Genet., 2016, 17, 175–188 CrossRef CAS PubMed.
  2. T. W. Murphy, Q. Zhang, L. B. Naler, S. Ma and C. Lu, Analyst, 2018, 143, 60–80 RSC.
  3. D. Schwartz, X. Li, L. Hernandez, S. Ramnarain, E. Huff and Y. Wang, Science, 1993, 262, 110–114 CrossRef CAS PubMed.
  4. K. Jo, D. M. Dhingra, T. Odijk, J. J. de Pablo, M. D. Graham, R. Runnheim, D. Forrest and D. C. Schwartz, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 2673 CrossRef CAS PubMed.
  5. W. Reisner, N. B. Larsen, A. Silahtaroglu, A. Kristensen, N. Tommerup, J. O. Tegenfeldt and H. Flyvbjerg, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 13294 CrossRef CAS PubMed.
  6. L. K. Nyberg, F. Persson, J. Berg, J. Bergstrom, E. Fransson, L. Olsson, M. Persson, A. Stalnacke, J. Wigenius, J. O. Tegenfeldt and F. Westerlund, Biochem. Biophys. Res. Commun., 2012, 417, 404–408 CrossRef CAS PubMed.
  7. J. Jeffet, A. Kobo, T. Su, A. Grunwald, O. Green, A. N. Nilsson, E. Eisenberg, T. Ambjörnsson, F. Westerlund, E. Weinhold, D. Shabat, P. K. Purohit and Y. Ebenstein, ACS Nano, 2016, 10, 9823–9830 CrossRef CAS PubMed.
  8. S. Lee, Y. Kawamoto, V. Thangavel, J. Park, J. Bae, J. Kim-Ha, H. Sugiyama and J. Kyubong, Nucleic Acids Res., 2018 DOI:10.1093/nar/gky531.
  9. B. Lang, N. Blot, E. Bouffartigues, M. Buckle, M. Geertz, C. O. Gualerzi, R. Mavathur, G. Muskhelishvili, C. L. Pon, S. Rimsky, S. Stella, M. M. Babu and A. Travers, Nucleic Acids Res., 2007, 35, 6330–6337 CrossRef CAS PubMed.
  10. R. Reeves and M. S. Nissen, J. Biol. Chem., 1990, 265, 8573–8582 CAS.
  11. P. L. Cable, C. A. Wilson, F. J. Calzone, F. J. Rauscher 3rd, R. Scully, D. M. Livingston, L. Li, C. B. Blackwell, P. A. Futreal and C. A. Afshari, Mol. Carcinog., 2003, 38, 85–96 CrossRef CAS PubMed.
  12. S. Lee, C. Wang, J. Song, D. G. Kim, Y. Oh, W. Ko, J. Lee, J. Park, H. S. Lee and K. Jo, RSC Adv., 2016, 6, 46291–46298 RSC.
  13. M. A. Tycon, C. F. Dial, K. Faison, W. Melvin and C. J. Fecko, Anal. Biochem., 2012, 426, 13–21 CrossRef CAS PubMed.
  14. C. U. Murade, V. Subramaniam, C. Otto and M. L. Bennink, Nucleic Acids Res., 2010, 38, 3423–3431 CrossRef CAS PubMed.
  15. S. Lee, C. Wang, J. Song, D. G. Kim, Y. Oh, W. Ko, J. Lee, J. Park, H. S. Lee and K. Jo, RSC Adv., 2016, 6, 46291–46298 RSC.
  16. S. Lee, Y. Oh, J. Lee, S. Choe, S. Lim, H. S. Lee, K. Jo and D. C. Schwartz, Nucleic Acids Res., 2016, 44, e6 CrossRef PubMed.
  17. K. I. Lim, R. Klimczak, J. H. Yu and D. V. Schaffer, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 12475–12480 CrossRef CAS PubMed.
  18. E. T. Dimalanta, A. Lim, R. Runnheim, C. Lamers, C. Churas, D. K. Forrest, J. J. de Pablo, M. D. Graham, S. N. Coppersmith, S. Goldstein and D. C. Schwartz, Anal. Chem., 2004, 76, 5293–5301 CrossRef CAS PubMed.
  19. S. Lee and K. Jo, J. Visualized Exp., 2016, e54141 Search PubMed.
  20. P. M. van Wezenbeek, T. J. Hulsebos and J. G. Schoenmakers, Gene, 1980, 11, 129–148 CrossRef CAS PubMed.
  21. S. Perryman, J. Nishio and B. Chesebro, Nucleic Acids Res., 1991, 19, 6950 CrossRef CAS PubMed.


Electronic supplementary information (ESI) available: Supplementary figure, full sequences of fluorescent proteins. See DOI: 10.1039/c8an01426d
These authors equally contributed to this work.

This journal is © The Royal Society of Chemistry 2019