2.1. Preparation of DNA-Beads

Beads with double-stranded DNA bound to their surface were prepared using the same method as previously reported ⁵. The pGADT7 plasmid (Clontech Laboratories, Mountain View, CA, USA; #K1612-1; 8 kbp) was amplified in Escherichia coli and purified using a plasmid purification kit. The plasmid was then linearized by restriction enzyme digestion, and biotinylation was performed at only one end of the linear DNA. Phosphorothioate-modified oligonucleotides were introduced at both ends to protect DNA from exonuclease-mediated degradation. The resulting biotinylated DNA was mixed with streptavidin-conjugated polystyrene beads (Dynabeads M-270; Invitrogen, Carlsbad, CA, USA; #65305; 2.8 μm in diameter) and incubated at room temperature for 3 h, resulting in the formation of DNA-beads in which one end of the DNA was immobilized on the bead surface through the strong biotin–streptavidin interaction. Unbound biotinylated DNA was removed by washing, and the beads were finally suspended in distilled water at a concentration of \(5 \times 10^7\) beads/mL. DNA binding to the beads was performed using the Dynabeads Kilobase Binder Kit (Invitrogen; #60101). Approximately 20 μg of biotinylated dsDNA was immobilized onto 50 μg of the beads (\(\sim\)75 pmoles of dsDNA per milligram of beads) as described previously ⁵.

2.2. Plasmid Construction

Plasmid DNA encoding cGAS-mRFP was prepared as follows. At first, to construct the PBEF1-MCS-mRFP-IRES-neo vector, the DNA fragment encoding mRFP was amplified from the plasmid encoding mRFP-BAF ⁵ using PCR by using the following primers: mRFP FWD InF PBEF1 Bam (5\('\)-ATTCGAATTTAAATCGGGATCCATGGCCTCCTCC­GAGGACGT-3\('\)) and mRFP REV InF PBEF1 Bam (5\('\)-CTCAGCGGCCGCGGATC­TTAGGCGCCGGTGGAG­TGGC-3\('\)). The PCR product was digested with BamHI, and the fragment was inserted into the PB-EF1-MCS-IRES-neo cDNA expression vector (System Biosciences, Palo Alto, CA, USA; PB533A-2) after digestion with \({Bam}\)HI to obtain the PB-EF1-MCS-mRFP-IRES-neo vector.

Next, to construct the PBEF1-MCS-mRFP-zeo vector, a fragment corresponding to nucleotides 2,388–2,545 of the PBEF1-MCS-IRES-neo vector was amplified by PCR using the following primers: PBEF1 Kpn FWD (5\('\)-CTGAAGGA­TGC­CCA­GAA­GGT­CCC­CAT­TGT-3\('\)) and PBEF1 KOZAK REV (5\('\)-TCCGGACG­CCA­TGG­TTG­TGG-3\('\)). In addition, the zeocin resistance gene was amplified from the pcDNA3.1/zeo vector using the following primers: Zeocin FWD for PBEF1 KOZAK (5\('\)-ACAACCATG­GCG­TCC­GGA­ATG­GCC­AAG­TTG­ACC­AGT­GCC­GTTCC-3\('\)) and Zeocin REV for PBEF1 SalI (5\('\)-TCCAGAGGTTGATTGT­CGA­CTC­AGT­CCT­GCT­CCT­CGGC­CAC­GAA-3\('\)). The PBEF1-MCS-mRFP-IRES-neo vector was digested with \({Kpn}\)I and \({Sal}\)I to generate a linearized backbone fragment. The three fragments were assembled using the In-Fusion® HD Cloning Kit (Clontech; #639649) according to the manufacturer’s instructions, resulting in the PBEF1-MCS-mRFP-zeo vector.

The coding sequence of cGAS was PCR-amplified from a cDNA pool derived from HeLa cells using the following primers: hcGAS FWD RI InF PB (5\('\)-TCTAGAGCTAGCG­AAT­TCG­ATG­CAG­CCT­TGG­CAC­GGA­AA-3\('\); reverse primer) and hcGAS REV RI InF PB (5\('\)-CGATTTAAATTC­GAA­TTC­TCA­AAATTC­ATC­AAA­AAC­TG-3\('\)).

The fragment was subcloned into the PBEF1-EGFP-MCS-neo vector ¹¹. The coding sequence was then amplified using the hcGAS FWD RI InF primer and the hcGAS REV2 N RI InF PB primer (5\('\)-CGATTTAAATTCG­AAT­TCA­AAT­TCA­TCA­AAA­ACT­G-3\('\)), in which the stop codon (TCA) was omitted, and inserted into the pBEF1-MCS-mRFP-zeo vector to generate the PBEF1-cGAS-mRFP-zeo plasmid.

DNA constructs expressing cGAS-GFP and mRFP-BAF were generated by swapping the GFP and mRFP tags between the original GFP-BAF and cGAS-mRFP constructs, using the cloning strategy described above.

2.3. Cell Culture

HeLa cells were obtained from RIKEN Cell Bank (Tsukuba, Japan) and maintained in DMEM containing 10% (v/v) calf serum. HeLa cells expressing both GFP-BAF and cGAS-mRFP (HeLa/GFP-BAF/cGAS-mRFP) were obtained as follows: HeLa cells were first transfected with plasmid DNA encoding cGAS-mRFP (PBEF1-cGAS-mRFP-zeo) and incubated in the presence of Zeocin (Life Technologies, Carlsbad, CA, USA; #R250-01) to obtain HeLa/cGAS-mRFP cells. The HeLa/cGAS-mRFP cells were then transfected with a plasmid encoding GFP-BAF ¹² under the EF1\(\alpha\) promoter and maintained in the presence of Zeocin (200 μg/mL) and Geneticin (200 μg/mL).

2.4. Bead Introduction into Living Cells

DNA-beads were introduced into HeLa cells using the Effectene-mediated method using the “enhancer reagent” equipped with the Effectene transfection reagent kit (Qiagen, Hilden, Germany; #301425) ⁵. One day before bead introduction, cells were seeded onto 35-mm glass-bottom culture dishes (MatTek) at a concentration of \(2 \times 10^{5}\) cells per dish. \(5 \times 10^{4}\) of the beads per dish were used in each experiment. After incubation at 37°C for 1 h, the cells were washed with culture medium to remove unattached beads, and then the cells were pre-stained with Hoechst33342 for DNA visualization prior to observation.

2.5. Indirect Immunofluorescence Staining

Indirect immunofluorescence staining was performed as previously described ⁴. Cells cultured in 35-mm glass-bottom culture dishes were fixed with 4% formaldehyde in PBS (pH 7.4) for 15 min. After washing with PBS, the cells were permeabilized with 0.1% Triton X-100 in PBS for 5 min, treated with 2% BSA in PBS for 1 h at room temperature for blocking, and then stained with specific antibodies against BAF and cGAS in 1% BSA/PBS for 20 h at 4°C. The primary antibodies used were anti-BANF1 mouse monoclonal antibody (Abnova, Taipei, Taiwan; #H00008815-M01) and anti-cGAS rabbit monoclonal antibody (Cell Signaling Technology, Danvers, MA, USA; #15102). Secondary antibodies labeled with Alexa Fluor 488 and 594 were purchased from Invitrogen (#A11029 and #A11037, respectively). After washing thrice with PBS, the cells were counterstained with 0.5 μg/mL DAPI for 15 min at room temperature. Fluorescence images were obtained using an Olympus oil-immersion objective lens PLAPON60xOSC (NA \(= 1.40\)) and a DeltaVision microscope.

2.6. Live-Cell Fluorescence Imaging and FRAP Analysis

DMEM (without phenol red) supplemented with 20 mM HEPES, 10% FBS, and 80 μg/mL kanamycin was used as the observation medium for live-cell imaging (1.5 mL/dish). A layer of mineral oil was overlaid on the medium to prevent the evaporation of the culture medium during observation. Time-lapse images were obtained through an Olympus oil-immersion objective lens UApo/340 (40\(\times\), NA \(= 1.35\)) on the DeltaVision microscope system (Applied Precision, Issaquah, WA, USA) placed in a temperature-controlled room (37°C).

Fluorescence recovery after photobleaching (FRAP) analysis was performed using a confocal laser-scanning microscope (LSM 880; Zeiss). A C-Apochromat 40\(\times\)/1.2 W Corr UV-VIS-IR water-immersion objective (Zeiss) was used.

DNA-beads positive for both GFP-BAF and cGAS-mRFP were photobleached using 488- and 561-nm lasers. Three pre-bleach images were acquired, and fluorescence recovery around the beads was monitored every 10 s. FRAP analysis was performed on DNA-beads that had escaped from the endosomes and were classified according to the elapsed time after escape as \(<10\) min (\(n = 4\)) or \(\ge 10\) min (\(n = 16\)). The fluorescence recovery curves were analyzed as described below.

For image analysis, the mean fluorescence intensity of the bleached region of interest (ROI) was measured at each time point and corrected for background fluorescence. The background-corrected fluorescence intensity, \({I}_\mathrm{corr}(t)\), was calculated as follows:

The fluorescence recovery curves were normalized by setting the average background-corrected fluorescence intensity of the three pre-bleach frames (\({I}_\mathrm{pre}\)) to 100% and the background-corrected fluorescence intensity immediately after photobleaching (\({I}_\mathrm{bleach}\)) to 0%. The normalized recovery fraction \({R_{f}}(t)\) was calculated as follows:

Normalized recovery curves were fitted by nonlinear regression using a single-exponential recovery model commonly used in FRAP analyses ¹³.

The half-recovery time (\({t}_{1/2}\)) was calculated from the fitted rate constants as follows:

2.7. 1,6-Hexanediol Treatment and CLEM Analysis

For medium exchange during time-lapse imaging, a live cell imaging fluidic system (Takasago Fluidic Systems, Nagoya, Japan) was used with minor modifications to allow mounting on the stage of the DeltaVision system. Medium exchange was performed by perfusing an observation medium containing 2.5% (v/v) 1,6-hexanediol into a glass-bottomed dish at a flow rate of 3 mL/min using two tubes, while an equivalent volume of the original medium was simultaneously removed from the opposite side. Statistical analyses were performed using R. A two-way repeated-measures ANOVA was conducted with fluorophores and treatment conditions as within-subject factors (Fig. 5(b)). Correlative light and electron microscopy (CLEM) of 2.5% (w/v) glutaraldehyde-fixed samples was performed as described previously ⁵. For quantitative analysis, the grayscale intensities within an annular region extending 550 nm from the bead surface were measured using Fiji. The measured intensities were normalized to the grayscale intensity of the corresponding beads and plotted for comparison between samples.

3.1. Intracellular Delivery of DNA-Beads and Accumulation of Endogenous Cytosolic DNA Sensors on the Beads

The experimental setup used in this study is illustrated in Fig. 1. We previously reported that when DNA-beads were introduced into HeLa cells using a transfection reagent, the cytosolic DNA sensor protein BAF rapidly accumulated on the beads immediately after endosomal membrane rupture, allowing the beads to evade autophagy ⁵. However, it has remained unclear whether cGAS (cyclic GMP-AMP synthase), which plays a critical role in the cGAS-STING pathway as an exogenous DNA ^7,8, accumulates on DNA-beads.

Therefore, we first examined whether endogenous BAF and cGAS accumulated on DNA-beads through immunofluorescence staining using specific antibodies against BAF and cGAS. Accumulation of endogenous BAF and cGAS on DNA-beads was observed 4 h after washout of unattached beads (Fig. 2). In contrast, no accumulation of BAF or cGAS was detected on the beads without DNA.

3.2. Live-Cell Imaging of BAF and cGAS Accumulation on DNA-Beads

To examine the dynamics of BAF and cGAS in the vicinity of DNA-beads, we generated a stable cell line co-expressing GFP-BAF and cGAS-mRFP (HeLa/GFP-BAF/cGAS-mRFP). When DNA-beads were introduced into this cell line, all GFP-BAF-positive beads were also positive for cGAS-mRFP in the cytoplasm of the cell 4 h after washout of unattached beads (117 BAF-positive beads were examined in total across three glass-bottom dishes) (Fig. 3(a)). These results indicate that this stable cell line is suitable for analyzing the dynamics of BAF and cGAS around DNA beads in the cytoplasm.

Next, DNA-beads were introduced into HeLa/GFP-BAF/cGAS-mRFP cells and subjected to fluorescent time-lapse imaging. Both BAF and cGAS began to accumulate on the DNA-beads nearly simultaneously, and fluorescence signals on the beads reached about 80% of their final intensity within approximately 10 min (Figs. 3(b) and (c)). The fluorescence accumulation half-times of GFP-BAF and cGAS-mRFP were 2.2 and 4.9 min, respectively (Fig. 3(c)). Based on our previous report showing that the accumulation of BAF on DNA-beads starts immediately after endosomal membrane rupture ⁵, these results indicate that both BAF and cGAS rapidly bind to double-stranded DNA on beads following endosomal disruption. These results partially mimic the state in which both BAF and cGAS simultaneously localize to DNA exposed to the cytosol after nuclear envelope rupture ¹⁴.

3.3. FRAP-Based Analysis of BAF and cGAS Dynamics Around DNA-Beads

Previous studies have shown that nuclear envelope-like membrane structures containing the BAF-binding inner nuclear membrane protein emerin form around DNA-beads within \(\sim\)10 min after their escape from endosomes into the cytosol ⁵. To investigate whether the dynamics of the BAF and cGAS around the DNA beads changed over time after endosomal escape, we performed FRAP analysis. When photobleaching was performed 3 min after the beads became positive for both BAF and cGAS, the recovery of fluorescence signals corresponding to at least 30% of the prebleach intensity was observed within 5 min after bleaching for both BAF and cGAS (Fig. 4(a)). In contrast, when photobleaching was performed \(\ge 10\) min after endosomal escape, fluorescence recovery was markedly reduced (Fig. 4(b)). Quantitative analysis revealed no significant difference in the half-recovery time (\({t}_{1/2}\)) of either BAF or cGAS between the two groups (Fig. 4(c), upper panels). In contrast, the recovery fraction (\(R_{f}\)) of cGAS was significantly reduced in DNA-beads that had resided in the cytosol for \(\ge 10\) min compared with those analyzed within 10 min after endosomal escape (Fig. 4(c), lower- right panel). These findings suggest that the mobile fraction of cGAS decreases over time after endosomal escape, resulting in a more stable association with DNA-beads.

3.4. Relationship Between 1,6-Hexanediol-Induced Localization Changes of cGAS-GFP and mRFP-BAF with Intracellular Structures Around DNA-Beads

Recent studies have reported that liquid–liquid phase separation (LLPS) is involved in the function of cGAS as a cytosolic DNA sensor ¹⁵. Therefore, we examined how treatment with 1,6-hexanediol, which is known to disrupt LLPS, affects the localization of BAF and cGAS on DNA beads. As the HeLa/GFP-BAF/cGAS-mRFP cell line was not sufficiently tolerant to 1,6-hexanediol treatment for quantitative analysis, we used a cell line in which the fluorophores were swapped between BAF and cGAS (cGAS-GFP and mRFP-BAF). Time-lapse imaging was performed to monitor the localization dynamics of BAF and cGAS during treatment with 2.5% (v/v) 1,6-hexanediol, which was delivered to the culture medium using a fluidic control device. Upon hexanediol treatment, the BAF signal associated with DNA-beads gradually decreased over time, whereas the cGAS signal remained largely unchanged (Fig. 5(a)). Quantitative analysis of the fluorescence intensities confirmed this observation. The fluorescence intensity of mRFP-BAF was significantly reduced at 10 min after exposure to 2.5% hexanediol and remained low after washout, whereas the fluorescence intensity of cGAS-GFP did not exhibit a significant change throughout the observation period (Fig. 5(b)). A two-way repeated-measures ANOVA revealed significant effects of fluorophore (\({F}(1,6) = 93.81\), \({p} = 6.95\times10^{-5}\)) and condition (\({F} (2,12) = 34.41\), \({p} = 1.07\times10^{-5}\)), as well as a significant fluorophore \(\times\) condition interaction (\({F} (2,12) = 30.75\), \({p} = 1.9\times10^{-5}\)). These results indicated that cGAS-GFP and mRFP-BAF exhibited distinct fluorescence intensity changes in response to hexanediol treatment and washout. Taken together, these results suggest that although both BAF and cGAS bind to double-stranded DNA coated on beads that enter the cytosol, they exist in distinct states with different sensitivities to 1,6-hexanediol.

We conducted a TEM analysis to examine whether the loss of BAF from the DNA-beads was accompanied by changes in the ultrastructure. Previous studies have shown that, after sufficient time following washout of unattached beads, DNA-beads introduced into cells become surrounded by nuclear envelope-like membrane structures, with an electron-lucent region between the bead and the surrounding membrane, distinct from the cytosolic region ⁵. Under these conditions, the beads subjected to 1,6-hexanediol treatment were chemically fixed with glutaraldehyde and examined by TEM. Compared with untreated samples, the electron density of the DNA-containing region between the beads and the nuclear envelope-like structures was markedly increased (Figs. 6(a) and (b)), becoming indistinguishable from that of the surrounding cytosol. Quantification of the electron density surrounding DNA-beads revealed lower normalized grayscale intensity following 1,6-hexanediol treatment compared to untreated controls (Fig. 6(c)). As lower normalized intensity values indicate higher electron density, these findings suggest that hexanediol treatment induces the accumulation of electron-dense molecules around the DNA-beads. As BAF has been reported to bridge double-stranded DNA molecules ¹⁶, the decrease in BAF associated with DNA-beads induced by hexanediol treatment may have altered the local DNA architecture, potentially permitting the accumulation of other intracellular factors around the beads. Taken together with the observed reduction in the BAF signals around the beads upon 1,6-hexanediol treatment (Fig. 5), these results suggest that BAF plays a role in restricting the access of cytosolic factors to the DNA on the beads.

Previous studies have shown that BAF efficiently binds to and bridges double-stranded DNA and outcompetes cGAS in DNA binding under certain conditions ^14,16. In addition, BAF recruits the LEM-domain inner nuclear membrane protein, emerin, to the exposed DNA and contributes to the formation of nuclear envelope-like membrane structures around the cytosolic DNA ⁵. Consistent with these observations, the rapid accumulation of BAF observed in this study may have facilitated the initial organization of the DNA-bead-associated structures. Furthermore, the progressive decrease in the mobile fraction of cGAS after endosomal escape suggests the stabilization of cGAS-DNA interactions over time. The reduced mobile fraction of cGAS observed at later time points (Fig. 4(c)) may, therefore, reflect the formation of a BAF-dependent nuclear envelope-like membrane structure around DNA-beads, which could contribute to the stable retention of cGAS within the DNA-associated compartment. Interestingly, although 1,6-hexanediol treatment reduced the accumulation of BAF around DNA-beads, cGAS localization was largely unaffected. Although the underlying mechanism remains unclear, this difference may reflect distinct modes of retention around DNA-beads, with cGAS remaining associated through direct DNA binding, whereas BAF accumulation may depend, in part, on higher-order DNA-associated assemblies or membrane-associated structures that are more sensitive to 1,6-hexanediol.