2.2. Photoactivated Conformational Change of BMP Receptors Initiates Smad1/5/8 Cascades

We then clarified how the BMP signaling was directly initiated by laser in the absence of BMP ligands. Two different inhibitors of BMPRs, Noggin and LDN193189, which respectively inhibit the extracellular binding site of BMPRs for ligands and the intracellular phosphorylation site of BMPRs, were used to treat cells respectively. These two inhibitors suppressed pSmad1/5/8 in the positive control by ligand BMP2, or DEX. However, only LDN193189 inhibited BMP signaling in the photoactivated ADSCs (Figure 3A). This result implied the photoactivation directly induced the intracellular phosphorylation of BMPRs instead of activating the extracellular BMP ligand binding sites. As a comparison, when the laser focus was moved down into the cytoplasm, the pSmad1/5/8 level was significantly lower than when activated by laser scanning on the cell membrane (Figure 3B). To further confirm that the photoactivation target was BMPRs, the BMPR1a was knocked down (KD), verified by Western Blot and immunofluorescence microscopy (Figure 3C,D). Then, in those KD ADSCs, the pSmad1/5/8 upregulation after photoactivation was greatly reduced. (Figure 3E). In this regard, we speculated the transient femtosecond laser scanning directly activated the intracellular phosphorylation site of BMPRs.

Open in a separate window

Figure 3

The femtosecond‐laser photostimulation activates BMP signaling specifically through BMPRs. A) The immunofluorescence of pSmad1/5/8 24 h after photostimulation (Laser, n = 10 fields in 3 independent trials; noggin, n = 20 fields in 4 independent trials; LDN, n = 9 fields in 3 independent trials), BMP2 (100 ng mL⁻¹) treated (n = 16 fields in 3 independent trials for each group) and DEX (10 nm) treated (DEX, n = 15 fields in 3 independent trials; noggin, n = 16 fields in 3 independent trials; LDN, n = 15 fields in 3 independent trials), respectively, with or without BMPR‐I inhibitor LDN193189 (1 µm) and BMP Ligand antagonist noggin (100 ng mL⁻¹), compared with Control group (n = 24 fields in 6 independent trials). Scale bar: 50 µm. Right panel: the quantified pSmad level from the fluorescence in the Left. B) The quantified pSmad1/5/8 level 24 h after photostimulation located in the cell membrane (n = 24 fields in 3 independent trials) and cytoplasm (n = 21 fields in 3 independent trials). C) Western blot of BMPR1a in the negative control (siNC, transfected with an invalid siRNA) and BMPR1a‐KD ADSCs. D) Immunofluorescence images of BMPR1a in siNC and BMPR1a‐KD ADSCs. Scale bar: 20 µm. E) The quantified pSmad1/5/8 level in siNC (n = 15 fields in 3 independent trials) and BMPR1a‐KD ADSCs (n = 18 fields in 3 independent trials) 24 h after photostimulation, normalized by the level in siNC group. ^* P < 0.05, ^**** P < 0.0001, by two‐tailed unpaired t‐test and one‐way ANOVA analysis corrected by Dunnet's post‐hoc. N.S., no significant difference.

We verified this hypothesis by labeling the N‐terminus of BMPR2 and BMPR1a with Cyan fluorescence protein (CFP) and Yellow fluorescence protein (YFP) respectively to examine the conformational change of these two receptors by measuring the Foster Resonance Energy Transfer (FRET) signals. Initially, the CFP‐BMPR2 exhibited significantly enhanced fluorescence when the YFP‐BMPR1a was bleached (by continuous 473 nm laser scanning) (Figure 4A), suggesting that the BMPR PFCs had already bound together at the N terminus. This binding caused the CFP fluorescence to rise after YFP bleaching, which stopped the CFP‐YFP FRET, and the transferred energy was thus returned to CFP. The 473‐nm laser, instead of green lasers at the YFP‐excitation, was used here to provide pure oxidative stress and avoid the reversible bleaching of YFP or photoactivation of CFP that could cause false‐positive signals in FRET microscopy.^{[ 30 , 31 ]} The CFP‐BMPR2 fluorescence (without YFP‐BMPR1a transfected) treated with such 473‐nm laser scanning did not show any difference (Figure 4A). We also checked the C‐terminus of BMPRs using this method and observed the same results (Figure 4B). These data suggest that BMPR PFCs had bound together at both N─ and C─ terminus, but the GS domain of BMPR1a was not phosphorylated by BMPR‐II before photoactivation.

Open in a separate window

Figure 4

The conformational change of BMPRs initiated by photoactivation. A) CFP‐BMPR2 and YFP‐BMPR1a FRET test indicated by CFP fluorescence by YFP photobleaching. Controls: ADSCs without YFP‐BMPR1a transfected, or without YFP photobleaching (n = 19‐20 cells of each group in 3 independent trials). Boxes, acceptor photobleaching region. Scale bar: 10 µm. Right panel: quantified CFP fluorescence after the intense 473‐nm laser scanning for YFP photobleaching. B) FRET test by BMPR2‐CFP and BMPR1a‐YFP before and after YFP photobleaching (n = 20‐24 cells of each group in 3 independent trials). Boxes: photobleaching region. Scale bar: 10 µm. Right panel: quantified CFP fluorescence after photobleaching. C) FRET test by CFP‐BMPR2 and YFP‐BMPR1a after photostimulation and subsequent photobleaching of YFP (acceptor) (n = 21 cells in 3 independent trials). Boxes, photostimulation, and photobleaching region. Scale bar: 10 µm. Right panel: quantified CFP fluorescence after photostimulation and then photobleaching. D) Similar FRET test by BMPR2‐CFP and BMPR1a‐YFP (n = 18 cells in 3 independent trials). Scale bar: 10 µm. (E,F) Pre‐bleached YFP‐BMPR1a E) and CFP‐BMPR2 F) recovered after photostimulation. Boxes, photostimulation region. Scale bar: 10 µm. G) Pre‐bleached YFP expressed in whole cells could not recover after photostimulation. Boxes, photostimulation region. Scale bar: 10 µm. H) The quantified recovered ratio of YFP fluorescence (Control, n = 12 cells in 3 independent trials; Laser, n = 13 cells in 3 independent trials), YFP‐BMPR1a (Control, n = 15 cells in 3 independent trials; Laser, n = 17 cells in 4 independent trials) and CFP‐BMPR2 (Control, n = 11 cells in 3 independent trials; Laser, n = 16 cells in 3 independent trials). I) The quantified recovered ratio of BMPR1a‐YFP fluorescence (Control, n = 14 cells in 3 independent trials; Laser, n = 17 cells in 3 independent trials) and BMPR2‐CFP (Control, n = 17 cells in 3 independent trials; Laser, n = 16 cells in 3 independent trials). ^**** P < 0.0001 by two‐tailed unpaired t‐test.

We then employed this method to assess the spatial distance between BMPRs during the photoactivation process. After photoactivation, the CFP‐BMPR1a showed increased fluorescence, indicating a decreased FRET and thus the separation of the N‐terminus of BMPR1a and BMPR2 (Figure 4C). An additional increase of CFP fluorescence could be observed when YFP was subsequently bleached (Figure 4C). Hence the distance between the N‐terminus of BMPR1a and BMPR2 was still < 30 nm, allowing for partial FRET. Such N‐terminus detachment of BMPRs was consistent with that induced by DEX (Figure S1, Supportingg Information). Similarly, the C‐terminus of BMPRs exhibited lower FRET signals after photoactivation and their distance was still quite close for partial FRET (Figure 4D). Given the upregulation of pSmad1/5/8 after photoactivation confirmed the phosphorylation of the GS domain in BMPR1a by BMPR2, in this regard, the photoactivation induced the detachment of both C‐ and N‐ terminuses of BMPRs but the GS domain of BMPR‐I binding with BMPR‐II, as illustrated in Figure 4C,D.

To rule out the possibility of photoconversion or photoswitching^{[ 32 , 33 ]} of fluorescent proteins (FPs) after photobleaching, ADSCs were transfected solely with YFP‐BMPR1a (no CFP transfected). The YFP‐BMPR1a was pre‐bleached by the 473‐nm laser scanning in the entire cell. Then, a subcellular region was stimulated with a single‐time femtosecond laser scanning. The YFP fluorescence immediately recovered partially (Figure 4E,H). The pre‐bleached CFP‐BMPR2 exhibited similar fluorescence recovery after the femtosecond‐laser photostimulation (Figure 4F,H). However, if the cells were transfected with only YFP (not fused with BMPRs) and the YFP was bleached, its fluorescence could not recover after photostimulation (Figure 4G,H). The fluorescence recovery of FPs fused with BMPRs induced by femtosecond‐laser photostimulation suggested the reorientation and transition of BMPRs.^{[ 34 ]} The bleaching could not turn down all FPs due to their anisotropy.^{[ 35 , 36 ]} After photostimulation, the reorientation and transition of BMPRs exposed the remaining FPs to fluorescence excitation. Consistently, if the CFP and YFP were fused with the C‐terminus of BMPR2 and BMPR1a, respectively, and transfected into cells separately, the pre‐bleached FPs partially recovered after the single‐time femtosecond‐laser photostimulation (Figure 4I and Figure S2, Supportingg Information).

We suspected flavin, the endogenous photosensitive molecules of cells, might be responsible for mediating the photoactivated conformational change of BMPRs.^{[ 37 ]} The autofluorescence of flavin^{[ 38 ]} in the photostimulation region, which directly indicated the local level of flavin molecules, was found significantly decreased after photoexcitation (Figure 5A). More importantly, the BMP photoactivation efficiency spectrum was quite close to the two‐photon excitation spectrum of flavins^{[ 39 ]} (Figure 5B). These clues were further verified by using the specific inhibitors of free and bonded flavin separately. The inhibitor of free flavins, potassium iodide (KI), showed significant inhibition to the BMPRs activation, whereas the inhibitor of flavoproteins (bonded flavin) DPI did not present any inhibition to conformational changes of BMPRs (Figure 5C). This result was confirmed by testing the pSmad 1/5/8 level in the presence of KI, diphenyliodonium chloride (DPI), and quinacrine dihydrochloride (QCDC) (another specific inhibitor of free flavins) separately. We found KI and QCDC significantly suppressed the pSmad1/5/8 expression after photoactivation (Figure 5D). Therefore, cellular endogenous flavins were the key mediators of BMPRs photoactivation.

Open in a separate window

Figure 5

Photoreceptors in the photoactivation of BMPRs. A) Autofluorescence of flavin molecules before and after photostimulation, with their difference shown in the heat map (n = 5 cells). Boxes: the photostimulation region. B) Spectrum of BMP photoactivation probability (n = 22–25 cells at each wavelength in 3 independent trials). C) FRET test indicated by CFP fluorescence in groups of photostimulation without any other treatments (n = 21 cells in 3 independent trials), in the presence of DPI (10 µm, n = 19 cells in 3 independent trials) and KI (25 mm, n = 16 cells in 3 independent trials). Insert: FRET test with KI. Boxes, photostimulation region. D) The quantified pSmad1/5/8 fluorescence 24 h after photostimulation in the presence of DPI (10 µm, n = 34 fields in 4 independent trials), KI (25 mm, n = 30 fields in 4 independent trials), QCDC (10 µm, n = 30 fields in 4 independent trials) and without any drugs (n = 33 fields in 4 independent trials). E) Simulated binding energy of photoexcited flavin with each animo site in BMPR1a (rats). The most probable binding site, C180, presented a potential binding energy ΔG = −4.4 kcal mol⁻¹ with photoexcited flavin. TM: transmembrane region. F) FRET test by CFP‐BMPR2 and YFP‐BMPR1a‐C180S after photostimulation (n = 21 cells in 3 independent trials for both mutation and control group). Left: FRET signals of YFP‐BMPR1a‐C180S. Boxes, photostimulation region. G) FRET test indicated by CFP after photostimulation in the presence of DTT (2 mm) and Tween‐20 (0.025%, v/v, 20 cells in 3 independent trials for both), compared with the group without any drugs (n = 21 cells in 3 independent trials). H) The pSmad1/5/8 level 24 h after photostimulation in the presence of DTT (2 mm) and Tween‐20 (0.025%, v/v), compared with the photostimulation group (n = 17 fields in 3 independent trials for each group). ^**** P < 0.0001 by two‐tailed unpaired t‐test and one‐way ANOVA analysis corrected by Dunnet's post‐hoc. N.S., no significant difference. Scale bar: 10 µm.

Previous studies suggested the photoexcited flavin could combine with cysteine (Cys) residues in proteins by forming a thioether bond which further generated hydrophobic cavities in the protein and mediated the formation of hydrophobic bonds between them.^{[ 40 , 41 ]} Hence we verified whether the phosphorylation of the GS domain in BMPR‐I by the kinase domain in BMPR‐II followed this mechanism. We calculated the binding energy of each Cys in BMPR1a and found C180 was the most probable binding site with the highest potential energy of −4.4 kCal mol⁻¹ with photoexcited flavin, which exactly located at the closest position to the GS domain (204‐233) (Figure 5E). If C180 in BMPR1a was mutant to serine (C180S), the photoactivation could not induce the conformational change of BMPRs (Figure 5F). We used Dithiothreitol (DTT) and low‐concentration Tween‐20 treatment to suppress the formation of thioether bonds (photoexcited flavins – Cys) and hydrophobic bonds (BMPR‐I – BMPR‐II), respectively. Each of both inhibited the BMPRs conformational changes activated by the femtosecond laser (Figure 5G). This result was finally confirmed with the suppressed pSmad 1/5/8 expression by them (Figure 5H). Hence the femtosecond laser excited cellular endogenous flavins located near the cell membrane which then mediated the phosphorylation of the GS site in BMPR‐I by BMPR‐II.

Materials

LDN193189 (HY‐12071), U0126 (HY‐12031), LGK974 (HY‐17545), and recombinant BMP2 protein (HY‐P7006) were purchased from MedChemExpress. TEMPO (426 369), α‐tocopherol (α‐T, T1539), rotenone (R8875), diphenyliodonium chloride (DPI, 43 088), potassium iodide (KI, 221 945), quinacrine dihydrochloride (QCDC, 222 992) were purchased from Sigma–Aldrich. Recombinant human noggin protein (6057‐NG) was purchased from R&D Systems. DTT (ST041) was purchased from Beyotime. Tween‐20 (A100777) was purchased from Sangon Biotech. Cells were treated with these reagents according to the protocols provided by the manufacturers. After the laser stimulation experiments, cells were incubated with these reagents for another 30 min before replacing the cell culture medium. While in BMP2 or DEX induction groups, cells were treated with them during the entire experiments.

The Ca²⁺‐free medium followed this formula (140 mm NaCl, 5 mm KCl, 1 mm MgCl₂, 10 mm glucose, 10 mm HEPES, 200 mm EGTA, pH 7.4). For further intracellular Ca²⁺ depletion, cells were treated with membrane‐permeable Ca²⁺ chelator BAPTA‐AM (10 µm, Invitrogen, B6769) in Ca²⁺‐free medium for 15 min before experiments.

Cells were incubated with 2 µm Fluo‐4/AM (ThermoFisher, F14202), 5 µm Dihydroergotamine (DHE, Beyotime, S0063), and 5 µg ml⁻¹ JC‐1 (Invitrogen, T3168) for 30 min at 37°C, then washed and replaced with clean medium to fluorescently indicate intracellular Ca²⁺, ROS, and MMP, respectively. The green and red fluorescence signals were excited at 473 and 543 nm and detected in spectral bands of 500–530 nm and 553–618 nm, respectively.

Photoactivation Scheme

A Yb‐doped fiber femtosecond laser (1030 nm, ≈220 fs, 1 MHz, MenloSystems, BlueCut) and a Ti: Sapphire femtosecond laser (tunable from 680–1080 nm, 140 ± 20 fs, 80 MHz, Chameleon Ultra II, Coherent) were coupled into an inverted confocal microscope (FV1200, Olympus). The femtosecond laser shared the same light path with the confocal microscopy for simultaneous photostimulation in any predefined region and continuous microscopy of the FOV. The laser beams were controlled by the galvo mirrors (for scanning in the predefined region) and a mechanical shutter that was synchronized. In experiments, the lasers were focused by a 30 ×, 1.05 NA, oil immersion objective. The focused femtosecond laser scanned the defined photostimulation region point by point for photoactivation. Its scanning plane was vertically tuned to the upper half (close to the top) of cells to avoid the laser passing through the cytoplasm.

The photoactivation was defined as a single microscopy frame, controlled by both galvo mirrors and the shutter, which could be inserted in a pre‐defined time‐lapse confocal microscopy sequence for real‐time microscopy and photoactivation. The photostimulation duration was thus the time of the frame.

Cells adherent on glass‐bottom (0.17 mm thick) dishes were observed by confocal microscopy and subjected to femtosecond‐laser scanning point by point for a single time in a predefined costume region in FOV. In the FRET experiments, cells were stimulated by the laser scanning in a predefined area (26 × 26 µm²) in the FOV (420 × 420 µm²) frame‐by‐frame (40 × 40 pixels, 2 µs/pixel, 0.06 s/frame) for 0.5 s at 10 mW using the femtosecond laser at 1030 nm. In other experiments, cells were stimulated by the laser at 1030 nm scanning in the whole FOV (420 × 420 µm²) for a single frame (640 × 640 pixels, 2 µs/pixel, 1.5 s) at 8 mW. For the BMP activation efficiency, the tunable femtosecond laser was used to activate cells in a predefined region (26 × 26 µm²) frame‐by‐frame (40 × 40 pixels, 2 µs/pixel, 0.06 s/frame) for 5 s. The power of 700–1000 nm laser at the specimen was kept at 30 mW.

siRNA and Plasmids Transfection

Synthetic small interference RNA (siRNA) sequences targeting BMP receptors type Ia was synthesized and purchased from Genomeditech, Shanghai. siRNA with a nontargeting sequence (siNC) was used as a negative control.

Plasmids of YFP‐BMPR1a, CFP‐BMPR2, and BMPR1a‐YFP were cloned and purchased from Tsingke Biotech, Beijing. BMPR2‐CFP and YFP‐BMPR1a‐C180S were cloned and purchased from GeneChem, Shanghai. The DNA fragments encoding fluorescent proteins (YFP and CFP) were inserted in‐frame to the N‐terminus or C‐terminus of DNA sequences encoding for the BMP receptors type Ia, II (downstream of the signal peptide encoding sequence when inserted to N‐terminus), respectively. YFP‐BMPR1a‐C180S was a mutation of YFP‐BMPR1a, with BMPR1a protein cysteine 180 mutated to serine (C180S). PCMV‐N‐YFP (D2716) was purchased from Beyotime and used as whole‐cell YFP fluorescence control.

Cells were transiently transfected with the siRNAs or plasmids through jetPRIME (Polyplus‐transfection) according to the manufacturer's protocol. Cells were prepared 24 h after the transfection for laser stimulation or silencing validation.

FRET Assay

FRET experiments were performed on cells co‐transfected simultaneously with (YFP‐BMPR1a and CFP‐BMPR2 plasmids) or (BMPR1a‐YFP and BMPR2‐CFP plasmids). The excitation wavelength of CFP (donor) was 405 nm. The excitation and bleaching wavelength of YFP (acceptor) was 473 nm.

We measured the CFP fluorescence intensity of the target cell at first, then bleached the acceptor YFP in this cell by continuous scanning of the 473 nm laser, and recorded CFP fluorescence intensity after photobleaching. To test the extent of photoactivated conformational changes of BMPRs, after recording the CFP fluorescence baseline, cells were photostimulated and the CFP fluorescence was then measured. The YFP was then bleached, and the CFP fluorescence was measured again. The CFP (donor) fluorescence fluctuations indicated the original transferred energy to YFP (acceptor), which thus reflected the efficiency of FRET.

Microscopy and Immunofluorescence Labeling

A confocal microscope (FV1200, Olympus) was used for confocal microscopy. Confocal images were usually acquired with 1024 × 1024 pixels/frame at 2 µs/pixel and further processed by ImageJ.

Primary antibodies, anti‐phospho‐Smad1/5/8 rabbit antibody (Cell Signaling Technology, 13 820, 1:800), anti‐Runx2 rabbit antibody (Cell Signaling Technology, 12 556, 1:5000), anti‐BMPR1a rabbit antibody (ThermoFisher, 38–6000, 2 µg mL⁻¹), anti‐phospho‐Smad2 rabbit antibody (Cell Signaling Technology, 18 338, 1:800), anti‐phospho‐p38 rabbit antibody (Cell Signaling Technology, 4511, 1:1600), anti‐phospho‐eIF4E rabbit antibody (Abcam, ab76256, 1:500), anti‐beta Catenin rabbit antibody (Abcam, ab32572, 1:500), were used according to the protocols from manufacturers.

Cells were first cultured in 35 mm glass‐bottom (0.17 mm thick) dishes. For immunofluorescence microscopy, cells were fixed with 5% paraformaldehyde (Beyotime) and permeabilized with 0.1% Triton X‐100 (Beyotime). After being blocked with 1% bovine serum albumin (Sigma–Aldrich), cells were incubated with primary antibodies at 4°C overnight. The sample was washed with PBS for 5 min, and then incubated with secondary antibodies Goat anti‐Rabbit IgG H&L‐Green (Alexa Fluor 488) (Abcam, ab150077, 1:1000) for 1.5 h at room temperature.

Western Blotting

Cells were dissected and homogenized in lysis buffer (RIPA: protease inhibitor cocktail = 100:1). Whole cell lysates were analyzed in a western blot experiment according to standard protocols. Antibodies used in immunoblotting included anti‐BMPR1a rabbit polyclonal antibody (ThermoFisher, 38–6000, 2 µg mL⁻¹), anti‐GAPDH mouse monoclonal antibody (Proteintech, 60004‐1‐Ig, 1:5000), Peroxidase AffiniPure goat anti‐rabbit IgG (H + L) (Jackson ImmunoResearch, 111‐035‐003, 1:10 000), and Peroxidase AffiniPure goat anti‐mouse IgG (H + L) (Jackson ImmunoResearch, 115‐035‐003, 1:10 000). Proteins were visualized using enhanced chemiluminescence. The images were acquired by Tanon 4200SF (Tanon).

Covalent Binding Site Screening

For the covalent binding Cys site screening analysis of flavin molecules and BMPR1a, flavin molecule was downloaded from the PUBCHEM based on CAS number and BMPR1a from the UniProt database. The structure of BMPR1a was evaluated for reasonableness using the protein structure evaluation function of MOE2015. The results showed that three amino acid residues (AN16, HIS31, SER28) were located in the impermissible region of the protein structure with a percentage of 0.6% (3/532), indicating that the protein structure could be used for molecular docking to screen for Cys amino acid residues that could form a thioether bond with the flavin molecule. Covalent docking experiments were performed using MOE2015 software. The docked Cys site was used as the reaction site, the reaction type was Michael acceptor, and other settings were used as parameters for molecular docking. The covalent docking results of all sites were counted and the binding possibility was judged based on the binding free energy. The results of each Cys in BMPR1a are shown in the following table.

This work was supported by the National Natural Science Foundation of China (NSFC 92054105, 62022056, and 61975118), and the Shanghai Jiao Tong University (YG2022QN063).