【Nat.Methods】“亮度解混”技术，实现单一通道3种靶标同时3D超分辨成像
文章标题：Brightness demixing for simultaneous multi-target imaging in 3D single-molecule localization microscopy
通讯作者：Sandrine Lévêque-Fort
文章概要
引言
单分子定位显微术（SMLM）极大地提高了生物成像的分辨率，但在实现多靶标成像时，传统方法主要依赖基于波长的光谱分离。这种策略不仅受到荧光团光谱重叠的固有限制，通常还需要复杂的色差校正和多通道探测器，从而降低了信噪比并增加了系统复杂性。为了解决这些痛点，研究团队提出了一种被称为亮度解混（Brightness demixing） 的新方法，通过开发和利用荧光团的固有光物理特性（光子通量），在单一探测通道内实现了无需额外滤光片的高效多显色超分辨成像。
Fig. 1: Two-target brightness demixing implementation.
a, Cells labeled for simultaneous SMLM imaging of two targets with low and high brightness dyes. b, Chronogram illustrating molecule blinking undersampling when the exposure time _t_exp is higher or equal to the mean ON time _τ_ON. c, Chronogram illustrating molecule blinking oversampling. The flux is determined by merging localizations of the same molecule, excluding transient binding frames. d–h, DNA-PAINT measurement of microtubules and clathrin-coated pits in N = 1 COS-7 cell in the FOV (representative of five independent experiments), imaged with I1-Atto647N and I3-Atto655, respectively: intensity color-coded localization image representing the total number of photons for each event (d); intensity histogram from d (e); flux histogram following oversampled detection associated to g, with two Gaussian-fitted peaks and a 70% specificity threshold (f); flux color-coded molecule image (g); and two color-coded image after classification (h). i, DNA-PAINT measurement of microtubules and clathrin-coated pits in N = 1 COS-7 cell in the FOV (representative of five independent experiments), imaged with I1-Atto647N and I3-Atto680, respectively. j, Flux histogram associated with i, with two Gaussian-fitted peaks and a 70% specificity threshold.
主要实验及结论
研究人员首先基于DNA-PAINT技术构建了实验体系，利用ASTER动态照明技术确保整个视野内的激发光强均匀分布。通过缩短相机曝光时间对荧光分子的闪烁事件进行过采样，算法能够精确追踪单个分子的完全开启状态并计算其光子通量（Photon flux），从而将其作为表征亮度的可靠指标。在COS-7细胞实验中，该方法成功区分了光谱高度重叠的远红外染料Atto647N和Atto655，清晰地将微管和网格蛋白包裹的小窝分离开来，其误判率极低且保留了远多于传统光谱解混法的有效定位点，并在DNA纳米尺和核孔复合物（NPC）的精细结构解析中验证了该定量分类的极高鲁棒性。
Fig. 2: Brightness demixing on nanorulers and NPCs.
a–c, Nanorulers with orthogonal docking strands imaged by F1-Atto647N and F4-Atto655: reconstruction of the nanorulers with flux color-coded image (top left corner) and after brightness demixing (bottom right corner) (a); flux histogram fitted with a sum of two Gaussians based on N = 670 nanorulers (b); and spot-wise spatial s.d. within nanorulers (c). d–f, NPCs imaged in U2OS cells labeled for Nup62 (I1 strand) and Nup96 (F3 strand), with I1-Atto647N and F3-Atto655: NPC reconstruction with flux color-coded image in N = 1 U2OS cell in the FOV (representative of five independent experiments) (d); flux versus radial distance from NPC center, shown as a density color map with contours lines at σ (with σ from the Gaussian fit) (e); and relative proportions of Nup62 and Nup96 given by the 2D Gaussian fit of e versus radial distance (f). g, Image of the distribution density (N = 1,144 NPCs) in the NPCs for the core Nup62-Atto647N, the crown Nup96-Atto655 and the composite image, obtained with a drift correction described in the Methods.
Fig. 3: Three-target brightness demixing in COS-7 cell.
a, DNA-PAINT measurement of microtubules (magenta), clathrin pits (cyan) and vimentin (yellow) in N = 1 COS-7 cell in the FOV (representative of three independent experiments) labeled for microtubules (magenta), clathrin (cyan) and vimentin (yellow) with P1, P3 and P7 strands, imaged with I1-Atto647N, I3-Atto680 and I7-Atto655 respectively. b, Flux histogram from a after merging with arbitrary bounds.
随后，团队进一步展示了该技术的拓展潜力，通过精细调整分类阈值，利用亮度比例仅为1.5的三个染料成功实现了微管、波形蛋白和网格蛋白的三靶标同时成像。更为重要的是，亮度解混技术被成功引入到3D SMLM成像中，无论是通过插入柱面镜的像散技术，还是通过基于超临界角荧光（SAF）检测的DONALD配置，研究人员都在没有增加物理探测通道的情况下，实现了对细胞结构的绝对轴向定位与多色三维重构，充分证明了该方法与现有主流三维成像技术的高兼容性。
Fig. 4: Three-dimensional brightness demixing in COS-7 cell.
a, Setup for astigmatism-based 3D SMLM with a cylindrical lens (Lcyl). A physical mask is placed in a conjugated plane of the back focal plane to block the SAF collection. c, Setup for SAF-3D SMLM with DONALD configuration. A 50:50 beamsplitter splits the light into two imaging channels: the UAF channel where a physical mask is blocking SAF collection, and the EPI channel where both UAF and SAF are collected. A HILO illumination is used in both configurations. b,d, DNA-PAINT measurement of microtubules and clathrin-coated pits in N = 1 COS-7 cell in the FOV (representative of three independent experiments), imaged with I1-Atto647 and I3-Atto655, respectively. Both orthogonal color bars indicate the relative axial position with respect to the cylindrical lens focus (b) and the absolute axial position relative to the coverslip (d).
总结及展望
亮度解混技术为超分辨显微镜的多靶标成像提供了一种简单而强大的新方案，它打破了依赖多波长激发和多探测通道的传统思维，从根本上规避了色差干扰，同时大幅缩短了实验采集时间。未来，随着针对亮度差异专门优化而非仅针对光谱分离的荧光染料的不断开发，以及该方法在光片显微镜等其他几何结构中的推广，亮度解混必将极大地加速空间蛋白质组学的发展，并为活细胞多靶标纳米尺度动力学的实时研究开辟全新的途径。