Fluolab【Nat.Methods】中红外光热荧光实现20纳米级单细胞器代谢解密
文章标题：FILM: mapping organellar metabolism by mid-infrared photothermal-modulated fluorescence
通讯作者：Ji-Xin Cheng, Meng C. Wang
文章概要
引言
细胞代谢是维持生命活动的核心过程，且高度 compartmentalized（区室化）在不同的细胞器中。作为真核细胞内重要的代谢中心，溶酶体不仅负责生物大分子的降解与回收，其代谢状态还动态影响着细胞信号转导、稳态维持以及机体的衰老与病理过程。尽管溶酶体代谢如此重要，但由于缺乏原位高分辨率的活体测量手段，科学家们此前很难在单个细胞器水平上定量解析其代谢异质性。红外吸收光谱虽能提供丰富的分子结构指纹信息，传统的红外光热显微镜在水相生物样本中却面临空间分辨率不足或背景干扰强的问题，而先前报道的荧光检测中红外光热技术又深陷严重的荧光光漂白泥潭，光子利用率极低。为了打破这一技术瓶颈，本研究开发出了一种创新的荧光检测中红外光热显微镜（FILM）系统，首次实现了在活体细胞和生物体内对单个溶酶体进行全指纹谱的高灵敏度代谢成像。
Fig. 1: FILM principle, instrumentation and spectral fidelity.
a, Principle of FILM microscopy depicted by energy diagram. b, The previous CW fluorescence excitation schematic recorded the entire IR-induced photothermal (PT) dynamics. c, The optical boxcar schematic selectively recorded the ‘hot’ and ‘cold’ states to remove noncontributing photons. d, Schematic of the experimental setup for the FILM microscope. CM, concave mirrors; F, filter; GM, galvo mirrors; M, reflection mirrors; Obj, objective; R obj, reflective objective; SL, scan lens; TL, tube lens. e, Photobleaching curves of standard fluorescence beads (n = 3) under different excitation duty cycles. Data are presented as mean ± s.d. Solid line represents the mean value and the shaded area indicates the s.d. of photobleaching measurements. f, FILM signal of Shigella flexneri expressing GFP measured with different visible light duty cycles (n = 5 independent measurements). Statistical data are presented as mean ± s.d. g, FILM images of S. aureus at 1,650 cm−1 and 1,780 cm−1. Representative results are shown from five independent experiments. h, FILM spectrum of single S. aureus. Scale bar, 10 μm.
主要实验及结论
研究团队首先在硬件层面实现了跨越式的升级，打破了以往连续波激光照射导致的严重光漂白限制。他们创新性地引入了频率解调光学盒车（Optical Boxcar）检测方案，通过将脉冲式可见光探测光与中红外脉冲激发光进行精确的同步时间门控，选择性地只记录温度升高的“热态”和随后冷却的“冷态”，从而完美滤除了不贡献光热信号的背景荧光光子。这一改进让荧光照射时间骤降100倍以上，将光漂白降低了数倍，并在高水相环境下展现出极强的抑制慢热扩散背景的能力。在获取到高质量的高光谱数据集后，针对快速成像带来的低信噪比挑战，团队开发了一种名为SPEND的自监督三维深度学习去噪算法。该算法不依赖外部标签，仅通过对单组高光谱数据进行谱段重排与训练，就将图像信噪比飙升了26.9倍。随后结合引入先验标准的增强型多元曲线解析-最小绝对收缩和选择算子（MCR-LASSO）谱图解卷积算法，成功实现了对大分子复合物成分的精准定量拆解。
Fig. 2: AI-assisted FILM hyperspectral imaging and analysis.
a, Workflow of AI-assisted hyperspectral data analysis. Left: a deep learning-based self-supervised denoising algorithm, called SPEND. The raw noisy hyperspectral data were first rearranged into two different sequences with permutation process. Next, the two sets of noisy data were served as the input and target for a U-Net training. The trained network was then applied to denoise raw hyperspectral data. In the schematic, concat denotes concatenate, conv(+BN)+ReLU indicates convolution followed by batch normalization and rectified linear unit activation, max pool represents max pooling, up-conv refers to up-convolution (transposed convolution) and conv denotes convolution. Right: the ratiometric analysis and MCR–LASSO spectral unmixing process. Reference spectrum of pure chemicals, acquired with the same instrument, were modified with augmented MCR on the basis of the lysosomal data and then fed to LASSO for spectral unmixing and quantification. b, The comparison of FILM images of lysosomes acquired with IR at 1,711 cm−1 and 1,797 cm−1 before and after SPEND denoising. c, Intensity profiles along the dotted red lines marked in b. d, The comparison of raw FILM spectrum without calibration before and after SPEND processing. e, Quantification of image SNR and spectral SNR before and after SPEND denoising (n = 13 lysosomes). f, LASSO unmixing with unmodified references and comparison of original calibrated input and reconstructed spectrum (n = 13 lysosomes). g, LASSO unmixing with MCR-modified references and comparison of original calibrated input and reconstructed spectrum (n = 13 lysosomes). h, The comparison of cosine similarity and Euclidean distance with and without augmented MCR modification (n = 13 lysosomes). Scale bar, 10 μm. In e, f, g and h, the boxes show the interquartile range (IQR), the center lines indicate medians and the lines outside the boxes extend to 1.5 times the IQR.
Fig. 3: Hydrolytic heterogeneity of lysosomes revealed by FILM.
a, Fluorescence image of C. elegans labeled with LysoSensor DND-189. b, FILM spectra of individual lysosomes and surrounding region marked in a. c, Ratiometric mapping of intensity ratios at 1,587 and 1,649 cm−1 (proteolytic activity) and 1,711 and 1,741 cm−1 (lipolytic activity), representing proteolysis activity and lipolysis activities, respectively. d, Classification of lysosomal subpopulations on the basis of the two ratios shown in c. e, Classification of lysosomal subpopulations of mammalian lysosomes. Scale bars, 10 μm. Representative results are shown from three independent experiments.
利用这套强大的AI辅助FILM显微系统，研究人员对活体秀丽隐杆线虫和哺乳动物细胞的溶水酶体展开了多维度的代谢功能剖析。实验ASSIGN了1587波数和1711波数分别对应氨基酸与游离脂肪酸的特征峰，并借此定义了反映蛋白质水解与脂肪水解活性的定量指标。结果首次揭示了即使在同一个细胞内部，不同溶酶体之间也存在着剧烈的代谢功能异质性。进一步的机体衰老追踪研究发现，随着线虫从成年第2天迈向第10天，溶酶体的蛋白质水解和脂肪水解活性均呈现出显著的阶段性下降，且这种代谢功能紊乱在成年第4天（即机体大规模衰老死亡前）就已经早期发生。此外，利用RNA干扰技术抑制线虫体内特定的溶酶体贮积症（LSD）相关基因，FILM系统成功定量捕捉到了大分子的异常堆积模式，发现不同病理基因型下甘油三酯的累积是这类疾病共有的代谢特征，并在尼曼匹克症克隆哺乳动物模型中验证了全球性大分子降解受阻的生化图景。除了溶酶体，该技术还成功泛化应用到了线粒体与脂质滴的特征分型中。
Fig. 4: Age-related metabolic changes at lysosomal scale.
a, Ratiometric mapping of intensity ratios at 1,587 and 1,649 cm−1 and 1,711 and 1,741 cm−1 across worms of different ages. b, Quantitative comparison of the two intensity ratios of lysosomes among four age groups (n = 81 for day 2, n = 61 for day 4, n = 28 for day 6 and n = 61 for day 10 (two-sided two-sample t-test compared to the day 2 group: for the ratio of 1,587 and 1,649 cm−1, _P_D4 versus D2 = 1.41 × 10−11, _P_D6 versus D2 = 1.26 × 10−5, _P_D10 versus D2 = 0.710; for the ratio of 1,711 and 1,741 cm−1, _P_D4 versus D2 = 2.40 × 10−37, _P_D6 versus D2 = 2.98 × 10−18, _P_D10 versus D2 = 3.01 × 10−34). c, Heat map of lysosomal fingerprint spectra extracted from worms in four age groups (n = 69 for day 2, n = 97 for day 4, n = 68 for day 6 and n = 80 for day 10 derived from five to seven independent biological experiments). Each row represents a lysosomal spectrum. d, Representative average spectra for each age group. Data are presented as mean ± s.d. Solid line represents the mean value, and shaded area indicates the s.d. e, Z-score heat map of different age groups. Red boxes highlight signal regions with the higher intensity for the day 2 group. Orange boxes indicate signal regions with the higher intensity for the day 4 group. Yellow boxes highlight signal regions with the higher intensity for the day 6 and day 10 groups. f, t-SNE visualization of all spectra. Each dot indicates a lysosomal spectrum. Shaded area indicates 85% confidence interval. g, Intracluster distance analysis from t-SNE, where larger distances indicate poorer clustering and greater heterogeneity within the data. h, High-content analysis of metabolic profiles across the four age groups (n = 69 for day 2, n = 97 for day 4, n = 68 for day 6 and n = 80 for day 10 derived from five to seven independent biological experiments). All comparisons were made relative to the day 2 group, whose lysosomal profiles are representative of a normal metabolic state (two-sided two-sample t-test, FFAs: _P_D4 versus D2 = 4.89 × 10−40, _P_D6 versus D2 = 8.29 × 10−56, _P_D10 versus D2 = 3.74 × 10−25; protein: _P_D4 versus D2 = 6.67 × 10−9, _P_D6 versus D2 = 1.16 × 10−13, _P_D10 versus D2 = 1.84 × 10−9; AAs: _P_D4 versus D2 = 0.726, _P_D6 versus D2 = 1.23 × 10−27, _P_D10 versus D2 = 1.94 × 10−9; DNA: _P_D4 versus D2 = 5.43 × 10−11, _P_D6 versus D2 = 1.02 × 10−12, _P_D10 versus D2 = 2.95 × 10−10; ceramides: _P_D4 versus D2 = 5.09 × 10−13, _P_D6 versus D2 = 1.02 × 10−3, _P_D10 versus D2 = 1.33 × 10−3; TAGs: _P_D4 versus D2 = 1.97 × 10−4, _P_D6 versus D2 = 3.16 × 10−27, _P_D10 versus D2 = 2.58 × 10−15; glycogens: _P_D4 versus D2 = 1.81 × 10−11, _P_D6 versus D2 = 1.76 × 10−14, _P_D10 versus D2 = 7.57 × 10−7; CE: _P_D4 versus D2 = 0.062, _P_D6 versus D2 = 1.17 × 10−4, _P_D10 versus D2 = 5.92 × 10−3). Scale bar, 10 μm. In b and h, the boxes show the IQR, the center lines indicate medians and the lines outside the boxes extend to 1.5 times the IQR.
Fig. 5: Profiling of metabolic changes associated with LSDs.
a, Heat map of fingerprint spectra extracted from lysosomes under different RNAi conditions (n = 22 for control, n = 32 for nuc-1, n = 35 for aagr-2, n = 21 for asah-2, n = 34 for lipl-3 and n = 24 for ncr-1 derived from two to four independent biological experiments). Each row represents a lysosomal spectrum. b, Representative average spectrum for each RNAi condition. Data are presented as mean ± s.d. Solid line represents the mean value, and shaded area indicates the s.d. c, High-content analysis of lysosomal contents across RNAi groups (n = 22 for control, n = 32 for nuc-1, n = 35 for aagr-2, n = 21 for asah-2, n = 34 for lipl-3 and n = 24 for ncr-1 derived from two to four independent biological experiments). All comparisons were made relative to the day 2 control group, whose lysosomal profiles represent the normal metabolic state (two-sided two-sample t-test, FFAs: P__nuc-1 versus control = 3.23 × 10−14, P__aagr-2 versus control = 0.258, P__asah-2 versus control = 2.67 × 10−7, P__lipl-3 versus control = 2.97 × 10−4, P__ncr-1 versus control = 1.65 × 10−5; protein: P__nuc-1 versus control = 5.49 × 10−6, P__aagr-2 versus control = 0.494, P__asah-2 versus control = 0.306, P__lipl-3 versus control = 0.573, P__ncr-1 versus control = 0.062; AAs: P__nuc-1 versus control = 1.22 × 10−6, P__aagr-2 versus control = 0.451, P__asah-2 versus contro_l_ = 6.95 × 10−5, P__lipl-3 versus control = 0.035, P__ncr-1 versus control = 0.134; DNA: P__nuc-1 versus control = 6.62 × 10−6, P__aagr-2 versus control = 0.118, P__asah-2 versus control = 5.88 × 10−5, P__lipl-3 versus control = 0.241, P__ncr-1 lipl-3 versus control = 0.072; ceramides: P__nuc-1 lipl-3 versus control = 2.51 × 10−4, P__aagr-2 lipl-3 versus control = 0.018, P__asah-2 lipl-3 versus control = 0.032, P__lipl-3 lipl-3 versus control = 0.066, P__ncr-1 lipl-3 versus control = 0.111; TAGs: P__nuc-1 lipl-3 versus control = 4.55 × 10−10, P__aagr-2 lipl-3 versus control = 6.27 × 10−4, P__asah-2 lipl-3 versus control = 1.44 × 10−3, P__lipl-3 lipl-3 versus control = 8.35 × 10−3, P__ncr-1 lipl-3 versus control = 3.44 × 10−3; glycogens: P__nuc-1 lipl-3 versus control = 3.99 × 10−3, P__aagr-2 lipl-3 versus control = 0.043, P__asah-2 lipl-3 versus control = 0.252, P__lipl-3 lipl-3 versus control = 0.017, P__ncr-1 lipl-3 versus control = 0.015; CE: P__nuc-1 lipl-3 versus control = 0.065, P__aagr-2 lipl-3 versus control = 7.06 × 10−4, P__asah-2 lipl-3 versus control = 0.011, P__lipl-3 lipl-3 versus control = 0.020, P__ncr-1 lipl-3 versus control = 6.84 × 10−3). d, Pearson correlation analysis of eight lysosomal contents from C. elegans samples visualized using a chord diagram. Blue curves represent negative correlations lower than −0.5, and red curves represent positive correlations higher than 0.5, with curve thickness indicating the strength of the correlation. e, Heat map of fingerprint spectra extracted from WT and NPC1KO of HEK293T cells (n = 59 for WT, n = 61 for NPC1KO derived from five independent biological experiments). f, Representative average spectra of WT and NPC1KO cell lines. Data are presented as mean ± s.d. Solid line represents the mean value, and shaded area indicates the s.d. g, High-content analysis with statistical comparison of lysosomal chemical contents between WT (n = 59 from five independent biological experiments) and NPC1KO groups (n = 61 from five independent biological experiments) (two-sided two-sample t-test. _P_Protein = 6.41 × 10−12, _P_FFA = 5.26 × 10−11, _P_DNA = 1.49 × 10−11, _P_AA = 0.078, _P_Cer = 3.45 × 10−15, _P_TAG = 2.26 × 10−6, _P_CE = 9.05 × 10−4, _P_Gly = 4.01 × 10−3). h, Pearson correlation analysis of eight lysosomal contents from mammalian cells visualized using a chord diagram. Blue curves indicate negative correlations lower than −0.5, and red curves indicate positive correlations higher than 0.5, with curve thickness reflecting correlation strength. In c and g, the boxes show the IQR, the center lines indicate medians and the lines outside the boxes extend to 1.5 times the IQR.
总结及展望
这项工作成功展示了FILM技术作为活体细胞器级代谢成像平台的巨大潜力。通过巧妙结合光学盒车调制、自监督智能去噪与解开复杂多组分光谱的数学模型，FILM跨越了传统振动光谱缺乏细胞器特异性以及传统荧光成像缺乏高通量化学谱特征的双重鸿沟。它不仅为研究活体状态下亚细胞结构的动态代谢异质性开辟了全新的生化视野，也为深入探究人类生命衰老机制、解密溶酶体贮积症等代谢相关疾病的病理演变规律提供了前所未有的高分辨率化学细胞图谱。未来，随着该技术在超快高光谱谱段覆盖度上的进一步拓展，它将在生物医药转化与临床前精准诊断中发挥出更加深远的影响。