Fig.1 (A) Schematic of RNA-seq analysis in patients with NMIBC. (B) PCA using the DEGs identified in tumor tissue (n = 11) and normal tissue (n = 11) from patients with bladder cancer. (C) Representative sashimi image of CD44 in tumors tissue and normal tissue. (D) Heatmaps from RNA-seq analysis of all CD44 exons in tumor tissue (n = 11) and normal tissue (n = 11). (E) mRNA expression level of each CD44 exon in clinical tissue samples. The horizontal bars represent the mean. (F) IHC analysis with anti-CD44v6 antibody in tumor and normal tissues. n = 3. Original magnification: 30× (left) and 60× (right). (G) IF staining of tumor and normal tissues. n = 3. (H) Quantitative MFI of CD44v6. Data (n = 3) are presented as means ± SD. Statistical differences were analyzed using the two-tailed unpaired Student’s t test. n.s. represents no statistical difference and *P < 0.05.#

基于CD44v6靶点，研究团队构建了由前体多肽DBCO-TPP与叠氮化荧光染料azide-Cy5组成的预靶向平台。如图2所示，前体多肽包含CD44v6靶向序列、基质金属蛋白酶2酶切位点及具有自组装能力的双苯丙氨酸片段。当探针遇到肿瘤微环境中高表达的基质金属蛋白酶2时，多肽发生特异性剪切，释放出残基片段并触发其自组装形成纳米纤维结构。光谱与结构表征表明，自组装形成的网络内部富含β-折叠与π-π堆叠作用，这种高度有序的组装基质有效限制了结合荧光染料的非辐射跃迁，使叠氮化染料在持续激光照射下的抗光漂白稳定性提高了数倍。

Fig.2 (A) Schematic diagram of the MMP2 enzyme cleavage and self-assembly into nanofibers. (B) HPLC analysis of DBCO-TPP, DBCO-RSP, and DBCO-TPP + MMP2. (C) DLS of DBCO-TPP incubated with MMP2. FTIR spectra (D) and CD (E) of DBCO-TPP (50 μM), DBCO-TPP (50 μM) + MMP2 (2 μg ml−1), and DBCO-RSP (50 μM). i: DBCO-TPP; ii: DBCO-TPP + MMP2; iii: DBCO-RSP. (F) High-content screening (HCS) quantitative analysis of DBCO-TPP (50 μM) + MMP2 (5 μg ml−1) incubated with ThT (30 μM) for different times. h, hours. (G) Zeta potential of DBCO-TPP, DBCO-TPP + MMP2, and DBCO-RSP. Fluorescence spectra (H) and kinetic curve (I) of the click reactions of DBCO-RSP (50 μM) and coumarin (50 μM) at different times (5, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, and 130 min). a.u., arbitrary units. (J) CLSM images of DBCO-RSP (50 μM) incubated with ThT (30 μM) and azide-Cy5 (60 μM). (K) Photostability investigation of each group under laser irradiation (660 nm, 0.165 W cm−2) with different times. G1: PBS + azide-Cy5; G2: DBCO-TPP + azide-Cy5; G3: DBCO-TPP + MMP2 + azide-Cy5. Data (n = 3) are presented as means ± SD. Statistical differences were analyzed using the two-way ANOVA method. n.s. represents no statistical difference and ****P < 0.0001#

研究团队接着在细胞与三维肿瘤球模型中验证了探针的原位组装特性。如图3所示，高表达CD44v6的膀胱癌EJ与RT112细胞表面能够诱导纳米纤维的特异性生成，而在低表达的内皮细胞与干细胞表面则无组装现象。在三维肿瘤球穿透实验中，前体多肽展现出深达120微米的优异穿透能力。全原子分子动力学模拟揭示了组装增强近邻黏附效应。如图4所示，前体多肽结合CD44v6受体后被拉近至细胞膜，经剪切释放的残基通过疏水作用与氢键稳固插入磷脂双分子层，原位延伸为稳定的纳米纤维网络，极大地延长了探针在肿瘤细胞膜上的滞留时间。

Fig.3 (A) CLSM images and (B) FI of EJ and HUVECs incubated with RhB-TPP (50 μM) for 2 hours. Scale bars, 20 μm. DAPI, 4′,6-diamidino-2-phenylindole. (C) CLSM images of HUVECs incubated with RhB-TPP in the presence and absence of MMP2. (D) The 3D construction diagram of the incubation with RhB-TPP in RT112 cells for 2 hours. (E) CLSM images of EJ MCTs (20 to 120 μm) taken at 20-μm intervals after incubation with RhB-TPP (50 μM) and RhB-RSP (50 μM) for 2 hours. Scale bar, 100 μm. (F) Schematic illustration of 3D CLSM imaging in tumor spheroid cross sections. (G) FI of RhB-TPP and RhB-RSP in the z-axis distance (20 to 120 μm). (H) SEM images of EJ cells incubated with PBS and DBCO-TPP for 12 hours. Scale bars, 20 μm. (I) Kinetic simulation and structural changes of the phospholipid membranes after binding of the CD44v6 protein to DBCO-TPP. Cyan: phospholipid membrane; blue: CD44v6 protein; magenta: DBCO-TPP. (J) Structural phase plots of the formation of DBCO-RSP and G-TP from DBCO-TPP after it is brought close to the phospholipid membranes and cleaved by MMP2. Data (n = 3) are expressed as means ± SD. Statistical differences were analyzed using the ANOVA method. n.s. represents no statistical difference and ****P < 0.0001.#

Fig.4 (A) Dynamic conformational changes of DBCO-RSP and G-TP + DBCO-RSP at 0, 15, 30, and 45 ns. (B) Representative configurations of peptides and the bilayer at an equilibrium stage in MD simulations. (C) RMSD, (D) intramolecular hydrogen bonds, (E) DBCO-RSP water hydrogen bonding, and (F) SASA during the self-assembly of DBCO-RSP in water. In the presence of G-TP, (G) RMSD, (H) intramolecular hydrogen bonding, (I) distance to the phospholipid bilayer, and (J) SASA were analyzed following the interaction of DBCO-RSP with the cell membrane. (K to M) Binding energy between different peptides and phospholipid membranes, including van der Waals energy (VDWAALS), electrostatic energy (EEL), polar solvation energy (EGB), nonpolar solvation energy (ESURF), total gas-phase free energy (Δ_G_ gas), and total solvation free energy (Δ_G_ solv). (N) Schematic illustration of the AEPA process. (O) Representative CLSM image of EJ cells treated with PBS, FITC-TP, and Cy5-RSP + FITC-TP. Scale bars, 20 μm. (P) Representative SEM image of EJ cells treated with PBS, DBCO-RSP, and DBCO-RSP + G-TP. Scale bars, 20 μm.#

借助原位生成的纳米纤维网络，研究团队在细胞膜表面构建了高密度的生物正交嵌合人工受体。如图5所示，定量评估显示，传统靶向探针受限于受体数量，荧光强度很快达到饱和；而人工受体策略大幅增加了细胞表面的结合位点，使叠氮化染料的富集量提升了5.0倍，并表现出优异的抗光漂白能力。随后，研究团队在小鼠模型中评估了体内成像性能。如图6所示，在皮下气囊与原位膀胱癌小鼠模型中，预靶向平台均展现出极高的高对比度成像能力。在原位膀胱癌模型中，该平台实现了高达5.1的肿瘤与正常组织成像对比比值，显著高于传统靶向探针的3.0和游离染料的1.8，清晰地描绘出肿瘤边界，同时全身毒性试验证实了该体系具备优异的体内生物安全性。

Fig.5 (A) Diagram illustrating the improved recognition efficiency of fluorescent molecules mediated by BCARs. (B) CLSM images and (C) FI quantitative analysis of EJ MCTs were pretreated with DBCO-TPP (50 μM) followed by the addition of azide-Cy5 (0.25 to 10 μM) for a further 30 min and TP-Cy5 (0.25 to 10 μM) as control group for 30 min. Scale bar, 200 μm. (D) FI quantitative analysis of TP-Cy5 in the concentration range from 0.25 to 100 μM. (E) CLSM images and (F) relative FI of DBCO-TPP (50 μM), DBCO-RSP (50 μM), and PBS coincubated with azide-Cy5 (2 μM) in EJ MCTs and irradiated under 660 nm for different times. Data (n = 3) were expressed as means ± SD, and statistical differences were analyzed by the two-way ANOVA method. n.s. represents no statistical difference, ***P < 0.001, and ****P < 0.0001.#

Fig.6 (A) Representative IVIS images and (B) corresponding normalized FI of mice following intratumoral injection of DBCO-TPP + azide-Cy5 or azide-Cy5. (C) Ex vivo imaging of tumors and major organs 10 hours after intratumoral injection. (D) Quantitative analysis of FI in tumors after 10 hours. (E) Bio-TEM images of EJ tumor after treatment with PBS and DBCO-TPP. (F) Schematic illustration of imaging in the APBC model. (G) IVIS images, (H) corresponding normalized FI, and the (I) T/N ratio of APBC imaged with G1 (azide-Cy5, 10 μM), G2 (TP-Cy5, 10 μM), and G3 (DBCO-TPP, 50 μM + azide-Cy5, 10 μM) and then irradiated with a 660-nm laser for 0 min and 2, 4, and 6 hours. Scale bars, 2 mm. (J) IVIS images, (K) corresponding normalized FI, and (L) the T/N ratio of APBC models with varying tumor volumes imaged with G3 (DBCO-TPP, 50 μM + azide-Cy5, 10 μM). Scale bar, 2 mm. (M) Schematic illustration of imaging in the orthotopic bladder cancer model. (N) IVIS images and (O) the corresponding T/N ratio of orthotopic bladder cancer imaged with G1 (azide-Cy5, 10 μM), G2 (TP-Cy5, 10 μM), and G3 (DBCO-TPP, 50 μM + azide-Cy5, 10 μM). Data (n = 3) are expressed as means ± SD, and statistical differences were analyzed using the two-way ANOVA method. n.s. represents no statistical difference.#

为了评估临床转化潜力，研究团队采集了14例膀胱癌患者手术切除的离体肿瘤及正常组织标本进行评估。如图7所示，离体染色成像结果显示，预靶向组在肿瘤组织中的平均荧光强度显著高于正常组织，其成像对比比值较游离染料组提升了2.4倍，诊断的受试者工作特征曲线下面积达到0.8017，敏感性为81.8%，特异性达90.9%。更重要的是，在过渡区组织的成像实验中，该平台能够精准区分肿瘤内部与边缘的信号差异，其边缘成像对比比值达到4.2，为术中实时精准判断切缘状态提供了强有力的证据。

Fig.7 (A) Schematic illustration of ex vivo imaging of human bladder tumor. (B) White light, fluorescence images and H&E images of freshly isolated normal and tumor bladder tissues incubated with azide-Cy5 (G1), TP-Cy5 (G2), and DBCO-TPP + azide-Cy5 (G3), along with (C) the corresponding MFI and (D) T/N ratio. Data are presented as means ± SD and were analyzed by the two/one-way ANOVA. n = 7. (E) Receiver operating characteristic (ROC) analysis and AUC of normal and tumor bladder tissue G3 imaging intensity. n = 11. (F) Fluorescence image and signal intensity of resected transition zone tissue from the patient and the FI under the red line. (G) Tumor margin fluorescence imaging in G1, G2, and G3, along with (H) the corresponding MFI and (I) T/N ratio. Data (n = 3) are presented as means ± SD and were analyzed by the two/one-way ANOVA. n.s. represents no statistical difference.#