Detection of the UV-vis silent biomarker trimethylamine-N-oxide via outer-sphere interactions in a lanthanide metal-organic framework

Trimethylamine-N-oxide (TMAO) is a biomarker of the cardiovascular disease that is one of the leading causes of worldwide death. Facile detection of TMAO can significantly improve the survival rate of this disease by allowing early prevention. However, the UV-vis silent nature of TMAO makes it intricated to be detected by conventional sensing materials or analytical instruments. Here we show a bilanthanide metal-organic framework functionalized by borono group for the recognition of TMAO. Superior sensitivity, selectivity and anti-interference ability were achieved by the inverse emission intensity changes of the two lanthanide centers. The limit of detection is 15.6 μM, covering the clinical urinary concentration range of TMAO. A smartphone application was developed based on the change in R-G-B chromaticity. The sensing mechanism via a well-matched outer-sphere interaction governing the sensing function was studied in detail, providing fundamentals in molecular level for the design of advanced sensing materials for UV-Vis silent molecules.

C onvenient recognition of biomarkers for early diagnosis of major diseases of human beings is a general strategy that requires continuous development 1 . Cardiovascular disease (CVD) is one of the leading causes of death around the world and hence is a major public health problem 2 . Early detection of CVD is extremely important for reducing mortality because of the suddenness of this disease 3 . Trimethylamine-N-oxide (TMAO), an intestinal gut flora metabolite of phosphatidylcholine and L-carnitine, has been recognized as an important biomarker for CVD [4][5][6] . TMAO is mostly eliminated through urine, and the concentration of TMAO in infarcted patients is approximately 2.2 times higher than that in healthy people [7][8][9] . Therefore, real-time monitoring of TMAO concentrations in urine is highly important for CVD prevention. To date, there have been limited reports on TMAO recognition, and most of them have to use timeconsuming and cumbersome chromatography or mass spectrometry 10,11 . Although electrochemical methods 12 , indicator displacement assays 13 and colorimetric sensor arrays 14 have been reported for sensing UVsilent biomarkers such as TMAO (Supplementary Table 1), there are deficiencies in the accuracy and convenience of such methods that have to be improved to satisfy the practical requirements. The development of a facile sensing material for quantitative recognition of TMAO to meet the requirement for early clinical diagnosis of CVD is highly desirable 3 . However, TMAO has no characteristic absorption or emission in the range of 200-800 nm ( Supplementary Fig. 1), making it difficult to be detected via well-known sensing mechanisms 15,16 .
Owing to the designable structures, metal-organic frameworks (MOFs) have received great attention as sensing materials, which are able to match the targeted chemical analytes in both structures and energy levels [17][18][19][20][21] . Well-designed lanthanide MOFs have shown excellent potential in this field because of their outstanding luminescent properties from lanthanide centers with long lifetimes and sharp line emissions [22][23][24][25] . The sensing functions of high sensitivity, high selectivity and low systematic error can be achieved by lanthanide MOFs with two emission centers because of their excellent self-calibration function and color gradient feature originated from the multi-emission centers [26][27][28][29][30][31] . To achieve high-performance recognition for UV-vissilent TMAO, a well-matched host-guest interaction between the inner and outer coordination sphere in lanthanide MOFs is essential 32 . Since TMAO is an electron donor, the functionalization of lanthanide MOFs by an electron acceptor is a rational approach, but it has not been realized in MOF chemistry.
In this contribution, a family of lanthanide MOFs, {[Ln 2 (BI-PA) 3 Fig. 2), influencing the energy transfer process and hence changing the emission intensity and color. However, no lanthanide MOFs with H 2 BIPA has been reported thus far [33][34][35][36][37] . B1-B10 were synthesized to optimize the sensing function for TMAO, while B11 and {[Tb 0.92 Eu 1.08 (I-PA) 3 (EG) 2 ]·H 2 O} n (B12, H 2 IPA = isophthalic acid) without a borono group were synthesized to study the recognition mechanism. By a well-designed mixed lanthanide strategy, highly selective and sensitive detection of TMAO in simulative urine was achieved by B7, based on which a facile smartphone application was successfully developed.

Results and Discussion
Structure and characterizations. Single-crystal X-ray diffraction studies showed that isostructural B1-B11 crystallized in the P2 1 /c space group (Supplementary data 1-3 and Supplementary  Table 2). The structure of B1 is described here as a representative. The asymmetric unit consists of two crystallographically independent Tb 3+ ions, three BIPA 2− ions, two water molecules and one EG molecule in the coordination sphere and one and a half DMA and six water molecules in the lattice ( Supplementary  Fig. 3a). Both Tb 3+ ions are eight-coordinated but have different coordination environments: bicapped trigonal prism and triangular dodecahedron ( Supplementary Fig. 3b). Tb1 is coordinated by six oxygen atoms from five BIPA 2− moieties and two oxygen atoms from two water molecules; Tb2 is coordinated by six oxygen atoms from five BIPA 2− moieties and two oxygen atoms from an EG molecule. The Tb-O bond lengths range from 2.309(3) to 2.832(4) Å. Tb1 and Tb2 are bridged by one μ 2 -η 2 :η 1carboxylate and three μ 2 -η 1 :η 1 -carboxylates from four BIPA 2− moieties to form a binuclear unit with a Tb···Tb distance of 4.045(4) Å ( Supplementary Fig. 3c). The binuclear units are connected by BIPA 2− to form two-dimensional networks in the ab plane ( Supplementary Fig. 3d), which are further connected by BIPA 2− along the c axis to form a three-dimensional framework (Fig. 1a). Hydrogen bonds are formed by the lattice water molecules and the coordinated EG molecules, water molecules and borono groups ( Supplementary Fig. 3e). By viewing BIPA 2− as 2-connected linker and the binuclear unit as 6-connected nodes, B1 can be simplified as a new 2,6-c net with a Schläfli symbol of {8 12 ;12 3 }{8} 3 ( Supplementary Fig. 3f).
The morphology observed from scanning electron microscope (SEM) images and Fourier transform infrared (FTIR) spectra of B1-B11 are almost the same ( Supplementary Fig. 4), and the powder X-ray diffraction (PXRD) patterns of B1-B11 are well consistent with the simulated peaks based on the single-crystal data of B1 (Fig. 1b), confirming the high-phase purity of these compounds. The water stability at different pH values of representative B1 was studied by FTIR and PXRD analysis ( Supplementary Fig. 5), indicating that B1 is stable in the pH range of 3-12. This high stability is attributed to hydrogen bond networks of lattice water molecules in the rigid framework, which increases the steric hindrance and prevents water molecules from attacking the Tb-O bond 38,39 . The thermogravimetric analysis (TGA) of B1-B11 showed similar weight losses in the temperature range of 40-800°C ( Supplementary Fig. 6). Taking B1 as an example, the first weight loss is from the release of six water molecules (calc. 8.44%, obs. 8.65%) in the range of 40-120°C. A weight loss of 9.50% was observed from 120 to 320°C, corresponding to the loss of one DMA and two coordinated water molecules (calc. 9.62%). The third weight loss of 8.08% was observed between 320 and 450°C, indicating the absence of half a DMA molecule and one coordinated EG molecule (calc. 8.25%). Further weight losses were attributed to the decomposition of the framework.
Luminescent properties. The solid-state luminescence spectra of B1-B10 were acquired at room temperature (Fig. 1c). Four wellresolved emission peaks at 488, 544, 583 and 621 nm were observed in the spectrum of B1 that can be assigned to the 5 D 4 → 7 F J (J = 6, 5, 4, 3) transitions of the Tb 3+ ion. The characteristic emissions at 593, 616, 653 and 703 nm in the spectrum of B10 are attributed to the 5 D 0 → 7 F J (J = 1, 2, 3, 4) transitions of Eu 3+ ions. The spectra of B2-B9 simultaneously show emission peaks attributed to Eu 3+ and Tb 3+ , indicating successful mixing of the two lanthanide ions. The lifetimes of excited states 5 D 4 (Tb 3+ ) and/or 5 D 0 (Eu 3+ ) in B1-B10 were measured to study the Tb 3+ -to-Eu 3+ energy transfer process (Supplementary Figs. 7,8). Compared with the lifetime of B1 ( 5 D 4 ), the lifetime of Tb 3+ in B2-B9 is shorter (Fig. 1d), which might be caused by the formation of a new energy transfer pathway (Supplementary Figs. 9, 10). The efficiency of energy transfer (E) can be quantitatively described using E = 1 -τ Tb-Eu /τ Tb 40 , where τ Tb-Eu and τ Tb are the excited-state lifetimes of Tb 3+ in B2-B9 and B1, respectively (Fig. 1e). The calculated E values reached 40% when the Eu content reached 90% (Supplementary Table 3), further proving the occurrence of the Tb 3+ -to-Eu 3+ energy transfer process in B2-B9. The quantum yields of B1 and B10 were determined to be 98.75 and 44.33%, respectively ( Supplementary Fig. 11). With increasing Eu content, the quantum yields of B2-B9 gradually decreased because of the additional energy transfer pathway from Tb 3+ to Eu 3+ .
Detection of TMAO. A screen of B1-B10 for the detection of TMAO was performed. When 1 mM or 10 mM TMAO was added, different degrees of color changes of the aqueous dispersions of B1-B10 were observed (Fig. 2a). Fluorescence titration experiments of TMAO were first performed for B1 and B10 with a single lanthanide center. The fluorescence intensities of B1 and B10 increased after the addition of TMAO ( Supplementary  Fig. 12). The emission intensities in the spectrum of B1 at 544 nm and the spectrum of B10 at 616 nm follow the Benesi-Hildebrand equation: 41,42 , where I 0 and I are the emission intensities without or with TMAO, K BH is the association constant, and [C] is the concentration of TMAO. The K BH values were calculated to be 1.57×10 4 M and 1.75×10 4 M for B1 and B10, respectively. These large K BH values indicate an effective interaction between B1/B10 and TMAO. Due to the isostructural nature of these compounds, the same interaction should exist in B2-B9 and influence the energy transfer process to induce the color change of the aqueous dispersions of B2-B9. The ratio-metric fluorescence responses of B2-B9 toward different concentrations of TMAO were studied ( Supplementary Fig. 13). B9 was found to be most sensitive to TMAO. However, the deep red color of B9 is not conducive to the visual detection of TMAO.
Considering that B4, B5, B6, and B7 can all be applied for visual detection of TMAO and that B7 shows the best comprehensive performance, the use of B7 for the detection of TMAO was further studied in detail.
The water stability of B7 at different pH values was studied by FTIR and PXRD ( Supplementary Fig. 14), indicating that B7 is stable in the pH range of 3-12. Considering that the specific lanthanide distribution can affect the energy transfer efficiency between Tb 3+ and Eu 3+ , thus affecting the detection result 27,28,43 , elemental mapping of B7 has been investigated ( Supplementary  Fig. 15). Elements of B, Eu and Tb are uniformly distributed in B7, indicating the arrangement of Tb and Eu atoms in B7 are controllable.
Aqueous dispersions containing different concentrations of B7 (0.15-0.30 mg/mL) showed stable emission intensity ratios of I 616 /I 544 ( Supplementary Fig. 16), indicating that water does not interfere with the fluorescence sensing results. The lack of obvious changes in the peak intensities in the time-dependent luminescence spectra of B7 (Supplementary Fig. 17) excludes the influence of particle aggregation or structure change, indicating the high stability of B7 in water. Luminescent titrations of TMAO to B7 aqueous dispersions were performed (Fig. 2b). With the addition of TMAO, the fluorescence intensities at 544 nm decrease, while those at 616 nm increase. The color of the aqueous dispersions of B7 changes from light yellow to pink after the addition of 9 mM TMAO under a 254 nm UV lamp (Fig. 2c, inset). The I 616 /I 544 ratios exhibit a good linear relationship with the concentration of TMAO. To improve the detection accuracy, three independent fluorescence titrations with different volumes of TMAO solutions were performed ( Supplementary Fig. 18). A combined result of these three titrations is shown in Fig. 2c, obeying the equation I 616 /I 544 = 0.706[C] + 0.760. The limit of detection was calculated to be 15.6 μM, which is lower than the clinical urinary TMAO concentration 9 .
Selectivity and anti-interference experiments toward common urine disruptors, including creatine, creatinine, glucose, uric acid, urea, KCl, NH 4 Cl and Na 2 SO 4 , were performed (Fig. 3). After adding these interferents, the fluorescence intensities at 544 nm or 616 nm exhibited moderate selectivity (Fig. 3a, b) but poor antiinterference ability (Fig. 3d, e). However, when using the ratio of I 616 /I 544 as indicator, excellent selectivity and anti-interference ability can be simultaneously obtained (Fig. 3c, f), reflecting the superior advantage of the dual-emission center approach.
Due to the obvious discoloration to allow detection of TMAO, a smartphone application based on R-G-B chromaticity was proposed ( Supplementary Fig. 19). Upon increasing the concentration of TMAO, real-time recording of the R-G-B chromaticity of B7 was performed by a smartphone to determine the concentration-dependent R, G and B values (Supplementary Table 4 Mechanism study. To clarify the sensing mechanism via the designed outer-sphere interaction, a series of characterizations and analyses were performed. PXRD of B7 immersed in 10 mM TMAO for 12 h was consistent with that of the original sample ( Supplementary Fig. 20), indicating the stability of B7 in the presence of TMAO. The UV-vis absorption spectrum of TMAO does not overlap with the excitation and emission spectra of B7 ( Supplementary Fig. 21), excluding the possibility of an internal filtration effect and fluorescence resonance energy transfer mechanism 15,16 . According to the DFT calculation at the B3LYP/ 6-31 G* level ( Supplementary Fig. 22), the LUMO energy level of TMAO is higher than that of H 2 BIPA, indicating the absence of photoinduced electron transfer progress 44 .
B7 can interact directly with TMAO via the borono group to change the triplet energy level of the ligand and eventually change the energy transfer process of the system, resulting in the sensing property. To determine the function of the borono group, B12 was synthesized with isophthalic acid, which has a molecular structure like that of H 2 BIPA but without the borono group (Supplementary data 4 and Supplementary Fig. 23). The emission intensities in the spectrum of B12 at both 544 nm and 616 nm decreased slightly, and the value of I 616 /I 544 did not change after adding TMAO ( Supplementary Fig. 24), indicating that there was no direct interaction between TMAO and B12. In fact, B7 has a one-dimensional channel with dimensions of 4.52 × 5.83 Å 2 , and the molecular size of TMAO is 2.95 × 3.51 × 4.16 Å 3 , indicating that TMAO can enter the channel of B7. The exposed borono group in the channel of B7 is the interaction site to interact with TMAO in outer sphere via two modes: bonding or nonbonding interactions (Fig. 4a).
To further verify the interaction between the borono group and TMAO, 1 H liquid and 11 B solid-state MAS NMR experiments were performed. The 1 H liquid NMR spectra of TMAO, H 2 BIPA, TMAO and H 2 BIPA mixture, and 5-hydroxy-isophthalic acid first exclude the possibility that H 2 BIPA is oxidized by TMAO to form 5-hydroxy-isophthalic acid in DMSO ( Supplementary  Fig. S25), suggesting that the borono group is stable in B7 45 . The 11 B MAS NMR spectra of B7 before and after treatment with  TMAO are similar, and both are characterized by an asymmetric and featureless center band with intense spinning sidebands (Fig. 4b). The observed NMR line shape is induced by adjacent paramagnetic ions around 11 B nuclei, i.e., the Tb 3+ and Eu 3+ within B7 framework. The spectral width only slightly decreases after TMAO treatment, giving rise to a decrease of the quadrupolar coupling constant (C Q ) from 4.4 MHz to 4.2 MHz. The change in 11 B chemical shift value (δ iso ) of −6 ppm clearly implies that B7 can interact with TMAO via certain interaction 13 , and the small differences indicate such interaction is nonbonding (see the comparison between DFT-predicted C Q and δ iso of bonding and nonbonding modes in Supplementary Table 5, 6). In addition, after inclusion of TMAO by B7, differential FTIR spectra features attributed to B-O, B-O-H, and C-B were observed at 1312 cm −1 , 1176 cm −1 and 1096 cm −1 , respectively 46 (Supplementary Fig. 26), which were attributed to the direct interaction of TMAO and B7.
The lifetimes of B7 with TMAO were monitored at 544 and 616 nm (Supplementary Figs. 27, 28). The fluorescence lifetime at 544 nm decreased with the addition of TMAO, whereas the lifetime at 616 nm increased (Fig. 4c). The ratio of the lifetimes at 544 and 616 nm had a linear relationship with the concentration of TMAO (Supplementary Fig. 29 and Supplementary Table 7). This result indicates that the energy transfer process of B7 changes by the interaction with TMAO via forming a B7···TMAO intermediate. The triplet energy levels (T 1 ) of the antennas were measured using B11 before and after interacting with TMAO (B11···TMAO) at 77 K, respectively (Fig. 4d). The energy of T 1 of the antenna decreased from 22675 to 21231 cm −1 after combination with TMAO. The energy gap between T 1 of the ligand and 5 D 0 of Eu 3+ (17500 cm −1 ) changed from 5175 cm −1 to 3731 cm −1 , which falls in the optimal energy gap range to excite Eu 3+ (2500-4000 cm −1 ) 47 . The energy gap between the T 1 of the ligand and 5 D 4 of Tb 3+ (20500 cm −1 ) changed from 2175 cm −1 to 731 cm −1 , which allows back-energy transfer 48 . These energy level changes led to an increase in Eu 3+ emissions and a decrease in Tb 3+ emissions (Fig. 4e). X-ray photoelectron spectroscopy (XPS) spectra of B7 were also measured before and after immersion in TMAO ( Supplementary Fig. 30). The B 1s peak of BIPA 2− was observed at 191.60 eV (+3 valence B) in B7, which shifted slightly to 191.45 eV after treatment with TMAO, indicating an interaction with TMAO and weak electron transfer from the O atom of TMAO to the B atom of BIPA 2− 49,50 .

Conclusions
A family of lanthanide metal-organic frameworks functionalized with borono groups were synthesized for the visual recognition of UV-vis silent molecules, demonstrated by TMAO that is the biomarker of cardiovascular disease. Well-matched interaction between the borono group of bilanthanide metal-organic framework and TMAO in the outer coordination sphere was successfully achieved, to provide a facile detection of TMAO. The superior sensitivity, selectivity and anti-interference ability of this borono-functionalized lanthanide metal-organic framework is based on the inverse variation trend of the emission intensities of two emission centers, which is originated from TMAO-influenced energy transfer process. This work not only provides a facile strategy for TMAO on-site detection that could be used to reduce the mortality rate of cardiovascular diseases but also reveal the fundamentals in molecular level for the design of advanced sensing materials for UV-vis silent molecules that related to public health and green environments.

Methods
Synthesis. H 2 BIPA (1 mmol, 0.208 g), 10 mL EG, 10 mL DMA and 40 mL H 2 O were added to a 100 mL beaker to form a solution under ultrasound. 4 mL as prepared solution and 1 mL terbium acetate tetrahydrate aqueous solution (0.1 M) were added to a 10 mL glass vial, which was then sealed and heated at 90°C for 12 h. The obtained crystals (B1) were washed with distilled water for three times and then dried in air. B2-B11 were synthesized similarly to B1, except 0.1 M terbium/europium/gadolinium acetate aqueous solution with different volumes (1 mL in total) were used. B12 was synthesized similarly to B7, except changing H 2 BIPA to H 2 IPA. Isomorphic B12-Tb was synthesized similarly to B12, except using 1 mL 0.1 M terbium acetate aqueous solution. Elemental analysis (EA), inductively coupled plasma-atomic emission spectrometry (ICP-AES) results and yields of all compounds were summarized in Supplementary Table 8.
Characterization. The single-crystal data were collected using a Rigaku SuperNova or a Rigaku XtaLAB Mini II single-crystal diffractometer equipped with graphitemonochromatic Mo-Kα radiation (λ = 0.71073 Å). The structures were solved by SHELXS (direct methods) and refined by SHELXL (full matrix least-squares techniques) in the Olex2 package 51,52 . PXRD measurements were performed using a Rigaku Smartlab SE X-ray diffractometer equipped with a Cu-tube and a graphite monochromator scanning over the range of 5-50°at the scan rate of 0.2°s −1 at room temperature. Simulations of the PXRD patterns were carried out with the single-crystal data and diffraction crystal module of the Mercury program available free of charge via http://www.ccdc.cam.ac.uk/mercury/. The SEM images and EDS elemental mapping were obtained using Hitachi SU3500 scanning electron microscopy equipped with Brooke energy spectrometer. The FTIR spectroscopy were carried out on a Bruker ALPHA spectrophotometer. TGA data were obtained on TGA 2 STAR e System of METTLER TOLEDO under nitrogen atmosphere with the heating rate of 10°C min −1 . EA for C, H and N were carried out using a Vario EL cube elemental analyzer. ICP-AES analyses were conducted using a Thermo IRIS Advantage instrument. UV-vis absorption spectra were measured with a SHIMADZU UV-2600 spectrophotometer. XPS were acquired using PHI5000-Versa probe equipped ESCALAB 250xi. 1 H NMR spectra were recorded on a Bruker AV400 spectrometer. The solid-state MAS 11 B NMR experiments were performed on a 400 MHz Digital Solid-State NMR Spectrometer with 2.5 mm MAS probe at 20 kHz. Luminescence spectra were recorded on an Edinburgh FS5 fluorescence spectrophotometer equipped with a xenon lamp and pulsed flash lamps at room temperature. Photos were taken by an iPhone 12 and chroma values (R, G and B) of every photo are acquired using Swatches app obtained from App Store.
DFT calculation. The ground-state structure optimization, the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO) energy levels of H 2 BIPA, TMAO and H 2 BIPA···TMAO were calculated by the DFT method at B3LYP/6-31 G* level by the Gaussian 09 W program package 53 . Molecular modeling simulation of B1···TMAO was performed at ultrafine level in Forcite module using Materials Studio. The energy and the force were set as 2.0×10 −5 kcal/mol and 0.001 kcal/mol/Å, respectively. The displacement was 1.0 ×10 −5 Å.
Luminescence measurements. The finely grounded sample (30 mg) was dispersed into distilled water (100 mL) to form aqueous dispersions. The mixture was sonicated for 10 min. The luminescence spectra of B2-B9 dispersions upon excitation at 254 nm were measured in situ after incremental addition of freshly prepared water solution containing 100 mM TMAO. Interference experiments were performed using 2 mL aqueous dispersions of B7. Fluorescent lifetimes of B7 without or with TMAO with different concentrations were obtained upon excitation at 254 nm. For the smartphone application, after adding 100 mM TMAO with different volumes to 0.5 mL aqueous dispersions of B7, respectively, images were taken by an iPhone 12.

Data availability
All the data of this study are available. The authors declare that the data supporting the findings of this study are available within the article and its Supplementary Information files. Supplementary Data 1-4 are X-ray crystallographic data for B1, B10, B11 and B12-Tb, which have been deposited at the Cambridge Crystallographic Data Centre (CCDC), under deposition numbers 2092326 -2092329, respectively. These data can also be obtained free of charge from the CCDC via www.ccdc.cam.ac.uk/data_request/cif. The data that support the findings of this study are available from the corresponding author.