*3.3. Selectivity*

Selectivity of the probes is one of the key parameters for evaluating their performances. We first checked the response of **L** to a wide range of the biologically relevant anions including Cl<sup>−</sup>, PO4<sup>3</sup><sup>−</sup>, HPO4<sup>2</sup><sup>−</sup>, H2PO4<sup>−</sup>, P2O7<sup>2</sup><sup>−</sup>, as well as ATP and its analogues ADP and AMP by absorption and emission spectra. As shown in Figure 1, after addition of 1 equiv. of these anions to aqueous solution of **L**, most of the solutions remained yellow with slight changes of λmax, except for that containing ADP, which gave orange solutions with the shifts of the absorption peaks from 450 to 468 nm. Compared with those anions, the most remarkable influence was observed upon adding ATP, which gave a purple solution along with the significant red-shifts of λmax. The significant color change indicated that **L** can serve as a "naked-eye" indicator for ATP in aqueous solution. On the other hand, the addition of 3 equiv. of ATP resulted in almost complete quenching (about 60-fold) in photoluminescence intensity of **L** (Figure 2). However, the addition of other related anions had no obvious effect on the fluorescence emission of **L**, except that adding ADP produced a slight decrease in emission intensity. These results indicated that the probe **L** had high selectivity to ATP over other tested anions in aqueous solution.

**Figure 1.** Absorption spectra of **L** (50 μM) upon addition of various bio-related sodium salts of anions including ATP, ADP, AMP, Cl<sup>−</sup>, PO4 3−, HPO4 2−, H2PO4 − and P2O7 2− (50 μM) in water at pH 7.4.

**Figure 2.** Fluorescent emission changes of **L** (10 μM) upon addition of various bio-related sodium salts of anions including ATP, ADP, AMP, Cl<sup>−</sup>, PO4 3−, HPO4 2−, H2PO4 − and P2O7 2− (30 μM) in water at pH 7.4.

#### *3.4. UV-Vis Spectral Response of Probe L to ATP*

The interaction of **L** with ATP in *tris*-HCl bu ffer solution (10 mM, pH = 7.4) at room temperature was carefully investigated by UV-Vis absorption spectroscopy. As displayed in Figure 3a, the *tris*-HCl bu ffer solution (10 mM, pH 7.4) of **L** (50 μM) exhibited a maximum absorption band at 450 nm with ε = 1.2 × 10<sup>4</sup> M−<sup>1</sup> cm<sup>−</sup>1, which is attributed to π–π\* transition of polythiophene backbone with a random-coil conformation [19,31,32]. As the concentrations of ATP increased, the λmax was gradually red-shifted from 450 to 540 and 588 nm, accompanied with a distinct solution color change from yellow to purple. The significant shift and the appearance of two vibronic bands are characteristic of the aggregation of the polythiophene backbone [19]. Jobs plot (Figure S3, Supplementary Materials) indicated a near 1:1 stoichiometry of complex formed between **L** and ATP. The binding constant of **L** with ATP was estimated to be about 4.02 × 10<sup>5</sup> M−<sup>1</sup> with *R*<sup>2</sup> of 0.995 based on the double-reciprocal

plot of 1/(*A* − *A*0) versus 1/(ATP), as shown in Figure 3b [33–35]. Further, the control homopolymer **PTB,** which just contains boronic acid groups on its side chains, exhibited the binding constant of 2.68 × 10<sup>4</sup> M−<sup>1</sup> (Figure S4, Supplementary Materials). In summary, comparing with other PT-based ATP probes which only contained quaternary ammonium groups [18,24] or boronic acid groups on their side chains, the relatively larger binding constant of **L** with ATP implied stronger affinity between the both, probably due to synergistic interactions.

**Figure 3.** (**a**) UV-Vis absorption spectra of **L** (50 μM) with the addition of ATP with concentrations ranging from 0 to 13.6 μM in *tris*-HCl buffer solution (10 mM, pH = 7.4); (**b**) Absorbance at 540 nm of **L** as a function of ATP concentration. *A*0 is the initial absorbance of **L** at 540 nm, and *A*i is the recorded absorbance of **L** in the presence of ATP with different concentrations. Error bars represent the standard deviations of three trials.

Further, we confirmed the formation of **L** supermolecular aggregates induced by ATP by circular dichroism (CD) measurements. As shown in Figure 4, both **L** and ATP are optically inactive, and no CD pattern in the π–π\* transition region can be detected. These results indicate that free **L** adopts an achiral random-coiled conformation in water. However, upon binding with ATP, an intense split-type induced CD (ICD) in the π–π\* transition region appeared, and as the ATP concentration increased, the ICD intensity increased gradually, suggesting the formation of chiral π-stacked aggregates of **L** in the presence of ATP [36].

**Figure 4.** Circular dichroism (CD) spectra of ATP (15 μM) and **L** (30 μM) in the absence and the presence of various amounts of ATP in water (pH 7.4).

#### *3.5. Fluorescence Spectral Response of Probe L to ATP*

Figure 5a displays the changes of fluorescent spectra of polymer **L** (10 μM) in the presence of different amounts of ATP in *tris*-HCl buffer solution (10 mM, pH 7.4). The fluorescence spectra of **L** exhibited a strong fluorescence at 570 nm upon excitation at 450 nm (ϕ = 0.304). The titration of ATP into **L** buffer solution caused that the fluorescent intensities of the emission band centered at 570 nm decreased gradually with the increasing concentrations of ATP. The significant quenching of the emission of **L** could be explained by the formation of π-stacked aggregates of **L** induced by ATP, as demonstrated by UV-Vis and CD studied, quenching light emission via other nonradiative pathways [24,25]. The quenching constant *K*SV was determined according to the well-known Stern-Volmer equation [37]. As seen from the Stern-Volmer plot (Figure 5b), there was a linear relationship between (*F*0 − *F*i)/*F*0, and the concentrations of ATP ranging from 0 to 0.8 μM and *K*SV were estimated to 1.08 × 10<sup>6</sup> M−<sup>1</sup> (*R*<sup>2</sup> = 0.997). The high quenching efficiency indicated the strong affinity between **L** and ATP. In addition, for our system, only 0.16 equiv. of ATP almost completely quenched the fluorescence of **L**, but in case of the PT-based ATP probe bearing only quaternary ammonium groups, about 1 equiv. of ATP was required to produce such a degree of quenching under similar detection conditions [19]. Thus, it is reasonable to believe that boronic acid groups were involved in binding of **L** with ATP, which promoted the conformational transformation and subsequent aggregation of **L**.

**Figure 5.** (**a**) Fluorescence spectra of **L** (10 μM) with the addition of various concentrations of ATP in *tris*-HCl buffer solution (10 mM, pH 7.4); (**b**) Plot of (*F*0 − *F*i)/*F*0 at 570 nm vs. concentrations of ATP. *F*0 is the initial emission intensity of **L** (10 μM), and *F*i is the recorded emission intensities of **L** in the presence of ATP with different concentrations. Excitation wavelength: 450 nm. Error bars represent the standard deviations of three trials.

From the changes in ATP concentrations-dependent fluorescence intensities (Figure S5, Supplementary Materials), the detection limit was estimated to be 2.07 × 10−<sup>9</sup> M based on the 3σ/slope (where σ is the standard deviation of the background and slope is the sensitivity) [24,38]. Such a low detection limit should be very competitive with most of previously reported fluorescence or colorimetric probes for ATP [18–25] (Table S1, Supplementary Materials). The high sensitivity of **L** for ATP should be closely related to the strong synergetic interactions between **L** and ATP.

In order to further confirm that cyclic borate can be formed between 3-pyridineboric acid of **L** and ribose of ATP in weakly basic aqueous solution, 1H NMR spectra of 3-pyridineboric acid, ATP, and their mixture at pH 8.0 in D2O were collected, as shown in Figure S6. From Figure S6 (Supplementary Materials), it is obviously seen that both the strong peak at δ 7.42 in 1H NMR of 3-pyridineboronic acid (Figure S6A, Supplementary Materials) assigned as the 5-position Ha and the peak at δ 6.08 in 1H NMR of ATP (Figure S6B, Supplementary Materials) corresponding to Hb of ribose in ATP moved towards higher field in 1H NMR of 3-pyridineboronic acid/ATP mixture, along with more complex splits (Figure S6C, Supplementary Materials). It could be explained as follows: the neutral form of the boronic acid group serves as an electron-withdrawing group, but after complex with vicinal diols of ribose in ATP, it is transformed into the anionic form, which will act as an electron donor group [28]. These results indicated the formation of covalent bonds between 3-pyridineboronic acid and ATP in a weakly alkaline environment.

#### *3.6. Assay for ADK*

Since **L** can selectively recognize ATP from phosphate-containing anions, particularly ADP and AMP, **L** was applied to set up a real-time fluorescence assay for ATP-relevant enzyme activity. For instance, adenylate kinase (ADK) is an important enzyme which is involved in maintaining cellular energy homeostasis. It can catalyze the reversible reaction: ATP + AMP - 2ADP. As shown in Figure 6a, **L** (50 μM) exhibited an absorption peak at about 450 nm in *tris*-HCl buffer solution (10 mM, pH 7.4) containing 0.2 mM MgCl2. When the ATP (20 μM) and AMP (20 μM) were added into the above solution, the absorption band of the solution red-shifted to 535 and 578 nm along with an obvious color change (Inset, Figure 6a). After incubating with ADK ((ADK) = 100 mU/mL) for 15 minutes at 37 ◦C, the conversion reaction from ATP and AMP to ADP happened, resulting in that the λmax returned to 450 nm and the solution color became yellow (Figure 6a). It was interesting to note that the resultant absorption spectra was almost identical with the control spectra of **L** with 40 μM of ADP, indicating the ATP and AMP can be converted to ADP thoroughly under the catalysis of ADK. These results demonstrated preliminarily that **L** can be used as a probe to monitor ADK activity in buffer solution.

**Figure 6.** (**a**) Absorption spectra of **L** before and after incubating with ADK in buffer solution. Inset: The photographs taken from **L** (1) and L/ATP/AMP before (2) and after (3) incubating with ADK. (**L**) = 50 μM, (ATP) = (AMP) = 20 μM, (ADP) = 40 μM, (ADK) = 100 mU/mL. Incubating time was 15 min; (**b**) Time-trace plots of the conversion from ATP and AMP to ADP catalyzed by various concentration of ADK (5–20 mU/mL) monitored by the emission ratio *F*i/*F*0. (**L**) = 10 μM, (ATP) = (AMP) = 2 μM. Excitation wavelength: 450 nm. All measurements were performed in *tris*-HCl buffer solution (10 mM, pH 7.4) containing 0.2 mM Mg<sup>2</sup>+ ions.

Figure 6b shows the dependence of the fluorescence intensity ratio (*F*i/*F*0) at 570 nm of **L** on the concentration of ADK and incubating time. In this study, the fluorescent spectra were measured at 2 min intervals. From Figure 6b, it is seen that the reaction rate was accelerated with the increase in ADK concentration and after about 10 min, all reaction reached equilibrium. These results indicated that **L** has grea<sup>t</sup> potential for facile real-time monitoring of enzyme activity to elucidate their functions in biological systems.

## *3.7. Cell Imaging*

Before applying to cell imaging, the biocompatibility of **L** was investigated by MTT assay. As shown in Figure S7 (Supplementary Materials), as the concentration of **L** increased, the average

cell viability was greater than 80%, suggesting low cytotoxicity of **L** to living cells in the range of tested concentrations.

The applicability of in situ sensing ATP using **L** as a probe in living cell was further investigated by a confocal laser scanning microscope (CLSM). After incubation of HeLa cells with **L** (10 μM) for 2 h at 37 ◦C, the strong green fluorescence of **L** was observed in the cells (Figure S8, Supplementary Materials), demonstrating that **L** can be uptaken by HeLa cells with good biocompatibility. The photostability measurement (Figure S8, Supplementary Materials) showed almost negligible changes in the fluorescence intensities of **L** in HeLa cells after continuously irradiating at 405 nm for 100 s.

The sensing ability of **L** in HeLa cells was subsequently assessed by extracellular addition of ATP (4 μM). Figure 7 shows the representative CLSM images of **L**-loaded HeLa cells that were cocultured with ATP for various times. It was observed that the fluorescence intensity of **L** was decreased gradually with the prolonged incubation time. After about 30 min, the fluorescence of **L** was almost completely quenched. As a control, the fluorescence intensity of **L** itself in HeLa cells did not change significantly within the same test time. These results suggested that **L** had the potential to detect the fluctuation of ATP concentration in living cells.

**Figure 7.** CLSM images of **L**-loaded HeLa cells incubated with ATP (4 μM) for various times from 0 to 34 min (**<sup>a</sup>**–**h**). (**i**): Brightfield; (**j**): Overlay of (a) and (i); (**k**): Dot representing the integrated optical density (IOD/area) of the probe **L** from the fluorescence images at each time point. Scale bars: 20 μm. λex = 405 nm, λem = 564–633 nm. Error bars represent the standard deviations of three independent experiments.
