We previously synthesized [Ir(pbz)2(DMSO)Cl] and [Ir(ppy)2(DMSO)Cl] as PL and ECL probes for histidine and its metabolites with good emission [25]. Herein, [Ir(dfppy)2(DMSO)Cl] was synthesized for the detection of 3-MeHis. The obtained [Ir(dfppy)2(DMSO)Cl] was characterized and verified by 1H NMR (Fig. S2) and MS (Fig. S3, five peaks centered at m/z [M]+ 649.0674, 650.0711, 651.0696, 652.0731, and 653.0709 for [Ir(dfppy)2(DMSO)Cl]).

The PL spectra of Ir-DMSO without and with 3-MeHis were firstly examined. A negligible PL emission with a maximum wavelength at 450 nm was observed for Ir-DMSO alone (Fig. 1A, I0 = 16710 a.u.). Ir-DMSO with 3-MeHis displays a strong PL emission with two maximum wavelengths at 458 nm and 486 nm. The ratio of Is/I0 at 458 nm was about 24. The ΦPL was 0.4% and 4.2% for Ir-DMSO without and with excess 3-MeHis, respectively, which was measured by a calibrated integrating sphere (errors < 3%). For Ir-DMSO, τ was 148.5 ns according to a monoexponential decay function; for Ir-DMSO in the presence of excess 3-MeHis, τ was 818.4 ns (137.2 ns, 37%; 1212.4 ns, 63%) according to a biexponential decay function (Fig. S4). The changes in the PL spectrum, ΦPL, and τ indicate the reaction between Ir-DMSO and 3-MeHis.

The interaction between Ir-DMSO and 3-MeHis was further confirmed by MS techniques. After adding 5 equivalents of 3-MeHis in the solution of Ir-DMSO, a new peak centered at m/z 742.1408 was observed, which is ascribed to [Ir(dfppy)2(DMSO)(3-MeHis)] (calculated, 742.1413) (Fig. 1B, Fig. S5). Based on our results and the results in references [25, 27], the coordination reaction between Ir-DMSO and 3-MeHis does occur. A new product (named Ir-3-MeHis) and its possible structure were shown in Fig. 1B, inset.

(A) PL emission spectra of 5 μM Ir-DMSO in the absence (a) and presence (b) of 50 μM 3-MeHis in 0.01 M PB (pH = 7.40). λex = 366 nm. (B) Mass spectrum of Ir-DMSO in the presence of 5 equivalents of 3-MeHis (ndfppy = 0.1 mmol, n3-MeHis = 0.5 mmol)

Inspired by the enhancement of PL intensity of Ir-DMSO with 3-MeHis, PL response to different concentrations of 3-MeHis was further checked. It was found that the PL intensity of Ir-DMSO increased when the 3-MeHis concentration increased from 10 to 250 μM (Fig. 2). The increased PL intensity at 458 nm (△I = Is − I0) was linearly proportional to 3-MeHis concentration between 10 and 100 μM (Fig. 2B). The equation was △I = 8.83 × 103C (μM) − 5.03 × 104 (R2 = 0.9986). A detection limit was 3.0 μM. For 1 μM 3-MeHis, a relative standard deviation (RSD) was 2.6% (n = 5). One PL method is feasible for quantifying 3-MeHis in neutral aqueous solution.

(A) PL emission spectra of 5 μM Ir-DMSO with different concentrations of 3-MeHis (1–250 μM) in 0.01 M PB (pH 7.40). λex = 366 nm. (B) The relationship between the increase of PL intensity with 3-MeHis concentration at 458 nm. Inset, calibration curve of 3-MeHis using the PL method

Inspired by the good PL performance of Ir-DMSO to 3-MeHis, we then investigated the ECL behaviors of Ir-DMSO without and with 3-MeHis. At GCE using TPA as coreactant, Ir-DMSO displayed a weak ECL emission with one broad peak potential at + 1.19 V (Fig. 3A, a, 1243 a.u.). For the mixture of 2 μM Ir-DMSO and 10 μM 3-MeHis, two obvious ECL peak emissions at + 1.19 V and + 1.30 V were observed in the presence of TPA at GCE (Fig. 3A, b, 2135 a.u. at + 1.19 V and 2598 a.u. at + 1.30 V). Increasing 3-MeHis concentration to 20 μM 3-MeHis, one obvious and higher ECL peak emission at + 1.30 V was observed (Fig. 3A, c, 11,365 a.u.). The ECL intensity at + 1.30 V increased as 3-MeHis concentration increased. As a control, no ECL emission was observed from Ir-DMSO alone (Fig S6, a) and the mixture of Ir-DMSO and 3-MeHis without TPA (Fig. S6, b, c). Thus, an ECL method can be proposed for the detection of 3-MeHis based on enhancement of ECL intensity of Ir-DMSO-TPA.

In order to illustrate the ECL mechanism, the ECL spectrum of Ir-DMSO-TPA with 3-MeHis was recorded. In 0.1 M PBS (pH 7.40) with 50 mM TPA, a weak ECL wavelength at 492 nm was obtained for Ir-DMSO alone (Fig. 3B, a, 1091 a.u.) while two strong ECL wavelengths at 465 nm (5142 a.u.) and 492 nm (6027 a.u.) were obtained for the mixture of 20 μM Ir-DMSO and 25 μM 3-MeHis (Fig. 3B, b). With the increase of the concentration of 3-MeHis to 100 μM, the ECL emissions increase correspondingly (Fig. 3B, c). In order to compare the PL and ECL spectra in same solution, 0.1 M PBS (pH 7.40) containing 50 mM TPA was used as solvent to record PL spectra. In 0.1 M PBS containing 50 mM TPA, two strong PL maximum wavelengths at 460 nm (3,455,040 a.u.) and 488 nm (2,955,560 a.u.) were obtained for the mixture of 20 μM Ir-DMSO and 100 μM 3-MeHis (Fig. S7). Compared PL behaviors in different solvent, PL intensity of Ir-DMSO is higher in 0.1 M PBS containing 50 mM TPA than that in 0.01 M PB and the maximum emission wavelength is nearly same (Fig. S7). Because of the decay and instability of ECL emissions from the mixture of Ir-DMSO and 3-MeHis in the presence of TPA using multi-potential pulse steps by pulsing between 0 V and + 1.35 V (Fig. S8), ECL spectrum recorded on the fluorescence spectrophotometer was not credible and ECL spectrum was recorded on the CCD camera. By comparing the PL spectra and ECL spectra of 20 μM Ir-DMSO with 100 μM 3-MeHis in 0.1 M PBS containing 50 mM TPA (pH 7.40) on different instrument, similar shape was obtained with slight difference (maximum wavelength and the intensity at maximum wavelength), in which the maximum PL wavelength was 460 nm and 487 nm while the maximum ECL wavelength was 468 nm and 492 nm; the PL intensity at 460 nm was higher than that at 487 nm while the ECL intensity at 468 nm was lower than that at 492 nm (Fig. S9). The difference between PL spectra and ECL spectra may be ascribed to the instruments used for PL and ECL spectra, in which PL spectra was recorded on the fluorescence spectrophotometer while ECL spectra was recorded on the CCD camera [28]. Although slightly different spectra were obtained, the excited state should be Ir-3-MeHis*, similar with that in reference [29]. More sophisticated techniques may be needed to monitor and capture the excited state within ECL emissions.

The electrochemical behaviors of Ir-DMSO without and with 3-MeHis were further studied. By considering the potential window of solvent, the solubility of Ir-DMSO and 3-MeHis (the solubility of Ir-DMSO in VDMSO:VPB = 1:99 is 20 μM, Fig. S10; and 3-MeHis is not soluble in DMSO), 0.1 M PBS-DMSO (VPBS:VDMSO = 1:1) was used for the electrochemical investigation of Ir-DMSO and the mixture of Ir-DMSO and 3-MeHis. pH of 0.1 M PBS-DMSO (VPBS:VDMSO = 1:1) was adjusted to be 7.40 with HCl. Cyclic voltammogram (CV) of 1 mM Ir-DMSO shows an irreversible oxidation peak at + 1.09 V in 0.1 M PBS-DMSO (VPBS:VDMSO = 1:1, pH 7.40, Fig. 3C), which is similar with that in DMSO solvent (Fig. S11). The mixture of 1 mM Ir-DMSO and 5 mM 3-MeHis displays an irreversible oxidation peak at + 1.30 V (Fig. 3D). The peak oxidation current of the mixed 1 mM Ir-DMSO and 5 mM 3-MeHis is higher than that of 1 mM Ir-DMSO alone because of the presence of 5 mM 3-MeHis. CV of 3-MeHis was recorded in 0.1 M NaOH because of positive oxidation potential of 3-MeHis. CV of 3-MeHis shows an irreversible oxidation peak at + 1.31 V in 0.1 M NaOH (Fig. S12). By comparison of electrochemical and ECL behaviors of Ir-DMSO without and with 3-MeHis, positive shift of oxidation peak potential, positive shift of ECL peak potential, the blue shift of the ECL maximum wavelength, and the enhancement of ECL intensity were observed for Ir-DMSO with 3-MeHis.

Thus, an ECL mechanism was proposed for the mixture of Ir-DMSO and 3-MeHis using TPA coreactant. ECL emissions at + 1.35 V should come from Ir-3-MeHis* after the reaction between the oxidized Ir-3-MeHis+• and radical TPA• (generated from the oxidation of TPA and the deprotonation of TPA+•).

(A) ECL intensity vs. potential profiles of 2 μM Ir-DMSO in the absence (a) and presence of 10 μM 3-MeHis (b) or 20 μM 3-MeHis (c) ECL measurement conditions: 0.1 M PBS (pH 7.40) containing 50 mM TPA. Scan rate, 0.10 V/s; PMT is − 900 V. (B) ECL spectra of 20 μM Ir-DMSO without (a) and with 25 μM 3-MeHis (b) 100 μM (c) 3-MeHis. ECL measurement conditions: 0.1 M PBS (pH 7.40) containing 50 mM TPA, multi-potential pulse steps by pulsing between 0 V and + 1.35 V. (C, D) Cyclic voltammograms of 1 mM Ir-DMSO (C) and the mixture of 1 mM Ir-DMSO and 5 mM 3-MeHis (D) in 0.1 M PBS-DMSO (VPBS:VDMSO = 1:1, pH 7.40) at GCE, Scan rate, 0.01 V/s

Based on the enhanced ECL intensity of Ir-DMSO after reacting with 3-MeHis, a “signal on” ECL method was proposed for 3-MeHis assay. Since the reaction time and reaction temperature between Ir-DMSO and 3-MeHis may affect the ECL response, the reaction time and the reaction temperature are optimized. The optimal reaction time and the reaction temperature are 2 h and 37 °C, respectively (Fig. S13). Under the optimal conditions, the ECL intensity of 2 μM Ir-DMSO increased when the 3-MeHis concentration increased from 5 to 25 μM (Fig. 4A). The linear equation was △I = 699.1C (μM) − 4031 (R2 = 0.9963, Fig. 4A, inset). The detection limit was 0.4 μM. Compared with that of analytical performances in references, lower detection limit was obtained in this work (Table S1). For 15 μM 3-MeHis, RSD was 2.7% (n = 5).

For application of the ECL method in real samples, the selectivity is important. In this work, we tested the selectivity of the developed ECL method to 16 types of possible interferents. As shown in Fig. 4B, for 2 μM Ir-DMSO, a great ECL increase (ΔI = 7000 a.u.) was observed for 15 μM 3-MeHis. Negligible ECL change was observed for 150 μM of l-glutamate(ΔI = 20 a.u.), l-tryptophan (ΔI = 100 a.u.), l-aspartic(ΔI = 50 a.u.), l-methionine(ΔI = 250 a.u.), l-cysteine(ΔI = 10 a.u.), l-glutamine(ΔI = 300 a.u.), l-asparagine(ΔI = 10 a.u.), l-glycine(ΔI = 100 a.u.), l-arginine(ΔI = 450 a.u.), l-threonine(ΔI = 10 a.u.), l-serine(ΔI = 20 a.u.), and l-proline (ΔI = 450 a.u.). When increasing the concentration of interferents to 1500 μM, negligible ECL change was observed for l-valine (ΔI = 20 a.u.), l-isoleucine (ΔI = 100 a.u.), l-leucine (ΔI = 100 a.u.), l-methionine (ΔI = 750 a.u.), l-glycine (ΔI = 550 a.u.), and l-arginine (ΔI = 750 a.u.) while there are small ECL responses to l-threonine (ΔI = 850 a.u.), l-cysteine (ΔI = 1050 a.u.), l-glutamine (ΔI = 1050 a.u.), l-asparagine (ΔI = 2750 a.u.), l-serine (ΔI = 950 a.u.), and l-proline (ΔI = 5550 a.u.). Further increasing the concentration to 15,000 μM, negligible ECL change was observed for l-alanine (ΔI = 100 a.u.) and relatively obvious ECL response was observed for l-glycine (ΔI = 950 a.u.), l-arginine (ΔI = 1550 a.u.), l-threonine (ΔI = 1550 a.u.), l-serine (ΔI = 1950 a.u.), and l-proline (ΔI = 13,550 a.u.). Good selectivity is obtained for 3-MeHis compared with these interferences.

We previously developed an ECL method for the quantification of His and its metabolites and found that iridium(III) solvent complex can coordinate with the imidazole group of histidine and its derivatives [25]. It was reported that methylation of histidine at either the N1 or N3 position of its imidazole ring could yield 1-methylhistidine (1-MeHis) or 3-methylhistidine [30], respectively (Chemical structures of them are shown in Fig. S14.). Then, it is expected that both His and 1-MeHis should react with Ir-DMSO. Thus, the performance of Ir-DMSO in the presence of His and 1-MeHis was also assessed. As expected, the ECL intensity of Ir-DMSO also enhanced in the presence of 1-MeHis and His, as shown in Fig. 4C. When the His and 1-MeHis concentration increased from 5 to 25 μM into 2 μM Ir-DMSO solution, the ECL intensity increased (Fig. 4D). The linear equation was △I = 110C (μM) − 966 (R2 = 0.9965) and △I = 1275C (μM) − 1350 (R2 = 0.9994) for His and 1-MeHis. Although there are ECL responses to His, 1-MeHis, and 3-MeHis, the ECL responses between them are different. The slopes of the linear range for 1-MeHis and His are 1275 and 110, respectively. The trend of the equation slope is S1-His > S3-MeHis > SHis. The detection limit is 2.8 μM for His and 0.3 μM for 1-MeHis. The discrimination of His, 1-MeHis, and 3-MeHis can be done by using the sensor array–based technique combined with principal component analysis [25] or the separation using chromatography or electrophoretic separation [10, 13]. Further work is going on in our lab.

(A) ECL intensity vs. potential profiles of 2 μM Ir-DMSO in the absence and presence of different concentrations of 3-MeHis. Inset, calibration curve for 3-MeHis. (B) ECL responses of Ir-DMSO to different amino acids (1, trimethylhistidine; 2, l-alanine; 3, l-glutamate; 4, l-tryptophan; 5, l-aspartic; 6, l-valine; 7, l-isoleucine; 8, l-leucine; 9, l-methionine; 10, l-cysteine; 11, l-glutamine; 12, l-asparagine; 13, l-glycine; 14, l-arginine; 15, l-threonine; 16, l-serine; 17, l-proline acid). (C) ECL responses to 1-MeHis and His. (D) Calibration curve for 1-MeHis (a), 3-MeHis (b), and His (c)

In order to apply the ECL method for 3-MeHis assay in real sample, the amount of 3-MeHis in actual human urines was detected. In original 1 × human urine samples, the ECL intensity of Ir-DMSO was inhibited (Fig. S15). Considering the complexity of urine, deproteinization and dilution pretreatments were done to avoid possible interferences [25]. We then proceeded to detect 3-MeHis in urine with different dilution ratios by the developed ECL method. For one representative urine sample, ECL intensity of Ir-DMSO-TPA increased when the concentrations of 3-MeHis increased from 5 to 15 μM in 50-fold, 100-fold, and 200-fold diluted urine samples (Fig. 5A, Fig. S16-18). The blank ECL intensity is 1614, 1810, and 3743 in 50-fold (Fig. S16), 100-fold (Fig. S17), and 200-fold (Fig. S18) diluted urine samples, respectively. The linear equation was △I = 164.8C (μM) + 956 (R2 = 0.6545), △I = 556.2 C (μM) − 481 (R2 = 0.9999), and △I = 632.2C (μM) + 2014 (R2 = 0.9297) in 50-fold, 100-fold, and 200-fold diluted urine sample, respectively. The trend of the equation slope is S50 diluted < S100 diluted < S200 diluted while the sequence of blank signal is I0-50 diluted < I0-100 diluted < I0-200 diluted, indicating that there are interferences in real urine samples. Nearly the same slope was obtained for 200-fold diluted urine samples and that in 0.1 M PBS (pH 7.40, Fig. S19). Calculating with the standard curve in Fig. 5A, recoveries were ranged from 90 to 108% in 200-fold diluted human serum samples (Table S2), indicating the feasibility of the ECL method for 3-MeHis assay in 200-fold diluted urine samples. Figure 5B shows ECL intensity vs. potential profiles of 2 μM Ir-DMSO without and with 3-MeHis in 200-fold diluted another urine sample. The ECL intensity increased when 3-MeHis concentration increased from 5 to 18 μM. The linear equation was △I = 696.5C (μM) − 4079 (R2 = 0.9313, Fig. 5C). Although different linear ranges were obtained for two different urine samples, the slope was nearly same (Fig. S20). Then, we move to test other urine samples in 200-fold diluted urine samples.

Inspired by the enhancement of ECL intensity, we then proceeded to measure the ECL intensity to 30 different urine samples from individuals with different ages. Urine was collected from 28 healthy volunteers (10 young females (age, 20–30), 7 young males (age, 20–30), 7 old females (age, 45–60), 4 old males (age, 50–70)) and from 2 patients hospitalized with obesity (1 patient, young male) and with chronic kidney disease (1 old female). As shown in Fig. 5D, the median ECL intensity measured from 28 healthy volunteers was 1100 a.u. for 10 young females, 1636 a.u. for 7 young males, 936 a.u. for 7 old females, and 1074 a.u. for 4 old males. The ECL intensity measured from 2 patients hospitalized with obesity (1 patient, young male) is 5500 a.u. and that of patient with chronic kidney disease (1 patient, old female) is 4833 a.u. Higher ECL intensity was obtained from patients compared with healthy volunteers, indicating high concentrations of 3-MeHis or histidine and its metabolites in patients [31, 32]. There was extreme significant difference between patients and healthy volunteers (p < 0.001, Fig. 5E) while there was no significant difference among healthy volunteers (p > 0.05, Fig. 5F) by using the ECL intensity as response. The urine samples from patients with obesity and kidney disease and different healthy people can be identified by the ECL responses of the Ir-DMSO-TPA system, which is promising in identification or screening large numbers of patients for routine analysis.

(A) Calibration curve for 3-MeHis in 50-fold (blue), 100-fold (red), and 200-fold (black) diluted one urine samples. (B) ECL intensity vs. potential profiles of 2 μM Ir-DMSO in the absence and presence of different concentrations of 3-MeHis in 200-fold diluted another urine sample. (C) Calibration curve for 3-MeHis in 200-fold diluted urine sample from Fig. 5B. (D) ECL intensity to 30 different urine samples. (E) Plot of the discrimination of ECL intensity between healthy volunteers and patients. (F) Plot of the discrimination of ECL intensity between healthy volunteers at different ages and different gender