An enzyme-responsive metal-enhanced near-infrared fluorescence sensor based on functionalized gold nanoparticles

An enzyme-responsive NIR nanosystem based on MEF was fabricated by surface functionalization of gold nanoparticles. Sensors based on this strategy are promising for enzyme detection in early diagnostic imaging and in vivo applications.


Introduction
Near-infrared (NIR) uorescence imaging continues to attract attention because, in the physiologically relevant optical window (700-900 nm), the features of this technique include minimal levels of interfering absorption from biological samples, reduced scattering, enhanced tissue penetration depth, and minimal levels of autouorescence. [1][2][3][4][5][6] However, most NIR uorophores exhibit uorescence quantum yields (F F ) much lower than those of visible-wavelength uorophores. For instance, indocyanine green (ICG), one of the most commonly used dyes in medical diagnostics, has a low F F of 0.012 in aqueous media, 7 whereas uorescein, a typical green uorescent dye, has a F F of 0.91. 8 The low F F of NIR uorophores signicantly limits their applications in medical imaging. Moreover, the development of NIR uorescent sensors based on uorescence resonance energy transfer or photoinduced electron transfer is more problematic and less efficient than for visible-wavelength uorophores. Therefore, the development of a general approach for NIR uorescence enhancement has become highly important and urgent.
Recently, many reports have shown that localized surface plasmon resonance (LSPR) of noble metals can enhance the uorescence of proximal uorophores when the architectures of metallic nanostructures and uorophores are properly engineered. [9][10][11][12][13][14][15][16][17][18][19][20][21] This phenomenon is named as metal-enhanced uorescence (MEF). 22 Plenty of surface-immobilized metallic nanostructures for MEF have been developed and the advantages of MEF include the increased photostability of uorophores, 23 and improved detection of protein 24 and DNA/ RNA. 25 Although the mechanism of MEF is not fully understood, it is indicated that the strong local electric eld of LSPR causes an increase in the radiative decay rate (k r ) of uorophores near the metallic nanostructure. 12,26 On the other hand, it is well known that noble metal nanostructures such as gold nanoparticles (AuNPs) strongly quench the uorescence of neighboring uorophores. When the distance between the gold surface and the uorophore is very short (<5 nm), or when the electromagnetic resonance coupling between the LSPR and the uorophore emission is very small, uorescence quenching is observed due to the predominance of the nonradiative decay rate (k nr ). [27][28][29][30][31] In principle, the change from uorescence enhancement to uorescence quenching depends on the relative magnitudes of k r and k nr of the uorophore. Thus, the uorescence enhancement caused by AuNPs is a function of both the proximity and the electromagnetic resonance coupling of the uorophores and AuNPs. So far, most studies on MEF have focused on the fabrication of systems that allow for a more selective control of the distance between the uorophore and the gold lm surface, and of the electromagnetic resonance coupling. [32][33][34][35][36][37][38] Nevertheless, stimuli-responsive enhancement of NIR uorescence using noble metal LSPR, especially that based on the tuning of LSPR, has hardly been developed, despite the fact that it shows more potential for signicant imaging applications in medicine.
Thus, we planned to develop MEF-based NIR uorescent sensors by adequately tuning the distance between the metal surface and uorophores in response to analytes. Recently, we reported that b-galactosidase (b-gal) can shi the LSPR absorption spectrum of AuNPs functionalized with specic b-gal substrate ligands from the visible to the NIR region. 39 This result prompted us to exploit enzyme-triggered LSPR changes to obtain MEF. Here we report a b-gal-responsive metal-enhanced NIR uorescence sensor based on functionalized AuNPs.

Results and discussion
The mechanism of the enzyme-responsive NIR uorescence enhancement is illustrated in Scheme 1. The prepared nanoparticles, NGal-NIR-AuNPs and CHO-NIR-AuNPs, were anchored with b-gal-substrate ligands (Lip-NGal) and with aldehyde ligands (Lip-CHO), respectively. NIR-uorophore ligands (Lip-Cy5.5m) and polyethylene glycol ligands (Lip-PEG 400 ) as co-ligands for stabilization in solution were attached to both types of AuNPs. The synthetic routes to the ligands Lip-NGal, Lip-CHO, and Lip-Cy5.5m are summarized in Schemes S1-S3 †. Details of the synthetic procedures, the characterization of the intermediates and ligands, the synthesis of gold nanoparticles, and the fabrication of functionalized gold nanoparticles (NGal-NIR-AuNPs and CHO-NIR-AuNPs) are also provided in the ESI. † The ratio of the attached ligands on the nanoparticle surface was determined by the dye molar extinction coefficient and HPLC analysis (Scheme S4 †). As a result, the ratio of Lip-NGal (or Lip-CHO), Lip-Cy5.5m, and Lip-PEG 400 was 1/0.12/1.26 and 1/0.10/1.19 for the NGal-NIR-AuNPs and CHO-NIR-AuNPs, respectively. The results show slight differences from those of the modication-reaction solutions (Lip-NGal (or Lip-CHO) : Lip-Cy5.5m : Lip-PEG 400 ¼ 1/0.2/1), presumably due to the nature of the ligand pendants (hydrophilic sugar versus hydrophobic dye), lengths of the linkers, and the overall sizes of the ligands. [40][41][42] The uorescence of the surface-attached Cy5.5m was expected to be quenched by AuNPs due to the negligibly small overlap between the LSPR absorption ($520 nm) and the emission ($707 nm) spectra (Fig. S1 †). Aer the enzymatic substrate hydrolysis of NGal-NIR-AuNPs, a selfimmolative elimination reaction via the formation of a quinone methide occurs and the subsequent decarboxylation rapidly produces a primary amine that reacts with an aldehyde on a neighboring CHO-NIR-AuNP to form a Schiff base. As a result, the inter-nanoparticle distance decreases and the LSPR shis towards the NIR region, as Mirkin et al. reported. 43 Such an increase in the spectral overlap between LSPR and emission enhances the electromagnetic resonance coupling, and the uorescence is enhanced.
To verify the b-gal-induced LSPR absorption spectral changes of the mixture of the two different gold nanoparticles (NGal-NIR-AuNPs/CHO-NIR-AuNPs), the evolution of the visible-NIR absorbance spectrum of NGal-NIR-AuNPs/CHO-NIR-AuNPs was measured over time (Fig. 1a). The spectra show a typical LSPR peak at 525 nm, which is 5 nm red-shied with respect to that of the citrate-capped gold nanoparticles due to the ligandexchange effect, 44 and they have a shoulder around 680 nm that arises from the absorbance of Cy5.5m in phosphate buffered saline (PBS) (Fig. S2 †). As expected, the addition of b-gal causes a decrease in the intensity of the absorbance at 525 nm over time, along with a gradual increase in absorption near 700 nm due to the aggregation of gold nanoparticles. In comparison, the absorption spectrum of NGal-NIR-AuNPs does not change in the presence of b-gal (Fig. S3 †). These results indicate that the formation of the Schiff base induces a shi of the LSPR absorption toward the NIR region.
The uorescence of NGal-NIR-AuNPs/CHO-NIR-AuNPs in PBS at 37 C was partially quenched relative to that of Lip-Cy5.5m, because the uorophores are deposited on the Au surface via a moderately long linker (>9 nm). Upon the addition of b-gal, the uorescence was enhanced in a time-dependent manner by a factor of 5.2 at 1 h, 6.7 at 2 h, and 7.8 at 6 h of incubation, respectively ( Fig. 1b and S4 †). These results indicate that the response time was relatively fast initially, and aerwards became slower. This could be satisfactory for in-demand imaging techniques, such as in vitro enzyme detection and early diagnosis. However, we realize the improvement of reaction kinetics is very important for versatile imaging applications as well as improvement of the enhancement factor. Relative to the uorescence of free ligand Lip-Cy5.5m (at the same concentration) in PBS, the uorescence was enhanced by a factor of 2.6 ( Fig. 1c and Scheme S4 †).
The excitation spectra were similarly enhanced (Fig. S5 †). A prolonged incubation time did not afford signicant further spectral changes. The uorescence enhancement was also examined in a biological solution containing BSA (0.1%, w/w) and glycine (0.25%, w/w). Under these conditions, the enhancement factor slightly decreased to 4.6-fold at 1 h and 6.2fold at 6 h, keeping around 90% and 81% enhancement, respectively, compared to that in PBS (Fig. S4 †). The small amount of disruption indicates its practicability in biological media, although complete bioorthogonality is more promising.
Although various ratios of Lip-NGal (or Lip-CHO) and Lip-Cy5.5m (from 10 : 1 to 1 : 1 at the concentration ratio in the modication reaction) were employed to modulate the properties of the nanoparticles, variation of these ratios proved to have no signicant effect on the enhancement factor (Fig. S6 †). The uorescence of NGal-NIR-AuNPs is hardly affected by the presence of b-gal (Fig. S7 †). This strongly indicates that the enzymeresponsive uorescence enhancement of NGal-NIR-AuNPs/ CHO-NIR-AuNPs is induced by an increase in the spectral overlap between the LSPR of gold nanoparticles and the emission of the uorophores, which causes the electromagnetic resonance coupling.
It was difficult to directly monitor the enzymatic hydrolysis and Schiff base formation on the AuNP surface. Alternatively, the enzymatic reaction of Lip-NGal was analyzed. The absorption spectra (Fig. S8 †) and HPLC chart (Fig. S9 †) show the reaction proceeded in a conversion yield of almost 100% within 10 min. The subsequent Schiff base formation with Lip-CHO proceeded relatively slowly, and the reaction yield was almost 50-60% aer 4 h (Fig. S10 †). These results suggest that the slow kinetics of the b-gal-triggered spectral changes of NIR-AuNPs ( Fig. 1a and b) is mainly due to the slow rate of the Schiff base formation.
The morphological variation was conrmed by transmission electron microscopy (TEM) (Fig. 2a). Without b-gal, NGal-NIR-AuNPs/CHO-NIR-AuNPs were quite well dispersed. Upon addition of b-gal, the nanoparticles aggregated. The hydrodynamic size of nanoparticles measured by dynamic light scattering (DLS) showed a gradual increase with b-gal from 25 nm to 137 nm (Fig. 2b). Time-dependent zeta-potential (z) measurements showed a drastic increase from À3.4 to +6.2 mV in 20 min and a subsequent slow decrease to +3.5 mV over time (Fig. S11 †). All data indicate that enzymatic production of primary amines on the nanoparticle surface occurs in the initial stage ($20 min) and the Schiff-base formation accompanied with the consumption of the amines progresses more slowly, in agreement with HPLC analysis (Fig. S9 and S10 †).
Next, to validate the mechanism of enzyme-responsive NIR uorescence enhancement, we synthesized hollow AuNPs (HGNs) with an LSPR peak at 700 nm (matching well with the emission of Cy5.5m), 45 and modied them with Lip-Cy5.5m (Fig. S12 †). The uorescence of these uorophore-modied Lip-Cy5.5m@HGNs was remarkably enhanced by a factor of 4.2 relative to that of Lip-Cy5.5m at the same concentration (Fig. S13 †), indicating that the sufficient spectral overlap between LSPR and NIR emission efficiently induces the uorescence enhancement.
Then, the essential mechanism of responsive MEF was investigated. The increase in scattering efficiency enhances the amount of light absorbed by the proximal uorophores and contributes to the uorescence enhancement. 12,14 To evaluate the light scattering effect, uorescent dye-free NGal-AuNPs/ CHO-AuNPs were prepared on the assumption that they show the same light scattering intensity changes as the dye-containing NGal-NIR-AuNPs/CHO-NIR-AuNPs. In the presence of b-gal, the scattering intensity of NGal-AuNPs/CHO-AuNPs drastically increased in the NIR region (Fig. 3).
The effect of AuNPs aggregation on the uorescence radiation process was also investigated. Fluorescence lifetime measurements revealed the average uorescence lifetimes (s ave ) of Lip-Cy5.5m, and NGal-NIR-AuNPs/CHO-NIR-AuNPs in the absence and presence of b-gal were 835 ps, 274 ps and 202 ps, respectively (Table 1 and S2, and Fig. S14 †). The F F of Lip-Cy5.5m and NGal-NIR-AuNPs/CHO-NIR-AuNPs without and with enzyme were 0.18, 0.057, and 0.36, respectively (Table 1). From the s ave and F F , the kinetic parameters of the radiation process were calculated ( Table 1). The nonradiative kinetic parameter k nr of NGal-NIR-AuNPs/CHO-NIR-AuNPs without enzyme is 3.3 Â 10 9 s À1 , which is a factor of 3.3 higher than that of Lip-Cy5.5m (1.0 Â 10 9 s À1 ). Concomitantly, the radiative kinetic parameter k r is 2.1 Â 10 8 s À1 , which is almost unchanged from that of Lip-Cy5.5m (2.2 Â 10 8 s À1 ). These results indicate that the AuNPs signicantly quench the uorescence of the surface-modied uorophores and that the small enhancement of k r is probably due to negligible coupling between LSPR and emission. Interestingly, k r with enzyme is 1.8 Â 10 9 s À1 , which is an enhancement by a factor of 8.2 and 8.6 relative to Lip-Cy5.5m and NGal-NIR-AuNPs/CHO-NIR-AuNPs without enzyme, respectively. On the other hand, k nr is 3.2 Â 10 9 s À1 , which is slightly lower than that without enzyme and about 3 times larger than that of Lip-Cy5.5m. Overall, NGal-NIR-AuNPs/CHO-NIR-AuNPs show an enhancement of the F F by a factor of approximately 6 in response to b-gal, and it is mostly due to a signicant increase in k r .
Given that k nr depends on the distance and orientation between the uorophore and AuNPs, 46 the slight change in k nr suggests that the enzyme-triggered aggregation induces a very small disruption of the architecture of the nanostructure, such as the distance between the uorophore and the AuNPs surface. The k r of Lip-Cy5.5m on HGNs drastically increases by a factor of 25, and k nr is similar to those of NGal-NIR-AuNPs/CHO-NIR-   AuNPs with and without enzyme ( Table 1). These results indicate that the enzyme-triggered enhancement of NIR uorescence results from an aggregation-induced LSPR shi that increases the electromagnetic resonance coupling, which in turn increases k r , but has little effect on k nr (Fig. 4).

Conclusions
In summary, we developed a novel enzyme-triggered NIR uorescence enhancement nanosystem, which was obtained by the deliberate design and construction of functionalized AuNPs. The NIR uorescence on the AuNPs was enhanced by a factor of 7.8 by b-gal. The increase induced by b-gal in electromagnetic resonance coupling between the AuNPs and the uorophores gave rise to an increase in k r and light scattering intensity, both of which substantially contribute to the uorescence enhancement. This novel strategy for the design of NIR uorescent sensors will be widely applicable for the development of various enzyme-detection systems by the introduction of specic substrates on AuNPs. The new NIR uorescence enhancement principle described here will be useful in the in-demand imaging techniques, such as in vivo enzyme detection and early diagnostic imaging. Although some challenges remain for the in vivo applications, they will be overcome by the integration with some advanced technologies such as stimuli-responsive polymers or active targeting technologies.