Near infrared bioluminescence resonance energy transfer from firefly luciferase—quantum dot bionanoconjugates

The hydrophobic QD(800) stock was purchased from Invitrogen (Life Sciences), and used as received. L-histidine (His, >99.8%), sodium borohydride (NaBH₄, >96%), sodium tetraborate (99.5%), boric acid (>99.5%), gly–gly (BioUltra, ≥99.5%), toluene (≥99.5%), chloroform (>99.8%), methanol (>99.8%), acetone (99.5%) were purchased from Sigma Aldrich. Ultrapure water (18.2 MΩ) was from a Sartorius Stedim Arium 61316 reverse osmosis unit combined with an Arium 611DI polishing unit. The Mg-ATP (bacterial source) was purchased from Sigma-Aldrich, and restriction endonucleases from New England Biolabs (Beverly, MA). Firefly luciferin (LH₂) was a generous gift made to the Branchini lab from Promega (Madison, WI).

2.2.1. PPy expression

2.2.2. Histidine-mediated QD phase transfer and characterization

2.2.3. PPyGRTS-QR conjugations and characterization

2.2.4. BRET measurement and analysis

In a typical BRET experiment, a mixture of 100 μL of 91 μM LH₂ (firefly luciferin) and 30 μL of 8.66 mM Mg-ATP in 25 mM gly–gly buffer (pH 7.8) is quickly added to the PPyGRTS-QR(800) conjugate solution in a 96-well plate and bioluminescence emission is immediately collected. The bioluminescence and BRET were collected on a Varian Cary-Eclipse spectrophotometer in bioluminescence/chemiluminescence mode using a 96-well plate reading accessory. The spectra are corrected for detector sensitivity by comparison of fluorescence standard emission intensities (450–950 nm) with the corrected detector on the Fluoromax-4 spectrophotometer. White 96-well plates were employed, with volumes ranging from 50–200 μL. Bioluminescence spectra were collected every 15 s for 7.5 min. The presented BRET results are the average of the first five spectra collected over 1.5 min after addition of LH₂. Finally, the BRET efficiencies of the systems were calculated as BR, which is defined as the ratio of peak area of the acceptor and donor emission respectively. Peak area was calculated by spectral deconvolution of each spectrum using the data analysis package in Igor Pro (Wavemetrics).

Theoretical assessment of BRET efficiency was conducted in the identical manner to FRET. In FRET, the Förster distance (R₀) is calculated using equation (1) [2]:

$$\begin{eqnarray}&&{{R}_{0}}\;=\;9.78\;\times \;{{10}^{3}}{{\left[ {{\kappa }^{2}}{{n}^{-4}}{{Q}_{D}}J\left( \lambda \right) \right]}^{1/6}},\end{eqnarray} \tag{ 1 }$$

where η is the refractive index of the medium (η_D = 1.33), k_p is the polarization parameter (κ = 2/3), Q_D is the donor quantum yield QY(GRTS)≈32%, and J is the spectral overlap integral. The J value can be calculated using equation (2):

$$\begin{eqnarray}&&J=\int {{f}_{D}}\left( \lambda \right){{\varepsilon }_{A}}\left( \lambda \right){{\lambda }^{4}}{\rm d}\lambda ,\end{eqnarray} \tag{ 2 }$$

where λ is the defined wavelengths of the donor–acceptor spectral overlap (λ = 450–650 nm), and f_D(λ) is the integrated donor emission and _A(λ) represents the integrated acceptor absorption using the acceptor extinction coefficient (_A = 2.0 × 10⁶ M⁻¹cm⁻¹). The values of R₀ = 10.6 nm, and 2.37 × 10⁻¹¹ M⁻¹ cm³ were calculated using equations (1) and (2) as well as the software PhotoChemCAD for QD(800). Using the R₀ values calculated above, the FRET efficiency, E, was calculated using equation (3) [2]:

$$\begin{eqnarray}&&E\;=\;R_{0}^{6}/\left( R_{0}^{6}+{{r}^{6}} \right),\end{eqnarray} \tag{ 3 }$$

where r is the distance between the active site of the PpyGRTS and the QR dipole.

The general QD functionalization strategy is outlined in scheme 1(a) and (b), which begins using commercially purchased alkyl-terminated hydrophobic QD(800) nIR QDs. Firstly, the QD(800) are phase transferred from chloroform to aqueous buffers using the L-histidine (His) mediated phase transfer route recently developed in our lab [21]. The His-capping is rather weakly bound [21], which facilitates additional monolayer exchange [21], ssDNA attachment [23, 28], or 6xHis tagged protein binding [14, 22–24]. Recently a His rich co-polymer was used to polymer-wrap QDs for improved stability for bioimaging [29]. After phase transfer, the His-QD(800) had a PL emission at ≈788 nm (figure 1(a)), a QY of 47%, and long, single exponential lifetime of τ = 77.8 ± 1.5 ns (figure 1(b)). Next, the His-capped QD(800) are incubated with PPyGRTS at increasing loading ratios, L = [PPyGRTS]/[QD(800)] = 0.5–10. The PPyGRTS were coordinated to the QD interface using the 6xHis tag [14, 28, 30], allowing for short donor–acceptor distances. If required, excess PPyGRTS could be removed by addition of nickel loaded colloids to absorb the 6xHis tag, followed by removal through centrifugation.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The PPyGRTS + QD(800) conjugates were then characterized. Figure 2 shows a TEM micrograph of the conjugates at L = 5 after negative staining with phosphotungstic acid (a), and before staining controls (b). The bright halos corresponding to the PPyGRTS layer can be observed surrounding the QD. Further analysis of the TEM resulted in a calculated QD(800) diameter of d = 4.0 ± 0.5 nm.

Zoom In Zoom Out Reset image size

Next, conjugation was characterized via TGA. Here, mass loss is directly related to the lightweight His monolayer, and the heavier PPyGRTS (MW ≈ 60 kDa). Figure 3 a shows a representative TGA profile for the QD(800) before (i), and after (ii) phase transfer, where a mass loss of 51% and 25% was measured, respectively. These results provide conformation of monolayer exchange and subsequent phase transfer, since the His has much lower MW compared to typical hydrophobic capping ligands like TOPO, etc. For the PPyGRTS + QD(800) conjugates, a mass loss increase was observed as a function of increased L. Specifically, as L increased from 11–43, a successive increase from 66 to 88% was observed (iii–v), indicating increased weight loss due to additional proteins residing on each QD.

Zoom In Zoom Out Reset image size

The conjugates were also studied via FTIR (figures 3(b) and (c)). Compared to the His-QD(800) (i), the PpyGRTS + QD(800) conjugates show sharp vibrations at ≈ 1627, 1551 cm⁻¹ for L = 10 (ii), which are characteristics of amide-I and amide-II modes originating from the protein [31, 32]. Amide-III modes are also present at ≈ 1290 cm⁻¹. While the amide-I modes are due to C=O stretching, amide-II and III are attributed to CN stretch and NH bend within the protein. There are also sharp vibrations at 3177–3184 cm⁻¹, indicating the presence of amide-A and B modes, which arise from NH stretching.

We next studied the emission properties of the PPyGRTS + QD(800) conjugates. Upon addition of the substrate luciferin (LH₂) to a solution of the conjugates, we observed little visible light associated with the PPyGRTS bioluminescence, suggesting the efficient energy transfer into the nIR. Figure 4 quantifies this energy transfer, and shows a typical set of BRET emission results at L = 1 (a), 2 (b), 5 (c), and 10 (d). We chose this range to limit the potential for free PPyGRTS in solution, since each QD800 can ideally pack 11–13 proteins, based on geometric models. Features of this data are the donor emission red-shifted by ∼250 nm, the absence of other external excitation, and the substrate based activation. The BRET efficiency was quantified by calculating the so-called BR [6, 10], in which the integrated emission of the acceptor (QD(800)) is divided by the integrated emission of the donor (PPyGRTS). For example, at L = 1 a BR of 3.9 was calculated. Also shown is BRET for conjugates at L = 2.0 (b), 5.0 (c), and 10 (e), in which we observe that the BR slightly fluctuates between 4 ∼ 5 for each, as shown in the histogram plotting BR versus L in the inset of figure 4(e). The observation of high BR at low loading ratios, L = 1.0–2.0, has been reported before [6, 14], and is a unique phenomenon for BRET systems compared to FRET analogs. Another characteristic of these systems is the relatively high noise value at L = 1–2 (a) and (b) compared to L = 5–10 (c) and (d). We speculate that this is due to the possibility that at low L some QD(800) are unconjugated due to stochastic binding at those ratios, since any unbound QD(800) will remain dark, which is reflected in the low intensity of emission (figure 4(a)). At higher L, a more equal population of QD(800) will be conjugated, allowing for higher numbers of QD(800) emitting, and higher intensity values (vide infra). The exact mechanism for the decrease BR at higher L is still under investigation. Recently, researchers showed that increased coverage of cytochrome C on a gold nanoparticle decreased the particles ability to act as an plasmonic quencher, revealing that a energy transfer gating process may be occurring, in which the dielectric change provided by the protein layer disrupts energy transfer [33]. In our system, this increased coverage may screen the bound LH₂ environment, which is the source of the bioluminescence.

Zoom In Zoom Out Reset image size

This phenomenon may also arise due to a small degree of uncertainty (∼20%) in regards to the molar concentrations of QD(800) that were estimated here based on supplier estimates of extinction coefficients (_QD(800) ≈ 2.0 × 10⁶ M⁻¹ cm⁻¹). It is important to note that the presence of emission at 550 nm does not necessarily correspond to free PPyGRTS, but instead, incomplete energy transfer. Moreover, multiple repeat experiments resulted in similar BRET efficiencies (figure S1), and control experiments without 6xHis tagged PPyGRTS resulted in no BRET. Finally, the approximate excess loading ratio was found to be L > 10, above which free PPyGRTS remain unbound in solution. Figure S2 shows the BRET spectra at L = 20 before and after incubating with Ni-sepharose beads, which bind and remove the free His tagged PPyGRTS. A BR increase from 2 to 4 was observed after purification, suggesting the removal of free proteins at those L values.

To better understand these results, the spectroscopic properties and FRET parameters were compared, as shown in figure 5. Since the QD(800) has a first absorption maximum of ∼790 nm, it has a significantly broad absorption profile that covers the visible spectrum and has a high extinction coefficient near the PPyGRTS emission maximum ( ≈ 2.0 × 10⁶ M⁻¹ cm⁻¹ @ 550 nm). As a result, the calculated spectral overlap integral, J, is 2.37 × 10⁻¹¹ M⁻¹ cm³, which in combination with the PPyGRTS donor [14], results in a Förster distance (R₀) of 10.6 nm [2]. If the QD(800) are treated as spherical, the maximum distance between the QD(800) core, to the active site on the PPyGRTS, is r ∼ 5.5 nm, a value well below the R₀ (figure 5(b)), and one that theoretically should result in a ≈97% efficiency under normal spherical dipole orientations. Compared to other BRET nanosystems [5, 6, 10, 11, 17–19], the present BR values are 2 ∼ 3x higher. This efficiency increase is attributed largely to the high QY of the PPyGRTS, and its direct attachment to the QD interface (i.e. r < R₀). One likely limit to even higher BRET efficiency is that compared to traditional CdSe/ZnS QDs (τ ≈ 10–20 ns), the longer lifetimes of the QD(800) (τ ≈ 77 ns), may also be introducing a bottle-neck in the energy transfer process.

Zoom In Zoom Out Reset image size

Because of the high BRs of ≈5.0, and emission at 788 nm, this BRET nanosystem is easily observed via modest nIR imaging. Shown in figure 6 are digital images collected from a home-built setup using commercially available night vision goggles. For example, figure 6(a) shows a digital image of a 96-well plate containing the PPyGRTS (left) and PPyGRTS + QD(800) (right) systems collected using a traditional digital camera. Only a faint PPyGRTS spot can be observed, and no light is detected from the PPyGRTS-QD(800) conjugate due to BRET. The identical system can be observed easily using night vision, however. Note that the false green colors observed are due to the goggles and are not the colors of the sample. Figure 6(b) is imaged using a nIR camera, and the PpyGRTS + QD800 is now clearly visible (right). This system is fully scalable, as is illustrated by figure S3, in which a large PDMS pattern is filled with PPyGRTS + QD(800) and imaged.

Zoom In Zoom Out Reset image size