2.

Materials and Methods

The method of photoacoustic imaging of FRET was reported in our previous article.²¹ FRET imaging involves nonradiative dipole–dipole interaction that transfers the excited state energy from a donor fluorophore to an acceptor chromophore.²² The excited state fluorophore generally decays through radiative $k_f$ and nonradiative $k_{\mathrm{nr}}$ pathways, generating fluorescence and thermal/photoacoustic emission, respectively. FRET adds another excited state decay mechanism with decay rate $k_t$. The fluorescence quantum yield $Q$ of the donor is determined by the ratio of the rate of fluorescence emission to the sum of all decay rates of the excited state:

With a nonfluorescence acceptor (quencher), the transferred energy eventually decays through the acceptor’s nonradiative pathway, yielding a 100% conversion efficiency of the transferred energy to thermal/photoacoustic emission. Therefore, in the case of a quenching acceptor, FRET manifests itself through a reduction of donor fluorescence and a concomitant increase of the total photoacoustic signal. The FRET efficiency $E$ can be computed from photoacoustic measurement as follows²¹:

The ratio of the fluorescence emission wavelength to the excitation wavelength ${\lambda }_f/{\lambda }_{\mathrm{ex}}$ accounts for the Stokes shift of fluorescence emission. For rhodamine 6G, ${\lambda }_f/({\lambda }_{\mathrm{ex}}{Q}_D)$ is measured to be 1.20 at an excitation wavelength of 523 nm.²¹

The photoacoustic emissions $P_D$ and $P_{\mathrm{DA}}$ in Eq. (4) are attributed to the excited state energy from light absorbed by the donor fluorophore. However, the spectral overlap of the donor and acceptor required for FRET imaging leads to the acceptor bleed-through (ABT) contamination, i.e., the direct excitation of the acceptor at the donor’s excitation wavelength.²³ Therefore, for quantitative treatment of FRET, the ABT background signal needs to be subtracted from the raw photoacoustic amplitude measured from the FRET dye system. The corrected photoacoustic signal $P_{\mathrm{DA}}$ at ${\lambda }_{\mathrm{ex}}$ is obtained as follows:

Figure 1 is a schematic of the photoacoustic tomography system whose design and performance have been detailed previously.^24,25 The photoacoustic tomography system employs a 512-element full-ring transducer array. The transducer array has a 5-MHz central frequency (80% bandwidth) and a 50-mm ring diameter. A tunable OPO laser (Newport, basiScan 280) with a 12-ns pulse duration and a 10-Hz pulse repetition rate was the illumination source. The laser beam was homogenized using an optical diffuser, and the maximum light intensity at the surface of the sample was approximately $10\mathrm{mJ}/{\mathrm{cm}}^2$. Within the imaging plane, the system provided 0.10 to 0.25 mm tangential resolution and relatively uniform 0.10-mm radial resolution.²⁶ The raw data from each element was first Wiener deconvolved to account for the ultrasonic transducer’s impulse response. A back-projection algorithm was then implemented using matlab to reconstruct photoacoustic images.

Fig. 1

Schematic of the photoacoustic tomography system.

We used solutions containing a controlled amount of donor rhodamine 6G and acceptor DQOCI to analyze FRET efficiencies. Seven stock ethanol solutions were prepared with concentrations of donor rhodamine 6G and acceptor DQOCI as tabulated in Fig. 2(a). Fluorescence imaging of the sample solutions sealed in glass tubes was performed.²⁷ Figure 2(b) shows the fluorescence emission from solution containing only donor rhodamine 6G and from mixtures containing both donor rhodamine 6G and acceptor DQOCI. It can be seen that more acceptor DQOCI quenched rhodamine 6G fluorescence more effectively. The fluorescence intensity acquired at 523 nm from the pure donor rhodamine 6G (tube no. 1) and the mixture solutions with different acceptor concentrations of 0.125, 0.25, and 0.5 mM (tube nos. 3, 5, and 7) are plotted in Fig. 2(c). From the ratios of the diminished and unperturbed fluorescence intensities, the absolute FRET efficiencies were quantified using Eq. (3) and are shown in Fig. 2(d).

Fig. 2

(a) Photograph of the tube phantom and tabulation of the donor (rhodamine 6G or R6G) and acceptor (DQOCI) concentrations. (b) Fluorescence microscopic image of the sample solutions acquired at 523 nm. (c) Fluorescence signals versus acceptor concentration, showing the FRET effect. (d) FRET efficiency map acquired using fluorescence microscopy.

3.

Results

To test photoacoustic tomography of FRET in biological tissue, the sample solutions were sequentially placed in a 0.3-mm diameter silastic tube beneath 1-cm thickness of chicken breast tissue. Figure 3(a) shows photoacoustic tomography images of the sample solutions acquired at 523 and 630 nm. The photoacoustic signals at 630 nm are dominated by the acceptor DQOCI absorption, as is evident by the equal photoacoustic amplitudes from solutions containing only DQOCI and from solutions containing both rhodamine 6G and DQOCI. The photoacoustic signals at 523 nm show the raw photoacoustic signals generated by both the donor rhodamine 6G and the acceptor DQOCI absorptions. After using Eq. (5) to subtract the acceptor DQOCI bleed-through photoacoustic background (tube nos. 2, 4, and 6) from the raw photoacoustic signals of the FRET pair (tube nos. 3, 5, and 7), the photoacoustic amplitude resulting from the donor rhodamine 6G excitation in the absence (tube no. 1) and the presence (tube nos. 3, 5, and 7) of different concentrations of acceptor DQOCI is plotted in Fig. 3(b). Here, the quenching of donor rhodamine 6G fluorescence leads to an enhancement of the photoacoustic signal. Figure 3(c) shows the FRET efficiency map of the tissue phantom calculated from the relative increase of photoacoustic amplitude using Eq. (4). The photoacoustic and fluorescence measurements of FRET efficiencies agree with a correlation coefficient of 0.93. The results demonstrate the potential for in vivo FRET imaging using photoacoustic tomography.

Fig. 3

(a) Photoacoustic tomography image of the sample solutions acquired at 523 and 630 nm below a 1-cm thickness of chicken breast tissue. (b) Photoacoustic signals after acceptor bleed-through correction versus acceptor concentration, showing the Förster resonance energy transfer (FRET) effect. (c) FRET efficiency map acquired using photoacoustic tomography.

The signal-to-noise ratio (SNR, the ratio of the signal amplitude to the standard deviation of the background or $20{\log }_{10}$ of the ratio in dB) of the photoacoustic FRET tomography was investigated by overlaying a 25 μM rhodamine 6G donor solution tube with a wedge of chicken breast tissue. The imaging depth was incremented from 0.5 to 1.0 cm [Fig. 4(a)]. Figure 4(a) shows a photoacoustic image of the tube covered with the tissue wedge. Figure 4(b) plots the SNR versus the depth of the tube. It can be seen that the SNR in decibels decreased linearly with increasing depth. The fitted penetration depth for $1/e$ decay at 523 nm in chicken breast tissue was 0.4 cm, and the extrapolated noise equivalent (i.e., $\mathrm{SNR}=0\mathrm{dB}$) penetration depth was 1.3 cm. With the laser fluence of $10\mathrm{mJ}/{\mathrm{cm}}^2$ (half of the ANSI safety limit²⁸), a 1-cm imaging depth was achieved with an SNR of 7 dB. Increasing the laser fluence can further improve the noise-equivalent penetration depth.

Fig. 4

Photoacoustic tomography of the donor solution in a tube acquired at 523 nm beneath a wedge of chicken breast tissue with a varying thickness from 0.5 to 1.0 cm. (a) Image of the tube at depths from 0.5 to 1.0 cm as labeled. (b) Photoacoustic signal-to-noise ratio (SNR) versus the depth of the tube.

In conclusion, photoacoustic tomography has been used to image FRET efficiencies beneath a 1-cm thickness of chicken breast tissue. Based on the relative increase in photoacoustic signals, absolute FRET efficiencies can be quantified. Compared to fluorescence-based methods, photoacoustic tomography enables deep penetration imaging in biological tissue at high ultrasonic resolution. This advantage should facilitate the application of FRET imaging in in vivo animal studies.