2.1.

Materials

The deuterated docosanol was obtained from GlaxoSmithKline. The powder form of the drug was used to acquire a standard spectrum using spontaneous Raman scattering and CARS spectroscopy. Dimethyl sulfoxide (DMSO), the solvent for the drug, was purchased from Sigma-Aldrich.

2.2.

Cell Culture

Both human dermal fibroblasts and human epidermal keratinocytes were purchased from ZenBio, Inc. Fibroblasts were cultured in dermal fibroblast culture medium (DF-1, ZenBio). Keratinocytes were cultured in adult keratinocyte growth medium (KM-2, ZenBio). Both cell cultures were grown in cell culture flasks at 37°C with 5% ${\mathrm{CO}}_2$ and passaged onto Petri dishes two days before imaging.

2.3.

Coherent Anti-Stokes Raman Scattering Imaging

A previously described, custom-made, fiber supercontinuum-based system (Fig. 1) was used for CARS imaging in the silent spectral region (2000 to $2500{\mathrm{cm}}^{-1}$) of docosanol in living cells.^22,23 Briefly, pulses (1041 nm, 220 fs, 80 MHz) from a Yb:KYM laser (femtoTRAIN IC model-Z, High Q Laser) were sent into a 21-cm long photonic crystal fiber (NL-1050-NEG-PM-FUD, NKT Photonics) to generate a widely coherent supercontinuum from 780 to 1320 nm. The supercontinuum output was separated into the pump beam (780 to 880 nm) and the Stokes beam (900 to 1320 nm) by a dichroic mirror (DMLP900, Thorlabs). The targeted wave number of this spectral-focusing-based CARS imaging system was determined by the temporal delay between the pump and the Stokes beam and was controlled by a motorized stage in the path of the Stokes beam. The two beams were then recombined by a second dichroic mirror, sent into a commercial microscope (BX61WI, Olympus), and focused by a superapochromat objective (UPLSAPO 60XW/IF, N.A. 1.20, Olympus). The optical power at the sample was 8 mW for the pump beam and 20 mW for the Stokes beam. The forward-directed CARS signals were collected by a second objective (LUMPlanFl/IR 60XW, N.A. 0.9, Olympus). The mapping relationship between the temporal delay and the true wave number was determined and calibrated using the spectral peaks of acetonitrile and the drug powder. The spectral resolution was $\sim 20$ to $30{\mathrm{cm}}^{-1}$. The pixel dwelling time in each CARS image was 1 ms.

Fig. 1

A schematic diagram of the CARS imaging system. DC, dichroic mirror; FW, filter wheel; Gl, glass; HWP, half-wave plate; IS, isolator; KYW, potassium yttrium tungstate; L, lens; NDF, neutral density filter; OBJ, objective; PCF, photonic crystal fiber; PM, parabolic mirror; PMT, photomultiplier tube; and Yb, ytterbium.

2.4.

Coherent Anti-Stokes Raman Scattering Imaging of Drug-Treated Cells

The deuterated drug was dissolved in DMSO at the concentration of 10 mM. For the drug treatment studies, $30\mu \mathrm{L}$ of the drug solution was added to a cell culture dish along with 3 mL of cell culture media (dermal fibroblast culture medium and adult keratinocyte growth medium for fibroblast and keratinocyte, respectively), resulting in a final concentration of the drug at $\sim 100\mu \mathrm{M}$. After 24 h of incubation, the cell media was replaced with fresh media and the cells were immediately imaged.

3.1.

Spontaneous and Coherent Raman Spectroscopy of the Drug Molecules

The spontaneous Raman spectrum of the drug molecule was obtained using a commercial Raman microscope (Horiba LabRAM HR 3D, Horiba). As shown in Fig. 2(a), several characteristic $\mathrm{C}─\mathrm{D}$ vibrational peaks appear around $2100{\mathrm{cm}}^{-1}$ in the cellular silent region (2000 to $2300{\mathrm{cm}}^{-1}$), which is attributed to the replacement of multiple $\mathrm{C}─\mathrm{H}$ bonds with $\mathrm{C}─\mathrm{D}$ bonds in the deuterated drug.^24,25 In particular, the unique strong peak around $2100{\mathrm{cm}}^{-1}$ was used as the spectral region of interest in this study. To examine the possible distortion of the CARS spectrum, the drug molecule was measured by CARS spectroscopy in the silent region to compare with the spontaneous Raman spectrum [Fig. 2(b)]. The high degree of similarity between these two spectra demonstrated the feasibility of this experiment. The noticeably reduced spectral resolution can be attributed to the limitation of the spectral-focusing-based CARS setup.

Fig. 2

Drug vibrational spectrum by (a) spontaneous Raman scattering spectroscopy and (b) CARS spectroscopy. The temporal delay of $0\mu \mathrm{m}$ in the CARS spectrum corresponds to the wave number of $2100{\mathrm{cm}}^{-1}$ in the spontaneous Raman spectrum, which was used as the drug-resonant peak in this study. The temporal delay of $100\mu \mathrm{m}$ in the CARS spectrum corresponds to the wave number of $2250{\mathrm{cm}}^{-1}$ in the spontaneous Raman spectrum, which was used as the nonresonant region in this study.

3.2.

Drug Accumulation in Fibroblasts

Drug-treated fibroblasts were imaged at two frequencies, $2100{\mathrm{cm}}^{-1}$ for the resonant frequency and $2250{\mathrm{cm}}^{-1}$ for the nonresonant frequency. Figure 3 shows the representative (a) drug-resonant CARS image, (b) the drug nonresonant CARS image, and (c) the composite image. Bright, densely packed punctate regions were observed inside the cells. Representative CARS spectra collected from pixels of interest within the CARS image are shown in Fig. 3(d). Compared with the spectrum acquired from the background of the cell body, the spectrum of the bright punctate regions has a unique, sharp, and strong peak at $2100{\mathrm{cm}}^{-1}$. Furthermore, the spectra representing the bright regions strongly resemble the drug spectra acquired from the spontaneous Raman scattering and CARS spectroscopy (Fig. 2). Interestingly, the drug areas appear much darker than the surroundings in the nonresonant image [Fig. 3(b)]. This can be explained by the lack of water component in the highly concentrated drug area, which results in the sharp decrease of nonresonant CARS signal compared with the nondrug areas. These observations support the hypothesis that the punctate regions within the fibroblasts are indeed the result of highly concentrated drug.

Fig. 3

Hyperspectral CARS imaging shows the accumulation of docosanol in living fibroblasts. Representative CARS image at (a) $2100{\mathrm{cm}}^{-1}$ and at (b) $2250{\mathrm{cm}}^{-1}$ of drug-treated living fibroblasts. (c) Spatial overlap of the drug-resonant CARS image (green) and the drug-nonresonant image (red). In the resulting composite image, the drug-resonant regions appear red, whereas the nonresonant regions appear yellow. (d) CARS spectra of the bright regions in the cell [red line, corresponding to the red arrow (right arrow)/small yellow box in (c)] and the cell body background [blue line, corresponding to the blue arrow (left arrow)/small blue box in (c)]. Image field of view: $128\times 128{\mu \mathrm{m}}^2$.

3.3.

Drug Accumulation in Keratinocytes

Following the demonstration of drug accumulation within living fibroblasts, a similar study was conducted with human epidermal keratinocytes, which are the targeted cells for docosanol treatment. Figure 4 shows (a) the drug-resonant CARS image, (b) the nonresonant CARS image, and (c) the composite image of the two frequencies. Interestingly, a bright ring was observed around the periphery of the keratinocytes, unlike the pattern of punctate regions within the fibroblasts. The CARS spectra, shown in Fig. 4(d), indicate that the bright ring is the result of drug accumulation in the keratinocytes. This observation is highly intriguing because this particular distribution pattern resonates with the predominant mechanism of action for docosanol, which is reported to be an interaction with the cell membrane to inhibit the fusion between the cell plasma membrane and the HSV envelope.^10,11 Further studies are needed to investigate whether this distribution of docosanol within the cell membrane is directly responsible for its mechanism of inhibiting viral entry.

Fig. 4

Hyperspectral CARS imaging shows the accumulation of docosanol in the living keratinocyte. Representative CARS image at (a) $2100{\mathrm{cm}}^{-1}$ and at (b) $2250{\mathrm{cm}}^{-1}$ of drug-treated living keratinocytes. (c) Spatial overlap of the drug-resonant CARS image (green) and the drug-nonresonant image (red). In the resulting composite image, drug-resonant regions appear red whereas the nonresonant regions appear yellow. (d) CARS spectra of the bright pixels in the cell [red line, corresponding to the red arrow (left arrow)/small yellow box in (c)] and the cell body background [blue line, corresponding to the blue arrow (right arrow)/small blue box in (c)]. Image field of view: $50\times 50{\mu \mathrm{m}}^2$.

In summary, our study visualizes the intracellular distribution of docosanol in fibroblasts and keratinocytes, and reveals cell-specific distribution patterns that may indicate specific mechanisms of drug action and, thus, have important therapeutic implications. These results suggest that docosanol docking at the periphery of keratinocytes might be one of the earlier steps for the inhibition of fusion at the plasma membrane during docosanol treatment. However, the precise mechanism of action for docosanol interference remains unclear. To fully understand the process of docosanol interference, more in-depth experiments on infected and subsequently treated keratinocytes will be essential for future studies. We wish to emphasize that the goal of this study is to present initial results demonstrating how a CARS imaging-based method can be used to characterize and observe the intracellular behavior of docosanol in living cells. The investigation model as well as the preliminary results presented in this study demonstrate the potential impact of the proposed methodology on improving the understanding of docosanol treatment.