Design and Synthesis of PA-HD-MGL and PA-HD-FAAH

The first criterion we considered when designing PA-HD-MGL and PA-HD-FAAH was to obtain excellent substrate specificity for their respective enzyme targets. Limiting cross-reactivity is essential for in vivo detection of MGL and FAAH because both enzymes are present in the GI tract. We hypothesized this could be achieved by appending AA to a PA reporter via an ester bond (MGL) or an amide linkage (FAAH). The second major criterion is the probes must be minimally perturbing to the system under investigation. The selection of a strong PA emitter was therefore prioritized since doing so would allow us to dose at notably lower concentrations. To this end, we proposed an initial design where AA was appended to PA-HD via the linkages discussed above. PA-HD is a newly developed congener of the classic hemicyanine dye (HD) platform¹⁹ but differs in that it has been optimized for PA imaging (4.8-fold higher sensitivity at the same dye concentration).²⁰ However, the resulting first-generation probes we prepared self-assembled into unreactive micelle-like particles due to the positively charge dye and hydrophobic tail components (Figure S1). We hypothesized that by replacing the N-ethyl group (cLogP=16.3) with a N-propyl sulfonate moiety (cLogP=9.4), reactivity would be restored because the inclusion of negatively charged arms should disfavor aggregation.

Briefly, the synthesis of both probes began from cyanine 1, which was prepared from the indolium sulfonate precursor in 60 % yield. PA-HDs 2 and 3 were obtained from the corresponding retro-Knoevenagel reactions using 4-chloro-3-methoxybenzenethiol and 3-aminothiophenol, followed by acylation with arachidonoyl chloride to obtain PA-HD-MGL and PA-HD-FAAH in 80 % and 78 % yield, respectively (Figure 1). Of note, the ortho-chloro substituent featured in PA-HD-MGL was strategically installed to attenuate spontaneous hydrolysis of the ester. Previous work in similar dye systems have shown this modification led to enhance stability which was due to donation of the halogen lone pair electrons to the π* orbital of the adjacent carbonyl.²¹

Photophysical Characterization and in vitro Evaluation

Following synthesis, we sought to determine the photophysical properties of PA-HD-MGL, PA-HD-FAAH, and their corresponding turned over products (2 and 3) to ensure each dye exhibited the desired properties for in vivo PA imaging. We calculated the PA brightness factor (PABF) by multiplying the extinction coefficient at the λ_abs by the term (one minus the fluorescent quantum yield (Φ_F)).²² We previously found that PABF values >10⁴ were necessary to reduce the amount of dye administered to animals while still maintaining a high PA output. In the case of 2 and 3, the large experimental PABFs (≈10⁵ range) can be attributed to strategic selection of the PA-HD platform which is highly efficient at harvesting incident light. Additionally, we found the λ_PA (wavelength of maximum PA signal output) for 2 and 3 to be 740 nm and 730 nm, respectively (Figure 2a). Excitation of PA-HD-MGL and PA-HD-FAAH at these wavelengths result in the negligible production of soundwaves, and this corresponds to theoretical signal enhancements of 7.29-fold and 4.15-fold, respectively. Of note, the PA spectral traces shown in Figure 2b and 2c were obtained using densely formulated phantoms comprised of milk and agarose at high concentrations as a proxy for deep-tissue imaging.²³

Prior to evaluating the responsiveness of PA-HD-MGL and PA-HD-FAAH toward their respective target enzymes, it was necessary to assess the hydrolytic stability of each probe under relevant in vivo conditions. Specifically, the pH of the gastrointestinal tract begins at 6.0 (duodenum) and gradually increases to 7.4 (terminal ileum). Upon exiting the small intestines, a second pH gradient is associated with the large intestines that range from 5.7 (caecum) to 6.7 (rectum).²⁴ Our analyses indicate there is minimal change in the signal intensity in this pH range and furthermore, extended incubation up to one hour (a typical in vivo imaging experiment is only 30 minutes) shows 0.30-fold decrease (Figure S2). We attribute the exceptional stability of this particular ester moiety to the inclusion of the ortho-chloro group. Likewise, the stability of PA-HD-FAAH was extraordinary even under basic conditions (Figure S3). Having determined the requisite baseline readings, we treated each probe with purified human MGL and FAAH (Figure S4 and S5). To our delight, MGL only activated PA-HD-MGL and FAAH selectively catalyzed the turnover of PA-HD-FAAH. We confirmed the expected products were being generated via mass spectrometric analysis of probe samples treated with the appropriate enzyme (Figure S25 and S26). We attribute the absence of cross-reactivity between PA-HD-MGL and FAAH to enhanced stability to our ortho-chloro ester trigger. Indeed, FAAH has been shown to hydrolyze a variety of amide and ester substrates with nearly identical rates..²⁵ Beyond MGL and FAAH, we also subjected our probes to other enzymes including members of the ECS that recognizes AA as a substrate (cyclooxygenase-1, cyclooxygenase-2, and arachidonate 5-lipoxygenase),²⁶ various cytochrome p450s (via rat liver microsomes), ester bond cleaving enzymes (esterase and lipase), as well as catalase (Figure 2d and 2e). Under no circumstances did we observe notable off-target reactivity greater than 10 %. Lastly, we evaluated whether PA-HD-MGL and PA-HD-FAAH would turnover in the presence of biologically relevant analytes (Figure S6). No hydrolysis of the ester or amide bonds were observed in the presence of metal ions exhibiting Lewis acidic properties (i.e., iron and copper). We also found that PA-HD-MGL and PA-HD-FAAH were stable to nucleophiles such as thiols. Because the dye scaffold is susceptible to decomposition by reactive oxygen and nitrogen species, we also examined potential reactivity but found no change in the absorbance at biologically relevant concentrations of hydroxyl radical, superoxide, peroxynitrite, and nitric oxide.

Detection of MGL and FAAH Activity in Cell Cultures

Prior to utilizing PA-HD-MGL and PA-HD-FAAH in vivo, it is critical to demonstrate probe activation occurs selectively by the target enzyme in cell culture. A careful survey of the literature revealed that LNCaP cells (a human prostate adenocarcinoma cell line) constitutively express both MGL and FAAH at appreciable levels.²⁷ Moreover, we found the application of PA-HD-MGL, PA-HD-FAAH, and their corresponding turned over products at experimentally relevant concentrations and incubation periods did not significantly reduce LNCaP viability (Figure S7–10). Thus, we hypothesized treatment of this cell line with appropriate small-molecule inhibitors would allow us to determine the molecular target responsible for activating each probe. Specifically, JZL184,²⁸ a highly selective irreversible inhibitor of MGL, and PF-3845,²⁹ a potent FAAH covalent inhibitor that modifies the active site serine residue, were chosen for our studies. We began by pre-treating LNCaP cells with either a vehicle control or a solution of JZL184, prior to staining with PA-HD-MGL (Figure 3a and 3b). Our results demonstrate that blocking MGL activity using JZL184 led to a statistically significant decrease in fluorescence of 0.72±0.08-fold (mean±SEM) (Figure 3g). Importantly, we found that applying the FAAH inhibitor instead of JZL184 did not impact PA-HD-MGL activation, which provides further support that the signal enhancement is due to MGL activity (Figure 3c and S11). A similar set of experiments were performed to assess the responsiveness of PA-HD-FAAH in LNCaP cells. As before, cell cultures were treated with either a vehicle, PF-3845, or the MGL inhibitor. Consistent with our in vitro testing, attenuation of probe activation was only observed when FAAH activity was blocked via inhibition (0.75±0.06-fold difference (mean±SEM)) (Figure 3d–f, 3h and S12). Lastly, in addition to live cell imaging, we performed a series of cell lysate experiments. LNCaP cells were lysed via sonication, and serial dilutions of these lysates were generated and treated with either PA-HD-MGL or PA-HD-FAAH for a total of 30 and 90 minutes, respectively (Figure S13 and S14). In both instances, probe activation was shown to be dose-dependent where the most pronounced turnover was associated with samples containing the highest cell count.

In vivo PA Imaging of MGL and FAAH Activity in a Murine Model of Obesity

Next, we established a murine model of obesity by feeding BALB/c mice (female, ≈six weeks old) research-grade chow where fat accounts for 60 % of the caloric profile for 12 weeks. Of note, female mice were selected to be consistent with a previous study conducted by our group to assess diet-induced inflammation of the mammary fat pad.^11t A second cohort of animals were fed a control diet where the calories from fat were standardized to only 10 %. Although the chows are isocaloric, we found that by allowing the animals to feed freely, subjects on a high-fat diet became obese, gaining an average of 15.5 % body weight over this period (Figure 4a). Since obesity is known to trigger an inflammatory response in the GI tract, we hypothesized the application of PA-HD-MGL and PA-HD-FAAH would allow us to discern how and to what extent this process involves the ECS. To facilitate in vivo PA imaging of the GI tract using our probes, which is a deep-seated region of the body, we turned to the use of an advanced MSOT (multispectral optoacoustic (photoacoustic) tomography) system because it is possible to visualize the entire animal and present the data in lateral and cross-sectional views (Figure 4b and 4c). Moreover, the PA signals from each probe (and their products) can be isolated from endogenous PA-active pigments such as hemoglobin found in blood and from each other via spectral unmixing.

PA-HD-MGL was then administered systemically via intraperitoneal injection to both groups of mice and the corresponding PA images were acquired at the 30-minute time point. Analysis of the spectrally unmixed images in lateral and cross-sectional views indicated PA signals corresponding to the turned over product was being generated. Moreover, the signals were isolated to the intestines. However, the strength of the signal output was 1.74-fold in the high-fat group compared to the low-fat animals, which was significantly lower (Figure 5a–c). This result indicates inflammation of the gastrointestinal tract in obese animals is linked to the upregulation of MGL activity.^{8a, 30} We corroborated our findings by excising intestines from sacrificed subjects for ex vivo PA imaging. Consistent with our in vivo data, the PA signal was greater in intestines from the high-fat diet group (Figure S15). On the contrary, the same PA pattern was not observed in other tissues (e.g., liver) that were analyzed. In these organ systems, there was no discernible difference in the PA signal between the two experimental groups (Figure S16). Having successfully demonstrated MGL activity is elevated in the obesity model, we turned our attention to uncover whether FAAH activity was altered in a similar manner. As before, both groups of animals were treated with PA-HD-FAAH and imaged using our MSOT system. Analysis of the data revealed comparable results to when PA-HD-MGL was employed. Specifically, we found the enhancement was greatest in the high-fat diet animals (1.59-fold signal enhancement) and PA signals originated from the gastrointestinal tract (Figure 5d–f). The corresponding ex vivo validation studies confirmed our in vivo results (Figure S17 and S18).