A Mouse Model of Endometriosis with Nanoparticle Labeling for In Vivo Photoacoustic Imaging

To generate an exogenous contrast agent suitable for distinguishing labeled tissue via PA imaging, we colloidally synthesized gold nanorods using seed-mediated growth methods by employing surfactants as directing agents [ 39 ]. We then coated the particles with silica shells via the silica sol-gel process and attached them with a distal fluorescein tag (FITC) using simple silane chemistry [ 35 ]. Transmission electron microscopy (TEM) showed the distinctive rod structure of gold particles (width: 12–14 nm, length: 50–55 nm), and the silica shells were distinctively detected on the surface of the gold nanorods ( Fig. 1A ). The UV–vis absorption spectrum of the nanoparticles showed a maximal absorbance peak at 780 nm with a broad and intense absorption in NIR, generating strong PA signals ( Fig. 1B ). The fluorescent FITC dye attached to the particles was detected by the fluorescent mode in microplate readers (excitation wavelength: 460 nm, emission wavelength: 516 nm), and the gold-FITC particles exhibited the typical emissions of fluorescein at 516 nm under UV excitation ( Fig. 1C , D ). In order to determine optimal gold-FITC nanoparticle tissue labeling conditions for use in a mouse model of endometriosis based on syngeneic uterine tissue transfer, we tested various incubation times and concentrations with uterine tissue fragments. Tissue donor mice were treated with estradiol (E2) for 3 days to synchronize the hormonal state of the uterine tissue. Beginning with a nanoparticle solution of approximately 300 μg Au/mL and 15 μg FITC/mL, we made a 20% dilution of the nanoparticles in supplemented tissue media and incubated tissue fragments for one hour, two hours, or three hours. Tissues were washed three times before imaging to clear unattached particles. Gross visualization of tissue fragments under fluorescence microscopy and independent fluorescence quantitation showed significantly increasing fluorescent intensity with increasing incubation time ( Fig. 2A , B ). Examination of tissue sections by fluorescence microscopy revealed that FITC signal was located at the exterior border of tissue fragments ( Fig. 2C ). Furthermore, a gold enhancement assay confirmed that gold molecules were also located at the exterior border of tissue fragments, matching the location of FITC signal ( Fig. 2D ). Due to concerns about tissue viability, we did not extend the incubation longer than 3 hours and proceeded with this condition to further experiments. To determine the dependence of successful tissue labeling on nanoparticle concentration, we performed a dose-comparison with 0%, 5%, 10%, 20%, and 40% of nanoparticles in media for the three-hour incubation period with uterine tissue fragments. Fluorescence microscopy and quantitation of fluorescent intensity demonstrated the dose-dependency of nanoparticle tissue labeling ( Fig. 3A , B ). Histological examination again showed that both fluorescent dye and gold particles were located on the outside edge of the tissue fragments ( Fig. 3C , D ). We determined the labeling coverage to be sufficient at a 20% dilution (60 μg Au/mL, 3 μg FITC/mL) for use in further experiments. To monitor the establishment of endometriosis-like lesions in mice in vivo, we induced endometriosis in intact, untreated wild-type mice using syngeneic uterine tissue transfer after ex vivo incubation with gold-FITC nanoparticles in media or with vehicle (media only) according to the previously determined conditions ( Fig. 4 ). After allowing time for endometriosis lesion establishment in the recipient mice (4 weeks), we performed noninvasive, full-body photoacoustic (PA) imaging followed by fluorescence-guided lesion isolation. In mice induced with gold-FITC nanoparticle-labeled endometriosis lesions, regions outlined in gold signal (yellow) were detected colocalized with oxyhemoglobin (HbO 2 ; red) and deoxyhemoglobin (Hb; blue) signal, but similar areas were not seen in controls by PA imaging with concurrent ultrasound ( Fig. 5A ; Suppl. Video 1 ; Suppl. Video 2 ). This observation indicates an anatomically distinct area of gold nanoparticle-labeled tissue with increased blood supply as would be expected in endometriosis-like lesions [ 40 , 41 ]. Nanoparticle labeling enabled the PA detection of endometriosis lesion-specific signals, where we detected the statistically significant increase of the gold signals in the region of labeled tissue compared to unlabeled control tissues (P=0.0004; Fig. 5B ). In contrast, the presence of the nanoparticle did not significantly affect local oxygen saturation (SO 2 ), a ratio of oxyhemoglobin to total Hb (p=0.9538; Suppl. Fig. 1 ). Upon dissection, FITC-positive endometriosis-like lesions were clearly visible in the peritoneal cavity of mice induced with gold-FITC nanoparticle-labeled lesions but not in controls ( Fig. 5C ). In addition to the visualization of nanoparticle-labeled endometriosis-like lesions, PA imaging combined with ultrasound imaging can also be used to detect implanted embryos and associated functional changes (e.g., vasculature, placental oxygenation) at various stages of pregnancy development [ 29 , 30 ]. Using the same noninvasive PA imaging platform as applied for the detection of endometriosis lesions, we detected the presence of embryos as early as gestation day (GD) 7.5 and on subsequent days of pregnancy based on ultrasound-derived anatomical structures (grey) and hemoglobin photoacoustic signals ( Fig. 5D ; Suppl. Video 3 – 7 ; oxyhemoglobin, red; deoxyhemoglobin, blue). This method allows longitudinal study of pregnancy progression that is amenable to combination with monitoring of nanoparticle-labeled endometriosis lesions. Bona fide endometriosis lesion tissue contains endometrial epithelial glands and stroma [ 3 ]. To assess the histology of FITC-positive endometriosis lesions isolated from endometriosis model mice, we performed immunostaining for E-cadherin and vimentin on tissue sections. E-cadherin-positive gland-like structures surrounded by vimentin-positive stromal cells with similar structures and staining strength to the eutopic endometrium were observed, confirming endometriosis-like tissue identity ( Fig. 6A , B ). Furthermore, we performed fluorescence imaging and gold enhancement on tissue sections from the endometriosis-like lesions and detected the presence of both FITC signal and gold particles in the lesions labeled with gold-FITC nanoparticles ( Fig. 6C , D ).

AuNRs were synthesized by the seed-mediated growth method with some modifications from the previous report [ 39 ]. First, the gold seed solution was prepared by adding cold sodium borohydride (NaBH 4 ; 0.01 M; Sigma Aldrich, St. Louis, MO, Cat. #71320) to an aqueous solution of 5 mL of hexadecyltrimethylammonium bromide (CTAB; 0.2 M, Sigma Aldrich, Cat. #H6269) and 5 mL of gold(III) chloride (AuCl 3 ; 0.005 M; >99%, Sigma Aldrich, Cat. #334049). The growth solution was prepared by adding 3.5 mL of L-Ascorbic acid (0.089 M; Sigma Aldrich, Cat. #A7506) to an aqueous solution containing 12 mL of silver nitrate (4 mM; AgNO 3 ; (≥99.0%, Sigma Aldrich, Cat. #209139), 250 mL of CTAB (0.2 M), and 250 mL of AuCl 3 (0.001 M). Next, 0.6 mL of the gold seed solution was added into the growth solution, and the reaction mixture became dark blue/purple/brown overtime. After 6 hours of reaction time, the mixture was then washed three times with distilled (DI) water by centrifugation (12,000 rpm, 20 min) to remove any extra CTAB. For silica coating, tetraethyl orthosilicate (TEOS; 6 μL, 99.0%, Sigma Aldrich, Cat. #86578) was added to the diluted stock of AuNRs (2.2 nM) in 10mL of DI water with the addition of 100 μL of 0.1 N sodium hydroxide (NaOH; Sigma Aldrich, Cat. #221465). The mixture solution was then vigorously stirred for 1 hour to ensure the complete coating of silica shells, washed with centrifugation (12,000 rpm, 20 min), re-dispersed in DI water, and sonicated for resuspension. 10 mg of fluorescein isothiocyanate (FITC; Sigma Aldrich, Cat. #46905) was reacted with 44 μL of 3- aminopropyltriethoxysilane (APTES; 99%, Sigma Aldrich, Cat. #440140; molar ratio of FITC:APTES = 1:10) in 0.75 mL of ethanol under dark conditions for 2 days. 50 μL of the prepared FITC-APTES was added into the AuNRs@silica nanoparticle solution in ethanol and was stirred for 3 hours. The particle suspension AuNRs@silica(FITC) nanoparticles were washed 3 times with ethanol to remove the unreacted species and dispersed in aqueous solution. Transmission electron microscopy (TEM) imaging was performed by using a 2200FS transmission electron microscope (JEOL, Japan). TEM specimens were prepared by dropping a small amount of nanoparticle suspension in ethanol onto carbon-coated Cu grids. ICP-OES (PerkinElmer Optima 3000DV) was used to quantify the amount of gold elements from the particles by using a gold standard solution (Sigma Aldrich, Cat. #38168). The hydrodynamic diameter and zeta potentials of nanoparticles were measured by DLS (Zetasizer ZS 90, Malvern Instruments). The UV–visible absorption and fluorescent spectrum of the FITC attached particles were analyzed with a microplate reader (SpectraMax; Molecular Devices). All mouse procedures were approved by Michigan State University’s Institutional Animal Care and Use Committee. All housing and breeding were done in a designated animal care facility at Michigan State University with controlled humidity and temperature conditions and a 12 hour light/dark cycle. Access to water and food (Envigo 8640 rodent diet) was ad libitum. Mice utilized for experiments were 8 to 12 weeks old wildtype mice from mixed background C57BL/6 and 129P2/OlaHsd strains. For breeding, one male mouse was normally housed with one female mouse. One male was occasionally placed with two females to increase breeding success, in which case females were separated with their pups until weaning. After weaning at P21-P28, male and female littermates were housed separately at 5 mice/cage maximum until use in experiments or further breeding. All tissues collected for histological analysis were fixed for 6 hours in 4% (vol/vol) paraformaldehyde (Fisher Scientific, Hampton, NH, Cat. #04042–500) followed cryopreservation in a series of sucrose solutions increasing from 10% to 15% to 20% sucrose in Hanks’ Balanced Salt Solution (HBSS; Gibco, Grand Island, NY, Cat. #14170–112) before freezing in Tissue-Tek optimal cutting temperature (OCT) compound (Sakura Finetek USA, Torrance, CA, Cat. # 4583). Induction of endometriosis was modified from previously described methods [ 10 , 12 ]. Intact wildtype tissue donor mice were injected with 100 μL of 1 μg/mL estradiol (E2; Sigma Aldrich, Cat. #E8875) in sesame oil daily for 3 days to prepare the donor uterine tissue. Approximately 6 hours after the final injection, the mouse was euthanized, and the uterus was removed. For each recipient mouse, one uterine horn was opened longitudinally with scissors and cut into small fragments of about 1 mm 3 in a petri dish with a scalpel. Tissue fragments were placed into 500 μL RPMI-1640 media (Gibco, Cat. #11835–030) supplemented with 10% fetal bovine serum (FBS; Gibco, Cat # 16000044), 0.1 mM sodium pyruvate (Gibco, Cat. #11360–070), and 1% penicillin streptomycin (P/S; Gibco, Cat #15140) with gold-FITC nanoparticles (60 μg Au/mL, 3 μg FITC/mL). Mixtures were incubated for 3 hours at 37°C, 5% CO 2 before washing 3 times in media to remove free floating nanoparticles before transferring to the recipient mouse. Under anesthesia, a small midline abdominal incision was made in the recipient mouse, and nanoparticle-labeled tissue was injected into the peritoneal cavity. The abdominal incision was closed with sutures for the peritoneum and wound clips for the skin. Photoacoustic imaging was performed using the inVision 512-echo preclinical multispectral optoacoustic tomographic imaging (MSOT) system (iThera Medical, Munich, Germany). The transmit central frequency of the transducer is set to be 5 MHz. MSOT functions to visualize images by irradiating the pulsed laser lights in multiple wavelengths (680–980 nm) and detecting the propagated ultrasound waves emitted from photo-absorbing molecules–oxyhemoglobin (HbO 2 ), deoxyhemoglobin (Hb), and imaging agents–from the tissue. Under isoflurane anesthesia, the mouse was shaved 360° around the abdominal area, all remaining hair in this region was removed with commercial hair removal cream, and warm ultrasound gel was applied. The mouse was then placed in a mouse holder with a thin polyethylene membrane to ensure acoustic coupling. Next, the assembled mouse holder was submerged in warm water in the imaging chamber. During imaging, the anesthetic isoflurane and oxygen were supplied through a breathing mask. Imaging took place with the movement of the imaging stage in a given scanning site (abdominal area). All acquisition was performed using 10 averages per illumination wavelength, with chosen wavelengths (680, 700, 730, 760, 800, and 850 nm) and step sizes of scanning (0.3 mm). The imaging took less than 20 minutes per mouse. The acquired images were reconstructed using a back-projection algorithm, and linear spectral unmixing was applied as implemented in the ViewMSOT software (iTheraMedical). The multispectral data analysis fit to the pixel-to-pixel intensities across different wavelengths to create component images for each individual absorber (gold, Hb, and HbO 2 ) from the composites. In PA imaging, all layers in a multispectral image are autoscaled using imaging threshold tool to eliminate the visualization of low signals. Oxygen saturation (SO 2 ) was calculated as a ratio of oxyhemoglobin to total Hb. PA imaging at specific times of pregnancy was performed using wildtype female mice after mating with wildtype male mice and defining morning of identification of a vaginal plug as gestation day (GD) 0.5. Pregnancy was visually confirmed by dissection after completing in vivo imaging. Fluorescence-guided dissection and brightfield imaging were performed with a Nikon fluorescence dissection microscope and NIS-Elements imaging software (Nikon Instruments, Melville, NY). Quantitation of fluorescence intensity was performed using the IVIS Spectrum in vivo imaging system and Living Image software (PerkinElmer, Waltham, MA). Measurements were made based on normalized Radiant Efficiency units from manual regions of interest after adaptive fluorescence background subtraction. For fluorescence imaging of tissue sections, frozen, OCT-embedded tissue samples were cut at 10 μM and mounted on slides before coverslipping with DAPI mounting media (Vector Laboratories, Burlingame, CA, Cat. #H-1800) for imaging. For immunostaining, frozen tissue sections were fixed in 4% (vol/vol) paraformaldehyde (Fisher Scientific, Cat. #04042–500), immersed in 0.3% hydrogen peroxide in methanol, and washed in 1/40 Triton-X 100 (Fisher Scientific, Cat. #BP151–500) before blocking with 10% normal goat serum (NGS; Vector Laboratories, Cat. #S-1000) in pH 7.5 PBS and incubating with primary antibodies diluted in 10% NGS in PBS overnight at 4°C. Primary antibodies were used at the following dilutions: 1:1000 for anti-E-Cadherin (CS-3195, Cell Signaling Technology, Danvers, MA) and 1:10,000 for anti-Vimentin (ab92547, Abcam, Cambridge, United Kingdom). An appropriate species-specific fluorescently tagged secondary antibody (Donkey anti-Rabbit IgG Alexa Fluor 594; Invitrogen, Carlsbad, CA, Cat. # A-21207) was then used before mounting and coverslipping with DAPI mounting media for imaging. Imaging was performed with a Nikon epi-fluorescence microscope and NIS-Elements imaging software (Nikon Instruments). Gold enhancement for microscopic visualization was performed with the GoldEnhance LM kit (Nanoprobes, Inc, Yaphank, NY, Cat. #2112–28) on frozen, OCT-embedded tissue samples cut at 10 μM and mounted on slides. The kit was used according to the manufacturer’s instructions. To assess statistical significance, we used one-way ANOVA followed by Tukey’s post hoc test for multiple group comparisons. A value of p < 0.05 was considered statistically significant. Statistical analyses were performed using the InStat 3 package from GraphPad (San Diego, CA).

Preclinical animal models are essential tools for gaining a mechanistic understanding of endometriosis development and its relationship to infertility and pain. They also enable preclinical proof of concept studies for future clinical diagnostics and treatment options. A wide variety of mouse models of endometriosis have been used for this purpose, but many of these have been limited by a lack of close similarity to human endometriosis pathophysiology and a need to use large numbers of mice to collect endpoint data without a way to monitor endometriosis development over time [ 42 ]. Here, we present a reproductively intact mouse model of endometriosis based on syngeneic transfer of uterine tissue tagged with gold nanoparticles conjugated to FITC dye and amenable to noninvasive detection. The benefits of this model are at least six-fold. First, the primary advance made by our model for endometriosis research is the novel application of PA imaging with an exogenous contrast agent for in vivo imaging of internally located endometriosis-like lesions in mice with the added benefit of pregnancy monitoring via endogenous PA signals and B-mode ultrasound. To our knowledge, this is the first reported endometriosis mouse model for in vivo imaging of lesions located deep in the peritoneal cavity that is not based on optical imaging, which is limited by penetration depth and lacks anatomical information. Photoacoustic, high-resolution ultrasound, and magnetic resonance imaging have each been applied to endometriosis mouse models previously, but these models involved suturing the donor endometrial tissue close to the surface of the animal, either subcutaneously or to the peritoneal wall [ 43 – 49 ]. Our model incorporates both endogenous PA signals (e.g., hemoglobin) and an exogenous contrast agent to locate endometriosis-like lesion tissue anywhere within the peritoneal cavity and also visualize developing embryos in the case of pregnant mice. The PA signals are combined with ultrasound imaging to provide concurrent anatomical data. Second, our exogenous gold nanoparticle contrast agent is conjugated to a FITC fluorescent dye, which enables identification of labeled tissues in multiple-length scales by fluorescent imaging (macroscopic/microscopic) in addition to PA imaging (macroscopic). This property is useful for fluorescence-guided dissection and isolation of lesions at the experiment endpoint as well as locating the nanoparticles in the tissue via histological analysis. Third, our use of an inert, biocompatible exogenous nanoparticle contrast agent rather than a genetically incorporated reporter alleviates the need for the time-consuming establishment of new genetically modified mouse lines. Rather, our system can be applied to any wild-type or genetically modified mouse of interest. Still, combining the current model for PA imaging with bioluminescence would be beneficial to track the viability of implanted tissues, which is a limitation of nanoparticle imaging markers. Fourth, our model is applied to recipient mice without surgical or hormonal disruption of reproductive function. Since suboptimal fertility is one of the major dysfunctions associated with endometriosis, the ability to study the effects of endometriosis and potential therapies on an intact reproductive tract is crucial. Additionally, chronic E2 treatment in other endometriosis models frequently produces cystic lesions that do not represent the typical endometriosis lesion histology in women [ 50 , 51 ]. Fifth, our model is based on the injection of finely chopped uterine tissue into the peritoneal cavity of the recipient mouse rather than suturing larger pieces of tissue. This injection method more closely mirrors the mechanism of human endometriosis etiology that is most commonly accepted, retrograde menstruation, where menstrual endometrial tissue flows backward into the peritoneal cavity via the fallopian tubes [ 3 ]. Sixth, our model is simple to employ. There is no need for major survival surgery in the recipient that would require a great degree of surgical skill, and there is only a simple three-day intraperitoneal E2 injection required to prepare the donor mouse rather than a lengthy and complicated hormonal regimen. These strengths highlight the utility of our new model as a tool for the field. In addition to the many benefits of our model, there are also important limitations to consider. Of course, synthesis of the gold-FITC nanoparticles as a contrast agent is required since they are not commercially available at this time. PA imaging systems are also currently less accessible than optical imagers, but this may change with the increased adoption of PA imaging methods in various fields of study [ 52 ]. Furthermore, an inert exogenous label like our gold-FITC nanoparticle will dissipate over time in living tissue as cells divide, die, and are recycled. This leads to diminished signal intensity, the possibility of off-target signals, and the inability to determine tissue viability based on signal presence or intensity. Moreover, though signal penetration for PA imaging is greatly improved over optical imaging, imaging depth could limit the application of this system in larger animals and in the clinic. There are several exciting potential applications of the model we present in this work with regard to preclinical animal experiments and future clinical studies. We view this model as an excellent candidate for the longitudinal study of the effects of endometriosis lesion number, size, and location on fertility outcomes. It also has potential as a method for drug candidate screening as it allows imaging lesion number and size before, during, and after treatment. Combined longitudinal in vivo data regarding fertility outcomes and lesion response to drug treatments would also be possible. Furthermore, we see no reason our nanoparticle-based imaging method could not also be applied to other endometriosis mouse models to compare treatment efficacy in multiple models and increase the rigor and reproducibility of preclinical findings. Additionally, past studies have shown the feasibility of in vivo imaging using intravenously injectable contrast agents that preferentially home to endometriosis lesions based on locally increased angiogenesis [ 22 , 40 , 53 ]. Our nanoparticle-based contrast agent could potentially be applied through this avenue for short term noninvasive PA imaging. Finally, this work contributes to a growing body of research suggesting the utility of PA imaging either with or without exogenous contrast agents for noninvasive clinical imaging [ 26 , 52 ]. Traditional ultrasound has already been applied as a tool for detection and surveillance of endometriosis, and Doppler ultrasound can even estimate blood flow [ 54 ]. Therefore, it is easy to envision how the incorporation of PA imaging technology would be a feasible and useful addition to the clinical arsenal of noninvasive endometriosis imaging. The ‘light in/sound out’ approach of PA imaging has the potential improve the imaging depth and resolution to identify and monitor lesion development more precisely [ 26 ]. Additionally, detection of endogenous chromophores such as hemoglobin and exogenous contrast agents like indocyanine green and methylene blue could enhance the ability for clinicians to noninvasively determine the lesion type and molecular and cellular characteristics [ 52 ]. Though there is still need for further preclinical development, the future possibility of implementation of portable PA imaging as an additional tool for noninvasive clinical diagnosis of endometriosis is exciting. Women suffering from endometriosis may someday finally be diagnosed and treated without waiting for years and without the need for invasive laparoscopy.

Endometriosis affects an estimated 1-in-10 women of reproductive age, equating to 176 million globally, but estimated delays from symptom onset to diagnosis range from 4 to 11 years [ 1 – 3 ]. With regard to pathophysiology, endometriosis is a condition of the female reproductive tract where non-malignant lesions composed of endometrium-like glands and stroma take root and grow outside the uterus, frequently leading to chronic pelvic pain and fertility problems [ 3 ]. Definitive diagnosis requires invasive laparoscopic visualization of lesions because no clinically reliable biomarkers are available, and only limited subtypes of endometriosis such as ovarian endometrioma or deep infiltrating endometriosis can be detected using noninvasive ultrasonography or MRI [ 3 – 5 ]. This lack of noninvasive diagnostic tools combined with heterogeneity in symptoms and lesion location among patients contributes to the marked delays in diagnosis and treatment [ 2 ]. Furthermore, the delay in diagnosis makes it difficult to directly study the early stages of endometriosis development and discern how it leads to infertility and pain in patients. To mechanistically study endometriosis development and the relationship between lesion presence and symptoms, several animal models have been developed utilizing rodents and non-human primates [ 6 ]. Mice, in particular, are attractive as an accessible and cost-effective species for modeling endometriosis due to their small size, rapid breeding, and receptivity to genetic manipulation [ 7 ]. A variety of approaches have been used to induce endometriosis-like lesions in mice that involve either heterologous transfer of human endometrial tissue to immunocompromised mice [ 8 , 9 ] or homologous murine uterine tissue transfer. Such homologous models have used autologous [ 10 , 11 ] or syngeneic [ 12 – 14 ] transfer of either sutured [ 15 – 17 ] or injected [ 18 ] uterine tissue. Some models utilize inductions performed after ovariectomy with exogenous hormone treatment to prepare the tissue [ 19 ], whereas some use intact, untreated mice [ 20 ]. Each method comes with unique benefits and limitations with regard to mimicking human endometriosis etiology and histology, identification of lesions, and maintenance of a functionally intact reproductive system [ 21 ]. Several past studies utilizing endometriosis mouse models have incorporated reporter systems for detection of endometriosis lesions. For example, genetically encoded fluorescence or bioluminescence reporters [ 14 , 22 , 23 ] and ex vivo adenoviral vector-mediated fluorescent labeling [ 8 , 13 ] have been successfully applied for in vivo imaging or post-mortem endometriosis lesion analysis. However, conventional optical imaging suffers from low imaging depth, thus limiting whole tomographic imaging in preclinical studies [ 24 ]. Therefore, the majority of these studies resort to mouse models that place the lesions subcutaneously or just beneath the peritoneal wall, except the work of Dorning et al., which achieved bioluminescent imaging in intraperitoneal injection models of endometriosis [ 14 , 24 ]. Photoacoustic (PA) imaging is an emerging imaging tool to noninvasively detect and longitudinally track the targeted cells or tissues in vivo [ 25 ]. In contrast to conventional optical imaging, PA imaging can substantially increase the signal penetration depth based on the ‘light in/sound out’ approach [ 26 ]. In PA imaging, the signals are generated by the ultrasonic pressure waves, which are emitted by the thermoelastic expansion under near-infrared (NIR) light illumination (680–980 nm) [ 27 , 28 ]. Since ultrasonic waves propagate much farther through the tissue than light, PA imaging can achieve deeper tissue penetration for imaging (up to several centimeters). As a hybrid imaging technique, PA imaging offers excellent temporal (100 ms) and spatial (50–150 μm) resolution with concurrent anatomical data from B-mode ultrasound imaging [ 25 ]. The ultrasound imaging component in the tool can also be used to analyze pregnancy, fetal gestational staging, and functional changes during pregnancy development [ 29 , 30 ]. Since PA imaging is a multi-spectral imaging technique, this tool can facilitate the imaging of target-specific signals via exogenous nanoparticle contrast agents alongside mapping with endogenous chromophores such as hemoglobin [ 31 ]. The quantification of total hemoglobin can allow imaging for angiogenic endometriosis lesion vasculature [ 32 ], and the addition of an exogenous contrast agent is beneficial to attain sensitive target-specific signals. Ideal photoacoustic contrast agents can be made with NIR light (680–980 nm) absorbing compounds, including small-molecule dye (e.g., indocyanine green, methylene blue), metallic nanoparticles (e.g., gold nanorods, carbon nanotubes), or organic nanostructures (porphyrin). The contrast agents can efficiently convert the irradiated light energy to heat in PA imaging to produce intense PA signals for detection [ 33 ]. Among them, gold nanorods are the most available exogenous imaging agent because they feature strong NIR light absorption (high molar extinction coefficient), inert nature, biocompatibility, and in vivo stability [ 34 ]. In addition, silica-coated gold nanorods have been reported to amplify the labelling efficacy to achieve sufficient particle loading in the targeted tissues [ 35 , 36 ]. Therefore, gold nanoparticles have been extensively used for PA molecular imaging of cells [ 35 ], proteins [ 37 ], and tumor tissues [ 38 ]. In this work, we utilized gold nanoparticles conjugated with a fluorescein isothiocyanate (FITC) dye as a contrast agent with sensitive PA signal to discriminate the endometriosis lesion-specific signals from endogenous tissues in mice. The multimodal imaging nanoparticles incorporating a fluorescent FITC dye on the particle surface were designed to be used dually for in vivo PA imaging and for fluorescence-guided tissue isolation by fluorescence dissection microscope. Upon ex vivo labelling of donor uterine tissue with the nanoparticles followed by transfer to the recipient mice, we monitored the distribution and retention of the labelled tissue by PA imaging. Furthermore, we optimized the treated particle dosages without inducing toxicity or perturbing the functionality of labelled endometriosis-like lesion tissues. Finally, we validated their utility for in vivo detection and fluorescence-guided dissection with histological confirmation.

Suppl. Figure 1 (Related to Fig. 5A ) In vivo imaging of endometriosis-like lesions based on oxygen saturation (SO 2 ). (A) Representative in vivo photoacoustic (PA) images from control endometriosis mice (no nanoparticle; left) and mice with gold-FITC nanoparticle-labeled lesions (right) showing PA signal detection for SO 2 (blue-red) four weeks after induction. The orange segmented circle (left panel) indicates a region containing an unlabeled lesion identified after dissection. The orange segmented circle (right panel) indicates lesion identified by PA signal and confirmed by dissection. (B) Mean intensities of SO 2 PA signals from each group. We plotted PA signals from ROIs drawn at the endometriotic lesions of at least three mice. The graphs represent the mean ± SEM (control n=4, gold-FITC nanoparticle n=3; ns, p>0.05). Scale bar = 5 mm. Suppl. Video 2 (Related to Figure 5A ) Video reconstruction from in vivo PA imaging of a mouse with gold-FITC nanoparticle-labeled lesions showing PA signal detection for gold (yellow), Hb (blue), and HbO 2 (red) four weeks after induction. PA signals were co-registered with mouse anatomy by B-mode ultrasound (grey). The viewing angle is transverse to the length of the mouse body, and the scanning direction is from cephalic to caudal. Scale bar = 5 mm. Suppl. Video 1 (Related to Figure 5A ) Video reconstruction from in vivo PA imaging of a control endometriosis mouse (no nanoparticle) showing PA signal detection for gold (yellow), Hb (blue), and HbO 2 (red) four weeks after induction. PA signals were co-registered with mouse anatomy by B-mode ultrasound (grey). The viewing angle is transverse to the length of the mouse body, and the scanning direction is from cephalic to caudal. Scale bar = 5 mm. Suppl. Video 7 (Related to Figure 5D ) Video reconstruction from in vivo PA imaging of a gestation day (GD) 11.5 wild-type mouse showing PA signal detection of Hb (blue) and HbO 2 (red) combined with ultrasound-derived anatomical structures (grey). The viewing angle is transverse to the length of the mouse body, and the scanning direction is from cephalic to caudal. Scale bar = 5 mm. Suppl. Video 3 (Related to Figure 5D ) Video reconstruction from in vivo PA imaging of a non-pregnant wild-type mouse showing PA signal detection of Hb (blue) and HbO 2 (red) combined with ultrasound-derived anatomical structures (grey). The viewing angle is transverse to the length of the mouse body, and the scanning direction is from cephalic to caudal. Scale bar = 5 mm. Suppl. Video 5 (Related to Figure 5D ) Video reconstruction from in vivo PA imaging of a gestation day (GD) 8.5 wild-type mouse showing PA signal detection of Hb (blue) and HbO 2 (red) combined with ultrasound-derived anatomical structures (grey). The viewing angle is transverse to the length of the mouse body, and the scanning direction is from cephalic to caudal. Scale bar = 5 mm. Suppl. Video 4 (Related to Figure 5D ) Video reconstruction from in vivo PA imaging of a gestation day (GD) 7.5 wild-type mouse showing PA signal detection of Hb (blue) and HbO 2 (red) combined with ultrasound-derived anatomical structures (grey). The viewing angle is transverse to the length of the mouse body, and the scanning direction is from cephalic to caudal. Scale bar = 5 mm. Suppl. Video 6 (Related to Figure 5D ) Video reconstruction from in vivo PA imaging of a gestation day (GD) 9.5 wild-type mouse showing PA signal detection of Hb (blue) and HbO 2 (red) combined with ultrasound-derived anatomical structures (grey). The viewing angle is transverse to the length of the mouse body, and the scanning direction is from cephalic to caudal. Scale bar = 5 mm.