Ultra-high-resolution 3D optical coherence tomography reveals inner structures of human placenta-derived trophoblast organoids.

OBJECTIVE
3D optical coherence tomography (OCT) is used for analyses of human placenta organoids in situ without sample preparation.


METHODS
The trophoblast organoids analyzed were derived from primary human trophoblast. In this study a custom made ultra-high-resolution spectral domain OCT system with uniform spatial and axial resolution of 2.48 m in organoid tissue was used. The obtained OCT results align to differentiation status tested via quantitative polymerase chain reaction, Western blot analyses, immunohistochemistry, and immunofluorescence of histological sections.


RESULTS
3D OCT enables a more detailed placenta organoid monitoring compared to brightfield microscopy. Inner architecture with light scattering bridges surrounding cavities were visualized and quantified in situ for the first time. The formation of these bridges and cavities is congruent to differentiated trophoblast organoids having developed syncytiotrophoblasts.


CONCLUSION
Using 3D OCT in living placenta organoids is a fast tool to assess the differentiation status and resolve internal structures in situ, which is not possible with standard live cell imaging modality.


SIGNIFICANCE
Only recently human placenta-derived organoids were established, allowing to have a highly reproducible and stable in vitro model to investigate not only developmental but also physiological and pathophysiological processes during early pregnancy. To our knowledge, this work is the first to analyze living human placenta organoids using 3D OCT. Thereby, the rapid and especially non- endpoint OCT qualitative analyses align to the differentiation stage of organoids, which will aid future advancement in this field.


I. INTRODUCTION
Cell culture is an indispensable preclinical model system to study physiological events, disease, and drug discovery. Conventionally, 2D cell culture are used in which cells adhere in a monolayer to the flat surface of a petri dish [1], [2]. However, cells in the body interact in a 3D microenvironment which influences cellular proliferation, differentiation, and function due to different signals, chemical gradients, and cellcell interactions [1], [3]. organoids, have been developed to address these shortcomings [1], [3]. The former is a simple form of aggregated cellular spheres, often derived from epithelial cancers and maintained as free-floating cultures [4]. In contrast, organoids are formed by stem or progenitor cells, that give rise to complex, selforganizing, and self-renewing 3D cell culture systems embedded in a matrix-gel recapitulating more closely features of in vivo organs [5], [6], [7]. Another advantage of organoids is their amenability for genetic manipulation, monitoring, and molecular assays compared to in vivo models [8]. A plethora of organoids of different organs were already established such as brain, breast, liver, colon, and pancreas [9]. Recently, selfrenewing placenta organoids, derived from isolated trophoblasts of first trimester placentae, were developed and are used as a model system for the early human placenta [10], [11]. These works set the fundament to elucidate physiological and pathophysiological processes aiming for placental disease modeling and fetal-maternal interaction during placentation. The human placenta is an extraembryonic organ and connects the mother to the fetus. Beside the exchange of nutrients and gases, the placenta produces pregnancy hormones, such as human chorion gonadotropin (hCG) required for the establishment and maintenance of pregnancy, and specified placental cells effectuate the maternal acceptance of the fetal allograft. Failures in placentation result in gestational disorders such as preeclampsia, intra-uterine growth restriction and recurrent abortion. In the first trimester of pregnancy, the human placenta has a tree-like (villous) structure that is composed of a stromal part being covered by a bi-layer of the so-called trophoblasts. Villous cytotrophoblast (vCTBs), directly contacting the stroma, undergo two differentiation routes. Firstly, vCTBs proliferate and fuse into the overlying syncytiotrophoblast (STB) layer, which is responsible for hormone production and transport functions. Secondly, intensive proliferation of vCTBs at villous tips lead to the generation of extravillous trophoblasts that invade maternal uterine tissues to transform maternal vessels, thereby providing an adequate blood flow to the fetus during the gestation, and to adapt the maternal immune system to the fetal semi-allograft [10]. However, the current understanding of the first weeks of human placentation were mainly derived from interpretations of hysterectomy specimen as well as studies in mice and great apes [12], [13]. 2D trophoblastic cell lines as well as 2D human primary trophoblasts isolated from first-trimester placentae were developed [14], but are not an adequate model as they rapidly terminate proliferation and differentiate in vitro. Further, they do not properly represent 3D in vivo morphology and cell composition [4]. In the last three years, the establishment of human trophoblast stem cells [15], and especially trophoblast organoids [10], [11] have substantially improved the field of placental research. Trophoblast organoids (TB-ORGs) are densely packed, consisting of an outer cell layer of proliferative vCTBs which differentiate into hormoneproducing STBs toward the center of organoids. However, most imaging analyses were conducted as endpoint studies in which the organoid cultures were fixed and processed for either immunofluorescence analyses or electron transmission microscopy [10], [11]. Longitudinal studies monitoring living organoids were performed by phase contrast/brightfield imaging, but due to the shallow penetration depth in large organoids (300 to 800 µm), their inner structures were neither clearly discernible nor quantifiable [10], [11]. Moreover, the above-mentioned endpoint analyses require the termination of the experiment at potentially sub-optimal time points [16]. Additionally, the inevitable fixation and clearing procedures cause alterations in 3D structures such as shrinking and collapsing of organoids [17], and cross sections are only snapshots of individual organoids [16]. Indeed, serial cross section staining and fluorescence Z-Stack images of the whole organoid can be obtained, but inherently do not eliminate the need of genetic fluorescent reporters or extensive sample preparation [17].
In organoid cultures such as organoids of the brain [18] or colon [17], light sheet and multiphoton imaging were used for volumetric imaging. Here, entire organoids can be imaged in detail, but fixation, labeling, and optical clearance need to be performed [17]. Further, micro-computed tomography (microCT) was established for analyzing dentin-pulp-like and retinal organoids, but again, sample fixation was required [19], [20]. Improvement of live imaging was obtained with lattice light sheet microscopy, but this technique still required fluorescence labeling [21]. Other methods, such as hyperspectral imaging and fluorescence lifetime imaging microscopy allows imaging of the metabolism of organoids without labeling [20], however, these are costly methods and are only providing intensity projections. In contrast, optical coherence tomography (OCT) allows optical sectioning of organoids enabling label free, longitudinal studies. OCT is based on low-coherence interferometry [22] and takes advantage of intrinsic scattering contrast arising from biological tissue due to different refractive indices [23]. OCT has been widely used in several clinical fields including ophthalmology [24] and dermatology [25], but also has started to being used in 3D cell cultures, such as organoids and tumor spheroids [26]- [28]. So far, stem cell-derived organoids of human retinal [16], [20], [29], [30] and intestinal origin [31] were analyzed by OCT.
In this work, we hypothesize that 3D OCT enables to determine the inner structures of living placenta organoids in situ without sample preparation and to align the obtained recording to the differentiation status of organoids.
Herein, we will demonstrate that the 3D OCT imaging technology can be used to monitor STB formation in individual TB-ORGs in a volumetric, non-invasive, and non-labelling manner. By using specific inhibitors blocking STB differentiation we provide controls that allow to verify the strong and direct relationship between OCT-obtained imaging data with conventional end-point-derived methods.

A. Trophoblast organoid formation and cultivation
Isolation, utilization of placenta tissue, and all experimental procedures were approved by the ethics board of the Medical University of Vienna (Number 084/2013) requiring informed consent of all patients and are in adherence with the Helsinki Declaration. Human first-trimester placental tissues from 6 th to 7 th week of gestation (n=3) were collected from elective pregnancy terminations. Trophoblast organoids were generated and cultivated as described previously [10]. In brief, villous cytotrophoblasts were obtained after sequential digestion of the placental tissue, and embedded in a semi-solid 60 % growth factor reduced-matrigel (GFR-M) diluted with organoid growth medium consisting of DMEM/F12 (Invitrogen) supplemented with 1x N2 (Gibco), 1x B27 (Gibco), 2 mM glutamine (Gibco), 10 mM HEPES (Gibco), 1 µM A83-01, 100 ng/ml recombinant human epidermal growth factor (rhEGF, R&D Systems) and 3 µM CHIR99021 (Tocris). Drops (40 µl) of cell-GFR-M mixtures were placed in the center of 4-well dishes (Nunclon Delta Surface, Thermo Scientific) and incubated for 1 minute at 37 °C and 5 % CO2 in a humidified incubator. Subsequently, plates were flipped upside down and incubated for another 15 minutes to ensure equal distribution of the cells in the solidifying GFR-M drops. Afterwards, the plates were flipped back and 0.5 ml of organoid growth medium was added to the dishes. After 7 to 10 days, TB-ORGs were split and treated with or without 10 µM p38 MAPK inhibitors (SB202190, Stemcell Technologies). The culture medium was changed every 2 to 4 days. After additional 10 days of cultivation, brightfield images were taken with the EVOS FL Cell Imaging System microscope (Life technologies) and OCT images were obtained. Next, organoids were either fixed for immunofluorescence analyses or harvested for ribonucleic acid (RNA) and protein isolation.

B. Optical coherence tomography system set-up
In the customized 3D bulk-optics spectral domain OCT we used a novel polarization-aligned three SLED broadband source (EBD290002-00, EXALOS AG) with a central wavelength of 846 nm and a 3 dB bandwidth of 133.3 nm [32]. The spectrometer for measuring the interference signals between the sample and the reference arm comprised a transmitting diffraction grating (3253-W-01, Wasatch Photonics), a lens system (85 mm f/2.0 Makro-Planar T, Zeiss) and a Basler Sprint spL4096-140km line scan camera. The sample arm applied galvanometer scanners (CTI6220H, Cambridge Technology), a 4 focal length lens system and an Olympus-UPlanSApo 4 telecentric microscope objective to perform raster scanning across the sample with an axial resolution of 3.40 µm in air. A refractive index of 1. 37 [33], similar to cancer spheroids, was assumed for quantitative axial measurements and 3D reconstructions, leading to an axial resolution of 2.48 µm in tissue. The transverse resolution was measured to be 2.48 µm. The transverse pixel step size was 1.27 µm. The optical power at the sample was approximately 500 µW for all experiments.

C. OCT data acquisition and reconstruction
For OCT imaging, the 4-well plates with the living organoids were placed in focus on the sample holder of the OCT system and tilted by 10° to prevent back reflection from the cell culture dish. OCT images were acquired with 5 kHz line rate and depending on the size of the organoid 200 to 800 A-scans per 200 to 800 B-scans were acquired. The obtained OCT interferogram was subjected to standard OCT signal processing (background subtraction, dispersion compensation, zeropadding interpolation, fast Fourier transform) [34]. For visualization, the imaging software ImageJ 1.52p (Wayne Rasband, National Institutes of Health, USA) was used for 2D images and Icy 2.0.3.0 (BioImage Analysis unit Institut Pasteur) for 3D rendering, respectively. ImageJ was used for semi-automated quantification of the organoids.

D. Immunofluorescence and H&E staining
After imaging of the placenta organoid with OCT the samples were subjected to fixation, paraffin embedding, hematoxylin and eosin (H&E), and immunofluorescence staining as previously described [10]. In brief, organoid domes were washed with phosphate-buffered saline (PBS) and fixed in 4 % formaldehyde solution (Honeywell) overnight at 4 °C. Afterwards, the organoids were washed with PBS twice, and incubated for 30 minutes with 70 % ethanol (EtOH) supplemented with 10 µl alcian blue (Merck) to enhance visibility. Next, the samples were dehydrated with rising concentrations of EtOH and lastly with xylene, 15 minutes each at room temperature (RT). Subsequently, samples were impregnated in pre-warmed paraffin (Histowax) at 65 °C and then allowed to solidify on an ice-cold plate. Serial sections (2 to 3 µm) were deparaffinized, rehydrated and either stained with H&E using standard procedures [35], [36], or immunofluorescence was performed. For the latter, antigen retrieval was done with 1x PT module buffer 1 (100x stock solution, Thermo Scientific) for 35 minutes at 93 °C using a KOS microwave histostation (Milestone). Then, sections were incubated with blocking solution (PBS/0.5 % fetal bovine serum/0.3 % Triton X-100) for one hour at RT. Afterwards, slides were incubated with primary antibodies for staining the protein of interests (diluted in PBS/1 % Bovine Serum Albumin/0.3 % Triton X-100) at 4 °C overnight. The following antibodies were used: rabbit polyclonal anti-SDC1 (Sigma, Cat. No. HPA006185, 1:200), mouse monoclonal anti-E-cadherin (BD Transduction Laboratories, Cat. No. 610181, 1:200). Next, the slides were washed three times with PBS and incubated with appropriate secondary antibodies (2 μg/ml; Alexa, Molecular Probes) for one hour at RT. Nuclei were stained with 1 μg/ml 4′,6-Diamidino-2-phenylindole dihydrochloride (DAPI, Roche) and sections were embedded using fluoromount G (Soubio). Stained sections were analyzed by fluorescence microscopy (Olympus, BX50) and digitally photographed (CellP software, Olympus).

G. Manual segmentation for OCT vs brightfield comparison and statistical analysis
ImageJ was used to quantify the area of organoids. Thereby, the organoid was circumvented and the region of interest (ROI) saved in ImageJ ROI manager. For processing brightfield images, the length of the incorporated scale bar was used to set the correct distance in pixels to obtain the area of the circumvented organoid. For OCT images, the scale was set to 0.79 pixels/µm based on the pixel step size. For comparing brightfield and OCT microscopy nine organoids were measured three times via ImageJ by three raters on OCT and brightfield images. To test interrater reliability intraclass correlation coefficient (ICC) estimates and their 95 % confidence intervals were calculated based on 2-way mixedeffects model, mean-rating (k = 3), absolute-agreement. As basis for evaluating the level of reliability 95 % confidence interval of the ICC estimate was used: Values less than 0.50 were indicative of poor reliability, values 0.50 to 0.75 moderate reliability, 0.75 to 0.90 good reliability, and values greater than 0.90 excellent reliability according to Koo and Li [38]. For comparing brightfield and OCT images paired t-test was used, thereby p-values smaller than 0.05 were considered as significantly different. Furthermore, a Bland-Altman plot was generated to visualize the agreement [39]. Thereby, the area of the OCT en face images were subtracted from the brightfield images to measure the difference. The regression linear coefficient was calculated to test for proportional bias. For obtaining the volume of the whole organoids (n=4) and their cavities, each OCT cross-sectional segmented area was multiplied by the z-axis pixel size (1.27 µm) and added together. The cavity volume was calculated as percentage with respect to the organoid whole volume (100 %). Statistical analyses were run on IBM SPSS Statistics for Windows, Version 25.0. Armonk, NY: IBM Corp.

A. OCT imaging of TB-ORG enables non-invasive, detailed evaluation of inner organoid structures
TB-ORG cultures (n=3 donors) were used to investigate if OCT provides advantages over standard imaging methods for 3D cell culture. We compared the images of the same living, in GRF-M embedded TB-ORGs via two in situ imaging techniques, brightfield microscopy using the EVOS FL Cell Imaging System microscope (Fig. 1a) and a customized OCT system ( Fig. 1b-d), respectively. A tread was used as a marker for orientation and facilitated locating identical organoids in the cell culture well for imaging. Fig. 1a illustrates two representative TB-ORGs. Compared to OCT, the brightfield microscope images revealed only limited structural detail (Fig.  1a). Using OCT, a volume scan enables different organoid assessments and visualization (Fig. 1b -1d), such as en face averaged images (Fig. 1b) similar to brightfield images and different cross sections (Fig. 1c). To investigate wether OCT and brightfield imaging techniques were in quantitative agreement, TB-ORG areas of brightfield and en face avaraged OCT images of 9 organoids were measured via ImageJ by three raters three times and compared. The average measurements of brightfield images and OCT showed no significant difference (p-value=0.621) when tested with paired samples t-test. To analyse interrater reliability, ICC of the raters were performed. The interrater reliability for raters for brightfield and OCT images were reported as excellent (brightfield ICC = 0.999 with 95 % confidence inveral = 0.996 -1.000; OCT ICC = 0.999 with 95 % confidence inveral = 0.996 -1.000). To assess the agreement between the two microscopy methods, a Bland-Altman plot was generated (Fig. 2). It visualizes the different sized organoids ranging from approximately 5000 µm 2 to 77000 µm 2 . The average difference (= 250.67 µm 2 ) of the quantified brightfield and OCT imaged organoids is shown by the red dotted line, indicating an offset resulting from brightfield images being non significantly bigger than OCT measurements. All analyzed organoids were within the 95 % limit of agreement (upper limit: 3095.10 µm 2 ; lower limit -2593.77 µm 2 ). No proportional bias was evident when testing linear regression (pvalue = 0.237).
Investigating the advantages of OCT to optically dissect the specimen, the smaller TB-ORG in Fig. 1c depicts a high scattering cell border with a lumen, which stands in contrast to the bigger TB-ORG with numerous heterogeneously sized cavities. Interestingly, the zoom-in in Fig. 1d allows to appreciate that cavities were filled to different degrees with light scattering content.  In conclusion, areal quantitative assessment of TB-ORG were similar between the brightfield and OCT imaging methods. However, OCT additionally enables to non-invasively reveal details of inner structures which were not possible via brightfield microscopy.

B. OCT allows appreciation of organoid 3D orientation to each other and of internal structures
From the en face images seen in Fig. 1 exact 3D cavity structures cannot be easily identified. However, analyzing and comparing different planes of TB-ORG (Fig. 3) revealed e.g. an elongated cavity spanning throughout the organoid (Fig. 3a, YX plane). The intersections of the yellow lines mark the same position (the center of the elongated cavity) in all three planes (Fig. 3a-c). The white arrows in Fig. 3a and b denote the marking thread at the same position in different planes, cross section/YX and en face/XZ plane, respectively. The XZ plane in Fig. 3b was taken at the height of the horizontal yellow line in Fig. 3a, whereas the cross section of YX plane is at the horizontal line depicted in Fig. 3b. The asterisks in Fig. 3b and c are also showing the marking thread at the side of the organoid. Fig. 3d shows a 3D rendering of the TB-ORGs, demonstrating that the bigger organoid was not round but had a more oval-shaped morphology, and the exact constellation of both TB-ORGs to each other in the Matrigel could be delineated. Fig. 3e (and Supplemental Video 1) is a rendering of the elongated cavity, whose orientation within the organoid can be appreciated in Fig. 3f (and Supplemental Video 2).
Taken together, OCT imaging of living TB-ORGs enables to evaluate morphology and orientation of the organoid as well as cavities within TB-ORGs, further indicating that this technology allows for better visualization and characterization of inner TB-ORG structures compared to standard microscopy.

C. OCT image aligns with STB formation
Literature showed that in TB-ORG, vCTBs differentiate into hormone producing STBs toward the centers of the organoids [5]. We therefore assessed whether the cavities seen in OCT images would align to STB formation in TB-ORG. To provide an adequate differentiation control we treated the organoids with or without p38 mitogen-activated protein kinases (MAPK) signaling inhibitor (Fig. 4). Upon p38 MAPK inhibition, differentiation of STBs, and consequently hormone production, is hampered [40]. Organoids were either treated or not treated (control) with 10 µM p38 inhibitor for 10 days. Then, TB-ORG were imaged without sample preparation with OCT and subsequently either fixed and embedded in paraffin or further processed for RNA as well as protein analyses. Fig. 4 upper panel shows control TB-ORG samples without p38 inhibitor. OCT images and H&E staining revealed the existence of cavities in the central region of the organoids. Immunofluorescence analyses further showed that those cavity structures lack E-cadherin expression and were positive for syndecan-1 (SDC1), two hallmarks of STBs indicating that these cavities indeed recapitulate STB-associated structures [41]. SDC1 further associated well with the high intensity bridges, further confirming that the observed cavities by OCT resembled STB structures. In contrast, p38 inhibitor treated TB-ORG (Fig. 4a, lower panel) failed to show those cavities, in both OCT obtained images and H&E stainings. Immunofluorescence analyses of those samples showed continuous E-cadherin staining throughout the organoid and hardly exhibit SDC1. Consistent with these findings were the results of other syncytiotrophoblast markers on mRNA and protein level (Fig. 4b, c). Western blot and qPCR analyses revealed that the STB marker ENDOU was only detected in the control TB-ORGs.
Further, the pregnancy-specific hormone chorion gonadotrophin β (CGβ) was highly expressed when organoids where not treated with STB differentiation inhibitor and undetectable upon p38 MAPK inhibition (Fig. 4b, c).
In summary, we suggest that the high scattering bridges in the inner part of the TB-ORG represent STB structures and our studies with p38 inhibitor confirmed that OCT images align with the differentiation status of trophoblasts within TB-ORG.

D. Organoid segmentation shows diverse degree of differentiation independent of size
To enable future longitudinal monitoring and staging of TB-ORGs according to STB and cavity formation, we sought to quantify the volume of the whole organoid and its cavities. Using ImageJ, three raters manually segmented non-treated whole organoids (n=4) including the cavities or cavities only. Interestingly, placenta organoids exhibited diverse shapes and structures. Fig. 5a-d (Supplemental Video 3) show an irregularly shaped organoid with a large, undifferentiated part and a differentiated, cavity-bearing fraction at the side. To quantify the size of the whole organoid, each TB-ORG cross section was evaluated, thereby the borders of the organoid were circumvented (Fig. 5b). Similarly, each cavity was segmented (Fig. 5c). The cavity in comparison to the whole organoid were calculated in percentage and plotted against the corresponding whole organoid size in Fig. 5e. The analyzed organoids were heterogenous in volume, and irrespective of their size, they bear different percentage range of cavities (TB-ORG1: 8.8 to 10.8 %; TB-ORG2: 21.9 to 30.0 %; TB-ORG3: 43.1 to 55.1 %; TB-ORG4: 41.6 to 52.4 %). Of note, TB-ORG3 and TB-ORG4 were the small and big organoids, respectively, displayed in Fig. 1 and Fig. 3. OCT revealed herein that TB-ORGs have heterogenous morphologies and dissimilar differentiation states even within one TB-ORG culture.

IV. DISCUSSION
The human placenta is the connecting organ between the mother and the fetus. Correct placental formation, differentiation, and function display an indispensable prerequisite for uncomplicated pregnancies. Hence, reliable cell culture models allowing to investigate trophoblast physiology are required. Placenta-derived trophoblast organoids provide a functional experimental model to study not only physiological changes in the placenta during early pregnancy, but also to investigate maternal-fetal transmission of drugs, proteins, and pathogens [10], [11]. However, current imaging methods characterizing inner TB-ORG structures -and thus the status of STB formation -require the termination of experiments. Fluorescence imaging methods including light sheet microscopy require exogenous contrast agents for molecular specificity. In contrast, OCT is a label-free and non-invasive volumetric imaging method allowing to assess unperturbed specimen derived from primary tissue. Micro-CT provides deeper imaging penetration than OCT, but requires terminal sample preparation for organoids and has a lower resolution, therefore overlooking morphological features which OCT detects [20]. Serial scanning electron microscopy delivers sub micrometer resolution, however, only thin samples after heavy metal staining, dehydration, and resin-embedding can be imaged.
OCT has e.g. been applied to stem-cell-derived organoids of intestinal and retinal origin [16], [20], [29], [30], [31]. Retinal layers are aimed to be mimicked in organoids and OCT has been used to investigate their structure in vitro and in in vivo transplantation studies [20], [29], [30]. Further, intestinal organoids derived from induced pluripotent human stem cells were also characterized via OCT. These organoids form a central lumen which does not provide signal and is enclosed by epithelial cells providing the high scattering contrast [31].
However, OCT has not yet been used in trophoblast organoids. Hence, in this work we imaged TB-ORGs of three donors with OCT to determine inner structures and, moreover, TBME-01565-2020 7 align these to the differentiation status of STBs in these organoids. Therefore, we analyzed placenta organoids with our custombuilt spectral domain OCT instrument. Similar to the intestinal organoid lumen [31], cavities in the trophoblast organoids were displayed by no or low OCT signal. In TB-ORG, differentiation of STBs occurs toward the center of organoid structures which is in stark contrast to in vivo placental tissue where STBs are on the outside in contact with maternal blood [10]. Nevertheless, encapsulated STBs in trophoblast organoids were functional and secreted the hormone hCG [10], [11]. In this study we demonstrate that the centrally-located differentiated STBs in TB-ORG align to the cavities shown via OCT. Since OCT is based on scattering contrast, it is likely that the dark cavities predominantly represent cell-free areas. However, some cavities seem to have scattering properties, a possible indication that they are filled with secreted products of STBs (hormones, transport vesicles) accumulating in the center of organoids [10], [11]. Tang et al. investigated the origin of interference pattern i.e. OCT signal on a subcellular level in glioblastoma cells grown in a semi-solid gel matrix. Thereby, scattering contrast derived from actin filaments, mitochondria, plasma membrane, boundary between cytoplasma and nucleus due to change in refractive index of their lipid, nucleic acid, and protein contents [42]. This was concordant with our results, as OCT scattering was associated to nuclei staining (H&E; DAPI in immune fluorescence) and cellular membrane (E-cadherin, SDC1).
We further tested whether OCT allows to non-invasively determine STB differentiation status in TB-ORGs. In literature, p38 signaling was shown to regulate CTB differentiation to STB, in which mononuclear CTB undergo cell fusion thus forming the syncytium [43], [44]. Therefore, organoids were cultured with the p38 inhibitor SB202190 for 10 days to block trophoblast differentiation and compared to non-treated and therefore differentiated TB-ORGs. The latter showed multinucleated cells towards the center of the organoids, which were positive for the STB membrane marker SDC1 [45]. H&E, immuno stainings and qPCR as well as Western blot analyses revealed that cavities and markers for STB differentiation (CGb, ENDOU, SDC1) were only detectable in non-treated TB-ORGs. These results fit to the images taken by OCT on nonfixed, in situ samples during cultivation. Therefore, we conclude that non-invasive, label free OCT enables to qualitatively relate cavity formation to the differentiation of STBs in TB-ORGs.
Additionally, TB-ORG do not only differentiate into STBs but can also be triggered to produce outgrowing extravillous trophoblasts invading the matrigel. Hence, OCT would also be useful depicting the exact orientation of outgrowing extravillous trophoblasts in 3D [10], [11].
We also investigated the differences between brightfield microscope and OCT images. The Bland-Altman plot is a statistical visualization method assessing agreement between two methods [39]. Here, we analyzed the difference between the images obtained via brightfield microscope and OCT compared to the mean organoid size. OCT en face averaged TB-ORG images were nonsignificant smaller than when measured with the brightfield microscope. Overall, the Bland-Altman plot and paired samples t-test show the agreement between the two imaging modalities.
TB-ORGs have variable outer and inner morphology which are not resolvable and quantifiable via brightfield images. Moreover, as shown in literature, using measured diameter and using the spherical volume formula of 3D cellular structures result in a size overestimation in comparison to voxel-based measurements [33]. In contrast, 3D OCT imaging not only enables to appreciate the volumetric size of the overall organoid, but optical dissection allows to quantify inner structure and cavities of the organoid. Manual segmentation then allowed us to identify differentiation variability in human trophoblast organoids. Similar observation was reported for human retinal organoids [16]. Possible cause may be the requirement of a 3D extracellular matrix which often is the semi-solid gel Matrigel (Corning), which is extracted from mouse sarcoma and therefore is susceptible to batch-to-batch production variation [46]. This may prompt further studies to evaluate possible developmental staging approaches to improve experimental reproducibility. Moreover, the same organoid can be repeatedly imaged over a longer time course. This may bring more insights towards a quantitative link between OCT results and differentiation status. Further, automatic segmentation would be useful for unbiased organoid identification and detailed analyses [47], [48].

V. CONCLUSION
Our results demonstrate the value of OCT in placenta organoid research. In situ brightfield microscopy only allows to monitor the outer organoid morphology. For characterizations of the inner architecture conventional methods such as H&E and fluorescence staining are used, but are endpoint analyses. In contrast, using OCT enables to visualize and quantify structures which is congruent to the differentiation status of organoids in situ and without sample preparation. These findings will simplify the experimental design and may become an invaluable tool in placental research. Moreover, 3D rendering allows to better appreciate the TB-ORG morphology and orientation, which will also aid further studies focusing on cellular outgrowth of extravillous trophoblasts from the trophoblast organoid. Taken together we suggest that OCT imaging might be routinely included in organoid research to facilitate the monitoring of in situ samples. APPENDIX Supplemental Video 1: 3D cavity Supplemental Video 2: Cavity within organoid structure Supplemental Video 3: Cross-section of TB-ORG1