Label-free three-photon imaging of intact human cerebral organoids for tracking early events in brain development and deficits in Rett syndrome

  1. Murat Yildirim  Is a corresponding author
  2. Chloe Delepine
  3. Danielle Feldman
  4. Vincent A Pham
  5. Stephanie Chou
  6. Jacque Ip
  7. Alexi Nott
  8. Li-Huei Tsai
  9. Guo-Li Ming
  10. Peter TC So
  11. Mriganka Sur  Is a corresponding author
  1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, United States
  2. Department of Neuroscience, Cleveland Clinic Lerner Research Institute, United States
  3. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, United States
  4. School of Biomedical Sciences, The Chinese University of Hong Kong, China
  5. Department of Neuroscience and Mahoney Institute for Neurosciences, Institute for Regenerative Medicine, Perelman School of Medicine, University of Pennsylvania, United States
  6. Deparment of Mechanical Engineering, Massachusetts Institute of Technology, United States

Abstract

Human cerebral organoids are unique in their development of progenitor-rich zones akin to ventricular zones from which neuronal progenitors differentiate and migrate radially. Analyses of cerebral organoids thus far have been performed in sectioned tissue or in superficial layers due to their high scattering properties. Here, we demonstrate label-free three-photon imaging of whole, uncleared intact organoids (~2 mm depth) to assess early events of early human brain development. Optimizing a custom-made three-photon microscope to image intact cerebral organoids generated from Rett Syndrome patients, we show defects in the ventricular zone volumetric structure of mutant organoids compared to isogenic control organoids. Long-term imaging live organoids reveals that shorter migration distances and slower migration speeds of mutant radially migrating neurons are associated with more tortuous trajectories. Our label-free imaging system constitutes a particularly useful platform for tracking normal and abnormal development in individual organoids, as well as for screening therapeutic molecules via intact organoid imaging.

Editor's evaluation

This manuscript will be of interest to stem cell and developmental biologists who aim to use newly emerging brain organoid models to understand the structure and function of the developing human brain. It presents a technological advance in imaging and describes an innovative method for labeling and tracking cells within organoids to enable the assessment of dynamic processes within the intact organoid. The method is validated in a disease model and addresses a challenge in the field of human stem cell modeling of assessing cells within the 3D structure.

https://doi.org/10.7554/eLife.78079.sa0

Introduction

Human cerebral organoids derived from embryonic or induced pluripotent stem cells are unique in their ability to recapitulate early events of embryonic brain development (Lancaster et al., 2013). These spheroid structures contain progenitor-rich zones around ventricle-like cavities akin to ventricular zones (VZ) from which neuronal progenitors migrate radially to generate the cortical plate (CP) (Lancaster et al., 2013; Qian et al., 2016; Paşca et al., 2015; Mariani et al., 2012; Kadoshima et al., 2013). Human cerebral organoids also recapitulate gene expression programs of the fetal cortex (Qian et al., 2016; Velasco et al., 2019; Quadrato et al., 2017; Pollen et al., 2019; Luo et al., 2016; Camp et al., 2015) as well as the fetal brain epigenome (Luo et al., 2016). Thus, they have been used to model neurogenesis-relevant human pathologies such as microcephaly (Lancaster et al., 2013; Zhang et al., 2019), licencephaly (Bershteyn et al., 2017), heterotopia (Klaus et al., 2019), Zika virus infection (Qian et al., 2016; Garcez et al., 2016; Dang et al., 2016), and idiopathic autism (Mariani et al., 2015). Rett Syndrome (RTT) is an X-linked neurodevelopmental disorder caused by mutations in the gene encoding Methyl CpG binding protein 2 (MeCP2). MeCP2 is a pleiotropic regulator of gene expression and impacts multiple components of brain development and function (Ip et al., 2018). Structural deficits described in postmortem RTT human brains include reduced cortical thickness, cell size and dendritic arborization (Armstrong et al., 1995; Bauman et al., 1995) and reduced cerebral volume in MR imaging of RTT patients (Carter et al., 2008). These deficits are paralleled by reductions in dendritic arborization, soma size and spine density described in RTT mouse models (Fukuda et al., 2005; Kishi and Macklis, 2004; Shahbazian et al., 2002; Smrt et al., 2007). 2D human stem cell models of RTT, generated by reprogramming of patient cells or genome editing, have revealed deficits in human RTT neurons including aberrant transcription (Lyst and Bird, 2015; Li et al., 2013; Chen et al., 2013; Gomes et al., 2020; Trujillo et al., 2021), impaired neuronal maturation and electrophysiological function (Li et al., 2013; Tang et al., 2016; Kim et al., 2011; Farra et al., 2012), and up- or down-regulation of key signaling pathways and activity-related genes (Li et al., 2013). Using fixed tissue slices from 3D cerebral organoids, we have recently demonstrated impaired proliferation of the progenitor pool and delayed maturation and presumed migration of neurons (Mellios et al., 2011), consistent with human postmortem deficits. However, the use of fixed tissues did not allow cell tracking and analysis of parameters such as speed and trajectory of neuronal displacement; thus the dynamics and cellular mechanisms of the presumptive migration deficit remain to be characterized. Significantly, live cell time-lapse imaging of radial migration of MECP2-deficient neurons from the VZ to CP inside intact live 3D organoids has not been performed so far.

The majority of analyses of cerebral organoids have been performed in sectioned tissue (Bershteyn et al., 2017; Klaus et al., 2019; Li et al., 2011; Lancaster and Knoblich, 2014; Andersen et al., 2020; Bagley et al., 2017; Karzbrun et al., 2018; Lancaster et al., 2017; Miura et al., 2020), although recent progress has been made in studying migration using labeled neurons in organoid models (Klaus et al., 2019; Bagley et al., 2017; Birey et al., 2017; Birey et al., 2022; Xiang et al., 2017; Bajaj et al., 2021), including assembloid preparations (Birey et al., 2017; Birey et al., 2022; Xiang et al., 2017). However, labeled live cell imaging generally requires phototoxic dyes and limited incubation conditions, and results as well in labeling mostly peripheral neurons and cells which are not ideal for long-term imaging and for finding ventricular like zones in 3D intact organoids. Even with viral labeling, subtle damage to neuronal integrity can be sometimes detected (Yildirim et al., 2019). Optically, cerebral organoids appear opaque due to the optical density of neuronal tissues and the large number of apoptotic cells in the center of 3D intact conditions (Lancaster and Knoblich, 2014). Thus, label-free and high-resolution deep-tissue imaging are ideally required to perform intact organoid imaging.

Third-harmonic generation (THG) is an intrinsic signal which results from tripling of the frequency of the excitation wavelength. THG signal is either generated at structural interfaces such as local transitions of the refractive index or inside the materials whose third order nonlinear susceptibility χ3 is higher. Thus, THG excitation of biological tissues occurs predominantly at interfaces that are formed between aqueous interstitial fluids and lipid-rich structures, such as cellular membrane (Rehberg et al., 2011) and lipid bodies (Débarre et al., 2006). Both standard three-photon fluorescence microscopy and THG microscopy rely on three-photon interaction between ultrashort pulses and tissues. Three-photon fluorescence microscopy and THG microscopy has been recently used to perform structural and functional brain imaging in anesthetized and awake mice (Yildirim et al., 2019; Ouzounov et al., 2017). These studies are based on utilizing a green (GCaMP) genetically engineered calcium indicator (exogenous fluorophores) with their excitation wavelength (1300 nm) which provides peak absorption cross-sections for this indicator. In these three-photon fluorescence microscopy studies, three photons with enough peak power at these excitation wavelengths excite electrons from ground state to excitation states of this indicator. Then, these electrons release their energy while they return to their ground states. During this relaxation, they release emitted photons which have a large 1/e2 bandwidth of fluorescence emission (70 nm for GCaMP6s/f). Therefore, an exogenous label is required to perform three-photon fluorescence structural or functional brain imaging. On the other hand, THG microscopy does not need any label for structural three-photon brain imaging. In addition, three photons with enough peak power at the excitation wavelengths excite electrons from ground state to virtual state so that these electrons do not lose any energy when they come back to the ground state. Therefore, they release emitted photons which have a small 1/e2 bandwidth of emission (~20 nm). Therefore, the emission spectrum of THG imaging occurs at exactly 1/3 of the excitation wavelength.

Overall, three-photon fluorescence microscopy is valuable for performing structural and functional brain imaging with fluorescent dyes for which the excitation wavelengths are limited by the peak absorption cross-sections of these fluorophores. However, three-photon fluorescence imaging is prone to phototoxicity and photobleaching particularly with high peak intensity pulses (Yildirim et al., 2019). In contrast, THG microscopy provides label-free structural brain imaging (Yildirim et al., 2020) without problems of phototoxicity and photobleaching - which are its strengths for long-term live cell imaging of cerebral organoids. Optimized designs using high power lasers and high sensitivity photomultipliers have enabled the application of THG microscopy to 3D tissue microscopy by improving the imaging depth in tissues with varying scattering coefficients (Débarre et al., 2006; Yildirim et al., 2015; Yelin and Silberberg, 1999; Gualda et al., 2008). THG microscopy has been applied recently to the non-invasive monitoring of human adipose tissue (Chang et al., 2013), cell nuclei and cytoplasm in liver tissue (Lin et al., 2014) and subcortical structures within an intact mouse brain (Yildirim et al., 2019; Ouzounov et al., 2017; Yildirim et al., 2020). THG imaging provides the advantages of micron scale resolution, label-free imaging, deep tissue penetration depth, and non-destructiveness.

Here, we describe a custom-made three-photon microscope with optimized laser and optics design to perform label-free THG imaging of intact organoids, and its use in imaging organoids generated from RTT and control patients to assess early events of early brain development. We demonstrate that high resolution label-free THG signal can be collected from intact cerebral organoids. Comparing organoids from RTT patients and isogenic controls, we show that the ventricular zone in mutant organoids has larger volume, larger surface area, and lower ventricular thickness compared to isogenic control organoids. Finally, label-free live cell imaging provides a unique way of observing neuronal migration inside the VZ/SVZ and CP of developing organoids without any phototoxicitiy or photobleaching, and reveals that the smaller ventricular thickness in mutant organoids is associated with shorter migration distances, more tortuous trajectories, and slower migration speeds of radially migrating neurons.

Results

Optimized system for label-free THG imaging of intact cerebral organoids

We developed a three-photon microscope (Yildirim et al., 2019) and modified it (Figure 1A–C, see Materials and methods) to perform label free THG imaging of intact, uncleared, fixed and live organoids. Since low average laser power is essential for live cell imaging, we focused on optimizing laser and microscopy parameters to reduce the average power requirement for intact organoid imaging. First, we utilized moderate repetition rate (<1 MHz) to maximize the energy per pulse and reduce the average laser power requirement for THG imaging. Second, we minimized the pulse width on the sample by building an external pulse compressor. The pulse width on the sample was reduced to 27 fs in the deepest part of the organoids (see Materials and methods) enabling us to further reduce the average power requirement by two- to threefold compared to other three-photon studies in the literature (Yildirim et al., 2019; Ouzounov et al., 2017; Weisenburger et al., 2019; Figure 1—figure supplements 1 and 2). Finally, we designed all intermediate optics in the excitation and emission path to maximize the generation and collection efficiency of THG signal (Figure 1—figure supplement 3). This optical design procedure enabled us to reduce the spherical aberration in the system and increase the collected signal by twofold compared to off-the shelf optics (Ouzounov et al., 2017; Figure 1—figure supplement 4).

Figure 1 with 4 supplements see all
Three-photon microscope and imaging system.

(A) Femtosecond laser pulses from a pump laser (1045 nm) were pumped through a noncollinear optical parametric amplifier (NOPA) to obtain 1300 nm excitation wavelength. Power control of these laser pulses was performed using a combination of a half-wave plate (HWP) and polarizing cube beam splitter (PCBS). A quarter-wave plate (QWP) was used to control the polarization state of the laser pulses to maximize the third harmonic generation (THG) signal. Laser beams were scanned by a pair of galvanometric scanning mirrors (SM), and passed through a scan lens (SL) and a tube lens (TL) on the back aperture of a 1.05 NA, 25×objective. The sample (S) was placed on a two-axis motorized stage, while the objective (OL) was placed on a one-axis motorized stage for nonlinear imaging. Emitted light was collected by a dichroic mirror (CM1), collection optics (CO), laser blocking filters (BF), and nonlinear imaging filters (F) and corresponding collection optics (COA, COB, and COC) for each photomultiplier tube (PMT A, PMT B, and PMT C). (B) Fixed intact organoids were placed inside imaging chamber gaskets on top of the bottom glass. Gaskets were filled with Hprotos and covered with a cover glass. (C) Live organoids were placed on top of the bottom plate of the incubator. Imaging gaskets were placed to secure organoids and cell medium was applied to fill the gaskets. Top plate was placed on top of the gaskets and the microincubator closed with 6 screws. Cell medium mixed with 5% CO2, 5%O2, and balanced N2 was pumped to the chamber; incubator temperature was set to 37 °C.

THG signals from intact cerebral organoids reveal distinct zones

Cerebral organoids were derived from two pairs of isogenic iPSC clonal lines (see Materials and methods). One pair (Line 1) was produced as previously described (Mellios et al., 2011) from one individual with RTT harboring a heterozygous single nucleotide deletion (frameshift 705delG) in the transcriptional repression domain of MeCP2. Because of the monoallelic expression of X-chromosome genes and clonal selection of iPSCs, the ‘RTT-WT’ line expressed exclusively the wild-type allele of MECP2, while the “RTT-MT” line expressed exclusively the mutated allele. The second pair of isogenic iPSC clonal lines (Line 2), produced as previously described (Nott et al., 2016), was from another individual with RTT harboring a heterozygous single nucleotide mutation MECP2R306C located in the transcriptional repression domain (TRD) or more specifically NCoR/SMRT interaction domain (NID) of MeCP2, which selectively blocks its interaction with the NCoR/SMRT complex (Lyst and Bird, 2015; Lyst et al., 2013). Because the MECP2R306C-derived iPSCs had skewed X-inactivation, and expressed only the MECP2R306C allele, the MECP2R306C RTT-WT line was generated from the MECP2R306C iPSCs using CRISPR/Cas9-mediated gene editing to correct the R306C mutation (Nott et al., 2016).

First, we established that cortex-like substructures could be distinguished in the organoids by cell density differences. In accordance with previous studies (Lancaster et al., 2013; Qian et al., 2016; Velasco et al., 2019), after 35 days of culture, the cerebral organoids demonstrated formation of well-organized structures composed of a ventricular zone (VZ)-like layer (KI67-positive proliferative cells) around a ventricle-like cavity and a surrounding cortical plate (CP)-like structure (Tuj1 or DCX-positive neurons) (Figure 2A–B, see details of the generation of organoids in Materials and methods). Notably, CP and VZ substructures demonstrated distinct cell density and morphology as shown by nuclear (7AAD) stain and plasma membrane specific dye (WGA) (Figure 2C). The VZ region, densely populated by neural progenitor cells and few immature neurons, could be markedly distinguished from the CP substructure with a less dense population of neurons. Thus, we subsequently used the cell density to localize the CP structures in three-photon fluorescence/THG imaging experiments.

Figure 2 with 5 supplements see all
Label-free Third Harmonic Generation (THG) imaging of intact cerebral organoids.

(A-B) Confocal imaging of immunolabeled 2D cerebral organoid slices where cerebral organoids present well-organized spherical structures located around ventricle-like cavities. Around these cavities the ventricular zone-like substructure (VZ), densely populated with progenitors (KI67 + cells) and young migrating neurons (TUJ1 + cells), can be distinguished from the cortical plate-like substructure (CP) with a less dense population of neurons (Line 1, TUJ1, DCX + cells). (C) Cortical plate and ventricular zone substructures presented distinct cell density and morphology (Line 2) as shown by nuclear (7AAD dye) and plasma membrane labeling (WGA dye). (D–E). Label-free THG imaging from intact fixed 3D cerebral organoids (Line 2) show distinct signals from the cortical plate and ventricular zone, as confirmed by the nuclear labeling (7AAD dye) observed with the same setting but using three-photon epifluorescence (3PEF). THG signal seems to occur at the plasma membranes in the VZ and to be brighter in the CP substructure. Scale bars are 100 µm.

Next, we demonstrated that high-resolution label-free THG signal could be collected from intact cerebral organoids. Nuclear stain was used as a control to localize the high cell density VZ from the low cell density CP (Figure 2D). Strikingly, THG imaging provided a clean delineation between VZ and CP as THG signal intensity was lower in the former area than the latter (Figure 2C–D). The intrinsic THG signal mostly arose distinctly from cell membranes. Consequently, the THG signal nicely revealed the typical radially elongated and bipolar radial glia cell morphology as well as the round-shaped somata of neurons (Figure 2D–E). Additionally, neurons in the CP and newborn neurons inside the VZ showed high intensity somatic signals (Figure 2D–E). Finally, we performed THG imaging of an intact organoid and then stained it with neuronal marker DCX. We were able to find the same field of view in our stained organoids and compare it with our THG imaging results (Figure 2—figure supplement 1). Our results confirmed that strong THG signal is mostly detected in the CP (DCX-rich, low cell density), rather than the VZ (DCX-negative, high cell density) (Figure 2—figure supplement 1).

To further validate that the intrinsic THG signal arose from neurons, we expressed green fluorescent protein (GFP) in control organoids by electroporation (see Materials and methods). We previously demonstrated that >90% of the GFP-positive cells in electroporated organoids are neurons (Delepine et al., 2021). Three-dimensional imaging of cells in organoids (~2 mm thickness) revealed that both WT and MT organoids in fixed and live conditions had >85% overlap between GFP and THG signals (Figure 2—figure supplements 25, see Methods).

Three-dimensional rendering of a~450-image z-stack with 2 µm increments revealed our system’s ability to perform high-resolution imaging (Figure 3—videos 1–2; Figure 3A–B, left). We performed depth-resolved imaging of a 250 × 250 × 900 µm3 region selected to include a cortex-like structure, centered around the ventricle-like cavity, for both RTT-WT (Figure 3A) and RTT-MT organoids (Figure 3B). The strong THG signal produced at the interface between an organoid’ssurface and the cover glass helped us determine boundaries of the intact organoids. Since our field of view (FOV) was 250 µm, serial imaging of multiple sites was necessary to delineate individual CP regions. Notably, we were able to collect high-resolution THG signal from the entire 900 µm-deep regions, highlighting the performance of three-photon fluorescence and THG detection for deep-tissue imaging, and in particular, intact uncleared whole organoid imaging. The maximum imaging depth was only limited by the working distance of the objective, which was approximately 2 mm in our experiments (Figure 3—video 3).

Figure 3 with 3 supplements see all
3-D label-free THG imaging of fixed organoids and characterizing their extinction lengths.

(A–B) 3D reconstruction of label-free THG signal from ventricular regions highlighted in Figure 2D–E. THG allows imaging of deep structures (here, till 900 µm). Scale bars represent 100 µm. (C–D) Characterization of the extinction lengths of a WT and a MT fixed organoid at 1300 nm excitation wavelength. Semi-logarithmic plot for ratio of PMT signal and cube of laser power with respect to imaging depth for third harmonic generation (THG) imaging. Slope of these curves result in 162.5 µm and 148.2 µm extinction lengths for the WT and MT fixed organoids, respectively. (E) Comparison of the extinction lengths of six WT and MT fixed organoids (Lines 1 and 2) at 1300 nm excitation wavelength. The average extinction length of wild-type (WT) organoids is significantly higher than that of mutant (MT) organoids (n=6 organoids, p<0.05, t-test, error bars are standard error of the mean [SEM]).

Label-free imaging of fixed cerebral organoids reveals differences between RTT and isogenic control organoids

First, we characterized the extinction (combined scattering and absorption) lengths of WT and MT organoids. Our label-free THG imaging characterization shows that the extinction lengths of a fixed WT and a MT organoid are 162.5 µm and 148.2 µm, respectively which are approximately half of the extinction length of a primary visual cortex of an awake mouse brain (Yildirim et al., 2019; Yildirim et al., 2020; Figure 3C–D). In other words, fixed WT and MT organoids are at least two times more scattering than the mouse visual cortex at 1300 nm excitation wavelength. We also examined different WT and MT organoids for comparison (Figure 3E). Our results show that average extinction length of WT organoids (152.1±3.5 µm) is significantly higher than that of MT organoids (125.9±4.9 µm, p<0.005, n=6 organoids, t-test).

To study the 3D organization of VZ structures in RTT-WT and RTT-MT organoids, depth-resolved serial imaging of whole organoids was performed automatically with a custom algorithm to reduce the total imaging duration (see Materials and methods). We acquired both 7AAD three-photon fluorescence and THG signal from RTT-WT (Figure 4A) and RTT-MT (Figure 4B) organoids. Then, we determined the boundaries of individual VZ structures using THG signal and rendered individual cortical regions in both RTT-WT (Figure 4A) and RTT-MT (Figure 4B) organoids (see Materials and methods, see Figure 4—video 1, see Figure 4—figure supplement 1). First, we quantified the total volume of each region of interest and found slightly (but not significantly) higher volumes in mutant organoids than in control ones (1.13±0.18 × 107 µm3 for RTT-WT, and 1.39±0.15 × 107 µm3 for RTT-MT organoids, p=0.3013; n=10 organoids, t-test; Figure 4C). However, the surface area of the VZ region in RTT-MT organoids was significantly higher than that in RTT-WT organoids (5.70±0.75 × 105 µm2 for RTT-WT, and 11.0±0.71 × 105 µm2 for RTT-MT organoids, p<0.0001; n=10 organoids, t-test; Figure 4C). In addition, the ratio of volume to area was used as a relative measure of the VZ thickness. The progenitor-rich structures in the RTT-MT organoids had significantly lower unit thickness relative to RTT-WT organoids (23.11±4.12 µm for RTT-WT, and 9.99±0.27 µm for RTT-MT organoids, p<0.0001, n=10 organoids, t-test; Figure 4C). Finally, the number of ventricles was significantly higher in RTT-MT organoids (12.3±0.7 for RTT-WT, and 22.0±1.3 for RTT-MT organoids, p<0.0001, n=10 organoids, t-test; Figure 4C).

Figure 4 with 6 supplements see all
Three-dimensional characterization of control (RTT-WT) and mutant (RTT-MT) VZ regions in intact organoids.

(A, B). THG signal acquired from the whole RTT-WT (A) and RTT-MT (B) organoids (Line 1) (left) was segmented into individual ventricular regions (right). (C) (left; top) RTT-MT organoids showed slightly higher VZ volume than those in RTT-WT organoids (1.13±0.18 × 107 µm3 for RTT-WT, and 1.39±0.15 × 107 µm3 for RTT-MT organoids, p=0.3013; n=10 organoids, t-test). (left; bottom) The surface area of the VZ region in RTT-MT organoids was significantly higher than that in RTT-WT organoids (5.70±0.75 × 105 µm2 for RTT-WT, and 11.0±0.71 × 105 µm2 for RTT-MT organoids, p<0.0001; n=10 organoids, t-test). (right; top) The VZ thickness (V/A ratio) in RTT-MT organoids was significantly lower than that in RTT-WT organoids (23.11±4.12 µm for RTT-WT, and 9.99±0.27 µm for RTT-MT organoids, p<0.0001, n=10 organoids, t-test). (right; bottom) The number of ventricles was significantly higher in RTT-MT organoids than that in RTT-WT organoids (12.3±0.7 for RTT-WT, and 22.0±1.3 for RTT-MT organoids, p<0.0001, n=10 organoids, t-test). The data are collected from two lines (Line 1 and Line 2, see Figure 4—figure supplements 25). Error bars are 90% of the confidence interval (CI).

In summary, using label-free intrinsic THG imaging we show that RTT-MT organoids contain an increased number of VZ structures with larger surface area and volume, but lower VZ thickness, compared to RTT-WT organoids (values for each line separately are shown in Figure 4—figure supplements 25).

Label-free imaging of live cerebral organoids captures lower migration speed and displacement in RTT organoids

Direct visualization and time-lapse characterization of neuronal migration dynamics with label-free methods have not been examined so far. Thus, we tracked migrating neurons in real time using THG microscopy. We performed live imaging of RTT-WT and RTT-MT organoids at 35 days in vitro (DIV) for 12–96 hr, with a volumetric acquisition interval of 20 min (Figure 5—figure supplement 1, see Materials and methods). Given the sparseness and high signal to noise ratio, we were able to track cells harboring a bright somatic THG signal in both RTT-WT and RTT-MT organoids (Figure 5A, see Figure 5—video 1Figure 5—videos 1; 2, see Materials and methods). Although the tracked cells in both RTT-WT and RTT-MT organoids migrated radially from the inner VZ to the outer VZ, RTT-MT cells showed decreased displacement, more tortuous trajectories, and reduced linear speed than RTT-WT cells (Figure 5B and C, also see Figure 6). Specifically, we observed in RTT-WT organoids an average migration speed of 23.4±0.2 µm/hr (n=6 organoids, N=208 cells). The average speed for each line was 20.5±0.1 µm/hr for Line 1 and 25.6±0.2 µm/hr for Line 2 (Figure 5—figure supplement 2). These average speed values are comparable to the speed recorded in neurons migrating off a human wild-type cerebral organoid onto a Matrigel-coated surface (Bershteyn et al., 2017) and in ferret cortical explants (Gertz and Kriegstein, 2015). In contrast, RTT-MT cells had a reduced average speed of 13.3±0.1 µm/hour (n=6 organoids, N=217 cells). The average speed for each line was 11.4±0.1 µm/hour for Line 1 and 14.5±0.1 µm/hr for Line 2 (Figure 5—figure supplement 2). Collectively, the average speed in WT organoids was significantly higher than the average speed in MT organoids (Figure 5C). In addition, we compared migration speeds of individual cells with respect to their location in each organoid as well as between organoids (Figure 5—figure supplement 3). In conclusion, we did not find any statistically significant difference in migration speeds of individual cells at different locations in the same organoid (Figure 5—figure supplement 3A-B) as well as between organoids of the same type (Figure 5—figure supplement 3C-D). We also quantified the straightness of trajectory by taking the ratio of displacement and total trajectory length. With a straightness value close to 1, RTT-WT cells (straightness value of 0.881±0.004) had significantly more straight trajectories than RTT-MT cells (0.658±0.006) which showed much more nonlinear migrating patterns (Figure 5A and C).

Figure 5 with 9 supplements see all
Dynamics of neuronal migration in RTT-WT and RTT-MT organoids.

(A) Representative time-lapse THG images of migrating cells in the ventricular zone in RTT-WT (top) and RTT-MT (bottom) organoids (Line 1) in every 2 hr. The migrating trajectory of each representative cells is shown on the right panel to the time-lapse images. Color bar represents the instantaneous speed of these cells. (B) Representative track speed over time for cells in RTT-WT (left) and RTT-MT (right) organoids. (C) Summary of displacement, average migration speed and straightness of the migration trajectory (n=6 RTT-WT and 6 RTT-MT organoids, 200 cells for WT and 210 cells for MT. Scale bar is 50 µm. **, p<0.01; ***, p<0.001; ****, p<0.0001, t-test). Error bars are 90% of the confidence interval (CI).

Dynamics of neuronal migration in RTT-WT and RTT-MT organoids.

Representative three-dimensional THG images of the live RTT-WT (A) and RTT-MT (C) organoids from Line 2. Dashed red lines show the planes of imaging presented in (B) and (D). Representative time-lapse, depth-resolved THG images of migrating cells in the VZ region for RTT-WT (B) and RTT-MT (D) organoids in every 2 hr. The migrating trajectory of each representative cells is shown on the bottom right panel to the time-lapse images. Color bar represents the instantaneous speed of these cells. (E) Representative track speed over time for cells in RTT-WT and RTT-MT organoids. (F) Summary of displacement, average migration speed and straightness of the migration trajectory. Scale bar is 50 µm.

Since the average speed of RTT-MT was lower and these cells had a less straight trajectory, net displacement of MT cells (133.2±2.4 µm) was significantly lower than that of WT cells (244.4±3.6 µm) (Figure 5C). The straightness and net displacement values of RTT-WT and RTT-MT in both lines showed similar trends (Figure 5—figure supplements 4 and 5). To check whether cells in the organoids were still viable after imaging session of ~18 hr, we performed Caspase3 immunostaining for both imaged and not-imaged control organoids as well as a 90-day-old organoid with a necrotic core as a positive control (Figure 5—figure supplement 6). Our results show that imaging live organoids in the mini-incubator with three-photon microscopy did not induce any significant cell death compared to that in organoids that were not kept in the mini-incubator or imaged (Figure 5—figure supplement 6).

Electroporation of organoids with GFP construct is a powerful method to label migrating neurons with high efficiency and low toxicity (Mellios et al., 2011). We performed a comparative live-cell imaging in GFP-electroporated wild-type and mutant organoids and confirmed that most cells expressing GFP also emitted prominent THG signal (Figure 2—figure supplement 4A-B). Indeed, we found that 23/25 GFP-positive cells in the WT organoids and 27/30 cells in the MT organoids emitted somatic THG signals (Figure 2—figure supplement 5). Cells expressing both GFP and THG signal demonstrated similar migration deficits in their migration speeds (Figure 2—figure supplement 4C-D) as described above (Figures 56).

Discussion

Since the demonstration that neural differentiation of human induced pluripotent stem cells could be induced in 3D and used to recapitulate the early stages of embryonic brain development, cerebral organoids have raised considerable interest as a model for understanding embryonic stages of cortical development and neurodevelopmental disorders (Lancaster et al., 2013; Qian et al., 2016; Paşca et al., 2015; Mariani et al., 2012; Kadoshima et al., 2013; Luo et al., 2016; Camp et al., 2015; Zhang et al., 2019; Bershteyn et al., 2017; Klaus et al., 2019; Garcez et al., 2016; Dang et al., 2016; Mariani et al., 2015; Ip et al., 2018) However, high-resolution 3D imaging of intact and uncleared organoids is challenging due to high light scattering. Thus, the majority of analyses of organoids has been limited to imaging tissue sections or intact organoids, with one-photon microscopy (Lancaster et al., 2013; Paşca et al., 2015; Bershteyn et al., 2017; Klaus et al., 2019; Birey et al., 2017; Birey et al., 2022; Xiang et al., 2017) or imaging superficial parts of intact organoids with two-photon microscopy (Mansour et al., 2018). Especially for live cell imaging, the one photon imaging depth is limited to 20 µm, with improvement to 40 µm with tunable refractive-index matching medium (Boothe et al., 2017). Two photon imaging depth is limited to a couple of hundred micrometers with live cerebral organoids (Mansour et al., 2018). To image intact and uncleared organoids, we thus developed custom three-photon microscopy to perform both fluorescence and label free (THG) three-photon imaging. Here, we demonstrate for the first time that three-photon fluorescence and THG microscopy is capable of high spatial resolution imaging deep within intact human cerebral organoids, enabling 3D imaging of intact cerebral organoids as well as long-term live cell time-lapse imaging of radially migrating neurons in intact cerebral organoids. Since we used moderate repetition rate of the laser (800 kHz), uniquely reduced the pulse width <30 fs in the deepest parts of the organoid for the first time (see Materials and methods), and optimized the intermediate optics in the excitation and emission paths, the average power on the surface of the organoids did not exceed 5 mW, which induced no noticeable tissue damage such as cell death (Figure 5—figure supplement 6). No specific processing of the fixed organoids, such as clearing techniques (Murray et al., 2015; Chung and Deisseroth, 2013) was necessary; however, incubation in index matching solution improved the imaging depth if necessary (Figure 5—figure supplement 7).

Most of the THG signal we observed in RTT-WT and RTT-MT organoids arose from cellular membranes or showed a high-intensity somatic profile observed mostly in the CP region. It was shown previously that the cell nucleus (Lin et al., 2014; Tsai et al., 2011) and several organelles in the cell cytoplasm could generate a THG signal (Yelin and Silberberg, 1999; Barzda et al., 2005). Thus, we propose that the bright somatic signal is the result of the high concentration of organelles in the neuronal soma. THG imaging of intact fixed organoids provided clear demarcation between VZ and CP structures, likely due to differences in cell composition, morphology and density in these regions. This clear label-free demarcation of VZ and CP further allowed us to characterize VZ 3D structures in RTT-WT and RTT-MT organoids. We characterized the extinction lengths of RTT-WT and RTT-MT organoids (Figure 3C–E) and found that average extinction length of WT organoids is significantly higher than that of MT organoids. We suggest that higher surface area with lower radial thickness in MT organoids resulted in irregular shapes of ventricular zones. In addition, the number of ventricular zones in MT organoids is significantly higher than in WT organoids. Therefore, higher number of irregular shaped ventricular zones may induce more scattering in MT organoids which results in their lower extinction lengths. Moreover, a bright THG signal arose from neuronal soma, allowing live cell tracking of migrating neurons in the VZ. To check the source of THG signal in the VZ, we labeled migrating neurons with GFP through electroporation. We showed that more than 85% of the GFP positive cells had also significant THG signal, suggesting that label-free THG imaging of live cerebral organoids can reveal the migration trajectories and deficits at the single-cell level (Figure 2—figure supplements 25). In addition, we also performed DCX staining to the same field of view that we imaged with our system. This staining also shows that THG signal is mostly overlapped with DCX signal which indicates that THG signal is mostly populated in the cortical plate but not in the ventricular zone (Figure 2—figure supplement 1).

Our findings of 3D morphological alterations in RTT-MT organoids including an increase in total ventricular zone volume, a significant increase in ventricular zone surface area, and a significant decrease in mean ventricle thickness, enable noninvasive and comprehensive quantification of organoids consistent with enhanced proliferation of the progenitor pool (Mellios et al., 2011). As VZ expansion is uniquely different in human versus mouse neurogenesis (Florio and Huttner, 2014), it is likely that this phenotype would go unobserved in mouse (or other lissencephalic) models of RTT. Moreover, we and others previously suggested with endpoint experiments that the putative migration of young neurons through the VZ is impaired in mice after in utero electroporation of Mecp2-sh RNA as well as in human RTT cerebral organoids (Mellios et al., 2011; Bedogni et al., 2016). However, the dynamics of the migration deficit were never previously characterized. Thus, here we performed label-free live cell tracking in intact and uncleared human cerebral organoids at different depths and locations to explore the dynamics of neuronal radial migration in RTT-WT and RTT-MT organoids. RTT-MT neurons demonstrated reduced migration speed and less straight trajectories and consequently a lowered displacement than RTT-WT cells in cerebral organoids at 5 weeks. Furthermore, these trends in migration speed were independent of the imaging depth or the imaging location in both MT and WT organoids. Although there was a range of radial thickness in MT organoids, lower migration speed and tortuous migration patterns were common. Altered cytoskeleton dynamics have been previously suggested in rat, mouse and human models of Rett syndrome and could contribute to migration defects (Kaufmann et al., 1995; Delépine et al., 2013; Delépine et al., 2016; Cortelazzo et al., 2014; Bhattacherjee et al., 2017). Non-cell autonomous effects could also contribute to the neuronal radial migration deficit, such as impaired cell-to-cell communication, adhesion and radial glia dysfunction. Further studies are required to study the mechanisms responsible for the migration deficit as well as to rescue the phenotype. Taken together, these results suggest that changes in ventricular wall thickness result not only from the decreased neurogenesis and increased proliferation of the progenitor pool but also from the abnormally slow displacement and tortuous trajectory of the neurons. Importantly, our findings directly demonstrate dynamic deficits in migration at very early stages of prenatal development which provide insight into previously described deficits in human RTT brains, including reduced cortical thickness, cell size and dendritic arborization, and reduced cerebral volume (Armstrong et al., 1995; Bauman et al., 1995; Carter et al., 2008).

Our study demonstrates the promise of three-photon and label-free THG imaging for monitoring spatiotemporal dynamics in intact and uncleared human cerebral organoids. We believe that our optimized three-photon microscopy system is straightforward to use, and can provide a new platform for monitoring structural and functional abnormalities in fixed and live organoids derived from patients with neurogenetic disorders as well as for therapeutic drug screening. In addition, we can study phenotypes in calcium activity in neurons especially VZ calcium dynamics such as waves (Weissman et al., 2004) as well as their migration deficits in cerebral organoids. In addition, we can capture structural phenotypes in myelinated axons and endothelial cells (microvasculature; Yildirim et al., 2020). We believe that imaging myelinated axons as well as endothelial cells can be used for studying brain disorders such as Multiple Sclerosis and Alzheimer’s disease with cerebral organoids since there is a significant loss of myelin as well as reduction of blood flow in these diseases, respectively. Finally, it should be possible to perform structural and functional three-photon and THG imaging for revealing structural and functional phenotypes for cancer applications such as glioblastoma (Bakker et al., 2022).

Materials and methods

Custom microscope for three-photon and THG imaging

Request a detailed protocol

To obtain an excitation wavelength of 1300 nm required for label-free THG imaging deep within intact organoids, ultrashort laser pulses (300 fs, 400 kHz,16W) at 1045 nm from a pump laser (Spirit, Spectra Physics) were sent to a noncollinear optical parametric amplifier (NOPA, Spectra Physics). In order to shorten the pulse width on the sample to maximize the generated THG signal, we built a two-prism based external compressor to prechirp the pulse before sending it to the microscope. The internal compressor in NOPA could compress the Gaussian pulse width to 22 fs (which was the transform limited pulse width for 110 nm spectral full width half maximum bandwidth). Due to the dispersion in the microscope, the Gaussian pulse width on the sample was ~250 fs. After dispersing the pulse with 65 cm long prism-pair, and silicon broadband window (Thorlabs, WG81050) the pulse width on the sample was reduced to 27 fs in the deepest part of the organoids (pulse widths <30 fs obtained, to our knowledge, for the first time). To measure the pulse width inside the organoids, we placed a live organoid (<1 mm thickness) on the external sensor of the autocorrelator (CARPE). Then, we measured the pulse width of the laser after it exits from the thick organoid. In addition, we built a delay line to double the repetition rate thereby increasing the frame rate of THG imaging. Subsequently, the frame rate at 512 × 512 pixel resolution was increased up to 2 Hz. Power control was performed with a combination of half-wave plate (AHWP05M-1600, Thorlabs) and low-GDD ultrafast beamsplitter (UFBS2080, Thorlabs) with 100:1 extinction ratio. The laser beams were scanned by a pair of galvonometric mirrors (6215 H, Cambridge Technologies) to image the laser spot on the back aperture of the objective using a pair of custom-designed scan and tube lenses. A pair of collection lenses for three photon-multiplying tubes (PMT) collected the emitted signal from the cerebral organoids. To maximize the signal detected in PMTs, optics in both excitation and emission paths were optimized considering the objective lens using Zemax software. Fluorescent signals from immunostaining labels (7AAD) were detected using GaAsP photomultiplier tubes (H7422A-40, Hamamatsu, Japan) and a bandpass filter (Semrock, 650/70 nm), and THG signal was detected using bialkali (BA) photomultiplier tube (R7600U-200) and a narrow-bandwidth bandpass filter (430/20 nm). A custom designed objective lens (25 x, 1.05 N.A., Olympus) was used. Automated image acquisition, control of scanners, and sample stage were carried out using ScanImage (Vidrio) (Pologruto et al., 2003). Imaged cells were located at a depth of 0–2000 μm below the surface of cerebral organoids. The images were taken with 2 µm and 5 µm axial increments in fixed and live organoids, respectively. In live cell imaging, we acquired 50–100 z-stack images to make a volumetric acquisition. This volumetric imaging took approximately 3–6 min and we repeated it every 20 min. Laser power ranged from 0.5 to 5 mW at the sample depending on depth and fluorescence expression levels. Both fixed and live organoids were placed on a two-axis motorized stage (MMBP, Scientifica) and the objective lens was placed on a single-axis motorized stage (MMBP, Scientifica) to move it in the axial direction.

Methods for fixed and live organoid imaging

Request a detailed protocol

For fixed organoid imaging, five-week-old cerebral organoids were placed into an imaging chamber gasket (C18161, ThermoFisher) filled with index-matching medium Hprotos (Murray et al., 2015). Top and bottom parts of the imaging chamber gasket were sandwiched between a glass slide (48311–703, VWR) and a cover glass (48366–227, VWR) (Figure 1B). To perform whole fixed organoid imaging, automatic serial imaging of multiple sites was performed via determining the coordinates of four corners of the entire field of view. A Matlab-based algorithm moved the stage in the required x and y axis locations to image the full field of view and then moved the focal plane in the z direction.

For live organoid imaging, we increased the imaging field of view from 250 µm to 320 µm. The 5-week-old live organoids were placed in a microincubator (RC-30, Warner Instruments). The microincubator temperature was controlled and maintained at 37 °C with a pair of resistive heater elements. A peristaltic pump was connected to the microincubator to provide continuous flow of cell medium supplemented with 5% CO2, 5%O2, and balanced N2 (Airgas) (Figure 1C). We performed Caspase3 immunostaining to check whether our imaging system induced any significant cell death. Compared with healthy and not-imaged organoids, the organoids which were imaged up to 18 hr in our imaging system did not show any increased cell death (Figure 5—figure supplement 6).

iPSC generation and colony maintenance

Request a detailed protocol

Induced pluripotent stem cells (iPSC) were reprogrammed from Rett Syndrome patient samples obtained from the Coriell cell line repository (Mellios et al., 2011; Nott et al., 2016). These lines were derived from: (1) a patient carrying a MECP2 V247X mutation (Coriell Institute biobank GM07982); (2) a patient with MECP2 R306C mutation (Coriell Institute biobank GM23298). All iPSC lines were primarily and routinely quality-controlled by karyotyping (G-band), mycoplasma testing and pluripotency testing (initially by teratoma test then routinely by immunostaining for pluripotency markers such as Sox2 and Oct4). We examined allele-specific transcription of the MECP2 gene in all the iPSC lines by performing RNA extraction, reverse transcription, PCR and Sanger sequencing of the amplified fragments. In all lines we were able to detect transcription of either the wildtype or the mutant alleles of the MECP2 gene, but not both, confirming the clonality and X inactivation status. We confirmed that the X inactivation profile is maintained at all stages of the organoid differentiation. Cell culture quality check and authentication were performed every ten iPSC passages. iPSC colonies were grown in iPSC media, consisting of 20% Knock-out Serum Replacement (KOSR) (Invitrogen), 1% penicillin/streptomycin (Invitrogen), 1% non-essential amino acids (Invitrogen), 0.5% L-glutamine (Invitrogen), 100 µM 2-mercaptoethanol (Bio-Rad), DMEM/F-12 (Invitrogen), supplemented with 10 ng/mL bFGF (Stemgent). Culture media was changed daily. iPSC colonies were passaged weekly onto 6-well plates coated with 0.1% gelatin (EMD Millipore) and pre-seeded with a feeder layer of irradiated mouse embryonic fibroblasts (MEFs) (GlobalStem), which were plated at a density of 200,000 cells/well. Passaging entailed lifting colonies with 2.5 mg/ml Collagenase, Type IV (ThermoFisher) for 1 hr and dissociation into smaller pieces through manual tritulation before seeding onto feeder layer of MEFs.

Cerebral organoid generation

Request a detailed protocol

iPSCs were detached from irradiated MEFs and plated at 9 × 104 cells per well of an ultralow attachment 96-well plate (Corning) in iPSC media supplemented with bFGF (10 ng/mL) and ROCK inhibitor (50 µM; Y-27632, Tocris) (Day 0). Embryoid bodies (EBs) were subsequently transferred (Day 6) in an ultra-low attachment 24-well plate (Corning) to neural induction media: 1% N2 supplement (Invitrogen), 1% Glutamax (Invitrogen), 1% non-essential amino acids (Invitrogen), 5 µg/mL heparin (Sigma), DMEM/F12 (Invitrogen), supplemented with 10 µM SB431542 (Tocris Bioscience) and 1 µM dorsomorphin (Stemgent). EBs were embedded in Matrigel (Corning) droplets on Day 11 and transferred to neural differentiation media: DMEM/F12: Neurobasal (Invitrogen), 0.5% N2 supplement, 1% Glutamax, 0.5% non-essential amino acids, 100 µM 2-mercaptoethanol, insulin, 1% Pen/Strep (Invitrogen) supplemented with 1% B27 without vitamin A (Gibco, Life Technologies). On Day 15, embedded EBs were transferred to a shaker and grown in neural differentiation media supplemented with B27 with vitamin A (Gibco, Life Technologies). Organoids were either fixed on Day 35 by a 3-hr incubation in 4% paraformaldehyde solution or imaged under the microscope for live cell characterization.

Cryosectioning and immunostaining

Request a detailed protocol

For cryosectioning, fixed organoids were incubated in sucrose solution overnight at 4 °C before embedding and freezing in O.C.T medium. Frozen organoid tissue was sliced into 20 μm sections using a cryostat. Permeabilization/blocking was performed using 3% BSA/0.1% TX100 in TBS. Incubation of sections from cerebral organoids in primary antibodies solution was performed overnight at 4 °C and in secondary antibodies solution at room temperature for 1 hr (Alexa Fluor, Molecular Probes). The following primary antibodies were used: DCX (Aves Labs, 1:200); PAX6 (Millipore; 1:200); TBR1 (Abcam 1:400); Cleaved capsase-3 (Cell Signaling, 1:500). Coverslips were affixed with ProLong Gold antifade reagent with DAPI (Life Technologies) and z-stacks were acquired using either a Leica TCS SP8 confocal microscope.

Intact organoid staining

Request a detailed protocol

For 3D imaging of whole organoids, intact fixed organoids were incubated with a 1:50 solution of the nuclear dye 7-Aminoactinomycin D (7AAD, MolecularProbes) in PBS for 1 hr at room temperature. Before imaging, organoids were transferred to a shaker and immersed twice for 3 hr in index matching solution (Murray et al., 2015). Z-stacks were acquired with the three-photon microscope described above.

Electroporation

Request a detailed protocol

Organoids (specifically, the ‘ventricles’ or progenitor zones) were electroporated with a GFP construct (pCAG-GFP) at 7 weeks post-EB formation. Plasmids were maxi-prepared using Qiagen Maxi kit. DNA solution (0.5 μg/μl) was mixed with Fast Green dye (0.1%) and injected into the organoids using a pulled-glass capillary microelectrode. Successful injection was confirmed by the visualization of Fast Green dye inside the injected organoid. Immediately after DNA injection, four 50 ms electrical pulses (40 V) were applied at 1 s intervals using a 5 mm electrode and an electroporator (EM830, BTX). The organoids were returned to the incubator after electroporation. After 3–4 days, the electroporated organoids were subjected to the three-photon microscope for imaging.

Statistical analysis

Request a detailed protocol

All the statistical comparisons presented in the figures and figure supplements (Figure 3E, Figure 4C, Figure 5C, Figure 6F, Figure 4—figure supplements 25, and Figure 5—figure supplement 2; Figure 5—figure supplement 4; Figure 5—figure supplement 5) are based on t-test method where the error bars are 90% of the confidence interval (CI). Points below and above this confidence interval are drawn as individual dots.

Area segmentation and rendering

Request a detailed protocol

The area segmentation was performed manually for individual 2-D THG images and confirmed with 7AAD images which are taken as ground truth (Figure 4A–B). Then, Surface function of Imaris software was used to render these 2-D segmented areas into a 3-D volume. We used both opaque (Figure 4A–B) and transparent modes (Figure 4—figure supplement 1C-D) of this function in order to represent 3-D structure of ventricular zones.

Live cell tracking

Request a detailed protocol

The three-dimensional THG images were used to characterize the speed, straightness, and displacement of cells in control and mutant organoids. To quantify these features of control and mutant cells, Spots function of Imaris software was used. This function first identifies the location of the individual cell, and tracks them in time-lapse images.

Colocalization of GFP, THG, and 7AAD signals

Request a detailed protocol

The colocalization of GFP, THG, and 7AAD signals were performed through Coloc module of Imaris software. We used our z-stack 2-D images (~1,000 images) to form three-channel three-dimensional data sets (Figure 2—figure supplement 2). Then, Coloc module quantified the percentage of overlapping between each channel.

Incubation of fixed organoids in index-matching solution

Request a detailed protocol

For imaging some of the fixed organoids, we incubated organoids in the H-protos index- matching solution for 3 hr. With this incubation, we observed improvement in maximum imaging depth up to 50% compared to the condition without the incubation (Figure 5—figure supplement 7).

Data availability

All raw data has been deposited to Dryad (https://dx.doi.org/10.5061/dryad.hqbzkh1jz) and is freely available.

The following data sets were generated
    1. Yildirim M
    2. Delepine C
    3. Feldman D
    4. Pham V
    5. Chou S
    6. Ip J
    7. Nott A
    8. Tsai L
    9. Ming G
    10. So P
    11. Sur M
    (2022) Dryad Digital Repository
    Label-free three-photon imaging of intact human cerebral organoids: Tracking early events in brain development and deficits in Rett Syndrome.
    https://doi.org/10.5061/dryad.hqbzkh1jz

References

Decision letter

  1. Jeannie Chin
    Senior and Reviewing Editor; Baylor College of Medicine, United States
  2. Anita Bhattacharyya
    Reviewer; University of Wisconsin, Madison, United States
  3. Simon R Schultz
    Reviewer; Imperial College London, United Kingdom

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Label-free three-photon imaging of intact human cerebral organoids: tracking early events in brain development and deficits in Rett Syndrome" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and Jeannie Chin as the Senior Editor. The following individuals involved in the review of your submission have agreed to reveal their identity: Anita Bhattacharyya (Reviewer #1); Simon R Schultz (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1) The prior literature in this area of research is not put in perspective well to reflect the current approaches in organoid imaging. There is a significant body of literature describing migration imaging using brain organoid models, including ones that use intact (not sliced) assembloid preparations not cited in the paper, as opposed to what's been suggested by the authors (Lines 60 and 61; Lines 258 and 259). These studies use viral labeling to visualize the cells of interest during long-term (days-long) migration imaging without seemingly affecting organoid health, again as opposed to what's been suggested (Line 66). Please revise to reflect more accurately what's been achievable by the current state of the art. This surely does not take away from the novelty of the 3P/THG method but better clarifies the limitations being addressed. Please see references in Reviewer 3's comments below.

2) A major issue in the field of stem cell modeling of disease is how well the model recapitulates the disease. Please comment on how the phenotypes (less migration) in organoids inform our understanding of RTT. How does phenotype relate to patient neuropathology?

3) Is the RTT migration phenotype caused solely by the more tortuous VZ region they need to navigate? Once they reach THG-high regions (cortical plate), does the phenotype recover? If not, does this indicate additional cell-intrinsic deficits?

4) Please tone down the narrative in the introduction in which the dichotomy between intact 3D and sliced 2D structures is highlighted. While these are certainly different, both are still 3D and will benefit from deep tissue imaging techniques.

5) For clarity of the manuscript to a wide audience, please include some discussion contrasting THG with standard three-photon excitation.

6) An expanded discussion could serve readers well to envision the true potential of the 3P/THG system. Do the authors envision additional features can be extracted in older (e.g. myelinated axons) or novel (e.g. organoids with endothelia) organoids using their models? Is THG-coupled calcium imaging possible, for example where one can couple VZ calcium dynamics, such as waves (Weissman et al., 2004), to neuronal dynamics? Can other disorders that cause structural abnormalities like cancers, tuberous sclerosis, or heterotopias be modeled using organoids and differentiated from one another using the 3P/THG signal?

7) On L343 "incubation in index matching solution improved the imaging depth" does not seem to be described in the Methods, please describe. Please also clarify whether this was quantified through attenuation length, maximum depth at which cells could be observed, or another measure.

8) Some explanations are proposed for the result that WT and MT organoids show a different attenuation length. Were any measurements made to validate these explanations? The stated structural differences would appear to be measureable from 3D organoid stacks, and thus may provide support to the conjecture.

9) To better understand how the two different RTT iPSC isogenic pairs are measured and presented throughout the results, please use different colors or shapes to indicate data from the two lines so that it is clear that two lines were assessed and the data combined. Please also include more information in the results or legends for clarification of the data being presented (different lines, differences between lines, number of experimental batches, etc).

10) Please report actual p values, degrees of freedom (when relevant) and n for each t-test, and define the error bars (SEM, 90% CI, etc).

11) In Figure 6, it looks like the data is pooled from organoids that were imaged for different periods of time. If so, it is misleading to report statistical differences between WT and MT groups. Please clarify.

Note: Several validation studies were suggested in the reviews. The translation of THG signal to its cellular correlates is important and likely to present ongoing issues if this becomes a more commonplace imaging practice. However, given the strength of the technical advance and that this is the first proof-of-principle demonstration, we are not requiring further validation experiments at this time. If you already have data to address the issues, we highly encourage you to incorporate them into your revised manuscript.

Reviewer #1 (Recommendations for the authors):

Two major issues remain:

Throughout the results, it is not clear how the two different RTT iPSC isogenic pairs are measured and presented. Although some information is in Supplementary figures, the data would be better presented with the two lines separated by color/shape to make it clear that two lines were assessed and the data combined. Information in the figure legend or results should provide additional clarification. Were all data from different lines combined? Are there differences between lines? How many experimental batches were done?

A major issue in the field of stem cell modeling of disease is how well the model recapitulates the disease. The authors should comment on how the phenotypes (less migration) in organoids inform our understanding of RTT. How does phenotype relate to patient neuropathology?

Reviewer #2 (Recommendations for the authors):

I think the Intro slightly over-eggs some things in making the justification for the work. Firstly it is noted that sectioning induces unwanted tissue damage and cell loss. Yes, but this neglects that sectioning is intrinsically part of some cerebral organoid protocols, in order to mature them: as increasing numbers of neurons are produced, they are progressively pushed inside and undergo necrosis. The way of getting around this (and maturing 3D organoid structures over much longer timeframes) has been to slice the organoids and then maintain organotypic organoid slice cultures over long time periods. Thus the dichotomy made between intact 3D structures and sliced 2D structures in prior work is not really valid – there is much ongoing work on engineered cerebral organoids that occupy a middle ground.

For clarity of the manuscript to a wide audience, I think it would help a lot to contrast THG with standard three-photon excitation. Some readers may be confused (and not all 3P is THG). Several sentences around lines 68-69 might suffice. Citations [41-43] are made about THG in the brain, but really only [42] is about THG, the others relate to 3PEF.

Figure 2 – scale bars are missing lengths.

Figure 3 – You use both attenuation length and extinction length – are you referring to the same thing? Perhaps keep terminology consistent.

Figure 3 – please report the p-value and n for the t-test, it is not acceptable in this day and age to just say p<0.05. (Actually, I think it would be more useful to report effect size.) In the main text also: L219 is that an S.E.M. or 90% confidence interval? L220 state the test, n, df (if relevant), and actually produced a p-value. Given that from the figure, n=4, I'm not convinced that this is sufficiently powered, it seems to be in the danger zone, so should be reported and described carefully. Statistical reporting from L229 seems much better.

L260 what is DIV? Also, do you mean 1 frame being captured every 20 minutes or a sequence of frames?

Figure 6 – I am slightly puzzled by the displacement statistics shown here. It looks like the data pooled is from organoids that are imaged for different periods of time, in which case it is misleading to report statistical differences between WT and MT groups. I imagine the "straightness" measure would also be biased by the inclusion of samples recorded from different imaging durations.

L343 "incubation in index matching solution improved the imaging depth". I could not find this in the Methods, can you please detail? Also, did you quantify this through attenuation length, maximum depth at which cells could be observed, or another measure?

L355 you propose some explanations for the result that WT and MT organoids show a different attenuation length. Did you make any measurements to validate these explanations? The stated structural differences would appear to be measurable from 3D organoid stacks, and thus may provide support to the conjecture.

L413 exits rather than exists.

Reviewer #3 (Recommendations for the authors):

I have three broad critiques in regard to the cited literature, validation experiments, and additional applications:

1) Prior literature: Authors should revise the text to better reflect the current approaches in organoid imaging. There is a significant body of literature describing migration imaging using brain organoid models (e.g. Birey et al., 2017; Bagley et al., 2017; Xiang et al., 2017; Klaus et al., 2019; Bajaj et al., 2021; Birey et al., 2022), including ones that use intact (not sliced) assembloid preparations not cited in the paper (Xiang et al., 2017; Birey et al., 2017, 2022), as opposed to what's been suggested by the authors (Lines 60 and 61; Lines 258 and 259). These studies use viral labeling to visualize the cells of interest during long-term (days-long) migration imaging without seemingly affecting organoid health, again as opposed to what's been suggested (Line 66). Authors should revise to reflect more accurately what's been achievable by the current state of the art. This surely does not take away from the novelty of the 3P/THG method but better clarifies the limitations being addressed.

2) Validation experiments: In early characterization experiments, authors convincingly show that THG signal is higher in regions populated with neurons compared to denser VZ-like regions populated with progenitors. As it is important to translate the THG single as accurately as possible, similar validation experiments are needed to better link THG dynamics to specific cellular events in the following phenotyping experiments. For example, the authors show a higher surface area and about a 2-fold increase in the number of ventricles in RTT organoids. An optical clearing approach for the SA quantification and/or a simple cryosection quantification for the number of VZ-like zones should be able to reveal a change of this magnitude. Authors electroporate a CAG-driven plasmid to visualize cells that they have previously reported, albeit in three samples, to almost exclusively (~95%) capture neurons (Delepine et al., 2021). As the rationale as to why a generic promoter shows lineage selectivity is unclear and as the imaging field of interest also includes other cell types that migrate in a similar fashion (i.e. radial glia that go inter-kinetic nuclear migration), a parallel labeling strategy, e.g. lineage-specific viral vectors or expression vectors with different promoters, to further delineate the specific lineage captured by the THG signal would be informative. For example, Figure 3A and B show VZ-like regions with high TGH signal within their cores, which are unlikely to be neurons. This could resolve questions of this nature.

3) Is the RTT migration phenotype caused solely by the more tortuous VZ region they need to navigate? Once they reach THG-high regions (cortical plate), does the phenotype recover? If not, does this indicate additional cell-intrinsic deficits?

4) An expanded discussion could serve the reader well to envision the true potential of the 3P/THG system. Do the authors envision additional features can be extracted in older (e.g. myelinated axons) or novel (e.g. organoids with endothelia) organoids using their models? Is THG-coupled calcium imaging possible, for example where one can couple VZ calcium dynamics, such as waves (Weissman et al., 2004), to neuronal dynamics? Can other disorders that cause structural abnormalities like cancers, tuberous sclerosis, or heterotopias be modeled using organoids and differentiated from one another using the 3P/THG signal?

References:

Bagley, J.A., Reumann, D., Bian, S., Lévi-Strauss, J., and Knoblich, J.A. (2017). Fused cerebral organoids model interactions between brain regions. Nat Methods 14, 743-751. https://doi.org/10.1038/nmeth.4304.

Bajaj, S., Bagley, J.A., Sommer, C., Vertesy, A., Nagumo Wong, S., Krenn, V., Lévi-Strauss, J., and Knoblich, J.A. (2021). Neurotransmitter signaling regulates distinct phases of multimodal human interneuron migration. EMBO J 40, e108714. https://doi.org/10.15252/embj.2021108714.

Birey, F., Andersen, J., Makinson, C.D., Islam, S., Wei, W., Huber, N., Fan, H.C., Metzler, K.R.C., Panagiotakos, G., Thom, N., et al. (2017). Assembly of functionally integrated human forebrain spheroids. Nature 545, 54-59. https://doi.org/10.1038/nature22330.

Birey, F., Li, M.-Y., Gordon, A., Thete, M.V., Valencia, A.M., Revah, O., Paşca, A.M., Geschwind, D.H., and Paşca, S.P. (2022). Dissecting the molecular basis of human interneuron migration in forebrain assembloids from Timothy syndrome. Cell Stem Cell https://doi.org/10.1016/j.stem.2021.11.011.

Delepine, C., Pham, V.A., Tsang, H.W.S., and Sur, M. (2021). GSK3ß inhibitor CHIR 99021 modulates cerebral organoid development through dose-dependent regulation of apoptosis, proliferation, differentiation and migration. PLOS ONE 16, e0251173. https://doi.org/10.1371/journal.pone.0251173.

Klaus, J., Kanton, S., Kyrousi, C., Ayo-Martin, A.C., Di Giaimo, R., Riesenberg, S., O'Neill, A.C., Camp, J.G., Tocco, C., Santel, M., et al. (2019). Altered neuronal migratory trajectories in human cerebral organoids derived from individuals with neuronal heterotopia. Nat Med 25, 561-568. https://doi.org/10.1038/s41591-019-0371-0.

Weissman, T.A., Riquelme, P.A., Ivic, L., Flint, A.C., and Kriegstein, A.R. (2004). Calcium Waves Propagate through Radial Glial Cells and Modulate Proliferation in the Developing Neocortex. Neuron 43, 647-661. https://doi.org/10.1016/j.neuron.2004.08.015.

Xiang, Y., Tanaka, Y., Patterson, B., Kang, Y.-J., Govindaiah, G., Roselaar, N., Cakir, B., Kim, K.-Y., Lombroso, A.P., Hwang, S.-M., et al. (2017). Fusion of Regionally Specified hPSC-Derived Organoids Models Human Brain Development and Interneuron Migration. Cell Stem Cell 21, 383-398.e7. https://doi.org/10.1016/j.stem.2017.07.007.

https://doi.org/10.7554/eLife.78079.sa1

Author response

Essential revisions:

1) The prior literature in this area of research is not put in perspective well to reflect the current approaches in organoid imaging. There is a significant body of literature describing migration imaging using brain organoid models, including ones that use intact (not sliced) assembloid preparations not cited in the paper, as opposed to what's been suggested by the authors (Lines 60 and 61; Lines 258 and 259). These studies use viral labeling to visualize the cells of interest during long-term (days-long) migration imaging without seemingly affecting organoid health, again as opposed to what's been suggested (Line 66). Please revise to reflect more accurately what's been achievable by the current state of the art. This surely does not take away from the novelty of the 3P/THG method but better clarifies the limitations being addressed. Please see references in Reviewer 3's comments below.

We thank the reviewer for this comment, and have now included the suggested references in the main text of the revised manuscript (Lines 64-66). As a small point, while live cell imaging has been done in intact organoids with viral labeling, it is not clear that a viral strategy is the best for long-term imaging and preserving organoid health. In our recent paper performed in awake mouse brain (Yildirim et al., 2019), we found laser power regimes which have adverse (but not visible) effects on the physiological responses of virally labeled neurons, in addition to the higher laser power regime which can induce visible damage and optical breakdown. We have explained this briefly (Lines 69-70) as a further justification for using label-free approaches, importantly in conjunction with high-resolution deep-tissue imaging.

2) A major issue in the field of stem cell modeling of disease is how well the model recapitulates the disease. Please comment on how the phenotypes (less migration) in organoids inform our understanding of RTT. How does phenotype relate to patient neuropathology?

The vast majority of the literature has focused on the role of MeCP2 in postnatal brain maturation and function, mainly in mouse models, and relatively little is known about its effects in prenatal brain development. Structural deficits described in postmortem RTT human brains include reduced cortical thickness, cell size and dendritic arborization (Bauman et al., 1995; Armstrong et al., 1995), and reduced cerebral volume in MR imaging of RTT patients (Carter et al., 2008). These deficits are paralleled by reductions in dendritic arborization, soma size and spine density described in RTT mouse models (Smrt et al., 2007; Fukuda et al., 2005; Kishi and Macklis, 2004; Shahbazian et al., 2002). This is now mentioned in the Introduction (Lines 48-52).

Our previous work demonstrated impaired proliferation of the progenitor pool and delayed maturation and presumed migration of neurons (Mellios et al., 2018), consistent with the human postmortem deficits. However, as we point out, the use of fixed tissues did not allow cell tracking and analysis of parameters such as speed and trajectory of neuronal displacement; thus the dynamics and mechanisms of the presumptive migration deficit remain to be characterized. Our present findings directly demonstrate deficits in migration at very early stages of prenatal development which, together with deficits mentioned above and described as ‘curtailment of development’ (Bauman et al., 1995), provide insight into early events which underlie the structural phenotypes of cortical development in RTT. This is now noted in the Discussion (Lines 427-430).

3) Is the RTT migration phenotype caused solely by the more tortuous VZ region they need to navigate? Once they reach THG-high regions (cortical plate), does the phenotype recover? If not, does this indicate additional cell-intrinsic deficits?

We are not certain whether the RTT migration phenotype caused solely by more tortuous VZ region since we do not have long-term (potentially spanning several months) imaging data to reveal how cells migrate in the cortical plate. However, we have two working hypothesis that we would like to test. In the first hypothesis, the migration phenotype recovers when cells start to form the cortical plate since their differentiation into different neuronal cell types is fairly complete. We can address this by labeling neurons in the cortical plate with different cell-specific markers in addition to analyzing their migration patterns. In the second hypothesis, migration phenotypes especially for inhibitory neurons may continue while cortical layers are being formed. In this case, delayed neurodevelopment may cause disproportionately small and layered brain regions as well as disruption of excitation and inhibition (EI) balance. Although rare in the clinic, it is reported that some male babies born with a mutant MECP2 allele display disproportionately small frontal and temporal lobes characteristic of a prenatal pathogenesis (Schule et al., 2008). Recent studies have also demonstrated cell-autonomous deficits in GABA signaling, including that GABA is less hyperpolarizing in MeCP2-deficient mice (Banerjee et al., 2016) and in iPSC derived neurons generated from RTT patients (Tang et al., 2016).

4) Please tone down the narrative in the introduction in which the dichotomy between intact 3D and sliced 2D structures is highlighted. While these are certainly different, both are still 3D and will benefit from deep tissue imaging techniques.

We have removed the narrative in the introduction about dichotomy between intact 3D and sliced 2D structures. We now focus on challenges in 3D imaging of intact organoids and how to deal with them with label-free 3-photon imaging (Lines 64-73).

5) For clarity of the manuscript to a wide audience, please include some discussion contrasting THG with standard three-photon excitation.

We have expanded on our earlier description of three-photon fluorescence microscopy and THG microscopy in the Introduction. As reviewers suggested, we have now added more explanation about these microscopy techniques (Lines 79-96).

Both standard three-photon fluorescence microscopy and THG microscopy occur based on a three-photon interaction between ultrashort pulses and tissues. Three-photon fluorescence microscopy has been recently used to perform functional and structural brain imaging in anesthetized and awake mice (Ouzounov et al., 2017, Yildirim et al., 2019). These studies are based on utilizing a green (GCaMP) genetically engineered calcium indicator (exogenous fluorophores) with their corresponding excitation wavelengths (1300 nm) which provide peak absorption cross-sections for this indicator. In these three-photon fluorescence microscopy studies, three photons with enough peak power at these excitation wavelengths excite electrons from ground state to excitation states of this indicator. Then, these electrons release their energy while they return to their ground states. During this relaxation, they release emitted photons which have a large 1/e2 bandwidth of fluorescence emission (70 nm for GcaMP6s/f). Therefore, they require an exogenous label with large bandwidth of emission spectrum to perform three-photon fluorescence structural or functional brain imaging. On the other hand, THG microscopy does not need to have any label to perform structural three-photon brain imaging. In addition, three photons with enough peak power at the excitation wavelengths excite electrons from ground state to virtual state so that these electrons do not lose any energy when they come back to the ground state. Therefore, they release emitted photons which have a small 1/e2 bandwidth of emission (~20 nm), and the emission spectrum of THG imaging happens at exactly 1/3 of the excitation wavelength.

Overall, three-photon fluorescence microscopy is valuable for performing structural and functional brain imaging with fluorescent dyes for which the excitation wavelengths are limited by the peak absorption cross-sections of these fluorophores. However, three-photon fluorescence imaging is prone to phototoxicity and photobleaching particularly with high peak intensity pulses (Yildirim et al., 2019). In contrast, THG microscopy provides label-free structural brain imaging (Yildirim et al., 2020) without problems of phototoxicity and photobleaching – which are its strengths for long-term live cell imaging of cerebral organoids.

6) An expanded discussion could serve readers well to envision the true potential of the 3P/THG system. Do the authors envision additional features can be extracted in older (e.g. myelinated axons) or novel (e.g. organoids with endothelia) organoids using their models? Is THG-coupled calcium imaging possible, for example where one can couple VZ calcium dynamics, such as waves (Weissman et al., 2004), to neuronal dynamics? Can other disorders that cause structural abnormalities like cancers, tuberous sclerosis, or heterotopias be odelled using organoids and differentiated from one another using the 3P/THG signal?

At the reviewers’ suggestion, we have added a paragraph (Lines 436-444) to discuss possible applications of our label-free imaging system coupled with three-photon fluorescence imaging for different disorders. First, it is feasible to combine label free THG imaging and three-photon calcium imaging as shown in our recent paper (Yildirim et al., 2019). Therefore, we can also capture phenotypes in calcium activity in neurons in addition to their migration deficits in cerebral organoids. In addition, we can also capture structural phenotypes in myelinated axons and endothelial cells (microvasculature). We showed recently (Yildirim et al., 2020) that one can image calcium dynamics, myelinated axons, as well as microvasculature (three-color imaging) in the cortex and in the white matter of primary and higher visual areas of awake mice. We believe that imaging myelinated axons as well as endothelial cells can be used for studying brain disorders such as Multiple Sclerosis and Alzheimer’s Disease since there is a significant loss of myelin as well as reduction of blood flow in these diseases, respectively. Finally, it is possible to perform structural, and functional three-photon and THG imaging for revealing structural and functional phenotypes for cancer applications such as glioblastoma (Bakker et al., 2022).

7) On L343 "incubation in index matching solution improved the imaging depth" does not seem to be described in the Methods, please describe. Please also clarify whether this was quantified through attenuation length, maximum depth at which cells could be observed, or another measure.

We thank the reviewers for highlighting the effect of index matching solution on the imaging depth. We did not need to apply any index matching solution for 60% of the WT and MT organoids that we used in this study. For the remaining organoids, index matching gel improved the imaging depth up to about 50% as we quantified Supp. Figure 19 in the revised manuscript. We described the incubation of cerebral organoids in index-matching solution in the Methods (Lines 597-600).

8) Some explanations are proposed for the result that WT and MT organoids show a different attenuation length. Were any measurements made to validate these explanations? The stated structural differences would appear to be measureable from 3D organoid stacks, and thus may provide support to the conjecture.

We explained the differences in extinction lengths of WT and MT organoids based on the results that we presented in Figure 3C. We suggest that higher surface area with lower radial thickness in MT organoids resulted in irregular shapes of ventricular zones. In addition, number of ventricular zones in MT organoids is significantly higher than that of WT organoids. Therefore, higher number of irregular shaped ventricular zones may induce more scattering in MT organoids which results in their lower extinction lengths. We have added this additional explanation in the Discussion (lines 390-394).

9) To better understand how the two different RTT iPSC isogenic pairs are measured and presented throughout the results, please use different colors or shapes to indicate data from the two lines so that it is clear that two lines were assessed and the data combined. Please also include more information in the results or legends for clarification of the data being presented (different lines, differences between lines, number of experimental batches, etc).

We have used different colors (blue for Line 1 and green for Line 2) for each line in the Supplementary Figures 9-12, 14, 16-17. Also, we described each line that we used in the figure captions (Figures2-6, Supp. Figures 6-8,13, 15, 18-21).

10) Please report actual p values, degrees of freedom (when relevant) and n for each t-test, and define the error bars (SEM, 90% CI, etc).

We have reported actual p values, and n for each t-test and defined the error bars (SEM, or 90% CI) in each figure caption and in the main text (lines 250, 260, 262, 266, 268).

11) In Figure 6, it looks like the data is pooled from organoids that were imaged for different periods of time. If so, it is misleading to report statistical differences between WT and MT groups. Please clarify.

As reviewers pointed out, we agree that pooling the data from the cells that are not imaged for 12 hours is not reasonable for reporting statistical differences especially in displacement values of cells. These cells generally disappeared from the field of view before finishing 12 hours of imaging, therefore their values are limited to shorter time durations. We removed the data from these cells and updated figures 5 and 6 and supplementary figures 14,16-17 in the revised manuscript. Also, we updated the corresponding figure captions and reported values in the manuscript (lines 300-301, 304, 306, 314-315, 318-319).

Note: Several validation studies were suggested in the reviews. The translation of THG signal to its cellular correlates is important and likely to present ongoing issues if this becomes a more commonplace imaging practice. However, given the strength of the technical advance and that this is the first proof-of-principle demonstration, we are not requiring further validation experiments at this time. If you already have data to address the issues, we highly encourage you to incorporate them into your revised manuscript.

We thank the reviewers again for appreciating the technical advance of our paper, and we agree that it is the first proof-of-principle of THG imaging for intact cerebral organoids with application to Rett syndrome.

Reviewer #1 (Recommendations for the authors):

Two major issues remain:

Throughout the results, it is not clear how the two different RTT iPSC isogenic pairs are measured and presented. Although some information is in Supplementary figures, the data would be better presented with the two lines separated by color/shape to make it clear that two lines were assessed and the data combined. Information in the figure legend or results should provide additional clarification. Were all data from different lines combined? Are there differences between lines? How many experimental batches were done?

We explained whether we combined the data from both lines or not in each figure in the corresponding figure captions (Figures 2- 6 and Supp. Figures 5-8, 13, 15, 18, 20).

In terms of structural parameters (volume, surface area, volume per area and number of cerebral organoids), we observed no differences between WT and MT organoids of both lines. For example, MT organoids have larger surface area, lower volume/area thickness, and higher number of ventricular regions than those in WT organoids independent of the lines. When we focus on individual parameters, we can see some slight differences in absolute values of these parameters in MT and WT organoids in each line. For example, MT and WT organoids in Line 2 have higher number of ventricular regions as well as higher surface area compared to those in Line 1. Whereas, MT and WT organoids in Line 1 have higher volume/area thickness compared that of Line 2. Finally, WT organoids in Line 2 have higher ventricular volume compared to that in Line 1 whereas MT organoids in Line 1 have higher ventricular volume than that of Line 2.

In terms of migration parameters (overall displacement, average speed, and straightness), we found no differences between WT and MT organoids of both lines. When we focus on individual parameters, we found that both WT and MT organoids in Line 2 have higher average speed, straightness and longer displacement than those in Line 1.

A major issue in the field of stem cell modeling of disease is how well the model recapitulates the disease. The authors should comment on how the phenotypes (less migration) in organoids inform our understanding of RTT. How does phenotype relate to patient neuropathology?

We have added information about the relevance of migration deficits to neuropathology of RTT in the Introduction and Discussion, as described in item 2 of Essential Revisions.

Reviewer #2 (Recommendations for the authors):

I think the Intro slightly over-eggs some things in making the justification for the work. Firstly it is noted that sectioning induces unwanted tissue damage and cell loss. Yes, but this neglects that sectioning is intrinsically part of some cerebral organoid protocols, in order to mature them: as increasing numbers of neurons are produced, they are progressively pushed inside and undergo necrosis. The way of getting around this (and maturing 3D organoid structures over much longer timeframes) has been to slice the organoids and then maintain organotypic organoid slice cultures over long time periods. Thus the dichotomy made between intact 3D structures and sliced 2D structures in prior work is not really valid – there is much ongoing work on engineered cerebral organoids that occupy a middle ground.

We changed the wording about dichotomy between 2d and 3d preparations of cerebral organoids in the Introduction and Discussion. Please see item 4 in Essential Revisions.

For clarity of the manuscript to a wide audience, I think it would help a lot to contrast THG with standard three-photon excitation. Some readers may be confused (and not all 3P is THG). Several sentences around lines 68-69 might suffice. Citations [41-43] are made about THG in the brain, but really only [42] is about THG, the others relate to 3PEF.

We have added a section to contrast THG and standard three-photon fluorescence imaging in the Introduction. Please refer to the item 5 in the Essential Revisions.

Figure 2 – scale bars are missing lengths.

We provided the lengths of all scale bars in Figure 2.

Figure 3 – You use both attenuation length and extinction length – are you referring to the same thing? Perhaps keep terminology consistent.

We termed all the lengths (combined scattering and absorption) as extinction lengths in the main text.

Figure 3 – please report the p-value and n for the t-test, it is not acceptable in this day and age to just say p<0.05. (Actually, I think it would be more useful to report effect size.) In the main text also: L219 is that an S.E.M. or 90% confidence interval? L220 state the test, n, df (if relevant), and actually produced a p-value. Given that from the figure, n=4, I'm not convinced that this is sufficiently powered, it seems to be in the danger zone, so should be reported and described carefully. Statistical reporting from L229 seems much better.

For the figure 3, we added two more samples for both WT and MT to increase the statistical power and updated Figure 3E.

L260 what is DIV? Also, do you mean 1 frame being captured every 20 minutes or a sequence of frames?

DIV means days in vitro. We captured 5 frames on the same plane and moved to another z plane which is 5 µm away from the previous plane. We continued getting these z-stack images of 50-100 different planes to make a volume. This volumetric imaging took approximately 3-6 minutes. Then, we repeated this volumetric acquisition at every 20 minutes. We updated the text for explaining the DIV term and speed of data acquisition (Lines 293,478-480).

Figure 6 – I am slightly puzzled by the displacement statistics shown here. It looks like the data pooled is from organoids that are imaged for different periods of time, in which case it is misleading to report statistical differences between WT and MT groups. I imagine the "straightness" measure would also be biased by the inclusion of samples recorded from different imaging durations.

We agree with the reviewer that comparing the displacement and straightness values of the cells which disappeared from the field of view before finalizing the overall imaging duration of 12 hours may not be reasonable. Therefore, we removed these data and updated Figure 6 as well as Figure 5. Please see item 11 in the Essential Revisions.

L343 "incubation in index matching solution improved the imaging depth". I could not find this in the Methods, can you please detail? Also, did you quantify this through attenuation length, maximum depth at which cells could be observed, or another measure?

We would like to thank the reviewer for highlighting the effect of index matching solution on the imaging depth. We did not need to apply any index matching solution for 60% of the WT and MT organoids that we used in this study. For the remaining part of the organoids, index matching gel improved the imaging depth maximum 50% as we quantified in Supp. Figure 19 in the revised manuscript. Please see item 7 above in Essential Revisions.

L355 you propose some explanations for the result that WT and MT organoids show a different attenuation length. Did you make any measurements to validate these explanations? The stated structural differences would appear to be measurable from 3D organoid stacks, and thus may provide support to the conjecture.

We explained the differences in extinction lengths of WT and MT organoids based on the results that we presented in Figure 3C. We suggest that higher surface area with lower radial thickness in MT organoids resulted in irregular shapes of ventricular zones. In addition, number of ventricular zones in MT organoids is significantly higher than that of WT organoids. Therefore, higher number of irregular shaped ventricular zones may induce more scattering in MT organoids which results in their lower extinction lengths. We added this additional explanation in the Discussion part (lines 390-394, also see item 8 in Essential Revisions).

L413 exits rather than exists.

We corrected this typo (Line 461)

Reviewer #3 (Recommendations for the authors):

I have three broad critiques in regard to the cited literature, validation experiments, and additional applications:

1) Prior literature: Authors should revise the text to better reflect the current approaches in organoid imaging. There is a significant body of literature describing migration imaging using brain organoid models (e.g. Birey et al., 2017; Bagley et al., 2017; Xiang et al., 2017; Klaus et al., 2019; Bajaj et al., 2021; Birey et al., 2022), including ones that use intact (not sliced) assembloid preparations not cited in the paper (Xiang et al., 2017; Birey et al., 2017, 2022), as opposed to what's been suggested by the authors (Lines 60 and 61; Lines 258 and 259). These studies use viral labeling to visualize the cells of interest during long-term (days-long) migration imaging without seemingly affecting organoid health, again as opposed to what's been suggested (Line 66). Authors should revise to reflect more accurately what's been achievable by the current state of the art. This surely does not take away from the novelty of the 3P/THG method but better clarifies the limitations being addressed.

We updated the Introduction by adding the references that Reviewer 3 suggested. Please refer to item 1 above in the Essential Revisions.

2) Validation experiments: In early characterization experiments, authors convincingly show that THG signal is higher in regions populated with neurons compared to denser VZ-like regions populated with progenitors. As it is important to translate the THG single as accurately as possible, similar validation experiments are needed to better link THG dynamics to specific cellular events in the following phenotyping experiments. For example, the authors show a higher surface area and about a 2-fold increase in the number of ventricles in RTT organoids. An optical clearing approach for the SA quantification and/or a simple cryosection quantification for the number of VZ-like zones should be able to reveal a change of this magnitude. Authors electroporate a CAG-driven plasmid to visualize cells that they have previously reported, albeit in three samples, to almost exclusively (~95%) capture neurons (Delepine et al., 2021). As the rationale as to why a generic promoter shows lineage selectivity is unclear and as the imaging field of interest also includes other cell types that migrate in a similar fashion (i.e. radial glia that go inter-kinetic nuclear migration), a parallel labeling strategy, e.g. lineage-specific viral vectors or expression vectors with different promoters, to further delineate the specific lineage captured by the THG signal would be informative. For example, Figure 3A and B show VZ-like regions with high TGH signal within their cores, which are unlikely to be neurons. This could resolve questions of this nature.

We would like to thank the reviewer for the suggestions. In future follow-up studies, multi-color multiphoton imaging (including THG) can be performed to delineate further the cell-type and/or lineage-type information we can obtain for migration experiments in details.

3) Is the RTT migration phenotype caused solely by the more tortuous VZ region they need to navigate? Once they reach THG-high regions (cortical plate), does the phenotype recover? If not, does this indicate additional cell-intrinsic deficits?

Please see item 3 in the Essential Revisions.

4) An expanded discussion could serve the reader well to envision the true potential of the 3P/THG system. Do the authors envision additional features can be extracted in older (e.g. myelinated axons) or novel (e.g. organoids with endothelia) organoids using their models? Is THG-coupled calcium imaging possible, for example where one can couple VZ calcium dynamics, such as waves (Weissman et al., 2004), to neuronal dynamics? Can other disorders that cause structural abnormalities like cancers, tuberous sclerosis, or heterotopias be modeled using organoids and differentiated from one another using the 3P/THG signal?

We have added this at the end of Discussion. Please see item 6 in the Essential Revisions.

https://doi.org/10.7554/eLife.78079.sa2

Article and author information

Author details

  1. Murat Yildirim

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Neuroscience, Cleveland Clinic Lerner Research Institute, Cleveland, United States
    3. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Conceptualization, Software, Formal analysis, Funding acquisition, Methodology, Writing – original draft
    For correspondence
    muraty@mit.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-0853-2557
  2. Chloe Delepine

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Methodology, Writing – original draft, Generated cerebral organoids
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1856-9583
  3. Danielle Feldman

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Methodology, Generated cerebral organoids
    Competing interests
    No competing interests declared
  4. Vincent A Pham

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Methodology, Generated cerebral organoids
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0001-7083-9360
  5. Stephanie Chou

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Methodology, Generated cerebral organoids
    Competing interests
    No competing interests declared
  6. Jacque Ip

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    3. School of Biomedical Sciences, The Chinese University of Hong Kong, Hong Kong, China
    Contribution
    Methodology, Carried out organoid electroporation
    Competing interests
    No competing interests declared
  7. Alexi Nott

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Methodology, Provided cell lines
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-2029-7193
  8. Li-Huei Tsai

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Methodology, Provided cell lines
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1262-0592
  9. Guo-Li Ming

    Department of Neuroscience and Mahoney Institute for Neurosciences, Institute for Regenerative Medicine, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States
    Contribution
    Methodology, Provided cell lines
    Competing interests
    No competing interests declared
  10. Peter TC So

    Deparment of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Methodology, Writing – original draft, Writing - review and editing
    Competing interests
    No competing interests declared
  11. Mriganka Sur

    1. Picower Institute of Learning and Memory, Massachusetts Institute of Technology, Cambridge, United States
    2. Department of Brain and Cognitive Sciences, Massachusetts Institute of Technology, Cambridge, United States
    Contribution
    Resources, Supervision, Funding acquisition, Writing – original draft, Writing - review and editing
    For correspondence
    msur@mit.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-2442-5671

Funding

National Institute of Biomedical Imaging and Bioengineering (K99EB027706)

  • Murat Yildirim

National Institute of Mental Health (MH085802)

  • Mriganka Sur

National Institute of Mental Health (NS090473)

  • Mriganka Sur

National Institute of Biomedical Imaging and Bioengineering (P41-EB015871)

  • Peter TC So

National Science Foundation (EF1451125)

  • Mriganka Sur

National Institute of Neurological Disorders and Stroke (R35NS097370)

  • Guo-Li Ming

JPB Foundation

  • Li-Huei Tsai

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

This work was supported by US National Institute of Health (NIH) grants MH085802 and NS090473 (MS), 4-P41-EB015871 (PTCS), K99EB027706 (MY), US National Science Foundation (NSF) grant EF1451125 (MS), a Picower Institute Engineering Collaboration Grant (MS, PTCS and MY), and an equipment grant from the Massachusetts Life Sciences Initiative. AN was supported by the UK Dementia Research Institute, which receives its funding from UK DRI Ltd, funded by the UK Medical Research Council, Alzheimer’s Society, and Alzheimer’s Research UK. LHT was supported by JPB Foundation. G-lM was supported by National Institutes of Health grant R35NS097370. We thank Dr. Alex Albenese and Dr. Kwunghun Chung for their assistance on Hprotos generation. We are grateful to members of the Sur and So labs for their help.

Senior and Reviewing Editor

  1. Jeannie Chin, Baylor College of Medicine, United States

Reviewers

  1. Anita Bhattacharyya, University of Wisconsin, Madison, United States
  2. Simon R Schultz, Imperial College London, United Kingdom

Publication history

  1. Received: February 22, 2022
  2. Preprint posted: March 1, 2022 (view preprint)
  3. Accepted: July 8, 2022
  4. Version of Record published: July 29, 2022 (version 1)

Copyright

© 2022, Yildirim et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 752
    Page views
  • 171
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Murat Yildirim
  2. Chloe Delepine
  3. Danielle Feldman
  4. Vincent A Pham
  5. Stephanie Chou
  6. Jacque Ip
  7. Alexi Nott
  8. Li-Huei Tsai
  9. Guo-Li Ming
  10. Peter TC So
  11. Mriganka Sur
(2022)
Label-free three-photon imaging of intact human cerebral organoids for tracking early events in brain development and deficits in Rett syndrome
eLife 11:e78079.
https://doi.org/10.7554/eLife.78079

Further reading

    1. Neuroscience
    Kevin J Miller et al.
    Research Article

    Humans and animals make predictions about the rewards they expect to receive in different situations. In formal models of behavior, these predictions are known as value representations, and they play two very different roles. Firstly, they drive choice: the expected values of available options are compared to one another, and the best option is selected. Secondly, they support learning: expected values are compared to rewards actually received, and future expectations are updated accordingly. Whether these different functions are mediated by different neural representations remains an open question. Here we employ a recently-developed multi-step task for rats that computationally separates learning from choosing. We investigate the role of value representations in the rodent orbitofrontal cortex, a key structure for value-based cognition. Electrophysiological recordings and optogenetic perturbations indicate that these representations do not directly drive choice. Instead, they signal expected reward information to a learning process elsewhere in the brain that updates choice mechanisms.

    1. Neuroscience
    Bin Wan et al.
    Research Article Updated

    The human cerebral cortex is symmetrically organized along large-scale axes but also presents inter-hemispheric differences in structure and function. The quantified contralateral homologous difference, that is asymmetry, is a key feature of the human brain left-right axis supporting functional processes, such as language. Here, we assessed whether the asymmetry of cortical functional organization is heritable and phylogenetically conserved between humans and macaques. Our findings indicate asymmetric organization along an axis describing a functional trajectory from perceptual/action to abstract cognition. Whereas language network showed leftward asymmetric organization, frontoparietal network showed rightward asymmetric organization in humans. These asymmetries were heritable in humans and showed a similar spatial distribution with macaques, in the case of intra-hemispheric asymmetry of functional hierarchy. This suggests (phylo)genetic conservation. However, both language and frontoparietal networks showed a qualitatively larger asymmetry in humans relative to macaques. Overall, our findings suggest a genetic basis for asymmetry in intrinsic functional organization, linked to higher order cognitive functions uniquely developed in humans.