Rostral Hypothalamic Differentiation With Minimal Exogenous Signals in Human Naive Induced Pluripotent Stem Cells


 Familial neurohypophyseal diabetes insipidus (FNDI) is a degenerative disease of vasopressin (AVP) neurons. Studies in mouse in vivo models indicate that accumulation of mutant AVP prehormone is associated with FNDI pathology. However, studying human FNDI pathology in vivo is technically challenging. Therefore, an in vitro human model needs to be developed. When exogenous signals are minimized in the early phase of differentiation in vitro, mouse embryonic stem cells (ESCs) differentiate into AVP neurons, whereas human ESCs/induced pluripotent stem cells (iPSCs) die. Human ES/iPSCs are generally more similar to mouse epiblast stem cells compared to mouse ESCs, which are termed as primed and naive, respectively. In this study, we converted human FNDI-specific iPSCs from primed to naive cells, and found improved cell survival under minimal exogenous signals and differentiation into rostral hypothalamic organoids. Overall, this method provides a simple and straightforward differentiation direction, which may improve the efficiency of hypothalamic differentiation.


Introduction
The hypothalamus plays an essential role in maintaining physiological homeostasis. One of these is the generation and secretion of arginine vasopressin (AVP), which controls water balance. Familial neurohypophyseal diabetes insipidus (FNDI) is characterized by a progressive decrease in AVP secretion caused by the degeneration of neurons in the hypothalamus. Mutations in the AVP gene locus have been identi ed in patients with FNDI, and the majority of these mutations are located in the region encoding the AVP carrier protein, neurophysin II (NPII) 1 . Therefore, we generated FNDI mice with a Cys98stop (previously called Cys67stop) mutation 2 , which occurs in patients with FNDI 3 , for pathological analysis.
Our previous studies using Cys98stop-knock-in FNDI mouse models [4][5][6][7][8][9][10][11][12][13] , indicated that endoplasmic reticulum (ER) stress caused by the accumulation of mutant proteins in the ER could be associated with FNDI pathology. However, further studies, including the use of human models, are needed to investigate the overall pathological mechanisms of FNDI. FNDI-related studies in humans are restricted by technical limitations because hypothalamic biopsies cannot be performed in living humans and the loss of AVP neurons in human patients with FNDI has been reported only in autopsy studies [14][15][16] . Therefore, AVP neurons derived from human FNDI-speci c induced pluripotent stem cells (iPSCs) are a promising human model for pathological analysis and drug development.

Pluripotent stem cells can differentiate into various cells or tissues in three-dimensional cultures termed
"serum-free culture of embryoid body-like aggregates with quick re-aggregation" (SFEBq) [17][18][19][20][21][22][23][24][25][26][27] . Based on the differentiation of hypothalamic neurons from mouse embryonic stem cells (mESCs) by SFEBq 28 , we improved the differentiation e ciency of mouse iPSCs 29 . In these methods, the differentiation of pluripotent stem cells into neural tissue is achieved by regulating the positional information added to the medium. The hypothalamus in the embryo originates from the most rostral part of the neural plate (Fig.   S1A). To differentiate the hypothalamus from mouse ES/iPSCs, it is important to remove exogenous signals strictly in the early stages of differentiation, and to control the positional information so that it represents the most rostral side 28,29 . The medium with minimized exogenous signals is called a growth factor-free chemically de ned medium (gfCDM) 28 . With gfCDM, mouse ES/iPSCs differentiated into the rostral hypothalamus, including AVP neurons.
However, this theoretical concept in mouse ES/iPSCs has not yet been realized in the differentiation of human ES cells (hESCs). Under conditions with minimal exogenous signals using gfCDM, hESCs cannot form aggregates and die within a few days. The addition of knock out serum replacement (KSR) allows hESCs to form aggregates, but results in their differentiation into telencephalic progenitor cells. This means that the positional information contained in KSR shifts the differentiating position to caudal and dorsal. Thus, to achieve hypothalamic differentiation, it is necessary to use KSR-containing medium to guide the cells toward the telencephalon in the rst step, and then force them to change their position to the hypothalamus by adding rostralizing and ventralizing signals in the second step 30 (Fig. S1B). We speculate that this two-step differentiation strategy causes less e cient differentiation to the hypothalamus. Therefore, it is necessary to improve the method for hESC/hiPSC differentiation into the hypothalamus. In this study, we focused on the difference in the starting points between hES/iPSCs and mES/iPSCs.
Mouse epiblast stem cells (mEpiSCs) derived from post-implantation embryos 31,32 have characteristics different from those of conventional mESCs derived from pre-implantation embryos. Nichols et al.
advocated the concept of 'naive' and 'primed', which correspond to mESCs and mEpiSCs, respectively 33 .
Human ESCs are recognized to be in a primed state, similar to mEpiSCs. In fact, hESCs and mEpiSCs share the same colony morphology and vulnerability to isolation. Therefore, in this study, we examined the hypothesis that if hESCs/iPSCs were converted to a naive state, they could be differentiated in the same way as mESCs.

Results
Primed FNDI-speci c hiPSCs fail to aggregate and differentiate poorly using conventional methods The differentiation of hypothalamic neurons from mESCs and hESCs was achieved by SFEBq 28,30 , but there were differences between the methods. With minimal exogenous signals, mESCs differentiated into hypothalamic neurons whereas hESCs failed to form aggregates. Therefore, following the hESC method, primed FNDI hiPSCs (FDI-02) were dissociated into single cells and suspended in gfCDM containing 5% KSR. Under these conditions, they aggregated quickly. When the concentration of KSR was maintained at 5%, the aggregates grew without collapsing. However, as previously reported, maintaining the concentration of KSR at 5% without other exogenous signals caused the hESCs to differentiate into FOXG1-positive telencephalic progenitors 30 . Addition of BMP4 and SAG caused hESCs to differentiate into hypothalamic precursors, but at a lower e ciency than that in the method using mESCs with minimal exogenous signals 28, 30 . To minimize the effects of exogenous signals, we reduced their concentrations and shortened their incubation period (Fig. S2A). When the concentration of KSR at day 0 was reduced to 2.5% or KSR-free gfCDM was used after day 3, the aggregates collapsed or formed cysts (Fig. S2B).
Our previous study using miPSCs showed that stopping miPSC colonies to clumps of approximately 10-20 cells, rather than isolating them to a single cell for SFEBq, can facilitate miPSC aggregation 29 . Accordingly, SFEBq using clumps was performed to prevent collapse and cyst formation. When the KSR concentration at day 0 was 2.5% and KSR-free gfCDM was used after day 3, the clumps quickly aggregated and grew without collapsing. However, even under these conditions, primed FDI-02 differentiated into FOXG1-positive telencephalic progenitors (Fig. S2B). When a lower concentration of KSR was examined on day 0, the aggregates collapsed or formed cysts. Although SFEBq using clumps could ease the effects of reduced KSR concentration to some extent, it was di cult for primed FDI-02 to differentiate into hypothalamic neurons with minimal exogenous signals.

Primed-to-naive conversion improves the differentiation of FDI-02 into hypothalamic precursors
To minimize the exogenous signals needed for FDI-02 differentiation, we assessed a method that enables human pluripotent stem cells to be treated in the same manner as mouse pluripotent stem cells.
Compared with conventional mESCs, human pluripotent stem cells show some distinctive features, including differences in developmental identity. In particular, mEpiSC derivation 31,32 shows alternative pluripotency, which is associated with primitive streak-stage late epiblasts 34 . The terms 'naive' and 'primed,' which correspond to mESCs and mEpiSCs, respectively, refer to the early and late stages of epiblast ontogeny 33 . Human pluripotent stem cells share de ning features with primed mEpiSCs rather than with naive mESCs. Takashima et al. and Guo et al. reported methods for converting cells from the primed state to the naive state [35][36][37] . We hypothesized that these methods could help hiPSCs differentiate into hypothalamic neurons by minimizing exogenous signals.
Primed FDI-02 cells were cultured under feeder-free conditions using mTeSR1 TM culture medium.
Thereafter, the medium was replaced with RSeT TM . The colony morphology changed from at to tightly packed and domed, and TFCP2L1, a naive marker 38 , was expressed ( Fig. 1A-D), con rming conversion from the primed to naive state.
SFEBq was performed with naive FDI-02 single cells and primed FDI-02 clumps. The use of a medium containing neither KSR nor Y-27632, a Rho-associated kinase (ROCK) inhibitor, from an early stage was compared with a medium containing both throughout the culture ( Fig. 2A). Even when the initial concentration of KSR was reduced to 0.7%, the single cells and clumps quickly aggregated under all conditions. Most primed aggregates formed cysts or collapsed over time, whereas naive aggregates maintained a healthy state. The naive aggregates showed signi cantly fewer changes on day 30 (Fig. 2B, C).
Aggregates were evaluated by immunohistochemistry. The expression of BRN2, an AVP precursor marker 39,40 , was signi cantly increased in naive aggregates in the shortened period condition with KSR and Y-27632 (Fig. 2D, E). This is consistent with increased hypothalamic differentiation with minimal exogenous signals.
In summary, conversion to the naive state enabled minimization of exogenous signals in SFEBq with hiPSCs. Furthermore, the minimization of exogenous signals improved the e ciency of differentiation into hypothalamic-like cells.
Hypothalamic organoids derived from FDI-02 expressed mutant NPII Based on these results, the early stage of the differentiation protocol was optimized. We have previously reported that the addition of FGF8b to late-stage culture improves the e ciency of AVP neuron differentiation 29 . Therefore, we adopted this method and established an overall protocol (Fig. 3A).
SFEBq using the naive human iPSC cell line, 201B7, was performed according to the established protocol.

Discussion
In this study, we investigated a novel perspective in the differentiation of AVP neurons, regarding whether cells were in a naive or primed state before differentiation, rather than considering the early or later stages of the differentiation protocol 28-30 . We found two advantages. First, the conversion of cells from primed to naive solved problems such as decreased survival, collapse of aggregates, and cyst formation by minimizing exogenous signals in the SFEBq method for the differentiation of human pluripotent stem cells into AVP neurons. Second, the strict removal of exogenous signals by conversion to the naive type increased the rate of cells positive for BRN2, which is an essential transcription factor for AVP differentiation [39][40][41][42][43] . These ndings indicate an improved e ciency of AVP neuron differentiation.
The differentiation potential of pluripotent stem cells varies among cell lines 44 . When differentiating the rostral part of the central nervous system such as the hypothalamus, the use of naive cells could be one way to solve this problem. It is di cult to differentiate the primed hESCs/hiPSCs into rostral hypothalamic neurons because that they cannot survive in culture conditions with minimal exogenous signals. Historically, it is known that dissociation of primed hESCs causes apoptosis due to the Rhohigh/Rac-low state 45 . Thus, treatment with ROCK inhibitors is very effective for the dissociation process of primed hESCs 46 . Using a ROCK inhibitor, primed hESCs/hiPSCs can be differentiated into non-rostral nervous systems such as the telencephalon 17,18 , cerebellum 22 , and spinal cord 24 . However, for inducing the rostral part of the central nervous system, such as AVP neurons, additional conditions are required to induce the rostral edge of the neural plate by using minimal exogenous signals as the positional information during differentiation 28, 29 . In the primed hESC/hiPSC cultures, use of the ROCK inhibitor alone cannot maintain their aggregation in the state of minimal exogenous signals. Nutrients such as KSR must be added to some extent for their survival, resulting in their differentiation into FOXG1-positive cerebral tissues. Additional BMP4 and SHH signals are necessary to primed hESCs/hiPSCs for repositioning from the cerebrum to the hypothalamus 30 . This involves a two-step process: once the cells are posteriorized, they are repositioned anteriorly. This is thought to be the reason for the previously low e ciency of hypothalamic differentiation. By converting hiPSCs from the primed-type to the naive type, addition of ROCK inhibitor and KSR can be reduced to the minimum, resulting in a simpler straightforward differentiation direction, which may improve the e ciency of differentiation into the rostral hypothalamus.
Analyses of our FNDI model mice, which possess the Cys98stop mutation 2,3 , revealed that mutant NPII was accumulated in the ER-associated compartment (ERAC) in AVP neurons 4,5 . In this study, we clearly demonstrated mutant NPII expression in FNDI-speci c hiPSC-derived AVP neurons. Accumulation of mutant proteins in the ER is implicated in the pathophysiology of many diseases, including FNDI 1, 6-13,47,48 ; therefore, FNDI-speci c hiPSC-derived AVP neurons are a promising human model of ER stress and are a valuable resource for drug development. We plan to analyze characteristic structures, such as the ERAC, which was con rmed in mice by electron microscopy, in FNDI-speci c hiPSC-derived AVP neurons. We would also like to undertake a functional investigation for pathological analysis and drug development in future studies.
In this study, conversion from the prime to naive state was performed based on the reports by Takashima et al. and Guo et al. [35][36][37] . However, there are several other reports concerning conversion to the naive type 49-57 and a novel concept termed 'formative type' has also been advocated between the naive and primed types 58 . Although we achieved some success by inducing conversion to the naive type, further studies are needed to determine the optimal state of the cells when initiating differentiation from hiPSCs. Based on this study, we will continue to improve the differentiation method, analyze the expression pattern of mutant proteins in detail, and research drug reactivity.

Methods
These experiments were approved by the ethics committee of Nagoya University and performed in accordance with relevant guidelines and regulations. Informed consent was obtained from all subjects.

Generation of FNDI hiPSCs and a control cell line
We generated iPSCs from a patient with FNDI with a Cys98Stop mutation 3 based on ethical approval from Nagoya University Committee (2013-0228-2). To establish patient-derived iPSCs, peripheral blood monocular cells of the patient were reprogrammed using episomal vectors (expressing OCT3/4, SOX2, KLF4, L-MYC, LIN28, EBNA1, and p53shRNA) as described previously 59,60 . The iPSCs showed embryonic stem cell-like morphology and normal karyotype (Fig. S3A). These cells expressed the undifferentiated markers (Fig. S3B, C) and could differentiate into the three germ layers in vitro (Fig. S3D), indicating that these cells were pluripotent (deposited in RIKEN BRC as HPS1011 and HPS1904). We named them as For passaging, primed hiPSC colonies were harvested by incubation in 0.25% (w/v) trypsin and 0.1 mg/mL collagenase IV in PBS containing 20% (v/v) KSR and 1 mM CaCl 2 for 6-8 min at 37°C. The harvested clumps were broken into smaller pieces by gentle pipetting. The passages were performed at a split ratio of 1:4-6.
Reset of hiPSCs from the primed state to naive state and maintenance incubation hiPSCs harvested during passaging, were suspended in mTeSR1 TM medium (ST-85850; STEMCELL Technologies, Vancouver, Canada), distributed on a Matrigel-coated dish (354277; Corning, Corning, New York, USA), and incubated under 5% CO 2 at 37°C. After 24-36 h, the mTeSR1 medium was replaced with RSeT TM feeder-free medium (ST-05975; STEMCELL Technologies) and cells were incubated under hypoxic conditions (5% O 2 , 5% CO 2 ) at 37°C. The medium was changed every second day. Reset to the naive state was con rmed by changes in colony morphology and immunohistochemistry.
For passaging, naive hiPSCs were harvested using TrypLE Express (12605-010; Thermo Fisher, Waltham, Massachusetts, USA). Cells were counted and 1.1 × 10 6 cells were distributed on a 10 cm Matrigel-coated dish. The medium was changed after 24-36 h and thereafter, every second day. This passaging procedure was performed every 4-6 days.

Immunohistochemistry
Aggregates were xed with 4% PFA for 10-15 min and embedded in OCT compound (4583; Sakura Finetek, Tokyo, Japan). Ten micron thick sections were cut using a cryostat, mounted on slides, and xed in 4% PFA for 10 min.

Statistics and reproducibility
We have described the exact n values for each experiment in the main text and gure legends. IBM SPSS Statistics (IBM, Armonk, New York, USA) was used for the statistical analyses. Two-group comparisons were performed using the two-tailed unpaired t-test. Signi cance was set at P < 0.05.

Declarations Data Availability
The datasets generated during and/or analyzed during this study are available from the corresponding author upon reasonable request.   Scheme of the culture protocol. Using a medium containing neither KSR nor Y-27632 from an early stage or a medium containing KSR and Y-27632 throughout, single naive hiPSCs (white arrows) were compared with primed hiPSC clumps (blue arrows). (B, C) On day 30, naive aggregates showed signi cantly fewer changes, such as cyst formation or collapse (n=3). Mean ± SEM. *P < 0.05; **P < 0.01, two-tailed unpaired t-test. On day 30, primed aggregates scarcely expressed BRN2; however, BRN2 expression was observed by conversion to the naive state (D), and was signi cantly increased by shortening the period of KSR and Y-27632 treatment (n=8) (E). Mean ± SEM. **P < 0.01, two-tailed unpaired t-test. OTP (green), BRN2 (red), FOXG1 (white). The increased expression of the AVP precursor marker, BRN2, indicated that minimization of exogenous signals improved the e ciency of differentiation into hypothalamic-like cells. For all relevant panels, nuclear counterstaining was performed with DAPI (blue). Scale bars, 100 µm.  FNDI-speci c hiPSCs generate mutant NPII. Expression of normal NPII (red) was observed in SFEBq from FDI-02. Additionally, the expression of mutant NPII (mNPII) (green) was con rmed, which was not expressed in 201B7-derived AVP neurons. For all relevant panels, nuclear counterstaining was performed with DAPI (blue). Scale bars, 50 µm.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.