Hypoimmunogenic hPSCs could be long-term maintained in culture and keep three germinal differentiation potency
We constructed hypoimmunogenic hPSCs with different strategies, biallelic lesion of B2M gene to remove all surface expression of classical and nonclassical HLA class I molecules (B2Mnull), biallelic homologous recombination of nonclassical HLA-G1 to the B2M loci to knockout B2M while express membrane-bound β2m-HLA-G1 fusion proteins (B2MmHLAG), and ectopic expression of soluble and secreted β2m-HLA-G5 fusion proteins in B2MmHLAG hPSCs (B2Mm/sHLAG). hPSCs constructed with these three strategies have shown robust immunotolerance to CD8+ T cells and NK cells both in vitro and in vivo allografts .
To study whether these engineered hypoimmunogenic hPSCs retain normal self-renewal and differentiation potency, we continuously cultured these cells to more than 30 passages. After passage, all three hypoimmunogenic hPSCs exhibited typical morphology with large nuclear/cytoplasmic ratios, multiple and prominent nucleoli, and round colonies with clear edges resembling of the wild type (WT) control (Fig. 1A). Immunostaining experiments revealed that all three hypoimmunogenic hPSCs had uniform nuclear expression of typical pluripotent transcription factors, including OCT4 and SOX2 (Fig. 1B). These data suggest that hypoimmunogenic hPSCs engineered with different strategies retained their pluripotency and could be finely maintained in culture. To transcriptionally characterize the engineered hypoimmunogenic hPSCs, we performed RNA-seq analyses on these cells at passage 55. Dimensionality reduction and clustering by principal component analysis (PCA) demonstrated that hypoimmunogenic hPSCs were clustered with WT hPSCs, but not in vivo differentiated teratomas (Fig. 1C, 1D). Moreover, similar to the WT control, all three hypoimmunogenic hPSCs showed robust expression of pluripotency genes while lack of expression of three germ layer-featured genes (Fig. 1E), highlighting a non-differentiated state of these hypoimmunogenic hPSCs.
To investigate their differentiation potency, three types of hypoimmunogenic hPSCs were subcutaneously injected into non-obese diabetic/severe combined immune-deficient (NOD/SCID) mice, with the WT hPSCs as a control, respectively. For all four groups, teratomas were visibly formed at similar occurrence rates after 2 months of injection. Teratomas were then resected, fixed in paraformaldehyde, followed by hematoxylin-eosin (H&E) staining (Fig. 1C). We observed ectodermal neural tube-like tissues, mesoderm-derived cartilage like tissues and endodermal intestine like tissues in both WT- and the three types of hypoimmunogenic hPSCs-injected groups. The dissected teratomas from all groups were also subjected to RNA-seq. Heatmap analyses revealed that teratomas from all three hypoimmunogenic hPSCs and WT hPSCs had similar gene expression patterns featured with all three germ layers (Fig. 1E). Of note, similar to the WT control, we did not observe elevated pluripotent marker genes in any teratomas derived from the three hypoimmunogenic hPSCs. Together, these results suggest that the hypoimmunogenic hPSCs with various HLA expression patterns and immune compromising spectrums could be efficiently differentiated into all three germ layers.
Neurons differentiated from hypoimmunogenic hPSCs functionally mature and form neural circuits
To study whether hypoimmunogenic hPSCs could be efficiently specified into functional tissue cells, we firstly differentiated WT hPSCs and all three hypoimmunogenic hPSCs toward cortical neurons via the Dual-Smad inhibition differentiation protocol as previously described [38–40] (Fig. 2A). Neural ectoderm (NE) cells appeared on post-differentiation day 7 and they formed typical neural tube-like rosettes on day 10 in all four groups. Immunostaining experiments showed that on day 10, most NE cells were positive for PAX6 (Fig. 2B), indicating synchronized neural induction in control and hypoimmunogenic hPSC groups. Quantification of the percentage of PAX6+ cells in NE cultures further strengthened the conclusion that hypoimmunogenic hPSCs were almost uniformly specified into the PAX6+ NE (Fig. 2C).
The specified NE were further differentiated into neural precursors and then to neurons (Fig. 2A). On post-differentiation day 30, neurons yielded from all three hypoimmunogenic hPSCs were positive for TUJ1 with extended long projections (Fig. 2D). The differentiated neurons from hypoimmunogenic hPSCs also expressed punctated synapsin1 (SYN1), a hallmark protein of presynaptic membrane, at 8 weeks post differentiation (Fig. 2D), indicating gradual synaptic maturation of WT and hypoimmunogenic hPSC-derived neurons.
To characterize the functional maturity of the three hypoimmunogenic hPSC-derived neurons, whole-cell patch-clamp electrophysiological recording experiments were performed. All three hypoimmunogenic hPSC-derived neurons had the ability to fire action potentials (APs) repetitively in response to current injection (Fig. 2E, 2F). AP properties were then quantified and compared to evaluate the electrophysiological maturity of the neurons. The resting membrane potentials (RMPs) of WT-, B2Mnull-, B2MmHLAG- and B2Mm/sHLAG-derived neurons were -57.18±2.510 mV, -55.50±1.564 mV, -61.58±4.026 mV, and -58.27±3.627 mV, respectively. The peak amplitudes were 83.70±2.386 mV, 79.26±6.053 mV, 95.16±6.068 mV, and 95.50±7.588 mV, respectively. There were no significant differences in all four groups by analyzing these AP parameters (Fig. 2G, 2H), indicating that the hypoimmunogenic hPSC-derived neurons attained the capacity to fire trains-of-action potentials by 8-10 weeks, the same time point for neurons derived from the WT hPSCs. In addition, there were no differences in synaptic connectivity amongst the WT and hypoimmunogenic hPSC-derived neurons (Fig. 2I, 2J, 2K). The frequencies of spontaneous synaptic activity from WT-, B2Mnull-, B2MmHLAG- and B2Mm/sHLAG-derived neurons were 1.33±0.4024 Hz, 4.032±1.676 Hz, 0.8269±0.2159 Hz, and 0.6134±0.1601 Hz, respectively. The amplitudes of spontaneous synaptic activity were 28.46±8.173 pA, 29.63±4.473 pA, 50.98±7.815 pA, and 48.94±21.18 pA, respectively. Taken together, these data suggest that the three types of hypoimmunogenic hPSCs are normally programmed into electrophysiologically mature neurons in culture.
Hypoimmunogenic hPSC-derived cardiomyocytes spontaneously contract and possess functionally electrophysiological characteristics
Cardiomyocyte transplantation has been considered as a replacement for heart transplantation and conventional regenerative therapies [41–43]. To study whether hypoimmunogenic hPSCs hold the ability to differentiate into functional and mature cardiomyocytes, WT, B2Mnull, B2MmHLAG and B2Mm/sHLAG hPSCs were differentiated toward a cardiomyocyte fate with a well-characterized protocol [44–45] (Fig. 3A). On day 8, cells differentiated from WT and all three hypoimmunogenic hPSCs were uniformly positive for NKX2.5, a cardiac transcription factor, suggesting synchronized cardiac fate specification (Fig. 3B). On day 12, cardiomyocytes derived from hypoimmunogenic hPSCs as well as WT hPSCs began to spontaneously contract, and these cardiomyocytes beat robustly even after 100 days of differentiation (Supplementary video 1-4). The hypoimmunogenic hPSC-derived cardiomyocytes were also positive for cardiac troponin T (cTnT), a highly cardiac-specific myofilament protein. Quantification studies revealed that the percentage of cTnT+ cardiomyocytes in the entire culture of WT and all three hypoimmunogenic hPSCs were over 95% with batch-to-batch consistency (Fig. 3B, 3C). To evaluate the cardiac sarcomere organization, cells were labeled with α-actinin, the Z-line marker of the sarcomere, and myosin light chain 2 atrial isoform (MLC2a), the A-band marker of the sarcomere, separately. Again, hypoimmunogenic hPSC-derived cardiomyocytes showed typical α-actinin and MLC2a staining (Fig. 3D). Taken together, immunolabeling of multiple myofilament proteins indicates that a well-organized sarcomeric structure can be similarly developed in all three hypoimmunogenic hPSC-derived cardiomyocytes.
To assess the maturity of cardiomyocytes derived from hypoimmunogenic hPSCs, we performed electrophysiological studies in cardiomyocytes derived from WT, B2Mnull, B2MmHLAG and B2Mm/sHLAG hPSCs 30-35 days post differentiation, a time window when ventricular-like cells being the predominant phenotype . A majority of hypoimmunogenic hPSC-derived cardiomyocytes exhibited spontaneous ventricular-like electrical activity, similarly to Burridge’s report . Representative recordings of ventricular-like action potentials were shown in Fig. 3E. Specifically, the RMPs of WT-, B2Mnull-, B2MmHLAG- and B2Mm/sHLAG-derived ventricular-like cells were -55.30±1.398 mV, -55.60±2.230 mV, -56.64±1.421 mV, and -56.65±1.779 mV, respectively. The action potential amplitudes (APAs) of WT-, B2Mnull-, B2MmHLAG- and B2Mm/sHLAG-derived ventricular-like cells were 101.0±1.610 mV, 97.39±1.530 mV, 95.52±2.343 mV, and 100.9±1.935 mV, respectively. The action potential durations (APDs) at different levels of repolarization (90% and 50%, APD90 and APD50) of WT-, B2Mnull-, B2MmHLAG- and B2Mm/sHLAG-derived ventricular-like cells were 221.6±35.50 ms and 173.7±25.68 ms, 288.1±27.89 ms and 233.2±25.85 ms, 278.9±21.61 ms and 213.3±16.27 ms, and 191.7±13.72 and 144.0±11.20 ms, respectively. The maximal rates of depolarization (dV/dtmax) of WT-, B2Mnull-, B2MmHLAG- and B2Mm/sHLAG-derived ventricular-like cells were 17.70±2.163 V/s, 20.70±4.778 V/s, 17.46±3.062 V/s, and 15.46±1.509 V/s, respectively. Quantification data of the AP properties of ventricular cells derived from each group were presented and they showed no significant differences among groups (Fig. 3F, 3G, 3H, 3I, 3J). These data suggest that the engineered hypoimmunogenic hPSCs by modifying HLA class I molecules could be faithfully differentiated into functionally mature cardiomyocytes with proper cytoskeleton morphology and electrophysiological activities.
Various from neurons and cardiomyocytes, which are typical for their induced or spontaneous electrophysiological excitability, hepatocytes have more complex cellular functions related to metabolic pathways, such as cargo transport, insulin-regulated glucose metabolism, and detoxification. We used a four-step protocol to drive hPSCs toward a definitive endoderm and then a hepatocyte fate [47–48] (Fig. 4A). Immunostaining experiments revealed that the vast majority of the differentiation derivatives from WT and all three hypoimmunogenic hPSCs were positive for alpha fetoprotein (AFP) on day 13 (Fig. 4B), suggesting a uniform hepatic precursor fate (HPCs) obtained in all cultures. On day 21, albumin (ALB), a marker of mature hepatocytes, was detected in hepatocyte-like cells (Fig. 4B). FACS analysis revealed that ~90% cells were positive for ALB in WT and all three hypoimmunogenic hPSCs cultures (Fig. 4C). These results suggest that hypoimmunogenic hPSCs are successfully specified into hepatocytes with high efficiency resembling those of the WT hPSCs.
ALB synthesis assays were further performed to specifically test the metabolic activities of differentiated hepatocytes. The concentrations of secreted ALB in the supernatants of day 21 hepatocytes cultures from WT, B2Mnull, B2MmHLAG and B2Mm/sHLAG hPSCs were 20.40±0.631 µg/ml, 22.18±2.424 µg/ml, 19.31±0.927 µg/ml, and 20.95±1.763 µg/ml, respectively, with no obvious differences within each group (Fig. 4D). Indocyanine green (ICG) uptake and release assays showed that all three hypoimmunogenic hPSC-derived hepatocytes exhibited clear ICG uptake and release within 6 h similar to the WT control (Fig. 4E). In addition, periodic acid schiff (PAS) staining revealed comparable glycogen storage in all four groups (Fig. 4F). These results indicate that functional hepatocyte-like cells can be derived from hypoimmunogenic hPSCs, and these hypoimmunogenic hPSCs-derived cells have regular metabolic functions, such as macromolecule transportation and glucose metabolism.