Long-term engraftment and maturation of autologous iPSC-derived cardiomyocytes in two rhesus macaques.

CD34+ hematopoietic stem and progenitor cells from two middle-aged RMs, A11E073 (NHP#1) and ZG32 (NHP#2), were reprogrammed into rhesus induced pluripotent stem cells (RhiPSCs) 19 , 27 , 28 . To enable non-invasive, non-toxic repetitive longitudinal PET/CT imaging of the persistence and distribution of RhiPSC-derived cells in vivo , the cDNA of the rhesus NIS gene ( SLC5A5 ) driven by the CAG promoter was knocked-in into the safe harbor AAVS1 locus of RhiPSCs via CRISPR/Cas9-mediated homologous recombination 26 , 29 . RhiPSC clones containing a single NIS cDNA inserted correctly into the AAVS1 locus were isolated for each animal. These clones expressed pluripotency genes and had normal karyotypes ( Figure S1A , B ). Following in vitro cardiac differentiation, the differentiation efficiency and purity of cardiomyocytes were confirmed by immunofluorescence staining and flow cytometric analysis for cardiac lineage markers ( Figure S1C ). Additional flow cytometric analysis was performed to characterize any residual non-cardiac lineage cells. Representative data from final cell products of NHP#2 showed no SSEA4+ residual pluripotent/undifferentiated cells. However, small proportions (1–3.3%) of CD31+, CD44+, CD90+, and/or CD105+ cells were detected, suggesting limited differentiation to other mesenchymal linages such as endothelial cells, mesenchymal stem cells, smooth muscle cells, or fibroblasts under these culture conditions ( Figure S1D ). Generation and comprehensive characterization of the large numbers of RhiPSC-NIS-CMs required for the in vivo studies involved cryopreserving batches of RhiPSC-NIS-CMs at day 10 post-differentiation. Upon thawing, we observed 50–60% viability. Thawed cells exhibited successful recovery after in vitro replating with spontaneous beating ( Video S1 ). Immunofluorescence staining of Ki67 at day 15 post-differentiation (5 days post-thawing) indicated a significant proportion of proliferating cardiomyocytes, with a noticeable decrease at day 30 ( Figure S2A ), suggesting their proliferative capacity at early stage. The majority of RhiPSC-NIS-CMs expressed MLC2a, with very few cells expressing MLC2v at day 15, while a larger fraction gained MLC2v expression by day 30 ( Figure S2B ). All these data demonstrated that the cryopreserved day 10 RhiPSC-NIS-CMs for transplantation were immature, but capable of maturing after long-term culture. The function of NIS in RhiPSC-NIS-CMs was tested via measurement of NIS-mediated in vitro uptake of 18 F-TFB, the PET probe to be used for in vivo imaging 30 . As expected, uptake of 18 F-TFB in RhiPSC-NIS-CMs derived from both animals was NIS mediated ( Figure S2C ). Note the data from NHP#1 iPSCs and CMs had been included in our prior NIS RhiPSC engineering report 26 . Notably, in both animals, there was an increase in 18 F-TFB incorporation as NIS-RhiPSCs differentiate into NIS-RhiPSC-CMs in vitro, with the difference statistically significant in NHP#2 ( Figure S2C ). MI was induced in each animal via 90-minute percutaneous balloon catheter occlusion of the mid left anterior descending (LAD) coronary artery ( Figure 1A ). Location of the occlusion was chosen to result in a large enough infarct to be visualized by cardiac magnetic resonance (CMR) imaging and potentially induce inflammation enhancing any immune response to autologous RhiPSC-NIS-CMs, but also small enough to reasonably ensure survival through long-term follow-up. The animals received prophylactic anti-arrhythmic agents before and during the MI procedure, but did not require treatment post MI. The CMR imaging at late gadolinium enhancement phase demonstrated small, transmural antero-apical infarcts in both animals, with preserved left ventricular ejection fraction (LVEF) ( Figure S3A ). Infarctions were apical-septal in NHP#1 and basal-mid septal in NHP#2. To model a clinically relevant time post-infarction that human patients might receive autologous iPSC-derived cardiomyocytes, autologous RhiPSC-NIS-CMs were delivered to the infarcted area 4 months post-infarction ( Figure 1A ), in contrast to earlier studies using a subacute model, where xenogeneic or allogeneic cells were delivered 2 weeks post-infarction 8 – 10 . A cardiac loop recorder was implanted at the time of cell delivery to monitor arrhythmias for the duration of animal follow-up. No immunosuppressive drugs, including corticosteroids, were administered at any time throughout the study. Pre-transplantation heart rate prior to the study were all within expected range for healthy rhesus macaques. Continuous electrocardiography monitoring via implanted devices revealed rapid engraftment arrhythmias in NHP#1 one week after cell transplantation, with spontaneous resolution by one week later ( Figure S3B ). In NHP#2, engraftment arrhythmias were noted two weeks after transplantation and spontaneously resolved 6 weeks later ( Figure S3B ). In both animals, engraftment arrhythmias were well tolerated and did not require additional treatment or intervention apart from the protocol-specified prophylactic anti-arrhythmic agent. PET/CT imaging with 18 F-TFB was used to monitor engraftment of viable RhiPSC-NIS-CMs in both animals. NIS protein expression results in intracellular uptake of this radiotracer, allowing detection and localization via PET/CT 26 . Maximum intensity projection PET images and PET/CT merged axial sections of the hearts of the macaques were generated at baseline and periodically through end of study ( Figure 1B and C ). In both animals, engrafted NIS-expressing cells were visible in the hearts on scans performed 10- or 14-days post cell injections. It is notable that the imaging was of sufficiently high sensitivity and resolution to clearly distinguish the multiple cell injection sites in each animal. Other than the expected high signal in thyroid and stomach on baseline and subsequent scans resulting from endogenous NIS expression, as well as signal from free tracer being excreted into the urinary tract (kidneys, ureters, and bladder), no ectopic NIS expression suggesting spread of delivered RhiPSC-NIS-derived cells outside the heart was detected in either animal ( Figure S3C ). LVEF did not change appreciably in either animal on repeat CMR imaging more than 6 months post cell injection (57% versus 51% in NHP#1 and 71% versus 69% in NHP#2, Figure S3A ) and neither animal had structural changes on imaging consistent with tumor formation. Quantitation of acquired PET images indicated significant increase in the signal strength and total uptake of 18 F-TFB to the grafts in the initial weeks post-delivery, likely reflecting proliferation and hypertrophy of the RhiPSC-NIS-CMs, and then reached a stable plateau within 2–4 months ( Figure 2A and B ; Figure S3D ). This pattern was more clearly observed in animal NHP#2 that underwent more frequent and longer follow-up PET imaging. Planned imaging between 2 and 6 months in NHP#1 was interrupted by COVID-19 pandemic restrictions, and prematurely terminated at 8 months due to complications from endometriosis requiring euthanasia, a known clinical syndrome in middle-aged female RMs unrelated to myocardial infarction or cell therapy. 18 F-TFB signal strength and uptakes were lower in NHP#1 compared to NHP#2, but they demonstrated a similar pattern in their relative changes after normalization to maximal values (right panels in Figure 2A , B ). This suggests that a smaller number of cells engrafted in NHP#1 than NHP#2, but that autologous RhiPSC-NIS-CMs showed similar dynamics in both recipients. Taken together, imaging of NIS activity by 18 F-TFB PET showed that grafts are detectable within 2 weeks of delivery, likely undergo maturation and expansion for the following 2–4 months, and then persist stably for 8–12 months. NHP#1 and NHP#2 underwent full necropsy after being euthanized at 8- and 14-months post cell injection, respectively. No macroscopic tumor formation or other abnormalities were detected in either animal, other than extensive endometriosis that had been suspected clinically and radiographically in NHP#1. The excised hearts were fixed, sliced, and embedded in paraffin for sectioning ( Figure S4A , B ). No abnormal microscopic lesions of undifferentiated cells or other signs of neoplasia were found in tissues and organs screened histologically, including the hearts, lungs, kidneys, and livers from both animals. As commercial anti-NIS antibodies only bind epitopes maintained in frozen but not fixed tissue, we designed a NIS RNA probe and performed RNAscope to identify engrafted RhiPSC-NIS-CMs derived NIS expressing cells. In the five slices of NHP#1’s ventricular areas, several small collections of NIS+ cells were identified by RNAscope ( Figure 3A , Figure S4C ). NHP#2 showed much larger collections of NIS+ cells, with RhiPSC-NIS-CM grafts detected in 4 out of the 8 total slices, primarily in the antero-lateral left ventricle ( Figure 3B , Figure S4D ). We used histomorphometry to estimate the total number of engrafted cells in each of the hearts (see Methods for details ). This yielded estimates of 4.2 million long-term engrafted cells in NHP#1 and 9.7 million cells in NHP#2. The 2.3-fold larger graft size by histology in NHP#2 compared to NHP#1 corroborates the 4-fold higher NIS SUV signals obtained by PET-CT ( Fig. 2A ). This low engraftment efficiency is in line with what has been reported in previous pre-clinical studies using PSC-derived cardiomyocytes 9 as well as dopaminergic neurons in large animal models 31 , 32 . The cardiac marker α-actinin was expressed in the vast majority of NIS+ cells, confirming the cells were cardiomyocytes. The graft cells had the characteristic rod shape of adult cardiomyocytes, suggesting substantial hypertrophic growth and maturation had taken place over the 8–14 months following delivery ( Figure 3A , B , C ). Grafts were located both in the infarct and peri-infarct zones, as anticipated based on delivery to multiple injection sites in the general area of the infarctions. In some regions, especially in NHP#1, the peri-infarct grafts were separated from host tissue by thin bands of collagen, while in other areas, had nearly seamless integration of the endogenous and engrafted cells (middle panels in Figure 3A , B ; Figure S4C , D ). Staining for the Z-disc protein α-actinin revealed well myofibril alignment and sarcomere organization in engrafted NIS+ cardiomyocytes in both NHP#1 and NHP#2, with more normal organization in NHP#2 ( Figure 3C right panels). The measurement of cell diameter ( Figure 3D ) and sarcomere length ( Figure 3E ) revealed that in NHP#1, cell diameter (10.89 µm) and sarcomere length (1.79 µm) of graft cardiomyocytes were shorter than those of the host (11.88 µm and 2.01 µm, respectively), whereas there was no significant difference in NHP#2 (diameter: 11.16 µm vs 11.33 µm, sarcomere length: 2.09 µm vs 2.05 µm, respectively). To investigate the proliferative capability of transplanted cardiomyocytes after 8 months and 14 months engraftment, we conducted Ki67/MLC2v antibody staining along with NIS RNAscope in 5 slides per animal across the sections. The data revealed that very rare engrafted cardiomyocytes were still proliferating (0.28% in NHP#1 and 0.03% in NHP#2, respectively). Representative fields with Ki67+ engrafted cardiomyocytes in both NHP#1 and NHP#2 heart sections were shown in Figure S5A . Co-detection of myosin light chain 2 isoforms (MLC2a and MLC2v) through immunostaining and NIS RNAscope indicated that all the engrafted cardiomyocytes expressed MLC2v, with no retained expression of MLC2a that was observed at the early stages before injection ( Figure S5B ). Connexin 43 (Cx43) is the primary component of gap junctions mediating electrical coupling between mature ventricular cardiomyocytes 33 , 34 . To evaluate the structural maturity of the transplanted cardiomyocytes, we co-stained for α-actinin and Cx43. In both NHP#1 and NHP#2, Cx43 was expressed and located at the cell-cell junctions between engrafted NIS+ cardiomyocytes. In the NHP#1 graft analyzed at 8 months post-injection, Cx43 staining was punctate and circumferential, whereas in NHP#2 graft analyzed 6 months later following cell delivery, Cx43 had polarized to the intercalated disk ( Figure 4A ). The intercellular adhesion protein N-Cadherin (N-cad) 35 is shown together with the cardiac marker myosin light chain 2v (MLC2v) in Figure 4B . N-cad was more concentrated at intercalated disks in the 8-month graft than was Cx43, and there was further restriction of this protein to the intercalated disk in the 14-month graft. Grafts expressed MLC2v uniformly at both time points. Although the majority of the engrafted NIS+ cardiomyocytes in NHP#1 were separated from the host tissue by collagen encapsulation, in a few areas we observed Cx43 and N-cad localization between engrafted and host cardiomyocytes (white arrows in top panels of Figure 4A , B ). In NHP#2, the integration of engrafted with host cardiomyocytes was much more frequent (white arrows in the bottom panels of Figure 4A , B ), providing structural evidence for electromechanical coupling between host and graft cardiomyocytes. Two isoforms of troponin I (TnI) are expressed at different stages during heart development and cardiomyocyte differentiation 36 . The slow-skeletal troponin I (ssTnI) isoform is expressed in immature cardiomyocytes but decreases during maturation. Conversely, expression of the cTnI isoform increases with cardiomyocyte maturation 9 , 37 . In NHP#1, the engrafted NIS+ cardiomyocytes still expressed ssTnI at 8 months post cell injection, and the intensity of cTnI expression was weaker than that of host cardiomyocytes ( Figure 5A top panels). In NHP#2 at 14 months post cell injection, only a few ssTnI-positive cardiomyocytes could be identified in the engrafted NIS+ regions (white arrows in bottom panels Figure 5A ), and there was no difference in the intensity of cTnI expression between engrafted NIS+ and host cardiomyocytes ( Figure 5A bottom panels). These findings suggest that maturation of engrafted cardiomyocytes in NHP#2 at 14 months post injection approached that of endogenous cardiomyocytes, while this process was still incomplete at 8 months post injection in NHP#1. To investigate vascularization in the engrafted areas, we performed RNAscope with a CD31 probe. Although we did not detect large vessels coursing through engrafted tissue areas, small CD31+ endothelial cells forming microvessels were observed within engrafted tissue areas in both animals ( Figure 5B ). Concurrent CD31 and NIS RNAscope indicated that the vast majority of CD31+ endothelial cells were NIS-negative, suggesting that these endothelial cells originated from the host tissue via blood vessel network extension rather than from engraftment of any endothelial cells derived from RhiPSCs during in vitro cardiac differentiation, consistent with previous findings 8 . To compare the immune response between autologous and allogeneic cardiomyocytes in the same model system, the same number of allogeneic RhiPSC-CMs (derived from a male iPSC line ZH26; not expressing NIS) were transplanted into 2 MHC-mismatched animals. The first recipient (Allo-NHP#1; A17010, male) underwent infarction induction identical to the autologous recipients, and the second recipient (Allo-NHP#2; A15050, female) was non-infarcted to assess whether the infarct milieu altered the allogeneic immune response, and to avoid any chance of post-MI mortality, given profound macaque shortages post-pandemic. The match to donor ratios were Allo-NHP#1: MHC-I 0%, MHC-II 10% and Allo-NHP#2: MHC-1 0%, MHC-II 0%, respectively. No immunosuppression was given to the allogeneic cell recipients. The two allogeneic recipients were followed for 8 weeks post cell injection. There were no arrhythmias noted on weekly EKGs. Planned euthanasia and heart processing was performed at 8 weeks post cell injection, based on prior macaque allogeneic studies predicting that active immune rejection would be present by that time point. Prior studies transplanting xenogeneic human embryonic stem cell (ESC) or mismatched allogeneic NHP iPSC-CMs in NHP MI models have required continuous high dose multi-drug immunosuppression including glucocorticosteroids, calcineurin inhibition and CTLA4 blockade to prevent or reduce rejection of transplanted cells 9 , 10 . To understand the immune response to allogeneic cardiomyocytes without immunosuppression, both allogeneic animals (Allo-NHP#1 and Allo-NHP#2) were euthanized at a pre-specified endpoint of 8 weeks post-transplantation, chosen based on pilot studies to fall within a window of predicted ongoing rejection. Since the injected ZH26 RhiPSC-CMs could not be distinguished from recipient cardiomyocytes via NIS expression, ssTnI antibody staining was used to distinguish immature engrafted cardiomyocytes (ssTnI positive) from mature host cardiomyocytes (ssTnI negative) given the early time point of the analysis. In Allo-NHP#1, all ssTnI+ graft regions were undergoing severe rejection (International Society for Heart and Lung Transplantation grade 3R) with intense infiltration by CD3+ T lymphocytes and CD20+ B lymphocytes ( Figure 6A ). In Allo-NHP#2, the rejection was so profound that no residual ssTnI+ engrafted cardiomyocytes could be detected 8 weeks after injection. However, dense, organized infiltrates of CD3+ T cells and CD20+ B cells were found in multiple foci within the injected areas ( Figure 6B ). Organized T or B cell infiltrates are not found in normal myocardium, strongly suggesting fulminant cellular rejection and clearance of the allogeneic cardiomyocytes. In contrast to the allogeneic RhiPSC-CM recipients, the hearts of NHP#1 and NHP#2 receiving autologous RhiPSC-NIS-CM grafts showed no immune cell infiltrates at 8 and 14 months respectively, with only widely scattered T and B cells in NIS+ engrafted areas, at no higher frequency than in surrounding host myocardium ( Figure 6C , D and Figure S6A ). These data demonstrate that RhiPSC-NIS-CMs derived, cultured, and differentiated in vitro did not evoke detectable inflammatory or immune responses in vivo , even in the absence of immunosuppression.

NHPs are a valuable preclinical model to investigate the potential of regenerative therapies, due to their physiological, immunological, and developmental similarities to humans. Previous studies have reported the successful transplantation of xenogeneic human ESC-CMs into the infarcted myocardium of NHPs, or allogeneic transplantation of iPSC-CM into MHC-matched cynomolgus monkeys, resulting in remuscularization and recovery of cardiac function 8 – 10 . A recent report also demonstrated that co-transplantation of human iPSC-CMs and iPSC-ECs (endothelial cells) into NHPs promoted maturity of CMs and improved cardiac function after infarction 11 . However, these studies required continuous intensive immunosuppression to prevent rejection of engrafted foreign cells, and this regimen could not be maintained further than 3–4 months, precluding long-term follow-up of safety and efficacy. Importantly, immunosuppression poses a significant barrier to clinical translation of this regenerative technology 2 . Another recent study reported that rhesus macaque islets, derived from engineered immune shielded ( B2M −/− CIITA −/− CD47 + ) iPSCs, achieved long-term survival over 40 weeks in allogeneic rhesus macaque recipient without immunosuppression 38 . These results, together with our evidence for lack of inherent immunogenicity of autologous iPSC-CMs, suggests that either autologous or potentially immune-shielded allogeneic iPSC-CM transplantation remains a potential long-term strategy for cellular therapy, as demonstrated in various preclinical and clinical studies 39 – 42 , despite formidable technical and logistical challenges with present day technologies. To the best of our knowledge, autologous transplantation of iPSC-CMs has not been evaluated in NHPs or human patients. This study is the first reporting on the safety, long-term engraftment, and characterization of autologous iPSC-CMs in NHPs, using a chronic MI model without immunosuppression. Severe chronic heart failure is a highly morbid condition with increasing prevalence in the United States and worldwide. Orthotopic heart transplantation remains the only curative therapy, but with donor organs available to less than 1% of the heart failure population, this disease is essentially incurable. Cell replacement therapy to regenerate dysfunctional or absent heart muscle is a promising strategy to prevent or even reverse chronic cardiomyopathy. Following MI, the damaged area enters an initial inflammatory phase within hours, followed by proliferative and maturation phases, resulting in wound healing and eventual fibrotic scar formation by several weeks or months 43 . It is currently difficult to predict which patients will progress into ischemic heart failure until several months post-infarction 44 , and this makes selection of appropriate candidates a tricky endeavor in the early phase of disease. While iPSC-CMs have demonstrated the ability to engraft and restore the function of the recently infarcted NHP heart when delivered sub-acutely, two weeks following MI 8 – 10 , there have been limited preclinical studies investigating whether cells can engraft and provide benefit in hearts with more chronic cardiomyopathy occurring following fibrosis and remodeling. Our previous work in small animal models suggested that injection of human ESC-CMs was not as effective in improving heart function or reversing adverse remodeling in chronic MI models compared to subacute transplantation, suggesting that chronic injury-related organized fibrous scar tissue with low vascularization may be a more hostile environment for cell engraftment 45 , 46 . Our current data demonstrate that engraftment and integration of iPSC-CMs in a chronic infarct region can occur in NHPs. Previous cell transplantation studies in NHPs have utilized the GCaMP or green fluorescent protein transgenes to label and study excitation coupling of implanted PSC-CMs 8 – 10 . However, these xenogeneic proteins would be immunogenic in autologous iPSC-CMs delivered to immunocompetent NHPs, and therefore would require immunosuppression following transplantation 47 . In addition, this type of labeling cannot be assessed in vivo non-invasively or longitudinally, requiring euthanasia for immunohistochemical assay. For these reasons, we selected the nonimmunogenic endogenous NIS protein to track iPSC-CMs non-invasively and with high sensitivity via repeated PET/CT imaging. In a previous pilot study of RhiPSC-NIS-CMs injected into hearts of NSG mice, as few as 1 × 10 5 cells were detectable via PET/CT imaging using a small animal scanner, and could be tracked and quantified longitudinally 26 . In the current study, we used a clinical PET/CT scanner equipped with an enhanced reconstruction algorithm for higher sensitivity visualization, and we were pleased to obtain images as early as 10- or 14-days post-cell injection showing multiple distinct sites of 18 F-TFB accumulation in the areas of the LV corresponding to individual injection locations. The 18 F-TFB signal strength as well as total uptake increased through 2–4 months post-engraftment and plateaued thereafter in both animals, suggesting the increased intake of the tracer into engrafted RhiPSC-NIS-CMs. We hypothesize the increased NIS expression is due to proliferation of immature RhiPSC-NIS-CMs immediately following transplantation followed by graft hypertrophy during in vivo cardiomyocyte maturation 9 . We observed an increase in 18 F-TFB incorporation as the cells differentiate towards cardiomyocytes in vitro in the current as well as previous 26 studies. Hypertrophy and maturation of rat and human iPSC-CM grafts in infarcted rat hearts were observed over 3-months follow-up in our previous study 48 . In the current study, RhiPSC-NIS-CMs cultured in vitro robustly expressed Ki67 at earlier time points, with a drop over time, indicating decline of proliferative capacity during maturation. At necropsy, we found only extremely rare Ki67+ graft cardiomyocytes. Graft cells were larger and expressed more mature marker patterns at 14 months versus 8 months, comparing the two animals. Altogether, these data support our hypothesis that the initial 18 F-TFB increase in signal over time was likely due to graft proliferation and maturation. In addition, it is possible that the low vascularity of the infarcted area, and thus poor 18 F-TFB delivery to the delivered cells at the earlier time points have contributed to lower tracer uptake, with improved vascularization over time. Most importantly, no teratomas or other abnormal cell collections were detected macroscopically or microscopically at the time of necropsy in the heart or other tissues. We did not detect 18 F-TFB signal outside the heart in unexpected tissues or organs on repeated longitudinal PET imaging, which is considered far more sensitive than other imaging modalities or microscopic examination 26 . The PET tracer 18 F-TFB has been used safely in clinical trials 30 . We believe this non-invasive PET/CT imaging platform could be useful in further preclinical and clinical studies to accurately evaluate novel strategies for enhancing cell delivery and engraftment and assess critical challenges for iPSC-based regenerative approaches 49 , 50 . Previous large animal studies have only evaluated the maturation process of transplanted cardiomyocytes for a few weeks to at most several months following delivery, in part due to the feasibility and toxicity of the intensive immunosuppression required to prevent rejection of xenogeneic or allogeneic cells. In our current study, two rhesus macaques receiving autologous grafts were followed up to 8 and 14 months, the longest follow-up reported for iPSC-derived CMs in vivo . In both cases, we observed maturation of the transplanted cardiomyocytes, with alignment of sarcomeres, maturation-related isoform switching of troponin I, as well as expression and localization of adhesion and gap junction proteins to the intercalated disk, required for electrical integration of the graft tissue with the host myocardium. This process appeared virtually complete in NHP#2 by 14 months, with loss of fetal ssTnI and comparable levels of adult cTnI expression to adjacent host cardiomyocytes. In vivo maturation of PSC-derived cardiomyocyte to this degree has, to our knowledge, not previously been achieved in any large animal model, including NHPs, and help predict the length of time expected for full maturation and graft stability in humans, although larger studies are clearly desirable in the future. Although the two autologous cell recipients received the same number of injected viable cells, the total tracer uptake in NHP#2 was ~4-fold larger than in NHP#1 by PET-CT scanning and 2.3-fold larger by histomorphometry. Cardiomyocyte graft size is known to be variable with current technologies, and similar 4–6-fold variation was seen in our previous studies involving human cardiomyocyte transplantation into infarcted macaque hearts 8 , 9 . The basis for this variability is only partly understood, including short-term retention (leakage from injection site, loss to venous circulation) and long-term cell survival and proliferation. Survival is influenced by input cell health (determined by harvest, cryopreservation, dose preparation), pro-survival interventions (heat shock or biochemical interventions to antagonize graft death), and host factors (inflammation, vascularization, etc.). The cells were extensively characterized prior to transplantation in both animals, and input populations appeared qualitatively and quantitatively similar, suggesting downstream factors may be responsible for the difference in graft size. In NHP#1, we observed two graft tissues of significant size, both encased by collagen bands that acted as barriers, separating them from the host tissues. This structural feature might constrain the available space for engrafted cardiomyocytes to grow further. The observed shorter cell diameter and sarcomere length in these grafts could potentially be attributed to the limited space for optimal growth. Importantly, both grafts exhibited similar kinetics in the NIS signal increase from baseline over time. This suggests that despite the lower number of engrafted cells in NHP#1, the growth or maturation kinetics of the graft in NHP#1 was comparable to NHP#2. While we are unable to make any conclusions about the relative arrhythmogenicity of autologous versus allogeneic iPSC-CM transplantation, both autologous animals exhibited transient engraftment arrhythmias, consistent with prior studies in xenogeneic or allogeneic settings 7 – 10 , 51 , 52 . This suggests that transient engraftment arrhythmias are caused by the automaticity of electrically immature PSC-CMs as they integrate with host cells, not immune-mediated inflammation 51 . Interestingly, the duration of arrhythmia for NHP#2 was four weeks, compared to just one week for NHP#1. The greater arrhythmia burden may be due to dose-effect of the larger graft size in NHP#2. Immune-mediated rejection of allogeneic or xenogeneic cells has been well documented in both NHPs and human patients 39 – 41 , 53 , 54 , and we confirm here that allogeneic cardiomyocytes are rejected in NHPs. Although allogeneic transplantation provides the opportunity for off-the-shelf products, the long-term requirement of immunosuppression in this setting increases the risk of a number of complications including infection, organ failure and tumor formation. This prompted us to test whether autologous transplantation would provide a path to heart regeneration without immunosuppression. Our current results demonstrate that PSC-derived autologous cardiomyocyte grafts showed no evidence of inherent immunogenicity. In addition to utilizing autologous iPSCs, cells were grown under “xeno-free” conditions using chemically defined culture media and Synthemax II coated plates, and the marker NIS gene was rhesus macaque in origin. Some but not all early murine studies had suggested that expression of fetal antigens by PSC-derived cells might be rejected as foreign by post-natal animals, even in a syngeneic setting 13 , 15 . Progress has been made in creating MHC-I/II deleted, engineered MHC–matched iPSCs, and/or other genetically-engineered immune “stealth” iPSCs for production of universal “off-the-shelf” regenerative cell therapies. However, some immunosuppression may still be required to prevent immune rejection, due to minor histocompatibility antigens, mutations acquired in cell culture, or other mechanisms 10 , 55 – 58 . In addition, we cannot exclude the possibility that neoantigen-expressing RhiPSC-NIS-CMs were under immune attack early after delivery, with clonal selection for non-immunogenic RhiPSC-NIS-CMs, resulting in graft survival and lack of signs of rejection by end of study. The only available clonal tracking study of human PSC-MCs in infarcted rat hearts documented an average of over 5 cells per clone at 6 weeks, confirming the ability of at least some cells to proliferate in vivo 59 . In summary, this study provides the first proof-of-concept in a relevant large animal model that, in the absence of immunosuppression, autologous iPSC-CMs can engraft, fully mature, and integrate into chronically infarcted myocardium on long-term follow-up of over one year, although limited in the sample size. These data should encourage further studies and development of immune-evasive regenerative therapies approaching autologous immunoreactivity for both heart disease and other target organ applications. Facilitation of more robust cell retention and engraftment will be required to move iPSC-CM therapies forward clinically, potentially via stimulation of vascularization 11 , inclusion of supportive scaffold materials, or other bioengineering approaches 60 , which can be evaluated in large animal models facilitated by NIS or other approaches for longitudinal imaging. This proof-of-concept study of autologous cardiac cell therapy was designed to demonstrate the long-term survival of autologous iPSC-CM transplantation using a non-invasive serial PET/CT imaging platform. The autologous study design was central to questions being examined, however, feasibility of enrolling large numbers of animals during the COVID-19 pandemic, requiring years of housing for each study animal during iPSC generation/cardiomyocyte differentiation and longitudinal follow-up, precluded entry of additional animals in this study. Given the absolute necessity of each animal surviving post-infarction in order to receive autologous iPSC-CM grafts, we purposefully created small, sub-clinical infarcts unlikely to result in immediate animal death or heart failure. This precluded analysis of any impact on cardiac function. In addition, the dose of iPSC-CMs delivered in this study was lower than the dose of allogeneic iPSC-CMs shown to improve function. Further studies would be needed to investigate whether functional improvement long-term could be achieved in the setting of larger infarcts in a chronic NHP model of ischemic cardiomyopathy. Finally, the small sample size precludes rigorous assessment of safety concerns such as arrhythmias. While engraftment arrythmias were observed in both autologous animals with timing and severity consistent with previous published data following allogeneic and xenogeneic PSC-CM delivery, conclusions regarding likelihood and severity cannot be made. Although our data suggest that autologous iPSC-CM did not evoke detectable inflammatory or immune responses in vivo , we cannot completely rule out a transient immune response or the selection of non-immunogenic/non-neoantigen expressing cells.

Ischemic heart disease is the leading causes of death worldwide 1 . Much of this morbidity and mortality is attributed to the heart’s limited regenerative ability. When the heart muscle is damaged due to myocardial infarction (MI), the damaged tissue is replaced with scar tissue, which eventually impacts contractility and leads to heart failure 2 . Despite advances in medical and device therapies for acute and subacute MI, there are no curative treatments for chronic heart failure except for orthotopic heart transplantation 2 . Pluripotent stem cell-derived cardiomyocytes (PSC-CMs) represent a promising cellular therapeutic option to induce remuscularization of the infarcted heart. Significant progress has been made to demonstrate efficacy of these cells in the context of acute or subacute MI injury using xeno- or allografted animal models 3 – 11 . However, graft survival and maturation kinetics in clinically relevant immunocompetent large animal models are poorly understood. The immunogenicity risk of autologous PSC-CMs, having undergone in vitro reprogramming and differentiation, is controversial 12 . Some studies have raised concerns about neoepitopes arising from early developmental genes or mutations arising during reprogramming/cell culture 13 – 15 , whereas others suggest that autologous or syngeneic cells are well tolerated 16 – 20 . Furthermore, long-term safety of these PSC-CMs is still largely unknown because the intensive immunosuppression required in these models is poorly tolerated long-term, limiting follow-up to 3–4 months in previous nonhuman primate (NHP) studies. In addition, transient arrhythmias are observed in pigs and NHPs following transplantation of PSC-CMs, and remain a serious concern for clinical translation, and it is not known whether rejection or immunosuppression toxicities contribute to these arrhythmias 21 , 22 . Therefore, evaluating arrhythmia risk in an autologous setting that allows long-term monitoring is warranted prior to clinical applications. The sodium/iodide symporter (NIS) is a transmembrane glycoprotein normally expressed at high levels in the thyroid, salivary glands, and stomach, but with limited expression in other tissues 23 . NIS transports iodide and several other anions into cells using the transmembrane sodium gradient, and this property is the basis of evaluating thyroid disease by scintigraphy or single-photon emission tomography 24 . As NIS is a nonimmunogenic endogenous protein, it has been repurposed as a reporter to track NIS-expressing cells via radionuclide imaging in immunocompetent animal models. The recently developed tracer fluorine-18 tetrafluoroborate ( 18 F-TFB) enables high sensitivity imaging of NIS-expressing cells via positron emission tomography (PET) 24 , 25 . In a proof-of-principle study in a murine cardiac infarction model, we have demonstrated that NIS expressing xeno-grafted iPSC-CMs could be successfully monitored by PET/computed tomography (CT) longitudinally following serial 18 F-TFB administrations for up to 10 weeks 26 . In this study, we transplanted induced pluripotent stem cell-derived NIS-labeled cardiomyocytes (iPSC-NIS-CMs) 4 months post MI induction in autologous rhesus macaques (RMs), a model with comparable cardiac and immunophysiology to humans, allowing long-term tracking and assessment of safety and graft survival. In addition, the same number of allogeneic RhiPSC-CMs were transplanted into major histocompatibility complex (MHC)-mismatched animals in the absence of immunosuppression to compare the immune response between autologous and allogeneic cardiomyocyte transplantations. We show autologous grafts could be efficiently monitored via non-invasive serial PET/CT imaging and demonstrated long-term engraftment and graft maturation up to 14 months, while allogeneic iPSC-CMs, without immunosuppression, were rejected within 8 weeks of transplantation. This study provides valuable insights towards clinical translation of PSC-based cardiac cell therapies.

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Cynthia E. Dunbar ( [email protected] ). Rhesus iPSC lines generated in this study are available upon MTA. Data reported in this paper will be shared by the lead contact upon request. This paper does not report original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request. The rhesus macaques used in this study were housed and handled in accordance with protocols approved by the Animal Care and Use Committees of the National Heart, Lung, and Blood Institute and the University of Washington. In the autologous transplantation study, NHP#1 was an 8-year-old female (9.4 kg) and NHP#2 was a 12-year-old male (8.9 kg) at the time of the transplant. In the allogeneic transplantation study, Allo-NHP#1 was a 7-years old, 16.6 kg male and Allo-NHP#2 was a 9 years old, 9 kg female. Due to humane considerations, NHP#1 had to be euthanized and tissues were collected at 8 months post-cell delivery due to development of refractory severe endometriosis resulting in ureteral obstruction, hydronephrosis and renal failure, a relatively common disease in middle-aged female macaques 61 . NHP#2 was followed for 14 months following RhiPSC-NIS-CM injection before planned study termination and tissue collection. Both allogeneic animals were euthanized at a pre-specified endpoint of 8 weeks post-transplantation for histology analysis. Rhesus CD34+ hematopoietic stem and progenitor cells were isolated from bone marrow and reprogrammed into RhiPSCs using the Cytotune ® 2.0 Sendai virus vectors encoding hOCT3/4, hSOX2, hKLF4 and hc-MYC, as previously described 29 . The cDNA of the rhesus NIS gene ( SLC5A5 ) driven by the CAG promoter was knocked-in to the safe harbor AAVS1 locus of NHP#1 and NHP#2 RhiPSCs via CRISPR/Cas9-mediated homologous recombination, as previously described 27 , 62 . RhiPSCs were maintained on Matrigel (Corning, 354230)-coated plates under one of two different culture conditions: #1) mouse embryonic fibroblast-conditioned medium (MEF-CM) with 20 ng/mL fibroblast growth factor 2 (FGF2) cultured under hypoxic (5% O 2 ) condition. The cells were grown in colonies and manually passaged every 2–4 days with the media changed every day 28 , 29 . #2) PluriSTEM medium (Millipore Sigma) and normoxic (20% O 2 ) condition. The cells were grown in uniformly spread monolayers and passaged via 0.5 mM ethylenediaminetetraacetic acid (EDTA) dissociation every 2–3 days. Manually picked RhiPSC colonies or EDTA dissociated RhiPSCs were cryopreserved in CryoStor CS10 medium (STEMCELL Technologies) and stored in liquid nitrogen. Frozen RhiPSCs from condition #1 were thawed and cultured in MEF-CM with 20 ng/ml FGF2 on pre-seeded MEF feeders, then manually passaged and placed on Matrigel-coated plates for routine culturing. Frozen RhiPSCs from condition #2 were thawed directly in PluriSTEM medium on Matrigel-coated plates. The RevitaCell Supplement (Thermo Fisher Scientific) was added to medium for first day of culture after thawing but removed on the next day. See the Supplementary materials and methods for details related to the maintenance of RhiPSCs used in allogeneic transplants. The process of cardiac differentiation for autologous transplantation was performed on plates coated with Synthemax II-SC (Corning), a fully chemically defined, xeno-free substrate, in order to avoid potential immune rejection of foreign antigens after transplantation, as Matrigel is derived from murine extracellular matrix 63 . Two different protocols were used to differentiate RhiPSCs into cardiomyocytes, based on whether the RhiPSCs had been cultured in conditions #1 or #2 above. Cardiomyocyte protocol #1 (CM-#1) was used for condition #1 RhiPSCs. Differentiation was induced by serially treating the cells with growth factors activin A + bone morphogenic protein 4 (BMP4) + FGF2 and a Wnt inhibitor (IWP2) as described in detail previously 28 . Cardiomyocyte protocol #2 (CM-#2) was used for condition #2 RhiPSCs. The cells were dissociated into single cells or small clumps with 0.5 mM EDTA at 70–80% confluence and seeded on Synthemax-coated plates in PluriSTEM medium at a 1:4 dilution (approximately 50,000 cells/cm 2 ) with PluriSTEM medium changed daily. Two days later, when the cells reached 50–70% confluence, differentiation was started (day 0) by changing the medium to cardiac differentiation basal medium (CDBM consisting of DMEM/F-12, 64 µg/ml L-ascorbic acid, 13.6 µg/L sodium selenium, transferrin 10 µg/mL, 1x chemically defined lipid concentrate, and 1x penicillin-streptomycin) with 5 mM CHIR99021 and 15 ng/ml Activin A added until day 1, then changed to CDBM plus 3 µM IWP2 with daily changes until day 7. After day 7, conditions were changed to maintenance medium (CDBM plus 20 µg/mL insulin), changed every other day. Both differentiation protocols generated cardiomyocytes initiating spontaneous beating at day 7–8, with comparable marker expression and viability, before and after thawing. By day 10, the RhiPSC-CMs could be dissociated by TrypLE Express (Thermo Fisher Scientific) and cryopreserved in CryoStor CS10 for future transplantation or continually cultured for further characterization. RhiPSCs (ZH26) were thawed and expanded in PluriSTEM on LN521-coated plasticware. The dissociated monolayer was seeded into PluriSTEM media supplemented with 0.5% bovine serum albumin (BSA, Thermo Fisher Scientific) in PBS-Mini Vertical Wheel Bioreactor (PBS Biotech) for 3 days with daily media changes. The following day, cardiomyocyte differentiation was induced on Day 0 using 10ng/mL activin A (R&D Systems) and 5μM CHIR (Tocris) in RPMI (Thermo Fisher Scientific) supplemented with 0.25% BSA and B27 Supplement Minus Insulin (Thermo Fisher Scientific). On Day 1, media was exchanged to RPMI supplemented with 0.25% BSA and B27 Supplement Minus Insulin containing 1ng/mL bone morphogenetic protein 4 (BMP4, R&D Systems). On Day 3, media was exchanged to RPMI with B27 Supplement Minus Insulin (Thermo Fisher Scientific) containing 5μM IWP4 (Tocris). On Day 5 and 7, media was exchanged to RPMI with B27 Supplement (Thermo Fisher Scientific). On Day 10, selection was initiated with glucose-free RPMI (Thermo Fisher Scientific) supplemented with 4mM L-Lactate (Sigma-Aldrich). After 4 days in selection media, media was exchanged to RPMI supplemented with B27 Supplement and exchanged every other day until harvest. On Day 17, RhiPSC-CMs were harvested with 1x TrypLE (Thermo Fisher Scientific) in Versene and cryopreserved in CryoStor10 using a controlled-rate liquid nitrogen freezer. Thawed cells were more than 85% viable using an NC200/Via1-cassette (Chemometec) and expressed more than 90% cTnT (Cardiac Troponin T Antibody, anti-human/mouse/rat, clone REA400, Miltenyi). Prior to the MI procedure, autologous animals (NHP#1 and NHP#2) underwent coronary CT angiography using intravenous contrast medium (Isovue-370, Bracco Diagnostics) injection to map the anatomy of the coronary arteries. Except for Allo-NHP#2, all macaques received MI induction procedures before cardiomyocyte transplantation. Closed-chest ischemia-reperfusion injury for MI induction was performed percutaneously as previously described 8 , 9 with minor modifications to increase survivability. Briefly, autologous animals were pre-treated with amiodarone (5 mg/kg/day) and atenolol (1 mg/kg/day) seven days prior to MI. During the MI induction procedure, the animals received intravenous lidocaine 0.7–1.4 mg/kg and amiodarone 2–5 mg/kg. Heparin was delivered intravenously to maintain activated clotting times of 250–350s to prevent thrombus formation. The mid-distal portion of the left anterior descending artery (LAD) was occluded via 90 minutes of balloon inflation, followed by deflation and reperfusion. The animals did not receive further anti-arrhythmic agents following the MI procedure. 17–18 weeks after infarction induction, viable RhiPSC-NIS-CMs were injected into the infarct region via thoracotomy and direct visualization of the target infarct zone as mapped by CMR imaging. Cryopreserved autologous RhiPSC-NIS-CMs were thawed, washed, and the total cell dose of 200 million viable cells for each animal was split into 5 aliquots of 200–250 µl and 5 discrete injections were performed in the target area with placement of an epicardial horizontal mattress suture for retention. A cardiac loop recorder from Medtronic (Reveal LINQ) was implanted subcutaneously at the time of cell delivery for assessment of arrhythmia burden in the autologous recipients. In addition, 12-lead electrocardiograms were performed each time an animal was sedated for PET/CT or MRI procedures. In allogeneic recipients, 12-lead electrocardiograms were performed weekly. No immunosuppressive drugs, including glucocorticosteroids, were administered to autologous or allogeneic recipient animals at any time throughout the study. Approximately three hours before transplantation, multiple vials of 1mL cryopreserved cardiomyocytes were thawed in a 37°C water bath. Phenol-red free RPMI supplemented with B27 Supplement and ≥200 Kunitz Units/mL DNase I (Millipore) was added to the cell suspension to dilute the cryopreservation media. Subsequent wash steps using phenol-red free RPMI basal media were performed in progressively smaller volumes to concentrate the cell suspension. The cell suspension was centrifuged such that sufficient volume of supernatant can be removed in order to achieve a target cell density for injection (2 × 10 8 viable cells in 0.6 – 0.75mL phenol-red free RPMI basal media). The final cell suspension volume was determined from cell counts sampled after the initial wash to remove cryopreservation media. Baseline CMR imaging with late gadolinium enhancement was performed post-infarction and pre cell injection to determine the location of the infarction as well as to measure infarct scar volume and ventricular ejection fraction. Additional CMR imaging was performed following cell injection (up to 29 weeks in NHP#1 and 48 weeks in NHP#2) to assess changes in myocardial structure, monitor for tumor formation, and measure cardiac function. A Siemens 3T Skyra MRI scanner was used for all NHP#1 scans and the baseline MRI for NHP#2. An Aera 1.5T MRI scanner was used for the post-cell injection MRI on NHP#2, due to sudden lack of availability of the original machine. In each animal, 0.2 mmol/kg gadolinium contrast (Gadovist, Bayer) was administered intravenously. The images were analyzed using suiteHEART v5.04 (NeoSoft, Pewaukee, WI). 18 F-TFB was synthesized in-house as previously described 26 . Each autologous recipient underwent baseline 18 F-TFB PET/CT imaging 2 weeks prior to RhiPSC-NIS-CM injection, and then periodically from 2 weeks to 25 weeks after RhiPSC-NIS-CM injection in NHP#1 and from 10 days to 52 weeks in NHP#2. PET/CT images were acquired on a GE Discovery MI DR PET/CT system 1 hour after intravenous injection of 21.4 ± 2.3 kBq 18 F-TFB at 2.4 ± 0.3 kBq/kg. The torso was scanned with 4 bed positions at 5 minutes per bed position. Acquired PET images were reconstructed using GE’s Q CLEAR algorithm. Maximum intensity projection PET images and PET/CT merged images were generated using MIM v7.1.5 (MIM Software). To quantitate the PET images at each time point, a volume of interest (VOI) was drawn delineating the area of SUV higher than 3.5 (SUV>3.5) associated with the engrafted RhiPSC-NIS-CMs in the LV wall to apex and a VOI of similar size in the corresponding LV wall to apex on the baseline image to estimate the background activity. SUV cutoff value of 3.5 for the RhiPSC-NIS-CM VOI was selected to be slightly higher than the maximum SUVs of the background and LV lumen that ranged SUV 2.5–3.3 to eliminate the signal ambiguity caused by motion artifacts. The 67% quantile SUV of the RhiPSC-NIS-CM VOI was first calculated and compared to that of control VOI drawn on the normal lateral LV wall. The areas demonstrating SUV above the 67% quantile SUV are considered to represent the locations of RhiPSC-NIS-CMs, and the average SUV (SUVmean) of the areas reflects the signal strength of RhiPSC-NIS-CMs on the images. The total uptake to the RhiPSC-NIS-CMs in the VOI was calculated by subtracting the background total signal obtained from the corresponding area on the baseline pre-injection image as follows. Total uptake = (SUVmean of SUV>3.5 area – SUVmean of background) x VOI volume (mL), where SUVmean = [activity in the VOI (kBq) / VOI volume (mL)] / [injected dose (kBq) / body weight (g)]. Relative changes of SUVmean of area above the 67% quantile SUV and those of total uptake were calculated for each animal, considering the values at 6 months in NHP#1 and 12 months in NHP#2 as 100%, respectively. To aid generation of tissue sections in the same angle as PET/CT images, a cardiac mold for NHP#2 was created which had the shape of the heart extracted from contours drawn on the CT scans of the heart. The region of interest contour outlining the heart wall was drawn on each CT slice using the MIM software and saved as JPEG images, which were imported into Solidworks software (Dassault Systemes Soldworks Corporation, MA) where a 3D volume was created by layering 3.27mm slabs, with the JPEG CT image fixed to the surface of each slab. The outline of the contour was used to form a closed boundary from which an extruded cut was performed on the slab. Layering each slab then creates a 3D volume containing a cutout in the shape of the heart. Once the 3D volume with the cutout was formed, a parallel set of rectangles, 1.4 mm wide, in 5 mm increments were used to create evenly spaced inserts to guide tissue slicing. The volume was then cut in half exposing the heart mold. The model was 3D printed using a Formlabs (Boston, MA) Form 3+ SLA printer. The heart harvested from NHP#1 was kept in cold PBS and sliced freshly by hand into 5 1-cm slices and fixed with 10% buffered formalin. The heart of NHP#2 was fixed with 10% buffered formalin and then sliced into 8 slices with a 3D printed mold at 5 mm thickness. The mold was created based on the CT images of the NHP#2 heart ( Supplementary Methods ) to generate tissue sections at the same angle as PET/CT images. Heart tissue slices were embedded in paraffin. Serial sections (5 μm) were cut for standard hematoxylin and eosin staining, Masson trichome staining, immunohistochemistry, and RNAscope. The whole slide images were scanned with an NDP digital slide scanner and viewed with NDP Viewer 2.0 software (Hamamatsu, Japan). Graft cardiomyocyte number was estimated from photomicrographs of sections stained with H&E and NIS reporter gene as described 9 . Briefly, NIS-positive regions in adjacent sections were photographed with a 20X objective. A defined-area grid was overlaid onto the images, and 5–6 fields of NIS-positive cell nuclei were counted for each heart and converted into the number of nuclei per mm 2 . We mathematically calculated nuclei/mm 3 as (nuclei/mm 2 ) 1 . 5 . Graft area on a particular slide was determined in whole-slide scanned H&E images. Graft volume was calculated as [(Σslice graft area) x (slice thickness)]. The number of graft cells for each heart was calculated as [(graft volume) x (graft nuclei/mm 3 )]. Assumptions for these calculations are 1) that the cells are isotropically packed; and 2) that each cardiomyocyte is mononucleated. We note that tissue sampling evolved from NHP#1 to NHP#2, including fixation prior to sectioning and the use of a mold to facilitate uniform sectioning. There is thus greater precision in the estimate of graft cell number in NHP#2. Staining of RhiPSC colonies and RhiPSC-derived cardiomyocytes for relevant markers were performed as previously described 28 , using the antibodies listed in STAR Methods . RNAscope/immunofluorescence co-detection was performed as per Advanced Cell Diagnostics (ACD)’s instructions using the RNAscope v2 fluorescent multiplex assay combined with the immunofluorescence kit (ACD, 323110). Briefly, paraffin-embedded slides were baked at 65 °C for 1 hour, followed by deparaffinization and hydrogen peroxide quenching. The slides were boiled in Co-Detection target retrieval reagents (ACD, 323165) for 15 minutes, followed by overnight incubation with primary antibodies at 4 °C. The following day, the slides were washed and fixed with 10% neutral buffered formalin at room temperature for 30 minutes, followed by protease incubation at 40 °C for 30 minutes. The slides were then hybridized with the RNA probe for 2 hours at 40 °C. The hybridization signal was amplified by RNAscope multiplex v2 amplify solutions and detected by horseradish peroxidase (HRP) labeling and Opal 520, 570, or 650 fluorophores (AKOYA Biosciences, Marlborough, MA) 1:3000 diluted in tyramide signal amplification buffer. After HRP blocking, the slides were incubated with the fluorophore-conjugated secondary antibodies for 1 hour and then counterstained with 4′,6-diamidino-2-phenylindole (DAPI, 300 nM) before mounting with Aqua-PolyMount (Polysciences) for confocal imaging. The NHP positive probe Mfa-Polr2a-C1 (Advanced Cell Diagnostics, 320901), and a negative probe for Bacillus subtilis (Advanced Cell Diagnostics, 320871) were used as positive and negative RNAscope controls respectively. The following RNA probes from Advanced Cell Diagnostics were used for RNAscope: rhesus NIS (Mmu-SLC5A5-C1, 1063321-C1) and rhesus CD31 (Mmu-PECAM1-C3, 499151-C3). Phase contrast and fluorescence images were captured via Zeiss Axio Observer A1 microscope using ZEN 2.3 pro software. Whole slide immunofluorescence images were obtained by NanoZoomer XR slide scanner system (Hamamatsu, Japan), stored as ndpi format, and analyzed using the NDP Viewer 2.0 software. Confocal fluorescence images were captured via Zeiss 880 confocal microscope using ZEN black 2.3 software or via Leica TCS SP8 confocal microscope using LASX software. All data are shown as mean ± SEM. Student’s t tests and Tukey multiple comparisons test were utilized for Figure 3 and Figure S2 , respectively. Values of p < 0.05 were considered statistically significant.

Supplemental video 1: Spontaneous beating of NHP#1 RhiPSC-NIS-CMs in culture, related to Figure 1 .