Hypoimmune stem cells and islets: hype or a true breakthrough in diabetes treatment?

Abstract

Immune-resistant pancreatic islets hold great promise for advancing diabetes cell therapy. Two key approaches, hypoimmunogenic pluripotent stem cells (PSCs) and hypoimmunogenic cadaveric islets, aim to overcome immune rejection in islet transplantation. Human PSCs provide a versatile source of insulin-producing cells, but immune rejection remains a major barrier. Recent advances in gene-editing technologies have enabled the modification of PSCs and cadaveric islets to reduce their immunogenicity. These cells can be engineered to express human leukocyte antigen (HLA)-negative profiles, while overexpressing immunoregulatory factors such as CD47, PD-L1, and HLA-G to evade T cell and natural killer (NK) cell immune-mediated responses. These modifications aim to generate “off-the-shelf” islet cell therapies compatible with a wide range of patients, potentially eliminating the need for immunosuppressants. However, ensuring long-term safety and functionality remains a challenge. Potential risks such as immune escape, viral infections, and tumorigenicity must be carefully addressed through additional safety measures. This review explores different approaches for generating hypoimmunogenic islets, recent advances in overcoming immune rejection, and key hurdles that need to be addressed for widespread clinical use for patients with diabetes. It also compares the potential benefits and limitations of hypoimmunogenic cadaveric islets versus hPSC-derived islets, providing insights into their future clinical applications.

Graphical abstract

Introduction

Diabetes mellitus (DM) is a chronic metabolic disorder characterized by hyperglycemia, resulting from defects in insulin secretion, insulin action, or both [1]. DM is a major public health issue and is one of the four priority noncommunicable diseases identified for global action by world leaders [2]. The number of diabetes cases in 2021 was 536.6 million people worldwide (10.5%), a number projected to rise to 783.2 million (12.2%) in 2045.

The types of diabetes include type 1 diabetes (T1D), type 2 diabetes (T2D), and monogenic diabetes (MD). T1D is an autoimmune disease where the immune system erroneously targets and destroys insulin-producing beta cells in the pancreas, leading to insulin deficiency. Typically, T1D manifests in childhood or adolescence, although it can develop at any age. Patients require lifelong insulin therapy to manage blood glucose levels. Without proper management, T1D can lead to complications such as diabetic ketoacidosis, cardiovascular diseases, neuropathy, retinopathy, and nephropathy [3]. T2D, the most common type, is characterized by insulin resistance in insulin target tissues and pancreatic beta cell dysfunction. Although it commonly develops in adults, it is increasingly seen in younger populations. Management of T2D includes lifestyle changes, oral hypoglycemic medications, and insulin therapy. Poorly managed T2D can also lead to severe complications similar to those seen in T1D [4]. MD, a rare type of diabetes (1–2% of all diabetes cases), results from inherited or spontaneous mutations in a single gene that regulates pancreas development and/or beta cell function [5]. Subtypes of MD include neonatal diabetes mellitus (NDM) typically diagnosed within the first 6 months of life, maturity-onset diabetes of the young (MODY) usually identified before the age of 25 years, and certain syndromic forms of diabetes.

Conventional diabetes treatments focus on managing symptoms but do not address the underlying issue of beta-cell deficiency. Cell therapy represents a promising approach to treat diabetes, particularly T1D, by restoring the body’s ability to produce insulin through transplanting insulin-producing cells, such as cadaveric pancreatic islets or stem cell (SC)-derived beta cells [6]. Cadaveric islet transplantation is limited by donor availability and the need for immunosuppressive therapy [7, 8]. Human pluripotent stem cells (hPSCs), including embryonic stem cells (hESCs) and induced pluripotent stem cells (iPSCs), can be differentiated into insulin-producing islets (SC-islets) (reviewed in [5, 9, 10]). These SC-islets are a scalable alternative to cadaveric sources.

This review discusses the potential of hypoimmunogenic SC-islets and cadaveric islets in revolutionizing cell therapy for diabetes. It addresses the challenges faced by both sources in terms of immunogenicity, scalability, and long-term function and explores different strategies to protect transplanted islets from immune attack. The review also discusses the limitations and hurdles associated with each approach and outlines future directions for research and clinical application.

Mechanisms of immune recognition, rejection, and regulation

Understanding the mechanisms of immune recognition and rejection is essential for developing strategies to mitigate these responses and improve the success of cell therapies. One mechanism of immune recognition involves antigen presentation of major histocompatibility complex (MHC) molecules (also known as human leukocyte antigen [HLA] in humans) to T cells. The HLA gene cluster contains genes that encode for HLA class I (HLA-A, HLA-B, and HLA-C) and HLA class II (HLA-DR, HLA-DP, and HLA-DQ) molecules [11]. HLA genes are highly polymorphic, with approximately 40,000 different alleles reported, resulting in a vast diversity of HLA molecules expressed in a population [12]. This diversity is further enhanced by the HLA genes’ codominant expression, where both maternal and paternal alleles are simultaneously expressed [13]. Immune rejection of a transplant occurs when T cells recognize foreign HLA molecules on transplanted cells, leading to an immune attack [11].

HLA class I molecules are expressed on all nucleated cells and are recognized by CD8⁺ cytotoxic T cells to enable the elimination of cells presenting cancerous or pathogenic viral antigens [14]. HLA class I molecules are downregulated in certain tumors and virus-infected cells to avoid recognition by CD8⁺ T cells [15]. In such cases, the innate immune system’s natural killer (NK) cells attack to eliminate cells with abnormal low or absent HLA class I expression [15]. The presentation of HLA class I molecules is mediated by β2 microglobulin (B2M) and transporter associated with antigen processing 1/2 (TAP1/2) [16]. B2M is required to form HLA class I heterodimers on the cell surface [14]. TAP1 and TAP2 play a role in loading endogenous peptides onto HLA class I molecules in the endoplasmic reticulum, facilitating the presentation of endogenous peptides on the cell surface [17]. HLA class II molecules are typically expressed only on the surface of professional antigen-presenting cells (APCs), primarily dendritic cells, macrophages, and B cells, but can also be expressed on other cell types under certain conditions [11, 18]. Pancreatic beta cells express HLA class II molecules upon stimulation with inflammatory cytokines, such as interferon gamma (IFN-γ) in T1D [19]. HLA class II molecules are recognized by CD4⁺ T helper cells that, upon recognition of foreign antigens, release cytokines to activate CD8⁺ T cells, B cells, dendritic cells, NK cells, and macrophages [20, 21]. CD4⁺ regulatory T cells (Tregs) suppress CD4⁺ T helper cells to maintain peripheral tolerance and self/non-self-discrimination, which are required to prevent autoimmunity and reduce allergy [22]. The expression of HLA class II molecules is controlled by RFXANK, RFX5, RFXAP, and the class II transactivator (CIITA) [23, 24]. RFXANK, RFX5, and RFXAP are components of a complex that binds to HLA-II promoters, enabling the recruitment of CIITA, a master transcriptional activator that then initiates the expression of the HLA class II genes: HLA-DR, HLA-DP, and HLA-DQ.

Given their essential role, some cells are granted immune privilege status by the immune system, such as in the eye, testis, and placenta, allowing them to remain unharmed [25]. Cells in these sites typically employ strategies such as downregulating HLA class I and II molecules to evade immune detection, secreting immunoregulatory cytokines to suppress immune activation, and expressing surface molecules to induce apoptosis in infiltrating immune cells. For instance, corneal endothelial cells in the eye lack HLA class I expression [26], and they do not express HLA class II molecules, limiting direct antigen presentation to the immune system. In the human placenta, trophoblast cells avoid recognition by the maternal immune system by not expressing classical HLA molecules [27]. To evade NK cell-mediated cytotoxicity, these cells express non-classical HLA molecules, such as HLA-G and HLA-E, which bind receptors on NK cells, thereby inhibiting their activity.

In addition to these mechanisms, immunoregulatory and anti-inflammatory factors play a role in enforcing immune regulation. For example, Tregs exert their immunosuppressive effects through the production of immunoregulatory cytokines, such as interleukin-10 (IL-10) and transforming growth factor-β (TGF-β) [28]. Similarly, dendritic cells near the immune-privileged placenta express high levels of anti-IL-10 and TGF-β to tolerate the presence of paternal and fetal tissue [29].

The expression of resident surface regulatory factors, such as the Fas ligand (FasL) and programmed death-ligand 1 (PD-L1), may also help in maintaining immune privilege status by inducing apoptosis in invading T cells through their interaction with the Fas and programmed death 1 (PD1) receptors, respectively. FasL is constitutively expressed in immune-privileged regions such as the placenta [30], the testis [31, 32], and the eye [33], while PD-L1 expression is expressed in immunosuppressive mesenchymal stem cells (MSCs) [34] and is also linked to immune evasion in tumors [35].

Overcoming immune rejection: strategies to protect transplanted islets from immune attack

The immune system, while critical for defending against pathogens and abnormal cells, presents significant challenges in cell therapy, particularly when transplanting allogeneic cells, such as cadaveric or SC-derived islets for diabetes treatment. Although administering immunosuppressive agents can improve the long-term survival of transplanted cells, they carry risks such as autoreactivity, infection, and malignancy [36]. Approaches to address these challenges include modifying cadaveric donor-derived islets to make them hypoimmunogenic or to use encapsulation strategies for protection. However, this requires a large number of donors. Alternatively, stem cells can provide an unlimited source of insulin-producing cells, but they also come with technical challenges that need to be addressed. To improve the success of cell therapy for diabetes, several strategies have been developed to protect transplanted cells from immune attack as summarized in the sections below.

Stem cells and cadaveric donors as cell sources for islet transplantation

There are two main sources for generating hypoimmunogenic islets: cadaveric donor islets and those derived from hPSCs. Recent studies highlight the potential of primary hypoimmunogenic islets to achieve insulin independence in allogeneic diabetic recipients without the need for immunosuppression [37]. In one study involving four non-human primate (NHP) donors, genetically engineered hypoimmunogenic islets (HIP) have successfully been transplanted into an allogeneic diabetic recipient, achieving long-term insulin independence without immunosuppression [37]. In a clinical trial to establish safety and function, a first-in-human study presented by the same team examined a low dose of cadaveric donor-derived HYP islets (UP421) [38, 39]. HIP islets transplanted intramuscularly into the forearm of a patient with T1D demonstrated survival and insulin production for up to 6 months post transplantation, indicated by detectable circulating C-peptide, a marker for insulin production, which increased in response to a meal. Magnetic resonance imaging (MRI) confirmed graft survival, providing proof-of-concept that hypoimmunogenic islets can evade immune recognition and function without immunosuppression, though further studies are needed to assess long-term glycemic benefits.

While these findings are promising, they require further validation in larger and more diverse cohorts. The study not only highlights the potential of donor-derived islets but also underscores the challenge of donor availability. Furthermore, the strategy employed to establish hypoimmunogenic cells involves extensive genetic modification and, while effective, the potential off-target effects in primary cells must be carefully managed [40–42]. Donor-derived hypoimmunogenic islets may face variability in quality and availability, compromising reproducibility—a critical factor in the clinical application of pancreatic islet transplantation [43]. In contrast, because hPSC-derived hypoimmunogenic cells can be grown indefinitely, screened, and stored in culture, they offer a more controlled and scalable production process, ensuring consistent quality and renewable supply for transplantation [44]. These advantages make stem cells a compelling option as a cell source for widespread clinical use.

The promise, progress, and challenges of stem cell-derived islets for diabetes treatment

Multiple studies have demonstrated that hPSCs can be differentiated into mature, functional beta cells in vitro, significantly enhancing glycemic control in diabetic models. In vitro differentiation protocols utilize a combination of small molecules and growth factors to regulate specific pathways at each stage of the differentiation process [5, 45]. It is established that pancreatic beta cells arise from pancreatic progenitors that coexpress PDX1 and NKX6.1. Recent protocols have been shown to produce approximately 80%–90% PDX1⁺/NKX6.1⁺ pancreatic progenitors [46–49], demonstrating the effectiveness and reliability of these differentiation methods for generating pancreatic progenitors. The generation of beta cells from stem cells has recently demonstrated improved efficiency and functionality, with around 30–60% of functional insulin-secreting cells coexpressing C-peptide and NKX6.1 [50–56]. However, the process of differentiating hPSCs into beta cells exhibits variability across different cell lines, particularly during the later stages of pancreatic beta cell development. This variability is influenced by the inherent differences among hPSC lines and the robustness of the differentiation protocols employed. While earlier studies indicated that immature beta cells lacked dynamic insulin secretion and matured only after transplantation into animal models [50, 51], more recent studies have shown significant advancements in the in vitro differentiation of hPSCs into functional pancreatic beta cells [52–54, 56–59]. This ability to produce functional insulin-secreting cells in a controlled environment is a significant step toward achieving insulin independence in patients with diabetes.

Thus, the transplantation of hPSC-derived islets represents a groundbreaking advancement in diabetes treatment. However, it faces two significant challenges: immunogenicity toward the transplanted islets, and tumorigenicity of residual undifferentiated cells. Vertex Pharmaceuticals developed fully differentiated insulin-producing cells (Zimislecel) from hPSCs and transplanted them into the hepatic portal vein in patients with T1D (currently in phase III of a phase I/II/III clinical trial: NCT04786262). The liver is the preferred site for clinical islet transplantation owing to its high vascularization, minimal invasiveness, and low complication rates [60]. In total, 12 patients were dosed with a single infusion of Zimislecel, and all of them demonstrated islet cell engraftment and glucose-responsive insulin secretion by day 90. Recently, presented results are promising, with all 12 patients achieving the recommended targets of an HbA1C below 7.0% and time-in-range above 70% on continuous glucose monitoring. Overall, 11 out of the 12 participants have met the secondary endpoint of reduced or eliminated need for insulin administration. Further trials with more patients are ongoing. This treatment, however, requires chronic immunosuppression, posing a trade-off between the administration of insulin and immunosuppressive agents.

The applicability of autologous SC-based cell therapies to mitigate the need for immunosuppression

One of the promising strategies for diabetes treatment to mitigate the need for immunosuppression is to transplant patient-specific, autologous SC-islets [61]. Recent successes have been reported by two independent groups in China, where they succeeded in transplanting SC-islets derived from the diabetic patients’ own iPSCs. A 59-year-old patient with T2D experienced significant improvements in glycemic control during the first 27 months post-intrahepatic implantation of islet tissue differentiated in vitro [62]. This report marks the first evidence that iPSC-derived islets can restore islet function in late-stage T2D [62]. However, this approach cannot be applied to autoimmune conditions such as T1D. In a separate study, a patient with T1D received autologous iPSC-derived islets implanted under the abdominal anterior rectus sheath and achieved insulin-independence starting on day 75, with follow-up extending up to 1 year [63]. However, this patient was on immunosuppressants owing to a previous liver transplantation, and although demonstrating potential for the abdominal transplantation site, the study provides limited insight on potential autoimmune response. Another limitation of employing autologous iPSCs is that the process of generating pancreatic islets for each patient may be challenging to reproduce on a large scale. In addition to variable efficiency, each distinct cell line would generate a unique therapeutic cell product, necessitating separate development processes and regulatory approvals. Furthermore, even autologous iPSCs are not completely free of immunogenicity, as genetic and epigenetic variations that may occur during reprogramming can lead to immunogenic responses in some iPSC lines [64, 65].

HLA-based stem cell banking

An alternative approach to generating a unique iPSC line for every patient is to generate an iPSC bank from donors with the most common HLA types that can cover the highest percentage of a population [66]. This approach can support cell therapies with reduced immune rejection and offers coverage for a large portion of the population with only a limited number of iPSC lines. Previous studies on cord blood and kidney grafting suggest that matching HLA-A, HLA-B, and HLA-DR loci is enough for long-term graft survival, without the need for immunosuppressive drugs [67, 68]. Multiple government-funded country-based HLA-homozygous iPSC banks have been or are currently being established in Japan [69], Korea [70], Spain [71], Germany [72], France [73], Norway [74], Australia [75], the USA [76], Saudi Arabia [77], and Taiwan [78].

The generation of haplotype-matched iPSC lines faces challenges such as having to screen a large number of potential donors and covering the full genetic diversity of a population. However, smaller populations or those with limited genetic diversity might simplify the generation of haplotype-matched lines. In Japan, a country characterized by low genetic diversity, 27 cell lines from seven donors have been generated to represent the four most common haplotypes, covering 40% of the Japanese population [69]. By 2022, these lines had been used in over ten clinical trials [79], demonstrating the practical potential of this approach in clinical settings. It has been estimated that around 50% of the Japanese population can be matched using 10 HLA-homozygous donors and 90% using 140 HLA-homozygous donors [80]. Similarly, analysis of UK solid organ donor HLA typing data by Taylor et al. revealed that the top 150 HLA-A, HLA-B, and HLA-DR homozygous combinations could cover 93% of UK patients [81]. In Korea, data from 4205 cord blood donations have been used to develop ten HLA-homozygous iPSC lines to cover 41.1% of the population [71]. Norway, with a small population of 5.26 million (83.2% ethnic Norwegians), selected nine donors representing common haplotypes to cover 70% of the ethnic Norwegian population by screening the HLA types of 4514 donors [75]. In Spain, a country with a greater genetic diversity, analyzed 30,000 high-resolution HLA-typed bone marrow donors and generated seven iPSC lines to cover 21% of the population [72].

In addition to public banks, at least two private companies are offering clinical-grade iPSC lines for cell therapy development [82]. Fujifilm Cellular Dynamics Inc. established two Good Manufacturing Practice (GMP)-compliant iPSC lines that are homozygous for HLA haplotypes, covering 19% of the US population [82], and are available for commercial use in clinical cell therapies. Similarly, CATALENT Biologics offers a GMP-compliant iPSC bank, covering a significant portion of HLA haplotypes in the Caucasian population [83].

Encapsulation techniques for transplanted islets

Encapsulation of the transplanted islets in a biocompatible material can shield them from the immune system while allowing nutrient and insulin exchange [84, 85]. However, immune responses to biomaterials and scaffold degradation products remain a concern [86–89]. For example, ViaCyte has developed an immunoprotective device loaded with pancreatic progenitor cells derived from hESCs (VC-01). Subcutaneous transplantation in patients with T1D showed inconsistent results, likely due to immunogenicity to the encapsulation device, and poor vascularization of this transplantation site, leading to inefficient blood sugar regulation and hypoxia [90, 91]. To improve outcomes, a modified device, VC-02, has been created with larger pores to allow better oxygenation and metabolic exchange, though it requires immunosuppression owing to lack of immune protection. Recent trials have showed that some patients have had detectable C-peptide responses, but insulin independence has not been achieved, likely owing to insufficient engrafted cells and fibrous tissue deposition [92]. Vertex Pharmaceuticals have also initiated a second phase I/II clinical trial (NCT05791201) with encapsulated insulin-producing cells (VX-264) in an effort to address the risks of immunosuppression with their first clinical trial. Although results from the first phase demonstrated safety, the increases in C-peptide were not sufficient [93]. These devices require further refinement to sustain long-term permeability for nutrients, oxygen, and insulin, while effectively isolating the encapsulated cells from the host’s immune system.

Promoting localized immune tolerance

MSCs possess immunomodulatory and regenerative properties, making them a promising tool for cell therapy applications [94]. Some strategies address immunogenicity by harnessing the immunomodulatory features and the hypoimmunogenic nature of these cells. The immune-evasive nature of MSCs is attributed to the low expression of HLA class I and lack of class II antigens [95, 96]. Furthermore, MSCs have the ability to home to injured tissues and secrete growth factors, cytokines, and extracellular vesicles that modulate the immune system, accelerating repair processes [97, 98]. Upon activation by proinflammatory cytokines, MSCs further exert their immunomodulatory effects through soluble factors and direct cell-to-cell interactions [99, 100]. Studies have demonstrated the use of MSCs in reducing beta cell immunogenicity in patients with T1D, as evaluated in several clinical trials [101–105]. These promising properties highlight the therapeutic potential of MSCs in diabetes management. However, the application of MSCs in cell therapy must be treated with caution owing to their possible tumorigenic properties [106–110].

A recent study demonstrated a novel approach to overcoming graft immune rejection by co-engineering hPSCs and Tregs [111]. By engineering hPSCs to express a truncated epidermal growth factor receptor (EGFRt) and generating chimeric antigen receptor (CAR)-Tregs targeting EGFRt, researchers achieved localized immune protection. This strategy effectively suppressed immune responses and protected SC-pancreatic beta-like cell grafts in vivo, providing proof of concept for combining hPSC and Treg engineering to enhance transplantation outcomes. These findings underscore the promise of innovative genetic and culture techniques in overcoming immune challenges and improving the efficacy of cell therapies.

Approaches for genetically engineering universal hypoimmunogenic islets

Recent studies have focused on generating universally compatible hypoimmunogenic islets by silencing or deleting HLA genes or genes crucial for HLA expression and function, and by expressing genes encoding immune-modulatory molecules [40, 112–118]. Hypoimmunogenic islets are genetically engineered to evade the immune response, making them a promising solution for diabetes treatment through transplantation. The ability of these islets to achieve insulin independence without requiring immunosuppression could significantly improve the quality of life for patients. In this section, we outline several key approaches and their outcomes in the context of reducing immunogenicity to facilitate transplantation without the need for long-term immunosuppression (Table 1; Fig. 1).

Table 1.

Genetic engineering strategies for developing hypoimmunogenic cells

Cell type	Genetic engineering strategy	Outcome		Reference
		In vitro	In vivo
CD8+ T cell evasion
hESCs	HLA-I KD by siRNA and blocking by intrabodies	Reduced T cell activity in PBMCs in vitro, and in vivo following transplantation into the hind limbs of immunocompetent mice Prolonged xenogeneic survival for up to 21 days in five and up to 42 days in two out of seven recipients	–	[135]
hiPSCs	TAP2 KO by ZFNs	Reduced CD8+ T cell activity toward derived myeloid precursor and myeloid cells in a co-cultured allo-reactive CD8+ T cell line	–	[17]
hESCs	B2M KO by AAV targeted delivery	Reduced T cell activity in co-culture assays with PBMCs	–	[113]
hESCs	B2M KO by TALENs	Reduced T cell activity in co-culture assays with PBMCs	Reduced T and NK cell infiltration 2 days post transplantation into tibialis anterior muscles of mice	[112]
hESCs	B2M KO by CRISPR/Cas9	–	Derived islets normalize hyperglycemia and delayed rejection post transplantation under the kidney capsule of humanized diabetic mice	[133]
CD4+ T cell evasion
hiPSCs	B2M and CIITA KO by CRISPR/Cas9	Diminished T cell activation by derived cardiomyocytes in co-culture assays with PBMC	–	[138]
hiPSCs	HLA-DR KO by TALENs	Reduced CD4+ T cell proliferation co-cultured with derived dendritic cells	–	[143]
hiPSCs	HLA-A, HLA-B, and HLA-DR KO by CRISPR/Cas9 HLA-C retention	Reduced CD4+ proliferation and NK cell activation	–	[118]
NK cell evasion
hESCs	HLA-G OE by piggyBac transposon	Reduced T cell proliferation and NK lysis of hESCs and derived epidermal progenitors	–	[120]
hESCs	B2M KO HLA-G1 knockin at B2M locus	Reduced T cell cytotoxicity and NK cell activation and cytotoxicity toward derived cardiomyocytes	Reduced NK cell infiltration 20 days following subcutaneous transplantation in humanized mice	[116]
hPSCs	B2M KO HLA-E AAV-mediated knockin at B2M locus	Evasion of CD8+ T cells diminished NK lysis of derived CD45+ hematopoietic cells	Enhanced survival of derived CD45+ hematopoietic cells over 3 days in immunodeficient mice injected with NK cells	[132]
hiPSCs	HLA-A, HLA-B, and CIITA KO by CRISPR/Cas9 HLA-C retention	Evasion of CD8+ T cell cytotoxicity and NK lysis of derived CD43+ blood cells and cardiomyocytes Evasion of CD4+ T cells recognition of CIITA-deficient derived CD43+ blood cells	Enhanced survival of derived CD43+ blood cells over 7 days in immunodeficient mice supplemented with either reactive T or NK cells	[114]
hiPSCs and miPSCs	B2M and CIITA KO by CRISPR/Cas9 CD47 OE by LV	Evasion of NK cell activation and killing of PSCs cells and derived endothelial and cardiomyocytes	Long-term teratoma survival for up to 50 days, diminished immune cell infiltration, and NK clearance in allogeneic hosts	[40]
Immune tolerance
hESCs	CTLA4-Ig, PD-L1 OE via HR at HPRT locus by BAC	–	Increased teratoma size and diminished CD4+ and CD8+ T cell infiltration 6 weeks post transplantation of modified hESCs and derived fibroblasts and cardiomyocytes in humanized mice Increased infiltration by Tregs that secrete immunosuppressive cytokines and promote localized immune protection to co-transplanted grafts	[126]
hESCs	IL-2 (N88D), IL-10, TGF-β OE	Prolonged survival of derived beta cells for up to 9 weeks post transplantation under the kidney capsule of autoimmune diabetic mice and normalized hyperglycemia	–	[133]
mESCs and hESCs	PD-L1, CD200, CD47, HLA-G/H2-M3, FASL, SERPINB9, CCL21, MFGE8 OE by piggyBac and Sleeping Beauty transposons HSV-TK safety switch	Diminished activation of PBMCs, monocytes, dendritic cells, and NK cells toward hESC-derived retinal pigment epithelium cells	Prolonged survival of mESCs for up to 17 days and teratomas for up to 9 months following subcutaneous injection into the neck or flank in different allogeneic hosts Ectopically transplanted cloaked mESCs provided localized immune tolerance to co-transplanted xenogeneic hESCs for up to 8 months and allogeneic islets for up to 4 months	[127]
Comprehensive approaches
hPSCs	HLA-A, HLA-B, HLA-C, and CIITA KO using CRISPR/Cas9 HLA-G, PD-L1, and CD47 OE by knockin at the AAVS1 safe harbor locus	Evasion from T cells, NK cells, and macrophage recognition	Prolonged teratoma survival in immunodeficient mice sensitized with allogeneic CD8+ T cells	[115]
Rhesus macaque iPSCs and hiPSCs	B2M and CIITA KO by CRISPR/Cas9 CD47 OE by LV	Diminished PBMC and T cell immunogenicity against iPSCs for up to 16 weeks	Diminished immune cell infiltration in grafts 16 weeks post transplantation Differentiated islets survive and maintain normal glucose levels in diabetic humanized mice for up to 30 days	[122]
Primary rhesus macaque islet cells	B2M and CIITA KO by CRISPR/Cas9 CD47 OE by LV	–	Modified islets are able to survive and maintain insulin independence in a diabetic allogeneic recipient for up to 6 months	[37]
hiPSCs	HLA-A, HLA-B, HLA-C, and RFXANK KO using CRISPR/Cas9 HLA-G, B2M, PD-L1, PD-L2, and RapaCasp9 OE by piggyBac transposon	Diminished T cell immunogenicity and NK cell lysis and macrophage phagocytosis	RapaCasp9 induced apoptosis to implants in immunodeficient mice following a rapamycin injection	[144]

hESCs: human embryonic stem cells; HLA-I: Human Leukocyte Antigen class I; KD: knock down; siRNA: small interfering RNA; PBMCs: peripheral blood mononuclear cells; hiPSCs; human induced pluripotent stem cells; KO: knock out; ZFNs: zinc-finger nucleases; AAV: adeno-associated viral vector; TALENs: transcription activator-like effector nucleases; NK: natural killer; OE: overexpression; LV: lentiviral vector; miPSCs: mouse induced pluripotent stem cells; PSCs: pluripotent stem cells; Tregs: regulatory T cells; HR: homologous recombination; BAC: bacterial artificial chromosome; mESCs: mouse embryonic stem cells; hPSCs: human pluripotent stem cells.

Fig. 1.

Strategies for genetically engineering universal hypoimmunogenic pluripotent stem cells. CD8⁺ T cell evasion: genes targeted to promote evasion from cytotoxic CD8⁺ T cell recognition include β2 microglobulin (β2M, necessary for HLA class I complex formation), HLA class I expressing genes (HLA-A, HLA-B, and HLA-C), and transporter associated with antigen processing (TAP2, crucial for loading antigens onto HLA class I molecules in the endoplasmic reticulum). CD4⁺ T cell evasion: genes targeted to promote evasion from helper CD4⁺ T cell recognition include HLA class II expressing genes (HLA-DR), class II transactivator (CIITA, transactivator for HLA class II expression), and RFXANK (recruits CIITA on HLA class II promoter). Natural killer (NK) cell evasion: retention of HLA-C, or overexpression of non-classical HLA class I molecules (HLA-E or HLA-G) can inhibit recognition of HLA class I deficient cells by NK cell subpopulations that express their respective killer-cell immunoglobulin-like (KIR) receptors. CD47 transmembrane protein overexpression interacts with signal regulatory protein alpha (SIRPα) on NK cells and macrophages, sending an inhibitory signal that protects cells lacking self-HLA class I molecules from innate immune attack. Immune tolerance: overexpressing surface regulatory factors such as programmed death-ligand 1 (PD-L1), and Fas ligand (FasL) can inhibit immune responses by inducing apoptosis in T cells. CTLA4-Ig can also promote immune tolerance by interacting with immune cell receptors. Secreting cytokines and anti-inflammatory factors such as CCL21, interleukin-10 (IL-10), interleukin-2 (IL-2) mutein (N88D), and transforming growth factor-β (TGF-β) can suppress local immune activation and improve tolerance to transplants. Safety switch: “suicide” genes can trigger apoptosis in unwanted cells when activated by a specific drug. The herpes simplex virus thymidine kinase (HSV-TK) phosphorylates ganciclovir (GCV), creating a toxic compound that incorporates into DNA and induces cell death. Another system, RapaCasp9, combines FRB and FKBP domains with the catalytic domain of caspase 9. Administering rapamycin binds to the FRB and FKBP domains, causing dimerization of caspase 9 and triggering apoptosis. The figure was created using Biorender

Methods to validate the hypoimmunogenicity of genetically modified cells prior to transplantation

Various methods to evaluate the hypoimmunogenicity of genetically modified hPSCs have been developed (Fig. 2). Initial in vitro assessments to confirm the removal of HLA molecules and to measure changes in gene expression and protein surface markers are paramount. Further in vitro assays involve testing the immune evasion capabilities of the modified stem cells using co-culture systems with immune cells to measure T cell, NK cell, dendritic cell, and macrophage immunogenic activities [112–115, 118–120]. Peripheral blood mononuclear cells (PBMCs) or isolated immune cells (such as CD8⁺ or CD4⁺ T cells) are often used in mixed leukocyte reactions (MLRs) to measure alloimmunogenicity. Typically, PBMCs contain 70–90% lymphocytes (of which, 70–85% CD3⁺ T cells, 50–10% B cells, and 5–20% NK cells), 10–20% monocytes, and 1–2% dendritic cells, making them a good in vitro model for measuring immunogenicity [121]. CD3⁺ T cells are composed of CD4⁺ and CD8⁺ T cells in a 2:1 ratio [121]. As a read-out, the respective immune cells’ cytokine production or proliferation rates are analyzed after co-culturing with the donor cells. In vivo assessment methods involve transplanting the modified cells into animal models to assess their survival and potential to evade immune recognition. To replicate the human immune microenvironment, humanized mouse (hu-mouse) models, such as hu-PBL, hu-SRC, NSG-SGM3, and hu-BLT mice, are employed [40, 122]. Several hu-mouse models are generated through reconstituting immunodeficient mice (such as NSG and NOG) with human PBMCs (hu-PBL) or HSCs (hu-SRC) and supplementation with human cytokines, such as interleukin-2 (IL-2) [123], to promote NK cell activation [124, 125]. However, hu-mice still do not fully reproduce the human immune system, and testing in NHPs that are phylogenetically closer to human beings is necessary for clinical translation. In addition, the engineered cells are re-harvested and examined histologically to assess immune cell infiltration, specifically looking for the presence of infiltrating T cells, NK cells, and macrophages to examine the level of immune response and rejection [126, 127].

Fig. 2.

Generation of hypoimmunogenic islets for diabetes treatment. Hypoimmunogenic hPSCs are first generated through various genetic engineering methods to remove HLA surface molecules and/or overexpress immunoregulatory factors. Tests to confirm the removal of HLA molecules and to validate the overexpression of protein surface markers on the modified hPSCs or tissue of interest include Sanger’s sequencing, western blot, reverse transcription polymerase chain reaction (RT-PCR), and flow cytometry. The modified hPSCs are further assessed for pluripotency, genomic stability, and their ability to differentiate to different lineages. Upon confirmation, the modified cells are differentiated into pancreatic islets via established protocols generating PDX1⁺/NKX6.1⁺ pancreatic progenitors (PPs), which give rise to INS⁺/NKX6.1⁺ stem-cell-derived islet cells (SC-islet) that secrete insulin. In vitro tests and omics analyses, such as pathway analysis, are carried out to confirm the functionality and expression of the marker proteins and biological pathways of pancreatic islets. Functional assays are carried out to ensure that the hypoimmunogenic SC-islets can secrete insulin in response to glucose. Further in vitro tests using co-culture systems with host or reactive immune cells are carried out to measure T cell, NK cell, dendritic cell, and macrophage immunogenic activities toward the SC-islets. In vivo evaluation involves transplanting the modified hPSCs or tissue of interest into animal models such as humanized mice (hu-mice) or non-human primates (NHPs) to assess their survival and potential to evade immune recognition. The transplanted cells are re-harvested and examined histologically to assess immune cell infiltration. The figure was created using Biorender

Further evaluations are necessary to assess safety and potential in clinical applications. For example, genomic stability, the potential to differentiate into multiple lineages, and functional assays are performed to ensure that genetic manipulation did not affect the cells’ ability to differentiate into normal functional cells. Transcriptomic profiling can help to confirm that hypoimmunogenic differentiated cells contain the appropriate expression patterns present in their unmodified counterparts and faithfully replicate their developmental trajectory [44, 128, 129]. Further functional assays, such as glucose-dependent insulin secretion for pancreatic islets, can be applied to the modified cells in vitro to ensure that they can still exert their intended function.

Modification of antigen presentation to tackle T cell and NK cell recognition

Targeting HLA class I to promote CD8⁺ T cell evasion

Several studies have demonstrated that inactivating B2M to disable HLA class I antigen presentation and evade T cell recognition could prolong graft survival [112, 113, 130, 131]. In 1993, Li and Faustman transplanted liver cell grafts from B2m^−/− mouse models, generated through homologous recombination, under the kidney capsule of immunocompetent recipients [130]. After 30 days, B2m^−/− allografts survived in 9 out of 15 mice, with lymphocyte infiltration observed in 2, compared with a complete rejection of all wild-type (WT) allografts. Similarly, 15-day embryonic bodies (EBs) differentiated from B2M^−/− hESCs, which are generated by AAV targeted delivery, reduced T cell IFN-γ expression in co-culture assays with human PBMCs [113]. However, EBs may not be the optimal model since they only express HLA class I but not HLA class II [132], and it is not known what the proportion of cells generated capable of antigen presentation is. Similarly, another report showed that B2M^−/− hESCs generated by transcription activator-like effector nucleases (TALENs) significantly attenuated IFN-γ secretion by human PBMCs in vitro compared with WT [112]. Furthermore, they exhibited a significant reduction in T and NK cell infiltration 2 days post transplantation into the tibialis anterior muscles of mice. Ideally, more time could be allowed for the hESCs to differentiate into cells capable of antigen presentation and for a more pronounced immunogenic reaction to occur for this assay [112]. In a more recent report by Gerace et al., B2M knockout using CRISPR/Cas9 significantly prolonged the survival of hESC-derived islets transplanted under the kidney capsule of humanized diabetic mice compared with WT and similarly reversed diabetes [133]. While WT islets were rejected within 2 weeks, the B2M^−/− islet survival was prolonged and was somewhat functional for up to 7 weeks following a PBMC injection. These studies underscore the critical role of HLA class I in T cell recognition and graft rejection and demonstrated that its depletion can be achieved solely via targeting the B2M gene.

Others directly targeted HLA class I molecules to improve graft acceptance [134, 135]. For example, HLA-A-deficient hESCs generated using zinc-finger nucleases (ZFNs) by Torikai et al. could be useful to increase the chances for donor matching at the other alleles (HLA-B, HLA-C and HLA-DRB1) [134]. Deuse et al. targeted a common region in HLA class I in hESCs via the combined use of small interfering RNA (siRNA) and intrabodies, resulting in a stable 99% reduction of HLA class I surface expression [135]. The modified hESCs showed a significant reduction in IFN-γ and interleukin-4 (IL-4) secretion by PBMCs in vitro, and in vivo following transplantation into the hind limbs of immunocompetent mice. Transplanted hESCs also showed significantly prolonged survival in the xenogeneic hosts for up to 21 days in five out of seven recipients, compared with normal hESCs, which were rejected within 5 days in all recipients. Furthermore, the remaining two transplants persisted beyond the assay’s 42-day duration. These studies demonstrated that HLA class I depletion can also be achieved by directly targeting HLA class I genes to achieve immune T cell evasion.

In an alternative approach, TAP2^−/− iPSCs, generated by ZFNs, and derived myeloid precursor and myeloid cells showed lower IFN-γ release by a co-cultured allo-reactive CD8⁺ T cell line [17]. However, the expression of HLA class I is present in the derived myeloid cells, albeit at a lower level than that detected in WT cells. This is because TAP^−/− cells may still express HLA class I molecules that present peptides originating from membrane or secreted proteins [17, 136]. The reduced immunogenic response to the TAP2^−/− deficient cells could be attributed to the reduced complexity of the peptides presented by these HLA class I molecules [17]. Although this approach highlights TAP2^−/− cells’ potential for reducing T cell evasion, TAP may not be an optimal target for generating hypoimmunogenic cells for universal application in cell therapy.

Targeting HLA class II to promote CD4⁺ T cell evasion

HLA class II molecules can still trigger an immunogenic reaction in some scenarios, and the depletion of HLA class II antigens is necessary to evade recognition by CD4⁺ T cells. The functional disruption of HLA class II in hESCs has been well demonstrated by targeting the CIITA gene [37, 40, 114, 122, 137, 138]. For example, CIITA KO using CRISPR/Cas9 achieves evasion of recognition by CD4⁺ T cells in vitro [114]. Simultaneously targeting B2M and CIITA has been demonstrated to effectively eliminate HLA class I and HLA class II expression in hiPSCs, respectively [138]. Unlike their WT counterparts, HLA class I/II double knockout (DKO)-derived cardiomyocytes do not activate T cells when cultured with PBMCs in vitro. However, it is important to consider the biological complexity of HLA expression pathways. While CIITA is the master regulator for the expression of HLA class II genes, it is not the only pathway. Some HLA class II genes can be expressed independently of CIITA, particularly in specific cell types or under certain conditions [139–141]. Furthermore, CIITA plays a role in the induction of the expression of the non-classical HLA class I molecules HLA-E and HLA-F, but not HLA-G, due to its divergent transcription regulation [138, 142].

Alternatively, directly targeting HLA class II genes has been demonstrated to also be a viable approach for eliminating HLA class II antigen presentation [143]. For example, HLA-DR^−/− iPSCs [118] and iPSC-derived dendritic cells [143] exhibit significantly reduced CD4⁺ T cell proliferation in vitro compared with WT dendritic cells.

Modification of antigen presentation to promote NK cell evasion

The complete absence of HLA class I molecules on the cell surface can trigger host NK cell recognition and attack [15]. Different groups addressed this through the selective retention of HLA-C, or overexpression of the non-classical HLA class I molecules, HLA-E or HLA-G [114–116, 118, 120, 127, 132, 144].

Xu et al. demonstrated that deleting HLA-A and HLA-B via CRISPR/Cas9, while retaining HLA-C in hiPSCs, can effectively evade T cell immunity and is sufficient to suppress NK cell activity [114]. CD43⁺ blood cells and cardiomyocytes differentiated from HLA-C-retained iPSCs or B2M^−/− iPSCs similarly evaded CD8⁺ T cell cytotoxicity in vitro, while those differentiated from WT iPSCs did not. Interestingly, unlike B2M^−/−, HLA-C-retained CD43⁺ blood cells further evaded NK cell lysis in vitro. Furthermore, intraperitoneally transplanted blood cells showed significantly better survival than WT and B2M^−/− over 7 days in immunodeficient mice supplemented with either reactive T or NK cells, respectively [114]. Similarly, a recent study demonstrated that HLA-C retention in iPSCs significantly decreases CD4⁺ and NK immunogenicity in vitro [118]. These studies highlight the advantages of selective HLA retention, a strategy that should also improve HLA-matching since there is no requirement to match for the highly polymorphic HLA-A and HLA-B and considering the low variability of the HLA-C allele. Furthermore, the authors chose to retain a common HLA-C allele, which increases the HLA-matching chances significantly.

HLA-G interacts with different immune cell receptors to inhibit immune effector cells, such as T and NK cells, and promote the expansion of immunosuppressive Tregs [145]. It is known to be expressed on cytotrophoblasts and plays a role in protecting the fetus from maternal immune attack and in tumor escape in cancer [119, 145]. Multiple research groups have adopted HLA-G overexpression into their strategy to create hypoimmunogenic hPSCs [115, 116, 120, 127, 144]. Interestingly, the overexpression of HLA-G alone, using the piggyBac transposon, significantly reduced T cell proliferation as well as NK lysis of hESCs and derived epidermal progenitors in vitro [120]. Furthermore, HLA-G overexpressing B2M-deficient hESCs were generated by knocking in HLA-G within the frame of the B2M gene [116]. In addition to T cell evasion, cardiomyocytes derived from these cells exhibited a marked reduction in NK cell activation and cytotoxicity in vitro compared with B2M-deficient cardiomyocytes. Furthermore, 20 days following subcutaneous transplantation in hu-mice, a significant reduction in NK cell infiltration was observed compared with B2M deficient transplants [116]. Similarly, overexpressing HLA-E via adeno-associated virus (AAV)-mediated knockin at the B2M locus completely reversed the NK-mediated lysis of derived CD45⁺ hematopoietic cells in vitro and enhanced their survival in immunodeficient mice injected with NK cells over 3 days [132].

A recent report suggested that a co-culture with NK cells alone may not be an optimal model that fully recapitulates primary NK cell activity [133]. HLA-E targeted at the GAPDH locus in B2M^−/− hESCs does not offer any extra protection to derived islet cells against co-cultured PBMCs and only provided protection against co-cultured NK cells in the absence of IL-2 activation [133]. IL-2, a cytokine produced by lymphocytes, upregulates several activating receptors on NK cells that are required for NK-mediated lysis [133, 146, 147]. With IL-2 pre-activated NK cells, HLA-E overexpression does not provide any protection to the islet cells in vitro and in vivo [133].

CD47 is a transmembrane protein that interacts with signal regulatory protein alpha (SIRPα) on NK cells and macrophages to protect cells lacking self-HLA class I molecules, such as red blood cells, from innate immune attack [146, 148]. Deuse et al. reported reduced HLA class I/II expression along with elevated levels of CD47 expression in syncytiotrophoblasts that prevent maternal immune attack of allogeneic paternal antigens during pregnancy [40]. In a comparison of their ability for IL-2-activated NK cell evasion, HLA-E, HLA-G, PD-L1, or CD47 were separately overexpressed in the HLA-deficient K562 cell line [122]. While HLA-E, HLA-G, and PD-L1 overexpressing cells are each able to evade only the subset of NK cells that specifically expressed their corresponding inhibitory receptors, CD47 overexpressing cells were protected from all NK cell killing in vitro and in vivo [122].

Furthermore, it has been demonstrated that CD47 effectively enables NK cell evasion in engineered HLA-deficient DKO (B2M^−/− CIITA^−/−) and CD47 overexpressing HIP human and mouse iPSCs [40]. Compared with DKO cells, HIP cells and derived endothelial and cardiomyocyte-like cells completely prevented NK cell activation and killing in vitro. Furthermore, they formed teratomas and achieved long-term in vivo survival for up to 50 days, were void of immune cell infiltration, and resisted NK clearance in allogeneic hosts, an effect reversed by a CD47 blocking antibody [40]. Similarly, HIP human and rhesus macaque iPSCs were generated by the same team and showed efficient NK and macrophage evasion in four rhesus macaques that received subcutaneous (hiPSCs) or intramuscular (rhesus macaque iPSCs) injections [122]. Of note, CD47 has been shown to be species-specific and requires a tissue-specific level of expression to achieve this NK suppressive effect [40, 122, 146].

Overexpressing surface markers or anti-inflammatory cytokines to promote localized immune tolerance

Immune responses to both foreign and self-antigens require precise and balanced actions to eliminate pathogens and tumors while preserving tolerance to self-tissues. Managing the equilibrium between T cell activation and inhibition depends on several factors, including PD-1 and its ligands, PD-L1 and PD-L2, and cytotoxic T lymphocyte antigen 4 (CTLA4), that curb effector T cell responses, thereby preventing autoimmunity [149, 150]. Engineering hESCs to express PD-L1 and CTLA4-immunoglobulin fusion protein (CTLA4-Ig) has been reported using a bacterial artificial chromosome (BAC) achieving homologous recombination at the HPRT locus [126]. This modification increases teratoma size and diminishes CD4⁺ and CD8⁺ T cell infiltration 6 weeks post subcutaneous transplantation of hESCs in hu-mice [126]. Similar results have been also achieved in derived fibroblasts and cardiomyocytes transplanted subcutaneously and intramuscularly, respectively. Notably, the teratomas formed by the transplanted hESCs contained Tregs that secreted significantly higher levels of immunosuppressive cytokines. When mixed and transplanted with WT hESCs, they provided localized immune protection that diminished T cell infiltration into all cells in the formed mixed teratomas [126]. The combined use of both PD-L1 and CTLA4-Ig ligands is required, and neither is sufficient to achieve these results. In a more recent report, the expression of PD-L1 was targeted to the GAPDH locus of hESCs [133]. The overexpression of PD-L1 was shown to not be sufficient, neither alone nor in combination with knocking out B2M, to protect derived islet cells transplanted under the kidney capsule in B6/albino mice. However, as demonstrated by the authors, mouse PD-1 exhibits a reduced affinity toward human PD-L1, which may have weakened its effect [133].

To demonstrate the feasibility of using immunomodulatory cytokines to also enhance local immune tolerance, viral interleukin-10 (vIL-10), an anti-inflammatory cytokine, was introduced into rat islets using an adenoviral vector [151]. This study showed that the combined use of vIL-10 with a subtherapeutic dose of cyclosporine, but neither alone, significantly reduced immunogenicity in vitro. In addition, this combination extended allograft survival in the liver to 29 days, compared with 5 days with control grafts [151]. Gerace et al. tested the protective effect of the overexpression of IL-2 mutein (N88D), IL-10, and TGF-β in hESC-derived islet cells in xenogeneic mouse models transplanted under the kidney capsule [133]. The IL-2 mutein signals the expansion of Tregs, with a minimal effect on CD4⁺ and CD8⁺ T cells, while IL-10 and TGF-β are required to promote the immunosuppressive phenotype of Tregs. These SC-beta cells exhibited significantly prolonged survival in vitro and in vivo compared with WT with absence of any rejection up to 9 weeks and further normalized hyperglycemia in autoimmune diabetic mice transplanted subcutaneously [133].

Harding et al. showed that the overexpression of eight immunomodulatory factors (PD-L1, CD200, CD47, H2-M3, FasL, SERPINB9, CCL21, and MFGE8) in murine ESCs (mESCs) and hESCs (HLA-G instead of H2-M3) via piggyBac and Sleeping Beauty transposons reduce the immunogenicity of cells and their derivatives and promote localized immune tolerance [127]. The so called cloaked mESCs survived for up to 17 days following subcutaneous injection into the neck or flank in different allogeneic hosts, while WT cells were cleared within 5–10 days [127]. These cells also formed teratomas that survived for up to 9 months. Ectopically transplanted cloaked mESCs provided localized immune tolerance to co-transplanted xenogeneic hESCs for up to 8 months and allogeneic islets for up to 4 months. Furthermore, cloaked hESC-derived retinal pigment epithelium cells showed diminished activation of PBMCs, monocytes, dendritic cells, and NK cells in co-culture assays [127]. These studies demonstrate the utility of immunomodulatory factors to generate immune privilege sites that can also provide protection toward co-transplanted tissue. However, transplanting immune-tolerizing cells might also lead to chronic immunosuppression in surrounding tissue, the dose of which may need to be optimized by transplanting a mixture of immunomodulatory-factor-expressing and non-expressing cells to achieve localized graft tolerance.

Safety strategies to minimize risks of hypoimmunogenic cell transplantation

Transplantation of hPSC-derived cells carries a risk of tumorigenesis from residual pluripotent cells [152, 153]. Furthermore, immune-opaque cells can pose a risk to the host if they become cancerous or virally infected and lack the ability to present endogenous antigens. A common safety strategy to mitigate these risks is the inclusion of inducible “suicide” genes in the genetic modification of hypoimmunogenic cells. “Suicide” genes can include toxin, apoptotic, tagging, and drug-conversion genes, all of which are widely used in cancer gene therapy [153]. Among these, drug-conversion genes are the most prevalent. These genes can be activated by administering a specific drug or compound, leading to the targeted destruction of the cells. For example, the herpes simplex virus thymidine kinase (HSV-TK) gene makes cells susceptible to ganciclovir (GCV), a well-known suicide gene system [154]. If hypoimmunogenic cells need to be removed, the suicide gene can be activated, causing cell death and preventing potential complications. This strategy has been recently employed to protect mESC and hESC grafts following subcutaneous injection in immunocompetent allogeneic and xenogeneic hosts [127]. In addition, the RapaCasp9 safety “switch,” which can induce apoptosis in engineered cells when required, was utilized [144]. The engineered cells demonstrated robust immune evasion and could be effectively controlled using the safety switch following subcutaneous injection in vivo, showcasing a comprehensive strategy for generating hypoimmunogenic stem cells. However, it has been demonstrated that the current suicide system induces ~95% apoptosis in hPSCs and their derivatives, requiring optimization for complete elimination [155]. Drug-conversion genes often convert pro-drugs into diffusible toxic drugs, enabling a strong bystander effect in cancer gene therapy by affecting surrounding cells. While this effect is preferred for cancer therapy owing to its bystander action, its use in cell therapy is challenging as it may harm nearby healthy tissue. Employing strategies that generate non-diffusible metabolites, such as HSV-TK, can help reduce this risk.

Comprehensive strategies: combined targeting of HLA and the overexpression of immunomodulatory factors

The need for a multifaceted approach to achieve the full out immune evasion required for clinical application has been highlighted by recent pioneering studies. Han et al. demonstrated a comprehensive approach in hPSCs by targeting HLA-A, HLA-B, and HLA-C to deplete HLA class I, and CIITA to deplete HLA class II, while overexpressing HLA-G, PD-L1, and CD47 to evade recognition by NK cells, T cells, and macrophages [115]. This approach achieved immune evasion from T cells, NK cells, and macrophages in vitro, and significantly prolonged subcutaneously injected teratoma survival in an immunodeficient mouse model sensitized with allogeneic CD8⁺ T cells. Tsuneyoshi et al. similarly targeted HLA-A^−/−, HLA-B^−/−, and HLA-C^−/− to deplete HLA class I, and RFXANK to deplete HLA class II, and further incorporated PD-L1 and PD-L2 for additional T cell evasion, as well as HLA-G and B2M for NK cell evasion [144]. In in vitro co-culture assays, the modified hiPSCs demonstrated diminished T cell immunogenicity compared with WT and significantly reduced NK cell lysis and macrophage phagocytosis compared with DKO. The integration of a RapaCasp9 safety switch has been employed and demonstrated to induce apoptosis to subcutaneous implants in immunodeficient mice following a rapamycin injection [144]. However, for the above studies, no evidence of NK cell evasion was provided in vivo. In addition, in such HLA-G overexpression approaches, it is imperative to keep B2M and alternatively target each of the HLA class I genes separately, which could prove labor intensive and may increase the chances of introducing off-target effects. Optimizing the level of CD47 expression may be more effective at NK cell evasion and sufficient to bypass the need for either HLA-G or PD-L1 [122, 146]. PD-1 ligands would thus be completely redundant in the absence of T cell recognition in this HLA-deficient scenario.

The HIP approach’s efficacy at evading different levels of immunogenicity was demonstrated [37, 40, 122]. While WT iPSCs endured rapid PBMC and T cell immunogenicity after 1 week, no evidence of immunogenicity has been reported against HIP iPSCs for up to 16 weeks. Furthermore, WT and DKO grafts endured B cell, T cell, and macrophage infiltration and were completely lost at 3 weeks and 1 week, respectively, while HIP grafts survived with no immune responses for up to 16 weeks. In addition, differentiated human HIP islets, unlike WT and DKO islets, survived and maintained normal glucose levels in Streptozotocin (STZ)-induced diabetic hu-mice for up to 30 days. Likewise, primary rhesus macaque HIP islets, but not WT, were able to survive and maintain insulin independence in a STZ-diabetic allogeneic recipient for up to 6 months [37, 122]. In an early phase 1 clinical trial to demonstrate the safety and efficacy of HIP-edited primary islets (UP421), results indicated survival and functionality, as evidenced by the detection of circulating C-peptide, a marker of insulin production, up to 6 months post transplantation [39]. C-peptide levels increased during a mixed-meal tolerance test (MMTT), reflecting meal-induced insulin secretion. Initially, the patient exhibited undetectable C-peptide levels both in fasting conditions and during the MMTT. Positron emission tomography (PET)-MRI scans at 12 weeks had also confirmed the presence of islet cells at the transplantation site in the forearm muscle [38]. The HIP platform has demonstrated proof-of-concept in humans, avoiding immune recognition and suggesting potential in allogeneic transplantation without the need for immunosuppression.

Challenges and future directions of hypoimmunogenic islets in diabetes therapy

There is no doubt that significant progress has been made in developing islet cells that can evade the immune system, thereby reducing or eliminating the need for immunosuppressants. This advancement is a major step forward, and continued development is expected to further improve the robustness of these hypoimmunogenic islets, as many research groups are actively working on enhancing these strategies. However, several challenges remain to be addressed before hypoimmunogenic islets can be widely used in patients.

Potential complications and mitigation of excessive immune evasion

The clearance of cancerous or virally infected cells requires the presentation of endogenous peptides through HLA class I, and its absence provides these cells with complete immune evasion, the risks of which should be appropriately addressed. Tumors can arise from hPSCs or their derivatives that have acquired genetic abnormalities during cell culture and differentiation [156, 157], or from multiple rounds of genetic manipulation [41, 42]. Studies have shown that hPSCs can develop mutations in TP53 with increasing cell passages [158]. The genetic ablation of proto-oncogenes, such as BIRC5, MYC, and E2F2 has been considered [159]. However, it is not feasible to identify and ablate all these genes, and some of the genes might be required for the pluripotency and differentiation of hPSCs. Alternatively, various “suicide” gene systems have been employed to selectively ablate unwanted cells [160–165]. However, these systems add an additional layer of genome modification that could increase the chances of off-target effects and genomic instability [41, 42]. Therefore, to qualify hypoimmunogenic hPSCs and their differentiated progenies intended for transplantation, it is essential to thoroughly screen them for tumorigenicity and genomic stability.

The risks of tumorigenicity can be enhanced by residual undifferentiated hPSCs left within differentiated grafts. Excluding these cells is crucial to ensure the purity of the transplant, which can be achieved using islet cell-specific surface markers such as CD49a [56] in the context of SC-islet transplantation, or other purification methods [166]. However, the specificity of surface markers such as CD49a should be carefully examined as it is not beta cell specific in adult human islets [167], and the use of multiple surface markers to enhance specificity and enrichment should be considered. Furthermore, human islets are composed of a diverse population of pancreatic endocrine cells that work together to maintain glucose homeostasis. Isolating a specific group of cells may not be the most effective strategy for diabetes therapy. Some reports suggest the use of a monoclonal antibody, named mAb 84, that specifically recognizes and causes apoptosis in hPSCs, but not their differentiated progeny, without the need for additional genetic manipulation [168–170].

Choice of cell source

When comparing hypoimmunogenic SC-islets versus cadaveric islets, each approach has its own advantages and limitations. A key benefit of using gene-edited cadaveric islets is that they are fully mature and free from non-pancreatic cells. In contrast, SC-islets may contain a mix of pancreatic and non-pancreatic cells. Some of these SC-derived cells can be highly proliferative, and such proliferation must be carefully controlled prior to transplantation to avoid these risks. One of the main advantages of using SC-islets over cadaveric islets is their availability. The demand for cadaveric islets is difficult to meet, as one patient may require islets from more than two donors. In contrast, if hypoimmunogenic hPSCs can be successfully generated, they could be maintained and stored for extended periods, providing an essentially unlimited supply of hypoimmunogenic islet cells and other types of cells. This represents a significant advantage, especially in treating a large number of patients.

Lack of long-term validation

The long-term effects of these hypoimmunogenic islets, regardless of whether they are derived from stem cells or cadaveric donors, remain uncertain, since they have yet to be demonstrated for functional integration and lifetime survival in a host. It is still unclear whether deleting HLA combinations and re-expressing surface markers such as CD47 is sufficient to ensure the transplanted cells’ long-term safety in patients. These questions will take time to answer, especially since clinical trials using gene-edited cadaveric islets have just begun, and patients have yet to receive the full dose of these edited islets. Furthermore, it is still unknown whether these patients will eventually require further transplants, or whether the transplanted cells could lead to any long-term complications, such as tumorigenesis, particularly in the complete absence of immune recognition of the transplanted islets.

While genome-edited hypoimmune cells present an appealing option for allogeneic cell therapy, their long-term safety and efficacy have yet to be established, as the majority of existing studies are limited to in vitro systems or animal models. A recent study indicated that while HLA-edited SC-islets showed enhanced survival in an in vitro PBMC co-culture system, these benefits did not translate to in vivo conditions [133]. Similarly, the PBMC hu-mouse used in this study does not fully recapitulate human immune responses and clinical translation of which is likely to face similar issues. T cells of most hu-mice are selected on mouse MHCs, a limitation overcome by hu-BLT mice that are implanted with human thymic tissue to facilitate HLA-restricted T cell responses [40]. Therefore, selecting appropriate hu-mouse and NHP models is essential for evaluating immune-evasion and immune-tolerizing genetic engineering strategies, as well as assessing the long-term impact of transplanted hypoimmunogenic islets, rather than only focusing on short-term outcomes. Furthermore, in vivo data should still be treated with caution owing to inherent immunological differences between animals and humans [171].

Choice of transplantation site

While some research has been conducted on potential islet transplantation sites, the optimal site for transplantation has not yet been definitively established [172–176]. The choice of transplantation site is critical as it can minimize the islets’ rejection and improve their survival and function, which directly translates to how many donors/islets are required for each patient. The choice of transplantation site depends on several factors, including: surgical accessibility for easily transplanting, monitoring, and retrieving the islets if needed; size or capacity of islets the site can accommodate without causing complications; vascularization to ensure rapid and adequate blood and oxygen supply, and nutrient exchange for islet survival and function [177–179]; innervation to modulate and control islet function [180, 181]; immune privilege to protect the islets [182]; and proximity to the liver for insulin delivery [183]. Outcomes of transplantation may also be influenced by external factors such as the transplantation techniques involved, donor source and islet volume, whether the patient is on immunosuppressants, and whether they have T1D or T2D. Using hypoimmunogenic islets should improve the success rate for long-term application since most of the immediate graft loss can be attributed to the recipient’s immune response [184, 185].

The pancreas, while the natural home for islet cells, is not currently the preferred site for islet transplantation owing to the technical challenges associated with the surgical procedure, the risk of digestive enzyme leakage from exocrine cells, and the anticipated autoimmune recurrence at this site in T1D [186, 187]. It is well vascularized and physiologically relevant, and animal tests achieved normoglycemia with minimal complications using a fraction of the amount of islets required at other transplantation sites [188]. While it is conceptually ideal, most of clinical islet transplantations have been performed in the liver via the portal vein after the success of the Edmonton protocol owing to its ease of access, adequate vascularization, and efficient insulin delivery [7, 8, 43, 60, 176, 183]. However, it is not ideal, and the search for the optimal site for islet transplantation is ongoing. Increased islet numbers transplanted via the portal vein are associated with vascular complications that may lead to liver ischemia, reflecting the low capacity of this transplantation site [189, 190]. Intraportal transplants are also vulnerable to instant blood-mediated inflammatory reactions (IBMIR) responsible for most of the early loss of islet mass [184]. Furthermore, proximity to portal circulation also means exposure to toxins released from intestinal microbiota that may contribute to graft rejection [60, 191]. For these reasons, alternative extrahepatic sites are being explored. The kidney capsule is convenient in research but is not as applicable in humans owing to anatomical differences [188, 192] and, similar to other extrahepatic sites, it delivers insulin peripherally. The anterior eye chamber and omentum offer excellent vascularization, innervation, and are immune privileged, but require invasive surgery and have unique complications (e.g., ocular complications, abdominal infection) [176]. The broad distribution of sites such as the bone marrow, and subcutaneous and intramuscular sites, provide accessibility and the capacity for multiple transplantations at different sites [176, 193]. Being extravascular with no direct contact with the blood, the bone marrow and intramuscular sites may overcome complications such as IBMIR associated with intraportal injection [194]. Subcutaneous and intramuscular sites are very accessible and retrievable, yet they exhibit poor neovascularization and are immunogenic and thus may require a larger mass of islets to be transplanted [193, 195]. SC-islets transplantation under the abdominal anterior rectus sheath has shown robust vascularization and better cell survival than in intramuscular or subcutaneous sites in NHPs [196]. Excitingly, a clinical trial is showing promising outcomes for this transplantation site [63]. Transplantation into adipose tissue is also being explored owing to its highly vascularized and anti-inflammatory microenvironment, which may improve graft acceptance [197, 198]. Further research into the development of biomaterials and strategies to create a transplantation space, promote angiogenesis, and enhance blood and oxygen supply could improve the engraftment of transplanted islets and their long-term survival [186, 199–203].

Off-target concerns in gene-edited hypoimmunogenic islets

Overall, key research steps must still be taken before hypoimmunogenic islets can be used to treat diabetes in patients. CRISPR/Cas9 off-target edits can occur even with more than six mismatches between the guide RNA (gRNA) and the off-target DNA site [48]. Advances in the design of gRNAs and improvements in Cas9 nuclease efficiency and in delivery methods will likely make gene editing more precise [48, 204, 205]. Several predictive algorithms for gRNA design are available that can rank gRNAs specificity to avoid the use of gRNAs with high possibilities of off-target activity [48, 49, 206]. Although useful, the predictions given by such tools still need to be validated technically through genome wide off-target analysis methods that are constantly being developed [48, 49]. Cell-free methods such as Digenome-seq, SITE-seq, and CIRCLE-seq involve the in vitro digestion of extracted genomic DNA using CRISPR/Cas9 and the gRNA to be evaluated followed by whole-genome sequencing (WGS). However, such tools fail to consider the epigenetic and the chromatin organization states of the native nuclear environment. Other cell-free methods that involve pre-incubation of CRISPR/Cas9 and gRNA with extracted nuclei (DIG-seq), or with suspended live cells (Extru-seq) prior to genome extraction address these issues. Undoubtedly, the gold standard is to assess the off-target effects directly in cells in culture using WGS. Since the majority of sequence data collected during WGS represents unedited genomic DNA, there is a low signal-to-noise ratio, which is addressed by methods that add target enrichment such as the cell-free SITE-seq and the cell-culture-based GUIDE-seq. Limitations of WGS include throughput, cost, and efficiency [48, 49].

Future outlook

Numerous genes involved in immune system regulation have not yet been considered, and analyzing the intrinsic ligand profile of islets might be helpful to elucidate the appropriate genome modification required to achieve a hypoimmunogenic phenotype. For example, novel strategies such as knocking out B7-H3 and CD155 ligands to target NK cell activation are being considered [207]. Furthermore, minimizing gene editing according to need would help reduce potential off-targets and avoid affecting islet function. For example, an alternative strategy has been devised by expressing viral US2 glycoprotein to deplete HLA class I while retaining non-classical HLA class I molecules to achieve both T cell and NK cell evasion [208]. Recently, transcriptomic profiling and whole-genome CRISPR screening of SC-islets under immune interaction with allogeneic PBMCs have been performed [209]. The generation and deep analysis of such datasets would help define the genes involved in relevant scenarios and tailor alternative gene-targeting strategies.

The recent clinical trial by Sana Biotechnology, which involved the intramuscular transplantation of hypoimmune cadaveric islets into a patient with T1D, has shown promising initial results. However, these findings are still in the early stages, as the trial has not yet used the full dose of the transplant. Moreover, the study has not yet been completed, making it premature to draw definitive conclusions. To fully evaluate the effectiveness and long-term viability of this approach, further testing with the complete transplant dose and longer follow-up periods, extending over several months, will be necessary. This will help determine whether the initial promising signs can translate into sustained therapeutic benefits. In addition, it remains unclear whether encapsulation of these transplanted hypoimmunogenic islets will be necessary to control the transplant and facilitate its removal, if needed. For long-term clinical translation, the incorporation of a safety strategy might be required to eliminate the grafted cells from the host to avoid potential complications.

Abbreviations

AAV

Adeno-associated virus

APCs

Antigen-presenting cells

B2M

β2 microglobulin

BAC

Bacterial artificial chromosome

CAR

Chimeric antigen receptor

CIITA

Class II transactivator

CTLA4-Ig

CTLA4-immunoglobulin fusion protein

DKO

Double knockout

DM

Diabetes mellitus

EBs

Embryoid bodies

EGFR

Epidermal growth factor receptor

EGFRt

Truncated epidermal growth factor receptor

GCV

Ganciclovir

GMP

Good Manufacturing Practice

gRNA

Guide RNA

hESCs

Human embryonic stem cells

HIP

Hypoimmune pluripotent

HLA

Human leukocyte antigen

hPSCs

Human pluripotent stem cells

HSV-TK

Herpes simplex virus thymidine kinase

Hu-mouse

Humanized mouse

IBMIR

Instant blood-mediated inflammatory reactions

IFN-γ

Interferon gamma

IL-10

Interleukin-10

IL-4

Interleukin-4

iPSCs

Induced pluripotent stem cells

MD

Monogenic diabetes

MHC

Major histocompatibility complex

MLRs

Mixed leukocyte reactions

MMTT

Mixed-meal tolerance test

MODY

Maturity-onset diabetes of the young

MRI

Magnetic resonance imaging

MSCs

Mesenchymal stem cells

NHP

Non-human primate

NK

Natural killer

PET

Positron emission tomography

PPs

Pancreatic progenitors

RT-PCR

Reverse transcription polymerase chain reaction

SC

Stem cell

siRNA

Small interfering RNA

SIRP-α

Signal regulatory protein alpha

STZ

Streptozotocin

T1D

Type 1 diabetes

T2D

Type 2 diabetes

TALENs

Transcription activator-like effector nucleases

TAP

Transporter associated with antigen processing

TGF-β

transforming growth factor-β

Tregs

Regulatory T cells

WGS

Whole-genome sequencing

WT

Wild-type

ZFNs

Zinc-finger nucleases

Author contributions

K.S. contributed to the design, data collection, and writing of the manuscript. E.M.A. contributed to the conception and design, review, and editing of the manuscript. All authors read and approved the final version of the manuscript.

Funding

This work was supported by a budget from Sidra Medicine (project no. SDR400215; SDR400217).

Availability of data and materials

Not applicable.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing Interests

All other authors declare no conflicts of interest.