ACADÉMIE NATIONALE DE MÉDECINE Genetic editing of human germline cells and embryos April 2016 In 2015, the “Académie Nationale de Médecine” appointed a Working Committee* to discuss research and the medical implications of altering the human germline and early human development. The specific objective was to identify potential medical indications for these new molecular genetics methods, to assess the risks and uncertainties of their use based on current knowledge, and to consider the ethical issues they raise. The resulting report has been submitted to the plenary session of the Academy on April 12, 2016 and has been adopted: 50 votes in favor, 20 against, 14 abstentions. Abstract: Interventions causing genome modifications that can be passed on to descendants have been prohibited in France since 1994. New methods, such as CRISPR-Cas9, have been developed, raising questions about their potential use on human germline cells and embryos. The only acceptable medical indication would be to prevent the transmission of a disease gene to the child. However, the necessary conditions have not yet been met for this technology to be considered for clinical use, particularly as concerns the efficacy and safety of these methods. There are also other ways for couples to achieve the goal of having children. The ethical questions raised by these technologies will require multidisciplinary discussions within the wider debate on all assisted reproductive technology procedures, which may affect the genome of the unborn child, and, possibly, of subsequent generations. However, this research, including that on germline cells and human embryos, should be carried out provided that it is scientifically and medically justified. *The following members of the Académie Nationale de Médecine participated to the Working Committee: Full members: Monique Adolphe, Jean-François Allilaire, Raymond Ardaillou, Claudine Bergoignan-Esper, Yves Chapuis, Francis Galibert, Alain Fischer, Pierre Jouannet (coordinator), Jean Yves Le Gall, Jean François Mattei, Jacques Milliez, Alfred Spira Corresponding members: Gérard Benoit, Nathalie Cartier-Lacave, Marc Delpech, Philippe Jeanteur, Yves Le Bouc, Jean Louis Mandel, Florent Soubrier And Anne Fagot-Largeault (Académie des Sciences) 2 Contents: 1 Introduction page 4 2 Legislative, institutional and organizational environment page 5 3 Challenges of human genome editing techniques that can affect the germline page 6 3.1 Potential clinical applications page 6 3.2 Mode of action page 8 3.2.1 Targeted modification of the embryo genome page 8 3.2.2 Targeted modification of the germline genome page 10 3.2.3 Efficiency and safety of genome editing techniques in embryonic And germline cells page 10 4 Research page 11 5 Ethics page 12 5.1 Ethical questions relating to research performed on human germline cells and embryos page 12 5.2 Ethical issues regarding the clinical use of technologies that can edit the genome of germline cells and human embryos page 13 6 Conclusions page 14 7 Recommendations page 15 References page 16 Annexe 1: Terminology page 20 Annexe 2: Experts Consulted page 21 3 1 Introduction There has been tremendous progress over the last 50 years in our understanding of the role of gene expression in cellular function and dysregulation. Initially, tools were developed to unravel the genome, generating a wealth of information on genetic variations and mutations underlying diseases. The next step of the research effort was to engineer the DNA for experimental or therapeutic purposes, by preventing, adjusting, or modifying gene expression. Clinical trials were progressively undertaken to treat patients with hereditary diseases and certain forms of leukemia and lymphoma [1, 2]. New molecular tools have recently been developed, based on the bacterial defense system against exogenous DNA (bacteriophages or plasmids), the CRISPR (clustered regularly interspaced short palindromic repeats) system [3]. A guide RNA, coupled to an endonuclease (Cas9), precisely targets any specific sequence in the genome. Then the two strands of DNA are cuts and, depending on the application, the targeted fragment of the DNA molecule is removed or replaced by inserting a new DNA sequence. The modified DNA molecule is then repaired by one of the two DNA repair systems present in all cell types: homologous directed recombination (HDR) or non-homologous end-joining (NHEJ). CRISPR-Cas9 has proved to be more effective than previous methods (TALEN, zinc finger nuclease, meganucleases). It is relatively simple to perform and affordable, explaining its rapid spread to many laboratories since its discovery [4-7]. There is considerable potential for further improvement, as for any new technology, including the recent use of bacterial nucleases with different modes of action [8, 9]. CRISPR-Cas9 research has been conducted exclusively in animal models or cell cultures. Clinical applications for this technology have been suggested for repairing single-gene disorders, treating cancer, or inducing potential therapeutic or protective effects against infectious diseases [10, 11]. These applications of CRISPR-Cas9 technology would be appropriate only for somatic cell-based gene therapy [12]. In 2015, a Chinese peer-reviewed publication made waves because it raised the possibility of editing the genome of human embryos. The purpose of this research, using triploid and therefore non-transferable embryos, was to determine to what extent the CRISPR/Cas9 system could be used to replace the mutated β-globin gene responsible for thalassemia [13]. The results were unconvincing, demonstrating low efficiency for modification of the target gene, with numerous undesirable modifications. Reactions were heated, criticizing the approach, and calling for a moratorium or even a ban on any research aiming to modify the genome of human embryos [14, 15]. In addition to the scientific discussion about the effectiveness and safety of the method, ethical and social concerns have been raised about the use of the technique for trivial or eugenic practices (the “slippery slope” argument) [16]. The presence of the induced DNA modifications in all cells, including those of the germ line, leading to their being passed on to subsequent generations, was also of concern while many uncertainties about unintended consequences remain. 4 2 Legislative, institutional and organizational environment In accordance with French law, interventions seeking to introduce any modification into the genome of any descendant have been clearly prohibited since 1994. Article 16-4 of the Civil Code states: “… Without prejudice to research seeking to prevent or to treat genetic diseases, no alteration can be made to genetic characteristics with the aim of modifying a person's offspring”. In addition, the Oviedo convention, ratified by France in 2011 and by most European countries, clarifies, in article 13, that “in every case, any intervention which aims to modify the human genome must be carried out for preventive, diagnostic or therapeutic purposes and only if its aim is not to introduce any modification in the genome of any descendants". Finally, although the Universal Declaration on the Human Genome and Human Rights published by UNESCO in 1997 does not explicitly apply to alteration of the human genome, the International Bioethics Committee of this institution published a report on October 2, 2015 specifying: “The international community of scientific researchers should be entrusted with the responsibility of assessing and ensuring the safety of procedures that modify the human genome. A thorough and constantly updated investigation on all the consequences of these technologies is required.” and that "It is important for States and governments to renounce the possibility of acting alone in relation to engineering the human genome and accept to cooperate on establishing a shared, global standard for this purpose.” The international position on this topic is heterogeneous, but many countries have taken steps to prohibit any modification of the genome that can be transmitted to subsequent generations through human germline cells. In addition, the provisions concerning genomic changes to the human germ line may become unclear when combined with other provisions concerning embryo research or genetically modified organisms (GMOs) in general [17]. Since the publication of the article by Liang et al. in 2015 [13], individual statements and concerns have been increasingly voiced. These statements come from scientists, scientific societies, but also from governments. In the U.S., the NIH stated that no clinical research protocols “using germline gene transfer” will be considered. On May 26, 2015, the White House declared that it supports any serious evaluation of the ethical issues in the field, while emphasizing that the modification of the human germ line for clinical purposes is a limit that should not yet be crossed. The U.S. National Academies of Science, Engineering and Medicine have launched a joint study in December 2015, with an international summit at which the scientific, ethical and political issues raised by “genome editing” were examined. Professional societies, such as the Society for Developmental Biology, have called for a moratorium on any manipulation of the preimplantation human embryo by "genome editing" [18]. The International Society for Stem Cell Research has also called for a moratorium restricted to clinical applications, with rigorous scientific studies evaluating the risks of the 5 method undertaken as part of a broad public discussion on the societal and ethical implications of the technique [19]. A view similar to that of the International Society for Stem Cell Research was also expressed in a review of EMBO Reports on this topic [20]. The Hinxton group formulated the same recommendation, while recognizing that once the questions of efficiency, safety, and governance have been settled, morally acceptable applications of the technology could be found in the field of human reproduction [21]. In the United Kingdom, the five principal agencies responsible for biomedical research, all of which believe that this technology has potential clinical applications, declared that they would support preclinical research using "genome editing", including the human embryo and germline cells. [22]. The Nuffield Council on Bioethics has also established a working group on this topic. In Germany, the government allocated funds for a debate on this issue. The Federation of European Academies of Medicine will organize a workshop in 2016. 3 Challenges of human genome editing techniques that can affect the germline 3.1 Potential clinical applications This technique could be used to prevent the transmission of a particularly serious hereditary disease to a child, if the causal genetic abnormality has been clearly identified and there is no treatment. The transmission of monogenic changes to the child can be avoided by prenatal diagnosis, potentially followed by medically assisted termination of the pregnancy or by preimplantation genetic diagnosis (PGD), generally performed on the third day of embryo development (8-cell stage). This makes it possible to transfer only embryos not carrying the genetic defect responsible for disease into the womb [23]. There are, however, rare cases in which PGD is not the answer to the problem. For example, if one of the two parents-to-be is homozygous for a dominant autosomal disease (Huntington's chorea), or if both parents-to-be are homozygous carriers of a recessive autosomal disease (cystic fibrosis), PGD cannot be used. In addition, some homoplasmic mutations of mitochondrial DNA (as in Leber’s hereditary optic neuropathy) also prevent the use of PGD. Other cases, in addition to these indisputable but rare occurrences, are those of couples who tried PGD but for whom no embryos could be transferred. Among the 119 PGDs performed at the Necker-Antoine Béclère Center between 1-1-2015 and 11-15-2015, embryo transfer was not possible for 22 couples (18%). In most cases, all the analyzed embryos were affected by a genetic disease. For these couples, PGD was suggested because of an autosomal dominant disease (11), autosomal recessive disease (8), an X-linked disease (2), or an abnormality of the mitochondrial DNA (1). The concerned patients often asked whether embryos could be "treated" rather than destroyed. This request from patients could become increasingly common [24]. Even if modifying the genome of embryos to avoid transmitting a genetic disease to a future child were acceptable, this approach is not currently clinically feasible (see below). It is also 6 important to note that there are other ways for couples to achieve their parental objectives: adoption, gamete donation, embryo donation, with all these options being legal and often used in France. Finally, one of the possible outcomes of research using methods such as CRISPR-Cas9 is the development of somatic gene therapies that could benefit the affected child after birth. Other indications could become integrated into an approach aiming to reduce the risk of common diseases developing or to "protect" the individual, sometimes referred to as “medical indications”. This approach has also been described as "transhumanism" by some, because it could be used to "enhance" the human being. There are, indeed, natural variants of the genome that can play a strong "protective" role against diseases such as diabetes (SLC30A8 gene), hypercholesterolemia (PCSK9 gene), and some viral infections (CCR5 gene). The introduction of these variants into individuals, by targeted modification of the germ line, could be viewed as “protection” [25]. The elimination of the e4 allele of the APOE gene may also decrease the risk of developing Alzheimer's disease [26]. Targeted modification of these variants would, thus, be a way to improve the performance of humans, by making them less vulnerable to certain diseases. However, this approach may have some flaws, because we do not yet completely understand the various roles played by these variants. For example, it has been suggested, but not yet confirmed, that the APOE e4 allele is associated with better memory in young adults [27]. Countless interventions on the human genome would be required to introduce new qualities because of the number of genes involved and the number of disease to which humans are susceptible, such as cardiovascular diseases, cancer, neurodegenerative diseases and infectious diseases. In addition, several identified genetic variants are poorly correlated with phenotype, are generally not the main variants involved (these variants often remain unidentified) and often interact with other genes. Their modification in the human genome would make such physiological changes to prevent the occurrence of disease completely illusory. Finally, this strictly genetic view of disease ignores other factors involved in disease susceptibility and development, and other existing ways to prevent or treat disease. Thus, the limitations of a project involving the targeted modification of the human genome to generate a “superman" are apparent, and such a project belongs to the realm of science fiction. Would it be possible to modify the genome of germline cells or the embryo to favor specific characteristics or traits in the unborn child? All new technologies developed for intervention in the process of human procreation (from in-vitro fertilization (IVF) to PGD) over the years have been debated and contested because they might be used to satisfy the aspirations of parents and physicians wishing to create "designer babies" or to promote a new form of eugenics. This threat has been raised again by many since the description of techniques for the effective modification of DNA structure, particularly since the publication of the article by Liang et al. on human embryos. [13] The risk of misusing these technologies to modify the genome of the embryo to choose the 7 physical or other early characteristics of the unborn child cannot be ignored, but success would be far from guaranteed in such a project. Indeed, most features would be unpredictable on the basis of a simple DNA modification during embryonic development for several reasons. First, the genomic structure of an individual is not stable throughout development [28]. Second, the expression of a gene is usually controlled by the expression of other genes and by epigenetic and/or environmental factors. There is, thus, no simple and direct relationship between the nucleotide sequence of an embryo and the phenotype of the resulting child, although some variants could clearly lead to striking phenotypic changes. The other risk is that changes might be made to the germline genome to "improve" humans or "enhance" their performance. This issue, raised by the theoretical possibilities offered by CRISPR-Cas9 and similar techniques, is covered by the broader debate on the evolution of medicine. Medicine is increasingly being asked to act outside its traditional mission of care. It now intervenes to replace faulty organs, tissues, cells, or genes. It can also intervene, with various degrees of invasiveness, to stimulate function or to improve performance in many areas, such as sports or military activity. These various forms of intervention have been facilitated by considerable technological progress. How do the new techniques altering DNA structure fit into this context? A classification of all technologies used to “enhance” the human body into four categories, according to their scope and duration, has been proposed [29]: - Localized and temporary or reversible actions, such as the use of removable prosthesis; - General and temporary actions, such as the use of doping agents; - Localized and definitive action, such as some somatic gene therapies; - General and definitive action, such as germline gene therapy. This last category is the most interventionist, and would affect not only the “treated” individual, but also their progeny. 3.2 Mode of action For decades, the modification of animal genomes at the embryonic stage in the laboratory has constituted an important area in basic science. This type of research has helped to identify genes correlated with a given phenotype, to create models of human diseases, and has been used for agronomic purposes [30]. The yield and efficiency of these techniques have been greatly increased by the use of nucleases, and, more recently, the CRISPR-Cas9 system. 3.2.1 Targeted modification of the embryo genome In the last three years, the birth of animals obtained after targeted modification of the embryonic genome by the CRISPR-Cas9 method has been reported for many species (Table 1). In most cases, the molecular material was microinjected in vitro at the zygote stage (first embryonic cell), either into the cytoplasm or directly into the pronuclei. Embryos were then transferred into the womb of surrogate recipients, either immediately or at the blastocyst stage. The number of live births was often very small. The aim was to abolish the expression of the targeted gene, to overexpress that gene or to modify it. In most cases, the desired genomic 8 modification was found only in a minority of newborn animals. Furthermore, mosaics were often observed. This incomplete modification of cells in the offspring may be explained by either the late action of CRISPR-Cas9 during embryonic development or by the genetic modification of the paternal and maternal alleles occurring at different times during the transition from gamete to embryo [31]. Finally, unexpected phenotypes often occurred in animals with modified genomes, particularly if the DNA molecule was repaired by non- homologous end-joining (NHEJ) as opposed to homologous directed recombination (HDR). Newborn Transferred Newborn Authors Species Target gene with modified embryos animals gene Wu 2013 [32] Mouse Crygc 472 78 36 (46%) Mizuno2014 [33] Mouse Tyrosinase 205 60 28 (46%) Whitworth 2014[34] Pig CD163 93 4 4 CD1D 110 4 2 Niu 2014 [35] C y n o m o l g u s P p a r γ , R a g 1 , 83 2 2 monkey Nrob1 Ménoret 2015[36] Rat Anks3 156 22 4 (18%) Zou 2015 [37] Dog Myostatin 35 27 2 (7%) Kou 2015 [38] Ferret Dcx 117 15 11 (73%) Aspm 64 12 8 (75%) Disc1 18 4 1 Crispo 2015 [39] Sheep Myostatin 53 22 10 (45%) Honda 2015 [40] Rabbit Tyrosinase 67 9 2 Wang 2015 [41] Sheep Myostatin, FGF5 416 98 26 (26%) Table 1: Genomic changes observed in newborn animals after the induction of genomic changes in vitro by the CRISPR-Cas9 method and microinjection at the zygote stage of embryonic development (one cell). The introduction of the components of the CRISPR-Cas9 system into zygotes by electroporation has been suggested, because microinjection is technically difficult and potentially dangerous for the embryo. In mice and rats, electroporation and microinjection have yielded similar results [42-44]. No attempts at targeted modification of the embryonic genome appear to have been made after the zygote stage and before embryo implantation. There are, thus, currently no experimental results indicating that it would be possible to act on embryos subject to PGD, at the eight-cell stage or beyond. In any case, from this stage onward, microinjection into the 9 embryonic cells would be difficult. Could other vectors, such as retroviruses, which have been successfully used in other cellular systems [46] and in somatic gene therapy, be used at this stage? Again, there are no experimental results for embryos to suggest that such techniques would be useful. 3.2.2 Targeted modification of the germline genome Another approach to modifying the genome in all the cells of a person would be to intervene before fertilization, by injecting the various components of the RNA-Cas9 complex into the mature oocyte (metaphase II) at the same time as the sperm cell or sequentially [32]. This technique of gene modification in oocytes has been experimentally tested in mice, to reduce the rate of mitochondrial mutation (using the TALEN system and not CRISPR-Cas9 as the endonuclease). However, in this experiment, there was no embryo transfer so no offspring [47]. Application of the technique at an earlier stage of oogenesis is not possible because the oocytes are not easily accessible in the ovary. In addition, oocytes in adult ovaries are at the meiotic prophase stage, with oogonial stem cells present only in the fetal ovary. The requirements are different in males, because the stem spermatogonia can easily be removed from the postpubertal testis. The cells can be treated in vitro by CRISPR-Cas9 and cultured to ensure their proliferation and the formation of cell colonies, on which all the necessary controls can be performed before transferring the modified spermatogonia into the testis, where in vivo spermatogenesis will occur. This technique was successfully used in mice to correct a mutation of the Crygc gene (Crygc-/-) causing cataracts. The modified gene was present in all newborns, which had a normal phenotype and in which no off-target effects were observed on whole-genome sequencing [48]. Other experiments in mice and rats have been less successful [49-50]. Indeed, it is necessary to transfer the modified spermatogonia into the testicles of a living animal to obtain gametes, because spermatogenesis cannot be reproduced in vitro. However, to ensure that all the produced sperm cells carry the desired genic modification, native germline cells present in the host testis must be eliminated, and this is not easy to achieve. This approach is useful for the production of transgenic animals or for studying male infertility of genetic origin. It is unlikely, however, that it could be used in humans to obtain "correct versions" of gametes in a homozygous carrier of a gene mutation, to prevent the transmission of a genetic defect to descendants. It would also be necessary to ensure that germline cell differentiation and maturation in the adult testis, and the in vitro processing and proliferation of these cells in culture did not lead to epigenetic abnormalities or genomic imprinting defects that might be transmitted to future generations. 3.2.3 Efficiency and safety of genome editing techniques in embryonic and germline cells When the objective is to modify all the cells of an individual, regardless of the timing of genome engineering (before or after fertilization), the efficiency and safety of the method must be much higher than for an experimental approach, or even somatic gene therapy. When 10