History of Iterated Embryo Selection
The multiple-invention (>3) history of the idea of extremely-powerful embryo selection by using gametogenesis to run many ‘generations’ in vitro.
The idea of iterated embryo selection—conducing multiple generations of embryo selection in a petri dish by exploiting gametogenesis or stem cells—has a complicated history. Tracing relevant papers back to 198935ya, the idea appears to have been invented independently at least 4
A predecessor was introduced by 1991 as “velogenetics”. Velogenetics led to what appears to be the first invention of IES, 1998’s “whizzogenetics”. It was then invented in 200915ya by Carl Shulman as “iterated embryo selection”/“IES”. It was reinvented a third time by 2013 as “in vitro eugenics”. And it was reinvented up to 3× in 2018, as “in vitro breeding”, by et al 2018 /et al 2018 /et al 2018 (whose relationship is unclear, as the latter two claim novelty but publish not just the same idea but same name, while the former, published before them and giving said name & idea, nevertheless does not claim novelty).
The history of IES as an idea is surprisingly young—tracing back only to 199826ya, as far as I can tell.
The idea developed out of livestock breeding where IVF and egg extraction/harvesting have long been practiced to allow particular individuals to have many more offspring than possible through normal breeding (eg. a cow can only have a few calves via normal pregnancy per year, but if her eggs are extracted, any number can be implanted into surrogate cows), but has appeared in human genetics as well, spurred by the steady progress of stem cell research, allowing for an even faster form of IES. I’ve noticed that animal genetics and human genetics have been oddly divergent for many decades—for example, SNP heritability methods appear in standard animal genetics textbooks like Lynch & Walsh 199826ya a decade before GCTA was applied to human research, and workhorses of animal genetics like the pervasiveness of additivity or the infinitesimal model were studiously ignored by medical researchers until GWASes forced their revival—and the many inventions of IES may reflect this lack of communication.
Et Al 1989
One of the earliest papers discussing combining egg manipulation in vitro with breeding systems is et al 1989’s “Potential genetic improvement of cattle by fertilization of fetal oocytes in vitro” discusses the technical feasibility of maturing embryos’ eggs in vitro to the point of being fertile and capable of creating new embryos in vitro, and the potential of creating, from an initial female egg, a line of successive embryos which could yield eggs which are then fertilized by a different elite male in each generation (different ones to avoid inbreeding and keep raising the genetic mean of the sequence of embryos). Such a system would rapidly bring the embryo line up to the average of the elite males, at which point the embryos would then be implanted in a surrogate and tested in the real world for their phenotypic fitness, and then the process can start over again with a new elite female and a new set of elite males, in a rapid “tick-tock” fashion. (“The average genetic merit of the offspring born from the fetal oocytes would rapidly become equal to the average merit of the selected sires; those offspring above and below average would be identified by a further round of progeny testing and so the genetic improvement would be continuous.”)
Betteridge et al 198935ya do not propose to use any direct measurements of either genetic potential or genetic relatedness to select among embryos, and so their system derives the value of embryos simply as the average of the breeding values of parents estimated from their own phenotypic data (such as offspring) & pedigree; this also puts a ceiling on the gain in each cycle—the embryos can’t exceed the sires on average, and so there’s no point in running the cycle for too many steps, since one needs to then measure the embryos into the real world to see which are better/worse (some embryos will have larger or smaller polygenic scores, but which embryos are unknown without sequencing, which was not feasible in 198935ya).
This clearly anticipates IES by establishing an in vitro multi-generational cycle for genetic gains, but is missing the key part of genomic selection, which allows eliminating the need for regular halts to slowly do full pregnancies & phenotyping.
1991
1991’s “Velogenetics, Or The Synergistic Use Of Marker Assisted And Germ-Line Manipulation”, proposes a system they call “velogenetics”. In that, one does IVF with elite sperm/eggs, which are then hypothetically genetically-sequenced & marker-selected among, but then one does not implant the embryo into a surrogate for normal pregnancy; instead, it is matured in vitro (an artificial womb) for several months until the earliest oocyte follicles are grown, then one extracts those and create a new embryo which is then implanted and born naturally, giving a generation time of “three to 6 months” (as opposed to the more usual 1 year+ required for pregnancy and then sexual maturation).1 Hence their other name for velogenetics, “generation skipping”- you skip one generation in vitro.
But they don’t go to full IES because they don’t propose using stem cells (which is much faster & creates sperm as well), or potentially doing indefinitely many generations/cycles in vitro, which is where much of the power of IES would come from. Georges & Massey 199133ya cite Betteridge et al 198935ya, and add in “marker-assisted selection” as a step; the use of marker-assisted selection avoids the major rate-limiting step in Betteridge et al 198935ya, and it would seem natural to extend it to indefinitely many steps, so in retrospect it’s a little surprising that they don’t propose full-blown IES by ‘closing the loop’.
1998
1998’s “Strategies to Utilize Marker-Quantitative Trait Loci Associations”, in a review of the potential of marker-assisted selection (or PGSes) for breeding, appears give the first true iterated embryo selection proposal by critiquing Georges & Massey 199133ya and proposing a small modification which they (continuing the ‘velo’ theme) dub “whizzogenetics” & illustrate in a diagram:
A major potential drawback in difficulty, cost, and welfare of these velogenetic schemes is the need to harvest oocytes from calves in utero. If the technology eventually develops to a stage at which cell differentiation can be controlled in vitro, then in vitro meiosis, followed by fertilization, may become possible. In this case, the step requiring transfer to the recipient female, followed by oocyte harvesting, would become redundant. Cell cultures derived from fertilized oocytes could be selected using markers and then induced to undergo meiosis. After fertilization (not necessarily via a true oocyte), the resulting cultures could again be selected on marker information, and the process could be repeated (Figure 5c). After a number of generations, once the desired genotypes are attained, animals are regenerated, possibly via nuclear transfer (43). Such a scheme would allow for rapid introgression when genotypic information alone is used, for example, introgression of the polled gene into Holstein-Frisian cattle with little genetic lag. With high density marker maps and knowledge of close marker-QTL associations, more generalized selection objectives could also be tackled in periods of less than one natural cattle generation…We think that MAS has much to offer animal breeding when used with other new technologies and when used carefully to complement selection based on phenotype. Rapidly developing genome technologies mean that the close linkages envisaged by Smith and Smith (34) will soon be the rule rather than the exception.
This is not exactly the same as later IES proposals, since they imagine maturation of each embryo to the point where gametes develop normally and can be harvested, rather than brute-forcing it by regressing cells to stem cells & then turning them into gametes, but that merely speeds up the process somewhat and doesn’t fundamentally change the capabilities or upper bounds on improvement, so I consider it a true IES. Judging from hits for “whizzogenetics” and checking citations of Haley & Visscher 199826ya, the idea of it & velogenetics was almost entirely forgotten. (The facetiousness of the name probably didn’t help.)
2009
The next appearance is in the 2009 MIRI “Uncertain Future” scenario planning tool FAQ, answer #7. The entry was written by Carl Shulman, who independently invented it (personal communication, 2017), giving it the much better name of “iterated embryo selection”:
Multi-generational in vitro embryo selection, also known as iterated embryo selection (IES), refers to a not-yet-developed biotechnological technique that could be used to select human embryos for specific genotypes. IES, like the currently-existing practice of pre-implantation genetic diagnosis (PGD), would allow “designer babies”, and hence might be controversial in some countries…Next, the third step is repeated again. The eggs and sperm are intelligently and selectively combined in the process of fertilization, creating “grandchildren zygotes” of the original embryonic stem cells from the first step. By sequencing the relevant genes at each stage of the process, then combining the gametes together in an intelligently orchestrated way, embryos can be created with the desired alleles. How specific the genes can be made is a function of how many iterations are used. Thus the term iterated embryo selection…This technology is a combination of 3 others: artificial gametes (not yet developed for humans), in vitro fertilization (invented 197846ya), and pre-implantation genetic diagnosis (invented 199034ya).
The idea here appears to be identical to whizzogenetics in all important aspects; presumably the idea was developed earlier than the December 200915ya FAQ snapshot, inasmuch as the “Uncertain Future” tool was not primarily genetics-related & it is simply one of the scenario options included for completeness. The “iterated embryo selection”/“IES” name has caught on much more than “whizzogenetics”; one early appearance was pg98–99 of the somewhat-popular futurology book Singularity Rising, Miller 201212ya, and the FAQ is cited (along with Miller 201212ya & Sparrow 201311ya) in Shulman’s later & much more widely-read 2014 discussion of embryo selection, which paper (or its overlapping discussion in 2014 book, Superintelligence: Paths, Dangers, Strategies) appears to be where most people at present have first encountered the idea of IES (eg. 2016, 2016).
2013
The next appearance of the idea is 2013’s “In vitro eugenics” discussion of it from a bioethics standpoint. Sparrow cites neither Haley & Visscher 199826ya nor Shulman 200915ya nor Miller 201212ya as the source of what he dubs “in vitro eugenics”. Introducing the idea, Sparrow writes:
However, the development of a technology of in vitro gametogenesis would also make possible other technological interventions into human reproduction, which as yet have barely been discussed at all. In particular, it might allow what I will call “in vitro eugenics”: the deliberate breeding of human beings in vitro by fusing sperm and egg derived from different stem-cell lines to create an embryo and then deriving new gametes from stem cells derived from that embryo, which in turn might be used in the creation of another embryo. Repeated iterations of this process would allow scientists to proceed through multiple human generations ‘in the lab’.ii…More controversially, it might also function as a powerful technology of ‘human enhancement’ by allowing researchers to use all the techniques of selective breeding to produce human individuals with a desired genotype.
ii. et al 2009 note the potential of in vitro gametogenesis to facilitate the ‘practice of in vitro genetics’ for research purposes, so that ‘scientists will be able to conduct multigenerational human genetic studies in a dish’.4 They also draw attention to the fact that in vitro gametogenesis may facilitate human enhancement by greatly increasing the power of PGD by removing current limits imposed by the small number of eggs salvaged in each cycle of IVF. et al 2012 investigate at length the power of in vitro gametogenesis to enhance the human genome and advocate its use to this end.6 However, to my knowledge, the current paper is the first to explicitly discuss the possibility of the iterative use of this technology for reproductive purposes and is the first full-length consideration of the challenges that will need to be overcome before it will be possible to use in vitro gametogenesis to breed human beings in vitro.
Does it appear in Mathews et al 200915ya or Bourne et al 201212ya? et al 2009 lays out some of the possible consequences of full control of stem cells for reproduction, but while Mathews et al 200915ya explicitly say that it could be used for selection, they appear to think only of a single selection step (ie. what I call “massive embryo selection”) rather than IES, and describe “in vitro human genetics” but without any mention of selection:
Germline genetic modification of humans, be it for the correction of disease mutations or genetic enhancement (for example, to confer disease resistance or increase height), will raise serious moral concerns for some. This technology may also facilitate the production of [substantially] larger quantities of eggs and, subsequently, embryos than current assisted reproductive technologies, vastly increasing the possibilities for embryo selection based on genetic profile. For example, if a couple is interested in selecting embryos for implantation based on multiple alleles, whether related to disease risk or phenotypic traits such as eye color, the potential mother’s PSC-derived eggs could be used to create hundreds of embryos, ensuring that all of the desired alleles are present together in at least one embryo.
…Finally, the ability to generate large numbers of human gametes (with random or designed genetic constitutions) will enable the practice of in vitro human genetics. That is, scientists will be able to conduct multigenerational human genetic studies in a dish, for example, to track the impact of various environmental conditions on the development of human disease or the impact of crossing specific genotypes. Such research may also facilitate the generation of ideal “universal donor” cells, with appropriate combinations of haplotypes at histocompatibility loci.
et al 2012 likewise appear to have only massive embryo selection in mind:
The ability to create large numbers of eggs or sperm through IVG greatly increases our capacity to select the best child possible. Selection could occur in two ways: (1) the most genetically desirable of this massive number of gametes could be selected and then used to create an embryo, or alternatively, (2) large numbers of embryos could be produced from these gametes and then the best embryo selected. Whatever the method, the advent of IVG could allow us to select for a much larger number of traits than is currently conceivable. IVG may also offer new possibilities for genetic enhancement. Cells could be modified prior to gamete formation, most probably once an ES cell culture has been established 25 However, in this paper we focus on possibilities for selection, rather than enhancement.
It is a relatively small step from the two papers’ mentions of massive embryo selection and et al 2009’s “in vitro human genetics” to “in vitro eugenics”, of course, and Sparrow seems to be the one to make it (independently of the previous two IES inventions). 2014 (responding to da et al 2014 ) mentions “A recent treatment of this topic by Shulman and Bostrom6 calls the same technology ‘iterated embryo selection’—a name that Matthews and Fujita et al may prefer.”, consistent with Sparrow independently inventing it.
Et Al 2018
et al 2018, “Efficient derivation of stable primed pluripotent embryonic stem cells from bovine blastocyst” (submitted to PNAS on 2017-09-13; February 2018 UC Davis press release) mentions as justification for livestock stem cell research:
The highly efficient derivation of bovine ESCs holds great potential for producing cattle with desired genetic value through genomic selection and/or genome editing as well as for in vitro breeding schemes through genomic selection, germ cell differentiation, and in vitro fertilization…The CTFR-bESCs described in this study were easy to derive from whole blastocysts, fast to obtain, highly efficient to establish, and easy to passage (single-cell dissociation using trypsin). These are all desirable features that will facilitate the creation of genetically superior cattle and the industrial production of valuable pharmaceuticals, as they allow efficient genomic selection through bESC derivation and facile genome editing and are amenable for NT cloning to generate live animals. Another potential use of these cells is the in vitro differentiation to gametes, facilitating in vitro breeding schemes that could result in multiple rounds of genomic selection, gamete production and fertilization, and bESC derivation to achieve genetically superior cattle within a [substantially] shorter generational intervals.
et al 2018 does not cite any sources or discuss “in vitro breeding” further, and the bibliography does not include any of the previous sources; the idea is treated as animal breeding ‘folklore’ —obscure enough to need to be explained, but not considered novel or necessary to reference thoroughly. The idea may have been circulating in animal breeding circles for a while before et al 2018 mentioned it. This appears to be the first appearance of the IVB term for IES specifically, suggesting that it doesn’t derive from either Haley & Visscher 199826ya or Shulman 200915ya, and is either a modification of Sparrow’s “in vitro eugenics” (since one might reasonably feel that ‘eugenics’ wouldn’t apply to a desire to apply it to animal breeding or is just too inflammatory) or another independent invention somewhere in the research circles of et al 2018’s authors.
Et Al 2018
et al 2018 in “In vitro breeding: application of embryonic stem cells to animal production” (apparently submitted before March 2018, as Alison Van Eenennaam in March 2018 posted a slide of Figure 1 on Twitter) proposes IVB in much more detail:
In this article, we present a strategy that will notably accelerate genetic improvement in livestock populations by reducing the generational interval, namely in vitro breeding (IVB). IVB combines genomic selection (GS), a widely used strategy for genetically improving livestock, with ESC derivation and in vitro differentiation of germ cells from pluripotent stem cells. We also review the most recent findings in the fields on which IVB is based. Evidence suggests this strategy will be soon within reach…In this article, we present and evaluate the benefits of a novel alternative to complement the GS methodology and increase genetic gain…In this context, we present a new strategy, in vitro breeding (IVB), which aims to combine these cutting-edge reproductive technologies with genomic selection to accelerate the genetic improvement of livestock populations (Figure 1).
…This type of combined strategy has been previously proposed in the early 1990s, prior to the implementation of the genomic technologies, which made possible the discovery of thousands of QTLs [16]. Such is the case of velogenesis, which initially proposed to grow, mature and fertilize prepubertal oocytes in vitro with sperm from selected progeny-tested bulls and to transfer them into postpubertal animals. The combination of velogenesis with marker-assisted selection gave rise to a new concept, velogenetics. This consisted of incorporating favorable mutations into a new genetic background by repeated backcrosses carried out by velogenesis. By the time the idea was conceived, the number of mapped QTLs was very limited, and markers were not available to explain a large part of the genetic variation for quantitative production traits. For this reason, the concept was initially conceived to be applied in cases such as disease resistance. Our strategy makes use of high throughput technologies like SNP arrays to determine hundreds of thousands of genotypes that might explain most of the genetic variance in production traits. It is also worth mentioning that the term “in vitro breeding” has been used for decades to refer to a number of in vitro plant breeding techniques, such as micropropagation, in vitro flowering, and in vitro pollination, among others [17]. Although these plant biotechnology techniques are grouped under the same name as our proposed strategy, the concept is considerably different.
While they cite Georges & Massey 199133ya for velogenetics and et al 2018 for recent work on stem cells, et al 2018 omits any mention or citation of Haley & Visscher 199826ya, Shulman 200915ya, or Sparrow 201311ya, and doesn’t mention et al 2018’s discussion of IVB. (Two of the authors, Jun Wu & Pablo J. Ross, appear on both & et al 2018.) So, appears to claim to invent both the idea & term IVB.
(As they note the term “in vitro breeding” is a bad choice as it is an established term in plant breeding dating back at least to 198935ya to refer to a wide array of manipulations going beyond IVF, such as changing polyploidyness of plants, mutagenesis, hybridizing them, using cryopreservation, micropropagation or micrografting plant embryos together, inducing flowering etc; for a recent discussion, see et al 2015 . It has also been used elsewhere in insect/animal research for still more things, such as maintaining parasites in the lab. It is unfortunate that they choose to use the term anyway.)
Et Al 2018
et al 2018 in “Revolutionize livestock breeding in the future: an animal embryo-stem cell breeding system in a dish”, propose (in July 2018) what they also call “in vitro breeding”, which they repeatedly describe as novel and claim to be introducing:
We propose a novel embryo-stem cell breeding system. Distinct from the conventional breeding system in farm animals that involves selecting and mating individuals, the novel breeding system completes breeding cycles from parental to offspring embryos directly by selecting and mating embryos in a dish…In vitro germline induction, together with subsequent in vitro fertilization (IVF) and ESC derivation, has successfully created new individuals and alternation of generations, thereby reconstituting an entire mammalian life cycle in vitro (Figure 1b.). This rapidly renewed life cycle can be used to constitute a recurrent animal breeding cycle by selecting and mating embryos directly, ie. IVF using PSC-derived gametes. Thus, we propose a novel animal breeding system termed animal embryo-stem cell breeding system that can revolutionize the design and implementation of current breeding programs in livestock…The embryo-stem cell breeding system has substantial advantages compared to the conventional breeding system, especially in shortening generation interval, increasing the number of female monotocous offspring, and selection intensity. Taking breeding dairy cows as an example, ideally, our envisioned system is expected to be 30–40× more efficient in comparison to the conventional system, meaning that 1-year genetic gain of in vitro breeding can be the same as that of 30–40 years of conventional breeding.
There are no citations or discussions of the previous proposals like Hayes & Visscher 199826ya, Shulman 200915ya, Sparrow 201311ya, or et al 2018 (and also no author overlap) which might explain a key difference which makes their version novel, but there are 2 citations to et al 2018, which, however, like , they do not mention as previously giving both the idea & name for their proposed novel system.
-
Haley & Visscher 199826ya summarize Georges & Massey 199133ya:
Georges and Massey (8) took this idea to the extreme by suggesting “velogenetics”. This scheme uses “velogenesis” (3) in which the generation interval is greatly reduced by harvesting oocytes from calves while still in utero. The harvested oocytes are matured and fertilized in vitro prior to being transferred to a recipient female. The process can be repeated by harvesting oocytes from these second generation animals with the generation interval being reduced to around 3–6 mo. In velogenetics, markers are added to such a scheme either to select the in utero calves from which to harvest oocytes or to type one or two cells from embryos while they are still in vitro to determine which to transfer to recipient females (Figure 5a). In principle, this procedure could be used for several successive generations, with each generation using markers for selection without ever generating an adult animal or measuring a phenotype. Hence, velogenetics might be used to introgress rapidly a gene from one breed to another using semen from the recipient line each generation.