News of the "deaf project"

Editing continues

Daria Spasskaya, N+1

For links, see the original article – VM.

Two years ago, the world scientific community actively discussed the intention of the Russian geneticist Denis Rebrikov to edit a human embryo in order to cure an unborn child from hereditary deafness. Then Rebrikov was already recruiting subjects for his experiment, although of all the proofs of the effectiveness of his procedure, he had only a publication about experiments on defective embryos, where he edited another gene. Today, his team at the Moscow conference on genomic sequencing and editing for the first time reported on the progress of work on the "deaf" project. We studied the presentation from the conference and talked with the speaker, Andrey Kriv, a researcher at the genomic editing laboratory of the Kulakov Center for Obstetrics, Gynecology and Perinatology.

The disease in question, autosomal recessive nonsyndromic deafness, is caused by a mutation in the GJB2 gene. The gene encodes a protein that is needed to establish contacts between neurons in the inner ear. In its absence, the vibrations of sensitive hairs do not lead to the generation of a signal that would be transmitted further to the brain – so a person loses the ability to process sound signals. In Russia, according to Rebrikov, the carrier of such a mutation (specifically, the loss of one letter at the beginning of the gene – 35delG) is every thirtieth.

However, in the heterozygous state – when the mutation in the GJB2 gene is only on one of the two copies of the chromosome – it is not dangerous. Serious hearing disorders occur when there is a mutation on both chromosomes, which is possible when a child is born to two carriers. The chance of such a pair of parents appearing "in a vacuum", it would seem, is small – but in reality they occur much more often because deaf people communicate and create families within their community. And if most hereditary diseases can be "caught" if desired and eliminated during IVF, since the parents are likely to be heterozygous for a harmful mutation, in the case of two deaf parents, it is impossible to select a "wild" embryo according to GJB2. That is why Denis Rebrikov stopped at this model, considering it ethically justified.

In the laboratory of Rebrikov, who heads the department at the Center of Obstetrics and Gynecology named after him. Kulakova, there have already been good developments in working with embryos – in 2018, scientists published the results of introducing a mutation in the CCR5 gene into the human zygote, protecting its owner from HIV infection. Then geneticists reported that five of the eight edited blastocysts contain the desired mutation. Making a deletion in CCR5 is a fairly well–studied model that was worked out even before the spread of CRISPR with other genetic editors. In particular, before the experiments of the Rebrikov group, Chinese researchers tried to introduce it into embryos, and some of them went so far as to even get edited children, although not very successfully.

A year and a half ago, discussing the correction of a mutation in the GJB2 gene in human embryos, Denis Rebrikov assured that the system was almost ready for this, and that there was literally a month left before its tests, and he even found a couple ready to give birth to an edited child. The main problems were expected from the regulators (Ministry of Health), since, unlike Jiankui He, the experiment was planned to be conducted strictly officially. Apparently, scientists expected that, by analogy with CCR5, there should be no special problems with editing itself.

To make an edit to the embryo genome using the CRISPR/Cas system, you need to choose an effective guide RNA against the desired gene, mixed with Cas9 enzyme and matrix to eliminate the mutation. After introducing the above mixture into the zygote, the RNA guide should "drive" to the target Cas9 gene so that it cuts the DNA double helix at this point. Then the cell repair system should come into play – and in order for it to patch up the hole in the genome in the way we need, a "patch" is used, a matrix with genetic material that you want to substitute instead of the target fragment. The output remains a corrected chromosome and a healthy embryo. This is how it looks in theory.

And what are the problems there?

In practice, a lot can go wrong:

• The RNA guide may contact the wrong place and an unplanned (off-target) mutation will appear on the chromosome;

• the repair system will not pay attention to the matrix, and some other mutation will occur at the site of the incision as a result of non-specific "stitching" of the free ends of DNA;

• while the editing system is running, the embryo will divide. A scenario is possible in which only one copy of the chromosome will be corrected in some cell, and the second one will remain unedited, and because of this, not all cells of the embryo will carry the changes we need (mosaicism).

To move from experiments to transplanting an edited embryo into the womb, you need to be sure that none of the above has happened. This can only be verified by sequencing, during which the cell is destroyed. If the whole embryo is sent for sequencing, there can be no question of any transplantation. You can extract several cells from the embryo (preferably as few as possible, because the whole body of the future person is built from them) and sequence them. But in order to trust the results of such an analysis, it must first be convincingly shown that sequencing of individual embryo cells reliably predicts the state of all others. To do this, it is necessary to isolate DNA from the entire embryo in pilot experiments and sequence the edited section.

So far, no one has created a CRISPR system that is fully insured against all the listed incidents. Therefore, editing of human embryos is still prohibited in most countries. In Russia, this issue still remains in the "gray zone" of legislation, although in the fall of 2019, the Ministry of Health separately commented on Rebrikov's initiative to switch to experiments with embryo transplantation to mothers, saying that it was "premature" to issue a permit for such a thing and referred to the fact that he fully agreed with WHO, whose position is strictly "against".

In addition, recently geneticists who work with CRISPR/Cas have a new concern. It turned out that the incisions made by Cas9 can lead to the loss of large sections of one of the copies of chromosomes. As a result, when analyzing DNA, only the remaining copy is read and a false impression arises that the cells are homozygous at this locus (for this reason, this event is called loss of heterozygosity). For example, in one of the recent articles in PNAS, the probability of restructuring in one of the edited loci was estimated at 16 percent. The point here is not the accuracy of the nuclease – after all, the loss occurs in the right place, but with the peculiarity of the repair proteins, which in some cases destroy the DNA around the incision in order to synthesize it again later. To account for the loss of heterozygosity, it is necessary to develop alternative methods of genome analysis, such as analysis of the number of copies of genes.

Attempt #1: CCR5

In 2018, Rebrikov's team, working on introducing the del32 mutation into the CCR5 gene, achieved the following:

1. The targeted replacement in the genome occurred with an efficiency of 60 percent (five out of eight embryos were homozygous for mutation), and only two out of eight successfully edited embryos turned out to be mosaics. That is, the repair systems used the matrix very effectively, although this is rather an exception to the rule – the efficiency of repair by this mechanism in mice is estimated at 5-20 percent, and even less in human cells, and most likely strongly depends on the gene – its DNA sequence and activity level. In a 2017 article, Shukhrat Mitalipov's group, which also edited embryos using CRISPR/Cas, reported that although they got the target gene in 42 out of 58 cases, the DNA patch for the MYBPC3 gene did not work for them.

2. The presence of non–target mutations was not mentioned in the article at all - although scientists noted that they were going to look for them in the future using high-performance DNA sequencing systems by comparing them with the genomes of parents. Scientists had all the necessary capacities for this.

Therefore, when Rebrikov's group changed its goal from CCR5 to GJB2, first of all, it had to deal with the questions that were asked for their previous work, and in addition, take into account the problem with the loss of heterozygosity that also surfaced.

Attempt #2: GJB2

The project began with the selection of guide RNA and testing it on connective tissue cells. However, there is no question of correcting the 35delG mutation yet: scientists do not even have suitable material for experiments. Instead, they are still trying to make "hearing" cells "deaf", that is, to introduce the 35delG mutation into the GJB2 gene.

At this stage, geneticists were able to select an RNA guide that would lead to cutting GJB2 with an efficiency close to 100 percent. With him, the scientists moved on to the experiment on zygotes.

Five days after the introduction of the editing cocktail into the zygote, the resulting blastocysts were sent for genetic analysis. It was planned not only to look for the presence of a mutation in GJB2, but also to sequence possible off-target sites, look for the presence of chromosomal rearrangements and check large areas around the incision site to detect the loss of heterozygosity. And to check the mosaic of embryos, it was planned to compare the results of sequencing of several embryo cells with data on the whole blastocyst.

According to Krivoy, currently the team's efforts are focused on the development of methods of genetic analysis of embryos. In his report, he presented the results of the so-called CNV analysis (copy number variation), which, after sequencing the genome by the representation of signals from its different regions, allows us to estimate the number of copies of sites larger than 50 nucleotides. In this way, it is possible to track deletions that occur only on one of the two chromosomes.

1. According to the first results of editing, it is really possible to make changes to the GJB2 gene quite effectively. Analysis of several embryos showed that small deletions corresponding to the event of nonspecific suturing of the incision are present both in the selected cells and in the whole blastocyst. A variant of the wild-type gene was not found, which indicates the absence of mosaicity.

2. But it has not yet been possible to force the repair systems to use a DNA patch - the effectiveness of the embedding turned out to be extremely low.

3. CNV analysis showed the presence of a much bigger problem – in all embryos, either one copy of the 13th chromosome was missing, or a significant fragment of it containing the GJB2 gene. According to Krivoy, this is due to the poorly selected concentration of cocktail components in the first experiment, and this was avoided in the second round of editing. Nevertheless, this fact shows that experts' concerns about the lack of development of the technology do not arise from scratch.

4. Checks for off-targets in new experiments, as well as when working with CCR5, have not yet been carried out.

Total

After switching from the CCR5 gene to GJB2, things are still going worse for the Rebrikov group:

• It has not yet been possible to achieve effective embedding of the DNA patch, although this mechanism worked well in the last project;

• there was a problem with the loss of the 13th chromosome;

• scientists have so far ignored the problem of inappropriate editing.

Krivoy says that the group will continue to select new conditions, change the concentration and chemically modify the DNA patch in order to increase the efficiency of its embedding, and will also analyze DNA for off-targets.

If in the last CCR5 project the team worked with a well-known model (moreover, it clearly did not check the obtained embryos for mutations enough to talk about success), no one except Russian scientists has tried to edit the human GJB2 gene yet (although last year the Chinese inserted a whole mutant human GJB2 into a pig).

In fact, the team is still at the very beginning of the path to therapeutic editing of deafness, the interlocutor N+1 admits. He estimates the time for bringing the system to mind and security checks to be two years.

However, it seems that they have nowhere to hurry – according to Denis Rebrikov, they currently do not have anyone willing to edit the child right now, as there are no GJB2 mutant donor eggs. However, it cannot be said that scientists are working in vain – mastering the subtleties of genome editing and genome-wide DNA analysis, exploring the "pitfalls" of technology cannot be called a waste of time.

Portal "Eternal youth" http://vechnayamolodost.ru