Diagnosis of genetic disease in preimplantation embryos

Preimplantation genetic diagnosis today Alan H.Handyside Institute of Obstetrics and Gynaecology, Royal Postgraduate Medical School, Hammersmith Hospital, Du Cane Road, London W12 ONN, UK

Introduction The use of assisted reproduction techniques allows preimplantation genetic diagnosis of an inherited disease in early human embryos before fertilization. Some genetic defects have even been identified before fertilization by analysis of the first polar body of the oocyte (Verlinsky et al., 1990, 1992). By selective transfer of only unaffected embryos, couples at risk of having affected children can avoid the possibility of terminating an affected pregnancy following more conventional diagnostic approaches later in gestation. Clinical experience is still limited but worldwide approaching 200 hundred cycles have now been reported resulting in 50 pregnancies and over 30 births (Table I) (Harper, 1996).

Collecting and biopsying human embryos Attempts to flush embryos from the uterus following superovulation and conception in vivo have so far failed. In any case, there would always be a risk that a potentially affected embryo remained in the uterus. Clinical application of preimplantation genetic diagnosis has therefore exclusively used the methods of superovulation and in-vitro fertilization (IVF) established for the treatment of infertile couples. This produces a number of normally fertilized embryos in a single cycle accessible for genetic screening. Most couples are known to be fertile since the previous birth of an affected child is often the reason they are aware of the risk in future pregnancies. Nevertheless, the use of established ovarian stimulation protocols following down-regulation with gonadotrophinreleasing hormone (GnRH) analogues is appropriate because the probability of an embryo being affected is high in many cases. Maximizing the number of embryos screened increases the likelihood of establishing a pregnancy. Human embryos have been successfully biopsied at cleavage stages on day 3 post-insemination, at the 6-l0-cell stage, and at the blastocyst stage on days 5 or 6 (Dokras et al., 1990, 1991; Muggleton Harris et al., 1995; Pickering and Human Reproduction Volume 11 Supplement 1 1996 © European Society for Human Reproduction and Embryology

139

A.H.Handyside Table I. Summary of world clinical experience with preimplantation genetic diagnosis following in-vitro fertilization (IVF) and cleavage stage biopsy to February, 1995" Patients

Cycles

Transfers

Pregnancies

Babies

X-linked (PCR)

41

62

53

14

II

X-linked (FISH)

49

70

56

15

II

Single gene defects Total

59

65

62

21

12

149

197

171

50

34

PCR = polymerase chain raction; FISH = fluorescence in-situ hybridization. Pregnancy rate = 25% per cycle, 29% per embryo transfer. Results collated from 14 centres. "Modified from Harper, (1996).

Muggleton Harris, 1995). Blastocyst biopsy has the advantage that a larger number of cells can be removed from the outer trophectoderm layer without affecting the inner cell mass from which the fetus later develops. However, too few embryos reach this stage and implant after transfer to be clinically viable for preimplantation genetic diagnosis at the present time. Removal of one or two cells at the equivalent of the 8-cell stage does not appear to affect development to the blastocyst stage in vitro (Hardy et aI., 1990) and has proved to be highly efficient in practice (Ao and Handyside, 1995). The procedure involves dissolving a hole in the zona pellucida using acidified Tyrode's solution (Figure 1) and then aspirating the cells with a second larger micropipette. Also, co-culture of the isolated cells with the biopsied embryo supports limited proliferation and may provide an alternative to the more difficult procedure of blastocyst biopsy (Geber et al., 1995). So far there have been no reports of serious abnormalities at birth following cleavage stage biopsy. Human embryos, like other mammalian embryos, appear to be able to regulate their development at these early stages presumably because cells are not yet irreversibly committed to specific fates.

Genetic analysis The genetic analysis of single cells has been made possible by the development of DNA amplification methods involving the polymerase chain reaction (PCR) and fluorescence in-situ hybridization (FISH) for rapid cytogenetic analysis of interphase nuclei. Single cells biopsied at cleavage stages are therefore prepared for analysis either by carefully placing the cells in lysis buffer in micro centrifuge tubes for PCR or by spreading the nucleus on a microscope slide for FISH.

Single cell analysis by peR To amplify sufficient DNA from a single cell for analysis by conventional gel electrophoresis, two rounds of amplification are necessary. 'Nesting' the second 140

Preimplantation genetic diagnosis

Figure 1. Scanning electron micrograph of a human cleavage stage embryo following zona drilling with Acid Tyrodes and biopsy to remove one of the blastomeres. Note the apparently double layered appearance of the zona and the extensive thinning of the outer layer. (Courtesy of Dr George Nikas).

pair of oligonucleotide primers, which define the region to be amplified, internally to the first pair has several advantages and provides a safeguard for product contamination carried over into the sample tubes (Figure 2). Theoretically, only a single molecule could be amplified and give an erroneous result. Following amplification, many approaches have been used to identify the presence of different mutations (Table II). Since the aim is to transfer selected embryos on the same day as the biopsy to maximise pregnancy rates, amplification and mutation detection have to be completed in 8-12 h. Pregnancies and births confirmed to be free of the inherited disease have been established in couples at risk of several common single gene defects including cystic fibrosis (CF) (Handyside et aI., 1992), Duchenne muscular dystrophy (DMD) (Liu et aI., 1995) and Tay Sachs disease (Gibbons et aI., 1995). Preimplantation genetic diagnosis has also been achieved for a specific mutation causing the rare X-linked condition, Lesch-Nyhan syndrome (E. Hughes, P.P. Ray, R.M. Winston and A.H. Handyside, unpublished). With CF and Tay Sachs, the common three base pair (bp) deletion and 4 bp insertion respectively, were detected by heteroduplex formation (Figure 3). A 3 or 4 bp size difference in the amplified fragment cannot be reliably discriminated by rapid gel electrophoresis. By mixing the amplified DNA from the single cell with DNA previously 141

A.H.Handyside

01

____ _ __'____Sl.¥.._ _ _ __

-----

------2S----~--

1 -20 cycles 102 11~

~

12

130 cycles 1 Figure 2. Diagrammatic representation of polymerase chain reaction (peR) using nested oligonucleotide primers. In the first amplification reaction, an outer pair of primers (0 I and 02) annealing to the sense and antisense strands of the genomic target sequence encompassing the mutation (tIiangles) is pm1ially amplified. In the second reaction, a second set of 'nested' pIimers (II and 12) annealing to the target sequence internal to that of the outer primers is used to amplify a smaller fragment from an aliquot of the first amplified product. Two rounds of amplification is necessary in most cases to amplify sufficient DNA from single cells for conventional analysis. The use of nested primers reduces non-specific amplification. It is also a safeguard to prevent carry-over contamination of samples tubes with the final amplification product since the outer primers should not anneal to this sequence.

Table II. Some methods for detecting mutations in amplified fragments for preimplantation genetic diagnosis Method

Typical application

Fragment length differences

Distinguishing gene and intronless pseudogene sequences Analysis of variable number tandem repeats

Heteroduplex formation

Rapid detection of small insertions or deletions

Restriction digestion

Detection of mutations that eliminate restriction sites

Allele specific oligonucleotides (ASOs)

Hybridization of normal and mutant ASOs detected by nonradioactive methods

Single strand conformation polymorphism (SSCP)

Detection of majority of mutations anywhere within the amplified fragment

Oligonucleotide ligation

Direct detection of point mutations

Microsequencing

Mutation detection by incorporation of radio labelled nucleotides-avoids gel electrophoresis

amplified from homozygous normal or affected individuals, denaturing the double stranded DNA fragments and allowing the mixtures to cool slowly, heteroduplexes are formed in those cases where both normal length fragments and mutated longer or shorter fragments are present. When these mixtures are electrophoresed, 142

Preimplantation genetic diagnosis CellI

~ + NN AA +

Cell 2

~

+

NN

Cell 3

~

+

+

+

AA

NN

AA

Cell 4

~

+

NN

+

AA

-- -------NN

NA

Amp failure

Figure 3. Diagrammatic representation of the use of heteroduplex formation for the rapid detection of, in this example, the common 3 base pair (bp) deletion, L'1F508, causing cystic fibrosis, Amplified product from each of four single cells is mixed with amplified product from either known homozygous normal (NN) or affected (L'1L'1) individuals. The pattern of heteroduplex bands obtained with the two mixtures allows all three possible genotypes as well as amplification failure to be identified on a small polyacrylamide gel run for as little as 30 min.

the heteroduplexes are significantly retarded on the gel and can be reliably interpreted for diagnosing the genotype of the cell biopsied from the embryo.

Single cell analysis by FISH Pioneering studies of the karyotype of early human embryos following IVF demonstrated a high incidence of chromosomal abnormalities and mosaicism (Angell et aI., 1986; Plachot et aI., 1987; Jamieson et aI., 1994). The low incidence of metaphases, with or without the use of spindle inhibitors, and the difficulty of banding the short chromosomes at these stages, however, make it impractical to use this approach for diagnostic purposes and prevents analysis of all nuclei. FISH is equally applicable to both interphase and metaphase nuclei and the use of multicolour probes allows the detection of more than one probe simultaneously. Dual FISH with X and Y specific probes have been used clinically for identifying the sex of embryos in couples at risk of X-linked diseases which only affect boys (Griffin et al., 1992, 1993, 1994). This approach has now been further refined using directly labelled probes and an additional autosomal probe to distinguish sex chromosome aneuploidy from abnormal ploidy. Combined with a new spreading procedure involving dissolving the cytoplasm of the single cells and attaching the nuclei to poly-L-lysine-coated slides, it is now possible to complete the analysis within 2 h (Harper et aI., 1994). In addition to identifying sex in X-linked disease, FISH analysis is being used for detecting trisomies in couples in which one partner is carrying a reciprocal or Robertsonian translocation. These couples are often at high risk of trisomic pregnancies and many have repeated miscarriages. These couples are good candidates for preimplantation genetic diagnosis since several embryos can be screened in a single cycle. The usual probes specific for repeated sequences in 143

A.H.Handyside

the centromeric region of particular chromosomes may not be suitable and more distal unique sequence yeast artificial chromosome (YAC) or cosmid contig probes are being developed. Combinations of probes, specific for the same chromosome, can also be used for accurate detection of particular trisomies, for example, in women believed to be gonadal mosaics (c. Conn, J. Harper and J.D.A. Delhanty, personal communication). Analysis of spare human embryos or those rejected for transfer following FISH analysis of the sex chromosomes has confirmed the relatively high incidence of aneuploidy (-25%) at conception (Munne et al., 1993). There also appears to be an unexpectedly marked increase with maternal age. In some studies this increased to almost 50% above the age of 40 (Munne et al., 1995). As the majority of aneuploidies arise during maternal meiosis especially meiosis I, first and/or second polar body analysis with combinations of probes detecting X, Y, 21, 18 and 13 are being used to screen embryos in older women undergoing IVF (Verlinsky et al., 1995). If sufficient euploid embryos can be identified this may improve pregnancy rates and decrease miscarriage rates in these women. Analysis of all nuclei in individual embryos at cleavage stages has confirmed that many embryos are chromosomally mosaic. In addition to aneuploidies arising during gametogenesis and abnormalities arising at fertilization, it is clear that there is also a high incidence of postzygotic abnormalities (Delhanty et al., 1996). Most of the nuclei involved, however, have an abnormal ploidy. For example, tetraploid nuclei are relatively common although haploid and triploid nuclei have also been detected. The origin of these cells is not known but it is possible that in some cases it results from delayed fertilization or pronucleus formation by supernumerary spermatozoa for example. Non-disjunction in early cleavage has been observed but is relatively uncommon. Finally, some embryos have apparently chaotic chromosomal complements in almost every nucleus and this appears to be associated with particular patients as it is consistently observed in successive cycles of IVE Human embryos like those of lower vertebrates and invertebrates, therefore, may lack the normal cell cycle checkpoints and accumulate chromosomal errors during cleavage (Delhanty and Handy-side, 1995).

The accuracy of preimplantation genetic diagnosis The accuracy of preimplantation genetic diagnosis remains to be assessed in clinical practice. Three misdiagnoses have already been reported (Harper and Handyside, 1994). However, these involved different diagnostic procedures and probably occurred for different reasons. All five children resulting from a recent series of 18 preimplantation genetic diagnosis cycles for the predominant ~F508 deletion causing CF were confirmed to be homozygous normal (Ao et al., 1995). However, analysis of each blastomere from the embryos which were not transferred did reveal some errors particularly in amplifying both parental alleles in heterozygous carrier cells. Often one parental allele failed to amplify, apparently randomly (allele dropout), resulting in an incorrect diagnosis as homozygous 144

Preimplantation genetic diagnosis

normal or affected. Further work with single heterozygous lymphocytes has now shown that this phenomenon is partly explained by incomplete denaturation of the genomic template DNA during the initial cycles of PCR (Ray and Handyside, 1996). Raising the temperature in the initial cycles improves the efficiency of denaturation and minimizes but does not eliminate allelic dropout (Figure 4). Fortunately in an autosomal recessive condition, allele dropout cannot cause a serious misdiagnosis leading to the transfer of a homozygous affected embryo (at least in these couples in which both partners were carrying the same mutation). For compound heterozygote detection or dominant conditions, however, this could theoretically occur in half the cases of allele dropout in which the affected allele did not amplify. One way of avoiding these errors for diagnosis of compound heterozygotes is to amplify a single DNA fragment encompassing both mutations. This has recently been demonstrated for ~-thalassaemia: major diagnosis with single lymphocytes from a boy who is a compound heterozygote for two common Indian mutations of the ~-globin gene (Ray et ai., 1996). A 208 bp fragment including ex on 1 and part of intervening sequence 1 encompassing 12 of the common ~-thalassaemia mutations and also the mutation causing sickle cell disease was amplified with high efficiency from single cells (Figure 5). Allele dropout did occur in a minority of cells but in these cases the genotype appeared to be homozygous at one locus for the normal allele and at the other for the mutation. Since this is an impossible combination of parental alleles the possibility of an error is avoided (Figure 6). Clinical application of preimplantation genetic diagnosis for couples at risk of ~-thalassaemia, many of whom carry different mutations, should therefore be possible using this approach. Four general sources of errors have been identified with single blastomere analysis at cleavage stages (Table III). However, careful monitoring of FISH or amplification efficiency and contamination levels and reducing allelic dropout I