Runs of homozygosity in spontaneous abortions from families with recurrent pregnancy loss

Recurrent pregnancy loss (RPL) is a severe reproductive pathology with a significant component of unexplained etiology. Extended homozygous regions as a possible etiological factor for RPL were sought in the genomes of embryos. Twenty-two paired first-trimester spontaneously aborted embryos from eleven women with recurrent miscarriage were analyzed. All embryos had normal karyotypes according to metaphase karyotyping and conventional comparative genomic hybridization. SurePrint G3 Human CGH + SNP 4 × 180K microarrays (Agilent Technologies) were used to search for homozygous regions. As a result, 39 runs of homozygosity (ROH) were identified in extraembryonic tissues of 15 abortuses. Verification of recurrent homozygous regions was performed by Sanger sequencing. The presence of occasional heterozygous SNPs was shown in 25 extended ROHs, which may indicate that they did not arise de novo but were inherited from parents. In the course of inheritance in a series of generations, they may accumulate mutations, leading to heterozygosity for several sites in the initially homozygous population-specific regions. Homozygotization of recessive mutations is one of the putative mechanisms of the influence of such inherited ROHs on RPL development. The high frequency of extended ROHs detected in the present study may point to a role of inbreeding in RPL etiology. Homozygous regions may also occur due to uniparental disomy, and abnormalities of genomic imprinting may be another mechanism responsible for the pathological manifestation of ROHs in embryogenesis. Indeed, five predicted imprinted genes were identified within ROHs according to the Geneimprint database: OBSCN, HIST3H2BB, LMX1B, CELF4, and FAM59A. This work reports the first finding of a high frequency of extended ROHs in spontaneously aborted embryos with normal karyotypes from families with RPL.


Introduction
Spontaneous abortions occur in approximately 15 % of pregnancies in different populations; however, in some families, pregnancy loss occurs more than once (Nikitina et al., 2016).According to the guidelines of the European Society of Human Reproduction and Embryology, a diagnosis of recurrent pregnancy loss (RPL) should be considered after the loss of two or more pregnancies (Goddijn, 2017).The prevalence of RPL varies from 2 to 5 % of the total number of pregnancies (El Hachem et al., 2017).Recurrent pregnancy loss may be caused by uterine abnormalities, parental chromosomal aberrations, antiphospholipid antibodies, polycystic ovarian syndrome, diabetes mellitus, and hyperthyroidism.In about half of all cases, the RPL cause remains unexplained (Nikitina et al., 2016).
Homozygotization of recessive mutations due to runs of homozygosity (ROH) or uniparental disomy (UPD) may be one of the causes (Robberecht et al., 2012;Niida et al., 2018).In a consanguine marriage with manifestations of recessive diseases, the homozygous risk locus is likely to fall into an extended homozygous region.Thus, pathogenic recessive variants can be identified in affected individuals from inbred population by identifying ROHs.This approach to the search for pathological mutations made it possible to map genes associated with many recessive Mendelian diseases (Botstein, Risch, 2003).
The emergence of regions with ROHs may result from inbreeding, but ROHs are found in human genomes even among outbred populations (Ceballos et al., 2018).Population history and cultural factors may influence homozygosity levels, and therefore ROHs are characterized by population specificity, while data on their prevalence in Russian populations are unavailable yet.
The aim of this paper was to study the role of homozygotization of various genome regions in the pathogenesis of RPL.

Materials and methods
Patients supervised by the Laboratory of Cytogenetics of the Research Institute of Medical Genetics (Tomsk, Russia) from 1987 to 2018 were analyzed.Material from first-trimester spontaneously aborted embryos was obtained from gynecologic and obstetric clinics in Tomsk and Seversk (Russia).Information was recorded about the maternal and paternal age, gynecological anamnesis of the women, the number and outcomes of previous pregnancies, and features of the present gestation.
Tissue samples obtained by curettage were transferred to the cytogenetic laboratory in sterile saline.The products of conception were examined, and extraembryonic tissues were separated from decidua and blood clots.All abortuses had normal karyotypes according to the results of metaphase karyotyping and conventional comparative genomic hybridiza tion of extraembryonic tissues.For conventional karyotyping, metaphase chromosomes were obtained after long-term culture of fibroblasts of extraembryonic mesoderm in DMEM/ F12 (1:1) medium (Sigma, United States) supplemented with 20 % fetal bovine serum (HyClone, United States).Colchicine (Sigma, United States) was added 4 h before chromosome harvesting, and the samples were processed by standard techniques.All specimens were G-banded with trypsin-Giemsa (Sigma, United States) to identify chromosomes.
For conventional comparative genomic hybridization (cCGH), test and reference DNA were labeled by nick translation (Rooney, Czepulkowski, 1992).Hybridization of the DNA libraries on metaphase plates derived from peripheral blood lymphocytes of a healthy male was carried out with suppression with 50-fold excess C 0 t-1 DNA (Agilent Technologies, United States) for 72 h at 37 °C in a Thermo-Brite hybridization chamber (Abbott Molecular, United States).Metaphase chromosomes were stained with DAPI.Hybridization signals were detected using Axio Imager Z2 fluorescence micro scope (Carl Zeiss, Germany) with a set of corresponding filters.The cCGH results were processed with Isis CGH software (Metasystems, Germany).
A search for homozygous regions was performed with SurePrint G3 Human CGH + SNP 4 × 180K microarrays (Agilent Technologies) in the extraembryonic tissues from 22 paired spontaneous abortions with the normal karyotype from 11 women with RPL.Results were visualized with Cytogenomics software (Agilent Technologies).Default Analysis Method (CGH+SNP v2) was used with a LOH threshold of 6.0.The resolution of this platform for LOH detection is about 2 Mb.
Some recurrent homozygous regions were verified by Sanger sequencing.To confirm the status of heterozygosity, 18 single nucleotide polymorphism (SNP) markers located in the analyzed chromosomal regions were chosen.Criteria for the choice of markers: the level of heterozygosity in Caucasians close to 0.5 and uniform distribution within assumed homozygosity regions.DNA sequences flanking the analyzed SNPs (155 to 350 bp for different markers) were analyzed for each marker.In the presence of other polymorphic variants in the analyzed fragments, their heterozygosity status was also taken into account.Sequencing was performed on an Applied Biosystems 3730 automated analyzer with the BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, United States) according to the manufacturer's protocol.Data analysis was performed with Sequencing Analysis v5.4 software.
The analysis of the gene composition was carried out with the "WEB-based GEne SeT AnaLysis Toolkit" program based on overrepresentation enrichment analysis (ORA) (http:// www.webgestalt.org).Gene categories were classified ac-

Results and discussion
Thirty-nine homozygous regions were identified in extraembryonic tissues of 15 embryos (Table 1).The average size of the identified regions was 4.6 Mb (from 2.0 to 9.6 Mb).
Due to the limitation of the resolution of the used platform, adequate comparison of the results with population-specific data available in the literature is difficult.However, comparisons were made with studies in which identified ROHs were ranked by size (McQuillan et al., 2008;Pemberton et al., 2012).It was shown that extended ROHs (>2.5 Mb) are not typical for unrelated individuals and for populations with low levels of inbreeding.(About 20 % of samples had extended ROHs.)In contrast, isolated endogamous populations had a high frequency of extended ROHs (~80 %) (McQuillan et al., 2008).The frequency of extended ROHs in our samples was 68.1 % (15/22).In addition, multiple extended ROHs were detected in 40 % of the samples (9/22).It is possible that the high frequency of extended ROHs is associated with a possible role of inbreeding in the etiology of RPL.
When comparing the repeated homozygous regions found in this study with the most frequent ROHs in other populations, no matches were found (Nothnagel et al., 2010;Pemberton et al., 2012).This may be indicative of a specific prevalence pattern of ROHs in the Siberian population or result from a strong sample bias in comparison with general populations.For more detailed analysis, characterization of Russian populations by ROH structure, location and frequency of occurrence is required.
Samples with recurrent ROHs were selected for verification by Sanger sequencing (Table 2).As a result, occasional heterozygous SNPs were identified in some regions.Detailed analysis of the results of the microarray study also revealed occasional heterozygous SNPs within the homozygous regions.Out of 39 regions of homozygosity, occasional heterozygous SNPs were present in 25 cases (64 %), whereas only homozygous SNPs were identified in 14 spontaneous abortions (36 %).The presence of heterozygous sites in extended homozygous regions suggests that these chromosome regions are not the result of de novo loss of heterozygosity.Such homozygous regions may occur when identical haplotypes are inherited from each parent (Peripolli et al., 2017).Mutations may accumulate in them in a series of generations, leading to heterozygosity in some sites of initially homozygous regions.
The information on clinical effects of ROHs or their associations with phenotypic traits is scarce.One of the possible  (Yang et al., 2012;Gamsiz et al., 2013;Christofidou et al., 2015;Ghani et al., 2015).Most of the ROHs in our study contained genes; the average number of genes per ROH was about 30.The region 22q12.3-q13.31 of 8 Mb in size contained the largest number of genes, 164 (sample 4b).There were no genes in three spontaneous abortions with corresponding homozygous regions (samples 4b, 10a, and 10b), no known genes were identified.All regions in these samples were mapped to 11p12.
To study the overrepresentation of groups of genes by biological functions among all genes localized in ROHs, enrichment analysis was carried out.As a result, a group of genes involved in the metabolism of flavonoids was identified (GO: 0009812, FDR = 0.0001).This functional group includes mainly genes from the UDP-glucuronosyltransferase family (UGT2B11, UGT2A1, UGT2B28, UGT2B4, UGT2B7, UGT2B10, UGT2B15, UGT2B17, UGT2A3, SULT1B1).However, a more detailed analysis showed that all genes from this group mapped to the same region 4q13.2-q21.1 in only one sample, 4b.Consequently, the overrepresentation of this group of genes is due to their nonrandom occurrence within one ROH region but not systemic overrepresentation in different samples.The absence of common pathogenetic pathways for different samples may be indicative of the role of homozygosity for different genes, which is quite possible in the light of the extended size of ROHs and the number of affected genes.
As mentioned above, in addition to regions with occasional heterozygous SNPs, regions containing only homozygous SNPs were identified.Perhaps such sites arise from de novo mitotic recombination, that is, UPD occurs.Papenhausen et al. (2011) showed that about 31 % of the samples (29/92) with homozygous regions had appeared from UPD.In our study, 36 % of regions (14/39) without heterozygous SNPs were identified, and they might have been generated by de novo mitotic recombination.Imprinting abnormalities can be one of the mechanisms by which such UPD regions contribute to the etiology of miscarriage.When analyzing the gene composition according to the Geneimprint database, five predicted imprinted genes were found: OBSCN, HIST3H2BB, LMX1B, CELF4, and FAM59A.

Conclusion
This is the first study to demonstrate a high frequency of ex tended ROHs in spontaneous abortions with a normal karyo type from families with RPL.Future studies should be dedicated to further analysis of the prevalence of ROHs in Russian populations and consideration of this phenomenon from the viewpoint of association with common diseases.

Table 1 .
Runs of homozygosity in spontaneous abortions from families with recurrent pregnancy loss List of ROHs

Table 2 .
Genotypes for highly polymorphic SNPs in recurrent ROHs analyzed by Sanger sequencing mechanisms of the influence of ROH on the pathogenesis of RPL may be homozygotization of recessive mutations significant for early embryo development.Some studies demonstrate that ROHs are significantly associated with many common diseases, such as Alzheimer's disease, rheumatoid arthritis, autism, and coronary artery disease