Advancements in Pathogenic Genes and Biomarkers for Non-syndromic Cleft Lip With or Without Cleft Palate Via Multiomics

Abstract

Non-syndromic cleft lip with or without cleft palate (nsCL/P) is a common congenital malformation influenced by a combination of environmental and genetic factors. nsCL/P is usually diagnosed using fetal ultrasound during the late second trimester; however, these results are often affected by factors such as instruments, fetal position, and maternal obesity. Moreover, by this time, structural anomalies in the fetuses are already formed and missed optimal time for intervention. Therefore, identifying more efficient and non-invasive biomarkers before fetal ultrasound is essential. In recent years, rapidly evolving omics technologies, including genomics, transcriptomics, proteomics, lipidomics, epigenomics, and single-cell omics, have been used to identify several nsCL/P-associated risk genes. Additionally, omics technologies have proven invaluable for investigating non-invasive biomarkers for prenatal diagnosis of nsCL/P. Therefore, this article reviews the current applications of multi-omics technologies in nsCL/P research, focusing on their use to identify pathogenic genes and the research advances in prenatal diagnosis. We highlighted the technological landscape and applications of multi-omics in nsCL/P, and explored the potential opportunities and challenges for future clinical practice.

Introduction

Oral and maxillofacial clefts are the most common congenital craniofacial anomalies and are categorized into non-syndromic and syndromic types, depending on the presence or absence of other congenital anomalies. Approximately 70% of cleft lip with or without cleft palate cases occur as non-syndromic and the remaining 30% are a part of syndromic phenotypes.¹ The incidence of non-syndromic cleft lip with or without cleft palate (nsCL/P) is higher and the causes are more complex. nsCL/P encompasses non-syndromic cleft lip (nsCL), nonsyndromic cleft palate (nsCP), and non-syndromic cleft lip with palate (nsCLP). The global prevalence of nsCL/P among live births ranges from 1 in 500 to 1 in 2,500, varying slightly among different regions.² The development of nsCL/P is influenced by a combination of genetic and environmental factors.³ Maternal nutritional status, smoking, alcohol consumption, passive smoking, air pollution, and heavy metal exposure are some of the environmental factors that may contribute to the development of nsCL/P.4, 5, 6, 7, 8, 9 Children with nsCL/P have facial malformations that affect their appearance and result in varying degrees of impaired swallowing, breathing, and speech functions. These children often require multiple surgeries, which poses serious psychological and financial burdens for patients as well as their families.

As palate morphological and genetic development in mice closely resembles that of humans, mice are considered the most suitable animal model for studying cleft palate. Common teratogens used in cleft palate models include all-trans retinoic acid (atRA) and 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). atRA is the main metabolically active product of vitamin A. Though, it plays major roles in embryonic organ development and is involved in cell proliferation, differentiation, and migration.¹⁰ Clinically, atRA is commonly used to treat skin diseases and cancers such as acute promyelocytic leukemia. However, excessive exposure to atRA induces cleft palate. TCDD, on the other hand, is a common environmental pollutant, primarily produced as a byproduct of various industrial activities such as metal smelting and waste incineration. TCDD can be absorbed into the human body through the gastrointestinal tract, respiratory system, and skin, inducing various physiological and toxicological responses and leading to endocrine disruption, reproductive and developmental defects, immunosuppression, and carcinogenesis.11, 12, 13 Previous studies have widely used human specimens and mouse models of cleft palate to explore pathogenic genes and the associated mechanisms. In humans, palatogenesis is initiated in the early 6^th week of gestation and palatal fusion is completed by the 12^th week of gestation.¹⁴ The secondary palate develops in three stages. In mice, the palate grows from the lateral aspects of the maxillary process on embryonic day E11.5 (corresponding to the 6^th week of gestation in humans). During E12.5-13.5 (gestation weeks 6-7 in humans), the palate grows vertically downward along the sides of the tongue and then ascends to a horizontal position on E14.0. Finally, on E14.5-E16.5 (gestation weeks 9-12 in humans), the bilateral palatine processes begin to converge along the midline and merge to form a complete palate.¹⁵ Disruptions at any point in this entire process can lead to the formation of cleft palate. Presently, multi-omics technologies have been performed on human and mice tissues.

Currently, ultrasound imaging and blood tests are the most common screening methods for detecting fetal developmental defects. Currently, nsCL/P is primarily diagnosed using fetal ultrasound during the late second trimester. However, factors such as instruments used, maternal obesity, fetal position, and physician skills may interfere with the accuracy of the diagnostic results. Ultrasound detection rates vary widely between different nations and regions, with second-trimester detection rates for nsCL/P ranging from 28% to 88%.¹⁶ Moreover, structural anomalies are already formed in the late second trimester, crossing the most appropriate time window for clinical intervention. Owing to its non-invasive nature, peripheral blood testing of pregnant women is considered a better method of prenatal diagnosis than amniocentesis or cord blood sampling. Moreover, peripheral blood can be easily collected and can be readily examined in primary hospitals to provide guidance for further clinical examinations. However, there are no clinical biomarkers for the diagnosis of nsCL/P.

In recent years, the rapid development in multi-omics technologies, including genomics, transcriptomics, proteomics, lipidomics, epigenomics and single-cell omics, have significantly contributed to the identification of genes responsible for congenital malformations. Genetic screening played an important role in the detection of diseases. Clinical data have revealed that the presence of 2 or 3 oral features was commonly seen in PTEN Hamartoma Tumour Syndrome (PHTS) patients.¹⁷ PTEN inhibitor could partially rescue the mouse embryonic CP induced by atRA.¹⁸ Dental practitioners are crucial for the early detection of PHTS by identifying oral features, such as high palate and cleft palate, which allows for timely genetic diagnosis and the initiation of preventative cancer care.^17,19 These technologies are now innovatively utilized in early screening for common congenital malformations, such as neural tube defects and congenital heart diseases.^20,21 Deng et al. identified a panel of six potential prenatal diagnostic genes for neural tube defects through integrating bulk- and scRNA-seq transcriptome.²² Multi-omics technologies are constantly emerging that have enabled researchers to access multi-layer information from the genome, transcriptome, proteome, metabolome, and more. Presently, there is no review summarized recent advances in biomarkers for the prenatal diagnosis of fetal nsCL/P or pathogenic genes based on multi-omics technologies. To delve deeper into this aspect, in this review, we aimed to discuss and summarize the studies that have applied multi-omics technologies to discover pathogenic genes associated with nsCL/P, as well as its prenatal biomarkers. The search strategy is presented in file S1. Here is the figure abstract of this study (Figure 1).

Fig. 1.

The technological landscape and applications of multi-omics technologies in nsCL/P. Multi-omics techniques were performed in human and mouse specimens to search for pathogenic genes and prenatal biomarkers of nsCL/P. yellow arrow, intact lip; 1, hard palate; 2, soft palate; 3, intact maxilla; white arrow, cleft lip; 4, hemimaxilla; red arrow, cleft palate.

Genomics

Various genomics research methods have been applied to screen for candidate causative genes of nsCL/P, including genome-wide association studies (GWASs), whole exome sequencing (WES), whole genome sequencing (WGS) and so on. GWAS is a conventional method for discovering mutation sites known as single-nucleotide polymorphisms (SNPs) in disease-associated risk genes. The first GWAS on nsCL/P identified the presence of a susceptibility locus on chromosome 8q24.²³ This finding was subsequently confirmed by other GWASs in several independent populations; in addition, these GWASs studies identified some additional susceptibility loci on chromosome 17q22, 10q25, and 9q22.24, 25, 26, 27 Bian et al. identified 14 susceptibility loci associated with nsCLP and confirmed 12 previously reported loci in the Chinese population.²⁸ A GWAS study identified 14 susceptibility loci for nsCL/P and suggested different subtypes harbor different genetic etiologies.²⁹ Many other nsCL/P-associated genes and SNPs within these genes, such as IRF6, MAFB, SHTN1, PAX7, FGFR2 and NOG have been identified through GWASs.30, 31, 32 Although GWASs have been highly successful in identifying susceptibility genes for nsCL/P, the genes or SNPs identified to date represent only a small fraction of the total genes involved in this condition.³³

Exons constitute only 1% of the human genome; however, 85% of disease-causing mutations in humans are located in these coding regions.³⁴ WES, also known as targeted exome capture, is a genomic analysis method that utilizes sequence-capturing techniques and exon DNA sequence microarrays to capture and enrich DNA from exons throughout the genome for high-throughput sequencing. WES has been performed in various populations to identify pathogenic SNPs and small insertions/deletions (indels) associated with nsCL/P.³⁵ Several candidate causative genes and loci associated with nsCL/P have been identified using WES, including CHD7, PDGFC, ACSS2, PHYH, LRP6, MTR, ARHGAP29, GLI2, TP63, IRF6, NRP1, RPL27A, FZD6, and PTCH1.36, 37, 38, 39, 40, 41, 42, 43, 44, 45 With an increase in the discovery of candidate causative genes and mutation loci, studies investigating their functions are gradually progressing. A mutation locus in the LAMA5 gene has been detected in nsCL/P patients from eastern China and subsequently found to be involved in the temporal and spatial development of the palate in mouse models of cleft palate.⁴⁶ WGS, unlike WES, can be used to detect SNPs and indels in non-coding regions. WGS analysis of case-patients with orofacial anomalies from European and Colombian revealed a new disease locus on chromosome 21.⁴⁷ Loss-of-function de novo mutations in IRF6, TFAP2A, and ZFHX4 have been identified in the tissues of patients with oral and maxillofacial anomalies.⁴⁸ The characteristics of the studies mentioned above are presented in Table 1. These findings imply that genomics technologies are a reliable methodological approach to identify the causative genes of nsCL/P with enhanced accuracy and efficiency, thereby significantly contributing to genetic counseling, prenatal diagnosis, and prevention. Circulating cell-free DNA (cfDNA) in maternal bodily fluids carry information concerning the fetal origin. Over the past decade, non-invasive prenatal test employing cfDNA to screen for fetal aneuploidy has been used in clinical prenatal care. As we look ahead, future studies may focus on the potential clinical applications of cfDNA in nsCL/P.

Table 1.

Characteristics of the genomics studies in nsCL/P.

Technologies	Subtypes of orofacial clefts	Study participants	Areas or races	Genetic susceptibility factors	References
GWAS	nsCL/P	224 cases, 383 controls	Central European	chr. 8q24	Birnbaum, S. et al.,23
	nsCL/P	111 cases, 5951 controls	Greater Philadelphia	chr. 8q24	Grant, S F. et al.,24
	nsCL/P	401 cases, 1323 controls	Central European	chr. 17q22 and 10q25.3	Mangold, E. et al.,25
	nsCL/P	149 cases, 303 controls	Mayan Mesoamerican populations	IRF6, chr.10q25, and 8q24	Rojas-Martinez A, et al.,26
	nsCL/P	515 cases, 151 controls	Honduran and Colombian populations	chr. 8q24, 9q22, 10q25, and 17q22	Lennon C J, et al.,27
	nsCLP	7404 cases, 16059 controls	Chinese populations	chr. 2p25.1, 4p16.2, 4q28.1,5p12, 6p24.3, 8p11.23, 8q22.1, 9q22.32, 12q13.13, 12q13.2, 12q21.1, 14q22.1, 14q32.13, and 17q21.32	Yu Y, et al.,28
	nsCL/P	6986 cases, 10165 controls	Chinese populations	chr. 3p22.1, 4p16.3, 11q14.2, 12q21.31, 13q32.3, 14q13.3, 14q32.2, 16q24.2, 19p13.11, 16q24.1, 1q32.2, 2p24.2, 10q25.3, and 20q12	Huang, L, et al.,29
	nsCL/P	1863 case–parent trios	European and Asian populations	chr. 8q24, IRF6, MAFB, and ABCA4	Beaty, T. H, et al.,30
	nsCL/P	1931 cases, 2258 controls	Chinese populations	chr. 10q25.3 and SHTIN1	Wang Y, et al.,31
	nsCL/P	1409 cases, 1409 controls	Asian and European populations	NTN1, PAX7, FGFR2, and NOG	Leslie EJ, et al.,32
WES	nsCL/P, syndromic CL/P	107 singleton pregnancies and their fetuses	Chinese populations	CHD7	Yan S, et al.,36
	nsCLP	52 case–parent trios	Honduran populations	ACSS2 and PHYH	Aylward A, et al., 37
	nsCP	30 cases, 30 controls	Brazilian populations	LRP6 and MTR	Machado RA, et al.,38
	nsCL/P	a four-generation family	Chinese populations	ARHGAP29	Tang JX, et al.,39
	nsCL/P	a three-generation family	Chinese populations	GLI2	Meng P, et al.,40
	nsCL/P	a three-generation family	Chinese populations	TP63	Xu T, et al.,41
	nsCL/P	a three-generation family	Chinese populations	IRF6	Wang Y, et al.,42
	nsCL/P	a four-generation family	Saudi populations	NRP1 and RPL27A	Al Mahdi HB, et al.,43
	nsCL/P	a three-generation family	Chinese populations	FZD6	Zhang J, et al.,44
	nsCL/P	a three-generation family	Chinese populations	PTCH1	Zhong W, et al.,45
	nsCL/P	30 cases	Chinese populations	LAMA5	Fu, et al.,46
WGS	nsCL/P	580 case–parent trios	European and Colombian populations	chr. 21q	Mukhopadhyay, et al.,47
	nsCL/P	756 case–parent trios	European, Colombian, and Taiwanese	IRF6, TFAP2A, and ZFHX4	Bishop MR, et al.,48

Transcriptomics

Figure 2. presented the applications of transcriptomics to exploring the pathogenic genes and biomarkers of nsCL/P. Cai et al. performed transcriptomics analysis of the upper lip and primary palate from C57BL/6J mouse embryos at three stages (E10.5, E11.5 and E12.5) and revealed a series of key genes involved during development.⁴⁹ Following the completion of the Human Genome Project, researchers discovered that over 80% of disease-causing mutations occur in the non-coding regions of genes.⁵⁰ Non-coding RNAs include long non-coding RNAs (lncRNAs), circular RNAs (circRNAs), microRNAs (miRNAs), and PIWI-interacting RNAs (piRNAs), among others. lncRNAs are non-coding RNAs longer than 200 nucleotides in length, and their function is closely associated with subcellular localization. In the cytoplasm, lncRNAs form sponge complexes with miRNAs and competitively exert the mechanism of competitive endogenous RNAs. In the nucleus, lncRNAs interact with various molecules, including DNA, RNA, and proteins, to maintain chromosome structure and function, regulate gene transcription in cis or trans, and influence mRNA splicing, stabilization, and translation.51, 52, 53 Many studies have reported the involvement of lncRNAs in nsCL/P development.54, 55, 56 Wang et al. systematically investigated lncRNAs with these functions in nsCL/P and characterized the lncRNA USP17L6P/miR-449c-5p/c-Myc regulatory network.⁵⁷ Liu et al. found that the lncRNA Meg3/TGF-β/Smad regulatory axis affects the proliferation and apoptosis of mouse embryonic palatal mesenchymal (MEPM) cells.⁵⁸ Li et al. found that lncRNA MIR31HG is present in the susceptibility locus of nsCL/P and affects the proliferation and migration of human embryonic palatal mesenchymal (HEPM) cells and human oral keratinocyte (HOK) cells.⁵⁹ Additionally, the lncRNA NONMMUT100923.1/miR-200a-3p/Cdsn regulatory axis contributes to the failure of palate fusion by impeding the epithelial-mesenchymal transition of cells in the midpalatal suture.⁶⁰ Similarly, circRNAs form sponge complexes with miRNAs and affect the post-transcriptional level regulation of miRNA target genes. Shu et al. proposed and validated the circRNA 0954/miRNA-881-3p/RKAR1α regulatory axis through omics data of mouse cleft palate tissues.⁶¹ RNA sequencing (RNA-seq) was used on palatal tissues at four stages (E10.5, E13.5, E15, and E17), three lncRNAs (H19, Malat1, and Miat) and four mRNAs (Cdh1, Irf6, Grhl3, and Efnb1) were validated as hub genes during palatogenesis.⁶² miRNAs are small non-coding RNAs of 18-25 nucleotides whose primary function in diseases is to inhibit target genes at the post-transcriptional level.⁶³ miRNAs have been demonstrated to be involved in spatiotemporal developmental processes in orofacial and palatal sites in mice and to participate in the regulatory network of causative genes of cleft palate.^64,65 Many similar studies have been reported to date. For example, miR-124-3p and miR-340-5p have been shown to inhibit the proliferation of MEPM cells and the neural crest stem cell line O9-1 by regulating the expression of target genes such as Sox5h, Trp53, and Tgfbr1; further, a cocktail of miR-124-3p and miR-340-5p inhibitors can reduce the malformation rate of cleft palate in mice.⁶⁶ miRNA-470-5p inhibits epithelial-mesenchymal transition in MEPM cells by regulating the expression of its target gene Fgfr1, which in turn results in failure of palate fusion.⁶⁷ miR-27b, miR-133b, and miR-205 inhibit the proliferation of lip mesenchymal cells in humans and mice, leading to the development of cleft lip.⁶⁸ The Let-7c-5p/PIGA and miR-193a-3p/TGFB2 regulatory axes decrease the proliferation and increase the apoptosis of HEPM cells and oral epithelial cells, which are involved in nsCL/P.⁶⁹ Tang et al. performed RNA-seq on plasma from nsCL/P patients, and found three differentially expressed miRNAs (miR-212-3p, miR-200b-3p and miR-130b-3p) and their related ceRNA network.⁷⁰ All of these studies suggest that miRNAs are involved in the process of lip and palate development. Large quantities of free RNA can be found in the peripheral blood of pregnant women and can be characterized using transcriptomics or microarray techniques. Exosomes, which are a class of lipid bilayer vesicles rich in miRNAs and resistant to enzyme degradation in body fluids, are present in the peripheral blood.⁷¹ Exosomes participate in embryonic development by facilitating cellular and cell-cell transport and communication. In a previous study, we highlighted the clinical applicability of the Let-7 family, either from maternal plasma or plasma-derived exosomes, in predicting nsCLP. We found that EN2 regulates LIN28A/hsa-let-7a-3p-suppressed hedgehog signaling pathway, inhibiting HOK cell proliferation by suppressing GLI2 activity.⁷² The genes involved in this regulatory axis were screened using RNA-seq analysis of lip tissues from nsCLP fetuses and further validated using additional lip specimens from nsCLP and control fetuses. Specifically, we identified eight miRNAs from the Let-7 family (hsa-let-7a-3p, hsa-let-7a-5p, hsa-let-7c-5p, hsa-let-7d-3p, hsa-let-7d-5p, hsa-let-7e-5p, hsa-let-7f-5p, and hsa-miR-98-5p) through small RNA sequencing of the plasma and exosomes derived from the plasma of pregnant women carrying nsCLP fetuses. Our findings indicate that these miRNAs hold promise as potential diagnostic biomarkers of nsCLP. The area under the receiver operating characteristic curve (AUC) of these eight miRNAs in the plasma was 0.893, with sensitivity and specificity values of 84.40% and 80.00%, respectively. When combined with diagnosis using plasma-derived exosomes, the AUC was 0.992 and the sensitivity and specificity were 100.00% and 93.30%, respectively. In a previous study, a miRNA microarray conducted on saliva collected from 12 children with cleft lip and palate and 12 healthy children revealed significant differences in miR-141, miR-223, and miR-324-3p expression levels.⁷³ Another study performed miRNA microarray analysis using plasma specimens from three children with cleft lip and three healthy children and subsequently conducted real-time quantitative reverse transcription polymerase chain reactions to validate the findings on three miRNAs (miR-16-2-3p, miR-365a-3p, and miR-877-5p) with high AUC values.⁷⁴ Li et al. included serum specimens from three children with cleft lip, three with cleft lip and palate, and three healthy children for miRNA microarray studies. Further sample validation revealed significant differences in the expression of six miRNAs (miR-340-5p, miR-877-5p, miR-3648, miR-1260a, miR-494-3p, and miR-1304-3p) in the patients with CL/P.⁷⁵ piRNAs, ranging from approximately 26-31 nucleotides in length, along with miRNAs, are a major class of small non-coding RNAs. These RNAs bind to PIWI proteins and play important roles in embryonic stem cell differentiation and early embryonic development.⁷⁶ In our previous study, we identified three piRNAs (hsa-piR-009228, hsa-piR-016659, and hsa-piR-020496) derived from the plasma exosomes of pregnant women as potential diagnostic biomarkers for the early diagnosis of nsCLP in the early second trimester (gestation weeks 15-19).⁷⁷ The characteristics of the studies mentioned above are presented in Table 2.

Figure 2.

Application of transcriptomics to exploring the pathogenic genes and biomarkers of nsCL/P. Transcriptomics includes microarray and RNA-seq analysis. LncRNA, circRNAs, miRNAs and mRNAs are found as pathogenic genes and participate in the regulatory networks of nsCL/P. miRNAs and piRNAs from serum/plasma, saliva and exosomes are as potential biomarkers for nsCL/P.

Table 2.

Characteristics of the transcriptomics studies in nsCL/P.

Subtypes of orofacial clefts	Samples	Species	RNA names	References
nsCL/P	peripheral blood samples	human	lncRNA USP17L6P, hsa-miR-449c-5p, and MYC	Wang X, et al.,57
nsCP	palatal shelf tissues	57BL/6N mice	lncRNA Meg3, TGF-β, and Smad	Liu X, et al.,58
nsCL/P	lip tissues	human	lncRNA MIR31HG	Li X, et al.,59
nsCP	palatal shelf tissues	C57BL/6J mice	lncRNA NONMMUT100923.1, miR-200a-3p, and Cdsn	Zhang M, et al.,60
nsCP	palatal shelf tissues	C57BL/6J mice	circRNA_0954, miRNA-881-3p, and PRKAR1α	Shu X, et al.,61
nsCP	palatal shelf tissues	ICR mouse	lncRNAs H19, lncRNAs Malat1, lncRNAs Miat, Cdh1, Irf6, Grhl3, and Efnb1	Huang W, et al.,62
nsCP	palatal shelf tissues	C57BL/6J mice	miR-129-5p, miR-340-5p, Sox5, Trp53, Chd7, Fign, and Tgfbr1	Yoshioka H, et al.,66
nsCP	palatal shelf tissues	C57BL/6J mice	miRNA-470-5p and Fgfr1	Zhou J, et al.,67
nsCL	lip tissues	human, mice	miR-27b, miR-133b, and miR-205	Yoshioka H, et al.,68
nsCL/P	palatal shelf tissues, lip tissues	human, mice	Let-7c-5p, miR-193a-3p, PIGA, and TGFB2	Fu C, et al.,69
nsCL/P	plasma	human	miR-212-3p, miR-200b-3p, and miR-130b-3p	Tang J, et al.,70
nsCLP	plasma and plasma-derived exosomes	human	hsa-let-7a-3p, hsa-let-7a-5p, hsa-let-7c-5p, hsa-let-7d-3p, hsa-let-7d-5p, hsa-let-7e-5p, hsa-let-7f-5p, and hsa-miR-98-5p	Jia S, et al.,72
nsCLP	saliva	human	miR-141, miR-223, and miR-324-3p	Grassia V, et al.,73
nsCL	plasma	human	miR-16-2-3p, miR-365a-3p, and miR-877-5p	Zou J, et al.,74
nsCL/P	serum	human	miR-340-5p, miR-877-5p, miR-3648, miR-1260a, miR-494-3p, and miR-1304-3p	Li J, et al.,75
nsCLP	plasma-derived exosomes	human	hsa-piR-009228, hsa-piR-016659, and hsa-piR-020496	Jia S, et al.,77

Compared with DNA, RNA is less stable, has more stringent storage requirements, and necessitates more complicated experimental handling. Thus, uniform, rigorous, and standardized systems should be used to ensure the reproducibility of the experimental results from such studies.

Proteomics

Being downstream of the genomics and transcriptomics analysis, the proteomics can reflect the combined effects of multiple upstream factors more accurately and comprehensively, helping us to understand the pathogenesis of the disease. The development of proteomics relies on two-dimensional gel electrophoresis (2-DE) and mass spectrometry. One conventional assay method in proteomics is 2-DE with matrix-assisted laser desorption/ionization time of flight mass spectrometry (MALDI-TOF-MS). Compared with 2-DE, liquid chromatography-tandem mass spectrometry (LC-MS/MS) has high detection throughput, accuracy, reproducibility, and can be used for comparative and quantitative proteomics analyses. The isobaric tags for relative and absolute quantification (iTRAQ) technique, in which amino acid residues in a protein sample are labeled by isobaric tags to produce specific mass signals in mass spectrometry analysis, enables the quantitative analysis of proteins. Shotgun proteomics is another LC-MS-based technique used in proteomics analysis. Proteins are first enzymatically digested, and peptides of appropriate size are detected by mass spectrometry. Next, the proteins corresponding to the peptides are determined through alignment with database entries. Finally, qualitative and quantitative analyses are performed.⁷⁸ Tandem mass tagging (TMT) is an in vitro peptide labeling technique whose principle is similar to that of iTRAQ. Peptide fragments are labeled by linking them to specific amino acid sites, and then tandem mass spectrometry is performed to monitor the tagged fragments and quantify the peptides, thus achieving higher throughput compared with iTRAQ. These common proteomics technologies have been used to study pathogenic proteins associated with nsCL/P and in prenatal diagnosis. The changes in protein profiles using mouse models of cleft palate have been previously investigated. For example, Yuan et al. constructed a mouse model of cleft palate using TCDD and performed 2-DE in conjunction with MALDI-TOF-MS on palate tissues. They found that peroxidase-1 is involved in energy metabolism, cell migration, and apoptosis in palate development in mice.⁷⁹ Huang et al. identified 10 hub proteins including motor proteins and extracellular matrix components in pathogenesis using iTRAQ proteomic analysis on mice model of cleft palate⁸⁰. Wang et al. constructed a mouse model of cleft palate using TCDD and atRA separately and extracted palate tissues for iTRAQ proteomic analysis with further validation through western blot, revealing significant enrichment of 14-3-3 sigma and annexin A1 protein expression in the cleft palate group.⁸¹ Moreover, Wang et al. investigated prenatal diagnostic proteins as biomarkers for cleft palate by constructing animal models. They used the teratogens atRA and TCDD for the construction of mouse models, extracted amniotic fluid for antibody array to screen for differentially expressed proteins, confirmed these results through enzyme-linked immunosorbent assay (ELISA) and immunohistochemistry, and revealed that receptor of advanced glycosylation end products (RAGE) and epiregulin were significantly decreased in amniotic fluid.⁸² Subsequently, they found that interleukin (IL)-12p40 and soluble RAGE were differentially expressed in the serum of mice with cleft palate, indicating that these differentially expressed proteins can be used as diagnostic biomarkers for cleft palate in mice.⁸³ Following initial investigations in mouse models, subsequent studies have shifted their focus on humans. Several differentially expressed secreted proteins, including actin, salivary cystatin, and keratin, were screened from saliva collected from children with cleft palate and healthy children using MALDI-TOF-MS.⁸⁴ Zhang et al. collected serum from children with nsCLP and healthy children and applied shotgun proteomics to analyze differentially expressed proteins and found that retinol-binding protein 4 and vitamin A were significantly decreased in the serum of children with nsCL/P.⁸⁵ This finding provides a theoretical basis for vitamin A supplementation during pregnancy to prevent the development of nsCL/P. Hou et al. conducted TMT-based quantitative proteomics analysis on serum from children with non-syndromic orofacial cleft compared with healthy children, followed by quantitative validation using parallel reaction monitoring techniques and found significant differences in the expression levels of four proteins (sex hormone-binding globulin, periostin, osteomodulin, and aggrecan core protein) in the non-syndromic orofacial cleft group.⁸⁶ Wang et al. categorized pregnant women in nsCLP, nsCL, and healthy control groups, collected serum samples from the women at 16-28 weeks of gestation for iTRAQ proteomic analysis, and validated them using MRM and ELISA methods.⁸⁷ They found that Haptoglobin was significantly downregulated in the nsCLP and nsCL groups, whereas levels of Apolipoprotein(a) and C-reactive protein were significantly increased in the nsCLP group. The combined diagnostic efficiency of these three proteins was better in the nsCLP group, with an AUC of 0.820 (95% confidence interval: 0.701-0.906). This finding suggests a practical avenue for investigating protein biomarkers for prenatal diagnosis of fetal cleft lip and palate. In contrast to traditional data acquisition, data-independent acquisition (DIA) as a sensitive proteomic method enables in-depth protein profiling with low sample requirements in an unbiased manner. Future studies may use DIA proteomics to explore more promising protein biomarkers for nsCL/P. The characteristics of the studies mentioned above are presented in Table 3.

Table 3.

Characteristics of the proteomics studies in nsCL/P.

Technologies	Subtypes of orofacial clefts	Study participants	Tissues	Protein names	References
2-DE with MALDI-TOF-MS	nsCP	C57BL/6J mice	palates	Peroxiredoxin-1	Yuan X, et al.,79
iTRAQ	nsCP	Kun Ming mice	palates	Tnnc2, Myl1, Myh8, Ckm, Myh3, Rps16, Col2a1, Col9a1, Matn3, and Acan	Huang Z, et al.,80
iTRAQ	nsCP	C57BL/6J mice	palates	14-3-3 sigma and annexin A1	Wang C, et al.,81
label-based mouse antibody array	nsCP	C57BL/6J mice	amniotic fluid	receptor for advanced glycation end products (RAGE) and epiregulin	Wang X, et al.,82
label-based mouse antibody array	nsCP	C57BL/6J mice	serum	interleukin (IL)-12p40 and soluble RAGE	Wang X, et al.,83
MALDI-TOF-MS	nsCLP	31 patients and 20 controls	saliva	actin, salivary cystatin, keratin, TGF-b3, and dermokine	Szabo GT, et al.,84
shotgun proteomics	nsCLP	13 patients and 10 controls	serum	Retinol binding protein 4 and vitamin A	Zhang J, et al.,85
TMT	nsCLP	5 patients and 5 controls	serum	sex hormone-binding globulin, periostin, osteomodulin, and aggrecan core protein	Hou Y, et al.,86
iTRAQ	nsCLP, nsCL	20 pregnant women carrying nsCLP/nsCL fetuses and 20 pregnant women with healthy ones	serum	Haptoglobin, Apolipoprotein(a), and C-reactive	Wang X, et al.,87

Lipidomics

Lipidomics is a relatively new high-throughput methodology compared with conventional omics technologies such as genomics, transcriptomics, and proteomics. Lipidomics is a major branch of metabolomics that has benefited from the development of mass spectrometry technology. It uses qualitative and quantitative analysis and involves the identification of lipid metabolites to reveal relationships between lipid synthesis, lipid metabolism, and physiological and pathological processes, thus enabling researchers to predict the development, progression, and outcomes of diseases. Lipidomics is categorized into untargeted Lipidomics (which is characterized by broad coverage) and targeted Lipidomics (which is characterized by high accuracy).^88,89 Lipids act as major components of cell membranes, energy stores, signaling molecules, and critical players in embryonic development. Zhang et al. found that unsaturated triglycerides play a crucial role in palate formation through non-targeted lipidomic analysis of palate tissues with atRA-treated cleft palate compared with control embryos.⁹⁰ Abnormal lipid metabolism also reportedly affects the proliferation and migration of MEPM cells. For example, intracellular lipid droplets were found to accumulate in a Tgfbr2 mutant mouse model of cleft palate, in which MEPM cells proliferation was impaired by abnormal lipid metabolism.⁹¹ Similarly, another study found decreased expression of the adipose triglyceride lipase Pnpla2 (a major enzyme involved in lipolysis process) in palate tissues from an atRA-induced mouse cleft palate model, led to impaired proliferation and migration of MEPM cells.⁹² Combining lipidomics and machine learning, we explored a panel of 3 lipid biomarkers showed great potential for nsCLP diagnosis.⁹³ Lipid-related studies are particularly important for investigating prenatal screening and the pathogenic mechanisms of nsCL/P (Figure 3A).

Figure 3.

Application of Lipidomics and single-cell omics to exploring the pathogenic genes of nsCL/P. Lipidomics and single-cell omics techniques were performed in nsCL/P studies. A: Lipidomics studies in nsCL/P. B: Single-cell omics studies in nsCL/P.

Epigenomics

Epigenomics is a comprehensive analysis of a range of epigenetics, including but not limited to, DNA methylation and histone modification. Epigenetics is a critical mediator of the interplay between genes and the environment, and influences phenotypic outcomes. As mentioned, nsCL/P is influenced by a combination of environmental and genetic factors. DNA methylation is well-recognized as one of the mechanisms of epigenetics and involved in the development of orofacial clefts. Aberrant DNA methylation levels result in craniofacial malformations, including orofacial clefts.94, 95, 96 DNA hypomethylation or hypermethylation depends on the tissue- and timing-dependent in orofacial development.⁹⁷ Different orofacial cleft subtypes have distinct methylation profiles.⁹⁸ Up till now, DNA methylation has been examined in patients with nsCL/P from multiple specimens, including lip, palate, whole blood, and saliva, as well as experimental studies in mice. DNMT1, an active DNA methyltransferase enzyme, regulates cranial neural crest cell (cNCC) proliferation and differentiation, and required for orofacial morphogenesis.⁹⁴ Zhang et al. illustrated the DNA methylation profile of lip tissues from nsCLP patients by whole-genome bisulfite sequencing (WGBS) and found the DNA hypomethylation of GLI2, the well- known susceptibility gene of nsCL/P.⁹⁹ Our previous study detected numerous aberrantly methylated loci in fetal lip tissues of nsCL fetus via Infinium HumanMethylation450 BeadChi array.¹⁰⁰ Two individual CpGs and multiple differentially methylated regions were identified in blood DNA between newborn babies with and without nsCL/P.¹⁰¹ Other studies have confirmed that saliva and blood samples between monozygotic twins discordant for nsCL/P present significant differences in DNA methylation.^102,103 Histone acetylation is involved in the critical palate fusion during E13.5-E15.5 in TCDD induced cleft palate formation.¹⁰⁴ Wilderman, A. et al. profiled human embryonic craniofacial samples from each of four distinct Carnegie stages (CSs) (CS13, CS14, CS15, and CS17) encompassing 4-5 post conception weeks (pcw) to 6 pcw and found multiple histone modifications via chromatin immunoprecipitation sequencing (CHIP-seq).¹⁰⁵ The characteristics of the studies mentioned above are presented in Table 4. Understanding epigenomics in orofacial clefts will contribute to the development of new genetic tools for early detection, the enrichment of prognostic information, and ultimately, the prevention of orofacial clefts.

Table 4.

Characteristics of the epigenomics studies in nsCL/P.

Technologies	Subtypes of orofacial clefts	Study participants	Areas or races	Tissues	References
Infinium Human Methylation 450 K Bead-Array	nsCL/P	238 patients and 246 controls	Brazil	whole-blood and lip	Alvizi, L. et al.,95
Illumina Infinium HumanMethylation450 BeadChip	CL/P	150 patients	United Kingdom	whole-blood, lip, and palate	Sharp, GC, et al.,98
WGBS	nsCLP	8 nsCLP patients	Chinese populations	lip	Zhang, et al.,99
Illumina Infinium HumanMethylation450 BeadChip	nsCL	3 nsCL fetuses and 3 control fetuses	Chinese populations	lip	Xu, et al.,100
Illumina Infinium HumanMethylation450 BeadChip	nsCL/P	308 patients and 436 controls	Norway	whole-blood	Xu, et al.,101
WGBS	nsCLP	6 monozygotic twin pairs	Non-Hispanic White and Hispanic populations	saliva	Young, JI., et al.,102
MethylationEPIC BeadChip	nsCL, nsCLP	6 monozygotic twin pairs	Chinese populations	whole-blood	Shi, et al.,103

Single-cell omics

Conventional approaches involve the use of small RNA microarray or bulk transcriptomic analysis, which fail to provide cell type-specific gene expression profiles owing to the inherent heterogeneity within tissue blocks. Under these circumstances, single-cell RNA sequencing (scRNA-seq) offers an accurate and efficient way to analyze the interactions between different cells involved in craniofacial development (Figure 3B). scRNA-seq has been performed on mouse facial tissues to identify and analyze gene expression profiles of ectodermal, mesenchymal, and endothelial cell populations involved in the development of lip and palate tissues.¹⁰⁶ An scRNA-seq study of palate tissue in mouse embryos at day E13.5, E14.5 and E15.5 revealed complex cellular heterogeneity during craniofacial development, describing the complex regulatory role of cells derived from cNCC in the development of the craniofacial musculoskeletal system.¹⁰⁷ Another scRNA analysis of the anterior palate in mouse embryos at day E13.5 revealed the heterogeneity of mesenchymal cells.¹⁰⁸ scRNA-seq analysis of the palate tissues of mice with atRA-induced cleft palate and normal mice on day E16.5 revealed different subcellular patterns of gene expression, such as the increased levels of M1-type macrophages and monocytes, as well as activation of the IL-17 signaling pathway in mice with cleft palate.¹⁰⁹ In another recent study, gene expression and chromatin accessibility in the same cells at different developmental periods (E12.5, E13.5, E14.0, and E14.5) were analyzed using single-cell multi-omics technologies, including scRNA-seq and single-cell assay for transposase-accessible chromatin sequencing (scATAC-seq), to elucidate the linkage between spatial and temporal gene expression in palatal cells at the transcriptomic and genomic levels.¹¹⁰ A recent study generated genome-wide maps of the enhancer-associated histone mark H3K27ac, accessible chromatin, and gene expression via multi-omics techniques (CHIP-seq, ATAC-seq and RNA-seq) from embryonic face tissue for CS 18, 19, 22, and 23, which are within the critical developmental window for palate formation.¹¹¹ Meanwhile, they combined the considerable temporal dynamics of human craniofacial enhancers with single-cell-resolved data of mouse embryonic face to define the regulatory landscape of craniofacial development at both bulk and single-cell resolution. The combination of conventional RNA sequencing methods with scRNA-seq analysis can help explore the role of known nsCL/P-related pathogenic genes during development.¹¹² Combining scRNA-seq and spatially enhanced resolution omics-sequencing (stereo-seq) at three critical stages of palate formation (E12.5, E14.5, and E16.5) establishes a transcriptomic regulatory landscape across single-cell and spatial scales during the development of palate in mice.¹¹³ Single-cell omics technologies can offer insights into gene expression at single-cell resolutions, allowing the study of cellular heterogeneity and the effective selection of cell types associated with lip and palate development.

Conclusion

Omics technologies are commonly employed in nsCL/P research to understand its causation and development and improve prenatal diagnosis. Utilizing multi-omics technologies enables a multifaceted search for the relevant causative genes of nsCL/P, providing a theoretical basis for further investigation of their molecular mechanisms. This article provides a comprehensive and timely review of multi-omics technologies in nsCL/P, effectively synthesizing their applications in identifying pathogenic genes and promising non-invasive prenatal biomarkers. These findings provide valuable clues for further understanding the underlying mechanisms and offer insights for the future development of prenatal diagnostics. Integrating multi-omics studies has contributed to a better knowledge of aetiology of nsCL/P. Nine novel candidate genes involved in nsCL/P were revealed by the integrative multi-omics analysis.¹¹⁴ Recently, a multi-omics approach was employed to provide genetic variants contributing to nsCL/P.¹¹⁵ Despite these advances, several shortcomings remain in the current body of research. Distinct genetic backgrounds across ethnic populations introduce significant genetic heterogeneity. This can lead to the omission of population-specific biomarkers and the generation of biased biological models, thereby compromising the generalizability and clinical applicability of multi-omics findings. The samples included in the research articles on prenatal diagnosis of nsCL/P presented in this review were primarily from single centers and included a small number of samples, which may have led to a diagnostic bias in biomarkers. In addition, detection techniques differed between studies, indicating a lack of uniformity and standardization. Therefore, future multicenter, large-sample studies with improved designs can enable timely translation of prenatal diagnosis research to the clinic. Data integration, standardization, and the substantial cost barriers that currently limit the widespread clinical implementation of multi-omics approaches. Further technological developments and cost reductions may eventually allow genomics, transcriptomics, proteomics, lipidomics, epigenomics, and single-cell multi-omics technologies to become routine clinical tests. Non-invasive prenatal diagnostic techniques based on omics technologies are likely to be key to early and accurate diagnosis of nsCL/P and will thus play a crucial role in prenatal medicine. Although multi-omics studies on nsCL/P have been carried out, integrated multi-omics analysis has rarely been reported. While multi-omics analysis provides a powerful tool for nsCL/P, it still faces multiple challenges, including data heterogeneity, computational complexity, difficulties in biological interpretation, and high demands for interdisciplinary collaboration. Artificial intelligence (AI) is particularly well-suited for the complexities of heterogeneous and unstructured data, excelling at discovering the non-linear, high-dimensional correlations present in multi-modal information.¹¹⁶ AI-based multi-omics analysis has great promise for prenatal diagnosis and mechanism research. Moreover, establishing an open-source database for nsCL/P multi-omics data would significantly advance research in this field.

Author contributions

CQ. Y., SS. J., and ZW. Y. wrote the main manuscript text. CQ. Y. and L.D., prepared the figures and tables. YZ. D., Y. W., and SY. C. edited the manuscript. All authors reviewed the manuscript.

Conflicts of interests

None disclosed.