Enhancing gene expression studies in bovine embryos fertilized in vitro: Identifying stable reference genes across blastocysts with different developmental speeds

Abstract

In vitro fertilization (IVF) is crucial for livestock reproduction; however, pregnancy rates after embryo transfer vary depending on the developmental speeds of the embryos. Although quantitative PCR (qPCR) is used to predict developmental potential, its reliability depends on the selection of appropriate reference genes (RGs) for normalization. To determine suitable RGs in bovine blastocysts with different developing speed, we evaluated the stability of eight candidate RGs (18S, ACTB, GAPDH, HMBS, PPIA, TBP, HPRT1, and SDHA) in early-, mid-, and late-developing IVF blastocysts (E-BL, M-BL, and L-BL, respectively) using RefFinder. Despite morphological similarities, E-BL, M-BL, and L-BL exhibited different biological features, including significantly lower pregnancy rates in L-BL than in the other groups, and less abundant transcript levels of five candidate RGs in L-BL than in E-BL. RefFinder revealed that ACTB was the most stable RG, whereas TBP was the least stable. To emphasize the critical importance of selecting stable RGs, we analyzed the expression of key developmental markers including those of the inner cell mass (ICM; OCT4, SOX2) and trophectoderm (TE; CDX2, GATA3, IFNτ), using various RGs for normalization. For ICM markers, normalization with ACTB showed results consistent with pregnancy rates, whereas moderately stable (18S) and less stable (TBP) RGs yielded contradictory outcomes. Normalization with unstable RGs produced inconsistent TE marker expression patterns (CDX2, GATA3) and overestimated (IFNτ) results across groups, compared with the results of ACTB. These results demonstrate that selecting inappropriate RGs for qPCR normalization can lead to misinterpretation, highlighting the necessity of proper RG evaluation to ensure accurate results in bovine embryo research.

In embryo production following in vitro fertilization (IVF), presumptive zygotes develop into blastocysts at different speeds from days 7 to 9 during in vitro culture (IVC), which is highly related to their fertility after embryo transfer (ET) into the recipients. Notably, pregnancy rates decrease in late-developing IVF blastocysts than in the early-developing blastocysts; however, the reason for this difference is not well understood [1, 2]. Each preimplantation embryo undergoes a complex series of developmental steps in which an undifferentiated cell biologically transforms into a totipotent embryo. Thus, a comparative analysis of IVF blastocyst characteristics under various conditions is necessary to increase the efficacy of ET for bovine blastocysts [3,4,5]. Presently, the most commonly used method for evaluating the viability and grade of IVF blastocysts is morphological evaluation using a microscope [6]. However, whether morphological evaluation results and embryo quality strongly correlate with morphological assessment and implantation ability, respectively, remains questionable [2, 6,7,8]. To overcome these limitations in the morphological evaluation of blastocysts, the analysis and interpretation of gene expression in transferable blastocysts are important for understanding the distinct molecular interactions under specific conditions and assessing the quality of different embryos [9].

Quantitative approaches for interpreting gene expression play a key role in understanding the underlying causes of biological phenomena [10, 11]. Notably, the use of quantitative reverse transcription polymerase chain reaction (qPCR) to analyze embryos has the advantages of accuracy, reliability, sensitivity, and reproducibility, along with the ability to test limited amounts of initial materials derived from a single embryo. Therefore, qPCR is widely applicable for gene expression studies in embryos. However, despite these advantages, inevitable variations may occur during qPCR because of inaccurate quantification of initial materials, poor purity of RNA extraction, nonspecific binding of primers, and differences in protocols, operators, and equipment [5, 10, 12, 13]. Therefore, target gene expression is normalized against that of a reference gene (RG) to address these variations. Indeed, normalization against an RG is considered the most reliable and widely used method in qPCR analysis because RG expression is essential for maintaining normal physiology and sustaining proper cellular function, and thus remains constant to ensure homeostasis. Additionally, RG expression is believed to remain unaffected by experimental conditions or sample variations [3, 6, 10, 11]. However, a universal RG whose expression does not change under different experimental conditions remains to be identified, as the expression levels of commonly used RGs are altered under different experimental conditions [12, 14, 15]. Notably, employing invalidated RGs during target gene normalization can lead to data misinterpretation and erroneous conclusions [6, 9, 14].

Therefore, to avoid misleading results from inadequate normalization, selection of the correct RGs under each experimental condition is a prerequisite for obtaining the most accurate and reliable qPCR results [3, 10, 11, 16]. Subsequently, several software algorithms, such as the comparative ΔCt method, GeNorm, NormFinder, and BestKeeper, have been employed to assess the stability of candidate RGs under each experimental condition [10, 11]. Because each algorithm applies different criteria to evaluate stability, RefFinder has recently been employed to obtain an integrated stability ranking from the data pool derived by each algorithm, which can be used to comprehensively determine the ideal RGs for any experimental condition [17].

Although bovine blastocysts with different developmental speeds exhibit different pregnancy potential, the molecular mechanisms underlying these differences remain to be elucidated. To address this, qPCR, which can be performed using limited initial materials, is essential; however, the most stable RGs in bovine blastocysts that develop at different speeds are yet to be identified. Therefore, this study attempted to evaluate the most stably expressed RGs in bovine blastocysts with different developmental speeds (early- to late-developing blastocysts), using various algorithms (RefFinder and the algorithms it incorporates, including GeNorm, NormFinder, BestKeeper, and the comparative ΔCt method). Further, we aimed to emphasize the importance of selecting an appropriate RG for the qPCR analysis of embryos by applying the evaluated RGs in the present study to normalize target genes.

Materials and Methods

Ethical statements

All animal experimental procedures were approved by the Institutional Animal Care and Use Committee of Kyungpook National University (approval number: 2023-0372).

In vitro production of bovine embryos with different developmental speeds

After culling at a local abattoir, bovine ovaries (Bos taurus coreanae) were obtained and transported to the laboratory in phosphate-buffered saline (PBS; Thermo Fisher Scientific, Waltham, MA, USA) at 37°C. Cumulus-oocyte complexes (COCs) were isolated from follicles < 10 mm using an 18 G needle (Becton, Dickinson and Company, Franklin Lakes, NJ, USA) and evaluated for cumulus cell compaction and oocyte viability. The selected COCs were incubated for in vitro maturation (IVM) in TCM-199 (Thermo Fisher Scientific) supplemented with 5 μg/ml follicle-stimulating hormone (Antrin R10, Kyoritsu Seiyaku Corporation, Tokyo, Japan), 10 IU/ml luteinizing hormone, 1 μg/ml 17β-estradiol, 10% fetal bovine serum, and 0.05 mg/ml gentamycin (all from Thermo Fisher Scientific) in an atmosphere with 5% CO₂ in humidified air at 38.5–39.0°C for 22–24 h. Two hours before COC maturation was completed, frozen semen was thawed in a water bath at 37°C and washed with Tyrode’s albumin lactate pyruvate (TALP) media by centrifugating at 1000 × g for 10 min. Motile sperm were then routinely isolated using the swim-up method for 1 h. The 10 mature COCs were inseminated with 1 × 10⁵ sperms in a 50 μl drop of TALP media supplemented with 20 μg/ml heparin (Thermo Fisher Scientific) at the same incubating condition as for IVM; the timing of COC insemination was considered as day 0. At 18–20 h post-insemination, the presumptive zygotes were denuded and further incubated as an IVC in synthetic oviductal fluid (SOF) media supplemented with 5 mg/ml bovine serum albumin (Thermo Fisher Scientific) and 0.05 mg/ml gentamycin in an atmosphere with both 5% CO₂ and O₂ in humidified air at 38.5–39.0°C (APM-30D, ASTEC, Fukuoka, Japan) until blastocysts appeared; during the IVC, cleaved zygotes were only collected on day 3 and cultured further. When blastocysts initially appeared during the IVC on days 7, 8, or 9, they were promptly collected for further study as early-, mid-, or late-developing blastocysts (E-BL, M-BL, or L-BL), respectively; to exclude the effects of blastocysts among the groups functioning in various developmental stages, only blastocysts ranked as stage 6–7 according to the International Embryo Technology Society (IETS) criteria were used in this study [18]. Blastocysts produced from four repeated IVF procedures were allocated to ET (N in E-BL, M-BL, L-BL = 19, 18, and 16, respectively; freshly transferred into recipients), morphological observation (N in E-BL, M-BL, L-BL = 13, 13, and 18, respectively; fixed using 4% paraformaldehyde), and molecular analysis by qPCR (each N = 18; snap-frozen using liquid nitrogen). The remaining blastocysts that were not allocated to any group were pooled and used for PCR efficiency tests.

Comparative characterization of bovine embryos with different developmental speeds

For morphological observations, the fixed blastocysts were washed twice using PBS, stained using 5 mg/ml 4′,6-diamidine-2′-phenylindole dihydrochloride (DAPI; Thermo Fisher Scientific) for 5 min, and mounted on a glass slide using Vectashield (Vector Laboratories Inc., Newark, CA, USA). Bright-field images of the mounted blastocysts were obtained to measure the blastocyst diameters. Thereafter, images of DAPI-positive cells for each blastocyst were obtained using an inverted fluorescence microscope (Olympus IX70; Olympus Corporation, Tokyo, Japan) at a wavelength of 461 nm to assess the total number of cells per blastocyst.

To compare the fertility of the E-BL, M-BL, and L-BL groups, ET was performed using recipients of another breed (Holstein heifers, approximately 1.5 years old, with a body condition score between 2.5 and 2.75). To prepare for ET using fresh E-BL, M-BL, and L-BL (stage 6–7 according to IETS criteria), recipients were synchronized to a 7-day-old corpus luteum (CL) using the widely adopted Ovsynch protocol, as established in previous studies [19,20,21], thus ensuring accurate comparison of molecular characteristics and pregnancy rates among blastocysts with different developmental speeds. A single fresh blastocyst from the E-BL, M-BL, and L-BL groups was gently loaded in a 0.25 ml mini straw (IMV Technologies, L’Aigle, France) with SOF media and transported to the farms of each recipient within a thermos set at 38°C. Upon arrival, the blastocysts were transferred to the same uterine horn as the CL using a nonsurgical method with an ET gun (IMV Technologies). The pregnancy diagnosis was performed 45–50 days post-ET, using an iScan with a 7.5 MHz linear transducer (DRAMIŃSKI, Sząbruk, Poland); ultrasonography showing the presence of a gestational vesicle with a fetus and fetal heartbeat was diagnosed as a pregnancy.

RNA extraction, cDNA synthesis, and qPCR to obtain the cycle threshold of candidate RGs

Total RNA was extracted from each single snap-frozen blastocyst using an RNeasy Micro kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions; the final elution was performed in 20 μl Rnase-free water (Qiagen). In agreement with a previous study [10], quantification of total RNA extracts was waived because of the extremely low amount of total RNA derived from a single blastocyst. The cDNA was synthesized with the entire volume (20 μl) of eluted total RNA using HiSenScript RH(-) RT PreMix kit (iNtRon Biotechnology, Seongnam-si, Korea) at 45°C for 60 min in a thermal cycler (Qiagen). Eight candidate RGs, 18S, ACTB, GAPDH, HMBS, PPIA, TBP, HPRT1, and SDHA (Supplementary Table 1) were selected for the present study. These are commonly used RGs in molecular analyses and are related to different intracellular biological functions. To obtain the cycle threshold (Ct) values for each RG in the E-BL, M-BL, and L-BL groups, qPCR was performed on a CFX Opus Real-Time PCR System (Bio-Rad Laboratories, Inc., Hercules, CA, USA) using TB Green^® Premix Ex Taq™ (Takara Bio, Shiga, Japan); a 20 μl mixture per reaction was applied, which comprised 10 μl TB Green Premix, 0.5 μl 0.2 μM forward/reward primer (Macrogen, Seoul, Korea), 2 μl template cDNA, and 7 μl purified water. The amplification procedure was initiated by holding the reaction mix at 95°C for 5 min, followed by 40 cycles of denaturation at 95°C for 5 min and annealing and extension at 60°C for 10 sec. Melting curves were generated from 60°C to 95°C, rising by 1°C at each step; similar to the previous study, owing to the small amount of cDNA product, qPCR was only performed once per blastocyst to obtain Ct values for each RG [10]. Finally, melting curve analysis of PCR amplicons and electrophoresis on a 1% agarose gel for the recovered qPCR products were performed to evaluate the appropriate amplification of a single PCR product and confirm the absence of unexpected hairpin or dimer formation.

qPCR efficiency of candidate RGs

To assess the qPCR efficiency of each RG, we employed a standard curve approach using a two-fold dilution series with six concentration points (1:1 to 1:32) of pooled cDNA samples; cDNA samples were pooled from the E-BL, M-BL, and L-BL groups to represent all experimental conditions. PCR amplification conditions were identical to those described previously. The resulting Ct values, corresponding to cDNA dilution factors, were used to calculate the key parameters, including slope, PCR efficiency (10^(1/–slope)–1), and correlation coefficient (R²), using Excel (Microsoft, Redmond, WA, USA), in accordance with a previous study [11].

Assessment for the stability of candidate RGs in bovine embryos with different developmental speeds

The stability of candidate RGs was evaluated using RefFinder, with the obtained Ct values. RefFinder is a web-based software (https://www.ciidirsinaloa.com.mx/RefFinder-master/) that evaluates stability of candidate RGs based on the results of four different well-known algorithms: GeNorm, NormFinder, BestKeeper (cut-off value as RG: standard error of the mean (SEM) of the coefficient of variance (CV) from Ct values higher than 1), and comparative ΔCt method [10, 11]. RefFinder calculates a comprehensive stability ranking to determine the best RGs [17].

Application of the validated RGs for normalization

To emphasize the importance of selecting the most stable RGs for each experimental condition, normalization of inner cell mass (ICM) markers such as OCT4 [19] and SOX2 [22, 23] and trophectoderm (TE) markers including CDX2 [24], GATA3 [25], and IFNτ [26] was performed using the various RGs analyzed in this study. The PCR amplification conditions used to obtain the Ct values for each primer set (Supplementary Table 1) were identical to those described previously. OCT4 maintains ICM pluripotency and regulates NANOG without causing transcriptome disruption or differentiation failure [19]. SOX2 is expressed in the ICM and is directly correlated with embryo quality and pregnancy rates [19, 22, 23]. CDX2 ensures blastocyst`s structural stability and promotes pregnancy signals (e.g., IFNτ) via TE differentiation [24, 27]. GATA3 sustains the TE lineage program by regulating NANOG and transcriptome balance [25], and IFNτ, secreted by TE cells, prevents luteolysis and sustains progesterone secretion [26]. Therefore, the expression of these early embryonic developmental genes in E-BL, M-BL, and L-BL was normalized using several RGs, including the most, moderately, and least stable RGs.

Statistical analysis

Statistical analyses were performed using SPSS (version 12.0: SPSS Inc., IL, USA). Total cell numbers, diameter, Ct values of each RG, and the relative expression level of target genes (OCT4, SOX2, CDX2, GATA3, and IFNτ) in the E-BL, M-BL, and L-BL were compared using ANOVA with the Games–Howell post hoc test. All analyzed results are described as the mean ± SEM. A value of P < 0.05 was considered statistically significant.

Results

Comparative characterization of bovine embryos with different developmental speeds

No significant differences were observed in the appearance of E-BL, M-BL, and L-BL based on morphological observation (Fig. 1A), including the average diameters (Fig. 1B) and total cell numbers (Fig. 1C). This result suggested that once different embryos reached the same stage of the blastocyst (stages 6–7 in the present study), the morphological characteristics of the blastocyst were similar, regardless of the developmental speed of the embryos. Further, in agreement with previous studies [10, 27], this finding also supported that the quantity of RNA extracted from single E-BLs, M-BLs, and L-BLs may be comparable without quantification because of the similar total cell numbers among groups. In contrast, a significant (P < 0.05) decrease in pregnancy rate (%) was found with the L-BL (19.5 ± 7%) relative to the E-BL (58.7 ± 5.9%) and M-BL (46.2 ± 10.1%), suggesting that molecular comparative analysis of blastocysts with different developmental speeds is necessary to understand their biological features and improve ET efficiency (Figs. 1D and 1E).

Fig. 1.

Morphological evaluation of the blastocysts produced in vitro as E-BL, M-BL, and L-BL. Representative bright-field and DAPI-stained images (A) suggest that the E-BL, M-BL, and L-BL are morphologically similar without significant differences in the average diameter (B) and total cell number (C). Pregnancy diagnosis by ultrasonography (D) presented significantly (P < 0.05) lower pregnancy rates in the L-BL compared to that in the other groups (E); pregnancy was confirmed when a gestational vesicle with the fetus and fetal heartbeat was observed. * indicates significant differences among groups. Graphs are presented as the mean ± SEM. Magnification: × 400. DAPI, 4′,6-diamidine-2′-phenylindole dihydrochloride; E-BL, early-developing blastocysts; M-BL, mid-developing blastocysts; L-BL, late-developing blastocysts.

Primer specificity, efficiency, and amplicon size

Following qPCR, amplified products from the eight candidate RGs were subjected to melting curve analysis and electrophoresis (Supplementary Fig. 1). Melting curve analysis demonstrated that the eight RGs in bovine embryos were amplified with a high peak of single amplicons, without nonspecific amplification. In electrophoresis, the expected size of the amplicons (Supplementary Table 1) was confirmed by a single band and the absence of contamination or improper amplification. Further, the qPCR efficiencies for the eight candidate RGs ranged between 0.93 and 1.04 (Supplementary Table 2). These results indicate that the assay systems for qPCR and candidate RGs are valid for quantifying transcripts in bovine embryos with different developmental speeds.

Average Ct values of candidate RGs in bovine embryos with different developmental speeds

The average Ct values obtained for the eight candidate RGs are presented in Fig. 2. Transcript levels of GAPDH, ACTB, and PPIA were consistent, regardless of the blastocyst developmental speed. However, the other five candidates (TBP, HMBS, HPRT1, SDHA, and 18S) showed significant (P < 0.05) differences among experimental groups; interestingly, the transcript levels of these RGs were lower (higher Ct values) in L-BL than in the E-BL. Along with the results shown in Fig. 1E, these findings suggest that the molecular features of bovine blastocysts at different developmental speeds may vary, highlighting the importance of evaluating the most stable RGs under experimental conditions.

Fig. 2.

Average Ct values of each reference gene in blastocysts with different developing speeds. Transcript levels of each RG in the E-BL, M-BL, and L-BL were statistically compared, resulting in significant (P < 0.05) differences in five candidate RGs, excluding GAPDH, ACTB, and PPIA. ^{a, b} Significant differences between groups are denoted by different letters above the bars (P < 0.05). Graphs are presented as the mean ± SEM. RG, reference gene; E-BL, early-developing blastocyst; M-BL, mid-developing blastocyst; L-BL, late-developing blastocyst.

Evaluation for the stability of eight candidate RGs in bovine embryos with different developmental speeds

The stability of each Ct value for the eight candidate RGs was analyzed using several algorithms (comparative ΔCt method, GeNorm, BestKeeper, and NormFinder) included in RefFinder, as shown in Table 1. First, according to the analysis using BestKeeper, all candidates exhibiting a SEM (± CV) of less than one were considered valuable RGs. In the stability results of the four algorithms, ACTB and HMBS were comprehensively included in the three most stable RGs, whereas TBP was placed in the three least stable RGs. Additionally, SDHA was identified as a stable RG by three algorithms, excluding BestKeeper, whereas GAPDH was among the least stable RGs. Based on the analysis of each algorithm, RefFinder provided a comprehensive result as the geomean of ranking values for the stability of each RG (Fig. 3). The stability rankings using RefFinder were as follows: ACTB ( most stable), HMBS, SDHA, PPIA, 18S, HPRT1, GAPDH, and TBP (least stable). Notably, classical RGs such as GAPDH and 18S showed less and moderate stability, respectively, in bovine blastocysts at different developmental speeds.

Table 1. Stability rankings of the eight candidate RGs.

Rank *	Comparative ΔCt		GeNorm		BestKeeper			NormFinder
	Gene	Average SD	Gene	Stability value	Gene	SEM (± Ct)	CV (% Ct)	Gene	Stability value
1	ACTB	0.719	HMBS, SDHA	0.470	ACTB	0.552	2.03	ACTB	0.340
2	HMBS	0.763			PPIA	0.583	2.14	HMBS	0.452
3	SDHA	0.765	ACTB	0.537	HMBS	0.656	2.01	SDHA	0.470
4	PPIA	0.859	PPIA	0.644	GAPDH	0.709	2.42	PPIA	0.613
5	18S	0.888	HPRT1	0.707	SDHA	0.775	2.44	18S	0.648
6	HPRT1	0.888	18S	0.760	18S	0.850	2.45	HPRT1	0.662
7	TBP	0.969	TBP	0.811	HPRT1	0.854	2.58	TBP	0.778
8	GAPDH	0.985	GAPDH	0.854	TBP	0.867	2.52	GAPDH	0.806

^* Stability values closer to 1 indicate increased stability, and conversely, greater instability as they approach 8. RG, reference gene; SD, standard deviation; SEM, standard error of the mean; CV, coefficient of variance; Ct, cycle threshold.

Fig. 3.

The geometric mean of ranking value for eight candidate reference genes from RefFinder. Based on the stability results of the RGs analyzed using GeNorm, NormFinder, BestKeeper, and the comparative ΔCt method in Table 1, RefFinder calculated the geometric mean of ranking to determine the most stable RGs in bovine blastocysts with different developmental speeds: ACTB was the most stable RG (with the lowest value). RG, reference gene.

Application of the validated RGs to normalization

To validate the stability of RGs and emphasize the critical importance of appropriate RG selection in each experimental condition, we analyzed the expression levels of ICM (OCT4 and SOX2) and TE (CDX2, GATA3, and IFNτ) markers using the most stable (ACTB), moderately stable (18S), least stable (TBP), and classical but less stable (GAPDH) RGs identified using RefFinder (Fig. 4). For ICM-specific markers, normalization with ACTB revealed significantly (P < 0.05) reduced relative expression of SOX2 and OCT4 in L-BL compared with that in E-BL and M-BL, consistent with the pregnancy rate patterns (Fig. 1E). In contrast, compared with the results from ACTB, normalization with 18S altered the intergroup significance, while TBP generated no significant differences between the groups (Fig. 4A and 4B). For TE markers, ACTB-normalized CDX2 and GATA3 showed no significant differences across developmental groups, whereas unstable RGs indicated significantly (P < 0.05) elevated expression in L-BL (CDX2 by 18S; GATA3 by 18S and TBP), contradicting the low pregnancy rate (Figs. 4C and 4D). All RGs indicated a significant increase in IFNτ expression from E-BL to L-BL (Fig. 4E); however, fold-change magnitudes varied drastically between E-BL and L-BL depending on the RGs: 19.6× (ACTB), 28.2× (18S), 7.6× (GAPDH), and 34.5× (TBP). GAPDH, while yielding ICM marker results similar to ACTB, introduced divergent TE marker patterns (GATA3 and IFNτ) compared with the results obtained using ACTB. These findings demonstrate that unstable RGs may generate unreliable or distorted gene expression profiles, underscoring the necessity of pre-experimental RG validation to ensure accurate data interpretation in embryo research.

Fig. 4.

The application of validated reference genes to normalization. The relative expression level of inner cell mass markers (OCT4 and SOX2) and trophectoderm markers (CDX2, GATA3 and IFNτ) of bovine blastocysts were normalized against the most stable (ACTB), moderately stable (18S), least stable (TBP), and classical (GAPDH) RGs. ^{a, b, c} Significant differences between groups are denoted by different letters above the bars (P < 0.05). Graphs are presented as the mean ± SEM. RG, reference gene.

Discussion

Studying gene expression in bovine blastocysts is crucial for enhancing our fundamental reproductive physiological knowledge and for developing strategies to improve ET efficiency by evaluating blastocysts that cannot be assessed through morphological evaluation [9, 10]. Although RNA sequencing (RNA-seq) has become a revolutionary tool for gene expression analysis in recent molecular biological studies, the importance of qPCR in blastocyst studies should not be overlooked. RNA-seq is expensive, requires complex data analysis, and may introduce noise in low-expression genes. In contrast, qPCR offers high sensitivity, reliability, and accuracy when measuring mRNA levels in small samples such as single blastocysts [6, 27]. It also allows the targeted analysis of a few specific genes crucial for embryo quality assessment [28]. However, PCR-based analysis of single blastocysts faces numerous challenges, including extremely low DNA content, risk of sample loss, limited PCR amplification, a limited number of genes that can be analyzed, and an increased risk of contamination [29, 30]. Moreover, the limited RNA content in single blastocysts poses challenges for both RNA-seq and qPCR, including potential sample loss and reduced amplification efficiency [30, 31]. Notably, recent research has shown that using a robust statistical approach, the pool of well-known RGs for PCR can be as effective as the genes selected from RNA-Seq data [32]. Therefore, this study focused on the selection of appropriate RGs for qPCR using four well-known robust statistical approaches because of their practicality in embryo research. qPCR enables the real-time monitoring of amplification, is more cost-effective for analyzing a limited number of genes, and is better suited for the minute quantities of RNA obtained from single blastocysts [30]. Careful experimental design, data interpretation, and preliminary analysis of the most stably expressed RGs under each experimental condition are crucial to ensure the accuracy and reliability of future experiments.

Preimplantation embryonic development involves complex gene expression patterns that are tightly regulated in a stage- and time-dependent manner. During oocyte maturation, numerous mRNAs and proteins are stored to support early embryonic development. Fertilized embryos undergo morphological changes, including cleavage division, compaction, and blastocyst formation, accompanied by embryonic genome activation [10]. Further, even when blastocysts are formed, their ability to implant can vary depending on their developmental speed. Research has consistently demonstrated that early-developing blastocysts, particularly those forming on days 6–7 of IVC, exhibit superior pregnancy rates compared to their late-developing counterparts (days 7–8 or later) across various cattle breeds [1, 2, 8]. Blastocysts at different developmental stages also exhibited distinct gene expression patterns. A previous study noted that differences in the mRNA expression patterns of stress-related (SOX, MnSOD, BAX, IFtau, and G6PD) and development-related (Glut5, Cx 43, IGF-II, and IGF-IR) genes were observed between early- and late-developing bovine embryos, suggesting that embryo developmental speed influences gene expression patterns, reflecting the response of the embryo to in vitro culture conditions [33]. Importantly, although RGs are believed to have constant and stable expression while unaffected by any experimental conditions, they show differential expression during early embryonic stages, complicating the accurate assessment of gene expression [6, 10, 14]. This highlights the need to carefully select RGs when studying preimplantation development. The present study confirmed that bovine blastocysts with a late developmental speed (L-BL), despite having a similar morphology to their counterparts, had lower pregnancy rates than those with E-BL and M-BL. The data also revealed that RG transcript levels differed between blastocysts with different developmental speeds, despite equalizing the initial sample amounts to a single blastocyst. By evaluating eight candidate RGs using RefFinder, ACTB was found to be the most stable RG in blastocysts with different developmental speeds, whereas other traditional RGs such as TBP, GAPDH, and 18S were found to be less stable.

In previous studies, when normalizing target genes using various RGs with different stability levels, unstable RGs often led to conclusions that differed from generally accepted biological facts [10, 11, 14]. Therefore, this study aimed to evaluate the potential impact of normalizing target genes using RGs with different levels of stability. SOX2, the ICM target gene in this study, plays a crucial role in the normal development of bovine embryos, first appearing at the 16-cell stage and becoming restricted to the ICM at the blastocyst stage [22, 23]. Further, SOX2 is essential for maintaining pluripotency and ICM formation, and its expression level is a key determinant of embryo quality and developmental competence; according to previous studies, L-BL, which has a lower implantation rate, showed a significant decrease in the proportion (%) of SOX2-expressing ICM within the blastocyst compared with that in the E-BL, which exhibits a high implantation rate, and consequently, the ICM/TE ratio (%) was also reduced [19, 22, 23]. OCT4 is primarily expressed in the ICM and partially in the TE of bovine blastocysts, where it is involved in regulating TE differentiation through the control of CDX2 expression and ICM maintenance through the regulation of other pluripotency markers such as NANOG [34]. Therefore, OCT4 deficiency or insufficiency leads to developmental arrest prior to blastocyst formation, failure to establish a normal ICM, and downregulation of gene clusters associated with stemness [35]. These findings suggest that SOX2 and OCT4 may serve as valuable molecular markers for assessing the quality of bovine blastocysts [19]. As shown in Fig. 1E, L-BL exhibited lower pregnancy rates than those of E-BL, indicating a decrease in the quality of late-developing blastocysts. Accordingly, when SOX2 and OCT4 were normalized using ACTB (Figs. 4A and 4B), the results were consistent with the observed pregnancy rates. However, when normalization was performed using the least stable RG (TBP) and moderately stable RG (18S), the expression patterns of both target genes differed from the results obtained with ACTB. Similar results were observed in the analysis of TE-specific genes (CDX2, GATA3, and IFNτ). CDX2 is a key transcription factor that ensures structural stability of the blastocyst by regulating TE differentiation and induces the expression of signals (such as IFNτ) essential for maintaining pregnancy [25]. Notably, reduced expression (or deficiency) of CDX2 in bovine blastocysts leads to decreased expression of TE-related genes (e.g. GATA3, IFNτ), reduced TE cell numbers, and greater problems with implantation and development [36]. Along with CDX2, GATA3 is specifically expressed in the TE of bovine blastocysts, directly binds to the CDX2 gene region to regulate its expression, and plays an important role in maintaining and differentiating the TE lineage program [25, 37]. Notably, in GATA3-deficient bovine blastocysts, NANOG signals in the ICM are reported to be weakened, resulting in reduced embryo quality [25]. The expression pattern of GATA3 is reported to be similar to that of CDX2 [36]. In the present study, when CDX2 and GATA3 were normalized using ACTB, no significant differences were observed among the bovine blastocysts at different developmental speeds. In contrast, when normalized with 18S and TBP, the expression of CDX2 and GATA3 was significantly increased in L-BL compared to that in E-BL, which had the lowest implantation rate, thus contradicting established knowledge [36] (Figs. 1E, 4C, and 4D). IFNτ secreted from the TE of bovine blastocysts acts as an implantation signal sent from the embryo to the mother, preventing luteolysis and maintaining progesterone secretion [26]. However, many studies emphasize that the precise timing of IFNτ expression, rather than its quantitative increase, is a key factor for successful pregnancy [38]. According to research, E-BL has higher developmental competence but lower IFNτ secretion, whereas L-BL has lower developmental competence but higher IFNτ secretion, indicating a negative correlation between IFNτ secretion and embryo quality [39]. Therefore, the amount of IFNτ secreted by the blastocyst can serve as an important indicator for assessing implantation competence. According to the results in Fig. 4E, all four selected RGs showed a significant increase in IFNτ from E-BL to L-BL. However, compared with the fold-change results between E-BL and L-BL presented by ACTB, those obtained with 18S and TBP were overestimated, whereas GAPDH underestimated the results. As an appropriate amount of IFNτ at the proper timing is necessary for a high pregnancy rate [38], unstable RGs can result in over- or underestimation, leading to incorrect assessment of embryo quality. Therefore, these findings strongly suggest that selecting an incorrect or unvalidated RG in blastocysts with low amounts of mRNA can result in misinterpretation of target gene expression.

Efforts to identify reliable RGs in pre-implantation embryos for each animal model are ongoing, including mice [13, 14], cows [3, 6, 16, 27], buffaloes [4, 5, 16], pigs [10], rabbits [12], sheep [40], and horses [15]. ACTB is associated with cytoskeletal structural proteins and is commonly selected as the RG to evaluate target genes [6]. Similar to our results, ACTB has been found to be a stable RG in the embryos of various animals, including the developing embryos of mice [13] and blastocysts of horses [15]. However, depending on the experimental conditions, ACTB can also be one of the least stable RGs. Indeed, ACTB has been noted as the least stable RG in bovine blastocysts produced by various biotechnological methods, including IVF, intracytoplasmic sperm injection (ICSI), and somatic cell nuclear transfer (SCNT) [3], in bovine embryos at different developmental stages (2-cell, 8-cell, and blastocyst) [6], and in mouse embryos produced both in vitro and in vivo [13]. GAPDH is regarded as being continuously expressed in most cells owing to its involvement in various cellular functions and fates, including glycolysis; therefore, GAPDH has been classically used as an RG, similar to ACTB [6]. Although GAPDH was shown to be a less stable RG in the present study, assessments of this gene in previous studies have varied widely. GAPDH was found to be the most reliable gene in bovine blastocysts produced by IVF, ICSI, and SCNT, as well as in buffalo oocytes and embryos during developmental stages [3, 4]. Further, although not ranked as the most stable, GAPDH was considered sufficiently stable in evaluations under various conditions of preimplantation embryos, ranking within the top three among more than ten candidate RGs [5, 15, 16, 40]. Reflecting on its fundamental cellular functions related to transcription, TBP is also expected to have consistent RNA expression levels, similar to other classical RGs such as ACTB and GAPDH. TBP stability in preimplantation embryos has shown conflicting results across species, ranking as the least stable in mouse embryos with different ploidies at various developmental stages [14], similar to our results, but the most stable in ovine embryos during parthenogenetic development [40]. Likewise, considering that the suitability of RGs for normalization can vary depending on experimental conditions, evaluating the stability and selecting appropriate RGs for each qPCR study condition is a prerequisite.

Taken together, various internal and external factors can exert intricate influences on the quality of embryos produced in vitro and on their ability to result in successful pregnancy outcomes. This study confirmed that even if blastocysts had a similar morphology, their RG expression levels could differ based on their developmental speed (E-BL, M-BL, and L-BL). Furthermore, ACTB was found to be the most suitable RG for the analysis of blastocysts by PCR. Overall, these findings will be useful in establishing gene expression experiments for a more successful ET strategy with IVF blastocysts in the future.

Conflict of interests

No potential conflicts of interest relevant to this article were reported.