All animal sample procedures were approved and performed in accordance with institutional guidelines.

Bovine tissue specimens, weighing 200–300 mg per sample, were obtained in triplicate from the foreleg muscle of a single adult male Simmental. The samples were immediately stored in liquid nitrogen and were moved into the freezer at –80 °C to be used for the experiment. All animal sample procedures were approved and performed in accordance with institutional guidelines.

High-quality total RNA was extracted from bovine foreleg muscle using the TRIzol Reagent (Invitrogen, USA). Briefly, approximately 1 mg of total RNA from muscle tissue was used to isolate polyA mRNA. The first-strand cDNA was synthesized using the All-in-one™ First-Strand cDNA Synthesis Kit (GeneCopoeia) according to the manufacturer's protocol. The ds-cDNA (double stranded-cDNA) was then divided into two parts, one was used to prepare clones in the pGADT7-Rec vector, and the other was used for PCR-amplification of full-length CMYA1 and the Xin-repeat fragment of CMYA1.

To construct the bovine muscle cDNA GAL4-AD yeast two-hybrid mating library, the yeast Y187 strain (Clontech) was co-transformed with the purified ds-cDNAs and the linearized pGADT7-Rec vector using the Make Your Own “Mate & Plate™” Library System technology (Cat.No.630490). Selection was performed using SD/-Leu agar medium. To identify the cDNA fragment length, a certain number of colonies were picked randomly and PCR-amplified with the common primers (AD-F CGCGTTTGGAATCACTACAGGG and AD-R GGAGACTTGACCAAACCTCTGGCG) (Oligo Synthesis was in Genewiz). The number of clones screened was calculated by counting the colonies from the DDO plates (dropout medium that lacks both leucine and tryptophan). No. of screened clones = No. of cfu/ml of diploids × resuspension volume (mL) (standard ≥ 1 million cfu). Mating Efficiency% = No. of CFU/mL of limiting partner × 100 (5% ≥ standard ≥ 2%). The resultant transformants were pooled and frozen stored at –80 °C. After thawing, the library titer was calculated. Library titer = No. of CFU/mL of diploids (standard ≥ 1 × 10⁷ CFU/mL).

To construct the yeast expression plasmids, NCBI BLAST was used to identify the conserved Xin-repeat region (144–2612 bp). Using the bovine muscle cDNA as a template, the full-length CMYA1 CDS (144–5606 bp) were PCR-amplified and directionally cloned into the pGBKT7 vector (Clontech) usingthe EcoRI and SalI restriction sites. The Xin-repeat region (144–2612 bp) was cloned using the EcoRI and BamHI restriction sites (Fig. 1A, Table 2). The plasmids pGBKT7-CMYA1 and pGBKT7-Xin repeats were transformed into DH5α competent cells for DNA preservation and proliferation. The resulting plasmids were transformed into the Y2HGold strain following the Clontech Yeastmaker™ Yeast Transformation System 2 User Manual (Cat#630439). To test the autoactivation and the toxicity of the bait proteins, yeast transformed with 100 ng of the bait vectors for CMYA1/Xin repeats were grown on SD/-Trp (dropout medium lacking tryptophan), or SD/-Trp/X (SD/-Trp supplemented with 40 μg/mL X-α-Gal), or SD/-Trp/X/A (SD/-Trp supplemented with 40 μg/mL X-α-Gal and 125 ng/mL aureobasidin A) at 30 °C for 3–5 days. The growth of white colonies on SD/-Trp and SD/-Trp/X mediums and no colony growth on SD/-Trp/X/A plates could indicate no autoactivation. White colonies that grow to a similar size, and at the same speed, as the pGBKT7 vector could indicate that there is no toxicity. Only bait vectors that show no autoactivation activity or toxicity could be used for the Y2H screen.

In our study, we performed a yeast two-hybrid screen twice following the protocol of the Clontech Matchmaker™ Gold Yeast Two-Hybrid System (Cat.No. 630489). First, for the initial selection of positively interacting clones, the bait strains that contained the pGBKT7-CMYA1 or pGBKT7-Xin repeats vectors were co-cultured with the bovine muscle cDNA library in a high-concentration YPDA (2×) liquid medium ideal for yeast mating to the fusion cell, and selection was performed using DDO plates. The positive fusion colonies were identified because they could grow on DDO/X/A (dropout medium of lacking both leucine and tryptophan supplemented with X-α-Gal and aureobasidin A) and appeared blue because of reporter genes of Leu/Trp/X-a-Ga/Aba that were expressed when interactions occurred between bait and prey. To eliminate false positives, all positive colonies were transformed and plated onto a more restrictive dropout minimal media. All blue colonies were picked using sterile pipet tips and streaked onto QDO/X/A plates (dropout medium lacking adenine, histidine, leucine, and tryptophan, supplemented with X-α-Gal and Aureobasidin A).

To identify the preys that were genuine positives, the prey inserts were sequenced using primers for the T7 promoter and were BLASTed on NCBI. The sequencing results of the positive prey inserts were BLASTed against the bovine genome (Bos_taurus_UMD_3.1.1 reference Annotation Release 104) to identify the corresponding bovine genes to which the sequence reads were in-frame. The majority of interacting proteins identified were partial segments of the CDS. To verify binding between the positive prey proteins and the bait proteins, CMYA1 and Xin repeats, a one-to-one mating Y2H assay was performed following the small-scale transformation protocol. The completed CDs for three candidate prey proteins were cloned into the vector pGADT7 and transformed into the yeast strain Y187. The vectors were co-cultured with bait-CMYA1 or bait-Xin repeats in YPDA media and screened using DDO/X or QDO/X/A medium using pGBKT7–53 and pGADT7-T as positive controls and pGBKT7-Lam and pGADT7-T as negative controls. The transformants were plated on DDO/X or QDO/X/A medium for 3–5 days at 30 °C for selection. Construction of the pGADT7-prey vectors and transformation into the Y187 stain with the GAL-BD was performed using the same methods that were described before (Fig. 1B). Growth of blue colonies on both DDO/X and QDO/X/A/confirmed that the candidate prey protein interacted with the bait; in contrast, in the control group, only white colonies grew on DDO/X, and no colonies grew on QDO/X/A.

Currently, to the best of our knowledge, there is no cDNA library of bovine muscle available to us for the detection of PPIs using the Y2H system. In this study, we have successfully generated a high-quality bovine muscle cDNA library for use in a GAL4 Y2H screen, which will be of great value to the study the regulatory networks that function during muscle development. The quality of the library was assessed by the following criteria: the number of clones screened, the sizes of the inserts, and recombination efficiency. The number of screened clones and recombination efficiency were calculated using gradient dilution and colony counting. In total, the volume of the cDNA library was 40 mL. The number of screened clones was 2.8 million, which satisfied the demand for more than 1 million clones (Fig. 2A, Table 1). To check the inserts size and to analyze the recombination efficiency, we performed PCR on approximately 100 random colonies from the library using pGADT7 specific primers (Table 2). Gel electrophoresis showed that the insert sizes of the muscle library ranged from 0.2 to 2.0 kb, and the average size was approximately 0.6 kb (Fig. 2C, Table 1). The mating efficiency was 3.27%, which satisfied the standard (2% ≥ standar ≥ 5%) (Fig. 5A, Table 1). The cDNA library titer was 3.5 × 10⁷ CFU/mL (Fig. 2B, Table 1).

Previously, we obtained the bovine CMYA1 protein partial sequence that is 1,132 aa with a predicted molecular mass of 124 kDa (GenBank: ADM45307.1). Conservation analysis was performed using BLAST on NCBI, and we found that the bovine CMYA1 protein is 73% identical (81% identity within the Xin repeat region) to the corresponding human protein (1,121 amino acids, 122 kDa, Sequence ID: NP_919269.2). Bovine CMYA1 also shares 80% identity within the Xin repeat region of mouse (Sequence ID: NP_035854.3). Whereas nearly 50% of the CMYA1N-terminus consists of 14 Xin repeats (Fig. 3A), there is only one Xin-repeat in C-terminus of CMYA1. Because it has been shown that the minimal number of Xin-repeats that have function is at least 3 [17], there is little chance that the one Xin-repeat in C-terminus is functional. Therefore, this one Xin-repeat was not targeted in this research. The consensus sequence of Xin-repeats was GDV(K/Q/R/S)XX(R/K/T)WLFET(Q/R/K/T)PLD (Fig. 3B).

To identify potential Xin-repeat binding proteins in bovine skeletal muscle, a yeast two-hybrid screen was performed on a bovine muscle cDNA library using full-length CMYA1 and Xin-repeats as the bait (Fig. 4A). CMYA1/Xin-repeats do not cause autoactivation or toxicity when introduced in yeast (Fig. 4C). A total of 2.8 million independent clones were screened producing 670 clones that interact with the full-length CMYA1 bait and 900 clones that interact with the Xin-repeats bait (Fig. 5B, Table 1). Positive clones were PCR-amplified, and inserts that were larger than 300 bp were selected for sequencing. After performing the screen twice, a total of 27 putative interacting proteins were identified (Table 3). There were 14 proteins that interacted with CMYA1, 16 proteins that interacted with the Xin-repeats, and three proteins (RPL35A, RPL21 and EIF3G) that interacted with both CMYA1 and Xin-repeats making them candidate interacting proteins (Table 3).

To determine if the three candidates (RPL35A, RPL21 and EIF3G) were CMYA1/Xin-repeats interacting proteins, the full-length CDS of each candidate was cloned into the pGADT7 prey vector to be used for a mating experiment (Fig. 6). One-to-one Y2H assays were performed between the three prey candidates and the two baits (CMYA1/Xin repeats). Notably, RPL35A, RPL21 and EIF3G all interacted with both CMYA1 and the Xin-repeats, which were also in accordance with the results from when the library was screened previously (Fig. 7).