Gene expression profiling of upregulated mRNAs in granulosa cells of bovine ovulatory follicles following stimulation with hCG

Normal cycling crossbred heifers were synchronized with one injection of PGF_2α (25 mg, im; Lutalyse, Upjohn, Kalamazoo, MI) given in the presence of a corpus luteum (CL). Behavioral estrus was monitored at 12 h intervals, from 48 h to 96 h following the PGF_2α injection. From the time of PGF_2α injection until ovariectomy, ovarian follicular development was monitored by daily transrectal ultrasonography performed with a real-time linear scanning ultrasound diagnostic system (LS-300; Tokyo Keiki Co, Ltd., Tokyo, Japan) equipped with a 7.5-MHz transducer probe. At each examination, the diameter of the CL and individual follicles ≥ 4 mm were measured at their largest cross-sectional area using internal calipers. Following estrus synchronization by PGF_2α, heifers were randomly assigned to the dominant follicle group (DF; n = 4) or the ovulatory hCG-induced follicle group (OF; n = 4). In the DF group, the ovary bearing the DF on the morning of day 5 of the estrous cycle (day 0 = day of estrus) was obtained by ovariectomy (via colpotomy). The DF was defined as >8 mm by ultrasonographic measurement and growing while subordinate follicles were either static or regressing [9]. The mean diameter of DF measured at the surface of the ovary was 10.4 ± 0.3 mm. The OF were obtained following an injection of 25 mg of PGF_2α (Lutalyse) on day 7 to induce luteolysis, thereby maintaining the development of the DF of the first follicular wave into a preovulatory follicle [10]. An ovulatory dose of hCG (3000 IU, iv; APL, Ayerst Lab, Montréal, QC) was injected 36 h after the induction of luteolysis, and the ovary bearing the hCG-induced OF was collected by ovariectomy at 24 h after hCG injection. The mean diameter of OF was 12.9 ± 0.3 mm. Follicular fluid and mural GC with whole cumulus-oocyte complex were collected separately from individual DF or OF as described previously [9]. Additionally, GC and follicular fluid were collected from 2 to 4 mm follicles that were obtained from slaughterhouse ovaries representing a total of three pools of 20 small follicles (SF). These experiments were approved by the Animal Ethics Committee of the Faculty of Veterinary Medicine of the Université de Montréal. Concentrations of progesterone (P₄), estradiol-17β (E₂) and their ratio (P₄/E₂) were analyzed by RIA of follicular fluid as previously described [9]. The E₂/P₄ ratios were calculated for each sample: 1) 0.008, 0.06 and 0.01 for the three SF pools; 2) 17.3, 19.7, 14.3 and 63.4 for individual DF at day 5 (n = 4); and 3) 0.44, 0.87, 0.64 and 0.27 (n = 4) for individual OF. CL at day 5 of the estrous cycle were obtained by ovariectomy from cows following ultrasound monitoring of follicular development and estrus synchronization as described above. The CL were dissected from the ovarian stroma, frozen in liquid nitrogen, and then stored at −80 °C until RNA extraction. Total RNA was isolated from GC or CL by homogenization in lysis buffer (4 M guanidium isothiocyanate, 0.5% Na-N-laurylsarcosine, 25 mM NaCitrate, pH 7), and total RNA sedimented on a cesium chloride cushion by ultracentrifugation as previously described [9]. The concentration of total RNA was quantified by measurement of optical density at 260 nm, and quality was evaluated by visualizing the 28S and 18S ribosomal bands following electrophoretic separation on a formaldehyde denaturing 1% agarose gel with ethidium bromide.

To compare gene expression in GC collected from OF versus DF, the suppression subtractive hybridization (SSH) was performed as previously described [8]. Identical amounts of total RNA (2 μg) from four OF or four DF were pooled within treatment groups to decrease inter animal variation. To generate sufficient amounts of double-stranded cDNA for an SSH experiment, both OF and DF cDNAs were amplified separately using the SMART PCR cDNA synthesis kit. One microgram of total RNA from each pooled group was reverse transcribed with an oligo-dT30 primer [CDS: 5′-AAGCAGTGGTAACAACGCAGAGTACT(30)(A/C/G/T) (A/G/C)-3′] and PowerScript (BD Biosciences Clontech) to generate the first strand cDNA. Second cDNA strands were produced with the SMART II 5′-anchored oligo and PCR-amplified for 15 cycles using Advantage 2 DNA polymerase (BD Biosciences Clontech). PCR-generated OF and DF cDNAs were digested with RsaI to generate blunt-ended cDNA fragments [from 0.2 to 2 kilobases (kb)]. The OF cDNAs were subtracted against DF cDNAs (forward reaction: OF-DF) using PCR-Select cDNA subtraction technology (User manual: PT1117-1; BD Biosciences Clontech) [8], and in a parallel experiment, the DF cDNAs were subtracted against the OF cDNAs (reverse reaction: DF-OF). The efficiency of subtraction was analyzed by comparing the abundance of cDNAs before and after subtraction by PCR using bovine gene-specific primers for a gene known to be induced by hCG, such as prostaglandin-endoperoxide synthase 2 (PTGS2), and a gene known to be downregulated by hCG, such as cytochrome P450, family 19, subfamily 1 (CYP19A1). PCR amplification was performed using Advantage 2 DNA polymerase (BD Biosciences Clontech), and 5-μl aliquots were removed following determined numbers of PCR cycles. The amplified products were resolved on a 2% agarose gel in TAE buffer (40 mM Tris-acetate, 1 mM EDTA, 0.5 μg/ml ethidium bromide). The difference in the number of cycles needed to generate an equal amount of the corresponding PCR product in subtracted and unsubtracted samples served to indicate the subtraction efficiency.

The subtracted cDNAs were cloned into the pT-Adv plasmid (BD Biosciences Clontech) to construct the OF-DF subtracted library and used to transform competent TOP10F’ Escherichia coli as previously described [11]. The subtracted OF-DF cDNA library (940 individual colonies) was used to establish macroarrays for differential screening following previously described methodologies [8]. The insert of each cDNA clone was amplified in 96-well plates by PCR (28 cycles) using the PCR-nested primers 1 and 2R and AmpliTaq DNA polymerase (Roche Molecular Systems Inc., Laval, QC). To establish the cDNA macroarrays, an aliquot of each amplification product was denatured in 0.3 M NaOH with 5% bromophenol blue, and 10 μl were vacuum transferred with a 96-well dot-blot apparatus onto nylon membranes (Hybond-N⁺, GE Healthcare Life Sciences), which were then exposed to 150 mJ ultraviolet light (UV) to perform DNA cross-linking (Gs Gene Linker; Bio-Rad, Mississauga, ON). Control cDNAs (CYP19A1, PTGS2) were transferred onto the macroarrays. For each 96-well plate, two identical cDNA macroarray replicate membranes were generated. The OF-DF and DF-OF cDNA pools were used to generate complex hybridization probes for differential screening of macroarrays of the OF-DF cDNA library. Probes were obtained by performing the secondary nested PCR and were then purified (QIAquick PCR Purification Kit; Qiagen Inc., Mississauga, ON). To prevent nonspecific interaction of the probes to cDNAs on macroarrays during hybridization, the adaptors were removed by three successive digestions with AfaI, SmaI, and EagI; the cDNA pools were again purified (QIAquick; PCR Purification Kit, Qiagen Inc.) and 100 ng were labeled with [α³²P]-dCTP by random priming (Megaprime DNA Labeling System; GE Healthcare Life Sciences). The radioactive probes were purified (QIAquick Nucleotide Removal kit; Qiagen Inc.) and quantified using a beta counter. The hybridization and washing conditions of macroarrays were performed as previously described [8]. Equal amounts of each heat-denatured cDNA probe (OF-DF, DF-OF) were used to hybridize each replicate of the OF-DF macroarray membrane. Following washing, membranes were exposed to a phosphor screen for 4 h and the images were digitized (Storm 840; GE Healthcare Life Sciences). The differentially hybridizing cDNA clones were characterized by DNA sequencing and their differential expression profiles were further validated by semiquantitative RT-PCR analysis from independent follicles obtained at different developmental stages.

Identification of differentially expressed cDNAs were obtained by sequencing. The cDNA clones identified as differentially expressed by the OF-DF subtracted probe were amplified by PCR for 15 cycles with the corresponding PCR-nested 1 and PCR-nested 2 oligos from the PCR product generated initially for the macroarrays. The PCR product was purified (Qiagen Inc.) and verified by agarose gel analysis for the presence of a single cDNA band before proceeding with sequencing. Sequencing reactions were performed on cDNA clones via the dideoxy sequencing method (Big Dye Terminator 3.0; ABI Prism, Applied BioSystem, PE, Branchburg, NJ) using the oligos PCR-Nested 1 or 2, and sequencing reactions were analyzed on an ABI Prism 310 sequencer (Applied Biosystems). Nucleic acid sequences were analyzed by BLAST (Basic Local Alignment Search Tool) against GenBank data banks.

The cDNA clones that were identified as differentially expressed in the SSH differential screening experiment were used to compare their differential expression pattern in GC collected from follicles at different developmental stages and CL, using semiquantitative RT-PCR/Southern blot analysis (also called virtual Northern blot analysis; User manual: PT1117-1; BD Biosciences Clontech). To perform semiquantitative RT-PCR/Southern blot analysis, total RNA (1 μg) from GC (SF, DF, OF) or CL (D5) were reverse transcribed using SMART PCR cDNA synthesis technology (BD Biosciences Clontech, Mississauga, ON) as previously described [8]. The cDNA products for each follicle or CL along with molecular weight standards (1 kb ladder, ϕX 174-RF/Hae III and l/Hind III; GE Healthcare Life Sciences) were separated on agarose gel, then transferred onto nylon membrane. Gene-specific probes derived from SSH cDNA fragments were generated by PCR (20 cycles) using the primers PCR Nested 1 and PCR-Nested 2R. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) probe was used as a constitutively expressed gene. Purified cDNA probes were labeled with [α³²P]-dCTP as described above. Semiquantitative RT-PCR/Southern membranes were prehybridized, hybridized, and washed as described [8]. Membranes were exposed to phosphor screen and the images were digitized (Storm 840; GE Healthcare Life Sciences).

Semiquantitative RT-PCR analysis was performed for genes that showed a weak signal by RT-PCR/Southern blot or to analyze differentially spliced transcripts [8]. Gene-specific primers were designed into the open reading frame of the cDNA sequence and are described in Table 1. SMART cDNAs were generated as described above, and 1 μl was used in a 25-μl PCR reaction using the Advantage 2 DNA polymerase. The number of PCR cycles were limited and optimized to fall in the linear range of amplification reaction for each gene to be analyzed. The PCR reactions were separated on a 2% TAE-agarose gel with ethidium bromide, PCR products were visualized by UV, and the images digitized. The digitized signals for each gene obtained either by semiquantitative RT-PCR/Southern blot or semiquantitative RT-PCR were analyzed by densitometry using ImageQuant software (GE Healthcare Life Sciences).

Isolation of full-length bovine periostin (POSTN), cysteine-rich secretory protein LCCL domain containing 2 (CRISPLD2) and lethal (3) malignant brain tumor-like protein (L3MBTL3) cDNAs was performed by screening a size-selected cDNA library for each of the cDNA as described [11], since their respective full-length cDNAs were not experimentally characterized in the bovine species. Initially, the respective sizes of the full-length cDNA were estimated by performing a semiquantitative RT-PCR/Southern blot analysis with cDNA pool generated from hCG-stimulated GC and hybridized with radioactive probes for POSTN, CRISPLD2 or L3MBTL3 derived from the SSH screening experiment (Table 2). Once the size of the full-length bovine POSTN, CRISPLD2 and L3MBTL3 cDNAs were determined, total SMART cDNAs from hCG-stimulated GC were size-fractionated by agarose gel electrophoresis, and cDNAs from 2.5 to 3.5 kb for POSTN, and from 4 to 6 kb for CRISPLD2 and L3MBTL3 were purified and used to construct a size-selected cDNA library based on the pDrive plasmid (Qiagen PCR cloning kit; Qiagen) that was then screened by radioactive hybridization as described [11]. Positive POSTN, CRISPLD2 and L3MBTL3 hybridizing bacterial colonies were grown, their plasmid contents were isolated (QIA-prep, Qiagen), and the size of the cloned cDNA was analyzed following an EcoR1 digestion and gel electrophoresis analysis. The cDNAs were sequenced via the dideoxy sequencing method (Big Dye Terminator 3.0; ABI Prism, Applied BioSystems, PE) that were analyzed on an ABI Prism 310 sequencer (Applied Biosystems). Nucleic acid sequences were analyzed by BLAST against GenBank data banks.

Gene-specific signals generated for semiquantitative RT-PCR/Southern blot and semiquantitative RT-PCR analyses were normalized with corresponding GAPDH signals for each sample. Homogeneity of variance between follicular groups and CL was verified by O’Brien and Brown-Forsythe tests. Corrected values of gene-specific mRNA levels were compared between follicular and CL groups by one-way ANOVA. When ANOVA indicated significant differences (P < 0.05), multiple comparisons of individual means for SF, DF, OF and CL groups were compared by the Tukey-Kramer test (P < 0.05). Pearson’s correlation (P < 0.05) analysis was used to compare amplicon levels for KITLG1 and KITLG2. Data were presented as least-square means ± SEM. Statistical analyses were performed using JMP software (SAS Institute, Inc.).

A cDNA library containing transcripts that are upregulated by hCG in GC was constructed by subtracting DF cDNAs from OF cDNAs (OF-DF). Reverse subtraction was also performed as a control and consisted of OF cDNAs subtracted from DF cDNAs (DF-OF). PCR amplification analysis was used to verify the efficiency of the subtraction procedure by comparing the expression of reference genes such as PTGS2 in OF and CYP19A1 in DF, before and after subtraction. In the OF sample, the PTGS2 PCR product was observed after 18 cycles but was undetectable in the DF sample. In the OF-DF subtracted sample, the PCR-amplified PTGS2 fragment was detected after 13 cycles, revealing that PTGS2 cDNA had been efficiently enriched in the subtracted OF-DF sample when compared with unsubtracted OF sample, confirming the effectiveness of the subtraction (validation of subtraction is presented in a Additional file 1: Figure S1). Subtracted cDNAs were then cloned into plasmid vector to generate the OF-DF cDNA library. Differential hybridization screening was performed on the randomly selected bacterial colonies to eliminate false-positive clones. Colonies were spotted onto two identical sets of macroarrays, and the OF-DF or DF-OF subtracted cDNA preparations were used as probes to hybridize macroarrays. On each membrane, PTGS2 and CYP19A1 cDNAs were spotted as controls. Selection of differentially expressed cDNA clones was achieved by comparing signal intensities between the two macroarrays as defined by the following criteria. Positive clones hybridized with the OF-DF subtracted probes but not with the reverse subtracted probes, DF-OF. Representative differential screening results are illustrated in Additional file 1: Figure S2). Of the initial 940 clones, differential screening identified 279 positives as defined by the selection criteria based on comparing differential intensity of the signal. After visualization on agarose gels of PCR products derived from these clones followed by their sequencing, 261 clones generated sequencing results of adequate quality to be analyzed by BLAST against GenBank databases. Table 2 lists all the compared sequences with an example for each entry that was deposited in GenBank, as well as their frequency of identification by differential screening of the OF-DF library. This comparison revealed that 89.7% (234/261) corresponded to 60 non-redundant known genes, and 10.3% (27/261) corresponded to express sequence tags (EST). The BLAST analysis of these ESTs against the bovine genome revealed that 24 ESTs were associated with transcribed loci, either introns or untranslated regions (UTR) of mRNA, of defined genes as indicated in brackets for each EST (Table 2) whereas three ESTs were considered novel transcribed loci.

Genes identified by macroarray analysis as upregulated in GC following stimulation by hCG were further validated on the basis of their frequency of identification in the OF-DF subtracted library (Table 2). Validation was performed initially by semiquantitative RT-PCR/Southern blot analysis using mRNA samples derived from GC collected from independent follicles at different developmental stages (SF, DF, OF) and CL obtained at Day 5 of the estrous cycle. For these analyses, mRNAs were reverse transcribed into cDNAs, separated on agarose gel, and transferred onto membranes that were then hybridized to cloned SSH cDNA fragments as described in Table 2. Messenger RNA expression for tissue plasminogen activator (PLAT), early growth response 1 (EGR1), tumor necrosis factor alpha-induced protein 6 (TNFAIP6), regulator of G-protein signaling 2 (RGS2), spermidine/spermine N1-acetyltransferase 1 (SAT1), glutamine-fructose-6-phosphate transaminase 2 (GFPT2), KIT proto-oncogen receptor tyrosine kinase (KIT) and periostin (POSTN), showed a statistically significant increase or induction by hCG in GC of OF (Fig. 1). When mRNA signals observed for GC of OF were compared with that of DF, expression of RGS2 and POSTN were induced, whereas for other genes, the increase was 32.8-fold for PLAT, 24.7-fold for EGR1, 7.9-fold for TNFAIP6, 18-fold for SAT1, 26.8-fold for GFPT2 and 23.6-fold for KIT.

A semiquantitative RT-PCR assay was performed on selected genes described in Table 2 to confirm and compare their mRNA differential expression pattern in GC of hCG-induced OF. For this procedure, cDNAs were generated from total RNA, and the number of PCR cycles was optimized for each gene. All of the genes analyzed in this manner showed a statistically significant increase in hCG-treated GC of OF compared to that of DF (Fig. 2). Messenger RNA expression for KIT ligand (KITLG1) and isoform 2, KITLG2, each showed a 6.7-fold increase in GC of OF compared to that of DF (Fig. 2a). Intensity of amplicon signals obtained for KITLG1 and KITLG2 were highly correlated (r = 0.956). The FK506 binding protein 5 (FKBP5) mRNA was mainly observed in OF with a 8.6-fold increase in expression when compared to DF. Conversely, FK506 binding protein 4 (FKBP4) used as control, showed no difference in expression between follicular stages but was reduced by 2.8-fold in CL when compared to OF. Syndecan 4 (SDC4) transcript was mainly expressed in GC of OF wherein a 4.5-fold higher expression level was observed compared to DF. Calpain 2 (m/II) large subunit (CAPN2) transcript was detected in all groups and showed a 7.1-fold higher expression level in GC of OF compared to DF. The prolyl 4-hydroxylase subunit alpha 3 (P4HA3) mRNA was mainly observed in OF compared to other groups, and showed a 13.6-fold increase in OF compared to DF (Fig. 2b). Prosaposin (PSAP) mRNA was slightly increased by 1.9-fold in OF when compared to DF. Tenascin C (TNC) was induced in GC of OF when compared to DF and SF. ADAM metallopeptidase with thrombospondin type 1 motif 9 (ADAMTS9) found as an EST corresponding to intron 28 (Table 2) was upregulated by 24.1-fold in OF compared to DF. Expression of the histone methyl-lysine binding protein (L3MBLT3) mRNA showed an increase of 7.5-fold in OF compared to DF whereas no expression was detected in CL (Fig. 2c). Cystein-rich secretory protein LCCL domain containing 2 (CRISPLD2) mRNA was induced in OF when compared to DF. A-Raf proto-oncogene serine/threonine kinase (ARAF) mRNA was upregulated by 7.2-fold in OF compared to DF, and maestro (MRO) mRNA was increased by 4-fold in OF compared to DF. TIMP metallopeptidase inhibitor 1 (TIMP1) was induced in GC of OF when compared to DF whereas TIMP metallopeptidase inhibitor 2 (TIMP2) was upregulated by only 1.5-fold (Fig. 2d). Matrix metallopeptidase 1 (MMP1) was induced in GC of OF when compared to SF and DF, and was reduced in CL. ADAM metallopeptidase with thrombospondin type 1 motif 1 (ADAMTS1) was upregulated by 18-fold in OF compared to DF. Expression of nudix hydroylase 10 (NUDT10) and nudix hydroxylase 11 (NUDT11) mRNAs where upregulated by 6.5-fold and 15.2-fold, respectively, reaching the lowest level in CL (Fig. 2e). Retinol-binding protein 1 (RBP1) mRNA was upregulated by 2.6-fold in OF, and ubiquitin specific peptidase 53 (USP53) was upregulated by 2.3-fold.

Bovine full-length cDNAs for POSTN, CRISPLD2 and L3MBTL3 cDNAs were not previously experimentally characterized in the bovine species. Since only cDNA fragments were obtained from the screening of the OF-DF subtracted library and that potential alternatively spliced isoforms may be expressed, we undertook their characterization in GC to ascertain the identity of the OF-DF cDNA fragments. Bovine cDNA fragments for POSTN, CRISPLD2 and L3MBTL3 were cloned from the SSH screening of the OF-DF subtracted library as presented in Table 1, and were used as probes to screen by hybridization size-selected cDNA generated from bovine GC that were collected 24 h following injection of hCG. The bovine POSTN cDNA consists of 3072 bp (Genbank: AY445072), and is composed of a 5′-UTR of 42 bp, an ORF of 2511 bp and a 3′-UTR of 519 bp containing two polyadenylation signals followed by a poly(A)⁺ tail. The coding region of bovine POSTN cDNA encodes a protein of 836 amino acids, with a theoretical molecular mass (Mr) of 93.1 kDa and an isoelectric point (pI) of 7. Amino acid homology analysis by PsiBlast revealed an overall identity level of 95% to human (NP_001129406). Bovine POSTN possesses a peptide signal (M¹-A²¹) that may be cleaved between A²¹-N²², a cystein-rich domain called EMI domain (G⁴⁰-A⁹⁴), and four fasciclin domains (T¹¹¹-T²³¹, E²⁴⁶-P³⁶⁷, Q³⁸¹-P⁴⁹⁴, R⁵⁰⁸-P⁶³⁰). Since five POSTN isoforms were reported in human, we verified by RT-PCR and sequencing if other POSTN isoforms were expressed in GC using primers targeting the entire ORF (Table 1). A single full-length cDNA of 2511 pb was characterized.

The bovine CRISPLD2 cDNA consists of 4020 bp (Genbank: AY369781), and is composed of a 5′-UTR of 210 bp, an ORF of 1488 bp, and a 3′-UTR of 2322 bp containing six AU-rich elements (ATTTA) as well as two polyadenylation signals followed by a poly(A)⁺ tail. The coding region of CRISPLD2 cDNA encodes a protein of 496 amino acids, with a theoretical Mr. of 55.6 kDa, pI of 8.2, and an overall identity level of 83% to human (NP_113664) and 76% to mouse (NP_084485) proteins. Bovine CRISPLD2 protein possesses a peptide signal (M¹-G²²) that may be cleaved between G²²-F²³, a cystein-rich secretory protein, antigen 5, and pathogenesis-related protein (known as CAP domain; L⁶⁰-Y²⁰⁰) and two Limulus factor C, Coch-5b2 and Lgl1 domains (LCCL-1: V²⁸⁷-F³⁷⁸; LCCL-2: L³⁸⁸-T⁴⁸²).

The bovine L3MBTL3 cDNA characterized from GC consists of 3892 bp (Genbank: AY437805), and is composed of a 5′-UTR of 205 bp, an open reading frame (ORF) of 2322 bp and a 3′-UTR of 1570 bp containing three AU-rich elements (ATTTA) as well as two polyadenylation signals followed by a poly(A)⁺ tail. The coding region of bovine L3MBTL3 cDNA encodes a protein of 773 amino acids, with a theoretical Mr. of 87.5 kDa, pI of 6, and an overall identity level of 96% to human (GenBank: NP_115814) and 87% to mouse (NP_766375) proteins. Since the bovine protein expressed in GC was missing the last seven amino acids at its carboxy-terminal end, corresponding to ⁷⁷⁴KNSHNEL⁷⁸⁰ when compared to human protein, we verified by RT-PCR and sequencing if other L3MBTL3 isoforms would be expressed in GC by using primers targeting the 3′-end of the open reading frame (Table 1). In parallel, RT-PCR analysis was performed using RNA extracted from a bovine endometrial epithelial and glandular cell line (Endo 8.3; [12]. In GC, a single alternatively spliced transcript was observed coding for the 773 amino acids protein that we previously characterized using the size selected cDNA library screening technique. This truncated L3MBTL3 isoform is referred as transcript 2 (GenBank: AY437805). Conversely, a single transcript was also observed in the endometrial cell line that corresponded to the complete protein of 780 amino acids with a theoretical Mr. of 88.3 kDa and pI of 6.1, referred as L3MBTL3 transcript 1 (GenBank: KT881243). Bovine L3MBTL3 protein isoforms expressed in GC and the endometrial cell line possesses three malignant brain tumor (MBT) repeat domains (MBT-1: M²⁶⁸-K³³⁶; MBT-2: M³⁷⁵-Y⁴⁴²; MBT-3: M⁴⁷⁹-P⁵⁴³), C2HC zinc finger domain (G⁵⁵⁷-S⁵⁸⁶) and a sterile alpha motif domain (SAM; S⁷⁰⁶-K⁷⁷⁰).