The human TMEM263 gene encodes a small protein of 116 amino acids with two predicted transmembrane helices (Fig. 1A). It is highly conserved across vertebrate species, with the fish (Danio rerio), frog (Xenopus tropicalis), chicken (Gallus gallus), and mouse (Mus musculus) orthologs sharing 73, 77, 90, and 97% amino acid identity with the full-length human TMEM263. In humans, the TMEM263 transcript is widely and variably expressed across tissues (Fig. 1B), with liver having the highest expression. The expression data is based on the consensus dataset of normalized expression (nTPM) levels across tissues by combining the Human Protein Atlas (HPA) (28) and Gene-Tissue Expression (GTEx) transcriptomics datasets (29). Reflecting the transcript expression profile, human TMEM263 protein is also widely expressed (www.proteinatlas.org) (28). The mouse Tmem263 is also variably and widely expressed across tissues (Fig. 1C).

Human TMEM263 protein is ∼20 kDa in size when expressed in mammalian HEK293 cells (Fig. 1D). The presence of two putative hydrophobic segments suggests that TMEM263 may be an integral membrane protein. However, different transmembrane prediction programs have given inconsistent results. To experimentally determine whether human TMEM263 is a membrane protein localized to the cell surface or other intracellular membrane compartments, we performed surface biotinylation followed by streptavidin pull-down and immunoblot analysis. The cell-impermeable biotinylation reagent labeled only proteins on the surface of intact cells. This analysis showed that TMEM263 is localized to the plasma membrane when expressed in HEK293 cells (Fig. 1E).

A genetic approach was used to determine the in vivo function of Tmem263. The mouse Tmem263 gene (NM_001013028) consists of three exons, and exon 3 encodes ∼81% of the full-length protein. To generate a loss-of-function mouse model, we used CRISPR/Cas9 method to delete the entire protein-coding portion of exon 3, thus ensuring a complete null allele (Fig. 2A). We performed genomic PCR and DNA sequencing to confirm the genotype and deletion of the Tmem263 gene (Fig. 2A-B). As expected, deletion of the Tmem263 gene resulted in complete absence of the transcript in liver and hypothalamus (Fig. 2C).

The genotype distribution of 82 pups at postnatal day 1 (P1) largely conformed to the expected Mendelian ratio for WT (+/+), heterozygous (+/-), and KO (-/-) mice (Fig. 2D), indicating that Tmem263 is not required for embryonic development. Loss of Tmem263 did not affect suckling at P1, as the milk spot was clearly visible in Tmem263-KO pups (Fig. 2E). No gross anatomical abnormalities in the axial skeleton, appendages, and cartilage were noted in Tmem263-KO pups at P1, indicating that Tmem263-KO mice were born normal (Fig. 2F). The birth weights of pups at P1 were also not different between genotypes (Fig. 2G). Together, these data indicate that Tmem263 is not required for embryonic development and prenatal growth.

From P1 to P14, the weights of Tmem263-KO pups were not different from WT controls (Fig. 2G), indicating no deficit in growth trajectory within the first two weeks of life. However, before and around the time of weaning at P21 when mice started to experience an accelerated growth spurt, coincident with the initiation of GH action (15), clear differences in body weight were noted in Tmem263-KO mice relative to WT (+/+) and heterozygous (+/-) controls (Fig. 2G). By adulthood (8 weeks old), a severe growth failure phenotype was apparent in the Tmem263-KO mice (Fig. 2H).

Both male and female KO mice exhibited the same striking degree of dwarfism, with body weight and body length dramatically reduced relative to WT controls (Fig. 2I-K). Due to their small size, the organ weights (liver, pancreas, heart, kidney, spleen, gastrocnemius, and brain) of the KO mice were also significantly lower relative to WT controls (Figure 2-figure supplement 2). When normalized to their body weight, Tmem263-KO mice also had lower relative weights of liver (male only), spleen (male only), and skeletal muscle (gastrocnemius) and higher kidney and brain weight (Figure 2-figure supplement 1). Partial loss of Tmem263 did not result in haploid insufficiency, as the heterozygous (+/-) mice were indistinguishable from WT littermates across all measured parameters. Both male and female Tmem263-KO mice were fertile, as indicated by viable mouse litters derived from multiple independent matings between male (-/-) and female (-/-) mice. The litter size in general was ∼30-40% smaller, though this observation was based only on a limited number of litters produced. The smaller litter size might reflect potential constrains imposed by the size of the KO dam. Mice born from homozygous KO (-/-) parents appeared largely normal with no gross physical or behavioral abnormalities noted except their pronounced dwarfism. Together, these data indicate that Tmem263 is required for postnatal growth and its absence causes postnatal growth retardation.

In accordance with postnatal growth failure, adult Tmem263-KO mice had much shorter femoral bones compared to WT and heterozygous controls (Fig. 3A-B). Quantification of various bone parameters in the distal femur indicated a marked reduction in trabecular bone volume, trabecular number, trabecular bone thickness, cortical tissue area, and cortical thickness in Tmem263-KO relative to WT and heterozygous mice (Fig. 3C-H, Regions of interest for both trabecular and cortical bone analysis were adjusted in proportion to the length of bone, See Methods). However, normalization of cortical tissue area and cortical thickness to femur length indicated that skeletal geometry was proportional to that in WT littermates (Fig. 3I-J). Histological analysis revealed a much thinner growth plate in Tmem263-KO mice (Fig. 3K). Examination of growth plate morphology in the proximal tibia revealed a reduction in total growth plate length that was secondary to reductions in proliferative zone length in Tmem263-KO mice relative to WT controls (Fig. 3L-M). The hypertrophic zone length was not different between genotypes (Fig. 3N). Together, these data indicate that Tmem263 deficiency severely affects postnatal skeletal growth and acquisition.

The GH/IGF-1 axis plays an essential role in postnatal growth and its deficiency results in growth failure (9, 11, 30). The growth retardation and skeletal dysplasia phenotypes seen in Tmem263-KO mice strongly suggest a potential deficit in the GH/IGF-1 axis. Tmem263-KO mice had a marginal, though not significant, increase in serum GH compared to WT controls (Fig. 4A). However, Serum IGF-1, IGF binding protein 3 (IGFBP3), and IGF acid labile subunit (IGFALS) levels were markedly lower in Tmem263-KO mice relative to WT controls (Fig. 4B-D). Serum insulin levels were significantly lower in KO female mice and trended lower in KO male mice relative to WT controls (Fig. 4E). Random-fed blood glucose levels, however, were significantly lower in Tmem263-KO mice (Fig. 4F). The combination of lower insulin and blood glucose levels suggests enhanced insulin sensitivity. These clinical features are also commonly seen in patients with GHR mutations (11). Given the bone phenotype seen in the Tmem263-KO mice, we also measured serum Ca²⁺ and phosphate levels. While serum Ca²⁺ levels were reduced in Tmem263-KO mice, serum phosphate levels and the Ca²⁺/P ratio were not different between genotypes (Fig. 4G-I). Together, these data indicate that low IGF-1, IGFBP3, and IGFALS levels likely contribute to growth retardation and skeletal dysplasia in Tmem263-KO mice.

The GH/IGF-1 axis (Fig. 5A) required for postnatal growth consists of 1) hypothalamic GH releasing hormone (GHRH) which induces pituitary GH release; 2) secreted GH then binds to and activates GHR in the liver, bone, and other peripheral tissues to induce the synthesis and secretion of IGF-1; 3) the direct effects of GH on bone, together with the endocrine and autocrine/paracrine actions of IGF-1, promote longitudinal bone growth (15, 31-33). A majority (∼75%) of the circulating IGF-1 is secreted by liver (31). Because Tmem263-KO mice had normal to marginally higher GH but low IGF-1 and IGFBP3 levels, the GH/IGF-1 axis is likely disrupted at the liver. Indeed, the expression of Ghr, Igf-1, and Igfals (IGF binding protein acid labile subunit) transcripts was significantly reduced (∼4 fold) in the liver of Tmem263-KO mice relative to WT controls (Fig. 5B). The expression of Igfbp3, however, was not different between genotypes. Consistent with the mRNA data, the hepatic Ghr protein level was similarly and greatly reduced in Tmem263-KO mice compared to WT controls (Fig. 5C-D).

Reduced Ghr transcript and protein expression would diminish the magnitude of GH-induced signaling, resulting in GH insensitivity. To test this, mice were injected with recombinant GH and GH-induced signaling in liver was assessed. Janus kinase 2 (JAK2) and Signal transducer and activator of transcription 5b (STAT5b) phosphorylation are the key essential signaling events downstream of the GHR (27, 34). At baseline when mice were not injected with GH, there were no detectable phospho-Jak2 and phopho-Stat5b signals (Fig. 5E). In contrast, WT male mice injected with GH showed a robust GH-stimulated phosphorylation of Jak2 and Stat5b (Fig. 5F). Relative to WT controls, GH injection elicited a greatly diminished response in Tmem263-KO mice, as indicated by the much lower magnitude (∼60-80% reduction) of Jak2 and Stat5b phosphorylation relative to WT controls (Fig. 5F-G). Together, these data point to reduced hepatic GHR expression and GH insensitivity as potentially contributing to growth failure and skeletal dysplasia seen in the Tmem263-KO animals.

Based on our current data and what is known in the literature (35, 36), impaired GH signaling results in widespread changes in the GH-regulated liver transcriptome of Tmem263-KO male mice. Indeed, 8.6% of the liver transcriptome in Tmem263-KO male mice were significantly altered relative to WT controls, with 1547 protein-coding genes upregulated and 862 protein-coding genes downregulated (Fig. 6A). Among the major classes of genes and pathways most affected by the loss of Tmem263 are those that involved in lipid and steroid metabolism, detoxification process, and xenobiotic handling (Fig. 6B); all of the affected genes and pathways are known to be markedly altered in male mice in which GH signaling is disrupted due to Stat5b deficiency or an altered GH secretory profile (i.e., loss of pulsatile release of GH) (21, 35-40).

It is well known that GH and its distinct secretory profiles (continuous in females vs. intermittent in males) plays a critical role in establishing and maintaining sexually dimorphic gene expression in the liver of male versus female mice (39-43). In male mice, when GH signaling or secretory profile is disrupted, the expression of male-bias genes is dramatically suppressed and, concomitantly, the expression of female-bias genes are upregulated due to their de-repression in male mice (39, 42, 44). This is thought to be the result of altered male-bias (e.g., Bcl6) and female-bias (e.g., Cux2) transcription factor expression that reciprocally regulates the hepatic expression of sex-bias genes in male and female mice (45-47). In male mice with disrupted GH signaling, Bcl6 expression is suppressed and Cux2 expression is markedly upregulated, leading to transcriptional repression of male-bias genes and upregulation of female-bias genes in the male liver (39). Remarkably, Cux2 was one of the most upregulated genes in general and was the most upregulated transcription factor gene in the liver of Tmem263-KO male mice (Fig. 6C). The expression of Bcl6, a male-bias transcription factor genes, were also significantly downregulated in the KO male liver (Fig. 6C). Consequently, all the well-known GH-regulated male-bias genes (growth and metabolism, cytochrome P450, and major urinary protein) were dramatically suppressed due to greatly diminished GH signaling, whereas the female-bias genes (e.g., distinct set of cytochrome P450) were concomitantly and markedly upregulated due to Cux2 overexpression in the male KO liver (Fig. 6D-F).

To further demonstrate the feminization of the liver transcriptome of Tmem263-KO male mice, we compared our liver DEGs to three published data sets of mouse liver gene expression: i) WT male vs female mice (45), ii) hypophysectomized vs sham control male mice (48), and iii.) Stat5b-KO vs WT male mice (35). Overlap analysis was conducted to determine which DEGs are shared among these three groups and the Tmem263-KO male liver (Fig. 6G-I). A complete list of all Tmem263-KO male liver DEGs that are also a DEG in at least one of the three published data sets is provided (Figure 6 – figure supplement 1). When compared to WT male vs female liver data, Tmem263-KO male mice down-regulate 110 genes considered to be male-specific and up-regulate 174 genes considered to be female-specific (Fig. 6G). This indicates in the male mouse, loss of Tmem263 results in a high degree of liver transcript feminization, a phenomenon usually prevented by normal GH signaling in male liver (39-43). When compared to the DEG list of hypophysectomized male mouse liver—which lacks pituitary hormones (including GH) due to surgical removal of the pituitary gland—Tmem263-KO male liver displays 176 shared up-regulated genes and 128 shared down-regulated genes (Fig. 6H). The high overlap of liver DEGs between Tmem263-KO and hypophysectomized male mice with a deficit in GH further suggests that Tmem263 deficiency disrupts GH action in the liver. Lastly, Tmem263-KO male mouse liver DEGs were compared to those of Stat5b-KO male mice (Fig. 6I), which exhibit dwarfism and GH pulse-resistance (21). We again observed a high degree of overlap between the two KO mouse models, as Tmem263-KO male liver shares 78 up-regulated and 91 down-regulated DEGs with Stat5b-KO male liver.

Altogether, these transcriptomic data provide supporting evidence that the underlying cause of the severe growth retardation and skeletal dysplasia seen in Tmem263-KO mice is most likely attributed to disrupted GH signaling in the liver and possibly other organs such as the bone (Fig. 6G).

The Tmem263 knockout (KO) mouse strain was generated at the Johns Hopkins University School of Medicine Transgenic Core facility. To obtain a Tmem263-null allele, exon 3 that encodes ∼81% of the full-length protein (GenBank # NP_689474) was deleted with the CRISPR/Cas9 method. The deletion encompasses amino acid 23 to 115. The two guide RNAs (gRNA) used were 5’-CTATCATATGTCTCTCATTAG-3’ and 5’- GCCACTGGCATCA TATGCTG-3’. The Tmem263 KO mice were generated and maintained on a C57BL/6J genetic background. Genotyping primers for Tmem263 wild-type (WT) allele were forward (263-F2) 5’- GCAAGAGCTCCTTACTTAC TCAG-3’ and reverse (263-R1) 5’- GATAAGGGCACTTTGTTTACAACTG -3’. The size of the WT band was 391 bp. Genotyping primers for the Tmem263 KO allele were forward (263-F2) 5’- GCAAGAGCTCCTTACTTACTCA G-3’ and reverse (263-R2) 5’-TCACCAATACTTTCAACACAGCAG-3’. The size of the KO band was 218 bp. The GoTaq Green Master Mix (Promega, M7123) was used for PCR genotyping. The genotyping PCR parameters were as follows: 94°C for 5 min, followed by 10 cycles of (94°C for 10 sec, 65°C for 15 sec, 72°C for 30 sec), then 25 cycles of (94°C for 10 sec, 55°C for 15 sec, 72°C for 30 sec), and lastly 72°C for 5 min. Genotyping PCR products from WT and KO mice were excised and confirmed by DNA sequencing. All mice were generated by intercrossing Tmem263 heterozygous (+/-) mice. Tmem263 WT (+/+), heterozygous (+/-), and KO (-/-) littermates were housed in polycarbonate cages on a 12-h light–dark photocycle with ad libitum access to water and food. Mice were fed a standard chow (Envigo; 2018SX). At termination of the study, all mice were fasted for 1-2 h and euthanized. Tissues were collected, snap-frozen in liquid nitrogen, and kept at -80°C until analysis. All mouse protocols (protocol # MO22M367) were approved by the Institutional Animal Care and Use Committee of the Johns Hopkins University School of Medicine. All animal experiments were conducted in accordance with the National Institute of Health guidelines and followed the standards established by the Animal Welfare Acts.

Mammalian expression plasmid encoding human TMEM263 with a C-terminal epitope tag (Myc-DDK) was obtained from Origene (RC203933). Control pCDNA3.1 empty plasmid was obtained from Invitrogen. Rabbit polyclonal anti-human C12orf23 (TMEM263) was from Origene (TA333490). Mouse monoclonal JAK2 (C-10), STAT5 (A-9), growth hormone receptor (GHR)(B-12) were obtained from Santa Cruz (sc-390539, and sc-74442, and sc-137184, respectively). Rabbit monoclonal antibody against phospho-JAK2 (Tyr1008)(D4A8) were obtained from Cell Signaling Technology (CST # 8082). Rabbit polyclonal antibody against phospho-STAT5 (Y694) was obtained from Cell Signaling Technology (CST#9351). Mouse monoclonal anti-FLAG M2 antibody (F1804) and mouse monoclonal anti-β-actin antibody (A1978) were obtained from Sigma. HRP-conjugated goat anti-rabbit (CST#7074) and goal anti-mouse (CST#7076) antibodies were from Cell Signaling Technology.

Cell surface proteins from transfected HEK293 cells were biotinylated using the Pierce Cell Surface Protein Isolation Kit (ThermoFisher Scientific., cat# 89881). Briefly, transfected cells were washed with ice-cold PBS, then incubated with the cell-impermeable biotinylation reagent (Sulfo-NHS-SS-Biotin) for 30 min at 4°C to prevent endocytosis. A quenching solution was then used to stop biotinylation. Cells were collected in a clean tube and centrifuged at 500 g for 3 min. The cell pellet was then lysed with RIPA buffer (10 mM Tris-HCl pH 8.0, 1 mM EDTA, 1% Triton X-100, 0.1% Sodium Deoxycholate, 0.1% Sodium Dodecyl Sulfate, 140 mM NaCl, 1 mM PMSF) supplemented with protease inhibitor cocktail (Roche, 11836153001) for 30 min on ice. Cell lysate was then centrifuged at 10,000 g for 2 min at 4°C and the resultant supernatant was collected. Biotinylated proteins were pulled-down with Pierce NeutrAvidin Agarose (ThermoFisher Scientific; 29200) at room temperature for 1 h mixed end-over-end. Bound proteins were eluted from the agarose by DTT (50 mM, final) and heat (94°C for 3 min). Fractions of the input cell lysate, flow-through/non-bound solution, and eluate were used for immunoblot analysis.

For Alcian blue/Alizarin red stained whole mounts, skeletal preps were preformed using P0 neonates following established protocols (77). Briefly, carcasses were macerated by removing skin, eyes, internal organs, and adipose tissue before dehydrating in 95% ethanol overnight and then fixing in acetone. Cartilage was stained by submerging in Alcian blue solution overnight at room temperature. Mineralized tissue was stained by submerging in Alizarin red solution at 4°C overnight.

For microcomputed tomographic analyses of bone structure, high resolution images of the mouse femur were acquired using a desktop microtomographic imaging system (Skyscan 1172, Bruker, Belgium) in accordance with the recommendation of the American Society for Bone and Mineral Research (78). Samples were scanned with an isotropic voxel size of 10μm at 65KeV and 153μA using a 1.0mm aluminum filter. The resulting images were reconstructed using NRecon (Bruker). Trabecular bone parameters in the distal femur of control mice were assessed in a region of interest 500μm proximal to the growth plate and extending for 2mm, while femoral cortical bone parameters were assessed in a 500μm region of interest centered on the mid-diaphysis. Due to the reduced size of the Tmem263-KO mice, regions of interest (ROI) were adjusted to be proportional for tissue length. Trabecular bone parameters were assessed in an ROI 300mm proximal to the growth plate and extending for 1.2mm; cortical bone parameters were assessed in a 300mm ROI centered on the mid-diaphysis.

Growth plate morphology was examined in the proximal tibia. Tibia were dehydrated in graded alcohol solutions and then embedded in methyl methacrylate. 7μm thin sections were cut with a Leica HistoCore Automated Rotary microtome (Leica, Wetzlar, Germany) and stained with Goldner’s modified trichrome. Total growth plate length, as well as the lengths of the hypertrophic and proliferative zones were measured with Bioquant Image Analysis software (Bioquant, Nashville, TN).

Liver protein was isolated using RIPA buffer as previously described (79). Liver protein lysates used for immunoblots were boiled for 5 min in a loading buffer (50 mM Tris, 2% SDS, 1% β-ME, 6% glycerol, 0.01% bromophenol blue). Total protein was quantified by BCA assay (Thermo Scientific, 23225), loaded in equal amounts and volume, and run on a 7.5% or 10% TGX gel (Bio-Rad, 4561023 and 4561033). Protein was transferred to PVDF membrane (Bio-Rad, 1620177) using the Trans Blot Turbo system (Bio-Rad) and blocked in PBS containing 0.2% Tween 20 and 5% non-fat milk for 1 h, then incubated overnight at 4°C on a shaker with primary antibody. The following primary antibodies were used: anti-TMEM263 (1:1000), anti-GH-R (1:100), anti-phospho-JAK2 (Tyr1008) (1:500), anti-phospho-STAT5 (Y694) (1:500), anti-JAK2 (1:100), anti-JAK5 (1:100), anti-β-actin (1:2000). Blots were washed at least 3 times (5 min each) with PBS containing 0.2% Tween 20 before secondary antibody was added. HRP-conjugated anti-rabbit or anti-mouse secondary antibody (1:500 to 1:1000; 1 hr at room temperature) were used to recognize the primary antibody. Blots were washed at least 3 times (5 min each) with PBS containing 0.2% Tween 20. Immunoblots were developed using HRP substrate ECL (GE Healthcare), visualized with a MultiImage III FluorChem Q (Alpha Innotech), and quantified with ImageJ (80).

Recombinant human GH from Peprotech (Cat # 100-40; Cranbury, NJ) was reconstituted in PBS at a concentration 0.2 μg/μL. Each mouse received GH injection at a dose of 3 μg/g body weight. Mice were sacrificed 25 min post GH injection, and the liver was immediately harvested and snap frozen in liquid nitrogen.

Tail vein blood samples were allowed to clot on ice and then centrifuged for 10 min at 10,000 x g. Serum samples were stored at -80°C until analyzed. Serum insulin (Crystal Chem, Elk Grove Village, IL; 90080), GH (Millipore Sigma; EZRMGH-45K, Burlington, MA), IGF-1 (Crystal Chem, 80574), IGFBP3 (Abcam; ab100692, Waltham, MA), and IGFALS (Cusabio; CSB-EL011094MO, Houston, TX) levels were measured by ELISA according to manufacturer’s instructions. Serum calcium and phosphate levels were quantified at the Molecular and Comparative Pathobiology Core at The Johns Hopkins University School of Medicine using the Respons®910VET Veterinary Chemistry Analyzer (DiaSys Diagnostic Systems, Wixom, MI, USA).

Total RNA was isolated from tissues using Trizol reagent (Thermo Fisher Scientific) according to the manufacturer’s instructions. Purified RNA was reverse transcribed using an iScript cDNA Synthesis Kit (Bio-rad). Real-time quantitative PCR analysis was performed on a CFX Connect Real-Time System (Bio-rad) using iTaq^TM Universal SYBR Green Supermix (Bio-rad) per manufacturer’s instructions. Data were normalized to either β-actin or 36B4 gene (encoding the acidic ribosomal phosphoprotein P0) and expressed as relative mRNA levels using the ΔΔCt method (81). Fold change data were log transformed to ensure normal distribution and statistics were performed. Real-time qPCR primers used were: mouse Tmem263 forward (qPCR-263-F1), 5’- CGCGGTGATCATGAATCAGG CAG-3’ and reverse (qPCR-263-R1), 5’-GCTCCCTTTGTAACACTGAAGA- 3’; growth hormone receptor (Ghr) forward, 5’-ACAGTGCCTACTTTTGTGAGTC-3’ and reverse, 5’- GTAGTGGTAAGGCTTTCTGTGG-3’; Igf-1 forward, 5’-GTGAGCCAAAGACACACCCA-3’ and reverse, 5’- ACCTCTGATTTTCCGAGTTGC-3’; Igfbp3 forward, 5’-CCAGGAAACATCAGTGAG TCC-3’ and reverse, 5’- GGATGGAACTTGGAATCGGTCA-3’; Igfals forward, 5’-CTGCCCGATAGCATCCCAG-3’ and reverse, 5’- GAAGCCAGACTTGGTGTGTGT-3’; 36B4 forward, 5’-AGATTCGGGATATGCTGTTGGC-3’ and reverse, 5’- TCGGGTCCTAGACCAGTGTTC-3’; β-actin forward, 5’-GGCACCACACCTTCTACAATG -3’ and reverse, 5’- GGGGTGTTGAAGGTCTCAAAC -3’.

Bulk RNA sequencing of WT (n = 8) and Tmem263-KO (n = 8) mouse liver were performed by Novogene (Sacramento, California, USA) on an Illumina platform (NovaSeq 6000) and pair-end reads were generated. Sequencing data was analyzed using the Novogene Analysis Pipeline. In brief, data analysis was performed using a combination of programs, including Fastp, Hisat2, and FeatureCounts. Differential expressions were determined through DESeq2. The resulting P-values were adjusted using the Benjamini and Hochberg’s approach for controlling the false discovery rate. Genes with an adjusted P-value ≤ 0.05 found by DESeq2 were assigned as differentially expressed. Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and Reactome (http://www.reactome.org) enrichment were implemented by ClusterProfiler. All volcano plots and heat maps were generated in Graphpad Prism 9 software. All statistics were performed on Log transformed data. All heat maps were generated from column z-score transformed data. The z-score of each column was determined by taking the column average, subtracting each sample’s individual expression value by said average then dividing that difference by the column standard deviation. Z-score = (value – column average)/column standard deviation. High-throughput sequencing data from this study have been submitted to the NCBI Sequence Read Archive (SRA) under accession number # PRJNA938158.

All results are expressed as mean ± standard error of the mean (SEM). Statistical analysis was performed with Graphpad Prism 9 software (GraphPad Software, San Diego, CA). Data were analyzed with two-tailed Student’s t-tests or by one-way ANOVA with Tukey’s multiple comparisons test. P < 0.05 was considered statistically significant.