Mechanical-Stress-Related Epigenetic Regulation of ZIC1 Transcription Factor in the Etiology of Postmenopausal Osteoporosis

Abstract

Mechanical loading exerts a profound influence on bone density and architecture, but the exact mechanism is unknown. Our study shows that expression of the neurological transcriptional factor zinc finger of the cerebellum 1 (ZIC1) is markedly increased in trabecular bone biopsies in the lumbar spine compared with the iliac crest, skeletal sites of high and low mechanical stress, respectively. Human trabecular bone transcriptome analyses revealed a strong association between ZIC1 mRNA levels and gene transcripts characteristically associated with osteoblasts, osteocytes and osteoclasts. This supposition is supported by higher ZIC1 expression in iliac bone biopsies from postmenopausal women with osteoporosis compared with age-matched control subjects, as well as strongly significant inverse correlation between ZIC1 mRNA levels and BMI-adjusted bone mineral density (BMD) (Z-score). ZIC1 promoter methylation was decreased in mechanically loaded vertebral bone compared to unloaded normal iliac bone, and its mRNA levels correlated inversely with ZIC1 promoter methylation, thus linking mechanical stress to epigenetic control of gene expression. The findings were corroborated in cultures of rat osteoblast progenitors and osteoblast-like cells. This study demonstrates for the first time how skeletal epigenetic changes that are affected by mechanical forces give rise to marked alteration in bone cell transcriptional activity and translate to human bone pathophysiology.

Keywords: ZIC1 transcription factor; bone; epigenetic; mechanical stress; osteoporosis.

Figure 1

Correlation observed between ZIC1 promotor methylation and ZIC1 expression. (A) Correlation between ZIC1 promotor methylation in CpG1 and ZIC1 expression in the male lumbar spine trabecular bone (Pearson r = −0.79, p = 2.4 × 10⁻⁶) (Table 1). (B) ZIC1 expression in the iliac bone of postmenopausal women showed a significant negative correlation with BMD. An inverse association between ZIC1 expression and total femoral BMD Z-score that had been adjusted for variation in body mass index (BMI) (r = −0.383; p = 0.0033) was observed. (C) A difference was observed in the methylation levels of ZIC1 promoter at CpG1 from iliac (IB) and lumbar regions (L3–L5) (expressed as %), skeletal sites of low and high mechanical stress, respectively.

Figure 2

Affymetrix ZIC1 signal values were correlated with signal values for the transcripts reflecting osteoblast matrix-forming activity. Transcripts for (A) COL1A1, IBSP*, SPARC*, BGLAP*, CDH11* and osteoblast differentiation (BMP2* and RUNX2*) showed strong positive correlation with ZIC1 in the lumbar spine but had weaker or no correlation in IB. The PTHR1* transcript used to indicate osteoblast number also showed a strong positive correlation with ZIC1 expression. (B) ACP5 and CTSK** transcripts that reflect osteoclast activity showed strong positive correlation with ZIC1 expression. Osteoclast-associated transcripts CALCR** and OSCAR** showed inverse correlation with ZIC1 expression in the lumbar spine but weaker and no correlation, respectively, in the iliacus, probably due to fewer samples. The expression of osteocyte-associated transcripts (C) SOST, PDPN*** and MEPE*** at both high (LS) and low (IB) bone turnover sites showed strong correlation with ZIC1 mRNA. (Please see Supplementary Figure S1. A*, B** and C*** for correlations).

Figure 2

Affymetrix ZIC1 signal values were correlated with signal values for the transcripts reflecting osteoblast matrix-forming activity. Transcripts for (A) COL1A1, IBSP*, SPARC*, BGLAP*, CDH11* and osteoblast differentiation (BMP2* and RUNX2*) showed strong positive correlation with ZIC1 in the lumbar spine but had weaker or no correlation in IB. The PTHR1* transcript used to indicate osteoblast number also showed a strong positive correlation with ZIC1 expression. (B) ACP5 and CTSK** transcripts that reflect osteoclast activity showed strong positive correlation with ZIC1 expression. Osteoclast-associated transcripts CALCR** and OSCAR** showed inverse correlation with ZIC1 expression in the lumbar spine but weaker and no correlation, respectively, in the iliacus, probably due to fewer samples. The expression of osteocyte-associated transcripts (C) SOST, PDPN*** and MEPE*** at both high (LS) and low (IB) bone turnover sites showed strong correlation with ZIC1 mRNA. (Please see Supplementary Figure S1. A*, B** and C*** for correlations).

Figure 3

The effect of fluid shear stress (FSS) on osteogenic genes. (A) Rat osteoprogenitor cells were cultured in osteogenic medium and subjected to FSS (F). Control cells are labelled S (stationary). The cells subjected to FSS for three days (lane 3F) had induction of the key osteogenic marker genes Runx2, ALP and Col1a1 when compared to controls (lane C). (B) The induction was confirmed by qRT-PCR in the FSS and control stationary cells. (C) Cells cultured in normal non-osteogenic medium, subjected to similar magnitude of FSS, showed no evidence of Zic1 or Col1a1 induction over three days (lane 3F vs. 3S). However, after prolonged shear stress, there was an induction of Col1a1, as well as ZIC1, in these cells (lanes 5F, 5S, 7F and 7S).

Figure 4

Rat osteoprogenitors (ROPs) cultured in osteogenic medium and maintained as static controls were fixed on 25 mm coverslips and immunostained at the end of day 3 with (A) anti-RUNX2 (green fluorescence, plate 1) and (B) anti-ZIC1 (red fluorescence, plate 2) antibodies. (C) Plates 1 and 2 were merged together to identify co-localisation of RUNX2 and ZIC1 antibodies (Plate 3). In the representative field of vision, many nuclei showed nuclear translocation of RUNX2, suggesting relevant osteogenic activity in these cells. A few nuclei showed co-localisation of RUNX2 and ZIC1, and it appears that expression of RUNX2 (yellow) was accompanied by some ZIC1 (red/green). The effect of FSS on the intracellular distribution of RUNX2 and ZIC1 in ROPs is seen in (D) plate 4 and (E) plate 5, respectively. (F) Plates 4 and 5 were merged together to identify co-localisation of RUNX2 and ZIC1 antibodies (Plate 6).

Figure 5

(A,B) The fluorescence intensity of RUNX2 and ZIC1 nuclei (expressed as mean, grey values) from ROPs cultured in osteogenic medium, maintained either as static control or subjected to FSS at the end of day 3. (C) The plot shows a direct relationship between translocation of RUNX2 and ZIC1 to the nucleus when subjected to FSS.

Figure 5

(A,B) The fluorescence intensity of RUNX2 and ZIC1 nuclei (expressed as mean, grey values) from ROPs cultured in osteogenic medium, maintained either as static control or subjected to FSS at the end of day 3. (C) The plot shows a direct relationship between translocation of RUNX2 and ZIC1 to the nucleus when subjected to FSS.