REVIEW
SHOX and sex difference in height: a hypothesis
2025 Volume 72 Issue 1 Pages 37-42
Details
2025 Volume 72 Issue 1 Pages 37-42
The mean height is taller in males than in females, except for early teens. In this regard, previous studies have revealed that (1) distribution of the mean adult heights in subjects with disorders accompanied by discordance between sex chromosome complement and bioactive sex steroids and in control subjects (the British height standards) indicates that, of the ~12.5 cm of sex difference in the mean adult height, ~9 cm is accounted for by the difference in the sex chromosome complement and the remaining ~3.5 cm is explained by the dimorphism in sex steroids (primarily due to the growth-promoting effect of gonadal androgens); (2) according to the infancy-childhood-puberty growth model, the sex difference in the childhood growth function produces height differences of ~1 cm in childhood and 8–10 cm at 18–20 years of age, whereas the sex difference in the pubertal growth function yields height difference of ~4.5 cm at 18–20 years of age; and (3) SHOX expression and methylation analyses using knee cartilage tissues and cultured chondrocytes have shown lower SHOX expression levels in female samples than in male samples and methylation patterns consistent with partial spreading of X-inactivation affecting SHOX in female samples. These findings suggest that small but persistent sex difference in SHOX expression dosage leads to the variation in the sex steroid independent childhood growth function, thereby yielding the sex difference in height which remains small in childhood but becomes obvious in adulthood.
The mean height is taller in males than in females with the height differences of ~1.0 cm in childhood and 12.5–13.0 cm in adulthood, except for early teens when girls become taller than boys because of the earlier onset of pubertal growth spurt [1, 2]. In this regard, previous studies have suggested that the sex difference in the adult height is explained by the presence or absence of GCY (growth-control gene on the Y chromosome) which promotes statural growth independently of the sex steroids and by the dimorphism in sex steroids which influence pubertal growth pattern [3–5]. Furthermore, in support of the presence of GCY, the distribution of the mean adult heights in subjects with sex chromosome aberrations is explained by the dosage effect of SHOX (short stature homeobox containing gene), the dosage effect of GCY, non-specific growth disadvantage caused by alteration of euchromatic or non-inactivated regions, and the dimorphism in sex steroids [4]. However, despite multiple attempts to identify GCY [6–8], GCY remains quite elusive even in the era of the whole genome sequencing.
SHOX is a DNA binding transcription factor gene isolated from the short arm pseudoautosomal region (PAR1) of the human sex chromosomes [9]. SHOX is involved in skeletal growth and development, and heterozygous SHOX deficiency causes diverse clinical features ranging from idiopathic short stature phenotype to Léri-Weill dyschondrosteosis (LWD) [10, 11]. Since the PAR1 harboring SHOX is shared by the X and the Y chromosomes, it is assumed that SHOX escapes X-inactivation and is present in two active copies in both males and females [9–11]. This implies that SHOX plays no major role in the phenotypic difference between sexes. Indeed, while LWD is predominantly manifested by pubertal and adult females, it has been postulated that gonadal estrogens exert a maturational effect on skeletal tissues that are susceptible to unbalanced premature fusion of growth plates because of SHOX deficiency, facilitating the development of LWD in a female-dominant and pubertal tempo-dependent fashion [10, 11].
Here, we review informative data for the deduction of underlying factor(s) leading to the sex difference in height, and propose that sex difference in the SHOX expression dosage constitutes the sex steroid independent determinant for the sex difference in height.
We have previously determined the mean adult heights in complete XY gonadal dysgenesis (CXYGD), complete XX gonadal dysgenesis (CXXGD), SRY (+) XX maleness (SRY(+)XXM), and complete androgen insensitivity syndrome (CAIS) which are associated with discordance between sex chromosome complement and bioactive sex steroids [3, 4]. All the patients satisfied the following selection criteria: (1) normal karyotype with no demonstrable mosaicism; (2) height recorded between 20 and 50 years of age, or confirmation of growth cessation; (3) apparent Caucasians of various nationalities; (4) no selection for height in the ascertainment of patients; (5) lack of other disorders that may affect statural growth; and (6) no intervention that may influence stature, such as sex steroid supplementation therapy in CXXGD and CXYGD and gonadectomy in CAIS. For controls, we utilized the British height standards which occupy a roughly median height position among Caucasian populations, with 12.5 cm of sex difference in the mean adult height [2, 12].
The mean adult heights are summarized in Table 1, and the distribution of the mean adult heights is shown in Fig. 1. Height comparisons between control males and SRY(+)XXM patients, between CAIS patients and control females, and between CXYGD patients and CXXGD patients, in whom the sex chromosome complement is different but the effect of bioactive sex steroids is similar, imply that the difference in sex chromosome complement yields the adult height difference of ~9 cm independently of sex steroids [3, 4]. Furthermore, height comparisons between control males and CXYGD patients, between SRY(+)XXM patients and CXXGD patients, between CAIS patients and CXYGD patients, between control females and CXXGD patients, between control males and CAIS patients, and between SRY(+)XXM patients and control females, in whom the sex chromosome complement is identical but the effect of bioactive sex steroids is different, argue that the dimorphism in sex steroids produces the adult height difference of ~3.5 cm, primarily due to the growth-promoting effect of gonadal androgens (gonadal estrogens are unlikely to have a major effect on the adult height) [3, 4].
| Karyotype | Bioactive sex steroids | Adult height (cm)* | |
|---|---|---|---|
| Complete XY gonadal dysgenesis (CXYGD) | 46,XY | None | 172.0 ± 7.0 (n = 24) |
| Complete XX gonadal dysgenesis (CXXGD) | 46,XX | None | 164.3 ± 7.7 (n = 22) |
| Complete androgen insensitivity syndrome (CAIS) | 46,XY | Estrogens | 172.2 ± 6.5 (n = 23) |
| SRY (+) XX maleness (SRY(+)XXM) | 46,XX | Androgens | 166.2 ± 5.8 (n = 14) |
| Control males | 46,XY | Androgens | 174.7 ± 6.7 |
| Control females | 46,XX | Estrogens | 162.2 ± 6.0 |
The above findings imply that the sex difference in the mean adult height is mainly caused by the sex steroid independent factor, i.e., the larger growth potential in the XY complement than in the XX complement, and partly caused by the sex steroid dependent factor, i.e., the growth promoting effect of gonadal androgens but not gonadal estrogens.
The ICP growth model is an excellent biologically-oriented growth standard produced by the mathematical analysis of auxological data [13, 14]. The main concept is that human growth is divided into three additive and partly superimposed components: (1) the “infancy” component which is expressed by an exponential function; (2) the “childhood” component which is expressed by a second-degree polynomial function; and (3) the “puberty” component which is expressed by a logistic function. Although the “infancy” component tails off by 3–4 years of age, the “childhood” component continues until the final height age. Thus, the height gain from childhood through adulthood is the sum of the height increment caused by the extension of the sex steroid independent childhood growth function and that added by the sex steroid dependent pubertal growth function [5].
According to the Swedish longitudinal growth study, the sex difference in the childhood growth function produces taller height in boys than in girls, with height differences of ~1 cm in childhood (around 6 years of age) and 8–10 cm at 18–20 years of age, whereas the sex difference in the pubertal growth function yields taller height in boys than in girls, with the height difference of ~4.5 cm at 18–20 years of age [5, 13]. Furthermore, in the absence of the puberty component, the childhood growth function alone can attain almost normal adult heights [15]. Thus, the actual growth patterns and the childhood component only growth patterns would be illustrated as shown in Fig. 2, although the childhood component growth patterns would virtually be decelerated with pubertal development and resultant skeletal maturation.
(A) Sex steroid dependent adult height difference (growth-promoting effect of gonadal androgens).
(B) Sex steroid independent adult height difference.
The above findings argue that the sex difference in the mean adult height is mainly yielded by the variation in the sex steroid independent childhood growth function and partly produced by that in the sex steroid dependent puberty growth function. In conjunction with the preceding arguments, it is inferred that the variation in the childhood growth function is closely related to the difference in sex chromosome complement and that in the puberty growth function is primarily due to the dimorphism in sex steroids.
Recent studies have demonstrated that transcript levels of X-linked genes escaping X-inactivation, including those on the PAR1, are lower on the inactive X chromosome than on the active X chromosome [16]. Indeed, transcript levels of PAR1 genes on the inactive X chromosome remain ~80% of those on the active X chromosome in various tissues [16]. Thus, average transcript levels of PAR1 genes are lower in females than in males. This phenomenon is explained by partial spreading of the X-inactivation into the PAR1 on the inactive X chromosome and resultant epigenetic modifications for PAR1 genes. In general, genes subject to X-inactivation are associated with relative hypermethylation at the promoter/enhancer regions and relative hypomethylation at the gene bodies and intergenic regions on the inactive X chromosome, whereas genes escaping X-inactivation show opposite methylation patterns [17]. It is predicted, therefore, that SHOX also undergoes partial X-inactivation on the inactive X chromosome with methylation patterns reminiscent of those of genes subject to X-inactivation. However, since SHOX is primarily expressed in the skeletal tissue of the distal limb region and the faciocervical region where Turner skeletal features are observed [18], it remained quite difficult to examine SHOX expression and methylation pattern in such tissues.
Recently, we had an opportunity to examine SHOX expression dosage and methylation pattern around SHOX, using knee cartilage tissues and cultured chondrocytes established from the phalangeal bones, in collaboration with Orthopedic Surgery group [19]. The data are summarized as follows (Fig. 3): (1) microarray-based transcriptome analysis for knee cartilage tissues obtained from four adults (two women and two men) and cultured chondrocytes established from phalangeal bones of 12 children (six girls and six boys) showed lower SHOX expression levels in the female samples than in the male samples, though not reaching statistical significance; (2) quantitative reverse-transcription PCR analysis for knee cartilage tissues obtained from 22 adolescent/adults (11 women and 11 men) and phalangeal cartilage tissues obtained from 14 children (five girls and nine boys) revealed lower SHOX expression levels in the female samples than in the male samples, with the difference in knee cartilage samples being statistically significant; (3) reduced representation bisulfite sequencing around SHOX in four knee cartilage tissues indicated male-dominant methylation pattern in a ~3.2 kb region at the position 3.9 kb upstream of SHOX and in a ~1.9 kb region at intron 5 ~ exon 6a, and female-dominant methylation pattern in a ~1.2 kb region at intron 2; (4) pyrosequencing for the above regions in 22 knee cartilage tissues reproduced the male-dominant methylation pattern in the upstream region and exon 6a; (5) SHOX expression dosage is positively correlated with the methylation indexes of the upstream region and exon 6a and negatively correlated with the methylation index of intron 2; and (6) the ~1.2 kb region at intron 2 is associated with CpG islands, candidate cis-regulatory elements, H3K4Me1 histone modification, and DNase hypersensitivity sites characteristic of an enhancer [19].
A. SHOX expression dosage. SHOX expression is presumably lower in the inactive X chromosome than in the active X and Y chromosome. This would cause the sex difference in the SHOX expression dosage, leading to the height difference between sexes.
B. Methylation patterns around SHOXa (the major functional SHOX isoform consisting of exons 1–6a). The difference in methylation indexes (MIs) between female (F) and male (M) samples indicates a female-dominant methylation pattern at intron 2 (an enhancer region) and male-dominant methylation patterns at the upstream region (an intergenic region) and exon 6a (a gene body).
C. Enhancer property of a ~1.2 kb region at intron 2. Negative correlation between relative SHOX expression dosage against TBP (TATA binding protein) and methylation index as well as lower SHOX expression dosage in females with relatively high MIs (orange circles) than in males with relatively low MIs (blue circles), and the presence of CpG island, candidate cis-regulatory elements (cCREs), H3K4Me1, and DNase hypersensitivity sites, imply that intron 2 has an enhancer property.
These results imply that the SHOX expression levels are somewhat lower in females than in males, in conjunction with methylation patterns similar to those of genes subject to X-inactivation. Indeed, the upstream region, exon 6a, and intron 2 have the property of an intergenic region, a gene body, and an enhancer, respectively [17]. In particular, the higher methylation pattern in female cells than in male cells at the ~1.2 kb region in intron 2 with an enhancer character is compatible with the lower SHOX expression dosage in female cells than in male cells, because of attenuated SHOX expression from the inactive X chromosome. It is inferred, therefore, that partial spreading of the X-inactivation into the PAR1 has caused epigenetic modification for SHOX, thereby leading to attenuated SHOX expression in females.
The above findings suggest that the small but persistent sex difference in SHOX expression dosage, i.e., the lower SHOX expression in females than in males, leads to the variation in the sex steroid independent childhood growth function, thereby yielding the sex difference in height which remains small in childhood but becomes obvious in adulthood. This notion explains why GCY has not been identified so far even after sequencing of the whole genome.
In this regard, several matters are worth pointing out. First, the lower SHOX expression dosage in females than in males would explain why the ratio of sitting height to height is larger in girls than in boys from childhood and becomes obviously different between sexes with puberty [20], because SHOX is primarily expressed in the distal limb regions (forearms and lower legs) but not in the vertebral bones [18], and growth plate fusion in the SHOX expression positive distal limb regions is facilitated by the skeletal maturing effects of gonadal estrogens in a female-dominant manner [10, 11]. Indeed, patients with SHOX haploinsufficiency exhibit markedly increased sitting height to height ratios from childhood [21], and a longitudinal auxological study in a girl with SHOX haploinsufficiency has revealed that the SDS for sitting height remains constant from childhood through puberty, whereas the SDSs for height, arm span, and lower leg length are drastically decreased with puberty [22]. Second, the sex difference in the SHOX expression dosage would also explain why the birth length is smaller in females than in males, because SHOX is expressed from the fetal life [18]. Third, the sex difference in the SHOX expression dosage would be involved in the female-dominant development of LWD, although the major factor for the female dominant and pubertal tempo dependent occurrence of LWD would be the skeletal maturing effects of gonadal estrogens [10, 11].
It should be pointed out, however, that multiple X-linked genes may show variable degrees of different expression dosages between sexes [16]. This may also be involved in the development of sex difference in height. Thus, further studies are required to clarify whether the sex difference in SHOX expression dosage constitutes the sole or major sex steroid independent determinant for the sex difference in height.
None of the authors have any potential conflicts of interest associated with this research. M.F. is a member of Endocrine Journal’s Editorial Board.
This work was supported by the grants from Japan Society for the Promotion of Science 22K15932, National Center for Child Health and Development, Canon Foundation, Japan Endocrine Society, and Takeda Science Foundation.
