Elevated levels of neutrophils with a pro-inflammatory profile in Turner syndrome across karyotypes

Introduction

Turner syndrome (TS) is a sex chromosome disorder in individuals who have a karyotype containing one X chromosome and complete or partial absence of the second sex chromosome, in association with one or more typical clinical manifestations, such as short stature, hypergonadotropic hypogonadism, early sensorineural hearing loss, and distinctive cardiac, renal, skeletal and autoimmune disorders¹. Complex epigenomic and genomic changes have been demonstrated in the proportion of individuals with TS with the 45,X karyotype. These changes affect both the coding and the non-coding transcriptome, as well as the methylome in a tissue specific manner, but with distinct similarities across tissues^2,3,4,5,6. Only about 35–45% of individuals with TS have the 45,X karyotype, while the remaining individuals have other karyotypes—i.e., 45,X/46,XX, 46,X,i(Xq), 46,X,i(Xp), 45,X/46,XY, karyotypes with a ring X, and others^7,8.

Primarily, 45,X cases have been studied in newer genomic studies^2,3,4,5,6, hence, it is unknown if other TS karyotypes affect the genomic landscape⁹. The karyotype-phenotype relation in TS is poorly understood^9,10, although newer, larger clinical studies clearly show that a relation may exist^7,11,12, with the 45,X karyotype having an increased comorbidity and mortality compared to other karyotypes^13,14. Especially TS individuals with the 45,X/46,XX mosaic karyotype seem to have a milder phenotype¹⁵, but there also appear to be phenotypic differences among other karyotypes^7,11,12. However, all individuals with TS currently undergo the same comorbidity screening protocols and treatment algorithms, which may induce overtreatment in a proportion of patients, while other groups of patients may receive sub-optimal treatment¹.

DNA-methylation (DNAm) and transcriptome (RNAseq) studies have revealed a distinct genome-wide DNAm and gene expression profile in TS with 45,X in blood^16,17, muscle and fat², emphasizing the importance of the X-chromosome (X) in controlling genome-wide epigenetic gene regulation^2,3,5,18. This is consistent with the discovery of the X-chromosome harboring candidate genes of epigenetic regulation^19,20. However, it is not yet clear which genes on the X chromosome are involved in the TS phenotype, but interest currently centers on dosage-sensitive genes, such as escape genes and genes of the pseudoautosomal regions (PAR), such as KDM6A²¹, KDM5C²², and SLC25A6²³. These genes have been proposed to influence the risk of TS comorbidity^2,16,22.

No data are available on the effects on the methylome and transcriptome of small losses of X material or of TS mosaicism, with previous studies being too small to draw conclusions¹⁷. Exploration into the DNAm and gene expression profile of TS individuals with other karyotypic representations, and the investigation of possible shared differences in the methylome of TS with other karyotypes and TS with 45,X, compared with female controls, could hint at markers of TS comorbidity when combined with clinical data.

The present study explores to which extent the gene expression and DNA-methylation profile in TS individuals with other karyotypical representations is perturbed in comparison with TS with 45,X and karyotypically normal females. We link the increased prevalence of autoimmune disorders in TS, which points to a fundamental imbalance in the immune system possibly encompassing low grade inflammation^24,25, increased risk of cardiovascular complications¹⁰ and abnormal body composition with increased fat mass and low muscle mass²⁶ with the differential methylome and transcriptome. Females (46,XX), males (46,XY) and males with Klinefelter syndrome (KS; 47,XXY) served as controls. These aspects of TS pathology may be critical, as these conditions not only contribute to the clinical burden, but also offer insights into underlying mechanisms at play. Considering this, we examine potential aberrations in blood cell profiles, which might shed light on how these cellular changes correlate with the broader immune dysregulation and genomic alterations.

Results

To explore the genomic and underlying pathological mechanisms of TS individuals across karyotypes, we utilized a “genomic cohort”, which included individuals with TS 45,X monosomy, TS other karyotypes and 46,XX controls (Table 1, Fig. 1). To validate our findings, we incorporated additional cohorts: “The extended cohort”, which included the same TS 45,X individuals and 46,XX controls as the genomic cohort, in addition to 46,XY control males and 47,XXY Klinefelter males; the flow cytometry cohort; the outpatient cohort and the epidemiological cohort. Each cohort was used for specific analyses to provide a comprehensive understanding of TS (Fig. 1). For inclusion criteria and further details (Cohorts, Material and Methods)

Table 1 Comparison of clinical variables between subgroups of Turner syndrome and controls in the genomic cohort

Full size table

Fig. 1: Overview of study cohorts and karyotype distributions.

A The table summarizes the composition of each cohort included in the study, highlighting the genomic cohort, extended cohort, flow cytometry cohort, outpatient cohort and epidemiological cohort along with the specific analyses performed on them such as RNA sequencing (RNAseq), DNA methylation (DNAm), clinical traits, and autoimmune disease history based on ICD10 classification. The color coding indicates the different karyotypes, including Turner syndrome (TS) variants and controls, as used throughout the study. Furthermore, if individuals were present in several cohorts, these were termed cross-over individuals (arrows). B Representative chromosome ideograms for selected karyotypes show the structural differences associated with TS, demonstrating the chromosomal diversity within the study group.

Full size image

DNA methylation and gene expression across karyotypes (genomic cohort)

In the “genomic cohort”, we performed DNA-methylation and transcriptome profiling on whole-blood samples to assess the genome-wide impact of karyotypes other than 45,X in TS (Fig. 1A, B). The present cohort constituted a subset of a larger cohort, which previously participated in a study investigating the cardiovascular phenotype in TS^27,28. As such, the TS other karyotypes group represented a karyotypically heterogeneous manifestation of TS.

Differences in clinical traits between these groups have previously been presented^29,30,31. Here, we made a new comparison specific for the subset included in this study (Table 1, Supplementary Table 1). No differences were observed between TS 45,X and TS other karyotypes with respect to the clinical traits. Thus, the two groups TS 45,X and TS other karyotypes, despite of the different karyotypes, seemed to share a near identical phenotype.

The transcriptomic profile

In the “genomic cohort” the transcriptomic profile of TS 45,X and TS other karyotypes was evaluated by RNA sequencing of whole blood. Overall, the expression of autosomal genes was similar in TS 45,X and TS other karyotypes, with no autosomal differentially expressed genes (DEGs). However, as expected, the expression of X chromosomal genes differed. A principal component analysis based on X chromosome gene expression showed heterogeneity in TS other karyotypes and illustrates how the TS other karyotypes represent a continuum of X chromosomal gene expression from individuals with TS 45,X to 46,XX (Fig. 2A). This clustering pattern was mainly driven by differential expression of XIST (Fig. 2B), and as such, low degree mosaicism and structural variants missing the X chromosome p-arm clustered closest to 46,XX.

TS 45,X and TS other karyotypes shared almost the exact same expression pattern of PAR1 genes when compared to 46,XX. That is, the expression of PAR1 genes was approximately half of that observed in 46,XX (Fig. 2C). No differences were observed for the expressed PAR2 genes. We further subdivided the X chromosomal genes into escape, inactive, variable, and unknown expression pattern based on the characterization by Tukiainen et al.³². Again, TS 45,X and TS other karyotypes shared an almost identical expression pattern of all X chromosomal genes regardless of the expression status assigned (Supplementary Fig. 1).

As described above, the largest difference between TS 45,X and TS other karyotypes was observed in the expression of XIST. The expression within the TS other karyotypes group indicated the presence of an XIST-low (n = 15) and XIST-high group (n = 9) (gene expression above 10,000 counts) (Fig. 2B). The XIST expression in the XIST-high group was comparable to that of 46,XX, whereas the XIST-low group was closer to the TS group. To investigate possible genome-wide effects of XIST expression, a differential expression analysis was conducted. Three X-chromosomal DEGs were found (XIST, TSIX, JPX), all within the X inactivation center, as expected. However, no autosomal DEGs were found, indicating that XIST expression do not have genome-wide effects on the expression of autosomal genes in TS.

The DNA methylation profile

Next, we investigated the methylation profile. Multidimensional scaling of CpG site methylation revealed a TS specific methylation profile on both autosomes and the X chromosome (Fig. 2D, E). Based on autosomal CpGs, most of the individuals in the TS other karyotypes group clustered with TS 45,X group, however, some of the TS other karyotypes represented an intermediate methylation pattern somewhat in between that of TS 45,X and 46,XX, mainly represented by karyotypes with the presence of an isochromosome Xq (45,X/46,X,I(Xq) -> 46X,I(Xq)), or low-degree mosaicism (45,X (3%)/46,XX (97%)) (Fig. 2D). This substantiated our finding, since most TS other karyotypes clustered along TS 45,X monosomy cases. Based on X chromosomal CpGs, a continuum between TS 45,X and 46,XX was observed (Fig. 2E). Again, this seemed to be driven by the degree of mosaicism and/or presence of an X chromosome q-arm. Only four differentially methylated positions (DMPs) were found when comparing TS 45,X and TS other karyotypes (Supplementary Data 1), and as such, TS other karyotypes in this cohort only affects the genome-wide methylation pattern minimally, if at all.

To assess the genome-wide methylation effect of XIST, we compared the methylation pattern between the XIST-high and XIST-low TS other karyotypes. We found 27 autosomal DMPs (Supplementary Data 2). Enrichment analyses based on the nearest gene of the 27 DMPs, did not reveal significant enrichment in any pathways. Two DMPs, with closest genes KRT77 and ARRB2, respectively, overlapped in the two contrasts (TS 45,X vs TS other karyotypes and TS other karyotypes XIST-high vs TS other karyotypes XIST-low) (Fig. 2F). As such, these were considered DMPs specific to TS other karyotypes with XIST expression comparable to that of 46,XX females. Interestingly, increased levels of ARRB2, coding for the beta-arrestin protein that regulates G-protein coupled receptor signaling, have been shown to be involved in the development of abdominal aortic aneurysm formation in mice³³.

Cell type proportion estimation

Using the DNAm data in the “genomic cohort”, we estimated the proportion of different blood cell types in each group using the Houseman algorithm³⁴. TS 45,X and TS other karyotypes had significantly higher counts of granulocytes compared to 46,XX (Fig. 3A). To delineate the granulocyte population into subtypes, we employed EpiDISH using a detailed reference of 12 blood cell subtypes³⁴. Based on this, we identified neutrophils as the main granulocyte subtype increased in TS (Fig. 3B, Supplementary Fig. 2). To further validate this finding, we downloaded a publicly available blood single cell RNA sequencing dataset (scRNAseq) and identified the top neutrophil markers³⁵ (Fig. 3C). These markers were expressed at higher levels in TS 45,X and TS other karyotypes compared with 46,XX females, and as such, confirmed the findings from the DNAm data (Fig. 3D–H).

**Fig. 3: Blood cell type distribution and neutrophil marker expression in Turner Syndrome (TS).**

Gene networks associated with TS clinical traits

Several clinical traits were measured in the “genomic cohort” (Table 1). To establish links between gene expression, phenotypic traits, and neutrophil proportion, we used weighted gene correlation network analysis (WGCNA) to identify gene networks, termed modules, associated with TS. TS 45,X and TS other karyotypes were collected into one TS group because of their shared phenotype for this analysis.

Four gene modules had a positive correlation to TS (lightgreen (214 genes), green-yellow (275 genes), blue (2077 genes) and magenta (469 genes) (p < 0.01)), reflecting that genes within these modules were generally upregulated in TS (Fig. 4A). For the green-yellow module, enrichment analysis showed that the genes within this module were involved in platelet activation, fibrin clot formation, blood coagulation and muscle contraction – and interestingly in a TS phenotype context – also integrin signaling, cell surface interactions at the vascular wall and molecules associated with elastic fibers (Supplementary Data 3). Furthermore, this module was negatively correlated to aortic distensibility and a phenotype with high BMI and waist-hip ratio, and increased levels of triglycerides. The genes within the blue module were enriched in markers of neutrophil degranulation, innate immune system, cytokine signaling, and interleukin signaling (Supplementary Data 4). This was consistent with the blue module being almost perfectly correlated to the proportion of neutrophils in the blood samples (Fig. 4B). Interestingly, the increased proportion of neutrophils in TS, as represented by the blue module, correlated positively to increased heart rate (day, night and 24 h) (Fig. 4C, D), and also to decreased aortic distensibility, both consistent features of TS, possibly through neutrophils that affect these clinical traits (Fig. 4A). Both day and night heart rates have previously been shown to be 8–10 beats per minute higher in TS individuals^36,37. Taking the specific increase in neutrophils into account, it seems likely that the increased night heart rate could be associated to increased neutrophil levels. The magenta module did not show any high-significant enrichment in any terms and was not associated to clinical traits besides height and waist-hip ratio.

**Fig. 4: Correlation of gene expression modules with clinical traits and neutrophil levels in Turner Syndrome (TS).**

One module, light-green, showed a strong negative correlation to TS, meaning that most of the genes in this network were downregulated in TS compared to 46,XX. Furthermore, this module was correlated to traits typically associated with TS. As expected, this module had a large proportion of X chromosome genes including escape genes, and as such, the correlation to all TS traits were confounded by X chromosomal gene expression (Supplementary Fig. 3).

Neutrophil, lymphocyte, and neutrophil-lymphocyte-ratio

While neutrophils have traditionally been recognized for their rapid response to infections, emerging evidence has highlighted their involvement in chronic inflammatory processes associated with various metabolic diseases. In conditions such as obesity and type 2 diabetes, dysregulated metabolic homeostasis triggers a sustained low-grade inflammation, often involving activated neutrophils that release pro-inflammatory mediators³⁸. Because of the strong correlation between a TS-associated gene network explained by neutrophil levels, and its correlation to increased heart rate and decreased descending aortic distensibility, we investigated this further in individuals with TS. The neutrophil proportion was higher in TS (Fig. 5A). Based on this, we also calculated the neutrophil-lymphocyte-ratio (NLR) as a proxy measure of the body’s inflammatory state³⁹ (Fig. 5B). The NLR was higher for individuals with TS compared to 46,XX. Estrogen replacement therapy (ERT), antihypertensive, statins or antidiabetic treatment did not affect neutrophil levels (Supplementary Fig. 4). The increase in neutrophil ratio in TS individuals was further shown in the extended cohort², including: 45,X; 46,XX; 46XY and 47,XXY (Fig. 5C). In this cohort, neutrophils were increased in TS individuals compared to all other karyotypes. Furthermore, 47,XXY KS males had lower levels of neutrophils, and as such, the number of sex chromosomes correlated negatively to neutrophil levels (Fig. 5C).

**Fig. 5: Analysis of neutrophil dynamics and inflammation markers in Turner Syndrome (TS).**

Granulocyte measurements using Flow cytometry

As the above neutrophil levels were based on cell proportion estimations from DNAm data, we included an independent cohort consisting of 20 individuals with TS and 107 female controls (the “flow cytometry cohort”), (Fig. 1A). We quantified blood cell types, and sub-populations of immune cells using flow cytometry. Again, granulocytes (CD3-, CD19-, CD14-) were more abundant in TS (Fig. 5D). Individuals with TS had a mean GLR (granulocyte/lymphocyte-ratio) residing around low-grade inflammation, compared to a normal level in controls (Fig. 5E). B-cell and T-cell counts, both numerically and proportionally, were comparable between individuals with TS and controls (Supplementary Data 5). One B-cell subtype: IgD-, CD27-, CD38-, a type of plasmablast, was numerically and proportionally higher in individuals with TS compared to controls. Besides this, the B-cell and T-cell profile of individuals with TS was like that of healthy controls (Supplementary Data 5). In addition, the granulocyte concentration was positively correlated with abdominal fat percentage in individuals with TS (Fig. 5F).

Neutrophil activation and NETosis

Pro-inflammatory neutrophils, when activated, can contribute to inflammation. Notably, the formation of neutrophil extracellular traps (NETs) has been associated with thrombosis and vascular inflammation in chronic inflammatory conditions⁴⁰. Thus, we investigated the key markers associated with a primed pro-inflammatory activated profile of the neutrophils. In the genomic cohort, we observed increased gene expression of the alarmins S100A8, S100A9 in TS, signaling activating immune responses – and further recruitment of neutrophils (Fig. 5G, H). In addition, we observed a strong upregulation of neutrophil elastase (ELANE) and myeloperoxidase (MPO) in a subset of the individuals with TS (Fig. 5I, J). Heightened activity of ELANE and MPO is of interest in chronic inflammation due to their roles in tissue damage, oxidative stress, and perpetuation of inflammatory cascades. Furthermore, ELANE and MPO are key contributors to the formation of NETs and sustained NETosis have been shown to contribute to tissue damage, exacerbating the inflammatory response⁴¹. Contrary to this, only negligible expression of ELANE and MPO was present in the control females. Thus, in addition to having higher number of neutrophils, individuals with TS also seemed to have an increased number of primed neutrophils (S100A8, S100A9), and increased levels of markers of NETosis (ELANE, MPO, PADI4), pointing towards an increased level of chronic inflammatory stress, possibly driven by neutrophils.

The genomic origin of increased neutrophil count

Next, we examined the genomic underpinnings of increased neutrophil count in TS individuals by correlating neutrophil proportion with the expression of genes in PAR1 and X/Y homologous genes. The most significant correlation across all karyotypes (extended cohort), as well as within specific karyotypes was with the gene expression of the gene TBL1X situated on the p-arm of the X-chromosome (Fig. 6A, B; Supplementary Fig. 5). Contrary to expectations, TBL1X expression was inversely correlated with the number of sex chromosomes (Fig. 6B); higher expression levels were associated with reduced sex chromosome count. This suggests that TBL1X plays a role in the modulation of neutrophil levels and hints at a possible genetic mechanism behind the elevated neutrophil levels observed in TS individuals.

**Fig. 6: Correlation of neutrophil fraction with *TBL1X* expression across different karyotypes.**

Neutrophil levels during routine outpatient visits in TS individuals

We analyzed absolute neutrophil counts in TS individuals managed as outpatients at Aarhus University Hospital, where routine leukocyte counts had been conducted (outpatient cohort). In this cohort, we examined neutrophil levels in individuals with KS compared to those with TS, using linear mixed models and adjusting for age (Fig. 6C, D). We observed a slight, non-significant increase in neutrophils with increasing age (p = 0.056), with no difference in the increase-with-age between KS and TS (p = 0.53). In adults, TS individuals had higher neutrophil levels compared to KS, with a fixed age difference (diff = 1.17·10⁹/l, 95% CI: 0.4 to 2.3, p = 0.04) (Fig. 6D). In children under 18 years, no association was found between age and neutrophil levels. After adjusting for age, TS individuals had neutrophils levels that were 2.7 10⁹/l higher than KS (95% CI: 0.47 to 4.9, p = 0.02, Fig. 6C). As such, the difference in neutrophils was present already in childhood. Given the higher incidence of autoimmune diseases in TS individuals, we examined neutrophil trends in a subgroup diagnosed with autoimmune disease (Supplementary Data 6), (n = 67), organizing data around the date of diagnosis (Fig. 6E, t = 0). We found that neutrophil counts rose leading up to the diagnosis and fell thereafter, likely due to the commencement of treatment. This suggested that rising neutrophil levels could also serve as an early indicator of developing autoimmune disease in TS individuals. Supporting this, data from the “epidemiological cohort” showed that TS individuals had a significantly higher risk of developing metabolic and autoimmune conditions compared with both KS and females with 47,XXX, while the risk of an infection was similar for all groups (Table 2). Thus, the difference in neutrophil levels (despite only using samples from routine checks) was not due to an increase in infections, and therefore, more likely to be a result of their respective karyotypes.

Table 2 Comparison of events related to autoimmunity, metabolic diagnoses, and infections in Turner syndrome, Klinefelter syndrome, and 47,XXX

Full size table

This table presents the risk of the first occurrence of autoimmune, metabolic, and infectious diseases in individuals with Klinefelter syndrome (KS), TS, and Triple X syndrome (47,XXX). KS risks were compared to matched male controls, while TS risks were compared to matched female controls. The relative risks for KS and TS were analyzed using a ratio of ratios. For the TS versus 47,XXX comparison, only age was adjusted for since sex differences were not relevant. For more details, see the statistics section.

Discussion

This comprehensive genomic analysis of individuals with TS 45,X and TS with other karyotypes provides insights into the shared molecular signatures and potential clinical implications associated with these different karyotypes. The investigation into the autosomal transcriptome and methylome reveals notable similarities between TS 45,X and TS other karyotypes, emphasizing a common methylome and transcriptome despite distinct karyotypic variations of the X chromosome. Furthermore, the observed presence of increased neutrophil levels and neutrophil activation offer additional insight into the molecular pathophysiology associated with TS and establishes new connections to the TS phenotype.

The shared autosomal transcriptome and methylome profiles between TS 45,X and TS other karyotypes suggest a common underlying mechanism, influencing gene expression and DNA methylation patterns. This similarity raises questions about the potential convergence of regulatory pathways in these disparate karyotypic backgrounds. Despite the heterogeneity within the TS other karyotypes, the near identical phenotype shared with the TS 45,X group suggests that downregulated PAR1 genes and X-Y homologous genes affecting autosomal transcriptome and methylome play pivotal roles in shaping the clinical manifestations of TS. Supporting this, 45,X TS, and TS other karyotypes share the expression profile for all PAR1 genes – indicating that the PAR1 genes are important for manifestations of the TS phenotype^2,3,42,43. This was further supported by the findings in individuals with a 46,X,i(Xq) karyotype, lacking the p-arm of one of the X chromosomes. As such, the X chromosomal p-arm, and perhaps in particular PAR1, appears to be the main driver influencing the autosomal transcriptome and methylome. The X-Y homologs, primarily situated on Xp, are also likely to play an important role^2,3,5,42.

The expression of XIST did not affect the autosomal transcriptome or methylome when we subdivided the TS individuals into an XIST-high, with expression comparable to that of 46,XX, and an XIST-low group. As such, this further strengthened the idea that the phenotypical traits of TS primarily stem from lacking the p-arm of the second X chromosome. The above could be exaggerated because of ascertainment bias in the selection of TS individuals for this study. Low-degree mosaicism and karyotypical representations where only the q-arm of the X chromosome is lacking, are not present in this study, except for one case with low level mosaicism in blood—possibly because TS traits are not present in low level 45,X/46,XX mosaicism and they thus remain undiagnosed. This was recently shown in the UK biobank database, where many cases with 45,X/46,XX seemed to have no, or almost no TS traits, and had never been diagnosed clinically¹⁵, emphasizing that the diagnosis of TS should indeed be based on a clinical diagnosis and not a karyotypic diagnosis¹.

Various clinical traits had previously been measured for the “genomic cohort”. We observed interesting associations between gene expression networks and clinical traits. We observed upregulation of three gene networks (green-yellow, blue, magenta) in TS. The genes within the green-yellow module were enriched in pathways related to platelet activation, blood coagulation, and muscle contraction, in addition to integrin signaling, cell surface interactions at the vascular wall, and molecules associated with elastic fibers. This emphasized its possible involvement in a cardiovascular context. Combined with the fact that this network was correlated to decreased aortic distensibility, aligns with the TS phenotype¹⁰. This correlation was particularly present in individuals with high BMI, increased waist-to-hip ratio, and elevated triglyceride levels, indicating a potential link between this gene network and cardiovascular and metabolic risk factors in TS^7,12,44. The upregulation of genes within this network may contribute to impaired vascular elasticity, providing a molecular basis for the observed cardiovascular and metabolic traits in TS. In this context, aortic distensibility is a key factor in maintaining arterial flexibility and preventing adverse cardiovascular events, including aortic dissection^28,45. The blue module, enriched in pathways related to neutrophil degranulation and activation, innate immune system, and cytokine signaling, exhibited an almost perfect correlation with neutrophil proportion. The neutrophil proportion was higher in TS compared to 46,XX, and the positive correlation between the increased proportion of neutrophils and elevated heart rate, abdominal body fat, along with reduced aortic distensibility, suggests a role of neutrophils in influencing these outcomes in TS.

In the last decade the significant role of neutrophils contributing to chronic inflammation in metabolic diseases has been highlighted, actively supporting inflammation by the release of serine proteases and formation of NETs³⁸. Our investigation into neutrophils, lymphocytes and NLR in TS unveils heightened neutrophil levels and an elevated NLR, indicating a chronic low-inflammatory state in individuals with TS. We verified increased granulocyte and neutrophil levels in 3 cohorts using 3 different methodologies. In the “genomic cohort” granulocytes and neutrophil levels were estimated from DNAm data from whole blood. Here, we showed that neutrophils, not basophils or eosinophil, were responsible for the increased levels of granulocytes. In the “flow cytometry cohort” granulocyte levels were quantified using flow cytometry and granulocyte levels in TS individuals were positively correlated to increased abdominal fat mass. In the “outpatient cohort” absolute neutrophil counts were quantified as part of standard follow-up visits. The slight, non-significant increase in neutrophils with age, alongside no observed difference between KS and TS in age-related increase, indicates that while age may impact neutrophil levels, the effect is not different between these two conditions. However, the observation that adults with TS have higher neutrophil levels than those with KS, with a fixed age difference, and that TS individuals have higher neutrophil levels than KS after adjusting for age, particularly in childhood, suggests a distinct neutrophil profile only present in TS individuals. Thus, neutrophils and low-grade inflammation may not play the same role in KS⁴⁶. This profile may be an intrinsic aspect of TS pathology, not merely a consequence of aging or secondary health conditions. The neutrophil trends in TS individuals diagnosed with autoimmune diseases are particularly interesting. The increase in neutrophil counts leading up to an autoimmune disease diagnosis and the subsequent decline post-diagnosis, possibly because of treatment initiation, suggests a potential predictive value of neutrophil monitoring. Considering these findings, the elevated neutrophil counts in TS raise questions about the mechanisms driving this increase. Is this a direct result of genetic abnormalities in TS, or does it reflect a complex interplay of genetic, environmental, and immune factors? Here, we show a strong correlation between neutrophil counts and TBL1X expression, particularly its inverse relation with the number of sex chromosomes. This highlights a potential genetic basis for the elevated neutrophil levels and low-grade inflammation in TS, where TBL1X may play a key role in neutrophil counts, and thus, modulating immune responses. The link between TBL1X and NF-kappa B signaling, especially its role in enhancing p65 transcriptional activity in response to inflammatory stimuli like TNF-alpha, could provide a molecular context for TBL1X involvement in inflammatory processes, including recruitment and activation of neutrophils⁴⁷. Moreover, studies have shown that neutrophils, and especially NE, have a significant role in the development of vascular inflammation, aortic aneurysm, and thoracic aortic dissection⁴⁸. In one study, TBL1X was identified as a downstream target of NE, implicating it in vascular pathologies, while inhibiting TBL1X impacted NE-driven signaling pathways, decreasing inflammatory cell migration and risk of aortic dissection⁴⁸.

These findings align with emerging evidence linking neutrophils with chronic inflammatory processes in metabolic diseases such as obesity and type 2 diabetes^49,50, autoimmune disease⁵¹ and cardiovascular risk⁵². We and others have previously found elevated levels of cytokines and low-grade inflammation in TS^26,53,54,55. In metabolic disorders like type 2 diabetes, hyperglycemia, and hyperlipidemia increase the generation of primed pro-inflammatory neutrophils, contributing to adipose tissue inflammation and the persistent inflammatory state associated with type 2 diabetes^49,50. ELANE and MPO, released during neutrophil degranulation, are implicated in oxidative bursts, while their excessive activity has also been linked to the formation of NETs. Thus, addressing subclinical metabolic inflammation might be important in TS. In this study, we observed a strong increase in gene expression of both MPO and ELANE, with this increase being driven by a subset of TS individuals that had more than 100 times higher expression levels than controls. As such, MPO and ELANE could be potential biomarkers to detect at-risk TS individuals with a pathophysiological immune profile.

Presently, we view the low-grade inflammatory condition among TS as multifactorial, being partially spurned by the specific methylation profile, affecting the transcriptome, and accentuated by increased abdominal body fat accumulation and perhaps a reduced physical fitness. There may well be additional influential factors not presently identified. In conclusion, we hypothesize that such a, perhaps life-long, low-grade inflammatory milieu may also explain some of the increased propensity for developing autoimmunity. If such a relation exists it implies that preventive measures lowering, or even ameliorating, the low-grade inflammatory condition could affect both the development of metabolic conditions, but also autoimmune conditions among individuals with TS.

Limitations

Our cross-sectional study identifies an association between increased neutrophil activation, low-grade inflammation, TS and TS co-morbidities, but, given the design, cannot establish causality. Our data indicates that elevated neutrophil counts are already present in young TS adults without metabolic conditions and that TBL1X expression is increased in TS and positively correlated with neutrophil levels. This suggests that increased neutrophil levels, and increased neutrophil activation, may precede the development of comorbidities in TS. However, it is also possible that the frequent metabolic conditions experienced by TS individuals worsen chronic inflammation. Additionally, comparing TS individuals to healthy controls, without matching for the higher prevalence of metabolic and cardiovascular conditions, may confound the observed associations. Future longitudinal and functional studies are necessary to establish causality and better understand the relationship between TBL1X, neutrophils, inflammation and adverse health outcomes in TS.

Methods

The cohorts (Fig. 1)

The genomic cohort

Individuals with karyotypically proven TS and their age-matched 46,XX controls, previously participating in a comprehensive study of the cardiovascular phenotype of TS (ClinicalTrial.gov #NCT00624949), were included in this study³⁰. DNA-methylation (Infinium HumanMethylation450) and gene expression (RNAseq) profiling have previously been published on the TS 45,X and 46,XX individuals^2,16 (DNAm: TS 45,X = 32, 46,XX = 33; RNAseq: TS 45X = 34, 46,XX = 33). RNAseq and DNAm for the TS other karyotypes were included specifically for this study (TS other karyotypes = 23). Informed written consent was obtained. Aarhus County Ethical Scientific Committee approved the trial protocol (#20010248) and all clinical investigation was conducted according to the principles expressed in the Declaration of Helsinki.

The extended cohort

To further assess neutrophil levels, we included the TS, 45X and 46,XX controls, also present in the genomic cohort (cross-over individuals), with the addition of 47,XXY Klinefelter males (n = 22) and 46,XY controls (n = 16). We included gene expression of PAR1 and X/Y homolog genes to correlate these to neutrophil levels. Gene expression data and DNAm from this cohort has previously been published². ClinicalTrials.gov (NCT00624949, NCT00999310, NCT02526628, NCT01678261).

The flow cytometry cohort

A flow cytometry cohort was established to verify the findings from the Genomic cohort. This cohort consisted of 20 Individuals with TS, 10 with 45,X and 12 with other karyotypes (45,X/46,XX: n = 7; 46,X del(X)(p21.3) n = 2; 45,X/46,XY: n = 1). These were recruited through the Department of Endocrinology at Aarhus University Hospital as part of project investigating endocrinological, cardiovascular, metabolic, and immunologic aspect of sex chromosome abnormalities with ethical approval (Region Midtjylland #1-10-72-186-21) and clinicalTrials.gov #NCT05425953. Inclusion criteria where age 18–70 and exclusion criteria were, acute illness. Apart from flow cytometry this cohort underwent clinical examination including DXA scan, 24 h blood pressure measurements and standard biochemistry. Age and gender matched controls for flow cytometry were recruited anonymously from a blood-donor registry and included 107 females.

The outpatient cohort

We included routinely sampled biochemistry data from another cohort of individuals with both TS (n = 135) and KS (n = 139) from the outpatient clinic at Aarhus University Hospital, who at some point had had a full examination of leukocytes performed as part of a general clinical workup, but not in conjunction with an acute illness. We retrieved the regional electronic patient records which also includes all biochemical analyses performed. The data utilized herein were deemed exempt by the Central Denmark Region Ethics Committee because no sensitive or personally identifying information was extracted (approved by the local ethics committee at Aarhus University Hospital).

The epidemiological cohort

We utilized data on TS (n = 1156)⁵⁶, KS (n = 1155)⁴⁶, and 47,XXX (n = 160)⁵⁷ from the national Danish Cytogenetic Registry, as previously described. From Statistics Denmark, the central authority on Danish national health care registries, we retrieved data on sex- and age-matched controls (1:100) and on admissions to the health care system. Statistics Denmark uses The International Classification of Diseases (ICD) 10th edition for the registration of admissions. All Danish citizens are assigned a unique identification number, which is used for data linkage across registries. Data were updated until 2014.

Recruitment, inclusion, and exclusion

In the genomic cohort, TS other karyotypes, in addition to the 45,X monosomy and 46,XX controls, were recruited through the Danish National Society for Turner Syndrome contact group and the endocrine outpatient clinic, while controls were recruited through websites and newspapers. The inclusion criteria were a verified diagnosis of TS and an age range of 18–70 years. The exclusion criteria included extreme obesity (due to the study involving MRI), other contraindications to an MRI scan, and malignant disease. Similar inclusion, exclusion, and recruitment criteria were applied to the Extended cohort and the Flow Cytometry cohort. For all cohorts, the TS individuals and KS males had age-matched 46,XX females controls and 46,XY males, respectively. For further details on recruitment, Inclusion, and exclusion, see previously published articles on these cohorts^2,30,58.

Blood sample collection and processing

Blood samples were collected from the antecubital vein and stored in PAXgene Blood RNA Tubes (RNAseq) or EDTA tubes (DNAm). The PAXgene samples were initially incubated at room temperature for 2 h and subsequently stored at -21 degrees Celsius overnight before long-term storage at -80 degrees Celsius. Total RNA was extracted using the PaxGene blood Kit 262174 (Qiagen), both before and after DNase treatment. The quality of RNA samples was assessed through on-chip electrophoresis using a Tapestation 4200 RNA Screen Tape System (Agilent) and by UV measurements on a Lunatic (Unchained Labs). Subsequently, directional RNA-seq libraries were prepared using the KAPA RNA HyperPrep with RiboErase Globin (HMR) (Roche) following the recommended procedure, with library preparation being automated using a Sciclone NGS (Caliper, Perkin Elmer) liquid handling robot. The RNA-Seq library qualities were evaluated using the Tapestation 4200 D100 Screen Tape System (Agilent), and library concentrations were estimated using the Qubit dsDNA HS Assay (Thermo Fisher). A total of 500 ng of total RNA was utilized as input material.

RNA sequencing

The RNA-seq libraries were multiplexed and subjected to paired-end sequencing on an Illumina NovaSeq 6000 platform (100 bp). Paired de-multiplexed fastq files underwent initial quality control using FastQC (Babraham Bioinformatics). Adapter removal and trimming of low-quality ends were performed using Trim Galore with default settings (Babraham Bioinformatics). Gene expression levels were quantified using Salmon, with a decoy-aware transcriptome index based on the hg38 transcriptome. Transcript abundances were summarized to the gene level using the R package Tximeta. Differential expression analysis was carried out using the R Bioconductor package DESeq2, with a false discovery rate (FDR) threshold of less than 0.05. Gene enrichment analysis was carried out using enrichR. The classification of X chromosome genes into distinct classes was based on the approach described by Tukiainen et al.³².

DNA methylation analysis

Infinium® HumanMethylation450 Beadchip Kit (Illumina, Inc.) data for TS other karyotypes was acquired in the exact same way, at the same time, as the 46,XX and 45,X data. In short, genomic DNA was purified and then subjected to bisulfite conversion using the Zymo EZ DNA-methylation Kit. Methylation levels were assessed using the Infinium® HumanMethylation450 Beadchip Kit (Illumina, Inc.) at Aros Applied Biotechnology A/S. Raw intensity values were imported into R (v. 4.1.1) and further processed using the R package Minfi. Cross-reactive probes and poorly performing probes, as indicated by a detection p < 0.01, were excluded from the analysis. The preprocessFunnorm normalization method was applied to remove between-array variation inferred by control probes, followed by the conversion of methylation values to M values (logit [beta]). For differential methylation analysis, M values were analyzed using LIMMA, with differentially methylated positions defined as having an adjusted p < 0.05. To investigate for differences in cell composition Minfi’s estimateCellCounts, the Houseman algorithm⁵⁹ was used to calculate the relative proportions of CD4+ and CD8+ T-cells, natural killer cells, monocytes, granulocytes, and B-cells in each sample^59,60. The EpiDISH algorithm was employed using a reference of 12 blood cell subtypes “cent12CT450k.m”³⁴

Weighted correlation network analysis (WGCNA)

To identify co-expressed genes from the RNA-seq data and relate them to the number of sex chromosomes, we applied weighted correlation network analysis (WGCNA, v1.70.3). A signed co-expression network was constructed for each tissue using a one-step approach, with adjacency calculated by selecting an appropriate soft thresholding power to approximate scale-free topology. Gene clustering was performed on the signed Topology Overlap Matrix using hierarchical clustering, and modules were identified via the blockwiseModules function with a minimum module size of 30 and a mergeCutHeight of 0.25. Module eigengenes were calculated using the moduleEigengenes function, and eigengene significance and corresponding p-values were obtained for each module-trait association. Hub genes for modules of interest were identified using the chooseTopHubInEachModule function.

Flow cytometry

The immunophenotyping assay involved determining both relative and absolute concentrations of prominent immunological cellular subsets in heparinized peripheral blood (200 µL). Three different panels of antibodies targeting the following antigens were utilized: Lineage-panel: CD45 [APC-H7], CD16 [APC], CD56 [BV421], CD14 [PE-Cy7], CD3 [FITC], and CD19 [PE]. T-cell panel: CD3 [APC-Cy7], CD8a [BV785], CD27 [BV421], CD4 [BV570], CD197 (CCR7) [PE], CD45RA [APC], TCR γδ [FITC], and HLA-DR [PE-Cy7]. B-cell panel: CD21 [PE-Cy7], Kappa-LC [APC], IgD [PE], CD27 [VioBright FITC], CD19 [SuperBright 600], CD38 [BV421], and CD20 [BV785]. The identification of immunological subsets was based on antibody staining patterns, including T cells, B cells, NK cells, NKT cells, monocytes, and granulocytes. Further classification included Helper T cells (CD4+) and Cytotoxic T cells (CD8+), each subclassified as Naïve, Central memory, or Effector memory, and assessed for activation status (HLA-DR) and TCR γδ expression. B cells were further identified as Transitional, Naïve, Class-switched memory, Plasma blasts, and Marginal-zone-like B cells, classified according to CD21 expression level.

Statistics

All data were presented in boxplots with individual data points present. We tested for normality using the Shapiro-Wilk normality test. For normally distributed data, a student’s t-test was used for comparing means. For comparing non-parametric data, a Wilcoxon sign-rank test was used. When comparing more than two groups, the ANOVA test was used to test differences between groups. We used linear mixed-models to assess the relationship between neutrophils and age at visit, incorporating Unique ID as a grouping variable to capture individual differences over time. This approach facilitated the analysis of fixed (age) and random (neutrophil variability) effects within a longitudinal framework. Differences in neutrophil levels between individuals with KS and TS were assessed similarly using linear mixed models. These models also included an interaction term between age and KS/TS status, allowing for the evaluation of age-related differences in neutrophil levels between the two conditions. This approach facilitated a fixed-effect comparison of age and condition on neutrophil levels. Analyses were performed using Stata ver. 17 ^©.

For the epidemiological study, we defined an admission to the health care system, as a measure of morbidity. We focused on admissions in three subgroups: autoimmunity (Supplementary Data 6), infections, corresponding to ICD-10 chapter A00 to B99, and metabolic conditions: Type 2 diabetes: DE11-DE119A overweight/obesity: D66-DE669, metabolic syndrome: DE888C, hypertension: DI109, acute myocardial infarction: DI21-DI219A, chronic ischemic heart disease: DI252-DI252C, hypercholesterolemia: DE780, and hyperlipidemia: DE785. For each comparison, we calculated first events only. For the comparison on morbidity for TS and KS, we adjusted for age and time at risk in 10-year intervals. Thus, we calculated the number of events in TS cases relative to the number of events in TS controls. We divided this ratio with the number of events in KS cases relative to the number of events in KS controls. For the direct comparison of events in 47,XXX and TS, we calculated Cox regression, adjusted for year of birth.