Serum levels of the hepcidin-20 isoform in a large general population: the Val Borbera study.

Hepcidin, a 25 amino-acid liver hormone, has recently emerged as the key regulator of iron homeostasis. Proteomic studies in limited number of subjects have shown that biological fluids can also contain truncated isoforms, whose role remains to be elucidated. We report, for the first time, data about serum levels of the hepcidin-20 isoform (hep-20) in a general population, taking advantage of the Val Borbera (VB) study where hepcidin-25 (hep-25) was measured by SELDI-TOF-MS. Detectable amount of hep-20 were found in sera from 854 out of 1577 subjects (54.2%), and its levels were about 14% of hep-25 levels. A small fraction of subjects (n=30, 1.9%) had detectable hep-20 but undetectable hep-25. In multivariate regression models, significant predictors of hep-20 were hep-25 and age in males, and hep-25, age, serum ferritin and body mass index in females. Of note, the hep-25:hep-20 ratio was not constant in the VB population, but increased progressively with increasing ferritin levels. This is not consistent with the simplistic view of hep-20 as a mere catabolic byproduct of hep-25. Although a possible active regulation of hep-20 production needs further confirmation, our results may also have implications for immunoassays for serum hepcidin based on antibodies lacking specificity for hep-25. This article is part of a Special Issue entitled: Integrated omics.

Introduction

In the last decade hepcidin, a small peptide hormone, has emerged as the key regulator of systemic iron homeostasis [1,2]. Hepcidin inhibits the intestinal absorption of dietary iron and the release of iron from macrophages through the interaction with the transmembrane iron exporter ferroportin, causing its internalization and degradation in lysosomes [3]. The iron bioactive form of hepcidin is a 25-amino acid peptide (hep-25) that shares high homology with defensins, a family of antimicrobial peptides of the innate immunity [4]. It is produced mainly by the liver as an 84-amino acid precursor that subsequently undergoes proteolytic cleavages to generate the mature form [5]. Further hep-25 processing can result in the generation of two amino-terminal truncated isoforms, hepcidin-22 (hep-22) and hepcidin-20 (hep-20), whose physiological role is still unclear [6]. The development of a reliable hepcidin assay has been proven difficult, particularly with classical immunological methods, yielding to a number of different approaches [7]. Among these, Mass Spectrometry (MS) based studies in limited number of subjects have identified and measured small amounts of hep-20 in both serum and urine, while hep-22 has been found only in urine [8]. Of note, functional studies have demonstrated that the two truncated isoforms almost completely loss the ability to interact with ferroportin [9]. Being inactive in iron regulation, they have been postulated to be degradation byproducts of hep-25. On the other hand, recent studies have suggested that hep-20 may retain greater antimicrobial and fungicidal activity than hep-25, particularly at acidic pH (pH 5.0) [10,11]. In the small series published until now, relatively high levels of hep-20 have been detected in heterogeneous pathological conditions like acute myocardial infarction (AMI) [12], anemia of chronic disease (ACD) [13], and, particularly, in chronic kidney disease (CKD) [14-16]. In our previous study using an improved Surface Enhanced Laser Desorption/Ionization Time of Flight-MS (SELDI-TOF-MS) approach where 54 patients in chronic hemodialysis were compared with 57 controls, hep-20 was detectable in 100% and 39%, respectively [15]. Kroot et al. [17] evaluated 186 patients with various diseases and 23 healthy controls by a weak cation exchange (WCX)-TOF-MS assay. No isoforms were found in sera from healthy subjects and patients with low hepcidin concentrations, while relevant amount of hep-20 were found in some conditions, again particularly in CKD. These studies highlighted that the contribution of hep-20 to “total” serum hepcidin measured by a non-selective ELISA assay is not constant but may vary substantially between healthy subjects and patients with different diseases. However, large-scale information about serum hep-20 concentration is lacking. Thus, the present study was aimed at quantifying serum hep-20 within the framework of the recently completed Iron Section of the Val Borbera (VB) study [18], where serum hepcidin was measured by SELDI-TOF-MS. To the best of our knowledge, we report here for the first time in a large general population the levels of hep-20 according to age and sex, their determinants, and the variations of hep-25:hep-20 ratio according to the iron status.

Material and methods

Subjects and biochemical analyses

This study included 1577 subjects aged 18–98 years enrolled in the Iron Section of the VB study. Details about the general design of this study, as well as the enrollment criteria have been extensively reported elsewhere [18,19]. The study was approved by the San Raffaele Hospital and Regione Piemonte ethical committees, and all subjects gave written informed consent. For each participant, anthropometric, complete blood cell count, and biochemical parameters including serum iron, transferrin, transferrin saturation, ferritin, creatinine, and markers of inflammation were available. Serum hepcidin isoforms were determined using a SELDI-TOF-MS assay previously used for their identification and characterization [8], and subsequently validated for quantification by several investigators including our group [12-18,20]. Briefly, copper loaded immobilized metal affinity capture ProteinChip arrays (IMAC-30 Cu2 +) were selected as chromatographic surface. A synthetic hepcidin analogue, hepcidin-24 (hep-24, purchased from Peptide International, Louisville, KY), was used as internal standard for quantification. Spectra were collected in duplicate for each serum samples, with or without spiking of the internal standard at a concentration of 10 nM (Fig. 1). With respect to hep-20 quantification, we used as reference peak hep-20 kindly provided by Dr. Elizabeta Nemeth (University of California Los Angeles, CA). Hep-20 concentration was expressed in nM resulting from the following equation: (sample 2192 m/z peak intensity) × 10 nM/(hep-24 spiked sample 2673 m/z peak intensity–nonspiked sample 2673 m/z peak intensity). Based on the measured background noise in MS spectra, the lower limit of detection (LLOD) for both hepcidin isoforms was 0.55 nM. It was calculated on 20 randomly selected VB sera, and was in agreement with the LLOD previously published by our group and others [20]. Adequate linear standard curve (y = 1.566 × − 0.254, R2 = 0.998) was obtained by serially diluting synthetic hepcidin-20 in blank serum (diluted 1:100 in water). Intra-day precision was determined by using three different concentrations of hep-20 (5, 10, and 25 nM), each with seven replicates. This was repeated on four separate days. Precision was assessed by coefficient of variation (CV%). The within- and between run precisions at the low concentrations similar to what found in VB subjects were 6.9% and 7.0%, respectively. The mean peak intensity ratio hepcidin-24/hepcidin-20 in blank serum spiked with both peptides at 7 different concentration combinations (25, 20, 15, 12.5, 10, 7.5, and 5 nM) was 0.85, i.e. similar to what obtained for the main isoform hepcidin-25 [15].

Fig. 1

Representative SELDI-TOF-MS profile of serum samples from Val Borbera cohort with (A) and without (B) hep-24 internal standard. The hepcidin isoforms hep-20, hep-24 (synthetic analogue), and hep-25 are indicated by rectangles.

Statistical analyses

Statistical analyses were performed using SPSS 17.0 software (SPSS Inc., Chicago, IL, USA). Continuous variables were expressed as means ± standard deviations, while those with a skewed distribution, including hep-25, hep-20, hep-25:hep-20 ratio, ferritin, CRP and creatinine, were log-transformed and expressed as geometric means with 95% confidence intervals (CIs). Considering that in many subjects the levels of hepcidin (in particular those of hep-20) were not detectable, not allowing a correct log-transformation, the relative statistical analyses in the whole study population were performed adding the value of 0.1 to each hepcidin value and considering not detectable hepcidin levels as 0 (thus, the log transformable 0.1 [0 + 0.1] for statistical analysis). For sake of completeness, the analyses on hep-25 and hep-20 were performed also in the subgroups with detectable levels. On the other hand, the analyses on hep-25:hep-20 ratio were performed only in subjects with detectable levels of both isoforms. Quantitative data were analyzed using the Student's t test or by analysis of variance (ANOVA) with polynomial contrasts for linear trend when indicated. Sex specific correlations between quantitative variables were assessed using Pearson's test. Qualitative data were analyzed using the chi-square test, with analysis for linear trend when indicated. Independent determinants of serum hep-20 and hep-25:hep-20 ratio were estimated by means of linear regression models, including all the variables significantly associated at univariate analysis. Two-sides P values < 0.05 were considered statistically significant.

Results

The main characteristics of the total population are listed in Table 1. This table also lists data stratified by gender, since we previously observed significant gender differences in serum levels of the bioactive hep-25 isoform, as reported in detail elsewhere [18]. Briefly recapitulating, hep-25 was significantly lower in women aged < 50 years as compared to men of same age, while after this time-point (nearly corresponding to women's menopause) hep-25 levels tended to be similar in both sexes and relatively stable over the following decades. This reflected mainly the lower iron status (i.e. lower ferritin levels) of women during the fertile period, where iron absorption from the gut needs to be allowed by low hep-25 to counterbalance iron losses with menses. Hep-20 was detectable in 854 out of 1577 (54.2%) subjects, at variance with hep-25, which was detectable in 89.1% of subjects. The main characteristics of the subjects stratified for detectable serum hepcidin isoform are reported in Supplementary Table 1. Of note, there was a group, albeit small (n = 30, 1.9%), of subjects in whom only the “minor” hep-20 isoform was detectable. In general, as compared with subjects with undetectable hep-20, subjects with detectable hep-20 were older and had higher body iron status (as reflected by transferrin saturation and ferritin levels) and hep-25 levels (Supplementary Table 2).

Table 1

Main characteristics of the total population.

	Total populationN = 1577	MalesN = 706	FemalesN = 871	P
Age (years)	55.7 ± 17.8	55.3 ± 17.6	56.0 ± 18.1	0.436
BMI	25.9 ± 4.4	26.5 ± 3.8	25.5 ± 4.8	< 0.001
s-iron (μg/dl)	97.58 ± 34.11	105.83 ± 36.15	90.86 ± 30.80	< 0.001
Transferrin (mg/dl)	242.04 ± 41.52	235.97 ± 36.89	246.98 ± 44.34	< 0.001
Transferrin Saturation %	29.30 ± 11.62	32.30 ± 12.11	26.85 ± 10.60	< 0.001
Ferritina (ng/ml)	69 (65–72)	120 (114–128)	44 (41–47)	< 0.001
Hb (g/dl)	14.4 ± 1.4	15.4 ± 1.2	13.7 ± 1.1	< 0.001
CRPa (mg/l)	0.17 (0.16–0.18)	0.16 (0.15–0.17)	0.17 (0.16–0.18)	0.220
Creatininea (μmol/l)	0.85 (0.84–0.86)	0.96 (0.95–0.98)	0.77 (0.76–0.78)	< 0.001
Hep-20a (nmol/l)b	0.69 (0.63–0.75)	0.82 (0.72–0.94)	0.60 (0.53–0.67)	< 0.001
Hep-20a (nmol/l)c	3.31 (3.17–3.46)	3.26 (3.06–3.47)	3.35 (3.16–3.56)	0.530
Hep-20 detectable %	54.2	59.1	50.2	–
Hep-25a (nmol/l)b	4.96 (4.59–5.35)	7.09 (6.46–7.78)	3.71 (3.31–4.16)	< 0.001
Hep-25a (nmol/l)c	7.86 (7.54–8.19)	8.89 (8.38–9.44)	7.02 (6.63–7.44)	< 0.001
Hep-25 detectable %	89.1	94.6	84.6	–

Variables not normally distributed are expressed as geometric means with 95% CIs.

Geometric mean of hep-20 and hep-25 with 95% CIs calculated on whole population (1577 subjects).

Geometric mean of hep-20 and hep-25 with 95% CIs calculated on subjects with detectable hepcidin levels i.e. n = 854 and n = 1405 respectively.

The two hepcidin isoforms were significantly and positively correlated in both sexes (males: r = 0.48, P < 0.001; females: r = 0.45, P < 0.001; Fig. 2). At univariate analyses, hep-20 also significantly correlated with age, hemoglobin, and C-Reactive Protein (CRP) in men, and with age, body mass index (BMI), ferritin, CRP and creatinine in women (Table 2). Multivariate linear regression models showed hep-25 and age as independent significant predictors of hep-20 in men, while hep-25, age, BMI, and ferritin were significant predictors of hep-20 in women (Table 3A and B). Of note, the beta coefficient of ferritin in women was negative (Table 3B). Fig. 3A and B shows the variations of hep-20 in the VB population according to different ranges of age and iron status (reflected by serum ferritin levels), respectively. In addition, Supplementary Fig. 1 shows the corresponding hep-25 levels after stratification for ferritin levels. We then focused on the ratio between the two isoforms, and particularly on its behavior in relation with age and iron parameters of VB subject (Table 4 and 5). This analysis could be properly done only in subjects with both the hepcidin isoforms detectable (n = 824). The hep-25:hep-20 ratio was clearly lower in women aged < 50 years (i.e. premenopausal) as compared to men of corresponding age, while differences attenuated in the subsequent decades (Fig. 4A). Of note, the hep-25:hep-20 ratio was not constant in the VB population, but increased progressively according to increasing ferritin levels (Fig. 4B). Fig. 5A and B summarizes the relative percentages of hep-25 and hep-20 according to increasing ferritin levels. In both sexes, the relative percentage of hep-20 progressively and significantly decreased with increased ferritin levels.

Fig. 2

Correlation plot between hep-20 and hep-25 (logarithmic scale).

Table 2

Sex specific correlation analysis of hepcidin-20.

	Males		Females
Correlation	P	Correlation	P
Hep-25 (nmol/l)	0.480	< 0.001	0.449	< 0.001
Age (years)	0.096	0.049	0.178	< 0.001
BMI	0.049	0.315	0.186	< 0.001
s-iron (μg/dl)	− 0.028	0.571	− 0.023	0.627
Transferrin (mg/dl)	− 0.039	0.430	− 0.049	0.308
Transferrin saturation %	− 0.005	0.912	− 0.013	0.789
Ferritina (ng/ml)	0.062	0.206	0.149	0.002
Hb (g/dl)	− 0.110	0.024	− 0.048	0.320
CRPa (mg/l)	0.115	0.052	0.173	0.002
Creatininea (μmol/l)	− 0.028	0.583	0.109	0.029

Values in bold are those considered statistically significant (P < 0.05).

Variables not normally distributed are expressed as geometric means with 95% CIs.

Table 3

Predictors of Hep-20 levels in males (A) and females (B).

A
	Males
β-coefficient	P
Hepcidin25 (nmol/l)	0.627	< 0.001
Age (years)	0.126	0.007
B
	Females
	β-coefficient	P
Hepcidin25 (nmol/l)	0.553	< 0.001
Age (years)	0.177	0.004
Ferritin (μg/l)	− 0.255	0.001
BMI	0.163	0.003

Fig. 3

Behavior of hep-20 in VB population according to different ranges of age (A) and ferritin (B), respectively. Males are indicated by continuous blue line, females by a red dotted line.

Fig. 4

Hep-25:hep-20 ratio in groups of individuals classified according to age (A) and ferritin levels (B), respectively. Males are indicated by blue continuous line, females by a red dotted line.

Fig. 5

Relative percentages of hep-25 and hep-20 according to increasing ferritin levels in males (A) and females (B), respectively. The percentage value over the column represents the proportion of hep-20 on total hep (calculated as hep-20 + hep-25).

Discussion

Although much is known on the mechanism of action of hepcidin through the binding with its receptor/cellular exporter ferroportin [3], the mechanism(s) of hepcidin processing, secretion, and catabolism are still poorly elucidated. Initial efforts to establish reliable assays for this hormone have indicated that the entire pre-pro-hormone (84 amino acid) is also present in the circulation [21], while, at variance with hep-25, its concentration correlates poorly with iron status [22]. Moreover, small studies by MS-based techniques [7,8] have found that two further N-terminal truncated isoforms, namely hep-22 and hep-20, are present in biological fluids, particularly hep-20 in certain disease conditions like CKD [15,17]. Even though hep-20 may be either quantitatively or qualitatively the most relevant isoform of “mature” hepcidin, no large-scale population study has been conducted so far on its serum levels relative to hep-25, as well as on its possible determinants. This study establishes for the first time that more than half individuals at population level have detectable amount of hep-20 in the circulation. These individuals were generally included among those with discrete amount of serum hep-25, who in turn represented approximately 89% of the total VB population. As described in detail elsewhere [18], the remaining 11% of individuals were mostly represented by pre-menopausal women with highly prevalent low iron status [23], implying the need to suppress hepcidin for up-regulating the absorption of dietary iron [2]. The large VB database also allowed us to investigate the relevant variables associated with hep-20 levels. To put our results into perspective, we will discuss in more detail two main possible practical implications.

Hepcidin-20 as a possible caveat in hepcidin assay for clinical purpose

To the best of our knowledge, the VB is the first large-scale population study on serum hepcidin using a MS-based assay. The high correlation of hep-25 with iron status was reported elsewhere [18], and paralleled what observed in a Dutch population by another group using an ELISA method [24]. Notwithstanding recent considerable progress in the complex field of hepcidin assay [7], we still lack a gold reference method [25], and each approach has relative advantages and caveats. In particular, ELISA methods are cheaper, easy, and have the potential for wide diffusion in clinical settings, but lack absolute specificity for hepcidin isoforms because of various degree of antibody cross-reactivity that is hard to eliminate. On the other hand, MS-based methods are costly and require dedicated personnel, but can properly distinguish the isoforms. Indeed, in the only study published so far that directly compared two second-generation hepcidin assays, Kroot et al. showed that the observed differences in absolute concentrations were explained, at least partially, by the isoforms detected by MS, as opposed to “total” hepcidin detected by ELISA [17]. Until now, data on hep-20 in healthy subjects were limited and could only be inferred from small case–control studies. We previously reported by SELDI-TOF-MS that hep-20 was detectable in sera from 35 out of 57 (62%) healthy controls, at variance with CKD patients in whom hep-20 was always detectable [15]. Kroot et al., using WCX-TOF-MS assay, compared data from 23 healthy controls with several groups of patients with heterogeneous disorders of iron metabolism [17]. Hep-20 was not detectable in the few of controls, but was frequently high in several diseases, particularly in CKD. Different percentage in controls with detectable hep-20 between the two studies may be explained by the fact that we used more stringent criteria for “control” definition, excluding subjects with even subclinical iron deficiency and hence virtually eliminating those with low/absent hepcidin production. Anyway, the present study suggests that the relative contribute of hep-20 to “total” serum hepcidin is not negligible at population level. In the total population, mean hep-20 levels were about 14% of corresponding hep-25 levels. On the other hand, when considering only subjects with detectable hep-25 levels (n = 1405), this percentage rose up to 42.1%.

Hepcidin-20: more than simply a “fixed” degradation products?

The presence of circulating truncated hepcidin isoforms raises questions about their origin and biological meaning. Until now, little is known about the processing of the 22-mer, and 20-mer peptides. Biochemical studies from Schranz and co-workers [26] proposed that the two truncated isoforms may result from the sequential action of unknown aminopeptidases on the mature hep-25 peptide. The Authors suggested Dipeptidylpeptidase IV as a strong candidate for the generation of hep-20 from hep-22, based on the presence of a proline at the cleavage site [27]. Whatever the molecular mechanism, whether or not the processing is actively regulated remains elusive. Theoretically, hep-20 may represent either the final inactive product of constitutive (i.e. not regulated) degradation of hep-25, or a possibly functional peptide whose production may be modulated by body's need relative to the need of hep-25. Clues to the latter hypothesis are the recent studies on the potentially relevant antimicrobial properties of hep-20 [10,11], as well as the pilot studies on its increased concentration in several diseases [12,13,15,16]. Although our study cannot inform on the putative hep-20 function, it may point toward a putative active regulation of hep-20 production and/or hep-25 degradation. This is illustrated particularly by analyzing the hep-25:hep-20 ratio. A simple degradation process should result in a relatively stable ratio between the bioactive peptide (hep-25) and its catabolic product (hep-20). On the other hand, as depicted in Fig. 4A and B, the ratio was consistently lower in women during the fertile period than in men of corresponding age, and increased progressively in both sexes with increasing ferritin levels. Accordingly, a multivariate model showed that ferritin was an independent predictor of hep-20 in women with a negative coefficient, and not positive as it would be expected if hep-20 would merely represent a constitutive degradation product of hep-25 (Table 5). Taken together, these observations suggest that in subjects with iron deficiency the few hep-25 produced may be, in addition, efficiently degraded to keep the iron bioactive peptide as low as possible to maximize intestinal iron absorption. On the other hand, in subjects with adequate or high iron status hep-25 degradation may proceed less efficiently to keep a normal iron balance and/or prevent dangerous iron load.

Table 5

Predictors of hep-25:hep-20 ratio in males and females.

	Males		Females
β-coefficient	P	β-coefficient	P
Ferritin (μg/l)	0.504	< 0.001	0.662	< 0.001
Age (years)	− 0.147	0.001	− 0.198	< 0.001

Study limitations

Although fascinating, the hypothesis of an active regulation of hep-20 needs further specific studies, being only indirectly supported by our data. Other limitations of our study are represented by the lack of mechanistic explanation on hep-20 formation/function, as well as the relatively high lower limit of detection of our assay that did not allow proper evaluation of the hep-25:hep-20 ratio in subjects with undetectable levels of both isoforms. Moreover, our data suffer from the lack of standardization of current hepcidin assays [25], and need confirmation once a “gold reference” assay will be established.

Conclusions

Considering the recent discovery of hepcidin, we are likely only at the beginning of a story in which much has to be yet discovered. While in the last decade we have learned much on hepcidin regulation at transcriptional level [1,2], times may be mature for in depth investigations on the post-translational regulation of this small peptide hormone. If confirmed, our data may suggest an active regulation of hep-25 degradation according to body iron need. The protease(s) responsible of hep-25 processing, once identified, might be therapeutic target(s) for the control of iron metabolism.

Conflict of interest disclosure

All the authors declare that they do not have any conflict of interest to disclose regarding this manuscript.

Table 4

Sex specific correlation analysis of hep-25:hep-20 ratio.

	Males		Females
Correlation	P	Correlation	P
Age (years)	− 0.119	0.015	0.136	0.006
BMI	− 0.027	0.588	0.073	0.137
s-iron (μg/dl)	0.098	0.046	0.118	0.016
Transferrin (mg/dl)	− 0.074	0.135	− 0.193	< 0.001
Transferrin saturation %	0.106	0.032	0.172	< 0.001
Ferritina (ng/ml)	0.497	< 0.001	0.564	< 0.001
Hb (g/dl)	− 0.009	0.850	0.134	0.006
CRPa (mg/l)	0.043	0.472	0.063	0.284
Creatininea (μmol/l)	− 0.061	0.234	0.037	0.470

Variables not normally distributed are expressed as geometric means with 95% CIs.