Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2021 Nov 1.
Published in final edited form as: Arterioscler Thromb Vasc Biol. 2020 Aug 20;40(11):2776–2784. doi: 10.1161/ATVBAHA.120.314599

High Mobility Group Box-1 is associated with Obesity, Inflammation and Subclinical Cardiovascular Risk among Young Adults: A Longitudinal Cohort Study

Li Chen 1, Haidong Zhu 1, Shaoyong Su 1, Gregory Harshfield 1, Jennifer Sullivan 2, Clinton Webb 2, James A Blumenthal 3, Xiaoling Wang 1, Ying Huang 1, Frank A Treiber 4,5, Gaston Kapuku 1, Wenjun Li 6, Yanbin Dong 1
PMCID: PMC7578115  NIHMSID: NIHMS1619712  PMID: 32814439

Abstract

Objective:

We aimed to characterize circulating high-mobility group box-1 (HMGB1) levels, one of the better-characterized damage-associated molecular patterns (DAMPs), with respect to age, sex, and race in the general population, and investigate the longitudinal associations of HMGB1 with inflammatory markers, obesity, and preclinical markers of cardiovascular disease.

Approach and Results:

The analyses included 489 participants (50% blacks, aged 24.6±3.3 years at the first visit) with up to four follow-up visits (1149 samples) over a maximum of 8.5 years. Systolic blood pressure (SBP), diastolic blood pressure (DBP), carotid-femoral pulse wave velocity (cf-PWV), and carotid intima-media thickness (cIMT) together with plasma HMGB1, high-sensitivity C-reactive protein (hs-CRP), interferon-gamma (IFNγ), interleukin-6 (IL-6), interleukin-10 (IL-10), and tumor necrosis factor-α (TNF-α) were measured at each visit. At baseline, plasma HMGB1 concentrations were higher in blacks compared to whites (3.86 vs 3.20 ng/ml, p<0.001), and in females compared to males (3.75 vs 3.30 ng/ml, p=0.005). HMGB1 concentrations increased with age (p=0.007), and higher levels of obesity measures (p<0.001). Without adjustment for age, sex, race, and body mass index, HMGB1 concentrations were positively associated with hs-CRP, IL-6, TNF-α, SBP, DBP, and cf-PWV (p<0.05), but not IL-10, IFN-γ or cIMT. After covariate adjustments, the associations of HMGB1 with hs-CRP, and cf-PWV remained statistically significant (p<0.05).

Conclusion:

This study demonstrate the age, sex, and race differences in circulating HMGB1. The increasing circulating concentrations of HMGB1 with age suggest a potential role of HMGB1 in the pathogenesis of chronic low-grade inflammation, obesity, and subclinical CVD risk.

Keywords: High-Mobility Group Box-1, obesity, inflammation, cardiovascular risk, inflammation, biomarkers, obesity, risk factors

Graphical abstract

graphic file with name nihms-1619712-f0004.jpg

Introduction

Damage-associated molecular patterns (DAMPs) are endogenous danger signals released upon cellular stress or tissue injury, which alert the body about danger and induce potent inflammatory responses by activating the innate immune system during non-infectious inflammation.1 High-mobility group box-1 (HMGB1), a 30 kDa nuclear and cytosolic ubiquitous protein, is one of the best-characterized DAMPs.2–6 HMGB1 is actively secreted by innate immune cells and released into the extracellular space, stimulating other cytokines and playing an important role in inflammatory and immune responses.7–11 Indeed, elevated circulating levels of HMGB1 have been found in acute inflammatory conditions, yet HMGB1 has been understudied in low-grade chronic inflammatory conditions with systemic and long-term effects, which are involved in the development of multiple diseases.

Although the etiology of obesity is complex, obesity is associated with low-grade chronic inflammation, where its status is conditioned by the innate immune system activation in adipose tissue that promotes an increase in pro-inflammatory cytokines.12 Studies have established a key role for low-grade systemic inflammation in the early stage of blood pressure (BP) elevation, arterial stiffness, atherosclerosis, and cardiovascular disease (CVD).13–16 Further, inflammatory mechanisms linking obesity to its metabolic and cardiovascular complications are already activated in childhood obesity.17 In a cross-sectional study, HMGB1 levels were higher in obese children (n=60) than their healthy controls (n=40), and circulating HMGB1 levels were associated with body mass index (BMI) and interleukin-6 (IL-6).18 In another cross-sectional study of Chinese adults, circulating HMGB1 levels were higher in participants with type II diabetes (n=76) versus healthy participants (n=79), and HMGB1 levels were also higher in the obese group than in the normal weight group.19 Moreover, HMGB1 levels positively correlated with waist-to-hip ratio (WHR), BP, and IL-6.19 Japanese participants with peripheral artery disease (PAD, n=24) showed higher HMGB1 levels compared to their healthy controls (n=10), and there was a significant difference in HMGB1 levels between the mild PAD (n=12) and severe PAD (n=12), although high-sensitivity C-reactive protein (hs-CRP) demonstrated no significant correlation with HMGB1.5 A recent cross-sectional study from Turkey demonstrated that serum HMGB1 concentrations were significantly increased in participants with coronary artery disease (n=55) compared to healthy participants (n=50).20 Despite numerous small studies suggesting increases in HMGB1 correlate with the low-grade inflammation found in metabolic and CVD, longitudinal studies with larger sample sizes are urgently needed.

The Georgia Stress and Heart (GSH) Study, a prospective multiethnic cohort study (~50% blacks) with repeated physiological and subclinical CVD risk measures over time, provided an excellent opportunity to examine the longitudinal relationships of HMGB1 with obesity and related CVD risk factors. For the first time, the present study evaluated the longitudinal associations of HMGB1 with inflammatory markers, obesity, and preclinical markers of CVD among young adults. Furthermore, to our knowledge this is the first study to describe the variations of circulating HMGB1 levels with respect to age, sex, and race.

Materials and Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Participants

The participants were from the longitudinal GSH study. 21 The inclusion criteria were (1) aged 5 to 16 years in 1989, (2) blacks or whites, (3) normotensive for age- and sex-based on BP screening, and (4) apparently healthy. Participants were classified as black if (1) both parents reported being of African heritage, (2) the parents and the child were born and raised in the US, and (3) parents considered themselves and their child to be black, African American, or Afro-American. Participants were classified as white if (1) both parents reported that they were of European ancestry, (2) the parents and the child were born and raised in the US, and (3) they considered themselves and their child to be white, European American, or Caucasian. All participants were screened for eligibility and recruited using family health history questionnaires obtained from a county-wide (Richmond County, Georgia) public school screening of children in kindergarten through 8th grade whose families volunteered for health prevention research. The original GSH cohort consisting of 740 participants was established in 1989. Since visit 9, blood samples have been collected and properly stored. A total of 489 participants (66% of the original 740 participants) with up to four visits, including 1149 blood samples with valid HMGB1 measurements, one from each person-visit were analyzed for the current study. The annualized attrition rate was <4% due to geographical relocation, official discontinuation of death, and loss of interest. There have been no significant differences in age, ethnicity and sex distributions between dropouts and the participants who remained in the study. The Institutional Review Board at the Medical College of Georgia gave approval for the study. Informed consent was provided by all participants, or by parents if participants were less than 18 years.

Anthropometry measurements

Height was measured to the nearest 0.1 cm by a wall-mounted stadiometer (Tanita Corporation of American, Arlington Heights, IL); weight was measured to the nearest 0.1 kg by a calibrated electronic scale with the participants not wearing shoes and in light clothing (model CN2OL; Cardinal Detecto, Webb City, MO). BMI was computed as weight (in kg) per square of height (in m2). Overweight was defined as BMI ≥ 25 kg/m2, and obese was defined as ≥ 30 kg/m2. Waist circumference was measured at the center of the umbilicus. Hip circumference was measured at the widest part of the hips. WHR was computed as waist circumference divided by hip circumference.

Preclinical CVD phenotypic measurements

BP was recorded by Dinamap (model 1864 SX). After attachment of an appropriately sized BP cuff to the right arm, the participant was placed in a supine position on a medical table with head propped on a pillow and then given instructions to relax as completely as possible for 15 minutes. BP measurements were taken from the Dinamap at the end of the 11th, 13th, and 15th minutes. As recommended by the National Health and Nutrition Examination Survey procedures for BP measurement, the average of the last two readings was used to represent resting systolic blood pressure (SBP) and diastolic blood pressure (DBP).

Arterial stiffness was measured among 253 participants on visit 15 using the SphygmoCor Pulse Wave Analysis System (SphygmoCor, Sydney, Australia). Carotid-femoral pulse wave velocity (cf-PWV) was automatically calculated from measurements of pulse transit time and the distance traveled by the pulse between the distal and proximal measurement sites. Carotid artery IMT was available among 299 participants on visit 15 and 271 participants on visit 16. A total of 348 participants were measured for IMT for both visits. Hewlett-Packard Sonos 5500 (Andover, MA) equipped with a 7.5 MHz linear array probe was used to measure the common carotid artery IMT. There were 570 valid observations for IMT measurement. Left and right common carotid, carotid bulb, internal carotid, and external carotid were first visualized in transverse then in longitudinal planes. Measurements were made at a point 2 cm proximal to the bifurcation on both near and far wall that showed the intima-media boundaries most clearly. IMT was derived from a computer program Vascular Tool (Medical Imaging Application, Iowa City, Iowa). Common carotid’s IMT was measured as the distance from leading edge of first echogenic line to that of the second echogenic line. Ten frames of common carotid artery were analyzed by one experienced sonographer. The mean carotid IMT for far wall was used in this analysis.

Plasma inflammatory markers measurement

hs-CRP, interferon-γ (IFN-γ), IL-10, IL-6, and tumor necrosis factor-α (TNF-α) were measured using Simple Plex assay, which was based on microfluidics and glass nanoreactor (GNR) technology (Simple Plex, Protein Simple corp, San Jose, California).22 The intra-assay and inter-assay CVs were <7.0%.

HMGB1 measurement

Plasma HMGB1 concentrations were quantified in 1149 samples by ELISA using the kit from IBL International according to the manufacturer’s instruction. High sensitive range of standard curve is from 0.2 to 10 ng/mL, the intra-assay and inter-assay CVs are 5.8% and 7.6%, respectively. Due to the skewed and tailed distribution, HMGB1 was log-transformed in regression models. Residuals of log-transformed HMGB1 from linear regressions against test plates were taken as adjusted values of HMGB1 to account for the plate bias.

Statistical analysis

Descriptive statistics of participant characteristics were summarized by visit and presented in Table 1. Bar charts in Figure 1 illustrated the distributions of baseline HMGB1 by sex and race. Longitudinal associations of HMGB1 with measures of obesity (BMI, waist circumference, and WHR) were evaluated using mixed-effects models adjusting for age, sex and race. The longitudinal associations of HMGB1 with inflammatory markers (hs-CRP, IL-6, IFN-γ, IL-10, and TNF-α) and CVD phenotypes (SBP, DBP, cf-PWV, and cIMT) were also investigated using mixed-effects models, with and without adjustments for age, race, sex, and BMI. Autoregressive structure of order 1 of the within-group errors was used to account for serial correlations among observations on the same participant. cf-PWV was analyzed cross-sectionally using multiple linear regression model since it was measured only on visit 15. Similar statistical analyses were carried out to test the interactions between HMGB1 and age, sex, or race with regard to the inflammatory markers and CVD phenotypes, and subgroup analyses were performed by age, sex, and race groups. HMGB1, cf-PWV, and all five inflammatory markers were log-transformed in the regression models. No potential non-linear associations were identified using scatter plot. A p-value < 0.05 was considered statistically significant. All statistical analyses were performed using Stata version 12.0 (College Station, Texas 77845 USA).

Table 1.

General characteristics in each visit

Characteristics 1st visit 2nd visit 3rd visit 4th visit
N 489 372 223 65
Age, years 24.6 ± 3.3 27.1 ± 3.5 29.4 ± 3.3 30.4 ± 3.4
Female, N (%) 249 (51) 193 (52) 115 (52) 31 (48)
Black, N (%) 246 (50) 194 (52) 123 (55) 36 (55)
BMI, kg/m2 29.1 ± 7.9 29.7 ± 8.1 30.0 ± 8.3 31.2 ± 9.5
Waist circumference, inch 36.4 ± 6.8 37.2 ± 7.0 37.7 ± 6.7 39.0 ± 7.3
WHR 0.84 ± 0.07 0.85 ± 0.08 0.86 ± 0.07 0.87 ± 0.07
HMGB1, ng/ml 3.5 ± 1.8 3.8 ± 1.8 4.5 ± 1.8 4.7 ± 1.5
hs-CRP, mg/L 3.3 ± 6.2 3.4 ± 5.5 3.4 ± 6.5 3.7 ± 4.9
IFN-γ, pg/mL 0.8 ± 1.0 0.8 ± 0.8 0.8 ± 0.9 0.9 ± 1.4
IL-10, pg/mL 2.4 ± 1.2 2.4 ± 1.0 2.3 ± 0.9 2.4 ± 1.1
IL-6, pg/mL 3.4 ± 2.7 3.2 ± 2.3 3.4 ± 3.1 4.6 ± 3.8
TNF-α, pg/mL 6.2 ± 1.7 6.1 ± 1.8 6.2 ± 1.8 6.2 ± 2.4
SBP, mmHg 114.5 ± 13.0 115.1 ± 13.3 115.1 ± 13.3 119.1 ± 13.7
DBP, mmHg 64.6 ± 8.1 66.9 ± 8.5 69.0 ± 9.4 70.1 ± 9.6
cf-PWV, m/s 5.9 ± 1.2 6.2 ± 1.4 6.7 ± 1.4 6.9 ± 1.1
cIMT, mm 0.5 ± 0.1 0.5 ± 0.1 0.5 ± 0.1 N/A
Open in a new tab

Note: The observations included in this analysis were reordered according the order of the appearance of each participants with valid HMGB1 measurement. Abbreviations: HMGB1, high-mobility group box-1; BMI, body mass index; WHR, waist-to-hip ratio; hs-CRP, high-sensitivity C-reactive protein; SBP, systolic blood pressure; DBP, diastolic blood pressure; cf-PWV, carotid-femoral pulse wave velocity; cIMT, carotid intima-media thickness.

Figure 1. The baseline differences in HMGB1 concentrations between sex and race.

Figure 1.

Open in a new tab

The bar charts present the baseline differences in HMGB1 regarding sex (a), race (b). HMGB1, high-mobility group box-1.

Results

General characteristics of the participants

The observations included in this analysis were reordered according to the order of the appearance of each participant with valid HMGB1 measurement. The 1st visit was the first observation of each participants that was included in this analysis. The 489 participants averaged 24.6 ± 3.3 years in age at the first visit that was included in this study (Table 1). Among those, 249 (51%) were female, and 246 (50%) were black.

Variations in plasma HMGB1 by participant characteristics

Using the first visit as the baseline, higher plasma HMGB1 concentrations were associated with older age (p=0.007), and higher BMI, waist circumference, and WHR (p<0.001). Figure 1 shows the differences in baseline HMGB1 concentrations by sex and race. Plasma HMGB1 concentrations were higher among females than males, and among blacks than whites (p<0.05).

Longitudinal associations of HMGB1 with obesity measures

Longitudinally, plasma concentrations of HMGB1 were positively associated with BMI, WHR, and waist circumference without (p<0.001) or with adjustment for age, sex, and race (p<0.001). The participants were categorized into groups according to the baseline measurements of BMI, waist circumference, and WHR. Among the 489 participants, 172 (35%) were lean, 156 (32%) were overweight, and 161 (33%) were obese. Compared to the lean participants (BMI<25 kg/m2), higher HMGB1 concentrations were observed among the obese participants (BMI>30 kg/m2) (p<0.001), but not among the overweight participants (BMI between 25 to 30 kg/m2) (p=0.078). Compared to those in the lower tertile, participants in the middle and upper tertiles of WHR had higher HMGB1 (p=0.010 and p<0.001, respectively), and participants in the middle and upper tertiles of waist circumference had higher HMGB1 (p<0.001 for both) (Figure 2). We further tested the associations between HMGB1 and obesity with additional adjustment of the inflammatory markers. The associations of HMGB1 with BMI and waist circumference remained statistically significant after further adjustment of IFN-γ, IL-10, IL-6, and TNF-α (p<0.05). HMGB1 was no longer associated with WHR with adjustment of any of the inflammatory markers (p>0.05). HMGB1 was no longer associated with any of the obesity measures when adjusted for hs-CRP (p>0.05).

Figure 2. Longitudinal associations between obesity measures and HMGB1.

Figure 2.

Open in a new tab

Marginal plots between obesity measures and HMGB1 are based on mixed-effects models adjusting for age, sex, and race. (a) Obesity; (b) tertiles of waist circumference; (c) tertiles of waist-to-hip ratio. p-values labeled in each plot are the significant levels of the corresponding groups compared to the lean group (a) or to the low tertile (b and c).

Longitudinal associations of HMGB1 with inflammatory markers

As shown in Table 2, plasma concentrations of HMGB1 were positively associated with hs-CRP, IL-6 and TNF-α without (p<0.01) or with adjustment for age, sex, and race (p<0.01). HMGB1 remained significantly associated with hs-CRP with further adjustment of BMI (p<0.001). In the fully adjusted model, a 10% increase in HMGB1 was associated with approximately 4% increase in hs-CRP. The associations of HMGB1 with IL-6 and TNF-α were not significant with further adjustment of BMI (p>0.05). The associations of HMGB1 with IFN-γ and IL-10 were not significant (p>0.05). In addition, we tested whether the associations of HMGB1 with the inflammatory markers were modified by age, sex, or race. There were significant modifications of HMGB1 effects on hs-CRP by age (β = −0.04, p = 0.009), on IFN-γ (β = −0.16, p = 0.038) and TNF-α by sex (β = −0.07, p = 0.005). No other significant interactions of HMGB1 by age, sex, or race was detected. Figure 3 shows age, sex and race differences in the associations of HMGB1 with hs-CRP, IFN-γ, and TNF-α. The increase of hs-CRP associated with HMGB1 was significant in each of the subgroups (p<0.05). HMGB1 was associated with increased IFN-γ among males, but not statistically significant (β = 0.10, p = 0.104). HMGB1 was significantly associated with increased TNF-α among the males (β = 0.05, p = 0.008), not the females (β = −0.00, p = 0.993).

Table 2.

The longitudinal associations of HMGB1 with inflammatory markers

hs-CRP (N=1129) IFN-γ (N=888) IL-10 (N=1122) IL-6 (N=643) TNF-α (N=1134)
β (SE) p β (SE) p β (SE) p β (SE) p β (SE) p
Model 1 0.67 (0.07) <0.001 0.04 (0.04) 0.308 0.02 (0.02) 0.348 0.15 (0.04) <0.001 0.03 (0.01) 0.006
Model 2 0.57 (0.07) <0.001 0.05 (0.04) 0.178 0.03 (0.02) 0.205 0.12 (0.04) 0.006 0.03 (0.01) 0.009
Model 3 0.42 (0.06) <0.001 0.04 (0.04) 0.331 0.01 (0.02) 0.658 0.06 (0.04) 0.139 0.02 (0.01) 0.147
Open in a new tab

Note: The numbers of observations were smaller than total sample size (1149 observations of 489 participants) due to missing values. Two-level mixed-effects models were used. Autoregressive structure of order 1 of the within-group errors was assumed to account for successive observations with the groups. Model 1 was unadjusted; model 2 was adjusted for age, race, and sex; model 3 was adjusted for body mass index in addition to model 2. There were significant interaction of age and HMGB1 on hs-CRP (β = −0.04, p = 0.009), of sex and HMGB1 on IFN-γ (β = −0.16, p = 0.038), and of sex and HMGB1 on TNF-α (β = −0.07, p = 0.005). No other significant interaction of age, sex, or race on the association between HMGB1 and inflammatory markers was detected. HMGB1 and all five inflammatory markers were log-transformed. Abbreviations: HMGB1, high-mobility group box-1; hs-CRP, high-sensitivity C-reactive protein; IFN-γ, interferon-γ; IL-10, interleukin-10; IL-6, interleukin-6; and TNF-α, tumor necrosis factor-α.

Figure 3. Subgroup analyses of the associations of HMGB1 with hs-CRP, IFN-γ, and TNF-α.

Figure 3.

Open in a new tab

Two-level mixed-effects models were used. Autoregressive structure of order 1 of the within-group errors was assumed to account for successive observations with the groups. Adjusted covariates included age, race, sex, and body mass index when appropriate. HMGB1, hs-CRP, IFN-γ, and TNF-α were log-transformed. Abbreviations: hs-CRP, C-reactive protein; IFN-γ, interferon-γ; and TNF-α, tumor necrosis factor-α.

Longitudinal associations of HMGB1 with preclinical CVD phenotypes

Table 3 presents the results of two-level mixed-effects models associating circulating HMGB1 levels with cardiovascular phenotypes. Without covariate adjustment, HMGB1 was associated with elevated SBP, DBP, and cf-PWV (p<0.05), but not cIMT. After adjustment for age, sex, race, and BMI, the association between HMGB1 and cf-PWV remained statistically significant (p=0.005). In the fully adjusted model, a 100% increase in HMGB1 was associated with approximately 7% increase in cf-PWV. However, the associations of HMGB1 with SBP or DBP were no longer significant (p>0.05), after the adjustment of the above confounding factors. No significant age, sex or race effect modification on the association between HMGB1 and cardiovascular phenotypes was detected. In addition, the associations of inflammatory markers with preclinical CVD phenotypes were studied with the same regression models that were done for HMGB1. None of the inflammatory markers was significantly associated with preclinical CVD phenotypes (p>0.05). Moreover, the positive association between HMGB1 and cf-PWV remained significant with further adjustment of each of the inflammatory makers (p<0.05).

Table 3.

The longitudinal associations of HMGB1 with cardiovascular phenotypes

SBP (N=1145) DBP (N=1145) cf-PWV (N=253) cIMT (N=570)
β (SE) p β (SE) p β (SE) p β (SE) p
Model 1 1.64 (0.66) 0.012 2.28 (0.46) <0.001 0.12 (0.03) <0.001 0.00 (0.01) 0.807
Model 2 0.60 (0.67) 0.371 0.34 (0.44) 0.438 0.08 (0.02) 0.001 −0.00 (0.01) 0.554
Model 3 −0.24 (0.65) 0.710 0.49 (0.44) 0.261 0.07 (0.02) 0.005 −0.01 (0.01) 0.101
Open in a new tab

Note: The numbers of observations were smaller than total sample size (1149 observations of 489 participants) due to missing values. Two-level mixed-effects models were used except for cf-PWV. Autoregressive structure of order 1 of the within-group errors was assumed to account for successive observations with the groups. Multiple linear regression model was used for cf-PWV. Model 1 was unadjusted; model 2 was adjusted for age, race, and sex; model 3 was adjusted for body mass index in addition to model 2. No significant interaction of age, sex, or race on the association between HMGB1 and cardiovascular phenotypes was detected. HMGB1 and cf-PWV were log-transformed. Abbreviations: HMGB1, high-mobility group box-1; SBP, systolic blood pressure; DBP, diastolic blood pressure; cf-PWV, carotid-femoral pulse wave velocity; cIMT, carotid intima-media thickness.

Discussion

Our study characterizes circulating HMGB1 levels in a bi-ethnic population by age, sex, and race. Our study demonstrates that circulating HMGB1 concentrations are higher in blacks than whites, and in females than males, and HMGB1 increases with age. This is the first report of the longitudinal associations of the circulating HMGB1 concentrations with obesity status, inflammation, and preclinical markers of CVD. The associations of HMGB1 with hs-CRP and cf-PWV are consistent across different age, sex, and race subgroups.

The plasma concentrations of HMGB1 increased with age in our cohort. Similar longitudinal patterns were seen in animal models. HMGB1 protein and mRNA were elevated in the hippocampus and cerebrospinal fluid of aged rats, and blocking the actions of HMGB1 might reduce age-associated inflammatory priming.11 In a cross-sectional analysis, human serum concentrations of HMGB1 were found to be significantly higher in older (≥70 years old, n=18) than young adults (18–30 years old, n=19).23 However, in another cross-sectional analysis, the serum levels of HMGB1 were lower in the aged group (>65 years, n=30) compared with those in the young group (~40 years, n=30).24 The discrepancies could be attributed to various factors such as the sample sizes, study designs (cross-sectional vs. longitudinal), age range, race, lifestyles and geographic regions. Nonetheless, given the plausible underlying mechanisms, the relationship between HMGB1 and aging warrants further investigations.

Notable differences in HMGB1 concentrations were observed between black and white races and between sexes in our cohort. Blacks had higher concentrations than whites. This is consistent with previous findings that blacks tend to have greater levels of plasma inflammatory markers such as hs-CRP and IL-6.25–27 Females had significantly higher HMGB1 than males, which could be related to females’ higher percent body fat. However, the role of sex in HMGB1 biology deserves more attention.

In the present study, the plasma concentration of HMGB1 was significantly correlated with obesity over time. This is biologically plausible and supported by various animal studies, including our own.11 In animal studies, it has been reported that serum levels of HMGB1 were increased in rats fed with a high-fat diet for 12 weeks.28 Adipose tissue secrets HMGB1, which was expressed 2-fold more in adipose tissue from obese compared to normal-weight individuals.29 Moreover, obesity-associated adipocyte death leads to the release of the intracellular contents such as HMGB1 into the extracellular milieu.30 HMGB1 activates resident immune cells in the adipose tissue once released from necrotic adipocytes. Activated resident immune cells then secrete additional HMGB1, which then activates and recruits additional immune cells.30 Therefore, HMGB1 plays a critical role in the initiation and maintenance of chronic inflammatory state in adipose tissue of obese participants.30 HMGB1 binding to toll-like receptor (TLR), specifically TLR2 and TLR4, leads to the activation of nuclear factor κB, mitogen-activated protein kinases and Jun N-terminal kinase, thereafter leading to immunoinflammatory response with the production of cytokines and other inflammatory molecules.31 HMGB1 is also able to activate receptor for advanced glycation end products pathway, whose products lead to the activation of several intracellular oxidative stress-depending signals including nuclear factor κB, PI3K/Akt, mitogen-activated protein kinases, and then upregulate several pro-inflammatory genes.32–34 In this study, we observed longitudinal associations of HMGB1 with hs-CRP, which supports these mechanistic findings in vitro and animal models.

The demonstrated associations of HMGB1 with the above inflammatory markers are of clinical importance because chronic inflammation plays a key role in the early stage of hypertension and CVD.35 In our cohort, higher HMGB1 was associated with elevated SBP, DBP, and cf-PWV, although the correlations between HMGB1 and BP were no longer statistically significant after the adjustments of age, sex, race, and BMI. Moreover, our data show that HMGB1 could be a superior biomarker of cf-PWV to the inflammatory factors in the young adult population. In our study, HMGB1 seems to outperform hs-CRP in predicting arterial stiffness (PWV), or that although HMGB1 and hs-CRP are closely correlated, HMGB1 may still comprises independent effect on CVD development. Given our sample size, our data had the power of 97.80% to detect a correlation between HMGB1 and cf-PWV. There are several potential mechanisms linking HMGB1 to cardiovascular phenotypes. HMGB1 may affect cardiovascular health through TLR mediated inflammation. An animal study showed that hypertensive rats exhibited higher TLR4 in the paraventricular nucleus, which is a major receptor of HMGB1.36 TLR4 inhibition decreased mean arterial pressure in rats.36 In a human study, TLR polymorphism was associated with SBP and pulse pressure among patients with coronary artery disease.37 Another study in vitro proposed that extracellular HMGB1 could enhance mechanical stress‑induced cardiac hypertrophy via the RAGE/ERK1/2 signaling pathway.38 HMGB1 may also play a role in Ang II-induced vascular smooth muscle cell phenotypic transformation.39 Sustained release of HMGB1 in response to persistent hypertensive stimuli promotes vascular remodeling, endothelial dysfunction, cardiac hypertrophy, and increases in BP in models of pulmonary hypertension.40, 41 Our longitudinal data highlight that HMGB1 might play a critical role in the inflammatory response in the pathogenesis of CVD in humans. The finding that HMGB1 seemed to be associated with IFN-γ and TNF-α among the male warrants further investigation.

A major strength in the present study is our longitudinal cohort with the multiple HMGB1 measurements over time; with an equal distribution of blacks and whites, males and females; and with repeated detailed obesity and preclinical CVD measures. However, the study also has several limitations, including: 1) limited age range and relatively young age of our sample; 2) cf-PWV and cIMT were measured in a subset of the participants; 3) its observational nature that may not support causal inferences; and 4) we did not collect dietary intake information, which could be a confounding factor given that diets may differentially affect intestinal permeability and thus, low grade chronic inflammation. 5) We only collected blood samples from the GSH cohort, and measured circulating HMGB1 in this study, which was not able to identify the potential source of HMGB1. Future studies examining the HMGB1 from peripheral blood mononuclear cells or adipose tissue are warranted. Future interventional studies are also needed to validate our results.

In conclusion, for the first time, we report the longitudinal associations of circulating HMGB1 levels with obesity, proinflammatory cytokines and preclinical CVD phenotypes in a cohort of white and black races. Race and sex differences in some of these associations are evident. Further investigations of the variations in and biological mechanisms of HMGB1 in free-living human population may lead to better understanding of the role of HMGB1, and may confirm the value of circulating HMGB1 as a clinical biomarker and the development of anti-HMGB1 antibodies or antagonists in therapeutics for the targeted prevention, management, and treatment of chronic inflammation, obesity and CVD.

Highlights.

  • To our knowledge, this is the first longitudinal study to comprehensively examine the associations of human HMGB1 with several inflammatory markers, obesity, and subclinical cardiovascular risk;

  • We observe race and sex differences in plasma HMGB1;

  • Circulating HMGB1 increases with age;

  • Circulating HMGB1 is associated with obesity, inflammation and subclinical CVD risk.

Acknowledgments

Sources of Funding

This study was supported by the National Institute of Health, 1R01HL136630-01A1.

Abbreviations

DAMPS

damage-associated molecular patterns

HMGB1

high-mobility group box-1

BP

blood pressure

CVD

cardiovascular disease

PAD

peripheral artery disease

BMI

body mass index

GSH

Georgia Stress and Heart

WHR

waist-to-hip ratio

hs-CRP

high-sensitivity C-reactive protein

IFN-γ

interferon-γ

IL-10

interleukin-10

IL-6

interleukin-6

TNF-α

tumor necrosis factor-α.

SBP

systolic blood pressure

DBP

diastolic blood pressure

cf-PWV

carotid-femoral pulse wave velocity

cIMT

carotid intima-media thickness

TLR

toll-like receptor

Footnotes

Disclosures:

None.

References

  • 1.Venereau E, Ceriotti C, Bianchi ME. Damps from cell death to new life. Front Immunol. 2015;6:422. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Tsukagawa T, Katsumata R, Fujita M, Yasui K, Akhoon C, Ono K, Dohi K, Aruga T. Elevated serum high-mobility group box-1 protein level is associated with poor functional outcome in ischemic stroke. J Stroke Cerebrovasc Dis. 2017;26:2404–2411 [DOI] [PubMed] [Google Scholar]
  • 3.Huang JM, Hu J, Chen N, Hu ML. Relationship between plasma high-mobility group box-1 levels and clinical outcomes of ischemic stroke. J Crit Care. 2013;28:792–797 [DOI] [PubMed] [Google Scholar]
  • 4.Chen Q, Wang ZY, Chen LY, Hu HY. Roles of high mobility group box 1 in cardiovascular calcification. Cell Physiol Biochem. 2017;42:427–440 [DOI] [PubMed] [Google Scholar]
  • 5.Oozawa S, Sano S, Nishibori M. Usefulness of high mobility group box 1 protein as a plasma biomarker in patient with peripheral artery disease. Acta Med Okayama. 2014;68:157–162 [DOI] [PubMed] [Google Scholar]
  • 6.Zhang L, Liu M, Jiang H, Yu Y, Yu P, Tong R, Wu J, Zhang S, Yao K, Zou Y, Ge J. Extracellular high-mobility group box 1 mediates pressure overload-induced cardiac hypertrophy and heart failure. J Cell Mol Med. 2016;20:459–470 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Klune JR, Billiar TR, Tsung A. Hmgb1 preconditioning: Therapeutic application for a danger signal? J Leukoc Biol. 2008;83:558–563 [DOI] [PubMed] [Google Scholar]
  • 8.Klune JR, Dhupar R, Cardinal J, Billiar TR, Tsung A. Hmgb1: Endogenous danger signaling. Mol Med. 2008;14:476–484 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Lu B, Wang C, Wang M, Li W, Chen F, Tracey KJ, Wang H. Molecular mechanism and therapeutic modulation of high mobility group box 1 release and action: An updated review. Expert Rev Clin Immunol. 2014;10:713–727 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Kang R, Chen R, Zhang Q, Hou W, Wu S, Cao L, Huang J, Yu Y, Fan XG, Yan Z, Sun X, Wang H, Wang Q, Tsung A, Billiar TR, Zeh HJ 3rd, Lotze MT, Tang D. Hmgb1 in health and disease. Mol Aspects Med. 2014 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Fonken LK, Frank MG, Kitt MM, D’Angelo HM, Norden DM, Weber MD, Barrientos RM, Godbout JP, Watkins LR, Maier SF. The alarmin hmgb1 mediates age-induced neuroinflammatory priming. The Journal of neuroscience : the official journal of the Society for Neuroscience. 2016;36:7946–7956 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Lumeng CN. Innate immune activation in obesity. Molecular aspects of medicine. 2013;34:12–29 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Lu X, Crowley SD. Inflammation in salt-sensitive hypertension and renal damage. Curr Hypertens Rep. 2018;20:103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Alexy T, Pais E, Wenby RB, Mack WJ, Hodis HN, Kono N, Wang J, Baskurt OK, Fisher TC, Meiselman HJ. Abnormal blood rheology and chronic low grade inflammation: Possible risk factors for accelerated atherosclerosis and coronary artery disease in lewis negative subjects. Atherosclerosis. 2015;239:248–251 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Li X, Deng YP, Yang M, Wu YW, Sun SX, Sun JZ. Low-grade inflammation and increased arterial stiffness in chinese youth and adolescents with newly-diagnosed type 2 diabetes mellitus. J Clin Res Pediatr Endocrinol. 2015;7:268–273 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Greve SV, Blicher MK, Kruger R, Sehestedt T, Gram-Kampmann E, Rasmussen S, Vishram JKK, Boutouyrie P, Laurent S, Olsen MH. Elevated estimated arterial age is associated with metabolic syndrome and low-grade inflammation. Journal of Hypertension. 2016;34:2410–2417 [DOI] [PubMed] [Google Scholar]
  • 17.Kelishadi R, Roufarshbaf M, Soheili S, Payghambarzadeh F, Masjedi M. Association of childhood obesity and the immune system: A systematic review of reviews. Child Obes. 2017;13:332–346 [DOI] [PubMed] [Google Scholar]
  • 18.Arrigo T, Chirico V, Salpietro V, Munafo C, Ferrau V, Gitto E, Lacquaniti A, Salpietro C. High-mobility group protein b1: A new biomarker of metabolic syndrome in obese children. European journal of endocrinology. 2013;168:631–638 [DOI] [PubMed] [Google Scholar]
  • 19.Wang H, Qu H, Deng H. Plasma hmgb-1 levels in subjects with obesity and type 2 diabetes: A cross-sectional study in china. PloS one. 2015;10:e0136564 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Benlier N, Erdogan MB, Kecioglu S, Orhan N, Cicek H. Association of high mobility group box 1 protein with coronary artery disease. Asian Cardiovasc Thorac Ann. 2019;27:251–255 [DOI] [PubMed] [Google Scholar]
  • 21.Hao G, Wang X, Treiber FA, Harshfield G, Kapuku G, Su S. Blood pressure trajectories from childhood to young adulthood associated with cardiovascular risk: Results from the 23-year longitudinal georgia stress and heart study. Hypertension. 2017;69:435–442 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Aldo P, Marusov G, Svancara D, David J, Mor G. Simple plex™: A novel multi-analyte, automated microfluidic immunoassay platform for the detection of human and mouse cytokines and chemokines. American Journal of Reproductive Immunology. 2016;75:678–693 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Qi YF, Goel R, Kim S, Richards EM, Carter CS, Pepine CJ, Raizada MK, Buford TW. Intestinal permeability biomarker zonulin is elevated in healthy aging. Journal of the American Medical Directors Association. 2017;18 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Fu GX, Chen AF, Zhong Y, Zhao J, Gu YJ. Decreased serum level of hmgb1 and myd88 during human aging progress in healthy individuals. Aging Clin Exp Res. 2016;28:175–180 [DOI] [PubMed] [Google Scholar]
  • 25.Stepanikova I, Bateman LB, Oates GR. Systemic inflammation in midlife: Race, socioeconomic status, and perceived discrimination. American journal of preventive medicine. 2017;52:S63–s76 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Starzak DE, Semple SJ, Smith LL, McKune AJ. Differing cytokine responses by ethnic groups to a bout of exercise-induced muscle damage: A preliminary report. The Journal of sports medicine and physical fitness. 2016;56:665–677 [PubMed] [Google Scholar]
  • 27.Paalani M, Lee JW, Haddad E, Tonstad S. Determinants of inflammatory markers in a bi-ethnic population. Ethnicity & disease. 2011;21:142–149 [PMC free article] [PubMed] [Google Scholar]
  • 28.Zhang W, Wang LW, Wang LK, Li X, Zhang H, Luo LP, Song JC, Gong ZJ. Betaine protects against high-fat-diet-induced liver injury by inhibition of high-mobility group box 1 and toll-like receptor 4 expression in rats. Dig Dis Sci. 2013;58:3198–3206 [DOI] [PubMed] [Google Scholar]
  • 29.Gunasekaran MK, Viranaicken W, Girard AC, Festy F, Cesari M, Roche R, Hoareau L. Inflammation triggers high mobility group box 1 (hmgb1) secretion in adipose tissue, a potential link to obesity. Cytokine. 2013;64:103–111 [DOI] [PubMed] [Google Scholar]
  • 30.Zhang J, Zhang L, Zhang S, Yu Q, Xiong F, Huang K, Wang CY, Yang P. Hmgb1, an innate alarmin, plays a critical role in chronic inflammation of adipose tissue in obesity. Mol Cell Endocrinol. 2017;454:103–111 [DOI] [PubMed] [Google Scholar]
  • 31.Cirillo P, Giallauria F, Di Palma V, Maresca F, Ziviello F, Bevilacqua M, Vigorito C, Trimarco B. Cardiovascular disease and high-mobility group box 1--is a new inflammatory killer in town? Angiology. 2013;64:343–355 [DOI] [PubMed] [Google Scholar]
  • 32.Guimaraes EL, Empsen C, Geerts A, van Grunsven LA. Advanced glycation end products induce production of reactive oxygen species via the activation of nadph oxidase in murine hepatic stellate cells. J Hepatol. 2010;52:389–397 [DOI] [PubMed] [Google Scholar]
  • 33.Fehrenbach H, Weiskirchen R, Kasper M, Gressner AM. Up-regulated expression of the receptor for advanced glycation end products in cultured rat hepatic stellate cells during transdifferentiation to myofibroblasts. Hepatology. 2001;34:943–952 [DOI] [PubMed] [Google Scholar]
  • 34.Paull TT, Haykinson MJ, Johnson RC. The nonspecific DNA-binding and -bending proteins hmg1 and hmg2 promote the assembly of complex nucleoprotein structures. Genes Dev. 1993;7:1521–1534 [DOI] [PubMed] [Google Scholar]
  • 35.Agita A, Alsagaff MT. Inflammation, immunity, and hypertension. Acta medica Indonesiana. 2017;49:158–165 [PubMed] [Google Scholar]
  • 36.Dange RB, Agarwal D, Teruyama R, Francis J. Toll-like receptor 4 inhibition within the paraventricular nucleus attenuates blood pressure and inflammatory response in a genetic model of hypertension. J Neuroinflammation. 2015;12:31. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Schneider S, Koch W, Hoppmann P, Ubrich R, Kemmner S, Steinlechner E, Heemann U, Laugwitz KL, Kastrati A, Baumann M. Association of toll-like receptor 4 polymorphism with age-dependent systolic blood pressure increase in patients with coronary artery disease. Immunity & ageing : I & A. 2015;12:4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Zhang L, Yang X, Jiang GL, Yu Y, Wu H, Su YG, Sun AJ, Zou YZ, Jiang H, Ge JB. Hmgb1 enhances mechanical stress-induced cardiomyocyte hypertrophy in vitro via the rage/erk1/2 signaling pathway. International Journal of Molecular Medicine. 2019;44:885–892 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Li FJ, Zhang CL, Luo XJ, Peng J, Yang TL. Involvement of the mir-181b-5p/hmgb1 pathway in ang ii-induced phenotypic transformation of smooth muscle cells in hypertension. Aging Dis. 2019;10:231–248 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Bauer EM, Shapiro R, Zheng H, Ahmad F, Ishizawar D, Comhair SA, Erzurum SC, Billiar TR, Bauer PM. High mobility group box 1 contributes to the pathogenesis of experimental pulmonary hypertension via activation of toll-like receptor 4. Mol Med. 2013;18:1509–1518 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Sadamura-Takenaka Y, Ito T, Noma S, Oyama Y, Yamada S, Kawahara K, Inoue H, Maruyama I. Hmgb1 promotes the development of pulmonary arterial hypertension in rats. PloS one. 2014;9:e102482 [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES