Involvement of the LPS-LPB-CD14-MD2-TLR4 inflammation pathway in HIV-1/HAART-associated lipodystrophy syndrome (HALS)

Abstract

Objectives

A relationship between obesity and intestinal bacterial translocation has been reported. Very little information is available with respect to the involvement of the bacterial translocation mechanistic pathway in HIV-1/highly active antiretroviral therapy (HAART)-associated lipodystrophy syndrome (HALS). We determined whether lipopolysaccharide (LPS)-binding protein (LBP), cluster of differentiation 14 (CD14), myeloid differentiation protein 2 (MD2) and toll-like receptor 4 (TLR4) single-nucleotide polymorphisms and LPS, LBP and soluble CD14 (sCD14) plasma levels are involved in HALS.

Patients and methods

This cross-sectional multicentre study involved 558 treated HIV-1-infected patients, 240 with overt HALS and 318 without HALS. Anthropometric, clinical, immunovirological and metabolic variables were determined. Polymorphisms were assessed by genotyping. Plasma levels were determined by ELISA in 163 patients (81 with HALS and 82 without HALS) whose stored plasma samples were available. Student's t-test, one-way ANOVA, two-way repeated measures ANOVA, the χ² test and Pearson and Spearman correlation analyses were carried out for statistical analysis.

Results

LBP rs2232582 T→C polymorphism was significantly associated with HALS (P = 0.01 and P = 0.048 for genotype and allele analyses, respectively). Plasma levels of LPS (P = 0.009) and LBP (P < 0.001) were significantly higher and sCD14 significantly lower (P < 0.001) in patients with HALS compared with subjects without HALS. LPS levels were independently predicted by triglycerides (P < 0.001) and hepatitis C virus (P = 0.038), LBP levels by HALS (P < 0.001) and sCD14 levels by age (P = 0.008), current HIV-1 viral load (P = 0.001) and protease inhibitor use (P = 0.018).

Conclusions

HALS is associated with LBP polymorphism and with higher bacterial translocation.

genetics, microbial translocation, inflammation

Introduction

After years of intensive research, the pathogenesis of HIV-1/highly active antiretroviral therapy (HAART)-associated lipodystrophy syndrome (HALS) is not completely known. Studies have sought candidates for the factors involved, among them being the type of antiretroviral drugs used,¹ disturbances in cytokine and adipokine synthesis, mainly produced by adipose tissue,^2,3 inflammation⁴ and immune activation.⁵ Noteworthy is that some morphological traits and pathological mechanisms in HALS are reminiscent of obesity.⁶ Recent information suggests that there is a relationship between immune activation and obesity or metabolic disturbances in the general population.^7,8 In this way, a close positive association between the circulating lipopolysaccharide (LPS)-binding protein (LBP) concentration and obesity-associated metabolic perturbations has been reported.⁹ HIV-1-infected subjects have chronic immune activation,¹⁰ which is tempered but not fully reverted with successful HAART.¹¹ An increase in circulating microbial products due to increased intestinal translocation, such as LPS, has been proposed as a potential stimulus that causes this immune activation.^12,13 The classic LPS signalling pathway involves LBP, cluster of differentiation 14 (CD14), myeloid differentiation protein 2 (MD2) and toll-like receptor 4 (TLR4).¹⁴

Given that the LPS signalling pathway has been implicated in obesity and metabolic disturbances in uninfected subjects and that HIV-1-infected patients show greater immune activation, we aimed here to study the LPS–LBP–CD14–MD2–TLR4 complex in treated HIV-1-infected patients with HALS. Since an interaction between LPS and TLR4 triggers the expression and release of proinflammatory cytokines and that the latter have been involved in the pathogenesis of HALS, our working hypothesis was that HALS could be linked to genetic variability in the LPS signalling pathway.

Patients and methods

Design, setting and participants

This was a multicentre, cross-sectional case–control study comprising 558 treated adult HIV-1-infected patients, 318 without HALS and 240 with overt HALS. The sample size was calculated to find a difference in the distribution of polymorphisms between the two subsets of 13%, assuming a proportion of 40% in the non-HALS risk category, a risk α value of 5% and a power of 80%; the number of individuals to be included was 240 per group. This number has proved to be useful and replicable for genetic analyses of HIV-1-infected populations.¹⁵ Studies of the LBP, CD14, MD2 and TLR4 genes were carried out in all 558 patients, while circulating LPS, LBP and soluble CD14 (sCD14) levels were assessed in a subset of 163 patients (82 without HALS and 81 with HALS) whose stored plasma samples, drawn when enrolled, were available. Patients were consecutively recruited between 2004 and 2006 at the HIV outpatient clinic of the participating hospitals. Patients were selected from among those who were receiving HAART, defined as the combination of two nucleoside reverse transcriptase inhibitors (NRTIs) plus either a non-nucleoside reverse transcriptase inhibitor (NNRTI) or protease inhibitor(s) [PI(s)]. All the selected patients had to fulfil the following inclusion criteria: age >18 years, presence of HIV-1 infection, stable ART regimen for at least 1 year and the presence or absence of HALS according to previously defined standardized criteria (see below). Exclusion criteria were the presence of active opportunistic infections, current inflammatory diseases or conditions, consumption of drugs with known metabolic effects, such as steroids (systemic, inhaled or topical), antidiabetic or hypolipidaemic drugs, and hormones, and plasma C reactive protein >1 mg/dL. A flow chart with patient selection and enrolment is provided in Figure S1 (available as Supplementary data at JAC Online). Ethics committees from the participating institutions approved the project. Informed consent was obtained from each participant.

Assessment of HALS

All patients were given a full physical examination to assess the type (lipoatrophy, lipohypertrophy or mixed) and degree (slight, moderate or severe) of lipodystrophy. Criteria for lipoatrophy were one or more of the following: loss of fat from the face, arms and legs, prominent veins in the arms and legs and a thin bottom. Lipohypertrophy was defined by the presence of one or more of the following criteria: increase in abdominal perimeter, breast and/or neck fat deposition. We defined mixed lipodystrophy as being when at least one characteristic of lipoatrophy and one of lipohypertrophy were concomitantly present in a given patient. Lipodystrophy was categorized in accordance with a previously validated scale:¹⁶ nil (0), slight (1), moderate (2) and severe (3). Doubtful cases were excluded. This categorization was evaluated for the face, arms, legs, buttocks, abdomen, neck and breasts. The sum of the values corresponding to each body area indicated the degree of lipodystrophy: nil (0), slight (1–6), moderate (7–12) and severe (13–18).^16–18 In this study we included only extreme lipodystrophy phenotypes (nil versus severe cases) in order to avoid superposition between groups. To objectively assess the distribution of visceral adipose tissue and subcutaneous adipose tissue, a single-slice CT scan was performed at the L4 level in the 558 patients included in this study. The surface of adipose tissue was measured in cm².

Laboratory methods

Blood measurements

HIV-1 infection-related parameters and plasma glucose, insulin, homeostasis model assessment for insulin resistance (HOMA-IR), total cholesterol, high-density lipoprotein cholesterol (HDLc), low-density lipoprotein cholesterol (LDLc) and triglycerides were measured as previously desribed.^17,18

Assessment of polymorphisms

All the single-nucleotide polymorphisms (SNPs) were validated by genotyping (KBioscience, Herts, UK) and were as follows: LBP, rs2232582 (T→C); rs2232596 (G→A) and rs2232618 (T→C);¹⁹CD14, rs2569190 (A→G);²⁰TLR4, rs4986790 (A→G, Asp299Gly);²¹ rs4986791(C→T, Thr399Ile);²² rs1927911 (C→T)²² and rs11536889 (G→C);²³MD-2, rs11465996 (C→G).²⁴

LPS, LBP and sCD14 plasma levels

Circulating LPS has been proposed as a direct marker of bacterial translocation²⁵ and LPS levels were determined with a commercial Limulus Amebocyte Lysate (LAL) kit (Lonza, Spain). Plasma LBP levels were measured using the Human LBP ELISA kit (HyCult Biotechnology bv; PB Uden, The Netherlands). Plasma sCD14 was assayed with a commercially available ELISA kit on sera diluted 0.01% in duplicate wells (R&D Systems, Abingdon, UK).

Statistical analyses

Statistical analyses were carried out using the SPSS/PC+ statistical package (version 17; SPSS, Chicago, IL, USA). Prior to the statistical analyses, normal distribution and homogeneity of the variances were tested. Normally distributed data are expressed as mean ± SD, whereas variables with a skewed distribution are expressed as IQR or transformed into a decimal logarithm. Categorical variables are expressed as number (%). Qualitative variables were analysed using the χ² test or Fisher's exact test as necessary. Student's t-test and one way ANOVA with the post hoc Bonferroni test were used to compare continuous variables between two groups and more than two groups, respectively; to compare variables that did not have a Gaussian distribution we used the Mann–Whitney U-test and Kruskal–Wallis test. CD4+ T cell variation or recovery and the HIV-1 plasma viral load between two timepoints, pre-HAART and current, were analysed using two-way repeated measures ANOVA. Associations between quantitative variables were evaluated using Pearson correlation analysis or the Spearman correlation for non-normally distributed variables. The independence of associations was evaluated by linear regression analysis. In all statistical tests, a P value <0.05 was considered significant.

Results

Characteristics of study population

Table 1 and Table S1 (available as Supplementary data at JAC Online) show demographic, clinical and metabolic characteristics of the patients studied categorized according to the presence or absence of HALS. All patients with HALS had marked lipoatrophy that was severe enough to be treated with facial implants and most had also lipohypertrophy.

Table 1.

Demographic and clinical characteristics of the patients studied, categorized according to the presence or absence of HALS

	Non-HALS (n = 318)	HALS (n = 240)	P
Age (years)	43.5 ± 9.5	45.5 ± 9.7	0.01
Male, n (%)	227 (71.4)	171 (71.3)	1
Body mass index (kg/m2)	22.9 ± 3.1	23.6 ± 2.8	0.01
Waist–hip circumference ratio	0.89 ± 0.09	0.93 ± 0.08	0.04
Subcutaneous abdominal fat tissue (cm2)a	128.9 ± 61.3	50 ± 30.7	0.004
Visceral abdominal fat tissue (cm2)a	37.2 ± 27	110.2 ± 68	0.01
HCV infection, n (%)	120 (37.7)	79 (32.9)	0.23
HIV-1 risk group, n (%)
homosexual	111 (34.9)	76 (31.7)	0.4
heterosexual	93 (29.2)	81 (33.8)	0.35
injection drug user	114 (35.8)	77 (32.1)	0.35
other/unknown	0	6 (2.5)	0.004
Duration of HIV infection (years)	11.7 ± 5.7	13.7 ± 5.9	0.002
AIDS, n (%)	85 (26.7)	118 (49.1)	< 0.001
CD4 cell count (cells/mm3)
pre-HAART	361 ± 267	346 ± 257	0.68
current	474 ± 297	560 ± 308	0.04
Log plasma HIV-1 load (copies/mL)
pre-HAART	4.4 ± 1.2	4 ± 1.5	0.03
current	2.6 ± 1.2	2.2 ± 0.8	<0.001
Current plasma HIV-1 load <200 copies/mL, n (%)	208 (65.4)	189 (78.8)	<0.001
Exposure to NRTI before HAART, n (%)	226 (71.1)	198 (82.5)	0.002
Duration of HAART (months)	81 (37–117)	66 (44–126)	0.9
NRTI consumption, n (%)	300 (94.3)	239 (99.6)	<0.001
Cumulative time on NRTI (months)	128 (72–168)	159 (98–180)	0.05
NNRTI consumption, n (%)	220 (69.2)	168 (70)	0.8
Cumulative time on NNRTI (months)	21 (11–36)	26 (15–54)	0.008
PI consumption, n (%)	251 (78.9)	217 (90.4)	<0.001
Cumulative time on PI (months)	47 (27–88)	55 (32–84)	0.15
Zidovudine consumption, n (%)	225 (70.8)	186 (77.5)	0.03
Cumulative time on zidovudine (months)	45 (24–80)	41 (10–61)	0.03
Stavudine consumption, n (%)	156 (49.1)	196 (81.7)	<0.001
Cumulative time on stavudine (months)	37 (19–72)	39 (24–60)	0.93

Data are expressed as mean ± SD or as median and IQR; however, qualitative variables are expressed as n (%).

^aAssessed by CT scan at the L4 level.

Genetic association study

Data are shown in Table 2. Distributions of genotypes and allele frequencies of all SNPs assessed were in accordance with the expected Hardy–Weinberg equilibrium. There was an association between the LBP rs2232582 T→C SNP and HALS, carriers of the T allele being significantly overrepresented in HALS patients in both genotype and allele analyses. There was a marginal but significant association between the CD14 rs2569190 A→G SNP and HALS, but in the genotype analysis only. LBP and sCD14 plasma levels were not influenced by LBP and CD14 polymorphism, respectively, either in the whole cohort or in the HALS and non-HALS subsets.

Table 2.

Association between polymorphisms in the LBP, TLR4, CD14 and MD2 genes and HALS

Polymorphism	Genotypes	Non-HALS (n = 318)	HALS (n = 240)	P
LBP rs2232582 T→C	T/T	70.8	80.8	0.01
	C/T	27	18.8
	C/C	2.2	0.4
	allele T	84.3	90	0.048
LBP rs2232596 G→A	G/G	23.9	27.5	0.51
	G/A	47.5	47.5
	A/A	28.6	25
	allele A	52.5	48.8	0.38
LBP rs2232618 T→C	T/T	86.8	86.7	0.94
	C/T	11.6	12.1
	C/C	1.6	1.2
	allele T	92.8	92.5	0.9
CD14 rs2569190 A→G	A/A	22.3	29.6	0.046
	G/A	52.2	42.1
	G/G	25.5	28.3
	allele G	51.6	49.6	0.64
MD2 rs11465996 C→G	C/C	54.4	52.5	0.85
	G/C	40.6	42.9
	G/G	5	4.6
	allele C	74.8	74.2	0.86
TLR4 rs4986791 C→T	C/C	87.4	89.6	0.43
	C/T	12.6	10.4
	allele C	93.7	95	0.52
TLR4 rs11536889 G→C	G/G	78	80.4	0.64
	C/G	18.2	17.1
	C/C	3.8	2.5
	allele G	87.1	89.2	0.46
TLR4 rs1927911 C→T	C/C	56.9	51.2	0.39
	C/T	36.8	40.8
	T/T	6.3	8
	allele C	75.2	71.7	0.35
TLR4 rs4986790 A→G	A/A	88.4	90	0.54
	G/A	11.6	10
	allele A	94.3	95	0.73

Values are expressed as percentages.

Plasma levels of LPS, LBP and sCD14

HALS study

Figure 1 shows the values of circulating LPS, LBP and sCD14 in the assessed patients categorized according to the presence or absence of lipodystrophy. Patients with HALS had significantly higher LPS and LBP levels and significantly lower circulating sCD14 levels compared with patients without HALS.

Figure 1.

Plasma LPS, LBP and sCD14 levels in subsets of 163 patients categorized according to the absence (n = 82) or presence (n = 81) of HALS. Data are log transformed and expressed as mean ± SD.

Correlations with metabolic and immunovirological parameters

We assessed the correlations between LPS, LBP and sCD14 and total cholesterol, LDLc, HDLc, triglycerides, glycaemia, insulin, HOMA-IR, pre-HAART and current plasma viral load, pre-HAART and current CD4+ T cell count, and CD4+ T cell gain due to HAART. Each correlation was assessed in the whole cohort and in the subsets with and without HALS separately. Correlations with metabolic data are shown in Tables S2, S3 and S4 (available as Supplementary data at JAC Online). Correlations with immunovirological data are shown in Tables S5, S6 and S7 (available as Supplementary data at JAC Online).

Relationship between LPS, LBP and sCD14

In the whole cohort, there was a correlation between plasma LPS and LBP levels (r = 0.28, P < 0.001) and between plasma LBP and sCD14 levels (r = −0.21, P = 0.008). In patients with HALS there was a correlation between plasma LBP and sCD14 (r = 0.24, P = 0.03), and in the non-HALS subset there was a correlation between plasma LPS and LBP (r = 0.33, P = 0.002).

Regression analyses

To investigate the strength of the associations, we constructed linear regression analyses considering the levels of LPS, LBP and sCD14 as dependent variables and including the above-mentioned bivariate correlations, adjusting for age and gender. For LPS, the model had a multiple correlation coefficient of R = 0.73 and plasma LPS levels were mainly predicted by triglycerides (B = 0.05; P < 0.001) and hepatitis C virus (HCV) (B = −0.09, P = 0.04). For LBP, the model had a multiple correlation coefficient of R = 0.64 and plasma LBP levels were independently predicted by the presence of HALS (B = 0.74, P < 0.001). For sCD14, the model had a multiple correlation coefficient of R = 0.61 and plasma sCD14 levels were predicted by age (B = −0.01; P = 0.008), PI consumption (B = −0.41, P = 0.018) and current viral load (B = 0.11; P = 0.001).

Discussion

This is the first report to assess the perturbations of both circulating and gene polymorphisms of the members of the LPS–LPB–CD14–MD2–TLR4 complex together in treated HIV-1-infected patients with HALS. We found that the LBP rs2232582 C→T and CD14 rs2569190 A→G SNPs were associated with HALS. LBP and CD14 genotypes did not modulate the plasma levels of the molecules for which they respectively encode. In patients with HALS there was an excess of circulating LPS and LPB while in subjects without HALS circulating sCD14 levels were lower. The strongest independent predictors of plasma LPS were triglycerides and HCV, HALS was the unique independent predictor of circulating LBP levels and sCD14 levels were independently determined by age, PI consumption and plasma HIV-1 RNA levels.

Since not all treated HIV-1-infected patients develop HALS, a host genetic individual vulnerability has been proposed.²⁶ In this study we have assessed the effect of LBP, CD14, MD2 and TLR4 genetic variants. The rationale behind this study is the involvement of these genes in obesity^7,27 and HALS is reminiscent of obesity.⁶ Our data indicate that the LBP rs2232582 C→T SNP may be implicated in HALS, carriers of the T allele having greater prevalence of HALS; carriage of the C allele could therefore partially protect against HALS. Note that our finding of a positive association regarding LBP polymorphism is likely to be further replicated since it is based on double contrasts (genotype and allele analyses).²⁸ We also found a marginal association between polymorphism in the CD14 gene and HALS, but in the genotype analysis only. No association with MD2 and TLR4 polymorphisms and HALS was observed, confirming previous investigations.²⁹ A question arises as to which pathophysiological mechanisms could explain the association between LBP genetic variants and HALS. We have shown here that HALS patients have high systemic LBP levels, but these do not seem to be under genetic control in our patients. Hence, the mechanistic pathways that could explain the association between LBP gene variants and HALS remain obscure and demand further research. Potential explanations are an impairment of LBP function by the T to C mutation and that patients with the TT genotype are simply much better able to collect LPS and bring it to TLR4, thus amplifying the effects of circulating LPS.

An additional finding in our study is that HALS patients have high circulating LPS and LBP and low sCD14 levels. Inflammation biomarkers such as LPS and LBP have been studied in the HIV-1 setting but reports have focused on immunopathogenesis and immune activation^{11–13,25,30} and on several clinical outcomes and co-morbidities.^31,32 Untreated HIV-1-infected patients have subclinical inflammation³³ and the inflammatory parameters tend to improve when on HAART, but normal levels are usually not reached, even in patients who achieve virological suppression.¹¹ Innovative investigations from Koethe et al.^5,34 showed that HIV-1-infected individuals on HAART who were overweight had lower sCD14 plasma levels compared with infected patients of normal weight. Our data agree with this, since we found that our HALS subset also had significantly lower circulating sCD14 levels. The reason for the decrease in sCD14 in these patients is not clear but it could perhaps be related to CD14 down-regulation as a protective mechanism against excess of LPS and LBP. New information from our study is that individuals with an extreme HALS phenotype have marked systemic overproduction of LPS and LBP, a previously unreported finding. This implies that individuals with HALS probably have high intestinal bacterial translocation, which is highlighted by the fact that the most robust independent determinant of LBP levels in our study was the presence of HALS. The origin of the increased LBP overproduction in these patients seems to be the adipose tissue, as has been demonstrated in obese subjects and in a small cohort of HIV-1-infected patients with HALS,²⁷ and this reinforces the strong interactions between adipose tissue and immune function in HIV-1-infected subjects.^35,36 This is highlighted by the finding that in our HALS patients plasma levels of both LPS and LBP were positively correlated with pre-HAART HIV-1 viral load and negatively correlated with CD4+ T cell count. This could reflect a higher degree of inflammation and of immune activation in these patients.

Overall, it should be noted that there was a discrepancy in the presence of an association between rs2232582 T→C and HALS, an association between LBP levels and HALS, but the lack of association between rs2232582 T→C and LBP. Despite this inconsistency, our findings may stimulate further research following this line. It could be useful to assess the relationship between the polymorphisms (and the circulating molecules) in the LPS pathway and downstream inflammatory mediators such as IL-6 and TNF-α. Also, the assessment of the effect on adipose tissue biology related to peroxisome proliferator-activated receptor-γ (PPAR-γ), lipid storage and release, inflammatory cell infiltration, fibrosis, and cytokine/adipokine gene expression should be of interest.

Our study has some limitations. Firstly, the cross-sectional nature of our design provides associations but not causality. Secondly, our genetic association findings need further replication. We believe, however, that our results are of value given the large number of patients studied, the good categorization of extreme phenotypes and the fact that our positive association findings are based on two genetic contrasts. Finally, we acknowledge that the multiple comparisons performed in this study make some of the marginal associations found less likely to be real.

In summary, when taken together our data suggest that genetic variability within the LBP gene may be associated with HALS. There is a strong relationship between HALS and systemic inflammation, and the presence of HALS predicts circulating LBP levels independently. These findings may open a new and interesting mechanistic pathway for investigating the causes of HALS.

Other members of the HALS Study Group and contributors to this paper

Alba Aguilar, Montserrat Vargas (Hospital Universitari Joan XXIII, IISPV, Universitat Rovira i Virgili, Tarragona, Spain); Àngels Fontanet, Mar Gutiérrez, Gràcia Mateo, Jessica Muñoz, M. Antònia Sambeat (Hospital de la Santa Creu i Sant Pau, Universitat Autònoma de Barcelona, Barcelona, Spain); Lander Egaña-Gorrondo [Institut d'Investigació Sanitària August Pi i Sunyer (IDIBAPS), Hospital Clinic, Barcelona, Spain].

Funding

This work has been partially financed by grants from the Fondo de Investigacion Sanitaria (PI08/1032, PI08/1195, PI09/1715, PI10/2635, PI11/2512, PI13/0796 and PI13/1912), Instituto de Salud Carlos III; Fondos Europeos para el Desarrollo Regional (FEDER); Ministerio de Sanidad, Política Social e Igualdad (EC11-293); Fundación para la Investigación y Prevención del Sida en España (FIPSE 06/36572 and 06/36610); Programa de Suport als Grups de Recerca AGAUR (2009 SGR 1061); Red de Investigación en Sida (RIS): RIS-IP01, RIS-EST29, and RD12/0017/0001, RD12/0017/0005, RD17/0017/0014, RD17/0017/0029, Instituto de Salud Carlos III. Spain. F. V. (INT11/240 and INT12/282) and P. D. (INT12/383 and INT13/232) are supported by a grant from the Programa de Intensificación de Investigadores, Instituto de Salud Carlos III. M. A. is supported by the Instituto de Salud Carlos III and the Departament de Salut de la Generalitat de Catalunya.

Transparency declarations

None to declare.

Acknowledgements

Phil Hoddy kindly improved the English text. The comments and criticisms of the anonymous referees greatly helped us in improving the manuscript.

Author notes

†

These authors contributed equally to this work.

‡

See the Acknowledgements section.