Original Article doi: 10.1111/joim.12228
The collagen cross-linking enzyme lysyl oxidase is associated with the healing of human atherosclerotic lesions O. A. Ovchinnikova1,2,*, L. Folkersen1,*, J. Persson3, J. H. N. Lindeman4, T. Ueland5,6, P. Aukrust5, N. Gavrisheva2, E. Shlyakhto2, G. Paulsson-Berne1, U. Hedin7, P. S. Olofsson8 & G. K. Hansson1 From the 1Department of Medicine, Center for Molecular Medicine, Karolinska University Hospital, Stockholm, Sweden; 2Almazov Federal Heart, Blood and Endocrinology Centre, St. Petersburg, Russia; 3Department of Clinical Sciences, Karolinska Institutet, Danderyd University Hospital, Stockholm, Sweden; 4Department of Vascular Surgery, Leiden University Medical Center, Leiden, The Netherlands; 5The Research Institute for Internal Medicine, Oslo University Hospital Rikshospitalet, University of Oslo; 6Section of Endocrinology, Oslo University Hospital Rikshospitalet, University of Oslo, Oslo, Norway; 7Department of Surgery, Karolinska University Hospital, Stockholm, Sweden; and 8The Feinstein Institute for Medical Research, Manhasset, NY, USA
Abstract. Ovchinnikova OA, Folkersen L , Persson J, Lindeman JHN, Ueland T, Aukrust P, Gavrisheva N, Shlyakhto E, Paulsson-Berne G, Hedin U, Olofsson PS, Hansson GK (Center for Molecular Medicine, Karolinska University Hospital, Stockholm, Sweden; Almazov Federal Heart, Blood and Endocrinology Centre, St. Petersburg, Russia; Center for Molecular Medicine, Karolinska University Hospital, Stockholm; Karolinska Institutet, Danderyd University Hospital, Stockholm, Sweden; Leiden University Medical Center, Leiden, The Netherlands; The Research Institute for Internal Medicine, Oslo University Hospital Rikshospitalet, University of Oslo, Oslo; Section of Endocrinology, Oslo University Hospital Rikshospitalet, University of Oslo, Oslo; The Research Institute for Internal Medicine, Oslo University Hospital Rikshospitalet, University of Oslo, Oslo, Norway; Almazov Federal Heart, Blood and Endocrinology Centre, St. Petersburg; Almazov Federal Heart, Blood and Endocrinology Centre, St. Petersburg, Russia; Center for Molecular Medicine, Karolinska University Hospital, Stockholm; Department of Surgery, Karolinska University Hospital, Stockholm, Sweden; The Feinstein Institute for Medical Research, Manhasset, NY, USA; Center for Molecular Medicine, Karolinska University Hospital, Stockholm, Sweden). The collagen cross-linking enzyme lysyl oxidase is associated with the healing of human atherosclerotic lesions. J Intern Med 2014; doi: 10.1111/joim.12228.
Background. Acute clinical complications of atherosclerosis such as myocardial infarction (MI) and ischaemic stroke are usually caused by thrombus formation on the ruptured plaque surface. Collagen, the main structural protein of the fibrous cap, provides mechanical strength to the atherosclerotic plaque. The integrity of the fibrous cap depends on collagen fibre cross-linking, a process controlled by the enzyme lysyl oxidase (LOX). Methods and results. We studied atherosclerotic plaques from human carotid endarterectomies. LOX was strongly expressed in atherosclerotic lesions and detected in the regions with ongoing fibrogenesis. Higher LOX levels were associated with a more stable phenotype of the plaque. In the studied population, LOX mRNA levels in carotid plaques predicted the risk for future MI. Within the lesion, LOX mRNA levels correlated positively with levels of osteoprotegerin (OPG) and negatively with markers of immune activation. The amount of LOXmediated collagen cross-links in plaques correlated positively also with serum levels of OPG. Conclusions. Lysyl oxidase may contribute to the healing of atherosclerotic lesions and to the prevention of its lethal complications. Mediators of inflammation may control LOX expression in plaques and hence plaque stability. Keywords: atherosclerotic oxidase.
plaque,
collagen,
lysyl
*These authors contributed equally.
ª 2014 The Association for the Publication of the Journal of Internal Medicine
1
O. A. Ovchinnikova et al.
Introduction Acute clinical complications of atherosclerosis such as myocardial infarction (MI) and ischaemic stroke are caused by the thrombus formation on the plaque surface [1, 2]. The most frequent pathoanatomical substrate for sudden arterial thrombosis is a rupture of the fibrous cap that overlies the lipid core of the plaque [1]. Fibrillar collagens types I and III, synthesized by vascular smooth muscle cells (SMCs), are abundantly present in the fibrous cap and determine the biomechanical stability of the lesion [3]. Consequently, expression of procollagen genes in the plaque contributes to a gene expression profile that is associated with clinical manifestations of atherosclerosis [4]. The strength of the fibrous cap depends not only on the amount of collagen, but also on cross-linking generated during the maturation of collagen fibres [5]. Nascent procollagen polypeptides undergo a series of post-translational modifications that result in aggregation into collagen fibres and their maturation by intermolecular cross-linking [5]. Lysyl oxidase (LOX) is an extracellular copper-dependent enzyme that catalyses the formation of aldehydes from lysyl and hydroxylysyl residues within the C- and N-terminal telopeptides of collagen fibrils. This reaction leads to the formation of covalent intermolecular mature cross-links, hydroxylysylpyridinoline (HP) and lysylpyridinoline (LP), which results in the insolubilization and stabilization of extracellular collagen [5, 6]. Lox gene deletion causes fetal death due to defects in the cardiovascular system including aortic aneurysms [7], illustrating its importance for vascular pathophysiology. Sites of atherosclerotic plaque rupture are characterized by the presence of activated macrophages, T cells and mast cells that can influence collagen strength within the fibrous cap by producing proinflammatory cytokines, proteases, coagulation factors, reactive oxygen species and vasoactive molecules [1]. Interferon (IFN)-c produced by Th1 cells strongly inhibits the proliferation of vascular SMCs and the production of interstitial collagens and LOX by these cells [8–10]. Other proinflammatory cytokines such as tumour necrosis factor (TNF)-a induce expression of collagenolytic proteases by activated macrophages and mast cells in plaques [1, 11]. Atherosclerotic lesions of mice with increased T-cell activation display reduced amounts of mature, cross-linked collagen due to inhibition of 2
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
Lysyl oxidase in plaque stability
LOX by cytokines of activated T cells [12]. Two risk factors for atherosclerotic heart disease, hypercholesterolaemia and hyperhomocysteinaemia, can decrease LOX levels in cells [10, 13]. Also, TNF-a, as a prototypical proinflammatory cytokine, is capable of stimulating LOX expression in cardiac fibroblasts through transforming growth factor (TGF)-b signalling [14]. Treatment of atherosclerosis-prone mice with osteoprotegerin (OPG), a decoy receptor for the receptor activator of nuclear factor jB ligand (RANKL), promotes vascular SMC accumulation, LOX-dependent collagen fibre formation and development of fibrous caps [15]. However, the data on LOX in human atherosclerosis are limited. We investigated LOX expression and regulation in human atherosclerosis and hypothesized that LOX-dependent collagen cross-linking plays an important role in limiting complications of atherosclerotic disease by stabilizing the atherosclerotic plaque and preventing fibrous cap rupture. Methods Human specimen collection Human atherosclerotic plaque tissue was obtained from the Biobank of Karolinska Endarterectomies (BiKE) [16] and from the Biobank of St. Petersburg Investigation of Carotid Endarterectomies (SPICE) (Table 1). Atherosclerotic plaques from carotid arteries were retrieved during carotid endarterectomy (CEA) after obtaining informed consent. The studies were approved by the respective ethics committees for human studies. Surgery was performed due to the presence of symptoms (transient ischaemic attack, nondisabling ischaemic stroke or retinal ischaemia) or ipsilateral high-grade carotid stenosis [17]. Asymptomatic patients (32.5% in BiKE and 54.8% in SPICE) were operated upon according to the asymptomatic carotid surgery trial results [18]. Details of sample preparation are available in Data S1. RNA analysis Gene expression profiling was performed in the BiKE database. One set of 126 RNA samples was measured on Affymetrix Gene Array U133 Plus 2.0 (Affymetrix, Santa Clara, CA, USA). Another set of 133 RNA samples was analysed by real-time reverse-transcription polymerase chain reaction (rt-RT-PCR). The two sets had an overlap of 77 samples. Further description of preprocessing, choice of probe sets and quality control is available in Data S1.
73 (46)
Symptomatic
4.2 1.8
4.3 1.3
Stenosis, %
Total cholesterol
87.3 (55)
95.2 (60)
Antihypertensive
Antiplatelet drugs
91.9 (57)
84.1 (53)
0.612
0.782
0.847
0.375
0.212
0.416
0.611
0.147
0.448
0.0524
94.2 (65)
80.8 (63)
82.6 (57)
4.6 1.2
80 0
54 (34)
25 4.6
76.8 (53)
72 11
72.5 (50)
67
87.3 (55)
87.3 (69)
79.7 (51)
4.5 1.7
80 10
64.4 (38)
26 3.8
78.8 (52)
74 10
83.3 (55)
66
0.361
0.602
0.564
0.865
0.0209
0.637
0.0893
0.922
0.923
0.626
P-Value
100 (21)
80 (16)
52.4 (11)
5.6 2.4
80 5
55 (11)
26 2.5
52.4 (11)
66 15
90.5 (19)
21
(protein)
High LOX
SPICE cohort
76.2 (16)
66.7 (14)
33.3 (7)
4.9 2.2
80 15
42.9 (9)
25 3.5
38.1 (8)
63 12
66.7 (14)
21
(protein)
Low LOX
0.411
0.715
0.346
0.961
0.427
0.655
0.59
0.491
0.623
0.384
P-Value
BMI, body mass index; BiKE, Biobank of Karolinska Endarterectomies; rt-RT-PCR, real-time reverse-transcription polymerase chain reaction; SPICE, St. Petersburg Investigation of Carotid Endarterectomies; LOX, lysyl oxidase. Categorical variables are indicated as percentage (number), with the P-value from a chi-squared test. Numerical variables are indicated as median interquartile range, with the P-value from a Wilcoxon rank sum test.
drugs
88.9 (56)
Statins
78.3 (47)
99 16
99 9
Smoking
(mol L1)
60 (33)
25 5
50 (29)
BMI, kg m2 26 3.8
62.9 (39)
74 13
Age, years
patients
82.3 (51) 70 14
73 (46)
Male
63
63
No. of patients
(mRNA)
Low LOX
(mRNA)
(mRNA)
(mRNA)
High LOX
Low LOX
High LOX P-Value
BiKE rt-RT-PCR cohort
BiKE microarray cohort
Table 1 Characteristics of patients in biobanks used in the study
O. A. Ovchinnikova et al. Lysyl oxidase in plaque stability
ª 2014 The Association for the Publication of the Journal of Internal Medicine
Journal of Internal Medicine
3
O. A. Ovchinnikova et al.
The population of SMCs in the atherosclerotic lesion exhibits extensive phenotypic diversity [19]. Therefore, to estimate the content of SMCs in atherosclerotic lesions, we calculated an SMC index that reflected the median mRNA levels for all SMC-specific genes as determined by gene expression array of plaque mRNA (Table S1). Immunohistochemistry and immunofluorescence To localize LOX in human atherosclerotic plaques, immunohistochemical and immunofluorescent staining was performed on 42 samples from SPICE. The list of primary antibodies, staining procedures and procedures for image analysis are described in the Data S1. Protein analysis Detailed description is available in Data S1. LOX was detected by Western blotting; collagen fibres were detected by Picrosirius Red and analysed under linear polarized light; enzymatic cross-links and nonenzymatic cross-links were analysed by reverse-phase HPLC [20]. Osteoprotegerin, interleukin (IL)-6 and C-reactive protein (CRP) levels in serum and OPG level in plaque tissue were measured with ELISA kits from R&D Systems (Minneapolis, MN, USA), neopterin – with a kit from BRAHMS (Hennigsdorf, Germany) and soluble RANKL – with a kit from Peprotech (Rocky Hill, NJ, USA). Prediction of MI Clinical events in BiKE patients were retrieved from the Swedish Hospital Discharge Register and the Swedish Cause of Death Register [21]. The followup time was 35 14 months for the rt-RT-PCR cohort (n = 133) and 22 13 months for the microarray cohort (n = 126). Statistical analysis Spearman0 s method was used to calculate correlation coefficients, and all correlations were adjusted for multiple comparisons by Bonferroni correction. The P-value threshold was set at 0.05. All twogroup comparisons were made using two-sided tests: Wilcoxon rank sum test for numerical data and chi-squared test for categorical data. Both tests were used as implemented in R version 2.11.0 (R Foundation for Statistical Computing, Vienna, 4
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
Lysyl oxidase in plaque stability
Austria), using default settings. Follow-up data were analysed using Cox regression as previously described [22]. A Cox regression model was fitted with the censored data of MI events as response, LOX expression as main effect, and age and sex as covariates. Other covariates were investigated as described in Table 1. The data used in Cox regression were log2-transformed before analysis. The calculation was performed using the coxph function from the survival package as implemented in R version 2.11.0. Results Lysyl oxidase is strongly expressed in human atherosclerotic lesions and forms ‘fibrogenic spots’ Atherosclerotic plaques were collected from patients undergoing CEA. Clinical parameters, RNA and tissues for histological analysis were collected in two biobanks (Table 1). LOX mRNA was highly expressed in human carotid plaques. In global expression analysis, LOX mRNA was detected amongst the top 10% of all genes (data not shown). Immunostaining of human atherosclerotic lesions demonstrated prominent LOX protein expression in the fibrous cap (Fig. 1). LOX staining was more pronounced in the deeper layers of fibrous cap and in close proximity to necrotic areas (Fig. 1). Lysyl oxidase mRNA levels correlated significantly with mRNA for the three main procollagen types present in vascular tissue: I, III and IV, but not with mRNA for elastin (Table 2). LOX levels also correlated with mRNA for collagen prolyl-4 hydroxylase (C-P4H) and lysyl hydroxylase (Table 2), enzymes that are important for cross-linking of collagen fibrils in the intracellular space [5]. Similar correlations were observed on the protein level. LOX+ staining showed a histological co-localization with C-P4H in atherosclerotic plaques (Fig. 1a,b) (r = 0.6; P < 0.001). There was a significant positive correlation between active LOX protein and the percentage of collagen in tissue (r = 0.54, P = 0.009; Fig. 2a), as well as the amount of enzymatically formed collagen cross-links (r = 0.55, P = 0.009; Fig. 2b). In contrast, there was no significant correlation between active LOX protein and nonenzymatic advanced glycation endproducts (AGE)-induced cross-links (pentosidine) (Fig. 2c). Moreover, LOX+ staining was partially co-localized with areas positive for fibrillar collagen
O. A. Ovchinnikova et al.
Lysyl oxidase in plaque stability
(a)
(b)
(c)
(d)
(e)
(f)
Fig. 1 Lysyl oxidase (LOX) protein in human carotid atherosclerotic plaques. (a–e) Serial sections were stained with antibodies against LOX (a), C-P4H (b), aSM-actin (c), collagen type I (d) and type IV (e). Immunoperoxidase micrographs show specific staining in red and haematoxylin-stained nuclei in blue, original magnification, 925. Black bars are 1 mm. (f). Micrograph shows collagen fibres in the fibrous cap stained with Picrosirius Red and visualized under polarized light. L, lumen; FC, fibrous cap; NC, necrotic core.
type I and basement membrane collagen type IV (Figs 1d,e and 3d). Fibrogenic cells are the main source of LOX in tissues [6]; therefore, we assessed whether LOX mRNA levels were dependent on the content of SMCs in plaques. To overcome phenotypic diversity and plasticity of SMCs in lesions [23], we calculated an SMC index that reflected the median mRNA levels for all SMC-related genes in vascular tissues (Table S1). Although LOX mRNA correlated with the SMC index, the relationship was weak, suggesting that any changes in LOX mRNA expression in atherosclerotic plaques occurred independently of SMC content (Table 2 and Figure S1).
Table 2 Correlation between mRNA levels of lysyl oxidase and factors involved in collagen biosynthesis as analysed by gene expression arrays of human carotid atherosclerotic lesions Spearman Transcript Smooth muscle
P-value
0.19
0.02
Procollagen a1(I)
0.41
The collagen cross-linking enzyme lysyl oxidase is associated with the healing of human atherosclerotic lesions O. A. Ovchinnikova1,2,*, L. Folkersen1,*, J. Persson3, J. H. N. Lindeman4, T. Ueland5,6, P. Aukrust5, N. Gavrisheva2, E. Shlyakhto2, G. Paulsson-Berne1, U. Hedin7, P. S. Olofsson8 & G. K. Hansson1 From the 1Department of Medicine, Center for Molecular Medicine, Karolinska University Hospital, Stockholm, Sweden; 2Almazov Federal Heart, Blood and Endocrinology Centre, St. Petersburg, Russia; 3Department of Clinical Sciences, Karolinska Institutet, Danderyd University Hospital, Stockholm, Sweden; 4Department of Vascular Surgery, Leiden University Medical Center, Leiden, The Netherlands; 5The Research Institute for Internal Medicine, Oslo University Hospital Rikshospitalet, University of Oslo; 6Section of Endocrinology, Oslo University Hospital Rikshospitalet, University of Oslo, Oslo, Norway; 7Department of Surgery, Karolinska University Hospital, Stockholm, Sweden; and 8The Feinstein Institute for Medical Research, Manhasset, NY, USA
Abstract. Ovchinnikova OA, Folkersen L , Persson J, Lindeman JHN, Ueland T, Aukrust P, Gavrisheva N, Shlyakhto E, Paulsson-Berne G, Hedin U, Olofsson PS, Hansson GK (Center for Molecular Medicine, Karolinska University Hospital, Stockholm, Sweden; Almazov Federal Heart, Blood and Endocrinology Centre, St. Petersburg, Russia; Center for Molecular Medicine, Karolinska University Hospital, Stockholm; Karolinska Institutet, Danderyd University Hospital, Stockholm, Sweden; Leiden University Medical Center, Leiden, The Netherlands; The Research Institute for Internal Medicine, Oslo University Hospital Rikshospitalet, University of Oslo, Oslo; Section of Endocrinology, Oslo University Hospital Rikshospitalet, University of Oslo, Oslo; The Research Institute for Internal Medicine, Oslo University Hospital Rikshospitalet, University of Oslo, Oslo, Norway; Almazov Federal Heart, Blood and Endocrinology Centre, St. Petersburg; Almazov Federal Heart, Blood and Endocrinology Centre, St. Petersburg, Russia; Center for Molecular Medicine, Karolinska University Hospital, Stockholm; Department of Surgery, Karolinska University Hospital, Stockholm, Sweden; The Feinstein Institute for Medical Research, Manhasset, NY, USA; Center for Molecular Medicine, Karolinska University Hospital, Stockholm, Sweden). The collagen cross-linking enzyme lysyl oxidase is associated with the healing of human atherosclerotic lesions. J Intern Med 2014; doi: 10.1111/joim.12228.
Background. Acute clinical complications of atherosclerosis such as myocardial infarction (MI) and ischaemic stroke are usually caused by thrombus formation on the ruptured plaque surface. Collagen, the main structural protein of the fibrous cap, provides mechanical strength to the atherosclerotic plaque. The integrity of the fibrous cap depends on collagen fibre cross-linking, a process controlled by the enzyme lysyl oxidase (LOX). Methods and results. We studied atherosclerotic plaques from human carotid endarterectomies. LOX was strongly expressed in atherosclerotic lesions and detected in the regions with ongoing fibrogenesis. Higher LOX levels were associated with a more stable phenotype of the plaque. In the studied population, LOX mRNA levels in carotid plaques predicted the risk for future MI. Within the lesion, LOX mRNA levels correlated positively with levels of osteoprotegerin (OPG) and negatively with markers of immune activation. The amount of LOXmediated collagen cross-links in plaques correlated positively also with serum levels of OPG. Conclusions. Lysyl oxidase may contribute to the healing of atherosclerotic lesions and to the prevention of its lethal complications. Mediators of inflammation may control LOX expression in plaques and hence plaque stability. Keywords: atherosclerotic oxidase.
plaque,
collagen,
lysyl
*These authors contributed equally.
ª 2014 The Association for the Publication of the Journal of Internal Medicine
1
O. A. Ovchinnikova et al.
Introduction Acute clinical complications of atherosclerosis such as myocardial infarction (MI) and ischaemic stroke are caused by the thrombus formation on the plaque surface [1, 2]. The most frequent pathoanatomical substrate for sudden arterial thrombosis is a rupture of the fibrous cap that overlies the lipid core of the plaque [1]. Fibrillar collagens types I and III, synthesized by vascular smooth muscle cells (SMCs), are abundantly present in the fibrous cap and determine the biomechanical stability of the lesion [3]. Consequently, expression of procollagen genes in the plaque contributes to a gene expression profile that is associated with clinical manifestations of atherosclerosis [4]. The strength of the fibrous cap depends not only on the amount of collagen, but also on cross-linking generated during the maturation of collagen fibres [5]. Nascent procollagen polypeptides undergo a series of post-translational modifications that result in aggregation into collagen fibres and their maturation by intermolecular cross-linking [5]. Lysyl oxidase (LOX) is an extracellular copper-dependent enzyme that catalyses the formation of aldehydes from lysyl and hydroxylysyl residues within the C- and N-terminal telopeptides of collagen fibrils. This reaction leads to the formation of covalent intermolecular mature cross-links, hydroxylysylpyridinoline (HP) and lysylpyridinoline (LP), which results in the insolubilization and stabilization of extracellular collagen [5, 6]. Lox gene deletion causes fetal death due to defects in the cardiovascular system including aortic aneurysms [7], illustrating its importance for vascular pathophysiology. Sites of atherosclerotic plaque rupture are characterized by the presence of activated macrophages, T cells and mast cells that can influence collagen strength within the fibrous cap by producing proinflammatory cytokines, proteases, coagulation factors, reactive oxygen species and vasoactive molecules [1]. Interferon (IFN)-c produced by Th1 cells strongly inhibits the proliferation of vascular SMCs and the production of interstitial collagens and LOX by these cells [8–10]. Other proinflammatory cytokines such as tumour necrosis factor (TNF)-a induce expression of collagenolytic proteases by activated macrophages and mast cells in plaques [1, 11]. Atherosclerotic lesions of mice with increased T-cell activation display reduced amounts of mature, cross-linked collagen due to inhibition of 2
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
Lysyl oxidase in plaque stability
LOX by cytokines of activated T cells [12]. Two risk factors for atherosclerotic heart disease, hypercholesterolaemia and hyperhomocysteinaemia, can decrease LOX levels in cells [10, 13]. Also, TNF-a, as a prototypical proinflammatory cytokine, is capable of stimulating LOX expression in cardiac fibroblasts through transforming growth factor (TGF)-b signalling [14]. Treatment of atherosclerosis-prone mice with osteoprotegerin (OPG), a decoy receptor for the receptor activator of nuclear factor jB ligand (RANKL), promotes vascular SMC accumulation, LOX-dependent collagen fibre formation and development of fibrous caps [15]. However, the data on LOX in human atherosclerosis are limited. We investigated LOX expression and regulation in human atherosclerosis and hypothesized that LOX-dependent collagen cross-linking plays an important role in limiting complications of atherosclerotic disease by stabilizing the atherosclerotic plaque and preventing fibrous cap rupture. Methods Human specimen collection Human atherosclerotic plaque tissue was obtained from the Biobank of Karolinska Endarterectomies (BiKE) [16] and from the Biobank of St. Petersburg Investigation of Carotid Endarterectomies (SPICE) (Table 1). Atherosclerotic plaques from carotid arteries were retrieved during carotid endarterectomy (CEA) after obtaining informed consent. The studies were approved by the respective ethics committees for human studies. Surgery was performed due to the presence of symptoms (transient ischaemic attack, nondisabling ischaemic stroke or retinal ischaemia) or ipsilateral high-grade carotid stenosis [17]. Asymptomatic patients (32.5% in BiKE and 54.8% in SPICE) were operated upon according to the asymptomatic carotid surgery trial results [18]. Details of sample preparation are available in Data S1. RNA analysis Gene expression profiling was performed in the BiKE database. One set of 126 RNA samples was measured on Affymetrix Gene Array U133 Plus 2.0 (Affymetrix, Santa Clara, CA, USA). Another set of 133 RNA samples was analysed by real-time reverse-transcription polymerase chain reaction (rt-RT-PCR). The two sets had an overlap of 77 samples. Further description of preprocessing, choice of probe sets and quality control is available in Data S1.
73 (46)
Symptomatic
4.2 1.8
4.3 1.3
Stenosis, %
Total cholesterol
87.3 (55)
95.2 (60)
Antihypertensive
Antiplatelet drugs
91.9 (57)
84.1 (53)
0.612
0.782
0.847
0.375
0.212
0.416
0.611
0.147
0.448
0.0524
94.2 (65)
80.8 (63)
82.6 (57)
4.6 1.2
80 0
54 (34)
25 4.6
76.8 (53)
72 11
72.5 (50)
67
87.3 (55)
87.3 (69)
79.7 (51)
4.5 1.7
80 10
64.4 (38)
26 3.8
78.8 (52)
74 10
83.3 (55)
66
0.361
0.602
0.564
0.865
0.0209
0.637
0.0893
0.922
0.923
0.626
P-Value
100 (21)
80 (16)
52.4 (11)
5.6 2.4
80 5
55 (11)
26 2.5
52.4 (11)
66 15
90.5 (19)
21
(protein)
High LOX
SPICE cohort
76.2 (16)
66.7 (14)
33.3 (7)
4.9 2.2
80 15
42.9 (9)
25 3.5
38.1 (8)
63 12
66.7 (14)
21
(protein)
Low LOX
0.411
0.715
0.346
0.961
0.427
0.655
0.59
0.491
0.623
0.384
P-Value
BMI, body mass index; BiKE, Biobank of Karolinska Endarterectomies; rt-RT-PCR, real-time reverse-transcription polymerase chain reaction; SPICE, St. Petersburg Investigation of Carotid Endarterectomies; LOX, lysyl oxidase. Categorical variables are indicated as percentage (number), with the P-value from a chi-squared test. Numerical variables are indicated as median interquartile range, with the P-value from a Wilcoxon rank sum test.
drugs
88.9 (56)
Statins
78.3 (47)
99 16
99 9
Smoking
(mol L1)
60 (33)
25 5
50 (29)
BMI, kg m2 26 3.8
62.9 (39)
74 13
Age, years
patients
82.3 (51) 70 14
73 (46)
Male
63
63
No. of patients
(mRNA)
Low LOX
(mRNA)
(mRNA)
(mRNA)
High LOX
Low LOX
High LOX P-Value
BiKE rt-RT-PCR cohort
BiKE microarray cohort
Table 1 Characteristics of patients in biobanks used in the study
O. A. Ovchinnikova et al. Lysyl oxidase in plaque stability
ª 2014 The Association for the Publication of the Journal of Internal Medicine
Journal of Internal Medicine
3
O. A. Ovchinnikova et al.
The population of SMCs in the atherosclerotic lesion exhibits extensive phenotypic diversity [19]. Therefore, to estimate the content of SMCs in atherosclerotic lesions, we calculated an SMC index that reflected the median mRNA levels for all SMC-specific genes as determined by gene expression array of plaque mRNA (Table S1). Immunohistochemistry and immunofluorescence To localize LOX in human atherosclerotic plaques, immunohistochemical and immunofluorescent staining was performed on 42 samples from SPICE. The list of primary antibodies, staining procedures and procedures for image analysis are described in the Data S1. Protein analysis Detailed description is available in Data S1. LOX was detected by Western blotting; collagen fibres were detected by Picrosirius Red and analysed under linear polarized light; enzymatic cross-links and nonenzymatic cross-links were analysed by reverse-phase HPLC [20]. Osteoprotegerin, interleukin (IL)-6 and C-reactive protein (CRP) levels in serum and OPG level in plaque tissue were measured with ELISA kits from R&D Systems (Minneapolis, MN, USA), neopterin – with a kit from BRAHMS (Hennigsdorf, Germany) and soluble RANKL – with a kit from Peprotech (Rocky Hill, NJ, USA). Prediction of MI Clinical events in BiKE patients were retrieved from the Swedish Hospital Discharge Register and the Swedish Cause of Death Register [21]. The followup time was 35 14 months for the rt-RT-PCR cohort (n = 133) and 22 13 months for the microarray cohort (n = 126). Statistical analysis Spearman0 s method was used to calculate correlation coefficients, and all correlations were adjusted for multiple comparisons by Bonferroni correction. The P-value threshold was set at 0.05. All twogroup comparisons were made using two-sided tests: Wilcoxon rank sum test for numerical data and chi-squared test for categorical data. Both tests were used as implemented in R version 2.11.0 (R Foundation for Statistical Computing, Vienna, 4
ª 2014 The Association for the Publication of the Journal of Internal Medicine Journal of Internal Medicine
Lysyl oxidase in plaque stability
Austria), using default settings. Follow-up data were analysed using Cox regression as previously described [22]. A Cox regression model was fitted with the censored data of MI events as response, LOX expression as main effect, and age and sex as covariates. Other covariates were investigated as described in Table 1. The data used in Cox regression were log2-transformed before analysis. The calculation was performed using the coxph function from the survival package as implemented in R version 2.11.0. Results Lysyl oxidase is strongly expressed in human atherosclerotic lesions and forms ‘fibrogenic spots’ Atherosclerotic plaques were collected from patients undergoing CEA. Clinical parameters, RNA and tissues for histological analysis were collected in two biobanks (Table 1). LOX mRNA was highly expressed in human carotid plaques. In global expression analysis, LOX mRNA was detected amongst the top 10% of all genes (data not shown). Immunostaining of human atherosclerotic lesions demonstrated prominent LOX protein expression in the fibrous cap (Fig. 1). LOX staining was more pronounced in the deeper layers of fibrous cap and in close proximity to necrotic areas (Fig. 1). Lysyl oxidase mRNA levels correlated significantly with mRNA for the three main procollagen types present in vascular tissue: I, III and IV, but not with mRNA for elastin (Table 2). LOX levels also correlated with mRNA for collagen prolyl-4 hydroxylase (C-P4H) and lysyl hydroxylase (Table 2), enzymes that are important for cross-linking of collagen fibrils in the intracellular space [5]. Similar correlations were observed on the protein level. LOX+ staining showed a histological co-localization with C-P4H in atherosclerotic plaques (Fig. 1a,b) (r = 0.6; P < 0.001). There was a significant positive correlation between active LOX protein and the percentage of collagen in tissue (r = 0.54, P = 0.009; Fig. 2a), as well as the amount of enzymatically formed collagen cross-links (r = 0.55, P = 0.009; Fig. 2b). In contrast, there was no significant correlation between active LOX protein and nonenzymatic advanced glycation endproducts (AGE)-induced cross-links (pentosidine) (Fig. 2c). Moreover, LOX+ staining was partially co-localized with areas positive for fibrillar collagen
O. A. Ovchinnikova et al.
Lysyl oxidase in plaque stability
(a)
(b)
(c)
(d)
(e)
(f)
Fig. 1 Lysyl oxidase (LOX) protein in human carotid atherosclerotic plaques. (a–e) Serial sections were stained with antibodies against LOX (a), C-P4H (b), aSM-actin (c), collagen type I (d) and type IV (e). Immunoperoxidase micrographs show specific staining in red and haematoxylin-stained nuclei in blue, original magnification, 925. Black bars are 1 mm. (f). Micrograph shows collagen fibres in the fibrous cap stained with Picrosirius Red and visualized under polarized light. L, lumen; FC, fibrous cap; NC, necrotic core.
type I and basement membrane collagen type IV (Figs 1d,e and 3d). Fibrogenic cells are the main source of LOX in tissues [6]; therefore, we assessed whether LOX mRNA levels were dependent on the content of SMCs in plaques. To overcome phenotypic diversity and plasticity of SMCs in lesions [23], we calculated an SMC index that reflected the median mRNA levels for all SMC-related genes in vascular tissues (Table S1). Although LOX mRNA correlated with the SMC index, the relationship was weak, suggesting that any changes in LOX mRNA expression in atherosclerotic plaques occurred independently of SMC content (Table 2 and Figure S1).
Table 2 Correlation between mRNA levels of lysyl oxidase and factors involved in collagen biosynthesis as analysed by gene expression arrays of human carotid atherosclerotic lesions Spearman Transcript Smooth muscle
P-value
0.19
0.02
Procollagen a1(I)
0.41
