Original article
Identification of microRNAs associated with abdominal aortic aneurysms and peripheral arterial disease P. W. Stather1 , N. Sylvius2 , D. A. Sidloff1 , N. Dattani1 , A. Verissimo1 , J. B. Wild1 , H. Z. Butt1 , E. Choke1 , R. D. Sayers1 and M. J. Bown1,3∗ Departments of 1 Cardiovascular Sciences and 2 Genetics and 3 National Institute for Health Research, University of Leicester, Leicester, UK Correspondence to: Mr P. W. Stather, Department of Cardiovascular Sciences, University of Leicester, Leicester LE2 7LX, UK (e-mail: [email protected])
Background: MicroRNAs are crucial in the regulation of cardiovascular disease and represent potential
therapeutic targets to decrease abdominal aortic aneurysm (AAA) expansion. The aim of this study was to identify circulating microRNAs associated with AAA. Methods: Some 754 microRNAs in whole-blood samples from 15 men with an AAA and ten control subjects were quantified using quantitative reverse transcriptase–PCR. MicroRNAs demonstrating a significant association with AAA were validated in peripheral blood and plasma samples of men in the following groups (40 in each): healthy controls, controls with peripheral arterial disease (PAD), men with a small AAA (30–54 mm), those with a large AAA (over 54 mm), and those following AAA repair. MicroRNA expression was also assessed in aortic tissue. Results: Twenty-nine differentially expressed microRNAs were identified in the discovery study. Validation study revealed that let-7e (fold change (FC) –1⋅80; P = 0⋅001), miR-15a (FC −2⋅24; P < 0⋅001) and miR-196b (FC −2⋅26; P < 0⋅001) were downregulated in peripheral blood from patients with an AAA, and miR-411 was upregulated (FC 5⋅90; P = 0⋅001). miR-196b was also downregulated in plasma from the same individuals (FC −3⋅75; P = 0⋅029). The same miRNAs were similarly expressed differentially in patients with PAD compared with healthy controls. Validated and predicted microRNA targets identified through miRWalk revealed that these miRNAs were all regulators of AAA-related genes (vascular cell adhesion molecule 1, intercellular cell adhesion molecule 1, DAB2 interacting protein, 𝛂1-antitrypsin, C-reactive protein, interleukin 6, osteoprotegerin, methylenetetrahydrofolate reductase, tumour necrosis factor 𝛂). Conclusion: In this study, circulating levels of let-7e, miR-15a, miR-196b and miR-411 were differentially expressed in men with an AAA compared with healthy controls, but also differentially expressed in men with PAD. Modulation of these miRNAs and their target genes may represent a new therapeutic pathway to affect the progression of AAA and atherosclerosis. ∗ Co-authors of this study can be found under the heading Contributors Presented to a meeting of the Vascular Society of Great Britain and Ireland, Manchester, UK, November 2013; published in abstract form as Br J Surg 2014; 101(Suppl 2): 1–2
Paper accepted 9 February 2015 Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.9802
Introduction
The underlying pathophysiology of abdominal aortic aneurysm (AAA) is unclear. Smoking is the most important environmental risk factor for the development of AAA; however, there is a well established genetic component as evidenced by familial1 – 3 and ethnicity4,5 studies. AAA is more common in men, increases with age and hypercholesterolaemia, and is less frequent in patients © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
with diabetes6 . Furthermore, the prevalence of AAA in patients with peripheral arterial disease (PAD) is reported to be significantly higher (9 per cent) than in the normal population7 . Meta-analysis8 of eight studies identified an increased risk of AAA in patients with PAD. AAA and PAD share many similarities, including risk factors such as smoking and hypertension, biomarkers such as fibrinogen and C-reactive protein (CRP), and genetic associations BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
such as LDLR and SORT1. However, diabetes is a negative indicator for AAA but strongly associated with PAD, male sex is strongly associated with AAA, and some recognized genetic determinants of atherosclerosis, such as Apo E polymorphisms, have no association with AAA9 . Four genetic loci have been identified through genome-wide association studies (DAB2IP, LRP1, SORT1 and LDLR)10 – 13 , and three candidate single-nucleotide polymorphisms have been identified as associated with AAA from meta-analysis (MMP9, ACE, MTHFR)14 . In addition, a wide range of potential biomarkers for AAA have been identified, with two recent meta-analyses15,16 reporting a significant association between AAA and 18 proteins (matrix metalloproteinase (MMP) 2, MMP-9, tissue inhibitor of metalloproteinase (TIMP) 1, interleukin (IL) 6, tumour necrosis factor (TNF) α, osteoprotegerin, osteopontin, interferon (IFN) γ, intercellular adhesion molecule (ICAM) 1, vascular cell adhesion molecule (VCAM) 1, D-dimer, CRP, α1-antitrypsin, fibrinogen, triglycerides, lipoprotein(a), apolipoprotein A and high-density lipoprotein). The specificity and sensitivity of these biomarkers is, however, poor with all reporter operator characteristics below 0⋅7; therefore, a different approach must be taken to understand further the underlying pathophysiology of AAA. MicroRNAs (miRNAs) are short, non-coding RNAs involved in gene regulation by binding to mRNAs, causing repression of translation and mRNA degradation, thus fine-tuning protein expression. Each single miRNA is capable of interacting with several different mRNAs, thereby affecting multiple pathways. There is a clear difference in miRNA signature between cell and tissue types. Several studies have looked at the abundance of miRNAs in specific tissues, identifying distinct patterns of miRNA expression in human skeletal muscle, heart, prostate, brain, liver, spleen and colon17 – 19 . Specific miRNAs are also found only in certain tissues, such as miR-122 in liver20 , and miR-1, miR-208 and miR-133 in muscle or heart21 . This is of particular use in acute conditions such as myocardial infarction, where heart-specific miRNAs are released into the circulation. Therefore, miRNAs may be specific to aortic aneurysm tissue or increased in aortic cells, leading to altered expression profiles. Two studies22,23 have assessed miRNA expression in human AAA tissue using a miRNA transcriptome (miRNome) approach, with conflicting results. In addition, several other studies and reviews24 – 28 have identified miR-21, miR-29b, miR-26a, miR-133, miR-145, miR-155, miR-195 and miR-221/222 as potential miRNAs associated with AAA in human and animal models. However, miRNA expression in patients with an AAA has not been
evaluated across tissue types, comparing whole blood, plasma and the aortic wall. This study therefore aimed to determine whether miRNAs can be used as circulating biomarkers for AAA by exploring whether any miRNAs are differentially expressed in whole blood, plasma and aortic tissue of patients with an AAA. A further aim was to determine whether these miRNAs remain dysregulated after surgical intervention for AAA, correlate with AAA diameter, or are specific to AAA, by assessing miRNA expression in subjects with PAD.
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
Methods
Ethical approval was obtained from Derby Research Ethics Committee for collection of blood and tissue samples (REC: 6819). Ethical approval was also obtained for cadaveric tissue samples (REC: 12/EM/0355). This study was performed in accordance with the declaration of Helsinki. Consent was obtained from all men before inclusion in the study. A two-stage study was conducted to determine which miRNAs were associated with AAA, and to establish whether these were specific to AAA or related to generalized atherosclerosis. An initial discovery study was performed quantifying 754 miRNAs in whole-blood samples taken from 15 men with a large AAA (over 54 mm) and ten healthy controls. MicroRNAs associated with AAA from the analysis of the discovery study were taken forward to a validation study of 200 men (40 in each group): healthy controls, controls with PAD, men with a small AAA (30–54 mm), those with a large AAA (over 54 mm) and men who had undergone AAA repair (sample collected 6 months after repair, both open and endovascular repair samples included; there were no postoperative complications). All 40 men in each cohort were included in the whole-blood study. For the plasma study 30 samples showed poor PCR curves, and were therefore excluded. This resulted in 28 healthy controls, 35 controls with PAD, 36 men with a small AAA, 36 with a large AAA, and 35 post-AAA repair. MicroRNAs were selected for the validation study based on a fold change (FC) threshold of more than 1⋅5 and a P value threshold of less than 0⋅100. A FC greater than 1 represented an upregulated miRNA, whereas a FC less than 1 represented a downregulated miRNA. A FC of 1 (or −1) was equivalent to no change in miRNA expression. In addition, miRNAs potentially associated with AAA taken from the literature were tested for association in the validation study. MicroRNAs included in the validation study were quantified in whole-blood samples, plasma samples and, BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
0 0·1 0·2 0·3 0·4 P
where patients with AAA were undergoing open surgery, AAA wall biopsies. Clinical information including medical and family history was obtained for all participants, including: age, smoking history, history of hypertension, hypercholesterolaemia, myocardial infarction, angina, stroke, diabetes, chronic obstructive pulmonary disease, peripheral vascular disease and any history of malignancy or autoimmune/inflammatory disease. A medication history was also taken, including: aspirin, beta-blockers, statins, clopidogrel, digoxin, warfarin, diuretics and angiotensin-converting enzyme inhibitors. All men underwent abdominal ultrasonography to measure aortic diameter using inner wall to inner wall measurements in both anteroposterior and lateral planes. Patients with a clinical history of PAD also had ultrasound imaging of the leg arteries to assess for arterial stenosis or occlusion. A small AAA was defined by a diameter of 30–54 mm, and large AAA by a diameter of more than 54 mm. The diagnosis of PAD was based on the presence of both symptoms of intermittent claudication and arterial stenosis/occlusion identified on ultrasound imaging. Whole blood and plasma were obtained from all men (except cadaveric donors). Aortic tissue samples were obtained from men undergoing open abdominal aortic surgery; samples were taken from the anterior aortic wall approximately 5 cm below the renal arteries. Samples were also taken from the infrarenal aorta of cadaveric donors (Appendix S1, supporting information). All aortic tissue samples had matched blood and plasma samples. RNA quality control was conducted using 260/280 ratios (only those with a ratio between 1⋅8 and 2⋅2 were included), RNA integrity numbers (RINs) and RNA concentration (more than 40 ng/μl for whole-blood and aortic tissue samples). RNA integrity was determined for a subgroup of samples, with a mean(s.d.) RIN of 8⋅60(0⋅81) in 11 whole-blood samples and 6⋅75(0⋅67) in six aortic tissue samples. These RIN results were consistently good, so RNA integrity was not assessed in all samples. Plasma samples lack a cellular component, and miRNAs are present owing to release from necrotic cells, apoptosis or intercellular communication. This results in a deficiency of large ribosomal RNAs, and as such the integrity of the RNA samples cannot be determined. Quality control for plasma samples was therefore done purely by ensuring the presence of RNA using a NanoDrop™ spectrophotometer (ThermoScientific, Waltham, Massachusetts, USA), with all samples requiring more than 1 ng/μl for use in subsequent processing. The following inclusion criteria were implemented in order to permit standardization and reduce bias when comparing cases and controls: men over 55 years; Caucasian;
0·5 0·6 0·7 0·8 0·9 1·0 –3
–2
–1 1
2
3
Fold change
Volcano plot showing the distribution of fold change and P value for each microRNA (miRNA) expression in the discovery study comparing 15 participants with a large adominal aortic aneurysm and ten healthy controls. The red dots represent the miRNAs that were significantly different; pink shaded areas represent a significant fold change. Thirteen miRNAs had a fold change greater than 3 or −3 (with 1 of these being significantly differentially expressed), and are therefore not included in this plot
Fig. 1
Differentially expressed microRNAs from the discovery study comparing 15 participants with large abdominal aortic aneurysm and ten healthy controls
Table 1
MicroRNA miR-672 miR-19a* miR-15a miR-16-1* miR-9* miR-125a-3p miR-196b miR-134 miR-181a miR-942 Let-7e miR-431 miR-1291 miR-27b miR-1249 miR-483-5p miR-624 miR-411 miR-589 miR-29b miR-1179 miR-1227 miR-138 miR-26a-1* miR-374a miR-125b miR-628-5p miR-148b* miR-15b*
Fold change −2⋅22 2⋅24 −2⋅04 −1⋅75 −2⋅22 69 866⋅67 −1⋅89 −2⋅63 1⋅62 −2⋅70 2⋅29 1⋅82 −1⋅85 2⋅48 −1⋅54 1⋅86 −2⋅04 −2⋅04 −1⋅51 −1⋅89 −2⋅00 1⋅70 2⋅22 −1⋅59 −1⋅69 1⋅56 −1⋅69 −1⋅52 −2⋅50
P† 0⋅003 0⋅006 0⋅019 0⋅019 0⋅024 0⋅031 0⋅031 0⋅036 0⋅041 0⋅041 0⋅041 0⋅042 0⋅048 0⋅051 0⋅055 0⋅055 0⋅057 0⋅058 0⋅068 0⋅076 0⋅081 0⋅081 0⋅081 0⋅081 0⋅081 0⋅091 0⋅091 0⋅096 0⋅096
MicroRNAs with P < 0⋅100 and more than 1⋅5-fold change were deemed to be statistically significant. †Mann–Whitney U test.
www.bjs.co.uk
BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
14 12 10 8 6
Fold change
4 2 0 –2
∗
–4 –6
∗
∗
∗
–8 –10 –12 –14
1
8
-41 miR
b
-13
-27
miR
-58
9 miR
5 -14 miR
miR
5
1 -43
-15 miR
1a -18
miR
miR
-22
1
-31
miR
b -29
e
5b
t-7
miR
miR
Le
p
-12
miR
3-5
a -15
-48 miR
miR
miR
-19
6b
–15
Whole-blood validation results showing the mean(s.d.) fold change for each microRNA for 80 participants with an abdominal aortic aneurysm compared with 40 healthy controls. *P ≤ 0⋅001 (one-way ANOVA)
Fig. 2
and no known systemic illness that could confound miRNA expression profiling, such as malignant, systemic autoimmune or inflammatory disease. Details of RNA isolation and miRNA expression profiling methods are provided in Appendix S1 (supporting information).
MicroRNA quantification For the discovery study, miRNome profiling was conducted by quantitative PCR (qPCR) using TaqMan® miRNA Cards A + B on an ABI 7900HT qPCR instrument (Life Technologies, Paisley, UK). FC was calculated using the relative standard curve method. For the validation study, selected miRNAs were profiled by digital PCR using TaqMan® OpenArray® MicroRNA assays on an OpenArray® Quantstudio™ 12 K Flex Real-Time PCR system (Life Technologies). FC in the validation study was calculated using the comparative threshold cycle (ΔCT) method. Data were normalized with respect to the geometric mean of RNU48 and MammU6. The expression curves of all miRNAs were inspected visually, and miRNAs with suboptimal amplification PCR curves (absence of clear S-shaped curves) were excluded. © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
Statistical analysis For the discovery study, background clinical data were analysed using Fisher’s exact and Mann–Whitney U tests; miRNA levels were compared between patients with an AAA and healthy controls by means of Mann–Whitney U test. For the validation study, an initial analysis was undertaken using ANOVA to determine any significant difference in miRNA expression in blood between controls, patients with an AAA and controls with PAD. Further post hoc analysis was then undertaken using t tests and binary logistic regression to determine whether significant differences were specific to AAA, or due to generalized atherosclerosis. Additional analysis was done to determine whether differences in miRNA expression were maintained after AAA repair, as this would determine whether miRNA expression is altered owing to strain on the aneurysm sac, or indicates an underlying predisposition to aneurysm formation. The correlation between aneurysm size and miRNA expression was also analysed by bivariable correlation. Aortic tissue miRNA expression was analysed using the Mann–Whitney U test and Kruskal–Wallis test. GraphPad Prism® 6 (GraphPad, La Jolla, California, USA) was used for graphical www.bjs.co.uk
BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
†
‡ ‡
‡ 6 10
ΔCT
ΔCT
4
5 2
0
a
Control
AAA
PAD
0
Postop.
b
Let-7e
Control
AAA
PAD
Postop.
PAD
Postop.
miR-15a
‡
‡
∗
†
† 15
10
ΔCT
ΔCT
10 5
5
0
0 Control
c
AAA
PAD
Control
Postop.
miR-196b
d
AAA
miR-411
Fig. 3 Mean(s.d.) comparative threshold cycle (ΔCT) values for a let-7e, b miR-15a, c miR-196b and d miR-411 in whole-blood samples from 80 patients with an abdominal aortic aneurysm (AAA), 40 with peripheral arterial disease (PAD), 40 postoperative patients who had undergone AAA repair and 40 control subjects. a P = 0⋅001, b P < 0⋅001, c P < 0⋅001, d P = 0⋅001 (ANOVA). *P < 0⋅001, †P < 0⋅010, ‡P < 0⋅050 (adjusted values on binary logistic regression)
representation of data. SPSS® version 20 was used for statistical analysis (IBM, Armonk, New York, USA).
MicroRNA validation and prediction algorithms
were used for predictions of mRNA targets. Owing to the variability in prediction tools and potential inaccuracies between algorithms, only gene targets predicted by five or more algorithms were included, to improve specificity.
MicroRNA validated targets and target prediction was performed using the miRWalk server that enables retrieval of genes predicted by ten or fewer algorithms29 . The algorithms Diana-microT, miRANDA, miRDB, miRWalk, RNAhybrid, PICTAR5, PITA, RNA22 and TargetScan
Results
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
Discovery study The initial discovery study was conducted in whole-blood samples from 15 men with an AAA and ten control subjects, BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
and had been identified in the literature as associated with AAA (miR-15a23 , miR-29b23,30,31 , miR-13832 and miR-181a22 ), and a further four miRNAs previously associated with AAA or vascular smooth muscle cell pathophysiology (miR-3123,33 , miR-14534,35 , miR-15523 and miR-22123,36,37 ). The selected miRNAs were validated individually in the whole blood and plasma from a second sample set of 200 men including 40 controls, 40 with PAD, 40 with a small AAA (30–54 mm), 40 with a large AAA (over 54 mm) and 40 men following AAA repair. Eighteen of the 40 postoperative samples were matched (collected from the same men as in the large AAA cohort). Aortic tissue samples were also collected from 22 men undergoing open AAA repair and 17 controls, as well as three men with aortoiliac occlusive disease.
8
6
ΔCT
4
2
0
–2 2·0
4·0
6·0
8·0
10·0
Aortic diameter (cm)
Mean comparative threshold cycle (ΔCT) values for let-7e in whole-blood samples in relation to aneurysm size for all 80 preoperative subjects with an abdominal aortic aneurysm. R2 linear = 0⋅081
Fig. 4
examining the expression of 754 miRNAs. Analysis of previous medical history revealed that the men with an AAA were older and more likely to be taking a statin (Table S1, supporting information). Some 395 miRNAs were excluded from further analysis as the qPCR curves were suboptimal (miRNAs not confidently detected). Analysis of the remaining 359 identified 29 miRNAs differentially expressed in men with an AAA compared with controls, with P < 0⋅100 and FC greater than 1⋅5 (Fig. 1, Table 1). P values and FCs for all miRNAs analysed are provided in Table S2 (supporting information).
Validation study For the validation study, data from the discovery study were combined with information from the literature to determine the most relevant miRNAs to take forward. The reliability of the expression of miRNAs found to be differentially expressed in the discovery cohort (defined as the number of samples expressing that miRNA) and their CT values were examined to select the optimum miRNAs for further analysis (let-7e, miR-125b, miR-27b, miR-196b, miR-411, miR-431, miR-483-5p and miR-589). Five miRNAs did not produce PCR curves for all samples (miR-672, miR-125a-3p, miR-134, miR-624 and miR-628-5p) and were therefore excluded. In addition, four miRNAs were selected that were significant in the discovery study © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
Whole-blood analysis Comparison of cardiovascular risk factors revealed that men with an AAA were significantly older, more likely to be smokers, and to have a history of hypertension and hypercholesterolaemia than healthy controls. Comparison of medication use revealed a significant difference in the use of aspirin, statins, clopidogrel, warfarin and diuretics, as would be expected in men with increased cardiovascular risk (Table S3, supporting information). Owing to these significant differences, binary logistic regression was used for further analysis, which yielded adjusted P values, to take these confounding factors into consideration. Initial one-way ANOVA was conducted to compare all 80 men with an AAA, 40 men with PAD and 40 controls. A significant difference was identified between the cohorts in let-7e (P = 0⋅001), miR-15a (P < 0⋅001), miR-196b (P < 0⋅001) and miR-411 (P = 0⋅001) (Fig. 2; Table S4, supporting information). No significant difference was identified for miR-27b, miR-29b, miR-31, miR-125b, miR-138, miR-145, miR-155, miR-181a, miR-221, miR-431, miR-483-5p and miR-589. For each miRNA identified as differentially expressed, further analyses were undertaken to determine whether there was a significant difference between any of the individual cohorts, and, in addition, whether there was any difference before and after surgical repair (open or endovascular). Analysis revealed a significant dysregulation of these miRNAs in men with an AAA compared with controls (FC −1⋅80, −2⋅24, −2⋅26 and 5⋅90 for let-7e, miR-15a, miR-196b and miR-411 respectively) (Fig. 2). P values were all 0⋅001 or less (adjusted P = 0⋅003–0⋅079); miR-411 was not significant on binary logistic regression (Fig. 3). There was also a significant dysregulation of each miRNA in men with PAD (FC −1⋅71, −2⋅19, −2⋅11 and 4⋅52 respectively), www.bjs.co.uk
BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
25 20 15
Fold change
10 5
∗
0 –5 –10 –15 –20 –25
8
9
-13
-58
miR
5b
miR
a
b -27
-12 miR
miR
1
-15 miR
5 -14
-43 miR
miR
1
-7e
-22
Let
1a
1
-31
miR
miR
-18 miR
p
-41
3-5
miR
b
5 -15
-48 miR
miR
-29 miR
miR
-19
6b
–30
Plasma validation results showing the mean(s.d.) fold change for each microRNA for 72 participants with an abdominal aortic aneurysm compared with 28 healthy controls. *P = 0⋅010 (one-way ANOVA)
Fig. 5
Plasma analysis For the validation study, plasma samples used were from the same men as in the whole-blood analysis. Thirty samples showed poor qPCR curves and were therefore excluded. Comparison of cardiovascular risk factors revealed a significant difference in age but not smoking status between the cohorts, and in previous medical history of hypertension and hypercholesterolaemia, and the use of aspirin, © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
10
5 ΔCT
but no difference between PAD and AAA cohorts. There was a significant difference in miRNA expression between men with previous AAA repair and controls for miR-15a (FC −1⋅80), miR-196b (FC −1⋅74) and miR-411 (FC 9⋅30). There was, however, no significant difference between preoperative and postoperative intervention cohorts, which was maintained when matched samples only were analysed, or on subgroup analysis of open and endovascular repair cohorts. A significant correlation between miRNA expression and aortic diameter was found for let-7e (R2 = 0⋅081, P = 0⋅011) (Fig. 4), but not for miR-15a (P = 0⋅834), miR-196b (P = 0⋅724) or miR-411 (P = 0⋅450).
0
–5
Control
AAA
PAD
Postop.
Fig. 6 Mean(s.d.) comparative threshold cycle (ΔCT) values for miR-196b in plasma samples from 72 patients with an abdominal aortic aneurysm (AAA), 35 with peripheral arterial disease (PAD), 35 postoperative patients who had undergone AAA repair and 28 control subjects. There were no significant differences on binary logistic regression
statins, warfarin and diuretics (Table S5, supporting information). Binary logistic regression was again used to take these confounding factors into consideration. www.bjs.co.uk
BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
Vascular cell adhesion molecule 1
α1-Antitrypsin
C-reactive protein
Interleukin 6 miR-15a
Let-7e
DAB2 interacting protein
Osteoprotegerin
Interferon
Low-density lipoprotein receptor-related protein 1
miR-196b
Low-density lipoprotein receptor
Methylenetetrahydrofolate reductase
Tumour necrosis factor
miR-411
Intercellular adhesion molecule 1
Predicted interaction Validated interaction
Diagram showing validated and predicted interactions between the microRNAs differentially expressed in blood in patients with an abdominal aortic aneurysm (AAA), and known genes and biomarkers found to be differentially expressed in subjects with an AAA
Fig. 7
One-way ANOVA revealed a significant difference between all cohorts for miR-196b only (P = 0⋅010) (Fig. 5; Table S6, supporting information). No significant difference was identified for let-7e, miR-15a, miR-27b, miR-29b, miR-31, miR-125b, miR-138, miR-145, miR-155, miR-181a, miR-221, miR-411, miR-431, miR-483-5p and miR-589. There was significant downregulation of miR-196b in men with an AAA compared with controls (FC −3⋅75; P = 0⋅029; adjusted P = 0⋅115), although this was not maintained on binary logistic regression (Fig. 6). There was also a significant downregulation of miR-196b in men with PAD compared with controls, although again this was not maintained on binary logistic regression (FC −5⋅15; P = 0⋅014; adjusted P = 0⋅219). There was no significant difference in miR-196b expression between AAA and PAD cohorts (adjusted P = 0⋅567), nor between men with previous AAA repair and controls (adjusted P = 0⋅814), whether open or endovascular. There was also no significant difference between preoperative and postoperative miR-196b expression in comparison of either all samples (adjusted P = 0⋅236) or matched samples (P = 0⋅146); subgroup analysis between open and endovascular samples revealed no significant difference. There was no correlation between miR-196b expression and aortic diameter (R2 = 0⋅003, P = 0⋅635). © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
Aortic tissue results For the initial validation study, aortic tissue was collected from 19 men with an AAA and ten cadaveric donors. For the additional validation study, aortic tissue was collected from 22 men having AAA repair, three men undergoing aortobifemoral bypass surgery for atherosclerosis, and 17 cadaveric donors. All samples showed good PCR curves and were therefore used for the analysis. Background data for both the initial and additional aortic tissue cohorts are summarized in Tables S7 and S8 (supporting information). For both the initial and additional aortic tissue cohorts, no significant difference was identified for any miRNA (Tables S9 and S10, supporting information).
Interactions between microRNAs identified and abdominal aortic aneurysm-related genes The miRWalk server was used to identify both validated and predicted interactions between let-7e, miR-15a, miR-196b and miR-411 and AAA biomarkers (MMP-2, MMP-9, TIMP-1, IL-6, TNF-α, osteoprotegerin, osteopontin, IFN-γ, ICAM-1, VCAM-1, D-dimer, CRP, α1-antitrypsin, fibrinogen, triglycerides, lipoprotein(a), apoprotein A, high-density lipoprotein, DAB2 interacting protein (DAB2IP), low-density lipoprotein receptor-related protein (LRP) 1, sortilin 1, low-density www.bjs.co.uk
BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
lipoprotein receptor (LDLR), angiotensin-converting enzyme, methylenetetrahydrofolate reductase (MTHFR)). Several validated interactions were identified with let-7e (IL-6, IFN, TNF-α), miR-15a (IL-6) and miR-196b (IFN). There were also several predicted interactions with let-7e (VCAM-1, CRP, osteoprotegerin, MTHFR), miR-15a (TNF-α, α1-antitrypsin, osteoprotegerin, CRP, MTHFR, DAB2IP), miR-196b (TNF-α, osteoprotegerin, LRP-1, MTHFR) and miR-411 (TNF-α, ICAM-1, osteoprotegerin, LDLR, MTHFR) (Fig. 7). Discussion
A discovery and validation study into miRNAs in AAA was undertaken. Significant upregulation of 29 miRNAs was identified in the discovery study, four of which were validated in blood (let-7e, miR-15a, miR-196b, miR-411), and one (miR-196b) further validated in plasma. Although no miRNA dysregulation was found in aortic tissue, all four miRNAs validated in blood were found to have significant interactions with previously identified AAA biomarkers15,16 . All four miRNAs were similarly dysregulated in patients with PAD, indicating that this dysregulation may be due to atherosclerosis rather than specifically to AAA. This study cannot be used to provide a biomarker for AAA, but it is possible to conclude that the miRNAs studied here were not altered by surgical intervention, and were not sufficiently correlated with aortic diameter to be useful. In addition, the miRNAs were not specific to AAA as they were also dysregulated in patients with PAD. The results of the validation study in whole-blood samples revealed three miRNAs to be significantly downregulated in patients with AAA compared with controls (let-7e, miR-15a, miR-196b), and one miRNA (miR-411) to be significantly upregulated; all except miR-411 remained significant when the analysis was corrected for confounding factors including age, smoking status, medication and cardiovascular risk factors. MicroRNAs miR-15a, miR-196b and miR-411 were also found to be differentially expressed in patients after AAA repair compared with controls, with miR-196b and miR-411 remaining significant after correction for confounding factors. No differences were found in miRNA expression between preoperative and postoperative samples from the same men, which suggested that exclusion of the aneurysm sac did not alter miRNA expression. Additionally, let-7e levels were significantly, but weakly, correlated with aneurysm diameter. Although these results were significant, it should be noted that both miR-411 and let-7e were dysregulated in the opposite direction to the findings in the discovery study. © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
An important aim of this study was to determine whether the miRNAs found to be differentially expressed in men with an AAA were specific to AAA, or were a feature of generalized atherosclerosis, as it has been hypothesised previously that these conditions share similar aetiology. Analysis of the validation cohort showed the differentially expressed miRNAs to be dysregulated in patients with PAD also, giving further weight to these conditions having a similar genetic basis. The reason why certain individuals develop atherosclerotic disease and others develop AAA remains to be elucidated. A single previous study38 showed let-7e to be associated with PAD; miR-15a, miR-196b and miR-411 have not previously been identified in PAD. The results from the plasma study reinforce the likelihood that miR-196b plays a role in the pathogenesis of AAA, but they also give further weight to the theory that AAA and PAD share a similar aetiology. The validation study in aortic tissue samples revealed no significant difference in expression levels of any of the miRNAs. This may be due to a type II error caused by an insufficient number of samples; however, as these results have been replicated in two independent cohorts it is likely that there is no correlation between circulating miRNA expression and aortic tissue expression of these miRNAs. Several validated and predicted interactions between the miRNAs found to be differentially expressed in whole blood in subjects with AAA and known genes or biomarkers of AAA have been identified. This highlights the plausibility of let-7e, miR-15a, miR-196b and miR-411 being dysregulated in subjects with AAA; however, these genes and biomarkers are not specific to AAA. Therefore, although these interactions add credibility to the miRNAs identified, the results should be viewed with caution, as the majority of these genes/biomarkers are also dysregulated in subjects with generalized atherosclerosis. To date, seven studies28,35,37,39 – 42 have investigated miRNA expression in animals, identifying significant dysregulation of miR-19a, miR-19b, miR-21, miR-24, miR-26a, miR-29b, miR-132, miR-143, miR-145, miR-195 and miR-222 in AAA. The first human studies40 – 42 of miRNA expression in AAA validated the animal models, revealing that miR-29b and miR-24 are significantly downregulated in aortic tissue, whereas miR-21 is significantly upregulated in AAA tissue. Recently, two miRNome studies have revealed further insights into aneurysm genesis. Pahl and co-workers22 looked at the expression of 847 miRNAs in five AAA and five control aortic tissue samples, identifying upregulation of miR-181*, miR-146 and miR-21, and downregulation of miR-133b, miR-133a, miR-331-3p, miR-30c-2* and miR-204. Validation work was undertaken using 25 elective AAA samples, www.bjs.co.uk
BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
11 ruptured AAA samples and seven control samples, which replicated the downregulated results, but not the upregulated ones. Kin and colleagues23 also used a miRNome approach, analysing the expression of 384 miRNAs in aortic tissue from 13 AAA samples and seven control samples. Initial microarray analysis identified 77 upregulated miRNAs, and validation work was carried out on a subset of these. A significant difference was found in the levels of let-7f, miR-21, miR-20a, miR-27a, miR-92a, miR-126, miR-221, miR-222, miR-29b, miR-124a, miR-146a, miR-155 and miR-223. Kin et al.23 then looked at miRNA expression in plasma samples from 23 patients with AAA, 17 with coronary artery disease and 12 controls; miR-126, miR-124a, miR-146a, miR-155, miR-223, miR-29b, miR-15a and miR-15b were significantly downregulated in AAA compared with controls. In addition, miR-124a, miR-155, miR-223 and miR-29b were downregulated in AAA compared with coronary artery disease. The present study was able to replicate these results with regard to miR-15a, but none of the other miRNAs identified above. This was possibly due to a difference in the tissue type studied: miR-29b, miR-31, miR-138, miR-145, miR-155, miR-181a and miR-221 were examined in whole blood, plasma and aortic tissue in the present study. A further study by Biros and co-workers27 looked at 124 miRNAs in aortic tissue and serum samples in ten patients undergoing AAA repair and ten controls. They identified differential expression of miR-155 in aortic aneurysm tissue compared with aortic neck samples, and also in the circulation, as found by Kin and colleagues23 but not in the present study. In addition, differential expression of miR-28, miR-30a-3p, miR-150, miR-302b*, miR-93 and miR-99a was identified. Although this study examined a wide range of miRNAs, it was possible to take only a subset of those identified in the discovery study through to validation. Future work, therefore, should initially evaluate the expression of these remaining miRNAs in blood, plasma and aortic tissue samples. In addition, these positive findings should be taken forward into experimental models to determine whether dysregulation of the miRNAs in cell culture and in vivo models alters expression of AAA-related biomarkers such as MMP-9, and whether their dysregulation leads to AAA development in animal models. Contributors
N. Sylvius (Department of Genetics, University of Leicester), D. A. Sidloff (Department of Cardiovascular Sciences, University of Leicester), N. Dattani © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
(Department of Cardiovascular Sciences, University of Leicester), A. Verissimo (Department of Cardiovascular Sciences, University of Leicester), J. B. Wild (Department of Cardiovascular Sciences, University of Leicester), H. Z. Butt (Department of Cardiovascular Sciences, University of Leicester), E. Choke (Department of Cardiovascular Sciences, University of Leicester), R. D. Sayers (Department of Cardiovascular Sciences, University of Leicester) and M. J. Bown (Department of Cardiovascular Sciences, University of Leicester and National Institute for Health Research (NIHR) Leicester Cardiovascular Biomedical Research Unit). Acknowledgements
This work was supported by a donation in memory of Bill Fowler and Tom Phillips. In addition, P.W.S. is funded by a Royal College of Surgeons/Dunhill Medical Trust Research Fellowship and a donation from Leicester NIHR Biomedical Research Unit. M.J.B. is funded by a Higher Education Funding Council for England Clinical Senior Lecturer Fellowship, the Leicester NIHR Cardiovascular Biomedical Research Unit and the Circulation Foundation President’s Early Career Award. Disclosure: The authors declare no conflict of interest. References 1 Kuivaniemi H, Shibamura H, Arthur C, Berguer R, Cole CW, Juvonen T et al. Familial abdominal aortic aneurysms: collection of 233 multiplex families. J Vasc Surg 2003; 37: 340–345. 2 Sakalihasan N, Limet R, Defawe OD. Abdominal aortic aneurysm. Lancet 2005; 365: 1577–1589. 3 Wahlgren CM, Larsson E, Magnusson PK, Hultgren R, Swedenborg J. Genetic and environmental contributions to abdominal aortic aneurysm development in a twin population. J Vasc Surg 2010; 51: 3–7. 4 Sandiford P, Mosquera D, Bramley D. Ethnic inequalities in incidence, survival and mortality from abdominal aortic aneurysm in New Zealand. J Epidemiol Community Health 2012; 66: 1097–1103. 5 Rossaak JI, Sporle A, Birks CL, van Rij AM. Abdominal aortic aneurysms in the New Zealand Maori population. Br J Surg 2003; 90: 1361–1366. 6 Sweeting MJ, Thompson SG, Brown LC, Powell JT; RESCAN collaborators. Meta-analysis of individual patient data to examine factors affecting growth and rupture of small abdominal aortic aneurysms. Br J Surg 2012; 99: 655–665. 7 Giugliano G, Laurenzano E, Rengo C, De Rosa G, Brevetti L, Sannino A et al. Abdominal aortic aneurysm in patients affected by intermittent claudication: prevalence and clinical predictors. BMC Surg 2012; 12(Suppl 1): S17.
www.bjs.co.uk
BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
8 Cornuz J, Sidoti Pinto C, Tevaearai H, Egger M. Risk factors for asymptomatic abdominal aortic aneurysm: systematic review and meta-analysis of population-based screening studies. Eur J Public Health 2004; 14: 343–349. 9 Golledge J, Norman PE. Atherosclerosis and abdominal aortic aneurysm: cause, response, or common risk factors? Arterioscler Thromb Vasc Biol 2010; 30: 1075–1077. 10 Gretarsdottir S, Baas AF, Thorleifsson G, Holm H, den Heijer M, de Vries JP et al. Genome-wide association study identifies a sequence variant within the DAB2IP gene conferring susceptibility to abdominal aortic aneurysm. Nat Genet 2010; 42: 692–697. 11 Jones GT, Bown MJ, Gretarsdottir S, Romaine SP, Helgadottir A, Yu G et al. A sequence variant associated with sortilin-1 (SORT1) on 1p13.3 is independently associated with abdominal aortic aneurysm. Hum Mol Genet 2013; 22: 2941–2947. 12 Bown MJ, Jones GT, Harrison SC, Wright BJ, Bumpstead S, Baas AF et al. Abdominal aortic aneurysm is associated with a variant in low-density lipoprotein receptor-related protein 1. Am J Hum Genet 2011; 89: 619–627. 13 Bradley DT, Hughes AE, Badger SA, Jones GT, Harrison SC, Wright BJ et al. A variant in LDLR is associated with abdominal aortic aneurysm. Circ Cardiovasc Genet 2013; 6: 498–504. 14 Thompson AR, Drenos F, Hafez H, Humphries SE. Candidate gene association studies in abdominal aortic aneurysm disease: a review and meta-analysis. Eur J Vasc Endovasc Surg 2008; 35: 19–30. 15 Sidloff DA, Stather PW, Choke E, Bown MJ, Sayers RD. A systematic review and meta-analysis of the association between markers of hemostasis and abdominal aortic aneurysm presence and size. J Vasc Surg 2014; 59: 528–535.e4. 16 Stather PW, Sidloff DA, Dattani N, Gokani VJ, Choke E, Sayers RD et al. Meta-analysis and meta-regression analysis of biomarkers for abdominal aortic aneurysm. Br J Surg 2014; 101: 1358–1372. 17 Lagos-Quintana M, Rauhut R, Yalcin A, Meyer J, Lendeckel W, Tuschl T. Identification of tissue-specific microRNAs from mouse. Curr Biol 2002; 12: 735–739. 18 Liu CG, Calin GA, Meloon B, Gamliel N, Sevignani C, Ferracin M et al. An oligonucleotide microchip for genome-wide microRNA profiling in human and mouse tissues. Proc Natl Acad Sci U S A 2004; 101: 9740–9744. 19 Tang Y, Liu D, Zhang L, Ingvarsson S, Chen H. Quantitative analysis of miRNA expression in seven human foetal and adult organs. PLoS One 2011; 6: e28730. 20 Jopling CL, Norman KL, Sarnow P. Positive and negative modulation of viral and cellular mRNAs by liver-specific microRNA miR-122. Cold Spring Harb Symp Quant Biol 2006; 71: 369–376. 21 Bostjancic E, Zidar N, Stajer D, Glavac D. MicroRNAs miR-1, miR-133a, miR-133b and miR-208 are dysregulated
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
22
23
24
25
26
27
28
29
30
31
32
33
34
35
in human myocardial infarction. Cardiology 2010; 115: 163–169. Pahl MC, Derr K, Gäbel G, Hinterseher I, Elmore JR, Schworer CM et al. MicroRNA expression signature in human abdominal aortic aneurysms. BMC Med Genomics 2012; 5: 25. Kin K, Miyagawa S, Fukushima S, Shirakawa Y, Torikai K, Shimamura K et al. Tissue- and plasma-specific microRNA signatures for atherosclerotic abdominal aortic aneurysm. J Am Heart Assoc 2012; 1: e000745. Maegdefessel L, Dalman RL, Tsao PS. Pathogenesis of abdominal aortic aneurysms: microRNAs, proteases, genetic associations. Annu Rev Med 2014; 65: 49–62. Stather P, Wild J, Sylvius N, Choke E, Sayers R, Bown M. MicroRNAs associated with the pathways involved in the pathogenesis of abdominal aortic aneurysms. Artery Research 2012; 7: 28–35. Adam M, Raaz U, Spin JM, Tsao PS. MicroRNAs in abdominal aortic aneurysm. Curr Vasc Pharmacol 2013; [Epub ahead of print]. Biros E, Moran CS, Wang Y, Walker PJ, Cardinal J, Golledge J. MicroRNA profiling in patients with abdominal aortic aneurysms: the significance of miR-155. Clin Sci (Lond) 2014; 126: 795–803. Zampetaki A, Attia R, Mayr U, Gomes RS, Phinikaridou A, Yin X et al. Role of miR-195 in aortic aneurysmal disease. Circ Res 2014; 115: 857–866. Dweep H, Sticht C, Pandey P, Gretz N. miRWalk – database: prediction of possible miRNA binding sites by ‘walking’ the genes of three genomes. J Biomed Inform 2011; 44: 839–847. Boon RA, Seeger T, Heydt S, Fischer A, Hergenreider E, Horrevoets AJ et al. MicroRNA-29 in aortic dilation: implications for aneurysm formation. Circ Res 2011; 109: 1115–1119. Maegdefessel L, Azuma J, Toh R, Merk DR, Deng A, Chin JT et al. Inhibition of microRNA-29b reduces murine abdominal aortic aneurysm development. J Clin Invest 2012; 122: 497–506. Lee ST, Chu K, Im WS, Yoon HJ, Im JY, Park JE et al. Altered microRNA regulation in Huntington’s disease models. Exp Neurol 2011; 227: 172–179. Liu X, Cheng Y, Chen X, Yang J, Xu L, Zhang C. MicroRNA-31 regulated by the extracellular regulated kinase is involved in vascular smooth muscle cell growth via large tumor suppressor homolog 2. J Biol Chem 2011; 286: 42 371–42 380. Cheng Y, Liu X, Yang J, Lin Y, Xu DZ, Lu Q et al. MicroRNA-145, a novel smooth muscle cell phenotypic marker and modulator, controls vascular neointimal lesion formation. Circ Res 2009; 105: 158–166. Damrauer SM, Csizmadia E, Studer P, Bhasin M, Da Silva CG, Peterson CR et al. Partial loss of A20 (Tnfaip3) promotes resistance to abdominal aortic aneurysms. ATVB 2010; 30: e317.
www.bjs.co.uk
BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
36 Davis BN, Hilyard AC, Nguyen PH, Lagna G, Hata A. Induction of microRNA-221 by platelet-derived growth factor signaling is critical for modulation of vascular smooth muscle phenotype. J Biol Chem 2009; 284: 3728–3738. 37 Liu G, Huang Y, Lu X, Lu M, Huang X, Li W et al. Identification and characteristics of microRNAs with altered expression patterns in a rat model of abdominal aortic aneurysms. Tohoku J Exp Med 2010; 222: 187–193. 38 Stather PW, Sylvius N, Wild JB, Choke E, Sayers RD, Bown MJ. Differential microRNA expression profiles in peripheral arterial disease. Circ Cardiovasc Genet 2013; 6: 490–497. 39 Leeper NJ, Raiesdana A, Kojima Y, Chun HJ, Azuma J, Maegdefessel L et al. MicroRNA-26a is a novel regulator of
vascular smooth muscle cell function. J Cell Physiol 2011; 226: 1035–1043. 40 Maegdefessel L, Azuma J, Toh R, Merk DR, Deng A, Chin JT et al. Inhibition of microRNA-29b reduces murine abdominal aortic aneurysm development. J Clin Invest 2012; 122: 497–506. 41 Maegdefessel L, Azuma J, Toh R, Deng A, Merk DR, Raiesdana A et al. MicroRNA-21 blocks abdominal aortic aneurysm development and nicotine-augmented expansion. Sci Transl Med 2012; 4: 122ra22. 42 Maegdefessel L, Raaz U, Toh R, Deng A, Chernogubova E, Eriksson P et al. MicroRNA-24 regulates inflammation and calcification during abdominal aortic aneurysm development through targeting Ykl-40. Circ Suppl 2012; 126: A14221.
Supporting information
Additional supporting information may be found in the online version of this article: Appendix S1 RNA isolation and microRNA expression profiling methods (Word document) Table S1 Background comparison between control patients and those with abdominal aortic aneursym included in discovery study (Word document) Table S2 Fold change in all microRNAs expressed in whole-blood samples from discovery study (Word document) Table S3 Background analysis of subjects included in whole-blood sample validation study (Word document) Table S4 Results of one-way ANOVA comparing relative microRNA expression in whole-blood samples between all 80 subjects with an abdominal aortic aneurysm, 40 with peripheral arterial disease and 40 controls (Word document) Table S5 Background analysis of subjects included in plasma sample validation study (Word document) Table S6 Results of one-way ANOVA comparing relative microRNA expression in plasma samples between all 72 subjects with an abdominal aortic aneurysm, 35 with peripheral arterial disease and 28 controls (Word document) Table S7 Background analysis of subjects included in initial aortic tissue study (Word document) Table S8 Background analysis of subjects included in additional aortic tissue study (Word document) Table S9 Comparison of microRNA expression in aortic tissue from 19 abdominal aortic aneurysm samples and samples from ten cadaveric donors (Word document) Table S10 Comparison of microRNA expression between abdominal aortic aneurym, peripheral arterial disease and cadveric donor cohorts (Word document)
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS
Identification of microRNAs associated with abdominal aortic aneurysms and peripheral arterial disease P. W. Stather1 , N. Sylvius2 , D. A. Sidloff1 , N. Dattani1 , A. Verissimo1 , J. B. Wild1 , H. Z. Butt1 , E. Choke1 , R. D. Sayers1 and M. J. Bown1,3∗ Departments of 1 Cardiovascular Sciences and 2 Genetics and 3 National Institute for Health Research, University of Leicester, Leicester, UK Correspondence to: Mr P. W. Stather, Department of Cardiovascular Sciences, University of Leicester, Leicester LE2 7LX, UK (e-mail: [email protected])
Background: MicroRNAs are crucial in the regulation of cardiovascular disease and represent potential
therapeutic targets to decrease abdominal aortic aneurysm (AAA) expansion. The aim of this study was to identify circulating microRNAs associated with AAA. Methods: Some 754 microRNAs in whole-blood samples from 15 men with an AAA and ten control subjects were quantified using quantitative reverse transcriptase–PCR. MicroRNAs demonstrating a significant association with AAA were validated in peripheral blood and plasma samples of men in the following groups (40 in each): healthy controls, controls with peripheral arterial disease (PAD), men with a small AAA (30–54 mm), those with a large AAA (over 54 mm), and those following AAA repair. MicroRNA expression was also assessed in aortic tissue. Results: Twenty-nine differentially expressed microRNAs were identified in the discovery study. Validation study revealed that let-7e (fold change (FC) –1⋅80; P = 0⋅001), miR-15a (FC −2⋅24; P < 0⋅001) and miR-196b (FC −2⋅26; P < 0⋅001) were downregulated in peripheral blood from patients with an AAA, and miR-411 was upregulated (FC 5⋅90; P = 0⋅001). miR-196b was also downregulated in plasma from the same individuals (FC −3⋅75; P = 0⋅029). The same miRNAs were similarly expressed differentially in patients with PAD compared with healthy controls. Validated and predicted microRNA targets identified through miRWalk revealed that these miRNAs were all regulators of AAA-related genes (vascular cell adhesion molecule 1, intercellular cell adhesion molecule 1, DAB2 interacting protein, 𝛂1-antitrypsin, C-reactive protein, interleukin 6, osteoprotegerin, methylenetetrahydrofolate reductase, tumour necrosis factor 𝛂). Conclusion: In this study, circulating levels of let-7e, miR-15a, miR-196b and miR-411 were differentially expressed in men with an AAA compared with healthy controls, but also differentially expressed in men with PAD. Modulation of these miRNAs and their target genes may represent a new therapeutic pathway to affect the progression of AAA and atherosclerosis. ∗ Co-authors of this study can be found under the heading Contributors Presented to a meeting of the Vascular Society of Great Britain and Ireland, Manchester, UK, November 2013; published in abstract form as Br J Surg 2014; 101(Suppl 2): 1–2
Paper accepted 9 February 2015 Published online in Wiley Online Library (www.bjs.co.uk). DOI: 10.1002/bjs.9802
Introduction
The underlying pathophysiology of abdominal aortic aneurysm (AAA) is unclear. Smoking is the most important environmental risk factor for the development of AAA; however, there is a well established genetic component as evidenced by familial1 – 3 and ethnicity4,5 studies. AAA is more common in men, increases with age and hypercholesterolaemia, and is less frequent in patients © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
with diabetes6 . Furthermore, the prevalence of AAA in patients with peripheral arterial disease (PAD) is reported to be significantly higher (9 per cent) than in the normal population7 . Meta-analysis8 of eight studies identified an increased risk of AAA in patients with PAD. AAA and PAD share many similarities, including risk factors such as smoking and hypertension, biomarkers such as fibrinogen and C-reactive protein (CRP), and genetic associations BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
such as LDLR and SORT1. However, diabetes is a negative indicator for AAA but strongly associated with PAD, male sex is strongly associated with AAA, and some recognized genetic determinants of atherosclerosis, such as Apo E polymorphisms, have no association with AAA9 . Four genetic loci have been identified through genome-wide association studies (DAB2IP, LRP1, SORT1 and LDLR)10 – 13 , and three candidate single-nucleotide polymorphisms have been identified as associated with AAA from meta-analysis (MMP9, ACE, MTHFR)14 . In addition, a wide range of potential biomarkers for AAA have been identified, with two recent meta-analyses15,16 reporting a significant association between AAA and 18 proteins (matrix metalloproteinase (MMP) 2, MMP-9, tissue inhibitor of metalloproteinase (TIMP) 1, interleukin (IL) 6, tumour necrosis factor (TNF) α, osteoprotegerin, osteopontin, interferon (IFN) γ, intercellular adhesion molecule (ICAM) 1, vascular cell adhesion molecule (VCAM) 1, D-dimer, CRP, α1-antitrypsin, fibrinogen, triglycerides, lipoprotein(a), apolipoprotein A and high-density lipoprotein). The specificity and sensitivity of these biomarkers is, however, poor with all reporter operator characteristics below 0⋅7; therefore, a different approach must be taken to understand further the underlying pathophysiology of AAA. MicroRNAs (miRNAs) are short, non-coding RNAs involved in gene regulation by binding to mRNAs, causing repression of translation and mRNA degradation, thus fine-tuning protein expression. Each single miRNA is capable of interacting with several different mRNAs, thereby affecting multiple pathways. There is a clear difference in miRNA signature between cell and tissue types. Several studies have looked at the abundance of miRNAs in specific tissues, identifying distinct patterns of miRNA expression in human skeletal muscle, heart, prostate, brain, liver, spleen and colon17 – 19 . Specific miRNAs are also found only in certain tissues, such as miR-122 in liver20 , and miR-1, miR-208 and miR-133 in muscle or heart21 . This is of particular use in acute conditions such as myocardial infarction, where heart-specific miRNAs are released into the circulation. Therefore, miRNAs may be specific to aortic aneurysm tissue or increased in aortic cells, leading to altered expression profiles. Two studies22,23 have assessed miRNA expression in human AAA tissue using a miRNA transcriptome (miRNome) approach, with conflicting results. In addition, several other studies and reviews24 – 28 have identified miR-21, miR-29b, miR-26a, miR-133, miR-145, miR-155, miR-195 and miR-221/222 as potential miRNAs associated with AAA in human and animal models. However, miRNA expression in patients with an AAA has not been
evaluated across tissue types, comparing whole blood, plasma and the aortic wall. This study therefore aimed to determine whether miRNAs can be used as circulating biomarkers for AAA by exploring whether any miRNAs are differentially expressed in whole blood, plasma and aortic tissue of patients with an AAA. A further aim was to determine whether these miRNAs remain dysregulated after surgical intervention for AAA, correlate with AAA diameter, or are specific to AAA, by assessing miRNA expression in subjects with PAD.
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
Methods
Ethical approval was obtained from Derby Research Ethics Committee for collection of blood and tissue samples (REC: 6819). Ethical approval was also obtained for cadaveric tissue samples (REC: 12/EM/0355). This study was performed in accordance with the declaration of Helsinki. Consent was obtained from all men before inclusion in the study. A two-stage study was conducted to determine which miRNAs were associated with AAA, and to establish whether these were specific to AAA or related to generalized atherosclerosis. An initial discovery study was performed quantifying 754 miRNAs in whole-blood samples taken from 15 men with a large AAA (over 54 mm) and ten healthy controls. MicroRNAs associated with AAA from the analysis of the discovery study were taken forward to a validation study of 200 men (40 in each group): healthy controls, controls with PAD, men with a small AAA (30–54 mm), those with a large AAA (over 54 mm) and men who had undergone AAA repair (sample collected 6 months after repair, both open and endovascular repair samples included; there were no postoperative complications). All 40 men in each cohort were included in the whole-blood study. For the plasma study 30 samples showed poor PCR curves, and were therefore excluded. This resulted in 28 healthy controls, 35 controls with PAD, 36 men with a small AAA, 36 with a large AAA, and 35 post-AAA repair. MicroRNAs were selected for the validation study based on a fold change (FC) threshold of more than 1⋅5 and a P value threshold of less than 0⋅100. A FC greater than 1 represented an upregulated miRNA, whereas a FC less than 1 represented a downregulated miRNA. A FC of 1 (or −1) was equivalent to no change in miRNA expression. In addition, miRNAs potentially associated with AAA taken from the literature were tested for association in the validation study. MicroRNAs included in the validation study were quantified in whole-blood samples, plasma samples and, BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
0 0·1 0·2 0·3 0·4 P
where patients with AAA were undergoing open surgery, AAA wall biopsies. Clinical information including medical and family history was obtained for all participants, including: age, smoking history, history of hypertension, hypercholesterolaemia, myocardial infarction, angina, stroke, diabetes, chronic obstructive pulmonary disease, peripheral vascular disease and any history of malignancy or autoimmune/inflammatory disease. A medication history was also taken, including: aspirin, beta-blockers, statins, clopidogrel, digoxin, warfarin, diuretics and angiotensin-converting enzyme inhibitors. All men underwent abdominal ultrasonography to measure aortic diameter using inner wall to inner wall measurements in both anteroposterior and lateral planes. Patients with a clinical history of PAD also had ultrasound imaging of the leg arteries to assess for arterial stenosis or occlusion. A small AAA was defined by a diameter of 30–54 mm, and large AAA by a diameter of more than 54 mm. The diagnosis of PAD was based on the presence of both symptoms of intermittent claudication and arterial stenosis/occlusion identified on ultrasound imaging. Whole blood and plasma were obtained from all men (except cadaveric donors). Aortic tissue samples were obtained from men undergoing open abdominal aortic surgery; samples were taken from the anterior aortic wall approximately 5 cm below the renal arteries. Samples were also taken from the infrarenal aorta of cadaveric donors (Appendix S1, supporting information). All aortic tissue samples had matched blood and plasma samples. RNA quality control was conducted using 260/280 ratios (only those with a ratio between 1⋅8 and 2⋅2 were included), RNA integrity numbers (RINs) and RNA concentration (more than 40 ng/μl for whole-blood and aortic tissue samples). RNA integrity was determined for a subgroup of samples, with a mean(s.d.) RIN of 8⋅60(0⋅81) in 11 whole-blood samples and 6⋅75(0⋅67) in six aortic tissue samples. These RIN results were consistently good, so RNA integrity was not assessed in all samples. Plasma samples lack a cellular component, and miRNAs are present owing to release from necrotic cells, apoptosis or intercellular communication. This results in a deficiency of large ribosomal RNAs, and as such the integrity of the RNA samples cannot be determined. Quality control for plasma samples was therefore done purely by ensuring the presence of RNA using a NanoDrop™ spectrophotometer (ThermoScientific, Waltham, Massachusetts, USA), with all samples requiring more than 1 ng/μl for use in subsequent processing. The following inclusion criteria were implemented in order to permit standardization and reduce bias when comparing cases and controls: men over 55 years; Caucasian;
0·5 0·6 0·7 0·8 0·9 1·0 –3
–2
–1 1
2
3
Fold change
Volcano plot showing the distribution of fold change and P value for each microRNA (miRNA) expression in the discovery study comparing 15 participants with a large adominal aortic aneurysm and ten healthy controls. The red dots represent the miRNAs that were significantly different; pink shaded areas represent a significant fold change. Thirteen miRNAs had a fold change greater than 3 or −3 (with 1 of these being significantly differentially expressed), and are therefore not included in this plot
Fig. 1
Differentially expressed microRNAs from the discovery study comparing 15 participants with large abdominal aortic aneurysm and ten healthy controls
Table 1
MicroRNA miR-672 miR-19a* miR-15a miR-16-1* miR-9* miR-125a-3p miR-196b miR-134 miR-181a miR-942 Let-7e miR-431 miR-1291 miR-27b miR-1249 miR-483-5p miR-624 miR-411 miR-589 miR-29b miR-1179 miR-1227 miR-138 miR-26a-1* miR-374a miR-125b miR-628-5p miR-148b* miR-15b*
Fold change −2⋅22 2⋅24 −2⋅04 −1⋅75 −2⋅22 69 866⋅67 −1⋅89 −2⋅63 1⋅62 −2⋅70 2⋅29 1⋅82 −1⋅85 2⋅48 −1⋅54 1⋅86 −2⋅04 −2⋅04 −1⋅51 −1⋅89 −2⋅00 1⋅70 2⋅22 −1⋅59 −1⋅69 1⋅56 −1⋅69 −1⋅52 −2⋅50
P† 0⋅003 0⋅006 0⋅019 0⋅019 0⋅024 0⋅031 0⋅031 0⋅036 0⋅041 0⋅041 0⋅041 0⋅042 0⋅048 0⋅051 0⋅055 0⋅055 0⋅057 0⋅058 0⋅068 0⋅076 0⋅081 0⋅081 0⋅081 0⋅081 0⋅081 0⋅091 0⋅091 0⋅096 0⋅096
MicroRNAs with P < 0⋅100 and more than 1⋅5-fold change were deemed to be statistically significant. †Mann–Whitney U test.
www.bjs.co.uk
BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
14 12 10 8 6
Fold change
4 2 0 –2
∗
–4 –6
∗
∗
∗
–8 –10 –12 –14
1
8
-41 miR
b
-13
-27
miR
-58
9 miR
5 -14 miR
miR
5
1 -43
-15 miR
1a -18
miR
miR
-22
1
-31
miR
b -29
e
5b
t-7
miR
miR
Le
p
-12
miR
3-5
a -15
-48 miR
miR
miR
-19
6b
–15
Whole-blood validation results showing the mean(s.d.) fold change for each microRNA for 80 participants with an abdominal aortic aneurysm compared with 40 healthy controls. *P ≤ 0⋅001 (one-way ANOVA)
Fig. 2
and no known systemic illness that could confound miRNA expression profiling, such as malignant, systemic autoimmune or inflammatory disease. Details of RNA isolation and miRNA expression profiling methods are provided in Appendix S1 (supporting information).
MicroRNA quantification For the discovery study, miRNome profiling was conducted by quantitative PCR (qPCR) using TaqMan® miRNA Cards A + B on an ABI 7900HT qPCR instrument (Life Technologies, Paisley, UK). FC was calculated using the relative standard curve method. For the validation study, selected miRNAs were profiled by digital PCR using TaqMan® OpenArray® MicroRNA assays on an OpenArray® Quantstudio™ 12 K Flex Real-Time PCR system (Life Technologies). FC in the validation study was calculated using the comparative threshold cycle (ΔCT) method. Data were normalized with respect to the geometric mean of RNU48 and MammU6. The expression curves of all miRNAs were inspected visually, and miRNAs with suboptimal amplification PCR curves (absence of clear S-shaped curves) were excluded. © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
Statistical analysis For the discovery study, background clinical data were analysed using Fisher’s exact and Mann–Whitney U tests; miRNA levels were compared between patients with an AAA and healthy controls by means of Mann–Whitney U test. For the validation study, an initial analysis was undertaken using ANOVA to determine any significant difference in miRNA expression in blood between controls, patients with an AAA and controls with PAD. Further post hoc analysis was then undertaken using t tests and binary logistic regression to determine whether significant differences were specific to AAA, or due to generalized atherosclerosis. Additional analysis was done to determine whether differences in miRNA expression were maintained after AAA repair, as this would determine whether miRNA expression is altered owing to strain on the aneurysm sac, or indicates an underlying predisposition to aneurysm formation. The correlation between aneurysm size and miRNA expression was also analysed by bivariable correlation. Aortic tissue miRNA expression was analysed using the Mann–Whitney U test and Kruskal–Wallis test. GraphPad Prism® 6 (GraphPad, La Jolla, California, USA) was used for graphical www.bjs.co.uk
BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
†
‡ ‡
‡ 6 10
ΔCT
ΔCT
4
5 2
0
a
Control
AAA
PAD
0
Postop.
b
Let-7e
Control
AAA
PAD
Postop.
PAD
Postop.
miR-15a
‡
‡
∗
†
† 15
10
ΔCT
ΔCT
10 5
5
0
0 Control
c
AAA
PAD
Control
Postop.
miR-196b
d
AAA
miR-411
Fig. 3 Mean(s.d.) comparative threshold cycle (ΔCT) values for a let-7e, b miR-15a, c miR-196b and d miR-411 in whole-blood samples from 80 patients with an abdominal aortic aneurysm (AAA), 40 with peripheral arterial disease (PAD), 40 postoperative patients who had undergone AAA repair and 40 control subjects. a P = 0⋅001, b P < 0⋅001, c P < 0⋅001, d P = 0⋅001 (ANOVA). *P < 0⋅001, †P < 0⋅010, ‡P < 0⋅050 (adjusted values on binary logistic regression)
representation of data. SPSS® version 20 was used for statistical analysis (IBM, Armonk, New York, USA).
MicroRNA validation and prediction algorithms
were used for predictions of mRNA targets. Owing to the variability in prediction tools and potential inaccuracies between algorithms, only gene targets predicted by five or more algorithms were included, to improve specificity.
MicroRNA validated targets and target prediction was performed using the miRWalk server that enables retrieval of genes predicted by ten or fewer algorithms29 . The algorithms Diana-microT, miRANDA, miRDB, miRWalk, RNAhybrid, PICTAR5, PITA, RNA22 and TargetScan
Results
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
Discovery study The initial discovery study was conducted in whole-blood samples from 15 men with an AAA and ten control subjects, BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
and had been identified in the literature as associated with AAA (miR-15a23 , miR-29b23,30,31 , miR-13832 and miR-181a22 ), and a further four miRNAs previously associated with AAA or vascular smooth muscle cell pathophysiology (miR-3123,33 , miR-14534,35 , miR-15523 and miR-22123,36,37 ). The selected miRNAs were validated individually in the whole blood and plasma from a second sample set of 200 men including 40 controls, 40 with PAD, 40 with a small AAA (30–54 mm), 40 with a large AAA (over 54 mm) and 40 men following AAA repair. Eighteen of the 40 postoperative samples were matched (collected from the same men as in the large AAA cohort). Aortic tissue samples were also collected from 22 men undergoing open AAA repair and 17 controls, as well as three men with aortoiliac occlusive disease.
8
6
ΔCT
4
2
0
–2 2·0
4·0
6·0
8·0
10·0
Aortic diameter (cm)
Mean comparative threshold cycle (ΔCT) values for let-7e in whole-blood samples in relation to aneurysm size for all 80 preoperative subjects with an abdominal aortic aneurysm. R2 linear = 0⋅081
Fig. 4
examining the expression of 754 miRNAs. Analysis of previous medical history revealed that the men with an AAA were older and more likely to be taking a statin (Table S1, supporting information). Some 395 miRNAs were excluded from further analysis as the qPCR curves were suboptimal (miRNAs not confidently detected). Analysis of the remaining 359 identified 29 miRNAs differentially expressed in men with an AAA compared with controls, with P < 0⋅100 and FC greater than 1⋅5 (Fig. 1, Table 1). P values and FCs for all miRNAs analysed are provided in Table S2 (supporting information).
Validation study For the validation study, data from the discovery study were combined with information from the literature to determine the most relevant miRNAs to take forward. The reliability of the expression of miRNAs found to be differentially expressed in the discovery cohort (defined as the number of samples expressing that miRNA) and their CT values were examined to select the optimum miRNAs for further analysis (let-7e, miR-125b, miR-27b, miR-196b, miR-411, miR-431, miR-483-5p and miR-589). Five miRNAs did not produce PCR curves for all samples (miR-672, miR-125a-3p, miR-134, miR-624 and miR-628-5p) and were therefore excluded. In addition, four miRNAs were selected that were significant in the discovery study © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
Whole-blood analysis Comparison of cardiovascular risk factors revealed that men with an AAA were significantly older, more likely to be smokers, and to have a history of hypertension and hypercholesterolaemia than healthy controls. Comparison of medication use revealed a significant difference in the use of aspirin, statins, clopidogrel, warfarin and diuretics, as would be expected in men with increased cardiovascular risk (Table S3, supporting information). Owing to these significant differences, binary logistic regression was used for further analysis, which yielded adjusted P values, to take these confounding factors into consideration. Initial one-way ANOVA was conducted to compare all 80 men with an AAA, 40 men with PAD and 40 controls. A significant difference was identified between the cohorts in let-7e (P = 0⋅001), miR-15a (P < 0⋅001), miR-196b (P < 0⋅001) and miR-411 (P = 0⋅001) (Fig. 2; Table S4, supporting information). No significant difference was identified for miR-27b, miR-29b, miR-31, miR-125b, miR-138, miR-145, miR-155, miR-181a, miR-221, miR-431, miR-483-5p and miR-589. For each miRNA identified as differentially expressed, further analyses were undertaken to determine whether there was a significant difference between any of the individual cohorts, and, in addition, whether there was any difference before and after surgical repair (open or endovascular). Analysis revealed a significant dysregulation of these miRNAs in men with an AAA compared with controls (FC −1⋅80, −2⋅24, −2⋅26 and 5⋅90 for let-7e, miR-15a, miR-196b and miR-411 respectively) (Fig. 2). P values were all 0⋅001 or less (adjusted P = 0⋅003–0⋅079); miR-411 was not significant on binary logistic regression (Fig. 3). There was also a significant dysregulation of each miRNA in men with PAD (FC −1⋅71, −2⋅19, −2⋅11 and 4⋅52 respectively), www.bjs.co.uk
BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
25 20 15
Fold change
10 5
∗
0 –5 –10 –15 –20 –25
8
9
-13
-58
miR
5b
miR
a
b -27
-12 miR
miR
1
-15 miR
5 -14
-43 miR
miR
1
-7e
-22
Let
1a
1
-31
miR
miR
-18 miR
p
-41
3-5
miR
b
5 -15
-48 miR
miR
-29 miR
miR
-19
6b
–30
Plasma validation results showing the mean(s.d.) fold change for each microRNA for 72 participants with an abdominal aortic aneurysm compared with 28 healthy controls. *P = 0⋅010 (one-way ANOVA)
Fig. 5
Plasma analysis For the validation study, plasma samples used were from the same men as in the whole-blood analysis. Thirty samples showed poor qPCR curves and were therefore excluded. Comparison of cardiovascular risk factors revealed a significant difference in age but not smoking status between the cohorts, and in previous medical history of hypertension and hypercholesterolaemia, and the use of aspirin, © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
10
5 ΔCT
but no difference between PAD and AAA cohorts. There was a significant difference in miRNA expression between men with previous AAA repair and controls for miR-15a (FC −1⋅80), miR-196b (FC −1⋅74) and miR-411 (FC 9⋅30). There was, however, no significant difference between preoperative and postoperative intervention cohorts, which was maintained when matched samples only were analysed, or on subgroup analysis of open and endovascular repair cohorts. A significant correlation between miRNA expression and aortic diameter was found for let-7e (R2 = 0⋅081, P = 0⋅011) (Fig. 4), but not for miR-15a (P = 0⋅834), miR-196b (P = 0⋅724) or miR-411 (P = 0⋅450).
0
–5
Control
AAA
PAD
Postop.
Fig. 6 Mean(s.d.) comparative threshold cycle (ΔCT) values for miR-196b in plasma samples from 72 patients with an abdominal aortic aneurysm (AAA), 35 with peripheral arterial disease (PAD), 35 postoperative patients who had undergone AAA repair and 28 control subjects. There were no significant differences on binary logistic regression
statins, warfarin and diuretics (Table S5, supporting information). Binary logistic regression was again used to take these confounding factors into consideration. www.bjs.co.uk
BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
Vascular cell adhesion molecule 1
α1-Antitrypsin
C-reactive protein
Interleukin 6 miR-15a
Let-7e
DAB2 interacting protein
Osteoprotegerin
Interferon
Low-density lipoprotein receptor-related protein 1
miR-196b
Low-density lipoprotein receptor
Methylenetetrahydrofolate reductase
Tumour necrosis factor
miR-411
Intercellular adhesion molecule 1
Predicted interaction Validated interaction
Diagram showing validated and predicted interactions between the microRNAs differentially expressed in blood in patients with an abdominal aortic aneurysm (AAA), and known genes and biomarkers found to be differentially expressed in subjects with an AAA
Fig. 7
One-way ANOVA revealed a significant difference between all cohorts for miR-196b only (P = 0⋅010) (Fig. 5; Table S6, supporting information). No significant difference was identified for let-7e, miR-15a, miR-27b, miR-29b, miR-31, miR-125b, miR-138, miR-145, miR-155, miR-181a, miR-221, miR-411, miR-431, miR-483-5p and miR-589. There was significant downregulation of miR-196b in men with an AAA compared with controls (FC −3⋅75; P = 0⋅029; adjusted P = 0⋅115), although this was not maintained on binary logistic regression (Fig. 6). There was also a significant downregulation of miR-196b in men with PAD compared with controls, although again this was not maintained on binary logistic regression (FC −5⋅15; P = 0⋅014; adjusted P = 0⋅219). There was no significant difference in miR-196b expression between AAA and PAD cohorts (adjusted P = 0⋅567), nor between men with previous AAA repair and controls (adjusted P = 0⋅814), whether open or endovascular. There was also no significant difference between preoperative and postoperative miR-196b expression in comparison of either all samples (adjusted P = 0⋅236) or matched samples (P = 0⋅146); subgroup analysis between open and endovascular samples revealed no significant difference. There was no correlation between miR-196b expression and aortic diameter (R2 = 0⋅003, P = 0⋅635). © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
Aortic tissue results For the initial validation study, aortic tissue was collected from 19 men with an AAA and ten cadaveric donors. For the additional validation study, aortic tissue was collected from 22 men having AAA repair, three men undergoing aortobifemoral bypass surgery for atherosclerosis, and 17 cadaveric donors. All samples showed good PCR curves and were therefore used for the analysis. Background data for both the initial and additional aortic tissue cohorts are summarized in Tables S7 and S8 (supporting information). For both the initial and additional aortic tissue cohorts, no significant difference was identified for any miRNA (Tables S9 and S10, supporting information).
Interactions between microRNAs identified and abdominal aortic aneurysm-related genes The miRWalk server was used to identify both validated and predicted interactions between let-7e, miR-15a, miR-196b and miR-411 and AAA biomarkers (MMP-2, MMP-9, TIMP-1, IL-6, TNF-α, osteoprotegerin, osteopontin, IFN-γ, ICAM-1, VCAM-1, D-dimer, CRP, α1-antitrypsin, fibrinogen, triglycerides, lipoprotein(a), apoprotein A, high-density lipoprotein, DAB2 interacting protein (DAB2IP), low-density lipoprotein receptor-related protein (LRP) 1, sortilin 1, low-density www.bjs.co.uk
BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
lipoprotein receptor (LDLR), angiotensin-converting enzyme, methylenetetrahydrofolate reductase (MTHFR)). Several validated interactions were identified with let-7e (IL-6, IFN, TNF-α), miR-15a (IL-6) and miR-196b (IFN). There were also several predicted interactions with let-7e (VCAM-1, CRP, osteoprotegerin, MTHFR), miR-15a (TNF-α, α1-antitrypsin, osteoprotegerin, CRP, MTHFR, DAB2IP), miR-196b (TNF-α, osteoprotegerin, LRP-1, MTHFR) and miR-411 (TNF-α, ICAM-1, osteoprotegerin, LDLR, MTHFR) (Fig. 7). Discussion
A discovery and validation study into miRNAs in AAA was undertaken. Significant upregulation of 29 miRNAs was identified in the discovery study, four of which were validated in blood (let-7e, miR-15a, miR-196b, miR-411), and one (miR-196b) further validated in plasma. Although no miRNA dysregulation was found in aortic tissue, all four miRNAs validated in blood were found to have significant interactions with previously identified AAA biomarkers15,16 . All four miRNAs were similarly dysregulated in patients with PAD, indicating that this dysregulation may be due to atherosclerosis rather than specifically to AAA. This study cannot be used to provide a biomarker for AAA, but it is possible to conclude that the miRNAs studied here were not altered by surgical intervention, and were not sufficiently correlated with aortic diameter to be useful. In addition, the miRNAs were not specific to AAA as they were also dysregulated in patients with PAD. The results of the validation study in whole-blood samples revealed three miRNAs to be significantly downregulated in patients with AAA compared with controls (let-7e, miR-15a, miR-196b), and one miRNA (miR-411) to be significantly upregulated; all except miR-411 remained significant when the analysis was corrected for confounding factors including age, smoking status, medication and cardiovascular risk factors. MicroRNAs miR-15a, miR-196b and miR-411 were also found to be differentially expressed in patients after AAA repair compared with controls, with miR-196b and miR-411 remaining significant after correction for confounding factors. No differences were found in miRNA expression between preoperative and postoperative samples from the same men, which suggested that exclusion of the aneurysm sac did not alter miRNA expression. Additionally, let-7e levels were significantly, but weakly, correlated with aneurysm diameter. Although these results were significant, it should be noted that both miR-411 and let-7e were dysregulated in the opposite direction to the findings in the discovery study. © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
An important aim of this study was to determine whether the miRNAs found to be differentially expressed in men with an AAA were specific to AAA, or were a feature of generalized atherosclerosis, as it has been hypothesised previously that these conditions share similar aetiology. Analysis of the validation cohort showed the differentially expressed miRNAs to be dysregulated in patients with PAD also, giving further weight to these conditions having a similar genetic basis. The reason why certain individuals develop atherosclerotic disease and others develop AAA remains to be elucidated. A single previous study38 showed let-7e to be associated with PAD; miR-15a, miR-196b and miR-411 have not previously been identified in PAD. The results from the plasma study reinforce the likelihood that miR-196b plays a role in the pathogenesis of AAA, but they also give further weight to the theory that AAA and PAD share a similar aetiology. The validation study in aortic tissue samples revealed no significant difference in expression levels of any of the miRNAs. This may be due to a type II error caused by an insufficient number of samples; however, as these results have been replicated in two independent cohorts it is likely that there is no correlation between circulating miRNA expression and aortic tissue expression of these miRNAs. Several validated and predicted interactions between the miRNAs found to be differentially expressed in whole blood in subjects with AAA and known genes or biomarkers of AAA have been identified. This highlights the plausibility of let-7e, miR-15a, miR-196b and miR-411 being dysregulated in subjects with AAA; however, these genes and biomarkers are not specific to AAA. Therefore, although these interactions add credibility to the miRNAs identified, the results should be viewed with caution, as the majority of these genes/biomarkers are also dysregulated in subjects with generalized atherosclerosis. To date, seven studies28,35,37,39 – 42 have investigated miRNA expression in animals, identifying significant dysregulation of miR-19a, miR-19b, miR-21, miR-24, miR-26a, miR-29b, miR-132, miR-143, miR-145, miR-195 and miR-222 in AAA. The first human studies40 – 42 of miRNA expression in AAA validated the animal models, revealing that miR-29b and miR-24 are significantly downregulated in aortic tissue, whereas miR-21 is significantly upregulated in AAA tissue. Recently, two miRNome studies have revealed further insights into aneurysm genesis. Pahl and co-workers22 looked at the expression of 847 miRNAs in five AAA and five control aortic tissue samples, identifying upregulation of miR-181*, miR-146 and miR-21, and downregulation of miR-133b, miR-133a, miR-331-3p, miR-30c-2* and miR-204. Validation work was undertaken using 25 elective AAA samples, www.bjs.co.uk
BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
11 ruptured AAA samples and seven control samples, which replicated the downregulated results, but not the upregulated ones. Kin and colleagues23 also used a miRNome approach, analysing the expression of 384 miRNAs in aortic tissue from 13 AAA samples and seven control samples. Initial microarray analysis identified 77 upregulated miRNAs, and validation work was carried out on a subset of these. A significant difference was found in the levels of let-7f, miR-21, miR-20a, miR-27a, miR-92a, miR-126, miR-221, miR-222, miR-29b, miR-124a, miR-146a, miR-155 and miR-223. Kin et al.23 then looked at miRNA expression in plasma samples from 23 patients with AAA, 17 with coronary artery disease and 12 controls; miR-126, miR-124a, miR-146a, miR-155, miR-223, miR-29b, miR-15a and miR-15b were significantly downregulated in AAA compared with controls. In addition, miR-124a, miR-155, miR-223 and miR-29b were downregulated in AAA compared with coronary artery disease. The present study was able to replicate these results with regard to miR-15a, but none of the other miRNAs identified above. This was possibly due to a difference in the tissue type studied: miR-29b, miR-31, miR-138, miR-145, miR-155, miR-181a and miR-221 were examined in whole blood, plasma and aortic tissue in the present study. A further study by Biros and co-workers27 looked at 124 miRNAs in aortic tissue and serum samples in ten patients undergoing AAA repair and ten controls. They identified differential expression of miR-155 in aortic aneurysm tissue compared with aortic neck samples, and also in the circulation, as found by Kin and colleagues23 but not in the present study. In addition, differential expression of miR-28, miR-30a-3p, miR-150, miR-302b*, miR-93 and miR-99a was identified. Although this study examined a wide range of miRNAs, it was possible to take only a subset of those identified in the discovery study through to validation. Future work, therefore, should initially evaluate the expression of these remaining miRNAs in blood, plasma and aortic tissue samples. In addition, these positive findings should be taken forward into experimental models to determine whether dysregulation of the miRNAs in cell culture and in vivo models alters expression of AAA-related biomarkers such as MMP-9, and whether their dysregulation leads to AAA development in animal models. Contributors
N. Sylvius (Department of Genetics, University of Leicester), D. A. Sidloff (Department of Cardiovascular Sciences, University of Leicester), N. Dattani © 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
(Department of Cardiovascular Sciences, University of Leicester), A. Verissimo (Department of Cardiovascular Sciences, University of Leicester), J. B. Wild (Department of Cardiovascular Sciences, University of Leicester), H. Z. Butt (Department of Cardiovascular Sciences, University of Leicester), E. Choke (Department of Cardiovascular Sciences, University of Leicester), R. D. Sayers (Department of Cardiovascular Sciences, University of Leicester) and M. J. Bown (Department of Cardiovascular Sciences, University of Leicester and National Institute for Health Research (NIHR) Leicester Cardiovascular Biomedical Research Unit). Acknowledgements
This work was supported by a donation in memory of Bill Fowler and Tom Phillips. In addition, P.W.S. is funded by a Royal College of Surgeons/Dunhill Medical Trust Research Fellowship and a donation from Leicester NIHR Biomedical Research Unit. M.J.B. is funded by a Higher Education Funding Council for England Clinical Senior Lecturer Fellowship, the Leicester NIHR Cardiovascular Biomedical Research Unit and the Circulation Foundation President’s Early Career Award. Disclosure: The authors declare no conflict of interest. References 1 Kuivaniemi H, Shibamura H, Arthur C, Berguer R, Cole CW, Juvonen T et al. Familial abdominal aortic aneurysms: collection of 233 multiplex families. J Vasc Surg 2003; 37: 340–345. 2 Sakalihasan N, Limet R, Defawe OD. Abdominal aortic aneurysm. Lancet 2005; 365: 1577–1589. 3 Wahlgren CM, Larsson E, Magnusson PK, Hultgren R, Swedenborg J. Genetic and environmental contributions to abdominal aortic aneurysm development in a twin population. J Vasc Surg 2010; 51: 3–7. 4 Sandiford P, Mosquera D, Bramley D. Ethnic inequalities in incidence, survival and mortality from abdominal aortic aneurysm in New Zealand. J Epidemiol Community Health 2012; 66: 1097–1103. 5 Rossaak JI, Sporle A, Birks CL, van Rij AM. Abdominal aortic aneurysms in the New Zealand Maori population. Br J Surg 2003; 90: 1361–1366. 6 Sweeting MJ, Thompson SG, Brown LC, Powell JT; RESCAN collaborators. Meta-analysis of individual patient data to examine factors affecting growth and rupture of small abdominal aortic aneurysms. Br J Surg 2012; 99: 655–665. 7 Giugliano G, Laurenzano E, Rengo C, De Rosa G, Brevetti L, Sannino A et al. Abdominal aortic aneurysm in patients affected by intermittent claudication: prevalence and clinical predictors. BMC Surg 2012; 12(Suppl 1): S17.
www.bjs.co.uk
BJS
MicroRNAs associated with abdominal aortic aneurysms and peripheral arterial disease
8 Cornuz J, Sidoti Pinto C, Tevaearai H, Egger M. Risk factors for asymptomatic abdominal aortic aneurysm: systematic review and meta-analysis of population-based screening studies. Eur J Public Health 2004; 14: 343–349. 9 Golledge J, Norman PE. Atherosclerosis and abdominal aortic aneurysm: cause, response, or common risk factors? Arterioscler Thromb Vasc Biol 2010; 30: 1075–1077. 10 Gretarsdottir S, Baas AF, Thorleifsson G, Holm H, den Heijer M, de Vries JP et al. Genome-wide association study identifies a sequence variant within the DAB2IP gene conferring susceptibility to abdominal aortic aneurysm. Nat Genet 2010; 42: 692–697. 11 Jones GT, Bown MJ, Gretarsdottir S, Romaine SP, Helgadottir A, Yu G et al. A sequence variant associated with sortilin-1 (SORT1) on 1p13.3 is independently associated with abdominal aortic aneurysm. Hum Mol Genet 2013; 22: 2941–2947. 12 Bown MJ, Jones GT, Harrison SC, Wright BJ, Bumpstead S, Baas AF et al. Abdominal aortic aneurysm is associated with a variant in low-density lipoprotein receptor-related protein 1. Am J Hum Genet 2011; 89: 619–627. 13 Bradley DT, Hughes AE, Badger SA, Jones GT, Harrison SC, Wright BJ et al. A variant in LDLR is associated with abdominal aortic aneurysm. Circ Cardiovasc Genet 2013; 6: 498–504. 14 Thompson AR, Drenos F, Hafez H, Humphries SE. Candidate gene association studies in abdominal aortic aneurysm disease: a review and meta-analysis. Eur J Vasc Endovasc Surg 2008; 35: 19–30. 15 Sidloff DA, Stather PW, Choke E, Bown MJ, Sayers RD. A systematic review and meta-analysis of the association between markers of hemostasis and abdominal aortic aneurysm presence and size. J Vasc Surg 2014; 59: 528–535.e4. 16 Stather PW, Sidloff DA, Dattani N, Gokani VJ, Choke E, Sayers RD et al. Meta-analysis and meta-regression analysis of biomarkers for abdominal aortic aneurysm. Br J Surg 2014; 101: 1358–1372. 17 Lagos-Quintana M, Rauhut R, Yalcin A, Meyer J, Lendeckel W, Tuschl T. Identification of tissue-specific microRNAs from mouse. Curr Biol 2002; 12: 735–739. 18 Liu CG, Calin GA, Meloon B, Gamliel N, Sevignani C, Ferracin M et al. An oligonucleotide microchip for genome-wide microRNA profiling in human and mouse tissues. Proc Natl Acad Sci U S A 2004; 101: 9740–9744. 19 Tang Y, Liu D, Zhang L, Ingvarsson S, Chen H. Quantitative analysis of miRNA expression in seven human foetal and adult organs. PLoS One 2011; 6: e28730. 20 Jopling CL, Norman KL, Sarnow P. Positive and negative modulation of viral and cellular mRNAs by liver-specific microRNA miR-122. Cold Spring Harb Symp Quant Biol 2006; 71: 369–376. 21 Bostjancic E, Zidar N, Stajer D, Glavac D. MicroRNAs miR-1, miR-133a, miR-133b and miR-208 are dysregulated
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
22
23
24
25
26
27
28
29
30
31
32
33
34
35
in human myocardial infarction. Cardiology 2010; 115: 163–169. Pahl MC, Derr K, Gäbel G, Hinterseher I, Elmore JR, Schworer CM et al. MicroRNA expression signature in human abdominal aortic aneurysms. BMC Med Genomics 2012; 5: 25. Kin K, Miyagawa S, Fukushima S, Shirakawa Y, Torikai K, Shimamura K et al. Tissue- and plasma-specific microRNA signatures for atherosclerotic abdominal aortic aneurysm. J Am Heart Assoc 2012; 1: e000745. Maegdefessel L, Dalman RL, Tsao PS. Pathogenesis of abdominal aortic aneurysms: microRNAs, proteases, genetic associations. Annu Rev Med 2014; 65: 49–62. Stather P, Wild J, Sylvius N, Choke E, Sayers R, Bown M. MicroRNAs associated with the pathways involved in the pathogenesis of abdominal aortic aneurysms. Artery Research 2012; 7: 28–35. Adam M, Raaz U, Spin JM, Tsao PS. MicroRNAs in abdominal aortic aneurysm. Curr Vasc Pharmacol 2013; [Epub ahead of print]. Biros E, Moran CS, Wang Y, Walker PJ, Cardinal J, Golledge J. MicroRNA profiling in patients with abdominal aortic aneurysms: the significance of miR-155. Clin Sci (Lond) 2014; 126: 795–803. Zampetaki A, Attia R, Mayr U, Gomes RS, Phinikaridou A, Yin X et al. Role of miR-195 in aortic aneurysmal disease. Circ Res 2014; 115: 857–866. Dweep H, Sticht C, Pandey P, Gretz N. miRWalk – database: prediction of possible miRNA binding sites by ‘walking’ the genes of three genomes. J Biomed Inform 2011; 44: 839–847. Boon RA, Seeger T, Heydt S, Fischer A, Hergenreider E, Horrevoets AJ et al. MicroRNA-29 in aortic dilation: implications for aneurysm formation. Circ Res 2011; 109: 1115–1119. Maegdefessel L, Azuma J, Toh R, Merk DR, Deng A, Chin JT et al. Inhibition of microRNA-29b reduces murine abdominal aortic aneurysm development. J Clin Invest 2012; 122: 497–506. Lee ST, Chu K, Im WS, Yoon HJ, Im JY, Park JE et al. Altered microRNA regulation in Huntington’s disease models. Exp Neurol 2011; 227: 172–179. Liu X, Cheng Y, Chen X, Yang J, Xu L, Zhang C. MicroRNA-31 regulated by the extracellular regulated kinase is involved in vascular smooth muscle cell growth via large tumor suppressor homolog 2. J Biol Chem 2011; 286: 42 371–42 380. Cheng Y, Liu X, Yang J, Lin Y, Xu DZ, Lu Q et al. MicroRNA-145, a novel smooth muscle cell phenotypic marker and modulator, controls vascular neointimal lesion formation. Circ Res 2009; 105: 158–166. Damrauer SM, Csizmadia E, Studer P, Bhasin M, Da Silva CG, Peterson CR et al. Partial loss of A20 (Tnfaip3) promotes resistance to abdominal aortic aneurysms. ATVB 2010; 30: e317.
www.bjs.co.uk
BJS
P. W. Stather, N. Sylvius, D. A. Sidloff, N. Dattani, A.Verissimo, J. B. Wild et al.
36 Davis BN, Hilyard AC, Nguyen PH, Lagna G, Hata A. Induction of microRNA-221 by platelet-derived growth factor signaling is critical for modulation of vascular smooth muscle phenotype. J Biol Chem 2009; 284: 3728–3738. 37 Liu G, Huang Y, Lu X, Lu M, Huang X, Li W et al. Identification and characteristics of microRNAs with altered expression patterns in a rat model of abdominal aortic aneurysms. Tohoku J Exp Med 2010; 222: 187–193. 38 Stather PW, Sylvius N, Wild JB, Choke E, Sayers RD, Bown MJ. Differential microRNA expression profiles in peripheral arterial disease. Circ Cardiovasc Genet 2013; 6: 490–497. 39 Leeper NJ, Raiesdana A, Kojima Y, Chun HJ, Azuma J, Maegdefessel L et al. MicroRNA-26a is a novel regulator of
vascular smooth muscle cell function. J Cell Physiol 2011; 226: 1035–1043. 40 Maegdefessel L, Azuma J, Toh R, Merk DR, Deng A, Chin JT et al. Inhibition of microRNA-29b reduces murine abdominal aortic aneurysm development. J Clin Invest 2012; 122: 497–506. 41 Maegdefessel L, Azuma J, Toh R, Deng A, Merk DR, Raiesdana A et al. MicroRNA-21 blocks abdominal aortic aneurysm development and nicotine-augmented expansion. Sci Transl Med 2012; 4: 122ra22. 42 Maegdefessel L, Raaz U, Toh R, Deng A, Chernogubova E, Eriksson P et al. MicroRNA-24 regulates inflammation and calcification during abdominal aortic aneurysm development through targeting Ykl-40. Circ Suppl 2012; 126: A14221.
Supporting information
Additional supporting information may be found in the online version of this article: Appendix S1 RNA isolation and microRNA expression profiling methods (Word document) Table S1 Background comparison between control patients and those with abdominal aortic aneursym included in discovery study (Word document) Table S2 Fold change in all microRNAs expressed in whole-blood samples from discovery study (Word document) Table S3 Background analysis of subjects included in whole-blood sample validation study (Word document) Table S4 Results of one-way ANOVA comparing relative microRNA expression in whole-blood samples between all 80 subjects with an abdominal aortic aneurysm, 40 with peripheral arterial disease and 40 controls (Word document) Table S5 Background analysis of subjects included in plasma sample validation study (Word document) Table S6 Results of one-way ANOVA comparing relative microRNA expression in plasma samples between all 72 subjects with an abdominal aortic aneurysm, 35 with peripheral arterial disease and 28 controls (Word document) Table S7 Background analysis of subjects included in initial aortic tissue study (Word document) Table S8 Background analysis of subjects included in additional aortic tissue study (Word document) Table S9 Comparison of microRNA expression in aortic tissue from 19 abdominal aortic aneurysm samples and samples from ten cadaveric donors (Word document) Table S10 Comparison of microRNA expression between abdominal aortic aneurym, peripheral arterial disease and cadveric donor cohorts (Word document)
© 2015 BJS Society Ltd Published by John Wiley & Sons Ltd
www.bjs.co.uk
BJS
