Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
1
Differential expression analyses on aortic tissue reveal novel genes and pathways associated
1
with abdominal aortic aneurysm onset and progression
2
Gerard Temprano-Sagrera MSc (gtemprano@santpau.cat)
1
, Begoña Soto MD PhD 3
(bsoto@santpau.cat)
1,2
, Jaume Dilmé MD PhD (jdilme@santpau.cat)
1,2,3
, Olga Peypoch MD 4
(opeypoch@santpau.cat)
1,2
, Laura Calsina Juscafresa MD PhD (lcalsina@psmar.cat)
4,5
, David 5
Davtian PhD
6
(2400580@dundee.ac.uk), Lluís Nieto MD (llnietofernandez@gmail.com)
4
, 6
Andrew Brown PhD (a.z.t.brown@dundee.ac.uk)
6
, José Román Escudero MD 7
(jescuderor@santpau.cat)
1,2
, Ana Viñuela PhD# (ana.vinuela@newcastle.ac.uk)
7
, Mercedes 8
Camacho PhD# (mcamacho@santpau.cat)
1
, Maria Sabater-Lleal PhD# 9
(msabater@santpau.cat)
1,8,9
10
11
#These authors equally contributed as senior authors. 12
13 Affiliations: 14 15
1. Unit of genomics of Complex Diseases, Institut de Recerca Sant Pau (IR SANT PAU), Sant Quintí 77-79,
16
08041 Barcelona, Spain
17
2. Servei d'Angiologia i Cirurgia Vascular i Endovascular, Hospital de la Santa Creu i Sant Pau, Sant Antoni
18
Maria Claret 167, 08025 Barcelona, Spain
19
3. Centro de Investigación Biomédica en Red de Enfermedades Cardiovasculares (CIBERECV), Madrid,
20
Spain
21
4. Department of Vascular and Endovascular Surgery, Hospital del M ar, Pa sseig Marítim 25-29, 08003,
22
Barcelona, Spain
23
5. Department of Medicine and Surgery, Universitat Pompeu Fabra, Barcelona, Spain
24
6. Population Health and Genomics, Ninewells Hospital and Medical School, University of Dundee,
25
Dundee, DD1 9SY, United Kingd om .
26
7. Biosciences Institut e, Faculty of Medical Sciences, University of Newcastle, Newcastl e upon Tyne, NE1
27
4EP, United Kingdom
28
8. Cardiovascular Medicine Unit, Department of Medicine, Karolinska Institutet, Stoc kholm, Sweden
29
9. Centro de Investigació n Biomédica en Red de Enfermedades Raras (CIBERER), Madrid, Spai n
30 31 32 Corresponding author: 33 34 Maria Sabater-Lleal, PhD 35 Genomics of Complex Disease Unit, Institut de Recerca Sant Pau (IR SANT PAU) 36 St Quintí 77-79, 08041, Barcelona, Spain 37 Phone +34932919000; Email: msabater@santpau.cat
38
39
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.
2
ABSTRACT
40
Background: 41
Abdominal aortic aneurysms (AAA) are focal dilatations of the abdominal aorta. They are 42
normally asymptomatic and progressively expand, increasing their risk of rupture. Rupture of 43
an AAA is associated with high mortality rates, but the mechanisms underlying the initiation, 44
expansion and rupture of AAA are not yet fully understood. This study aims to characterize and 45
identify new genes associated with the pathophysiology of AAA through differential expression 46
analyses between dilated and non-dilated aortic tissue samples, and between AAA of different 47
diameters. Our study used RNA-seq data on 140 samples, becoming the largest RNA-seq 48
dataset for differential expression studies of AAA. 49
Results: 50
We identified 7,454 differentially expressed genes (DEGs) between AAA and controls, 2,851 of 51
which were new compared to previous microarray studies. Notably, a novel cluster on 52
adenosine triphosphate synthesis regulation emerged as strongly associated with AAA. 53
Additionally, exploring AAA of different diameters identified eight genes (
EXTL3
,
ZFR
,
DUSP8
, 54
DISP1
,
USP33
,
VPS37C
,
ZNF784
,
RFX1
) that overlapped with the DEGs between AAA and 55
controls, implying roles in both disease onset and progression. Seven genes (
SPP1
,
FHL1
,
GNAS
, 56
MORF4L2
,
HMGN1
,
ARL1
,
RNASE4
) with differential splicing patterns were also DEGs between 57
AAA and controls, suggesting that splicing differences contribute to the observed expression 58
changes and the disease development. 59
Conclusions: 60
This study identified new genes and pathways associated with AAA onset and progression and 61
validated previous relevant roles of inflammation and intracellular calcium regulation. These 62
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
3
findings provide insights into the complex mechanisms underlying AAA and indicate potential 63
targets to limit AAA progression and mortality risk. 64
KEYWORDS
: Abdominal aortic aneurysm, RNAseq, transcriptomics, differential expression, 65
ischemic time, alternative splicing, allelic specific expression. 66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
4
BACKGROUND:
84
Abdominal aortic aneurysms (AAA) are characterized by a local dilation of the infrarenal 85
abdominal aorta to about 1.5 times the normal adjacent aortic diameter or more than 3 cm in 86
maximum diameter[1]. AAA is accompanied by chronic inflammation, apoptosis of vascular 87
smooth muscle cells and neovascularization[2,3 ]. Additionally, extracellular matrix degradation, 88
microcalcification, and oxidative stress contribute to the degeneration of the aorta[1,2]. The 89
disease is progressive, and most aneurysms develop without causing symptoms[1]. However, in 90
the event of AAA rupture, mortality rates can reach 80 %[4]. The only effective treatment 91
currently available for AAA is aortic tissue repair, either through open surgery or endovascular 92
repair[1,5]. 93
Some risk factors are known for developing AAA, including age, male sex, smoking, and family 94
history of AAA[1]. Smoking, in addition, is also known to increase the risk of rupture[6]. 95
Additionally, recent genomic studies have revealed 121 loci associated with risk of developing 96
AAA, contributing to the knowledge of the possible pathways leading to this disease[7]. 97
However, there is still an insufficient understanding of the clear mechanisms that underlie the 98
initiation, propagation, and rupture of AAA. 99
The study of gene expression, known as transcriptomics, is a valuable tool for understanding 100
human disease and revealing new therapeutic targets [8,9]. Several studies have been 101
performed to study the differentially expressed genes (DEGs) between dilated aortic tissue and 102
non-dilated control aorta using microarray technology, detecting DEGs especially associated 103
with the immune and inflammatory responses, extracellular matrix remodeling and 104
angiogenesis[10–15]. In the present study, we performed RNA sequencing (RNAseq) of the 105
complete transcriptome in 140 abdominal aortic tissue samples (96 dilated aortas and 44 106
control aortas from deceased donors) from the Triple A Barcelona Study (TABS) cohort, to 107
identify new DEGs and pathways associated with the pathophysiology of AAA initiation and 108
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
5
progression, allowing for a more comprehensive analysis of the transcriptome and becoming 109
the largest RNAseq dataset for AAA tissue. Additionally, we aimed to investigate the differences 110
in alternative splicing patterns in the context of AAA, and the role of genetic variants in gene 111
expression in AAA tissue. The study design is described in
Figure 1
. 112
RESUL TS:
113
Participants characteristics: 114
Table 1
shows the participants demographic and clinical data. Aortic tissue samples from 96 115
AAA patients and 44 controls from the TABS cohort were used for RNAseq analysis. 116
Table 1: Characteristics of study participants
Controls
(N = 44)
AAA
(N = 96) P-value Missing
values (%)
Age (Years)
61.66 (21-82) 70.38 (53-87) 0.0006 0 (0)
Sex (Male)
21 (47.73) 92 (95.83) 1.00E-10 0 (0)
Smoking (Current)
8 (21.62) 27 (32.14) 0.3378 19 (13.57)
Smoking (Past)
3 (8.11) 42 (50) 2.81E-05 19 (13.57)
Aortic Diameter (mm)
NA 65.57 (38-100.12) - 0 (0)
Hypertension (Yes)
15 (40.54) 54 (64.29) 0.0256 19 (13.57)
Dyslipidemia (Yes)
10 (27.03) 46 (54.76) 0.0087 19 (13.57)
Diabetes mellitus (Yes)
5 (13.51) 12 (14.29) 1 19 (13.57)
Peripheral arterial disease (Yes)
NA 23 (27.71) - 57 (59.38)
Other aneurysms (Yes)
NA 25 (29.76) - 12 (12.5)
Cerebrovascular Disease (Yes)
3 (8.11) 36 (43.9) 0.0003 21 (15)
Cardiovascular Disease (Yes)
1 (3.34) 16 (19.05) 0.0846 27 (19.29)
Chronic obstructive pulmonary disease (Yes)
1 (2.7) 15 (17.86) 0.0481 19 (13.57)
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
6
Chronic kidney disease (Yes)
2 (5.4) 17 (20.24) 0.0726 19 (13.57)
Continuous variables are presented as mean (range), and categorical variables are presented as
%. Two-sample t-tests and chi-squared tests were used to compare the means of continuous
phenotypes and the distribution of categorical phenotypes, between AAA and control groups,
respectively. Missing values were excluded from the calculations of each variable. Hypertension
was defined based on clinical his tory and the use of an tihypertensive m e dic ation. Dyslipide mia
was diagnosed through clinical history and the use of hypolipidemic medication. Diabetes
mellitus was identified by clinical history and the use of insulin or oral hypoglycemic
medications, without differentiation between type 1 or type 2. Peripheral arterial disease was
assessed based on clinical symptoms and clinical history. Other aneurysms included thoracic
and visceral aortic aneurysms, iliac artery aneurysms, and popliteal artery aneurysms, and
were diagnosed using computed tomography or ultrasound. Cerebrovascular diseases were
determined by a history of transient ischemic attack or stroke. Cardiovascular diseases were
assessed by history of acute myocardial infarction or angina pectoris, or admission with clinical
symptoms, electrocardiogram changes, or a positive enzymatic curve diagnosed by a
cardiologist. Chronic obstructive pulmonary disease was identified based on clinical history.
Chronic kidney disease was assessed by clinical history.
We examined demographic and clinical variables between AAA and controls to determine 117
whether these might influence our expression levels differences. We found significant 118
differences in sex (
Additional file 1: Figure S1A
), age (
Additional file 1: Figure S1B
), smoking 119
status, and the prevalence of hypertension, dyslipidemia, cerebrovascular disease, and chronic 120
obstructive pulmonary disease between AAA patients and controls. In relation to smoking 121
status (N = 122), which is a known risk factor for AAA development and rupture, 70.3 % of 122
controls were never smokers, compared to 17.86 % of AAA patients. Among current and past 123
smokers, there were also differences, with 50 % of former smokers and 32.14 % of current 124
smokers in AAA, compared to only 8.11 % and 21.62 % respectively, in the controls (
Additional
125
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
7
file 1: Figure S3C
). We adjusted our regression analyses for age and sex, acknowledging their 126
potential influence on expression level differences. 127
Differential expression analyses between AAA and controls: 128
The analysis of differential expression between aortic samples from 96 AAA patients and 44 129
controls revealed 7,454 genes displaying significant differences in expression (adjusted p-value 130
< 0.05) (
Additional file 1: Figure S2A and Additional file 2: Table S1
). This number exceeds by 131
26.57 % the previous DEGs identified in comparable analyses using microarray 132
technologies[10–15]. Using GO and KEGG enrichment analyses, we found a total of 1,152 and 133
89 enriched terms, respectively (
Additional file 1: Figure S3A and Figure S4A
). The complete 134
results of the enriched terms for GO and KEGG are shown in
Additional files 3 and 4: Tables S2
135
and S3
. To better characterize the biological processes associated with the DEGs in the 136
enrichment analysis, we performed a cluster analysis of GO pathways. We found that most of 137
the DEGs were associated with the immune system: regulation of mononuclear cell 138
proliferation, leukocyte chemotaxis, regulation of adaptive immune response based on somatic 139
recombination of immune receptors built from immunoglobulin superfamily domains, mast cell 140
degranulation, major histocompatibility complex (MHC) class II protein complex, positive 141
regulation of T cell activation, and CD4-positive alpha-beta T cell differentiation (
Figure 2A)
. 142
Other represented metabolic pathways were related to sequestering of calcium ion, regulation 143
of actin filament length and adenosine triphosphate (ATP) synthesis coupled electron transport 144
(
Figure 2A
). While these analyses corroborated previous associations with inflammatory, actin 145
filament regulation and intracellular calcium regulation processes, the ATP synthesis regulation 146
pathway was a new differential expressed pathway in AAA tissue. 147
Control aortic samples from deceased organ donors have been used by us and others [10–148
13,15]. To account for differences in gene expression between AAA and control tissue that 149
could be attributed to ischemic time (time between the donor’s death and sample collection 150
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
8
when blood flow is interrupted), we removed 10,737 DEGs associated to ischemic time in the 151
GTEx aorta samples (N = 387)[16] from the total DEGs found between AAA and controls, 152
leaving 3,002 DEGs (
Additional file 1: Figure S2B and Additional file 5: Table S4
). 153
We then performed a new enrichment analysis and identified 424 enriched GO terms and 65 154
KEGG pathways (
Additional file 1: Figure S3B and Figure S4B
) (Complete results are available 155
at
Additional files 6 and 7: Tables S5 and S6
, respectively), which represented removal of 728 156
and 24 pathways susceptible of being caused by ischemic time, respectively. Cluster analysis of 157
GO enriched terms confirmed identified clusters related to the regulation of calcium ion 158
retention, ATP synthesis coupled electron transport, and immune response centered on T cell 159
activation (MHC class II protein complex, positive regulation of T cell activation). On the other 160
hand, other clusters also associated with the immune system were no longer represented 161
(regulation of mononuclear cell proliferation, leukocyte chemotaxis regulation of adaptive 162
immune response based on somatic recombination of immune receptors built from 163
immunoglobulin superfamily domains, CD4-positive, alpha-beta T cell differentiation, mast cell 164
degranulation) together with the regulation of actin filament length (
Figure 2B
). By accounting 165
for genes whose expression was altered by ischemic time, we identified a set of genes that are 166
less likely to be affected by the experimental limitations of these types of studies. 167
Vascular inflammation has been previously associated to AAA development and 168
progression[17,18]. Even in our most stringent analysis, which removed pathways possibly 169
caused by ischemic time, there was a notable enrichment of pathways associated with the 170
immune response. Consequently, we decided to investigate the influence of the inflammatory 171
infiltrate on AAA by comparing the abundance of 22 immune cell types from gene expression 172
profiles between our AAA and control samples. After correcting for multiple testing, we found 173
significant differences on the abundances of CD8 T-cells, natural killer (NK) resting cells, and 174
dendritic activated cells (
Figure 2C
). AAA samples had a higher proportion of CD8 T-cells, while 175
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
9
controls had more NK resting cells and dendritic activated cells.
Table 2
provides a summary of 176
the three cell populations with the highest and lowest presence in each group. The findings 177
support the previously reported involvement of CD8 T-cells and NK cells in AAA.[17,18] The 178
lower levels of dendritic cells detected in AAA were unexpected, as previous studies found 179
significantly higher levels of dendritic cells in AAA samples compared to controls[19]. 180
Table 2:
Summary of the three most and least prevalent cell populations in AAA and controls.
181
Cell Populations (AAA) Percentage (%) Cell Populations (Controls) Percentage (%)
Plasma B cells
9.46
Resting mast cells
12.34
M2 macrophages
8.68
Resting memory CD4 T cells
8.31
Resting mast cells
6.99
M2 Macrophages
8.23
M1 macrophages
1.43
Eosinophils
2.03
Eosinophils
1.37
T cell follicular helper cells
1.94
Activated dendritic cells
1.13
M1 Macrophages
1.45
182
Study of alternative splicing: 183
The relevance of alternative splicing in the development of diseases, such as cancer, 184
neurological, and cardiovascular diseases has been well-established for years.[20] However, 185
capturing the complexity of alternative splicing has been challenging. With the recent 186
improvements in sequencing techniques it is now possible to study alternative splicing in more 187
depth.[20] We investigated alternative splicing patterns between the AAA and control groups 188
to identify specific splicing patterns associated with AAA. We identified 15 significant 189
alternative splicing events on eleven unique genes (
FHL1
,
GNAS
,
ASAH1
,
SPP1
,
ARL1
,
MORF4L2
, 190
CYCS
,
HMGB1
,
HMGN1
,
SELENOP
,
RNASE4
) between AAA and controls. The analysis revealed 191
that, as anticipated, the number of altered alternative first exon events was more represented 192
than any other splice event among AAA and control samples (
Figure 3A
), consistent with 193
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
10
previous work suggesting that it is the most frequent splice event in the human genome[21] 194
(Complete results are available in
Additional file 8: Table S7
). A functional enrichment analysis 195
was conducted on the eleven genes with significant alternative splicing events, but no 196
significantly enriched metabolic pathways were found. Interestingly, seven out of the eleven 197
genes were also differentially expressed between AAA and controls, suggesting that the 198
expression of specific splicing variants could be altered in AAA (
Figure 3B
). 199
Differential expression analysis by aortic diameter: 200
The diameter of AAA is a significant risk factor for rupture. We analyzed the DEGs by diameter 201
to identify alterations in gene expression throughout the progression of the disease. We 202
observed a total of 32 DEGs among aneurysms of varying diameters (N = 84), although no 203
enriched pathways were identified (
Additional file 1: Figure S2C and Additional file 9: Table
204
S8
). Of the 32 DEGs by diameter, eight were also DEG s between AA A and controls (
Figure 3C
), 205
suggesting that these eight genes are relevant for disease formation and also during disease 206
progression. 207
Allelic specific expression: 208
We investigated the potential effect of AAA-associated genetic variants on gene expression in 209
diseased tissue by studying allele specific expression in twelve AAA samples with available 210
genetic data. On average, we identified 529 genes with significant allele specific expression 211
(adjusted p-value < 0.05) in the twelve AAA samples. Among these genes, 90 exhibited 212
significant allele specific expression in more than five of the twelve AAA samples. Additionally, 213
to determine whether these associations were related to AAA or were characteristic of the 214
aortic tissue, we compared allele specific expression patterns between our AAA samples and 215
387 GTEx aortic samples, used as controls. The comparison between AAA and control samples 216
identified 1,815 genes with significant differences in the allele specific expression patterns 217
(adjusted p-value < 0.05) (
Figure 4A
). An enrichment analysis on GO terms for these 1,815 218
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
11
genes revealed 91 enriched pathways. The posterior cluster analysis revealed three clusters 219
strongly related to the immune system: MHC protein complex, positive regulation of T cell-220
mediated cytotoxicity , and regulation of T cell activation (
Figure 4B
). The association between 221
the immune system and AAA was also observed in the differential expression analysis between 222
AAA samples and controls, validating the robustness of these results. 223
Finally, among the 90 genes that exhibited significant allele specific expression in more than 224
five AAA samples, we selected those that also showed significant differential allelic specific 225
expression analysis between AAA and GTEx control samples, and those present in loci 226
identified in the largest genome-wide association study (GWAS) on AAA risk[7], in order to 227
identify haplotypes associated with AAA risk. Among the selected genes,
SNURF
was the only 228
gene that also presented differential allele specific expression patterns between AAA and 229
control tissues, and
SPP1
and
THBS2
were prioritized based on their presence in a locus 230
identified in the previous GWAS on AAA. This allowed to hypothesize that the presence of 231
particular genetic haplotypes in these three genes determined their differential expression 232
associated with risk of AAA. 233
DISCUSSION
: 234
This study analyzes differential expression between AAA aortic tissue samples and control 235
aortic samples using whole transcriptome data obtained through RNAseq. In addition, we 236
studied the effect of ischemic time on gene expression, to obtain a more credible list of genes 237
associated with AAA development. Using our RNAseq data, which provides superior alternative 238
splicing analysis compared to microarrays[22], we conducted a novel exploration of alternative 239
splicing between AAA and control samples, to identify potential causes of the observed 240
differential expression. Furthermore, we analyzed the differential expression between AAA of 241
different diameters to study the genes altered during disease progression. Finally, we analyzed 242
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
12
allele specific expression to gain insights into how genetic variants impact expression in the 243
diseased tissue. 244
Study of ischemic time-independent pathways involved in AAA development: 245
Clustering analysis with the enriched pathways after accounting for the ischemic time effect 246
revealed a strong association with MHC class II protein complex, positive regulation of T-cells, 247
and intracellular calcium ion regulation. Additionally, we have for the first time identified the 248
regulation of the ATP synthesis pathway in a differential expression analysis of aortic samples. 249
While the detection of the ATP synthesis regulatory pathway is novel, it is in line with previous 250
work associ ating mitochondrial disfunction and AAA [23,24]. On the o ther hand, previous 251
differential expression studies in microarrays between AAA and control aortic tissue have 252
consistently found associations of immune system p athways with AAA.[10–13,15] Our re sults 253
confirm these associations and demonstrate that these are independent of the ischemic time, 254
which is a confounder factor in most studies using donor samples. Finally, the regulation of 255
intracellular calcium was previously detected in one study of differential expression between 256
dilated and non-dilated aortic samples[11]. Our analyses confirm that this association is 257
independent of ischemic time. 258
We identified for the first time several enriched signaling pathways, with a large presence of 259
genes that code for subunits of complexes I (NADH ubiquinone oxidoreductase)), III (Ubiquinol-260
cytochrome c reductase) and IV (cytochrome c oxidase)) of the electron transport chain. Our 261
results indicate that 88 % (22 / 25) of the DEGs coding for the subunits of the complexes that 262
form the electron transport chain are expressed to a lower extent in AAA, suggesting a lower 263
synthesis of ATP in AAA tissue. Mitochondrial dysfunction has previously been studied in the 264
development of AAA[23,24] and other cardiovascular diseases[25] due to its key role in some 265
of the cellular alterations characteristic of cardiovascular diseases, including excessive 266
production of reactive oxygen species, energy depletion, endoplasmic reticulum stress, and 267
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
13
activation of apoptosis. However, this is the first time that these metabolic pathways have been 268
characterized in a DEGs study of AAA, confirming the role of mitochondrial dysfunction on AAA. 269
Among the genes present in all the enriched signaling pathways related to intracellular calcium 270
regulation,
APLNR
[26],
F2R
[27],
GPER1
[19],
JPH2
[28],
PKD2
[29] and
THY1
/
CD90
[30] have 271
already been investigated for their role in AAA. However, six additional genes are present in all 272
pathways:
ABL1
,
CALM1
,
CALM2
,
RYR2
,
SRI
and
CD19
. Except for
CALM2
, all of them have been 273
previously identified as DEGs in previous microarray studies between AAA and control 274
samples[10–12,15], but none of these genes have been further investigated in either functional 275
or epidemiological studies. These genes are closely related to intracellular calcium metabolism. 276
ABL1
participates in both the release of stored intracellular calcium and extracellular calcium 277
entry[31].
CALM1
and
CALM2
code for two isoforms of the calmodulin protein, which plays a 278
crucial role in the contraction of vascular and cardiac tissue through the detection of 279
intracellular calcium[32].
RYR2
is mainly expressed in cardiac tissue and codes for the main 280
regulator of sarcopla smic calcium relea se[3 3].
SRI
codes for the main binding protein of the 281
RYR2
gene product[34]. The comparison between our AAA and control samples shows a 282
downregulation of all these genes, suggesting a decrease in intracellular calcium levels in 283
smooth muscle cells, consistent with loss of vascular contractility in the dilated aorta[35]. 284
On the other hand, we observed upregulation of the
CD19
gene in AAA. The activation of the 285
surface protein encoded by
CD19
triggers the release of intracellular calcium, which contrasts 286
with the previous results[36]. However,
CD19
i s also a biomarker of B-cell development [36] 287
which also play a key role in the development of AAA[37]. 288
Among the pathways that were previously identified in differential expression analyses 289
between AAA and control tissue is inflammation. There is a widely studied inflammatory 290
component in AAA development involving both, adaptative and innate immune responses 291
[17,18]. The presence of inflammatory infiltrates in AAA tissue have been widely 292
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
14
demonstrated, which play a key role in the development of the disease[17,18]. The results of 293
our cluster analysis also corroborated the association with the immune system after accounting 294
for ischemic time, highlighting the key role of T cells on AAA development[38,39]. 295
To better understand the effects of the inflammatory infiltrate in AAA development, we 296
compared the proportions of inflammatory infiltrates between AAA and control samples. Some 297
previous studies have analyzed the inflammatory infiltrates in AAA tissue[15,19,40]. In one 298
study[15], immune cell proportions were estimated in AAA tissue layers (media and adventitia), 299
without comparing with controls. Our results strongly corroborate their findings, suggesting 300
that plasma B cells and M2 macrophages were the two most represented cell populations in 301
both layers, and M1 macrophages, eosinophils, and activated dendritic cells were among the 302
least represented cells in both layers. Surprisingly, resting mast cells emerged as our third most 303
represented cell group, while in the layer-specific analysis, mast cells represented a small 304
percentage of the total inflammatory cells. This contrasts with another study that compared 305
whole-tissue samples between AAA and controls[19],where resting mast cells were among the 306
most frequent cell groups in AAA samples. In addition, our results align with this study in 307
detecting higher proportions of CD8 T-cells and lower proportions of resting NK cells in AAA 308
samples. However, a discrepancy was noted in the levels of activated dendritic cells, which 309
were more present in controls in our study but slightly higher in AAA samples in the previous 310
work. Additional single-cell data demonstrated a higher proportion of T follicular helper cells 311
and lower proportions of M1 and M2 macrophages in AAA samples compared to controls[40]. 312
To rule out the effect of death as potential cause of variability, we used a reference work that 313
evaluated the effect of death in blood samples,[41] which found higher levels of resting NK 314
cells and CD8 T-cells in post-mortem samples. These results suggest that the increase in CD8 T-315
cells levels in AAA tissue could be even greater than the one we observed, and that the greater 316
presence of NK resting cells in control samples could be due, in part, to their origin from organ 317
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
15
donors. Consistent with this hypothesis, previous studies have shown that levels of NK cells are 318
higher in the peripheral blood of AAA patients compared to controls,[42] and that these cells 319
play a role in the development of the disease[18]. These results also suggest the existence of a 320
highly cytotoxic environment led by CD8 T-cells in AAA tissue. 321
The lower levels of activated dendritic cells in the AAA samples compared to the controls was 322
unexpected, given their established role as AAA inducers[17,19]. These results suggest that, 323
although dendritic cells may participate in the development of AAA, they are not part of the 324
inflammatory infiltrate. 325
Study of alternative splicing between AAA and controls: 326
This is the first study to elucidate the role of splicing in AAA development. We compared splice 327
events between AAA and controls and identified eleven genes (
SPP1
,
FHL1
,
GNAS,
MORF4L2
, 328
HMGN1
,
ARL1
,
RNASE4
,
A
SAH1
,
CYCS
,
HMGB1
,
SELENOP
) with differentially represented 329
splicing variants. We compared the splice events types identified in our comparison between 330
AAA and controls samples with the presence of splicing events in the whole genome [21] and 331
the observed proportions were comparable to the expected values genome-wide. The most 332
frequent splicing types were alternative first exon (60 %) and skipping exon (20 %), while the 333
least frequent were alternative 5’ splice-site (13.3 %) and mutually exclusive exon (6.67 %). On 334
the other hand, it was surprising that our results did not include alternative last exon, 335
alternative 3’ splice-site, and retained intron, despite their considerable genome-wide 336
frequencies (10.72 %, 9.2 % and 3.54 %, respectively). This may be due to the limited sample 337
size, as only 15 events were identified. 338
We observed seven genes that showed differential expression between AAA and control 339
samples and had a splicing variant significantly more represented in AAA or controls (
SPP1
, 340
FHL1
,
GNAS
,
MORF4L2
,
HMGN1
,
ARL1
,
RNASE4
), indicating that splicing differences could be 341
explaining the observed differential expression. Among these genes,
SPP1
and
FHL1
have been 342
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
16
previously characterized in relation to AAA[43,44], whereas
GNAS
is a new DEG identified 343
between AAA and control tissue. The evidence of differential splicing events validates
GNAS
as 344
a new robust DEG between AAA and control tissue, and suggests that alternative splicing in this 345
gene explains the differential expression and its implication to AAA. Finally,
MORF4L2
,
HMGN1
,
346
ARL1
and
RNASE4
although they have been identified in previous differential expression 347
studies in relation to AAA[10–15], their specific role in AAA has not been studied. For these 348
genes, our results contribute to understand the molecular mechanism leading to differential 349
expression in AAA tissue.
350
SPP1
codes for the osteopontin protein, an important regulator of inflammation that has 351
described functions in cardiovascular diseases[45].
SPP1
is more expressed in AAA tissue than 352
controls, both in animal models and in humans, and it participates in AAA-associated 353
extracellular matrix degradation[46,47] through the nuclear factor kappa B signaling pathway. 354
It is also known that the
SPP1
gene undergoes splicing and gives rise to 3 distinct isoforms 355
osteopontin a, osteopontin b and osteopontin c, with specific characteristics, which have not 356
been characterized in AAA. Consistent with previous data, our results found increased 357
expression in AAA tissue, and identified for the first time that that skipping of exon 3 on
SPP1
358
gene is more frequent in AAA than in controls, suggesting that this form of alternative splicing 359
may be important for the development of AAA. 360
FHL1
codes for a protein that is highly expressed in skeletal and cardiac muscle.
FHL1
has been 361
shown to be a promising blood biomarker for human ascending thoracic aortic aneurysm as a 362
modulator of metalloproteases.[44] Our findings, and those obtained in previous microarray 363
studies[10–13,15], indicate that
FHL1
levels are lower in AAA than in controls. We have 364
detected for the first time that an alternative 5' splicing-site form in this gene occurs more 365
frequently in the control group, suggesting that AAA tissue would have reduced expression of 366
this alternative isoform and reduced levels of
FHL1
, leading to higher risk of AAA development. 367
. CC-BY-NC-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted February 27, 2024. ; https://doi.org/10.1101/2024.02.26.24303384doi: medRxiv preprint
17
Although not previously found in transcriptomic studies, mutations in the
GNAS
gene have 368
been studied in mice for their effect promoting AAA[48].
GNAS
codes for the alpha subunit of 369
the heterotrimeric G stimulatory protein (Gsα). Gsα may play a protective role in AAA 370
development through regulation of vascular muscle tissue and is considered a potential 371
therapeutic target[48]. Consistent with this protective role, our results confirmed lower 372
expression
GNAS
levels in AAA. Moreover, our results add a mechanistic insight by revealing an 373
alternative first exon splicing variant that occurs more frequently in controls and that could 374
increase expression levels of the final protein and protect against AAA. 375
MORF4L2
and
HMGN1
are DNA repair related genes, which had significantly lower and higher 376
expression levels in AAA tissue compared to control tissue, respectively.
MORF4L2
has been 377
associated with atheroma plaque progression in atherosclerosis[49]. Both genes present 378
alternative splicing events that are less frequent in AAA tissue. Further