The authors have declared that no competing interests exist.
Conceived and designed the experiments: KB QT PMT ADK LBO. Performed the experiments: KB PMT. Analyzed the data: PMT MJL. Contributed reagents/materials/analysis tools: PMT ADK. Wrote the paper: PMT.
Hereditary Haemorrhagic Telangiectasia (HHT) is an autosomal dominantly inherited vascular disease characterized by the presence of mucocutaneous telangiectasia and arteriovenous malformations in visceral organs. HHT is predominantly caused by mutations in
Hereditary Haemorrhagic Telangiectasia (HHT) is an autosomal dominantly inherited vascular disease characterized by the presence of mucocutaneous telangiectasia and arteriovenous malformations (AVMs) in visceral organs, primarily the lungs, liver and brain. The most common clinical manifestation is spontaneous and recurrent epistaxis
HHT manifestations are thought to result from an imbalance in the process of angiogenesis. Angiogenesis is the development of new capillary blood vessels from pre-existing vessels and is controlled by different cytokines such as vascular endothelial growth factor (VEGF) and transforming growth factor beta 1 (TGFβ1).
Our knowledge of the non protein-coding part of the human transcriptome is expanding rapidly, and recently attention has shifted towards the most numerous but still poorly understood group—the long non-coding RNAs (lncRNAs). Long non-coding RNAs (lncRNAs) are defined as eukaryote RNAs longer than 200 nucleotides in length, without protein coding capacity. This arbitrary limit of 200 nucleotides distinguishes lncRNAs from small regulatory RNAs, such as microRNA and short interfering RNA (siRNA). Studies indicate that lncRNAs are generated through pathways similar to that of protein-coding genes, with similar histone-modification profiles, splicing signals, and exon/intron lengths
Of the many identified lncRNAs, only few have been characterized functionally
To identify potential HHT therapeutic targets, further knowledge on how a disturbance of the TGF-β signalling pathway leads to HHT manifestations is needed. As lncRNAs are increasingly recognized as key regulators of gene expression we assessed the lncRNA expression in HHT tissue. Thus, the purpose of this study was to investigate the possible involvement of lncRNAs in the molecular pathogenesis of the telangiectasia formation in HHT patients. This was done by expression profiling of lncRNAs in 40 telangiectasial and 40 non-telangiectasial samples from HHT patients using microarray technology.
This study was approved by the ethics committee of Southern Denmark (S-20090131), and the patients provided verbal and written informed consent to participate in this study in accordance with the Declaration of Helsinki.
Nasal mucosal biopsy specimens of telangiectasial and non-telangiectasial tissue in pairs from HHT1 (n = 19) and HHT2 (n = 21) patients were used in this study. The nasal mucosa was anaesthetized with gauze impregnated with Lidocain phenylephrinehydrocloride. No infiltration anaesthesia was used. The samples were collected with a Weil Nasal Forceps by an experienced rhinologist and contained either macroscopicly visible telangiectasia or natural mucosa. All patients showed signs of HHT according to the Curaçao Criteria
Gene | Family No. | Nucleotid change | Amino acid change | Age | Sex | Phenotype |
9 | c.360C>A | p.Tyr120* | 49 | M | E,T,P,G,F | |
9 | c.360C>A | p.Tyr120* | 53 | F | E,T,F | |
13 | c.360C>A | p.Tyr120* | 54 | F | E,T,P,G,F | |
13 | c.360C>A | p.Tyr120* | 30 | F | E,T,P,F | |
15 | c.361-2A>G | p.? | 47 | F | E,T,G,F | |
15 | c.361-2A>G | p.? | 38 | M | E,T,P,F | |
15 | c.361-2A>G | p.? | 40 | F | E,T,F | |
24 | c.360C>A | p.Tyr120* | 40 | F | E,T,F | |
24 | c.360C>A | p.Tyr120* | 46 | F | E,T,P,G,F | |
37 | c.360C>A | p.Tyr120* | 51 | M | E,T,P,F | |
38 | c.360C>A | p.Tyr120* | 37 | F | E,T,F | |
49 | c.821C>T | p.Thr274Ile | 47 | F | E,T,F | |
61 | c.817-3T>G | p.? | 49 | F | E,T,F | |
65 | c.1166_1168delTCT | p.Phe389del | 35 | F | T,P,F | |
87 | c.808C>T | p.Gln270* | 38 | M | T,P,F | |
92 | c.219+1G>T | p.? | 44 | M | E,T,P | |
93 | c.1550_1551delTG | p.Val517Glufs*10 | 37 | M | E,T,F | |
94 | p.1582_1583del | p.Pro528Alafs*38 | 30 | M | E,T,F | |
96 | c.277C>T | p.Arg93* | 46 | M | E,T,P,F | |
8 | c.1120C>T | p.Arg374Trp | 55 | M | E,T,F | |
8 | c.1120C>T | p.Arg374Trp | 28 | M | E,T,F | |
18 | c.1468C>T | p.Gln490* | 43 | M | E,T,F | |
18 | c.1468C>T | p.Gln490* | 44 | F | E,T,P,F | |
18 | c.1468C>T | p.Gln490* | 41 | M | E,T,F | |
20 | c.430C>T | p.Arg144* | 47 | M | E,T | |
42 | c.1135G>A | p.Glu379Lys | 32 | F | E,T,F | |
43 | c.626-3C>G | p.? | 38 | F | E,T,F | |
43 | c.626-3C>G | p.? | 36 | F | T,F | |
46 | c.1013T>A | p.Val338Asp | 39 | M | E,T,F | |
47 | c.1120C>T | p.Arg374Trp | 40 | F | E,T,F | |
56 | c.1-?_1048+?del | p.0? | 44 | F | E,T,F | |
57 | c.143G>A | p.Gly48Glu | 34 | M | E,T,F | |
57 | c.143G>A | p.Gly48Glu | 32 | F | T,F | |
67 | c.266G>A | p.Cys89Tyr | 24 | F | E,T,F | |
67 | c.266G>A | p.Cys89Tyr | 31 | M | E,T,F | |
69 | c.1232G>A | p.Arg411Gln | 39 | M | E,T,F | |
71 | c.143G>A | p.Gly48Glu | 44 | F | E,T,F | |
82 | c.139_140insCG | p.Arg47Profs*8 | 48 | M | E,T,F | |
88 | c.155del | p.Thr52Lysfs*2 | 56 | F | E,T,G,F | |
88 | c.155del | p.Thr52Lysfs*2 | 50 | M | E,T,F |
Abbreviations: AVM, arteriovenous malformation; E, epistaxis; T, telangiectasia; P, pulmonary AVM; C, cerebral AVM; G, gastrointestinal telangiectasia/gastrointestinal bleeding; H, Hepatic AVM; F, family history. M, male; F, female.
Samples were collected in RNA
Sample labelling and array hybridization were performed according to the Two-Color Microarray-Based Gene Expression Analysis—Low Input Quick Amp Labeling—protocol (Agilent Technologies) using the SurePrint G3 Human Gene Expression 8×60 microarray format (Agilent Technologies). Samples were labelled with Cy5 and Universal Human Reference RNA (Stratagene) was labelled with Cy3 and used as a common reference on all arrays.
Agilent Feature Extraction software v. 10.7.3.1 (Agilent technologies) was used to analyse acquired array images. Data were then within-array normalized by Loess normalization method and between-array normalized by Quantile normalization. The normalized values were used to calculate log2 transformed Cy5/Cy3 ratios. Replicate probes were collapsed calculating the median. Missing expression values were imputed by
All the 42,164 probes of the Agilent SurePrint G3 array were re-annotated using GENCODE v.16 gene annotation database (
Data analyses were performed using the Qlucore Omics Explorer 2.3 software (Qlucore). Differentially expressed lncRNAs comparing telangiectasial and non-telangiectasial tissue were ranked according to statistical significance determined by two-group comparison (paired t-test). This was done for the groups HHT1 and HHT2 seperately and subsequently for the total group of HHT. Multiple testing was adjusted for by the Benjamini-Hochberg method. Differentially expressed lncRNAs were chosen for further evaluation (q<0.15). Principal component analysis (PCA) and hierarchical clustering were performed in Qlucore Omics Explorer 2.3 to examine whether telangiectasial and non-telangiectasial samples could be separated.
To assess the potential
To identify potential
All the 42,164 probes of the Agilent SurePrint G3 array were re-annotated using GENCODE v.16 gene annotation database (
We detected 617 statistically significantly differentially expressed lncRNAs (q<0.05, paired t-rest) when comparing telangiectasial and non-telangiectasial samples in the total HHT group (80 paired samples), of which 42 were highly statistically significant (q<0.001). Of the 42 differentially expressed lncRNAs (listed in
Ensembl gene ID | gencode.v16_GenomicCoordinates | q-value (FDR) | Fold change | Gene biotype | HGNC symbol |
ENSG00000249772.1 | chr5:80409204-80410671_R | 1.83E-06 | 0.86 | antisense | - |
ENSG00000230544.1 | chr13:114586640-114588308_F | 1.50E-05 | 0.88 | lincRNA | LINC00453 |
ENSG00000215231.3 | chr5:5034472-5070117_F | 1.50E-05 | 0.88 | lincRNA | - |
ENSG00000237548.1 | chr9:124646915-124725998_R | 1.50E-05 | 0.86 | processed_transcript | TTLL11-IT1 |
ENSG00000263753.1 | chr18:5232875-5246507_F | 0.00011 | 0.90 | lincRNA | LINC00667 |
ENSG00000231133.1 | chr20:61727150-61733631_R | 0.00014 | 0.90 | processed_transcript | HAR1B |
ENSG00000256218.1 | chr12:5475214-5476940_R | 0.00015 | 0.90 | lincRNA | - |
ENSG00000241269.1, ENSG00000188365.3 | chr7:5459458-5462753_F | 0.00022 | 0.85 | antisense | - |
ENSG00000226496.1 | chr21:42513427-42520060_R | 0.00023 | 1.14 | antisense | LINC00323 |
ENSG00000248176.1 | chr4:29119930-29204392_F | 0.00039 | 0.89 | lincRNA | - |
ENSG00000259484.1 | chr15:57151866-57210697_R | 0.00039 | 0.86 | processed_transcript | - |
ENSG00000206129.3 | chr18:53670844-53858493_R | 0.00042 | 1.11 | lincRNA | - |
ENSG00000235285.1 | chr13:44720606-44732358_R | 0.00042 | 0.90 | sense_intronic | SMIM2-IT1 |
ENSG00000147753.5 | chrY:6317509-6325947_F | 0.00042 | 1.13 | lincRNA | TTTY7 |
ENSG00000240453.1 | chr1:745489-753092_R | 0.00042 | 0.84 | processed_transcript | - |
ENSG00000259150.1 | chr15:26360960-26378184_F | 0.00042 | 1.09 | lincRNA | LINC00929 |
ENSG00000255471.1 |
chr11:86603256-86636079_R | 0.00045 | 0.81 | antisense | - |
ENSG00000237036.3 | chr10:31596646-31608810_R | 0.00056 | 1.11 | antisense | - |
ENSG00000233154.1 | chr1:116966346-117021464_R | 0.00065 | 0.91 | lincRNA | - |
ENSG00000197251.3 | chr6:33553883-33561115_R | 0.00068 | 1.10 | antisense | LINC00336 |
ENSG00000248176.1 | chr4:29119930-29204392_F | 0.00073 | 0.92 | lincRNA | - |
ENSG00000254154.3 | chr1:177897923-178007142_R | 0.00073 | 0.88 | processed_transcript | - |
ENSG00000215374.4 | chr8:7159133-7212876_R | 0.00073 | 0.90 | processed_transcript | - |
ENSG00000196096.3 | chr2:214141276-214148929_R | 0.00073 | 0.92 | processed_transcript | - |
ENSG00000259758.1 | chr8:141530255-141539600_R | 0.00073 | 1.13 | lincRNA | CASC7 |
ENSG00000264772.1, ENSG00000265500.1 | chr17:7466604-7485342_F | 0.00074 | 1.26 | processed_transcript | - |
ENSG00000229563.1 | chrX:45364633-45489447_F | 0.00076 | 1.10 | processed_transcript | - |
ENSG00000203325.3 | chr1:32517892-32539075_R | 0.00076 | 0.88 | antisense | - |
ENSG00000232956.3 |
chr7:45022622-45026560_R | 0.00076 | 1.20 | lincRNA | - |
ENSG00000232021.2 |
chr4:109088681-109177992_F | 0.00076 | 0.90 | processed_transcript | LEF1-AS1 |
ENSG00000250195.1 | chr4:139741108-139933800_R | 0.00076 | 1.21 | antisense | - |
ENSG00000250608.1 | chr3:131043936-131100319_R | 0.00076 | 1.11 | processed_transcript | - |
ENSG00000266952.1 | chr18:61880317-61927290_R | 0.00078 | 0.91 | lincRNA | - |
ENSG00000259334.1 | chr14:24391456-24403777_R | 0.00079 | 1.11 | lincRNA | LINC00596 |
ENSG00000249364.1 | chr5:66675206-67101066_F | 0.00079 | 0.91 | processed_transcript | - |
ENSG00000231185.2 |
chr5:141704858-142051566_F | 0.00086 | 0.93 | processed_transcript | - |
ENSG00000232046.1 | chr2:66801162-66957289_F | 0.00086 | 0.91 | processed_transcript | - |
ENSG00000230133.1 | chr20:24180403-24205224_F | 0.00086 | 0.90 | lincRNA | - |
ENSG00000215808.2 | chr1:238643684-238649323_R | 0.00090 | 1.11 | processed_transcript | - |
ENSG00000135253.9 | chr7:128502505-128550773_R | 0.00094 | 1.08 | processed_transcript | KCP |
ENSG00000233251.3 | chr2:56400669-56412905_R | 0.00094 | 1.09 | processed_transcript | - |
ENSG00000245910.3 | chr8:67833919-67838633_R | 0.0010 | 0.85 | processed_transcript | SNHG6 |
*Are part of the lncRNAs identified by GREAT analysis.
Analysis of the HHT1 samples resulted in 70 differentially expressed lncRNAs (q<0.05, paired t-test) when comparing telangiectasial and non-telangiectasial samples, and in HHT2 analysis resulted in 114 differentially expressed lncRNAs (q<0.05, paired t-test). The number of common lncRNAs (q<0.05) in the three groups is rather small as shown in
The number of statistically significantly differentially expressed long non-coding RNAs in the three groups (q<0.05). The numbers in the overlapped areas indicate the lncRNAs that are differentially expressed in the multiple groups.
The complete results of the paired t-tests in HHT and the subgroups HHT1 and HHT2 are provided in
PCA applied to the differentially expressed lncRNAs (q<0.15) revealed a clear separation of the telangiectasial and non-telangiectasial samples, regarding HHT1 (
a. PCA applied to the differentially expressed long non-coding RNAs (lncRNAs)(paired t-test, q<0.15) revealed a clear separation of the telangiectasial and non-telangiectasial samples, regarding HHT1. b. PCA applied to the differentially expressed lncRNAs (paired t-test, q<0.15) revealed a clear separation of the telangiectasial and non-telangiectasial samples, regarding HHT2. c. PCA applied to the differentially expressed lncRNAs (paired t-test, q<0.15) revealed a clear separation of the telangiectasial and non-telangiectasial samples in the total group of HHT. There is no clustering into subgroups.
Hierarchical clustering: amongst the differentially expressed lncRNAs (q<0.001) in the HHT group both up- and down-regulation was observed (
Hierarchical clustering for the 80 HHT paired samples comparing the expression in telangiectasial (yellow) and non-telangiectasial (pink) tissue, using the top variant lncRNAs (n = 42)(q<0.001). In the heat map rows correspond to long non-coding RNAs (lncRNAs) and columns to samples. Red indicates elevated expression, green indicates reduced expression. In the 42 differentially expressed lncRNAs; 16 are up-regulated and 26 are down-regulated.
The differentially expressed lncRNAs with q<0.15 (n = 617) were loaded into GREAT. Of these, 31% map within 50 kilobases (kb) of an annotated gene, and 89% map within 500 kb of a gene (
Graphical view of the gene ontology (GO) terms tree showing part of the GO terms in the ontology biological process and their connections. Our three GO terms; blood vessel development, blood vessel development, and vasculogenesis (in red boxes), are the more narrow and specific terms in the tree and overlap by a number of genes.
Ontology | Term name | Term ID | Binom FDR Q-value | Binom fold enrichment | Hyper FDR Q-value | Hyper fold enrichment |
GO Biological Process | blood vessel morphogenesis | GO:0048514 | 9.21e-7 | 2.29 | 4.59602e-5 | 2.72 |
GO Biological Process | blood vessel development | GO:0001568 | 6.81e-6 | 2.07 | 5.89303e-5 | 2.52 |
GO Biological Process | immune system development | GO:0002520 | 7.30e-5 | 2.06 | 5.60561e-3 | 2.06 |
GO Biological Process | hemopoietic or lymphoid organ development | GO:0048534 | 1.74e-4 | 2.04 | 1.14257e-2 | 2.04 |
GO Biological Process | hemopoiesis | GO:0030097 | 4.03e-4 | 2.06 | 1.65733e-2 | 2.06 |
GO Biological Process | vasculogenesis | GO:0001570 | 6.00e-4 | 3.43 | 5.61511e-4 | 4.90 |
GO Blood vessel morphogenesis (HHT1) | GO Blood vessel development (HHT1) | GO Vasculogenesis (HHT1) | GO Vasculogenesis (HHT2) | GO Vasculogenesis (HHT) | GO Blood vessel morphogenesis (HHT) | Gene located on Chromosome | Foldchange (ENSG) |
2 | 1.09 | ||||||
8 | 1.13 | ||||||
14 | 1.06–1.07 | ||||||
14 | 0.92–0.93 | ||||||
7 | |||||||
7 | |||||||
16 | 0.89 | ||||||
6 | 0.92–0.93 | ||||||
10 | 0.89–0.92 | ||||||
2 | 1.10 | ||||||
2 | 0.89–0.90 | ||||||
5 | 0.09–1.16 | ||||||
2 | 0.88–1.22 | ||||||
16 | |||||||
13 | 1.11 | ||||||
11 | |||||||
15 | 1.09 | ||||||
12 | 0.91–0.92 | ||||||
3 | 0.91–1.14 | ||||||
10 | 1.09 | ||||||
12 | 1.19–1.20 | ||||||
20 | 1.27 | ||||||
11 | 0.95 | ||||||
4 | 0.87–0.90 | ||||||
5 | 1.09–1.11 | ||||||
2 | 0.57–1.14 | ||||||
15 | 0.93 | ||||||
15 | 0.92 | ||||||
10 | 1.09 | ||||||
2 | 0.94–0.95 | ||||||
X | 1.08 | ||||||
3 | 1.08–1.20 | ||||||
11 | |||||||
5 | |||||||
7 | 0.88–1.09 | ||||||
6 | 0.92–1.12 | ||||||
12 | 0.92–0.94 | ||||||
6 | 1.16–1.28 | ||||||
1 | 1.08–1.13 | ||||||
15 | 0.93–1.17 | ||||||
3 | |||||||
6 | 1.07–1.24 | ||||||
14 | 0.92 | ||||||
7 | 0.93 | ||||||
3 | 0.91–1.10 | ||||||
22 | 0.90 | ||||||
1 | 0.92 | ||||||
8 | 1.13–1.23 | ||||||
10 |
q-values can be seen of tables S4–S9. Genes that are present for all six GO–terms are highlighted in bold.
The differentially expressed lncRNAs with q<0.15 (n = 640) were loaded into GREAT. Of these, 31% map within 50 kilobases (kb) of an annotated gene, and 89% map within 500 kb of a gene (
Ontology | Term name | Term ID | Binom FDR Q-value | Binom fold enrichment | Hyper FDR Q-value | Hyper fold enrichment |
GO Biological Process | vasculogenesis | GO:0001570 | 1.90e-5 | 3.80 | 6.20e-5 | 4.91 |
GO Biological Process | positive regulation of neuron differentiation | GO:0045666 | 4.95e-5 | 3.13 | 4.51e-4 | 4.41 |
GO Biological Process | regulation of mesenchymal cell proliferation | GO:0010464 | 3.90e-4 | 3.28 | 3.45e-3 | 5.00 |
GO Biological Process | regulation of DNA binding | GO:0051101 | 7.35e-4 | 3.24 | 2.21e-5 | 5.00 |
GO Biological Process | regulation of binding | GO:0051098 | 8.32e-4 | 2.43 | 1.13e-5 | 3.38 |
The differentially expressed lncRNAs with q<0.05 (n = 617) were loaded into GREAT. A more stringent cut-off of q<0.05 was applied here, as the sample size was larger. Of these lncRNAs, 31% map within 50 kilobases (kb) of an annotated gene, and 90% map within 500 kb of a gene (
Ontology | Term name | Term ID | Binom FDR Q-value | Binom fold enrichment | Hyper FDR Q-value | Hyper fold enrichment |
GO biological process | vasculogenesis | GO:0001570 | 1.71e-5 | 3.94 | 4.92e-5 | 5.21 |
GO biological process | lactation | GO:0007595 | 2.06e-5 | 6.51 | 9.57e-3 | 4.85 |
GO biological process | blood vessel morphogenesis | GO:0048514 | 1.92e-4 | 2.01 | 4.06e-4 | 2.40 |
By GREAT analysis, 63 lncRNAs were identified, which is equal to 10.2% of the differentially expressed lncRNAs in the HHT group (n = 617, q<0.05). Concordingly, 89.8% of statistically significantly differentially expressed lncRNAs in the HHT group are not part of the three selected statistically significant GO terms in GREAT.
To identify potential
Manhattan plots showing significance of correlation between the three top statistically significantly differentially expressed long non-coding RNAs' (lncRNAs) expression and expression of all other genes at the microarray. The negative logarithm (-log10) p-values of the Pearson correlation were plotted across chromosomes. The Bonferoni-corrected significance level is indicated by the dashed line (p<0.05). a. ENSG00000249772.1 (Chromosome 5) has 144 statistically significantly correlated transcripts, of which 62.5% are other lncRNAs. b. ENSG00000230544.1 (Chromosome 13) has 178 statistically significantly correlated transcripts, of which 61% are other lncRNAs. c. ENSG00000215231.3 (Chromosome 5) has 158 statistically significantly correlated transcripts, of which 63% are other lncRNAs. For comparison, only 11% of the probes across the microarray map to lncRNAs.
To our knowledge, this is the first study to assess the regulatory effects of long noncoding RNAs (lncRNAs) in HHT affected tissue. Using microarray technology, we identified lncRNAs that are statistically significantly differentially expressed in HHT telangiectasial tissue compared with HHT non-telangiectasial nasal tissue. Using GREAT, a tool which assumes
The lncRNA list used in this study was retrieved from the GENCODE v.16 dataset, which uses a combination of manual annotation, computational analysis and targeted experimental validation and is the largest catalogue of human lncRNAs to date. It was used in order to retrieve optimal lncRNA annotation. To minimize the number of probes mapping only or mostly to mRNAs, we filtered out probes mapping to overlapping mRNA exons and lncRNA introns.
LncRNAs appear to control expression of protein-coding genes through both
In most of the cases of multiple lncRNAs neighbouring a gene, a clear correlation of direction was observed. However, a correlation of direction was not present in all cases, which could indicate that a more fine-tuned regulation is being played out. Each GO term involved both up- and downregulated lncRNAs, which may suggest local
The three GO terms contain repeatedly mentioned genes. Accordingly, that indicates a central group of genes that seems to be
Another approach in the data analysis was the correlation study. The correlations between the expression levels of the 10 highest ranking differentially expressed lncRNAs in the HHT total group and every transcript present in the Agilent microarray were assessed by calculating Pearson's correlation coefficient. Multiple statistically significant correlations, mapping to many different chromosomes, were present for each lncRNA in question. Hence, the correlation analyses may indicate that multiple
In 2007, two gene expression microarray studies were published describing HHT samples.
The strength of our study is that affected HHT tissue was analysed in a paired design with samples from a relatively large number of patients. This resulted in a large number of differently expressed lncRNAs, even though we applied a strict cut-off. GREAT is a useful tool to relate lncRNAs to known biology, but it is limited to
In summary, our study identified lncRNAs that are aberrantly expressed in HHT telangiectasia and indicates that lncRNAs may contribute to regulate protein-coding loci in HHT, suggesting that the lncRNA component of the transcriptome deserves more attention in HHT research. Thus, a deeper understanding of lncRNAs and their role in telangiectasia formation possesses potential for discovering therapeutic targets and for identifying new biomarkers.
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(TIFF)
(XLS)
(XLS)
(XLS)
(DOCX)
We thank the patients and families for their participation in the study. The technical assistance of Jette Moeller is greatly appreciated.