Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

A Boolean approach for novel hypoxia-related gene discovery

  • Tsering Stobdan,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing

    Affiliation Department of Pediatrics, Division of Respiratory Medicine, University of California San Diego, La Jolla, California, United States of America

  • Debashis Sahoo,

    Roles Conceptualization, Data curation, Methodology, Software, Supervision

    Affiliations Department of Pediatrics, Division of Respiratory Medicine, University of California San Diego, La Jolla, California, United States of America, Department of Computer Science and Engineering, University of California San Diego, La Jolla, California, United States of America

  • Gabriel G. Haddad

    Roles Conceptualization, Funding acquisition, Investigation, Project administration, Resources, Supervision, Visualization, Writing – review & editing

    Affiliations Department of Pediatrics, Division of Respiratory Medicine, University of California San Diego, La Jolla, California, United States of America, Department of Neurosciences, University of California San Diego, La Jolla, California, United States of America, Rady Children’s Hospital, San Diego, California, United States of America


Hypoxia plays a major role in the etiology and pathogenesis of most of the leading causes of morbidity and mortality, whether cardiovascular diseases, cancer, respiratory diseases or stroke. Despite active research on hypoxia-signaling pathways, the understanding of regulatory mechanisms, especially in specific tissues, still remain elusive. With the accessibility of thousands of potentially diverse genomic datasets, computational methods are utilized to generate new hypotheses. Here we utilized Boolean implication relationship, a powerful method to probe symmetrically and asymmetrically related genes, to identify novel hypoxia related genes. We used a well-known hypoxia-responsive gene, VEGFA, with very large human expression datasets (n = 25,955) to identify novel hypoxia-responsive candidate gene/s. Further, we utilized in-vitro analysis using human endothelial cells exposed to 1% O2 environment for 2, 8, 24 and 48 hours to validate top candidate genes. Out of the top candidate genes (n = 19), 84% genes were previously reported as hypoxia related, validating our results. However, we identified FAM114A1 as a novel candidate gene significantly upregulated in the endothelial cells at 8, 24 and 48 hours of 1% O2 environment. Additional evidence, particularly the localization of intronic miRNA and numerous HREs further support and strengthen our finding. Current results on FAM114A1 provide an example demonstrating the utility of powerful computational methods, like Boolean implications, in playing a major role in hypothesis building and discovery.


Oxygen is vital to the living cells, especially critical in high-energy requiring tissues like brain, heart, liver and kidneys, and therefore it plays a dominant role in the pathogenesis and pathophysiology of most of the major diseases [13]. Since an impaired oxygen supply (hypoxia) is the basis for these diseases i.e., cardiac ischemia, stroke and chronic obstructive pulmonary disease (COPD), all advances in the treatments are focused on methods that could maintain a steady O2 supply, little is known about treating or preserving the affected cells, especially the terminally differentiated cardiomyocytes and neurons [4, 5]. Although the identification of hypoxia inducible factor (HIF), the central regulator of hypoxia [6], and numerous other hypoxia related genes, have recognized potential therapeutic targets [7, 8], our knowledge on the molecular mechanisms that can be targeted to protect or increase the tolerance of these cells to hypoxia is primitive.

The conventional approaches like the candidate gene assessment or ‘omics’ approaches such as genomics, transcriptomics, proteomics and metabolomics, so far have helped us to identify critical targets [7]. However, it is clear that identification of new or next generation of targets will require data from diverse samples types and enormous sample numbers and a newer method to analyze. The emerging tools of network analysis or systems biology that utilizes data from databases that has large-scale diverse sample types, offers a platform to understand the complexity and to visualize and prioritize targets [9, 10], which addresses the current challenges to some extent. Since most of these tools are based on the symmetric relationship between a pair of genes [9, 10], overlooking the larger set of the asymmetric relationship, newer approaches are evolving to identify both symmetric and asymmetric relationships [11]. In the current study, our goal was to identify novel hypoxia-related gene using Boolean implications relationship i.e., that follow simple “if-then” rules, [11, 12]. For this we used MiDReG (mining developmentally regulated genes), a bioinformatics method initially developed that identifies developmentally regulated genes but has the potential to examine the transcriptional relationships, symmetric and asymmetric, between any known genes from thousands of microarray experiments.


Boolean implication analysis predicts hypoxia-responsive genes

In order to identify novel hypoxia-responsive genes, we utilized more than 25,955 publicly available human dataset and perform Boolean implication analysis to filter candidate genes that have a Boolean relationship with the seed gene VEGFA (vascular endothelial growth factor A, Gene ID: 7422, Fig 1A) i.e., a well-established hypoxia responsive/sensitive gene. Out of the four probesets (cDNA fragment or oligonucletides that represent genes) targeting VEGFA i.e., 210512_s_at, 210513_s_at, 211527_x_at and 212171_x_at, we used probeset 210512_s_at in our primary analysis because of its robust signal. The software MiDReG (mining developmentally regulated genes) which couples gene expression patterns with “if-then” rules (Boolean implications) was used to predict functionally important VEGFA/hypoxia-related genes relationships [11, 12]. At a BooleanNet statistical threshold of SThr > 40 and pThr < 0.2 MiDReG detected 310 probesets (230 genes) that pairs with VEGFA (S1 Table). Out of the six potential gene relationships discovered by BooleanNet, two symmetric (Equivalent and Opposite) and four asymmetric, the 310 probesets show only five Boolean relationship and no Boolean Opposite relationship (S1 Table). For example, out of 230 genes two genes shows high = >high implication, one shows high = >low implication, three shows low = >high implication and 216 shows low = >low implication with VEGFA. We picked 22 probesets (19 genes) that have a strong relationship with VEGFA (S2 Table). Boolean relationships here include ‘equivalent’ for SEPT10, TEAD1, FKBP9; ‘high = >low’ for RASGRP2; ‘low = >high’ for VAV1, PRKCB1, MFNG, CUGBP2; ‘low = >low’ for YAP1, SH3D19, FAM114A1, EMP2, LGR4, NR2F2, PPIC, ERRFI1 and high = >high for LAPTM4B and ITGAV. Interestingly, eleven genes (85%) were previously known to be hypoxia related (S2 Table).

Fig 1. Boolean implication analysis to identify novel hypoxia-responsive genes.

(a) Schematics of approach utilized to filter hypoxia-related genes. We used VEGFA as the seed gene to identify genes responding to hypoxia. Filter includes ranking, conserved relationship in mice and novelty. (b—e) Scatter plots depicting the expression profile of 25,955 samples for VEGFA (probset 210512_s_at, x-axis) and the additional VEGFA probset 210513_s_at (c) and top correlated genes i.e., SDC4 (d), ERRFI1 (e) and ITGAV (f). (f—i) Validation of candidate seed gene (VEGFA) and SDC4 (h), ERRFl1 (i) and ITGAV (j) in HPAECs in normoxia (21% O2) and at 2, 8, 24 and 48 hours of constant 1% O2. (*, P<0.05).

We subsequently validated the hypoxia response of some of the candidate genes discovered, along with the seed gene, VEGFA, in human pulmonary artery endothelial cells (HPAECs) exposed to 1% O2 environment for 2, 8, 24 and 48 hours (Methods). The results showed an upregulation of VEGFA and some of the candidate genes at the 8 hours of hypoxia treatment (Fig 1). For example, the expression levels of VEGFA and the tested genes SDC4, ERRFI1 and ITGAV increased significantly after 8 hours of 1% O2 (Fig 1). As anticipated, the expression of HIF1a does not change during the course of hypoxia (S1 Fig).

Boolean implication analysis identifies FAM114A1 as a novel hypoxia-related gene

At BooleanNet statistical value of SThr = 48.06 and pThr  =  0.08, FAM114A1 is the top candidate with Boolean implication VEGFA low = > FAM114A1 low relationship (Fig 2A, S1 Table). Furthermore, analysis of the mouse data using 11,758 publicly available samples also reveal a similar relationship i.e., Vegfa high = > 9130005N14Rik high (Fig 2B). In order to validate the hypoxia sensitivity of FAM114A1, we measured its transcript levels in HPAECs treated with 1% O2 for 2, 8, 24 and 48 hours. The expression level increases significantly at 8, 24 and 48 hours of hypoxia treatment (P<0.05; Fig 2C) which is consistent with upregulation of VEGFA post 8 hours of hypoxia (Fig 1F). However, the fold change in FAM114A1 is modest, i.e., 1.18, 1.3 and 1.4 fold increase, when compared to the VEGFA where the fold change is 4.8, 8.8 and 8.3 fold at 8, 24 and 48 hours respectively. Interestingly, at around 2 hours of 1% O2 treatment the FAM114A1 levels are significantly low (P<0.05; Fig 2C).

Fig 2. Identification of FAM114A1 as a novel hypoxia responsive gene.

(a) Scatter plot depicting the expression profile of VEGFA (probe ID, 210512_s_at; x-axis) and FAM114A1 (probe ID, 226697_at; y-axis) in n = 25,955 ‘human’ samples. (b) Scatter plot depicting the expression profile of Vegfa (probe ID, 1420909_at; x-axis) and Fam114a1 (9130005-N14Rik, probe ID, 1417272_at; y-axis) in n = 11,758 ‘mice’ samples (Correlation = 0.451606). (c) RT-PCR results depicting expression level of FAM114A1 under normoxia (21% O2) and at 2, 8, 24 and 48 hours of constant 1% O2 measured in HPAEC. Insert, gene expression dynamics of VEGFA and FAM114A1 along the boolean path. (*, P<0.05).

Distinctive properties of FAM114A1 and its role in hypoxia

Although the gene was first identified in the nervous system (Noxp20), [13] its expression is low in the nervous system (Fig 3A). Therefore, in order to better understand the role of FAM114A1, particularly under hypoxic environment, we systematically examined its distinctive properties.

Fig 3. Distinctive properties of FAM114A1 and its role in hypoxia.

(a) FAM114A1 is expressed in most of the tissues but its expression is low in the nervous system. (b) Position of hypoxia-response elements (HREs) consenses in FAM114A1 and intronic miRNA, miR-574 in intron-1. The region constituting miR-574 is evolutionarily conserved. (c) Network analysis reveals physical interaction of FAM114A1 with AKT1S1 (AKT1 substrate 1) and N4BP1 (NEDD4 binding protein 1).

Sequence analyses of FAM114A1 indicate numerous hypoxia-response elements (HREs, 5′-RCGTG-3′) [14] that are located in the promoter and 5’ untranslated region (Fig 3B). Additionally, intron-1 has five HREs of which four are part of an independently transcribed intronic miRNA, miR-574 (at positions chr4:38,868,032–38,868,127, GRCh38/hg38; Fig 3B). Further, the region containing miR-574 is evolutionarily conserved (Fig 3B) and since miR-574-3p is reported to reduce VEGFA [15], we believe that there is a functional link between FAM114A1 and VEGFA.

The gene network estimation analysis, using GeneMANIA, reveals FAM114A1 having a shared protein domain with its homologue FAM114A2 (Fig 3C). The top 20 FAM114A1 interacting genes include KDELR2, KDELR3, P4HA2 and PON2 which are part of the broader list of genes having Boolean implication ‘low = > low’ with VEGFA. Beside this, the interaction analysis reveals AKT1S1 (AKT1 substrate 1 or proline-rich Akt substrate of 40 kDa (PRAS40)) and N4BP1 (NEDD4 binding protein), as the two most physically interacting genes (Fig 3C).


In the present study, we used a Boolean implication relationship on publicly available microarray datasets and identified FAM114A1 (Family with Sequence Similarity 114 Member A1) as a novel hypoxia related gene (Fig 2). Since the computational approach is reliant on the dynamic range of expression of the seed gene, we used VEGFA as seed gene due to its robust hypoxia sensitivity [16], and checked the Boolean implications of VEGFA to the other genes. We observed previously known hypoxia-related gene (>85% hypoxia related genes), which is important since such findings validate to a degree our novel findings [11]. The two important advantages of this approach a) is the coverage of a wide range of samples from a publicly available datasets of any conventional experiments and b) this method can identify important genes that are functionally conserved in humans and mice.

The Boolean implication ‘VEGFA low⇒FAM114A1 low’ enabled us to hypothesize a relationship at a molecular level, particularly in certain specific cell types. As expected, we discovered a putative relationship of similar pattern i.e., simultaneous upregulation of both VEGFA and FAM114A1 at 8 hours of 1% O2 (Fig 2). We also uncovered similar responses in some of the previous studies [17, 18]. For example, both genes were significantly upregulated in HPAECs and cardiac microvascular endothelial cells treated with 1% O2 (S2 Fig) [17]. Importantly, the hypoxia-induced upregulation was not immediate, much like in previous studies [17]. Similar responses were also revealed in lymphatic endothelial cells cultured in 1% O2 for 24 hours [19] and Burkitt’s lymphoma cell line P493-6 cultured at 0.1% O2 for 29 hours [18], all having common cell lineage. By contrast, FAM114A1 fails to respond in a similar way when endothelial cells are exogenously treated with VEGF (S3A and S3B Fig) [20]. Further, the relationship was absent in human primary renal proximal tubule epithelial cells treated with 1% O2 for 24 h (S3C and S3D Fig) [21], indicating an endothelial cell type-dependent response. It is worth noting that most of these observations were not highlighted in the previous studies. For example, the fold change of FAM114A1 was not among the top 25 genes with >4 fold change as reported in Kim et al., (S3E and S3F Fig) [18], or the differences are revealed only after we normalized Affymetrix platform using RMA (Robust Multichip Average) [11, 17, 22, 23].

Additional evidence that supports and strengthens our discovery of FAM114A1 as a hypoxia-related gene includes a) the presence of numerous HREs in its promoter and 5’ untranslated region (Fig 3B), which are hypoxia-inducible factor (HIF) binding site that lead to hypoxia-induced transcriptional response in the target genes [24, 25] and b) the presence of an evolutionarily conserved and functionally related miRNA-574 embedded in the intron-1 of FAM114A1, miRNA-574, which negatively regulates VEGFA translation (Fig 3B) [15].

To distinguish whether FAM114A1 upregulation is a hypoxia related or predominantly related to VEGFA activation, we explored into conditions involving non-hypoxia related VEGFA activation such as in cell growth, apoptosis, cell proliferation and tumor development [26]. A similar relationship was observed during neuronal differentiation of human SH-SY5Y neuronal cells (S4A and S4B Fig) and in methotrexate sensitive (low) vs resistant (high) HT29 colon cancer cells (S4C and S4D Fig) [27, 28]. A recent study report an increase FAM114A1 expression in the failing heart [29], which we believe could be related to VEGFA [30]. However, VEGFA expression does not change when FAM114A1 is upregulated in breast cell line subpopulation (MDA-MB-231) (S4E and S3F Fig) [31]. Overall, these results indicate that the interaction between VEGFA and FAM114A1 is typically maintained in endothelial or closely related cell lineages when exposed to hypoxic conditions but seldom in non-endothelial and non-hypoxic conditions.

When we explored previous studies on FAM114A1 to understand its function, particularly under hypoxia, we found that apart from its role in neuronal cell development [13], melanocyte apoptosis [32] and an association with ankylosing spondylitis [33], there was very little known about this gene or its homologue FAM114A2. A recent study in Fam114a1−/− mice showed that cardiac function was markedly restored in the knockout mice, in angiotensin II infusion model of hypertension, when compared to the controls [29], indicating its role in oxidative stress. Although, this study reported that FAM114A1 regulated the expression of angiotensin type 1 receptor (AGT1R), there was not an apparent link between the two genes in the conventional network analysis (Fig 3C). Interestingly, we observed a Boolean implication FAM114A1 low = > AGTR1 low (S5 Fig), clearly highlighting the importance of our approach. On further exploring AGTR1 related pathway, i.e., Renin Angiontensin system, we noticed that FAM114A1 has a Boolean relationship with ACE2 (angiotensin-converting enzyme 2 also a functional receptor on cell surfaces through which SARS-CoV-2 enters the host cells) and AGT (angiotensinogen) but not with ACE (angiotensin-converting enzyme) and AGTR2 (angiotensin type 2 receptor) (S5 Fig). This is highly informative because a Boolean implications of FAM114A1 with ACE2 and AGTR1 and ‘no relation’ with the leading candidate of RAS i.e., ACE, would depict its preferential interaction or regulation of RAS, especially when ACE and ACE2 has opposing role in the same pathway [34, 35]. Additionally, both ACE2 and AGTR1 are upregulated in hypoxia [36, 37] and both has role in failing human heart [29, 38].

One of the previous study labelled the protein product of this gene as ‘nervous system over-expressed protein 20’ (NOXP20), detected from in situ hybridization of mouse cryostat sections [13]. However, the transcript levels in different tissue-wide expression datasets e.g., GTEx, Illumina bodyMap2 transcriptome etc., (Fig 3A) is lowest in the nervous system (brain and spinal cord) when compared to all the other tissues [39, 40]. We therefore call this gene OXSI1 (oxidative stress induced 1), representing the noticeable characteristic from ours and numerous other studies. The lack of any supporting literature led us to focus on gene network estimation, which indicates that FAM114A1 closely interact with AKT1S1 (PRAS40), a subunit of mTORC1, and N4BP1. In the context to our current finding, AKT1S1 is known to be activated by hypoxia [41]. Its activation in the nerve cells protect neuronal cell from damage [41] and in endothelial cells, it suppresses atherogenesis [42], both through inhibition of mTORC1 signaling. Interestingly, both AKT1S1 and N4BP1 are known to oppositely regulate NF-κB transcriptional activity. While N4BP1 inhibits TLR-dependent (Toll-Like Receptor-dependent) activation of NF-κB by interacting with IκB kinase γ [43, 44], AKT1S1 promotes NF-κB transcriptional activity through its association with p65 [45]. While the activation of NF-κB in hypoxia is well known, the several proposed mechanisms still needs consolidation [4649]. Additionally, the genomic proximity of FAM114A1 with some of the TLR genes i.e., TLR1, TLR6 and TLR10, specifically TLR6, which has an overlapping promoter region but transcribing in opposite direction, may indicate that genes in proximity are driven by their shared biological function [50]. After appreciating the fact that FAM114A1 interacts with two oppositely regulating entities of NF-κB one can anticipate the critical role it may play under different conditions. Therefore, future studies on FAM114A1, including its potential role in NF-κB signaling through its interaction with ATK1S1 and/or N4BP1, is critical.

Overall, we provide a computational method for identifying hypoxia related gene that bypasses the conventional approach that are time consuming and costly. Since the method utilizes expression data from thousands of diverse samples, it holds the potential to reveal novel candidate markers. Our result suggest that FAM114A1 is a hypoxia related gene, with a role in oxidative stress and several additional evidences, including it hosting a hypoxia related miRNA, support this observation. However, due to the lack of sufficient literature on FAM114A1, particularly our findings on the network interaction with NF-κB, it will be critical to further investigate mechanisms underlying its activation and its implications in physiology and pathophysiology.


Boolean implications analysis

We utilized more than 25,955 publicly available human dataset to perform Boolean implication analysis. The seed gene was VEGFA (vascular endothelial growth factor A, Gene ID: 7422, Fig 1A) which is a well-established hypoxia responsive/sensitive gene. We explored ANGPTL4, PPARG, PTGIS and INHBA as potential seed genes, one at a time, because of their higher fold change in HPAECs, as seen in previous microarray study, when exposed to hypoxia [51]. However, since the basal expression of these genes was low in most of the tissues (S6 Fig), we chose VEGFA as a more appropriate seed gene for hypoxia related gene discovery, especially when using endothelial cells for further validation. Interestingly, when we individually used ANGPTL4, PPARG, PTGIS and INHBA as the seed genes, FAM114A1 was the only common gene in the top 100 Boolean related genes listed for each seed gene (S7 Fig).

Out of the four probesets targeting VEGFA i.e., 210512_s_at, 210513_s_at, 211527_x_at and 212171_x_at, we used probeset 210512_s_at in our primary analysis because of its robust signal. The software MiDReG (mining developmentally regulated genes) which couples gene expression patterns with “if-then” rules (Boolean implications) was used to predict functionally important VEGFA/hypoxia-related genes [11, 12]. The BooleanNet statistical threshold for this analysis was set at SThr > 40 and pThr < 0.2. At this stringent cutoff threshold we identified 310 probesets (230 genes) that has Boolean relation with VEGFA. We then use multilayer filters: 1) inclusion of strong relationship with VEGFA, 2) a similar relationship with the other three probesets of VEGFA, 3) conserve relationship in mice and 4) the candidate gene is not reported previously as a hypoxia related genes. The selected gene/s were proceeded for in-vitro validation.

Cell culture

Primary Pulmonary Artery Endothelial Cells (HPAEC) was purchased from ATCC (PCS-100-022™). For maintaining normal growth, we followed protocol as indicated by ATCC. The Endothelial Cell Growth Kit-BBE (ATCC® PCS-100-040) was added as indicated. We passaged the cells when cultures reached approximately 80% confluence. For the hypoxia experiments, equal numbers of cells were plated in five 60mm cell culture dishes and maintained in regular incubator of room air, 5% CO2 and 37⁰C. On day three of cell expansion i.e., ~70% confluence, four plates were transferred to an incubator with 1% O2, 5% CO2 and 37⁰C. Cells from each dish were used to isolate RNA at 0, 2, 8, 24 and 48 hours of hypoxia exposure. In order to get a robust reproducible readout the technical replicates used for RT-PCR are from three different passages.

Real-Time qRT-PCR

RNA was isolated from tissue samples using RNeasy Mini Kit (Qiagen, US). We used SYBR® Green Master Mix for RT-PCR which is a pre-formulated, optimized, universal 2X master mix for real-time PCR workflows. RT-PCR was performed on CFX96 Real-Time PCR System. The specific primers used for RT-PCR are listed in S3 Table. The relative transcript levels in hypoxia is compared to its levels in normoxia after normalizing to a housekeeping gene i.e., GAPDH (S3 Table).

Network estimation analysis

To predict the molecular function of FAM114A1 we used geneMANIA ( on Cytospace program (v3.4.0). This helps us identify various types of interactions with the other genes in the network as it uses a very large set of functional association data [9] to find out other genes that are related to FAM114A1. We used the default values for this query i.e., maximum resultant genes = 20 and maximum resultant attributes = 10. The information on physical and genetic interactions, pathways, co-expression, co-localization and protein domain similarity were obtained as a readout.

Statistical analysis

The data are shown as means ± standard errors (SEM). Paired or unpaired Student’s t-test and one-way analysis of variance (ANOVA) with Bonferroni multiple comparison test were used for statistical analysis. A p value <0.05 was considered as statistically significant.

Supporting information

S1 Fig. Expression profile of HIF1A in HPAEC under normoxia (21% O2) i.e., at 0 hour and at 2, 8, 24 and 48 hours of constant 1% O2.


S2 Fig. Expression profile of VEGFA and FAM114A1 in the data extracted from Costello et al., 2008 (GEO accession: GSE12792, PMID: 18469115).

Expression profile of VEGFA and FAM114A1 in the pulmonary microvascular endothelial cells (a, b) and cardiac microvascular endothelium (c, d) under normoxia (21% O2) and at 3, 24 and 48 hours of constant 1% O2 as reported in Costello et al., Am J Physiol Lung Cell Mol Physiol 2008. (*, P<0.05 when compared to the normoxia).


S3 Fig. Hypoxia-induced changes in the expression profile of VEGFA and FAM114A1 in different cell lineage from previously reported data.

(a, b) VEGF-treated endothelial cells treated for 0, 2 and 4 hours (GEO accession: GSE18913; PMID: 19965691). (c, d) Human renal proximal tubule epithelial cells (RPTEC, obtained from Lonza) exposed to 1% oxygen for 24 h. (GEO accession: GSE12792; PMID: 18984585). (e, f) P493-6 cells (Human Lymphoblastoid Cell Line) incubated in normoxic (20% O2) or hypoxic condition (0.1% O2) for 29 hr. (GEO accession: GSE4086; PMID: 16517405). *, P<0.05 when compared to the normoxia.


S4 Fig. Non-hypoxia-related changes in the expression profile of VEGFA and FAM114A1.

(a, b) Similar changes in VEGFA and FAM114A1 in the undifferentiated neuroblastoma cells (SH-SY5Y), i.e. before PMA induced differentiation (UD), 48 hours differentiated and untransfected cells (D-UT), 48 hours differentiated and MeCP2 decoy transfected cells (D-MD) and 48 hours differentiated and Control decoy transfected cells (D-CD) (GEO accession: GSE4600; PMID: 16682435). (c, d) Comparison of human HT29 colon cancer cells that are sensitive and resistant to methotrexate (GEO accession: GSE11440; PMID: 18694510). (e, f) CXCR4-positive subpopulation (expressing CXCR4), CXCR4-positive subpopulation treated with SDF-1 (CXCR4 treated with SDF-1alpha, ligand for CXCR4, also called CXCL12) for one hour and CXCR4-negative subpopulation (not expressing CXCR4) (GEO accession: GSE15893; PMID: 20603605). *, P<0.05.


S5 Fig. Boolean relationships between FAM114A1 and candidate genes from Renin Angiotensin system (RAS) taken from the Affymetrix Human U133 Plus 2.0 dataset.

At SThr = 10 and pThr = 0.1 no relation between FAM114A1 and ACE (a), FAM114A1 low = > ACE2 low (b), FAM114A1 low = > AGTR1 low (c), no relation for FAM114A1 vs AGTR2 (d), no relation for FAM114A1 vs AGT (e) and FAM114A1 low = > REN low (f).


S6 Fig. Basal expression of VEGFA, ANGPTL4, PPARG, PTGIS and INHBA in different tissues as indicated in GTEx database.

Y-axis indicates the median TPM (Transcript Per Million).


S7 Fig. Venn diagram indicating the common shared genes (among the top 100 Boolean related genes) when ANGPTL4, PPARG, PTGIS and INHBA, respectively, are used seed genes one at a time.

FAM114A1 (red circle) is the common for all seed genes.


S1 Table. The probesets (genes) detected by MiDReG that pairs with VEGFA at a BooleanNet statistical threshold of SThr > 40 and pThr < 0.2.


S2 Table. Top probesets (genes) that have a strong relationship with VEGFA.


S3 Table. Fold change in the expression levels of candidate genes at 2, 8, 24 and 48 hours of 1% O2 treatment compared to its baseline expression levels at 21% O2.



  1. 1. Heiss WD. The ischemic penumbra: how does tissue injury evolve? Ann N Y Acad Sci. 2012;1268(26–34). pmid:22994218
  2. 2. Abe H, Semba H, and Takeda N. The Roles of Hypoxia Signaling in the Pathogenesis of Cardiovascular Diseases. J Atheroscler Thromb. 2017;24(9):884–94. pmid:28757538
  3. 3. Eckardt KU, Bernhardt WM, Weidemann A, Warnecke C, Rosenberger C, Wiesener MS, et al. Role of hypoxia in the pathogenesis of renal disease. Kidney Int Suppl. 200599):S46–51.
  4. 4. Puente BN, Kimura W, Muralidhar SA, Moon J, Amatruda JF, Phelps KL, et al. The oxygen-rich postnatal environment induces cardiomyocyte cell-cycle arrest through DNA damage response. Cell. 2014;157(3):565–79. pmid:24766806
  5. 5. Galderisi U, Jori FP, and Giordano A. Cell cycle regulation and neural differentiation. Oncogene. 2003;22(33):5208–19. pmid:12910258
  6. 6. Iyer NV, Kotch LE, Agani F, Leung SW, Laughner E, Wenger RH, et al. Cellular and developmental control of O2 homeostasis by hypoxia-inducible factor 1 alpha. Genes Dev. 1998;12(2):149–62. pmid:9436976
  7. 7. Baqlouq L, Zihlif M, Hammad H, and Thaib TMA. Determining the Relative Gene Expression Level of Hypoxia Related Genes in Different Cancer Cell Lines. Curr Mol Pharmacol. 2021;14(1):52–9. pmid:32436837
  8. 8. Wilson WR, and Hay MP. Targeting hypoxia in cancer therapy. Nat Rev Cancer. 2011;11(6):393–410. pmid:21606941
  9. 9. Warde-Farley D, Donaldson SL, Comes O, Zuberi K, Badrawi R, Chao P, et al. The GeneMANIA prediction server: biological network integration for gene prioritization and predicting gene function. Nucleic Acids Res. 2010;38(Web Server issue):W214–20. pmid:20576703
  10. 10. Szklarczyk D, Gable AL, Lyon D, Junge A, Wyder S, Huerta-Cepas J, et al. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res. 2019;47(D1):D607–D13. pmid:30476243
  11. 11. Sahoo D, Dill DL, Gentles AJ, Tibshirani R, and Plevritis SK. Boolean implication networks derived from large scale, whole genome microarray datasets. Genome Biol. 2008;9(10):R157. pmid:18973690
  12. 12. Sahoo D, Seita J, Bhattacharya D, Inlay MA, Weissman IL, Plevritis SK, et al. MiDReG: a method of mining developmentally regulated genes using Boolean implications. Proc Natl Acad Sci U S A. 2010;107(13):5732–7. pmid:20231483
  13. 13. Boucquey M, De Plaen E, Locker M, Poliard A, Mouillet-Richard S, Boon T, et al. Noxp20 and Noxp70, two new markers of early neuronal differentiation, detected in teratocarcinoma-derived neuroectodermic precursor cells. J Neurochem. 2006;99(2):657–69. pmid:17029606
  14. 14. Semenza GL. HIF-1 and mechanisms of hypoxia sensing. Curr Opin Cell Biol. 2001;13(2):167–71. pmid:11248550
  15. 15. Yao P, Wu J, Lindner D, and Fox PL. Interplay between miR-574-3p and hnRNP L regulates VEGFA mRNA translation and tumorigenesis. Nucleic Acids Res. 2017;45(13):7950–64. pmid:28520992
  16. 16. Shweiki D, Itin A, Soffer D, and Keshet E. Vascular endothelial growth factor induced by hypoxia may mediate hypoxia-initiated angiogenesis. Nature. 1992;359(6398):843–5. pmid:1279431
  17. 17. Costello CM, Howell K, Cahill E, McBryan J, Konigshoff M, Eickelberg O, et al. Lung-selective gene responses to alveolar hypoxia: potential role for the bone morphogenetic antagonist gremlin in pulmonary hypertension. Am J Physiol Lung Cell Mol Physiol. 2008;295(2):L272–84. pmid:18469115
  18. 18. Kim JW, Tchernyshyov I, Semenza GL, and Dang CV. HIF-1-mediated expression of pyruvate dehydrogenase kinase: a metabolic switch required for cellular adaptation to hypoxia. Cell Metab. 2006;3(3):177–85. pmid:16517405
  19. 19. Irigoyen M, Anso E, Martinez E, Garayoa M, Martinez-Irujo JJ, and Rouzaut A. Hypoxia alters the adhesive properties of lymphatic endothelial cells. A transcriptional and functional study. Biochim Biophys Acta. 2007;1773(6):880–90. pmid:17442415
  20. 20. Suehiro J, Hamakubo T, Kodama T, Aird WC, and Minami T. Vascular endothelial growth factor activation of endothelial cells is mediated by early growth response-3. Blood. 2010;115(12):2520–32. pmid:19965691
  21. 21. Beyer S, Kristensen MM, Jensen KS, Johansen JV, and Staller P. The histone demethylases JMJD1A and JMJD2B are transcriptional targets of hypoxia-inducible factor HIF. J Biol Chem. 2008;283(52):36542–52. pmid:18984585
  22. 22. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, and Speed TP. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids Res. 2003;31(4):e15. pmid:12582260
  23. 23. Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003;4(2):249–64. pmid:12925520
  24. 24. O’Rourke JF, Dachs GU, Gleadle JM, Maxwell PH, Pugh CW, Stratford IJ, et al. Hypoxia response elements. Oncol Res. 1997;9(6–7):327–32. pmid:9406238
  25. 25. Ratcliffe PJ, O’Rourke JF, Maxwell PH, and Pugh CW. Oxygen sensing, hypoxia-inducible factor-1 and the regulation of mammalian gene expression. J Exp Biol. 1998;201(Pt 8):1153–62. pmid:9510527
  26. 26. Zeng FC, Zeng MQ, Huang L, Li YL, Gao BM, Chen JJ, et al. Downregulation of VEGFA inhibits proliferation, promotes apoptosis, and suppresses migration and invasion of renal clear cell carcinoma. Onco Targets Ther. 2016;9(2131–41.
  27. 27. Peddada S, Yasui DH, and LaSalle JM. Inhibitors of differentiation (ID1, ID2, ID3 and ID4) genes are neuronal targets of MeCP2 that are elevated in Rett syndrome. Hum Mol Genet. 2006;15(12):2003–14. pmid:16682435
  28. 28. Selga E, Morales C, Noe V, Peinado MA, and Ciudad CJ. Role of caveolin 1, E-cadherin, Enolase 2 and PKCalpha on resistance to methotrexate in human HT29 colon cancer cells. BMC Med Genomics. 2008;1(35. pmid:18694510
  29. 29. Subbaiah KCV, Wu J, Tang WHW, and Yao P. FAM114A1 Influences Cardiac Fibrosis by Regulating Angiotensin II Signaling in Cardiac Fibroblasts. 2021.
  30. 30. ErZen B, Silar M, and Sabovic M. Stable phase post-MI patients have elevated VEGF levels correlated with inflammation markers, but not with atherosclerotic burden. BMC Cardiovasc Disord. 2014;14(166.
  31. 31. Appaiah H, Bhat-Nakshatri P, Mehta R, Thorat M, Badve S, and Nakshatri H. ITF2 is a target of CXCR4 in MDA-MB-231 breast cancer cells and is associated with reduced survival in estrogen receptor-negative breast cancer. Cancer Biol Ther. 2010;10(6):600–14. pmid:20603605
  32. 32. Zhou M, Lin F, Wu X, Ping Z, Xu W, Jin R, et al. Inhibition of Fam114A1 protects melanocytes from apoptosis through higher RACK1 expression. Aging (Albany NY). 2021;13(22):24740–52. pmid:34837888
  33. 33. Robinson PC, Leo PJ, Pointon JJ, Harris J, Cremin K, Bradbury LA, et al. Exome-wide study of ankylosing spondylitis demonstrates additional shared genetic background with inflammatory bowel disease. NPJ Genom Med. 2016;1(16008. pmid:29263810
  34. 34. Donoghue M, Hsieh F, Baronas E, Godbout K, Gosselin M, Stagliano N, et al. A novel angiotensin-converting enzyme-related carboxypeptidase (ACE2) converts angiotensin I to angiotensin 1–9. Circ Res. 2000;87(5):E1–9. pmid:10969042
  35. 35. Burrell LM, Risvanis J, Kubota E, Dean RG, MacDonald PS, Lu S, et al. Myocardial infarction increases ACE2 expression in rat and humans. Eur Heart J. 2005;26(4):369–75; discussion 22–4. pmid:15671045
  36. 36. Joshi S, Wollenzien H, Leclerc E, and Jarajapu YP. Hypoxic regulation of angiotensin-converting enzyme 2 and Mas receptor in human CD34(+) cells. J Cell Physiol. 2019;234(11):20420–31. pmid:30989646
  37. 37. Delforce SJ, Wang Y, Van-Aalst ME, Corbisier de Meaultsart C, Morris BJ, Broughton-Pipkin F, et al. Effect of oxygen on the expression of renin-angiotensin system components in a human trophoblast cell line. Placenta. 2016;37(1–6.
  38. 38. Zisman LS, Keller RS, Weaver B, Lin Q, Speth R, Bristow MR, et al. Increased angiotensin-(1–7)-forming activity in failing human heart ventricles: evidence for upregulation of the angiotensin-converting enzyme Homologue ACE2. Circulation. 2003;108(14):1707–12. pmid:14504186
  39. 39. Duff MO, Olson S, Wei X, Garrett SC, Osman A, Bolisetty M, et al. Genome-wide identification of zero nucleotide recursive splicing in Drosophila. Nature. 2015;521(7552):376–9. pmid:25970244
  40. 40. Fagerberg L, Hallstrom BM, Oksvold P, Kampf C, Djureinovic D, Odeberg J, et al. Analysis of the human tissue-specific expression by genome-wide integration of transcriptomics and antibody-based proteomics. Mol Cell Proteomics. 2014;13(2):397–406. pmid:24309898
  41. 41. Shin MJ, Kim DW, Jo HS, Cho SB, Park JH, Lee CH, et al. Tat-PRAS40 prevent hippocampal HT-22 cell death and oxidative stress induced animal brain ischemic insults. Free Radic Biol Med. 2016;97(250–62. pmid:27317854
  42. 42. Zhang KS, Schecker J, Krull A, Riechert E, Jurgensen L, Kamuf-Schenk V, et al. PRAS40 suppresses atherogenesis through inhibition of mTORC1-dependent pro-inflammatory signaling in endothelial cells. Sci Rep. 2019;9(1):16787. pmid:31728028
  43. 43. Gitlin AD, Heger K, Schubert AF, Reja R, Yan D, Pham VC, et al. Integration of innate immune signalling by caspase-8 cleavage of N4BP1. Nature. 2020;587(7833):275–80. pmid:32971525
  44. 44. Shi H, Sun L, Wang Y, Liu A, Zhan X, Li X, et al. N4BP1 negatively regulates NF-kappaB by binding and inhibiting NEMO oligomerization. Nat Commun. 2021;12(1):1379.
  45. 45. Zhu G, Qi Q, Havel JJ, Li Z, Du Y, Zhang X, et al. PRAS40 promotes NF-kappaB transcriptional activity through association with p65. Oncogenesis. 2017;6(9):e381.
  46. 46. Cummins EP, Berra E, Comerford KM, Ginouves A, Fitzgerald KT, Seeballuck F, et al. Prolyl hydroxylase-1 negatively regulates IkappaB kinase-beta, giving insight into hypoxia-induced NFkappaB activity. Proc Natl Acad Sci U S A. 2006;103(48):18154–9. pmid:17114296
  47. 47. Fu J, and Taubman MB. EGLN3 inhibition of NF-kappaB is mediated by prolyl hydroxylase-independent inhibition of IkappaB kinase gamma ubiquitination. Mol Cell Biol. 2013;33(15):3050–61.
  48. 48. Chandel NS, Trzyna WC, McClintock DS, and Schumacker PT. Role of oxidants in NF-kappa B activation and TNF-alpha gene transcription induced by hypoxia and endotoxin. J Immunol. 2000;165(2):1013–21. pmid:10878378
  49. 49. Koong AC, Chen EY, and Giaccia AJ. Hypoxia causes the activation of nuclear factor kappa B through the phosphorylation of I kappa B alpha on tyrosine residues. Cancer Res. 1994;54(6):1425–30. pmid:8137243
  50. 50. Santoni D, Castiglione F, and Paci P. Identifying correlations between chromosomal proximity of genes and distance of their products in protein-protein interaction networks of yeast. PLoS One. 2013;8(3):e57707. pmid:23483922
  51. 51. Manalo DJ, Rowan A, Lavoie T, Natarajan L, Kelly BD, Ye SQ, et al. Transcriptional regulation of vascular endothelial cell responses to hypoxia by HIF-1. Blood. 2005;105(2):659–69. pmid:15374877