Impact of Polyunsaturated Fatty Acids on miRNA Profiles of Monocytes/Macrophages and Endothelial Cells—A Pilot Study

Alteration of miRNAs and dietary polyunsaturated fatty acids (PUFAs) underlies vascular inflammation. PUFAs are known to be incorporated into the cell membrane of monocytes/macrophages or endothelial cells, the major cellular players of vascular diseases, thereby affecting cellular signal transduction. Nevertheless, there are no investigations concerning the PUFA impact on miRNA expression by these cells. With regard to the key role miRNAs play for overall cellular functionality, this study aims to elucidate whether PUFAs affect miRNA expression profiles. To this end, the monocyte/macrophage cell line RAW264.7 and the endothelial cell line TIME were enriched with either docosahexaenoic acid (DHA; n3-PUFA) or arachidonic acid (AA; n6-PUFA) until reaching a stable incorporation into the plasma membrane and, at least in part, exposed to an inflammatory milieu. Expressed miRNAs were determined by deep sequencing, and compared to unsupplemented/unstimulated controls. Data gained clearly show that PUFAs in fact modulate miRNA expression of both cell types analyzed regardless the presence/absence of an inflammatory stimulator. Moreover, certain miRNAs already linked to vascular inflammation were found to be affected by cellular PUFA enrichment. Hence, vascular inflammation appears to be influenced by dietary fatty acids, inter alia, via PUFA-mediated modulation of the type and amount of miRNAs synthesized by cells involved in the inflammatory process.


Introduction
Fine-tuning of protein expression is of great significance for overall cellular functionality. Gene transcription and translation not only need to be adjusted to meet the demands of the single eukaryotic cell but also to ensure an optimum interaction in higher multi-cellular organisms. Dysregulation of these biological processes are frequently associated with disease development and progression.
Non-coding RNAs, in particular microRNAs (miRNAs), play an important role in gene regulation. Preponderantly, miRNAs are able to inhibit the expression of a certain protein by means of translation repression or even mRNA degradation, but some miRNAs even can stimulate translation [1][2][3]. Each of these short, endogenous, regulatory RNAs can target a wide variety of gene transcripts due to imperfect base pairing, thereby mediating post-transcriptional gene silencing [1][2][3][4]. What is more, a particular mRNA can be co-regulated by several miRNAs resulting in a complex, interconnected, and coordinated regulatory network [1,2,4]. Consequently, aberrant processes and a large number of disease states including inflammatory disorders and cardiovascular conditions have been implicated

Results
In this study, miRNA expression profiles of the monocyte/macrophage cell line RAW264.7 as well as the endothelial cell line TIME were analyzed. Up-or downregulated miRNAs were identified due to (i) an inflammatory milieu; (ii) PUFA supplementation; or (iii) a combination of inflammatory milieu and PUFA supplementation. Expression levels of a miRNA were considered when there was at least a two-fold disparity between the comparison groups (n = 3 for each subgroup).
Supplementation of RAW264.7 with the n6-PUFA arachidonic acid (AA) resulted in the alternative expression of 29 miRNAs. Compared to unsupplemented control cells, seven miRNAs were upregulated, and eight miRNAs were downregulated in RAW264.7 cultured in AA enriched medium (Table 3). Seven miRNAs could only be identified in AA treated RAW264.7, and a further seven miRNAs could not be detected in the AA in contrast to the control group (Table 3).  It has to be noted that there are distinct differences in the miRNAs influenced by the PUFAs tested. As shown in Figure 1a, a certain proportion of miRNAs modulated by DHA was not affected by AA and vice versa. Only two miRNAs were increased by both DHA and AA. From the miRNAs characterized by a decreased expression profile, nine were concordantly modulated by both DHA and AA.
It has to be noted that there are distinct differences in the miRNAs influenced by the PUFAs tested. As shown in Figure 1a, a certain proportion of miRNAs modulated by DHA was not affected by AA and vice versa. Only two miRNAs were increased by both DHA and AA. From the miRNAs characterized by a decreased expression profile, nine were concordantly modulated by both DHA and AA.

Impact of a Combination of LPS Stimulation and PUFA Supplementation
LPS stimulation of RAW264.7 previously enriched with the n3-PUFA DHA resulted in the alternative expression of 50 miRNAs. Compared to control cells, 5 miRNAs were upregulated, and 32 miRNAs were downregulated in DHA-supplemented RAW264.7 triggered with LPS (Table 4). Of these, five miRNAs showed an expression reduction of at least five-fold. A total of 3 miRNAs were detected in RAW264.7 treated with DHA plus LPS but not in the untreated control group, whereas 10 miRNAs could be identified in control cells only (Table 4). Combined treatment of RAW264.7 with LPS plus the n6-PUFA AA resulted in the alternative expression of 86 miRNAs. Compared to the control group, 8 miRNAs were upregulated, and 59 miRNAs were downregulated in RAW264.7 enriched in AA and stimulated with LPS (Table 5). For 13 of these miRNAs, downregulation attained a fold-change of at least 0.2. In particular, miR-143-3p was characterized by an expression reduction by a factor of 500. There was one miRNA that could only be identified in LPS-triggered RAW264.7 cultivated in AA supplemented medium (Table 5). In addition, a total of 18 miRNAs could solely be detected in untreated control cells but not in RAW264.7 after combined treatment using LPS and AA (Table 5). Again, there are distinct differences in the miRNAs influenced by the PUFA tested ( Figure 2). Three miRNAs were found to be increased by both DHA and AA. Regarding downregulated miRNAs, a total of 28 were concordantly modulated by both DHA and AA. Of note, there is almost no overlap between the miRNAs alternatively expressed in RAW264.7 treated with a combination of PUFA supplementation and LPS stimulation and the miRNAs up-or downregulated in RAW264.7 exclusively treated with either PUFA or LPS.
Three miRNAs were found to be increased by both DHA and AA. Regarding downregulated miRNAs, a total of 28 were concordantly modulated by both DHA and AA. Of note, there is almost no overlap between the miRNAs alternatively expressed in RAW264.7 treated with a combination of PUFA supplementation and LPS stimulation and the miRNAs up-or downregulated in RAW264.7 exclusively treated with either PUFA or LPS.

Impact of Cytokine Stimulation
Following exposure to the pro-inflammatory cytokines IL-1β, TNF-α, and IFN-γ 173 miRNAs were identified to be alternatively expressed in stimulated TIME, compared to control cells. These included 151 upregulated miRNAs and 4 downregulated miRNAs (Table 6). Of these, 20 miRNAs showed differences greater than five-fold. Furthermore, 18 miRNAs could only be identified in

Impact of Cytokine Stimulation
Following exposure to the pro-inflammatory cytokines IL-1β, TNF-α, and IFN-γ 173 miRNAs were identified to be alternatively expressed in stimulated TIME, compared to control cells. These included 151 upregulated miRNAs and 4 downregulated miRNAs (Table 6). Of these, 20 miRNAs showed differences greater than five-fold. Furthermore, 18 miRNAs could only be identified in cytokine treated TIME but not in the unstimulated control group (Table 6). There was no disappearance of miRNAs due to cytokine stimulation. Table 6. Significantly differentially expressed miRNAs (sequence read >10; regulated greater than 2-fold) in TIME due to cytokine stimulation (24 h, IL-1β + TNF-α + IFN-γ 5 ng/mL each).   Enrichment of TIME with the n3-PUFA DHA resulted in the alternative expression of 21 miRNAs. Compared to unsupplemented control cells, 16 miRNAs were upregulated, and one miRNA was downregulated in TIME cultured in DHA-enriched medium (Table 7). Three miRNAs could be identified in DHA-treated TIME only (Table 7). Furthermore, one miRNA could be detected in control cells but not in cells enriched in DHA (Table 7). Table 7. Significantly differentially expressed miRNAs (sequence read >10; regulated greater than 2-fold) in TIME due to DHA supplementation (72 h, 15 µM).

Fold-Change miRNAs
solely detected in experimental miR-9-5p, miR-3177-3p, miR-3158-3p Supplementation of TIME with the n6-PUFA AA resulted in the alternative expression of 140 miRNAs. Compared to unsupplemented control cells, 131 miRNAs were upregulated in TIME cultured in AA enriched medium with three miRNAs being increased five-fold and beyond (Table 8). Moreover, nine miRNAs were solely detectable in AA treated TIME (Table 8). AA supplementation did not result in a downregulation or disappearance of any miRNAs.
The number of miRNAs concurrently influenced by both DHA and AA is shown in Figure 3. Despite the great differences in the absolute numbers of affected miRNAs, there is a considerable degree of convergence in the nature of miRNAs biased by PUFA enrichment in general. From the miRNAs characterized by an increased expression profile, a total of 14 were concordantly modulated by both DHA and AA.  2.0 miR-23a-3p, miR-28-3p, miR-148b-3p, miR-181a-2-3p, miR-199a-3p, miR-320a, miR-410-3p, miR-629-5p, miR-769-5p, miR-889-3p, miR-2110, miR-2682-5p solely detected in experimental miR-30b-3p, miR-145-3p, miR-183-5p, miR-216a-5p, miR-539-5p, miR-548o-3p, miR-3130-3p, miR-3158-3p, miR-4662a-5p The number of miRNAs concurrently influenced by both DHA and AA is shown in Figure 3. Despite the great differences in the absolute numbers of affected miRNAs, there is a considerable degree of convergence in the nature of miRNAs biased by PUFA enrichment in general. From the miRNAs characterized by an increased expression profile, a total of 14 were concordantly modulated by both DHA and AA.    Combined treatment of TIME with cytokines plus the n3-PUFA DHA resulted in the alternative expression of 46 miRNAs. Compared to control cells, four miRNAs were upregulated, and 31 miRNAs were downregulated in DHA-supplemented TIME triggered with the cytokines IL-1β, TNF-α, and IFN-γ (Table 9). Of these, the miRNA miR-1-3p showed an expression increase of about 118-fold.
Beyond that, two miRNAs were detected in TIME treated with DHA plus cytokines but not in the untreated control group (Table 9). Nine miRNAs could solely be identified in control cells (Table 9). Table 9. Significantly differentially expressed miRNAs (sequence read >10; regulated greater than 2-fold) in TIME due to combined DHA supplementation (72 h, 15 µM) and cytokine stimulation (24 h, IL-1β + TNF-α + IFN-γ 5 ng/mL each). Cytokine stimulation of TIME previously enriched with the n6-PUFA AA resulted in the alternative expression of 16 miRNAs. Compared to the control group, three miRNAs were upregulated, and eight miRNAs were downregulated in TIME enriched in AA and stimulated with IL-1β and TNF-α plus IFN-γ (Table 10). There was one miRNA, which could only be identified in cytokine-triggered TIME cultivated in AA supplemented medium (Table 10). In addition, a total of four miRNAs could solely be detected in untreated control cells but not in TIME after combined treatment using cytokines and AA (Table 10). There are distinct differences in the miRNAs influenced by either DHA or AA. The majority of miRNAs modulated by DHA was not affected by AA and vice versa (Figure 4). Only one miRNA was increased by both DHA and AA. From the miRNAs characterized by a decreased expression profile, six were concordantly modulated by both DHA and AA. As seen for the monocyte/macrophage cell line RAW264.7, there is no or almost no overlap between the miRNAs alternatively expressed in endothelial cells treated with a combination of PUFA supplementation and cytokine stimulation and the miRNAs up-or downregulated in endothelial cells exclusively treated with either PUFA or cytokines. was increased by both DHA and AA. From the miRNAs characterized by a decreased expression profile, six were concordantly modulated by both DHA and AA. As seen for the monocyte/ macrophage cell line RAW264.7, there is no or almost no overlap between the miRNAs alternatively expressed in endothelial cells treated with a combination of PUFA supplementation and cytokine stimulation and the miRNAs up-or downregulated in endothelial cells exclusively treated with either PUFA or cytokines.

In Silico Analysis
Identified PUFA-regulated miRNAs were subjected to an in silico analysis by means of the miRWalk 2.0 database in order to determine previously validated corresponding target genes. The Database for Annotation, Visualization, and Integrated Discovery (DAVID) functional annotation clustering tool was used to perform a KEGG pathway and GO term enrichment analysis on tagged miRNA target genes. Next, Kyoto Encyclopedia of Genes and Genoms-(KEGG) and Gene Ontology (GO)-related clusters were related to basic biological processes, namely metabolism, signal transduction, growth/differentiation, apoptosis/necrosis, gene expression, cytoskeleton, and barrier function, transport, as well as immune defense. It was found that major targets of miRNAs alternatively expressed in PUFA-enriched cells are linked to gene expression and signal transduction without tangible differences between the cell types or inflammatory conditions analyzed ( Figure 5). This was followed by immune defense-, growth/differentiation-, and transport-associated genes ( Figure 5). Key miRNAs that are particularly altered by PUFAs and validated to play a role in inflammation are displayed in Table 11 for the monocyte/macrophage cell line RAW264.7 and in Table 12 for the endothelial cell line TIME. For both cell lines investigated, there are a wide variety of target genes and thus immunological processes under the regulatory control of PUFA-affected

In Silico Analysis
Identified PUFA-regulated miRNAs were subjected to an in silico analysis by means of the miRWalk 2.0 database in order to determine previously validated corresponding target genes. The Database for Annotation, Visualization, and Integrated Discovery (DAVID) functional annotation clustering tool was used to perform a KEGG pathway and GO term enrichment analysis on tagged miRNA target genes. Next, Kyoto Encyclopedia of Genes and Genoms-(KEGG) and Gene Ontology (GO)-related clusters were related to basic biological processes, namely metabolism, signal transduction, growth/differentiation, apoptosis/necrosis, gene expression, cytoskeleton, and barrier function, transport, as well as immune defense. It was found that major targets of miRNAs alternatively expressed in PUFA-enriched cells are linked to gene expression and signal transduction without tangible differences between the cell types or inflammatory conditions analyzed ( Figure 5). This was followed by immune defense-, growth/differentiation-, and transport-associated genes ( Figure 5). was increased by both DHA and AA. From the miRNAs characterized by a decreased expression profile, six were concordantly modulated by both DHA and AA. As seen for the monocyte/ macrophage cell line RAW264.7, there is no or almost no overlap between the miRNAs alternatively expressed in endothelial cells treated with a combination of PUFA supplementation and cytokine stimulation and the miRNAs up-or downregulated in endothelial cells exclusively treated with either PUFA or cytokines.

In Silico Analysis
Identified PUFA-regulated miRNAs were subjected to an in silico analysis by means of the miRWalk 2.0 database in order to determine previously validated corresponding target genes. The Database for Annotation, Visualization, and Integrated Discovery (DAVID) functional annotation clustering tool was used to perform a KEGG pathway and GO term enrichment analysis on tagged miRNA target genes. Next, Kyoto Encyclopedia of Genes and Genoms-(KEGG) and Gene Ontology (GO)-related clusters were related to basic biological processes, namely metabolism, signal transduction, growth/differentiation, apoptosis/necrosis, gene expression, cytoskeleton, and barrier function, transport, as well as immune defense. It was found that major targets of miRNAs alternatively expressed in PUFA-enriched cells are linked to gene expression and signal transduction without tangible differences between the cell types or inflammatory conditions analyzed ( Figure 5). This was followed by immune defense-, growth/differentiation-, and transport-associated genes ( Figure 5). Key miRNAs that are particularly altered by PUFAs and validated to play a role in inflammation are displayed in Table 11 for the monocyte/macrophage cell line RAW264.7 and in Table 12 for the endothelial cell line TIME. For both cell lines investigated, there are a wide variety of target genes and thus immunological processes under the regulatory control of PUFA-affected Key miRNAs that are particularly altered by PUFAs and validated to play a role in inflammation are displayed in Table 11 for the monocyte/macrophage cell line RAW264.7 and in Table 12 for the endothelial cell line TIME. For both cell lines investigated, there are a wide variety of target genes and thus immunological processes under the regulatory control of PUFA-affected miRNAs. It should be noted that several miRNAs biased by PUFA supplementation have concurring target genes. An outstanding number of immunological relevant target genes were found for hsa-miR-335-5p, which in TIME cells is modulated by supplementation of both DHA and AA. Inhibitor of kappaB kinase epsilon (Ikbke) 56489       Immunoglobulin superfamily member 6 (IGSF6) 10261 Immunoglobulin-like transcript 1 (LILRA2) 11027 Lymphocyte transmembrane adaptor 1 (LAX1) 54900 Tumor necrosis factor receptor 13B (TNFRSF13B) 23495

Discussion
Here we present the first analysis of the PUFA impact on monocytes/macrophage and endothelial cell miRNA expression profiles in either a normal or an inflammatory milieu. For both cell types, our data provide evidence that PUFA enrichment affects the kind and the amount of particular miRNAs synthesized. It should be noted that, in our study, cellular miRNA profiles were measured as a function of long-term PUFA supplementation, which leads to an incorporation of the fatty acids into the plasma membrane. The periods of supplementation used in our experiments are already proved by us to result in a membrane fatty acid steady state [23,24]. From this perspective, our data hint toward an innovative mechanism of PUFA action. It seems plausible that there is a causal link between the plasma membrane lipid composition and the miRNA expression of a cell.
So far, studies concerning the interrelation between PUFA and miRNA expression are limited, concentrating primarily on tumor cell biology [25,26]. Information on monocytes/macrophages or endothelial cells is missing although these cells are in constant blood contact, making them particularly susceptible to nutrition-based influences. Both cell types are directly involved in inflammatory processes of the vascular system [5][6][7] and in turn have been said to be influenced by unsaturated fatty acids [8,9]. Inflammatory states of the vasculature have been correlated to both miRNA expression [1,3] and fatty acid supply; nevertheless, miRNA expression and fatty acid supplyhave not yet been related.
With the recognition that there is a direct connection between the dietary PUFA content and the plasma membrane lipid composition of monocytes/macrophages and endothelial cells on the one hand, and the miRNA expression of these cells on the other hand, our study points the way to a new understanding of the PUFA-mediated modulation of vascular inflammatory processes.
With a view to the literature, certain miRNAs have already been linked to vascular inflammation, whether they are in acute inflammatory states as in the case of sepsis or chronic inflammatory conditions as in the case of atherosclerosis. Several of these miRNAs also pop up in our study in case of LPS-stimulated RAW264.7 or cytokine-stimulated TIME, which supports the validity of the data set. Beyond that, PUFA supplementation itself demonstrably impacts these inflammation-associated miRNAs; however, it turns out that there are distinct differences in PUFA effects between the cell types tested. Looking into the details, an upregulation of miR-10a, miR-17-3p, miR-125a, miR-155, and miR-181b was observed due to AA supplementation of endothelial cells in the absence of an inflammatory stimulus. All of these miRNAs are described to possess anti-inflammatory actions, attenuating the expression of adhesion molecules and inflammation-driving cytokines [27,28]. DHA in turn was observed to downregulate miR-146a in the context of an inflammatory milieu. This miRNA is known to contribute to the induction of vascular inflammation [29]. It can therefore be stated that, by this means, both DHA and AA seem to support an anti-inflammatory endothelial state. With regard to the monocytes/macrophages tested, the PUFA-mediated impact on proven inflammation-associated miRNAs is even more pronounced. Under inflammatory conditions, both DHA and AA downregulate numerous miRNAs, which are well-known for their involvement in Toll-like receptor signaling and macrophage differentiation. Namely expression of miR-21, miR-125a, miR146a, miR-146b, and miR-155, which are all described as miRNA targets of TLR signaling [27,28,30], are decreased due to RAW264.7 enrichment with either DHA or AA. PUFAs, therefore, may be seen as modulators of macrophage phenotype and inflammatory response. This is in accordance with our previously published functional analyses demonstrating PUFAs such as DHA or AA to impact virtually all macrophage features including cytokine synthesis, respiratory burst, phagocytosis, and the expression of adhesion molecules [15,31,32].
Besides these literature-described miRNAs, DAVID functional annotation clustering indicates that there are numerous further PUFA-targeted miRNAs, which are linked to immune defense. About 10% of all miRNAs alternatively expressed in either RAW264.7 or TIME due to PUFA enrichment belong in this group, thus opening an exciting field in vascular inflammation research. For example, miR-1-3p, whose expression is increased by a factor of 118 in cytokine-stimulated, DHA-supplemented endothelial cells, may influence the interplay between cytokines and their corresponding receptors, and is predicted to interact with various members of the Jak-STAT signaling pathway.
Comparing the PUFAs analyzed, it is found that there are distinct differences in the miRNA targets of the fatty acids. It is interesting to note that, beside the deviations observed for particular miRNAs, there seems to be a common focus, as can be seen in the context of the literature-described miRNAs discussed above.

Materials
All chemicals and reagents were obtained from Sigma-Aldrich (Taufkirchen, Germany) unless noted otherwise. Cell culture flasks were purchased from Greiner Bio-One (Frickenhausen, Germany). HEPES (25 mmol/L)-buffered RPMI 1640 culture medium containing 300 mg/L L-glutamine was acquired from Pan-Biotech (Aidenbach, Germany). Microvascular endothelial growth medium as per customer specification according to ATCC recommendations was acquired from Provitro AG (Berlin, Germany).

RNA Isolation, Deep Sequencing, and Analysis of Deep Sequencing Data
Total RNA extraction was performed using a standard liquid-liquid extraction protocol based on TRIzol LS (Thermo Fisher Scientific, Dreieich, Germany) according to the manufacturer's instructions. Quality of RNA gained was analyzed by means of the NanoDrop spectrophotometer (Thermo Fisher Scientific, Dreieich, Germany) as well as the Agilent Bioanalyzer (Agilent Technologies, Waldbronn, Germany). Analysis of miRNA expression by high-throughput sequencing was performed in the Core Unit DNA, Leipzig University by means of a Illumina HiScanSQ (Illumina Inc., San Diego, CA, USA). Briefly, samples were prepared using the TruSeq Small RNA Prep kit v2 (Illumina Inc., San Diego, CA, USA) according to the manufacturer's instructions. The barcoded libraries were size restricted (140-165 bp), purified, and quantified using the Library quantification kit-Illumina/Universal (KAPA Biosystems, Woburn, MA, USA) according to the manufacturer's instructions. A pool of up to 10 libraries was used for cluster generation per lane using an Illumina cBot. Sequencing of 50 bp was performed with an Illumina HighScanSQ sequencer using version 3 chemistry and flowcell according to the manufacturer's instructions. For analysis of the deep sequencing data, adapter sequences were removed from raw sequences using Cutadapt software [33]. From the remaining sequences, only those 15-27 bases long were kept for further analysis. Alignment of these reads to murine/human genome as well as mature sequences of miRBase v21 was done using the bowtie aligner [34]. Data were compressed with Samtools [35] to bam format. Mapped reads were counted with R/Bioconductor programming environment [36] by application of the ShortRead library [37]. An error rate of 1 nt per mature miRNA sequence was allowed. Normalization of data was performed by the two independent procedures DESeq2 and the TMM algorithm.

Statistical Analysis
Two-way analysis of variance followed by unpaired Student's t-test was used to identify significant differences between means. The statistical analysis was carried out by means of the program GraphPad Prism 6 (GraphPad Software, La Jolla, CA, USA). In all cases, p < 0.05 was considered to indicate significant differences. Significantly differentially expressed miRNAs, which had a sequence read >10 and were regulated greater than 2-fold, were depicted in the data tables.

Conclusions
As this is a pilot study, details regarding expression levels of individual miRNAs should be taken with caution. Clearly, a validation of data, for example, by means of a TaqMan-based polymerase chain reaction, should be performed for miRNAs of interest. Nevertheless, with regard to the data presented here, the following conclusions can be drawn: (1) PUFAs impact the miRNA profiles of monocytes/macrophages and endothelial cells; (2) PUFA-mediated alteration of membrane micro-domains modulate cellular signal transduction, thereby influencing the transcription of non-coding RNAs such as miRNAs; (3) in doing so, PUFAs are able to affect vascular inflammation in a miRNA-based manner. Altogether, this study leads the way to a new understanding of PUFA-mediated cellular signal transduction in general and, in particular, of the dietary modulation of inflammatory processes taking place in the vascular system.