Selenoprotein T Protects Endothelial Cells against Lipopolysaccharide-Induced Activation and Apoptosis

Sepsis is an exaggerated immune response upon infection with lipopolysaccharide (LPS) as the main causative agent. LPS-induced activation and apoptosis of endothelial cells (EC) can lead to organ dysfunction and finally organ failure. We previously demonstrated that the first twenty amino acids of the Apurinic/Apyrimidinic Endodeoxyribonuclease 1 (APEX1) are sufficient to inhibit EC apoptosis. To identify genes whose regulation by LPS is affected by this N-terminal APEX1 peptide, EC were transduced with an expression vector for the APEX1 peptide or an empty control vector and treated with LPS. Following RNA deep sequencing, genes upregulated in LPS-treated EC expressing the APEX1 peptide were identified bioinformatically. Selected candidates were validated by semi-quantitative real time PCR, a promising one was Selenoprotein T (SELENOT). For functional analyses, an expression vector for SELENOT was generated. To study the effect of SELENOT expression on LPS-induced EC activation and apoptosis, the SELENOT vector was transfected in EC. Immunostaining showed that SELENOT was expressed and localized in the ER. EC transfected with the SELENOT plasmid showed no activation and reduced apoptosis induced by LPS. SELENOT as well as APEX1(1-20) can protect EC against activation and apoptosis and could provide new therapeutic approaches in the treatment of sepsis.


Introduction
Sepsis can best be described as an overwhelming inflammatory condition, in which the body responds to an infection in a hyperactive, dysregulated way, which in turn results in life-threatening organ dysfunction and eventually septic shock. According to an estimate of the World Health Organization (WHO), sepsis affects more than 48 million people every year, potentially leading to 11 million deaths [1]. The basis for the pathophysiological responses in the context of sepsis is multifactorial. Therefore, except for the introduction of vasopressor agents 40 years ago, no new therapeutic principle for the treatment of sepsis has been developed until today.
Lipopolysaccharide (LPS) is an outer membrane component of Gram-negative bacteria. Most bacterial LPS molecules are thermostable and generate a pro-inflammatory stimulus

RNA Sequencing and Bioinformatic Analysis
RNA sequencing data were obtained from quadruplicate total RNA samples. Total RNAs used for transcriptome analyses were quantified using the Qubit TM RNA HS Assay kit (Thermo Fisher Scientific, Dreieich, Germany) and quality was determined by capillary electrophoresis using the FragmentAnalyzer and the Total RNA Standard Sensitivity Assay (Agilent Technologies, Santa Clara, CA, USA). All samples in this study showed highest RNA Quality Numbers (RQN 10.0). Library construction and sequencing were performed at the Genomics and Transcriptomics Laboratory at the Biological Medical Research Centre (BMFZ) of the Heinrich-Heine University Düsseldorf. Library preparation was performed according to the manufacturer's protocol using the TruSeq Stranded mRNA Assay kit (Illumina, San Diego, CA, USA). Briefly, 500 ng total RNA was used for mRNA capturing, fragmentation, synthesis of cDNA, adapter ligation and library amplification. Bead purified libraries were normalized and finally sequenced on the HiSeq 3000 system (Illumina San Diego, CA, USA) with a read setup of 1 × 150 bp. The bcl2fastq2 (version 2.17.1.4) tool was used to convert the bcl files to fastq files as well for adapter trimming and demultiplexing. GC-content, base-calling quality, adapter content and read length were measured using the tool FASTQC by Andrews (http://www.bioinformatics.babraham.ac.uk/projects/ fastqc/ accessed on 17 August 2021) and MultiQC [9]. Reads were then trimmed or discarded based on their base calling quality and adapter content with Trimmomatic version 0.36 [10]. Subsequently, with the help of the SortMeRNA algorithm version 2.1b [11], the extent of rRNA depletion was measured by mapping the reads to rRNA databases. For alignment and the following analyses, the human genomic reference sequence (GRCh38) and annotation data (release 101) were downloaded from Ensembl [12] and BioMart [13]. For splice site usage analysis, the reads were then aligned to the human reference genome using the two-pass mapping protocol of the STAR aligner (2.5.4b) [14]. With help of the SAMtools software package [15], uniquely mapped reads were selected for creation of a gap table, listing the coordinates of every gap found in the alignment of the reads and the number of overlapping reads. For DGE analysis with the R package DESeq2 version 1.18.1 [16], count matrices were generated using the software salmon version 0.9.1 [17]. Significantly enriched gene sets were calculated, using the R package GOseq [18]. Scripts used for this work are publicly available at https://github.com/caggtaagtat/SELENOT (accessed on 17 August 2021). FASTQ file preparation and alignment were accomplished using custom BASH shell scripts in the environment of the High Performing Cluster of the Heinrich-Heine University Düsseldorf.

cDNA Synthesis
Total cellular RNA was reverse transcribed using the QuantiTect Reverse Transcription kit (Qiagen, Hilden Germany) according to the manufacturer's instructions.

Polymerase Chain Reaction (PCR)
Endpoint PCRs were performed with MyTaq™ HS DNA Polymerase (Biocat, Heidelberg, Germany) according to manufacturer's recommendations in a Bio-Rad T100 Thermal Cycler (Bio-Rad, Feldkirchen Germany). Reaction products were resolved on standard agarose gels.
Relative transcript levels were determined by semi-quantitative real-time PCR using cDNA as a template and the primaQUANT 2x qPCR-SYBR-Green-MasterMix (Steinbrenner, Wiesenbach, Germany), the transcript for the ribosomal protein L32 (RPL32), served as a reference. The PCR reactions were performed in a Rotor-Gene Q instrument (Qiagen, Hilden, Germany). Relative expression was calculated by the ∆C t method [19].
The sequences of all primer used for PCR are listed in Supplementary Table S1.

Plasmids
A lentiviral expression vector for the first twenty amino acids of APEX1 was constructed by transferring the coding sequence for APEX1 (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) with a C-terminal myc-tag Antioxidants 2021, 10, 1427 4 of 14 from the previously published expression vector [6] into a lentiviral transfer vector, in which the transgene is expressed under the transcriptional control of the cytomegalovirus immediate early promoter/enhancer [8]. To generate an expression vector for human SELENOT with an N-terminal FLAG-tag, the SELENOT coding sequence together with the first 179 bp of the 3 -untranslated region of the human SELENOT gene containing the selenocysteine insertion sequence were amplified from a human EC cDNA using Q5 ® High-Fidelity DNA Polymerase (New England Biolabs, Frankfurt, Germany). This fragment was inserted into pFLAG-CMV-2 (Sigma-Aldrich, Deisenhofen, Germany) opened with Not I and Xba I using the Gibson Assembly ® Cloning kit (New England Biolabs, Frankfurt, Germany) according to the manufacturer's protocol. The construct was verified by DNA sequencing. Cloning details and the complete plasmid sequence are available upon request.

Transient Transfection of EC
Transient transfections of EC with plasmid DNA were performed using Superfect (Qiagen, Hilden, Germany) as previously described [20,21]. In detail, EC were transfected on 6 cm culture dishes with 3 µg plasmid DNA and 22.5 µL Superfect, or in 6-well plates with 1.2 µg plasmid DNA and 12 µL Superfect per well.
The following day, membranes were incubated with secondary antibodies coupled to horseradish peroxidase (ECL TM Anti-Rabbit or Anti-Mouse IgG, Horseradish Peroxidase linked whole antibody (from sheep), 1:5000; Cat. Nos. NA934V and NA931V, GE healthcare, Solingen, Germany). Detection was performed using ECL substrate (GE healthcare, Solingen, Germany) and X-ray films. Semi-quantitative analyses were performed on scanned X-ray films using Fiji [22].

Statistics
The number of experiments (n) given in the figure legends represents independent biological replicates, the data shown are mean ± SEM. Normal distribution for all data sets was confirmed by a Shapiro-Wilk test; homogeneity of variances (from means) between groups was verified by Levene's test. Multiple comparisons were performed using one-way ANOVA with post-hoc Tukey LSD test.
Notably, we observed clearly different LPS responses in cells expressing APEX1(1-20) when compared to cells transduced with the empty vector (Figure 1C-F and Supplementary  Tables S7-S10).
For functional studies, we focused on genes whose expression is upregulated by LPS only in cells expressing the small APEX1 peptide as the corresponding proteins might evoke APEX1(1-20)-dependent protective effects in EC, which could be of interest in a therapeutic setting.  Tables S7-S10).
For functional studies, we focused on genes whose expression is upregulated by LPS only in cells expressing the small APEX1 peptide as the corresponding proteins might evoke APEX1(1-20)-dependent protective effects in EC, which could be of interest in a therapeutic setting.

Expression of PXDN and SELENOT Is Specifically Upregulated after LPS Treatment of EC Expressing APEX1(1-20)
As a prerequisite for functional studies, we first validated the regulation of the topranked candidates, which, according to the RNA sequencing data, should be expressed to levels allowing reliable detection and quantification.
Antioxidants 2021, 10, 1427 7 of 14 IL1RL encodes an Interleukin 1 Receptor-like protein, which belongs to a family of ten distinct but structurally related receptors. These proteins serve either as ligand binding or accessory chains and some act as signaling inhibitors. Moreover, two members of this family are orphan receptors [23]. Therefore, IL1RL1 is part of a complex signaling network and one could easily envision that-due to this redundancy-interference with this network might be compensated or evoke unwanted side effects.
Peroxidasin (PXDN), originally described as Vascular Peroxidase 1, is a heme-containing peroxidase, which shows highest expression in the heart and the vascular wall [24]. The protein is rapidly secreted [25] and required for formation of the vascular basement membrane by reinforcing fibrillar network assembly in the extracellular matrix through formation of sulfilimine bonds [26]. It has recently been shown that PXDN promotes angiogenesis [27] and, furthermore, is essential for endothelial cell survival [28].
Selenoprotein T (SELENOT) is a member of the selenoprotein family, whose members are characterized by containing one or more selenocysteine residues, frequently in enzymatically active sites [29]. SELENOT is the most highly conserved selenoprotein throughout evolution [30], suggestive of an essential function, which is underscored by the early embryonic lethality of mice in which the selenot gene is constitutively disrupted [31]. SELENOT is one of 7 out of 25 human selenoproteins localized to the ER [32]. The expression of SELENOT, like all other selenoproteins, depends on dietary selenium as shown by a reduced expression in chicken stomach after 55 days on a selenium-deficient diet. Moreover, this regimen resulted in stress injuries [33]. In addition, SELENOT protects kidney cells against cisplatin-induced apoptosis [34]. These observations go along with the notion that ER-resident selenoproteins are critical in cellular stress responses [35].
For the reasons explained above, we did not follow up on IL1RL, but validated the regulation of PXDN and SELENOT by semi-quantitative real-time PCR (Figure 2).

Expression of PXDN and SELENOT Is Specifically Upregulated after LPS Treatment of EC Expressing APEX1(1-20)
As a prerequisite for functional studies, we first validated the regulation of the topranked candidates, which, according to the RNA sequencing data, should be expressed to levels allowing reliable detection and quantification.
IL1RL encodes an Interleukin 1 Receptor-like protein, which belongs to a family of ten distinct but structurally related receptors. These proteins serve either as ligand binding or accessory chains and some act as signaling inhibitors. Moreover, two members of this family are orphan receptors [23]. Therefore, IL1RL1 is part of a complex signaling network and one could easily envision that-due to this redundancy-interference with this network might be compensated or evoke unwanted side effects.
Peroxidasin (PXDN), originally described as Vascular Peroxidase 1, is a heme-containing peroxidase, which shows highest expression in the heart and the vascular wall [24]. The protein is rapidly secreted [25] and required for formation of the vascular basement membrane by reinforcing fibrillar network assembly in the extracellular matrix through formation of sulfilimine bonds [26]. It has recently been shown that PXDN promotes angiogenesis [27] and, furthermore, is essential for endothelial cell survival [28].
Selenoprotein T (SELENOT) is a member of the selenoprotein family, whose members are characterized by containing one or more selenocysteine residues, frequently in enzymatically active sites [29]. SELENOT is the most highly conserved selenoprotein throughout evolution [30], suggestive of an essential function, which is underscored by the early embryonic lethality of mice in which the selenot gene is constitutively disrupted [31]. SELENOT is one of 7 out of 25 human selenoproteins localized to the ER [32]. The expression of SELENOT, like all other selenoproteins, depends on dietary selenium as shown by a reduced expression in chicken stomach after 55 days on a selenium-deficient diet. Moreover, this regimen resulted in stress injuries [33]. In addition, SELENOT protects kidney cells against cisplatin-induced apoptosis [34]. These observations go along with the notion that ER-resident selenoproteins are critical in cellular stress responses [35].

Generation of a SELENOT Expression Vector and Intracellular Localization of the Overexpressed Protein
To study the impact of SELENOT on endothelial cell functions affected by LPS, we generated an expression vector, which contained a FLAG-epitope tag allowing the identification of the overexpressed protein. For the generation of this expression vector, an aspect unique to selenoproteins had to be taken into account. Selenocysteine (Sec) residues in selenoproteins are not the product of a post-translational modification, but are rather incorporated already during translation by using one of the translation termination codons, namely UGA, for binding of the selenocysteine tRNA (tRNA Sec ) to the mRNA. This translational recoding of the UGA codon involves a so-called selenocysteine insertion sequence (SECIS) in the 3 -untranslated region (UTR) of the transcript. The SECIS, which is not highly conserved on the sequence level, forms a stem-loop structure that is required for recruitment of the tRNA Sec to the UGA codon [36]. Consequently, the lack of a SECIS leads to premature translation termination, when the ribosome encounters the first UGA within the open reading frame. Therefore, we included-besides the SELENOT open reading frame-a portion of the SELENOT 3 -UTR including the SECIS in the expression vector.
We first analyzed the expression of FLAG-SELENOT after transient transfection of EC on the RNA level by reverse transcriptase PCR ( Figure 3A). We then determined the intracellular localization of the overexpressed FLAG-SELENOT protein by immunofluorescence. As demonstrated by colocalization with the ER-resident protein Calnexin ( Figure 3B), FLAG-SELENOT was localized in the ER.

Generation of a SELENOT Expression Vector and Intracellular Localization of the Overexpressed Protein
To study the impact of SELENOT on endothelial cell functions affected by LPS, we generated an expression vector, which contained a FLAG-epitope tag allowing the identification of the overexpressed protein. For the generation of this expression vector, an aspect unique to selenoproteins had to be taken into account. Selenocysteine (Sec) residues in selenoproteins are not the product of a post-translational modification, but are rather incorporated already during translation by using one of the translation termination codons, namely UGA, for binding of the selenocysteine tRNA (tRNA Sec ) to the mRNA. This translational recoding of the UGA codon involves a so-called selenocysteine insertion sequence (SECIS) in the 3′-untranslated region (UTR) of the transcript. The SECIS, which is not highly conserved on the sequence level, forms a stem-loop structure that is required for recruitment of the tRNA Sec to the UGA codon [36]. Consequently, the lack of a SECIS leads to premature translation termination, when the ribosome encounters the first UGA within the open reading frame. Therefore, we included-besides the SELENOT open reading frame-a portion of the SELENOT 3′-UTR including the SECIS in the expression vector.
We first analyzed the expression of FLAG-SELENOT after transient transfection of EC on the RNA level by reverse transcriptase PCR ( Figure 3A). We then determined the intracellular localization of the overexpressed FLAG-SELENOT protein by immunofluorescence. As demonstrated by colocalization with the ER-resident protein Calnexin (Figure 3B), FLAG-SELENOT was localized in the ER.

SELENOT Overexpression Inhibits LPS-Induced Endothelial Cell Activation
Having demonstrated that FLAG-SELENOT is localized in the ER, we next investigated the effect of SELENOT on LPS-induced endothelial cell activation. Therefore, FLAG-SELENOT was expressed in EC as before. After treatment with 150 ng/mL LPS for 18 h, ICAM1-a marker for endothelial cell activation-was detected. As expected, LPS upregulated ICAM1 protein levels in empty vector transfected EC. This upregulation was completely inhibited in cells, in which SELENOT is overexpressed (Figure 4).
SELENOT fusion transcript, the housekeeping gene RPL32 served as control. Amplification products were resolved by agarose gel electrophoresis, the expected fragment sizes are specified, numbers on the left indicate selected DNA size markers (M). (B) Localization of FLAG-SELENOT was examined by immunostaining and fluorescence microscopy. Cells were stained with an antibody directed against Calnexin (CANX), a marker for the ER (green) and an anti-FLAG antibody (red). Nuclei were counterstained with DAPI (blue); merge is the overlay of all channels (scale bar = 30 μm).

SELENOT Overexpression Inhibits LPS-Induced Endothelial Cell Activation
Having demonstrated that FLAG-SELENOT is localized in the ER, we next investigated the effect of SELENOT on LPS-induced endothelial cell activation. Therefore, FLAG-SELENOT was expressed in EC as before. After treatment with 150 ng/mL LPS for 18 h, ICAM1-a marker for endothelial cell activation-was detected. As expected, LPS upregulated ICAM1 protein levels in empty vector transfected EC. This upregulation was completely inhibited in cells, in which SELENOT is overexpressed (Figure 4).

SELENOT Overexpression Inhibits LPS-Induced Endothelial Cell Apoptosis
Besides endothelial cell activation, LPS also induces apoptosis of EC [37]. Therefore, we determined Caspase 3 cleavage as a marker for apoptosis in EC. As for ICAM1, LPS increased Caspase 3 cleavage in cells not expressing SELENOT. On the contrary, overexpression of SELENOT completely blunted apoptosis induction by LPS ( Figure 5).

SELENOT Overexpression Inhibits LPS-Induced Endothelial Cell Apoptosis
Besides endothelial cell activation, LPS also induces apoptosis of EC [37]. Therefore, we determined Caspase 3 cleavage as a marker for apoptosis in EC. As for ICAM1, LPS increased Caspase 3 cleavage in cells not expressing SELENOT. On the contrary, overexpression of SELENOT completely blunted apoptosis induction by LPS ( Figure 5).
SELENOT is an ER-resident selenoprotein, which is associated with the ER membrane and required to maintain ER redox homeostasis. It is needed to cope with intracellular stress conditions and is one of the most important selenoproteins [30].
As mentioned before, the expression of all selenoproteins depends on selenium. However, there seems to be a hierarchy in the sensitivity of different selenoproteins with respect to selenium levels and SELENOT seems to respond more avidly to selenium depletion than several other proteins of this family [40]. It has been estimated that up to one
SELENOT is an ER-resident selenoprotein, which is associated with the ER membrane and required to maintain ER redox homeostasis. It is needed to cope with intracellular stress conditions and is one of the most important selenoproteins [30].
As mentioned before, the expression of all selenoproteins depends on selenium. However, there seems to be a hierarchy in the sensitivity of different selenoproteins with respect to selenium levels and SELENOT seems to respond more avidly to selenium depletion than several other proteins of this family [40]. It has been estimated that up to one in seven people worldwide have a low dietary selenium intake [41] and it is clear that proper endothelial functionality depends on an adequate selenium supply [42]. Even more interesting is the observation that selenium serum levels are dramatically reduced in critically ill patients with sepsis [43]. Therefore, selenium supplementation seems to be an obvious supplementary treatment option for sepsis and possibly the protection of the endothelium in this disease. In this context, it is interesting to note that selenium pretreatment or supplementation alleviates some of the deleterious effects of LPS. In the murine macrophage cell line RAW264.7, LPS induced immunological stress as shown by the upregulation of multiple inflammation-related genes. This was accompanied by a reduction in the relative selenot mRNA level. Pretreatment with selenium partially rescued this downregulation and had only a very modest effect on the expression of the inflammation-related genes [44]. In mice, LPS-induced myocardial dysfunction, oxidative stress and apoptosis in the heart could be attenuated when the animals were put on a selenium-supplemented diet 2 weeks prior to LPS treatment [45]. Again, this pretreatment did not completely restore heart functionality or prevent oxidative stress and apoptosis induction evoked by LPS. Our experiments did not show a significant downregulation of selenot expression in LPS-treated EC, although there seems to be a trend in this direction. On the contrary, the cells expressing APEX1 (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20) showed an upregulation of selenot RNA levels of approximately threefold after LPS treatment. This clearly indicates that the small APEX1 peptide can convey a protective outcome, which is much stronger than the effects observed with selenium supplementation or pretreatment.
Up to now, the precise molecular functions of SELENOT have not been elucidated. Nevertheless, a peptide derived from SELENOT has already been used in animal models. Rocca et al. demonstrated that this SELENOT-derived peptide-including the active catalytic site corresponding to the sequence FQICVSUGYR-applied after ischemia and prior to reperfusion is able to protect the heart from ischemia/reperfusion injury. This protection was attributed to a reduction in oxidative stress and inhibition of apoptosis [46]. This is in accordance with our study presented here, in which we demonstrate that SELENOT completely inhibited LPS-induced activation and apoptosis in human primary EC.
The same peptide was applied in a cell-permeable form in a mouse model for Parkinson's disease, where it protected dopaminergic neurons. This effect was also associated with reduced oxidative stress and Caspase 3 activity [47].
Based on the protective effects of this SELENOT peptide in such different organs as the brain and the heart, it is conceivable that it could exert its protective functions also in the vasculature in the setting of sepsis.
Given the high numbers of patients and the up to 11 million deaths per year due to sepsis, a protection of the endothelium as an additional additive therapy could be of tremendous importance. The metabolic response to sepsis entails the rapid breakdown of intracellular reserves of proteins, carbohydrates and fat. This is accompanied by an increase in ER stress. An increase in SELENOT or application of a peptide could dampen this stress and maintain the ER homeostasis, counteracting the overshooting responses of the body to sepsis.