Peucedanum ostruthium Inhibits E-Selectin and VCAM-1 Expression in Endothelial Cells through Interference with NF-κB Signaling

Twenty natural remedies traditionally used against different inflammatory diseases were probed for their potential to suppress the expression of the inflammatory markers E-selectin and VCAM-1 in a model system of IL-1 stimulated human umbilical vein endothelial cells (HUVEC). One third of the tested extracts showed in vitro inhibitory effects comparable to the positive control oxozeaenol, an inhibitor of TAK1. Among them, the extract derived from the roots and rhizomes of Peucedanum ostruthium (i.e., Radix Imperatoriae), also known as masterwort, showed a pronounced and dose-dependent inhibitory effect. Reporter gene analysis demonstrated that inhibition takes place on the transcriptional level and involves the transcription factor NF-κB. A more detailed analysis revealed that the P. ostruthium extract (PO) affected the phosphorylation, degradation, and resynthesis of IκBα, the activation of IKKs, and the nuclear translocation of the NF-κB subunit RelA. Strikingly, early effects on this pathway were less affected as compared to later ones, suggesting that PO may act on mechanism(s) that are downstream of nuclear translocation. As the majority of cognate NF-κB inhibitors affect upstream events such as IKK2, these findings could indicate the existence of targetable signaling events at later stages of NF-κB activation.


Introduction
The inflammatory reaction is a common feature of several diseases, including cardiovascular, gastrointestinal, and neurodegenerative, as well as joint and skin disorders. Inflammatory mediators such as IL-1 and TNF or bacterial LPS evoke the expression of a complex set of genes in the endothelium that encode, e.g., chemoattractants and adhesion molecules, a pre-requisite for the adherence and transmigration of immune cells through the vascular wall into the underlying tissue, where they migrate towards the site of injury. However, they also fulfill other functions that are related to proliferation, cell and tissue dynamics, metabolism, survival and apoptosis [1]. Examples of induced genes include the adhesion molecules E-selectin, ICAM-1, VCAM-1, the cytokines IL-1, IL-6, IL-8, chemokines and their receptors, cyclooxygenase, and anti-apoptotic proteins of the IAP family and A20.
The transcription factor NF-κB has been demonstrated to play a major role in the expression of many of these genes, and inhibition of its activation was shown to prevent or at least attenuate pro-inflammatory gene expression. NF-κB comprises a family of five members that can form homo-

Cytotoxicity Assays
HUVECs were incubated for 4 h with different concentrations of PO (50, 25, 12.5 µg/mL), and the CellTox Green Cytotoxicity Assay (Promega, Madison, WI, USA, #G8741) was performed according to the manufacturer's recommendations. Cytotoxicity was judged by morphological examination.

Real-Time PCR
Total RNA was isolated using the PeqGold Total RNA Isolation Kit (VWR International, Radnor, USA, #732-2868) according to the manufacturer's instructions. A total of 1 µg RNA was reverse-transcribed using random hexamers (Fisher Scientific, Schwerte, Germany; #SO142) and murine leukemia virus reverse transcriptase (Fisher Scientific, #10338842). Primers were designed using the software "Primer3", sequences are given in Table S2 (Supplementary Materials). Real-time PCR was performed with the SsoAdvanced Universal SYBR Green Supermix (BioRad, Vienna, Austria, #1725272) using the StepOnePlus instrument (Applied Biosystems, Foster City, CA, USA), and relative mRNA expression normalized to GAPDH. Fold changes in mRNA expression were calculated according to the 2-∆∆Ct method. Results are shown as mean fold induction of averaged Ct-values of triplicates.  The TAK inhibitor (TAK-I) was used as positive control at 10, 5, 2.5 µM. Relative levels of E-selectin as compared to IL-1 alone are shown. Samples marked with an "+" indicates apparent cytotoxicity as judged by morphological examination.

Western Blotting
Post-confluent HUVEC grown in 6-well plates were pre-incubated for 30 min with 50 µg/mL PO, stimulated with IL-1ß (5 ng/mL) for different periods of time (0, 10, 30, 90 min) and lysed in Laemmli buffer. Western blotting for IκBα, IKK2, their phosphorylated forms, as well as ß-actin, was done as described [21]. Densitometric analysis was done with Image J.

Transfection and Reporter Gene Assays
HUVEC were grown as described above and transfected with pNL3.2.NF-κB-RE (Promega) and pmaxGFP (Amaxa/Origene, Rockville, MD, USA) by electroporation using a BioRad Gene Pulser with the settings 200 V/960 µF. Then, 5 × 10 6 cells were electroporated in 400 µL RPMI medium in 0.4 cm cuvettes with a total of 10 µg plasmid DNA. Cells were seeded into 6-well plates and grown for two days before stimulation. Luciferase levels were analyzed using the NanoGlo Luciferase Assay (Promega, #N1110) according to the manufacturer's protocol and normalized to EGFP fluorescence.

Ultra Performance Liquid Chromatography
UPLC analysis was performed on a Waters Acquity UPLC system (Waters Corporation, Milford, MA, USA; H-class) equipped with a quaternary solvent manager, a sample manager, a column manager, an isocratic solvent manager, a photodiode array (PDA) detector, an evaporative light scattering detector (ELSD), and a fraction collector. A Waters Acquity UPLC BEH Phenyl column (1.7 µm, 2.1 × 100 mm) was used for analytical experiments. Data acquisition and processing were conducted using the operating software Waters Empower 3. PO was chromatographed over UPLC using a binary mobile phase system consisting of (A) H2O and (B) CH 3 CN. The gradient was from 13-98% B in 12 min followed by 5 min re-equilibration. Method in detail: 13% B for 0.5 min, 13-18% B in 0.5 min, 18-45% B in 1 min, isocratic 45% B for 1.7 min, 45-73% B in 2.8 min, 73-98% B in 0.3 min, isocratic 98% B for 5 min, 98-13% B in 0.1 min, isocratic 13% B for 0.1 min; Conditions: temperature, 40 • C; flow rate, 0.300 mL/min; injection volume, 1 µL. Detection of compounds using PDA and ELSD. PDA conditions: 210 nm and full range spectra 192-400 nm. Further, the UPLC system was coupled to an Acquity QDa mass detector with an electrospray ionization source. An isocratic solvent manager was used as a make-up pump and positioned before the mass detector. The main flow stream was then split (1:10) and compounds detected in the positive ionization mode. (Ultrahigh-)gradient grade solvents from VWR Chemicals were used for all analytical experiments.

Statistical Significance Calculations
Differences between samples were analyzed by Ordinary one-way ANOVA with Dunnett´s multiple comparisons test using Graph Pad Prism software, San Diego, CA, USA. *, **, and *** indicate p < 0.05, 0.01, and 0.001, respectively.

Results
Within this study, natural materials from 17 plant species and three polypore species have been selected based on several criteria, primarily because of their long-standing traditional use in the field of inflammation and related areas, as well as anti-inflammatory effects described in the literature without in depth knowledge of their molecular mechanisms. Besides the 17 traditional herbal drugs, three polypore species were selected for the preparation of extracts, since they have been extensively used not only in TCM for the treatment of various ailments, but also for their health-promoting effects around the world [22]. A summary of the herbal and fungal materials used in this study, along with their traditional use and cognate pre-clinical and clinical studies, is given in Table 2.
Extracts of the selected herbal drugs and mushrooms were prepared to cope with a broad spectrum of metabolites endowed with drug like properties [11,12]. As a first assessment of their in vitro anti-inflammatory activity, we used a cell ELISA to screen these extracts for suppression of IL-1 stimulated E-selectin expression in human umbilical vein endothelial cells (HUVEC). The adhesion molecule E-selectin was chosen as a read-out, since it mediates one of the key steps in the inflammatory reaction, the initial weak adhesion ("rolling") of leucocytes on the endothelial wall. As shown in Figure 1, many of the tested extracts inhibited E-selectin expression in a dose-dependent manner. Table 2. Herbal and fungal materials with anti-inflammatory activity selected for this study, including their traditional application and reported in vitro and in vivo data; clinical evidence also includes studies not directly linked to inflammation.
Effective against allergic conjunctivitis and conjunctivitis due to external irritants or other causes [47].
We selected the extract of P. ostruthium (PO) for further study based on the strength and robustness in repetitive experiments of its inhibitory effect. PO was characterized by PDA/ELSD-UPLC, and by means of UPLC-ESI-MS its main constituents could be dereplicated ( Figure S1; Supplementary Materials). In Table S1 (Supplementary Materials) the results from the dereplication via literature search in SciFinder (accessed 2020/05/11) are presented. As a first step into the mechanistic investigation of E-selectin expression, we analyzed the mRNA levels after pre-incubation of HUVEC with different concentrations of PO (i.e., 50, 25 and 12 µg/mL, respectively) and stimulation with 5 ng/mL IL-1. In addition, another adhesion molecule, VCAM-1, which mediates the firm adhesion of leucocytes to the endothelium, was analyzed. The choice of pre-incubation time was done based on previous experience with herbal extracts [20] and on experiments analyzing different times of pre-incubation depending on the concentration ( Figure S2; Supplementary Materials). PO did not show toxicity at these concentrations ( Figure S3; Supplementary Materials). As shown in Figure 2, PO strongly inhibited the expression of both adhesion molecules on the mRNA level in a dose-dependent manner, ranging from 50 to 15 µg/mL. Since mRNA levels can be regulated on the level of transcription but also depend on other factors such as mRNA stability, we then determined whether the inhibition of E-selectin expression occurs on the transcriptional level. HUVEC were transfected with an E-selectin promoter luciferase reporter construct and, two days later, after reaching post-confluency, they were pre-incubated with PO and stimulated with IL-1. As shown in Figure 3A, PO inhibited the activity of the E-selectin promoter. Since the expression of E-selectin (and also VCAM-1) is dependent on NF-κB, we investigated whether PO might act via inhibition of this transcription factor. Therefore, HUVEC were transfected with a construct containing a multimerized NF-κB binding site driving the expression of a luciferase Since mRNA levels can be regulated on the level of transcription but also depend on other factors such as mRNA stability, we then determined whether the inhibition of E-selectin expression occurs on the transcriptional level. HUVEC were transfected with an E-selectin promoter luciferase reporter construct and, two days later, after reaching post-confluency, they were pre-incubated with PO and stimulated with IL-1. As shown in Figure 3A, PO inhibited the activity of the E-selectin promoter. Since the expression of E-selectin (and also VCAM-1) is dependent on NF-κB, we investigated whether PO might act via inhibition of this transcription factor. Therefore, HUVEC were transfected with a construct containing a multimerized NF-κB binding site driving the expression of a luciferase reporter gene, treated as above and analyzed for luciferase expression. PO diminished NF-κB activity at all concentrations tested (although statistical significance was not reached; Figure 3B). Since one of the key steps in NF-κB signaling is the phosphorylation and degradation of its inhibitor IκBα, which takes place within minutes after stimulation and is followed by re-synthesis that starts approx. one hour later [20,119], we assayed IκBα and phospho-IκBα levels by Western blotting. HUVEC were pre-incubated and stimulated as above, except that the times of stimulation were adjusted to assay for both the early and late events. As shown in Figure 4, IκBα was degraded within 10 min following IL-1 stimulation and re-synthesized at 90 min; in PO-treated cells, degradation was slightly delayed, but re-synthesis entirely inhibited. Accordingly, and in contrast to IL-1 stimulation alone, IκBα phosphorylation was still visible at 10 min in IL-1+PO-treated cells, suggesting a delay in degradation. Previously, the group of D. Baltimore described "oscillations" of NF-κB nuclear-cytoplasmic translocation due to several rounds of degradation and re-synthesis of IκBα as well as other IκB species [120]. Whereas this phenomenon can be seen best in T cells where multiple waves can be observed, endothelial cells also show at least one additional wave of NF-κB nuclear translocation and the associated IκBα phosphorylation and degradation. In our experiments, Figure 3. PO inhibits E-selectin promoter and NF-κB activity. HUVEC were transfected with either an E-selectin (A) or a NF-κB luciferase reporter construct (5xNF-κB luc; (B)) and EGFP as internal control, pretreated with PO (or TAK inhibitor as positive control) and stimulated with IL-1 as indicated. Luciferase levels were determined after 16 h and are shown as relative levels normalized to EGFP. Triplicate samples were analyzed. * p < 0.05, ** p < 0.01 as compared to IL-1 stimulation; C: unstimulated control.
Since one of the key steps in NF-κB signaling is the phosphorylation and degradation of its inhibitor IκBα, which takes place within minutes after stimulation and is followed by re-synthesis that starts approx. one hour later [20,119], we assayed IκBα and phospho-IκBα levels by Western blotting. HUVEC were pre-incubated and stimulated as above, except that the times of stimulation were adjusted to assay for both the early and late events. As shown in Figure 4, IκBα was degraded within 10 min following IL-1 stimulation and re-synthesized at 90 min; in PO-treated cells, degradation was slightly delayed, but re-synthesis entirely inhibited. Accordingly, and in contrast to IL-1 stimulation alone, IκBα phosphorylation was still visible at 10 min in IL-1+PO-treated cells, suggesting a delay in degradation. Previously, the group of D. Baltimore described "oscillations" of NF-κB nuclear-cytoplasmic translocation due to several rounds of degradation and re-synthesis of IκBα as well as other IκB species [120]. Whereas this phenomenon can be seen best in T cells where multiple waves can be observed, endothelial cells also show at least one additional wave of NF-κB nuclear translocation and the associated IκBα phosphorylation and degradation. In our experiments, this second wave is indicated by the re-appearance of pIκBα between 30 and 90 min; it is lacking in PO (as well as in TAK-I) pretreated cells. In addition, the phosphorylation of IKKs in the activation loop, which is a measurement for their activity [121], did not show, besides a weak prolongation, a pronounced difference between IL-1-and IL-1+PO-treated HUVEC; together, these data indicate that the initial steps of activation, namely IKK activation and IκBα phosphorylation/degradation, are weakly affected, whereas the later steps are predominant. This suggests that in HUVEC, PO displays two types of activity towards NF-κB, one (weaker) that affects the initial steps of activation, and a second, more pronounced one that is directed towards a downstream part of the NF-κB signaling cascade.
Biomolecules 2020, 10, x FOR PEER REVIEW 12 of 23 this second wave is indicated by the re-appearance of pIκBα between 30 and 90 min; it is lacking in PO (as well as in TAK-I) pretreated cells. In addition, the phosphorylation of IKKs in the activation loop, which is a measurement for their activity [121], did not show, besides a weak prolongation, a pronounced difference between IL-1-and IL-1+PO-treated HUVEC; together, these data indicate that the initial steps of activation, namely IKK activation and IκBα phosphorylation/degradation, are weakly affected, whereas the later steps are predominant. This suggests that in HUVEC, PO displays two types of activity towards NF-κB, one (weaker) that affects the initial steps of activation, and a second, more pronounced one that is directed towards a downstream part of the NF-κB signaling cascade.  In order to further analyze the mechanism and kinetics of PO on NF-κB activation, we performed immunostaining to asses NF-κB nuclear translocation. HUVEC were pre-treated as above, and stimulated with 5 ng/mL IL-1 for 15, 30, and 90 min. As shown in Figure 5A, the p65 subunit of NF-κB (RelA) translocated into the nucleus within 15 min, whereas this was delayed in PO-treated cells ( Figure 5B). The later time points were not affected. This confirmed that PO does have an effect on an early step of NF-κB signaling, but that additional mechanisms at later stages of the signaling cascade are operative, that add to the later events such as the lack of IκBα re-synthesis, the overall diminished activity in the reporter gene assay and, as a result, diminished adhesion molecule expression.
In order to further analyze the mechanism and kinetics of PO on NF-κB activation, we performed immunostaining to asses NF-κB nuclear translocation. HUVEC were pre-treated as above, and stimulated with 5 ng/mL IL-1 for 15, 30, and 90 min. As shown in Figure 5A, the p65 subunit of NF-κB (RelA) translocated into the nucleus within 15 min, whereas this was delayed in PO-treated cells ( Figure 5B). The later time points were not affected. This confirmed that PO does have an effect on an early step of NF-κB signaling, but that additional mechanisms at later stages of the signaling cascade are operative, that add to the later events such as the lack of IκBα re-synthesis, the overall diminished activity in the reporter gene assay and, as a result, diminished adhesion molecule expression.

Discussion
Based on the long-standing traditional use of masterwort in the Alpine region and further studies in the scientific literature, but limited knowledge of its mode of action, we selected PO from a set of extracts with anti-inflammatory activity for further analysis in our model system. Peucedanum Figure 5. PO delays nuclear translocation of NF-κB. HUVEC were either (A) stimulated with 5 ng/mL IL-1 alone for the indicated times or (B) pre-treated for 30 min with 50 ng/mL PO before IL-1 stimulation, and immunostained for the p65/RelA subunit of NF-κB (green) and actin (red). Nuclei were stained with DAPI (blue). Merged pictures are shown on the right.

Discussion
Based on the long-standing traditional use of masterwort in the Alpine region and further studies in the scientific literature, but limited knowledge of its mode of action, we selected PO from a set of extracts with anti-inflammatory activity for further analysis in our model system. Peucedanum (Apiaceae) comprises approximately 120 species in several continents including Europe, Africa, Asia, and North America. Many of these have been used in traditional medicine for the treatment of a variety of diverse disorders, e.g., pyrexia, cardiovascular, gastrointestinal or neurological diseases. This is likely due to the broad spectrum of constituents, including various coumarins, phenolic acids, amines, glycosides, flavonoids, diterpenes and components of their essential oils [122].
Only recently, we identified and isolated several bioactive compounds from PO by applying a novel biochemometric approach named Eliciting Nature's Activities (ELINA; [18,130]) with an NF-kB reporter-gene assay plus two assays for E-selectin and VCAM-1 as biological read-out. However, the anti-inflammatory activity could only partially recovered (as compared to the total extract) by the combination of imperatorin and peucenin as a mixture, suggesting additional combinatorial effects [18].
Therefore, we here investigated the anti-inflammatory activity of the total extract in more detail and provide first insight into its molecular mechanism(s) of action. In IL-1 stimulated endothelial cells, PO suppressed the expression of the adhesion molecule E-selectin. Inhibition was found on both the mRNA and protein levels. Moreover, PO inhibited the activity of E-selectin on the transcriptional level, as demonstrated by reporter gene experiments ( Figure 3A). Moreover, since E-selectin transcription is dependent on NF-κB, we also investigated the activity of PO on this transcription factor by reporter gene analysis and demonstrate that PO strongly inhibited NF-κB activity ( Figure 3B).
One of the key steps in the NF-κB signaling pathway is the phosphorylation and ubiquitin-dependent degradation of the inhibitor IκBα, which forms a cytoplasmic complex with NF-κB p65/p50 to prevent its nuclear translocation. In contrast to IL-1 alone, IL-1 in combination with PO treated cells showed a slight delay in IκBα phosphorylation and degradation, and a total lack of IκBα re-synthesis (which is NF-κB-dependent). The re-appearance of phospho-IκBα specific species at 30 and 90 min after treatment, indicative of a second wave of NF-κB activation, is lacking in PO-treated cells. This second wave has been observed in other cells, and was most pronounced in T cells where even multiple oscillations can occur that result from repeated degradation and resynthesis in the NF-κB/IκBα (and other IκBs such as -ß and -e) system [120]. It has been speculated that these have differential effects on the expression of NF-κB-dependent genes [131].
To investigate upstream events, we additionally probed for activation of IKK2, the main kinase responsible for IκBα phosphorylation and found that, in line with the IκBα data, its activation was only slightly inhibited. In contrast, our positive control (5Z)-7-oxozeaenol that inhibits TAK1, a kinase that together with its co-activators TAB1 and -2/3 transmits the signal from the IL-1 (and also TNF) receptor to activate IKK2, completely prevented IKK2 phosphorylation and subsequent IκBα degradation. Together, it can be concluded that PO displays a weak effect on either IKK2 or an upstream signaling molecule, but that additional more downstream effects have to be operative. This was subsequently confirmed by immunofluorescence studies, which showed a delay in nuclear translocation of NF-κB ( Figure 5, compare 15 min timepoint between IL-1 (A) and IL-1+PO, (B)). Surprisingly, however, at 30 and 90 min, NF-κB nuclear translocation was indistinguishable between these conditions. This suggests that, at these timepoints, although NF-κB resides in the nucleus, its activity must be impaired by other mechanisms. At this point, we can only speculate about their nature, but several possibilities have been described, e.g., post-translational mechanisms such as phosphorylation, acetylation, S-nitrosylation that affect either DNA-binding or transactivation, miRNAs, or the crosstalk with other signaling pathways such as PPARα and -ß, ATF3, or STAT3. These will be the subject of further investigations.

Conclusions
The anti-inflammatory activity that has been ascribed to PO by traditional medicine and previous reports could be substantiated in this study and attributed to the inhibition of gene expression of the pro-inflammatory cell adhesion molecules E-selectin and VCAM-1. Thereby, PO acts by inhibiting predominantly, but probably not exclusively, the activity of NF-κB. Although this inhibition does involve, to a certain degree, the initial steps of activation including nuclear translocation, a major part of the inhibitory activity takes place at later stages. From a therapeutic point of view NF-κB inhibition holds promise for the treatment of a wide variety of mainly inflammation-related disorders, but also others such as, e.g., therapy-resistance in cancer. However, most cognate NF-κB inhibitors target signaling molecules that act early in the pathway such as IKK2, TAK1, or IκBα (through inhibition of ubiquitination). Our observation that PO also inhibits later stages of activation could be of interest, since it suggests that it affects signaling molecules which are distinct from cognate drug targets. Identification of these may open the way for the identification of lead compounds and, subsequently, novel drugs.  Table S1: Results from the dereplication of PO via literature search; Table S2: Primers for real-time PCR.