LMWF5A suppresses distinct pro-inflammatory cytokines in LPS-stimulated PBMC
We have previously shown that LMWF5A exhibits anti-inflammatory properties with an ≈35% inhibition of TNFα from LPS-induced human ex vivo PBMC cultures [22]. A 48-plex ELISA array of cytokines and chemokines was employed to assess a broader effect in the same model. Consistent with historical findings, LMWF5A treatment resulted in a significant (p ≤ 0.05) 38 ± 6.7% inhibition or 0.62 ± 0.07-fold change in the measured optical density (OD) signal for TNFα as compared to controls (Table 1). Also observed in conditioned media from LMWF5A-treated cells was significant suppression of CXCL9 (0.48 ± 0.01 fold change), CXCL10 (0.36 ± 0.15-fold change), CXCL11 (0.58 ± 0.08-fold change), IL-1β (0.71 ± 0.05-fold change), and IL-12 (0.34 ± 0.07-fold change) (Table 1). These findings demonstrate that in addition to TNFα, LMWF5A treatment of LPS-stimulated PBMC results in a distinct cytokine signature, with reduced release of specific pro-inflammatory signals after 24 hours in culture.
Pathway enrichment analysis of LMWF5A differentially expressed gene sets identifies associations with CD4+ Th1 inflammatory and M1 polarization markers
Next, gene set enrichment analysis was used to gain mechanistic insight into the biologic processes underlying the identified differentially expressed cytokines by querying Enrichr using cytokine gene symbols [27,28]. As expected for our LPS-stimulated PBMC model, the two most significant enrichment terms returned from the Wikipathways library involve LPS-related Toll-like receptor signaling pathways (Pathway#s WP75 and WP1449) (Table 2). However, these are closely followed by overrepresentation in pathway terms for allograft rejection as the result of CD4+ T cell differentiation (Pathway# WP2328), the differentiation or polarization of innate lymphoid progenitor cells (Pathway# WP3893), and the AhR signaling pathway (Pathway# WP2873) (Table 2). The top three pathways identified in the Biocarta library are associated with IL-2/IFNγ/STAT-induced Th1 differentiation (Systemic names M6231 and M4319) and the activation of NF-κB (Systemic name M2821) (Table 2). Interestingly, Biocarta also found significant overrepresentation in two well-known anti-inflammatory pathways: the PPARγ-related obesity pathway (Systemic name M22017) and the IL-10/JAK/STAT signaling pathway that result in the repression of TNFα, IL-1, and IL-6 (Systemic name M6778) (Table 2). Finally, when mined by literature in the Elsevier pathway collection, M1 macrophage polarization is associated with the LMWF5A gene set (Table 2). Together these findings suggest that the cytokine genes found to be influenced by LMWF5A treatment of LPS-stimulated PBMC are linked to classical CD4+ Th1 T cell activation, M1 macrophage polarization phenotypes, and the modulation of pro- and anti-inflammatory pathways, including PPARγ, AhR, NF-κB, and STAT.
Transcription factor enrichment analysis suggests that LMWF5A treatment involves changes in NF-kB and STAT activity
To identify transcription factors that are potentially responsible for the LMWF5A-reduced cytokine signature, differentially abundant cytokines were input into Chea3 [29]. When sorted by mean rank, the most highly associated transcription factor predicted to regulate this set of cytokines is basic leucine zipper transcription factor ATF-like 3 (BATF3), which heterodimerizes with AP-1 to function as a transcriptional repressor and may play a role in the fate of T-cell differentiation [30] (Table 3). Importantly, pro-inflammatory NF-κB family member subunits (NFKB2 and REL) and STAT1 are also represented (Table 3). In support of the initial rankings, the NF-κB subunits RELA, NFKB2, and RELB as well as STAT1 are overrepresented when assembled by gross number of overlapping genes (Table 3). In addition, enrichment analysis by gross number indicates that most of the genes submitted have upstream c-Jun AP-1 promoter elements (Table 3). Thus, these data imply that the cytokines that are reduced by LMWF5A in PBMCs are, in part, regulated by NF-κB, STAT, and AP-1.
LMWF5A reduces transcription factor/DNA-binding activity
The regulatory picture provided by enrichment analysis suggests that LMWF5A treatment of LPS-stimulated PBMC results in reduced Th1/M1 differentiation or activation through attenuation of the activity of hallmark pro-inflammatory TFs, such as AP-1, NF-κB, and STAT. To test this hypothesis, nuclear protein extracts were collected from resting and stimulated PBMC cultured in the presence of saline or LMWF5A for 24 hours and TF activation was assessed by specific DNA-binding ELISAs. Phosphorylated c-Jun antibody was chosen for ELISA quantification to represent AP-1 activation, but no changes in optical density (OD) were observed between diluent controls and LMWF5A treatment groups (data not shown). However, as presented in Figure 1A and 1B, significant reductions in ELISA OD measurements for the canonical p65 and non-canonical RelB NF-κB subunits were observed in LPS-stimulated, LMWF5A-treated nuclear protein samples. In addition, STAT1α and STAT3 DNA binding was reduced in the nuclear compartment in both resting and stimulated PBMC following LMWF5A treatment (Figure 1C and 1D).
To gauge the magnitude and repeatability of these responses, fold changes in OD between saline control and LMWF5A treatments for 4-6 independent experiments were then non-parametrically compared to a hypothetical fold change of 1.0 for both unstimulated and LPS-stimulated cells. In addition, fold changes in the LPS-induced activation signal, adjusted to account for constituent resting cell activity, were also calculated by subtracting the basal resting OD of the unstimulated, saline-treated sample for each experiment prior to fold change calculation. Due to inherent variability in the LMWF5A-induced fold change when measured at this time point, no significant change in NF-κB p65 was observed from hypothetical 1.0 through non-parametric analysis; however, the fold change in p65 activity trends to be lower with LMWF5A treatment compared to the saline control in unstimulated cultures (Median = 0.87, Interquartile Range [IQR] = 0.79 to 0.92, p = 0.14), LPS-stimulated cultures (Median = 0.78, IQR = 0.75 to 0.85, p = 0.06), and the overall activation signal (Median = 0.69, IQR = 0.62 to 0.81, p = 0.10) (Figure 2 A,B,C). Interestingly, RelB activity was not detected in resting cells and replicates exhibited a broad distribution in fold change (Median = 0.44, IQR = 0.26 to 0.71, p = 0.20, Figure 2B) following stimulation, suggesting that this target is temporally diverse. As for STAT1α, LMWF5A treatment results in a significant reduction in the OD measurement quantifying DNA binding in unstimulated cells (Median = 0.58, IQR = 0.47 to 0.77, p = 0.036), stimulated cells (Median = 0.66, IQR = 0.58 to 0.73, p = 0.036), and the overall activation signal (Median = 0.49, IQR = 0.38 to 0.57, p = 0.036) (Figure 2 A,B,C). When STAT3 activity was evaluated, no significant change was observed in unstimulated cells (Median = 0.69, IQR = 0.57 to 0.90, p = 0.59) while stimulated cells (Median = 0.74, IQR = 0.68 to 0.88, p = 0.059) and the overall activation signal (Median = 0.65, IQR = 0.63 to 0.72, p = 0.059) trend to exhibit reduced DNA-binding in LMWF5A treatment groups (Figure 2 A,B,C). These findings suggest that LMWF5A reduces STAT1α activity, as measured by protein binding to specific DNA motifs, 24 hours after treatment. Furthermore, LMWF5A may also reduce, to a lesser degree, the detectable DNA-binding activity of STAT3 as well as p65 and RelB NF-κB family members subunits.
LMWF5A reduces NF-κB luciferase reporter activity
NF-κB signaling is the culmination of hetero- and homodimers, consisting of five family member TFs, that can form a least 12 distinct combinations [31]. To address these complexities and capture the broader impact of LMWF5A on NF-κB signaling, a luciferase reporter system was employed. Because PBMC present technical difficulties for this model, a stably transfected NF-κB promoter-driven HEK293 reporter line (Invivogen HEK-Dual TNFα) was purchased and grown under selection conditions. The stably transfected construct consists of Lucia luciferase under the control of an IFN-β minimal promoter fused to five copies of the NF-κB consensus translational response element and three copies of the c-Rel binding site. HEK-Dual TNFα cells were cultured for 24 hours with saline or serially diluted LMWF5A, and luciferase expression was evaluated 2 hours after stimulation. Because HEK293 cells do not express innate pattern-recognition machinery in a similar fashion to PBMC, these cells were alternatively stimulated with 2 ng/ml TNFα, instead of LPS, to trigger robust signaling and activation. LMWF5A treatment resulted in the dose-dependent inhibition of TNFα-induced luciferase expression, ranging from 46 ± 1.2% to 21 ± 3.8%, as compared to saline-treated, TNFα-stimulated controls (R2 = 0.98, Figure 3). When applying without saline dilution, LMWF5A was found to significantly inhibit reporter expression by 42 ± 6% across a total of 21 replicates (data not shown). This observation suggests that while LMWF5A does not significantly reduce p65 and RelB DNA binding, treatment effects the functional NF-κB-driven expression of luciferase.
PPARγ and AhR antagonists reduce LMWF5A drug potency
Based on bioinformatic analysis and historical findings, we hypothesize that the biologic mechanisms surrounding LMWF5A’s anti-inflammatory activity involve activation of immunoregulatory signaling pathways and ligand-activated TFs. Our group previously reported that LMWF5A-induced AhR activation is involved in the inhibition of IL-6 release from LPS-stimulated, macrophage-like THP-1 cells [24]. In support of this, the AhR immunoregulatory pathway was identified by gene enrichment analysis in the current study using our PBMC model. Furthermore, it has been established that sodium caprylate, one of the identified active components found in LMWF5A [22], can serve as an agonist for PPARγ [32]. Importantly, a large body of evidence demonstrates that these pathways can regulate the activity of pro-inflammatory TFs as well as drive differentiation of immune cell subsets [33–35]. Consequently, activation of these pathways may contribute to the ability of LMWF5A to suppress the release of pro-inflammatory cytokines from activated immune cells.
Relative potency (REP) was chosen as a robust metric for examination of our hypotheses. To establish overall drug activity, a relative potency assay was developed and validated, in adherence to USP 1032, 1033, and 1034 guidelines, based on the ability of LMWF5A to inhibit TNFα release in our established PBMC model. In brief, PBMC were stimulated overnight with LPS in the presence of 1.4-fold serial dilutions of LMWF5A and then TNFα release into the medium was determined by ELISA. Following this dilution scheme, LMWF5A-induced TNFα inhibition exhibits a log-linear dose response conducive to parallel-line REP calculation. Thus, separation in parallel responses can then be used to calculate the biologic activity of tested samples in relation to reference materials. Horizontal shifts in transformed TNFα inhibition dose response curves are observed when testing LMWF5A samples of known activity using this model (Supplementary Figure 1 A-C). Intermediate precision regression analysis, which assesses the accuracy and closeness of analytical samples, shows a correlation of 0.987 between expected and measured REP of reference LMWF5A samples manipulated into having differing activities (Supplementary Figure 1D). These data demonstrate that this method provides a highly accurate and precise bioassay for determining changes in LMWF5A drug potency.
The involvement of suspected pathways in the biologic activity of LMWF5A was then evaluated via this bioassay using specific antagonists for both PPARγ (GW9662, MilliporeSigma) and AhR (CH223191, MilliporeSigma). Exposure of PBMC to 0.5 µM and 0.05 µM final concentrations of GW9662 and CH223191 respectively, resulted in shifts towards reduced potency in the log-linear dose response for LMWF5A-induced TNFα inhibition (Figure 4A and 4B). Additionally, IL-1β release exhibited a similar log-linear dose response in this serial dilution range, and as a result, relative potency for IL-1β was calculated. As with TNFα, PPARγ and AhR antagonism resulted in linear shifts in IL-1β dose responses that reflect a loss of drug potency (Figure 4C and 4D). REP calculated for 3 independent experiments resulted in the reduction of mean REP to 0.74 ± 0.05 and 0.63 ± 0.14 for TNFα and IL-1β, respectively, when cells were treated with GW9662 in concert with LMWF5A (Table 4). Treatment with CH223191 also resulted in significant reduction in TNFα and IL-1β % inhibition potency to mean REP of 0.76 ± 0.09 and 0.72 ± 0.09, respectively. Collectively, these findings show that chemical antagonism of PPARγ and AhR transcription factors interfered with the overall anti-inflammatory activity of LMWF5A.