The physiological role of serine antiproteases has been extensively studied over the past two decades [9], mainly due to their contribution to homeostatic equilibrium through the control of proteases, inflammation, and infections [10], [11]. Together with other proteins, such as snake venom neurotoxins [12] and whey acidic protein [13], serine antiproteases share an evolutionary conserved canonical cysteine-rich four-disulfide core (FDC) domain, or the WAP domain, involved in protease inhibition [14]. Trappin-2 (9.9 kDa) (or pre-elafin) is a secreted and unglycosylated protein of 95-amino acids (aa) [15] that contains an N-terminal cementoin domain (38-aa) [16] and elafin (5.9 kDa), a C-terminal inhibitory WAP (57-aa) domain [14], [16]. Elafin is released from the N-terminus of Tr by proteolysis, arguably most efficiently by mast cell tryptase [17], [18]. Antiprotease activity and wound repair were the first described properties of Tr and E (Tr/E), similar to SLPI. Unlike ps20, SLPI along with Tr/E are functional neutrophil serine protease inhibitors [7], [11]. Inhibition of human neutrophil elastase (HNE) and proteinase 3 by the inhibitory loop on a WAP domain allows Tr/E to control tissues proteolysis associated with excessive inflammation in a neutrophil-rich environment. In turn, cross-linking between repeated hexapeptide motifs (GQDPVK) on the N-terminal portion of each Tr/E [15], [19] and extracellular matrix proteins arguably allows Tr/E to repair compromised tissue integrity [19], [20]. In addition, due to their cationic nature, but not exclusively [21], Tr/E were shown to possess antimicrobial activity against Gram-negative and Gram-positive bacteria [21], [22], [23] and certain fungal infections [24]. Worth mentioning is that similar to SLPI, antibacterial activity of Tr/E appeared to be independent of their antiprotease function [22]. Later, anti-inflammatory features of the antiproteases were also described, showing that Tr/E and SLPI were capable of reducing activation of NF-κB and AP-1 by altering IκB activation [25] and proteosomal degradation [26], respectively, in response to inflammatory and bacterial stimulation. More recent studies, however, also reported immunomodulatory properties of Tr/E. Indeed, depending on the environment, Tr/E can either dampen inflammation [20], [26] or promote immunostimulatory events and prime the immune system [27], [28]. Both Tr/E are found at mucosal surfaces [6], [29], in tissues [30], [31], [32], [33], [34] and multiple cell types, including genital ECs [6], [31] and regarded as alarm antiproteases, as they are mainly produced in response to pro-inflammatory stimuli like LPS [35], TNFα [36], and IL-1β [31], [37]. Interestingly, ECs from the female genital tract (FGT) produce Tr/E constitutively, with uterine cells capable of producing even greater amounts of Tr/E in response to a viral ligand, polyinosine-polycytidylic acid (polyIC) [6], indicating the significance of these molecules in controlling the local milieu in the FGT.

Viral double-stranded RNA (dsRNA) is a PAMP generated during the life cycle of most, if not all, viruses [4], [38]. Double-stranded RNA, including viral dsRNA and its synthetic mimic polyIC, are recognized by at least two families of PRRs: Toll-like receptors (TLRs), including TLR 3 [39], [40], and RNA helicases, namely retinoic acid inducible gene-I (RIG-I) [41], [42] and melanoma differentiation associated gene 5 (MDA5) [43]. Following recognition of dsRNA, activated PRRs initiate a series of signaling events, triggering phosphorylation, homodimerization and translocation into the nucleus of a set of transcription factors like interferon (IFN) regulatory factor 3 (IRF3), IRF7, NF-κB, and ATF2/c-Jun (c-Jun) [38], [41], [44], [45]. Inside the nucleus, these transcription factors either work independently or interact with each other [46] in triggering the transcription of antiviral and inflammatory gene products, such as type I IFNs and IL-8, IL-6, and TNFα. Specifically, IRF3 alone can directly bind to the IFN-stimulated response element in the promoter region of interferon-stimulated genes (ISGs) and activate a set of ISGs and their products in the absence of type I IFN production [47], [48]. Such antiviral cascade is devoid of excessive inflammatory responses and is induced when a low viral stimulation is detected. Alternatively, in response to a high viral load, IRF3 associates with NF-κB and c-Jun to trigger the production of IFNβ and the induction of ISGs in an IFN-dependent mode [45], [49], which also triggers robust inflammation, target cell recruitment, and/or tissue damage due to the production of pro-inflammatory cytokines and chemokines along with antiviral type I IFNs molecules [39], [48].

Treatment with polyIC has been shown to induce potent antiviral protection in vitro and in vivo, making polyIC an attractive candidate for microbicide or vaccine adjuvant trials against STIs [39], [40], [50]. However, in addition to antiviral activity, polyIC also triggers the release of pro-inflammatory mediators [39], [40], therefore potentially negating its beneficial effects in the FGT. While pro-inflammatory factors are important for immune cell recruitment and activation, if poorly controlled, they may also be detrimental in FGT, since acquisition and pathogenesis of common STIs are associated with immune activation and inflammation [51], [52], [53]. Hence, interventions leading to better control of inflammatory responses may prove to be more beneficial for increased antiviral protection and overall health in the FGT. Recently, new evidence has accumulated on the role of Tr/E in protection against viruses [6], [29]. Indeed, Tr/E have been associated with resistance to HIV mucosal transmission in commercial sex workers (CSWs) in Kenya [29]. Later, Ghosh et al. reported an anti-HIV feature of E in in vitro study [6]. Furthermore, Roghanian et al. documented that Tr expression increased Ad/LacZ viral clearance, as well as secondary immune responses, in a murine model in vivo [27].

This is the first study that comparatively assessed the presence and potential mechanisms of antiviral activity of each Tr/E. Here, we presented evidence showing two distinct, but likely complimentary antiviral properties of Tr/E: (i) direct antiviral activity exerted during the virus/cell interaction and targeting virus, but not cells; and (ii) indirect and cell-associated immunomodulatory activity, targeting polyIC-triggered cellular antiviral responses. The virus-mediated activity was observed in the absence of polyIC stimulation and in the presence of Tr in Ad/Tr-cells (presumably present in both Ad/Tr-sups and Ad/Tr-cells). Because only Tr was detected in polyIC-untreated Ad/Tr-cells, and because the expression of IFNβ was low and not different between untreated Ad/dl and Ad/Tr groups, we conclude that in the absence of polyIC, Tr alone was mediating antiviral activity in Ad/Tr-cells by acting either directly on virus or indirectly through cells. However, we failed to transfer this direct antiviral effect of Tr via Ad/Tr-sups, possibly due to the absence of an additive protective effect from augmented intracellular Tr (as would be expected in Ad/Tr-cells),or due to an inefficient delivery of Tr in supernatants and a “diluting" effect from other antiviral factors released in response to Ad/dl delivery (Fig. 2E, insert).

Additionally, the presence of both Tr/E, as in rTr preparation, was also protective against VSV-GFP challenge in the absence of polyIC treatment. Although no reports describing direct antiviral activity of Tr are published to date and no precise mechanisms of Tr/E direct antiviral effect have been identified, our observation with rTr is in line with Ghosh et al. findings, describing a direct antiviral effect of E as a mode of action against HIV [6]. Interestingly, rTr also appeared to have only virus- and not cell-mediated protective effects, similar to E mentioned earlier [6]. In contrast, a close WAP member SLPI, was shown to have only cell-mediated antiviral effects, at least against HIV [72] and herpes simplex virus (HSV) [73]. Examples of same-family members having differential antiviral mechanisms have also been described for other innate molecules, such as human defensins against HIV [74], [75] and HSV [76]. These reports suggest that molecules even from the same group, may possess their own, potentially different in potency and targets, exquisite antiviral activities and yet still uniquely contribute to overall mucosal protection against STIs.

Our results further suggest that the presence of both Tr/E might be required for each virus- (direct) and cell-associated (indirect) antiviral effect, as was evident from our data using Ad/Tr-cells, Ad/Tr-sups, and rTr, indicating that perhaps the most efficient antiviral protection depends on collaborative work of both Tr/E. Interestingly, when HIV-susceptible, but uninfected, CSWs were followed prospectively, those who remained HIV-negative had elevated levels of both Tr/E detected in CVLs [29]. Additionally, when characterizing the specificity of proteins secreted after Ad/Tr infection, our ELISA and WB results also showed that Ad/Tr-cells secreted both Tr/E independently in response to polyIC, while only Tr was detected without polyIC stimulation. Although Ghosh et al. also reported that primary uterine EC produced Tr/E in response to polyIC [6], the independent production of E was never demonstrated. Further, only Tr (13–16 kDa) was previously identified in supernatants from LPS-stimulated alveolar ECs [37]. On conjuncture, these results suggest that expression of each Tr/E could be a tissue/cell- or ligand-specific defense mechanism against an unknown protease that was potentially activated in response to a viral ligand. It might be important in the future to clarify whether primary genital EC from the FGT produce each Tr/E independently in response to polyIC, similarly to Ad/Tr-cells.

It is unclear why tested rE failed to show antiviral activity against VSV-GFP; it could be attributed, however, to a HAT-tag insertion at the N-terminus of rE. Indeed, all rTr, rE, and secreted Tr were equally functional against HNE (data not shown) and capable of inhibiting IL-8 production in response to polyIC (Figure 8). Yet, while rTr with a His-C-terminus tag exhibited antiviral activity, rE with a HAT-N-terminus tag did not. This observation suggests that blocking N-terminus, but not C-terminus, appears to be critical for antiviral activity of rE. An earlier study by McMichael et al. supports this argument, since they showed that the N-terminus of Tr had a better affinity for LPS than its C-terminus end [28]. Additionally, it is unclear why we observed increased levels of IL-8 with higher concentrations of rTr and rE. But one possible explanation could be that the proteins were initially delivered and left on cells in serum-free conditions, thus promoting the activation of pro-inflammatory events as was previously shown for Tr/E in response to LPS [28].

The second antiviral property of Tr/E observed in our study was an indirect cell-mediated immunomodulatory activity of Tr/E, targeting polyIC-induced antiviral cellular responses. In contrast to responses to bacterial or pro-inflammatory stimuli [25], [26], the scope and specific mechanism(s) of viral ligand-triggered immunomodulatory activity of Tr/E have never been fully investigated. Our data demonstrate that this indirect cell-associated activity is targeting viral recognition through modulation of RNA helicase expression as well as the induction of key inflammatory and antiviral innate signaling pathways and mediators.

Our results showed that polyIC-triggered antiviral cellular protection was significantly enhanced in the presence of Tr (in Ad/Tr-sups) and Tr/E (in Ad/Tr-cells and in rTr). We also showed that polyIC-mediated activation of IRF3 was further induced in Ad/Tr-cells, compared to controls, whereas IFNβ expression was dampened. Interestingly, human β defensin 3 [77] and cathelicidin LL37 [78] that were previously shown to have antiviral, including anti-HIV, activity [79], [80], were also reported to inhibit IFNβ production in vitro in response to LPS and polyIC, respectively. These observations further support our results and strengthen the earlier argument of Tr/E acting either directly against VSV-GFP or through cells and additional factors/pathways in Ad/Tr-cells. Furthermore, moderation of immune-inflammatory responses and thus curbing undesirable immune activation might be one of the protective mechanisms of innate antimicrobials at mucosal sites. Additionally, in searching for ISGs typically associated with antiviral protection and IRF3 activation [62], we found that expression of ISG15 was not significantly changed, unlike ISG56 being reduced and in agreement with IFNβ data. It is not entirely understood why such discordance was observed; however, it could be due to the fact that ISG15 was shown to be regulated by either IRF3 or IFNβ [44], [81], unlike ISG56 that was shown to be under the regulation of IFNβ or viruses [82], [83] and thus following IFNβ pattern of induction as shown in our study. The alternative explanation could be that these two genes follow a different temporal pattern of activation that was overlooked here.

This is the first report on the involvement of serine antiproteases, Tr/E in particular, in antiviral signaling pathways. As no prior data are available on the role of Tr/E in IFNβ and IRF3 induction, further and more detailed investigations might be required to explain why in Ad/Tr-cells IFNβ and ISG56 levels were reduced while IRF3 activation was increased. We hypothesize, however, that this phenomenon could be an attempt of Tr/E to control antiviral inflammatory events through RIG-I/MDA5 and NF-κB downregulation while increasing cellular protection through activation of IRF3 and/or alternative factors or pathways. Although most of the studies show ISG56 to be associated with upregulated IRF3 [62], our finding is in line with data from Li et al. showing that a knockdown of ISG56 was associated with increased IRF3 activation and inhibition of VSV-GFP replication [84] as a result of ISG56 mediating MITA-TBK1 interaction and subsequent downstream activation of IRF3. It is also possible that alternative factors/pathways, in addition to IRF3, regulate IFNβ and ISG56 expression and contribute to Tr/E-enhanced antiviral protection, which is also supported by our IRF3 siRNA data. Collectively, these data indicate that in the presence of Tr/E, antiviral protection is increased and that direct or indirect antiviral effect(s) of Tr/E depend, but not exclusively, on IRF3 and other factors, perhaps upstream of IRF3.

Inflammation is one of the leading factors predisposing to acquisition and disease progression of STIs in the FGT [52], [53], [85]. This notion is supported by the fact that “immune quiescence" and reduced immune activation are crucial for resistance against STIs [86], while dysregulated TLR expression and immune-inflammatory responses are detrimental [52], [53], [85]. Here, we showed that Tr/E individually or as a mixture, as well as in Ad/Tr-cells and as secreted/soluble proteins in Ad/Tr-sups, were capable of reducing IL-8, IL-6, and TNFα expression in response to polyIC. Moreover, in Ad/Tr-cells we also observed significantly reduced activation and transcriptional activity of NF-κB. The IL-8 inhibitory effect was not specific to human endometrial ECs, or to polyIC, indicating that similar effects could be observed at other mucosal surfaces and in response to different microbial ligands. We further showed that, compared to controls, mRNA and protein levels of RIG-I and MDA5 (mainly at a later time point), but not TLR3, were significantly diminished in response to polyIC and in presence of Tr/E. Immunomodulatory properties of both Tr/E demonstrated in models of pro-inflammatory and bacterial (LPS) stimulations were shown to depend on inhibition of NF-κB and AP-1 activation [25], [26], thus further supporting our NF-κB data. However, Tr/E inhibitory effect targeting antiviral immune responses, including viral sensing, has not been previously reported. Hence, modulation of expression of RIG-I, MDA5, and pro-inflammatory mediators shown here could represent novel antiviral functions of Tr/E, possibly even executed at different levels, namely receptors and transcription factors. That we observed differential pattern of RIG-I, MDA5, and TLR3 expression could indicate either different temporal kinetics of responses of these sensors, or that each RIG-I, MDA5, and TLR3 respond differentially to polyIC, being a mixture of variable lengths of dsRNA [43]. Further, the lack of polyIC-triggered TLR3 induction in HEC-1A compared to primary genital ECs, also likely reflects tissue or structure-dependent differences between the cells, suggesting that primary genital ECs may exhibit distinct results. Collectively, these observations suggest that Tr/E can alter innate viral recognition and mounting of antiviral immune-inflammatory responses.

The precise mechanism(s) of immunomodulatory effects of Tr/E, as both secreted and/or intracellularly expressed proteins, on viral sensors, cytokines, and IFNβ in response to polyIC treatment is still largely unknown. This is partly because the existence of the cognate receptor for Tr/E remains elusive, and it is equally unknown whether Tr/E require a receptor to function. We propose that reduced levels of IL-8, IL-6, TNFα and IFNβ in Ad/Tr cells in response to polyIC are likely a result of overall attenuation of RIG-I and MDA5 levels, as they are known to regulate the expression of pro-inflammatory and antiviral mediators [4], [41], [87] through activation of main signaling pathways, such as NF-κB that is downregulated in our study [41]. It remains to be elucidated, however, how Tr/E specifically inhibit RIG-I and MDA5 expression. It is plausible that Tr/E directly bind to polyIC, as was shown for binding of LL37 to polyIC [78], as well as Tr/E binding to LPS [28]. Such an interaction may alter binding/recognition of polyIC by its cognate receptors, including cell-surface scavenger receptor A or intracellular sensors RIG-I, MDA5, and TLR3, which in turn could explain our reduced expression levels of RIG-I and MDA5. Another possible site of inhibition by Tr/E could be downstream of receptors/viral sensors and involve Tr/E binding to DNA and competing for specific DNA binding sites with transcription factors including NF-κB, as was shown for SLPI as one of its anti-inflammatory mechanisms in response to LPS [88].

A noteworthy observation of this study is that while Tr from Ad/Tr-sups and rE were found functional against HNE and able to inhibit polyIC-induced IL-8 production, they did not show any antiviral activity, suggesting that antiprotease, anti-inflammatory, and antiviral activities of the tested proteins may not necessarily be co-dependent or predictive of each other; nonetheless, they can be complimentary. This observation is supported by earlier reports, showing both a protease non-inhibitory N-terminus and an inhibitory C-terminus of Tr exhibiting comparable antibacterial and antifungal functions [22], [24]. In contrast, Mulligan et al. showed that SLPI Gly(72) mutant, unlike other mutants tested in that study, lost its in vivo immunosuppressive activity against NF-κB activation and neutrophil recruitment in the lungs that appeared to be most closely related to SLPI's trypsin-inhibiting activity [89]. Although being an important property of both Tr/E, the inhibition of HNE activity is not considered a critical function for our studies, since epithelial cells do not make neutrophil elastase [90] and thus, the earlier discussed Tr/E-mediated changes are most unlikely attributed to antielastase activity of the proteins. The above observations indicate that perhaps additional structure-function studies might be warranted in the future to specifically address the cross-talk between antiprotease, anti-inflammatory, and antiviral properties of Tr/E and their specific roles in defense against viruses.