Human Hemoglobin Subunit Beta Functions as a Pleiotropic Regulator of RIG-I/MDA5-Mediated Antiviral Innate Immune Responses

Hemoglobin, the most important oxygen-carrying protein, is involved in the regulation of innate immune responses. We have previously reported that the porcine hemoglobin subunit beta (HB) exerts antiviral activity through regulation of type I interferon production. However, the antiviral activities and the underlying mechanisms of HBs originating from other animals have been poorly understood. Here, we identified human HB (hHB) as a pleiotropic regulator of the replication of RNA viruses through regulation of RIG-I/MDA5-mediated signaling pathways. hHB enhances RIG-I-mediated antiviral responses by promoting RIG-I ubiquitination depending on the hHB-induced reactive oxygen species (ROS), while it blocks MDA5-mediated antiviral signaling by suppressing the MDA5-dsRNA interaction. Our results contribute to an understanding of the crucial roles of hHB in the regulation of the RIG-I/MDA5-mediated signaling pathways. We also provide novel insight into the correlation of the intercellular redox state with the regulation of antiviral innate immunity.

pathogen-associated molecular patterns (PAMPs) (2). Viral nucleic acid is one of the well-characterized PAMPs. Depending on the genetic nature of viruses, retinoic acidinducible gene I (RIG-I)-like receptors (RLRs) function as the key viral RNA sensors and mediators of IFN production (3,4).
RIG-I and melanoma differentiation-associated gene 5 (MDA5) are major members of RLRs and contain a central DEXD/H box helicase domain, which is responsible for recognizing viral RNA, and two caspase recruitment domains (CARD) at their N-terminal regions (5). Upon sensing viral RNA, RIG-I and MDA5 undergo conformational alterations and interact with the adaptor mitochondrial antiviral signaling protein (MAVS, also called IPS-1, VISA, or Cardif) through CARD (6,7). Acting as a central adaptor, MAVS initiates downstream antiviral signaling by activating the downstream IKK␣, -␤, and -␥ and TBK1/IKKi kinases, resulting in the activation of NF-B and interferon regulatory factor 3 (IRF3)/IRF7 to transcriptionally induce type I IFNs (8)(9)(10).
RIG-I and MDA5 share high structural homologies and signaling features (11). However, they sense different species and types of viral RNAs (12). It is now well established that RIG-I primarily senses the 5=-triphosphate (5=PPP)-containing viral RNAs and some specific sequence motifs in the viral RNA, such as poly(U/UC) (13)(14)(15). In contrast to motifs recognized by RIG-I, the characteristics of viral PAMPs sensed by MDA5 remain elusive. It has been proposed that MDA5 can recognize long doublestranded RNA (dsRNA) as well as web-like RNA aggregates (16). As a countermeasure, it has been shown that RIG-I-and MDA5-mediated signaling pathways are differentially regulated (5). Some viral proteins exert different effects on the RIG-I-and MDA5mediated pathways. For instance, the paramyxovirus V protein regulates the MDA5-but not RIG-I-mediated signaling (17). In addition to viral proteins, numerous host molecules have been identified to regulate the RIG-I-mediated signaling. The ARF-like protein 16 (Arl16) and the anti-apoptotic protein A20 interact with RIG-I to inhibit antiviral responses (18,19); cylindromatosis (CYLD) and ubiquitin-specific peptidase 21 (USP21) remove K63-linked polyubiquitin chains to suppress RIG-I-mediated signaling (20,21), and ring-finger protein 125 (RNF125) can trigger the proteasome-mediated degradation of RIG-I (22). Several MDA5-associated host proteins have also been identified, such as ADP-ribosylation factor-like protein 5B (Arl5B) and dihydroacetone kinase (DAK) (23,24). However, the regulatory mechanisms of these two RLRs are insufficiently understood.
Hemoglobin is the main oxygen-carrying protein in vertebrates and many invertebrates, and in adult humans, it exists as a tetramer composed of two ␣-chains and two ␤-chains (25). It was previously thought that hemoglobin is expressed only by erythroid cells. However, this belief has been challenged by the recent findings that hemoglobin is expressed in a wide variety of nonerythrocytes, including hepatocytes, alveolar cells, neuronal/glial cells, and endometrial cells (26)(27)(28)(29). Hemoglobin exerts multiple functions and plays important roles in resistance to the invasion of pathogens and in the regulation of innate immunity (30,31). Peptides derived from hemoglobin have great potential as therapeutic drug candidates (32,33). However, the antiviral activities and the underlying mechanisms of hemoglobin have been poorly explored. Previously, we corroborated for the first time that porcine hemoglobin subunit beta (pHB) is able to suppress the growth of classical swine fever virus (CSFV) through the regulation of RIG-I-mediated type I IFN responses (34). However, the roles of the HBs of other species in innate immunity have not yet been determined. Due to the significant amino acid homology (84.4%) between human HB (hHB) and pHB, we speculated that they have functional homologies in regulating antiviral innate immunity.
In the present study, we identified hHB as a pleiotropic regulator of innate antiviral immunity through regulation of RIG-I/MDA5-mediated signaling pathways. We investigated the molecular mechanisms underlying hHB-induced differential regulation of RIG-I-and MDA5-mediated type I IFN responses in humans. Our results illustrate the importance of hHB in regulating antiviral responses and provide novel insights into functional differences in RIG-I-and MDA5-mediated antiviral innate immunity. short-poly(I·C)-induced activation of the IFN-␤ promoter in a dose-dependent manner and significantly enhanced the transcription of IFN-␤ mRNA (Fig. 2L). Moreover, the short-poly(I·C)-induced transcription of IFN-␤ was lower in MDA5 Ϫ/Ϫ hHB Ϫ/Ϫ cells than that in the MDA5 Ϫ/Ϫ cells (Fig. 2M). These results highlight the involvement of RIG-I in hHB-mediated regulation of type I IFNs. hHB inhibits the MDA5-mediated antiviral signaling. We next investigated the contribution of hHB in MDA5-mediated type I IFN signaling. We examined the activation of IFN-␤ promoter in p3ϫFlag-hHB and pMyc-MDA5 cotransfected HEK293T cells. The results demonstrated that overexpression of hHB significantly decreased MDA5induced IFN-␤ promoter activation in a dose-dependent manner (Fig. 3A). The suppressive effects of hHB on IFN-␤ promoter activation were also evident in the cells stimulated with long poly(I·C) (1.5 to 8 kb; an MDA5 agonist) (Fig. 3B). Overexpression of hHB also inhibited the transcription of IFN-␤, GBP1, and ISG56 induced by MDA5 or long poly(I·C) (Fig. 3C to E).
To further verify the contribution of hHB in MDA5-mediated signaling, we examined the impact of hHB knockout on IFN-␤ transcription. We observed that the activation of IFN-␤ promoter and the transcription of IFN-␤, GBP1, and ISG56 in response to long poly(I·C) or MDA5 stimulation were significantly higher in the hHB Ϫ/Ϫ cells than in the WT cells ( Fig. 3F to I). We also generated RIG-I-deficient HEK293T (RIG-I Ϫ/Ϫ ) cells and double-knockout cells deficient in both RIG-I and hHB (RIG-I Ϫ/Ϫ hHB Ϫ/Ϫ ) to exclude the possible disturbance of RIG-I (Fig. 3J). Overexpression of hHB in RIG-I Ϫ/Ϫ cells inhibited long-poly(I·C)-induced activation of the IFN-␤ promoter in a dose-dependent manner and significantly suppressed the transcription of IFN-␤ mRNA (Fig. 3K). Moreover, long-poly(I·C)-induced transcription of IFN-␤ was higher in RIG-I Ϫ/Ϫ hHB Ϫ/Ϫ cells than that in the RIG-I Ϫ/Ϫ cells (Fig. 3L). Based on these findings, it is plausible that hHB can distinctly regulate IFN-␤ production through MDA5-and RIG-I-mediated signaling pathways.
hHB regulates the replication of RNA viruses through RIG-I-and MDA5mediated signaling pathways. We further verified the contribution of hHB-mediated regulation of the RIG-I-and MDA5-mediated signaling to the replication of RNA viruses. We first investigated the relevance of hHB overexpression and the replication of RNA viruses in MDA5 Ϫ/Ϫ and RIG-I Ϫ/Ϫ cells. The results showed that overexpression of hHB led to resistance against SeV reproduction and inhibited the replication of SeV, VSV, and NDV in MDA5 Ϫ/Ϫ cells (Fig. 4A). Conversely, in RIG-I Ϫ/Ϫ cells, overexpression of hHB enhanced the replication of SeV, VSV, and NDV (Fig. 4B). Moreover, although hHB overexpression also increased the viral titer and RNA replication of EMCV in RIG-I Ϫ/Ϫ cells, it had little impact on the viral titer and RNA replication of EMCV in MDA5 Ϫ/Ϫ cells ( Fig. 4C and D).
In order to further assess the combinatorial impacts of RLR and hHB on viruses, we compared the replication levels of SeV, VSV, NDV, and EMCV between MDA5 Ϫ/Ϫ hHB Ϫ/Ϫ and MDA5 Ϫ/Ϫ cells and between RIG-I Ϫ/Ϫ hHB Ϫ/Ϫ and RIG-I Ϫ/Ϫ cells. The MDA5 Ϫ/Ϫ hHB Ϫ/Ϫ cells showed higher viral titers and RNA replication levels of SeV  Journal of Virology than MDA5 Ϫ/Ϫ cells, and the replication levels of VSV and NDV were also higher in MDA5 Ϫ/Ϫ hHB Ϫ/Ϫ cells than levels in MDA5 Ϫ/Ϫ cells (Fig. 4E). In contrast, deficiency of hHB in RIG-I Ϫ/Ϫ cells established resistance to SeV, VSV, and NDV (Fig. 4F). Additionally, deficiency of hHB suppressed the viral titer and RNA replication of EMCV in RIG-I Ϫ/Ϫ cells but exhibited little effect on the replication of EMCV in MDA5 Ϫ/Ϫ cells ( Fig. 4G and H). Taken together, these results indicate that hHB can differentially regulate the defense response of host cells to RNA viruses through the RIG-I-and MDA5-mediated antiviral signaling pathway. hHB has no effect on the expression of RIG-I or MDA5. After establishing the relationship between hHB and the MDA5/RIG-I pathways in regulating IFN-␤, we explored whether hHB affects the expression of RIG-I or MDA5. Overexpression of hHB failed to alter the exogenous protein expression of RIG-I and MDA5 at various doses ( Fig. 5A and B). In addition, the mRNA levels of RIG-I and MDA5 remained unchanged upon overexpression of hHB ( Fig. 5C and D). We were unable to observe a difference between the mRNA levels of RIG-I or MDA5 in hHB Ϫ/Ϫ and WT cells (Fig. 5E). Correspondingly, the endogenous expression of RIG-I and MDA5 remained unaffected in hHB-overexpressing cells (Fig. 5F). Moreover, there were no differences observed in the endogenous expression of RIG-I or MDA5 in hHB Ϫ/Ϫ cells and the WT cells (Fig. 5G).
hHB interferes with the MDA5-dsRNA interaction and enhances the ubiquitination of RIG-I. The results presented so far clearly articulate the involvement of RIG-I/MDA5 in mediating hHB-dependent regulation of the type I IFN pathway without affecting RIG-I/MDA5 protein expression. We next mechanistically investigated whether hHB affects the activation of these RLRs. Previous studies have demonstrated that sensing different types of viral RNA is required for the activation of MDA5 and RIG-I to initiate their signal transductions (38)(39)(40). RIG-I and MDA5 also recognize short poly(I·C) and long poly(I·C) as synthetic dsRNA analogues, respectively (35). Thus, we examined whether hHB affected the interaction between RIG-I or MDA5 and dsRNA using a shortor long-poly(I·C)-binding assay. We overexpressed exogenous hHB and RIG-I in HEK293T cells and examined the binding of the photobiotin-labeled short poly(I·C) with the exogenous RIG-I. The results showed that short poly(I·C) interacted with the exogenous RIG-I, which is independent of hHB (Fig. 6A). However, overexpression of hHB significantly inhibited the interaction of exogenous MDA5 with long poly(I·C) (Fig.  6B). Furthermore, exogenous MDA5 was also overexpressed in hHB Ϫ/Ϫ and WT cells, and a poly(I·C)-binding assay indicated that long poly(I·C) interacted with exogenous MDA5 more effectively in the hHB Ϫ/Ϫ cells than in the WT cells (Fig. 6C). These results indicate that hHB acts as a repressor of MDA5 activation by inhibiting the MDA5-dsRNA interaction. To further verify this possibility, we also used long poly(I·C) to pull down endogenous MDA5 in hHB Ϫ/Ϫ and WT cells. As expected, the binding of long poly(I·C) with the endogenous MDA5 was also stronger in hHB Ϫ/Ϫ cells than that in WT cells (Fig.  6D). Interestingly, when we determined the interaction between hHB and dsRNA, we found that hHB was precipitated with long poly(I·C) but not with short poly(I·C) (Fig. 6E). Due to the association of hHB with dsRNA and the involvement of hHB in MDA5-dsRNA interaction, we were interested to investigate if the interaction with the dsRNA occurs at the interface of hHB and MDA5. Coimmunoprecipitation (co-IP) analysis indicated no identifiable interaction of hHB with either MDA5 or RIG-I (Fig. 6F). These results underline the possibility that the binding of hHB to dsRNA may compete for the interaction of MDA5 with the dsRNA ligand and that this competition may result in a reduction of IFN-␤ induction.
Upon interaction with dsRNA ligands, RIG-I or MDA5 is ubiquitinated before recruitment to the mitochondrion-associated membrane and binding to MAVS (41). This suggested that the ubiquitination of RIG-I and MDA5 is crucial for the activation of RLR signaling. Thus, we overexpressed exogenous hHB, ubiquitin, and RIG-I or MDA5 in HEK293T cells to evaluate the ubiquitination of exogenous RIG-I and MDA5 upon hHB overexpression. Based on the disruption of the MDA5-dsRNA interaction by hHB, the ubiquitination of MDA5 was certainly suppressed by hHB in a dose-dependent manner (Fig. 6G). However, the ubiquitination of exogenous RIG-I was potentiated by increasing hHB protein expression (Fig. 6H). To further confirm the effect of hHB on the ubiquitination of RIG-I, we compared the levels of endogenous ubiquitination of RIG-I in hHB Ϫ/Ϫ and WT cells after SeV infection. The results showed that the level of endogenous ubiquitination of RIG-I was lower in the hHB Ϫ/Ϫ cells than that in the WT cells  ( Fig. 6I). It has been shown that RIG-I has different ubiquitination forms and that K63-linked ubiquitination of RIG-I is positively required for RIG-I activation, whereas K48-linked ubiquitination will result in the destabilization of RIG-I (22,42). To verify if hHB promotes RIG-I activation through potentiating RIG-I ubiquitination, we constructed two ubiquitin mutants in which all lysine residues were replaced with arginine except K48 or K63 (HA-K48Ub or HA-K63Ub, respectively; HA is hemagglutinin and Ub is ubiquitin) to further investigate the polyubiquitination of RIG-I regulated by hHB. The results showed that hHB enhanced K63-linked but not K48-linked RIG-I ubiquitination (Fig. 6J). Moreover, the total ubiquitination of RIG-I and the K63-linked RIG-I ubiquitination were lower in hHB Ϫ/Ϫ cells than levels in WT cells (Fig. 6K). In addition, to uncover whether the action of hHB is characteristic or noncharacteristic, we also examined whether the alpha subunit of hemoglobin (hHA) has similar effects on the activation of RIG-I or MDA5. The results showed that both the RIG-I-short poly(I·C) and the MDA5long poly(I·C) interactions were independent of hHA (Fig. 6L). Furthermore, the ubiquitination of MDA5 remained unchanged by hHA, but the ubiquitination of RIG-I was enhanced by hHA in a dose-dependent manner (Fig. 6M). Moreover, hHA could promote the short-poly(I·C)-induced transcription of IFN-␤ but not the long-poly(I·C)induced transcription of IFN-␤ (Fig. 6N). These data imply that the inhibition of the MDA5-dsRNA interaction is a characteristic of hHB.

ROS is required for the hHB-induced upregulation of the RIG-I signaling pathway.
Considering that hemoglobin is able to regulate the production of reactive oxygen species (ROS), which is a key factor for the host cell to trigger efficient activation of immunity (43)(44)(45), we evaluated if ROS was involved in the hHB-mediated regulation of the RIG-I or MDA5 signaling pathway. So, we first explored the links among hHB, ROS, and virus infections. hHB was overexpressed in HEK293T cells, and ROS production was monitored using the oxidant-sensitive fluorescent detection probe dichlorodihydrofluorescein diacetate (DCFH-DA). The results demonstrated that hHB increased intracellular ROS accumulation in a dose-dependent manner (Fig. 7A). In addition, intracellular ROS accumulation was also upregulated by SeV in a dose-dependent manner (Fig.  7B). Thus, we also monitored the effect of hHB on ROS accumulation in SeV-infected cells. The overexpression of hHB promoted ROS production in a dose-dependent manner at an early time of SeV infection (Fig. 7C). However, intracellular ROS accumulation was far more impacted by robust viral replication than by hHB at a later stage of SeV infection. Thus, the hHB-overexpressed cells showed lower intracellular ROS accu- mulation at a later time of SeV infection as the replication of SeV was suppressed (Fig.  7D). Consistently, SeV induced lower ROS production in hHB Ϫ/Ϫ cells than in WT cells at the earlier time, but the hHB Ϫ/Ϫ cells accumulated more intracellular ROS at the later time of SeV infection (Fig. 7E).
Next, we investigated if ROS is required for hHB-mediated regulation of the MDA5 signaling pathway. As shown in Fig. 8A and B, MDA5-or RIG-induced IFN-␤ transcription was significantly reduced by tempol (an ROS inhibitor). When intracellular ROS was suppressed by tempol, hHB still inhibited long-poly(I·C)-induced activation of IFN-␤ promoter (Fig. 8C). Accordingly, tempol treatment could not absolutely counteract the inhibition of the MDA5-or long-poly(I·C)-induced transcription of IFN-␤ by hHB overexpression (Fig. 8D). To exclude the unspecific effects of tempol and the disturbance of the RIG-I-mediated signaling pathway, two other ROS inhibitors, diphenyleneiodonium chloride (DPI) and N-acetyl-L-cysteine (NAC), were also tested in the RIG-I Ϫ/Ϫ cells. After DPI or NAC treatment, hHB-mediated inhibition of the MDA5-induced or long-poly(I·C)induced transcription of IFN-␤ was still observed in RIG-I Ϫ/Ϫ cells (Fig. 8E). However, hHB-induced upregulation of the IFN-␤ promoter activation in response to short poly(I·C) was inhibited by tempol (Fig. 8F). Tempol also suppressed hHB-induced upregulation of RIG-I-induced or short-poly(I·C)-induced transcription of IFN-␤ (Fig. 8G). Moreover, DPI or NAC also obviously counteracted the hHB-mediated facilitation of RIG-I-induced or short-poly(I·C)-induced IFN-␤ transcription in the MDA5 Ϫ/Ϫ cells (Fig. 8H). These findings indicate that the upregulation of the RIG-I signaling by hHB is related to the hHB-induced ROS, whereas hHB can regulate MDA5 signaling through a pathway which does not entirely depend on the hHB-induced ROS.
We have previously shown that the MDA5-mediated signaling pathway is disrupted by hHB but not by hHA. However, the RIG-I-but not MDA5-mediated signaling pathway is certainly promoted by both hHB and hHA. Thus, we also verified the role of ROS in the hHA-mediated regulation of RIG-I-mediated signaling as a reference. Our results showed that hHA could also promote the intracellular accumulation of ROS (Fig. 8I). Moreover, tempol or DPI treatment counteracted the hHA-mediated enhancement of short-poly(I·C)-induced transcription of IFN-␤ (Fig. 8J). These results implied that the ability of hHB to promote the activation of RIG-I by inducing ROS might be a general characteristic of hemoglobin. Furthermore, we evaluated the effects of hHB on the ubiquitination of RIG-I upon tempol treatment. Suppression of ROS by tempol significantly inhibited the total ubiquitination and the K63-linked ubiquitination of RIG-I (Fig. 8K). Consistent with previous findings, when ROS was suppressed, the ubiquitination of RIG-I was no longer enhanced by increasing hHB (Fig. 8L). To further verify the role of ROS in the hHBmediated regulation of RIG-I activation, we investigated the effects of hHB-induced ROS on K63-linked RIG-I ubiquitination. The results showed that K63-linked ubiquitination of RIG-I was no longer enhanced by hHB when ROS accumulation was inhibited (Fig. 8M).

DISCUSSION
Generally, RIG-I and MDA5, the cytoplasmic RNA helicase proteins, are the main sensors of RNA viruses in triggering type I IFNs in eukaryotes (3,4). Exploring the molecular events of the RIG-I/MDA5 signaling pathway is critical for understanding complex innate immune responses against RNA viruses. In this study, we identified hHB as a novel innate immune regulator of the RIG-I/MDA5-mediated antiviral signaling pathways, which further advances our understanding of the regulatory mechanisms involved in RLR-mediated signaling pathways.
The activation of RIG-I or MDA5 is a complex regulatory process including viral RNA binding, structural rearrangement, dephosphorylation, ubiquitination, and binding to MAVS to activate downstream antiviral signaling (46)(47)(48). Despite functional overlap between the RIG-I and MDA5 pathways, our results showed that hHB could differentially regulate these signaling pathways. RIG-I and MDA5 recognize differential viral RNAs (12). In our study, hHB significantly inhibited the replication of SeV, VSV, and NDV, which are mainly sensed by RIG-I, but enhanced the growth of EMCV, which almost only activates MDA5-mediated signaling ( Fig. 1 and 4). Moreover, we showed that hHB promoted RIG-I signaling and remarkably inhibited MDA5-mediated type I IFN produc- hHB Regulates RIG-I/MDA5-Mediated Signaling Journal of Virology tion ( Fig. 2 and 3). Therefore, it is plausible that hHB differentially regulates RIG-I and MDA5 activation upstream of the MAVS-mediated signaling. RIG-I and MDA5 share similar structural frameworks implicated in viral dsRNA recognition and detection of short and long poly(I·C) as the synthetic dsRNA analogues, respectively (35,49). Although RIG-I recognition of viral RNA has been mostly clarified, how MDA5 recognizes viral RNA is yet to be determined. Our results demonstrated that hHB had no obvious influence on the binding of short poly(I·C) to RIG-I, but hHB acted as a direct repressor of MDA5 by interfering with the interaction between MDA5 and long poly(I·C) (Fig. 6B and C). In addition, hHB could bind to the long poly(I·C), but it failed to interact with short poly(I·C) or MDA5 and RIG-I (Fig. 6D to F). These data imply that hHB-dsRNA may compete with the MDA5-dsRNA interaction and thus negatively regulates the MDA5mediated IFN pathway. Moreover, ubiquitination plays a critical role in the regulation of RIG-I and MDA5 activation (3). The E3 ubiquitin ligases TRIM25 and TRIM65 catalyze the K63-linked ubiquitination of RIG-I and MDA5, respectively, and thus positively regulate the RIG-I-and MDA5-mediated signaling pathways, respectively (42,50). In the present study, we found that hHB promoted the K63-linked ubiquitination of RIG-I whereas it inhibited MDA5 ubiquitination ( Fig. 6G to I). Interestingly, our results showed that hHA had a similar effect on the activation of RIG-I and that it promoted RIG-I ubiquitination after RNA virus infection. These findings suggest that the action of hHB on the MDA5-dsRNA interaction is characteristic, but the ability of hHB to facilitate the activation of RIG-I may be due to the general characteristics of hemoglobin. Hemoglobin is the main respiratory protein in vertebrates and many invertebrates. It exerts multiple functions and plays an important role in resistance to pathogen invasion (31). In addition to functioning as a major host respiratory protein, hemoglobin can also be specifically activated by pathogens to produce ROS to constitute a part of the host defense strategy (43,51,52). For example, human hemoglobin significantly enhances ROS production under microbial protease stimulation but not host protease stimulation (53). ROS plays a key role in immunity and pathogen killing (54)(55)(56). The host respiratory proteins directly exploit the invasion of microbes to produce ROS, resulting in localized cytotoxicity to rapidly kill the neighboring pathogens (53,57). Recently, the association of ROS with RLR signaling has been reported. The host cell requires ROS to efficiently trigger RIG-I-mediated IRF3 activation and IFN-␤ expression (58). This indicates that ROS may provide a mediator for hHB to be involved the regulation of RIG-I signaling. Here, we demonstrated that hHB increases the ROS level in cells and that the antioxidant inhibitors, including tempol, DPI, and NAC, could counteract the hHB-mediated upregulation of the RIG-I-mediated signaling pathway but could not eliminate the effects of hHB on MDA5 signaling (Fig. 8). Moreover, inhibition of ROS by tempol suppressed hHB-mediated facilitation of RIG-I ubiquitination, especially the K63-linked ubiquitination of RIG-I. These findings certify that ROS is required for hHB-mediated regulation of RIG-I ubiquitination, which indirectly promotes the activation of RIG-I signaling. Interestingly, our results showed that hHA also promotes the RIG-I-mediated signaling pathway by enhancing the intercellular ROS accumulation. These findings suggest that the action of hHB on the MDA5-dsRNA interaction is canonical but that the ability of hHB to facilitate the activation of RIG-I might be a general attribute of hemoglobin. The innate immune system has evolved various strategies to prevent harmful overproduction of type I IFNs during viral infection. Thus, several host molecules are capable of regulating type I IFN production via multiple pathways, including the RIG-I/MDA5 signaling pathway. For example, DAK is a specific repressor of MDA5-mediated signaling, and the deubiquitinating activity of A20 inhibits RIG-I-mediated signaling (19,23). It is also worth noting that several host factors adopted multiple ways to regulate RIG-I/MDA5 signaling. For example, IFN-␤ levels are increased following stimulation with activators of RIG-I signaling in protein kinase R (PKR)-null cells, and the absence of PKR severely impairs MDA5-mediated IFN induction (59). In this study, we determined that hHB was a pleiotropic regulator of the RIG-I/MDA5-mediated signaling pathway. Moreover, hHB could affect RIG-I/MDA5 signaling in a direct or indirect manner. hHB directly inhibits binding of MDA5 to dsRNA and negatively regulates MDA5-mediated IFN production (Fig. 9). On the other hand, hHB is involved in the regulation of cellular oxidative stress to enhance RIG-I ubiquitination, which indirectly promotes RIG-Imediated IFN production (Fig. 9). These findings imply that hHB contributes to protection mechanisms needed for controlling the RLR signaling pathway.
Currently, many regulators have been identified to have direct effects on a single point or a single pathway of antiviral innate immunity (60)(61)(62). For these regulators, the direct effects are appreciated and emphatically studied. However, their indirect impact on the intercellular microenvironment, such as redox state, pH, and ion leakage, are relatively ignored. In the present study, although hHB can directly inhibit type I IFN production by interfering with the MDA5-mediated signaling pathway, an hHB-induced change in the intercellular redox state will concurrently impede this inhibition by promoting the RIG-I-mediated signaling pathway. This reveals the importance of the intercellular microenvironment in the regulation of antiviral innate immunity and suggests why the effects of some regulators are always fluctuating in response to the nature of the stimuli. The indirect effects resulting from hHB's influence on the cellular microenvironment suggest that hHB-mediated innate immune regulation may be dependent on the cellular state and stimulus types. Therefore, future work is required to further understand the regulatory mechanisms of antiviral innate immunity and to improve the effectiveness of some regulators.
In summary, we identified hHB as a novel innate immune regulator of RNA viruses through multifunctional and pleiotropic regulation of the RIG-I/MDA5 signaling pathways. On one hand, hHB promoted the RIG-I-mediated signaling pathway by enhancing RIG-I ubiquitination. On the other hand, hHB remarkably inhibited MDA5-mediated type I IFN production by interfering with the MDA5-dsRNA interaction. We mechanistically illustrated the crucial roles of hHB in regulating and protecting antiviral innate immunity. Our findings also highlight the importance of the intercellular microenvi-