TREM2 suppresses the proinflammatory response to facilitate PRRSV infection via PI3K/NF-κB signaling

Triggering receptor expressed on myeloid cells 2 (TREM2) serves as an anti-inflammatory receptor, negatively regulating the innate immune response. TREM2 is mainly expressed on dendritic cells and macrophages, the target cells of porcine reproductive and respiratory syndrome virus (PRRSV). Thus, we investigated the potential role of TREM2 in PRRSV infection in porcine alveolar macrophages (PAMs). We found that there was an increased expression of TREM2 upon PRRSV infection in vitro. TREM2 silencing restrained the replication of PRRSV, whereas TREM2 overexpression facilitated viral replication. The cytoplasmic tail domain of TREM2 interacted with PRRSV Nsp2 to promote infection. TREM2 downregulation led to early activation of PI3K/NF-κB signaling, thus reinforcing the expression of proinflammatory cytokines and type I interferons. Due to the enhanced cytokine expression, a disintegrin and metalloproteinase 17 was activated to promote the cleavage of membrane CD163, which resulted in suppression of infection. Furthermore, exogenous soluble TREM2 (sTREM2)-mediated inhibition of PRRSV attachment might be attributed to its competitive binding to viral envelope proteins. In pigs, following PRRSV challenge in vivo, the expression of TREM2 in lungs and lymph nodes as well as the production of sTREM2 were significantly increased. These novel findings indicate that TREM2 plays a role in regulating PRRSV replication via the inflammatory response. Therefore, our work describes a novel antiviral mechanism against PRRSV infection and suggests that targeting TREM2 could be a new approach in the control of the PRRSV infection.

Introduction migration of microglial cells [40]. TREM-2 expression is significantly increased in alveolar monocytes/macrophages of influenza virus-infected pigs [41]. Adeno-associated virus infection induces microglial TREM2 overexpression in the hippocampus of wild-type mice [42]. DAP12 knockdown also intensifies PRRSV-induced proinflammatory cytokine production and inhibits viral replication [43]. However, little is known about TREM2 in the regulation of PRRSV infection.
In the current study, we determined the role of TREM2 in PRRSV infection and elucidated the underlying mechanisms of TREM2-mediated PRRSV infection. We demonstrate that the expression of TREM2 has a positive correlation with PRRSV infection in vivo and in vitro. TREM2 and its downstream signaling negatively regulated the host proinflammatory response and enhanced the expression of CD163, which contributed to PRRSV infection. The cytoplasmic tail domain of TREM2 was responsible for promoting the replication of PRRSV via binding to PRRSV Nsp2. Moreover, exogenous sTREM2 could restrain PRRSV attachment to the cell surface through interaction with PRRSV envelope proteins.

TREM2 facilitates PRRSV infection
PAMs are the primary targets of PRRSV [44]. TREM2 is widely expressed on macrophages and dendritic cells and serves as an immunoregulatory receptor [45]. Accordingly, we investigated the role of TREM2 in PRRSV infection. First, the kinetics of TREM2 expression in PAMs were analyzed by western blot at various hours post-infection (hpi) or multiplicities of infection (MOIs) following PRRSV infection. There were increased levels of TREM2 mRNA and protein in PAMs, concomitant with PRRSV replication (Fig 1A, 1B and S1A and S1B Fig). Subsequently, three siRNAs were designed for reducing the expression of TREM2 (Fig 1C). TREM2 knockdown suppressed PRRSV replication as shown by decreased levels of viral nucleocapsid (N) protein ( Fig 1D). Similarly, viral ORF7 mRNA was also markedly reduced at 12, 24, and 36 hpi following TREM2 knockdown (Fig 1E). Silencing of TREM2 also inhibited PRRSV replication at different MOIs ( Fig 1F). When virus titers were measured at different hours post PRRSV infection, the results showed that TREM2 silencing induced lower virulence at 12, 24, and 36 hpi compared to controls ( Fig 1G). Immunofluorescence analysis also demonstrated a significant inhibition of viral replication in PAMs with TREM2 knockdown ( Fig  1H). To identify whether overexpression of TREM2 had the opposite effect on PRRSV infection, TREM2 was overexpressed in PAMs and Marc-145 cells. As expected, TREM2 overexpression facilitated the replication of PRRSV and the expression of PRRSV N protein both in PAMs (Fig 1I and 1J) and Marc-145 cells (Fig 1L and 1M). Compared with the control vector, high expression of TREM2 significantly enhanced viral yields in both PAMs ( Fig 1K) and Marc-145 cells (Fig 1N). Taken together, TREM2 facilitates the replication of PRRSV.

The cytoplasmic tail domain of TREM2 upregulates PRRSV replication through interaction with PRRSV Nsp2
To map the domain(s) of TREM2 associated with PRRSV replication, Marc-145 cells were transfected with Myc-tagged full-length or truncated TREM2 and then infected with PRRSV (Fig 2A). We found that TREM2 fragments lacking the cytoplasmic tail region did not influence the expression of PRRSV N, suggesting that the TREM2 cytoplasmic tail domain is required for regulation of PRRSV (Fig 2B and 2C). Next, we screened the non-structural proteins (NSPs) of PRRSV responsible for the upregulation of TREM2. Ten mCherry-tagged NSPs were transfected into PAMs, respectively. It was found that the level of TREM2 was increased in PAMs that were expressing mCherry-tagged Nsp2 (Fig 2D). Myc-tagged TREM2 and ten mCherry-tagged NSPs were then co-transfected into HEK293T cells, respectively. We observed that Nsp2, Nsp3, Nsp5, and Nsp7 co-localized with TREM2 ( Fig 2E). Likewise, we performed co-IP assays to screen for TREM2-interacting viral NSPs. We found that exogenous TREM2 co-precipitated with Nsp2 in HEK293T cells (Fig 2F), while no interaction between TREM2 and Nsp3, Nsp5, or Nsp7 was detected (S2B Fig). To further investigate the positive interaction between TREM2 and Nsp2, Myc-tagged TREM2 was overexpressed in Marc-145 cells, and then cells were infected with PRRSV. As expected, a Myc antibody immunoprecipitated PRRSV Nsp2 (Fig 2G). Moreover, PAMs were mock-infected or infected with PRRSV, and a TREM2 antibody immunoprecipitated PRRSV Nsp2 in PRRSV-infected cells (S2A Fig). To map the domain(s) required for the TREM2/Nsp2 interaction, we co-transfected mCherrytagged Nsp2 and Myc-tagged full-length or truncated TREM2 into HEK293T cells. We found that TREM2 constructs comprising the cytoplasmic tail region co-precipitated with Nsp2, while TREM2 fragments lacking the cytoplasmic tail region had no interaction with Nsp2, indicating that the cytoplasmic tail domain might contain the binding region of viral Nsp2 (Fig 2H). To explore whether the TREM2/Nsp2 interaction affects TREM2 expression, we constructed two mCherry-tagged truncated Nsp2 (Nsp2-N and Nsp2-M) [46] and transfected them into PAMs (S2C Fig). We observed transcriptional upregulation of TREM2 in Nsp2-transfected cells and Nsp2-N-transfected cells, but there was no significant difference in TREM2 expression between Nsp2-M-transfected cells and control cells (S2D Fig). Furthermore, we co-transfected Myc-tagged TREM2 and mCherry-tagged Nsp2 or truncated Nsp2 into HEK293T cells and Marc-145 cells. The results revealed that the truncated Nsp2-N retained the ability to interact with TREM2, while the truncated Nsp2-M did not (S2E and S2F Fig), indicating that Nsp2-N (aa1-405) is responsible for the interaction with TREM2 and therefore affects TREM2 levels. Taken together, the cytoplasmic tail domain of TREM2 is required and sufficient for facilitating PRRSV replication via binding to PRRSV Nsp2.  (B) at different MOIs for 24 h,  protein levels of TREM2, PRRSV N, and GAPDH are shown, as detected by western blot using the indicated antibodies. GAPDH is shown as an internal control. (C to H) TREM2 knockdown restrains PRRSV infection. (C) PAMs were transfected with scrambled siRNA (siNC) or three TREM2 siRNAs (siTREM2) for 24 h, gene expression of TREM2 is shown using qRT-PCR analysis. (D) PAMs were transfected with siNC or three siTREM2 for 24 h and then infected with PRRSV (MOI = 1) for 24 h, protein levels of TREM2 and PRRSV N are shown, as detected by western blot. (E and F) PAMs were treated with siNC or siTREM2-1 and additionally infected with PRRSV (MOI = 1) for the indicated time points (E) or at different MOIs (F), the expression of viral ORF7 (PRRSV N) is shown, as detected by qRT-PCR. (G) Cell supernatants from different periods post-infection were collected for TCID 50 . (H) PAMs were treated as described in E, viral load at 24 h post-infection is shown using immunofluorescence analysis (Bar, 200 μm). (I-N) TREM2 overexpression facilitates the replication of PRRSV. PAMs or Marc-145 cells were transfected with pcDNA3.1-control (vector) or pcDNA3.1-TREM2 for 24 h, and then infected with PRRSV (MOI = 1) for another 24 h. PAMs were collected and the transcription levels of TREM2 and PRRSV N are shown using qRT-PCR (I); protein levels are shown using western blot analysis (J). (K) Viral production in PAMs was measured and is shown as TCID 50 . (L-N) Marc-145 cells were collected; protein levels of TREM2 and PRRSV N are shown (L), transcription levels are shown (M) and the TCID 50 is shown from cell supernatants (N) . Data are representative of the results of three independent experiments (mean ± SE). Significant differences are indicated as follows: � (P < .05), �� (P < .01) and ��� (P < .001).

PLOS PATHOGENS
TREM2 regulates PRRSV via inflammation inhibitor R406 and PI3K inhibitor Wortmannin were used to inhibit the phosphorylation of Syk and Akt, respectively. R406 significantly decreased the phosphorylation of Syk and the combination between phosphorylated Syk and DAP12 during PRRSV infection ( Fig 3C). Interestingly, in TREM2 knockdown cells treated with inhibitors, the PRRSV N protein was increasingly rescued compared to those not treated with an inhibitor (Fig 3D and 3E). These data illustrate that TREM2 silencing provokes the recruitment of Syk and the activation of the PI3K and ERK1/2 pathways, which are associated with PRRSV replication.

TREM2 knockdown leads to early activation of the TLR4-NF-κB pathway during PRRSV infection
Previous studies have shown that the expression of TREM2 on the cell surface is regulated by TLR stimulation and TREM2 is downregulated upon LPS challenge in peritoneal macrophages [47]. TREM2 plays a role in fine-tuning innate responses by dampening signals induced by TLRs [29,30]. We speculated TREM2 controlled the immune response against PRRSV infection through TLRs and their signaling. To address this hypothesis, we used TLR agonists Zymosan, Pam 3 CSK 4 , LPS, and poly (I:C) to stimulate PAMs and then detected the expression of TREM2. As shown in Fig 4A, Zymosan, Pam 3 CSK 4 , and poly (I:C) promoted the transcription of TREM2. Surprisingly, TREM2 expression gradually decreased when PAMs were stimulated with LPS, a specific TLR4 agonist, indicating that TLR4 is involved in TREM2 signaling. Next, we detected TLR4 expression on the cell membrane using flow cytometry when TREM2 was knocked down. Silencing of TREM2 boosted the expression of membrane-associated TLR4 in virus-infected or mock-infected cells ( Fig 4B). Meanwhile, TREM2 knockdown enhanced the mRNA expression of TLR4 at the early stage of infection ( Fig 4C). Furthermore, we determined whether TREM2 activates signaling downstream of the TLR4 pathway, such as NF-κB signaling. We conducted a TREM2 interference assay and found that TREM2 knockdown increased the transcription of TLR4, MyD88, and NF-κB ( Fig 4D). Additionally, NF-κB mRNA was induced as early as 6 hpi in TREM2-silenced PAMs and maintained a high transcription level throughout the early stage of infection ( Fig 4E). In contrast, TREM2 overexpression decreased the transcription of TLR4 and NF-κB induced by PRRSV (Fig 4F). Western blot analysis revealed that phosphorylation of IκBα and p65 occurred much earlier (0 hpi) in TREM2-silenced PAMs compared to control PAMs. Moreover, the phosphorylation of IκBα and p65 was clearly increased and reached a peak at 12 hpi in TREM2-silenced PAMs ( Fig  4G). Immunofluorescence analysis of the nuclear translocation of p65, which represents the activation of NF-κB, was conducted. Compared with the negative control, translocation of p65 (green) into the nucleus of infected cells occurred much earlier following TREM2 silencing and reached a maximum (~80%) at 24 hpi (Fig 4H and 4I). Accordingly, we performed a cell 145 cells. After overexpression for 24 h, cells were infected with PRRSV (MOI = 1) for 24 h. The transcription levels (B) and protein levels (C) of PRRSV N are shown, as analyzed by qRT-PCR and western blot, respectively. Asterisks mark the expressed Myc-fusion proteins of full-length or other truncated TREM2. GAPDH is shown as an internal control. (D) mCherry control vector and ten mCherry-tagged viral NSPs were transfected into PAMs for 36 h, respectively. The transcription of TREM2 is shown, as measured by qRT-PCR. (E) Myc-tagged TREM2 and the different mCherry-tagged NSPs were co-transfected into HEK293T for 24 h. Immunofluorescence analysis of the co-localization between TREM2 (green) and NSPs (red) is shown. (Bar, 5 μm). (F) Myc-tagged TREM2 and mCherry-tagged Nsp2 were separately transfected or co-transfected into HEK293T for 36 h. Cell lysates were immunoprecipitated with Myc antibody, and the immunoblots with Myc and mCherry antibodies are shown. GAPDH is shown as an internal control. (G) Myc-tagged TREM2 was transfected into Marc-145 cells for 24 h and infected with PRRSV (MOI = 1) for 24 h. TREM2 was immunoprecipitated with Myc antibody and immunoblots with PRRSV Nsp2 and Myc antibodies are shown. GAPDH is shown as an internal control. (H) mCherry-tagged Nsp2 was co-transfected with Myc-tagged full-length or other truncated TREM2 in HEK293T. IP assays with anti-Myc antibody were performed to determine their interaction; the Nsp2-mCherry immunoblot with anti-mCherry antibody is shown. GAPDH is shown as an internal control. Asterisks mark the expressed Myc-fusion proteins of full-length or other truncated TREM2. Data are representative of the results of three independent experiments (mean ± SE). Significant differences are indicated as follows: � (P < .05), �� (P < .01) and ��� (P < .001).  fractionation assay and found that nuclear accumulation of endogenous p65, measured following infection, occurred at earlier time points and to a greater magnitude in TREM2-silenced PAMs, as compared with controls ( Fig 4J). A rescue experiment was implemented using BAY11-7082, an NF-κB inhibitor, which partially rescued the replication of PRRSV following TREM2 knockdown treatment (Fig 4K). Taken together, TREM2 knockdown dramatically accelerates the activation of NF-κB induced by PRRSV.

TREM2 knockdown accelerates the proinflammatory response and increases type I interferon production
TREM2 is involved in regulation of the virus-triggered inflammatory response [38]. To summarize the above findings, NF-κB signaling was activated at early time points following RNA interference of TREM2, prior to PRRSV-induced NF-κB activation. Accordingly, we investigated whether TREM2 regulates the production of PRRSV-induced inflammatory cytokines and type I interferons in PAMs. As shown in Fig 5A, PAMs with a TREM2 knockdown produced more (about 3~5 fold) inflammatory cytokines (IL-1β, IL-6, IL-8, and TNF-α) and type I interferons (IFN-α and IFN-β) than those with the negative control. The protein levels of IL-1β, IL-4, IL-10, and TNF-α were significantly increased in TREM2-silenced PAMs ( Fig 5B). Next, after PAMs were infected with PRRSV following TREM2 knockdown, we detected the transcription of proinflammatory cytokines at the early stage of infection. Proinflammatory cytokines and type I interferons were clearly induced as early as 3 hpi and were significantly augmented from 6-12 hpi in TREM2-silenced PAMs ( Fig 5C). TREM2 knockdown also boosted PRRSV-induced cytokine (IL-1β, TNF-α, IFN-β) transcription at different MOIs ( Fig  5D). These results illustrate that TREM2 knockdown potentiates the transcription of proinflammatory cytokines and type I interferons at the early stage of PRRSV infection.

TREM2 increases the expression of CD163 and decreases the cleavage of membrane CD163 mediated by ADAM17
CD163 is a key factor in the initiation of PRRSV infection. It is regulated by many proinflammatory cytokines and is cleaved by ADAM17 [18,48]. As previously covered, TREM2 silencing resulted in an early proinflammatory cytokine and type I interferon response. We investigated whether TREM2 regulates the expression of ADAM17 and CD163 via proinflammatory cytokines. Therefore, we used siRNA to deplete cells of TREM2 and detected the expression of ADAM17 and CD163. As expected, TREM2 knockdown clearly enhanced the protein level of ADAM17, while decreasing the expression of CD163 following PRRSV infection (Fig 6A and 6B). ADAM17 mRNA levels were significantly increased, but CD163 mRNA levels showed a significant reduction in each case when compared to the levels in PAMs transfected with negative siRNA (Fig 6C and 6D). Similarly, an immunofluorescence assay revealed that the expression of ADAM17 was increased (Fig 6E), whereas the fluorescence intensity of the CD163 protein was obviously reduced in TREM2-silenced PAMs following PRRSV infection ( Fig 6F). Since activated ADAM17 cleaves the ectodomain of CD163, which mediates CD163 shedding from the cell membrane and the production of sCD163, we detected membrane CD163 expression and sCD163 production following the RNA interference assay. As expected, TREM2 knockdown also induced a significant reduction of membrane-associated CD163 in PAMs based on flow cytometry (Fig 6G and 6H). Consistently, sCD163 production Akt inhibitor, 1 μM) (E) for 24 h. Western blot analysis was conducted to illustrate the protein levels of phosphorylated Syk, phosphorylated Akt, and PRRSV N, respectively. GAPDH is shown as an internal control. Data are representative of the results of three independent experiments. https://doi.org/10.1371/journal.ppat.1008543.g003 To demonstrate that TREM2 knockdown reduces cytokine-mediated CD163 expression to inhibit virus infection via Syk/PI3K and TLR4/NF-κB signaling, we first explored whether TREM2 knockdown promotes the expression of proinflammatory cytokines and type I interferons via these pathways. PAMs were mock-treated or treated with Syk inhibitor R406, PI3K inhibitor Wortmannin, or NF-κB inhibitor BAY11-7082 in conditions of TREM2 knockdown and were then mock-infected or infected with PRRSV. We found that the mRNA expression However, this upregulation mediated by TREM2 silencing was significantly decreased when cells were exposed to inhibitors in the presence or absence of PRRSV, suggesting that TREM2 silencing intensifies the expression of cytokines via Syk/PI3K and NF-κB signaling, which may affect the expression of CD163 [15,49]. To understand whether TREM2 regulates ADAM17 and CD163 expression through Syk/PI3K and NF-κB signaling, we next detected the expression of ADAM17 and CD163. As expected, the upregulation of ADAM17 induced by TREM2 knockdown was significantly decreased when cells were treated with inhibitors ( Fig 6J, 6K and 6L and S4G Fig). Consistently, the protein and mRNA expression of CD163 was decreased in TREM2-silenced PAMs without inhibitor treatment. However, the TREM2-silencing-mediated inhibition of CD163 was clearly rescued when Syk, PI3K, or NF-κB signaling was blocked (Fig 6J, 6K and  6L and S4H Fig). These results suggest that silencing of TREM2 upregulates ADAM17 and downregulates CD163 via Syk/PI3K and NF-κB signaling. To further investigate whether CD163 downregulation triggered through TREM2 silencing is dependent on cytokines, we used TLR4 inhibitor TAK-242 or dexamethasone (an anti-inflammatory drug) [50] to treat TREM2 knockdown cells. We found that the decreased CD163 expression was rescued in TREM2 knockdown cells treated with inhibitor or dexamethasone compared to the cells without treatment (S4I, S4J and S4K Fig), indicating that the effect of TREM2 on CD163 is cytokine-mediated. Inhibitors partially rescued PRRSV replication in TREM2 knockdown cells, since CD163 expression was rescued (S4J, S4K, and S4L Fig). Taken together, these data demonstrate that TREM2 knockdown decreases cytokine-mediated CD163 expression to inhibit virus infection via Syk/PI3K and TLR4/NF-κB signaling.

TREM2 regulates PRRSV via inflammation
qRT-PCR, at the early stage of infection. (F) PAMs were transfected with pcDNA3.1-control (vector) or pcDNA3.1-TREM2 for 24 h, and infected with PRRSV (MOI = 1) for 24 h. qRT-PCR is shown to depict levels of TLR4 and NF-κB mRNA. (G) PAMs were infected with PRRSV (MOI = 1) for the indicated periods after transfection with siNC or siTREM2-1 for 24 h. Western blots are shown to detect phosphorylated IκBα, NF-κB p65, and IKKα. GAPDH is shown as an internal control. (H) Immunofluorescence analysis of the nuclear translocation of NF-κB p65 and the co-localization between NF-κB p65 (green) and cell nucleus (blue) (Bar, 25 μm). (I) Quantification of NF-κB p65 translocation based on the visual images in H.~300 cells for each time point. (J) PAMs were treated as described in G, cells were collected, and a cell fractionation assay was performed. Western blots are shown to detect NF-κB p65. Histone H3 is used as a nuclear internal control. (K) PAMs with TREM2 knockdown were infected with PRRSV (MOI = 1) in the presence or absence of BAY11-7082 (NF-κB inhibitor, 10 μM) and western blots are shown to detect protein levels of phosphorylated IκBα and PRRSV N. GAPDH is shown as an internal control. Data are representative of the results of three independent experiments (mean ± SE). Significant differences compared to the control group are denoted by � (P < .05), �� (P < .01) and ��� (P < .001). https://doi.org/10.1371/journal.ppat.1008543.g004

TREM2 regulates PRRSV via inflammation
Exogenous sTREM2 inhibition of PRRSV infection might be attributed to its competitive binding with viral envelope proteins sTREM2, an important form of TREM2, is generated via cleavage by ADAM10 and ADAM17, which plays a role in promoting an inflammatory response and macrophage survival [38,51,52]. To identify the potential function of sTREM2 during PRRSV infection, we first determined whether there is production of sTREM2 in cell supernatants following PRRSV challenge. We found a marked increase of sTREM2 in supernatants at different MOIs and time points after PRRSV infection (S5A and S5B Fig). Subsequently, sTREM2 and GFP protein were expressed and purified in Escherichia coli BL21 cells (DE3) (S5C and S5D Fig). Purified GFP protein served as a negative control. The PRRSV N protein was reduced in the presence of different concentrations of sTREM2 (Fig 7A). Meanwhile, sTREM2 showed potent antagonism against PRRSV in a dose-and time-dependent manner in PAMs (Fig 7B). The endogenous TREM2 expression in cells treated with sTREM2 was comparable to that in GFP proteintreated cells at 0, 12, and 24 hpi or different MOIs (Fig 7A and 7B), suggesting that the virus inhibition is indeed attributed to the effect of sTREM2. To identify which stage of viral infection was blocked by sTREM2, we carried out viral attachment and entry assays. The results revealed that sTREM2 reduced virus particles attaching to the cell surface (Fig 7C, 7D and 7E), but had no influence on the entry of the virus particles in Marc-145 cells, suggesting that sTREM2 restrains PRRSV infection by blocking viral attachment (S5E Fig). Given that TREM2 is a transmembrane glycoprotein (GP), we speculated that there was an interaction between sTREM2 and viral structural proteins. To test this hypothesis, Myc-tagged TREM2 and different mCherry-tagged viral structural proteins were co-transfected into HEK293T cells. Immunofluorescence assays showed that TREM2 co-localized with glycoprotein (GP) 2a, GP3, GP4, GP5, and M protein, but not with N protein (Fig 7F). IP assays further showed that TREM2 interacted with PRRSV GP2a, GP3, GP4, GP5, and M protein, but there was no interaction between TREM2 and N protein (Fig 7G). To screen the domain(s) of TREM2 required for its interaction with viral structural proteins, we selected GP4 and GP5, which were largely precipitated by TREM2. To conduct IP assays. Myc-tagged full-length or truncated TREM2 and mCherry-tagged GP4 or GP5 were co-transfected into HEK293T. We found that TREM2 constructs comprising the Ig-like V-type domain co-precipitated GP4 or GP5, suggesting that an Ig-like V-type domain of TREM2 might be the binding region of viral GP4 and GP5 (S5F and S5G Fig). Collectively, sTREM2 plays a role in protection against PRRSV infection.

TREM2 expression is upregulated post PRRSV infection in piglets
To further validate the above results, we performed a PRRSV challenge in vivo. First, we detected the expression of TREM2 in different tissues and found that TREM2 was largely expressed in the lungs, which are the main target of PRRSV (Fig 8A). Based on the pathology  assessed using visual examination and H&E staining, we found severe hemorrhage, congestion, and diffuse interstitial pneumonia occurred in virus-infected lungs (Fig 8B and 8C). qRT-PCR analysis demonstrated that TREM2 mRNA was remarkably induced in lungs of infected piglets on 3 and 7 dpi, consistent with PRRSV N expression (Fig 8D and 8E). Similarly, TREM2 protein level in infected pigs was significantly higher than that of controls on 3 and 7 dpi (Fig 8F). The results of immunofluorescence indicated that the expression of TREM2 was enhanced in infected lungs and lymph nodes compared to PBS controls. TREM2 (red) and PRRSV (green) specifically co-localized in PAMs, and their colocalization was increased after PRRSV infection (Fig 8I). Blood samples were collected on days 3 and 7 postchallenge for the detection of sTREM2 and sCD163. It was clear that the production of sTREM2 and sCD163 in infected piglets was significantly higher than in uninfected controls (Fig 8G and 8H). Collectively, TREM2 is associated with PRRSV infection in vivo, and PRRSV potentiates the expression of TREM2 in lungs and lymph codes, as well as the production of sTREM2 in plasma.

Discussion
TREM2 is a novel immunoregulatory receptor with important functions within the antiinflammatory response [22,24,27]. This report identified TREM2 as a regulatory factor for PRRSV infection. It particularly covers the following (Fig 9). First, the expression of TREM2 was increased with PRRSV challenge in vivo and in vitro. TREM2 promotes PRRSV infection via the interaction between its cytoplasmic region and PRRSV Nsp2. Second, silencing of TREM2 triggered downstream PI3K/AKT and TLR4-NF-κB signaling, which resulted in an early proinflammatory cytokine and type I interferon response. Third, increased proinflammatory cytokines in TREM2-silenced PAMs reduced CD163 expression and enhanced the shedding of sCD163 cleaved by ADAM17, leading to the inhibition of PRRSV infection. Finally, sTREM2 was upregulated upon PRRSV infection and exogenous sTREM2 inhibited PRRSV infection in vitro. Taken together, TREM2 functions as an immunoregulatory receptor in PRRSV infection.
The current study provides at least three novel insights regarding the role of TREM2 in PRRSV infection to be discussed. First, CD163 was not only indispensable for PRRSV infection, but was also associated with an inflammatory response [9,53]. CD163 receptors are downregulated by various inflammatory factors, including IL-1α, IL-1β, IL-4, IL-8, LPS, TNFα, IFN-γ, TLR4, etc. Under inflammatory conditions, ADAM17 is activated and rapidly cleaves CD163 from the surface of inflammatory macrophages [16]. Many studies have highlighted that inflammatory responses are regulated by the TREM2/DAP12 complex. Gao X reported that silencing of TREM2 enhances the inflammatory response of alveolar macrophages to LPS [54]. TREM2 -/and DAP12 -/mice were generated to show that the activation of macrophages produces high levels of inflammatory cytokines compared to WT mice [26]. In this study, TREM2 knockdown in PAMs intensified the PRRSV-induced proinflammatory response and amplified the release of proinflammatory cytokines and type I interferons, particularly at the early stage of infection ( Fig 5). As expected, CD163 expression was downregulated, and the shedding of sCD163 mediated by ADAM17 was increased in TREM2-silenced PAMs (Fig 6). Early induction of proinflammatory cytokines and type I interferons in GP4, GP5, M, and N) were co-transfected into HEK293T for 24 h, respectively. Immunofluorescence analysis of the co-localization between TREM2 (green) and the structural proteins (red) (Bar, 5 μm) was performed. (G) Likewise, HEK293T cells were transfected with Myc-tagged TREM2 and these mCherry-tagged structural proteins. Cell lysates were immunoprecipitated with a Myc antibody. Immunoprecipitated proteins were subjected to immunoblot with the mCherry antibody. Asterisks mark the expressed Myc-fusion proteins of TREM2 fragments. Data are representative of the results of three independent experiments. https://doi.org/10.1371/journal.ppat.1008543.g007
Second, it was reported that NF-κB, ERK1/2 and p38 MAPK signaling participated in antiviral immunity following PRRSV infection [55][56][57]. There are dual roles of NF-κB regulation by PRRSV: inhibition and activation. It mainly manifests in the inhibition of the inflammatory response and type I interferon production in early infection while accelerating the production of proinflammatory cytokines later in the infection [58,59]. In the current study, we demonstrated that TREM2 knockdown triggered the activation of NF-κB as early as 0 hpi in PAMs, prior to PRRSV-induced activation (Fig 4G-4K). The early activation of NF-κB leading to an

PLOS PATHOGENS
TREM2 regulates PRRSV via inflammation intensive proinflammatory and type I interferon response is a possible means of inhibiting PRRSV infection. However, the NF-κB signaling pathway is reported to influence the expression of TREM1 and TREM2. More studies need to be performed to determine whether TREM2 expression is conversely regulated by the NF-κB pathway. TREM-2 interaction with TLRs plays a role in dampening immune responses. Studies showed that TREM2 or DAP12 deficient mice promote cytokine production by macrophages in response to LPS, zymosan, and CpG and suggested that TREM2 plays a role in reducing activation [26,29]. Some studies reported that LPS reduces TREM2 expression and promotes the activation of cell surface markers [60]. In our study, zymosan and Pam 3 CSK 4 upregulated TREM2 expression, while LPS downregulated TREM2 expression, which suggests TLR4 is involved in TREM2 knockdown signaling (Fig 4A).
Third, sTREM2 is an important form of TREM2, which is generated by the cleavage of ADAM10 and ADAM17 at the H157-S158 peptide bond [51,52,61]. sTREM2 was measured in plasma, cerebrospinal fluid, and bronchoalveolar lavage fluid in humans or animals suffering from bacterial, viral, and fungal infections. sTREM2 serves as an inflammatory marker with biological activity [62][63][64][65]. Here we found that the level of sTREM2 in serum dramatically increased at the early stage of infection (3-7 dpi) compared to controls in vivo (Fig 8G). Similarly, sTREM2 in cell supernatants also showed an increase during PRRSV infection in vitro (S5A and S5B Fig), indicating that an acute inflammatory response had occurred. Therefore, sTREM2 is a possible inflammatory biomarker in PRRSV infection. Interestingly, we observed that exogenous sTREM2 restrained the replication of PRRSV by blocking viral attachment (Fig 7C, 7D and 7E). We have no explanation for the higher level of TREM2 in GFP-treated cells than in the sTREM2-treated cells at 36 hpi (Fig 7B). Given that TREM2 is a transmembrane glycoprotein, we speculated that sTREM2 antagonized PRRSV binding to the cell surface via competitive binding with PRRSV structural proteins. As expected, IP assays indicated TREM2 had interactions with viral envelope proteins rather than the nucleocapsid protein (Fig 7F and 7G). Further investigation should be carried out to illustrate the binding mechanism between TREM2 and PRRSV. According to our results, sTREM2 inhibits PRRSV infection via blocking viral attachment in vitro, which shows potential therapeutic opportunities. However, sTREM2 has no effects on viral entry stage. Taking this into account, the antiviral effects of sTREM2 may work in the early stage of infection. Meanwhile, the dosage, administration route, and side effects should be taken into account when sTREM2 is tested in infected models in vivo. The emergence of sTREM2 in PRRSV-infected pigs suggests that it may serve as a potential diagnostic marker for disease prevention.
In summary, we demonstrated that knockdown of the TREM2 significantly accelerated the release of proinflammatory cytokines and type I interferons, which reduced the expression of CD163 and ultimately suppressed PRRSV infection. In addition, sTREM2 was increased upon PRRSV infection and exogenous sTREM2 inhibited PRRSV by blocking viral attachment. In general, this work highlights TREM2, a novel receptor that positively promotes PRRSV infection via an anti-inflammatory response, which provides a new perspective to cellular immune regulation following PRRSV infection as well as a new antiviral strategy.

Ethics statement
All animal experiments described in this study have been performed on pigs. These piglets were kept in housing with suitable temperature and humidity where water and food were available ad libitum. The protocol license number was IACUC-DD-16-0901, which was approved by the Institutional Animal Care and Use Committee (IACUC) of Sun Yat-sen University. All animal work was carried out under the Laboratory Animals-Guideline of welfare and ethics written by the General Administration of Quality Supervision, Inspection and Quarantine of the People's Republic of China.

Cells and viruses
PAMs were isolated from 6-week-old specific-pathogen free piglets. Piglets were euthanized and the lungs were removed in a sterile environment. Sterile phosphate buffer solution (PBS) (Corning, USA) was lavaged into the lungs and gentle massage was used after lung lavage three times. Cells were collected by centrifugation for 10 min at 1000 rpm. After washing with RPMI-1640 medium (Gibco, USA), cells were frozen in 40% RPMI-1640 medium, 50% fetal bovine serum (FBS) (PAN, Germany), and 10% DMSO (Sigma, USA). Finally, cells were gradually frozen and stored in liquid nitrogen. PAMs were cultured in RPMI 1640 medium with 10% FBS at 37˚C in 5% CO 2 . Marc-145 cell, derived from an African green monkey embryonic kidney cell, is an immortalized cell line which is permissive to PRRSV replication and commonly used in laboratories. Marc-145 cells and HEK-293T cells (China Center for Type Culture Collection, China) were maintained in Dulbecco's modified Eagle's medium (DMEM) (Corning, USA) containing 10% FBS. Two PRRSV strains, CHR6 (PRRSV-2, used for in vitro challenge assay) and Li10 (PRRSV-2, used for in vivo challenge assay) were provided by Dr. Heng Wang from South China Agricultural University. The two virus strains were propagated in Marc-145 cells and titrated to 50% tissue culture infective dose (TCID 50 ).

Challenge with PRRSV strain Li10 in pigs
Twenty piglets aged 4-6 weeks were selected for the viral challenge in vivo. Prior to infection, piglets were tested for both PRRSV antigen and antibody and were negative for both. Piglets were divided into two groups, 10 piglets per group. These piglets were kept in housing with suitable temperature and humidity where water and food were available ad libitum. After one week of acclimation, one group of piglets was challenged with the HP-PRRSV Li10 strain using nasal drip, and the other group was injected with PBS. The dose of intranasal injection was 2 × 10 5 TCID 50 Li10 in 3 mL culture medium. On days 3 and 7 post-challenge, we selected five piglets from each group. Selected piglets were euthanized, and different tissues were separated for tissue section, immunoblotting, and qRT-PCR. Blood samples were collected on the above-mentioned days for the detection of sTREM2 and soluble CD163 (sCD163).

RNA interference
All small interference RNAs (siRNAs) against TREM2 and siRNA-negative control (NC) were designed and synthesized by Invitrogen. PAMs were seeded in six-well plates at 2×10 6 cells/ well, adherent cells were transfected with the indicated siRNAs at a final concentration of 10 nM using Lipofectamine RNAiMAX (13778, Invitrogen, USA) according to the manufacturer's instructions for 24 h. Cells were infected with PRRSV for the indicated hours, 24 h posttransfection. Indicated siRNAs are listed in S1 Table.

Protein expression and purification
The extracellular domain of TREM2 gene and GFP gene were cloned into the expression vector PET32a and expressed in Escherichia coli BL21 cells (DE3) (CD601, TransGen Biotech, China). sTREM2 and GFP protein were induced with 1 mM IPTG (9030, Takara, Japan) at 37˚C with shaking at 220 rpm for 12 h. After centrifugation, the sediments were resuspended with equilibration buffer and bacteria were lysed using the Ultrasonic Cell Disruptor. Verified sTREM2 and GFP protein were purified with Ni-NTA agarose (R90101, Invitrogen, USA), according to the manufacturer's instructions. The concentration of purified sTREM2 and GFP protein was determined using the BCA Protein Assay Kit (23227, Thermo Scientific, USA).

Immunofluorescence assay
Twelve-well plate cell samples were fixed with paraformaldehyde (Ruishu, China) for 10 min and then immediately permeabilized by 0.5% Triton (Ruishu, China) for 15 min at room temperature. After rinsing with PBS three times, cells were blocked with 1% BSA for 30 min and incubated with the primary antibody overnight at 4˚C. Cells were incubated with the secondary antibody for 1 h, followed by staining with 4',6-Diamidino-2-phenylindole dihydrochloride (DAPI) (C1002, Beyotime, China) in PBS for an additional 5 min. Fluorescent images were acquired with an inverted fluorescence microscope (NIKON ECLIPSE Ti2-U, Japan) or a confocal laser scanning microscope (TCS-SP5, LEICA, Germany).

Measurement of sTREM2 and sCD163
Serum was separated from blood samples and the supernatants were collected from cultured cells. These samples were used for the detection of sTREM2 and sCD163. The porcine TREM2 ELISA kit and porcine CD163 ELISA kit (JL30780 and JL27715, Laibio, China) were used to measure the absorbance (450 nm) according to the manufacturer's instructions.

Flow cytometry
PAMs were harvested and fixed with paraformaldehyde for 10 min. Then, cells were gently rinsed with PBS three times. PAMs were stained with antibodies in 1% BSA for 1 h at room temperature. After rinsing with PBS three times, cells were resuspended in PBS at a concentration of 1 × 10 6 cells/mL. Data were collected on a FACSCalibur flow cytometer (BD Bioscience, USA), and analyses were performed using the FlowJo software.

Quantitative real-time reverse-transcription polymerase chain reaction (qRT-PCR)
Total RNA was extracted from cultured cells using the TRIzol reagent (Magen, China). Total RNA (1 μg) was reverse transcribed into cDNA using the Reverse Transcription System (A3500, Promega, USA) in a 10 μL reaction volume following the manufacturer's protocol. Oligo (dT) 15 primer (C110A, Promega, USA) and random primer (C118A, Promega, USA) were the reverse-transcription primers. Reverse-transcription products were amplified with 2×RealStar Green Power Mixture (GenStar, China) on a LightCycler 480 Real-Time PCR System (LC480, Roche, Switzerland). The primers used for qRT-PCR are listed in S2 Table. GAPDH or HPRT1 was used as a housekeeping gene. Relative expression of target genes was calculated using the 2 −ΔΔCt method and normalized with the mean Ct of GAPDH or HPRT1.

Co-immunoprecipitation (Co-IP) and immunoblotting
Cells were lysed in radio immunoprecipitation buffer (RIPA) (Beyotime, China), supplemented with PMSF and phosphatase inhibitors (Beyotime, China). Immunoprecipitation was performed according to the manufacturer's instructions of the Dynabeads Protein G Immunoprecipitation Kit (10007D, Invitrogen, USA), 3 μg of the indicated antibodies were first bound to 40 μL Dynabeads for 1 h at room temperature. Cell lysate protein (1 mg) was incubated with the Dynabeads-Ab complex for 1 h at room temperature. Dynabeads-Ab-Ag complexes were washed with Washing Buffer three times and retained in Elution Buffer. The precipitates and PLOS PATHOGENS TREM2 regulates PRRSV via inflammation whole-cell lysates were subjected to 10-15% SDS-PAGE and potentially associated proteins were tested by western blot as stated above.

Hematoxylin-Eosin (H&E) staining
When piglets were euthanized on the indicated day post PRRSV challenge, lungs and other tissues were isolated. Pictures were taken from the dorsal side of the lungs to evaluate pathology. Lungs and lymph nodes were cut into appropriate tissue blocks and fixed in 4% neutral buffered formalin (Ruishu, China) for 18 h. After dehydration and embedding in paraffin, tissue samples were cut into 5 μm thick sections and mounted onto glass slides. For histopathology, sections were stained with H&E (NJJCTECH, China) according to the manufacturer's instructions.

Statistical analysis
All experiments were performed with at least three independent replicates. Student's t-test and one-way ANOVA were used to analyze the data. Statistical analysis was performed using SPSS 17.0 and GraphPad Prism 5.0. Shown are means and error bars represent SE in all graphs. P < 0.05 was considered to be significant. Lane 1 is the control. Lane 2 is the purified GFP protein (27 kDa). M is the protein molecular weight marker. (E) For the entry assay, Marc-145 cells were initially challenged with PRRSV (MOI = 5) for 3 h at 4˚C. Then, unbound viral particles were removed, and cells were cultured at 37˚C in the presence of various concentrations of sTREM2 or GFP protein (20 and 40 pg/ mL) for 6 h. These proteins were added at 0, 2, or 4 h after the temperature shift to 37˚C. After washing with PBS three times, cells were incubated for another 24 h at 37˚C. Western blot was used to detect the expression of PRRSV N. Purified GFP protein serves as a negative control. (F and G) mCherry-tagged GP4 (F) or mCherry-tagged GP5 (G) was co-transfected with Myctagged full-length or other truncated TREM2 in HEK293T. IP assays with anti-Myc antibody were performed to determine their interaction, and the GP4-mCherry or GP5-mCherry was detected by IB with the anti-mCherry antibody. Asterisks mark the expressed Myc-fusion proteins of TREM2 fragments. Data are representative of the results of three independent experiments. (TIF) S1