Double-stranded RNA analog and type I interferon regulate expression of Trem paired receptors in murine myeloid cells

Triggering receptors expressed on myeloid cells (Trem) proteins are a family of cell surface receptors used to control innate immune responses such as proinflammatory cytokine production in mice. Trem genes belong to a rapidly expanding family of receptors that include activating and inhibitory paired-isoforms. By comparative genomic analysis, we found that Trem4, Trem5 and Trem-like transcript-6 (Treml6) genes typically paired receptors. These paired Trem genes were murine-specific and originated from an immunoreceptor tyrosine-based inhibition motif (ITIM)-containing gene. Treml6 encoded ITIM, whereas Trem4 and Trem5 lacked the ITIM but possessed positively-charged residues to associate with DNAX activating protein of 12 kDa (DAP12). DAP12 was directly associated with Trem4 and Trem5, and DAP12 coupling was mandatory for their expression on the cell surface. In bone marrow-derived dendritic cells (BMDCs) and macrophages (BMDMs), and splenic DC subsets, polyinosinic-polycytidylic acid (polyI:C) followed by type I interferon (IFN) production induced Trem4 and Treml6 whereas polyI:C or other TLR agonists failed to induce the expression of Trem5. PolyI:C induced Treml6 and Trem4 more efficiently in BMDMs than BMDCs. Treml6 was more potentially up-regulated in conventional DC (cDCs) and plasmacytoid DC (pDCs) than Trem4 in mice upon in vivo stimulation with polyI:C. Treml6-dependent inhibitory signal would be dominant in viral infection compared to resting state. Though no direct ligands of these Trem receptors have been determined, the results infer that a set of Trem receptors are up-regulated in response to viral RNA to regulate myeloid cell activation through modulation of DAP12-associated Trem4 and ITIM-containing Treml6.


Background
Some multiple immune receptors, such as signal regulatory protein (SIRP), CD200R, killer inhibitory receptor (KIR), lymphocyte antigen 49 (Ly49), CD300, have a highly homologous extracellular domain [1][2][3][4]. These receptors are termed "paired receptors" that are generally defined by the following characteristics: 1. they have a high level of amino acid sequence identity within their extracellular domains, 2. they are encoded by different genes, but located in the same gene cluster, 3. they have both activating receptor associated by immunoreceptor tyrosine-based activation motif (ITAM)-containing adaptor, such as DNAX activating protein of 12 kDa (DAP12), and immunoreceptor tyrosine-based inhibition motif (ITIM)-containing inhibitory receptors [4][5][6]. Their activating and inhibitory receptors recognize endogenous/exogenous components and often differ in their binding specificity. Paired receptors expressed on myeloid cells including dendritic cell (DC) subsets have been poorly understood in their functional features concerning innateadaptive interplay, whereas paired NK receptors including Ly49, KIR, and Natural Killer group 2 (NKG2)/CD94, have been characterized to explain the "missing self" activity of NK cells [3,7]. Therefore, we attempted to investigate the role of uncharacterized paired receptors in myeloid cells. We first searched for genes of myeloid paired receptors in mammalian genome database and analyzed their expression profiling and immune-related function.
Triggering receptors expressed on myeloid cells (Trem) proteins are a family of cell surface receptors expressed broadly on myeloid cells [8]. The receptors are a structurally-related protein family that generally contain a single immunoglobulin (Ig) domain in the extracellular region, and their gene cluster is located on mouse chromosome 17C and human chromosome 6p21.1 [9,10]. Inhibitory Trem proteins, containing cytoplasmic ITIM recruit protein tyrosine phosphatases such as Src homology 2 domain-containing phosphatase 1 (SHP-1) and SHP-2 [11]. On the other hand, activating Trem proteins lack cytoplasmic signaling elements, but contain a charged residue in their transmembrane domain that facilitates the association with DAP12, which functions as docking sites for Syk tyrosine kinases that mediate PLCγ and PI3K activation [12,13].
Trem and Trem-like genes such as novel immune type receptors have also been reported in teleost [14][15][16]. Human Trem1 is an activating receptor expressed at high levels on neutrophils and monocytes that infiltrate human tissues infected with bacteria [17]. Recent studies have shown that Trem1 has a critical role in negative regulation of proinflammatory cytokine production in vivo [18,19]. Trem2 is also activating receptor expressed on DCs, osteoclasts and microglia and control not only immune function such as maturation of DCs but also osteoclastogenesis [20,21]. Functional DAP12associated Trem3 is present in mice, but not primates [22]. Treml4 binds to dead cells and accelerates antigen presentation and priming of CD4 + and CD8 + T cells in vivo [23,24]. Trem4 (also known as pDC-Trem) is involved in IFN-α production from Fms-related tyrosine kinase 3 (Flt3) ligand-induced plasmacytoid DCs (pDC), and TLR signals are essential for Trem4 expression [25]. Nevertheless, other Trem and Treml receptors have been poorly characterized in mice, although they show myeloid cell-dominant expression [8]. There is so far no report on the comprehensive analysis of the mouse Trem/Treml gene cluster, and we found paired receptorlike signatures in several Trem receptors on database search.
In this study, we focused on orphan and uncharacterized Trem genes, named murine Trem4 (deposited as pDC-Trem [25]), Trem5 (known as 9830107B12Rik) and Treml6 (known as B430306N03Rik), and elucidated their evolutional stands in vertebrates. The ligands of these receptors remain undetermined, but we noticed that polyI:C signal dynamically up-regulated the Trem4 and Treml6. Although the extracellular immunoglobulin (Ig) domain of Trem4, Trem5 and Treml6 were closely related to each other, Treml6 encoded ITIM in its cytoplasmic region, whereas Trem4 and Trem5 lacked ITIM but had positively-charged residues to associate with DAP12. Together, these expression features indicate that these receptors are typical paired receptors, which modulate innate signal in myeloid cells like the case of Trem1 and Trem2 [26]. Trem4 and Treml6 may be up-regulated on the cell-surface in myeloid cells by viral RNA to counteract or modulate presetting immune response.

Mice
All mice were backcrossed with C57BL/6 mice more than seven times before use. Ifnar1 −/− mice were obtained from the Jackson Laboratory. Ticam-1 −/− and Mavs −/− mice were generated in our laboratory [27,28]. All mice were maintained under specific pathogen-free conditions in the animal facility of the Hokkaido University Graduate School of Medicine. Female mice 8-12 weeks of age were used in all experiments, all of which were performed according to the guidelines issued by Hokkaido University Animal Care and Use Committee, who approved this study as ID number: 08-0243, "Analysis of immune modulation by toll-like receptors". Mice were sacrificed according to the rule of animal euthanasia. Animal experiments protocols and guidelines were approved by the Animal Safety Center, Hokkaido University, Japan.

Database analysis
A BLAST search of the vertebrate Trem genes was performed by the amino acid sequences of human and mouse Trem genes as queries using the Ensembl genome databases (http://www.ensembl.org/index.html). The genome assembly versions of species used in this paper were as follows: GRCh37 of human, GRCm38 of mouse, Ory-Cun2.0 of Rabbit, CanFam3.1 of Dog, EquCab2 of Horse, UMD3.1 of Cow, Sscrofa10.2 of Pig, BROADO5 of opossum, Galgal4 of chicken, JGI_4.2 of frog and Zv9 of zebrafish. Domain structures of Trem proteins were analyzed by SMART program (http://smart.embl-heidelberg.de). A phylogenetic tree based on the amino acid sequences of Ig domain was constructed by the Neighbor-joining (NJ) method in the ClustalX version 2 program and the MEGA version 5 program [29,30]. The distance matrix was obtained by calculating p-distances for all pairs of sequences. Sites containing gaps were excluded from the analysis using the pairwise deletion option. The reliability of branching patterns was assessed by bootstrap analysis (1000 replications). Vertebrate Trem genes used for phylogenetic and comparative genomic analysis are listed in supplementary Table 1.

RNA isolation and cDNA synthesis
Total RNA from BMDCs, BMDMs, splenic DC subsets, and tissues was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA, USA), after which 0.1-1 mg RNA was reverse transcribed using a high-capacity cDNA transcription kit with an RNase inhibitor kit (Applied Biosystems Foster City, CA, USA), according to the manufacturer's instructions.

Identification of murine paired trem genes
All murine Trem proteins are type-I transmembrane proteins containing single extracellular Ig domains [10]. To assess the homology of these Trem proteins, we analyzed amino acid sequence identities of the murine Trem proteins (  1). Trem4, Trem5 and Treml6 all encoded a threonine-riched stalk between the Ig domain and transmembrane region (TM), suggesting that they recognize a similar signature of the ligand. Treml6 encoded two ITIMs that formed a conserved sequence of amino acids (S/I/V/LxYxxI/V/L) in the cytoplasmic region (CPR), whereas Trem4 and Trem5 lacking the ITIM of positively-charged residues to enable them to associate with DAP12 in the TM. Consequently, these features indicate that these Trem proteins are typical paired receptors.
Murine paired trem genes are murine-specific and related most closely to ITIM-containing Treml7 genes In BLAST search of the jawed vertebrate genome data deposited in Ensembl genome databases, we found additional two ITIM-containing Trem genes, named Treml7 and Treml8 and identified in dog/horse and rabbit/opossum, respectively ( Fig. 1, Additional file 1: Table S1). Their amino acid sequences indicated the presence of two ITIMs in putative CPR of Treml7 and Treml8.
To clarify the evolutional history of the vertebrate Trem gene cluster, we performed comparative genome analysis to compare the human and murine Trem genomic regions with the corresponding regions of the rabbit, dog, horse, cow, pig, opossum, chicken, frog (X. tropicalis) and teleost (zebrafish). Our BLAST search of the genome data failed to identify obvious Trem genes in the teleost genome, whereas we found a single Trem gene showing high similarity to vertebrate Trem2 in the X. tropicalis genome (Fig. 2a). The tetrapod Trem genes are located between the Nfya and Foxp4 genes, and these syntenic blocks are highly conserved between the human and frog genomes. The Treml6, Treml7 and Treml8 genes are all located between the mammalian Trem2 and Treml2 genes. On the other hand, the Trem4 and Trem5 genes are located between the murine Trem1 and Foxp4 genes. To address the relationship between the paired Trem genes and other Trem family members, we constructed a neighbor-joining tree (Fig. 2b). We found that three paired Trem genes formed a unique cluster independent of other vertebrate Trem genes. Hence, murine paired Trem genes are murine-specific. Additionally, these paired Trem genes are related most closely to dog and horse ITIM-containing Treml7 genes by a high bootstrap value. On the other hand, Treml8 genes formed a unique cluster separate from Treml7 and paired Trem genes. A possible process of gene-clustering of the Trem family is depicted in Fig. 2c.

Trem4 and Trem5 associated with DAP12
DAP12 is directly associated with Trem4 (pDC-Trem) [25]. This association is essential for expression of Trem4 on the cell surface and its signal transduction.
In addition, association of DAP12 with immune receptors is required for post-translational modification including glycosylation in the Golgi apparatus [32]. We predicted the N-glycosylation sites of paired Trem receptors using NetNGlyc program (http://www.cbs.dtu.dk/ services/NetNGlyc/), and subsequently identified several sites in the amino acid sequences of these Trem proteins (Fig. 1). To investigate whether DAP12 is required for post-translational modification of paired Trem proteins, we performed immunoblotting analysis using HEK293FT cells transfected with HA-tagged Trem with/without FLAG-tagged DAP12. Immunoblotting of Treml6 and/or DAP12 lysates showed that two bands, mature (high molecular weight band) and immature (low molecular weight band) Treml6 proteins, were detected regardless of co-transfection of the DAP12 plasmid (Fig. 3a). However, immunoblotting of Trem4 and Trem5 lysates from DAP12-untransfected cells showed that only singleimmature Trem4 and Trem5 proteins were detected. In contrast, immunoblotting of the lysates from DAP12transfected cells detected mature Trem4 and Trem5 proteins, which were heavily glycosylated (Fig. 3a). Hence, DAP12-binding is required for post-translational modification of Trem4 and Trem5, but not Treml6. To investigate the association of Trem proteins with DAP12, various combinations of the HA-tagged Trem and FLAGtagged DAP12 were transfected into HEK293FT cells and were then subjected to immunoprecipitation analyses. DAP12 was co-immunoprecipitated with Trem4 and Trem5, but not Treml6, indicating that DAP12 binds directly to Trem4 and Trem5 in HEK293 cells (Fig. 3b). To investigate whether DAP12 is essential for the Trem4/5's expression on the cell surface, we performed FACS analysis of Trem and/or DAP12transfected HEK293FT cells. The expression of Trem4 and Trem5 on the cell surface was induced by cotransfection with DAP12, whereas Treml6 was constitutively surface-expressed regardless of DAP12 transfection under the same conditions (Fig. 3c). Taken together, these data suggest that association with DAP12 is essentially required for post-translational modification and surface-expression of Trem4 and Trem5 proteins.

Expression of paired trem receptors in BMDCs and BMDMs
Trem4 expression on pDCs is correlated with TLRdependent type I IFN production [25]. Treml6 gene expression is induced by type I IFN produced in response to mouse cytomegalovirus infection in CD8α + and CD11b + conventional dendritic cells (cDCs) [33]. Therefore, Trem4 and Treml6 are likely induced via nucleic acids-sensing pathways. To investigate the IFN-inducible pathway for paired Trem gene expression, we performed qPCR analysis for BMDCs stimulated with various TLR agonists. None of the TLR agonists tested stimulated the expression of the Trem5 gene in BMDCs, and Trem5 basal expression was rather decreased in response to      TLR2 agonist (Pam3CSK4). In contrast, polyI:C strongly induced the expression of Trem4 and Treml6 genes,~10-fold over the basal levels (Fig. 4a). Treminducing profiles were also observed with macrophages (BMDM) in responses to polyI:C (Fig. 5).
Myeloid cells, macrophages and some DC subsets including cDC play a critical rule in both innate and adaptive immunity. Activation of cDCs by polyI:C robustly promoted the production of type I IFN [34]. In cDCs, both the TLR3/Toll-IL-1 receptor domaincontaining adaptor molecule 1 (TICAM-1) and MDA5/ mitochondrial antiviral signaling protein (MAVS) pathways lead to production of type I IFN in response to polyI:C [35,36]. To investigate the role of the TICAM-1 and MAVS pathways in type I IFN production by BMDCs, we stimulated BMDCs derived from Ticam-1-or Mavs-deficient mice. The results showed that the production of type I IFN was mainly dependent on the MAVS pathway, but not the TICAM-1 pathway (Fig. 4b). To clarify the mechanism of Trem4 and Treml6 induction, we investigated the Trem expression in various genetically-manipulated mice. The expression of Trem4 and Treml6 genes was abolished in BMDCs derived from Mavs-and Ifnar1-deficient mice (Fig. 4c). Additionally, the expression of the Trem4 and Treml6 genes was completely blocked by treatment with IFNAR1-neutralizing antibody (Fig. 4d). Ultimately, type I IFN signaling is required for the expression of Trem4 and Treml6 in BMDCs. Hence, our data indicate that expression of DAP12-associated Trem4 and Trem5 genes is differentially regulated in BMDCs, whereas Trem4 and Treml6 genes are regulated by polyI:C-induced type I IFN. Likewise, polyI:C-inducible properties of Trem4 and Treml6 were confirmed in the BMDMs (Fig. 5). The mRNA levels of Trem4 and Treml6 tended to be more upregulated in BMDMs than in BMDCs in response to polyI:C (Fig. 4a vs. Fig. 5b).

Expression of paired trem genes in tissues and splenic DC subsets
We compared the tissue distributions of paired Trem transcripts in wild-type mice by qPCR analysis. Among the three paired Trem genes, Trem4 and Trem5 were expressed to similar levels in spleen, thymus, small intestine, liver and lung, whereas Treml6 was mainly expressed in spleen (Fig. 6a). Thus, all these paired Trem transcripts are highly expressed in the spleen, only where Treml6 ITIM-inhibitory effect can be exerted. To examine whether polyI:C-MAVS stimulation induces expression of the paired Trem genes, mice were administered with polyI:C by intraperitoneal injection, and monitored their expression levels in the spleen by qPCR analysis. In conformity with the results observed in BMDCs, polyI:C stimulation was able to induce the expression of Trem4 and Treml6, but not Trem5 (Fig. 6b). Therefore, these data indicate that expression of DAP12associated Trem4 and Trem5 genes are differentially regulated in vivo. To investigate whether type I IFN is required for mRNA induction of Trem4 and Treml6 in splenic cDC and pDC, we monitored the expression levels of Trem4 and Treml6 mRNA in FACS-sorted cDC and pDC from wild-type and Ifnar1-deficient mice (Fig. 6c). We found that the mRNA expression of Trem4 and Treml6 is induced in both cDCs and pDCs in wild-type mice following polyI:C-injection, whereas their expression is abolished in Ifnar1-deficient mice. These results indicate that type I IFN is required for Trem4 and Treml6 induction in spleen-resident cDCs and pDCs as in BMDCs.

Discussion
This report is the first to provide evidence that mouse has paired Trem receptors in myeloid cells and increases the expression levels in response to type I IFN. The term "paired receptor" is commonly used to describe families of membrane receptors that have similar extracellular regions but different TM and CPR. Trem4/ Trem5 vs. Treml6 are typical paired receptors in the murine linage (Figs. 1 and 2b). PolyI:C stimulation triggers the mRNA expression of Trem4 and Treml6 via type I IFN, which is an output produced through the MDA5/ MAVS pathway in BMDCs (Figs. 4b, c, d and 6c). All paired Trem genes are highly expressed in secondary lymphoid organs such as the spleen, which implies that the pair receptor effect coincides with an event (See figure on previous page.) Fig. 2 Comparative genome analysis of Trem genes in vertebrates. a Synteny block including Trem and their neighbor genes. The arrows represent the genes and their transcriptional orientations annotated in the Ensembl genome data for each species. Gray and black arrows indicate ITIM-containing Trem and other Trem genes, respectively. This figure is not drawn to scale. b Neighbor-joining tree of vertebrate Trem genes based on extracellular Ig domains. Numbers at the nodes represent the bootstrap confidence level in percentage. Only bootstrap values over 40 are indicated. Hs; human (Homo sapiens), Mm; mouse (Mus musculus), Oc; rabbit (Oryctolagus cuniculus), Cf; dog (Canis lupus familiaris), Ec; horse (Equus caballus), Bt; cow (Bos taurus), Ss; pig (Sus scrofa), Md; opossum (Monodelphis domestica), Gg; chicken (Gallus gallus), Xt; frog (Xenopus tropicalis). c Evolutionary model of murine paired Trem genes. Firstly, ancestral paired Trem gene and Treml7 rose from common ancestral gene that encored two ITIMs in its CPR. After brunching to murine linage, ancestral paired Trem gene containing ITIMs were duplicated, and the one of them evolved murine Treml6 gene. Another duplicated gene was translocated between Trem1 and Foxp4 genes and lost ITIMs. Finally, the common ancestor of Trem5 and Trem4 genes was duplicated, and evolved murine Trem5 and Trem4 of type I IFN induction. The Syk/SLP76-PKC or PI3K pathways are known to be activated downstream of their signal, where type I IFN also participates. Therefore, Trem4, Trem5 and Treml6 may participate in innate immune response in a myeloid-specific manner.
We also identified two novel ITIM-containing Trem genes, named as Treml7 in the dog and horse genomes and Treml8 in the rabbit and opossum genomes (Fig. 2a, b). Interestingly, either of these genes are lacking in other mammalian genomes. This finding indicates that species-specific construction occurs in paired Trem receptors. Phylogenetic analysis shows that the murine paired Trem genes are most closely related to Treml7 genes, but not Treml8. Hence, murine paired Trem genes and Treml7 are likely to originate from a common ancestral gene. A possible evolution of murine paired Trem genes is summarized in Fig. 2c. Initially, the ancestral paired Trem gene and Treml7 arose from a common ancestral gene that encoded two ITIMs in its CPR. After branching off of the murine linage, the ancestral paired Trem gene containing ITIMs was duplicated, and the murine Treml6 gene evolved from one of the duplicated genes. Another duplicated gene was translocated between the Trem1 and Foxp4 genes and lost its ITIMs. Finally, the common ancestor of Trem4 and Trem5 genes was duplicated, and evolved into murine Trem4 and Trem5.
Notably, the above-mentioned model clearly shows that activating Trem4 and Trem5 are derived from the ancestral ITIM-containing Trem gene. In human, siglec-14 or human/murine NK receptors fall into the same family [1,[37][38][39]. Barclay and Hatherley propose the "counterbalance theory" to explain the evolution of paired receptors [6]. In this theory, the driving force in the evolution of paired receptors is that pathogens target the inhibitory receptors on immune cells and then activating receptors evolve from these inhibitory receptors [6]. Although macrophages and DCs express Trem4, Trem5 and Treml6, natural ligands for these Trems have not been identified in either endogenous or microbial origin. How these receptors interact with NK cells, therefore, remains undetermined. Trem proteins other than Trem4/Treml6 are involved in inflammatory response [26]. Trem1 is a pro-inflammatory mediator that amplifies endotoxin-induced shock in sepsis [17,40]. Trem1 constitutively associates with DAP12 and recognizes endogenous ligands, which is indispensable for induction of intracellular signals [17,41]. The Trem1 signal appears to act in concert with other pattern-recognition receptors (PRRs) including TLRs and modify the cytokine/chemokine profile in myeloid cells, but its physiological mechanism remains largely elusive. Trem2 also couples with DAP12, and make a crosstalk with TLR signaling [42]. Trem2 essentially (but not simply) suppresses inflammatory responses, although their endogenous ligands are largely unknown [26,41,42]. Trem1 and Trem2 appear differentially work in the context of inflammation and microbial environment. Watarai et al. showed that the Trem4-DAP12 complex mediates positive signaling for type I IFN production in Flt3 ligand-induced pDCs [25]. Here we find that direct association of DAP12 with Trem4 and Trem5 is required for their post-translational modification and cell surface expression (Fig. 3). Both receptors may lead to activation of Src-family kinases and subsequent phosphorylation of paired tyrosine residues in the ITAM of DAP12. On the other hand, activation of the MAVS pathway and type I IFN are required for surface-expression of Trem4. PolyI:C induce high surface-expression of Trem4 in the presence of DAP12, while polyI:C alone sustains high expression of Treml6 independent of DAP12. Notably, none of the TLR agonists tested participate in their expression. There is a gene-inducing program in DCs that evokes DC priming in response to internalization of viral RNA [43], where DAP12-associated Trem4 induction would be involved. Since infected DCs usually induce cell death [44], viral RNA is likely to be extrinsically taken up into DC/macrophages in the form of exosome or debris. What is responsible for Trem5 expression in combination with DAP12 is unknown. Trem5 expression might be controlled by other innate receptors such as C-type lectin receptors and NOD-like receptors.
BMDMs more efficently induce up-regulation of Trem4 and Treml6 in response to polyI:C compared to BMDCs (Fig. 4a vs. Fig. 5b). Trem-DAP12 complex may facilitate macrophage activation by enhancing phagocytosis and cytokine production. In fact, the Trem2-DAP12 axis plays a critical role in evoking phagocytosis in addition to production of inflammatory cytokines in macrophages and granulocytes in mice challenged with LPS [45]. Taken together with our results that both Trem4 and Treml6 are expressed through type I IFN signaling in macrophages and DC subsets, Trem proteins may be engaged in fine-tuning of DC maturation and myeloid immune response via the DAP12 axis and IFN-inducing pathways. The repertoire of Trem receptors has been evolutionally selected depending on species and environment. In mouse, Trem proteins modify allergic states and host response to some infections [46,47], which will be etiologically connected with the profiles of the Trem expression.

Ethics approval
This research does not involve human materials. All animal research protocols for this work were reviewed and approved by the Animal Safety Center, Hokkaido University, Japan.

Availability of data and materials
The dataset supporting the conclusions of this article is included within the article. Wild-type mice were intraperitoneally injected with polyI:C (200 μg/ml). After 3 h, total RNA was isolated from spleen in each mouse. c Treml6 and Trem4 expression after polyI:C stimulation in splenic cDCs and pDCs. Wild-type mice were intraperitoneally injected with polyI:C (200 μg/ml). After 3 h, total RNA was isolated from FACS-sorted cDCs and pDCs. In these experiments, mRNA expressions of paired Trem genes were measured by qPCR and their expression levels were normalized to that of Gapdh mRNA. Data are representative of three independent experiments. The data shown are representative of at least two independent experiments. Data are means ± SD of three independent samples. *p < 0.05. **p < 0.01