Evolution and functional divergence of NLRPgenes in mammalian reproductive systems

1The innate immune system is an ancestral and ubiquitous system of defense against microbial infection and other potential threats to the host. The first mammalian molecules shown to be involved in innate immune recognition of, and defense against, microbial pathogens are the Toll-like receptors (TLRs), which constitute the main sensors for detection of extracellular microbes. Recent research has shown that Nod-like receptors (NLRs) act as intracellular regulators of bacterial-induced inflammation.

2The NLR protein family is separated into several subfamilies based on the different N-terminal effecter domains. The NLRP subfamily (known also as NALP family) is a new identified NLR group characterized by a PYRIN domain. Initially, research on the function of NLRP proteins focused on their roles in apoptotic and inflammatory signaling pathways via the formation of a large signaling platform (named inflammasome) and the activation of caspases in innate immunity. Interestingly, more and more researches have recently revealed that some NLRP genes play roles in mammalian reproduction. For instance, the mouse Nlrp5 (known also as MATER) was one of the first identified mammalian maternal effect genes, i.e. it encodes mRNA required for successful development of a fertilized oocyte. In particular, the targeted invalidation of mouse oocyte-specific Nlrp5 leads to the sterility of females due to a blockage at the two-cell stage in the development of embryos. In vitro knockdown experiments in mouse fertilized eggs also revealed that a decrease of mouse germ-cell-specific Nlrp14 transcripts results in an arrest of development between the one-cell and eight-cell stages of the embryos. In the mouse, exclusive duplications of Nlrp4 and Nlrp9 have been detected with specific expression profiles restricted to the oocyte, whereas the mouse Nlrp4a and the cattle NLRP9 have also been reported to be expressed in the testis. In humans, the mutations of NLRP7 are found to cause recurrent hydatidiform moles, spontaneous abortions, stillbirths and intrauterine growth retardation. Additionally, the recent expression analyses of NLRP genes in the human and the rhesus macaque (Macaca mulatta) have shown that most if not all NLRP genes are expressed in primate gametes and early embryos, suggesting a role of NLRPs in primate pre-implantation development.

3Overall, it seems that the NLRP genes play roles not only in innate immunity but also in the reproductive system of mammals. Moreover, Nalp5 and Nalp14 invalidation data in the mouse suggests that there is no functional compensation of each of these oocyte-specific genes. In this context, one question is to study how these reproduction-related NLRP genes evolved and diverged in function after their duplication, as well as whether natural selection played a role during their evolution.

4The availability of several completely sequenced vertebrate genomes allows us to use a phylogenomic approach to identify the NLRP orthologues from different mammals and to evaluate the evolutionary features of NLRPs in different lineages. In this paper, we focus on the evolution and functional divergence of reproduction-related NLRPs, as well as lineage-specific expansion of Nlrp4 and Nlrp9 in rodents. The aim of the present work is to provide new information in understanding how NLRP genes evolve in reproductive system and how different mammals adapted to develop their own NLRP gene copies particularly in reproductive biology.

5A total of 83 mammalian NLRP amino acid sequences with chicken NLRP3-like as outgroup were used for the reconstruction of phylogenetic trees. A total of 122 positions are included in the final dataset.

6In the consensus phylogenetic tree (Figure 1), each NLRP protein, except for NLRP2, is shown to be monophyletic by all four methods with high bootstrap values. NLRP2 encompasses the primate-specific NLRP7, suggesting the origin of NLRP7 from a duplication of the NLRP2/7 ancestor in primates. NLRP11 is also primate-specific and might be related to the reproduction-related NLRP4 and 9. NLRP4 proteins are only identified in primates and rodents, indicating the potential birth of the ancestral NLRP4 gene before the divergence between primates and rodents. Otherwise, the other NLRP genes seem to have duplicated before the divergence of mammals.

7In rodents, three Nlrp genes show lineage-specific duplications including Nlrp1, Nlrp4 and Nlrp9 (Figure 1). More specifically, Nlrp1 exclusively duplicated in the mouse, Nlrp9 duplicated independently in the rodents after the separation of mouse and rat, whereas the duplication of Nlrp4 is likely to have occurred both before and after this separation.

8Unexpectedly, NLRP8 and NLRP13 seem to be universal in mammals except for rodents. We speculate that these two genes might have been lost during evolution of the rodents. Unfortunately, our tBLASTn search (using the human NLRP8 and 13 respectively as queries against the mouse genome) failed to find any trace of pseudogene of neither NLRP8 nor NLRP13 within the mouse conserved syntenic region. We found that, on the human chromosome 19, these two genes are located side by side between NLRP4 and NLRP5, whereas the mouse Nlrp4 gene has been expansively duplicated and the duplicates are located around Nlrp5 on the mouse chromosome 7. In rodents, the biological significance of the probable loss of NLRP8 and 13 on one hand, and the parallel expansion of Nlrp4 on the other hand, remains unknown.

9Remarkably, our phylogenetic analyses identified a well supported (bootstrap values >90% by ME, NJ and ML methods) reproduction-related clade including nine NLRP proteins: NLRP2, 4, 5, 7, 8, 9, 11, 13 and 14 (Figure 1). All these NLRP genes have been shown to be expressed in the human oocyte and pre-implantation embryos. NLRP4, 5, 8, 9 and 14 are exclusively expressed in germ-cells, especially in the oocyte of mammals. The mutations of NLRP7 are associated with abnormal embryo development in the human. The expression of mouse Nlrp2 has been found to be highly enriched in fully grown oocytes but diminished in the 2-cell embryos upon embryo genome activation. In both the human and the rhesus macaque, NLRP11 and 13 have been found specifically or preferentially expressed in the oocyte, with a similar expression pattern to other oocyte-specific NLRP genes, i.e., enriched in maturing oocytes and then progressively diminished in embryos, indicating an exclusive maternal origin of these transcripts. Unfortunately, the biological study of these two genes is not sufficient to address their putative roles in reproduction.

10Moreover, this reproduction-related clade is also supported by the syntenic analysis (Figure 2). In the human genome, all the reproduction-related NLRP genes, except for NLRP14, are tandemly located on the chromosome 19q13.42, suggesting that multiple tandem duplication events might have given birth to these genes. The similar arrangements of the reproduction-related NLRPs are also found in cattle and dog genomes (Figure 2). In the mouse genome, the reproduction-related Nlrps (except for Nlrp14) are not located side by side, but interrupted by some other types of genes, especially and mainly by certain V1r genes (discussed below) on chromosome 7A1 and 7A3. Although NLRP14 is not located in the syntenic region encompassing other reproduction-related NLRPs in all the studied mammalian genomes (involving cattle, dog, human, chimpanzee, mouse and rat), our phylogenetic result and the published expression profile show that this gene is close to other reproduction-related NLRPs. This suggests that early (before the divergence of mammals) genomic rearrangements might have resulted in the separation of NLRP14 from its relatives.

11We identified the other NLRP members (NLRP1, 3, 6, 10 and 12) as non-reproduction-related NLRP genes, because they are not in the reproduction-related clade in the phylogenetic trees and ubiquitously expressed. These genes have been shown to participate in inflammatory and immune responses by regulating the activation of other signaling factors, such as NF-κB, caspase-1 and IL-1β. In contrast, little is known about the function of the reproduction-related NLRP proteins in inflammation and immunity, except that the human NLRP2, 4 and 7 may be able to inhibit IL-1β and/or NF-κB, and thus may contribute to modulate the inflammatory response. It is known that inflammation and bacterial infection can cause infertility, ectopic pregnancy and abortion. Thus, NLRP proteins might play roles both in innate immunity and in reproductive biology.

12In the reproduction-related NLRP cluster referred by the phylogenetic analysis above, there are several oocyte-specific NLRP genes including NLRP4, 5, 8, 9 and 14. To investigate whether functional divergence have occurred following the duplications of these oocyte-specific NLRPs, four oocyte-expressed NLRP genes (NLRP4, 5, 9 and 14) were estimated for functional divergence by using the DIVERGE program. NLRP8, which was not found in rodents, was therefore not included in this analysis in order to compare the level of divergence across all candidate species.

13After alignment and removal of the sites with gaps, a total of 650 amino acid sites are included in the analysis. Pairwise comparisons of each of the four oocyte-expressed NLRP gene clusters (Additional file 1) were carried out and the rate of amino acid evolution at each sequence position was estimated. As shown in Table 1, functional divergence is significant between each comparison (θ > 0 with p < 0.001) by both MFE and MLE methods, indicating that site-specific altered selective constraints should contribute to the functional evolution of the oocyte-specific NLRP genes after their duplication.

14Furthermore, the important amino acid residues, responsible for functional divergence, were predicted by calculating the site-specific profile based on posterior analysis for all pairs of clusters with functional divergence. In order to extensively reduce false positive, cutoff values, i.e., the minimal posterior probabilities for RFD (Residues with predicted Functional Divergence) were established empirically by progressively removing the highest scoring residues from the alignment until θ dropped to zero. As shown in Table 1, the least RFD (33 residues, covering 5.1% of a total of 650 aligned sites) are observed between NLRP5 and NLRP14, and the most RFD (118 residues, covering 18.2% of a total of 650 aligned sites) are observed between NLRP4 and NLRP14. In general, the RFD are detected in all three functional domains (NACHT, PYRIN and LRRs) of NLRP proteins, implying that shifted functional constrains might have acted on each protein domain.

15In the mouse, there are seven Nlrp4 gene copies (named from Nlrp4a to 4g) which are specifically expressed in oocytes. Our lab has also found the expression of Nlrp4a in the testis by RT-PCR. Nlrp4a-4e genes are located on chromosome 7A, but Nlrp4f and 4g are located on chromosomes 13B3 and chromosome 9 respectively. By the tBLASTn method, we identified six additional putative pseudogenes on chromosome 7A (named here ΨNlrp4h-4m, Figure 3a). Based on the phylogenetic tree (Figure 3a), multiple duplications of mouse Nlrp4 could be depicted as follows. Nlrp4b might be the earliest Nlrp4 copy in the mouse genome with an orthologue present in the rat genome, suggesting an origin before the separation of mouse and rat. The duplication of ancestral Nlrp4b is supposed to have given rise to Nlrp4d, and the subsequent duplications might have resulted in Nlrp4f/4g and Nlrp4a/4c. Subsequent chromosomal rearrangements might have caused the independent locations of Nlrp4f and Nlrp4g with other Nlrp4 duplicates. All the putative Nlrp4h-4m pseudogene sequences show high identities with each other as well as with Nlrp4e. In fact, in all the pseudogenes, a common frameshift of 1-bp deletion was detected at the position corresponding to 358aa of Nlrp4e and it caused a subsequent prematural stop codon at the position of 434aa, suggesting that the pseudogenization had happened before duplications of these copies.

16In the mouse genome, Nlrp4 duplicates (except for Nlrp4f and 4g) are located between 6.00 Mb and 27.32 Mb on the chromosome 7. Interestingly, all these Nlrp4 duplicates are found next to the vomeronasal 1 receptor encoding genes (V1r), such as V1rd and V1re genes (Figure 3a). The phylogeny and genomic location of these V1r genes reveal that V1r genes might have a similar duplication model to their neighbors, Nlrp4 genes (Figure 3a). Thus, one may hypothesize that Nlrp4 may have not duplicated one by one, but were rather involved in segmental duplications encompassing several different genes, such as V1r genes.

17Vomeronasal organ, which is well developed in rodents in comparison with human, detects pheromones and other chemical signals implicated in innate reproductive and social behaviours between the members of the same species. The V1r gene family is supposed to have emerged during the divergence of placental mammals (80-110 MYA), but many subfamilies including V1rd and V1re originated only in rodents after their separation from the primates, which is consistent with the rodent-specific duplication of Nlrp4. Thus, the phylogenetic relationship and genomic location (Figure 3a) of these V1r genes, that are located in the segmental duplication region, could help us to better understand the segmental duplications. Overall, as shown in Figure 3b, a presumed itinerary of segmental duplication was deduced by combining phylogeny with genomic location of both Nlrp4 duplicates and V1r genes. Although we can not confirm the order of duplication events by the present data, we can state that segmental duplication events characterize the evolution of this genomic region.

18Several examples have shown that genes in duplication blocks could be maintained by positive selection. Here, we implemented PAML4 to investigate the model of selective force acting on mouse Nlrp4 and Nlrp9 duplicates during their evolution.

19By using Site Models, estimates of the parameter values under M2a and M8 (Table 2) indicate that a fraction of sites (about 5-6%) may be under positive selection in both datasets (Nlrp4_mouse and Nlrp9_mouse). LRTs of all comparisons (including M1a versus M2a, M7 versus M8 and M8a versus M8) are significant (p < 0.001) for both datasets (Table 2), implying that selective forces varied among amino acid sites and that the mouse Nlrp4 and 9 genes might have been subjected to positive selection after their expansion. Furthermore, the comparison of model MEC (AICc = 15913.11 for Nlrp4_mouse and AICc = 14196.22 for Nlrp9_mouse) versus model M8a (AICc = 16003.68 for Nlrp4_mouse and AICc = 14293.43 for Nlrp9_mouse) also supports positive selection in these two gene clusters. The AICc scores of MEC are lower than those of M8a, indicating that MEC better fits the data, i.e. positive selection is significant.

20Subsequently, we wanted to know if the target of positive selection had changed during evolution in these two gene clades. We compared the potential sites subjected to positive selection between the two datasets (Table 2). In Nlrp4_mouse dataset, 2 sites (133C and 677V) are identified as positively selected sites at p > 95% level by both M2a and M8 models, and 2 more sites (153S and 657N) are identified only by the M8 model. In Nlrp9_mouse dataset, 10 sites (106H, 123 E, 356F, 360L, 378R, 513E, 536V, 571V, 661L and 948A) are identified as potential targets of positive selection by both M2a and M8 models, and 1 more sites (278L) is identified only by the M8 model. We note that no common sites are shared by the two datasets as positively selected sites at p > 95% level, suggesting a dramatic shift in the target of positive selection between the Nlrp4 and Nlrp9 genes in the mouse. When the 3D structures or models of these proteins are available in the future, we could investigate whether these positively selected sites are located in functionally pivotal regions.

21Phylogenetic analyses show that NLRP genes have been originated and duplicated before the divergence of mammals. Certain genes are well conserved during their evolution, such as NLRP3 (82% identity between the human and the mouse), whereas many genes involved in reproduction have rapidly evolved resulting in higher sequence divergence among lineages, such as NLRP5 (50% identity between the human and the mouse) and NLRP14 (62% identity between the human and the mouse). The pairwise estimates of evolutionary divergence between the reproduction-related paralogues are ranged from 61% to 75% (except that the pairwise distance between NLRP2 and its recent primate-specific duplicate, NLRP7, is 45%), which is obviously higher than the one between the non-reproduction-related paralogues (ranged from 43% to 69%). Moreover, gene duplication and/or gene loss is found to have occurred independently in different mammalian lineages. On the one hand, Nlrp4 (originated before the divergence of primates and rodents) and Nlrp9 have been extensively duplicated in rodents, and other lineage-specific gene duplication events concern the specific duplication of Nlrp1 (three paralogues) in the mouse as well as the unique origin of NLRP7 and NLRP11 in primates. On the other hand, Nlrp8 and Nlrp13 are seemed to be lost from the genomes of rodents. Interestingly, the major gene duplication and gene loss events are found to be associated with the reproduction-related genes (such as NLRP4, 7, 8, 9, 11 and 13), implying that reproduction-related genes have undertaken a fast and diverged evolution among different mammalian lineages. Given note that major reproduction-related NLRP genes are germ-cell specific, whereas the other non-reproduction-related genes are usually expressed in multiple tissues from a single organism. Recent large-scale gene expression studies have shown that the tissue specificity of genes is correlated positively with gene evolution rates. Furthermore, our result reinforces the hypothesis that reproduction-related genes are most highly divergent and evolve more rapidly than genes expressed in other organs.

22Gene duplication is thought to be the essential source of gene novelty, with new or altered functions, as shown by widespread existence of gene families. Among NLRP genes, there are at least four oocyte specifically expressed in mammals, including NLRP4, 5, 9 and 14. These genes have been found to be restricted expressed in oocytes and early embryos, suggesting their important roles in oogenesis and preimplantation embryo development. Additionally, their temporal expression in oocytes and degradation during preimplantation development coincide with the timing of gene expression transition from maternal to zygotic control, indicating their roles as maternal effect genes.

23In the present work, by using an in silico method (with the help of the program DIVERGE), we demonstrate that the functional divergence could have occurred between each pair of oocyte-specific NLRP genes. This result is supported by certain published experimental data. For example, the knock-out of mouse Nlrp5 induced female infertility due to a blockage of early embryonic development; and the injection of siRNA against Nlrp14 into fertilized eggs resulted in arrested development of embryos between 1-cell and 8-cell stages. These results suggest that the expression of other oocyte-specific NLRP genes is not able to compensate the absence of neither NLRP5 nor NLRP14 genes. Additionally, recent research has shown that NLRP14 is also expressed in the testis and its mutation might cause spermatogenic failure in the human. Thus, we presume that the expressional divergence of NLRP5 and NLRP14 during the development of mouse germ-cells and embryos may lead to functional specialization. On the other hand, target invalidation/inhibition of NLRP5 and 14 did not reveal any deleterious effect on ovarian folliculogenesis but rather in early embryonic development, suggesting other genes might play the similar roles during folliculogenesis and meiotic maturation. For these latter functions, plausible candidates could be other oocyte-specific NLRP genes, such as NLRP4 and NLRP9. To investigate this possibility, further functional studies such as knock-out models or other targeted inhibition experiments on these genes should be carried out.

24Gene duplication is thought to be a major driving force in enabling the evolution of tissue specialization. By using microarray gene expression data from mammals, fruit flies and yeast, it has recently shown that as gene family size enlarges, there is a general trend for paralogous genes with decreased breadth and increased specificity of expression. In particular, by studying the relationship between gene family size and expression breadth of 1249 protein families in the mouse, Freilich and his colleagues have recently demonstrated that duplicates that arose through post-multicellularity duplication events show a tendency to become more specifically expressed, supporting the view suggested by the subfunctionalization model. In our case, especially with the evidence that NLRP5 and NLRP14 genes are significantly involved in different developmental stages of embryos, we hypothesize that the functional divergence of oocyte-specific NLRP duplicates might be derived by the expression specialization of duplicate genes, which could provide the mammals reproductive advantages such as in adaptation.

25The phylogenetic research indicates the rodent-specific expansion of Nlrp4 and NLRP9, and the furthermore genomic analyses find that these Nlrp duplicates, especially Nlrp4, have been duplicated together with other genes, such as V1r copies. This segmental duplication is not rare during the mouse and human genomic evolution. It has recently revealed that the segmental duplication constitutes about 5% of mouse genome, and its distribution is in a highly non-random fashion. Interestingly, the duplication blocks account for 32% of the first 50 Mb of the chromosome 7, where the segmental duplications concerning Nlrp4 and V1r genes are located. Nlrp9 has been also duplicated in this region, but its duplication scenario is not clear and might be due to subsequent recombination.

26As found in the human, the segmental duplication regions in the mouse might also be "hot spots" for the occurrence of non-allelic homologous recombination, leading to genomic mutations such as deletion, duplication, inversion or translocation. Thus, these instable regions might have a biological significance in form of genome evolution.

27In the mouse, Nlrp4 and Nlrp9 are specifically extensive in gene copies. By using Site-Models implemented in PAML4, we evaluated the variation of selective pressure acting on Nlrp4 and Nlrp9 duplicates, respectively. Our result shows that positive selection is significant in both datasets, which consistent with other reports referring that the rapid divergence of the reproductive genes may be driven by positive selection. In particular, in the mouse, several gene cluster concerning lineage-specific expansion of reproduction-related genes have been found to be under adaptive evolution (positive selection), such as the Sva (seminal vesicle autoantigen) gene cluster, the Prame (or oogenesin) gene cluster, the Psg (pregnancy-specific glycoprotein) gene cluster, the Rhox4 gene cluster and the Svs (seminal vesicle secretion proteins) gene cluster. The lineage-independent expansion and subsequently rapid evolution of such genes might contribute to speciation or adaptation, such as genetic barriers between species and hybrid incompatibilities, or provide the species specification in reproductive processes such as sperm competition, host immunity to pathogens, and manipulation of female/male reproductive physiology and behavior. In the case of NLRP4 and 9 genes, their biological functions are still unclear, so we can not evaluate the significance of their specific duplications in the reproduction of the mouse. However, the similar evolutionary mechanism driven by positive selection suggests that after duplication, these NLRP gene copies diverged and acquired abilities (probably by subfunctionalization) to adapt the new environment. In the subsequent research, the adaptive evolution of the other reproduction-related NLRP genes could be further evaluated. Hopefully, when the biological functions of these NLRP genes are elucidated, the nature of the selective pressure acting on NLRPs will be better understood.

28Interestingly, the evolution of V1r genes, the other genes involved in the segmental duplication with Nlrp4 and 9, is also driven by positive selection in rodents. V1R proteins are thought to be responsible for the detection of pheromones that induce innate reproductive behaviors between members of the same species. Thus, it is presumed that the adaptive evolution of the V1r genes might play an important role in reinforcing pre-zygotic barriers among species of rodents.

29NLRP genes have originated and duplicated mainly before the divergence of mammals. During evolution, NLRPs, especially the reproduction-related NLRP genes, have undergone a fast and independent diversification in different mammalian lineages. The expansion of reproduction-related NLRP genes has been associated with functional divergence after duplication, suggesting that each NLRP oocyte-specific gene might play an essential role in oogenesis and early preimplantation embryo development. The mouse-specific expansion of Nlrp4 and Nlrp9, concerning segmental duplication events, has been driven by positive evolution. Our founding suggests that the duplication and functional divergence of NLRP might provide mammals advantages in reproductive biology.

30The protein sequences of all the 14 known human NLRPs were retrieved from the GenBank http://www.ncbi.nlm.nih.gov/. The NLRP proteins from other species were searched by PSI-BLAST with human NLRP protein sequences as queries against the protein databases (NCBI: RefSeq protein databases) of chimpanzee (Pan troglodytes), dog (Canis familiaris), cattle (Bos taurus), mouse (Mus musculus), rat (Rattus norvegicus) and chicken (Gallus gallus). The predicted coding sequences of the best hit proteins were retrieved when the hits presented more than 80% in length to be aligned with the query sequence (with E values < 10-100). These settings could effectively detect the potential NLRP members from different species but avoid involving the relative NLR proteins from other protein subfamilies. After removal of redundant sequences, the initial data set (Additional file 2) for NLRP phylogenetic studies includes 83 protein sequences from 6 mammals and 1 sequence from chicken.

31Analyses of the orthologous and paralogous relationships among different species were carried out by combining the phylogenetic reconstruction of the gene family with the syntenic comparison. Multiple sequence alignments were performed using the Clustal W algorithm and then manually edited. The alignment was submitted as Online Additional file 3. All alignment gap sites were eliminated before phylogenetic analyses. The phylogenetic trees were reconstructed with Neighbor Joining (NJ), Minimum-Evolution (ME) and Maximum Parsimony (MP) methods implemented in MEGA4, as well as with Maximum Likelihood (ML) method implemented in PhyML. The consensus phylogenetic tree was generated by the fusion of the independent trees reconstructed by the four methods. In all analyses, chicken NLRP protein (Q5F3J4) was treated as the root of all the mammalian NLRPs. The bootstrap values were estimated with 1000 replications. The syntenic comparison is based on Ensembl http://www.ensembl.org/index.html utilities such as "orthologue prediction" and "view syntenic regions".

32The program DIVERGE was used to estimate type I functional divergence between oocyte-expressed NLRP paralogues (NLRP4, 5, 9 and 14). Type I sites represent amino acid residues conserved in one clade (gene cluster) but highly variable in another, suggesting that these residues have been subjected to different functional constraints. Statistically, this functional divergence between two clades is measured by the coefficient of functional divergence, θ, ranging from 0 to 1. A null hypothesis of θ = 0 indicates that the evolutionary rate is virtually the same between two duplicate genes at each site. If the null hypothesis was rejected, a site-specific profile was then used to predict the critical amino acid residues most likely to be responsible for the detected functional divergence.

33A set of 33 protein sequences from mammalian NLRP4, 5, 9 and 14 is included in the analysis. Sequences of rat NLRP4d, 9a and 14 were excluded from the alignment because of their shorter lengths. Before the use of DIVERGE, our sequence dataset was examined to satisfy the conditions recommended by Gu et Vander Velden (2002), which permit us to obtain higher efficiency of detecting functional divergence-related residues: 1) each cluster has more than four amino acid sequences; (2) all pairwise sequence identities are <90%; and (3) multiple alignment is reliable. The phylogenetic tree used for DIVERGE was reconstructed by MEGA4 with ME method. The coefficients of functional divergence (θ) between gene clusters are calculated by Model-Free Estimation (MFE) and Maximum-Likelihood Estimation under two-state model (MLE). To detect amino acid residues reflecting functional divergence, all four reproduction-related NLRP gene clades were compared to each other.

34In order to investigate the predicted segmental duplications concerning Nlrp4 duplicates on the mouse chromosome 7, a tBLASTn search was implemented by using the mouse Nlrp4a (the longest Nlrp4 sequence) as a query against the mouse genome. This method permitted us to identify both expressed Nlrp4 genes and traces of the pseudogenes. Then we combined phylogenetic analyses and genomic mapping to identify the segmental duplication. The phylogenetic tree of mouse Nlrp4 genes and pseudogenes was reconstructed by using nucleotide sequences instead of protein sequences. The phylogenetic tree was also reconstructed for vomeronasal 1 receptor proteins (V1R), because their encoding genes are located paralleling to Nlrp4 duplicates in the mouse genome and they are presumed to have duplicated together with Nlrp4. The accession numbers of the V1R protein sequences used here are listed in Additional file 4. The genomic organization of all the concerned genes and pseudogenes in the mouse is inferred from Ensembl (release 49).

35To examine whether the duplicates of Nlrp4 and Nlrp9 in the mouse have been submitted to adaptive evolution, an analysis of variation in selective pressure following gene duplication events was carried out with the CODEML program implemented in PAML4. The alignments were resulted from Clustal W and PAL2NAL. The shorter mouse Nlrp4g was excluded in order to obtain more informative sites.

36Two datasets named Nlrp4_mouse (including 7 sequences) and Nlrp9_mouse (including 3 sequences) were investigated for different Site Models implemented in PAML4. In this study, we employed three pairs of models including M1a (NearlyNeutral: 0< ω0 <1 and ω1 = 1) versus M2a (PositiveSelection: 0< ω0 <1, ω1 = 1 and ωs >1), M7 (beta: 0< ω <1) versus M8 (beta and ω: 0< ω <1 and ωs >1), and M8a (beta &ωs = 1: 0< ω <1 and ωs = 1) versus M8. LRTs (Likelihood Ratio Tests) were used to test for significant differences in the fit of the models incorporating selection relative to their (nested) counterparts that did not allow positive selection. These tests provide a useful series of metrices for interpreting the significance of the results and a degree of protection against false positives. Bayes empirical Bayes (BEB) method implemented in PAML4 was used to estimate posterior probabilities of selection on each codon. Furthermore, a newly described model MEC was also employed on the Selecton Server to compare with the results from other models for positive selection. The advantage of the MEC model over the other models used here is that by treating specifically each amino-acid replacement, Ka is computed differently. So under the MEC model, a position with radical replacements will obtain a higher Ka value than a position with more moderate replacements.

37XT performed the main data collection and analyses. GP helped to guide in the bioinformatic analyses. PM conceived the study and helped to guide in the general analyses. All coauthors participated in manuscript preparation and review.