Expression Analysis of the NLRP Gene Family Suggests a Role in Human Preimplantation Development

Background The NLRP (Nucleotide-binding oligomerization domain, Leucine rich Repeat and Pyrin domain containing) family, also referred to as NALP family, is well known for its roles in apoptosis and inflammation. Several NLRPs have been indicated as being involved in reproduction as well. Methodology We studied, using the unique human gametes and embryo materials, the expression of the NLRP family in human gametes and preimplantation embryos at different developmental stages, and compared the expression levels between normal and abnormal embryos using real-time PCR. Principal Findings Among 14 members of the NLRP family, twelve were detected in human oocytes and preimplantation embryos, whereas seven were detected in spermatozoa. Eight NLRPs (NLRP4, 5, 8, 9, 11, 12, 13, and 14) showed a similar expression pattern: their expression levels were high in oocytes and then decreased progressively in embryos, resulting in a very low level in day 5 embryos. However, NLRP2 and NLRP7 showed a different expression pattern: their expression decreased from oocytes to the lowest level by day 3, but increased again by day 5. The expression levels of NLRP5, 9, and 12 were lower in day 1 abnormal embryos but higher in day3 and day5 arrested embryos, when compared with normal embryos at the same stages. NLRP7 was down-regulated in day 1 and day 5 abnormal embryos but over-expressed in day3 arrested embryos. Conclusions According to our results, different NLRPs possibly work in a stage-dependent manner during human preimplantation development.

Several NLRPs have been found to be related to reproduction. Down-regulation of NLRP expression is connected with oocyte aging in mice [5]. NLRP5 (also known as Mater) null female mice are sterile because their embryos arrest at the two-cell stage [6]. Another NLRP member in mice, NLRP iota, is also required for normal preimplantation development. Injection of siRNA against NLRP iota into fertilized eggs results in arrested development of embryos between 1-cell and 8-cell stages [5]. A mutation in human NLRP7 has been found to be associated with recurrent hydatidiform moles, spontaneous abortions, stillbirths and intrauterine growth retardation [7,8]. Additionally, over-expression of NLRP7 is related to development of testicular seminomas in humans [9]. NLRP14 mutation was discovered in five of 157 men with azoospermaia or severe oligozoospermia [10]. Transcripts of NLRP5, NLRP8 and NLRP9 [11], and the protein of NLRP5 [12], have been detected in bovine oocytes and preimplantation embryos.
We reported earlier the expression of NLRP5 in human fullygrown GV oocytes [13]. Here, we studied, using the unique human gametes and embryo materials, the expression of all the 14 NLRP members in human gametes and preimplantation embryos at day 2 (D2), day3 (D3), day 5 (D5), and compared their expression levels between normal and abnormal embryos. According to our results, different NLRPs possibly work in a stage-dependent manner during early human development.

Selection of PSMB6 as an internal control gene
In order to compare gene expression levels across several developmental stages, it is essential to relate the test gene to control genes (''housekeeping genes'') that are stably expressed at the different developmental stages. We started by studying the expression of two commonly used internal control genes: betaactin and GAPDH in human oocytes and embryos, but found that the expression levels of these two genes in human oocytes and embryos were not stable enough to act as good internal controls. The instability of beta-actin and GAPDH has also been observed in both bovine and mouse oocytes and embryos [14,15]. Hence, we then selected housekeeping/maintenance genes from the list provided by Affymextrix. They have tested 11 different human adult and fetal tissues and found 47 transcripts expressed at the same level in all the tissues [16]. PSMB6 and EEF1A1 are listed among these 47 genes. Our microarray data (Zhang et al. unpublished) showed that both were expressed at similar levels in human oocytes and embryos. We verified the results by testing the expression of PSMB6 and EEF1A1 at five developmental stages in oocytes (GV, MI, MII) and embryos (D2, D3, D5). Real-time PCR showed the same pattern for PSMB6, but not for EEF1A1. The average Ct value of PSMB6 in individual oocytes and embryos was 29.0760,79 (Mean6SD). PSMB6 was therefore chosen as a stable internal control in real-time PCR when human oocytes and/or embryos are used, especially when the tested genes have similar expression levels as PSMB6.
Expression of NLRP family genes in normal development from oocytes to day 5 embryos We used Affymetrix arrays to monitor global gene expression during five stages of early human development (Zhang et al., unpublished). The results suggested that almost all NLRP family genes were expressed during development. In order to verify and extend the findings, we performed quantitative RT-PCR assays for each NLRP gene. The specificity of the PCR product was confirmed by both dissociation curve and gel electrophoresis. A single peak was observed in all the dissociation curves and a single band was seen on the gel at the correct product size for each gene (Fig. 1).
Among the ten NLRPs (2,4,5,7,8,9,11,12,13,14) that were detected in all the oocytes and embryos, microarrays showed two different expression patterns (Fig 2A and B). Except for NLRP2 and NLRP7, the rest were highly expressed in oocytes and then gradually decreased in embryos with a very low level in D5 embryos. The expression of NLRP2 and NLRP7 followed a similar pattern up to D3, but then showed a sharp increase in D5. These expression patterns were consistent with those revealed by realtime PCR (Fig. 3A and B).

Expression of NLRPs in developmentally arrested embryos
To figure out the possible roles of NLRPs in oocyte maturation, fertilization and early embryonic development, we compared the expression of NLRPs between normal and abnormal embryos. As we were unable to acquire normal day 1 (D1) embryos, we compared D1 abnormal embryos with normal D2 embryos (see explanation in discussion part). NLRP5 and NLRP9 showed 5 times lower expression levels in unfertilized oocytes than in normal D2 embryos, while NLRP12 and NLRP7 showed 3 times and 2.5 times lower levels respectively. In 1PN and 3PN embryos, the expression of NLRP9 NLRP3 was only detected in GV, MI oocytes and spermatozoa. NLRP5, 8,9,13 were detected in all oocytes and embryos, but not in spermatozoa. NLPR6 was only detected in spermatozoa and NLRP10 was not detected in any gametes or embryos (figure not shown). doi:10.1371/journal.pone.0002755.g001 decreased 3 times when compared with normal D2, whereas the decrease of NLRP5, 12, and 7 was not significant (Fig. 4).
We also compared the expression levels of NLRP5, 9, 12 and 7 between development-arrested embryos and normal ones at D3 and D5. All four genes had a higher expression level in development-arrested embryos at D3 than in normal ones. Their expression levels increased 1.5, 2.0, 3.4 and 1.3 times respectively (Fig. 5A). In D5 development-arrested embryos, NLRP5, NLRP9 and NLRP12 had a higher expression level than in normal D5 embryos. The fold change was 32, 29 and 17 respectively. Oppositely, the expression of NLRP7 was 50 times lower in development-arrested D5 embryos (Fig. 5B).

Phylogenetic analysis of NLRP genes in mammals
To further predict the structure and function of the NLRP family, we did phylogenetic analysis of the NLRP family using protein sequences from both mice and humans. The results (Fig. 6) showed a high homology of NLRP members between mouse and human, except for NLRP8. The cluster of the NLRP members by the phylogenetic analysis fitted well with their expression patterns in human oocytes and embryos, such as NLRP2 and NLRP7. Another cluster was NLRP1, 6, 10, 3. All 4 of these genes were expressed at relatively low levels in oocytes and embryos. Notably, the gene closest to NLRP5 was NLRP13, suggesting that NLRP13 might have a similar function to NLPR5. It would be interesting to further test the function of NLRP13, both in mice and humans.

Discussion
For the first time, we report here a comprehensive expression picture of the NLRP family in human gametes and preimplantation embryos. Their expression patterns were confirmed by both microarray and real-time PCR. We found downregulation of NLRP5, 9, 12 and 7 in D1 abnormal embryos and upregulation of NLRP5, 9, and 12 in development-arrested embryos both at D3 and D5. NLRP7 showed upregulation in D3 arrested embryos but downregulation in D5 when compared with normal embryos at the same stages.
It is not easy to obtain human oocytes and embryos for research, but in our large in vitro fertilisation programme, we have ethics approvals to get immature oocytes, which cannot be used for intracytoplasmic sperm injection, and mature them in vitro to metaphase II. We can also get cryo-preserved human embryos that the couples do not want to use for infertility treatment any more. Clearly abnormal embryos are also available for research in some occasions. Any functional experiments using these scarce human oocytes and embryos are not feasible. We could, however, using microarrays, real-time PCR, extensive bioinformatics  methods, and comparative studies with several animal species, obtain a large amount of really new information regarding the function of these genes in human early development.

NLRP5 and NLRP9
NLRP5 (Mater) is essential for mouse early embryonic development beyond the two-cell stage [6]. In mice, transcripts of NLRP5 are detected in oocytes from the primary follicle onwards, but they are not detectable in any stages of preimplantation embryos. NLRP5 protein is present until the late blastocyst stage in mice [17]. In bovine, the transcripts of NLRP5 have the same expression pattern as we found in humans [12]: NLRP5 is expressed at a high level in oocytes, and remains detectable after fertilization. It declines in preimplantation embryos and becomes hardly detectable in blastocysts.
The expression of NLRP 9 has been thought to be restricted to oocytes and the testis because it was not detected in somatic tissues in bovine [11,18]. NLRP9 is also found exclusively expressed in oocytes in mice [19]. According to public databases, NLRP9 is also expressed in lung, placenta, thymus, intestine, brain and prostate. We observed that NLRP9 declined gradually from oocytes to blastocysts. Consistent with our study, similar expression patterns are also reported by two other studies on bovine [11,18].
NLRP9 had a similar expression picture as NLRP5 in both normal and abnormal embryos. The decrease of NLRP9 in D1 abnormal embryos implies its possible roles in the fertilization and zygote development; while the increase of NLRP9 in D3, D5 arrested embryos suggests it is probably not needed in normal D3 and D5 embryos.

NLRP7 and NLRP14
Opposite to other NLRPs, the expression of NLRP7 did not decrease in D5 embryos. It increased from D3 to D5, and was downregulated in developmentally-arrested D5 embryos. All these results suggest its potential role in later embryonic development in humans. This hypothesis is supported by the association of NLRP7 mutation with abnormal embryo development in humans [8]. The link between overexpression of NLRP7 and tumor development suggests that NLRP7 participates in cell proliferation and/ or cell differentiation. The involvement of NLRP7 in inflammation (where NLRP7 is referred to as PYPAF3) implies possible participation of NLRP7 in other pathways [20].
NLRP14 has been thought to be a testis-specific gene in humans and it might participate in inflammation as well [10]. However, mouse NLRP14 protein is found in oocytes at all stages of follicles, except primordial follicles; whereas no signal is detected in testis in the same study [21]. In our study, NLRP14 was detected in oocytes, spermatozoa and embryos. It shared a similar expression pattern with NLRP5 and NLRP9. The inconsistency in the results for NLRP14 could be due to species variance, or due to the diverse nominating of the same genes.

NLRP12
NLRP12 is also known as PYPAF7. The expression of NLRP12 has been reported to be highly restricted to immune cells, there NLRP12 activates inflammatory signaling pathways [22,23]. So far, there is no literature showing any relationship between NLRP12 and reproduction. The expression levels of NLRP12 in oocytes, spermatozoa and embryos were relatively low when compared to that of NLRP5. The expression pattern of NLRP12 was similar to that of NLRP5 and NLRP9, both in normal and abnormal embryos. Similar to NLRP9, NLRP12 is most likely involved in the fertilization and zygote development, but in D3, D5 development.

The indirect comparison between D1 abnormal embryos and D2 normal embryos
Although the comparison between D1 abnormal embryos and D2 normal embryos was not direct, we could still show that there was a downregulation of NLRP5, 7, 9 and 12 in D1 abnormal embryos. According to the overall expression patterns, D1 embryos should contain a higher level of NLRPs than D2 embryos. Here we saw the opposite trend instead. It suggests that the expression level of NLRP5, 7, and 9 in D1 abnormal embryos was much lower than in normal D1 embryos. There was a downregulation of these four NLRPs in D1 abnormal embryos. We can also exclude the possibility that this downregulation is due to degradation of mRNA in abnormal embryos by the observation that NLRP 5, 9,12 hade very high expression levels in D3 and D5 abnormal embryos. If there was a degradation of mRNA in abnormal embryos, the expression levels of these genes in abnormal D3 and D5 embryos would have been much lower.
Overall, the information presented in our study provides clear clues for better understanding of the roles of all NLRP family members in human reproduction. The changed expression of NLRPs in abnormal oocytes and embryos further suggests that they have potential roles in preimplantation development in humans.

Collection of human oocytes and embryos
This study was approved by the Ethics Committees of Karolinska Institute, Karolinska University Hospital Huddinge and Ö rebro University Hospital. Informed consent was obtained from all the oocyte, sperm and embryo donors. None of the donors received any financial reimbursement. The researchers of this study did not participate in obtaining consent.
Morphologically normal fully-grown germinal vesicle (GV) and metaphase I (MI) oocytes were donated by healthy women undergoing intracytoplasmic sperm injection (ICSI) treatment due to male factor infertility as described elsewhere [13]. Such oocytes Only embryos that could not be used in the infertility treatment were used for this study. They were from either conventional in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI). They had been cryopreserved at the 2-8-cell stage using a threestep propanediol cryopreservation kit (Freeze kit 1; Vitrolife AB, Gothenburg, Sweden). In this study, these embryos were then thawed (Sydney IVF thawing kit, CooK IVF, Brisbane, Australia) and used at the 4-cell (D2) or 8-cell (D3) stage, or further cultured to the blastocyst stage (D5) in either BlastAssist System (Medicult, Jyllinge, Denmark) or blastocyst sequential media (Sydney IVF Blastocyst medium, CooK IVF, Brisbane, Australia).
Abnormal oocytes and embryos were collected from the clinic on the day they were to be discarded. They included unfertilized oocytes at day 1 (Unfert. oocyte), fertilized oocytes with 1 pronucleus at day 1 (1PN), fertilized oocytes with 3 pronuclei at day 1 (3PN), and embryos that stopped developing (stop dev.) at D3 and D5.
Oocytes and embryos were stored in RNAlater at 270uC. Prior to cDNA synthesis, they were thawed at room temperature in RNAlater, then moved to RNase-free PBS and washed three times before being frozen in 2 ml RNase-free H 2 O. Three to five oocytes or embryos were pooled together as one sample. Re-thawed oocytes and embryos were ready for cell-direct cDNA synthesis.

Collection of human spermatozoa
Semen samples were donated by healthy males who underwent IVF treatment due to female infertility. Spermatozoa were obtained after standard preparation from ejaculated semen samples. Swimup technique was used for spermatozoa preparation in all cases. Concentration and motility of spermatozoa were measured using Crismas computer spermatozoa analyzer (Image House, Denmark). Only samples with normal concentration and spermatozoa motility between 95 and 100% in final suspension were used for this study. After being washed three times in RNase-free PBS, the spermatozoa were transferred to 350 ml of RNA lysis buffer (Qiagen, Hilden, Germany) and stored at 270uC until RNA extraction.

Real-time PCR (RT-PCR)
Materials used for real-time were different from those used in microarray. Experiments were repeated three times for each sample type using different materials.
Oocyte and embryo cDNA was synthesized using Super-Script TM III CellsDirect cDNA Synthesis system (Invitrogen, Stockholm, Sweden). Spermatozoa cDNA was synthesized with SuperScript TM III First-Strand Synthesis system for RT-PCR (Invitrogen, Stockholm, Sweden) after RNA extraction using RNeasy mini kit (Qiagen, Hilden, Germany). Oligo-dT was used for reverse transcription for all the samples. The total volume of reverse transcription is 21 ml. All the primers (Table 1) are designed by Primer Press 3.0, software for optimal primer/probe design provided by Applied Biosystems. All primer pairs span intron-exon boundaries. SYBR Green technology was applied for all the assays with ABI 7500 standard qPCR system (Applied Biosystems, Foster City, CA). The total reaction volume was 10 ml, including 5 ml 2X Power SYBR Green PCR master mix (Applied Biosystems, Warrington, UK), 0.4 ml of 5 mM forward primer, 0.4 ml of 5 mM reverse primer, 1 ml cDNA template, and 3.2 ml H 2 O. Negative controls without template were added each time. Proteasome subunit, beta type 6, also named Proteasome subunit Y (PSMB6), was used as an internal control. The PCR program started with 50uC, 2 minutes; 95uC, 10 minutes; followed by 40 cycles of 95uC, 15 seconds; 60uC, 50 seconds; and ended with 95uC, 15 seconds; 60uC, 1 minute; 95uC, 15 seconds. The ending step of PCR is performed to acquire the dissociation curve, validating the specificity of the PCR products. Comparative method was used for fold-change calculation. The PCR products were taken to run agarose gel (2.5%; stained with 50 _g/mL ethidium bromide). A 50 -base pair DNA ladder (Invitrogen) was used as the size marker.

Phylogenetic analysis of the NLRP family
Multiple alignment of the sequences was performed with the software ClustalX [24], while the trees where produced with TreeView 1.6.6 [25]