Phylogenetic Analysis of EPPIN and SEMG1

The Catarrhini (old world monkeys, great apes, gibbons, and humans) emerged 43.5 million years ago (MYA) from a common ancestor and subsequently diverged into the Hominoidea and Cercopithecoidea 31.6 MYA [26]. The Homininae emerged 8.3 MYA and underwent rapid speciation into three genera (Gorilla, Pan, Homo); the Gorilla branching rapidly away from the Homo-Pan genera [26]. We examined the molecular phylogeny of EPPIN and SEMG1 within the Hominoidea and of Macaca mulatta and Papio anubis within the Cercopithecoidea and found 2.2 more substitutions per site in the SEMG1 lineage (Figure 1A, B). The Maximum Likelihood method gave the same tree for EPPIN (Figure 1A) as all other methods and conformed to the expected evolutionary tree for primates [26]. However, using the same methodology, all the methods gave an unexpected tree for the SEMG1 sequence (Figure 1B) showing the Gorilla branching from the common ancestor of the Cercopithecoidea rather than the common ancestor of the Homininae. A test of the homogeneity of substitution patterns between sequences revealed that the Gorilla EPPIN sequence exhibits heterogeneity with Homo, Pan and Pongo (Table 1, asterisks in row 1 vs. column 3, row 2 vs. column 3, and row 3 vs. column 4), whereas the Gorilla SEMG1 sequence exhibits heterogeneity with Macaca mulatta (Table 2, asterisks in row 3 vs. column 6) and, although not statistically significant, a trend towards heterogeneity with Homo sapiens (Table 2, trend symbol in row 1 vs. column 3, row 3 vs. column 6). Although there was an overabundance of non-synonymous substitutions in several codons in each lineage, only SEMG1 showed positively selected codons that were statistically significant (Table S1 and Table S2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Molecular phylogenetic analysis for EPPIN and SEMG1 within the primate lineage.

EPPIN (A) and SEMG1 (B) trees were drawn to scale with branch lengths measured in the number of substitutions per site. GenBank accession number for each DNA sequence used for EPPIN and SEMG1 analyses are shown in Table 6.

https://doi.org/10.1371/journal.pone.0082014.g001

We found no evidence of positive selection in any site within the primate phylogenetic tree of EPPIN (Table S1). On the other hand, detailed examination of the full-length EPPIN sequence revealed that when the Hominoidea and Cercopithecoidea Superfamilies emerged from the Catarrhini 31.6 MYA the Hominoidea underwent a distinct change in seven amino acid residues (Lys35, Lys50, Asp51, Gln55, Lys81, Tyr89, Leu91, Met104, and Lys120) that appears to be a directional selection throughout the Hominidae and Hylobatidae Families (Table 3), separating them from the Papionini. An examination of the EPPIN sequence within the SEMG1 binding site in the Kunitz domain [23] revealed only two residue changes that were different between Homo and all the other Hominoidea, namely His92 and Asp99. Only His92 projects into the binding pocket [23] and would represent a significant change from Arg92 in charge and size.

Analysis of the SEMG1 sequence in fragment 74-8 (R165-Q247; Figure 2), a fragment still retaining EPPIN binding and sperm motility inhibitory activities (see below), revealed that the Hominoidea underwent a distinct change in four amino acid residues (Trp167, His169, Ser193, and Gln215), which might be indicative of a directional selection in Hominidae (Table 4). Additionally, Bayes Empirical Bayes (BEB) analysis [27] revealed eleven positively selected codons in this SEMG1 sequence (Table 4; Table S2); only two (Trp167, His169) were throughout the Hominoidea. Although there were no amino acid residue differences between Homo sapiens and all other Hominoidea, residues Asp225 and Leu234 in the Gorilla were distinct from other Hominidae and Hylobatidae residues, Glu225 and Val234 respectively, but identical to Macaca mulatta and Papio anubis residues (Table 4). Residue His239 was unique in the Gorilla SEMG1 in comparison to SEMG1 from other Hominidae and Hylobatidae (Cys239), as well as Macaca mulatta and Papio anubis (Arg239). Residue Cys239 and His169 were reported previously to be sites under positive selection [18] and our BEB analysis confirmed that the Cys239 codon underwent positive selection (Table 4; Table S2). Homo sapiens Cys239, which is necessary for SEMG1 binding to EPPIN [24,25], underwent a distinct change when the Hominidae and Hylobatidae Families separated from the Cercopithecidae (R239C). Consequently, based on our phylogenetic analysis, it is likely that the positive selection of eleven of SEMG1’s codons within the WFDC locus of the Hominoidea and the co-evolution of EPPIN’s His92 resulted in the high affinity binding of SEMG1 to EPPIN in humans.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Human semenogelin I (SEMG1) sequence.

A) SEMG1 primary sequence. Signal peptide is indicated by a dashed underline. B) Schematic representation of SEMG1 containing repeats Ia, Ib, IIa, IIb, IIIa, and IIIb. Recombinant SEMG1 fragments used in AlphaScreen assay and CASA experiments: G26-R281 (SEMG1^214-42), R165-R281 (SEMG1^74-42), R165-Q247 (SEMG1^74-8), Q207-R281 (SEMG1^32-42), Y220-L262 (SEMG1^19-22), E229-L269 (SEMG1^10-30), and Q207-Q247 (SEMG1^32-8) are shown. Cysteine residue at position 239 is indicated by a box in (A) and a vertical arrow in (B).

https://doi.org/10.1371/journal.pone.0082014.g002

Binding of SEMG1 fragments and Cys239 mutants to EPPIN

As we reported previously [23], the binding of human recombinant SEMG1^214-42 fragment to EPPIN was time-dependent (Figure S1) and saturable (Figure 3A). Experiments performed using increasing concentrations of SEMG1^214-42 (0.04 - 7.5 µM) in the presence of a constant concentration of EPPIN indicated a concentration-dependent increase in the signal up to 300 nM SEMG1^214-42 (Figure 3A and Figure S1). At higher SEMG1^214-42 concentrations, we observed a progressive drop in the signal due to oversaturation of Ni-NTA-chelate donor beads with SEMG1^214-42 (the hook effect [28], Figure S1). In order to determine the smallest region within SEMG1^214-42 critical for EPPIN binding, we cloned and expressed six SEMG1 truncated fragments lacking N-terminal and/or C-terminal sequences flanking Cys239 residue, namely SEMG1^74-42, SEMG1^74-8, SEMG1^32-42, SEMG1^19-22, SEMG1^10-30, and SEMG1^32-8 (Figure 2B). We designed these truncations because they correspond to SEMG1 fragments containing the Cys239 residue generated by PSA digestion [6] and are part of the 15-kDa SEMG1 fragment protected from PSA cleavage by EPPIN [7]. Thus, it is likely they are naturally generated during the semen liquefaction. Among these fragments, only SEMG1^74-42 and SEMG1^74-8 bound to EPPIN, albeit at much higher concentrations (1 - 7.5 µM) in comparison to the largest fragment SEMG1^214-42 (Figure 3A). We observed no saturation when SEMG1^74-42 and SEMG1^74-8 fragments were titrated in the presence of EPPIN under similar conditions (Figure 3A).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. Binding of recombinant SEMG1 fragments to EPPIN in the AlphaScreen assay.

A) Concentration-response curve for SEMG1^214-42, SEMG1^74-42, SEMG1^74-8, and SEMG1^32-8 in the presence of a constant concentration of EPPIN. B) Displacement of bt-SEMG1^214-42 from its binding site on EPPIN by SEMG1 constructs. Specific signal for each data point was normalized as percentage of specific signal in the absence of SEMG1 competitors. Data are expressed as mean ± SD from two independent experiments, each performed in four replicates. C) Concentration-response curve for SEMG1^214-42 (wild-type) and Cys239 mutants (C239G, C239D, C239H, C239R, C239S) in the presence of a constant concentration of EPPIN. Data points in (A) and (C) represent mean ± SD of specific signal from a representative experiment of (A) four (SEMG1^214-42 and SEMG1^74-42) and two (SEMG1^74-8) experiments; and (C) three experiments, each performed in four replicates. cps = counts per second.

https://doi.org/10.1371/journal.pone.0082014.g003

To confirm these results, we performed competition experiments to investigate the capacity of SEMG1 fragments to compete with biotinylated (bt)-SEMG1^214-42 for its binding site on EPPIN. SEMG1^214-42 decreased bt-SEMG1^214-42/EPPIN interaction in a concentration-dependent manner (Figure 3B) with a calculated IC₅₀ value (95% confidence interval) of 8.1 nM (6.8-9.7 nM). Both SEMG1^74-42 and SEMG1^74-8 inhibited the binding of bt-SEMG1^214-42 to EPPIN at higher concentrations in comparison to SEMG1^214-42 (Figure 3B). The calculated IC₅₀ values were 123.7 nM (92.4-165.7 nM) and 384.7 nM (104.2-1421 nM) for SEMG1^74-42 and SEMG1^74-8, respectively. Curiously, SEMG1^32-8 competed with bt-SEMG1^214-42 for binding EPPIN at even higher concentrations (Figure 3B), with an IC₅₀ of 5.1 µM (0.3-75.2 µM).

Next, we measured the binding of the SEMG1^214-42 Cys239 point mutants to EPPIN in the AlphaScreen assay. All point mutations negatively affected SEMG1/EPPIN binding as demonstrated by right-shifted concentration-response curves (Figure 3C) and consequently increased EC₅₀ values (Table 5) in comparison to the wild-type when SEMG1^214-42 mutants were incubated with EPPIN under similar conditions. We observed a ~4-fold, ~8-fold and ~20-fold increase in the EC₅₀ values of SEMG1/EPPIN binding when glycine/aspartic acid, histidine or serine/arginine residues replaced the Cys239 residue, respectively (Table 5). Altogether, we confirmed previous observations that the Cys239 residue within human SEMG1 primary sequence plays a critical role in the SEMG1/EPPIN interaction. Additionally, our results indicated that sequences within SEMG1 repeats IIIa, IIa and IIb flanking Cys239 residue may provide interacting interfaces that significantly enhance SEMG1/EPPIN interaction.

Effect of SEMG1 fragments and Cys239 mutants on sperm motility

We have previously demonstrated that the recombinant SEMG1^214-42 fragment inhibited different sperm movement-related parameters, including motility, progressive motility and VSL (straight line velocity) of human ejaculate spermatozoa in a time- and concentration-dependent manner [24]. To determine the minimum sequence within SEMG1^214-42 that inhibits sperm motility, we incubated spermatozoa with different recombinant SEMG1 fragments (Figure 2) and assessed by Computer-assisted sperm analysis (CASA). Based on observations that both sperm motility and velocity are directly correlated to the fertilizing ability of human spermatozoa [29] and that SEMG1 forms a gelatinous mass after ejaculation restricting sperm movement, we developed an index (iRMI) correlating both %motility and VSL parameters, which allowed the normalization of each data set against its respective control, reducing inter-assay variation due to differences in sperm quality in different semen samples.

As expected, SEMG1^214-42 at a concentration of 1.5 fmol/sperm (2 µM) significantly decreased the normalized iRMI in comparison to control group (Figure 4A). Surprisingly, all SEMG1 fragments tested significantly affected sperm motility, including those that did not bind EPPIN in the AlphaScreen assay, as demonstrated by a decrease in the normalized iRMI equivalent to that observed with SEMG1^214-42 at the same concentration (Figure 4A). To investigate the nature of these results, we performed concentration-response experiments in which spermatozoa were incubated with increasing concentrations of recombinant SEMG1^214-42, SEMG1^74-42 and SEMG1^32-8 fragments under similar conditions and the normalized iRMI was then calculated for each concentration point (Figure 4B). All fragments tested decreased the normalized iRMI in a concentration-dependent fashion (Figure 4B). The calculated EC₅₀ values (95% confidence interval) were 1.69 µM (0.77 - 3.7 µM), 3.56 µM (1.67 - 7.63 µM), and 1.75 µM (0.50 - 6.06 µM) for SEMG1^214-42, SEMG1^74-42 and SEMG1^32-8 fragments, respectively. Based on these results, we conclude that a region within the sequence E229-Q247 of SEMG1 contains a domain responsible for the inhibition of sperm motility.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Effect of recombinant SEMG1 constructs on human sperm motility.

Spermatozoa were incubated with the same concentration of recombinant SEMG1 fragments (A) or Cys239 mutants (C) and assessed by CASA. Concentration-response curves for spermatozoa incubated in the presence of increasing concentrations of SEMG1^214-42, SEMG1^74-42, and SEMG1^32-8 fragments (B) or C239H-SEMG1^214-42 and C239R-SEMG1^214-42 (D) were obtained. Results are presented as normalized iRMI (%motility*VSL of sample/%motility*VSL of control). Control untreated experiments were performed with buffer only. Data are representative of mean ± SD of independent experiments performed with ejaculate spermatozoa from six (A) or three (B, C and D) donors. *p < 0.05, ** p < 0.01, and *** p < 0.001 when compared to control group (ANOVA followed by Bonferroni test).

https://doi.org/10.1371/journal.pone.0082014.g004

CASA experiments performed with SEMG1^214-42 Cys239 mutants demonstrated that only C239H-SEMG1 mutant (His239 in the Gorilla, Table 4) significantly decreased the normalized iRMI in a similar extent to the wild-type at a concentration of 0.5 fmol/sperm (Figure 4C, left panel). When the concentration was increased to 1.5 fmol/sperm all the SEMG1 mutants, except C239G-SEMG1, significantly decreased the normalized iRMI in comparison to control (Figure 4C, right panel). To confirm the concentration-dependent sperm motility inhibition of the SEMG1 Cys239 mutants, we performed concentration-response experiments using C239H-SEMG1 and C239R-SEMG1 (Figure 4D). Both mutants decreased the normalized iRMI in a concentration-dependent manner (Figure 4D). The calculated EC₅₀ values (95% confidence interval) were 1.08 µM (0.41 - 2.84 µM) and 2.21 µM (1.14 - 4.29 µM) C239H-SEMG1 and C239R-SEMG1, respectively.

Phylogenetic analysis

We aligned nucleotides sequences of EPPIN and SEMG1 from Homo sapiens, Pan troglodytes, Gorilla gorilla gorilla, Pongo abelii, Nomascus leucogenys, Macaca mulatta, Papio anubis, and Saimiri boliviensis boliviensis (GenBank accession numbers in Table 6) with ClustalW. We used MEGA5 molecular evolutionary genetic analysis software package (version 5.1) [36] to obtain phylogenetic trees using the Maximum Likelihood method based on the Tamura-Nei model [37] and to perform the test of homogeneity of substitution patterns between sequences for EPPIN and SEMG1 using the neighbor-joining method [38]. Codon positions included were 1st+2nd+3rd. All positions containing gaps and missing data were eliminated. For technical details of these tests, please refer to their original descriptions [36-38]. We chose the phylogenetic trees with the highest log likelihood (EPPIN) -926.2396 or (SEMG1) -2108.0861 and used them for further analysis. We investigated evidence of positive selection codon-by-codon in the primate phylogenetic tree of EPPIN and SEMG1 using the CodeML software in the PAML 4 package [27,31]. In this case, we compared the likelihood of a nearly-neutral model M1a with that of a selection model M2a. Next, we compared the likelihood of a neutral model M7 with that of a selection model M8. In both cases we used a Likelihood Ratio Test (LRT) to determine if the selection model was a better fit to the data than the neutral model by comparing the (-2[Log(neutral) – Log(positive)]) value between the two models with the χ²-distribution [18,27]. When the calculations suggested positive selection, we used the BEB method to calculate the probability that each codon experienced positive selection under models M2a and M8 [27].

Production and Expression of Recombinant Proteins

Human recombinant SEMG1 (SEMG1^214-42, wild-type, residues G26-R281, Figure 2) and EPPIN (residues P22-P133) were previously cloned into pET100/D-TOPO vector (Invitrogen, Carlsbad, CA) [23]. Additionally, C-terminal and N-terminal truncated fragments flanking the residue Cys239 from the SEMG1 primary sequence, corresponding to residues R165-R281 (SEMG1^74-42), R165-Q247 (SEMG1^74-8), Q207-R281 (SEMG1^32-42), Y220-L262 (SEMG1^19-22), E229-L269 (SEMG1^10-30), and Q207-Q247 (SEMG1^32-8) (Figure 2B) were produced and cloned into the same vector. All constructs were verified by DNA sequencing. Recombinant proteins were expressed either in BL21 Star^TM (DE3) One Shot^® E. coli (Invitrogen; SEMG1 constructs) or E. coli Rosetta-Gami 2^TM (DE3) (Novagen, Madison, WI; EPPIN construct) and purified under denaturing conditions using nickel-nitrilotriacetic acid agarose (Ni-NTA) beads (Qiagen, Valencia, CA) as described previously [23].

Removal of N-terminal 6X His-tag from recombinant EPPIN was performed using recombinant enterokinase cleavage/capture kit (Novagen), according to the manufacturer’s instructions. For removal of undigested 6X His-tag-EPPIN, the cleavage reaction was incubated with Ni-NTA beads at room temperature for 30 min with gentle mixing and tag-free EPPIN was then eluted in PBS buffer containing 1 M urea. Recombinant SEMG1^214-42 was biotinylated as described previously [23].

In Vitro Site-Directed Mutagenesis

Site-directed mutagenesis of wild-type SEMG1^214-42 of residue Cys239 to glycine (C239G-SEMG1), aspartic acid (C239D-SEMG1), histidine (C239H-SEMG1), serine (C239S-SEMG1), or arginine (C239R-SEMG1) was performed using the Gene Tailor site-directed mutagenesis system (Invitrogen), according to the manufacturer’s instructions. Mutagenesis products were transformed into DH5a-T^1R E. coli and positive clones selected, followed by DNA sequencing to confirm the mutation. Recombinant SEMG1 mutants were expressed and purified as described above.

AlphaScreen^® Assay

The AlphaScreen^® assay was carried as previously described [23], with the following modifications. Briefly, recombinant EPPIN was pre-incubated with anti-EPPIN Q20E antibody and Protein A acceptor beads for 30 minutes. In parallel, increasing concentrations of each recombinant SEMG1 construct was incubated with Ni-NTA-chelate donor beads under the same conditions. Equal volumes of each EPPIN/Q20E/Protein A acceptor beads and SEMG1/Ni-NTA-chelate donor beads were pipetted into in white opaque 384-well microplates (OptiPlate-384; PerkinElmer, Waltham, MA) in a final volume of 30 µl. Plates were covered with top seal and transferred to a Synergy 2 Multiplatform automated plate reader (Biotek, Winooski, VT). After shaking for 2 min, plates were read every 2 h during 16 h: excitation using a 680/30 filter, emission using a 570/100 filter and data acquired using Gen5 software (Biotek). A total of 9 time points were generated during each experiment. Each set of samples was pipetted in 4 replicates. The final concentration of assay components was 29 nM EPPIN, 2 nM Q20E antibody, 0.04-7.5 µM SEMG1, and 10 µg/ml beads. Negative controls were performed under the same conditions in the absence of EPPIN or SEMG1 and in the presence of beads only. A specific signal for each time point was calculated by subtracting the background signal (obtained in the absence of SEMG1) from its respective total signal. EC50 values for the binding of EPPIN to each SEMG1 construct were calculated by non-linear regression curve fitting using the specific signal obtained after 16 h of incubation.

Competition experiments using AlphaScreen assay platform were performed as previously described [23]. Briefly, EPPIN (10 nM) and biotinylated (bt-)SEMG1^214-42 (1 nM) were incubated in the presence of increasing concentrations of SEMG1^214-42, SEMG1^74-42, SEMG1^74-8, or SEMG1^32-8 (10 pM – 8 µM) in a 30 µl reaction. The bead concentration was 15 µg/ml. The specific signal for each competitor concentration point was calculated as described above and the IC₅₀ values were calculated by nonlinear regression curve fitting.

Preparation of spermatozoa and analysis of sperm motility

Human semen samples were obtained from the Department of Obstetrics and Gynecology, University of North Carolina Memorial Hospital, Chapel Hill, NC. This study was approved by the Committee on the Protection of the Rights of Human Subjects at the University of North Carolina, School of Medicine Chapel Hill, NC. The Institutional Review Board determined that this study did not constitute human subjects research as defined under U.S. federal regulations [45 CFR 46.102 (d or f) and 21 CFR 56.102(c)(e)(l)] and the need for informed consent was waived. All samples were de-identified before further processing. Semen samples were allowed to liquefy for 30 min and subjected to standard semen analysis to determine acceptability and either used fresh or stored in liquid nitrogen. An isolate gradient (Irving Scientific, Irving, CA, cat. #99264) was used to prepare spermatozoa for further analysis [15].

Aliquots of sperm suspension (1500 cells/µl) were incubated with increasing concentrations of each SEMG1 construct (0 - 35.6 µM; diluted in modified M16 medium no phenol red (Millipore, Billerica, MA, cat. #MR/010/D) supplemented with 50 µM ZnCl₂ in 12- X 75-mm glass tubes at 37°C and 5% CO2 for 1-2 h. We repeated each experiment with spermatozoa from at least three different ejaculates and evaluated at least 100 cells (range 100-1000 cells) from each data point.

The analysis of sperm motility after incubation with SEMG1 constructs was carried out using computer-assisted sperm analysis (CASA; Ceros version 14.8 software, Hamilton-Thorne), as previously described [24]. In each case, the percentage sperm motility (%motility) and straight-line velocity (VSL; the average velocity measured in a straight line from the beginning to the end of track in µm/sec) were used to calculate the index of relative motility inhibition (iRMI) as: %motility*VSL. Normalized iRMI was calculated by dividing the iRMI of each experimental condition by its respective control (absence of SEMG1).

Statistical analysis

Results from AlphaScreen assays and CASA experiments were expressed as mean ± standard deviation (SD) or standard error of the mean (SEM) from the indicated number of independent experiments. For statistical significance, one-way analysis of variance (ANOVA) followed by the Bonferroni test was performed using GraphPad Prism 5.0 software. p < 0.05 was considered statistically significant.